{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.687870982192855,
  "eval_steps": 500,
  "global_step": 48000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 5.599731212901781e-05,
      "grad_norm": 1.5208414793014526,
      "learning_rate": 5e-09,
      "loss": 0.5181,
      "step": 1
    },
    {
      "epoch": 0.00011199462425803562,
      "grad_norm": 1.3712694644927979,
      "learning_rate": 1e-08,
      "loss": 0.376,
      "step": 2
    },
    {
      "epoch": 0.0001679919363870534,
      "grad_norm": 1.4184719324111938,
      "learning_rate": 1.5e-08,
      "loss": 0.4098,
      "step": 3
    },
    {
      "epoch": 0.00022398924851607123,
      "grad_norm": 1.4291776418685913,
      "learning_rate": 2e-08,
      "loss": 0.4771,
      "step": 4
    },
    {
      "epoch": 0.00027998656064508903,
      "grad_norm": 1.239039659500122,
      "learning_rate": 2.5000000000000002e-08,
      "loss": 0.5122,
      "step": 5
    },
    {
      "epoch": 0.0003359838727741068,
      "grad_norm": 1.3924657106399536,
      "learning_rate": 3e-08,
      "loss": 0.5332,
      "step": 6
    },
    {
      "epoch": 0.00039198118490312467,
      "grad_norm": 1.0782504081726074,
      "learning_rate": 3.5e-08,
      "loss": 0.3749,
      "step": 7
    },
    {
      "epoch": 0.00044797849703214247,
      "grad_norm": 1.5758692026138306,
      "learning_rate": 4e-08,
      "loss": 0.5218,
      "step": 8
    },
    {
      "epoch": 0.0005039758091611603,
      "grad_norm": 1.3339658975601196,
      "learning_rate": 4.5e-08,
      "loss": 0.5131,
      "step": 9
    },
    {
      "epoch": 0.0005599731212901781,
      "grad_norm": 1.5984278917312622,
      "learning_rate": 5.0000000000000004e-08,
      "loss": 0.5228,
      "step": 10
    },
    {
      "epoch": 0.0006159704334191959,
      "grad_norm": 1.2421988248825073,
      "learning_rate": 5.5e-08,
      "loss": 0.3985,
      "step": 11
    },
    {
      "epoch": 0.0006719677455482136,
      "grad_norm": 1.3071264028549194,
      "learning_rate": 6e-08,
      "loss": 0.4958,
      "step": 12
    },
    {
      "epoch": 0.0007279650576772315,
      "grad_norm": 1.7665575742721558,
      "learning_rate": 6.5e-08,
      "loss": 0.5927,
      "step": 13
    },
    {
      "epoch": 0.0007839623698062493,
      "grad_norm": 1.494536280632019,
      "learning_rate": 7e-08,
      "loss": 0.4472,
      "step": 14
    },
    {
      "epoch": 0.0008399596819352671,
      "grad_norm": 1.1757087707519531,
      "learning_rate": 7.500000000000001e-08,
      "loss": 0.4116,
      "step": 15
    },
    {
      "epoch": 0.0008959569940642849,
      "grad_norm": 1.6829992532730103,
      "learning_rate": 8e-08,
      "loss": 0.5333,
      "step": 16
    },
    {
      "epoch": 0.0009519543061933027,
      "grad_norm": 1.836796522140503,
      "learning_rate": 8.5e-08,
      "loss": 0.4205,
      "step": 17
    },
    {
      "epoch": 0.0010079516183223206,
      "grad_norm": 1.6840550899505615,
      "learning_rate": 9e-08,
      "loss": 0.4523,
      "step": 18
    },
    {
      "epoch": 0.0010639489304513383,
      "grad_norm": 1.5171109437942505,
      "learning_rate": 9.5e-08,
      "loss": 0.5163,
      "step": 19
    },
    {
      "epoch": 0.0011199462425803561,
      "grad_norm": 1.4950929880142212,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 0.4597,
      "step": 20
    },
    {
      "epoch": 0.001175943554709374,
      "grad_norm": 1.510266900062561,
      "learning_rate": 1.05e-07,
      "loss": 0.4991,
      "step": 21
    },
    {
      "epoch": 0.0012319408668383918,
      "grad_norm": 1.2690305709838867,
      "learning_rate": 1.1e-07,
      "loss": 0.4515,
      "step": 22
    },
    {
      "epoch": 0.0012879381789674097,
      "grad_norm": 1.4038795232772827,
      "learning_rate": 1.15e-07,
      "loss": 0.4929,
      "step": 23
    },
    {
      "epoch": 0.0013439354910964273,
      "grad_norm": 1.4907742738723755,
      "learning_rate": 1.2e-07,
      "loss": 0.766,
      "step": 24
    },
    {
      "epoch": 0.0013999328032254451,
      "grad_norm": 1.5836879014968872,
      "learning_rate": 1.2500000000000002e-07,
      "loss": 0.5249,
      "step": 25
    },
    {
      "epoch": 0.001455930115354463,
      "grad_norm": 2.3851399421691895,
      "learning_rate": 1.3e-07,
      "loss": 0.5666,
      "step": 26
    },
    {
      "epoch": 0.0015119274274834808,
      "grad_norm": 1.2829978466033936,
      "learning_rate": 1.35e-07,
      "loss": 0.4299,
      "step": 27
    },
    {
      "epoch": 0.0015679247396124987,
      "grad_norm": 1.6352183818817139,
      "learning_rate": 1.4e-07,
      "loss": 0.5148,
      "step": 28
    },
    {
      "epoch": 0.0016239220517415163,
      "grad_norm": 1.3571499586105347,
      "learning_rate": 1.45e-07,
      "loss": 0.4276,
      "step": 29
    },
    {
      "epoch": 0.0016799193638705342,
      "grad_norm": 1.4235525131225586,
      "learning_rate": 1.5000000000000002e-07,
      "loss": 0.6501,
      "step": 30
    },
    {
      "epoch": 0.001735916675999552,
      "grad_norm": 1.5542964935302734,
      "learning_rate": 1.55e-07,
      "loss": 0.504,
      "step": 31
    },
    {
      "epoch": 0.0017919139881285699,
      "grad_norm": 1.2780183553695679,
      "learning_rate": 1.6e-07,
      "loss": 0.4262,
      "step": 32
    },
    {
      "epoch": 0.0018479113002575877,
      "grad_norm": 1.2457411289215088,
      "learning_rate": 1.65e-07,
      "loss": 0.4542,
      "step": 33
    },
    {
      "epoch": 0.0019039086123866053,
      "grad_norm": 1.409238576889038,
      "learning_rate": 1.7e-07,
      "loss": 0.5514,
      "step": 34
    },
    {
      "epoch": 0.0019599059245156234,
      "grad_norm": 1.564186453819275,
      "learning_rate": 1.7500000000000002e-07,
      "loss": 0.4696,
      "step": 35
    },
    {
      "epoch": 0.0020159032366446413,
      "grad_norm": 1.1453927755355835,
      "learning_rate": 1.8e-07,
      "loss": 0.4193,
      "step": 36
    },
    {
      "epoch": 0.0020719005487736587,
      "grad_norm": 1.3826923370361328,
      "learning_rate": 1.8500000000000003e-07,
      "loss": 0.457,
      "step": 37
    },
    {
      "epoch": 0.0021278978609026765,
      "grad_norm": 1.4963278770446777,
      "learning_rate": 1.9e-07,
      "loss": 0.6355,
      "step": 38
    },
    {
      "epoch": 0.0021838951730316944,
      "grad_norm": 2.089097261428833,
      "learning_rate": 1.95e-07,
      "loss": 0.4094,
      "step": 39
    },
    {
      "epoch": 0.0022398924851607122,
      "grad_norm": 1.8463817834854126,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.5716,
      "step": 40
    },
    {
      "epoch": 0.00229588979728973,
      "grad_norm": 1.464522123336792,
      "learning_rate": 2.0500000000000002e-07,
      "loss": 0.519,
      "step": 41
    },
    {
      "epoch": 0.002351887109418748,
      "grad_norm": 1.4796329736709595,
      "learning_rate": 2.1e-07,
      "loss": 0.5112,
      "step": 42
    },
    {
      "epoch": 0.0024078844215477658,
      "grad_norm": 1.595494270324707,
      "learning_rate": 2.15e-07,
      "loss": 0.6506,
      "step": 43
    },
    {
      "epoch": 0.0024638817336767836,
      "grad_norm": 1.8938987255096436,
      "learning_rate": 2.2e-07,
      "loss": 0.4963,
      "step": 44
    },
    {
      "epoch": 0.0025198790458058015,
      "grad_norm": 1.4932397603988647,
      "learning_rate": 2.25e-07,
      "loss": 0.4818,
      "step": 45
    },
    {
      "epoch": 0.0025758763579348193,
      "grad_norm": 1.2402743101119995,
      "learning_rate": 2.3e-07,
      "loss": 0.4129,
      "step": 46
    },
    {
      "epoch": 0.0026318736700638367,
      "grad_norm": 1.3202379941940308,
      "learning_rate": 2.3500000000000003e-07,
      "loss": 0.4353,
      "step": 47
    },
    {
      "epoch": 0.0026878709821928546,
      "grad_norm": 1.5665568113327026,
      "learning_rate": 2.4e-07,
      "loss": 0.6673,
      "step": 48
    },
    {
      "epoch": 0.0027438682943218724,
      "grad_norm": 1.557573676109314,
      "learning_rate": 2.45e-07,
      "loss": 0.4768,
      "step": 49
    },
    {
      "epoch": 0.0027998656064508903,
      "grad_norm": 1.5255318880081177,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.528,
      "step": 50
    },
    {
      "epoch": 0.002855862918579908,
      "grad_norm": 1.3243638277053833,
      "learning_rate": 2.5500000000000005e-07,
      "loss": 0.5488,
      "step": 51
    },
    {
      "epoch": 0.002911860230708926,
      "grad_norm": 1.2211617231369019,
      "learning_rate": 2.6e-07,
      "loss": 0.44,
      "step": 52
    },
    {
      "epoch": 0.002967857542837944,
      "grad_norm": 2.0557689666748047,
      "learning_rate": 2.65e-07,
      "loss": 0.5321,
      "step": 53
    },
    {
      "epoch": 0.0030238548549669617,
      "grad_norm": 1.4214329719543457,
      "learning_rate": 2.7e-07,
      "loss": 0.3911,
      "step": 54
    },
    {
      "epoch": 0.0030798521670959795,
      "grad_norm": 1.35947847366333,
      "learning_rate": 2.75e-07,
      "loss": 0.4427,
      "step": 55
    },
    {
      "epoch": 0.0031358494792249974,
      "grad_norm": 1.4284133911132812,
      "learning_rate": 2.8e-07,
      "loss": 0.4277,
      "step": 56
    },
    {
      "epoch": 0.0031918467913540152,
      "grad_norm": 1.4069606065750122,
      "learning_rate": 2.85e-07,
      "loss": 0.4372,
      "step": 57
    },
    {
      "epoch": 0.0032478441034830326,
      "grad_norm": 2.8065545558929443,
      "learning_rate": 2.9e-07,
      "loss": 0.4607,
      "step": 58
    },
    {
      "epoch": 0.0033038414156120505,
      "grad_norm": 1.6844050884246826,
      "learning_rate": 2.95e-07,
      "loss": 0.6309,
      "step": 59
    },
    {
      "epoch": 0.0033598387277410683,
      "grad_norm": 1.4104372262954712,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 0.4987,
      "step": 60
    },
    {
      "epoch": 0.003415836039870086,
      "grad_norm": 1.5404714345932007,
      "learning_rate": 3.0500000000000004e-07,
      "loss": 0.5942,
      "step": 61
    },
    {
      "epoch": 0.003471833351999104,
      "grad_norm": 1.5271331071853638,
      "learning_rate": 3.1e-07,
      "loss": 0.4563,
      "step": 62
    },
    {
      "epoch": 0.003527830664128122,
      "grad_norm": 1.4416390657424927,
      "learning_rate": 3.15e-07,
      "loss": 0.5131,
      "step": 63
    },
    {
      "epoch": 0.0035838279762571397,
      "grad_norm": 1.416885256767273,
      "learning_rate": 3.2e-07,
      "loss": 0.4983,
      "step": 64
    },
    {
      "epoch": 0.0036398252883861576,
      "grad_norm": 1.6866464614868164,
      "learning_rate": 3.25e-07,
      "loss": 0.4553,
      "step": 65
    },
    {
      "epoch": 0.0036958226005151754,
      "grad_norm": 1.4955333471298218,
      "learning_rate": 3.3e-07,
      "loss": 0.4709,
      "step": 66
    },
    {
      "epoch": 0.0037518199126441933,
      "grad_norm": 1.3900035619735718,
      "learning_rate": 3.35e-07,
      "loss": 0.5626,
      "step": 67
    },
    {
      "epoch": 0.0038078172247732107,
      "grad_norm": 1.3770145177841187,
      "learning_rate": 3.4e-07,
      "loss": 0.5259,
      "step": 68
    },
    {
      "epoch": 0.0038638145369022285,
      "grad_norm": 1.2875261306762695,
      "learning_rate": 3.4500000000000003e-07,
      "loss": 0.4388,
      "step": 69
    },
    {
      "epoch": 0.003919811849031247,
      "grad_norm": 1.3735551834106445,
      "learning_rate": 3.5000000000000004e-07,
      "loss": 0.5785,
      "step": 70
    },
    {
      "epoch": 0.003975809161160264,
      "grad_norm": 1.9509772062301636,
      "learning_rate": 3.5500000000000004e-07,
      "loss": 0.5238,
      "step": 71
    },
    {
      "epoch": 0.0040318064732892825,
      "grad_norm": 1.8565374612808228,
      "learning_rate": 3.6e-07,
      "loss": 0.7029,
      "step": 72
    },
    {
      "epoch": 0.0040878037854183,
      "grad_norm": 1.4239002466201782,
      "learning_rate": 3.65e-07,
      "loss": 0.4697,
      "step": 73
    },
    {
      "epoch": 0.004143801097547317,
      "grad_norm": 1.467603087425232,
      "learning_rate": 3.7000000000000006e-07,
      "loss": 0.4178,
      "step": 74
    },
    {
      "epoch": 0.004199798409676336,
      "grad_norm": 1.1795387268066406,
      "learning_rate": 3.75e-07,
      "loss": 0.4297,
      "step": 75
    },
    {
      "epoch": 0.004255795721805353,
      "grad_norm": 1.1643917560577393,
      "learning_rate": 3.8e-07,
      "loss": 0.4823,
      "step": 76
    },
    {
      "epoch": 0.004311793033934371,
      "grad_norm": 1.4003437757492065,
      "learning_rate": 3.85e-07,
      "loss": 0.4341,
      "step": 77
    },
    {
      "epoch": 0.004367790346063389,
      "grad_norm": 1.246448278427124,
      "learning_rate": 3.9e-07,
      "loss": 0.4935,
      "step": 78
    },
    {
      "epoch": 0.004423787658192407,
      "grad_norm": 1.449789047241211,
      "learning_rate": 3.950000000000001e-07,
      "loss": 0.4214,
      "step": 79
    },
    {
      "epoch": 0.0044797849703214244,
      "grad_norm": 1.5086668729782104,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.5524,
      "step": 80
    },
    {
      "epoch": 0.004535782282450443,
      "grad_norm": 1.5325603485107422,
      "learning_rate": 4.05e-07,
      "loss": 0.4823,
      "step": 81
    },
    {
      "epoch": 0.00459177959457946,
      "grad_norm": 1.292855143547058,
      "learning_rate": 4.1000000000000004e-07,
      "loss": 0.5164,
      "step": 82
    },
    {
      "epoch": 0.0046477769067084776,
      "grad_norm": 1.3947184085845947,
      "learning_rate": 4.1500000000000005e-07,
      "loss": 0.5706,
      "step": 83
    },
    {
      "epoch": 0.004703774218837496,
      "grad_norm": 2.1091814041137695,
      "learning_rate": 4.2e-07,
      "loss": 0.5663,
      "step": 84
    },
    {
      "epoch": 0.004759771530966513,
      "grad_norm": 1.1866250038146973,
      "learning_rate": 4.2500000000000006e-07,
      "loss": 0.4289,
      "step": 85
    },
    {
      "epoch": 0.0048157688430955315,
      "grad_norm": 1.5813519954681396,
      "learning_rate": 4.3e-07,
      "loss": 0.8407,
      "step": 86
    },
    {
      "epoch": 0.004871766155224549,
      "grad_norm": 1.4272483587265015,
      "learning_rate": 4.3499999999999996e-07,
      "loss": 0.5547,
      "step": 87
    },
    {
      "epoch": 0.004927763467353567,
      "grad_norm": 1.3648320436477661,
      "learning_rate": 4.4e-07,
      "loss": 0.6105,
      "step": 88
    },
    {
      "epoch": 0.004983760779482585,
      "grad_norm": 1.4899494647979736,
      "learning_rate": 4.4500000000000003e-07,
      "loss": 0.5103,
      "step": 89
    },
    {
      "epoch": 0.005039758091611603,
      "grad_norm": 1.5137723684310913,
      "learning_rate": 4.5e-07,
      "loss": 0.6131,
      "step": 90
    },
    {
      "epoch": 0.00509575540374062,
      "grad_norm": 2.122572660446167,
      "learning_rate": 4.5500000000000004e-07,
      "loss": 0.4913,
      "step": 91
    },
    {
      "epoch": 0.005151752715869639,
      "grad_norm": 1.4125502109527588,
      "learning_rate": 4.6e-07,
      "loss": 0.4826,
      "step": 92
    },
    {
      "epoch": 0.005207750027998656,
      "grad_norm": 1.275382399559021,
      "learning_rate": 4.65e-07,
      "loss": 0.4789,
      "step": 93
    },
    {
      "epoch": 0.0052637473401276735,
      "grad_norm": 1.5469411611557007,
      "learning_rate": 4.7000000000000005e-07,
      "loss": 0.5324,
      "step": 94
    },
    {
      "epoch": 0.005319744652256692,
      "grad_norm": 1.503063440322876,
      "learning_rate": 4.75e-07,
      "loss": 0.5175,
      "step": 95
    },
    {
      "epoch": 0.005375741964385709,
      "grad_norm": 1.4756922721862793,
      "learning_rate": 4.8e-07,
      "loss": 0.6135,
      "step": 96
    },
    {
      "epoch": 0.0054317392765147274,
      "grad_norm": 1.2535488605499268,
      "learning_rate": 4.85e-07,
      "loss": 0.4542,
      "step": 97
    },
    {
      "epoch": 0.005487736588643745,
      "grad_norm": 1.3292878866195679,
      "learning_rate": 4.9e-07,
      "loss": 0.6138,
      "step": 98
    },
    {
      "epoch": 0.005543733900772763,
      "grad_norm": 1.4198826551437378,
      "learning_rate": 4.95e-07,
      "loss": 0.5016,
      "step": 99
    },
    {
      "epoch": 0.0055997312129017806,
      "grad_norm": 1.4598007202148438,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.4207,
      "step": 100
    },
    {
      "epoch": 0.005655728525030799,
      "grad_norm": 1.2995636463165283,
      "learning_rate": 5.05e-07,
      "loss": 0.6009,
      "step": 101
    },
    {
      "epoch": 0.005711725837159816,
      "grad_norm": 1.3568147420883179,
      "learning_rate": 5.100000000000001e-07,
      "loss": 0.405,
      "step": 102
    },
    {
      "epoch": 0.0057677231492888345,
      "grad_norm": 1.1702243089675903,
      "learning_rate": 5.15e-07,
      "loss": 0.3981,
      "step": 103
    },
    {
      "epoch": 0.005823720461417852,
      "grad_norm": 1.4444057941436768,
      "learning_rate": 5.2e-07,
      "loss": 0.4031,
      "step": 104
    },
    {
      "epoch": 0.005879717773546869,
      "grad_norm": 1.5215481519699097,
      "learning_rate": 5.250000000000001e-07,
      "loss": 0.6405,
      "step": 105
    },
    {
      "epoch": 0.005935715085675888,
      "grad_norm": 1.5095003843307495,
      "learning_rate": 5.3e-07,
      "loss": 0.6124,
      "step": 106
    },
    {
      "epoch": 0.005991712397804905,
      "grad_norm": 1.3150241374969482,
      "learning_rate": 5.35e-07,
      "loss": 0.4757,
      "step": 107
    },
    {
      "epoch": 0.006047709709933923,
      "grad_norm": 1.1489843130111694,
      "learning_rate": 5.4e-07,
      "loss": 0.4294,
      "step": 108
    },
    {
      "epoch": 0.006103707022062941,
      "grad_norm": 1.4761924743652344,
      "learning_rate": 5.450000000000001e-07,
      "loss": 0.5482,
      "step": 109
    },
    {
      "epoch": 0.006159704334191959,
      "grad_norm": 1.4262828826904297,
      "learning_rate": 5.5e-07,
      "loss": 0.4507,
      "step": 110
    },
    {
      "epoch": 0.0062157016463209765,
      "grad_norm": 1.3081737756729126,
      "learning_rate": 5.550000000000001e-07,
      "loss": 0.4249,
      "step": 111
    },
    {
      "epoch": 0.006271698958449995,
      "grad_norm": 2.10310435295105,
      "learning_rate": 5.6e-07,
      "loss": 0.5236,
      "step": 112
    },
    {
      "epoch": 0.006327696270579012,
      "grad_norm": 1.2382475137710571,
      "learning_rate": 5.65e-07,
      "loss": 0.3968,
      "step": 113
    },
    {
      "epoch": 0.0063836935827080304,
      "grad_norm": 2.0119569301605225,
      "learning_rate": 5.7e-07,
      "loss": 0.6051,
      "step": 114
    },
    {
      "epoch": 0.006439690894837048,
      "grad_norm": 1.5267064571380615,
      "learning_rate": 5.75e-07,
      "loss": 0.5475,
      "step": 115
    },
    {
      "epoch": 0.006495688206966065,
      "grad_norm": 2.7379636764526367,
      "learning_rate": 5.8e-07,
      "loss": 0.3968,
      "step": 116
    },
    {
      "epoch": 0.0065516855190950836,
      "grad_norm": 1.4146085977554321,
      "learning_rate": 5.85e-07,
      "loss": 0.6878,
      "step": 117
    },
    {
      "epoch": 0.006607682831224101,
      "grad_norm": 1.2528458833694458,
      "learning_rate": 5.9e-07,
      "loss": 0.4599,
      "step": 118
    },
    {
      "epoch": 0.006663680143353119,
      "grad_norm": 1.3054895401000977,
      "learning_rate": 5.95e-07,
      "loss": 0.4442,
      "step": 119
    },
    {
      "epoch": 0.006719677455482137,
      "grad_norm": 1.1482446193695068,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.4932,
      "step": 120
    },
    {
      "epoch": 0.006775674767611155,
      "grad_norm": 1.2546861171722412,
      "learning_rate": 6.05e-07,
      "loss": 0.4436,
      "step": 121
    },
    {
      "epoch": 0.006831672079740172,
      "grad_norm": 1.4483879804611206,
      "learning_rate": 6.100000000000001e-07,
      "loss": 0.5461,
      "step": 122
    },
    {
      "epoch": 0.006887669391869191,
      "grad_norm": 1.6083807945251465,
      "learning_rate": 6.15e-07,
      "loss": 0.5924,
      "step": 123
    },
    {
      "epoch": 0.006943666703998208,
      "grad_norm": 1.2077231407165527,
      "learning_rate": 6.2e-07,
      "loss": 0.4581,
      "step": 124
    },
    {
      "epoch": 0.0069996640161272255,
      "grad_norm": 1.2149949073791504,
      "learning_rate": 6.25e-07,
      "loss": 0.4393,
      "step": 125
    },
    {
      "epoch": 0.007055661328256244,
      "grad_norm": 1.4178581237792969,
      "learning_rate": 6.3e-07,
      "loss": 0.422,
      "step": 126
    },
    {
      "epoch": 0.007111658640385261,
      "grad_norm": 1.2476677894592285,
      "learning_rate": 6.35e-07,
      "loss": 0.496,
      "step": 127
    },
    {
      "epoch": 0.0071676559525142795,
      "grad_norm": 1.5778127908706665,
      "learning_rate": 6.4e-07,
      "loss": 0.6051,
      "step": 128
    },
    {
      "epoch": 0.007223653264643297,
      "grad_norm": 1.2792600393295288,
      "learning_rate": 6.450000000000001e-07,
      "loss": 0.467,
      "step": 129
    },
    {
      "epoch": 0.007279650576772315,
      "grad_norm": 2.7617485523223877,
      "learning_rate": 6.5e-07,
      "loss": 0.486,
      "step": 130
    },
    {
      "epoch": 0.007335647888901333,
      "grad_norm": 1.6663296222686768,
      "learning_rate": 6.550000000000001e-07,
      "loss": 0.5529,
      "step": 131
    },
    {
      "epoch": 0.007391645201030351,
      "grad_norm": 1.3751574754714966,
      "learning_rate": 6.6e-07,
      "loss": 0.5267,
      "step": 132
    },
    {
      "epoch": 0.007447642513159368,
      "grad_norm": 1.3827310800552368,
      "learning_rate": 6.65e-07,
      "loss": 0.4651,
      "step": 133
    },
    {
      "epoch": 0.0075036398252883866,
      "grad_norm": 1.3962993621826172,
      "learning_rate": 6.7e-07,
      "loss": 0.5695,
      "step": 134
    },
    {
      "epoch": 0.007559637137417404,
      "grad_norm": 1.3722511529922485,
      "learning_rate": 6.75e-07,
      "loss": 0.482,
      "step": 135
    },
    {
      "epoch": 0.007615634449546421,
      "grad_norm": 1.6061739921569824,
      "learning_rate": 6.8e-07,
      "loss": 0.4326,
      "step": 136
    },
    {
      "epoch": 0.00767163176167544,
      "grad_norm": 1.2789545059204102,
      "learning_rate": 6.85e-07,
      "loss": 0.3909,
      "step": 137
    },
    {
      "epoch": 0.007727629073804457,
      "grad_norm": 1.573917269706726,
      "learning_rate": 6.900000000000001e-07,
      "loss": 0.5517,
      "step": 138
    },
    {
      "epoch": 0.007783626385933475,
      "grad_norm": Infinity,
      "learning_rate": 6.900000000000001e-07,
      "loss": 0.5032,
      "step": 139
    },
    {
      "epoch": 0.007839623698062494,
      "grad_norm": 1.4517173767089844,
      "learning_rate": 6.95e-07,
      "loss": 0.5369,
      "step": 140
    },
    {
      "epoch": 0.007895621010191511,
      "grad_norm": 1.7139356136322021,
      "learning_rate": 7.000000000000001e-07,
      "loss": 0.5134,
      "step": 141
    },
    {
      "epoch": 0.007951618322320528,
      "grad_norm": 2.1147677898406982,
      "learning_rate": 7.05e-07,
      "loss": 0.6408,
      "step": 142
    },
    {
      "epoch": 0.008007615634449546,
      "grad_norm": 1.1497327089309692,
      "learning_rate": 7.100000000000001e-07,
      "loss": 0.4417,
      "step": 143
    },
    {
      "epoch": 0.008063612946578565,
      "grad_norm": 1.518349051475525,
      "learning_rate": 7.15e-07,
      "loss": 0.5329,
      "step": 144
    },
    {
      "epoch": 0.008119610258707582,
      "grad_norm": 1.3756242990493774,
      "learning_rate": 7.2e-07,
      "loss": 0.4445,
      "step": 145
    },
    {
      "epoch": 0.0081756075708366,
      "grad_norm": 1.3819228410720825,
      "learning_rate": 7.25e-07,
      "loss": 0.4424,
      "step": 146
    },
    {
      "epoch": 0.008231604882965617,
      "grad_norm": 1.4747849702835083,
      "learning_rate": 7.3e-07,
      "loss": 0.5919,
      "step": 147
    },
    {
      "epoch": 0.008287602195094635,
      "grad_norm": 1.4428592920303345,
      "learning_rate": 7.350000000000001e-07,
      "loss": 0.6117,
      "step": 148
    },
    {
      "epoch": 0.008343599507223654,
      "grad_norm": 1.2393968105316162,
      "learning_rate": 7.400000000000001e-07,
      "loss": 0.4504,
      "step": 149
    },
    {
      "epoch": 0.008399596819352671,
      "grad_norm": 1.219472885131836,
      "learning_rate": 7.450000000000001e-07,
      "loss": 0.4756,
      "step": 150
    },
    {
      "epoch": 0.008455594131481689,
      "grad_norm": 1.123125433921814,
      "learning_rate": 7.5e-07,
      "loss": 0.4298,
      "step": 151
    },
    {
      "epoch": 0.008511591443610706,
      "grad_norm": 1.373967170715332,
      "learning_rate": 7.550000000000001e-07,
      "loss": 0.4997,
      "step": 152
    },
    {
      "epoch": 0.008567588755739725,
      "grad_norm": 1.5358420610427856,
      "learning_rate": 7.6e-07,
      "loss": 0.6462,
      "step": 153
    },
    {
      "epoch": 0.008623586067868743,
      "grad_norm": 2.188631296157837,
      "learning_rate": 7.65e-07,
      "loss": 0.4562,
      "step": 154
    },
    {
      "epoch": 0.00867958337999776,
      "grad_norm": 1.5309580564498901,
      "learning_rate": 7.7e-07,
      "loss": 0.4204,
      "step": 155
    },
    {
      "epoch": 0.008735580692126777,
      "grad_norm": 1.2409296035766602,
      "learning_rate": 7.75e-07,
      "loss": 0.4228,
      "step": 156
    },
    {
      "epoch": 0.008791578004255795,
      "grad_norm": Infinity,
      "learning_rate": 7.75e-07,
      "loss": 0.5002,
      "step": 157
    },
    {
      "epoch": 0.008847575316384814,
      "grad_norm": 1.2380938529968262,
      "learning_rate": 7.8e-07,
      "loss": 0.4157,
      "step": 158
    },
    {
      "epoch": 0.008903572628513831,
      "grad_norm": 1.5206270217895508,
      "learning_rate": 7.85e-07,
      "loss": 0.5231,
      "step": 159
    },
    {
      "epoch": 0.008959569940642849,
      "grad_norm": 1.4741681814193726,
      "learning_rate": 7.900000000000002e-07,
      "loss": 0.4099,
      "step": 160
    },
    {
      "epoch": 0.009015567252771866,
      "grad_norm": 1.3287562131881714,
      "learning_rate": 7.950000000000001e-07,
      "loss": 0.3698,
      "step": 161
    },
    {
      "epoch": 0.009071564564900885,
      "grad_norm": 1.2945431470870972,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.4772,
      "step": 162
    },
    {
      "epoch": 0.009127561877029903,
      "grad_norm": 1.1704779863357544,
      "learning_rate": 8.05e-07,
      "loss": 0.5405,
      "step": 163
    },
    {
      "epoch": 0.00918355918915892,
      "grad_norm": 1.4360202550888062,
      "learning_rate": 8.1e-07,
      "loss": 0.6144,
      "step": 164
    },
    {
      "epoch": 0.009239556501287938,
      "grad_norm": 1.7820653915405273,
      "learning_rate": 8.149999999999999e-07,
      "loss": 0.6294,
      "step": 165
    },
    {
      "epoch": 0.009295553813416955,
      "grad_norm": 1.488885760307312,
      "learning_rate": 8.200000000000001e-07,
      "loss": 0.5041,
      "step": 166
    },
    {
      "epoch": 0.009351551125545974,
      "grad_norm": 1.0918800830841064,
      "learning_rate": 8.25e-07,
      "loss": 0.3615,
      "step": 167
    },
    {
      "epoch": 0.009407548437674992,
      "grad_norm": 1.3130441904067993,
      "learning_rate": 8.300000000000001e-07,
      "loss": 0.4489,
      "step": 168
    },
    {
      "epoch": 0.009463545749804009,
      "grad_norm": 1.2139885425567627,
      "learning_rate": 8.35e-07,
      "loss": 0.4585,
      "step": 169
    },
    {
      "epoch": 0.009519543061933027,
      "grad_norm": 1.5473610162734985,
      "learning_rate": 8.4e-07,
      "loss": 0.5744,
      "step": 170
    },
    {
      "epoch": 0.009575540374062046,
      "grad_norm": 1.2141534090042114,
      "learning_rate": 8.45e-07,
      "loss": 0.4876,
      "step": 171
    },
    {
      "epoch": 0.009631537686191063,
      "grad_norm": 1.2391549348831177,
      "learning_rate": 8.500000000000001e-07,
      "loss": 0.3436,
      "step": 172
    },
    {
      "epoch": 0.00968753499832008,
      "grad_norm": 1.445334792137146,
      "learning_rate": 8.550000000000001e-07,
      "loss": 0.4885,
      "step": 173
    },
    {
      "epoch": 0.009743532310449098,
      "grad_norm": 1.298776626586914,
      "learning_rate": 8.6e-07,
      "loss": 0.4514,
      "step": 174
    },
    {
      "epoch": 0.009799529622578117,
      "grad_norm": 1.5566740036010742,
      "learning_rate": 8.65e-07,
      "loss": 0.4906,
      "step": 175
    },
    {
      "epoch": 0.009855526934707134,
      "grad_norm": 1.181283712387085,
      "learning_rate": 8.699999999999999e-07,
      "loss": 0.5238,
      "step": 176
    },
    {
      "epoch": 0.009911524246836152,
      "grad_norm": 1.2867265939712524,
      "learning_rate": 8.750000000000001e-07,
      "loss": 0.4365,
      "step": 177
    },
    {
      "epoch": 0.00996752155896517,
      "grad_norm": 1.2805672883987427,
      "learning_rate": 8.8e-07,
      "loss": 0.4703,
      "step": 178
    },
    {
      "epoch": 0.010023518871094187,
      "grad_norm": 1.4181572198867798,
      "learning_rate": 8.850000000000001e-07,
      "loss": 0.545,
      "step": 179
    },
    {
      "epoch": 0.010079516183223206,
      "grad_norm": 1.273695945739746,
      "learning_rate": 8.900000000000001e-07,
      "loss": 0.4459,
      "step": 180
    },
    {
      "epoch": 0.010135513495352223,
      "grad_norm": 1.3415385484695435,
      "learning_rate": 8.95e-07,
      "loss": 0.5312,
      "step": 181
    },
    {
      "epoch": 0.01019151080748124,
      "grad_norm": 1.2442187070846558,
      "learning_rate": 9e-07,
      "loss": 0.4855,
      "step": 182
    },
    {
      "epoch": 0.010247508119610258,
      "grad_norm": 1.607743740081787,
      "learning_rate": 9.050000000000001e-07,
      "loss": 0.636,
      "step": 183
    },
    {
      "epoch": 0.010303505431739277,
      "grad_norm": 1.3131659030914307,
      "learning_rate": 9.100000000000001e-07,
      "loss": 0.4415,
      "step": 184
    },
    {
      "epoch": 0.010359502743868295,
      "grad_norm": 1.1934609413146973,
      "learning_rate": 9.15e-07,
      "loss": 0.4209,
      "step": 185
    },
    {
      "epoch": 0.010415500055997312,
      "grad_norm": 1.2915111780166626,
      "learning_rate": 9.2e-07,
      "loss": 0.4771,
      "step": 186
    },
    {
      "epoch": 0.01047149736812633,
      "grad_norm": 1.5296729803085327,
      "learning_rate": 9.25e-07,
      "loss": 0.6426,
      "step": 187
    },
    {
      "epoch": 0.010527494680255347,
      "grad_norm": 2.22491717338562,
      "learning_rate": 9.3e-07,
      "loss": 0.5904,
      "step": 188
    },
    {
      "epoch": 0.010583491992384366,
      "grad_norm": 1.2684471607208252,
      "learning_rate": 9.350000000000002e-07,
      "loss": 0.5164,
      "step": 189
    },
    {
      "epoch": 0.010639489304513383,
      "grad_norm": 1.7580739259719849,
      "learning_rate": 9.400000000000001e-07,
      "loss": 0.471,
      "step": 190
    },
    {
      "epoch": 0.010695486616642401,
      "grad_norm": 1.3328614234924316,
      "learning_rate": 9.450000000000001e-07,
      "loss": 0.5199,
      "step": 191
    },
    {
      "epoch": 0.010751483928771418,
      "grad_norm": 10.069008827209473,
      "learning_rate": 9.5e-07,
      "loss": 0.3754,
      "step": 192
    },
    {
      "epoch": 0.010807481240900437,
      "grad_norm": 1.2109394073486328,
      "learning_rate": 9.55e-07,
      "loss": 0.5417,
      "step": 193
    },
    {
      "epoch": 0.010863478553029455,
      "grad_norm": 1.279045581817627,
      "learning_rate": 9.6e-07,
      "loss": 0.6722,
      "step": 194
    },
    {
      "epoch": 0.010919475865158472,
      "grad_norm": 1.482125997543335,
      "learning_rate": 9.65e-07,
      "loss": 0.4102,
      "step": 195
    },
    {
      "epoch": 0.01097547317728749,
      "grad_norm": 1.2239766120910645,
      "learning_rate": 9.7e-07,
      "loss": 0.4908,
      "step": 196
    },
    {
      "epoch": 0.011031470489416509,
      "grad_norm": 1.3760238885879517,
      "learning_rate": 9.75e-07,
      "loss": 0.4594,
      "step": 197
    },
    {
      "epoch": 0.011087467801545526,
      "grad_norm": 1.3297518491744995,
      "learning_rate": 9.8e-07,
      "loss": 0.5052,
      "step": 198
    },
    {
      "epoch": 0.011143465113674544,
      "grad_norm": 1.2966132164001465,
      "learning_rate": 9.849999999999999e-07,
      "loss": 0.4952,
      "step": 199
    },
    {
      "epoch": 0.011199462425803561,
      "grad_norm": 1.4916179180145264,
      "learning_rate": 9.9e-07,
      "loss": 0.4624,
      "step": 200
    },
    {
      "epoch": 0.011255459737932579,
      "grad_norm": 1.8758313655853271,
      "learning_rate": 9.95e-07,
      "loss": 0.6089,
      "step": 201
    },
    {
      "epoch": 0.011311457050061598,
      "grad_norm": 1.1678751707077026,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.4251,
      "step": 202
    },
    {
      "epoch": 0.011367454362190615,
      "grad_norm": 1.1295413970947266,
      "learning_rate": 1.0050000000000001e-06,
      "loss": 0.394,
      "step": 203
    },
    {
      "epoch": 0.011423451674319633,
      "grad_norm": 1.2641093730926514,
      "learning_rate": 1.01e-06,
      "loss": 0.4225,
      "step": 204
    },
    {
      "epoch": 0.01147944898644865,
      "grad_norm": 1.326363205909729,
      "learning_rate": 1.015e-06,
      "loss": 0.5714,
      "step": 205
    },
    {
      "epoch": 0.011535446298577669,
      "grad_norm": 0.9996131062507629,
      "learning_rate": 1.0200000000000002e-06,
      "loss": 0.3477,
      "step": 206
    },
    {
      "epoch": 0.011591443610706686,
      "grad_norm": 1.3686496019363403,
      "learning_rate": 1.0250000000000001e-06,
      "loss": 0.4609,
      "step": 207
    },
    {
      "epoch": 0.011647440922835704,
      "grad_norm": 1.1996912956237793,
      "learning_rate": 1.03e-06,
      "loss": 0.5012,
      "step": 208
    },
    {
      "epoch": 0.011703438234964721,
      "grad_norm": 1.2017878293991089,
      "learning_rate": 1.035e-06,
      "loss": 0.4679,
      "step": 209
    },
    {
      "epoch": 0.011759435547093739,
      "grad_norm": 1.3492447137832642,
      "learning_rate": 1.04e-06,
      "loss": 0.6228,
      "step": 210
    },
    {
      "epoch": 0.011815432859222758,
      "grad_norm": 1.0122253894805908,
      "learning_rate": 1.045e-06,
      "loss": 0.4258,
      "step": 211
    },
    {
      "epoch": 0.011871430171351775,
      "grad_norm": 1.3060179948806763,
      "learning_rate": 1.0500000000000001e-06,
      "loss": 0.4753,
      "step": 212
    },
    {
      "epoch": 0.011927427483480793,
      "grad_norm": 1.1057965755462646,
      "learning_rate": 1.055e-06,
      "loss": 0.372,
      "step": 213
    },
    {
      "epoch": 0.01198342479560981,
      "grad_norm": 1.383988857269287,
      "learning_rate": 1.06e-06,
      "loss": 0.6251,
      "step": 214
    },
    {
      "epoch": 0.01203942210773883,
      "grad_norm": 1.1000547409057617,
      "learning_rate": 1.065e-06,
      "loss": 0.3334,
      "step": 215
    },
    {
      "epoch": 0.012095419419867847,
      "grad_norm": 1.307716727256775,
      "learning_rate": 1.07e-06,
      "loss": 0.4021,
      "step": 216
    },
    {
      "epoch": 0.012151416731996864,
      "grad_norm": 1.3594478368759155,
      "learning_rate": 1.0749999999999999e-06,
      "loss": 0.4034,
      "step": 217
    },
    {
      "epoch": 0.012207414044125882,
      "grad_norm": 1.0731086730957031,
      "learning_rate": 1.08e-06,
      "loss": 0.4315,
      "step": 218
    },
    {
      "epoch": 0.012263411356254899,
      "grad_norm": 1.0504783391952515,
      "learning_rate": 1.085e-06,
      "loss": 0.3754,
      "step": 219
    },
    {
      "epoch": 0.012319408668383918,
      "grad_norm": 1.1919739246368408,
      "learning_rate": 1.0900000000000002e-06,
      "loss": 0.4962,
      "step": 220
    },
    {
      "epoch": 0.012375405980512936,
      "grad_norm": 1.6576035022735596,
      "learning_rate": 1.095e-06,
      "loss": 0.6879,
      "step": 221
    },
    {
      "epoch": 0.012431403292641953,
      "grad_norm": 1.518970251083374,
      "learning_rate": 1.1e-06,
      "loss": 0.6722,
      "step": 222
    },
    {
      "epoch": 0.01248740060477097,
      "grad_norm": 1.672376036643982,
      "learning_rate": 1.1050000000000002e-06,
      "loss": 0.5444,
      "step": 223
    },
    {
      "epoch": 0.01254339791689999,
      "grad_norm": 1.0940489768981934,
      "learning_rate": 1.1100000000000002e-06,
      "loss": 0.3958,
      "step": 224
    },
    {
      "epoch": 0.012599395229029007,
      "grad_norm": 1.4551310539245605,
      "learning_rate": 1.1150000000000001e-06,
      "loss": 0.5054,
      "step": 225
    },
    {
      "epoch": 0.012655392541158024,
      "grad_norm": 1.3679637908935547,
      "learning_rate": 1.12e-06,
      "loss": 0.4151,
      "step": 226
    },
    {
      "epoch": 0.012711389853287042,
      "grad_norm": 1.242140293121338,
      "learning_rate": 1.125e-06,
      "loss": 0.401,
      "step": 227
    },
    {
      "epoch": 0.012767387165416061,
      "grad_norm": 1.0123270750045776,
      "learning_rate": 1.13e-06,
      "loss": 0.4147,
      "step": 228
    },
    {
      "epoch": 0.012823384477545078,
      "grad_norm": 1.3932843208312988,
      "learning_rate": 1.1350000000000001e-06,
      "loss": 0.4361,
      "step": 229
    },
    {
      "epoch": 0.012879381789674096,
      "grad_norm": 1.0555158853530884,
      "learning_rate": 1.14e-06,
      "loss": 0.4174,
      "step": 230
    },
    {
      "epoch": 0.012935379101803113,
      "grad_norm": 1.380471110343933,
      "learning_rate": 1.145e-06,
      "loss": 0.4259,
      "step": 231
    },
    {
      "epoch": 0.01299137641393213,
      "grad_norm": 1.2852363586425781,
      "learning_rate": 1.15e-06,
      "loss": 0.4641,
      "step": 232
    },
    {
      "epoch": 0.01304737372606115,
      "grad_norm": 1.1543710231781006,
      "learning_rate": 1.155e-06,
      "loss": 0.5133,
      "step": 233
    },
    {
      "epoch": 0.013103371038190167,
      "grad_norm": 1.174673318862915,
      "learning_rate": 1.16e-06,
      "loss": 0.3946,
      "step": 234
    },
    {
      "epoch": 0.013159368350319185,
      "grad_norm": 1.1139841079711914,
      "learning_rate": 1.165e-06,
      "loss": 0.4283,
      "step": 235
    },
    {
      "epoch": 0.013215365662448202,
      "grad_norm": 1.6161155700683594,
      "learning_rate": 1.17e-06,
      "loss": 0.5241,
      "step": 236
    },
    {
      "epoch": 0.013271362974577221,
      "grad_norm": 1.313589334487915,
      "learning_rate": 1.175e-06,
      "loss": 0.5173,
      "step": 237
    },
    {
      "epoch": 0.013327360286706239,
      "grad_norm": 1.4072622060775757,
      "learning_rate": 1.18e-06,
      "loss": 0.7632,
      "step": 238
    },
    {
      "epoch": 0.013383357598835256,
      "grad_norm": 1.611053228378296,
      "learning_rate": 1.185e-06,
      "loss": 0.526,
      "step": 239
    },
    {
      "epoch": 0.013439354910964273,
      "grad_norm": 1.2626253366470337,
      "learning_rate": 1.19e-06,
      "loss": 0.6914,
      "step": 240
    },
    {
      "epoch": 0.01349535222309329,
      "grad_norm": 1.1688967943191528,
      "learning_rate": 1.1950000000000002e-06,
      "loss": 0.5033,
      "step": 241
    },
    {
      "epoch": 0.01355134953522231,
      "grad_norm": 1.1805068254470825,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.5048,
      "step": 242
    },
    {
      "epoch": 0.013607346847351327,
      "grad_norm": 1.3243757486343384,
      "learning_rate": 1.2050000000000001e-06,
      "loss": 0.5549,
      "step": 243
    },
    {
      "epoch": 0.013663344159480345,
      "grad_norm": 1.4326049089431763,
      "learning_rate": 1.21e-06,
      "loss": 0.733,
      "step": 244
    },
    {
      "epoch": 0.013719341471609362,
      "grad_norm": 1.2837655544281006,
      "learning_rate": 1.215e-06,
      "loss": 0.5065,
      "step": 245
    },
    {
      "epoch": 0.013775338783738381,
      "grad_norm": 1.687691330909729,
      "learning_rate": 1.2200000000000002e-06,
      "loss": 0.605,
      "step": 246
    },
    {
      "epoch": 0.013831336095867399,
      "grad_norm": 1.4184305667877197,
      "learning_rate": 1.2250000000000001e-06,
      "loss": 0.4447,
      "step": 247
    },
    {
      "epoch": 0.013887333407996416,
      "grad_norm": 1.5195060968399048,
      "learning_rate": 1.23e-06,
      "loss": 0.4551,
      "step": 248
    },
    {
      "epoch": 0.013943330720125434,
      "grad_norm": 1.5335530042648315,
      "learning_rate": 1.235e-06,
      "loss": 0.5239,
      "step": 249
    },
    {
      "epoch": 0.013999328032254451,
      "grad_norm": 1.2263363599777222,
      "learning_rate": 1.24e-06,
      "loss": 0.4089,
      "step": 250
    },
    {
      "epoch": 0.01405532534438347,
      "grad_norm": 1.3511929512023926,
      "learning_rate": 1.245e-06,
      "loss": 0.5311,
      "step": 251
    },
    {
      "epoch": 0.014111322656512488,
      "grad_norm": 1.4361425638198853,
      "learning_rate": 1.25e-06,
      "loss": 0.5113,
      "step": 252
    },
    {
      "epoch": 0.014167319968641505,
      "grad_norm": 1.3185176849365234,
      "learning_rate": 1.255e-06,
      "loss": 0.5712,
      "step": 253
    },
    {
      "epoch": 0.014223317280770522,
      "grad_norm": 1.7490395307540894,
      "learning_rate": 1.26e-06,
      "loss": 0.4938,
      "step": 254
    },
    {
      "epoch": 0.014279314592899542,
      "grad_norm": 1.2518575191497803,
      "learning_rate": 1.265e-06,
      "loss": 0.4094,
      "step": 255
    },
    {
      "epoch": 0.014335311905028559,
      "grad_norm": 1.9215528964996338,
      "learning_rate": 1.27e-06,
      "loss": 0.5999,
      "step": 256
    },
    {
      "epoch": 0.014391309217157576,
      "grad_norm": 1.1665399074554443,
      "learning_rate": 1.275e-06,
      "loss": 0.4608,
      "step": 257
    },
    {
      "epoch": 0.014447306529286594,
      "grad_norm": 1.461685299873352,
      "learning_rate": 1.28e-06,
      "loss": 0.593,
      "step": 258
    },
    {
      "epoch": 0.014503303841415613,
      "grad_norm": 1.0450584888458252,
      "learning_rate": 1.2850000000000002e-06,
      "loss": 0.3811,
      "step": 259
    },
    {
      "epoch": 0.01455930115354463,
      "grad_norm": 1.2816405296325684,
      "learning_rate": 1.2900000000000001e-06,
      "loss": 0.4941,
      "step": 260
    },
    {
      "epoch": 0.014615298465673648,
      "grad_norm": 1.2557902336120605,
      "learning_rate": 1.295e-06,
      "loss": 0.459,
      "step": 261
    },
    {
      "epoch": 0.014671295777802665,
      "grad_norm": 1.1175843477249146,
      "learning_rate": 1.3e-06,
      "loss": 0.4936,
      "step": 262
    },
    {
      "epoch": 0.014727293089931683,
      "grad_norm": 1.479242205619812,
      "learning_rate": 1.3050000000000002e-06,
      "loss": 0.4825,
      "step": 263
    },
    {
      "epoch": 0.014783290402060702,
      "grad_norm": 1.5817683935165405,
      "learning_rate": 1.3100000000000002e-06,
      "loss": 0.3805,
      "step": 264
    },
    {
      "epoch": 0.01483928771418972,
      "grad_norm": 1.5696752071380615,
      "learning_rate": 1.3150000000000001e-06,
      "loss": 0.6506,
      "step": 265
    },
    {
      "epoch": 0.014895285026318737,
      "grad_norm": 1.4629154205322266,
      "learning_rate": 1.32e-06,
      "loss": 0.4585,
      "step": 266
    },
    {
      "epoch": 0.014951282338447754,
      "grad_norm": 1.5436128377914429,
      "learning_rate": 1.325e-06,
      "loss": 0.5556,
      "step": 267
    },
    {
      "epoch": 0.015007279650576773,
      "grad_norm": 1.4215129613876343,
      "learning_rate": 1.33e-06,
      "loss": 0.4365,
      "step": 268
    },
    {
      "epoch": 0.01506327696270579,
      "grad_norm": 1.3057754039764404,
      "learning_rate": 1.3350000000000001e-06,
      "loss": 0.4787,
      "step": 269
    },
    {
      "epoch": 0.015119274274834808,
      "grad_norm": 1.1739662885665894,
      "learning_rate": 1.34e-06,
      "loss": 0.5333,
      "step": 270
    },
    {
      "epoch": 0.015175271586963825,
      "grad_norm": 1.51952064037323,
      "learning_rate": 1.345e-06,
      "loss": 0.3872,
      "step": 271
    },
    {
      "epoch": 0.015231268899092843,
      "grad_norm": 1.7925301790237427,
      "learning_rate": 1.35e-06,
      "loss": 0.5211,
      "step": 272
    },
    {
      "epoch": 0.015287266211221862,
      "grad_norm": 1.2592473030090332,
      "learning_rate": 1.355e-06,
      "loss": 0.4538,
      "step": 273
    },
    {
      "epoch": 0.01534326352335088,
      "grad_norm": 1.247963547706604,
      "learning_rate": 1.36e-06,
      "loss": 0.4779,
      "step": 274
    },
    {
      "epoch": 0.015399260835479897,
      "grad_norm": 1.4560402631759644,
      "learning_rate": 1.365e-06,
      "loss": 0.4341,
      "step": 275
    },
    {
      "epoch": 0.015455258147608914,
      "grad_norm": 1.4166802167892456,
      "learning_rate": 1.37e-06,
      "loss": 0.5152,
      "step": 276
    },
    {
      "epoch": 0.015511255459737933,
      "grad_norm": 1.4385263919830322,
      "learning_rate": 1.3750000000000002e-06,
      "loss": 0.4547,
      "step": 277
    },
    {
      "epoch": 0.01556725277186695,
      "grad_norm": 1.5150035619735718,
      "learning_rate": 1.3800000000000001e-06,
      "loss": 0.5362,
      "step": 278
    },
    {
      "epoch": 0.015623250083995968,
      "grad_norm": 3.9275166988372803,
      "learning_rate": 1.385e-06,
      "loss": 0.5216,
      "step": 279
    },
    {
      "epoch": 0.015679247396124987,
      "grad_norm": 1.3215997219085693,
      "learning_rate": 1.39e-06,
      "loss": 0.6029,
      "step": 280
    },
    {
      "epoch": 0.015735244708254003,
      "grad_norm": 1.7647961378097534,
      "learning_rate": 1.3950000000000002e-06,
      "loss": 0.4077,
      "step": 281
    },
    {
      "epoch": 0.015791242020383022,
      "grad_norm": 1.5197356939315796,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.6688,
      "step": 282
    },
    {
      "epoch": 0.015847239332512038,
      "grad_norm": 1.4369826316833496,
      "learning_rate": 1.405e-06,
      "loss": 0.4598,
      "step": 283
    },
    {
      "epoch": 0.015903236644641057,
      "grad_norm": 1.3526933193206787,
      "learning_rate": 1.41e-06,
      "loss": 0.5358,
      "step": 284
    },
    {
      "epoch": 0.015959233956770076,
      "grad_norm": 1.2903668880462646,
      "learning_rate": 1.415e-06,
      "loss": 0.3934,
      "step": 285
    },
    {
      "epoch": 0.016015231268899092,
      "grad_norm": 1.2543387413024902,
      "learning_rate": 1.4200000000000002e-06,
      "loss": 0.496,
      "step": 286
    },
    {
      "epoch": 0.01607122858102811,
      "grad_norm": 1.3359887599945068,
      "learning_rate": 1.4250000000000001e-06,
      "loss": 0.5165,
      "step": 287
    },
    {
      "epoch": 0.01612722589315713,
      "grad_norm": 1.1500910520553589,
      "learning_rate": 1.43e-06,
      "loss": 0.3095,
      "step": 288
    },
    {
      "epoch": 0.016183223205286146,
      "grad_norm": 1.53554368019104,
      "learning_rate": 1.435e-06,
      "loss": 0.6023,
      "step": 289
    },
    {
      "epoch": 0.016239220517415165,
      "grad_norm": 1.2718192338943481,
      "learning_rate": 1.44e-06,
      "loss": 0.4301,
      "step": 290
    },
    {
      "epoch": 0.01629521782954418,
      "grad_norm": 1.4737380743026733,
      "learning_rate": 1.445e-06,
      "loss": 0.4529,
      "step": 291
    },
    {
      "epoch": 0.0163512151416732,
      "grad_norm": 1.2656103372573853,
      "learning_rate": 1.45e-06,
      "loss": 0.5206,
      "step": 292
    },
    {
      "epoch": 0.01640721245380222,
      "grad_norm": 1.2045220136642456,
      "learning_rate": 1.455e-06,
      "loss": 0.4806,
      "step": 293
    },
    {
      "epoch": 0.016463209765931235,
      "grad_norm": 1.7034382820129395,
      "learning_rate": 1.46e-06,
      "loss": 0.4972,
      "step": 294
    },
    {
      "epoch": 0.016519207078060254,
      "grad_norm": 1.1198574304580688,
      "learning_rate": 1.465e-06,
      "loss": 0.5209,
      "step": 295
    },
    {
      "epoch": 0.01657520439018927,
      "grad_norm": 1.133908748626709,
      "learning_rate": 1.4700000000000001e-06,
      "loss": 0.5018,
      "step": 296
    },
    {
      "epoch": 0.01663120170231829,
      "grad_norm": 1.2515907287597656,
      "learning_rate": 1.475e-06,
      "loss": 0.5826,
      "step": 297
    },
    {
      "epoch": 0.016687199014447308,
      "grad_norm": 1.297999382019043,
      "learning_rate": 1.4800000000000002e-06,
      "loss": 0.4961,
      "step": 298
    },
    {
      "epoch": 0.016743196326576323,
      "grad_norm": 1.4551876783370972,
      "learning_rate": 1.4850000000000002e-06,
      "loss": 0.5899,
      "step": 299
    },
    {
      "epoch": 0.016799193638705343,
      "grad_norm": 1.346625566482544,
      "learning_rate": 1.4900000000000001e-06,
      "loss": 0.5737,
      "step": 300
    },
    {
      "epoch": 0.016855190950834358,
      "grad_norm": 1.3084245920181274,
      "learning_rate": 1.495e-06,
      "loss": 0.5083,
      "step": 301
    },
    {
      "epoch": 0.016911188262963377,
      "grad_norm": 1.501230001449585,
      "learning_rate": 1.5e-06,
      "loss": 0.5419,
      "step": 302
    },
    {
      "epoch": 0.016967185575092397,
      "grad_norm": 1.3253297805786133,
      "learning_rate": 1.505e-06,
      "loss": 0.5299,
      "step": 303
    },
    {
      "epoch": 0.017023182887221412,
      "grad_norm": 1.3683420419692993,
      "learning_rate": 1.5100000000000002e-06,
      "loss": 0.5981,
      "step": 304
    },
    {
      "epoch": 0.01707918019935043,
      "grad_norm": 1.3841209411621094,
      "learning_rate": 1.5150000000000001e-06,
      "loss": 0.4693,
      "step": 305
    },
    {
      "epoch": 0.01713517751147945,
      "grad_norm": 1.3428679704666138,
      "learning_rate": 1.52e-06,
      "loss": 0.445,
      "step": 306
    },
    {
      "epoch": 0.017191174823608466,
      "grad_norm": 1.248568058013916,
      "learning_rate": 1.525e-06,
      "loss": 0.3911,
      "step": 307
    },
    {
      "epoch": 0.017247172135737485,
      "grad_norm": 1.3845674991607666,
      "learning_rate": 1.53e-06,
      "loss": 0.546,
      "step": 308
    },
    {
      "epoch": 0.0173031694478665,
      "grad_norm": 1.465867519378662,
      "learning_rate": 1.5350000000000001e-06,
      "loss": 0.6043,
      "step": 309
    },
    {
      "epoch": 0.01735916675999552,
      "grad_norm": 1.1764181852340698,
      "learning_rate": 1.54e-06,
      "loss": 0.377,
      "step": 310
    },
    {
      "epoch": 0.01741516407212454,
      "grad_norm": 1.1894500255584717,
      "learning_rate": 1.545e-06,
      "loss": 0.4326,
      "step": 311
    },
    {
      "epoch": 0.017471161384253555,
      "grad_norm": 1.270615577697754,
      "learning_rate": 1.55e-06,
      "loss": 0.448,
      "step": 312
    },
    {
      "epoch": 0.017527158696382574,
      "grad_norm": 1.9738068580627441,
      "learning_rate": 1.555e-06,
      "loss": 0.5594,
      "step": 313
    },
    {
      "epoch": 0.01758315600851159,
      "grad_norm": 1.4984630346298218,
      "learning_rate": 1.56e-06,
      "loss": 0.5236,
      "step": 314
    },
    {
      "epoch": 0.01763915332064061,
      "grad_norm": 1.8723793029785156,
      "learning_rate": 1.565e-06,
      "loss": 0.6548,
      "step": 315
    },
    {
      "epoch": 0.017695150632769628,
      "grad_norm": 1.298327922821045,
      "learning_rate": 1.57e-06,
      "loss": 0.4244,
      "step": 316
    },
    {
      "epoch": 0.017751147944898644,
      "grad_norm": 1.0812171697616577,
      "learning_rate": 1.5750000000000002e-06,
      "loss": 0.4079,
      "step": 317
    },
    {
      "epoch": 0.017807145257027663,
      "grad_norm": 1.3327000141143799,
      "learning_rate": 1.5800000000000003e-06,
      "loss": 0.6322,
      "step": 318
    },
    {
      "epoch": 0.017863142569156682,
      "grad_norm": 1.8076618909835815,
      "learning_rate": 1.585e-06,
      "loss": 0.409,
      "step": 319
    },
    {
      "epoch": 0.017919139881285698,
      "grad_norm": 1.5142481327056885,
      "learning_rate": 1.5900000000000002e-06,
      "loss": 0.5574,
      "step": 320
    },
    {
      "epoch": 0.017975137193414717,
      "grad_norm": 1.940686821937561,
      "learning_rate": 1.595e-06,
      "loss": 0.533,
      "step": 321
    },
    {
      "epoch": 0.018031134505543733,
      "grad_norm": 1.33735990524292,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.4811,
      "step": 322
    },
    {
      "epoch": 0.018087131817672752,
      "grad_norm": 1.296083927154541,
      "learning_rate": 1.6049999999999999e-06,
      "loss": 0.4361,
      "step": 323
    },
    {
      "epoch": 0.01814312912980177,
      "grad_norm": 1.1967061758041382,
      "learning_rate": 1.61e-06,
      "loss": 0.4532,
      "step": 324
    },
    {
      "epoch": 0.018199126441930787,
      "grad_norm": 1.6449378728866577,
      "learning_rate": 1.6150000000000002e-06,
      "loss": 0.7581,
      "step": 325
    },
    {
      "epoch": 0.018255123754059806,
      "grad_norm": 1.3829392194747925,
      "learning_rate": 1.62e-06,
      "loss": 0.5744,
      "step": 326
    },
    {
      "epoch": 0.01831112106618882,
      "grad_norm": 1.4353978633880615,
      "learning_rate": 1.6250000000000001e-06,
      "loss": 0.5701,
      "step": 327
    },
    {
      "epoch": 0.01836711837831784,
      "grad_norm": 1.405627965927124,
      "learning_rate": 1.6299999999999999e-06,
      "loss": 0.4762,
      "step": 328
    },
    {
      "epoch": 0.01842311569044686,
      "grad_norm": 1.2175483703613281,
      "learning_rate": 1.635e-06,
      "loss": 0.4396,
      "step": 329
    },
    {
      "epoch": 0.018479113002575875,
      "grad_norm": 1.2880176305770874,
      "learning_rate": 1.6400000000000002e-06,
      "loss": 0.3867,
      "step": 330
    },
    {
      "epoch": 0.018535110314704895,
      "grad_norm": 1.3141285181045532,
      "learning_rate": 1.645e-06,
      "loss": 0.4472,
      "step": 331
    },
    {
      "epoch": 0.01859110762683391,
      "grad_norm": 1.2125033140182495,
      "learning_rate": 1.65e-06,
      "loss": 0.5351,
      "step": 332
    },
    {
      "epoch": 0.01864710493896293,
      "grad_norm": 1.4135538339614868,
      "learning_rate": 1.655e-06,
      "loss": 0.6807,
      "step": 333
    },
    {
      "epoch": 0.01870310225109195,
      "grad_norm": 1.2327980995178223,
      "learning_rate": 1.6600000000000002e-06,
      "loss": 0.4119,
      "step": 334
    },
    {
      "epoch": 0.018759099563220964,
      "grad_norm": 1.4258291721343994,
      "learning_rate": 1.6650000000000002e-06,
      "loss": 0.6192,
      "step": 335
    },
    {
      "epoch": 0.018815096875349983,
      "grad_norm": 1.2788861989974976,
      "learning_rate": 1.67e-06,
      "loss": 0.5347,
      "step": 336
    },
    {
      "epoch": 0.018871094187479003,
      "grad_norm": 1.8292855024337769,
      "learning_rate": 1.6750000000000003e-06,
      "loss": 0.5791,
      "step": 337
    },
    {
      "epoch": 0.018927091499608018,
      "grad_norm": 1.2971700429916382,
      "learning_rate": 1.68e-06,
      "loss": 0.7015,
      "step": 338
    },
    {
      "epoch": 0.018983088811737037,
      "grad_norm": 1.2348295450210571,
      "learning_rate": 1.6850000000000002e-06,
      "loss": 0.3874,
      "step": 339
    },
    {
      "epoch": 0.019039086123866053,
      "grad_norm": 1.1175373792648315,
      "learning_rate": 1.69e-06,
      "loss": 0.3748,
      "step": 340
    },
    {
      "epoch": 0.019095083435995072,
      "grad_norm": 1.308707594871521,
      "learning_rate": 1.695e-06,
      "loss": 0.6144,
      "step": 341
    },
    {
      "epoch": 0.01915108074812409,
      "grad_norm": 1.1951675415039062,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 0.4266,
      "step": 342
    },
    {
      "epoch": 0.019207078060253107,
      "grad_norm": 1.2016880512237549,
      "learning_rate": 1.705e-06,
      "loss": 0.457,
      "step": 343
    },
    {
      "epoch": 0.019263075372382126,
      "grad_norm": 1.9292396306991577,
      "learning_rate": 1.7100000000000001e-06,
      "loss": 0.508,
      "step": 344
    },
    {
      "epoch": 0.019319072684511142,
      "grad_norm": 1.3462649583816528,
      "learning_rate": 1.7149999999999999e-06,
      "loss": 0.5321,
      "step": 345
    },
    {
      "epoch": 0.01937506999664016,
      "grad_norm": 2.1647658348083496,
      "learning_rate": 1.72e-06,
      "loss": 0.6202,
      "step": 346
    },
    {
      "epoch": 0.01943106730876918,
      "grad_norm": 1.140655755996704,
      "learning_rate": 1.7250000000000002e-06,
      "loss": 0.486,
      "step": 347
    },
    {
      "epoch": 0.019487064620898196,
      "grad_norm": 1.2482906579971313,
      "learning_rate": 1.73e-06,
      "loss": 0.446,
      "step": 348
    },
    {
      "epoch": 0.019543061933027215,
      "grad_norm": 1.4118690490722656,
      "learning_rate": 1.7350000000000001e-06,
      "loss": 0.7418,
      "step": 349
    },
    {
      "epoch": 0.019599059245156234,
      "grad_norm": 1.7058286666870117,
      "learning_rate": 1.7399999999999999e-06,
      "loss": 0.4212,
      "step": 350
    },
    {
      "epoch": 0.01965505655728525,
      "grad_norm": 1.3479478359222412,
      "learning_rate": 1.745e-06,
      "loss": 0.4631,
      "step": 351
    },
    {
      "epoch": 0.01971105386941427,
      "grad_norm": 1.2280378341674805,
      "learning_rate": 1.7500000000000002e-06,
      "loss": 0.6453,
      "step": 352
    },
    {
      "epoch": 0.019767051181543285,
      "grad_norm": 1.2790160179138184,
      "learning_rate": 1.7550000000000001e-06,
      "loss": 0.4572,
      "step": 353
    },
    {
      "epoch": 0.019823048493672304,
      "grad_norm": 1.3132696151733398,
      "learning_rate": 1.76e-06,
      "loss": 0.612,
      "step": 354
    },
    {
      "epoch": 0.019879045805801323,
      "grad_norm": 1.4868576526641846,
      "learning_rate": 1.765e-06,
      "loss": 0.5342,
      "step": 355
    },
    {
      "epoch": 0.01993504311793034,
      "grad_norm": 1.3354017734527588,
      "learning_rate": 1.7700000000000002e-06,
      "loss": 0.5269,
      "step": 356
    },
    {
      "epoch": 0.019991040430059358,
      "grad_norm": 1.2551268339157104,
      "learning_rate": 1.775e-06,
      "loss": 0.4749,
      "step": 357
    },
    {
      "epoch": 0.020047037742188373,
      "grad_norm": 1.1117607355117798,
      "learning_rate": 1.7800000000000001e-06,
      "loss": 0.3936,
      "step": 358
    },
    {
      "epoch": 0.020103035054317393,
      "grad_norm": 1.2144349813461304,
      "learning_rate": 1.7850000000000003e-06,
      "loss": 0.5228,
      "step": 359
    },
    {
      "epoch": 0.020159032366446412,
      "grad_norm": 1.3747631311416626,
      "learning_rate": 1.79e-06,
      "loss": 0.3782,
      "step": 360
    },
    {
      "epoch": 0.020215029678575427,
      "grad_norm": 1.2662698030471802,
      "learning_rate": 1.7950000000000002e-06,
      "loss": 0.4512,
      "step": 361
    },
    {
      "epoch": 0.020271026990704447,
      "grad_norm": 1.4247366189956665,
      "learning_rate": 1.8e-06,
      "loss": 0.5507,
      "step": 362
    },
    {
      "epoch": 0.020327024302833462,
      "grad_norm": 1.3011784553527832,
      "learning_rate": 1.805e-06,
      "loss": 0.4161,
      "step": 363
    },
    {
      "epoch": 0.02038302161496248,
      "grad_norm": 1.6431331634521484,
      "learning_rate": 1.8100000000000002e-06,
      "loss": 0.4019,
      "step": 364
    },
    {
      "epoch": 0.0204390189270915,
      "grad_norm": 1.3516236543655396,
      "learning_rate": 1.815e-06,
      "loss": 0.5224,
      "step": 365
    },
    {
      "epoch": 0.020495016239220516,
      "grad_norm": 1.4772320985794067,
      "learning_rate": 1.8200000000000002e-06,
      "loss": 0.4568,
      "step": 366
    },
    {
      "epoch": 0.020551013551349535,
      "grad_norm": 1.2806977033615112,
      "learning_rate": 1.8249999999999999e-06,
      "loss": 0.3767,
      "step": 367
    },
    {
      "epoch": 0.020607010863478555,
      "grad_norm": 1.4820971488952637,
      "learning_rate": 1.83e-06,
      "loss": 0.5307,
      "step": 368
    },
    {
      "epoch": 0.02066300817560757,
      "grad_norm": 1.237572193145752,
      "learning_rate": 1.8350000000000002e-06,
      "loss": 0.4643,
      "step": 369
    },
    {
      "epoch": 0.02071900548773659,
      "grad_norm": 1.4231642484664917,
      "learning_rate": 1.84e-06,
      "loss": 0.5078,
      "step": 370
    },
    {
      "epoch": 0.020775002799865605,
      "grad_norm": 1.2499034404754639,
      "learning_rate": 1.8450000000000001e-06,
      "loss": 0.5165,
      "step": 371
    },
    {
      "epoch": 0.020831000111994624,
      "grad_norm": 1.2820899486541748,
      "learning_rate": 1.85e-06,
      "loss": 0.5586,
      "step": 372
    },
    {
      "epoch": 0.020886997424123643,
      "grad_norm": 1.3260475397109985,
      "learning_rate": 1.8550000000000002e-06,
      "loss": 0.3488,
      "step": 373
    },
    {
      "epoch": 0.02094299473625266,
      "grad_norm": 2.5342912673950195,
      "learning_rate": 1.86e-06,
      "loss": 0.7005,
      "step": 374
    },
    {
      "epoch": 0.020998992048381678,
      "grad_norm": 1.2824656963348389,
      "learning_rate": 1.8650000000000001e-06,
      "loss": 0.4665,
      "step": 375
    },
    {
      "epoch": 0.021054989360510694,
      "grad_norm": 1.4731974601745605,
      "learning_rate": 1.8700000000000003e-06,
      "loss": 0.5446,
      "step": 376
    },
    {
      "epoch": 0.021110986672639713,
      "grad_norm": 1.3010905981063843,
      "learning_rate": 1.875e-06,
      "loss": 0.5301,
      "step": 377
    },
    {
      "epoch": 0.021166983984768732,
      "grad_norm": 1.2400481700897217,
      "learning_rate": 1.8800000000000002e-06,
      "loss": 0.6083,
      "step": 378
    },
    {
      "epoch": 0.021222981296897748,
      "grad_norm": 1.2115167379379272,
      "learning_rate": 1.885e-06,
      "loss": 0.4465,
      "step": 379
    },
    {
      "epoch": 0.021278978609026767,
      "grad_norm": 1.2475694417953491,
      "learning_rate": 1.8900000000000001e-06,
      "loss": 0.5166,
      "step": 380
    },
    {
      "epoch": 0.021334975921155786,
      "grad_norm": 1.2450370788574219,
      "learning_rate": 1.8950000000000003e-06,
      "loss": 0.4768,
      "step": 381
    },
    {
      "epoch": 0.021390973233284802,
      "grad_norm": 1.3486833572387695,
      "learning_rate": 1.9e-06,
      "loss": 0.63,
      "step": 382
    },
    {
      "epoch": 0.02144697054541382,
      "grad_norm": 1.5615993738174438,
      "learning_rate": 1.9050000000000002e-06,
      "loss": 0.5499,
      "step": 383
    },
    {
      "epoch": 0.021502967857542837,
      "grad_norm": 1.0564674139022827,
      "learning_rate": 1.91e-06,
      "loss": 0.4291,
      "step": 384
    },
    {
      "epoch": 0.021558965169671856,
      "grad_norm": 1.1341657638549805,
      "learning_rate": 1.9150000000000003e-06,
      "loss": 0.4337,
      "step": 385
    },
    {
      "epoch": 0.021614962481800875,
      "grad_norm": 1.2850117683410645,
      "learning_rate": 1.92e-06,
      "loss": 0.3848,
      "step": 386
    },
    {
      "epoch": 0.02167095979392989,
      "grad_norm": 1.4669718742370605,
      "learning_rate": 1.925e-06,
      "loss": 0.4459,
      "step": 387
    },
    {
      "epoch": 0.02172695710605891,
      "grad_norm": 1.3067212104797363,
      "learning_rate": 1.93e-06,
      "loss": 0.4179,
      "step": 388
    },
    {
      "epoch": 0.021782954418187925,
      "grad_norm": 1.338586688041687,
      "learning_rate": 1.935e-06,
      "loss": 0.468,
      "step": 389
    },
    {
      "epoch": 0.021838951730316945,
      "grad_norm": 1.3039277791976929,
      "learning_rate": 1.94e-06,
      "loss": 0.4632,
      "step": 390
    },
    {
      "epoch": 0.021894949042445964,
      "grad_norm": 1.2103965282440186,
      "learning_rate": 1.945e-06,
      "loss": 0.4425,
      "step": 391
    },
    {
      "epoch": 0.02195094635457498,
      "grad_norm": 1.1756149530410767,
      "learning_rate": 1.95e-06,
      "loss": 0.5389,
      "step": 392
    },
    {
      "epoch": 0.022006943666704,
      "grad_norm": 1.6295759677886963,
      "learning_rate": 1.9550000000000003e-06,
      "loss": 0.5089,
      "step": 393
    },
    {
      "epoch": 0.022062940978833018,
      "grad_norm": 1.2471020221710205,
      "learning_rate": 1.96e-06,
      "loss": 0.4591,
      "step": 394
    },
    {
      "epoch": 0.022118938290962033,
      "grad_norm": 1.276649832725525,
      "learning_rate": 1.9650000000000002e-06,
      "loss": 0.4814,
      "step": 395
    },
    {
      "epoch": 0.022174935603091053,
      "grad_norm": 1.3960410356521606,
      "learning_rate": 1.9699999999999998e-06,
      "loss": 0.3976,
      "step": 396
    },
    {
      "epoch": 0.022230932915220068,
      "grad_norm": 1.5828909873962402,
      "learning_rate": 1.975e-06,
      "loss": 0.4629,
      "step": 397
    },
    {
      "epoch": 0.022286930227349087,
      "grad_norm": 1.1500638723373413,
      "learning_rate": 1.98e-06,
      "loss": 0.4718,
      "step": 398
    },
    {
      "epoch": 0.022342927539478107,
      "grad_norm": 1.217605471611023,
      "learning_rate": 1.985e-06,
      "loss": 0.5962,
      "step": 399
    },
    {
      "epoch": 0.022398924851607122,
      "grad_norm": 1.4575655460357666,
      "learning_rate": 1.99e-06,
      "loss": 0.5309,
      "step": 400
    },
    {
      "epoch": 0.02245492216373614,
      "grad_norm": 1.4009506702423096,
      "learning_rate": 1.995e-06,
      "loss": 0.5567,
      "step": 401
    },
    {
      "epoch": 0.022510919475865157,
      "grad_norm": 1.2377105951309204,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.4204,
      "step": 402
    },
    {
      "epoch": 0.022566916787994176,
      "grad_norm": 1.214349389076233,
      "learning_rate": 2.005e-06,
      "loss": 0.5886,
      "step": 403
    },
    {
      "epoch": 0.022622914100123195,
      "grad_norm": 1.275333046913147,
      "learning_rate": 2.0100000000000002e-06,
      "loss": 0.4612,
      "step": 404
    },
    {
      "epoch": 0.02267891141225221,
      "grad_norm": 1.3972492218017578,
      "learning_rate": 2.015e-06,
      "loss": 0.6052,
      "step": 405
    },
    {
      "epoch": 0.02273490872438123,
      "grad_norm": 1.0658378601074219,
      "learning_rate": 2.02e-06,
      "loss": 0.3883,
      "step": 406
    },
    {
      "epoch": 0.022790906036510246,
      "grad_norm": 1.2928189039230347,
      "learning_rate": 2.025e-06,
      "loss": 0.6167,
      "step": 407
    },
    {
      "epoch": 0.022846903348639265,
      "grad_norm": 1.3358932733535767,
      "learning_rate": 2.03e-06,
      "loss": 0.3844,
      "step": 408
    },
    {
      "epoch": 0.022902900660768284,
      "grad_norm": 1.479500412940979,
      "learning_rate": 2.035e-06,
      "loss": 0.4249,
      "step": 409
    },
    {
      "epoch": 0.0229588979728973,
      "grad_norm": 1.2579445838928223,
      "learning_rate": 2.0400000000000004e-06,
      "loss": 0.4825,
      "step": 410
    },
    {
      "epoch": 0.02301489528502632,
      "grad_norm": 1.2056101560592651,
      "learning_rate": 2.045e-06,
      "loss": 0.4273,
      "step": 411
    },
    {
      "epoch": 0.023070892597155338,
      "grad_norm": 1.3829147815704346,
      "learning_rate": 2.0500000000000003e-06,
      "loss": 0.6324,
      "step": 412
    },
    {
      "epoch": 0.023126889909284354,
      "grad_norm": 1.4959462881088257,
      "learning_rate": 2.055e-06,
      "loss": 0.5117,
      "step": 413
    },
    {
      "epoch": 0.023182887221413373,
      "grad_norm": 1.593402624130249,
      "learning_rate": 2.06e-06,
      "loss": 0.573,
      "step": 414
    },
    {
      "epoch": 0.02323888453354239,
      "grad_norm": 1.2119688987731934,
      "learning_rate": 2.065e-06,
      "loss": 0.4263,
      "step": 415
    },
    {
      "epoch": 0.023294881845671408,
      "grad_norm": 1.2972121238708496,
      "learning_rate": 2.07e-06,
      "loss": 0.4812,
      "step": 416
    },
    {
      "epoch": 0.023350879157800427,
      "grad_norm": 1.3286548852920532,
      "learning_rate": 2.075e-06,
      "loss": 0.4082,
      "step": 417
    },
    {
      "epoch": 0.023406876469929443,
      "grad_norm": 1.6062296628952026,
      "learning_rate": 2.08e-06,
      "loss": 0.5704,
      "step": 418
    },
    {
      "epoch": 0.023462873782058462,
      "grad_norm": 1.2025117874145508,
      "learning_rate": 2.085e-06,
      "loss": 0.4505,
      "step": 419
    },
    {
      "epoch": 0.023518871094187477,
      "grad_norm": 1.1478592157363892,
      "learning_rate": 2.09e-06,
      "loss": 0.4226,
      "step": 420
    },
    {
      "epoch": 0.023574868406316497,
      "grad_norm": 1.5359517335891724,
      "learning_rate": 2.0950000000000003e-06,
      "loss": 0.5153,
      "step": 421
    },
    {
      "epoch": 0.023630865718445516,
      "grad_norm": 1.400523066520691,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.5782,
      "step": 422
    },
    {
      "epoch": 0.02368686303057453,
      "grad_norm": 1.2018365859985352,
      "learning_rate": 2.105e-06,
      "loss": 0.4398,
      "step": 423
    },
    {
      "epoch": 0.02374286034270355,
      "grad_norm": 1.286759853363037,
      "learning_rate": 2.11e-06,
      "loss": 0.4881,
      "step": 424
    },
    {
      "epoch": 0.02379885765483257,
      "grad_norm": 1.2115166187286377,
      "learning_rate": 2.115e-06,
      "loss": 0.5765,
      "step": 425
    },
    {
      "epoch": 0.023854854966961585,
      "grad_norm": 1.3774797916412354,
      "learning_rate": 2.12e-06,
      "loss": 0.4356,
      "step": 426
    },
    {
      "epoch": 0.023910852279090605,
      "grad_norm": 1.211519479751587,
      "learning_rate": 2.1250000000000004e-06,
      "loss": 0.5455,
      "step": 427
    },
    {
      "epoch": 0.02396684959121962,
      "grad_norm": 1.1984440088272095,
      "learning_rate": 2.13e-06,
      "loss": 0.4587,
      "step": 428
    },
    {
      "epoch": 0.02402284690334864,
      "grad_norm": 1.484640121459961,
      "learning_rate": 2.1350000000000003e-06,
      "loss": 0.4316,
      "step": 429
    },
    {
      "epoch": 0.02407884421547766,
      "grad_norm": 1.4276765584945679,
      "learning_rate": 2.14e-06,
      "loss": 0.5961,
      "step": 430
    },
    {
      "epoch": 0.024134841527606674,
      "grad_norm": 0.9489923715591431,
      "learning_rate": 2.1450000000000002e-06,
      "loss": 0.3301,
      "step": 431
    },
    {
      "epoch": 0.024190838839735693,
      "grad_norm": 1.2276958227157593,
      "learning_rate": 2.1499999999999997e-06,
      "loss": 0.3866,
      "step": 432
    },
    {
      "epoch": 0.02424683615186471,
      "grad_norm": 1.308908462524414,
      "learning_rate": 2.155e-06,
      "loss": 0.4123,
      "step": 433
    },
    {
      "epoch": 0.024302833463993728,
      "grad_norm": 1.2939643859863281,
      "learning_rate": 2.16e-06,
      "loss": 0.4278,
      "step": 434
    },
    {
      "epoch": 0.024358830776122747,
      "grad_norm": 1.3001083135604858,
      "learning_rate": 2.165e-06,
      "loss": 0.657,
      "step": 435
    },
    {
      "epoch": 0.024414828088251763,
      "grad_norm": 1.6597774028778076,
      "learning_rate": 2.17e-06,
      "loss": 0.6464,
      "step": 436
    },
    {
      "epoch": 0.024470825400380782,
      "grad_norm": 1.2446776628494263,
      "learning_rate": 2.175e-06,
      "loss": 0.4831,
      "step": 437
    },
    {
      "epoch": 0.024526822712509798,
      "grad_norm": 0.962810218334198,
      "learning_rate": 2.1800000000000003e-06,
      "loss": 0.3997,
      "step": 438
    },
    {
      "epoch": 0.024582820024638817,
      "grad_norm": 1.5243562459945679,
      "learning_rate": 2.1850000000000003e-06,
      "loss": 0.3864,
      "step": 439
    },
    {
      "epoch": 0.024638817336767836,
      "grad_norm": 1.2688652276992798,
      "learning_rate": 2.19e-06,
      "loss": 0.4964,
      "step": 440
    },
    {
      "epoch": 0.024694814648896852,
      "grad_norm": 1.4606717824935913,
      "learning_rate": 2.195e-06,
      "loss": 0.6258,
      "step": 441
    },
    {
      "epoch": 0.02475081196102587,
      "grad_norm": 1.2517849206924438,
      "learning_rate": 2.2e-06,
      "loss": 0.4499,
      "step": 442
    },
    {
      "epoch": 0.02480680927315489,
      "grad_norm": 1.210240125656128,
      "learning_rate": 2.205e-06,
      "loss": 0.5199,
      "step": 443
    },
    {
      "epoch": 0.024862806585283906,
      "grad_norm": 1.3706921339035034,
      "learning_rate": 2.2100000000000004e-06,
      "loss": 0.4387,
      "step": 444
    },
    {
      "epoch": 0.024918803897412925,
      "grad_norm": 1.144190788269043,
      "learning_rate": 2.215e-06,
      "loss": 0.3526,
      "step": 445
    },
    {
      "epoch": 0.02497480120954194,
      "grad_norm": 1.22667396068573,
      "learning_rate": 2.2200000000000003e-06,
      "loss": 0.5217,
      "step": 446
    },
    {
      "epoch": 0.02503079852167096,
      "grad_norm": 1.5109798908233643,
      "learning_rate": 2.225e-06,
      "loss": 0.4878,
      "step": 447
    },
    {
      "epoch": 0.02508679583379998,
      "grad_norm": 1.2600774765014648,
      "learning_rate": 2.2300000000000002e-06,
      "loss": 0.4519,
      "step": 448
    },
    {
      "epoch": 0.025142793145928995,
      "grad_norm": 1.29214346408844,
      "learning_rate": 2.2349999999999998e-06,
      "loss": 0.459,
      "step": 449
    },
    {
      "epoch": 0.025198790458058014,
      "grad_norm": 1.1631250381469727,
      "learning_rate": 2.24e-06,
      "loss": 0.4691,
      "step": 450
    },
    {
      "epoch": 0.02525478777018703,
      "grad_norm": 1.379044771194458,
      "learning_rate": 2.245e-06,
      "loss": 0.524,
      "step": 451
    },
    {
      "epoch": 0.02531078508231605,
      "grad_norm": 1.169918417930603,
      "learning_rate": 2.25e-06,
      "loss": 0.4829,
      "step": 452
    },
    {
      "epoch": 0.025366782394445068,
      "grad_norm": 1.133783221244812,
      "learning_rate": 2.255e-06,
      "loss": 0.536,
      "step": 453
    },
    {
      "epoch": 0.025422779706574083,
      "grad_norm": 1.431279182434082,
      "learning_rate": 2.26e-06,
      "loss": 0.5463,
      "step": 454
    },
    {
      "epoch": 0.025478777018703103,
      "grad_norm": 1.282882809638977,
      "learning_rate": 2.265e-06,
      "loss": 0.4981,
      "step": 455
    },
    {
      "epoch": 0.025534774330832122,
      "grad_norm": 1.1804708242416382,
      "learning_rate": 2.2700000000000003e-06,
      "loss": 0.419,
      "step": 456
    },
    {
      "epoch": 0.025590771642961137,
      "grad_norm": 1.298541784286499,
      "learning_rate": 2.2750000000000002e-06,
      "loss": 0.4705,
      "step": 457
    },
    {
      "epoch": 0.025646768955090157,
      "grad_norm": 1.1998291015625,
      "learning_rate": 2.28e-06,
      "loss": 0.3948,
      "step": 458
    },
    {
      "epoch": 0.025702766267219172,
      "grad_norm": 1.3842189311981201,
      "learning_rate": 2.285e-06,
      "loss": 0.4904,
      "step": 459
    },
    {
      "epoch": 0.02575876357934819,
      "grad_norm": 1.264269232749939,
      "learning_rate": 2.29e-06,
      "loss": 0.4574,
      "step": 460
    },
    {
      "epoch": 0.02581476089147721,
      "grad_norm": 1.2546792030334473,
      "learning_rate": 2.2950000000000005e-06,
      "loss": 0.3574,
      "step": 461
    },
    {
      "epoch": 0.025870758203606226,
      "grad_norm": 1.0641919374465942,
      "learning_rate": 2.3e-06,
      "loss": 0.5158,
      "step": 462
    },
    {
      "epoch": 0.025926755515735245,
      "grad_norm": 1.35035240650177,
      "learning_rate": 2.3050000000000004e-06,
      "loss": 0.5019,
      "step": 463
    },
    {
      "epoch": 0.02598275282786426,
      "grad_norm": 1.248283863067627,
      "learning_rate": 2.31e-06,
      "loss": 0.4133,
      "step": 464
    },
    {
      "epoch": 0.02603875013999328,
      "grad_norm": 1.6036901473999023,
      "learning_rate": 2.3150000000000003e-06,
      "loss": 0.6371,
      "step": 465
    },
    {
      "epoch": 0.0260947474521223,
      "grad_norm": 1.4229565858840942,
      "learning_rate": 2.32e-06,
      "loss": 0.4239,
      "step": 466
    },
    {
      "epoch": 0.026150744764251315,
      "grad_norm": 1.245241403579712,
      "learning_rate": 2.325e-06,
      "loss": 0.4052,
      "step": 467
    },
    {
      "epoch": 0.026206742076380334,
      "grad_norm": 1.4726648330688477,
      "learning_rate": 2.33e-06,
      "loss": 0.5263,
      "step": 468
    },
    {
      "epoch": 0.02626273938850935,
      "grad_norm": 1.1756625175476074,
      "learning_rate": 2.335e-06,
      "loss": 0.4515,
      "step": 469
    },
    {
      "epoch": 0.02631873670063837,
      "grad_norm": 1.3073421716690063,
      "learning_rate": 2.34e-06,
      "loss": 0.6833,
      "step": 470
    },
    {
      "epoch": 0.026374734012767388,
      "grad_norm": 1.1485313177108765,
      "learning_rate": 2.345e-06,
      "loss": 0.3888,
      "step": 471
    },
    {
      "epoch": 0.026430731324896404,
      "grad_norm": 1.120802402496338,
      "learning_rate": 2.35e-06,
      "loss": 0.3704,
      "step": 472
    },
    {
      "epoch": 0.026486728637025423,
      "grad_norm": 1.3014438152313232,
      "learning_rate": 2.3550000000000003e-06,
      "loss": 0.6238,
      "step": 473
    },
    {
      "epoch": 0.026542725949154442,
      "grad_norm": 1.2936598062515259,
      "learning_rate": 2.36e-06,
      "loss": 0.5,
      "step": 474
    },
    {
      "epoch": 0.026598723261283458,
      "grad_norm": 1.2880456447601318,
      "learning_rate": 2.3650000000000002e-06,
      "loss": 0.41,
      "step": 475
    },
    {
      "epoch": 0.026654720573412477,
      "grad_norm": 1.2736504077911377,
      "learning_rate": 2.37e-06,
      "loss": 0.4159,
      "step": 476
    },
    {
      "epoch": 0.026710717885541493,
      "grad_norm": 1.3126493692398071,
      "learning_rate": 2.375e-06,
      "loss": 0.3981,
      "step": 477
    },
    {
      "epoch": 0.026766715197670512,
      "grad_norm": 1.1644043922424316,
      "learning_rate": 2.38e-06,
      "loss": 0.4143,
      "step": 478
    },
    {
      "epoch": 0.02682271250979953,
      "grad_norm": 1.2235571146011353,
      "learning_rate": 2.385e-06,
      "loss": 0.4354,
      "step": 479
    },
    {
      "epoch": 0.026878709821928547,
      "grad_norm": 1.503503680229187,
      "learning_rate": 2.3900000000000004e-06,
      "loss": 0.4693,
      "step": 480
    },
    {
      "epoch": 0.026934707134057566,
      "grad_norm": 1.1442761421203613,
      "learning_rate": 2.395e-06,
      "loss": 0.4132,
      "step": 481
    },
    {
      "epoch": 0.02699070444618658,
      "grad_norm": 1.382543921470642,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.4063,
      "step": 482
    },
    {
      "epoch": 0.0270467017583156,
      "grad_norm": 1.2548270225524902,
      "learning_rate": 2.405e-06,
      "loss": 0.3863,
      "step": 483
    },
    {
      "epoch": 0.02710269907044462,
      "grad_norm": 1.2861868143081665,
      "learning_rate": 2.4100000000000002e-06,
      "loss": 0.4733,
      "step": 484
    },
    {
      "epoch": 0.027158696382573635,
      "grad_norm": 1.3512135744094849,
      "learning_rate": 2.415e-06,
      "loss": 0.4445,
      "step": 485
    },
    {
      "epoch": 0.027214693694702655,
      "grad_norm": 16.69654655456543,
      "learning_rate": 2.42e-06,
      "loss": 0.5016,
      "step": 486
    },
    {
      "epoch": 0.027270691006831674,
      "grad_norm": 1.4881073236465454,
      "learning_rate": 2.425e-06,
      "loss": 0.5744,
      "step": 487
    },
    {
      "epoch": 0.02732668831896069,
      "grad_norm": 1.1898785829544067,
      "learning_rate": 2.43e-06,
      "loss": 0.526,
      "step": 488
    },
    {
      "epoch": 0.02738268563108971,
      "grad_norm": 1.2967259883880615,
      "learning_rate": 2.435e-06,
      "loss": 0.5289,
      "step": 489
    },
    {
      "epoch": 0.027438682943218724,
      "grad_norm": 1.49234139919281,
      "learning_rate": 2.4400000000000004e-06,
      "loss": 0.6374,
      "step": 490
    },
    {
      "epoch": 0.027494680255347743,
      "grad_norm": 1.1433918476104736,
      "learning_rate": 2.445e-06,
      "loss": 0.3643,
      "step": 491
    },
    {
      "epoch": 0.027550677567476763,
      "grad_norm": 1.375954508781433,
      "learning_rate": 2.4500000000000003e-06,
      "loss": 0.655,
      "step": 492
    },
    {
      "epoch": 0.02760667487960578,
      "grad_norm": 1.0267375707626343,
      "learning_rate": 2.4550000000000002e-06,
      "loss": 0.5066,
      "step": 493
    },
    {
      "epoch": 0.027662672191734797,
      "grad_norm": 1.6544849872589111,
      "learning_rate": 2.46e-06,
      "loss": 0.4229,
      "step": 494
    },
    {
      "epoch": 0.027718669503863813,
      "grad_norm": 1.0966876745224,
      "learning_rate": 2.465e-06,
      "loss": 0.446,
      "step": 495
    },
    {
      "epoch": 0.027774666815992832,
      "grad_norm": 1.3126972913742065,
      "learning_rate": 2.47e-06,
      "loss": 0.4629,
      "step": 496
    },
    {
      "epoch": 0.02783066412812185,
      "grad_norm": 1.245058298110962,
      "learning_rate": 2.4750000000000004e-06,
      "loss": 0.618,
      "step": 497
    },
    {
      "epoch": 0.027886661440250867,
      "grad_norm": 1.700477957725525,
      "learning_rate": 2.48e-06,
      "loss": 0.5081,
      "step": 498
    },
    {
      "epoch": 0.027942658752379886,
      "grad_norm": 1.3252248764038086,
      "learning_rate": 2.4850000000000003e-06,
      "loss": 0.4537,
      "step": 499
    },
    {
      "epoch": 0.027998656064508902,
      "grad_norm": 1.6630816459655762,
      "learning_rate": 2.49e-06,
      "loss": 0.3542,
      "step": 500
    },
    {
      "epoch": 0.02805465337663792,
      "grad_norm": 1.546900749206543,
      "learning_rate": 2.4950000000000003e-06,
      "loss": 0.6161,
      "step": 501
    },
    {
      "epoch": 0.02811065068876694,
      "grad_norm": 1.234569787979126,
      "learning_rate": 2.5e-06,
      "loss": 0.436,
      "step": 502
    },
    {
      "epoch": 0.028166648000895956,
      "grad_norm": 1.4984447956085205,
      "learning_rate": 2.505e-06,
      "loss": 0.4401,
      "step": 503
    },
    {
      "epoch": 0.028222645313024975,
      "grad_norm": 1.3634111881256104,
      "learning_rate": 2.51e-06,
      "loss": 0.4091,
      "step": 504
    },
    {
      "epoch": 0.028278642625153994,
      "grad_norm": 1.5010645389556885,
      "learning_rate": 2.515e-06,
      "loss": 0.4773,
      "step": 505
    },
    {
      "epoch": 0.02833463993728301,
      "grad_norm": 1.1970478296279907,
      "learning_rate": 2.52e-06,
      "loss": 0.4239,
      "step": 506
    },
    {
      "epoch": 0.02839063724941203,
      "grad_norm": 1.2129427194595337,
      "learning_rate": 2.5250000000000004e-06,
      "loss": 0.444,
      "step": 507
    },
    {
      "epoch": 0.028446634561541045,
      "grad_norm": 1.586907148361206,
      "learning_rate": 2.53e-06,
      "loss": 0.5241,
      "step": 508
    },
    {
      "epoch": 0.028502631873670064,
      "grad_norm": 1.338651418685913,
      "learning_rate": 2.5350000000000003e-06,
      "loss": 0.5268,
      "step": 509
    },
    {
      "epoch": 0.028558629185799083,
      "grad_norm": 1.1952580213546753,
      "learning_rate": 2.54e-06,
      "loss": 0.4454,
      "step": 510
    },
    {
      "epoch": 0.0286146264979281,
      "grad_norm": 1.2241628170013428,
      "learning_rate": 2.545e-06,
      "loss": 0.4019,
      "step": 511
    },
    {
      "epoch": 0.028670623810057118,
      "grad_norm": 1.2646491527557373,
      "learning_rate": 2.55e-06,
      "loss": 0.5136,
      "step": 512
    },
    {
      "epoch": 0.028726621122186134,
      "grad_norm": 1.3502599000930786,
      "learning_rate": 2.555e-06,
      "loss": 0.557,
      "step": 513
    },
    {
      "epoch": 0.028782618434315153,
      "grad_norm": 1.107664704322815,
      "learning_rate": 2.56e-06,
      "loss": 0.3778,
      "step": 514
    },
    {
      "epoch": 0.028838615746444172,
      "grad_norm": 1.457395076751709,
      "learning_rate": 2.565e-06,
      "loss": 0.7625,
      "step": 515
    },
    {
      "epoch": 0.028894613058573188,
      "grad_norm": 1.2146413326263428,
      "learning_rate": 2.5700000000000004e-06,
      "loss": 0.4772,
      "step": 516
    },
    {
      "epoch": 0.028950610370702207,
      "grad_norm": 1.2828574180603027,
      "learning_rate": 2.575e-06,
      "loss": 0.4514,
      "step": 517
    },
    {
      "epoch": 0.029006607682831226,
      "grad_norm": 1.3281506299972534,
      "learning_rate": 2.5800000000000003e-06,
      "loss": 0.5961,
      "step": 518
    },
    {
      "epoch": 0.02906260499496024,
      "grad_norm": 1.1464653015136719,
      "learning_rate": 2.5850000000000002e-06,
      "loss": 0.3494,
      "step": 519
    },
    {
      "epoch": 0.02911860230708926,
      "grad_norm": 1.368131399154663,
      "learning_rate": 2.59e-06,
      "loss": 0.5817,
      "step": 520
    },
    {
      "epoch": 0.029174599619218276,
      "grad_norm": 1.3296947479248047,
      "learning_rate": 2.595e-06,
      "loss": 0.615,
      "step": 521
    },
    {
      "epoch": 0.029230596931347295,
      "grad_norm": 1.1147074699401855,
      "learning_rate": 2.6e-06,
      "loss": 0.437,
      "step": 522
    },
    {
      "epoch": 0.029286594243476315,
      "grad_norm": 1.301010012626648,
      "learning_rate": 2.605e-06,
      "loss": 0.4163,
      "step": 523
    },
    {
      "epoch": 0.02934259155560533,
      "grad_norm": 1.564185380935669,
      "learning_rate": 2.6100000000000004e-06,
      "loss": 0.6379,
      "step": 524
    },
    {
      "epoch": 0.02939858886773435,
      "grad_norm": 1.2892937660217285,
      "learning_rate": 2.615e-06,
      "loss": 0.5075,
      "step": 525
    },
    {
      "epoch": 0.029454586179863365,
      "grad_norm": 1.5146722793579102,
      "learning_rate": 2.6200000000000003e-06,
      "loss": 0.5486,
      "step": 526
    },
    {
      "epoch": 0.029510583491992384,
      "grad_norm": 1.3844255208969116,
      "learning_rate": 2.625e-06,
      "loss": 0.5322,
      "step": 527
    },
    {
      "epoch": 0.029566580804121403,
      "grad_norm": 1.4252188205718994,
      "learning_rate": 2.6300000000000002e-06,
      "loss": 0.696,
      "step": 528
    },
    {
      "epoch": 0.02962257811625042,
      "grad_norm": 1.1514140367507935,
      "learning_rate": 2.6349999999999998e-06,
      "loss": 0.3972,
      "step": 529
    },
    {
      "epoch": 0.02967857542837944,
      "grad_norm": 1.384243369102478,
      "learning_rate": 2.64e-06,
      "loss": 0.4538,
      "step": 530
    },
    {
      "epoch": 0.029734572740508454,
      "grad_norm": 1.5517767667770386,
      "learning_rate": 2.645e-06,
      "loss": 0.7234,
      "step": 531
    },
    {
      "epoch": 0.029790570052637473,
      "grad_norm": 1.220450520515442,
      "learning_rate": 2.65e-06,
      "loss": 0.4006,
      "step": 532
    },
    {
      "epoch": 0.029846567364766492,
      "grad_norm": 1.5837041139602661,
      "learning_rate": 2.655e-06,
      "loss": 0.5086,
      "step": 533
    },
    {
      "epoch": 0.029902564676895508,
      "grad_norm": 2.34853196144104,
      "learning_rate": 2.66e-06,
      "loss": 0.5333,
      "step": 534
    },
    {
      "epoch": 0.029958561989024527,
      "grad_norm": 1.433725118637085,
      "learning_rate": 2.6650000000000003e-06,
      "loss": 0.6396,
      "step": 535
    },
    {
      "epoch": 0.030014559301153546,
      "grad_norm": 1.336133599281311,
      "learning_rate": 2.6700000000000003e-06,
      "loss": 0.5378,
      "step": 536
    },
    {
      "epoch": 0.030070556613282562,
      "grad_norm": 1.3267064094543457,
      "learning_rate": 2.6750000000000002e-06,
      "loss": 0.6231,
      "step": 537
    },
    {
      "epoch": 0.03012655392541158,
      "grad_norm": 1.4553563594818115,
      "learning_rate": 2.68e-06,
      "loss": 0.5325,
      "step": 538
    },
    {
      "epoch": 0.030182551237540597,
      "grad_norm": 1.1272870302200317,
      "learning_rate": 2.685e-06,
      "loss": 0.4525,
      "step": 539
    },
    {
      "epoch": 0.030238548549669616,
      "grad_norm": 1.8186594247817993,
      "learning_rate": 2.69e-06,
      "loss": 0.4983,
      "step": 540
    },
    {
      "epoch": 0.030294545861798635,
      "grad_norm": 1.2522361278533936,
      "learning_rate": 2.6950000000000005e-06,
      "loss": 0.5709,
      "step": 541
    },
    {
      "epoch": 0.03035054317392765,
      "grad_norm": 1.3136264085769653,
      "learning_rate": 2.7e-06,
      "loss": 0.4878,
      "step": 542
    },
    {
      "epoch": 0.03040654048605667,
      "grad_norm": 1.222240686416626,
      "learning_rate": 2.7050000000000004e-06,
      "loss": 0.4881,
      "step": 543
    },
    {
      "epoch": 0.030462537798185686,
      "grad_norm": 1.3070313930511475,
      "learning_rate": 2.71e-06,
      "loss": 0.4469,
      "step": 544
    },
    {
      "epoch": 0.030518535110314705,
      "grad_norm": 1.2443084716796875,
      "learning_rate": 2.7150000000000003e-06,
      "loss": 0.5433,
      "step": 545
    },
    {
      "epoch": 0.030574532422443724,
      "grad_norm": 1.4681586027145386,
      "learning_rate": 2.72e-06,
      "loss": 0.5622,
      "step": 546
    },
    {
      "epoch": 0.03063052973457274,
      "grad_norm": 1.0752031803131104,
      "learning_rate": 2.725e-06,
      "loss": 0.3916,
      "step": 547
    },
    {
      "epoch": 0.03068652704670176,
      "grad_norm": 1.338870644569397,
      "learning_rate": 2.73e-06,
      "loss": 0.4114,
      "step": 548
    },
    {
      "epoch": 0.030742524358830778,
      "grad_norm": 1.2860264778137207,
      "learning_rate": 2.735e-06,
      "loss": 0.4996,
      "step": 549
    },
    {
      "epoch": 0.030798521670959794,
      "grad_norm": 10.62662124633789,
      "learning_rate": 2.74e-06,
      "loss": 0.474,
      "step": 550
    },
    {
      "epoch": 0.030854518983088813,
      "grad_norm": 1.2164727449417114,
      "learning_rate": 2.745e-06,
      "loss": 0.5298,
      "step": 551
    },
    {
      "epoch": 0.03091051629521783,
      "grad_norm": 1.0920971632003784,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.4692,
      "step": 552
    },
    {
      "epoch": 0.030966513607346847,
      "grad_norm": 1.2918709516525269,
      "learning_rate": 2.7550000000000003e-06,
      "loss": 0.3756,
      "step": 553
    },
    {
      "epoch": 0.031022510919475867,
      "grad_norm": 1.6300508975982666,
      "learning_rate": 2.7600000000000003e-06,
      "loss": 0.4695,
      "step": 554
    },
    {
      "epoch": 0.031078508231604882,
      "grad_norm": 1.2380396127700806,
      "learning_rate": 2.765e-06,
      "loss": 0.5224,
      "step": 555
    },
    {
      "epoch": 0.0311345055437339,
      "grad_norm": 1.748967170715332,
      "learning_rate": 2.77e-06,
      "loss": 0.6471,
      "step": 556
    },
    {
      "epoch": 0.031190502855862917,
      "grad_norm": 1.3673230409622192,
      "learning_rate": 2.775e-06,
      "loss": 0.576,
      "step": 557
    },
    {
      "epoch": 0.031246500167991936,
      "grad_norm": 1.1652144193649292,
      "learning_rate": 2.78e-06,
      "loss": 0.4436,
      "step": 558
    },
    {
      "epoch": 0.031302497480120955,
      "grad_norm": 1.121286153793335,
      "learning_rate": 2.785e-06,
      "loss": 0.4163,
      "step": 559
    },
    {
      "epoch": 0.031358494792249975,
      "grad_norm": 1.2735199928283691,
      "learning_rate": 2.7900000000000004e-06,
      "loss": 0.4716,
      "step": 560
    },
    {
      "epoch": 0.03141449210437899,
      "grad_norm": 1.420231819152832,
      "learning_rate": 2.795e-06,
      "loss": 0.618,
      "step": 561
    },
    {
      "epoch": 0.031470489416508006,
      "grad_norm": 1.2376223802566528,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.3637,
      "step": 562
    },
    {
      "epoch": 0.031526486728637025,
      "grad_norm": 1.3908500671386719,
      "learning_rate": 2.805e-06,
      "loss": 0.6301,
      "step": 563
    },
    {
      "epoch": 0.031582484040766044,
      "grad_norm": 1.2173155546188354,
      "learning_rate": 2.81e-06,
      "loss": 0.4616,
      "step": 564
    },
    {
      "epoch": 0.03163848135289506,
      "grad_norm": 1.752326488494873,
      "learning_rate": 2.815e-06,
      "loss": 0.403,
      "step": 565
    },
    {
      "epoch": 0.031694478665024076,
      "grad_norm": 1.6780214309692383,
      "learning_rate": 2.82e-06,
      "loss": 0.5124,
      "step": 566
    },
    {
      "epoch": 0.031750475977153095,
      "grad_norm": 1.2604048252105713,
      "learning_rate": 2.825e-06,
      "loss": 0.4851,
      "step": 567
    },
    {
      "epoch": 0.031806473289282114,
      "grad_norm": 1.31491219997406,
      "learning_rate": 2.83e-06,
      "loss": 0.4711,
      "step": 568
    },
    {
      "epoch": 0.03186247060141113,
      "grad_norm": 4.062485218048096,
      "learning_rate": 2.835e-06,
      "loss": 0.5685,
      "step": 569
    },
    {
      "epoch": 0.03191846791354015,
      "grad_norm": 1.2817109823226929,
      "learning_rate": 2.8400000000000003e-06,
      "loss": 0.427,
      "step": 570
    },
    {
      "epoch": 0.031974465225669164,
      "grad_norm": 1.2257922887802124,
      "learning_rate": 2.8450000000000003e-06,
      "loss": 0.485,
      "step": 571
    },
    {
      "epoch": 0.032030462537798184,
      "grad_norm": 1.1841542720794678,
      "learning_rate": 2.8500000000000002e-06,
      "loss": 0.4409,
      "step": 572
    },
    {
      "epoch": 0.0320864598499272,
      "grad_norm": 1.4808175563812256,
      "learning_rate": 2.855e-06,
      "loss": 0.4201,
      "step": 573
    },
    {
      "epoch": 0.03214245716205622,
      "grad_norm": 1.2412053346633911,
      "learning_rate": 2.86e-06,
      "loss": 0.4371,
      "step": 574
    },
    {
      "epoch": 0.03219845447418524,
      "grad_norm": 1.184926986694336,
      "learning_rate": 2.865e-06,
      "loss": 0.5311,
      "step": 575
    },
    {
      "epoch": 0.03225445178631426,
      "grad_norm": 1.2667714357376099,
      "learning_rate": 2.87e-06,
      "loss": 0.4015,
      "step": 576
    },
    {
      "epoch": 0.03231044909844327,
      "grad_norm": 1.4195163249969482,
      "learning_rate": 2.8750000000000004e-06,
      "loss": 0.4566,
      "step": 577
    },
    {
      "epoch": 0.03236644641057229,
      "grad_norm": 1.3507108688354492,
      "learning_rate": 2.88e-06,
      "loss": 0.5713,
      "step": 578
    },
    {
      "epoch": 0.03242244372270131,
      "grad_norm": 1.4129518270492554,
      "learning_rate": 2.8850000000000003e-06,
      "loss": 0.5334,
      "step": 579
    },
    {
      "epoch": 0.03247844103483033,
      "grad_norm": 1.3056820631027222,
      "learning_rate": 2.89e-06,
      "loss": 0.4025,
      "step": 580
    },
    {
      "epoch": 0.03253443834695935,
      "grad_norm": 1.4413847923278809,
      "learning_rate": 2.8950000000000002e-06,
      "loss": 0.5258,
      "step": 581
    },
    {
      "epoch": 0.03259043565908836,
      "grad_norm": 1.270606279373169,
      "learning_rate": 2.9e-06,
      "loss": 0.4603,
      "step": 582
    },
    {
      "epoch": 0.03264643297121738,
      "grad_norm": 1.3087241649627686,
      "learning_rate": 2.905e-06,
      "loss": 0.5114,
      "step": 583
    },
    {
      "epoch": 0.0327024302833464,
      "grad_norm": 1.163053274154663,
      "learning_rate": 2.91e-06,
      "loss": 0.4819,
      "step": 584
    },
    {
      "epoch": 0.03275842759547542,
      "grad_norm": 1.5739001035690308,
      "learning_rate": 2.915e-06,
      "loss": 0.6085,
      "step": 585
    },
    {
      "epoch": 0.03281442490760444,
      "grad_norm": 1.167637825012207,
      "learning_rate": 2.92e-06,
      "loss": 0.4609,
      "step": 586
    },
    {
      "epoch": 0.03287042221973345,
      "grad_norm": 1.22612464427948,
      "learning_rate": 2.9250000000000004e-06,
      "loss": 0.5268,
      "step": 587
    },
    {
      "epoch": 0.03292641953186247,
      "grad_norm": 1.5374791622161865,
      "learning_rate": 2.93e-06,
      "loss": 0.5035,
      "step": 588
    },
    {
      "epoch": 0.03298241684399149,
      "grad_norm": 1.3516424894332886,
      "learning_rate": 2.9350000000000003e-06,
      "loss": 0.4111,
      "step": 589
    },
    {
      "epoch": 0.03303841415612051,
      "grad_norm": 1.4715403318405151,
      "learning_rate": 2.9400000000000002e-06,
      "loss": 0.371,
      "step": 590
    },
    {
      "epoch": 0.03309441146824953,
      "grad_norm": 1.4850103855133057,
      "learning_rate": 2.945e-06,
      "loss": 0.4527,
      "step": 591
    },
    {
      "epoch": 0.03315040878037854,
      "grad_norm": 1.4048868417739868,
      "learning_rate": 2.95e-06,
      "loss": 0.458,
      "step": 592
    },
    {
      "epoch": 0.03320640609250756,
      "grad_norm": 1.741652011871338,
      "learning_rate": 2.955e-06,
      "loss": 0.4839,
      "step": 593
    },
    {
      "epoch": 0.03326240340463658,
      "grad_norm": 1.2577733993530273,
      "learning_rate": 2.9600000000000005e-06,
      "loss": 0.5411,
      "step": 594
    },
    {
      "epoch": 0.033318400716765596,
      "grad_norm": 1.3856233358383179,
      "learning_rate": 2.965e-06,
      "loss": 0.5322,
      "step": 595
    },
    {
      "epoch": 0.033374398028894615,
      "grad_norm": 1.447217345237732,
      "learning_rate": 2.9700000000000004e-06,
      "loss": 0.6493,
      "step": 596
    },
    {
      "epoch": 0.03343039534102363,
      "grad_norm": 1.4229252338409424,
      "learning_rate": 2.975e-06,
      "loss": 0.3921,
      "step": 597
    },
    {
      "epoch": 0.03348639265315265,
      "grad_norm": 11.00639820098877,
      "learning_rate": 2.9800000000000003e-06,
      "loss": 0.3584,
      "step": 598
    },
    {
      "epoch": 0.033542389965281666,
      "grad_norm": 1.2174385786056519,
      "learning_rate": 2.9850000000000002e-06,
      "loss": 0.3824,
      "step": 599
    },
    {
      "epoch": 0.033598387277410685,
      "grad_norm": 1.1561325788497925,
      "learning_rate": 2.99e-06,
      "loss": 0.4915,
      "step": 600
    },
    {
      "epoch": 0.033654384589539704,
      "grad_norm": 1.4047815799713135,
      "learning_rate": 2.995e-06,
      "loss": 0.5279,
      "step": 601
    },
    {
      "epoch": 0.033710381901668716,
      "grad_norm": 4.640164852142334,
      "learning_rate": 3e-06,
      "loss": 0.4194,
      "step": 602
    },
    {
      "epoch": 0.033766379213797736,
      "grad_norm": 1.154270052909851,
      "learning_rate": 3.005e-06,
      "loss": 0.4438,
      "step": 603
    },
    {
      "epoch": 0.033822376525926755,
      "grad_norm": 1.1494734287261963,
      "learning_rate": 3.01e-06,
      "loss": 0.4764,
      "step": 604
    },
    {
      "epoch": 0.033878373838055774,
      "grad_norm": 1.1763715744018555,
      "learning_rate": 3.015e-06,
      "loss": 0.5438,
      "step": 605
    },
    {
      "epoch": 0.03393437115018479,
      "grad_norm": 1.0460625886917114,
      "learning_rate": 3.0200000000000003e-06,
      "loss": 0.4528,
      "step": 606
    },
    {
      "epoch": 0.03399036846231381,
      "grad_norm": 5.100392818450928,
      "learning_rate": 3.0250000000000003e-06,
      "loss": 0.4503,
      "step": 607
    },
    {
      "epoch": 0.034046365774442824,
      "grad_norm": 1.0152342319488525,
      "learning_rate": 3.0300000000000002e-06,
      "loss": 0.4112,
      "step": 608
    },
    {
      "epoch": 0.034102363086571844,
      "grad_norm": 1.1221469640731812,
      "learning_rate": 3.035e-06,
      "loss": 0.4216,
      "step": 609
    },
    {
      "epoch": 0.03415836039870086,
      "grad_norm": 2.1545469760894775,
      "learning_rate": 3.04e-06,
      "loss": 0.5182,
      "step": 610
    },
    {
      "epoch": 0.03421435771082988,
      "grad_norm": 1.3261489868164062,
      "learning_rate": 3.0450000000000005e-06,
      "loss": 0.7225,
      "step": 611
    },
    {
      "epoch": 0.0342703550229589,
      "grad_norm": 1.443620204925537,
      "learning_rate": 3.05e-06,
      "loss": 0.6039,
      "step": 612
    },
    {
      "epoch": 0.03432635233508791,
      "grad_norm": 1.3014883995056152,
      "learning_rate": 3.0550000000000004e-06,
      "loss": 0.5219,
      "step": 613
    },
    {
      "epoch": 0.03438234964721693,
      "grad_norm": 2.6879069805145264,
      "learning_rate": 3.06e-06,
      "loss": 0.501,
      "step": 614
    },
    {
      "epoch": 0.03443834695934595,
      "grad_norm": 1.3093355894088745,
      "learning_rate": 3.0650000000000003e-06,
      "loss": 0.4093,
      "step": 615
    },
    {
      "epoch": 0.03449434427147497,
      "grad_norm": 1.1495851278305054,
      "learning_rate": 3.0700000000000003e-06,
      "loss": 0.388,
      "step": 616
    },
    {
      "epoch": 0.03455034158360399,
      "grad_norm": 1.318150281906128,
      "learning_rate": 3.075e-06,
      "loss": 0.4241,
      "step": 617
    },
    {
      "epoch": 0.034606338895733,
      "grad_norm": 1.5912282466888428,
      "learning_rate": 3.08e-06,
      "loss": 0.5434,
      "step": 618
    },
    {
      "epoch": 0.03466233620786202,
      "grad_norm": 1.113492488861084,
      "learning_rate": 3.085e-06,
      "loss": 0.5174,
      "step": 619
    },
    {
      "epoch": 0.03471833351999104,
      "grad_norm": 1.3931667804718018,
      "learning_rate": 3.09e-06,
      "loss": 0.3684,
      "step": 620
    },
    {
      "epoch": 0.03477433083212006,
      "grad_norm": 1.3992459774017334,
      "learning_rate": 3.095e-06,
      "loss": 0.472,
      "step": 621
    },
    {
      "epoch": 0.03483032814424908,
      "grad_norm": 1.2402055263519287,
      "learning_rate": 3.1e-06,
      "loss": 0.3491,
      "step": 622
    },
    {
      "epoch": 0.03488632545637809,
      "grad_norm": 1.3450154066085815,
      "learning_rate": 3.1050000000000003e-06,
      "loss": 0.5987,
      "step": 623
    },
    {
      "epoch": 0.03494232276850711,
      "grad_norm": 1.214292287826538,
      "learning_rate": 3.11e-06,
      "loss": 0.4254,
      "step": 624
    },
    {
      "epoch": 0.03499832008063613,
      "grad_norm": 1.2098424434661865,
      "learning_rate": 3.1150000000000002e-06,
      "loss": 0.3846,
      "step": 625
    },
    {
      "epoch": 0.03505431739276515,
      "grad_norm": 1.3445996046066284,
      "learning_rate": 3.12e-06,
      "loss": 0.5421,
      "step": 626
    },
    {
      "epoch": 0.03511031470489417,
      "grad_norm": 1.2277252674102783,
      "learning_rate": 3.125e-06,
      "loss": 0.6699,
      "step": 627
    },
    {
      "epoch": 0.03516631201702318,
      "grad_norm": 1.179675817489624,
      "learning_rate": 3.13e-06,
      "loss": 0.3626,
      "step": 628
    },
    {
      "epoch": 0.0352223093291522,
      "grad_norm": 12.111026763916016,
      "learning_rate": 3.1350000000000005e-06,
      "loss": 0.3788,
      "step": 629
    },
    {
      "epoch": 0.03527830664128122,
      "grad_norm": 1.2397090196609497,
      "learning_rate": 3.14e-06,
      "loss": 0.4611,
      "step": 630
    },
    {
      "epoch": 0.03533430395341024,
      "grad_norm": 1.2845584154129028,
      "learning_rate": 3.145e-06,
      "loss": 0.4117,
      "step": 631
    },
    {
      "epoch": 0.035390301265539256,
      "grad_norm": 1.0865263938903809,
      "learning_rate": 3.1500000000000003e-06,
      "loss": 0.3984,
      "step": 632
    },
    {
      "epoch": 0.03544629857766827,
      "grad_norm": 1.6106657981872559,
      "learning_rate": 3.1550000000000003e-06,
      "loss": 0.5957,
      "step": 633
    },
    {
      "epoch": 0.03550229588979729,
      "grad_norm": 1.0239503383636475,
      "learning_rate": 3.1600000000000007e-06,
      "loss": 0.4252,
      "step": 634
    },
    {
      "epoch": 0.03555829320192631,
      "grad_norm": 1.4274102449417114,
      "learning_rate": 3.1649999999999998e-06,
      "loss": 0.5052,
      "step": 635
    },
    {
      "epoch": 0.035614290514055326,
      "grad_norm": 1.3243954181671143,
      "learning_rate": 3.17e-06,
      "loss": 0.5587,
      "step": 636
    },
    {
      "epoch": 0.035670287826184345,
      "grad_norm": 1.0345747470855713,
      "learning_rate": 3.175e-06,
      "loss": 0.3972,
      "step": 637
    },
    {
      "epoch": 0.035726285138313364,
      "grad_norm": 1.2989063262939453,
      "learning_rate": 3.1800000000000005e-06,
      "loss": 0.536,
      "step": 638
    },
    {
      "epoch": 0.035782282450442376,
      "grad_norm": 1.3336299657821655,
      "learning_rate": 3.1850000000000004e-06,
      "loss": 0.4359,
      "step": 639
    },
    {
      "epoch": 0.035838279762571396,
      "grad_norm": 1.2301009893417358,
      "learning_rate": 3.19e-06,
      "loss": 0.4369,
      "step": 640
    },
    {
      "epoch": 0.035894277074700415,
      "grad_norm": 1.0255048274993896,
      "learning_rate": 3.195e-06,
      "loss": 0.4257,
      "step": 641
    },
    {
      "epoch": 0.035950274386829434,
      "grad_norm": 1.262315273284912,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.409,
      "step": 642
    },
    {
      "epoch": 0.03600627169895845,
      "grad_norm": 1.1630877256393433,
      "learning_rate": 3.2050000000000002e-06,
      "loss": 0.3283,
      "step": 643
    },
    {
      "epoch": 0.036062269011087465,
      "grad_norm": 1.3334583044052124,
      "learning_rate": 3.2099999999999998e-06,
      "loss": 0.4425,
      "step": 644
    },
    {
      "epoch": 0.036118266323216484,
      "grad_norm": 1.649351954460144,
      "learning_rate": 3.215e-06,
      "loss": 0.5322,
      "step": 645
    },
    {
      "epoch": 0.036174263635345504,
      "grad_norm": 1.298487901687622,
      "learning_rate": 3.22e-06,
      "loss": 0.4998,
      "step": 646
    },
    {
      "epoch": 0.03623026094747452,
      "grad_norm": 1.2624866962432861,
      "learning_rate": 3.225e-06,
      "loss": 0.3944,
      "step": 647
    },
    {
      "epoch": 0.03628625825960354,
      "grad_norm": 1.4487106800079346,
      "learning_rate": 3.2300000000000004e-06,
      "loss": 0.4767,
      "step": 648
    },
    {
      "epoch": 0.036342255571732554,
      "grad_norm": 1.2317790985107422,
      "learning_rate": 3.235e-06,
      "loss": 0.4973,
      "step": 649
    },
    {
      "epoch": 0.03639825288386157,
      "grad_norm": 1.002759337425232,
      "learning_rate": 3.24e-06,
      "loss": 0.3591,
      "step": 650
    },
    {
      "epoch": 0.03645425019599059,
      "grad_norm": 1.2259923219680786,
      "learning_rate": 3.2450000000000003e-06,
      "loss": 0.3969,
      "step": 651
    },
    {
      "epoch": 0.03651024750811961,
      "grad_norm": 1.3137469291687012,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.4246,
      "step": 652
    },
    {
      "epoch": 0.03656624482024863,
      "grad_norm": 1.1844868659973145,
      "learning_rate": 3.2550000000000006e-06,
      "loss": 0.4375,
      "step": 653
    },
    {
      "epoch": 0.03662224213237764,
      "grad_norm": 1.4780162572860718,
      "learning_rate": 3.2599999999999997e-06,
      "loss": 0.5578,
      "step": 654
    },
    {
      "epoch": 0.03667823944450666,
      "grad_norm": 1.3112592697143555,
      "learning_rate": 3.265e-06,
      "loss": 0.354,
      "step": 655
    },
    {
      "epoch": 0.03673423675663568,
      "grad_norm": 1.0690807104110718,
      "learning_rate": 3.27e-06,
      "loss": 0.4815,
      "step": 656
    },
    {
      "epoch": 0.0367902340687647,
      "grad_norm": 1.209903359413147,
      "learning_rate": 3.2750000000000004e-06,
      "loss": 0.3635,
      "step": 657
    },
    {
      "epoch": 0.03684623138089372,
      "grad_norm": 1.1705541610717773,
      "learning_rate": 3.2800000000000004e-06,
      "loss": 0.5315,
      "step": 658
    },
    {
      "epoch": 0.03690222869302273,
      "grad_norm": 1.2070939540863037,
      "learning_rate": 3.285e-06,
      "loss": 0.3196,
      "step": 659
    },
    {
      "epoch": 0.03695822600515175,
      "grad_norm": 1.1473920345306396,
      "learning_rate": 3.29e-06,
      "loss": 0.4252,
      "step": 660
    },
    {
      "epoch": 0.03701422331728077,
      "grad_norm": 1.2592459917068481,
      "learning_rate": 3.2950000000000002e-06,
      "loss": 0.4727,
      "step": 661
    },
    {
      "epoch": 0.03707022062940979,
      "grad_norm": 2.0471949577331543,
      "learning_rate": 3.3e-06,
      "loss": 0.6152,
      "step": 662
    },
    {
      "epoch": 0.03712621794153881,
      "grad_norm": 1.3031131029129028,
      "learning_rate": 3.3050000000000005e-06,
      "loss": 0.7303,
      "step": 663
    },
    {
      "epoch": 0.03718221525366782,
      "grad_norm": 1.2334582805633545,
      "learning_rate": 3.31e-06,
      "loss": 0.5149,
      "step": 664
    },
    {
      "epoch": 0.03723821256579684,
      "grad_norm": 1.376380205154419,
      "learning_rate": 3.315e-06,
      "loss": 0.4672,
      "step": 665
    },
    {
      "epoch": 0.03729420987792586,
      "grad_norm": 1.150626301765442,
      "learning_rate": 3.3200000000000004e-06,
      "loss": 0.506,
      "step": 666
    },
    {
      "epoch": 0.03735020719005488,
      "grad_norm": 1.3962228298187256,
      "learning_rate": 3.3250000000000004e-06,
      "loss": 0.6496,
      "step": 667
    },
    {
      "epoch": 0.0374062045021839,
      "grad_norm": 1.3976243734359741,
      "learning_rate": 3.3300000000000003e-06,
      "loss": 0.5132,
      "step": 668
    },
    {
      "epoch": 0.037462201814312916,
      "grad_norm": 1.2252672910690308,
      "learning_rate": 3.335e-06,
      "loss": 0.3773,
      "step": 669
    },
    {
      "epoch": 0.03751819912644193,
      "grad_norm": 1.3803831338882446,
      "learning_rate": 3.34e-06,
      "loss": 0.4245,
      "step": 670
    },
    {
      "epoch": 0.03757419643857095,
      "grad_norm": 1.45225989818573,
      "learning_rate": 3.345e-06,
      "loss": 0.7511,
      "step": 671
    },
    {
      "epoch": 0.03763019375069997,
      "grad_norm": 1.3100789785385132,
      "learning_rate": 3.3500000000000005e-06,
      "loss": 0.4476,
      "step": 672
    },
    {
      "epoch": 0.037686191062828986,
      "grad_norm": 1.2988783121109009,
      "learning_rate": 3.3550000000000005e-06,
      "loss": 0.5123,
      "step": 673
    },
    {
      "epoch": 0.037742188374958005,
      "grad_norm": 1.1171373128890991,
      "learning_rate": 3.36e-06,
      "loss": 0.4446,
      "step": 674
    },
    {
      "epoch": 0.03779818568708702,
      "grad_norm": 1.5306018590927124,
      "learning_rate": 3.365e-06,
      "loss": 0.4832,
      "step": 675
    },
    {
      "epoch": 0.037854182999216036,
      "grad_norm": 1.277156114578247,
      "learning_rate": 3.3700000000000003e-06,
      "loss": 0.4448,
      "step": 676
    },
    {
      "epoch": 0.037910180311345056,
      "grad_norm": 1.3082318305969238,
      "learning_rate": 3.3750000000000003e-06,
      "loss": 0.4844,
      "step": 677
    },
    {
      "epoch": 0.037966177623474075,
      "grad_norm": 1.2136180400848389,
      "learning_rate": 3.38e-06,
      "loss": 0.5297,
      "step": 678
    },
    {
      "epoch": 0.038022174935603094,
      "grad_norm": 1.3228617906570435,
      "learning_rate": 3.3849999999999998e-06,
      "loss": 0.4339,
      "step": 679
    },
    {
      "epoch": 0.038078172247732106,
      "grad_norm": 1.2094823122024536,
      "learning_rate": 3.39e-06,
      "loss": 0.4614,
      "step": 680
    },
    {
      "epoch": 0.038134169559861125,
      "grad_norm": 1.2327871322631836,
      "learning_rate": 3.395e-06,
      "loss": 0.4556,
      "step": 681
    },
    {
      "epoch": 0.038190166871990144,
      "grad_norm": 1.4336551427841187,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.5025,
      "step": 682
    },
    {
      "epoch": 0.038246164184119164,
      "grad_norm": 1.1964848041534424,
      "learning_rate": 3.405e-06,
      "loss": 0.3432,
      "step": 683
    },
    {
      "epoch": 0.03830216149624818,
      "grad_norm": 1.4089807271957397,
      "learning_rate": 3.41e-06,
      "loss": 0.5415,
      "step": 684
    },
    {
      "epoch": 0.038358158808377195,
      "grad_norm": 1.3728796243667603,
      "learning_rate": 3.4150000000000003e-06,
      "loss": 0.4414,
      "step": 685
    },
    {
      "epoch": 0.038414156120506214,
      "grad_norm": 1.1719013452529907,
      "learning_rate": 3.4200000000000003e-06,
      "loss": 0.4603,
      "step": 686
    },
    {
      "epoch": 0.03847015343263523,
      "grad_norm": 1.244240164756775,
      "learning_rate": 3.4250000000000002e-06,
      "loss": 0.5535,
      "step": 687
    },
    {
      "epoch": 0.03852615074476425,
      "grad_norm": 1.2287418842315674,
      "learning_rate": 3.4299999999999998e-06,
      "loss": 0.4683,
      "step": 688
    },
    {
      "epoch": 0.03858214805689327,
      "grad_norm": 1.5462092161178589,
      "learning_rate": 3.435e-06,
      "loss": 0.5485,
      "step": 689
    },
    {
      "epoch": 0.038638145369022284,
      "grad_norm": 1.2673622369766235,
      "learning_rate": 3.44e-06,
      "loss": 0.6002,
      "step": 690
    },
    {
      "epoch": 0.0386941426811513,
      "grad_norm": 1.442499041557312,
      "learning_rate": 3.4450000000000005e-06,
      "loss": 0.6001,
      "step": 691
    },
    {
      "epoch": 0.03875013999328032,
      "grad_norm": 1.2574551105499268,
      "learning_rate": 3.4500000000000004e-06,
      "loss": 0.4971,
      "step": 692
    },
    {
      "epoch": 0.03880613730540934,
      "grad_norm": 1.633691668510437,
      "learning_rate": 3.455e-06,
      "loss": 0.6757,
      "step": 693
    },
    {
      "epoch": 0.03886213461753836,
      "grad_norm": 1.447800636291504,
      "learning_rate": 3.46e-06,
      "loss": 0.506,
      "step": 694
    },
    {
      "epoch": 0.03891813192966737,
      "grad_norm": 1.336168646812439,
      "learning_rate": 3.4650000000000003e-06,
      "loss": 0.4252,
      "step": 695
    },
    {
      "epoch": 0.03897412924179639,
      "grad_norm": 1.0794219970703125,
      "learning_rate": 3.4700000000000002e-06,
      "loss": 0.4782,
      "step": 696
    },
    {
      "epoch": 0.03903012655392541,
      "grad_norm": 1.2878397703170776,
      "learning_rate": 3.4750000000000006e-06,
      "loss": 0.6454,
      "step": 697
    },
    {
      "epoch": 0.03908612386605443,
      "grad_norm": 1.3281662464141846,
      "learning_rate": 3.4799999999999997e-06,
      "loss": 0.4986,
      "step": 698
    },
    {
      "epoch": 0.03914212117818345,
      "grad_norm": 1.8202744722366333,
      "learning_rate": 3.485e-06,
      "loss": 0.6661,
      "step": 699
    },
    {
      "epoch": 0.03919811849031247,
      "grad_norm": 1.1826694011688232,
      "learning_rate": 3.49e-06,
      "loss": 0.4021,
      "step": 700
    },
    {
      "epoch": 0.03925411580244148,
      "grad_norm": 2.1669232845306396,
      "learning_rate": 3.4950000000000004e-06,
      "loss": 0.4714,
      "step": 701
    },
    {
      "epoch": 0.0393101131145705,
      "grad_norm": 1.3901138305664062,
      "learning_rate": 3.5000000000000004e-06,
      "loss": 0.5407,
      "step": 702
    },
    {
      "epoch": 0.03936611042669952,
      "grad_norm": 1.0436351299285889,
      "learning_rate": 3.505e-06,
      "loss": 0.3554,
      "step": 703
    },
    {
      "epoch": 0.03942210773882854,
      "grad_norm": 1.0835562944412231,
      "learning_rate": 3.5100000000000003e-06,
      "loss": 0.4768,
      "step": 704
    },
    {
      "epoch": 0.03947810505095756,
      "grad_norm": 1.1899420022964478,
      "learning_rate": 3.5150000000000002e-06,
      "loss": 0.539,
      "step": 705
    },
    {
      "epoch": 0.03953410236308657,
      "grad_norm": 1.0693700313568115,
      "learning_rate": 3.52e-06,
      "loss": 0.4778,
      "step": 706
    },
    {
      "epoch": 0.03959009967521559,
      "grad_norm": 1.595470666885376,
      "learning_rate": 3.5249999999999997e-06,
      "loss": 0.5446,
      "step": 707
    },
    {
      "epoch": 0.03964609698734461,
      "grad_norm": 1.35414719581604,
      "learning_rate": 3.53e-06,
      "loss": 0.5125,
      "step": 708
    },
    {
      "epoch": 0.03970209429947363,
      "grad_norm": 1.272335410118103,
      "learning_rate": 3.535e-06,
      "loss": 0.4132,
      "step": 709
    },
    {
      "epoch": 0.039758091611602646,
      "grad_norm": 3.9252395629882812,
      "learning_rate": 3.5400000000000004e-06,
      "loss": 0.488,
      "step": 710
    },
    {
      "epoch": 0.03981408892373166,
      "grad_norm": 1.3578699827194214,
      "learning_rate": 3.5450000000000004e-06,
      "loss": 0.6098,
      "step": 711
    },
    {
      "epoch": 0.03987008623586068,
      "grad_norm": 1.3184001445770264,
      "learning_rate": 3.55e-06,
      "loss": 0.3705,
      "step": 712
    },
    {
      "epoch": 0.039926083547989696,
      "grad_norm": 1.26827871799469,
      "learning_rate": 3.555e-06,
      "loss": 0.4091,
      "step": 713
    },
    {
      "epoch": 0.039982080860118716,
      "grad_norm": 1.2401190996170044,
      "learning_rate": 3.5600000000000002e-06,
      "loss": 0.5061,
      "step": 714
    },
    {
      "epoch": 0.040038078172247735,
      "grad_norm": 1.6993461847305298,
      "learning_rate": 3.565e-06,
      "loss": 0.5484,
      "step": 715
    },
    {
      "epoch": 0.04009407548437675,
      "grad_norm": 1.1220078468322754,
      "learning_rate": 3.5700000000000005e-06,
      "loss": 0.5662,
      "step": 716
    },
    {
      "epoch": 0.040150072796505766,
      "grad_norm": 1.306207299232483,
      "learning_rate": 3.575e-06,
      "loss": 0.4847,
      "step": 717
    },
    {
      "epoch": 0.040206070108634785,
      "grad_norm": 1.3487942218780518,
      "learning_rate": 3.58e-06,
      "loss": 0.428,
      "step": 718
    },
    {
      "epoch": 0.040262067420763804,
      "grad_norm": 1.2893807888031006,
      "learning_rate": 3.585e-06,
      "loss": 0.439,
      "step": 719
    },
    {
      "epoch": 0.040318064732892823,
      "grad_norm": 1.2305127382278442,
      "learning_rate": 3.5900000000000004e-06,
      "loss": 0.4885,
      "step": 720
    },
    {
      "epoch": 0.040374062045021836,
      "grad_norm": 1.2074695825576782,
      "learning_rate": 3.5950000000000003e-06,
      "loss": 0.4735,
      "step": 721
    },
    {
      "epoch": 0.040430059357150855,
      "grad_norm": 1.2309809923171997,
      "learning_rate": 3.6e-06,
      "loss": 0.4212,
      "step": 722
    },
    {
      "epoch": 0.040486056669279874,
      "grad_norm": 1.4647796154022217,
      "learning_rate": 3.6050000000000002e-06,
      "loss": 0.4403,
      "step": 723
    },
    {
      "epoch": 0.04054205398140889,
      "grad_norm": 2.394265651702881,
      "learning_rate": 3.61e-06,
      "loss": 0.5968,
      "step": 724
    },
    {
      "epoch": 0.04059805129353791,
      "grad_norm": 1.4066598415374756,
      "learning_rate": 3.6150000000000005e-06,
      "loss": 0.4851,
      "step": 725
    },
    {
      "epoch": 0.040654048605666925,
      "grad_norm": 1.1987708806991577,
      "learning_rate": 3.6200000000000005e-06,
      "loss": 0.4466,
      "step": 726
    },
    {
      "epoch": 0.040710045917795944,
      "grad_norm": 1.0690430402755737,
      "learning_rate": 3.625e-06,
      "loss": 0.5593,
      "step": 727
    },
    {
      "epoch": 0.04076604322992496,
      "grad_norm": 1.1197233200073242,
      "learning_rate": 3.63e-06,
      "loss": 0.5191,
      "step": 728
    },
    {
      "epoch": 0.04082204054205398,
      "grad_norm": 1.168726921081543,
      "learning_rate": 3.6350000000000003e-06,
      "loss": 0.3816,
      "step": 729
    },
    {
      "epoch": 0.040878037854183,
      "grad_norm": 1.2563281059265137,
      "learning_rate": 3.6400000000000003e-06,
      "loss": 0.4761,
      "step": 730
    },
    {
      "epoch": 0.04093403516631202,
      "grad_norm": 1.3992208242416382,
      "learning_rate": 3.6450000000000007e-06,
      "loss": 0.5434,
      "step": 731
    },
    {
      "epoch": 0.04099003247844103,
      "grad_norm": 0.9997865557670593,
      "learning_rate": 3.6499999999999998e-06,
      "loss": 0.4953,
      "step": 732
    },
    {
      "epoch": 0.04104602979057005,
      "grad_norm": 1.3411892652511597,
      "learning_rate": 3.655e-06,
      "loss": 0.4478,
      "step": 733
    },
    {
      "epoch": 0.04110202710269907,
      "grad_norm": 6.3067450523376465,
      "learning_rate": 3.66e-06,
      "loss": 0.4111,
      "step": 734
    },
    {
      "epoch": 0.04115802441482809,
      "grad_norm": 1.3528656959533691,
      "learning_rate": 3.6650000000000005e-06,
      "loss": 0.522,
      "step": 735
    },
    {
      "epoch": 0.04121402172695711,
      "grad_norm": 1.1481785774230957,
      "learning_rate": 3.6700000000000004e-06,
      "loss": 0.4935,
      "step": 736
    },
    {
      "epoch": 0.04127001903908612,
      "grad_norm": 1.2303099632263184,
      "learning_rate": 3.675e-06,
      "loss": 0.4304,
      "step": 737
    },
    {
      "epoch": 0.04132601635121514,
      "grad_norm": 1.5268962383270264,
      "learning_rate": 3.68e-06,
      "loss": 0.4484,
      "step": 738
    },
    {
      "epoch": 0.04138201366334416,
      "grad_norm": 1.2627424001693726,
      "learning_rate": 3.6850000000000003e-06,
      "loss": 0.489,
      "step": 739
    },
    {
      "epoch": 0.04143801097547318,
      "grad_norm": 1.397262692451477,
      "learning_rate": 3.6900000000000002e-06,
      "loss": 0.5543,
      "step": 740
    },
    {
      "epoch": 0.0414940082876022,
      "grad_norm": 1.2485219240188599,
      "learning_rate": 3.6949999999999998e-06,
      "loss": 0.4055,
      "step": 741
    },
    {
      "epoch": 0.04155000559973121,
      "grad_norm": 1.1135332584381104,
      "learning_rate": 3.7e-06,
      "loss": 0.3795,
      "step": 742
    },
    {
      "epoch": 0.04160600291186023,
      "grad_norm": 1.4126280546188354,
      "learning_rate": 3.705e-06,
      "loss": 0.6212,
      "step": 743
    },
    {
      "epoch": 0.04166200022398925,
      "grad_norm": 1.4079738855361938,
      "learning_rate": 3.7100000000000005e-06,
      "loss": 0.7047,
      "step": 744
    },
    {
      "epoch": 0.04171799753611827,
      "grad_norm": 1.4724724292755127,
      "learning_rate": 3.7150000000000004e-06,
      "loss": 0.4877,
      "step": 745
    },
    {
      "epoch": 0.04177399484824729,
      "grad_norm": 1.661381721496582,
      "learning_rate": 3.72e-06,
      "loss": 0.4769,
      "step": 746
    },
    {
      "epoch": 0.0418299921603763,
      "grad_norm": 1.7881790399551392,
      "learning_rate": 3.725e-06,
      "loss": 0.563,
      "step": 747
    },
    {
      "epoch": 0.04188598947250532,
      "grad_norm": 1.3039181232452393,
      "learning_rate": 3.7300000000000003e-06,
      "loss": 0.5058,
      "step": 748
    },
    {
      "epoch": 0.04194198678463434,
      "grad_norm": 1.1592111587524414,
      "learning_rate": 3.7350000000000002e-06,
      "loss": 0.4011,
      "step": 749
    },
    {
      "epoch": 0.041997984096763356,
      "grad_norm": 1.2834731340408325,
      "learning_rate": 3.7400000000000006e-06,
      "loss": 0.4299,
      "step": 750
    },
    {
      "epoch": 0.042053981408892376,
      "grad_norm": 1.2477951049804688,
      "learning_rate": 3.7449999999999997e-06,
      "loss": 0.4805,
      "step": 751
    },
    {
      "epoch": 0.04210997872102139,
      "grad_norm": 1.5107780694961548,
      "learning_rate": 3.75e-06,
      "loss": 0.4794,
      "step": 752
    },
    {
      "epoch": 0.04216597603315041,
      "grad_norm": 1.288833737373352,
      "learning_rate": 3.755e-06,
      "loss": 0.5569,
      "step": 753
    },
    {
      "epoch": 0.042221973345279426,
      "grad_norm": 1.2187424898147583,
      "learning_rate": 3.7600000000000004e-06,
      "loss": 0.3778,
      "step": 754
    },
    {
      "epoch": 0.042277970657408445,
      "grad_norm": 1.2177382707595825,
      "learning_rate": 3.7650000000000004e-06,
      "loss": 0.4573,
      "step": 755
    },
    {
      "epoch": 0.042333967969537464,
      "grad_norm": 1.2576156854629517,
      "learning_rate": 3.77e-06,
      "loss": 0.443,
      "step": 756
    },
    {
      "epoch": 0.042389965281666483,
      "grad_norm": 1.2027933597564697,
      "learning_rate": 3.775e-06,
      "loss": 0.5188,
      "step": 757
    },
    {
      "epoch": 0.042445962593795496,
      "grad_norm": 1.3134757280349731,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 0.5296,
      "step": 758
    },
    {
      "epoch": 0.042501959905924515,
      "grad_norm": 1.1032532453536987,
      "learning_rate": 3.785e-06,
      "loss": 0.4721,
      "step": 759
    },
    {
      "epoch": 0.042557957218053534,
      "grad_norm": 1.1537388563156128,
      "learning_rate": 3.7900000000000006e-06,
      "loss": 0.3652,
      "step": 760
    },
    {
      "epoch": 0.04261395453018255,
      "grad_norm": 1.1453958749771118,
      "learning_rate": 3.795e-06,
      "loss": 0.3934,
      "step": 761
    },
    {
      "epoch": 0.04266995184231157,
      "grad_norm": 1.3031656742095947,
      "learning_rate": 3.8e-06,
      "loss": 0.4986,
      "step": 762
    },
    {
      "epoch": 0.042725949154440584,
      "grad_norm": 1.8433433771133423,
      "learning_rate": 3.8050000000000004e-06,
      "loss": 0.615,
      "step": 763
    },
    {
      "epoch": 0.042781946466569604,
      "grad_norm": 1.5257457494735718,
      "learning_rate": 3.8100000000000004e-06,
      "loss": 0.5185,
      "step": 764
    },
    {
      "epoch": 0.04283794377869862,
      "grad_norm": 1.3897265195846558,
      "learning_rate": 3.815000000000001e-06,
      "loss": 0.5186,
      "step": 765
    },
    {
      "epoch": 0.04289394109082764,
      "grad_norm": 1.1871623992919922,
      "learning_rate": 3.82e-06,
      "loss": 0.4082,
      "step": 766
    },
    {
      "epoch": 0.04294993840295666,
      "grad_norm": 2.2760584354400635,
      "learning_rate": 3.825e-06,
      "loss": 0.8009,
      "step": 767
    },
    {
      "epoch": 0.04300593571508567,
      "grad_norm": 1.3228431940078735,
      "learning_rate": 3.830000000000001e-06,
      "loss": 0.4165,
      "step": 768
    },
    {
      "epoch": 0.04306193302721469,
      "grad_norm": 1.8395787477493286,
      "learning_rate": 3.8350000000000006e-06,
      "loss": 0.6306,
      "step": 769
    },
    {
      "epoch": 0.04311793033934371,
      "grad_norm": 1.3083232641220093,
      "learning_rate": 3.84e-06,
      "loss": 0.393,
      "step": 770
    },
    {
      "epoch": 0.04317392765147273,
      "grad_norm": 1.3711607456207275,
      "learning_rate": 3.845e-06,
      "loss": 0.5634,
      "step": 771
    },
    {
      "epoch": 0.04322992496360175,
      "grad_norm": 1.2909704446792603,
      "learning_rate": 3.85e-06,
      "loss": 0.4414,
      "step": 772
    },
    {
      "epoch": 0.04328592227573076,
      "grad_norm": 1.2462844848632812,
      "learning_rate": 3.855e-06,
      "loss": 0.4788,
      "step": 773
    },
    {
      "epoch": 0.04334191958785978,
      "grad_norm": 1.2240833044052124,
      "learning_rate": 3.86e-06,
      "loss": 0.5835,
      "step": 774
    },
    {
      "epoch": 0.0433979168999888,
      "grad_norm": 1.207229733467102,
      "learning_rate": 3.865e-06,
      "loss": 0.4087,
      "step": 775
    },
    {
      "epoch": 0.04345391421211782,
      "grad_norm": 1.5298811197280884,
      "learning_rate": 3.87e-06,
      "loss": 0.5086,
      "step": 776
    },
    {
      "epoch": 0.04350991152424684,
      "grad_norm": 1.5431842803955078,
      "learning_rate": 3.875e-06,
      "loss": 0.5255,
      "step": 777
    },
    {
      "epoch": 0.04356590883637585,
      "grad_norm": 1.2514537572860718,
      "learning_rate": 3.88e-06,
      "loss": 0.4644,
      "step": 778
    },
    {
      "epoch": 0.04362190614850487,
      "grad_norm": 1.1557564735412598,
      "learning_rate": 3.885e-06,
      "loss": 0.4376,
      "step": 779
    },
    {
      "epoch": 0.04367790346063389,
      "grad_norm": 1.2817264795303345,
      "learning_rate": 3.89e-06,
      "loss": 0.5082,
      "step": 780
    },
    {
      "epoch": 0.04373390077276291,
      "grad_norm": 1.2424750328063965,
      "learning_rate": 3.895e-06,
      "loss": 0.506,
      "step": 781
    },
    {
      "epoch": 0.04378989808489193,
      "grad_norm": 1.014085054397583,
      "learning_rate": 3.9e-06,
      "loss": 0.5267,
      "step": 782
    },
    {
      "epoch": 0.04384589539702094,
      "grad_norm": 1.3631815910339355,
      "learning_rate": 3.905000000000001e-06,
      "loss": 0.4526,
      "step": 783
    },
    {
      "epoch": 0.04390189270914996,
      "grad_norm": 1.462369441986084,
      "learning_rate": 3.910000000000001e-06,
      "loss": 0.3593,
      "step": 784
    },
    {
      "epoch": 0.04395789002127898,
      "grad_norm": 1.1282445192337036,
      "learning_rate": 3.915e-06,
      "loss": 0.4777,
      "step": 785
    },
    {
      "epoch": 0.044013887333408,
      "grad_norm": 1.2128945589065552,
      "learning_rate": 3.92e-06,
      "loss": 0.3664,
      "step": 786
    },
    {
      "epoch": 0.044069884645537016,
      "grad_norm": 1.27531099319458,
      "learning_rate": 3.9250000000000005e-06,
      "loss": 0.5199,
      "step": 787
    },
    {
      "epoch": 0.044125881957666035,
      "grad_norm": 1.6103917360305786,
      "learning_rate": 3.9300000000000005e-06,
      "loss": 0.7764,
      "step": 788
    },
    {
      "epoch": 0.04418187926979505,
      "grad_norm": 1.2535672187805176,
      "learning_rate": 3.9350000000000004e-06,
      "loss": 0.5684,
      "step": 789
    },
    {
      "epoch": 0.04423787658192407,
      "grad_norm": 1.2681461572647095,
      "learning_rate": 3.9399999999999995e-06,
      "loss": 0.4236,
      "step": 790
    },
    {
      "epoch": 0.044293873894053086,
      "grad_norm": 1.3095355033874512,
      "learning_rate": 3.945e-06,
      "loss": 0.5838,
      "step": 791
    },
    {
      "epoch": 0.044349871206182105,
      "grad_norm": 1.3122279644012451,
      "learning_rate": 3.95e-06,
      "loss": 0.4812,
      "step": 792
    },
    {
      "epoch": 0.044405868518311124,
      "grad_norm": 1.420548915863037,
      "learning_rate": 3.955e-06,
      "loss": 0.512,
      "step": 793
    },
    {
      "epoch": 0.044461865830440137,
      "grad_norm": 1.1349029541015625,
      "learning_rate": 3.96e-06,
      "loss": 0.3778,
      "step": 794
    },
    {
      "epoch": 0.044517863142569156,
      "grad_norm": 1.3369938135147095,
      "learning_rate": 3.965e-06,
      "loss": 0.4454,
      "step": 795
    },
    {
      "epoch": 0.044573860454698175,
      "grad_norm": 1.3689488172531128,
      "learning_rate": 3.97e-06,
      "loss": 0.4819,
      "step": 796
    },
    {
      "epoch": 0.044629857766827194,
      "grad_norm": 6.419103145599365,
      "learning_rate": 3.975e-06,
      "loss": 0.5101,
      "step": 797
    },
    {
      "epoch": 0.04468585507895621,
      "grad_norm": 1.4806280136108398,
      "learning_rate": 3.98e-06,
      "loss": 0.6097,
      "step": 798
    },
    {
      "epoch": 0.044741852391085225,
      "grad_norm": 1.527851939201355,
      "learning_rate": 3.985e-06,
      "loss": 0.6501,
      "step": 799
    },
    {
      "epoch": 0.044797849703214244,
      "grad_norm": 1.3300751447677612,
      "learning_rate": 3.99e-06,
      "loss": 0.4639,
      "step": 800
    },
    {
      "epoch": 0.044853847015343264,
      "grad_norm": 1.3300977945327759,
      "learning_rate": 3.995e-06,
      "loss": 0.4535,
      "step": 801
    },
    {
      "epoch": 0.04490984432747228,
      "grad_norm": 1.2579883337020874,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.5186,
      "step": 802
    },
    {
      "epoch": 0.0449658416396013,
      "grad_norm": 1.0359623432159424,
      "learning_rate": 4.005000000000001e-06,
      "loss": 0.4602,
      "step": 803
    },
    {
      "epoch": 0.045021838951730314,
      "grad_norm": 1.2562466859817505,
      "learning_rate": 4.01e-06,
      "loss": 0.6528,
      "step": 804
    },
    {
      "epoch": 0.04507783626385933,
      "grad_norm": 1.155367374420166,
      "learning_rate": 4.015e-06,
      "loss": 0.4138,
      "step": 805
    },
    {
      "epoch": 0.04513383357598835,
      "grad_norm": 1.1500927209854126,
      "learning_rate": 4.0200000000000005e-06,
      "loss": 0.4477,
      "step": 806
    },
    {
      "epoch": 0.04518983088811737,
      "grad_norm": 1.2277973890304565,
      "learning_rate": 4.0250000000000004e-06,
      "loss": 0.4754,
      "step": 807
    },
    {
      "epoch": 0.04524582820024639,
      "grad_norm": 1.1716548204421997,
      "learning_rate": 4.03e-06,
      "loss": 0.3876,
      "step": 808
    },
    {
      "epoch": 0.0453018255123754,
      "grad_norm": 1.2428687810897827,
      "learning_rate": 4.0349999999999995e-06,
      "loss": 0.4257,
      "step": 809
    },
    {
      "epoch": 0.04535782282450442,
      "grad_norm": 1.1642229557037354,
      "learning_rate": 4.04e-06,
      "loss": 0.4966,
      "step": 810
    },
    {
      "epoch": 0.04541382013663344,
      "grad_norm": 1.3695844411849976,
      "learning_rate": 4.045e-06,
      "loss": 0.4443,
      "step": 811
    },
    {
      "epoch": 0.04546981744876246,
      "grad_norm": 1.215888500213623,
      "learning_rate": 4.05e-06,
      "loss": 0.4461,
      "step": 812
    },
    {
      "epoch": 0.04552581476089148,
      "grad_norm": 1.8594948053359985,
      "learning_rate": 4.055e-06,
      "loss": 0.5258,
      "step": 813
    },
    {
      "epoch": 0.04558181207302049,
      "grad_norm": 1.3100166320800781,
      "learning_rate": 4.06e-06,
      "loss": 0.4898,
      "step": 814
    },
    {
      "epoch": 0.04563780938514951,
      "grad_norm": 1.0542774200439453,
      "learning_rate": 4.065e-06,
      "loss": 0.3433,
      "step": 815
    },
    {
      "epoch": 0.04569380669727853,
      "grad_norm": 1.1762754917144775,
      "learning_rate": 4.07e-06,
      "loss": 0.4285,
      "step": 816
    },
    {
      "epoch": 0.04574980400940755,
      "grad_norm": 1.7509896755218506,
      "learning_rate": 4.075e-06,
      "loss": 0.7001,
      "step": 817
    },
    {
      "epoch": 0.04580580132153657,
      "grad_norm": 2.022645950317383,
      "learning_rate": 4.080000000000001e-06,
      "loss": 0.4217,
      "step": 818
    },
    {
      "epoch": 0.04586179863366559,
      "grad_norm": 1.2015148401260376,
      "learning_rate": 4.085e-06,
      "loss": 0.4548,
      "step": 819
    },
    {
      "epoch": 0.0459177959457946,
      "grad_norm": 1.2890264987945557,
      "learning_rate": 4.09e-06,
      "loss": 0.4291,
      "step": 820
    },
    {
      "epoch": 0.04597379325792362,
      "grad_norm": 1.2706619501113892,
      "learning_rate": 4.095000000000001e-06,
      "loss": 0.4252,
      "step": 821
    },
    {
      "epoch": 0.04602979057005264,
      "grad_norm": 1.3321216106414795,
      "learning_rate": 4.1000000000000006e-06,
      "loss": 0.4343,
      "step": 822
    },
    {
      "epoch": 0.04608578788218166,
      "grad_norm": 1.4125432968139648,
      "learning_rate": 4.1050000000000005e-06,
      "loss": 0.6235,
      "step": 823
    },
    {
      "epoch": 0.046141785194310676,
      "grad_norm": 1.304635763168335,
      "learning_rate": 4.11e-06,
      "loss": 0.6007,
      "step": 824
    },
    {
      "epoch": 0.04619778250643969,
      "grad_norm": 1.1754714250564575,
      "learning_rate": 4.115e-06,
      "loss": 0.4203,
      "step": 825
    },
    {
      "epoch": 0.04625377981856871,
      "grad_norm": 1.2974853515625,
      "learning_rate": 4.12e-06,
      "loss": 0.6425,
      "step": 826
    },
    {
      "epoch": 0.04630977713069773,
      "grad_norm": 1.3489078283309937,
      "learning_rate": 4.125e-06,
      "loss": 0.5505,
      "step": 827
    },
    {
      "epoch": 0.046365774442826746,
      "grad_norm": 1.4762914180755615,
      "learning_rate": 4.13e-06,
      "loss": 0.4373,
      "step": 828
    },
    {
      "epoch": 0.046421771754955765,
      "grad_norm": 1.038651943206787,
      "learning_rate": 4.135e-06,
      "loss": 0.4102,
      "step": 829
    },
    {
      "epoch": 0.04647776906708478,
      "grad_norm": 1.2074174880981445,
      "learning_rate": 4.14e-06,
      "loss": 0.557,
      "step": 830
    },
    {
      "epoch": 0.046533766379213796,
      "grad_norm": 1.2075752019882202,
      "learning_rate": 4.145e-06,
      "loss": 0.4114,
      "step": 831
    },
    {
      "epoch": 0.046589763691342816,
      "grad_norm": 1.296036720275879,
      "learning_rate": 4.15e-06,
      "loss": 0.5151,
      "step": 832
    },
    {
      "epoch": 0.046645761003471835,
      "grad_norm": 1.5932459831237793,
      "learning_rate": 4.155e-06,
      "loss": 0.6072,
      "step": 833
    },
    {
      "epoch": 0.046701758315600854,
      "grad_norm": 1.1555140018463135,
      "learning_rate": 4.16e-06,
      "loss": 0.4633,
      "step": 834
    },
    {
      "epoch": 0.046757755627729866,
      "grad_norm": 1.7015200853347778,
      "learning_rate": 4.165e-06,
      "loss": 0.5737,
      "step": 835
    },
    {
      "epoch": 0.046813752939858885,
      "grad_norm": 1.2817109823226929,
      "learning_rate": 4.17e-06,
      "loss": 0.6031,
      "step": 836
    },
    {
      "epoch": 0.046869750251987904,
      "grad_norm": 1.0388692617416382,
      "learning_rate": 4.175000000000001e-06,
      "loss": 0.3362,
      "step": 837
    },
    {
      "epoch": 0.046925747564116924,
      "grad_norm": 1.156796932220459,
      "learning_rate": 4.18e-06,
      "loss": 0.4485,
      "step": 838
    },
    {
      "epoch": 0.04698174487624594,
      "grad_norm": 1.2893918752670288,
      "learning_rate": 4.185e-06,
      "loss": 0.5426,
      "step": 839
    },
    {
      "epoch": 0.047037742188374955,
      "grad_norm": 1.2383168935775757,
      "learning_rate": 4.1900000000000005e-06,
      "loss": 0.4145,
      "step": 840
    },
    {
      "epoch": 0.047093739500503974,
      "grad_norm": 1.4720534086227417,
      "learning_rate": 4.1950000000000005e-06,
      "loss": 0.5713,
      "step": 841
    },
    {
      "epoch": 0.04714973681263299,
      "grad_norm": 5.215790748596191,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.4335,
      "step": 842
    },
    {
      "epoch": 0.04720573412476201,
      "grad_norm": 1.6487631797790527,
      "learning_rate": 4.2049999999999996e-06,
      "loss": 0.466,
      "step": 843
    },
    {
      "epoch": 0.04726173143689103,
      "grad_norm": 1.4052884578704834,
      "learning_rate": 4.21e-06,
      "loss": 0.5244,
      "step": 844
    },
    {
      "epoch": 0.047317728749020044,
      "grad_norm": 1.196442723274231,
      "learning_rate": 4.215e-06,
      "loss": 0.5315,
      "step": 845
    },
    {
      "epoch": 0.04737372606114906,
      "grad_norm": 1.341271996498108,
      "learning_rate": 4.22e-06,
      "loss": 0.5769,
      "step": 846
    },
    {
      "epoch": 0.04742972337327808,
      "grad_norm": 1.5218042135238647,
      "learning_rate": 4.225e-06,
      "loss": 0.5436,
      "step": 847
    },
    {
      "epoch": 0.0474857206854071,
      "grad_norm": 1.12349271774292,
      "learning_rate": 4.23e-06,
      "loss": 0.4701,
      "step": 848
    },
    {
      "epoch": 0.04754171799753612,
      "grad_norm": 1.3710609674453735,
      "learning_rate": 4.235e-06,
      "loss": 0.5771,
      "step": 849
    },
    {
      "epoch": 0.04759771530966514,
      "grad_norm": 1.0842112302780151,
      "learning_rate": 4.24e-06,
      "loss": 0.3687,
      "step": 850
    },
    {
      "epoch": 0.04765371262179415,
      "grad_norm": 1.6881595849990845,
      "learning_rate": 4.245e-06,
      "loss": 0.5157,
      "step": 851
    },
    {
      "epoch": 0.04770970993392317,
      "grad_norm": 5.3729047775268555,
      "learning_rate": 4.250000000000001e-06,
      "loss": 0.5579,
      "step": 852
    },
    {
      "epoch": 0.04776570724605219,
      "grad_norm": 1.4211618900299072,
      "learning_rate": 4.255e-06,
      "loss": 0.288,
      "step": 853
    },
    {
      "epoch": 0.04782170455818121,
      "grad_norm": 1.1432451009750366,
      "learning_rate": 4.26e-06,
      "loss": 0.6627,
      "step": 854
    },
    {
      "epoch": 0.04787770187031023,
      "grad_norm": 1.4285887479782104,
      "learning_rate": 4.265e-06,
      "loss": 0.5766,
      "step": 855
    },
    {
      "epoch": 0.04793369918243924,
      "grad_norm": 1.2693086862564087,
      "learning_rate": 4.270000000000001e-06,
      "loss": 0.5203,
      "step": 856
    },
    {
      "epoch": 0.04798969649456826,
      "grad_norm": 1.0911688804626465,
      "learning_rate": 4.2750000000000006e-06,
      "loss": 0.373,
      "step": 857
    },
    {
      "epoch": 0.04804569380669728,
      "grad_norm": 1.2234435081481934,
      "learning_rate": 4.28e-06,
      "loss": 0.4711,
      "step": 858
    },
    {
      "epoch": 0.0481016911188263,
      "grad_norm": 1.1317834854125977,
      "learning_rate": 4.2850000000000005e-06,
      "loss": 0.3715,
      "step": 859
    },
    {
      "epoch": 0.04815768843095532,
      "grad_norm": 1.4064030647277832,
      "learning_rate": 4.2900000000000004e-06,
      "loss": 0.4611,
      "step": 860
    },
    {
      "epoch": 0.04821368574308433,
      "grad_norm": 1.5491019487380981,
      "learning_rate": 4.295e-06,
      "loss": 0.4954,
      "step": 861
    },
    {
      "epoch": 0.04826968305521335,
      "grad_norm": 1.1596893072128296,
      "learning_rate": 4.2999999999999995e-06,
      "loss": 0.4147,
      "step": 862
    },
    {
      "epoch": 0.04832568036734237,
      "grad_norm": 1.39393150806427,
      "learning_rate": 4.305e-06,
      "loss": 0.4504,
      "step": 863
    },
    {
      "epoch": 0.04838167767947139,
      "grad_norm": 0.9863626956939697,
      "learning_rate": 4.31e-06,
      "loss": 0.4018,
      "step": 864
    },
    {
      "epoch": 0.048437674991600406,
      "grad_norm": 1.2500860691070557,
      "learning_rate": 4.315e-06,
      "loss": 0.3177,
      "step": 865
    },
    {
      "epoch": 0.04849367230372942,
      "grad_norm": 1.2027941942214966,
      "learning_rate": 4.32e-06,
      "loss": 0.5849,
      "step": 866
    },
    {
      "epoch": 0.04854966961585844,
      "grad_norm": 1.3106149435043335,
      "learning_rate": 4.325e-06,
      "loss": 0.3912,
      "step": 867
    },
    {
      "epoch": 0.048605666927987456,
      "grad_norm": 2.14656400680542,
      "learning_rate": 4.33e-06,
      "loss": 0.5411,
      "step": 868
    },
    {
      "epoch": 0.048661664240116476,
      "grad_norm": 1.165299415588379,
      "learning_rate": 4.335e-06,
      "loss": 0.3608,
      "step": 869
    },
    {
      "epoch": 0.048717661552245495,
      "grad_norm": 1.5042529106140137,
      "learning_rate": 4.34e-06,
      "loss": 0.3919,
      "step": 870
    },
    {
      "epoch": 0.04877365886437451,
      "grad_norm": 1.0934454202651978,
      "learning_rate": 4.345000000000001e-06,
      "loss": 0.3231,
      "step": 871
    },
    {
      "epoch": 0.048829656176503526,
      "grad_norm": 1.1957106590270996,
      "learning_rate": 4.35e-06,
      "loss": 0.5768,
      "step": 872
    },
    {
      "epoch": 0.048885653488632545,
      "grad_norm": 1.435402750968933,
      "learning_rate": 4.355e-06,
      "loss": 0.5095,
      "step": 873
    },
    {
      "epoch": 0.048941650800761564,
      "grad_norm": 1.366929531097412,
      "learning_rate": 4.360000000000001e-06,
      "loss": 0.4974,
      "step": 874
    },
    {
      "epoch": 0.048997648112890584,
      "grad_norm": 1.3479841947555542,
      "learning_rate": 4.3650000000000006e-06,
      "loss": 0.4395,
      "step": 875
    },
    {
      "epoch": 0.049053645425019596,
      "grad_norm": 1.4555615186691284,
      "learning_rate": 4.3700000000000005e-06,
      "loss": 0.7476,
      "step": 876
    },
    {
      "epoch": 0.049109642737148615,
      "grad_norm": 1.3223801851272583,
      "learning_rate": 4.375e-06,
      "loss": 0.4592,
      "step": 877
    },
    {
      "epoch": 0.049165640049277634,
      "grad_norm": 1.015773057937622,
      "learning_rate": 4.38e-06,
      "loss": 0.3394,
      "step": 878
    },
    {
      "epoch": 0.04922163736140665,
      "grad_norm": 1.224155068397522,
      "learning_rate": 4.385e-06,
      "loss": 0.4811,
      "step": 879
    },
    {
      "epoch": 0.04927763467353567,
      "grad_norm": 1.174251675605774,
      "learning_rate": 4.39e-06,
      "loss": 0.3923,
      "step": 880
    },
    {
      "epoch": 0.04933363198566469,
      "grad_norm": 1.5335127115249634,
      "learning_rate": 4.395e-06,
      "loss": 0.531,
      "step": 881
    },
    {
      "epoch": 0.049389629297793704,
      "grad_norm": 1.2524313926696777,
      "learning_rate": 4.4e-06,
      "loss": 0.5577,
      "step": 882
    },
    {
      "epoch": 0.04944562660992272,
      "grad_norm": 1.1989326477050781,
      "learning_rate": 4.405e-06,
      "loss": 0.5136,
      "step": 883
    },
    {
      "epoch": 0.04950162392205174,
      "grad_norm": 1.1970378160476685,
      "learning_rate": 4.41e-06,
      "loss": 0.3874,
      "step": 884
    },
    {
      "epoch": 0.04955762123418076,
      "grad_norm": 1.2088544368743896,
      "learning_rate": 4.415e-06,
      "loss": 0.4222,
      "step": 885
    },
    {
      "epoch": 0.04961361854630978,
      "grad_norm": 1.228928565979004,
      "learning_rate": 4.420000000000001e-06,
      "loss": 0.4155,
      "step": 886
    },
    {
      "epoch": 0.04966961585843879,
      "grad_norm": 1.324442744255066,
      "learning_rate": 4.425e-06,
      "loss": 0.4051,
      "step": 887
    },
    {
      "epoch": 0.04972561317056781,
      "grad_norm": 1.2726175785064697,
      "learning_rate": 4.43e-06,
      "loss": 0.3889,
      "step": 888
    },
    {
      "epoch": 0.04978161048269683,
      "grad_norm": 1.198268175125122,
      "learning_rate": 4.435e-06,
      "loss": 0.4054,
      "step": 889
    },
    {
      "epoch": 0.04983760779482585,
      "grad_norm": 1.3631871938705444,
      "learning_rate": 4.440000000000001e-06,
      "loss": 0.5341,
      "step": 890
    },
    {
      "epoch": 0.04989360510695487,
      "grad_norm": 1.1524522304534912,
      "learning_rate": 4.445000000000001e-06,
      "loss": 0.584,
      "step": 891
    },
    {
      "epoch": 0.04994960241908388,
      "grad_norm": 1.4735164642333984,
      "learning_rate": 4.45e-06,
      "loss": 0.5243,
      "step": 892
    },
    {
      "epoch": 0.0500055997312129,
      "grad_norm": 1.1556566953659058,
      "learning_rate": 4.4550000000000005e-06,
      "loss": 0.3543,
      "step": 893
    },
    {
      "epoch": 0.05006159704334192,
      "grad_norm": 1.2251462936401367,
      "learning_rate": 4.4600000000000005e-06,
      "loss": 0.4228,
      "step": 894
    },
    {
      "epoch": 0.05011759435547094,
      "grad_norm": 1.2216525077819824,
      "learning_rate": 4.4650000000000004e-06,
      "loss": 0.5462,
      "step": 895
    },
    {
      "epoch": 0.05017359166759996,
      "grad_norm": 3.547142744064331,
      "learning_rate": 4.4699999999999996e-06,
      "loss": 0.7819,
      "step": 896
    },
    {
      "epoch": 0.05022958897972897,
      "grad_norm": 1.2285093069076538,
      "learning_rate": 4.475e-06,
      "loss": 0.6265,
      "step": 897
    },
    {
      "epoch": 0.05028558629185799,
      "grad_norm": 1.074223518371582,
      "learning_rate": 4.48e-06,
      "loss": 0.3546,
      "step": 898
    },
    {
      "epoch": 0.05034158360398701,
      "grad_norm": 1.1250219345092773,
      "learning_rate": 4.485e-06,
      "loss": 0.42,
      "step": 899
    },
    {
      "epoch": 0.05039758091611603,
      "grad_norm": 1.5750139951705933,
      "learning_rate": 4.49e-06,
      "loss": 0.5449,
      "step": 900
    },
    {
      "epoch": 0.05045357822824505,
      "grad_norm": 1.213463306427002,
      "learning_rate": 4.495e-06,
      "loss": 0.4779,
      "step": 901
    },
    {
      "epoch": 0.05050957554037406,
      "grad_norm": 1.3777439594268799,
      "learning_rate": 4.5e-06,
      "loss": 0.4854,
      "step": 902
    },
    {
      "epoch": 0.05056557285250308,
      "grad_norm": 1.180437445640564,
      "learning_rate": 4.505e-06,
      "loss": 0.5171,
      "step": 903
    },
    {
      "epoch": 0.0506215701646321,
      "grad_norm": 1.2369575500488281,
      "learning_rate": 4.51e-06,
      "loss": 0.5179,
      "step": 904
    },
    {
      "epoch": 0.050677567476761116,
      "grad_norm": 1.1850531101226807,
      "learning_rate": 4.515000000000001e-06,
      "loss": 0.4336,
      "step": 905
    },
    {
      "epoch": 0.050733564788890136,
      "grad_norm": 1.65920889377594,
      "learning_rate": 4.52e-06,
      "loss": 0.4392,
      "step": 906
    },
    {
      "epoch": 0.05078956210101915,
      "grad_norm": 1.313531517982483,
      "learning_rate": 4.525e-06,
      "loss": 0.4201,
      "step": 907
    },
    {
      "epoch": 0.05084555941314817,
      "grad_norm": 1.7814490795135498,
      "learning_rate": 4.53e-06,
      "loss": 0.424,
      "step": 908
    },
    {
      "epoch": 0.050901556725277186,
      "grad_norm": 1.093064785003662,
      "learning_rate": 4.535000000000001e-06,
      "loss": 0.4134,
      "step": 909
    },
    {
      "epoch": 0.050957554037406205,
      "grad_norm": 1.2600054740905762,
      "learning_rate": 4.540000000000001e-06,
      "loss": 0.4557,
      "step": 910
    },
    {
      "epoch": 0.051013551349535224,
      "grad_norm": 1.131227970123291,
      "learning_rate": 4.545e-06,
      "loss": 0.4015,
      "step": 911
    },
    {
      "epoch": 0.051069548661664244,
      "grad_norm": 1.0976067781448364,
      "learning_rate": 4.5500000000000005e-06,
      "loss": 0.4971,
      "step": 912
    },
    {
      "epoch": 0.051125545973793256,
      "grad_norm": 1.1147174835205078,
      "learning_rate": 4.5550000000000004e-06,
      "loss": 0.3695,
      "step": 913
    },
    {
      "epoch": 0.051181543285922275,
      "grad_norm": 1.4934289455413818,
      "learning_rate": 4.56e-06,
      "loss": 0.6644,
      "step": 914
    },
    {
      "epoch": 0.051237540598051294,
      "grad_norm": 1.0716238021850586,
      "learning_rate": 4.565e-06,
      "loss": 0.3123,
      "step": 915
    },
    {
      "epoch": 0.05129353791018031,
      "grad_norm": 1.3327045440673828,
      "learning_rate": 4.57e-06,
      "loss": 0.5055,
      "step": 916
    },
    {
      "epoch": 0.05134953522230933,
      "grad_norm": 1.1724621057510376,
      "learning_rate": 4.575e-06,
      "loss": 0.5311,
      "step": 917
    },
    {
      "epoch": 0.051405532534438345,
      "grad_norm": 1.1683303117752075,
      "learning_rate": 4.58e-06,
      "loss": 0.5039,
      "step": 918
    },
    {
      "epoch": 0.051461529846567364,
      "grad_norm": 1.2588891983032227,
      "learning_rate": 4.585e-06,
      "loss": 0.5484,
      "step": 919
    },
    {
      "epoch": 0.05151752715869638,
      "grad_norm": 1.3223018646240234,
      "learning_rate": 4.590000000000001e-06,
      "loss": 0.5563,
      "step": 920
    },
    {
      "epoch": 0.0515735244708254,
      "grad_norm": 1.3173508644104004,
      "learning_rate": 4.595e-06,
      "loss": 0.466,
      "step": 921
    },
    {
      "epoch": 0.05162952178295442,
      "grad_norm": 1.736454963684082,
      "learning_rate": 4.6e-06,
      "loss": 0.5395,
      "step": 922
    },
    {
      "epoch": 0.05168551909508343,
      "grad_norm": 1.2006267309188843,
      "learning_rate": 4.605e-06,
      "loss": 0.5065,
      "step": 923
    },
    {
      "epoch": 0.05174151640721245,
      "grad_norm": 1.3227877616882324,
      "learning_rate": 4.610000000000001e-06,
      "loss": 0.4938,
      "step": 924
    },
    {
      "epoch": 0.05179751371934147,
      "grad_norm": 1.1532106399536133,
      "learning_rate": 4.615e-06,
      "loss": 0.6579,
      "step": 925
    },
    {
      "epoch": 0.05185351103147049,
      "grad_norm": 1.1137820482254028,
      "learning_rate": 4.62e-06,
      "loss": 0.4266,
      "step": 926
    },
    {
      "epoch": 0.05190950834359951,
      "grad_norm": 1.6235989332199097,
      "learning_rate": 4.625e-06,
      "loss": 0.4395,
      "step": 927
    },
    {
      "epoch": 0.05196550565572852,
      "grad_norm": 1.167560338973999,
      "learning_rate": 4.6300000000000006e-06,
      "loss": 0.3794,
      "step": 928
    },
    {
      "epoch": 0.05202150296785754,
      "grad_norm": 1.2593872547149658,
      "learning_rate": 4.6350000000000005e-06,
      "loss": 0.4781,
      "step": 929
    },
    {
      "epoch": 0.05207750027998656,
      "grad_norm": 1.4556719064712524,
      "learning_rate": 4.64e-06,
      "loss": 0.5134,
      "step": 930
    },
    {
      "epoch": 0.05213349759211558,
      "grad_norm": 1.1882364749908447,
      "learning_rate": 4.645e-06,
      "loss": 0.4005,
      "step": 931
    },
    {
      "epoch": 0.0521894949042446,
      "grad_norm": 1.14823579788208,
      "learning_rate": 4.65e-06,
      "loss": 0.4556,
      "step": 932
    },
    {
      "epoch": 0.05224549221637361,
      "grad_norm": 1.373916506767273,
      "learning_rate": 4.655e-06,
      "loss": 0.4299,
      "step": 933
    },
    {
      "epoch": 0.05230148952850263,
      "grad_norm": 1.1341737508773804,
      "learning_rate": 4.66e-06,
      "loss": 0.4239,
      "step": 934
    },
    {
      "epoch": 0.05235748684063165,
      "grad_norm": 1.2210662364959717,
      "learning_rate": 4.665e-06,
      "loss": 0.5539,
      "step": 935
    },
    {
      "epoch": 0.05241348415276067,
      "grad_norm": 1.1791493892669678,
      "learning_rate": 4.67e-06,
      "loss": 0.4105,
      "step": 936
    },
    {
      "epoch": 0.05246948146488969,
      "grad_norm": 1.2142149209976196,
      "learning_rate": 4.675e-06,
      "loss": 0.4108,
      "step": 937
    },
    {
      "epoch": 0.0525254787770187,
      "grad_norm": 1.3496472835540771,
      "learning_rate": 4.68e-06,
      "loss": 0.396,
      "step": 938
    },
    {
      "epoch": 0.05258147608914772,
      "grad_norm": 1.1289809942245483,
      "learning_rate": 4.685000000000001e-06,
      "loss": 0.3934,
      "step": 939
    },
    {
      "epoch": 0.05263747340127674,
      "grad_norm": 1.210899829864502,
      "learning_rate": 4.69e-06,
      "loss": 0.4609,
      "step": 940
    },
    {
      "epoch": 0.05269347071340576,
      "grad_norm": 1.090699553489685,
      "learning_rate": 4.695e-06,
      "loss": 0.5039,
      "step": 941
    },
    {
      "epoch": 0.052749468025534776,
      "grad_norm": 1.3170802593231201,
      "learning_rate": 4.7e-06,
      "loss": 0.498,
      "step": 942
    },
    {
      "epoch": 0.052805465337663796,
      "grad_norm": 1.2018908262252808,
      "learning_rate": 4.705000000000001e-06,
      "loss": 0.393,
      "step": 943
    },
    {
      "epoch": 0.05286146264979281,
      "grad_norm": 1.3873302936553955,
      "learning_rate": 4.710000000000001e-06,
      "loss": 0.478,
      "step": 944
    },
    {
      "epoch": 0.05291745996192183,
      "grad_norm": 1.3100378513336182,
      "learning_rate": 4.715e-06,
      "loss": 0.4197,
      "step": 945
    },
    {
      "epoch": 0.052973457274050846,
      "grad_norm": 1.6671725511550903,
      "learning_rate": 4.72e-06,
      "loss": 0.6221,
      "step": 946
    },
    {
      "epoch": 0.053029454586179865,
      "grad_norm": 1.6387559175491333,
      "learning_rate": 4.7250000000000005e-06,
      "loss": 0.5495,
      "step": 947
    },
    {
      "epoch": 0.053085451898308884,
      "grad_norm": 1.2692862749099731,
      "learning_rate": 4.7300000000000005e-06,
      "loss": 0.4587,
      "step": 948
    },
    {
      "epoch": 0.0531414492104379,
      "grad_norm": 1.299742341041565,
      "learning_rate": 4.735e-06,
      "loss": 0.4027,
      "step": 949
    },
    {
      "epoch": 0.053197446522566916,
      "grad_norm": 1.3480663299560547,
      "learning_rate": 4.74e-06,
      "loss": 0.4369,
      "step": 950
    },
    {
      "epoch": 0.053253443834695935,
      "grad_norm": 1.2166800498962402,
      "learning_rate": 4.745e-06,
      "loss": 0.5011,
      "step": 951
    },
    {
      "epoch": 0.053309441146824954,
      "grad_norm": 1.2832262516021729,
      "learning_rate": 4.75e-06,
      "loss": 0.402,
      "step": 952
    },
    {
      "epoch": 0.05336543845895397,
      "grad_norm": 1.4792896509170532,
      "learning_rate": 4.755e-06,
      "loss": 0.5747,
      "step": 953
    },
    {
      "epoch": 0.053421435771082985,
      "grad_norm": 1.568776249885559,
      "learning_rate": 4.76e-06,
      "loss": 0.4724,
      "step": 954
    },
    {
      "epoch": 0.053477433083212005,
      "grad_norm": 1.1982817649841309,
      "learning_rate": 4.765e-06,
      "loss": 0.5061,
      "step": 955
    },
    {
      "epoch": 0.053533430395341024,
      "grad_norm": 1.3185970783233643,
      "learning_rate": 4.77e-06,
      "loss": 0.5286,
      "step": 956
    },
    {
      "epoch": 0.05358942770747004,
      "grad_norm": 1.147926688194275,
      "learning_rate": 4.775e-06,
      "loss": 0.3612,
      "step": 957
    },
    {
      "epoch": 0.05364542501959906,
      "grad_norm": 1.26372230052948,
      "learning_rate": 4.780000000000001e-06,
      "loss": 0.5493,
      "step": 958
    },
    {
      "epoch": 0.053701422331728074,
      "grad_norm": 1.3605045080184937,
      "learning_rate": 4.785e-06,
      "loss": 0.6715,
      "step": 959
    },
    {
      "epoch": 0.05375741964385709,
      "grad_norm": 1.3014461994171143,
      "learning_rate": 4.79e-06,
      "loss": 0.4488,
      "step": 960
    },
    {
      "epoch": 0.05381341695598611,
      "grad_norm": 1.37398362159729,
      "learning_rate": 4.795e-06,
      "loss": 0.4222,
      "step": 961
    },
    {
      "epoch": 0.05386941426811513,
      "grad_norm": 1.230643391609192,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.4864,
      "step": 962
    },
    {
      "epoch": 0.05392541158024415,
      "grad_norm": 1.095284104347229,
      "learning_rate": 4.805000000000001e-06,
      "loss": 0.5956,
      "step": 963
    },
    {
      "epoch": 0.05398140889237316,
      "grad_norm": 1.1946420669555664,
      "learning_rate": 4.81e-06,
      "loss": 0.4388,
      "step": 964
    },
    {
      "epoch": 0.05403740620450218,
      "grad_norm": 1.0814542770385742,
      "learning_rate": 4.8150000000000005e-06,
      "loss": 0.4001,
      "step": 965
    },
    {
      "epoch": 0.0540934035166312,
      "grad_norm": 1.1272335052490234,
      "learning_rate": 4.8200000000000004e-06,
      "loss": 0.4352,
      "step": 966
    },
    {
      "epoch": 0.05414940082876022,
      "grad_norm": 1.2352651357650757,
      "learning_rate": 4.825e-06,
      "loss": 0.4212,
      "step": 967
    },
    {
      "epoch": 0.05420539814088924,
      "grad_norm": 1.3018606901168823,
      "learning_rate": 4.83e-06,
      "loss": 0.429,
      "step": 968
    },
    {
      "epoch": 0.05426139545301825,
      "grad_norm": 1.3863848447799683,
      "learning_rate": 4.835e-06,
      "loss": 0.5049,
      "step": 969
    },
    {
      "epoch": 0.05431739276514727,
      "grad_norm": 1.1036901473999023,
      "learning_rate": 4.84e-06,
      "loss": 0.4213,
      "step": 970
    },
    {
      "epoch": 0.05437339007727629,
      "grad_norm": 1.420832633972168,
      "learning_rate": 4.845e-06,
      "loss": 0.42,
      "step": 971
    },
    {
      "epoch": 0.05442938738940531,
      "grad_norm": 1.2023344039916992,
      "learning_rate": 4.85e-06,
      "loss": 0.5156,
      "step": 972
    },
    {
      "epoch": 0.05448538470153433,
      "grad_norm": 1.184357762336731,
      "learning_rate": 4.855e-06,
      "loss": 0.4431,
      "step": 973
    },
    {
      "epoch": 0.05454138201366335,
      "grad_norm": 1.2339346408843994,
      "learning_rate": 4.86e-06,
      "loss": 0.4456,
      "step": 974
    },
    {
      "epoch": 0.05459737932579236,
      "grad_norm": 1.1052863597869873,
      "learning_rate": 4.865e-06,
      "loss": 0.3762,
      "step": 975
    },
    {
      "epoch": 0.05465337663792138,
      "grad_norm": 1.15488600730896,
      "learning_rate": 4.87e-06,
      "loss": 0.4208,
      "step": 976
    },
    {
      "epoch": 0.0547093739500504,
      "grad_norm": 1.5858561992645264,
      "learning_rate": 4.875000000000001e-06,
      "loss": 0.4824,
      "step": 977
    },
    {
      "epoch": 0.05476537126217942,
      "grad_norm": 1.2476688623428345,
      "learning_rate": 4.880000000000001e-06,
      "loss": 0.6081,
      "step": 978
    },
    {
      "epoch": 0.054821368574308436,
      "grad_norm": 0.9850094318389893,
      "learning_rate": 4.885e-06,
      "loss": 0.4135,
      "step": 979
    },
    {
      "epoch": 0.05487736588643745,
      "grad_norm": 1.3759442567825317,
      "learning_rate": 4.89e-06,
      "loss": 0.5223,
      "step": 980
    },
    {
      "epoch": 0.05493336319856647,
      "grad_norm": 1.2599742412567139,
      "learning_rate": 4.8950000000000006e-06,
      "loss": 0.5832,
      "step": 981
    },
    {
      "epoch": 0.05498936051069549,
      "grad_norm": 2.0648694038391113,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 0.8202,
      "step": 982
    },
    {
      "epoch": 0.055045357822824506,
      "grad_norm": 1.2609869241714478,
      "learning_rate": 4.9050000000000005e-06,
      "loss": 0.4878,
      "step": 983
    },
    {
      "epoch": 0.055101355134953525,
      "grad_norm": 1.1653976440429688,
      "learning_rate": 4.9100000000000004e-06,
      "loss": 0.4716,
      "step": 984
    },
    {
      "epoch": 0.05515735244708254,
      "grad_norm": 1.4410059452056885,
      "learning_rate": 4.915e-06,
      "loss": 0.5038,
      "step": 985
    },
    {
      "epoch": 0.05521334975921156,
      "grad_norm": 1.3019509315490723,
      "learning_rate": 4.92e-06,
      "loss": 0.417,
      "step": 986
    },
    {
      "epoch": 0.055269347071340576,
      "grad_norm": 1.4015498161315918,
      "learning_rate": 4.925e-06,
      "loss": 0.6574,
      "step": 987
    },
    {
      "epoch": 0.055325344383469595,
      "grad_norm": 1.9603328704833984,
      "learning_rate": 4.93e-06,
      "loss": 0.5267,
      "step": 988
    },
    {
      "epoch": 0.055381341695598614,
      "grad_norm": 1.3965656757354736,
      "learning_rate": 4.935e-06,
      "loss": 0.4231,
      "step": 989
    },
    {
      "epoch": 0.055437339007727626,
      "grad_norm": 1.1022210121154785,
      "learning_rate": 4.94e-06,
      "loss": 0.4136,
      "step": 990
    },
    {
      "epoch": 0.055493336319856645,
      "grad_norm": 1.0762064456939697,
      "learning_rate": 4.945e-06,
      "loss": 0.4043,
      "step": 991
    },
    {
      "epoch": 0.055549333631985665,
      "grad_norm": 1.375928282737732,
      "learning_rate": 4.950000000000001e-06,
      "loss": 0.4881,
      "step": 992
    },
    {
      "epoch": 0.055605330944114684,
      "grad_norm": 1.385292649269104,
      "learning_rate": 4.955e-06,
      "loss": 0.4528,
      "step": 993
    },
    {
      "epoch": 0.0556613282562437,
      "grad_norm": 1.17954683303833,
      "learning_rate": 4.96e-06,
      "loss": 0.4497,
      "step": 994
    },
    {
      "epoch": 0.055717325568372715,
      "grad_norm": 1.412698745727539,
      "learning_rate": 4.965e-06,
      "loss": 0.4283,
      "step": 995
    },
    {
      "epoch": 0.055773322880501734,
      "grad_norm": 1.191758155822754,
      "learning_rate": 4.970000000000001e-06,
      "loss": 0.3965,
      "step": 996
    },
    {
      "epoch": 0.05582932019263075,
      "grad_norm": 6.146461486816406,
      "learning_rate": 4.975000000000001e-06,
      "loss": 0.5264,
      "step": 997
    },
    {
      "epoch": 0.05588531750475977,
      "grad_norm": 1.2692173719406128,
      "learning_rate": 4.98e-06,
      "loss": 0.4529,
      "step": 998
    },
    {
      "epoch": 0.05594131481688879,
      "grad_norm": 1.5443811416625977,
      "learning_rate": 4.985e-06,
      "loss": 0.4728,
      "step": 999
    },
    {
      "epoch": 0.055997312129017804,
      "grad_norm": 1.2640265226364136,
      "learning_rate": 4.9900000000000005e-06,
      "loss": 0.5565,
      "step": 1000
    },
    {
      "epoch": 0.05605330944114682,
      "grad_norm": 1.2944355010986328,
      "learning_rate": 4.9950000000000005e-06,
      "loss": 0.5138,
      "step": 1001
    },
    {
      "epoch": 0.05610930675327584,
      "grad_norm": 1.1413021087646484,
      "learning_rate": 5e-06,
      "loss": 0.4785,
      "step": 1002
    },
    {
      "epoch": 0.05616530406540486,
      "grad_norm": 1.3369749784469604,
      "learning_rate": 5.005e-06,
      "loss": 0.4315,
      "step": 1003
    },
    {
      "epoch": 0.05622130137753388,
      "grad_norm": 1.085232138633728,
      "learning_rate": 5.01e-06,
      "loss": 0.3798,
      "step": 1004
    },
    {
      "epoch": 0.0562772986896629,
      "grad_norm": 1.620424747467041,
      "learning_rate": 5.015e-06,
      "loss": 0.4397,
      "step": 1005
    },
    {
      "epoch": 0.05633329600179191,
      "grad_norm": 1.3571701049804688,
      "learning_rate": 5.02e-06,
      "loss": 0.5498,
      "step": 1006
    },
    {
      "epoch": 0.05638929331392093,
      "grad_norm": 1.3521087169647217,
      "learning_rate": 5.025e-06,
      "loss": 0.3857,
      "step": 1007
    },
    {
      "epoch": 0.05644529062604995,
      "grad_norm": 1.1394175291061401,
      "learning_rate": 5.03e-06,
      "loss": 0.5012,
      "step": 1008
    },
    {
      "epoch": 0.05650128793817897,
      "grad_norm": 1.5971336364746094,
      "learning_rate": 5.035e-06,
      "loss": 0.3897,
      "step": 1009
    },
    {
      "epoch": 0.05655728525030799,
      "grad_norm": 1.226791262626648,
      "learning_rate": 5.04e-06,
      "loss": 0.4669,
      "step": 1010
    },
    {
      "epoch": 0.056613282562437,
      "grad_norm": 1.140771746635437,
      "learning_rate": 5.045000000000001e-06,
      "loss": 0.4582,
      "step": 1011
    },
    {
      "epoch": 0.05666927987456602,
      "grad_norm": 1.2240629196166992,
      "learning_rate": 5.050000000000001e-06,
      "loss": 0.5498,
      "step": 1012
    },
    {
      "epoch": 0.05672527718669504,
      "grad_norm": 1.2744507789611816,
      "learning_rate": 5.055e-06,
      "loss": 0.5201,
      "step": 1013
    },
    {
      "epoch": 0.05678127449882406,
      "grad_norm": 1.2416425943374634,
      "learning_rate": 5.06e-06,
      "loss": 0.455,
      "step": 1014
    },
    {
      "epoch": 0.05683727181095308,
      "grad_norm": 1.136209487915039,
      "learning_rate": 5.065000000000001e-06,
      "loss": 0.4787,
      "step": 1015
    },
    {
      "epoch": 0.05689326912308209,
      "grad_norm": 1.156378149986267,
      "learning_rate": 5.070000000000001e-06,
      "loss": 0.4328,
      "step": 1016
    },
    {
      "epoch": 0.05694926643521111,
      "grad_norm": 1.2695443630218506,
      "learning_rate": 5.0750000000000005e-06,
      "loss": 0.4819,
      "step": 1017
    },
    {
      "epoch": 0.05700526374734013,
      "grad_norm": 1.191765308380127,
      "learning_rate": 5.08e-06,
      "loss": 0.3793,
      "step": 1018
    },
    {
      "epoch": 0.05706126105946915,
      "grad_norm": 1.2647852897644043,
      "learning_rate": 5.0850000000000004e-06,
      "loss": 0.3896,
      "step": 1019
    },
    {
      "epoch": 0.057117258371598166,
      "grad_norm": 1.2763571739196777,
      "learning_rate": 5.09e-06,
      "loss": 0.6245,
      "step": 1020
    },
    {
      "epoch": 0.05717325568372718,
      "grad_norm": 1.239418864250183,
      "learning_rate": 5.095e-06,
      "loss": 0.4452,
      "step": 1021
    },
    {
      "epoch": 0.0572292529958562,
      "grad_norm": 1.3589504957199097,
      "learning_rate": 5.1e-06,
      "loss": 0.514,
      "step": 1022
    },
    {
      "epoch": 0.05728525030798522,
      "grad_norm": 1.1839314699172974,
      "learning_rate": 5.105e-06,
      "loss": 0.4385,
      "step": 1023
    },
    {
      "epoch": 0.057341247620114236,
      "grad_norm": 1.2520166635513306,
      "learning_rate": 5.11e-06,
      "loss": 0.5166,
      "step": 1024
    },
    {
      "epoch": 0.057397244932243255,
      "grad_norm": 1.7305189371109009,
      "learning_rate": 5.115e-06,
      "loss": 0.4474,
      "step": 1025
    },
    {
      "epoch": 0.05745324224437227,
      "grad_norm": 1.2060898542404175,
      "learning_rate": 5.12e-06,
      "loss": 0.509,
      "step": 1026
    },
    {
      "epoch": 0.057509239556501286,
      "grad_norm": 1.473137617111206,
      "learning_rate": 5.125e-06,
      "loss": 0.4312,
      "step": 1027
    },
    {
      "epoch": 0.057565236868630305,
      "grad_norm": 1.0966969728469849,
      "learning_rate": 5.13e-06,
      "loss": 0.4443,
      "step": 1028
    },
    {
      "epoch": 0.057621234180759325,
      "grad_norm": 1.2823864221572876,
      "learning_rate": 5.135e-06,
      "loss": 0.533,
      "step": 1029
    },
    {
      "epoch": 0.057677231492888344,
      "grad_norm": 1.2967363595962524,
      "learning_rate": 5.140000000000001e-06,
      "loss": 0.4279,
      "step": 1030
    },
    {
      "epoch": 0.057733228805017356,
      "grad_norm": 1.336930751800537,
      "learning_rate": 5.145000000000001e-06,
      "loss": 0.5224,
      "step": 1031
    },
    {
      "epoch": 0.057789226117146375,
      "grad_norm": 1.5669735670089722,
      "learning_rate": 5.15e-06,
      "loss": 0.5194,
      "step": 1032
    },
    {
      "epoch": 0.057845223429275394,
      "grad_norm": 1.1038568019866943,
      "learning_rate": 5.155e-06,
      "loss": 0.4,
      "step": 1033
    },
    {
      "epoch": 0.05790122074140441,
      "grad_norm": 1.100191354751587,
      "learning_rate": 5.1600000000000006e-06,
      "loss": 0.4871,
      "step": 1034
    },
    {
      "epoch": 0.05795721805353343,
      "grad_norm": 1.2605558633804321,
      "learning_rate": 5.1650000000000005e-06,
      "loss": 0.4253,
      "step": 1035
    },
    {
      "epoch": 0.05801321536566245,
      "grad_norm": 1.4248777627944946,
      "learning_rate": 5.1700000000000005e-06,
      "loss": 0.3997,
      "step": 1036
    },
    {
      "epoch": 0.058069212677791464,
      "grad_norm": 1.2265772819519043,
      "learning_rate": 5.175e-06,
      "loss": 0.4736,
      "step": 1037
    },
    {
      "epoch": 0.05812520998992048,
      "grad_norm": 1.3616083860397339,
      "learning_rate": 5.18e-06,
      "loss": 0.4686,
      "step": 1038
    },
    {
      "epoch": 0.0581812073020495,
      "grad_norm": 1.1219232082366943,
      "learning_rate": 5.185e-06,
      "loss": 0.3539,
      "step": 1039
    },
    {
      "epoch": 0.05823720461417852,
      "grad_norm": 1.4267109632492065,
      "learning_rate": 5.19e-06,
      "loss": 0.5069,
      "step": 1040
    },
    {
      "epoch": 0.05829320192630754,
      "grad_norm": 1.2390908002853394,
      "learning_rate": 5.195e-06,
      "loss": 0.5129,
      "step": 1041
    },
    {
      "epoch": 0.05834919923843655,
      "grad_norm": 1.2516577243804932,
      "learning_rate": 5.2e-06,
      "loss": 0.5162,
      "step": 1042
    },
    {
      "epoch": 0.05840519655056557,
      "grad_norm": 1.876985788345337,
      "learning_rate": 5.205e-06,
      "loss": 0.3696,
      "step": 1043
    },
    {
      "epoch": 0.05846119386269459,
      "grad_norm": 1.3517183065414429,
      "learning_rate": 5.21e-06,
      "loss": 0.5062,
      "step": 1044
    },
    {
      "epoch": 0.05851719117482361,
      "grad_norm": 1.160207748413086,
      "learning_rate": 5.215e-06,
      "loss": 0.4457,
      "step": 1045
    },
    {
      "epoch": 0.05857318848695263,
      "grad_norm": 1.3615148067474365,
      "learning_rate": 5.220000000000001e-06,
      "loss": 0.5174,
      "step": 1046
    },
    {
      "epoch": 0.05862918579908164,
      "grad_norm": 1.04086434841156,
      "learning_rate": 5.225e-06,
      "loss": 0.4562,
      "step": 1047
    },
    {
      "epoch": 0.05868518311121066,
      "grad_norm": 1.3939971923828125,
      "learning_rate": 5.23e-06,
      "loss": 0.4531,
      "step": 1048
    },
    {
      "epoch": 0.05874118042333968,
      "grad_norm": 1.3616175651550293,
      "learning_rate": 5.235000000000001e-06,
      "loss": 0.478,
      "step": 1049
    },
    {
      "epoch": 0.0587971777354687,
      "grad_norm": 1.2056399583816528,
      "learning_rate": 5.240000000000001e-06,
      "loss": 0.5392,
      "step": 1050
    },
    {
      "epoch": 0.05885317504759772,
      "grad_norm": 1.2128673791885376,
      "learning_rate": 5.245e-06,
      "loss": 0.3881,
      "step": 1051
    },
    {
      "epoch": 0.05890917235972673,
      "grad_norm": 1.4392180442810059,
      "learning_rate": 5.25e-06,
      "loss": 0.4251,
      "step": 1052
    },
    {
      "epoch": 0.05896516967185575,
      "grad_norm": 1.199028730392456,
      "learning_rate": 5.2550000000000005e-06,
      "loss": 0.3948,
      "step": 1053
    },
    {
      "epoch": 0.05902116698398477,
      "grad_norm": 1.0078922510147095,
      "learning_rate": 5.2600000000000005e-06,
      "loss": 0.3426,
      "step": 1054
    },
    {
      "epoch": 0.05907716429611379,
      "grad_norm": 1.1445984840393066,
      "learning_rate": 5.265e-06,
      "loss": 0.3774,
      "step": 1055
    },
    {
      "epoch": 0.05913316160824281,
      "grad_norm": 1.3585809469223022,
      "learning_rate": 5.2699999999999995e-06,
      "loss": 0.6656,
      "step": 1056
    },
    {
      "epoch": 0.05918915892037182,
      "grad_norm": 1.065910816192627,
      "learning_rate": 5.275e-06,
      "loss": 0.4658,
      "step": 1057
    },
    {
      "epoch": 0.05924515623250084,
      "grad_norm": 1.2044261693954468,
      "learning_rate": 5.28e-06,
      "loss": 0.5031,
      "step": 1058
    },
    {
      "epoch": 0.05930115354462986,
      "grad_norm": 1.6254397630691528,
      "learning_rate": 5.285e-06,
      "loss": 0.4542,
      "step": 1059
    },
    {
      "epoch": 0.05935715085675888,
      "grad_norm": 1.1306949853897095,
      "learning_rate": 5.29e-06,
      "loss": 0.4671,
      "step": 1060
    },
    {
      "epoch": 0.059413148168887896,
      "grad_norm": 1.2099978923797607,
      "learning_rate": 5.295e-06,
      "loss": 0.4892,
      "step": 1061
    },
    {
      "epoch": 0.05946914548101691,
      "grad_norm": 1.401309609413147,
      "learning_rate": 5.3e-06,
      "loss": 0.5684,
      "step": 1062
    },
    {
      "epoch": 0.05952514279314593,
      "grad_norm": 1.8093961477279663,
      "learning_rate": 5.305e-06,
      "loss": 0.4251,
      "step": 1063
    },
    {
      "epoch": 0.059581140105274946,
      "grad_norm": 1.5952280759811401,
      "learning_rate": 5.31e-06,
      "loss": 0.5887,
      "step": 1064
    },
    {
      "epoch": 0.059637137417403965,
      "grad_norm": 1.223723292350769,
      "learning_rate": 5.315000000000001e-06,
      "loss": 0.3764,
      "step": 1065
    },
    {
      "epoch": 0.059693134729532984,
      "grad_norm": 1.25835120677948,
      "learning_rate": 5.32e-06,
      "loss": 0.4321,
      "step": 1066
    },
    {
      "epoch": 0.059749132041662004,
      "grad_norm": 1.1497814655303955,
      "learning_rate": 5.325e-06,
      "loss": 0.4415,
      "step": 1067
    },
    {
      "epoch": 0.059805129353791016,
      "grad_norm": 1.0686076879501343,
      "learning_rate": 5.330000000000001e-06,
      "loss": 0.3571,
      "step": 1068
    },
    {
      "epoch": 0.059861126665920035,
      "grad_norm": 1.2597800493240356,
      "learning_rate": 5.335000000000001e-06,
      "loss": 0.4686,
      "step": 1069
    },
    {
      "epoch": 0.059917123978049054,
      "grad_norm": 1.2790300846099854,
      "learning_rate": 5.3400000000000005e-06,
      "loss": 0.4832,
      "step": 1070
    },
    {
      "epoch": 0.05997312129017807,
      "grad_norm": 1.1553374528884888,
      "learning_rate": 5.345e-06,
      "loss": 0.5143,
      "step": 1071
    },
    {
      "epoch": 0.06002911860230709,
      "grad_norm": 1.1005035638809204,
      "learning_rate": 5.3500000000000004e-06,
      "loss": 0.4896,
      "step": 1072
    },
    {
      "epoch": 0.060085115914436105,
      "grad_norm": 1.3469942808151245,
      "learning_rate": 5.355e-06,
      "loss": 0.4591,
      "step": 1073
    },
    {
      "epoch": 0.060141113226565124,
      "grad_norm": 1.2604517936706543,
      "learning_rate": 5.36e-06,
      "loss": 0.5521,
      "step": 1074
    },
    {
      "epoch": 0.06019711053869414,
      "grad_norm": 1.243295431137085,
      "learning_rate": 5.365e-06,
      "loss": 0.6024,
      "step": 1075
    },
    {
      "epoch": 0.06025310785082316,
      "grad_norm": 1.3653740882873535,
      "learning_rate": 5.37e-06,
      "loss": 0.4207,
      "step": 1076
    },
    {
      "epoch": 0.06030910516295218,
      "grad_norm": 1.1624484062194824,
      "learning_rate": 5.375e-06,
      "loss": 0.4241,
      "step": 1077
    },
    {
      "epoch": 0.060365102475081193,
      "grad_norm": 1.260948657989502,
      "learning_rate": 5.38e-06,
      "loss": 0.4985,
      "step": 1078
    },
    {
      "epoch": 0.06042109978721021,
      "grad_norm": 1.099338173866272,
      "learning_rate": 5.385e-06,
      "loss": 0.4063,
      "step": 1079
    },
    {
      "epoch": 0.06047709709933923,
      "grad_norm": 1.4310381412506104,
      "learning_rate": 5.390000000000001e-06,
      "loss": 0.5739,
      "step": 1080
    },
    {
      "epoch": 0.06053309441146825,
      "grad_norm": 1.2879993915557861,
      "learning_rate": 5.395e-06,
      "loss": 0.4956,
      "step": 1081
    },
    {
      "epoch": 0.06058909172359727,
      "grad_norm": 1.1941678524017334,
      "learning_rate": 5.4e-06,
      "loss": 0.3758,
      "step": 1082
    },
    {
      "epoch": 0.06064508903572628,
      "grad_norm": 1.1191668510437012,
      "learning_rate": 5.405e-06,
      "loss": 0.5986,
      "step": 1083
    },
    {
      "epoch": 0.0607010863478553,
      "grad_norm": 1.4496634006500244,
      "learning_rate": 5.410000000000001e-06,
      "loss": 0.4977,
      "step": 1084
    },
    {
      "epoch": 0.06075708365998432,
      "grad_norm": 1.4725685119628906,
      "learning_rate": 5.415e-06,
      "loss": 0.5158,
      "step": 1085
    },
    {
      "epoch": 0.06081308097211334,
      "grad_norm": 1.0966806411743164,
      "learning_rate": 5.42e-06,
      "loss": 0.4611,
      "step": 1086
    },
    {
      "epoch": 0.06086907828424236,
      "grad_norm": 1.06184720993042,
      "learning_rate": 5.4250000000000006e-06,
      "loss": 0.4319,
      "step": 1087
    },
    {
      "epoch": 0.06092507559637137,
      "grad_norm": 1.874305248260498,
      "learning_rate": 5.4300000000000005e-06,
      "loss": 0.5097,
      "step": 1088
    },
    {
      "epoch": 0.06098107290850039,
      "grad_norm": 2.8478968143463135,
      "learning_rate": 5.4350000000000005e-06,
      "loss": 0.3575,
      "step": 1089
    },
    {
      "epoch": 0.06103707022062941,
      "grad_norm": 1.3469611406326294,
      "learning_rate": 5.44e-06,
      "loss": 0.428,
      "step": 1090
    },
    {
      "epoch": 0.06109306753275843,
      "grad_norm": 1.0713883638381958,
      "learning_rate": 5.445e-06,
      "loss": 0.3892,
      "step": 1091
    },
    {
      "epoch": 0.06114906484488745,
      "grad_norm": 1.3371243476867676,
      "learning_rate": 5.45e-06,
      "loss": 0.577,
      "step": 1092
    },
    {
      "epoch": 0.06120506215701646,
      "grad_norm": 1.6126753091812134,
      "learning_rate": 5.455e-06,
      "loss": 0.4554,
      "step": 1093
    },
    {
      "epoch": 0.06126105946914548,
      "grad_norm": 1.1165636777877808,
      "learning_rate": 5.46e-06,
      "loss": 0.358,
      "step": 1094
    },
    {
      "epoch": 0.0613170567812745,
      "grad_norm": 1.371865153312683,
      "learning_rate": 5.465e-06,
      "loss": 0.4045,
      "step": 1095
    },
    {
      "epoch": 0.06137305409340352,
      "grad_norm": 1.3662270307540894,
      "learning_rate": 5.47e-06,
      "loss": 0.3902,
      "step": 1096
    },
    {
      "epoch": 0.061429051405532537,
      "grad_norm": 1.5756611824035645,
      "learning_rate": 5.475e-06,
      "loss": 0.8939,
      "step": 1097
    },
    {
      "epoch": 0.061485048717661556,
      "grad_norm": 1.336142897605896,
      "learning_rate": 5.48e-06,
      "loss": 0.549,
      "step": 1098
    },
    {
      "epoch": 0.06154104602979057,
      "grad_norm": 1.1127017736434937,
      "learning_rate": 5.485000000000001e-06,
      "loss": 0.462,
      "step": 1099
    },
    {
      "epoch": 0.06159704334191959,
      "grad_norm": 1.4991849660873413,
      "learning_rate": 5.49e-06,
      "loss": 0.4478,
      "step": 1100
    },
    {
      "epoch": 0.061653040654048606,
      "grad_norm": 1.2078429460525513,
      "learning_rate": 5.495e-06,
      "loss": 0.3859,
      "step": 1101
    },
    {
      "epoch": 0.061709037966177625,
      "grad_norm": 1.0736446380615234,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.3938,
      "step": 1102
    },
    {
      "epoch": 0.061765035278306644,
      "grad_norm": 1.1046690940856934,
      "learning_rate": 5.505000000000001e-06,
      "loss": 0.335,
      "step": 1103
    },
    {
      "epoch": 0.06182103259043566,
      "grad_norm": 1.3569049835205078,
      "learning_rate": 5.510000000000001e-06,
      "loss": 0.3924,
      "step": 1104
    },
    {
      "epoch": 0.061877029902564676,
      "grad_norm": 2.3091189861297607,
      "learning_rate": 5.515e-06,
      "loss": 0.5528,
      "step": 1105
    },
    {
      "epoch": 0.061933027214693695,
      "grad_norm": 1.2597846984863281,
      "learning_rate": 5.5200000000000005e-06,
      "loss": 0.4925,
      "step": 1106
    },
    {
      "epoch": 0.061989024526822714,
      "grad_norm": 1.195900321006775,
      "learning_rate": 5.5250000000000005e-06,
      "loss": 0.5321,
      "step": 1107
    },
    {
      "epoch": 0.06204502183895173,
      "grad_norm": 1.2691329717636108,
      "learning_rate": 5.53e-06,
      "loss": 0.4351,
      "step": 1108
    },
    {
      "epoch": 0.062101019151080745,
      "grad_norm": 1.3498635292053223,
      "learning_rate": 5.535e-06,
      "loss": 0.4116,
      "step": 1109
    },
    {
      "epoch": 0.062157016463209765,
      "grad_norm": 1.189966082572937,
      "learning_rate": 5.54e-06,
      "loss": 0.4857,
      "step": 1110
    },
    {
      "epoch": 0.062213013775338784,
      "grad_norm": 1.262734293937683,
      "learning_rate": 5.545e-06,
      "loss": 0.6124,
      "step": 1111
    },
    {
      "epoch": 0.0622690110874678,
      "grad_norm": 1.9112929105758667,
      "learning_rate": 5.55e-06,
      "loss": 0.4752,
      "step": 1112
    },
    {
      "epoch": 0.06232500839959682,
      "grad_norm": 1.508002519607544,
      "learning_rate": 5.555e-06,
      "loss": 0.4645,
      "step": 1113
    },
    {
      "epoch": 0.062381005711725834,
      "grad_norm": 1.417464017868042,
      "learning_rate": 5.56e-06,
      "loss": 0.449,
      "step": 1114
    },
    {
      "epoch": 0.06243700302385485,
      "grad_norm": 1.1880978345870972,
      "learning_rate": 5.565e-06,
      "loss": 0.4766,
      "step": 1115
    },
    {
      "epoch": 0.06249300033598387,
      "grad_norm": 1.3594896793365479,
      "learning_rate": 5.57e-06,
      "loss": 0.4782,
      "step": 1116
    },
    {
      "epoch": 0.06254899764811289,
      "grad_norm": 1.158437967300415,
      "learning_rate": 5.575e-06,
      "loss": 0.5021,
      "step": 1117
    },
    {
      "epoch": 0.06260499496024191,
      "grad_norm": 1.074735164642334,
      "learning_rate": 5.580000000000001e-06,
      "loss": 0.4648,
      "step": 1118
    },
    {
      "epoch": 0.06266099227237093,
      "grad_norm": 1.4043564796447754,
      "learning_rate": 5.585e-06,
      "loss": 0.5121,
      "step": 1119
    },
    {
      "epoch": 0.06271698958449995,
      "grad_norm": 1.1679229736328125,
      "learning_rate": 5.59e-06,
      "loss": 0.3579,
      "step": 1120
    },
    {
      "epoch": 0.06277298689662897,
      "grad_norm": 1.1285709142684937,
      "learning_rate": 5.595000000000001e-06,
      "loss": 0.2937,
      "step": 1121
    },
    {
      "epoch": 0.06282898420875797,
      "grad_norm": 1.2264790534973145,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.5157,
      "step": 1122
    },
    {
      "epoch": 0.06288498152088699,
      "grad_norm": 1.1236133575439453,
      "learning_rate": 5.6050000000000005e-06,
      "loss": 0.4314,
      "step": 1123
    },
    {
      "epoch": 0.06294097883301601,
      "grad_norm": 1.0496784448623657,
      "learning_rate": 5.61e-06,
      "loss": 0.3789,
      "step": 1124
    },
    {
      "epoch": 0.06299697614514503,
      "grad_norm": 1.0616122484207153,
      "learning_rate": 5.6150000000000005e-06,
      "loss": 0.4315,
      "step": 1125
    },
    {
      "epoch": 0.06305297345727405,
      "grad_norm": 1.397759199142456,
      "learning_rate": 5.62e-06,
      "loss": 0.7762,
      "step": 1126
    },
    {
      "epoch": 0.06310897076940307,
      "grad_norm": 1.163167119026184,
      "learning_rate": 5.625e-06,
      "loss": 0.4815,
      "step": 1127
    },
    {
      "epoch": 0.06316496808153209,
      "grad_norm": 1.3902819156646729,
      "learning_rate": 5.63e-06,
      "loss": 0.5662,
      "step": 1128
    },
    {
      "epoch": 0.06322096539366111,
      "grad_norm": 1.2462685108184814,
      "learning_rate": 5.635e-06,
      "loss": 0.458,
      "step": 1129
    },
    {
      "epoch": 0.06327696270579013,
      "grad_norm": 1.4905333518981934,
      "learning_rate": 5.64e-06,
      "loss": 0.4204,
      "step": 1130
    },
    {
      "epoch": 0.06333296001791915,
      "grad_norm": 1.259095549583435,
      "learning_rate": 5.645e-06,
      "loss": 0.4723,
      "step": 1131
    },
    {
      "epoch": 0.06338895733004815,
      "grad_norm": 1.1537303924560547,
      "learning_rate": 5.65e-06,
      "loss": 0.4517,
      "step": 1132
    },
    {
      "epoch": 0.06344495464217717,
      "grad_norm": 1.472391963005066,
      "learning_rate": 5.655000000000001e-06,
      "loss": 0.451,
      "step": 1133
    },
    {
      "epoch": 0.06350095195430619,
      "grad_norm": 1.1070297956466675,
      "learning_rate": 5.66e-06,
      "loss": 0.4294,
      "step": 1134
    },
    {
      "epoch": 0.06355694926643521,
      "grad_norm": 1.4566556215286255,
      "learning_rate": 5.665e-06,
      "loss": 0.6163,
      "step": 1135
    },
    {
      "epoch": 0.06361294657856423,
      "grad_norm": 1.146721363067627,
      "learning_rate": 5.67e-06,
      "loss": 0.3609,
      "step": 1136
    },
    {
      "epoch": 0.06366894389069325,
      "grad_norm": 1.228152871131897,
      "learning_rate": 5.675000000000001e-06,
      "loss": 0.5157,
      "step": 1137
    },
    {
      "epoch": 0.06372494120282227,
      "grad_norm": 1.262759804725647,
      "learning_rate": 5.680000000000001e-06,
      "loss": 0.5129,
      "step": 1138
    },
    {
      "epoch": 0.06378093851495129,
      "grad_norm": 1.3428313732147217,
      "learning_rate": 5.685e-06,
      "loss": 0.4894,
      "step": 1139
    },
    {
      "epoch": 0.0638369358270803,
      "grad_norm": 1.4424748420715332,
      "learning_rate": 5.690000000000001e-06,
      "loss": 0.4817,
      "step": 1140
    },
    {
      "epoch": 0.06389293313920932,
      "grad_norm": 1.4872840642929077,
      "learning_rate": 5.6950000000000005e-06,
      "loss": 0.4142,
      "step": 1141
    },
    {
      "epoch": 0.06394893045133833,
      "grad_norm": 1.3789427280426025,
      "learning_rate": 5.7000000000000005e-06,
      "loss": 0.4699,
      "step": 1142
    },
    {
      "epoch": 0.06400492776346735,
      "grad_norm": 1.0845141410827637,
      "learning_rate": 5.705e-06,
      "loss": 0.4109,
      "step": 1143
    },
    {
      "epoch": 0.06406092507559637,
      "grad_norm": 1.1307657957077026,
      "learning_rate": 5.71e-06,
      "loss": 0.4811,
      "step": 1144
    },
    {
      "epoch": 0.06411692238772539,
      "grad_norm": 1.0611789226531982,
      "learning_rate": 5.715e-06,
      "loss": 0.4281,
      "step": 1145
    },
    {
      "epoch": 0.0641729196998544,
      "grad_norm": 1.1467761993408203,
      "learning_rate": 5.72e-06,
      "loss": 0.3959,
      "step": 1146
    },
    {
      "epoch": 0.06422891701198342,
      "grad_norm": 1.3303271532058716,
      "learning_rate": 5.725e-06,
      "loss": 0.4611,
      "step": 1147
    },
    {
      "epoch": 0.06428491432411244,
      "grad_norm": 1.106090784072876,
      "learning_rate": 5.73e-06,
      "loss": 0.451,
      "step": 1148
    },
    {
      "epoch": 0.06434091163624146,
      "grad_norm": 1.0233514308929443,
      "learning_rate": 5.735e-06,
      "loss": 0.4053,
      "step": 1149
    },
    {
      "epoch": 0.06439690894837048,
      "grad_norm": 0.9405644536018372,
      "learning_rate": 5.74e-06,
      "loss": 0.3488,
      "step": 1150
    },
    {
      "epoch": 0.0644529062604995,
      "grad_norm": 1.0472259521484375,
      "learning_rate": 5.745e-06,
      "loss": 0.4616,
      "step": 1151
    },
    {
      "epoch": 0.06450890357262852,
      "grad_norm": 1.146802544593811,
      "learning_rate": 5.750000000000001e-06,
      "loss": 0.4724,
      "step": 1152
    },
    {
      "epoch": 0.06456490088475753,
      "grad_norm": 1.3439722061157227,
      "learning_rate": 5.755e-06,
      "loss": 0.5802,
      "step": 1153
    },
    {
      "epoch": 0.06462089819688654,
      "grad_norm": 1.4531556367874146,
      "learning_rate": 5.76e-06,
      "loss": 0.5034,
      "step": 1154
    },
    {
      "epoch": 0.06467689550901556,
      "grad_norm": 1.0436081886291504,
      "learning_rate": 5.765e-06,
      "loss": 0.3639,
      "step": 1155
    },
    {
      "epoch": 0.06473289282114458,
      "grad_norm": 1.1610552072525024,
      "learning_rate": 5.770000000000001e-06,
      "loss": 0.4778,
      "step": 1156
    },
    {
      "epoch": 0.0647888901332736,
      "grad_norm": 1.209583044052124,
      "learning_rate": 5.775000000000001e-06,
      "loss": 0.4406,
      "step": 1157
    },
    {
      "epoch": 0.06484488744540262,
      "grad_norm": 1.3594990968704224,
      "learning_rate": 5.78e-06,
      "loss": 0.5006,
      "step": 1158
    },
    {
      "epoch": 0.06490088475753164,
      "grad_norm": 1.2716442346572876,
      "learning_rate": 5.7850000000000005e-06,
      "loss": 0.5065,
      "step": 1159
    },
    {
      "epoch": 0.06495688206966066,
      "grad_norm": 1.2829830646514893,
      "learning_rate": 5.7900000000000005e-06,
      "loss": 0.6007,
      "step": 1160
    },
    {
      "epoch": 0.06501287938178968,
      "grad_norm": 1.3459588289260864,
      "learning_rate": 5.795e-06,
      "loss": 0.5943,
      "step": 1161
    },
    {
      "epoch": 0.0650688766939187,
      "grad_norm": 1.4075367450714111,
      "learning_rate": 5.8e-06,
      "loss": 0.5364,
      "step": 1162
    },
    {
      "epoch": 0.0651248740060477,
      "grad_norm": 1.419149398803711,
      "learning_rate": 5.805e-06,
      "loss": 0.4434,
      "step": 1163
    },
    {
      "epoch": 0.06518087131817672,
      "grad_norm": 1.4249473810195923,
      "learning_rate": 5.81e-06,
      "loss": 0.5376,
      "step": 1164
    },
    {
      "epoch": 0.06523686863030574,
      "grad_norm": 1.5048781633377075,
      "learning_rate": 5.815e-06,
      "loss": 0.5568,
      "step": 1165
    },
    {
      "epoch": 0.06529286594243476,
      "grad_norm": 1.059531807899475,
      "learning_rate": 5.82e-06,
      "loss": 0.3689,
      "step": 1166
    },
    {
      "epoch": 0.06534886325456378,
      "grad_norm": 1.1879932880401611,
      "learning_rate": 5.825000000000001e-06,
      "loss": 0.4556,
      "step": 1167
    },
    {
      "epoch": 0.0654048605666928,
      "grad_norm": 2.4681849479675293,
      "learning_rate": 5.83e-06,
      "loss": 0.4627,
      "step": 1168
    },
    {
      "epoch": 0.06546085787882182,
      "grad_norm": 1.6035898923873901,
      "learning_rate": 5.835e-06,
      "loss": 0.4652,
      "step": 1169
    },
    {
      "epoch": 0.06551685519095084,
      "grad_norm": 1.147926688194275,
      "learning_rate": 5.84e-06,
      "loss": 0.3389,
      "step": 1170
    },
    {
      "epoch": 0.06557285250307986,
      "grad_norm": 1.3321911096572876,
      "learning_rate": 5.845000000000001e-06,
      "loss": 0.5156,
      "step": 1171
    },
    {
      "epoch": 0.06562884981520888,
      "grad_norm": 1.3565380573272705,
      "learning_rate": 5.850000000000001e-06,
      "loss": 0.4251,
      "step": 1172
    },
    {
      "epoch": 0.06568484712733788,
      "grad_norm": 1.0869241952896118,
      "learning_rate": 5.855e-06,
      "loss": 0.4787,
      "step": 1173
    },
    {
      "epoch": 0.0657408444394669,
      "grad_norm": 1.4895256757736206,
      "learning_rate": 5.86e-06,
      "loss": 0.6179,
      "step": 1174
    },
    {
      "epoch": 0.06579684175159592,
      "grad_norm": 1.591769814491272,
      "learning_rate": 5.865000000000001e-06,
      "loss": 0.5253,
      "step": 1175
    },
    {
      "epoch": 0.06585283906372494,
      "grad_norm": 1.162840485572815,
      "learning_rate": 5.8700000000000005e-06,
      "loss": 0.4676,
      "step": 1176
    },
    {
      "epoch": 0.06590883637585396,
      "grad_norm": 1.1410831212997437,
      "learning_rate": 5.875e-06,
      "loss": 0.3855,
      "step": 1177
    },
    {
      "epoch": 0.06596483368798298,
      "grad_norm": 1.4543075561523438,
      "learning_rate": 5.8800000000000005e-06,
      "loss": 0.4941,
      "step": 1178
    },
    {
      "epoch": 0.066020831000112,
      "grad_norm": 1.4013811349868774,
      "learning_rate": 5.885e-06,
      "loss": 0.476,
      "step": 1179
    },
    {
      "epoch": 0.06607682831224101,
      "grad_norm": 2.4917755126953125,
      "learning_rate": 5.89e-06,
      "loss": 0.3922,
      "step": 1180
    },
    {
      "epoch": 0.06613282562437003,
      "grad_norm": 1.086756944656372,
      "learning_rate": 5.895e-06,
      "loss": 0.4143,
      "step": 1181
    },
    {
      "epoch": 0.06618882293649905,
      "grad_norm": 1.2095462083816528,
      "learning_rate": 5.9e-06,
      "loss": 0.4772,
      "step": 1182
    },
    {
      "epoch": 0.06624482024862807,
      "grad_norm": 1.4608217477798462,
      "learning_rate": 5.905e-06,
      "loss": 0.4348,
      "step": 1183
    },
    {
      "epoch": 0.06630081756075708,
      "grad_norm": 1.523728609085083,
      "learning_rate": 5.91e-06,
      "loss": 0.6067,
      "step": 1184
    },
    {
      "epoch": 0.0663568148728861,
      "grad_norm": 1.1437195539474487,
      "learning_rate": 5.915e-06,
      "loss": 0.3993,
      "step": 1185
    },
    {
      "epoch": 0.06641281218501512,
      "grad_norm": 1.146959662437439,
      "learning_rate": 5.920000000000001e-06,
      "loss": 0.474,
      "step": 1186
    },
    {
      "epoch": 0.06646880949714414,
      "grad_norm": 1.1326645612716675,
      "learning_rate": 5.925e-06,
      "loss": 0.3891,
      "step": 1187
    },
    {
      "epoch": 0.06652480680927315,
      "grad_norm": 1.0731227397918701,
      "learning_rate": 5.93e-06,
      "loss": 0.3448,
      "step": 1188
    },
    {
      "epoch": 0.06658080412140217,
      "grad_norm": 1.418973684310913,
      "learning_rate": 5.935e-06,
      "loss": 0.4704,
      "step": 1189
    },
    {
      "epoch": 0.06663680143353119,
      "grad_norm": 1.1930632591247559,
      "learning_rate": 5.940000000000001e-06,
      "loss": 0.3829,
      "step": 1190
    },
    {
      "epoch": 0.06669279874566021,
      "grad_norm": 1.1238776445388794,
      "learning_rate": 5.945000000000001e-06,
      "loss": 0.3387,
      "step": 1191
    },
    {
      "epoch": 0.06674879605778923,
      "grad_norm": 1.3016777038574219,
      "learning_rate": 5.95e-06,
      "loss": 0.5368,
      "step": 1192
    },
    {
      "epoch": 0.06680479336991825,
      "grad_norm": 1.2945044040679932,
      "learning_rate": 5.955000000000001e-06,
      "loss": 0.4609,
      "step": 1193
    },
    {
      "epoch": 0.06686079068204726,
      "grad_norm": 1.4390751123428345,
      "learning_rate": 5.9600000000000005e-06,
      "loss": 0.476,
      "step": 1194
    },
    {
      "epoch": 0.06691678799417627,
      "grad_norm": 1.038458228111267,
      "learning_rate": 5.9650000000000005e-06,
      "loss": 0.3525,
      "step": 1195
    },
    {
      "epoch": 0.0669727853063053,
      "grad_norm": 1.1566381454467773,
      "learning_rate": 5.9700000000000004e-06,
      "loss": 0.4043,
      "step": 1196
    },
    {
      "epoch": 0.06702878261843431,
      "grad_norm": 1.3829346895217896,
      "learning_rate": 5.975e-06,
      "loss": 0.3849,
      "step": 1197
    },
    {
      "epoch": 0.06708477993056333,
      "grad_norm": 1.5022857189178467,
      "learning_rate": 5.98e-06,
      "loss": 0.5313,
      "step": 1198
    },
    {
      "epoch": 0.06714077724269235,
      "grad_norm": 1.3618924617767334,
      "learning_rate": 5.985e-06,
      "loss": 0.4925,
      "step": 1199
    },
    {
      "epoch": 0.06719677455482137,
      "grad_norm": 1.1289489269256592,
      "learning_rate": 5.99e-06,
      "loss": 0.4691,
      "step": 1200
    },
    {
      "epoch": 0.06725277186695039,
      "grad_norm": 1.4186307191848755,
      "learning_rate": 5.995e-06,
      "loss": 0.5976,
      "step": 1201
    },
    {
      "epoch": 0.06730876917907941,
      "grad_norm": 1.2453892230987549,
      "learning_rate": 6e-06,
      "loss": 0.7152,
      "step": 1202
    },
    {
      "epoch": 0.06736476649120843,
      "grad_norm": 1.31748628616333,
      "learning_rate": 6.005e-06,
      "loss": 0.6554,
      "step": 1203
    },
    {
      "epoch": 0.06742076380333743,
      "grad_norm": 1.271621584892273,
      "learning_rate": 6.01e-06,
      "loss": 0.5174,
      "step": 1204
    },
    {
      "epoch": 0.06747676111546645,
      "grad_norm": 1.4273734092712402,
      "learning_rate": 6.015000000000001e-06,
      "loss": 0.5568,
      "step": 1205
    },
    {
      "epoch": 0.06753275842759547,
      "grad_norm": 1.1785576343536377,
      "learning_rate": 6.02e-06,
      "loss": 0.4569,
      "step": 1206
    },
    {
      "epoch": 0.06758875573972449,
      "grad_norm": 1.4180866479873657,
      "learning_rate": 6.025e-06,
      "loss": 0.5308,
      "step": 1207
    },
    {
      "epoch": 0.06764475305185351,
      "grad_norm": 1.1484123468399048,
      "learning_rate": 6.03e-06,
      "loss": 0.3796,
      "step": 1208
    },
    {
      "epoch": 0.06770075036398253,
      "grad_norm": 1.2170040607452393,
      "learning_rate": 6.035000000000001e-06,
      "loss": 0.4532,
      "step": 1209
    },
    {
      "epoch": 0.06775674767611155,
      "grad_norm": 1.2905791997909546,
      "learning_rate": 6.040000000000001e-06,
      "loss": 0.4595,
      "step": 1210
    },
    {
      "epoch": 0.06781274498824057,
      "grad_norm": 1.2938952445983887,
      "learning_rate": 6.045e-06,
      "loss": 0.4142,
      "step": 1211
    },
    {
      "epoch": 0.06786874230036959,
      "grad_norm": 1.3278106451034546,
      "learning_rate": 6.0500000000000005e-06,
      "loss": 0.4678,
      "step": 1212
    },
    {
      "epoch": 0.0679247396124986,
      "grad_norm": 1.311219334602356,
      "learning_rate": 6.0550000000000005e-06,
      "loss": 0.3523,
      "step": 1213
    },
    {
      "epoch": 0.06798073692462762,
      "grad_norm": 1.160986304283142,
      "learning_rate": 6.0600000000000004e-06,
      "loss": 0.4355,
      "step": 1214
    },
    {
      "epoch": 0.06803673423675663,
      "grad_norm": 1.3731441497802734,
      "learning_rate": 6.065e-06,
      "loss": 0.5311,
      "step": 1215
    },
    {
      "epoch": 0.06809273154888565,
      "grad_norm": 1.191652536392212,
      "learning_rate": 6.07e-06,
      "loss": 0.4536,
      "step": 1216
    },
    {
      "epoch": 0.06814872886101467,
      "grad_norm": 1.0599191188812256,
      "learning_rate": 6.075e-06,
      "loss": 0.4172,
      "step": 1217
    },
    {
      "epoch": 0.06820472617314369,
      "grad_norm": 1.1530780792236328,
      "learning_rate": 6.08e-06,
      "loss": 0.4636,
      "step": 1218
    },
    {
      "epoch": 0.0682607234852727,
      "grad_norm": 1.4733500480651855,
      "learning_rate": 6.085e-06,
      "loss": 0.6163,
      "step": 1219
    },
    {
      "epoch": 0.06831672079740173,
      "grad_norm": 1.335119605064392,
      "learning_rate": 6.090000000000001e-06,
      "loss": 0.5694,
      "step": 1220
    },
    {
      "epoch": 0.06837271810953074,
      "grad_norm": 1.1457488536834717,
      "learning_rate": 6.095e-06,
      "loss": 0.4272,
      "step": 1221
    },
    {
      "epoch": 0.06842871542165976,
      "grad_norm": 1.125653862953186,
      "learning_rate": 6.1e-06,
      "loss": 0.3819,
      "step": 1222
    },
    {
      "epoch": 0.06848471273378878,
      "grad_norm": 1.4815268516540527,
      "learning_rate": 6.105e-06,
      "loss": 0.6059,
      "step": 1223
    },
    {
      "epoch": 0.0685407100459178,
      "grad_norm": 1.072304606437683,
      "learning_rate": 6.110000000000001e-06,
      "loss": 0.5435,
      "step": 1224
    },
    {
      "epoch": 0.06859670735804681,
      "grad_norm": 1.2699904441833496,
      "learning_rate": 6.115000000000001e-06,
      "loss": 0.4013,
      "step": 1225
    },
    {
      "epoch": 0.06865270467017583,
      "grad_norm": 1.4592626094818115,
      "learning_rate": 6.12e-06,
      "loss": 0.4995,
      "step": 1226
    },
    {
      "epoch": 0.06870870198230485,
      "grad_norm": 1.1897928714752197,
      "learning_rate": 6.125e-06,
      "loss": 0.3772,
      "step": 1227
    },
    {
      "epoch": 0.06876469929443386,
      "grad_norm": 1.142614483833313,
      "learning_rate": 6.130000000000001e-06,
      "loss": 0.3635,
      "step": 1228
    },
    {
      "epoch": 0.06882069660656288,
      "grad_norm": 1.2323251962661743,
      "learning_rate": 6.1350000000000006e-06,
      "loss": 0.4454,
      "step": 1229
    },
    {
      "epoch": 0.0688766939186919,
      "grad_norm": 1.40288507938385,
      "learning_rate": 6.1400000000000005e-06,
      "loss": 0.5085,
      "step": 1230
    },
    {
      "epoch": 0.06893269123082092,
      "grad_norm": 1.1993368864059448,
      "learning_rate": 6.1450000000000005e-06,
      "loss": 0.3992,
      "step": 1231
    },
    {
      "epoch": 0.06898868854294994,
      "grad_norm": 1.6297578811645508,
      "learning_rate": 6.15e-06,
      "loss": 0.5697,
      "step": 1232
    },
    {
      "epoch": 0.06904468585507896,
      "grad_norm": 1.4021323919296265,
      "learning_rate": 6.155e-06,
      "loss": 0.3727,
      "step": 1233
    },
    {
      "epoch": 0.06910068316720798,
      "grad_norm": 1.0993765592575073,
      "learning_rate": 6.16e-06,
      "loss": 0.6237,
      "step": 1234
    },
    {
      "epoch": 0.06915668047933698,
      "grad_norm": 1.3789044618606567,
      "learning_rate": 6.165e-06,
      "loss": 0.5462,
      "step": 1235
    },
    {
      "epoch": 0.069212677791466,
      "grad_norm": 1.5529453754425049,
      "learning_rate": 6.17e-06,
      "loss": 0.5599,
      "step": 1236
    },
    {
      "epoch": 0.06926867510359502,
      "grad_norm": 1.2780702114105225,
      "learning_rate": 6.175e-06,
      "loss": 0.6287,
      "step": 1237
    },
    {
      "epoch": 0.06932467241572404,
      "grad_norm": 0.9809923768043518,
      "learning_rate": 6.18e-06,
      "loss": 0.3876,
      "step": 1238
    },
    {
      "epoch": 0.06938066972785306,
      "grad_norm": 1.0424203872680664,
      "learning_rate": 6.185000000000001e-06,
      "loss": 0.3569,
      "step": 1239
    },
    {
      "epoch": 0.06943666703998208,
      "grad_norm": 1.1957716941833496,
      "learning_rate": 6.19e-06,
      "loss": 0.5428,
      "step": 1240
    },
    {
      "epoch": 0.0694926643521111,
      "grad_norm": 1.3562878370285034,
      "learning_rate": 6.195e-06,
      "loss": 0.4914,
      "step": 1241
    },
    {
      "epoch": 0.06954866166424012,
      "grad_norm": 1.4439308643341064,
      "learning_rate": 6.2e-06,
      "loss": 0.5092,
      "step": 1242
    },
    {
      "epoch": 0.06960465897636914,
      "grad_norm": 1.3014452457427979,
      "learning_rate": 6.205000000000001e-06,
      "loss": 0.4253,
      "step": 1243
    },
    {
      "epoch": 0.06966065628849816,
      "grad_norm": 1.09708571434021,
      "learning_rate": 6.210000000000001e-06,
      "loss": 0.4735,
      "step": 1244
    },
    {
      "epoch": 0.06971665360062718,
      "grad_norm": 1.2579667568206787,
      "learning_rate": 6.215e-06,
      "loss": 0.3417,
      "step": 1245
    },
    {
      "epoch": 0.06977265091275618,
      "grad_norm": 1.1711158752441406,
      "learning_rate": 6.22e-06,
      "loss": 0.5578,
      "step": 1246
    },
    {
      "epoch": 0.0698286482248852,
      "grad_norm": 1.3848479986190796,
      "learning_rate": 6.2250000000000005e-06,
      "loss": 0.3832,
      "step": 1247
    },
    {
      "epoch": 0.06988464553701422,
      "grad_norm": 1.2990936040878296,
      "learning_rate": 6.2300000000000005e-06,
      "loss": 0.5252,
      "step": 1248
    },
    {
      "epoch": 0.06994064284914324,
      "grad_norm": 1.3192658424377441,
      "learning_rate": 6.2350000000000004e-06,
      "loss": 0.6234,
      "step": 1249
    },
    {
      "epoch": 0.06999664016127226,
      "grad_norm": 1.1719008684158325,
      "learning_rate": 6.24e-06,
      "loss": 0.4339,
      "step": 1250
    },
    {
      "epoch": 0.07005263747340128,
      "grad_norm": 1.21258544921875,
      "learning_rate": 6.245e-06,
      "loss": 0.4835,
      "step": 1251
    },
    {
      "epoch": 0.0701086347855303,
      "grad_norm": 1.2766406536102295,
      "learning_rate": 6.25e-06,
      "loss": 0.5146,
      "step": 1252
    },
    {
      "epoch": 0.07016463209765932,
      "grad_norm": 1.305820107460022,
      "learning_rate": 6.254999999999999e-06,
      "loss": 0.3805,
      "step": 1253
    },
    {
      "epoch": 0.07022062940978833,
      "grad_norm": 1.2711191177368164,
      "learning_rate": 6.26e-06,
      "loss": 0.4137,
      "step": 1254
    },
    {
      "epoch": 0.07027662672191735,
      "grad_norm": 1.3470959663391113,
      "learning_rate": 6.265e-06,
      "loss": 0.4417,
      "step": 1255
    },
    {
      "epoch": 0.07033262403404636,
      "grad_norm": 0.9316703677177429,
      "learning_rate": 6.270000000000001e-06,
      "loss": 0.3569,
      "step": 1256
    },
    {
      "epoch": 0.07038862134617538,
      "grad_norm": 1.3561193943023682,
      "learning_rate": 6.275e-06,
      "loss": 0.5043,
      "step": 1257
    },
    {
      "epoch": 0.0704446186583044,
      "grad_norm": 1.0972591638565063,
      "learning_rate": 6.28e-06,
      "loss": 0.4823,
      "step": 1258
    },
    {
      "epoch": 0.07050061597043342,
      "grad_norm": 1.1062606573104858,
      "learning_rate": 6.285000000000001e-06,
      "loss": 0.3591,
      "step": 1259
    },
    {
      "epoch": 0.07055661328256244,
      "grad_norm": 1.186181902885437,
      "learning_rate": 6.29e-06,
      "loss": 0.6183,
      "step": 1260
    },
    {
      "epoch": 0.07061261059469146,
      "grad_norm": 1.2585527896881104,
      "learning_rate": 6.295000000000001e-06,
      "loss": 0.5375,
      "step": 1261
    },
    {
      "epoch": 0.07066860790682047,
      "grad_norm": 1.2941310405731201,
      "learning_rate": 6.300000000000001e-06,
      "loss": 0.4806,
      "step": 1262
    },
    {
      "epoch": 0.0707246052189495,
      "grad_norm": 1.3581117391586304,
      "learning_rate": 6.305e-06,
      "loss": 0.5579,
      "step": 1263
    },
    {
      "epoch": 0.07078060253107851,
      "grad_norm": 1.5740511417388916,
      "learning_rate": 6.3100000000000006e-06,
      "loss": 0.5084,
      "step": 1264
    },
    {
      "epoch": 0.07083659984320753,
      "grad_norm": 1.3985000848770142,
      "learning_rate": 6.315e-06,
      "loss": 0.4166,
      "step": 1265
    },
    {
      "epoch": 0.07089259715533654,
      "grad_norm": 1.2136940956115723,
      "learning_rate": 6.320000000000001e-06,
      "loss": 0.4584,
      "step": 1266
    },
    {
      "epoch": 0.07094859446746556,
      "grad_norm": 1.1227129697799683,
      "learning_rate": 6.3250000000000004e-06,
      "loss": 0.388,
      "step": 1267
    },
    {
      "epoch": 0.07100459177959458,
      "grad_norm": 1.2944831848144531,
      "learning_rate": 6.3299999999999995e-06,
      "loss": 0.5666,
      "step": 1268
    },
    {
      "epoch": 0.0710605890917236,
      "grad_norm": 1.0914361476898193,
      "learning_rate": 6.335e-06,
      "loss": 0.3782,
      "step": 1269
    },
    {
      "epoch": 0.07111658640385261,
      "grad_norm": 1.227116346359253,
      "learning_rate": 6.34e-06,
      "loss": 0.5257,
      "step": 1270
    },
    {
      "epoch": 0.07117258371598163,
      "grad_norm": 1.1540474891662598,
      "learning_rate": 6.345000000000001e-06,
      "loss": 0.3348,
      "step": 1271
    },
    {
      "epoch": 0.07122858102811065,
      "grad_norm": 1.340309977531433,
      "learning_rate": 6.35e-06,
      "loss": 0.465,
      "step": 1272
    },
    {
      "epoch": 0.07128457834023967,
      "grad_norm": 1.0309977531433105,
      "learning_rate": 6.355e-06,
      "loss": 0.4303,
      "step": 1273
    },
    {
      "epoch": 0.07134057565236869,
      "grad_norm": 1.1087217330932617,
      "learning_rate": 6.360000000000001e-06,
      "loss": 0.3617,
      "step": 1274
    },
    {
      "epoch": 0.07139657296449771,
      "grad_norm": 1.0575168132781982,
      "learning_rate": 6.365e-06,
      "loss": 0.4675,
      "step": 1275
    },
    {
      "epoch": 0.07145257027662673,
      "grad_norm": 3.266204833984375,
      "learning_rate": 6.370000000000001e-06,
      "loss": 0.5266,
      "step": 1276
    },
    {
      "epoch": 0.07150856758875573,
      "grad_norm": 1.163043737411499,
      "learning_rate": 6.375000000000001e-06,
      "loss": 0.3054,
      "step": 1277
    },
    {
      "epoch": 0.07156456490088475,
      "grad_norm": 1.5021004676818848,
      "learning_rate": 6.38e-06,
      "loss": 0.6384,
      "step": 1278
    },
    {
      "epoch": 0.07162056221301377,
      "grad_norm": 1.5890547037124634,
      "learning_rate": 6.385000000000001e-06,
      "loss": 0.6123,
      "step": 1279
    },
    {
      "epoch": 0.07167655952514279,
      "grad_norm": 1.2791961431503296,
      "learning_rate": 6.39e-06,
      "loss": 0.6111,
      "step": 1280
    },
    {
      "epoch": 0.07173255683727181,
      "grad_norm": 1.395918846130371,
      "learning_rate": 6.395000000000001e-06,
      "loss": 0.3987,
      "step": 1281
    },
    {
      "epoch": 0.07178855414940083,
      "grad_norm": 1.292610764503479,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.4165,
      "step": 1282
    },
    {
      "epoch": 0.07184455146152985,
      "grad_norm": 1.6642959117889404,
      "learning_rate": 6.405e-06,
      "loss": 0.4285,
      "step": 1283
    },
    {
      "epoch": 0.07190054877365887,
      "grad_norm": 1.4356571435928345,
      "learning_rate": 6.4100000000000005e-06,
      "loss": 0.4188,
      "step": 1284
    },
    {
      "epoch": 0.07195654608578789,
      "grad_norm": 1.263612985610962,
      "learning_rate": 6.415e-06,
      "loss": 0.4058,
      "step": 1285
    },
    {
      "epoch": 0.0720125433979169,
      "grad_norm": 1.3215962648391724,
      "learning_rate": 6.4199999999999995e-06,
      "loss": 0.5013,
      "step": 1286
    },
    {
      "epoch": 0.07206854071004591,
      "grad_norm": 1.2451215982437134,
      "learning_rate": 6.425e-06,
      "loss": 0.5058,
      "step": 1287
    },
    {
      "epoch": 0.07212453802217493,
      "grad_norm": 1.1814186573028564,
      "learning_rate": 6.43e-06,
      "loss": 0.3944,
      "step": 1288
    },
    {
      "epoch": 0.07218053533430395,
      "grad_norm": 1.1279220581054688,
      "learning_rate": 6.435000000000001e-06,
      "loss": 0.4033,
      "step": 1289
    },
    {
      "epoch": 0.07223653264643297,
      "grad_norm": 1.166107416152954,
      "learning_rate": 6.44e-06,
      "loss": 0.4462,
      "step": 1290
    },
    {
      "epoch": 0.07229252995856199,
      "grad_norm": 1.1797864437103271,
      "learning_rate": 6.444999999999999e-06,
      "loss": 0.3905,
      "step": 1291
    },
    {
      "epoch": 0.07234852727069101,
      "grad_norm": 1.0709209442138672,
      "learning_rate": 6.45e-06,
      "loss": 0.3772,
      "step": 1292
    },
    {
      "epoch": 0.07240452458282003,
      "grad_norm": 1.115615963935852,
      "learning_rate": 6.455e-06,
      "loss": 0.3869,
      "step": 1293
    },
    {
      "epoch": 0.07246052189494905,
      "grad_norm": 1.7949514389038086,
      "learning_rate": 6.460000000000001e-06,
      "loss": 0.5933,
      "step": 1294
    },
    {
      "epoch": 0.07251651920707806,
      "grad_norm": 1.6825324296951294,
      "learning_rate": 6.465e-06,
      "loss": 0.5273,
      "step": 1295
    },
    {
      "epoch": 0.07257251651920708,
      "grad_norm": 1.1637367010116577,
      "learning_rate": 6.47e-06,
      "loss": 0.4111,
      "step": 1296
    },
    {
      "epoch": 0.07262851383133609,
      "grad_norm": 1.2039698362350464,
      "learning_rate": 6.475000000000001e-06,
      "loss": 0.5101,
      "step": 1297
    },
    {
      "epoch": 0.07268451114346511,
      "grad_norm": 1.3971374034881592,
      "learning_rate": 6.48e-06,
      "loss": 0.4956,
      "step": 1298
    },
    {
      "epoch": 0.07274050845559413,
      "grad_norm": 1.3009939193725586,
      "learning_rate": 6.485000000000001e-06,
      "loss": 0.371,
      "step": 1299
    },
    {
      "epoch": 0.07279650576772315,
      "grad_norm": 1.4169080257415771,
      "learning_rate": 6.4900000000000005e-06,
      "loss": 0.7171,
      "step": 1300
    },
    {
      "epoch": 0.07285250307985217,
      "grad_norm": 2.2091469764709473,
      "learning_rate": 6.495e-06,
      "loss": 0.565,
      "step": 1301
    },
    {
      "epoch": 0.07290850039198118,
      "grad_norm": 1.427626371383667,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.39,
      "step": 1302
    },
    {
      "epoch": 0.0729644977041102,
      "grad_norm": 1.2434399127960205,
      "learning_rate": 6.505e-06,
      "loss": 0.4394,
      "step": 1303
    },
    {
      "epoch": 0.07302049501623922,
      "grad_norm": 1.6202256679534912,
      "learning_rate": 6.510000000000001e-06,
      "loss": 0.5287,
      "step": 1304
    },
    {
      "epoch": 0.07307649232836824,
      "grad_norm": 2.5532824993133545,
      "learning_rate": 6.515e-06,
      "loss": 0.577,
      "step": 1305
    },
    {
      "epoch": 0.07313248964049726,
      "grad_norm": 1.2064380645751953,
      "learning_rate": 6.519999999999999e-06,
      "loss": 0.6086,
      "step": 1306
    },
    {
      "epoch": 0.07318848695262628,
      "grad_norm": 1.3083627223968506,
      "learning_rate": 6.525e-06,
      "loss": 0.4166,
      "step": 1307
    },
    {
      "epoch": 0.07324448426475529,
      "grad_norm": 1.352712631225586,
      "learning_rate": 6.53e-06,
      "loss": 0.4438,
      "step": 1308
    },
    {
      "epoch": 0.0733004815768843,
      "grad_norm": 1.2328424453735352,
      "learning_rate": 6.535000000000001e-06,
      "loss": 0.4751,
      "step": 1309
    },
    {
      "epoch": 0.07335647888901332,
      "grad_norm": 1.2588063478469849,
      "learning_rate": 6.54e-06,
      "loss": 0.5114,
      "step": 1310
    },
    {
      "epoch": 0.07341247620114234,
      "grad_norm": 1.082762360572815,
      "learning_rate": 6.545e-06,
      "loss": 0.4373,
      "step": 1311
    },
    {
      "epoch": 0.07346847351327136,
      "grad_norm": 1.2910832166671753,
      "learning_rate": 6.550000000000001e-06,
      "loss": 0.4089,
      "step": 1312
    },
    {
      "epoch": 0.07352447082540038,
      "grad_norm": 1.1955432891845703,
      "learning_rate": 6.555e-06,
      "loss": 0.41,
      "step": 1313
    },
    {
      "epoch": 0.0735804681375294,
      "grad_norm": 1.530601143836975,
      "learning_rate": 6.560000000000001e-06,
      "loss": 0.7009,
      "step": 1314
    },
    {
      "epoch": 0.07363646544965842,
      "grad_norm": 1.209916353225708,
      "learning_rate": 6.565000000000001e-06,
      "loss": 0.6077,
      "step": 1315
    },
    {
      "epoch": 0.07369246276178744,
      "grad_norm": 1.1398082971572876,
      "learning_rate": 6.57e-06,
      "loss": 0.4088,
      "step": 1316
    },
    {
      "epoch": 0.07374846007391646,
      "grad_norm": 1.4278721809387207,
      "learning_rate": 6.5750000000000006e-06,
      "loss": 0.5536,
      "step": 1317
    },
    {
      "epoch": 0.07380445738604546,
      "grad_norm": 1.253266453742981,
      "learning_rate": 6.58e-06,
      "loss": 0.4177,
      "step": 1318
    },
    {
      "epoch": 0.07386045469817448,
      "grad_norm": 1.234006404876709,
      "learning_rate": 6.5850000000000005e-06,
      "loss": 0.3958,
      "step": 1319
    },
    {
      "epoch": 0.0739164520103035,
      "grad_norm": 1.3969980478286743,
      "learning_rate": 6.5900000000000004e-06,
      "loss": 0.5297,
      "step": 1320
    },
    {
      "epoch": 0.07397244932243252,
      "grad_norm": 1.5530579090118408,
      "learning_rate": 6.5949999999999995e-06,
      "loss": 0.5444,
      "step": 1321
    },
    {
      "epoch": 0.07402844663456154,
      "grad_norm": 1.1098066568374634,
      "learning_rate": 6.6e-06,
      "loss": 0.4511,
      "step": 1322
    },
    {
      "epoch": 0.07408444394669056,
      "grad_norm": 1.1944104433059692,
      "learning_rate": 6.605e-06,
      "loss": 0.3559,
      "step": 1323
    },
    {
      "epoch": 0.07414044125881958,
      "grad_norm": 1.2156531810760498,
      "learning_rate": 6.610000000000001e-06,
      "loss": 0.4142,
      "step": 1324
    },
    {
      "epoch": 0.0741964385709486,
      "grad_norm": 1.2404552698135376,
      "learning_rate": 6.615e-06,
      "loss": 0.4893,
      "step": 1325
    },
    {
      "epoch": 0.07425243588307762,
      "grad_norm": 1.3357646465301514,
      "learning_rate": 6.62e-06,
      "loss": 0.4867,
      "step": 1326
    },
    {
      "epoch": 0.07430843319520664,
      "grad_norm": 1.24103844165802,
      "learning_rate": 6.625000000000001e-06,
      "loss": 0.563,
      "step": 1327
    },
    {
      "epoch": 0.07436443050733564,
      "grad_norm": 0.9767637848854065,
      "learning_rate": 6.63e-06,
      "loss": 0.4116,
      "step": 1328
    },
    {
      "epoch": 0.07442042781946466,
      "grad_norm": 1.257356882095337,
      "learning_rate": 6.635000000000001e-06,
      "loss": 0.487,
      "step": 1329
    },
    {
      "epoch": 0.07447642513159368,
      "grad_norm": 1.5664671659469604,
      "learning_rate": 6.640000000000001e-06,
      "loss": 0.4498,
      "step": 1330
    },
    {
      "epoch": 0.0745324224437227,
      "grad_norm": 1.2948386669158936,
      "learning_rate": 6.645e-06,
      "loss": 0.4589,
      "step": 1331
    },
    {
      "epoch": 0.07458841975585172,
      "grad_norm": 1.3933333158493042,
      "learning_rate": 6.650000000000001e-06,
      "loss": 0.5161,
      "step": 1332
    },
    {
      "epoch": 0.07464441706798074,
      "grad_norm": 1.3152962923049927,
      "learning_rate": 6.655e-06,
      "loss": 0.539,
      "step": 1333
    },
    {
      "epoch": 0.07470041438010976,
      "grad_norm": 1.059032678604126,
      "learning_rate": 6.660000000000001e-06,
      "loss": 0.5141,
      "step": 1334
    },
    {
      "epoch": 0.07475641169223878,
      "grad_norm": 1.3351902961730957,
      "learning_rate": 6.6650000000000006e-06,
      "loss": 0.4759,
      "step": 1335
    },
    {
      "epoch": 0.0748124090043678,
      "grad_norm": 1.3436874151229858,
      "learning_rate": 6.67e-06,
      "loss": 0.5057,
      "step": 1336
    },
    {
      "epoch": 0.07486840631649681,
      "grad_norm": 1.3562463521957397,
      "learning_rate": 6.6750000000000005e-06,
      "loss": 0.419,
      "step": 1337
    },
    {
      "epoch": 0.07492440362862583,
      "grad_norm": 1.434891939163208,
      "learning_rate": 6.68e-06,
      "loss": 0.4996,
      "step": 1338
    },
    {
      "epoch": 0.07498040094075484,
      "grad_norm": 1.249218463897705,
      "learning_rate": 6.685000000000001e-06,
      "loss": 0.3987,
      "step": 1339
    },
    {
      "epoch": 0.07503639825288386,
      "grad_norm": 1.236860990524292,
      "learning_rate": 6.69e-06,
      "loss": 0.5033,
      "step": 1340
    },
    {
      "epoch": 0.07509239556501288,
      "grad_norm": 1.1586852073669434,
      "learning_rate": 6.695e-06,
      "loss": 0.4026,
      "step": 1341
    },
    {
      "epoch": 0.0751483928771419,
      "grad_norm": 1.1438720226287842,
      "learning_rate": 6.700000000000001e-06,
      "loss": 0.411,
      "step": 1342
    },
    {
      "epoch": 0.07520439018927091,
      "grad_norm": 1.0562424659729004,
      "learning_rate": 6.705e-06,
      "loss": 0.346,
      "step": 1343
    },
    {
      "epoch": 0.07526038750139993,
      "grad_norm": 1.7857998609542847,
      "learning_rate": 6.710000000000001e-06,
      "loss": 0.3514,
      "step": 1344
    },
    {
      "epoch": 0.07531638481352895,
      "grad_norm": 1.415966510772705,
      "learning_rate": 6.715e-06,
      "loss": 0.4802,
      "step": 1345
    },
    {
      "epoch": 0.07537238212565797,
      "grad_norm": 1.238795280456543,
      "learning_rate": 6.72e-06,
      "loss": 0.4715,
      "step": 1346
    },
    {
      "epoch": 0.07542837943778699,
      "grad_norm": 1.2985560894012451,
      "learning_rate": 6.725000000000001e-06,
      "loss": 0.4027,
      "step": 1347
    },
    {
      "epoch": 0.07548437674991601,
      "grad_norm": 1.053097128868103,
      "learning_rate": 6.73e-06,
      "loss": 0.3804,
      "step": 1348
    },
    {
      "epoch": 0.07554037406204502,
      "grad_norm": 1.1594445705413818,
      "learning_rate": 6.735e-06,
      "loss": 0.4952,
      "step": 1349
    },
    {
      "epoch": 0.07559637137417403,
      "grad_norm": 1.3159594535827637,
      "learning_rate": 6.740000000000001e-06,
      "loss": 0.3801,
      "step": 1350
    },
    {
      "epoch": 0.07565236868630305,
      "grad_norm": 1.2488571405410767,
      "learning_rate": 6.745e-06,
      "loss": 0.5075,
      "step": 1351
    },
    {
      "epoch": 0.07570836599843207,
      "grad_norm": 1.1797436475753784,
      "learning_rate": 6.750000000000001e-06,
      "loss": 0.4398,
      "step": 1352
    },
    {
      "epoch": 0.07576436331056109,
      "grad_norm": 1.1828269958496094,
      "learning_rate": 6.7550000000000005e-06,
      "loss": 0.5035,
      "step": 1353
    },
    {
      "epoch": 0.07582036062269011,
      "grad_norm": 1.5088539123535156,
      "learning_rate": 6.76e-06,
      "loss": 0.4993,
      "step": 1354
    },
    {
      "epoch": 0.07587635793481913,
      "grad_norm": 1.2271095514297485,
      "learning_rate": 6.7650000000000005e-06,
      "loss": 0.4532,
      "step": 1355
    },
    {
      "epoch": 0.07593235524694815,
      "grad_norm": 1.0573952198028564,
      "learning_rate": 6.7699999999999996e-06,
      "loss": 0.3829,
      "step": 1356
    },
    {
      "epoch": 0.07598835255907717,
      "grad_norm": 1.1782804727554321,
      "learning_rate": 6.775000000000001e-06,
      "loss": 0.4338,
      "step": 1357
    },
    {
      "epoch": 0.07604434987120619,
      "grad_norm": 1.1794538497924805,
      "learning_rate": 6.78e-06,
      "loss": 0.4669,
      "step": 1358
    },
    {
      "epoch": 0.07610034718333519,
      "grad_norm": 1.2195823192596436,
      "learning_rate": 6.784999999999999e-06,
      "loss": 0.614,
      "step": 1359
    },
    {
      "epoch": 0.07615634449546421,
      "grad_norm": 1.3329391479492188,
      "learning_rate": 6.79e-06,
      "loss": 0.4395,
      "step": 1360
    },
    {
      "epoch": 0.07621234180759323,
      "grad_norm": 1.1854500770568848,
      "learning_rate": 6.795e-06,
      "loss": 0.5083,
      "step": 1361
    },
    {
      "epoch": 0.07626833911972225,
      "grad_norm": 1.340144395828247,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.4426,
      "step": 1362
    },
    {
      "epoch": 0.07632433643185127,
      "grad_norm": 1.565130352973938,
      "learning_rate": 6.805e-06,
      "loss": 0.4845,
      "step": 1363
    },
    {
      "epoch": 0.07638033374398029,
      "grad_norm": 2.765289306640625,
      "learning_rate": 6.81e-06,
      "loss": 0.5165,
      "step": 1364
    },
    {
      "epoch": 0.07643633105610931,
      "grad_norm": 1.2596156597137451,
      "learning_rate": 6.815000000000001e-06,
      "loss": 0.619,
      "step": 1365
    },
    {
      "epoch": 0.07649232836823833,
      "grad_norm": 1.4560434818267822,
      "learning_rate": 6.82e-06,
      "loss": 0.4165,
      "step": 1366
    },
    {
      "epoch": 0.07654832568036735,
      "grad_norm": 1.0805444717407227,
      "learning_rate": 6.825000000000001e-06,
      "loss": 0.4847,
      "step": 1367
    },
    {
      "epoch": 0.07660432299249637,
      "grad_norm": 1.2908236980438232,
      "learning_rate": 6.830000000000001e-06,
      "loss": 0.4831,
      "step": 1368
    },
    {
      "epoch": 0.07666032030462538,
      "grad_norm": 1.1790556907653809,
      "learning_rate": 6.835e-06,
      "loss": 0.4685,
      "step": 1369
    },
    {
      "epoch": 0.07671631761675439,
      "grad_norm": 1.2401238679885864,
      "learning_rate": 6.840000000000001e-06,
      "loss": 0.4726,
      "step": 1370
    },
    {
      "epoch": 0.07677231492888341,
      "grad_norm": 1.1133661270141602,
      "learning_rate": 6.845e-06,
      "loss": 0.3749,
      "step": 1371
    },
    {
      "epoch": 0.07682831224101243,
      "grad_norm": 1.0683152675628662,
      "learning_rate": 6.8500000000000005e-06,
      "loss": 0.3595,
      "step": 1372
    },
    {
      "epoch": 0.07688430955314145,
      "grad_norm": 1.19899582862854,
      "learning_rate": 6.8550000000000004e-06,
      "loss": 0.4688,
      "step": 1373
    },
    {
      "epoch": 0.07694030686527047,
      "grad_norm": 1.1337486505508423,
      "learning_rate": 6.8599999999999995e-06,
      "loss": 0.454,
      "step": 1374
    },
    {
      "epoch": 0.07699630417739949,
      "grad_norm": 1.4259730577468872,
      "learning_rate": 6.865e-06,
      "loss": 0.5977,
      "step": 1375
    },
    {
      "epoch": 0.0770523014895285,
      "grad_norm": 0.9646322727203369,
      "learning_rate": 6.87e-06,
      "loss": 0.4321,
      "step": 1376
    },
    {
      "epoch": 0.07710829880165752,
      "grad_norm": 1.3179805278778076,
      "learning_rate": 6.875000000000001e-06,
      "loss": 0.4031,
      "step": 1377
    },
    {
      "epoch": 0.07716429611378654,
      "grad_norm": 1.463568925857544,
      "learning_rate": 6.88e-06,
      "loss": 0.5436,
      "step": 1378
    },
    {
      "epoch": 0.07722029342591556,
      "grad_norm": 1.2384649515151978,
      "learning_rate": 6.885e-06,
      "loss": 0.4111,
      "step": 1379
    },
    {
      "epoch": 0.07727629073804457,
      "grad_norm": 1.4996534585952759,
      "learning_rate": 6.890000000000001e-06,
      "loss": 0.4076,
      "step": 1380
    },
    {
      "epoch": 0.07733228805017359,
      "grad_norm": 1.232329249382019,
      "learning_rate": 6.895e-06,
      "loss": 0.4954,
      "step": 1381
    },
    {
      "epoch": 0.0773882853623026,
      "grad_norm": 1.4218552112579346,
      "learning_rate": 6.900000000000001e-06,
      "loss": 0.5603,
      "step": 1382
    },
    {
      "epoch": 0.07744428267443162,
      "grad_norm": 1.2759547233581543,
      "learning_rate": 6.905e-06,
      "loss": 0.5696,
      "step": 1383
    },
    {
      "epoch": 0.07750027998656064,
      "grad_norm": 1.9820356369018555,
      "learning_rate": 6.91e-06,
      "loss": 0.3646,
      "step": 1384
    },
    {
      "epoch": 0.07755627729868966,
      "grad_norm": 1.2480307817459106,
      "learning_rate": 6.915000000000001e-06,
      "loss": 0.4816,
      "step": 1385
    },
    {
      "epoch": 0.07761227461081868,
      "grad_norm": 1.3854848146438599,
      "learning_rate": 6.92e-06,
      "loss": 0.4986,
      "step": 1386
    },
    {
      "epoch": 0.0776682719229477,
      "grad_norm": 1.5433881282806396,
      "learning_rate": 6.925000000000001e-06,
      "loss": 0.4568,
      "step": 1387
    },
    {
      "epoch": 0.07772426923507672,
      "grad_norm": 1.2369163036346436,
      "learning_rate": 6.9300000000000006e-06,
      "loss": 0.5573,
      "step": 1388
    },
    {
      "epoch": 0.07778026654720574,
      "grad_norm": 1.256779432296753,
      "learning_rate": 6.935e-06,
      "loss": 0.436,
      "step": 1389
    },
    {
      "epoch": 0.07783626385933475,
      "grad_norm": 1.3208945989608765,
      "learning_rate": 6.9400000000000005e-06,
      "loss": 0.5535,
      "step": 1390
    },
    {
      "epoch": 0.07789226117146376,
      "grad_norm": 1.238890290260315,
      "learning_rate": 6.945e-06,
      "loss": 0.4379,
      "step": 1391
    },
    {
      "epoch": 0.07794825848359278,
      "grad_norm": 1.3645505905151367,
      "learning_rate": 6.950000000000001e-06,
      "loss": 0.6061,
      "step": 1392
    },
    {
      "epoch": 0.0780042557957218,
      "grad_norm": 1.1675653457641602,
      "learning_rate": 6.955e-06,
      "loss": 0.3521,
      "step": 1393
    },
    {
      "epoch": 0.07806025310785082,
      "grad_norm": 1.2153840065002441,
      "learning_rate": 6.9599999999999994e-06,
      "loss": 0.4654,
      "step": 1394
    },
    {
      "epoch": 0.07811625041997984,
      "grad_norm": 1.3639699220657349,
      "learning_rate": 6.965000000000001e-06,
      "loss": 0.497,
      "step": 1395
    },
    {
      "epoch": 0.07817224773210886,
      "grad_norm": 1.2933604717254639,
      "learning_rate": 6.97e-06,
      "loss": 0.6267,
      "step": 1396
    },
    {
      "epoch": 0.07822824504423788,
      "grad_norm": 1.2269926071166992,
      "learning_rate": 6.975000000000001e-06,
      "loss": 0.4726,
      "step": 1397
    },
    {
      "epoch": 0.0782842423563669,
      "grad_norm": 1.5721561908721924,
      "learning_rate": 6.98e-06,
      "loss": 0.4109,
      "step": 1398
    },
    {
      "epoch": 0.07834023966849592,
      "grad_norm": 1.0359388589859009,
      "learning_rate": 6.985e-06,
      "loss": 0.4389,
      "step": 1399
    },
    {
      "epoch": 0.07839623698062494,
      "grad_norm": 4.911037445068359,
      "learning_rate": 6.990000000000001e-06,
      "loss": 0.45,
      "step": 1400
    },
    {
      "epoch": 0.07845223429275394,
      "grad_norm": 1.179355263710022,
      "learning_rate": 6.995e-06,
      "loss": 0.4637,
      "step": 1401
    },
    {
      "epoch": 0.07850823160488296,
      "grad_norm": 1.384724736213684,
      "learning_rate": 7.000000000000001e-06,
      "loss": 0.5596,
      "step": 1402
    },
    {
      "epoch": 0.07856422891701198,
      "grad_norm": 1.3643035888671875,
      "learning_rate": 7.005000000000001e-06,
      "loss": 0.5343,
      "step": 1403
    },
    {
      "epoch": 0.078620226229141,
      "grad_norm": 1.1445242166519165,
      "learning_rate": 7.01e-06,
      "loss": 0.5881,
      "step": 1404
    },
    {
      "epoch": 0.07867622354127002,
      "grad_norm": 1.1189980506896973,
      "learning_rate": 7.015000000000001e-06,
      "loss": 0.379,
      "step": 1405
    },
    {
      "epoch": 0.07873222085339904,
      "grad_norm": 1.3632959127426147,
      "learning_rate": 7.0200000000000006e-06,
      "loss": 0.4533,
      "step": 1406
    },
    {
      "epoch": 0.07878821816552806,
      "grad_norm": 1.0132023096084595,
      "learning_rate": 7.025000000000001e-06,
      "loss": 0.4028,
      "step": 1407
    },
    {
      "epoch": 0.07884421547765708,
      "grad_norm": 1.574580192565918,
      "learning_rate": 7.0300000000000005e-06,
      "loss": 0.7049,
      "step": 1408
    },
    {
      "epoch": 0.0789002127897861,
      "grad_norm": 1.207859754562378,
      "learning_rate": 7.0349999999999996e-06,
      "loss": 0.3993,
      "step": 1409
    },
    {
      "epoch": 0.07895621010191511,
      "grad_norm": 1.1706397533416748,
      "learning_rate": 7.04e-06,
      "loss": 0.3902,
      "step": 1410
    },
    {
      "epoch": 0.07901220741404412,
      "grad_norm": 1.36715829372406,
      "learning_rate": 7.045e-06,
      "loss": 0.5922,
      "step": 1411
    },
    {
      "epoch": 0.07906820472617314,
      "grad_norm": 3.39699649810791,
      "learning_rate": 7.049999999999999e-06,
      "loss": 0.4428,
      "step": 1412
    },
    {
      "epoch": 0.07912420203830216,
      "grad_norm": 1.1224071979522705,
      "learning_rate": 7.055e-06,
      "loss": 0.4317,
      "step": 1413
    },
    {
      "epoch": 0.07918019935043118,
      "grad_norm": 1.2464991807937622,
      "learning_rate": 7.06e-06,
      "loss": 0.4403,
      "step": 1414
    },
    {
      "epoch": 0.0792361966625602,
      "grad_norm": 1.354768991470337,
      "learning_rate": 7.065000000000001e-06,
      "loss": 0.4857,
      "step": 1415
    },
    {
      "epoch": 0.07929219397468922,
      "grad_norm": 1.5229969024658203,
      "learning_rate": 7.07e-06,
      "loss": 0.5088,
      "step": 1416
    },
    {
      "epoch": 0.07934819128681823,
      "grad_norm": 1.3952661752700806,
      "learning_rate": 7.075e-06,
      "loss": 0.6225,
      "step": 1417
    },
    {
      "epoch": 0.07940418859894725,
      "grad_norm": 1.2597805261611938,
      "learning_rate": 7.080000000000001e-06,
      "loss": 0.3716,
      "step": 1418
    },
    {
      "epoch": 0.07946018591107627,
      "grad_norm": 1.479173183441162,
      "learning_rate": 7.085e-06,
      "loss": 0.6558,
      "step": 1419
    },
    {
      "epoch": 0.07951618322320529,
      "grad_norm": 1.175714135169983,
      "learning_rate": 7.090000000000001e-06,
      "loss": 0.6377,
      "step": 1420
    },
    {
      "epoch": 0.0795721805353343,
      "grad_norm": 1.1564451456069946,
      "learning_rate": 7.095000000000001e-06,
      "loss": 0.5012,
      "step": 1421
    },
    {
      "epoch": 0.07962817784746332,
      "grad_norm": 1.0812937021255493,
      "learning_rate": 7.1e-06,
      "loss": 0.4637,
      "step": 1422
    },
    {
      "epoch": 0.07968417515959234,
      "grad_norm": 1.143661379814148,
      "learning_rate": 7.105000000000001e-06,
      "loss": 0.3334,
      "step": 1423
    },
    {
      "epoch": 0.07974017247172135,
      "grad_norm": 1.3311735391616821,
      "learning_rate": 7.11e-06,
      "loss": 0.4989,
      "step": 1424
    },
    {
      "epoch": 0.07979616978385037,
      "grad_norm": 1.2069625854492188,
      "learning_rate": 7.1150000000000005e-06,
      "loss": 0.3392,
      "step": 1425
    },
    {
      "epoch": 0.07985216709597939,
      "grad_norm": 1.180067539215088,
      "learning_rate": 7.1200000000000004e-06,
      "loss": 0.3681,
      "step": 1426
    },
    {
      "epoch": 0.07990816440810841,
      "grad_norm": 1.2177414894104004,
      "learning_rate": 7.1249999999999995e-06,
      "loss": 0.4858,
      "step": 1427
    },
    {
      "epoch": 0.07996416172023743,
      "grad_norm": 1.3053133487701416,
      "learning_rate": 7.13e-06,
      "loss": 0.4826,
      "step": 1428
    },
    {
      "epoch": 0.08002015903236645,
      "grad_norm": 1.080035924911499,
      "learning_rate": 7.135e-06,
      "loss": 0.3143,
      "step": 1429
    },
    {
      "epoch": 0.08007615634449547,
      "grad_norm": 1.509613275527954,
      "learning_rate": 7.140000000000001e-06,
      "loss": 0.4389,
      "step": 1430
    },
    {
      "epoch": 0.08013215365662449,
      "grad_norm": 1.1446751356124878,
      "learning_rate": 7.145e-06,
      "loss": 0.4379,
      "step": 1431
    },
    {
      "epoch": 0.0801881509687535,
      "grad_norm": 1.4199235439300537,
      "learning_rate": 7.15e-06,
      "loss": 0.4777,
      "step": 1432
    },
    {
      "epoch": 0.08024414828088251,
      "grad_norm": 1.5133382081985474,
      "learning_rate": 7.155000000000001e-06,
      "loss": 0.6484,
      "step": 1433
    },
    {
      "epoch": 0.08030014559301153,
      "grad_norm": 1.3345826864242554,
      "learning_rate": 7.16e-06,
      "loss": 0.4601,
      "step": 1434
    },
    {
      "epoch": 0.08035614290514055,
      "grad_norm": 1.524214267730713,
      "learning_rate": 7.165000000000001e-06,
      "loss": 0.4004,
      "step": 1435
    },
    {
      "epoch": 0.08041214021726957,
      "grad_norm": 1.2183082103729248,
      "learning_rate": 7.17e-06,
      "loss": 0.4514,
      "step": 1436
    },
    {
      "epoch": 0.08046813752939859,
      "grad_norm": 1.2108083963394165,
      "learning_rate": 7.175e-06,
      "loss": 0.4932,
      "step": 1437
    },
    {
      "epoch": 0.08052413484152761,
      "grad_norm": 1.309736728668213,
      "learning_rate": 7.180000000000001e-06,
      "loss": 0.3938,
      "step": 1438
    },
    {
      "epoch": 0.08058013215365663,
      "grad_norm": 0.9972428679466248,
      "learning_rate": 7.185e-06,
      "loss": 0.322,
      "step": 1439
    },
    {
      "epoch": 0.08063612946578565,
      "grad_norm": 1.2740402221679688,
      "learning_rate": 7.190000000000001e-06,
      "loss": 0.531,
      "step": 1440
    },
    {
      "epoch": 0.08069212677791467,
      "grad_norm": 1.4080537557601929,
      "learning_rate": 7.1950000000000006e-06,
      "loss": 0.5996,
      "step": 1441
    },
    {
      "epoch": 0.08074812409004367,
      "grad_norm": 1.7083243131637573,
      "learning_rate": 7.2e-06,
      "loss": 0.6026,
      "step": 1442
    },
    {
      "epoch": 0.08080412140217269,
      "grad_norm": 1.209126591682434,
      "learning_rate": 7.2050000000000005e-06,
      "loss": 0.4386,
      "step": 1443
    },
    {
      "epoch": 0.08086011871430171,
      "grad_norm": 1.3964226245880127,
      "learning_rate": 7.2100000000000004e-06,
      "loss": 0.4701,
      "step": 1444
    },
    {
      "epoch": 0.08091611602643073,
      "grad_norm": 1.3449779748916626,
      "learning_rate": 7.215000000000001e-06,
      "loss": 0.4253,
      "step": 1445
    },
    {
      "epoch": 0.08097211333855975,
      "grad_norm": 1.313974380493164,
      "learning_rate": 7.22e-06,
      "loss": 0.4444,
      "step": 1446
    },
    {
      "epoch": 0.08102811065068877,
      "grad_norm": 1.2853071689605713,
      "learning_rate": 7.2249999999999994e-06,
      "loss": 0.5527,
      "step": 1447
    },
    {
      "epoch": 0.08108410796281779,
      "grad_norm": 1.1168649196624756,
      "learning_rate": 7.230000000000001e-06,
      "loss": 0.4293,
      "step": 1448
    },
    {
      "epoch": 0.0811401052749468,
      "grad_norm": 1.1112141609191895,
      "learning_rate": 7.235e-06,
      "loss": 0.5183,
      "step": 1449
    },
    {
      "epoch": 0.08119610258707582,
      "grad_norm": 1.6637715101242065,
      "learning_rate": 7.240000000000001e-06,
      "loss": 0.4883,
      "step": 1450
    },
    {
      "epoch": 0.08125209989920484,
      "grad_norm": 1.3402663469314575,
      "learning_rate": 7.245e-06,
      "loss": 0.3671,
      "step": 1451
    },
    {
      "epoch": 0.08130809721133385,
      "grad_norm": 1.248765230178833,
      "learning_rate": 7.25e-06,
      "loss": 0.3786,
      "step": 1452
    },
    {
      "epoch": 0.08136409452346287,
      "grad_norm": 1.4092042446136475,
      "learning_rate": 7.255000000000001e-06,
      "loss": 0.4625,
      "step": 1453
    },
    {
      "epoch": 0.08142009183559189,
      "grad_norm": 1.480547547340393,
      "learning_rate": 7.26e-06,
      "loss": 0.6167,
      "step": 1454
    },
    {
      "epoch": 0.0814760891477209,
      "grad_norm": 1.3180210590362549,
      "learning_rate": 7.265000000000001e-06,
      "loss": 0.4995,
      "step": 1455
    },
    {
      "epoch": 0.08153208645984993,
      "grad_norm": 1.1281007528305054,
      "learning_rate": 7.270000000000001e-06,
      "loss": 0.4142,
      "step": 1456
    },
    {
      "epoch": 0.08158808377197894,
      "grad_norm": 1.519881248474121,
      "learning_rate": 7.275e-06,
      "loss": 0.5344,
      "step": 1457
    },
    {
      "epoch": 0.08164408108410796,
      "grad_norm": 1.4132120609283447,
      "learning_rate": 7.280000000000001e-06,
      "loss": 0.5994,
      "step": 1458
    },
    {
      "epoch": 0.08170007839623698,
      "grad_norm": 1.0918807983398438,
      "learning_rate": 7.2850000000000006e-06,
      "loss": 0.3358,
      "step": 1459
    },
    {
      "epoch": 0.081756075708366,
      "grad_norm": 1.292568325996399,
      "learning_rate": 7.290000000000001e-06,
      "loss": 0.4402,
      "step": 1460
    },
    {
      "epoch": 0.08181207302049502,
      "grad_norm": 1.1813576221466064,
      "learning_rate": 7.2950000000000005e-06,
      "loss": 0.4083,
      "step": 1461
    },
    {
      "epoch": 0.08186807033262404,
      "grad_norm": 1.139785647392273,
      "learning_rate": 7.2999999999999996e-06,
      "loss": 0.5056,
      "step": 1462
    },
    {
      "epoch": 0.08192406764475305,
      "grad_norm": 1.500008225440979,
      "learning_rate": 7.305e-06,
      "loss": 0.4989,
      "step": 1463
    },
    {
      "epoch": 0.08198006495688206,
      "grad_norm": 1.4263228178024292,
      "learning_rate": 7.31e-06,
      "loss": 0.4714,
      "step": 1464
    },
    {
      "epoch": 0.08203606226901108,
      "grad_norm": 1.2919131517410278,
      "learning_rate": 7.315000000000001e-06,
      "loss": 0.4517,
      "step": 1465
    },
    {
      "epoch": 0.0820920595811401,
      "grad_norm": 1.3762054443359375,
      "learning_rate": 7.32e-06,
      "loss": 0.4877,
      "step": 1466
    },
    {
      "epoch": 0.08214805689326912,
      "grad_norm": 1.3548452854156494,
      "learning_rate": 7.325e-06,
      "loss": 0.4628,
      "step": 1467
    },
    {
      "epoch": 0.08220405420539814,
      "grad_norm": 1.3240776062011719,
      "learning_rate": 7.330000000000001e-06,
      "loss": 0.5224,
      "step": 1468
    },
    {
      "epoch": 0.08226005151752716,
      "grad_norm": 1.2442258596420288,
      "learning_rate": 7.335e-06,
      "loss": 0.4429,
      "step": 1469
    },
    {
      "epoch": 0.08231604882965618,
      "grad_norm": 1.2015458345413208,
      "learning_rate": 7.340000000000001e-06,
      "loss": 0.4401,
      "step": 1470
    },
    {
      "epoch": 0.0823720461417852,
      "grad_norm": 1.0796226263046265,
      "learning_rate": 7.345000000000001e-06,
      "loss": 0.3873,
      "step": 1471
    },
    {
      "epoch": 0.08242804345391422,
      "grad_norm": 1.2062454223632812,
      "learning_rate": 7.35e-06,
      "loss": 0.4338,
      "step": 1472
    },
    {
      "epoch": 0.08248404076604322,
      "grad_norm": 1.2619911432266235,
      "learning_rate": 7.355000000000001e-06,
      "loss": 0.4712,
      "step": 1473
    },
    {
      "epoch": 0.08254003807817224,
      "grad_norm": 1.6455378532409668,
      "learning_rate": 7.36e-06,
      "loss": 0.4193,
      "step": 1474
    },
    {
      "epoch": 0.08259603539030126,
      "grad_norm": 1.209715485572815,
      "learning_rate": 7.365e-06,
      "loss": 0.5695,
      "step": 1475
    },
    {
      "epoch": 0.08265203270243028,
      "grad_norm": 1.3279260396957397,
      "learning_rate": 7.370000000000001e-06,
      "loss": 0.5192,
      "step": 1476
    },
    {
      "epoch": 0.0827080300145593,
      "grad_norm": 1.1530653238296509,
      "learning_rate": 7.375e-06,
      "loss": 0.3153,
      "step": 1477
    },
    {
      "epoch": 0.08276402732668832,
      "grad_norm": 1.9557344913482666,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 0.6641,
      "step": 1478
    },
    {
      "epoch": 0.08282002463881734,
      "grad_norm": 1.6863083839416504,
      "learning_rate": 7.3850000000000004e-06,
      "loss": 0.566,
      "step": 1479
    },
    {
      "epoch": 0.08287602195094636,
      "grad_norm": 1.282794713973999,
      "learning_rate": 7.3899999999999995e-06,
      "loss": 0.5705,
      "step": 1480
    },
    {
      "epoch": 0.08293201926307538,
      "grad_norm": 1.4731370210647583,
      "learning_rate": 7.395e-06,
      "loss": 0.5086,
      "step": 1481
    },
    {
      "epoch": 0.0829880165752044,
      "grad_norm": 1.4147073030471802,
      "learning_rate": 7.4e-06,
      "loss": 0.5286,
      "step": 1482
    },
    {
      "epoch": 0.0830440138873334,
      "grad_norm": 1.2918366193771362,
      "learning_rate": 7.405000000000001e-06,
      "loss": 0.3988,
      "step": 1483
    },
    {
      "epoch": 0.08310001119946242,
      "grad_norm": 0.9884164929389954,
      "learning_rate": 7.41e-06,
      "loss": 0.4509,
      "step": 1484
    },
    {
      "epoch": 0.08315600851159144,
      "grad_norm": 1.1498831510543823,
      "learning_rate": 7.414999999999999e-06,
      "loss": 0.3887,
      "step": 1485
    },
    {
      "epoch": 0.08321200582372046,
      "grad_norm": 1.2837716341018677,
      "learning_rate": 7.420000000000001e-06,
      "loss": 0.52,
      "step": 1486
    },
    {
      "epoch": 0.08326800313584948,
      "grad_norm": 1.303924560546875,
      "learning_rate": 7.425e-06,
      "loss": 0.5503,
      "step": 1487
    },
    {
      "epoch": 0.0833240004479785,
      "grad_norm": 1.507111668586731,
      "learning_rate": 7.430000000000001e-06,
      "loss": 0.6404,
      "step": 1488
    },
    {
      "epoch": 0.08337999776010752,
      "grad_norm": 1.2963229417800903,
      "learning_rate": 7.435e-06,
      "loss": 0.5778,
      "step": 1489
    },
    {
      "epoch": 0.08343599507223654,
      "grad_norm": 1.512855052947998,
      "learning_rate": 7.44e-06,
      "loss": 0.5745,
      "step": 1490
    },
    {
      "epoch": 0.08349199238436555,
      "grad_norm": 1.5575251579284668,
      "learning_rate": 7.445000000000001e-06,
      "loss": 0.4233,
      "step": 1491
    },
    {
      "epoch": 0.08354798969649457,
      "grad_norm": 1.0272166728973389,
      "learning_rate": 7.45e-06,
      "loss": 0.3971,
      "step": 1492
    },
    {
      "epoch": 0.08360398700862359,
      "grad_norm": 1.458645224571228,
      "learning_rate": 7.455000000000001e-06,
      "loss": 0.5315,
      "step": 1493
    },
    {
      "epoch": 0.0836599843207526,
      "grad_norm": 1.2564878463745117,
      "learning_rate": 7.4600000000000006e-06,
      "loss": 0.498,
      "step": 1494
    },
    {
      "epoch": 0.08371598163288162,
      "grad_norm": 1.3385682106018066,
      "learning_rate": 7.465e-06,
      "loss": 0.6363,
      "step": 1495
    },
    {
      "epoch": 0.08377197894501064,
      "grad_norm": 1.3317980766296387,
      "learning_rate": 7.4700000000000005e-06,
      "loss": 0.4346,
      "step": 1496
    },
    {
      "epoch": 0.08382797625713966,
      "grad_norm": 1.1899913549423218,
      "learning_rate": 7.4750000000000004e-06,
      "loss": 0.5111,
      "step": 1497
    },
    {
      "epoch": 0.08388397356926867,
      "grad_norm": 1.3336282968521118,
      "learning_rate": 7.480000000000001e-06,
      "loss": 0.5527,
      "step": 1498
    },
    {
      "epoch": 0.0839399708813977,
      "grad_norm": 1.2331520318984985,
      "learning_rate": 7.485e-06,
      "loss": 0.4771,
      "step": 1499
    },
    {
      "epoch": 0.08399596819352671,
      "grad_norm": 1.0488063097000122,
      "learning_rate": 7.4899999999999994e-06,
      "loss": 0.4086,
      "step": 1500
    },
    {
      "epoch": 0.08405196550565573,
      "grad_norm": 1.606632947921753,
      "learning_rate": 7.495e-06,
      "loss": 0.5363,
      "step": 1501
    },
    {
      "epoch": 0.08410796281778475,
      "grad_norm": 1.2820321321487427,
      "learning_rate": 7.5e-06,
      "loss": 0.4214,
      "step": 1502
    },
    {
      "epoch": 0.08416396012991377,
      "grad_norm": 1.3019225597381592,
      "learning_rate": 7.505000000000001e-06,
      "loss": 0.5302,
      "step": 1503
    },
    {
      "epoch": 0.08421995744204278,
      "grad_norm": 1.2436977624893188,
      "learning_rate": 7.51e-06,
      "loss": 0.555,
      "step": 1504
    },
    {
      "epoch": 0.0842759547541718,
      "grad_norm": 1.3431636095046997,
      "learning_rate": 7.515e-06,
      "loss": 0.4918,
      "step": 1505
    },
    {
      "epoch": 0.08433195206630081,
      "grad_norm": 1.1182070970535278,
      "learning_rate": 7.520000000000001e-06,
      "loss": 0.4375,
      "step": 1506
    },
    {
      "epoch": 0.08438794937842983,
      "grad_norm": 1.193709135055542,
      "learning_rate": 7.525e-06,
      "loss": 0.3783,
      "step": 1507
    },
    {
      "epoch": 0.08444394669055885,
      "grad_norm": 1.1308382749557495,
      "learning_rate": 7.530000000000001e-06,
      "loss": 0.4268,
      "step": 1508
    },
    {
      "epoch": 0.08449994400268787,
      "grad_norm": 0.9366526007652283,
      "learning_rate": 7.535000000000001e-06,
      "loss": 0.4236,
      "step": 1509
    },
    {
      "epoch": 0.08455594131481689,
      "grad_norm": 1.1923974752426147,
      "learning_rate": 7.54e-06,
      "loss": 0.3839,
      "step": 1510
    },
    {
      "epoch": 0.08461193862694591,
      "grad_norm": 1.1978589296340942,
      "learning_rate": 7.545000000000001e-06,
      "loss": 0.3851,
      "step": 1511
    },
    {
      "epoch": 0.08466793593907493,
      "grad_norm": 1.1501808166503906,
      "learning_rate": 7.55e-06,
      "loss": 0.4263,
      "step": 1512
    },
    {
      "epoch": 0.08472393325120395,
      "grad_norm": 1.415138840675354,
      "learning_rate": 7.555000000000001e-06,
      "loss": 0.5748,
      "step": 1513
    },
    {
      "epoch": 0.08477993056333297,
      "grad_norm": 1.2347288131713867,
      "learning_rate": 7.5600000000000005e-06,
      "loss": 0.4694,
      "step": 1514
    },
    {
      "epoch": 0.08483592787546197,
      "grad_norm": 1.1840616464614868,
      "learning_rate": 7.5649999999999996e-06,
      "loss": 0.5046,
      "step": 1515
    },
    {
      "epoch": 0.08489192518759099,
      "grad_norm": 1.046579122543335,
      "learning_rate": 7.57e-06,
      "loss": 0.4277,
      "step": 1516
    },
    {
      "epoch": 0.08494792249972001,
      "grad_norm": 1.0324609279632568,
      "learning_rate": 7.575e-06,
      "loss": 0.3577,
      "step": 1517
    },
    {
      "epoch": 0.08500391981184903,
      "grad_norm": 1.2944605350494385,
      "learning_rate": 7.580000000000001e-06,
      "loss": 0.6079,
      "step": 1518
    },
    {
      "epoch": 0.08505991712397805,
      "grad_norm": 1.3902816772460938,
      "learning_rate": 7.585e-06,
      "loss": 0.4648,
      "step": 1519
    },
    {
      "epoch": 0.08511591443610707,
      "grad_norm": 1.3388081789016724,
      "learning_rate": 7.59e-06,
      "loss": 0.4821,
      "step": 1520
    },
    {
      "epoch": 0.08517191174823609,
      "grad_norm": 1.249360203742981,
      "learning_rate": 7.595000000000001e-06,
      "loss": 0.4765,
      "step": 1521
    },
    {
      "epoch": 0.0852279090603651,
      "grad_norm": 1.153415560722351,
      "learning_rate": 7.6e-06,
      "loss": 0.3804,
      "step": 1522
    },
    {
      "epoch": 0.08528390637249413,
      "grad_norm": 1.3876703977584839,
      "learning_rate": 7.605000000000001e-06,
      "loss": 0.6177,
      "step": 1523
    },
    {
      "epoch": 0.08533990368462314,
      "grad_norm": 1.1727885007858276,
      "learning_rate": 7.610000000000001e-06,
      "loss": 0.4695,
      "step": 1524
    },
    {
      "epoch": 0.08539590099675215,
      "grad_norm": 1.174000859260559,
      "learning_rate": 7.615e-06,
      "loss": 0.4969,
      "step": 1525
    },
    {
      "epoch": 0.08545189830888117,
      "grad_norm": 1.5841765403747559,
      "learning_rate": 7.620000000000001e-06,
      "loss": 0.3613,
      "step": 1526
    },
    {
      "epoch": 0.08550789562101019,
      "grad_norm": 1.0650402307510376,
      "learning_rate": 7.625e-06,
      "loss": 0.4859,
      "step": 1527
    },
    {
      "epoch": 0.08556389293313921,
      "grad_norm": 1.0996922254562378,
      "learning_rate": 7.630000000000001e-06,
      "loss": 0.4277,
      "step": 1528
    },
    {
      "epoch": 0.08561989024526823,
      "grad_norm": 1.2042404413223267,
      "learning_rate": 7.635e-06,
      "loss": 0.4736,
      "step": 1529
    },
    {
      "epoch": 0.08567588755739725,
      "grad_norm": 1.3029634952545166,
      "learning_rate": 7.64e-06,
      "loss": 0.4859,
      "step": 1530
    },
    {
      "epoch": 0.08573188486952626,
      "grad_norm": 1.5617334842681885,
      "learning_rate": 7.645e-06,
      "loss": 0.5699,
      "step": 1531
    },
    {
      "epoch": 0.08578788218165528,
      "grad_norm": 1.1617907285690308,
      "learning_rate": 7.65e-06,
      "loss": 0.4353,
      "step": 1532
    },
    {
      "epoch": 0.0858438794937843,
      "grad_norm": 1.383780837059021,
      "learning_rate": 7.655e-06,
      "loss": 0.5618,
      "step": 1533
    },
    {
      "epoch": 0.08589987680591332,
      "grad_norm": 1.1526728868484497,
      "learning_rate": 7.660000000000001e-06,
      "loss": 0.4117,
      "step": 1534
    },
    {
      "epoch": 0.08595587411804233,
      "grad_norm": 1.1477508544921875,
      "learning_rate": 7.665e-06,
      "loss": 0.3695,
      "step": 1535
    },
    {
      "epoch": 0.08601187143017135,
      "grad_norm": 1.674328327178955,
      "learning_rate": 7.670000000000001e-06,
      "loss": 0.4579,
      "step": 1536
    },
    {
      "epoch": 0.08606786874230037,
      "grad_norm": 1.103920817375183,
      "learning_rate": 7.675e-06,
      "loss": 0.4256,
      "step": 1537
    },
    {
      "epoch": 0.08612386605442938,
      "grad_norm": 1.1993048191070557,
      "learning_rate": 7.68e-06,
      "loss": 0.4485,
      "step": 1538
    },
    {
      "epoch": 0.0861798633665584,
      "grad_norm": 1.079773187637329,
      "learning_rate": 7.685e-06,
      "loss": 0.3587,
      "step": 1539
    },
    {
      "epoch": 0.08623586067868742,
      "grad_norm": 1.1587178707122803,
      "learning_rate": 7.69e-06,
      "loss": 0.4912,
      "step": 1540
    },
    {
      "epoch": 0.08629185799081644,
      "grad_norm": 1.2015721797943115,
      "learning_rate": 7.695e-06,
      "loss": 0.5274,
      "step": 1541
    },
    {
      "epoch": 0.08634785530294546,
      "grad_norm": 1.0844711065292358,
      "learning_rate": 7.7e-06,
      "loss": 0.3939,
      "step": 1542
    },
    {
      "epoch": 0.08640385261507448,
      "grad_norm": 1.1569340229034424,
      "learning_rate": 7.705e-06,
      "loss": 0.5162,
      "step": 1543
    },
    {
      "epoch": 0.0864598499272035,
      "grad_norm": 1.127315640449524,
      "learning_rate": 7.71e-06,
      "loss": 0.4135,
      "step": 1544
    },
    {
      "epoch": 0.08651584723933252,
      "grad_norm": 1.184778094291687,
      "learning_rate": 7.715e-06,
      "loss": 0.477,
      "step": 1545
    },
    {
      "epoch": 0.08657184455146152,
      "grad_norm": 1.3671724796295166,
      "learning_rate": 7.72e-06,
      "loss": 0.4705,
      "step": 1546
    },
    {
      "epoch": 0.08662784186359054,
      "grad_norm": 1.2034422159194946,
      "learning_rate": 7.725e-06,
      "loss": 0.4117,
      "step": 1547
    },
    {
      "epoch": 0.08668383917571956,
      "grad_norm": 2.615835428237915,
      "learning_rate": 7.73e-06,
      "loss": 0.4523,
      "step": 1548
    },
    {
      "epoch": 0.08673983648784858,
      "grad_norm": 1.4498186111450195,
      "learning_rate": 7.735000000000001e-06,
      "loss": 0.4638,
      "step": 1549
    },
    {
      "epoch": 0.0867958337999776,
      "grad_norm": 1.169029951095581,
      "learning_rate": 7.74e-06,
      "loss": 0.3688,
      "step": 1550
    },
    {
      "epoch": 0.08685183111210662,
      "grad_norm": 1.2343640327453613,
      "learning_rate": 7.745000000000001e-06,
      "loss": 0.476,
      "step": 1551
    },
    {
      "epoch": 0.08690782842423564,
      "grad_norm": 1.3974204063415527,
      "learning_rate": 7.75e-06,
      "loss": 0.422,
      "step": 1552
    },
    {
      "epoch": 0.08696382573636466,
      "grad_norm": 1.21696937084198,
      "learning_rate": 7.755e-06,
      "loss": 0.4368,
      "step": 1553
    },
    {
      "epoch": 0.08701982304849368,
      "grad_norm": 1.798683524131775,
      "learning_rate": 7.76e-06,
      "loss": 0.4386,
      "step": 1554
    },
    {
      "epoch": 0.0870758203606227,
      "grad_norm": 1.2375297546386719,
      "learning_rate": 7.765e-06,
      "loss": 0.6109,
      "step": 1555
    },
    {
      "epoch": 0.0871318176727517,
      "grad_norm": 1.1786304712295532,
      "learning_rate": 7.77e-06,
      "loss": 0.4902,
      "step": 1556
    },
    {
      "epoch": 0.08718781498488072,
      "grad_norm": 1.4570947885513306,
      "learning_rate": 7.775000000000001e-06,
      "loss": 0.7328,
      "step": 1557
    },
    {
      "epoch": 0.08724381229700974,
      "grad_norm": 1.12009859085083,
      "learning_rate": 7.78e-06,
      "loss": 0.4181,
      "step": 1558
    },
    {
      "epoch": 0.08729980960913876,
      "grad_norm": 1.328429102897644,
      "learning_rate": 7.785000000000001e-06,
      "loss": 0.7427,
      "step": 1559
    },
    {
      "epoch": 0.08735580692126778,
      "grad_norm": 1.4115002155303955,
      "learning_rate": 7.79e-06,
      "loss": 0.4477,
      "step": 1560
    },
    {
      "epoch": 0.0874118042333968,
      "grad_norm": 1.53958261013031,
      "learning_rate": 7.795e-06,
      "loss": 0.5461,
      "step": 1561
    },
    {
      "epoch": 0.08746780154552582,
      "grad_norm": 2.6513330936431885,
      "learning_rate": 7.8e-06,
      "loss": 0.5128,
      "step": 1562
    },
    {
      "epoch": 0.08752379885765484,
      "grad_norm": 1.325663685798645,
      "learning_rate": 7.805e-06,
      "loss": 0.5435,
      "step": 1563
    },
    {
      "epoch": 0.08757979616978386,
      "grad_norm": 0.9753627777099609,
      "learning_rate": 7.810000000000001e-06,
      "loss": 0.3439,
      "step": 1564
    },
    {
      "epoch": 0.08763579348191287,
      "grad_norm": 1.3109617233276367,
      "learning_rate": 7.815e-06,
      "loss": 0.4922,
      "step": 1565
    },
    {
      "epoch": 0.08769179079404188,
      "grad_norm": 1.272173523902893,
      "learning_rate": 7.820000000000001e-06,
      "loss": 0.4498,
      "step": 1566
    },
    {
      "epoch": 0.0877477881061709,
      "grad_norm": 1.77103853225708,
      "learning_rate": 7.825e-06,
      "loss": 0.7537,
      "step": 1567
    },
    {
      "epoch": 0.08780378541829992,
      "grad_norm": 1.1595690250396729,
      "learning_rate": 7.83e-06,
      "loss": 0.4236,
      "step": 1568
    },
    {
      "epoch": 0.08785978273042894,
      "grad_norm": 1.303318738937378,
      "learning_rate": 7.835e-06,
      "loss": 0.4757,
      "step": 1569
    },
    {
      "epoch": 0.08791578004255796,
      "grad_norm": 1.357015609741211,
      "learning_rate": 7.84e-06,
      "loss": 0.659,
      "step": 1570
    },
    {
      "epoch": 0.08797177735468698,
      "grad_norm": 1.5197786092758179,
      "learning_rate": 7.845e-06,
      "loss": 0.5869,
      "step": 1571
    },
    {
      "epoch": 0.088027774666816,
      "grad_norm": 1.1078104972839355,
      "learning_rate": 7.850000000000001e-06,
      "loss": 0.3845,
      "step": 1572
    },
    {
      "epoch": 0.08808377197894501,
      "grad_norm": 1.304028868675232,
      "learning_rate": 7.855e-06,
      "loss": 0.4724,
      "step": 1573
    },
    {
      "epoch": 0.08813976929107403,
      "grad_norm": 1.3804655075073242,
      "learning_rate": 7.860000000000001e-06,
      "loss": 0.4343,
      "step": 1574
    },
    {
      "epoch": 0.08819576660320305,
      "grad_norm": 1.1329188346862793,
      "learning_rate": 7.865e-06,
      "loss": 0.4534,
      "step": 1575
    },
    {
      "epoch": 0.08825176391533207,
      "grad_norm": 1.2403630018234253,
      "learning_rate": 7.870000000000001e-06,
      "loss": 0.4721,
      "step": 1576
    },
    {
      "epoch": 0.08830776122746108,
      "grad_norm": 1.1998299360275269,
      "learning_rate": 7.875e-06,
      "loss": 0.6336,
      "step": 1577
    },
    {
      "epoch": 0.0883637585395901,
      "grad_norm": 1.285543441772461,
      "learning_rate": 7.879999999999999e-06,
      "loss": 0.5604,
      "step": 1578
    },
    {
      "epoch": 0.08841975585171911,
      "grad_norm": 1.1370717287063599,
      "learning_rate": 7.885e-06,
      "loss": 0.3956,
      "step": 1579
    },
    {
      "epoch": 0.08847575316384813,
      "grad_norm": 1.230810284614563,
      "learning_rate": 7.89e-06,
      "loss": 0.5079,
      "step": 1580
    },
    {
      "epoch": 0.08853175047597715,
      "grad_norm": 2.2659618854522705,
      "learning_rate": 7.895000000000001e-06,
      "loss": 0.5141,
      "step": 1581
    },
    {
      "epoch": 0.08858774778810617,
      "grad_norm": 1.1236450672149658,
      "learning_rate": 7.9e-06,
      "loss": 0.3948,
      "step": 1582
    },
    {
      "epoch": 0.08864374510023519,
      "grad_norm": 1.294722557067871,
      "learning_rate": 7.905e-06,
      "loss": 0.444,
      "step": 1583
    },
    {
      "epoch": 0.08869974241236421,
      "grad_norm": 1.1199108362197876,
      "learning_rate": 7.91e-06,
      "loss": 0.5846,
      "step": 1584
    },
    {
      "epoch": 0.08875573972449323,
      "grad_norm": 1.7658997774124146,
      "learning_rate": 7.915e-06,
      "loss": 0.649,
      "step": 1585
    },
    {
      "epoch": 0.08881173703662225,
      "grad_norm": 1.241074800491333,
      "learning_rate": 7.92e-06,
      "loss": 0.4549,
      "step": 1586
    },
    {
      "epoch": 0.08886773434875125,
      "grad_norm": 1.1961077451705933,
      "learning_rate": 7.925000000000001e-06,
      "loss": 0.4492,
      "step": 1587
    },
    {
      "epoch": 0.08892373166088027,
      "grad_norm": 1.2907549142837524,
      "learning_rate": 7.93e-06,
      "loss": 0.3696,
      "step": 1588
    },
    {
      "epoch": 0.08897972897300929,
      "grad_norm": 1.470672845840454,
      "learning_rate": 7.935000000000001e-06,
      "loss": 0.4875,
      "step": 1589
    },
    {
      "epoch": 0.08903572628513831,
      "grad_norm": 1.1282093524932861,
      "learning_rate": 7.94e-06,
      "loss": 0.348,
      "step": 1590
    },
    {
      "epoch": 0.08909172359726733,
      "grad_norm": 1.446938395500183,
      "learning_rate": 7.945000000000001e-06,
      "loss": 0.4021,
      "step": 1591
    },
    {
      "epoch": 0.08914772090939635,
      "grad_norm": 1.3912911415100098,
      "learning_rate": 7.95e-06,
      "loss": 0.6479,
      "step": 1592
    },
    {
      "epoch": 0.08920371822152537,
      "grad_norm": 1.2886724472045898,
      "learning_rate": 7.955e-06,
      "loss": 0.4321,
      "step": 1593
    },
    {
      "epoch": 0.08925971553365439,
      "grad_norm": 1.1775058507919312,
      "learning_rate": 7.96e-06,
      "loss": 0.5141,
      "step": 1594
    },
    {
      "epoch": 0.08931571284578341,
      "grad_norm": 1.33034348487854,
      "learning_rate": 7.965e-06,
      "loss": 0.4683,
      "step": 1595
    },
    {
      "epoch": 0.08937171015791243,
      "grad_norm": 1.1250934600830078,
      "learning_rate": 7.97e-06,
      "loss": 0.5636,
      "step": 1596
    },
    {
      "epoch": 0.08942770747004143,
      "grad_norm": 1.177208423614502,
      "learning_rate": 7.975e-06,
      "loss": 0.4972,
      "step": 1597
    },
    {
      "epoch": 0.08948370478217045,
      "grad_norm": 1.2766159772872925,
      "learning_rate": 7.98e-06,
      "loss": 0.4352,
      "step": 1598
    },
    {
      "epoch": 0.08953970209429947,
      "grad_norm": 1.3314900398254395,
      "learning_rate": 7.985e-06,
      "loss": 0.5088,
      "step": 1599
    },
    {
      "epoch": 0.08959569940642849,
      "grad_norm": 1.1538054943084717,
      "learning_rate": 7.99e-06,
      "loss": 0.4706,
      "step": 1600
    },
    {
      "epoch": 0.08965169671855751,
      "grad_norm": 1.2058525085449219,
      "learning_rate": 7.995e-06,
      "loss": 0.4178,
      "step": 1601
    },
    {
      "epoch": 0.08970769403068653,
      "grad_norm": 1.1523019075393677,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.5341,
      "step": 1602
    },
    {
      "epoch": 0.08976369134281555,
      "grad_norm": 1.2555052042007446,
      "learning_rate": 8.005e-06,
      "loss": 0.4499,
      "step": 1603
    },
    {
      "epoch": 0.08981968865494457,
      "grad_norm": 1.2619866132736206,
      "learning_rate": 8.010000000000001e-06,
      "loss": 0.4104,
      "step": 1604
    },
    {
      "epoch": 0.08987568596707358,
      "grad_norm": 1.5733071565628052,
      "learning_rate": 8.015e-06,
      "loss": 0.4929,
      "step": 1605
    },
    {
      "epoch": 0.0899316832792026,
      "grad_norm": 1.2147235870361328,
      "learning_rate": 8.02e-06,
      "loss": 0.5144,
      "step": 1606
    },
    {
      "epoch": 0.08998768059133162,
      "grad_norm": 1.2371256351470947,
      "learning_rate": 8.025e-06,
      "loss": 0.4795,
      "step": 1607
    },
    {
      "epoch": 0.09004367790346063,
      "grad_norm": 1.2636479139328003,
      "learning_rate": 8.03e-06,
      "loss": 0.4319,
      "step": 1608
    },
    {
      "epoch": 0.09009967521558965,
      "grad_norm": 1.4651589393615723,
      "learning_rate": 8.035e-06,
      "loss": 0.6121,
      "step": 1609
    },
    {
      "epoch": 0.09015567252771867,
      "grad_norm": 1.4587225914001465,
      "learning_rate": 8.040000000000001e-06,
      "loss": 0.3772,
      "step": 1610
    },
    {
      "epoch": 0.09021166983984769,
      "grad_norm": 1.1905932426452637,
      "learning_rate": 8.045e-06,
      "loss": 0.464,
      "step": 1611
    },
    {
      "epoch": 0.0902676671519767,
      "grad_norm": 1.5001569986343384,
      "learning_rate": 8.050000000000001e-06,
      "loss": 0.4785,
      "step": 1612
    },
    {
      "epoch": 0.09032366446410572,
      "grad_norm": 1.0893656015396118,
      "learning_rate": 8.055e-06,
      "loss": 0.4889,
      "step": 1613
    },
    {
      "epoch": 0.09037966177623474,
      "grad_norm": 1.2985970973968506,
      "learning_rate": 8.06e-06,
      "loss": 0.4723,
      "step": 1614
    },
    {
      "epoch": 0.09043565908836376,
      "grad_norm": 1.1527200937271118,
      "learning_rate": 8.065e-06,
      "loss": 0.4792,
      "step": 1615
    },
    {
      "epoch": 0.09049165640049278,
      "grad_norm": 1.253238558769226,
      "learning_rate": 8.069999999999999e-06,
      "loss": 0.4372,
      "step": 1616
    },
    {
      "epoch": 0.0905476537126218,
      "grad_norm": 1.2246673107147217,
      "learning_rate": 8.075000000000001e-06,
      "loss": 0.4444,
      "step": 1617
    },
    {
      "epoch": 0.0906036510247508,
      "grad_norm": 0.9724764823913574,
      "learning_rate": 8.08e-06,
      "loss": 0.426,
      "step": 1618
    },
    {
      "epoch": 0.09065964833687983,
      "grad_norm": 1.9928661584854126,
      "learning_rate": 8.085000000000001e-06,
      "loss": 0.5301,
      "step": 1619
    },
    {
      "epoch": 0.09071564564900884,
      "grad_norm": 1.286141276359558,
      "learning_rate": 8.09e-06,
      "loss": 0.5638,
      "step": 1620
    },
    {
      "epoch": 0.09077164296113786,
      "grad_norm": 1.0990967750549316,
      "learning_rate": 8.095e-06,
      "loss": 0.421,
      "step": 1621
    },
    {
      "epoch": 0.09082764027326688,
      "grad_norm": 1.2366746664047241,
      "learning_rate": 8.1e-06,
      "loss": 0.4869,
      "step": 1622
    },
    {
      "epoch": 0.0908836375853959,
      "grad_norm": 1.4013638496398926,
      "learning_rate": 8.105e-06,
      "loss": 0.4185,
      "step": 1623
    },
    {
      "epoch": 0.09093963489752492,
      "grad_norm": 1.1708909273147583,
      "learning_rate": 8.11e-06,
      "loss": 0.4643,
      "step": 1624
    },
    {
      "epoch": 0.09099563220965394,
      "grad_norm": 1.323397159576416,
      "learning_rate": 8.115000000000001e-06,
      "loss": 0.4741,
      "step": 1625
    },
    {
      "epoch": 0.09105162952178296,
      "grad_norm": 1.1295795440673828,
      "learning_rate": 8.12e-06,
      "loss": 0.4907,
      "step": 1626
    },
    {
      "epoch": 0.09110762683391198,
      "grad_norm": 1.4290260076522827,
      "learning_rate": 8.125000000000001e-06,
      "loss": 0.3986,
      "step": 1627
    },
    {
      "epoch": 0.09116362414604098,
      "grad_norm": 1.2899596691131592,
      "learning_rate": 8.13e-06,
      "loss": 0.4426,
      "step": 1628
    },
    {
      "epoch": 0.09121962145817,
      "grad_norm": 1.6422119140625,
      "learning_rate": 8.135000000000001e-06,
      "loss": 0.5782,
      "step": 1629
    },
    {
      "epoch": 0.09127561877029902,
      "grad_norm": 1.2225149869918823,
      "learning_rate": 8.14e-06,
      "loss": 0.4683,
      "step": 1630
    },
    {
      "epoch": 0.09133161608242804,
      "grad_norm": 1.1593647003173828,
      "learning_rate": 8.144999999999999e-06,
      "loss": 0.445,
      "step": 1631
    },
    {
      "epoch": 0.09138761339455706,
      "grad_norm": 1.1714057922363281,
      "learning_rate": 8.15e-06,
      "loss": 0.4355,
      "step": 1632
    },
    {
      "epoch": 0.09144361070668608,
      "grad_norm": 1.0047953128814697,
      "learning_rate": 8.155e-06,
      "loss": 0.4127,
      "step": 1633
    },
    {
      "epoch": 0.0914996080188151,
      "grad_norm": 1.2258692979812622,
      "learning_rate": 8.160000000000001e-06,
      "loss": 0.5346,
      "step": 1634
    },
    {
      "epoch": 0.09155560533094412,
      "grad_norm": 1.2532273530960083,
      "learning_rate": 8.165e-06,
      "loss": 0.4842,
      "step": 1635
    },
    {
      "epoch": 0.09161160264307314,
      "grad_norm": 1.0819934606552124,
      "learning_rate": 8.17e-06,
      "loss": 0.495,
      "step": 1636
    },
    {
      "epoch": 0.09166759995520216,
      "grad_norm": 1.4493647813796997,
      "learning_rate": 8.175e-06,
      "loss": 0.5362,
      "step": 1637
    },
    {
      "epoch": 0.09172359726733118,
      "grad_norm": 1.7464864253997803,
      "learning_rate": 8.18e-06,
      "loss": 0.4703,
      "step": 1638
    },
    {
      "epoch": 0.09177959457946018,
      "grad_norm": 1.2716431617736816,
      "learning_rate": 8.185e-06,
      "loss": 0.5344,
      "step": 1639
    },
    {
      "epoch": 0.0918355918915892,
      "grad_norm": 1.320268154144287,
      "learning_rate": 8.190000000000001e-06,
      "loss": 0.3806,
      "step": 1640
    },
    {
      "epoch": 0.09189158920371822,
      "grad_norm": 1.4997364282608032,
      "learning_rate": 8.195e-06,
      "loss": 0.4835,
      "step": 1641
    },
    {
      "epoch": 0.09194758651584724,
      "grad_norm": 1.2244665622711182,
      "learning_rate": 8.200000000000001e-06,
      "loss": 0.4452,
      "step": 1642
    },
    {
      "epoch": 0.09200358382797626,
      "grad_norm": 1.1460756063461304,
      "learning_rate": 8.205e-06,
      "loss": 0.4294,
      "step": 1643
    },
    {
      "epoch": 0.09205958114010528,
      "grad_norm": 1.1829346418380737,
      "learning_rate": 8.210000000000001e-06,
      "loss": 0.4801,
      "step": 1644
    },
    {
      "epoch": 0.0921155784522343,
      "grad_norm": 1.3419283628463745,
      "learning_rate": 8.215e-06,
      "loss": 0.5275,
      "step": 1645
    },
    {
      "epoch": 0.09217157576436331,
      "grad_norm": 1.430458426475525,
      "learning_rate": 8.22e-06,
      "loss": 0.5378,
      "step": 1646
    },
    {
      "epoch": 0.09222757307649233,
      "grad_norm": 1.3106675148010254,
      "learning_rate": 8.225e-06,
      "loss": 0.4741,
      "step": 1647
    },
    {
      "epoch": 0.09228357038862135,
      "grad_norm": 1.2035247087478638,
      "learning_rate": 8.23e-06,
      "loss": 0.5643,
      "step": 1648
    },
    {
      "epoch": 0.09233956770075036,
      "grad_norm": 1.4303395748138428,
      "learning_rate": 8.235000000000002e-06,
      "loss": 0.4945,
      "step": 1649
    },
    {
      "epoch": 0.09239556501287938,
      "grad_norm": 1.3392964601516724,
      "learning_rate": 8.24e-06,
      "loss": 0.3851,
      "step": 1650
    },
    {
      "epoch": 0.0924515623250084,
      "grad_norm": 1.2980291843414307,
      "learning_rate": 8.245e-06,
      "loss": 0.5338,
      "step": 1651
    },
    {
      "epoch": 0.09250755963713742,
      "grad_norm": 1.2684897184371948,
      "learning_rate": 8.25e-06,
      "loss": 0.4215,
      "step": 1652
    },
    {
      "epoch": 0.09256355694926643,
      "grad_norm": 1.3237746953964233,
      "learning_rate": 8.255e-06,
      "loss": 0.5577,
      "step": 1653
    },
    {
      "epoch": 0.09261955426139545,
      "grad_norm": 1.4197481870651245,
      "learning_rate": 8.26e-06,
      "loss": 0.4287,
      "step": 1654
    },
    {
      "epoch": 0.09267555157352447,
      "grad_norm": 1.2387175559997559,
      "learning_rate": 8.265000000000001e-06,
      "loss": 0.5273,
      "step": 1655
    },
    {
      "epoch": 0.09273154888565349,
      "grad_norm": 20.208898544311523,
      "learning_rate": 8.27e-06,
      "loss": 0.4626,
      "step": 1656
    },
    {
      "epoch": 0.09278754619778251,
      "grad_norm": 1.0856859683990479,
      "learning_rate": 8.275000000000001e-06,
      "loss": 0.4981,
      "step": 1657
    },
    {
      "epoch": 0.09284354350991153,
      "grad_norm": 1.3090143203735352,
      "learning_rate": 8.28e-06,
      "loss": 0.5757,
      "step": 1658
    },
    {
      "epoch": 0.09289954082204054,
      "grad_norm": 1.2446850538253784,
      "learning_rate": 8.285e-06,
      "loss": 0.4142,
      "step": 1659
    },
    {
      "epoch": 0.09295553813416955,
      "grad_norm": 1.3336029052734375,
      "learning_rate": 8.29e-06,
      "loss": 0.395,
      "step": 1660
    },
    {
      "epoch": 0.09301153544629857,
      "grad_norm": 1.2740191221237183,
      "learning_rate": 8.295e-06,
      "loss": 0.4564,
      "step": 1661
    },
    {
      "epoch": 0.09306753275842759,
      "grad_norm": 1.667121410369873,
      "learning_rate": 8.3e-06,
      "loss": 0.4583,
      "step": 1662
    },
    {
      "epoch": 0.09312353007055661,
      "grad_norm": 1.074404001235962,
      "learning_rate": 8.305000000000001e-06,
      "loss": 0.4812,
      "step": 1663
    },
    {
      "epoch": 0.09317952738268563,
      "grad_norm": 1.130149245262146,
      "learning_rate": 8.31e-06,
      "loss": 0.4388,
      "step": 1664
    },
    {
      "epoch": 0.09323552469481465,
      "grad_norm": 1.5548359155654907,
      "learning_rate": 8.315000000000001e-06,
      "loss": 0.4344,
      "step": 1665
    },
    {
      "epoch": 0.09329152200694367,
      "grad_norm": 2.0083305835723877,
      "learning_rate": 8.32e-06,
      "loss": 0.4691,
      "step": 1666
    },
    {
      "epoch": 0.09334751931907269,
      "grad_norm": 1.2877039909362793,
      "learning_rate": 8.325e-06,
      "loss": 0.6106,
      "step": 1667
    },
    {
      "epoch": 0.09340351663120171,
      "grad_norm": 1.2804948091506958,
      "learning_rate": 8.33e-06,
      "loss": 0.4049,
      "step": 1668
    },
    {
      "epoch": 0.09345951394333073,
      "grad_norm": 1.088975429534912,
      "learning_rate": 8.334999999999999e-06,
      "loss": 0.3796,
      "step": 1669
    },
    {
      "epoch": 0.09351551125545973,
      "grad_norm": 1.1175159215927124,
      "learning_rate": 8.34e-06,
      "loss": 0.4282,
      "step": 1670
    },
    {
      "epoch": 0.09357150856758875,
      "grad_norm": 1.0835729837417603,
      "learning_rate": 8.345e-06,
      "loss": 0.372,
      "step": 1671
    },
    {
      "epoch": 0.09362750587971777,
      "grad_norm": 1.7679165601730347,
      "learning_rate": 8.350000000000001e-06,
      "loss": 0.9209,
      "step": 1672
    },
    {
      "epoch": 0.09368350319184679,
      "grad_norm": 1.1253948211669922,
      "learning_rate": 8.355e-06,
      "loss": 0.4427,
      "step": 1673
    },
    {
      "epoch": 0.09373950050397581,
      "grad_norm": 1.2487722635269165,
      "learning_rate": 8.36e-06,
      "loss": 0.4141,
      "step": 1674
    },
    {
      "epoch": 0.09379549781610483,
      "grad_norm": 1.62444269657135,
      "learning_rate": 8.365e-06,
      "loss": 0.3576,
      "step": 1675
    },
    {
      "epoch": 0.09385149512823385,
      "grad_norm": 1.2208722829818726,
      "learning_rate": 8.37e-06,
      "loss": 0.4182,
      "step": 1676
    },
    {
      "epoch": 0.09390749244036287,
      "grad_norm": 1.2999380826950073,
      "learning_rate": 8.375e-06,
      "loss": 0.4973,
      "step": 1677
    },
    {
      "epoch": 0.09396348975249189,
      "grad_norm": 1.1313492059707642,
      "learning_rate": 8.380000000000001e-06,
      "loss": 0.5136,
      "step": 1678
    },
    {
      "epoch": 0.0940194870646209,
      "grad_norm": 1.0610212087631226,
      "learning_rate": 8.385e-06,
      "loss": 0.3811,
      "step": 1679
    },
    {
      "epoch": 0.09407548437674991,
      "grad_norm": 1.2607392072677612,
      "learning_rate": 8.390000000000001e-06,
      "loss": 0.4823,
      "step": 1680
    },
    {
      "epoch": 0.09413148168887893,
      "grad_norm": 1.3894352912902832,
      "learning_rate": 8.395e-06,
      "loss": 0.5465,
      "step": 1681
    },
    {
      "epoch": 0.09418747900100795,
      "grad_norm": 1.0513899326324463,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.4264,
      "step": 1682
    },
    {
      "epoch": 0.09424347631313697,
      "grad_norm": 0.9940479397773743,
      "learning_rate": 8.405e-06,
      "loss": 0.3418,
      "step": 1683
    },
    {
      "epoch": 0.09429947362526599,
      "grad_norm": 1.2676581144332886,
      "learning_rate": 8.409999999999999e-06,
      "loss": 0.4437,
      "step": 1684
    },
    {
      "epoch": 0.094355470937395,
      "grad_norm": 1.2914938926696777,
      "learning_rate": 8.415e-06,
      "loss": 0.5123,
      "step": 1685
    },
    {
      "epoch": 0.09441146824952402,
      "grad_norm": 1.2333265542984009,
      "learning_rate": 8.42e-06,
      "loss": 0.5305,
      "step": 1686
    },
    {
      "epoch": 0.09446746556165304,
      "grad_norm": 1.0795105695724487,
      "learning_rate": 8.425000000000001e-06,
      "loss": 0.441,
      "step": 1687
    },
    {
      "epoch": 0.09452346287378206,
      "grad_norm": 1.2224481105804443,
      "learning_rate": 8.43e-06,
      "loss": 0.3985,
      "step": 1688
    },
    {
      "epoch": 0.09457946018591108,
      "grad_norm": 1.626680612564087,
      "learning_rate": 8.435e-06,
      "loss": 0.5514,
      "step": 1689
    },
    {
      "epoch": 0.09463545749804009,
      "grad_norm": 1.2383500337600708,
      "learning_rate": 8.44e-06,
      "loss": 0.4813,
      "step": 1690
    },
    {
      "epoch": 0.0946914548101691,
      "grad_norm": 1.3884074687957764,
      "learning_rate": 8.445e-06,
      "loss": 0.4207,
      "step": 1691
    },
    {
      "epoch": 0.09474745212229813,
      "grad_norm": 1.1471067667007446,
      "learning_rate": 8.45e-06,
      "loss": 0.4253,
      "step": 1692
    },
    {
      "epoch": 0.09480344943442715,
      "grad_norm": 1.3975670337677002,
      "learning_rate": 8.455000000000001e-06,
      "loss": 0.5008,
      "step": 1693
    },
    {
      "epoch": 0.09485944674655616,
      "grad_norm": 1.1151567697525024,
      "learning_rate": 8.46e-06,
      "loss": 0.3075,
      "step": 1694
    },
    {
      "epoch": 0.09491544405868518,
      "grad_norm": 0.992013692855835,
      "learning_rate": 8.465000000000001e-06,
      "loss": 0.348,
      "step": 1695
    },
    {
      "epoch": 0.0949714413708142,
      "grad_norm": 1.1896450519561768,
      "learning_rate": 8.47e-06,
      "loss": 0.5221,
      "step": 1696
    },
    {
      "epoch": 0.09502743868294322,
      "grad_norm": 1.4490216970443726,
      "learning_rate": 8.475000000000001e-06,
      "loss": 0.4325,
      "step": 1697
    },
    {
      "epoch": 0.09508343599507224,
      "grad_norm": 1.1228501796722412,
      "learning_rate": 8.48e-06,
      "loss": 0.4076,
      "step": 1698
    },
    {
      "epoch": 0.09513943330720126,
      "grad_norm": 1.3249083757400513,
      "learning_rate": 8.485e-06,
      "loss": 0.456,
      "step": 1699
    },
    {
      "epoch": 0.09519543061933028,
      "grad_norm": 1.3075578212738037,
      "learning_rate": 8.49e-06,
      "loss": 0.5684,
      "step": 1700
    },
    {
      "epoch": 0.09525142793145928,
      "grad_norm": 1.3035677671432495,
      "learning_rate": 8.495e-06,
      "loss": 0.5176,
      "step": 1701
    },
    {
      "epoch": 0.0953074252435883,
      "grad_norm": 1.1941081285476685,
      "learning_rate": 8.500000000000002e-06,
      "loss": 0.4353,
      "step": 1702
    },
    {
      "epoch": 0.09536342255571732,
      "grad_norm": 1.3645747900009155,
      "learning_rate": 8.505e-06,
      "loss": 0.4457,
      "step": 1703
    },
    {
      "epoch": 0.09541941986784634,
      "grad_norm": 1.3234845399856567,
      "learning_rate": 8.51e-06,
      "loss": 0.5028,
      "step": 1704
    },
    {
      "epoch": 0.09547541717997536,
      "grad_norm": 1.5293045043945312,
      "learning_rate": 8.515e-06,
      "loss": 0.6054,
      "step": 1705
    },
    {
      "epoch": 0.09553141449210438,
      "grad_norm": 1.2742122411727905,
      "learning_rate": 8.52e-06,
      "loss": 0.3425,
      "step": 1706
    },
    {
      "epoch": 0.0955874118042334,
      "grad_norm": 1.2540256977081299,
      "learning_rate": 8.525e-06,
      "loss": 0.4417,
      "step": 1707
    },
    {
      "epoch": 0.09564340911636242,
      "grad_norm": 1.1341816186904907,
      "learning_rate": 8.53e-06,
      "loss": 0.4395,
      "step": 1708
    },
    {
      "epoch": 0.09569940642849144,
      "grad_norm": 1.4354275465011597,
      "learning_rate": 8.535e-06,
      "loss": 0.5021,
      "step": 1709
    },
    {
      "epoch": 0.09575540374062046,
      "grad_norm": 1.21577787399292,
      "learning_rate": 8.540000000000001e-06,
      "loss": 0.4162,
      "step": 1710
    },
    {
      "epoch": 0.09581140105274946,
      "grad_norm": 1.385469913482666,
      "learning_rate": 8.545e-06,
      "loss": 0.3978,
      "step": 1711
    },
    {
      "epoch": 0.09586739836487848,
      "grad_norm": 1.0504804849624634,
      "learning_rate": 8.550000000000001e-06,
      "loss": 0.46,
      "step": 1712
    },
    {
      "epoch": 0.0959233956770075,
      "grad_norm": 1.2518787384033203,
      "learning_rate": 8.555e-06,
      "loss": 0.4717,
      "step": 1713
    },
    {
      "epoch": 0.09597939298913652,
      "grad_norm": 1.2391563653945923,
      "learning_rate": 8.56e-06,
      "loss": 0.3796,
      "step": 1714
    },
    {
      "epoch": 0.09603539030126554,
      "grad_norm": 1.352744460105896,
      "learning_rate": 8.565e-06,
      "loss": 0.4809,
      "step": 1715
    },
    {
      "epoch": 0.09609138761339456,
      "grad_norm": 1.2199307680130005,
      "learning_rate": 8.570000000000001e-06,
      "loss": 0.5013,
      "step": 1716
    },
    {
      "epoch": 0.09614738492552358,
      "grad_norm": 2.9449145793914795,
      "learning_rate": 8.575000000000002e-06,
      "loss": 0.4527,
      "step": 1717
    },
    {
      "epoch": 0.0962033822376526,
      "grad_norm": 1.0867520570755005,
      "learning_rate": 8.580000000000001e-06,
      "loss": 0.5052,
      "step": 1718
    },
    {
      "epoch": 0.09625937954978162,
      "grad_norm": 1.5014264583587646,
      "learning_rate": 8.585e-06,
      "loss": 0.4327,
      "step": 1719
    },
    {
      "epoch": 0.09631537686191063,
      "grad_norm": 1.358459711074829,
      "learning_rate": 8.59e-06,
      "loss": 0.4432,
      "step": 1720
    },
    {
      "epoch": 0.09637137417403964,
      "grad_norm": 1.1366547346115112,
      "learning_rate": 8.595e-06,
      "loss": 0.4062,
      "step": 1721
    },
    {
      "epoch": 0.09642737148616866,
      "grad_norm": 1.6252096891403198,
      "learning_rate": 8.599999999999999e-06,
      "loss": 0.4598,
      "step": 1722
    },
    {
      "epoch": 0.09648336879829768,
      "grad_norm": 1.2072232961654663,
      "learning_rate": 8.605e-06,
      "loss": 0.49,
      "step": 1723
    },
    {
      "epoch": 0.0965393661104267,
      "grad_norm": 1.2367745637893677,
      "learning_rate": 8.61e-06,
      "loss": 0.4035,
      "step": 1724
    },
    {
      "epoch": 0.09659536342255572,
      "grad_norm": 1.3461731672286987,
      "learning_rate": 8.615000000000001e-06,
      "loss": 0.4942,
      "step": 1725
    },
    {
      "epoch": 0.09665136073468474,
      "grad_norm": 1.1359554529190063,
      "learning_rate": 8.62e-06,
      "loss": 0.4236,
      "step": 1726
    },
    {
      "epoch": 0.09670735804681375,
      "grad_norm": 1.090421438217163,
      "learning_rate": 8.625e-06,
      "loss": 0.4333,
      "step": 1727
    },
    {
      "epoch": 0.09676335535894277,
      "grad_norm": 5.683356285095215,
      "learning_rate": 8.63e-06,
      "loss": 0.6335,
      "step": 1728
    },
    {
      "epoch": 0.09681935267107179,
      "grad_norm": 1.271315336227417,
      "learning_rate": 8.635e-06,
      "loss": 0.4435,
      "step": 1729
    },
    {
      "epoch": 0.09687534998320081,
      "grad_norm": 1.2029268741607666,
      "learning_rate": 8.64e-06,
      "loss": 0.3945,
      "step": 1730
    },
    {
      "epoch": 0.09693134729532983,
      "grad_norm": 1.723079800605774,
      "learning_rate": 8.645000000000001e-06,
      "loss": 0.4809,
      "step": 1731
    },
    {
      "epoch": 0.09698734460745884,
      "grad_norm": 1.122281551361084,
      "learning_rate": 8.65e-06,
      "loss": 0.4269,
      "step": 1732
    },
    {
      "epoch": 0.09704334191958786,
      "grad_norm": 1.2079497575759888,
      "learning_rate": 8.655000000000001e-06,
      "loss": 0.4232,
      "step": 1733
    },
    {
      "epoch": 0.09709933923171687,
      "grad_norm": 1.2280787229537964,
      "learning_rate": 8.66e-06,
      "loss": 0.3697,
      "step": 1734
    },
    {
      "epoch": 0.0971553365438459,
      "grad_norm": 1.0389596223831177,
      "learning_rate": 8.665000000000001e-06,
      "loss": 0.4221,
      "step": 1735
    },
    {
      "epoch": 0.09721133385597491,
      "grad_norm": 1.22295343875885,
      "learning_rate": 8.67e-06,
      "loss": 0.4129,
      "step": 1736
    },
    {
      "epoch": 0.09726733116810393,
      "grad_norm": 1.3335331678390503,
      "learning_rate": 8.674999999999999e-06,
      "loss": 0.4507,
      "step": 1737
    },
    {
      "epoch": 0.09732332848023295,
      "grad_norm": 1.3346177339553833,
      "learning_rate": 8.68e-06,
      "loss": 0.3771,
      "step": 1738
    },
    {
      "epoch": 0.09737932579236197,
      "grad_norm": 1.0787492990493774,
      "learning_rate": 8.685e-06,
      "loss": 0.4032,
      "step": 1739
    },
    {
      "epoch": 0.09743532310449099,
      "grad_norm": 1.352504849433899,
      "learning_rate": 8.690000000000002e-06,
      "loss": 0.4517,
      "step": 1740
    },
    {
      "epoch": 0.09749132041662001,
      "grad_norm": 1.0957409143447876,
      "learning_rate": 8.695e-06,
      "loss": 0.3548,
      "step": 1741
    },
    {
      "epoch": 0.09754731772874901,
      "grad_norm": 1.1434050798416138,
      "learning_rate": 8.7e-06,
      "loss": 0.4702,
      "step": 1742
    },
    {
      "epoch": 0.09760331504087803,
      "grad_norm": 1.2777423858642578,
      "learning_rate": 8.705e-06,
      "loss": 0.3896,
      "step": 1743
    },
    {
      "epoch": 0.09765931235300705,
      "grad_norm": 1.4034701585769653,
      "learning_rate": 8.71e-06,
      "loss": 0.4506,
      "step": 1744
    },
    {
      "epoch": 0.09771530966513607,
      "grad_norm": 1.207825779914856,
      "learning_rate": 8.715e-06,
      "loss": 0.5022,
      "step": 1745
    },
    {
      "epoch": 0.09777130697726509,
      "grad_norm": 1.3076139688491821,
      "learning_rate": 8.720000000000001e-06,
      "loss": 0.3915,
      "step": 1746
    },
    {
      "epoch": 0.09782730428939411,
      "grad_norm": 1.2862523794174194,
      "learning_rate": 8.725e-06,
      "loss": 0.5892,
      "step": 1747
    },
    {
      "epoch": 0.09788330160152313,
      "grad_norm": 1.442941665649414,
      "learning_rate": 8.730000000000001e-06,
      "loss": 0.4577,
      "step": 1748
    },
    {
      "epoch": 0.09793929891365215,
      "grad_norm": 1.251220941543579,
      "learning_rate": 8.735e-06,
      "loss": 0.4834,
      "step": 1749
    },
    {
      "epoch": 0.09799529622578117,
      "grad_norm": 1.1590315103530884,
      "learning_rate": 8.740000000000001e-06,
      "loss": 0.5074,
      "step": 1750
    },
    {
      "epoch": 0.09805129353791019,
      "grad_norm": 1.3057984113693237,
      "learning_rate": 8.745e-06,
      "loss": 0.4082,
      "step": 1751
    },
    {
      "epoch": 0.09810729085003919,
      "grad_norm": 1.29001784324646,
      "learning_rate": 8.75e-06,
      "loss": 0.4628,
      "step": 1752
    },
    {
      "epoch": 0.09816328816216821,
      "grad_norm": 1.0873615741729736,
      "learning_rate": 8.755e-06,
      "loss": 0.3739,
      "step": 1753
    },
    {
      "epoch": 0.09821928547429723,
      "grad_norm": 1.2104002237319946,
      "learning_rate": 8.76e-06,
      "loss": 0.4093,
      "step": 1754
    },
    {
      "epoch": 0.09827528278642625,
      "grad_norm": 1.1916160583496094,
      "learning_rate": 8.765000000000002e-06,
      "loss": 0.5169,
      "step": 1755
    },
    {
      "epoch": 0.09833128009855527,
      "grad_norm": 1.1350438594818115,
      "learning_rate": 8.77e-06,
      "loss": 0.4636,
      "step": 1756
    },
    {
      "epoch": 0.09838727741068429,
      "grad_norm": 1.4811171293258667,
      "learning_rate": 8.775e-06,
      "loss": 0.5471,
      "step": 1757
    },
    {
      "epoch": 0.0984432747228133,
      "grad_norm": 1.5752403736114502,
      "learning_rate": 8.78e-06,
      "loss": 0.5765,
      "step": 1758
    },
    {
      "epoch": 0.09849927203494233,
      "grad_norm": 2.1089720726013184,
      "learning_rate": 8.785e-06,
      "loss": 0.4019,
      "step": 1759
    },
    {
      "epoch": 0.09855526934707134,
      "grad_norm": 1.8355612754821777,
      "learning_rate": 8.79e-06,
      "loss": 0.7225,
      "step": 1760
    },
    {
      "epoch": 0.09861126665920036,
      "grad_norm": 1.0810399055480957,
      "learning_rate": 8.795e-06,
      "loss": 0.3344,
      "step": 1761
    },
    {
      "epoch": 0.09866726397132938,
      "grad_norm": 1.1854368448257446,
      "learning_rate": 8.8e-06,
      "loss": 0.3407,
      "step": 1762
    },
    {
      "epoch": 0.09872326128345839,
      "grad_norm": 1.2420506477355957,
      "learning_rate": 8.805000000000001e-06,
      "loss": 0.415,
      "step": 1763
    },
    {
      "epoch": 0.09877925859558741,
      "grad_norm": 1.075506329536438,
      "learning_rate": 8.81e-06,
      "loss": 0.4851,
      "step": 1764
    },
    {
      "epoch": 0.09883525590771643,
      "grad_norm": 1.1143125295639038,
      "learning_rate": 8.815000000000001e-06,
      "loss": 0.4585,
      "step": 1765
    },
    {
      "epoch": 0.09889125321984545,
      "grad_norm": 1.1370161771774292,
      "learning_rate": 8.82e-06,
      "loss": 0.3993,
      "step": 1766
    },
    {
      "epoch": 0.09894725053197446,
      "grad_norm": 1.1423094272613525,
      "learning_rate": 8.825e-06,
      "loss": 0.3881,
      "step": 1767
    },
    {
      "epoch": 0.09900324784410348,
      "grad_norm": 1.2995854616165161,
      "learning_rate": 8.83e-06,
      "loss": 0.4175,
      "step": 1768
    },
    {
      "epoch": 0.0990592451562325,
      "grad_norm": 1.3148881196975708,
      "learning_rate": 8.835000000000001e-06,
      "loss": 0.6037,
      "step": 1769
    },
    {
      "epoch": 0.09911524246836152,
      "grad_norm": 1.084141731262207,
      "learning_rate": 8.840000000000002e-06,
      "loss": 0.3905,
      "step": 1770
    },
    {
      "epoch": 0.09917123978049054,
      "grad_norm": 1.2053523063659668,
      "learning_rate": 8.845000000000001e-06,
      "loss": 0.3216,
      "step": 1771
    },
    {
      "epoch": 0.09922723709261956,
      "grad_norm": 1.2783927917480469,
      "learning_rate": 8.85e-06,
      "loss": 0.4215,
      "step": 1772
    },
    {
      "epoch": 0.09928323440474857,
      "grad_norm": 1.163581132888794,
      "learning_rate": 8.855e-06,
      "loss": 0.5082,
      "step": 1773
    },
    {
      "epoch": 0.09933923171687759,
      "grad_norm": 1.0954021215438843,
      "learning_rate": 8.86e-06,
      "loss": 0.372,
      "step": 1774
    },
    {
      "epoch": 0.0993952290290066,
      "grad_norm": 1.236472487449646,
      "learning_rate": 8.865e-06,
      "loss": 0.3974,
      "step": 1775
    },
    {
      "epoch": 0.09945122634113562,
      "grad_norm": 1.2931112051010132,
      "learning_rate": 8.87e-06,
      "loss": 0.43,
      "step": 1776
    },
    {
      "epoch": 0.09950722365326464,
      "grad_norm": 1.1061400175094604,
      "learning_rate": 8.875e-06,
      "loss": 0.4575,
      "step": 1777
    },
    {
      "epoch": 0.09956322096539366,
      "grad_norm": 1.1908855438232422,
      "learning_rate": 8.880000000000001e-06,
      "loss": 0.4363,
      "step": 1778
    },
    {
      "epoch": 0.09961921827752268,
      "grad_norm": 1.2525837421417236,
      "learning_rate": 8.885e-06,
      "loss": 0.5029,
      "step": 1779
    },
    {
      "epoch": 0.0996752155896517,
      "grad_norm": 1.2395910024642944,
      "learning_rate": 8.890000000000001e-06,
      "loss": 0.4891,
      "step": 1780
    },
    {
      "epoch": 0.09973121290178072,
      "grad_norm": 1.4050637483596802,
      "learning_rate": 8.895e-06,
      "loss": 0.5142,
      "step": 1781
    },
    {
      "epoch": 0.09978721021390974,
      "grad_norm": 1.4121829271316528,
      "learning_rate": 8.9e-06,
      "loss": 0.6583,
      "step": 1782
    },
    {
      "epoch": 0.09984320752603874,
      "grad_norm": 1.1940996646881104,
      "learning_rate": 8.905e-06,
      "loss": 0.4022,
      "step": 1783
    },
    {
      "epoch": 0.09989920483816776,
      "grad_norm": 1.557766318321228,
      "learning_rate": 8.910000000000001e-06,
      "loss": 0.6801,
      "step": 1784
    },
    {
      "epoch": 0.09995520215029678,
      "grad_norm": 1.3310229778289795,
      "learning_rate": 8.915e-06,
      "loss": 0.3957,
      "step": 1785
    },
    {
      "epoch": 0.1000111994624258,
      "grad_norm": 1.4797648191452026,
      "learning_rate": 8.920000000000001e-06,
      "loss": 0.5639,
      "step": 1786
    },
    {
      "epoch": 0.10006719677455482,
      "grad_norm": 0.9999483227729797,
      "learning_rate": 8.925e-06,
      "loss": 0.4347,
      "step": 1787
    },
    {
      "epoch": 0.10012319408668384,
      "grad_norm": 1.196457862854004,
      "learning_rate": 8.930000000000001e-06,
      "loss": 0.5074,
      "step": 1788
    },
    {
      "epoch": 0.10017919139881286,
      "grad_norm": 1.206497073173523,
      "learning_rate": 8.935e-06,
      "loss": 0.2889,
      "step": 1789
    },
    {
      "epoch": 0.10023518871094188,
      "grad_norm": 1.3581793308258057,
      "learning_rate": 8.939999999999999e-06,
      "loss": 0.3974,
      "step": 1790
    },
    {
      "epoch": 0.1002911860230709,
      "grad_norm": 1.3118083477020264,
      "learning_rate": 8.945e-06,
      "loss": 0.5182,
      "step": 1791
    },
    {
      "epoch": 0.10034718333519992,
      "grad_norm": 1.631662368774414,
      "learning_rate": 8.95e-06,
      "loss": 0.4934,
      "step": 1792
    },
    {
      "epoch": 0.10040318064732894,
      "grad_norm": 1.0491358041763306,
      "learning_rate": 8.955000000000002e-06,
      "loss": 0.4696,
      "step": 1793
    },
    {
      "epoch": 0.10045917795945794,
      "grad_norm": 1.2039273977279663,
      "learning_rate": 8.96e-06,
      "loss": 0.5036,
      "step": 1794
    },
    {
      "epoch": 0.10051517527158696,
      "grad_norm": 1.1414377689361572,
      "learning_rate": 8.965e-06,
      "loss": 0.5919,
      "step": 1795
    },
    {
      "epoch": 0.10057117258371598,
      "grad_norm": 1.2074891328811646,
      "learning_rate": 8.97e-06,
      "loss": 0.5084,
      "step": 1796
    },
    {
      "epoch": 0.100627169895845,
      "grad_norm": 1.3807848691940308,
      "learning_rate": 8.975e-06,
      "loss": 0.373,
      "step": 1797
    },
    {
      "epoch": 0.10068316720797402,
      "grad_norm": 1.1689538955688477,
      "learning_rate": 8.98e-06,
      "loss": 0.4797,
      "step": 1798
    },
    {
      "epoch": 0.10073916452010304,
      "grad_norm": 1.133975625038147,
      "learning_rate": 8.985e-06,
      "loss": 0.4301,
      "step": 1799
    },
    {
      "epoch": 0.10079516183223206,
      "grad_norm": 1.4214868545532227,
      "learning_rate": 8.99e-06,
      "loss": 0.4385,
      "step": 1800
    },
    {
      "epoch": 0.10085115914436107,
      "grad_norm": 1.0573962926864624,
      "learning_rate": 8.995000000000001e-06,
      "loss": 0.3093,
      "step": 1801
    },
    {
      "epoch": 0.1009071564564901,
      "grad_norm": 1.1686052083969116,
      "learning_rate": 9e-06,
      "loss": 0.4314,
      "step": 1802
    },
    {
      "epoch": 0.10096315376861911,
      "grad_norm": 1.1324526071548462,
      "learning_rate": 9.005000000000001e-06,
      "loss": 0.4602,
      "step": 1803
    },
    {
      "epoch": 0.10101915108074812,
      "grad_norm": 1.3244366645812988,
      "learning_rate": 9.01e-06,
      "loss": 0.4623,
      "step": 1804
    },
    {
      "epoch": 0.10107514839287714,
      "grad_norm": 1.2081764936447144,
      "learning_rate": 9.015e-06,
      "loss": 0.6177,
      "step": 1805
    },
    {
      "epoch": 0.10113114570500616,
      "grad_norm": 1.2845309972763062,
      "learning_rate": 9.02e-06,
      "loss": 0.4678,
      "step": 1806
    },
    {
      "epoch": 0.10118714301713518,
      "grad_norm": 1.27948796749115,
      "learning_rate": 9.025e-06,
      "loss": 0.4194,
      "step": 1807
    },
    {
      "epoch": 0.1012431403292642,
      "grad_norm": 1.3915820121765137,
      "learning_rate": 9.030000000000002e-06,
      "loss": 0.4707,
      "step": 1808
    },
    {
      "epoch": 0.10129913764139321,
      "grad_norm": 1.4097074270248413,
      "learning_rate": 9.035e-06,
      "loss": 0.5651,
      "step": 1809
    },
    {
      "epoch": 0.10135513495352223,
      "grad_norm": 1.1681487560272217,
      "learning_rate": 9.04e-06,
      "loss": 0.4335,
      "step": 1810
    },
    {
      "epoch": 0.10141113226565125,
      "grad_norm": 1.217890977859497,
      "learning_rate": 9.045e-06,
      "loss": 0.522,
      "step": 1811
    },
    {
      "epoch": 0.10146712957778027,
      "grad_norm": 1.242827296257019,
      "learning_rate": 9.05e-06,
      "loss": 0.5948,
      "step": 1812
    },
    {
      "epoch": 0.10152312688990929,
      "grad_norm": 1.3299055099487305,
      "learning_rate": 9.055e-06,
      "loss": 0.656,
      "step": 1813
    },
    {
      "epoch": 0.1015791242020383,
      "grad_norm": 1.1744288206100464,
      "learning_rate": 9.06e-06,
      "loss": 0.4626,
      "step": 1814
    },
    {
      "epoch": 0.10163512151416731,
      "grad_norm": 1.298693299293518,
      "learning_rate": 9.065e-06,
      "loss": 0.5209,
      "step": 1815
    },
    {
      "epoch": 0.10169111882629633,
      "grad_norm": 1.4027745723724365,
      "learning_rate": 9.070000000000001e-06,
      "loss": 0.4836,
      "step": 1816
    },
    {
      "epoch": 0.10174711613842535,
      "grad_norm": 1.2797088623046875,
      "learning_rate": 9.075e-06,
      "loss": 0.5703,
      "step": 1817
    },
    {
      "epoch": 0.10180311345055437,
      "grad_norm": 1.4140253067016602,
      "learning_rate": 9.080000000000001e-06,
      "loss": 0.5811,
      "step": 1818
    },
    {
      "epoch": 0.10185911076268339,
      "grad_norm": 1.6313427686691284,
      "learning_rate": 9.085e-06,
      "loss": 0.5481,
      "step": 1819
    },
    {
      "epoch": 0.10191510807481241,
      "grad_norm": 1.1847879886627197,
      "learning_rate": 9.09e-06,
      "loss": 0.4082,
      "step": 1820
    },
    {
      "epoch": 0.10197110538694143,
      "grad_norm": 1.5288350582122803,
      "learning_rate": 9.095e-06,
      "loss": 0.4765,
      "step": 1821
    },
    {
      "epoch": 0.10202710269907045,
      "grad_norm": 1.3034580945968628,
      "learning_rate": 9.100000000000001e-06,
      "loss": 0.3948,
      "step": 1822
    },
    {
      "epoch": 0.10208310001119947,
      "grad_norm": 1.2626841068267822,
      "learning_rate": 9.105000000000002e-06,
      "loss": 0.4911,
      "step": 1823
    },
    {
      "epoch": 0.10213909732332849,
      "grad_norm": 1.275468111038208,
      "learning_rate": 9.110000000000001e-06,
      "loss": 0.4396,
      "step": 1824
    },
    {
      "epoch": 0.10219509463545749,
      "grad_norm": 1.1918511390686035,
      "learning_rate": 9.115e-06,
      "loss": 0.4379,
      "step": 1825
    },
    {
      "epoch": 0.10225109194758651,
      "grad_norm": 1.9813170433044434,
      "learning_rate": 9.12e-06,
      "loss": 0.7549,
      "step": 1826
    },
    {
      "epoch": 0.10230708925971553,
      "grad_norm": 1.3841813802719116,
      "learning_rate": 9.125e-06,
      "loss": 0.4712,
      "step": 1827
    },
    {
      "epoch": 0.10236308657184455,
      "grad_norm": 1.1758801937103271,
      "learning_rate": 9.13e-06,
      "loss": 0.5183,
      "step": 1828
    },
    {
      "epoch": 0.10241908388397357,
      "grad_norm": 1.1382752656936646,
      "learning_rate": 9.135e-06,
      "loss": 0.3649,
      "step": 1829
    },
    {
      "epoch": 0.10247508119610259,
      "grad_norm": 1.2009247541427612,
      "learning_rate": 9.14e-06,
      "loss": 0.4377,
      "step": 1830
    },
    {
      "epoch": 0.10253107850823161,
      "grad_norm": 1.2643797397613525,
      "learning_rate": 9.145000000000001e-06,
      "loss": 0.4512,
      "step": 1831
    },
    {
      "epoch": 0.10258707582036063,
      "grad_norm": 1.1410797834396362,
      "learning_rate": 9.15e-06,
      "loss": 0.4931,
      "step": 1832
    },
    {
      "epoch": 0.10264307313248965,
      "grad_norm": 1.101846694946289,
      "learning_rate": 9.155000000000001e-06,
      "loss": 0.5352,
      "step": 1833
    },
    {
      "epoch": 0.10269907044461866,
      "grad_norm": 1.570682168006897,
      "learning_rate": 9.16e-06,
      "loss": 0.4423,
      "step": 1834
    },
    {
      "epoch": 0.10275506775674767,
      "grad_norm": 1.3165384531021118,
      "learning_rate": 9.165e-06,
      "loss": 0.4251,
      "step": 1835
    },
    {
      "epoch": 0.10281106506887669,
      "grad_norm": 1.0519487857818604,
      "learning_rate": 9.17e-06,
      "loss": 0.4528,
      "step": 1836
    },
    {
      "epoch": 0.10286706238100571,
      "grad_norm": 1.3688673973083496,
      "learning_rate": 9.175000000000001e-06,
      "loss": 0.4195,
      "step": 1837
    },
    {
      "epoch": 0.10292305969313473,
      "grad_norm": 1.401411533355713,
      "learning_rate": 9.180000000000002e-06,
      "loss": 0.55,
      "step": 1838
    },
    {
      "epoch": 0.10297905700526375,
      "grad_norm": 1.0831941366195679,
      "learning_rate": 9.185000000000001e-06,
      "loss": 0.3569,
      "step": 1839
    },
    {
      "epoch": 0.10303505431739277,
      "grad_norm": 1.2040417194366455,
      "learning_rate": 9.19e-06,
      "loss": 0.4522,
      "step": 1840
    },
    {
      "epoch": 0.10309105162952178,
      "grad_norm": 1.1933237314224243,
      "learning_rate": 9.195000000000001e-06,
      "loss": 0.6391,
      "step": 1841
    },
    {
      "epoch": 0.1031470489416508,
      "grad_norm": 1.2009390592575073,
      "learning_rate": 9.2e-06,
      "loss": 0.5783,
      "step": 1842
    },
    {
      "epoch": 0.10320304625377982,
      "grad_norm": 1.155548095703125,
      "learning_rate": 9.205e-06,
      "loss": 0.3664,
      "step": 1843
    },
    {
      "epoch": 0.10325904356590884,
      "grad_norm": 1.0223307609558105,
      "learning_rate": 9.21e-06,
      "loss": 0.3759,
      "step": 1844
    },
    {
      "epoch": 0.10331504087803785,
      "grad_norm": 1.281187891960144,
      "learning_rate": 9.215e-06,
      "loss": 0.43,
      "step": 1845
    },
    {
      "epoch": 0.10337103819016687,
      "grad_norm": 1.2086399793624878,
      "learning_rate": 9.220000000000002e-06,
      "loss": 0.3932,
      "step": 1846
    },
    {
      "epoch": 0.10342703550229589,
      "grad_norm": 1.6127561330795288,
      "learning_rate": 9.225e-06,
      "loss": 0.5527,
      "step": 1847
    },
    {
      "epoch": 0.1034830328144249,
      "grad_norm": 1.1773858070373535,
      "learning_rate": 9.23e-06,
      "loss": 0.4325,
      "step": 1848
    },
    {
      "epoch": 0.10353903012655392,
      "grad_norm": 1.2475886344909668,
      "learning_rate": 9.235e-06,
      "loss": 0.4141,
      "step": 1849
    },
    {
      "epoch": 0.10359502743868294,
      "grad_norm": 1.3503400087356567,
      "learning_rate": 9.24e-06,
      "loss": 0.5307,
      "step": 1850
    },
    {
      "epoch": 0.10365102475081196,
      "grad_norm": 1.1808948516845703,
      "learning_rate": 9.245e-06,
      "loss": 0.5951,
      "step": 1851
    },
    {
      "epoch": 0.10370702206294098,
      "grad_norm": 1.2310532331466675,
      "learning_rate": 9.25e-06,
      "loss": 0.5327,
      "step": 1852
    },
    {
      "epoch": 0.10376301937507,
      "grad_norm": 1.1280689239501953,
      "learning_rate": 9.255e-06,
      "loss": 0.4972,
      "step": 1853
    },
    {
      "epoch": 0.10381901668719902,
      "grad_norm": 1.2415132522583008,
      "learning_rate": 9.260000000000001e-06,
      "loss": 0.5418,
      "step": 1854
    },
    {
      "epoch": 0.10387501399932804,
      "grad_norm": 1.0911405086517334,
      "learning_rate": 9.265e-06,
      "loss": 0.402,
      "step": 1855
    },
    {
      "epoch": 0.10393101131145704,
      "grad_norm": 1.1728242635726929,
      "learning_rate": 9.270000000000001e-06,
      "loss": 0.3286,
      "step": 1856
    },
    {
      "epoch": 0.10398700862358606,
      "grad_norm": 1.2124569416046143,
      "learning_rate": 9.275e-06,
      "loss": 0.4761,
      "step": 1857
    },
    {
      "epoch": 0.10404300593571508,
      "grad_norm": 1.1276865005493164,
      "learning_rate": 9.28e-06,
      "loss": 0.3177,
      "step": 1858
    },
    {
      "epoch": 0.1040990032478441,
      "grad_norm": 1.0694518089294434,
      "learning_rate": 9.285e-06,
      "loss": 0.2832,
      "step": 1859
    },
    {
      "epoch": 0.10415500055997312,
      "grad_norm": 1.132310390472412,
      "learning_rate": 9.29e-06,
      "loss": 0.4409,
      "step": 1860
    },
    {
      "epoch": 0.10421099787210214,
      "grad_norm": 1.23562490940094,
      "learning_rate": 9.295000000000002e-06,
      "loss": 0.4543,
      "step": 1861
    },
    {
      "epoch": 0.10426699518423116,
      "grad_norm": 1.2050089836120605,
      "learning_rate": 9.3e-06,
      "loss": 0.4191,
      "step": 1862
    },
    {
      "epoch": 0.10432299249636018,
      "grad_norm": 1.4085443019866943,
      "learning_rate": 9.305e-06,
      "loss": 0.4678,
      "step": 1863
    },
    {
      "epoch": 0.1043789898084892,
      "grad_norm": 1.2895890474319458,
      "learning_rate": 9.31e-06,
      "loss": 0.5659,
      "step": 1864
    },
    {
      "epoch": 0.10443498712061822,
      "grad_norm": 1.4111061096191406,
      "learning_rate": 9.315e-06,
      "loss": 0.3764,
      "step": 1865
    },
    {
      "epoch": 0.10449098443274722,
      "grad_norm": 1.1714211702346802,
      "learning_rate": 9.32e-06,
      "loss": 0.4289,
      "step": 1866
    },
    {
      "epoch": 0.10454698174487624,
      "grad_norm": 1.271815299987793,
      "learning_rate": 9.325e-06,
      "loss": 0.4254,
      "step": 1867
    },
    {
      "epoch": 0.10460297905700526,
      "grad_norm": 1.2217464447021484,
      "learning_rate": 9.33e-06,
      "loss": 0.4228,
      "step": 1868
    },
    {
      "epoch": 0.10465897636913428,
      "grad_norm": 1.1268996000289917,
      "learning_rate": 9.335000000000001e-06,
      "loss": 0.4438,
      "step": 1869
    },
    {
      "epoch": 0.1047149736812633,
      "grad_norm": 1.1250314712524414,
      "learning_rate": 9.34e-06,
      "loss": 0.3935,
      "step": 1870
    },
    {
      "epoch": 0.10477097099339232,
      "grad_norm": 1.4297596216201782,
      "learning_rate": 9.345000000000001e-06,
      "loss": 0.5209,
      "step": 1871
    },
    {
      "epoch": 0.10482696830552134,
      "grad_norm": 1.139918565750122,
      "learning_rate": 9.35e-06,
      "loss": 0.4825,
      "step": 1872
    },
    {
      "epoch": 0.10488296561765036,
      "grad_norm": 1.2917416095733643,
      "learning_rate": 9.355e-06,
      "loss": 0.5089,
      "step": 1873
    },
    {
      "epoch": 0.10493896292977938,
      "grad_norm": 1.4938721656799316,
      "learning_rate": 9.36e-06,
      "loss": 0.5283,
      "step": 1874
    },
    {
      "epoch": 0.1049949602419084,
      "grad_norm": 1.5548522472381592,
      "learning_rate": 9.365000000000001e-06,
      "loss": 0.4841,
      "step": 1875
    },
    {
      "epoch": 0.1050509575540374,
      "grad_norm": 1.1434475183486938,
      "learning_rate": 9.370000000000002e-06,
      "loss": 0.4877,
      "step": 1876
    },
    {
      "epoch": 0.10510695486616642,
      "grad_norm": 2.284370183944702,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.4799,
      "step": 1877
    },
    {
      "epoch": 0.10516295217829544,
      "grad_norm": 1.4365899562835693,
      "learning_rate": 9.38e-06,
      "loss": 0.5354,
      "step": 1878
    },
    {
      "epoch": 0.10521894949042446,
      "grad_norm": 1.1886478662490845,
      "learning_rate": 9.385e-06,
      "loss": 0.4485,
      "step": 1879
    },
    {
      "epoch": 0.10527494680255348,
      "grad_norm": 1.3608007431030273,
      "learning_rate": 9.39e-06,
      "loss": 0.4481,
      "step": 1880
    },
    {
      "epoch": 0.1053309441146825,
      "grad_norm": 1.356736421585083,
      "learning_rate": 9.395e-06,
      "loss": 0.5918,
      "step": 1881
    },
    {
      "epoch": 0.10538694142681151,
      "grad_norm": 1.3387682437896729,
      "learning_rate": 9.4e-06,
      "loss": 0.5101,
      "step": 1882
    },
    {
      "epoch": 0.10544293873894053,
      "grad_norm": 1.227960467338562,
      "learning_rate": 9.405e-06,
      "loss": 0.4004,
      "step": 1883
    },
    {
      "epoch": 0.10549893605106955,
      "grad_norm": 1.068472146987915,
      "learning_rate": 9.410000000000001e-06,
      "loss": 0.4885,
      "step": 1884
    },
    {
      "epoch": 0.10555493336319857,
      "grad_norm": 1.46451735496521,
      "learning_rate": 9.415e-06,
      "loss": 0.4849,
      "step": 1885
    },
    {
      "epoch": 0.10561093067532759,
      "grad_norm": 1.1203140020370483,
      "learning_rate": 9.420000000000001e-06,
      "loss": 0.3321,
      "step": 1886
    },
    {
      "epoch": 0.1056669279874566,
      "grad_norm": 1.3179823160171509,
      "learning_rate": 9.425e-06,
      "loss": 0.3947,
      "step": 1887
    },
    {
      "epoch": 0.10572292529958562,
      "grad_norm": 1.2362351417541504,
      "learning_rate": 9.43e-06,
      "loss": 0.4712,
      "step": 1888
    },
    {
      "epoch": 0.10577892261171463,
      "grad_norm": 1.349965214729309,
      "learning_rate": 9.435e-06,
      "loss": 0.4748,
      "step": 1889
    },
    {
      "epoch": 0.10583491992384365,
      "grad_norm": 1.158205509185791,
      "learning_rate": 9.44e-06,
      "loss": 0.3502,
      "step": 1890
    },
    {
      "epoch": 0.10589091723597267,
      "grad_norm": 1.2987258434295654,
      "learning_rate": 9.445000000000002e-06,
      "loss": 0.4718,
      "step": 1891
    },
    {
      "epoch": 0.10594691454810169,
      "grad_norm": 1.2764886617660522,
      "learning_rate": 9.450000000000001e-06,
      "loss": 0.4185,
      "step": 1892
    },
    {
      "epoch": 0.10600291186023071,
      "grad_norm": 1.5539253950119019,
      "learning_rate": 9.455e-06,
      "loss": 0.4658,
      "step": 1893
    },
    {
      "epoch": 0.10605890917235973,
      "grad_norm": 1.2473183870315552,
      "learning_rate": 9.460000000000001e-06,
      "loss": 0.3696,
      "step": 1894
    },
    {
      "epoch": 0.10611490648448875,
      "grad_norm": 1.3843179941177368,
      "learning_rate": 9.465e-06,
      "loss": 0.4765,
      "step": 1895
    },
    {
      "epoch": 0.10617090379661777,
      "grad_norm": 1.3900247812271118,
      "learning_rate": 9.47e-06,
      "loss": 0.606,
      "step": 1896
    },
    {
      "epoch": 0.10622690110874677,
      "grad_norm": 1.1595515012741089,
      "learning_rate": 9.475e-06,
      "loss": 0.4623,
      "step": 1897
    },
    {
      "epoch": 0.1062828984208758,
      "grad_norm": 2.569775104522705,
      "learning_rate": 9.48e-06,
      "loss": 0.3973,
      "step": 1898
    },
    {
      "epoch": 0.10633889573300481,
      "grad_norm": 1.197704792022705,
      "learning_rate": 9.485000000000002e-06,
      "loss": 0.4226,
      "step": 1899
    },
    {
      "epoch": 0.10639489304513383,
      "grad_norm": 2.009443521499634,
      "learning_rate": 9.49e-06,
      "loss": 0.5467,
      "step": 1900
    },
    {
      "epoch": 0.10645089035726285,
      "grad_norm": 1.1895710229873657,
      "learning_rate": 9.495000000000001e-06,
      "loss": 0.4143,
      "step": 1901
    },
    {
      "epoch": 0.10650688766939187,
      "grad_norm": 1.5191071033477783,
      "learning_rate": 9.5e-06,
      "loss": 0.4761,
      "step": 1902
    },
    {
      "epoch": 0.10656288498152089,
      "grad_norm": 1.3583176136016846,
      "learning_rate": 9.505e-06,
      "loss": 0.5819,
      "step": 1903
    },
    {
      "epoch": 0.10661888229364991,
      "grad_norm": 1.112642526626587,
      "learning_rate": 9.51e-06,
      "loss": 0.4729,
      "step": 1904
    },
    {
      "epoch": 0.10667487960577893,
      "grad_norm": 1.0252685546875,
      "learning_rate": 9.515e-06,
      "loss": 0.4811,
      "step": 1905
    },
    {
      "epoch": 0.10673087691790795,
      "grad_norm": 1.255096673965454,
      "learning_rate": 9.52e-06,
      "loss": 0.441,
      "step": 1906
    },
    {
      "epoch": 0.10678687423003695,
      "grad_norm": 1.3634824752807617,
      "learning_rate": 9.525000000000001e-06,
      "loss": 0.4749,
      "step": 1907
    },
    {
      "epoch": 0.10684287154216597,
      "grad_norm": 1.1139448881149292,
      "learning_rate": 9.53e-06,
      "loss": 0.3083,
      "step": 1908
    },
    {
      "epoch": 0.10689886885429499,
      "grad_norm": 1.3352259397506714,
      "learning_rate": 9.535000000000001e-06,
      "loss": 0.3185,
      "step": 1909
    },
    {
      "epoch": 0.10695486616642401,
      "grad_norm": 1.2440032958984375,
      "learning_rate": 9.54e-06,
      "loss": 0.474,
      "step": 1910
    },
    {
      "epoch": 0.10701086347855303,
      "grad_norm": 1.1798661947250366,
      "learning_rate": 9.545e-06,
      "loss": 0.4714,
      "step": 1911
    },
    {
      "epoch": 0.10706686079068205,
      "grad_norm": 1.2166504859924316,
      "learning_rate": 9.55e-06,
      "loss": 0.5616,
      "step": 1912
    },
    {
      "epoch": 0.10712285810281107,
      "grad_norm": 1.2180215120315552,
      "learning_rate": 9.555e-06,
      "loss": 0.4465,
      "step": 1913
    },
    {
      "epoch": 0.10717885541494009,
      "grad_norm": 1.3125149011611938,
      "learning_rate": 9.560000000000002e-06,
      "loss": 0.5186,
      "step": 1914
    },
    {
      "epoch": 0.1072348527270691,
      "grad_norm": 1.1238898038864136,
      "learning_rate": 9.565e-06,
      "loss": 0.3985,
      "step": 1915
    },
    {
      "epoch": 0.10729085003919812,
      "grad_norm": 1.2942389249801636,
      "learning_rate": 9.57e-06,
      "loss": 0.4492,
      "step": 1916
    },
    {
      "epoch": 0.10734684735132714,
      "grad_norm": 1.153488039970398,
      "learning_rate": 9.575e-06,
      "loss": 0.5233,
      "step": 1917
    },
    {
      "epoch": 0.10740284466345615,
      "grad_norm": 1.1370173692703247,
      "learning_rate": 9.58e-06,
      "loss": 0.4421,
      "step": 1918
    },
    {
      "epoch": 0.10745884197558517,
      "grad_norm": 1.3307583332061768,
      "learning_rate": 9.585e-06,
      "loss": 0.4855,
      "step": 1919
    },
    {
      "epoch": 0.10751483928771419,
      "grad_norm": 1.103548526763916,
      "learning_rate": 9.59e-06,
      "loss": 0.39,
      "step": 1920
    },
    {
      "epoch": 0.1075708365998432,
      "grad_norm": 1.2443008422851562,
      "learning_rate": 9.595e-06,
      "loss": 0.5393,
      "step": 1921
    },
    {
      "epoch": 0.10762683391197223,
      "grad_norm": 1.0949368476867676,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.3842,
      "step": 1922
    },
    {
      "epoch": 0.10768283122410124,
      "grad_norm": 1.3167270421981812,
      "learning_rate": 9.605e-06,
      "loss": 0.4478,
      "step": 1923
    },
    {
      "epoch": 0.10773882853623026,
      "grad_norm": 1.3839845657348633,
      "learning_rate": 9.610000000000001e-06,
      "loss": 0.3765,
      "step": 1924
    },
    {
      "epoch": 0.10779482584835928,
      "grad_norm": 1.3814969062805176,
      "learning_rate": 9.615e-06,
      "loss": 0.5044,
      "step": 1925
    },
    {
      "epoch": 0.1078508231604883,
      "grad_norm": 1.2561522722244263,
      "learning_rate": 9.62e-06,
      "loss": 0.4028,
      "step": 1926
    },
    {
      "epoch": 0.10790682047261732,
      "grad_norm": 1.1426435708999634,
      "learning_rate": 9.625e-06,
      "loss": 0.3741,
      "step": 1927
    },
    {
      "epoch": 0.10796281778474633,
      "grad_norm": 1.2376185655593872,
      "learning_rate": 9.630000000000001e-06,
      "loss": 0.639,
      "step": 1928
    },
    {
      "epoch": 0.10801881509687535,
      "grad_norm": 1.2048944234848022,
      "learning_rate": 9.635000000000002e-06,
      "loss": 0.4114,
      "step": 1929
    },
    {
      "epoch": 0.10807481240900436,
      "grad_norm": 1.0896660089492798,
      "learning_rate": 9.640000000000001e-06,
      "loss": 0.4479,
      "step": 1930
    },
    {
      "epoch": 0.10813080972113338,
      "grad_norm": 1.411189079284668,
      "learning_rate": 9.645e-06,
      "loss": 0.4999,
      "step": 1931
    },
    {
      "epoch": 0.1081868070332624,
      "grad_norm": 1.2659348249435425,
      "learning_rate": 9.65e-06,
      "loss": 0.5425,
      "step": 1932
    },
    {
      "epoch": 0.10824280434539142,
      "grad_norm": 3.3288135528564453,
      "learning_rate": 9.655e-06,
      "loss": 0.3585,
      "step": 1933
    },
    {
      "epoch": 0.10829880165752044,
      "grad_norm": 1.1483445167541504,
      "learning_rate": 9.66e-06,
      "loss": 0.4003,
      "step": 1934
    },
    {
      "epoch": 0.10835479896964946,
      "grad_norm": 1.2844260931015015,
      "learning_rate": 9.665e-06,
      "loss": 0.5146,
      "step": 1935
    },
    {
      "epoch": 0.10841079628177848,
      "grad_norm": 1.4117387533187866,
      "learning_rate": 9.67e-06,
      "loss": 0.5105,
      "step": 1936
    },
    {
      "epoch": 0.1084667935939075,
      "grad_norm": 1.4861122369766235,
      "learning_rate": 9.675000000000001e-06,
      "loss": 0.5692,
      "step": 1937
    },
    {
      "epoch": 0.1085227909060365,
      "grad_norm": 1.534714937210083,
      "learning_rate": 9.68e-06,
      "loss": 0.4468,
      "step": 1938
    },
    {
      "epoch": 0.10857878821816552,
      "grad_norm": 1.1656982898712158,
      "learning_rate": 9.685000000000001e-06,
      "loss": 0.4292,
      "step": 1939
    },
    {
      "epoch": 0.10863478553029454,
      "grad_norm": 1.238632321357727,
      "learning_rate": 9.69e-06,
      "loss": 0.4331,
      "step": 1940
    },
    {
      "epoch": 0.10869078284242356,
      "grad_norm": 1.5374995470046997,
      "learning_rate": 9.695e-06,
      "loss": 0.6631,
      "step": 1941
    },
    {
      "epoch": 0.10874678015455258,
      "grad_norm": 1.3231433629989624,
      "learning_rate": 9.7e-06,
      "loss": 0.3684,
      "step": 1942
    },
    {
      "epoch": 0.1088027774666816,
      "grad_norm": 1.2549045085906982,
      "learning_rate": 9.705e-06,
      "loss": 0.3888,
      "step": 1943
    },
    {
      "epoch": 0.10885877477881062,
      "grad_norm": 0.9933781623840332,
      "learning_rate": 9.71e-06,
      "loss": 0.4307,
      "step": 1944
    },
    {
      "epoch": 0.10891477209093964,
      "grad_norm": 1.3064879179000854,
      "learning_rate": 9.715000000000001e-06,
      "loss": 0.495,
      "step": 1945
    },
    {
      "epoch": 0.10897076940306866,
      "grad_norm": 1.2886368036270142,
      "learning_rate": 9.72e-06,
      "loss": 0.4827,
      "step": 1946
    },
    {
      "epoch": 0.10902676671519768,
      "grad_norm": 1.11863112449646,
      "learning_rate": 9.725000000000001e-06,
      "loss": 0.39,
      "step": 1947
    },
    {
      "epoch": 0.1090827640273267,
      "grad_norm": 1.2331268787384033,
      "learning_rate": 9.73e-06,
      "loss": 0.4058,
      "step": 1948
    },
    {
      "epoch": 0.1091387613394557,
      "grad_norm": 4.416565895080566,
      "learning_rate": 9.735e-06,
      "loss": 0.5548,
      "step": 1949
    },
    {
      "epoch": 0.10919475865158472,
      "grad_norm": 1.183388113975525,
      "learning_rate": 9.74e-06,
      "loss": 0.4693,
      "step": 1950
    },
    {
      "epoch": 0.10925075596371374,
      "grad_norm": 1.4820784330368042,
      "learning_rate": 9.745e-06,
      "loss": 0.4546,
      "step": 1951
    },
    {
      "epoch": 0.10930675327584276,
      "grad_norm": 1.2141631841659546,
      "learning_rate": 9.750000000000002e-06,
      "loss": 0.4229,
      "step": 1952
    },
    {
      "epoch": 0.10936275058797178,
      "grad_norm": 1.192245602607727,
      "learning_rate": 9.755e-06,
      "loss": 0.4096,
      "step": 1953
    },
    {
      "epoch": 0.1094187479001008,
      "grad_norm": 1.2988064289093018,
      "learning_rate": 9.760000000000001e-06,
      "loss": 0.5635,
      "step": 1954
    },
    {
      "epoch": 0.10947474521222982,
      "grad_norm": 1.2981138229370117,
      "learning_rate": 9.765e-06,
      "loss": 0.4759,
      "step": 1955
    },
    {
      "epoch": 0.10953074252435883,
      "grad_norm": 1.0608023405075073,
      "learning_rate": 9.77e-06,
      "loss": 0.4461,
      "step": 1956
    },
    {
      "epoch": 0.10958673983648785,
      "grad_norm": 1.3462153673171997,
      "learning_rate": 9.775e-06,
      "loss": 0.4433,
      "step": 1957
    },
    {
      "epoch": 0.10964273714861687,
      "grad_norm": 1.4526351690292358,
      "learning_rate": 9.78e-06,
      "loss": 0.4635,
      "step": 1958
    },
    {
      "epoch": 0.10969873446074588,
      "grad_norm": 1.1988767385482788,
      "learning_rate": 9.785e-06,
      "loss": 0.4356,
      "step": 1959
    },
    {
      "epoch": 0.1097547317728749,
      "grad_norm": 1.0474293231964111,
      "learning_rate": 9.790000000000001e-06,
      "loss": 0.3663,
      "step": 1960
    },
    {
      "epoch": 0.10981072908500392,
      "grad_norm": 1.190447211265564,
      "learning_rate": 9.795e-06,
      "loss": 0.459,
      "step": 1961
    },
    {
      "epoch": 0.10986672639713294,
      "grad_norm": 1.1437945365905762,
      "learning_rate": 9.800000000000001e-06,
      "loss": 0.4235,
      "step": 1962
    },
    {
      "epoch": 0.10992272370926195,
      "grad_norm": 1.2026954889297485,
      "learning_rate": 9.805e-06,
      "loss": 0.5103,
      "step": 1963
    },
    {
      "epoch": 0.10997872102139097,
      "grad_norm": 1.231103539466858,
      "learning_rate": 9.810000000000001e-06,
      "loss": 0.5811,
      "step": 1964
    },
    {
      "epoch": 0.11003471833351999,
      "grad_norm": 1.1028512716293335,
      "learning_rate": 9.815e-06,
      "loss": 0.381,
      "step": 1965
    },
    {
      "epoch": 0.11009071564564901,
      "grad_norm": 1.0779800415039062,
      "learning_rate": 9.820000000000001e-06,
      "loss": 0.3742,
      "step": 1966
    },
    {
      "epoch": 0.11014671295777803,
      "grad_norm": 1.243393063545227,
      "learning_rate": 9.825000000000002e-06,
      "loss": 0.4914,
      "step": 1967
    },
    {
      "epoch": 0.11020271026990705,
      "grad_norm": 1.29071044921875,
      "learning_rate": 9.83e-06,
      "loss": 0.4967,
      "step": 1968
    },
    {
      "epoch": 0.11025870758203606,
      "grad_norm": 1.126316785812378,
      "learning_rate": 9.835000000000002e-06,
      "loss": 0.4539,
      "step": 1969
    },
    {
      "epoch": 0.11031470489416507,
      "grad_norm": 1.5042952299118042,
      "learning_rate": 9.84e-06,
      "loss": 0.6822,
      "step": 1970
    },
    {
      "epoch": 0.1103707022062941,
      "grad_norm": 1.3173643350601196,
      "learning_rate": 9.845e-06,
      "loss": 0.4644,
      "step": 1971
    },
    {
      "epoch": 0.11042669951842311,
      "grad_norm": 1.1041350364685059,
      "learning_rate": 9.85e-06,
      "loss": 0.3427,
      "step": 1972
    },
    {
      "epoch": 0.11048269683055213,
      "grad_norm": 1.0766938924789429,
      "learning_rate": 9.855e-06,
      "loss": 0.4033,
      "step": 1973
    },
    {
      "epoch": 0.11053869414268115,
      "grad_norm": 1.1862468719482422,
      "learning_rate": 9.86e-06,
      "loss": 0.477,
      "step": 1974
    },
    {
      "epoch": 0.11059469145481017,
      "grad_norm": 1.4422067403793335,
      "learning_rate": 9.865000000000001e-06,
      "loss": 0.5174,
      "step": 1975
    },
    {
      "epoch": 0.11065068876693919,
      "grad_norm": 1.438631296157837,
      "learning_rate": 9.87e-06,
      "loss": 0.4254,
      "step": 1976
    },
    {
      "epoch": 0.11070668607906821,
      "grad_norm": 1.2510143518447876,
      "learning_rate": 9.875000000000001e-06,
      "loss": 0.4414,
      "step": 1977
    },
    {
      "epoch": 0.11076268339119723,
      "grad_norm": 1.0611276626586914,
      "learning_rate": 9.88e-06,
      "loss": 0.3732,
      "step": 1978
    },
    {
      "epoch": 0.11081868070332625,
      "grad_norm": 1.4915348291397095,
      "learning_rate": 9.885e-06,
      "loss": 0.5958,
      "step": 1979
    },
    {
      "epoch": 0.11087467801545525,
      "grad_norm": 1.314444899559021,
      "learning_rate": 9.89e-06,
      "loss": 0.4871,
      "step": 1980
    },
    {
      "epoch": 0.11093067532758427,
      "grad_norm": 1.2102175951004028,
      "learning_rate": 9.895e-06,
      "loss": 0.4774,
      "step": 1981
    },
    {
      "epoch": 0.11098667263971329,
      "grad_norm": 1.1883585453033447,
      "learning_rate": 9.900000000000002e-06,
      "loss": 0.3676,
      "step": 1982
    },
    {
      "epoch": 0.11104266995184231,
      "grad_norm": 1.1065423488616943,
      "learning_rate": 9.905000000000001e-06,
      "loss": 0.3954,
      "step": 1983
    },
    {
      "epoch": 0.11109866726397133,
      "grad_norm": 1.7444603443145752,
      "learning_rate": 9.91e-06,
      "loss": 0.7777,
      "step": 1984
    },
    {
      "epoch": 0.11115466457610035,
      "grad_norm": 1.5046581029891968,
      "learning_rate": 9.915e-06,
      "loss": 0.535,
      "step": 1985
    },
    {
      "epoch": 0.11121066188822937,
      "grad_norm": 1.9591126441955566,
      "learning_rate": 9.92e-06,
      "loss": 0.4007,
      "step": 1986
    },
    {
      "epoch": 0.11126665920035839,
      "grad_norm": 1.109308123588562,
      "learning_rate": 9.925e-06,
      "loss": 0.4983,
      "step": 1987
    },
    {
      "epoch": 0.1113226565124874,
      "grad_norm": 1.1649729013442993,
      "learning_rate": 9.93e-06,
      "loss": 0.4802,
      "step": 1988
    },
    {
      "epoch": 0.11137865382461642,
      "grad_norm": 1.3624181747436523,
      "learning_rate": 9.935e-06,
      "loss": 0.5429,
      "step": 1989
    },
    {
      "epoch": 0.11143465113674543,
      "grad_norm": 1.2421095371246338,
      "learning_rate": 9.940000000000001e-06,
      "loss": 0.4532,
      "step": 1990
    },
    {
      "epoch": 0.11149064844887445,
      "grad_norm": 1.2075271606445312,
      "learning_rate": 9.945e-06,
      "loss": 0.4258,
      "step": 1991
    },
    {
      "epoch": 0.11154664576100347,
      "grad_norm": 1.2682304382324219,
      "learning_rate": 9.950000000000001e-06,
      "loss": 0.5251,
      "step": 1992
    },
    {
      "epoch": 0.11160264307313249,
      "grad_norm": 1.2718613147735596,
      "learning_rate": 9.955e-06,
      "loss": 0.4165,
      "step": 1993
    },
    {
      "epoch": 0.1116586403852615,
      "grad_norm": 1.4027260541915894,
      "learning_rate": 9.96e-06,
      "loss": 0.6875,
      "step": 1994
    },
    {
      "epoch": 0.11171463769739053,
      "grad_norm": 1.2574266195297241,
      "learning_rate": 9.965e-06,
      "loss": 0.4169,
      "step": 1995
    },
    {
      "epoch": 0.11177063500951955,
      "grad_norm": 1.2845044136047363,
      "learning_rate": 9.97e-06,
      "loss": 0.4741,
      "step": 1996
    },
    {
      "epoch": 0.11182663232164856,
      "grad_norm": 1.383315086364746,
      "learning_rate": 9.975e-06,
      "loss": 0.5904,
      "step": 1997
    },
    {
      "epoch": 0.11188262963377758,
      "grad_norm": 1.2629704475402832,
      "learning_rate": 9.980000000000001e-06,
      "loss": 0.4515,
      "step": 1998
    },
    {
      "epoch": 0.1119386269459066,
      "grad_norm": 1.3212579488754272,
      "learning_rate": 9.985e-06,
      "loss": 0.5825,
      "step": 1999
    },
    {
      "epoch": 0.11199462425803561,
      "grad_norm": 1.274903655052185,
      "learning_rate": 9.990000000000001e-06,
      "loss": 0.5872,
      "step": 2000
    },
    {
      "epoch": 0.11205062157016463,
      "grad_norm": 1.2747188806533813,
      "learning_rate": 9.995e-06,
      "loss": 0.3708,
      "step": 2001
    },
    {
      "epoch": 0.11210661888229365,
      "grad_norm": 1.472104549407959,
      "learning_rate": 1e-05,
      "loss": 0.5602,
      "step": 2002
    },
    {
      "epoch": 0.11216261619442267,
      "grad_norm": 1.256666660308838,
      "learning_rate": 1.0005e-05,
      "loss": 0.4248,
      "step": 2003
    },
    {
      "epoch": 0.11221861350655168,
      "grad_norm": 1.092176079750061,
      "learning_rate": 1.001e-05,
      "loss": 0.4048,
      "step": 2004
    },
    {
      "epoch": 0.1122746108186807,
      "grad_norm": 1.1835278272628784,
      "learning_rate": 1.0015000000000002e-05,
      "loss": 0.5522,
      "step": 2005
    },
    {
      "epoch": 0.11233060813080972,
      "grad_norm": 17.31488800048828,
      "learning_rate": 1.002e-05,
      "loss": 0.354,
      "step": 2006
    },
    {
      "epoch": 0.11238660544293874,
      "grad_norm": 1.2700695991516113,
      "learning_rate": 1.0025000000000001e-05,
      "loss": 0.6672,
      "step": 2007
    },
    {
      "epoch": 0.11244260275506776,
      "grad_norm": 1.2675044536590576,
      "learning_rate": 1.003e-05,
      "loss": 0.5727,
      "step": 2008
    },
    {
      "epoch": 0.11249860006719678,
      "grad_norm": 1.354222059249878,
      "learning_rate": 1.0035e-05,
      "loss": 0.553,
      "step": 2009
    },
    {
      "epoch": 0.1125545973793258,
      "grad_norm": 1.2412958145141602,
      "learning_rate": 1.004e-05,
      "loss": 0.3739,
      "step": 2010
    },
    {
      "epoch": 0.1126105946914548,
      "grad_norm": 1.1372991800308228,
      "learning_rate": 1.0045e-05,
      "loss": 0.3767,
      "step": 2011
    },
    {
      "epoch": 0.11266659200358382,
      "grad_norm": 1.4014700651168823,
      "learning_rate": 1.005e-05,
      "loss": 0.611,
      "step": 2012
    },
    {
      "epoch": 0.11272258931571284,
      "grad_norm": 1.648236632347107,
      "learning_rate": 1.0055000000000001e-05,
      "loss": 0.546,
      "step": 2013
    },
    {
      "epoch": 0.11277858662784186,
      "grad_norm": 1.0424588918685913,
      "learning_rate": 1.006e-05,
      "loss": 0.3135,
      "step": 2014
    },
    {
      "epoch": 0.11283458393997088,
      "grad_norm": 1.1952458620071411,
      "learning_rate": 1.0065000000000001e-05,
      "loss": 0.4185,
      "step": 2015
    },
    {
      "epoch": 0.1128905812520999,
      "grad_norm": 1.2783581018447876,
      "learning_rate": 1.007e-05,
      "loss": 0.5532,
      "step": 2016
    },
    {
      "epoch": 0.11294657856422892,
      "grad_norm": 1.3510452508926392,
      "learning_rate": 1.0075000000000001e-05,
      "loss": 0.3781,
      "step": 2017
    },
    {
      "epoch": 0.11300257587635794,
      "grad_norm": 1.534055471420288,
      "learning_rate": 1.008e-05,
      "loss": 0.5066,
      "step": 2018
    },
    {
      "epoch": 0.11305857318848696,
      "grad_norm": 1.1314103603363037,
      "learning_rate": 1.0085e-05,
      "loss": 0.3968,
      "step": 2019
    },
    {
      "epoch": 0.11311457050061598,
      "grad_norm": 1.1797354221343994,
      "learning_rate": 1.0090000000000002e-05,
      "loss": 0.2986,
      "step": 2020
    },
    {
      "epoch": 0.11317056781274498,
      "grad_norm": 1.1401501893997192,
      "learning_rate": 1.0095e-05,
      "loss": 0.4593,
      "step": 2021
    },
    {
      "epoch": 0.113226565124874,
      "grad_norm": 1.2980574369430542,
      "learning_rate": 1.0100000000000002e-05,
      "loss": 0.6112,
      "step": 2022
    },
    {
      "epoch": 0.11328256243700302,
      "grad_norm": 1.022275447845459,
      "learning_rate": 1.0105e-05,
      "loss": 0.4415,
      "step": 2023
    },
    {
      "epoch": 0.11333855974913204,
      "grad_norm": 1.181807518005371,
      "learning_rate": 1.011e-05,
      "loss": 0.4308,
      "step": 2024
    },
    {
      "epoch": 0.11339455706126106,
      "grad_norm": 1.1795766353607178,
      "learning_rate": 1.0115e-05,
      "loss": 0.4049,
      "step": 2025
    },
    {
      "epoch": 0.11345055437339008,
      "grad_norm": 1.262852668762207,
      "learning_rate": 1.012e-05,
      "loss": 0.4993,
      "step": 2026
    },
    {
      "epoch": 0.1135065516855191,
      "grad_norm": 1.3481334447860718,
      "learning_rate": 1.0125e-05,
      "loss": 0.5144,
      "step": 2027
    },
    {
      "epoch": 0.11356254899764812,
      "grad_norm": 1.1121824979782104,
      "learning_rate": 1.0130000000000001e-05,
      "loss": 0.3884,
      "step": 2028
    },
    {
      "epoch": 0.11361854630977714,
      "grad_norm": 1.3804322481155396,
      "learning_rate": 1.0135e-05,
      "loss": 0.4579,
      "step": 2029
    },
    {
      "epoch": 0.11367454362190615,
      "grad_norm": 1.271407127380371,
      "learning_rate": 1.0140000000000001e-05,
      "loss": 0.6424,
      "step": 2030
    },
    {
      "epoch": 0.11373054093403516,
      "grad_norm": 1.1223245859146118,
      "learning_rate": 1.0145e-05,
      "loss": 0.4934,
      "step": 2031
    },
    {
      "epoch": 0.11378653824616418,
      "grad_norm": 1.3676866292953491,
      "learning_rate": 1.0150000000000001e-05,
      "loss": 0.4716,
      "step": 2032
    },
    {
      "epoch": 0.1138425355582932,
      "grad_norm": 1.60161292552948,
      "learning_rate": 1.0155e-05,
      "loss": 0.5407,
      "step": 2033
    },
    {
      "epoch": 0.11389853287042222,
      "grad_norm": 1.5236585140228271,
      "learning_rate": 1.016e-05,
      "loss": 0.4677,
      "step": 2034
    },
    {
      "epoch": 0.11395453018255124,
      "grad_norm": 1.2690157890319824,
      "learning_rate": 1.0165e-05,
      "loss": 0.4658,
      "step": 2035
    },
    {
      "epoch": 0.11401052749468026,
      "grad_norm": 1.9310686588287354,
      "learning_rate": 1.0170000000000001e-05,
      "loss": 0.3959,
      "step": 2036
    },
    {
      "epoch": 0.11406652480680927,
      "grad_norm": 1.4214739799499512,
      "learning_rate": 1.0175e-05,
      "loss": 0.5339,
      "step": 2037
    },
    {
      "epoch": 0.1141225221189383,
      "grad_norm": 1.6591894626617432,
      "learning_rate": 1.018e-05,
      "loss": 0.8932,
      "step": 2038
    },
    {
      "epoch": 0.11417851943106731,
      "grad_norm": 1.1825146675109863,
      "learning_rate": 1.0185e-05,
      "loss": 0.4305,
      "step": 2039
    },
    {
      "epoch": 0.11423451674319633,
      "grad_norm": 1.151872158050537,
      "learning_rate": 1.019e-05,
      "loss": 0.4143,
      "step": 2040
    },
    {
      "epoch": 0.11429051405532535,
      "grad_norm": 1.2419713735580444,
      "learning_rate": 1.0195e-05,
      "loss": 0.6054,
      "step": 2041
    },
    {
      "epoch": 0.11434651136745436,
      "grad_norm": 2.2976205348968506,
      "learning_rate": 1.02e-05,
      "loss": 0.5601,
      "step": 2042
    },
    {
      "epoch": 0.11440250867958338,
      "grad_norm": 1.0336729288101196,
      "learning_rate": 1.0205000000000001e-05,
      "loss": 0.3298,
      "step": 2043
    },
    {
      "epoch": 0.1144585059917124,
      "grad_norm": 13.27386474609375,
      "learning_rate": 1.021e-05,
      "loss": 0.4519,
      "step": 2044
    },
    {
      "epoch": 0.11451450330384141,
      "grad_norm": 1.9197330474853516,
      "learning_rate": 1.0215000000000001e-05,
      "loss": 0.4625,
      "step": 2045
    },
    {
      "epoch": 0.11457050061597043,
      "grad_norm": 1.1858235597610474,
      "learning_rate": 1.022e-05,
      "loss": 0.4263,
      "step": 2046
    },
    {
      "epoch": 0.11462649792809945,
      "grad_norm": 1.409785509109497,
      "learning_rate": 1.0225e-05,
      "loss": 0.5989,
      "step": 2047
    },
    {
      "epoch": 0.11468249524022847,
      "grad_norm": 1.406874656677246,
      "learning_rate": 1.023e-05,
      "loss": 0.4339,
      "step": 2048
    },
    {
      "epoch": 0.11473849255235749,
      "grad_norm": 2.0559372901916504,
      "learning_rate": 1.0235e-05,
      "loss": 0.4359,
      "step": 2049
    },
    {
      "epoch": 0.11479448986448651,
      "grad_norm": 1.2511943578720093,
      "learning_rate": 1.024e-05,
      "loss": 0.4325,
      "step": 2050
    },
    {
      "epoch": 0.11485048717661553,
      "grad_norm": 1.1281684637069702,
      "learning_rate": 1.0245000000000001e-05,
      "loss": 0.378,
      "step": 2051
    },
    {
      "epoch": 0.11490648448874453,
      "grad_norm": 1.5144761800765991,
      "learning_rate": 1.025e-05,
      "loss": 0.409,
      "step": 2052
    },
    {
      "epoch": 0.11496248180087355,
      "grad_norm": 1.0558453798294067,
      "learning_rate": 1.0255000000000001e-05,
      "loss": 0.3365,
      "step": 2053
    },
    {
      "epoch": 0.11501847911300257,
      "grad_norm": 1.1261556148529053,
      "learning_rate": 1.026e-05,
      "loss": 0.3968,
      "step": 2054
    },
    {
      "epoch": 0.11507447642513159,
      "grad_norm": 1.3315730094909668,
      "learning_rate": 1.0265e-05,
      "loss": 0.5828,
      "step": 2055
    },
    {
      "epoch": 0.11513047373726061,
      "grad_norm": 1.191011905670166,
      "learning_rate": 1.027e-05,
      "loss": 0.5525,
      "step": 2056
    },
    {
      "epoch": 0.11518647104938963,
      "grad_norm": 1.2746089696884155,
      "learning_rate": 1.0275e-05,
      "loss": 0.4807,
      "step": 2057
    },
    {
      "epoch": 0.11524246836151865,
      "grad_norm": 1.04375159740448,
      "learning_rate": 1.0280000000000002e-05,
      "loss": 0.3857,
      "step": 2058
    },
    {
      "epoch": 0.11529846567364767,
      "grad_norm": 1.2364025115966797,
      "learning_rate": 1.0285e-05,
      "loss": 0.4123,
      "step": 2059
    },
    {
      "epoch": 0.11535446298577669,
      "grad_norm": 1.1593568325042725,
      "learning_rate": 1.0290000000000001e-05,
      "loss": 0.3481,
      "step": 2060
    },
    {
      "epoch": 0.1154104602979057,
      "grad_norm": 1.2940306663513184,
      "learning_rate": 1.0295e-05,
      "loss": 0.4622,
      "step": 2061
    },
    {
      "epoch": 0.11546645761003471,
      "grad_norm": 1.0955636501312256,
      "learning_rate": 1.03e-05,
      "loss": 0.3879,
      "step": 2062
    },
    {
      "epoch": 0.11552245492216373,
      "grad_norm": 1.4514633417129517,
      "learning_rate": 1.0305e-05,
      "loss": 0.3795,
      "step": 2063
    },
    {
      "epoch": 0.11557845223429275,
      "grad_norm": 1.148901104927063,
      "learning_rate": 1.031e-05,
      "loss": 0.4643,
      "step": 2064
    },
    {
      "epoch": 0.11563444954642177,
      "grad_norm": 1.4800533056259155,
      "learning_rate": 1.0315e-05,
      "loss": 0.4183,
      "step": 2065
    },
    {
      "epoch": 0.11569044685855079,
      "grad_norm": 1.5105128288269043,
      "learning_rate": 1.0320000000000001e-05,
      "loss": 0.6919,
      "step": 2066
    },
    {
      "epoch": 0.11574644417067981,
      "grad_norm": 1.1953701972961426,
      "learning_rate": 1.0325e-05,
      "loss": 0.4238,
      "step": 2067
    },
    {
      "epoch": 0.11580244148280883,
      "grad_norm": 1.1766806840896606,
      "learning_rate": 1.0330000000000001e-05,
      "loss": 0.3962,
      "step": 2068
    },
    {
      "epoch": 0.11585843879493785,
      "grad_norm": 1.2753911018371582,
      "learning_rate": 1.0335e-05,
      "loss": 0.5049,
      "step": 2069
    },
    {
      "epoch": 0.11591443610706686,
      "grad_norm": 1.0954228639602661,
      "learning_rate": 1.0340000000000001e-05,
      "loss": 0.4322,
      "step": 2070
    },
    {
      "epoch": 0.11597043341919588,
      "grad_norm": 1.5553898811340332,
      "learning_rate": 1.0345e-05,
      "loss": 0.4716,
      "step": 2071
    },
    {
      "epoch": 0.1160264307313249,
      "grad_norm": 1.0009772777557373,
      "learning_rate": 1.035e-05,
      "loss": 0.3666,
      "step": 2072
    },
    {
      "epoch": 0.11608242804345391,
      "grad_norm": 1.017662525177002,
      "learning_rate": 1.0355000000000002e-05,
      "loss": 0.4066,
      "step": 2073
    },
    {
      "epoch": 0.11613842535558293,
      "grad_norm": 1.2353122234344482,
      "learning_rate": 1.036e-05,
      "loss": 0.5978,
      "step": 2074
    },
    {
      "epoch": 0.11619442266771195,
      "grad_norm": 1.3603179454803467,
      "learning_rate": 1.0365000000000002e-05,
      "loss": 0.5137,
      "step": 2075
    },
    {
      "epoch": 0.11625041997984097,
      "grad_norm": 1.1561442613601685,
      "learning_rate": 1.037e-05,
      "loss": 0.5091,
      "step": 2076
    },
    {
      "epoch": 0.11630641729196999,
      "grad_norm": 1.3805582523345947,
      "learning_rate": 1.0375e-05,
      "loss": 0.3834,
      "step": 2077
    },
    {
      "epoch": 0.116362414604099,
      "grad_norm": 1.298583745956421,
      "learning_rate": 1.038e-05,
      "loss": 0.4891,
      "step": 2078
    },
    {
      "epoch": 0.11641841191622802,
      "grad_norm": 1.1770060062408447,
      "learning_rate": 1.0385e-05,
      "loss": 0.513,
      "step": 2079
    },
    {
      "epoch": 0.11647440922835704,
      "grad_norm": 1.1579817533493042,
      "learning_rate": 1.039e-05,
      "loss": 0.3686,
      "step": 2080
    },
    {
      "epoch": 0.11653040654048606,
      "grad_norm": 0.9921775460243225,
      "learning_rate": 1.0395000000000001e-05,
      "loss": 0.3811,
      "step": 2081
    },
    {
      "epoch": 0.11658640385261508,
      "grad_norm": 1.3118376731872559,
      "learning_rate": 1.04e-05,
      "loss": 0.4713,
      "step": 2082
    },
    {
      "epoch": 0.11664240116474409,
      "grad_norm": 1.1961724758148193,
      "learning_rate": 1.0405000000000001e-05,
      "loss": 0.507,
      "step": 2083
    },
    {
      "epoch": 0.1166983984768731,
      "grad_norm": 1.0339444875717163,
      "learning_rate": 1.041e-05,
      "loss": 0.4746,
      "step": 2084
    },
    {
      "epoch": 0.11675439578900212,
      "grad_norm": 1.3444161415100098,
      "learning_rate": 1.0415000000000001e-05,
      "loss": 0.5601,
      "step": 2085
    },
    {
      "epoch": 0.11681039310113114,
      "grad_norm": 1.2207037210464478,
      "learning_rate": 1.042e-05,
      "loss": 0.6489,
      "step": 2086
    },
    {
      "epoch": 0.11686639041326016,
      "grad_norm": 1.1908115148544312,
      "learning_rate": 1.0425e-05,
      "loss": 0.4842,
      "step": 2087
    },
    {
      "epoch": 0.11692238772538918,
      "grad_norm": 1.1665292978286743,
      "learning_rate": 1.043e-05,
      "loss": 0.3457,
      "step": 2088
    },
    {
      "epoch": 0.1169783850375182,
      "grad_norm": 1.1667604446411133,
      "learning_rate": 1.0435000000000001e-05,
      "loss": 0.4657,
      "step": 2089
    },
    {
      "epoch": 0.11703438234964722,
      "grad_norm": 1.1318879127502441,
      "learning_rate": 1.0440000000000002e-05,
      "loss": 0.5898,
      "step": 2090
    },
    {
      "epoch": 0.11709037966177624,
      "grad_norm": 1.0579155683517456,
      "learning_rate": 1.0445e-05,
      "loss": 0.3713,
      "step": 2091
    },
    {
      "epoch": 0.11714637697390526,
      "grad_norm": 1.1594823598861694,
      "learning_rate": 1.045e-05,
      "loss": 0.5215,
      "step": 2092
    },
    {
      "epoch": 0.11720237428603426,
      "grad_norm": 1.0567612648010254,
      "learning_rate": 1.0455e-05,
      "loss": 0.3901,
      "step": 2093
    },
    {
      "epoch": 0.11725837159816328,
      "grad_norm": 1.176743507385254,
      "learning_rate": 1.046e-05,
      "loss": 0.522,
      "step": 2094
    },
    {
      "epoch": 0.1173143689102923,
      "grad_norm": 1.27535879611969,
      "learning_rate": 1.0465e-05,
      "loss": 0.3966,
      "step": 2095
    },
    {
      "epoch": 0.11737036622242132,
      "grad_norm": 1.0443687438964844,
      "learning_rate": 1.0470000000000001e-05,
      "loss": 0.3516,
      "step": 2096
    },
    {
      "epoch": 0.11742636353455034,
      "grad_norm": 1.2144715785980225,
      "learning_rate": 1.0475e-05,
      "loss": 0.3338,
      "step": 2097
    },
    {
      "epoch": 0.11748236084667936,
      "grad_norm": 1.1741278171539307,
      "learning_rate": 1.0480000000000001e-05,
      "loss": 0.5254,
      "step": 2098
    },
    {
      "epoch": 0.11753835815880838,
      "grad_norm": 1.267101526260376,
      "learning_rate": 1.0485e-05,
      "loss": 0.5502,
      "step": 2099
    },
    {
      "epoch": 0.1175943554709374,
      "grad_norm": 1.103272557258606,
      "learning_rate": 1.049e-05,
      "loss": 0.4883,
      "step": 2100
    },
    {
      "epoch": 0.11765035278306642,
      "grad_norm": 1.1672580242156982,
      "learning_rate": 1.0495e-05,
      "loss": 0.407,
      "step": 2101
    },
    {
      "epoch": 0.11770635009519544,
      "grad_norm": 1.3263310194015503,
      "learning_rate": 1.05e-05,
      "loss": 0.5444,
      "step": 2102
    },
    {
      "epoch": 0.11776234740732446,
      "grad_norm": 1.1278438568115234,
      "learning_rate": 1.0505e-05,
      "loss": 0.4182,
      "step": 2103
    },
    {
      "epoch": 0.11781834471945346,
      "grad_norm": 1.1948771476745605,
      "learning_rate": 1.0510000000000001e-05,
      "loss": 0.3452,
      "step": 2104
    },
    {
      "epoch": 0.11787434203158248,
      "grad_norm": 1.6579550504684448,
      "learning_rate": 1.0515e-05,
      "loss": 0.4885,
      "step": 2105
    },
    {
      "epoch": 0.1179303393437115,
      "grad_norm": 1.0571820735931396,
      "learning_rate": 1.0520000000000001e-05,
      "loss": 0.3004,
      "step": 2106
    },
    {
      "epoch": 0.11798633665584052,
      "grad_norm": 1.228864073753357,
      "learning_rate": 1.0525e-05,
      "loss": 0.4447,
      "step": 2107
    },
    {
      "epoch": 0.11804233396796954,
      "grad_norm": 1.2009637355804443,
      "learning_rate": 1.053e-05,
      "loss": 0.5708,
      "step": 2108
    },
    {
      "epoch": 0.11809833128009856,
      "grad_norm": 1.3223775625228882,
      "learning_rate": 1.0535e-05,
      "loss": 0.4903,
      "step": 2109
    },
    {
      "epoch": 0.11815432859222758,
      "grad_norm": 1.6229612827301025,
      "learning_rate": 1.0539999999999999e-05,
      "loss": 0.4234,
      "step": 2110
    },
    {
      "epoch": 0.1182103259043566,
      "grad_norm": 1.176546573638916,
      "learning_rate": 1.0545000000000002e-05,
      "loss": 0.4466,
      "step": 2111
    },
    {
      "epoch": 0.11826632321648561,
      "grad_norm": 1.011500358581543,
      "learning_rate": 1.055e-05,
      "loss": 0.3573,
      "step": 2112
    },
    {
      "epoch": 0.11832232052861463,
      "grad_norm": 1.4040199518203735,
      "learning_rate": 1.0555000000000001e-05,
      "loss": 0.5961,
      "step": 2113
    },
    {
      "epoch": 0.11837831784074364,
      "grad_norm": 1.226555585861206,
      "learning_rate": 1.056e-05,
      "loss": 0.386,
      "step": 2114
    },
    {
      "epoch": 0.11843431515287266,
      "grad_norm": 0.9699138402938843,
      "learning_rate": 1.0565e-05,
      "loss": 0.3447,
      "step": 2115
    },
    {
      "epoch": 0.11849031246500168,
      "grad_norm": 1.2999788522720337,
      "learning_rate": 1.057e-05,
      "loss": 0.4002,
      "step": 2116
    },
    {
      "epoch": 0.1185463097771307,
      "grad_norm": 1.2441914081573486,
      "learning_rate": 1.0575e-05,
      "loss": 0.4244,
      "step": 2117
    },
    {
      "epoch": 0.11860230708925971,
      "grad_norm": 1.3216584920883179,
      "learning_rate": 1.058e-05,
      "loss": 0.753,
      "step": 2118
    },
    {
      "epoch": 0.11865830440138873,
      "grad_norm": 1.2601038217544556,
      "learning_rate": 1.0585000000000001e-05,
      "loss": 0.4507,
      "step": 2119
    },
    {
      "epoch": 0.11871430171351775,
      "grad_norm": 1.233767032623291,
      "learning_rate": 1.059e-05,
      "loss": 0.44,
      "step": 2120
    },
    {
      "epoch": 0.11877029902564677,
      "grad_norm": 1.421810269355774,
      "learning_rate": 1.0595000000000001e-05,
      "loss": 0.4378,
      "step": 2121
    },
    {
      "epoch": 0.11882629633777579,
      "grad_norm": 1.1413922309875488,
      "learning_rate": 1.06e-05,
      "loss": 0.4579,
      "step": 2122
    },
    {
      "epoch": 0.11888229364990481,
      "grad_norm": 1.8250983953475952,
      "learning_rate": 1.0605000000000001e-05,
      "loss": 0.4029,
      "step": 2123
    },
    {
      "epoch": 0.11893829096203382,
      "grad_norm": 1.1074143648147583,
      "learning_rate": 1.061e-05,
      "loss": 0.4235,
      "step": 2124
    },
    {
      "epoch": 0.11899428827416283,
      "grad_norm": 1.3174351453781128,
      "learning_rate": 1.0615e-05,
      "loss": 0.4614,
      "step": 2125
    },
    {
      "epoch": 0.11905028558629185,
      "grad_norm": 1.3979164361953735,
      "learning_rate": 1.062e-05,
      "loss": 0.5064,
      "step": 2126
    },
    {
      "epoch": 0.11910628289842087,
      "grad_norm": 1.1409600973129272,
      "learning_rate": 1.0625e-05,
      "loss": 0.3579,
      "step": 2127
    },
    {
      "epoch": 0.11916228021054989,
      "grad_norm": 1.1139148473739624,
      "learning_rate": 1.0630000000000002e-05,
      "loss": 0.4408,
      "step": 2128
    },
    {
      "epoch": 0.11921827752267891,
      "grad_norm": 1.3749010562896729,
      "learning_rate": 1.0635e-05,
      "loss": 0.4861,
      "step": 2129
    },
    {
      "epoch": 0.11927427483480793,
      "grad_norm": 1.289667010307312,
      "learning_rate": 1.064e-05,
      "loss": 0.4676,
      "step": 2130
    },
    {
      "epoch": 0.11933027214693695,
      "grad_norm": 1.2668863534927368,
      "learning_rate": 1.0645e-05,
      "loss": 0.6976,
      "step": 2131
    },
    {
      "epoch": 0.11938626945906597,
      "grad_norm": 1.4886201620101929,
      "learning_rate": 1.065e-05,
      "loss": 0.5122,
      "step": 2132
    },
    {
      "epoch": 0.11944226677119499,
      "grad_norm": 15.780389785766602,
      "learning_rate": 1.0655e-05,
      "loss": 0.3455,
      "step": 2133
    },
    {
      "epoch": 0.11949826408332401,
      "grad_norm": 1.3982264995574951,
      "learning_rate": 1.0660000000000001e-05,
      "loss": 0.569,
      "step": 2134
    },
    {
      "epoch": 0.11955426139545301,
      "grad_norm": 1.2477631568908691,
      "learning_rate": 1.0665e-05,
      "loss": 0.4225,
      "step": 2135
    },
    {
      "epoch": 0.11961025870758203,
      "grad_norm": 1.1074628829956055,
      "learning_rate": 1.0670000000000001e-05,
      "loss": 0.3849,
      "step": 2136
    },
    {
      "epoch": 0.11966625601971105,
      "grad_norm": 1.1565700769424438,
      "learning_rate": 1.0675e-05,
      "loss": 0.3819,
      "step": 2137
    },
    {
      "epoch": 0.11972225333184007,
      "grad_norm": 1.2876352071762085,
      "learning_rate": 1.0680000000000001e-05,
      "loss": 0.4623,
      "step": 2138
    },
    {
      "epoch": 0.11977825064396909,
      "grad_norm": 1.0950559377670288,
      "learning_rate": 1.0685e-05,
      "loss": 0.3622,
      "step": 2139
    },
    {
      "epoch": 0.11983424795609811,
      "grad_norm": 1.1870139837265015,
      "learning_rate": 1.069e-05,
      "loss": 0.3958,
      "step": 2140
    },
    {
      "epoch": 0.11989024526822713,
      "grad_norm": 1.1585975885391235,
      "learning_rate": 1.0695e-05,
      "loss": 0.3928,
      "step": 2141
    },
    {
      "epoch": 0.11994624258035615,
      "grad_norm": 1.1703457832336426,
      "learning_rate": 1.0700000000000001e-05,
      "loss": 0.3748,
      "step": 2142
    },
    {
      "epoch": 0.12000223989248517,
      "grad_norm": 1.3534200191497803,
      "learning_rate": 1.0705000000000002e-05,
      "loss": 0.4758,
      "step": 2143
    },
    {
      "epoch": 0.12005823720461418,
      "grad_norm": 1.3202478885650635,
      "learning_rate": 1.071e-05,
      "loss": 0.461,
      "step": 2144
    },
    {
      "epoch": 0.12011423451674319,
      "grad_norm": 1.2206354141235352,
      "learning_rate": 1.0715e-05,
      "loss": 0.3884,
      "step": 2145
    },
    {
      "epoch": 0.12017023182887221,
      "grad_norm": 1.0906764268875122,
      "learning_rate": 1.072e-05,
      "loss": 0.431,
      "step": 2146
    },
    {
      "epoch": 0.12022622914100123,
      "grad_norm": 1.2466380596160889,
      "learning_rate": 1.0725e-05,
      "loss": 0.4785,
      "step": 2147
    },
    {
      "epoch": 0.12028222645313025,
      "grad_norm": 1.2545379400253296,
      "learning_rate": 1.073e-05,
      "loss": 0.4105,
      "step": 2148
    },
    {
      "epoch": 0.12033822376525927,
      "grad_norm": 1.18149733543396,
      "learning_rate": 1.0735000000000001e-05,
      "loss": 0.5563,
      "step": 2149
    },
    {
      "epoch": 0.12039422107738829,
      "grad_norm": 1.1253278255462646,
      "learning_rate": 1.074e-05,
      "loss": 0.4171,
      "step": 2150
    },
    {
      "epoch": 0.1204502183895173,
      "grad_norm": 1.4532430171966553,
      "learning_rate": 1.0745000000000001e-05,
      "loss": 0.4619,
      "step": 2151
    },
    {
      "epoch": 0.12050621570164632,
      "grad_norm": 1.6640323400497437,
      "learning_rate": 1.075e-05,
      "loss": 0.494,
      "step": 2152
    },
    {
      "epoch": 0.12056221301377534,
      "grad_norm": 1.1227400302886963,
      "learning_rate": 1.0755000000000001e-05,
      "loss": 0.362,
      "step": 2153
    },
    {
      "epoch": 0.12061821032590436,
      "grad_norm": 0.9881584048271179,
      "learning_rate": 1.076e-05,
      "loss": 0.3443,
      "step": 2154
    },
    {
      "epoch": 0.12067420763803337,
      "grad_norm": 1.3102385997772217,
      "learning_rate": 1.0765e-05,
      "loss": 0.489,
      "step": 2155
    },
    {
      "epoch": 0.12073020495016239,
      "grad_norm": 1.0849230289459229,
      "learning_rate": 1.077e-05,
      "loss": 0.4264,
      "step": 2156
    },
    {
      "epoch": 0.1207862022622914,
      "grad_norm": 1.1320273876190186,
      "learning_rate": 1.0775000000000001e-05,
      "loss": 0.4323,
      "step": 2157
    },
    {
      "epoch": 0.12084219957442043,
      "grad_norm": 1.3121912479400635,
      "learning_rate": 1.0780000000000002e-05,
      "loss": 0.5436,
      "step": 2158
    },
    {
      "epoch": 0.12089819688654944,
      "grad_norm": 1.8350551128387451,
      "learning_rate": 1.0785000000000001e-05,
      "loss": 0.4331,
      "step": 2159
    },
    {
      "epoch": 0.12095419419867846,
      "grad_norm": 1.2972077131271362,
      "learning_rate": 1.079e-05,
      "loss": 0.3937,
      "step": 2160
    },
    {
      "epoch": 0.12101019151080748,
      "grad_norm": 1.2910444736480713,
      "learning_rate": 1.0795e-05,
      "loss": 0.4605,
      "step": 2161
    },
    {
      "epoch": 0.1210661888229365,
      "grad_norm": 1.1541708707809448,
      "learning_rate": 1.08e-05,
      "loss": 0.3115,
      "step": 2162
    },
    {
      "epoch": 0.12112218613506552,
      "grad_norm": 1.3426966667175293,
      "learning_rate": 1.0804999999999999e-05,
      "loss": 0.4149,
      "step": 2163
    },
    {
      "epoch": 0.12117818344719454,
      "grad_norm": 1.162108302116394,
      "learning_rate": 1.081e-05,
      "loss": 0.3712,
      "step": 2164
    },
    {
      "epoch": 0.12123418075932356,
      "grad_norm": 1.4027280807495117,
      "learning_rate": 1.0815e-05,
      "loss": 0.4655,
      "step": 2165
    },
    {
      "epoch": 0.12129017807145256,
      "grad_norm": 1.7777239084243774,
      "learning_rate": 1.0820000000000001e-05,
      "loss": 0.5956,
      "step": 2166
    },
    {
      "epoch": 0.12134617538358158,
      "grad_norm": 1.2932366132736206,
      "learning_rate": 1.0825e-05,
      "loss": 0.4147,
      "step": 2167
    },
    {
      "epoch": 0.1214021726957106,
      "grad_norm": 1.4258068799972534,
      "learning_rate": 1.083e-05,
      "loss": 0.5727,
      "step": 2168
    },
    {
      "epoch": 0.12145817000783962,
      "grad_norm": 1.4387284517288208,
      "learning_rate": 1.0835e-05,
      "loss": 0.4624,
      "step": 2169
    },
    {
      "epoch": 0.12151416731996864,
      "grad_norm": 1.250396490097046,
      "learning_rate": 1.084e-05,
      "loss": 0.4046,
      "step": 2170
    },
    {
      "epoch": 0.12157016463209766,
      "grad_norm": 1.2278938293457031,
      "learning_rate": 1.0845e-05,
      "loss": 0.4158,
      "step": 2171
    },
    {
      "epoch": 0.12162616194422668,
      "grad_norm": 1.2678710222244263,
      "learning_rate": 1.0850000000000001e-05,
      "loss": 0.3746,
      "step": 2172
    },
    {
      "epoch": 0.1216821592563557,
      "grad_norm": 1.1719387769699097,
      "learning_rate": 1.0855e-05,
      "loss": 0.3382,
      "step": 2173
    },
    {
      "epoch": 0.12173815656848472,
      "grad_norm": 1.2299120426177979,
      "learning_rate": 1.0860000000000001e-05,
      "loss": 0.4751,
      "step": 2174
    },
    {
      "epoch": 0.12179415388061374,
      "grad_norm": 1.2288856506347656,
      "learning_rate": 1.0865e-05,
      "loss": 0.5048,
      "step": 2175
    },
    {
      "epoch": 0.12185015119274274,
      "grad_norm": 1.4430570602416992,
      "learning_rate": 1.0870000000000001e-05,
      "loss": 0.4628,
      "step": 2176
    },
    {
      "epoch": 0.12190614850487176,
      "grad_norm": 1.255171775817871,
      "learning_rate": 1.0875e-05,
      "loss": 0.405,
      "step": 2177
    },
    {
      "epoch": 0.12196214581700078,
      "grad_norm": 1.0752614736557007,
      "learning_rate": 1.088e-05,
      "loss": 0.5331,
      "step": 2178
    },
    {
      "epoch": 0.1220181431291298,
      "grad_norm": 1.3410518169403076,
      "learning_rate": 1.0885e-05,
      "loss": 0.4597,
      "step": 2179
    },
    {
      "epoch": 0.12207414044125882,
      "grad_norm": 1.2840832471847534,
      "learning_rate": 1.089e-05,
      "loss": 0.5856,
      "step": 2180
    },
    {
      "epoch": 0.12213013775338784,
      "grad_norm": 1.24722158908844,
      "learning_rate": 1.0895000000000002e-05,
      "loss": 0.4622,
      "step": 2181
    },
    {
      "epoch": 0.12218613506551686,
      "grad_norm": 1.6612390279769897,
      "learning_rate": 1.09e-05,
      "loss": 0.605,
      "step": 2182
    },
    {
      "epoch": 0.12224213237764588,
      "grad_norm": 1.183188796043396,
      "learning_rate": 1.0905e-05,
      "loss": 0.4119,
      "step": 2183
    },
    {
      "epoch": 0.1222981296897749,
      "grad_norm": 1.1382944583892822,
      "learning_rate": 1.091e-05,
      "loss": 0.4015,
      "step": 2184
    },
    {
      "epoch": 0.12235412700190391,
      "grad_norm": 1.2495747804641724,
      "learning_rate": 1.0915e-05,
      "loss": 0.3978,
      "step": 2185
    },
    {
      "epoch": 0.12241012431403292,
      "grad_norm": 1.3227219581604004,
      "learning_rate": 1.092e-05,
      "loss": 0.4539,
      "step": 2186
    },
    {
      "epoch": 0.12246612162616194,
      "grad_norm": 1.1677227020263672,
      "learning_rate": 1.0925000000000001e-05,
      "loss": 0.35,
      "step": 2187
    },
    {
      "epoch": 0.12252211893829096,
      "grad_norm": 1.3256142139434814,
      "learning_rate": 1.093e-05,
      "loss": 0.4132,
      "step": 2188
    },
    {
      "epoch": 0.12257811625041998,
      "grad_norm": 1.0301586389541626,
      "learning_rate": 1.0935000000000001e-05,
      "loss": 0.4447,
      "step": 2189
    },
    {
      "epoch": 0.122634113562549,
      "grad_norm": 1.344541311264038,
      "learning_rate": 1.094e-05,
      "loss": 0.5965,
      "step": 2190
    },
    {
      "epoch": 0.12269011087467802,
      "grad_norm": 1.0082032680511475,
      "learning_rate": 1.0945000000000001e-05,
      "loss": 0.3608,
      "step": 2191
    },
    {
      "epoch": 0.12274610818680703,
      "grad_norm": 1.1168889999389648,
      "learning_rate": 1.095e-05,
      "loss": 0.4109,
      "step": 2192
    },
    {
      "epoch": 0.12280210549893605,
      "grad_norm": 1.1556227207183838,
      "learning_rate": 1.0955e-05,
      "loss": 0.357,
      "step": 2193
    },
    {
      "epoch": 0.12285810281106507,
      "grad_norm": 1.1768267154693604,
      "learning_rate": 1.096e-05,
      "loss": 0.527,
      "step": 2194
    },
    {
      "epoch": 0.12291410012319409,
      "grad_norm": 1.3603123426437378,
      "learning_rate": 1.0965000000000001e-05,
      "loss": 0.5147,
      "step": 2195
    },
    {
      "epoch": 0.12297009743532311,
      "grad_norm": 1.364448070526123,
      "learning_rate": 1.0970000000000002e-05,
      "loss": 0.451,
      "step": 2196
    },
    {
      "epoch": 0.12302609474745212,
      "grad_norm": 1.1654924154281616,
      "learning_rate": 1.0975e-05,
      "loss": 0.3761,
      "step": 2197
    },
    {
      "epoch": 0.12308209205958114,
      "grad_norm": 1.1860575675964355,
      "learning_rate": 1.098e-05,
      "loss": 0.4211,
      "step": 2198
    },
    {
      "epoch": 0.12313808937171015,
      "grad_norm": 1.3204054832458496,
      "learning_rate": 1.0985e-05,
      "loss": 0.6236,
      "step": 2199
    },
    {
      "epoch": 0.12319408668383917,
      "grad_norm": 1.0727225542068481,
      "learning_rate": 1.099e-05,
      "loss": 0.4035,
      "step": 2200
    },
    {
      "epoch": 0.1232500839959682,
      "grad_norm": 1.1432641744613647,
      "learning_rate": 1.0995e-05,
      "loss": 0.3848,
      "step": 2201
    },
    {
      "epoch": 0.12330608130809721,
      "grad_norm": 1.2635337114334106,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.4212,
      "step": 2202
    },
    {
      "epoch": 0.12336207862022623,
      "grad_norm": 1.2043837308883667,
      "learning_rate": 1.1005e-05,
      "loss": 0.5291,
      "step": 2203
    },
    {
      "epoch": 0.12341807593235525,
      "grad_norm": 1.2756997346878052,
      "learning_rate": 1.1010000000000001e-05,
      "loss": 0.4819,
      "step": 2204
    },
    {
      "epoch": 0.12347407324448427,
      "grad_norm": 1.1286869049072266,
      "learning_rate": 1.1015e-05,
      "loss": 0.3912,
      "step": 2205
    },
    {
      "epoch": 0.12353007055661329,
      "grad_norm": 1.1184715032577515,
      "learning_rate": 1.1020000000000001e-05,
      "loss": 0.3552,
      "step": 2206
    },
    {
      "epoch": 0.1235860678687423,
      "grad_norm": 1.805359959602356,
      "learning_rate": 1.1025e-05,
      "loss": 0.5879,
      "step": 2207
    },
    {
      "epoch": 0.12364206518087131,
      "grad_norm": 1.4591259956359863,
      "learning_rate": 1.103e-05,
      "loss": 0.6414,
      "step": 2208
    },
    {
      "epoch": 0.12369806249300033,
      "grad_norm": 1.0697208642959595,
      "learning_rate": 1.1035e-05,
      "loss": 0.3618,
      "step": 2209
    },
    {
      "epoch": 0.12375405980512935,
      "grad_norm": 1.4364018440246582,
      "learning_rate": 1.1040000000000001e-05,
      "loss": 0.4088,
      "step": 2210
    },
    {
      "epoch": 0.12381005711725837,
      "grad_norm": 1.1499199867248535,
      "learning_rate": 1.1045000000000002e-05,
      "loss": 0.371,
      "step": 2211
    },
    {
      "epoch": 0.12386605442938739,
      "grad_norm": 1.1446573734283447,
      "learning_rate": 1.1050000000000001e-05,
      "loss": 0.3733,
      "step": 2212
    },
    {
      "epoch": 0.12392205174151641,
      "grad_norm": 1.2948927879333496,
      "learning_rate": 1.1055e-05,
      "loss": 0.5075,
      "step": 2213
    },
    {
      "epoch": 0.12397804905364543,
      "grad_norm": 1.4152624607086182,
      "learning_rate": 1.106e-05,
      "loss": 0.5764,
      "step": 2214
    },
    {
      "epoch": 0.12403404636577445,
      "grad_norm": 1.479748249053955,
      "learning_rate": 1.1065e-05,
      "loss": 0.5262,
      "step": 2215
    },
    {
      "epoch": 0.12409004367790347,
      "grad_norm": 1.7374500036239624,
      "learning_rate": 1.107e-05,
      "loss": 0.5617,
      "step": 2216
    },
    {
      "epoch": 0.12414604099003247,
      "grad_norm": 1.1393375396728516,
      "learning_rate": 1.1075e-05,
      "loss": 0.4754,
      "step": 2217
    },
    {
      "epoch": 0.12420203830216149,
      "grad_norm": 1.1788643598556519,
      "learning_rate": 1.108e-05,
      "loss": 0.4177,
      "step": 2218
    },
    {
      "epoch": 0.12425803561429051,
      "grad_norm": 1.2186473608016968,
      "learning_rate": 1.1085000000000001e-05,
      "loss": 0.6167,
      "step": 2219
    },
    {
      "epoch": 0.12431403292641953,
      "grad_norm": 1.2026934623718262,
      "learning_rate": 1.109e-05,
      "loss": 0.3933,
      "step": 2220
    },
    {
      "epoch": 0.12437003023854855,
      "grad_norm": 1.1073105335235596,
      "learning_rate": 1.1095e-05,
      "loss": 0.5107,
      "step": 2221
    },
    {
      "epoch": 0.12442602755067757,
      "grad_norm": 1.339430570602417,
      "learning_rate": 1.11e-05,
      "loss": 0.4354,
      "step": 2222
    },
    {
      "epoch": 0.12448202486280659,
      "grad_norm": 1.357672095298767,
      "learning_rate": 1.1105e-05,
      "loss": 0.4647,
      "step": 2223
    },
    {
      "epoch": 0.1245380221749356,
      "grad_norm": 1.2059109210968018,
      "learning_rate": 1.111e-05,
      "loss": 0.4529,
      "step": 2224
    },
    {
      "epoch": 0.12459401948706463,
      "grad_norm": 1.2800805568695068,
      "learning_rate": 1.1115000000000001e-05,
      "loss": 0.4615,
      "step": 2225
    },
    {
      "epoch": 0.12465001679919364,
      "grad_norm": 1.1725817918777466,
      "learning_rate": 1.112e-05,
      "loss": 0.3642,
      "step": 2226
    },
    {
      "epoch": 0.12470601411132266,
      "grad_norm": 1.4158438444137573,
      "learning_rate": 1.1125000000000001e-05,
      "loss": 0.3824,
      "step": 2227
    },
    {
      "epoch": 0.12476201142345167,
      "grad_norm": 1.0686370134353638,
      "learning_rate": 1.113e-05,
      "loss": 0.4248,
      "step": 2228
    },
    {
      "epoch": 0.12481800873558069,
      "grad_norm": 1.151619553565979,
      "learning_rate": 1.1135000000000001e-05,
      "loss": 0.4378,
      "step": 2229
    },
    {
      "epoch": 0.1248740060477097,
      "grad_norm": 1.1525603532791138,
      "learning_rate": 1.114e-05,
      "loss": 0.3432,
      "step": 2230
    },
    {
      "epoch": 0.12493000335983873,
      "grad_norm": 1.104702353477478,
      "learning_rate": 1.1145e-05,
      "loss": 0.4535,
      "step": 2231
    },
    {
      "epoch": 0.12498600067196775,
      "grad_norm": 1.0402547121047974,
      "learning_rate": 1.115e-05,
      "loss": 0.3269,
      "step": 2232
    },
    {
      "epoch": 0.12504199798409676,
      "grad_norm": 1.2609214782714844,
      "learning_rate": 1.1155e-05,
      "loss": 0.4081,
      "step": 2233
    },
    {
      "epoch": 0.12509799529622578,
      "grad_norm": 1.3451021909713745,
      "learning_rate": 1.1160000000000002e-05,
      "loss": 0.5328,
      "step": 2234
    },
    {
      "epoch": 0.1251539926083548,
      "grad_norm": 1.1217625141143799,
      "learning_rate": 1.1165e-05,
      "loss": 0.4598,
      "step": 2235
    },
    {
      "epoch": 0.12520998992048382,
      "grad_norm": 1.4307420253753662,
      "learning_rate": 1.117e-05,
      "loss": 0.4804,
      "step": 2236
    },
    {
      "epoch": 0.12526598723261284,
      "grad_norm": 1.2126822471618652,
      "learning_rate": 1.1175e-05,
      "loss": 0.5237,
      "step": 2237
    },
    {
      "epoch": 0.12532198454474186,
      "grad_norm": 2.062070846557617,
      "learning_rate": 1.118e-05,
      "loss": 0.3978,
      "step": 2238
    },
    {
      "epoch": 0.12537798185687088,
      "grad_norm": 1.2071523666381836,
      "learning_rate": 1.1185e-05,
      "loss": 0.421,
      "step": 2239
    },
    {
      "epoch": 0.1254339791689999,
      "grad_norm": 1.3621774911880493,
      "learning_rate": 1.1190000000000001e-05,
      "loss": 0.4159,
      "step": 2240
    },
    {
      "epoch": 0.12548997648112892,
      "grad_norm": 1.3251384496688843,
      "learning_rate": 1.1195e-05,
      "loss": 0.5032,
      "step": 2241
    },
    {
      "epoch": 0.12554597379325794,
      "grad_norm": 1.3219131231307983,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.4851,
      "step": 2242
    },
    {
      "epoch": 0.12560197110538693,
      "grad_norm": 1.1742477416992188,
      "learning_rate": 1.1205e-05,
      "loss": 0.415,
      "step": 2243
    },
    {
      "epoch": 0.12565796841751595,
      "grad_norm": 1.6238731145858765,
      "learning_rate": 1.1210000000000001e-05,
      "loss": 0.5411,
      "step": 2244
    },
    {
      "epoch": 0.12571396572964497,
      "grad_norm": 1.1181552410125732,
      "learning_rate": 1.1215e-05,
      "loss": 0.5741,
      "step": 2245
    },
    {
      "epoch": 0.12576996304177399,
      "grad_norm": 1.1299830675125122,
      "learning_rate": 1.122e-05,
      "loss": 0.3773,
      "step": 2246
    },
    {
      "epoch": 0.125825960353903,
      "grad_norm": 1.1737785339355469,
      "learning_rate": 1.1225e-05,
      "loss": 0.4077,
      "step": 2247
    },
    {
      "epoch": 0.12588195766603202,
      "grad_norm": 1.0765039920806885,
      "learning_rate": 1.1230000000000001e-05,
      "loss": 0.5425,
      "step": 2248
    },
    {
      "epoch": 0.12593795497816104,
      "grad_norm": 1.0482008457183838,
      "learning_rate": 1.1235000000000002e-05,
      "loss": 0.403,
      "step": 2249
    },
    {
      "epoch": 0.12599395229029006,
      "grad_norm": 1.0707420110702515,
      "learning_rate": 1.124e-05,
      "loss": 0.4714,
      "step": 2250
    },
    {
      "epoch": 0.12604994960241908,
      "grad_norm": 1.2540937662124634,
      "learning_rate": 1.1245e-05,
      "loss": 0.3512,
      "step": 2251
    },
    {
      "epoch": 0.1261059469145481,
      "grad_norm": 1.2286819219589233,
      "learning_rate": 1.125e-05,
      "loss": 0.397,
      "step": 2252
    },
    {
      "epoch": 0.12616194422667712,
      "grad_norm": 1.150970458984375,
      "learning_rate": 1.1255e-05,
      "loss": 0.4503,
      "step": 2253
    },
    {
      "epoch": 0.12621794153880614,
      "grad_norm": 1.4410957098007202,
      "learning_rate": 1.126e-05,
      "loss": 0.5917,
      "step": 2254
    },
    {
      "epoch": 0.12627393885093516,
      "grad_norm": 1.1711270809173584,
      "learning_rate": 1.1265e-05,
      "loss": 0.4428,
      "step": 2255
    },
    {
      "epoch": 0.12632993616306418,
      "grad_norm": 1.3283476829528809,
      "learning_rate": 1.127e-05,
      "loss": 0.5402,
      "step": 2256
    },
    {
      "epoch": 0.1263859334751932,
      "grad_norm": 1.4431616067886353,
      "learning_rate": 1.1275000000000001e-05,
      "loss": 0.5657,
      "step": 2257
    },
    {
      "epoch": 0.12644193078732222,
      "grad_norm": 1.122452974319458,
      "learning_rate": 1.128e-05,
      "loss": 0.4835,
      "step": 2258
    },
    {
      "epoch": 0.12649792809945123,
      "grad_norm": 1.05092191696167,
      "learning_rate": 1.1285000000000001e-05,
      "loss": 0.4517,
      "step": 2259
    },
    {
      "epoch": 0.12655392541158025,
      "grad_norm": 1.0658460855484009,
      "learning_rate": 1.129e-05,
      "loss": 0.4475,
      "step": 2260
    },
    {
      "epoch": 0.12660992272370927,
      "grad_norm": 1.3967729806900024,
      "learning_rate": 1.1295e-05,
      "loss": 0.4501,
      "step": 2261
    },
    {
      "epoch": 0.1266659200358383,
      "grad_norm": 1.4232945442199707,
      "learning_rate": 1.13e-05,
      "loss": 0.5232,
      "step": 2262
    },
    {
      "epoch": 0.1267219173479673,
      "grad_norm": 1.4301387071609497,
      "learning_rate": 1.1305000000000001e-05,
      "loss": 0.6171,
      "step": 2263
    },
    {
      "epoch": 0.1267779146600963,
      "grad_norm": 1.4325543642044067,
      "learning_rate": 1.1310000000000002e-05,
      "loss": 0.5432,
      "step": 2264
    },
    {
      "epoch": 0.12683391197222532,
      "grad_norm": 1.292751669883728,
      "learning_rate": 1.1315000000000001e-05,
      "loss": 0.5031,
      "step": 2265
    },
    {
      "epoch": 0.12688990928435434,
      "grad_norm": 1.2938718795776367,
      "learning_rate": 1.132e-05,
      "loss": 0.5866,
      "step": 2266
    },
    {
      "epoch": 0.12694590659648336,
      "grad_norm": 1.285792589187622,
      "learning_rate": 1.1325e-05,
      "loss": 0.4874,
      "step": 2267
    },
    {
      "epoch": 0.12700190390861238,
      "grad_norm": 1.2472320795059204,
      "learning_rate": 1.133e-05,
      "loss": 0.5701,
      "step": 2268
    },
    {
      "epoch": 0.1270579012207414,
      "grad_norm": 1.1968580484390259,
      "learning_rate": 1.1335e-05,
      "loss": 0.4281,
      "step": 2269
    },
    {
      "epoch": 0.12711389853287042,
      "grad_norm": 1.3408206701278687,
      "learning_rate": 1.134e-05,
      "loss": 0.4181,
      "step": 2270
    },
    {
      "epoch": 0.12716989584499944,
      "grad_norm": 1.2121702432632446,
      "learning_rate": 1.1345e-05,
      "loss": 0.4036,
      "step": 2271
    },
    {
      "epoch": 0.12722589315712846,
      "grad_norm": 1.0712432861328125,
      "learning_rate": 1.1350000000000001e-05,
      "loss": 0.4192,
      "step": 2272
    },
    {
      "epoch": 0.12728189046925747,
      "grad_norm": 1.1871991157531738,
      "learning_rate": 1.1355e-05,
      "loss": 0.3987,
      "step": 2273
    },
    {
      "epoch": 0.1273378877813865,
      "grad_norm": 1.0942261219024658,
      "learning_rate": 1.1360000000000001e-05,
      "loss": 0.4153,
      "step": 2274
    },
    {
      "epoch": 0.1273938850935155,
      "grad_norm": 1.2354952096939087,
      "learning_rate": 1.1365e-05,
      "loss": 0.5993,
      "step": 2275
    },
    {
      "epoch": 0.12744988240564453,
      "grad_norm": 1.4431953430175781,
      "learning_rate": 1.137e-05,
      "loss": 0.4021,
      "step": 2276
    },
    {
      "epoch": 0.12750587971777355,
      "grad_norm": 1.1220649480819702,
      "learning_rate": 1.1375e-05,
      "loss": 0.4295,
      "step": 2277
    },
    {
      "epoch": 0.12756187702990257,
      "grad_norm": 1.1681586503982544,
      "learning_rate": 1.1380000000000001e-05,
      "loss": 0.4202,
      "step": 2278
    },
    {
      "epoch": 0.1276178743420316,
      "grad_norm": 1.1978394985198975,
      "learning_rate": 1.1385000000000002e-05,
      "loss": 0.5624,
      "step": 2279
    },
    {
      "epoch": 0.1276738716541606,
      "grad_norm": 1.1839028596878052,
      "learning_rate": 1.1390000000000001e-05,
      "loss": 0.4514,
      "step": 2280
    },
    {
      "epoch": 0.12772986896628963,
      "grad_norm": 1.1288633346557617,
      "learning_rate": 1.1395e-05,
      "loss": 0.5819,
      "step": 2281
    },
    {
      "epoch": 0.12778586627841865,
      "grad_norm": 0.9166799783706665,
      "learning_rate": 1.1400000000000001e-05,
      "loss": 0.3726,
      "step": 2282
    },
    {
      "epoch": 0.12784186359054767,
      "grad_norm": 1.317753791809082,
      "learning_rate": 1.1405e-05,
      "loss": 0.3707,
      "step": 2283
    },
    {
      "epoch": 0.12789786090267666,
      "grad_norm": 1.3430582284927368,
      "learning_rate": 1.141e-05,
      "loss": 0.5908,
      "step": 2284
    },
    {
      "epoch": 0.12795385821480568,
      "grad_norm": 1.2576439380645752,
      "learning_rate": 1.1415e-05,
      "loss": 0.5255,
      "step": 2285
    },
    {
      "epoch": 0.1280098555269347,
      "grad_norm": 13.189178466796875,
      "learning_rate": 1.142e-05,
      "loss": 0.4606,
      "step": 2286
    },
    {
      "epoch": 0.12806585283906372,
      "grad_norm": 1.1022242307662964,
      "learning_rate": 1.1425000000000002e-05,
      "loss": 0.3685,
      "step": 2287
    },
    {
      "epoch": 0.12812185015119273,
      "grad_norm": 1.0560503005981445,
      "learning_rate": 1.143e-05,
      "loss": 0.4096,
      "step": 2288
    },
    {
      "epoch": 0.12817784746332175,
      "grad_norm": 1.2018636465072632,
      "learning_rate": 1.1435e-05,
      "loss": 0.4187,
      "step": 2289
    },
    {
      "epoch": 0.12823384477545077,
      "grad_norm": 1.204622745513916,
      "learning_rate": 1.144e-05,
      "loss": 0.4063,
      "step": 2290
    },
    {
      "epoch": 0.1282898420875798,
      "grad_norm": 1.3194224834442139,
      "learning_rate": 1.1445e-05,
      "loss": 0.6101,
      "step": 2291
    },
    {
      "epoch": 0.1283458393997088,
      "grad_norm": 1.1247808933258057,
      "learning_rate": 1.145e-05,
      "loss": 0.4507,
      "step": 2292
    },
    {
      "epoch": 0.12840183671183783,
      "grad_norm": 1.17900550365448,
      "learning_rate": 1.1455000000000001e-05,
      "loss": 0.4489,
      "step": 2293
    },
    {
      "epoch": 0.12845783402396685,
      "grad_norm": 1.0201923847198486,
      "learning_rate": 1.146e-05,
      "loss": 0.4107,
      "step": 2294
    },
    {
      "epoch": 0.12851383133609587,
      "grad_norm": 1.1454983949661255,
      "learning_rate": 1.1465000000000001e-05,
      "loss": 0.4719,
      "step": 2295
    },
    {
      "epoch": 0.1285698286482249,
      "grad_norm": 1.003904104232788,
      "learning_rate": 1.147e-05,
      "loss": 0.3516,
      "step": 2296
    },
    {
      "epoch": 0.1286258259603539,
      "grad_norm": 1.0594944953918457,
      "learning_rate": 1.1475000000000001e-05,
      "loss": 0.4612,
      "step": 2297
    },
    {
      "epoch": 0.12868182327248293,
      "grad_norm": 1.2085590362548828,
      "learning_rate": 1.148e-05,
      "loss": 0.4192,
      "step": 2298
    },
    {
      "epoch": 0.12873782058461194,
      "grad_norm": 1.2151544094085693,
      "learning_rate": 1.1485e-05,
      "loss": 0.3859,
      "step": 2299
    },
    {
      "epoch": 0.12879381789674096,
      "grad_norm": 1.1104528903961182,
      "learning_rate": 1.149e-05,
      "loss": 0.387,
      "step": 2300
    },
    {
      "epoch": 0.12884981520886998,
      "grad_norm": 1.1801403760910034,
      "learning_rate": 1.1495000000000001e-05,
      "loss": 0.4741,
      "step": 2301
    },
    {
      "epoch": 0.128905812520999,
      "grad_norm": 1.357081651687622,
      "learning_rate": 1.1500000000000002e-05,
      "loss": 0.5217,
      "step": 2302
    },
    {
      "epoch": 0.12896180983312802,
      "grad_norm": 1.403161883354187,
      "learning_rate": 1.1505e-05,
      "loss": 0.5856,
      "step": 2303
    },
    {
      "epoch": 0.12901780714525704,
      "grad_norm": 1.230513334274292,
      "learning_rate": 1.151e-05,
      "loss": 0.4272,
      "step": 2304
    },
    {
      "epoch": 0.12907380445738603,
      "grad_norm": 1.083959937095642,
      "learning_rate": 1.1515e-05,
      "loss": 0.4267,
      "step": 2305
    },
    {
      "epoch": 0.12912980176951505,
      "grad_norm": 1.5055561065673828,
      "learning_rate": 1.152e-05,
      "loss": 0.5365,
      "step": 2306
    },
    {
      "epoch": 0.12918579908164407,
      "grad_norm": 1.0667353868484497,
      "learning_rate": 1.1525e-05,
      "loss": 0.3674,
      "step": 2307
    },
    {
      "epoch": 0.1292417963937731,
      "grad_norm": 1.5784897804260254,
      "learning_rate": 1.153e-05,
      "loss": 0.5412,
      "step": 2308
    },
    {
      "epoch": 0.1292977937059021,
      "grad_norm": 1.2324721813201904,
      "learning_rate": 1.1535e-05,
      "loss": 0.4253,
      "step": 2309
    },
    {
      "epoch": 0.12935379101803113,
      "grad_norm": 1.1168930530548096,
      "learning_rate": 1.1540000000000001e-05,
      "loss": 0.565,
      "step": 2310
    },
    {
      "epoch": 0.12940978833016015,
      "grad_norm": 4.88847541809082,
      "learning_rate": 1.1545e-05,
      "loss": 0.3856,
      "step": 2311
    },
    {
      "epoch": 0.12946578564228917,
      "grad_norm": 1.3440314531326294,
      "learning_rate": 1.1550000000000001e-05,
      "loss": 0.4282,
      "step": 2312
    },
    {
      "epoch": 0.12952178295441819,
      "grad_norm": 1.2424455881118774,
      "learning_rate": 1.1555e-05,
      "loss": 0.5565,
      "step": 2313
    },
    {
      "epoch": 0.1295777802665472,
      "grad_norm": 1.1632091999053955,
      "learning_rate": 1.156e-05,
      "loss": 0.5432,
      "step": 2314
    },
    {
      "epoch": 0.12963377757867622,
      "grad_norm": 1.2187398672103882,
      "learning_rate": 1.1565e-05,
      "loss": 0.5051,
      "step": 2315
    },
    {
      "epoch": 0.12968977489080524,
      "grad_norm": 1.359572172164917,
      "learning_rate": 1.1570000000000001e-05,
      "loss": 0.4581,
      "step": 2316
    },
    {
      "epoch": 0.12974577220293426,
      "grad_norm": 1.4037069082260132,
      "learning_rate": 1.1575000000000002e-05,
      "loss": 0.5692,
      "step": 2317
    },
    {
      "epoch": 0.12980176951506328,
      "grad_norm": 1.292794942855835,
      "learning_rate": 1.1580000000000001e-05,
      "loss": 0.4582,
      "step": 2318
    },
    {
      "epoch": 0.1298577668271923,
      "grad_norm": 1.0918874740600586,
      "learning_rate": 1.1585e-05,
      "loss": 0.3554,
      "step": 2319
    },
    {
      "epoch": 0.12991376413932132,
      "grad_norm": 1.0295218229293823,
      "learning_rate": 1.159e-05,
      "loss": 0.3185,
      "step": 2320
    },
    {
      "epoch": 0.12996976145145034,
      "grad_norm": 1.1888872385025024,
      "learning_rate": 1.1595e-05,
      "loss": 0.3827,
      "step": 2321
    },
    {
      "epoch": 0.13002575876357936,
      "grad_norm": 1.218995451927185,
      "learning_rate": 1.16e-05,
      "loss": 0.4396,
      "step": 2322
    },
    {
      "epoch": 0.13008175607570838,
      "grad_norm": 1.046060562133789,
      "learning_rate": 1.1605e-05,
      "loss": 0.3439,
      "step": 2323
    },
    {
      "epoch": 0.1301377533878374,
      "grad_norm": 1.1816738843917847,
      "learning_rate": 1.161e-05,
      "loss": 0.4889,
      "step": 2324
    },
    {
      "epoch": 0.13019375069996642,
      "grad_norm": 1.1769605875015259,
      "learning_rate": 1.1615000000000001e-05,
      "loss": 0.5111,
      "step": 2325
    },
    {
      "epoch": 0.1302497480120954,
      "grad_norm": 11.965676307678223,
      "learning_rate": 1.162e-05,
      "loss": 0.4498,
      "step": 2326
    },
    {
      "epoch": 0.13030574532422443,
      "grad_norm": 1.245835542678833,
      "learning_rate": 1.1625000000000001e-05,
      "loss": 0.4435,
      "step": 2327
    },
    {
      "epoch": 0.13036174263635344,
      "grad_norm": 1.261703372001648,
      "learning_rate": 1.163e-05,
      "loss": 0.3417,
      "step": 2328
    },
    {
      "epoch": 0.13041773994848246,
      "grad_norm": 1.4906578063964844,
      "learning_rate": 1.1635e-05,
      "loss": 0.4779,
      "step": 2329
    },
    {
      "epoch": 0.13047373726061148,
      "grad_norm": 1.5169082880020142,
      "learning_rate": 1.164e-05,
      "loss": 0.5212,
      "step": 2330
    },
    {
      "epoch": 0.1305297345727405,
      "grad_norm": 1.599611520767212,
      "learning_rate": 1.1645000000000001e-05,
      "loss": 0.4286,
      "step": 2331
    },
    {
      "epoch": 0.13058573188486952,
      "grad_norm": 1.9279402494430542,
      "learning_rate": 1.1650000000000002e-05,
      "loss": 0.466,
      "step": 2332
    },
    {
      "epoch": 0.13064172919699854,
      "grad_norm": 1.1912360191345215,
      "learning_rate": 1.1655000000000001e-05,
      "loss": 0.4664,
      "step": 2333
    },
    {
      "epoch": 0.13069772650912756,
      "grad_norm": 1.3635510206222534,
      "learning_rate": 1.166e-05,
      "loss": 0.4613,
      "step": 2334
    },
    {
      "epoch": 0.13075372382125658,
      "grad_norm": 1.300800085067749,
      "learning_rate": 1.1665000000000001e-05,
      "loss": 0.5828,
      "step": 2335
    },
    {
      "epoch": 0.1308097211333856,
      "grad_norm": 1.1869381666183472,
      "learning_rate": 1.167e-05,
      "loss": 0.3331,
      "step": 2336
    },
    {
      "epoch": 0.13086571844551462,
      "grad_norm": 1.4955719709396362,
      "learning_rate": 1.1675000000000001e-05,
      "loss": 0.5642,
      "step": 2337
    },
    {
      "epoch": 0.13092171575764364,
      "grad_norm": 1.167914867401123,
      "learning_rate": 1.168e-05,
      "loss": 0.4144,
      "step": 2338
    },
    {
      "epoch": 0.13097771306977266,
      "grad_norm": 1.3202574253082275,
      "learning_rate": 1.1685e-05,
      "loss": 0.4789,
      "step": 2339
    },
    {
      "epoch": 0.13103371038190167,
      "grad_norm": 1.2649132013320923,
      "learning_rate": 1.1690000000000002e-05,
      "loss": 0.4761,
      "step": 2340
    },
    {
      "epoch": 0.1310897076940307,
      "grad_norm": 1.065508484840393,
      "learning_rate": 1.1695e-05,
      "loss": 0.4014,
      "step": 2341
    },
    {
      "epoch": 0.1311457050061597,
      "grad_norm": 0.9914081692695618,
      "learning_rate": 1.1700000000000001e-05,
      "loss": 0.3879,
      "step": 2342
    },
    {
      "epoch": 0.13120170231828873,
      "grad_norm": 1.1917837858200073,
      "learning_rate": 1.1705e-05,
      "loss": 0.4464,
      "step": 2343
    },
    {
      "epoch": 0.13125769963041775,
      "grad_norm": 1.1857242584228516,
      "learning_rate": 1.171e-05,
      "loss": 0.3995,
      "step": 2344
    },
    {
      "epoch": 0.13131369694254677,
      "grad_norm": 1.0622613430023193,
      "learning_rate": 1.1715e-05,
      "loss": 0.3963,
      "step": 2345
    },
    {
      "epoch": 0.13136969425467576,
      "grad_norm": 1.1169847249984741,
      "learning_rate": 1.172e-05,
      "loss": 0.3781,
      "step": 2346
    },
    {
      "epoch": 0.13142569156680478,
      "grad_norm": 1.2596418857574463,
      "learning_rate": 1.1725e-05,
      "loss": 0.4689,
      "step": 2347
    },
    {
      "epoch": 0.1314816888789338,
      "grad_norm": 1.2164725065231323,
      "learning_rate": 1.1730000000000001e-05,
      "loss": 0.4248,
      "step": 2348
    },
    {
      "epoch": 0.13153768619106282,
      "grad_norm": 1.1801409721374512,
      "learning_rate": 1.1735e-05,
      "loss": 0.4889,
      "step": 2349
    },
    {
      "epoch": 0.13159368350319184,
      "grad_norm": 0.9698452949523926,
      "learning_rate": 1.1740000000000001e-05,
      "loss": 0.4187,
      "step": 2350
    },
    {
      "epoch": 0.13164968081532086,
      "grad_norm": 1.3726561069488525,
      "learning_rate": 1.1745e-05,
      "loss": 0.6082,
      "step": 2351
    },
    {
      "epoch": 0.13170567812744988,
      "grad_norm": 1.310638666152954,
      "learning_rate": 1.175e-05,
      "loss": 0.4059,
      "step": 2352
    },
    {
      "epoch": 0.1317616754395789,
      "grad_norm": 1.0972820520401,
      "learning_rate": 1.1755e-05,
      "loss": 0.5182,
      "step": 2353
    },
    {
      "epoch": 0.13181767275170791,
      "grad_norm": 1.407310128211975,
      "learning_rate": 1.1760000000000001e-05,
      "loss": 0.4274,
      "step": 2354
    },
    {
      "epoch": 0.13187367006383693,
      "grad_norm": 1.238159418106079,
      "learning_rate": 1.1765000000000002e-05,
      "loss": 0.4569,
      "step": 2355
    },
    {
      "epoch": 0.13192966737596595,
      "grad_norm": 1.4061623811721802,
      "learning_rate": 1.177e-05,
      "loss": 0.5208,
      "step": 2356
    },
    {
      "epoch": 0.13198566468809497,
      "grad_norm": 1.0815753936767578,
      "learning_rate": 1.1775e-05,
      "loss": 0.5048,
      "step": 2357
    },
    {
      "epoch": 0.132041662000224,
      "grad_norm": 1.5622210502624512,
      "learning_rate": 1.178e-05,
      "loss": 0.4615,
      "step": 2358
    },
    {
      "epoch": 0.132097659312353,
      "grad_norm": 1.1877309083938599,
      "learning_rate": 1.1785e-05,
      "loss": 0.4608,
      "step": 2359
    },
    {
      "epoch": 0.13215365662448203,
      "grad_norm": 1.7222483158111572,
      "learning_rate": 1.179e-05,
      "loss": 0.4566,
      "step": 2360
    },
    {
      "epoch": 0.13220965393661105,
      "grad_norm": 0.9789606928825378,
      "learning_rate": 1.1795e-05,
      "loss": 0.3139,
      "step": 2361
    },
    {
      "epoch": 0.13226565124874007,
      "grad_norm": 1.3129633665084839,
      "learning_rate": 1.18e-05,
      "loss": 0.5787,
      "step": 2362
    },
    {
      "epoch": 0.1323216485608691,
      "grad_norm": 1.207702398300171,
      "learning_rate": 1.1805000000000001e-05,
      "loss": 0.5146,
      "step": 2363
    },
    {
      "epoch": 0.1323776458729981,
      "grad_norm": 1.3435063362121582,
      "learning_rate": 1.181e-05,
      "loss": 0.4662,
      "step": 2364
    },
    {
      "epoch": 0.13243364318512713,
      "grad_norm": 1.13418710231781,
      "learning_rate": 1.1815000000000001e-05,
      "loss": 0.3667,
      "step": 2365
    },
    {
      "epoch": 0.13248964049725614,
      "grad_norm": 1.2030316591262817,
      "learning_rate": 1.182e-05,
      "loss": 0.3939,
      "step": 2366
    },
    {
      "epoch": 0.13254563780938514,
      "grad_norm": 1.1721278429031372,
      "learning_rate": 1.1825e-05,
      "loss": 0.5194,
      "step": 2367
    },
    {
      "epoch": 0.13260163512151416,
      "grad_norm": 1.1768015623092651,
      "learning_rate": 1.183e-05,
      "loss": 0.3973,
      "step": 2368
    },
    {
      "epoch": 0.13265763243364317,
      "grad_norm": 1.225443720817566,
      "learning_rate": 1.1835000000000001e-05,
      "loss": 0.4661,
      "step": 2369
    },
    {
      "epoch": 0.1327136297457722,
      "grad_norm": 1.093108892440796,
      "learning_rate": 1.1840000000000002e-05,
      "loss": 0.3354,
      "step": 2370
    },
    {
      "epoch": 0.1327696270579012,
      "grad_norm": 1.1921272277832031,
      "learning_rate": 1.1845000000000001e-05,
      "loss": 0.4253,
      "step": 2371
    },
    {
      "epoch": 0.13282562437003023,
      "grad_norm": 1.2796928882598877,
      "learning_rate": 1.185e-05,
      "loss": 0.4149,
      "step": 2372
    },
    {
      "epoch": 0.13288162168215925,
      "grad_norm": 1.2408664226531982,
      "learning_rate": 1.1855e-05,
      "loss": 0.4325,
      "step": 2373
    },
    {
      "epoch": 0.13293761899428827,
      "grad_norm": 1.502134919166565,
      "learning_rate": 1.186e-05,
      "loss": 0.5105,
      "step": 2374
    },
    {
      "epoch": 0.1329936163064173,
      "grad_norm": 1.290805459022522,
      "learning_rate": 1.1865e-05,
      "loss": 0.5634,
      "step": 2375
    },
    {
      "epoch": 0.1330496136185463,
      "grad_norm": 1.6113225221633911,
      "learning_rate": 1.187e-05,
      "loss": 0.5746,
      "step": 2376
    },
    {
      "epoch": 0.13310561093067533,
      "grad_norm": 1.0114905834197998,
      "learning_rate": 1.1875e-05,
      "loss": 0.4165,
      "step": 2377
    },
    {
      "epoch": 0.13316160824280435,
      "grad_norm": 1.1047070026397705,
      "learning_rate": 1.1880000000000001e-05,
      "loss": 0.4955,
      "step": 2378
    },
    {
      "epoch": 0.13321760555493337,
      "grad_norm": 1.0724822282791138,
      "learning_rate": 1.1885e-05,
      "loss": 0.4241,
      "step": 2379
    },
    {
      "epoch": 0.13327360286706239,
      "grad_norm": 1.5702725648880005,
      "learning_rate": 1.1890000000000001e-05,
      "loss": 0.3546,
      "step": 2380
    },
    {
      "epoch": 0.1333296001791914,
      "grad_norm": 1.3315016031265259,
      "learning_rate": 1.1895e-05,
      "loss": 0.4321,
      "step": 2381
    },
    {
      "epoch": 0.13338559749132042,
      "grad_norm": 2.0095887184143066,
      "learning_rate": 1.19e-05,
      "loss": 0.4651,
      "step": 2382
    },
    {
      "epoch": 0.13344159480344944,
      "grad_norm": 1.2531713247299194,
      "learning_rate": 1.1905e-05,
      "loss": 0.4597,
      "step": 2383
    },
    {
      "epoch": 0.13349759211557846,
      "grad_norm": 1.2469711303710938,
      "learning_rate": 1.1910000000000001e-05,
      "loss": 0.4425,
      "step": 2384
    },
    {
      "epoch": 0.13355358942770748,
      "grad_norm": 1.7293294668197632,
      "learning_rate": 1.1915000000000002e-05,
      "loss": 0.4855,
      "step": 2385
    },
    {
      "epoch": 0.1336095867398365,
      "grad_norm": 1.332313895225525,
      "learning_rate": 1.1920000000000001e-05,
      "loss": 0.3875,
      "step": 2386
    },
    {
      "epoch": 0.13366558405196552,
      "grad_norm": 1.1652700901031494,
      "learning_rate": 1.1925e-05,
      "loss": 0.5291,
      "step": 2387
    },
    {
      "epoch": 0.1337215813640945,
      "grad_norm": 1.0502643585205078,
      "learning_rate": 1.1930000000000001e-05,
      "loss": 0.3097,
      "step": 2388
    },
    {
      "epoch": 0.13377757867622353,
      "grad_norm": 1.5209193229675293,
      "learning_rate": 1.1935e-05,
      "loss": 0.4903,
      "step": 2389
    },
    {
      "epoch": 0.13383357598835255,
      "grad_norm": 1.1833019256591797,
      "learning_rate": 1.1940000000000001e-05,
      "loss": 0.4808,
      "step": 2390
    },
    {
      "epoch": 0.13388957330048157,
      "grad_norm": 1.3718583583831787,
      "learning_rate": 1.1945e-05,
      "loss": 0.4008,
      "step": 2391
    },
    {
      "epoch": 0.1339455706126106,
      "grad_norm": 1.2104849815368652,
      "learning_rate": 1.195e-05,
      "loss": 0.4653,
      "step": 2392
    },
    {
      "epoch": 0.1340015679247396,
      "grad_norm": 1.029047966003418,
      "learning_rate": 1.1955000000000002e-05,
      "loss": 0.4049,
      "step": 2393
    },
    {
      "epoch": 0.13405756523686863,
      "grad_norm": 3.4568305015563965,
      "learning_rate": 1.196e-05,
      "loss": 0.4246,
      "step": 2394
    },
    {
      "epoch": 0.13411356254899764,
      "grad_norm": 1.1154710054397583,
      "learning_rate": 1.1965000000000001e-05,
      "loss": 0.4391,
      "step": 2395
    },
    {
      "epoch": 0.13416955986112666,
      "grad_norm": 1.150034785270691,
      "learning_rate": 1.197e-05,
      "loss": 0.3921,
      "step": 2396
    },
    {
      "epoch": 0.13422555717325568,
      "grad_norm": 1.2795981168746948,
      "learning_rate": 1.1975e-05,
      "loss": 0.3264,
      "step": 2397
    },
    {
      "epoch": 0.1342815544853847,
      "grad_norm": 1.226135015487671,
      "learning_rate": 1.198e-05,
      "loss": 0.5624,
      "step": 2398
    },
    {
      "epoch": 0.13433755179751372,
      "grad_norm": 1.088365077972412,
      "learning_rate": 1.1985e-05,
      "loss": 0.3313,
      "step": 2399
    },
    {
      "epoch": 0.13439354910964274,
      "grad_norm": 1.1446475982666016,
      "learning_rate": 1.199e-05,
      "loss": 0.4507,
      "step": 2400
    },
    {
      "epoch": 0.13444954642177176,
      "grad_norm": 1.3302208185195923,
      "learning_rate": 1.1995000000000001e-05,
      "loss": 0.4239,
      "step": 2401
    },
    {
      "epoch": 0.13450554373390078,
      "grad_norm": 1.1120394468307495,
      "learning_rate": 1.2e-05,
      "loss": 0.4509,
      "step": 2402
    },
    {
      "epoch": 0.1345615410460298,
      "grad_norm": 1.155068278312683,
      "learning_rate": 1.2005000000000001e-05,
      "loss": 0.434,
      "step": 2403
    },
    {
      "epoch": 0.13461753835815882,
      "grad_norm": 1.1789742708206177,
      "learning_rate": 1.201e-05,
      "loss": 0.2487,
      "step": 2404
    },
    {
      "epoch": 0.13467353567028784,
      "grad_norm": 1.2293139696121216,
      "learning_rate": 1.2015000000000001e-05,
      "loss": 0.5815,
      "step": 2405
    },
    {
      "epoch": 0.13472953298241686,
      "grad_norm": 1.2889957427978516,
      "learning_rate": 1.202e-05,
      "loss": 0.6689,
      "step": 2406
    },
    {
      "epoch": 0.13478553029454587,
      "grad_norm": 1.4162606000900269,
      "learning_rate": 1.2025000000000001e-05,
      "loss": 0.4166,
      "step": 2407
    },
    {
      "epoch": 0.13484152760667487,
      "grad_norm": 1.2048757076263428,
      "learning_rate": 1.2030000000000002e-05,
      "loss": 0.4165,
      "step": 2408
    },
    {
      "epoch": 0.13489752491880388,
      "grad_norm": 1.0300843715667725,
      "learning_rate": 1.2035e-05,
      "loss": 0.3927,
      "step": 2409
    },
    {
      "epoch": 0.1349535222309329,
      "grad_norm": 1.2152204513549805,
      "learning_rate": 1.204e-05,
      "loss": 0.507,
      "step": 2410
    },
    {
      "epoch": 0.13500951954306192,
      "grad_norm": 1.2543202638626099,
      "learning_rate": 1.2045e-05,
      "loss": 0.3755,
      "step": 2411
    },
    {
      "epoch": 0.13506551685519094,
      "grad_norm": 1.1340587139129639,
      "learning_rate": 1.205e-05,
      "loss": 0.4536,
      "step": 2412
    },
    {
      "epoch": 0.13512151416731996,
      "grad_norm": 1.5051394701004028,
      "learning_rate": 1.2055e-05,
      "loss": 0.4643,
      "step": 2413
    },
    {
      "epoch": 0.13517751147944898,
      "grad_norm": 1.0842450857162476,
      "learning_rate": 1.206e-05,
      "loss": 0.3299,
      "step": 2414
    },
    {
      "epoch": 0.135233508791578,
      "grad_norm": 1.1488196849822998,
      "learning_rate": 1.2065e-05,
      "loss": 0.445,
      "step": 2415
    },
    {
      "epoch": 0.13528950610370702,
      "grad_norm": 1.1640756130218506,
      "learning_rate": 1.2070000000000001e-05,
      "loss": 0.4106,
      "step": 2416
    },
    {
      "epoch": 0.13534550341583604,
      "grad_norm": 1.4307230710983276,
      "learning_rate": 1.2075e-05,
      "loss": 0.4711,
      "step": 2417
    },
    {
      "epoch": 0.13540150072796506,
      "grad_norm": 1.7485276460647583,
      "learning_rate": 1.2080000000000001e-05,
      "loss": 0.6179,
      "step": 2418
    },
    {
      "epoch": 0.13545749804009408,
      "grad_norm": 1.2945975065231323,
      "learning_rate": 1.2085e-05,
      "loss": 0.5587,
      "step": 2419
    },
    {
      "epoch": 0.1355134953522231,
      "grad_norm": 1.1517056226730347,
      "learning_rate": 1.209e-05,
      "loss": 0.3638,
      "step": 2420
    },
    {
      "epoch": 0.13556949266435211,
      "grad_norm": 1.4271259307861328,
      "learning_rate": 1.2095e-05,
      "loss": 0.4494,
      "step": 2421
    },
    {
      "epoch": 0.13562548997648113,
      "grad_norm": 1.8111350536346436,
      "learning_rate": 1.2100000000000001e-05,
      "loss": 0.5689,
      "step": 2422
    },
    {
      "epoch": 0.13568148728861015,
      "grad_norm": 1.0819214582443237,
      "learning_rate": 1.2105000000000002e-05,
      "loss": 0.4173,
      "step": 2423
    },
    {
      "epoch": 0.13573748460073917,
      "grad_norm": 1.2219858169555664,
      "learning_rate": 1.2110000000000001e-05,
      "loss": 0.4282,
      "step": 2424
    },
    {
      "epoch": 0.1357934819128682,
      "grad_norm": 1.6722608804702759,
      "learning_rate": 1.2115e-05,
      "loss": 0.5747,
      "step": 2425
    },
    {
      "epoch": 0.1358494792249972,
      "grad_norm": 1.2269113063812256,
      "learning_rate": 1.2120000000000001e-05,
      "loss": 0.3372,
      "step": 2426
    },
    {
      "epoch": 0.13590547653712623,
      "grad_norm": 1.5652239322662354,
      "learning_rate": 1.2125e-05,
      "loss": 0.4925,
      "step": 2427
    },
    {
      "epoch": 0.13596147384925525,
      "grad_norm": 1.2126359939575195,
      "learning_rate": 1.213e-05,
      "loss": 0.3988,
      "step": 2428
    },
    {
      "epoch": 0.13601747116138424,
      "grad_norm": 1.5335955619812012,
      "learning_rate": 1.2135e-05,
      "loss": 0.4273,
      "step": 2429
    },
    {
      "epoch": 0.13607346847351326,
      "grad_norm": 1.3891562223434448,
      "learning_rate": 1.214e-05,
      "loss": 0.4148,
      "step": 2430
    },
    {
      "epoch": 0.13612946578564228,
      "grad_norm": 1.6771533489227295,
      "learning_rate": 1.2145000000000001e-05,
      "loss": 0.5713,
      "step": 2431
    },
    {
      "epoch": 0.1361854630977713,
      "grad_norm": 1.255646824836731,
      "learning_rate": 1.215e-05,
      "loss": 0.4457,
      "step": 2432
    },
    {
      "epoch": 0.13624146040990032,
      "grad_norm": 1.1224887371063232,
      "learning_rate": 1.2155000000000001e-05,
      "loss": 0.3463,
      "step": 2433
    },
    {
      "epoch": 0.13629745772202934,
      "grad_norm": 1.122126817703247,
      "learning_rate": 1.216e-05,
      "loss": 0.4545,
      "step": 2434
    },
    {
      "epoch": 0.13635345503415836,
      "grad_norm": 1.541749119758606,
      "learning_rate": 1.2165e-05,
      "loss": 0.4453,
      "step": 2435
    },
    {
      "epoch": 0.13640945234628737,
      "grad_norm": 1.2225465774536133,
      "learning_rate": 1.217e-05,
      "loss": 0.5011,
      "step": 2436
    },
    {
      "epoch": 0.1364654496584164,
      "grad_norm": 1.091254711151123,
      "learning_rate": 1.2175e-05,
      "loss": 0.4753,
      "step": 2437
    },
    {
      "epoch": 0.1365214469705454,
      "grad_norm": 1.122660517692566,
      "learning_rate": 1.2180000000000002e-05,
      "loss": 0.438,
      "step": 2438
    },
    {
      "epoch": 0.13657744428267443,
      "grad_norm": 1.0282984972000122,
      "learning_rate": 1.2185000000000001e-05,
      "loss": 0.4516,
      "step": 2439
    },
    {
      "epoch": 0.13663344159480345,
      "grad_norm": 1.0135828256607056,
      "learning_rate": 1.219e-05,
      "loss": 0.4791,
      "step": 2440
    },
    {
      "epoch": 0.13668943890693247,
      "grad_norm": 1.4245688915252686,
      "learning_rate": 1.2195000000000001e-05,
      "loss": 0.462,
      "step": 2441
    },
    {
      "epoch": 0.1367454362190615,
      "grad_norm": 1.4579600095748901,
      "learning_rate": 1.22e-05,
      "loss": 0.5111,
      "step": 2442
    },
    {
      "epoch": 0.1368014335311905,
      "grad_norm": 1.064908742904663,
      "learning_rate": 1.2205000000000001e-05,
      "loss": 0.368,
      "step": 2443
    },
    {
      "epoch": 0.13685743084331953,
      "grad_norm": 1.4000297784805298,
      "learning_rate": 1.221e-05,
      "loss": 0.3009,
      "step": 2444
    },
    {
      "epoch": 0.13691342815544855,
      "grad_norm": 1.2969307899475098,
      "learning_rate": 1.2215e-05,
      "loss": 0.3928,
      "step": 2445
    },
    {
      "epoch": 0.13696942546757757,
      "grad_norm": 3.748359441757202,
      "learning_rate": 1.2220000000000002e-05,
      "loss": 0.3718,
      "step": 2446
    },
    {
      "epoch": 0.13702542277970658,
      "grad_norm": 1.4786757230758667,
      "learning_rate": 1.2225e-05,
      "loss": 0.4797,
      "step": 2447
    },
    {
      "epoch": 0.1370814200918356,
      "grad_norm": 1.3004447221755981,
      "learning_rate": 1.2230000000000001e-05,
      "loss": 0.5356,
      "step": 2448
    },
    {
      "epoch": 0.13713741740396462,
      "grad_norm": 1.1472911834716797,
      "learning_rate": 1.2235e-05,
      "loss": 0.3354,
      "step": 2449
    },
    {
      "epoch": 0.13719341471609361,
      "grad_norm": 1.274685025215149,
      "learning_rate": 1.224e-05,
      "loss": 0.4278,
      "step": 2450
    },
    {
      "epoch": 0.13724941202822263,
      "grad_norm": 1.412546157836914,
      "learning_rate": 1.2245e-05,
      "loss": 0.5157,
      "step": 2451
    },
    {
      "epoch": 0.13730540934035165,
      "grad_norm": 1.0137228965759277,
      "learning_rate": 1.225e-05,
      "loss": 0.4446,
      "step": 2452
    },
    {
      "epoch": 0.13736140665248067,
      "grad_norm": 2.3677423000335693,
      "learning_rate": 1.2255e-05,
      "loss": 0.5029,
      "step": 2453
    },
    {
      "epoch": 0.1374174039646097,
      "grad_norm": 1.2410162687301636,
      "learning_rate": 1.2260000000000001e-05,
      "loss": 0.4446,
      "step": 2454
    },
    {
      "epoch": 0.1374734012767387,
      "grad_norm": 1.2138422727584839,
      "learning_rate": 1.2265e-05,
      "loss": 0.4604,
      "step": 2455
    },
    {
      "epoch": 0.13752939858886773,
      "grad_norm": 1.2067878246307373,
      "learning_rate": 1.2270000000000001e-05,
      "loss": 0.4757,
      "step": 2456
    },
    {
      "epoch": 0.13758539590099675,
      "grad_norm": 1.2064911127090454,
      "learning_rate": 1.2275e-05,
      "loss": 0.4175,
      "step": 2457
    },
    {
      "epoch": 0.13764139321312577,
      "grad_norm": 1.3723119497299194,
      "learning_rate": 1.2280000000000001e-05,
      "loss": 0.4115,
      "step": 2458
    },
    {
      "epoch": 0.1376973905252548,
      "grad_norm": 1.29324209690094,
      "learning_rate": 1.2285e-05,
      "loss": 0.4133,
      "step": 2459
    },
    {
      "epoch": 0.1377533878373838,
      "grad_norm": 1.12079656124115,
      "learning_rate": 1.2290000000000001e-05,
      "loss": 0.4476,
      "step": 2460
    },
    {
      "epoch": 0.13780938514951283,
      "grad_norm": 1.2613892555236816,
      "learning_rate": 1.2295000000000002e-05,
      "loss": 0.5619,
      "step": 2461
    },
    {
      "epoch": 0.13786538246164184,
      "grad_norm": 1.4234511852264404,
      "learning_rate": 1.23e-05,
      "loss": 0.5854,
      "step": 2462
    },
    {
      "epoch": 0.13792137977377086,
      "grad_norm": 1.1991769075393677,
      "learning_rate": 1.2305000000000002e-05,
      "loss": 0.4518,
      "step": 2463
    },
    {
      "epoch": 0.13797737708589988,
      "grad_norm": 1.281947135925293,
      "learning_rate": 1.231e-05,
      "loss": 0.4044,
      "step": 2464
    },
    {
      "epoch": 0.1380333743980289,
      "grad_norm": 1.3853332996368408,
      "learning_rate": 1.2315e-05,
      "loss": 0.6167,
      "step": 2465
    },
    {
      "epoch": 0.13808937171015792,
      "grad_norm": 1.23692786693573,
      "learning_rate": 1.232e-05,
      "loss": 0.4404,
      "step": 2466
    },
    {
      "epoch": 0.13814536902228694,
      "grad_norm": 14.155194282531738,
      "learning_rate": 1.2325e-05,
      "loss": 0.7027,
      "step": 2467
    },
    {
      "epoch": 0.13820136633441596,
      "grad_norm": 1.47590172290802,
      "learning_rate": 1.233e-05,
      "loss": 0.3997,
      "step": 2468
    },
    {
      "epoch": 0.13825736364654498,
      "grad_norm": 1.3184940814971924,
      "learning_rate": 1.2335000000000001e-05,
      "loss": 0.3687,
      "step": 2469
    },
    {
      "epoch": 0.13831336095867397,
      "grad_norm": 1.1062796115875244,
      "learning_rate": 1.234e-05,
      "loss": 0.4121,
      "step": 2470
    },
    {
      "epoch": 0.138369358270803,
      "grad_norm": 1.1457158327102661,
      "learning_rate": 1.2345000000000001e-05,
      "loss": 0.5695,
      "step": 2471
    },
    {
      "epoch": 0.138425355582932,
      "grad_norm": 1.1479376554489136,
      "learning_rate": 1.235e-05,
      "loss": 0.5058,
      "step": 2472
    },
    {
      "epoch": 0.13848135289506103,
      "grad_norm": 1.411997675895691,
      "learning_rate": 1.2355e-05,
      "loss": 0.5817,
      "step": 2473
    },
    {
      "epoch": 0.13853735020719005,
      "grad_norm": 1.0818930864334106,
      "learning_rate": 1.236e-05,
      "loss": 0.5058,
      "step": 2474
    },
    {
      "epoch": 0.13859334751931907,
      "grad_norm": 1.1070023775100708,
      "learning_rate": 1.2365e-05,
      "loss": 0.3422,
      "step": 2475
    },
    {
      "epoch": 0.13864934483144808,
      "grad_norm": 1.2352498769760132,
      "learning_rate": 1.2370000000000002e-05,
      "loss": 0.3878,
      "step": 2476
    },
    {
      "epoch": 0.1387053421435771,
      "grad_norm": 1.0397839546203613,
      "learning_rate": 1.2375000000000001e-05,
      "loss": 0.3501,
      "step": 2477
    },
    {
      "epoch": 0.13876133945570612,
      "grad_norm": 1.1352453231811523,
      "learning_rate": 1.238e-05,
      "loss": 0.3688,
      "step": 2478
    },
    {
      "epoch": 0.13881733676783514,
      "grad_norm": 1.3301494121551514,
      "learning_rate": 1.2385000000000001e-05,
      "loss": 0.5199,
      "step": 2479
    },
    {
      "epoch": 0.13887333407996416,
      "grad_norm": 1.426438570022583,
      "learning_rate": 1.239e-05,
      "loss": 0.453,
      "step": 2480
    },
    {
      "epoch": 0.13892933139209318,
      "grad_norm": 1.164884090423584,
      "learning_rate": 1.2395e-05,
      "loss": 0.323,
      "step": 2481
    },
    {
      "epoch": 0.1389853287042222,
      "grad_norm": 1.1007789373397827,
      "learning_rate": 1.24e-05,
      "loss": 0.3785,
      "step": 2482
    },
    {
      "epoch": 0.13904132601635122,
      "grad_norm": 1.163779854774475,
      "learning_rate": 1.2405e-05,
      "loss": 0.4914,
      "step": 2483
    },
    {
      "epoch": 0.13909732332848024,
      "grad_norm": 1.3361643552780151,
      "learning_rate": 1.2410000000000001e-05,
      "loss": 0.4419,
      "step": 2484
    },
    {
      "epoch": 0.13915332064060926,
      "grad_norm": 1.0904892683029175,
      "learning_rate": 1.2415e-05,
      "loss": 0.4141,
      "step": 2485
    },
    {
      "epoch": 0.13920931795273828,
      "grad_norm": 1.2341094017028809,
      "learning_rate": 1.2420000000000001e-05,
      "loss": 0.4362,
      "step": 2486
    },
    {
      "epoch": 0.1392653152648673,
      "grad_norm": 1.0510798692703247,
      "learning_rate": 1.2425e-05,
      "loss": 0.415,
      "step": 2487
    },
    {
      "epoch": 0.13932131257699631,
      "grad_norm": 0.9391849637031555,
      "learning_rate": 1.243e-05,
      "loss": 0.4643,
      "step": 2488
    },
    {
      "epoch": 0.13937730988912533,
      "grad_norm": 1.0622152090072632,
      "learning_rate": 1.2435e-05,
      "loss": 0.3353,
      "step": 2489
    },
    {
      "epoch": 0.13943330720125435,
      "grad_norm": 1.3571791648864746,
      "learning_rate": 1.244e-05,
      "loss": 0.4687,
      "step": 2490
    },
    {
      "epoch": 0.13948930451338334,
      "grad_norm": 1.1573200225830078,
      "learning_rate": 1.2445e-05,
      "loss": 0.3999,
      "step": 2491
    },
    {
      "epoch": 0.13954530182551236,
      "grad_norm": 1.1886022090911865,
      "learning_rate": 1.2450000000000001e-05,
      "loss": 0.3521,
      "step": 2492
    },
    {
      "epoch": 0.13960129913764138,
      "grad_norm": 2.901169538497925,
      "learning_rate": 1.2455e-05,
      "loss": 0.4332,
      "step": 2493
    },
    {
      "epoch": 0.1396572964497704,
      "grad_norm": 2.1086270809173584,
      "learning_rate": 1.2460000000000001e-05,
      "loss": 0.4126,
      "step": 2494
    },
    {
      "epoch": 0.13971329376189942,
      "grad_norm": 1.3888367414474487,
      "learning_rate": 1.2465e-05,
      "loss": 0.4817,
      "step": 2495
    },
    {
      "epoch": 0.13976929107402844,
      "grad_norm": 1.083674669265747,
      "learning_rate": 1.2470000000000001e-05,
      "loss": 0.383,
      "step": 2496
    },
    {
      "epoch": 0.13982528838615746,
      "grad_norm": 1.3220771551132202,
      "learning_rate": 1.2475e-05,
      "loss": 0.5275,
      "step": 2497
    },
    {
      "epoch": 0.13988128569828648,
      "grad_norm": 1.1915440559387207,
      "learning_rate": 1.248e-05,
      "loss": 0.4635,
      "step": 2498
    },
    {
      "epoch": 0.1399372830104155,
      "grad_norm": 1.0927143096923828,
      "learning_rate": 1.2485000000000002e-05,
      "loss": 0.4208,
      "step": 2499
    },
    {
      "epoch": 0.13999328032254452,
      "grad_norm": 1.2048420906066895,
      "learning_rate": 1.249e-05,
      "loss": 0.4241,
      "step": 2500
    },
    {
      "epoch": 0.14004927763467354,
      "grad_norm": 1.1832644939422607,
      "learning_rate": 1.2495000000000001e-05,
      "loss": 0.5051,
      "step": 2501
    },
    {
      "epoch": 0.14010527494680255,
      "grad_norm": 1.1782584190368652,
      "learning_rate": 1.25e-05,
      "loss": 0.5329,
      "step": 2502
    },
    {
      "epoch": 0.14016127225893157,
      "grad_norm": 1.2893174886703491,
      "learning_rate": 1.2505e-05,
      "loss": 0.4063,
      "step": 2503
    },
    {
      "epoch": 0.1402172695710606,
      "grad_norm": 1.0172559022903442,
      "learning_rate": 1.2509999999999999e-05,
      "loss": 0.3683,
      "step": 2504
    },
    {
      "epoch": 0.1402732668831896,
      "grad_norm": 1.2625501155853271,
      "learning_rate": 1.2515000000000001e-05,
      "loss": 0.3565,
      "step": 2505
    },
    {
      "epoch": 0.14032926419531863,
      "grad_norm": 1.2588109970092773,
      "learning_rate": 1.252e-05,
      "loss": 0.4849,
      "step": 2506
    },
    {
      "epoch": 0.14038526150744765,
      "grad_norm": 1.3349969387054443,
      "learning_rate": 1.2525000000000001e-05,
      "loss": 0.5433,
      "step": 2507
    },
    {
      "epoch": 0.14044125881957667,
      "grad_norm": 1.1840282678604126,
      "learning_rate": 1.253e-05,
      "loss": 0.5035,
      "step": 2508
    },
    {
      "epoch": 0.1404972561317057,
      "grad_norm": 1.236934781074524,
      "learning_rate": 1.2535e-05,
      "loss": 0.3751,
      "step": 2509
    },
    {
      "epoch": 0.1405532534438347,
      "grad_norm": 1.3336421251296997,
      "learning_rate": 1.2540000000000002e-05,
      "loss": 0.5805,
      "step": 2510
    },
    {
      "epoch": 0.14060925075596373,
      "grad_norm": 1.1406078338623047,
      "learning_rate": 1.2545000000000001e-05,
      "loss": 0.4124,
      "step": 2511
    },
    {
      "epoch": 0.14066524806809272,
      "grad_norm": 1.195516586303711,
      "learning_rate": 1.255e-05,
      "loss": 0.4189,
      "step": 2512
    },
    {
      "epoch": 0.14072124538022174,
      "grad_norm": 1.338142991065979,
      "learning_rate": 1.2555000000000001e-05,
      "loss": 0.46,
      "step": 2513
    },
    {
      "epoch": 0.14077724269235076,
      "grad_norm": 1.2000536918640137,
      "learning_rate": 1.256e-05,
      "loss": 0.5513,
      "step": 2514
    },
    {
      "epoch": 0.14083324000447978,
      "grad_norm": 1.066293716430664,
      "learning_rate": 1.2565000000000003e-05,
      "loss": 0.3948,
      "step": 2515
    },
    {
      "epoch": 0.1408892373166088,
      "grad_norm": 1.0766243934631348,
      "learning_rate": 1.2570000000000002e-05,
      "loss": 0.5143,
      "step": 2516
    },
    {
      "epoch": 0.14094523462873781,
      "grad_norm": 1.296629786491394,
      "learning_rate": 1.2575e-05,
      "loss": 0.4408,
      "step": 2517
    },
    {
      "epoch": 0.14100123194086683,
      "grad_norm": 1.11188542842865,
      "learning_rate": 1.258e-05,
      "loss": 0.4171,
      "step": 2518
    },
    {
      "epoch": 0.14105722925299585,
      "grad_norm": 1.1227508783340454,
      "learning_rate": 1.2584999999999999e-05,
      "loss": 0.571,
      "step": 2519
    },
    {
      "epoch": 0.14111322656512487,
      "grad_norm": 1.1131638288497925,
      "learning_rate": 1.2590000000000001e-05,
      "loss": 0.5317,
      "step": 2520
    },
    {
      "epoch": 0.1411692238772539,
      "grad_norm": 1.5281466245651245,
      "learning_rate": 1.2595e-05,
      "loss": 0.3774,
      "step": 2521
    },
    {
      "epoch": 0.1412252211893829,
      "grad_norm": 1.1421079635620117,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 0.4091,
      "step": 2522
    },
    {
      "epoch": 0.14128121850151193,
      "grad_norm": 1.5237642526626587,
      "learning_rate": 1.2605e-05,
      "loss": 0.446,
      "step": 2523
    },
    {
      "epoch": 0.14133721581364095,
      "grad_norm": 1.4897043704986572,
      "learning_rate": 1.261e-05,
      "loss": 0.5319,
      "step": 2524
    },
    {
      "epoch": 0.14139321312576997,
      "grad_norm": 1.6626853942871094,
      "learning_rate": 1.2615000000000002e-05,
      "loss": 0.5172,
      "step": 2525
    },
    {
      "epoch": 0.141449210437899,
      "grad_norm": 1.3545503616333008,
      "learning_rate": 1.2620000000000001e-05,
      "loss": 0.5435,
      "step": 2526
    },
    {
      "epoch": 0.141505207750028,
      "grad_norm": 3.209622383117676,
      "learning_rate": 1.2625e-05,
      "loss": 0.528,
      "step": 2527
    },
    {
      "epoch": 0.14156120506215703,
      "grad_norm": 1.1368006467819214,
      "learning_rate": 1.263e-05,
      "loss": 0.4018,
      "step": 2528
    },
    {
      "epoch": 0.14161720237428604,
      "grad_norm": 1.019187092781067,
      "learning_rate": 1.2635e-05,
      "loss": 0.329,
      "step": 2529
    },
    {
      "epoch": 0.14167319968641506,
      "grad_norm": 1.2222051620483398,
      "learning_rate": 1.2640000000000003e-05,
      "loss": 0.3975,
      "step": 2530
    },
    {
      "epoch": 0.14172919699854408,
      "grad_norm": 1.1408919095993042,
      "learning_rate": 1.2645000000000002e-05,
      "loss": 0.4006,
      "step": 2531
    },
    {
      "epoch": 0.14178519431067307,
      "grad_norm": 1.2872763872146606,
      "learning_rate": 1.2650000000000001e-05,
      "loss": 0.3789,
      "step": 2532
    },
    {
      "epoch": 0.1418411916228021,
      "grad_norm": 2.181398868560791,
      "learning_rate": 1.2655e-05,
      "loss": 0.3909,
      "step": 2533
    },
    {
      "epoch": 0.1418971889349311,
      "grad_norm": 1.4929990768432617,
      "learning_rate": 1.2659999999999999e-05,
      "loss": 0.4905,
      "step": 2534
    },
    {
      "epoch": 0.14195318624706013,
      "grad_norm": 1.4117430448532104,
      "learning_rate": 1.2665000000000002e-05,
      "loss": 0.4487,
      "step": 2535
    },
    {
      "epoch": 0.14200918355918915,
      "grad_norm": 1.4963791370391846,
      "learning_rate": 1.267e-05,
      "loss": 0.5112,
      "step": 2536
    },
    {
      "epoch": 0.14206518087131817,
      "grad_norm": 1.105371117591858,
      "learning_rate": 1.2675000000000001e-05,
      "loss": 0.3966,
      "step": 2537
    },
    {
      "epoch": 0.1421211781834472,
      "grad_norm": 1.4377449750900269,
      "learning_rate": 1.268e-05,
      "loss": 0.4421,
      "step": 2538
    },
    {
      "epoch": 0.1421771754955762,
      "grad_norm": 1.1643390655517578,
      "learning_rate": 1.2685e-05,
      "loss": 0.3825,
      "step": 2539
    },
    {
      "epoch": 0.14223317280770523,
      "grad_norm": 1.2283012866973877,
      "learning_rate": 1.2690000000000002e-05,
      "loss": 0.4719,
      "step": 2540
    },
    {
      "epoch": 0.14228917011983425,
      "grad_norm": 1.1322824954986572,
      "learning_rate": 1.2695000000000001e-05,
      "loss": 0.3118,
      "step": 2541
    },
    {
      "epoch": 0.14234516743196327,
      "grad_norm": 1.1129308938980103,
      "learning_rate": 1.27e-05,
      "loss": 0.3436,
      "step": 2542
    },
    {
      "epoch": 0.14240116474409228,
      "grad_norm": 1.2738077640533447,
      "learning_rate": 1.2705e-05,
      "loss": 0.3745,
      "step": 2543
    },
    {
      "epoch": 0.1424571620562213,
      "grad_norm": 1.1257846355438232,
      "learning_rate": 1.271e-05,
      "loss": 0.3746,
      "step": 2544
    },
    {
      "epoch": 0.14251315936835032,
      "grad_norm": 1.2664601802825928,
      "learning_rate": 1.2715000000000001e-05,
      "loss": 0.5354,
      "step": 2545
    },
    {
      "epoch": 0.14256915668047934,
      "grad_norm": 1.3434889316558838,
      "learning_rate": 1.2720000000000002e-05,
      "loss": 0.4078,
      "step": 2546
    },
    {
      "epoch": 0.14262515399260836,
      "grad_norm": 1.602239727973938,
      "learning_rate": 1.2725000000000001e-05,
      "loss": 0.4273,
      "step": 2547
    },
    {
      "epoch": 0.14268115130473738,
      "grad_norm": 1.0283304452896118,
      "learning_rate": 1.273e-05,
      "loss": 0.3713,
      "step": 2548
    },
    {
      "epoch": 0.1427371486168664,
      "grad_norm": 1.1225849390029907,
      "learning_rate": 1.2735e-05,
      "loss": 0.5288,
      "step": 2549
    },
    {
      "epoch": 0.14279314592899542,
      "grad_norm": 1.2365565299987793,
      "learning_rate": 1.2740000000000002e-05,
      "loss": 0.3731,
      "step": 2550
    },
    {
      "epoch": 0.14284914324112444,
      "grad_norm": 1.0750316381454468,
      "learning_rate": 1.2745e-05,
      "loss": 0.3926,
      "step": 2551
    },
    {
      "epoch": 0.14290514055325346,
      "grad_norm": 1.3363348245620728,
      "learning_rate": 1.2750000000000002e-05,
      "loss": 0.4665,
      "step": 2552
    },
    {
      "epoch": 0.14296113786538245,
      "grad_norm": 1.021571159362793,
      "learning_rate": 1.2755e-05,
      "loss": 0.4289,
      "step": 2553
    },
    {
      "epoch": 0.14301713517751147,
      "grad_norm": 1.1585955619812012,
      "learning_rate": 1.276e-05,
      "loss": 0.3102,
      "step": 2554
    },
    {
      "epoch": 0.1430731324896405,
      "grad_norm": 1.2394684553146362,
      "learning_rate": 1.2765000000000002e-05,
      "loss": 0.4,
      "step": 2555
    },
    {
      "epoch": 0.1431291298017695,
      "grad_norm": 1.216558575630188,
      "learning_rate": 1.2770000000000001e-05,
      "loss": 0.4016,
      "step": 2556
    },
    {
      "epoch": 0.14318512711389852,
      "grad_norm": 1.206903338432312,
      "learning_rate": 1.2775e-05,
      "loss": 0.4442,
      "step": 2557
    },
    {
      "epoch": 0.14324112442602754,
      "grad_norm": 0.900111198425293,
      "learning_rate": 1.278e-05,
      "loss": 0.3385,
      "step": 2558
    },
    {
      "epoch": 0.14329712173815656,
      "grad_norm": 1.3177870512008667,
      "learning_rate": 1.2785e-05,
      "loss": 0.3729,
      "step": 2559
    },
    {
      "epoch": 0.14335311905028558,
      "grad_norm": 1.273997187614441,
      "learning_rate": 1.2790000000000001e-05,
      "loss": 0.3623,
      "step": 2560
    },
    {
      "epoch": 0.1434091163624146,
      "grad_norm": 1.188696026802063,
      "learning_rate": 1.2795000000000002e-05,
      "loss": 0.3905,
      "step": 2561
    },
    {
      "epoch": 0.14346511367454362,
      "grad_norm": 1.4021480083465576,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.5325,
      "step": 2562
    },
    {
      "epoch": 0.14352111098667264,
      "grad_norm": 1.4283920526504517,
      "learning_rate": 1.2805e-05,
      "loss": 0.4506,
      "step": 2563
    },
    {
      "epoch": 0.14357710829880166,
      "grad_norm": 1.5277773141860962,
      "learning_rate": 1.281e-05,
      "loss": 0.5579,
      "step": 2564
    },
    {
      "epoch": 0.14363310561093068,
      "grad_norm": 1.1661988496780396,
      "learning_rate": 1.2814999999999998e-05,
      "loss": 0.3988,
      "step": 2565
    },
    {
      "epoch": 0.1436891029230597,
      "grad_norm": 1.3075827360153198,
      "learning_rate": 1.2820000000000001e-05,
      "loss": 0.4105,
      "step": 2566
    },
    {
      "epoch": 0.14374510023518872,
      "grad_norm": 1.5823769569396973,
      "learning_rate": 1.2825000000000002e-05,
      "loss": 0.5679,
      "step": 2567
    },
    {
      "epoch": 0.14380109754731774,
      "grad_norm": 1.0940037965774536,
      "learning_rate": 1.283e-05,
      "loss": 0.3605,
      "step": 2568
    },
    {
      "epoch": 0.14385709485944675,
      "grad_norm": 1.1388133764266968,
      "learning_rate": 1.2835e-05,
      "loss": 0.3079,
      "step": 2569
    },
    {
      "epoch": 0.14391309217157577,
      "grad_norm": 1.4478360414505005,
      "learning_rate": 1.2839999999999999e-05,
      "loss": 0.4968,
      "step": 2570
    },
    {
      "epoch": 0.1439690894837048,
      "grad_norm": 1.2097203731536865,
      "learning_rate": 1.2845000000000002e-05,
      "loss": 0.4297,
      "step": 2571
    },
    {
      "epoch": 0.1440250867958338,
      "grad_norm": 1.618012547492981,
      "learning_rate": 1.285e-05,
      "loss": 0.419,
      "step": 2572
    },
    {
      "epoch": 0.14408108410796283,
      "grad_norm": 1.4660046100616455,
      "learning_rate": 1.2855e-05,
      "loss": 0.3765,
      "step": 2573
    },
    {
      "epoch": 0.14413708142009182,
      "grad_norm": 1.2173787355422974,
      "learning_rate": 1.286e-05,
      "loss": 0.3228,
      "step": 2574
    },
    {
      "epoch": 0.14419307873222084,
      "grad_norm": 1.3873306512832642,
      "learning_rate": 1.2865e-05,
      "loss": 0.5869,
      "step": 2575
    },
    {
      "epoch": 0.14424907604434986,
      "grad_norm": 1.5580289363861084,
      "learning_rate": 1.2870000000000002e-05,
      "loss": 0.4932,
      "step": 2576
    },
    {
      "epoch": 0.14430507335647888,
      "grad_norm": 1.7026492357254028,
      "learning_rate": 1.2875000000000001e-05,
      "loss": 0.5149,
      "step": 2577
    },
    {
      "epoch": 0.1443610706686079,
      "grad_norm": 1.3750256299972534,
      "learning_rate": 1.288e-05,
      "loss": 0.495,
      "step": 2578
    },
    {
      "epoch": 0.14441706798073692,
      "grad_norm": 1.2004586458206177,
      "learning_rate": 1.2885e-05,
      "loss": 0.4588,
      "step": 2579
    },
    {
      "epoch": 0.14447306529286594,
      "grad_norm": 1.4534201622009277,
      "learning_rate": 1.2889999999999999e-05,
      "loss": 0.5208,
      "step": 2580
    },
    {
      "epoch": 0.14452906260499496,
      "grad_norm": 1.2705984115600586,
      "learning_rate": 1.2895000000000001e-05,
      "loss": 0.3537,
      "step": 2581
    },
    {
      "epoch": 0.14458505991712398,
      "grad_norm": 1.1209765672683716,
      "learning_rate": 1.29e-05,
      "loss": 0.4387,
      "step": 2582
    },
    {
      "epoch": 0.144641057229253,
      "grad_norm": 1.1241347789764404,
      "learning_rate": 1.2905000000000001e-05,
      "loss": 0.3922,
      "step": 2583
    },
    {
      "epoch": 0.14469705454138201,
      "grad_norm": 1.6147980690002441,
      "learning_rate": 1.291e-05,
      "loss": 0.4985,
      "step": 2584
    },
    {
      "epoch": 0.14475305185351103,
      "grad_norm": 1.542745590209961,
      "learning_rate": 1.2915e-05,
      "loss": 0.6905,
      "step": 2585
    },
    {
      "epoch": 0.14480904916564005,
      "grad_norm": 1.2592793703079224,
      "learning_rate": 1.2920000000000002e-05,
      "loss": 0.4949,
      "step": 2586
    },
    {
      "epoch": 0.14486504647776907,
      "grad_norm": 1.1622910499572754,
      "learning_rate": 1.2925e-05,
      "loss": 0.3956,
      "step": 2587
    },
    {
      "epoch": 0.1449210437898981,
      "grad_norm": 1.2540090084075928,
      "learning_rate": 1.293e-05,
      "loss": 0.5471,
      "step": 2588
    },
    {
      "epoch": 0.1449770411020271,
      "grad_norm": 1.256731390953064,
      "learning_rate": 1.2935e-05,
      "loss": 0.3732,
      "step": 2589
    },
    {
      "epoch": 0.14503303841415613,
      "grad_norm": 1.5238062143325806,
      "learning_rate": 1.294e-05,
      "loss": 0.6349,
      "step": 2590
    },
    {
      "epoch": 0.14508903572628515,
      "grad_norm": 1.0773160457611084,
      "learning_rate": 1.2945000000000002e-05,
      "loss": 0.4166,
      "step": 2591
    },
    {
      "epoch": 0.14514503303841417,
      "grad_norm": 1.1376806497573853,
      "learning_rate": 1.2950000000000001e-05,
      "loss": 0.4292,
      "step": 2592
    },
    {
      "epoch": 0.1452010303505432,
      "grad_norm": 1.2171424627304077,
      "learning_rate": 1.2955e-05,
      "loss": 0.3987,
      "step": 2593
    },
    {
      "epoch": 0.14525702766267218,
      "grad_norm": 1.1606625318527222,
      "learning_rate": 1.296e-05,
      "loss": 0.4594,
      "step": 2594
    },
    {
      "epoch": 0.1453130249748012,
      "grad_norm": 1.031876802444458,
      "learning_rate": 1.2964999999999999e-05,
      "loss": 0.5854,
      "step": 2595
    },
    {
      "epoch": 0.14536902228693022,
      "grad_norm": 1.1607109308242798,
      "learning_rate": 1.2970000000000001e-05,
      "loss": 0.4157,
      "step": 2596
    },
    {
      "epoch": 0.14542501959905924,
      "grad_norm": 1.0473979711532593,
      "learning_rate": 1.2975e-05,
      "loss": 0.426,
      "step": 2597
    },
    {
      "epoch": 0.14548101691118825,
      "grad_norm": 1.1430045366287231,
      "learning_rate": 1.2980000000000001e-05,
      "loss": 0.4142,
      "step": 2598
    },
    {
      "epoch": 0.14553701422331727,
      "grad_norm": 1.156032919883728,
      "learning_rate": 1.2985e-05,
      "loss": 0.4853,
      "step": 2599
    },
    {
      "epoch": 0.1455930115354463,
      "grad_norm": 1.1408398151397705,
      "learning_rate": 1.299e-05,
      "loss": 0.4186,
      "step": 2600
    },
    {
      "epoch": 0.1456490088475753,
      "grad_norm": 1.059762716293335,
      "learning_rate": 1.2995000000000002e-05,
      "loss": 0.4544,
      "step": 2601
    },
    {
      "epoch": 0.14570500615970433,
      "grad_norm": 1.4854395389556885,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.7862,
      "step": 2602
    },
    {
      "epoch": 0.14576100347183335,
      "grad_norm": 1.234014868736267,
      "learning_rate": 1.3005e-05,
      "loss": 0.4272,
      "step": 2603
    },
    {
      "epoch": 0.14581700078396237,
      "grad_norm": 5.808340072631836,
      "learning_rate": 1.301e-05,
      "loss": 0.4147,
      "step": 2604
    },
    {
      "epoch": 0.1458729980960914,
      "grad_norm": 1.2030842304229736,
      "learning_rate": 1.3015e-05,
      "loss": 0.3735,
      "step": 2605
    },
    {
      "epoch": 0.1459289954082204,
      "grad_norm": 1.3522212505340576,
      "learning_rate": 1.3020000000000002e-05,
      "loss": 0.3872,
      "step": 2606
    },
    {
      "epoch": 0.14598499272034943,
      "grad_norm": 1.1150144338607788,
      "learning_rate": 1.3025000000000002e-05,
      "loss": 0.5558,
      "step": 2607
    },
    {
      "epoch": 0.14604099003247845,
      "grad_norm": 1.0971496105194092,
      "learning_rate": 1.303e-05,
      "loss": 0.3065,
      "step": 2608
    },
    {
      "epoch": 0.14609698734460747,
      "grad_norm": 1.3269139528274536,
      "learning_rate": 1.3035e-05,
      "loss": 0.4228,
      "step": 2609
    },
    {
      "epoch": 0.14615298465673648,
      "grad_norm": 1.054341435432434,
      "learning_rate": 1.3039999999999999e-05,
      "loss": 0.3975,
      "step": 2610
    },
    {
      "epoch": 0.1462089819688655,
      "grad_norm": 1.7736667394638062,
      "learning_rate": 1.3045000000000001e-05,
      "loss": 0.4209,
      "step": 2611
    },
    {
      "epoch": 0.14626497928099452,
      "grad_norm": 1.1408703327178955,
      "learning_rate": 1.305e-05,
      "loss": 0.3248,
      "step": 2612
    },
    {
      "epoch": 0.14632097659312354,
      "grad_norm": 1.2324352264404297,
      "learning_rate": 1.3055000000000001e-05,
      "loss": 0.6046,
      "step": 2613
    },
    {
      "epoch": 0.14637697390525256,
      "grad_norm": 1.1829913854599,
      "learning_rate": 1.306e-05,
      "loss": 0.364,
      "step": 2614
    },
    {
      "epoch": 0.14643297121738155,
      "grad_norm": 1.1477619409561157,
      "learning_rate": 1.3065e-05,
      "loss": 0.3573,
      "step": 2615
    },
    {
      "epoch": 0.14648896852951057,
      "grad_norm": 1.2593798637390137,
      "learning_rate": 1.3070000000000002e-05,
      "loss": 0.3189,
      "step": 2616
    },
    {
      "epoch": 0.1465449658416396,
      "grad_norm": 1.5565099716186523,
      "learning_rate": 1.3075000000000001e-05,
      "loss": 0.4649,
      "step": 2617
    },
    {
      "epoch": 0.1466009631537686,
      "grad_norm": 1.105677604675293,
      "learning_rate": 1.308e-05,
      "loss": 0.4293,
      "step": 2618
    },
    {
      "epoch": 0.14665696046589763,
      "grad_norm": 1.015621304512024,
      "learning_rate": 1.3085e-05,
      "loss": 0.3262,
      "step": 2619
    },
    {
      "epoch": 0.14671295777802665,
      "grad_norm": 1.252323031425476,
      "learning_rate": 1.309e-05,
      "loss": 0.4461,
      "step": 2620
    },
    {
      "epoch": 0.14676895509015567,
      "grad_norm": 1.135851502418518,
      "learning_rate": 1.3095000000000003e-05,
      "loss": 0.4974,
      "step": 2621
    },
    {
      "epoch": 0.1468249524022847,
      "grad_norm": 1.3516892194747925,
      "learning_rate": 1.3100000000000002e-05,
      "loss": 0.5207,
      "step": 2622
    },
    {
      "epoch": 0.1468809497144137,
      "grad_norm": 1.4236551523208618,
      "learning_rate": 1.3105e-05,
      "loss": 0.448,
      "step": 2623
    },
    {
      "epoch": 0.14693694702654272,
      "grad_norm": 1.2360954284667969,
      "learning_rate": 1.311e-05,
      "loss": 0.4699,
      "step": 2624
    },
    {
      "epoch": 0.14699294433867174,
      "grad_norm": 1.2243598699569702,
      "learning_rate": 1.3114999999999999e-05,
      "loss": 0.4643,
      "step": 2625
    },
    {
      "epoch": 0.14704894165080076,
      "grad_norm": 1.3075956106185913,
      "learning_rate": 1.3120000000000001e-05,
      "loss": 0.6034,
      "step": 2626
    },
    {
      "epoch": 0.14710493896292978,
      "grad_norm": 1.52328360080719,
      "learning_rate": 1.3125e-05,
      "loss": 0.4696,
      "step": 2627
    },
    {
      "epoch": 0.1471609362750588,
      "grad_norm": 1.355431318283081,
      "learning_rate": 1.3130000000000001e-05,
      "loss": 0.5116,
      "step": 2628
    },
    {
      "epoch": 0.14721693358718782,
      "grad_norm": 1.3428106307983398,
      "learning_rate": 1.3135e-05,
      "loss": 0.3957,
      "step": 2629
    },
    {
      "epoch": 0.14727293089931684,
      "grad_norm": 1.2088851928710938,
      "learning_rate": 1.314e-05,
      "loss": 0.5184,
      "step": 2630
    },
    {
      "epoch": 0.14732892821144586,
      "grad_norm": 1.3796359300613403,
      "learning_rate": 1.3145000000000002e-05,
      "loss": 0.3684,
      "step": 2631
    },
    {
      "epoch": 0.14738492552357488,
      "grad_norm": 1.294643759727478,
      "learning_rate": 1.3150000000000001e-05,
      "loss": 0.4408,
      "step": 2632
    },
    {
      "epoch": 0.1474409228357039,
      "grad_norm": 1.2840946912765503,
      "learning_rate": 1.3155e-05,
      "loss": 0.4452,
      "step": 2633
    },
    {
      "epoch": 0.14749692014783292,
      "grad_norm": 1.0767340660095215,
      "learning_rate": 1.316e-05,
      "loss": 0.3368,
      "step": 2634
    },
    {
      "epoch": 0.14755291745996194,
      "grad_norm": 1.0889029502868652,
      "learning_rate": 1.3165e-05,
      "loss": 0.3129,
      "step": 2635
    },
    {
      "epoch": 0.14760891477209093,
      "grad_norm": 1.5532374382019043,
      "learning_rate": 1.3170000000000001e-05,
      "loss": 0.4134,
      "step": 2636
    },
    {
      "epoch": 0.14766491208421995,
      "grad_norm": 1.4018138647079468,
      "learning_rate": 1.3175000000000002e-05,
      "loss": 0.6566,
      "step": 2637
    },
    {
      "epoch": 0.14772090939634897,
      "grad_norm": 1.1475685834884644,
      "learning_rate": 1.3180000000000001e-05,
      "loss": 0.4866,
      "step": 2638
    },
    {
      "epoch": 0.14777690670847798,
      "grad_norm": 1.2613779306411743,
      "learning_rate": 1.3185e-05,
      "loss": 0.3669,
      "step": 2639
    },
    {
      "epoch": 0.147832904020607,
      "grad_norm": 1.3189911842346191,
      "learning_rate": 1.3189999999999999e-05,
      "loss": 0.4983,
      "step": 2640
    },
    {
      "epoch": 0.14788890133273602,
      "grad_norm": 1.0426316261291504,
      "learning_rate": 1.3195000000000002e-05,
      "loss": 0.3501,
      "step": 2641
    },
    {
      "epoch": 0.14794489864486504,
      "grad_norm": 1.1406067609786987,
      "learning_rate": 1.32e-05,
      "loss": 0.3735,
      "step": 2642
    },
    {
      "epoch": 0.14800089595699406,
      "grad_norm": 1.1951979398727417,
      "learning_rate": 1.3205000000000001e-05,
      "loss": 0.4212,
      "step": 2643
    },
    {
      "epoch": 0.14805689326912308,
      "grad_norm": 1.0712980031967163,
      "learning_rate": 1.321e-05,
      "loss": 0.3888,
      "step": 2644
    },
    {
      "epoch": 0.1481128905812521,
      "grad_norm": 1.1973940134048462,
      "learning_rate": 1.3215e-05,
      "loss": 0.4139,
      "step": 2645
    },
    {
      "epoch": 0.14816888789338112,
      "grad_norm": 1.3590928316116333,
      "learning_rate": 1.3220000000000002e-05,
      "loss": 0.5349,
      "step": 2646
    },
    {
      "epoch": 0.14822488520551014,
      "grad_norm": 1.2918621301651,
      "learning_rate": 1.3225000000000001e-05,
      "loss": 0.442,
      "step": 2647
    },
    {
      "epoch": 0.14828088251763916,
      "grad_norm": 1.199952483177185,
      "learning_rate": 1.323e-05,
      "loss": 0.5537,
      "step": 2648
    },
    {
      "epoch": 0.14833687982976818,
      "grad_norm": 1.1175743341445923,
      "learning_rate": 1.3235e-05,
      "loss": 0.4777,
      "step": 2649
    },
    {
      "epoch": 0.1483928771418972,
      "grad_norm": 1.4558801651000977,
      "learning_rate": 1.324e-05,
      "loss": 0.6484,
      "step": 2650
    },
    {
      "epoch": 0.14844887445402621,
      "grad_norm": 1.170845866203308,
      "learning_rate": 1.3245000000000001e-05,
      "loss": 0.4357,
      "step": 2651
    },
    {
      "epoch": 0.14850487176615523,
      "grad_norm": 1.2923656702041626,
      "learning_rate": 1.3250000000000002e-05,
      "loss": 0.5244,
      "step": 2652
    },
    {
      "epoch": 0.14856086907828425,
      "grad_norm": 1.153900146484375,
      "learning_rate": 1.3255000000000001e-05,
      "loss": 0.4252,
      "step": 2653
    },
    {
      "epoch": 0.14861686639041327,
      "grad_norm": 1.2607039213180542,
      "learning_rate": 1.326e-05,
      "loss": 0.4234,
      "step": 2654
    },
    {
      "epoch": 0.1486728637025423,
      "grad_norm": 1.284332513809204,
      "learning_rate": 1.3265e-05,
      "loss": 0.4065,
      "step": 2655
    },
    {
      "epoch": 0.14872886101467128,
      "grad_norm": 1.0736377239227295,
      "learning_rate": 1.3270000000000002e-05,
      "loss": 0.4597,
      "step": 2656
    },
    {
      "epoch": 0.1487848583268003,
      "grad_norm": 1.2907679080963135,
      "learning_rate": 1.3275e-05,
      "loss": 0.4447,
      "step": 2657
    },
    {
      "epoch": 0.14884085563892932,
      "grad_norm": 1.1318539381027222,
      "learning_rate": 1.3280000000000002e-05,
      "loss": 0.3366,
      "step": 2658
    },
    {
      "epoch": 0.14889685295105834,
      "grad_norm": 1.27273690700531,
      "learning_rate": 1.3285e-05,
      "loss": 0.5273,
      "step": 2659
    },
    {
      "epoch": 0.14895285026318736,
      "grad_norm": 1.3122705221176147,
      "learning_rate": 1.329e-05,
      "loss": 0.5537,
      "step": 2660
    },
    {
      "epoch": 0.14900884757531638,
      "grad_norm": 1.3192979097366333,
      "learning_rate": 1.3295000000000002e-05,
      "loss": 0.5078,
      "step": 2661
    },
    {
      "epoch": 0.1490648448874454,
      "grad_norm": 1.4099100828170776,
      "learning_rate": 1.3300000000000001e-05,
      "loss": 0.4531,
      "step": 2662
    },
    {
      "epoch": 0.14912084219957442,
      "grad_norm": 6.571522235870361,
      "learning_rate": 1.3305e-05,
      "loss": 0.4682,
      "step": 2663
    },
    {
      "epoch": 0.14917683951170344,
      "grad_norm": 1.0490186214447021,
      "learning_rate": 1.331e-05,
      "loss": 0.34,
      "step": 2664
    },
    {
      "epoch": 0.14923283682383245,
      "grad_norm": 1.4035292863845825,
      "learning_rate": 1.3315e-05,
      "loss": 0.4918,
      "step": 2665
    },
    {
      "epoch": 0.14928883413596147,
      "grad_norm": 1.2334212064743042,
      "learning_rate": 1.3320000000000001e-05,
      "loss": 0.543,
      "step": 2666
    },
    {
      "epoch": 0.1493448314480905,
      "grad_norm": 1.3419967889785767,
      "learning_rate": 1.3325000000000002e-05,
      "loss": 0.3962,
      "step": 2667
    },
    {
      "epoch": 0.1494008287602195,
      "grad_norm": 1.2629648447036743,
      "learning_rate": 1.3330000000000001e-05,
      "loss": 0.5449,
      "step": 2668
    },
    {
      "epoch": 0.14945682607234853,
      "grad_norm": 1.0612081289291382,
      "learning_rate": 1.3335e-05,
      "loss": 0.3091,
      "step": 2669
    },
    {
      "epoch": 0.14951282338447755,
      "grad_norm": 1.045214056968689,
      "learning_rate": 1.334e-05,
      "loss": 0.3962,
      "step": 2670
    },
    {
      "epoch": 0.14956882069660657,
      "grad_norm": 1.0915015935897827,
      "learning_rate": 1.3345000000000002e-05,
      "loss": 0.4691,
      "step": 2671
    },
    {
      "epoch": 0.1496248180087356,
      "grad_norm": 1.229937195777893,
      "learning_rate": 1.3350000000000001e-05,
      "loss": 0.3836,
      "step": 2672
    },
    {
      "epoch": 0.1496808153208646,
      "grad_norm": 1.303233027458191,
      "learning_rate": 1.3355e-05,
      "loss": 0.4157,
      "step": 2673
    },
    {
      "epoch": 0.14973681263299363,
      "grad_norm": 1.1198750734329224,
      "learning_rate": 1.336e-05,
      "loss": 0.4337,
      "step": 2674
    },
    {
      "epoch": 0.14979280994512265,
      "grad_norm": 1.3870460987091064,
      "learning_rate": 1.3365e-05,
      "loss": 0.5762,
      "step": 2675
    },
    {
      "epoch": 0.14984880725725166,
      "grad_norm": 1.2103842496871948,
      "learning_rate": 1.3370000000000002e-05,
      "loss": 0.4496,
      "step": 2676
    },
    {
      "epoch": 0.14990480456938066,
      "grad_norm": 0.9885027408599854,
      "learning_rate": 1.3375000000000002e-05,
      "loss": 0.3836,
      "step": 2677
    },
    {
      "epoch": 0.14996080188150968,
      "grad_norm": 1.1948119401931763,
      "learning_rate": 1.338e-05,
      "loss": 0.441,
      "step": 2678
    },
    {
      "epoch": 0.1500167991936387,
      "grad_norm": 1.1679539680480957,
      "learning_rate": 1.3385e-05,
      "loss": 0.4136,
      "step": 2679
    },
    {
      "epoch": 0.1500727965057677,
      "grad_norm": 1.3134140968322754,
      "learning_rate": 1.339e-05,
      "loss": 0.7125,
      "step": 2680
    },
    {
      "epoch": 0.15012879381789673,
      "grad_norm": 1.2312588691711426,
      "learning_rate": 1.3395000000000001e-05,
      "loss": 0.3682,
      "step": 2681
    },
    {
      "epoch": 0.15018479113002575,
      "grad_norm": 1.028341293334961,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 0.2944,
      "step": 2682
    },
    {
      "epoch": 0.15024078844215477,
      "grad_norm": 1.3490902185440063,
      "learning_rate": 1.3405000000000001e-05,
      "loss": 0.4722,
      "step": 2683
    },
    {
      "epoch": 0.1502967857542838,
      "grad_norm": 1.3073066473007202,
      "learning_rate": 1.341e-05,
      "loss": 0.4111,
      "step": 2684
    },
    {
      "epoch": 0.1503527830664128,
      "grad_norm": 1.1631131172180176,
      "learning_rate": 1.3415e-05,
      "loss": 0.4215,
      "step": 2685
    },
    {
      "epoch": 0.15040878037854183,
      "grad_norm": 1.4284385442733765,
      "learning_rate": 1.3420000000000002e-05,
      "loss": 0.5716,
      "step": 2686
    },
    {
      "epoch": 0.15046477769067085,
      "grad_norm": 1.2655913829803467,
      "learning_rate": 1.3425000000000001e-05,
      "loss": 0.353,
      "step": 2687
    },
    {
      "epoch": 0.15052077500279987,
      "grad_norm": 1.0791429281234741,
      "learning_rate": 1.343e-05,
      "loss": 0.4433,
      "step": 2688
    },
    {
      "epoch": 0.1505767723149289,
      "grad_norm": 1.2483781576156616,
      "learning_rate": 1.3435000000000001e-05,
      "loss": 0.4055,
      "step": 2689
    },
    {
      "epoch": 0.1506327696270579,
      "grad_norm": 1.4182034730911255,
      "learning_rate": 1.344e-05,
      "loss": 0.4274,
      "step": 2690
    },
    {
      "epoch": 0.15068876693918692,
      "grad_norm": 1.1914207935333252,
      "learning_rate": 1.3445e-05,
      "loss": 0.3799,
      "step": 2691
    },
    {
      "epoch": 0.15074476425131594,
      "grad_norm": 1.1741597652435303,
      "learning_rate": 1.3450000000000002e-05,
      "loss": 0.4406,
      "step": 2692
    },
    {
      "epoch": 0.15080076156344496,
      "grad_norm": 1.1351898908615112,
      "learning_rate": 1.3455e-05,
      "loss": 0.3233,
      "step": 2693
    },
    {
      "epoch": 0.15085675887557398,
      "grad_norm": 1.2189592123031616,
      "learning_rate": 1.346e-05,
      "loss": 0.4002,
      "step": 2694
    },
    {
      "epoch": 0.150912756187703,
      "grad_norm": 1.1392837762832642,
      "learning_rate": 1.3465e-05,
      "loss": 0.3374,
      "step": 2695
    },
    {
      "epoch": 0.15096875349983202,
      "grad_norm": 1.401028037071228,
      "learning_rate": 1.347e-05,
      "loss": 0.4197,
      "step": 2696
    },
    {
      "epoch": 0.15102475081196104,
      "grad_norm": 1.1850882768630981,
      "learning_rate": 1.3475000000000002e-05,
      "loss": 0.4155,
      "step": 2697
    },
    {
      "epoch": 0.15108074812409003,
      "grad_norm": 1.4200987815856934,
      "learning_rate": 1.3480000000000001e-05,
      "loss": 0.381,
      "step": 2698
    },
    {
      "epoch": 0.15113674543621905,
      "grad_norm": 1.0592139959335327,
      "learning_rate": 1.3485e-05,
      "loss": 0.3618,
      "step": 2699
    },
    {
      "epoch": 0.15119274274834807,
      "grad_norm": 1.2360087633132935,
      "learning_rate": 1.349e-05,
      "loss": 0.3661,
      "step": 2700
    },
    {
      "epoch": 0.1512487400604771,
      "grad_norm": 1.242908239364624,
      "learning_rate": 1.3494999999999999e-05,
      "loss": 0.514,
      "step": 2701
    },
    {
      "epoch": 0.1513047373726061,
      "grad_norm": 1.3020732402801514,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.4683,
      "step": 2702
    },
    {
      "epoch": 0.15136073468473513,
      "grad_norm": 1.5411075353622437,
      "learning_rate": 1.3505e-05,
      "loss": 0.421,
      "step": 2703
    },
    {
      "epoch": 0.15141673199686415,
      "grad_norm": 1.2832776308059692,
      "learning_rate": 1.3510000000000001e-05,
      "loss": 0.4048,
      "step": 2704
    },
    {
      "epoch": 0.15147272930899316,
      "grad_norm": 1.2762134075164795,
      "learning_rate": 1.3515e-05,
      "loss": 0.5117,
      "step": 2705
    },
    {
      "epoch": 0.15152872662112218,
      "grad_norm": 1.3318418264389038,
      "learning_rate": 1.352e-05,
      "loss": 0.4159,
      "step": 2706
    },
    {
      "epoch": 0.1515847239332512,
      "grad_norm": 1.4038443565368652,
      "learning_rate": 1.3525000000000002e-05,
      "loss": 0.5075,
      "step": 2707
    },
    {
      "epoch": 0.15164072124538022,
      "grad_norm": 1.3062500953674316,
      "learning_rate": 1.3530000000000001e-05,
      "loss": 0.6217,
      "step": 2708
    },
    {
      "epoch": 0.15169671855750924,
      "grad_norm": 1.1369431018829346,
      "learning_rate": 1.3535e-05,
      "loss": 0.4208,
      "step": 2709
    },
    {
      "epoch": 0.15175271586963826,
      "grad_norm": 1.464113473892212,
      "learning_rate": 1.3539999999999999e-05,
      "loss": 0.4137,
      "step": 2710
    },
    {
      "epoch": 0.15180871318176728,
      "grad_norm": 1.2051522731781006,
      "learning_rate": 1.3545e-05,
      "loss": 0.5511,
      "step": 2711
    },
    {
      "epoch": 0.1518647104938963,
      "grad_norm": 1.232174038887024,
      "learning_rate": 1.3550000000000002e-05,
      "loss": 0.4952,
      "step": 2712
    },
    {
      "epoch": 0.15192070780602532,
      "grad_norm": 1.2316820621490479,
      "learning_rate": 1.3555000000000002e-05,
      "loss": 0.4071,
      "step": 2713
    },
    {
      "epoch": 0.15197670511815434,
      "grad_norm": 1.3428975343704224,
      "learning_rate": 1.356e-05,
      "loss": 0.486,
      "step": 2714
    },
    {
      "epoch": 0.15203270243028336,
      "grad_norm": 1.1460320949554443,
      "learning_rate": 1.3565e-05,
      "loss": 0.4263,
      "step": 2715
    },
    {
      "epoch": 0.15208869974241238,
      "grad_norm": 1.3443206548690796,
      "learning_rate": 1.3569999999999999e-05,
      "loss": 0.4956,
      "step": 2716
    },
    {
      "epoch": 0.1521446970545414,
      "grad_norm": 1.0955899953842163,
      "learning_rate": 1.3575000000000001e-05,
      "loss": 0.3872,
      "step": 2717
    },
    {
      "epoch": 0.15220069436667039,
      "grad_norm": 1.1686948537826538,
      "learning_rate": 1.358e-05,
      "loss": 0.487,
      "step": 2718
    },
    {
      "epoch": 0.1522566916787994,
      "grad_norm": 1.1902086734771729,
      "learning_rate": 1.3585000000000001e-05,
      "loss": 0.4301,
      "step": 2719
    },
    {
      "epoch": 0.15231268899092842,
      "grad_norm": 1.2101625204086304,
      "learning_rate": 1.359e-05,
      "loss": 0.4894,
      "step": 2720
    },
    {
      "epoch": 0.15236868630305744,
      "grad_norm": 1.1747902631759644,
      "learning_rate": 1.3595e-05,
      "loss": 0.3868,
      "step": 2721
    },
    {
      "epoch": 0.15242468361518646,
      "grad_norm": 1.752328634262085,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.5366,
      "step": 2722
    },
    {
      "epoch": 0.15248068092731548,
      "grad_norm": 1.3041211366653442,
      "learning_rate": 1.3605000000000001e-05,
      "loss": 0.4857,
      "step": 2723
    },
    {
      "epoch": 0.1525366782394445,
      "grad_norm": 1.0765955448150635,
      "learning_rate": 1.361e-05,
      "loss": 0.345,
      "step": 2724
    },
    {
      "epoch": 0.15259267555157352,
      "grad_norm": 1.4420080184936523,
      "learning_rate": 1.3615e-05,
      "loss": 0.5749,
      "step": 2725
    },
    {
      "epoch": 0.15264867286370254,
      "grad_norm": 1.1885454654693604,
      "learning_rate": 1.362e-05,
      "loss": 0.5055,
      "step": 2726
    },
    {
      "epoch": 0.15270467017583156,
      "grad_norm": 1.2062809467315674,
      "learning_rate": 1.3625e-05,
      "loss": 0.4599,
      "step": 2727
    },
    {
      "epoch": 0.15276066748796058,
      "grad_norm": 1.3495934009552002,
      "learning_rate": 1.3630000000000002e-05,
      "loss": 0.4824,
      "step": 2728
    },
    {
      "epoch": 0.1528166648000896,
      "grad_norm": 1.6765187978744507,
      "learning_rate": 1.3635e-05,
      "loss": 0.5063,
      "step": 2729
    },
    {
      "epoch": 0.15287266211221862,
      "grad_norm": 1.3617475032806396,
      "learning_rate": 1.364e-05,
      "loss": 0.475,
      "step": 2730
    },
    {
      "epoch": 0.15292865942434763,
      "grad_norm": 1.0638370513916016,
      "learning_rate": 1.3644999999999999e-05,
      "loss": 0.3662,
      "step": 2731
    },
    {
      "epoch": 0.15298465673647665,
      "grad_norm": 1.3291374444961548,
      "learning_rate": 1.3650000000000001e-05,
      "loss": 0.4763,
      "step": 2732
    },
    {
      "epoch": 0.15304065404860567,
      "grad_norm": 0.9607083797454834,
      "learning_rate": 1.3655e-05,
      "loss": 0.3983,
      "step": 2733
    },
    {
      "epoch": 0.1530966513607347,
      "grad_norm": 1.1476843357086182,
      "learning_rate": 1.3660000000000001e-05,
      "loss": 0.5913,
      "step": 2734
    },
    {
      "epoch": 0.1531526486728637,
      "grad_norm": 1.1318851709365845,
      "learning_rate": 1.3665e-05,
      "loss": 0.4258,
      "step": 2735
    },
    {
      "epoch": 0.15320864598499273,
      "grad_norm": 1.1070514917373657,
      "learning_rate": 1.367e-05,
      "loss": 0.4925,
      "step": 2736
    },
    {
      "epoch": 0.15326464329712175,
      "grad_norm": 1.3675343990325928,
      "learning_rate": 1.3675000000000002e-05,
      "loss": 0.5905,
      "step": 2737
    },
    {
      "epoch": 0.15332064060925077,
      "grad_norm": 1.3040297031402588,
      "learning_rate": 1.3680000000000001e-05,
      "loss": 0.4395,
      "step": 2738
    },
    {
      "epoch": 0.15337663792137976,
      "grad_norm": 1.1907511949539185,
      "learning_rate": 1.3685e-05,
      "loss": 0.3728,
      "step": 2739
    },
    {
      "epoch": 0.15343263523350878,
      "grad_norm": 1.1401329040527344,
      "learning_rate": 1.369e-05,
      "loss": 0.446,
      "step": 2740
    },
    {
      "epoch": 0.1534886325456378,
      "grad_norm": 1.0914613008499146,
      "learning_rate": 1.3695e-05,
      "loss": 0.3913,
      "step": 2741
    },
    {
      "epoch": 0.15354462985776682,
      "grad_norm": 1.1461143493652344,
      "learning_rate": 1.3700000000000001e-05,
      "loss": 0.4044,
      "step": 2742
    },
    {
      "epoch": 0.15360062716989584,
      "grad_norm": 1.229974627494812,
      "learning_rate": 1.3705000000000002e-05,
      "loss": 0.5203,
      "step": 2743
    },
    {
      "epoch": 0.15365662448202486,
      "grad_norm": 1.1887822151184082,
      "learning_rate": 1.3710000000000001e-05,
      "loss": 0.3773,
      "step": 2744
    },
    {
      "epoch": 0.15371262179415388,
      "grad_norm": 1.0333051681518555,
      "learning_rate": 1.3715e-05,
      "loss": 0.4415,
      "step": 2745
    },
    {
      "epoch": 0.1537686191062829,
      "grad_norm": 1.1656396389007568,
      "learning_rate": 1.3719999999999999e-05,
      "loss": 0.4399,
      "step": 2746
    },
    {
      "epoch": 0.1538246164184119,
      "grad_norm": 1.2455742359161377,
      "learning_rate": 1.3725000000000002e-05,
      "loss": 0.4364,
      "step": 2747
    },
    {
      "epoch": 0.15388061373054093,
      "grad_norm": 1.0791863203048706,
      "learning_rate": 1.373e-05,
      "loss": 0.3751,
      "step": 2748
    },
    {
      "epoch": 0.15393661104266995,
      "grad_norm": 1.1460646390914917,
      "learning_rate": 1.3735000000000001e-05,
      "loss": 0.4815,
      "step": 2749
    },
    {
      "epoch": 0.15399260835479897,
      "grad_norm": 1.4752322435379028,
      "learning_rate": 1.374e-05,
      "loss": 0.6131,
      "step": 2750
    },
    {
      "epoch": 0.154048605666928,
      "grad_norm": 1.1885502338409424,
      "learning_rate": 1.3745e-05,
      "loss": 0.5409,
      "step": 2751
    },
    {
      "epoch": 0.154104602979057,
      "grad_norm": 1.6608283519744873,
      "learning_rate": 1.3750000000000002e-05,
      "loss": 0.4526,
      "step": 2752
    },
    {
      "epoch": 0.15416060029118603,
      "grad_norm": 1.4488664865493774,
      "learning_rate": 1.3755000000000001e-05,
      "loss": 0.3633,
      "step": 2753
    },
    {
      "epoch": 0.15421659760331505,
      "grad_norm": 1.2881131172180176,
      "learning_rate": 1.376e-05,
      "loss": 0.5018,
      "step": 2754
    },
    {
      "epoch": 0.15427259491544407,
      "grad_norm": 1.3838692903518677,
      "learning_rate": 1.3765e-05,
      "loss": 0.5169,
      "step": 2755
    },
    {
      "epoch": 0.15432859222757309,
      "grad_norm": 1.2695553302764893,
      "learning_rate": 1.377e-05,
      "loss": 0.5031,
      "step": 2756
    },
    {
      "epoch": 0.1543845895397021,
      "grad_norm": 1.3294891119003296,
      "learning_rate": 1.3775000000000001e-05,
      "loss": 0.5022,
      "step": 2757
    },
    {
      "epoch": 0.15444058685183112,
      "grad_norm": 1.100566029548645,
      "learning_rate": 1.3780000000000002e-05,
      "loss": 0.4378,
      "step": 2758
    },
    {
      "epoch": 0.15449658416396014,
      "grad_norm": 1.1493943929672241,
      "learning_rate": 1.3785000000000001e-05,
      "loss": 0.5951,
      "step": 2759
    },
    {
      "epoch": 0.15455258147608913,
      "grad_norm": 1.3388559818267822,
      "learning_rate": 1.379e-05,
      "loss": 0.4385,
      "step": 2760
    },
    {
      "epoch": 0.15460857878821815,
      "grad_norm": 1.2471098899841309,
      "learning_rate": 1.3795e-05,
      "loss": 0.3845,
      "step": 2761
    },
    {
      "epoch": 0.15466457610034717,
      "grad_norm": 1.0742604732513428,
      "learning_rate": 1.3800000000000002e-05,
      "loss": 0.343,
      "step": 2762
    },
    {
      "epoch": 0.1547205734124762,
      "grad_norm": 1.1153868436813354,
      "learning_rate": 1.3805e-05,
      "loss": 0.3902,
      "step": 2763
    },
    {
      "epoch": 0.1547765707246052,
      "grad_norm": 1.2439930438995361,
      "learning_rate": 1.381e-05,
      "loss": 0.4284,
      "step": 2764
    },
    {
      "epoch": 0.15483256803673423,
      "grad_norm": 2.0175135135650635,
      "learning_rate": 1.3815e-05,
      "loss": 0.3932,
      "step": 2765
    },
    {
      "epoch": 0.15488856534886325,
      "grad_norm": 0.9710941910743713,
      "learning_rate": 1.382e-05,
      "loss": 0.2777,
      "step": 2766
    },
    {
      "epoch": 0.15494456266099227,
      "grad_norm": 1.301509976387024,
      "learning_rate": 1.3825000000000002e-05,
      "loss": 0.418,
      "step": 2767
    },
    {
      "epoch": 0.1550005599731213,
      "grad_norm": 1.266956090927124,
      "learning_rate": 1.3830000000000001e-05,
      "loss": 0.4642,
      "step": 2768
    },
    {
      "epoch": 0.1550565572852503,
      "grad_norm": 1.3388291597366333,
      "learning_rate": 1.3835e-05,
      "loss": 0.5071,
      "step": 2769
    },
    {
      "epoch": 0.15511255459737933,
      "grad_norm": 1.240614414215088,
      "learning_rate": 1.384e-05,
      "loss": 0.4912,
      "step": 2770
    },
    {
      "epoch": 0.15516855190950835,
      "grad_norm": 1.2789167165756226,
      "learning_rate": 1.3845e-05,
      "loss": 0.4411,
      "step": 2771
    },
    {
      "epoch": 0.15522454922163736,
      "grad_norm": 1.1685004234313965,
      "learning_rate": 1.3850000000000001e-05,
      "loss": 0.4822,
      "step": 2772
    },
    {
      "epoch": 0.15528054653376638,
      "grad_norm": 1.5487463474273682,
      "learning_rate": 1.3855000000000002e-05,
      "loss": 0.4153,
      "step": 2773
    },
    {
      "epoch": 0.1553365438458954,
      "grad_norm": 1.0213351249694824,
      "learning_rate": 1.3860000000000001e-05,
      "loss": 0.4142,
      "step": 2774
    },
    {
      "epoch": 0.15539254115802442,
      "grad_norm": 1.3269951343536377,
      "learning_rate": 1.3865e-05,
      "loss": 0.3837,
      "step": 2775
    },
    {
      "epoch": 0.15544853847015344,
      "grad_norm": 1.3169128894805908,
      "learning_rate": 1.387e-05,
      "loss": 0.3842,
      "step": 2776
    },
    {
      "epoch": 0.15550453578228246,
      "grad_norm": 1.3252760171890259,
      "learning_rate": 1.3875000000000002e-05,
      "loss": 0.3117,
      "step": 2777
    },
    {
      "epoch": 0.15556053309441148,
      "grad_norm": 1.3185728788375854,
      "learning_rate": 1.3880000000000001e-05,
      "loss": 0.5022,
      "step": 2778
    },
    {
      "epoch": 0.1556165304065405,
      "grad_norm": 1.1187517642974854,
      "learning_rate": 1.3885e-05,
      "loss": 0.4589,
      "step": 2779
    },
    {
      "epoch": 0.1556725277186695,
      "grad_norm": 1.306776762008667,
      "learning_rate": 1.389e-05,
      "loss": 0.6016,
      "step": 2780
    },
    {
      "epoch": 0.1557285250307985,
      "grad_norm": 1.2258596420288086,
      "learning_rate": 1.3895e-05,
      "loss": 0.506,
      "step": 2781
    },
    {
      "epoch": 0.15578452234292753,
      "grad_norm": 1.2695173025131226,
      "learning_rate": 1.3900000000000002e-05,
      "loss": 0.5451,
      "step": 2782
    },
    {
      "epoch": 0.15584051965505655,
      "grad_norm": 1.2983871698379517,
      "learning_rate": 1.3905000000000002e-05,
      "loss": 0.4632,
      "step": 2783
    },
    {
      "epoch": 0.15589651696718557,
      "grad_norm": 1.0823630094528198,
      "learning_rate": 1.391e-05,
      "loss": 0.3751,
      "step": 2784
    },
    {
      "epoch": 0.15595251427931459,
      "grad_norm": 1.118971586227417,
      "learning_rate": 1.3915e-05,
      "loss": 0.4565,
      "step": 2785
    },
    {
      "epoch": 0.1560085115914436,
      "grad_norm": 1.221749186515808,
      "learning_rate": 1.3919999999999999e-05,
      "loss": 0.415,
      "step": 2786
    },
    {
      "epoch": 0.15606450890357262,
      "grad_norm": 1.352028727531433,
      "learning_rate": 1.3925000000000001e-05,
      "loss": 0.4915,
      "step": 2787
    },
    {
      "epoch": 0.15612050621570164,
      "grad_norm": 1.284001111984253,
      "learning_rate": 1.3930000000000002e-05,
      "loss": 0.4595,
      "step": 2788
    },
    {
      "epoch": 0.15617650352783066,
      "grad_norm": 1.2041031122207642,
      "learning_rate": 1.3935000000000001e-05,
      "loss": 0.4143,
      "step": 2789
    },
    {
      "epoch": 0.15623250083995968,
      "grad_norm": 1.3998245000839233,
      "learning_rate": 1.394e-05,
      "loss": 0.5409,
      "step": 2790
    },
    {
      "epoch": 0.1562884981520887,
      "grad_norm": 1.4810839891433716,
      "learning_rate": 1.3945e-05,
      "loss": 0.4637,
      "step": 2791
    },
    {
      "epoch": 0.15634449546421772,
      "grad_norm": 1.242527961730957,
      "learning_rate": 1.3950000000000002e-05,
      "loss": 0.3989,
      "step": 2792
    },
    {
      "epoch": 0.15640049277634674,
      "grad_norm": 1.5063345432281494,
      "learning_rate": 1.3955000000000001e-05,
      "loss": 0.4764,
      "step": 2793
    },
    {
      "epoch": 0.15645649008847576,
      "grad_norm": 1.2918951511383057,
      "learning_rate": 1.396e-05,
      "loss": 0.469,
      "step": 2794
    },
    {
      "epoch": 0.15651248740060478,
      "grad_norm": 1.5124133825302124,
      "learning_rate": 1.3965000000000001e-05,
      "loss": 0.6527,
      "step": 2795
    },
    {
      "epoch": 0.1565684847127338,
      "grad_norm": 1.2914239168167114,
      "learning_rate": 1.397e-05,
      "loss": 0.4341,
      "step": 2796
    },
    {
      "epoch": 0.15662448202486282,
      "grad_norm": 1.5263826847076416,
      "learning_rate": 1.3975000000000003e-05,
      "loss": 0.5055,
      "step": 2797
    },
    {
      "epoch": 0.15668047933699183,
      "grad_norm": 1.289559245109558,
      "learning_rate": 1.3980000000000002e-05,
      "loss": 0.4069,
      "step": 2798
    },
    {
      "epoch": 0.15673647664912085,
      "grad_norm": 1.3386255502700806,
      "learning_rate": 1.3985e-05,
      "loss": 0.5132,
      "step": 2799
    },
    {
      "epoch": 0.15679247396124987,
      "grad_norm": 1.1883652210235596,
      "learning_rate": 1.399e-05,
      "loss": 0.5696,
      "step": 2800
    },
    {
      "epoch": 0.15684847127337886,
      "grad_norm": 1.2485971450805664,
      "learning_rate": 1.3994999999999999e-05,
      "loss": 0.3799,
      "step": 2801
    },
    {
      "epoch": 0.15690446858550788,
      "grad_norm": 1.091121792793274,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 0.3849,
      "step": 2802
    },
    {
      "epoch": 0.1569604658976369,
      "grad_norm": 1.335418462753296,
      "learning_rate": 1.4005000000000002e-05,
      "loss": 0.5873,
      "step": 2803
    },
    {
      "epoch": 0.15701646320976592,
      "grad_norm": 1.2520933151245117,
      "learning_rate": 1.4010000000000001e-05,
      "loss": 0.388,
      "step": 2804
    },
    {
      "epoch": 0.15707246052189494,
      "grad_norm": 1.3190689086914062,
      "learning_rate": 1.4015e-05,
      "loss": 0.5136,
      "step": 2805
    },
    {
      "epoch": 0.15712845783402396,
      "grad_norm": 1.1924923658370972,
      "learning_rate": 1.402e-05,
      "loss": 0.5447,
      "step": 2806
    },
    {
      "epoch": 0.15718445514615298,
      "grad_norm": 1.170854091644287,
      "learning_rate": 1.4025000000000002e-05,
      "loss": 0.5238,
      "step": 2807
    },
    {
      "epoch": 0.157240452458282,
      "grad_norm": 1.2387036085128784,
      "learning_rate": 1.4030000000000001e-05,
      "loss": 0.5482,
      "step": 2808
    },
    {
      "epoch": 0.15729644977041102,
      "grad_norm": 1.2129141092300415,
      "learning_rate": 1.4035e-05,
      "loss": 0.4282,
      "step": 2809
    },
    {
      "epoch": 0.15735244708254004,
      "grad_norm": 1.2074551582336426,
      "learning_rate": 1.4040000000000001e-05,
      "loss": 0.4376,
      "step": 2810
    },
    {
      "epoch": 0.15740844439466906,
      "grad_norm": 1.2769153118133545,
      "learning_rate": 1.4045e-05,
      "loss": 0.5132,
      "step": 2811
    },
    {
      "epoch": 0.15746444170679808,
      "grad_norm": 1.108242154121399,
      "learning_rate": 1.4050000000000003e-05,
      "loss": 0.4453,
      "step": 2812
    },
    {
      "epoch": 0.1575204390189271,
      "grad_norm": 1.230486512184143,
      "learning_rate": 1.4055000000000002e-05,
      "loss": 0.3676,
      "step": 2813
    },
    {
      "epoch": 0.1575764363310561,
      "grad_norm": 1.2049204111099243,
      "learning_rate": 1.4060000000000001e-05,
      "loss": 0.4022,
      "step": 2814
    },
    {
      "epoch": 0.15763243364318513,
      "grad_norm": 1.43659508228302,
      "learning_rate": 1.4065e-05,
      "loss": 0.3815,
      "step": 2815
    },
    {
      "epoch": 0.15768843095531415,
      "grad_norm": 1.3100559711456299,
      "learning_rate": 1.4069999999999999e-05,
      "loss": 0.4313,
      "step": 2816
    },
    {
      "epoch": 0.15774442826744317,
      "grad_norm": 1.345204472541809,
      "learning_rate": 1.4075e-05,
      "loss": 0.3449,
      "step": 2817
    },
    {
      "epoch": 0.1578004255795722,
      "grad_norm": 1.1812174320220947,
      "learning_rate": 1.408e-05,
      "loss": 0.4178,
      "step": 2818
    },
    {
      "epoch": 0.1578564228917012,
      "grad_norm": 1.0329145193099976,
      "learning_rate": 1.4085000000000002e-05,
      "loss": 0.3366,
      "step": 2819
    },
    {
      "epoch": 0.15791242020383023,
      "grad_norm": 1.1096454858779907,
      "learning_rate": 1.409e-05,
      "loss": 0.4205,
      "step": 2820
    },
    {
      "epoch": 0.15796841751595925,
      "grad_norm": 1.0906081199645996,
      "learning_rate": 1.4095e-05,
      "loss": 0.41,
      "step": 2821
    },
    {
      "epoch": 0.15802441482808824,
      "grad_norm": 1.1314079761505127,
      "learning_rate": 1.4099999999999999e-05,
      "loss": 0.4788,
      "step": 2822
    },
    {
      "epoch": 0.15808041214021726,
      "grad_norm": 1.343037486076355,
      "learning_rate": 1.4105000000000001e-05,
      "loss": 0.4959,
      "step": 2823
    },
    {
      "epoch": 0.15813640945234628,
      "grad_norm": 1.146909236907959,
      "learning_rate": 1.411e-05,
      "loss": 0.4298,
      "step": 2824
    },
    {
      "epoch": 0.1581924067644753,
      "grad_norm": 1.0961363315582275,
      "learning_rate": 1.4115000000000001e-05,
      "loss": 0.3701,
      "step": 2825
    },
    {
      "epoch": 0.15824840407660432,
      "grad_norm": 1.1320148706436157,
      "learning_rate": 1.412e-05,
      "loss": 0.365,
      "step": 2826
    },
    {
      "epoch": 0.15830440138873333,
      "grad_norm": 1.2287229299545288,
      "learning_rate": 1.4125e-05,
      "loss": 0.3599,
      "step": 2827
    },
    {
      "epoch": 0.15836039870086235,
      "grad_norm": 1.4046834707260132,
      "learning_rate": 1.4130000000000002e-05,
      "loss": 0.4867,
      "step": 2828
    },
    {
      "epoch": 0.15841639601299137,
      "grad_norm": 1.079779863357544,
      "learning_rate": 1.4135000000000001e-05,
      "loss": 0.3229,
      "step": 2829
    },
    {
      "epoch": 0.1584723933251204,
      "grad_norm": 1.482107400894165,
      "learning_rate": 1.414e-05,
      "loss": 0.5059,
      "step": 2830
    },
    {
      "epoch": 0.1585283906372494,
      "grad_norm": 1.0793694257736206,
      "learning_rate": 1.4145e-05,
      "loss": 0.3989,
      "step": 2831
    },
    {
      "epoch": 0.15858438794937843,
      "grad_norm": 1.906036376953125,
      "learning_rate": 1.415e-05,
      "loss": 0.4956,
      "step": 2832
    },
    {
      "epoch": 0.15864038526150745,
      "grad_norm": 1.4872894287109375,
      "learning_rate": 1.4155000000000001e-05,
      "loss": 0.4636,
      "step": 2833
    },
    {
      "epoch": 0.15869638257363647,
      "grad_norm": 1.1432147026062012,
      "learning_rate": 1.4160000000000002e-05,
      "loss": 0.4302,
      "step": 2834
    },
    {
      "epoch": 0.1587523798857655,
      "grad_norm": 1.4246662855148315,
      "learning_rate": 1.4165e-05,
      "loss": 0.3449,
      "step": 2835
    },
    {
      "epoch": 0.1588083771978945,
      "grad_norm": 1.111423134803772,
      "learning_rate": 1.417e-05,
      "loss": 0.3431,
      "step": 2836
    },
    {
      "epoch": 0.15886437451002353,
      "grad_norm": 1.2800912857055664,
      "learning_rate": 1.4174999999999999e-05,
      "loss": 0.5825,
      "step": 2837
    },
    {
      "epoch": 0.15892037182215255,
      "grad_norm": 1.2523735761642456,
      "learning_rate": 1.4180000000000001e-05,
      "loss": 0.5138,
      "step": 2838
    },
    {
      "epoch": 0.15897636913428156,
      "grad_norm": 1.3163225650787354,
      "learning_rate": 1.4185e-05,
      "loss": 0.4908,
      "step": 2839
    },
    {
      "epoch": 0.15903236644641058,
      "grad_norm": 1.0494143962860107,
      "learning_rate": 1.4190000000000001e-05,
      "loss": 0.3915,
      "step": 2840
    },
    {
      "epoch": 0.1590883637585396,
      "grad_norm": 1.2246289253234863,
      "learning_rate": 1.4195e-05,
      "loss": 0.4301,
      "step": 2841
    },
    {
      "epoch": 0.1591443610706686,
      "grad_norm": 1.6092647314071655,
      "learning_rate": 1.42e-05,
      "loss": 0.6129,
      "step": 2842
    },
    {
      "epoch": 0.1592003583827976,
      "grad_norm": 1.4280298948287964,
      "learning_rate": 1.4205000000000002e-05,
      "loss": 0.517,
      "step": 2843
    },
    {
      "epoch": 0.15925635569492663,
      "grad_norm": 1.171018362045288,
      "learning_rate": 1.4210000000000001e-05,
      "loss": 0.3967,
      "step": 2844
    },
    {
      "epoch": 0.15931235300705565,
      "grad_norm": 1.0142732858657837,
      "learning_rate": 1.4215e-05,
      "loss": 0.3814,
      "step": 2845
    },
    {
      "epoch": 0.15936835031918467,
      "grad_norm": 1.1765437126159668,
      "learning_rate": 1.422e-05,
      "loss": 0.4179,
      "step": 2846
    },
    {
      "epoch": 0.1594243476313137,
      "grad_norm": 1.0797767639160156,
      "learning_rate": 1.4225e-05,
      "loss": 0.3455,
      "step": 2847
    },
    {
      "epoch": 0.1594803449434427,
      "grad_norm": 1.056195855140686,
      "learning_rate": 1.4230000000000001e-05,
      "loss": 0.4499,
      "step": 2848
    },
    {
      "epoch": 0.15953634225557173,
      "grad_norm": 1.177141785621643,
      "learning_rate": 1.4235000000000002e-05,
      "loss": 0.4186,
      "step": 2849
    },
    {
      "epoch": 0.15959233956770075,
      "grad_norm": 1.1607415676116943,
      "learning_rate": 1.4240000000000001e-05,
      "loss": 0.4834,
      "step": 2850
    },
    {
      "epoch": 0.15964833687982977,
      "grad_norm": 1.601799488067627,
      "learning_rate": 1.4245e-05,
      "loss": 0.455,
      "step": 2851
    },
    {
      "epoch": 0.15970433419195879,
      "grad_norm": 1.3250458240509033,
      "learning_rate": 1.4249999999999999e-05,
      "loss": 0.4554,
      "step": 2852
    },
    {
      "epoch": 0.1597603315040878,
      "grad_norm": 1.2698512077331543,
      "learning_rate": 1.4255000000000002e-05,
      "loss": 0.5106,
      "step": 2853
    },
    {
      "epoch": 0.15981632881621682,
      "grad_norm": 1.3289108276367188,
      "learning_rate": 1.426e-05,
      "loss": 0.4777,
      "step": 2854
    },
    {
      "epoch": 0.15987232612834584,
      "grad_norm": 1.2498283386230469,
      "learning_rate": 1.4265e-05,
      "loss": 0.471,
      "step": 2855
    },
    {
      "epoch": 0.15992832344047486,
      "grad_norm": 1.1666674613952637,
      "learning_rate": 1.427e-05,
      "loss": 0.4196,
      "step": 2856
    },
    {
      "epoch": 0.15998432075260388,
      "grad_norm": 1.1459076404571533,
      "learning_rate": 1.4275e-05,
      "loss": 0.4377,
      "step": 2857
    },
    {
      "epoch": 0.1600403180647329,
      "grad_norm": 1.3304330110549927,
      "learning_rate": 1.4280000000000002e-05,
      "loss": 0.6097,
      "step": 2858
    },
    {
      "epoch": 0.16009631537686192,
      "grad_norm": 1.171502947807312,
      "learning_rate": 1.4285000000000001e-05,
      "loss": 0.3562,
      "step": 2859
    },
    {
      "epoch": 0.16015231268899094,
      "grad_norm": 1.9733786582946777,
      "learning_rate": 1.429e-05,
      "loss": 0.4004,
      "step": 2860
    },
    {
      "epoch": 0.16020831000111996,
      "grad_norm": 1.3946818113327026,
      "learning_rate": 1.4295e-05,
      "loss": 0.4732,
      "step": 2861
    },
    {
      "epoch": 0.16026430731324898,
      "grad_norm": 1.0983641147613525,
      "learning_rate": 1.43e-05,
      "loss": 0.3928,
      "step": 2862
    },
    {
      "epoch": 0.16032030462537797,
      "grad_norm": 1.106260895729065,
      "learning_rate": 1.4305000000000001e-05,
      "loss": 0.3349,
      "step": 2863
    },
    {
      "epoch": 0.160376301937507,
      "grad_norm": 1.3615952730178833,
      "learning_rate": 1.4310000000000002e-05,
      "loss": 0.4601,
      "step": 2864
    },
    {
      "epoch": 0.160432299249636,
      "grad_norm": 1.24410879611969,
      "learning_rate": 1.4315000000000001e-05,
      "loss": 0.4041,
      "step": 2865
    },
    {
      "epoch": 0.16048829656176503,
      "grad_norm": 1.5198066234588623,
      "learning_rate": 1.432e-05,
      "loss": 0.4549,
      "step": 2866
    },
    {
      "epoch": 0.16054429387389405,
      "grad_norm": 1.0579211711883545,
      "learning_rate": 1.4325e-05,
      "loss": 0.414,
      "step": 2867
    },
    {
      "epoch": 0.16060029118602306,
      "grad_norm": 1.3212554454803467,
      "learning_rate": 1.4330000000000002e-05,
      "loss": 0.6388,
      "step": 2868
    },
    {
      "epoch": 0.16065628849815208,
      "grad_norm": 1.2160145044326782,
      "learning_rate": 1.4335e-05,
      "loss": 0.4994,
      "step": 2869
    },
    {
      "epoch": 0.1607122858102811,
      "grad_norm": 1.075178623199463,
      "learning_rate": 1.434e-05,
      "loss": 0.3752,
      "step": 2870
    },
    {
      "epoch": 0.16076828312241012,
      "grad_norm": 1.1443240642547607,
      "learning_rate": 1.4345e-05,
      "loss": 0.5189,
      "step": 2871
    },
    {
      "epoch": 0.16082428043453914,
      "grad_norm": 1.1939789056777954,
      "learning_rate": 1.435e-05,
      "loss": 0.4313,
      "step": 2872
    },
    {
      "epoch": 0.16088027774666816,
      "grad_norm": 1.0715328454971313,
      "learning_rate": 1.4355000000000002e-05,
      "loss": 0.3825,
      "step": 2873
    },
    {
      "epoch": 0.16093627505879718,
      "grad_norm": 1.2756158113479614,
      "learning_rate": 1.4360000000000001e-05,
      "loss": 0.5611,
      "step": 2874
    },
    {
      "epoch": 0.1609922723709262,
      "grad_norm": 1.0364662408828735,
      "learning_rate": 1.4365e-05,
      "loss": 0.3575,
      "step": 2875
    },
    {
      "epoch": 0.16104826968305522,
      "grad_norm": 1.26275634765625,
      "learning_rate": 1.437e-05,
      "loss": 0.5086,
      "step": 2876
    },
    {
      "epoch": 0.16110426699518424,
      "grad_norm": 1.209248423576355,
      "learning_rate": 1.4374999999999999e-05,
      "loss": 0.4786,
      "step": 2877
    },
    {
      "epoch": 0.16116026430731326,
      "grad_norm": 1.212764859199524,
      "learning_rate": 1.4380000000000001e-05,
      "loss": 0.404,
      "step": 2878
    },
    {
      "epoch": 0.16121626161944227,
      "grad_norm": 1.119140625,
      "learning_rate": 1.4385000000000002e-05,
      "loss": 0.3736,
      "step": 2879
    },
    {
      "epoch": 0.1612722589315713,
      "grad_norm": 2.300510883331299,
      "learning_rate": 1.4390000000000001e-05,
      "loss": 0.4957,
      "step": 2880
    },
    {
      "epoch": 0.1613282562437003,
      "grad_norm": 1.0119082927703857,
      "learning_rate": 1.4395e-05,
      "loss": 0.4613,
      "step": 2881
    },
    {
      "epoch": 0.16138425355582933,
      "grad_norm": 1.2868393659591675,
      "learning_rate": 1.44e-05,
      "loss": 0.7153,
      "step": 2882
    },
    {
      "epoch": 0.16144025086795835,
      "grad_norm": 1.080954909324646,
      "learning_rate": 1.4405000000000002e-05,
      "loss": 0.4354,
      "step": 2883
    },
    {
      "epoch": 0.16149624818008734,
      "grad_norm": 1.0970935821533203,
      "learning_rate": 1.4410000000000001e-05,
      "loss": 0.4083,
      "step": 2884
    },
    {
      "epoch": 0.16155224549221636,
      "grad_norm": 2.6635384559631348,
      "learning_rate": 1.4415e-05,
      "loss": 0.5151,
      "step": 2885
    },
    {
      "epoch": 0.16160824280434538,
      "grad_norm": 1.5100364685058594,
      "learning_rate": 1.4420000000000001e-05,
      "loss": 0.414,
      "step": 2886
    },
    {
      "epoch": 0.1616642401164744,
      "grad_norm": 1.3002201318740845,
      "learning_rate": 1.4425e-05,
      "loss": 0.5592,
      "step": 2887
    },
    {
      "epoch": 0.16172023742860342,
      "grad_norm": 1.2284952402114868,
      "learning_rate": 1.4430000000000002e-05,
      "loss": 0.4932,
      "step": 2888
    },
    {
      "epoch": 0.16177623474073244,
      "grad_norm": 1.1353243589401245,
      "learning_rate": 1.4435000000000002e-05,
      "loss": 0.455,
      "step": 2889
    },
    {
      "epoch": 0.16183223205286146,
      "grad_norm": 1.4194884300231934,
      "learning_rate": 1.444e-05,
      "loss": 0.4574,
      "step": 2890
    },
    {
      "epoch": 0.16188822936499048,
      "grad_norm": 0.9726737141609192,
      "learning_rate": 1.4445e-05,
      "loss": 0.4064,
      "step": 2891
    },
    {
      "epoch": 0.1619442266771195,
      "grad_norm": 1.0060579776763916,
      "learning_rate": 1.4449999999999999e-05,
      "loss": 0.3954,
      "step": 2892
    },
    {
      "epoch": 0.16200022398924852,
      "grad_norm": 1.358452320098877,
      "learning_rate": 1.4455000000000001e-05,
      "loss": 0.5166,
      "step": 2893
    },
    {
      "epoch": 0.16205622130137753,
      "grad_norm": 1.0128717422485352,
      "learning_rate": 1.4460000000000002e-05,
      "loss": 0.364,
      "step": 2894
    },
    {
      "epoch": 0.16211221861350655,
      "grad_norm": 1.2332308292388916,
      "learning_rate": 1.4465000000000001e-05,
      "loss": 0.3993,
      "step": 2895
    },
    {
      "epoch": 0.16216821592563557,
      "grad_norm": 1.2258899211883545,
      "learning_rate": 1.447e-05,
      "loss": 0.4307,
      "step": 2896
    },
    {
      "epoch": 0.1622242132377646,
      "grad_norm": 1.2921556234359741,
      "learning_rate": 1.4475e-05,
      "loss": 0.4426,
      "step": 2897
    },
    {
      "epoch": 0.1622802105498936,
      "grad_norm": 1.0930753946304321,
      "learning_rate": 1.4480000000000002e-05,
      "loss": 0.3127,
      "step": 2898
    },
    {
      "epoch": 0.16233620786202263,
      "grad_norm": 1.4140493869781494,
      "learning_rate": 1.4485000000000001e-05,
      "loss": 0.4241,
      "step": 2899
    },
    {
      "epoch": 0.16239220517415165,
      "grad_norm": 1.1093642711639404,
      "learning_rate": 1.449e-05,
      "loss": 0.4391,
      "step": 2900
    },
    {
      "epoch": 0.16244820248628067,
      "grad_norm": 0.9811723232269287,
      "learning_rate": 1.4495000000000001e-05,
      "loss": 0.3002,
      "step": 2901
    },
    {
      "epoch": 0.1625041997984097,
      "grad_norm": 1.0897340774536133,
      "learning_rate": 1.45e-05,
      "loss": 0.4447,
      "step": 2902
    },
    {
      "epoch": 0.1625601971105387,
      "grad_norm": 1.60068941116333,
      "learning_rate": 1.4505000000000003e-05,
      "loss": 0.4836,
      "step": 2903
    },
    {
      "epoch": 0.1626161944226677,
      "grad_norm": 1.0105966329574585,
      "learning_rate": 1.4510000000000002e-05,
      "loss": 0.4024,
      "step": 2904
    },
    {
      "epoch": 0.16267219173479672,
      "grad_norm": 1.241045594215393,
      "learning_rate": 1.4515e-05,
      "loss": 0.452,
      "step": 2905
    },
    {
      "epoch": 0.16272818904692574,
      "grad_norm": 1.3478405475616455,
      "learning_rate": 1.452e-05,
      "loss": 0.6681,
      "step": 2906
    },
    {
      "epoch": 0.16278418635905476,
      "grad_norm": 1.3035074472427368,
      "learning_rate": 1.4524999999999999e-05,
      "loss": 0.4031,
      "step": 2907
    },
    {
      "epoch": 0.16284018367118377,
      "grad_norm": 1.215703010559082,
      "learning_rate": 1.4530000000000001e-05,
      "loss": 0.4691,
      "step": 2908
    },
    {
      "epoch": 0.1628961809833128,
      "grad_norm": 1.289297103881836,
      "learning_rate": 1.4535e-05,
      "loss": 0.3669,
      "step": 2909
    },
    {
      "epoch": 0.1629521782954418,
      "grad_norm": 1.149066686630249,
      "learning_rate": 1.4540000000000001e-05,
      "loss": 0.4264,
      "step": 2910
    },
    {
      "epoch": 0.16300817560757083,
      "grad_norm": 1.2812926769256592,
      "learning_rate": 1.4545e-05,
      "loss": 0.4938,
      "step": 2911
    },
    {
      "epoch": 0.16306417291969985,
      "grad_norm": 1.0730267763137817,
      "learning_rate": 1.455e-05,
      "loss": 0.4693,
      "step": 2912
    },
    {
      "epoch": 0.16312017023182887,
      "grad_norm": 1.1804014444351196,
      "learning_rate": 1.4555000000000002e-05,
      "loss": 0.4372,
      "step": 2913
    },
    {
      "epoch": 0.1631761675439579,
      "grad_norm": 1.2339531183242798,
      "learning_rate": 1.4560000000000001e-05,
      "loss": 0.4266,
      "step": 2914
    },
    {
      "epoch": 0.1632321648560869,
      "grad_norm": 1.1102869510650635,
      "learning_rate": 1.4565e-05,
      "loss": 0.4657,
      "step": 2915
    },
    {
      "epoch": 0.16328816216821593,
      "grad_norm": 1.4539523124694824,
      "learning_rate": 1.4570000000000001e-05,
      "loss": 0.4176,
      "step": 2916
    },
    {
      "epoch": 0.16334415948034495,
      "grad_norm": 1.2633984088897705,
      "learning_rate": 1.4575e-05,
      "loss": 0.4214,
      "step": 2917
    },
    {
      "epoch": 0.16340015679247397,
      "grad_norm": 1.2158887386322021,
      "learning_rate": 1.4580000000000003e-05,
      "loss": 0.4991,
      "step": 2918
    },
    {
      "epoch": 0.16345615410460299,
      "grad_norm": 1.266831636428833,
      "learning_rate": 1.4585000000000002e-05,
      "loss": 0.5787,
      "step": 2919
    },
    {
      "epoch": 0.163512151416732,
      "grad_norm": 1.1851109266281128,
      "learning_rate": 1.4590000000000001e-05,
      "loss": 0.5018,
      "step": 2920
    },
    {
      "epoch": 0.16356814872886102,
      "grad_norm": 1.096369981765747,
      "learning_rate": 1.4595e-05,
      "loss": 0.3453,
      "step": 2921
    },
    {
      "epoch": 0.16362414604099004,
      "grad_norm": 1.7543258666992188,
      "learning_rate": 1.4599999999999999e-05,
      "loss": 0.5218,
      "step": 2922
    },
    {
      "epoch": 0.16368014335311906,
      "grad_norm": 1.2711646556854248,
      "learning_rate": 1.4605000000000002e-05,
      "loss": 0.3713,
      "step": 2923
    },
    {
      "epoch": 0.16373614066524808,
      "grad_norm": 1.3825998306274414,
      "learning_rate": 1.461e-05,
      "loss": 0.4534,
      "step": 2924
    },
    {
      "epoch": 0.16379213797737707,
      "grad_norm": 1.39090096950531,
      "learning_rate": 1.4615000000000002e-05,
      "loss": 0.6809,
      "step": 2925
    },
    {
      "epoch": 0.1638481352895061,
      "grad_norm": 1.1341646909713745,
      "learning_rate": 1.462e-05,
      "loss": 0.2352,
      "step": 2926
    },
    {
      "epoch": 0.1639041326016351,
      "grad_norm": 1.1790781021118164,
      "learning_rate": 1.4625e-05,
      "loss": 0.4092,
      "step": 2927
    },
    {
      "epoch": 0.16396012991376413,
      "grad_norm": 1.4217054843902588,
      "learning_rate": 1.4630000000000002e-05,
      "loss": 0.5113,
      "step": 2928
    },
    {
      "epoch": 0.16401612722589315,
      "grad_norm": 1.1839795112609863,
      "learning_rate": 1.4635000000000001e-05,
      "loss": 0.4317,
      "step": 2929
    },
    {
      "epoch": 0.16407212453802217,
      "grad_norm": 1.1309460401535034,
      "learning_rate": 1.464e-05,
      "loss": 0.3266,
      "step": 2930
    },
    {
      "epoch": 0.1641281218501512,
      "grad_norm": 1.3069939613342285,
      "learning_rate": 1.4645e-05,
      "loss": 0.5372,
      "step": 2931
    },
    {
      "epoch": 0.1641841191622802,
      "grad_norm": 1.2711175680160522,
      "learning_rate": 1.465e-05,
      "loss": 0.4304,
      "step": 2932
    },
    {
      "epoch": 0.16424011647440923,
      "grad_norm": 1.1943514347076416,
      "learning_rate": 1.4655000000000003e-05,
      "loss": 0.3806,
      "step": 2933
    },
    {
      "epoch": 0.16429611378653824,
      "grad_norm": 1.2070180177688599,
      "learning_rate": 1.4660000000000002e-05,
      "loss": 0.4196,
      "step": 2934
    },
    {
      "epoch": 0.16435211109866726,
      "grad_norm": 1.223737120628357,
      "learning_rate": 1.4665000000000001e-05,
      "loss": 0.4812,
      "step": 2935
    },
    {
      "epoch": 0.16440810841079628,
      "grad_norm": 1.440280556678772,
      "learning_rate": 1.467e-05,
      "loss": 0.4457,
      "step": 2936
    },
    {
      "epoch": 0.1644641057229253,
      "grad_norm": 1.0822083950042725,
      "learning_rate": 1.4675e-05,
      "loss": 0.3355,
      "step": 2937
    },
    {
      "epoch": 0.16452010303505432,
      "grad_norm": 1.3040459156036377,
      "learning_rate": 1.4680000000000002e-05,
      "loss": 0.4158,
      "step": 2938
    },
    {
      "epoch": 0.16457610034718334,
      "grad_norm": 1.1827491521835327,
      "learning_rate": 1.4685000000000001e-05,
      "loss": 0.4322,
      "step": 2939
    },
    {
      "epoch": 0.16463209765931236,
      "grad_norm": 1.0448668003082275,
      "learning_rate": 1.4690000000000002e-05,
      "loss": 0.3462,
      "step": 2940
    },
    {
      "epoch": 0.16468809497144138,
      "grad_norm": 1.0867873430252075,
      "learning_rate": 1.4695e-05,
      "loss": 0.4471,
      "step": 2941
    },
    {
      "epoch": 0.1647440922835704,
      "grad_norm": 1.357195496559143,
      "learning_rate": 1.47e-05,
      "loss": 0.491,
      "step": 2942
    },
    {
      "epoch": 0.16480008959569942,
      "grad_norm": 1.3898372650146484,
      "learning_rate": 1.4704999999999999e-05,
      "loss": 0.5863,
      "step": 2943
    },
    {
      "epoch": 0.16485608690782844,
      "grad_norm": 1.168304681777954,
      "learning_rate": 1.4710000000000001e-05,
      "loss": 0.3575,
      "step": 2944
    },
    {
      "epoch": 0.16491208421995746,
      "grad_norm": 1.2201783657073975,
      "learning_rate": 1.4715e-05,
      "loss": 0.4379,
      "step": 2945
    },
    {
      "epoch": 0.16496808153208645,
      "grad_norm": 1.1552715301513672,
      "learning_rate": 1.472e-05,
      "loss": 0.4842,
      "step": 2946
    },
    {
      "epoch": 0.16502407884421547,
      "grad_norm": 1.0281933546066284,
      "learning_rate": 1.4725e-05,
      "loss": 0.344,
      "step": 2947
    },
    {
      "epoch": 0.16508007615634449,
      "grad_norm": 1.1922563314437866,
      "learning_rate": 1.473e-05,
      "loss": 0.4426,
      "step": 2948
    },
    {
      "epoch": 0.1651360734684735,
      "grad_norm": 1.4431538581848145,
      "learning_rate": 1.4735000000000002e-05,
      "loss": 0.5686,
      "step": 2949
    },
    {
      "epoch": 0.16519207078060252,
      "grad_norm": 1.0972483158111572,
      "learning_rate": 1.4740000000000001e-05,
      "loss": 0.36,
      "step": 2950
    },
    {
      "epoch": 0.16524806809273154,
      "grad_norm": 1.3279143571853638,
      "learning_rate": 1.4745e-05,
      "loss": 0.4609,
      "step": 2951
    },
    {
      "epoch": 0.16530406540486056,
      "grad_norm": 1.2672828435897827,
      "learning_rate": 1.475e-05,
      "loss": 0.5439,
      "step": 2952
    },
    {
      "epoch": 0.16536006271698958,
      "grad_norm": 1.2774341106414795,
      "learning_rate": 1.4755e-05,
      "loss": 0.3992,
      "step": 2953
    },
    {
      "epoch": 0.1654160600291186,
      "grad_norm": 1.1626168489456177,
      "learning_rate": 1.4760000000000001e-05,
      "loss": 0.4066,
      "step": 2954
    },
    {
      "epoch": 0.16547205734124762,
      "grad_norm": 1.351238489151001,
      "learning_rate": 1.4765000000000002e-05,
      "loss": 0.4526,
      "step": 2955
    },
    {
      "epoch": 0.16552805465337664,
      "grad_norm": 1.3523439168930054,
      "learning_rate": 1.4770000000000001e-05,
      "loss": 0.2999,
      "step": 2956
    },
    {
      "epoch": 0.16558405196550566,
      "grad_norm": 1.2778282165527344,
      "learning_rate": 1.4775e-05,
      "loss": 0.4493,
      "step": 2957
    },
    {
      "epoch": 0.16564004927763468,
      "grad_norm": 1.4594963788986206,
      "learning_rate": 1.4779999999999999e-05,
      "loss": 0.4177,
      "step": 2958
    },
    {
      "epoch": 0.1656960465897637,
      "grad_norm": 1.1730430126190186,
      "learning_rate": 1.4785000000000002e-05,
      "loss": 0.4972,
      "step": 2959
    },
    {
      "epoch": 0.16575204390189271,
      "grad_norm": 1.3805487155914307,
      "learning_rate": 1.479e-05,
      "loss": 0.4852,
      "step": 2960
    },
    {
      "epoch": 0.16580804121402173,
      "grad_norm": 1.6004854440689087,
      "learning_rate": 1.4795e-05,
      "loss": 0.4528,
      "step": 2961
    },
    {
      "epoch": 0.16586403852615075,
      "grad_norm": 1.24202299118042,
      "learning_rate": 1.48e-05,
      "loss": 0.4216,
      "step": 2962
    },
    {
      "epoch": 0.16592003583827977,
      "grad_norm": 1.1655060052871704,
      "learning_rate": 1.4805e-05,
      "loss": 0.298,
      "step": 2963
    },
    {
      "epoch": 0.1659760331504088,
      "grad_norm": 1.1057674884796143,
      "learning_rate": 1.4810000000000002e-05,
      "loss": 0.3616,
      "step": 2964
    },
    {
      "epoch": 0.1660320304625378,
      "grad_norm": 1.1026337146759033,
      "learning_rate": 1.4815000000000001e-05,
      "loss": 0.4615,
      "step": 2965
    },
    {
      "epoch": 0.1660880277746668,
      "grad_norm": 1.2011353969573975,
      "learning_rate": 1.482e-05,
      "loss": 0.4535,
      "step": 2966
    },
    {
      "epoch": 0.16614402508679582,
      "grad_norm": 1.1062843799591064,
      "learning_rate": 1.4825e-05,
      "loss": 0.3815,
      "step": 2967
    },
    {
      "epoch": 0.16620002239892484,
      "grad_norm": 1.2528876066207886,
      "learning_rate": 1.4829999999999999e-05,
      "loss": 0.4937,
      "step": 2968
    },
    {
      "epoch": 0.16625601971105386,
      "grad_norm": 1.081746220588684,
      "learning_rate": 1.4835000000000001e-05,
      "loss": 0.4291,
      "step": 2969
    },
    {
      "epoch": 0.16631201702318288,
      "grad_norm": 1.4553874731063843,
      "learning_rate": 1.4840000000000002e-05,
      "loss": 0.5446,
      "step": 2970
    },
    {
      "epoch": 0.1663680143353119,
      "grad_norm": 1.3763076066970825,
      "learning_rate": 1.4845000000000001e-05,
      "loss": 0.4954,
      "step": 2971
    },
    {
      "epoch": 0.16642401164744092,
      "grad_norm": 1.634247899055481,
      "learning_rate": 1.485e-05,
      "loss": 0.5409,
      "step": 2972
    },
    {
      "epoch": 0.16648000895956994,
      "grad_norm": 1.0740395784378052,
      "learning_rate": 1.4855e-05,
      "loss": 0.4845,
      "step": 2973
    },
    {
      "epoch": 0.16653600627169896,
      "grad_norm": 1.0615090131759644,
      "learning_rate": 1.4860000000000002e-05,
      "loss": 0.431,
      "step": 2974
    },
    {
      "epoch": 0.16659200358382797,
      "grad_norm": 1.1746183633804321,
      "learning_rate": 1.4865e-05,
      "loss": 0.4175,
      "step": 2975
    },
    {
      "epoch": 0.166648000895957,
      "grad_norm": 1.1579493284225464,
      "learning_rate": 1.487e-05,
      "loss": 0.5676,
      "step": 2976
    },
    {
      "epoch": 0.166703998208086,
      "grad_norm": 1.1894301176071167,
      "learning_rate": 1.4875e-05,
      "loss": 0.3935,
      "step": 2977
    },
    {
      "epoch": 0.16675999552021503,
      "grad_norm": 1.4352726936340332,
      "learning_rate": 1.488e-05,
      "loss": 0.525,
      "step": 2978
    },
    {
      "epoch": 0.16681599283234405,
      "grad_norm": 1.2767298221588135,
      "learning_rate": 1.4885000000000002e-05,
      "loss": 0.4904,
      "step": 2979
    },
    {
      "epoch": 0.16687199014447307,
      "grad_norm": 1.2601221799850464,
      "learning_rate": 1.4890000000000001e-05,
      "loss": 0.4483,
      "step": 2980
    },
    {
      "epoch": 0.1669279874566021,
      "grad_norm": 1.524616003036499,
      "learning_rate": 1.4895e-05,
      "loss": 0.6168,
      "step": 2981
    },
    {
      "epoch": 0.1669839847687311,
      "grad_norm": 1.2325519323349,
      "learning_rate": 1.49e-05,
      "loss": 0.4345,
      "step": 2982
    },
    {
      "epoch": 0.16703998208086013,
      "grad_norm": 1.2944282293319702,
      "learning_rate": 1.4904999999999999e-05,
      "loss": 0.5398,
      "step": 2983
    },
    {
      "epoch": 0.16709597939298915,
      "grad_norm": 1.2927069664001465,
      "learning_rate": 1.4910000000000001e-05,
      "loss": 0.6528,
      "step": 2984
    },
    {
      "epoch": 0.16715197670511817,
      "grad_norm": 1.38869047164917,
      "learning_rate": 1.4915000000000002e-05,
      "loss": 0.4316,
      "step": 2985
    },
    {
      "epoch": 0.16720797401724719,
      "grad_norm": 1.2885438203811646,
      "learning_rate": 1.4920000000000001e-05,
      "loss": 0.4888,
      "step": 2986
    },
    {
      "epoch": 0.16726397132937618,
      "grad_norm": 1.0417410135269165,
      "learning_rate": 1.4925e-05,
      "loss": 0.3589,
      "step": 2987
    },
    {
      "epoch": 0.1673199686415052,
      "grad_norm": 1.3367822170257568,
      "learning_rate": 1.493e-05,
      "loss": 0.495,
      "step": 2988
    },
    {
      "epoch": 0.16737596595363421,
      "grad_norm": 1.2681429386138916,
      "learning_rate": 1.4935000000000002e-05,
      "loss": 0.411,
      "step": 2989
    },
    {
      "epoch": 0.16743196326576323,
      "grad_norm": 1.2044123411178589,
      "learning_rate": 1.4940000000000001e-05,
      "loss": 0.4475,
      "step": 2990
    },
    {
      "epoch": 0.16748796057789225,
      "grad_norm": 1.0016268491744995,
      "learning_rate": 1.4945e-05,
      "loss": 0.3603,
      "step": 2991
    },
    {
      "epoch": 0.16754395789002127,
      "grad_norm": 1.2403982877731323,
      "learning_rate": 1.4950000000000001e-05,
      "loss": 0.3746,
      "step": 2992
    },
    {
      "epoch": 0.1675999552021503,
      "grad_norm": 1.594496250152588,
      "learning_rate": 1.4955e-05,
      "loss": 0.625,
      "step": 2993
    },
    {
      "epoch": 0.1676559525142793,
      "grad_norm": 1.1329015493392944,
      "learning_rate": 1.4960000000000002e-05,
      "loss": 0.3695,
      "step": 2994
    },
    {
      "epoch": 0.16771194982640833,
      "grad_norm": 1.2726589441299438,
      "learning_rate": 1.4965000000000002e-05,
      "loss": 0.4215,
      "step": 2995
    },
    {
      "epoch": 0.16776794713853735,
      "grad_norm": 1.1244759559631348,
      "learning_rate": 1.497e-05,
      "loss": 0.4554,
      "step": 2996
    },
    {
      "epoch": 0.16782394445066637,
      "grad_norm": 1.028041958808899,
      "learning_rate": 1.4975e-05,
      "loss": 0.4082,
      "step": 2997
    },
    {
      "epoch": 0.1678799417627954,
      "grad_norm": 1.3320670127868652,
      "learning_rate": 1.4979999999999999e-05,
      "loss": 0.5336,
      "step": 2998
    },
    {
      "epoch": 0.1679359390749244,
      "grad_norm": 1.1810829639434814,
      "learning_rate": 1.4985000000000001e-05,
      "loss": 0.4698,
      "step": 2999
    },
    {
      "epoch": 0.16799193638705343,
      "grad_norm": 1.2294703722000122,
      "learning_rate": 1.499e-05,
      "loss": 0.5342,
      "step": 3000
    },
    {
      "epoch": 0.16804793369918244,
      "grad_norm": 13.666152954101562,
      "learning_rate": 1.4995000000000001e-05,
      "loss": 0.4374,
      "step": 3001
    },
    {
      "epoch": 0.16810393101131146,
      "grad_norm": 1.1388001441955566,
      "learning_rate": 1.5e-05,
      "loss": 0.4293,
      "step": 3002
    },
    {
      "epoch": 0.16815992832344048,
      "grad_norm": 1.140563726425171,
      "learning_rate": 1.5005e-05,
      "loss": 0.411,
      "step": 3003
    },
    {
      "epoch": 0.1682159256355695,
      "grad_norm": 1.1854054927825928,
      "learning_rate": 1.5010000000000002e-05,
      "loss": 0.3789,
      "step": 3004
    },
    {
      "epoch": 0.16827192294769852,
      "grad_norm": 1.3162811994552612,
      "learning_rate": 1.5015000000000001e-05,
      "loss": 0.5886,
      "step": 3005
    },
    {
      "epoch": 0.16832792025982754,
      "grad_norm": 1.1015150547027588,
      "learning_rate": 1.502e-05,
      "loss": 0.3127,
      "step": 3006
    },
    {
      "epoch": 0.16838391757195656,
      "grad_norm": 1.1969358921051025,
      "learning_rate": 1.5025000000000001e-05,
      "loss": 0.4865,
      "step": 3007
    },
    {
      "epoch": 0.16843991488408555,
      "grad_norm": 1.3611640930175781,
      "learning_rate": 1.503e-05,
      "loss": 0.5478,
      "step": 3008
    },
    {
      "epoch": 0.16849591219621457,
      "grad_norm": 1.1446624994277954,
      "learning_rate": 1.5035000000000003e-05,
      "loss": 0.4432,
      "step": 3009
    },
    {
      "epoch": 0.1685519095083436,
      "grad_norm": 1.3180516958236694,
      "learning_rate": 1.5040000000000002e-05,
      "loss": 0.4095,
      "step": 3010
    },
    {
      "epoch": 0.1686079068204726,
      "grad_norm": 1.0477768182754517,
      "learning_rate": 1.5045e-05,
      "loss": 0.4066,
      "step": 3011
    },
    {
      "epoch": 0.16866390413260163,
      "grad_norm": 1.051212191581726,
      "learning_rate": 1.505e-05,
      "loss": 0.3413,
      "step": 3012
    },
    {
      "epoch": 0.16871990144473065,
      "grad_norm": 1.187883734703064,
      "learning_rate": 1.5054999999999999e-05,
      "loss": 0.3827,
      "step": 3013
    },
    {
      "epoch": 0.16877589875685967,
      "grad_norm": 1.0885032415390015,
      "learning_rate": 1.5060000000000001e-05,
      "loss": 0.4144,
      "step": 3014
    },
    {
      "epoch": 0.16883189606898868,
      "grad_norm": 1.3088181018829346,
      "learning_rate": 1.5065e-05,
      "loss": 0.4413,
      "step": 3015
    },
    {
      "epoch": 0.1688878933811177,
      "grad_norm": 1.3181278705596924,
      "learning_rate": 1.5070000000000001e-05,
      "loss": 0.3683,
      "step": 3016
    },
    {
      "epoch": 0.16894389069324672,
      "grad_norm": 1.3163723945617676,
      "learning_rate": 1.5075e-05,
      "loss": 0.4511,
      "step": 3017
    },
    {
      "epoch": 0.16899988800537574,
      "grad_norm": 1.29971444606781,
      "learning_rate": 1.508e-05,
      "loss": 0.4471,
      "step": 3018
    },
    {
      "epoch": 0.16905588531750476,
      "grad_norm": 1.2885290384292603,
      "learning_rate": 1.5085000000000002e-05,
      "loss": 0.4663,
      "step": 3019
    },
    {
      "epoch": 0.16911188262963378,
      "grad_norm": 1.0854727029800415,
      "learning_rate": 1.5090000000000001e-05,
      "loss": 0.4764,
      "step": 3020
    },
    {
      "epoch": 0.1691678799417628,
      "grad_norm": 1.3393666744232178,
      "learning_rate": 1.5095e-05,
      "loss": 0.4239,
      "step": 3021
    },
    {
      "epoch": 0.16922387725389182,
      "grad_norm": 1.4022681713104248,
      "learning_rate": 1.51e-05,
      "loss": 0.5483,
      "step": 3022
    },
    {
      "epoch": 0.16927987456602084,
      "grad_norm": 1.2516348361968994,
      "learning_rate": 1.5105e-05,
      "loss": 0.4618,
      "step": 3023
    },
    {
      "epoch": 0.16933587187814986,
      "grad_norm": 1.2458224296569824,
      "learning_rate": 1.5110000000000003e-05,
      "loss": 0.4085,
      "step": 3024
    },
    {
      "epoch": 0.16939186919027888,
      "grad_norm": 1.6810840368270874,
      "learning_rate": 1.5115000000000002e-05,
      "loss": 0.3476,
      "step": 3025
    },
    {
      "epoch": 0.1694478665024079,
      "grad_norm": 1.6540249586105347,
      "learning_rate": 1.5120000000000001e-05,
      "loss": 0.4504,
      "step": 3026
    },
    {
      "epoch": 0.16950386381453691,
      "grad_norm": 1.420183777809143,
      "learning_rate": 1.5125e-05,
      "loss": 0.5645,
      "step": 3027
    },
    {
      "epoch": 0.16955986112666593,
      "grad_norm": 1.0648268461227417,
      "learning_rate": 1.5129999999999999e-05,
      "loss": 0.3991,
      "step": 3028
    },
    {
      "epoch": 0.16961585843879493,
      "grad_norm": 1.2405067682266235,
      "learning_rate": 1.5135000000000002e-05,
      "loss": 0.5124,
      "step": 3029
    },
    {
      "epoch": 0.16967185575092394,
      "grad_norm": 1.7114293575286865,
      "learning_rate": 1.514e-05,
      "loss": 0.5756,
      "step": 3030
    },
    {
      "epoch": 0.16972785306305296,
      "grad_norm": 1.1942071914672852,
      "learning_rate": 1.5145000000000002e-05,
      "loss": 0.4559,
      "step": 3031
    },
    {
      "epoch": 0.16978385037518198,
      "grad_norm": 1.2123992443084717,
      "learning_rate": 1.515e-05,
      "loss": 0.3711,
      "step": 3032
    },
    {
      "epoch": 0.169839847687311,
      "grad_norm": 1.1135144233703613,
      "learning_rate": 1.5155e-05,
      "loss": 0.5915,
      "step": 3033
    },
    {
      "epoch": 0.16989584499944002,
      "grad_norm": 1.4512019157409668,
      "learning_rate": 1.5160000000000002e-05,
      "loss": 0.4455,
      "step": 3034
    },
    {
      "epoch": 0.16995184231156904,
      "grad_norm": 2.1916909217834473,
      "learning_rate": 1.5165000000000001e-05,
      "loss": 0.4776,
      "step": 3035
    },
    {
      "epoch": 0.17000783962369806,
      "grad_norm": 1.0539445877075195,
      "learning_rate": 1.517e-05,
      "loss": 0.4514,
      "step": 3036
    },
    {
      "epoch": 0.17006383693582708,
      "grad_norm": 1.3222472667694092,
      "learning_rate": 1.5175e-05,
      "loss": 0.4183,
      "step": 3037
    },
    {
      "epoch": 0.1701198342479561,
      "grad_norm": 1.3580297231674194,
      "learning_rate": 1.518e-05,
      "loss": 0.5505,
      "step": 3038
    },
    {
      "epoch": 0.17017583156008512,
      "grad_norm": 1.196560025215149,
      "learning_rate": 1.5185000000000003e-05,
      "loss": 0.437,
      "step": 3039
    },
    {
      "epoch": 0.17023182887221414,
      "grad_norm": 1.0886143445968628,
      "learning_rate": 1.5190000000000002e-05,
      "loss": 0.332,
      "step": 3040
    },
    {
      "epoch": 0.17028782618434316,
      "grad_norm": 1.1416614055633545,
      "learning_rate": 1.5195000000000001e-05,
      "loss": 0.3313,
      "step": 3041
    },
    {
      "epoch": 0.17034382349647217,
      "grad_norm": 1.246839165687561,
      "learning_rate": 1.52e-05,
      "loss": 0.5246,
      "step": 3042
    },
    {
      "epoch": 0.1703998208086012,
      "grad_norm": 1.0934566259384155,
      "learning_rate": 1.5205e-05,
      "loss": 0.488,
      "step": 3043
    },
    {
      "epoch": 0.1704558181207302,
      "grad_norm": 1.2822872400283813,
      "learning_rate": 1.5210000000000002e-05,
      "loss": 0.4606,
      "step": 3044
    },
    {
      "epoch": 0.17051181543285923,
      "grad_norm": 1.2843599319458008,
      "learning_rate": 1.5215000000000001e-05,
      "loss": 0.4081,
      "step": 3045
    },
    {
      "epoch": 0.17056781274498825,
      "grad_norm": 1.3425087928771973,
      "learning_rate": 1.5220000000000002e-05,
      "loss": 0.4268,
      "step": 3046
    },
    {
      "epoch": 0.17062381005711727,
      "grad_norm": 1.0990785360336304,
      "learning_rate": 1.5225e-05,
      "loss": 0.3835,
      "step": 3047
    },
    {
      "epoch": 0.1706798073692463,
      "grad_norm": 1.4088256359100342,
      "learning_rate": 1.523e-05,
      "loss": 0.8543,
      "step": 3048
    },
    {
      "epoch": 0.17073580468137528,
      "grad_norm": 2.091273546218872,
      "learning_rate": 1.5235000000000002e-05,
      "loss": 0.4991,
      "step": 3049
    },
    {
      "epoch": 0.1707918019935043,
      "grad_norm": 1.6073461771011353,
      "learning_rate": 1.5240000000000001e-05,
      "loss": 0.3343,
      "step": 3050
    },
    {
      "epoch": 0.17084779930563332,
      "grad_norm": 1.0276155471801758,
      "learning_rate": 1.5245e-05,
      "loss": 0.3591,
      "step": 3051
    },
    {
      "epoch": 0.17090379661776234,
      "grad_norm": 1.3987497091293335,
      "learning_rate": 1.525e-05,
      "loss": 0.7108,
      "step": 3052
    },
    {
      "epoch": 0.17095979392989136,
      "grad_norm": 1.3775502443313599,
      "learning_rate": 1.5255e-05,
      "loss": 0.3814,
      "step": 3053
    },
    {
      "epoch": 0.17101579124202038,
      "grad_norm": 1.2856470346450806,
      "learning_rate": 1.5260000000000003e-05,
      "loss": 0.4444,
      "step": 3054
    },
    {
      "epoch": 0.1710717885541494,
      "grad_norm": 1.4564297199249268,
      "learning_rate": 1.5265e-05,
      "loss": 0.5069,
      "step": 3055
    },
    {
      "epoch": 0.17112778586627841,
      "grad_norm": 1.328133225440979,
      "learning_rate": 1.527e-05,
      "loss": 0.5168,
      "step": 3056
    },
    {
      "epoch": 0.17118378317840743,
      "grad_norm": 1.12852144241333,
      "learning_rate": 1.5275000000000002e-05,
      "loss": 0.4805,
      "step": 3057
    },
    {
      "epoch": 0.17123978049053645,
      "grad_norm": 1.114654541015625,
      "learning_rate": 1.528e-05,
      "loss": 0.4181,
      "step": 3058
    },
    {
      "epoch": 0.17129577780266547,
      "grad_norm": 1.1266412734985352,
      "learning_rate": 1.5285000000000004e-05,
      "loss": 0.3826,
      "step": 3059
    },
    {
      "epoch": 0.1713517751147945,
      "grad_norm": 1.3138937950134277,
      "learning_rate": 1.529e-05,
      "loss": 0.4112,
      "step": 3060
    },
    {
      "epoch": 0.1714077724269235,
      "grad_norm": 1.2326487302780151,
      "learning_rate": 1.5295000000000002e-05,
      "loss": 0.4183,
      "step": 3061
    },
    {
      "epoch": 0.17146376973905253,
      "grad_norm": 1.2712304592132568,
      "learning_rate": 1.53e-05,
      "loss": 0.5426,
      "step": 3062
    },
    {
      "epoch": 0.17151976705118155,
      "grad_norm": 2.414008617401123,
      "learning_rate": 1.5305e-05,
      "loss": 0.5064,
      "step": 3063
    },
    {
      "epoch": 0.17157576436331057,
      "grad_norm": 1.1649055480957031,
      "learning_rate": 1.531e-05,
      "loss": 0.4934,
      "step": 3064
    },
    {
      "epoch": 0.1716317616754396,
      "grad_norm": 1.3338874578475952,
      "learning_rate": 1.5315e-05,
      "loss": 0.3964,
      "step": 3065
    },
    {
      "epoch": 0.1716877589875686,
      "grad_norm": 1.398659348487854,
      "learning_rate": 1.5320000000000002e-05,
      "loss": 0.4128,
      "step": 3066
    },
    {
      "epoch": 0.17174375629969763,
      "grad_norm": 1.2956663370132446,
      "learning_rate": 1.5325e-05,
      "loss": 0.4504,
      "step": 3067
    },
    {
      "epoch": 0.17179975361182664,
      "grad_norm": 1.2123852968215942,
      "learning_rate": 1.533e-05,
      "loss": 0.3877,
      "step": 3068
    },
    {
      "epoch": 0.17185575092395566,
      "grad_norm": 1.2996089458465576,
      "learning_rate": 1.5334999999999998e-05,
      "loss": 0.3504,
      "step": 3069
    },
    {
      "epoch": 0.17191174823608465,
      "grad_norm": 1.2874212265014648,
      "learning_rate": 1.5340000000000002e-05,
      "loss": 0.4792,
      "step": 3070
    },
    {
      "epoch": 0.17196774554821367,
      "grad_norm": 1.4098880290985107,
      "learning_rate": 1.5345e-05,
      "loss": 0.5226,
      "step": 3071
    },
    {
      "epoch": 0.1720237428603427,
      "grad_norm": 1.5210481882095337,
      "learning_rate": 1.535e-05,
      "loss": 0.6552,
      "step": 3072
    },
    {
      "epoch": 0.1720797401724717,
      "grad_norm": 1.165636658668518,
      "learning_rate": 1.5355e-05,
      "loss": 0.3838,
      "step": 3073
    },
    {
      "epoch": 0.17213573748460073,
      "grad_norm": 1.2327104806900024,
      "learning_rate": 1.536e-05,
      "loss": 0.5711,
      "step": 3074
    },
    {
      "epoch": 0.17219173479672975,
      "grad_norm": 1.2012931108474731,
      "learning_rate": 1.5365000000000003e-05,
      "loss": 0.3197,
      "step": 3075
    },
    {
      "epoch": 0.17224773210885877,
      "grad_norm": 1.347421407699585,
      "learning_rate": 1.537e-05,
      "loss": 0.4665,
      "step": 3076
    },
    {
      "epoch": 0.1723037294209878,
      "grad_norm": 1.07391357421875,
      "learning_rate": 1.5375e-05,
      "loss": 0.3609,
      "step": 3077
    },
    {
      "epoch": 0.1723597267331168,
      "grad_norm": 1.2701846361160278,
      "learning_rate": 1.538e-05,
      "loss": 0.4394,
      "step": 3078
    },
    {
      "epoch": 0.17241572404524583,
      "grad_norm": 1.3560216426849365,
      "learning_rate": 1.5385e-05,
      "loss": 0.3959,
      "step": 3079
    },
    {
      "epoch": 0.17247172135737485,
      "grad_norm": 1.2126895189285278,
      "learning_rate": 1.539e-05,
      "loss": 0.4248,
      "step": 3080
    },
    {
      "epoch": 0.17252771866950387,
      "grad_norm": 1.192703127861023,
      "learning_rate": 1.5395e-05,
      "loss": 0.4845,
      "step": 3081
    },
    {
      "epoch": 0.17258371598163288,
      "grad_norm": 1.2717024087905884,
      "learning_rate": 1.54e-05,
      "loss": 0.3706,
      "step": 3082
    },
    {
      "epoch": 0.1726397132937619,
      "grad_norm": 1.38326096534729,
      "learning_rate": 1.5405e-05,
      "loss": 0.3618,
      "step": 3083
    },
    {
      "epoch": 0.17269571060589092,
      "grad_norm": 1.286408543586731,
      "learning_rate": 1.541e-05,
      "loss": 0.4486,
      "step": 3084
    },
    {
      "epoch": 0.17275170791801994,
      "grad_norm": 1.0604596138000488,
      "learning_rate": 1.5415e-05,
      "loss": 0.3375,
      "step": 3085
    },
    {
      "epoch": 0.17280770523014896,
      "grad_norm": 1.3629469871520996,
      "learning_rate": 1.542e-05,
      "loss": 0.4036,
      "step": 3086
    },
    {
      "epoch": 0.17286370254227798,
      "grad_norm": 1.4825447797775269,
      "learning_rate": 1.5425000000000002e-05,
      "loss": 0.4507,
      "step": 3087
    },
    {
      "epoch": 0.172919699854407,
      "grad_norm": 1.2464237213134766,
      "learning_rate": 1.543e-05,
      "loss": 0.4137,
      "step": 3088
    },
    {
      "epoch": 0.17297569716653602,
      "grad_norm": 1.3638365268707275,
      "learning_rate": 1.5435e-05,
      "loss": 0.3957,
      "step": 3089
    },
    {
      "epoch": 0.17303169447866504,
      "grad_norm": 1.085571050643921,
      "learning_rate": 1.544e-05,
      "loss": 0.4331,
      "step": 3090
    },
    {
      "epoch": 0.17308769179079403,
      "grad_norm": 1.0439016819000244,
      "learning_rate": 1.5445000000000002e-05,
      "loss": 0.4425,
      "step": 3091
    },
    {
      "epoch": 0.17314368910292305,
      "grad_norm": 1.42640221118927,
      "learning_rate": 1.545e-05,
      "loss": 0.5665,
      "step": 3092
    },
    {
      "epoch": 0.17319968641505207,
      "grad_norm": 1.3006322383880615,
      "learning_rate": 1.5455e-05,
      "loss": 0.5638,
      "step": 3093
    },
    {
      "epoch": 0.1732556837271811,
      "grad_norm": 1.4400304555892944,
      "learning_rate": 1.546e-05,
      "loss": 0.4891,
      "step": 3094
    },
    {
      "epoch": 0.1733116810393101,
      "grad_norm": 1.1101337671279907,
      "learning_rate": 1.5465000000000002e-05,
      "loss": 0.3336,
      "step": 3095
    },
    {
      "epoch": 0.17336767835143913,
      "grad_norm": 1.6164960861206055,
      "learning_rate": 1.5470000000000003e-05,
      "loss": 0.3791,
      "step": 3096
    },
    {
      "epoch": 0.17342367566356814,
      "grad_norm": 1.3061712980270386,
      "learning_rate": 1.5475e-05,
      "loss": 0.4327,
      "step": 3097
    },
    {
      "epoch": 0.17347967297569716,
      "grad_norm": 1.2732435464859009,
      "learning_rate": 1.548e-05,
      "loss": 0.4474,
      "step": 3098
    },
    {
      "epoch": 0.17353567028782618,
      "grad_norm": 1.1919748783111572,
      "learning_rate": 1.5484999999999998e-05,
      "loss": 0.3302,
      "step": 3099
    },
    {
      "epoch": 0.1735916675999552,
      "grad_norm": 1.0183470249176025,
      "learning_rate": 1.5490000000000002e-05,
      "loss": 0.3668,
      "step": 3100
    },
    {
      "epoch": 0.17364766491208422,
      "grad_norm": 1.2144869565963745,
      "learning_rate": 1.5495e-05,
      "loss": 0.3694,
      "step": 3101
    },
    {
      "epoch": 0.17370366222421324,
      "grad_norm": 1.2006837129592896,
      "learning_rate": 1.55e-05,
      "loss": 0.3806,
      "step": 3102
    },
    {
      "epoch": 0.17375965953634226,
      "grad_norm": 1.1483407020568848,
      "learning_rate": 1.5505e-05,
      "loss": 0.4532,
      "step": 3103
    },
    {
      "epoch": 0.17381565684847128,
      "grad_norm": 1.499038577079773,
      "learning_rate": 1.551e-05,
      "loss": 0.4937,
      "step": 3104
    },
    {
      "epoch": 0.1738716541606003,
      "grad_norm": 1.1571745872497559,
      "learning_rate": 1.5515000000000003e-05,
      "loss": 0.3937,
      "step": 3105
    },
    {
      "epoch": 0.17392765147272932,
      "grad_norm": 1.2323365211486816,
      "learning_rate": 1.552e-05,
      "loss": 0.407,
      "step": 3106
    },
    {
      "epoch": 0.17398364878485834,
      "grad_norm": 4.923003673553467,
      "learning_rate": 1.5525e-05,
      "loss": 0.3192,
      "step": 3107
    },
    {
      "epoch": 0.17403964609698735,
      "grad_norm": 1.1060408353805542,
      "learning_rate": 1.553e-05,
      "loss": 0.4702,
      "step": 3108
    },
    {
      "epoch": 0.17409564340911637,
      "grad_norm": 1.3280766010284424,
      "learning_rate": 1.5535e-05,
      "loss": 0.6155,
      "step": 3109
    },
    {
      "epoch": 0.1741516407212454,
      "grad_norm": 1.5248066186904907,
      "learning_rate": 1.554e-05,
      "loss": 0.5402,
      "step": 3110
    },
    {
      "epoch": 0.17420763803337438,
      "grad_norm": 1.3715664148330688,
      "learning_rate": 1.5545e-05,
      "loss": 0.4532,
      "step": 3111
    },
    {
      "epoch": 0.1742636353455034,
      "grad_norm": 1.1363284587860107,
      "learning_rate": 1.5550000000000002e-05,
      "loss": 0.5191,
      "step": 3112
    },
    {
      "epoch": 0.17431963265763242,
      "grad_norm": 1.1898376941680908,
      "learning_rate": 1.5555e-05,
      "loss": 0.3865,
      "step": 3113
    },
    {
      "epoch": 0.17437562996976144,
      "grad_norm": 1.2565094232559204,
      "learning_rate": 1.556e-05,
      "loss": 0.762,
      "step": 3114
    },
    {
      "epoch": 0.17443162728189046,
      "grad_norm": 1.0949702262878418,
      "learning_rate": 1.5565e-05,
      "loss": 0.4829,
      "step": 3115
    },
    {
      "epoch": 0.17448762459401948,
      "grad_norm": 1.1513077020645142,
      "learning_rate": 1.5570000000000002e-05,
      "loss": 0.5075,
      "step": 3116
    },
    {
      "epoch": 0.1745436219061485,
      "grad_norm": 1.1390165090560913,
      "learning_rate": 1.5575e-05,
      "loss": 0.4113,
      "step": 3117
    },
    {
      "epoch": 0.17459961921827752,
      "grad_norm": 1.1877434253692627,
      "learning_rate": 1.558e-05,
      "loss": 0.3997,
      "step": 3118
    },
    {
      "epoch": 0.17465561653040654,
      "grad_norm": 1.2542484998703003,
      "learning_rate": 1.5585e-05,
      "loss": 0.3045,
      "step": 3119
    },
    {
      "epoch": 0.17471161384253556,
      "grad_norm": 1.3228293657302856,
      "learning_rate": 1.559e-05,
      "loss": 0.4405,
      "step": 3120
    },
    {
      "epoch": 0.17476761115466458,
      "grad_norm": 1.401941180229187,
      "learning_rate": 1.5595000000000002e-05,
      "loss": 0.4046,
      "step": 3121
    },
    {
      "epoch": 0.1748236084667936,
      "grad_norm": 1.18621826171875,
      "learning_rate": 1.56e-05,
      "loss": 0.3908,
      "step": 3122
    },
    {
      "epoch": 0.17487960577892261,
      "grad_norm": 1.0453542470932007,
      "learning_rate": 1.5605e-05,
      "loss": 0.2901,
      "step": 3123
    },
    {
      "epoch": 0.17493560309105163,
      "grad_norm": 1.1331298351287842,
      "learning_rate": 1.561e-05,
      "loss": 0.3826,
      "step": 3124
    },
    {
      "epoch": 0.17499160040318065,
      "grad_norm": 1.3581092357635498,
      "learning_rate": 1.5615000000000002e-05,
      "loss": 0.585,
      "step": 3125
    },
    {
      "epoch": 0.17504759771530967,
      "grad_norm": 1.1543517112731934,
      "learning_rate": 1.5620000000000003e-05,
      "loss": 0.3554,
      "step": 3126
    },
    {
      "epoch": 0.1751035950274387,
      "grad_norm": 1.265600562095642,
      "learning_rate": 1.5625e-05,
      "loss": 0.4692,
      "step": 3127
    },
    {
      "epoch": 0.1751595923395677,
      "grad_norm": 1.3256622552871704,
      "learning_rate": 1.563e-05,
      "loss": 0.4647,
      "step": 3128
    },
    {
      "epoch": 0.17521558965169673,
      "grad_norm": 1.303101897239685,
      "learning_rate": 1.5635e-05,
      "loss": 0.359,
      "step": 3129
    },
    {
      "epoch": 0.17527158696382575,
      "grad_norm": 1.1615647077560425,
      "learning_rate": 1.5640000000000003e-05,
      "loss": 0.5086,
      "step": 3130
    },
    {
      "epoch": 0.17532758427595477,
      "grad_norm": 1.257786750793457,
      "learning_rate": 1.5645e-05,
      "loss": 0.4834,
      "step": 3131
    },
    {
      "epoch": 0.17538358158808376,
      "grad_norm": 1.3199831247329712,
      "learning_rate": 1.565e-05,
      "loss": 0.4673,
      "step": 3132
    },
    {
      "epoch": 0.17543957890021278,
      "grad_norm": 1.4349507093429565,
      "learning_rate": 1.5655000000000002e-05,
      "loss": 0.5471,
      "step": 3133
    },
    {
      "epoch": 0.1754955762123418,
      "grad_norm": 1.3560909032821655,
      "learning_rate": 1.566e-05,
      "loss": 0.4311,
      "step": 3134
    },
    {
      "epoch": 0.17555157352447082,
      "grad_norm": 1.2484694719314575,
      "learning_rate": 1.5665000000000003e-05,
      "loss": 0.4119,
      "step": 3135
    },
    {
      "epoch": 0.17560757083659984,
      "grad_norm": 1.0213234424591064,
      "learning_rate": 1.567e-05,
      "loss": 0.3423,
      "step": 3136
    },
    {
      "epoch": 0.17566356814872885,
      "grad_norm": 1.153052806854248,
      "learning_rate": 1.5675e-05,
      "loss": 0.4602,
      "step": 3137
    },
    {
      "epoch": 0.17571956546085787,
      "grad_norm": 1.1225149631500244,
      "learning_rate": 1.568e-05,
      "loss": 0.3676,
      "step": 3138
    },
    {
      "epoch": 0.1757755627729869,
      "grad_norm": 1.1485469341278076,
      "learning_rate": 1.5685e-05,
      "loss": 0.5003,
      "step": 3139
    },
    {
      "epoch": 0.1758315600851159,
      "grad_norm": 1.3653138875961304,
      "learning_rate": 1.569e-05,
      "loss": 0.497,
      "step": 3140
    },
    {
      "epoch": 0.17588755739724493,
      "grad_norm": 1.5089391469955444,
      "learning_rate": 1.5695e-05,
      "loss": 0.3692,
      "step": 3141
    },
    {
      "epoch": 0.17594355470937395,
      "grad_norm": 1.2435405254364014,
      "learning_rate": 1.5700000000000002e-05,
      "loss": 0.4499,
      "step": 3142
    },
    {
      "epoch": 0.17599955202150297,
      "grad_norm": 1.225533366203308,
      "learning_rate": 1.5705e-05,
      "loss": 0.4798,
      "step": 3143
    },
    {
      "epoch": 0.176055549333632,
      "grad_norm": 1.2281639575958252,
      "learning_rate": 1.571e-05,
      "loss": 0.4124,
      "step": 3144
    },
    {
      "epoch": 0.176111546645761,
      "grad_norm": 1.0934711694717407,
      "learning_rate": 1.5715e-05,
      "loss": 0.385,
      "step": 3145
    },
    {
      "epoch": 0.17616754395789003,
      "grad_norm": 1.3133177757263184,
      "learning_rate": 1.5720000000000002e-05,
      "loss": 0.6233,
      "step": 3146
    },
    {
      "epoch": 0.17622354127001905,
      "grad_norm": 0.9715774059295654,
      "learning_rate": 1.5725e-05,
      "loss": 0.3162,
      "step": 3147
    },
    {
      "epoch": 0.17627953858214807,
      "grad_norm": 1.1965703964233398,
      "learning_rate": 1.573e-05,
      "loss": 0.3621,
      "step": 3148
    },
    {
      "epoch": 0.17633553589427708,
      "grad_norm": 1.2315959930419922,
      "learning_rate": 1.5735e-05,
      "loss": 0.4043,
      "step": 3149
    },
    {
      "epoch": 0.1763915332064061,
      "grad_norm": 0.9554809927940369,
      "learning_rate": 1.5740000000000002e-05,
      "loss": 0.3764,
      "step": 3150
    },
    {
      "epoch": 0.17644753051853512,
      "grad_norm": 1.1647918224334717,
      "learning_rate": 1.5745000000000003e-05,
      "loss": 0.6171,
      "step": 3151
    },
    {
      "epoch": 0.17650352783066414,
      "grad_norm": 1.220215082168579,
      "learning_rate": 1.575e-05,
      "loss": 0.5622,
      "step": 3152
    },
    {
      "epoch": 0.17655952514279313,
      "grad_norm": 1.1583800315856934,
      "learning_rate": 1.5755e-05,
      "loss": 0.3858,
      "step": 3153
    },
    {
      "epoch": 0.17661552245492215,
      "grad_norm": 1.3775138854980469,
      "learning_rate": 1.5759999999999998e-05,
      "loss": 0.5959,
      "step": 3154
    },
    {
      "epoch": 0.17667151976705117,
      "grad_norm": 1.2710376977920532,
      "learning_rate": 1.5765000000000002e-05,
      "loss": 0.4621,
      "step": 3155
    },
    {
      "epoch": 0.1767275170791802,
      "grad_norm": 1.2439836263656616,
      "learning_rate": 1.577e-05,
      "loss": 0.4974,
      "step": 3156
    },
    {
      "epoch": 0.1767835143913092,
      "grad_norm": 1.2212399244308472,
      "learning_rate": 1.5775e-05,
      "loss": 0.5128,
      "step": 3157
    },
    {
      "epoch": 0.17683951170343823,
      "grad_norm": 1.3426244258880615,
      "learning_rate": 1.578e-05,
      "loss": 0.5533,
      "step": 3158
    },
    {
      "epoch": 0.17689550901556725,
      "grad_norm": 1.1366629600524902,
      "learning_rate": 1.5785e-05,
      "loss": 0.4811,
      "step": 3159
    },
    {
      "epoch": 0.17695150632769627,
      "grad_norm": 1.3456504344940186,
      "learning_rate": 1.5790000000000003e-05,
      "loss": 0.4341,
      "step": 3160
    },
    {
      "epoch": 0.1770075036398253,
      "grad_norm": 1.1240496635437012,
      "learning_rate": 1.5795e-05,
      "loss": 0.4233,
      "step": 3161
    },
    {
      "epoch": 0.1770635009519543,
      "grad_norm": 1.4094661474227905,
      "learning_rate": 1.58e-05,
      "loss": 0.5234,
      "step": 3162
    },
    {
      "epoch": 0.17711949826408332,
      "grad_norm": 1.1591876745224,
      "learning_rate": 1.5805000000000002e-05,
      "loss": 0.5352,
      "step": 3163
    },
    {
      "epoch": 0.17717549557621234,
      "grad_norm": 1.3503307104110718,
      "learning_rate": 1.581e-05,
      "loss": 0.5129,
      "step": 3164
    },
    {
      "epoch": 0.17723149288834136,
      "grad_norm": 1.1911805868148804,
      "learning_rate": 1.5815000000000004e-05,
      "loss": 0.4462,
      "step": 3165
    },
    {
      "epoch": 0.17728749020047038,
      "grad_norm": 1.2154439687728882,
      "learning_rate": 1.582e-05,
      "loss": 0.4113,
      "step": 3166
    },
    {
      "epoch": 0.1773434875125994,
      "grad_norm": 1.424020528793335,
      "learning_rate": 1.5825000000000002e-05,
      "loss": 0.4254,
      "step": 3167
    },
    {
      "epoch": 0.17739948482472842,
      "grad_norm": 1.2356969118118286,
      "learning_rate": 1.583e-05,
      "loss": 0.4238,
      "step": 3168
    },
    {
      "epoch": 0.17745548213685744,
      "grad_norm": 1.2835613489151,
      "learning_rate": 1.5835e-05,
      "loss": 0.4716,
      "step": 3169
    },
    {
      "epoch": 0.17751147944898646,
      "grad_norm": 1.5604387521743774,
      "learning_rate": 1.584e-05,
      "loss": 0.5455,
      "step": 3170
    },
    {
      "epoch": 0.17756747676111548,
      "grad_norm": 1.066110372543335,
      "learning_rate": 1.5845e-05,
      "loss": 0.4306,
      "step": 3171
    },
    {
      "epoch": 0.1776234740732445,
      "grad_norm": 1.3697227239608765,
      "learning_rate": 1.5850000000000002e-05,
      "loss": 0.4028,
      "step": 3172
    },
    {
      "epoch": 0.1776794713853735,
      "grad_norm": 1.1536370515823364,
      "learning_rate": 1.5855e-05,
      "loss": 0.5904,
      "step": 3173
    },
    {
      "epoch": 0.1777354686975025,
      "grad_norm": 1.0349966287612915,
      "learning_rate": 1.586e-05,
      "loss": 0.3471,
      "step": 3174
    },
    {
      "epoch": 0.17779146600963153,
      "grad_norm": 1.278933048248291,
      "learning_rate": 1.5865e-05,
      "loss": 0.4645,
      "step": 3175
    },
    {
      "epoch": 0.17784746332176055,
      "grad_norm": 1.2004108428955078,
      "learning_rate": 1.5870000000000002e-05,
      "loss": 0.5068,
      "step": 3176
    },
    {
      "epoch": 0.17790346063388957,
      "grad_norm": 1.3695133924484253,
      "learning_rate": 1.5875e-05,
      "loss": 0.5353,
      "step": 3177
    },
    {
      "epoch": 0.17795945794601858,
      "grad_norm": 1.1981608867645264,
      "learning_rate": 1.588e-05,
      "loss": 0.3492,
      "step": 3178
    },
    {
      "epoch": 0.1780154552581476,
      "grad_norm": 1.4457505941390991,
      "learning_rate": 1.5885e-05,
      "loss": 0.6995,
      "step": 3179
    },
    {
      "epoch": 0.17807145257027662,
      "grad_norm": 1.5259352922439575,
      "learning_rate": 1.5890000000000002e-05,
      "loss": 0.7483,
      "step": 3180
    },
    {
      "epoch": 0.17812744988240564,
      "grad_norm": 1.0876466035842896,
      "learning_rate": 1.5895000000000003e-05,
      "loss": 0.4789,
      "step": 3181
    },
    {
      "epoch": 0.17818344719453466,
      "grad_norm": 1.0859925746917725,
      "learning_rate": 1.59e-05,
      "loss": 0.4018,
      "step": 3182
    },
    {
      "epoch": 0.17823944450666368,
      "grad_norm": 1.1666775941848755,
      "learning_rate": 1.5905e-05,
      "loss": 0.4115,
      "step": 3183
    },
    {
      "epoch": 0.1782954418187927,
      "grad_norm": 1.1463675498962402,
      "learning_rate": 1.591e-05,
      "loss": 0.3401,
      "step": 3184
    },
    {
      "epoch": 0.17835143913092172,
      "grad_norm": 1.154799461364746,
      "learning_rate": 1.5915000000000003e-05,
      "loss": 0.3368,
      "step": 3185
    },
    {
      "epoch": 0.17840743644305074,
      "grad_norm": 1.1332404613494873,
      "learning_rate": 1.592e-05,
      "loss": 0.4372,
      "step": 3186
    },
    {
      "epoch": 0.17846343375517976,
      "grad_norm": 1.2234429121017456,
      "learning_rate": 1.5925e-05,
      "loss": 0.4924,
      "step": 3187
    },
    {
      "epoch": 0.17851943106730878,
      "grad_norm": 1.1957119703292847,
      "learning_rate": 1.593e-05,
      "loss": 0.4729,
      "step": 3188
    },
    {
      "epoch": 0.1785754283794378,
      "grad_norm": 1.2825809717178345,
      "learning_rate": 1.5935e-05,
      "loss": 0.4555,
      "step": 3189
    },
    {
      "epoch": 0.17863142569156681,
      "grad_norm": 3.089897871017456,
      "learning_rate": 1.594e-05,
      "loss": 0.4048,
      "step": 3190
    },
    {
      "epoch": 0.17868742300369583,
      "grad_norm": 1.099593997001648,
      "learning_rate": 1.5945e-05,
      "loss": 0.3809,
      "step": 3191
    },
    {
      "epoch": 0.17874342031582485,
      "grad_norm": 1.1275124549865723,
      "learning_rate": 1.595e-05,
      "loss": 0.3926,
      "step": 3192
    },
    {
      "epoch": 0.17879941762795387,
      "grad_norm": 1.413540005683899,
      "learning_rate": 1.5955e-05,
      "loss": 0.4899,
      "step": 3193
    },
    {
      "epoch": 0.17885541494008286,
      "grad_norm": 1.2551677227020264,
      "learning_rate": 1.596e-05,
      "loss": 0.4831,
      "step": 3194
    },
    {
      "epoch": 0.17891141225221188,
      "grad_norm": 1.3640986680984497,
      "learning_rate": 1.5965e-05,
      "loss": 0.6391,
      "step": 3195
    },
    {
      "epoch": 0.1789674095643409,
      "grad_norm": 1.2376562356948853,
      "learning_rate": 1.597e-05,
      "loss": 0.4542,
      "step": 3196
    },
    {
      "epoch": 0.17902340687646992,
      "grad_norm": 1.3746341466903687,
      "learning_rate": 1.5975000000000002e-05,
      "loss": 0.5066,
      "step": 3197
    },
    {
      "epoch": 0.17907940418859894,
      "grad_norm": 0.9634360074996948,
      "learning_rate": 1.598e-05,
      "loss": 0.3555,
      "step": 3198
    },
    {
      "epoch": 0.17913540150072796,
      "grad_norm": 1.433107852935791,
      "learning_rate": 1.5985e-05,
      "loss": 0.4511,
      "step": 3199
    },
    {
      "epoch": 0.17919139881285698,
      "grad_norm": 1.2713160514831543,
      "learning_rate": 1.599e-05,
      "loss": 0.3922,
      "step": 3200
    },
    {
      "epoch": 0.179247396124986,
      "grad_norm": 1.3845282793045044,
      "learning_rate": 1.5995000000000002e-05,
      "loss": 0.5921,
      "step": 3201
    },
    {
      "epoch": 0.17930339343711502,
      "grad_norm": 1.1872634887695312,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.4316,
      "step": 3202
    },
    {
      "epoch": 0.17935939074924404,
      "grad_norm": 1.2111296653747559,
      "learning_rate": 1.6005e-05,
      "loss": 0.3921,
      "step": 3203
    },
    {
      "epoch": 0.17941538806137305,
      "grad_norm": 1.0769726037979126,
      "learning_rate": 1.601e-05,
      "loss": 0.3636,
      "step": 3204
    },
    {
      "epoch": 0.17947138537350207,
      "grad_norm": 1.1083983182907104,
      "learning_rate": 1.6014999999999998e-05,
      "loss": 0.3965,
      "step": 3205
    },
    {
      "epoch": 0.1795273826856311,
      "grad_norm": 1.2002310752868652,
      "learning_rate": 1.6020000000000002e-05,
      "loss": 0.369,
      "step": 3206
    },
    {
      "epoch": 0.1795833799977601,
      "grad_norm": 1.1439322233200073,
      "learning_rate": 1.6025e-05,
      "loss": 0.369,
      "step": 3207
    },
    {
      "epoch": 0.17963937730988913,
      "grad_norm": 1.3816206455230713,
      "learning_rate": 1.603e-05,
      "loss": 0.5202,
      "step": 3208
    },
    {
      "epoch": 0.17969537462201815,
      "grad_norm": 1.112925410270691,
      "learning_rate": 1.6035e-05,
      "loss": 0.3286,
      "step": 3209
    },
    {
      "epoch": 0.17975137193414717,
      "grad_norm": 1.315228819847107,
      "learning_rate": 1.604e-05,
      "loss": 0.4827,
      "step": 3210
    },
    {
      "epoch": 0.1798073692462762,
      "grad_norm": 2.721423387527466,
      "learning_rate": 1.6045000000000003e-05,
      "loss": 0.5205,
      "step": 3211
    },
    {
      "epoch": 0.1798633665584052,
      "grad_norm": 1.248304009437561,
      "learning_rate": 1.605e-05,
      "loss": 0.4123,
      "step": 3212
    },
    {
      "epoch": 0.17991936387053423,
      "grad_norm": 1.1161116361618042,
      "learning_rate": 1.6055e-05,
      "loss": 0.4418,
      "step": 3213
    },
    {
      "epoch": 0.17997536118266325,
      "grad_norm": 1.2112042903900146,
      "learning_rate": 1.606e-05,
      "loss": 0.5014,
      "step": 3214
    },
    {
      "epoch": 0.18003135849479224,
      "grad_norm": 1.0611644983291626,
      "learning_rate": 1.6065e-05,
      "loss": 0.3713,
      "step": 3215
    },
    {
      "epoch": 0.18008735580692126,
      "grad_norm": 1.357844352722168,
      "learning_rate": 1.607e-05,
      "loss": 0.5438,
      "step": 3216
    },
    {
      "epoch": 0.18014335311905028,
      "grad_norm": 0.9168925285339355,
      "learning_rate": 1.6075e-05,
      "loss": 0.3858,
      "step": 3217
    },
    {
      "epoch": 0.1801993504311793,
      "grad_norm": 1.3100770711898804,
      "learning_rate": 1.6080000000000002e-05,
      "loss": 0.5269,
      "step": 3218
    },
    {
      "epoch": 0.18025534774330831,
      "grad_norm": 1.1514203548431396,
      "learning_rate": 1.6085e-05,
      "loss": 0.4493,
      "step": 3219
    },
    {
      "epoch": 0.18031134505543733,
      "grad_norm": 1.176634669303894,
      "learning_rate": 1.609e-05,
      "loss": 0.4187,
      "step": 3220
    },
    {
      "epoch": 0.18036734236756635,
      "grad_norm": 1.0469194650650024,
      "learning_rate": 1.6095e-05,
      "loss": 0.43,
      "step": 3221
    },
    {
      "epoch": 0.18042333967969537,
      "grad_norm": 1.5243301391601562,
      "learning_rate": 1.6100000000000002e-05,
      "loss": 0.4707,
      "step": 3222
    },
    {
      "epoch": 0.1804793369918244,
      "grad_norm": 1.0046426057815552,
      "learning_rate": 1.6105e-05,
      "loss": 0.3883,
      "step": 3223
    },
    {
      "epoch": 0.1805353343039534,
      "grad_norm": 1.3109506368637085,
      "learning_rate": 1.611e-05,
      "loss": 0.5103,
      "step": 3224
    },
    {
      "epoch": 0.18059133161608243,
      "grad_norm": 1.446347951889038,
      "learning_rate": 1.6115e-05,
      "loss": 0.4054,
      "step": 3225
    },
    {
      "epoch": 0.18064732892821145,
      "grad_norm": 1.2103548049926758,
      "learning_rate": 1.612e-05,
      "loss": 0.5018,
      "step": 3226
    },
    {
      "epoch": 0.18070332624034047,
      "grad_norm": 1.17299222946167,
      "learning_rate": 1.6125000000000002e-05,
      "loss": 0.4382,
      "step": 3227
    },
    {
      "epoch": 0.1807593235524695,
      "grad_norm": 1.3501638174057007,
      "learning_rate": 1.613e-05,
      "loss": 0.5293,
      "step": 3228
    },
    {
      "epoch": 0.1808153208645985,
      "grad_norm": 1.369954228401184,
      "learning_rate": 1.6135e-05,
      "loss": 0.4372,
      "step": 3229
    },
    {
      "epoch": 0.18087131817672752,
      "grad_norm": 1.442979097366333,
      "learning_rate": 1.6139999999999998e-05,
      "loss": 0.3399,
      "step": 3230
    },
    {
      "epoch": 0.18092731548885654,
      "grad_norm": 1.3130812644958496,
      "learning_rate": 1.6145000000000002e-05,
      "loss": 0.3832,
      "step": 3231
    },
    {
      "epoch": 0.18098331280098556,
      "grad_norm": 1.3480647802352905,
      "learning_rate": 1.6150000000000003e-05,
      "loss": 0.4964,
      "step": 3232
    },
    {
      "epoch": 0.18103931011311458,
      "grad_norm": 1.1867642402648926,
      "learning_rate": 1.6155e-05,
      "loss": 0.3486,
      "step": 3233
    },
    {
      "epoch": 0.1810953074252436,
      "grad_norm": 1.3922346830368042,
      "learning_rate": 1.616e-05,
      "loss": 0.5325,
      "step": 3234
    },
    {
      "epoch": 0.1811513047373726,
      "grad_norm": 1.1357862949371338,
      "learning_rate": 1.6165e-05,
      "loss": 0.637,
      "step": 3235
    },
    {
      "epoch": 0.1812073020495016,
      "grad_norm": 1.2021393775939941,
      "learning_rate": 1.6170000000000003e-05,
      "loss": 0.4567,
      "step": 3236
    },
    {
      "epoch": 0.18126329936163063,
      "grad_norm": 1.1709843873977661,
      "learning_rate": 1.6175e-05,
      "loss": 0.4624,
      "step": 3237
    },
    {
      "epoch": 0.18131929667375965,
      "grad_norm": 1.1683293581008911,
      "learning_rate": 1.618e-05,
      "loss": 0.3118,
      "step": 3238
    },
    {
      "epoch": 0.18137529398588867,
      "grad_norm": 1.096571922302246,
      "learning_rate": 1.6185000000000002e-05,
      "loss": 0.3676,
      "step": 3239
    },
    {
      "epoch": 0.1814312912980177,
      "grad_norm": 1.2501559257507324,
      "learning_rate": 1.619e-05,
      "loss": 0.4443,
      "step": 3240
    },
    {
      "epoch": 0.1814872886101467,
      "grad_norm": 1.254988670349121,
      "learning_rate": 1.6195000000000003e-05,
      "loss": 0.5011,
      "step": 3241
    },
    {
      "epoch": 0.18154328592227573,
      "grad_norm": 1.2877148389816284,
      "learning_rate": 1.62e-05,
      "loss": 0.4825,
      "step": 3242
    },
    {
      "epoch": 0.18159928323440475,
      "grad_norm": 1.2468076944351196,
      "learning_rate": 1.6205e-05,
      "loss": 0.4051,
      "step": 3243
    },
    {
      "epoch": 0.18165528054653376,
      "grad_norm": 1.257940411567688,
      "learning_rate": 1.621e-05,
      "loss": 0.5026,
      "step": 3244
    },
    {
      "epoch": 0.18171127785866278,
      "grad_norm": 1.0413216352462769,
      "learning_rate": 1.6215e-05,
      "loss": 0.3557,
      "step": 3245
    },
    {
      "epoch": 0.1817672751707918,
      "grad_norm": 1.2333520650863647,
      "learning_rate": 1.622e-05,
      "loss": 0.4615,
      "step": 3246
    },
    {
      "epoch": 0.18182327248292082,
      "grad_norm": 1.1986799240112305,
      "learning_rate": 1.6225e-05,
      "loss": 0.4174,
      "step": 3247
    },
    {
      "epoch": 0.18187926979504984,
      "grad_norm": 1.9241962432861328,
      "learning_rate": 1.6230000000000002e-05,
      "loss": 0.4915,
      "step": 3248
    },
    {
      "epoch": 0.18193526710717886,
      "grad_norm": 1.2649354934692383,
      "learning_rate": 1.6235e-05,
      "loss": 0.654,
      "step": 3249
    },
    {
      "epoch": 0.18199126441930788,
      "grad_norm": 1.273526906967163,
      "learning_rate": 1.624e-05,
      "loss": 0.413,
      "step": 3250
    },
    {
      "epoch": 0.1820472617314369,
      "grad_norm": 1.438672661781311,
      "learning_rate": 1.6245e-05,
      "loss": 0.406,
      "step": 3251
    },
    {
      "epoch": 0.18210325904356592,
      "grad_norm": 1.3299928903579712,
      "learning_rate": 1.6250000000000002e-05,
      "loss": 0.4551,
      "step": 3252
    },
    {
      "epoch": 0.18215925635569494,
      "grad_norm": 1.1286715269088745,
      "learning_rate": 1.6255e-05,
      "loss": 0.437,
      "step": 3253
    },
    {
      "epoch": 0.18221525366782396,
      "grad_norm": 1.193117618560791,
      "learning_rate": 1.626e-05,
      "loss": 0.5152,
      "step": 3254
    },
    {
      "epoch": 0.18227125097995298,
      "grad_norm": 2.6552906036376953,
      "learning_rate": 1.6265e-05,
      "loss": 0.4532,
      "step": 3255
    },
    {
      "epoch": 0.18232724829208197,
      "grad_norm": 1.1225849390029907,
      "learning_rate": 1.6270000000000002e-05,
      "loss": 0.4489,
      "step": 3256
    },
    {
      "epoch": 0.182383245604211,
      "grad_norm": 1.1070002317428589,
      "learning_rate": 1.6275000000000003e-05,
      "loss": 0.2876,
      "step": 3257
    },
    {
      "epoch": 0.18243924291634,
      "grad_norm": 2.4725143909454346,
      "learning_rate": 1.628e-05,
      "loss": 0.3886,
      "step": 3258
    },
    {
      "epoch": 0.18249524022846902,
      "grad_norm": 1.3732391595840454,
      "learning_rate": 1.6285e-05,
      "loss": 0.5453,
      "step": 3259
    },
    {
      "epoch": 0.18255123754059804,
      "grad_norm": 1.3322089910507202,
      "learning_rate": 1.6289999999999998e-05,
      "loss": 0.4839,
      "step": 3260
    },
    {
      "epoch": 0.18260723485272706,
      "grad_norm": 1.0850098133087158,
      "learning_rate": 1.6295000000000002e-05,
      "loss": 0.4464,
      "step": 3261
    },
    {
      "epoch": 0.18266323216485608,
      "grad_norm": 1.1595327854156494,
      "learning_rate": 1.63e-05,
      "loss": 0.3385,
      "step": 3262
    },
    {
      "epoch": 0.1827192294769851,
      "grad_norm": 1.194762945175171,
      "learning_rate": 1.6305e-05,
      "loss": 0.4882,
      "step": 3263
    },
    {
      "epoch": 0.18277522678911412,
      "grad_norm": 1.1758840084075928,
      "learning_rate": 1.631e-05,
      "loss": 0.437,
      "step": 3264
    },
    {
      "epoch": 0.18283122410124314,
      "grad_norm": 1.1860482692718506,
      "learning_rate": 1.6315e-05,
      "loss": 0.6276,
      "step": 3265
    },
    {
      "epoch": 0.18288722141337216,
      "grad_norm": 1.4705352783203125,
      "learning_rate": 1.6320000000000003e-05,
      "loss": 0.4491,
      "step": 3266
    },
    {
      "epoch": 0.18294321872550118,
      "grad_norm": 1.955003023147583,
      "learning_rate": 1.6325e-05,
      "loss": 0.5621,
      "step": 3267
    },
    {
      "epoch": 0.1829992160376302,
      "grad_norm": 3.5246810913085938,
      "learning_rate": 1.633e-05,
      "loss": 0.4047,
      "step": 3268
    },
    {
      "epoch": 0.18305521334975922,
      "grad_norm": 1.4596385955810547,
      "learning_rate": 1.6335e-05,
      "loss": 0.4728,
      "step": 3269
    },
    {
      "epoch": 0.18311121066188824,
      "grad_norm": 1.2304860353469849,
      "learning_rate": 1.634e-05,
      "loss": 0.4617,
      "step": 3270
    },
    {
      "epoch": 0.18316720797401725,
      "grad_norm": 1.2931616306304932,
      "learning_rate": 1.6345000000000004e-05,
      "loss": 0.4159,
      "step": 3271
    },
    {
      "epoch": 0.18322320528614627,
      "grad_norm": 1.710740566253662,
      "learning_rate": 1.635e-05,
      "loss": 0.3767,
      "step": 3272
    },
    {
      "epoch": 0.1832792025982753,
      "grad_norm": 1.0282319784164429,
      "learning_rate": 1.6355000000000002e-05,
      "loss": 0.3537,
      "step": 3273
    },
    {
      "epoch": 0.1833351999104043,
      "grad_norm": 1.1052045822143555,
      "learning_rate": 1.636e-05,
      "loss": 0.5242,
      "step": 3274
    },
    {
      "epoch": 0.18339119722253333,
      "grad_norm": 1.0453317165374756,
      "learning_rate": 1.6365e-05,
      "loss": 0.4622,
      "step": 3275
    },
    {
      "epoch": 0.18344719453466235,
      "grad_norm": 1.3618289232254028,
      "learning_rate": 1.637e-05,
      "loss": 0.3347,
      "step": 3276
    },
    {
      "epoch": 0.18350319184679134,
      "grad_norm": 1.2078756093978882,
      "learning_rate": 1.6375e-05,
      "loss": 0.3666,
      "step": 3277
    },
    {
      "epoch": 0.18355918915892036,
      "grad_norm": 1.0806200504302979,
      "learning_rate": 1.6380000000000002e-05,
      "loss": 0.3669,
      "step": 3278
    },
    {
      "epoch": 0.18361518647104938,
      "grad_norm": 1.468646764755249,
      "learning_rate": 1.6385e-05,
      "loss": 0.4764,
      "step": 3279
    },
    {
      "epoch": 0.1836711837831784,
      "grad_norm": 1.1237543821334839,
      "learning_rate": 1.639e-05,
      "loss": 0.4137,
      "step": 3280
    },
    {
      "epoch": 0.18372718109530742,
      "grad_norm": 1.1732734441757202,
      "learning_rate": 1.6395e-05,
      "loss": 0.4237,
      "step": 3281
    },
    {
      "epoch": 0.18378317840743644,
      "grad_norm": 1.1597936153411865,
      "learning_rate": 1.6400000000000002e-05,
      "loss": 0.3797,
      "step": 3282
    },
    {
      "epoch": 0.18383917571956546,
      "grad_norm": 2.0686988830566406,
      "learning_rate": 1.6405e-05,
      "loss": 0.7229,
      "step": 3283
    },
    {
      "epoch": 0.18389517303169448,
      "grad_norm": 1.333763599395752,
      "learning_rate": 1.641e-05,
      "loss": 0.4357,
      "step": 3284
    },
    {
      "epoch": 0.1839511703438235,
      "grad_norm": 1.0924245119094849,
      "learning_rate": 1.6415e-05,
      "loss": 0.3907,
      "step": 3285
    },
    {
      "epoch": 0.1840071676559525,
      "grad_norm": 1.6450637578964233,
      "learning_rate": 1.6420000000000002e-05,
      "loss": 0.4214,
      "step": 3286
    },
    {
      "epoch": 0.18406316496808153,
      "grad_norm": 1.07072114944458,
      "learning_rate": 1.6425000000000003e-05,
      "loss": 0.4035,
      "step": 3287
    },
    {
      "epoch": 0.18411916228021055,
      "grad_norm": 1.138118028640747,
      "learning_rate": 1.643e-05,
      "loss": 0.3068,
      "step": 3288
    },
    {
      "epoch": 0.18417515959233957,
      "grad_norm": 1.1400213241577148,
      "learning_rate": 1.6435e-05,
      "loss": 0.4134,
      "step": 3289
    },
    {
      "epoch": 0.1842311569044686,
      "grad_norm": 1.229209303855896,
      "learning_rate": 1.644e-05,
      "loss": 0.6559,
      "step": 3290
    },
    {
      "epoch": 0.1842871542165976,
      "grad_norm": 1.224164605140686,
      "learning_rate": 1.6445000000000003e-05,
      "loss": 0.4453,
      "step": 3291
    },
    {
      "epoch": 0.18434315152872663,
      "grad_norm": 1.1119213104248047,
      "learning_rate": 1.645e-05,
      "loss": 0.3873,
      "step": 3292
    },
    {
      "epoch": 0.18439914884085565,
      "grad_norm": 1.228772759437561,
      "learning_rate": 1.6455e-05,
      "loss": 0.5373,
      "step": 3293
    },
    {
      "epoch": 0.18445514615298467,
      "grad_norm": 1.1952736377716064,
      "learning_rate": 1.646e-05,
      "loss": 0.4511,
      "step": 3294
    },
    {
      "epoch": 0.1845111434651137,
      "grad_norm": 1.3519757986068726,
      "learning_rate": 1.6465e-05,
      "loss": 0.5203,
      "step": 3295
    },
    {
      "epoch": 0.1845671407772427,
      "grad_norm": 1.3381385803222656,
      "learning_rate": 1.6470000000000003e-05,
      "loss": 0.4271,
      "step": 3296
    },
    {
      "epoch": 0.1846231380893717,
      "grad_norm": 1.0474534034729004,
      "learning_rate": 1.6475e-05,
      "loss": 0.4076,
      "step": 3297
    },
    {
      "epoch": 0.18467913540150072,
      "grad_norm": 1.3297818899154663,
      "learning_rate": 1.648e-05,
      "loss": 0.4738,
      "step": 3298
    },
    {
      "epoch": 0.18473513271362973,
      "grad_norm": 1.1280972957611084,
      "learning_rate": 1.6485e-05,
      "loss": 0.4053,
      "step": 3299
    },
    {
      "epoch": 0.18479113002575875,
      "grad_norm": 1.257645606994629,
      "learning_rate": 1.649e-05,
      "loss": 0.4157,
      "step": 3300
    },
    {
      "epoch": 0.18484712733788777,
      "grad_norm": 1.3946285247802734,
      "learning_rate": 1.6495e-05,
      "loss": 0.4413,
      "step": 3301
    },
    {
      "epoch": 0.1849031246500168,
      "grad_norm": 1.6009902954101562,
      "learning_rate": 1.65e-05,
      "loss": 0.6021,
      "step": 3302
    },
    {
      "epoch": 0.1849591219621458,
      "grad_norm": 1.2088947296142578,
      "learning_rate": 1.6505000000000002e-05,
      "loss": 0.4489,
      "step": 3303
    },
    {
      "epoch": 0.18501511927427483,
      "grad_norm": 1.2706345319747925,
      "learning_rate": 1.651e-05,
      "loss": 0.5796,
      "step": 3304
    },
    {
      "epoch": 0.18507111658640385,
      "grad_norm": 1.1603660583496094,
      "learning_rate": 1.6515e-05,
      "loss": 0.4107,
      "step": 3305
    },
    {
      "epoch": 0.18512711389853287,
      "grad_norm": 1.2458375692367554,
      "learning_rate": 1.652e-05,
      "loss": 0.4275,
      "step": 3306
    },
    {
      "epoch": 0.1851831112106619,
      "grad_norm": 1.3546783924102783,
      "learning_rate": 1.6525000000000002e-05,
      "loss": 0.4559,
      "step": 3307
    },
    {
      "epoch": 0.1852391085227909,
      "grad_norm": 1.1928722858428955,
      "learning_rate": 1.6530000000000003e-05,
      "loss": 0.455,
      "step": 3308
    },
    {
      "epoch": 0.18529510583491993,
      "grad_norm": 0.9799600839614868,
      "learning_rate": 1.6535e-05,
      "loss": 0.3385,
      "step": 3309
    },
    {
      "epoch": 0.18535110314704895,
      "grad_norm": 1.222354769706726,
      "learning_rate": 1.654e-05,
      "loss": 0.4966,
      "step": 3310
    },
    {
      "epoch": 0.18540710045917796,
      "grad_norm": 1.098609209060669,
      "learning_rate": 1.6545e-05,
      "loss": 0.3555,
      "step": 3311
    },
    {
      "epoch": 0.18546309777130698,
      "grad_norm": 1.2803016901016235,
      "learning_rate": 1.6550000000000002e-05,
      "loss": 0.4859,
      "step": 3312
    },
    {
      "epoch": 0.185519095083436,
      "grad_norm": 1.0965261459350586,
      "learning_rate": 1.6555e-05,
      "loss": 0.498,
      "step": 3313
    },
    {
      "epoch": 0.18557509239556502,
      "grad_norm": 1.2349803447723389,
      "learning_rate": 1.656e-05,
      "loss": 0.3836,
      "step": 3314
    },
    {
      "epoch": 0.18563108970769404,
      "grad_norm": 1.229987621307373,
      "learning_rate": 1.6565e-05,
      "loss": 0.4132,
      "step": 3315
    },
    {
      "epoch": 0.18568708701982306,
      "grad_norm": 1.1460660696029663,
      "learning_rate": 1.657e-05,
      "loss": 0.4971,
      "step": 3316
    },
    {
      "epoch": 0.18574308433195208,
      "grad_norm": 1.3114126920700073,
      "learning_rate": 1.6575000000000003e-05,
      "loss": 0.6052,
      "step": 3317
    },
    {
      "epoch": 0.18579908164408107,
      "grad_norm": 1.1734764575958252,
      "learning_rate": 1.658e-05,
      "loss": 0.4576,
      "step": 3318
    },
    {
      "epoch": 0.1858550789562101,
      "grad_norm": 1.2834408283233643,
      "learning_rate": 1.6585e-05,
      "loss": 0.4598,
      "step": 3319
    },
    {
      "epoch": 0.1859110762683391,
      "grad_norm": 1.509110689163208,
      "learning_rate": 1.659e-05,
      "loss": 0.4597,
      "step": 3320
    },
    {
      "epoch": 0.18596707358046813,
      "grad_norm": 1.226487398147583,
      "learning_rate": 1.6595e-05,
      "loss": 0.3691,
      "step": 3321
    },
    {
      "epoch": 0.18602307089259715,
      "grad_norm": 1.135722041130066,
      "learning_rate": 1.66e-05,
      "loss": 0.5141,
      "step": 3322
    },
    {
      "epoch": 0.18607906820472617,
      "grad_norm": 1.3855825662612915,
      "learning_rate": 1.6605e-05,
      "loss": 0.4158,
      "step": 3323
    },
    {
      "epoch": 0.18613506551685519,
      "grad_norm": 1.1391615867614746,
      "learning_rate": 1.6610000000000002e-05,
      "loss": 0.4244,
      "step": 3324
    },
    {
      "epoch": 0.1861910628289842,
      "grad_norm": 1.1902307271957397,
      "learning_rate": 1.6615e-05,
      "loss": 0.4648,
      "step": 3325
    },
    {
      "epoch": 0.18624706014111322,
      "grad_norm": 1.1830710172653198,
      "learning_rate": 1.662e-05,
      "loss": 0.4394,
      "step": 3326
    },
    {
      "epoch": 0.18630305745324224,
      "grad_norm": 1.219680905342102,
      "learning_rate": 1.6625e-05,
      "loss": 0.3814,
      "step": 3327
    },
    {
      "epoch": 0.18635905476537126,
      "grad_norm": 1.0361963510513306,
      "learning_rate": 1.6630000000000002e-05,
      "loss": 0.3575,
      "step": 3328
    },
    {
      "epoch": 0.18641505207750028,
      "grad_norm": 1.4810428619384766,
      "learning_rate": 1.6635e-05,
      "loss": 0.5595,
      "step": 3329
    },
    {
      "epoch": 0.1864710493896293,
      "grad_norm": 1.082762598991394,
      "learning_rate": 1.664e-05,
      "loss": 0.5685,
      "step": 3330
    },
    {
      "epoch": 0.18652704670175832,
      "grad_norm": 1.3045333623886108,
      "learning_rate": 1.6645e-05,
      "loss": 0.6185,
      "step": 3331
    },
    {
      "epoch": 0.18658304401388734,
      "grad_norm": 1.1538302898406982,
      "learning_rate": 1.665e-05,
      "loss": 0.4312,
      "step": 3332
    },
    {
      "epoch": 0.18663904132601636,
      "grad_norm": 1.336602807044983,
      "learning_rate": 1.6655000000000002e-05,
      "loss": 0.3832,
      "step": 3333
    },
    {
      "epoch": 0.18669503863814538,
      "grad_norm": 1.2260713577270508,
      "learning_rate": 1.666e-05,
      "loss": 0.3917,
      "step": 3334
    },
    {
      "epoch": 0.1867510359502744,
      "grad_norm": 1.078220248222351,
      "learning_rate": 1.6665e-05,
      "loss": 0.4161,
      "step": 3335
    },
    {
      "epoch": 0.18680703326240342,
      "grad_norm": 1.0990285873413086,
      "learning_rate": 1.6669999999999998e-05,
      "loss": 0.4115,
      "step": 3336
    },
    {
      "epoch": 0.18686303057453243,
      "grad_norm": 1.2535604238510132,
      "learning_rate": 1.6675000000000002e-05,
      "loss": 0.4842,
      "step": 3337
    },
    {
      "epoch": 0.18691902788666145,
      "grad_norm": 1.0767483711242676,
      "learning_rate": 1.668e-05,
      "loss": 0.3885,
      "step": 3338
    },
    {
      "epoch": 0.18697502519879045,
      "grad_norm": 1.1079645156860352,
      "learning_rate": 1.6685e-05,
      "loss": 0.4516,
      "step": 3339
    },
    {
      "epoch": 0.18703102251091946,
      "grad_norm": 1.3449302911758423,
      "learning_rate": 1.669e-05,
      "loss": 0.4821,
      "step": 3340
    },
    {
      "epoch": 0.18708701982304848,
      "grad_norm": 1.2095441818237305,
      "learning_rate": 1.6695e-05,
      "loss": 0.3834,
      "step": 3341
    },
    {
      "epoch": 0.1871430171351775,
      "grad_norm": 1.2906239032745361,
      "learning_rate": 1.6700000000000003e-05,
      "loss": 0.7066,
      "step": 3342
    },
    {
      "epoch": 0.18719901444730652,
      "grad_norm": 1.172589659690857,
      "learning_rate": 1.6705e-05,
      "loss": 0.4537,
      "step": 3343
    },
    {
      "epoch": 0.18725501175943554,
      "grad_norm": 1.1754515171051025,
      "learning_rate": 1.671e-05,
      "loss": 0.4595,
      "step": 3344
    },
    {
      "epoch": 0.18731100907156456,
      "grad_norm": 1.052424669265747,
      "learning_rate": 1.6715000000000002e-05,
      "loss": 0.3947,
      "step": 3345
    },
    {
      "epoch": 0.18736700638369358,
      "grad_norm": 1.1512207984924316,
      "learning_rate": 1.672e-05,
      "loss": 0.3732,
      "step": 3346
    },
    {
      "epoch": 0.1874230036958226,
      "grad_norm": 1.1828744411468506,
      "learning_rate": 1.6725000000000003e-05,
      "loss": 0.2665,
      "step": 3347
    },
    {
      "epoch": 0.18747900100795162,
      "grad_norm": 1.670555830001831,
      "learning_rate": 1.673e-05,
      "loss": 0.5061,
      "step": 3348
    },
    {
      "epoch": 0.18753499832008064,
      "grad_norm": 1.185208797454834,
      "learning_rate": 1.6735e-05,
      "loss": 0.3772,
      "step": 3349
    },
    {
      "epoch": 0.18759099563220966,
      "grad_norm": 1.3466157913208008,
      "learning_rate": 1.674e-05,
      "loss": 0.5246,
      "step": 3350
    },
    {
      "epoch": 0.18764699294433868,
      "grad_norm": 1.374879240989685,
      "learning_rate": 1.6745e-05,
      "loss": 0.5011,
      "step": 3351
    },
    {
      "epoch": 0.1877029902564677,
      "grad_norm": 1.1384507417678833,
      "learning_rate": 1.675e-05,
      "loss": 0.3486,
      "step": 3352
    },
    {
      "epoch": 0.1877589875685967,
      "grad_norm": 1.1417534351348877,
      "learning_rate": 1.6755e-05,
      "loss": 0.3765,
      "step": 3353
    },
    {
      "epoch": 0.18781498488072573,
      "grad_norm": 1.1946250200271606,
      "learning_rate": 1.6760000000000002e-05,
      "loss": 0.3828,
      "step": 3354
    },
    {
      "epoch": 0.18787098219285475,
      "grad_norm": 1.1479038000106812,
      "learning_rate": 1.6765e-05,
      "loss": 0.4193,
      "step": 3355
    },
    {
      "epoch": 0.18792697950498377,
      "grad_norm": 1.2469680309295654,
      "learning_rate": 1.677e-05,
      "loss": 0.3967,
      "step": 3356
    },
    {
      "epoch": 0.1879829768171128,
      "grad_norm": 1.0604054927825928,
      "learning_rate": 1.6775e-05,
      "loss": 0.3472,
      "step": 3357
    },
    {
      "epoch": 0.1880389741292418,
      "grad_norm": 1.4340620040893555,
      "learning_rate": 1.6780000000000002e-05,
      "loss": 0.4665,
      "step": 3358
    },
    {
      "epoch": 0.1880949714413708,
      "grad_norm": 1.5655089616775513,
      "learning_rate": 1.6785e-05,
      "loss": 0.4712,
      "step": 3359
    },
    {
      "epoch": 0.18815096875349982,
      "grad_norm": 1.0947039127349854,
      "learning_rate": 1.679e-05,
      "loss": 0.4547,
      "step": 3360
    },
    {
      "epoch": 0.18820696606562884,
      "grad_norm": 1.078403353691101,
      "learning_rate": 1.6795e-05,
      "loss": 0.4219,
      "step": 3361
    },
    {
      "epoch": 0.18826296337775786,
      "grad_norm": 1.148415446281433,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.4868,
      "step": 3362
    },
    {
      "epoch": 0.18831896068988688,
      "grad_norm": 1.633833646774292,
      "learning_rate": 1.6805000000000003e-05,
      "loss": 0.5315,
      "step": 3363
    },
    {
      "epoch": 0.1883749580020159,
      "grad_norm": 1.1244282722473145,
      "learning_rate": 1.681e-05,
      "loss": 0.3964,
      "step": 3364
    },
    {
      "epoch": 0.18843095531414492,
      "grad_norm": 1.420272946357727,
      "learning_rate": 1.6815e-05,
      "loss": 0.4208,
      "step": 3365
    },
    {
      "epoch": 0.18848695262627393,
      "grad_norm": 1.3265632390975952,
      "learning_rate": 1.6819999999999998e-05,
      "loss": 0.3931,
      "step": 3366
    },
    {
      "epoch": 0.18854294993840295,
      "grad_norm": 1.3832519054412842,
      "learning_rate": 1.6825000000000002e-05,
      "loss": 0.4967,
      "step": 3367
    },
    {
      "epoch": 0.18859894725053197,
      "grad_norm": 1.217711329460144,
      "learning_rate": 1.683e-05,
      "loss": 0.4278,
      "step": 3368
    },
    {
      "epoch": 0.188654944562661,
      "grad_norm": 1.2732466459274292,
      "learning_rate": 1.6835e-05,
      "loss": 0.5072,
      "step": 3369
    },
    {
      "epoch": 0.18871094187479,
      "grad_norm": 1.0388083457946777,
      "learning_rate": 1.684e-05,
      "loss": 0.438,
      "step": 3370
    },
    {
      "epoch": 0.18876693918691903,
      "grad_norm": 1.2363613843917847,
      "learning_rate": 1.6845e-05,
      "loss": 0.4776,
      "step": 3371
    },
    {
      "epoch": 0.18882293649904805,
      "grad_norm": 1.2334297895431519,
      "learning_rate": 1.6850000000000003e-05,
      "loss": 0.4249,
      "step": 3372
    },
    {
      "epoch": 0.18887893381117707,
      "grad_norm": 1.1314977407455444,
      "learning_rate": 1.6855e-05,
      "loss": 0.3967,
      "step": 3373
    },
    {
      "epoch": 0.1889349311233061,
      "grad_norm": 1.0096185207366943,
      "learning_rate": 1.686e-05,
      "loss": 0.3685,
      "step": 3374
    },
    {
      "epoch": 0.1889909284354351,
      "grad_norm": 1.3609317541122437,
      "learning_rate": 1.6865e-05,
      "loss": 0.4632,
      "step": 3375
    },
    {
      "epoch": 0.18904692574756413,
      "grad_norm": 1.0902884006500244,
      "learning_rate": 1.687e-05,
      "loss": 0.3116,
      "step": 3376
    },
    {
      "epoch": 0.18910292305969315,
      "grad_norm": 1.6312155723571777,
      "learning_rate": 1.6875000000000004e-05,
      "loss": 0.4591,
      "step": 3377
    },
    {
      "epoch": 0.18915892037182216,
      "grad_norm": 1.2797203063964844,
      "learning_rate": 1.688e-05,
      "loss": 0.3855,
      "step": 3378
    },
    {
      "epoch": 0.18921491768395118,
      "grad_norm": 1.2569507360458374,
      "learning_rate": 1.6885000000000002e-05,
      "loss": 0.4277,
      "step": 3379
    },
    {
      "epoch": 0.18927091499608018,
      "grad_norm": 1.1022034883499146,
      "learning_rate": 1.689e-05,
      "loss": 0.3475,
      "step": 3380
    },
    {
      "epoch": 0.1893269123082092,
      "grad_norm": 1.084113359451294,
      "learning_rate": 1.6895e-05,
      "loss": 0.415,
      "step": 3381
    },
    {
      "epoch": 0.1893829096203382,
      "grad_norm": 1.3184095621109009,
      "learning_rate": 1.69e-05,
      "loss": 0.3926,
      "step": 3382
    },
    {
      "epoch": 0.18943890693246723,
      "grad_norm": 1.2320349216461182,
      "learning_rate": 1.6905e-05,
      "loss": 0.4394,
      "step": 3383
    },
    {
      "epoch": 0.18949490424459625,
      "grad_norm": 2.3696770668029785,
      "learning_rate": 1.6910000000000002e-05,
      "loss": 0.3514,
      "step": 3384
    },
    {
      "epoch": 0.18955090155672527,
      "grad_norm": 1.1212445497512817,
      "learning_rate": 1.6915e-05,
      "loss": 0.6016,
      "step": 3385
    },
    {
      "epoch": 0.1896068988688543,
      "grad_norm": 1.0768115520477295,
      "learning_rate": 1.692e-05,
      "loss": 0.4472,
      "step": 3386
    },
    {
      "epoch": 0.1896628961809833,
      "grad_norm": 1.5229803323745728,
      "learning_rate": 1.6925e-05,
      "loss": 0.4502,
      "step": 3387
    },
    {
      "epoch": 0.18971889349311233,
      "grad_norm": 1.1422045230865479,
      "learning_rate": 1.6930000000000002e-05,
      "loss": 0.2921,
      "step": 3388
    },
    {
      "epoch": 0.18977489080524135,
      "grad_norm": 1.3426090478897095,
      "learning_rate": 1.6935e-05,
      "loss": 0.4891,
      "step": 3389
    },
    {
      "epoch": 0.18983088811737037,
      "grad_norm": 1.250209093093872,
      "learning_rate": 1.694e-05,
      "loss": 0.446,
      "step": 3390
    },
    {
      "epoch": 0.18988688542949939,
      "grad_norm": 1.2611923217773438,
      "learning_rate": 1.6945e-05,
      "loss": 0.4895,
      "step": 3391
    },
    {
      "epoch": 0.1899428827416284,
      "grad_norm": 1.2335516214370728,
      "learning_rate": 1.6950000000000002e-05,
      "loss": 0.4183,
      "step": 3392
    },
    {
      "epoch": 0.18999888005375742,
      "grad_norm": 1.0964800119400024,
      "learning_rate": 1.6955000000000003e-05,
      "loss": 0.4134,
      "step": 3393
    },
    {
      "epoch": 0.19005487736588644,
      "grad_norm": 1.8114066123962402,
      "learning_rate": 1.696e-05,
      "loss": 0.5505,
      "step": 3394
    },
    {
      "epoch": 0.19011087467801546,
      "grad_norm": 1.3944091796875,
      "learning_rate": 1.6965e-05,
      "loss": 0.6571,
      "step": 3395
    },
    {
      "epoch": 0.19016687199014448,
      "grad_norm": 1.537448763847351,
      "learning_rate": 1.697e-05,
      "loss": 0.486,
      "step": 3396
    },
    {
      "epoch": 0.1902228693022735,
      "grad_norm": 1.3557907342910767,
      "learning_rate": 1.6975000000000003e-05,
      "loss": 0.4855,
      "step": 3397
    },
    {
      "epoch": 0.19027886661440252,
      "grad_norm": 1.2067703008651733,
      "learning_rate": 1.698e-05,
      "loss": 0.5486,
      "step": 3398
    },
    {
      "epoch": 0.19033486392653154,
      "grad_norm": 1.1005481481552124,
      "learning_rate": 1.6985e-05,
      "loss": 0.4072,
      "step": 3399
    },
    {
      "epoch": 0.19039086123866056,
      "grad_norm": 1.3335630893707275,
      "learning_rate": 1.699e-05,
      "loss": 0.4883,
      "step": 3400
    },
    {
      "epoch": 0.19044685855078955,
      "grad_norm": 1.3925936222076416,
      "learning_rate": 1.6995e-05,
      "loss": 0.4764,
      "step": 3401
    },
    {
      "epoch": 0.19050285586291857,
      "grad_norm": 2.9519593715667725,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 0.5681,
      "step": 3402
    },
    {
      "epoch": 0.1905588531750476,
      "grad_norm": 1.0862815380096436,
      "learning_rate": 1.7005e-05,
      "loss": 0.3684,
      "step": 3403
    },
    {
      "epoch": 0.1906148504871766,
      "grad_norm": 1.3254729509353638,
      "learning_rate": 1.701e-05,
      "loss": 0.4242,
      "step": 3404
    },
    {
      "epoch": 0.19067084779930563,
      "grad_norm": 1.2939419746398926,
      "learning_rate": 1.7015e-05,
      "loss": 0.6702,
      "step": 3405
    },
    {
      "epoch": 0.19072684511143465,
      "grad_norm": 1.1743439435958862,
      "learning_rate": 1.702e-05,
      "loss": 0.4808,
      "step": 3406
    },
    {
      "epoch": 0.19078284242356366,
      "grad_norm": 1.2036492824554443,
      "learning_rate": 1.7025e-05,
      "loss": 0.3705,
      "step": 3407
    },
    {
      "epoch": 0.19083883973569268,
      "grad_norm": 1.1415599584579468,
      "learning_rate": 1.703e-05,
      "loss": 0.5025,
      "step": 3408
    },
    {
      "epoch": 0.1908948370478217,
      "grad_norm": 1.1346678733825684,
      "learning_rate": 1.7035000000000002e-05,
      "loss": 0.5311,
      "step": 3409
    },
    {
      "epoch": 0.19095083435995072,
      "grad_norm": 1.3246341943740845,
      "learning_rate": 1.704e-05,
      "loss": 0.5128,
      "step": 3410
    },
    {
      "epoch": 0.19100683167207974,
      "grad_norm": 1.2443675994873047,
      "learning_rate": 1.7045e-05,
      "loss": 0.6092,
      "step": 3411
    },
    {
      "epoch": 0.19106282898420876,
      "grad_norm": 1.1285314559936523,
      "learning_rate": 1.705e-05,
      "loss": 0.3824,
      "step": 3412
    },
    {
      "epoch": 0.19111882629633778,
      "grad_norm": 1.1717637777328491,
      "learning_rate": 1.7055000000000002e-05,
      "loss": 0.446,
      "step": 3413
    },
    {
      "epoch": 0.1911748236084668,
      "grad_norm": 1.0986918210983276,
      "learning_rate": 1.706e-05,
      "loss": 0.4009,
      "step": 3414
    },
    {
      "epoch": 0.19123082092059582,
      "grad_norm": 1.2700912952423096,
      "learning_rate": 1.7065e-05,
      "loss": 0.4518,
      "step": 3415
    },
    {
      "epoch": 0.19128681823272484,
      "grad_norm": 1.0021355152130127,
      "learning_rate": 1.707e-05,
      "loss": 0.3094,
      "step": 3416
    },
    {
      "epoch": 0.19134281554485386,
      "grad_norm": 1.3428599834442139,
      "learning_rate": 1.7075e-05,
      "loss": 0.3574,
      "step": 3417
    },
    {
      "epoch": 0.19139881285698288,
      "grad_norm": 1.2496545314788818,
      "learning_rate": 1.7080000000000002e-05,
      "loss": 0.4016,
      "step": 3418
    },
    {
      "epoch": 0.1914548101691119,
      "grad_norm": 1.1841208934783936,
      "learning_rate": 1.7085e-05,
      "loss": 0.4462,
      "step": 3419
    },
    {
      "epoch": 0.1915108074812409,
      "grad_norm": 1.4078991413116455,
      "learning_rate": 1.709e-05,
      "loss": 0.4315,
      "step": 3420
    },
    {
      "epoch": 0.1915668047933699,
      "grad_norm": 1.3215162754058838,
      "learning_rate": 1.7095e-05,
      "loss": 0.4588,
      "step": 3421
    },
    {
      "epoch": 0.19162280210549892,
      "grad_norm": 1.2057838439941406,
      "learning_rate": 1.7100000000000002e-05,
      "loss": 0.4078,
      "step": 3422
    },
    {
      "epoch": 0.19167879941762794,
      "grad_norm": 1.2553930282592773,
      "learning_rate": 1.7105000000000003e-05,
      "loss": 0.3858,
      "step": 3423
    },
    {
      "epoch": 0.19173479672975696,
      "grad_norm": 1.4046380519866943,
      "learning_rate": 1.711e-05,
      "loss": 0.6379,
      "step": 3424
    },
    {
      "epoch": 0.19179079404188598,
      "grad_norm": 1.4144824743270874,
      "learning_rate": 1.7115e-05,
      "loss": 0.4791,
      "step": 3425
    },
    {
      "epoch": 0.191846791354015,
      "grad_norm": 1.804913878440857,
      "learning_rate": 1.712e-05,
      "loss": 0.5213,
      "step": 3426
    },
    {
      "epoch": 0.19190278866614402,
      "grad_norm": 1.282642126083374,
      "learning_rate": 1.7125000000000003e-05,
      "loss": 0.5367,
      "step": 3427
    },
    {
      "epoch": 0.19195878597827304,
      "grad_norm": 1.3015738725662231,
      "learning_rate": 1.713e-05,
      "loss": 0.4267,
      "step": 3428
    },
    {
      "epoch": 0.19201478329040206,
      "grad_norm": 1.5885628461837769,
      "learning_rate": 1.7135e-05,
      "loss": 0.5507,
      "step": 3429
    },
    {
      "epoch": 0.19207078060253108,
      "grad_norm": 1.3165570497512817,
      "learning_rate": 1.7140000000000002e-05,
      "loss": 0.4275,
      "step": 3430
    },
    {
      "epoch": 0.1921267779146601,
      "grad_norm": 1.299818992614746,
      "learning_rate": 1.7145e-05,
      "loss": 0.5218,
      "step": 3431
    },
    {
      "epoch": 0.19218277522678912,
      "grad_norm": 1.2481540441513062,
      "learning_rate": 1.7150000000000004e-05,
      "loss": 0.4117,
      "step": 3432
    },
    {
      "epoch": 0.19223877253891813,
      "grad_norm": 1.2244377136230469,
      "learning_rate": 1.7155e-05,
      "loss": 0.4895,
      "step": 3433
    },
    {
      "epoch": 0.19229476985104715,
      "grad_norm": 1.1141343116760254,
      "learning_rate": 1.7160000000000002e-05,
      "loss": 0.488,
      "step": 3434
    },
    {
      "epoch": 0.19235076716317617,
      "grad_norm": 1.5054800510406494,
      "learning_rate": 1.7165e-05,
      "loss": 0.5853,
      "step": 3435
    },
    {
      "epoch": 0.1924067644753052,
      "grad_norm": 1.1070597171783447,
      "learning_rate": 1.717e-05,
      "loss": 0.4004,
      "step": 3436
    },
    {
      "epoch": 0.1924627617874342,
      "grad_norm": 1.2049047946929932,
      "learning_rate": 1.7175e-05,
      "loss": 0.437,
      "step": 3437
    },
    {
      "epoch": 0.19251875909956323,
      "grad_norm": 1.857879638671875,
      "learning_rate": 1.718e-05,
      "loss": 0.4563,
      "step": 3438
    },
    {
      "epoch": 0.19257475641169225,
      "grad_norm": 1.1342371702194214,
      "learning_rate": 1.7185000000000002e-05,
      "loss": 0.3864,
      "step": 3439
    },
    {
      "epoch": 0.19263075372382127,
      "grad_norm": 1.1872518062591553,
      "learning_rate": 1.719e-05,
      "loss": 0.5765,
      "step": 3440
    },
    {
      "epoch": 0.1926867510359503,
      "grad_norm": 1.1592533588409424,
      "learning_rate": 1.7195e-05,
      "loss": 0.3613,
      "step": 3441
    },
    {
      "epoch": 0.19274274834807928,
      "grad_norm": 1.2872483730316162,
      "learning_rate": 1.7199999999999998e-05,
      "loss": 0.4437,
      "step": 3442
    },
    {
      "epoch": 0.1927987456602083,
      "grad_norm": NaN,
      "learning_rate": 1.7199999999999998e-05,
      "loss": 0.2918,
      "step": 3443
    },
    {
      "epoch": 0.19285474297233732,
      "grad_norm": 1.2049381732940674,
      "learning_rate": 1.7205000000000002e-05,
      "loss": 0.383,
      "step": 3444
    },
    {
      "epoch": 0.19291074028446634,
      "grad_norm": 1.1656842231750488,
      "learning_rate": 1.721e-05,
      "loss": 0.4182,
      "step": 3445
    },
    {
      "epoch": 0.19296673759659536,
      "grad_norm": 1.127423644065857,
      "learning_rate": 1.7215e-05,
      "loss": 0.4357,
      "step": 3446
    },
    {
      "epoch": 0.19302273490872437,
      "grad_norm": 1.9054218530654907,
      "learning_rate": 1.722e-05,
      "loss": 0.3881,
      "step": 3447
    },
    {
      "epoch": 0.1930787322208534,
      "grad_norm": 1.3814903497695923,
      "learning_rate": 1.7225e-05,
      "loss": 0.6128,
      "step": 3448
    },
    {
      "epoch": 0.1931347295329824,
      "grad_norm": 1.302217960357666,
      "learning_rate": 1.7230000000000003e-05,
      "loss": 0.4546,
      "step": 3449
    },
    {
      "epoch": 0.19319072684511143,
      "grad_norm": 1.1936508417129517,
      "learning_rate": 1.7235e-05,
      "loss": 0.4362,
      "step": 3450
    },
    {
      "epoch": 0.19324672415724045,
      "grad_norm": 1.2464027404785156,
      "learning_rate": 1.724e-05,
      "loss": 0.4663,
      "step": 3451
    },
    {
      "epoch": 0.19330272146936947,
      "grad_norm": 1.2320516109466553,
      "learning_rate": 1.7245e-05,
      "loss": 0.497,
      "step": 3452
    },
    {
      "epoch": 0.1933587187814985,
      "grad_norm": 1.2153738737106323,
      "learning_rate": 1.725e-05,
      "loss": 0.5814,
      "step": 3453
    },
    {
      "epoch": 0.1934147160936275,
      "grad_norm": 1.3801615238189697,
      "learning_rate": 1.7255000000000003e-05,
      "loss": 0.478,
      "step": 3454
    },
    {
      "epoch": 0.19347071340575653,
      "grad_norm": 1.465854525566101,
      "learning_rate": 1.726e-05,
      "loss": 0.4406,
      "step": 3455
    },
    {
      "epoch": 0.19352671071788555,
      "grad_norm": 2.0683443546295166,
      "learning_rate": 1.7265e-05,
      "loss": 0.5539,
      "step": 3456
    },
    {
      "epoch": 0.19358270803001457,
      "grad_norm": 1.6779159307479858,
      "learning_rate": 1.727e-05,
      "loss": 0.5283,
      "step": 3457
    },
    {
      "epoch": 0.19363870534214359,
      "grad_norm": 1.1865346431732178,
      "learning_rate": 1.7275e-05,
      "loss": 0.5156,
      "step": 3458
    },
    {
      "epoch": 0.1936947026542726,
      "grad_norm": 1.2769807577133179,
      "learning_rate": 1.728e-05,
      "loss": 0.5397,
      "step": 3459
    },
    {
      "epoch": 0.19375069996640162,
      "grad_norm": 1.1956596374511719,
      "learning_rate": 1.7285e-05,
      "loss": 0.3929,
      "step": 3460
    },
    {
      "epoch": 0.19380669727853064,
      "grad_norm": 1.1267244815826416,
      "learning_rate": 1.7290000000000002e-05,
      "loss": 0.3968,
      "step": 3461
    },
    {
      "epoch": 0.19386269459065966,
      "grad_norm": 1.2932826280593872,
      "learning_rate": 1.7295e-05,
      "loss": 0.4719,
      "step": 3462
    },
    {
      "epoch": 0.19391869190278865,
      "grad_norm": 1.1786478757858276,
      "learning_rate": 1.73e-05,
      "loss": 0.4781,
      "step": 3463
    },
    {
      "epoch": 0.19397468921491767,
      "grad_norm": 1.1676743030548096,
      "learning_rate": 1.7305e-05,
      "loss": 0.3853,
      "step": 3464
    },
    {
      "epoch": 0.1940306865270467,
      "grad_norm": 1.2847338914871216,
      "learning_rate": 1.7310000000000002e-05,
      "loss": 0.5865,
      "step": 3465
    },
    {
      "epoch": 0.1940866838391757,
      "grad_norm": 1.2821464538574219,
      "learning_rate": 1.7315e-05,
      "loss": 0.4091,
      "step": 3466
    },
    {
      "epoch": 0.19414268115130473,
      "grad_norm": 1.1550301313400269,
      "learning_rate": 1.732e-05,
      "loss": 0.4035,
      "step": 3467
    },
    {
      "epoch": 0.19419867846343375,
      "grad_norm": 1.3662973642349243,
      "learning_rate": 1.7325e-05,
      "loss": 0.4234,
      "step": 3468
    },
    {
      "epoch": 0.19425467577556277,
      "grad_norm": 1.301316261291504,
      "learning_rate": 1.7330000000000002e-05,
      "loss": 0.4344,
      "step": 3469
    },
    {
      "epoch": 0.1943106730876918,
      "grad_norm": 1.1564757823944092,
      "learning_rate": 1.7335000000000003e-05,
      "loss": 0.5202,
      "step": 3470
    },
    {
      "epoch": 0.1943666703998208,
      "grad_norm": 1.756524682044983,
      "learning_rate": 1.734e-05,
      "loss": 0.5682,
      "step": 3471
    },
    {
      "epoch": 0.19442266771194983,
      "grad_norm": 1.184188961982727,
      "learning_rate": 1.7345e-05,
      "loss": 0.4627,
      "step": 3472
    },
    {
      "epoch": 0.19447866502407885,
      "grad_norm": 1.332406997680664,
      "learning_rate": 1.7349999999999998e-05,
      "loss": 0.4546,
      "step": 3473
    },
    {
      "epoch": 0.19453466233620786,
      "grad_norm": 1.2890806198120117,
      "learning_rate": 1.7355000000000002e-05,
      "loss": 0.4304,
      "step": 3474
    },
    {
      "epoch": 0.19459065964833688,
      "grad_norm": 1.150179386138916,
      "learning_rate": 1.736e-05,
      "loss": 0.4157,
      "step": 3475
    },
    {
      "epoch": 0.1946466569604659,
      "grad_norm": 1.432388186454773,
      "learning_rate": 1.7365e-05,
      "loss": 0.5514,
      "step": 3476
    },
    {
      "epoch": 0.19470265427259492,
      "grad_norm": 1.2167162895202637,
      "learning_rate": 1.737e-05,
      "loss": 0.4673,
      "step": 3477
    },
    {
      "epoch": 0.19475865158472394,
      "grad_norm": 1.4046660661697388,
      "learning_rate": 1.7375e-05,
      "loss": 0.5461,
      "step": 3478
    },
    {
      "epoch": 0.19481464889685296,
      "grad_norm": 1.2196552753448486,
      "learning_rate": 1.7380000000000003e-05,
      "loss": 0.425,
      "step": 3479
    },
    {
      "epoch": 0.19487064620898198,
      "grad_norm": 1.8962860107421875,
      "learning_rate": 1.7385e-05,
      "loss": 0.5941,
      "step": 3480
    },
    {
      "epoch": 0.194926643521111,
      "grad_norm": 1.1011854410171509,
      "learning_rate": 1.739e-05,
      "loss": 0.4297,
      "step": 3481
    },
    {
      "epoch": 0.19498264083324002,
      "grad_norm": 1.023834466934204,
      "learning_rate": 1.7395e-05,
      "loss": 0.3723,
      "step": 3482
    },
    {
      "epoch": 0.195038638145369,
      "grad_norm": 1.4861829280853271,
      "learning_rate": 1.74e-05,
      "loss": 0.4788,
      "step": 3483
    },
    {
      "epoch": 0.19509463545749803,
      "grad_norm": 1.1437004804611206,
      "learning_rate": 1.7405e-05,
      "loss": 0.364,
      "step": 3484
    },
    {
      "epoch": 0.19515063276962705,
      "grad_norm": 1.8231146335601807,
      "learning_rate": 1.741e-05,
      "loss": 0.5312,
      "step": 3485
    },
    {
      "epoch": 0.19520663008175607,
      "grad_norm": 1.5955363512039185,
      "learning_rate": 1.7415000000000002e-05,
      "loss": 0.4753,
      "step": 3486
    },
    {
      "epoch": 0.19526262739388509,
      "grad_norm": 1.3428277969360352,
      "learning_rate": 1.742e-05,
      "loss": 0.4872,
      "step": 3487
    },
    {
      "epoch": 0.1953186247060141,
      "grad_norm": 1.1213557720184326,
      "learning_rate": 1.7425e-05,
      "loss": 0.3974,
      "step": 3488
    },
    {
      "epoch": 0.19537462201814312,
      "grad_norm": 1.1225180625915527,
      "learning_rate": 1.743e-05,
      "loss": 0.3424,
      "step": 3489
    },
    {
      "epoch": 0.19543061933027214,
      "grad_norm": 1.5477772951126099,
      "learning_rate": 1.7435e-05,
      "loss": 0.6466,
      "step": 3490
    },
    {
      "epoch": 0.19548661664240116,
      "grad_norm": 1.1222776174545288,
      "learning_rate": 1.7440000000000002e-05,
      "loss": 0.4342,
      "step": 3491
    },
    {
      "epoch": 0.19554261395453018,
      "grad_norm": 1.058448314666748,
      "learning_rate": 1.7445e-05,
      "loss": 0.3581,
      "step": 3492
    },
    {
      "epoch": 0.1955986112666592,
      "grad_norm": 1.4802191257476807,
      "learning_rate": 1.745e-05,
      "loss": 0.4271,
      "step": 3493
    },
    {
      "epoch": 0.19565460857878822,
      "grad_norm": 1.29079008102417,
      "learning_rate": 1.7455e-05,
      "loss": 0.4136,
      "step": 3494
    },
    {
      "epoch": 0.19571060589091724,
      "grad_norm": 1.099254846572876,
      "learning_rate": 1.7460000000000002e-05,
      "loss": 0.3975,
      "step": 3495
    },
    {
      "epoch": 0.19576660320304626,
      "grad_norm": 1.9378408193588257,
      "learning_rate": 1.7465e-05,
      "loss": 0.5841,
      "step": 3496
    },
    {
      "epoch": 0.19582260051517528,
      "grad_norm": 1.5394763946533203,
      "learning_rate": 1.747e-05,
      "loss": 0.4734,
      "step": 3497
    },
    {
      "epoch": 0.1958785978273043,
      "grad_norm": 1.3304824829101562,
      "learning_rate": 1.7475e-05,
      "loss": 0.3832,
      "step": 3498
    },
    {
      "epoch": 0.19593459513943332,
      "grad_norm": 1.273561954498291,
      "learning_rate": 1.7480000000000002e-05,
      "loss": 0.5094,
      "step": 3499
    },
    {
      "epoch": 0.19599059245156233,
      "grad_norm": 1.340970516204834,
      "learning_rate": 1.7485000000000003e-05,
      "loss": 0.4204,
      "step": 3500
    },
    {
      "epoch": 0.19604658976369135,
      "grad_norm": 1.1625409126281738,
      "learning_rate": 1.749e-05,
      "loss": 0.4155,
      "step": 3501
    },
    {
      "epoch": 0.19610258707582037,
      "grad_norm": 0.9949480295181274,
      "learning_rate": 1.7495e-05,
      "loss": 0.4649,
      "step": 3502
    },
    {
      "epoch": 0.1961585843879494,
      "grad_norm": 1.329535961151123,
      "learning_rate": 1.75e-05,
      "loss": 0.5627,
      "step": 3503
    },
    {
      "epoch": 0.19621458170007838,
      "grad_norm": 1.1956413984298706,
      "learning_rate": 1.7505000000000003e-05,
      "loss": 0.3473,
      "step": 3504
    },
    {
      "epoch": 0.1962705790122074,
      "grad_norm": 1.3843954801559448,
      "learning_rate": 1.751e-05,
      "loss": 0.5833,
      "step": 3505
    },
    {
      "epoch": 0.19632657632433642,
      "grad_norm": 1.2624551057815552,
      "learning_rate": 1.7515e-05,
      "loss": 0.3751,
      "step": 3506
    },
    {
      "epoch": 0.19638257363646544,
      "grad_norm": 2.117694854736328,
      "learning_rate": 1.752e-05,
      "loss": 0.3899,
      "step": 3507
    },
    {
      "epoch": 0.19643857094859446,
      "grad_norm": 1.484227180480957,
      "learning_rate": 1.7525e-05,
      "loss": 0.4206,
      "step": 3508
    },
    {
      "epoch": 0.19649456826072348,
      "grad_norm": 1.1634652614593506,
      "learning_rate": 1.7530000000000003e-05,
      "loss": 0.3729,
      "step": 3509
    },
    {
      "epoch": 0.1965505655728525,
      "grad_norm": 1.2138384580612183,
      "learning_rate": 1.7535e-05,
      "loss": 0.3716,
      "step": 3510
    },
    {
      "epoch": 0.19660656288498152,
      "grad_norm": 1.0454835891723633,
      "learning_rate": 1.754e-05,
      "loss": 0.3648,
      "step": 3511
    },
    {
      "epoch": 0.19666256019711054,
      "grad_norm": 1.5565184354782104,
      "learning_rate": 1.7545e-05,
      "loss": 0.5028,
      "step": 3512
    },
    {
      "epoch": 0.19671855750923956,
      "grad_norm": 1.8158458471298218,
      "learning_rate": 1.755e-05,
      "loss": 0.4712,
      "step": 3513
    },
    {
      "epoch": 0.19677455482136857,
      "grad_norm": 0.9922124743461609,
      "learning_rate": 1.7555e-05,
      "loss": 0.4019,
      "step": 3514
    },
    {
      "epoch": 0.1968305521334976,
      "grad_norm": 1.0502554178237915,
      "learning_rate": 1.756e-05,
      "loss": 0.3467,
      "step": 3515
    },
    {
      "epoch": 0.1968865494456266,
      "grad_norm": 4.462704181671143,
      "learning_rate": 1.7565000000000002e-05,
      "loss": 0.5373,
      "step": 3516
    },
    {
      "epoch": 0.19694254675775563,
      "grad_norm": 1.2560471296310425,
      "learning_rate": 1.757e-05,
      "loss": 0.4816,
      "step": 3517
    },
    {
      "epoch": 0.19699854406988465,
      "grad_norm": 1.1651909351348877,
      "learning_rate": 1.7575e-05,
      "loss": 0.4926,
      "step": 3518
    },
    {
      "epoch": 0.19705454138201367,
      "grad_norm": 1.382475733757019,
      "learning_rate": 1.758e-05,
      "loss": 0.4079,
      "step": 3519
    },
    {
      "epoch": 0.1971105386941427,
      "grad_norm": 1.362947940826416,
      "learning_rate": 1.7585000000000002e-05,
      "loss": 0.6844,
      "step": 3520
    },
    {
      "epoch": 0.1971665360062717,
      "grad_norm": 1.4244499206542969,
      "learning_rate": 1.759e-05,
      "loss": 0.3927,
      "step": 3521
    },
    {
      "epoch": 0.19722253331840073,
      "grad_norm": 1.4018924236297607,
      "learning_rate": 1.7595e-05,
      "loss": 0.4086,
      "step": 3522
    },
    {
      "epoch": 0.19727853063052975,
      "grad_norm": 1.0669325590133667,
      "learning_rate": 1.76e-05,
      "loss": 0.3776,
      "step": 3523
    },
    {
      "epoch": 0.19733452794265877,
      "grad_norm": 1.1211423873901367,
      "learning_rate": 1.7605000000000002e-05,
      "loss": 0.3565,
      "step": 3524
    },
    {
      "epoch": 0.19739052525478776,
      "grad_norm": 1.3023756742477417,
      "learning_rate": 1.7610000000000002e-05,
      "loss": 0.4846,
      "step": 3525
    },
    {
      "epoch": 0.19744652256691678,
      "grad_norm": 1.5412284135818481,
      "learning_rate": 1.7615e-05,
      "loss": 0.6275,
      "step": 3526
    },
    {
      "epoch": 0.1975025198790458,
      "grad_norm": 1.2327158451080322,
      "learning_rate": 1.762e-05,
      "loss": 0.3694,
      "step": 3527
    },
    {
      "epoch": 0.19755851719117482,
      "grad_norm": 1.0797861814498901,
      "learning_rate": 1.7625e-05,
      "loss": 0.3528,
      "step": 3528
    },
    {
      "epoch": 0.19761451450330383,
      "grad_norm": 1.3560792207717896,
      "learning_rate": 1.7630000000000002e-05,
      "loss": 0.5039,
      "step": 3529
    },
    {
      "epoch": 0.19767051181543285,
      "grad_norm": 1.2430589199066162,
      "learning_rate": 1.7635000000000003e-05,
      "loss": 0.4804,
      "step": 3530
    },
    {
      "epoch": 0.19772650912756187,
      "grad_norm": 1.2607671022415161,
      "learning_rate": 1.764e-05,
      "loss": 0.4166,
      "step": 3531
    },
    {
      "epoch": 0.1977825064396909,
      "grad_norm": 1.0509237051010132,
      "learning_rate": 1.7645e-05,
      "loss": 0.3564,
      "step": 3532
    },
    {
      "epoch": 0.1978385037518199,
      "grad_norm": 1.234286904335022,
      "learning_rate": 1.765e-05,
      "loss": 0.4267,
      "step": 3533
    },
    {
      "epoch": 0.19789450106394893,
      "grad_norm": 1.2905102968215942,
      "learning_rate": 1.7655000000000003e-05,
      "loss": 0.425,
      "step": 3534
    },
    {
      "epoch": 0.19795049837607795,
      "grad_norm": 1.1756755113601685,
      "learning_rate": 1.766e-05,
      "loss": 0.3897,
      "step": 3535
    },
    {
      "epoch": 0.19800649568820697,
      "grad_norm": 1.2536176443099976,
      "learning_rate": 1.7665e-05,
      "loss": 0.4033,
      "step": 3536
    },
    {
      "epoch": 0.198062493000336,
      "grad_norm": 1.1823030710220337,
      "learning_rate": 1.7670000000000002e-05,
      "loss": 0.4084,
      "step": 3537
    },
    {
      "epoch": 0.198118490312465,
      "grad_norm": 1.295543909072876,
      "learning_rate": 1.7675e-05,
      "loss": 0.3927,
      "step": 3538
    },
    {
      "epoch": 0.19817448762459403,
      "grad_norm": 1.1278620958328247,
      "learning_rate": 1.7680000000000004e-05,
      "loss": 0.4166,
      "step": 3539
    },
    {
      "epoch": 0.19823048493672304,
      "grad_norm": 1.3221131563186646,
      "learning_rate": 1.7685e-05,
      "loss": 0.4792,
      "step": 3540
    },
    {
      "epoch": 0.19828648224885206,
      "grad_norm": 1.1316249370574951,
      "learning_rate": 1.7690000000000002e-05,
      "loss": 0.6107,
      "step": 3541
    },
    {
      "epoch": 0.19834247956098108,
      "grad_norm": 1.281059741973877,
      "learning_rate": 1.7695e-05,
      "loss": 0.4603,
      "step": 3542
    },
    {
      "epoch": 0.1983984768731101,
      "grad_norm": 1.519079327583313,
      "learning_rate": 1.77e-05,
      "loss": 0.9461,
      "step": 3543
    },
    {
      "epoch": 0.19845447418523912,
      "grad_norm": 1.1021186113357544,
      "learning_rate": 1.7705e-05,
      "loss": 0.4054,
      "step": 3544
    },
    {
      "epoch": 0.1985104714973681,
      "grad_norm": 1.2137590646743774,
      "learning_rate": 1.771e-05,
      "loss": 0.3485,
      "step": 3545
    },
    {
      "epoch": 0.19856646880949713,
      "grad_norm": 1.8117284774780273,
      "learning_rate": 1.7715000000000002e-05,
      "loss": 0.4094,
      "step": 3546
    },
    {
      "epoch": 0.19862246612162615,
      "grad_norm": 1.588281273841858,
      "learning_rate": 1.772e-05,
      "loss": 0.4485,
      "step": 3547
    },
    {
      "epoch": 0.19867846343375517,
      "grad_norm": 1.4057315587997437,
      "learning_rate": 1.7725e-05,
      "loss": 0.5001,
      "step": 3548
    },
    {
      "epoch": 0.1987344607458842,
      "grad_norm": 1.0868313312530518,
      "learning_rate": 1.773e-05,
      "loss": 0.42,
      "step": 3549
    },
    {
      "epoch": 0.1987904580580132,
      "grad_norm": 1.1870468854904175,
      "learning_rate": 1.7735000000000002e-05,
      "loss": 0.4441,
      "step": 3550
    },
    {
      "epoch": 0.19884645537014223,
      "grad_norm": 1.0263683795928955,
      "learning_rate": 1.774e-05,
      "loss": 0.4931,
      "step": 3551
    },
    {
      "epoch": 0.19890245268227125,
      "grad_norm": 1.1085573434829712,
      "learning_rate": 1.7745e-05,
      "loss": 0.4928,
      "step": 3552
    },
    {
      "epoch": 0.19895844999440027,
      "grad_norm": 1.0907138586044312,
      "learning_rate": 1.775e-05,
      "loss": 0.481,
      "step": 3553
    },
    {
      "epoch": 0.19901444730652929,
      "grad_norm": 1.4030075073242188,
      "learning_rate": 1.7755000000000002e-05,
      "loss": 0.6967,
      "step": 3554
    },
    {
      "epoch": 0.1990704446186583,
      "grad_norm": 1.2095900774002075,
      "learning_rate": 1.7760000000000003e-05,
      "loss": 0.3484,
      "step": 3555
    },
    {
      "epoch": 0.19912644193078732,
      "grad_norm": 1.1291601657867432,
      "learning_rate": 1.7765e-05,
      "loss": 0.3844,
      "step": 3556
    },
    {
      "epoch": 0.19918243924291634,
      "grad_norm": 1.094610571861267,
      "learning_rate": 1.777e-05,
      "loss": 0.4705,
      "step": 3557
    },
    {
      "epoch": 0.19923843655504536,
      "grad_norm": 1.1452468633651733,
      "learning_rate": 1.7775e-05,
      "loss": 0.4597,
      "step": 3558
    },
    {
      "epoch": 0.19929443386717438,
      "grad_norm": 1.232744812965393,
      "learning_rate": 1.7780000000000003e-05,
      "loss": 0.5608,
      "step": 3559
    },
    {
      "epoch": 0.1993504311793034,
      "grad_norm": 1.1906026601791382,
      "learning_rate": 1.7785e-05,
      "loss": 0.3942,
      "step": 3560
    },
    {
      "epoch": 0.19940642849143242,
      "grad_norm": 0.969206690788269,
      "learning_rate": 1.779e-05,
      "loss": 0.2845,
      "step": 3561
    },
    {
      "epoch": 0.19946242580356144,
      "grad_norm": 1.2132002115249634,
      "learning_rate": 1.7795e-05,
      "loss": 0.4097,
      "step": 3562
    },
    {
      "epoch": 0.19951842311569046,
      "grad_norm": 1.3831021785736084,
      "learning_rate": 1.78e-05,
      "loss": 0.4446,
      "step": 3563
    },
    {
      "epoch": 0.19957442042781948,
      "grad_norm": 1.3817284107208252,
      "learning_rate": 1.7805000000000003e-05,
      "loss": 0.3767,
      "step": 3564
    },
    {
      "epoch": 0.1996304177399485,
      "grad_norm": 1.3107645511627197,
      "learning_rate": 1.781e-05,
      "loss": 0.4725,
      "step": 3565
    },
    {
      "epoch": 0.1996864150520775,
      "grad_norm": 1.0882208347320557,
      "learning_rate": 1.7815e-05,
      "loss": 0.4093,
      "step": 3566
    },
    {
      "epoch": 0.1997424123642065,
      "grad_norm": 1.105525255203247,
      "learning_rate": 1.7820000000000002e-05,
      "loss": 0.3445,
      "step": 3567
    },
    {
      "epoch": 0.19979840967633553,
      "grad_norm": 1.329720377922058,
      "learning_rate": 1.7825e-05,
      "loss": 0.4393,
      "step": 3568
    },
    {
      "epoch": 0.19985440698846454,
      "grad_norm": 1.0822575092315674,
      "learning_rate": 1.783e-05,
      "loss": 0.3839,
      "step": 3569
    },
    {
      "epoch": 0.19991040430059356,
      "grad_norm": 1.0894161462783813,
      "learning_rate": 1.7835e-05,
      "loss": 0.344,
      "step": 3570
    },
    {
      "epoch": 0.19996640161272258,
      "grad_norm": 1.1637275218963623,
      "learning_rate": 1.7840000000000002e-05,
      "loss": 0.4248,
      "step": 3571
    },
    {
      "epoch": 0.2000223989248516,
      "grad_norm": 1.4276413917541504,
      "learning_rate": 1.7845e-05,
      "loss": 0.4069,
      "step": 3572
    },
    {
      "epoch": 0.20007839623698062,
      "grad_norm": 1.1614675521850586,
      "learning_rate": 1.785e-05,
      "loss": 0.3726,
      "step": 3573
    },
    {
      "epoch": 0.20013439354910964,
      "grad_norm": 1.1339586973190308,
      "learning_rate": 1.7855e-05,
      "loss": 0.6507,
      "step": 3574
    },
    {
      "epoch": 0.20019039086123866,
      "grad_norm": 1.5252258777618408,
      "learning_rate": 1.7860000000000002e-05,
      "loss": 0.4611,
      "step": 3575
    },
    {
      "epoch": 0.20024638817336768,
      "grad_norm": 1.1716711521148682,
      "learning_rate": 1.7865000000000003e-05,
      "loss": 0.3847,
      "step": 3576
    },
    {
      "epoch": 0.2003023854854967,
      "grad_norm": 1.0965520143508911,
      "learning_rate": 1.787e-05,
      "loss": 0.4186,
      "step": 3577
    },
    {
      "epoch": 0.20035838279762572,
      "grad_norm": 1.3718756437301636,
      "learning_rate": 1.7875e-05,
      "loss": 0.5508,
      "step": 3578
    },
    {
      "epoch": 0.20041438010975474,
      "grad_norm": 1.2677582502365112,
      "learning_rate": 1.7879999999999998e-05,
      "loss": 0.4108,
      "step": 3579
    },
    {
      "epoch": 0.20047037742188376,
      "grad_norm": 1.3012787103652954,
      "learning_rate": 1.7885000000000002e-05,
      "loss": 0.5334,
      "step": 3580
    },
    {
      "epoch": 0.20052637473401277,
      "grad_norm": 1.152340292930603,
      "learning_rate": 1.789e-05,
      "loss": 0.3269,
      "step": 3581
    },
    {
      "epoch": 0.2005823720461418,
      "grad_norm": 1.3426048755645752,
      "learning_rate": 1.7895e-05,
      "loss": 0.5069,
      "step": 3582
    },
    {
      "epoch": 0.2006383693582708,
      "grad_norm": 1.2388484477996826,
      "learning_rate": 1.79e-05,
      "loss": 0.544,
      "step": 3583
    },
    {
      "epoch": 0.20069436667039983,
      "grad_norm": 1.2543814182281494,
      "learning_rate": 1.7905e-05,
      "loss": 0.4034,
      "step": 3584
    },
    {
      "epoch": 0.20075036398252885,
      "grad_norm": 1.060953140258789,
      "learning_rate": 1.7910000000000003e-05,
      "loss": 0.4458,
      "step": 3585
    },
    {
      "epoch": 0.20080636129465787,
      "grad_norm": 1.3129549026489258,
      "learning_rate": 1.7915e-05,
      "loss": 0.5294,
      "step": 3586
    },
    {
      "epoch": 0.20086235860678686,
      "grad_norm": 1.1931536197662354,
      "learning_rate": 1.792e-05,
      "loss": 0.4113,
      "step": 3587
    },
    {
      "epoch": 0.20091835591891588,
      "grad_norm": 1.254481315612793,
      "learning_rate": 1.7925e-05,
      "loss": 0.4261,
      "step": 3588
    },
    {
      "epoch": 0.2009743532310449,
      "grad_norm": 1.1662412881851196,
      "learning_rate": 1.793e-05,
      "loss": 0.348,
      "step": 3589
    },
    {
      "epoch": 0.20103035054317392,
      "grad_norm": 1.0701323747634888,
      "learning_rate": 1.7935e-05,
      "loss": 0.4522,
      "step": 3590
    },
    {
      "epoch": 0.20108634785530294,
      "grad_norm": 1.308609127998352,
      "learning_rate": 1.794e-05,
      "loss": 0.4527,
      "step": 3591
    },
    {
      "epoch": 0.20114234516743196,
      "grad_norm": 1.1434340476989746,
      "learning_rate": 1.7945000000000002e-05,
      "loss": 0.3729,
      "step": 3592
    },
    {
      "epoch": 0.20119834247956098,
      "grad_norm": 1.10639488697052,
      "learning_rate": 1.795e-05,
      "loss": 0.402,
      "step": 3593
    },
    {
      "epoch": 0.20125433979169,
      "grad_norm": 2.03780460357666,
      "learning_rate": 1.7955e-05,
      "loss": 0.4185,
      "step": 3594
    },
    {
      "epoch": 0.20131033710381901,
      "grad_norm": 1.222267508506775,
      "learning_rate": 1.796e-05,
      "loss": 0.4905,
      "step": 3595
    },
    {
      "epoch": 0.20136633441594803,
      "grad_norm": 1.2427198886871338,
      "learning_rate": 1.7965e-05,
      "loss": 0.4275,
      "step": 3596
    },
    {
      "epoch": 0.20142233172807705,
      "grad_norm": 1.4295467138290405,
      "learning_rate": 1.797e-05,
      "loss": 0.4698,
      "step": 3597
    },
    {
      "epoch": 0.20147832904020607,
      "grad_norm": 1.0984424352645874,
      "learning_rate": 1.7975e-05,
      "loss": 0.4006,
      "step": 3598
    },
    {
      "epoch": 0.2015343263523351,
      "grad_norm": 1.083338975906372,
      "learning_rate": 1.798e-05,
      "loss": 0.4357,
      "step": 3599
    },
    {
      "epoch": 0.2015903236644641,
      "grad_norm": 1.1363531351089478,
      "learning_rate": 1.7985e-05,
      "loss": 0.4991,
      "step": 3600
    },
    {
      "epoch": 0.20164632097659313,
      "grad_norm": 1.5071083307266235,
      "learning_rate": 1.7990000000000002e-05,
      "loss": 0.4102,
      "step": 3601
    },
    {
      "epoch": 0.20170231828872215,
      "grad_norm": 1.1894534826278687,
      "learning_rate": 1.7995e-05,
      "loss": 0.3997,
      "step": 3602
    },
    {
      "epoch": 0.20175831560085117,
      "grad_norm": 1.1389291286468506,
      "learning_rate": 1.8e-05,
      "loss": 0.5243,
      "step": 3603
    },
    {
      "epoch": 0.2018143129129802,
      "grad_norm": 1.225938081741333,
      "learning_rate": 1.8005e-05,
      "loss": 0.3694,
      "step": 3604
    },
    {
      "epoch": 0.2018703102251092,
      "grad_norm": 1.445648431777954,
      "learning_rate": 1.8010000000000002e-05,
      "loss": 0.3522,
      "step": 3605
    },
    {
      "epoch": 0.20192630753723823,
      "grad_norm": 1.427200436592102,
      "learning_rate": 1.8015000000000003e-05,
      "loss": 0.5047,
      "step": 3606
    },
    {
      "epoch": 0.20198230484936722,
      "grad_norm": 1.5686529874801636,
      "learning_rate": 1.802e-05,
      "loss": 0.6975,
      "step": 3607
    },
    {
      "epoch": 0.20203830216149624,
      "grad_norm": 1.1451834440231323,
      "learning_rate": 1.8025e-05,
      "loss": 0.4568,
      "step": 3608
    },
    {
      "epoch": 0.20209429947362526,
      "grad_norm": 1.3459903001785278,
      "learning_rate": 1.803e-05,
      "loss": 0.5214,
      "step": 3609
    },
    {
      "epoch": 0.20215029678575427,
      "grad_norm": 1.2125604152679443,
      "learning_rate": 1.8035000000000003e-05,
      "loss": 0.3619,
      "step": 3610
    },
    {
      "epoch": 0.2022062940978833,
      "grad_norm": 0.9751166105270386,
      "learning_rate": 1.804e-05,
      "loss": 0.5279,
      "step": 3611
    },
    {
      "epoch": 0.2022622914100123,
      "grad_norm": 1.3576555252075195,
      "learning_rate": 1.8045e-05,
      "loss": 0.4481,
      "step": 3612
    },
    {
      "epoch": 0.20231828872214133,
      "grad_norm": 2.4999935626983643,
      "learning_rate": 1.805e-05,
      "loss": 0.4593,
      "step": 3613
    },
    {
      "epoch": 0.20237428603427035,
      "grad_norm": 0.9820563197135925,
      "learning_rate": 1.8055e-05,
      "loss": 0.3046,
      "step": 3614
    },
    {
      "epoch": 0.20243028334639937,
      "grad_norm": 1.426029920578003,
      "learning_rate": 1.8060000000000003e-05,
      "loss": 0.4896,
      "step": 3615
    },
    {
      "epoch": 0.2024862806585284,
      "grad_norm": 1.2235302925109863,
      "learning_rate": 1.8065e-05,
      "loss": 0.3901,
      "step": 3616
    },
    {
      "epoch": 0.2025422779706574,
      "grad_norm": 1.4202797412872314,
      "learning_rate": 1.807e-05,
      "loss": 0.4293,
      "step": 3617
    },
    {
      "epoch": 0.20259827528278643,
      "grad_norm": 1.1697888374328613,
      "learning_rate": 1.8075e-05,
      "loss": 0.4151,
      "step": 3618
    },
    {
      "epoch": 0.20265427259491545,
      "grad_norm": 1.2334222793579102,
      "learning_rate": 1.808e-05,
      "loss": 0.4224,
      "step": 3619
    },
    {
      "epoch": 0.20271026990704447,
      "grad_norm": 1.3797261714935303,
      "learning_rate": 1.8085e-05,
      "loss": 0.488,
      "step": 3620
    },
    {
      "epoch": 0.20276626721917348,
      "grad_norm": 0.9996871948242188,
      "learning_rate": 1.809e-05,
      "loss": 0.3075,
      "step": 3621
    },
    {
      "epoch": 0.2028222645313025,
      "grad_norm": 1.453730583190918,
      "learning_rate": 1.8095000000000002e-05,
      "loss": 0.394,
      "step": 3622
    },
    {
      "epoch": 0.20287826184343152,
      "grad_norm": 1.36322820186615,
      "learning_rate": 1.81e-05,
      "loss": 0.3731,
      "step": 3623
    },
    {
      "epoch": 0.20293425915556054,
      "grad_norm": 1.352899432182312,
      "learning_rate": 1.8105e-05,
      "loss": 0.5815,
      "step": 3624
    },
    {
      "epoch": 0.20299025646768956,
      "grad_norm": 1.3098249435424805,
      "learning_rate": 1.811e-05,
      "loss": 0.5279,
      "step": 3625
    },
    {
      "epoch": 0.20304625377981858,
      "grad_norm": 1.2320561408996582,
      "learning_rate": 1.8115000000000002e-05,
      "loss": 0.4337,
      "step": 3626
    },
    {
      "epoch": 0.2031022510919476,
      "grad_norm": 1.6283172369003296,
      "learning_rate": 1.812e-05,
      "loss": 0.3657,
      "step": 3627
    },
    {
      "epoch": 0.2031582484040766,
      "grad_norm": 1.1894358396530151,
      "learning_rate": 1.8125e-05,
      "loss": 0.4657,
      "step": 3628
    },
    {
      "epoch": 0.2032142457162056,
      "grad_norm": 1.1958972215652466,
      "learning_rate": 1.813e-05,
      "loss": 0.4009,
      "step": 3629
    },
    {
      "epoch": 0.20327024302833463,
      "grad_norm": 1.127791404724121,
      "learning_rate": 1.8135000000000002e-05,
      "loss": 0.4063,
      "step": 3630
    },
    {
      "epoch": 0.20332624034046365,
      "grad_norm": 1.1458134651184082,
      "learning_rate": 1.8140000000000003e-05,
      "loss": 0.4677,
      "step": 3631
    },
    {
      "epoch": 0.20338223765259267,
      "grad_norm": 1.2265417575836182,
      "learning_rate": 1.8145e-05,
      "loss": 0.4175,
      "step": 3632
    },
    {
      "epoch": 0.2034382349647217,
      "grad_norm": 1.3974417448043823,
      "learning_rate": 1.815e-05,
      "loss": 0.4813,
      "step": 3633
    },
    {
      "epoch": 0.2034942322768507,
      "grad_norm": 1.2176661491394043,
      "learning_rate": 1.8154999999999998e-05,
      "loss": 0.4138,
      "step": 3634
    },
    {
      "epoch": 0.20355022958897973,
      "grad_norm": 1.1092934608459473,
      "learning_rate": 1.8160000000000002e-05,
      "loss": 0.4615,
      "step": 3635
    },
    {
      "epoch": 0.20360622690110874,
      "grad_norm": 1.218031883239746,
      "learning_rate": 1.8165000000000003e-05,
      "loss": 0.5333,
      "step": 3636
    },
    {
      "epoch": 0.20366222421323776,
      "grad_norm": 1.1778126955032349,
      "learning_rate": 1.817e-05,
      "loss": 0.4052,
      "step": 3637
    },
    {
      "epoch": 0.20371822152536678,
      "grad_norm": 1.1710573434829712,
      "learning_rate": 1.8175e-05,
      "loss": 0.3803,
      "step": 3638
    },
    {
      "epoch": 0.2037742188374958,
      "grad_norm": 1.1477383375167847,
      "learning_rate": 1.818e-05,
      "loss": 0.4362,
      "step": 3639
    },
    {
      "epoch": 0.20383021614962482,
      "grad_norm": 1.2956640720367432,
      "learning_rate": 1.8185000000000003e-05,
      "loss": 0.4826,
      "step": 3640
    },
    {
      "epoch": 0.20388621346175384,
      "grad_norm": 1.3115049600601196,
      "learning_rate": 1.819e-05,
      "loss": 0.56,
      "step": 3641
    },
    {
      "epoch": 0.20394221077388286,
      "grad_norm": 1.4746919870376587,
      "learning_rate": 1.8195e-05,
      "loss": 0.4711,
      "step": 3642
    },
    {
      "epoch": 0.20399820808601188,
      "grad_norm": 1.923878788948059,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 0.5686,
      "step": 3643
    },
    {
      "epoch": 0.2040542053981409,
      "grad_norm": 1.1285382509231567,
      "learning_rate": 1.8205e-05,
      "loss": 0.4183,
      "step": 3644
    },
    {
      "epoch": 0.20411020271026992,
      "grad_norm": 1.5769509077072144,
      "learning_rate": 1.8210000000000004e-05,
      "loss": 0.4979,
      "step": 3645
    },
    {
      "epoch": 0.20416620002239894,
      "grad_norm": 1.2398681640625,
      "learning_rate": 1.8215e-05,
      "loss": 0.4363,
      "step": 3646
    },
    {
      "epoch": 0.20422219733452796,
      "grad_norm": 1.1484941244125366,
      "learning_rate": 1.8220000000000002e-05,
      "loss": 0.391,
      "step": 3647
    },
    {
      "epoch": 0.20427819464665697,
      "grad_norm": 1.3237804174423218,
      "learning_rate": 1.8225e-05,
      "loss": 0.6652,
      "step": 3648
    },
    {
      "epoch": 0.20433419195878597,
      "grad_norm": 1.0015528202056885,
      "learning_rate": 1.823e-05,
      "loss": 0.3516,
      "step": 3649
    },
    {
      "epoch": 0.20439018927091498,
      "grad_norm": 1.793995976448059,
      "learning_rate": 1.8235e-05,
      "loss": 0.7392,
      "step": 3650
    },
    {
      "epoch": 0.204446186583044,
      "grad_norm": 1.415885329246521,
      "learning_rate": 1.824e-05,
      "loss": 0.3627,
      "step": 3651
    },
    {
      "epoch": 0.20450218389517302,
      "grad_norm": 1.268105149269104,
      "learning_rate": 1.8245000000000002e-05,
      "loss": 0.4041,
      "step": 3652
    },
    {
      "epoch": 0.20455818120730204,
      "grad_norm": 1.5501044988632202,
      "learning_rate": 1.825e-05,
      "loss": 0.4487,
      "step": 3653
    },
    {
      "epoch": 0.20461417851943106,
      "grad_norm": 2.0684571266174316,
      "learning_rate": 1.8255e-05,
      "loss": 0.4629,
      "step": 3654
    },
    {
      "epoch": 0.20467017583156008,
      "grad_norm": 1.2597365379333496,
      "learning_rate": 1.826e-05,
      "loss": 0.4913,
      "step": 3655
    },
    {
      "epoch": 0.2047261731436891,
      "grad_norm": 0.9715607166290283,
      "learning_rate": 1.8265000000000002e-05,
      "loss": 0.5863,
      "step": 3656
    },
    {
      "epoch": 0.20478217045581812,
      "grad_norm": 1.2507647275924683,
      "learning_rate": 1.827e-05,
      "loss": 0.4449,
      "step": 3657
    },
    {
      "epoch": 0.20483816776794714,
      "grad_norm": 1.1853148937225342,
      "learning_rate": 1.8275e-05,
      "loss": 0.4402,
      "step": 3658
    },
    {
      "epoch": 0.20489416508007616,
      "grad_norm": 1.6511564254760742,
      "learning_rate": 1.828e-05,
      "loss": 0.4787,
      "step": 3659
    },
    {
      "epoch": 0.20495016239220518,
      "grad_norm": 1.2150965929031372,
      "learning_rate": 1.8285000000000002e-05,
      "loss": 0.4149,
      "step": 3660
    },
    {
      "epoch": 0.2050061597043342,
      "grad_norm": 1.2021454572677612,
      "learning_rate": 1.8290000000000003e-05,
      "loss": 0.4282,
      "step": 3661
    },
    {
      "epoch": 0.20506215701646321,
      "grad_norm": 1.0750913619995117,
      "learning_rate": 1.8295e-05,
      "loss": 0.411,
      "step": 3662
    },
    {
      "epoch": 0.20511815432859223,
      "grad_norm": 1.0906314849853516,
      "learning_rate": 1.83e-05,
      "loss": 0.4069,
      "step": 3663
    },
    {
      "epoch": 0.20517415164072125,
      "grad_norm": 1.1319752931594849,
      "learning_rate": 1.8305e-05,
      "loss": 0.4575,
      "step": 3664
    },
    {
      "epoch": 0.20523014895285027,
      "grad_norm": 1.0121747255325317,
      "learning_rate": 1.8310000000000003e-05,
      "loss": 0.4482,
      "step": 3665
    },
    {
      "epoch": 0.2052861462649793,
      "grad_norm": 1.1749929189682007,
      "learning_rate": 1.8315e-05,
      "loss": 0.3555,
      "step": 3666
    },
    {
      "epoch": 0.2053421435771083,
      "grad_norm": 1.1696017980575562,
      "learning_rate": 1.832e-05,
      "loss": 0.3813,
      "step": 3667
    },
    {
      "epoch": 0.20539814088923733,
      "grad_norm": 1.187969446182251,
      "learning_rate": 1.8325e-05,
      "loss": 0.5685,
      "step": 3668
    },
    {
      "epoch": 0.20545413820136632,
      "grad_norm": 1.239634394645691,
      "learning_rate": 1.833e-05,
      "loss": 0.4809,
      "step": 3669
    },
    {
      "epoch": 0.20551013551349534,
      "grad_norm": 1.0861446857452393,
      "learning_rate": 1.8335000000000003e-05,
      "loss": 0.3965,
      "step": 3670
    },
    {
      "epoch": 0.20556613282562436,
      "grad_norm": 1.1064029932022095,
      "learning_rate": 1.834e-05,
      "loss": 0.5242,
      "step": 3671
    },
    {
      "epoch": 0.20562213013775338,
      "grad_norm": 1.0597418546676636,
      "learning_rate": 1.8345e-05,
      "loss": 0.4466,
      "step": 3672
    },
    {
      "epoch": 0.2056781274498824,
      "grad_norm": 1.2875657081604004,
      "learning_rate": 1.8350000000000002e-05,
      "loss": 0.3883,
      "step": 3673
    },
    {
      "epoch": 0.20573412476201142,
      "grad_norm": 1.3540278673171997,
      "learning_rate": 1.8355e-05,
      "loss": 0.5198,
      "step": 3674
    },
    {
      "epoch": 0.20579012207414044,
      "grad_norm": 1.4080015420913696,
      "learning_rate": 1.8360000000000004e-05,
      "loss": 0.4586,
      "step": 3675
    },
    {
      "epoch": 0.20584611938626945,
      "grad_norm": 1.2323344945907593,
      "learning_rate": 1.8365e-05,
      "loss": 0.5165,
      "step": 3676
    },
    {
      "epoch": 0.20590211669839847,
      "grad_norm": 1.1104493141174316,
      "learning_rate": 1.8370000000000002e-05,
      "loss": 0.3237,
      "step": 3677
    },
    {
      "epoch": 0.2059581140105275,
      "grad_norm": 1.1640260219573975,
      "learning_rate": 1.8375e-05,
      "loss": 0.4496,
      "step": 3678
    },
    {
      "epoch": 0.2060141113226565,
      "grad_norm": 1.1987717151641846,
      "learning_rate": 1.838e-05,
      "loss": 0.383,
      "step": 3679
    },
    {
      "epoch": 0.20607010863478553,
      "grad_norm": 1.0047123432159424,
      "learning_rate": 1.8385e-05,
      "loss": 0.4024,
      "step": 3680
    },
    {
      "epoch": 0.20612610594691455,
      "grad_norm": 1.3158748149871826,
      "learning_rate": 1.8390000000000002e-05,
      "loss": 0.3559,
      "step": 3681
    },
    {
      "epoch": 0.20618210325904357,
      "grad_norm": 1.124266505241394,
      "learning_rate": 1.8395000000000003e-05,
      "loss": 0.3636,
      "step": 3682
    },
    {
      "epoch": 0.2062381005711726,
      "grad_norm": 1.1218291521072388,
      "learning_rate": 1.84e-05,
      "loss": 0.4406,
      "step": 3683
    },
    {
      "epoch": 0.2062940978833016,
      "grad_norm": 1.4445165395736694,
      "learning_rate": 1.8405e-05,
      "loss": 0.6085,
      "step": 3684
    },
    {
      "epoch": 0.20635009519543063,
      "grad_norm": 1.0143694877624512,
      "learning_rate": 1.841e-05,
      "loss": 0.3806,
      "step": 3685
    },
    {
      "epoch": 0.20640609250755965,
      "grad_norm": 1.389467477798462,
      "learning_rate": 1.8415000000000002e-05,
      "loss": 0.5131,
      "step": 3686
    },
    {
      "epoch": 0.20646208981968867,
      "grad_norm": 1.1578280925750732,
      "learning_rate": 1.842e-05,
      "loss": 0.4549,
      "step": 3687
    },
    {
      "epoch": 0.20651808713181768,
      "grad_norm": 1.009496808052063,
      "learning_rate": 1.8425e-05,
      "loss": 0.3502,
      "step": 3688
    },
    {
      "epoch": 0.2065740844439467,
      "grad_norm": 1.3251539468765259,
      "learning_rate": 1.843e-05,
      "loss": 0.4412,
      "step": 3689
    },
    {
      "epoch": 0.2066300817560757,
      "grad_norm": 1.3728529214859009,
      "learning_rate": 1.8435000000000002e-05,
      "loss": 0.4851,
      "step": 3690
    },
    {
      "epoch": 0.20668607906820471,
      "grad_norm": 1.215119481086731,
      "learning_rate": 1.8440000000000003e-05,
      "loss": 0.4707,
      "step": 3691
    },
    {
      "epoch": 0.20674207638033373,
      "grad_norm": 1.2378907203674316,
      "learning_rate": 1.8445e-05,
      "loss": 0.4372,
      "step": 3692
    },
    {
      "epoch": 0.20679807369246275,
      "grad_norm": 1.4404093027114868,
      "learning_rate": 1.845e-05,
      "loss": 0.4095,
      "step": 3693
    },
    {
      "epoch": 0.20685407100459177,
      "grad_norm": 1.1341023445129395,
      "learning_rate": 1.8455e-05,
      "loss": 0.5088,
      "step": 3694
    },
    {
      "epoch": 0.2069100683167208,
      "grad_norm": 1.326324701309204,
      "learning_rate": 1.846e-05,
      "loss": 0.3672,
      "step": 3695
    },
    {
      "epoch": 0.2069660656288498,
      "grad_norm": 1.1963605880737305,
      "learning_rate": 1.8465e-05,
      "loss": 0.4715,
      "step": 3696
    },
    {
      "epoch": 0.20702206294097883,
      "grad_norm": 1.351576566696167,
      "learning_rate": 1.847e-05,
      "loss": 0.538,
      "step": 3697
    },
    {
      "epoch": 0.20707806025310785,
      "grad_norm": 1.4372308254241943,
      "learning_rate": 1.8475000000000002e-05,
      "loss": 0.3944,
      "step": 3698
    },
    {
      "epoch": 0.20713405756523687,
      "grad_norm": 1.6565874814987183,
      "learning_rate": 1.848e-05,
      "loss": 0.4954,
      "step": 3699
    },
    {
      "epoch": 0.2071900548773659,
      "grad_norm": 1.2670085430145264,
      "learning_rate": 1.8485e-05,
      "loss": 0.3647,
      "step": 3700
    },
    {
      "epoch": 0.2072460521894949,
      "grad_norm": 1.227007508277893,
      "learning_rate": 1.849e-05,
      "loss": 0.3455,
      "step": 3701
    },
    {
      "epoch": 0.20730204950162393,
      "grad_norm": 1.4911811351776123,
      "learning_rate": 1.8495e-05,
      "loss": 0.5222,
      "step": 3702
    },
    {
      "epoch": 0.20735804681375294,
      "grad_norm": 1.2862108945846558,
      "learning_rate": 1.85e-05,
      "loss": 0.4129,
      "step": 3703
    },
    {
      "epoch": 0.20741404412588196,
      "grad_norm": 1.2085920572280884,
      "learning_rate": 1.8505e-05,
      "loss": 0.4963,
      "step": 3704
    },
    {
      "epoch": 0.20747004143801098,
      "grad_norm": 1.2827470302581787,
      "learning_rate": 1.851e-05,
      "loss": 0.4467,
      "step": 3705
    },
    {
      "epoch": 0.20752603875014,
      "grad_norm": 1.2292492389678955,
      "learning_rate": 1.8515e-05,
      "loss": 0.5528,
      "step": 3706
    },
    {
      "epoch": 0.20758203606226902,
      "grad_norm": 1.120884895324707,
      "learning_rate": 1.8520000000000002e-05,
      "loss": 0.3977,
      "step": 3707
    },
    {
      "epoch": 0.20763803337439804,
      "grad_norm": 1.2557817697525024,
      "learning_rate": 1.8525e-05,
      "loss": 0.3356,
      "step": 3708
    },
    {
      "epoch": 0.20769403068652706,
      "grad_norm": 1.2214707136154175,
      "learning_rate": 1.853e-05,
      "loss": 0.3781,
      "step": 3709
    },
    {
      "epoch": 0.20775002799865608,
      "grad_norm": 1.42682945728302,
      "learning_rate": 1.8535e-05,
      "loss": 0.5651,
      "step": 3710
    },
    {
      "epoch": 0.20780602531078507,
      "grad_norm": 1.4732400178909302,
      "learning_rate": 1.8540000000000002e-05,
      "loss": 0.5905,
      "step": 3711
    },
    {
      "epoch": 0.2078620226229141,
      "grad_norm": 1.1726034879684448,
      "learning_rate": 1.8545000000000003e-05,
      "loss": 0.5278,
      "step": 3712
    },
    {
      "epoch": 0.2079180199350431,
      "grad_norm": 1.4470715522766113,
      "learning_rate": 1.855e-05,
      "loss": 0.4314,
      "step": 3713
    },
    {
      "epoch": 0.20797401724717213,
      "grad_norm": 2.5902199745178223,
      "learning_rate": 1.8555e-05,
      "loss": 0.3967,
      "step": 3714
    },
    {
      "epoch": 0.20803001455930115,
      "grad_norm": 1.2021496295928955,
      "learning_rate": 1.856e-05,
      "loss": 0.3446,
      "step": 3715
    },
    {
      "epoch": 0.20808601187143017,
      "grad_norm": 1.0416159629821777,
      "learning_rate": 1.8565000000000003e-05,
      "loss": 0.3733,
      "step": 3716
    },
    {
      "epoch": 0.20814200918355918,
      "grad_norm": 1.1497995853424072,
      "learning_rate": 1.857e-05,
      "loss": 0.4278,
      "step": 3717
    },
    {
      "epoch": 0.2081980064956882,
      "grad_norm": 1.2033779621124268,
      "learning_rate": 1.8575e-05,
      "loss": 0.4386,
      "step": 3718
    },
    {
      "epoch": 0.20825400380781722,
      "grad_norm": 1.0341720581054688,
      "learning_rate": 1.858e-05,
      "loss": 0.4024,
      "step": 3719
    },
    {
      "epoch": 0.20831000111994624,
      "grad_norm": 1.464414358139038,
      "learning_rate": 1.8585e-05,
      "loss": 0.3869,
      "step": 3720
    },
    {
      "epoch": 0.20836599843207526,
      "grad_norm": 1.1235709190368652,
      "learning_rate": 1.8590000000000003e-05,
      "loss": 0.5638,
      "step": 3721
    },
    {
      "epoch": 0.20842199574420428,
      "grad_norm": 1.870194911956787,
      "learning_rate": 1.8595e-05,
      "loss": 0.4259,
      "step": 3722
    },
    {
      "epoch": 0.2084779930563333,
      "grad_norm": 1.689988613128662,
      "learning_rate": 1.86e-05,
      "loss": 0.5645,
      "step": 3723
    },
    {
      "epoch": 0.20853399036846232,
      "grad_norm": 1.531330943107605,
      "learning_rate": 1.8605e-05,
      "loss": 0.3632,
      "step": 3724
    },
    {
      "epoch": 0.20858998768059134,
      "grad_norm": 1.2564024925231934,
      "learning_rate": 1.861e-05,
      "loss": 0.3609,
      "step": 3725
    },
    {
      "epoch": 0.20864598499272036,
      "grad_norm": 1.226952314376831,
      "learning_rate": 1.8615e-05,
      "loss": 0.4338,
      "step": 3726
    },
    {
      "epoch": 0.20870198230484938,
      "grad_norm": 1.4002666473388672,
      "learning_rate": 1.862e-05,
      "loss": 0.4459,
      "step": 3727
    },
    {
      "epoch": 0.2087579796169784,
      "grad_norm": 1.1037551164627075,
      "learning_rate": 1.8625000000000002e-05,
      "loss": 0.4196,
      "step": 3728
    },
    {
      "epoch": 0.20881397692910741,
      "grad_norm": 1.1568371057510376,
      "learning_rate": 1.863e-05,
      "loss": 0.6159,
      "step": 3729
    },
    {
      "epoch": 0.20886997424123643,
      "grad_norm": 1.2428112030029297,
      "learning_rate": 1.8635e-05,
      "loss": 0.5167,
      "step": 3730
    },
    {
      "epoch": 0.20892597155336542,
      "grad_norm": 1.3571763038635254,
      "learning_rate": 1.864e-05,
      "loss": 0.4118,
      "step": 3731
    },
    {
      "epoch": 0.20898196886549444,
      "grad_norm": 1.3883105516433716,
      "learning_rate": 1.8645000000000002e-05,
      "loss": 0.4858,
      "step": 3732
    },
    {
      "epoch": 0.20903796617762346,
      "grad_norm": 1.0853148698806763,
      "learning_rate": 1.865e-05,
      "loss": 0.4217,
      "step": 3733
    },
    {
      "epoch": 0.20909396348975248,
      "grad_norm": 1.1567766666412354,
      "learning_rate": 1.8655e-05,
      "loss": 0.4391,
      "step": 3734
    },
    {
      "epoch": 0.2091499608018815,
      "grad_norm": 1.1084656715393066,
      "learning_rate": 1.866e-05,
      "loss": 0.4222,
      "step": 3735
    },
    {
      "epoch": 0.20920595811401052,
      "grad_norm": 1.4658235311508179,
      "learning_rate": 1.8665000000000002e-05,
      "loss": 0.4633,
      "step": 3736
    },
    {
      "epoch": 0.20926195542613954,
      "grad_norm": 1.1171648502349854,
      "learning_rate": 1.8670000000000003e-05,
      "loss": 0.4406,
      "step": 3737
    },
    {
      "epoch": 0.20931795273826856,
      "grad_norm": 1.4457217454910278,
      "learning_rate": 1.8675e-05,
      "loss": 0.4031,
      "step": 3738
    },
    {
      "epoch": 0.20937395005039758,
      "grad_norm": 1.199464201927185,
      "learning_rate": 1.868e-05,
      "loss": 0.4826,
      "step": 3739
    },
    {
      "epoch": 0.2094299473625266,
      "grad_norm": 1.1101138591766357,
      "learning_rate": 1.8684999999999998e-05,
      "loss": 0.4125,
      "step": 3740
    },
    {
      "epoch": 0.20948594467465562,
      "grad_norm": 1.6544722318649292,
      "learning_rate": 1.8690000000000002e-05,
      "loss": 0.4482,
      "step": 3741
    },
    {
      "epoch": 0.20954194198678464,
      "grad_norm": 1.0128848552703857,
      "learning_rate": 1.8695e-05,
      "loss": 0.3927,
      "step": 3742
    },
    {
      "epoch": 0.20959793929891365,
      "grad_norm": 1.020386815071106,
      "learning_rate": 1.87e-05,
      "loss": 0.317,
      "step": 3743
    },
    {
      "epoch": 0.20965393661104267,
      "grad_norm": 1.2744553089141846,
      "learning_rate": 1.8705e-05,
      "loss": 0.4031,
      "step": 3744
    },
    {
      "epoch": 0.2097099339231717,
      "grad_norm": 1.2086420059204102,
      "learning_rate": 1.871e-05,
      "loss": 0.4046,
      "step": 3745
    },
    {
      "epoch": 0.2097659312353007,
      "grad_norm": 1.2720385789871216,
      "learning_rate": 1.8715000000000003e-05,
      "loss": 0.4306,
      "step": 3746
    },
    {
      "epoch": 0.20982192854742973,
      "grad_norm": 1.245012640953064,
      "learning_rate": 1.872e-05,
      "loss": 0.4181,
      "step": 3747
    },
    {
      "epoch": 0.20987792585955875,
      "grad_norm": 1.219249963760376,
      "learning_rate": 1.8725e-05,
      "loss": 0.4987,
      "step": 3748
    },
    {
      "epoch": 0.20993392317168777,
      "grad_norm": 1.035294532775879,
      "learning_rate": 1.8730000000000002e-05,
      "loss": 0.3916,
      "step": 3749
    },
    {
      "epoch": 0.2099899204838168,
      "grad_norm": 1.2753604650497437,
      "learning_rate": 1.8735e-05,
      "loss": 0.4066,
      "step": 3750
    },
    {
      "epoch": 0.2100459177959458,
      "grad_norm": 2.5616674423217773,
      "learning_rate": 1.8740000000000004e-05,
      "loss": 0.5153,
      "step": 3751
    },
    {
      "epoch": 0.2101019151080748,
      "grad_norm": 1.2599841356277466,
      "learning_rate": 1.8745e-05,
      "loss": 0.4677,
      "step": 3752
    },
    {
      "epoch": 0.21015791242020382,
      "grad_norm": 1.5064753293991089,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.559,
      "step": 3753
    },
    {
      "epoch": 0.21021390973233284,
      "grad_norm": 1.3997769355773926,
      "learning_rate": 1.8755e-05,
      "loss": 0.4659,
      "step": 3754
    },
    {
      "epoch": 0.21026990704446186,
      "grad_norm": 1.1182029247283936,
      "learning_rate": 1.876e-05,
      "loss": 0.2785,
      "step": 3755
    },
    {
      "epoch": 0.21032590435659088,
      "grad_norm": 1.0288310050964355,
      "learning_rate": 1.8765e-05,
      "loss": 0.3188,
      "step": 3756
    },
    {
      "epoch": 0.2103819016687199,
      "grad_norm": 1.3543319702148438,
      "learning_rate": 1.877e-05,
      "loss": 0.4312,
      "step": 3757
    },
    {
      "epoch": 0.21043789898084891,
      "grad_norm": 1.339989423751831,
      "learning_rate": 1.8775000000000002e-05,
      "loss": 0.5552,
      "step": 3758
    },
    {
      "epoch": 0.21049389629297793,
      "grad_norm": 1.3205184936523438,
      "learning_rate": 1.878e-05,
      "loss": 0.6186,
      "step": 3759
    },
    {
      "epoch": 0.21054989360510695,
      "grad_norm": 1.0515899658203125,
      "learning_rate": 1.8785e-05,
      "loss": 0.2768,
      "step": 3760
    },
    {
      "epoch": 0.21060589091723597,
      "grad_norm": 1.117658257484436,
      "learning_rate": 1.879e-05,
      "loss": 0.5759,
      "step": 3761
    },
    {
      "epoch": 0.210661888229365,
      "grad_norm": 1.1721341609954834,
      "learning_rate": 1.8795000000000002e-05,
      "loss": 0.369,
      "step": 3762
    },
    {
      "epoch": 0.210717885541494,
      "grad_norm": 1.2887459993362427,
      "learning_rate": 1.88e-05,
      "loss": 0.4338,
      "step": 3763
    },
    {
      "epoch": 0.21077388285362303,
      "grad_norm": 1.096023440361023,
      "learning_rate": 1.8805e-05,
      "loss": 0.3805,
      "step": 3764
    },
    {
      "epoch": 0.21082988016575205,
      "grad_norm": 1.596562147140503,
      "learning_rate": 1.881e-05,
      "loss": 0.4453,
      "step": 3765
    },
    {
      "epoch": 0.21088587747788107,
      "grad_norm": 1.3703455924987793,
      "learning_rate": 1.8815000000000002e-05,
      "loss": 0.4875,
      "step": 3766
    },
    {
      "epoch": 0.2109418747900101,
      "grad_norm": 1.202178716659546,
      "learning_rate": 1.8820000000000003e-05,
      "loss": 0.4183,
      "step": 3767
    },
    {
      "epoch": 0.2109978721021391,
      "grad_norm": 1.2643823623657227,
      "learning_rate": 1.8825e-05,
      "loss": 0.3607,
      "step": 3768
    },
    {
      "epoch": 0.21105386941426812,
      "grad_norm": 1.2469091415405273,
      "learning_rate": 1.883e-05,
      "loss": 0.5585,
      "step": 3769
    },
    {
      "epoch": 0.21110986672639714,
      "grad_norm": 1.277901530265808,
      "learning_rate": 1.8835e-05,
      "loss": 0.4921,
      "step": 3770
    },
    {
      "epoch": 0.21116586403852616,
      "grad_norm": 1.6774253845214844,
      "learning_rate": 1.8840000000000003e-05,
      "loss": 0.7642,
      "step": 3771
    },
    {
      "epoch": 0.21122186135065518,
      "grad_norm": 1.413717269897461,
      "learning_rate": 1.8845e-05,
      "loss": 0.4913,
      "step": 3772
    },
    {
      "epoch": 0.21127785866278417,
      "grad_norm": 1.151466965675354,
      "learning_rate": 1.885e-05,
      "loss": 0.3912,
      "step": 3773
    },
    {
      "epoch": 0.2113338559749132,
      "grad_norm": 1.195562720298767,
      "learning_rate": 1.8855e-05,
      "loss": 0.3835,
      "step": 3774
    },
    {
      "epoch": 0.2113898532870422,
      "grad_norm": 1.1106895208358765,
      "learning_rate": 1.886e-05,
      "loss": 0.4649,
      "step": 3775
    },
    {
      "epoch": 0.21144585059917123,
      "grad_norm": 1.3207541704177856,
      "learning_rate": 1.8865000000000003e-05,
      "loss": 0.5763,
      "step": 3776
    },
    {
      "epoch": 0.21150184791130025,
      "grad_norm": 1.8818163871765137,
      "learning_rate": 1.887e-05,
      "loss": 0.3591,
      "step": 3777
    },
    {
      "epoch": 0.21155784522342927,
      "grad_norm": 1.256289005279541,
      "learning_rate": 1.8875e-05,
      "loss": 0.4754,
      "step": 3778
    },
    {
      "epoch": 0.2116138425355583,
      "grad_norm": 1.5186047554016113,
      "learning_rate": 1.888e-05,
      "loss": 0.6175,
      "step": 3779
    },
    {
      "epoch": 0.2116698398476873,
      "grad_norm": 1.3236587047576904,
      "learning_rate": 1.8885e-05,
      "loss": 0.4407,
      "step": 3780
    },
    {
      "epoch": 0.21172583715981633,
      "grad_norm": 1.3085664510726929,
      "learning_rate": 1.8890000000000004e-05,
      "loss": 0.4197,
      "step": 3781
    },
    {
      "epoch": 0.21178183447194535,
      "grad_norm": 1.1369835138320923,
      "learning_rate": 1.8895e-05,
      "loss": 0.3776,
      "step": 3782
    },
    {
      "epoch": 0.21183783178407437,
      "grad_norm": 1.1113334894180298,
      "learning_rate": 1.8900000000000002e-05,
      "loss": 0.3816,
      "step": 3783
    },
    {
      "epoch": 0.21189382909620338,
      "grad_norm": 1.2961411476135254,
      "learning_rate": 1.8905e-05,
      "loss": 0.4108,
      "step": 3784
    },
    {
      "epoch": 0.2119498264083324,
      "grad_norm": 1.045507550239563,
      "learning_rate": 1.891e-05,
      "loss": 0.3978,
      "step": 3785
    },
    {
      "epoch": 0.21200582372046142,
      "grad_norm": 1.2163691520690918,
      "learning_rate": 1.8915e-05,
      "loss": 0.4174,
      "step": 3786
    },
    {
      "epoch": 0.21206182103259044,
      "grad_norm": 1.1749398708343506,
      "learning_rate": 1.8920000000000002e-05,
      "loss": 0.5375,
      "step": 3787
    },
    {
      "epoch": 0.21211781834471946,
      "grad_norm": 1.2830992937088013,
      "learning_rate": 1.8925000000000003e-05,
      "loss": 0.4793,
      "step": 3788
    },
    {
      "epoch": 0.21217381565684848,
      "grad_norm": 1.1235780715942383,
      "learning_rate": 1.893e-05,
      "loss": 0.4108,
      "step": 3789
    },
    {
      "epoch": 0.2122298129689775,
      "grad_norm": 1.4002350568771362,
      "learning_rate": 1.8935e-05,
      "loss": 0.6034,
      "step": 3790
    },
    {
      "epoch": 0.21228581028110652,
      "grad_norm": 1.2956151962280273,
      "learning_rate": 1.894e-05,
      "loss": 0.5659,
      "step": 3791
    },
    {
      "epoch": 0.21234180759323554,
      "grad_norm": 1.1184310913085938,
      "learning_rate": 1.8945000000000002e-05,
      "loss": 0.3904,
      "step": 3792
    },
    {
      "epoch": 0.21239780490536453,
      "grad_norm": 1.0486329793930054,
      "learning_rate": 1.895e-05,
      "loss": 0.3441,
      "step": 3793
    },
    {
      "epoch": 0.21245380221749355,
      "grad_norm": 1.1602412462234497,
      "learning_rate": 1.8955e-05,
      "loss": 0.4124,
      "step": 3794
    },
    {
      "epoch": 0.21250979952962257,
      "grad_norm": 1.1501165628433228,
      "learning_rate": 1.896e-05,
      "loss": 0.4024,
      "step": 3795
    },
    {
      "epoch": 0.2125657968417516,
      "grad_norm": 1.0928531885147095,
      "learning_rate": 1.8965000000000002e-05,
      "loss": 0.3906,
      "step": 3796
    },
    {
      "epoch": 0.2126217941538806,
      "grad_norm": 1.1243789196014404,
      "learning_rate": 1.8970000000000003e-05,
      "loss": 0.3966,
      "step": 3797
    },
    {
      "epoch": 0.21267779146600962,
      "grad_norm": 1.264003872871399,
      "learning_rate": 1.8975e-05,
      "loss": 0.5284,
      "step": 3798
    },
    {
      "epoch": 0.21273378877813864,
      "grad_norm": 1.400215983390808,
      "learning_rate": 1.898e-05,
      "loss": 0.4005,
      "step": 3799
    },
    {
      "epoch": 0.21278978609026766,
      "grad_norm": 1.1315284967422485,
      "learning_rate": 1.8985e-05,
      "loss": 0.3573,
      "step": 3800
    },
    {
      "epoch": 0.21284578340239668,
      "grad_norm": 1.10823392868042,
      "learning_rate": 1.8990000000000003e-05,
      "loss": 0.4071,
      "step": 3801
    },
    {
      "epoch": 0.2129017807145257,
      "grad_norm": 1.2107127904891968,
      "learning_rate": 1.8995e-05,
      "loss": 0.4024,
      "step": 3802
    },
    {
      "epoch": 0.21295777802665472,
      "grad_norm": 1.29421067237854,
      "learning_rate": 1.9e-05,
      "loss": 0.4936,
      "step": 3803
    },
    {
      "epoch": 0.21301377533878374,
      "grad_norm": 2.305823802947998,
      "learning_rate": 1.9005000000000002e-05,
      "loss": 0.326,
      "step": 3804
    },
    {
      "epoch": 0.21306977265091276,
      "grad_norm": 1.4411567449569702,
      "learning_rate": 1.901e-05,
      "loss": 0.5825,
      "step": 3805
    },
    {
      "epoch": 0.21312576996304178,
      "grad_norm": 0.961778461933136,
      "learning_rate": 1.9015000000000003e-05,
      "loss": 0.3066,
      "step": 3806
    },
    {
      "epoch": 0.2131817672751708,
      "grad_norm": 1.2145847082138062,
      "learning_rate": 1.902e-05,
      "loss": 0.4319,
      "step": 3807
    },
    {
      "epoch": 0.21323776458729982,
      "grad_norm": 1.2316093444824219,
      "learning_rate": 1.9025e-05,
      "loss": 0.3642,
      "step": 3808
    },
    {
      "epoch": 0.21329376189942884,
      "grad_norm": 1.1941778659820557,
      "learning_rate": 1.903e-05,
      "loss": 0.4714,
      "step": 3809
    },
    {
      "epoch": 0.21334975921155785,
      "grad_norm": 1.402944564819336,
      "learning_rate": 1.9035e-05,
      "loss": 0.4408,
      "step": 3810
    },
    {
      "epoch": 0.21340575652368687,
      "grad_norm": 1.1802160739898682,
      "learning_rate": 1.904e-05,
      "loss": 0.3552,
      "step": 3811
    },
    {
      "epoch": 0.2134617538358159,
      "grad_norm": 1.1805145740509033,
      "learning_rate": 1.9045e-05,
      "loss": 0.4566,
      "step": 3812
    },
    {
      "epoch": 0.2135177511479449,
      "grad_norm": 1.1233844757080078,
      "learning_rate": 1.9050000000000002e-05,
      "loss": 0.3404,
      "step": 3813
    },
    {
      "epoch": 0.2135737484600739,
      "grad_norm": 1.4245500564575195,
      "learning_rate": 1.9055e-05,
      "loss": 0.4563,
      "step": 3814
    },
    {
      "epoch": 0.21362974577220292,
      "grad_norm": 1.065510869026184,
      "learning_rate": 1.906e-05,
      "loss": 0.3891,
      "step": 3815
    },
    {
      "epoch": 0.21368574308433194,
      "grad_norm": 1.2466912269592285,
      "learning_rate": 1.9064999999999998e-05,
      "loss": 0.4299,
      "step": 3816
    },
    {
      "epoch": 0.21374174039646096,
      "grad_norm": 1.3753843307495117,
      "learning_rate": 1.9070000000000002e-05,
      "loss": 0.5251,
      "step": 3817
    },
    {
      "epoch": 0.21379773770858998,
      "grad_norm": 1.0985254049301147,
      "learning_rate": 1.9075000000000003e-05,
      "loss": 0.4184,
      "step": 3818
    },
    {
      "epoch": 0.213853735020719,
      "grad_norm": 1.2996129989624023,
      "learning_rate": 1.908e-05,
      "loss": 0.3592,
      "step": 3819
    },
    {
      "epoch": 0.21390973233284802,
      "grad_norm": 1.3616174459457397,
      "learning_rate": 1.9085e-05,
      "loss": 0.4883,
      "step": 3820
    },
    {
      "epoch": 0.21396572964497704,
      "grad_norm": 1.2674696445465088,
      "learning_rate": 1.909e-05,
      "loss": 0.5054,
      "step": 3821
    },
    {
      "epoch": 0.21402172695710606,
      "grad_norm": 1.243328332901001,
      "learning_rate": 1.9095000000000003e-05,
      "loss": 0.3761,
      "step": 3822
    },
    {
      "epoch": 0.21407772426923508,
      "grad_norm": 1.2152774333953857,
      "learning_rate": 1.91e-05,
      "loss": 0.4544,
      "step": 3823
    },
    {
      "epoch": 0.2141337215813641,
      "grad_norm": 1.3071707487106323,
      "learning_rate": 1.9105e-05,
      "loss": 0.4506,
      "step": 3824
    },
    {
      "epoch": 0.21418971889349311,
      "grad_norm": 1.2705812454223633,
      "learning_rate": 1.911e-05,
      "loss": 0.4259,
      "step": 3825
    },
    {
      "epoch": 0.21424571620562213,
      "grad_norm": 1.104065179824829,
      "learning_rate": 1.9115e-05,
      "loss": 0.4035,
      "step": 3826
    },
    {
      "epoch": 0.21430171351775115,
      "grad_norm": 1.5568335056304932,
      "learning_rate": 1.9120000000000003e-05,
      "loss": 0.3943,
      "step": 3827
    },
    {
      "epoch": 0.21435771082988017,
      "grad_norm": 1.2089629173278809,
      "learning_rate": 1.9125e-05,
      "loss": 0.4169,
      "step": 3828
    },
    {
      "epoch": 0.2144137081420092,
      "grad_norm": 1.3171229362487793,
      "learning_rate": 1.913e-05,
      "loss": 0.4887,
      "step": 3829
    },
    {
      "epoch": 0.2144697054541382,
      "grad_norm": 1.2432467937469482,
      "learning_rate": 1.9135e-05,
      "loss": 0.3612,
      "step": 3830
    },
    {
      "epoch": 0.21452570276626723,
      "grad_norm": 1.291901707649231,
      "learning_rate": 1.914e-05,
      "loss": 0.4334,
      "step": 3831
    },
    {
      "epoch": 0.21458170007839625,
      "grad_norm": 1.3384559154510498,
      "learning_rate": 1.9145e-05,
      "loss": 0.648,
      "step": 3832
    },
    {
      "epoch": 0.21463769739052527,
      "grad_norm": 1.176081657409668,
      "learning_rate": 1.915e-05,
      "loss": 0.4228,
      "step": 3833
    },
    {
      "epoch": 0.2146936947026543,
      "grad_norm": 1.0742981433868408,
      "learning_rate": 1.9155000000000002e-05,
      "loss": 0.386,
      "step": 3834
    },
    {
      "epoch": 0.21474969201478328,
      "grad_norm": 1.1499167680740356,
      "learning_rate": 1.916e-05,
      "loss": 0.457,
      "step": 3835
    },
    {
      "epoch": 0.2148056893269123,
      "grad_norm": 1.1521497964859009,
      "learning_rate": 1.9165e-05,
      "loss": 0.4305,
      "step": 3836
    },
    {
      "epoch": 0.21486168663904132,
      "grad_norm": 1.220842957496643,
      "learning_rate": 1.917e-05,
      "loss": 0.4883,
      "step": 3837
    },
    {
      "epoch": 0.21491768395117034,
      "grad_norm": 1.277765154838562,
      "learning_rate": 1.9175000000000002e-05,
      "loss": 0.5959,
      "step": 3838
    },
    {
      "epoch": 0.21497368126329935,
      "grad_norm": 1.2280707359313965,
      "learning_rate": 1.918e-05,
      "loss": 0.3969,
      "step": 3839
    },
    {
      "epoch": 0.21502967857542837,
      "grad_norm": 1.1846415996551514,
      "learning_rate": 1.9185e-05,
      "loss": 0.4591,
      "step": 3840
    },
    {
      "epoch": 0.2150856758875574,
      "grad_norm": 1.2512959241867065,
      "learning_rate": 1.919e-05,
      "loss": 0.4162,
      "step": 3841
    },
    {
      "epoch": 0.2151416731996864,
      "grad_norm": 1.3382731676101685,
      "learning_rate": 1.9195000000000002e-05,
      "loss": 0.4439,
      "step": 3842
    },
    {
      "epoch": 0.21519767051181543,
      "grad_norm": 1.1869760751724243,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.3983,
      "step": 3843
    },
    {
      "epoch": 0.21525366782394445,
      "grad_norm": 1.209533929824829,
      "learning_rate": 1.9205e-05,
      "loss": 0.4118,
      "step": 3844
    },
    {
      "epoch": 0.21530966513607347,
      "grad_norm": 1.108127474784851,
      "learning_rate": 1.921e-05,
      "loss": 0.4376,
      "step": 3845
    },
    {
      "epoch": 0.2153656624482025,
      "grad_norm": 1.2668966054916382,
      "learning_rate": 1.9214999999999998e-05,
      "loss": 0.4207,
      "step": 3846
    },
    {
      "epoch": 0.2154216597603315,
      "grad_norm": 1.006713628768921,
      "learning_rate": 1.9220000000000002e-05,
      "loss": 0.4193,
      "step": 3847
    },
    {
      "epoch": 0.21547765707246053,
      "grad_norm": 1.2234450578689575,
      "learning_rate": 1.9225e-05,
      "loss": 0.4321,
      "step": 3848
    },
    {
      "epoch": 0.21553365438458955,
      "grad_norm": 1.158087134361267,
      "learning_rate": 1.923e-05,
      "loss": 0.4531,
      "step": 3849
    },
    {
      "epoch": 0.21558965169671856,
      "grad_norm": 1.0506086349487305,
      "learning_rate": 1.9235e-05,
      "loss": 0.3995,
      "step": 3850
    },
    {
      "epoch": 0.21564564900884758,
      "grad_norm": 1.504860520362854,
      "learning_rate": 1.924e-05,
      "loss": 0.4866,
      "step": 3851
    },
    {
      "epoch": 0.2157016463209766,
      "grad_norm": 1.173839807510376,
      "learning_rate": 1.9245000000000003e-05,
      "loss": 0.3839,
      "step": 3852
    },
    {
      "epoch": 0.21575764363310562,
      "grad_norm": 1.140916347503662,
      "learning_rate": 1.925e-05,
      "loss": 0.4344,
      "step": 3853
    },
    {
      "epoch": 0.21581364094523464,
      "grad_norm": 1.2121093273162842,
      "learning_rate": 1.9255e-05,
      "loss": 0.4094,
      "step": 3854
    },
    {
      "epoch": 0.21586963825736363,
      "grad_norm": 0.9509993195533752,
      "learning_rate": 1.9260000000000002e-05,
      "loss": 0.4051,
      "step": 3855
    },
    {
      "epoch": 0.21592563556949265,
      "grad_norm": 1.1066420078277588,
      "learning_rate": 1.9265e-05,
      "loss": 0.5323,
      "step": 3856
    },
    {
      "epoch": 0.21598163288162167,
      "grad_norm": 1.3245099782943726,
      "learning_rate": 1.9270000000000004e-05,
      "loss": 0.3931,
      "step": 3857
    },
    {
      "epoch": 0.2160376301937507,
      "grad_norm": 1.0722681283950806,
      "learning_rate": 1.9275e-05,
      "loss": 0.3758,
      "step": 3858
    },
    {
      "epoch": 0.2160936275058797,
      "grad_norm": 1.0039154291152954,
      "learning_rate": 1.9280000000000002e-05,
      "loss": 0.294,
      "step": 3859
    },
    {
      "epoch": 0.21614962481800873,
      "grad_norm": 1.2737786769866943,
      "learning_rate": 1.9285e-05,
      "loss": 0.3626,
      "step": 3860
    },
    {
      "epoch": 0.21620562213013775,
      "grad_norm": 1.5499736070632935,
      "learning_rate": 1.929e-05,
      "loss": 0.4792,
      "step": 3861
    },
    {
      "epoch": 0.21626161944226677,
      "grad_norm": 1.185977578163147,
      "learning_rate": 1.9295e-05,
      "loss": 0.3455,
      "step": 3862
    },
    {
      "epoch": 0.2163176167543958,
      "grad_norm": 1.3004982471466064,
      "learning_rate": 1.93e-05,
      "loss": 0.4465,
      "step": 3863
    },
    {
      "epoch": 0.2163736140665248,
      "grad_norm": 1.3049736022949219,
      "learning_rate": 1.9305000000000002e-05,
      "loss": 0.4994,
      "step": 3864
    },
    {
      "epoch": 0.21642961137865382,
      "grad_norm": 1.0772696733474731,
      "learning_rate": 1.931e-05,
      "loss": 0.3288,
      "step": 3865
    },
    {
      "epoch": 0.21648560869078284,
      "grad_norm": 1.2598570585250854,
      "learning_rate": 1.9315e-05,
      "loss": 0.4872,
      "step": 3866
    },
    {
      "epoch": 0.21654160600291186,
      "grad_norm": 1.141111135482788,
      "learning_rate": 1.932e-05,
      "loss": 0.3861,
      "step": 3867
    },
    {
      "epoch": 0.21659760331504088,
      "grad_norm": 1.1527847051620483,
      "learning_rate": 1.9325000000000002e-05,
      "loss": 0.3721,
      "step": 3868
    },
    {
      "epoch": 0.2166536006271699,
      "grad_norm": 1.2805637121200562,
      "learning_rate": 1.933e-05,
      "loss": 0.4316,
      "step": 3869
    },
    {
      "epoch": 0.21670959793929892,
      "grad_norm": 1.1219383478164673,
      "learning_rate": 1.9335e-05,
      "loss": 0.3921,
      "step": 3870
    },
    {
      "epoch": 0.21676559525142794,
      "grad_norm": 1.264259696006775,
      "learning_rate": 1.934e-05,
      "loss": 0.518,
      "step": 3871
    },
    {
      "epoch": 0.21682159256355696,
      "grad_norm": 1.3242542743682861,
      "learning_rate": 1.9345000000000002e-05,
      "loss": 0.7451,
      "step": 3872
    },
    {
      "epoch": 0.21687758987568598,
      "grad_norm": 1.1558231115341187,
      "learning_rate": 1.9350000000000003e-05,
      "loss": 0.3191,
      "step": 3873
    },
    {
      "epoch": 0.216933587187815,
      "grad_norm": 1.0650345087051392,
      "learning_rate": 1.9355e-05,
      "loss": 0.3282,
      "step": 3874
    },
    {
      "epoch": 0.21698958449994402,
      "grad_norm": 1.3427801132202148,
      "learning_rate": 1.936e-05,
      "loss": 0.4338,
      "step": 3875
    },
    {
      "epoch": 0.217045581812073,
      "grad_norm": 1.2418391704559326,
      "learning_rate": 1.9365e-05,
      "loss": 0.3861,
      "step": 3876
    },
    {
      "epoch": 0.21710157912420203,
      "grad_norm": 1.1236191987991333,
      "learning_rate": 1.9370000000000003e-05,
      "loss": 0.3656,
      "step": 3877
    },
    {
      "epoch": 0.21715757643633105,
      "grad_norm": 1.2333683967590332,
      "learning_rate": 1.9375e-05,
      "loss": 0.455,
      "step": 3878
    },
    {
      "epoch": 0.21721357374846006,
      "grad_norm": 1.2741773128509521,
      "learning_rate": 1.938e-05,
      "loss": 0.5378,
      "step": 3879
    },
    {
      "epoch": 0.21726957106058908,
      "grad_norm": 1.2516076564788818,
      "learning_rate": 1.9385e-05,
      "loss": 0.5302,
      "step": 3880
    },
    {
      "epoch": 0.2173255683727181,
      "grad_norm": 1.4284601211547852,
      "learning_rate": 1.939e-05,
      "loss": 0.5134,
      "step": 3881
    },
    {
      "epoch": 0.21738156568484712,
      "grad_norm": 1.147981882095337,
      "learning_rate": 1.9395000000000003e-05,
      "loss": 0.4111,
      "step": 3882
    },
    {
      "epoch": 0.21743756299697614,
      "grad_norm": 1.3357466459274292,
      "learning_rate": 1.94e-05,
      "loss": 0.4595,
      "step": 3883
    },
    {
      "epoch": 0.21749356030910516,
      "grad_norm": 1.2503118515014648,
      "learning_rate": 1.9405e-05,
      "loss": 0.4039,
      "step": 3884
    },
    {
      "epoch": 0.21754955762123418,
      "grad_norm": 1.1886709928512573,
      "learning_rate": 1.941e-05,
      "loss": 0.5563,
      "step": 3885
    },
    {
      "epoch": 0.2176055549333632,
      "grad_norm": 1.2082996368408203,
      "learning_rate": 1.9415e-05,
      "loss": 0.386,
      "step": 3886
    },
    {
      "epoch": 0.21766155224549222,
      "grad_norm": 1.1673897504806519,
      "learning_rate": 1.942e-05,
      "loss": 0.4305,
      "step": 3887
    },
    {
      "epoch": 0.21771754955762124,
      "grad_norm": 1.125820279121399,
      "learning_rate": 1.9425e-05,
      "loss": 0.3498,
      "step": 3888
    },
    {
      "epoch": 0.21777354686975026,
      "grad_norm": 1.2262734174728394,
      "learning_rate": 1.9430000000000002e-05,
      "loss": 0.4673,
      "step": 3889
    },
    {
      "epoch": 0.21782954418187928,
      "grad_norm": 1.3357839584350586,
      "learning_rate": 1.9435e-05,
      "loss": 0.3753,
      "step": 3890
    },
    {
      "epoch": 0.2178855414940083,
      "grad_norm": 1.0969672203063965,
      "learning_rate": 1.944e-05,
      "loss": 0.3779,
      "step": 3891
    },
    {
      "epoch": 0.2179415388061373,
      "grad_norm": 1.1183667182922363,
      "learning_rate": 1.9445e-05,
      "loss": 0.5111,
      "step": 3892
    },
    {
      "epoch": 0.21799753611826633,
      "grad_norm": 1.522701621055603,
      "learning_rate": 1.9450000000000002e-05,
      "loss": 0.5003,
      "step": 3893
    },
    {
      "epoch": 0.21805353343039535,
      "grad_norm": 1.3292375802993774,
      "learning_rate": 1.9455000000000003e-05,
      "loss": 0.4663,
      "step": 3894
    },
    {
      "epoch": 0.21810953074252437,
      "grad_norm": 1.110558032989502,
      "learning_rate": 1.946e-05,
      "loss": 0.4689,
      "step": 3895
    },
    {
      "epoch": 0.2181655280546534,
      "grad_norm": 1.293588638305664,
      "learning_rate": 1.9465e-05,
      "loss": 0.3785,
      "step": 3896
    },
    {
      "epoch": 0.21822152536678238,
      "grad_norm": 1.0624083280563354,
      "learning_rate": 1.947e-05,
      "loss": 0.3824,
      "step": 3897
    },
    {
      "epoch": 0.2182775226789114,
      "grad_norm": 1.5652928352355957,
      "learning_rate": 1.9475000000000002e-05,
      "loss": 0.491,
      "step": 3898
    },
    {
      "epoch": 0.21833351999104042,
      "grad_norm": 1.1526103019714355,
      "learning_rate": 1.948e-05,
      "loss": 0.4683,
      "step": 3899
    },
    {
      "epoch": 0.21838951730316944,
      "grad_norm": 0.9201048612594604,
      "learning_rate": 1.9485e-05,
      "loss": 0.2474,
      "step": 3900
    },
    {
      "epoch": 0.21844551461529846,
      "grad_norm": 1.3379740715026855,
      "learning_rate": 1.949e-05,
      "loss": 0.4053,
      "step": 3901
    },
    {
      "epoch": 0.21850151192742748,
      "grad_norm": 1.0218881368637085,
      "learning_rate": 1.9495000000000002e-05,
      "loss": 0.3582,
      "step": 3902
    },
    {
      "epoch": 0.2185575092395565,
      "grad_norm": 1.1562334299087524,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 0.6028,
      "step": 3903
    },
    {
      "epoch": 0.21861350655168552,
      "grad_norm": 1.180961012840271,
      "learning_rate": 1.9505e-05,
      "loss": 0.5034,
      "step": 3904
    },
    {
      "epoch": 0.21866950386381453,
      "grad_norm": 1.1061652898788452,
      "learning_rate": 1.951e-05,
      "loss": 0.3482,
      "step": 3905
    },
    {
      "epoch": 0.21872550117594355,
      "grad_norm": 1.178895115852356,
      "learning_rate": 1.9515e-05,
      "loss": 0.4866,
      "step": 3906
    },
    {
      "epoch": 0.21878149848807257,
      "grad_norm": 1.2782617807388306,
      "learning_rate": 1.9520000000000003e-05,
      "loss": 0.4267,
      "step": 3907
    },
    {
      "epoch": 0.2188374958002016,
      "grad_norm": 1.481217622756958,
      "learning_rate": 1.9525e-05,
      "loss": 0.4508,
      "step": 3908
    },
    {
      "epoch": 0.2188934931123306,
      "grad_norm": 1.2187516689300537,
      "learning_rate": 1.953e-05,
      "loss": 0.5677,
      "step": 3909
    },
    {
      "epoch": 0.21894949042445963,
      "grad_norm": 1.1630263328552246,
      "learning_rate": 1.9535000000000002e-05,
      "loss": 0.4421,
      "step": 3910
    },
    {
      "epoch": 0.21900548773658865,
      "grad_norm": 1.4248945713043213,
      "learning_rate": 1.954e-05,
      "loss": 0.4959,
      "step": 3911
    },
    {
      "epoch": 0.21906148504871767,
      "grad_norm": 1.3881953954696655,
      "learning_rate": 1.9545000000000003e-05,
      "loss": 0.5092,
      "step": 3912
    },
    {
      "epoch": 0.2191174823608467,
      "grad_norm": 1.2719542980194092,
      "learning_rate": 1.955e-05,
      "loss": 0.5339,
      "step": 3913
    },
    {
      "epoch": 0.2191734796729757,
      "grad_norm": 1.3698889017105103,
      "learning_rate": 1.9555e-05,
      "loss": 0.452,
      "step": 3914
    },
    {
      "epoch": 0.21922947698510473,
      "grad_norm": 1.281928300857544,
      "learning_rate": 1.956e-05,
      "loss": 0.4297,
      "step": 3915
    },
    {
      "epoch": 0.21928547429723375,
      "grad_norm": 1.2747917175292969,
      "learning_rate": 1.9565e-05,
      "loss": 0.4263,
      "step": 3916
    },
    {
      "epoch": 0.21934147160936274,
      "grad_norm": 1.3958020210266113,
      "learning_rate": 1.957e-05,
      "loss": 0.4027,
      "step": 3917
    },
    {
      "epoch": 0.21939746892149176,
      "grad_norm": 1.191336989402771,
      "learning_rate": 1.9575e-05,
      "loss": 0.4204,
      "step": 3918
    },
    {
      "epoch": 0.21945346623362078,
      "grad_norm": 1.225386142730713,
      "learning_rate": 1.9580000000000002e-05,
      "loss": 0.4535,
      "step": 3919
    },
    {
      "epoch": 0.2195094635457498,
      "grad_norm": 1.052686095237732,
      "learning_rate": 1.9585e-05,
      "loss": 0.3126,
      "step": 3920
    },
    {
      "epoch": 0.2195654608578788,
      "grad_norm": 1.013994574546814,
      "learning_rate": 1.959e-05,
      "loss": 0.4611,
      "step": 3921
    },
    {
      "epoch": 0.21962145817000783,
      "grad_norm": 1.1470279693603516,
      "learning_rate": 1.9595e-05,
      "loss": 0.3578,
      "step": 3922
    },
    {
      "epoch": 0.21967745548213685,
      "grad_norm": 1.25358247756958,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 0.3754,
      "step": 3923
    },
    {
      "epoch": 0.21973345279426587,
      "grad_norm": 1.0770961046218872,
      "learning_rate": 1.9605e-05,
      "loss": 0.4533,
      "step": 3924
    },
    {
      "epoch": 0.2197894501063949,
      "grad_norm": 1.4587650299072266,
      "learning_rate": 1.961e-05,
      "loss": 0.465,
      "step": 3925
    },
    {
      "epoch": 0.2198454474185239,
      "grad_norm": 1.3744086027145386,
      "learning_rate": 1.9615e-05,
      "loss": 0.5234,
      "step": 3926
    },
    {
      "epoch": 0.21990144473065293,
      "grad_norm": 5.708791732788086,
      "learning_rate": 1.9620000000000002e-05,
      "loss": 0.3558,
      "step": 3927
    },
    {
      "epoch": 0.21995744204278195,
      "grad_norm": 1.502600908279419,
      "learning_rate": 1.9625000000000003e-05,
      "loss": 0.6503,
      "step": 3928
    },
    {
      "epoch": 0.22001343935491097,
      "grad_norm": 1.393741488456726,
      "learning_rate": 1.963e-05,
      "loss": 0.4353,
      "step": 3929
    },
    {
      "epoch": 0.22006943666703999,
      "grad_norm": 1.2048612833023071,
      "learning_rate": 1.9635e-05,
      "loss": 0.4566,
      "step": 3930
    },
    {
      "epoch": 0.220125433979169,
      "grad_norm": 1.2519006729125977,
      "learning_rate": 1.9640000000000002e-05,
      "loss": 0.6378,
      "step": 3931
    },
    {
      "epoch": 0.22018143129129802,
      "grad_norm": 1.1331911087036133,
      "learning_rate": 1.9645000000000002e-05,
      "loss": 0.3402,
      "step": 3932
    },
    {
      "epoch": 0.22023742860342704,
      "grad_norm": 1.2613248825073242,
      "learning_rate": 1.9650000000000003e-05,
      "loss": 0.4721,
      "step": 3933
    },
    {
      "epoch": 0.22029342591555606,
      "grad_norm": 1.1872540712356567,
      "learning_rate": 1.9655e-05,
      "loss": 0.4616,
      "step": 3934
    },
    {
      "epoch": 0.22034942322768508,
      "grad_norm": 1.1382859945297241,
      "learning_rate": 1.966e-05,
      "loss": 0.3649,
      "step": 3935
    },
    {
      "epoch": 0.2204054205398141,
      "grad_norm": 1.139190673828125,
      "learning_rate": 1.9665e-05,
      "loss": 0.4429,
      "step": 3936
    },
    {
      "epoch": 0.22046141785194312,
      "grad_norm": 1.1289645433425903,
      "learning_rate": 1.9670000000000003e-05,
      "loss": 0.4831,
      "step": 3937
    },
    {
      "epoch": 0.2205174151640721,
      "grad_norm": 1.1161208152770996,
      "learning_rate": 1.9675e-05,
      "loss": 0.516,
      "step": 3938
    },
    {
      "epoch": 0.22057341247620113,
      "grad_norm": 1.1716969013214111,
      "learning_rate": 1.968e-05,
      "loss": 0.3657,
      "step": 3939
    },
    {
      "epoch": 0.22062940978833015,
      "grad_norm": 1.2149205207824707,
      "learning_rate": 1.9685000000000002e-05,
      "loss": 0.5391,
      "step": 3940
    },
    {
      "epoch": 0.22068540710045917,
      "grad_norm": 1.528893232345581,
      "learning_rate": 1.969e-05,
      "loss": 0.4852,
      "step": 3941
    },
    {
      "epoch": 0.2207414044125882,
      "grad_norm": 1.6157225370407104,
      "learning_rate": 1.9695e-05,
      "loss": 0.4268,
      "step": 3942
    },
    {
      "epoch": 0.2207974017247172,
      "grad_norm": 1.2216227054595947,
      "learning_rate": 1.97e-05,
      "loss": 0.4365,
      "step": 3943
    },
    {
      "epoch": 0.22085339903684623,
      "grad_norm": 1.2056119441986084,
      "learning_rate": 1.9705000000000002e-05,
      "loss": 0.3892,
      "step": 3944
    },
    {
      "epoch": 0.22090939634897525,
      "grad_norm": 1.2248709201812744,
      "learning_rate": 1.971e-05,
      "loss": 0.4097,
      "step": 3945
    },
    {
      "epoch": 0.22096539366110426,
      "grad_norm": 1.1583555936813354,
      "learning_rate": 1.9715e-05,
      "loss": 0.3983,
      "step": 3946
    },
    {
      "epoch": 0.22102139097323328,
      "grad_norm": 1.0924626588821411,
      "learning_rate": 1.972e-05,
      "loss": 0.3504,
      "step": 3947
    },
    {
      "epoch": 0.2210773882853623,
      "grad_norm": 1.202951431274414,
      "learning_rate": 1.9725000000000002e-05,
      "loss": 0.5088,
      "step": 3948
    },
    {
      "epoch": 0.22113338559749132,
      "grad_norm": 1.0647072792053223,
      "learning_rate": 1.9730000000000003e-05,
      "loss": 0.4107,
      "step": 3949
    },
    {
      "epoch": 0.22118938290962034,
      "grad_norm": 1.2809765338897705,
      "learning_rate": 1.9735e-05,
      "loss": 0.4364,
      "step": 3950
    },
    {
      "epoch": 0.22124538022174936,
      "grad_norm": 1.095226526260376,
      "learning_rate": 1.974e-05,
      "loss": 0.4434,
      "step": 3951
    },
    {
      "epoch": 0.22130137753387838,
      "grad_norm": 1.2039659023284912,
      "learning_rate": 1.9744999999999998e-05,
      "loss": 0.4319,
      "step": 3952
    },
    {
      "epoch": 0.2213573748460074,
      "grad_norm": 1.1134363412857056,
      "learning_rate": 1.9750000000000002e-05,
      "loss": 0.3769,
      "step": 3953
    },
    {
      "epoch": 0.22141337215813642,
      "grad_norm": 1.2964189052581787,
      "learning_rate": 1.9755e-05,
      "loss": 0.4453,
      "step": 3954
    },
    {
      "epoch": 0.22146936947026544,
      "grad_norm": 1.1339832544326782,
      "learning_rate": 1.976e-05,
      "loss": 0.4056,
      "step": 3955
    },
    {
      "epoch": 0.22152536678239446,
      "grad_norm": 1.169204831123352,
      "learning_rate": 1.9765e-05,
      "loss": 0.5727,
      "step": 3956
    },
    {
      "epoch": 0.22158136409452348,
      "grad_norm": 1.220909833908081,
      "learning_rate": 1.977e-05,
      "loss": 0.4262,
      "step": 3957
    },
    {
      "epoch": 0.2216373614066525,
      "grad_norm": 3.000483512878418,
      "learning_rate": 1.9775000000000003e-05,
      "loss": 0.5093,
      "step": 3958
    },
    {
      "epoch": 0.22169335871878149,
      "grad_norm": 1.0850926637649536,
      "learning_rate": 1.978e-05,
      "loss": 0.3535,
      "step": 3959
    },
    {
      "epoch": 0.2217493560309105,
      "grad_norm": 1.3514493703842163,
      "learning_rate": 1.9785e-05,
      "loss": 0.5564,
      "step": 3960
    },
    {
      "epoch": 0.22180535334303952,
      "grad_norm": 1.3057911396026611,
      "learning_rate": 1.979e-05,
      "loss": 0.4899,
      "step": 3961
    },
    {
      "epoch": 0.22186135065516854,
      "grad_norm": 1.3611317873001099,
      "learning_rate": 1.9795e-05,
      "loss": 0.461,
      "step": 3962
    },
    {
      "epoch": 0.22191734796729756,
      "grad_norm": 0.9021110534667969,
      "learning_rate": 1.9800000000000004e-05,
      "loss": 0.3009,
      "step": 3963
    },
    {
      "epoch": 0.22197334527942658,
      "grad_norm": 1.1194170713424683,
      "learning_rate": 1.9805e-05,
      "loss": 0.4004,
      "step": 3964
    },
    {
      "epoch": 0.2220293425915556,
      "grad_norm": 1.1084462404251099,
      "learning_rate": 1.9810000000000002e-05,
      "loss": 0.4546,
      "step": 3965
    },
    {
      "epoch": 0.22208533990368462,
      "grad_norm": 1.375201940536499,
      "learning_rate": 1.9815e-05,
      "loss": 0.4415,
      "step": 3966
    },
    {
      "epoch": 0.22214133721581364,
      "grad_norm": 1.2064272165298462,
      "learning_rate": 1.982e-05,
      "loss": 0.3728,
      "step": 3967
    },
    {
      "epoch": 0.22219733452794266,
      "grad_norm": 1.4434298276901245,
      "learning_rate": 1.9825e-05,
      "loss": 0.3807,
      "step": 3968
    },
    {
      "epoch": 0.22225333184007168,
      "grad_norm": 1.2069168090820312,
      "learning_rate": 1.983e-05,
      "loss": 0.616,
      "step": 3969
    },
    {
      "epoch": 0.2223093291522007,
      "grad_norm": 1.4005426168441772,
      "learning_rate": 1.9835000000000002e-05,
      "loss": 0.4133,
      "step": 3970
    },
    {
      "epoch": 0.22236532646432972,
      "grad_norm": 1.2770768404006958,
      "learning_rate": 1.984e-05,
      "loss": 0.4056,
      "step": 3971
    },
    {
      "epoch": 0.22242132377645873,
      "grad_norm": 1.7044603824615479,
      "learning_rate": 1.9845e-05,
      "loss": 0.5747,
      "step": 3972
    },
    {
      "epoch": 0.22247732108858775,
      "grad_norm": 1.3524123430252075,
      "learning_rate": 1.985e-05,
      "loss": 0.5833,
      "step": 3973
    },
    {
      "epoch": 0.22253331840071677,
      "grad_norm": 1.2796204090118408,
      "learning_rate": 1.9855000000000002e-05,
      "loss": 0.472,
      "step": 3974
    },
    {
      "epoch": 0.2225893157128458,
      "grad_norm": 1.0586299896240234,
      "learning_rate": 1.986e-05,
      "loss": 0.4362,
      "step": 3975
    },
    {
      "epoch": 0.2226453130249748,
      "grad_norm": 1.1387319564819336,
      "learning_rate": 1.9865e-05,
      "loss": 0.5358,
      "step": 3976
    },
    {
      "epoch": 0.22270131033710383,
      "grad_norm": 1.3744721412658691,
      "learning_rate": 1.987e-05,
      "loss": 0.5624,
      "step": 3977
    },
    {
      "epoch": 0.22275730764923285,
      "grad_norm": 1.2997620105743408,
      "learning_rate": 1.9875000000000002e-05,
      "loss": 0.4796,
      "step": 3978
    },
    {
      "epoch": 0.22281330496136184,
      "grad_norm": 1.1262180805206299,
      "learning_rate": 1.9880000000000003e-05,
      "loss": 0.5009,
      "step": 3979
    },
    {
      "epoch": 0.22286930227349086,
      "grad_norm": 1.2399729490280151,
      "learning_rate": 1.9885e-05,
      "loss": 0.4173,
      "step": 3980
    },
    {
      "epoch": 0.22292529958561988,
      "grad_norm": 1.1392662525177002,
      "learning_rate": 1.989e-05,
      "loss": 0.5183,
      "step": 3981
    },
    {
      "epoch": 0.2229812968977489,
      "grad_norm": 1.3595386743545532,
      "learning_rate": 1.9895e-05,
      "loss": 0.4768,
      "step": 3982
    },
    {
      "epoch": 0.22303729420987792,
      "grad_norm": 1.2113465070724487,
      "learning_rate": 1.9900000000000003e-05,
      "loss": 0.4527,
      "step": 3983
    },
    {
      "epoch": 0.22309329152200694,
      "grad_norm": 1.2641816139221191,
      "learning_rate": 1.9905e-05,
      "loss": 0.4014,
      "step": 3984
    },
    {
      "epoch": 0.22314928883413596,
      "grad_norm": 1.269322156906128,
      "learning_rate": 1.991e-05,
      "loss": 0.4136,
      "step": 3985
    },
    {
      "epoch": 0.22320528614626498,
      "grad_norm": 1.7250635623931885,
      "learning_rate": 1.9915e-05,
      "loss": 0.5067,
      "step": 3986
    },
    {
      "epoch": 0.223261283458394,
      "grad_norm": 1.4223933219909668,
      "learning_rate": 1.992e-05,
      "loss": 0.3975,
      "step": 3987
    },
    {
      "epoch": 0.223317280770523,
      "grad_norm": 1.2370339632034302,
      "learning_rate": 1.9925000000000003e-05,
      "loss": 0.4593,
      "step": 3988
    },
    {
      "epoch": 0.22337327808265203,
      "grad_norm": 1.584632396697998,
      "learning_rate": 1.993e-05,
      "loss": 0.4526,
      "step": 3989
    },
    {
      "epoch": 0.22342927539478105,
      "grad_norm": 1.1744674444198608,
      "learning_rate": 1.9935e-05,
      "loss": 0.4607,
      "step": 3990
    },
    {
      "epoch": 0.22348527270691007,
      "grad_norm": 1.1270041465759277,
      "learning_rate": 1.994e-05,
      "loss": 0.3616,
      "step": 3991
    },
    {
      "epoch": 0.2235412700190391,
      "grad_norm": 1.543769359588623,
      "learning_rate": 1.9945e-05,
      "loss": 0.439,
      "step": 3992
    },
    {
      "epoch": 0.2235972673311681,
      "grad_norm": 1.010480284690857,
      "learning_rate": 1.995e-05,
      "loss": 0.3787,
      "step": 3993
    },
    {
      "epoch": 0.22365326464329713,
      "grad_norm": 1.2416563034057617,
      "learning_rate": 1.9955e-05,
      "loss": 0.4614,
      "step": 3994
    },
    {
      "epoch": 0.22370926195542615,
      "grad_norm": 1.3135097026824951,
      "learning_rate": 1.9960000000000002e-05,
      "loss": 0.499,
      "step": 3995
    },
    {
      "epoch": 0.22376525926755517,
      "grad_norm": 1.3679999113082886,
      "learning_rate": 1.9965e-05,
      "loss": 0.3731,
      "step": 3996
    },
    {
      "epoch": 0.22382125657968419,
      "grad_norm": 1.2174789905548096,
      "learning_rate": 1.997e-05,
      "loss": 0.3717,
      "step": 3997
    },
    {
      "epoch": 0.2238772538918132,
      "grad_norm": 1.1787962913513184,
      "learning_rate": 1.9975e-05,
      "loss": 0.4345,
      "step": 3998
    },
    {
      "epoch": 0.22393325120394222,
      "grad_norm": 1.3835020065307617,
      "learning_rate": 1.9980000000000002e-05,
      "loss": 0.4765,
      "step": 3999
    },
    {
      "epoch": 0.22398924851607122,
      "grad_norm": 1.0391042232513428,
      "learning_rate": 1.9985000000000003e-05,
      "loss": 0.4041,
      "step": 4000
    },
    {
      "epoch": 0.22404524582820023,
      "grad_norm": 1.2916306257247925,
      "learning_rate": 1.999e-05,
      "loss": 0.5135,
      "step": 4001
    },
    {
      "epoch": 0.22410124314032925,
      "grad_norm": 1.3005999326705933,
      "learning_rate": 1.9995e-05,
      "loss": 0.3908,
      "step": 4002
    },
    {
      "epoch": 0.22415724045245827,
      "grad_norm": 1.5253877639770508,
      "learning_rate": 2e-05,
      "loss": 0.5192,
      "step": 4003
    },
    {
      "epoch": 0.2242132377645873,
      "grad_norm": 1.100231647491455,
      "learning_rate": 2.0005000000000002e-05,
      "loss": 0.4016,
      "step": 4004
    },
    {
      "epoch": 0.2242692350767163,
      "grad_norm": 1.447638988494873,
      "learning_rate": 2.001e-05,
      "loss": 0.4658,
      "step": 4005
    },
    {
      "epoch": 0.22432523238884533,
      "grad_norm": 1.2481557130813599,
      "learning_rate": 2.0015e-05,
      "loss": 0.4316,
      "step": 4006
    },
    {
      "epoch": 0.22438122970097435,
      "grad_norm": 1.5969074964523315,
      "learning_rate": 2.002e-05,
      "loss": 0.5882,
      "step": 4007
    },
    {
      "epoch": 0.22443722701310337,
      "grad_norm": 1.2705366611480713,
      "learning_rate": 2.0025000000000002e-05,
      "loss": 0.5284,
      "step": 4008
    },
    {
      "epoch": 0.2244932243252324,
      "grad_norm": 1.133697509765625,
      "learning_rate": 2.0030000000000003e-05,
      "loss": 0.4072,
      "step": 4009
    },
    {
      "epoch": 0.2245492216373614,
      "grad_norm": 0.9591661691665649,
      "learning_rate": 2.0035e-05,
      "loss": 0.4083,
      "step": 4010
    },
    {
      "epoch": 0.22460521894949043,
      "grad_norm": 1.754600167274475,
      "learning_rate": 2.004e-05,
      "loss": 0.4694,
      "step": 4011
    },
    {
      "epoch": 0.22466121626161945,
      "grad_norm": 1.204201340675354,
      "learning_rate": 2.0045e-05,
      "loss": 0.4599,
      "step": 4012
    },
    {
      "epoch": 0.22471721357374846,
      "grad_norm": 1.0342845916748047,
      "learning_rate": 2.0050000000000003e-05,
      "loss": 0.2756,
      "step": 4013
    },
    {
      "epoch": 0.22477321088587748,
      "grad_norm": 1.1289149522781372,
      "learning_rate": 2.0055e-05,
      "loss": 0.2926,
      "step": 4014
    },
    {
      "epoch": 0.2248292081980065,
      "grad_norm": 1.4004709720611572,
      "learning_rate": 2.006e-05,
      "loss": 0.4329,
      "step": 4015
    },
    {
      "epoch": 0.22488520551013552,
      "grad_norm": 1.1849826574325562,
      "learning_rate": 2.0065000000000002e-05,
      "loss": 0.4633,
      "step": 4016
    },
    {
      "epoch": 0.22494120282226454,
      "grad_norm": 1.387434482574463,
      "learning_rate": 2.007e-05,
      "loss": 0.535,
      "step": 4017
    },
    {
      "epoch": 0.22499720013439356,
      "grad_norm": 1.1173462867736816,
      "learning_rate": 2.0075000000000003e-05,
      "loss": 0.385,
      "step": 4018
    },
    {
      "epoch": 0.22505319744652258,
      "grad_norm": 1.1251437664031982,
      "learning_rate": 2.008e-05,
      "loss": 0.4063,
      "step": 4019
    },
    {
      "epoch": 0.2251091947586516,
      "grad_norm": 1.127577304840088,
      "learning_rate": 2.0085e-05,
      "loss": 0.3449,
      "step": 4020
    },
    {
      "epoch": 0.2251651920707806,
      "grad_norm": 1.073022723197937,
      "learning_rate": 2.009e-05,
      "loss": 0.4331,
      "step": 4021
    },
    {
      "epoch": 0.2252211893829096,
      "grad_norm": 1.1744555234909058,
      "learning_rate": 2.0095e-05,
      "loss": 0.3727,
      "step": 4022
    },
    {
      "epoch": 0.22527718669503863,
      "grad_norm": 1.1534401178359985,
      "learning_rate": 2.01e-05,
      "loss": 0.4825,
      "step": 4023
    },
    {
      "epoch": 0.22533318400716765,
      "grad_norm": 3.915041208267212,
      "learning_rate": 2.0105e-05,
      "loss": 0.3737,
      "step": 4024
    },
    {
      "epoch": 0.22538918131929667,
      "grad_norm": 1.0764929056167603,
      "learning_rate": 2.0110000000000002e-05,
      "loss": 0.3697,
      "step": 4025
    },
    {
      "epoch": 0.22544517863142569,
      "grad_norm": 1.2173855304718018,
      "learning_rate": 2.0115e-05,
      "loss": 0.4043,
      "step": 4026
    },
    {
      "epoch": 0.2255011759435547,
      "grad_norm": 1.309613585472107,
      "learning_rate": 2.012e-05,
      "loss": 0.4685,
      "step": 4027
    },
    {
      "epoch": 0.22555717325568372,
      "grad_norm": 1.1317449808120728,
      "learning_rate": 2.0125e-05,
      "loss": 0.5362,
      "step": 4028
    },
    {
      "epoch": 0.22561317056781274,
      "grad_norm": 1.1389638185501099,
      "learning_rate": 2.0130000000000002e-05,
      "loss": 0.336,
      "step": 4029
    },
    {
      "epoch": 0.22566916787994176,
      "grad_norm": 1.1004079580307007,
      "learning_rate": 2.0135e-05,
      "loss": 0.3205,
      "step": 4030
    },
    {
      "epoch": 0.22572516519207078,
      "grad_norm": 1.545257329940796,
      "learning_rate": 2.014e-05,
      "loss": 0.5128,
      "step": 4031
    },
    {
      "epoch": 0.2257811625041998,
      "grad_norm": 1.1331626176834106,
      "learning_rate": 2.0145e-05,
      "loss": 0.4678,
      "step": 4032
    },
    {
      "epoch": 0.22583715981632882,
      "grad_norm": 1.08872652053833,
      "learning_rate": 2.0150000000000002e-05,
      "loss": 0.4114,
      "step": 4033
    },
    {
      "epoch": 0.22589315712845784,
      "grad_norm": 1.0941650867462158,
      "learning_rate": 2.0155000000000003e-05,
      "loss": 0.4096,
      "step": 4034
    },
    {
      "epoch": 0.22594915444058686,
      "grad_norm": 1.186246633529663,
      "learning_rate": 2.016e-05,
      "loss": 0.407,
      "step": 4035
    },
    {
      "epoch": 0.22600515175271588,
      "grad_norm": 1.2471370697021484,
      "learning_rate": 2.0165e-05,
      "loss": 0.4056,
      "step": 4036
    },
    {
      "epoch": 0.2260611490648449,
      "grad_norm": 1.18431556224823,
      "learning_rate": 2.017e-05,
      "loss": 0.3996,
      "step": 4037
    },
    {
      "epoch": 0.22611714637697392,
      "grad_norm": 1.24180006980896,
      "learning_rate": 2.0175000000000003e-05,
      "loss": 0.4061,
      "step": 4038
    },
    {
      "epoch": 0.22617314368910293,
      "grad_norm": 1.2733039855957031,
      "learning_rate": 2.0180000000000003e-05,
      "loss": 0.4062,
      "step": 4039
    },
    {
      "epoch": 0.22622914100123195,
      "grad_norm": 1.229092001914978,
      "learning_rate": 2.0185e-05,
      "loss": 0.4002,
      "step": 4040
    },
    {
      "epoch": 0.22628513831336095,
      "grad_norm": 1.216475009918213,
      "learning_rate": 2.019e-05,
      "loss": 0.4704,
      "step": 4041
    },
    {
      "epoch": 0.22634113562548996,
      "grad_norm": 1.1875768899917603,
      "learning_rate": 2.0195e-05,
      "loss": 0.3753,
      "step": 4042
    },
    {
      "epoch": 0.22639713293761898,
      "grad_norm": 1.3118387460708618,
      "learning_rate": 2.0200000000000003e-05,
      "loss": 0.3932,
      "step": 4043
    },
    {
      "epoch": 0.226453130249748,
      "grad_norm": 1.1569195985794067,
      "learning_rate": 2.0205e-05,
      "loss": 0.4384,
      "step": 4044
    },
    {
      "epoch": 0.22650912756187702,
      "grad_norm": 1.3034249544143677,
      "learning_rate": 2.021e-05,
      "loss": 0.4687,
      "step": 4045
    },
    {
      "epoch": 0.22656512487400604,
      "grad_norm": 1.364123821258545,
      "learning_rate": 2.0215000000000002e-05,
      "loss": 0.4705,
      "step": 4046
    },
    {
      "epoch": 0.22662112218613506,
      "grad_norm": 1.1235162019729614,
      "learning_rate": 2.022e-05,
      "loss": 0.373,
      "step": 4047
    },
    {
      "epoch": 0.22667711949826408,
      "grad_norm": 1.688078761100769,
      "learning_rate": 2.0225000000000004e-05,
      "loss": 0.4103,
      "step": 4048
    },
    {
      "epoch": 0.2267331168103931,
      "grad_norm": 1.3599395751953125,
      "learning_rate": 2.023e-05,
      "loss": 0.4554,
      "step": 4049
    },
    {
      "epoch": 0.22678911412252212,
      "grad_norm": 1.1683887243270874,
      "learning_rate": 2.0235000000000002e-05,
      "loss": 0.5024,
      "step": 4050
    },
    {
      "epoch": 0.22684511143465114,
      "grad_norm": 1.1350637674331665,
      "learning_rate": 2.024e-05,
      "loss": 0.3886,
      "step": 4051
    },
    {
      "epoch": 0.22690110874678016,
      "grad_norm": 1.2215251922607422,
      "learning_rate": 2.0245e-05,
      "loss": 0.4779,
      "step": 4052
    },
    {
      "epoch": 0.22695710605890917,
      "grad_norm": 1.2877695560455322,
      "learning_rate": 2.025e-05,
      "loss": 0.5126,
      "step": 4053
    },
    {
      "epoch": 0.2270131033710382,
      "grad_norm": 1.397469162940979,
      "learning_rate": 2.0255000000000002e-05,
      "loss": 0.5584,
      "step": 4054
    },
    {
      "epoch": 0.2270691006831672,
      "grad_norm": 1.0369408130645752,
      "learning_rate": 2.0260000000000003e-05,
      "loss": 0.3528,
      "step": 4055
    },
    {
      "epoch": 0.22712509799529623,
      "grad_norm": 1.805389165878296,
      "learning_rate": 2.0265e-05,
      "loss": 0.4645,
      "step": 4056
    },
    {
      "epoch": 0.22718109530742525,
      "grad_norm": 1.299943208694458,
      "learning_rate": 2.027e-05,
      "loss": 0.4551,
      "step": 4057
    },
    {
      "epoch": 0.22723709261955427,
      "grad_norm": 1.1022294759750366,
      "learning_rate": 2.0275e-05,
      "loss": 0.3104,
      "step": 4058
    },
    {
      "epoch": 0.2272930899316833,
      "grad_norm": 1.477035641670227,
      "learning_rate": 2.0280000000000002e-05,
      "loss": 0.4018,
      "step": 4059
    },
    {
      "epoch": 0.2273490872438123,
      "grad_norm": 1.0145372152328491,
      "learning_rate": 2.0285e-05,
      "loss": 0.3613,
      "step": 4060
    },
    {
      "epoch": 0.22740508455594133,
      "grad_norm": 1.1505661010742188,
      "learning_rate": 2.029e-05,
      "loss": 0.4782,
      "step": 4061
    },
    {
      "epoch": 0.22746108186807032,
      "grad_norm": 1.1380447149276733,
      "learning_rate": 2.0295e-05,
      "loss": 0.3925,
      "step": 4062
    },
    {
      "epoch": 0.22751707918019934,
      "grad_norm": 1.3335957527160645,
      "learning_rate": 2.0300000000000002e-05,
      "loss": 0.3813,
      "step": 4063
    },
    {
      "epoch": 0.22757307649232836,
      "grad_norm": 1.0564080476760864,
      "learning_rate": 2.0305000000000003e-05,
      "loss": 0.4491,
      "step": 4064
    },
    {
      "epoch": 0.22762907380445738,
      "grad_norm": 1.1848081350326538,
      "learning_rate": 2.031e-05,
      "loss": 0.4345,
      "step": 4065
    },
    {
      "epoch": 0.2276850711165864,
      "grad_norm": 1.1922739744186401,
      "learning_rate": 2.0315e-05,
      "loss": 0.4108,
      "step": 4066
    },
    {
      "epoch": 0.22774106842871542,
      "grad_norm": 1.2298710346221924,
      "learning_rate": 2.032e-05,
      "loss": 0.474,
      "step": 4067
    },
    {
      "epoch": 0.22779706574084443,
      "grad_norm": 1.4893807172775269,
      "learning_rate": 2.0325e-05,
      "loss": 0.567,
      "step": 4068
    },
    {
      "epoch": 0.22785306305297345,
      "grad_norm": 1.437177062034607,
      "learning_rate": 2.033e-05,
      "loss": 0.4056,
      "step": 4069
    },
    {
      "epoch": 0.22790906036510247,
      "grad_norm": 1.269019365310669,
      "learning_rate": 2.0335e-05,
      "loss": 0.4033,
      "step": 4070
    },
    {
      "epoch": 0.2279650576772315,
      "grad_norm": 1.282313585281372,
      "learning_rate": 2.0340000000000002e-05,
      "loss": 0.4129,
      "step": 4071
    },
    {
      "epoch": 0.2280210549893605,
      "grad_norm": 1.614296317100525,
      "learning_rate": 2.0345e-05,
      "loss": 0.4413,
      "step": 4072
    },
    {
      "epoch": 0.22807705230148953,
      "grad_norm": 1.1022065877914429,
      "learning_rate": 2.035e-05,
      "loss": 0.2764,
      "step": 4073
    },
    {
      "epoch": 0.22813304961361855,
      "grad_norm": 1.1078407764434814,
      "learning_rate": 2.0355e-05,
      "loss": 0.4086,
      "step": 4074
    },
    {
      "epoch": 0.22818904692574757,
      "grad_norm": 1.0584485530853271,
      "learning_rate": 2.036e-05,
      "loss": 0.4224,
      "step": 4075
    },
    {
      "epoch": 0.2282450442378766,
      "grad_norm": 11.007890701293945,
      "learning_rate": 2.0365000000000002e-05,
      "loss": 0.6179,
      "step": 4076
    },
    {
      "epoch": 0.2283010415500056,
      "grad_norm": 1.19613778591156,
      "learning_rate": 2.037e-05,
      "loss": 0.4075,
      "step": 4077
    },
    {
      "epoch": 0.22835703886213463,
      "grad_norm": 1.2499363422393799,
      "learning_rate": 2.0375e-05,
      "loss": 0.3791,
      "step": 4078
    },
    {
      "epoch": 0.22841303617426365,
      "grad_norm": 1.310246229171753,
      "learning_rate": 2.038e-05,
      "loss": 0.4022,
      "step": 4079
    },
    {
      "epoch": 0.22846903348639266,
      "grad_norm": 1.4538041353225708,
      "learning_rate": 2.0385000000000002e-05,
      "loss": 0.5483,
      "step": 4080
    },
    {
      "epoch": 0.22852503079852168,
      "grad_norm": 1.1844452619552612,
      "learning_rate": 2.039e-05,
      "loss": 0.4425,
      "step": 4081
    },
    {
      "epoch": 0.2285810281106507,
      "grad_norm": 1.2127856016159058,
      "learning_rate": 2.0395e-05,
      "loss": 0.4141,
      "step": 4082
    },
    {
      "epoch": 0.2286370254227797,
      "grad_norm": 1.1689798831939697,
      "learning_rate": 2.04e-05,
      "loss": 0.4201,
      "step": 4083
    },
    {
      "epoch": 0.2286930227349087,
      "grad_norm": 1.2544913291931152,
      "learning_rate": 2.0405000000000002e-05,
      "loss": 0.5006,
      "step": 4084
    },
    {
      "epoch": 0.22874902004703773,
      "grad_norm": 1.3648995161056519,
      "learning_rate": 2.0410000000000003e-05,
      "loss": 0.386,
      "step": 4085
    },
    {
      "epoch": 0.22880501735916675,
      "grad_norm": 1.1508864164352417,
      "learning_rate": 2.0415e-05,
      "loss": 0.364,
      "step": 4086
    },
    {
      "epoch": 0.22886101467129577,
      "grad_norm": 1.1088776588439941,
      "learning_rate": 2.042e-05,
      "loss": 0.397,
      "step": 4087
    },
    {
      "epoch": 0.2289170119834248,
      "grad_norm": 1.291202425956726,
      "learning_rate": 2.0425e-05,
      "loss": 0.4527,
      "step": 4088
    },
    {
      "epoch": 0.2289730092955538,
      "grad_norm": 1.4738807678222656,
      "learning_rate": 2.0430000000000003e-05,
      "loss": 0.4579,
      "step": 4089
    },
    {
      "epoch": 0.22902900660768283,
      "grad_norm": 1.0973998308181763,
      "learning_rate": 2.0435e-05,
      "loss": 0.2876,
      "step": 4090
    },
    {
      "epoch": 0.22908500391981185,
      "grad_norm": 1.024002194404602,
      "learning_rate": 2.044e-05,
      "loss": 0.4462,
      "step": 4091
    },
    {
      "epoch": 0.22914100123194087,
      "grad_norm": 1.2808518409729004,
      "learning_rate": 2.0445e-05,
      "loss": 0.3622,
      "step": 4092
    },
    {
      "epoch": 0.22919699854406989,
      "grad_norm": 1.0888512134552002,
      "learning_rate": 2.045e-05,
      "loss": 0.3681,
      "step": 4093
    },
    {
      "epoch": 0.2292529958561989,
      "grad_norm": 0.9593205451965332,
      "learning_rate": 2.0455000000000003e-05,
      "loss": 0.3167,
      "step": 4094
    },
    {
      "epoch": 0.22930899316832792,
      "grad_norm": 1.2199854850769043,
      "learning_rate": 2.046e-05,
      "loss": 0.441,
      "step": 4095
    },
    {
      "epoch": 0.22936499048045694,
      "grad_norm": 1.281033992767334,
      "learning_rate": 2.0465e-05,
      "loss": 0.5411,
      "step": 4096
    },
    {
      "epoch": 0.22942098779258596,
      "grad_norm": 1.123522162437439,
      "learning_rate": 2.047e-05,
      "loss": 0.3339,
      "step": 4097
    },
    {
      "epoch": 0.22947698510471498,
      "grad_norm": 0.994236409664154,
      "learning_rate": 2.0475e-05,
      "loss": 0.3565,
      "step": 4098
    },
    {
      "epoch": 0.229532982416844,
      "grad_norm": 1.1076732873916626,
      "learning_rate": 2.048e-05,
      "loss": 0.4364,
      "step": 4099
    },
    {
      "epoch": 0.22958897972897302,
      "grad_norm": 1.6202208995819092,
      "learning_rate": 2.0485e-05,
      "loss": 0.4889,
      "step": 4100
    },
    {
      "epoch": 0.22964497704110204,
      "grad_norm": 1.2983157634735107,
      "learning_rate": 2.0490000000000002e-05,
      "loss": 0.5233,
      "step": 4101
    },
    {
      "epoch": 0.22970097435323106,
      "grad_norm": 1.2962462902069092,
      "learning_rate": 2.0495e-05,
      "loss": 0.5092,
      "step": 4102
    },
    {
      "epoch": 0.22975697166536005,
      "grad_norm": 1.4778438806533813,
      "learning_rate": 2.05e-05,
      "loss": 0.5732,
      "step": 4103
    },
    {
      "epoch": 0.22981296897748907,
      "grad_norm": 1.2814544439315796,
      "learning_rate": 2.0505e-05,
      "loss": 0.4424,
      "step": 4104
    },
    {
      "epoch": 0.2298689662896181,
      "grad_norm": 1.4536662101745605,
      "learning_rate": 2.0510000000000002e-05,
      "loss": 0.5741,
      "step": 4105
    },
    {
      "epoch": 0.2299249636017471,
      "grad_norm": 1.1617083549499512,
      "learning_rate": 2.0515e-05,
      "loss": 0.4934,
      "step": 4106
    },
    {
      "epoch": 0.22998096091387613,
      "grad_norm": 1.3995189666748047,
      "learning_rate": 2.052e-05,
      "loss": 0.3733,
      "step": 4107
    },
    {
      "epoch": 0.23003695822600514,
      "grad_norm": 0.9614766836166382,
      "learning_rate": 2.0525e-05,
      "loss": 0.3484,
      "step": 4108
    },
    {
      "epoch": 0.23009295553813416,
      "grad_norm": 1.136983036994934,
      "learning_rate": 2.053e-05,
      "loss": 0.3716,
      "step": 4109
    },
    {
      "epoch": 0.23014895285026318,
      "grad_norm": 1.257845163345337,
      "learning_rate": 2.0535000000000002e-05,
      "loss": 0.4114,
      "step": 4110
    },
    {
      "epoch": 0.2302049501623922,
      "grad_norm": 1.6189639568328857,
      "learning_rate": 2.054e-05,
      "loss": 0.6255,
      "step": 4111
    },
    {
      "epoch": 0.23026094747452122,
      "grad_norm": 1.0244381427764893,
      "learning_rate": 2.0545e-05,
      "loss": 0.2547,
      "step": 4112
    },
    {
      "epoch": 0.23031694478665024,
      "grad_norm": 1.1978487968444824,
      "learning_rate": 2.055e-05,
      "loss": 0.3887,
      "step": 4113
    },
    {
      "epoch": 0.23037294209877926,
      "grad_norm": 1.0918502807617188,
      "learning_rate": 2.0555000000000002e-05,
      "loss": 0.3757,
      "step": 4114
    },
    {
      "epoch": 0.23042893941090828,
      "grad_norm": 1.8906546831130981,
      "learning_rate": 2.0560000000000003e-05,
      "loss": 0.4998,
      "step": 4115
    },
    {
      "epoch": 0.2304849367230373,
      "grad_norm": 1.2287076711654663,
      "learning_rate": 2.0565e-05,
      "loss": 0.4845,
      "step": 4116
    },
    {
      "epoch": 0.23054093403516632,
      "grad_norm": 1.193910002708435,
      "learning_rate": 2.057e-05,
      "loss": 0.3618,
      "step": 4117
    },
    {
      "epoch": 0.23059693134729534,
      "grad_norm": 1.2260141372680664,
      "learning_rate": 2.0575e-05,
      "loss": 0.414,
      "step": 4118
    },
    {
      "epoch": 0.23065292865942436,
      "grad_norm": 1.1606512069702148,
      "learning_rate": 2.0580000000000003e-05,
      "loss": 0.3471,
      "step": 4119
    },
    {
      "epoch": 0.23070892597155337,
      "grad_norm": 1.0225188732147217,
      "learning_rate": 2.0585e-05,
      "loss": 0.4305,
      "step": 4120
    },
    {
      "epoch": 0.2307649232836824,
      "grad_norm": 1.2030072212219238,
      "learning_rate": 2.059e-05,
      "loss": 0.3368,
      "step": 4121
    },
    {
      "epoch": 0.2308209205958114,
      "grad_norm": 1.2328826189041138,
      "learning_rate": 2.0595000000000002e-05,
      "loss": 0.5293,
      "step": 4122
    },
    {
      "epoch": 0.23087691790794043,
      "grad_norm": 1.2912899255752563,
      "learning_rate": 2.06e-05,
      "loss": 0.5007,
      "step": 4123
    },
    {
      "epoch": 0.23093291522006942,
      "grad_norm": 1.3425554037094116,
      "learning_rate": 2.0605000000000003e-05,
      "loss": 0.5251,
      "step": 4124
    },
    {
      "epoch": 0.23098891253219844,
      "grad_norm": 1.1544756889343262,
      "learning_rate": 2.061e-05,
      "loss": 0.3617,
      "step": 4125
    },
    {
      "epoch": 0.23104490984432746,
      "grad_norm": 1.1377413272857666,
      "learning_rate": 2.0615e-05,
      "loss": 0.3603,
      "step": 4126
    },
    {
      "epoch": 0.23110090715645648,
      "grad_norm": 1.1988937854766846,
      "learning_rate": 2.062e-05,
      "loss": 0.473,
      "step": 4127
    },
    {
      "epoch": 0.2311569044685855,
      "grad_norm": 1.0015439987182617,
      "learning_rate": 2.0625e-05,
      "loss": 0.4049,
      "step": 4128
    },
    {
      "epoch": 0.23121290178071452,
      "grad_norm": 1.2456467151641846,
      "learning_rate": 2.063e-05,
      "loss": 0.403,
      "step": 4129
    },
    {
      "epoch": 0.23126889909284354,
      "grad_norm": 1.124685287475586,
      "learning_rate": 2.0635e-05,
      "loss": 0.405,
      "step": 4130
    },
    {
      "epoch": 0.23132489640497256,
      "grad_norm": 1.2655459642410278,
      "learning_rate": 2.0640000000000002e-05,
      "loss": 0.354,
      "step": 4131
    },
    {
      "epoch": 0.23138089371710158,
      "grad_norm": 1.039529800415039,
      "learning_rate": 2.0645e-05,
      "loss": 0.354,
      "step": 4132
    },
    {
      "epoch": 0.2314368910292306,
      "grad_norm": 1.238182783126831,
      "learning_rate": 2.065e-05,
      "loss": 0.3983,
      "step": 4133
    },
    {
      "epoch": 0.23149288834135962,
      "grad_norm": 1.2212363481521606,
      "learning_rate": 2.0655e-05,
      "loss": 0.4486,
      "step": 4134
    },
    {
      "epoch": 0.23154888565348863,
      "grad_norm": 1.1392161846160889,
      "learning_rate": 2.0660000000000002e-05,
      "loss": 0.3777,
      "step": 4135
    },
    {
      "epoch": 0.23160488296561765,
      "grad_norm": 1.199353575706482,
      "learning_rate": 2.0665e-05,
      "loss": 0.2687,
      "step": 4136
    },
    {
      "epoch": 0.23166088027774667,
      "grad_norm": 1.4285528659820557,
      "learning_rate": 2.067e-05,
      "loss": 0.2833,
      "step": 4137
    },
    {
      "epoch": 0.2317168775898757,
      "grad_norm": 1.139581561088562,
      "learning_rate": 2.0675e-05,
      "loss": 0.4413,
      "step": 4138
    },
    {
      "epoch": 0.2317728749020047,
      "grad_norm": 1.2959959506988525,
      "learning_rate": 2.0680000000000002e-05,
      "loss": 0.4027,
      "step": 4139
    },
    {
      "epoch": 0.23182887221413373,
      "grad_norm": 1.315169095993042,
      "learning_rate": 2.0685000000000003e-05,
      "loss": 0.4878,
      "step": 4140
    },
    {
      "epoch": 0.23188486952626275,
      "grad_norm": 1.1356548070907593,
      "learning_rate": 2.069e-05,
      "loss": 0.4531,
      "step": 4141
    },
    {
      "epoch": 0.23194086683839177,
      "grad_norm": 1.150536060333252,
      "learning_rate": 2.0695e-05,
      "loss": 0.4032,
      "step": 4142
    },
    {
      "epoch": 0.2319968641505208,
      "grad_norm": 1.3828458786010742,
      "learning_rate": 2.07e-05,
      "loss": 0.6142,
      "step": 4143
    },
    {
      "epoch": 0.2320528614626498,
      "grad_norm": 1.0190892219543457,
      "learning_rate": 2.0705000000000003e-05,
      "loss": 0.4076,
      "step": 4144
    },
    {
      "epoch": 0.2321088587747788,
      "grad_norm": 1.242693543434143,
      "learning_rate": 2.0710000000000003e-05,
      "loss": 0.4358,
      "step": 4145
    },
    {
      "epoch": 0.23216485608690782,
      "grad_norm": 1.4580390453338623,
      "learning_rate": 2.0715e-05,
      "loss": 0.5017,
      "step": 4146
    },
    {
      "epoch": 0.23222085339903684,
      "grad_norm": 1.630743384361267,
      "learning_rate": 2.072e-05,
      "loss": 0.5403,
      "step": 4147
    },
    {
      "epoch": 0.23227685071116586,
      "grad_norm": 1.3127927780151367,
      "learning_rate": 2.0725e-05,
      "loss": 0.5772,
      "step": 4148
    },
    {
      "epoch": 0.23233284802329487,
      "grad_norm": 1.32776939868927,
      "learning_rate": 2.0730000000000003e-05,
      "loss": 0.4103,
      "step": 4149
    },
    {
      "epoch": 0.2323888453354239,
      "grad_norm": 1.1280535459518433,
      "learning_rate": 2.0735e-05,
      "loss": 0.3944,
      "step": 4150
    },
    {
      "epoch": 0.2324448426475529,
      "grad_norm": 1.0871868133544922,
      "learning_rate": 2.074e-05,
      "loss": 0.4578,
      "step": 4151
    },
    {
      "epoch": 0.23250083995968193,
      "grad_norm": 1.211596965789795,
      "learning_rate": 2.0745000000000002e-05,
      "loss": 0.3587,
      "step": 4152
    },
    {
      "epoch": 0.23255683727181095,
      "grad_norm": 1.4957422018051147,
      "learning_rate": 2.075e-05,
      "loss": 0.4731,
      "step": 4153
    },
    {
      "epoch": 0.23261283458393997,
      "grad_norm": 1.413071632385254,
      "learning_rate": 2.0755000000000004e-05,
      "loss": 0.5488,
      "step": 4154
    },
    {
      "epoch": 0.232668831896069,
      "grad_norm": 1.5196342468261719,
      "learning_rate": 2.076e-05,
      "loss": 0.4164,
      "step": 4155
    },
    {
      "epoch": 0.232724829208198,
      "grad_norm": 4.492232799530029,
      "learning_rate": 2.0765000000000002e-05,
      "loss": 0.589,
      "step": 4156
    },
    {
      "epoch": 0.23278082652032703,
      "grad_norm": 1.2077269554138184,
      "learning_rate": 2.077e-05,
      "loss": 0.3548,
      "step": 4157
    },
    {
      "epoch": 0.23283682383245605,
      "grad_norm": 0.9782938957214355,
      "learning_rate": 2.0775e-05,
      "loss": 0.301,
      "step": 4158
    },
    {
      "epoch": 0.23289282114458507,
      "grad_norm": 1.1544853448867798,
      "learning_rate": 2.078e-05,
      "loss": 0.428,
      "step": 4159
    },
    {
      "epoch": 0.23294881845671409,
      "grad_norm": 1.3924914598464966,
      "learning_rate": 2.0785000000000002e-05,
      "loss": 0.3908,
      "step": 4160
    },
    {
      "epoch": 0.2330048157688431,
      "grad_norm": 1.404755711555481,
      "learning_rate": 2.0790000000000003e-05,
      "loss": 0.5763,
      "step": 4161
    },
    {
      "epoch": 0.23306081308097212,
      "grad_norm": 1.3203518390655518,
      "learning_rate": 2.0795e-05,
      "loss": 0.4953,
      "step": 4162
    },
    {
      "epoch": 0.23311681039310114,
      "grad_norm": 1.2612299919128418,
      "learning_rate": 2.08e-05,
      "loss": 0.3418,
      "step": 4163
    },
    {
      "epoch": 0.23317280770523016,
      "grad_norm": 1.3685733079910278,
      "learning_rate": 2.0805e-05,
      "loss": 0.4777,
      "step": 4164
    },
    {
      "epoch": 0.23322880501735915,
      "grad_norm": 1.4660611152648926,
      "learning_rate": 2.0810000000000002e-05,
      "loss": 0.5761,
      "step": 4165
    },
    {
      "epoch": 0.23328480232948817,
      "grad_norm": 2.6750452518463135,
      "learning_rate": 2.0815e-05,
      "loss": 0.3986,
      "step": 4166
    },
    {
      "epoch": 0.2333407996416172,
      "grad_norm": 1.106176495552063,
      "learning_rate": 2.082e-05,
      "loss": 0.4198,
      "step": 4167
    },
    {
      "epoch": 0.2333967969537462,
      "grad_norm": 1.3173162937164307,
      "learning_rate": 2.0825e-05,
      "loss": 0.6655,
      "step": 4168
    },
    {
      "epoch": 0.23345279426587523,
      "grad_norm": 1.1135119199752808,
      "learning_rate": 2.0830000000000002e-05,
      "loss": 0.4669,
      "step": 4169
    },
    {
      "epoch": 0.23350879157800425,
      "grad_norm": 1.1758356094360352,
      "learning_rate": 2.0835000000000003e-05,
      "loss": 0.4082,
      "step": 4170
    },
    {
      "epoch": 0.23356478889013327,
      "grad_norm": 1.1846275329589844,
      "learning_rate": 2.084e-05,
      "loss": 0.4446,
      "step": 4171
    },
    {
      "epoch": 0.2336207862022623,
      "grad_norm": 1.242761492729187,
      "learning_rate": 2.0845e-05,
      "loss": 0.3611,
      "step": 4172
    },
    {
      "epoch": 0.2336767835143913,
      "grad_norm": 1.046606183052063,
      "learning_rate": 2.085e-05,
      "loss": 0.424,
      "step": 4173
    },
    {
      "epoch": 0.23373278082652033,
      "grad_norm": 3.2558372020721436,
      "learning_rate": 2.0855000000000003e-05,
      "loss": 0.4671,
      "step": 4174
    },
    {
      "epoch": 0.23378877813864934,
      "grad_norm": 1.3819587230682373,
      "learning_rate": 2.086e-05,
      "loss": 0.4429,
      "step": 4175
    },
    {
      "epoch": 0.23384477545077836,
      "grad_norm": 0.9712210297584534,
      "learning_rate": 2.0865e-05,
      "loss": 0.3406,
      "step": 4176
    },
    {
      "epoch": 0.23390077276290738,
      "grad_norm": 1.188122034072876,
      "learning_rate": 2.0870000000000002e-05,
      "loss": 0.4788,
      "step": 4177
    },
    {
      "epoch": 0.2339567700750364,
      "grad_norm": 1.0752432346343994,
      "learning_rate": 2.0875e-05,
      "loss": 0.471,
      "step": 4178
    },
    {
      "epoch": 0.23401276738716542,
      "grad_norm": 1.4987022876739502,
      "learning_rate": 2.0880000000000003e-05,
      "loss": 0.4276,
      "step": 4179
    },
    {
      "epoch": 0.23406876469929444,
      "grad_norm": 1.4521417617797852,
      "learning_rate": 2.0885e-05,
      "loss": 0.3481,
      "step": 4180
    },
    {
      "epoch": 0.23412476201142346,
      "grad_norm": 1.2257338762283325,
      "learning_rate": 2.089e-05,
      "loss": 0.4603,
      "step": 4181
    },
    {
      "epoch": 0.23418075932355248,
      "grad_norm": 1.2818821668624878,
      "learning_rate": 2.0895e-05,
      "loss": 0.3515,
      "step": 4182
    },
    {
      "epoch": 0.2342367566356815,
      "grad_norm": 1.5747284889221191,
      "learning_rate": 2.09e-05,
      "loss": 0.4984,
      "step": 4183
    },
    {
      "epoch": 0.23429275394781052,
      "grad_norm": 1.2921171188354492,
      "learning_rate": 2.0905000000000004e-05,
      "loss": 0.5154,
      "step": 4184
    },
    {
      "epoch": 0.23434875125993954,
      "grad_norm": 1.3256133794784546,
      "learning_rate": 2.091e-05,
      "loss": 0.4911,
      "step": 4185
    },
    {
      "epoch": 0.23440474857206853,
      "grad_norm": 1.235826015472412,
      "learning_rate": 2.0915000000000002e-05,
      "loss": 0.3948,
      "step": 4186
    },
    {
      "epoch": 0.23446074588419755,
      "grad_norm": 1.1025710105895996,
      "learning_rate": 2.092e-05,
      "loss": 0.3685,
      "step": 4187
    },
    {
      "epoch": 0.23451674319632657,
      "grad_norm": 1.6832563877105713,
      "learning_rate": 2.0925e-05,
      "loss": 0.4591,
      "step": 4188
    },
    {
      "epoch": 0.23457274050845558,
      "grad_norm": 1.313822627067566,
      "learning_rate": 2.093e-05,
      "loss": 0.483,
      "step": 4189
    },
    {
      "epoch": 0.2346287378205846,
      "grad_norm": 1.299112319946289,
      "learning_rate": 2.0935000000000002e-05,
      "loss": 0.5008,
      "step": 4190
    },
    {
      "epoch": 0.23468473513271362,
      "grad_norm": 1.257529616355896,
      "learning_rate": 2.0940000000000003e-05,
      "loss": 0.5007,
      "step": 4191
    },
    {
      "epoch": 0.23474073244484264,
      "grad_norm": 3.786053419113159,
      "learning_rate": 2.0945e-05,
      "loss": 0.4312,
      "step": 4192
    },
    {
      "epoch": 0.23479672975697166,
      "grad_norm": 1.1573549509048462,
      "learning_rate": 2.095e-05,
      "loss": 0.3606,
      "step": 4193
    },
    {
      "epoch": 0.23485272706910068,
      "grad_norm": 1.4084893465042114,
      "learning_rate": 2.0955e-05,
      "loss": 0.5302,
      "step": 4194
    },
    {
      "epoch": 0.2349087243812297,
      "grad_norm": 1.0694756507873535,
      "learning_rate": 2.0960000000000003e-05,
      "loss": 0.309,
      "step": 4195
    },
    {
      "epoch": 0.23496472169335872,
      "grad_norm": 1.0330634117126465,
      "learning_rate": 2.0965e-05,
      "loss": 0.3656,
      "step": 4196
    },
    {
      "epoch": 0.23502071900548774,
      "grad_norm": 1.2291839122772217,
      "learning_rate": 2.097e-05,
      "loss": 0.3507,
      "step": 4197
    },
    {
      "epoch": 0.23507671631761676,
      "grad_norm": 1.1577256917953491,
      "learning_rate": 2.0975e-05,
      "loss": 0.3888,
      "step": 4198
    },
    {
      "epoch": 0.23513271362974578,
      "grad_norm": 1.1568584442138672,
      "learning_rate": 2.098e-05,
      "loss": 0.5337,
      "step": 4199
    },
    {
      "epoch": 0.2351887109418748,
      "grad_norm": 1.1812026500701904,
      "learning_rate": 2.0985000000000003e-05,
      "loss": 0.4119,
      "step": 4200
    },
    {
      "epoch": 0.23524470825400381,
      "grad_norm": 1.2275822162628174,
      "learning_rate": 2.099e-05,
      "loss": 0.353,
      "step": 4201
    },
    {
      "epoch": 0.23530070556613283,
      "grad_norm": 1.1405919790267944,
      "learning_rate": 2.0995e-05,
      "loss": 0.4058,
      "step": 4202
    },
    {
      "epoch": 0.23535670287826185,
      "grad_norm": 1.1904031038284302,
      "learning_rate": 2.1e-05,
      "loss": 0.3385,
      "step": 4203
    },
    {
      "epoch": 0.23541270019039087,
      "grad_norm": 1.063535213470459,
      "learning_rate": 2.1005e-05,
      "loss": 0.3502,
      "step": 4204
    },
    {
      "epoch": 0.2354686975025199,
      "grad_norm": 1.3525657653808594,
      "learning_rate": 2.101e-05,
      "loss": 0.5297,
      "step": 4205
    },
    {
      "epoch": 0.2355246948146489,
      "grad_norm": 1.166908621788025,
      "learning_rate": 2.1015e-05,
      "loss": 0.3891,
      "step": 4206
    },
    {
      "epoch": 0.2355806921267779,
      "grad_norm": 1.3686550855636597,
      "learning_rate": 2.1020000000000002e-05,
      "loss": 0.497,
      "step": 4207
    },
    {
      "epoch": 0.23563668943890692,
      "grad_norm": 1.311750888824463,
      "learning_rate": 2.1025e-05,
      "loss": 0.4479,
      "step": 4208
    },
    {
      "epoch": 0.23569268675103594,
      "grad_norm": 1.2289323806762695,
      "learning_rate": 2.103e-05,
      "loss": 0.4434,
      "step": 4209
    },
    {
      "epoch": 0.23574868406316496,
      "grad_norm": 1.266406536102295,
      "learning_rate": 2.1035e-05,
      "loss": 0.4205,
      "step": 4210
    },
    {
      "epoch": 0.23580468137529398,
      "grad_norm": 1.1448001861572266,
      "learning_rate": 2.1040000000000002e-05,
      "loss": 0.3897,
      "step": 4211
    },
    {
      "epoch": 0.235860678687423,
      "grad_norm": 1.3236624002456665,
      "learning_rate": 2.1045e-05,
      "loss": 0.4088,
      "step": 4212
    },
    {
      "epoch": 0.23591667599955202,
      "grad_norm": 1.2581989765167236,
      "learning_rate": 2.105e-05,
      "loss": 0.379,
      "step": 4213
    },
    {
      "epoch": 0.23597267331168104,
      "grad_norm": 1.0380505323410034,
      "learning_rate": 2.1055e-05,
      "loss": 0.3879,
      "step": 4214
    },
    {
      "epoch": 0.23602867062381006,
      "grad_norm": 1.1456139087677002,
      "learning_rate": 2.106e-05,
      "loss": 0.4532,
      "step": 4215
    },
    {
      "epoch": 0.23608466793593907,
      "grad_norm": 1.1638354063034058,
      "learning_rate": 2.1065000000000002e-05,
      "loss": 0.4703,
      "step": 4216
    },
    {
      "epoch": 0.2361406652480681,
      "grad_norm": 1.608616590499878,
      "learning_rate": 2.107e-05,
      "loss": 0.5295,
      "step": 4217
    },
    {
      "epoch": 0.2361966625601971,
      "grad_norm": 1.3021283149719238,
      "learning_rate": 2.1075e-05,
      "loss": 0.3815,
      "step": 4218
    },
    {
      "epoch": 0.23625265987232613,
      "grad_norm": 5.347463607788086,
      "learning_rate": 2.1079999999999998e-05,
      "loss": 0.3893,
      "step": 4219
    },
    {
      "epoch": 0.23630865718445515,
      "grad_norm": 1.0997823476791382,
      "learning_rate": 2.1085000000000002e-05,
      "loss": 0.3264,
      "step": 4220
    },
    {
      "epoch": 0.23636465449658417,
      "grad_norm": 1.0885735750198364,
      "learning_rate": 2.1090000000000003e-05,
      "loss": 0.3894,
      "step": 4221
    },
    {
      "epoch": 0.2364206518087132,
      "grad_norm": 1.0570690631866455,
      "learning_rate": 2.1095e-05,
      "loss": 0.374,
      "step": 4222
    },
    {
      "epoch": 0.2364766491208422,
      "grad_norm": 1.38384211063385,
      "learning_rate": 2.11e-05,
      "loss": 0.4647,
      "step": 4223
    },
    {
      "epoch": 0.23653264643297123,
      "grad_norm": 1.2475320100784302,
      "learning_rate": 2.1105e-05,
      "loss": 0.5061,
      "step": 4224
    },
    {
      "epoch": 0.23658864374510025,
      "grad_norm": 1.3273953199386597,
      "learning_rate": 2.1110000000000003e-05,
      "loss": 0.4165,
      "step": 4225
    },
    {
      "epoch": 0.23664464105722927,
      "grad_norm": 1.0664968490600586,
      "learning_rate": 2.1115e-05,
      "loss": 0.3721,
      "step": 4226
    },
    {
      "epoch": 0.23670063836935826,
      "grad_norm": 1.255656361579895,
      "learning_rate": 2.112e-05,
      "loss": 0.4174,
      "step": 4227
    },
    {
      "epoch": 0.23675663568148728,
      "grad_norm": 1.2060269117355347,
      "learning_rate": 2.1125000000000002e-05,
      "loss": 0.4766,
      "step": 4228
    },
    {
      "epoch": 0.2368126329936163,
      "grad_norm": 1.2741271257400513,
      "learning_rate": 2.113e-05,
      "loss": 0.4044,
      "step": 4229
    },
    {
      "epoch": 0.23686863030574531,
      "grad_norm": 1.3632087707519531,
      "learning_rate": 2.1135000000000003e-05,
      "loss": 0.4712,
      "step": 4230
    },
    {
      "epoch": 0.23692462761787433,
      "grad_norm": 1.3004961013793945,
      "learning_rate": 2.114e-05,
      "loss": 0.4695,
      "step": 4231
    },
    {
      "epoch": 0.23698062493000335,
      "grad_norm": 1.64434814453125,
      "learning_rate": 2.1145e-05,
      "loss": 0.463,
      "step": 4232
    },
    {
      "epoch": 0.23703662224213237,
      "grad_norm": 1.0939937829971313,
      "learning_rate": 2.115e-05,
      "loss": 0.4293,
      "step": 4233
    },
    {
      "epoch": 0.2370926195542614,
      "grad_norm": 1.2327302694320679,
      "learning_rate": 2.1155e-05,
      "loss": 0.4478,
      "step": 4234
    },
    {
      "epoch": 0.2371486168663904,
      "grad_norm": 1.1831134557724,
      "learning_rate": 2.116e-05,
      "loss": 0.4402,
      "step": 4235
    },
    {
      "epoch": 0.23720461417851943,
      "grad_norm": 1.0770121812820435,
      "learning_rate": 2.1165e-05,
      "loss": 0.6058,
      "step": 4236
    },
    {
      "epoch": 0.23726061149064845,
      "grad_norm": 1.3046811819076538,
      "learning_rate": 2.1170000000000002e-05,
      "loss": 0.4307,
      "step": 4237
    },
    {
      "epoch": 0.23731660880277747,
      "grad_norm": 1.263245701789856,
      "learning_rate": 2.1175e-05,
      "loss": 0.4071,
      "step": 4238
    },
    {
      "epoch": 0.2373726061149065,
      "grad_norm": 1.2050762176513672,
      "learning_rate": 2.118e-05,
      "loss": 0.5376,
      "step": 4239
    },
    {
      "epoch": 0.2374286034270355,
      "grad_norm": 1.1449065208435059,
      "learning_rate": 2.1185e-05,
      "loss": 0.4948,
      "step": 4240
    },
    {
      "epoch": 0.23748460073916453,
      "grad_norm": 1.1334929466247559,
      "learning_rate": 2.1190000000000002e-05,
      "loss": 0.3879,
      "step": 4241
    },
    {
      "epoch": 0.23754059805129354,
      "grad_norm": 1.0854896306991577,
      "learning_rate": 2.1195e-05,
      "loss": 0.3292,
      "step": 4242
    },
    {
      "epoch": 0.23759659536342256,
      "grad_norm": 1.4370067119598389,
      "learning_rate": 2.12e-05,
      "loss": 0.6236,
      "step": 4243
    },
    {
      "epoch": 0.23765259267555158,
      "grad_norm": 1.1262603998184204,
      "learning_rate": 2.1205e-05,
      "loss": 0.4683,
      "step": 4244
    },
    {
      "epoch": 0.2377085899876806,
      "grad_norm": 1.2429581880569458,
      "learning_rate": 2.1210000000000002e-05,
      "loss": 0.4212,
      "step": 4245
    },
    {
      "epoch": 0.23776458729980962,
      "grad_norm": 1.2421106100082397,
      "learning_rate": 2.1215000000000003e-05,
      "loss": 0.6051,
      "step": 4246
    },
    {
      "epoch": 0.23782058461193864,
      "grad_norm": 1.3536938428878784,
      "learning_rate": 2.122e-05,
      "loss": 0.5184,
      "step": 4247
    },
    {
      "epoch": 0.23787658192406763,
      "grad_norm": 1.1302820444107056,
      "learning_rate": 2.1225e-05,
      "loss": 0.4098,
      "step": 4248
    },
    {
      "epoch": 0.23793257923619665,
      "grad_norm": 1.2246413230895996,
      "learning_rate": 2.123e-05,
      "loss": 0.4006,
      "step": 4249
    },
    {
      "epoch": 0.23798857654832567,
      "grad_norm": 2.1079390048980713,
      "learning_rate": 2.1235000000000003e-05,
      "loss": 0.483,
      "step": 4250
    },
    {
      "epoch": 0.2380445738604547,
      "grad_norm": 1.4226256608963013,
      "learning_rate": 2.124e-05,
      "loss": 0.4082,
      "step": 4251
    },
    {
      "epoch": 0.2381005711725837,
      "grad_norm": 1.766867995262146,
      "learning_rate": 2.1245e-05,
      "loss": 0.3793,
      "step": 4252
    },
    {
      "epoch": 0.23815656848471273,
      "grad_norm": 1.7468613386154175,
      "learning_rate": 2.125e-05,
      "loss": 0.6336,
      "step": 4253
    },
    {
      "epoch": 0.23821256579684175,
      "grad_norm": 1.261499285697937,
      "learning_rate": 2.1255e-05,
      "loss": 0.4244,
      "step": 4254
    },
    {
      "epoch": 0.23826856310897077,
      "grad_norm": 2.6207706928253174,
      "learning_rate": 2.1260000000000003e-05,
      "loss": 0.4324,
      "step": 4255
    },
    {
      "epoch": 0.23832456042109978,
      "grad_norm": 1.1398931741714478,
      "learning_rate": 2.1265e-05,
      "loss": 0.4442,
      "step": 4256
    },
    {
      "epoch": 0.2383805577332288,
      "grad_norm": 1.4394181966781616,
      "learning_rate": 2.127e-05,
      "loss": 0.4277,
      "step": 4257
    },
    {
      "epoch": 0.23843655504535782,
      "grad_norm": 1.1773593425750732,
      "learning_rate": 2.1275000000000002e-05,
      "loss": 0.4644,
      "step": 4258
    },
    {
      "epoch": 0.23849255235748684,
      "grad_norm": 1.4122238159179688,
      "learning_rate": 2.128e-05,
      "loss": 0.5031,
      "step": 4259
    },
    {
      "epoch": 0.23854854966961586,
      "grad_norm": 1.206392765045166,
      "learning_rate": 2.1285000000000004e-05,
      "loss": 0.407,
      "step": 4260
    },
    {
      "epoch": 0.23860454698174488,
      "grad_norm": 1.189375638961792,
      "learning_rate": 2.129e-05,
      "loss": 0.3432,
      "step": 4261
    },
    {
      "epoch": 0.2386605442938739,
      "grad_norm": 1.1893759965896606,
      "learning_rate": 2.1295000000000002e-05,
      "loss": 0.4062,
      "step": 4262
    },
    {
      "epoch": 0.23871654160600292,
      "grad_norm": 1.5574356317520142,
      "learning_rate": 2.13e-05,
      "loss": 0.3862,
      "step": 4263
    },
    {
      "epoch": 0.23877253891813194,
      "grad_norm": 1.6357659101486206,
      "learning_rate": 2.1305e-05,
      "loss": 0.4128,
      "step": 4264
    },
    {
      "epoch": 0.23882853623026096,
      "grad_norm": 1.0961475372314453,
      "learning_rate": 2.131e-05,
      "loss": 0.303,
      "step": 4265
    },
    {
      "epoch": 0.23888453354238998,
      "grad_norm": 1.287390947341919,
      "learning_rate": 2.1315000000000002e-05,
      "loss": 0.5198,
      "step": 4266
    },
    {
      "epoch": 0.238940530854519,
      "grad_norm": 1.3111846446990967,
      "learning_rate": 2.1320000000000003e-05,
      "loss": 0.4619,
      "step": 4267
    },
    {
      "epoch": 0.23899652816664801,
      "grad_norm": 1.5583844184875488,
      "learning_rate": 2.1325e-05,
      "loss": 0.3147,
      "step": 4268
    },
    {
      "epoch": 0.239052525478777,
      "grad_norm": 1.2882252931594849,
      "learning_rate": 2.133e-05,
      "loss": 0.3848,
      "step": 4269
    },
    {
      "epoch": 0.23910852279090603,
      "grad_norm": 1.083142638206482,
      "learning_rate": 2.1335e-05,
      "loss": 0.4959,
      "step": 4270
    },
    {
      "epoch": 0.23916452010303504,
      "grad_norm": 1.1915721893310547,
      "learning_rate": 2.1340000000000002e-05,
      "loss": 0.3234,
      "step": 4271
    },
    {
      "epoch": 0.23922051741516406,
      "grad_norm": 1.1333130598068237,
      "learning_rate": 2.1345e-05,
      "loss": 0.371,
      "step": 4272
    },
    {
      "epoch": 0.23927651472729308,
      "grad_norm": 1.2794235944747925,
      "learning_rate": 2.135e-05,
      "loss": 0.415,
      "step": 4273
    },
    {
      "epoch": 0.2393325120394221,
      "grad_norm": 1.1736987829208374,
      "learning_rate": 2.1355e-05,
      "loss": 0.3776,
      "step": 4274
    },
    {
      "epoch": 0.23938850935155112,
      "grad_norm": 1.090275526046753,
      "learning_rate": 2.1360000000000002e-05,
      "loss": 0.433,
      "step": 4275
    },
    {
      "epoch": 0.23944450666368014,
      "grad_norm": 1.1244940757751465,
      "learning_rate": 2.1365000000000003e-05,
      "loss": 0.5467,
      "step": 4276
    },
    {
      "epoch": 0.23950050397580916,
      "grad_norm": 1.1872576475143433,
      "learning_rate": 2.137e-05,
      "loss": 0.4813,
      "step": 4277
    },
    {
      "epoch": 0.23955650128793818,
      "grad_norm": 1.2008262872695923,
      "learning_rate": 2.1375e-05,
      "loss": 0.4253,
      "step": 4278
    },
    {
      "epoch": 0.2396124986000672,
      "grad_norm": 1.2897939682006836,
      "learning_rate": 2.138e-05,
      "loss": 0.415,
      "step": 4279
    },
    {
      "epoch": 0.23966849591219622,
      "grad_norm": 1.1022006273269653,
      "learning_rate": 2.1385000000000003e-05,
      "loss": 0.4679,
      "step": 4280
    },
    {
      "epoch": 0.23972449322432524,
      "grad_norm": 1.2431297302246094,
      "learning_rate": 2.139e-05,
      "loss": 0.3932,
      "step": 4281
    },
    {
      "epoch": 0.23978049053645425,
      "grad_norm": 1.5248690843582153,
      "learning_rate": 2.1395e-05,
      "loss": 0.4623,
      "step": 4282
    },
    {
      "epoch": 0.23983648784858327,
      "grad_norm": 1.0249652862548828,
      "learning_rate": 2.1400000000000002e-05,
      "loss": 0.3644,
      "step": 4283
    },
    {
      "epoch": 0.2398924851607123,
      "grad_norm": 1.4337689876556396,
      "learning_rate": 2.1405e-05,
      "loss": 0.4221,
      "step": 4284
    },
    {
      "epoch": 0.2399484824728413,
      "grad_norm": 1.283555269241333,
      "learning_rate": 2.1410000000000003e-05,
      "loss": 0.453,
      "step": 4285
    },
    {
      "epoch": 0.24000447978497033,
      "grad_norm": 1.3956892490386963,
      "learning_rate": 2.1415e-05,
      "loss": 0.4818,
      "step": 4286
    },
    {
      "epoch": 0.24006047709709935,
      "grad_norm": 1.484602451324463,
      "learning_rate": 2.142e-05,
      "loss": 0.5885,
      "step": 4287
    },
    {
      "epoch": 0.24011647440922837,
      "grad_norm": 1.120192050933838,
      "learning_rate": 2.1425e-05,
      "loss": 0.366,
      "step": 4288
    },
    {
      "epoch": 0.24017247172135736,
      "grad_norm": 5.211393356323242,
      "learning_rate": 2.143e-05,
      "loss": 0.451,
      "step": 4289
    },
    {
      "epoch": 0.24022846903348638,
      "grad_norm": 1.0974055528640747,
      "learning_rate": 2.1435000000000004e-05,
      "loss": 0.3342,
      "step": 4290
    },
    {
      "epoch": 0.2402844663456154,
      "grad_norm": 1.5038377046585083,
      "learning_rate": 2.144e-05,
      "loss": 0.4989,
      "step": 4291
    },
    {
      "epoch": 0.24034046365774442,
      "grad_norm": 0.933678150177002,
      "learning_rate": 2.1445000000000002e-05,
      "loss": 0.2991,
      "step": 4292
    },
    {
      "epoch": 0.24039646096987344,
      "grad_norm": 1.149804949760437,
      "learning_rate": 2.145e-05,
      "loss": 0.5103,
      "step": 4293
    },
    {
      "epoch": 0.24045245828200246,
      "grad_norm": 1.0830732583999634,
      "learning_rate": 2.1455e-05,
      "loss": 0.4144,
      "step": 4294
    },
    {
      "epoch": 0.24050845559413148,
      "grad_norm": 1.2888356447219849,
      "learning_rate": 2.146e-05,
      "loss": 0.4835,
      "step": 4295
    },
    {
      "epoch": 0.2405644529062605,
      "grad_norm": 5.33606481552124,
      "learning_rate": 2.1465000000000002e-05,
      "loss": 0.5273,
      "step": 4296
    },
    {
      "epoch": 0.24062045021838951,
      "grad_norm": 1.4628833532333374,
      "learning_rate": 2.1470000000000003e-05,
      "loss": 0.4482,
      "step": 4297
    },
    {
      "epoch": 0.24067644753051853,
      "grad_norm": 1.1477042436599731,
      "learning_rate": 2.1475e-05,
      "loss": 0.5156,
      "step": 4298
    },
    {
      "epoch": 0.24073244484264755,
      "grad_norm": 1.2502121925354004,
      "learning_rate": 2.148e-05,
      "loss": 0.3495,
      "step": 4299
    },
    {
      "epoch": 0.24078844215477657,
      "grad_norm": 1.1966519355773926,
      "learning_rate": 2.1485000000000002e-05,
      "loss": 0.4084,
      "step": 4300
    },
    {
      "epoch": 0.2408444394669056,
      "grad_norm": 1.095430850982666,
      "learning_rate": 2.1490000000000003e-05,
      "loss": 0.4155,
      "step": 4301
    },
    {
      "epoch": 0.2409004367790346,
      "grad_norm": 1.3317145109176636,
      "learning_rate": 2.1495e-05,
      "loss": 0.3883,
      "step": 4302
    },
    {
      "epoch": 0.24095643409116363,
      "grad_norm": 1.2294490337371826,
      "learning_rate": 2.15e-05,
      "loss": 0.5213,
      "step": 4303
    },
    {
      "epoch": 0.24101243140329265,
      "grad_norm": 1.0221933126449585,
      "learning_rate": 2.1505e-05,
      "loss": 0.2962,
      "step": 4304
    },
    {
      "epoch": 0.24106842871542167,
      "grad_norm": 1.0682564973831177,
      "learning_rate": 2.1510000000000002e-05,
      "loss": 0.4275,
      "step": 4305
    },
    {
      "epoch": 0.2411244260275507,
      "grad_norm": 1.2987785339355469,
      "learning_rate": 2.1515000000000003e-05,
      "loss": 0.3958,
      "step": 4306
    },
    {
      "epoch": 0.2411804233396797,
      "grad_norm": 1.083464503288269,
      "learning_rate": 2.152e-05,
      "loss": 0.4409,
      "step": 4307
    },
    {
      "epoch": 0.24123642065180873,
      "grad_norm": 1.1261757612228394,
      "learning_rate": 2.1525e-05,
      "loss": 0.4545,
      "step": 4308
    },
    {
      "epoch": 0.24129241796393774,
      "grad_norm": 1.1840627193450928,
      "learning_rate": 2.153e-05,
      "loss": 0.4417,
      "step": 4309
    },
    {
      "epoch": 0.24134841527606674,
      "grad_norm": 2.27750301361084,
      "learning_rate": 2.1535000000000003e-05,
      "loss": 0.5773,
      "step": 4310
    },
    {
      "epoch": 0.24140441258819575,
      "grad_norm": 1.263756513595581,
      "learning_rate": 2.154e-05,
      "loss": 0.4707,
      "step": 4311
    },
    {
      "epoch": 0.24146040990032477,
      "grad_norm": 1.3066948652267456,
      "learning_rate": 2.1545e-05,
      "loss": 0.4886,
      "step": 4312
    },
    {
      "epoch": 0.2415164072124538,
      "grad_norm": 1.117456078529358,
      "learning_rate": 2.1550000000000002e-05,
      "loss": 0.4034,
      "step": 4313
    },
    {
      "epoch": 0.2415724045245828,
      "grad_norm": 1.612994909286499,
      "learning_rate": 2.1555e-05,
      "loss": 0.3843,
      "step": 4314
    },
    {
      "epoch": 0.24162840183671183,
      "grad_norm": 1.3376786708831787,
      "learning_rate": 2.1560000000000004e-05,
      "loss": 0.5396,
      "step": 4315
    },
    {
      "epoch": 0.24168439914884085,
      "grad_norm": 1.0367299318313599,
      "learning_rate": 2.1565e-05,
      "loss": 0.3505,
      "step": 4316
    },
    {
      "epoch": 0.24174039646096987,
      "grad_norm": 1.2690515518188477,
      "learning_rate": 2.1570000000000002e-05,
      "loss": 0.4122,
      "step": 4317
    },
    {
      "epoch": 0.2417963937730989,
      "grad_norm": 1.1069188117980957,
      "learning_rate": 2.1575e-05,
      "loss": 0.4649,
      "step": 4318
    },
    {
      "epoch": 0.2418523910852279,
      "grad_norm": 1.1699316501617432,
      "learning_rate": 2.158e-05,
      "loss": 0.531,
      "step": 4319
    },
    {
      "epoch": 0.24190838839735693,
      "grad_norm": 1.2313868999481201,
      "learning_rate": 2.1585e-05,
      "loss": 0.4042,
      "step": 4320
    },
    {
      "epoch": 0.24196438570948595,
      "grad_norm": 1.5419354438781738,
      "learning_rate": 2.159e-05,
      "loss": 0.5642,
      "step": 4321
    },
    {
      "epoch": 0.24202038302161497,
      "grad_norm": 1.1893784999847412,
      "learning_rate": 2.1595000000000002e-05,
      "loss": 0.3844,
      "step": 4322
    },
    {
      "epoch": 0.24207638033374398,
      "grad_norm": 1.4173463582992554,
      "learning_rate": 2.16e-05,
      "loss": 0.5144,
      "step": 4323
    },
    {
      "epoch": 0.242132377645873,
      "grad_norm": 0.9899412393569946,
      "learning_rate": 2.1605e-05,
      "loss": 0.3475,
      "step": 4324
    },
    {
      "epoch": 0.24218837495800202,
      "grad_norm": 1.3944251537322998,
      "learning_rate": 2.1609999999999998e-05,
      "loss": 0.4864,
      "step": 4325
    },
    {
      "epoch": 0.24224437227013104,
      "grad_norm": 1.2878979444503784,
      "learning_rate": 2.1615000000000002e-05,
      "loss": 0.3878,
      "step": 4326
    },
    {
      "epoch": 0.24230036958226006,
      "grad_norm": 1.1655343770980835,
      "learning_rate": 2.162e-05,
      "loss": 0.284,
      "step": 4327
    },
    {
      "epoch": 0.24235636689438908,
      "grad_norm": 1.329520583152771,
      "learning_rate": 2.1625e-05,
      "loss": 0.518,
      "step": 4328
    },
    {
      "epoch": 0.2424123642065181,
      "grad_norm": 1.3053680658340454,
      "learning_rate": 2.163e-05,
      "loss": 0.5209,
      "step": 4329
    },
    {
      "epoch": 0.24246836151864712,
      "grad_norm": 1.5850566625595093,
      "learning_rate": 2.1635e-05,
      "loss": 0.4758,
      "step": 4330
    },
    {
      "epoch": 0.2425243588307761,
      "grad_norm": 1.0623127222061157,
      "learning_rate": 2.1640000000000003e-05,
      "loss": 0.4205,
      "step": 4331
    },
    {
      "epoch": 0.24258035614290513,
      "grad_norm": 1.486602783203125,
      "learning_rate": 2.1645e-05,
      "loss": 0.5736,
      "step": 4332
    },
    {
      "epoch": 0.24263635345503415,
      "grad_norm": 1.1264127492904663,
      "learning_rate": 2.165e-05,
      "loss": 0.3344,
      "step": 4333
    },
    {
      "epoch": 0.24269235076716317,
      "grad_norm": 1.114925503730774,
      "learning_rate": 2.1655000000000002e-05,
      "loss": 0.3654,
      "step": 4334
    },
    {
      "epoch": 0.2427483480792922,
      "grad_norm": 0.9926008582115173,
      "learning_rate": 2.166e-05,
      "loss": 0.4218,
      "step": 4335
    },
    {
      "epoch": 0.2428043453914212,
      "grad_norm": 1.0779087543487549,
      "learning_rate": 2.1665000000000003e-05,
      "loss": 0.3215,
      "step": 4336
    },
    {
      "epoch": 0.24286034270355022,
      "grad_norm": 1.2170718908309937,
      "learning_rate": 2.167e-05,
      "loss": 0.6787,
      "step": 4337
    },
    {
      "epoch": 0.24291634001567924,
      "grad_norm": 1.7566055059432983,
      "learning_rate": 2.1675e-05,
      "loss": 0.5167,
      "step": 4338
    },
    {
      "epoch": 0.24297233732780826,
      "grad_norm": 1.25746488571167,
      "learning_rate": 2.168e-05,
      "loss": 0.4718,
      "step": 4339
    },
    {
      "epoch": 0.24302833463993728,
      "grad_norm": 1.1558181047439575,
      "learning_rate": 2.1685e-05,
      "loss": 0.475,
      "step": 4340
    },
    {
      "epoch": 0.2430843319520663,
      "grad_norm": 9.28864860534668,
      "learning_rate": 2.169e-05,
      "loss": 0.4129,
      "step": 4341
    },
    {
      "epoch": 0.24314032926419532,
      "grad_norm": 1.0949761867523193,
      "learning_rate": 2.1695e-05,
      "loss": 0.3915,
      "step": 4342
    },
    {
      "epoch": 0.24319632657632434,
      "grad_norm": 1.1187132596969604,
      "learning_rate": 2.1700000000000002e-05,
      "loss": 0.3921,
      "step": 4343
    },
    {
      "epoch": 0.24325232388845336,
      "grad_norm": 1.0224883556365967,
      "learning_rate": 2.1705e-05,
      "loss": 0.3613,
      "step": 4344
    },
    {
      "epoch": 0.24330832120058238,
      "grad_norm": 1.2836576700210571,
      "learning_rate": 2.171e-05,
      "loss": 0.4491,
      "step": 4345
    },
    {
      "epoch": 0.2433643185127114,
      "grad_norm": 1.13035249710083,
      "learning_rate": 2.1715e-05,
      "loss": 0.3769,
      "step": 4346
    },
    {
      "epoch": 0.24342031582484042,
      "grad_norm": 1.0483314990997314,
      "learning_rate": 2.1720000000000002e-05,
      "loss": 0.4401,
      "step": 4347
    },
    {
      "epoch": 0.24347631313696944,
      "grad_norm": 1.3753732442855835,
      "learning_rate": 2.1725e-05,
      "loss": 0.5379,
      "step": 4348
    },
    {
      "epoch": 0.24353231044909845,
      "grad_norm": 4.958712100982666,
      "learning_rate": 2.173e-05,
      "loss": 0.3993,
      "step": 4349
    },
    {
      "epoch": 0.24358830776122747,
      "grad_norm": 1.0264718532562256,
      "learning_rate": 2.1735e-05,
      "loss": 0.3764,
      "step": 4350
    },
    {
      "epoch": 0.24364430507335647,
      "grad_norm": 1.439278483390808,
      "learning_rate": 2.1740000000000002e-05,
      "loss": 0.651,
      "step": 4351
    },
    {
      "epoch": 0.24370030238548548,
      "grad_norm": 1.1682460308074951,
      "learning_rate": 2.1745000000000003e-05,
      "loss": 0.3624,
      "step": 4352
    },
    {
      "epoch": 0.2437562996976145,
      "grad_norm": 1.3046634197235107,
      "learning_rate": 2.175e-05,
      "loss": 0.4369,
      "step": 4353
    },
    {
      "epoch": 0.24381229700974352,
      "grad_norm": 1.123918890953064,
      "learning_rate": 2.1755e-05,
      "loss": 0.4421,
      "step": 4354
    },
    {
      "epoch": 0.24386829432187254,
      "grad_norm": 1.194492220878601,
      "learning_rate": 2.176e-05,
      "loss": 0.3583,
      "step": 4355
    },
    {
      "epoch": 0.24392429163400156,
      "grad_norm": 1.5091285705566406,
      "learning_rate": 2.1765000000000003e-05,
      "loss": 0.557,
      "step": 4356
    },
    {
      "epoch": 0.24398028894613058,
      "grad_norm": 1.6220325231552124,
      "learning_rate": 2.177e-05,
      "loss": 0.563,
      "step": 4357
    },
    {
      "epoch": 0.2440362862582596,
      "grad_norm": 1.1459437608718872,
      "learning_rate": 2.1775e-05,
      "loss": 0.4278,
      "step": 4358
    },
    {
      "epoch": 0.24409228357038862,
      "grad_norm": 1.2954154014587402,
      "learning_rate": 2.178e-05,
      "loss": 0.5298,
      "step": 4359
    },
    {
      "epoch": 0.24414828088251764,
      "grad_norm": 1.2371245622634888,
      "learning_rate": 2.1785e-05,
      "loss": 0.343,
      "step": 4360
    },
    {
      "epoch": 0.24420427819464666,
      "grad_norm": 1.1733187437057495,
      "learning_rate": 2.1790000000000003e-05,
      "loss": 0.3858,
      "step": 4361
    },
    {
      "epoch": 0.24426027550677568,
      "grad_norm": 1.2497729063034058,
      "learning_rate": 2.1795e-05,
      "loss": 0.3206,
      "step": 4362
    },
    {
      "epoch": 0.2443162728189047,
      "grad_norm": 1.3116555213928223,
      "learning_rate": 2.18e-05,
      "loss": 0.5136,
      "step": 4363
    },
    {
      "epoch": 0.24437227013103371,
      "grad_norm": 1.1510710716247559,
      "learning_rate": 2.1805e-05,
      "loss": 0.4088,
      "step": 4364
    },
    {
      "epoch": 0.24442826744316273,
      "grad_norm": 1.3727792501449585,
      "learning_rate": 2.181e-05,
      "loss": 0.4046,
      "step": 4365
    },
    {
      "epoch": 0.24448426475529175,
      "grad_norm": 1.2303552627563477,
      "learning_rate": 2.1815000000000004e-05,
      "loss": 0.3942,
      "step": 4366
    },
    {
      "epoch": 0.24454026206742077,
      "grad_norm": 1.3398981094360352,
      "learning_rate": 2.182e-05,
      "loss": 0.4133,
      "step": 4367
    },
    {
      "epoch": 0.2445962593795498,
      "grad_norm": 1.425965428352356,
      "learning_rate": 2.1825000000000002e-05,
      "loss": 0.4376,
      "step": 4368
    },
    {
      "epoch": 0.2446522566916788,
      "grad_norm": 1.1586977243423462,
      "learning_rate": 2.183e-05,
      "loss": 0.4468,
      "step": 4369
    },
    {
      "epoch": 0.24470825400380783,
      "grad_norm": 1.0181491374969482,
      "learning_rate": 2.1835e-05,
      "loss": 0.4791,
      "step": 4370
    },
    {
      "epoch": 0.24476425131593685,
      "grad_norm": 1.1308424472808838,
      "learning_rate": 2.184e-05,
      "loss": 0.3961,
      "step": 4371
    },
    {
      "epoch": 0.24482024862806584,
      "grad_norm": 1.4323550462722778,
      "learning_rate": 2.1845000000000002e-05,
      "loss": 0.5017,
      "step": 4372
    },
    {
      "epoch": 0.24487624594019486,
      "grad_norm": 1.1525380611419678,
      "learning_rate": 2.1850000000000003e-05,
      "loss": 0.4749,
      "step": 4373
    },
    {
      "epoch": 0.24493224325232388,
      "grad_norm": 1.2878707647323608,
      "learning_rate": 2.1855e-05,
      "loss": 0.4322,
      "step": 4374
    },
    {
      "epoch": 0.2449882405644529,
      "grad_norm": 1.210702896118164,
      "learning_rate": 2.186e-05,
      "loss": 0.4982,
      "step": 4375
    },
    {
      "epoch": 0.24504423787658192,
      "grad_norm": 1.2443805932998657,
      "learning_rate": 2.1865e-05,
      "loss": 0.3842,
      "step": 4376
    },
    {
      "epoch": 0.24510023518871094,
      "grad_norm": 1.083103060722351,
      "learning_rate": 2.1870000000000002e-05,
      "loss": 0.3987,
      "step": 4377
    },
    {
      "epoch": 0.24515623250083995,
      "grad_norm": 1.1591500043869019,
      "learning_rate": 2.1875e-05,
      "loss": 0.3353,
      "step": 4378
    },
    {
      "epoch": 0.24521222981296897,
      "grad_norm": 1.045145869255066,
      "learning_rate": 2.188e-05,
      "loss": 0.4127,
      "step": 4379
    },
    {
      "epoch": 0.245268227125098,
      "grad_norm": 1.5242241621017456,
      "learning_rate": 2.1885e-05,
      "loss": 0.5309,
      "step": 4380
    },
    {
      "epoch": 0.245324224437227,
      "grad_norm": 1.0925313234329224,
      "learning_rate": 2.1890000000000002e-05,
      "loss": 0.456,
      "step": 4381
    },
    {
      "epoch": 0.24538022174935603,
      "grad_norm": 1.1851149797439575,
      "learning_rate": 2.1895000000000003e-05,
      "loss": 0.3898,
      "step": 4382
    },
    {
      "epoch": 0.24543621906148505,
      "grad_norm": 1.1446529626846313,
      "learning_rate": 2.19e-05,
      "loss": 0.357,
      "step": 4383
    },
    {
      "epoch": 0.24549221637361407,
      "grad_norm": 1.3524234294891357,
      "learning_rate": 2.1905e-05,
      "loss": 0.4854,
      "step": 4384
    },
    {
      "epoch": 0.2455482136857431,
      "grad_norm": 1.198757290840149,
      "learning_rate": 2.191e-05,
      "loss": 0.4094,
      "step": 4385
    },
    {
      "epoch": 0.2456042109978721,
      "grad_norm": 1.1887050867080688,
      "learning_rate": 2.1915000000000003e-05,
      "loss": 0.3984,
      "step": 4386
    },
    {
      "epoch": 0.24566020831000113,
      "grad_norm": 1.2815239429473877,
      "learning_rate": 2.192e-05,
      "loss": 0.3801,
      "step": 4387
    },
    {
      "epoch": 0.24571620562213015,
      "grad_norm": 1.220858097076416,
      "learning_rate": 2.1925e-05,
      "loss": 0.429,
      "step": 4388
    },
    {
      "epoch": 0.24577220293425917,
      "grad_norm": 1.093279480934143,
      "learning_rate": 2.1930000000000002e-05,
      "loss": 0.3266,
      "step": 4389
    },
    {
      "epoch": 0.24582820024638818,
      "grad_norm": 1.432355523109436,
      "learning_rate": 2.1935e-05,
      "loss": 0.4068,
      "step": 4390
    },
    {
      "epoch": 0.2458841975585172,
      "grad_norm": 1.2643381357192993,
      "learning_rate": 2.1940000000000003e-05,
      "loss": 0.5241,
      "step": 4391
    },
    {
      "epoch": 0.24594019487064622,
      "grad_norm": 1.1043672561645508,
      "learning_rate": 2.1945e-05,
      "loss": 0.4859,
      "step": 4392
    },
    {
      "epoch": 0.24599619218277521,
      "grad_norm": 1.317258596420288,
      "learning_rate": 2.195e-05,
      "loss": 0.5257,
      "step": 4393
    },
    {
      "epoch": 0.24605218949490423,
      "grad_norm": 1.2772438526153564,
      "learning_rate": 2.1955e-05,
      "loss": 0.4314,
      "step": 4394
    },
    {
      "epoch": 0.24610818680703325,
      "grad_norm": 3.390718460083008,
      "learning_rate": 2.196e-05,
      "loss": 0.4308,
      "step": 4395
    },
    {
      "epoch": 0.24616418411916227,
      "grad_norm": 1.1928256750106812,
      "learning_rate": 2.1965e-05,
      "loss": 0.4142,
      "step": 4396
    },
    {
      "epoch": 0.2462201814312913,
      "grad_norm": 1.1900676488876343,
      "learning_rate": 2.197e-05,
      "loss": 0.3713,
      "step": 4397
    },
    {
      "epoch": 0.2462761787434203,
      "grad_norm": 1.3439291715621948,
      "learning_rate": 2.1975000000000002e-05,
      "loss": 0.5114,
      "step": 4398
    },
    {
      "epoch": 0.24633217605554933,
      "grad_norm": 1.282044768333435,
      "learning_rate": 2.198e-05,
      "loss": 0.5487,
      "step": 4399
    },
    {
      "epoch": 0.24638817336767835,
      "grad_norm": 1.2520606517791748,
      "learning_rate": 2.1985e-05,
      "loss": 0.4649,
      "step": 4400
    },
    {
      "epoch": 0.24644417067980737,
      "grad_norm": 1.1088248491287231,
      "learning_rate": 2.199e-05,
      "loss": 0.3805,
      "step": 4401
    },
    {
      "epoch": 0.2465001679919364,
      "grad_norm": 1.6109001636505127,
      "learning_rate": 2.1995000000000002e-05,
      "loss": 0.3532,
      "step": 4402
    },
    {
      "epoch": 0.2465561653040654,
      "grad_norm": 1.2714914083480835,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.4865,
      "step": 4403
    },
    {
      "epoch": 0.24661216261619442,
      "grad_norm": 1.204676628112793,
      "learning_rate": 2.2005e-05,
      "loss": 0.5088,
      "step": 4404
    },
    {
      "epoch": 0.24666815992832344,
      "grad_norm": 1.2741031646728516,
      "learning_rate": 2.201e-05,
      "loss": 0.4747,
      "step": 4405
    },
    {
      "epoch": 0.24672415724045246,
      "grad_norm": 1.3649706840515137,
      "learning_rate": 2.2015000000000002e-05,
      "loss": 0.5744,
      "step": 4406
    },
    {
      "epoch": 0.24678015455258148,
      "grad_norm": 1.3730049133300781,
      "learning_rate": 2.2020000000000003e-05,
      "loss": 0.5204,
      "step": 4407
    },
    {
      "epoch": 0.2468361518647105,
      "grad_norm": 1.2699775695800781,
      "learning_rate": 2.2025e-05,
      "loss": 0.4023,
      "step": 4408
    },
    {
      "epoch": 0.24689214917683952,
      "grad_norm": 1.1627721786499023,
      "learning_rate": 2.203e-05,
      "loss": 0.4179,
      "step": 4409
    },
    {
      "epoch": 0.24694814648896854,
      "grad_norm": 1.1745370626449585,
      "learning_rate": 2.2035e-05,
      "loss": 0.4344,
      "step": 4410
    },
    {
      "epoch": 0.24700414380109756,
      "grad_norm": 1.1994861364364624,
      "learning_rate": 2.2040000000000002e-05,
      "loss": 0.4245,
      "step": 4411
    },
    {
      "epoch": 0.24706014111322658,
      "grad_norm": 0.9576858282089233,
      "learning_rate": 2.2045000000000003e-05,
      "loss": 0.3195,
      "step": 4412
    },
    {
      "epoch": 0.24711613842535557,
      "grad_norm": 1.258575201034546,
      "learning_rate": 2.205e-05,
      "loss": 0.425,
      "step": 4413
    },
    {
      "epoch": 0.2471721357374846,
      "grad_norm": 1.6441184282302856,
      "learning_rate": 2.2055e-05,
      "loss": 0.511,
      "step": 4414
    },
    {
      "epoch": 0.2472281330496136,
      "grad_norm": 1.2258238792419434,
      "learning_rate": 2.206e-05,
      "loss": 0.391,
      "step": 4415
    },
    {
      "epoch": 0.24728413036174263,
      "grad_norm": 1.3207824230194092,
      "learning_rate": 2.2065000000000003e-05,
      "loss": 0.4906,
      "step": 4416
    },
    {
      "epoch": 0.24734012767387165,
      "grad_norm": 1.6287943124771118,
      "learning_rate": 2.207e-05,
      "loss": 0.6048,
      "step": 4417
    },
    {
      "epoch": 0.24739612498600067,
      "grad_norm": 1.5828886032104492,
      "learning_rate": 2.2075e-05,
      "loss": 0.4398,
      "step": 4418
    },
    {
      "epoch": 0.24745212229812968,
      "grad_norm": 1.2290335893630981,
      "learning_rate": 2.2080000000000002e-05,
      "loss": 0.4713,
      "step": 4419
    },
    {
      "epoch": 0.2475081196102587,
      "grad_norm": 1.4252212047576904,
      "learning_rate": 2.2085e-05,
      "loss": 0.4958,
      "step": 4420
    },
    {
      "epoch": 0.24756411692238772,
      "grad_norm": 1.0584359169006348,
      "learning_rate": 2.2090000000000004e-05,
      "loss": 0.4633,
      "step": 4421
    },
    {
      "epoch": 0.24762011423451674,
      "grad_norm": 0.9757269024848938,
      "learning_rate": 2.2095e-05,
      "loss": 0.3672,
      "step": 4422
    },
    {
      "epoch": 0.24767611154664576,
      "grad_norm": 1.0999170541763306,
      "learning_rate": 2.2100000000000002e-05,
      "loss": 0.4658,
      "step": 4423
    },
    {
      "epoch": 0.24773210885877478,
      "grad_norm": 1.0702378749847412,
      "learning_rate": 2.2105e-05,
      "loss": 0.3745,
      "step": 4424
    },
    {
      "epoch": 0.2477881061709038,
      "grad_norm": 1.335506796836853,
      "learning_rate": 2.211e-05,
      "loss": 0.423,
      "step": 4425
    },
    {
      "epoch": 0.24784410348303282,
      "grad_norm": 1.1725655794143677,
      "learning_rate": 2.2115e-05,
      "loss": 0.4771,
      "step": 4426
    },
    {
      "epoch": 0.24790010079516184,
      "grad_norm": 1.1515663862228394,
      "learning_rate": 2.212e-05,
      "loss": 0.3655,
      "step": 4427
    },
    {
      "epoch": 0.24795609810729086,
      "grad_norm": 1.1827774047851562,
      "learning_rate": 2.2125000000000002e-05,
      "loss": 0.4368,
      "step": 4428
    },
    {
      "epoch": 0.24801209541941988,
      "grad_norm": 1.267818570137024,
      "learning_rate": 2.213e-05,
      "loss": 0.4739,
      "step": 4429
    },
    {
      "epoch": 0.2480680927315489,
      "grad_norm": 1.034224510192871,
      "learning_rate": 2.2135e-05,
      "loss": 0.3464,
      "step": 4430
    },
    {
      "epoch": 0.24812409004367791,
      "grad_norm": 1.0499874353408813,
      "learning_rate": 2.214e-05,
      "loss": 0.44,
      "step": 4431
    },
    {
      "epoch": 0.24818008735580693,
      "grad_norm": 1.1445691585540771,
      "learning_rate": 2.2145000000000002e-05,
      "loss": 0.3484,
      "step": 4432
    },
    {
      "epoch": 0.24823608466793595,
      "grad_norm": 1.0334786176681519,
      "learning_rate": 2.215e-05,
      "loss": 0.5041,
      "step": 4433
    },
    {
      "epoch": 0.24829208198006494,
      "grad_norm": 1.2405943870544434,
      "learning_rate": 2.2155e-05,
      "loss": 0.4128,
      "step": 4434
    },
    {
      "epoch": 0.24834807929219396,
      "grad_norm": 1.390884280204773,
      "learning_rate": 2.216e-05,
      "loss": 0.5113,
      "step": 4435
    },
    {
      "epoch": 0.24840407660432298,
      "grad_norm": 1.0858657360076904,
      "learning_rate": 2.2165000000000002e-05,
      "loss": 0.363,
      "step": 4436
    },
    {
      "epoch": 0.248460073916452,
      "grad_norm": 1.241275429725647,
      "learning_rate": 2.2170000000000003e-05,
      "loss": 0.3941,
      "step": 4437
    },
    {
      "epoch": 0.24851607122858102,
      "grad_norm": 1.1823296546936035,
      "learning_rate": 2.2175e-05,
      "loss": 0.3404,
      "step": 4438
    },
    {
      "epoch": 0.24857206854071004,
      "grad_norm": 1.2853224277496338,
      "learning_rate": 2.218e-05,
      "loss": 0.3665,
      "step": 4439
    },
    {
      "epoch": 0.24862806585283906,
      "grad_norm": 1.2950419187545776,
      "learning_rate": 2.2185000000000002e-05,
      "loss": 0.4266,
      "step": 4440
    },
    {
      "epoch": 0.24868406316496808,
      "grad_norm": 1.5352532863616943,
      "learning_rate": 2.219e-05,
      "loss": 0.6076,
      "step": 4441
    },
    {
      "epoch": 0.2487400604770971,
      "grad_norm": 1.2307060956954956,
      "learning_rate": 2.2195000000000003e-05,
      "loss": 0.4616,
      "step": 4442
    },
    {
      "epoch": 0.24879605778922612,
      "grad_norm": 1.073006510734558,
      "learning_rate": 2.22e-05,
      "loss": 0.3265,
      "step": 4443
    },
    {
      "epoch": 0.24885205510135514,
      "grad_norm": 1.1062570810317993,
      "learning_rate": 2.2205000000000002e-05,
      "loss": 0.3564,
      "step": 4444
    },
    {
      "epoch": 0.24890805241348415,
      "grad_norm": 1.2196799516677856,
      "learning_rate": 2.221e-05,
      "loss": 0.4025,
      "step": 4445
    },
    {
      "epoch": 0.24896404972561317,
      "grad_norm": 1.1566624641418457,
      "learning_rate": 2.2215e-05,
      "loss": 0.3357,
      "step": 4446
    },
    {
      "epoch": 0.2490200470377422,
      "grad_norm": 1.2082751989364624,
      "learning_rate": 2.222e-05,
      "loss": 0.442,
      "step": 4447
    },
    {
      "epoch": 0.2490760443498712,
      "grad_norm": 1.264952540397644,
      "learning_rate": 2.2225e-05,
      "loss": 0.5093,
      "step": 4448
    },
    {
      "epoch": 0.24913204166200023,
      "grad_norm": 1.0614506006240845,
      "learning_rate": 2.2230000000000002e-05,
      "loss": 0.3028,
      "step": 4449
    },
    {
      "epoch": 0.24918803897412925,
      "grad_norm": 1.1484986543655396,
      "learning_rate": 2.2235e-05,
      "loss": 0.4714,
      "step": 4450
    },
    {
      "epoch": 0.24924403628625827,
      "grad_norm": 1.2769389152526855,
      "learning_rate": 2.224e-05,
      "loss": 0.5447,
      "step": 4451
    },
    {
      "epoch": 0.2493000335983873,
      "grad_norm": 1.1708201169967651,
      "learning_rate": 2.2245e-05,
      "loss": 0.3064,
      "step": 4452
    },
    {
      "epoch": 0.2493560309105163,
      "grad_norm": 1.3028026819229126,
      "learning_rate": 2.2250000000000002e-05,
      "loss": 0.4678,
      "step": 4453
    },
    {
      "epoch": 0.24941202822264533,
      "grad_norm": 1.1337742805480957,
      "learning_rate": 2.2255e-05,
      "loss": 0.4317,
      "step": 4454
    },
    {
      "epoch": 0.24946802553477432,
      "grad_norm": 1.2387633323669434,
      "learning_rate": 2.226e-05,
      "loss": 0.4296,
      "step": 4455
    },
    {
      "epoch": 0.24952402284690334,
      "grad_norm": 1.2744488716125488,
      "learning_rate": 2.2265e-05,
      "loss": 0.4337,
      "step": 4456
    },
    {
      "epoch": 0.24958002015903236,
      "grad_norm": 1.4655698537826538,
      "learning_rate": 2.2270000000000002e-05,
      "loss": 0.3449,
      "step": 4457
    },
    {
      "epoch": 0.24963601747116138,
      "grad_norm": 1.068646788597107,
      "learning_rate": 2.2275000000000003e-05,
      "loss": 0.3799,
      "step": 4458
    },
    {
      "epoch": 0.2496920147832904,
      "grad_norm": 1.1122599840164185,
      "learning_rate": 2.228e-05,
      "loss": 0.4005,
      "step": 4459
    },
    {
      "epoch": 0.2497480120954194,
      "grad_norm": 1.5223112106323242,
      "learning_rate": 2.2285e-05,
      "loss": 0.4872,
      "step": 4460
    },
    {
      "epoch": 0.24980400940754843,
      "grad_norm": 1.0510022640228271,
      "learning_rate": 2.229e-05,
      "loss": 0.3729,
      "step": 4461
    },
    {
      "epoch": 0.24986000671967745,
      "grad_norm": 1.342260718345642,
      "learning_rate": 2.2295000000000003e-05,
      "loss": 0.5689,
      "step": 4462
    },
    {
      "epoch": 0.24991600403180647,
      "grad_norm": 1.2196329832077026,
      "learning_rate": 2.23e-05,
      "loss": 0.4905,
      "step": 4463
    },
    {
      "epoch": 0.2499720013439355,
      "grad_norm": 1.9157620668411255,
      "learning_rate": 2.2305e-05,
      "loss": 0.4368,
      "step": 4464
    },
    {
      "epoch": 0.2500279986560645,
      "grad_norm": 1.316776990890503,
      "learning_rate": 2.231e-05,
      "loss": 0.3954,
      "step": 4465
    },
    {
      "epoch": 0.25008399596819353,
      "grad_norm": 1.326110601425171,
      "learning_rate": 2.2315e-05,
      "loss": 0.4895,
      "step": 4466
    },
    {
      "epoch": 0.25013999328032255,
      "grad_norm": 1.2966643571853638,
      "learning_rate": 2.2320000000000003e-05,
      "loss": 0.451,
      "step": 4467
    },
    {
      "epoch": 0.25019599059245157,
      "grad_norm": 1.364499807357788,
      "learning_rate": 2.2325e-05,
      "loss": 0.4063,
      "step": 4468
    },
    {
      "epoch": 0.2502519879045806,
      "grad_norm": 1.3014533519744873,
      "learning_rate": 2.233e-05,
      "loss": 0.3914,
      "step": 4469
    },
    {
      "epoch": 0.2503079852167096,
      "grad_norm": 0.955682098865509,
      "learning_rate": 2.2335e-05,
      "loss": 0.4266,
      "step": 4470
    },
    {
      "epoch": 0.2503639825288386,
      "grad_norm": 1.215097188949585,
      "learning_rate": 2.234e-05,
      "loss": 0.3631,
      "step": 4471
    },
    {
      "epoch": 0.25041997984096764,
      "grad_norm": 1.2024786472320557,
      "learning_rate": 2.2345e-05,
      "loss": 0.4752,
      "step": 4472
    },
    {
      "epoch": 0.25047597715309666,
      "grad_norm": 1.1775621175765991,
      "learning_rate": 2.235e-05,
      "loss": 0.5077,
      "step": 4473
    },
    {
      "epoch": 0.2505319744652257,
      "grad_norm": 1.1632726192474365,
      "learning_rate": 2.2355000000000002e-05,
      "loss": 0.428,
      "step": 4474
    },
    {
      "epoch": 0.2505879717773547,
      "grad_norm": 1.2444415092468262,
      "learning_rate": 2.236e-05,
      "loss": 0.4456,
      "step": 4475
    },
    {
      "epoch": 0.2506439690894837,
      "grad_norm": 1.2137200832366943,
      "learning_rate": 2.2365e-05,
      "loss": 0.3735,
      "step": 4476
    },
    {
      "epoch": 0.25069996640161274,
      "grad_norm": 1.350601315498352,
      "learning_rate": 2.237e-05,
      "loss": 0.4603,
      "step": 4477
    },
    {
      "epoch": 0.25075596371374176,
      "grad_norm": 1.1135917901992798,
      "learning_rate": 2.2375000000000002e-05,
      "loss": 0.4001,
      "step": 4478
    },
    {
      "epoch": 0.2508119610258708,
      "grad_norm": 1.1618399620056152,
      "learning_rate": 2.2380000000000003e-05,
      "loss": 0.4734,
      "step": 4479
    },
    {
      "epoch": 0.2508679583379998,
      "grad_norm": 1.1615564823150635,
      "learning_rate": 2.2385e-05,
      "loss": 0.4212,
      "step": 4480
    },
    {
      "epoch": 0.2509239556501288,
      "grad_norm": 2.1652941703796387,
      "learning_rate": 2.239e-05,
      "loss": 0.424,
      "step": 4481
    },
    {
      "epoch": 0.25097995296225784,
      "grad_norm": 1.3567856550216675,
      "learning_rate": 2.2395e-05,
      "loss": 0.7095,
      "step": 4482
    },
    {
      "epoch": 0.25103595027438685,
      "grad_norm": 1.3855979442596436,
      "learning_rate": 2.2400000000000002e-05,
      "loss": 0.5495,
      "step": 4483
    },
    {
      "epoch": 0.2510919475865159,
      "grad_norm": 1.2875161170959473,
      "learning_rate": 2.2405e-05,
      "loss": 0.3818,
      "step": 4484
    },
    {
      "epoch": 0.2511479448986449,
      "grad_norm": 1.3533802032470703,
      "learning_rate": 2.241e-05,
      "loss": 0.4189,
      "step": 4485
    },
    {
      "epoch": 0.25120394221077386,
      "grad_norm": 1.444612741470337,
      "learning_rate": 2.2415e-05,
      "loss": 0.4597,
      "step": 4486
    },
    {
      "epoch": 0.2512599395229029,
      "grad_norm": 1.1549077033996582,
      "learning_rate": 2.2420000000000002e-05,
      "loss": 0.4765,
      "step": 4487
    },
    {
      "epoch": 0.2513159368350319,
      "grad_norm": 1.3089532852172852,
      "learning_rate": 2.2425000000000003e-05,
      "loss": 0.5499,
      "step": 4488
    },
    {
      "epoch": 0.2513719341471609,
      "grad_norm": 1.3971365690231323,
      "learning_rate": 2.243e-05,
      "loss": 0.6296,
      "step": 4489
    },
    {
      "epoch": 0.25142793145928993,
      "grad_norm": 1.27325439453125,
      "learning_rate": 2.2435e-05,
      "loss": 0.4193,
      "step": 4490
    },
    {
      "epoch": 0.25148392877141895,
      "grad_norm": 1.5754708051681519,
      "learning_rate": 2.244e-05,
      "loss": 0.4805,
      "step": 4491
    },
    {
      "epoch": 0.25153992608354797,
      "grad_norm": 1.695016622543335,
      "learning_rate": 2.2445000000000003e-05,
      "loss": 0.5813,
      "step": 4492
    },
    {
      "epoch": 0.251595923395677,
      "grad_norm": 1.0356920957565308,
      "learning_rate": 2.245e-05,
      "loss": 0.4802,
      "step": 4493
    },
    {
      "epoch": 0.251651920707806,
      "grad_norm": 1.0675708055496216,
      "learning_rate": 2.2455e-05,
      "loss": 0.3035,
      "step": 4494
    },
    {
      "epoch": 0.25170791801993503,
      "grad_norm": 1.0566827058792114,
      "learning_rate": 2.2460000000000002e-05,
      "loss": 0.3935,
      "step": 4495
    },
    {
      "epoch": 0.25176391533206405,
      "grad_norm": 1.114900827407837,
      "learning_rate": 2.2465e-05,
      "loss": 0.38,
      "step": 4496
    },
    {
      "epoch": 0.25181991264419307,
      "grad_norm": 1.1412712335586548,
      "learning_rate": 2.2470000000000003e-05,
      "loss": 0.463,
      "step": 4497
    },
    {
      "epoch": 0.2518759099563221,
      "grad_norm": 1.238297939300537,
      "learning_rate": 2.2475e-05,
      "loss": 0.3508,
      "step": 4498
    },
    {
      "epoch": 0.2519319072684511,
      "grad_norm": 1.077498197555542,
      "learning_rate": 2.248e-05,
      "loss": 0.3731,
      "step": 4499
    },
    {
      "epoch": 0.2519879045805801,
      "grad_norm": 1.1671607494354248,
      "learning_rate": 2.2485e-05,
      "loss": 0.2541,
      "step": 4500
    },
    {
      "epoch": 0.25204390189270914,
      "grad_norm": 1.0954142808914185,
      "learning_rate": 2.249e-05,
      "loss": 0.3176,
      "step": 4501
    },
    {
      "epoch": 0.25209989920483816,
      "grad_norm": 1.1871159076690674,
      "learning_rate": 2.2495e-05,
      "loss": 0.3175,
      "step": 4502
    },
    {
      "epoch": 0.2521558965169672,
      "grad_norm": 1.5153894424438477,
      "learning_rate": 2.25e-05,
      "loss": 0.4673,
      "step": 4503
    },
    {
      "epoch": 0.2522118938290962,
      "grad_norm": 1.1715959310531616,
      "learning_rate": 2.2505000000000002e-05,
      "loss": 0.4213,
      "step": 4504
    },
    {
      "epoch": 0.2522678911412252,
      "grad_norm": 1.47342050075531,
      "learning_rate": 2.251e-05,
      "loss": 0.4968,
      "step": 4505
    },
    {
      "epoch": 0.25232388845335424,
      "grad_norm": 12.813328742980957,
      "learning_rate": 2.2515e-05,
      "loss": 0.4618,
      "step": 4506
    },
    {
      "epoch": 0.25237988576548326,
      "grad_norm": 1.367955207824707,
      "learning_rate": 2.252e-05,
      "loss": 0.4697,
      "step": 4507
    },
    {
      "epoch": 0.2524358830776123,
      "grad_norm": 1.470633625984192,
      "learning_rate": 2.2525000000000002e-05,
      "loss": 0.4447,
      "step": 4508
    },
    {
      "epoch": 0.2524918803897413,
      "grad_norm": 1.6107702255249023,
      "learning_rate": 2.253e-05,
      "loss": 0.5733,
      "step": 4509
    },
    {
      "epoch": 0.2525478777018703,
      "grad_norm": 1.2405142784118652,
      "learning_rate": 2.2535e-05,
      "loss": 0.5126,
      "step": 4510
    },
    {
      "epoch": 0.25260387501399933,
      "grad_norm": 1.1081312894821167,
      "learning_rate": 2.254e-05,
      "loss": 0.326,
      "step": 4511
    },
    {
      "epoch": 0.25265987232612835,
      "grad_norm": 1.2331016063690186,
      "learning_rate": 2.2545000000000002e-05,
      "loss": 0.4817,
      "step": 4512
    },
    {
      "epoch": 0.2527158696382574,
      "grad_norm": 1.1779156923294067,
      "learning_rate": 2.2550000000000003e-05,
      "loss": 0.4658,
      "step": 4513
    },
    {
      "epoch": 0.2527718669503864,
      "grad_norm": 1.5347821712493896,
      "learning_rate": 2.2555e-05,
      "loss": 0.4134,
      "step": 4514
    },
    {
      "epoch": 0.2528278642625154,
      "grad_norm": 1.191646695137024,
      "learning_rate": 2.256e-05,
      "loss": 0.5631,
      "step": 4515
    },
    {
      "epoch": 0.25288386157464443,
      "grad_norm": 1.340574860572815,
      "learning_rate": 2.2565e-05,
      "loss": 0.4469,
      "step": 4516
    },
    {
      "epoch": 0.25293985888677345,
      "grad_norm": 1.1392805576324463,
      "learning_rate": 2.2570000000000002e-05,
      "loss": 0.4925,
      "step": 4517
    },
    {
      "epoch": 0.25299585619890247,
      "grad_norm": 1.2708933353424072,
      "learning_rate": 2.2575000000000003e-05,
      "loss": 0.4406,
      "step": 4518
    },
    {
      "epoch": 0.2530518535110315,
      "grad_norm": 1.3625980615615845,
      "learning_rate": 2.258e-05,
      "loss": 0.497,
      "step": 4519
    },
    {
      "epoch": 0.2531078508231605,
      "grad_norm": 0.9875698685646057,
      "learning_rate": 2.2585e-05,
      "loss": 0.3762,
      "step": 4520
    },
    {
      "epoch": 0.2531638481352895,
      "grad_norm": 1.1133662462234497,
      "learning_rate": 2.259e-05,
      "loss": 0.3702,
      "step": 4521
    },
    {
      "epoch": 0.25321984544741855,
      "grad_norm": 1.538754940032959,
      "learning_rate": 2.2595000000000003e-05,
      "loss": 0.5965,
      "step": 4522
    },
    {
      "epoch": 0.25327584275954756,
      "grad_norm": 1.3976726531982422,
      "learning_rate": 2.26e-05,
      "loss": 0.3976,
      "step": 4523
    },
    {
      "epoch": 0.2533318400716766,
      "grad_norm": 1.1165637969970703,
      "learning_rate": 2.2605e-05,
      "loss": 0.389,
      "step": 4524
    },
    {
      "epoch": 0.2533878373838056,
      "grad_norm": 1.373055338859558,
      "learning_rate": 2.2610000000000002e-05,
      "loss": 0.4778,
      "step": 4525
    },
    {
      "epoch": 0.2534438346959346,
      "grad_norm": 1.2408130168914795,
      "learning_rate": 2.2615e-05,
      "loss": 0.6502,
      "step": 4526
    },
    {
      "epoch": 0.2534998320080636,
      "grad_norm": 0.9629552364349365,
      "learning_rate": 2.2620000000000004e-05,
      "loss": 0.4091,
      "step": 4527
    },
    {
      "epoch": 0.2535558293201926,
      "grad_norm": 1.3871678113937378,
      "learning_rate": 2.2625e-05,
      "loss": 0.4843,
      "step": 4528
    },
    {
      "epoch": 0.2536118266323216,
      "grad_norm": 1.508987545967102,
      "learning_rate": 2.2630000000000002e-05,
      "loss": 0.4926,
      "step": 4529
    },
    {
      "epoch": 0.25366782394445064,
      "grad_norm": 1.3006188869476318,
      "learning_rate": 2.2635e-05,
      "loss": 0.4367,
      "step": 4530
    },
    {
      "epoch": 0.25372382125657966,
      "grad_norm": 1.3513522148132324,
      "learning_rate": 2.264e-05,
      "loss": 0.4645,
      "step": 4531
    },
    {
      "epoch": 0.2537798185687087,
      "grad_norm": 1.414544939994812,
      "learning_rate": 2.2645e-05,
      "loss": 0.6809,
      "step": 4532
    },
    {
      "epoch": 0.2538358158808377,
      "grad_norm": 1.3013323545455933,
      "learning_rate": 2.265e-05,
      "loss": 0.5165,
      "step": 4533
    },
    {
      "epoch": 0.2538918131929667,
      "grad_norm": 1.2263951301574707,
      "learning_rate": 2.2655000000000002e-05,
      "loss": 0.5141,
      "step": 4534
    },
    {
      "epoch": 0.25394781050509574,
      "grad_norm": 1.914273738861084,
      "learning_rate": 2.266e-05,
      "loss": 0.4634,
      "step": 4535
    },
    {
      "epoch": 0.25400380781722476,
      "grad_norm": 1.4267395734786987,
      "learning_rate": 2.2665e-05,
      "loss": 0.4487,
      "step": 4536
    },
    {
      "epoch": 0.2540598051293538,
      "grad_norm": 1.1767939329147339,
      "learning_rate": 2.267e-05,
      "loss": 0.434,
      "step": 4537
    },
    {
      "epoch": 0.2541158024414828,
      "grad_norm": 1.0303541421890259,
      "learning_rate": 2.2675000000000002e-05,
      "loss": 0.3804,
      "step": 4538
    },
    {
      "epoch": 0.2541717997536118,
      "grad_norm": 1.2321490049362183,
      "learning_rate": 2.268e-05,
      "loss": 0.4092,
      "step": 4539
    },
    {
      "epoch": 0.25422779706574083,
      "grad_norm": 1.1032359600067139,
      "learning_rate": 2.2685e-05,
      "loss": 0.4038,
      "step": 4540
    },
    {
      "epoch": 0.25428379437786985,
      "grad_norm": 1.090529441833496,
      "learning_rate": 2.269e-05,
      "loss": 0.4244,
      "step": 4541
    },
    {
      "epoch": 0.2543397916899989,
      "grad_norm": 1.335461139678955,
      "learning_rate": 2.2695000000000002e-05,
      "loss": 0.5703,
      "step": 4542
    },
    {
      "epoch": 0.2543957890021279,
      "grad_norm": 1.1057159900665283,
      "learning_rate": 2.2700000000000003e-05,
      "loss": 0.3951,
      "step": 4543
    },
    {
      "epoch": 0.2544517863142569,
      "grad_norm": 1.1145004034042358,
      "learning_rate": 2.2705e-05,
      "loss": 0.3985,
      "step": 4544
    },
    {
      "epoch": 0.25450778362638593,
      "grad_norm": 1.1383142471313477,
      "learning_rate": 2.271e-05,
      "loss": 0.4392,
      "step": 4545
    },
    {
      "epoch": 0.25456378093851495,
      "grad_norm": 1.4747370481491089,
      "learning_rate": 2.2715e-05,
      "loss": 0.4301,
      "step": 4546
    },
    {
      "epoch": 0.25461977825064397,
      "grad_norm": 1.2272024154663086,
      "learning_rate": 2.2720000000000003e-05,
      "loss": 0.3965,
      "step": 4547
    },
    {
      "epoch": 0.254675775562773,
      "grad_norm": 1.4545232057571411,
      "learning_rate": 2.2725000000000003e-05,
      "loss": 0.4497,
      "step": 4548
    },
    {
      "epoch": 0.254731772874902,
      "grad_norm": 1.2840518951416016,
      "learning_rate": 2.273e-05,
      "loss": 0.3729,
      "step": 4549
    },
    {
      "epoch": 0.254787770187031,
      "grad_norm": 1.1751893758773804,
      "learning_rate": 2.2735000000000002e-05,
      "loss": 0.3626,
      "step": 4550
    },
    {
      "epoch": 0.25484376749916005,
      "grad_norm": 1.1711511611938477,
      "learning_rate": 2.274e-05,
      "loss": 0.3928,
      "step": 4551
    },
    {
      "epoch": 0.25489976481128906,
      "grad_norm": 1.4283241033554077,
      "learning_rate": 2.2745000000000003e-05,
      "loss": 0.4136,
      "step": 4552
    },
    {
      "epoch": 0.2549557621234181,
      "grad_norm": 1.297680139541626,
      "learning_rate": 2.275e-05,
      "loss": 0.4825,
      "step": 4553
    },
    {
      "epoch": 0.2550117594355471,
      "grad_norm": 1.3692355155944824,
      "learning_rate": 2.2755e-05,
      "loss": 0.4102,
      "step": 4554
    },
    {
      "epoch": 0.2550677567476761,
      "grad_norm": 1.418385624885559,
      "learning_rate": 2.2760000000000002e-05,
      "loss": 0.5126,
      "step": 4555
    },
    {
      "epoch": 0.25512375405980514,
      "grad_norm": 1.3031096458435059,
      "learning_rate": 2.2765e-05,
      "loss": 0.5565,
      "step": 4556
    },
    {
      "epoch": 0.25517975137193416,
      "grad_norm": 1.1401340961456299,
      "learning_rate": 2.2770000000000004e-05,
      "loss": 0.4568,
      "step": 4557
    },
    {
      "epoch": 0.2552357486840632,
      "grad_norm": 1.2938733100891113,
      "learning_rate": 2.2775e-05,
      "loss": 0.4326,
      "step": 4558
    },
    {
      "epoch": 0.2552917459961922,
      "grad_norm": 1.1616764068603516,
      "learning_rate": 2.2780000000000002e-05,
      "loss": 0.36,
      "step": 4559
    },
    {
      "epoch": 0.2553477433083212,
      "grad_norm": 1.2217586040496826,
      "learning_rate": 2.2785e-05,
      "loss": 0.5322,
      "step": 4560
    },
    {
      "epoch": 0.25540374062045024,
      "grad_norm": 1.309910774230957,
      "learning_rate": 2.279e-05,
      "loss": 0.6896,
      "step": 4561
    },
    {
      "epoch": 0.25545973793257926,
      "grad_norm": 1.2356170415878296,
      "learning_rate": 2.2795e-05,
      "loss": 0.4242,
      "step": 4562
    },
    {
      "epoch": 0.2555157352447083,
      "grad_norm": 1.4275574684143066,
      "learning_rate": 2.2800000000000002e-05,
      "loss": 0.5966,
      "step": 4563
    },
    {
      "epoch": 0.2555717325568373,
      "grad_norm": 1.1685105562210083,
      "learning_rate": 2.2805000000000003e-05,
      "loss": 0.3881,
      "step": 4564
    },
    {
      "epoch": 0.2556277298689663,
      "grad_norm": 1.184205174446106,
      "learning_rate": 2.281e-05,
      "loss": 0.5306,
      "step": 4565
    },
    {
      "epoch": 0.25568372718109533,
      "grad_norm": 1.4231535196304321,
      "learning_rate": 2.2815e-05,
      "loss": 0.3465,
      "step": 4566
    },
    {
      "epoch": 0.25573972449322435,
      "grad_norm": 1.3884516954421997,
      "learning_rate": 2.282e-05,
      "loss": 0.4995,
      "step": 4567
    },
    {
      "epoch": 0.2557957218053533,
      "grad_norm": 1.2389317750930786,
      "learning_rate": 2.2825000000000003e-05,
      "loss": 0.3583,
      "step": 4568
    },
    {
      "epoch": 0.25585171911748233,
      "grad_norm": 1.3554555177688599,
      "learning_rate": 2.283e-05,
      "loss": 0.4887,
      "step": 4569
    },
    {
      "epoch": 0.25590771642961135,
      "grad_norm": 1.494165301322937,
      "learning_rate": 2.2835e-05,
      "loss": 0.44,
      "step": 4570
    },
    {
      "epoch": 0.2559637137417404,
      "grad_norm": 1.0958404541015625,
      "learning_rate": 2.284e-05,
      "loss": 0.499,
      "step": 4571
    },
    {
      "epoch": 0.2560197110538694,
      "grad_norm": 1.5103360414505005,
      "learning_rate": 2.2845e-05,
      "loss": 0.5023,
      "step": 4572
    },
    {
      "epoch": 0.2560757083659984,
      "grad_norm": 1.204907774925232,
      "learning_rate": 2.2850000000000003e-05,
      "loss": 0.4267,
      "step": 4573
    },
    {
      "epoch": 0.25613170567812743,
      "grad_norm": 1.4339368343353271,
      "learning_rate": 2.2855e-05,
      "loss": 0.5432,
      "step": 4574
    },
    {
      "epoch": 0.25618770299025645,
      "grad_norm": 1.1898574829101562,
      "learning_rate": 2.286e-05,
      "loss": 0.3848,
      "step": 4575
    },
    {
      "epoch": 0.25624370030238547,
      "grad_norm": 1.053621530532837,
      "learning_rate": 2.2865e-05,
      "loss": 0.3681,
      "step": 4576
    },
    {
      "epoch": 0.2562996976145145,
      "grad_norm": 1.312021017074585,
      "learning_rate": 2.287e-05,
      "loss": 0.464,
      "step": 4577
    },
    {
      "epoch": 0.2563556949266435,
      "grad_norm": 1.321693778038025,
      "learning_rate": 2.2875e-05,
      "loss": 0.4943,
      "step": 4578
    },
    {
      "epoch": 0.2564116922387725,
      "grad_norm": 1.1725354194641113,
      "learning_rate": 2.288e-05,
      "loss": 0.3536,
      "step": 4579
    },
    {
      "epoch": 0.25646768955090155,
      "grad_norm": 1.213531732559204,
      "learning_rate": 2.2885000000000002e-05,
      "loss": 0.3764,
      "step": 4580
    },
    {
      "epoch": 0.25652368686303056,
      "grad_norm": 1.1212034225463867,
      "learning_rate": 2.289e-05,
      "loss": 0.5052,
      "step": 4581
    },
    {
      "epoch": 0.2565796841751596,
      "grad_norm": 1.2362000942230225,
      "learning_rate": 2.2895e-05,
      "loss": 0.4205,
      "step": 4582
    },
    {
      "epoch": 0.2566356814872886,
      "grad_norm": 1.1426849365234375,
      "learning_rate": 2.29e-05,
      "loss": 0.4204,
      "step": 4583
    },
    {
      "epoch": 0.2566916787994176,
      "grad_norm": 1.246262788772583,
      "learning_rate": 2.2905000000000002e-05,
      "loss": 0.3154,
      "step": 4584
    },
    {
      "epoch": 0.25674767611154664,
      "grad_norm": 1.2284398078918457,
      "learning_rate": 2.2910000000000003e-05,
      "loss": 0.4645,
      "step": 4585
    },
    {
      "epoch": 0.25680367342367566,
      "grad_norm": 1.2259362936019897,
      "learning_rate": 2.2915e-05,
      "loss": 0.4662,
      "step": 4586
    },
    {
      "epoch": 0.2568596707358047,
      "grad_norm": 1.1868433952331543,
      "learning_rate": 2.292e-05,
      "loss": 0.4249,
      "step": 4587
    },
    {
      "epoch": 0.2569156680479337,
      "grad_norm": 1.2697373628616333,
      "learning_rate": 2.2925e-05,
      "loss": 0.4276,
      "step": 4588
    },
    {
      "epoch": 0.2569716653600627,
      "grad_norm": 1.3835608959197998,
      "learning_rate": 2.2930000000000002e-05,
      "loss": 0.6027,
      "step": 4589
    },
    {
      "epoch": 0.25702766267219174,
      "grad_norm": 1.2019926309585571,
      "learning_rate": 2.2935e-05,
      "loss": 0.4143,
      "step": 4590
    },
    {
      "epoch": 0.25708365998432076,
      "grad_norm": 1.1837432384490967,
      "learning_rate": 2.294e-05,
      "loss": 0.4683,
      "step": 4591
    },
    {
      "epoch": 0.2571396572964498,
      "grad_norm": 1.3229570388793945,
      "learning_rate": 2.2945e-05,
      "loss": 0.6599,
      "step": 4592
    },
    {
      "epoch": 0.2571956546085788,
      "grad_norm": 1.2265925407409668,
      "learning_rate": 2.2950000000000002e-05,
      "loss": 0.4594,
      "step": 4593
    },
    {
      "epoch": 0.2572516519207078,
      "grad_norm": 1.4257705211639404,
      "learning_rate": 2.2955000000000003e-05,
      "loss": 0.3496,
      "step": 4594
    },
    {
      "epoch": 0.25730764923283683,
      "grad_norm": 1.018993854522705,
      "learning_rate": 2.296e-05,
      "loss": 0.4226,
      "step": 4595
    },
    {
      "epoch": 0.25736364654496585,
      "grad_norm": 1.0288788080215454,
      "learning_rate": 2.2965e-05,
      "loss": 0.3671,
      "step": 4596
    },
    {
      "epoch": 0.25741964385709487,
      "grad_norm": 1.3952146768569946,
      "learning_rate": 2.297e-05,
      "loss": 0.3726,
      "step": 4597
    },
    {
      "epoch": 0.2574756411692239,
      "grad_norm": 1.4173377752304077,
      "learning_rate": 2.2975000000000003e-05,
      "loss": 0.549,
      "step": 4598
    },
    {
      "epoch": 0.2575316384813529,
      "grad_norm": 1.2097254991531372,
      "learning_rate": 2.298e-05,
      "loss": 0.5329,
      "step": 4599
    },
    {
      "epoch": 0.25758763579348193,
      "grad_norm": 1.0523312091827393,
      "learning_rate": 2.2985e-05,
      "loss": 0.4327,
      "step": 4600
    },
    {
      "epoch": 0.25764363310561095,
      "grad_norm": 1.171477198600769,
      "learning_rate": 2.2990000000000002e-05,
      "loss": 0.3262,
      "step": 4601
    },
    {
      "epoch": 0.25769963041773997,
      "grad_norm": 1.3108364343643188,
      "learning_rate": 2.2995e-05,
      "loss": 0.4516,
      "step": 4602
    },
    {
      "epoch": 0.257755627729869,
      "grad_norm": 1.0478096008300781,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 0.3064,
      "step": 4603
    },
    {
      "epoch": 0.257811625041998,
      "grad_norm": 1.145826816558838,
      "learning_rate": 2.3005e-05,
      "loss": 0.4085,
      "step": 4604
    },
    {
      "epoch": 0.257867622354127,
      "grad_norm": 1.2956432104110718,
      "learning_rate": 2.301e-05,
      "loss": 0.4424,
      "step": 4605
    },
    {
      "epoch": 0.25792361966625604,
      "grad_norm": 1.2124707698822021,
      "learning_rate": 2.3015e-05,
      "loss": 0.4248,
      "step": 4606
    },
    {
      "epoch": 0.25797961697838506,
      "grad_norm": 1.0245577096939087,
      "learning_rate": 2.302e-05,
      "loss": 0.4546,
      "step": 4607
    },
    {
      "epoch": 0.2580356142905141,
      "grad_norm": 1.367613434791565,
      "learning_rate": 2.3025e-05,
      "loss": 0.4073,
      "step": 4608
    },
    {
      "epoch": 0.2580916116026431,
      "grad_norm": 1.5224664211273193,
      "learning_rate": 2.303e-05,
      "loss": 0.4822,
      "step": 4609
    },
    {
      "epoch": 0.25814760891477206,
      "grad_norm": 1.2754371166229248,
      "learning_rate": 2.3035000000000002e-05,
      "loss": 0.5098,
      "step": 4610
    },
    {
      "epoch": 0.2582036062269011,
      "grad_norm": 1.6209826469421387,
      "learning_rate": 2.304e-05,
      "loss": 0.4425,
      "step": 4611
    },
    {
      "epoch": 0.2582596035390301,
      "grad_norm": 1.2482712268829346,
      "learning_rate": 2.3045e-05,
      "loss": 0.5628,
      "step": 4612
    },
    {
      "epoch": 0.2583156008511591,
      "grad_norm": 1.3711938858032227,
      "learning_rate": 2.305e-05,
      "loss": 0.4381,
      "step": 4613
    },
    {
      "epoch": 0.25837159816328814,
      "grad_norm": 1.0197200775146484,
      "learning_rate": 2.3055000000000002e-05,
      "loss": 0.3743,
      "step": 4614
    },
    {
      "epoch": 0.25842759547541716,
      "grad_norm": 1.2958955764770508,
      "learning_rate": 2.306e-05,
      "loss": 0.386,
      "step": 4615
    },
    {
      "epoch": 0.2584835927875462,
      "grad_norm": 1.3102233409881592,
      "learning_rate": 2.3065e-05,
      "loss": 0.5122,
      "step": 4616
    },
    {
      "epoch": 0.2585395900996752,
      "grad_norm": 1.064424991607666,
      "learning_rate": 2.307e-05,
      "loss": 0.4189,
      "step": 4617
    },
    {
      "epoch": 0.2585955874118042,
      "grad_norm": 1.2254605293273926,
      "learning_rate": 2.3075000000000002e-05,
      "loss": 0.4455,
      "step": 4618
    },
    {
      "epoch": 0.25865158472393324,
      "grad_norm": 1.271575927734375,
      "learning_rate": 2.3080000000000003e-05,
      "loss": 0.4599,
      "step": 4619
    },
    {
      "epoch": 0.25870758203606226,
      "grad_norm": 1.0416302680969238,
      "learning_rate": 2.3085e-05,
      "loss": 0.4125,
      "step": 4620
    },
    {
      "epoch": 0.2587635793481913,
      "grad_norm": 1.469797134399414,
      "learning_rate": 2.309e-05,
      "loss": 0.4717,
      "step": 4621
    },
    {
      "epoch": 0.2588195766603203,
      "grad_norm": 1.3082259893417358,
      "learning_rate": 2.3095e-05,
      "loss": 0.4036,
      "step": 4622
    },
    {
      "epoch": 0.2588755739724493,
      "grad_norm": 1.3652042150497437,
      "learning_rate": 2.3100000000000002e-05,
      "loss": 0.5413,
      "step": 4623
    },
    {
      "epoch": 0.25893157128457833,
      "grad_norm": 1.3426355123519897,
      "learning_rate": 2.3105000000000003e-05,
      "loss": 0.5378,
      "step": 4624
    },
    {
      "epoch": 0.25898756859670735,
      "grad_norm": 1.111953854560852,
      "learning_rate": 2.311e-05,
      "loss": 0.2973,
      "step": 4625
    },
    {
      "epoch": 0.25904356590883637,
      "grad_norm": 1.1020268201828003,
      "learning_rate": 2.3115e-05,
      "loss": 0.4528,
      "step": 4626
    },
    {
      "epoch": 0.2590995632209654,
      "grad_norm": 1.0999373197555542,
      "learning_rate": 2.312e-05,
      "loss": 0.3916,
      "step": 4627
    },
    {
      "epoch": 0.2591555605330944,
      "grad_norm": 1.2418850660324097,
      "learning_rate": 2.3125000000000003e-05,
      "loss": 0.483,
      "step": 4628
    },
    {
      "epoch": 0.25921155784522343,
      "grad_norm": 1.1667850017547607,
      "learning_rate": 2.313e-05,
      "loss": 0.4024,
      "step": 4629
    },
    {
      "epoch": 0.25926755515735245,
      "grad_norm": 1.1083531379699707,
      "learning_rate": 2.3135e-05,
      "loss": 0.4499,
      "step": 4630
    },
    {
      "epoch": 0.25932355246948147,
      "grad_norm": 1.260475516319275,
      "learning_rate": 2.3140000000000002e-05,
      "loss": 0.5076,
      "step": 4631
    },
    {
      "epoch": 0.2593795497816105,
      "grad_norm": 1.1606007814407349,
      "learning_rate": 2.3145e-05,
      "loss": 0.4015,
      "step": 4632
    },
    {
      "epoch": 0.2594355470937395,
      "grad_norm": 1.4891753196716309,
      "learning_rate": 2.3150000000000004e-05,
      "loss": 0.5287,
      "step": 4633
    },
    {
      "epoch": 0.2594915444058685,
      "grad_norm": 1.2093613147735596,
      "learning_rate": 2.3155e-05,
      "loss": 0.5178,
      "step": 4634
    },
    {
      "epoch": 0.25954754171799754,
      "grad_norm": 1.1132348775863647,
      "learning_rate": 2.3160000000000002e-05,
      "loss": 0.5762,
      "step": 4635
    },
    {
      "epoch": 0.25960353903012656,
      "grad_norm": 1.8130476474761963,
      "learning_rate": 2.3165e-05,
      "loss": 0.553,
      "step": 4636
    },
    {
      "epoch": 0.2596595363422556,
      "grad_norm": 1.2123457193374634,
      "learning_rate": 2.317e-05,
      "loss": 0.6021,
      "step": 4637
    },
    {
      "epoch": 0.2597155336543846,
      "grad_norm": 1.457024335861206,
      "learning_rate": 2.3175e-05,
      "loss": 0.5136,
      "step": 4638
    },
    {
      "epoch": 0.2597715309665136,
      "grad_norm": 1.319503664970398,
      "learning_rate": 2.318e-05,
      "loss": 0.4162,
      "step": 4639
    },
    {
      "epoch": 0.25982752827864264,
      "grad_norm": 1.2477689981460571,
      "learning_rate": 2.3185000000000002e-05,
      "loss": 0.4707,
      "step": 4640
    },
    {
      "epoch": 0.25988352559077166,
      "grad_norm": 1.1592180728912354,
      "learning_rate": 2.319e-05,
      "loss": 0.2992,
      "step": 4641
    },
    {
      "epoch": 0.2599395229029007,
      "grad_norm": 1.3189729452133179,
      "learning_rate": 2.3195e-05,
      "loss": 0.52,
      "step": 4642
    },
    {
      "epoch": 0.2599955202150297,
      "grad_norm": 1.237096905708313,
      "learning_rate": 2.32e-05,
      "loss": 0.6822,
      "step": 4643
    },
    {
      "epoch": 0.2600515175271587,
      "grad_norm": 1.4841668605804443,
      "learning_rate": 2.3205000000000002e-05,
      "loss": 0.4308,
      "step": 4644
    },
    {
      "epoch": 0.26010751483928773,
      "grad_norm": 1.2012847661972046,
      "learning_rate": 2.321e-05,
      "loss": 0.5243,
      "step": 4645
    },
    {
      "epoch": 0.26016351215141675,
      "grad_norm": 1.1921275854110718,
      "learning_rate": 2.3215e-05,
      "loss": 0.4918,
      "step": 4646
    },
    {
      "epoch": 0.2602195094635458,
      "grad_norm": 1.1845691204071045,
      "learning_rate": 2.322e-05,
      "loss": 0.384,
      "step": 4647
    },
    {
      "epoch": 0.2602755067756748,
      "grad_norm": 0.9477851986885071,
      "learning_rate": 2.3225000000000002e-05,
      "loss": 0.3095,
      "step": 4648
    },
    {
      "epoch": 0.2603315040878038,
      "grad_norm": 1.0472500324249268,
      "learning_rate": 2.3230000000000003e-05,
      "loss": 0.349,
      "step": 4649
    },
    {
      "epoch": 0.26038750139993283,
      "grad_norm": 1.1473208665847778,
      "learning_rate": 2.3235e-05,
      "loss": 0.3444,
      "step": 4650
    },
    {
      "epoch": 0.2604434987120618,
      "grad_norm": 1.0765466690063477,
      "learning_rate": 2.324e-05,
      "loss": 0.356,
      "step": 4651
    },
    {
      "epoch": 0.2604994960241908,
      "grad_norm": 1.3327019214630127,
      "learning_rate": 2.3245e-05,
      "loss": 0.4309,
      "step": 4652
    },
    {
      "epoch": 0.26055549333631983,
      "grad_norm": 1.0411831140518188,
      "learning_rate": 2.3250000000000003e-05,
      "loss": 0.4155,
      "step": 4653
    },
    {
      "epoch": 0.26061149064844885,
      "grad_norm": 1.1315257549285889,
      "learning_rate": 2.3255e-05,
      "loss": 0.3979,
      "step": 4654
    },
    {
      "epoch": 0.26066748796057787,
      "grad_norm": 1.1296149492263794,
      "learning_rate": 2.326e-05,
      "loss": 0.3826,
      "step": 4655
    },
    {
      "epoch": 0.2607234852727069,
      "grad_norm": 1.3033127784729004,
      "learning_rate": 2.3265000000000002e-05,
      "loss": 0.4008,
      "step": 4656
    },
    {
      "epoch": 0.2607794825848359,
      "grad_norm": 1.210599660873413,
      "learning_rate": 2.327e-05,
      "loss": 0.5096,
      "step": 4657
    },
    {
      "epoch": 0.26083547989696493,
      "grad_norm": 1.4044979810714722,
      "learning_rate": 2.3275000000000003e-05,
      "loss": 0.4934,
      "step": 4658
    },
    {
      "epoch": 0.26089147720909395,
      "grad_norm": 4.5497307777404785,
      "learning_rate": 2.328e-05,
      "loss": 0.3402,
      "step": 4659
    },
    {
      "epoch": 0.26094747452122297,
      "grad_norm": 1.3142727613449097,
      "learning_rate": 2.3285e-05,
      "loss": 0.397,
      "step": 4660
    },
    {
      "epoch": 0.261003471833352,
      "grad_norm": 1.4795187711715698,
      "learning_rate": 2.3290000000000002e-05,
      "loss": 0.4657,
      "step": 4661
    },
    {
      "epoch": 0.261059469145481,
      "grad_norm": 1.0734316110610962,
      "learning_rate": 2.3295e-05,
      "loss": 0.4536,
      "step": 4662
    },
    {
      "epoch": 0.26111546645761,
      "grad_norm": 1.2687352895736694,
      "learning_rate": 2.3300000000000004e-05,
      "loss": 0.4132,
      "step": 4663
    },
    {
      "epoch": 0.26117146376973904,
      "grad_norm": 1.1907485723495483,
      "learning_rate": 2.3305e-05,
      "loss": 0.5572,
      "step": 4664
    },
    {
      "epoch": 0.26122746108186806,
      "grad_norm": 1.0302767753601074,
      "learning_rate": 2.3310000000000002e-05,
      "loss": 0.3827,
      "step": 4665
    },
    {
      "epoch": 0.2612834583939971,
      "grad_norm": 2.662245750427246,
      "learning_rate": 2.3315e-05,
      "loss": 0.3729,
      "step": 4666
    },
    {
      "epoch": 0.2613394557061261,
      "grad_norm": 1.0700167417526245,
      "learning_rate": 2.332e-05,
      "loss": 0.5333,
      "step": 4667
    },
    {
      "epoch": 0.2613954530182551,
      "grad_norm": 1.3027064800262451,
      "learning_rate": 2.3325e-05,
      "loss": 0.449,
      "step": 4668
    },
    {
      "epoch": 0.26145145033038414,
      "grad_norm": 1.1364097595214844,
      "learning_rate": 2.3330000000000002e-05,
      "loss": 0.3691,
      "step": 4669
    },
    {
      "epoch": 0.26150744764251316,
      "grad_norm": 1.3689920902252197,
      "learning_rate": 2.3335000000000003e-05,
      "loss": 0.5063,
      "step": 4670
    },
    {
      "epoch": 0.2615634449546422,
      "grad_norm": 1.2601542472839355,
      "learning_rate": 2.334e-05,
      "loss": 0.48,
      "step": 4671
    },
    {
      "epoch": 0.2616194422667712,
      "grad_norm": 1.0073635578155518,
      "learning_rate": 2.3345e-05,
      "loss": 0.3379,
      "step": 4672
    },
    {
      "epoch": 0.2616754395789002,
      "grad_norm": 1.1468099355697632,
      "learning_rate": 2.3350000000000002e-05,
      "loss": 0.3992,
      "step": 4673
    },
    {
      "epoch": 0.26173143689102923,
      "grad_norm": 1.1332155466079712,
      "learning_rate": 2.3355000000000003e-05,
      "loss": 0.4549,
      "step": 4674
    },
    {
      "epoch": 0.26178743420315825,
      "grad_norm": 4.049678325653076,
      "learning_rate": 2.336e-05,
      "loss": 0.4229,
      "step": 4675
    },
    {
      "epoch": 0.2618434315152873,
      "grad_norm": 0.9749440550804138,
      "learning_rate": 2.3365e-05,
      "loss": 0.3084,
      "step": 4676
    },
    {
      "epoch": 0.2618994288274163,
      "grad_norm": 1.0869070291519165,
      "learning_rate": 2.337e-05,
      "loss": 0.4879,
      "step": 4677
    },
    {
      "epoch": 0.2619554261395453,
      "grad_norm": 1.350502848625183,
      "learning_rate": 2.3375000000000002e-05,
      "loss": 0.483,
      "step": 4678
    },
    {
      "epoch": 0.26201142345167433,
      "grad_norm": 1.2524774074554443,
      "learning_rate": 2.3380000000000003e-05,
      "loss": 0.4149,
      "step": 4679
    },
    {
      "epoch": 0.26206742076380335,
      "grad_norm": 1.0579578876495361,
      "learning_rate": 2.3385e-05,
      "loss": 0.3859,
      "step": 4680
    },
    {
      "epoch": 0.26212341807593237,
      "grad_norm": 1.2938880920410156,
      "learning_rate": 2.339e-05,
      "loss": 0.4728,
      "step": 4681
    },
    {
      "epoch": 0.2621794153880614,
      "grad_norm": 1.2401838302612305,
      "learning_rate": 2.3395e-05,
      "loss": 0.4062,
      "step": 4682
    },
    {
      "epoch": 0.2622354127001904,
      "grad_norm": 1.2220298051834106,
      "learning_rate": 2.3400000000000003e-05,
      "loss": 0.3831,
      "step": 4683
    },
    {
      "epoch": 0.2622914100123194,
      "grad_norm": 1.1159085035324097,
      "learning_rate": 2.3405e-05,
      "loss": 0.4694,
      "step": 4684
    },
    {
      "epoch": 0.26234740732444845,
      "grad_norm": 1.4172265529632568,
      "learning_rate": 2.341e-05,
      "loss": 0.376,
      "step": 4685
    },
    {
      "epoch": 0.26240340463657746,
      "grad_norm": 1.4832614660263062,
      "learning_rate": 2.3415000000000002e-05,
      "loss": 0.4544,
      "step": 4686
    },
    {
      "epoch": 0.2624594019487065,
      "grad_norm": 1.2928845882415771,
      "learning_rate": 2.342e-05,
      "loss": 0.5758,
      "step": 4687
    },
    {
      "epoch": 0.2625153992608355,
      "grad_norm": 1.3649362325668335,
      "learning_rate": 2.3425000000000004e-05,
      "loss": 0.435,
      "step": 4688
    },
    {
      "epoch": 0.2625713965729645,
      "grad_norm": 1.32871413230896,
      "learning_rate": 2.343e-05,
      "loss": 0.4684,
      "step": 4689
    },
    {
      "epoch": 0.26262739388509354,
      "grad_norm": 1.632665753364563,
      "learning_rate": 2.3435000000000002e-05,
      "loss": 0.5861,
      "step": 4690
    },
    {
      "epoch": 0.26268339119722256,
      "grad_norm": 1.1872894763946533,
      "learning_rate": 2.344e-05,
      "loss": 0.4616,
      "step": 4691
    },
    {
      "epoch": 0.2627393885093515,
      "grad_norm": 1.101766586303711,
      "learning_rate": 2.3445e-05,
      "loss": 0.3728,
      "step": 4692
    },
    {
      "epoch": 0.26279538582148054,
      "grad_norm": 1.2853742837905884,
      "learning_rate": 2.345e-05,
      "loss": 0.4049,
      "step": 4693
    },
    {
      "epoch": 0.26285138313360956,
      "grad_norm": 1.1566184759140015,
      "learning_rate": 2.3455e-05,
      "loss": 0.3942,
      "step": 4694
    },
    {
      "epoch": 0.2629073804457386,
      "grad_norm": 1.2635637521743774,
      "learning_rate": 2.3460000000000002e-05,
      "loss": 0.5816,
      "step": 4695
    },
    {
      "epoch": 0.2629633777578676,
      "grad_norm": 1.4154095649719238,
      "learning_rate": 2.3465e-05,
      "loss": 0.4768,
      "step": 4696
    },
    {
      "epoch": 0.2630193750699966,
      "grad_norm": 1.0704032182693481,
      "learning_rate": 2.347e-05,
      "loss": 0.3733,
      "step": 4697
    },
    {
      "epoch": 0.26307537238212564,
      "grad_norm": 1.040293574333191,
      "learning_rate": 2.3475e-05,
      "loss": 0.4653,
      "step": 4698
    },
    {
      "epoch": 0.26313136969425466,
      "grad_norm": 1.2215741872787476,
      "learning_rate": 2.3480000000000002e-05,
      "loss": 0.3926,
      "step": 4699
    },
    {
      "epoch": 0.2631873670063837,
      "grad_norm": 1.3420178890228271,
      "learning_rate": 2.3485000000000003e-05,
      "loss": 0.5665,
      "step": 4700
    },
    {
      "epoch": 0.2632433643185127,
      "grad_norm": 1.307134985923767,
      "learning_rate": 2.349e-05,
      "loss": 0.3772,
      "step": 4701
    },
    {
      "epoch": 0.2632993616306417,
      "grad_norm": 1.4651503562927246,
      "learning_rate": 2.3495e-05,
      "loss": 0.5187,
      "step": 4702
    },
    {
      "epoch": 0.26335535894277073,
      "grad_norm": 1.3955353498458862,
      "learning_rate": 2.35e-05,
      "loss": 0.3472,
      "step": 4703
    },
    {
      "epoch": 0.26341135625489975,
      "grad_norm": 1.3448361158370972,
      "learning_rate": 2.3505000000000003e-05,
      "loss": 0.5255,
      "step": 4704
    },
    {
      "epoch": 0.2634673535670288,
      "grad_norm": 1.1564198732376099,
      "learning_rate": 2.351e-05,
      "loss": 0.4308,
      "step": 4705
    },
    {
      "epoch": 0.2635233508791578,
      "grad_norm": 1.274255633354187,
      "learning_rate": 2.3515e-05,
      "loss": 0.4997,
      "step": 4706
    },
    {
      "epoch": 0.2635793481912868,
      "grad_norm": 1.2160266637802124,
      "learning_rate": 2.3520000000000002e-05,
      "loss": 0.4688,
      "step": 4707
    },
    {
      "epoch": 0.26363534550341583,
      "grad_norm": 1.0255680084228516,
      "learning_rate": 2.3525e-05,
      "loss": 0.291,
      "step": 4708
    },
    {
      "epoch": 0.26369134281554485,
      "grad_norm": 1.1800979375839233,
      "learning_rate": 2.3530000000000003e-05,
      "loss": 0.4176,
      "step": 4709
    },
    {
      "epoch": 0.26374734012767387,
      "grad_norm": 1.404802918434143,
      "learning_rate": 2.3535e-05,
      "loss": 0.4543,
      "step": 4710
    },
    {
      "epoch": 0.2638033374398029,
      "grad_norm": 3.526089668273926,
      "learning_rate": 2.354e-05,
      "loss": 0.4589,
      "step": 4711
    },
    {
      "epoch": 0.2638593347519319,
      "grad_norm": 1.169753909111023,
      "learning_rate": 2.3545e-05,
      "loss": 0.3912,
      "step": 4712
    },
    {
      "epoch": 0.2639153320640609,
      "grad_norm": 1.3709254264831543,
      "learning_rate": 2.355e-05,
      "loss": 0.5337,
      "step": 4713
    },
    {
      "epoch": 0.26397132937618994,
      "grad_norm": 1.6002558469772339,
      "learning_rate": 2.3555e-05,
      "loss": 0.3886,
      "step": 4714
    },
    {
      "epoch": 0.26402732668831896,
      "grad_norm": 1.47130286693573,
      "learning_rate": 2.356e-05,
      "loss": 0.4143,
      "step": 4715
    },
    {
      "epoch": 0.264083324000448,
      "grad_norm": 1.2650200128555298,
      "learning_rate": 2.3565000000000002e-05,
      "loss": 0.4887,
      "step": 4716
    },
    {
      "epoch": 0.264139321312577,
      "grad_norm": 1.438278317451477,
      "learning_rate": 2.357e-05,
      "loss": 0.5166,
      "step": 4717
    },
    {
      "epoch": 0.264195318624706,
      "grad_norm": 1.1774498224258423,
      "learning_rate": 2.3575e-05,
      "loss": 0.3873,
      "step": 4718
    },
    {
      "epoch": 0.26425131593683504,
      "grad_norm": 1.2495230436325073,
      "learning_rate": 2.358e-05,
      "loss": 0.4097,
      "step": 4719
    },
    {
      "epoch": 0.26430731324896406,
      "grad_norm": 1.2106897830963135,
      "learning_rate": 2.3585000000000002e-05,
      "loss": 0.4468,
      "step": 4720
    },
    {
      "epoch": 0.2643633105610931,
      "grad_norm": 1.0755109786987305,
      "learning_rate": 2.359e-05,
      "loss": 0.4026,
      "step": 4721
    },
    {
      "epoch": 0.2644193078732221,
      "grad_norm": 1.3256511688232422,
      "learning_rate": 2.3595e-05,
      "loss": 0.4691,
      "step": 4722
    },
    {
      "epoch": 0.2644753051853511,
      "grad_norm": 1.1157100200653076,
      "learning_rate": 2.36e-05,
      "loss": 0.4084,
      "step": 4723
    },
    {
      "epoch": 0.26453130249748014,
      "grad_norm": 1.227482795715332,
      "learning_rate": 2.3605000000000002e-05,
      "loss": 0.463,
      "step": 4724
    },
    {
      "epoch": 0.26458729980960916,
      "grad_norm": 1.2092044353485107,
      "learning_rate": 2.3610000000000003e-05,
      "loss": 0.4472,
      "step": 4725
    },
    {
      "epoch": 0.2646432971217382,
      "grad_norm": 1.1431828737258911,
      "learning_rate": 2.3615e-05,
      "loss": 0.3422,
      "step": 4726
    },
    {
      "epoch": 0.2646992944338672,
      "grad_norm": 1.1278163194656372,
      "learning_rate": 2.362e-05,
      "loss": 0.3907,
      "step": 4727
    },
    {
      "epoch": 0.2647552917459962,
      "grad_norm": 2.944944381713867,
      "learning_rate": 2.3624999999999998e-05,
      "loss": 0.3977,
      "step": 4728
    },
    {
      "epoch": 0.26481128905812523,
      "grad_norm": 1.0838154554367065,
      "learning_rate": 2.3630000000000002e-05,
      "loss": 0.4222,
      "step": 4729
    },
    {
      "epoch": 0.26486728637025425,
      "grad_norm": 1.3145902156829834,
      "learning_rate": 2.3635000000000003e-05,
      "loss": 0.4228,
      "step": 4730
    },
    {
      "epoch": 0.26492328368238327,
      "grad_norm": 1.237244725227356,
      "learning_rate": 2.364e-05,
      "loss": 0.496,
      "step": 4731
    },
    {
      "epoch": 0.2649792809945123,
      "grad_norm": 1.3676848411560059,
      "learning_rate": 2.3645e-05,
      "loss": 0.4149,
      "step": 4732
    },
    {
      "epoch": 0.2650352783066413,
      "grad_norm": 1.361025333404541,
      "learning_rate": 2.365e-05,
      "loss": 0.3989,
      "step": 4733
    },
    {
      "epoch": 0.2650912756187703,
      "grad_norm": 1.2569867372512817,
      "learning_rate": 2.3655000000000003e-05,
      "loss": 0.4032,
      "step": 4734
    },
    {
      "epoch": 0.2651472729308993,
      "grad_norm": 1.121619701385498,
      "learning_rate": 2.366e-05,
      "loss": 0.4134,
      "step": 4735
    },
    {
      "epoch": 0.2652032702430283,
      "grad_norm": 1.3902980089187622,
      "learning_rate": 2.3665e-05,
      "loss": 0.4361,
      "step": 4736
    },
    {
      "epoch": 0.26525926755515733,
      "grad_norm": 1.2956088781356812,
      "learning_rate": 2.3670000000000002e-05,
      "loss": 0.5531,
      "step": 4737
    },
    {
      "epoch": 0.26531526486728635,
      "grad_norm": 1.6822242736816406,
      "learning_rate": 2.3675e-05,
      "loss": 0.5115,
      "step": 4738
    },
    {
      "epoch": 0.26537126217941537,
      "grad_norm": 1.040383219718933,
      "learning_rate": 2.3680000000000004e-05,
      "loss": 0.4566,
      "step": 4739
    },
    {
      "epoch": 0.2654272594915444,
      "grad_norm": 1.2056008577346802,
      "learning_rate": 2.3685e-05,
      "loss": 0.3719,
      "step": 4740
    },
    {
      "epoch": 0.2654832568036734,
      "grad_norm": 1.303937554359436,
      "learning_rate": 2.3690000000000002e-05,
      "loss": 0.4788,
      "step": 4741
    },
    {
      "epoch": 0.2655392541158024,
      "grad_norm": 1.337628960609436,
      "learning_rate": 2.3695e-05,
      "loss": 0.45,
      "step": 4742
    },
    {
      "epoch": 0.26559525142793144,
      "grad_norm": 1.2148423194885254,
      "learning_rate": 2.37e-05,
      "loss": 0.448,
      "step": 4743
    },
    {
      "epoch": 0.26565124874006046,
      "grad_norm": 1.4554351568222046,
      "learning_rate": 2.3705e-05,
      "loss": 0.3779,
      "step": 4744
    },
    {
      "epoch": 0.2657072460521895,
      "grad_norm": 1.3510658740997314,
      "learning_rate": 2.371e-05,
      "loss": 0.4273,
      "step": 4745
    },
    {
      "epoch": 0.2657632433643185,
      "grad_norm": 1.185594081878662,
      "learning_rate": 2.3715000000000002e-05,
      "loss": 0.5057,
      "step": 4746
    },
    {
      "epoch": 0.2658192406764475,
      "grad_norm": 1.1551557779312134,
      "learning_rate": 2.372e-05,
      "loss": 0.3967,
      "step": 4747
    },
    {
      "epoch": 0.26587523798857654,
      "grad_norm": 1.1626313924789429,
      "learning_rate": 2.3725e-05,
      "loss": 0.423,
      "step": 4748
    },
    {
      "epoch": 0.26593123530070556,
      "grad_norm": 1.1862657070159912,
      "learning_rate": 2.373e-05,
      "loss": 0.4447,
      "step": 4749
    },
    {
      "epoch": 0.2659872326128346,
      "grad_norm": 0.9451472163200378,
      "learning_rate": 2.3735000000000002e-05,
      "loss": 0.3167,
      "step": 4750
    },
    {
      "epoch": 0.2660432299249636,
      "grad_norm": 1.0458323955535889,
      "learning_rate": 2.374e-05,
      "loss": 0.5032,
      "step": 4751
    },
    {
      "epoch": 0.2660992272370926,
      "grad_norm": 1.0098243951797485,
      "learning_rate": 2.3745e-05,
      "loss": 0.399,
      "step": 4752
    },
    {
      "epoch": 0.26615522454922164,
      "grad_norm": 1.2983150482177734,
      "learning_rate": 2.375e-05,
      "loss": 0.4968,
      "step": 4753
    },
    {
      "epoch": 0.26621122186135066,
      "grad_norm": 1.238157868385315,
      "learning_rate": 2.3755000000000002e-05,
      "loss": 0.4248,
      "step": 4754
    },
    {
      "epoch": 0.2662672191734797,
      "grad_norm": 1.2264575958251953,
      "learning_rate": 2.3760000000000003e-05,
      "loss": 0.4707,
      "step": 4755
    },
    {
      "epoch": 0.2663232164856087,
      "grad_norm": 1.0791478157043457,
      "learning_rate": 2.3765e-05,
      "loss": 0.3632,
      "step": 4756
    },
    {
      "epoch": 0.2663792137977377,
      "grad_norm": 1.2311959266662598,
      "learning_rate": 2.377e-05,
      "loss": 0.4551,
      "step": 4757
    },
    {
      "epoch": 0.26643521110986673,
      "grad_norm": 1.0826047658920288,
      "learning_rate": 2.3775e-05,
      "loss": 0.3833,
      "step": 4758
    },
    {
      "epoch": 0.26649120842199575,
      "grad_norm": 1.1972943544387817,
      "learning_rate": 2.3780000000000003e-05,
      "loss": 0.4986,
      "step": 4759
    },
    {
      "epoch": 0.26654720573412477,
      "grad_norm": 1.5173351764678955,
      "learning_rate": 2.3785e-05,
      "loss": 0.4581,
      "step": 4760
    },
    {
      "epoch": 0.2666032030462538,
      "grad_norm": 1.081595540046692,
      "learning_rate": 2.379e-05,
      "loss": 0.4176,
      "step": 4761
    },
    {
      "epoch": 0.2666592003583828,
      "grad_norm": 1.485235333442688,
      "learning_rate": 2.3795000000000002e-05,
      "loss": 0.4463,
      "step": 4762
    },
    {
      "epoch": 0.2667151976705118,
      "grad_norm": 1.043321132659912,
      "learning_rate": 2.38e-05,
      "loss": 0.3156,
      "step": 4763
    },
    {
      "epoch": 0.26677119498264085,
      "grad_norm": 1.2274316549301147,
      "learning_rate": 2.3805000000000003e-05,
      "loss": 0.6176,
      "step": 4764
    },
    {
      "epoch": 0.26682719229476987,
      "grad_norm": 1.2730541229248047,
      "learning_rate": 2.381e-05,
      "loss": 0.4126,
      "step": 4765
    },
    {
      "epoch": 0.2668831896068989,
      "grad_norm": 1.2899019718170166,
      "learning_rate": 2.3815e-05,
      "loss": 0.3958,
      "step": 4766
    },
    {
      "epoch": 0.2669391869190279,
      "grad_norm": 1.1440935134887695,
      "learning_rate": 2.3820000000000002e-05,
      "loss": 0.3716,
      "step": 4767
    },
    {
      "epoch": 0.2669951842311569,
      "grad_norm": 1.1878410577774048,
      "learning_rate": 2.3825e-05,
      "loss": 0.4676,
      "step": 4768
    },
    {
      "epoch": 0.26705118154328594,
      "grad_norm": 1.21403169631958,
      "learning_rate": 2.3830000000000004e-05,
      "loss": 0.4589,
      "step": 4769
    },
    {
      "epoch": 0.26710717885541496,
      "grad_norm": 1.1536064147949219,
      "learning_rate": 2.3835e-05,
      "loss": 0.4081,
      "step": 4770
    },
    {
      "epoch": 0.267163176167544,
      "grad_norm": 1.207689642906189,
      "learning_rate": 2.3840000000000002e-05,
      "loss": 0.6582,
      "step": 4771
    },
    {
      "epoch": 0.267219173479673,
      "grad_norm": 1.1608504056930542,
      "learning_rate": 2.3845e-05,
      "loss": 0.4031,
      "step": 4772
    },
    {
      "epoch": 0.267275170791802,
      "grad_norm": 1.3046009540557861,
      "learning_rate": 2.385e-05,
      "loss": 0.4721,
      "step": 4773
    },
    {
      "epoch": 0.26733116810393104,
      "grad_norm": 1.1998666524887085,
      "learning_rate": 2.3855e-05,
      "loss": 0.3622,
      "step": 4774
    },
    {
      "epoch": 0.26738716541606,
      "grad_norm": 6.5772223472595215,
      "learning_rate": 2.3860000000000002e-05,
      "loss": 0.4529,
      "step": 4775
    },
    {
      "epoch": 0.267443162728189,
      "grad_norm": 1.2209452390670776,
      "learning_rate": 2.3865000000000003e-05,
      "loss": 0.3514,
      "step": 4776
    },
    {
      "epoch": 0.26749916004031804,
      "grad_norm": 1.2021710872650146,
      "learning_rate": 2.387e-05,
      "loss": 0.4096,
      "step": 4777
    },
    {
      "epoch": 0.26755515735244706,
      "grad_norm": 1.1438333988189697,
      "learning_rate": 2.3875e-05,
      "loss": 0.3843,
      "step": 4778
    },
    {
      "epoch": 0.2676111546645761,
      "grad_norm": 1.3314814567565918,
      "learning_rate": 2.3880000000000002e-05,
      "loss": 0.4809,
      "step": 4779
    },
    {
      "epoch": 0.2676671519767051,
      "grad_norm": 1.1786727905273438,
      "learning_rate": 2.3885000000000003e-05,
      "loss": 0.3968,
      "step": 4780
    },
    {
      "epoch": 0.2677231492888341,
      "grad_norm": 1.0819017887115479,
      "learning_rate": 2.389e-05,
      "loss": 0.3484,
      "step": 4781
    },
    {
      "epoch": 0.26777914660096314,
      "grad_norm": 1.17559015750885,
      "learning_rate": 2.3895e-05,
      "loss": 0.4232,
      "step": 4782
    },
    {
      "epoch": 0.26783514391309216,
      "grad_norm": 1.4684512615203857,
      "learning_rate": 2.39e-05,
      "loss": 0.4547,
      "step": 4783
    },
    {
      "epoch": 0.2678911412252212,
      "grad_norm": 1.2338345050811768,
      "learning_rate": 2.3905000000000002e-05,
      "loss": 0.4311,
      "step": 4784
    },
    {
      "epoch": 0.2679471385373502,
      "grad_norm": 1.3646914958953857,
      "learning_rate": 2.3910000000000003e-05,
      "loss": 0.5086,
      "step": 4785
    },
    {
      "epoch": 0.2680031358494792,
      "grad_norm": 1.2001941204071045,
      "learning_rate": 2.3915e-05,
      "loss": 0.385,
      "step": 4786
    },
    {
      "epoch": 0.26805913316160823,
      "grad_norm": 1.1692379713058472,
      "learning_rate": 2.392e-05,
      "loss": 0.6629,
      "step": 4787
    },
    {
      "epoch": 0.26811513047373725,
      "grad_norm": 1.214898943901062,
      "learning_rate": 2.3925e-05,
      "loss": 0.415,
      "step": 4788
    },
    {
      "epoch": 0.26817112778586627,
      "grad_norm": 1.2813019752502441,
      "learning_rate": 2.3930000000000003e-05,
      "loss": 0.4171,
      "step": 4789
    },
    {
      "epoch": 0.2682271250979953,
      "grad_norm": 1.1769672632217407,
      "learning_rate": 2.3935e-05,
      "loss": 0.3184,
      "step": 4790
    },
    {
      "epoch": 0.2682831224101243,
      "grad_norm": 1.01447594165802,
      "learning_rate": 2.394e-05,
      "loss": 0.3295,
      "step": 4791
    },
    {
      "epoch": 0.2683391197222533,
      "grad_norm": 1.1792190074920654,
      "learning_rate": 2.3945000000000002e-05,
      "loss": 0.3586,
      "step": 4792
    },
    {
      "epoch": 0.26839511703438235,
      "grad_norm": 1.2017855644226074,
      "learning_rate": 2.395e-05,
      "loss": 0.3741,
      "step": 4793
    },
    {
      "epoch": 0.26845111434651137,
      "grad_norm": 1.1356403827667236,
      "learning_rate": 2.3955000000000004e-05,
      "loss": 0.5755,
      "step": 4794
    },
    {
      "epoch": 0.2685071116586404,
      "grad_norm": 1.1810275316238403,
      "learning_rate": 2.396e-05,
      "loss": 0.5044,
      "step": 4795
    },
    {
      "epoch": 0.2685631089707694,
      "grad_norm": 1.492426872253418,
      "learning_rate": 2.3965000000000002e-05,
      "loss": 0.6112,
      "step": 4796
    },
    {
      "epoch": 0.2686191062828984,
      "grad_norm": 1.587020993232727,
      "learning_rate": 2.397e-05,
      "loss": 0.4583,
      "step": 4797
    },
    {
      "epoch": 0.26867510359502744,
      "grad_norm": 1.438361644744873,
      "learning_rate": 2.3975e-05,
      "loss": 0.4967,
      "step": 4798
    },
    {
      "epoch": 0.26873110090715646,
      "grad_norm": 1.1004462242126465,
      "learning_rate": 2.398e-05,
      "loss": 0.5319,
      "step": 4799
    },
    {
      "epoch": 0.2687870982192855,
      "grad_norm": 1.253659963607788,
      "learning_rate": 2.3985e-05,
      "loss": 0.4341,
      "step": 4800
    },
    {
      "epoch": 0.2688430955314145,
      "grad_norm": 1.118528962135315,
      "learning_rate": 2.3990000000000002e-05,
      "loss": 0.3381,
      "step": 4801
    },
    {
      "epoch": 0.2688990928435435,
      "grad_norm": 1.5777441263198853,
      "learning_rate": 2.3995e-05,
      "loss": 0.3385,
      "step": 4802
    },
    {
      "epoch": 0.26895509015567254,
      "grad_norm": 1.1764894723892212,
      "learning_rate": 2.4e-05,
      "loss": 0.4373,
      "step": 4803
    },
    {
      "epoch": 0.26901108746780156,
      "grad_norm": 1.0975189208984375,
      "learning_rate": 2.4005e-05,
      "loss": 0.3497,
      "step": 4804
    },
    {
      "epoch": 0.2690670847799306,
      "grad_norm": 1.1311278343200684,
      "learning_rate": 2.4010000000000002e-05,
      "loss": 0.3729,
      "step": 4805
    },
    {
      "epoch": 0.2691230820920596,
      "grad_norm": 1.2189937829971313,
      "learning_rate": 2.4015000000000003e-05,
      "loss": 0.5284,
      "step": 4806
    },
    {
      "epoch": 0.2691790794041886,
      "grad_norm": 2.796288251876831,
      "learning_rate": 2.402e-05,
      "loss": 0.3669,
      "step": 4807
    },
    {
      "epoch": 0.26923507671631763,
      "grad_norm": 1.243399977684021,
      "learning_rate": 2.4025e-05,
      "loss": 0.4037,
      "step": 4808
    },
    {
      "epoch": 0.26929107402844665,
      "grad_norm": 1.410094976425171,
      "learning_rate": 2.4030000000000002e-05,
      "loss": 0.5426,
      "step": 4809
    },
    {
      "epoch": 0.26934707134057567,
      "grad_norm": 1.1329196691513062,
      "learning_rate": 2.4035000000000003e-05,
      "loss": 0.3994,
      "step": 4810
    },
    {
      "epoch": 0.2694030686527047,
      "grad_norm": 1.3636788129806519,
      "learning_rate": 2.404e-05,
      "loss": 0.4161,
      "step": 4811
    },
    {
      "epoch": 0.2694590659648337,
      "grad_norm": 1.0871763229370117,
      "learning_rate": 2.4045e-05,
      "loss": 0.3582,
      "step": 4812
    },
    {
      "epoch": 0.26951506327696273,
      "grad_norm": 1.1364787817001343,
      "learning_rate": 2.4050000000000002e-05,
      "loss": 0.4485,
      "step": 4813
    },
    {
      "epoch": 0.26957106058909175,
      "grad_norm": 1.3407330513000488,
      "learning_rate": 2.4055000000000003e-05,
      "loss": 0.4873,
      "step": 4814
    },
    {
      "epoch": 0.26962705790122077,
      "grad_norm": 1.1668158769607544,
      "learning_rate": 2.4060000000000003e-05,
      "loss": 0.6243,
      "step": 4815
    },
    {
      "epoch": 0.26968305521334973,
      "grad_norm": 1.3728643655776978,
      "learning_rate": 2.4065e-05,
      "loss": 0.4324,
      "step": 4816
    },
    {
      "epoch": 0.26973905252547875,
      "grad_norm": 1.2123281955718994,
      "learning_rate": 2.407e-05,
      "loss": 0.4673,
      "step": 4817
    },
    {
      "epoch": 0.26979504983760777,
      "grad_norm": 1.2647933959960938,
      "learning_rate": 2.4075e-05,
      "loss": 0.355,
      "step": 4818
    },
    {
      "epoch": 0.2698510471497368,
      "grad_norm": 1.1391644477844238,
      "learning_rate": 2.408e-05,
      "loss": 0.4842,
      "step": 4819
    },
    {
      "epoch": 0.2699070444618658,
      "grad_norm": 1.1071332693099976,
      "learning_rate": 2.4085e-05,
      "loss": 0.3956,
      "step": 4820
    },
    {
      "epoch": 0.2699630417739948,
      "grad_norm": 1.0850937366485596,
      "learning_rate": 2.409e-05,
      "loss": 0.4921,
      "step": 4821
    },
    {
      "epoch": 0.27001903908612385,
      "grad_norm": 1.157414197921753,
      "learning_rate": 2.4095000000000002e-05,
      "loss": 0.4969,
      "step": 4822
    },
    {
      "epoch": 0.27007503639825287,
      "grad_norm": 1.525781273841858,
      "learning_rate": 2.41e-05,
      "loss": 0.4886,
      "step": 4823
    },
    {
      "epoch": 0.2701310337103819,
      "grad_norm": 1.0770370960235596,
      "learning_rate": 2.4105e-05,
      "loss": 0.4066,
      "step": 4824
    },
    {
      "epoch": 0.2701870310225109,
      "grad_norm": 1.1031999588012695,
      "learning_rate": 2.411e-05,
      "loss": 0.3887,
      "step": 4825
    },
    {
      "epoch": 0.2702430283346399,
      "grad_norm": 1.0829763412475586,
      "learning_rate": 2.4115000000000002e-05,
      "loss": 0.3964,
      "step": 4826
    },
    {
      "epoch": 0.27029902564676894,
      "grad_norm": 1.165955901145935,
      "learning_rate": 2.412e-05,
      "loss": 0.3823,
      "step": 4827
    },
    {
      "epoch": 0.27035502295889796,
      "grad_norm": 0.955703854560852,
      "learning_rate": 2.4125e-05,
      "loss": 0.3086,
      "step": 4828
    },
    {
      "epoch": 0.270411020271027,
      "grad_norm": 1.2926037311553955,
      "learning_rate": 2.413e-05,
      "loss": 0.4862,
      "step": 4829
    },
    {
      "epoch": 0.270467017583156,
      "grad_norm": 1.4200623035430908,
      "learning_rate": 2.4135000000000002e-05,
      "loss": 0.591,
      "step": 4830
    },
    {
      "epoch": 0.270523014895285,
      "grad_norm": 1.1878716945648193,
      "learning_rate": 2.4140000000000003e-05,
      "loss": 0.3831,
      "step": 4831
    },
    {
      "epoch": 0.27057901220741404,
      "grad_norm": 1.1649898290634155,
      "learning_rate": 2.4145e-05,
      "loss": 0.3933,
      "step": 4832
    },
    {
      "epoch": 0.27063500951954306,
      "grad_norm": 1.396077275276184,
      "learning_rate": 2.415e-05,
      "loss": 0.3464,
      "step": 4833
    },
    {
      "epoch": 0.2706910068316721,
      "grad_norm": 1.1895679235458374,
      "learning_rate": 2.4154999999999998e-05,
      "loss": 0.4139,
      "step": 4834
    },
    {
      "epoch": 0.2707470041438011,
      "grad_norm": 1.3763668537139893,
      "learning_rate": 2.4160000000000002e-05,
      "loss": 0.451,
      "step": 4835
    },
    {
      "epoch": 0.2708030014559301,
      "grad_norm": 1.2594642639160156,
      "learning_rate": 2.4165e-05,
      "loss": 0.3893,
      "step": 4836
    },
    {
      "epoch": 0.27085899876805913,
      "grad_norm": 1.2071425914764404,
      "learning_rate": 2.417e-05,
      "loss": 0.4495,
      "step": 4837
    },
    {
      "epoch": 0.27091499608018815,
      "grad_norm": 1.3346881866455078,
      "learning_rate": 2.4175e-05,
      "loss": 0.5326,
      "step": 4838
    },
    {
      "epoch": 0.27097099339231717,
      "grad_norm": 1.264471173286438,
      "learning_rate": 2.418e-05,
      "loss": 0.5254,
      "step": 4839
    },
    {
      "epoch": 0.2710269907044462,
      "grad_norm": 1.154289960861206,
      "learning_rate": 2.4185000000000003e-05,
      "loss": 0.4498,
      "step": 4840
    },
    {
      "epoch": 0.2710829880165752,
      "grad_norm": 1.2578608989715576,
      "learning_rate": 2.419e-05,
      "loss": 0.3984,
      "step": 4841
    },
    {
      "epoch": 0.27113898532870423,
      "grad_norm": 1.193702220916748,
      "learning_rate": 2.4195e-05,
      "loss": 0.4636,
      "step": 4842
    },
    {
      "epoch": 0.27119498264083325,
      "grad_norm": 6.047183990478516,
      "learning_rate": 2.4200000000000002e-05,
      "loss": 0.4836,
      "step": 4843
    },
    {
      "epoch": 0.27125097995296227,
      "grad_norm": 1.906936764717102,
      "learning_rate": 2.4205e-05,
      "loss": 0.4673,
      "step": 4844
    },
    {
      "epoch": 0.2713069772650913,
      "grad_norm": 1.150715708732605,
      "learning_rate": 2.4210000000000004e-05,
      "loss": 0.5078,
      "step": 4845
    },
    {
      "epoch": 0.2713629745772203,
      "grad_norm": 1.1989617347717285,
      "learning_rate": 2.4215e-05,
      "loss": 0.479,
      "step": 4846
    },
    {
      "epoch": 0.2714189718893493,
      "grad_norm": 1.1006189584732056,
      "learning_rate": 2.4220000000000002e-05,
      "loss": 0.4624,
      "step": 4847
    },
    {
      "epoch": 0.27147496920147834,
      "grad_norm": 1.0710850954055786,
      "learning_rate": 2.4225e-05,
      "loss": 0.3288,
      "step": 4848
    },
    {
      "epoch": 0.27153096651360736,
      "grad_norm": 1.242953896522522,
      "learning_rate": 2.423e-05,
      "loss": 0.4696,
      "step": 4849
    },
    {
      "epoch": 0.2715869638257364,
      "grad_norm": 1.2073276042938232,
      "learning_rate": 2.4235e-05,
      "loss": 0.4357,
      "step": 4850
    },
    {
      "epoch": 0.2716429611378654,
      "grad_norm": 1.162598967552185,
      "learning_rate": 2.4240000000000002e-05,
      "loss": 0.4673,
      "step": 4851
    },
    {
      "epoch": 0.2716989584499944,
      "grad_norm": 1.40058171749115,
      "learning_rate": 2.4245000000000002e-05,
      "loss": 0.5045,
      "step": 4852
    },
    {
      "epoch": 0.27175495576212344,
      "grad_norm": 1.334394931793213,
      "learning_rate": 2.425e-05,
      "loss": 0.4338,
      "step": 4853
    },
    {
      "epoch": 0.27181095307425246,
      "grad_norm": 1.1808335781097412,
      "learning_rate": 2.4255e-05,
      "loss": 0.5973,
      "step": 4854
    },
    {
      "epoch": 0.2718669503863815,
      "grad_norm": 1.288703441619873,
      "learning_rate": 2.426e-05,
      "loss": 0.4282,
      "step": 4855
    },
    {
      "epoch": 0.2719229476985105,
      "grad_norm": 1.2336734533309937,
      "learning_rate": 2.4265000000000002e-05,
      "loss": 0.6002,
      "step": 4856
    },
    {
      "epoch": 0.2719789450106395,
      "grad_norm": 1.362411618232727,
      "learning_rate": 2.427e-05,
      "loss": 0.4054,
      "step": 4857
    },
    {
      "epoch": 0.2720349423227685,
      "grad_norm": 1.4240525960922241,
      "learning_rate": 2.4275e-05,
      "loss": 0.5253,
      "step": 4858
    },
    {
      "epoch": 0.2720909396348975,
      "grad_norm": 1.1944591999053955,
      "learning_rate": 2.428e-05,
      "loss": 0.481,
      "step": 4859
    },
    {
      "epoch": 0.2721469369470265,
      "grad_norm": 1.2200754880905151,
      "learning_rate": 2.4285000000000002e-05,
      "loss": 0.281,
      "step": 4860
    },
    {
      "epoch": 0.27220293425915554,
      "grad_norm": 1.2856251001358032,
      "learning_rate": 2.4290000000000003e-05,
      "loss": 0.3931,
      "step": 4861
    },
    {
      "epoch": 0.27225893157128456,
      "grad_norm": 1.1528245210647583,
      "learning_rate": 2.4295e-05,
      "loss": 0.4851,
      "step": 4862
    },
    {
      "epoch": 0.2723149288834136,
      "grad_norm": 1.2538505792617798,
      "learning_rate": 2.43e-05,
      "loss": 0.4604,
      "step": 4863
    },
    {
      "epoch": 0.2723709261955426,
      "grad_norm": 1.1997997760772705,
      "learning_rate": 2.4305e-05,
      "loss": 0.4507,
      "step": 4864
    },
    {
      "epoch": 0.2724269235076716,
      "grad_norm": 1.1665483713150024,
      "learning_rate": 2.4310000000000003e-05,
      "loss": 0.3933,
      "step": 4865
    },
    {
      "epoch": 0.27248292081980063,
      "grad_norm": 1.152035117149353,
      "learning_rate": 2.4315e-05,
      "loss": 0.5699,
      "step": 4866
    },
    {
      "epoch": 0.27253891813192965,
      "grad_norm": 1.3405358791351318,
      "learning_rate": 2.432e-05,
      "loss": 0.525,
      "step": 4867
    },
    {
      "epoch": 0.27259491544405867,
      "grad_norm": 1.1431670188903809,
      "learning_rate": 2.4325000000000002e-05,
      "loss": 0.3511,
      "step": 4868
    },
    {
      "epoch": 0.2726509127561877,
      "grad_norm": 1.0083286762237549,
      "learning_rate": 2.433e-05,
      "loss": 0.3615,
      "step": 4869
    },
    {
      "epoch": 0.2727069100683167,
      "grad_norm": 1.4361419677734375,
      "learning_rate": 2.4335000000000003e-05,
      "loss": 0.3746,
      "step": 4870
    },
    {
      "epoch": 0.27276290738044573,
      "grad_norm": 1.2798997163772583,
      "learning_rate": 2.434e-05,
      "loss": 0.5562,
      "step": 4871
    },
    {
      "epoch": 0.27281890469257475,
      "grad_norm": 1.5994699001312256,
      "learning_rate": 2.4345e-05,
      "loss": 0.4758,
      "step": 4872
    },
    {
      "epoch": 0.27287490200470377,
      "grad_norm": 1.4659181833267212,
      "learning_rate": 2.435e-05,
      "loss": 0.4897,
      "step": 4873
    },
    {
      "epoch": 0.2729308993168328,
      "grad_norm": 1.0324087142944336,
      "learning_rate": 2.4355e-05,
      "loss": 0.3914,
      "step": 4874
    },
    {
      "epoch": 0.2729868966289618,
      "grad_norm": 1.11372709274292,
      "learning_rate": 2.4360000000000004e-05,
      "loss": 0.4483,
      "step": 4875
    },
    {
      "epoch": 0.2730428939410908,
      "grad_norm": 1.0878872871398926,
      "learning_rate": 2.4365e-05,
      "loss": 0.4028,
      "step": 4876
    },
    {
      "epoch": 0.27309889125321984,
      "grad_norm": 1.1511904001235962,
      "learning_rate": 2.4370000000000002e-05,
      "loss": 0.2844,
      "step": 4877
    },
    {
      "epoch": 0.27315488856534886,
      "grad_norm": 1.0558775663375854,
      "learning_rate": 2.4375e-05,
      "loss": 0.4916,
      "step": 4878
    },
    {
      "epoch": 0.2732108858774779,
      "grad_norm": 1.0329701900482178,
      "learning_rate": 2.438e-05,
      "loss": 0.3883,
      "step": 4879
    },
    {
      "epoch": 0.2732668831896069,
      "grad_norm": 1.5741260051727295,
      "learning_rate": 2.4385e-05,
      "loss": 0.468,
      "step": 4880
    },
    {
      "epoch": 0.2733228805017359,
      "grad_norm": 1.1136219501495361,
      "learning_rate": 2.4390000000000002e-05,
      "loss": 0.3881,
      "step": 4881
    },
    {
      "epoch": 0.27337887781386494,
      "grad_norm": 1.2174041271209717,
      "learning_rate": 2.4395000000000003e-05,
      "loss": 0.4176,
      "step": 4882
    },
    {
      "epoch": 0.27343487512599396,
      "grad_norm": 1.137325644493103,
      "learning_rate": 2.44e-05,
      "loss": 0.4695,
      "step": 4883
    },
    {
      "epoch": 0.273490872438123,
      "grad_norm": 1.1314067840576172,
      "learning_rate": 2.4405e-05,
      "loss": 0.3367,
      "step": 4884
    },
    {
      "epoch": 0.273546869750252,
      "grad_norm": 1.1522043943405151,
      "learning_rate": 2.4410000000000002e-05,
      "loss": 0.4068,
      "step": 4885
    },
    {
      "epoch": 0.273602867062381,
      "grad_norm": 1.210898995399475,
      "learning_rate": 2.4415000000000003e-05,
      "loss": 0.5382,
      "step": 4886
    },
    {
      "epoch": 0.27365886437451004,
      "grad_norm": 1.389197826385498,
      "learning_rate": 2.442e-05,
      "loss": 0.5014,
      "step": 4887
    },
    {
      "epoch": 0.27371486168663905,
      "grad_norm": 1.416251540184021,
      "learning_rate": 2.4425e-05,
      "loss": 0.5832,
      "step": 4888
    },
    {
      "epoch": 0.2737708589987681,
      "grad_norm": 1.4145478010177612,
      "learning_rate": 2.443e-05,
      "loss": 0.5117,
      "step": 4889
    },
    {
      "epoch": 0.2738268563108971,
      "grad_norm": 1.226279854774475,
      "learning_rate": 2.4435000000000002e-05,
      "loss": 0.4941,
      "step": 4890
    },
    {
      "epoch": 0.2738828536230261,
      "grad_norm": 0.9975038766860962,
      "learning_rate": 2.4440000000000003e-05,
      "loss": 0.3055,
      "step": 4891
    },
    {
      "epoch": 0.27393885093515513,
      "grad_norm": 1.38507080078125,
      "learning_rate": 2.4445e-05,
      "loss": 0.4014,
      "step": 4892
    },
    {
      "epoch": 0.27399484824728415,
      "grad_norm": 1.3080692291259766,
      "learning_rate": 2.445e-05,
      "loss": 0.502,
      "step": 4893
    },
    {
      "epoch": 0.27405084555941317,
      "grad_norm": 1.0307468175888062,
      "learning_rate": 2.4455e-05,
      "loss": 0.3938,
      "step": 4894
    },
    {
      "epoch": 0.2741068428715422,
      "grad_norm": 1.2235243320465088,
      "learning_rate": 2.4460000000000003e-05,
      "loss": 0.3463,
      "step": 4895
    },
    {
      "epoch": 0.2741628401836712,
      "grad_norm": 1.427044153213501,
      "learning_rate": 2.4465e-05,
      "loss": 0.5066,
      "step": 4896
    },
    {
      "epoch": 0.2742188374958002,
      "grad_norm": 1.1579972505569458,
      "learning_rate": 2.447e-05,
      "loss": 0.4708,
      "step": 4897
    },
    {
      "epoch": 0.27427483480792925,
      "grad_norm": 1.1244254112243652,
      "learning_rate": 2.4475000000000002e-05,
      "loss": 0.3776,
      "step": 4898
    },
    {
      "epoch": 0.2743308321200582,
      "grad_norm": 1.1676359176635742,
      "learning_rate": 2.448e-05,
      "loss": 0.4447,
      "step": 4899
    },
    {
      "epoch": 0.27438682943218723,
      "grad_norm": 1.186279058456421,
      "learning_rate": 2.4485000000000004e-05,
      "loss": 0.5561,
      "step": 4900
    },
    {
      "epoch": 0.27444282674431625,
      "grad_norm": 1.1873775720596313,
      "learning_rate": 2.449e-05,
      "loss": 0.4251,
      "step": 4901
    },
    {
      "epoch": 0.27449882405644527,
      "grad_norm": 1.4521654844284058,
      "learning_rate": 2.4495000000000002e-05,
      "loss": 0.5958,
      "step": 4902
    },
    {
      "epoch": 0.2745548213685743,
      "grad_norm": 1.1759424209594727,
      "learning_rate": 2.45e-05,
      "loss": 0.3257,
      "step": 4903
    },
    {
      "epoch": 0.2746108186807033,
      "grad_norm": 1.4622273445129395,
      "learning_rate": 2.4505e-05,
      "loss": 0.4231,
      "step": 4904
    },
    {
      "epoch": 0.2746668159928323,
      "grad_norm": 1.1844336986541748,
      "learning_rate": 2.451e-05,
      "loss": 0.465,
      "step": 4905
    },
    {
      "epoch": 0.27472281330496134,
      "grad_norm": 1.2939742803573608,
      "learning_rate": 2.4515e-05,
      "loss": 0.4654,
      "step": 4906
    },
    {
      "epoch": 0.27477881061709036,
      "grad_norm": 1.2372941970825195,
      "learning_rate": 2.4520000000000002e-05,
      "loss": 0.4132,
      "step": 4907
    },
    {
      "epoch": 0.2748348079292194,
      "grad_norm": 1.1188563108444214,
      "learning_rate": 2.4525e-05,
      "loss": 0.37,
      "step": 4908
    },
    {
      "epoch": 0.2748908052413484,
      "grad_norm": 1.455541729927063,
      "learning_rate": 2.453e-05,
      "loss": 0.4644,
      "step": 4909
    },
    {
      "epoch": 0.2749468025534774,
      "grad_norm": 1.6214908361434937,
      "learning_rate": 2.4535e-05,
      "loss": 0.4694,
      "step": 4910
    },
    {
      "epoch": 0.27500279986560644,
      "grad_norm": 1.17264986038208,
      "learning_rate": 2.4540000000000002e-05,
      "loss": 0.3957,
      "step": 4911
    },
    {
      "epoch": 0.27505879717773546,
      "grad_norm": 1.0448565483093262,
      "learning_rate": 2.4545000000000003e-05,
      "loss": 0.3585,
      "step": 4912
    },
    {
      "epoch": 0.2751147944898645,
      "grad_norm": 1.3260798454284668,
      "learning_rate": 2.455e-05,
      "loss": 0.4594,
      "step": 4913
    },
    {
      "epoch": 0.2751707918019935,
      "grad_norm": 1.7412939071655273,
      "learning_rate": 2.4555e-05,
      "loss": 0.4749,
      "step": 4914
    },
    {
      "epoch": 0.2752267891141225,
      "grad_norm": 1.4049450159072876,
      "learning_rate": 2.4560000000000002e-05,
      "loss": 0.4712,
      "step": 4915
    },
    {
      "epoch": 0.27528278642625154,
      "grad_norm": 1.2056244611740112,
      "learning_rate": 2.4565000000000003e-05,
      "loss": 0.4464,
      "step": 4916
    },
    {
      "epoch": 0.27533878373838055,
      "grad_norm": 1.6052457094192505,
      "learning_rate": 2.457e-05,
      "loss": 0.4002,
      "step": 4917
    },
    {
      "epoch": 0.2753947810505096,
      "grad_norm": 1.1523957252502441,
      "learning_rate": 2.4575e-05,
      "loss": 0.425,
      "step": 4918
    },
    {
      "epoch": 0.2754507783626386,
      "grad_norm": 1.3647980690002441,
      "learning_rate": 2.4580000000000002e-05,
      "loss": 0.536,
      "step": 4919
    },
    {
      "epoch": 0.2755067756747676,
      "grad_norm": 1.340335726737976,
      "learning_rate": 2.4585000000000003e-05,
      "loss": 0.4708,
      "step": 4920
    },
    {
      "epoch": 0.27556277298689663,
      "grad_norm": 1.038291096687317,
      "learning_rate": 2.4590000000000003e-05,
      "loss": 0.3694,
      "step": 4921
    },
    {
      "epoch": 0.27561877029902565,
      "grad_norm": 1.2275313138961792,
      "learning_rate": 2.4595e-05,
      "loss": 0.5274,
      "step": 4922
    },
    {
      "epoch": 0.27567476761115467,
      "grad_norm": 1.0066707134246826,
      "learning_rate": 2.46e-05,
      "loss": 0.332,
      "step": 4923
    },
    {
      "epoch": 0.2757307649232837,
      "grad_norm": 1.0508103370666504,
      "learning_rate": 2.4605e-05,
      "loss": 0.4163,
      "step": 4924
    },
    {
      "epoch": 0.2757867622354127,
      "grad_norm": 1.1964588165283203,
      "learning_rate": 2.4610000000000003e-05,
      "loss": 0.419,
      "step": 4925
    },
    {
      "epoch": 0.2758427595475417,
      "grad_norm": 1.2009811401367188,
      "learning_rate": 2.4615e-05,
      "loss": 0.4048,
      "step": 4926
    },
    {
      "epoch": 0.27589875685967075,
      "grad_norm": 1.2661802768707275,
      "learning_rate": 2.462e-05,
      "loss": 0.3793,
      "step": 4927
    },
    {
      "epoch": 0.27595475417179977,
      "grad_norm": 1.1443170309066772,
      "learning_rate": 2.4625000000000002e-05,
      "loss": 0.375,
      "step": 4928
    },
    {
      "epoch": 0.2760107514839288,
      "grad_norm": 1.1995105743408203,
      "learning_rate": 2.463e-05,
      "loss": 0.3884,
      "step": 4929
    },
    {
      "epoch": 0.2760667487960578,
      "grad_norm": 1.284255862236023,
      "learning_rate": 2.4635000000000004e-05,
      "loss": 0.3677,
      "step": 4930
    },
    {
      "epoch": 0.2761227461081868,
      "grad_norm": 1.3781988620758057,
      "learning_rate": 2.464e-05,
      "loss": 0.459,
      "step": 4931
    },
    {
      "epoch": 0.27617874342031584,
      "grad_norm": 1.1450660228729248,
      "learning_rate": 2.4645000000000002e-05,
      "loss": 0.3256,
      "step": 4932
    },
    {
      "epoch": 0.27623474073244486,
      "grad_norm": 1.0806472301483154,
      "learning_rate": 2.465e-05,
      "loss": 0.4452,
      "step": 4933
    },
    {
      "epoch": 0.2762907380445739,
      "grad_norm": 1.2479525804519653,
      "learning_rate": 2.4655e-05,
      "loss": 0.3918,
      "step": 4934
    },
    {
      "epoch": 0.2763467353567029,
      "grad_norm": 1.154831051826477,
      "learning_rate": 2.466e-05,
      "loss": 0.3139,
      "step": 4935
    },
    {
      "epoch": 0.2764027326688319,
      "grad_norm": 1.1486061811447144,
      "learning_rate": 2.4665000000000002e-05,
      "loss": 0.3987,
      "step": 4936
    },
    {
      "epoch": 0.27645872998096094,
      "grad_norm": 1.382432222366333,
      "learning_rate": 2.4670000000000003e-05,
      "loss": 0.3491,
      "step": 4937
    },
    {
      "epoch": 0.27651472729308996,
      "grad_norm": 1.035082221031189,
      "learning_rate": 2.4675e-05,
      "loss": 0.3382,
      "step": 4938
    },
    {
      "epoch": 0.276570724605219,
      "grad_norm": 1.1332824230194092,
      "learning_rate": 2.468e-05,
      "loss": 0.3603,
      "step": 4939
    },
    {
      "epoch": 0.27662672191734794,
      "grad_norm": 1.4208630323410034,
      "learning_rate": 2.4685e-05,
      "loss": 0.4386,
      "step": 4940
    },
    {
      "epoch": 0.27668271922947696,
      "grad_norm": 1.26140296459198,
      "learning_rate": 2.4690000000000002e-05,
      "loss": 0.4126,
      "step": 4941
    },
    {
      "epoch": 0.276738716541606,
      "grad_norm": 1.047723412513733,
      "learning_rate": 2.4695e-05,
      "loss": 0.3948,
      "step": 4942
    },
    {
      "epoch": 0.276794713853735,
      "grad_norm": 1.1719717979431152,
      "learning_rate": 2.47e-05,
      "loss": 0.5086,
      "step": 4943
    },
    {
      "epoch": 0.276850711165864,
      "grad_norm": 1.5609759092330933,
      "learning_rate": 2.4705e-05,
      "loss": 0.6026,
      "step": 4944
    },
    {
      "epoch": 0.27690670847799304,
      "grad_norm": 1.3194032907485962,
      "learning_rate": 2.471e-05,
      "loss": 0.3067,
      "step": 4945
    },
    {
      "epoch": 0.27696270579012205,
      "grad_norm": 1.4805982112884521,
      "learning_rate": 2.4715000000000003e-05,
      "loss": 0.5675,
      "step": 4946
    },
    {
      "epoch": 0.2770187031022511,
      "grad_norm": 1.563145399093628,
      "learning_rate": 2.472e-05,
      "loss": 0.3775,
      "step": 4947
    },
    {
      "epoch": 0.2770747004143801,
      "grad_norm": 1.1520642042160034,
      "learning_rate": 2.4725e-05,
      "loss": 0.4203,
      "step": 4948
    },
    {
      "epoch": 0.2771306977265091,
      "grad_norm": 1.2293246984481812,
      "learning_rate": 2.473e-05,
      "loss": 0.4264,
      "step": 4949
    },
    {
      "epoch": 0.27718669503863813,
      "grad_norm": 1.1341722011566162,
      "learning_rate": 2.4735e-05,
      "loss": 0.3552,
      "step": 4950
    },
    {
      "epoch": 0.27724269235076715,
      "grad_norm": 1.4974896907806396,
      "learning_rate": 2.4740000000000004e-05,
      "loss": 0.4868,
      "step": 4951
    },
    {
      "epoch": 0.27729868966289617,
      "grad_norm": 1.3844859600067139,
      "learning_rate": 2.4745e-05,
      "loss": 0.4703,
      "step": 4952
    },
    {
      "epoch": 0.2773546869750252,
      "grad_norm": 1.133491039276123,
      "learning_rate": 2.4750000000000002e-05,
      "loss": 0.3953,
      "step": 4953
    },
    {
      "epoch": 0.2774106842871542,
      "grad_norm": 1.1602778434753418,
      "learning_rate": 2.4755e-05,
      "loss": 0.4219,
      "step": 4954
    },
    {
      "epoch": 0.2774666815992832,
      "grad_norm": 1.3232240676879883,
      "learning_rate": 2.476e-05,
      "loss": 0.3785,
      "step": 4955
    },
    {
      "epoch": 0.27752267891141225,
      "grad_norm": 1.514744520187378,
      "learning_rate": 2.4765e-05,
      "loss": 0.4758,
      "step": 4956
    },
    {
      "epoch": 0.27757867622354127,
      "grad_norm": 1.1780812740325928,
      "learning_rate": 2.4770000000000002e-05,
      "loss": 0.3897,
      "step": 4957
    },
    {
      "epoch": 0.2776346735356703,
      "grad_norm": 1.2148231267929077,
      "learning_rate": 2.4775000000000003e-05,
      "loss": 0.3999,
      "step": 4958
    },
    {
      "epoch": 0.2776906708477993,
      "grad_norm": 1.3397095203399658,
      "learning_rate": 2.478e-05,
      "loss": 0.3621,
      "step": 4959
    },
    {
      "epoch": 0.2777466681599283,
      "grad_norm": 1.153542160987854,
      "learning_rate": 2.4785e-05,
      "loss": 0.3358,
      "step": 4960
    },
    {
      "epoch": 0.27780266547205734,
      "grad_norm": 1.0039702653884888,
      "learning_rate": 2.479e-05,
      "loss": 0.3771,
      "step": 4961
    },
    {
      "epoch": 0.27785866278418636,
      "grad_norm": 1.048505425453186,
      "learning_rate": 2.4795000000000002e-05,
      "loss": 0.4062,
      "step": 4962
    },
    {
      "epoch": 0.2779146600963154,
      "grad_norm": 1.4633502960205078,
      "learning_rate": 2.48e-05,
      "loss": 0.4179,
      "step": 4963
    },
    {
      "epoch": 0.2779706574084444,
      "grad_norm": 1.3561614751815796,
      "learning_rate": 2.4805e-05,
      "loss": 0.395,
      "step": 4964
    },
    {
      "epoch": 0.2780266547205734,
      "grad_norm": 1.1100013256072998,
      "learning_rate": 2.481e-05,
      "loss": 0.4721,
      "step": 4965
    },
    {
      "epoch": 0.27808265203270244,
      "grad_norm": 1.2962788343429565,
      "learning_rate": 2.4815000000000002e-05,
      "loss": 0.5376,
      "step": 4966
    },
    {
      "epoch": 0.27813864934483146,
      "grad_norm": 1.1644928455352783,
      "learning_rate": 2.4820000000000003e-05,
      "loss": 0.4759,
      "step": 4967
    },
    {
      "epoch": 0.2781946466569605,
      "grad_norm": 1.102250337600708,
      "learning_rate": 2.4825e-05,
      "loss": 0.358,
      "step": 4968
    },
    {
      "epoch": 0.2782506439690895,
      "grad_norm": 0.9479377269744873,
      "learning_rate": 2.483e-05,
      "loss": 0.3354,
      "step": 4969
    },
    {
      "epoch": 0.2783066412812185,
      "grad_norm": 1.2795408964157104,
      "learning_rate": 2.4835e-05,
      "loss": 0.3787,
      "step": 4970
    },
    {
      "epoch": 0.27836263859334753,
      "grad_norm": 1.070845365524292,
      "learning_rate": 2.4840000000000003e-05,
      "loss": 0.4005,
      "step": 4971
    },
    {
      "epoch": 0.27841863590547655,
      "grad_norm": 1.3072528839111328,
      "learning_rate": 2.4845e-05,
      "loss": 0.5137,
      "step": 4972
    },
    {
      "epoch": 0.27847463321760557,
      "grad_norm": 1.3732484579086304,
      "learning_rate": 2.485e-05,
      "loss": 0.5127,
      "step": 4973
    },
    {
      "epoch": 0.2785306305297346,
      "grad_norm": 1.2259336709976196,
      "learning_rate": 2.4855000000000002e-05,
      "loss": 0.4656,
      "step": 4974
    },
    {
      "epoch": 0.2785866278418636,
      "grad_norm": 1.290574073791504,
      "learning_rate": 2.486e-05,
      "loss": 0.6261,
      "step": 4975
    },
    {
      "epoch": 0.27864262515399263,
      "grad_norm": 1.3530707359313965,
      "learning_rate": 2.4865000000000003e-05,
      "loss": 0.649,
      "step": 4976
    },
    {
      "epoch": 0.27869862246612165,
      "grad_norm": 1.1110451221466064,
      "learning_rate": 2.487e-05,
      "loss": 0.4133,
      "step": 4977
    },
    {
      "epoch": 0.27875461977825067,
      "grad_norm": 1.2195521593093872,
      "learning_rate": 2.4875e-05,
      "loss": 0.4227,
      "step": 4978
    },
    {
      "epoch": 0.2788106170903797,
      "grad_norm": 1.1598562002182007,
      "learning_rate": 2.488e-05,
      "loss": 0.4384,
      "step": 4979
    },
    {
      "epoch": 0.2788666144025087,
      "grad_norm": 1.4165732860565186,
      "learning_rate": 2.4885e-05,
      "loss": 0.4252,
      "step": 4980
    },
    {
      "epoch": 0.2789226117146377,
      "grad_norm": 1.760197639465332,
      "learning_rate": 2.489e-05,
      "loss": 0.3759,
      "step": 4981
    },
    {
      "epoch": 0.2789786090267667,
      "grad_norm": 1.1311376094818115,
      "learning_rate": 2.4895e-05,
      "loss": 0.37,
      "step": 4982
    },
    {
      "epoch": 0.2790346063388957,
      "grad_norm": 1.2808183431625366,
      "learning_rate": 2.4900000000000002e-05,
      "loss": 0.4915,
      "step": 4983
    },
    {
      "epoch": 0.2790906036510247,
      "grad_norm": 1.4143792390823364,
      "learning_rate": 2.4905e-05,
      "loss": 0.5387,
      "step": 4984
    },
    {
      "epoch": 0.27914660096315375,
      "grad_norm": 1.1529659032821655,
      "learning_rate": 2.491e-05,
      "loss": 0.4424,
      "step": 4985
    },
    {
      "epoch": 0.27920259827528277,
      "grad_norm": 1.3599112033843994,
      "learning_rate": 2.4915e-05,
      "loss": 0.3906,
      "step": 4986
    },
    {
      "epoch": 0.2792585955874118,
      "grad_norm": 1.1522306203842163,
      "learning_rate": 2.4920000000000002e-05,
      "loss": 0.448,
      "step": 4987
    },
    {
      "epoch": 0.2793145928995408,
      "grad_norm": 1.0313642024993896,
      "learning_rate": 2.4925000000000003e-05,
      "loss": 0.4001,
      "step": 4988
    },
    {
      "epoch": 0.2793705902116698,
      "grad_norm": 1.1960699558258057,
      "learning_rate": 2.493e-05,
      "loss": 0.3986,
      "step": 4989
    },
    {
      "epoch": 0.27942658752379884,
      "grad_norm": 1.370259404182434,
      "learning_rate": 2.4935e-05,
      "loss": 0.4686,
      "step": 4990
    },
    {
      "epoch": 0.27948258483592786,
      "grad_norm": 1.4210662841796875,
      "learning_rate": 2.4940000000000002e-05,
      "loss": 0.4857,
      "step": 4991
    },
    {
      "epoch": 0.2795385821480569,
      "grad_norm": 1.4025802612304688,
      "learning_rate": 2.4945000000000003e-05,
      "loss": 0.4234,
      "step": 4992
    },
    {
      "epoch": 0.2795945794601859,
      "grad_norm": 2.0384538173675537,
      "learning_rate": 2.495e-05,
      "loss": 0.4057,
      "step": 4993
    },
    {
      "epoch": 0.2796505767723149,
      "grad_norm": 1.112663745880127,
      "learning_rate": 2.4955e-05,
      "loss": 0.3578,
      "step": 4994
    },
    {
      "epoch": 0.27970657408444394,
      "grad_norm": 1.6824556589126587,
      "learning_rate": 2.496e-05,
      "loss": 0.5093,
      "step": 4995
    },
    {
      "epoch": 0.27976257139657296,
      "grad_norm": 1.216307282447815,
      "learning_rate": 2.4965000000000002e-05,
      "loss": 0.3959,
      "step": 4996
    },
    {
      "epoch": 0.279818568708702,
      "grad_norm": 1.1118022203445435,
      "learning_rate": 2.4970000000000003e-05,
      "loss": 0.4064,
      "step": 4997
    },
    {
      "epoch": 0.279874566020831,
      "grad_norm": 1.4133844375610352,
      "learning_rate": 2.4975e-05,
      "loss": 0.4684,
      "step": 4998
    },
    {
      "epoch": 0.27993056333296,
      "grad_norm": 1.2977321147918701,
      "learning_rate": 2.498e-05,
      "loss": 0.4066,
      "step": 4999
    },
    {
      "epoch": 0.27998656064508903,
      "grad_norm": 1.1011120080947876,
      "learning_rate": 2.4985e-05,
      "loss": 0.3026,
      "step": 5000
    },
    {
      "epoch": 0.28004255795721805,
      "grad_norm": 1.4148776531219482,
      "learning_rate": 2.4990000000000003e-05,
      "loss": 0.5396,
      "step": 5001
    },
    {
      "epoch": 0.28009855526934707,
      "grad_norm": 1.0133522748947144,
      "learning_rate": 2.4995e-05,
      "loss": 0.3985,
      "step": 5002
    },
    {
      "epoch": 0.2801545525814761,
      "grad_norm": 1.296737790107727,
      "learning_rate": 2.5e-05,
      "loss": 0.4786,
      "step": 5003
    },
    {
      "epoch": 0.2802105498936051,
      "grad_norm": 1.4818477630615234,
      "learning_rate": 2.5005000000000002e-05,
      "loss": 0.5541,
      "step": 5004
    },
    {
      "epoch": 0.28026654720573413,
      "grad_norm": 1.2512661218643188,
      "learning_rate": 2.501e-05,
      "loss": 0.4498,
      "step": 5005
    },
    {
      "epoch": 0.28032254451786315,
      "grad_norm": 1.0812656879425049,
      "learning_rate": 2.5015e-05,
      "loss": 0.3263,
      "step": 5006
    },
    {
      "epoch": 0.28037854182999217,
      "grad_norm": 1.1355310678482056,
      "learning_rate": 2.5019999999999998e-05,
      "loss": 0.4299,
      "step": 5007
    },
    {
      "epoch": 0.2804345391421212,
      "grad_norm": 1.1510392427444458,
      "learning_rate": 2.5025e-05,
      "loss": 0.4134,
      "step": 5008
    },
    {
      "epoch": 0.2804905364542502,
      "grad_norm": 1.1566981077194214,
      "learning_rate": 2.5030000000000003e-05,
      "loss": 0.3365,
      "step": 5009
    },
    {
      "epoch": 0.2805465337663792,
      "grad_norm": 1.1639187335968018,
      "learning_rate": 2.5035000000000003e-05,
      "loss": 0.4559,
      "step": 5010
    },
    {
      "epoch": 0.28060253107850824,
      "grad_norm": 1.1393630504608154,
      "learning_rate": 2.504e-05,
      "loss": 0.3888,
      "step": 5011
    },
    {
      "epoch": 0.28065852839063726,
      "grad_norm": 1.2431306838989258,
      "learning_rate": 2.5045e-05,
      "loss": 0.5282,
      "step": 5012
    },
    {
      "epoch": 0.2807145257027663,
      "grad_norm": 1.3750158548355103,
      "learning_rate": 2.5050000000000002e-05,
      "loss": 0.6099,
      "step": 5013
    },
    {
      "epoch": 0.2807705230148953,
      "grad_norm": 1.3600974082946777,
      "learning_rate": 2.5055e-05,
      "loss": 0.3914,
      "step": 5014
    },
    {
      "epoch": 0.2808265203270243,
      "grad_norm": 1.1310267448425293,
      "learning_rate": 2.506e-05,
      "loss": 0.3774,
      "step": 5015
    },
    {
      "epoch": 0.28088251763915334,
      "grad_norm": 1.2618962526321411,
      "learning_rate": 2.5064999999999998e-05,
      "loss": 0.4592,
      "step": 5016
    },
    {
      "epoch": 0.28093851495128236,
      "grad_norm": 1.211580514907837,
      "learning_rate": 2.507e-05,
      "loss": 0.4236,
      "step": 5017
    },
    {
      "epoch": 0.2809945122634114,
      "grad_norm": 1.16752028465271,
      "learning_rate": 2.5075e-05,
      "loss": 0.3997,
      "step": 5018
    },
    {
      "epoch": 0.2810505095755404,
      "grad_norm": 1.2811846733093262,
      "learning_rate": 2.5080000000000004e-05,
      "loss": 0.4685,
      "step": 5019
    },
    {
      "epoch": 0.2811065068876694,
      "grad_norm": 1.071851372718811,
      "learning_rate": 2.5085000000000005e-05,
      "loss": 0.3922,
      "step": 5020
    },
    {
      "epoch": 0.28116250419979844,
      "grad_norm": 1.2484806776046753,
      "learning_rate": 2.5090000000000002e-05,
      "loss": 0.4412,
      "step": 5021
    },
    {
      "epoch": 0.28121850151192745,
      "grad_norm": 1.1056876182556152,
      "learning_rate": 2.5095000000000003e-05,
      "loss": 0.4058,
      "step": 5022
    },
    {
      "epoch": 0.2812744988240564,
      "grad_norm": 1.1852781772613525,
      "learning_rate": 2.51e-05,
      "loss": 0.4035,
      "step": 5023
    },
    {
      "epoch": 0.28133049613618544,
      "grad_norm": 1.5772156715393066,
      "learning_rate": 2.5105e-05,
      "loss": 0.4305,
      "step": 5024
    },
    {
      "epoch": 0.28138649344831446,
      "grad_norm": 1.1575450897216797,
      "learning_rate": 2.5110000000000002e-05,
      "loss": 0.4063,
      "step": 5025
    },
    {
      "epoch": 0.2814424907604435,
      "grad_norm": 1.2748372554779053,
      "learning_rate": 2.5115e-05,
      "loss": 0.4925,
      "step": 5026
    },
    {
      "epoch": 0.2814984880725725,
      "grad_norm": 1.3930341005325317,
      "learning_rate": 2.512e-05,
      "loss": 0.441,
      "step": 5027
    },
    {
      "epoch": 0.2815544853847015,
      "grad_norm": 1.5599757432937622,
      "learning_rate": 2.5124999999999997e-05,
      "loss": 0.4644,
      "step": 5028
    },
    {
      "epoch": 0.28161048269683053,
      "grad_norm": 1.35392427444458,
      "learning_rate": 2.5130000000000005e-05,
      "loss": 0.5732,
      "step": 5029
    },
    {
      "epoch": 0.28166648000895955,
      "grad_norm": 2.7221648693084717,
      "learning_rate": 2.5135000000000002e-05,
      "loss": 0.3834,
      "step": 5030
    },
    {
      "epoch": 0.28172247732108857,
      "grad_norm": 1.168181300163269,
      "learning_rate": 2.5140000000000003e-05,
      "loss": 0.4182,
      "step": 5031
    },
    {
      "epoch": 0.2817784746332176,
      "grad_norm": 1.1578096151351929,
      "learning_rate": 2.5145e-05,
      "loss": 0.3981,
      "step": 5032
    },
    {
      "epoch": 0.2818344719453466,
      "grad_norm": 1.066641926765442,
      "learning_rate": 2.515e-05,
      "loss": 0.4481,
      "step": 5033
    },
    {
      "epoch": 0.28189046925747563,
      "grad_norm": 1.20148766040802,
      "learning_rate": 2.5155000000000002e-05,
      "loss": 0.4609,
      "step": 5034
    },
    {
      "epoch": 0.28194646656960465,
      "grad_norm": 1.1593502759933472,
      "learning_rate": 2.516e-05,
      "loss": 0.5019,
      "step": 5035
    },
    {
      "epoch": 0.28200246388173367,
      "grad_norm": 1.2151565551757812,
      "learning_rate": 2.5165e-05,
      "loss": 0.4255,
      "step": 5036
    },
    {
      "epoch": 0.2820584611938627,
      "grad_norm": 1.204663872718811,
      "learning_rate": 2.5169999999999998e-05,
      "loss": 0.4624,
      "step": 5037
    },
    {
      "epoch": 0.2821144585059917,
      "grad_norm": 1.0738803148269653,
      "learning_rate": 2.5175e-05,
      "loss": 0.4004,
      "step": 5038
    },
    {
      "epoch": 0.2821704558181207,
      "grad_norm": 1.1140012741088867,
      "learning_rate": 2.5180000000000003e-05,
      "loss": 0.3801,
      "step": 5039
    },
    {
      "epoch": 0.28222645313024974,
      "grad_norm": 1.0218623876571655,
      "learning_rate": 2.5185000000000004e-05,
      "loss": 0.3478,
      "step": 5040
    },
    {
      "epoch": 0.28228245044237876,
      "grad_norm": 1.285437822341919,
      "learning_rate": 2.519e-05,
      "loss": 0.4935,
      "step": 5041
    },
    {
      "epoch": 0.2823384477545078,
      "grad_norm": 1.3076173067092896,
      "learning_rate": 2.5195000000000002e-05,
      "loss": 0.3867,
      "step": 5042
    },
    {
      "epoch": 0.2823944450666368,
      "grad_norm": 1.273854374885559,
      "learning_rate": 2.5200000000000003e-05,
      "loss": 0.4336,
      "step": 5043
    },
    {
      "epoch": 0.2824504423787658,
      "grad_norm": 1.2694003582000732,
      "learning_rate": 2.5205e-05,
      "loss": 0.4235,
      "step": 5044
    },
    {
      "epoch": 0.28250643969089484,
      "grad_norm": 1.1363548040390015,
      "learning_rate": 2.521e-05,
      "loss": 0.3504,
      "step": 5045
    },
    {
      "epoch": 0.28256243700302386,
      "grad_norm": 1.2498666048049927,
      "learning_rate": 2.5214999999999998e-05,
      "loss": 0.4245,
      "step": 5046
    },
    {
      "epoch": 0.2826184343151529,
      "grad_norm": 1.1381645202636719,
      "learning_rate": 2.522e-05,
      "loss": 0.3956,
      "step": 5047
    },
    {
      "epoch": 0.2826744316272819,
      "grad_norm": 1.3356531858444214,
      "learning_rate": 2.5225e-05,
      "loss": 0.4546,
      "step": 5048
    },
    {
      "epoch": 0.2827304289394109,
      "grad_norm": 1.0721495151519775,
      "learning_rate": 2.5230000000000004e-05,
      "loss": 0.5557,
      "step": 5049
    },
    {
      "epoch": 0.28278642625153994,
      "grad_norm": 1.2902302742004395,
      "learning_rate": 2.5235e-05,
      "loss": 0.5123,
      "step": 5050
    },
    {
      "epoch": 0.28284242356366895,
      "grad_norm": 1.2670661211013794,
      "learning_rate": 2.5240000000000002e-05,
      "loss": 0.4056,
      "step": 5051
    },
    {
      "epoch": 0.282898420875798,
      "grad_norm": 1.1348469257354736,
      "learning_rate": 2.5245000000000003e-05,
      "loss": 0.4311,
      "step": 5052
    },
    {
      "epoch": 0.282954418187927,
      "grad_norm": 1.2487995624542236,
      "learning_rate": 2.525e-05,
      "loss": 0.4668,
      "step": 5053
    },
    {
      "epoch": 0.283010415500056,
      "grad_norm": 1.5302445888519287,
      "learning_rate": 2.5255e-05,
      "loss": 0.4978,
      "step": 5054
    },
    {
      "epoch": 0.28306641281218503,
      "grad_norm": 1.219663143157959,
      "learning_rate": 2.526e-05,
      "loss": 0.3641,
      "step": 5055
    },
    {
      "epoch": 0.28312241012431405,
      "grad_norm": 1.0319263935089111,
      "learning_rate": 2.5265e-05,
      "loss": 0.3338,
      "step": 5056
    },
    {
      "epoch": 0.28317840743644307,
      "grad_norm": 1.1401755809783936,
      "learning_rate": 2.527e-05,
      "loss": 0.3796,
      "step": 5057
    },
    {
      "epoch": 0.2832344047485721,
      "grad_norm": 1.1127512454986572,
      "learning_rate": 2.5274999999999998e-05,
      "loss": 0.5296,
      "step": 5058
    },
    {
      "epoch": 0.2832904020607011,
      "grad_norm": 1.3074079751968384,
      "learning_rate": 2.5280000000000005e-05,
      "loss": 0.3188,
      "step": 5059
    },
    {
      "epoch": 0.2833463993728301,
      "grad_norm": 1.3833627700805664,
      "learning_rate": 2.5285000000000003e-05,
      "loss": 0.5894,
      "step": 5060
    },
    {
      "epoch": 0.28340239668495915,
      "grad_norm": 1.3308225870132446,
      "learning_rate": 2.5290000000000004e-05,
      "loss": 0.4272,
      "step": 5061
    },
    {
      "epoch": 0.28345839399708816,
      "grad_norm": 1.2617563009262085,
      "learning_rate": 2.5295e-05,
      "loss": 0.3797,
      "step": 5062
    },
    {
      "epoch": 0.2835143913092172,
      "grad_norm": 1.2066013813018799,
      "learning_rate": 2.5300000000000002e-05,
      "loss": 0.4103,
      "step": 5063
    },
    {
      "epoch": 0.28357038862134615,
      "grad_norm": 1.3341902494430542,
      "learning_rate": 2.5305000000000003e-05,
      "loss": 0.3874,
      "step": 5064
    },
    {
      "epoch": 0.28362638593347517,
      "grad_norm": 1.2941293716430664,
      "learning_rate": 2.531e-05,
      "loss": 0.3671,
      "step": 5065
    },
    {
      "epoch": 0.2836823832456042,
      "grad_norm": 0.9018704891204834,
      "learning_rate": 2.5315e-05,
      "loss": 0.3604,
      "step": 5066
    },
    {
      "epoch": 0.2837383805577332,
      "grad_norm": 4.461050510406494,
      "learning_rate": 2.5319999999999998e-05,
      "loss": 0.4973,
      "step": 5067
    },
    {
      "epoch": 0.2837943778698622,
      "grad_norm": 1.8063386678695679,
      "learning_rate": 2.5325e-05,
      "loss": 0.4332,
      "step": 5068
    },
    {
      "epoch": 0.28385037518199124,
      "grad_norm": 1.1380176544189453,
      "learning_rate": 2.5330000000000003e-05,
      "loss": 0.3468,
      "step": 5069
    },
    {
      "epoch": 0.28390637249412026,
      "grad_norm": 1.3340381383895874,
      "learning_rate": 2.5335000000000004e-05,
      "loss": 0.4126,
      "step": 5070
    },
    {
      "epoch": 0.2839623698062493,
      "grad_norm": 1.4871255159378052,
      "learning_rate": 2.534e-05,
      "loss": 0.5937,
      "step": 5071
    },
    {
      "epoch": 0.2840183671183783,
      "grad_norm": 1.2137213945388794,
      "learning_rate": 2.5345000000000002e-05,
      "loss": 0.3692,
      "step": 5072
    },
    {
      "epoch": 0.2840743644305073,
      "grad_norm": 1.7437152862548828,
      "learning_rate": 2.5350000000000003e-05,
      "loss": 0.4339,
      "step": 5073
    },
    {
      "epoch": 0.28413036174263634,
      "grad_norm": 1.2067270278930664,
      "learning_rate": 2.5355e-05,
      "loss": 0.4506,
      "step": 5074
    },
    {
      "epoch": 0.28418635905476536,
      "grad_norm": 1.5555673837661743,
      "learning_rate": 2.536e-05,
      "loss": 0.4121,
      "step": 5075
    },
    {
      "epoch": 0.2842423563668944,
      "grad_norm": 1.3813008069992065,
      "learning_rate": 2.5365e-05,
      "loss": 0.4734,
      "step": 5076
    },
    {
      "epoch": 0.2842983536790234,
      "grad_norm": 1.4264832735061646,
      "learning_rate": 2.537e-05,
      "loss": 0.4616,
      "step": 5077
    },
    {
      "epoch": 0.2843543509911524,
      "grad_norm": 1.092544674873352,
      "learning_rate": 2.5375e-05,
      "loss": 0.3485,
      "step": 5078
    },
    {
      "epoch": 0.28441034830328144,
      "grad_norm": 1.1917117834091187,
      "learning_rate": 2.5380000000000004e-05,
      "loss": 0.5205,
      "step": 5079
    },
    {
      "epoch": 0.28446634561541045,
      "grad_norm": 1.355815052986145,
      "learning_rate": 2.5385000000000002e-05,
      "loss": 0.4565,
      "step": 5080
    },
    {
      "epoch": 0.2845223429275395,
      "grad_norm": 1.0870096683502197,
      "learning_rate": 2.5390000000000003e-05,
      "loss": 0.4144,
      "step": 5081
    },
    {
      "epoch": 0.2845783402396685,
      "grad_norm": 1.094519853591919,
      "learning_rate": 2.5395000000000003e-05,
      "loss": 0.4035,
      "step": 5082
    },
    {
      "epoch": 0.2846343375517975,
      "grad_norm": 1.3015124797821045,
      "learning_rate": 2.54e-05,
      "loss": 0.4412,
      "step": 5083
    },
    {
      "epoch": 0.28469033486392653,
      "grad_norm": 1.3095399141311646,
      "learning_rate": 2.5405e-05,
      "loss": 0.3885,
      "step": 5084
    },
    {
      "epoch": 0.28474633217605555,
      "grad_norm": 1.2276149988174438,
      "learning_rate": 2.541e-05,
      "loss": 0.4867,
      "step": 5085
    },
    {
      "epoch": 0.28480232948818457,
      "grad_norm": 1.4077141284942627,
      "learning_rate": 2.5415e-05,
      "loss": 0.4127,
      "step": 5086
    },
    {
      "epoch": 0.2848583268003136,
      "grad_norm": 7.861964702606201,
      "learning_rate": 2.542e-05,
      "loss": 0.372,
      "step": 5087
    },
    {
      "epoch": 0.2849143241124426,
      "grad_norm": 1.4856375455856323,
      "learning_rate": 2.5424999999999998e-05,
      "loss": 0.5014,
      "step": 5088
    },
    {
      "epoch": 0.2849703214245716,
      "grad_norm": 1.215444564819336,
      "learning_rate": 2.5430000000000002e-05,
      "loss": 0.4289,
      "step": 5089
    },
    {
      "epoch": 0.28502631873670065,
      "grad_norm": 1.1757819652557373,
      "learning_rate": 2.5435000000000003e-05,
      "loss": 0.4394,
      "step": 5090
    },
    {
      "epoch": 0.28508231604882966,
      "grad_norm": 1.2638258934020996,
      "learning_rate": 2.5440000000000004e-05,
      "loss": 0.4226,
      "step": 5091
    },
    {
      "epoch": 0.2851383133609587,
      "grad_norm": 1.574288010597229,
      "learning_rate": 2.5445e-05,
      "loss": 0.3677,
      "step": 5092
    },
    {
      "epoch": 0.2851943106730877,
      "grad_norm": 1.1244778633117676,
      "learning_rate": 2.5450000000000002e-05,
      "loss": 0.4044,
      "step": 5093
    },
    {
      "epoch": 0.2852503079852167,
      "grad_norm": 1.2923802137374878,
      "learning_rate": 2.5455e-05,
      "loss": 0.4191,
      "step": 5094
    },
    {
      "epoch": 0.28530630529734574,
      "grad_norm": 1.1234909296035767,
      "learning_rate": 2.546e-05,
      "loss": 0.5451,
      "step": 5095
    },
    {
      "epoch": 0.28536230260947476,
      "grad_norm": 0.9855756759643555,
      "learning_rate": 2.5465e-05,
      "loss": 0.4568,
      "step": 5096
    },
    {
      "epoch": 0.2854182999216038,
      "grad_norm": 1.175218105316162,
      "learning_rate": 2.547e-05,
      "loss": 0.3712,
      "step": 5097
    },
    {
      "epoch": 0.2854742972337328,
      "grad_norm": 0.9445306658744812,
      "learning_rate": 2.5475e-05,
      "loss": 0.3229,
      "step": 5098
    },
    {
      "epoch": 0.2855302945458618,
      "grad_norm": 1.227358102798462,
      "learning_rate": 2.5480000000000003e-05,
      "loss": 0.4326,
      "step": 5099
    },
    {
      "epoch": 0.28558629185799084,
      "grad_norm": 1.0580247640609741,
      "learning_rate": 2.5485000000000004e-05,
      "loss": 0.3549,
      "step": 5100
    },
    {
      "epoch": 0.28564228917011986,
      "grad_norm": 1.212801218032837,
      "learning_rate": 2.549e-05,
      "loss": 0.4072,
      "step": 5101
    },
    {
      "epoch": 0.2856982864822489,
      "grad_norm": 1.2784329652786255,
      "learning_rate": 2.5495000000000002e-05,
      "loss": 0.4814,
      "step": 5102
    },
    {
      "epoch": 0.2857542837943779,
      "grad_norm": 1.2336822748184204,
      "learning_rate": 2.5500000000000003e-05,
      "loss": 0.4123,
      "step": 5103
    },
    {
      "epoch": 0.2858102811065069,
      "grad_norm": 1.3024848699569702,
      "learning_rate": 2.5505e-05,
      "loss": 0.3914,
      "step": 5104
    },
    {
      "epoch": 0.28586627841863593,
      "grad_norm": 1.5754663944244385,
      "learning_rate": 2.551e-05,
      "loss": 0.5698,
      "step": 5105
    },
    {
      "epoch": 0.2859222757307649,
      "grad_norm": 1.086853265762329,
      "learning_rate": 2.5515e-05,
      "loss": 0.4066,
      "step": 5106
    },
    {
      "epoch": 0.2859782730428939,
      "grad_norm": 1.2721822261810303,
      "learning_rate": 2.552e-05,
      "loss": 0.4683,
      "step": 5107
    },
    {
      "epoch": 0.28603427035502293,
      "grad_norm": 1.3510558605194092,
      "learning_rate": 2.5525e-05,
      "loss": 0.4937,
      "step": 5108
    },
    {
      "epoch": 0.28609026766715195,
      "grad_norm": 1.0186558961868286,
      "learning_rate": 2.5530000000000005e-05,
      "loss": 0.3634,
      "step": 5109
    },
    {
      "epoch": 0.286146264979281,
      "grad_norm": 1.2649930715560913,
      "learning_rate": 2.5535000000000002e-05,
      "loss": 0.4996,
      "step": 5110
    },
    {
      "epoch": 0.28620226229141,
      "grad_norm": 1.2046400308609009,
      "learning_rate": 2.5540000000000003e-05,
      "loss": 0.4747,
      "step": 5111
    },
    {
      "epoch": 0.286258259603539,
      "grad_norm": 1.1831685304641724,
      "learning_rate": 2.5545000000000004e-05,
      "loss": 0.4545,
      "step": 5112
    },
    {
      "epoch": 0.28631425691566803,
      "grad_norm": 1.6218205690383911,
      "learning_rate": 2.555e-05,
      "loss": 0.6066,
      "step": 5113
    },
    {
      "epoch": 0.28637025422779705,
      "grad_norm": 1.156431794166565,
      "learning_rate": 2.5555000000000002e-05,
      "loss": 0.3627,
      "step": 5114
    },
    {
      "epoch": 0.28642625153992607,
      "grad_norm": 1.2976090908050537,
      "learning_rate": 2.556e-05,
      "loss": 0.4457,
      "step": 5115
    },
    {
      "epoch": 0.2864822488520551,
      "grad_norm": 1.2460031509399414,
      "learning_rate": 2.5565e-05,
      "loss": 0.6375,
      "step": 5116
    },
    {
      "epoch": 0.2865382461641841,
      "grad_norm": 1.0107266902923584,
      "learning_rate": 2.557e-05,
      "loss": 0.3809,
      "step": 5117
    },
    {
      "epoch": 0.2865942434763131,
      "grad_norm": 1.262149691581726,
      "learning_rate": 2.5574999999999998e-05,
      "loss": 0.3705,
      "step": 5118
    },
    {
      "epoch": 0.28665024078844215,
      "grad_norm": 1.5213390588760376,
      "learning_rate": 2.5580000000000002e-05,
      "loss": 0.4707,
      "step": 5119
    },
    {
      "epoch": 0.28670623810057116,
      "grad_norm": 1.0707709789276123,
      "learning_rate": 2.5585000000000003e-05,
      "loss": 0.4222,
      "step": 5120
    },
    {
      "epoch": 0.2867622354127002,
      "grad_norm": 1.4900261163711548,
      "learning_rate": 2.5590000000000004e-05,
      "loss": 0.4853,
      "step": 5121
    },
    {
      "epoch": 0.2868182327248292,
      "grad_norm": 1.1637797355651855,
      "learning_rate": 2.5595e-05,
      "loss": 0.3548,
      "step": 5122
    },
    {
      "epoch": 0.2868742300369582,
      "grad_norm": 1.3067561388015747,
      "learning_rate": 2.5600000000000002e-05,
      "loss": 0.5567,
      "step": 5123
    },
    {
      "epoch": 0.28693022734908724,
      "grad_norm": 1.1704126596450806,
      "learning_rate": 2.5605e-05,
      "loss": 0.4554,
      "step": 5124
    },
    {
      "epoch": 0.28698622466121626,
      "grad_norm": 2.44575834274292,
      "learning_rate": 2.561e-05,
      "loss": 0.4063,
      "step": 5125
    },
    {
      "epoch": 0.2870422219733453,
      "grad_norm": 1.180340051651001,
      "learning_rate": 2.5615e-05,
      "loss": 0.3816,
      "step": 5126
    },
    {
      "epoch": 0.2870982192854743,
      "grad_norm": 1.2813947200775146,
      "learning_rate": 2.562e-05,
      "loss": 0.5276,
      "step": 5127
    },
    {
      "epoch": 0.2871542165976033,
      "grad_norm": 1.2095357179641724,
      "learning_rate": 2.5625e-05,
      "loss": 0.3904,
      "step": 5128
    },
    {
      "epoch": 0.28721021390973234,
      "grad_norm": 1.0724081993103027,
      "learning_rate": 2.5629999999999997e-05,
      "loss": 0.4676,
      "step": 5129
    },
    {
      "epoch": 0.28726621122186136,
      "grad_norm": 1.3954498767852783,
      "learning_rate": 2.5635000000000004e-05,
      "loss": 0.4394,
      "step": 5130
    },
    {
      "epoch": 0.2873222085339904,
      "grad_norm": 1.6469731330871582,
      "learning_rate": 2.5640000000000002e-05,
      "loss": 0.5405,
      "step": 5131
    },
    {
      "epoch": 0.2873782058461194,
      "grad_norm": 1.2044121026992798,
      "learning_rate": 2.5645000000000003e-05,
      "loss": 0.4052,
      "step": 5132
    },
    {
      "epoch": 0.2874342031582484,
      "grad_norm": 1.0413527488708496,
      "learning_rate": 2.5650000000000003e-05,
      "loss": 0.2997,
      "step": 5133
    },
    {
      "epoch": 0.28749020047037743,
      "grad_norm": 1.354501485824585,
      "learning_rate": 2.5655e-05,
      "loss": 0.3402,
      "step": 5134
    },
    {
      "epoch": 0.28754619778250645,
      "grad_norm": 1.4380000829696655,
      "learning_rate": 2.566e-05,
      "loss": 0.4822,
      "step": 5135
    },
    {
      "epoch": 0.28760219509463547,
      "grad_norm": 1.2316100597381592,
      "learning_rate": 2.5665e-05,
      "loss": 0.5041,
      "step": 5136
    },
    {
      "epoch": 0.2876581924067645,
      "grad_norm": 1.3962647914886475,
      "learning_rate": 2.567e-05,
      "loss": 0.5794,
      "step": 5137
    },
    {
      "epoch": 0.2877141897188935,
      "grad_norm": 1.6493854522705078,
      "learning_rate": 2.5675e-05,
      "loss": 0.5601,
      "step": 5138
    },
    {
      "epoch": 0.28777018703102253,
      "grad_norm": 1.103853464126587,
      "learning_rate": 2.5679999999999998e-05,
      "loss": 0.617,
      "step": 5139
    },
    {
      "epoch": 0.28782618434315155,
      "grad_norm": 1.2031441926956177,
      "learning_rate": 2.5685000000000002e-05,
      "loss": 0.4776,
      "step": 5140
    },
    {
      "epoch": 0.28788218165528057,
      "grad_norm": 1.3442846536636353,
      "learning_rate": 2.5690000000000003e-05,
      "loss": 0.4194,
      "step": 5141
    },
    {
      "epoch": 0.2879381789674096,
      "grad_norm": 1.297261118888855,
      "learning_rate": 2.5695000000000004e-05,
      "loss": 0.5212,
      "step": 5142
    },
    {
      "epoch": 0.2879941762795386,
      "grad_norm": 1.2500529289245605,
      "learning_rate": 2.57e-05,
      "loss": 0.4298,
      "step": 5143
    },
    {
      "epoch": 0.2880501735916676,
      "grad_norm": 1.1209262609481812,
      "learning_rate": 2.5705000000000002e-05,
      "loss": 0.3535,
      "step": 5144
    },
    {
      "epoch": 0.28810617090379664,
      "grad_norm": 1.3996305465698242,
      "learning_rate": 2.571e-05,
      "loss": 0.4195,
      "step": 5145
    },
    {
      "epoch": 0.28816216821592566,
      "grad_norm": 1.3168365955352783,
      "learning_rate": 2.5715e-05,
      "loss": 0.4505,
      "step": 5146
    },
    {
      "epoch": 0.2882181655280546,
      "grad_norm": 1.3115452527999878,
      "learning_rate": 2.572e-05,
      "loss": 0.4509,
      "step": 5147
    },
    {
      "epoch": 0.28827416284018365,
      "grad_norm": 1.2151601314544678,
      "learning_rate": 2.5725e-05,
      "loss": 0.5359,
      "step": 5148
    },
    {
      "epoch": 0.28833016015231266,
      "grad_norm": 1.0965923070907593,
      "learning_rate": 2.573e-05,
      "loss": 0.4087,
      "step": 5149
    },
    {
      "epoch": 0.2883861574644417,
      "grad_norm": 1.2941659688949585,
      "learning_rate": 2.5735000000000003e-05,
      "loss": 0.4825,
      "step": 5150
    },
    {
      "epoch": 0.2884421547765707,
      "grad_norm": 1.1040208339691162,
      "learning_rate": 2.5740000000000004e-05,
      "loss": 0.4537,
      "step": 5151
    },
    {
      "epoch": 0.2884981520886997,
      "grad_norm": 1.198751449584961,
      "learning_rate": 2.5745e-05,
      "loss": 0.4545,
      "step": 5152
    },
    {
      "epoch": 0.28855414940082874,
      "grad_norm": 1.3784911632537842,
      "learning_rate": 2.5750000000000002e-05,
      "loss": 0.4048,
      "step": 5153
    },
    {
      "epoch": 0.28861014671295776,
      "grad_norm": 1.2006009817123413,
      "learning_rate": 2.5755e-05,
      "loss": 0.4406,
      "step": 5154
    },
    {
      "epoch": 0.2886661440250868,
      "grad_norm": 1.1796938180923462,
      "learning_rate": 2.576e-05,
      "loss": 0.4501,
      "step": 5155
    },
    {
      "epoch": 0.2887221413372158,
      "grad_norm": 1.0965385437011719,
      "learning_rate": 2.5765e-05,
      "loss": 0.3665,
      "step": 5156
    },
    {
      "epoch": 0.2887781386493448,
      "grad_norm": 1.062350869178772,
      "learning_rate": 2.577e-05,
      "loss": 0.3916,
      "step": 5157
    },
    {
      "epoch": 0.28883413596147384,
      "grad_norm": 1.24132239818573,
      "learning_rate": 2.5775e-05,
      "loss": 0.3841,
      "step": 5158
    },
    {
      "epoch": 0.28889013327360286,
      "grad_norm": 1.4969040155410767,
      "learning_rate": 2.5779999999999997e-05,
      "loss": 0.5066,
      "step": 5159
    },
    {
      "epoch": 0.2889461305857319,
      "grad_norm": 1.4512512683868408,
      "learning_rate": 2.5785000000000005e-05,
      "loss": 0.4499,
      "step": 5160
    },
    {
      "epoch": 0.2890021278978609,
      "grad_norm": 1.1753685474395752,
      "learning_rate": 2.5790000000000002e-05,
      "loss": 0.3539,
      "step": 5161
    },
    {
      "epoch": 0.2890581252099899,
      "grad_norm": 1.3325785398483276,
      "learning_rate": 2.5795000000000003e-05,
      "loss": 0.5696,
      "step": 5162
    },
    {
      "epoch": 0.28911412252211893,
      "grad_norm": 1.0597665309906006,
      "learning_rate": 2.58e-05,
      "loss": 0.3675,
      "step": 5163
    },
    {
      "epoch": 0.28917011983424795,
      "grad_norm": 1.442156195640564,
      "learning_rate": 2.5805e-05,
      "loss": 0.4321,
      "step": 5164
    },
    {
      "epoch": 0.28922611714637697,
      "grad_norm": 1.0994163751602173,
      "learning_rate": 2.5810000000000002e-05,
      "loss": 0.3789,
      "step": 5165
    },
    {
      "epoch": 0.289282114458506,
      "grad_norm": 1.2429791688919067,
      "learning_rate": 2.5815e-05,
      "loss": 0.4146,
      "step": 5166
    },
    {
      "epoch": 0.289338111770635,
      "grad_norm": 1.4033571481704712,
      "learning_rate": 2.582e-05,
      "loss": 0.4169,
      "step": 5167
    },
    {
      "epoch": 0.28939410908276403,
      "grad_norm": 1.0820399522781372,
      "learning_rate": 2.5824999999999998e-05,
      "loss": 0.3315,
      "step": 5168
    },
    {
      "epoch": 0.28945010639489305,
      "grad_norm": 1.3229745626449585,
      "learning_rate": 2.583e-05,
      "loss": 0.4174,
      "step": 5169
    },
    {
      "epoch": 0.28950610370702207,
      "grad_norm": 1.4531428813934326,
      "learning_rate": 2.5835000000000003e-05,
      "loss": 0.5334,
      "step": 5170
    },
    {
      "epoch": 0.2895621010191511,
      "grad_norm": 1.3618649244308472,
      "learning_rate": 2.5840000000000003e-05,
      "loss": 0.4907,
      "step": 5171
    },
    {
      "epoch": 0.2896180983312801,
      "grad_norm": 1.25014066696167,
      "learning_rate": 2.5845000000000004e-05,
      "loss": 0.465,
      "step": 5172
    },
    {
      "epoch": 0.2896740956434091,
      "grad_norm": 1.3573648929595947,
      "learning_rate": 2.585e-05,
      "loss": 0.4262,
      "step": 5173
    },
    {
      "epoch": 0.28973009295553814,
      "grad_norm": 1.3631477355957031,
      "learning_rate": 2.5855000000000002e-05,
      "loss": 0.5082,
      "step": 5174
    },
    {
      "epoch": 0.28978609026766716,
      "grad_norm": 1.305877447128296,
      "learning_rate": 2.586e-05,
      "loss": 0.497,
      "step": 5175
    },
    {
      "epoch": 0.2898420875797962,
      "grad_norm": 1.152976632118225,
      "learning_rate": 2.5865e-05,
      "loss": 0.437,
      "step": 5176
    },
    {
      "epoch": 0.2898980848919252,
      "grad_norm": 1.2623867988586426,
      "learning_rate": 2.587e-05,
      "loss": 0.4915,
      "step": 5177
    },
    {
      "epoch": 0.2899540822040542,
      "grad_norm": 1.0398708581924438,
      "learning_rate": 2.5875e-05,
      "loss": 0.435,
      "step": 5178
    },
    {
      "epoch": 0.29001007951618324,
      "grad_norm": 1.1778053045272827,
      "learning_rate": 2.588e-05,
      "loss": 0.524,
      "step": 5179
    },
    {
      "epoch": 0.29006607682831226,
      "grad_norm": 1.231372356414795,
      "learning_rate": 2.5885000000000004e-05,
      "loss": 0.4914,
      "step": 5180
    },
    {
      "epoch": 0.2901220741404413,
      "grad_norm": 1.2939479351043701,
      "learning_rate": 2.5890000000000005e-05,
      "loss": 0.3787,
      "step": 5181
    },
    {
      "epoch": 0.2901780714525703,
      "grad_norm": 1.2962067127227783,
      "learning_rate": 2.5895000000000002e-05,
      "loss": 0.468,
      "step": 5182
    },
    {
      "epoch": 0.2902340687646993,
      "grad_norm": 1.5894113779067993,
      "learning_rate": 2.5900000000000003e-05,
      "loss": 0.5801,
      "step": 5183
    },
    {
      "epoch": 0.29029006607682833,
      "grad_norm": 1.2547224760055542,
      "learning_rate": 2.5905e-05,
      "loss": 0.4094,
      "step": 5184
    },
    {
      "epoch": 0.29034606338895735,
      "grad_norm": 2.165175676345825,
      "learning_rate": 2.591e-05,
      "loss": 0.4425,
      "step": 5185
    },
    {
      "epoch": 0.2904020607010864,
      "grad_norm": 1.166042685508728,
      "learning_rate": 2.5915000000000002e-05,
      "loss": 0.376,
      "step": 5186
    },
    {
      "epoch": 0.2904580580132154,
      "grad_norm": 1.2118500471115112,
      "learning_rate": 2.592e-05,
      "loss": 0.3256,
      "step": 5187
    },
    {
      "epoch": 0.29051405532534436,
      "grad_norm": 1.4399436712265015,
      "learning_rate": 2.5925e-05,
      "loss": 0.5189,
      "step": 5188
    },
    {
      "epoch": 0.2905700526374734,
      "grad_norm": 1.0340296030044556,
      "learning_rate": 2.5929999999999997e-05,
      "loss": 0.2661,
      "step": 5189
    },
    {
      "epoch": 0.2906260499496024,
      "grad_norm": 1.224493384361267,
      "learning_rate": 2.5935000000000005e-05,
      "loss": 0.4585,
      "step": 5190
    },
    {
      "epoch": 0.2906820472617314,
      "grad_norm": 1.3201141357421875,
      "learning_rate": 2.5940000000000002e-05,
      "loss": 0.4362,
      "step": 5191
    },
    {
      "epoch": 0.29073804457386043,
      "grad_norm": 1.560363531112671,
      "learning_rate": 2.5945000000000003e-05,
      "loss": 0.5274,
      "step": 5192
    },
    {
      "epoch": 0.29079404188598945,
      "grad_norm": 1.2800922393798828,
      "learning_rate": 2.595e-05,
      "loss": 0.3426,
      "step": 5193
    },
    {
      "epoch": 0.29085003919811847,
      "grad_norm": 1.2105079889297485,
      "learning_rate": 2.5955e-05,
      "loss": 0.362,
      "step": 5194
    },
    {
      "epoch": 0.2909060365102475,
      "grad_norm": 1.2776095867156982,
      "learning_rate": 2.5960000000000002e-05,
      "loss": 0.3604,
      "step": 5195
    },
    {
      "epoch": 0.2909620338223765,
      "grad_norm": 1.4299017190933228,
      "learning_rate": 2.5965e-05,
      "loss": 0.4469,
      "step": 5196
    },
    {
      "epoch": 0.29101803113450553,
      "grad_norm": 1.1238104104995728,
      "learning_rate": 2.597e-05,
      "loss": 0.3655,
      "step": 5197
    },
    {
      "epoch": 0.29107402844663455,
      "grad_norm": 1.1724493503570557,
      "learning_rate": 2.5974999999999998e-05,
      "loss": 0.4825,
      "step": 5198
    },
    {
      "epoch": 0.29113002575876357,
      "grad_norm": 1.156067967414856,
      "learning_rate": 2.598e-05,
      "loss": 0.4406,
      "step": 5199
    },
    {
      "epoch": 0.2911860230708926,
      "grad_norm": 1.392245888710022,
      "learning_rate": 2.5985000000000003e-05,
      "loss": 0.4914,
      "step": 5200
    },
    {
      "epoch": 0.2912420203830216,
      "grad_norm": 1.2607609033584595,
      "learning_rate": 2.5990000000000004e-05,
      "loss": 0.3389,
      "step": 5201
    },
    {
      "epoch": 0.2912980176951506,
      "grad_norm": 1.2951900959014893,
      "learning_rate": 2.5995000000000004e-05,
      "loss": 0.5232,
      "step": 5202
    },
    {
      "epoch": 0.29135401500727964,
      "grad_norm": 1.1071255207061768,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 0.3389,
      "step": 5203
    },
    {
      "epoch": 0.29141001231940866,
      "grad_norm": 1.5917093753814697,
      "learning_rate": 2.6005000000000003e-05,
      "loss": 0.4165,
      "step": 5204
    },
    {
      "epoch": 0.2914660096315377,
      "grad_norm": 1.740160346031189,
      "learning_rate": 2.601e-05,
      "loss": 0.5188,
      "step": 5205
    },
    {
      "epoch": 0.2915220069436667,
      "grad_norm": 1.6251461505889893,
      "learning_rate": 2.6015e-05,
      "loss": 0.6349,
      "step": 5206
    },
    {
      "epoch": 0.2915780042557957,
      "grad_norm": 1.1782416105270386,
      "learning_rate": 2.602e-05,
      "loss": 0.4572,
      "step": 5207
    },
    {
      "epoch": 0.29163400156792474,
      "grad_norm": 1.1010757684707642,
      "learning_rate": 2.6025e-05,
      "loss": 0.3086,
      "step": 5208
    },
    {
      "epoch": 0.29168999888005376,
      "grad_norm": 1.2914388179779053,
      "learning_rate": 2.603e-05,
      "loss": 0.4165,
      "step": 5209
    },
    {
      "epoch": 0.2917459961921828,
      "grad_norm": 1.0075737237930298,
      "learning_rate": 2.6035000000000004e-05,
      "loss": 0.3249,
      "step": 5210
    },
    {
      "epoch": 0.2918019935043118,
      "grad_norm": 1.1916017532348633,
      "learning_rate": 2.6040000000000005e-05,
      "loss": 0.4307,
      "step": 5211
    },
    {
      "epoch": 0.2918579908164408,
      "grad_norm": 1.2970917224884033,
      "learning_rate": 2.6045000000000002e-05,
      "loss": 0.5413,
      "step": 5212
    },
    {
      "epoch": 0.29191398812856983,
      "grad_norm": 1.0869152545928955,
      "learning_rate": 2.6050000000000003e-05,
      "loss": 0.3936,
      "step": 5213
    },
    {
      "epoch": 0.29196998544069885,
      "grad_norm": 1.3564529418945312,
      "learning_rate": 2.6055e-05,
      "loss": 0.3797,
      "step": 5214
    },
    {
      "epoch": 0.2920259827528279,
      "grad_norm": 1.2981070280075073,
      "learning_rate": 2.606e-05,
      "loss": 0.4802,
      "step": 5215
    },
    {
      "epoch": 0.2920819800649569,
      "grad_norm": 1.3548084497451782,
      "learning_rate": 2.6065000000000002e-05,
      "loss": 0.3205,
      "step": 5216
    },
    {
      "epoch": 0.2921379773770859,
      "grad_norm": 1.379685878753662,
      "learning_rate": 2.607e-05,
      "loss": 0.4269,
      "step": 5217
    },
    {
      "epoch": 0.29219397468921493,
      "grad_norm": 1.2204341888427734,
      "learning_rate": 2.6075e-05,
      "loss": 0.4196,
      "step": 5218
    },
    {
      "epoch": 0.29224997200134395,
      "grad_norm": 1.3498417139053345,
      "learning_rate": 2.6079999999999998e-05,
      "loss": 0.3891,
      "step": 5219
    },
    {
      "epoch": 0.29230596931347297,
      "grad_norm": 1.0870332717895508,
      "learning_rate": 2.6085000000000005e-05,
      "loss": 0.3441,
      "step": 5220
    },
    {
      "epoch": 0.292361966625602,
      "grad_norm": 1.2560654878616333,
      "learning_rate": 2.6090000000000003e-05,
      "loss": 0.4986,
      "step": 5221
    },
    {
      "epoch": 0.292417963937731,
      "grad_norm": 1.2160749435424805,
      "learning_rate": 2.6095000000000003e-05,
      "loss": 0.5128,
      "step": 5222
    },
    {
      "epoch": 0.29247396124986,
      "grad_norm": 1.0488581657409668,
      "learning_rate": 2.61e-05,
      "loss": 0.334,
      "step": 5223
    },
    {
      "epoch": 0.29252995856198905,
      "grad_norm": 1.336931824684143,
      "learning_rate": 2.6105e-05,
      "loss": 0.4615,
      "step": 5224
    },
    {
      "epoch": 0.29258595587411806,
      "grad_norm": 1.1971161365509033,
      "learning_rate": 2.6110000000000002e-05,
      "loss": 0.5235,
      "step": 5225
    },
    {
      "epoch": 0.2926419531862471,
      "grad_norm": 1.2192145586013794,
      "learning_rate": 2.6115e-05,
      "loss": 0.4185,
      "step": 5226
    },
    {
      "epoch": 0.2926979504983761,
      "grad_norm": 1.3168421983718872,
      "learning_rate": 2.612e-05,
      "loss": 0.4166,
      "step": 5227
    },
    {
      "epoch": 0.2927539478105051,
      "grad_norm": 1.271488070487976,
      "learning_rate": 2.6124999999999998e-05,
      "loss": 0.5113,
      "step": 5228
    },
    {
      "epoch": 0.29280994512263414,
      "grad_norm": 1.1534937620162964,
      "learning_rate": 2.613e-05,
      "loss": 0.4285,
      "step": 5229
    },
    {
      "epoch": 0.2928659424347631,
      "grad_norm": 1.4117984771728516,
      "learning_rate": 2.6135000000000003e-05,
      "loss": 0.4157,
      "step": 5230
    },
    {
      "epoch": 0.2929219397468921,
      "grad_norm": 1.0014228820800781,
      "learning_rate": 2.6140000000000004e-05,
      "loss": 0.3379,
      "step": 5231
    },
    {
      "epoch": 0.29297793705902114,
      "grad_norm": 1.0868574380874634,
      "learning_rate": 2.6145e-05,
      "loss": 0.4443,
      "step": 5232
    },
    {
      "epoch": 0.29303393437115016,
      "grad_norm": 1.707353949546814,
      "learning_rate": 2.6150000000000002e-05,
      "loss": 0.4213,
      "step": 5233
    },
    {
      "epoch": 0.2930899316832792,
      "grad_norm": 1.113995909690857,
      "learning_rate": 2.6155000000000003e-05,
      "loss": 0.2867,
      "step": 5234
    },
    {
      "epoch": 0.2931459289954082,
      "grad_norm": 0.9036622047424316,
      "learning_rate": 2.616e-05,
      "loss": 0.2777,
      "step": 5235
    },
    {
      "epoch": 0.2932019263075372,
      "grad_norm": 5.723801136016846,
      "learning_rate": 2.6165e-05,
      "loss": 0.4799,
      "step": 5236
    },
    {
      "epoch": 0.29325792361966624,
      "grad_norm": 1.1289016008377075,
      "learning_rate": 2.617e-05,
      "loss": 0.3623,
      "step": 5237
    },
    {
      "epoch": 0.29331392093179526,
      "grad_norm": 1.2886630296707153,
      "learning_rate": 2.6175e-05,
      "loss": 0.3451,
      "step": 5238
    },
    {
      "epoch": 0.2933699182439243,
      "grad_norm": 1.478747844696045,
      "learning_rate": 2.618e-05,
      "loss": 0.4712,
      "step": 5239
    },
    {
      "epoch": 0.2934259155560533,
      "grad_norm": 1.2470319271087646,
      "learning_rate": 2.6185000000000004e-05,
      "loss": 0.4373,
      "step": 5240
    },
    {
      "epoch": 0.2934819128681823,
      "grad_norm": 1.5660260915756226,
      "learning_rate": 2.6190000000000005e-05,
      "loss": 0.5966,
      "step": 5241
    },
    {
      "epoch": 0.29353791018031133,
      "grad_norm": 1.3090187311172485,
      "learning_rate": 2.6195000000000002e-05,
      "loss": 0.4731,
      "step": 5242
    },
    {
      "epoch": 0.29359390749244035,
      "grad_norm": 1.6193612813949585,
      "learning_rate": 2.6200000000000003e-05,
      "loss": 0.4805,
      "step": 5243
    },
    {
      "epoch": 0.2936499048045694,
      "grad_norm": 1.3604973554611206,
      "learning_rate": 2.6205e-05,
      "loss": 0.4926,
      "step": 5244
    },
    {
      "epoch": 0.2937059021166984,
      "grad_norm": 1.1638587713241577,
      "learning_rate": 2.621e-05,
      "loss": 0.4017,
      "step": 5245
    },
    {
      "epoch": 0.2937618994288274,
      "grad_norm": 1.0777242183685303,
      "learning_rate": 2.6215000000000002e-05,
      "loss": 0.371,
      "step": 5246
    },
    {
      "epoch": 0.29381789674095643,
      "grad_norm": 1.2868236303329468,
      "learning_rate": 2.622e-05,
      "loss": 0.4956,
      "step": 5247
    },
    {
      "epoch": 0.29387389405308545,
      "grad_norm": 1.0613162517547607,
      "learning_rate": 2.6225e-05,
      "loss": 0.3956,
      "step": 5248
    },
    {
      "epoch": 0.29392989136521447,
      "grad_norm": 1.9690322875976562,
      "learning_rate": 2.6229999999999998e-05,
      "loss": 0.4669,
      "step": 5249
    },
    {
      "epoch": 0.2939858886773435,
      "grad_norm": 1.1180588006973267,
      "learning_rate": 2.6235000000000005e-05,
      "loss": 0.4524,
      "step": 5250
    },
    {
      "epoch": 0.2940418859894725,
      "grad_norm": 1.1822772026062012,
      "learning_rate": 2.6240000000000003e-05,
      "loss": 0.4241,
      "step": 5251
    },
    {
      "epoch": 0.2940978833016015,
      "grad_norm": 1.2001798152923584,
      "learning_rate": 2.6245000000000004e-05,
      "loss": 0.4325,
      "step": 5252
    },
    {
      "epoch": 0.29415388061373055,
      "grad_norm": 1.5586347579956055,
      "learning_rate": 2.625e-05,
      "loss": 0.6026,
      "step": 5253
    },
    {
      "epoch": 0.29420987792585956,
      "grad_norm": 1.2429914474487305,
      "learning_rate": 2.6255000000000002e-05,
      "loss": 0.4933,
      "step": 5254
    },
    {
      "epoch": 0.2942658752379886,
      "grad_norm": 1.1091340780258179,
      "learning_rate": 2.6260000000000003e-05,
      "loss": 0.416,
      "step": 5255
    },
    {
      "epoch": 0.2943218725501176,
      "grad_norm": 1.7527451515197754,
      "learning_rate": 2.6265e-05,
      "loss": 0.656,
      "step": 5256
    },
    {
      "epoch": 0.2943778698622466,
      "grad_norm": 1.5141398906707764,
      "learning_rate": 2.627e-05,
      "loss": 0.6055,
      "step": 5257
    },
    {
      "epoch": 0.29443386717437564,
      "grad_norm": 1.19267737865448,
      "learning_rate": 2.6275e-05,
      "loss": 0.3927,
      "step": 5258
    },
    {
      "epoch": 0.29448986448650466,
      "grad_norm": 1.6720216274261475,
      "learning_rate": 2.628e-05,
      "loss": 0.5716,
      "step": 5259
    },
    {
      "epoch": 0.2945458617986337,
      "grad_norm": 1.1752185821533203,
      "learning_rate": 2.6285e-05,
      "loss": 0.4485,
      "step": 5260
    },
    {
      "epoch": 0.2946018591107627,
      "grad_norm": 1.1701414585113525,
      "learning_rate": 2.6290000000000004e-05,
      "loss": 0.4355,
      "step": 5261
    },
    {
      "epoch": 0.2946578564228917,
      "grad_norm": 1.6818978786468506,
      "learning_rate": 2.6295e-05,
      "loss": 0.5101,
      "step": 5262
    },
    {
      "epoch": 0.29471385373502074,
      "grad_norm": 1.0774511098861694,
      "learning_rate": 2.6300000000000002e-05,
      "loss": 0.3872,
      "step": 5263
    },
    {
      "epoch": 0.29476985104714976,
      "grad_norm": 1.1634888648986816,
      "learning_rate": 2.6305000000000003e-05,
      "loss": 0.4338,
      "step": 5264
    },
    {
      "epoch": 0.2948258483592788,
      "grad_norm": 1.1789129972457886,
      "learning_rate": 2.631e-05,
      "loss": 0.4044,
      "step": 5265
    },
    {
      "epoch": 0.2948818456714078,
      "grad_norm": 1.176689624786377,
      "learning_rate": 2.6315e-05,
      "loss": 0.4996,
      "step": 5266
    },
    {
      "epoch": 0.2949378429835368,
      "grad_norm": 1.0302584171295166,
      "learning_rate": 2.632e-05,
      "loss": 0.3189,
      "step": 5267
    },
    {
      "epoch": 0.29499384029566583,
      "grad_norm": 2.2208893299102783,
      "learning_rate": 2.6325e-05,
      "loss": 0.6289,
      "step": 5268
    },
    {
      "epoch": 0.29504983760779485,
      "grad_norm": 1.3514935970306396,
      "learning_rate": 2.633e-05,
      "loss": 0.5208,
      "step": 5269
    },
    {
      "epoch": 0.29510583491992387,
      "grad_norm": 1.4031091928482056,
      "learning_rate": 2.6334999999999998e-05,
      "loss": 0.4056,
      "step": 5270
    },
    {
      "epoch": 0.29516183223205283,
      "grad_norm": 1.3235684633255005,
      "learning_rate": 2.6340000000000002e-05,
      "loss": 0.4818,
      "step": 5271
    },
    {
      "epoch": 0.29521782954418185,
      "grad_norm": 1.2179088592529297,
      "learning_rate": 2.6345000000000003e-05,
      "loss": 0.4226,
      "step": 5272
    },
    {
      "epoch": 0.2952738268563109,
      "grad_norm": 1.4105010032653809,
      "learning_rate": 2.6350000000000004e-05,
      "loss": 0.5017,
      "step": 5273
    },
    {
      "epoch": 0.2953298241684399,
      "grad_norm": 3.7600419521331787,
      "learning_rate": 2.6355e-05,
      "loss": 0.4592,
      "step": 5274
    },
    {
      "epoch": 0.2953858214805689,
      "grad_norm": 1.1268954277038574,
      "learning_rate": 2.6360000000000002e-05,
      "loss": 0.4334,
      "step": 5275
    },
    {
      "epoch": 0.29544181879269793,
      "grad_norm": 1.2404820919036865,
      "learning_rate": 2.6365e-05,
      "loss": 0.3669,
      "step": 5276
    },
    {
      "epoch": 0.29549781610482695,
      "grad_norm": 1.1679812669754028,
      "learning_rate": 2.637e-05,
      "loss": 0.478,
      "step": 5277
    },
    {
      "epoch": 0.29555381341695597,
      "grad_norm": 1.0536789894104004,
      "learning_rate": 2.6375e-05,
      "loss": 0.3539,
      "step": 5278
    },
    {
      "epoch": 0.295609810729085,
      "grad_norm": 1.0712404251098633,
      "learning_rate": 2.6379999999999998e-05,
      "loss": 0.4798,
      "step": 5279
    },
    {
      "epoch": 0.295665808041214,
      "grad_norm": 1.2229100465774536,
      "learning_rate": 2.6385e-05,
      "loss": 0.3869,
      "step": 5280
    },
    {
      "epoch": 0.295721805353343,
      "grad_norm": 1.2663319110870361,
      "learning_rate": 2.6390000000000003e-05,
      "loss": 0.4911,
      "step": 5281
    },
    {
      "epoch": 0.29577780266547204,
      "grad_norm": 1.721091628074646,
      "learning_rate": 2.6395000000000004e-05,
      "loss": 0.5603,
      "step": 5282
    },
    {
      "epoch": 0.29583379997760106,
      "grad_norm": 1.3844891786575317,
      "learning_rate": 2.64e-05,
      "loss": 0.4141,
      "step": 5283
    },
    {
      "epoch": 0.2958897972897301,
      "grad_norm": 1.3836339712142944,
      "learning_rate": 2.6405000000000002e-05,
      "loss": 0.485,
      "step": 5284
    },
    {
      "epoch": 0.2959457946018591,
      "grad_norm": 1.2061278820037842,
      "learning_rate": 2.6410000000000003e-05,
      "loss": 0.3872,
      "step": 5285
    },
    {
      "epoch": 0.2960017919139881,
      "grad_norm": 1.4442470073699951,
      "learning_rate": 2.6415e-05,
      "loss": 0.3686,
      "step": 5286
    },
    {
      "epoch": 0.29605778922611714,
      "grad_norm": 1.3720992803573608,
      "learning_rate": 2.642e-05,
      "loss": 0.5564,
      "step": 5287
    },
    {
      "epoch": 0.29611378653824616,
      "grad_norm": 1.1148436069488525,
      "learning_rate": 2.6425e-05,
      "loss": 0.3856,
      "step": 5288
    },
    {
      "epoch": 0.2961697838503752,
      "grad_norm": 1.1165697574615479,
      "learning_rate": 2.643e-05,
      "loss": 0.3491,
      "step": 5289
    },
    {
      "epoch": 0.2962257811625042,
      "grad_norm": 1.0927644968032837,
      "learning_rate": 2.6435e-05,
      "loss": 0.3341,
      "step": 5290
    },
    {
      "epoch": 0.2962817784746332,
      "grad_norm": 1.2208930253982544,
      "learning_rate": 2.6440000000000004e-05,
      "loss": 0.4179,
      "step": 5291
    },
    {
      "epoch": 0.29633777578676224,
      "grad_norm": 1.1353740692138672,
      "learning_rate": 2.6445000000000002e-05,
      "loss": 0.3374,
      "step": 5292
    },
    {
      "epoch": 0.29639377309889126,
      "grad_norm": 1.3508774042129517,
      "learning_rate": 2.6450000000000003e-05,
      "loss": 0.4931,
      "step": 5293
    },
    {
      "epoch": 0.2964497704110203,
      "grad_norm": 1.188261866569519,
      "learning_rate": 2.6455000000000003e-05,
      "loss": 0.3865,
      "step": 5294
    },
    {
      "epoch": 0.2965057677231493,
      "grad_norm": 1.3219794034957886,
      "learning_rate": 2.646e-05,
      "loss": 0.4982,
      "step": 5295
    },
    {
      "epoch": 0.2965617650352783,
      "grad_norm": 1.3908417224884033,
      "learning_rate": 2.6465e-05,
      "loss": 0.5614,
      "step": 5296
    },
    {
      "epoch": 0.29661776234740733,
      "grad_norm": 0.9926604628562927,
      "learning_rate": 2.647e-05,
      "loss": 0.3857,
      "step": 5297
    },
    {
      "epoch": 0.29667375965953635,
      "grad_norm": 1.7230104207992554,
      "learning_rate": 2.6475e-05,
      "loss": 0.5869,
      "step": 5298
    },
    {
      "epoch": 0.29672975697166537,
      "grad_norm": 1.4793049097061157,
      "learning_rate": 2.648e-05,
      "loss": 0.4726,
      "step": 5299
    },
    {
      "epoch": 0.2967857542837944,
      "grad_norm": 1.1072078943252563,
      "learning_rate": 2.6484999999999998e-05,
      "loss": 0.3418,
      "step": 5300
    },
    {
      "epoch": 0.2968417515959234,
      "grad_norm": 1.1886978149414062,
      "learning_rate": 2.6490000000000002e-05,
      "loss": 0.395,
      "step": 5301
    },
    {
      "epoch": 0.29689774890805243,
      "grad_norm": 1.1559021472930908,
      "learning_rate": 2.6495000000000003e-05,
      "loss": 0.4542,
      "step": 5302
    },
    {
      "epoch": 0.29695374622018145,
      "grad_norm": 1.074407935142517,
      "learning_rate": 2.6500000000000004e-05,
      "loss": 0.4714,
      "step": 5303
    },
    {
      "epoch": 0.29700974353231047,
      "grad_norm": 1.2388737201690674,
      "learning_rate": 2.6505e-05,
      "loss": 0.4807,
      "step": 5304
    },
    {
      "epoch": 0.2970657408444395,
      "grad_norm": 1.353883981704712,
      "learning_rate": 2.6510000000000002e-05,
      "loss": 0.372,
      "step": 5305
    },
    {
      "epoch": 0.2971217381565685,
      "grad_norm": 0.8923586010932922,
      "learning_rate": 2.6515e-05,
      "loss": 0.254,
      "step": 5306
    },
    {
      "epoch": 0.2971777354686975,
      "grad_norm": 1.645772933959961,
      "learning_rate": 2.652e-05,
      "loss": 0.5328,
      "step": 5307
    },
    {
      "epoch": 0.29723373278082654,
      "grad_norm": 1.2057169675827026,
      "learning_rate": 2.6525e-05,
      "loss": 0.4626,
      "step": 5308
    },
    {
      "epoch": 0.29728973009295556,
      "grad_norm": 1.567265510559082,
      "learning_rate": 2.653e-05,
      "loss": 0.6087,
      "step": 5309
    },
    {
      "epoch": 0.2973457274050846,
      "grad_norm": 1.189358115196228,
      "learning_rate": 2.6535e-05,
      "loss": 0.5019,
      "step": 5310
    },
    {
      "epoch": 0.2974017247172136,
      "grad_norm": 1.2605701684951782,
      "learning_rate": 2.6540000000000003e-05,
      "loss": 0.4445,
      "step": 5311
    },
    {
      "epoch": 0.29745772202934256,
      "grad_norm": 1.3523656129837036,
      "learning_rate": 2.6545000000000004e-05,
      "loss": 0.4203,
      "step": 5312
    },
    {
      "epoch": 0.2975137193414716,
      "grad_norm": 1.2546820640563965,
      "learning_rate": 2.655e-05,
      "loss": 0.3697,
      "step": 5313
    },
    {
      "epoch": 0.2975697166536006,
      "grad_norm": 1.0982375144958496,
      "learning_rate": 2.6555000000000002e-05,
      "loss": 0.3868,
      "step": 5314
    },
    {
      "epoch": 0.2976257139657296,
      "grad_norm": 1.1533113718032837,
      "learning_rate": 2.6560000000000003e-05,
      "loss": 0.4109,
      "step": 5315
    },
    {
      "epoch": 0.29768171127785864,
      "grad_norm": 1.237347960472107,
      "learning_rate": 2.6565e-05,
      "loss": 0.4632,
      "step": 5316
    },
    {
      "epoch": 0.29773770858998766,
      "grad_norm": 1.1130071878433228,
      "learning_rate": 2.657e-05,
      "loss": 0.4277,
      "step": 5317
    },
    {
      "epoch": 0.2977937059021167,
      "grad_norm": 1.2339191436767578,
      "learning_rate": 2.6575e-05,
      "loss": 0.3036,
      "step": 5318
    },
    {
      "epoch": 0.2978497032142457,
      "grad_norm": 1.2246606349945068,
      "learning_rate": 2.658e-05,
      "loss": 0.4144,
      "step": 5319
    },
    {
      "epoch": 0.2979057005263747,
      "grad_norm": 1.2826143503189087,
      "learning_rate": 2.6585e-05,
      "loss": 0.5132,
      "step": 5320
    },
    {
      "epoch": 0.29796169783850374,
      "grad_norm": 1.2820706367492676,
      "learning_rate": 2.6590000000000005e-05,
      "loss": 0.3742,
      "step": 5321
    },
    {
      "epoch": 0.29801769515063276,
      "grad_norm": 1.1958434581756592,
      "learning_rate": 2.6595000000000002e-05,
      "loss": 0.4186,
      "step": 5322
    },
    {
      "epoch": 0.2980736924627618,
      "grad_norm": 1.3425551652908325,
      "learning_rate": 2.6600000000000003e-05,
      "loss": 0.3936,
      "step": 5323
    },
    {
      "epoch": 0.2981296897748908,
      "grad_norm": 1.3036679029464722,
      "learning_rate": 2.6605000000000004e-05,
      "loss": 0.4485,
      "step": 5324
    },
    {
      "epoch": 0.2981856870870198,
      "grad_norm": 1.1589664220809937,
      "learning_rate": 2.661e-05,
      "loss": 0.384,
      "step": 5325
    },
    {
      "epoch": 0.29824168439914883,
      "grad_norm": 1.0981476306915283,
      "learning_rate": 2.6615000000000002e-05,
      "loss": 0.3953,
      "step": 5326
    },
    {
      "epoch": 0.29829768171127785,
      "grad_norm": 1.6495417356491089,
      "learning_rate": 2.662e-05,
      "loss": 0.4254,
      "step": 5327
    },
    {
      "epoch": 0.29835367902340687,
      "grad_norm": 2.0696768760681152,
      "learning_rate": 2.6625e-05,
      "loss": 0.4943,
      "step": 5328
    },
    {
      "epoch": 0.2984096763355359,
      "grad_norm": 1.2074174880981445,
      "learning_rate": 2.663e-05,
      "loss": 0.3937,
      "step": 5329
    },
    {
      "epoch": 0.2984656736476649,
      "grad_norm": 1.1499083042144775,
      "learning_rate": 2.6634999999999998e-05,
      "loss": 0.3489,
      "step": 5330
    },
    {
      "epoch": 0.2985216709597939,
      "grad_norm": 1.2360188961029053,
      "learning_rate": 2.6640000000000002e-05,
      "loss": 0.4464,
      "step": 5331
    },
    {
      "epoch": 0.29857766827192295,
      "grad_norm": 1.1654911041259766,
      "learning_rate": 2.6645000000000003e-05,
      "loss": 0.4109,
      "step": 5332
    },
    {
      "epoch": 0.29863366558405197,
      "grad_norm": 1.2289232015609741,
      "learning_rate": 2.6650000000000004e-05,
      "loss": 0.3535,
      "step": 5333
    },
    {
      "epoch": 0.298689662896181,
      "grad_norm": 1.474075436592102,
      "learning_rate": 2.6655e-05,
      "loss": 0.5,
      "step": 5334
    },
    {
      "epoch": 0.29874566020831,
      "grad_norm": 1.3659262657165527,
      "learning_rate": 2.6660000000000002e-05,
      "loss": 0.5725,
      "step": 5335
    },
    {
      "epoch": 0.298801657520439,
      "grad_norm": 1.4412240982055664,
      "learning_rate": 2.6665e-05,
      "loss": 0.4145,
      "step": 5336
    },
    {
      "epoch": 0.29885765483256804,
      "grad_norm": 1.2272512912750244,
      "learning_rate": 2.667e-05,
      "loss": 0.5215,
      "step": 5337
    },
    {
      "epoch": 0.29891365214469706,
      "grad_norm": 1.1624239683151245,
      "learning_rate": 2.6675e-05,
      "loss": 0.545,
      "step": 5338
    },
    {
      "epoch": 0.2989696494568261,
      "grad_norm": 1.381621241569519,
      "learning_rate": 2.668e-05,
      "loss": 0.6109,
      "step": 5339
    },
    {
      "epoch": 0.2990256467689551,
      "grad_norm": 1.4536820650100708,
      "learning_rate": 2.6685e-05,
      "loss": 0.5272,
      "step": 5340
    },
    {
      "epoch": 0.2990816440810841,
      "grad_norm": 0.9184737801551819,
      "learning_rate": 2.6690000000000004e-05,
      "loss": 0.3607,
      "step": 5341
    },
    {
      "epoch": 0.29913764139321314,
      "grad_norm": 1.2206549644470215,
      "learning_rate": 2.6695000000000004e-05,
      "loss": 0.4182,
      "step": 5342
    },
    {
      "epoch": 0.29919363870534216,
      "grad_norm": 1.0115301609039307,
      "learning_rate": 2.6700000000000002e-05,
      "loss": 0.3466,
      "step": 5343
    },
    {
      "epoch": 0.2992496360174712,
      "grad_norm": 1.1144261360168457,
      "learning_rate": 2.6705000000000003e-05,
      "loss": 0.387,
      "step": 5344
    },
    {
      "epoch": 0.2993056333296002,
      "grad_norm": 1.4079464673995972,
      "learning_rate": 2.671e-05,
      "loss": 0.6056,
      "step": 5345
    },
    {
      "epoch": 0.2993616306417292,
      "grad_norm": 1.191615104675293,
      "learning_rate": 2.6715e-05,
      "loss": 0.4494,
      "step": 5346
    },
    {
      "epoch": 0.29941762795385823,
      "grad_norm": 0.9942251443862915,
      "learning_rate": 2.672e-05,
      "loss": 0.3627,
      "step": 5347
    },
    {
      "epoch": 0.29947362526598725,
      "grad_norm": 1.1981528997421265,
      "learning_rate": 2.6725e-05,
      "loss": 0.5028,
      "step": 5348
    },
    {
      "epoch": 0.2995296225781163,
      "grad_norm": 1.0284557342529297,
      "learning_rate": 2.673e-05,
      "loss": 0.4042,
      "step": 5349
    },
    {
      "epoch": 0.2995856198902453,
      "grad_norm": 1.6800273656845093,
      "learning_rate": 2.6734999999999997e-05,
      "loss": 0.5364,
      "step": 5350
    },
    {
      "epoch": 0.2996416172023743,
      "grad_norm": 1.4248425960540771,
      "learning_rate": 2.6740000000000005e-05,
      "loss": 0.5269,
      "step": 5351
    },
    {
      "epoch": 0.29969761451450333,
      "grad_norm": 1.2523330450057983,
      "learning_rate": 2.6745000000000002e-05,
      "loss": 0.5303,
      "step": 5352
    },
    {
      "epoch": 0.29975361182663235,
      "grad_norm": 1.0883934497833252,
      "learning_rate": 2.6750000000000003e-05,
      "loss": 0.3841,
      "step": 5353
    },
    {
      "epoch": 0.2998096091387613,
      "grad_norm": 1.0700803995132446,
      "learning_rate": 2.6755000000000004e-05,
      "loss": 0.393,
      "step": 5354
    },
    {
      "epoch": 0.29986560645089033,
      "grad_norm": 1.1892080307006836,
      "learning_rate": 2.676e-05,
      "loss": 0.299,
      "step": 5355
    },
    {
      "epoch": 0.29992160376301935,
      "grad_norm": 0.9703457355499268,
      "learning_rate": 2.6765000000000002e-05,
      "loss": 0.3694,
      "step": 5356
    },
    {
      "epoch": 0.29997760107514837,
      "grad_norm": 4.798460483551025,
      "learning_rate": 2.677e-05,
      "loss": 0.4071,
      "step": 5357
    },
    {
      "epoch": 0.3000335983872774,
      "grad_norm": 1.0188037157058716,
      "learning_rate": 2.6775e-05,
      "loss": 0.4911,
      "step": 5358
    },
    {
      "epoch": 0.3000895956994064,
      "grad_norm": 1.1308927536010742,
      "learning_rate": 2.678e-05,
      "loss": 0.4291,
      "step": 5359
    },
    {
      "epoch": 0.3001455930115354,
      "grad_norm": 1.2467466592788696,
      "learning_rate": 2.6785e-05,
      "loss": 0.4322,
      "step": 5360
    },
    {
      "epoch": 0.30020159032366445,
      "grad_norm": 1.0073515176773071,
      "learning_rate": 2.6790000000000003e-05,
      "loss": 0.3015,
      "step": 5361
    },
    {
      "epoch": 0.30025758763579347,
      "grad_norm": 1.4517663717269897,
      "learning_rate": 2.6795000000000003e-05,
      "loss": 0.5408,
      "step": 5362
    },
    {
      "epoch": 0.3003135849479225,
      "grad_norm": 1.2505501508712769,
      "learning_rate": 2.6800000000000004e-05,
      "loss": 0.4191,
      "step": 5363
    },
    {
      "epoch": 0.3003695822600515,
      "grad_norm": 1.104689359664917,
      "learning_rate": 2.6805000000000002e-05,
      "loss": 0.3857,
      "step": 5364
    },
    {
      "epoch": 0.3004255795721805,
      "grad_norm": 1.179711103439331,
      "learning_rate": 2.6810000000000003e-05,
      "loss": 0.5031,
      "step": 5365
    },
    {
      "epoch": 0.30048157688430954,
      "grad_norm": 1.175329327583313,
      "learning_rate": 2.6815e-05,
      "loss": 0.3959,
      "step": 5366
    },
    {
      "epoch": 0.30053757419643856,
      "grad_norm": 1.5125898122787476,
      "learning_rate": 2.682e-05,
      "loss": 0.345,
      "step": 5367
    },
    {
      "epoch": 0.3005935715085676,
      "grad_norm": 1.0912848711013794,
      "learning_rate": 2.6825e-05,
      "loss": 0.4557,
      "step": 5368
    },
    {
      "epoch": 0.3006495688206966,
      "grad_norm": 1.1836665868759155,
      "learning_rate": 2.683e-05,
      "loss": 0.4663,
      "step": 5369
    },
    {
      "epoch": 0.3007055661328256,
      "grad_norm": 1.2546424865722656,
      "learning_rate": 2.6835e-05,
      "loss": 0.4343,
      "step": 5370
    },
    {
      "epoch": 0.30076156344495464,
      "grad_norm": 1.251499891281128,
      "learning_rate": 2.6840000000000004e-05,
      "loss": 0.5248,
      "step": 5371
    },
    {
      "epoch": 0.30081756075708366,
      "grad_norm": 1.3481758832931519,
      "learning_rate": 2.6845000000000005e-05,
      "loss": 0.3643,
      "step": 5372
    },
    {
      "epoch": 0.3008735580692127,
      "grad_norm": 1.3178684711456299,
      "learning_rate": 2.6850000000000002e-05,
      "loss": 0.4057,
      "step": 5373
    },
    {
      "epoch": 0.3009295553813417,
      "grad_norm": 2.1568729877471924,
      "learning_rate": 2.6855000000000003e-05,
      "loss": 0.4088,
      "step": 5374
    },
    {
      "epoch": 0.3009855526934707,
      "grad_norm": 1.8466178178787231,
      "learning_rate": 2.686e-05,
      "loss": 0.5227,
      "step": 5375
    },
    {
      "epoch": 0.30104155000559973,
      "grad_norm": 1.2152745723724365,
      "learning_rate": 2.6865e-05,
      "loss": 0.5257,
      "step": 5376
    },
    {
      "epoch": 0.30109754731772875,
      "grad_norm": 1.4119421243667603,
      "learning_rate": 2.6870000000000002e-05,
      "loss": 0.3585,
      "step": 5377
    },
    {
      "epoch": 0.3011535446298578,
      "grad_norm": 1.2167298793792725,
      "learning_rate": 2.6875e-05,
      "loss": 0.4351,
      "step": 5378
    },
    {
      "epoch": 0.3012095419419868,
      "grad_norm": 1.2719478607177734,
      "learning_rate": 2.688e-05,
      "loss": 0.4216,
      "step": 5379
    },
    {
      "epoch": 0.3012655392541158,
      "grad_norm": 1.229483962059021,
      "learning_rate": 2.6884999999999998e-05,
      "loss": 0.4044,
      "step": 5380
    },
    {
      "epoch": 0.30132153656624483,
      "grad_norm": 1.5179332494735718,
      "learning_rate": 2.689e-05,
      "loss": 0.3877,
      "step": 5381
    },
    {
      "epoch": 0.30137753387837385,
      "grad_norm": 1.4955464601516724,
      "learning_rate": 2.6895000000000003e-05,
      "loss": 0.5172,
      "step": 5382
    },
    {
      "epoch": 0.30143353119050287,
      "grad_norm": 1.1615431308746338,
      "learning_rate": 2.6900000000000003e-05,
      "loss": 0.443,
      "step": 5383
    },
    {
      "epoch": 0.3014895285026319,
      "grad_norm": 1.117387294769287,
      "learning_rate": 2.6905e-05,
      "loss": 0.3921,
      "step": 5384
    },
    {
      "epoch": 0.3015455258147609,
      "grad_norm": 1.1309614181518555,
      "learning_rate": 2.691e-05,
      "loss": 0.4126,
      "step": 5385
    },
    {
      "epoch": 0.3016015231268899,
      "grad_norm": 1.202384352684021,
      "learning_rate": 2.6915000000000002e-05,
      "loss": 0.4628,
      "step": 5386
    },
    {
      "epoch": 0.30165752043901894,
      "grad_norm": 1.3168935775756836,
      "learning_rate": 2.692e-05,
      "loss": 0.4685,
      "step": 5387
    },
    {
      "epoch": 0.30171351775114796,
      "grad_norm": 1.3338346481323242,
      "learning_rate": 2.6925e-05,
      "loss": 0.4951,
      "step": 5388
    },
    {
      "epoch": 0.301769515063277,
      "grad_norm": 1.2364870309829712,
      "learning_rate": 2.693e-05,
      "loss": 0.3585,
      "step": 5389
    },
    {
      "epoch": 0.301825512375406,
      "grad_norm": 1.512573003768921,
      "learning_rate": 2.6935e-05,
      "loss": 0.4172,
      "step": 5390
    },
    {
      "epoch": 0.301881509687535,
      "grad_norm": 1.2886953353881836,
      "learning_rate": 2.694e-05,
      "loss": 0.4066,
      "step": 5391
    },
    {
      "epoch": 0.30193750699966404,
      "grad_norm": 1.155105471611023,
      "learning_rate": 2.6945000000000004e-05,
      "loss": 0.3305,
      "step": 5392
    },
    {
      "epoch": 0.30199350431179306,
      "grad_norm": 1.0350037813186646,
      "learning_rate": 2.6950000000000005e-05,
      "loss": 0.4095,
      "step": 5393
    },
    {
      "epoch": 0.3020495016239221,
      "grad_norm": 1.0273566246032715,
      "learning_rate": 2.6955000000000002e-05,
      "loss": 0.3563,
      "step": 5394
    },
    {
      "epoch": 0.30210549893605104,
      "grad_norm": 1.1574289798736572,
      "learning_rate": 2.6960000000000003e-05,
      "loss": 0.5187,
      "step": 5395
    },
    {
      "epoch": 0.30216149624818006,
      "grad_norm": 1.1745827198028564,
      "learning_rate": 2.6965e-05,
      "loss": 0.3549,
      "step": 5396
    },
    {
      "epoch": 0.3022174935603091,
      "grad_norm": 1.4617410898208618,
      "learning_rate": 2.697e-05,
      "loss": 0.5529,
      "step": 5397
    },
    {
      "epoch": 0.3022734908724381,
      "grad_norm": 2.2632431983947754,
      "learning_rate": 2.6975000000000002e-05,
      "loss": 0.5178,
      "step": 5398
    },
    {
      "epoch": 0.3023294881845671,
      "grad_norm": 1.0546752214431763,
      "learning_rate": 2.698e-05,
      "loss": 0.3277,
      "step": 5399
    },
    {
      "epoch": 0.30238548549669614,
      "grad_norm": 1.3697587251663208,
      "learning_rate": 2.6985e-05,
      "loss": 0.4477,
      "step": 5400
    },
    {
      "epoch": 0.30244148280882516,
      "grad_norm": 1.1841217279434204,
      "learning_rate": 2.6989999999999997e-05,
      "loss": 0.6534,
      "step": 5401
    },
    {
      "epoch": 0.3024974801209542,
      "grad_norm": 1.2722182273864746,
      "learning_rate": 2.6995000000000005e-05,
      "loss": 0.3931,
      "step": 5402
    },
    {
      "epoch": 0.3025534774330832,
      "grad_norm": 1.2848879098892212,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 0.3851,
      "step": 5403
    },
    {
      "epoch": 0.3026094747452122,
      "grad_norm": 1.2802610397338867,
      "learning_rate": 2.7005000000000003e-05,
      "loss": 0.5178,
      "step": 5404
    },
    {
      "epoch": 0.30266547205734123,
      "grad_norm": 1.2036280632019043,
      "learning_rate": 2.701e-05,
      "loss": 0.4296,
      "step": 5405
    },
    {
      "epoch": 0.30272146936947025,
      "grad_norm": 1.3071026802062988,
      "learning_rate": 2.7015e-05,
      "loss": 0.5502,
      "step": 5406
    },
    {
      "epoch": 0.30277746668159927,
      "grad_norm": 1.4859977960586548,
      "learning_rate": 2.7020000000000002e-05,
      "loss": 0.5912,
      "step": 5407
    },
    {
      "epoch": 0.3028334639937283,
      "grad_norm": 1.4547327756881714,
      "learning_rate": 2.7025e-05,
      "loss": 0.4606,
      "step": 5408
    },
    {
      "epoch": 0.3028894613058573,
      "grad_norm": 1.0255621671676636,
      "learning_rate": 2.703e-05,
      "loss": 0.4704,
      "step": 5409
    },
    {
      "epoch": 0.30294545861798633,
      "grad_norm": 29.907569885253906,
      "learning_rate": 2.7034999999999998e-05,
      "loss": 0.3666,
      "step": 5410
    },
    {
      "epoch": 0.30300145593011535,
      "grad_norm": 1.1806755065917969,
      "learning_rate": 2.704e-05,
      "loss": 0.443,
      "step": 5411
    },
    {
      "epoch": 0.30305745324224437,
      "grad_norm": 1.1206432580947876,
      "learning_rate": 2.7045000000000003e-05,
      "loss": 0.3466,
      "step": 5412
    },
    {
      "epoch": 0.3031134505543734,
      "grad_norm": 1.1510664224624634,
      "learning_rate": 2.7050000000000004e-05,
      "loss": 0.5531,
      "step": 5413
    },
    {
      "epoch": 0.3031694478665024,
      "grad_norm": 1.128298044204712,
      "learning_rate": 2.7055e-05,
      "loss": 0.4528,
      "step": 5414
    },
    {
      "epoch": 0.3032254451786314,
      "grad_norm": 1.5585412979125977,
      "learning_rate": 2.7060000000000002e-05,
      "loss": 0.6488,
      "step": 5415
    },
    {
      "epoch": 0.30328144249076044,
      "grad_norm": 1.4108123779296875,
      "learning_rate": 2.7065000000000003e-05,
      "loss": 0.4687,
      "step": 5416
    },
    {
      "epoch": 0.30333743980288946,
      "grad_norm": 0.9559827446937561,
      "learning_rate": 2.707e-05,
      "loss": 0.2918,
      "step": 5417
    },
    {
      "epoch": 0.3033934371150185,
      "grad_norm": 1.0924289226531982,
      "learning_rate": 2.7075e-05,
      "loss": 0.4397,
      "step": 5418
    },
    {
      "epoch": 0.3034494344271475,
      "grad_norm": 1.3185755014419556,
      "learning_rate": 2.7079999999999998e-05,
      "loss": 0.4814,
      "step": 5419
    },
    {
      "epoch": 0.3035054317392765,
      "grad_norm": 1.189787745475769,
      "learning_rate": 2.7085e-05,
      "loss": 0.4277,
      "step": 5420
    },
    {
      "epoch": 0.30356142905140554,
      "grad_norm": 1.108835220336914,
      "learning_rate": 2.709e-05,
      "loss": 0.4128,
      "step": 5421
    },
    {
      "epoch": 0.30361742636353456,
      "grad_norm": 1.2378427982330322,
      "learning_rate": 2.7095000000000004e-05,
      "loss": 0.3862,
      "step": 5422
    },
    {
      "epoch": 0.3036734236756636,
      "grad_norm": 1.2699310779571533,
      "learning_rate": 2.7100000000000005e-05,
      "loss": 0.3449,
      "step": 5423
    },
    {
      "epoch": 0.3037294209877926,
      "grad_norm": 1.3861756324768066,
      "learning_rate": 2.7105000000000002e-05,
      "loss": 0.5577,
      "step": 5424
    },
    {
      "epoch": 0.3037854182999216,
      "grad_norm": 1.676414966583252,
      "learning_rate": 2.7110000000000003e-05,
      "loss": 0.4003,
      "step": 5425
    },
    {
      "epoch": 0.30384141561205064,
      "grad_norm": 1.1581931114196777,
      "learning_rate": 2.7115e-05,
      "loss": 0.4549,
      "step": 5426
    },
    {
      "epoch": 0.30389741292417966,
      "grad_norm": 1.4321802854537964,
      "learning_rate": 2.712e-05,
      "loss": 0.53,
      "step": 5427
    },
    {
      "epoch": 0.3039534102363087,
      "grad_norm": 1.5010488033294678,
      "learning_rate": 2.7125000000000002e-05,
      "loss": 0.4593,
      "step": 5428
    },
    {
      "epoch": 0.3040094075484377,
      "grad_norm": 1.455575704574585,
      "learning_rate": 2.713e-05,
      "loss": 0.4798,
      "step": 5429
    },
    {
      "epoch": 0.3040654048605667,
      "grad_norm": 1.265520691871643,
      "learning_rate": 2.7135e-05,
      "loss": 0.6806,
      "step": 5430
    },
    {
      "epoch": 0.30412140217269573,
      "grad_norm": 1.3100696802139282,
      "learning_rate": 2.7139999999999998e-05,
      "loss": 0.6165,
      "step": 5431
    },
    {
      "epoch": 0.30417739948482475,
      "grad_norm": 1.181958556175232,
      "learning_rate": 2.7145000000000005e-05,
      "loss": 0.4008,
      "step": 5432
    },
    {
      "epoch": 0.30423339679695377,
      "grad_norm": 1.2225289344787598,
      "learning_rate": 2.7150000000000003e-05,
      "loss": 0.3583,
      "step": 5433
    },
    {
      "epoch": 0.3042893941090828,
      "grad_norm": 1.6007293462753296,
      "learning_rate": 2.7155000000000003e-05,
      "loss": 0.4714,
      "step": 5434
    },
    {
      "epoch": 0.3043453914212118,
      "grad_norm": 1.3652026653289795,
      "learning_rate": 2.716e-05,
      "loss": 0.4413,
      "step": 5435
    },
    {
      "epoch": 0.30440138873334077,
      "grad_norm": 1.3229869604110718,
      "learning_rate": 2.7165e-05,
      "loss": 0.3679,
      "step": 5436
    },
    {
      "epoch": 0.3044573860454698,
      "grad_norm": 1.6404391527175903,
      "learning_rate": 2.7170000000000002e-05,
      "loss": 0.4618,
      "step": 5437
    },
    {
      "epoch": 0.3045133833575988,
      "grad_norm": 1.1321698427200317,
      "learning_rate": 2.7175e-05,
      "loss": 0.4808,
      "step": 5438
    },
    {
      "epoch": 0.30456938066972783,
      "grad_norm": 1.2999236583709717,
      "learning_rate": 2.718e-05,
      "loss": 0.433,
      "step": 5439
    },
    {
      "epoch": 0.30462537798185685,
      "grad_norm": 1.1787896156311035,
      "learning_rate": 2.7184999999999998e-05,
      "loss": 0.5162,
      "step": 5440
    },
    {
      "epoch": 0.30468137529398587,
      "grad_norm": 1.1056458950042725,
      "learning_rate": 2.719e-05,
      "loss": 0.368,
      "step": 5441
    },
    {
      "epoch": 0.3047373726061149,
      "grad_norm": 1.2190624475479126,
      "learning_rate": 2.7195000000000003e-05,
      "loss": 0.4291,
      "step": 5442
    },
    {
      "epoch": 0.3047933699182439,
      "grad_norm": 1.6064512729644775,
      "learning_rate": 2.7200000000000004e-05,
      "loss": 0.493,
      "step": 5443
    },
    {
      "epoch": 0.3048493672303729,
      "grad_norm": 1.5153958797454834,
      "learning_rate": 2.7205e-05,
      "loss": 0.4283,
      "step": 5444
    },
    {
      "epoch": 0.30490536454250194,
      "grad_norm": 1.2916572093963623,
      "learning_rate": 2.7210000000000002e-05,
      "loss": 0.4106,
      "step": 5445
    },
    {
      "epoch": 0.30496136185463096,
      "grad_norm": 1.1816558837890625,
      "learning_rate": 2.7215000000000003e-05,
      "loss": 0.3662,
      "step": 5446
    },
    {
      "epoch": 0.30501735916676,
      "grad_norm": 1.0784937143325806,
      "learning_rate": 2.722e-05,
      "loss": 0.4712,
      "step": 5447
    },
    {
      "epoch": 0.305073356478889,
      "grad_norm": 1.6013766527175903,
      "learning_rate": 2.7225e-05,
      "loss": 0.5618,
      "step": 5448
    },
    {
      "epoch": 0.305129353791018,
      "grad_norm": 1.3034296035766602,
      "learning_rate": 2.723e-05,
      "loss": 0.5378,
      "step": 5449
    },
    {
      "epoch": 0.30518535110314704,
      "grad_norm": 1.2664868831634521,
      "learning_rate": 2.7235e-05,
      "loss": 0.4298,
      "step": 5450
    },
    {
      "epoch": 0.30524134841527606,
      "grad_norm": 1.2620458602905273,
      "learning_rate": 2.724e-05,
      "loss": 0.3901,
      "step": 5451
    },
    {
      "epoch": 0.3052973457274051,
      "grad_norm": 1.152858018875122,
      "learning_rate": 2.7245000000000004e-05,
      "loss": 0.4303,
      "step": 5452
    },
    {
      "epoch": 0.3053533430395341,
      "grad_norm": 1.1570677757263184,
      "learning_rate": 2.725e-05,
      "loss": 0.4018,
      "step": 5453
    },
    {
      "epoch": 0.3054093403516631,
      "grad_norm": 1.1236780881881714,
      "learning_rate": 2.7255000000000002e-05,
      "loss": 0.3916,
      "step": 5454
    },
    {
      "epoch": 0.30546533766379214,
      "grad_norm": 1.2584707736968994,
      "learning_rate": 2.7260000000000003e-05,
      "loss": 0.4971,
      "step": 5455
    },
    {
      "epoch": 0.30552133497592115,
      "grad_norm": 1.286197304725647,
      "learning_rate": 2.7265e-05,
      "loss": 0.4123,
      "step": 5456
    },
    {
      "epoch": 0.3055773322880502,
      "grad_norm": 1.1834160089492798,
      "learning_rate": 2.727e-05,
      "loss": 0.4101,
      "step": 5457
    },
    {
      "epoch": 0.3056333296001792,
      "grad_norm": 0.9636449217796326,
      "learning_rate": 2.7275e-05,
      "loss": 0.3525,
      "step": 5458
    },
    {
      "epoch": 0.3056893269123082,
      "grad_norm": 1.1810790300369263,
      "learning_rate": 2.728e-05,
      "loss": 0.4266,
      "step": 5459
    },
    {
      "epoch": 0.30574532422443723,
      "grad_norm": 1.2433044910430908,
      "learning_rate": 2.7285e-05,
      "loss": 0.4085,
      "step": 5460
    },
    {
      "epoch": 0.30580132153656625,
      "grad_norm": 1.2087873220443726,
      "learning_rate": 2.7289999999999998e-05,
      "loss": 0.4045,
      "step": 5461
    },
    {
      "epoch": 0.30585731884869527,
      "grad_norm": 1.1909713745117188,
      "learning_rate": 2.7295000000000005e-05,
      "loss": 0.4767,
      "step": 5462
    },
    {
      "epoch": 0.3059133161608243,
      "grad_norm": 1.2529692649841309,
      "learning_rate": 2.7300000000000003e-05,
      "loss": 0.3872,
      "step": 5463
    },
    {
      "epoch": 0.3059693134729533,
      "grad_norm": 1.424086093902588,
      "learning_rate": 2.7305000000000004e-05,
      "loss": 0.5771,
      "step": 5464
    },
    {
      "epoch": 0.3060253107850823,
      "grad_norm": 1.120654821395874,
      "learning_rate": 2.731e-05,
      "loss": 0.3698,
      "step": 5465
    },
    {
      "epoch": 0.30608130809721135,
      "grad_norm": 1.3193612098693848,
      "learning_rate": 2.7315000000000002e-05,
      "loss": 0.417,
      "step": 5466
    },
    {
      "epoch": 0.30613730540934037,
      "grad_norm": 1.2496871948242188,
      "learning_rate": 2.7320000000000003e-05,
      "loss": 0.4239,
      "step": 5467
    },
    {
      "epoch": 0.3061933027214694,
      "grad_norm": 1.2125095129013062,
      "learning_rate": 2.7325e-05,
      "loss": 0.3446,
      "step": 5468
    },
    {
      "epoch": 0.3062493000335984,
      "grad_norm": 1.3269721269607544,
      "learning_rate": 2.733e-05,
      "loss": 0.4778,
      "step": 5469
    },
    {
      "epoch": 0.3063052973457274,
      "grad_norm": 1.2976957559585571,
      "learning_rate": 2.7335e-05,
      "loss": 0.4254,
      "step": 5470
    },
    {
      "epoch": 0.30636129465785644,
      "grad_norm": 1.2981770038604736,
      "learning_rate": 2.734e-05,
      "loss": 0.4417,
      "step": 5471
    },
    {
      "epoch": 0.30641729196998546,
      "grad_norm": 1.0654748678207397,
      "learning_rate": 2.7345000000000003e-05,
      "loss": 0.3958,
      "step": 5472
    },
    {
      "epoch": 0.3064732892821145,
      "grad_norm": 1.086867094039917,
      "learning_rate": 2.7350000000000004e-05,
      "loss": 0.4133,
      "step": 5473
    },
    {
      "epoch": 0.3065292865942435,
      "grad_norm": 1.3270715475082397,
      "learning_rate": 2.7355e-05,
      "loss": 0.5076,
      "step": 5474
    },
    {
      "epoch": 0.3065852839063725,
      "grad_norm": 1.3053569793701172,
      "learning_rate": 2.7360000000000002e-05,
      "loss": 0.5096,
      "step": 5475
    },
    {
      "epoch": 0.30664128121850154,
      "grad_norm": 1.1925352811813354,
      "learning_rate": 2.7365000000000003e-05,
      "loss": 0.4149,
      "step": 5476
    },
    {
      "epoch": 0.30669727853063056,
      "grad_norm": 1.3278213739395142,
      "learning_rate": 2.737e-05,
      "loss": 0.4896,
      "step": 5477
    },
    {
      "epoch": 0.3067532758427595,
      "grad_norm": 1.2239010334014893,
      "learning_rate": 2.7375e-05,
      "loss": 0.529,
      "step": 5478
    },
    {
      "epoch": 0.30680927315488854,
      "grad_norm": 1.0836764574050903,
      "learning_rate": 2.738e-05,
      "loss": 0.3727,
      "step": 5479
    },
    {
      "epoch": 0.30686527046701756,
      "grad_norm": 1.2639875411987305,
      "learning_rate": 2.7385e-05,
      "loss": 0.4574,
      "step": 5480
    },
    {
      "epoch": 0.3069212677791466,
      "grad_norm": 1.3477859497070312,
      "learning_rate": 2.739e-05,
      "loss": 0.5604,
      "step": 5481
    },
    {
      "epoch": 0.3069772650912756,
      "grad_norm": 1.1119790077209473,
      "learning_rate": 2.7395000000000005e-05,
      "loss": 0.4352,
      "step": 5482
    },
    {
      "epoch": 0.3070332624034046,
      "grad_norm": 1.3080546855926514,
      "learning_rate": 2.7400000000000002e-05,
      "loss": 0.4564,
      "step": 5483
    },
    {
      "epoch": 0.30708925971553364,
      "grad_norm": 1.1802984476089478,
      "learning_rate": 2.7405000000000003e-05,
      "loss": 0.4341,
      "step": 5484
    },
    {
      "epoch": 0.30714525702766265,
      "grad_norm": 1.335644006729126,
      "learning_rate": 2.7410000000000004e-05,
      "loss": 0.6299,
      "step": 5485
    },
    {
      "epoch": 0.3072012543397917,
      "grad_norm": 1.1943070888519287,
      "learning_rate": 2.7415e-05,
      "loss": 0.4315,
      "step": 5486
    },
    {
      "epoch": 0.3072572516519207,
      "grad_norm": 1.4698803424835205,
      "learning_rate": 2.7420000000000002e-05,
      "loss": 0.4656,
      "step": 5487
    },
    {
      "epoch": 0.3073132489640497,
      "grad_norm": 1.4018625020980835,
      "learning_rate": 2.7425e-05,
      "loss": 0.3524,
      "step": 5488
    },
    {
      "epoch": 0.30736924627617873,
      "grad_norm": 1.2805378437042236,
      "learning_rate": 2.743e-05,
      "loss": 0.4766,
      "step": 5489
    },
    {
      "epoch": 0.30742524358830775,
      "grad_norm": 1.030600905418396,
      "learning_rate": 2.7435e-05,
      "loss": 0.365,
      "step": 5490
    },
    {
      "epoch": 0.30748124090043677,
      "grad_norm": 1.0351371765136719,
      "learning_rate": 2.7439999999999998e-05,
      "loss": 0.3058,
      "step": 5491
    },
    {
      "epoch": 0.3075372382125658,
      "grad_norm": 1.1799652576446533,
      "learning_rate": 2.7445000000000002e-05,
      "loss": 0.4724,
      "step": 5492
    },
    {
      "epoch": 0.3075932355246948,
      "grad_norm": 1.1889554262161255,
      "learning_rate": 2.7450000000000003e-05,
      "loss": 0.3925,
      "step": 5493
    },
    {
      "epoch": 0.3076492328368238,
      "grad_norm": 1.3212112188339233,
      "learning_rate": 2.7455000000000004e-05,
      "loss": 0.4217,
      "step": 5494
    },
    {
      "epoch": 0.30770523014895285,
      "grad_norm": 1.0993441343307495,
      "learning_rate": 2.746e-05,
      "loss": 0.3788,
      "step": 5495
    },
    {
      "epoch": 0.30776122746108187,
      "grad_norm": 1.5850281715393066,
      "learning_rate": 2.7465000000000002e-05,
      "loss": 0.4674,
      "step": 5496
    },
    {
      "epoch": 0.3078172247732109,
      "grad_norm": 1.312206506729126,
      "learning_rate": 2.7470000000000003e-05,
      "loss": 0.4479,
      "step": 5497
    },
    {
      "epoch": 0.3078732220853399,
      "grad_norm": 1.1603442430496216,
      "learning_rate": 2.7475e-05,
      "loss": 0.4297,
      "step": 5498
    },
    {
      "epoch": 0.3079292193974689,
      "grad_norm": 1.311500072479248,
      "learning_rate": 2.748e-05,
      "loss": 0.4327,
      "step": 5499
    },
    {
      "epoch": 0.30798521670959794,
      "grad_norm": 1.2915374040603638,
      "learning_rate": 2.7485e-05,
      "loss": 0.5015,
      "step": 5500
    },
    {
      "epoch": 0.30804121402172696,
      "grad_norm": 1.0879619121551514,
      "learning_rate": 2.749e-05,
      "loss": 0.3734,
      "step": 5501
    },
    {
      "epoch": 0.308097211333856,
      "grad_norm": 1.2090144157409668,
      "learning_rate": 2.7495000000000004e-05,
      "loss": 0.4948,
      "step": 5502
    },
    {
      "epoch": 0.308153208645985,
      "grad_norm": 1.6563881635665894,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 0.4062,
      "step": 5503
    },
    {
      "epoch": 0.308209205958114,
      "grad_norm": 1.4525583982467651,
      "learning_rate": 2.7505000000000002e-05,
      "loss": 0.3904,
      "step": 5504
    },
    {
      "epoch": 0.30826520327024304,
      "grad_norm": 1.367253303527832,
      "learning_rate": 2.7510000000000003e-05,
      "loss": 0.3965,
      "step": 5505
    },
    {
      "epoch": 0.30832120058237206,
      "grad_norm": 1.3779171705245972,
      "learning_rate": 2.7515000000000003e-05,
      "loss": 0.4004,
      "step": 5506
    },
    {
      "epoch": 0.3083771978945011,
      "grad_norm": 1.220894694328308,
      "learning_rate": 2.752e-05,
      "loss": 0.4829,
      "step": 5507
    },
    {
      "epoch": 0.3084331952066301,
      "grad_norm": 1.0637099742889404,
      "learning_rate": 2.7525e-05,
      "loss": 0.3627,
      "step": 5508
    },
    {
      "epoch": 0.3084891925187591,
      "grad_norm": 1.20713210105896,
      "learning_rate": 2.753e-05,
      "loss": 0.3328,
      "step": 5509
    },
    {
      "epoch": 0.30854518983088813,
      "grad_norm": 1.1569844484329224,
      "learning_rate": 2.7535e-05,
      "loss": 0.5242,
      "step": 5510
    },
    {
      "epoch": 0.30860118714301715,
      "grad_norm": 1.340274453163147,
      "learning_rate": 2.754e-05,
      "loss": 0.4154,
      "step": 5511
    },
    {
      "epoch": 0.30865718445514617,
      "grad_norm": 1.2659398317337036,
      "learning_rate": 2.7544999999999998e-05,
      "loss": 0.368,
      "step": 5512
    },
    {
      "epoch": 0.3087131817672752,
      "grad_norm": 0.991502046585083,
      "learning_rate": 2.7550000000000002e-05,
      "loss": 0.3658,
      "step": 5513
    },
    {
      "epoch": 0.3087691790794042,
      "grad_norm": 1.056311845779419,
      "learning_rate": 2.7555000000000003e-05,
      "loss": 0.4026,
      "step": 5514
    },
    {
      "epoch": 0.30882517639153323,
      "grad_norm": 1.4860235452651978,
      "learning_rate": 2.7560000000000004e-05,
      "loss": 0.4851,
      "step": 5515
    },
    {
      "epoch": 0.30888117370366225,
      "grad_norm": 1.1402136087417603,
      "learning_rate": 2.7565e-05,
      "loss": 0.315,
      "step": 5516
    },
    {
      "epoch": 0.30893717101579127,
      "grad_norm": 1.2493200302124023,
      "learning_rate": 2.7570000000000002e-05,
      "loss": 0.467,
      "step": 5517
    },
    {
      "epoch": 0.3089931683279203,
      "grad_norm": 1.1811987161636353,
      "learning_rate": 2.7575e-05,
      "loss": 0.3523,
      "step": 5518
    },
    {
      "epoch": 0.30904916564004925,
      "grad_norm": 2.222553253173828,
      "learning_rate": 2.758e-05,
      "loss": 0.542,
      "step": 5519
    },
    {
      "epoch": 0.30910516295217827,
      "grad_norm": 1.4518640041351318,
      "learning_rate": 2.7585e-05,
      "loss": 0.4165,
      "step": 5520
    },
    {
      "epoch": 0.3091611602643073,
      "grad_norm": 1.4288419485092163,
      "learning_rate": 2.759e-05,
      "loss": 0.4544,
      "step": 5521
    },
    {
      "epoch": 0.3092171575764363,
      "grad_norm": 1.4100167751312256,
      "learning_rate": 2.7595e-05,
      "loss": 0.5303,
      "step": 5522
    },
    {
      "epoch": 0.3092731548885653,
      "grad_norm": 1.3606417179107666,
      "learning_rate": 2.7600000000000003e-05,
      "loss": 0.4296,
      "step": 5523
    },
    {
      "epoch": 0.30932915220069435,
      "grad_norm": 1.1433569192886353,
      "learning_rate": 2.7605000000000004e-05,
      "loss": 0.389,
      "step": 5524
    },
    {
      "epoch": 0.30938514951282337,
      "grad_norm": 1.2820134162902832,
      "learning_rate": 2.761e-05,
      "loss": 0.5378,
      "step": 5525
    },
    {
      "epoch": 0.3094411468249524,
      "grad_norm": 1.375495433807373,
      "learning_rate": 2.7615000000000002e-05,
      "loss": 0.4937,
      "step": 5526
    },
    {
      "epoch": 0.3094971441370814,
      "grad_norm": 1.3101216554641724,
      "learning_rate": 2.762e-05,
      "loss": 0.4153,
      "step": 5527
    },
    {
      "epoch": 0.3095531414492104,
      "grad_norm": 1.2939258813858032,
      "learning_rate": 2.7625e-05,
      "loss": 0.4385,
      "step": 5528
    },
    {
      "epoch": 0.30960913876133944,
      "grad_norm": 1.3447812795639038,
      "learning_rate": 2.763e-05,
      "loss": 0.5507,
      "step": 5529
    },
    {
      "epoch": 0.30966513607346846,
      "grad_norm": 1.0491257905960083,
      "learning_rate": 2.7635e-05,
      "loss": 0.388,
      "step": 5530
    },
    {
      "epoch": 0.3097211333855975,
      "grad_norm": 1.0880898237228394,
      "learning_rate": 2.764e-05,
      "loss": 0.4203,
      "step": 5531
    },
    {
      "epoch": 0.3097771306977265,
      "grad_norm": 1.275537371635437,
      "learning_rate": 2.7644999999999997e-05,
      "loss": 0.4122,
      "step": 5532
    },
    {
      "epoch": 0.3098331280098555,
      "grad_norm": 1.2868367433547974,
      "learning_rate": 2.7650000000000005e-05,
      "loss": 0.4093,
      "step": 5533
    },
    {
      "epoch": 0.30988912532198454,
      "grad_norm": 1.0064719915390015,
      "learning_rate": 2.7655000000000002e-05,
      "loss": 0.4153,
      "step": 5534
    },
    {
      "epoch": 0.30994512263411356,
      "grad_norm": 1.1891820430755615,
      "learning_rate": 2.7660000000000003e-05,
      "loss": 0.4099,
      "step": 5535
    },
    {
      "epoch": 0.3100011199462426,
      "grad_norm": 1.3737549781799316,
      "learning_rate": 2.7665000000000004e-05,
      "loss": 0.5673,
      "step": 5536
    },
    {
      "epoch": 0.3100571172583716,
      "grad_norm": 1.3942443132400513,
      "learning_rate": 2.767e-05,
      "loss": 0.4315,
      "step": 5537
    },
    {
      "epoch": 0.3101131145705006,
      "grad_norm": 1.105025291442871,
      "learning_rate": 2.7675000000000002e-05,
      "loss": 0.4325,
      "step": 5538
    },
    {
      "epoch": 0.31016911188262963,
      "grad_norm": 1.2726826667785645,
      "learning_rate": 2.768e-05,
      "loss": 0.419,
      "step": 5539
    },
    {
      "epoch": 0.31022510919475865,
      "grad_norm": 1.1925272941589355,
      "learning_rate": 2.7685e-05,
      "loss": 0.3958,
      "step": 5540
    },
    {
      "epoch": 0.31028110650688767,
      "grad_norm": 1.500717043876648,
      "learning_rate": 2.769e-05,
      "loss": 0.3668,
      "step": 5541
    },
    {
      "epoch": 0.3103371038190167,
      "grad_norm": 1.0971508026123047,
      "learning_rate": 2.7694999999999998e-05,
      "loss": 0.33,
      "step": 5542
    },
    {
      "epoch": 0.3103931011311457,
      "grad_norm": 1.1650553941726685,
      "learning_rate": 2.7700000000000002e-05,
      "loss": 0.4002,
      "step": 5543
    },
    {
      "epoch": 0.31044909844327473,
      "grad_norm": 1.1508090496063232,
      "learning_rate": 2.7705000000000003e-05,
      "loss": 0.5138,
      "step": 5544
    },
    {
      "epoch": 0.31050509575540375,
      "grad_norm": 1.3230493068695068,
      "learning_rate": 2.7710000000000004e-05,
      "loss": 0.5699,
      "step": 5545
    },
    {
      "epoch": 0.31056109306753277,
      "grad_norm": 1.0767821073532104,
      "learning_rate": 2.7715e-05,
      "loss": 0.4049,
      "step": 5546
    },
    {
      "epoch": 0.3106170903796618,
      "grad_norm": 1.247046947479248,
      "learning_rate": 2.7720000000000002e-05,
      "loss": 0.4443,
      "step": 5547
    },
    {
      "epoch": 0.3106730876917908,
      "grad_norm": 1.0499744415283203,
      "learning_rate": 2.7725e-05,
      "loss": 0.4459,
      "step": 5548
    },
    {
      "epoch": 0.3107290850039198,
      "grad_norm": 1.4015235900878906,
      "learning_rate": 2.773e-05,
      "loss": 0.5551,
      "step": 5549
    },
    {
      "epoch": 0.31078508231604884,
      "grad_norm": 1.2208417654037476,
      "learning_rate": 2.7735e-05,
      "loss": 0.4329,
      "step": 5550
    },
    {
      "epoch": 0.31084107962817786,
      "grad_norm": 1.5305595397949219,
      "learning_rate": 2.774e-05,
      "loss": 0.4818,
      "step": 5551
    },
    {
      "epoch": 0.3108970769403069,
      "grad_norm": 1.173004388809204,
      "learning_rate": 2.7745e-05,
      "loss": 0.4242,
      "step": 5552
    },
    {
      "epoch": 0.3109530742524359,
      "grad_norm": 1.2876002788543701,
      "learning_rate": 2.7750000000000004e-05,
      "loss": 0.36,
      "step": 5553
    },
    {
      "epoch": 0.3110090715645649,
      "grad_norm": 1.264246940612793,
      "learning_rate": 2.7755000000000004e-05,
      "loss": 0.3828,
      "step": 5554
    },
    {
      "epoch": 0.31106506887669394,
      "grad_norm": 1.0596487522125244,
      "learning_rate": 2.7760000000000002e-05,
      "loss": 0.3565,
      "step": 5555
    },
    {
      "epoch": 0.31112106618882296,
      "grad_norm": 1.1578136682510376,
      "learning_rate": 2.7765000000000003e-05,
      "loss": 0.4483,
      "step": 5556
    },
    {
      "epoch": 0.311177063500952,
      "grad_norm": 1.1922684907913208,
      "learning_rate": 2.777e-05,
      "loss": 0.3374,
      "step": 5557
    },
    {
      "epoch": 0.311233060813081,
      "grad_norm": 1.4355252981185913,
      "learning_rate": 2.7775e-05,
      "loss": 0.5503,
      "step": 5558
    },
    {
      "epoch": 0.31128905812521,
      "grad_norm": 1.160586953163147,
      "learning_rate": 2.778e-05,
      "loss": 0.3731,
      "step": 5559
    },
    {
      "epoch": 0.311345055437339,
      "grad_norm": 1.2258049249649048,
      "learning_rate": 2.7785e-05,
      "loss": 0.3638,
      "step": 5560
    },
    {
      "epoch": 0.311401052749468,
      "grad_norm": 1.42200767993927,
      "learning_rate": 2.779e-05,
      "loss": 0.4598,
      "step": 5561
    },
    {
      "epoch": 0.311457050061597,
      "grad_norm": 1.2269688844680786,
      "learning_rate": 2.7794999999999997e-05,
      "loss": 0.3328,
      "step": 5562
    },
    {
      "epoch": 0.31151304737372604,
      "grad_norm": 1.2529104948043823,
      "learning_rate": 2.7800000000000005e-05,
      "loss": 0.5513,
      "step": 5563
    },
    {
      "epoch": 0.31156904468585506,
      "grad_norm": 1.2473069429397583,
      "learning_rate": 2.7805000000000002e-05,
      "loss": 0.5143,
      "step": 5564
    },
    {
      "epoch": 0.3116250419979841,
      "grad_norm": 1.274416446685791,
      "learning_rate": 2.7810000000000003e-05,
      "loss": 0.4876,
      "step": 5565
    },
    {
      "epoch": 0.3116810393101131,
      "grad_norm": 1.2429513931274414,
      "learning_rate": 2.7815e-05,
      "loss": 0.4585,
      "step": 5566
    },
    {
      "epoch": 0.3117370366222421,
      "grad_norm": 1.2814960479736328,
      "learning_rate": 2.782e-05,
      "loss": 0.4002,
      "step": 5567
    },
    {
      "epoch": 0.31179303393437113,
      "grad_norm": 1.2824534177780151,
      "learning_rate": 2.7825000000000002e-05,
      "loss": 0.4641,
      "step": 5568
    },
    {
      "epoch": 0.31184903124650015,
      "grad_norm": 1.194035291671753,
      "learning_rate": 2.783e-05,
      "loss": 0.4814,
      "step": 5569
    },
    {
      "epoch": 0.31190502855862917,
      "grad_norm": 1.5696885585784912,
      "learning_rate": 2.7835e-05,
      "loss": 0.5196,
      "step": 5570
    },
    {
      "epoch": 0.3119610258707582,
      "grad_norm": 1.2227907180786133,
      "learning_rate": 2.7839999999999998e-05,
      "loss": 0.3883,
      "step": 5571
    },
    {
      "epoch": 0.3120170231828872,
      "grad_norm": 0.9320754408836365,
      "learning_rate": 2.7845e-05,
      "loss": 0.3771,
      "step": 5572
    },
    {
      "epoch": 0.31207302049501623,
      "grad_norm": 1.4009031057357788,
      "learning_rate": 2.7850000000000003e-05,
      "loss": 0.3545,
      "step": 5573
    },
    {
      "epoch": 0.31212901780714525,
      "grad_norm": 1.3426100015640259,
      "learning_rate": 2.7855000000000004e-05,
      "loss": 0.4465,
      "step": 5574
    },
    {
      "epoch": 0.31218501511927427,
      "grad_norm": 1.4786632061004639,
      "learning_rate": 2.7860000000000004e-05,
      "loss": 0.4563,
      "step": 5575
    },
    {
      "epoch": 0.3122410124314033,
      "grad_norm": 1.123327612876892,
      "learning_rate": 2.7865000000000002e-05,
      "loss": 0.4662,
      "step": 5576
    },
    {
      "epoch": 0.3122970097435323,
      "grad_norm": 1.3635311126708984,
      "learning_rate": 2.7870000000000003e-05,
      "loss": 0.5339,
      "step": 5577
    },
    {
      "epoch": 0.3123530070556613,
      "grad_norm": 7.306573390960693,
      "learning_rate": 2.7875e-05,
      "loss": 0.4075,
      "step": 5578
    },
    {
      "epoch": 0.31240900436779034,
      "grad_norm": 1.197943925857544,
      "learning_rate": 2.788e-05,
      "loss": 0.456,
      "step": 5579
    },
    {
      "epoch": 0.31246500167991936,
      "grad_norm": 1.428166151046753,
      "learning_rate": 2.7885e-05,
      "loss": 0.3866,
      "step": 5580
    },
    {
      "epoch": 0.3125209989920484,
      "grad_norm": 1.0873808860778809,
      "learning_rate": 2.789e-05,
      "loss": 0.3884,
      "step": 5581
    },
    {
      "epoch": 0.3125769963041774,
      "grad_norm": 1.3680000305175781,
      "learning_rate": 2.7895e-05,
      "loss": 0.4507,
      "step": 5582
    },
    {
      "epoch": 0.3126329936163064,
      "grad_norm": 1.1258691549301147,
      "learning_rate": 2.7900000000000004e-05,
      "loss": 0.4959,
      "step": 5583
    },
    {
      "epoch": 0.31268899092843544,
      "grad_norm": 1.0404407978057861,
      "learning_rate": 2.7905000000000005e-05,
      "loss": 0.4571,
      "step": 5584
    },
    {
      "epoch": 0.31274498824056446,
      "grad_norm": 1.3987343311309814,
      "learning_rate": 2.7910000000000002e-05,
      "loss": 0.4018,
      "step": 5585
    },
    {
      "epoch": 0.3128009855526935,
      "grad_norm": 1.401898980140686,
      "learning_rate": 2.7915000000000003e-05,
      "loss": 0.3769,
      "step": 5586
    },
    {
      "epoch": 0.3128569828648225,
      "grad_norm": 1.24025559425354,
      "learning_rate": 2.792e-05,
      "loss": 0.401,
      "step": 5587
    },
    {
      "epoch": 0.3129129801769515,
      "grad_norm": 1.172914743423462,
      "learning_rate": 2.7925e-05,
      "loss": 0.4351,
      "step": 5588
    },
    {
      "epoch": 0.31296897748908054,
      "grad_norm": 1.3653998374938965,
      "learning_rate": 2.7930000000000002e-05,
      "loss": 0.4432,
      "step": 5589
    },
    {
      "epoch": 0.31302497480120955,
      "grad_norm": 1.4646999835968018,
      "learning_rate": 2.7935e-05,
      "loss": 0.5552,
      "step": 5590
    },
    {
      "epoch": 0.3130809721133386,
      "grad_norm": 1.4121801853179932,
      "learning_rate": 2.794e-05,
      "loss": 0.391,
      "step": 5591
    },
    {
      "epoch": 0.3131369694254676,
      "grad_norm": 1.7955048084259033,
      "learning_rate": 2.7944999999999998e-05,
      "loss": 0.53,
      "step": 5592
    },
    {
      "epoch": 0.3131929667375966,
      "grad_norm": 1.044741153717041,
      "learning_rate": 2.7950000000000005e-05,
      "loss": 0.401,
      "step": 5593
    },
    {
      "epoch": 0.31324896404972563,
      "grad_norm": 1.3014229536056519,
      "learning_rate": 2.7955000000000003e-05,
      "loss": 0.4103,
      "step": 5594
    },
    {
      "epoch": 0.31330496136185465,
      "grad_norm": 1.2207649946212769,
      "learning_rate": 2.7960000000000003e-05,
      "loss": 0.4124,
      "step": 5595
    },
    {
      "epoch": 0.31336095867398367,
      "grad_norm": 1.241531252861023,
      "learning_rate": 2.7965e-05,
      "loss": 0.3804,
      "step": 5596
    },
    {
      "epoch": 0.3134169559861127,
      "grad_norm": 1.4336752891540527,
      "learning_rate": 2.797e-05,
      "loss": 0.4088,
      "step": 5597
    },
    {
      "epoch": 0.3134729532982417,
      "grad_norm": 1.685155987739563,
      "learning_rate": 2.7975000000000002e-05,
      "loss": 0.5316,
      "step": 5598
    },
    {
      "epoch": 0.3135289506103707,
      "grad_norm": 1.1184717416763306,
      "learning_rate": 2.798e-05,
      "loss": 0.3776,
      "step": 5599
    },
    {
      "epoch": 0.31358494792249975,
      "grad_norm": 1.1784001588821411,
      "learning_rate": 2.7985e-05,
      "loss": 0.3587,
      "step": 5600
    },
    {
      "epoch": 0.31364094523462877,
      "grad_norm": 1.4661425352096558,
      "learning_rate": 2.7989999999999998e-05,
      "loss": 0.628,
      "step": 5601
    },
    {
      "epoch": 0.31369694254675773,
      "grad_norm": 1.212763786315918,
      "learning_rate": 2.7995e-05,
      "loss": 0.3783,
      "step": 5602
    },
    {
      "epoch": 0.31375293985888675,
      "grad_norm": 1.6633150577545166,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.4354,
      "step": 5603
    },
    {
      "epoch": 0.31380893717101577,
      "grad_norm": 1.237869381904602,
      "learning_rate": 2.8005000000000004e-05,
      "loss": 0.4162,
      "step": 5604
    },
    {
      "epoch": 0.3138649344831448,
      "grad_norm": 1.1660290956497192,
      "learning_rate": 2.8010000000000005e-05,
      "loss": 0.3373,
      "step": 5605
    },
    {
      "epoch": 0.3139209317952738,
      "grad_norm": 1.2090556621551514,
      "learning_rate": 2.8015000000000002e-05,
      "loss": 0.5203,
      "step": 5606
    },
    {
      "epoch": 0.3139769291074028,
      "grad_norm": 1.4198402166366577,
      "learning_rate": 2.8020000000000003e-05,
      "loss": 0.4898,
      "step": 5607
    },
    {
      "epoch": 0.31403292641953184,
      "grad_norm": 1.1494313478469849,
      "learning_rate": 2.8025e-05,
      "loss": 0.4086,
      "step": 5608
    },
    {
      "epoch": 0.31408892373166086,
      "grad_norm": 1.2049280405044556,
      "learning_rate": 2.803e-05,
      "loss": 0.3833,
      "step": 5609
    },
    {
      "epoch": 0.3141449210437899,
      "grad_norm": 1.3241567611694336,
      "learning_rate": 2.8035000000000002e-05,
      "loss": 0.5409,
      "step": 5610
    },
    {
      "epoch": 0.3142009183559189,
      "grad_norm": 1.1381441354751587,
      "learning_rate": 2.804e-05,
      "loss": 0.416,
      "step": 5611
    },
    {
      "epoch": 0.3142569156680479,
      "grad_norm": 1.4359488487243652,
      "learning_rate": 2.8045e-05,
      "loss": 0.4532,
      "step": 5612
    },
    {
      "epoch": 0.31431291298017694,
      "grad_norm": 1.5256915092468262,
      "learning_rate": 2.8050000000000004e-05,
      "loss": 0.5876,
      "step": 5613
    },
    {
      "epoch": 0.31436891029230596,
      "grad_norm": 1.111158847808838,
      "learning_rate": 2.8055000000000005e-05,
      "loss": 0.3589,
      "step": 5614
    },
    {
      "epoch": 0.314424907604435,
      "grad_norm": 1.328284740447998,
      "learning_rate": 2.8060000000000002e-05,
      "loss": 0.4164,
      "step": 5615
    },
    {
      "epoch": 0.314480904916564,
      "grad_norm": 1.2936842441558838,
      "learning_rate": 2.8065000000000003e-05,
      "loss": 0.6145,
      "step": 5616
    },
    {
      "epoch": 0.314536902228693,
      "grad_norm": 1.5181560516357422,
      "learning_rate": 2.807e-05,
      "loss": 0.4223,
      "step": 5617
    },
    {
      "epoch": 0.31459289954082204,
      "grad_norm": 1.1503472328186035,
      "learning_rate": 2.8075e-05,
      "loss": 0.4605,
      "step": 5618
    },
    {
      "epoch": 0.31464889685295105,
      "grad_norm": 1.248275637626648,
      "learning_rate": 2.8080000000000002e-05,
      "loss": 0.3591,
      "step": 5619
    },
    {
      "epoch": 0.3147048941650801,
      "grad_norm": 1.1627516746520996,
      "learning_rate": 2.8085e-05,
      "loss": 0.3392,
      "step": 5620
    },
    {
      "epoch": 0.3147608914772091,
      "grad_norm": 1.1381075382232666,
      "learning_rate": 2.809e-05,
      "loss": 0.4387,
      "step": 5621
    },
    {
      "epoch": 0.3148168887893381,
      "grad_norm": 1.2067393064498901,
      "learning_rate": 2.8094999999999998e-05,
      "loss": 0.4865,
      "step": 5622
    },
    {
      "epoch": 0.31487288610146713,
      "grad_norm": 1.7616690397262573,
      "learning_rate": 2.8100000000000005e-05,
      "loss": 0.4576,
      "step": 5623
    },
    {
      "epoch": 0.31492888341359615,
      "grad_norm": 1.2296456098556519,
      "learning_rate": 2.8105000000000003e-05,
      "loss": 0.403,
      "step": 5624
    },
    {
      "epoch": 0.31498488072572517,
      "grad_norm": 1.0594286918640137,
      "learning_rate": 2.8110000000000004e-05,
      "loss": 0.4521,
      "step": 5625
    },
    {
      "epoch": 0.3150408780378542,
      "grad_norm": 1.0713024139404297,
      "learning_rate": 2.8115e-05,
      "loss": 0.3054,
      "step": 5626
    },
    {
      "epoch": 0.3150968753499832,
      "grad_norm": 1.0795053243637085,
      "learning_rate": 2.8120000000000002e-05,
      "loss": 0.3588,
      "step": 5627
    },
    {
      "epoch": 0.3151528726621122,
      "grad_norm": 1.4612751007080078,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 0.5633,
      "step": 5628
    },
    {
      "epoch": 0.31520886997424125,
      "grad_norm": 1.1460578441619873,
      "learning_rate": 2.813e-05,
      "loss": 0.3124,
      "step": 5629
    },
    {
      "epoch": 0.31526486728637027,
      "grad_norm": 1.154659628868103,
      "learning_rate": 2.8135e-05,
      "loss": 0.4332,
      "step": 5630
    },
    {
      "epoch": 0.3153208645984993,
      "grad_norm": 1.8693031072616577,
      "learning_rate": 2.8139999999999998e-05,
      "loss": 0.5789,
      "step": 5631
    },
    {
      "epoch": 0.3153768619106283,
      "grad_norm": 1.6017730236053467,
      "learning_rate": 2.8145e-05,
      "loss": 0.473,
      "step": 5632
    },
    {
      "epoch": 0.3154328592227573,
      "grad_norm": 4.376621246337891,
      "learning_rate": 2.815e-05,
      "loss": 0.4659,
      "step": 5633
    },
    {
      "epoch": 0.31548885653488634,
      "grad_norm": 1.125675916671753,
      "learning_rate": 2.8155000000000004e-05,
      "loss": 0.4019,
      "step": 5634
    },
    {
      "epoch": 0.31554485384701536,
      "grad_norm": 1.2513413429260254,
      "learning_rate": 2.816e-05,
      "loss": 0.4662,
      "step": 5635
    },
    {
      "epoch": 0.3156008511591444,
      "grad_norm": 0.9438375234603882,
      "learning_rate": 2.8165000000000002e-05,
      "loss": 0.3095,
      "step": 5636
    },
    {
      "epoch": 0.3156568484712734,
      "grad_norm": 1.113909363746643,
      "learning_rate": 2.8170000000000003e-05,
      "loss": 0.5185,
      "step": 5637
    },
    {
      "epoch": 0.3157128457834024,
      "grad_norm": 1.4468607902526855,
      "learning_rate": 2.8175e-05,
      "loss": 0.4859,
      "step": 5638
    },
    {
      "epoch": 0.31576884309553144,
      "grad_norm": 1.0653977394104004,
      "learning_rate": 2.818e-05,
      "loss": 0.3885,
      "step": 5639
    },
    {
      "epoch": 0.31582484040766046,
      "grad_norm": 1.389087438583374,
      "learning_rate": 2.8185e-05,
      "loss": 0.3641,
      "step": 5640
    },
    {
      "epoch": 0.3158808377197895,
      "grad_norm": 1.0898454189300537,
      "learning_rate": 2.819e-05,
      "loss": 0.3619,
      "step": 5641
    },
    {
      "epoch": 0.3159368350319185,
      "grad_norm": 1.2313554286956787,
      "learning_rate": 2.8195e-05,
      "loss": 0.3261,
      "step": 5642
    },
    {
      "epoch": 0.31599283234404746,
      "grad_norm": 1.072247862815857,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 0.375,
      "step": 5643
    },
    {
      "epoch": 0.3160488296561765,
      "grad_norm": 1.2600675821304321,
      "learning_rate": 2.8205000000000005e-05,
      "loss": 0.3451,
      "step": 5644
    },
    {
      "epoch": 0.3161048269683055,
      "grad_norm": 1.0342261791229248,
      "learning_rate": 2.8210000000000003e-05,
      "loss": 0.3224,
      "step": 5645
    },
    {
      "epoch": 0.3161608242804345,
      "grad_norm": 1.348740816116333,
      "learning_rate": 2.8215000000000003e-05,
      "loss": 0.5904,
      "step": 5646
    },
    {
      "epoch": 0.31621682159256354,
      "grad_norm": 1.1729305982589722,
      "learning_rate": 2.822e-05,
      "loss": 0.3687,
      "step": 5647
    },
    {
      "epoch": 0.31627281890469255,
      "grad_norm": 1.4918018579483032,
      "learning_rate": 2.8225e-05,
      "loss": 0.5176,
      "step": 5648
    },
    {
      "epoch": 0.3163288162168216,
      "grad_norm": 1.546281337738037,
      "learning_rate": 2.8230000000000002e-05,
      "loss": 0.5422,
      "step": 5649
    },
    {
      "epoch": 0.3163848135289506,
      "grad_norm": 1.0162818431854248,
      "learning_rate": 2.8235e-05,
      "loss": 0.4217,
      "step": 5650
    },
    {
      "epoch": 0.3164408108410796,
      "grad_norm": 1.5955443382263184,
      "learning_rate": 2.824e-05,
      "loss": 0.4193,
      "step": 5651
    },
    {
      "epoch": 0.31649680815320863,
      "grad_norm": 1.1460846662521362,
      "learning_rate": 2.8244999999999998e-05,
      "loss": 0.3683,
      "step": 5652
    },
    {
      "epoch": 0.31655280546533765,
      "grad_norm": 1.2553859949111938,
      "learning_rate": 2.825e-05,
      "loss": 0.5352,
      "step": 5653
    },
    {
      "epoch": 0.31660880277746667,
      "grad_norm": 1.315063714981079,
      "learning_rate": 2.8255000000000003e-05,
      "loss": 0.3756,
      "step": 5654
    },
    {
      "epoch": 0.3166648000895957,
      "grad_norm": 1.206869125366211,
      "learning_rate": 2.8260000000000004e-05,
      "loss": 0.452,
      "step": 5655
    },
    {
      "epoch": 0.3167207974017247,
      "grad_norm": 1.225957989692688,
      "learning_rate": 2.8265e-05,
      "loss": 0.539,
      "step": 5656
    },
    {
      "epoch": 0.3167767947138537,
      "grad_norm": 1.1983261108398438,
      "learning_rate": 2.8270000000000002e-05,
      "loss": 0.4334,
      "step": 5657
    },
    {
      "epoch": 0.31683279202598275,
      "grad_norm": 1.1372549533843994,
      "learning_rate": 2.8275000000000003e-05,
      "loss": 0.2428,
      "step": 5658
    },
    {
      "epoch": 0.31688878933811176,
      "grad_norm": 1.3845407962799072,
      "learning_rate": 2.828e-05,
      "loss": 0.4571,
      "step": 5659
    },
    {
      "epoch": 0.3169447866502408,
      "grad_norm": 1.270957350730896,
      "learning_rate": 2.8285e-05,
      "loss": 0.3377,
      "step": 5660
    },
    {
      "epoch": 0.3170007839623698,
      "grad_norm": 1.497532606124878,
      "learning_rate": 2.829e-05,
      "loss": 0.3983,
      "step": 5661
    },
    {
      "epoch": 0.3170567812744988,
      "grad_norm": 1.2636052370071411,
      "learning_rate": 2.8295e-05,
      "loss": 0.5258,
      "step": 5662
    },
    {
      "epoch": 0.31711277858662784,
      "grad_norm": 1.3569202423095703,
      "learning_rate": 2.83e-05,
      "loss": 0.5553,
      "step": 5663
    },
    {
      "epoch": 0.31716877589875686,
      "grad_norm": 1.2245830297470093,
      "learning_rate": 2.8305000000000004e-05,
      "loss": 0.3707,
      "step": 5664
    },
    {
      "epoch": 0.3172247732108859,
      "grad_norm": 1.4354572296142578,
      "learning_rate": 2.8310000000000002e-05,
      "loss": 0.4876,
      "step": 5665
    },
    {
      "epoch": 0.3172807705230149,
      "grad_norm": 1.2246136665344238,
      "learning_rate": 2.8315000000000002e-05,
      "loss": 0.5164,
      "step": 5666
    },
    {
      "epoch": 0.3173367678351439,
      "grad_norm": 1.0793654918670654,
      "learning_rate": 2.8320000000000003e-05,
      "loss": 0.3274,
      "step": 5667
    },
    {
      "epoch": 0.31739276514727294,
      "grad_norm": 1.2143782377243042,
      "learning_rate": 2.8325e-05,
      "loss": 0.5446,
      "step": 5668
    },
    {
      "epoch": 0.31744876245940196,
      "grad_norm": 1.2183564901351929,
      "learning_rate": 2.833e-05,
      "loss": 0.495,
      "step": 5669
    },
    {
      "epoch": 0.317504759771531,
      "grad_norm": 1.2809075117111206,
      "learning_rate": 2.8335e-05,
      "loss": 0.487,
      "step": 5670
    },
    {
      "epoch": 0.31756075708366,
      "grad_norm": 1.1521151065826416,
      "learning_rate": 2.834e-05,
      "loss": 0.4489,
      "step": 5671
    },
    {
      "epoch": 0.317616754395789,
      "grad_norm": 1.0466850996017456,
      "learning_rate": 2.8345e-05,
      "loss": 0.3761,
      "step": 5672
    },
    {
      "epoch": 0.31767275170791803,
      "grad_norm": 1.2535468339920044,
      "learning_rate": 2.8349999999999998e-05,
      "loss": 0.3309,
      "step": 5673
    },
    {
      "epoch": 0.31772874902004705,
      "grad_norm": 1.3420789241790771,
      "learning_rate": 2.8355000000000002e-05,
      "loss": 0.4376,
      "step": 5674
    },
    {
      "epoch": 0.31778474633217607,
      "grad_norm": 1.7165738344192505,
      "learning_rate": 2.8360000000000003e-05,
      "loss": 0.4187,
      "step": 5675
    },
    {
      "epoch": 0.3178407436443051,
      "grad_norm": 1.2799228429794312,
      "learning_rate": 2.8365000000000004e-05,
      "loss": 0.5057,
      "step": 5676
    },
    {
      "epoch": 0.3178967409564341,
      "grad_norm": 1.2187089920043945,
      "learning_rate": 2.837e-05,
      "loss": 0.4693,
      "step": 5677
    },
    {
      "epoch": 0.31795273826856313,
      "grad_norm": 1.1975899934768677,
      "learning_rate": 2.8375000000000002e-05,
      "loss": 0.5128,
      "step": 5678
    },
    {
      "epoch": 0.31800873558069215,
      "grad_norm": 1.2510820627212524,
      "learning_rate": 2.8380000000000003e-05,
      "loss": 0.3987,
      "step": 5679
    },
    {
      "epoch": 0.31806473289282117,
      "grad_norm": 1.257769227027893,
      "learning_rate": 2.8385e-05,
      "loss": 0.4245,
      "step": 5680
    },
    {
      "epoch": 0.3181207302049502,
      "grad_norm": 1.138027310371399,
      "learning_rate": 2.839e-05,
      "loss": 0.4206,
      "step": 5681
    },
    {
      "epoch": 0.3181767275170792,
      "grad_norm": 1.360088586807251,
      "learning_rate": 2.8395e-05,
      "loss": 0.6143,
      "step": 5682
    },
    {
      "epoch": 0.3182327248292082,
      "grad_norm": 1.3086761236190796,
      "learning_rate": 2.84e-05,
      "loss": 0.4096,
      "step": 5683
    },
    {
      "epoch": 0.3182887221413372,
      "grad_norm": 1.507885217666626,
      "learning_rate": 2.8405000000000003e-05,
      "loss": 0.4123,
      "step": 5684
    },
    {
      "epoch": 0.3183447194534662,
      "grad_norm": 1.4379240274429321,
      "learning_rate": 2.8410000000000004e-05,
      "loss": 0.3721,
      "step": 5685
    },
    {
      "epoch": 0.3184007167655952,
      "grad_norm": 1.6149991750717163,
      "learning_rate": 2.8415e-05,
      "loss": 0.4733,
      "step": 5686
    },
    {
      "epoch": 0.31845671407772425,
      "grad_norm": 1.240517497062683,
      "learning_rate": 2.8420000000000002e-05,
      "loss": 0.4846,
      "step": 5687
    },
    {
      "epoch": 0.31851271138985326,
      "grad_norm": 1.319047451019287,
      "learning_rate": 2.8425000000000003e-05,
      "loss": 0.3765,
      "step": 5688
    },
    {
      "epoch": 0.3185687087019823,
      "grad_norm": 1.7602545022964478,
      "learning_rate": 2.843e-05,
      "loss": 0.476,
      "step": 5689
    },
    {
      "epoch": 0.3186247060141113,
      "grad_norm": 1.3483355045318604,
      "learning_rate": 2.8435e-05,
      "loss": 0.3854,
      "step": 5690
    },
    {
      "epoch": 0.3186807033262403,
      "grad_norm": 1.6905882358551025,
      "learning_rate": 2.844e-05,
      "loss": 0.5992,
      "step": 5691
    },
    {
      "epoch": 0.31873670063836934,
      "grad_norm": 1.2765854597091675,
      "learning_rate": 2.8445e-05,
      "loss": 0.4451,
      "step": 5692
    },
    {
      "epoch": 0.31879269795049836,
      "grad_norm": 1.7190817594528198,
      "learning_rate": 2.845e-05,
      "loss": 0.5721,
      "step": 5693
    },
    {
      "epoch": 0.3188486952626274,
      "grad_norm": 1.3409048318862915,
      "learning_rate": 2.8455000000000005e-05,
      "loss": 0.6477,
      "step": 5694
    },
    {
      "epoch": 0.3189046925747564,
      "grad_norm": 1.4718717336654663,
      "learning_rate": 2.8460000000000002e-05,
      "loss": 0.5517,
      "step": 5695
    },
    {
      "epoch": 0.3189606898868854,
      "grad_norm": 1.3900972604751587,
      "learning_rate": 2.8465000000000003e-05,
      "loss": 0.5744,
      "step": 5696
    },
    {
      "epoch": 0.31901668719901444,
      "grad_norm": 1.1296007633209229,
      "learning_rate": 2.8470000000000004e-05,
      "loss": 0.3525,
      "step": 5697
    },
    {
      "epoch": 0.31907268451114346,
      "grad_norm": 1.5070109367370605,
      "learning_rate": 2.8475e-05,
      "loss": 0.6085,
      "step": 5698
    },
    {
      "epoch": 0.3191286818232725,
      "grad_norm": 1.1078076362609863,
      "learning_rate": 2.8480000000000002e-05,
      "loss": 0.4084,
      "step": 5699
    },
    {
      "epoch": 0.3191846791354015,
      "grad_norm": 1.2499613761901855,
      "learning_rate": 2.8485e-05,
      "loss": 0.3357,
      "step": 5700
    },
    {
      "epoch": 0.3192406764475305,
      "grad_norm": 1.331945538520813,
      "learning_rate": 2.849e-05,
      "loss": 0.5819,
      "step": 5701
    },
    {
      "epoch": 0.31929667375965953,
      "grad_norm": 1.3197909593582153,
      "learning_rate": 2.8495e-05,
      "loss": 0.3637,
      "step": 5702
    },
    {
      "epoch": 0.31935267107178855,
      "grad_norm": 1.1270290613174438,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 0.4476,
      "step": 5703
    },
    {
      "epoch": 0.31940866838391757,
      "grad_norm": 3.274280309677124,
      "learning_rate": 2.8505000000000002e-05,
      "loss": 0.4659,
      "step": 5704
    },
    {
      "epoch": 0.3194646656960466,
      "grad_norm": 1.2685467004776,
      "learning_rate": 2.8510000000000003e-05,
      "loss": 0.3911,
      "step": 5705
    },
    {
      "epoch": 0.3195206630081756,
      "grad_norm": 1.19549560546875,
      "learning_rate": 2.8515000000000004e-05,
      "loss": 0.4517,
      "step": 5706
    },
    {
      "epoch": 0.31957666032030463,
      "grad_norm": 1.320735216140747,
      "learning_rate": 2.852e-05,
      "loss": 0.4496,
      "step": 5707
    },
    {
      "epoch": 0.31963265763243365,
      "grad_norm": 1.380415678024292,
      "learning_rate": 2.8525000000000002e-05,
      "loss": 0.4642,
      "step": 5708
    },
    {
      "epoch": 0.31968865494456267,
      "grad_norm": 1.2442623376846313,
      "learning_rate": 2.853e-05,
      "loss": 0.465,
      "step": 5709
    },
    {
      "epoch": 0.3197446522566917,
      "grad_norm": 1.3137507438659668,
      "learning_rate": 2.8535e-05,
      "loss": 0.4749,
      "step": 5710
    },
    {
      "epoch": 0.3198006495688207,
      "grad_norm": 1.3102186918258667,
      "learning_rate": 2.854e-05,
      "loss": 0.4644,
      "step": 5711
    },
    {
      "epoch": 0.3198566468809497,
      "grad_norm": 1.0463544130325317,
      "learning_rate": 2.8545e-05,
      "loss": 0.3313,
      "step": 5712
    },
    {
      "epoch": 0.31991264419307874,
      "grad_norm": 1.3802485466003418,
      "learning_rate": 2.855e-05,
      "loss": 0.4341,
      "step": 5713
    },
    {
      "epoch": 0.31996864150520776,
      "grad_norm": 3.6623709201812744,
      "learning_rate": 2.8555000000000004e-05,
      "loss": 0.5099,
      "step": 5714
    },
    {
      "epoch": 0.3200246388173368,
      "grad_norm": 1.4516994953155518,
      "learning_rate": 2.8560000000000004e-05,
      "loss": 0.4278,
      "step": 5715
    },
    {
      "epoch": 0.3200806361294658,
      "grad_norm": 1.276397466659546,
      "learning_rate": 2.8565000000000002e-05,
      "loss": 0.4023,
      "step": 5716
    },
    {
      "epoch": 0.3201366334415948,
      "grad_norm": 1.2371009588241577,
      "learning_rate": 2.8570000000000003e-05,
      "loss": 0.445,
      "step": 5717
    },
    {
      "epoch": 0.32019263075372384,
      "grad_norm": 1.18202543258667,
      "learning_rate": 2.8575000000000003e-05,
      "loss": 0.5315,
      "step": 5718
    },
    {
      "epoch": 0.32024862806585286,
      "grad_norm": 1.3691107034683228,
      "learning_rate": 2.858e-05,
      "loss": 0.3928,
      "step": 5719
    },
    {
      "epoch": 0.3203046253779819,
      "grad_norm": 1.5681381225585938,
      "learning_rate": 2.8585e-05,
      "loss": 0.4085,
      "step": 5720
    },
    {
      "epoch": 0.3203606226901109,
      "grad_norm": 1.030142068862915,
      "learning_rate": 2.859e-05,
      "loss": 0.3923,
      "step": 5721
    },
    {
      "epoch": 0.3204166200022399,
      "grad_norm": 1.2838459014892578,
      "learning_rate": 2.8595e-05,
      "loss": 0.5295,
      "step": 5722
    },
    {
      "epoch": 0.32047261731436893,
      "grad_norm": 1.3073571920394897,
      "learning_rate": 2.86e-05,
      "loss": 0.4917,
      "step": 5723
    },
    {
      "epoch": 0.32052861462649795,
      "grad_norm": 1.3806993961334229,
      "learning_rate": 2.8605000000000005e-05,
      "loss": 0.5803,
      "step": 5724
    },
    {
      "epoch": 0.320584611938627,
      "grad_norm": 1.4577785730361938,
      "learning_rate": 2.8610000000000002e-05,
      "loss": 0.4319,
      "step": 5725
    },
    {
      "epoch": 0.32064060925075594,
      "grad_norm": 1.2400891780853271,
      "learning_rate": 2.8615000000000003e-05,
      "loss": 0.5681,
      "step": 5726
    },
    {
      "epoch": 0.32069660656288496,
      "grad_norm": 1.2736188173294067,
      "learning_rate": 2.8620000000000004e-05,
      "loss": 0.496,
      "step": 5727
    },
    {
      "epoch": 0.320752603875014,
      "grad_norm": 1.7031136751174927,
      "learning_rate": 2.8625e-05,
      "loss": 0.314,
      "step": 5728
    },
    {
      "epoch": 0.320808601187143,
      "grad_norm": 1.4622594118118286,
      "learning_rate": 2.8630000000000002e-05,
      "loss": 0.4218,
      "step": 5729
    },
    {
      "epoch": 0.320864598499272,
      "grad_norm": 1.282241702079773,
      "learning_rate": 2.8635e-05,
      "loss": 0.5403,
      "step": 5730
    },
    {
      "epoch": 0.32092059581140103,
      "grad_norm": 1.379214882850647,
      "learning_rate": 2.864e-05,
      "loss": 0.3999,
      "step": 5731
    },
    {
      "epoch": 0.32097659312353005,
      "grad_norm": 1.029502272605896,
      "learning_rate": 2.8645e-05,
      "loss": 0.3157,
      "step": 5732
    },
    {
      "epoch": 0.32103259043565907,
      "grad_norm": 1.2138969898223877,
      "learning_rate": 2.865e-05,
      "loss": 0.4902,
      "step": 5733
    },
    {
      "epoch": 0.3210885877477881,
      "grad_norm": 1.1390117406845093,
      "learning_rate": 2.8655000000000003e-05,
      "loss": 0.3516,
      "step": 5734
    },
    {
      "epoch": 0.3211445850599171,
      "grad_norm": 1.102679967880249,
      "learning_rate": 2.8660000000000003e-05,
      "loss": 0.4028,
      "step": 5735
    },
    {
      "epoch": 0.32120058237204613,
      "grad_norm": 1.5025776624679565,
      "learning_rate": 2.8665000000000004e-05,
      "loss": 0.5073,
      "step": 5736
    },
    {
      "epoch": 0.32125657968417515,
      "grad_norm": 2.1668403148651123,
      "learning_rate": 2.867e-05,
      "loss": 0.4717,
      "step": 5737
    },
    {
      "epoch": 0.32131257699630417,
      "grad_norm": 1.0500065088272095,
      "learning_rate": 2.8675000000000002e-05,
      "loss": 0.4076,
      "step": 5738
    },
    {
      "epoch": 0.3213685743084332,
      "grad_norm": 1.29715096950531,
      "learning_rate": 2.868e-05,
      "loss": 0.5435,
      "step": 5739
    },
    {
      "epoch": 0.3214245716205622,
      "grad_norm": 2.245598316192627,
      "learning_rate": 2.8685e-05,
      "loss": 0.4601,
      "step": 5740
    },
    {
      "epoch": 0.3214805689326912,
      "grad_norm": 1.4639257192611694,
      "learning_rate": 2.869e-05,
      "loss": 0.3894,
      "step": 5741
    },
    {
      "epoch": 0.32153656624482024,
      "grad_norm": 1.1997876167297363,
      "learning_rate": 2.8695e-05,
      "loss": 0.3474,
      "step": 5742
    },
    {
      "epoch": 0.32159256355694926,
      "grad_norm": 1.290321707725525,
      "learning_rate": 2.87e-05,
      "loss": 0.4773,
      "step": 5743
    },
    {
      "epoch": 0.3216485608690783,
      "grad_norm": 1.3255780935287476,
      "learning_rate": 2.8705000000000004e-05,
      "loss": 0.54,
      "step": 5744
    },
    {
      "epoch": 0.3217045581812073,
      "grad_norm": 5.353636264801025,
      "learning_rate": 2.8710000000000005e-05,
      "loss": 0.4118,
      "step": 5745
    },
    {
      "epoch": 0.3217605554933363,
      "grad_norm": 1.1815154552459717,
      "learning_rate": 2.8715000000000002e-05,
      "loss": 0.327,
      "step": 5746
    },
    {
      "epoch": 0.32181655280546534,
      "grad_norm": 1.3262768983840942,
      "learning_rate": 2.8720000000000003e-05,
      "loss": 0.4177,
      "step": 5747
    },
    {
      "epoch": 0.32187255011759436,
      "grad_norm": 1.1953827142715454,
      "learning_rate": 2.8725e-05,
      "loss": 0.4005,
      "step": 5748
    },
    {
      "epoch": 0.3219285474297234,
      "grad_norm": 1.3174980878829956,
      "learning_rate": 2.873e-05,
      "loss": 0.4097,
      "step": 5749
    },
    {
      "epoch": 0.3219845447418524,
      "grad_norm": 1.2677258253097534,
      "learning_rate": 2.8735000000000002e-05,
      "loss": 0.3018,
      "step": 5750
    },
    {
      "epoch": 0.3220405420539814,
      "grad_norm": 1.2160332202911377,
      "learning_rate": 2.874e-05,
      "loss": 0.5729,
      "step": 5751
    },
    {
      "epoch": 0.32209653936611043,
      "grad_norm": 1.1369692087173462,
      "learning_rate": 2.8745e-05,
      "loss": 0.387,
      "step": 5752
    },
    {
      "epoch": 0.32215253667823945,
      "grad_norm": 1.3775033950805664,
      "learning_rate": 2.8749999999999997e-05,
      "loss": 0.5746,
      "step": 5753
    },
    {
      "epoch": 0.3222085339903685,
      "grad_norm": 1.3365129232406616,
      "learning_rate": 2.8754999999999998e-05,
      "loss": 0.4738,
      "step": 5754
    },
    {
      "epoch": 0.3222645313024975,
      "grad_norm": 1.5569610595703125,
      "learning_rate": 2.8760000000000002e-05,
      "loss": 0.4895,
      "step": 5755
    },
    {
      "epoch": 0.3223205286146265,
      "grad_norm": 1.3198693990707397,
      "learning_rate": 2.8765000000000003e-05,
      "loss": 0.3842,
      "step": 5756
    },
    {
      "epoch": 0.32237652592675553,
      "grad_norm": 1.2834845781326294,
      "learning_rate": 2.8770000000000004e-05,
      "loss": 0.4862,
      "step": 5757
    },
    {
      "epoch": 0.32243252323888455,
      "grad_norm": 1.2282882928848267,
      "learning_rate": 2.8775e-05,
      "loss": 0.3435,
      "step": 5758
    },
    {
      "epoch": 0.32248852055101357,
      "grad_norm": 1.120464563369751,
      "learning_rate": 2.8780000000000002e-05,
      "loss": 0.3643,
      "step": 5759
    },
    {
      "epoch": 0.3225445178631426,
      "grad_norm": 1.117025375366211,
      "learning_rate": 2.8785e-05,
      "loss": 0.4665,
      "step": 5760
    },
    {
      "epoch": 0.3226005151752716,
      "grad_norm": 1.0791250467300415,
      "learning_rate": 2.879e-05,
      "loss": 0.3406,
      "step": 5761
    },
    {
      "epoch": 0.3226565124874006,
      "grad_norm": 1.3383996486663818,
      "learning_rate": 2.8795e-05,
      "loss": 0.3864,
      "step": 5762
    },
    {
      "epoch": 0.32271250979952965,
      "grad_norm": 1.1708077192306519,
      "learning_rate": 2.88e-05,
      "loss": 0.3926,
      "step": 5763
    },
    {
      "epoch": 0.32276850711165866,
      "grad_norm": 1.342558741569519,
      "learning_rate": 2.8805e-05,
      "loss": 0.3435,
      "step": 5764
    },
    {
      "epoch": 0.3228245044237877,
      "grad_norm": 1.3170289993286133,
      "learning_rate": 2.8810000000000004e-05,
      "loss": 0.4149,
      "step": 5765
    },
    {
      "epoch": 0.3228805017359167,
      "grad_norm": 1.3614591360092163,
      "learning_rate": 2.8815000000000004e-05,
      "loss": 0.3713,
      "step": 5766
    },
    {
      "epoch": 0.32293649904804567,
      "grad_norm": 1.175022840499878,
      "learning_rate": 2.8820000000000002e-05,
      "loss": 0.4052,
      "step": 5767
    },
    {
      "epoch": 0.3229924963601747,
      "grad_norm": 0.9799365997314453,
      "learning_rate": 2.8825000000000003e-05,
      "loss": 0.2728,
      "step": 5768
    },
    {
      "epoch": 0.3230484936723037,
      "grad_norm": 1.7954214811325073,
      "learning_rate": 2.883e-05,
      "loss": 0.5044,
      "step": 5769
    },
    {
      "epoch": 0.3231044909844327,
      "grad_norm": 1.3334827423095703,
      "learning_rate": 2.8835e-05,
      "loss": 0.4983,
      "step": 5770
    },
    {
      "epoch": 0.32316048829656174,
      "grad_norm": 1.1819570064544678,
      "learning_rate": 2.8840000000000002e-05,
      "loss": 0.4125,
      "step": 5771
    },
    {
      "epoch": 0.32321648560869076,
      "grad_norm": 1.1033072471618652,
      "learning_rate": 2.8845e-05,
      "loss": 0.3211,
      "step": 5772
    },
    {
      "epoch": 0.3232724829208198,
      "grad_norm": 1.1614561080932617,
      "learning_rate": 2.885e-05,
      "loss": 0.5159,
      "step": 5773
    },
    {
      "epoch": 0.3233284802329488,
      "grad_norm": 1.4699541330337524,
      "learning_rate": 2.8854999999999997e-05,
      "loss": 0.6175,
      "step": 5774
    },
    {
      "epoch": 0.3233844775450778,
      "grad_norm": 0.9454047083854675,
      "learning_rate": 2.8860000000000005e-05,
      "loss": 0.3131,
      "step": 5775
    },
    {
      "epoch": 0.32344047485720684,
      "grad_norm": 1.3530412912368774,
      "learning_rate": 2.8865000000000002e-05,
      "loss": 0.4906,
      "step": 5776
    },
    {
      "epoch": 0.32349647216933586,
      "grad_norm": 1.1419742107391357,
      "learning_rate": 2.8870000000000003e-05,
      "loss": 0.4197,
      "step": 5777
    },
    {
      "epoch": 0.3235524694814649,
      "grad_norm": 1.2685377597808838,
      "learning_rate": 2.8875e-05,
      "loss": 0.4257,
      "step": 5778
    },
    {
      "epoch": 0.3236084667935939,
      "grad_norm": 1.2069966793060303,
      "learning_rate": 2.888e-05,
      "loss": 0.479,
      "step": 5779
    },
    {
      "epoch": 0.3236644641057229,
      "grad_norm": 1.2854137420654297,
      "learning_rate": 2.8885000000000002e-05,
      "loss": 0.2771,
      "step": 5780
    },
    {
      "epoch": 0.32372046141785193,
      "grad_norm": 1.4132953882217407,
      "learning_rate": 2.889e-05,
      "loss": 0.5198,
      "step": 5781
    },
    {
      "epoch": 0.32377645872998095,
      "grad_norm": 1.051746129989624,
      "learning_rate": 2.8895e-05,
      "loss": 0.3799,
      "step": 5782
    },
    {
      "epoch": 0.32383245604211,
      "grad_norm": 1.1334714889526367,
      "learning_rate": 2.8899999999999998e-05,
      "loss": 0.3857,
      "step": 5783
    },
    {
      "epoch": 0.323888453354239,
      "grad_norm": 1.6783074140548706,
      "learning_rate": 2.8905e-05,
      "loss": 0.5421,
      "step": 5784
    },
    {
      "epoch": 0.323944450666368,
      "grad_norm": 1.5858482122421265,
      "learning_rate": 2.8910000000000003e-05,
      "loss": 0.5914,
      "step": 5785
    },
    {
      "epoch": 0.32400044797849703,
      "grad_norm": 1.5772172212600708,
      "learning_rate": 2.8915000000000004e-05,
      "loss": 0.4384,
      "step": 5786
    },
    {
      "epoch": 0.32405644529062605,
      "grad_norm": 1.386919379234314,
      "learning_rate": 2.8920000000000004e-05,
      "loss": 0.5433,
      "step": 5787
    },
    {
      "epoch": 0.32411244260275507,
      "grad_norm": 1.2350701093673706,
      "learning_rate": 2.8925000000000002e-05,
      "loss": 0.2706,
      "step": 5788
    },
    {
      "epoch": 0.3241684399148841,
      "grad_norm": 1.2195590734481812,
      "learning_rate": 2.8930000000000003e-05,
      "loss": 0.4213,
      "step": 5789
    },
    {
      "epoch": 0.3242244372270131,
      "grad_norm": 1.4639583826065063,
      "learning_rate": 2.8935e-05,
      "loss": 0.5474,
      "step": 5790
    },
    {
      "epoch": 0.3242804345391421,
      "grad_norm": 1.1940836906433105,
      "learning_rate": 2.894e-05,
      "loss": 0.489,
      "step": 5791
    },
    {
      "epoch": 0.32433643185127115,
      "grad_norm": 1.021995186805725,
      "learning_rate": 2.8945e-05,
      "loss": 0.3867,
      "step": 5792
    },
    {
      "epoch": 0.32439242916340016,
      "grad_norm": 1.2153525352478027,
      "learning_rate": 2.895e-05,
      "loss": 0.3555,
      "step": 5793
    },
    {
      "epoch": 0.3244484264755292,
      "grad_norm": 1.187088966369629,
      "learning_rate": 2.8955e-05,
      "loss": 0.4573,
      "step": 5794
    },
    {
      "epoch": 0.3245044237876582,
      "grad_norm": 6.31917142868042,
      "learning_rate": 2.8960000000000004e-05,
      "loss": 0.3281,
      "step": 5795
    },
    {
      "epoch": 0.3245604210997872,
      "grad_norm": 1.2404654026031494,
      "learning_rate": 2.8965000000000005e-05,
      "loss": 0.372,
      "step": 5796
    },
    {
      "epoch": 0.32461641841191624,
      "grad_norm": 1.165102243423462,
      "learning_rate": 2.8970000000000002e-05,
      "loss": 0.3417,
      "step": 5797
    },
    {
      "epoch": 0.32467241572404526,
      "grad_norm": 1.2347089052200317,
      "learning_rate": 2.8975000000000003e-05,
      "loss": 0.4466,
      "step": 5798
    },
    {
      "epoch": 0.3247284130361743,
      "grad_norm": 1.1590970754623413,
      "learning_rate": 2.898e-05,
      "loss": 0.4122,
      "step": 5799
    },
    {
      "epoch": 0.3247844103483033,
      "grad_norm": 1.2014491558074951,
      "learning_rate": 2.8985e-05,
      "loss": 0.5527,
      "step": 5800
    },
    {
      "epoch": 0.3248404076604323,
      "grad_norm": 1.2441595792770386,
      "learning_rate": 2.8990000000000002e-05,
      "loss": 0.6121,
      "step": 5801
    },
    {
      "epoch": 0.32489640497256134,
      "grad_norm": 0.9242920875549316,
      "learning_rate": 2.8995e-05,
      "loss": 0.3175,
      "step": 5802
    },
    {
      "epoch": 0.32495240228469036,
      "grad_norm": 1.5692812204360962,
      "learning_rate": 2.9e-05,
      "loss": 0.6651,
      "step": 5803
    },
    {
      "epoch": 0.3250083995968194,
      "grad_norm": 1.4126709699630737,
      "learning_rate": 2.9004999999999998e-05,
      "loss": 0.3985,
      "step": 5804
    },
    {
      "epoch": 0.3250643969089484,
      "grad_norm": 1.0389025211334229,
      "learning_rate": 2.9010000000000005e-05,
      "loss": 0.3518,
      "step": 5805
    },
    {
      "epoch": 0.3251203942210774,
      "grad_norm": 1.4933347702026367,
      "learning_rate": 2.9015000000000003e-05,
      "loss": 0.4727,
      "step": 5806
    },
    {
      "epoch": 0.32517639153320643,
      "grad_norm": 1.28818941116333,
      "learning_rate": 2.9020000000000003e-05,
      "loss": 0.4388,
      "step": 5807
    },
    {
      "epoch": 0.3252323888453354,
      "grad_norm": 1.238250494003296,
      "learning_rate": 2.9025e-05,
      "loss": 0.5254,
      "step": 5808
    },
    {
      "epoch": 0.3252883861574644,
      "grad_norm": 5.895007610321045,
      "learning_rate": 2.903e-05,
      "loss": 0.3471,
      "step": 5809
    },
    {
      "epoch": 0.32534438346959343,
      "grad_norm": 1.2327629327774048,
      "learning_rate": 2.9035000000000002e-05,
      "loss": 0.3286,
      "step": 5810
    },
    {
      "epoch": 0.32540038078172245,
      "grad_norm": 1.2726235389709473,
      "learning_rate": 2.904e-05,
      "loss": 0.474,
      "step": 5811
    },
    {
      "epoch": 0.3254563780938515,
      "grad_norm": 1.2369911670684814,
      "learning_rate": 2.9045e-05,
      "loss": 0.4916,
      "step": 5812
    },
    {
      "epoch": 0.3255123754059805,
      "grad_norm": 1.1697537899017334,
      "learning_rate": 2.9049999999999998e-05,
      "loss": 0.4745,
      "step": 5813
    },
    {
      "epoch": 0.3255683727181095,
      "grad_norm": 1.4516698122024536,
      "learning_rate": 2.9055e-05,
      "loss": 0.4536,
      "step": 5814
    },
    {
      "epoch": 0.32562437003023853,
      "grad_norm": 1.222223162651062,
      "learning_rate": 2.9060000000000003e-05,
      "loss": 0.455,
      "step": 5815
    },
    {
      "epoch": 0.32568036734236755,
      "grad_norm": 1.2108324766159058,
      "learning_rate": 2.9065000000000004e-05,
      "loss": 0.4331,
      "step": 5816
    },
    {
      "epoch": 0.32573636465449657,
      "grad_norm": 1.1987415552139282,
      "learning_rate": 2.907e-05,
      "loss": 0.4094,
      "step": 5817
    },
    {
      "epoch": 0.3257923619666256,
      "grad_norm": 1.2771337032318115,
      "learning_rate": 2.9075000000000002e-05,
      "loss": 0.4081,
      "step": 5818
    },
    {
      "epoch": 0.3258483592787546,
      "grad_norm": 1.3818776607513428,
      "learning_rate": 2.9080000000000003e-05,
      "loss": 0.5237,
      "step": 5819
    },
    {
      "epoch": 0.3259043565908836,
      "grad_norm": 1.0041522979736328,
      "learning_rate": 2.9085e-05,
      "loss": 0.4008,
      "step": 5820
    },
    {
      "epoch": 0.32596035390301265,
      "grad_norm": 1.1087275743484497,
      "learning_rate": 2.909e-05,
      "loss": 0.3765,
      "step": 5821
    },
    {
      "epoch": 0.32601635121514166,
      "grad_norm": 1.0786428451538086,
      "learning_rate": 2.9095e-05,
      "loss": 0.4555,
      "step": 5822
    },
    {
      "epoch": 0.3260723485272707,
      "grad_norm": 1.2589988708496094,
      "learning_rate": 2.91e-05,
      "loss": 0.3989,
      "step": 5823
    },
    {
      "epoch": 0.3261283458393997,
      "grad_norm": 1.1303046941757202,
      "learning_rate": 2.9105e-05,
      "loss": 0.4542,
      "step": 5824
    },
    {
      "epoch": 0.3261843431515287,
      "grad_norm": 1.238532543182373,
      "learning_rate": 2.9110000000000004e-05,
      "loss": 0.4887,
      "step": 5825
    },
    {
      "epoch": 0.32624034046365774,
      "grad_norm": 1.1298034191131592,
      "learning_rate": 2.9115000000000005e-05,
      "loss": 0.3971,
      "step": 5826
    },
    {
      "epoch": 0.32629633777578676,
      "grad_norm": 1.25812828540802,
      "learning_rate": 2.9120000000000002e-05,
      "loss": 0.3751,
      "step": 5827
    },
    {
      "epoch": 0.3263523350879158,
      "grad_norm": 2.0425925254821777,
      "learning_rate": 2.9125000000000003e-05,
      "loss": 0.6093,
      "step": 5828
    },
    {
      "epoch": 0.3264083324000448,
      "grad_norm": 1.1871418952941895,
      "learning_rate": 2.913e-05,
      "loss": 0.3733,
      "step": 5829
    },
    {
      "epoch": 0.3264643297121738,
      "grad_norm": 1.1049633026123047,
      "learning_rate": 2.9135e-05,
      "loss": 0.3699,
      "step": 5830
    },
    {
      "epoch": 0.32652032702430284,
      "grad_norm": 1.097481369972229,
      "learning_rate": 2.9140000000000002e-05,
      "loss": 0.3666,
      "step": 5831
    },
    {
      "epoch": 0.32657632433643186,
      "grad_norm": 1.218788743019104,
      "learning_rate": 2.9145e-05,
      "loss": 0.4946,
      "step": 5832
    },
    {
      "epoch": 0.3266323216485609,
      "grad_norm": 1.073500394821167,
      "learning_rate": 2.915e-05,
      "loss": 0.4256,
      "step": 5833
    },
    {
      "epoch": 0.3266883189606899,
      "grad_norm": 1.147639513015747,
      "learning_rate": 2.9154999999999998e-05,
      "loss": 0.4234,
      "step": 5834
    },
    {
      "epoch": 0.3267443162728189,
      "grad_norm": 1.3192564249038696,
      "learning_rate": 2.9160000000000005e-05,
      "loss": 0.398,
      "step": 5835
    },
    {
      "epoch": 0.32680031358494793,
      "grad_norm": 1.2515380382537842,
      "learning_rate": 2.9165000000000003e-05,
      "loss": 0.3749,
      "step": 5836
    },
    {
      "epoch": 0.32685631089707695,
      "grad_norm": 1.0096896886825562,
      "learning_rate": 2.9170000000000004e-05,
      "loss": 0.377,
      "step": 5837
    },
    {
      "epoch": 0.32691230820920597,
      "grad_norm": 1.4385930299758911,
      "learning_rate": 2.9175e-05,
      "loss": 0.42,
      "step": 5838
    },
    {
      "epoch": 0.326968305521335,
      "grad_norm": 1.1319615840911865,
      "learning_rate": 2.9180000000000002e-05,
      "loss": 0.4285,
      "step": 5839
    },
    {
      "epoch": 0.327024302833464,
      "grad_norm": 1.2909079790115356,
      "learning_rate": 2.9185000000000003e-05,
      "loss": 0.3371,
      "step": 5840
    },
    {
      "epoch": 0.32708030014559303,
      "grad_norm": 1.3051002025604248,
      "learning_rate": 2.919e-05,
      "loss": 0.4579,
      "step": 5841
    },
    {
      "epoch": 0.32713629745772205,
      "grad_norm": 1.2702544927597046,
      "learning_rate": 2.9195e-05,
      "loss": 0.4637,
      "step": 5842
    },
    {
      "epoch": 0.32719229476985107,
      "grad_norm": 1.6683790683746338,
      "learning_rate": 2.9199999999999998e-05,
      "loss": 0.3962,
      "step": 5843
    },
    {
      "epoch": 0.3272482920819801,
      "grad_norm": 1.2359310388565063,
      "learning_rate": 2.9205e-05,
      "loss": 0.4835,
      "step": 5844
    },
    {
      "epoch": 0.3273042893941091,
      "grad_norm": 1.5957162380218506,
      "learning_rate": 2.9210000000000003e-05,
      "loss": 0.529,
      "step": 5845
    },
    {
      "epoch": 0.3273602867062381,
      "grad_norm": 1.5851261615753174,
      "learning_rate": 2.9215000000000004e-05,
      "loss": 0.6152,
      "step": 5846
    },
    {
      "epoch": 0.32741628401836714,
      "grad_norm": 1.1549354791641235,
      "learning_rate": 2.922e-05,
      "loss": 0.5002,
      "step": 5847
    },
    {
      "epoch": 0.32747228133049616,
      "grad_norm": 1.7438483238220215,
      "learning_rate": 2.9225000000000002e-05,
      "loss": 0.5194,
      "step": 5848
    },
    {
      "epoch": 0.3275282786426252,
      "grad_norm": 1.0823826789855957,
      "learning_rate": 2.9230000000000003e-05,
      "loss": 0.3454,
      "step": 5849
    },
    {
      "epoch": 0.32758427595475414,
      "grad_norm": 1.1117078065872192,
      "learning_rate": 2.9235e-05,
      "loss": 0.3429,
      "step": 5850
    },
    {
      "epoch": 0.32764027326688316,
      "grad_norm": 1.3410907983779907,
      "learning_rate": 2.924e-05,
      "loss": 0.4798,
      "step": 5851
    },
    {
      "epoch": 0.3276962705790122,
      "grad_norm": 1.2882479429244995,
      "learning_rate": 2.9245e-05,
      "loss": 0.4065,
      "step": 5852
    },
    {
      "epoch": 0.3277522678911412,
      "grad_norm": 1.0705493688583374,
      "learning_rate": 2.925e-05,
      "loss": 0.4158,
      "step": 5853
    },
    {
      "epoch": 0.3278082652032702,
      "grad_norm": 1.15053129196167,
      "learning_rate": 2.9255e-05,
      "loss": 0.44,
      "step": 5854
    },
    {
      "epoch": 0.32786426251539924,
      "grad_norm": 1.1476150751113892,
      "learning_rate": 2.9260000000000004e-05,
      "loss": 0.4863,
      "step": 5855
    },
    {
      "epoch": 0.32792025982752826,
      "grad_norm": 1.0913232564926147,
      "learning_rate": 2.9265000000000002e-05,
      "loss": 0.3419,
      "step": 5856
    },
    {
      "epoch": 0.3279762571396573,
      "grad_norm": 1.147476315498352,
      "learning_rate": 2.9270000000000003e-05,
      "loss": 0.524,
      "step": 5857
    },
    {
      "epoch": 0.3280322544517863,
      "grad_norm": 1.0973833799362183,
      "learning_rate": 2.9275000000000003e-05,
      "loss": 0.4205,
      "step": 5858
    },
    {
      "epoch": 0.3280882517639153,
      "grad_norm": 1.4630779027938843,
      "learning_rate": 2.928e-05,
      "loss": 0.5707,
      "step": 5859
    },
    {
      "epoch": 0.32814424907604434,
      "grad_norm": 1.075952172279358,
      "learning_rate": 2.9285e-05,
      "loss": 0.276,
      "step": 5860
    },
    {
      "epoch": 0.32820024638817336,
      "grad_norm": 1.5002340078353882,
      "learning_rate": 2.929e-05,
      "loss": 0.4981,
      "step": 5861
    },
    {
      "epoch": 0.3282562437003024,
      "grad_norm": 0.9377226233482361,
      "learning_rate": 2.9295e-05,
      "loss": 0.3876,
      "step": 5862
    },
    {
      "epoch": 0.3283122410124314,
      "grad_norm": 1.3477425575256348,
      "learning_rate": 2.93e-05,
      "loss": 0.4883,
      "step": 5863
    },
    {
      "epoch": 0.3283682383245604,
      "grad_norm": 1.6481932401657104,
      "learning_rate": 2.9304999999999998e-05,
      "loss": 0.5014,
      "step": 5864
    },
    {
      "epoch": 0.32842423563668943,
      "grad_norm": 1.1377789974212646,
      "learning_rate": 2.9310000000000006e-05,
      "loss": 0.3973,
      "step": 5865
    },
    {
      "epoch": 0.32848023294881845,
      "grad_norm": 1.1514617204666138,
      "learning_rate": 2.9315000000000003e-05,
      "loss": 0.4386,
      "step": 5866
    },
    {
      "epoch": 0.32853623026094747,
      "grad_norm": 1.2926123142242432,
      "learning_rate": 2.9320000000000004e-05,
      "loss": 0.4478,
      "step": 5867
    },
    {
      "epoch": 0.3285922275730765,
      "grad_norm": 1.2897146940231323,
      "learning_rate": 2.9325e-05,
      "loss": 0.3578,
      "step": 5868
    },
    {
      "epoch": 0.3286482248852055,
      "grad_norm": 1.2622413635253906,
      "learning_rate": 2.9330000000000002e-05,
      "loss": 0.4091,
      "step": 5869
    },
    {
      "epoch": 0.32870422219733453,
      "grad_norm": 1.2460603713989258,
      "learning_rate": 2.9335000000000003e-05,
      "loss": 0.3957,
      "step": 5870
    },
    {
      "epoch": 0.32876021950946355,
      "grad_norm": 1.0655896663665771,
      "learning_rate": 2.934e-05,
      "loss": 0.4109,
      "step": 5871
    },
    {
      "epoch": 0.32881621682159257,
      "grad_norm": 1.2387356758117676,
      "learning_rate": 2.9345e-05,
      "loss": 0.3508,
      "step": 5872
    },
    {
      "epoch": 0.3288722141337216,
      "grad_norm": 1.2638161182403564,
      "learning_rate": 2.935e-05,
      "loss": 0.4697,
      "step": 5873
    },
    {
      "epoch": 0.3289282114458506,
      "grad_norm": 1.4944034814834595,
      "learning_rate": 2.9355e-05,
      "loss": 0.4519,
      "step": 5874
    },
    {
      "epoch": 0.3289842087579796,
      "grad_norm": 1.0900880098342896,
      "learning_rate": 2.9360000000000003e-05,
      "loss": 0.4708,
      "step": 5875
    },
    {
      "epoch": 0.32904020607010864,
      "grad_norm": 1.1817508935928345,
      "learning_rate": 2.9365000000000004e-05,
      "loss": 0.4162,
      "step": 5876
    },
    {
      "epoch": 0.32909620338223766,
      "grad_norm": 1.2182482481002808,
      "learning_rate": 2.9370000000000002e-05,
      "loss": 0.4709,
      "step": 5877
    },
    {
      "epoch": 0.3291522006943667,
      "grad_norm": 1.1931767463684082,
      "learning_rate": 2.9375000000000003e-05,
      "loss": 0.3835,
      "step": 5878
    },
    {
      "epoch": 0.3292081980064957,
      "grad_norm": 1.245471715927124,
      "learning_rate": 2.9380000000000003e-05,
      "loss": 0.4692,
      "step": 5879
    },
    {
      "epoch": 0.3292641953186247,
      "grad_norm": 1.108049750328064,
      "learning_rate": 2.9385e-05,
      "loss": 0.5946,
      "step": 5880
    },
    {
      "epoch": 0.32932019263075374,
      "grad_norm": 1.0938096046447754,
      "learning_rate": 2.939e-05,
      "loss": 0.3256,
      "step": 5881
    },
    {
      "epoch": 0.32937618994288276,
      "grad_norm": 1.4213799238204956,
      "learning_rate": 2.9395e-05,
      "loss": 0.3809,
      "step": 5882
    },
    {
      "epoch": 0.3294321872550118,
      "grad_norm": 0.97994464635849,
      "learning_rate": 2.94e-05,
      "loss": 0.2664,
      "step": 5883
    },
    {
      "epoch": 0.3294881845671408,
      "grad_norm": 1.332228183746338,
      "learning_rate": 2.9405e-05,
      "loss": 0.4286,
      "step": 5884
    },
    {
      "epoch": 0.3295441818792698,
      "grad_norm": 1.1238287687301636,
      "learning_rate": 2.9409999999999998e-05,
      "loss": 0.3891,
      "step": 5885
    },
    {
      "epoch": 0.32960017919139883,
      "grad_norm": 1.1983394622802734,
      "learning_rate": 2.9415000000000002e-05,
      "loss": 0.4452,
      "step": 5886
    },
    {
      "epoch": 0.32965617650352785,
      "grad_norm": 1.1007509231567383,
      "learning_rate": 2.9420000000000003e-05,
      "loss": 0.3467,
      "step": 5887
    },
    {
      "epoch": 0.3297121738156569,
      "grad_norm": 1.1530849933624268,
      "learning_rate": 2.9425000000000004e-05,
      "loss": 0.3385,
      "step": 5888
    },
    {
      "epoch": 0.3297681711277859,
      "grad_norm": 1.2139053344726562,
      "learning_rate": 2.943e-05,
      "loss": 0.3979,
      "step": 5889
    },
    {
      "epoch": 0.3298241684399149,
      "grad_norm": 1.1592355966567993,
      "learning_rate": 2.9435000000000002e-05,
      "loss": 0.4563,
      "step": 5890
    },
    {
      "epoch": 0.3298801657520439,
      "grad_norm": 1.191756248474121,
      "learning_rate": 2.944e-05,
      "loss": 0.3815,
      "step": 5891
    },
    {
      "epoch": 0.3299361630641729,
      "grad_norm": 1.5103602409362793,
      "learning_rate": 2.9445e-05,
      "loss": 0.5304,
      "step": 5892
    },
    {
      "epoch": 0.3299921603763019,
      "grad_norm": 1.2559441328048706,
      "learning_rate": 2.945e-05,
      "loss": 0.4914,
      "step": 5893
    },
    {
      "epoch": 0.33004815768843093,
      "grad_norm": 1.2686264514923096,
      "learning_rate": 2.9455e-05,
      "loss": 0.4153,
      "step": 5894
    },
    {
      "epoch": 0.33010415500055995,
      "grad_norm": 1.3157553672790527,
      "learning_rate": 2.946e-05,
      "loss": 0.3492,
      "step": 5895
    },
    {
      "epoch": 0.33016015231268897,
      "grad_norm": 1.3160196542739868,
      "learning_rate": 2.9465000000000003e-05,
      "loss": 0.4295,
      "step": 5896
    },
    {
      "epoch": 0.330216149624818,
      "grad_norm": 1.253178596496582,
      "learning_rate": 2.9470000000000004e-05,
      "loss": 0.3636,
      "step": 5897
    },
    {
      "epoch": 0.330272146936947,
      "grad_norm": 1.1357029676437378,
      "learning_rate": 2.9475e-05,
      "loss": 0.3601,
      "step": 5898
    },
    {
      "epoch": 0.33032814424907603,
      "grad_norm": 1.3248504400253296,
      "learning_rate": 2.9480000000000002e-05,
      "loss": 0.5028,
      "step": 5899
    },
    {
      "epoch": 0.33038414156120505,
      "grad_norm": 0.962448239326477,
      "learning_rate": 2.9485000000000003e-05,
      "loss": 0.296,
      "step": 5900
    },
    {
      "epoch": 0.33044013887333407,
      "grad_norm": 1.3095908164978027,
      "learning_rate": 2.949e-05,
      "loss": 0.3847,
      "step": 5901
    },
    {
      "epoch": 0.3304961361854631,
      "grad_norm": 0.9791283011436462,
      "learning_rate": 2.9495e-05,
      "loss": 0.2507,
      "step": 5902
    },
    {
      "epoch": 0.3305521334975921,
      "grad_norm": 1.371442198753357,
      "learning_rate": 2.95e-05,
      "loss": 0.3895,
      "step": 5903
    },
    {
      "epoch": 0.3306081308097211,
      "grad_norm": 1.227273941040039,
      "learning_rate": 2.9505e-05,
      "loss": 0.3704,
      "step": 5904
    },
    {
      "epoch": 0.33066412812185014,
      "grad_norm": 1.2468384504318237,
      "learning_rate": 2.951e-05,
      "loss": 0.4539,
      "step": 5905
    },
    {
      "epoch": 0.33072012543397916,
      "grad_norm": 1.2714205980300903,
      "learning_rate": 2.9515000000000005e-05,
      "loss": 0.4572,
      "step": 5906
    },
    {
      "epoch": 0.3307761227461082,
      "grad_norm": 2.2266745567321777,
      "learning_rate": 2.9520000000000002e-05,
      "loss": 0.4953,
      "step": 5907
    },
    {
      "epoch": 0.3308321200582372,
      "grad_norm": 1.3567014932632446,
      "learning_rate": 2.9525000000000003e-05,
      "loss": 0.5034,
      "step": 5908
    },
    {
      "epoch": 0.3308881173703662,
      "grad_norm": 0.9874616265296936,
      "learning_rate": 2.9530000000000004e-05,
      "loss": 0.3476,
      "step": 5909
    },
    {
      "epoch": 0.33094411468249524,
      "grad_norm": 1.2961267232894897,
      "learning_rate": 2.9535e-05,
      "loss": 0.6049,
      "step": 5910
    },
    {
      "epoch": 0.33100011199462426,
      "grad_norm": 1.4374582767486572,
      "learning_rate": 2.9540000000000002e-05,
      "loss": 0.4624,
      "step": 5911
    },
    {
      "epoch": 0.3310561093067533,
      "grad_norm": 2.4435040950775146,
      "learning_rate": 2.9545e-05,
      "loss": 0.5334,
      "step": 5912
    },
    {
      "epoch": 0.3311121066188823,
      "grad_norm": 1.7158364057540894,
      "learning_rate": 2.955e-05,
      "loss": 0.5466,
      "step": 5913
    },
    {
      "epoch": 0.3311681039310113,
      "grad_norm": 1.1626513004302979,
      "learning_rate": 2.9555e-05,
      "loss": 0.3731,
      "step": 5914
    },
    {
      "epoch": 0.33122410124314033,
      "grad_norm": 1.1088207960128784,
      "learning_rate": 2.9559999999999998e-05,
      "loss": 0.5113,
      "step": 5915
    },
    {
      "epoch": 0.33128009855526935,
      "grad_norm": 1.0978524684906006,
      "learning_rate": 2.9565000000000002e-05,
      "loss": 0.3087,
      "step": 5916
    },
    {
      "epoch": 0.3313360958673984,
      "grad_norm": 1.5576510429382324,
      "learning_rate": 2.9570000000000003e-05,
      "loss": 0.3773,
      "step": 5917
    },
    {
      "epoch": 0.3313920931795274,
      "grad_norm": 1.8396542072296143,
      "learning_rate": 2.9575000000000004e-05,
      "loss": 0.4837,
      "step": 5918
    },
    {
      "epoch": 0.3314480904916564,
      "grad_norm": 1.2947006225585938,
      "learning_rate": 2.958e-05,
      "loss": 0.4933,
      "step": 5919
    },
    {
      "epoch": 0.33150408780378543,
      "grad_norm": 1.352721095085144,
      "learning_rate": 2.9585000000000002e-05,
      "loss": 0.4512,
      "step": 5920
    },
    {
      "epoch": 0.33156008511591445,
      "grad_norm": 1.2786140441894531,
      "learning_rate": 2.959e-05,
      "loss": 0.5311,
      "step": 5921
    },
    {
      "epoch": 0.33161608242804347,
      "grad_norm": 1.3983354568481445,
      "learning_rate": 2.9595e-05,
      "loss": 0.4592,
      "step": 5922
    },
    {
      "epoch": 0.3316720797401725,
      "grad_norm": 1.2974272966384888,
      "learning_rate": 2.96e-05,
      "loss": 0.3691,
      "step": 5923
    },
    {
      "epoch": 0.3317280770523015,
      "grad_norm": 1.0443476438522339,
      "learning_rate": 2.9605e-05,
      "loss": 0.4561,
      "step": 5924
    },
    {
      "epoch": 0.3317840743644305,
      "grad_norm": 1.1290427446365356,
      "learning_rate": 2.961e-05,
      "loss": 0.3246,
      "step": 5925
    },
    {
      "epoch": 0.33184007167655954,
      "grad_norm": 1.2534984350204468,
      "learning_rate": 2.9615000000000004e-05,
      "loss": 0.4785,
      "step": 5926
    },
    {
      "epoch": 0.33189606898868856,
      "grad_norm": 1.167789340019226,
      "learning_rate": 2.9620000000000004e-05,
      "loss": 0.3557,
      "step": 5927
    },
    {
      "epoch": 0.3319520663008176,
      "grad_norm": 1.1558021306991577,
      "learning_rate": 2.9625000000000002e-05,
      "loss": 0.4502,
      "step": 5928
    },
    {
      "epoch": 0.3320080636129466,
      "grad_norm": 1.1212310791015625,
      "learning_rate": 2.9630000000000003e-05,
      "loss": 0.3688,
      "step": 5929
    },
    {
      "epoch": 0.3320640609250756,
      "grad_norm": 1.2709614038467407,
      "learning_rate": 2.9635e-05,
      "loss": 0.5522,
      "step": 5930
    },
    {
      "epoch": 0.33212005823720464,
      "grad_norm": 1.2580879926681519,
      "learning_rate": 2.964e-05,
      "loss": 0.3608,
      "step": 5931
    },
    {
      "epoch": 0.3321760555493336,
      "grad_norm": 1.2257146835327148,
      "learning_rate": 2.9645e-05,
      "loss": 0.3542,
      "step": 5932
    },
    {
      "epoch": 0.3322320528614626,
      "grad_norm": 2.1395881175994873,
      "learning_rate": 2.965e-05,
      "loss": 0.3611,
      "step": 5933
    },
    {
      "epoch": 0.33228805017359164,
      "grad_norm": 1.2016584873199463,
      "learning_rate": 2.9655e-05,
      "loss": 0.3914,
      "step": 5934
    },
    {
      "epoch": 0.33234404748572066,
      "grad_norm": 1.1124950647354126,
      "learning_rate": 2.9659999999999997e-05,
      "loss": 0.3443,
      "step": 5935
    },
    {
      "epoch": 0.3324000447978497,
      "grad_norm": 0.9924523234367371,
      "learning_rate": 2.9665000000000005e-05,
      "loss": 0.343,
      "step": 5936
    },
    {
      "epoch": 0.3324560421099787,
      "grad_norm": 1.3250898122787476,
      "learning_rate": 2.9670000000000002e-05,
      "loss": 0.5017,
      "step": 5937
    },
    {
      "epoch": 0.3325120394221077,
      "grad_norm": 1.2025245428085327,
      "learning_rate": 2.9675000000000003e-05,
      "loss": 0.4645,
      "step": 5938
    },
    {
      "epoch": 0.33256803673423674,
      "grad_norm": 1.294891119003296,
      "learning_rate": 2.9680000000000004e-05,
      "loss": 0.3881,
      "step": 5939
    },
    {
      "epoch": 0.33262403404636576,
      "grad_norm": 1.1268936395645142,
      "learning_rate": 2.9685e-05,
      "loss": 0.3423,
      "step": 5940
    },
    {
      "epoch": 0.3326800313584948,
      "grad_norm": 1.1726906299591064,
      "learning_rate": 2.9690000000000002e-05,
      "loss": 0.4491,
      "step": 5941
    },
    {
      "epoch": 0.3327360286706238,
      "grad_norm": 1.1992634534835815,
      "learning_rate": 2.9695e-05,
      "loss": 0.527,
      "step": 5942
    },
    {
      "epoch": 0.3327920259827528,
      "grad_norm": 1.3620399236679077,
      "learning_rate": 2.97e-05,
      "loss": 0.4762,
      "step": 5943
    },
    {
      "epoch": 0.33284802329488183,
      "grad_norm": 1.216871738433838,
      "learning_rate": 2.9705e-05,
      "loss": 0.468,
      "step": 5944
    },
    {
      "epoch": 0.33290402060701085,
      "grad_norm": 1.324371099472046,
      "learning_rate": 2.971e-05,
      "loss": 0.3604,
      "step": 5945
    },
    {
      "epoch": 0.3329600179191399,
      "grad_norm": 1.2822390794754028,
      "learning_rate": 2.9715000000000003e-05,
      "loss": 0.6002,
      "step": 5946
    },
    {
      "epoch": 0.3330160152312689,
      "grad_norm": 1.1421071290969849,
      "learning_rate": 2.9720000000000003e-05,
      "loss": 0.3899,
      "step": 5947
    },
    {
      "epoch": 0.3330720125433979,
      "grad_norm": 1.2771941423416138,
      "learning_rate": 2.9725000000000004e-05,
      "loss": 0.4653,
      "step": 5948
    },
    {
      "epoch": 0.33312800985552693,
      "grad_norm": 1.1364119052886963,
      "learning_rate": 2.973e-05,
      "loss": 0.4113,
      "step": 5949
    },
    {
      "epoch": 0.33318400716765595,
      "grad_norm": 1.1866883039474487,
      "learning_rate": 2.9735000000000002e-05,
      "loss": 0.3082,
      "step": 5950
    },
    {
      "epoch": 0.33324000447978497,
      "grad_norm": 1.2553757429122925,
      "learning_rate": 2.974e-05,
      "loss": 0.4513,
      "step": 5951
    },
    {
      "epoch": 0.333296001791914,
      "grad_norm": 1.2461782693862915,
      "learning_rate": 2.9745e-05,
      "loss": 0.4556,
      "step": 5952
    },
    {
      "epoch": 0.333351999104043,
      "grad_norm": 1.210750699043274,
      "learning_rate": 2.975e-05,
      "loss": 0.4,
      "step": 5953
    },
    {
      "epoch": 0.333407996416172,
      "grad_norm": 1.0719366073608398,
      "learning_rate": 2.9755e-05,
      "loss": 0.315,
      "step": 5954
    },
    {
      "epoch": 0.33346399372830104,
      "grad_norm": 2.0919089317321777,
      "learning_rate": 2.976e-05,
      "loss": 0.4499,
      "step": 5955
    },
    {
      "epoch": 0.33351999104043006,
      "grad_norm": 1.1930122375488281,
      "learning_rate": 2.9765000000000004e-05,
      "loss": 0.4203,
      "step": 5956
    },
    {
      "epoch": 0.3335759883525591,
      "grad_norm": 1.2995959520339966,
      "learning_rate": 2.9770000000000005e-05,
      "loss": 0.4431,
      "step": 5957
    },
    {
      "epoch": 0.3336319856646881,
      "grad_norm": 1.2213457822799683,
      "learning_rate": 2.9775000000000002e-05,
      "loss": 0.4014,
      "step": 5958
    },
    {
      "epoch": 0.3336879829768171,
      "grad_norm": 1.2256245613098145,
      "learning_rate": 2.9780000000000003e-05,
      "loss": 0.3807,
      "step": 5959
    },
    {
      "epoch": 0.33374398028894614,
      "grad_norm": 1.2050431966781616,
      "learning_rate": 2.9785e-05,
      "loss": 0.4637,
      "step": 5960
    },
    {
      "epoch": 0.33379997760107516,
      "grad_norm": 1.3042926788330078,
      "learning_rate": 2.979e-05,
      "loss": 0.4021,
      "step": 5961
    },
    {
      "epoch": 0.3338559749132042,
      "grad_norm": 1.3158693313598633,
      "learning_rate": 2.9795000000000002e-05,
      "loss": 0.3686,
      "step": 5962
    },
    {
      "epoch": 0.3339119722253332,
      "grad_norm": 1.1798471212387085,
      "learning_rate": 2.98e-05,
      "loss": 0.5074,
      "step": 5963
    },
    {
      "epoch": 0.3339679695374622,
      "grad_norm": 4.180051326751709,
      "learning_rate": 2.9805e-05,
      "loss": 0.4549,
      "step": 5964
    },
    {
      "epoch": 0.33402396684959124,
      "grad_norm": 1.3571497201919556,
      "learning_rate": 2.9809999999999997e-05,
      "loss": 0.5164,
      "step": 5965
    },
    {
      "epoch": 0.33407996416172026,
      "grad_norm": 1.260483980178833,
      "learning_rate": 2.9815000000000005e-05,
      "loss": 0.4295,
      "step": 5966
    },
    {
      "epoch": 0.3341359614738493,
      "grad_norm": 4.337609767913818,
      "learning_rate": 2.9820000000000002e-05,
      "loss": 0.5291,
      "step": 5967
    },
    {
      "epoch": 0.3341919587859783,
      "grad_norm": 1.2313674688339233,
      "learning_rate": 2.9825000000000003e-05,
      "loss": 0.3771,
      "step": 5968
    },
    {
      "epoch": 0.3342479560981073,
      "grad_norm": 1.0947515964508057,
      "learning_rate": 2.9830000000000004e-05,
      "loss": 0.3706,
      "step": 5969
    },
    {
      "epoch": 0.33430395341023633,
      "grad_norm": 1.1112008094787598,
      "learning_rate": 2.9835e-05,
      "loss": 0.3836,
      "step": 5970
    },
    {
      "epoch": 0.33435995072236535,
      "grad_norm": 1.6684881448745728,
      "learning_rate": 2.9840000000000002e-05,
      "loss": 0.777,
      "step": 5971
    },
    {
      "epoch": 0.33441594803449437,
      "grad_norm": 1.222672939300537,
      "learning_rate": 2.9845e-05,
      "loss": 0.3733,
      "step": 5972
    },
    {
      "epoch": 0.3344719453466234,
      "grad_norm": 1.3642115592956543,
      "learning_rate": 2.985e-05,
      "loss": 0.4429,
      "step": 5973
    },
    {
      "epoch": 0.33452794265875235,
      "grad_norm": 1.3494887351989746,
      "learning_rate": 2.9855e-05,
      "loss": 0.5372,
      "step": 5974
    },
    {
      "epoch": 0.33458393997088137,
      "grad_norm": 1.0940035581588745,
      "learning_rate": 2.986e-05,
      "loss": 0.3721,
      "step": 5975
    },
    {
      "epoch": 0.3346399372830104,
      "grad_norm": 1.239007830619812,
      "learning_rate": 2.9865000000000003e-05,
      "loss": 0.4012,
      "step": 5976
    },
    {
      "epoch": 0.3346959345951394,
      "grad_norm": 1.5419095754623413,
      "learning_rate": 2.9870000000000004e-05,
      "loss": 0.4955,
      "step": 5977
    },
    {
      "epoch": 0.33475193190726843,
      "grad_norm": 1.0845848321914673,
      "learning_rate": 2.9875000000000004e-05,
      "loss": 0.4384,
      "step": 5978
    },
    {
      "epoch": 0.33480792921939745,
      "grad_norm": 1.198300838470459,
      "learning_rate": 2.9880000000000002e-05,
      "loss": 0.4088,
      "step": 5979
    },
    {
      "epoch": 0.33486392653152647,
      "grad_norm": 1.2984102964401245,
      "learning_rate": 2.9885000000000003e-05,
      "loss": 0.5256,
      "step": 5980
    },
    {
      "epoch": 0.3349199238436555,
      "grad_norm": 1.0907014608383179,
      "learning_rate": 2.989e-05,
      "loss": 0.317,
      "step": 5981
    },
    {
      "epoch": 0.3349759211557845,
      "grad_norm": 1.1750301122665405,
      "learning_rate": 2.9895e-05,
      "loss": 0.3831,
      "step": 5982
    },
    {
      "epoch": 0.3350319184679135,
      "grad_norm": 0.9335674047470093,
      "learning_rate": 2.9900000000000002e-05,
      "loss": 0.3426,
      "step": 5983
    },
    {
      "epoch": 0.33508791578004254,
      "grad_norm": 1.2949601411819458,
      "learning_rate": 2.9905e-05,
      "loss": 0.5453,
      "step": 5984
    },
    {
      "epoch": 0.33514391309217156,
      "grad_norm": 1.1657267808914185,
      "learning_rate": 2.991e-05,
      "loss": 0.4275,
      "step": 5985
    },
    {
      "epoch": 0.3351999104043006,
      "grad_norm": 0.9324960708618164,
      "learning_rate": 2.9915000000000004e-05,
      "loss": 0.247,
      "step": 5986
    },
    {
      "epoch": 0.3352559077164296,
      "grad_norm": 1.3305065631866455,
      "learning_rate": 2.9920000000000005e-05,
      "loss": 0.3834,
      "step": 5987
    },
    {
      "epoch": 0.3353119050285586,
      "grad_norm": 1.611104965209961,
      "learning_rate": 2.9925000000000002e-05,
      "loss": 0.4356,
      "step": 5988
    },
    {
      "epoch": 0.33536790234068764,
      "grad_norm": 1.1590913534164429,
      "learning_rate": 2.9930000000000003e-05,
      "loss": 0.4541,
      "step": 5989
    },
    {
      "epoch": 0.33542389965281666,
      "grad_norm": 1.1663968563079834,
      "learning_rate": 2.9935e-05,
      "loss": 0.3898,
      "step": 5990
    },
    {
      "epoch": 0.3354798969649457,
      "grad_norm": 0.9980918169021606,
      "learning_rate": 2.994e-05,
      "loss": 0.4104,
      "step": 5991
    },
    {
      "epoch": 0.3355358942770747,
      "grad_norm": 1.218346118927002,
      "learning_rate": 2.9945000000000002e-05,
      "loss": 0.3951,
      "step": 5992
    },
    {
      "epoch": 0.3355918915892037,
      "grad_norm": 1.4852898120880127,
      "learning_rate": 2.995e-05,
      "loss": 0.5311,
      "step": 5993
    },
    {
      "epoch": 0.33564788890133274,
      "grad_norm": 1.5939909219741821,
      "learning_rate": 2.9955e-05,
      "loss": 0.5085,
      "step": 5994
    },
    {
      "epoch": 0.33570388621346176,
      "grad_norm": 1.1286392211914062,
      "learning_rate": 2.9959999999999998e-05,
      "loss": 0.3972,
      "step": 5995
    },
    {
      "epoch": 0.3357598835255908,
      "grad_norm": 1.3640192747116089,
      "learning_rate": 2.9965000000000005e-05,
      "loss": 0.6368,
      "step": 5996
    },
    {
      "epoch": 0.3358158808377198,
      "grad_norm": 1.1870054006576538,
      "learning_rate": 2.9970000000000003e-05,
      "loss": 0.3638,
      "step": 5997
    },
    {
      "epoch": 0.3358718781498488,
      "grad_norm": 1.1932722330093384,
      "learning_rate": 2.9975000000000004e-05,
      "loss": 0.3375,
      "step": 5998
    },
    {
      "epoch": 0.33592787546197783,
      "grad_norm": 1.1448731422424316,
      "learning_rate": 2.998e-05,
      "loss": 0.4382,
      "step": 5999
    },
    {
      "epoch": 0.33598387277410685,
      "grad_norm": 1.493430256843567,
      "learning_rate": 2.9985000000000002e-05,
      "loss": 0.3791,
      "step": 6000
    },
    {
      "epoch": 0.33603987008623587,
      "grad_norm": 1.4207919836044312,
      "learning_rate": 2.9990000000000003e-05,
      "loss": 0.4886,
      "step": 6001
    },
    {
      "epoch": 0.3360958673983649,
      "grad_norm": 1.2372599840164185,
      "learning_rate": 2.9995e-05,
      "loss": 0.3462,
      "step": 6002
    },
    {
      "epoch": 0.3361518647104939,
      "grad_norm": 1.1469321250915527,
      "learning_rate": 3e-05,
      "loss": 0.2863,
      "step": 6003
    },
    {
      "epoch": 0.3362078620226229,
      "grad_norm": 1.5369642972946167,
      "learning_rate": 3.0004999999999998e-05,
      "loss": 0.42,
      "step": 6004
    },
    {
      "epoch": 0.33626385933475195,
      "grad_norm": 1.1388967037200928,
      "learning_rate": 3.001e-05,
      "loss": 0.3926,
      "step": 6005
    },
    {
      "epoch": 0.33631985664688097,
      "grad_norm": 1.1151459217071533,
      "learning_rate": 3.0015e-05,
      "loss": 0.32,
      "step": 6006
    },
    {
      "epoch": 0.33637585395901,
      "grad_norm": 1.7051790952682495,
      "learning_rate": 3.0020000000000004e-05,
      "loss": 0.3744,
      "step": 6007
    },
    {
      "epoch": 0.336431851271139,
      "grad_norm": 1.3285667896270752,
      "learning_rate": 3.0025000000000005e-05,
      "loss": 0.4068,
      "step": 6008
    },
    {
      "epoch": 0.336487848583268,
      "grad_norm": 1.3671685457229614,
      "learning_rate": 3.0030000000000002e-05,
      "loss": 0.476,
      "step": 6009
    },
    {
      "epoch": 0.33654384589539704,
      "grad_norm": 1.2600650787353516,
      "learning_rate": 3.0035000000000003e-05,
      "loss": 0.4219,
      "step": 6010
    },
    {
      "epoch": 0.33659984320752606,
      "grad_norm": 1.5153084993362427,
      "learning_rate": 3.004e-05,
      "loss": 0.5961,
      "step": 6011
    },
    {
      "epoch": 0.3366558405196551,
      "grad_norm": 1.4338983297348022,
      "learning_rate": 3.0045e-05,
      "loss": 0.4875,
      "step": 6012
    },
    {
      "epoch": 0.3367118378317841,
      "grad_norm": 1.2349830865859985,
      "learning_rate": 3.0050000000000002e-05,
      "loss": 0.3986,
      "step": 6013
    },
    {
      "epoch": 0.3367678351439131,
      "grad_norm": 1.2845345735549927,
      "learning_rate": 3.0055e-05,
      "loss": 0.3558,
      "step": 6014
    },
    {
      "epoch": 0.3368238324560421,
      "grad_norm": 1.5571088790893555,
      "learning_rate": 3.006e-05,
      "loss": 0.3869,
      "step": 6015
    },
    {
      "epoch": 0.3368798297681711,
      "grad_norm": 1.8339508771896362,
      "learning_rate": 3.0064999999999998e-05,
      "loss": 0.3918,
      "step": 6016
    },
    {
      "epoch": 0.3369358270803001,
      "grad_norm": 1.1436814069747925,
      "learning_rate": 3.0070000000000005e-05,
      "loss": 0.327,
      "step": 6017
    },
    {
      "epoch": 0.33699182439242914,
      "grad_norm": 1.402848720550537,
      "learning_rate": 3.0075000000000003e-05,
      "loss": 0.4811,
      "step": 6018
    },
    {
      "epoch": 0.33704782170455816,
      "grad_norm": 1.5328079462051392,
      "learning_rate": 3.0080000000000003e-05,
      "loss": 0.4196,
      "step": 6019
    },
    {
      "epoch": 0.3371038190166872,
      "grad_norm": 1.229766607284546,
      "learning_rate": 3.0085e-05,
      "loss": 0.4686,
      "step": 6020
    },
    {
      "epoch": 0.3371598163288162,
      "grad_norm": 1.4152299165725708,
      "learning_rate": 3.009e-05,
      "loss": 0.3973,
      "step": 6021
    },
    {
      "epoch": 0.3372158136409452,
      "grad_norm": 1.3779040575027466,
      "learning_rate": 3.0095000000000002e-05,
      "loss": 0.5503,
      "step": 6022
    },
    {
      "epoch": 0.33727181095307424,
      "grad_norm": 1.532416820526123,
      "learning_rate": 3.01e-05,
      "loss": 0.4722,
      "step": 6023
    },
    {
      "epoch": 0.33732780826520326,
      "grad_norm": 1.2227298021316528,
      "learning_rate": 3.0105e-05,
      "loss": 0.3752,
      "step": 6024
    },
    {
      "epoch": 0.3373838055773323,
      "grad_norm": 1.1379321813583374,
      "learning_rate": 3.0109999999999998e-05,
      "loss": 0.5022,
      "step": 6025
    },
    {
      "epoch": 0.3374398028894613,
      "grad_norm": 1.117943525314331,
      "learning_rate": 3.0115e-05,
      "loss": 0.4367,
      "step": 6026
    },
    {
      "epoch": 0.3374958002015903,
      "grad_norm": 1.1439199447631836,
      "learning_rate": 3.0120000000000003e-05,
      "loss": 0.372,
      "step": 6027
    },
    {
      "epoch": 0.33755179751371933,
      "grad_norm": 1.347422480583191,
      "learning_rate": 3.0125000000000004e-05,
      "loss": 0.5321,
      "step": 6028
    },
    {
      "epoch": 0.33760779482584835,
      "grad_norm": 1.2958378791809082,
      "learning_rate": 3.013e-05,
      "loss": 0.6347,
      "step": 6029
    },
    {
      "epoch": 0.33766379213797737,
      "grad_norm": 1.301550269126892,
      "learning_rate": 3.0135000000000002e-05,
      "loss": 0.3741,
      "step": 6030
    },
    {
      "epoch": 0.3377197894501064,
      "grad_norm": 1.3525784015655518,
      "learning_rate": 3.0140000000000003e-05,
      "loss": 0.3619,
      "step": 6031
    },
    {
      "epoch": 0.3377757867622354,
      "grad_norm": 1.347469449043274,
      "learning_rate": 3.0145e-05,
      "loss": 0.4806,
      "step": 6032
    },
    {
      "epoch": 0.3378317840743644,
      "grad_norm": 1.193375825881958,
      "learning_rate": 3.015e-05,
      "loss": 0.4192,
      "step": 6033
    },
    {
      "epoch": 0.33788778138649345,
      "grad_norm": 1.3086562156677246,
      "learning_rate": 3.0155e-05,
      "loss": 0.3881,
      "step": 6034
    },
    {
      "epoch": 0.33794377869862247,
      "grad_norm": 1.2944972515106201,
      "learning_rate": 3.016e-05,
      "loss": 0.4319,
      "step": 6035
    },
    {
      "epoch": 0.3379997760107515,
      "grad_norm": 1.7294611930847168,
      "learning_rate": 3.0165e-05,
      "loss": 0.4066,
      "step": 6036
    },
    {
      "epoch": 0.3380557733228805,
      "grad_norm": 1.4674289226531982,
      "learning_rate": 3.0170000000000004e-05,
      "loss": 0.4316,
      "step": 6037
    },
    {
      "epoch": 0.3381117706350095,
      "grad_norm": 1.7740062475204468,
      "learning_rate": 3.0175e-05,
      "loss": 0.402,
      "step": 6038
    },
    {
      "epoch": 0.33816776794713854,
      "grad_norm": 1.2527861595153809,
      "learning_rate": 3.0180000000000002e-05,
      "loss": 0.3912,
      "step": 6039
    },
    {
      "epoch": 0.33822376525926756,
      "grad_norm": 1.3782498836517334,
      "learning_rate": 3.0185000000000003e-05,
      "loss": 0.5425,
      "step": 6040
    },
    {
      "epoch": 0.3382797625713966,
      "grad_norm": 1.044850468635559,
      "learning_rate": 3.019e-05,
      "loss": 0.4195,
      "step": 6041
    },
    {
      "epoch": 0.3383357598835256,
      "grad_norm": 1.184483528137207,
      "learning_rate": 3.0195e-05,
      "loss": 0.3694,
      "step": 6042
    },
    {
      "epoch": 0.3383917571956546,
      "grad_norm": 5.612485885620117,
      "learning_rate": 3.02e-05,
      "loss": 0.6494,
      "step": 6043
    },
    {
      "epoch": 0.33844775450778364,
      "grad_norm": 1.1050502061843872,
      "learning_rate": 3.0205e-05,
      "loss": 0.4542,
      "step": 6044
    },
    {
      "epoch": 0.33850375181991266,
      "grad_norm": 1.2551862001419067,
      "learning_rate": 3.021e-05,
      "loss": 0.4103,
      "step": 6045
    },
    {
      "epoch": 0.3385597491320417,
      "grad_norm": 1.382351040840149,
      "learning_rate": 3.0214999999999998e-05,
      "loss": 0.4621,
      "step": 6046
    },
    {
      "epoch": 0.3386157464441707,
      "grad_norm": 0.9998094439506531,
      "learning_rate": 3.0220000000000005e-05,
      "loss": 0.3793,
      "step": 6047
    },
    {
      "epoch": 0.3386717437562997,
      "grad_norm": 1.1794143915176392,
      "learning_rate": 3.0225000000000003e-05,
      "loss": 0.3553,
      "step": 6048
    },
    {
      "epoch": 0.33872774106842873,
      "grad_norm": 1.4121116399765015,
      "learning_rate": 3.0230000000000004e-05,
      "loss": 0.3912,
      "step": 6049
    },
    {
      "epoch": 0.33878373838055775,
      "grad_norm": 1.1807981729507446,
      "learning_rate": 3.0235e-05,
      "loss": 0.4023,
      "step": 6050
    },
    {
      "epoch": 0.33883973569268677,
      "grad_norm": 1.160921573638916,
      "learning_rate": 3.0240000000000002e-05,
      "loss": 0.3538,
      "step": 6051
    },
    {
      "epoch": 0.3388957330048158,
      "grad_norm": 1.1883474588394165,
      "learning_rate": 3.0245000000000003e-05,
      "loss": 0.3711,
      "step": 6052
    },
    {
      "epoch": 0.3389517303169448,
      "grad_norm": 1.2117105722427368,
      "learning_rate": 3.025e-05,
      "loss": 0.3819,
      "step": 6053
    },
    {
      "epoch": 0.33900772762907383,
      "grad_norm": 1.452558159828186,
      "learning_rate": 3.0255e-05,
      "loss": 0.4038,
      "step": 6054
    },
    {
      "epoch": 0.33906372494120285,
      "grad_norm": 1.2886942625045776,
      "learning_rate": 3.0259999999999998e-05,
      "loss": 0.5197,
      "step": 6055
    },
    {
      "epoch": 0.33911972225333187,
      "grad_norm": 1.525838017463684,
      "learning_rate": 3.0265e-05,
      "loss": 0.4201,
      "step": 6056
    },
    {
      "epoch": 0.33917571956546083,
      "grad_norm": 1.502649188041687,
      "learning_rate": 3.0270000000000003e-05,
      "loss": 0.5808,
      "step": 6057
    },
    {
      "epoch": 0.33923171687758985,
      "grad_norm": 1.1106334924697876,
      "learning_rate": 3.0275000000000004e-05,
      "loss": 0.4563,
      "step": 6058
    },
    {
      "epoch": 0.33928771418971887,
      "grad_norm": 1.3896900415420532,
      "learning_rate": 3.028e-05,
      "loss": 0.4421,
      "step": 6059
    },
    {
      "epoch": 0.3393437115018479,
      "grad_norm": 1.3931236267089844,
      "learning_rate": 3.0285000000000002e-05,
      "loss": 0.3937,
      "step": 6060
    },
    {
      "epoch": 0.3393997088139769,
      "grad_norm": 1.1702464818954468,
      "learning_rate": 3.0290000000000003e-05,
      "loss": 0.4756,
      "step": 6061
    },
    {
      "epoch": 0.3394557061261059,
      "grad_norm": 1.3623554706573486,
      "learning_rate": 3.0295e-05,
      "loss": 0.4031,
      "step": 6062
    },
    {
      "epoch": 0.33951170343823495,
      "grad_norm": 1.37270188331604,
      "learning_rate": 3.03e-05,
      "loss": 0.503,
      "step": 6063
    },
    {
      "epoch": 0.33956770075036397,
      "grad_norm": 1.3576231002807617,
      "learning_rate": 3.0305e-05,
      "loss": 0.5454,
      "step": 6064
    },
    {
      "epoch": 0.339623698062493,
      "grad_norm": 1.1892614364624023,
      "learning_rate": 3.031e-05,
      "loss": 0.4471,
      "step": 6065
    },
    {
      "epoch": 0.339679695374622,
      "grad_norm": 1.1943871974945068,
      "learning_rate": 3.0315e-05,
      "loss": 0.5917,
      "step": 6066
    },
    {
      "epoch": 0.339735692686751,
      "grad_norm": 1.1706382036209106,
      "learning_rate": 3.0320000000000004e-05,
      "loss": 0.5538,
      "step": 6067
    },
    {
      "epoch": 0.33979168999888004,
      "grad_norm": 1.2243510484695435,
      "learning_rate": 3.0325000000000002e-05,
      "loss": 0.2955,
      "step": 6068
    },
    {
      "epoch": 0.33984768731100906,
      "grad_norm": 1.0546960830688477,
      "learning_rate": 3.0330000000000003e-05,
      "loss": 0.3759,
      "step": 6069
    },
    {
      "epoch": 0.3399036846231381,
      "grad_norm": 1.1734250783920288,
      "learning_rate": 3.0335000000000003e-05,
      "loss": 0.3993,
      "step": 6070
    },
    {
      "epoch": 0.3399596819352671,
      "grad_norm": 1.121383547782898,
      "learning_rate": 3.034e-05,
      "loss": 0.4006,
      "step": 6071
    },
    {
      "epoch": 0.3400156792473961,
      "grad_norm": 1.3144326210021973,
      "learning_rate": 3.0345e-05,
      "loss": 0.4595,
      "step": 6072
    },
    {
      "epoch": 0.34007167655952514,
      "grad_norm": 1.1103761196136475,
      "learning_rate": 3.035e-05,
      "loss": 0.3824,
      "step": 6073
    },
    {
      "epoch": 0.34012767387165416,
      "grad_norm": 1.0838195085525513,
      "learning_rate": 3.0355e-05,
      "loss": 0.344,
      "step": 6074
    },
    {
      "epoch": 0.3401836711837832,
      "grad_norm": 1.0776463747024536,
      "learning_rate": 3.036e-05,
      "loss": 0.3973,
      "step": 6075
    },
    {
      "epoch": 0.3402396684959122,
      "grad_norm": 1.2838622331619263,
      "learning_rate": 3.0364999999999998e-05,
      "loss": 0.5617,
      "step": 6076
    },
    {
      "epoch": 0.3402956658080412,
      "grad_norm": 1.385256290435791,
      "learning_rate": 3.0370000000000006e-05,
      "loss": 0.7115,
      "step": 6077
    },
    {
      "epoch": 0.34035166312017023,
      "grad_norm": 1.182480812072754,
      "learning_rate": 3.0375000000000003e-05,
      "loss": 0.4803,
      "step": 6078
    },
    {
      "epoch": 0.34040766043229925,
      "grad_norm": 1.7291700839996338,
      "learning_rate": 3.0380000000000004e-05,
      "loss": 0.5289,
      "step": 6079
    },
    {
      "epoch": 0.34046365774442827,
      "grad_norm": 1.4423400163650513,
      "learning_rate": 3.0385e-05,
      "loss": 0.521,
      "step": 6080
    },
    {
      "epoch": 0.3405196550565573,
      "grad_norm": 2.5962367057800293,
      "learning_rate": 3.0390000000000002e-05,
      "loss": 0.5251,
      "step": 6081
    },
    {
      "epoch": 0.3405756523686863,
      "grad_norm": 1.3955833911895752,
      "learning_rate": 3.0395000000000003e-05,
      "loss": 0.4798,
      "step": 6082
    },
    {
      "epoch": 0.34063164968081533,
      "grad_norm": 1.157979130744934,
      "learning_rate": 3.04e-05,
      "loss": 0.3445,
      "step": 6083
    },
    {
      "epoch": 0.34068764699294435,
      "grad_norm": 1.4732646942138672,
      "learning_rate": 3.0405e-05,
      "loss": 0.6209,
      "step": 6084
    },
    {
      "epoch": 0.34074364430507337,
      "grad_norm": 1.9608231782913208,
      "learning_rate": 3.041e-05,
      "loss": 0.5513,
      "step": 6085
    },
    {
      "epoch": 0.3407996416172024,
      "grad_norm": 1.2772008180618286,
      "learning_rate": 3.0415e-05,
      "loss": 0.4158,
      "step": 6086
    },
    {
      "epoch": 0.3408556389293314,
      "grad_norm": 1.2368332147598267,
      "learning_rate": 3.0420000000000004e-05,
      "loss": 0.5146,
      "step": 6087
    },
    {
      "epoch": 0.3409116362414604,
      "grad_norm": 1.4972248077392578,
      "learning_rate": 3.0425000000000004e-05,
      "loss": 0.5596,
      "step": 6088
    },
    {
      "epoch": 0.34096763355358944,
      "grad_norm": 1.2302608489990234,
      "learning_rate": 3.0430000000000002e-05,
      "loss": 0.3766,
      "step": 6089
    },
    {
      "epoch": 0.34102363086571846,
      "grad_norm": 1.5643738508224487,
      "learning_rate": 3.0435000000000003e-05,
      "loss": 0.3371,
      "step": 6090
    },
    {
      "epoch": 0.3410796281778475,
      "grad_norm": 1.244739294052124,
      "learning_rate": 3.0440000000000003e-05,
      "loss": 0.4079,
      "step": 6091
    },
    {
      "epoch": 0.3411356254899765,
      "grad_norm": 1.2407468557357788,
      "learning_rate": 3.0445e-05,
      "loss": 0.4041,
      "step": 6092
    },
    {
      "epoch": 0.3411916228021055,
      "grad_norm": 1.6655876636505127,
      "learning_rate": 3.045e-05,
      "loss": 0.3608,
      "step": 6093
    },
    {
      "epoch": 0.34124762011423454,
      "grad_norm": 1.256253957748413,
      "learning_rate": 3.0455e-05,
      "loss": 0.4899,
      "step": 6094
    },
    {
      "epoch": 0.34130361742636356,
      "grad_norm": 1.1335060596466064,
      "learning_rate": 3.046e-05,
      "loss": 0.4612,
      "step": 6095
    },
    {
      "epoch": 0.3413596147384926,
      "grad_norm": 1.1727112531661987,
      "learning_rate": 3.0465e-05,
      "loss": 0.3644,
      "step": 6096
    },
    {
      "epoch": 0.3414156120506216,
      "grad_norm": 1.1319737434387207,
      "learning_rate": 3.0470000000000005e-05,
      "loss": 0.3832,
      "step": 6097
    },
    {
      "epoch": 0.34147160936275056,
      "grad_norm": 1.1234557628631592,
      "learning_rate": 3.0475000000000002e-05,
      "loss": 0.3799,
      "step": 6098
    },
    {
      "epoch": 0.3415276066748796,
      "grad_norm": 1.4665559530258179,
      "learning_rate": 3.0480000000000003e-05,
      "loss": 0.5543,
      "step": 6099
    },
    {
      "epoch": 0.3415836039870086,
      "grad_norm": 1.3194949626922607,
      "learning_rate": 3.0485000000000004e-05,
      "loss": 0.4193,
      "step": 6100
    },
    {
      "epoch": 0.3416396012991376,
      "grad_norm": 1.2255374193191528,
      "learning_rate": 3.049e-05,
      "loss": 0.4217,
      "step": 6101
    },
    {
      "epoch": 0.34169559861126664,
      "grad_norm": 1.2952872514724731,
      "learning_rate": 3.0495000000000002e-05,
      "loss": 0.4849,
      "step": 6102
    },
    {
      "epoch": 0.34175159592339566,
      "grad_norm": 1.278414011001587,
      "learning_rate": 3.05e-05,
      "loss": 0.5239,
      "step": 6103
    },
    {
      "epoch": 0.3418075932355247,
      "grad_norm": 1.3654206991195679,
      "learning_rate": 3.0505e-05,
      "loss": 0.3876,
      "step": 6104
    },
    {
      "epoch": 0.3418635905476537,
      "grad_norm": 1.132808804512024,
      "learning_rate": 3.051e-05,
      "loss": 0.3932,
      "step": 6105
    },
    {
      "epoch": 0.3419195878597827,
      "grad_norm": 1.2543237209320068,
      "learning_rate": 3.0515e-05,
      "loss": 0.4998,
      "step": 6106
    },
    {
      "epoch": 0.34197558517191173,
      "grad_norm": 1.3329524993896484,
      "learning_rate": 3.0520000000000006e-05,
      "loss": 0.4027,
      "step": 6107
    },
    {
      "epoch": 0.34203158248404075,
      "grad_norm": 1.3539350032806396,
      "learning_rate": 3.0525e-05,
      "loss": 0.5102,
      "step": 6108
    },
    {
      "epoch": 0.34208757979616977,
      "grad_norm": 1.3504351377487183,
      "learning_rate": 3.053e-05,
      "loss": 0.3668,
      "step": 6109
    },
    {
      "epoch": 0.3421435771082988,
      "grad_norm": 1.2522965669631958,
      "learning_rate": 3.0535000000000005e-05,
      "loss": 0.4975,
      "step": 6110
    },
    {
      "epoch": 0.3421995744204278,
      "grad_norm": 1.115391492843628,
      "learning_rate": 3.054e-05,
      "loss": 0.3742,
      "step": 6111
    },
    {
      "epoch": 0.34225557173255683,
      "grad_norm": 1.205297827720642,
      "learning_rate": 3.0545e-05,
      "loss": 0.3266,
      "step": 6112
    },
    {
      "epoch": 0.34231156904468585,
      "grad_norm": 1.2326608896255493,
      "learning_rate": 3.0550000000000004e-05,
      "loss": 0.3953,
      "step": 6113
    },
    {
      "epoch": 0.34236756635681487,
      "grad_norm": 1.3323166370391846,
      "learning_rate": 3.0555e-05,
      "loss": 0.4736,
      "step": 6114
    },
    {
      "epoch": 0.3424235636689439,
      "grad_norm": 1.5239416360855103,
      "learning_rate": 3.056e-05,
      "loss": 0.4383,
      "step": 6115
    },
    {
      "epoch": 0.3424795609810729,
      "grad_norm": 1.2085906267166138,
      "learning_rate": 3.0564999999999996e-05,
      "loss": 0.3913,
      "step": 6116
    },
    {
      "epoch": 0.3425355582932019,
      "grad_norm": 1.346198320388794,
      "learning_rate": 3.057000000000001e-05,
      "loss": 0.3611,
      "step": 6117
    },
    {
      "epoch": 0.34259155560533094,
      "grad_norm": 1.3381444215774536,
      "learning_rate": 3.0575000000000005e-05,
      "loss": 0.3948,
      "step": 6118
    },
    {
      "epoch": 0.34264755291745996,
      "grad_norm": 1.4198280572891235,
      "learning_rate": 3.058e-05,
      "loss": 0.5018,
      "step": 6119
    },
    {
      "epoch": 0.342703550229589,
      "grad_norm": 1.2343220710754395,
      "learning_rate": 3.0585e-05,
      "loss": 0.4075,
      "step": 6120
    },
    {
      "epoch": 0.342759547541718,
      "grad_norm": 1.4251677989959717,
      "learning_rate": 3.0590000000000004e-05,
      "loss": 0.3703,
      "step": 6121
    },
    {
      "epoch": 0.342815544853847,
      "grad_norm": 1.425933837890625,
      "learning_rate": 3.0595e-05,
      "loss": 0.5023,
      "step": 6122
    },
    {
      "epoch": 0.34287154216597604,
      "grad_norm": 1.3435081243515015,
      "learning_rate": 3.06e-05,
      "loss": 0.422,
      "step": 6123
    },
    {
      "epoch": 0.34292753947810506,
      "grad_norm": 1.4446288347244263,
      "learning_rate": 3.0605e-05,
      "loss": 0.5023,
      "step": 6124
    },
    {
      "epoch": 0.3429835367902341,
      "grad_norm": 1.1390122175216675,
      "learning_rate": 3.061e-05,
      "loss": 0.363,
      "step": 6125
    },
    {
      "epoch": 0.3430395341023631,
      "grad_norm": 1.2695108652114868,
      "learning_rate": 3.0615e-05,
      "loss": 0.4795,
      "step": 6126
    },
    {
      "epoch": 0.3430955314144921,
      "grad_norm": 1.1526217460632324,
      "learning_rate": 3.062e-05,
      "loss": 0.4332,
      "step": 6127
    },
    {
      "epoch": 0.34315152872662114,
      "grad_norm": 1.4268572330474854,
      "learning_rate": 3.0625000000000006e-05,
      "loss": 0.4968,
      "step": 6128
    },
    {
      "epoch": 0.34320752603875015,
      "grad_norm": 1.3458688259124756,
      "learning_rate": 3.063e-05,
      "loss": 0.4948,
      "step": 6129
    },
    {
      "epoch": 0.3432635233508792,
      "grad_norm": 1.4268699884414673,
      "learning_rate": 3.0635e-05,
      "loss": 0.5031,
      "step": 6130
    },
    {
      "epoch": 0.3433195206630082,
      "grad_norm": 1.0886454582214355,
      "learning_rate": 3.0640000000000005e-05,
      "loss": 0.3386,
      "step": 6131
    },
    {
      "epoch": 0.3433755179751372,
      "grad_norm": 1.8940868377685547,
      "learning_rate": 3.0645e-05,
      "loss": 0.4986,
      "step": 6132
    },
    {
      "epoch": 0.34343151528726623,
      "grad_norm": 1.2075221538543701,
      "learning_rate": 3.065e-05,
      "loss": 0.3753,
      "step": 6133
    },
    {
      "epoch": 0.34348751259939525,
      "grad_norm": 1.0235775709152222,
      "learning_rate": 3.0655e-05,
      "loss": 0.3847,
      "step": 6134
    },
    {
      "epoch": 0.34354350991152427,
      "grad_norm": 1.2436648607254028,
      "learning_rate": 3.066e-05,
      "loss": 0.5126,
      "step": 6135
    },
    {
      "epoch": 0.3435995072236533,
      "grad_norm": 9.105645179748535,
      "learning_rate": 3.0665e-05,
      "loss": 0.4121,
      "step": 6136
    },
    {
      "epoch": 0.3436555045357823,
      "grad_norm": 1.3691940307617188,
      "learning_rate": 3.0669999999999996e-05,
      "loss": 0.4754,
      "step": 6137
    },
    {
      "epoch": 0.3437115018479113,
      "grad_norm": 1.1193677186965942,
      "learning_rate": 3.067500000000001e-05,
      "loss": 0.3744,
      "step": 6138
    },
    {
      "epoch": 0.3437674991600403,
      "grad_norm": 1.0840986967086792,
      "learning_rate": 3.0680000000000004e-05,
      "loss": 0.4402,
      "step": 6139
    },
    {
      "epoch": 0.3438234964721693,
      "grad_norm": 1.5513265132904053,
      "learning_rate": 3.0685e-05,
      "loss": 0.3624,
      "step": 6140
    },
    {
      "epoch": 0.34387949378429833,
      "grad_norm": 1.279994010925293,
      "learning_rate": 3.069e-05,
      "loss": 0.4096,
      "step": 6141
    },
    {
      "epoch": 0.34393549109642735,
      "grad_norm": 1.1125075817108154,
      "learning_rate": 3.0695000000000003e-05,
      "loss": 0.4031,
      "step": 6142
    },
    {
      "epoch": 0.34399148840855637,
      "grad_norm": 1.304520606994629,
      "learning_rate": 3.07e-05,
      "loss": 0.367,
      "step": 6143
    },
    {
      "epoch": 0.3440474857206854,
      "grad_norm": 1.2914345264434814,
      "learning_rate": 3.0705e-05,
      "loss": 0.4538,
      "step": 6144
    },
    {
      "epoch": 0.3441034830328144,
      "grad_norm": 1.1114362478256226,
      "learning_rate": 3.071e-05,
      "loss": 0.3721,
      "step": 6145
    },
    {
      "epoch": 0.3441594803449434,
      "grad_norm": 1.2352871894836426,
      "learning_rate": 3.0715e-05,
      "loss": 0.4383,
      "step": 6146
    },
    {
      "epoch": 0.34421547765707244,
      "grad_norm": 1.0656942129135132,
      "learning_rate": 3.072e-05,
      "loss": 0.4117,
      "step": 6147
    },
    {
      "epoch": 0.34427147496920146,
      "grad_norm": 1.1222028732299805,
      "learning_rate": 3.0725e-05,
      "loss": 0.3683,
      "step": 6148
    },
    {
      "epoch": 0.3443274722813305,
      "grad_norm": 1.280902624130249,
      "learning_rate": 3.0730000000000006e-05,
      "loss": 0.5762,
      "step": 6149
    },
    {
      "epoch": 0.3443834695934595,
      "grad_norm": 1.1570007801055908,
      "learning_rate": 3.0735e-05,
      "loss": 0.366,
      "step": 6150
    },
    {
      "epoch": 0.3444394669055885,
      "grad_norm": 1.252055048942566,
      "learning_rate": 3.074e-05,
      "loss": 0.3943,
      "step": 6151
    },
    {
      "epoch": 0.34449546421771754,
      "grad_norm": 1.2108405828475952,
      "learning_rate": 3.0745000000000005e-05,
      "loss": 0.3738,
      "step": 6152
    },
    {
      "epoch": 0.34455146152984656,
      "grad_norm": 1.0659451484680176,
      "learning_rate": 3.075e-05,
      "loss": 0.4407,
      "step": 6153
    },
    {
      "epoch": 0.3446074588419756,
      "grad_norm": 1.2742671966552734,
      "learning_rate": 3.0755e-05,
      "loss": 0.409,
      "step": 6154
    },
    {
      "epoch": 0.3446634561541046,
      "grad_norm": 1.1289572715759277,
      "learning_rate": 3.076e-05,
      "loss": 0.4658,
      "step": 6155
    },
    {
      "epoch": 0.3447194534662336,
      "grad_norm": 1.2440177202224731,
      "learning_rate": 3.0765e-05,
      "loss": 0.4304,
      "step": 6156
    },
    {
      "epoch": 0.34477545077836264,
      "grad_norm": 1.1720653772354126,
      "learning_rate": 3.077e-05,
      "loss": 0.4095,
      "step": 6157
    },
    {
      "epoch": 0.34483144809049165,
      "grad_norm": 7.197729110717773,
      "learning_rate": 3.0775e-05,
      "loss": 0.5063,
      "step": 6158
    },
    {
      "epoch": 0.3448874454026207,
      "grad_norm": 1.2122024297714233,
      "learning_rate": 3.078e-05,
      "loss": 0.4376,
      "step": 6159
    },
    {
      "epoch": 0.3449434427147497,
      "grad_norm": 1.476108193397522,
      "learning_rate": 3.0785000000000004e-05,
      "loss": 0.5088,
      "step": 6160
    },
    {
      "epoch": 0.3449994400268787,
      "grad_norm": 1.1375744342803955,
      "learning_rate": 3.079e-05,
      "loss": 0.4141,
      "step": 6161
    },
    {
      "epoch": 0.34505543733900773,
      "grad_norm": 1.2847084999084473,
      "learning_rate": 3.0795e-05,
      "loss": 0.458,
      "step": 6162
    },
    {
      "epoch": 0.34511143465113675,
      "grad_norm": 1.247577428817749,
      "learning_rate": 3.08e-05,
      "loss": 0.4325,
      "step": 6163
    },
    {
      "epoch": 0.34516743196326577,
      "grad_norm": 1.1049706935882568,
      "learning_rate": 3.0805e-05,
      "loss": 0.4173,
      "step": 6164
    },
    {
      "epoch": 0.3452234292753948,
      "grad_norm": 1.5211737155914307,
      "learning_rate": 3.081e-05,
      "loss": 0.6595,
      "step": 6165
    },
    {
      "epoch": 0.3452794265875238,
      "grad_norm": 1.4209784269332886,
      "learning_rate": 3.0815e-05,
      "loss": 0.461,
      "step": 6166
    },
    {
      "epoch": 0.3453354238996528,
      "grad_norm": 1.1894135475158691,
      "learning_rate": 3.082e-05,
      "loss": 0.6291,
      "step": 6167
    },
    {
      "epoch": 0.34539142121178185,
      "grad_norm": 1.2749963998794556,
      "learning_rate": 3.0825000000000004e-05,
      "loss": 0.4693,
      "step": 6168
    },
    {
      "epoch": 0.34544741852391087,
      "grad_norm": 1.2753534317016602,
      "learning_rate": 3.083e-05,
      "loss": 0.4812,
      "step": 6169
    },
    {
      "epoch": 0.3455034158360399,
      "grad_norm": 1.4958391189575195,
      "learning_rate": 3.0835000000000005e-05,
      "loss": 0.4399,
      "step": 6170
    },
    {
      "epoch": 0.3455594131481689,
      "grad_norm": 1.2740752696990967,
      "learning_rate": 3.084e-05,
      "loss": 0.3982,
      "step": 6171
    },
    {
      "epoch": 0.3456154104602979,
      "grad_norm": 1.0687915086746216,
      "learning_rate": 3.0845e-05,
      "loss": 0.3227,
      "step": 6172
    },
    {
      "epoch": 0.34567140777242694,
      "grad_norm": 1.1849191188812256,
      "learning_rate": 3.0850000000000004e-05,
      "loss": 0.3809,
      "step": 6173
    },
    {
      "epoch": 0.34572740508455596,
      "grad_norm": 1.2271841764450073,
      "learning_rate": 3.0855e-05,
      "loss": 0.4786,
      "step": 6174
    },
    {
      "epoch": 0.345783402396685,
      "grad_norm": 1.063286304473877,
      "learning_rate": 3.086e-05,
      "loss": 0.3983,
      "step": 6175
    },
    {
      "epoch": 0.345839399708814,
      "grad_norm": 0.9903095364570618,
      "learning_rate": 3.0865e-05,
      "loss": 0.3459,
      "step": 6176
    },
    {
      "epoch": 0.345895397020943,
      "grad_norm": 1.322891354560852,
      "learning_rate": 3.087e-05,
      "loss": 0.4494,
      "step": 6177
    },
    {
      "epoch": 0.34595139433307204,
      "grad_norm": 1.1799235343933105,
      "learning_rate": 3.0875000000000005e-05,
      "loss": 0.3814,
      "step": 6178
    },
    {
      "epoch": 0.34600739164520106,
      "grad_norm": 1.1407090425491333,
      "learning_rate": 3.088e-05,
      "loss": 0.409,
      "step": 6179
    },
    {
      "epoch": 0.3460633889573301,
      "grad_norm": 1.314505934715271,
      "learning_rate": 3.0885e-05,
      "loss": 0.412,
      "step": 6180
    },
    {
      "epoch": 0.34611938626945904,
      "grad_norm": 1.1176612377166748,
      "learning_rate": 3.0890000000000004e-05,
      "loss": 0.4532,
      "step": 6181
    },
    {
      "epoch": 0.34617538358158806,
      "grad_norm": 1.3744513988494873,
      "learning_rate": 3.0895e-05,
      "loss": 0.5124,
      "step": 6182
    },
    {
      "epoch": 0.3462313808937171,
      "grad_norm": 1.1842596530914307,
      "learning_rate": 3.09e-05,
      "loss": 0.451,
      "step": 6183
    },
    {
      "epoch": 0.3462873782058461,
      "grad_norm": 1.231649398803711,
      "learning_rate": 3.0905e-05,
      "loss": 0.3933,
      "step": 6184
    },
    {
      "epoch": 0.3463433755179751,
      "grad_norm": 1.2014415264129639,
      "learning_rate": 3.091e-05,
      "loss": 0.3315,
      "step": 6185
    },
    {
      "epoch": 0.34639937283010414,
      "grad_norm": 5.304755687713623,
      "learning_rate": 3.0915e-05,
      "loss": 0.4169,
      "step": 6186
    },
    {
      "epoch": 0.34645537014223315,
      "grad_norm": 1.3249987363815308,
      "learning_rate": 3.092e-05,
      "loss": 0.4412,
      "step": 6187
    },
    {
      "epoch": 0.3465113674543622,
      "grad_norm": 3.466057062149048,
      "learning_rate": 3.0925000000000006e-05,
      "loss": 0.4771,
      "step": 6188
    },
    {
      "epoch": 0.3465673647664912,
      "grad_norm": 1.0502848625183105,
      "learning_rate": 3.0930000000000004e-05,
      "loss": 0.3134,
      "step": 6189
    },
    {
      "epoch": 0.3466233620786202,
      "grad_norm": 1.3575304746627808,
      "learning_rate": 3.0935e-05,
      "loss": 0.4014,
      "step": 6190
    },
    {
      "epoch": 0.34667935939074923,
      "grad_norm": 1.1503523588180542,
      "learning_rate": 3.0940000000000005e-05,
      "loss": 0.4059,
      "step": 6191
    },
    {
      "epoch": 0.34673535670287825,
      "grad_norm": 1.267883062362671,
      "learning_rate": 3.0945e-05,
      "loss": 0.5432,
      "step": 6192
    },
    {
      "epoch": 0.34679135401500727,
      "grad_norm": 2.9755451679229736,
      "learning_rate": 3.095e-05,
      "loss": 0.3337,
      "step": 6193
    },
    {
      "epoch": 0.3468473513271363,
      "grad_norm": 1.181565523147583,
      "learning_rate": 3.0955e-05,
      "loss": 0.3977,
      "step": 6194
    },
    {
      "epoch": 0.3469033486392653,
      "grad_norm": 1.145254373550415,
      "learning_rate": 3.096e-05,
      "loss": 0.4545,
      "step": 6195
    },
    {
      "epoch": 0.3469593459513943,
      "grad_norm": 1.4250879287719727,
      "learning_rate": 3.0965e-05,
      "loss": 0.484,
      "step": 6196
    },
    {
      "epoch": 0.34701534326352335,
      "grad_norm": 1.1087721586227417,
      "learning_rate": 3.0969999999999997e-05,
      "loss": 0.458,
      "step": 6197
    },
    {
      "epoch": 0.34707134057565237,
      "grad_norm": 1.4464209079742432,
      "learning_rate": 3.0975e-05,
      "loss": 0.5284,
      "step": 6198
    },
    {
      "epoch": 0.3471273378877814,
      "grad_norm": 1.2910740375518799,
      "learning_rate": 3.0980000000000005e-05,
      "loss": 0.3903,
      "step": 6199
    },
    {
      "epoch": 0.3471833351999104,
      "grad_norm": 1.1853680610656738,
      "learning_rate": 3.0985e-05,
      "loss": 0.3896,
      "step": 6200
    },
    {
      "epoch": 0.3472393325120394,
      "grad_norm": 1.0837509632110596,
      "learning_rate": 3.099e-05,
      "loss": 0.427,
      "step": 6201
    },
    {
      "epoch": 0.34729532982416844,
      "grad_norm": 1.038487434387207,
      "learning_rate": 3.0995000000000004e-05,
      "loss": 0.4323,
      "step": 6202
    },
    {
      "epoch": 0.34735132713629746,
      "grad_norm": 1.403473138809204,
      "learning_rate": 3.1e-05,
      "loss": 0.4597,
      "step": 6203
    },
    {
      "epoch": 0.3474073244484265,
      "grad_norm": 1.3064563274383545,
      "learning_rate": 3.1005e-05,
      "loss": 0.5115,
      "step": 6204
    },
    {
      "epoch": 0.3474633217605555,
      "grad_norm": 1.1536040306091309,
      "learning_rate": 3.101e-05,
      "loss": 0.3336,
      "step": 6205
    },
    {
      "epoch": 0.3475193190726845,
      "grad_norm": 1.3112775087356567,
      "learning_rate": 3.1015e-05,
      "loss": 0.4123,
      "step": 6206
    },
    {
      "epoch": 0.34757531638481354,
      "grad_norm": 1.2330294847488403,
      "learning_rate": 3.102e-05,
      "loss": 0.3784,
      "step": 6207
    },
    {
      "epoch": 0.34763131369694256,
      "grad_norm": 1.2423793077468872,
      "learning_rate": 3.1025e-05,
      "loss": 0.4646,
      "step": 6208
    },
    {
      "epoch": 0.3476873110090716,
      "grad_norm": 1.0161997079849243,
      "learning_rate": 3.1030000000000006e-05,
      "loss": 0.3805,
      "step": 6209
    },
    {
      "epoch": 0.3477433083212006,
      "grad_norm": 1.2694252729415894,
      "learning_rate": 3.1035000000000004e-05,
      "loss": 0.385,
      "step": 6210
    },
    {
      "epoch": 0.3477993056333296,
      "grad_norm": 1.343778133392334,
      "learning_rate": 3.104e-05,
      "loss": 0.4388,
      "step": 6211
    },
    {
      "epoch": 0.34785530294545863,
      "grad_norm": 1.2166775465011597,
      "learning_rate": 3.1045000000000005e-05,
      "loss": 0.3247,
      "step": 6212
    },
    {
      "epoch": 0.34791130025758765,
      "grad_norm": 1.2452661991119385,
      "learning_rate": 3.105e-05,
      "loss": 0.339,
      "step": 6213
    },
    {
      "epoch": 0.34796729756971667,
      "grad_norm": 1.1289180517196655,
      "learning_rate": 3.1055e-05,
      "loss": 0.4189,
      "step": 6214
    },
    {
      "epoch": 0.3480232948818457,
      "grad_norm": 0.9198927879333496,
      "learning_rate": 3.106e-05,
      "loss": 0.361,
      "step": 6215
    },
    {
      "epoch": 0.3480792921939747,
      "grad_norm": 1.1886038780212402,
      "learning_rate": 3.1065e-05,
      "loss": 0.5015,
      "step": 6216
    },
    {
      "epoch": 0.34813528950610373,
      "grad_norm": 1.1647417545318604,
      "learning_rate": 3.107e-05,
      "loss": 0.3268,
      "step": 6217
    },
    {
      "epoch": 0.34819128681823275,
      "grad_norm": 1.2943861484527588,
      "learning_rate": 3.1075e-05,
      "loss": 0.3943,
      "step": 6218
    },
    {
      "epoch": 0.34824728413036177,
      "grad_norm": 1.125791072845459,
      "learning_rate": 3.108e-05,
      "loss": 0.3445,
      "step": 6219
    },
    {
      "epoch": 0.3483032814424908,
      "grad_norm": 1.4532657861709595,
      "learning_rate": 3.1085000000000005e-05,
      "loss": 0.4646,
      "step": 6220
    },
    {
      "epoch": 0.3483592787546198,
      "grad_norm": 1.043227195739746,
      "learning_rate": 3.109e-05,
      "loss": 0.3693,
      "step": 6221
    },
    {
      "epoch": 0.34841527606674877,
      "grad_norm": 1.2873167991638184,
      "learning_rate": 3.1095e-05,
      "loss": 0.4276,
      "step": 6222
    },
    {
      "epoch": 0.3484712733788778,
      "grad_norm": 1.4242085218429565,
      "learning_rate": 3.1100000000000004e-05,
      "loss": 0.5273,
      "step": 6223
    },
    {
      "epoch": 0.3485272706910068,
      "grad_norm": 1.1898281574249268,
      "learning_rate": 3.1105e-05,
      "loss": 0.4503,
      "step": 6224
    },
    {
      "epoch": 0.3485832680031358,
      "grad_norm": 1.1694384813308716,
      "learning_rate": 3.111e-05,
      "loss": 0.3438,
      "step": 6225
    },
    {
      "epoch": 0.34863926531526485,
      "grad_norm": 1.185025930404663,
      "learning_rate": 3.1115e-05,
      "loss": 0.337,
      "step": 6226
    },
    {
      "epoch": 0.34869526262739386,
      "grad_norm": 1.0493113994598389,
      "learning_rate": 3.112e-05,
      "loss": 0.3695,
      "step": 6227
    },
    {
      "epoch": 0.3487512599395229,
      "grad_norm": 1.441767692565918,
      "learning_rate": 3.1125000000000004e-05,
      "loss": 0.3575,
      "step": 6228
    },
    {
      "epoch": 0.3488072572516519,
      "grad_norm": 1.06027090549469,
      "learning_rate": 3.113e-05,
      "loss": 0.2929,
      "step": 6229
    },
    {
      "epoch": 0.3488632545637809,
      "grad_norm": 1.1034297943115234,
      "learning_rate": 3.1135000000000006e-05,
      "loss": 0.4501,
      "step": 6230
    },
    {
      "epoch": 0.34891925187590994,
      "grad_norm": 1.1654653549194336,
      "learning_rate": 3.1140000000000003e-05,
      "loss": 0.4183,
      "step": 6231
    },
    {
      "epoch": 0.34897524918803896,
      "grad_norm": 1.605057954788208,
      "learning_rate": 3.1145e-05,
      "loss": 0.4588,
      "step": 6232
    },
    {
      "epoch": 0.349031246500168,
      "grad_norm": 1.2390538454055786,
      "learning_rate": 3.115e-05,
      "loss": 0.4808,
      "step": 6233
    },
    {
      "epoch": 0.349087243812297,
      "grad_norm": 1.6129049062728882,
      "learning_rate": 3.1155e-05,
      "loss": 0.6286,
      "step": 6234
    },
    {
      "epoch": 0.349143241124426,
      "grad_norm": 1.207960605621338,
      "learning_rate": 3.116e-05,
      "loss": 0.5001,
      "step": 6235
    },
    {
      "epoch": 0.34919923843655504,
      "grad_norm": 1.1692736148834229,
      "learning_rate": 3.1165e-05,
      "loss": 0.3719,
      "step": 6236
    },
    {
      "epoch": 0.34925523574868406,
      "grad_norm": 1.7736616134643555,
      "learning_rate": 3.117e-05,
      "loss": 0.609,
      "step": 6237
    },
    {
      "epoch": 0.3493112330608131,
      "grad_norm": 1.7082831859588623,
      "learning_rate": 3.1175000000000006e-05,
      "loss": 0.4509,
      "step": 6238
    },
    {
      "epoch": 0.3493672303729421,
      "grad_norm": 2.39262056350708,
      "learning_rate": 3.118e-05,
      "loss": 0.6077,
      "step": 6239
    },
    {
      "epoch": 0.3494232276850711,
      "grad_norm": 1.184689998626709,
      "learning_rate": 3.1185e-05,
      "loss": 0.4831,
      "step": 6240
    },
    {
      "epoch": 0.34947922499720013,
      "grad_norm": 1.4418725967407227,
      "learning_rate": 3.1190000000000005e-05,
      "loss": 0.5638,
      "step": 6241
    },
    {
      "epoch": 0.34953522230932915,
      "grad_norm": 1.3676230907440186,
      "learning_rate": 3.1195e-05,
      "loss": 0.5351,
      "step": 6242
    },
    {
      "epoch": 0.34959121962145817,
      "grad_norm": 1.1918115615844727,
      "learning_rate": 3.12e-05,
      "loss": 0.488,
      "step": 6243
    },
    {
      "epoch": 0.3496472169335872,
      "grad_norm": 1.1994301080703735,
      "learning_rate": 3.1205000000000004e-05,
      "loss": 0.4618,
      "step": 6244
    },
    {
      "epoch": 0.3497032142457162,
      "grad_norm": 1.205239176750183,
      "learning_rate": 3.121e-05,
      "loss": 0.4535,
      "step": 6245
    },
    {
      "epoch": 0.34975921155784523,
      "grad_norm": 1.0689464807510376,
      "learning_rate": 3.1215e-05,
      "loss": 0.4064,
      "step": 6246
    },
    {
      "epoch": 0.34981520886997425,
      "grad_norm": 1.5116850137710571,
      "learning_rate": 3.122e-05,
      "loss": 0.5589,
      "step": 6247
    },
    {
      "epoch": 0.34987120618210327,
      "grad_norm": 1.1299906969070435,
      "learning_rate": 3.122500000000001e-05,
      "loss": 0.361,
      "step": 6248
    },
    {
      "epoch": 0.3499272034942323,
      "grad_norm": 1.441096544265747,
      "learning_rate": 3.1230000000000004e-05,
      "loss": 0.488,
      "step": 6249
    },
    {
      "epoch": 0.3499832008063613,
      "grad_norm": 1.3029266595840454,
      "learning_rate": 3.1235e-05,
      "loss": 0.4782,
      "step": 6250
    },
    {
      "epoch": 0.3500391981184903,
      "grad_norm": 1.3709391355514526,
      "learning_rate": 3.1240000000000006e-05,
      "loss": 0.5139,
      "step": 6251
    },
    {
      "epoch": 0.35009519543061934,
      "grad_norm": 1.1097272634506226,
      "learning_rate": 3.1245e-05,
      "loss": 0.4393,
      "step": 6252
    },
    {
      "epoch": 0.35015119274274836,
      "grad_norm": 1.2350056171417236,
      "learning_rate": 3.125e-05,
      "loss": 0.4682,
      "step": 6253
    },
    {
      "epoch": 0.3502071900548774,
      "grad_norm": 1.2020292282104492,
      "learning_rate": 3.1255e-05,
      "loss": 0.4677,
      "step": 6254
    },
    {
      "epoch": 0.3502631873670064,
      "grad_norm": 1.249942421913147,
      "learning_rate": 3.126e-05,
      "loss": 0.3179,
      "step": 6255
    },
    {
      "epoch": 0.3503191846791354,
      "grad_norm": 1.1817785501480103,
      "learning_rate": 3.1265e-05,
      "loss": 0.4654,
      "step": 6256
    },
    {
      "epoch": 0.35037518199126444,
      "grad_norm": 1.414600133895874,
      "learning_rate": 3.127e-05,
      "loss": 0.5809,
      "step": 6257
    },
    {
      "epoch": 0.35043117930339346,
      "grad_norm": 1.3949699401855469,
      "learning_rate": 3.1275e-05,
      "loss": 0.3887,
      "step": 6258
    },
    {
      "epoch": 0.3504871766155225,
      "grad_norm": 1.4005590677261353,
      "learning_rate": 3.1280000000000005e-05,
      "loss": 0.4218,
      "step": 6259
    },
    {
      "epoch": 0.3505431739276515,
      "grad_norm": 1.2349687814712524,
      "learning_rate": 3.1285e-05,
      "loss": 0.4686,
      "step": 6260
    },
    {
      "epoch": 0.3505991712397805,
      "grad_norm": 1.2598515748977661,
      "learning_rate": 3.129e-05,
      "loss": 0.3772,
      "step": 6261
    },
    {
      "epoch": 0.35065516855190954,
      "grad_norm": 1.1200759410858154,
      "learning_rate": 3.1295000000000004e-05,
      "loss": 0.3523,
      "step": 6262
    },
    {
      "epoch": 0.3507111658640385,
      "grad_norm": 1.159464955329895,
      "learning_rate": 3.13e-05,
      "loss": 0.3585,
      "step": 6263
    },
    {
      "epoch": 0.3507671631761675,
      "grad_norm": 1.3332213163375854,
      "learning_rate": 3.1305e-05,
      "loss": 0.4859,
      "step": 6264
    },
    {
      "epoch": 0.35082316048829654,
      "grad_norm": 0.9965168833732605,
      "learning_rate": 3.1310000000000003e-05,
      "loss": 0.3704,
      "step": 6265
    },
    {
      "epoch": 0.35087915780042556,
      "grad_norm": 1.3566290140151978,
      "learning_rate": 3.1315e-05,
      "loss": 0.5692,
      "step": 6266
    },
    {
      "epoch": 0.3509351551125546,
      "grad_norm": 1.304772138595581,
      "learning_rate": 3.132e-05,
      "loss": 0.3989,
      "step": 6267
    },
    {
      "epoch": 0.3509911524246836,
      "grad_norm": 1.2333190441131592,
      "learning_rate": 3.1324999999999996e-05,
      "loss": 0.429,
      "step": 6268
    },
    {
      "epoch": 0.3510471497368126,
      "grad_norm": 1.3603678941726685,
      "learning_rate": 3.133000000000001e-05,
      "loss": 0.3178,
      "step": 6269
    },
    {
      "epoch": 0.35110314704894163,
      "grad_norm": 1.4109737873077393,
      "learning_rate": 3.1335000000000004e-05,
      "loss": 0.6681,
      "step": 6270
    },
    {
      "epoch": 0.35115914436107065,
      "grad_norm": 1.0344195365905762,
      "learning_rate": 3.134e-05,
      "loss": 0.2755,
      "step": 6271
    },
    {
      "epoch": 0.35121514167319967,
      "grad_norm": 1.2682287693023682,
      "learning_rate": 3.1345e-05,
      "loss": 0.5373,
      "step": 6272
    },
    {
      "epoch": 0.3512711389853287,
      "grad_norm": 1.3705300092697144,
      "learning_rate": 3.135e-05,
      "loss": 0.497,
      "step": 6273
    },
    {
      "epoch": 0.3513271362974577,
      "grad_norm": 1.1386823654174805,
      "learning_rate": 3.1355e-05,
      "loss": 0.4038,
      "step": 6274
    },
    {
      "epoch": 0.35138313360958673,
      "grad_norm": 1.5180588960647583,
      "learning_rate": 3.136e-05,
      "loss": 0.4209,
      "step": 6275
    },
    {
      "epoch": 0.35143913092171575,
      "grad_norm": 1.1405197381973267,
      "learning_rate": 3.1365e-05,
      "loss": 0.3937,
      "step": 6276
    },
    {
      "epoch": 0.35149512823384477,
      "grad_norm": 1.2759274244308472,
      "learning_rate": 3.137e-05,
      "loss": 0.4227,
      "step": 6277
    },
    {
      "epoch": 0.3515511255459738,
      "grad_norm": 1.3097624778747559,
      "learning_rate": 3.1375e-05,
      "loss": 0.3797,
      "step": 6278
    },
    {
      "epoch": 0.3516071228581028,
      "grad_norm": 1.594214677810669,
      "learning_rate": 3.138e-05,
      "loss": 0.3762,
      "step": 6279
    },
    {
      "epoch": 0.3516631201702318,
      "grad_norm": 1.5521080493927002,
      "learning_rate": 3.1385000000000005e-05,
      "loss": 0.4929,
      "step": 6280
    },
    {
      "epoch": 0.35171911748236084,
      "grad_norm": 1.3778748512268066,
      "learning_rate": 3.139e-05,
      "loss": 0.4782,
      "step": 6281
    },
    {
      "epoch": 0.35177511479448986,
      "grad_norm": 1.325209617614746,
      "learning_rate": 3.1395e-05,
      "loss": 0.3956,
      "step": 6282
    },
    {
      "epoch": 0.3518311121066189,
      "grad_norm": 1.1175185441970825,
      "learning_rate": 3.1400000000000004e-05,
      "loss": 0.3477,
      "step": 6283
    },
    {
      "epoch": 0.3518871094187479,
      "grad_norm": 1.3731361627578735,
      "learning_rate": 3.1405e-05,
      "loss": 0.3739,
      "step": 6284
    },
    {
      "epoch": 0.3519431067308769,
      "grad_norm": 1.2805825471878052,
      "learning_rate": 3.141e-05,
      "loss": 0.4728,
      "step": 6285
    },
    {
      "epoch": 0.35199910404300594,
      "grad_norm": 1.4022307395935059,
      "learning_rate": 3.1415e-05,
      "loss": 0.5399,
      "step": 6286
    },
    {
      "epoch": 0.35205510135513496,
      "grad_norm": 1.5376461744308472,
      "learning_rate": 3.142e-05,
      "loss": 0.4116,
      "step": 6287
    },
    {
      "epoch": 0.352111098667264,
      "grad_norm": 1.169945240020752,
      "learning_rate": 3.1425e-05,
      "loss": 0.4146,
      "step": 6288
    },
    {
      "epoch": 0.352167095979393,
      "grad_norm": 1.2208759784698486,
      "learning_rate": 3.143e-05,
      "loss": 0.3352,
      "step": 6289
    },
    {
      "epoch": 0.352223093291522,
      "grad_norm": 1.0417909622192383,
      "learning_rate": 3.1435000000000007e-05,
      "loss": 0.3359,
      "step": 6290
    },
    {
      "epoch": 0.35227909060365103,
      "grad_norm": 1.1003497838974,
      "learning_rate": 3.1440000000000004e-05,
      "loss": 0.4339,
      "step": 6291
    },
    {
      "epoch": 0.35233508791578005,
      "grad_norm": 1.071714162826538,
      "learning_rate": 3.1445e-05,
      "loss": 0.4731,
      "step": 6292
    },
    {
      "epoch": 0.3523910852279091,
      "grad_norm": 1.254522681236267,
      "learning_rate": 3.145e-05,
      "loss": 0.4213,
      "step": 6293
    },
    {
      "epoch": 0.3524470825400381,
      "grad_norm": 1.2709230184555054,
      "learning_rate": 3.1455e-05,
      "loss": 0.4648,
      "step": 6294
    },
    {
      "epoch": 0.3525030798521671,
      "grad_norm": 1.3310257196426392,
      "learning_rate": 3.146e-05,
      "loss": 0.4866,
      "step": 6295
    },
    {
      "epoch": 0.35255907716429613,
      "grad_norm": 1.053220272064209,
      "learning_rate": 3.1465e-05,
      "loss": 0.3499,
      "step": 6296
    },
    {
      "epoch": 0.35261507447642515,
      "grad_norm": 1.2098735570907593,
      "learning_rate": 3.147e-05,
      "loss": 0.4933,
      "step": 6297
    },
    {
      "epoch": 0.35267107178855417,
      "grad_norm": 1.2267911434173584,
      "learning_rate": 3.1475e-05,
      "loss": 0.4664,
      "step": 6298
    },
    {
      "epoch": 0.3527270691006832,
      "grad_norm": 1.0496174097061157,
      "learning_rate": 3.1480000000000004e-05,
      "loss": 0.3439,
      "step": 6299
    },
    {
      "epoch": 0.3527830664128122,
      "grad_norm": 1.759591817855835,
      "learning_rate": 3.1485e-05,
      "loss": 0.3957,
      "step": 6300
    },
    {
      "epoch": 0.3528390637249412,
      "grad_norm": 1.3334206342697144,
      "learning_rate": 3.1490000000000005e-05,
      "loss": 0.5113,
      "step": 6301
    },
    {
      "epoch": 0.35289506103707025,
      "grad_norm": 0.9650449156761169,
      "learning_rate": 3.1495e-05,
      "loss": 0.3999,
      "step": 6302
    },
    {
      "epoch": 0.35295105834919926,
      "grad_norm": 1.84669029712677,
      "learning_rate": 3.15e-05,
      "loss": 0.4424,
      "step": 6303
    },
    {
      "epoch": 0.3530070556613283,
      "grad_norm": 1.1730831861495972,
      "learning_rate": 3.1505000000000004e-05,
      "loss": 0.5259,
      "step": 6304
    },
    {
      "epoch": 0.35306305297345725,
      "grad_norm": 1.1850773096084595,
      "learning_rate": 3.151e-05,
      "loss": 0.4617,
      "step": 6305
    },
    {
      "epoch": 0.35311905028558627,
      "grad_norm": 1.1357048749923706,
      "learning_rate": 3.1515e-05,
      "loss": 0.4219,
      "step": 6306
    },
    {
      "epoch": 0.3531750475977153,
      "grad_norm": 1.2091214656829834,
      "learning_rate": 3.1519999999999996e-05,
      "loss": 0.397,
      "step": 6307
    },
    {
      "epoch": 0.3532310449098443,
      "grad_norm": 1.161665678024292,
      "learning_rate": 3.1525e-05,
      "loss": 0.539,
      "step": 6308
    },
    {
      "epoch": 0.3532870422219733,
      "grad_norm": 1.1879373788833618,
      "learning_rate": 3.1530000000000005e-05,
      "loss": 0.4209,
      "step": 6309
    },
    {
      "epoch": 0.35334303953410234,
      "grad_norm": 1.4861319065093994,
      "learning_rate": 3.1535e-05,
      "loss": 0.4973,
      "step": 6310
    },
    {
      "epoch": 0.35339903684623136,
      "grad_norm": 1.3899377584457397,
      "learning_rate": 3.154e-05,
      "loss": 0.4838,
      "step": 6311
    },
    {
      "epoch": 0.3534550341583604,
      "grad_norm": 1.3385703563690186,
      "learning_rate": 3.1545000000000004e-05,
      "loss": 0.5009,
      "step": 6312
    },
    {
      "epoch": 0.3535110314704894,
      "grad_norm": 1.191323161125183,
      "learning_rate": 3.155e-05,
      "loss": 0.4146,
      "step": 6313
    },
    {
      "epoch": 0.3535670287826184,
      "grad_norm": 1.427211880683899,
      "learning_rate": 3.1555e-05,
      "loss": 0.4842,
      "step": 6314
    },
    {
      "epoch": 0.35362302609474744,
      "grad_norm": 1.081580638885498,
      "learning_rate": 3.156e-05,
      "loss": 0.4151,
      "step": 6315
    },
    {
      "epoch": 0.35367902340687646,
      "grad_norm": 1.265127182006836,
      "learning_rate": 3.1565e-05,
      "loss": 0.3041,
      "step": 6316
    },
    {
      "epoch": 0.3537350207190055,
      "grad_norm": 1.2927141189575195,
      "learning_rate": 3.157e-05,
      "loss": 0.5042,
      "step": 6317
    },
    {
      "epoch": 0.3537910180311345,
      "grad_norm": 1.1545687913894653,
      "learning_rate": 3.1575e-05,
      "loss": 0.386,
      "step": 6318
    },
    {
      "epoch": 0.3538470153432635,
      "grad_norm": 1.1653871536254883,
      "learning_rate": 3.1580000000000006e-05,
      "loss": 0.3908,
      "step": 6319
    },
    {
      "epoch": 0.35390301265539253,
      "grad_norm": 1.319900393486023,
      "learning_rate": 3.1585e-05,
      "loss": 0.4542,
      "step": 6320
    },
    {
      "epoch": 0.35395900996752155,
      "grad_norm": 0.9685698747634888,
      "learning_rate": 3.159e-05,
      "loss": 0.3191,
      "step": 6321
    },
    {
      "epoch": 0.3540150072796506,
      "grad_norm": 1.3467005491256714,
      "learning_rate": 3.1595000000000005e-05,
      "loss": 0.4291,
      "step": 6322
    },
    {
      "epoch": 0.3540710045917796,
      "grad_norm": 2.2519562244415283,
      "learning_rate": 3.16e-05,
      "loss": 0.4011,
      "step": 6323
    },
    {
      "epoch": 0.3541270019039086,
      "grad_norm": 1.185945987701416,
      "learning_rate": 3.1605e-05,
      "loss": 0.3492,
      "step": 6324
    },
    {
      "epoch": 0.35418299921603763,
      "grad_norm": 1.2187808752059937,
      "learning_rate": 3.1610000000000004e-05,
      "loss": 0.4182,
      "step": 6325
    },
    {
      "epoch": 0.35423899652816665,
      "grad_norm": 1.1608519554138184,
      "learning_rate": 3.1615e-05,
      "loss": 0.4505,
      "step": 6326
    },
    {
      "epoch": 0.35429499384029567,
      "grad_norm": 1.1894186735153198,
      "learning_rate": 3.162e-05,
      "loss": 0.5263,
      "step": 6327
    },
    {
      "epoch": 0.3543509911524247,
      "grad_norm": 1.2153644561767578,
      "learning_rate": 3.1624999999999996e-05,
      "loss": 0.4624,
      "step": 6328
    },
    {
      "epoch": 0.3544069884645537,
      "grad_norm": 1.2270259857177734,
      "learning_rate": 3.163000000000001e-05,
      "loss": 0.455,
      "step": 6329
    },
    {
      "epoch": 0.3544629857766827,
      "grad_norm": 0.9415087103843689,
      "learning_rate": 3.1635000000000005e-05,
      "loss": 0.2646,
      "step": 6330
    },
    {
      "epoch": 0.35451898308881175,
      "grad_norm": 1.7275992631912231,
      "learning_rate": 3.164e-05,
      "loss": 0.4851,
      "step": 6331
    },
    {
      "epoch": 0.35457498040094076,
      "grad_norm": 1.3599194288253784,
      "learning_rate": 3.1645e-05,
      "loss": 0.43,
      "step": 6332
    },
    {
      "epoch": 0.3546309777130698,
      "grad_norm": 1.7129243612289429,
      "learning_rate": 3.1650000000000004e-05,
      "loss": 0.4948,
      "step": 6333
    },
    {
      "epoch": 0.3546869750251988,
      "grad_norm": 1.1714155673980713,
      "learning_rate": 3.1655e-05,
      "loss": 0.3515,
      "step": 6334
    },
    {
      "epoch": 0.3547429723373278,
      "grad_norm": 1.2481791973114014,
      "learning_rate": 3.166e-05,
      "loss": 0.4746,
      "step": 6335
    },
    {
      "epoch": 0.35479896964945684,
      "grad_norm": 1.1287357807159424,
      "learning_rate": 3.1665e-05,
      "loss": 0.4145,
      "step": 6336
    },
    {
      "epoch": 0.35485496696158586,
      "grad_norm": 1.4542763233184814,
      "learning_rate": 3.167e-05,
      "loss": 0.3924,
      "step": 6337
    },
    {
      "epoch": 0.3549109642737149,
      "grad_norm": 1.4236118793487549,
      "learning_rate": 3.1675e-05,
      "loss": 0.4936,
      "step": 6338
    },
    {
      "epoch": 0.3549669615858439,
      "grad_norm": 1.5000228881835938,
      "learning_rate": 3.168e-05,
      "loss": 0.6088,
      "step": 6339
    },
    {
      "epoch": 0.3550229588979729,
      "grad_norm": 1.1576695442199707,
      "learning_rate": 3.1685000000000006e-05,
      "loss": 0.4053,
      "step": 6340
    },
    {
      "epoch": 0.35507895621010194,
      "grad_norm": 1.2773702144622803,
      "learning_rate": 3.169e-05,
      "loss": 0.4207,
      "step": 6341
    },
    {
      "epoch": 0.35513495352223096,
      "grad_norm": 1.1120816469192505,
      "learning_rate": 3.1695e-05,
      "loss": 0.3528,
      "step": 6342
    },
    {
      "epoch": 0.35519095083436,
      "grad_norm": 1.0784268379211426,
      "learning_rate": 3.1700000000000005e-05,
      "loss": 0.3445,
      "step": 6343
    },
    {
      "epoch": 0.355246948146489,
      "grad_norm": 1.5653364658355713,
      "learning_rate": 3.1705e-05,
      "loss": 0.5955,
      "step": 6344
    },
    {
      "epoch": 0.355302945458618,
      "grad_norm": 1.2151880264282227,
      "learning_rate": 3.171e-05,
      "loss": 0.4883,
      "step": 6345
    },
    {
      "epoch": 0.355358942770747,
      "grad_norm": 1.3910470008850098,
      "learning_rate": 3.1715e-05,
      "loss": 0.542,
      "step": 6346
    },
    {
      "epoch": 0.355414940082876,
      "grad_norm": 1.0868487358093262,
      "learning_rate": 3.172e-05,
      "loss": 0.3592,
      "step": 6347
    },
    {
      "epoch": 0.355470937395005,
      "grad_norm": 1.9859716892242432,
      "learning_rate": 3.1725e-05,
      "loss": 0.3893,
      "step": 6348
    },
    {
      "epoch": 0.35552693470713403,
      "grad_norm": 1.0183755159378052,
      "learning_rate": 3.173e-05,
      "loss": 0.401,
      "step": 6349
    },
    {
      "epoch": 0.35558293201926305,
      "grad_norm": 1.2667468786239624,
      "learning_rate": 3.1735e-05,
      "loss": 0.434,
      "step": 6350
    },
    {
      "epoch": 0.3556389293313921,
      "grad_norm": 1.2418057918548584,
      "learning_rate": 3.1740000000000004e-05,
      "loss": 0.58,
      "step": 6351
    },
    {
      "epoch": 0.3556949266435211,
      "grad_norm": 1.1524723768234253,
      "learning_rate": 3.1745e-05,
      "loss": 0.4516,
      "step": 6352
    },
    {
      "epoch": 0.3557509239556501,
      "grad_norm": 1.9870307445526123,
      "learning_rate": 3.175e-05,
      "loss": 0.457,
      "step": 6353
    },
    {
      "epoch": 0.35580692126777913,
      "grad_norm": 1.1945549249649048,
      "learning_rate": 3.1755000000000003e-05,
      "loss": 0.6001,
      "step": 6354
    },
    {
      "epoch": 0.35586291857990815,
      "grad_norm": 1.2876054048538208,
      "learning_rate": 3.176e-05,
      "loss": 0.444,
      "step": 6355
    },
    {
      "epoch": 0.35591891589203717,
      "grad_norm": 1.264521837234497,
      "learning_rate": 3.1765e-05,
      "loss": 0.3927,
      "step": 6356
    },
    {
      "epoch": 0.3559749132041662,
      "grad_norm": 1.2872791290283203,
      "learning_rate": 3.177e-05,
      "loss": 0.3936,
      "step": 6357
    },
    {
      "epoch": 0.3560309105162952,
      "grad_norm": 1.345687747001648,
      "learning_rate": 3.1775e-05,
      "loss": 0.4547,
      "step": 6358
    },
    {
      "epoch": 0.3560869078284242,
      "grad_norm": 1.096648097038269,
      "learning_rate": 3.1780000000000004e-05,
      "loss": 0.3369,
      "step": 6359
    },
    {
      "epoch": 0.35614290514055325,
      "grad_norm": 1.0514757633209229,
      "learning_rate": 3.1785e-05,
      "loss": 0.3599,
      "step": 6360
    },
    {
      "epoch": 0.35619890245268226,
      "grad_norm": 1.674454927444458,
      "learning_rate": 3.1790000000000006e-05,
      "loss": 0.6327,
      "step": 6361
    },
    {
      "epoch": 0.3562548997648113,
      "grad_norm": 1.044646143913269,
      "learning_rate": 3.1795e-05,
      "loss": 0.3736,
      "step": 6362
    },
    {
      "epoch": 0.3563108970769403,
      "grad_norm": 1.229859709739685,
      "learning_rate": 3.18e-05,
      "loss": 0.3778,
      "step": 6363
    },
    {
      "epoch": 0.3563668943890693,
      "grad_norm": 1.2639544010162354,
      "learning_rate": 3.1805000000000005e-05,
      "loss": 0.496,
      "step": 6364
    },
    {
      "epoch": 0.35642289170119834,
      "grad_norm": 1.1227244138717651,
      "learning_rate": 3.181e-05,
      "loss": 0.3985,
      "step": 6365
    },
    {
      "epoch": 0.35647888901332736,
      "grad_norm": 1.4106838703155518,
      "learning_rate": 3.1815e-05,
      "loss": 0.4,
      "step": 6366
    },
    {
      "epoch": 0.3565348863254564,
      "grad_norm": 1.2340143918991089,
      "learning_rate": 3.182e-05,
      "loss": 0.4061,
      "step": 6367
    },
    {
      "epoch": 0.3565908836375854,
      "grad_norm": 1.538635492324829,
      "learning_rate": 3.1825e-05,
      "loss": 0.5551,
      "step": 6368
    },
    {
      "epoch": 0.3566468809497144,
      "grad_norm": 1.2566502094268799,
      "learning_rate": 3.1830000000000005e-05,
      "loss": 0.5342,
      "step": 6369
    },
    {
      "epoch": 0.35670287826184344,
      "grad_norm": 1.134588599205017,
      "learning_rate": 3.1835e-05,
      "loss": 0.5413,
      "step": 6370
    },
    {
      "epoch": 0.35675887557397246,
      "grad_norm": 1.3127809762954712,
      "learning_rate": 3.184e-05,
      "loss": 0.3591,
      "step": 6371
    },
    {
      "epoch": 0.3568148728861015,
      "grad_norm": 1.045646071434021,
      "learning_rate": 3.1845000000000004e-05,
      "loss": 0.4485,
      "step": 6372
    },
    {
      "epoch": 0.3568708701982305,
      "grad_norm": 1.19120454788208,
      "learning_rate": 3.185e-05,
      "loss": 0.4151,
      "step": 6373
    },
    {
      "epoch": 0.3569268675103595,
      "grad_norm": 1.1111185550689697,
      "learning_rate": 3.1855e-05,
      "loss": 0.5048,
      "step": 6374
    },
    {
      "epoch": 0.35698286482248853,
      "grad_norm": 1.6205518245697021,
      "learning_rate": 3.186e-05,
      "loss": 0.4923,
      "step": 6375
    },
    {
      "epoch": 0.35703886213461755,
      "grad_norm": 1.3513946533203125,
      "learning_rate": 3.1865e-05,
      "loss": 0.4694,
      "step": 6376
    },
    {
      "epoch": 0.35709485944674657,
      "grad_norm": 1.2300530672073364,
      "learning_rate": 3.187e-05,
      "loss": 0.4352,
      "step": 6377
    },
    {
      "epoch": 0.3571508567588756,
      "grad_norm": 1.2493767738342285,
      "learning_rate": 3.1875e-05,
      "loss": 0.5207,
      "step": 6378
    },
    {
      "epoch": 0.3572068540710046,
      "grad_norm": 1.2247203588485718,
      "learning_rate": 3.188e-05,
      "loss": 0.32,
      "step": 6379
    },
    {
      "epoch": 0.35726285138313363,
      "grad_norm": 1.0390961170196533,
      "learning_rate": 3.1885000000000004e-05,
      "loss": 0.3408,
      "step": 6380
    },
    {
      "epoch": 0.35731884869526265,
      "grad_norm": 3.3146679401397705,
      "learning_rate": 3.189e-05,
      "loss": 0.4566,
      "step": 6381
    },
    {
      "epoch": 0.35737484600739167,
      "grad_norm": 1.019629955291748,
      "learning_rate": 3.1895000000000005e-05,
      "loss": 0.3753,
      "step": 6382
    },
    {
      "epoch": 0.3574308433195207,
      "grad_norm": 1.3227936029434204,
      "learning_rate": 3.19e-05,
      "loss": 0.6134,
      "step": 6383
    },
    {
      "epoch": 0.3574868406316497,
      "grad_norm": 1.313201904296875,
      "learning_rate": 3.1905e-05,
      "loss": 0.4308,
      "step": 6384
    },
    {
      "epoch": 0.3575428379437787,
      "grad_norm": 1.2985275983810425,
      "learning_rate": 3.191e-05,
      "loss": 0.4359,
      "step": 6385
    },
    {
      "epoch": 0.35759883525590774,
      "grad_norm": 1.136992335319519,
      "learning_rate": 3.1915e-05,
      "loss": 0.4,
      "step": 6386
    },
    {
      "epoch": 0.3576548325680367,
      "grad_norm": 1.094826102256775,
      "learning_rate": 3.192e-05,
      "loss": 0.3882,
      "step": 6387
    },
    {
      "epoch": 0.3577108298801657,
      "grad_norm": 1.4285268783569336,
      "learning_rate": 3.1925e-05,
      "loss": 0.4504,
      "step": 6388
    },
    {
      "epoch": 0.35776682719229475,
      "grad_norm": 1.1870620250701904,
      "learning_rate": 3.193e-05,
      "loss": 0.5015,
      "step": 6389
    },
    {
      "epoch": 0.35782282450442376,
      "grad_norm": 1.3874123096466064,
      "learning_rate": 3.1935000000000005e-05,
      "loss": 0.6506,
      "step": 6390
    },
    {
      "epoch": 0.3578788218165528,
      "grad_norm": 1.2326061725616455,
      "learning_rate": 3.194e-05,
      "loss": 0.3939,
      "step": 6391
    },
    {
      "epoch": 0.3579348191286818,
      "grad_norm": 1.4296495914459229,
      "learning_rate": 3.1945e-05,
      "loss": 0.6558,
      "step": 6392
    },
    {
      "epoch": 0.3579908164408108,
      "grad_norm": 1.5314677953720093,
      "learning_rate": 3.1950000000000004e-05,
      "loss": 0.3221,
      "step": 6393
    },
    {
      "epoch": 0.35804681375293984,
      "grad_norm": 1.0734069347381592,
      "learning_rate": 3.1955e-05,
      "loss": 0.3728,
      "step": 6394
    },
    {
      "epoch": 0.35810281106506886,
      "grad_norm": 1.264939308166504,
      "learning_rate": 3.196e-05,
      "loss": 0.4224,
      "step": 6395
    },
    {
      "epoch": 0.3581588083771979,
      "grad_norm": 1.4076383113861084,
      "learning_rate": 3.1965e-05,
      "loss": 0.4076,
      "step": 6396
    },
    {
      "epoch": 0.3582148056893269,
      "grad_norm": 1.0512858629226685,
      "learning_rate": 3.197e-05,
      "loss": 0.4198,
      "step": 6397
    },
    {
      "epoch": 0.3582708030014559,
      "grad_norm": 1.4213348627090454,
      "learning_rate": 3.1975e-05,
      "loss": 0.3753,
      "step": 6398
    },
    {
      "epoch": 0.35832680031358494,
      "grad_norm": 1.445689082145691,
      "learning_rate": 3.198e-05,
      "loss": 0.4111,
      "step": 6399
    },
    {
      "epoch": 0.35838279762571396,
      "grad_norm": 1.4471769332885742,
      "learning_rate": 3.1985000000000006e-05,
      "loss": 0.3986,
      "step": 6400
    },
    {
      "epoch": 0.358438794937843,
      "grad_norm": 1.280672311782837,
      "learning_rate": 3.1990000000000004e-05,
      "loss": 0.4805,
      "step": 6401
    },
    {
      "epoch": 0.358494792249972,
      "grad_norm": 1.2855925559997559,
      "learning_rate": 3.1995e-05,
      "loss": 0.4193,
      "step": 6402
    },
    {
      "epoch": 0.358550789562101,
      "grad_norm": 1.4518502950668335,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.4552,
      "step": 6403
    },
    {
      "epoch": 0.35860678687423003,
      "grad_norm": 1.4236962795257568,
      "learning_rate": 3.2005e-05,
      "loss": 0.5373,
      "step": 6404
    },
    {
      "epoch": 0.35866278418635905,
      "grad_norm": 1.2707767486572266,
      "learning_rate": 3.201e-05,
      "loss": 0.3918,
      "step": 6405
    },
    {
      "epoch": 0.35871878149848807,
      "grad_norm": 1.2630311250686646,
      "learning_rate": 3.2015e-05,
      "loss": 0.5197,
      "step": 6406
    },
    {
      "epoch": 0.3587747788106171,
      "grad_norm": 1.2726080417633057,
      "learning_rate": 3.202e-05,
      "loss": 0.3782,
      "step": 6407
    },
    {
      "epoch": 0.3588307761227461,
      "grad_norm": 1.275527000427246,
      "learning_rate": 3.2025e-05,
      "loss": 0.4598,
      "step": 6408
    },
    {
      "epoch": 0.35888677343487513,
      "grad_norm": 1.2320438623428345,
      "learning_rate": 3.2029999999999997e-05,
      "loss": 0.4221,
      "step": 6409
    },
    {
      "epoch": 0.35894277074700415,
      "grad_norm": 1.4411370754241943,
      "learning_rate": 3.2035e-05,
      "loss": 0.5037,
      "step": 6410
    },
    {
      "epoch": 0.35899876805913317,
      "grad_norm": 1.2561625242233276,
      "learning_rate": 3.2040000000000005e-05,
      "loss": 0.3273,
      "step": 6411
    },
    {
      "epoch": 0.3590547653712622,
      "grad_norm": 1.2090235948562622,
      "learning_rate": 3.2045e-05,
      "loss": 0.4458,
      "step": 6412
    },
    {
      "epoch": 0.3591107626833912,
      "grad_norm": 1.248092532157898,
      "learning_rate": 3.205e-05,
      "loss": 0.4784,
      "step": 6413
    },
    {
      "epoch": 0.3591667599955202,
      "grad_norm": 1.1258567571640015,
      "learning_rate": 3.2055000000000004e-05,
      "loss": 0.4477,
      "step": 6414
    },
    {
      "epoch": 0.35922275730764924,
      "grad_norm": 1.2157437801361084,
      "learning_rate": 3.206e-05,
      "loss": 0.4207,
      "step": 6415
    },
    {
      "epoch": 0.35927875461977826,
      "grad_norm": 1.399810791015625,
      "learning_rate": 3.2065e-05,
      "loss": 0.3596,
      "step": 6416
    },
    {
      "epoch": 0.3593347519319073,
      "grad_norm": 1.2439607381820679,
      "learning_rate": 3.207e-05,
      "loss": 0.467,
      "step": 6417
    },
    {
      "epoch": 0.3593907492440363,
      "grad_norm": 1.3221555948257446,
      "learning_rate": 3.2075e-05,
      "loss": 0.4603,
      "step": 6418
    },
    {
      "epoch": 0.3594467465561653,
      "grad_norm": 1.1428906917572021,
      "learning_rate": 3.208e-05,
      "loss": 0.3442,
      "step": 6419
    },
    {
      "epoch": 0.35950274386829434,
      "grad_norm": 1.3556814193725586,
      "learning_rate": 3.2085e-05,
      "loss": 0.4593,
      "step": 6420
    },
    {
      "epoch": 0.35955874118042336,
      "grad_norm": 1.1192015409469604,
      "learning_rate": 3.2090000000000006e-05,
      "loss": 0.3109,
      "step": 6421
    },
    {
      "epoch": 0.3596147384925524,
      "grad_norm": 1.3313668966293335,
      "learning_rate": 3.2095000000000004e-05,
      "loss": 0.5065,
      "step": 6422
    },
    {
      "epoch": 0.3596707358046814,
      "grad_norm": 1.2944973707199097,
      "learning_rate": 3.21e-05,
      "loss": 0.4034,
      "step": 6423
    },
    {
      "epoch": 0.3597267331168104,
      "grad_norm": 1.1574398279190063,
      "learning_rate": 3.2105e-05,
      "loss": 0.3418,
      "step": 6424
    },
    {
      "epoch": 0.35978273042893943,
      "grad_norm": 1.1419357061386108,
      "learning_rate": 3.211e-05,
      "loss": 0.349,
      "step": 6425
    },
    {
      "epoch": 0.35983872774106845,
      "grad_norm": 1.5139812231063843,
      "learning_rate": 3.2115e-05,
      "loss": 0.3879,
      "step": 6426
    },
    {
      "epoch": 0.3598947250531975,
      "grad_norm": 1.2404354810714722,
      "learning_rate": 3.212e-05,
      "loss": 0.4619,
      "step": 6427
    },
    {
      "epoch": 0.3599507223653265,
      "grad_norm": 1.1889538764953613,
      "learning_rate": 3.2125e-05,
      "loss": 0.4278,
      "step": 6428
    },
    {
      "epoch": 0.36000671967745546,
      "grad_norm": 1.2059705257415771,
      "learning_rate": 3.213e-05,
      "loss": 0.4227,
      "step": 6429
    },
    {
      "epoch": 0.3600627169895845,
      "grad_norm": 1.1609004735946655,
      "learning_rate": 3.2135e-05,
      "loss": 0.404,
      "step": 6430
    },
    {
      "epoch": 0.3601187143017135,
      "grad_norm": 1.007348656654358,
      "learning_rate": 3.214e-05,
      "loss": 0.4094,
      "step": 6431
    },
    {
      "epoch": 0.3601747116138425,
      "grad_norm": 1.18724524974823,
      "learning_rate": 3.2145000000000005e-05,
      "loss": 0.4617,
      "step": 6432
    },
    {
      "epoch": 0.36023070892597153,
      "grad_norm": 1.2899388074874878,
      "learning_rate": 3.215e-05,
      "loss": 0.5475,
      "step": 6433
    },
    {
      "epoch": 0.36028670623810055,
      "grad_norm": 1.316030740737915,
      "learning_rate": 3.2155e-05,
      "loss": 0.5201,
      "step": 6434
    },
    {
      "epoch": 0.36034270355022957,
      "grad_norm": 1.1159597635269165,
      "learning_rate": 3.2160000000000004e-05,
      "loss": 0.4133,
      "step": 6435
    },
    {
      "epoch": 0.3603987008623586,
      "grad_norm": 1.3339399099349976,
      "learning_rate": 3.2165e-05,
      "loss": 0.4596,
      "step": 6436
    },
    {
      "epoch": 0.3604546981744876,
      "grad_norm": 1.3317457437515259,
      "learning_rate": 3.217e-05,
      "loss": 0.4083,
      "step": 6437
    },
    {
      "epoch": 0.36051069548661663,
      "grad_norm": 1.309380054473877,
      "learning_rate": 3.2175e-05,
      "loss": 0.4438,
      "step": 6438
    },
    {
      "epoch": 0.36056669279874565,
      "grad_norm": 1.1014667749404907,
      "learning_rate": 3.218e-05,
      "loss": 0.5102,
      "step": 6439
    },
    {
      "epoch": 0.36062269011087467,
      "grad_norm": 1.2829416990280151,
      "learning_rate": 3.2185000000000004e-05,
      "loss": 0.4968,
      "step": 6440
    },
    {
      "epoch": 0.3606786874230037,
      "grad_norm": 1.0472525358200073,
      "learning_rate": 3.219e-05,
      "loss": 0.3369,
      "step": 6441
    },
    {
      "epoch": 0.3607346847351327,
      "grad_norm": 1.2059707641601562,
      "learning_rate": 3.2195000000000006e-05,
      "loss": 0.4293,
      "step": 6442
    },
    {
      "epoch": 0.3607906820472617,
      "grad_norm": 1.0743045806884766,
      "learning_rate": 3.2200000000000003e-05,
      "loss": 0.2988,
      "step": 6443
    },
    {
      "epoch": 0.36084667935939074,
      "grad_norm": 1.1963176727294922,
      "learning_rate": 3.2205e-05,
      "loss": 0.4514,
      "step": 6444
    },
    {
      "epoch": 0.36090267667151976,
      "grad_norm": 1.4290324449539185,
      "learning_rate": 3.221e-05,
      "loss": 0.4848,
      "step": 6445
    },
    {
      "epoch": 0.3609586739836488,
      "grad_norm": 1.4431935548782349,
      "learning_rate": 3.2215e-05,
      "loss": 0.4874,
      "step": 6446
    },
    {
      "epoch": 0.3610146712957778,
      "grad_norm": 1.5518471002578735,
      "learning_rate": 3.222e-05,
      "loss": 0.6615,
      "step": 6447
    },
    {
      "epoch": 0.3610706686079068,
      "grad_norm": 1.2346594333648682,
      "learning_rate": 3.2225e-05,
      "loss": 0.4037,
      "step": 6448
    },
    {
      "epoch": 0.36112666592003584,
      "grad_norm": 1.3797889947891235,
      "learning_rate": 3.223e-05,
      "loss": 0.3868,
      "step": 6449
    },
    {
      "epoch": 0.36118266323216486,
      "grad_norm": 1.1214135885238647,
      "learning_rate": 3.2235000000000006e-05,
      "loss": 0.3973,
      "step": 6450
    },
    {
      "epoch": 0.3612386605442939,
      "grad_norm": 1.1265441179275513,
      "learning_rate": 3.224e-05,
      "loss": 0.3838,
      "step": 6451
    },
    {
      "epoch": 0.3612946578564229,
      "grad_norm": 1.2349735498428345,
      "learning_rate": 3.2245e-05,
      "loss": 0.388,
      "step": 6452
    },
    {
      "epoch": 0.3613506551685519,
      "grad_norm": 1.1909857988357544,
      "learning_rate": 3.2250000000000005e-05,
      "loss": 0.4089,
      "step": 6453
    },
    {
      "epoch": 0.36140665248068093,
      "grad_norm": 1.2351858615875244,
      "learning_rate": 3.2255e-05,
      "loss": 0.4081,
      "step": 6454
    },
    {
      "epoch": 0.36146264979280995,
      "grad_norm": 1.1014418601989746,
      "learning_rate": 3.226e-05,
      "loss": 0.3513,
      "step": 6455
    },
    {
      "epoch": 0.361518647104939,
      "grad_norm": 1.296622395515442,
      "learning_rate": 3.2265000000000004e-05,
      "loss": 0.4217,
      "step": 6456
    },
    {
      "epoch": 0.361574644417068,
      "grad_norm": 1.2340731620788574,
      "learning_rate": 3.227e-05,
      "loss": 0.4694,
      "step": 6457
    },
    {
      "epoch": 0.361630641729197,
      "grad_norm": 1.2117183208465576,
      "learning_rate": 3.2275e-05,
      "loss": 0.3697,
      "step": 6458
    },
    {
      "epoch": 0.36168663904132603,
      "grad_norm": 1.2185648679733276,
      "learning_rate": 3.2279999999999996e-05,
      "loss": 0.4544,
      "step": 6459
    },
    {
      "epoch": 0.36174263635345505,
      "grad_norm": 1.2371495962142944,
      "learning_rate": 3.228500000000001e-05,
      "loss": 0.4588,
      "step": 6460
    },
    {
      "epoch": 0.36179863366558407,
      "grad_norm": 1.1846920251846313,
      "learning_rate": 3.2290000000000004e-05,
      "loss": 0.4448,
      "step": 6461
    },
    {
      "epoch": 0.3618546309777131,
      "grad_norm": 1.308303713798523,
      "learning_rate": 3.2295e-05,
      "loss": 0.4576,
      "step": 6462
    },
    {
      "epoch": 0.3619106282898421,
      "grad_norm": 1.1310920715332031,
      "learning_rate": 3.2300000000000006e-05,
      "loss": 0.4626,
      "step": 6463
    },
    {
      "epoch": 0.3619666256019711,
      "grad_norm": 1.205828070640564,
      "learning_rate": 3.2305e-05,
      "loss": 0.4259,
      "step": 6464
    },
    {
      "epoch": 0.36202262291410015,
      "grad_norm": 1.215994954109192,
      "learning_rate": 3.231e-05,
      "loss": 0.3851,
      "step": 6465
    },
    {
      "epoch": 0.36207862022622916,
      "grad_norm": 1.2289583683013916,
      "learning_rate": 3.2315e-05,
      "loss": 0.4878,
      "step": 6466
    },
    {
      "epoch": 0.3621346175383582,
      "grad_norm": 1.1727473735809326,
      "learning_rate": 3.232e-05,
      "loss": 0.3297,
      "step": 6467
    },
    {
      "epoch": 0.3621906148504872,
      "grad_norm": 1.1282618045806885,
      "learning_rate": 3.2325e-05,
      "loss": 0.402,
      "step": 6468
    },
    {
      "epoch": 0.3622466121626162,
      "grad_norm": 1.4074574708938599,
      "learning_rate": 3.233e-05,
      "loss": 0.4783,
      "step": 6469
    },
    {
      "epoch": 0.3623026094747452,
      "grad_norm": 2.0473852157592773,
      "learning_rate": 3.2335e-05,
      "loss": 0.5243,
      "step": 6470
    },
    {
      "epoch": 0.3623586067868742,
      "grad_norm": 1.3587144613265991,
      "learning_rate": 3.2340000000000005e-05,
      "loss": 0.385,
      "step": 6471
    },
    {
      "epoch": 0.3624146040990032,
      "grad_norm": 1.1086276769638062,
      "learning_rate": 3.2345e-05,
      "loss": 0.3856,
      "step": 6472
    },
    {
      "epoch": 0.36247060141113224,
      "grad_norm": 1.2624282836914062,
      "learning_rate": 3.235e-05,
      "loss": 0.5057,
      "step": 6473
    },
    {
      "epoch": 0.36252659872326126,
      "grad_norm": 1.1324260234832764,
      "learning_rate": 3.2355000000000004e-05,
      "loss": 0.4098,
      "step": 6474
    },
    {
      "epoch": 0.3625825960353903,
      "grad_norm": 1.260992407798767,
      "learning_rate": 3.236e-05,
      "loss": 0.3632,
      "step": 6475
    },
    {
      "epoch": 0.3626385933475193,
      "grad_norm": 1.2799453735351562,
      "learning_rate": 3.2365e-05,
      "loss": 0.3703,
      "step": 6476
    },
    {
      "epoch": 0.3626945906596483,
      "grad_norm": 1.233601689338684,
      "learning_rate": 3.2370000000000003e-05,
      "loss": 0.5573,
      "step": 6477
    },
    {
      "epoch": 0.36275058797177734,
      "grad_norm": 1.1402902603149414,
      "learning_rate": 3.2375e-05,
      "loss": 0.3808,
      "step": 6478
    },
    {
      "epoch": 0.36280658528390636,
      "grad_norm": 1.103020429611206,
      "learning_rate": 3.238e-05,
      "loss": 0.386,
      "step": 6479
    },
    {
      "epoch": 0.3628625825960354,
      "grad_norm": 1.1356475353240967,
      "learning_rate": 3.2385e-05,
      "loss": 0.406,
      "step": 6480
    },
    {
      "epoch": 0.3629185799081644,
      "grad_norm": 1.1693097352981567,
      "learning_rate": 3.239000000000001e-05,
      "loss": 0.4771,
      "step": 6481
    },
    {
      "epoch": 0.3629745772202934,
      "grad_norm": 1.4075266122817993,
      "learning_rate": 3.2395000000000004e-05,
      "loss": 0.5045,
      "step": 6482
    },
    {
      "epoch": 0.36303057453242243,
      "grad_norm": 1.3106882572174072,
      "learning_rate": 3.24e-05,
      "loss": 0.4642,
      "step": 6483
    },
    {
      "epoch": 0.36308657184455145,
      "grad_norm": 1.2524840831756592,
      "learning_rate": 3.2405e-05,
      "loss": 0.4903,
      "step": 6484
    },
    {
      "epoch": 0.3631425691566805,
      "grad_norm": 1.4604068994522095,
      "learning_rate": 3.241e-05,
      "loss": 0.4731,
      "step": 6485
    },
    {
      "epoch": 0.3631985664688095,
      "grad_norm": 1.555456280708313,
      "learning_rate": 3.2415e-05,
      "loss": 0.4913,
      "step": 6486
    },
    {
      "epoch": 0.3632545637809385,
      "grad_norm": 1.2731359004974365,
      "learning_rate": 3.242e-05,
      "loss": 0.3683,
      "step": 6487
    },
    {
      "epoch": 0.36331056109306753,
      "grad_norm": 1.37197744846344,
      "learning_rate": 3.2425e-05,
      "loss": 0.418,
      "step": 6488
    },
    {
      "epoch": 0.36336655840519655,
      "grad_norm": 1.0760390758514404,
      "learning_rate": 3.243e-05,
      "loss": 0.364,
      "step": 6489
    },
    {
      "epoch": 0.36342255571732557,
      "grad_norm": 1.191428542137146,
      "learning_rate": 3.2435000000000004e-05,
      "loss": 0.4739,
      "step": 6490
    },
    {
      "epoch": 0.3634785530294546,
      "grad_norm": 2.9587643146514893,
      "learning_rate": 3.244e-05,
      "loss": 0.4657,
      "step": 6491
    },
    {
      "epoch": 0.3635345503415836,
      "grad_norm": 1.2364811897277832,
      "learning_rate": 3.2445000000000005e-05,
      "loss": 0.53,
      "step": 6492
    },
    {
      "epoch": 0.3635905476537126,
      "grad_norm": 1.3529722690582275,
      "learning_rate": 3.245e-05,
      "loss": 0.4389,
      "step": 6493
    },
    {
      "epoch": 0.36364654496584164,
      "grad_norm": 1.2198383808135986,
      "learning_rate": 3.2455e-05,
      "loss": 0.4862,
      "step": 6494
    },
    {
      "epoch": 0.36370254227797066,
      "grad_norm": 1.2174612283706665,
      "learning_rate": 3.2460000000000004e-05,
      "loss": 0.4445,
      "step": 6495
    },
    {
      "epoch": 0.3637585395900997,
      "grad_norm": 0.9913917183876038,
      "learning_rate": 3.2465e-05,
      "loss": 0.4996,
      "step": 6496
    },
    {
      "epoch": 0.3638145369022287,
      "grad_norm": 1.5098800659179688,
      "learning_rate": 3.247e-05,
      "loss": 0.393,
      "step": 6497
    },
    {
      "epoch": 0.3638705342143577,
      "grad_norm": 1.3465111255645752,
      "learning_rate": 3.2474999999999997e-05,
      "loss": 0.4528,
      "step": 6498
    },
    {
      "epoch": 0.36392653152648674,
      "grad_norm": 1.342793345451355,
      "learning_rate": 3.248e-05,
      "loss": 0.5098,
      "step": 6499
    },
    {
      "epoch": 0.36398252883861576,
      "grad_norm": 1.5947413444519043,
      "learning_rate": 3.2485000000000005e-05,
      "loss": 0.4775,
      "step": 6500
    },
    {
      "epoch": 0.3640385261507448,
      "grad_norm": 1.0939754247665405,
      "learning_rate": 3.249e-05,
      "loss": 0.3968,
      "step": 6501
    },
    {
      "epoch": 0.3640945234628738,
      "grad_norm": 1.2496519088745117,
      "learning_rate": 3.2495000000000007e-05,
      "loss": 0.4816,
      "step": 6502
    },
    {
      "epoch": 0.3641505207750028,
      "grad_norm": 1.2200695276260376,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.449,
      "step": 6503
    },
    {
      "epoch": 0.36420651808713184,
      "grad_norm": 1.184432029724121,
      "learning_rate": 3.2505e-05,
      "loss": 0.4768,
      "step": 6504
    },
    {
      "epoch": 0.36426251539926086,
      "grad_norm": 1.0775400400161743,
      "learning_rate": 3.251e-05,
      "loss": 0.3957,
      "step": 6505
    },
    {
      "epoch": 0.3643185127113899,
      "grad_norm": 1.1742106676101685,
      "learning_rate": 3.2515e-05,
      "loss": 0.4073,
      "step": 6506
    },
    {
      "epoch": 0.3643745100235189,
      "grad_norm": 1.1494220495224,
      "learning_rate": 3.252e-05,
      "loss": 0.4473,
      "step": 6507
    },
    {
      "epoch": 0.3644305073356479,
      "grad_norm": 1.1055777072906494,
      "learning_rate": 3.2525e-05,
      "loss": 0.3384,
      "step": 6508
    },
    {
      "epoch": 0.36448650464777693,
      "grad_norm": 1.275314211845398,
      "learning_rate": 3.253e-05,
      "loss": 0.4473,
      "step": 6509
    },
    {
      "epoch": 0.36454250195990595,
      "grad_norm": 1.1592401266098022,
      "learning_rate": 3.2535e-05,
      "loss": 0.4372,
      "step": 6510
    },
    {
      "epoch": 0.3645984992720349,
      "grad_norm": 1.2728155851364136,
      "learning_rate": 3.2540000000000004e-05,
      "loss": 0.4949,
      "step": 6511
    },
    {
      "epoch": 0.36465449658416393,
      "grad_norm": 1.1379468441009521,
      "learning_rate": 3.2545e-05,
      "loss": 0.4844,
      "step": 6512
    },
    {
      "epoch": 0.36471049389629295,
      "grad_norm": 1.2028402090072632,
      "learning_rate": 3.2550000000000005e-05,
      "loss": 0.4344,
      "step": 6513
    },
    {
      "epoch": 0.364766491208422,
      "grad_norm": 1.402000069618225,
      "learning_rate": 3.2555e-05,
      "loss": 0.3926,
      "step": 6514
    },
    {
      "epoch": 0.364822488520551,
      "grad_norm": 1.4037359952926636,
      "learning_rate": 3.256e-05,
      "loss": 0.2992,
      "step": 6515
    },
    {
      "epoch": 0.36487848583268,
      "grad_norm": 1.2170379161834717,
      "learning_rate": 3.2565000000000004e-05,
      "loss": 0.5853,
      "step": 6516
    },
    {
      "epoch": 0.36493448314480903,
      "grad_norm": 1.0671933889389038,
      "learning_rate": 3.257e-05,
      "loss": 0.4087,
      "step": 6517
    },
    {
      "epoch": 0.36499048045693805,
      "grad_norm": 1.2431203126907349,
      "learning_rate": 3.2575e-05,
      "loss": 0.4131,
      "step": 6518
    },
    {
      "epoch": 0.36504647776906707,
      "grad_norm": 1.300535798072815,
      "learning_rate": 3.2579999999999996e-05,
      "loss": 0.3265,
      "step": 6519
    },
    {
      "epoch": 0.3651024750811961,
      "grad_norm": 1.442633867263794,
      "learning_rate": 3.2585e-05,
      "loss": 0.4488,
      "step": 6520
    },
    {
      "epoch": 0.3651584723933251,
      "grad_norm": 1.062958002090454,
      "learning_rate": 3.2590000000000005e-05,
      "loss": 0.2811,
      "step": 6521
    },
    {
      "epoch": 0.3652144697054541,
      "grad_norm": 1.154906153678894,
      "learning_rate": 3.2595e-05,
      "loss": 0.3751,
      "step": 6522
    },
    {
      "epoch": 0.36527046701758314,
      "grad_norm": 1.2711113691329956,
      "learning_rate": 3.26e-05,
      "loss": 0.5827,
      "step": 6523
    },
    {
      "epoch": 0.36532646432971216,
      "grad_norm": 1.325225591659546,
      "learning_rate": 3.2605000000000004e-05,
      "loss": 0.319,
      "step": 6524
    },
    {
      "epoch": 0.3653824616418412,
      "grad_norm": 1.3151487112045288,
      "learning_rate": 3.261e-05,
      "loss": 0.4581,
      "step": 6525
    },
    {
      "epoch": 0.3654384589539702,
      "grad_norm": 1.2263975143432617,
      "learning_rate": 3.2615e-05,
      "loss": 0.4808,
      "step": 6526
    },
    {
      "epoch": 0.3654944562660992,
      "grad_norm": 1.1650111675262451,
      "learning_rate": 3.262e-05,
      "loss": 0.3016,
      "step": 6527
    },
    {
      "epoch": 0.36555045357822824,
      "grad_norm": 1.2386382818222046,
      "learning_rate": 3.2625e-05,
      "loss": 0.4782,
      "step": 6528
    },
    {
      "epoch": 0.36560645089035726,
      "grad_norm": 1.1024527549743652,
      "learning_rate": 3.263e-05,
      "loss": 0.3295,
      "step": 6529
    },
    {
      "epoch": 0.3656624482024863,
      "grad_norm": 1.4803990125656128,
      "learning_rate": 3.2635e-05,
      "loss": 0.5028,
      "step": 6530
    },
    {
      "epoch": 0.3657184455146153,
      "grad_norm": 1.300776720046997,
      "learning_rate": 3.2640000000000006e-05,
      "loss": 0.5235,
      "step": 6531
    },
    {
      "epoch": 0.3657744428267443,
      "grad_norm": 1.1828978061676025,
      "learning_rate": 3.2645e-05,
      "loss": 0.3642,
      "step": 6532
    },
    {
      "epoch": 0.36583044013887334,
      "grad_norm": 1.1352730989456177,
      "learning_rate": 3.265e-05,
      "loss": 0.4256,
      "step": 6533
    },
    {
      "epoch": 0.36588643745100236,
      "grad_norm": 1.3790315389633179,
      "learning_rate": 3.2655000000000005e-05,
      "loss": 0.4952,
      "step": 6534
    },
    {
      "epoch": 0.3659424347631314,
      "grad_norm": 1.1742215156555176,
      "learning_rate": 3.266e-05,
      "loss": 0.4102,
      "step": 6535
    },
    {
      "epoch": 0.3659984320752604,
      "grad_norm": 1.2382011413574219,
      "learning_rate": 3.2665e-05,
      "loss": 0.4639,
      "step": 6536
    },
    {
      "epoch": 0.3660544293873894,
      "grad_norm": 1.2191896438598633,
      "learning_rate": 3.267e-05,
      "loss": 0.3776,
      "step": 6537
    },
    {
      "epoch": 0.36611042669951843,
      "grad_norm": 1.1040349006652832,
      "learning_rate": 3.2675e-05,
      "loss": 0.4356,
      "step": 6538
    },
    {
      "epoch": 0.36616642401164745,
      "grad_norm": 1.3455814123153687,
      "learning_rate": 3.268e-05,
      "loss": 0.4472,
      "step": 6539
    },
    {
      "epoch": 0.36622242132377647,
      "grad_norm": 1.3614933490753174,
      "learning_rate": 3.2684999999999996e-05,
      "loss": 0.6831,
      "step": 6540
    },
    {
      "epoch": 0.3662784186359055,
      "grad_norm": 1.1957603693008423,
      "learning_rate": 3.269000000000001e-05,
      "loss": 0.5281,
      "step": 6541
    },
    {
      "epoch": 0.3663344159480345,
      "grad_norm": 1.596968173980713,
      "learning_rate": 3.2695000000000005e-05,
      "loss": 0.4646,
      "step": 6542
    },
    {
      "epoch": 0.3663904132601635,
      "grad_norm": 1.2392882108688354,
      "learning_rate": 3.27e-05,
      "loss": 0.5475,
      "step": 6543
    },
    {
      "epoch": 0.36644641057229255,
      "grad_norm": 1.2253211736679077,
      "learning_rate": 3.2705e-05,
      "loss": 0.3998,
      "step": 6544
    },
    {
      "epoch": 0.36650240788442157,
      "grad_norm": 1.2598199844360352,
      "learning_rate": 3.2710000000000004e-05,
      "loss": 0.3963,
      "step": 6545
    },
    {
      "epoch": 0.3665584051965506,
      "grad_norm": 1.208227276802063,
      "learning_rate": 3.2715e-05,
      "loss": 0.5363,
      "step": 6546
    },
    {
      "epoch": 0.3666144025086796,
      "grad_norm": 1.0527280569076538,
      "learning_rate": 3.272e-05,
      "loss": 0.3906,
      "step": 6547
    },
    {
      "epoch": 0.3666703998208086,
      "grad_norm": 1.2148113250732422,
      "learning_rate": 3.2725e-05,
      "loss": 0.4611,
      "step": 6548
    },
    {
      "epoch": 0.36672639713293764,
      "grad_norm": 1.420346736907959,
      "learning_rate": 3.273e-05,
      "loss": 0.5132,
      "step": 6549
    },
    {
      "epoch": 0.36678239444506666,
      "grad_norm": 1.3327460289001465,
      "learning_rate": 3.2735e-05,
      "loss": 0.3674,
      "step": 6550
    },
    {
      "epoch": 0.3668383917571957,
      "grad_norm": 1.1744295358657837,
      "learning_rate": 3.274e-05,
      "loss": 0.3614,
      "step": 6551
    },
    {
      "epoch": 0.3668943890693247,
      "grad_norm": 1.3436514139175415,
      "learning_rate": 3.2745000000000006e-05,
      "loss": 0.3978,
      "step": 6552
    },
    {
      "epoch": 0.36695038638145366,
      "grad_norm": 1.2072014808654785,
      "learning_rate": 3.275e-05,
      "loss": 0.3727,
      "step": 6553
    },
    {
      "epoch": 0.3670063836935827,
      "grad_norm": 1.1398029327392578,
      "learning_rate": 3.2755e-05,
      "loss": 0.4489,
      "step": 6554
    },
    {
      "epoch": 0.3670623810057117,
      "grad_norm": 1.4085496664047241,
      "learning_rate": 3.2760000000000005e-05,
      "loss": 0.694,
      "step": 6555
    },
    {
      "epoch": 0.3671183783178407,
      "grad_norm": 1.164756417274475,
      "learning_rate": 3.2765e-05,
      "loss": 0.4144,
      "step": 6556
    },
    {
      "epoch": 0.36717437562996974,
      "grad_norm": 1.3206162452697754,
      "learning_rate": 3.277e-05,
      "loss": 0.4299,
      "step": 6557
    },
    {
      "epoch": 0.36723037294209876,
      "grad_norm": 1.059890866279602,
      "learning_rate": 3.2775e-05,
      "loss": 0.3972,
      "step": 6558
    },
    {
      "epoch": 0.3672863702542278,
      "grad_norm": 1.2624162435531616,
      "learning_rate": 3.278e-05,
      "loss": 0.4341,
      "step": 6559
    },
    {
      "epoch": 0.3673423675663568,
      "grad_norm": 1.3863948583602905,
      "learning_rate": 3.2785e-05,
      "loss": 0.5112,
      "step": 6560
    },
    {
      "epoch": 0.3673983648784858,
      "grad_norm": 1.219224214553833,
      "learning_rate": 3.279e-05,
      "loss": 0.4561,
      "step": 6561
    },
    {
      "epoch": 0.36745436219061484,
      "grad_norm": 1.2431224584579468,
      "learning_rate": 3.2795e-05,
      "loss": 0.4304,
      "step": 6562
    },
    {
      "epoch": 0.36751035950274386,
      "grad_norm": 1.2224704027175903,
      "learning_rate": 3.2800000000000004e-05,
      "loss": 0.3912,
      "step": 6563
    },
    {
      "epoch": 0.3675663568148729,
      "grad_norm": 2.1497962474823,
      "learning_rate": 3.2805e-05,
      "loss": 0.668,
      "step": 6564
    },
    {
      "epoch": 0.3676223541270019,
      "grad_norm": 1.4057832956314087,
      "learning_rate": 3.281e-05,
      "loss": 0.7726,
      "step": 6565
    },
    {
      "epoch": 0.3676783514391309,
      "grad_norm": 1.4491353034973145,
      "learning_rate": 3.2815000000000003e-05,
      "loss": 0.4135,
      "step": 6566
    },
    {
      "epoch": 0.36773434875125993,
      "grad_norm": 1.0267125368118286,
      "learning_rate": 3.282e-05,
      "loss": 0.3927,
      "step": 6567
    },
    {
      "epoch": 0.36779034606338895,
      "grad_norm": 1.2104021310806274,
      "learning_rate": 3.2825e-05,
      "loss": 0.3914,
      "step": 6568
    },
    {
      "epoch": 0.36784634337551797,
      "grad_norm": 1.4252570867538452,
      "learning_rate": 3.283e-05,
      "loss": 0.4011,
      "step": 6569
    },
    {
      "epoch": 0.367902340687647,
      "grad_norm": 1.1146646738052368,
      "learning_rate": 3.2835e-05,
      "loss": 0.5136,
      "step": 6570
    },
    {
      "epoch": 0.367958337999776,
      "grad_norm": 1.1819242238998413,
      "learning_rate": 3.2840000000000004e-05,
      "loss": 0.3871,
      "step": 6571
    },
    {
      "epoch": 0.368014335311905,
      "grad_norm": 1.1665692329406738,
      "learning_rate": 3.2845e-05,
      "loss": 0.3557,
      "step": 6572
    },
    {
      "epoch": 0.36807033262403405,
      "grad_norm": 1.575913906097412,
      "learning_rate": 3.2850000000000006e-05,
      "loss": 0.4912,
      "step": 6573
    },
    {
      "epoch": 0.36812632993616307,
      "grad_norm": 1.8078118562698364,
      "learning_rate": 3.2855e-05,
      "loss": 0.3965,
      "step": 6574
    },
    {
      "epoch": 0.3681823272482921,
      "grad_norm": 1.1450138092041016,
      "learning_rate": 3.286e-05,
      "loss": 0.319,
      "step": 6575
    },
    {
      "epoch": 0.3682383245604211,
      "grad_norm": 1.0713926553726196,
      "learning_rate": 3.2865000000000005e-05,
      "loss": 0.3741,
      "step": 6576
    },
    {
      "epoch": 0.3682943218725501,
      "grad_norm": 1.2838785648345947,
      "learning_rate": 3.287e-05,
      "loss": 0.3915,
      "step": 6577
    },
    {
      "epoch": 0.36835031918467914,
      "grad_norm": 1.1639436483383179,
      "learning_rate": 3.2875e-05,
      "loss": 0.4957,
      "step": 6578
    },
    {
      "epoch": 0.36840631649680816,
      "grad_norm": 1.3116776943206787,
      "learning_rate": 3.288e-05,
      "loss": 0.4058,
      "step": 6579
    },
    {
      "epoch": 0.3684623138089372,
      "grad_norm": 1.3612570762634277,
      "learning_rate": 3.2885e-05,
      "loss": 0.4285,
      "step": 6580
    },
    {
      "epoch": 0.3685183111210662,
      "grad_norm": 1.1726627349853516,
      "learning_rate": 3.2890000000000005e-05,
      "loss": 0.3682,
      "step": 6581
    },
    {
      "epoch": 0.3685743084331952,
      "grad_norm": 1.2552701234817505,
      "learning_rate": 3.2895e-05,
      "loss": 0.404,
      "step": 6582
    },
    {
      "epoch": 0.36863030574532424,
      "grad_norm": 1.3821288347244263,
      "learning_rate": 3.29e-05,
      "loss": 0.5341,
      "step": 6583
    },
    {
      "epoch": 0.36868630305745326,
      "grad_norm": 1.3465137481689453,
      "learning_rate": 3.2905000000000004e-05,
      "loss": 0.5916,
      "step": 6584
    },
    {
      "epoch": 0.3687423003695823,
      "grad_norm": 1.141660451889038,
      "learning_rate": 3.291e-05,
      "loss": 0.4408,
      "step": 6585
    },
    {
      "epoch": 0.3687982976817113,
      "grad_norm": 1.1541905403137207,
      "learning_rate": 3.2915e-05,
      "loss": 0.3921,
      "step": 6586
    },
    {
      "epoch": 0.3688542949938403,
      "grad_norm": 1.2907469272613525,
      "learning_rate": 3.292e-05,
      "loss": 0.51,
      "step": 6587
    },
    {
      "epoch": 0.36891029230596933,
      "grad_norm": 1.2633461952209473,
      "learning_rate": 3.2925e-05,
      "loss": 0.4051,
      "step": 6588
    },
    {
      "epoch": 0.36896628961809835,
      "grad_norm": 1.078071117401123,
      "learning_rate": 3.293e-05,
      "loss": 0.3231,
      "step": 6589
    },
    {
      "epoch": 0.3690222869302274,
      "grad_norm": 1.319933295249939,
      "learning_rate": 3.2935e-05,
      "loss": 0.4022,
      "step": 6590
    },
    {
      "epoch": 0.3690782842423564,
      "grad_norm": 1.2563138008117676,
      "learning_rate": 3.2940000000000006e-05,
      "loss": 0.4468,
      "step": 6591
    },
    {
      "epoch": 0.3691342815544854,
      "grad_norm": 1.109421730041504,
      "learning_rate": 3.2945000000000004e-05,
      "loss": 0.3579,
      "step": 6592
    },
    {
      "epoch": 0.36919027886661443,
      "grad_norm": 1.5637937784194946,
      "learning_rate": 3.295e-05,
      "loss": 0.5781,
      "step": 6593
    },
    {
      "epoch": 0.3692462761787434,
      "grad_norm": 1.3978664875030518,
      "learning_rate": 3.2955000000000006e-05,
      "loss": 0.4671,
      "step": 6594
    },
    {
      "epoch": 0.3693022734908724,
      "grad_norm": 1.0309531688690186,
      "learning_rate": 3.296e-05,
      "loss": 0.3874,
      "step": 6595
    },
    {
      "epoch": 0.36935827080300143,
      "grad_norm": 1.4158213138580322,
      "learning_rate": 3.2965e-05,
      "loss": 0.3398,
      "step": 6596
    },
    {
      "epoch": 0.36941426811513045,
      "grad_norm": 1.2440460920333862,
      "learning_rate": 3.297e-05,
      "loss": 0.4682,
      "step": 6597
    },
    {
      "epoch": 0.36947026542725947,
      "grad_norm": 1.157631516456604,
      "learning_rate": 3.2975e-05,
      "loss": 0.363,
      "step": 6598
    },
    {
      "epoch": 0.3695262627393885,
      "grad_norm": 1.1247055530548096,
      "learning_rate": 3.298e-05,
      "loss": 0.4482,
      "step": 6599
    },
    {
      "epoch": 0.3695822600515175,
      "grad_norm": 1.1412831544876099,
      "learning_rate": 3.2985e-05,
      "loss": 0.3781,
      "step": 6600
    },
    {
      "epoch": 0.3696382573636465,
      "grad_norm": 1.1583365201950073,
      "learning_rate": 3.299e-05,
      "loss": 0.3765,
      "step": 6601
    },
    {
      "epoch": 0.36969425467577555,
      "grad_norm": 1.2251805067062378,
      "learning_rate": 3.2995000000000005e-05,
      "loss": 0.4022,
      "step": 6602
    },
    {
      "epoch": 0.36975025198790457,
      "grad_norm": 1.2362018823623657,
      "learning_rate": 3.3e-05,
      "loss": 0.5551,
      "step": 6603
    },
    {
      "epoch": 0.3698062493000336,
      "grad_norm": 1.3730159997940063,
      "learning_rate": 3.3005e-05,
      "loss": 0.4925,
      "step": 6604
    },
    {
      "epoch": 0.3698622466121626,
      "grad_norm": 1.365923285484314,
      "learning_rate": 3.3010000000000004e-05,
      "loss": 0.4339,
      "step": 6605
    },
    {
      "epoch": 0.3699182439242916,
      "grad_norm": 1.1870639324188232,
      "learning_rate": 3.3015e-05,
      "loss": 0.3789,
      "step": 6606
    },
    {
      "epoch": 0.36997424123642064,
      "grad_norm": 4.544864177703857,
      "learning_rate": 3.302e-05,
      "loss": 0.5686,
      "step": 6607
    },
    {
      "epoch": 0.37003023854854966,
      "grad_norm": 1.421617865562439,
      "learning_rate": 3.3025e-05,
      "loss": 0.5282,
      "step": 6608
    },
    {
      "epoch": 0.3700862358606787,
      "grad_norm": 1.3793761730194092,
      "learning_rate": 3.303e-05,
      "loss": 0.4925,
      "step": 6609
    },
    {
      "epoch": 0.3701422331728077,
      "grad_norm": 1.3852094411849976,
      "learning_rate": 3.3035e-05,
      "loss": 0.4386,
      "step": 6610
    },
    {
      "epoch": 0.3701982304849367,
      "grad_norm": 1.2681360244750977,
      "learning_rate": 3.304e-05,
      "loss": 0.3768,
      "step": 6611
    },
    {
      "epoch": 0.37025422779706574,
      "grad_norm": 1.3386889696121216,
      "learning_rate": 3.3045000000000006e-05,
      "loss": 0.4566,
      "step": 6612
    },
    {
      "epoch": 0.37031022510919476,
      "grad_norm": 1.2781789302825928,
      "learning_rate": 3.3050000000000004e-05,
      "loss": 0.4486,
      "step": 6613
    },
    {
      "epoch": 0.3703662224213238,
      "grad_norm": 1.4412894248962402,
      "learning_rate": 3.3055e-05,
      "loss": 0.4828,
      "step": 6614
    },
    {
      "epoch": 0.3704222197334528,
      "grad_norm": 1.2869528532028198,
      "learning_rate": 3.3060000000000005e-05,
      "loss": 0.4226,
      "step": 6615
    },
    {
      "epoch": 0.3704782170455818,
      "grad_norm": 1.3932974338531494,
      "learning_rate": 3.3065e-05,
      "loss": 0.4495,
      "step": 6616
    },
    {
      "epoch": 0.37053421435771083,
      "grad_norm": 1.136078119277954,
      "learning_rate": 3.307e-05,
      "loss": 0.4614,
      "step": 6617
    },
    {
      "epoch": 0.37059021166983985,
      "grad_norm": 1.113147497177124,
      "learning_rate": 3.3075e-05,
      "loss": 0.3735,
      "step": 6618
    },
    {
      "epoch": 0.37064620898196887,
      "grad_norm": 1.1008528470993042,
      "learning_rate": 3.308e-05,
      "loss": 0.4377,
      "step": 6619
    },
    {
      "epoch": 0.3707022062940979,
      "grad_norm": 1.2893664836883545,
      "learning_rate": 3.3085e-05,
      "loss": 0.4508,
      "step": 6620
    },
    {
      "epoch": 0.3707582036062269,
      "grad_norm": 1.1024919748306274,
      "learning_rate": 3.309e-05,
      "loss": 0.3329,
      "step": 6621
    },
    {
      "epoch": 0.37081420091835593,
      "grad_norm": 1.320804476737976,
      "learning_rate": 3.3095e-05,
      "loss": 0.4158,
      "step": 6622
    },
    {
      "epoch": 0.37087019823048495,
      "grad_norm": 1.3588051795959473,
      "learning_rate": 3.3100000000000005e-05,
      "loss": 0.5241,
      "step": 6623
    },
    {
      "epoch": 0.37092619554261397,
      "grad_norm": 1.2875909805297852,
      "learning_rate": 3.3105e-05,
      "loss": 0.3818,
      "step": 6624
    },
    {
      "epoch": 0.370982192854743,
      "grad_norm": 1.308329463005066,
      "learning_rate": 3.311e-05,
      "loss": 0.5868,
      "step": 6625
    },
    {
      "epoch": 0.371038190166872,
      "grad_norm": 1.737322449684143,
      "learning_rate": 3.3115000000000004e-05,
      "loss": 0.4847,
      "step": 6626
    },
    {
      "epoch": 0.371094187479001,
      "grad_norm": 4.365012168884277,
      "learning_rate": 3.312e-05,
      "loss": 0.4465,
      "step": 6627
    },
    {
      "epoch": 0.37115018479113004,
      "grad_norm": 1.3691514730453491,
      "learning_rate": 3.3125e-05,
      "loss": 0.4063,
      "step": 6628
    },
    {
      "epoch": 0.37120618210325906,
      "grad_norm": 1.4230196475982666,
      "learning_rate": 3.313e-05,
      "loss": 0.461,
      "step": 6629
    },
    {
      "epoch": 0.3712621794153881,
      "grad_norm": 1.1951203346252441,
      "learning_rate": 3.3135e-05,
      "loss": 0.4172,
      "step": 6630
    },
    {
      "epoch": 0.3713181767275171,
      "grad_norm": 1.3649755716323853,
      "learning_rate": 3.314e-05,
      "loss": 0.3652,
      "step": 6631
    },
    {
      "epoch": 0.3713741740396461,
      "grad_norm": 1.3521397113800049,
      "learning_rate": 3.3145e-05,
      "loss": 0.4778,
      "step": 6632
    },
    {
      "epoch": 0.37143017135177514,
      "grad_norm": 1.3003733158111572,
      "learning_rate": 3.3150000000000006e-05,
      "loss": 0.3717,
      "step": 6633
    },
    {
      "epoch": 0.37148616866390416,
      "grad_norm": 1.3330631256103516,
      "learning_rate": 3.3155000000000004e-05,
      "loss": 0.4405,
      "step": 6634
    },
    {
      "epoch": 0.3715421659760331,
      "grad_norm": 1.1360352039337158,
      "learning_rate": 3.316e-05,
      "loss": 0.4608,
      "step": 6635
    },
    {
      "epoch": 0.37159816328816214,
      "grad_norm": 1.2262935638427734,
      "learning_rate": 3.3165e-05,
      "loss": 0.4187,
      "step": 6636
    },
    {
      "epoch": 0.37165416060029116,
      "grad_norm": 1.1865484714508057,
      "learning_rate": 3.317e-05,
      "loss": 0.403,
      "step": 6637
    },
    {
      "epoch": 0.3717101579124202,
      "grad_norm": 1.1020766496658325,
      "learning_rate": 3.3175e-05,
      "loss": 0.4161,
      "step": 6638
    },
    {
      "epoch": 0.3717661552245492,
      "grad_norm": 1.1455082893371582,
      "learning_rate": 3.318e-05,
      "loss": 0.4625,
      "step": 6639
    },
    {
      "epoch": 0.3718221525366782,
      "grad_norm": 1.2724734544754028,
      "learning_rate": 3.3185e-05,
      "loss": 0.4234,
      "step": 6640
    },
    {
      "epoch": 0.37187814984880724,
      "grad_norm": 1.2132536172866821,
      "learning_rate": 3.319e-05,
      "loss": 0.3902,
      "step": 6641
    },
    {
      "epoch": 0.37193414716093626,
      "grad_norm": 1.353208303451538,
      "learning_rate": 3.3195e-05,
      "loss": 0.5316,
      "step": 6642
    },
    {
      "epoch": 0.3719901444730653,
      "grad_norm": 1.3749722242355347,
      "learning_rate": 3.32e-05,
      "loss": 0.3778,
      "step": 6643
    },
    {
      "epoch": 0.3720461417851943,
      "grad_norm": 1.0999027490615845,
      "learning_rate": 3.3205000000000005e-05,
      "loss": 0.3242,
      "step": 6644
    },
    {
      "epoch": 0.3721021390973233,
      "grad_norm": 1.3657257556915283,
      "learning_rate": 3.321e-05,
      "loss": 0.5076,
      "step": 6645
    },
    {
      "epoch": 0.37215813640945233,
      "grad_norm": 1.0309100151062012,
      "learning_rate": 3.3215e-05,
      "loss": 0.3803,
      "step": 6646
    },
    {
      "epoch": 0.37221413372158135,
      "grad_norm": 1.1874897480010986,
      "learning_rate": 3.3220000000000004e-05,
      "loss": 0.4817,
      "step": 6647
    },
    {
      "epoch": 0.37227013103371037,
      "grad_norm": 1.455005407333374,
      "learning_rate": 3.3225e-05,
      "loss": 0.4271,
      "step": 6648
    },
    {
      "epoch": 0.3723261283458394,
      "grad_norm": 1.4285304546356201,
      "learning_rate": 3.323e-05,
      "loss": 0.3734,
      "step": 6649
    },
    {
      "epoch": 0.3723821256579684,
      "grad_norm": 1.0313078165054321,
      "learning_rate": 3.3235e-05,
      "loss": 0.3734,
      "step": 6650
    },
    {
      "epoch": 0.37243812297009743,
      "grad_norm": 0.9433008432388306,
      "learning_rate": 3.324e-05,
      "loss": 0.318,
      "step": 6651
    },
    {
      "epoch": 0.37249412028222645,
      "grad_norm": 1.1286393404006958,
      "learning_rate": 3.3245000000000004e-05,
      "loss": 0.459,
      "step": 6652
    },
    {
      "epoch": 0.37255011759435547,
      "grad_norm": 0.9914267659187317,
      "learning_rate": 3.325e-05,
      "loss": 0.3964,
      "step": 6653
    },
    {
      "epoch": 0.3726061149064845,
      "grad_norm": 1.1322031021118164,
      "learning_rate": 3.3255000000000006e-05,
      "loss": 0.5777,
      "step": 6654
    },
    {
      "epoch": 0.3726621122186135,
      "grad_norm": 1.4263185262680054,
      "learning_rate": 3.3260000000000003e-05,
      "loss": 0.5518,
      "step": 6655
    },
    {
      "epoch": 0.3727181095307425,
      "grad_norm": 1.1430028676986694,
      "learning_rate": 3.3265e-05,
      "loss": 0.4088,
      "step": 6656
    },
    {
      "epoch": 0.37277410684287154,
      "grad_norm": 1.332008957862854,
      "learning_rate": 3.327e-05,
      "loss": 0.4516,
      "step": 6657
    },
    {
      "epoch": 0.37283010415500056,
      "grad_norm": 1.0811599493026733,
      "learning_rate": 3.3275e-05,
      "loss": 0.3934,
      "step": 6658
    },
    {
      "epoch": 0.3728861014671296,
      "grad_norm": 1.345110297203064,
      "learning_rate": 3.328e-05,
      "loss": 0.4315,
      "step": 6659
    },
    {
      "epoch": 0.3729420987792586,
      "grad_norm": 1.412089467048645,
      "learning_rate": 3.3285e-05,
      "loss": 0.5837,
      "step": 6660
    },
    {
      "epoch": 0.3729980960913876,
      "grad_norm": 1.514289379119873,
      "learning_rate": 3.329e-05,
      "loss": 0.4549,
      "step": 6661
    },
    {
      "epoch": 0.37305409340351664,
      "grad_norm": 1.1381101608276367,
      "learning_rate": 3.3295000000000006e-05,
      "loss": 0.3819,
      "step": 6662
    },
    {
      "epoch": 0.37311009071564566,
      "grad_norm": 1.1717133522033691,
      "learning_rate": 3.33e-05,
      "loss": 0.4161,
      "step": 6663
    },
    {
      "epoch": 0.3731660880277747,
      "grad_norm": 1.2617908716201782,
      "learning_rate": 3.3305e-05,
      "loss": 0.4284,
      "step": 6664
    },
    {
      "epoch": 0.3732220853399037,
      "grad_norm": 1.1256204843521118,
      "learning_rate": 3.3310000000000005e-05,
      "loss": 0.3324,
      "step": 6665
    },
    {
      "epoch": 0.3732780826520327,
      "grad_norm": 1.3104017972946167,
      "learning_rate": 3.3315e-05,
      "loss": 0.603,
      "step": 6666
    },
    {
      "epoch": 0.37333407996416174,
      "grad_norm": 1.2038390636444092,
      "learning_rate": 3.332e-05,
      "loss": 0.5616,
      "step": 6667
    },
    {
      "epoch": 0.37339007727629075,
      "grad_norm": 1.3213260173797607,
      "learning_rate": 3.3325000000000004e-05,
      "loss": 0.4087,
      "step": 6668
    },
    {
      "epoch": 0.3734460745884198,
      "grad_norm": 1.4307289123535156,
      "learning_rate": 3.333e-05,
      "loss": 0.5434,
      "step": 6669
    },
    {
      "epoch": 0.3735020719005488,
      "grad_norm": 1.0907399654388428,
      "learning_rate": 3.3335e-05,
      "loss": 0.3285,
      "step": 6670
    },
    {
      "epoch": 0.3735580692126778,
      "grad_norm": 1.2606312036514282,
      "learning_rate": 3.3339999999999996e-05,
      "loss": 0.3946,
      "step": 6671
    },
    {
      "epoch": 0.37361406652480683,
      "grad_norm": 1.20966637134552,
      "learning_rate": 3.334500000000001e-05,
      "loss": 0.3584,
      "step": 6672
    },
    {
      "epoch": 0.37367006383693585,
      "grad_norm": 1.0477896928787231,
      "learning_rate": 3.3350000000000004e-05,
      "loss": 0.3188,
      "step": 6673
    },
    {
      "epoch": 0.37372606114906487,
      "grad_norm": 1.3895022869110107,
      "learning_rate": 3.3355e-05,
      "loss": 0.6125,
      "step": 6674
    },
    {
      "epoch": 0.3737820584611939,
      "grad_norm": 1.870864748954773,
      "learning_rate": 3.336e-05,
      "loss": 0.4728,
      "step": 6675
    },
    {
      "epoch": 0.3738380557733229,
      "grad_norm": 1.2265498638153076,
      "learning_rate": 3.3365e-05,
      "loss": 0.3698,
      "step": 6676
    },
    {
      "epoch": 0.37389405308545187,
      "grad_norm": 1.2869940996170044,
      "learning_rate": 3.337e-05,
      "loss": 0.5283,
      "step": 6677
    },
    {
      "epoch": 0.3739500503975809,
      "grad_norm": 1.4047949314117432,
      "learning_rate": 3.3375e-05,
      "loss": 0.3931,
      "step": 6678
    },
    {
      "epoch": 0.3740060477097099,
      "grad_norm": 1.167543888092041,
      "learning_rate": 3.338e-05,
      "loss": 0.4388,
      "step": 6679
    },
    {
      "epoch": 0.37406204502183893,
      "grad_norm": 1.3333889245986938,
      "learning_rate": 3.3385e-05,
      "loss": 0.3845,
      "step": 6680
    },
    {
      "epoch": 0.37411804233396795,
      "grad_norm": 1.2822749614715576,
      "learning_rate": 3.339e-05,
      "loss": 0.4201,
      "step": 6681
    },
    {
      "epoch": 0.37417403964609697,
      "grad_norm": 1.1371928453445435,
      "learning_rate": 3.3395e-05,
      "loss": 0.3391,
      "step": 6682
    },
    {
      "epoch": 0.374230036958226,
      "grad_norm": 1.3467376232147217,
      "learning_rate": 3.3400000000000005e-05,
      "loss": 0.4459,
      "step": 6683
    },
    {
      "epoch": 0.374286034270355,
      "grad_norm": 1.1777925491333008,
      "learning_rate": 3.3405e-05,
      "loss": 0.4282,
      "step": 6684
    },
    {
      "epoch": 0.374342031582484,
      "grad_norm": 1.192771077156067,
      "learning_rate": 3.341e-05,
      "loss": 0.4234,
      "step": 6685
    },
    {
      "epoch": 0.37439802889461304,
      "grad_norm": 1.1964887380599976,
      "learning_rate": 3.3415000000000004e-05,
      "loss": 0.4521,
      "step": 6686
    },
    {
      "epoch": 0.37445402620674206,
      "grad_norm": 1.1602240800857544,
      "learning_rate": 3.342e-05,
      "loss": 0.5186,
      "step": 6687
    },
    {
      "epoch": 0.3745100235188711,
      "grad_norm": 1.2414394617080688,
      "learning_rate": 3.3425e-05,
      "loss": 0.3399,
      "step": 6688
    },
    {
      "epoch": 0.3745660208310001,
      "grad_norm": 1.127994418144226,
      "learning_rate": 3.3430000000000003e-05,
      "loss": 0.5001,
      "step": 6689
    },
    {
      "epoch": 0.3746220181431291,
      "grad_norm": 1.3012959957122803,
      "learning_rate": 3.3435e-05,
      "loss": 0.4527,
      "step": 6690
    },
    {
      "epoch": 0.37467801545525814,
      "grad_norm": 1.074070692062378,
      "learning_rate": 3.344e-05,
      "loss": 0.401,
      "step": 6691
    },
    {
      "epoch": 0.37473401276738716,
      "grad_norm": 1.3016616106033325,
      "learning_rate": 3.3445e-05,
      "loss": 0.4612,
      "step": 6692
    },
    {
      "epoch": 0.3747900100795162,
      "grad_norm": 1.3945709466934204,
      "learning_rate": 3.345000000000001e-05,
      "loss": 0.4831,
      "step": 6693
    },
    {
      "epoch": 0.3748460073916452,
      "grad_norm": 1.1574516296386719,
      "learning_rate": 3.3455000000000004e-05,
      "loss": 0.5866,
      "step": 6694
    },
    {
      "epoch": 0.3749020047037742,
      "grad_norm": 1.5415083169937134,
      "learning_rate": 3.346e-05,
      "loss": 0.4111,
      "step": 6695
    },
    {
      "epoch": 0.37495800201590324,
      "grad_norm": 1.3542542457580566,
      "learning_rate": 3.3465e-05,
      "loss": 0.4482,
      "step": 6696
    },
    {
      "epoch": 0.37501399932803225,
      "grad_norm": 1.5849406719207764,
      "learning_rate": 3.347e-05,
      "loss": 0.3569,
      "step": 6697
    },
    {
      "epoch": 0.3750699966401613,
      "grad_norm": 1.2463269233703613,
      "learning_rate": 3.3475e-05,
      "loss": 0.4923,
      "step": 6698
    },
    {
      "epoch": 0.3751259939522903,
      "grad_norm": 1.2878328561782837,
      "learning_rate": 3.348e-05,
      "loss": 0.4517,
      "step": 6699
    },
    {
      "epoch": 0.3751819912644193,
      "grad_norm": 1.3001095056533813,
      "learning_rate": 3.3485e-05,
      "loss": 0.4187,
      "step": 6700
    },
    {
      "epoch": 0.37523798857654833,
      "grad_norm": 1.1260403394699097,
      "learning_rate": 3.349e-05,
      "loss": 0.3583,
      "step": 6701
    },
    {
      "epoch": 0.37529398588867735,
      "grad_norm": 1.2898516654968262,
      "learning_rate": 3.3495000000000004e-05,
      "loss": 0.4044,
      "step": 6702
    },
    {
      "epoch": 0.37534998320080637,
      "grad_norm": 1.0880471467971802,
      "learning_rate": 3.35e-05,
      "loss": 0.3948,
      "step": 6703
    },
    {
      "epoch": 0.3754059805129354,
      "grad_norm": 0.983112096786499,
      "learning_rate": 3.3505000000000005e-05,
      "loss": 0.3603,
      "step": 6704
    },
    {
      "epoch": 0.3754619778250644,
      "grad_norm": 1.2240639925003052,
      "learning_rate": 3.351e-05,
      "loss": 0.4636,
      "step": 6705
    },
    {
      "epoch": 0.3755179751371934,
      "grad_norm": 1.7151166200637817,
      "learning_rate": 3.3515e-05,
      "loss": 0.5167,
      "step": 6706
    },
    {
      "epoch": 0.37557397244932245,
      "grad_norm": 1.0267263650894165,
      "learning_rate": 3.3520000000000004e-05,
      "loss": 0.3929,
      "step": 6707
    },
    {
      "epoch": 0.37562996976145147,
      "grad_norm": 2.0249452590942383,
      "learning_rate": 3.3525e-05,
      "loss": 0.4115,
      "step": 6708
    },
    {
      "epoch": 0.3756859670735805,
      "grad_norm": 1.2309340238571167,
      "learning_rate": 3.353e-05,
      "loss": 0.4488,
      "step": 6709
    },
    {
      "epoch": 0.3757419643857095,
      "grad_norm": 1.1514673233032227,
      "learning_rate": 3.3534999999999997e-05,
      "loss": 0.3519,
      "step": 6710
    },
    {
      "epoch": 0.3757979616978385,
      "grad_norm": 0.9589496850967407,
      "learning_rate": 3.354e-05,
      "loss": 0.4595,
      "step": 6711
    },
    {
      "epoch": 0.37585395900996754,
      "grad_norm": 1.345346450805664,
      "learning_rate": 3.3545000000000005e-05,
      "loss": 0.4561,
      "step": 6712
    },
    {
      "epoch": 0.37590995632209656,
      "grad_norm": 1.1441172361373901,
      "learning_rate": 3.355e-05,
      "loss": 0.3547,
      "step": 6713
    },
    {
      "epoch": 0.3759659536342256,
      "grad_norm": 1.4294555187225342,
      "learning_rate": 3.3555e-05,
      "loss": 0.4995,
      "step": 6714
    },
    {
      "epoch": 0.3760219509463546,
      "grad_norm": 1.6530765295028687,
      "learning_rate": 3.3560000000000004e-05,
      "loss": 0.4534,
      "step": 6715
    },
    {
      "epoch": 0.3760779482584836,
      "grad_norm": 1.140238642692566,
      "learning_rate": 3.3565e-05,
      "loss": 0.4516,
      "step": 6716
    },
    {
      "epoch": 0.37613394557061264,
      "grad_norm": 1.2830810546875,
      "learning_rate": 3.357e-05,
      "loss": 0.4988,
      "step": 6717
    },
    {
      "epoch": 0.3761899428827416,
      "grad_norm": 1.5989134311676025,
      "learning_rate": 3.3575e-05,
      "loss": 0.4608,
      "step": 6718
    },
    {
      "epoch": 0.3762459401948706,
      "grad_norm": 1.3424721956253052,
      "learning_rate": 3.358e-05,
      "loss": 0.6089,
      "step": 6719
    },
    {
      "epoch": 0.37630193750699964,
      "grad_norm": 1.6010934114456177,
      "learning_rate": 3.3585e-05,
      "loss": 0.6814,
      "step": 6720
    },
    {
      "epoch": 0.37635793481912866,
      "grad_norm": 1.0401952266693115,
      "learning_rate": 3.359e-05,
      "loss": 0.3748,
      "step": 6721
    },
    {
      "epoch": 0.3764139321312577,
      "grad_norm": 1.324708104133606,
      "learning_rate": 3.3595000000000006e-05,
      "loss": 0.3935,
      "step": 6722
    },
    {
      "epoch": 0.3764699294433867,
      "grad_norm": 1.4354658126831055,
      "learning_rate": 3.3600000000000004e-05,
      "loss": 0.4073,
      "step": 6723
    },
    {
      "epoch": 0.3765259267555157,
      "grad_norm": 1.1871925592422485,
      "learning_rate": 3.3605e-05,
      "loss": 0.3696,
      "step": 6724
    },
    {
      "epoch": 0.37658192406764474,
      "grad_norm": 1.34526789188385,
      "learning_rate": 3.3610000000000005e-05,
      "loss": 0.4582,
      "step": 6725
    },
    {
      "epoch": 0.37663792137977375,
      "grad_norm": 1.047061562538147,
      "learning_rate": 3.3615e-05,
      "loss": 0.401,
      "step": 6726
    },
    {
      "epoch": 0.3766939186919028,
      "grad_norm": 1.6513768434524536,
      "learning_rate": 3.362e-05,
      "loss": 0.4484,
      "step": 6727
    },
    {
      "epoch": 0.3767499160040318,
      "grad_norm": 1.3245466947555542,
      "learning_rate": 3.3625000000000004e-05,
      "loss": 0.4729,
      "step": 6728
    },
    {
      "epoch": 0.3768059133161608,
      "grad_norm": 2.3662190437316895,
      "learning_rate": 3.363e-05,
      "loss": 0.4553,
      "step": 6729
    },
    {
      "epoch": 0.37686191062828983,
      "grad_norm": 1.32633638381958,
      "learning_rate": 3.3635e-05,
      "loss": 0.4385,
      "step": 6730
    },
    {
      "epoch": 0.37691790794041885,
      "grad_norm": 1.370506763458252,
      "learning_rate": 3.3639999999999996e-05,
      "loss": 0.5247,
      "step": 6731
    },
    {
      "epoch": 0.37697390525254787,
      "grad_norm": 1.3929905891418457,
      "learning_rate": 3.364500000000001e-05,
      "loss": 0.5001,
      "step": 6732
    },
    {
      "epoch": 0.3770299025646769,
      "grad_norm": 1.440767526626587,
      "learning_rate": 3.3650000000000005e-05,
      "loss": 0.5486,
      "step": 6733
    },
    {
      "epoch": 0.3770858998768059,
      "grad_norm": 1.2727113962173462,
      "learning_rate": 3.3655e-05,
      "loss": 0.5041,
      "step": 6734
    },
    {
      "epoch": 0.3771418971889349,
      "grad_norm": 1.1316595077514648,
      "learning_rate": 3.366e-05,
      "loss": 0.4268,
      "step": 6735
    },
    {
      "epoch": 0.37719789450106395,
      "grad_norm": 1.1441484689712524,
      "learning_rate": 3.3665000000000004e-05,
      "loss": 0.3752,
      "step": 6736
    },
    {
      "epoch": 0.37725389181319297,
      "grad_norm": 1.4344395399093628,
      "learning_rate": 3.367e-05,
      "loss": 0.4142,
      "step": 6737
    },
    {
      "epoch": 0.377309889125322,
      "grad_norm": 1.3655970096588135,
      "learning_rate": 3.3675e-05,
      "loss": 0.3492,
      "step": 6738
    },
    {
      "epoch": 0.377365886437451,
      "grad_norm": 1.1607861518859863,
      "learning_rate": 3.368e-05,
      "loss": 0.451,
      "step": 6739
    },
    {
      "epoch": 0.37742188374958,
      "grad_norm": 3.041842460632324,
      "learning_rate": 3.3685e-05,
      "loss": 0.5925,
      "step": 6740
    },
    {
      "epoch": 0.37747788106170904,
      "grad_norm": 1.3775370121002197,
      "learning_rate": 3.369e-05,
      "loss": 0.3979,
      "step": 6741
    },
    {
      "epoch": 0.37753387837383806,
      "grad_norm": 1.4027009010314941,
      "learning_rate": 3.3695e-05,
      "loss": 0.459,
      "step": 6742
    },
    {
      "epoch": 0.3775898756859671,
      "grad_norm": 1.8847804069519043,
      "learning_rate": 3.3700000000000006e-05,
      "loss": 0.6248,
      "step": 6743
    },
    {
      "epoch": 0.3776458729980961,
      "grad_norm": 1.2024933099746704,
      "learning_rate": 3.3705000000000003e-05,
      "loss": 0.4764,
      "step": 6744
    },
    {
      "epoch": 0.3777018703102251,
      "grad_norm": 1.3369094133377075,
      "learning_rate": 3.371e-05,
      "loss": 0.4192,
      "step": 6745
    },
    {
      "epoch": 0.37775786762235414,
      "grad_norm": 1.276689887046814,
      "learning_rate": 3.3715000000000005e-05,
      "loss": 0.4924,
      "step": 6746
    },
    {
      "epoch": 0.37781386493448316,
      "grad_norm": 1.388626217842102,
      "learning_rate": 3.372e-05,
      "loss": 0.4662,
      "step": 6747
    },
    {
      "epoch": 0.3778698622466122,
      "grad_norm": 1.1112879514694214,
      "learning_rate": 3.3725e-05,
      "loss": 0.3071,
      "step": 6748
    },
    {
      "epoch": 0.3779258595587412,
      "grad_norm": 1.2572911977767944,
      "learning_rate": 3.373e-05,
      "loss": 0.4241,
      "step": 6749
    },
    {
      "epoch": 0.3779818568708702,
      "grad_norm": 2.8464155197143555,
      "learning_rate": 3.3735e-05,
      "loss": 0.5338,
      "step": 6750
    },
    {
      "epoch": 0.37803785418299923,
      "grad_norm": 1.3102375268936157,
      "learning_rate": 3.374e-05,
      "loss": 0.3824,
      "step": 6751
    },
    {
      "epoch": 0.37809385149512825,
      "grad_norm": 1.555628776550293,
      "learning_rate": 3.3745e-05,
      "loss": 0.5413,
      "step": 6752
    },
    {
      "epoch": 0.37814984880725727,
      "grad_norm": 1.146290898323059,
      "learning_rate": 3.375000000000001e-05,
      "loss": 0.3548,
      "step": 6753
    },
    {
      "epoch": 0.3782058461193863,
      "grad_norm": 1.0477023124694824,
      "learning_rate": 3.3755000000000005e-05,
      "loss": 0.3244,
      "step": 6754
    },
    {
      "epoch": 0.3782618434315153,
      "grad_norm": 1.1944644451141357,
      "learning_rate": 3.376e-05,
      "loss": 0.4431,
      "step": 6755
    },
    {
      "epoch": 0.37831784074364433,
      "grad_norm": 1.236586332321167,
      "learning_rate": 3.3765e-05,
      "loss": 0.4362,
      "step": 6756
    },
    {
      "epoch": 0.37837383805577335,
      "grad_norm": 13.946683883666992,
      "learning_rate": 3.3770000000000004e-05,
      "loss": 0.399,
      "step": 6757
    },
    {
      "epoch": 0.37842983536790237,
      "grad_norm": 1.2594490051269531,
      "learning_rate": 3.3775e-05,
      "loss": 0.3616,
      "step": 6758
    },
    {
      "epoch": 0.37848583268003133,
      "grad_norm": 1.4111195802688599,
      "learning_rate": 3.378e-05,
      "loss": 0.4966,
      "step": 6759
    },
    {
      "epoch": 0.37854182999216035,
      "grad_norm": 1.3201584815979004,
      "learning_rate": 3.3785e-05,
      "loss": 0.386,
      "step": 6760
    },
    {
      "epoch": 0.37859782730428937,
      "grad_norm": 1.3126658201217651,
      "learning_rate": 3.379e-05,
      "loss": 0.506,
      "step": 6761
    },
    {
      "epoch": 0.3786538246164184,
      "grad_norm": 1.1481921672821045,
      "learning_rate": 3.3795e-05,
      "loss": 0.3953,
      "step": 6762
    },
    {
      "epoch": 0.3787098219285474,
      "grad_norm": 1.41203773021698,
      "learning_rate": 3.38e-05,
      "loss": 0.3521,
      "step": 6763
    },
    {
      "epoch": 0.3787658192406764,
      "grad_norm": 1.215990424156189,
      "learning_rate": 3.3805000000000006e-05,
      "loss": 0.4788,
      "step": 6764
    },
    {
      "epoch": 0.37882181655280545,
      "grad_norm": 1.1901772022247314,
      "learning_rate": 3.381e-05,
      "loss": 0.3711,
      "step": 6765
    },
    {
      "epoch": 0.37887781386493447,
      "grad_norm": 1.2417351007461548,
      "learning_rate": 3.3815e-05,
      "loss": 0.4276,
      "step": 6766
    },
    {
      "epoch": 0.3789338111770635,
      "grad_norm": 1.1263600587844849,
      "learning_rate": 3.3820000000000005e-05,
      "loss": 0.4496,
      "step": 6767
    },
    {
      "epoch": 0.3789898084891925,
      "grad_norm": 1.247110366821289,
      "learning_rate": 3.3825e-05,
      "loss": 0.4089,
      "step": 6768
    },
    {
      "epoch": 0.3790458058013215,
      "grad_norm": 1.2428250312805176,
      "learning_rate": 3.383e-05,
      "loss": 0.4079,
      "step": 6769
    },
    {
      "epoch": 0.37910180311345054,
      "grad_norm": 1.5003178119659424,
      "learning_rate": 3.3835e-05,
      "loss": 0.5265,
      "step": 6770
    },
    {
      "epoch": 0.37915780042557956,
      "grad_norm": 1.2733747959136963,
      "learning_rate": 3.384e-05,
      "loss": 0.4329,
      "step": 6771
    },
    {
      "epoch": 0.3792137977377086,
      "grad_norm": 1.390162706375122,
      "learning_rate": 3.3845e-05,
      "loss": 0.4958,
      "step": 6772
    },
    {
      "epoch": 0.3792697950498376,
      "grad_norm": 1.2606457471847534,
      "learning_rate": 3.385e-05,
      "loss": 0.4001,
      "step": 6773
    },
    {
      "epoch": 0.3793257923619666,
      "grad_norm": 1.286514401435852,
      "learning_rate": 3.3855e-05,
      "loss": 0.3534,
      "step": 6774
    },
    {
      "epoch": 0.37938178967409564,
      "grad_norm": 1.1558682918548584,
      "learning_rate": 3.3860000000000004e-05,
      "loss": 0.4396,
      "step": 6775
    },
    {
      "epoch": 0.37943778698622466,
      "grad_norm": 1.3537018299102783,
      "learning_rate": 3.3865e-05,
      "loss": 0.4584,
      "step": 6776
    },
    {
      "epoch": 0.3794937842983537,
      "grad_norm": 1.4790765047073364,
      "learning_rate": 3.387e-05,
      "loss": 0.5101,
      "step": 6777
    },
    {
      "epoch": 0.3795497816104827,
      "grad_norm": 1.381075382232666,
      "learning_rate": 3.3875000000000003e-05,
      "loss": 0.4182,
      "step": 6778
    },
    {
      "epoch": 0.3796057789226117,
      "grad_norm": 1.909953236579895,
      "learning_rate": 3.388e-05,
      "loss": 0.4372,
      "step": 6779
    },
    {
      "epoch": 0.37966177623474073,
      "grad_norm": 1.7746893167495728,
      "learning_rate": 3.3885e-05,
      "loss": 0.4639,
      "step": 6780
    },
    {
      "epoch": 0.37971777354686975,
      "grad_norm": 1.4533138275146484,
      "learning_rate": 3.389e-05,
      "loss": 0.4153,
      "step": 6781
    },
    {
      "epoch": 0.37977377085899877,
      "grad_norm": 1.1535168886184692,
      "learning_rate": 3.3895e-05,
      "loss": 0.3526,
      "step": 6782
    },
    {
      "epoch": 0.3798297681711278,
      "grad_norm": 1.2879915237426758,
      "learning_rate": 3.3900000000000004e-05,
      "loss": 0.3643,
      "step": 6783
    },
    {
      "epoch": 0.3798857654832568,
      "grad_norm": 1.2191060781478882,
      "learning_rate": 3.3905e-05,
      "loss": 0.5161,
      "step": 6784
    },
    {
      "epoch": 0.37994176279538583,
      "grad_norm": 1.0810246467590332,
      "learning_rate": 3.3910000000000006e-05,
      "loss": 0.4164,
      "step": 6785
    },
    {
      "epoch": 0.37999776010751485,
      "grad_norm": 1.4238454103469849,
      "learning_rate": 3.3915e-05,
      "loss": 0.426,
      "step": 6786
    },
    {
      "epoch": 0.38005375741964387,
      "grad_norm": 1.1587119102478027,
      "learning_rate": 3.392e-05,
      "loss": 0.3992,
      "step": 6787
    },
    {
      "epoch": 0.3801097547317729,
      "grad_norm": 1.2623403072357178,
      "learning_rate": 3.3925e-05,
      "loss": 0.4614,
      "step": 6788
    },
    {
      "epoch": 0.3801657520439019,
      "grad_norm": 1.398996114730835,
      "learning_rate": 3.393e-05,
      "loss": 0.4433,
      "step": 6789
    },
    {
      "epoch": 0.3802217493560309,
      "grad_norm": 1.2736495733261108,
      "learning_rate": 3.3935e-05,
      "loss": 0.3597,
      "step": 6790
    },
    {
      "epoch": 0.38027774666815994,
      "grad_norm": 1.165753960609436,
      "learning_rate": 3.394e-05,
      "loss": 0.3583,
      "step": 6791
    },
    {
      "epoch": 0.38033374398028896,
      "grad_norm": 1.2651578187942505,
      "learning_rate": 3.3945e-05,
      "loss": 0.4531,
      "step": 6792
    },
    {
      "epoch": 0.380389741292418,
      "grad_norm": 1.4925222396850586,
      "learning_rate": 3.3950000000000005e-05,
      "loss": 0.4597,
      "step": 6793
    },
    {
      "epoch": 0.380445738604547,
      "grad_norm": 1.1772981882095337,
      "learning_rate": 3.3955e-05,
      "loss": 0.4551,
      "step": 6794
    },
    {
      "epoch": 0.380501735916676,
      "grad_norm": 1.0072182416915894,
      "learning_rate": 3.396e-05,
      "loss": 0.3487,
      "step": 6795
    },
    {
      "epoch": 0.38055773322880504,
      "grad_norm": 1.4475287199020386,
      "learning_rate": 3.3965000000000004e-05,
      "loss": 0.5036,
      "step": 6796
    },
    {
      "epoch": 0.38061373054093406,
      "grad_norm": 1.2004691362380981,
      "learning_rate": 3.397e-05,
      "loss": 0.4327,
      "step": 6797
    },
    {
      "epoch": 0.3806697278530631,
      "grad_norm": 1.0906919240951538,
      "learning_rate": 3.3975e-05,
      "loss": 0.3438,
      "step": 6798
    },
    {
      "epoch": 0.3807257251651921,
      "grad_norm": 1.304298996925354,
      "learning_rate": 3.398e-05,
      "loss": 0.4037,
      "step": 6799
    },
    {
      "epoch": 0.3807817224773211,
      "grad_norm": 1.767535924911499,
      "learning_rate": 3.3985e-05,
      "loss": 0.5008,
      "step": 6800
    },
    {
      "epoch": 0.3808377197894501,
      "grad_norm": 1.1377581357955933,
      "learning_rate": 3.399e-05,
      "loss": 0.3894,
      "step": 6801
    },
    {
      "epoch": 0.3808937171015791,
      "grad_norm": 1.1364119052886963,
      "learning_rate": 3.3995e-05,
      "loss": 0.433,
      "step": 6802
    },
    {
      "epoch": 0.3809497144137081,
      "grad_norm": 1.2736361026763916,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.4333,
      "step": 6803
    },
    {
      "epoch": 0.38100571172583714,
      "grad_norm": 1.2721407413482666,
      "learning_rate": 3.4005000000000004e-05,
      "loss": 0.3837,
      "step": 6804
    },
    {
      "epoch": 0.38106170903796616,
      "grad_norm": 0.9903769493103027,
      "learning_rate": 3.401e-05,
      "loss": 0.3657,
      "step": 6805
    },
    {
      "epoch": 0.3811177063500952,
      "grad_norm": 1.243751883506775,
      "learning_rate": 3.4015000000000006e-05,
      "loss": 0.415,
      "step": 6806
    },
    {
      "epoch": 0.3811737036622242,
      "grad_norm": 1.1863563060760498,
      "learning_rate": 3.402e-05,
      "loss": 0.527,
      "step": 6807
    },
    {
      "epoch": 0.3812297009743532,
      "grad_norm": 1.2471858263015747,
      "learning_rate": 3.4025e-05,
      "loss": 0.5276,
      "step": 6808
    },
    {
      "epoch": 0.38128569828648223,
      "grad_norm": 1.4560558795928955,
      "learning_rate": 3.403e-05,
      "loss": 0.3881,
      "step": 6809
    },
    {
      "epoch": 0.38134169559861125,
      "grad_norm": 1.310857892036438,
      "learning_rate": 3.4035e-05,
      "loss": 0.6719,
      "step": 6810
    },
    {
      "epoch": 0.38139769291074027,
      "grad_norm": 1.2389583587646484,
      "learning_rate": 3.404e-05,
      "loss": 0.4477,
      "step": 6811
    },
    {
      "epoch": 0.3814536902228693,
      "grad_norm": 1.1277942657470703,
      "learning_rate": 3.4045e-05,
      "loss": 0.3727,
      "step": 6812
    },
    {
      "epoch": 0.3815096875349983,
      "grad_norm": 1.286651611328125,
      "learning_rate": 3.405e-05,
      "loss": 0.3979,
      "step": 6813
    },
    {
      "epoch": 0.38156568484712733,
      "grad_norm": 1.2730367183685303,
      "learning_rate": 3.4055000000000005e-05,
      "loss": 0.3411,
      "step": 6814
    },
    {
      "epoch": 0.38162168215925635,
      "grad_norm": 1.1979669332504272,
      "learning_rate": 3.406e-05,
      "loss": 0.3594,
      "step": 6815
    },
    {
      "epoch": 0.38167767947138537,
      "grad_norm": 1.4920653104782104,
      "learning_rate": 3.4065e-05,
      "loss": 0.4879,
      "step": 6816
    },
    {
      "epoch": 0.3817336767835144,
      "grad_norm": 1.131361722946167,
      "learning_rate": 3.4070000000000004e-05,
      "loss": 0.3472,
      "step": 6817
    },
    {
      "epoch": 0.3817896740956434,
      "grad_norm": 1.2309305667877197,
      "learning_rate": 3.4075e-05,
      "loss": 0.3836,
      "step": 6818
    },
    {
      "epoch": 0.3818456714077724,
      "grad_norm": 1.2341068983078003,
      "learning_rate": 3.408e-05,
      "loss": 0.5457,
      "step": 6819
    },
    {
      "epoch": 0.38190166871990144,
      "grad_norm": 1.188979983329773,
      "learning_rate": 3.4085e-05,
      "loss": 0.4289,
      "step": 6820
    },
    {
      "epoch": 0.38195766603203046,
      "grad_norm": 1.1993087530136108,
      "learning_rate": 3.409e-05,
      "loss": 0.5416,
      "step": 6821
    },
    {
      "epoch": 0.3820136633441595,
      "grad_norm": 0.96775883436203,
      "learning_rate": 3.4095e-05,
      "loss": 0.426,
      "step": 6822
    },
    {
      "epoch": 0.3820696606562885,
      "grad_norm": 2.017106056213379,
      "learning_rate": 3.41e-05,
      "loss": 0.5721,
      "step": 6823
    },
    {
      "epoch": 0.3821256579684175,
      "grad_norm": 1.203216791152954,
      "learning_rate": 3.4105000000000006e-05,
      "loss": 0.3714,
      "step": 6824
    },
    {
      "epoch": 0.38218165528054654,
      "grad_norm": 1.0981453657150269,
      "learning_rate": 3.4110000000000004e-05,
      "loss": 0.3762,
      "step": 6825
    },
    {
      "epoch": 0.38223765259267556,
      "grad_norm": 1.0832266807556152,
      "learning_rate": 3.4115e-05,
      "loss": 0.4779,
      "step": 6826
    },
    {
      "epoch": 0.3822936499048046,
      "grad_norm": 1.519089698791504,
      "learning_rate": 3.412e-05,
      "loss": 0.6131,
      "step": 6827
    },
    {
      "epoch": 0.3823496472169336,
      "grad_norm": 1.205888271331787,
      "learning_rate": 3.4125e-05,
      "loss": 0.4936,
      "step": 6828
    },
    {
      "epoch": 0.3824056445290626,
      "grad_norm": 1.3022352457046509,
      "learning_rate": 3.413e-05,
      "loss": 0.4341,
      "step": 6829
    },
    {
      "epoch": 0.38246164184119164,
      "grad_norm": 1.164675235748291,
      "learning_rate": 3.4135e-05,
      "loss": 0.4531,
      "step": 6830
    },
    {
      "epoch": 0.38251763915332065,
      "grad_norm": 1.1891977787017822,
      "learning_rate": 3.414e-05,
      "loss": 0.531,
      "step": 6831
    },
    {
      "epoch": 0.3825736364654497,
      "grad_norm": 1.1762791872024536,
      "learning_rate": 3.4145e-05,
      "loss": 0.4263,
      "step": 6832
    },
    {
      "epoch": 0.3826296337775787,
      "grad_norm": 1.4778733253479004,
      "learning_rate": 3.415e-05,
      "loss": 0.733,
      "step": 6833
    },
    {
      "epoch": 0.3826856310897077,
      "grad_norm": 1.2638355493545532,
      "learning_rate": 3.4155e-05,
      "loss": 0.404,
      "step": 6834
    },
    {
      "epoch": 0.38274162840183673,
      "grad_norm": 1.3806214332580566,
      "learning_rate": 3.4160000000000005e-05,
      "loss": 0.537,
      "step": 6835
    },
    {
      "epoch": 0.38279762571396575,
      "grad_norm": 1.2141627073287964,
      "learning_rate": 3.4165e-05,
      "loss": 0.457,
      "step": 6836
    },
    {
      "epoch": 0.38285362302609477,
      "grad_norm": 1.233931303024292,
      "learning_rate": 3.417e-05,
      "loss": 0.4319,
      "step": 6837
    },
    {
      "epoch": 0.3829096203382238,
      "grad_norm": 1.4215152263641357,
      "learning_rate": 3.4175000000000004e-05,
      "loss": 0.5703,
      "step": 6838
    },
    {
      "epoch": 0.3829656176503528,
      "grad_norm": 1.1874788999557495,
      "learning_rate": 3.418e-05,
      "loss": 0.5177,
      "step": 6839
    },
    {
      "epoch": 0.3830216149624818,
      "grad_norm": 1.167649269104004,
      "learning_rate": 3.4185e-05,
      "loss": 0.3902,
      "step": 6840
    },
    {
      "epoch": 0.38307761227461085,
      "grad_norm": 1.2854527235031128,
      "learning_rate": 3.419e-05,
      "loss": 0.4082,
      "step": 6841
    },
    {
      "epoch": 0.3831336095867398,
      "grad_norm": 1.080438256263733,
      "learning_rate": 3.4195e-05,
      "loss": 0.4204,
      "step": 6842
    },
    {
      "epoch": 0.38318960689886883,
      "grad_norm": 1.3386930227279663,
      "learning_rate": 3.4200000000000005e-05,
      "loss": 0.5462,
      "step": 6843
    },
    {
      "epoch": 0.38324560421099785,
      "grad_norm": 1.4374494552612305,
      "learning_rate": 3.4205e-05,
      "loss": 0.4368,
      "step": 6844
    },
    {
      "epoch": 0.38330160152312687,
      "grad_norm": 1.7632851600646973,
      "learning_rate": 3.4210000000000006e-05,
      "loss": 0.3249,
      "step": 6845
    },
    {
      "epoch": 0.3833575988352559,
      "grad_norm": 1.5548715591430664,
      "learning_rate": 3.4215000000000004e-05,
      "loss": 0.6289,
      "step": 6846
    },
    {
      "epoch": 0.3834135961473849,
      "grad_norm": 1.3940459489822388,
      "learning_rate": 3.422e-05,
      "loss": 0.4671,
      "step": 6847
    },
    {
      "epoch": 0.3834695934595139,
      "grad_norm": 1.5930670499801636,
      "learning_rate": 3.4225e-05,
      "loss": 0.6257,
      "step": 6848
    },
    {
      "epoch": 0.38352559077164294,
      "grad_norm": 1.5224014520645142,
      "learning_rate": 3.423e-05,
      "loss": 0.4133,
      "step": 6849
    },
    {
      "epoch": 0.38358158808377196,
      "grad_norm": 1.4070916175842285,
      "learning_rate": 3.4235e-05,
      "loss": 0.3598,
      "step": 6850
    },
    {
      "epoch": 0.383637585395901,
      "grad_norm": 1.2294327020645142,
      "learning_rate": 3.424e-05,
      "loss": 0.4053,
      "step": 6851
    },
    {
      "epoch": 0.38369358270803,
      "grad_norm": 1.1390637159347534,
      "learning_rate": 3.4245e-05,
      "loss": 0.4106,
      "step": 6852
    },
    {
      "epoch": 0.383749580020159,
      "grad_norm": 1.3385306596755981,
      "learning_rate": 3.4250000000000006e-05,
      "loss": 0.4459,
      "step": 6853
    },
    {
      "epoch": 0.38380557733228804,
      "grad_norm": 1.1420754194259644,
      "learning_rate": 3.4255e-05,
      "loss": 0.3797,
      "step": 6854
    },
    {
      "epoch": 0.38386157464441706,
      "grad_norm": 1.1515679359436035,
      "learning_rate": 3.426e-05,
      "loss": 0.3679,
      "step": 6855
    },
    {
      "epoch": 0.3839175719565461,
      "grad_norm": 1.1968247890472412,
      "learning_rate": 3.4265000000000005e-05,
      "loss": 0.4082,
      "step": 6856
    },
    {
      "epoch": 0.3839735692686751,
      "grad_norm": 1.1582374572753906,
      "learning_rate": 3.427e-05,
      "loss": 0.3743,
      "step": 6857
    },
    {
      "epoch": 0.3840295665808041,
      "grad_norm": 1.2968791723251343,
      "learning_rate": 3.4275e-05,
      "loss": 0.4489,
      "step": 6858
    },
    {
      "epoch": 0.38408556389293314,
      "grad_norm": 1.2227622270584106,
      "learning_rate": 3.4280000000000004e-05,
      "loss": 0.4424,
      "step": 6859
    },
    {
      "epoch": 0.38414156120506215,
      "grad_norm": 1.312872290611267,
      "learning_rate": 3.4285e-05,
      "loss": 0.4034,
      "step": 6860
    },
    {
      "epoch": 0.3841975585171912,
      "grad_norm": 1.2574517726898193,
      "learning_rate": 3.429e-05,
      "loss": 0.4268,
      "step": 6861
    },
    {
      "epoch": 0.3842535558293202,
      "grad_norm": 1.277334451675415,
      "learning_rate": 3.4294999999999996e-05,
      "loss": 0.4557,
      "step": 6862
    },
    {
      "epoch": 0.3843095531414492,
      "grad_norm": 1.556626558303833,
      "learning_rate": 3.430000000000001e-05,
      "loss": 0.851,
      "step": 6863
    },
    {
      "epoch": 0.38436555045357823,
      "grad_norm": 2.4134163856506348,
      "learning_rate": 3.4305000000000004e-05,
      "loss": 0.49,
      "step": 6864
    },
    {
      "epoch": 0.38442154776570725,
      "grad_norm": 1.1647013425827026,
      "learning_rate": 3.431e-05,
      "loss": 0.6369,
      "step": 6865
    },
    {
      "epoch": 0.38447754507783627,
      "grad_norm": 1.2150636911392212,
      "learning_rate": 3.4315000000000006e-05,
      "loss": 0.5115,
      "step": 6866
    },
    {
      "epoch": 0.3845335423899653,
      "grad_norm": 1.3191980123519897,
      "learning_rate": 3.4320000000000003e-05,
      "loss": 0.428,
      "step": 6867
    },
    {
      "epoch": 0.3845895397020943,
      "grad_norm": 1.2424001693725586,
      "learning_rate": 3.4325e-05,
      "loss": 0.4363,
      "step": 6868
    },
    {
      "epoch": 0.3846455370142233,
      "grad_norm": 1.0711196660995483,
      "learning_rate": 3.433e-05,
      "loss": 0.45,
      "step": 6869
    },
    {
      "epoch": 0.38470153432635235,
      "grad_norm": 1.1021277904510498,
      "learning_rate": 3.4335e-05,
      "loss": 0.3696,
      "step": 6870
    },
    {
      "epoch": 0.38475753163848136,
      "grad_norm": 1.745692491531372,
      "learning_rate": 3.434e-05,
      "loss": 0.4261,
      "step": 6871
    },
    {
      "epoch": 0.3848135289506104,
      "grad_norm": 1.2101738452911377,
      "learning_rate": 3.4345e-05,
      "loss": 0.4441,
      "step": 6872
    },
    {
      "epoch": 0.3848695262627394,
      "grad_norm": 1.2984495162963867,
      "learning_rate": 3.435e-05,
      "loss": 0.6471,
      "step": 6873
    },
    {
      "epoch": 0.3849255235748684,
      "grad_norm": 1.0604979991912842,
      "learning_rate": 3.4355000000000006e-05,
      "loss": 0.4003,
      "step": 6874
    },
    {
      "epoch": 0.38498152088699744,
      "grad_norm": 1.4221092462539673,
      "learning_rate": 3.436e-05,
      "loss": 0.4043,
      "step": 6875
    },
    {
      "epoch": 0.38503751819912646,
      "grad_norm": 1.2192177772521973,
      "learning_rate": 3.4365e-05,
      "loss": 0.4698,
      "step": 6876
    },
    {
      "epoch": 0.3850935155112555,
      "grad_norm": 1.405113697052002,
      "learning_rate": 3.4370000000000005e-05,
      "loss": 0.4693,
      "step": 6877
    },
    {
      "epoch": 0.3851495128233845,
      "grad_norm": 1.3599504232406616,
      "learning_rate": 3.4375e-05,
      "loss": 0.4721,
      "step": 6878
    },
    {
      "epoch": 0.3852055101355135,
      "grad_norm": 2.1586360931396484,
      "learning_rate": 3.438e-05,
      "loss": 0.4071,
      "step": 6879
    },
    {
      "epoch": 0.38526150744764254,
      "grad_norm": 1.1666386127471924,
      "learning_rate": 3.4385000000000004e-05,
      "loss": 0.4566,
      "step": 6880
    },
    {
      "epoch": 0.38531750475977156,
      "grad_norm": 1.0913177728652954,
      "learning_rate": 3.439e-05,
      "loss": 0.3272,
      "step": 6881
    },
    {
      "epoch": 0.3853735020719006,
      "grad_norm": 1.3614965677261353,
      "learning_rate": 3.4395e-05,
      "loss": 0.3895,
      "step": 6882
    },
    {
      "epoch": 0.38542949938402954,
      "grad_norm": 1.186721920967102,
      "learning_rate": 3.4399999999999996e-05,
      "loss": 0.364,
      "step": 6883
    },
    {
      "epoch": 0.38548549669615856,
      "grad_norm": 1.7788968086242676,
      "learning_rate": 3.440500000000001e-05,
      "loss": 0.8067,
      "step": 6884
    },
    {
      "epoch": 0.3855414940082876,
      "grad_norm": 1.1849788427352905,
      "learning_rate": 3.4410000000000004e-05,
      "loss": 0.4643,
      "step": 6885
    },
    {
      "epoch": 0.3855974913204166,
      "grad_norm": 1.4890176057815552,
      "learning_rate": 3.4415e-05,
      "loss": 0.4669,
      "step": 6886
    },
    {
      "epoch": 0.3856534886325456,
      "grad_norm": 1.48883056640625,
      "learning_rate": 3.442e-05,
      "loss": 0.3893,
      "step": 6887
    },
    {
      "epoch": 0.38570948594467463,
      "grad_norm": 1.2005877494812012,
      "learning_rate": 3.4425e-05,
      "loss": 0.5971,
      "step": 6888
    },
    {
      "epoch": 0.38576548325680365,
      "grad_norm": 1.5162949562072754,
      "learning_rate": 3.443e-05,
      "loss": 0.4271,
      "step": 6889
    },
    {
      "epoch": 0.3858214805689327,
      "grad_norm": 1.1552103757858276,
      "learning_rate": 3.4435e-05,
      "loss": 0.37,
      "step": 6890
    },
    {
      "epoch": 0.3858774778810617,
      "grad_norm": 1.5426958799362183,
      "learning_rate": 3.444e-05,
      "loss": 0.4643,
      "step": 6891
    },
    {
      "epoch": 0.3859334751931907,
      "grad_norm": 2.051938056945801,
      "learning_rate": 3.4445e-05,
      "loss": 0.4913,
      "step": 6892
    },
    {
      "epoch": 0.38598947250531973,
      "grad_norm": 1.3490667343139648,
      "learning_rate": 3.445e-05,
      "loss": 0.4701,
      "step": 6893
    },
    {
      "epoch": 0.38604546981744875,
      "grad_norm": 1.3706315755844116,
      "learning_rate": 3.4455e-05,
      "loss": 0.5357,
      "step": 6894
    },
    {
      "epoch": 0.38610146712957777,
      "grad_norm": 1.1113626956939697,
      "learning_rate": 3.4460000000000005e-05,
      "loss": 0.355,
      "step": 6895
    },
    {
      "epoch": 0.3861574644417068,
      "grad_norm": 3.260390281677246,
      "learning_rate": 3.4465e-05,
      "loss": 0.47,
      "step": 6896
    },
    {
      "epoch": 0.3862134617538358,
      "grad_norm": 1.1313581466674805,
      "learning_rate": 3.447e-05,
      "loss": 0.3574,
      "step": 6897
    },
    {
      "epoch": 0.3862694590659648,
      "grad_norm": 1.3877431154251099,
      "learning_rate": 3.4475000000000005e-05,
      "loss": 0.5123,
      "step": 6898
    },
    {
      "epoch": 0.38632545637809385,
      "grad_norm": 1.3723994493484497,
      "learning_rate": 3.448e-05,
      "loss": 0.4964,
      "step": 6899
    },
    {
      "epoch": 0.38638145369022286,
      "grad_norm": 1.228934407234192,
      "learning_rate": 3.4485e-05,
      "loss": 0.4702,
      "step": 6900
    },
    {
      "epoch": 0.3864374510023519,
      "grad_norm": 1.1545768976211548,
      "learning_rate": 3.449e-05,
      "loss": 0.4698,
      "step": 6901
    },
    {
      "epoch": 0.3864934483144809,
      "grad_norm": 1.2166193723678589,
      "learning_rate": 3.4495e-05,
      "loss": 0.4343,
      "step": 6902
    },
    {
      "epoch": 0.3865494456266099,
      "grad_norm": 1.4506065845489502,
      "learning_rate": 3.45e-05,
      "loss": 0.4393,
      "step": 6903
    },
    {
      "epoch": 0.38660544293873894,
      "grad_norm": 1.083709716796875,
      "learning_rate": 3.4505e-05,
      "loss": 0.3242,
      "step": 6904
    },
    {
      "epoch": 0.38666144025086796,
      "grad_norm": 1.1156466007232666,
      "learning_rate": 3.451000000000001e-05,
      "loss": 0.397,
      "step": 6905
    },
    {
      "epoch": 0.386717437562997,
      "grad_norm": 1.5354053974151611,
      "learning_rate": 3.4515000000000004e-05,
      "loss": 0.5582,
      "step": 6906
    },
    {
      "epoch": 0.386773434875126,
      "grad_norm": 1.1571729183197021,
      "learning_rate": 3.452e-05,
      "loss": 0.3811,
      "step": 6907
    },
    {
      "epoch": 0.386829432187255,
      "grad_norm": 1.1743639707565308,
      "learning_rate": 3.4525e-05,
      "loss": 0.5123,
      "step": 6908
    },
    {
      "epoch": 0.38688542949938404,
      "grad_norm": 1.0858750343322754,
      "learning_rate": 3.453e-05,
      "loss": 0.4123,
      "step": 6909
    },
    {
      "epoch": 0.38694142681151306,
      "grad_norm": 1.1950575113296509,
      "learning_rate": 3.4535e-05,
      "loss": 0.4931,
      "step": 6910
    },
    {
      "epoch": 0.3869974241236421,
      "grad_norm": 1.3804479837417603,
      "learning_rate": 3.454e-05,
      "loss": 0.4097,
      "step": 6911
    },
    {
      "epoch": 0.3870534214357711,
      "grad_norm": 1.3190300464630127,
      "learning_rate": 3.4545e-05,
      "loss": 0.3901,
      "step": 6912
    },
    {
      "epoch": 0.3871094187479001,
      "grad_norm": 1.4819968938827515,
      "learning_rate": 3.455e-05,
      "loss": 0.499,
      "step": 6913
    },
    {
      "epoch": 0.38716541606002913,
      "grad_norm": 1.4721925258636475,
      "learning_rate": 3.4555000000000004e-05,
      "loss": 0.4597,
      "step": 6914
    },
    {
      "epoch": 0.38722141337215815,
      "grad_norm": 1.1170094013214111,
      "learning_rate": 3.456e-05,
      "loss": 0.5047,
      "step": 6915
    },
    {
      "epoch": 0.38727741068428717,
      "grad_norm": 1.3028054237365723,
      "learning_rate": 3.4565000000000005e-05,
      "loss": 0.4317,
      "step": 6916
    },
    {
      "epoch": 0.3873334079964162,
      "grad_norm": 1.096895456314087,
      "learning_rate": 3.457e-05,
      "loss": 0.3691,
      "step": 6917
    },
    {
      "epoch": 0.3873894053085452,
      "grad_norm": 1.234584927558899,
      "learning_rate": 3.4575e-05,
      "loss": 0.4658,
      "step": 6918
    },
    {
      "epoch": 0.38744540262067423,
      "grad_norm": 1.1240357160568237,
      "learning_rate": 3.4580000000000004e-05,
      "loss": 0.3645,
      "step": 6919
    },
    {
      "epoch": 0.38750139993280325,
      "grad_norm": 2.325113296508789,
      "learning_rate": 3.4585e-05,
      "loss": 0.3758,
      "step": 6920
    },
    {
      "epoch": 0.38755739724493227,
      "grad_norm": 1.31596839427948,
      "learning_rate": 3.459e-05,
      "loss": 0.5102,
      "step": 6921
    },
    {
      "epoch": 0.3876133945570613,
      "grad_norm": 1.3582637310028076,
      "learning_rate": 3.4594999999999997e-05,
      "loss": 0.4485,
      "step": 6922
    },
    {
      "epoch": 0.3876693918691903,
      "grad_norm": 1.4770441055297852,
      "learning_rate": 3.46e-05,
      "loss": 0.4524,
      "step": 6923
    },
    {
      "epoch": 0.3877253891813193,
      "grad_norm": 1.3404148817062378,
      "learning_rate": 3.4605000000000005e-05,
      "loss": 0.5851,
      "step": 6924
    },
    {
      "epoch": 0.3877813864934483,
      "grad_norm": 1.0689105987548828,
      "learning_rate": 3.461e-05,
      "loss": 0.3645,
      "step": 6925
    },
    {
      "epoch": 0.3878373838055773,
      "grad_norm": 1.2504135370254517,
      "learning_rate": 3.4615e-05,
      "loss": 0.4709,
      "step": 6926
    },
    {
      "epoch": 0.3878933811177063,
      "grad_norm": 1.2252495288848877,
      "learning_rate": 3.4620000000000004e-05,
      "loss": 0.4627,
      "step": 6927
    },
    {
      "epoch": 0.38794937842983535,
      "grad_norm": 1.0927207469940186,
      "learning_rate": 3.4625e-05,
      "loss": 0.388,
      "step": 6928
    },
    {
      "epoch": 0.38800537574196436,
      "grad_norm": 1.1858336925506592,
      "learning_rate": 3.463e-05,
      "loss": 0.4278,
      "step": 6929
    },
    {
      "epoch": 0.3880613730540934,
      "grad_norm": 1.2361561059951782,
      "learning_rate": 3.4635e-05,
      "loss": 0.3855,
      "step": 6930
    },
    {
      "epoch": 0.3881173703662224,
      "grad_norm": 1.2411030530929565,
      "learning_rate": 3.464e-05,
      "loss": 0.4048,
      "step": 6931
    },
    {
      "epoch": 0.3881733676783514,
      "grad_norm": 1.1442571878433228,
      "learning_rate": 3.4645e-05,
      "loss": 0.4149,
      "step": 6932
    },
    {
      "epoch": 0.38822936499048044,
      "grad_norm": 1.3089278936386108,
      "learning_rate": 3.465e-05,
      "loss": 0.5776,
      "step": 6933
    },
    {
      "epoch": 0.38828536230260946,
      "grad_norm": 1.1244102716445923,
      "learning_rate": 3.4655000000000006e-05,
      "loss": 0.3307,
      "step": 6934
    },
    {
      "epoch": 0.3883413596147385,
      "grad_norm": 1.507246732711792,
      "learning_rate": 3.4660000000000004e-05,
      "loss": 0.5329,
      "step": 6935
    },
    {
      "epoch": 0.3883973569268675,
      "grad_norm": 4.239734649658203,
      "learning_rate": 3.4665e-05,
      "loss": 0.4608,
      "step": 6936
    },
    {
      "epoch": 0.3884533542389965,
      "grad_norm": 1.2836531400680542,
      "learning_rate": 3.4670000000000005e-05,
      "loss": 0.3798,
      "step": 6937
    },
    {
      "epoch": 0.38850935155112554,
      "grad_norm": 1.5413106679916382,
      "learning_rate": 3.4675e-05,
      "loss": 0.5689,
      "step": 6938
    },
    {
      "epoch": 0.38856534886325456,
      "grad_norm": 1.3127338886260986,
      "learning_rate": 3.468e-05,
      "loss": 0.4873,
      "step": 6939
    },
    {
      "epoch": 0.3886213461753836,
      "grad_norm": 1.2186760902404785,
      "learning_rate": 3.4685000000000004e-05,
      "loss": 0.3925,
      "step": 6940
    },
    {
      "epoch": 0.3886773434875126,
      "grad_norm": 1.1795878410339355,
      "learning_rate": 3.469e-05,
      "loss": 0.3932,
      "step": 6941
    },
    {
      "epoch": 0.3887333407996416,
      "grad_norm": 1.3185460567474365,
      "learning_rate": 3.4695e-05,
      "loss": 0.3922,
      "step": 6942
    },
    {
      "epoch": 0.38878933811177063,
      "grad_norm": 1.2021520137786865,
      "learning_rate": 3.4699999999999996e-05,
      "loss": 0.3763,
      "step": 6943
    },
    {
      "epoch": 0.38884533542389965,
      "grad_norm": 1.2987655401229858,
      "learning_rate": 3.470500000000001e-05,
      "loss": 0.5663,
      "step": 6944
    },
    {
      "epoch": 0.38890133273602867,
      "grad_norm": 1.2102904319763184,
      "learning_rate": 3.4710000000000005e-05,
      "loss": 0.4477,
      "step": 6945
    },
    {
      "epoch": 0.3889573300481577,
      "grad_norm": 1.288582444190979,
      "learning_rate": 3.4715e-05,
      "loss": 0.3822,
      "step": 6946
    },
    {
      "epoch": 0.3890133273602867,
      "grad_norm": 1.1250205039978027,
      "learning_rate": 3.472e-05,
      "loss": 0.435,
      "step": 6947
    },
    {
      "epoch": 0.38906932467241573,
      "grad_norm": 1.4243067502975464,
      "learning_rate": 3.4725000000000004e-05,
      "loss": 0.5506,
      "step": 6948
    },
    {
      "epoch": 0.38912532198454475,
      "grad_norm": 1.173213243484497,
      "learning_rate": 3.473e-05,
      "loss": 0.4241,
      "step": 6949
    },
    {
      "epoch": 0.38918131929667377,
      "grad_norm": 1.061355471611023,
      "learning_rate": 3.4735e-05,
      "loss": 0.3489,
      "step": 6950
    },
    {
      "epoch": 0.3892373166088028,
      "grad_norm": 1.2567192316055298,
      "learning_rate": 3.474e-05,
      "loss": 0.5298,
      "step": 6951
    },
    {
      "epoch": 0.3892933139209318,
      "grad_norm": 1.2359381914138794,
      "learning_rate": 3.4745e-05,
      "loss": 0.3426,
      "step": 6952
    },
    {
      "epoch": 0.3893493112330608,
      "grad_norm": 1.3463321924209595,
      "learning_rate": 3.475e-05,
      "loss": 0.412,
      "step": 6953
    },
    {
      "epoch": 0.38940530854518984,
      "grad_norm": 1.8644763231277466,
      "learning_rate": 3.4755e-05,
      "loss": 0.5142,
      "step": 6954
    },
    {
      "epoch": 0.38946130585731886,
      "grad_norm": 0.9919539093971252,
      "learning_rate": 3.4760000000000006e-05,
      "loss": 0.4033,
      "step": 6955
    },
    {
      "epoch": 0.3895173031694479,
      "grad_norm": 1.1118013858795166,
      "learning_rate": 3.4765000000000003e-05,
      "loss": 0.4297,
      "step": 6956
    },
    {
      "epoch": 0.3895733004815769,
      "grad_norm": 1.2252516746520996,
      "learning_rate": 3.477e-05,
      "loss": 0.457,
      "step": 6957
    },
    {
      "epoch": 0.3896292977937059,
      "grad_norm": 1.1286903619766235,
      "learning_rate": 3.4775000000000005e-05,
      "loss": 0.3852,
      "step": 6958
    },
    {
      "epoch": 0.38968529510583494,
      "grad_norm": 1.4658342599868774,
      "learning_rate": 3.478e-05,
      "loss": 0.3999,
      "step": 6959
    },
    {
      "epoch": 0.38974129241796396,
      "grad_norm": 1.2423683404922485,
      "learning_rate": 3.4785e-05,
      "loss": 0.51,
      "step": 6960
    },
    {
      "epoch": 0.389797289730093,
      "grad_norm": 1.2950979471206665,
      "learning_rate": 3.479e-05,
      "loss": 0.3804,
      "step": 6961
    },
    {
      "epoch": 0.389853287042222,
      "grad_norm": 1.4664896726608276,
      "learning_rate": 3.4795e-05,
      "loss": 0.3658,
      "step": 6962
    },
    {
      "epoch": 0.389909284354351,
      "grad_norm": 1.451002836227417,
      "learning_rate": 3.48e-05,
      "loss": 0.5165,
      "step": 6963
    },
    {
      "epoch": 0.38996528166648003,
      "grad_norm": 1.3125041723251343,
      "learning_rate": 3.4805e-05,
      "loss": 0.4174,
      "step": 6964
    },
    {
      "epoch": 0.39002127897860905,
      "grad_norm": 1.2604563236236572,
      "learning_rate": 3.481e-05,
      "loss": 0.3917,
      "step": 6965
    },
    {
      "epoch": 0.390077276290738,
      "grad_norm": 1.142025113105774,
      "learning_rate": 3.4815000000000005e-05,
      "loss": 0.3222,
      "step": 6966
    },
    {
      "epoch": 0.39013327360286704,
      "grad_norm": 1.1667081117630005,
      "learning_rate": 3.482e-05,
      "loss": 0.4721,
      "step": 6967
    },
    {
      "epoch": 0.39018927091499606,
      "grad_norm": 1.1558924913406372,
      "learning_rate": 3.4825e-05,
      "loss": 0.3383,
      "step": 6968
    },
    {
      "epoch": 0.3902452682271251,
      "grad_norm": 1.4228202104568481,
      "learning_rate": 3.4830000000000004e-05,
      "loss": 0.4312,
      "step": 6969
    },
    {
      "epoch": 0.3903012655392541,
      "grad_norm": 1.2872172594070435,
      "learning_rate": 3.4835e-05,
      "loss": 0.4468,
      "step": 6970
    },
    {
      "epoch": 0.3903572628513831,
      "grad_norm": 1.1830133199691772,
      "learning_rate": 3.484e-05,
      "loss": 0.44,
      "step": 6971
    },
    {
      "epoch": 0.39041326016351213,
      "grad_norm": 3.1067752838134766,
      "learning_rate": 3.4845e-05,
      "loss": 0.3564,
      "step": 6972
    },
    {
      "epoch": 0.39046925747564115,
      "grad_norm": 1.09810209274292,
      "learning_rate": 3.485e-05,
      "loss": 0.4307,
      "step": 6973
    },
    {
      "epoch": 0.39052525478777017,
      "grad_norm": 1.3250813484191895,
      "learning_rate": 3.4855000000000004e-05,
      "loss": 0.3495,
      "step": 6974
    },
    {
      "epoch": 0.3905812520998992,
      "grad_norm": 1.537216067314148,
      "learning_rate": 3.486e-05,
      "loss": 0.6629,
      "step": 6975
    },
    {
      "epoch": 0.3906372494120282,
      "grad_norm": 1.3258821964263916,
      "learning_rate": 3.4865000000000006e-05,
      "loss": 0.4447,
      "step": 6976
    },
    {
      "epoch": 0.39069324672415723,
      "grad_norm": 1.4364511966705322,
      "learning_rate": 3.487e-05,
      "loss": 0.535,
      "step": 6977
    },
    {
      "epoch": 0.39074924403628625,
      "grad_norm": 1.1638134717941284,
      "learning_rate": 3.4875e-05,
      "loss": 0.4004,
      "step": 6978
    },
    {
      "epoch": 0.39080524134841527,
      "grad_norm": 1.2769513130187988,
      "learning_rate": 3.4880000000000005e-05,
      "loss": 0.6045,
      "step": 6979
    },
    {
      "epoch": 0.3908612386605443,
      "grad_norm": 1.2911280393600464,
      "learning_rate": 3.4885e-05,
      "loss": 0.4701,
      "step": 6980
    },
    {
      "epoch": 0.3909172359726733,
      "grad_norm": 1.2202502489089966,
      "learning_rate": 3.489e-05,
      "loss": 0.4383,
      "step": 6981
    },
    {
      "epoch": 0.3909732332848023,
      "grad_norm": 1.4804140329360962,
      "learning_rate": 3.4895e-05,
      "loss": 0.381,
      "step": 6982
    },
    {
      "epoch": 0.39102923059693134,
      "grad_norm": 1.1945704221725464,
      "learning_rate": 3.49e-05,
      "loss": 0.4915,
      "step": 6983
    },
    {
      "epoch": 0.39108522790906036,
      "grad_norm": 1.3579177856445312,
      "learning_rate": 3.4905000000000005e-05,
      "loss": 0.5499,
      "step": 6984
    },
    {
      "epoch": 0.3911412252211894,
      "grad_norm": 1.2986173629760742,
      "learning_rate": 3.491e-05,
      "loss": 0.4338,
      "step": 6985
    },
    {
      "epoch": 0.3911972225333184,
      "grad_norm": 1.363681674003601,
      "learning_rate": 3.4915e-05,
      "loss": 0.6142,
      "step": 6986
    },
    {
      "epoch": 0.3912532198454474,
      "grad_norm": 1.0938702821731567,
      "learning_rate": 3.4920000000000004e-05,
      "loss": 0.3773,
      "step": 6987
    },
    {
      "epoch": 0.39130921715757644,
      "grad_norm": 1.4995903968811035,
      "learning_rate": 3.4925e-05,
      "loss": 0.5526,
      "step": 6988
    },
    {
      "epoch": 0.39136521446970546,
      "grad_norm": 1.4191012382507324,
      "learning_rate": 3.493e-05,
      "loss": 0.4109,
      "step": 6989
    },
    {
      "epoch": 0.3914212117818345,
      "grad_norm": 1.1995792388916016,
      "learning_rate": 3.4935000000000003e-05,
      "loss": 0.4719,
      "step": 6990
    },
    {
      "epoch": 0.3914772090939635,
      "grad_norm": 1.0581377744674683,
      "learning_rate": 3.494e-05,
      "loss": 0.3769,
      "step": 6991
    },
    {
      "epoch": 0.3915332064060925,
      "grad_norm": 1.1604219675064087,
      "learning_rate": 3.4945e-05,
      "loss": 0.3578,
      "step": 6992
    },
    {
      "epoch": 0.39158920371822153,
      "grad_norm": 1.4558964967727661,
      "learning_rate": 3.495e-05,
      "loss": 0.672,
      "step": 6993
    },
    {
      "epoch": 0.39164520103035055,
      "grad_norm": 1.0141115188598633,
      "learning_rate": 3.495500000000001e-05,
      "loss": 0.363,
      "step": 6994
    },
    {
      "epoch": 0.3917011983424796,
      "grad_norm": 1.1952109336853027,
      "learning_rate": 3.4960000000000004e-05,
      "loss": 0.4501,
      "step": 6995
    },
    {
      "epoch": 0.3917571956546086,
      "grad_norm": 1.6617878675460815,
      "learning_rate": 3.4965e-05,
      "loss": 0.4786,
      "step": 6996
    },
    {
      "epoch": 0.3918131929667376,
      "grad_norm": 1.493242621421814,
      "learning_rate": 3.4970000000000006e-05,
      "loss": 0.4174,
      "step": 6997
    },
    {
      "epoch": 0.39186919027886663,
      "grad_norm": 15.541422843933105,
      "learning_rate": 3.4975e-05,
      "loss": 0.4603,
      "step": 6998
    },
    {
      "epoch": 0.39192518759099565,
      "grad_norm": 1.217512607574463,
      "learning_rate": 3.498e-05,
      "loss": 0.4167,
      "step": 6999
    },
    {
      "epoch": 0.39198118490312467,
      "grad_norm": 1.1900042295455933,
      "learning_rate": 3.4985e-05,
      "loss": 0.453,
      "step": 7000
    },
    {
      "epoch": 0.3920371822152537,
      "grad_norm": 1.1864182949066162,
      "learning_rate": 3.499e-05,
      "loss": 0.3501,
      "step": 7001
    },
    {
      "epoch": 0.3920931795273827,
      "grad_norm": 1.2844698429107666,
      "learning_rate": 3.4995e-05,
      "loss": 0.4631,
      "step": 7002
    },
    {
      "epoch": 0.3921491768395117,
      "grad_norm": 1.172251582145691,
      "learning_rate": 3.5e-05,
      "loss": 0.4115,
      "step": 7003
    },
    {
      "epoch": 0.39220517415164075,
      "grad_norm": 1.2889574766159058,
      "learning_rate": 3.5005e-05,
      "loss": 0.4025,
      "step": 7004
    },
    {
      "epoch": 0.39226117146376976,
      "grad_norm": 1.1393121480941772,
      "learning_rate": 3.5010000000000005e-05,
      "loss": 0.4025,
      "step": 7005
    },
    {
      "epoch": 0.3923171687758988,
      "grad_norm": 1.583275556564331,
      "learning_rate": 3.5015e-05,
      "loss": 0.6657,
      "step": 7006
    },
    {
      "epoch": 0.39237316608802775,
      "grad_norm": 1.3567055463790894,
      "learning_rate": 3.502e-05,
      "loss": 0.5812,
      "step": 7007
    },
    {
      "epoch": 0.39242916340015677,
      "grad_norm": 1.3116973638534546,
      "learning_rate": 3.5025000000000004e-05,
      "loss": 0.527,
      "step": 7008
    },
    {
      "epoch": 0.3924851607122858,
      "grad_norm": 1.4356811046600342,
      "learning_rate": 3.503e-05,
      "loss": 0.4215,
      "step": 7009
    },
    {
      "epoch": 0.3925411580244148,
      "grad_norm": 1.354910969734192,
      "learning_rate": 3.5035e-05,
      "loss": 0.3817,
      "step": 7010
    },
    {
      "epoch": 0.3925971553365438,
      "grad_norm": 1.2192323207855225,
      "learning_rate": 3.504e-05,
      "loss": 0.3654,
      "step": 7011
    },
    {
      "epoch": 0.39265315264867284,
      "grad_norm": 1.248289942741394,
      "learning_rate": 3.5045e-05,
      "loss": 0.4381,
      "step": 7012
    },
    {
      "epoch": 0.39270914996080186,
      "grad_norm": 1.6048190593719482,
      "learning_rate": 3.505e-05,
      "loss": 0.6216,
      "step": 7013
    },
    {
      "epoch": 0.3927651472729309,
      "grad_norm": 1.28898024559021,
      "learning_rate": 3.5055e-05,
      "loss": 0.5206,
      "step": 7014
    },
    {
      "epoch": 0.3928211445850599,
      "grad_norm": 1.372096300125122,
      "learning_rate": 3.5060000000000007e-05,
      "loss": 0.5168,
      "step": 7015
    },
    {
      "epoch": 0.3928771418971889,
      "grad_norm": 1.2836592197418213,
      "learning_rate": 3.5065000000000004e-05,
      "loss": 0.3966,
      "step": 7016
    },
    {
      "epoch": 0.39293313920931794,
      "grad_norm": 1.5730881690979004,
      "learning_rate": 3.507e-05,
      "loss": 0.4068,
      "step": 7017
    },
    {
      "epoch": 0.39298913652144696,
      "grad_norm": 1.4202806949615479,
      "learning_rate": 3.5075000000000006e-05,
      "loss": 0.5073,
      "step": 7018
    },
    {
      "epoch": 0.393045133833576,
      "grad_norm": 1.3695437908172607,
      "learning_rate": 3.508e-05,
      "loss": 0.4575,
      "step": 7019
    },
    {
      "epoch": 0.393101131145705,
      "grad_norm": 1.159608006477356,
      "learning_rate": 3.5085e-05,
      "loss": 0.4209,
      "step": 7020
    },
    {
      "epoch": 0.393157128457834,
      "grad_norm": 1.1785948276519775,
      "learning_rate": 3.509e-05,
      "loss": 0.3444,
      "step": 7021
    },
    {
      "epoch": 0.39321312576996303,
      "grad_norm": 1.4187601804733276,
      "learning_rate": 3.5095e-05,
      "loss": 0.3834,
      "step": 7022
    },
    {
      "epoch": 0.39326912308209205,
      "grad_norm": 1.3515288829803467,
      "learning_rate": 3.51e-05,
      "loss": 0.6947,
      "step": 7023
    },
    {
      "epoch": 0.3933251203942211,
      "grad_norm": 1.3403400182724,
      "learning_rate": 3.5105e-05,
      "loss": 0.4919,
      "step": 7024
    },
    {
      "epoch": 0.3933811177063501,
      "grad_norm": 1.4202855825424194,
      "learning_rate": 3.511e-05,
      "loss": 0.3962,
      "step": 7025
    },
    {
      "epoch": 0.3934371150184791,
      "grad_norm": 1.132371187210083,
      "learning_rate": 3.5115000000000005e-05,
      "loss": 0.3486,
      "step": 7026
    },
    {
      "epoch": 0.39349311233060813,
      "grad_norm": 1.3173556327819824,
      "learning_rate": 3.512e-05,
      "loss": 0.4557,
      "step": 7027
    },
    {
      "epoch": 0.39354910964273715,
      "grad_norm": 1.1719353199005127,
      "learning_rate": 3.5125e-05,
      "loss": 0.3156,
      "step": 7028
    },
    {
      "epoch": 0.39360510695486617,
      "grad_norm": 1.105429768562317,
      "learning_rate": 3.5130000000000004e-05,
      "loss": 0.3959,
      "step": 7029
    },
    {
      "epoch": 0.3936611042669952,
      "grad_norm": 1.223471999168396,
      "learning_rate": 3.5135e-05,
      "loss": 0.4871,
      "step": 7030
    },
    {
      "epoch": 0.3937171015791242,
      "grad_norm": 1.2651889324188232,
      "learning_rate": 3.514e-05,
      "loss": 0.419,
      "step": 7031
    },
    {
      "epoch": 0.3937730988912532,
      "grad_norm": 1.2285356521606445,
      "learning_rate": 3.5145e-05,
      "loss": 0.4938,
      "step": 7032
    },
    {
      "epoch": 0.39382909620338225,
      "grad_norm": 1.3484148979187012,
      "learning_rate": 3.515e-05,
      "loss": 0.5286,
      "step": 7033
    },
    {
      "epoch": 0.39388509351551126,
      "grad_norm": 1.1736124753952026,
      "learning_rate": 3.5155e-05,
      "loss": 0.3927,
      "step": 7034
    },
    {
      "epoch": 0.3939410908276403,
      "grad_norm": 1.4963797330856323,
      "learning_rate": 3.516e-05,
      "loss": 0.5774,
      "step": 7035
    },
    {
      "epoch": 0.3939970881397693,
      "grad_norm": 1.4173800945281982,
      "learning_rate": 3.5165000000000006e-05,
      "loss": 0.6018,
      "step": 7036
    },
    {
      "epoch": 0.3940530854518983,
      "grad_norm": 1.1178412437438965,
      "learning_rate": 3.5170000000000004e-05,
      "loss": 0.5061,
      "step": 7037
    },
    {
      "epoch": 0.39410908276402734,
      "grad_norm": 1.119616985321045,
      "learning_rate": 3.5175e-05,
      "loss": 0.3985,
      "step": 7038
    },
    {
      "epoch": 0.39416508007615636,
      "grad_norm": 1.2967212200164795,
      "learning_rate": 3.518e-05,
      "loss": 0.4044,
      "step": 7039
    },
    {
      "epoch": 0.3942210773882854,
      "grad_norm": 1.1283607482910156,
      "learning_rate": 3.5185e-05,
      "loss": 0.3502,
      "step": 7040
    },
    {
      "epoch": 0.3942770747004144,
      "grad_norm": 1.2243571281433105,
      "learning_rate": 3.519e-05,
      "loss": 0.474,
      "step": 7041
    },
    {
      "epoch": 0.3943330720125434,
      "grad_norm": 1.5098562240600586,
      "learning_rate": 3.5195e-05,
      "loss": 0.4933,
      "step": 7042
    },
    {
      "epoch": 0.39438906932467244,
      "grad_norm": 1.3509429693222046,
      "learning_rate": 3.52e-05,
      "loss": 0.4049,
      "step": 7043
    },
    {
      "epoch": 0.39444506663680146,
      "grad_norm": 1.1621174812316895,
      "learning_rate": 3.5205e-05,
      "loss": 0.374,
      "step": 7044
    },
    {
      "epoch": 0.3945010639489305,
      "grad_norm": 1.142918586730957,
      "learning_rate": 3.5210000000000003e-05,
      "loss": 0.3954,
      "step": 7045
    },
    {
      "epoch": 0.3945570612610595,
      "grad_norm": 1.1373687982559204,
      "learning_rate": 3.5215e-05,
      "loss": 0.325,
      "step": 7046
    },
    {
      "epoch": 0.3946130585731885,
      "grad_norm": 1.0039230585098267,
      "learning_rate": 3.5220000000000005e-05,
      "loss": 0.408,
      "step": 7047
    },
    {
      "epoch": 0.39466905588531753,
      "grad_norm": 1.5211772918701172,
      "learning_rate": 3.5225e-05,
      "loss": 0.5512,
      "step": 7048
    },
    {
      "epoch": 0.3947250531974465,
      "grad_norm": 1.3901233673095703,
      "learning_rate": 3.523e-05,
      "loss": 0.571,
      "step": 7049
    },
    {
      "epoch": 0.3947810505095755,
      "grad_norm": 2.409402370452881,
      "learning_rate": 3.5235000000000004e-05,
      "loss": 0.3791,
      "step": 7050
    },
    {
      "epoch": 0.39483704782170453,
      "grad_norm": 1.1406456232070923,
      "learning_rate": 3.524e-05,
      "loss": 0.4096,
      "step": 7051
    },
    {
      "epoch": 0.39489304513383355,
      "grad_norm": 1.1457037925720215,
      "learning_rate": 3.5245e-05,
      "loss": 0.376,
      "step": 7052
    },
    {
      "epoch": 0.3949490424459626,
      "grad_norm": 1.4743554592132568,
      "learning_rate": 3.525e-05,
      "loss": 0.5204,
      "step": 7053
    },
    {
      "epoch": 0.3950050397580916,
      "grad_norm": 1.1844478845596313,
      "learning_rate": 3.5255e-05,
      "loss": 0.4048,
      "step": 7054
    },
    {
      "epoch": 0.3950610370702206,
      "grad_norm": 1.2880264520645142,
      "learning_rate": 3.5260000000000005e-05,
      "loss": 0.4198,
      "step": 7055
    },
    {
      "epoch": 0.39511703438234963,
      "grad_norm": 1.3818327188491821,
      "learning_rate": 3.5265e-05,
      "loss": 0.5185,
      "step": 7056
    },
    {
      "epoch": 0.39517303169447865,
      "grad_norm": 1.1847329139709473,
      "learning_rate": 3.5270000000000006e-05,
      "loss": 0.5541,
      "step": 7057
    },
    {
      "epoch": 0.39522902900660767,
      "grad_norm": 1.0822230577468872,
      "learning_rate": 3.5275000000000004e-05,
      "loss": 0.3495,
      "step": 7058
    },
    {
      "epoch": 0.3952850263187367,
      "grad_norm": 5.522929668426514,
      "learning_rate": 3.528e-05,
      "loss": 0.3949,
      "step": 7059
    },
    {
      "epoch": 0.3953410236308657,
      "grad_norm": 1.1144723892211914,
      "learning_rate": 3.5285e-05,
      "loss": 0.3869,
      "step": 7060
    },
    {
      "epoch": 0.3953970209429947,
      "grad_norm": 1.2210427522659302,
      "learning_rate": 3.529e-05,
      "loss": 0.3968,
      "step": 7061
    },
    {
      "epoch": 0.39545301825512374,
      "grad_norm": 1.253044605255127,
      "learning_rate": 3.5295e-05,
      "loss": 0.396,
      "step": 7062
    },
    {
      "epoch": 0.39550901556725276,
      "grad_norm": 1.2485785484313965,
      "learning_rate": 3.53e-05,
      "loss": 0.3947,
      "step": 7063
    },
    {
      "epoch": 0.3955650128793818,
      "grad_norm": 1.1033822298049927,
      "learning_rate": 3.5305e-05,
      "loss": 0.3477,
      "step": 7064
    },
    {
      "epoch": 0.3956210101915108,
      "grad_norm": 1.2442518472671509,
      "learning_rate": 3.5310000000000006e-05,
      "loss": 0.5305,
      "step": 7065
    },
    {
      "epoch": 0.3956770075036398,
      "grad_norm": 1.0721766948699951,
      "learning_rate": 3.5315e-05,
      "loss": 0.3685,
      "step": 7066
    },
    {
      "epoch": 0.39573300481576884,
      "grad_norm": 1.1637009382247925,
      "learning_rate": 3.532e-05,
      "loss": 0.5006,
      "step": 7067
    },
    {
      "epoch": 0.39578900212789786,
      "grad_norm": 1.4650561809539795,
      "learning_rate": 3.5325000000000005e-05,
      "loss": 0.4508,
      "step": 7068
    },
    {
      "epoch": 0.3958449994400269,
      "grad_norm": 1.0735374689102173,
      "learning_rate": 3.533e-05,
      "loss": 0.4061,
      "step": 7069
    },
    {
      "epoch": 0.3959009967521559,
      "grad_norm": 1.3330549001693726,
      "learning_rate": 3.5335e-05,
      "loss": 0.4362,
      "step": 7070
    },
    {
      "epoch": 0.3959569940642849,
      "grad_norm": 1.250771403312683,
      "learning_rate": 3.5340000000000004e-05,
      "loss": 0.5554,
      "step": 7071
    },
    {
      "epoch": 0.39601299137641394,
      "grad_norm": 1.3783133029937744,
      "learning_rate": 3.5345e-05,
      "loss": 0.51,
      "step": 7072
    },
    {
      "epoch": 0.39606898868854296,
      "grad_norm": 1.0835919380187988,
      "learning_rate": 3.535e-05,
      "loss": 0.445,
      "step": 7073
    },
    {
      "epoch": 0.396124986000672,
      "grad_norm": 1.3178541660308838,
      "learning_rate": 3.5354999999999996e-05,
      "loss": 0.4205,
      "step": 7074
    },
    {
      "epoch": 0.396180983312801,
      "grad_norm": 1.3751288652420044,
      "learning_rate": 3.536000000000001e-05,
      "loss": 0.4258,
      "step": 7075
    },
    {
      "epoch": 0.39623698062493,
      "grad_norm": 1.0545991659164429,
      "learning_rate": 3.5365000000000004e-05,
      "loss": 0.3546,
      "step": 7076
    },
    {
      "epoch": 0.39629297793705903,
      "grad_norm": 1.3124701976776123,
      "learning_rate": 3.537e-05,
      "loss": 0.3171,
      "step": 7077
    },
    {
      "epoch": 0.39634897524918805,
      "grad_norm": 1.2867190837860107,
      "learning_rate": 3.5375e-05,
      "loss": 0.4597,
      "step": 7078
    },
    {
      "epoch": 0.39640497256131707,
      "grad_norm": 1.1991990804672241,
      "learning_rate": 3.5380000000000003e-05,
      "loss": 0.4953,
      "step": 7079
    },
    {
      "epoch": 0.3964609698734461,
      "grad_norm": 1.238351821899414,
      "learning_rate": 3.5385e-05,
      "loss": 0.4385,
      "step": 7080
    },
    {
      "epoch": 0.3965169671855751,
      "grad_norm": 1.2495702505111694,
      "learning_rate": 3.539e-05,
      "loss": 0.4106,
      "step": 7081
    },
    {
      "epoch": 0.39657296449770413,
      "grad_norm": 3.731426477432251,
      "learning_rate": 3.5395e-05,
      "loss": 0.4307,
      "step": 7082
    },
    {
      "epoch": 0.39662896180983315,
      "grad_norm": 1.1573306322097778,
      "learning_rate": 3.54e-05,
      "loss": 0.3633,
      "step": 7083
    },
    {
      "epoch": 0.39668495912196217,
      "grad_norm": 1.0320273637771606,
      "learning_rate": 3.5405e-05,
      "loss": 0.3824,
      "step": 7084
    },
    {
      "epoch": 0.3967409564340912,
      "grad_norm": 1.2094013690948486,
      "learning_rate": 3.541e-05,
      "loss": 0.3714,
      "step": 7085
    },
    {
      "epoch": 0.3967969537462202,
      "grad_norm": 1.1214096546173096,
      "learning_rate": 3.5415000000000006e-05,
      "loss": 0.4957,
      "step": 7086
    },
    {
      "epoch": 0.3968529510583492,
      "grad_norm": 1.2384463548660278,
      "learning_rate": 3.542e-05,
      "loss": 0.4128,
      "step": 7087
    },
    {
      "epoch": 0.39690894837047824,
      "grad_norm": 1.5187432765960693,
      "learning_rate": 3.5425e-05,
      "loss": 0.5801,
      "step": 7088
    },
    {
      "epoch": 0.39696494568260726,
      "grad_norm": 1.1081868410110474,
      "learning_rate": 3.5430000000000005e-05,
      "loss": 0.4494,
      "step": 7089
    },
    {
      "epoch": 0.3970209429947362,
      "grad_norm": 1.179426670074463,
      "learning_rate": 3.5435e-05,
      "loss": 0.4237,
      "step": 7090
    },
    {
      "epoch": 0.39707694030686524,
      "grad_norm": 1.200624942779541,
      "learning_rate": 3.544e-05,
      "loss": 0.4315,
      "step": 7091
    },
    {
      "epoch": 0.39713293761899426,
      "grad_norm": 1.3659522533416748,
      "learning_rate": 3.5445000000000004e-05,
      "loss": 0.5273,
      "step": 7092
    },
    {
      "epoch": 0.3971889349311233,
      "grad_norm": 1.3462220430374146,
      "learning_rate": 3.545e-05,
      "loss": 0.3893,
      "step": 7093
    },
    {
      "epoch": 0.3972449322432523,
      "grad_norm": 1.2855170965194702,
      "learning_rate": 3.5455e-05,
      "loss": 0.3949,
      "step": 7094
    },
    {
      "epoch": 0.3973009295553813,
      "grad_norm": 1.4168190956115723,
      "learning_rate": 3.546e-05,
      "loss": 0.4308,
      "step": 7095
    },
    {
      "epoch": 0.39735692686751034,
      "grad_norm": 1.4847972393035889,
      "learning_rate": 3.546500000000001e-05,
      "loss": 0.4734,
      "step": 7096
    },
    {
      "epoch": 0.39741292417963936,
      "grad_norm": 1.2195695638656616,
      "learning_rate": 3.5470000000000004e-05,
      "loss": 0.3766,
      "step": 7097
    },
    {
      "epoch": 0.3974689214917684,
      "grad_norm": 1.1849185228347778,
      "learning_rate": 3.5475e-05,
      "loss": 0.3926,
      "step": 7098
    },
    {
      "epoch": 0.3975249188038974,
      "grad_norm": 1.1611347198486328,
      "learning_rate": 3.548e-05,
      "loss": 0.3221,
      "step": 7099
    },
    {
      "epoch": 0.3975809161160264,
      "grad_norm": 1.236920952796936,
      "learning_rate": 3.5485e-05,
      "loss": 0.5203,
      "step": 7100
    },
    {
      "epoch": 0.39763691342815544,
      "grad_norm": 1.2539856433868408,
      "learning_rate": 3.549e-05,
      "loss": 0.4427,
      "step": 7101
    },
    {
      "epoch": 0.39769291074028446,
      "grad_norm": 1.3313424587249756,
      "learning_rate": 3.5495e-05,
      "loss": 0.517,
      "step": 7102
    },
    {
      "epoch": 0.3977489080524135,
      "grad_norm": 1.177718162536621,
      "learning_rate": 3.55e-05,
      "loss": 0.5253,
      "step": 7103
    },
    {
      "epoch": 0.3978049053645425,
      "grad_norm": 1.362850546836853,
      "learning_rate": 3.5505e-05,
      "loss": 0.421,
      "step": 7104
    },
    {
      "epoch": 0.3978609026766715,
      "grad_norm": 1.25775146484375,
      "learning_rate": 3.5510000000000004e-05,
      "loss": 0.5547,
      "step": 7105
    },
    {
      "epoch": 0.39791689998880053,
      "grad_norm": 1.2604835033416748,
      "learning_rate": 3.5515e-05,
      "loss": 0.4058,
      "step": 7106
    },
    {
      "epoch": 0.39797289730092955,
      "grad_norm": 1.0059744119644165,
      "learning_rate": 3.5520000000000006e-05,
      "loss": 0.3841,
      "step": 7107
    },
    {
      "epoch": 0.39802889461305857,
      "grad_norm": 1.1106350421905518,
      "learning_rate": 3.5525e-05,
      "loss": 0.3949,
      "step": 7108
    },
    {
      "epoch": 0.3980848919251876,
      "grad_norm": 1.8010390996932983,
      "learning_rate": 3.553e-05,
      "loss": 0.7235,
      "step": 7109
    },
    {
      "epoch": 0.3981408892373166,
      "grad_norm": 1.3277437686920166,
      "learning_rate": 3.5535000000000005e-05,
      "loss": 0.4169,
      "step": 7110
    },
    {
      "epoch": 0.39819688654944563,
      "grad_norm": 1.2066757678985596,
      "learning_rate": 3.554e-05,
      "loss": 0.4722,
      "step": 7111
    },
    {
      "epoch": 0.39825288386157465,
      "grad_norm": 1.8611897230148315,
      "learning_rate": 3.5545e-05,
      "loss": 0.452,
      "step": 7112
    },
    {
      "epoch": 0.39830888117370367,
      "grad_norm": 1.1620473861694336,
      "learning_rate": 3.555e-05,
      "loss": 0.498,
      "step": 7113
    },
    {
      "epoch": 0.3983648784858327,
      "grad_norm": 1.1954268217086792,
      "learning_rate": 3.5555e-05,
      "loss": 0.4771,
      "step": 7114
    },
    {
      "epoch": 0.3984208757979617,
      "grad_norm": 1.1766629219055176,
      "learning_rate": 3.5560000000000005e-05,
      "loss": 0.3488,
      "step": 7115
    },
    {
      "epoch": 0.3984768731100907,
      "grad_norm": 1.1939324140548706,
      "learning_rate": 3.5565e-05,
      "loss": 0.4501,
      "step": 7116
    },
    {
      "epoch": 0.39853287042221974,
      "grad_norm": 0.9701012372970581,
      "learning_rate": 3.557e-05,
      "loss": 0.389,
      "step": 7117
    },
    {
      "epoch": 0.39858886773434876,
      "grad_norm": 1.6646636724472046,
      "learning_rate": 3.5575000000000004e-05,
      "loss": 0.5372,
      "step": 7118
    },
    {
      "epoch": 0.3986448650464778,
      "grad_norm": 1.4646655321121216,
      "learning_rate": 3.558e-05,
      "loss": 0.3593,
      "step": 7119
    },
    {
      "epoch": 0.3987008623586068,
      "grad_norm": 1.3005801439285278,
      "learning_rate": 3.5585e-05,
      "loss": 0.5581,
      "step": 7120
    },
    {
      "epoch": 0.3987568596707358,
      "grad_norm": 1.3204537630081177,
      "learning_rate": 3.559e-05,
      "loss": 0.3271,
      "step": 7121
    },
    {
      "epoch": 0.39881285698286484,
      "grad_norm": 1.1972370147705078,
      "learning_rate": 3.5595e-05,
      "loss": 0.6395,
      "step": 7122
    },
    {
      "epoch": 0.39886885429499386,
      "grad_norm": 1.33364999294281,
      "learning_rate": 3.56e-05,
      "loss": 0.4278,
      "step": 7123
    },
    {
      "epoch": 0.3989248516071229,
      "grad_norm": 1.370723009109497,
      "learning_rate": 3.5605e-05,
      "loss": 0.4126,
      "step": 7124
    },
    {
      "epoch": 0.3989808489192519,
      "grad_norm": 1.1895666122436523,
      "learning_rate": 3.5610000000000006e-05,
      "loss": 0.584,
      "step": 7125
    },
    {
      "epoch": 0.3990368462313809,
      "grad_norm": 1.1378017663955688,
      "learning_rate": 3.5615000000000004e-05,
      "loss": 0.3344,
      "step": 7126
    },
    {
      "epoch": 0.39909284354350993,
      "grad_norm": 1.3302061557769775,
      "learning_rate": 3.562e-05,
      "loss": 0.3386,
      "step": 7127
    },
    {
      "epoch": 0.39914884085563895,
      "grad_norm": 1.2411571741104126,
      "learning_rate": 3.5625000000000005e-05,
      "loss": 0.4365,
      "step": 7128
    },
    {
      "epoch": 0.399204838167768,
      "grad_norm": 1.3164284229278564,
      "learning_rate": 3.563e-05,
      "loss": 0.4375,
      "step": 7129
    },
    {
      "epoch": 0.399260835479897,
      "grad_norm": 1.1458563804626465,
      "learning_rate": 3.5635e-05,
      "loss": 0.3704,
      "step": 7130
    },
    {
      "epoch": 0.39931683279202596,
      "grad_norm": 1.1261699199676514,
      "learning_rate": 3.5640000000000004e-05,
      "loss": 0.3955,
      "step": 7131
    },
    {
      "epoch": 0.399372830104155,
      "grad_norm": 1.163615345954895,
      "learning_rate": 3.5645e-05,
      "loss": 0.3663,
      "step": 7132
    },
    {
      "epoch": 0.399428827416284,
      "grad_norm": 1.8783775568008423,
      "learning_rate": 3.565e-05,
      "loss": 0.5242,
      "step": 7133
    },
    {
      "epoch": 0.399484824728413,
      "grad_norm": 1.1717891693115234,
      "learning_rate": 3.5654999999999997e-05,
      "loss": 0.4164,
      "step": 7134
    },
    {
      "epoch": 0.39954082204054203,
      "grad_norm": 1.0024054050445557,
      "learning_rate": 3.566e-05,
      "loss": 0.4352,
      "step": 7135
    },
    {
      "epoch": 0.39959681935267105,
      "grad_norm": 1.2847247123718262,
      "learning_rate": 3.5665000000000005e-05,
      "loss": 0.4291,
      "step": 7136
    },
    {
      "epoch": 0.39965281666480007,
      "grad_norm": 1.2100944519042969,
      "learning_rate": 3.567e-05,
      "loss": 0.4851,
      "step": 7137
    },
    {
      "epoch": 0.3997088139769291,
      "grad_norm": 1.32924222946167,
      "learning_rate": 3.5675e-05,
      "loss": 0.5232,
      "step": 7138
    },
    {
      "epoch": 0.3997648112890581,
      "grad_norm": 1.2437349557876587,
      "learning_rate": 3.5680000000000004e-05,
      "loss": 0.4545,
      "step": 7139
    },
    {
      "epoch": 0.3998208086011871,
      "grad_norm": 1.652258038520813,
      "learning_rate": 3.5685e-05,
      "loss": 0.6215,
      "step": 7140
    },
    {
      "epoch": 0.39987680591331615,
      "grad_norm": 2.5210049152374268,
      "learning_rate": 3.569e-05,
      "loss": 0.5923,
      "step": 7141
    },
    {
      "epoch": 0.39993280322544517,
      "grad_norm": 1.4418772459030151,
      "learning_rate": 3.5695e-05,
      "loss": 0.4272,
      "step": 7142
    },
    {
      "epoch": 0.3999888005375742,
      "grad_norm": 1.1753559112548828,
      "learning_rate": 3.57e-05,
      "loss": 0.4517,
      "step": 7143
    },
    {
      "epoch": 0.4000447978497032,
      "grad_norm": 1.2177468538284302,
      "learning_rate": 3.5705e-05,
      "loss": 0.3785,
      "step": 7144
    },
    {
      "epoch": 0.4001007951618322,
      "grad_norm": 1.270458698272705,
      "learning_rate": 3.571e-05,
      "loss": 0.5715,
      "step": 7145
    },
    {
      "epoch": 0.40015679247396124,
      "grad_norm": 1.3269884586334229,
      "learning_rate": 3.5715000000000006e-05,
      "loss": 0.402,
      "step": 7146
    },
    {
      "epoch": 0.40021278978609026,
      "grad_norm": 1.286996603012085,
      "learning_rate": 3.5720000000000004e-05,
      "loss": 0.4246,
      "step": 7147
    },
    {
      "epoch": 0.4002687870982193,
      "grad_norm": 1.4019207954406738,
      "learning_rate": 3.5725e-05,
      "loss": 0.3634,
      "step": 7148
    },
    {
      "epoch": 0.4003247844103483,
      "grad_norm": 1.0722886323928833,
      "learning_rate": 3.5730000000000005e-05,
      "loss": 0.3802,
      "step": 7149
    },
    {
      "epoch": 0.4003807817224773,
      "grad_norm": 1.2400652170181274,
      "learning_rate": 3.5735e-05,
      "loss": 0.3321,
      "step": 7150
    },
    {
      "epoch": 0.40043677903460634,
      "grad_norm": 1.1441702842712402,
      "learning_rate": 3.574e-05,
      "loss": 0.4854,
      "step": 7151
    },
    {
      "epoch": 0.40049277634673536,
      "grad_norm": 1.040071964263916,
      "learning_rate": 3.5745e-05,
      "loss": 0.3512,
      "step": 7152
    },
    {
      "epoch": 0.4005487736588644,
      "grad_norm": 1.3732068538665771,
      "learning_rate": 3.575e-05,
      "loss": 0.3955,
      "step": 7153
    },
    {
      "epoch": 0.4006047709709934,
      "grad_norm": 1.2217940092086792,
      "learning_rate": 3.5755e-05,
      "loss": 0.5224,
      "step": 7154
    },
    {
      "epoch": 0.4006607682831224,
      "grad_norm": 1.4754819869995117,
      "learning_rate": 3.5759999999999996e-05,
      "loss": 0.5832,
      "step": 7155
    },
    {
      "epoch": 0.40071676559525143,
      "grad_norm": 1.1872299909591675,
      "learning_rate": 3.576500000000001e-05,
      "loss": 0.4427,
      "step": 7156
    },
    {
      "epoch": 0.40077276290738045,
      "grad_norm": 1.497488021850586,
      "learning_rate": 3.5770000000000005e-05,
      "loss": 0.4887,
      "step": 7157
    },
    {
      "epoch": 0.4008287602195095,
      "grad_norm": 1.1495366096496582,
      "learning_rate": 3.5775e-05,
      "loss": 0.4121,
      "step": 7158
    },
    {
      "epoch": 0.4008847575316385,
      "grad_norm": 1.237281084060669,
      "learning_rate": 3.578e-05,
      "loss": 0.4101,
      "step": 7159
    },
    {
      "epoch": 0.4009407548437675,
      "grad_norm": 1.3852341175079346,
      "learning_rate": 3.5785000000000004e-05,
      "loss": 0.4432,
      "step": 7160
    },
    {
      "epoch": 0.40099675215589653,
      "grad_norm": 1.3871299028396606,
      "learning_rate": 3.579e-05,
      "loss": 0.4908,
      "step": 7161
    },
    {
      "epoch": 0.40105274946802555,
      "grad_norm": 1.0768440961837769,
      "learning_rate": 3.5795e-05,
      "loss": 0.3331,
      "step": 7162
    },
    {
      "epoch": 0.40110874678015457,
      "grad_norm": 1.0352228879928589,
      "learning_rate": 3.58e-05,
      "loss": 0.4007,
      "step": 7163
    },
    {
      "epoch": 0.4011647440922836,
      "grad_norm": 1.4716922044754028,
      "learning_rate": 3.5805e-05,
      "loss": 0.5157,
      "step": 7164
    },
    {
      "epoch": 0.4012207414044126,
      "grad_norm": 1.0265246629714966,
      "learning_rate": 3.581e-05,
      "loss": 0.3717,
      "step": 7165
    },
    {
      "epoch": 0.4012767387165416,
      "grad_norm": 1.1104140281677246,
      "learning_rate": 3.5815e-05,
      "loss": 0.3902,
      "step": 7166
    },
    {
      "epoch": 0.40133273602867064,
      "grad_norm": 1.0913019180297852,
      "learning_rate": 3.5820000000000006e-05,
      "loss": 0.4035,
      "step": 7167
    },
    {
      "epoch": 0.40138873334079966,
      "grad_norm": 1.4303982257843018,
      "learning_rate": 3.5825000000000003e-05,
      "loss": 0.4168,
      "step": 7168
    },
    {
      "epoch": 0.4014447306529287,
      "grad_norm": 3.429379463195801,
      "learning_rate": 3.583e-05,
      "loss": 0.7091,
      "step": 7169
    },
    {
      "epoch": 0.4015007279650577,
      "grad_norm": 1.3319716453552246,
      "learning_rate": 3.5835000000000005e-05,
      "loss": 0.421,
      "step": 7170
    },
    {
      "epoch": 0.4015567252771867,
      "grad_norm": 1.5306086540222168,
      "learning_rate": 3.584e-05,
      "loss": 0.4384,
      "step": 7171
    },
    {
      "epoch": 0.40161272258931574,
      "grad_norm": 1.2060859203338623,
      "learning_rate": 3.5845e-05,
      "loss": 0.4859,
      "step": 7172
    },
    {
      "epoch": 0.4016687199014447,
      "grad_norm": 1.3723657131195068,
      "learning_rate": 3.585e-05,
      "loss": 0.3922,
      "step": 7173
    },
    {
      "epoch": 0.4017247172135737,
      "grad_norm": 1.451207160949707,
      "learning_rate": 3.5855e-05,
      "loss": 0.468,
      "step": 7174
    },
    {
      "epoch": 0.40178071452570274,
      "grad_norm": 1.2338693141937256,
      "learning_rate": 3.586e-05,
      "loss": 0.5271,
      "step": 7175
    },
    {
      "epoch": 0.40183671183783176,
      "grad_norm": 1.2146055698394775,
      "learning_rate": 3.5865e-05,
      "loss": 0.4202,
      "step": 7176
    },
    {
      "epoch": 0.4018927091499608,
      "grad_norm": 1.243673324584961,
      "learning_rate": 3.587e-05,
      "loss": 0.4702,
      "step": 7177
    },
    {
      "epoch": 0.4019487064620898,
      "grad_norm": 1.3768872022628784,
      "learning_rate": 3.5875000000000005e-05,
      "loss": 0.4021,
      "step": 7178
    },
    {
      "epoch": 0.4020047037742188,
      "grad_norm": 1.3724054098129272,
      "learning_rate": 3.588e-05,
      "loss": 0.4104,
      "step": 7179
    },
    {
      "epoch": 0.40206070108634784,
      "grad_norm": 1.1506850719451904,
      "learning_rate": 3.5885e-05,
      "loss": 0.344,
      "step": 7180
    },
    {
      "epoch": 0.40211669839847686,
      "grad_norm": 1.2319324016571045,
      "learning_rate": 3.5890000000000004e-05,
      "loss": 0.4383,
      "step": 7181
    },
    {
      "epoch": 0.4021726957106059,
      "grad_norm": 1.4333292245864868,
      "learning_rate": 3.5895e-05,
      "loss": 0.5512,
      "step": 7182
    },
    {
      "epoch": 0.4022286930227349,
      "grad_norm": 1.2146166563034058,
      "learning_rate": 3.59e-05,
      "loss": 0.5041,
      "step": 7183
    },
    {
      "epoch": 0.4022846903348639,
      "grad_norm": 1.3595854043960571,
      "learning_rate": 3.5905e-05,
      "loss": 0.4679,
      "step": 7184
    },
    {
      "epoch": 0.40234068764699293,
      "grad_norm": 1.2343111038208008,
      "learning_rate": 3.591e-05,
      "loss": 0.3349,
      "step": 7185
    },
    {
      "epoch": 0.40239668495912195,
      "grad_norm": 1.4902734756469727,
      "learning_rate": 3.5915000000000004e-05,
      "loss": 0.6286,
      "step": 7186
    },
    {
      "epoch": 0.40245268227125097,
      "grad_norm": 1.4743362665176392,
      "learning_rate": 3.592e-05,
      "loss": 0.4208,
      "step": 7187
    },
    {
      "epoch": 0.40250867958338,
      "grad_norm": 1.2193795442581177,
      "learning_rate": 3.5925000000000006e-05,
      "loss": 0.5368,
      "step": 7188
    },
    {
      "epoch": 0.402564676895509,
      "grad_norm": 1.2408791780471802,
      "learning_rate": 3.593e-05,
      "loss": 0.4042,
      "step": 7189
    },
    {
      "epoch": 0.40262067420763803,
      "grad_norm": 1.4323499202728271,
      "learning_rate": 3.5935e-05,
      "loss": 0.4446,
      "step": 7190
    },
    {
      "epoch": 0.40267667151976705,
      "grad_norm": 1.3486984968185425,
      "learning_rate": 3.594e-05,
      "loss": 0.4092,
      "step": 7191
    },
    {
      "epoch": 0.40273266883189607,
      "grad_norm": 1.1780016422271729,
      "learning_rate": 3.5945e-05,
      "loss": 0.5441,
      "step": 7192
    },
    {
      "epoch": 0.4027886661440251,
      "grad_norm": 1.5800676345825195,
      "learning_rate": 3.595e-05,
      "loss": 0.7775,
      "step": 7193
    },
    {
      "epoch": 0.4028446634561541,
      "grad_norm": 1.375052809715271,
      "learning_rate": 3.5955e-05,
      "loss": 0.4215,
      "step": 7194
    },
    {
      "epoch": 0.4029006607682831,
      "grad_norm": 1.192807674407959,
      "learning_rate": 3.596e-05,
      "loss": 0.4601,
      "step": 7195
    },
    {
      "epoch": 0.40295665808041214,
      "grad_norm": 1.272210717201233,
      "learning_rate": 3.5965000000000005e-05,
      "loss": 0.4006,
      "step": 7196
    },
    {
      "epoch": 0.40301265539254116,
      "grad_norm": 1.3375262022018433,
      "learning_rate": 3.597e-05,
      "loss": 0.5038,
      "step": 7197
    },
    {
      "epoch": 0.4030686527046702,
      "grad_norm": 1.3216993808746338,
      "learning_rate": 3.5975e-05,
      "loss": 0.4402,
      "step": 7198
    },
    {
      "epoch": 0.4031246500167992,
      "grad_norm": 1.2181096076965332,
      "learning_rate": 3.5980000000000004e-05,
      "loss": 0.355,
      "step": 7199
    },
    {
      "epoch": 0.4031806473289282,
      "grad_norm": 1.183577060699463,
      "learning_rate": 3.5985e-05,
      "loss": 0.471,
      "step": 7200
    },
    {
      "epoch": 0.40323664464105724,
      "grad_norm": 1.8316634893417358,
      "learning_rate": 3.599e-05,
      "loss": 0.4463,
      "step": 7201
    },
    {
      "epoch": 0.40329264195318626,
      "grad_norm": 1.3434016704559326,
      "learning_rate": 3.5995000000000004e-05,
      "loss": 0.5099,
      "step": 7202
    },
    {
      "epoch": 0.4033486392653153,
      "grad_norm": 1.1872845888137817,
      "learning_rate": 3.6e-05,
      "loss": 0.3647,
      "step": 7203
    },
    {
      "epoch": 0.4034046365774443,
      "grad_norm": 1.775923728942871,
      "learning_rate": 3.6005e-05,
      "loss": 0.4471,
      "step": 7204
    },
    {
      "epoch": 0.4034606338895733,
      "grad_norm": 1.5498169660568237,
      "learning_rate": 3.601e-05,
      "loss": 0.4645,
      "step": 7205
    },
    {
      "epoch": 0.40351663120170234,
      "grad_norm": 1.1243666410446167,
      "learning_rate": 3.601500000000001e-05,
      "loss": 0.4595,
      "step": 7206
    },
    {
      "epoch": 0.40357262851383136,
      "grad_norm": 1.136042833328247,
      "learning_rate": 3.6020000000000004e-05,
      "loss": 0.5057,
      "step": 7207
    },
    {
      "epoch": 0.4036286258259604,
      "grad_norm": 1.2992314100265503,
      "learning_rate": 3.6025e-05,
      "loss": 0.4082,
      "step": 7208
    },
    {
      "epoch": 0.4036846231380894,
      "grad_norm": 1.7175463438034058,
      "learning_rate": 3.6030000000000006e-05,
      "loss": 0.4009,
      "step": 7209
    },
    {
      "epoch": 0.4037406204502184,
      "grad_norm": 1.1723322868347168,
      "learning_rate": 3.6035e-05,
      "loss": 0.3874,
      "step": 7210
    },
    {
      "epoch": 0.40379661776234743,
      "grad_norm": 1.2163728475570679,
      "learning_rate": 3.604e-05,
      "loss": 0.4348,
      "step": 7211
    },
    {
      "epoch": 0.40385261507447645,
      "grad_norm": 0.9793835282325745,
      "learning_rate": 3.6045e-05,
      "loss": 0.3209,
      "step": 7212
    },
    {
      "epoch": 0.40390861238660547,
      "grad_norm": 1.1643357276916504,
      "learning_rate": 3.605e-05,
      "loss": 0.4987,
      "step": 7213
    },
    {
      "epoch": 0.40396460969873443,
      "grad_norm": 1.122285008430481,
      "learning_rate": 3.6055e-05,
      "loss": 0.4352,
      "step": 7214
    },
    {
      "epoch": 0.40402060701086345,
      "grad_norm": 1.0734235048294067,
      "learning_rate": 3.606e-05,
      "loss": 0.5808,
      "step": 7215
    },
    {
      "epoch": 0.40407660432299247,
      "grad_norm": 1.249040961265564,
      "learning_rate": 3.6065e-05,
      "loss": 0.6445,
      "step": 7216
    },
    {
      "epoch": 0.4041326016351215,
      "grad_norm": 1.110620379447937,
      "learning_rate": 3.6070000000000005e-05,
      "loss": 0.3976,
      "step": 7217
    },
    {
      "epoch": 0.4041885989472505,
      "grad_norm": 1.151766300201416,
      "learning_rate": 3.6075e-05,
      "loss": 0.35,
      "step": 7218
    },
    {
      "epoch": 0.40424459625937953,
      "grad_norm": 1.2607028484344482,
      "learning_rate": 3.608e-05,
      "loss": 0.4196,
      "step": 7219
    },
    {
      "epoch": 0.40430059357150855,
      "grad_norm": 1.0967851877212524,
      "learning_rate": 3.6085000000000004e-05,
      "loss": 0.4373,
      "step": 7220
    },
    {
      "epoch": 0.40435659088363757,
      "grad_norm": 1.0691070556640625,
      "learning_rate": 3.609e-05,
      "loss": 0.3334,
      "step": 7221
    },
    {
      "epoch": 0.4044125881957666,
      "grad_norm": 1.171049952507019,
      "learning_rate": 3.6095e-05,
      "loss": 0.5588,
      "step": 7222
    },
    {
      "epoch": 0.4044685855078956,
      "grad_norm": 1.1470245122909546,
      "learning_rate": 3.61e-05,
      "loss": 0.4066,
      "step": 7223
    },
    {
      "epoch": 0.4045245828200246,
      "grad_norm": 1.4739209413528442,
      "learning_rate": 3.6105e-05,
      "loss": 0.5064,
      "step": 7224
    },
    {
      "epoch": 0.40458058013215364,
      "grad_norm": 1.940086007118225,
      "learning_rate": 3.611e-05,
      "loss": 0.4543,
      "step": 7225
    },
    {
      "epoch": 0.40463657744428266,
      "grad_norm": 1.4099403619766235,
      "learning_rate": 3.6115e-05,
      "loss": 0.4356,
      "step": 7226
    },
    {
      "epoch": 0.4046925747564117,
      "grad_norm": 1.4548096656799316,
      "learning_rate": 3.6120000000000007e-05,
      "loss": 0.5859,
      "step": 7227
    },
    {
      "epoch": 0.4047485720685407,
      "grad_norm": 1.3317774534225464,
      "learning_rate": 3.6125000000000004e-05,
      "loss": 0.4717,
      "step": 7228
    },
    {
      "epoch": 0.4048045693806697,
      "grad_norm": 1.229290246963501,
      "learning_rate": 3.613e-05,
      "loss": 0.4747,
      "step": 7229
    },
    {
      "epoch": 0.40486056669279874,
      "grad_norm": 1.174689769744873,
      "learning_rate": 3.6135000000000006e-05,
      "loss": 0.483,
      "step": 7230
    },
    {
      "epoch": 0.40491656400492776,
      "grad_norm": 1.3518184423446655,
      "learning_rate": 3.614e-05,
      "loss": 0.4043,
      "step": 7231
    },
    {
      "epoch": 0.4049725613170568,
      "grad_norm": 1.2385170459747314,
      "learning_rate": 3.6145e-05,
      "loss": 0.4015,
      "step": 7232
    },
    {
      "epoch": 0.4050285586291858,
      "grad_norm": 1.234592080116272,
      "learning_rate": 3.615e-05,
      "loss": 0.4218,
      "step": 7233
    },
    {
      "epoch": 0.4050845559413148,
      "grad_norm": 1.2679672241210938,
      "learning_rate": 3.6155e-05,
      "loss": 0.4764,
      "step": 7234
    },
    {
      "epoch": 0.40514055325344384,
      "grad_norm": 1.4722155332565308,
      "learning_rate": 3.616e-05,
      "loss": 0.4824,
      "step": 7235
    },
    {
      "epoch": 0.40519655056557285,
      "grad_norm": 0.9777321219444275,
      "learning_rate": 3.6165000000000004e-05,
      "loss": 0.3926,
      "step": 7236
    },
    {
      "epoch": 0.4052525478777019,
      "grad_norm": 1.7210427522659302,
      "learning_rate": 3.617e-05,
      "loss": 0.5682,
      "step": 7237
    },
    {
      "epoch": 0.4053085451898309,
      "grad_norm": 1.0800691843032837,
      "learning_rate": 3.6175000000000005e-05,
      "loss": 0.4105,
      "step": 7238
    },
    {
      "epoch": 0.4053645425019599,
      "grad_norm": 1.1063060760498047,
      "learning_rate": 3.618e-05,
      "loss": 0.392,
      "step": 7239
    },
    {
      "epoch": 0.40542053981408893,
      "grad_norm": 1.069801688194275,
      "learning_rate": 3.6185e-05,
      "loss": 0.3287,
      "step": 7240
    },
    {
      "epoch": 0.40547653712621795,
      "grad_norm": 1.3083385229110718,
      "learning_rate": 3.6190000000000004e-05,
      "loss": 0.495,
      "step": 7241
    },
    {
      "epoch": 0.40553253443834697,
      "grad_norm": 1.2802073955535889,
      "learning_rate": 3.6195e-05,
      "loss": 0.4546,
      "step": 7242
    },
    {
      "epoch": 0.405588531750476,
      "grad_norm": 1.2145832777023315,
      "learning_rate": 3.62e-05,
      "loss": 0.4236,
      "step": 7243
    },
    {
      "epoch": 0.405644529062605,
      "grad_norm": 1.1301569938659668,
      "learning_rate": 3.6205e-05,
      "loss": 0.4901,
      "step": 7244
    },
    {
      "epoch": 0.405700526374734,
      "grad_norm": 1.6309202909469604,
      "learning_rate": 3.621e-05,
      "loss": 0.6519,
      "step": 7245
    },
    {
      "epoch": 0.40575652368686305,
      "grad_norm": 1.1026517152786255,
      "learning_rate": 3.6215000000000005e-05,
      "loss": 0.4012,
      "step": 7246
    },
    {
      "epoch": 0.40581252099899207,
      "grad_norm": 1.5104278326034546,
      "learning_rate": 3.622e-05,
      "loss": 0.4446,
      "step": 7247
    },
    {
      "epoch": 0.4058685183111211,
      "grad_norm": 1.4630019664764404,
      "learning_rate": 3.6225000000000006e-05,
      "loss": 0.4647,
      "step": 7248
    },
    {
      "epoch": 0.4059245156232501,
      "grad_norm": 1.742742657661438,
      "learning_rate": 3.6230000000000004e-05,
      "loss": 0.4791,
      "step": 7249
    },
    {
      "epoch": 0.4059805129353791,
      "grad_norm": 1.2679790258407593,
      "learning_rate": 3.6235e-05,
      "loss": 0.4973,
      "step": 7250
    },
    {
      "epoch": 0.40603651024750814,
      "grad_norm": 1.2643063068389893,
      "learning_rate": 3.624e-05,
      "loss": 0.5072,
      "step": 7251
    },
    {
      "epoch": 0.40609250755963716,
      "grad_norm": 1.1733382940292358,
      "learning_rate": 3.6245e-05,
      "loss": 0.3236,
      "step": 7252
    },
    {
      "epoch": 0.4061485048717662,
      "grad_norm": 1.5944677591323853,
      "learning_rate": 3.625e-05,
      "loss": 0.445,
      "step": 7253
    },
    {
      "epoch": 0.4062045021838952,
      "grad_norm": 1.2801021337509155,
      "learning_rate": 3.6255e-05,
      "loss": 0.5094,
      "step": 7254
    },
    {
      "epoch": 0.40626049949602416,
      "grad_norm": 1.1235573291778564,
      "learning_rate": 3.626e-05,
      "loss": 0.5119,
      "step": 7255
    },
    {
      "epoch": 0.4063164968081532,
      "grad_norm": 1.1305443048477173,
      "learning_rate": 3.6265e-05,
      "loss": 0.4362,
      "step": 7256
    },
    {
      "epoch": 0.4063724941202822,
      "grad_norm": 1.908765196800232,
      "learning_rate": 3.6270000000000003e-05,
      "loss": 0.6886,
      "step": 7257
    },
    {
      "epoch": 0.4064284914324112,
      "grad_norm": 1.2548452615737915,
      "learning_rate": 3.6275e-05,
      "loss": 0.401,
      "step": 7258
    },
    {
      "epoch": 0.40648448874454024,
      "grad_norm": 1.400665044784546,
      "learning_rate": 3.6280000000000005e-05,
      "loss": 0.4434,
      "step": 7259
    },
    {
      "epoch": 0.40654048605666926,
      "grad_norm": 1.52109694480896,
      "learning_rate": 3.6285e-05,
      "loss": 0.525,
      "step": 7260
    },
    {
      "epoch": 0.4065964833687983,
      "grad_norm": 1.3359694480895996,
      "learning_rate": 3.629e-05,
      "loss": 0.4099,
      "step": 7261
    },
    {
      "epoch": 0.4066524806809273,
      "grad_norm": 1.321385383605957,
      "learning_rate": 3.6295000000000004e-05,
      "loss": 0.4814,
      "step": 7262
    },
    {
      "epoch": 0.4067084779930563,
      "grad_norm": 1.3202811479568481,
      "learning_rate": 3.63e-05,
      "loss": 0.3902,
      "step": 7263
    },
    {
      "epoch": 0.40676447530518534,
      "grad_norm": 1.1171973943710327,
      "learning_rate": 3.6305e-05,
      "loss": 0.3772,
      "step": 7264
    },
    {
      "epoch": 0.40682047261731435,
      "grad_norm": 1.2634572982788086,
      "learning_rate": 3.6309999999999996e-05,
      "loss": 0.4029,
      "step": 7265
    },
    {
      "epoch": 0.4068764699294434,
      "grad_norm": 1.2522668838500977,
      "learning_rate": 3.6315e-05,
      "loss": 0.3694,
      "step": 7266
    },
    {
      "epoch": 0.4069324672415724,
      "grad_norm": 1.335279107093811,
      "learning_rate": 3.6320000000000005e-05,
      "loss": 0.5055,
      "step": 7267
    },
    {
      "epoch": 0.4069884645537014,
      "grad_norm": 1.239645004272461,
      "learning_rate": 3.6325e-05,
      "loss": 0.3645,
      "step": 7268
    },
    {
      "epoch": 0.40704446186583043,
      "grad_norm": 1.522817611694336,
      "learning_rate": 3.6330000000000006e-05,
      "loss": 0.4993,
      "step": 7269
    },
    {
      "epoch": 0.40710045917795945,
      "grad_norm": 1.3160094022750854,
      "learning_rate": 3.6335000000000004e-05,
      "loss": 0.5277,
      "step": 7270
    },
    {
      "epoch": 0.40715645649008847,
      "grad_norm": 1.2404543161392212,
      "learning_rate": 3.634e-05,
      "loss": 0.5226,
      "step": 7271
    },
    {
      "epoch": 0.4072124538022175,
      "grad_norm": 1.3096650838851929,
      "learning_rate": 3.6345e-05,
      "loss": 0.4599,
      "step": 7272
    },
    {
      "epoch": 0.4072684511143465,
      "grad_norm": 1.3579174280166626,
      "learning_rate": 3.635e-05,
      "loss": 0.5192,
      "step": 7273
    },
    {
      "epoch": 0.4073244484264755,
      "grad_norm": 1.1769434213638306,
      "learning_rate": 3.6355e-05,
      "loss": 0.3975,
      "step": 7274
    },
    {
      "epoch": 0.40738044573860455,
      "grad_norm": 1.0503480434417725,
      "learning_rate": 3.636e-05,
      "loss": 0.3213,
      "step": 7275
    },
    {
      "epoch": 0.40743644305073357,
      "grad_norm": 1.0357081890106201,
      "learning_rate": 3.6365e-05,
      "loss": 0.3951,
      "step": 7276
    },
    {
      "epoch": 0.4074924403628626,
      "grad_norm": 1.1854957342147827,
      "learning_rate": 3.6370000000000006e-05,
      "loss": 0.3392,
      "step": 7277
    },
    {
      "epoch": 0.4075484376749916,
      "grad_norm": 1.3929799795150757,
      "learning_rate": 3.6375e-05,
      "loss": 0.437,
      "step": 7278
    },
    {
      "epoch": 0.4076044349871206,
      "grad_norm": 1.2759002447128296,
      "learning_rate": 3.638e-05,
      "loss": 0.5598,
      "step": 7279
    },
    {
      "epoch": 0.40766043229924964,
      "grad_norm": 1.2573679685592651,
      "learning_rate": 3.6385000000000005e-05,
      "loss": 0.3609,
      "step": 7280
    },
    {
      "epoch": 0.40771642961137866,
      "grad_norm": 1.2415151596069336,
      "learning_rate": 3.639e-05,
      "loss": 0.4047,
      "step": 7281
    },
    {
      "epoch": 0.4077724269235077,
      "grad_norm": 1.0531760454177856,
      "learning_rate": 3.6395e-05,
      "loss": 0.4488,
      "step": 7282
    },
    {
      "epoch": 0.4078284242356367,
      "grad_norm": 1.5336650609970093,
      "learning_rate": 3.6400000000000004e-05,
      "loss": 0.4572,
      "step": 7283
    },
    {
      "epoch": 0.4078844215477657,
      "grad_norm": 1.1323792934417725,
      "learning_rate": 3.6405e-05,
      "loss": 0.4861,
      "step": 7284
    },
    {
      "epoch": 0.40794041885989474,
      "grad_norm": 1.2384064197540283,
      "learning_rate": 3.641e-05,
      "loss": 0.468,
      "step": 7285
    },
    {
      "epoch": 0.40799641617202376,
      "grad_norm": 1.3712652921676636,
      "learning_rate": 3.6414999999999996e-05,
      "loss": 0.3852,
      "step": 7286
    },
    {
      "epoch": 0.4080524134841528,
      "grad_norm": 1.0916869640350342,
      "learning_rate": 3.642000000000001e-05,
      "loss": 0.3724,
      "step": 7287
    },
    {
      "epoch": 0.4081084107962818,
      "grad_norm": 1.1505745649337769,
      "learning_rate": 3.6425000000000004e-05,
      "loss": 0.3085,
      "step": 7288
    },
    {
      "epoch": 0.4081644081084108,
      "grad_norm": 1.4061627388000488,
      "learning_rate": 3.643e-05,
      "loss": 0.4416,
      "step": 7289
    },
    {
      "epoch": 0.40822040542053983,
      "grad_norm": 1.1372989416122437,
      "learning_rate": 3.6435e-05,
      "loss": 0.4334,
      "step": 7290
    },
    {
      "epoch": 0.40827640273266885,
      "grad_norm": 1.311160683631897,
      "learning_rate": 3.6440000000000003e-05,
      "loss": 0.417,
      "step": 7291
    },
    {
      "epoch": 0.40833240004479787,
      "grad_norm": 1.776376485824585,
      "learning_rate": 3.6445e-05,
      "loss": 0.4312,
      "step": 7292
    },
    {
      "epoch": 0.4083883973569269,
      "grad_norm": 1.5318913459777832,
      "learning_rate": 3.645e-05,
      "loss": 0.4828,
      "step": 7293
    },
    {
      "epoch": 0.4084443946690559,
      "grad_norm": 1.1611205339431763,
      "learning_rate": 3.6455e-05,
      "loss": 0.3558,
      "step": 7294
    },
    {
      "epoch": 0.40850039198118493,
      "grad_norm": 1.2419720888137817,
      "learning_rate": 3.646e-05,
      "loss": 0.5373,
      "step": 7295
    },
    {
      "epoch": 0.40855638929331395,
      "grad_norm": 1.2146140336990356,
      "learning_rate": 3.6465e-05,
      "loss": 0.4384,
      "step": 7296
    },
    {
      "epoch": 0.4086123866054429,
      "grad_norm": 1.3515194654464722,
      "learning_rate": 3.647e-05,
      "loss": 0.4263,
      "step": 7297
    },
    {
      "epoch": 0.40866838391757193,
      "grad_norm": 1.2659810781478882,
      "learning_rate": 3.6475000000000006e-05,
      "loss": 0.4124,
      "step": 7298
    },
    {
      "epoch": 0.40872438122970095,
      "grad_norm": 1.2609952688217163,
      "learning_rate": 3.648e-05,
      "loss": 0.4871,
      "step": 7299
    },
    {
      "epoch": 0.40878037854182997,
      "grad_norm": 1.183382272720337,
      "learning_rate": 3.6485e-05,
      "loss": 0.3869,
      "step": 7300
    },
    {
      "epoch": 0.408836375853959,
      "grad_norm": 1.1628130674362183,
      "learning_rate": 3.6490000000000005e-05,
      "loss": 0.4279,
      "step": 7301
    },
    {
      "epoch": 0.408892373166088,
      "grad_norm": 1.176312804222107,
      "learning_rate": 3.6495e-05,
      "loss": 0.3122,
      "step": 7302
    },
    {
      "epoch": 0.408948370478217,
      "grad_norm": 1.3133997917175293,
      "learning_rate": 3.65e-05,
      "loss": 0.5109,
      "step": 7303
    },
    {
      "epoch": 0.40900436779034605,
      "grad_norm": 1.2249528169631958,
      "learning_rate": 3.6505e-05,
      "loss": 0.4877,
      "step": 7304
    },
    {
      "epoch": 0.40906036510247507,
      "grad_norm": 1.3445569276809692,
      "learning_rate": 3.651e-05,
      "loss": 0.5041,
      "step": 7305
    },
    {
      "epoch": 0.4091163624146041,
      "grad_norm": 1.1757526397705078,
      "learning_rate": 3.6515e-05,
      "loss": 0.4316,
      "step": 7306
    },
    {
      "epoch": 0.4091723597267331,
      "grad_norm": 1.3491877317428589,
      "learning_rate": 3.652e-05,
      "loss": 0.493,
      "step": 7307
    },
    {
      "epoch": 0.4092283570388621,
      "grad_norm": 1.0845398902893066,
      "learning_rate": 3.652500000000001e-05,
      "loss": 0.525,
      "step": 7308
    },
    {
      "epoch": 0.40928435435099114,
      "grad_norm": 1.3383963108062744,
      "learning_rate": 3.6530000000000004e-05,
      "loss": 0.575,
      "step": 7309
    },
    {
      "epoch": 0.40934035166312016,
      "grad_norm": 1.0717452764511108,
      "learning_rate": 3.6535e-05,
      "loss": 0.4511,
      "step": 7310
    },
    {
      "epoch": 0.4093963489752492,
      "grad_norm": 1.304875373840332,
      "learning_rate": 3.654e-05,
      "loss": 0.4678,
      "step": 7311
    },
    {
      "epoch": 0.4094523462873782,
      "grad_norm": 1.4814187288284302,
      "learning_rate": 3.6545e-05,
      "loss": 0.477,
      "step": 7312
    },
    {
      "epoch": 0.4095083435995072,
      "grad_norm": 1.102741003036499,
      "learning_rate": 3.655e-05,
      "loss": 0.4875,
      "step": 7313
    },
    {
      "epoch": 0.40956434091163624,
      "grad_norm": 1.2406634092330933,
      "learning_rate": 3.6555e-05,
      "loss": 0.4823,
      "step": 7314
    },
    {
      "epoch": 0.40962033822376526,
      "grad_norm": 1.372307300567627,
      "learning_rate": 3.656e-05,
      "loss": 0.347,
      "step": 7315
    },
    {
      "epoch": 0.4096763355358943,
      "grad_norm": 1.330670714378357,
      "learning_rate": 3.6565e-05,
      "loss": 0.5762,
      "step": 7316
    },
    {
      "epoch": 0.4097323328480233,
      "grad_norm": 1.1185837984085083,
      "learning_rate": 3.6570000000000004e-05,
      "loss": 0.4308,
      "step": 7317
    },
    {
      "epoch": 0.4097883301601523,
      "grad_norm": 1.2783232927322388,
      "learning_rate": 3.6575e-05,
      "loss": 0.4272,
      "step": 7318
    },
    {
      "epoch": 0.40984432747228133,
      "grad_norm": 1.0761449337005615,
      "learning_rate": 3.6580000000000006e-05,
      "loss": 0.3062,
      "step": 7319
    },
    {
      "epoch": 0.40990032478441035,
      "grad_norm": 1.0321117639541626,
      "learning_rate": 3.6585e-05,
      "loss": 0.3292,
      "step": 7320
    },
    {
      "epoch": 0.40995632209653937,
      "grad_norm": 1.1968425512313843,
      "learning_rate": 3.659e-05,
      "loss": 0.4818,
      "step": 7321
    },
    {
      "epoch": 0.4100123194086684,
      "grad_norm": 1.2820180654525757,
      "learning_rate": 3.6595000000000005e-05,
      "loss": 0.4734,
      "step": 7322
    },
    {
      "epoch": 0.4100683167207974,
      "grad_norm": 4.161149024963379,
      "learning_rate": 3.66e-05,
      "loss": 0.5211,
      "step": 7323
    },
    {
      "epoch": 0.41012431403292643,
      "grad_norm": 1.2682455778121948,
      "learning_rate": 3.6605e-05,
      "loss": 0.4286,
      "step": 7324
    },
    {
      "epoch": 0.41018031134505545,
      "grad_norm": 1.7455681562423706,
      "learning_rate": 3.661e-05,
      "loss": 0.6589,
      "step": 7325
    },
    {
      "epoch": 0.41023630865718447,
      "grad_norm": 1.2011220455169678,
      "learning_rate": 3.6615e-05,
      "loss": 0.3598,
      "step": 7326
    },
    {
      "epoch": 0.4102923059693135,
      "grad_norm": 1.5919878482818604,
      "learning_rate": 3.6620000000000005e-05,
      "loss": 0.4307,
      "step": 7327
    },
    {
      "epoch": 0.4103483032814425,
      "grad_norm": 1.0798871517181396,
      "learning_rate": 3.6625e-05,
      "loss": 0.3439,
      "step": 7328
    },
    {
      "epoch": 0.4104043005935715,
      "grad_norm": 1.3776450157165527,
      "learning_rate": 3.663e-05,
      "loss": 0.4705,
      "step": 7329
    },
    {
      "epoch": 0.41046029790570054,
      "grad_norm": 1.2709369659423828,
      "learning_rate": 3.6635000000000004e-05,
      "loss": 0.3923,
      "step": 7330
    },
    {
      "epoch": 0.41051629521782956,
      "grad_norm": 1.1078665256500244,
      "learning_rate": 3.664e-05,
      "loss": 0.4569,
      "step": 7331
    },
    {
      "epoch": 0.4105722925299586,
      "grad_norm": 1.228965401649475,
      "learning_rate": 3.6645e-05,
      "loss": 0.4598,
      "step": 7332
    },
    {
      "epoch": 0.4106282898420876,
      "grad_norm": 1.068197250366211,
      "learning_rate": 3.665e-05,
      "loss": 0.3491,
      "step": 7333
    },
    {
      "epoch": 0.4106842871542166,
      "grad_norm": 1.3216149806976318,
      "learning_rate": 3.6655e-05,
      "loss": 0.6087,
      "step": 7334
    },
    {
      "epoch": 0.41074028446634564,
      "grad_norm": 1.2426646947860718,
      "learning_rate": 3.666e-05,
      "loss": 0.3737,
      "step": 7335
    },
    {
      "epoch": 0.41079628177847466,
      "grad_norm": 1.1281229257583618,
      "learning_rate": 3.6665e-05,
      "loss": 0.4539,
      "step": 7336
    },
    {
      "epoch": 0.4108522790906037,
      "grad_norm": 1.2939091920852661,
      "learning_rate": 3.6670000000000006e-05,
      "loss": 0.4575,
      "step": 7337
    },
    {
      "epoch": 0.41090827640273264,
      "grad_norm": 1.2367515563964844,
      "learning_rate": 3.6675000000000004e-05,
      "loss": 0.4127,
      "step": 7338
    },
    {
      "epoch": 0.41096427371486166,
      "grad_norm": 1.260865569114685,
      "learning_rate": 3.668e-05,
      "loss": 0.6281,
      "step": 7339
    },
    {
      "epoch": 0.4110202710269907,
      "grad_norm": 1.2586859464645386,
      "learning_rate": 3.6685000000000005e-05,
      "loss": 0.4123,
      "step": 7340
    },
    {
      "epoch": 0.4110762683391197,
      "grad_norm": 1.1080645322799683,
      "learning_rate": 3.669e-05,
      "loss": 0.4533,
      "step": 7341
    },
    {
      "epoch": 0.4111322656512487,
      "grad_norm": 1.2502843141555786,
      "learning_rate": 3.6695e-05,
      "loss": 0.4435,
      "step": 7342
    },
    {
      "epoch": 0.41118826296337774,
      "grad_norm": 1.20815908908844,
      "learning_rate": 3.6700000000000004e-05,
      "loss": 0.4348,
      "step": 7343
    },
    {
      "epoch": 0.41124426027550676,
      "grad_norm": 1.056998610496521,
      "learning_rate": 3.6705e-05,
      "loss": 0.3169,
      "step": 7344
    },
    {
      "epoch": 0.4113002575876358,
      "grad_norm": 1.3029078245162964,
      "learning_rate": 3.671e-05,
      "loss": 0.5681,
      "step": 7345
    },
    {
      "epoch": 0.4113562548997648,
      "grad_norm": 1.2896829843521118,
      "learning_rate": 3.6714999999999997e-05,
      "loss": 0.7173,
      "step": 7346
    },
    {
      "epoch": 0.4114122522118938,
      "grad_norm": 1.2883481979370117,
      "learning_rate": 3.672000000000001e-05,
      "loss": 0.4045,
      "step": 7347
    },
    {
      "epoch": 0.41146824952402283,
      "grad_norm": 1.1023024320602417,
      "learning_rate": 3.6725000000000005e-05,
      "loss": 0.3875,
      "step": 7348
    },
    {
      "epoch": 0.41152424683615185,
      "grad_norm": 1.1380733251571655,
      "learning_rate": 3.673e-05,
      "loss": 0.3352,
      "step": 7349
    },
    {
      "epoch": 0.41158024414828087,
      "grad_norm": 1.166361689567566,
      "learning_rate": 3.6735e-05,
      "loss": 0.5477,
      "step": 7350
    },
    {
      "epoch": 0.4116362414604099,
      "grad_norm": 1.4538735151290894,
      "learning_rate": 3.6740000000000004e-05,
      "loss": 0.4344,
      "step": 7351
    },
    {
      "epoch": 0.4116922387725389,
      "grad_norm": 1.8776609897613525,
      "learning_rate": 3.6745e-05,
      "loss": 0.4424,
      "step": 7352
    },
    {
      "epoch": 0.41174823608466793,
      "grad_norm": 1.1175944805145264,
      "learning_rate": 3.675e-05,
      "loss": 0.4074,
      "step": 7353
    },
    {
      "epoch": 0.41180423339679695,
      "grad_norm": 2.575025796890259,
      "learning_rate": 3.6755e-05,
      "loss": 0.4766,
      "step": 7354
    },
    {
      "epoch": 0.41186023070892597,
      "grad_norm": 1.4082902669906616,
      "learning_rate": 3.676e-05,
      "loss": 0.5161,
      "step": 7355
    },
    {
      "epoch": 0.411916228021055,
      "grad_norm": 1.2325143814086914,
      "learning_rate": 3.6765e-05,
      "loss": 0.4787,
      "step": 7356
    },
    {
      "epoch": 0.411972225333184,
      "grad_norm": 1.1196098327636719,
      "learning_rate": 3.677e-05,
      "loss": 0.3825,
      "step": 7357
    },
    {
      "epoch": 0.412028222645313,
      "grad_norm": 1.7802287340164185,
      "learning_rate": 3.6775000000000006e-05,
      "loss": 0.4514,
      "step": 7358
    },
    {
      "epoch": 0.41208421995744204,
      "grad_norm": 1.1851742267608643,
      "learning_rate": 3.6780000000000004e-05,
      "loss": 0.3664,
      "step": 7359
    },
    {
      "epoch": 0.41214021726957106,
      "grad_norm": 1.0434215068817139,
      "learning_rate": 3.6785e-05,
      "loss": 0.3649,
      "step": 7360
    },
    {
      "epoch": 0.4121962145817001,
      "grad_norm": 1.520704746246338,
      "learning_rate": 3.6790000000000005e-05,
      "loss": 0.5598,
      "step": 7361
    },
    {
      "epoch": 0.4122522118938291,
      "grad_norm": 1.5654765367507935,
      "learning_rate": 3.6795e-05,
      "loss": 0.4388,
      "step": 7362
    },
    {
      "epoch": 0.4123082092059581,
      "grad_norm": 1.2206003665924072,
      "learning_rate": 3.68e-05,
      "loss": 0.5292,
      "step": 7363
    },
    {
      "epoch": 0.41236420651808714,
      "grad_norm": 1.1813868284225464,
      "learning_rate": 3.6805e-05,
      "loss": 0.4441,
      "step": 7364
    },
    {
      "epoch": 0.41242020383021616,
      "grad_norm": 1.269207239151001,
      "learning_rate": 3.681e-05,
      "loss": 0.3842,
      "step": 7365
    },
    {
      "epoch": 0.4124762011423452,
      "grad_norm": 1.1340774297714233,
      "learning_rate": 3.6815e-05,
      "loss": 0.3907,
      "step": 7366
    },
    {
      "epoch": 0.4125321984544742,
      "grad_norm": 1.211429476737976,
      "learning_rate": 3.682e-05,
      "loss": 0.4474,
      "step": 7367
    },
    {
      "epoch": 0.4125881957666032,
      "grad_norm": 1.2279529571533203,
      "learning_rate": 3.6825e-05,
      "loss": 0.6188,
      "step": 7368
    },
    {
      "epoch": 0.41264419307873224,
      "grad_norm": 1.2425464391708374,
      "learning_rate": 3.6830000000000005e-05,
      "loss": 0.3226,
      "step": 7369
    },
    {
      "epoch": 0.41270019039086125,
      "grad_norm": 1.0841212272644043,
      "learning_rate": 3.6835e-05,
      "loss": 0.3205,
      "step": 7370
    },
    {
      "epoch": 0.4127561877029903,
      "grad_norm": 1.2456731796264648,
      "learning_rate": 3.684e-05,
      "loss": 0.4568,
      "step": 7371
    },
    {
      "epoch": 0.4128121850151193,
      "grad_norm": 1.2336543798446655,
      "learning_rate": 3.6845000000000004e-05,
      "loss": 0.4919,
      "step": 7372
    },
    {
      "epoch": 0.4128681823272483,
      "grad_norm": 1.2150521278381348,
      "learning_rate": 3.685e-05,
      "loss": 0.3569,
      "step": 7373
    },
    {
      "epoch": 0.41292417963937733,
      "grad_norm": 1.4870706796646118,
      "learning_rate": 3.6855e-05,
      "loss": 0.4903,
      "step": 7374
    },
    {
      "epoch": 0.41298017695150635,
      "grad_norm": 1.4366742372512817,
      "learning_rate": 3.686e-05,
      "loss": 0.5873,
      "step": 7375
    },
    {
      "epoch": 0.41303617426363537,
      "grad_norm": 1.1463624238967896,
      "learning_rate": 3.6865e-05,
      "loss": 0.3585,
      "step": 7376
    },
    {
      "epoch": 0.4130921715757644,
      "grad_norm": 1.1975924968719482,
      "learning_rate": 3.6870000000000004e-05,
      "loss": 0.4505,
      "step": 7377
    },
    {
      "epoch": 0.4131481688878934,
      "grad_norm": 1.291324496269226,
      "learning_rate": 3.6875e-05,
      "loss": 0.3405,
      "step": 7378
    },
    {
      "epoch": 0.41320416620002237,
      "grad_norm": 1.1325887441635132,
      "learning_rate": 3.6880000000000006e-05,
      "loss": 0.4735,
      "step": 7379
    },
    {
      "epoch": 0.4132601635121514,
      "grad_norm": 1.3903597593307495,
      "learning_rate": 3.6885000000000003e-05,
      "loss": 0.4485,
      "step": 7380
    },
    {
      "epoch": 0.4133161608242804,
      "grad_norm": 1.4702013731002808,
      "learning_rate": 3.689e-05,
      "loss": 0.4377,
      "step": 7381
    },
    {
      "epoch": 0.41337215813640943,
      "grad_norm": 1.2567180395126343,
      "learning_rate": 3.6895000000000005e-05,
      "loss": 0.347,
      "step": 7382
    },
    {
      "epoch": 0.41342815544853845,
      "grad_norm": 1.202655553817749,
      "learning_rate": 3.69e-05,
      "loss": 0.3415,
      "step": 7383
    },
    {
      "epoch": 0.41348415276066747,
      "grad_norm": 1.4150012731552124,
      "learning_rate": 3.6905e-05,
      "loss": 0.4606,
      "step": 7384
    },
    {
      "epoch": 0.4135401500727965,
      "grad_norm": 1.4006381034851074,
      "learning_rate": 3.691e-05,
      "loss": 0.6619,
      "step": 7385
    },
    {
      "epoch": 0.4135961473849255,
      "grad_norm": 1.134314775466919,
      "learning_rate": 3.6915e-05,
      "loss": 0.3623,
      "step": 7386
    },
    {
      "epoch": 0.4136521446970545,
      "grad_norm": 1.419512152671814,
      "learning_rate": 3.692e-05,
      "loss": 0.3783,
      "step": 7387
    },
    {
      "epoch": 0.41370814200918354,
      "grad_norm": 1.394559383392334,
      "learning_rate": 3.6925e-05,
      "loss": 0.3843,
      "step": 7388
    },
    {
      "epoch": 0.41376413932131256,
      "grad_norm": 1.3517988920211792,
      "learning_rate": 3.693e-05,
      "loss": 0.5111,
      "step": 7389
    },
    {
      "epoch": 0.4138201366334416,
      "grad_norm": 1.3542639017105103,
      "learning_rate": 3.6935000000000005e-05,
      "loss": 0.4957,
      "step": 7390
    },
    {
      "epoch": 0.4138761339455706,
      "grad_norm": 1.3230736255645752,
      "learning_rate": 3.694e-05,
      "loss": 0.3711,
      "step": 7391
    },
    {
      "epoch": 0.4139321312576996,
      "grad_norm": 1.388952374458313,
      "learning_rate": 3.6945e-05,
      "loss": 0.3725,
      "step": 7392
    },
    {
      "epoch": 0.41398812856982864,
      "grad_norm": 1.247883677482605,
      "learning_rate": 3.6950000000000004e-05,
      "loss": 0.4009,
      "step": 7393
    },
    {
      "epoch": 0.41404412588195766,
      "grad_norm": 1.362732172012329,
      "learning_rate": 3.6955e-05,
      "loss": 0.4289,
      "step": 7394
    },
    {
      "epoch": 0.4141001231940867,
      "grad_norm": 1.1690155267715454,
      "learning_rate": 3.696e-05,
      "loss": 0.4162,
      "step": 7395
    },
    {
      "epoch": 0.4141561205062157,
      "grad_norm": 1.3039852380752563,
      "learning_rate": 3.6965e-05,
      "loss": 0.624,
      "step": 7396
    },
    {
      "epoch": 0.4142121178183447,
      "grad_norm": 1.0486277341842651,
      "learning_rate": 3.697e-05,
      "loss": 0.4451,
      "step": 7397
    },
    {
      "epoch": 0.41426811513047374,
      "grad_norm": 1.4914668798446655,
      "learning_rate": 3.6975000000000004e-05,
      "loss": 0.4872,
      "step": 7398
    },
    {
      "epoch": 0.41432411244260275,
      "grad_norm": 1.2971842288970947,
      "learning_rate": 3.698e-05,
      "loss": 0.4778,
      "step": 7399
    },
    {
      "epoch": 0.4143801097547318,
      "grad_norm": 1.277298927307129,
      "learning_rate": 3.6985000000000006e-05,
      "loss": 0.4799,
      "step": 7400
    },
    {
      "epoch": 0.4144361070668608,
      "grad_norm": 1.1096839904785156,
      "learning_rate": 3.699e-05,
      "loss": 0.3887,
      "step": 7401
    },
    {
      "epoch": 0.4144921043789898,
      "grad_norm": 1.6496092081069946,
      "learning_rate": 3.6995e-05,
      "loss": 0.3864,
      "step": 7402
    },
    {
      "epoch": 0.41454810169111883,
      "grad_norm": 1.2638951539993286,
      "learning_rate": 3.7e-05,
      "loss": 0.4364,
      "step": 7403
    },
    {
      "epoch": 0.41460409900324785,
      "grad_norm": 1.452673077583313,
      "learning_rate": 3.7005e-05,
      "loss": 0.4836,
      "step": 7404
    },
    {
      "epoch": 0.41466009631537687,
      "grad_norm": 1.1603468656539917,
      "learning_rate": 3.701e-05,
      "loss": 0.4391,
      "step": 7405
    },
    {
      "epoch": 0.4147160936275059,
      "grad_norm": 1.2490837574005127,
      "learning_rate": 3.7015e-05,
      "loss": 0.377,
      "step": 7406
    },
    {
      "epoch": 0.4147720909396349,
      "grad_norm": 1.301253080368042,
      "learning_rate": 3.702e-05,
      "loss": 0.4884,
      "step": 7407
    },
    {
      "epoch": 0.4148280882517639,
      "grad_norm": 1.1982135772705078,
      "learning_rate": 3.7025000000000005e-05,
      "loss": 0.4737,
      "step": 7408
    },
    {
      "epoch": 0.41488408556389295,
      "grad_norm": 1.099167823791504,
      "learning_rate": 3.703e-05,
      "loss": 0.3405,
      "step": 7409
    },
    {
      "epoch": 0.41494008287602197,
      "grad_norm": 1.174721121788025,
      "learning_rate": 3.7035e-05,
      "loss": 0.3617,
      "step": 7410
    },
    {
      "epoch": 0.414996080188151,
      "grad_norm": 1.277965784072876,
      "learning_rate": 3.7040000000000005e-05,
      "loss": 0.4473,
      "step": 7411
    },
    {
      "epoch": 0.41505207750028,
      "grad_norm": 1.1440922021865845,
      "learning_rate": 3.7045e-05,
      "loss": 0.3614,
      "step": 7412
    },
    {
      "epoch": 0.415108074812409,
      "grad_norm": 1.3988312482833862,
      "learning_rate": 3.705e-05,
      "loss": 0.5083,
      "step": 7413
    },
    {
      "epoch": 0.41516407212453804,
      "grad_norm": 1.0867133140563965,
      "learning_rate": 3.7055000000000004e-05,
      "loss": 0.3373,
      "step": 7414
    },
    {
      "epoch": 0.41522006943666706,
      "grad_norm": 1.3099242448806763,
      "learning_rate": 3.706e-05,
      "loss": 0.758,
      "step": 7415
    },
    {
      "epoch": 0.4152760667487961,
      "grad_norm": 1.2499079704284668,
      "learning_rate": 3.7065e-05,
      "loss": 0.4199,
      "step": 7416
    },
    {
      "epoch": 0.4153320640609251,
      "grad_norm": 1.0558418035507202,
      "learning_rate": 3.707e-05,
      "loss": 0.392,
      "step": 7417
    },
    {
      "epoch": 0.4153880613730541,
      "grad_norm": 1.1931686401367188,
      "learning_rate": 3.707500000000001e-05,
      "loss": 0.3167,
      "step": 7418
    },
    {
      "epoch": 0.41544405868518314,
      "grad_norm": 1.4347862005233765,
      "learning_rate": 3.7080000000000004e-05,
      "loss": 0.4799,
      "step": 7419
    },
    {
      "epoch": 0.41550005599731216,
      "grad_norm": 1.5621248483657837,
      "learning_rate": 3.7085e-05,
      "loss": 0.4901,
      "step": 7420
    },
    {
      "epoch": 0.4155560533094411,
      "grad_norm": 1.4398621320724487,
      "learning_rate": 3.7090000000000006e-05,
      "loss": 0.4738,
      "step": 7421
    },
    {
      "epoch": 0.41561205062157014,
      "grad_norm": 2.3422293663024902,
      "learning_rate": 3.7095e-05,
      "loss": 0.5476,
      "step": 7422
    },
    {
      "epoch": 0.41566804793369916,
      "grad_norm": 1.1191027164459229,
      "learning_rate": 3.71e-05,
      "loss": 0.3922,
      "step": 7423
    },
    {
      "epoch": 0.4157240452458282,
      "grad_norm": 1.2718764543533325,
      "learning_rate": 3.7105e-05,
      "loss": 0.4628,
      "step": 7424
    },
    {
      "epoch": 0.4157800425579572,
      "grad_norm": 1.2488576173782349,
      "learning_rate": 3.711e-05,
      "loss": 0.3959,
      "step": 7425
    },
    {
      "epoch": 0.4158360398700862,
      "grad_norm": 1.1072102785110474,
      "learning_rate": 3.7115e-05,
      "loss": 0.4925,
      "step": 7426
    },
    {
      "epoch": 0.41589203718221524,
      "grad_norm": 1.272878885269165,
      "learning_rate": 3.712e-05,
      "loss": 0.3726,
      "step": 7427
    },
    {
      "epoch": 0.41594803449434425,
      "grad_norm": 1.1713122129440308,
      "learning_rate": 3.7125e-05,
      "loss": 0.3643,
      "step": 7428
    },
    {
      "epoch": 0.4160040318064733,
      "grad_norm": 1.1248369216918945,
      "learning_rate": 3.7130000000000005e-05,
      "loss": 0.2857,
      "step": 7429
    },
    {
      "epoch": 0.4160600291186023,
      "grad_norm": 1.3147884607315063,
      "learning_rate": 3.7135e-05,
      "loss": 0.4309,
      "step": 7430
    },
    {
      "epoch": 0.4161160264307313,
      "grad_norm": 1.1655421257019043,
      "learning_rate": 3.714e-05,
      "loss": 0.454,
      "step": 7431
    },
    {
      "epoch": 0.41617202374286033,
      "grad_norm": 1.1496081352233887,
      "learning_rate": 3.7145000000000004e-05,
      "loss": 0.3775,
      "step": 7432
    },
    {
      "epoch": 0.41622802105498935,
      "grad_norm": 1.1635730266571045,
      "learning_rate": 3.715e-05,
      "loss": 0.3961,
      "step": 7433
    },
    {
      "epoch": 0.41628401836711837,
      "grad_norm": 1.2568323612213135,
      "learning_rate": 3.7155e-05,
      "loss": 0.5334,
      "step": 7434
    },
    {
      "epoch": 0.4163400156792474,
      "grad_norm": 1.2780178785324097,
      "learning_rate": 3.716e-05,
      "loss": 0.4151,
      "step": 7435
    },
    {
      "epoch": 0.4163960129913764,
      "grad_norm": 1.2436951398849487,
      "learning_rate": 3.7165e-05,
      "loss": 0.3052,
      "step": 7436
    },
    {
      "epoch": 0.4164520103035054,
      "grad_norm": 1.131870150566101,
      "learning_rate": 3.717e-05,
      "loss": 0.3319,
      "step": 7437
    },
    {
      "epoch": 0.41650800761563445,
      "grad_norm": 1.9683313369750977,
      "learning_rate": 3.7175e-05,
      "loss": 0.4952,
      "step": 7438
    },
    {
      "epoch": 0.41656400492776346,
      "grad_norm": 1.0045701265335083,
      "learning_rate": 3.7180000000000007e-05,
      "loss": 0.3867,
      "step": 7439
    },
    {
      "epoch": 0.4166200022398925,
      "grad_norm": 1.1301097869873047,
      "learning_rate": 3.7185000000000004e-05,
      "loss": 0.4603,
      "step": 7440
    },
    {
      "epoch": 0.4166759995520215,
      "grad_norm": 1.4950412511825562,
      "learning_rate": 3.719e-05,
      "loss": 0.7327,
      "step": 7441
    },
    {
      "epoch": 0.4167319968641505,
      "grad_norm": 1.2363497018814087,
      "learning_rate": 3.7195e-05,
      "loss": 0.4815,
      "step": 7442
    },
    {
      "epoch": 0.41678799417627954,
      "grad_norm": 1.0679094791412354,
      "learning_rate": 3.72e-05,
      "loss": 0.3766,
      "step": 7443
    },
    {
      "epoch": 0.41684399148840856,
      "grad_norm": 1.2219816446304321,
      "learning_rate": 3.7205e-05,
      "loss": 0.4915,
      "step": 7444
    },
    {
      "epoch": 0.4168999888005376,
      "grad_norm": 1.3547922372817993,
      "learning_rate": 3.721e-05,
      "loss": 0.4618,
      "step": 7445
    },
    {
      "epoch": 0.4169559861126666,
      "grad_norm": 1.3519762754440308,
      "learning_rate": 3.7215e-05,
      "loss": 0.5428,
      "step": 7446
    },
    {
      "epoch": 0.4170119834247956,
      "grad_norm": 1.3374686241149902,
      "learning_rate": 3.722e-05,
      "loss": 0.4427,
      "step": 7447
    },
    {
      "epoch": 0.41706798073692464,
      "grad_norm": 1.1669195890426636,
      "learning_rate": 3.7225000000000004e-05,
      "loss": 0.4255,
      "step": 7448
    },
    {
      "epoch": 0.41712397804905366,
      "grad_norm": 1.2421183586120605,
      "learning_rate": 3.723e-05,
      "loss": 0.3645,
      "step": 7449
    },
    {
      "epoch": 0.4171799753611827,
      "grad_norm": 1.3754910230636597,
      "learning_rate": 3.7235000000000005e-05,
      "loss": 0.5434,
      "step": 7450
    },
    {
      "epoch": 0.4172359726733117,
      "grad_norm": 1.1193515062332153,
      "learning_rate": 3.724e-05,
      "loss": 0.4259,
      "step": 7451
    },
    {
      "epoch": 0.4172919699854407,
      "grad_norm": 1.3255189657211304,
      "learning_rate": 3.7245e-05,
      "loss": 0.4716,
      "step": 7452
    },
    {
      "epoch": 0.41734796729756973,
      "grad_norm": 1.2888439893722534,
      "learning_rate": 3.7250000000000004e-05,
      "loss": 0.4525,
      "step": 7453
    },
    {
      "epoch": 0.41740396460969875,
      "grad_norm": 1.7321168184280396,
      "learning_rate": 3.7255e-05,
      "loss": 0.4392,
      "step": 7454
    },
    {
      "epoch": 0.41745996192182777,
      "grad_norm": 1.167161226272583,
      "learning_rate": 3.726e-05,
      "loss": 0.3948,
      "step": 7455
    },
    {
      "epoch": 0.4175159592339568,
      "grad_norm": 1.6288481950759888,
      "learning_rate": 3.7265e-05,
      "loss": 0.4082,
      "step": 7456
    },
    {
      "epoch": 0.4175719565460858,
      "grad_norm": 1.244638442993164,
      "learning_rate": 3.727e-05,
      "loss": 0.4842,
      "step": 7457
    },
    {
      "epoch": 0.41762795385821483,
      "grad_norm": 1.3217008113861084,
      "learning_rate": 3.7275000000000005e-05,
      "loss": 0.4173,
      "step": 7458
    },
    {
      "epoch": 0.41768395117034385,
      "grad_norm": 1.2548723220825195,
      "learning_rate": 3.728e-05,
      "loss": 0.4547,
      "step": 7459
    },
    {
      "epoch": 0.41773994848247287,
      "grad_norm": 1.2116191387176514,
      "learning_rate": 3.7285000000000006e-05,
      "loss": 0.337,
      "step": 7460
    },
    {
      "epoch": 0.4177959457946019,
      "grad_norm": 1.359483242034912,
      "learning_rate": 3.7290000000000004e-05,
      "loss": 0.4193,
      "step": 7461
    },
    {
      "epoch": 0.41785194310673085,
      "grad_norm": 1.196447491645813,
      "learning_rate": 3.7295e-05,
      "loss": 0.48,
      "step": 7462
    },
    {
      "epoch": 0.41790794041885987,
      "grad_norm": 1.2658195495605469,
      "learning_rate": 3.73e-05,
      "loss": 0.5135,
      "step": 7463
    },
    {
      "epoch": 0.4179639377309889,
      "grad_norm": 1.276366949081421,
      "learning_rate": 3.7305e-05,
      "loss": 0.4656,
      "step": 7464
    },
    {
      "epoch": 0.4180199350431179,
      "grad_norm": 1.5904545783996582,
      "learning_rate": 3.731e-05,
      "loss": 0.4994,
      "step": 7465
    },
    {
      "epoch": 0.4180759323552469,
      "grad_norm": 1.1834349632263184,
      "learning_rate": 3.7315e-05,
      "loss": 0.3434,
      "step": 7466
    },
    {
      "epoch": 0.41813192966737595,
      "grad_norm": 1.1932505369186401,
      "learning_rate": 3.732e-05,
      "loss": 0.4881,
      "step": 7467
    },
    {
      "epoch": 0.41818792697950496,
      "grad_norm": 1.2890348434448242,
      "learning_rate": 3.7325000000000006e-05,
      "loss": 0.4165,
      "step": 7468
    },
    {
      "epoch": 0.418243924291634,
      "grad_norm": 1.2739930152893066,
      "learning_rate": 3.7330000000000003e-05,
      "loss": 0.4243,
      "step": 7469
    },
    {
      "epoch": 0.418299921603763,
      "grad_norm": 1.0256266593933105,
      "learning_rate": 3.7335e-05,
      "loss": 0.3585,
      "step": 7470
    },
    {
      "epoch": 0.418355918915892,
      "grad_norm": 1.309674859046936,
      "learning_rate": 3.7340000000000005e-05,
      "loss": 0.4201,
      "step": 7471
    },
    {
      "epoch": 0.41841191622802104,
      "grad_norm": 1.2147879600524902,
      "learning_rate": 3.7345e-05,
      "loss": 0.3943,
      "step": 7472
    },
    {
      "epoch": 0.41846791354015006,
      "grad_norm": 1.3038194179534912,
      "learning_rate": 3.735e-05,
      "loss": 0.3817,
      "step": 7473
    },
    {
      "epoch": 0.4185239108522791,
      "grad_norm": 1.4162400960922241,
      "learning_rate": 3.7355000000000004e-05,
      "loss": 0.462,
      "step": 7474
    },
    {
      "epoch": 0.4185799081644081,
      "grad_norm": 1.294368028640747,
      "learning_rate": 3.736e-05,
      "loss": 0.4256,
      "step": 7475
    },
    {
      "epoch": 0.4186359054765371,
      "grad_norm": 1.3236455917358398,
      "learning_rate": 3.7365e-05,
      "loss": 0.3296,
      "step": 7476
    },
    {
      "epoch": 0.41869190278866614,
      "grad_norm": 1.3321163654327393,
      "learning_rate": 3.7369999999999996e-05,
      "loss": 0.4964,
      "step": 7477
    },
    {
      "epoch": 0.41874790010079516,
      "grad_norm": 1.361327886581421,
      "learning_rate": 3.737500000000001e-05,
      "loss": 0.4242,
      "step": 7478
    },
    {
      "epoch": 0.4188038974129242,
      "grad_norm": 1.1381531953811646,
      "learning_rate": 3.7380000000000005e-05,
      "loss": 0.3938,
      "step": 7479
    },
    {
      "epoch": 0.4188598947250532,
      "grad_norm": 1.175363302230835,
      "learning_rate": 3.7385e-05,
      "loss": 0.3535,
      "step": 7480
    },
    {
      "epoch": 0.4189158920371822,
      "grad_norm": 1.2960225343704224,
      "learning_rate": 3.739e-05,
      "loss": 0.4894,
      "step": 7481
    },
    {
      "epoch": 0.41897188934931123,
      "grad_norm": 1.4263495206832886,
      "learning_rate": 3.7395000000000004e-05,
      "loss": 0.5933,
      "step": 7482
    },
    {
      "epoch": 0.41902788666144025,
      "grad_norm": 1.2767161130905151,
      "learning_rate": 3.74e-05,
      "loss": 0.4764,
      "step": 7483
    },
    {
      "epoch": 0.41908388397356927,
      "grad_norm": 1.178892970085144,
      "learning_rate": 3.7405e-05,
      "loss": 0.4257,
      "step": 7484
    },
    {
      "epoch": 0.4191398812856983,
      "grad_norm": 1.3218621015548706,
      "learning_rate": 3.741e-05,
      "loss": 0.4084,
      "step": 7485
    },
    {
      "epoch": 0.4191958785978273,
      "grad_norm": 1.385350227355957,
      "learning_rate": 3.7415e-05,
      "loss": 0.4606,
      "step": 7486
    },
    {
      "epoch": 0.41925187590995633,
      "grad_norm": 1.311037302017212,
      "learning_rate": 3.742e-05,
      "loss": 0.4947,
      "step": 7487
    },
    {
      "epoch": 0.41930787322208535,
      "grad_norm": 1.311902403831482,
      "learning_rate": 3.7425e-05,
      "loss": 0.4066,
      "step": 7488
    },
    {
      "epoch": 0.41936387053421437,
      "grad_norm": 1.1940802335739136,
      "learning_rate": 3.7430000000000006e-05,
      "loss": 0.3556,
      "step": 7489
    },
    {
      "epoch": 0.4194198678463434,
      "grad_norm": 1.77219557762146,
      "learning_rate": 3.7435e-05,
      "loss": 0.4119,
      "step": 7490
    },
    {
      "epoch": 0.4194758651584724,
      "grad_norm": 1.2206567525863647,
      "learning_rate": 3.744e-05,
      "loss": 0.4377,
      "step": 7491
    },
    {
      "epoch": 0.4195318624706014,
      "grad_norm": 1.3089338541030884,
      "learning_rate": 3.7445000000000005e-05,
      "loss": 0.5447,
      "step": 7492
    },
    {
      "epoch": 0.41958785978273044,
      "grad_norm": 1.2232393026351929,
      "learning_rate": 3.745e-05,
      "loss": 0.3733,
      "step": 7493
    },
    {
      "epoch": 0.41964385709485946,
      "grad_norm": 0.9914332032203674,
      "learning_rate": 3.7455e-05,
      "loss": 0.4314,
      "step": 7494
    },
    {
      "epoch": 0.4196998544069885,
      "grad_norm": 1.6882667541503906,
      "learning_rate": 3.7460000000000004e-05,
      "loss": 0.5703,
      "step": 7495
    },
    {
      "epoch": 0.4197558517191175,
      "grad_norm": 1.0649563074111938,
      "learning_rate": 3.7465e-05,
      "loss": 0.4024,
      "step": 7496
    },
    {
      "epoch": 0.4198118490312465,
      "grad_norm": 1.0853193998336792,
      "learning_rate": 3.747e-05,
      "loss": 0.3227,
      "step": 7497
    },
    {
      "epoch": 0.41986784634337554,
      "grad_norm": 1.2998113632202148,
      "learning_rate": 3.7475e-05,
      "loss": 0.4718,
      "step": 7498
    },
    {
      "epoch": 0.41992384365550456,
      "grad_norm": 1.1960525512695312,
      "learning_rate": 3.748000000000001e-05,
      "loss": 0.4201,
      "step": 7499
    },
    {
      "epoch": 0.4199798409676336,
      "grad_norm": 1.2740609645843506,
      "learning_rate": 3.7485000000000004e-05,
      "loss": 0.459,
      "step": 7500
    },
    {
      "epoch": 0.4200358382797626,
      "grad_norm": 1.428773045539856,
      "learning_rate": 3.749e-05,
      "loss": 0.5769,
      "step": 7501
    },
    {
      "epoch": 0.4200918355918916,
      "grad_norm": 1.283211350440979,
      "learning_rate": 3.7495e-05,
      "loss": 0.3808,
      "step": 7502
    },
    {
      "epoch": 0.4201478329040206,
      "grad_norm": 1.177197813987732,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.4485,
      "step": 7503
    },
    {
      "epoch": 0.4202038302161496,
      "grad_norm": 1.2770239114761353,
      "learning_rate": 3.7505e-05,
      "loss": 0.4191,
      "step": 7504
    },
    {
      "epoch": 0.4202598275282786,
      "grad_norm": 1.404529333114624,
      "learning_rate": 3.751e-05,
      "loss": 0.4719,
      "step": 7505
    },
    {
      "epoch": 0.42031582484040764,
      "grad_norm": 1.4240243434906006,
      "learning_rate": 3.7515e-05,
      "loss": 0.7294,
      "step": 7506
    },
    {
      "epoch": 0.42037182215253666,
      "grad_norm": 1.0826585292816162,
      "learning_rate": 3.752e-05,
      "loss": 0.339,
      "step": 7507
    },
    {
      "epoch": 0.4204278194646657,
      "grad_norm": 1.2525144815444946,
      "learning_rate": 3.7525e-05,
      "loss": 0.4028,
      "step": 7508
    },
    {
      "epoch": 0.4204838167767947,
      "grad_norm": 1.1017965078353882,
      "learning_rate": 3.753e-05,
      "loss": 0.4051,
      "step": 7509
    },
    {
      "epoch": 0.4205398140889237,
      "grad_norm": 1.0927131175994873,
      "learning_rate": 3.7535000000000006e-05,
      "loss": 0.3306,
      "step": 7510
    },
    {
      "epoch": 0.42059581140105273,
      "grad_norm": 1.099252462387085,
      "learning_rate": 3.754e-05,
      "loss": 0.3469,
      "step": 7511
    },
    {
      "epoch": 0.42065180871318175,
      "grad_norm": 1.1844866275787354,
      "learning_rate": 3.7545e-05,
      "loss": 0.3762,
      "step": 7512
    },
    {
      "epoch": 0.42070780602531077,
      "grad_norm": 1.3952338695526123,
      "learning_rate": 3.7550000000000005e-05,
      "loss": 0.4501,
      "step": 7513
    },
    {
      "epoch": 0.4207638033374398,
      "grad_norm": 1.1330647468566895,
      "learning_rate": 3.7555e-05,
      "loss": 0.4405,
      "step": 7514
    },
    {
      "epoch": 0.4208198006495688,
      "grad_norm": 1.39003324508667,
      "learning_rate": 3.756e-05,
      "loss": 0.5793,
      "step": 7515
    },
    {
      "epoch": 0.42087579796169783,
      "grad_norm": 1.1815036535263062,
      "learning_rate": 3.7565e-05,
      "loss": 0.5452,
      "step": 7516
    },
    {
      "epoch": 0.42093179527382685,
      "grad_norm": 1.0811365842819214,
      "learning_rate": 3.757e-05,
      "loss": 0.396,
      "step": 7517
    },
    {
      "epoch": 0.42098779258595587,
      "grad_norm": 1.2157245874404907,
      "learning_rate": 3.7575e-05,
      "loss": 0.3333,
      "step": 7518
    },
    {
      "epoch": 0.4210437898980849,
      "grad_norm": 1.5133262872695923,
      "learning_rate": 3.758e-05,
      "loss": 0.5342,
      "step": 7519
    },
    {
      "epoch": 0.4210997872102139,
      "grad_norm": 1.161711573600769,
      "learning_rate": 3.758500000000001e-05,
      "loss": 0.373,
      "step": 7520
    },
    {
      "epoch": 0.4211557845223429,
      "grad_norm": 1.3511238098144531,
      "learning_rate": 3.7590000000000004e-05,
      "loss": 0.3456,
      "step": 7521
    },
    {
      "epoch": 0.42121178183447194,
      "grad_norm": 1.7794889211654663,
      "learning_rate": 3.7595e-05,
      "loss": 0.503,
      "step": 7522
    },
    {
      "epoch": 0.42126777914660096,
      "grad_norm": 1.5192196369171143,
      "learning_rate": 3.76e-05,
      "loss": 0.4161,
      "step": 7523
    },
    {
      "epoch": 0.42132377645873,
      "grad_norm": 1.1321442127227783,
      "learning_rate": 3.7605e-05,
      "loss": 0.3586,
      "step": 7524
    },
    {
      "epoch": 0.421379773770859,
      "grad_norm": 1.3278142213821411,
      "learning_rate": 3.761e-05,
      "loss": 0.4832,
      "step": 7525
    },
    {
      "epoch": 0.421435771082988,
      "grad_norm": 1.2671815156936646,
      "learning_rate": 3.7615e-05,
      "loss": 0.4192,
      "step": 7526
    },
    {
      "epoch": 0.42149176839511704,
      "grad_norm": 1.416520595550537,
      "learning_rate": 3.762e-05,
      "loss": 0.4862,
      "step": 7527
    },
    {
      "epoch": 0.42154776570724606,
      "grad_norm": 1.217038631439209,
      "learning_rate": 3.7625e-05,
      "loss": 0.4498,
      "step": 7528
    },
    {
      "epoch": 0.4216037630193751,
      "grad_norm": 2.6948342323303223,
      "learning_rate": 3.7630000000000004e-05,
      "loss": 0.6703,
      "step": 7529
    },
    {
      "epoch": 0.4216597603315041,
      "grad_norm": 1.1527397632598877,
      "learning_rate": 3.7635e-05,
      "loss": 0.4959,
      "step": 7530
    },
    {
      "epoch": 0.4217157576436331,
      "grad_norm": 1.5450860261917114,
      "learning_rate": 3.7640000000000006e-05,
      "loss": 0.4654,
      "step": 7531
    },
    {
      "epoch": 0.42177175495576213,
      "grad_norm": 1.394510269165039,
      "learning_rate": 3.7645e-05,
      "loss": 0.5322,
      "step": 7532
    },
    {
      "epoch": 0.42182775226789115,
      "grad_norm": 1.3331291675567627,
      "learning_rate": 3.765e-05,
      "loss": 0.4591,
      "step": 7533
    },
    {
      "epoch": 0.4218837495800202,
      "grad_norm": 1.3717551231384277,
      "learning_rate": 3.7655000000000005e-05,
      "loss": 0.5099,
      "step": 7534
    },
    {
      "epoch": 0.4219397468921492,
      "grad_norm": 1.1739580631256104,
      "learning_rate": 3.766e-05,
      "loss": 0.3854,
      "step": 7535
    },
    {
      "epoch": 0.4219957442042782,
      "grad_norm": 1.1824437379837036,
      "learning_rate": 3.7665e-05,
      "loss": 0.4561,
      "step": 7536
    },
    {
      "epoch": 0.42205174151640723,
      "grad_norm": 1.332931637763977,
      "learning_rate": 3.767e-05,
      "loss": 0.3535,
      "step": 7537
    },
    {
      "epoch": 0.42210773882853625,
      "grad_norm": 1.2487376928329468,
      "learning_rate": 3.7675e-05,
      "loss": 0.4024,
      "step": 7538
    },
    {
      "epoch": 0.42216373614066527,
      "grad_norm": 1.5085293054580688,
      "learning_rate": 3.7680000000000005e-05,
      "loss": 0.3954,
      "step": 7539
    },
    {
      "epoch": 0.4222197334527943,
      "grad_norm": 1.2047690153121948,
      "learning_rate": 3.7685e-05,
      "loss": 0.517,
      "step": 7540
    },
    {
      "epoch": 0.4222757307649233,
      "grad_norm": 1.1977018117904663,
      "learning_rate": 3.769e-05,
      "loss": 0.3594,
      "step": 7541
    },
    {
      "epoch": 0.4223317280770523,
      "grad_norm": 1.2525365352630615,
      "learning_rate": 3.7695000000000004e-05,
      "loss": 0.5402,
      "step": 7542
    },
    {
      "epoch": 0.42238772538918135,
      "grad_norm": 4.599066734313965,
      "learning_rate": 3.77e-05,
      "loss": 0.3296,
      "step": 7543
    },
    {
      "epoch": 0.42244372270131036,
      "grad_norm": 1.1863445043563843,
      "learning_rate": 3.7705e-05,
      "loss": 0.3454,
      "step": 7544
    },
    {
      "epoch": 0.42249972001343933,
      "grad_norm": 1.478772521018982,
      "learning_rate": 3.771e-05,
      "loss": 0.5987,
      "step": 7545
    },
    {
      "epoch": 0.42255571732556835,
      "grad_norm": 1.3632503747940063,
      "learning_rate": 3.7715e-05,
      "loss": 0.4575,
      "step": 7546
    },
    {
      "epoch": 0.42261171463769737,
      "grad_norm": 1.4847928285598755,
      "learning_rate": 3.772e-05,
      "loss": 0.3567,
      "step": 7547
    },
    {
      "epoch": 0.4226677119498264,
      "grad_norm": 1.3456907272338867,
      "learning_rate": 3.7725e-05,
      "loss": 0.542,
      "step": 7548
    },
    {
      "epoch": 0.4227237092619554,
      "grad_norm": 1.2731678485870361,
      "learning_rate": 3.7730000000000006e-05,
      "loss": 0.4282,
      "step": 7549
    },
    {
      "epoch": 0.4227797065740844,
      "grad_norm": 4.894685745239258,
      "learning_rate": 3.7735000000000004e-05,
      "loss": 0.6533,
      "step": 7550
    },
    {
      "epoch": 0.42283570388621344,
      "grad_norm": 1.2923585176467896,
      "learning_rate": 3.774e-05,
      "loss": 0.6542,
      "step": 7551
    },
    {
      "epoch": 0.42289170119834246,
      "grad_norm": 1.7255858182907104,
      "learning_rate": 3.7745000000000005e-05,
      "loss": 0.5819,
      "step": 7552
    },
    {
      "epoch": 0.4229476985104715,
      "grad_norm": 1.4177074432373047,
      "learning_rate": 3.775e-05,
      "loss": 0.5332,
      "step": 7553
    },
    {
      "epoch": 0.4230036958226005,
      "grad_norm": 1.2697304487228394,
      "learning_rate": 3.7755e-05,
      "loss": 0.48,
      "step": 7554
    },
    {
      "epoch": 0.4230596931347295,
      "grad_norm": 1.4944721460342407,
      "learning_rate": 3.776e-05,
      "loss": 0.6667,
      "step": 7555
    },
    {
      "epoch": 0.42311569044685854,
      "grad_norm": 1.0259102582931519,
      "learning_rate": 3.7765e-05,
      "loss": 0.3976,
      "step": 7556
    },
    {
      "epoch": 0.42317168775898756,
      "grad_norm": 1.1391164064407349,
      "learning_rate": 3.777e-05,
      "loss": 0.526,
      "step": 7557
    },
    {
      "epoch": 0.4232276850711166,
      "grad_norm": 1.4234325885772705,
      "learning_rate": 3.7775e-05,
      "loss": 0.5159,
      "step": 7558
    },
    {
      "epoch": 0.4232836823832456,
      "grad_norm": 1.290900468826294,
      "learning_rate": 3.778000000000001e-05,
      "loss": 0.4166,
      "step": 7559
    },
    {
      "epoch": 0.4233396796953746,
      "grad_norm": 1.2310255765914917,
      "learning_rate": 3.7785000000000005e-05,
      "loss": 0.4356,
      "step": 7560
    },
    {
      "epoch": 0.42339567700750363,
      "grad_norm": 1.2476540803909302,
      "learning_rate": 3.779e-05,
      "loss": 0.3219,
      "step": 7561
    },
    {
      "epoch": 0.42345167431963265,
      "grad_norm": 1.3199870586395264,
      "learning_rate": 3.7795e-05,
      "loss": 0.487,
      "step": 7562
    },
    {
      "epoch": 0.4235076716317617,
      "grad_norm": 1.2570263147354126,
      "learning_rate": 3.7800000000000004e-05,
      "loss": 0.4006,
      "step": 7563
    },
    {
      "epoch": 0.4235636689438907,
      "grad_norm": 1.365960955619812,
      "learning_rate": 3.7805e-05,
      "loss": 0.4384,
      "step": 7564
    },
    {
      "epoch": 0.4236196662560197,
      "grad_norm": 1.5694210529327393,
      "learning_rate": 3.781e-05,
      "loss": 0.528,
      "step": 7565
    },
    {
      "epoch": 0.42367566356814873,
      "grad_norm": 1.0389361381530762,
      "learning_rate": 3.7815e-05,
      "loss": 0.3886,
      "step": 7566
    },
    {
      "epoch": 0.42373166088027775,
      "grad_norm": Infinity,
      "learning_rate": 3.7815e-05,
      "loss": 0.4799,
      "step": 7567
    },
    {
      "epoch": 0.42378765819240677,
      "grad_norm": 1.392661690711975,
      "learning_rate": 3.782e-05,
      "loss": 0.4247,
      "step": 7568
    },
    {
      "epoch": 0.4238436555045358,
      "grad_norm": 1.1096845865249634,
      "learning_rate": 3.7825e-05,
      "loss": 0.354,
      "step": 7569
    },
    {
      "epoch": 0.4238996528166648,
      "grad_norm": 1.1711372137069702,
      "learning_rate": 3.783e-05,
      "loss": 0.4522,
      "step": 7570
    },
    {
      "epoch": 0.4239556501287938,
      "grad_norm": 1.2791259288787842,
      "learning_rate": 3.7835000000000006e-05,
      "loss": 0.3417,
      "step": 7571
    },
    {
      "epoch": 0.42401164744092285,
      "grad_norm": 1.1541465520858765,
      "learning_rate": 3.7840000000000004e-05,
      "loss": 0.4978,
      "step": 7572
    },
    {
      "epoch": 0.42406764475305186,
      "grad_norm": 1.3474169969558716,
      "learning_rate": 3.7845e-05,
      "loss": 0.4274,
      "step": 7573
    },
    {
      "epoch": 0.4241236420651809,
      "grad_norm": 1.2171330451965332,
      "learning_rate": 3.7850000000000005e-05,
      "loss": 0.4049,
      "step": 7574
    },
    {
      "epoch": 0.4241796393773099,
      "grad_norm": 1.2262377738952637,
      "learning_rate": 3.7855e-05,
      "loss": 0.3458,
      "step": 7575
    },
    {
      "epoch": 0.4242356366894389,
      "grad_norm": 2.331810474395752,
      "learning_rate": 3.786e-05,
      "loss": 0.4114,
      "step": 7576
    },
    {
      "epoch": 0.42429163400156794,
      "grad_norm": 1.209799885749817,
      "learning_rate": 3.7865e-05,
      "loss": 0.4726,
      "step": 7577
    },
    {
      "epoch": 0.42434763131369696,
      "grad_norm": 1.1870396137237549,
      "learning_rate": 3.787e-05,
      "loss": 0.4475,
      "step": 7578
    },
    {
      "epoch": 0.424403628625826,
      "grad_norm": 1.3213865756988525,
      "learning_rate": 3.7875e-05,
      "loss": 0.4684,
      "step": 7579
    },
    {
      "epoch": 0.424459625937955,
      "grad_norm": 1.2779390811920166,
      "learning_rate": 3.788e-05,
      "loss": 0.4014,
      "step": 7580
    },
    {
      "epoch": 0.424515623250084,
      "grad_norm": 1.1260337829589844,
      "learning_rate": 3.7885e-05,
      "loss": 0.4165,
      "step": 7581
    },
    {
      "epoch": 0.42457162056221304,
      "grad_norm": 1.3278111219406128,
      "learning_rate": 3.7890000000000005e-05,
      "loss": 0.4884,
      "step": 7582
    },
    {
      "epoch": 0.42462761787434206,
      "grad_norm": 1.2361208200454712,
      "learning_rate": 3.7895e-05,
      "loss": 0.4114,
      "step": 7583
    },
    {
      "epoch": 0.4246836151864711,
      "grad_norm": 1.1926347017288208,
      "learning_rate": 3.79e-05,
      "loss": 0.4686,
      "step": 7584
    },
    {
      "epoch": 0.4247396124986001,
      "grad_norm": 1.308377981185913,
      "learning_rate": 3.7905000000000004e-05,
      "loss": 0.5798,
      "step": 7585
    },
    {
      "epoch": 0.42479560981072906,
      "grad_norm": 1.328100562095642,
      "learning_rate": 3.791e-05,
      "loss": 0.4104,
      "step": 7586
    },
    {
      "epoch": 0.4248516071228581,
      "grad_norm": 1.2501837015151978,
      "learning_rate": 3.7915e-05,
      "loss": 0.3796,
      "step": 7587
    },
    {
      "epoch": 0.4249076044349871,
      "grad_norm": 1.293765664100647,
      "learning_rate": 3.792e-05,
      "loss": 0.3548,
      "step": 7588
    },
    {
      "epoch": 0.4249636017471161,
      "grad_norm": 1.2599366903305054,
      "learning_rate": 3.7925e-05,
      "loss": 0.3834,
      "step": 7589
    },
    {
      "epoch": 0.42501959905924513,
      "grad_norm": 1.2977622747421265,
      "learning_rate": 3.7930000000000004e-05,
      "loss": 0.5509,
      "step": 7590
    },
    {
      "epoch": 0.42507559637137415,
      "grad_norm": 1.1092506647109985,
      "learning_rate": 3.7935e-05,
      "loss": 0.4144,
      "step": 7591
    },
    {
      "epoch": 0.4251315936835032,
      "grad_norm": 1.2002649307250977,
      "learning_rate": 3.7940000000000006e-05,
      "loss": 0.3521,
      "step": 7592
    },
    {
      "epoch": 0.4251875909956322,
      "grad_norm": 1.210794448852539,
      "learning_rate": 3.7945000000000003e-05,
      "loss": 0.3235,
      "step": 7593
    },
    {
      "epoch": 0.4252435883077612,
      "grad_norm": 1.4287084341049194,
      "learning_rate": 3.795e-05,
      "loss": 0.4229,
      "step": 7594
    },
    {
      "epoch": 0.42529958561989023,
      "grad_norm": 1.0724073648452759,
      "learning_rate": 3.7955e-05,
      "loss": 0.3281,
      "step": 7595
    },
    {
      "epoch": 0.42535558293201925,
      "grad_norm": 1.2447272539138794,
      "learning_rate": 3.796e-05,
      "loss": 0.4698,
      "step": 7596
    },
    {
      "epoch": 0.42541158024414827,
      "grad_norm": 1.1794023513793945,
      "learning_rate": 3.7965e-05,
      "loss": 0.4844,
      "step": 7597
    },
    {
      "epoch": 0.4254675775562773,
      "grad_norm": 1.178626298904419,
      "learning_rate": 3.797e-05,
      "loss": 0.4751,
      "step": 7598
    },
    {
      "epoch": 0.4255235748684063,
      "grad_norm": 1.1559756994247437,
      "learning_rate": 3.7975e-05,
      "loss": 0.3661,
      "step": 7599
    },
    {
      "epoch": 0.4255795721805353,
      "grad_norm": 1.593385934829712,
      "learning_rate": 3.7980000000000006e-05,
      "loss": 0.6293,
      "step": 7600
    },
    {
      "epoch": 0.42563556949266435,
      "grad_norm": 1.361419439315796,
      "learning_rate": 3.7985e-05,
      "loss": 0.5309,
      "step": 7601
    },
    {
      "epoch": 0.42569156680479336,
      "grad_norm": 1.2820345163345337,
      "learning_rate": 3.799e-05,
      "loss": 0.5431,
      "step": 7602
    },
    {
      "epoch": 0.4257475641169224,
      "grad_norm": 1.487322449684143,
      "learning_rate": 3.7995000000000005e-05,
      "loss": 0.548,
      "step": 7603
    },
    {
      "epoch": 0.4258035614290514,
      "grad_norm": 1.1518892049789429,
      "learning_rate": 3.8e-05,
      "loss": 0.5031,
      "step": 7604
    },
    {
      "epoch": 0.4258595587411804,
      "grad_norm": 1.4109389781951904,
      "learning_rate": 3.8005e-05,
      "loss": 0.5499,
      "step": 7605
    },
    {
      "epoch": 0.42591555605330944,
      "grad_norm": 1.2256593704223633,
      "learning_rate": 3.8010000000000004e-05,
      "loss": 0.4154,
      "step": 7606
    },
    {
      "epoch": 0.42597155336543846,
      "grad_norm": 1.0937228202819824,
      "learning_rate": 3.8015e-05,
      "loss": 0.3778,
      "step": 7607
    },
    {
      "epoch": 0.4260275506775675,
      "grad_norm": 1.2796554565429688,
      "learning_rate": 3.802e-05,
      "loss": 0.4817,
      "step": 7608
    },
    {
      "epoch": 0.4260835479896965,
      "grad_norm": 1.2420028448104858,
      "learning_rate": 3.8025e-05,
      "loss": 0.4238,
      "step": 7609
    },
    {
      "epoch": 0.4261395453018255,
      "grad_norm": 1.066815733909607,
      "learning_rate": 3.803000000000001e-05,
      "loss": 0.4001,
      "step": 7610
    },
    {
      "epoch": 0.42619554261395454,
      "grad_norm": 1.2149802446365356,
      "learning_rate": 3.8035000000000004e-05,
      "loss": 0.3767,
      "step": 7611
    },
    {
      "epoch": 0.42625153992608356,
      "grad_norm": 1.2878392934799194,
      "learning_rate": 3.804e-05,
      "loss": 0.419,
      "step": 7612
    },
    {
      "epoch": 0.4263075372382126,
      "grad_norm": 1.2985421419143677,
      "learning_rate": 3.8045000000000006e-05,
      "loss": 0.447,
      "step": 7613
    },
    {
      "epoch": 0.4263635345503416,
      "grad_norm": 1.4540281295776367,
      "learning_rate": 3.805e-05,
      "loss": 0.3851,
      "step": 7614
    },
    {
      "epoch": 0.4264195318624706,
      "grad_norm": 1.2379271984100342,
      "learning_rate": 3.8055e-05,
      "loss": 0.4092,
      "step": 7615
    },
    {
      "epoch": 0.42647552917459963,
      "grad_norm": 1.2478660345077515,
      "learning_rate": 3.806e-05,
      "loss": 0.3936,
      "step": 7616
    },
    {
      "epoch": 0.42653152648672865,
      "grad_norm": 1.6623990535736084,
      "learning_rate": 3.8065e-05,
      "loss": 0.5192,
      "step": 7617
    },
    {
      "epoch": 0.42658752379885767,
      "grad_norm": 1.3103492259979248,
      "learning_rate": 3.807e-05,
      "loss": 0.4512,
      "step": 7618
    },
    {
      "epoch": 0.4266435211109867,
      "grad_norm": 1.294898509979248,
      "learning_rate": 3.8075e-05,
      "loss": 0.4323,
      "step": 7619
    },
    {
      "epoch": 0.4266995184231157,
      "grad_norm": 1.214297890663147,
      "learning_rate": 3.808e-05,
      "loss": 0.3929,
      "step": 7620
    },
    {
      "epoch": 0.42675551573524473,
      "grad_norm": 1.2719229459762573,
      "learning_rate": 3.8085000000000006e-05,
      "loss": 0.4642,
      "step": 7621
    },
    {
      "epoch": 0.42681151304737375,
      "grad_norm": 1.2533601522445679,
      "learning_rate": 3.809e-05,
      "loss": 0.3769,
      "step": 7622
    },
    {
      "epoch": 0.42686751035950277,
      "grad_norm": 1.5504531860351562,
      "learning_rate": 3.8095e-05,
      "loss": 0.4822,
      "step": 7623
    },
    {
      "epoch": 0.4269235076716318,
      "grad_norm": 1.2235283851623535,
      "learning_rate": 3.8100000000000005e-05,
      "loss": 0.3915,
      "step": 7624
    },
    {
      "epoch": 0.4269795049837608,
      "grad_norm": 1.3170377016067505,
      "learning_rate": 3.8105e-05,
      "loss": 0.4689,
      "step": 7625
    },
    {
      "epoch": 0.4270355022958898,
      "grad_norm": 1.1815876960754395,
      "learning_rate": 3.811e-05,
      "loss": 0.4741,
      "step": 7626
    },
    {
      "epoch": 0.4270914996080188,
      "grad_norm": 1.4471092224121094,
      "learning_rate": 3.8115000000000004e-05,
      "loss": 0.5063,
      "step": 7627
    },
    {
      "epoch": 0.4271474969201478,
      "grad_norm": 1.2353376150131226,
      "learning_rate": 3.812e-05,
      "loss": 0.4274,
      "step": 7628
    },
    {
      "epoch": 0.4272034942322768,
      "grad_norm": 1.41990065574646,
      "learning_rate": 3.8125e-05,
      "loss": 0.4,
      "step": 7629
    },
    {
      "epoch": 0.42725949154440584,
      "grad_norm": 1.7524484395980835,
      "learning_rate": 3.8129999999999996e-05,
      "loss": 0.4488,
      "step": 7630
    },
    {
      "epoch": 0.42731548885653486,
      "grad_norm": 1.1383867263793945,
      "learning_rate": 3.813500000000001e-05,
      "loss": 0.5019,
      "step": 7631
    },
    {
      "epoch": 0.4273714861686639,
      "grad_norm": 1.3128741979599,
      "learning_rate": 3.8140000000000004e-05,
      "loss": 0.3711,
      "step": 7632
    },
    {
      "epoch": 0.4274274834807929,
      "grad_norm": 1.2062790393829346,
      "learning_rate": 3.8145e-05,
      "loss": 0.4014,
      "step": 7633
    },
    {
      "epoch": 0.4274834807929219,
      "grad_norm": 1.2138209342956543,
      "learning_rate": 3.8150000000000006e-05,
      "loss": 0.4114,
      "step": 7634
    },
    {
      "epoch": 0.42753947810505094,
      "grad_norm": 1.4575783014297485,
      "learning_rate": 3.8155e-05,
      "loss": 0.6132,
      "step": 7635
    },
    {
      "epoch": 0.42759547541717996,
      "grad_norm": 1.1320710182189941,
      "learning_rate": 3.816e-05,
      "loss": 0.3843,
      "step": 7636
    },
    {
      "epoch": 0.427651472729309,
      "grad_norm": 1.2448700666427612,
      "learning_rate": 3.8165e-05,
      "loss": 0.5245,
      "step": 7637
    },
    {
      "epoch": 0.427707470041438,
      "grad_norm": 0.9886170029640198,
      "learning_rate": 3.817e-05,
      "loss": 0.3921,
      "step": 7638
    },
    {
      "epoch": 0.427763467353567,
      "grad_norm": 1.1423417329788208,
      "learning_rate": 3.8175e-05,
      "loss": 0.3704,
      "step": 7639
    },
    {
      "epoch": 0.42781946466569604,
      "grad_norm": 1.307464838027954,
      "learning_rate": 3.818e-05,
      "loss": 0.4182,
      "step": 7640
    },
    {
      "epoch": 0.42787546197782506,
      "grad_norm": 1.1473870277404785,
      "learning_rate": 3.8185e-05,
      "loss": 0.4207,
      "step": 7641
    },
    {
      "epoch": 0.4279314592899541,
      "grad_norm": 1.1849149465560913,
      "learning_rate": 3.8190000000000005e-05,
      "loss": 0.526,
      "step": 7642
    },
    {
      "epoch": 0.4279874566020831,
      "grad_norm": 1.3337353467941284,
      "learning_rate": 3.8195e-05,
      "loss": 0.3972,
      "step": 7643
    },
    {
      "epoch": 0.4280434539142121,
      "grad_norm": 1.0717594623565674,
      "learning_rate": 3.82e-05,
      "loss": 0.3909,
      "step": 7644
    },
    {
      "epoch": 0.42809945122634113,
      "grad_norm": 1.5796343088150024,
      "learning_rate": 3.8205000000000004e-05,
      "loss": 0.6941,
      "step": 7645
    },
    {
      "epoch": 0.42815544853847015,
      "grad_norm": 1.196513295173645,
      "learning_rate": 3.821e-05,
      "loss": 0.4088,
      "step": 7646
    },
    {
      "epoch": 0.42821144585059917,
      "grad_norm": 1.0890085697174072,
      "learning_rate": 3.8215e-05,
      "loss": 0.4185,
      "step": 7647
    },
    {
      "epoch": 0.4282674431627282,
      "grad_norm": 1.166717767715454,
      "learning_rate": 3.822e-05,
      "loss": 0.4052,
      "step": 7648
    },
    {
      "epoch": 0.4283234404748572,
      "grad_norm": 1.1860620975494385,
      "learning_rate": 3.8225e-05,
      "loss": 0.5055,
      "step": 7649
    },
    {
      "epoch": 0.42837943778698623,
      "grad_norm": 1.2802859544754028,
      "learning_rate": 3.823e-05,
      "loss": 0.3698,
      "step": 7650
    },
    {
      "epoch": 0.42843543509911525,
      "grad_norm": 1.1276755332946777,
      "learning_rate": 3.8235e-05,
      "loss": 0.4438,
      "step": 7651
    },
    {
      "epoch": 0.42849143241124427,
      "grad_norm": 1.3856489658355713,
      "learning_rate": 3.8240000000000007e-05,
      "loss": 0.5302,
      "step": 7652
    },
    {
      "epoch": 0.4285474297233733,
      "grad_norm": 1.2472093105316162,
      "learning_rate": 3.8245000000000004e-05,
      "loss": 0.3277,
      "step": 7653
    },
    {
      "epoch": 0.4286034270355023,
      "grad_norm": 1.3688454627990723,
      "learning_rate": 3.825e-05,
      "loss": 0.3755,
      "step": 7654
    },
    {
      "epoch": 0.4286594243476313,
      "grad_norm": 1.4699273109436035,
      "learning_rate": 3.8255e-05,
      "loss": 0.4998,
      "step": 7655
    },
    {
      "epoch": 0.42871542165976034,
      "grad_norm": 1.7625372409820557,
      "learning_rate": 3.826e-05,
      "loss": 0.6153,
      "step": 7656
    },
    {
      "epoch": 0.42877141897188936,
      "grad_norm": 1.4056943655014038,
      "learning_rate": 3.8265e-05,
      "loss": 0.4559,
      "step": 7657
    },
    {
      "epoch": 0.4288274162840184,
      "grad_norm": 1.6552726030349731,
      "learning_rate": 3.827e-05,
      "loss": 0.4928,
      "step": 7658
    },
    {
      "epoch": 0.4288834135961474,
      "grad_norm": 1.1531238555908203,
      "learning_rate": 3.8275e-05,
      "loss": 0.3823,
      "step": 7659
    },
    {
      "epoch": 0.4289394109082764,
      "grad_norm": 1.2474634647369385,
      "learning_rate": 3.828e-05,
      "loss": 0.4445,
      "step": 7660
    },
    {
      "epoch": 0.42899540822040544,
      "grad_norm": 1.1210706233978271,
      "learning_rate": 3.8285000000000004e-05,
      "loss": 0.3965,
      "step": 7661
    },
    {
      "epoch": 0.42905140553253446,
      "grad_norm": 1.180922508239746,
      "learning_rate": 3.829e-05,
      "loss": 0.3521,
      "step": 7662
    },
    {
      "epoch": 0.4291074028446635,
      "grad_norm": 0.9694157838821411,
      "learning_rate": 3.8295000000000005e-05,
      "loss": 0.4278,
      "step": 7663
    },
    {
      "epoch": 0.4291634001567925,
      "grad_norm": 1.468010663986206,
      "learning_rate": 3.83e-05,
      "loss": 0.5533,
      "step": 7664
    },
    {
      "epoch": 0.4292193974689215,
      "grad_norm": 1.3029483556747437,
      "learning_rate": 3.8305e-05,
      "loss": 0.3681,
      "step": 7665
    },
    {
      "epoch": 0.42927539478105053,
      "grad_norm": 1.5082733631134033,
      "learning_rate": 3.8310000000000004e-05,
      "loss": 0.4196,
      "step": 7666
    },
    {
      "epoch": 0.42933139209317955,
      "grad_norm": 1.4620344638824463,
      "learning_rate": 3.8315e-05,
      "loss": 0.4516,
      "step": 7667
    },
    {
      "epoch": 0.4293873894053086,
      "grad_norm": 1.3122128248214722,
      "learning_rate": 3.832e-05,
      "loss": 0.5609,
      "step": 7668
    },
    {
      "epoch": 0.42944338671743754,
      "grad_norm": 1.352057695388794,
      "learning_rate": 3.8324999999999996e-05,
      "loss": 0.7109,
      "step": 7669
    },
    {
      "epoch": 0.42949938402956656,
      "grad_norm": 1.085706353187561,
      "learning_rate": 3.833e-05,
      "loss": 0.3828,
      "step": 7670
    },
    {
      "epoch": 0.4295553813416956,
      "grad_norm": 1.1122759580612183,
      "learning_rate": 3.8335000000000005e-05,
      "loss": 0.3223,
      "step": 7671
    },
    {
      "epoch": 0.4296113786538246,
      "grad_norm": 1.310568928718567,
      "learning_rate": 3.834e-05,
      "loss": 0.3985,
      "step": 7672
    },
    {
      "epoch": 0.4296673759659536,
      "grad_norm": 1.6231038570404053,
      "learning_rate": 3.8345000000000006e-05,
      "loss": 0.531,
      "step": 7673
    },
    {
      "epoch": 0.42972337327808263,
      "grad_norm": 1.6763800382614136,
      "learning_rate": 3.8350000000000004e-05,
      "loss": 0.4758,
      "step": 7674
    },
    {
      "epoch": 0.42977937059021165,
      "grad_norm": 1.440268874168396,
      "learning_rate": 3.8355e-05,
      "loss": 0.5195,
      "step": 7675
    },
    {
      "epoch": 0.42983536790234067,
      "grad_norm": 1.206406593322754,
      "learning_rate": 3.836e-05,
      "loss": 0.3597,
      "step": 7676
    },
    {
      "epoch": 0.4298913652144697,
      "grad_norm": 1.423871636390686,
      "learning_rate": 3.8365e-05,
      "loss": 0.3996,
      "step": 7677
    },
    {
      "epoch": 0.4299473625265987,
      "grad_norm": 1.1362015008926392,
      "learning_rate": 3.837e-05,
      "loss": 0.3242,
      "step": 7678
    },
    {
      "epoch": 0.43000335983872773,
      "grad_norm": 1.2650737762451172,
      "learning_rate": 3.8375e-05,
      "loss": 0.3938,
      "step": 7679
    },
    {
      "epoch": 0.43005935715085675,
      "grad_norm": 1.2767328023910522,
      "learning_rate": 3.838e-05,
      "loss": 0.3629,
      "step": 7680
    },
    {
      "epoch": 0.43011535446298577,
      "grad_norm": 1.2091460227966309,
      "learning_rate": 3.8385000000000006e-05,
      "loss": 0.3916,
      "step": 7681
    },
    {
      "epoch": 0.4301713517751148,
      "grad_norm": 1.4345455169677734,
      "learning_rate": 3.8390000000000003e-05,
      "loss": 0.4058,
      "step": 7682
    },
    {
      "epoch": 0.4302273490872438,
      "grad_norm": 1.095362663269043,
      "learning_rate": 3.8395e-05,
      "loss": 0.4278,
      "step": 7683
    },
    {
      "epoch": 0.4302833463993728,
      "grad_norm": 1.3193087577819824,
      "learning_rate": 3.8400000000000005e-05,
      "loss": 0.3639,
      "step": 7684
    },
    {
      "epoch": 0.43033934371150184,
      "grad_norm": 1.5311903953552246,
      "learning_rate": 3.8405e-05,
      "loss": 0.4734,
      "step": 7685
    },
    {
      "epoch": 0.43039534102363086,
      "grad_norm": 1.3788670301437378,
      "learning_rate": 3.841e-05,
      "loss": 0.4635,
      "step": 7686
    },
    {
      "epoch": 0.4304513383357599,
      "grad_norm": 1.1755338907241821,
      "learning_rate": 3.8415000000000004e-05,
      "loss": 0.3556,
      "step": 7687
    },
    {
      "epoch": 0.4305073356478889,
      "grad_norm": 1.3992962837219238,
      "learning_rate": 3.842e-05,
      "loss": 0.4594,
      "step": 7688
    },
    {
      "epoch": 0.4305633329600179,
      "grad_norm": 1.4017950296401978,
      "learning_rate": 3.8425e-05,
      "loss": 0.4544,
      "step": 7689
    },
    {
      "epoch": 0.43061933027214694,
      "grad_norm": 1.3413540124893188,
      "learning_rate": 3.8429999999999996e-05,
      "loss": 0.4211,
      "step": 7690
    },
    {
      "epoch": 0.43067532758427596,
      "grad_norm": 1.6711502075195312,
      "learning_rate": 3.843500000000001e-05,
      "loss": 0.4686,
      "step": 7691
    },
    {
      "epoch": 0.430731324896405,
      "grad_norm": 1.2665237188339233,
      "learning_rate": 3.8440000000000005e-05,
      "loss": 0.3576,
      "step": 7692
    },
    {
      "epoch": 0.430787322208534,
      "grad_norm": 1.1450090408325195,
      "learning_rate": 3.8445e-05,
      "loss": 0.304,
      "step": 7693
    },
    {
      "epoch": 0.430843319520663,
      "grad_norm": 1.2788779735565186,
      "learning_rate": 3.845e-05,
      "loss": 0.5303,
      "step": 7694
    },
    {
      "epoch": 0.43089931683279203,
      "grad_norm": 1.3919241428375244,
      "learning_rate": 3.8455000000000004e-05,
      "loss": 0.5688,
      "step": 7695
    },
    {
      "epoch": 0.43095531414492105,
      "grad_norm": 1.1815086603164673,
      "learning_rate": 3.846e-05,
      "loss": 0.4524,
      "step": 7696
    },
    {
      "epoch": 0.4310113114570501,
      "grad_norm": 1.1565638780593872,
      "learning_rate": 3.8465e-05,
      "loss": 0.4198,
      "step": 7697
    },
    {
      "epoch": 0.4310673087691791,
      "grad_norm": 1.2445493936538696,
      "learning_rate": 3.847e-05,
      "loss": 0.6976,
      "step": 7698
    },
    {
      "epoch": 0.4311233060813081,
      "grad_norm": 1.1842223405838013,
      "learning_rate": 3.8475e-05,
      "loss": 0.4367,
      "step": 7699
    },
    {
      "epoch": 0.43117930339343713,
      "grad_norm": 1.1668330430984497,
      "learning_rate": 3.848e-05,
      "loss": 0.386,
      "step": 7700
    },
    {
      "epoch": 0.43123530070556615,
      "grad_norm": 1.2867923974990845,
      "learning_rate": 3.8485e-05,
      "loss": 0.4306,
      "step": 7701
    },
    {
      "epoch": 0.43129129801769517,
      "grad_norm": 1.5501301288604736,
      "learning_rate": 3.8490000000000006e-05,
      "loss": 0.4631,
      "step": 7702
    },
    {
      "epoch": 0.4313472953298242,
      "grad_norm": 1.4100300073623657,
      "learning_rate": 3.8495e-05,
      "loss": 0.5668,
      "step": 7703
    },
    {
      "epoch": 0.4314032926419532,
      "grad_norm": 1.12372887134552,
      "learning_rate": 3.85e-05,
      "loss": 0.3659,
      "step": 7704
    },
    {
      "epoch": 0.4314592899540822,
      "grad_norm": 1.3030691146850586,
      "learning_rate": 3.8505000000000005e-05,
      "loss": 0.353,
      "step": 7705
    },
    {
      "epoch": 0.43151528726621124,
      "grad_norm": 1.2100201845169067,
      "learning_rate": 3.851e-05,
      "loss": 0.3685,
      "step": 7706
    },
    {
      "epoch": 0.43157128457834026,
      "grad_norm": 1.2769105434417725,
      "learning_rate": 3.8515e-05,
      "loss": 0.4037,
      "step": 7707
    },
    {
      "epoch": 0.4316272818904693,
      "grad_norm": 1.2264678478240967,
      "learning_rate": 3.8520000000000004e-05,
      "loss": 0.4961,
      "step": 7708
    },
    {
      "epoch": 0.4316832792025983,
      "grad_norm": 1.1368542909622192,
      "learning_rate": 3.8525e-05,
      "loss": 0.5397,
      "step": 7709
    },
    {
      "epoch": 0.43173927651472727,
      "grad_norm": 1.2379624843597412,
      "learning_rate": 3.853e-05,
      "loss": 0.4158,
      "step": 7710
    },
    {
      "epoch": 0.4317952738268563,
      "grad_norm": 1.4977108240127563,
      "learning_rate": 3.8535e-05,
      "loss": 0.4232,
      "step": 7711
    },
    {
      "epoch": 0.4318512711389853,
      "grad_norm": 1.2705695629119873,
      "learning_rate": 3.854000000000001e-05,
      "loss": 0.3418,
      "step": 7712
    },
    {
      "epoch": 0.4319072684511143,
      "grad_norm": 1.5933589935302734,
      "learning_rate": 3.8545000000000004e-05,
      "loss": 0.4735,
      "step": 7713
    },
    {
      "epoch": 0.43196326576324334,
      "grad_norm": 1.3652877807617188,
      "learning_rate": 3.855e-05,
      "loss": 0.463,
      "step": 7714
    },
    {
      "epoch": 0.43201926307537236,
      "grad_norm": 1.5132670402526855,
      "learning_rate": 3.8555e-05,
      "loss": 0.4214,
      "step": 7715
    },
    {
      "epoch": 0.4320752603875014,
      "grad_norm": 1.5290735960006714,
      "learning_rate": 3.8560000000000004e-05,
      "loss": 0.4887,
      "step": 7716
    },
    {
      "epoch": 0.4321312576996304,
      "grad_norm": 1.0978847742080688,
      "learning_rate": 3.8565e-05,
      "loss": 0.318,
      "step": 7717
    },
    {
      "epoch": 0.4321872550117594,
      "grad_norm": 1.5986554622650146,
      "learning_rate": 3.857e-05,
      "loss": 0.3287,
      "step": 7718
    },
    {
      "epoch": 0.43224325232388844,
      "grad_norm": 1.3401552438735962,
      "learning_rate": 3.8575e-05,
      "loss": 0.4553,
      "step": 7719
    },
    {
      "epoch": 0.43229924963601746,
      "grad_norm": 1.586480975151062,
      "learning_rate": 3.858e-05,
      "loss": 0.5544,
      "step": 7720
    },
    {
      "epoch": 0.4323552469481465,
      "grad_norm": 1.1275185346603394,
      "learning_rate": 3.8585000000000004e-05,
      "loss": 0.3601,
      "step": 7721
    },
    {
      "epoch": 0.4324112442602755,
      "grad_norm": 1.3213917016983032,
      "learning_rate": 3.859e-05,
      "loss": 0.4303,
      "step": 7722
    },
    {
      "epoch": 0.4324672415724045,
      "grad_norm": 1.2904232740402222,
      "learning_rate": 3.8595000000000006e-05,
      "loss": 0.4344,
      "step": 7723
    },
    {
      "epoch": 0.43252323888453353,
      "grad_norm": 1.3175349235534668,
      "learning_rate": 3.86e-05,
      "loss": 0.492,
      "step": 7724
    },
    {
      "epoch": 0.43257923619666255,
      "grad_norm": 1.0625821352005005,
      "learning_rate": 3.8605e-05,
      "loss": 0.3455,
      "step": 7725
    },
    {
      "epoch": 0.4326352335087916,
      "grad_norm": 1.3834384679794312,
      "learning_rate": 3.8610000000000005e-05,
      "loss": 0.395,
      "step": 7726
    },
    {
      "epoch": 0.4326912308209206,
      "grad_norm": 1.3488636016845703,
      "learning_rate": 3.8615e-05,
      "loss": 0.4571,
      "step": 7727
    },
    {
      "epoch": 0.4327472281330496,
      "grad_norm": 1.7581381797790527,
      "learning_rate": 3.862e-05,
      "loss": 0.4489,
      "step": 7728
    },
    {
      "epoch": 0.43280322544517863,
      "grad_norm": 1.182015299797058,
      "learning_rate": 3.8625e-05,
      "loss": 0.2721,
      "step": 7729
    },
    {
      "epoch": 0.43285922275730765,
      "grad_norm": 1.2194050550460815,
      "learning_rate": 3.863e-05,
      "loss": 0.4309,
      "step": 7730
    },
    {
      "epoch": 0.43291522006943667,
      "grad_norm": 1.1201328039169312,
      "learning_rate": 3.8635000000000005e-05,
      "loss": 0.3698,
      "step": 7731
    },
    {
      "epoch": 0.4329712173815657,
      "grad_norm": 1.129077672958374,
      "learning_rate": 3.864e-05,
      "loss": 0.4306,
      "step": 7732
    },
    {
      "epoch": 0.4330272146936947,
      "grad_norm": 1.4392932653427124,
      "learning_rate": 3.8645e-05,
      "loss": 0.4341,
      "step": 7733
    },
    {
      "epoch": 0.4330832120058237,
      "grad_norm": 1.3381160497665405,
      "learning_rate": 3.8650000000000004e-05,
      "loss": 0.3719,
      "step": 7734
    },
    {
      "epoch": 0.43313920931795274,
      "grad_norm": 1.306389570236206,
      "learning_rate": 3.8655e-05,
      "loss": 0.6106,
      "step": 7735
    },
    {
      "epoch": 0.43319520663008176,
      "grad_norm": 1.388902187347412,
      "learning_rate": 3.866e-05,
      "loss": 0.8112,
      "step": 7736
    },
    {
      "epoch": 0.4332512039422108,
      "grad_norm": 1.2088311910629272,
      "learning_rate": 3.8665e-05,
      "loss": 0.5203,
      "step": 7737
    },
    {
      "epoch": 0.4333072012543398,
      "grad_norm": 1.1478239297866821,
      "learning_rate": 3.867e-05,
      "loss": 0.4369,
      "step": 7738
    },
    {
      "epoch": 0.4333631985664688,
      "grad_norm": 1.2467337846755981,
      "learning_rate": 3.8675e-05,
      "loss": 0.5301,
      "step": 7739
    },
    {
      "epoch": 0.43341919587859784,
      "grad_norm": 1.1686300039291382,
      "learning_rate": 3.868e-05,
      "loss": 0.395,
      "step": 7740
    },
    {
      "epoch": 0.43347519319072686,
      "grad_norm": 1.4347566366195679,
      "learning_rate": 3.8685000000000007e-05,
      "loss": 0.4974,
      "step": 7741
    },
    {
      "epoch": 0.4335311905028559,
      "grad_norm": 1.3105829954147339,
      "learning_rate": 3.8690000000000004e-05,
      "loss": 0.433,
      "step": 7742
    },
    {
      "epoch": 0.4335871878149849,
      "grad_norm": 1.149224042892456,
      "learning_rate": 3.8695e-05,
      "loss": 0.5755,
      "step": 7743
    },
    {
      "epoch": 0.4336431851271139,
      "grad_norm": 1.3344298601150513,
      "learning_rate": 3.8700000000000006e-05,
      "loss": 0.5189,
      "step": 7744
    },
    {
      "epoch": 0.43369918243924294,
      "grad_norm": 1.017473816871643,
      "learning_rate": 3.8705e-05,
      "loss": 0.3963,
      "step": 7745
    },
    {
      "epoch": 0.43375517975137196,
      "grad_norm": 1.1155141592025757,
      "learning_rate": 3.871e-05,
      "loss": 0.389,
      "step": 7746
    },
    {
      "epoch": 0.433811177063501,
      "grad_norm": 1.467545986175537,
      "learning_rate": 3.8715000000000005e-05,
      "loss": 0.4094,
      "step": 7747
    },
    {
      "epoch": 0.43386717437563,
      "grad_norm": 1.255529522895813,
      "learning_rate": 3.872e-05,
      "loss": 0.4239,
      "step": 7748
    },
    {
      "epoch": 0.433923171687759,
      "grad_norm": 1.2360903024673462,
      "learning_rate": 3.8725e-05,
      "loss": 0.4332,
      "step": 7749
    },
    {
      "epoch": 0.43397916899988803,
      "grad_norm": 1.3143188953399658,
      "learning_rate": 3.873e-05,
      "loss": 0.4957,
      "step": 7750
    },
    {
      "epoch": 0.434035166312017,
      "grad_norm": 1.3152927160263062,
      "learning_rate": 3.873500000000001e-05,
      "loss": 0.4215,
      "step": 7751
    },
    {
      "epoch": 0.434091163624146,
      "grad_norm": 1.1224337816238403,
      "learning_rate": 3.8740000000000005e-05,
      "loss": 0.3767,
      "step": 7752
    },
    {
      "epoch": 0.43414716093627503,
      "grad_norm": 1.1857683658599854,
      "learning_rate": 3.8745e-05,
      "loss": 0.4179,
      "step": 7753
    },
    {
      "epoch": 0.43420315824840405,
      "grad_norm": 1.1691921949386597,
      "learning_rate": 3.875e-05,
      "loss": 0.3126,
      "step": 7754
    },
    {
      "epoch": 0.43425915556053307,
      "grad_norm": 1.1144546270370483,
      "learning_rate": 3.8755000000000004e-05,
      "loss": 0.4183,
      "step": 7755
    },
    {
      "epoch": 0.4343151528726621,
      "grad_norm": 1.328009009361267,
      "learning_rate": 3.876e-05,
      "loss": 0.5218,
      "step": 7756
    },
    {
      "epoch": 0.4343711501847911,
      "grad_norm": 1.1573644876480103,
      "learning_rate": 3.8765e-05,
      "loss": 0.3924,
      "step": 7757
    },
    {
      "epoch": 0.43442714749692013,
      "grad_norm": 1.4506826400756836,
      "learning_rate": 3.877e-05,
      "loss": 0.5673,
      "step": 7758
    },
    {
      "epoch": 0.43448314480904915,
      "grad_norm": 1.173905611038208,
      "learning_rate": 3.8775e-05,
      "loss": 0.4158,
      "step": 7759
    },
    {
      "epoch": 0.43453914212117817,
      "grad_norm": 1.3109742403030396,
      "learning_rate": 3.878e-05,
      "loss": 0.3942,
      "step": 7760
    },
    {
      "epoch": 0.4345951394333072,
      "grad_norm": 1.3459542989730835,
      "learning_rate": 3.8785e-05,
      "loss": 0.4062,
      "step": 7761
    },
    {
      "epoch": 0.4346511367454362,
      "grad_norm": 1.3858788013458252,
      "learning_rate": 3.8790000000000006e-05,
      "loss": 0.4552,
      "step": 7762
    },
    {
      "epoch": 0.4347071340575652,
      "grad_norm": 1.2540960311889648,
      "learning_rate": 3.8795000000000004e-05,
      "loss": 0.5135,
      "step": 7763
    },
    {
      "epoch": 0.43476313136969424,
      "grad_norm": 1.2480825185775757,
      "learning_rate": 3.88e-05,
      "loss": 0.5709,
      "step": 7764
    },
    {
      "epoch": 0.43481912868182326,
      "grad_norm": 1.3746126890182495,
      "learning_rate": 3.8805000000000005e-05,
      "loss": 0.349,
      "step": 7765
    },
    {
      "epoch": 0.4348751259939523,
      "grad_norm": 1.1591088771820068,
      "learning_rate": 3.881e-05,
      "loss": 0.3539,
      "step": 7766
    },
    {
      "epoch": 0.4349311233060813,
      "grad_norm": 1.2983176708221436,
      "learning_rate": 3.8815e-05,
      "loss": 0.5045,
      "step": 7767
    },
    {
      "epoch": 0.4349871206182103,
      "grad_norm": 1.2469282150268555,
      "learning_rate": 3.882e-05,
      "loss": 0.3675,
      "step": 7768
    },
    {
      "epoch": 0.43504311793033934,
      "grad_norm": 1.2342485189437866,
      "learning_rate": 3.8825e-05,
      "loss": 0.4768,
      "step": 7769
    },
    {
      "epoch": 0.43509911524246836,
      "grad_norm": 1.05726957321167,
      "learning_rate": 3.883e-05,
      "loss": 0.3183,
      "step": 7770
    },
    {
      "epoch": 0.4351551125545974,
      "grad_norm": 1.0779683589935303,
      "learning_rate": 3.8835e-05,
      "loss": 0.3627,
      "step": 7771
    },
    {
      "epoch": 0.4352111098667264,
      "grad_norm": 1.2601139545440674,
      "learning_rate": 3.884e-05,
      "loss": 0.3713,
      "step": 7772
    },
    {
      "epoch": 0.4352671071788554,
      "grad_norm": 1.1393673419952393,
      "learning_rate": 3.8845000000000005e-05,
      "loss": 0.3637,
      "step": 7773
    },
    {
      "epoch": 0.43532310449098444,
      "grad_norm": 1.3326117992401123,
      "learning_rate": 3.885e-05,
      "loss": 0.311,
      "step": 7774
    },
    {
      "epoch": 0.43537910180311346,
      "grad_norm": 1.242078423500061,
      "learning_rate": 3.8855e-05,
      "loss": 0.4623,
      "step": 7775
    },
    {
      "epoch": 0.4354350991152425,
      "grad_norm": 1.4016425609588623,
      "learning_rate": 3.8860000000000004e-05,
      "loss": 0.3817,
      "step": 7776
    },
    {
      "epoch": 0.4354910964273715,
      "grad_norm": 1.1242419481277466,
      "learning_rate": 3.8865e-05,
      "loss": 0.3917,
      "step": 7777
    },
    {
      "epoch": 0.4355470937395005,
      "grad_norm": 1.3407025337219238,
      "learning_rate": 3.887e-05,
      "loss": 0.3847,
      "step": 7778
    },
    {
      "epoch": 0.43560309105162953,
      "grad_norm": 1.1474888324737549,
      "learning_rate": 3.8875e-05,
      "loss": 0.3273,
      "step": 7779
    },
    {
      "epoch": 0.43565908836375855,
      "grad_norm": 1.417089819908142,
      "learning_rate": 3.888e-05,
      "loss": 0.5501,
      "step": 7780
    },
    {
      "epoch": 0.43571508567588757,
      "grad_norm": 1.3149428367614746,
      "learning_rate": 3.8885e-05,
      "loss": 0.6506,
      "step": 7781
    },
    {
      "epoch": 0.4357710829880166,
      "grad_norm": 1.202264666557312,
      "learning_rate": 3.889e-05,
      "loss": 0.5165,
      "step": 7782
    },
    {
      "epoch": 0.4358270803001456,
      "grad_norm": 1.2116482257843018,
      "learning_rate": 3.8895000000000006e-05,
      "loss": 0.4614,
      "step": 7783
    },
    {
      "epoch": 0.4358830776122746,
      "grad_norm": 1.240391492843628,
      "learning_rate": 3.8900000000000004e-05,
      "loss": 0.4139,
      "step": 7784
    },
    {
      "epoch": 0.43593907492440365,
      "grad_norm": 1.3046212196350098,
      "learning_rate": 3.8905e-05,
      "loss": 0.3925,
      "step": 7785
    },
    {
      "epoch": 0.43599507223653267,
      "grad_norm": 1.100469708442688,
      "learning_rate": 3.8910000000000005e-05,
      "loss": 0.452,
      "step": 7786
    },
    {
      "epoch": 0.4360510695486617,
      "grad_norm": 1.89679753780365,
      "learning_rate": 3.8915e-05,
      "loss": 0.685,
      "step": 7787
    },
    {
      "epoch": 0.4361070668607907,
      "grad_norm": 1.1426291465759277,
      "learning_rate": 3.892e-05,
      "loss": 0.4713,
      "step": 7788
    },
    {
      "epoch": 0.4361630641729197,
      "grad_norm": 1.0942741632461548,
      "learning_rate": 3.8925e-05,
      "loss": 0.363,
      "step": 7789
    },
    {
      "epoch": 0.43621906148504874,
      "grad_norm": 1.3065084218978882,
      "learning_rate": 3.893e-05,
      "loss": 0.435,
      "step": 7790
    },
    {
      "epoch": 0.43627505879717776,
      "grad_norm": 1.1707557439804077,
      "learning_rate": 3.8935e-05,
      "loss": 0.476,
      "step": 7791
    },
    {
      "epoch": 0.4363310561093068,
      "grad_norm": 1.4024662971496582,
      "learning_rate": 3.894e-05,
      "loss": 0.4503,
      "step": 7792
    },
    {
      "epoch": 0.43638705342143574,
      "grad_norm": 1.111179232597351,
      "learning_rate": 3.8945e-05,
      "loss": 0.4734,
      "step": 7793
    },
    {
      "epoch": 0.43644305073356476,
      "grad_norm": 1.2108148336410522,
      "learning_rate": 3.8950000000000005e-05,
      "loss": 0.4128,
      "step": 7794
    },
    {
      "epoch": 0.4364990480456938,
      "grad_norm": 1.275070071220398,
      "learning_rate": 3.8955e-05,
      "loss": 0.3925,
      "step": 7795
    },
    {
      "epoch": 0.4365550453578228,
      "grad_norm": 1.3724416494369507,
      "learning_rate": 3.896e-05,
      "loss": 0.4165,
      "step": 7796
    },
    {
      "epoch": 0.4366110426699518,
      "grad_norm": 1.3684978485107422,
      "learning_rate": 3.8965000000000004e-05,
      "loss": 0.4433,
      "step": 7797
    },
    {
      "epoch": 0.43666703998208084,
      "grad_norm": 1.1701958179473877,
      "learning_rate": 3.897e-05,
      "loss": 0.4411,
      "step": 7798
    },
    {
      "epoch": 0.43672303729420986,
      "grad_norm": 1.102452278137207,
      "learning_rate": 3.8975e-05,
      "loss": 0.392,
      "step": 7799
    },
    {
      "epoch": 0.4367790346063389,
      "grad_norm": 1.1435978412628174,
      "learning_rate": 3.898e-05,
      "loss": 0.3787,
      "step": 7800
    },
    {
      "epoch": 0.4368350319184679,
      "grad_norm": 1.4434677362442017,
      "learning_rate": 3.8985e-05,
      "loss": 0.5624,
      "step": 7801
    },
    {
      "epoch": 0.4368910292305969,
      "grad_norm": 1.0614144802093506,
      "learning_rate": 3.8990000000000004e-05,
      "loss": 0.3437,
      "step": 7802
    },
    {
      "epoch": 0.43694702654272594,
      "grad_norm": 1.2822693586349487,
      "learning_rate": 3.8995e-05,
      "loss": 0.4794,
      "step": 7803
    },
    {
      "epoch": 0.43700302385485496,
      "grad_norm": 1.189255714416504,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 0.3455,
      "step": 7804
    },
    {
      "epoch": 0.437059021166984,
      "grad_norm": 1.1027987003326416,
      "learning_rate": 3.9005000000000003e-05,
      "loss": 0.5032,
      "step": 7805
    },
    {
      "epoch": 0.437115018479113,
      "grad_norm": 1.2029386758804321,
      "learning_rate": 3.901e-05,
      "loss": 0.3261,
      "step": 7806
    },
    {
      "epoch": 0.437171015791242,
      "grad_norm": 1.1270817518234253,
      "learning_rate": 3.9015e-05,
      "loss": 0.393,
      "step": 7807
    },
    {
      "epoch": 0.43722701310337103,
      "grad_norm": 1.2903618812561035,
      "learning_rate": 3.902e-05,
      "loss": 0.3469,
      "step": 7808
    },
    {
      "epoch": 0.43728301041550005,
      "grad_norm": 1.3289403915405273,
      "learning_rate": 3.9025e-05,
      "loss": 0.4608,
      "step": 7809
    },
    {
      "epoch": 0.43733900772762907,
      "grad_norm": 1.1192606687545776,
      "learning_rate": 3.903e-05,
      "loss": 0.4093,
      "step": 7810
    },
    {
      "epoch": 0.4373950050397581,
      "grad_norm": 1.3133131265640259,
      "learning_rate": 3.9035e-05,
      "loss": 0.3993,
      "step": 7811
    },
    {
      "epoch": 0.4374510023518871,
      "grad_norm": 1.2583736181259155,
      "learning_rate": 3.9040000000000006e-05,
      "loss": 0.4404,
      "step": 7812
    },
    {
      "epoch": 0.4375069996640161,
      "grad_norm": 0.9704046845436096,
      "learning_rate": 3.9045e-05,
      "loss": 0.3489,
      "step": 7813
    },
    {
      "epoch": 0.43756299697614515,
      "grad_norm": 1.8509451150894165,
      "learning_rate": 3.905e-05,
      "loss": 0.4609,
      "step": 7814
    },
    {
      "epoch": 0.43761899428827417,
      "grad_norm": 1.298453688621521,
      "learning_rate": 3.9055000000000005e-05,
      "loss": 0.4883,
      "step": 7815
    },
    {
      "epoch": 0.4376749916004032,
      "grad_norm": 1.572828769683838,
      "learning_rate": 3.906e-05,
      "loss": 0.665,
      "step": 7816
    },
    {
      "epoch": 0.4377309889125322,
      "grad_norm": 1.1312446594238281,
      "learning_rate": 3.9065e-05,
      "loss": 0.368,
      "step": 7817
    },
    {
      "epoch": 0.4377869862246612,
      "grad_norm": 1.2823508977890015,
      "learning_rate": 3.9070000000000004e-05,
      "loss": 0.4562,
      "step": 7818
    },
    {
      "epoch": 0.43784298353679024,
      "grad_norm": 1.3399925231933594,
      "learning_rate": 3.9075e-05,
      "loss": 0.4222,
      "step": 7819
    },
    {
      "epoch": 0.43789898084891926,
      "grad_norm": 1.2265499830245972,
      "learning_rate": 3.908e-05,
      "loss": 0.3734,
      "step": 7820
    },
    {
      "epoch": 0.4379549781610483,
      "grad_norm": 1.4076920747756958,
      "learning_rate": 3.9085e-05,
      "loss": 0.3937,
      "step": 7821
    },
    {
      "epoch": 0.4380109754731773,
      "grad_norm": 1.2663395404815674,
      "learning_rate": 3.909000000000001e-05,
      "loss": 0.3566,
      "step": 7822
    },
    {
      "epoch": 0.4380669727853063,
      "grad_norm": 1.4407652616500854,
      "learning_rate": 3.9095000000000004e-05,
      "loss": 0.4386,
      "step": 7823
    },
    {
      "epoch": 0.43812297009743534,
      "grad_norm": 1.1476681232452393,
      "learning_rate": 3.91e-05,
      "loss": 0.3476,
      "step": 7824
    },
    {
      "epoch": 0.43817896740956436,
      "grad_norm": 1.0374912023544312,
      "learning_rate": 3.9105000000000006e-05,
      "loss": 0.3338,
      "step": 7825
    },
    {
      "epoch": 0.4382349647216934,
      "grad_norm": 1.0831481218338013,
      "learning_rate": 3.911e-05,
      "loss": 0.4214,
      "step": 7826
    },
    {
      "epoch": 0.4382909620338224,
      "grad_norm": 1.1083035469055176,
      "learning_rate": 3.9115e-05,
      "loss": 0.318,
      "step": 7827
    },
    {
      "epoch": 0.4383469593459514,
      "grad_norm": 1.2883092164993286,
      "learning_rate": 3.912e-05,
      "loss": 0.4401,
      "step": 7828
    },
    {
      "epoch": 0.43840295665808043,
      "grad_norm": 1.313269019126892,
      "learning_rate": 3.9125e-05,
      "loss": 0.4767,
      "step": 7829
    },
    {
      "epoch": 0.43845895397020945,
      "grad_norm": 1.169537901878357,
      "learning_rate": 3.913e-05,
      "loss": 0.3928,
      "step": 7830
    },
    {
      "epoch": 0.43851495128233847,
      "grad_norm": 1.2053309679031372,
      "learning_rate": 3.9135e-05,
      "loss": 0.4535,
      "step": 7831
    },
    {
      "epoch": 0.4385709485944675,
      "grad_norm": 1.3004990816116333,
      "learning_rate": 3.914e-05,
      "loss": 0.4291,
      "step": 7832
    },
    {
      "epoch": 0.4386269459065965,
      "grad_norm": 1.3423177003860474,
      "learning_rate": 3.9145000000000006e-05,
      "loss": 0.3814,
      "step": 7833
    },
    {
      "epoch": 0.4386829432187255,
      "grad_norm": 1.397086262702942,
      "learning_rate": 3.915e-05,
      "loss": 0.423,
      "step": 7834
    },
    {
      "epoch": 0.4387389405308545,
      "grad_norm": 1.1477322578430176,
      "learning_rate": 3.9155e-05,
      "loss": 0.451,
      "step": 7835
    },
    {
      "epoch": 0.4387949378429835,
      "grad_norm": 1.1389315128326416,
      "learning_rate": 3.9160000000000005e-05,
      "loss": 0.4046,
      "step": 7836
    },
    {
      "epoch": 0.43885093515511253,
      "grad_norm": 1.3055751323699951,
      "learning_rate": 3.9165e-05,
      "loss": 0.4471,
      "step": 7837
    },
    {
      "epoch": 0.43890693246724155,
      "grad_norm": 1.1227158308029175,
      "learning_rate": 3.917e-05,
      "loss": 0.3757,
      "step": 7838
    },
    {
      "epoch": 0.43896292977937057,
      "grad_norm": 1.2383346557617188,
      "learning_rate": 3.9175000000000004e-05,
      "loss": 0.4505,
      "step": 7839
    },
    {
      "epoch": 0.4390189270914996,
      "grad_norm": 1.3493623733520508,
      "learning_rate": 3.918e-05,
      "loss": 0.5136,
      "step": 7840
    },
    {
      "epoch": 0.4390749244036286,
      "grad_norm": 1.250669240951538,
      "learning_rate": 3.9185e-05,
      "loss": 0.2715,
      "step": 7841
    },
    {
      "epoch": 0.4391309217157576,
      "grad_norm": 1.2727452516555786,
      "learning_rate": 3.919e-05,
      "loss": 0.4297,
      "step": 7842
    },
    {
      "epoch": 0.43918691902788665,
      "grad_norm": 1.1866668462753296,
      "learning_rate": 3.919500000000001e-05,
      "loss": 0.3604,
      "step": 7843
    },
    {
      "epoch": 0.43924291634001567,
      "grad_norm": 1.1657792329788208,
      "learning_rate": 3.9200000000000004e-05,
      "loss": 0.3871,
      "step": 7844
    },
    {
      "epoch": 0.4392989136521447,
      "grad_norm": 1.3171136379241943,
      "learning_rate": 3.9205e-05,
      "loss": 0.4378,
      "step": 7845
    },
    {
      "epoch": 0.4393549109642737,
      "grad_norm": 1.1899813413619995,
      "learning_rate": 3.921e-05,
      "loss": 0.3927,
      "step": 7846
    },
    {
      "epoch": 0.4394109082764027,
      "grad_norm": 1.1432589292526245,
      "learning_rate": 3.9215e-05,
      "loss": 0.3734,
      "step": 7847
    },
    {
      "epoch": 0.43946690558853174,
      "grad_norm": 1.1378377676010132,
      "learning_rate": 3.922e-05,
      "loss": 0.3443,
      "step": 7848
    },
    {
      "epoch": 0.43952290290066076,
      "grad_norm": 1.4463274478912354,
      "learning_rate": 3.9225e-05,
      "loss": 0.5706,
      "step": 7849
    },
    {
      "epoch": 0.4395789002127898,
      "grad_norm": 1.6434000730514526,
      "learning_rate": 3.923e-05,
      "loss": 0.5424,
      "step": 7850
    },
    {
      "epoch": 0.4396348975249188,
      "grad_norm": 1.0726593732833862,
      "learning_rate": 3.9235e-05,
      "loss": 0.3867,
      "step": 7851
    },
    {
      "epoch": 0.4396908948370478,
      "grad_norm": 1.233354926109314,
      "learning_rate": 3.9240000000000004e-05,
      "loss": 0.4477,
      "step": 7852
    },
    {
      "epoch": 0.43974689214917684,
      "grad_norm": 1.4947172403335571,
      "learning_rate": 3.9245e-05,
      "loss": 0.3989,
      "step": 7853
    },
    {
      "epoch": 0.43980288946130586,
      "grad_norm": 1.1785705089569092,
      "learning_rate": 3.9250000000000005e-05,
      "loss": 0.343,
      "step": 7854
    },
    {
      "epoch": 0.4398588867734349,
      "grad_norm": 1.3784432411193848,
      "learning_rate": 3.9255e-05,
      "loss": 0.4166,
      "step": 7855
    },
    {
      "epoch": 0.4399148840855639,
      "grad_norm": 1.8765945434570312,
      "learning_rate": 3.926e-05,
      "loss": 0.4711,
      "step": 7856
    },
    {
      "epoch": 0.4399708813976929,
      "grad_norm": 1.1780389547348022,
      "learning_rate": 3.9265000000000004e-05,
      "loss": 0.437,
      "step": 7857
    },
    {
      "epoch": 0.44002687870982193,
      "grad_norm": 1.1090939044952393,
      "learning_rate": 3.927e-05,
      "loss": 0.3735,
      "step": 7858
    },
    {
      "epoch": 0.44008287602195095,
      "grad_norm": 1.3783758878707886,
      "learning_rate": 3.9275e-05,
      "loss": 0.521,
      "step": 7859
    },
    {
      "epoch": 0.44013887333407997,
      "grad_norm": 1.1683554649353027,
      "learning_rate": 3.9280000000000003e-05,
      "loss": 0.4162,
      "step": 7860
    },
    {
      "epoch": 0.440194870646209,
      "grad_norm": 1.0943423509597778,
      "learning_rate": 3.9285e-05,
      "loss": 0.5115,
      "step": 7861
    },
    {
      "epoch": 0.440250867958338,
      "grad_norm": 1.2970889806747437,
      "learning_rate": 3.9290000000000005e-05,
      "loss": 0.5817,
      "step": 7862
    },
    {
      "epoch": 0.44030686527046703,
      "grad_norm": 1.1418607234954834,
      "learning_rate": 3.9295e-05,
      "loss": 0.3782,
      "step": 7863
    },
    {
      "epoch": 0.44036286258259605,
      "grad_norm": 1.0735609531402588,
      "learning_rate": 3.9300000000000007e-05,
      "loss": 0.3981,
      "step": 7864
    },
    {
      "epoch": 0.44041885989472507,
      "grad_norm": 1.6328279972076416,
      "learning_rate": 3.9305000000000004e-05,
      "loss": 0.4539,
      "step": 7865
    },
    {
      "epoch": 0.4404748572068541,
      "grad_norm": 1.295376181602478,
      "learning_rate": 3.931e-05,
      "loss": 0.46,
      "step": 7866
    },
    {
      "epoch": 0.4405308545189831,
      "grad_norm": 1.2095136642456055,
      "learning_rate": 3.9315e-05,
      "loss": 0.3505,
      "step": 7867
    },
    {
      "epoch": 0.4405868518311121,
      "grad_norm": 1.2632331848144531,
      "learning_rate": 3.932e-05,
      "loss": 0.5245,
      "step": 7868
    },
    {
      "epoch": 0.44064284914324114,
      "grad_norm": 1.3711323738098145,
      "learning_rate": 3.9325e-05,
      "loss": 0.4808,
      "step": 7869
    },
    {
      "epoch": 0.44069884645537016,
      "grad_norm": 1.4462391138076782,
      "learning_rate": 3.933e-05,
      "loss": 0.4866,
      "step": 7870
    },
    {
      "epoch": 0.4407548437674992,
      "grad_norm": 1.1605569124221802,
      "learning_rate": 3.9335e-05,
      "loss": 0.4799,
      "step": 7871
    },
    {
      "epoch": 0.4408108410796282,
      "grad_norm": 1.3611857891082764,
      "learning_rate": 3.9340000000000006e-05,
      "loss": 0.314,
      "step": 7872
    },
    {
      "epoch": 0.4408668383917572,
      "grad_norm": 1.3676567077636719,
      "learning_rate": 3.9345000000000004e-05,
      "loss": 0.4342,
      "step": 7873
    },
    {
      "epoch": 0.44092283570388624,
      "grad_norm": 1.602771282196045,
      "learning_rate": 3.935e-05,
      "loss": 0.5213,
      "step": 7874
    },
    {
      "epoch": 0.4409788330160152,
      "grad_norm": 1.1253604888916016,
      "learning_rate": 3.9355000000000005e-05,
      "loss": 0.3175,
      "step": 7875
    },
    {
      "epoch": 0.4410348303281442,
      "grad_norm": 1.1032730340957642,
      "learning_rate": 3.936e-05,
      "loss": 0.3832,
      "step": 7876
    },
    {
      "epoch": 0.44109082764027324,
      "grad_norm": 1.2083414793014526,
      "learning_rate": 3.9365e-05,
      "loss": 0.321,
      "step": 7877
    },
    {
      "epoch": 0.44114682495240226,
      "grad_norm": 1.108339548110962,
      "learning_rate": 3.9370000000000004e-05,
      "loss": 0.3668,
      "step": 7878
    },
    {
      "epoch": 0.4412028222645313,
      "grad_norm": 1.3832827806472778,
      "learning_rate": 3.9375e-05,
      "loss": 0.3733,
      "step": 7879
    },
    {
      "epoch": 0.4412588195766603,
      "grad_norm": 1.0996882915496826,
      "learning_rate": 3.938e-05,
      "loss": 0.365,
      "step": 7880
    },
    {
      "epoch": 0.4413148168887893,
      "grad_norm": 1.3138083219528198,
      "learning_rate": 3.9384999999999996e-05,
      "loss": 0.46,
      "step": 7881
    },
    {
      "epoch": 0.44137081420091834,
      "grad_norm": 2.8446733951568604,
      "learning_rate": 3.939e-05,
      "loss": 0.4353,
      "step": 7882
    },
    {
      "epoch": 0.44142681151304736,
      "grad_norm": 1.1881431341171265,
      "learning_rate": 3.9395000000000005e-05,
      "loss": 0.4411,
      "step": 7883
    },
    {
      "epoch": 0.4414828088251764,
      "grad_norm": 1.2280796766281128,
      "learning_rate": 3.94e-05,
      "loss": 0.568,
      "step": 7884
    },
    {
      "epoch": 0.4415388061373054,
      "grad_norm": 1.5344942808151245,
      "learning_rate": 3.9405e-05,
      "loss": 0.381,
      "step": 7885
    },
    {
      "epoch": 0.4415948034494344,
      "grad_norm": 1.349732756614685,
      "learning_rate": 3.9410000000000004e-05,
      "loss": 0.4975,
      "step": 7886
    },
    {
      "epoch": 0.44165080076156343,
      "grad_norm": 1.2575490474700928,
      "learning_rate": 3.9415e-05,
      "loss": 0.4531,
      "step": 7887
    },
    {
      "epoch": 0.44170679807369245,
      "grad_norm": 1.1180615425109863,
      "learning_rate": 3.942e-05,
      "loss": 0.4269,
      "step": 7888
    },
    {
      "epoch": 0.44176279538582147,
      "grad_norm": 1.3497393131256104,
      "learning_rate": 3.9425e-05,
      "loss": 0.5359,
      "step": 7889
    },
    {
      "epoch": 0.4418187926979505,
      "grad_norm": 1.1241097450256348,
      "learning_rate": 3.943e-05,
      "loss": 0.463,
      "step": 7890
    },
    {
      "epoch": 0.4418747900100795,
      "grad_norm": 1.443230390548706,
      "learning_rate": 3.9435e-05,
      "loss": 0.6691,
      "step": 7891
    },
    {
      "epoch": 0.44193078732220853,
      "grad_norm": 1.1160078048706055,
      "learning_rate": 3.944e-05,
      "loss": 0.3997,
      "step": 7892
    },
    {
      "epoch": 0.44198678463433755,
      "grad_norm": 1.2423415184020996,
      "learning_rate": 3.9445000000000006e-05,
      "loss": 0.4094,
      "step": 7893
    },
    {
      "epoch": 0.44204278194646657,
      "grad_norm": 1.3356132507324219,
      "learning_rate": 3.9450000000000003e-05,
      "loss": 0.4605,
      "step": 7894
    },
    {
      "epoch": 0.4420987792585956,
      "grad_norm": 1.4842458963394165,
      "learning_rate": 3.9455e-05,
      "loss": 0.4993,
      "step": 7895
    },
    {
      "epoch": 0.4421547765707246,
      "grad_norm": 1.5475319623947144,
      "learning_rate": 3.9460000000000005e-05,
      "loss": 0.3421,
      "step": 7896
    },
    {
      "epoch": 0.4422107738828536,
      "grad_norm": 1.338922381401062,
      "learning_rate": 3.9465e-05,
      "loss": 0.4591,
      "step": 7897
    },
    {
      "epoch": 0.44226677119498264,
      "grad_norm": 1.264709234237671,
      "learning_rate": 3.947e-05,
      "loss": 0.5858,
      "step": 7898
    },
    {
      "epoch": 0.44232276850711166,
      "grad_norm": 1.154921293258667,
      "learning_rate": 3.9475000000000004e-05,
      "loss": 0.4254,
      "step": 7899
    },
    {
      "epoch": 0.4423787658192407,
      "grad_norm": 1.3502569198608398,
      "learning_rate": 3.948e-05,
      "loss": 0.4737,
      "step": 7900
    },
    {
      "epoch": 0.4424347631313697,
      "grad_norm": 1.159049153327942,
      "learning_rate": 3.9485e-05,
      "loss": 0.4563,
      "step": 7901
    },
    {
      "epoch": 0.4424907604434987,
      "grad_norm": 1.2016639709472656,
      "learning_rate": 3.9489999999999996e-05,
      "loss": 0.5038,
      "step": 7902
    },
    {
      "epoch": 0.44254675775562774,
      "grad_norm": 1.406581997871399,
      "learning_rate": 3.949500000000001e-05,
      "loss": 0.5906,
      "step": 7903
    },
    {
      "epoch": 0.44260275506775676,
      "grad_norm": 1.1897046566009521,
      "learning_rate": 3.9500000000000005e-05,
      "loss": 0.4028,
      "step": 7904
    },
    {
      "epoch": 0.4426587523798858,
      "grad_norm": 1.1853468418121338,
      "learning_rate": 3.9505e-05,
      "loss": 0.3767,
      "step": 7905
    },
    {
      "epoch": 0.4427147496920148,
      "grad_norm": 1.3921821117401123,
      "learning_rate": 3.951e-05,
      "loss": 0.4843,
      "step": 7906
    },
    {
      "epoch": 0.4427707470041438,
      "grad_norm": 1.3302898406982422,
      "learning_rate": 3.9515000000000004e-05,
      "loss": 0.4561,
      "step": 7907
    },
    {
      "epoch": 0.44282674431627284,
      "grad_norm": 1.3814650774002075,
      "learning_rate": 3.952e-05,
      "loss": 0.4598,
      "step": 7908
    },
    {
      "epoch": 0.44288274162840185,
      "grad_norm": 1.281061053276062,
      "learning_rate": 3.9525e-05,
      "loss": 0.4592,
      "step": 7909
    },
    {
      "epoch": 0.4429387389405309,
      "grad_norm": 1.0565319061279297,
      "learning_rate": 3.953e-05,
      "loss": 0.3522,
      "step": 7910
    },
    {
      "epoch": 0.4429947362526599,
      "grad_norm": 1.2696032524108887,
      "learning_rate": 3.9535e-05,
      "loss": 0.4059,
      "step": 7911
    },
    {
      "epoch": 0.4430507335647889,
      "grad_norm": 1.0670174360275269,
      "learning_rate": 3.954e-05,
      "loss": 0.3732,
      "step": 7912
    },
    {
      "epoch": 0.44310673087691793,
      "grad_norm": 1.371474027633667,
      "learning_rate": 3.9545e-05,
      "loss": 0.4391,
      "step": 7913
    },
    {
      "epoch": 0.44316272818904695,
      "grad_norm": 1.1317119598388672,
      "learning_rate": 3.9550000000000006e-05,
      "loss": 0.3645,
      "step": 7914
    },
    {
      "epoch": 0.44321872550117597,
      "grad_norm": 1.0873996019363403,
      "learning_rate": 3.9555e-05,
      "loss": 0.4618,
      "step": 7915
    },
    {
      "epoch": 0.443274722813305,
      "grad_norm": 1.397986888885498,
      "learning_rate": 3.956e-05,
      "loss": 0.539,
      "step": 7916
    },
    {
      "epoch": 0.44333072012543395,
      "grad_norm": 1.3856488466262817,
      "learning_rate": 3.9565000000000005e-05,
      "loss": 0.6002,
      "step": 7917
    },
    {
      "epoch": 0.44338671743756297,
      "grad_norm": 1.1605008840560913,
      "learning_rate": 3.957e-05,
      "loss": 0.3676,
      "step": 7918
    },
    {
      "epoch": 0.443442714749692,
      "grad_norm": 1.2629355192184448,
      "learning_rate": 3.9575e-05,
      "loss": 0.3796,
      "step": 7919
    },
    {
      "epoch": 0.443498712061821,
      "grad_norm": 1.28914213180542,
      "learning_rate": 3.958e-05,
      "loss": 0.4858,
      "step": 7920
    },
    {
      "epoch": 0.44355470937395003,
      "grad_norm": 1.1475467681884766,
      "learning_rate": 3.9585e-05,
      "loss": 0.3757,
      "step": 7921
    },
    {
      "epoch": 0.44361070668607905,
      "grad_norm": 1.3344305753707886,
      "learning_rate": 3.959e-05,
      "loss": 0.5062,
      "step": 7922
    },
    {
      "epoch": 0.44366670399820807,
      "grad_norm": 1.1459726095199585,
      "learning_rate": 3.9595e-05,
      "loss": 0.3602,
      "step": 7923
    },
    {
      "epoch": 0.4437227013103371,
      "grad_norm": 1.5019686222076416,
      "learning_rate": 3.960000000000001e-05,
      "loss": 0.4824,
      "step": 7924
    },
    {
      "epoch": 0.4437786986224661,
      "grad_norm": 1.6637812852859497,
      "learning_rate": 3.9605000000000005e-05,
      "loss": 0.5459,
      "step": 7925
    },
    {
      "epoch": 0.4438346959345951,
      "grad_norm": 1.2453186511993408,
      "learning_rate": 3.961e-05,
      "loss": 0.3356,
      "step": 7926
    },
    {
      "epoch": 0.44389069324672414,
      "grad_norm": 1.4240748882293701,
      "learning_rate": 3.9615e-05,
      "loss": 0.3513,
      "step": 7927
    },
    {
      "epoch": 0.44394669055885316,
      "grad_norm": 1.5543899536132812,
      "learning_rate": 3.9620000000000004e-05,
      "loss": 0.4535,
      "step": 7928
    },
    {
      "epoch": 0.4440026878709822,
      "grad_norm": 1.4476829767227173,
      "learning_rate": 3.9625e-05,
      "loss": 0.4958,
      "step": 7929
    },
    {
      "epoch": 0.4440586851831112,
      "grad_norm": 1.2234768867492676,
      "learning_rate": 3.963e-05,
      "loss": 0.3737,
      "step": 7930
    },
    {
      "epoch": 0.4441146824952402,
      "grad_norm": 1.4079090356826782,
      "learning_rate": 3.9635e-05,
      "loss": 0.4697,
      "step": 7931
    },
    {
      "epoch": 0.44417067980736924,
      "grad_norm": 1.09403657913208,
      "learning_rate": 3.964e-05,
      "loss": 0.3721,
      "step": 7932
    },
    {
      "epoch": 0.44422667711949826,
      "grad_norm": 1.1749703884124756,
      "learning_rate": 3.9645000000000004e-05,
      "loss": 0.3767,
      "step": 7933
    },
    {
      "epoch": 0.4442826744316273,
      "grad_norm": 1.252769112586975,
      "learning_rate": 3.965e-05,
      "loss": 0.3794,
      "step": 7934
    },
    {
      "epoch": 0.4443386717437563,
      "grad_norm": 1.4599592685699463,
      "learning_rate": 3.9655000000000006e-05,
      "loss": 0.394,
      "step": 7935
    },
    {
      "epoch": 0.4443946690558853,
      "grad_norm": 1.324910044670105,
      "learning_rate": 3.966e-05,
      "loss": 0.4233,
      "step": 7936
    },
    {
      "epoch": 0.44445066636801434,
      "grad_norm": 1.1552929878234863,
      "learning_rate": 3.9665e-05,
      "loss": 0.482,
      "step": 7937
    },
    {
      "epoch": 0.44450666368014335,
      "grad_norm": 1.4286192655563354,
      "learning_rate": 3.9670000000000005e-05,
      "loss": 0.3971,
      "step": 7938
    },
    {
      "epoch": 0.4445626609922724,
      "grad_norm": 1.4314066171646118,
      "learning_rate": 3.9675e-05,
      "loss": 0.5845,
      "step": 7939
    },
    {
      "epoch": 0.4446186583044014,
      "grad_norm": 1.1849724054336548,
      "learning_rate": 3.968e-05,
      "loss": 0.3866,
      "step": 7940
    },
    {
      "epoch": 0.4446746556165304,
      "grad_norm": 1.1440964937210083,
      "learning_rate": 3.9685e-05,
      "loss": 0.5633,
      "step": 7941
    },
    {
      "epoch": 0.44473065292865943,
      "grad_norm": 1.0037082433700562,
      "learning_rate": 3.969e-05,
      "loss": 0.3313,
      "step": 7942
    },
    {
      "epoch": 0.44478665024078845,
      "grad_norm": 1.0694080591201782,
      "learning_rate": 3.9695000000000005e-05,
      "loss": 0.4466,
      "step": 7943
    },
    {
      "epoch": 0.44484264755291747,
      "grad_norm": 1.5107334852218628,
      "learning_rate": 3.97e-05,
      "loss": 0.5122,
      "step": 7944
    },
    {
      "epoch": 0.4448986448650465,
      "grad_norm": 1.397767424583435,
      "learning_rate": 3.9705e-05,
      "loss": 0.4112,
      "step": 7945
    },
    {
      "epoch": 0.4449546421771755,
      "grad_norm": 1.3833537101745605,
      "learning_rate": 3.9710000000000004e-05,
      "loss": 0.3841,
      "step": 7946
    },
    {
      "epoch": 0.4450106394893045,
      "grad_norm": 1.301817536354065,
      "learning_rate": 3.9715e-05,
      "loss": 0.5478,
      "step": 7947
    },
    {
      "epoch": 0.44506663680143355,
      "grad_norm": 1.5289353132247925,
      "learning_rate": 3.972e-05,
      "loss": 0.4269,
      "step": 7948
    },
    {
      "epoch": 0.44512263411356257,
      "grad_norm": 1.3700004816055298,
      "learning_rate": 3.9725e-05,
      "loss": 0.5458,
      "step": 7949
    },
    {
      "epoch": 0.4451786314256916,
      "grad_norm": 1.2713686227798462,
      "learning_rate": 3.973e-05,
      "loss": 0.4822,
      "step": 7950
    },
    {
      "epoch": 0.4452346287378206,
      "grad_norm": 1.1312288045883179,
      "learning_rate": 3.9735e-05,
      "loss": 0.3152,
      "step": 7951
    },
    {
      "epoch": 0.4452906260499496,
      "grad_norm": 1.3867393732070923,
      "learning_rate": 3.974e-05,
      "loss": 0.3717,
      "step": 7952
    },
    {
      "epoch": 0.44534662336207864,
      "grad_norm": 1.2697700262069702,
      "learning_rate": 3.9745000000000007e-05,
      "loss": 0.659,
      "step": 7953
    },
    {
      "epoch": 0.44540262067420766,
      "grad_norm": 1.2105894088745117,
      "learning_rate": 3.9750000000000004e-05,
      "loss": 0.4468,
      "step": 7954
    },
    {
      "epoch": 0.4454586179863367,
      "grad_norm": 1.2118644714355469,
      "learning_rate": 3.9755e-05,
      "loss": 0.3919,
      "step": 7955
    },
    {
      "epoch": 0.4455146152984657,
      "grad_norm": 1.1268454790115356,
      "learning_rate": 3.9760000000000006e-05,
      "loss": 0.3192,
      "step": 7956
    },
    {
      "epoch": 0.4455706126105947,
      "grad_norm": 1.1013085842132568,
      "learning_rate": 3.9765e-05,
      "loss": 0.3714,
      "step": 7957
    },
    {
      "epoch": 0.4456266099227237,
      "grad_norm": 1.3739217519760132,
      "learning_rate": 3.977e-05,
      "loss": 0.4496,
      "step": 7958
    },
    {
      "epoch": 0.4456826072348527,
      "grad_norm": 0.9961612224578857,
      "learning_rate": 3.9775e-05,
      "loss": 0.3866,
      "step": 7959
    },
    {
      "epoch": 0.4457386045469817,
      "grad_norm": 1.3658822774887085,
      "learning_rate": 3.978e-05,
      "loss": 0.4917,
      "step": 7960
    },
    {
      "epoch": 0.44579460185911074,
      "grad_norm": 1.3278608322143555,
      "learning_rate": 3.9785e-05,
      "loss": 0.3447,
      "step": 7961
    },
    {
      "epoch": 0.44585059917123976,
      "grad_norm": 1.2886418104171753,
      "learning_rate": 3.979e-05,
      "loss": 0.4889,
      "step": 7962
    },
    {
      "epoch": 0.4459065964833688,
      "grad_norm": 1.2320998907089233,
      "learning_rate": 3.979500000000001e-05,
      "loss": 0.4547,
      "step": 7963
    },
    {
      "epoch": 0.4459625937954978,
      "grad_norm": 1.3106369972229004,
      "learning_rate": 3.9800000000000005e-05,
      "loss": 0.4884,
      "step": 7964
    },
    {
      "epoch": 0.4460185911076268,
      "grad_norm": 1.3171268701553345,
      "learning_rate": 3.9805e-05,
      "loss": 0.2768,
      "step": 7965
    },
    {
      "epoch": 0.44607458841975584,
      "grad_norm": 1.265672206878662,
      "learning_rate": 3.981e-05,
      "loss": 0.5514,
      "step": 7966
    },
    {
      "epoch": 0.44613058573188485,
      "grad_norm": 1.2978721857070923,
      "learning_rate": 3.9815000000000004e-05,
      "loss": 0.4264,
      "step": 7967
    },
    {
      "epoch": 0.4461865830440139,
      "grad_norm": 1.543387770652771,
      "learning_rate": 3.982e-05,
      "loss": 0.5642,
      "step": 7968
    },
    {
      "epoch": 0.4462425803561429,
      "grad_norm": 1.3579707145690918,
      "learning_rate": 3.9825e-05,
      "loss": 0.3669,
      "step": 7969
    },
    {
      "epoch": 0.4462985776682719,
      "grad_norm": 1.495080828666687,
      "learning_rate": 3.983e-05,
      "loss": 0.4153,
      "step": 7970
    },
    {
      "epoch": 0.44635457498040093,
      "grad_norm": 1.2408615350723267,
      "learning_rate": 3.9835e-05,
      "loss": 0.4271,
      "step": 7971
    },
    {
      "epoch": 0.44641057229252995,
      "grad_norm": 1.266782283782959,
      "learning_rate": 3.984e-05,
      "loss": 0.3797,
      "step": 7972
    },
    {
      "epoch": 0.44646656960465897,
      "grad_norm": 1.8094933032989502,
      "learning_rate": 3.9845e-05,
      "loss": 0.582,
      "step": 7973
    },
    {
      "epoch": 0.446522566916788,
      "grad_norm": 1.3468682765960693,
      "learning_rate": 3.9850000000000006e-05,
      "loss": 0.473,
      "step": 7974
    },
    {
      "epoch": 0.446578564228917,
      "grad_norm": 1.3554326295852661,
      "learning_rate": 3.9855000000000004e-05,
      "loss": 0.5442,
      "step": 7975
    },
    {
      "epoch": 0.446634561541046,
      "grad_norm": 1.2305324077606201,
      "learning_rate": 3.986e-05,
      "loss": 0.4173,
      "step": 7976
    },
    {
      "epoch": 0.44669055885317505,
      "grad_norm": 0.9923909902572632,
      "learning_rate": 3.9865000000000005e-05,
      "loss": 0.3104,
      "step": 7977
    },
    {
      "epoch": 0.44674655616530407,
      "grad_norm": 1.1111879348754883,
      "learning_rate": 3.987e-05,
      "loss": 0.4633,
      "step": 7978
    },
    {
      "epoch": 0.4468025534774331,
      "grad_norm": 1.341869592666626,
      "learning_rate": 3.9875e-05,
      "loss": 0.3594,
      "step": 7979
    },
    {
      "epoch": 0.4468585507895621,
      "grad_norm": 1.3253010511398315,
      "learning_rate": 3.988e-05,
      "loss": 0.5988,
      "step": 7980
    },
    {
      "epoch": 0.4469145481016911,
      "grad_norm": 1.386620044708252,
      "learning_rate": 3.9885e-05,
      "loss": 0.4883,
      "step": 7981
    },
    {
      "epoch": 0.44697054541382014,
      "grad_norm": 1.2293016910552979,
      "learning_rate": 3.989e-05,
      "loss": 0.4162,
      "step": 7982
    },
    {
      "epoch": 0.44702654272594916,
      "grad_norm": 1.0457274913787842,
      "learning_rate": 3.9895000000000003e-05,
      "loss": 0.3897,
      "step": 7983
    },
    {
      "epoch": 0.4470825400380782,
      "grad_norm": 1.4893736839294434,
      "learning_rate": 3.99e-05,
      "loss": 0.4356,
      "step": 7984
    },
    {
      "epoch": 0.4471385373502072,
      "grad_norm": 1.1966415643692017,
      "learning_rate": 3.9905000000000005e-05,
      "loss": 0.3838,
      "step": 7985
    },
    {
      "epoch": 0.4471945346623362,
      "grad_norm": 1.4571969509124756,
      "learning_rate": 3.991e-05,
      "loss": 0.5493,
      "step": 7986
    },
    {
      "epoch": 0.44725053197446524,
      "grad_norm": 1.3030469417572021,
      "learning_rate": 3.9915e-05,
      "loss": 0.6022,
      "step": 7987
    },
    {
      "epoch": 0.44730652928659426,
      "grad_norm": 1.2906101942062378,
      "learning_rate": 3.9920000000000004e-05,
      "loss": 0.6794,
      "step": 7988
    },
    {
      "epoch": 0.4473625265987233,
      "grad_norm": 1.244630217552185,
      "learning_rate": 3.9925e-05,
      "loss": 0.3808,
      "step": 7989
    },
    {
      "epoch": 0.4474185239108523,
      "grad_norm": 1.330586314201355,
      "learning_rate": 3.993e-05,
      "loss": 0.5,
      "step": 7990
    },
    {
      "epoch": 0.4474745212229813,
      "grad_norm": 1.4680848121643066,
      "learning_rate": 3.9935e-05,
      "loss": 0.4274,
      "step": 7991
    },
    {
      "epoch": 0.44753051853511033,
      "grad_norm": 1.2660781145095825,
      "learning_rate": 3.994e-05,
      "loss": 0.4114,
      "step": 7992
    },
    {
      "epoch": 0.44758651584723935,
      "grad_norm": 1.614398717880249,
      "learning_rate": 3.9945000000000005e-05,
      "loss": 0.5062,
      "step": 7993
    },
    {
      "epoch": 0.44764251315936837,
      "grad_norm": 1.0764151811599731,
      "learning_rate": 3.995e-05,
      "loss": 0.3746,
      "step": 7994
    },
    {
      "epoch": 0.4476985104714974,
      "grad_norm": 1.3683723211288452,
      "learning_rate": 3.9955000000000006e-05,
      "loss": 0.5567,
      "step": 7995
    },
    {
      "epoch": 0.4477545077836264,
      "grad_norm": 1.0107738971710205,
      "learning_rate": 3.9960000000000004e-05,
      "loss": 0.3768,
      "step": 7996
    },
    {
      "epoch": 0.44781050509575543,
      "grad_norm": 1.3045953512191772,
      "learning_rate": 3.9965e-05,
      "loss": 0.5264,
      "step": 7997
    },
    {
      "epoch": 0.44786650240788445,
      "grad_norm": 1.2756482362747192,
      "learning_rate": 3.9970000000000005e-05,
      "loss": 0.4646,
      "step": 7998
    },
    {
      "epoch": 0.4479224997200134,
      "grad_norm": 1.2149758338928223,
      "learning_rate": 3.9975e-05,
      "loss": 0.4615,
      "step": 7999
    },
    {
      "epoch": 0.44797849703214243,
      "grad_norm": 1.2461477518081665,
      "learning_rate": 3.998e-05,
      "loss": 0.4821,
      "step": 8000
    },
    {
      "epoch": 0.44803449434427145,
      "grad_norm": 1.3961334228515625,
      "learning_rate": 3.9985e-05,
      "loss": 0.4495,
      "step": 8001
    },
    {
      "epoch": 0.44809049165640047,
      "grad_norm": 1.2145496606826782,
      "learning_rate": 3.999e-05,
      "loss": 0.3813,
      "step": 8002
    },
    {
      "epoch": 0.4481464889685295,
      "grad_norm": 1.1947134733200073,
      "learning_rate": 3.9995000000000006e-05,
      "loss": 0.4029,
      "step": 8003
    },
    {
      "epoch": 0.4482024862806585,
      "grad_norm": 1.1587896347045898,
      "learning_rate": 4e-05,
      "loss": 0.4203,
      "step": 8004
    },
    {
      "epoch": 0.4482584835927875,
      "grad_norm": 1.4524353742599487,
      "learning_rate": 4.0005e-05,
      "loss": 0.5276,
      "step": 8005
    },
    {
      "epoch": 0.44831448090491655,
      "grad_norm": 1.0707982778549194,
      "learning_rate": 4.0010000000000005e-05,
      "loss": 0.3752,
      "step": 8006
    },
    {
      "epoch": 0.44837047821704556,
      "grad_norm": 1.0258558988571167,
      "learning_rate": 4.0015e-05,
      "loss": 0.3766,
      "step": 8007
    },
    {
      "epoch": 0.4484264755291746,
      "grad_norm": 1.292534589767456,
      "learning_rate": 4.002e-05,
      "loss": 0.4509,
      "step": 8008
    },
    {
      "epoch": 0.4484824728413036,
      "grad_norm": 1.4285045862197876,
      "learning_rate": 4.0025000000000004e-05,
      "loss": 0.4806,
      "step": 8009
    },
    {
      "epoch": 0.4485384701534326,
      "grad_norm": 1.1258915662765503,
      "learning_rate": 4.003e-05,
      "loss": 0.3962,
      "step": 8010
    },
    {
      "epoch": 0.44859446746556164,
      "grad_norm": 1.215669870376587,
      "learning_rate": 4.0035e-05,
      "loss": 0.5454,
      "step": 8011
    },
    {
      "epoch": 0.44865046477769066,
      "grad_norm": 1.133365273475647,
      "learning_rate": 4.004e-05,
      "loss": 0.4184,
      "step": 8012
    },
    {
      "epoch": 0.4487064620898197,
      "grad_norm": 1.6083720922470093,
      "learning_rate": 4.0045e-05,
      "loss": 0.4896,
      "step": 8013
    },
    {
      "epoch": 0.4487624594019487,
      "grad_norm": 1.1899293661117554,
      "learning_rate": 4.0050000000000004e-05,
      "loss": 0.4205,
      "step": 8014
    },
    {
      "epoch": 0.4488184567140777,
      "grad_norm": 1.310274600982666,
      "learning_rate": 4.0055e-05,
      "loss": 0.4943,
      "step": 8015
    },
    {
      "epoch": 0.44887445402620674,
      "grad_norm": 1.2298719882965088,
      "learning_rate": 4.0060000000000006e-05,
      "loss": 0.3656,
      "step": 8016
    },
    {
      "epoch": 0.44893045133833576,
      "grad_norm": 1.258176565170288,
      "learning_rate": 4.0065000000000003e-05,
      "loss": 0.3348,
      "step": 8017
    },
    {
      "epoch": 0.4489864486504648,
      "grad_norm": 1.1516302824020386,
      "learning_rate": 4.007e-05,
      "loss": 0.5148,
      "step": 8018
    },
    {
      "epoch": 0.4490424459625938,
      "grad_norm": 1.500391960144043,
      "learning_rate": 4.0075e-05,
      "loss": 0.4206,
      "step": 8019
    },
    {
      "epoch": 0.4490984432747228,
      "grad_norm": 1.087298035621643,
      "learning_rate": 4.008e-05,
      "loss": 0.3701,
      "step": 8020
    },
    {
      "epoch": 0.44915444058685183,
      "grad_norm": 1.2705739736557007,
      "learning_rate": 4.0085e-05,
      "loss": 0.4524,
      "step": 8021
    },
    {
      "epoch": 0.44921043789898085,
      "grad_norm": 1.171284794807434,
      "learning_rate": 4.009e-05,
      "loss": 0.3866,
      "step": 8022
    },
    {
      "epoch": 0.44926643521110987,
      "grad_norm": 1.157090425491333,
      "learning_rate": 4.0095e-05,
      "loss": 0.4116,
      "step": 8023
    },
    {
      "epoch": 0.4493224325232389,
      "grad_norm": 1.1388421058654785,
      "learning_rate": 4.0100000000000006e-05,
      "loss": 0.4194,
      "step": 8024
    },
    {
      "epoch": 0.4493784298353679,
      "grad_norm": 1.231940746307373,
      "learning_rate": 4.0105e-05,
      "loss": 0.3605,
      "step": 8025
    },
    {
      "epoch": 0.44943442714749693,
      "grad_norm": 1.280762791633606,
      "learning_rate": 4.011e-05,
      "loss": 0.5177,
      "step": 8026
    },
    {
      "epoch": 0.44949042445962595,
      "grad_norm": 1.1017980575561523,
      "learning_rate": 4.0115000000000005e-05,
      "loss": 0.3673,
      "step": 8027
    },
    {
      "epoch": 0.44954642177175497,
      "grad_norm": 1.2504467964172363,
      "learning_rate": 4.012e-05,
      "loss": 0.512,
      "step": 8028
    },
    {
      "epoch": 0.449602419083884,
      "grad_norm": 1.2567552328109741,
      "learning_rate": 4.0125e-05,
      "loss": 0.3979,
      "step": 8029
    },
    {
      "epoch": 0.449658416396013,
      "grad_norm": 1.2547028064727783,
      "learning_rate": 4.0130000000000004e-05,
      "loss": 0.429,
      "step": 8030
    },
    {
      "epoch": 0.449714413708142,
      "grad_norm": 1.2034847736358643,
      "learning_rate": 4.0135e-05,
      "loss": 0.4543,
      "step": 8031
    },
    {
      "epoch": 0.44977041102027104,
      "grad_norm": 1.3718706369400024,
      "learning_rate": 4.014e-05,
      "loss": 0.4695,
      "step": 8032
    },
    {
      "epoch": 0.44982640833240006,
      "grad_norm": 1.1253634691238403,
      "learning_rate": 4.0144999999999996e-05,
      "loss": 0.4362,
      "step": 8033
    },
    {
      "epoch": 0.4498824056445291,
      "grad_norm": 1.0990970134735107,
      "learning_rate": 4.015000000000001e-05,
      "loss": 0.5066,
      "step": 8034
    },
    {
      "epoch": 0.4499384029566581,
      "grad_norm": 1.5450721979141235,
      "learning_rate": 4.0155000000000004e-05,
      "loss": 0.4642,
      "step": 8035
    },
    {
      "epoch": 0.4499944002687871,
      "grad_norm": 1.2929738759994507,
      "learning_rate": 4.016e-05,
      "loss": 0.5573,
      "step": 8036
    },
    {
      "epoch": 0.45005039758091614,
      "grad_norm": 1.633539080619812,
      "learning_rate": 4.0165000000000006e-05,
      "loss": 0.4439,
      "step": 8037
    },
    {
      "epoch": 0.45010639489304516,
      "grad_norm": 1.2135896682739258,
      "learning_rate": 4.017e-05,
      "loss": 0.3989,
      "step": 8038
    },
    {
      "epoch": 0.4501623922051742,
      "grad_norm": 1.394967794418335,
      "learning_rate": 4.0175e-05,
      "loss": 0.4351,
      "step": 8039
    },
    {
      "epoch": 0.4502183895173032,
      "grad_norm": 1.177072525024414,
      "learning_rate": 4.018e-05,
      "loss": 0.4132,
      "step": 8040
    },
    {
      "epoch": 0.45027438682943216,
      "grad_norm": 1.293129563331604,
      "learning_rate": 4.0185e-05,
      "loss": 0.3174,
      "step": 8041
    },
    {
      "epoch": 0.4503303841415612,
      "grad_norm": 1.3859515190124512,
      "learning_rate": 4.019e-05,
      "loss": 0.4402,
      "step": 8042
    },
    {
      "epoch": 0.4503863814536902,
      "grad_norm": 1.4350817203521729,
      "learning_rate": 4.0195e-05,
      "loss": 0.4825,
      "step": 8043
    },
    {
      "epoch": 0.4504423787658192,
      "grad_norm": 1.1298954486846924,
      "learning_rate": 4.02e-05,
      "loss": 0.4487,
      "step": 8044
    },
    {
      "epoch": 0.45049837607794824,
      "grad_norm": 1.2463808059692383,
      "learning_rate": 4.0205000000000006e-05,
      "loss": 0.3645,
      "step": 8045
    },
    {
      "epoch": 0.45055437339007726,
      "grad_norm": 1.2132930755615234,
      "learning_rate": 4.021e-05,
      "loss": 0.3821,
      "step": 8046
    },
    {
      "epoch": 0.4506103707022063,
      "grad_norm": 1.4917255640029907,
      "learning_rate": 4.0215e-05,
      "loss": 0.5085,
      "step": 8047
    },
    {
      "epoch": 0.4506663680143353,
      "grad_norm": 1.0278327465057373,
      "learning_rate": 4.0220000000000005e-05,
      "loss": 0.4188,
      "step": 8048
    },
    {
      "epoch": 0.4507223653264643,
      "grad_norm": 1.4655405282974243,
      "learning_rate": 4.0225e-05,
      "loss": 0.5233,
      "step": 8049
    },
    {
      "epoch": 0.45077836263859333,
      "grad_norm": 1.2491194009780884,
      "learning_rate": 4.023e-05,
      "loss": 0.4767,
      "step": 8050
    },
    {
      "epoch": 0.45083435995072235,
      "grad_norm": 1.330393671989441,
      "learning_rate": 4.0235000000000004e-05,
      "loss": 0.4438,
      "step": 8051
    },
    {
      "epoch": 0.45089035726285137,
      "grad_norm": 1.231075406074524,
      "learning_rate": 4.024e-05,
      "loss": 0.4787,
      "step": 8052
    },
    {
      "epoch": 0.4509463545749804,
      "grad_norm": 1.1820240020751953,
      "learning_rate": 4.0245e-05,
      "loss": 0.4129,
      "step": 8053
    },
    {
      "epoch": 0.4510023518871094,
      "grad_norm": 1.0467157363891602,
      "learning_rate": 4.025e-05,
      "loss": 0.3787,
      "step": 8054
    },
    {
      "epoch": 0.45105834919923843,
      "grad_norm": 1.2204055786132812,
      "learning_rate": 4.025500000000001e-05,
      "loss": 0.3706,
      "step": 8055
    },
    {
      "epoch": 0.45111434651136745,
      "grad_norm": 1.5015779733657837,
      "learning_rate": 4.0260000000000004e-05,
      "loss": 0.4293,
      "step": 8056
    },
    {
      "epoch": 0.45117034382349647,
      "grad_norm": 1.1499462127685547,
      "learning_rate": 4.0265e-05,
      "loss": 0.4028,
      "step": 8057
    },
    {
      "epoch": 0.4512263411356255,
      "grad_norm": 1.630139708518982,
      "learning_rate": 4.027e-05,
      "loss": 0.4948,
      "step": 8058
    },
    {
      "epoch": 0.4512823384477545,
      "grad_norm": 1.1332985162734985,
      "learning_rate": 4.0275e-05,
      "loss": 0.3303,
      "step": 8059
    },
    {
      "epoch": 0.4513383357598835,
      "grad_norm": 1.9683868885040283,
      "learning_rate": 4.028e-05,
      "loss": 0.4903,
      "step": 8060
    },
    {
      "epoch": 0.45139433307201254,
      "grad_norm": 1.0710465908050537,
      "learning_rate": 4.0285e-05,
      "loss": 0.3492,
      "step": 8061
    },
    {
      "epoch": 0.45145033038414156,
      "grad_norm": 1.514875888824463,
      "learning_rate": 4.029e-05,
      "loss": 0.4255,
      "step": 8062
    },
    {
      "epoch": 0.4515063276962706,
      "grad_norm": 1.4187906980514526,
      "learning_rate": 4.0295e-05,
      "loss": 0.4634,
      "step": 8063
    },
    {
      "epoch": 0.4515623250083996,
      "grad_norm": 1.4659744501113892,
      "learning_rate": 4.0300000000000004e-05,
      "loss": 0.4806,
      "step": 8064
    },
    {
      "epoch": 0.4516183223205286,
      "grad_norm": 1.1748239994049072,
      "learning_rate": 4.0305e-05,
      "loss": 0.3964,
      "step": 8065
    },
    {
      "epoch": 0.45167431963265764,
      "grad_norm": 1.0754516124725342,
      "learning_rate": 4.0310000000000005e-05,
      "loss": 0.394,
      "step": 8066
    },
    {
      "epoch": 0.45173031694478666,
      "grad_norm": 1.1976890563964844,
      "learning_rate": 4.0315e-05,
      "loss": 0.3775,
      "step": 8067
    },
    {
      "epoch": 0.4517863142569157,
      "grad_norm": 1.5069286823272705,
      "learning_rate": 4.032e-05,
      "loss": 0.6658,
      "step": 8068
    },
    {
      "epoch": 0.4518423115690447,
      "grad_norm": 1.6875029802322388,
      "learning_rate": 4.0325000000000004e-05,
      "loss": 0.4667,
      "step": 8069
    },
    {
      "epoch": 0.4518983088811737,
      "grad_norm": 1.1968203783035278,
      "learning_rate": 4.033e-05,
      "loss": 0.4361,
      "step": 8070
    },
    {
      "epoch": 0.45195430619330274,
      "grad_norm": 1.481873869895935,
      "learning_rate": 4.0335e-05,
      "loss": 0.4308,
      "step": 8071
    },
    {
      "epoch": 0.45201030350543175,
      "grad_norm": 1.190264344215393,
      "learning_rate": 4.034e-05,
      "loss": 0.4504,
      "step": 8072
    },
    {
      "epoch": 0.4520663008175608,
      "grad_norm": 1.1881425380706787,
      "learning_rate": 4.0345e-05,
      "loss": 0.4565,
      "step": 8073
    },
    {
      "epoch": 0.4521222981296898,
      "grad_norm": 1.15525484085083,
      "learning_rate": 4.0350000000000005e-05,
      "loss": 0.3779,
      "step": 8074
    },
    {
      "epoch": 0.4521782954418188,
      "grad_norm": 1.2618846893310547,
      "learning_rate": 4.0355e-05,
      "loss": 0.4439,
      "step": 8075
    },
    {
      "epoch": 0.45223429275394783,
      "grad_norm": 1.3922920227050781,
      "learning_rate": 4.0360000000000007e-05,
      "loss": 0.4628,
      "step": 8076
    },
    {
      "epoch": 0.45229029006607685,
      "grad_norm": 1.2423697710037231,
      "learning_rate": 4.0365000000000004e-05,
      "loss": 0.4043,
      "step": 8077
    },
    {
      "epoch": 0.45234628737820587,
      "grad_norm": 1.279617190361023,
      "learning_rate": 4.037e-05,
      "loss": 0.5224,
      "step": 8078
    },
    {
      "epoch": 0.4524022846903349,
      "grad_norm": 1.2444432973861694,
      "learning_rate": 4.0375e-05,
      "loss": 0.3951,
      "step": 8079
    },
    {
      "epoch": 0.4524582820024639,
      "grad_norm": 1.1708821058273315,
      "learning_rate": 4.038e-05,
      "loss": 0.3757,
      "step": 8080
    },
    {
      "epoch": 0.4525142793145929,
      "grad_norm": 1.2084872722625732,
      "learning_rate": 4.0385e-05,
      "loss": 0.3436,
      "step": 8081
    },
    {
      "epoch": 0.4525702766267219,
      "grad_norm": 1.4237982034683228,
      "learning_rate": 4.039e-05,
      "loss": 0.4698,
      "step": 8082
    },
    {
      "epoch": 0.4526262739388509,
      "grad_norm": 1.3786941766738892,
      "learning_rate": 4.0395e-05,
      "loss": 0.3887,
      "step": 8083
    },
    {
      "epoch": 0.45268227125097993,
      "grad_norm": 1.1412067413330078,
      "learning_rate": 4.0400000000000006e-05,
      "loss": 0.4617,
      "step": 8084
    },
    {
      "epoch": 0.45273826856310895,
      "grad_norm": 1.231732964515686,
      "learning_rate": 4.0405000000000004e-05,
      "loss": 0.3878,
      "step": 8085
    },
    {
      "epoch": 0.45279426587523797,
      "grad_norm": 1.242849588394165,
      "learning_rate": 4.041e-05,
      "loss": 0.4939,
      "step": 8086
    },
    {
      "epoch": 0.452850263187367,
      "grad_norm": 1.2185568809509277,
      "learning_rate": 4.0415000000000005e-05,
      "loss": 0.3916,
      "step": 8087
    },
    {
      "epoch": 0.452906260499496,
      "grad_norm": 1.1475306749343872,
      "learning_rate": 4.042e-05,
      "loss": 0.4227,
      "step": 8088
    },
    {
      "epoch": 0.452962257811625,
      "grad_norm": 1.4434412717819214,
      "learning_rate": 4.0425e-05,
      "loss": 0.489,
      "step": 8089
    },
    {
      "epoch": 0.45301825512375404,
      "grad_norm": 1.2608963251113892,
      "learning_rate": 4.0430000000000004e-05,
      "loss": 0.4135,
      "step": 8090
    },
    {
      "epoch": 0.45307425243588306,
      "grad_norm": 1.2011988162994385,
      "learning_rate": 4.0435e-05,
      "loss": 0.4414,
      "step": 8091
    },
    {
      "epoch": 0.4531302497480121,
      "grad_norm": 1.0195295810699463,
      "learning_rate": 4.044e-05,
      "loss": 0.4124,
      "step": 8092
    },
    {
      "epoch": 0.4531862470601411,
      "grad_norm": 1.2460380792617798,
      "learning_rate": 4.0444999999999996e-05,
      "loss": 0.408,
      "step": 8093
    },
    {
      "epoch": 0.4532422443722701,
      "grad_norm": 1.3952407836914062,
      "learning_rate": 4.045000000000001e-05,
      "loss": 0.5317,
      "step": 8094
    },
    {
      "epoch": 0.45329824168439914,
      "grad_norm": 1.2081043720245361,
      "learning_rate": 4.0455000000000005e-05,
      "loss": 0.37,
      "step": 8095
    },
    {
      "epoch": 0.45335423899652816,
      "grad_norm": 1.1570297479629517,
      "learning_rate": 4.046e-05,
      "loss": 0.365,
      "step": 8096
    },
    {
      "epoch": 0.4534102363086572,
      "grad_norm": 1.3316103219985962,
      "learning_rate": 4.0465e-05,
      "loss": 0.4904,
      "step": 8097
    },
    {
      "epoch": 0.4534662336207862,
      "grad_norm": 1.4848228693008423,
      "learning_rate": 4.0470000000000004e-05,
      "loss": 0.4557,
      "step": 8098
    },
    {
      "epoch": 0.4535222309329152,
      "grad_norm": 1.5339934825897217,
      "learning_rate": 4.0475e-05,
      "loss": 0.4291,
      "step": 8099
    },
    {
      "epoch": 0.45357822824504423,
      "grad_norm": 1.0706136226654053,
      "learning_rate": 4.048e-05,
      "loss": 0.3559,
      "step": 8100
    },
    {
      "epoch": 0.45363422555717325,
      "grad_norm": 1.1460086107254028,
      "learning_rate": 4.0485e-05,
      "loss": 0.3852,
      "step": 8101
    },
    {
      "epoch": 0.4536902228693023,
      "grad_norm": 1.1236015558242798,
      "learning_rate": 4.049e-05,
      "loss": 0.4457,
      "step": 8102
    },
    {
      "epoch": 0.4537462201814313,
      "grad_norm": 1.4457634687423706,
      "learning_rate": 4.0495e-05,
      "loss": 0.5892,
      "step": 8103
    },
    {
      "epoch": 0.4538022174935603,
      "grad_norm": 1.413439393043518,
      "learning_rate": 4.05e-05,
      "loss": 0.555,
      "step": 8104
    },
    {
      "epoch": 0.45385821480568933,
      "grad_norm": 1.2305419445037842,
      "learning_rate": 4.0505000000000006e-05,
      "loss": 0.3831,
      "step": 8105
    },
    {
      "epoch": 0.45391421211781835,
      "grad_norm": 1.3862773180007935,
      "learning_rate": 4.0510000000000003e-05,
      "loss": 0.4541,
      "step": 8106
    },
    {
      "epoch": 0.45397020942994737,
      "grad_norm": 1.3567650318145752,
      "learning_rate": 4.0515e-05,
      "loss": 0.5618,
      "step": 8107
    },
    {
      "epoch": 0.4540262067420764,
      "grad_norm": 1.232162356376648,
      "learning_rate": 4.0520000000000005e-05,
      "loss": 0.3998,
      "step": 8108
    },
    {
      "epoch": 0.4540822040542054,
      "grad_norm": 1.2350515127182007,
      "learning_rate": 4.0525e-05,
      "loss": 0.4445,
      "step": 8109
    },
    {
      "epoch": 0.4541382013663344,
      "grad_norm": 1.6478445529937744,
      "learning_rate": 4.053e-05,
      "loss": 0.4283,
      "step": 8110
    },
    {
      "epoch": 0.45419419867846345,
      "grad_norm": 1.372441053390503,
      "learning_rate": 4.0535000000000004e-05,
      "loss": 0.3788,
      "step": 8111
    },
    {
      "epoch": 0.45425019599059246,
      "grad_norm": 1.1326618194580078,
      "learning_rate": 4.054e-05,
      "loss": 0.4098,
      "step": 8112
    },
    {
      "epoch": 0.4543061933027215,
      "grad_norm": 1.2759462594985962,
      "learning_rate": 4.0545e-05,
      "loss": 0.4116,
      "step": 8113
    },
    {
      "epoch": 0.4543621906148505,
      "grad_norm": 1.4068135023117065,
      "learning_rate": 4.055e-05,
      "loss": 0.3877,
      "step": 8114
    },
    {
      "epoch": 0.4544181879269795,
      "grad_norm": 1.5296299457550049,
      "learning_rate": 4.055500000000001e-05,
      "loss": 0.5493,
      "step": 8115
    },
    {
      "epoch": 0.45447418523910854,
      "grad_norm": 1.0567477941513062,
      "learning_rate": 4.0560000000000005e-05,
      "loss": 0.3859,
      "step": 8116
    },
    {
      "epoch": 0.45453018255123756,
      "grad_norm": 1.291696310043335,
      "learning_rate": 4.0565e-05,
      "loss": 0.4156,
      "step": 8117
    },
    {
      "epoch": 0.4545861798633666,
      "grad_norm": 1.5526171922683716,
      "learning_rate": 4.057e-05,
      "loss": 0.5647,
      "step": 8118
    },
    {
      "epoch": 0.4546421771754956,
      "grad_norm": 1.2019410133361816,
      "learning_rate": 4.0575000000000004e-05,
      "loss": 0.3844,
      "step": 8119
    },
    {
      "epoch": 0.4546981744876246,
      "grad_norm": 1.3499386310577393,
      "learning_rate": 4.058e-05,
      "loss": 0.5272,
      "step": 8120
    },
    {
      "epoch": 0.45475417179975364,
      "grad_norm": 1.597776174545288,
      "learning_rate": 4.0585e-05,
      "loss": 0.5078,
      "step": 8121
    },
    {
      "epoch": 0.45481016911188266,
      "grad_norm": 1.053828477859497,
      "learning_rate": 4.059e-05,
      "loss": 0.282,
      "step": 8122
    },
    {
      "epoch": 0.4548661664240116,
      "grad_norm": 1.2141889333724976,
      "learning_rate": 4.0595e-05,
      "loss": 0.3871,
      "step": 8123
    },
    {
      "epoch": 0.45492216373614064,
      "grad_norm": 1.2278751134872437,
      "learning_rate": 4.0600000000000004e-05,
      "loss": 0.371,
      "step": 8124
    },
    {
      "epoch": 0.45497816104826966,
      "grad_norm": 1.1336435079574585,
      "learning_rate": 4.0605e-05,
      "loss": 0.3757,
      "step": 8125
    },
    {
      "epoch": 0.4550341583603987,
      "grad_norm": 1.213625431060791,
      "learning_rate": 4.0610000000000006e-05,
      "loss": 0.5468,
      "step": 8126
    },
    {
      "epoch": 0.4550901556725277,
      "grad_norm": 1.0659953355789185,
      "learning_rate": 4.0615e-05,
      "loss": 0.3782,
      "step": 8127
    },
    {
      "epoch": 0.4551461529846567,
      "grad_norm": 1.1252938508987427,
      "learning_rate": 4.062e-05,
      "loss": 0.3667,
      "step": 8128
    },
    {
      "epoch": 0.45520215029678573,
      "grad_norm": 1.2927285432815552,
      "learning_rate": 4.0625000000000005e-05,
      "loss": 0.4203,
      "step": 8129
    },
    {
      "epoch": 0.45525814760891475,
      "grad_norm": 1.2861311435699463,
      "learning_rate": 4.063e-05,
      "loss": 0.4138,
      "step": 8130
    },
    {
      "epoch": 0.4553141449210438,
      "grad_norm": 1.390527606010437,
      "learning_rate": 4.0635e-05,
      "loss": 0.4645,
      "step": 8131
    },
    {
      "epoch": 0.4553701422331728,
      "grad_norm": 1.4087966680526733,
      "learning_rate": 4.064e-05,
      "loss": 0.4163,
      "step": 8132
    },
    {
      "epoch": 0.4554261395453018,
      "grad_norm": 1.4138132333755493,
      "learning_rate": 4.0645e-05,
      "loss": 0.4528,
      "step": 8133
    },
    {
      "epoch": 0.45548213685743083,
      "grad_norm": 1.3514633178710938,
      "learning_rate": 4.065e-05,
      "loss": 0.3211,
      "step": 8134
    },
    {
      "epoch": 0.45553813416955985,
      "grad_norm": 1.0520461797714233,
      "learning_rate": 4.0655e-05,
      "loss": 0.4123,
      "step": 8135
    },
    {
      "epoch": 0.45559413148168887,
      "grad_norm": 1.2497740983963013,
      "learning_rate": 4.066e-05,
      "loss": 0.39,
      "step": 8136
    },
    {
      "epoch": 0.4556501287938179,
      "grad_norm": 1.0242879390716553,
      "learning_rate": 4.0665000000000005e-05,
      "loss": 0.3915,
      "step": 8137
    },
    {
      "epoch": 0.4557061261059469,
      "grad_norm": 1.367636799812317,
      "learning_rate": 4.067e-05,
      "loss": 0.362,
      "step": 8138
    },
    {
      "epoch": 0.4557621234180759,
      "grad_norm": 1.2019015550613403,
      "learning_rate": 4.0675e-05,
      "loss": 0.5287,
      "step": 8139
    },
    {
      "epoch": 0.45581812073020495,
      "grad_norm": 1.3564034700393677,
      "learning_rate": 4.0680000000000004e-05,
      "loss": 0.5151,
      "step": 8140
    },
    {
      "epoch": 0.45587411804233396,
      "grad_norm": 1.424517035484314,
      "learning_rate": 4.0685e-05,
      "loss": 0.5017,
      "step": 8141
    },
    {
      "epoch": 0.455930115354463,
      "grad_norm": 1.372143030166626,
      "learning_rate": 4.069e-05,
      "loss": 0.3535,
      "step": 8142
    },
    {
      "epoch": 0.455986112666592,
      "grad_norm": 1.584250569343567,
      "learning_rate": 4.0695e-05,
      "loss": 0.5072,
      "step": 8143
    },
    {
      "epoch": 0.456042109978721,
      "grad_norm": 1.085153579711914,
      "learning_rate": 4.07e-05,
      "loss": 0.4053,
      "step": 8144
    },
    {
      "epoch": 0.45609810729085004,
      "grad_norm": 1.2137624025344849,
      "learning_rate": 4.0705000000000004e-05,
      "loss": 0.4806,
      "step": 8145
    },
    {
      "epoch": 0.45615410460297906,
      "grad_norm": 1.4045476913452148,
      "learning_rate": 4.071e-05,
      "loss": 0.6196,
      "step": 8146
    },
    {
      "epoch": 0.4562101019151081,
      "grad_norm": 1.3317464590072632,
      "learning_rate": 4.0715000000000006e-05,
      "loss": 0.4997,
      "step": 8147
    },
    {
      "epoch": 0.4562660992272371,
      "grad_norm": 1.1357710361480713,
      "learning_rate": 4.072e-05,
      "loss": 0.3664,
      "step": 8148
    },
    {
      "epoch": 0.4563220965393661,
      "grad_norm": 1.2137690782546997,
      "learning_rate": 4.0725e-05,
      "loss": 0.3473,
      "step": 8149
    },
    {
      "epoch": 0.45637809385149514,
      "grad_norm": 1.4994457960128784,
      "learning_rate": 4.0730000000000005e-05,
      "loss": 0.5448,
      "step": 8150
    },
    {
      "epoch": 0.45643409116362416,
      "grad_norm": 1.363540768623352,
      "learning_rate": 4.0735e-05,
      "loss": 0.4171,
      "step": 8151
    },
    {
      "epoch": 0.4564900884757532,
      "grad_norm": 1.1198484897613525,
      "learning_rate": 4.074e-05,
      "loss": 0.3726,
      "step": 8152
    },
    {
      "epoch": 0.4565460857878822,
      "grad_norm": 1.3675659894943237,
      "learning_rate": 4.0745e-05,
      "loss": 0.5798,
      "step": 8153
    },
    {
      "epoch": 0.4566020831000112,
      "grad_norm": 1.2910454273223877,
      "learning_rate": 4.075e-05,
      "loss": 0.3636,
      "step": 8154
    },
    {
      "epoch": 0.45665808041214023,
      "grad_norm": 1.163226842880249,
      "learning_rate": 4.0755000000000005e-05,
      "loss": 0.3997,
      "step": 8155
    },
    {
      "epoch": 0.45671407772426925,
      "grad_norm": 1.3678209781646729,
      "learning_rate": 4.076e-05,
      "loss": 0.5217,
      "step": 8156
    },
    {
      "epoch": 0.45677007503639827,
      "grad_norm": 1.0722938776016235,
      "learning_rate": 4.0765e-05,
      "loss": 0.3597,
      "step": 8157
    },
    {
      "epoch": 0.4568260723485273,
      "grad_norm": 1.2889726161956787,
      "learning_rate": 4.0770000000000004e-05,
      "loss": 0.5934,
      "step": 8158
    },
    {
      "epoch": 0.4568820696606563,
      "grad_norm": 1.3214764595031738,
      "learning_rate": 4.0775e-05,
      "loss": 0.432,
      "step": 8159
    },
    {
      "epoch": 0.45693806697278533,
      "grad_norm": 1.1166839599609375,
      "learning_rate": 4.078e-05,
      "loss": 0.3357,
      "step": 8160
    },
    {
      "epoch": 0.45699406428491435,
      "grad_norm": 1.1353657245635986,
      "learning_rate": 4.0785e-05,
      "loss": 0.4701,
      "step": 8161
    },
    {
      "epoch": 0.45705006159704337,
      "grad_norm": 1.2369425296783447,
      "learning_rate": 4.079e-05,
      "loss": 0.3867,
      "step": 8162
    },
    {
      "epoch": 0.4571060589091724,
      "grad_norm": 1.2651628255844116,
      "learning_rate": 4.0795e-05,
      "loss": 0.4669,
      "step": 8163
    },
    {
      "epoch": 0.4571620562213014,
      "grad_norm": 1.2256214618682861,
      "learning_rate": 4.08e-05,
      "loss": 0.3618,
      "step": 8164
    },
    {
      "epoch": 0.45721805353343037,
      "grad_norm": 3.2736332416534424,
      "learning_rate": 4.0805000000000007e-05,
      "loss": 0.4127,
      "step": 8165
    },
    {
      "epoch": 0.4572740508455594,
      "grad_norm": 1.2997982501983643,
      "learning_rate": 4.0810000000000004e-05,
      "loss": 0.4168,
      "step": 8166
    },
    {
      "epoch": 0.4573300481576884,
      "grad_norm": 1.2261103391647339,
      "learning_rate": 4.0815e-05,
      "loss": 0.6351,
      "step": 8167
    },
    {
      "epoch": 0.4573860454698174,
      "grad_norm": 1.2697564363479614,
      "learning_rate": 4.0820000000000006e-05,
      "loss": 0.4485,
      "step": 8168
    },
    {
      "epoch": 0.45744204278194645,
      "grad_norm": 1.2801098823547363,
      "learning_rate": 4.0825e-05,
      "loss": 0.4397,
      "step": 8169
    },
    {
      "epoch": 0.45749804009407546,
      "grad_norm": 1.3548836708068848,
      "learning_rate": 4.083e-05,
      "loss": 0.4038,
      "step": 8170
    },
    {
      "epoch": 0.4575540374062045,
      "grad_norm": 1.4707496166229248,
      "learning_rate": 4.0835e-05,
      "loss": 0.4418,
      "step": 8171
    },
    {
      "epoch": 0.4576100347183335,
      "grad_norm": 1.1236820220947266,
      "learning_rate": 4.084e-05,
      "loss": 0.3652,
      "step": 8172
    },
    {
      "epoch": 0.4576660320304625,
      "grad_norm": 1.1483045816421509,
      "learning_rate": 4.0845e-05,
      "loss": 0.3515,
      "step": 8173
    },
    {
      "epoch": 0.45772202934259154,
      "grad_norm": 1.2674108743667603,
      "learning_rate": 4.085e-05,
      "loss": 0.5687,
      "step": 8174
    },
    {
      "epoch": 0.45777802665472056,
      "grad_norm": 1.2336115837097168,
      "learning_rate": 4.0855e-05,
      "loss": 0.4017,
      "step": 8175
    },
    {
      "epoch": 0.4578340239668496,
      "grad_norm": 1.3384089469909668,
      "learning_rate": 4.0860000000000005e-05,
      "loss": 0.4603,
      "step": 8176
    },
    {
      "epoch": 0.4578900212789786,
      "grad_norm": 1.5389137268066406,
      "learning_rate": 4.0865e-05,
      "loss": 0.468,
      "step": 8177
    },
    {
      "epoch": 0.4579460185911076,
      "grad_norm": 1.3750388622283936,
      "learning_rate": 4.087e-05,
      "loss": 0.4232,
      "step": 8178
    },
    {
      "epoch": 0.45800201590323664,
      "grad_norm": 1.1116269826889038,
      "learning_rate": 4.0875000000000004e-05,
      "loss": 0.3124,
      "step": 8179
    },
    {
      "epoch": 0.45805801321536566,
      "grad_norm": 1.4280734062194824,
      "learning_rate": 4.088e-05,
      "loss": 0.4786,
      "step": 8180
    },
    {
      "epoch": 0.4581140105274947,
      "grad_norm": 1.4413557052612305,
      "learning_rate": 4.0885e-05,
      "loss": 0.4382,
      "step": 8181
    },
    {
      "epoch": 0.4581700078396237,
      "grad_norm": 7.288043022155762,
      "learning_rate": 4.089e-05,
      "loss": 0.5532,
      "step": 8182
    },
    {
      "epoch": 0.4582260051517527,
      "grad_norm": 1.6042112112045288,
      "learning_rate": 4.0895e-05,
      "loss": 0.6748,
      "step": 8183
    },
    {
      "epoch": 0.45828200246388173,
      "grad_norm": 1.3809505701065063,
      "learning_rate": 4.09e-05,
      "loss": 0.3865,
      "step": 8184
    },
    {
      "epoch": 0.45833799977601075,
      "grad_norm": 1.2860163450241089,
      "learning_rate": 4.0905e-05,
      "loss": 0.3698,
      "step": 8185
    },
    {
      "epoch": 0.45839399708813977,
      "grad_norm": 1.2628626823425293,
      "learning_rate": 4.0910000000000006e-05,
      "loss": 0.6904,
      "step": 8186
    },
    {
      "epoch": 0.4584499944002688,
      "grad_norm": 1.2822555303573608,
      "learning_rate": 4.0915000000000004e-05,
      "loss": 0.3674,
      "step": 8187
    },
    {
      "epoch": 0.4585059917123978,
      "grad_norm": 1.1814857721328735,
      "learning_rate": 4.092e-05,
      "loss": 0.4019,
      "step": 8188
    },
    {
      "epoch": 0.45856198902452683,
      "grad_norm": 1.335870623588562,
      "learning_rate": 4.0925000000000005e-05,
      "loss": 0.4589,
      "step": 8189
    },
    {
      "epoch": 0.45861798633665585,
      "grad_norm": 1.0942025184631348,
      "learning_rate": 4.093e-05,
      "loss": 0.4158,
      "step": 8190
    },
    {
      "epoch": 0.45867398364878487,
      "grad_norm": 1.0841032266616821,
      "learning_rate": 4.0935e-05,
      "loss": 0.2678,
      "step": 8191
    },
    {
      "epoch": 0.4587299809609139,
      "grad_norm": 1.435125708580017,
      "learning_rate": 4.094e-05,
      "loss": 0.4795,
      "step": 8192
    },
    {
      "epoch": 0.4587859782730429,
      "grad_norm": 1.2814698219299316,
      "learning_rate": 4.0945e-05,
      "loss": 0.2863,
      "step": 8193
    },
    {
      "epoch": 0.4588419755851719,
      "grad_norm": 1.0510350465774536,
      "learning_rate": 4.095e-05,
      "loss": 0.4812,
      "step": 8194
    },
    {
      "epoch": 0.45889797289730094,
      "grad_norm": 1.5333740711212158,
      "learning_rate": 4.0955000000000003e-05,
      "loss": 0.4996,
      "step": 8195
    },
    {
      "epoch": 0.45895397020942996,
      "grad_norm": 1.2973487377166748,
      "learning_rate": 4.096e-05,
      "loss": 0.3833,
      "step": 8196
    },
    {
      "epoch": 0.459009967521559,
      "grad_norm": 1.12303626537323,
      "learning_rate": 4.0965000000000005e-05,
      "loss": 0.3749,
      "step": 8197
    },
    {
      "epoch": 0.459065964833688,
      "grad_norm": 1.536573052406311,
      "learning_rate": 4.097e-05,
      "loss": 0.3857,
      "step": 8198
    },
    {
      "epoch": 0.459121962145817,
      "grad_norm": 1.2383390665054321,
      "learning_rate": 4.0975e-05,
      "loss": 0.5466,
      "step": 8199
    },
    {
      "epoch": 0.45917795945794604,
      "grad_norm": 1.5300266742706299,
      "learning_rate": 4.0980000000000004e-05,
      "loss": 0.4991,
      "step": 8200
    },
    {
      "epoch": 0.45923395677007506,
      "grad_norm": 1.4450125694274902,
      "learning_rate": 4.0985e-05,
      "loss": 0.493,
      "step": 8201
    },
    {
      "epoch": 0.4592899540822041,
      "grad_norm": 1.259688377380371,
      "learning_rate": 4.099e-05,
      "loss": 0.3361,
      "step": 8202
    },
    {
      "epoch": 0.4593459513943331,
      "grad_norm": 1.6072922945022583,
      "learning_rate": 4.0995e-05,
      "loss": 0.368,
      "step": 8203
    },
    {
      "epoch": 0.4594019487064621,
      "grad_norm": 1.303582787513733,
      "learning_rate": 4.1e-05,
      "loss": 0.4323,
      "step": 8204
    },
    {
      "epoch": 0.45945794601859113,
      "grad_norm": 1.08994460105896,
      "learning_rate": 4.1005000000000005e-05,
      "loss": 0.3298,
      "step": 8205
    },
    {
      "epoch": 0.4595139433307201,
      "grad_norm": 1.1609015464782715,
      "learning_rate": 4.101e-05,
      "loss": 0.3274,
      "step": 8206
    },
    {
      "epoch": 0.4595699406428491,
      "grad_norm": 1.1675680875778198,
      "learning_rate": 4.1015000000000006e-05,
      "loss": 0.3734,
      "step": 8207
    },
    {
      "epoch": 0.45962593795497814,
      "grad_norm": 1.1078368425369263,
      "learning_rate": 4.1020000000000004e-05,
      "loss": 0.2963,
      "step": 8208
    },
    {
      "epoch": 0.45968193526710716,
      "grad_norm": 1.1644442081451416,
      "learning_rate": 4.1025e-05,
      "loss": 0.4045,
      "step": 8209
    },
    {
      "epoch": 0.4597379325792362,
      "grad_norm": 1.2316521406173706,
      "learning_rate": 4.103e-05,
      "loss": 0.3405,
      "step": 8210
    },
    {
      "epoch": 0.4597939298913652,
      "grad_norm": 1.2362315654754639,
      "learning_rate": 4.1035e-05,
      "loss": 0.4789,
      "step": 8211
    },
    {
      "epoch": 0.4598499272034942,
      "grad_norm": 1.1099541187286377,
      "learning_rate": 4.104e-05,
      "loss": 0.3682,
      "step": 8212
    },
    {
      "epoch": 0.45990592451562323,
      "grad_norm": 1.1598701477050781,
      "learning_rate": 4.1045e-05,
      "loss": 0.4685,
      "step": 8213
    },
    {
      "epoch": 0.45996192182775225,
      "grad_norm": 1.2803281545639038,
      "learning_rate": 4.105e-05,
      "loss": 0.4272,
      "step": 8214
    },
    {
      "epoch": 0.46001791913988127,
      "grad_norm": 1.3603127002716064,
      "learning_rate": 4.1055000000000006e-05,
      "loss": 0.4071,
      "step": 8215
    },
    {
      "epoch": 0.4600739164520103,
      "grad_norm": 1.4698951244354248,
      "learning_rate": 4.106e-05,
      "loss": 0.5065,
      "step": 8216
    },
    {
      "epoch": 0.4601299137641393,
      "grad_norm": 1.216223120689392,
      "learning_rate": 4.1065e-05,
      "loss": 0.42,
      "step": 8217
    },
    {
      "epoch": 0.46018591107626833,
      "grad_norm": 1.1401433944702148,
      "learning_rate": 4.1070000000000005e-05,
      "loss": 0.3659,
      "step": 8218
    },
    {
      "epoch": 0.46024190838839735,
      "grad_norm": 1.258061408996582,
      "learning_rate": 4.1075e-05,
      "loss": 0.3915,
      "step": 8219
    },
    {
      "epoch": 0.46029790570052637,
      "grad_norm": 1.2542272806167603,
      "learning_rate": 4.108e-05,
      "loss": 0.3817,
      "step": 8220
    },
    {
      "epoch": 0.4603539030126554,
      "grad_norm": 1.1371816396713257,
      "learning_rate": 4.1085000000000004e-05,
      "loss": 0.3568,
      "step": 8221
    },
    {
      "epoch": 0.4604099003247844,
      "grad_norm": 1.3532615900039673,
      "learning_rate": 4.109e-05,
      "loss": 0.5167,
      "step": 8222
    },
    {
      "epoch": 0.4604658976369134,
      "grad_norm": 1.1306337118148804,
      "learning_rate": 4.1095e-05,
      "loss": 0.3148,
      "step": 8223
    },
    {
      "epoch": 0.46052189494904244,
      "grad_norm": 1.223226547241211,
      "learning_rate": 4.11e-05,
      "loss": 0.4107,
      "step": 8224
    },
    {
      "epoch": 0.46057789226117146,
      "grad_norm": 1.410191297531128,
      "learning_rate": 4.110500000000001e-05,
      "loss": 0.398,
      "step": 8225
    },
    {
      "epoch": 0.4606338895733005,
      "grad_norm": 1.2490471601486206,
      "learning_rate": 4.1110000000000005e-05,
      "loss": 0.4474,
      "step": 8226
    },
    {
      "epoch": 0.4606898868854295,
      "grad_norm": 1.339874267578125,
      "learning_rate": 4.1115e-05,
      "loss": 0.4839,
      "step": 8227
    },
    {
      "epoch": 0.4607458841975585,
      "grad_norm": 1.209472417831421,
      "learning_rate": 4.1120000000000006e-05,
      "loss": 0.3883,
      "step": 8228
    },
    {
      "epoch": 0.46080188150968754,
      "grad_norm": 1.391992211341858,
      "learning_rate": 4.1125000000000004e-05,
      "loss": 0.4707,
      "step": 8229
    },
    {
      "epoch": 0.46085787882181656,
      "grad_norm": 1.5319911241531372,
      "learning_rate": 4.113e-05,
      "loss": 0.529,
      "step": 8230
    },
    {
      "epoch": 0.4609138761339456,
      "grad_norm": 1.330243468284607,
      "learning_rate": 4.1135e-05,
      "loss": 0.6138,
      "step": 8231
    },
    {
      "epoch": 0.4609698734460746,
      "grad_norm": 1.2333359718322754,
      "learning_rate": 4.114e-05,
      "loss": 0.4012,
      "step": 8232
    },
    {
      "epoch": 0.4610258707582036,
      "grad_norm": 1.2454456090927124,
      "learning_rate": 4.1145e-05,
      "loss": 0.3726,
      "step": 8233
    },
    {
      "epoch": 0.46108186807033263,
      "grad_norm": 1.2721551656723022,
      "learning_rate": 4.115e-05,
      "loss": 0.485,
      "step": 8234
    },
    {
      "epoch": 0.46113786538246165,
      "grad_norm": 1.3370745182037354,
      "learning_rate": 4.1155e-05,
      "loss": 0.4843,
      "step": 8235
    },
    {
      "epoch": 0.4611938626945907,
      "grad_norm": 1.1030439138412476,
      "learning_rate": 4.1160000000000006e-05,
      "loss": 0.4659,
      "step": 8236
    },
    {
      "epoch": 0.4612498600067197,
      "grad_norm": 1.1367319822311401,
      "learning_rate": 4.1165e-05,
      "loss": 0.3827,
      "step": 8237
    },
    {
      "epoch": 0.4613058573188487,
      "grad_norm": 1.2777172327041626,
      "learning_rate": 4.117e-05,
      "loss": 0.4422,
      "step": 8238
    },
    {
      "epoch": 0.46136185463097773,
      "grad_norm": 1.0879982709884644,
      "learning_rate": 4.1175000000000005e-05,
      "loss": 0.4092,
      "step": 8239
    },
    {
      "epoch": 0.46141785194310675,
      "grad_norm": 1.1981239318847656,
      "learning_rate": 4.118e-05,
      "loss": 0.3226,
      "step": 8240
    },
    {
      "epoch": 0.46147384925523577,
      "grad_norm": 1.2820683717727661,
      "learning_rate": 4.1185e-05,
      "loss": 0.3804,
      "step": 8241
    },
    {
      "epoch": 0.4615298465673648,
      "grad_norm": 1.2135425806045532,
      "learning_rate": 4.1190000000000004e-05,
      "loss": 0.4471,
      "step": 8242
    },
    {
      "epoch": 0.4615858438794938,
      "grad_norm": 1.178632378578186,
      "learning_rate": 4.1195e-05,
      "loss": 0.3996,
      "step": 8243
    },
    {
      "epoch": 0.4616418411916228,
      "grad_norm": 1.5495091676712036,
      "learning_rate": 4.12e-05,
      "loss": 0.4909,
      "step": 8244
    },
    {
      "epoch": 0.46169783850375185,
      "grad_norm": 1.735106348991394,
      "learning_rate": 4.1205e-05,
      "loss": 0.559,
      "step": 8245
    },
    {
      "epoch": 0.46175383581588086,
      "grad_norm": 1.3498954772949219,
      "learning_rate": 4.121000000000001e-05,
      "loss": 0.456,
      "step": 8246
    },
    {
      "epoch": 0.46180983312800983,
      "grad_norm": 1.32374906539917,
      "learning_rate": 4.1215000000000004e-05,
      "loss": 0.4899,
      "step": 8247
    },
    {
      "epoch": 0.46186583044013885,
      "grad_norm": 1.8373581171035767,
      "learning_rate": 4.122e-05,
      "loss": 0.5818,
      "step": 8248
    },
    {
      "epoch": 0.46192182775226787,
      "grad_norm": 1.378580093383789,
      "learning_rate": 4.1225e-05,
      "loss": 0.4885,
      "step": 8249
    },
    {
      "epoch": 0.4619778250643969,
      "grad_norm": 1.1680793762207031,
      "learning_rate": 4.123e-05,
      "loss": 0.3525,
      "step": 8250
    },
    {
      "epoch": 0.4620338223765259,
      "grad_norm": 1.231508731842041,
      "learning_rate": 4.1235e-05,
      "loss": 0.4798,
      "step": 8251
    },
    {
      "epoch": 0.4620898196886549,
      "grad_norm": 1.0578612089157104,
      "learning_rate": 4.124e-05,
      "loss": 0.4774,
      "step": 8252
    },
    {
      "epoch": 0.46214581700078394,
      "grad_norm": 1.1454347372055054,
      "learning_rate": 4.1245e-05,
      "loss": 0.5126,
      "step": 8253
    },
    {
      "epoch": 0.46220181431291296,
      "grad_norm": 1.1849184036254883,
      "learning_rate": 4.125e-05,
      "loss": 0.4887,
      "step": 8254
    },
    {
      "epoch": 0.462257811625042,
      "grad_norm": 1.3541444540023804,
      "learning_rate": 4.1255e-05,
      "loss": 0.444,
      "step": 8255
    },
    {
      "epoch": 0.462313808937171,
      "grad_norm": 1.303864598274231,
      "learning_rate": 4.126e-05,
      "loss": 0.4418,
      "step": 8256
    },
    {
      "epoch": 0.4623698062493,
      "grad_norm": 1.2061721086502075,
      "learning_rate": 4.1265000000000006e-05,
      "loss": 0.3507,
      "step": 8257
    },
    {
      "epoch": 0.46242580356142904,
      "grad_norm": 1.4810192584991455,
      "learning_rate": 4.127e-05,
      "loss": 0.4216,
      "step": 8258
    },
    {
      "epoch": 0.46248180087355806,
      "grad_norm": 1.2376545667648315,
      "learning_rate": 4.1275e-05,
      "loss": 0.3474,
      "step": 8259
    },
    {
      "epoch": 0.4625377981856871,
      "grad_norm": 1.3552175760269165,
      "learning_rate": 4.1280000000000005e-05,
      "loss": 0.5276,
      "step": 8260
    },
    {
      "epoch": 0.4625937954978161,
      "grad_norm": 1.53526771068573,
      "learning_rate": 4.1285e-05,
      "loss": 0.5101,
      "step": 8261
    },
    {
      "epoch": 0.4626497928099451,
      "grad_norm": 1.2176669836044312,
      "learning_rate": 4.129e-05,
      "loss": 0.3766,
      "step": 8262
    },
    {
      "epoch": 0.46270579012207413,
      "grad_norm": 1.3207225799560547,
      "learning_rate": 4.1295000000000004e-05,
      "loss": 0.3536,
      "step": 8263
    },
    {
      "epoch": 0.46276178743420315,
      "grad_norm": 1.5815467834472656,
      "learning_rate": 4.13e-05,
      "loss": 0.3933,
      "step": 8264
    },
    {
      "epoch": 0.4628177847463322,
      "grad_norm": 1.1484071016311646,
      "learning_rate": 4.1305e-05,
      "loss": 0.3941,
      "step": 8265
    },
    {
      "epoch": 0.4628737820584612,
      "grad_norm": 1.2536993026733398,
      "learning_rate": 4.131e-05,
      "loss": 0.3645,
      "step": 8266
    },
    {
      "epoch": 0.4629297793705902,
      "grad_norm": 1.1876859664916992,
      "learning_rate": 4.131500000000001e-05,
      "loss": 0.5404,
      "step": 8267
    },
    {
      "epoch": 0.46298577668271923,
      "grad_norm": 1.2260714769363403,
      "learning_rate": 4.1320000000000004e-05,
      "loss": 0.5652,
      "step": 8268
    },
    {
      "epoch": 0.46304177399484825,
      "grad_norm": 1.2548279762268066,
      "learning_rate": 4.1325e-05,
      "loss": 0.4476,
      "step": 8269
    },
    {
      "epoch": 0.46309777130697727,
      "grad_norm": 1.3838043212890625,
      "learning_rate": 4.133e-05,
      "loss": 0.3975,
      "step": 8270
    },
    {
      "epoch": 0.4631537686191063,
      "grad_norm": 1.2394155263900757,
      "learning_rate": 4.1335e-05,
      "loss": 0.4295,
      "step": 8271
    },
    {
      "epoch": 0.4632097659312353,
      "grad_norm": 1.3314915895462036,
      "learning_rate": 4.134e-05,
      "loss": 0.5559,
      "step": 8272
    },
    {
      "epoch": 0.4632657632433643,
      "grad_norm": 1.9199459552764893,
      "learning_rate": 4.1345e-05,
      "loss": 0.5407,
      "step": 8273
    },
    {
      "epoch": 0.46332176055549334,
      "grad_norm": 1.1639423370361328,
      "learning_rate": 4.135e-05,
      "loss": 0.4417,
      "step": 8274
    },
    {
      "epoch": 0.46337775786762236,
      "grad_norm": 1.4173465967178345,
      "learning_rate": 4.1355e-05,
      "loss": 0.3688,
      "step": 8275
    },
    {
      "epoch": 0.4634337551797514,
      "grad_norm": 1.2867845296859741,
      "learning_rate": 4.1360000000000004e-05,
      "loss": 0.387,
      "step": 8276
    },
    {
      "epoch": 0.4634897524918804,
      "grad_norm": 1.2569302320480347,
      "learning_rate": 4.1365e-05,
      "loss": 0.4712,
      "step": 8277
    },
    {
      "epoch": 0.4635457498040094,
      "grad_norm": 1.2919987440109253,
      "learning_rate": 4.1370000000000005e-05,
      "loss": 0.5301,
      "step": 8278
    },
    {
      "epoch": 0.46360174711613844,
      "grad_norm": 1.2371081113815308,
      "learning_rate": 4.1375e-05,
      "loss": 0.4337,
      "step": 8279
    },
    {
      "epoch": 0.46365774442826746,
      "grad_norm": 1.2740482091903687,
      "learning_rate": 4.138e-05,
      "loss": 0.5003,
      "step": 8280
    },
    {
      "epoch": 0.4637137417403965,
      "grad_norm": 1.199916124343872,
      "learning_rate": 4.1385000000000004e-05,
      "loss": 0.4129,
      "step": 8281
    },
    {
      "epoch": 0.4637697390525255,
      "grad_norm": 1.294663667678833,
      "learning_rate": 4.139e-05,
      "loss": 0.4201,
      "step": 8282
    },
    {
      "epoch": 0.4638257363646545,
      "grad_norm": 1.6683567762374878,
      "learning_rate": 4.1395e-05,
      "loss": 0.4427,
      "step": 8283
    },
    {
      "epoch": 0.46388173367678354,
      "grad_norm": 1.3331955671310425,
      "learning_rate": 4.14e-05,
      "loss": 0.3179,
      "step": 8284
    },
    {
      "epoch": 0.46393773098891256,
      "grad_norm": 1.1765532493591309,
      "learning_rate": 4.1405e-05,
      "loss": 0.2859,
      "step": 8285
    },
    {
      "epoch": 0.4639937283010416,
      "grad_norm": 1.1770352125167847,
      "learning_rate": 4.1410000000000005e-05,
      "loss": 0.4553,
      "step": 8286
    },
    {
      "epoch": 0.4640497256131706,
      "grad_norm": 1.4300110340118408,
      "learning_rate": 4.1415e-05,
      "loss": 0.3925,
      "step": 8287
    },
    {
      "epoch": 0.4641057229252996,
      "grad_norm": 1.2035119533538818,
      "learning_rate": 4.142000000000001e-05,
      "loss": 0.4124,
      "step": 8288
    },
    {
      "epoch": 0.4641617202374286,
      "grad_norm": 1.1478502750396729,
      "learning_rate": 4.1425000000000004e-05,
      "loss": 0.483,
      "step": 8289
    },
    {
      "epoch": 0.4642177175495576,
      "grad_norm": 1.1492865085601807,
      "learning_rate": 4.143e-05,
      "loss": 0.3764,
      "step": 8290
    },
    {
      "epoch": 0.4642737148616866,
      "grad_norm": 1.1740813255310059,
      "learning_rate": 4.1435e-05,
      "loss": 0.4237,
      "step": 8291
    },
    {
      "epoch": 0.46432971217381563,
      "grad_norm": 1.120821237564087,
      "learning_rate": 4.144e-05,
      "loss": 0.3322,
      "step": 8292
    },
    {
      "epoch": 0.46438570948594465,
      "grad_norm": 1.0857949256896973,
      "learning_rate": 4.1445e-05,
      "loss": 0.3412,
      "step": 8293
    },
    {
      "epoch": 0.4644417067980737,
      "grad_norm": 1.2305532693862915,
      "learning_rate": 4.145e-05,
      "loss": 0.4732,
      "step": 8294
    },
    {
      "epoch": 0.4644977041102027,
      "grad_norm": 3.610482692718506,
      "learning_rate": 4.1455e-05,
      "loss": 0.5287,
      "step": 8295
    },
    {
      "epoch": 0.4645537014223317,
      "grad_norm": 1.2439639568328857,
      "learning_rate": 4.1460000000000006e-05,
      "loss": 0.4171,
      "step": 8296
    },
    {
      "epoch": 0.46460969873446073,
      "grad_norm": 1.2058327198028564,
      "learning_rate": 4.1465000000000004e-05,
      "loss": 0.4941,
      "step": 8297
    },
    {
      "epoch": 0.46466569604658975,
      "grad_norm": 1.315577507019043,
      "learning_rate": 4.147e-05,
      "loss": 0.3526,
      "step": 8298
    },
    {
      "epoch": 0.46472169335871877,
      "grad_norm": 1.451165795326233,
      "learning_rate": 4.1475000000000005e-05,
      "loss": 0.4023,
      "step": 8299
    },
    {
      "epoch": 0.4647776906708478,
      "grad_norm": 1.1746854782104492,
      "learning_rate": 4.148e-05,
      "loss": 0.3519,
      "step": 8300
    },
    {
      "epoch": 0.4648336879829768,
      "grad_norm": 1.617592692375183,
      "learning_rate": 4.1485e-05,
      "loss": 0.429,
      "step": 8301
    },
    {
      "epoch": 0.4648896852951058,
      "grad_norm": 1.423981785774231,
      "learning_rate": 4.1490000000000004e-05,
      "loss": 0.5623,
      "step": 8302
    },
    {
      "epoch": 0.46494568260723484,
      "grad_norm": 1.2950692176818848,
      "learning_rate": 4.1495e-05,
      "loss": 0.4532,
      "step": 8303
    },
    {
      "epoch": 0.46500167991936386,
      "grad_norm": 1.4684441089630127,
      "learning_rate": 4.15e-05,
      "loss": 0.5641,
      "step": 8304
    },
    {
      "epoch": 0.4650576772314929,
      "grad_norm": 1.3021410703659058,
      "learning_rate": 4.1504999999999996e-05,
      "loss": 0.4803,
      "step": 8305
    },
    {
      "epoch": 0.4651136745436219,
      "grad_norm": 1.3266184329986572,
      "learning_rate": 4.151000000000001e-05,
      "loss": 0.5234,
      "step": 8306
    },
    {
      "epoch": 0.4651696718557509,
      "grad_norm": 1.1590931415557861,
      "learning_rate": 4.1515000000000005e-05,
      "loss": 0.3411,
      "step": 8307
    },
    {
      "epoch": 0.46522566916787994,
      "grad_norm": 1.4431616067886353,
      "learning_rate": 4.152e-05,
      "loss": 0.4403,
      "step": 8308
    },
    {
      "epoch": 0.46528166648000896,
      "grad_norm": 1.0642368793487549,
      "learning_rate": 4.1525e-05,
      "loss": 0.3634,
      "step": 8309
    },
    {
      "epoch": 0.465337663792138,
      "grad_norm": 1.254712462425232,
      "learning_rate": 4.1530000000000004e-05,
      "loss": 0.4495,
      "step": 8310
    },
    {
      "epoch": 0.465393661104267,
      "grad_norm": 1.1482433080673218,
      "learning_rate": 4.1535e-05,
      "loss": 0.3812,
      "step": 8311
    },
    {
      "epoch": 0.465449658416396,
      "grad_norm": 1.107596755027771,
      "learning_rate": 4.154e-05,
      "loss": 0.4688,
      "step": 8312
    },
    {
      "epoch": 0.46550565572852504,
      "grad_norm": 1.2421313524246216,
      "learning_rate": 4.1545e-05,
      "loss": 0.3898,
      "step": 8313
    },
    {
      "epoch": 0.46556165304065406,
      "grad_norm": 1.170276165008545,
      "learning_rate": 4.155e-05,
      "loss": 0.3996,
      "step": 8314
    },
    {
      "epoch": 0.4656176503527831,
      "grad_norm": 1.4561901092529297,
      "learning_rate": 4.1555e-05,
      "loss": 0.4181,
      "step": 8315
    },
    {
      "epoch": 0.4656736476649121,
      "grad_norm": 1.160518765449524,
      "learning_rate": 4.156e-05,
      "loss": 0.524,
      "step": 8316
    },
    {
      "epoch": 0.4657296449770411,
      "grad_norm": 1.5591087341308594,
      "learning_rate": 4.1565000000000006e-05,
      "loss": 0.5717,
      "step": 8317
    },
    {
      "epoch": 0.46578564228917013,
      "grad_norm": 1.2539479732513428,
      "learning_rate": 4.1570000000000003e-05,
      "loss": 0.5269,
      "step": 8318
    },
    {
      "epoch": 0.46584163960129915,
      "grad_norm": 1.2426033020019531,
      "learning_rate": 4.1575e-05,
      "loss": 0.5566,
      "step": 8319
    },
    {
      "epoch": 0.46589763691342817,
      "grad_norm": 1.1913050413131714,
      "learning_rate": 4.1580000000000005e-05,
      "loss": 0.3839,
      "step": 8320
    },
    {
      "epoch": 0.4659536342255572,
      "grad_norm": 1.2415966987609863,
      "learning_rate": 4.1585e-05,
      "loss": 0.3982,
      "step": 8321
    },
    {
      "epoch": 0.4660096315376862,
      "grad_norm": 1.1320737600326538,
      "learning_rate": 4.159e-05,
      "loss": 0.4148,
      "step": 8322
    },
    {
      "epoch": 0.46606562884981523,
      "grad_norm": 1.401557445526123,
      "learning_rate": 4.1595e-05,
      "loss": 0.4431,
      "step": 8323
    },
    {
      "epoch": 0.46612162616194425,
      "grad_norm": 1.2837218046188354,
      "learning_rate": 4.16e-05,
      "loss": 0.4153,
      "step": 8324
    },
    {
      "epoch": 0.46617762347407327,
      "grad_norm": 1.4516154527664185,
      "learning_rate": 4.1605e-05,
      "loss": 0.493,
      "step": 8325
    },
    {
      "epoch": 0.4662336207862023,
      "grad_norm": 1.219534993171692,
      "learning_rate": 4.161e-05,
      "loss": 0.4909,
      "step": 8326
    },
    {
      "epoch": 0.4662896180983313,
      "grad_norm": 1.3851169347763062,
      "learning_rate": 4.161500000000001e-05,
      "loss": 0.3984,
      "step": 8327
    },
    {
      "epoch": 0.4663456154104603,
      "grad_norm": 1.273726224899292,
      "learning_rate": 4.1620000000000005e-05,
      "loss": 0.3899,
      "step": 8328
    },
    {
      "epoch": 0.46640161272258934,
      "grad_norm": 2.3882343769073486,
      "learning_rate": 4.1625e-05,
      "loss": 0.6668,
      "step": 8329
    },
    {
      "epoch": 0.4664576100347183,
      "grad_norm": 1.1493287086486816,
      "learning_rate": 4.163e-05,
      "loss": 0.3776,
      "step": 8330
    },
    {
      "epoch": 0.4665136073468473,
      "grad_norm": 1.4332467317581177,
      "learning_rate": 4.1635000000000004e-05,
      "loss": 0.4647,
      "step": 8331
    },
    {
      "epoch": 0.46656960465897634,
      "grad_norm": 1.2122044563293457,
      "learning_rate": 4.164e-05,
      "loss": 0.3503,
      "step": 8332
    },
    {
      "epoch": 0.46662560197110536,
      "grad_norm": 1.2885552644729614,
      "learning_rate": 4.1645e-05,
      "loss": 0.587,
      "step": 8333
    },
    {
      "epoch": 0.4666815992832344,
      "grad_norm": 1.2309215068817139,
      "learning_rate": 4.165e-05,
      "loss": 0.5463,
      "step": 8334
    },
    {
      "epoch": 0.4667375965953634,
      "grad_norm": 1.1431134939193726,
      "learning_rate": 4.1655e-05,
      "loss": 0.3593,
      "step": 8335
    },
    {
      "epoch": 0.4667935939074924,
      "grad_norm": 1.6323133707046509,
      "learning_rate": 4.1660000000000004e-05,
      "loss": 0.4827,
      "step": 8336
    },
    {
      "epoch": 0.46684959121962144,
      "grad_norm": 1.3194406032562256,
      "learning_rate": 4.1665e-05,
      "loss": 0.4625,
      "step": 8337
    },
    {
      "epoch": 0.46690558853175046,
      "grad_norm": 1.180116891860962,
      "learning_rate": 4.1670000000000006e-05,
      "loss": 0.4089,
      "step": 8338
    },
    {
      "epoch": 0.4669615858438795,
      "grad_norm": 1.1480462551116943,
      "learning_rate": 4.1675e-05,
      "loss": 0.426,
      "step": 8339
    },
    {
      "epoch": 0.4670175831560085,
      "grad_norm": 1.3161413669586182,
      "learning_rate": 4.168e-05,
      "loss": 0.4495,
      "step": 8340
    },
    {
      "epoch": 0.4670735804681375,
      "grad_norm": 1.3676503896713257,
      "learning_rate": 4.1685000000000005e-05,
      "loss": 0.4018,
      "step": 8341
    },
    {
      "epoch": 0.46712957778026654,
      "grad_norm": 1.2224594354629517,
      "learning_rate": 4.169e-05,
      "loss": 0.3976,
      "step": 8342
    },
    {
      "epoch": 0.46718557509239556,
      "grad_norm": 1.1920114755630493,
      "learning_rate": 4.1695e-05,
      "loss": 0.3581,
      "step": 8343
    },
    {
      "epoch": 0.4672415724045246,
      "grad_norm": 1.3474382162094116,
      "learning_rate": 4.17e-05,
      "loss": 0.4556,
      "step": 8344
    },
    {
      "epoch": 0.4672975697166536,
      "grad_norm": 1.1781539916992188,
      "learning_rate": 4.1705e-05,
      "loss": 0.4035,
      "step": 8345
    },
    {
      "epoch": 0.4673535670287826,
      "grad_norm": 1.3382846117019653,
      "learning_rate": 4.1710000000000006e-05,
      "loss": 0.4421,
      "step": 8346
    },
    {
      "epoch": 0.46740956434091163,
      "grad_norm": 1.2219523191452026,
      "learning_rate": 4.1715e-05,
      "loss": 0.342,
      "step": 8347
    },
    {
      "epoch": 0.46746556165304065,
      "grad_norm": 1.2026203870773315,
      "learning_rate": 4.172e-05,
      "loss": 0.3207,
      "step": 8348
    },
    {
      "epoch": 0.46752155896516967,
      "grad_norm": 1.270452857017517,
      "learning_rate": 4.1725000000000005e-05,
      "loss": 0.3315,
      "step": 8349
    },
    {
      "epoch": 0.4675775562772987,
      "grad_norm": 1.188292145729065,
      "learning_rate": 4.173e-05,
      "loss": 0.4851,
      "step": 8350
    },
    {
      "epoch": 0.4676335535894277,
      "grad_norm": 1.2668474912643433,
      "learning_rate": 4.1735e-05,
      "loss": 0.5031,
      "step": 8351
    },
    {
      "epoch": 0.4676895509015567,
      "grad_norm": 1.1694830656051636,
      "learning_rate": 4.1740000000000004e-05,
      "loss": 0.399,
      "step": 8352
    },
    {
      "epoch": 0.46774554821368575,
      "grad_norm": Infinity,
      "learning_rate": 4.1740000000000004e-05,
      "loss": 0.3961,
      "step": 8353
    },
    {
      "epoch": 0.46780154552581477,
      "grad_norm": 1.2740497589111328,
      "learning_rate": 4.1745e-05,
      "loss": 0.3757,
      "step": 8354
    },
    {
      "epoch": 0.4678575428379438,
      "grad_norm": 1.4216548204421997,
      "learning_rate": 4.175e-05,
      "loss": 0.5351,
      "step": 8355
    },
    {
      "epoch": 0.4679135401500728,
      "grad_norm": 1.3175214529037476,
      "learning_rate": 4.1755e-05,
      "loss": 0.4135,
      "step": 8356
    },
    {
      "epoch": 0.4679695374622018,
      "grad_norm": 1.3715314865112305,
      "learning_rate": 4.176000000000001e-05,
      "loss": 0.4682,
      "step": 8357
    },
    {
      "epoch": 0.46802553477433084,
      "grad_norm": 1.3775523900985718,
      "learning_rate": 4.1765000000000004e-05,
      "loss": 0.3678,
      "step": 8358
    },
    {
      "epoch": 0.46808153208645986,
      "grad_norm": 0.9641080498695374,
      "learning_rate": 4.177e-05,
      "loss": 0.352,
      "step": 8359
    },
    {
      "epoch": 0.4681375293985889,
      "grad_norm": 1.3386561870574951,
      "learning_rate": 4.1775000000000006e-05,
      "loss": 0.3755,
      "step": 8360
    },
    {
      "epoch": 0.4681935267107179,
      "grad_norm": 1.8422759771347046,
      "learning_rate": 4.178e-05,
      "loss": 0.4397,
      "step": 8361
    },
    {
      "epoch": 0.4682495240228469,
      "grad_norm": 1.231330394744873,
      "learning_rate": 4.1785e-05,
      "loss": 0.4097,
      "step": 8362
    },
    {
      "epoch": 0.46830552133497594,
      "grad_norm": 1.1578826904296875,
      "learning_rate": 4.179e-05,
      "loss": 0.4185,
      "step": 8363
    },
    {
      "epoch": 0.46836151864710496,
      "grad_norm": 1.117853045463562,
      "learning_rate": 4.1795e-05,
      "loss": 0.4176,
      "step": 8364
    },
    {
      "epoch": 0.468417515959234,
      "grad_norm": 1.170587420463562,
      "learning_rate": 4.18e-05,
      "loss": 0.3724,
      "step": 8365
    },
    {
      "epoch": 0.468473513271363,
      "grad_norm": 1.0577501058578491,
      "learning_rate": 4.1805e-05,
      "loss": 0.4205,
      "step": 8366
    },
    {
      "epoch": 0.468529510583492,
      "grad_norm": 1.3998960256576538,
      "learning_rate": 4.181000000000001e-05,
      "loss": 0.5887,
      "step": 8367
    },
    {
      "epoch": 0.46858550789562103,
      "grad_norm": 1.2609167098999023,
      "learning_rate": 4.1815000000000005e-05,
      "loss": 0.4566,
      "step": 8368
    },
    {
      "epoch": 0.46864150520775005,
      "grad_norm": 1.1134520769119263,
      "learning_rate": 4.182e-05,
      "loss": 0.4511,
      "step": 8369
    },
    {
      "epoch": 0.4686975025198791,
      "grad_norm": 1.513018250465393,
      "learning_rate": 4.1825e-05,
      "loss": 0.4166,
      "step": 8370
    },
    {
      "epoch": 0.46875349983200804,
      "grad_norm": 1.2183899879455566,
      "learning_rate": 4.1830000000000004e-05,
      "loss": 0.3214,
      "step": 8371
    },
    {
      "epoch": 0.46880949714413706,
      "grad_norm": 1.2080955505371094,
      "learning_rate": 4.1835e-05,
      "loss": 0.4435,
      "step": 8372
    },
    {
      "epoch": 0.4688654944562661,
      "grad_norm": 1.2426809072494507,
      "learning_rate": 4.184e-05,
      "loss": 0.5445,
      "step": 8373
    },
    {
      "epoch": 0.4689214917683951,
      "grad_norm": 1.1984511613845825,
      "learning_rate": 4.1845000000000003e-05,
      "loss": 0.3413,
      "step": 8374
    },
    {
      "epoch": 0.4689774890805241,
      "grad_norm": 1.206897497177124,
      "learning_rate": 4.185e-05,
      "loss": 0.6036,
      "step": 8375
    },
    {
      "epoch": 0.46903348639265313,
      "grad_norm": 1.1841862201690674,
      "learning_rate": 4.1855e-05,
      "loss": 0.3707,
      "step": 8376
    },
    {
      "epoch": 0.46908948370478215,
      "grad_norm": 1.1985225677490234,
      "learning_rate": 4.186e-05,
      "loss": 0.4071,
      "step": 8377
    },
    {
      "epoch": 0.46914548101691117,
      "grad_norm": 1.1437454223632812,
      "learning_rate": 4.1865000000000007e-05,
      "loss": 0.3683,
      "step": 8378
    },
    {
      "epoch": 0.4692014783290402,
      "grad_norm": 1.1463754177093506,
      "learning_rate": 4.1870000000000004e-05,
      "loss": 0.3746,
      "step": 8379
    },
    {
      "epoch": 0.4692574756411692,
      "grad_norm": 1.1942380666732788,
      "learning_rate": 4.1875e-05,
      "loss": 0.3742,
      "step": 8380
    },
    {
      "epoch": 0.4693134729532982,
      "grad_norm": 1.176209807395935,
      "learning_rate": 4.1880000000000006e-05,
      "loss": 0.4211,
      "step": 8381
    },
    {
      "epoch": 0.46936947026542725,
      "grad_norm": 1.2539499998092651,
      "learning_rate": 4.1885e-05,
      "loss": 0.4638,
      "step": 8382
    },
    {
      "epoch": 0.46942546757755627,
      "grad_norm": 1.2557951211929321,
      "learning_rate": 4.189e-05,
      "loss": 0.3983,
      "step": 8383
    },
    {
      "epoch": 0.4694814648896853,
      "grad_norm": 1.3948652744293213,
      "learning_rate": 4.1895e-05,
      "loss": 0.4609,
      "step": 8384
    },
    {
      "epoch": 0.4695374622018143,
      "grad_norm": 1.2727895975112915,
      "learning_rate": 4.19e-05,
      "loss": 0.3754,
      "step": 8385
    },
    {
      "epoch": 0.4695934595139433,
      "grad_norm": 1.0155041217803955,
      "learning_rate": 4.1905e-05,
      "loss": 0.3234,
      "step": 8386
    },
    {
      "epoch": 0.46964945682607234,
      "grad_norm": 1.5105594396591187,
      "learning_rate": 4.191e-05,
      "loss": 0.4672,
      "step": 8387
    },
    {
      "epoch": 0.46970545413820136,
      "grad_norm": 1.1224207878112793,
      "learning_rate": 4.1915e-05,
      "loss": 0.483,
      "step": 8388
    },
    {
      "epoch": 0.4697614514503304,
      "grad_norm": 1.2138935327529907,
      "learning_rate": 4.1920000000000005e-05,
      "loss": 0.3697,
      "step": 8389
    },
    {
      "epoch": 0.4698174487624594,
      "grad_norm": 1.4979029893875122,
      "learning_rate": 4.1925e-05,
      "loss": 0.6758,
      "step": 8390
    },
    {
      "epoch": 0.4698734460745884,
      "grad_norm": 1.3279343843460083,
      "learning_rate": 4.193e-05,
      "loss": 0.3267,
      "step": 8391
    },
    {
      "epoch": 0.46992944338671744,
      "grad_norm": 1.2012410163879395,
      "learning_rate": 4.1935000000000004e-05,
      "loss": 0.3242,
      "step": 8392
    },
    {
      "epoch": 0.46998544069884646,
      "grad_norm": 1.4319946765899658,
      "learning_rate": 4.194e-05,
      "loss": 0.601,
      "step": 8393
    },
    {
      "epoch": 0.4700414380109755,
      "grad_norm": 1.10093355178833,
      "learning_rate": 4.1945e-05,
      "loss": 0.4318,
      "step": 8394
    },
    {
      "epoch": 0.4700974353231045,
      "grad_norm": 1.4722977876663208,
      "learning_rate": 4.195e-05,
      "loss": 0.5919,
      "step": 8395
    },
    {
      "epoch": 0.4701534326352335,
      "grad_norm": 1.5779893398284912,
      "learning_rate": 4.1955e-05,
      "loss": 0.4911,
      "step": 8396
    },
    {
      "epoch": 0.47020942994736253,
      "grad_norm": 1.1077783107757568,
      "learning_rate": 4.196e-05,
      "loss": 0.4134,
      "step": 8397
    },
    {
      "epoch": 0.47026542725949155,
      "grad_norm": 1.2830719947814941,
      "learning_rate": 4.1965e-05,
      "loss": 0.4634,
      "step": 8398
    },
    {
      "epoch": 0.47032142457162057,
      "grad_norm": 1.5814772844314575,
      "learning_rate": 4.1970000000000006e-05,
      "loss": 0.4829,
      "step": 8399
    },
    {
      "epoch": 0.4703774218837496,
      "grad_norm": 1.3765275478363037,
      "learning_rate": 4.1975000000000004e-05,
      "loss": 0.4915,
      "step": 8400
    },
    {
      "epoch": 0.4704334191958786,
      "grad_norm": 1.1571701765060425,
      "learning_rate": 4.198e-05,
      "loss": 0.3825,
      "step": 8401
    },
    {
      "epoch": 0.47048941650800763,
      "grad_norm": 1.8681832551956177,
      "learning_rate": 4.1985000000000005e-05,
      "loss": 0.7184,
      "step": 8402
    },
    {
      "epoch": 0.47054541382013665,
      "grad_norm": 1.151965618133545,
      "learning_rate": 4.199e-05,
      "loss": 0.3928,
      "step": 8403
    },
    {
      "epoch": 0.47060141113226567,
      "grad_norm": 1.2336509227752686,
      "learning_rate": 4.1995e-05,
      "loss": 0.4089,
      "step": 8404
    },
    {
      "epoch": 0.4706574084443947,
      "grad_norm": 1.2102365493774414,
      "learning_rate": 4.2e-05,
      "loss": 0.4185,
      "step": 8405
    },
    {
      "epoch": 0.4707134057565237,
      "grad_norm": 1.2155449390411377,
      "learning_rate": 4.2005e-05,
      "loss": 0.4994,
      "step": 8406
    },
    {
      "epoch": 0.4707694030686527,
      "grad_norm": 1.1627260446548462,
      "learning_rate": 4.201e-05,
      "loss": 0.4654,
      "step": 8407
    },
    {
      "epoch": 0.47082540038078174,
      "grad_norm": 1.1251270771026611,
      "learning_rate": 4.2015000000000003e-05,
      "loss": 0.4778,
      "step": 8408
    },
    {
      "epoch": 0.47088139769291076,
      "grad_norm": 1.2458096742630005,
      "learning_rate": 4.202e-05,
      "loss": 0.3236,
      "step": 8409
    },
    {
      "epoch": 0.4709373950050398,
      "grad_norm": 1.3230431079864502,
      "learning_rate": 4.2025000000000005e-05,
      "loss": 0.456,
      "step": 8410
    },
    {
      "epoch": 0.4709933923171688,
      "grad_norm": 1.3141199350357056,
      "learning_rate": 4.203e-05,
      "loss": 0.5702,
      "step": 8411
    },
    {
      "epoch": 0.4710493896292978,
      "grad_norm": 1.612363338470459,
      "learning_rate": 4.2035e-05,
      "loss": 0.5587,
      "step": 8412
    },
    {
      "epoch": 0.4711053869414268,
      "grad_norm": 1.1775351762771606,
      "learning_rate": 4.2040000000000004e-05,
      "loss": 0.4286,
      "step": 8413
    },
    {
      "epoch": 0.4711613842535558,
      "grad_norm": 1.340772032737732,
      "learning_rate": 4.2045e-05,
      "loss": 0.476,
      "step": 8414
    },
    {
      "epoch": 0.4712173815656848,
      "grad_norm": 1.1946804523468018,
      "learning_rate": 4.205e-05,
      "loss": 0.5355,
      "step": 8415
    },
    {
      "epoch": 0.47127337887781384,
      "grad_norm": 1.1417064666748047,
      "learning_rate": 4.2055e-05,
      "loss": 0.4496,
      "step": 8416
    },
    {
      "epoch": 0.47132937618994286,
      "grad_norm": 1.1817623376846313,
      "learning_rate": 4.206e-05,
      "loss": 0.4119,
      "step": 8417
    },
    {
      "epoch": 0.4713853735020719,
      "grad_norm": 1.3147894144058228,
      "learning_rate": 4.2065000000000005e-05,
      "loss": 0.5118,
      "step": 8418
    },
    {
      "epoch": 0.4714413708142009,
      "grad_norm": 1.122429370880127,
      "learning_rate": 4.207e-05,
      "loss": 0.3612,
      "step": 8419
    },
    {
      "epoch": 0.4714973681263299,
      "grad_norm": 1.4243937730789185,
      "learning_rate": 4.2075000000000006e-05,
      "loss": 0.5584,
      "step": 8420
    },
    {
      "epoch": 0.47155336543845894,
      "grad_norm": 1.4853757619857788,
      "learning_rate": 4.2080000000000004e-05,
      "loss": 0.3971,
      "step": 8421
    },
    {
      "epoch": 0.47160936275058796,
      "grad_norm": 1.1240071058273315,
      "learning_rate": 4.2085e-05,
      "loss": 0.4124,
      "step": 8422
    },
    {
      "epoch": 0.471665360062717,
      "grad_norm": 1.3261979818344116,
      "learning_rate": 4.209e-05,
      "loss": 0.445,
      "step": 8423
    },
    {
      "epoch": 0.471721357374846,
      "grad_norm": 1.1619199514389038,
      "learning_rate": 4.2095e-05,
      "loss": 0.3402,
      "step": 8424
    },
    {
      "epoch": 0.471777354686975,
      "grad_norm": 1.2499545812606812,
      "learning_rate": 4.21e-05,
      "loss": 0.3863,
      "step": 8425
    },
    {
      "epoch": 0.47183335199910403,
      "grad_norm": 1.130560040473938,
      "learning_rate": 4.2105e-05,
      "loss": 0.3989,
      "step": 8426
    },
    {
      "epoch": 0.47188934931123305,
      "grad_norm": 1.137787938117981,
      "learning_rate": 4.211e-05,
      "loss": 0.3587,
      "step": 8427
    },
    {
      "epoch": 0.47194534662336207,
      "grad_norm": 1.1024013757705688,
      "learning_rate": 4.2115000000000006e-05,
      "loss": 0.416,
      "step": 8428
    },
    {
      "epoch": 0.4720013439354911,
      "grad_norm": 1.1284549236297607,
      "learning_rate": 4.212e-05,
      "loss": 0.4004,
      "step": 8429
    },
    {
      "epoch": 0.4720573412476201,
      "grad_norm": 1.240843415260315,
      "learning_rate": 4.2125e-05,
      "loss": 0.4798,
      "step": 8430
    },
    {
      "epoch": 0.47211333855974913,
      "grad_norm": 1.1080268621444702,
      "learning_rate": 4.2130000000000005e-05,
      "loss": 0.389,
      "step": 8431
    },
    {
      "epoch": 0.47216933587187815,
      "grad_norm": 1.2600219249725342,
      "learning_rate": 4.2135e-05,
      "loss": 0.4573,
      "step": 8432
    },
    {
      "epoch": 0.47222533318400717,
      "grad_norm": 1.1538052558898926,
      "learning_rate": 4.214e-05,
      "loss": 0.305,
      "step": 8433
    },
    {
      "epoch": 0.4722813304961362,
      "grad_norm": 1.1136164665222168,
      "learning_rate": 4.2145000000000004e-05,
      "loss": 0.4308,
      "step": 8434
    },
    {
      "epoch": 0.4723373278082652,
      "grad_norm": 1.4552466869354248,
      "learning_rate": 4.215e-05,
      "loss": 0.491,
      "step": 8435
    },
    {
      "epoch": 0.4723933251203942,
      "grad_norm": 1.4417535066604614,
      "learning_rate": 4.2155e-05,
      "loss": 0.4447,
      "step": 8436
    },
    {
      "epoch": 0.47244932243252324,
      "grad_norm": 1.243757724761963,
      "learning_rate": 4.2159999999999996e-05,
      "loss": 0.3696,
      "step": 8437
    },
    {
      "epoch": 0.47250531974465226,
      "grad_norm": 1.1235696077346802,
      "learning_rate": 4.216500000000001e-05,
      "loss": 0.3932,
      "step": 8438
    },
    {
      "epoch": 0.4725613170567813,
      "grad_norm": 1.2974927425384521,
      "learning_rate": 4.2170000000000005e-05,
      "loss": 0.4324,
      "step": 8439
    },
    {
      "epoch": 0.4726173143689103,
      "grad_norm": 1.5832685232162476,
      "learning_rate": 4.2175e-05,
      "loss": 0.5752,
      "step": 8440
    },
    {
      "epoch": 0.4726733116810393,
      "grad_norm": 1.3697322607040405,
      "learning_rate": 4.2180000000000006e-05,
      "loss": 0.3736,
      "step": 8441
    },
    {
      "epoch": 0.47272930899316834,
      "grad_norm": 1.3663676977157593,
      "learning_rate": 4.2185000000000004e-05,
      "loss": 0.5753,
      "step": 8442
    },
    {
      "epoch": 0.47278530630529736,
      "grad_norm": 1.1377853155136108,
      "learning_rate": 4.219e-05,
      "loss": 0.4741,
      "step": 8443
    },
    {
      "epoch": 0.4728413036174264,
      "grad_norm": 1.0805935859680176,
      "learning_rate": 4.2195e-05,
      "loss": 0.4755,
      "step": 8444
    },
    {
      "epoch": 0.4728973009295554,
      "grad_norm": 1.3271127939224243,
      "learning_rate": 4.22e-05,
      "loss": 0.4224,
      "step": 8445
    },
    {
      "epoch": 0.4729532982416844,
      "grad_norm": 1.1402291059494019,
      "learning_rate": 4.2205e-05,
      "loss": 0.6048,
      "step": 8446
    },
    {
      "epoch": 0.47300929555381344,
      "grad_norm": 1.3297299146652222,
      "learning_rate": 4.221e-05,
      "loss": 0.4885,
      "step": 8447
    },
    {
      "epoch": 0.47306529286594245,
      "grad_norm": 1.0995362997055054,
      "learning_rate": 4.2215e-05,
      "loss": 0.4252,
      "step": 8448
    },
    {
      "epoch": 0.4731212901780715,
      "grad_norm": 1.1222656965255737,
      "learning_rate": 4.2220000000000006e-05,
      "loss": 0.3197,
      "step": 8449
    },
    {
      "epoch": 0.4731772874902005,
      "grad_norm": 1.27969229221344,
      "learning_rate": 4.2225e-05,
      "loss": 0.4638,
      "step": 8450
    },
    {
      "epoch": 0.4732332848023295,
      "grad_norm": 1.1901013851165771,
      "learning_rate": 4.223e-05,
      "loss": 0.4299,
      "step": 8451
    },
    {
      "epoch": 0.47328928211445853,
      "grad_norm": 1.2603455781936646,
      "learning_rate": 4.2235000000000005e-05,
      "loss": 0.3355,
      "step": 8452
    },
    {
      "epoch": 0.47334527942658755,
      "grad_norm": 1.155975341796875,
      "learning_rate": 4.224e-05,
      "loss": 0.3084,
      "step": 8453
    },
    {
      "epoch": 0.4734012767387165,
      "grad_norm": 1.3510417938232422,
      "learning_rate": 4.2245e-05,
      "loss": 0.5243,
      "step": 8454
    },
    {
      "epoch": 0.47345727405084553,
      "grad_norm": 1.082627773284912,
      "learning_rate": 4.2250000000000004e-05,
      "loss": 0.3634,
      "step": 8455
    },
    {
      "epoch": 0.47351327136297455,
      "grad_norm": 1.5972204208374023,
      "learning_rate": 4.2255e-05,
      "loss": 0.4188,
      "step": 8456
    },
    {
      "epoch": 0.47356926867510357,
      "grad_norm": 1.4694019556045532,
      "learning_rate": 4.226e-05,
      "loss": 0.3224,
      "step": 8457
    },
    {
      "epoch": 0.4736252659872326,
      "grad_norm": 1.2422003746032715,
      "learning_rate": 4.2265e-05,
      "loss": 0.5619,
      "step": 8458
    },
    {
      "epoch": 0.4736812632993616,
      "grad_norm": 1.2314832210540771,
      "learning_rate": 4.227000000000001e-05,
      "loss": 0.3839,
      "step": 8459
    },
    {
      "epoch": 0.47373726061149063,
      "grad_norm": 1.3565618991851807,
      "learning_rate": 4.2275000000000004e-05,
      "loss": 0.4406,
      "step": 8460
    },
    {
      "epoch": 0.47379325792361965,
      "grad_norm": 3.4537088871002197,
      "learning_rate": 4.228e-05,
      "loss": 0.5471,
      "step": 8461
    },
    {
      "epoch": 0.47384925523574867,
      "grad_norm": 1.3622307777404785,
      "learning_rate": 4.2285e-05,
      "loss": 0.4902,
      "step": 8462
    },
    {
      "epoch": 0.4739052525478777,
      "grad_norm": 1.2124581336975098,
      "learning_rate": 4.229e-05,
      "loss": 0.4515,
      "step": 8463
    },
    {
      "epoch": 0.4739612498600067,
      "grad_norm": 1.5086095333099365,
      "learning_rate": 4.2295e-05,
      "loss": 0.4574,
      "step": 8464
    },
    {
      "epoch": 0.4740172471721357,
      "grad_norm": 1.3772732019424438,
      "learning_rate": 4.23e-05,
      "loss": 0.4398,
      "step": 8465
    },
    {
      "epoch": 0.47407324448426474,
      "grad_norm": 1.1363698244094849,
      "learning_rate": 4.2305e-05,
      "loss": 0.4022,
      "step": 8466
    },
    {
      "epoch": 0.47412924179639376,
      "grad_norm": 1.329436182975769,
      "learning_rate": 4.231e-05,
      "loss": 0.4389,
      "step": 8467
    },
    {
      "epoch": 0.4741852391085228,
      "grad_norm": 1.2213752269744873,
      "learning_rate": 4.2315000000000004e-05,
      "loss": 0.3593,
      "step": 8468
    },
    {
      "epoch": 0.4742412364206518,
      "grad_norm": 1.3433914184570312,
      "learning_rate": 4.232e-05,
      "loss": 0.381,
      "step": 8469
    },
    {
      "epoch": 0.4742972337327808,
      "grad_norm": 1.1535258293151855,
      "learning_rate": 4.2325000000000006e-05,
      "loss": 0.3721,
      "step": 8470
    },
    {
      "epoch": 0.47435323104490984,
      "grad_norm": 1.349773645401001,
      "learning_rate": 4.233e-05,
      "loss": 0.4592,
      "step": 8471
    },
    {
      "epoch": 0.47440922835703886,
      "grad_norm": 1.6597414016723633,
      "learning_rate": 4.2335e-05,
      "loss": 0.4148,
      "step": 8472
    },
    {
      "epoch": 0.4744652256691679,
      "grad_norm": 1.2057273387908936,
      "learning_rate": 4.2340000000000005e-05,
      "loss": 0.3378,
      "step": 8473
    },
    {
      "epoch": 0.4745212229812969,
      "grad_norm": 1.6307529211044312,
      "learning_rate": 4.2345e-05,
      "loss": 0.421,
      "step": 8474
    },
    {
      "epoch": 0.4745772202934259,
      "grad_norm": 1.352099895477295,
      "learning_rate": 4.235e-05,
      "loss": 0.3547,
      "step": 8475
    },
    {
      "epoch": 0.47463321760555494,
      "grad_norm": 1.4897671937942505,
      "learning_rate": 4.2355000000000004e-05,
      "loss": 0.4558,
      "step": 8476
    },
    {
      "epoch": 0.47468921491768395,
      "grad_norm": 1.261129379272461,
      "learning_rate": 4.236e-05,
      "loss": 0.42,
      "step": 8477
    },
    {
      "epoch": 0.474745212229813,
      "grad_norm": 1.3802372217178345,
      "learning_rate": 4.2365000000000005e-05,
      "loss": 0.3786,
      "step": 8478
    },
    {
      "epoch": 0.474801209541942,
      "grad_norm": 1.2289320230484009,
      "learning_rate": 4.237e-05,
      "loss": 0.4587,
      "step": 8479
    },
    {
      "epoch": 0.474857206854071,
      "grad_norm": 1.220415711402893,
      "learning_rate": 4.237500000000001e-05,
      "loss": 0.4528,
      "step": 8480
    },
    {
      "epoch": 0.47491320416620003,
      "grad_norm": 1.6037752628326416,
      "learning_rate": 4.2380000000000004e-05,
      "loss": 0.4565,
      "step": 8481
    },
    {
      "epoch": 0.47496920147832905,
      "grad_norm": 1.3115330934524536,
      "learning_rate": 4.2385e-05,
      "loss": 0.4804,
      "step": 8482
    },
    {
      "epoch": 0.47502519879045807,
      "grad_norm": 1.4712803363800049,
      "learning_rate": 4.239e-05,
      "loss": 0.3827,
      "step": 8483
    },
    {
      "epoch": 0.4750811961025871,
      "grad_norm": 1.428632378578186,
      "learning_rate": 4.2395e-05,
      "loss": 0.5751,
      "step": 8484
    },
    {
      "epoch": 0.4751371934147161,
      "grad_norm": 1.1504849195480347,
      "learning_rate": 4.24e-05,
      "loss": 0.513,
      "step": 8485
    },
    {
      "epoch": 0.4751931907268451,
      "grad_norm": 1.1852437257766724,
      "learning_rate": 4.2405e-05,
      "loss": 0.5621,
      "step": 8486
    },
    {
      "epoch": 0.47524918803897415,
      "grad_norm": 1.389432430267334,
      "learning_rate": 4.241e-05,
      "loss": 0.264,
      "step": 8487
    },
    {
      "epoch": 0.47530518535110317,
      "grad_norm": 1.338192105293274,
      "learning_rate": 4.2415000000000006e-05,
      "loss": 0.4091,
      "step": 8488
    },
    {
      "epoch": 0.4753611826632322,
      "grad_norm": 1.1868879795074463,
      "learning_rate": 4.2420000000000004e-05,
      "loss": 0.4872,
      "step": 8489
    },
    {
      "epoch": 0.4754171799753612,
      "grad_norm": 1.2049229145050049,
      "learning_rate": 4.2425e-05,
      "loss": 0.3504,
      "step": 8490
    },
    {
      "epoch": 0.4754731772874902,
      "grad_norm": 1.1247048377990723,
      "learning_rate": 4.2430000000000005e-05,
      "loss": 0.3739,
      "step": 8491
    },
    {
      "epoch": 0.47552917459961924,
      "grad_norm": 1.4636249542236328,
      "learning_rate": 4.2435e-05,
      "loss": 0.4197,
      "step": 8492
    },
    {
      "epoch": 0.47558517191174826,
      "grad_norm": 1.7594317197799683,
      "learning_rate": 4.244e-05,
      "loss": 0.4573,
      "step": 8493
    },
    {
      "epoch": 0.4756411692238773,
      "grad_norm": 1.2627742290496826,
      "learning_rate": 4.2445000000000004e-05,
      "loss": 0.4462,
      "step": 8494
    },
    {
      "epoch": 0.47569716653600624,
      "grad_norm": 1.0841120481491089,
      "learning_rate": 4.245e-05,
      "loss": 0.3393,
      "step": 8495
    },
    {
      "epoch": 0.47575316384813526,
      "grad_norm": 1.362037181854248,
      "learning_rate": 4.2455e-05,
      "loss": 0.6053,
      "step": 8496
    },
    {
      "epoch": 0.4758091611602643,
      "grad_norm": 1.2184008359909058,
      "learning_rate": 4.246e-05,
      "loss": 0.4431,
      "step": 8497
    },
    {
      "epoch": 0.4758651584723933,
      "grad_norm": 1.2519798278808594,
      "learning_rate": 4.246500000000001e-05,
      "loss": 0.4529,
      "step": 8498
    },
    {
      "epoch": 0.4759211557845223,
      "grad_norm": 1.2431145906448364,
      "learning_rate": 4.2470000000000005e-05,
      "loss": 0.366,
      "step": 8499
    },
    {
      "epoch": 0.47597715309665134,
      "grad_norm": 1.226942539215088,
      "learning_rate": 4.2475e-05,
      "loss": 0.5221,
      "step": 8500
    },
    {
      "epoch": 0.47603315040878036,
      "grad_norm": 1.2117550373077393,
      "learning_rate": 4.248e-05,
      "loss": 0.5224,
      "step": 8501
    },
    {
      "epoch": 0.4760891477209094,
      "grad_norm": 1.5471678972244263,
      "learning_rate": 4.2485000000000004e-05,
      "loss": 0.5953,
      "step": 8502
    },
    {
      "epoch": 0.4761451450330384,
      "grad_norm": 1.3753056526184082,
      "learning_rate": 4.249e-05,
      "loss": 0.5152,
      "step": 8503
    },
    {
      "epoch": 0.4762011423451674,
      "grad_norm": 1.514878511428833,
      "learning_rate": 4.2495e-05,
      "loss": 0.7663,
      "step": 8504
    },
    {
      "epoch": 0.47625713965729644,
      "grad_norm": 1.3028452396392822,
      "learning_rate": 4.25e-05,
      "loss": 0.4354,
      "step": 8505
    },
    {
      "epoch": 0.47631313696942545,
      "grad_norm": 1.2990000247955322,
      "learning_rate": 4.2505e-05,
      "loss": 0.3929,
      "step": 8506
    },
    {
      "epoch": 0.4763691342815545,
      "grad_norm": 1.1618504524230957,
      "learning_rate": 4.251e-05,
      "loss": 0.3683,
      "step": 8507
    },
    {
      "epoch": 0.4764251315936835,
      "grad_norm": 1.2800405025482178,
      "learning_rate": 4.2515e-05,
      "loss": 0.4431,
      "step": 8508
    },
    {
      "epoch": 0.4764811289058125,
      "grad_norm": 1.145199179649353,
      "learning_rate": 4.2520000000000006e-05,
      "loss": 0.4034,
      "step": 8509
    },
    {
      "epoch": 0.47653712621794153,
      "grad_norm": 1.1259734630584717,
      "learning_rate": 4.2525000000000004e-05,
      "loss": 0.453,
      "step": 8510
    },
    {
      "epoch": 0.47659312353007055,
      "grad_norm": 1.203216791152954,
      "learning_rate": 4.253e-05,
      "loss": 0.4499,
      "step": 8511
    },
    {
      "epoch": 0.47664912084219957,
      "grad_norm": 1.1637547016143799,
      "learning_rate": 4.2535000000000005e-05,
      "loss": 0.415,
      "step": 8512
    },
    {
      "epoch": 0.4767051181543286,
      "grad_norm": 1.1568801403045654,
      "learning_rate": 4.254e-05,
      "loss": 0.4233,
      "step": 8513
    },
    {
      "epoch": 0.4767611154664576,
      "grad_norm": 1.6813509464263916,
      "learning_rate": 4.2545e-05,
      "loss": 0.5052,
      "step": 8514
    },
    {
      "epoch": 0.4768171127785866,
      "grad_norm": 1.1524626016616821,
      "learning_rate": 4.2550000000000004e-05,
      "loss": 0.4701,
      "step": 8515
    },
    {
      "epoch": 0.47687311009071565,
      "grad_norm": 1.2931450605392456,
      "learning_rate": 4.2555e-05,
      "loss": 0.3449,
      "step": 8516
    },
    {
      "epoch": 0.47692910740284467,
      "grad_norm": 1.2546114921569824,
      "learning_rate": 4.256e-05,
      "loss": 0.41,
      "step": 8517
    },
    {
      "epoch": 0.4769851047149737,
      "grad_norm": 1.379087209701538,
      "learning_rate": 4.2564999999999997e-05,
      "loss": 0.5385,
      "step": 8518
    },
    {
      "epoch": 0.4770411020271027,
      "grad_norm": 1.0247600078582764,
      "learning_rate": 4.257000000000001e-05,
      "loss": 0.3114,
      "step": 8519
    },
    {
      "epoch": 0.4770970993392317,
      "grad_norm": 1.1872286796569824,
      "learning_rate": 4.2575000000000005e-05,
      "loss": 0.3339,
      "step": 8520
    },
    {
      "epoch": 0.47715309665136074,
      "grad_norm": 1.1792237758636475,
      "learning_rate": 4.258e-05,
      "loss": 0.6009,
      "step": 8521
    },
    {
      "epoch": 0.47720909396348976,
      "grad_norm": 1.1308717727661133,
      "learning_rate": 4.2585e-05,
      "loss": 0.4187,
      "step": 8522
    },
    {
      "epoch": 0.4772650912756188,
      "grad_norm": 1.3926209211349487,
      "learning_rate": 4.2590000000000004e-05,
      "loss": 0.4622,
      "step": 8523
    },
    {
      "epoch": 0.4773210885877478,
      "grad_norm": 1.264739990234375,
      "learning_rate": 4.2595e-05,
      "loss": 0.3727,
      "step": 8524
    },
    {
      "epoch": 0.4773770858998768,
      "grad_norm": 1.2509269714355469,
      "learning_rate": 4.26e-05,
      "loss": 0.5148,
      "step": 8525
    },
    {
      "epoch": 0.47743308321200584,
      "grad_norm": 1.2540090084075928,
      "learning_rate": 4.2605e-05,
      "loss": 0.4493,
      "step": 8526
    },
    {
      "epoch": 0.47748908052413486,
      "grad_norm": 1.2280350923538208,
      "learning_rate": 4.261e-05,
      "loss": 0.3851,
      "step": 8527
    },
    {
      "epoch": 0.4775450778362639,
      "grad_norm": 1.5690923929214478,
      "learning_rate": 4.2615e-05,
      "loss": 0.3389,
      "step": 8528
    },
    {
      "epoch": 0.4776010751483929,
      "grad_norm": 1.2065706253051758,
      "learning_rate": 4.262e-05,
      "loss": 0.4382,
      "step": 8529
    },
    {
      "epoch": 0.4776570724605219,
      "grad_norm": 1.1458523273468018,
      "learning_rate": 4.2625000000000006e-05,
      "loss": 0.3564,
      "step": 8530
    },
    {
      "epoch": 0.47771306977265093,
      "grad_norm": 1.0513176918029785,
      "learning_rate": 4.2630000000000004e-05,
      "loss": 0.3677,
      "step": 8531
    },
    {
      "epoch": 0.47776906708477995,
      "grad_norm": 1.4468929767608643,
      "learning_rate": 4.2635e-05,
      "loss": 0.5722,
      "step": 8532
    },
    {
      "epoch": 0.47782506439690897,
      "grad_norm": 1.1844818592071533,
      "learning_rate": 4.2640000000000005e-05,
      "loss": 0.3671,
      "step": 8533
    },
    {
      "epoch": 0.477881061709038,
      "grad_norm": 1.572331428527832,
      "learning_rate": 4.2645e-05,
      "loss": 0.4609,
      "step": 8534
    },
    {
      "epoch": 0.477937059021167,
      "grad_norm": 1.3428393602371216,
      "learning_rate": 4.265e-05,
      "loss": 0.5407,
      "step": 8535
    },
    {
      "epoch": 0.47799305633329603,
      "grad_norm": 1.227597713470459,
      "learning_rate": 4.2655e-05,
      "loss": 0.4428,
      "step": 8536
    },
    {
      "epoch": 0.478049053645425,
      "grad_norm": 1.580487608909607,
      "learning_rate": 4.266e-05,
      "loss": 0.3856,
      "step": 8537
    },
    {
      "epoch": 0.478105050957554,
      "grad_norm": 1.3850281238555908,
      "learning_rate": 4.2665e-05,
      "loss": 0.4733,
      "step": 8538
    },
    {
      "epoch": 0.47816104826968303,
      "grad_norm": 1.440145492553711,
      "learning_rate": 4.267e-05,
      "loss": 0.655,
      "step": 8539
    },
    {
      "epoch": 0.47821704558181205,
      "grad_norm": 1.1501185894012451,
      "learning_rate": 4.2675e-05,
      "loss": 0.4654,
      "step": 8540
    },
    {
      "epoch": 0.47827304289394107,
      "grad_norm": 1.074999213218689,
      "learning_rate": 4.2680000000000005e-05,
      "loss": 0.3818,
      "step": 8541
    },
    {
      "epoch": 0.4783290402060701,
      "grad_norm": 1.189826488494873,
      "learning_rate": 4.2685e-05,
      "loss": 0.4428,
      "step": 8542
    },
    {
      "epoch": 0.4783850375181991,
      "grad_norm": 1.0231224298477173,
      "learning_rate": 4.269e-05,
      "loss": 0.3145,
      "step": 8543
    },
    {
      "epoch": 0.4784410348303281,
      "grad_norm": 1.1621557474136353,
      "learning_rate": 4.2695000000000004e-05,
      "loss": 0.3708,
      "step": 8544
    },
    {
      "epoch": 0.47849703214245715,
      "grad_norm": 1.0861577987670898,
      "learning_rate": 4.27e-05,
      "loss": 0.444,
      "step": 8545
    },
    {
      "epoch": 0.47855302945458617,
      "grad_norm": 1.2717492580413818,
      "learning_rate": 4.2705e-05,
      "loss": 0.413,
      "step": 8546
    },
    {
      "epoch": 0.4786090267667152,
      "grad_norm": 1.286747694015503,
      "learning_rate": 4.271e-05,
      "loss": 0.4528,
      "step": 8547
    },
    {
      "epoch": 0.4786650240788442,
      "grad_norm": 1.2454153299331665,
      "learning_rate": 4.2715e-05,
      "loss": 0.4464,
      "step": 8548
    },
    {
      "epoch": 0.4787210213909732,
      "grad_norm": 1.143328070640564,
      "learning_rate": 4.2720000000000004e-05,
      "loss": 0.3162,
      "step": 8549
    },
    {
      "epoch": 0.47877701870310224,
      "grad_norm": 1.269060492515564,
      "learning_rate": 4.2725e-05,
      "loss": 0.6328,
      "step": 8550
    },
    {
      "epoch": 0.47883301601523126,
      "grad_norm": 1.2941027879714966,
      "learning_rate": 4.2730000000000006e-05,
      "loss": 0.4825,
      "step": 8551
    },
    {
      "epoch": 0.4788890133273603,
      "grad_norm": 1.3494073152542114,
      "learning_rate": 4.2735e-05,
      "loss": 0.5508,
      "step": 8552
    },
    {
      "epoch": 0.4789450106394893,
      "grad_norm": 1.3683594465255737,
      "learning_rate": 4.274e-05,
      "loss": 0.3838,
      "step": 8553
    },
    {
      "epoch": 0.4790010079516183,
      "grad_norm": 1.2412092685699463,
      "learning_rate": 4.2745000000000005e-05,
      "loss": 0.401,
      "step": 8554
    },
    {
      "epoch": 0.47905700526374734,
      "grad_norm": 1.2897952795028687,
      "learning_rate": 4.275e-05,
      "loss": 0.4075,
      "step": 8555
    },
    {
      "epoch": 0.47911300257587636,
      "grad_norm": 1.2257930040359497,
      "learning_rate": 4.2755e-05,
      "loss": 0.4128,
      "step": 8556
    },
    {
      "epoch": 0.4791689998880054,
      "grad_norm": 1.193520188331604,
      "learning_rate": 4.276e-05,
      "loss": 0.3751,
      "step": 8557
    },
    {
      "epoch": 0.4792249972001344,
      "grad_norm": 1.2806997299194336,
      "learning_rate": 4.2765e-05,
      "loss": 0.3365,
      "step": 8558
    },
    {
      "epoch": 0.4792809945122634,
      "grad_norm": 1.281146764755249,
      "learning_rate": 4.2770000000000006e-05,
      "loss": 0.4771,
      "step": 8559
    },
    {
      "epoch": 0.47933699182439243,
      "grad_norm": 1.3099502325057983,
      "learning_rate": 4.2775e-05,
      "loss": 0.3065,
      "step": 8560
    },
    {
      "epoch": 0.47939298913652145,
      "grad_norm": 1.3549703359603882,
      "learning_rate": 4.278e-05,
      "loss": 0.3679,
      "step": 8561
    },
    {
      "epoch": 0.47944898644865047,
      "grad_norm": 1.138702154159546,
      "learning_rate": 4.2785000000000005e-05,
      "loss": 0.3429,
      "step": 8562
    },
    {
      "epoch": 0.4795049837607795,
      "grad_norm": 1.219333529472351,
      "learning_rate": 4.279e-05,
      "loss": 0.2814,
      "step": 8563
    },
    {
      "epoch": 0.4795609810729085,
      "grad_norm": 1.2598845958709717,
      "learning_rate": 4.2795e-05,
      "loss": 0.35,
      "step": 8564
    },
    {
      "epoch": 0.47961697838503753,
      "grad_norm": 1.1545138359069824,
      "learning_rate": 4.2800000000000004e-05,
      "loss": 0.4768,
      "step": 8565
    },
    {
      "epoch": 0.47967297569716655,
      "grad_norm": 1.149704933166504,
      "learning_rate": 4.2805e-05,
      "loss": 0.5022,
      "step": 8566
    },
    {
      "epoch": 0.47972897300929557,
      "grad_norm": 1.1559488773345947,
      "learning_rate": 4.281e-05,
      "loss": 0.368,
      "step": 8567
    },
    {
      "epoch": 0.4797849703214246,
      "grad_norm": 1.2474576234817505,
      "learning_rate": 4.2815e-05,
      "loss": 0.3818,
      "step": 8568
    },
    {
      "epoch": 0.4798409676335536,
      "grad_norm": 1.2686368227005005,
      "learning_rate": 4.282000000000001e-05,
      "loss": 0.3357,
      "step": 8569
    },
    {
      "epoch": 0.4798969649456826,
      "grad_norm": 1.383123755455017,
      "learning_rate": 4.2825000000000004e-05,
      "loss": 0.4867,
      "step": 8570
    },
    {
      "epoch": 0.47995296225781164,
      "grad_norm": 1.1608155965805054,
      "learning_rate": 4.283e-05,
      "loss": 0.4371,
      "step": 8571
    },
    {
      "epoch": 0.48000895956994066,
      "grad_norm": 1.2059683799743652,
      "learning_rate": 4.2835000000000006e-05,
      "loss": 0.4577,
      "step": 8572
    },
    {
      "epoch": 0.4800649568820697,
      "grad_norm": 1.4019975662231445,
      "learning_rate": 4.284e-05,
      "loss": 0.38,
      "step": 8573
    },
    {
      "epoch": 0.4801209541941987,
      "grad_norm": 1.38151216506958,
      "learning_rate": 4.2845e-05,
      "loss": 0.5864,
      "step": 8574
    },
    {
      "epoch": 0.4801769515063277,
      "grad_norm": 1.21138334274292,
      "learning_rate": 4.285e-05,
      "loss": 0.4912,
      "step": 8575
    },
    {
      "epoch": 0.48023294881845674,
      "grad_norm": 1.4562879800796509,
      "learning_rate": 4.2855e-05,
      "loss": 0.427,
      "step": 8576
    },
    {
      "epoch": 0.48028894613058576,
      "grad_norm": 1.2790573835372925,
      "learning_rate": 4.286e-05,
      "loss": 0.4197,
      "step": 8577
    },
    {
      "epoch": 0.4803449434427147,
      "grad_norm": 1.2116519212722778,
      "learning_rate": 4.2865e-05,
      "loss": 0.3213,
      "step": 8578
    },
    {
      "epoch": 0.48040094075484374,
      "grad_norm": 1.5755831003189087,
      "learning_rate": 4.287000000000001e-05,
      "loss": 0.5585,
      "step": 8579
    },
    {
      "epoch": 0.48045693806697276,
      "grad_norm": 1.244770884513855,
      "learning_rate": 4.2875000000000005e-05,
      "loss": 0.4356,
      "step": 8580
    },
    {
      "epoch": 0.4805129353791018,
      "grad_norm": 1.1025171279907227,
      "learning_rate": 4.288e-05,
      "loss": 0.4699,
      "step": 8581
    },
    {
      "epoch": 0.4805689326912308,
      "grad_norm": 1.2222015857696533,
      "learning_rate": 4.2885e-05,
      "loss": 0.3861,
      "step": 8582
    },
    {
      "epoch": 0.4806249300033598,
      "grad_norm": 1.071301817893982,
      "learning_rate": 4.2890000000000004e-05,
      "loss": 0.3008,
      "step": 8583
    },
    {
      "epoch": 0.48068092731548884,
      "grad_norm": 1.2025933265686035,
      "learning_rate": 4.2895e-05,
      "loss": 0.3804,
      "step": 8584
    },
    {
      "epoch": 0.48073692462761786,
      "grad_norm": 1.428401231765747,
      "learning_rate": 4.29e-05,
      "loss": 0.4237,
      "step": 8585
    },
    {
      "epoch": 0.4807929219397469,
      "grad_norm": 1.6277501583099365,
      "learning_rate": 4.2905000000000003e-05,
      "loss": 0.4789,
      "step": 8586
    },
    {
      "epoch": 0.4808489192518759,
      "grad_norm": 1.3520350456237793,
      "learning_rate": 4.291e-05,
      "loss": 0.5518,
      "step": 8587
    },
    {
      "epoch": 0.4809049165640049,
      "grad_norm": 1.2806624174118042,
      "learning_rate": 4.2915e-05,
      "loss": 0.522,
      "step": 8588
    },
    {
      "epoch": 0.48096091387613393,
      "grad_norm": 1.2821451425552368,
      "learning_rate": 4.292e-05,
      "loss": 0.4721,
      "step": 8589
    },
    {
      "epoch": 0.48101691118826295,
      "grad_norm": 1.2855767011642456,
      "learning_rate": 4.2925000000000007e-05,
      "loss": 0.3422,
      "step": 8590
    },
    {
      "epoch": 0.48107290850039197,
      "grad_norm": 1.1831409931182861,
      "learning_rate": 4.2930000000000004e-05,
      "loss": 0.4138,
      "step": 8591
    },
    {
      "epoch": 0.481128905812521,
      "grad_norm": 1.2454614639282227,
      "learning_rate": 4.2935e-05,
      "loss": 0.4348,
      "step": 8592
    },
    {
      "epoch": 0.48118490312465,
      "grad_norm": 1.504920482635498,
      "learning_rate": 4.2940000000000006e-05,
      "loss": 0.5542,
      "step": 8593
    },
    {
      "epoch": 0.48124090043677903,
      "grad_norm": 1.6143132448196411,
      "learning_rate": 4.2945e-05,
      "loss": 0.4956,
      "step": 8594
    },
    {
      "epoch": 0.48129689774890805,
      "grad_norm": 1.2793389558792114,
      "learning_rate": 4.295e-05,
      "loss": 0.385,
      "step": 8595
    },
    {
      "epoch": 0.48135289506103707,
      "grad_norm": 1.141887903213501,
      "learning_rate": 4.2955e-05,
      "loss": 0.3376,
      "step": 8596
    },
    {
      "epoch": 0.4814088923731661,
      "grad_norm": 1.3110154867172241,
      "learning_rate": 4.296e-05,
      "loss": 0.4257,
      "step": 8597
    },
    {
      "epoch": 0.4814648896852951,
      "grad_norm": 1.4544636011123657,
      "learning_rate": 4.2965e-05,
      "loss": 0.4742,
      "step": 8598
    },
    {
      "epoch": 0.4815208869974241,
      "grad_norm": 1.5760170221328735,
      "learning_rate": 4.2970000000000004e-05,
      "loss": 0.5291,
      "step": 8599
    },
    {
      "epoch": 0.48157688430955314,
      "grad_norm": 1.3123674392700195,
      "learning_rate": 4.2975e-05,
      "loss": 0.4305,
      "step": 8600
    },
    {
      "epoch": 0.48163288162168216,
      "grad_norm": 1.5963950157165527,
      "learning_rate": 4.2980000000000005e-05,
      "loss": 0.5198,
      "step": 8601
    },
    {
      "epoch": 0.4816888789338112,
      "grad_norm": 1.2458404302597046,
      "learning_rate": 4.2985e-05,
      "loss": 0.492,
      "step": 8602
    },
    {
      "epoch": 0.4817448762459402,
      "grad_norm": 1.2439574003219604,
      "learning_rate": 4.299e-05,
      "loss": 0.5217,
      "step": 8603
    },
    {
      "epoch": 0.4818008735580692,
      "grad_norm": 1.4142569303512573,
      "learning_rate": 4.2995000000000004e-05,
      "loss": 0.3955,
      "step": 8604
    },
    {
      "epoch": 0.48185687087019824,
      "grad_norm": 1.3641350269317627,
      "learning_rate": 4.3e-05,
      "loss": 0.5114,
      "step": 8605
    },
    {
      "epoch": 0.48191286818232726,
      "grad_norm": 1.1383672952651978,
      "learning_rate": 4.3005e-05,
      "loss": 0.411,
      "step": 8606
    },
    {
      "epoch": 0.4819688654944563,
      "grad_norm": 1.6961710453033447,
      "learning_rate": 4.301e-05,
      "loss": 0.4136,
      "step": 8607
    },
    {
      "epoch": 0.4820248628065853,
      "grad_norm": 1.1815787553787231,
      "learning_rate": 4.3015e-05,
      "loss": 0.3027,
      "step": 8608
    },
    {
      "epoch": 0.4820808601187143,
      "grad_norm": 1.0456570386886597,
      "learning_rate": 4.3020000000000005e-05,
      "loss": 0.378,
      "step": 8609
    },
    {
      "epoch": 0.48213685743084334,
      "grad_norm": 1.0018742084503174,
      "learning_rate": 4.3025e-05,
      "loss": 0.384,
      "step": 8610
    },
    {
      "epoch": 0.48219285474297235,
      "grad_norm": 1.1864042282104492,
      "learning_rate": 4.3030000000000006e-05,
      "loss": 0.528,
      "step": 8611
    },
    {
      "epoch": 0.4822488520551014,
      "grad_norm": 1.4641215801239014,
      "learning_rate": 4.3035000000000004e-05,
      "loss": 0.5839,
      "step": 8612
    },
    {
      "epoch": 0.4823048493672304,
      "grad_norm": 1.232445478439331,
      "learning_rate": 4.304e-05,
      "loss": 0.449,
      "step": 8613
    },
    {
      "epoch": 0.4823608466793594,
      "grad_norm": 1.4176139831542969,
      "learning_rate": 4.3045e-05,
      "loss": 0.4188,
      "step": 8614
    },
    {
      "epoch": 0.48241684399148843,
      "grad_norm": 2.108597993850708,
      "learning_rate": 4.305e-05,
      "loss": 0.3727,
      "step": 8615
    },
    {
      "epoch": 0.48247284130361745,
      "grad_norm": 1.249324083328247,
      "learning_rate": 4.3055e-05,
      "loss": 0.4655,
      "step": 8616
    },
    {
      "epoch": 0.48252883861574647,
      "grad_norm": 1.4487379789352417,
      "learning_rate": 4.306e-05,
      "loss": 0.4904,
      "step": 8617
    },
    {
      "epoch": 0.4825848359278755,
      "grad_norm": 1.620957612991333,
      "learning_rate": 4.3065e-05,
      "loss": 0.3909,
      "step": 8618
    },
    {
      "epoch": 0.48264083324000445,
      "grad_norm": 1.402370810508728,
      "learning_rate": 4.3070000000000006e-05,
      "loss": 0.6593,
      "step": 8619
    },
    {
      "epoch": 0.48269683055213347,
      "grad_norm": 1.1624068021774292,
      "learning_rate": 4.3075000000000003e-05,
      "loss": 0.4366,
      "step": 8620
    },
    {
      "epoch": 0.4827528278642625,
      "grad_norm": 1.351397156715393,
      "learning_rate": 4.308e-05,
      "loss": 0.5857,
      "step": 8621
    },
    {
      "epoch": 0.4828088251763915,
      "grad_norm": 1.030238151550293,
      "learning_rate": 4.3085000000000005e-05,
      "loss": 0.3954,
      "step": 8622
    },
    {
      "epoch": 0.48286482248852053,
      "grad_norm": 1.4337857961654663,
      "learning_rate": 4.309e-05,
      "loss": 0.4261,
      "step": 8623
    },
    {
      "epoch": 0.48292081980064955,
      "grad_norm": 1.1636183261871338,
      "learning_rate": 4.3095e-05,
      "loss": 0.2829,
      "step": 8624
    },
    {
      "epoch": 0.48297681711277857,
      "grad_norm": 1.1158586740493774,
      "learning_rate": 4.3100000000000004e-05,
      "loss": 0.4353,
      "step": 8625
    },
    {
      "epoch": 0.4830328144249076,
      "grad_norm": 2.748267889022827,
      "learning_rate": 4.3105e-05,
      "loss": 0.4636,
      "step": 8626
    },
    {
      "epoch": 0.4830888117370366,
      "grad_norm": 1.0934284925460815,
      "learning_rate": 4.311e-05,
      "loss": 0.3526,
      "step": 8627
    },
    {
      "epoch": 0.4831448090491656,
      "grad_norm": 1.2375462055206299,
      "learning_rate": 4.3115e-05,
      "loss": 0.444,
      "step": 8628
    },
    {
      "epoch": 0.48320080636129464,
      "grad_norm": 1.3731330633163452,
      "learning_rate": 4.312000000000001e-05,
      "loss": 0.4732,
      "step": 8629
    },
    {
      "epoch": 0.48325680367342366,
      "grad_norm": 1.1914840936660767,
      "learning_rate": 4.3125000000000005e-05,
      "loss": 0.3496,
      "step": 8630
    },
    {
      "epoch": 0.4833128009855527,
      "grad_norm": 1.2739779949188232,
      "learning_rate": 4.313e-05,
      "loss": 0.6161,
      "step": 8631
    },
    {
      "epoch": 0.4833687982976817,
      "grad_norm": 1.5741862058639526,
      "learning_rate": 4.3135000000000006e-05,
      "loss": 0.6004,
      "step": 8632
    },
    {
      "epoch": 0.4834247956098107,
      "grad_norm": 1.4872034788131714,
      "learning_rate": 4.3140000000000004e-05,
      "loss": 0.6158,
      "step": 8633
    },
    {
      "epoch": 0.48348079292193974,
      "grad_norm": 2.9754796028137207,
      "learning_rate": 4.3145e-05,
      "loss": 0.4436,
      "step": 8634
    },
    {
      "epoch": 0.48353679023406876,
      "grad_norm": 1.236561894416809,
      "learning_rate": 4.315e-05,
      "loss": 0.5231,
      "step": 8635
    },
    {
      "epoch": 0.4835927875461978,
      "grad_norm": 1.2106773853302002,
      "learning_rate": 4.3155e-05,
      "loss": 0.3766,
      "step": 8636
    },
    {
      "epoch": 0.4836487848583268,
      "grad_norm": 1.407659649848938,
      "learning_rate": 4.316e-05,
      "loss": 0.5709,
      "step": 8637
    },
    {
      "epoch": 0.4837047821704558,
      "grad_norm": 1.1345996856689453,
      "learning_rate": 4.3165e-05,
      "loss": 0.363,
      "step": 8638
    },
    {
      "epoch": 0.48376077948258484,
      "grad_norm": 1.2164907455444336,
      "learning_rate": 4.317e-05,
      "loss": 0.509,
      "step": 8639
    },
    {
      "epoch": 0.48381677679471385,
      "grad_norm": 1.0939404964447021,
      "learning_rate": 4.3175000000000006e-05,
      "loss": 0.468,
      "step": 8640
    },
    {
      "epoch": 0.4838727741068429,
      "grad_norm": 1.355599045753479,
      "learning_rate": 4.318e-05,
      "loss": 0.4164,
      "step": 8641
    },
    {
      "epoch": 0.4839287714189719,
      "grad_norm": 1.1362967491149902,
      "learning_rate": 4.3185e-05,
      "loss": 0.3244,
      "step": 8642
    },
    {
      "epoch": 0.4839847687311009,
      "grad_norm": 1.142841100692749,
      "learning_rate": 4.3190000000000005e-05,
      "loss": 0.439,
      "step": 8643
    },
    {
      "epoch": 0.48404076604322993,
      "grad_norm": 1.155008316040039,
      "learning_rate": 4.3195e-05,
      "loss": 0.5317,
      "step": 8644
    },
    {
      "epoch": 0.48409676335535895,
      "grad_norm": 1.1879433393478394,
      "learning_rate": 4.32e-05,
      "loss": 0.3743,
      "step": 8645
    },
    {
      "epoch": 0.48415276066748797,
      "grad_norm": 1.4973292350769043,
      "learning_rate": 4.3205000000000004e-05,
      "loss": 0.544,
      "step": 8646
    },
    {
      "epoch": 0.484208757979617,
      "grad_norm": 1.404468059539795,
      "learning_rate": 4.321e-05,
      "loss": 0.4304,
      "step": 8647
    },
    {
      "epoch": 0.484264755291746,
      "grad_norm": 1.1524723768234253,
      "learning_rate": 4.3215e-05,
      "loss": 0.3708,
      "step": 8648
    },
    {
      "epoch": 0.484320752603875,
      "grad_norm": 1.2423940896987915,
      "learning_rate": 4.3219999999999996e-05,
      "loss": 0.509,
      "step": 8649
    },
    {
      "epoch": 0.48437674991600405,
      "grad_norm": 1.4346377849578857,
      "learning_rate": 4.322500000000001e-05,
      "loss": 0.3965,
      "step": 8650
    },
    {
      "epoch": 0.48443274722813306,
      "grad_norm": 1.4229187965393066,
      "learning_rate": 4.3230000000000005e-05,
      "loss": 0.4362,
      "step": 8651
    },
    {
      "epoch": 0.4844887445402621,
      "grad_norm": 1.2607970237731934,
      "learning_rate": 4.3235e-05,
      "loss": 0.4363,
      "step": 8652
    },
    {
      "epoch": 0.4845447418523911,
      "grad_norm": 1.0456748008728027,
      "learning_rate": 4.324e-05,
      "loss": 0.379,
      "step": 8653
    },
    {
      "epoch": 0.4846007391645201,
      "grad_norm": 1.325932502746582,
      "learning_rate": 4.3245000000000004e-05,
      "loss": 0.4285,
      "step": 8654
    },
    {
      "epoch": 0.48465673647664914,
      "grad_norm": 1.3755015134811401,
      "learning_rate": 4.325e-05,
      "loss": 0.3669,
      "step": 8655
    },
    {
      "epoch": 0.48471273378877816,
      "grad_norm": 1.0719605684280396,
      "learning_rate": 4.3255e-05,
      "loss": 0.3082,
      "step": 8656
    },
    {
      "epoch": 0.4847687311009072,
      "grad_norm": 1.1531341075897217,
      "learning_rate": 4.326e-05,
      "loss": 0.4137,
      "step": 8657
    },
    {
      "epoch": 0.4848247284130362,
      "grad_norm": 1.3665847778320312,
      "learning_rate": 4.3265e-05,
      "loss": 0.4001,
      "step": 8658
    },
    {
      "epoch": 0.4848807257251652,
      "grad_norm": 1.3036551475524902,
      "learning_rate": 4.327e-05,
      "loss": 0.4852,
      "step": 8659
    },
    {
      "epoch": 0.48493672303729424,
      "grad_norm": 1.1637921333312988,
      "learning_rate": 4.3275e-05,
      "loss": 0.426,
      "step": 8660
    },
    {
      "epoch": 0.4849927203494232,
      "grad_norm": 1.3472542762756348,
      "learning_rate": 4.3280000000000006e-05,
      "loss": 0.4741,
      "step": 8661
    },
    {
      "epoch": 0.4850487176615522,
      "grad_norm": 1.2393300533294678,
      "learning_rate": 4.3285e-05,
      "loss": 0.317,
      "step": 8662
    },
    {
      "epoch": 0.48510471497368124,
      "grad_norm": 1.2900378704071045,
      "learning_rate": 4.329e-05,
      "loss": 0.4287,
      "step": 8663
    },
    {
      "epoch": 0.48516071228581026,
      "grad_norm": 1.5040082931518555,
      "learning_rate": 4.3295000000000005e-05,
      "loss": 0.4619,
      "step": 8664
    },
    {
      "epoch": 0.4852167095979393,
      "grad_norm": 1.1562973260879517,
      "learning_rate": 4.33e-05,
      "loss": 0.4546,
      "step": 8665
    },
    {
      "epoch": 0.4852727069100683,
      "grad_norm": 1.3024035692214966,
      "learning_rate": 4.3305e-05,
      "loss": 0.3741,
      "step": 8666
    },
    {
      "epoch": 0.4853287042221973,
      "grad_norm": 2.070420742034912,
      "learning_rate": 4.3310000000000004e-05,
      "loss": 0.4455,
      "step": 8667
    },
    {
      "epoch": 0.48538470153432633,
      "grad_norm": 1.1950358152389526,
      "learning_rate": 4.3315e-05,
      "loss": 0.4381,
      "step": 8668
    },
    {
      "epoch": 0.48544069884645535,
      "grad_norm": 1.0591496229171753,
      "learning_rate": 4.332e-05,
      "loss": 0.3745,
      "step": 8669
    },
    {
      "epoch": 0.4854966961585844,
      "grad_norm": 1.0986610651016235,
      "learning_rate": 4.3325e-05,
      "loss": 0.4661,
      "step": 8670
    },
    {
      "epoch": 0.4855526934707134,
      "grad_norm": 1.3725085258483887,
      "learning_rate": 4.333000000000001e-05,
      "loss": 0.5387,
      "step": 8671
    },
    {
      "epoch": 0.4856086907828424,
      "grad_norm": 1.3331035375595093,
      "learning_rate": 4.3335000000000004e-05,
      "loss": 0.445,
      "step": 8672
    },
    {
      "epoch": 0.48566468809497143,
      "grad_norm": 1.120967984199524,
      "learning_rate": 4.334e-05,
      "loss": 0.3554,
      "step": 8673
    },
    {
      "epoch": 0.48572068540710045,
      "grad_norm": 1.117554783821106,
      "learning_rate": 4.3345e-05,
      "loss": 0.3826,
      "step": 8674
    },
    {
      "epoch": 0.48577668271922947,
      "grad_norm": 1.3135879039764404,
      "learning_rate": 4.335e-05,
      "loss": 0.4429,
      "step": 8675
    },
    {
      "epoch": 0.4858326800313585,
      "grad_norm": 1.6566495895385742,
      "learning_rate": 4.3355e-05,
      "loss": 0.3585,
      "step": 8676
    },
    {
      "epoch": 0.4858886773434875,
      "grad_norm": 1.137291431427002,
      "learning_rate": 4.336e-05,
      "loss": 0.3587,
      "step": 8677
    },
    {
      "epoch": 0.4859446746556165,
      "grad_norm": 7.002615928649902,
      "learning_rate": 4.3365e-05,
      "loss": 0.3849,
      "step": 8678
    },
    {
      "epoch": 0.48600067196774555,
      "grad_norm": 1.2987697124481201,
      "learning_rate": 4.337e-05,
      "loss": 0.5691,
      "step": 8679
    },
    {
      "epoch": 0.48605666927987456,
      "grad_norm": 1.201171636581421,
      "learning_rate": 4.3375000000000004e-05,
      "loss": 0.3965,
      "step": 8680
    },
    {
      "epoch": 0.4861126665920036,
      "grad_norm": 1.2575432062149048,
      "learning_rate": 4.338e-05,
      "loss": 0.4293,
      "step": 8681
    },
    {
      "epoch": 0.4861686639041326,
      "grad_norm": 1.2194340229034424,
      "learning_rate": 4.3385000000000006e-05,
      "loss": 0.4773,
      "step": 8682
    },
    {
      "epoch": 0.4862246612162616,
      "grad_norm": 1.40916907787323,
      "learning_rate": 4.339e-05,
      "loss": 0.6009,
      "step": 8683
    },
    {
      "epoch": 0.48628065852839064,
      "grad_norm": 1.3053512573242188,
      "learning_rate": 4.3395e-05,
      "loss": 0.4454,
      "step": 8684
    },
    {
      "epoch": 0.48633665584051966,
      "grad_norm": 1.3352710008621216,
      "learning_rate": 4.3400000000000005e-05,
      "loss": 0.422,
      "step": 8685
    },
    {
      "epoch": 0.4863926531526487,
      "grad_norm": 1.613464117050171,
      "learning_rate": 4.3405e-05,
      "loss": 0.461,
      "step": 8686
    },
    {
      "epoch": 0.4864486504647777,
      "grad_norm": 1.0982309579849243,
      "learning_rate": 4.341e-05,
      "loss": 0.3205,
      "step": 8687
    },
    {
      "epoch": 0.4865046477769067,
      "grad_norm": 1.5111366510391235,
      "learning_rate": 4.3415e-05,
      "loss": 0.4938,
      "step": 8688
    },
    {
      "epoch": 0.48656064508903574,
      "grad_norm": 1.0738887786865234,
      "learning_rate": 4.342e-05,
      "loss": 0.4375,
      "step": 8689
    },
    {
      "epoch": 0.48661664240116476,
      "grad_norm": 1.1899917125701904,
      "learning_rate": 4.3425000000000005e-05,
      "loss": 0.3337,
      "step": 8690
    },
    {
      "epoch": 0.4866726397132938,
      "grad_norm": 1.386510968208313,
      "learning_rate": 4.343e-05,
      "loss": 0.4249,
      "step": 8691
    },
    {
      "epoch": 0.4867286370254228,
      "grad_norm": 1.2535392045974731,
      "learning_rate": 4.343500000000001e-05,
      "loss": 0.4321,
      "step": 8692
    },
    {
      "epoch": 0.4867846343375518,
      "grad_norm": 1.122766137123108,
      "learning_rate": 4.3440000000000004e-05,
      "loss": 0.4399,
      "step": 8693
    },
    {
      "epoch": 0.48684063164968083,
      "grad_norm": 1.240119218826294,
      "learning_rate": 4.3445e-05,
      "loss": 0.533,
      "step": 8694
    },
    {
      "epoch": 0.48689662896180985,
      "grad_norm": 1.1166800260543823,
      "learning_rate": 4.345e-05,
      "loss": 0.4065,
      "step": 8695
    },
    {
      "epoch": 0.48695262627393887,
      "grad_norm": 1.2940329313278198,
      "learning_rate": 4.3455e-05,
      "loss": 0.4668,
      "step": 8696
    },
    {
      "epoch": 0.4870086235860679,
      "grad_norm": 1.1920511722564697,
      "learning_rate": 4.346e-05,
      "loss": 0.4026,
      "step": 8697
    },
    {
      "epoch": 0.4870646208981969,
      "grad_norm": 1.1582311391830444,
      "learning_rate": 4.3465e-05,
      "loss": 0.5435,
      "step": 8698
    },
    {
      "epoch": 0.48712061821032593,
      "grad_norm": 1.4122322797775269,
      "learning_rate": 4.347e-05,
      "loss": 0.5706,
      "step": 8699
    },
    {
      "epoch": 0.48717661552245495,
      "grad_norm": 1.1645272970199585,
      "learning_rate": 4.3475000000000006e-05,
      "loss": 0.4621,
      "step": 8700
    },
    {
      "epoch": 0.48723261283458397,
      "grad_norm": 1.3535151481628418,
      "learning_rate": 4.3480000000000004e-05,
      "loss": 0.39,
      "step": 8701
    },
    {
      "epoch": 0.48728861014671293,
      "grad_norm": 1.1694250106811523,
      "learning_rate": 4.3485e-05,
      "loss": 0.4385,
      "step": 8702
    },
    {
      "epoch": 0.48734460745884195,
      "grad_norm": 1.1841671466827393,
      "learning_rate": 4.3490000000000005e-05,
      "loss": 0.5778,
      "step": 8703
    },
    {
      "epoch": 0.48740060477097097,
      "grad_norm": 1.0413399934768677,
      "learning_rate": 4.3495e-05,
      "loss": 0.4254,
      "step": 8704
    },
    {
      "epoch": 0.4874566020831,
      "grad_norm": 1.1953142881393433,
      "learning_rate": 4.35e-05,
      "loss": 0.4206,
      "step": 8705
    },
    {
      "epoch": 0.487512599395229,
      "grad_norm": 1.3137667179107666,
      "learning_rate": 4.3505000000000004e-05,
      "loss": 0.4972,
      "step": 8706
    },
    {
      "epoch": 0.487568596707358,
      "grad_norm": 1.1413288116455078,
      "learning_rate": 4.351e-05,
      "loss": 0.3851,
      "step": 8707
    },
    {
      "epoch": 0.48762459401948705,
      "grad_norm": 1.2208783626556396,
      "learning_rate": 4.3515e-05,
      "loss": 0.4572,
      "step": 8708
    },
    {
      "epoch": 0.48768059133161606,
      "grad_norm": 1.3009867668151855,
      "learning_rate": 4.352e-05,
      "loss": 0.441,
      "step": 8709
    },
    {
      "epoch": 0.4877365886437451,
      "grad_norm": 1.4197602272033691,
      "learning_rate": 4.352500000000001e-05,
      "loss": 0.4225,
      "step": 8710
    },
    {
      "epoch": 0.4877925859558741,
      "grad_norm": 1.5292143821716309,
      "learning_rate": 4.3530000000000005e-05,
      "loss": 0.4581,
      "step": 8711
    },
    {
      "epoch": 0.4878485832680031,
      "grad_norm": 1.1709518432617188,
      "learning_rate": 4.3535e-05,
      "loss": 0.308,
      "step": 8712
    },
    {
      "epoch": 0.48790458058013214,
      "grad_norm": 1.5590686798095703,
      "learning_rate": 4.354e-05,
      "loss": 0.351,
      "step": 8713
    },
    {
      "epoch": 0.48796057789226116,
      "grad_norm": 1.4348865747451782,
      "learning_rate": 4.3545000000000004e-05,
      "loss": 0.5042,
      "step": 8714
    },
    {
      "epoch": 0.4880165752043902,
      "grad_norm": 1.520550012588501,
      "learning_rate": 4.355e-05,
      "loss": 0.4064,
      "step": 8715
    },
    {
      "epoch": 0.4880725725165192,
      "grad_norm": 1.2306742668151855,
      "learning_rate": 4.3555e-05,
      "loss": 0.4183,
      "step": 8716
    },
    {
      "epoch": 0.4881285698286482,
      "grad_norm": 1.1449452638626099,
      "learning_rate": 4.356e-05,
      "loss": 0.4367,
      "step": 8717
    },
    {
      "epoch": 0.48818456714077724,
      "grad_norm": 1.4547386169433594,
      "learning_rate": 4.3565e-05,
      "loss": 0.3432,
      "step": 8718
    },
    {
      "epoch": 0.48824056445290626,
      "grad_norm": 1.3006969690322876,
      "learning_rate": 4.357e-05,
      "loss": 0.442,
      "step": 8719
    },
    {
      "epoch": 0.4882965617650353,
      "grad_norm": 1.424604058265686,
      "learning_rate": 4.3575e-05,
      "loss": 0.6239,
      "step": 8720
    },
    {
      "epoch": 0.4883525590771643,
      "grad_norm": 1.3461222648620605,
      "learning_rate": 4.3580000000000006e-05,
      "loss": 0.3196,
      "step": 8721
    },
    {
      "epoch": 0.4884085563892933,
      "grad_norm": 1.2990061044692993,
      "learning_rate": 4.3585000000000004e-05,
      "loss": 0.4009,
      "step": 8722
    },
    {
      "epoch": 0.48846455370142233,
      "grad_norm": 1.274626612663269,
      "learning_rate": 4.359e-05,
      "loss": 0.4871,
      "step": 8723
    },
    {
      "epoch": 0.48852055101355135,
      "grad_norm": 1.3779020309448242,
      "learning_rate": 4.3595000000000005e-05,
      "loss": 0.4717,
      "step": 8724
    },
    {
      "epoch": 0.48857654832568037,
      "grad_norm": 1.209932804107666,
      "learning_rate": 4.36e-05,
      "loss": 0.4969,
      "step": 8725
    },
    {
      "epoch": 0.4886325456378094,
      "grad_norm": 1.4006141424179077,
      "learning_rate": 4.3605e-05,
      "loss": 0.5306,
      "step": 8726
    },
    {
      "epoch": 0.4886885429499384,
      "grad_norm": 1.6883563995361328,
      "learning_rate": 4.361e-05,
      "loss": 0.4729,
      "step": 8727
    },
    {
      "epoch": 0.48874454026206743,
      "grad_norm": 1.2552179098129272,
      "learning_rate": 4.3615e-05,
      "loss": 0.5425,
      "step": 8728
    },
    {
      "epoch": 0.48880053757419645,
      "grad_norm": 1.6334960460662842,
      "learning_rate": 4.362e-05,
      "loss": 0.3721,
      "step": 8729
    },
    {
      "epoch": 0.48885653488632547,
      "grad_norm": 1.4374152421951294,
      "learning_rate": 4.3625e-05,
      "loss": 0.3604,
      "step": 8730
    },
    {
      "epoch": 0.4889125321984545,
      "grad_norm": 1.3681845664978027,
      "learning_rate": 4.363000000000001e-05,
      "loss": 0.4152,
      "step": 8731
    },
    {
      "epoch": 0.4889685295105835,
      "grad_norm": 1.2419909238815308,
      "learning_rate": 4.3635000000000005e-05,
      "loss": 0.4418,
      "step": 8732
    },
    {
      "epoch": 0.4890245268227125,
      "grad_norm": 1.1012789011001587,
      "learning_rate": 4.364e-05,
      "loss": 0.3571,
      "step": 8733
    },
    {
      "epoch": 0.48908052413484154,
      "grad_norm": 1.400465965270996,
      "learning_rate": 4.3645e-05,
      "loss": 0.4592,
      "step": 8734
    },
    {
      "epoch": 0.48913652144697056,
      "grad_norm": 1.4451253414154053,
      "learning_rate": 4.3650000000000004e-05,
      "loss": 0.5828,
      "step": 8735
    },
    {
      "epoch": 0.4891925187590996,
      "grad_norm": 1.1183686256408691,
      "learning_rate": 4.3655e-05,
      "loss": 0.3976,
      "step": 8736
    },
    {
      "epoch": 0.4892485160712286,
      "grad_norm": 1.1781045198440552,
      "learning_rate": 4.366e-05,
      "loss": 0.3226,
      "step": 8737
    },
    {
      "epoch": 0.4893045133833576,
      "grad_norm": 1.074252963066101,
      "learning_rate": 4.3665e-05,
      "loss": 0.428,
      "step": 8738
    },
    {
      "epoch": 0.48936051069548664,
      "grad_norm": 1.3820005655288696,
      "learning_rate": 4.367e-05,
      "loss": 0.4251,
      "step": 8739
    },
    {
      "epoch": 0.48941650800761566,
      "grad_norm": 1.3468071222305298,
      "learning_rate": 4.3675000000000005e-05,
      "loss": 0.4603,
      "step": 8740
    },
    {
      "epoch": 0.4894725053197447,
      "grad_norm": 1.370620846748352,
      "learning_rate": 4.368e-05,
      "loss": 0.4452,
      "step": 8741
    },
    {
      "epoch": 0.4895285026318737,
      "grad_norm": 1.138685703277588,
      "learning_rate": 4.3685000000000006e-05,
      "loss": 0.419,
      "step": 8742
    },
    {
      "epoch": 0.48958449994400266,
      "grad_norm": 1.067796230316162,
      "learning_rate": 4.3690000000000004e-05,
      "loss": 0.4796,
      "step": 8743
    },
    {
      "epoch": 0.4896404972561317,
      "grad_norm": 1.22769033908844,
      "learning_rate": 4.3695e-05,
      "loss": 0.3606,
      "step": 8744
    },
    {
      "epoch": 0.4896964945682607,
      "grad_norm": 1.3409802913665771,
      "learning_rate": 4.3700000000000005e-05,
      "loss": 0.3688,
      "step": 8745
    },
    {
      "epoch": 0.4897524918803897,
      "grad_norm": 1.0355149507522583,
      "learning_rate": 4.3705e-05,
      "loss": 0.4116,
      "step": 8746
    },
    {
      "epoch": 0.48980848919251874,
      "grad_norm": 1.2071590423583984,
      "learning_rate": 4.371e-05,
      "loss": 0.3694,
      "step": 8747
    },
    {
      "epoch": 0.48986448650464776,
      "grad_norm": 1.3578791618347168,
      "learning_rate": 4.3715e-05,
      "loss": 0.5592,
      "step": 8748
    },
    {
      "epoch": 0.4899204838167768,
      "grad_norm": 1.3043264150619507,
      "learning_rate": 4.372e-05,
      "loss": 0.4845,
      "step": 8749
    },
    {
      "epoch": 0.4899764811289058,
      "grad_norm": 1.3237848281860352,
      "learning_rate": 4.3725000000000006e-05,
      "loss": 0.4548,
      "step": 8750
    },
    {
      "epoch": 0.4900324784410348,
      "grad_norm": 1.137434959411621,
      "learning_rate": 4.373e-05,
      "loss": 0.4235,
      "step": 8751
    },
    {
      "epoch": 0.49008847575316383,
      "grad_norm": 1.2952182292938232,
      "learning_rate": 4.3735e-05,
      "loss": 0.4302,
      "step": 8752
    },
    {
      "epoch": 0.49014447306529285,
      "grad_norm": 1.2217761278152466,
      "learning_rate": 4.3740000000000005e-05,
      "loss": 0.4094,
      "step": 8753
    },
    {
      "epoch": 0.49020047037742187,
      "grad_norm": 1.3580907583236694,
      "learning_rate": 4.3745e-05,
      "loss": 0.4641,
      "step": 8754
    },
    {
      "epoch": 0.4902564676895509,
      "grad_norm": 1.3834844827651978,
      "learning_rate": 4.375e-05,
      "loss": 0.5332,
      "step": 8755
    },
    {
      "epoch": 0.4903124650016799,
      "grad_norm": 1.376662254333496,
      "learning_rate": 4.3755000000000004e-05,
      "loss": 0.4227,
      "step": 8756
    },
    {
      "epoch": 0.49036846231380893,
      "grad_norm": 1.4165754318237305,
      "learning_rate": 4.376e-05,
      "loss": 0.5348,
      "step": 8757
    },
    {
      "epoch": 0.49042445962593795,
      "grad_norm": 0.9695160388946533,
      "learning_rate": 4.3765e-05,
      "loss": 0.294,
      "step": 8758
    },
    {
      "epoch": 0.49048045693806697,
      "grad_norm": 1.2843620777130127,
      "learning_rate": 4.377e-05,
      "loss": 0.5706,
      "step": 8759
    },
    {
      "epoch": 0.490536454250196,
      "grad_norm": 1.5324783325195312,
      "learning_rate": 4.3775e-05,
      "loss": 0.4855,
      "step": 8760
    },
    {
      "epoch": 0.490592451562325,
      "grad_norm": 1.4297372102737427,
      "learning_rate": 4.3780000000000004e-05,
      "loss": 0.458,
      "step": 8761
    },
    {
      "epoch": 0.490648448874454,
      "grad_norm": 1.6255004405975342,
      "learning_rate": 4.3785e-05,
      "loss": 0.4558,
      "step": 8762
    },
    {
      "epoch": 0.49070444618658304,
      "grad_norm": 1.2813847064971924,
      "learning_rate": 4.3790000000000006e-05,
      "loss": 0.3361,
      "step": 8763
    },
    {
      "epoch": 0.49076044349871206,
      "grad_norm": 1.2937064170837402,
      "learning_rate": 4.3795e-05,
      "loss": 0.3665,
      "step": 8764
    },
    {
      "epoch": 0.4908164408108411,
      "grad_norm": 1.2819708585739136,
      "learning_rate": 4.38e-05,
      "loss": 0.3996,
      "step": 8765
    },
    {
      "epoch": 0.4908724381229701,
      "grad_norm": 1.1545991897583008,
      "learning_rate": 4.3805000000000005e-05,
      "loss": 0.3478,
      "step": 8766
    },
    {
      "epoch": 0.4909284354350991,
      "grad_norm": 1.3991823196411133,
      "learning_rate": 4.381e-05,
      "loss": 0.4191,
      "step": 8767
    },
    {
      "epoch": 0.49098443274722814,
      "grad_norm": 1.263278603553772,
      "learning_rate": 4.3815e-05,
      "loss": 0.384,
      "step": 8768
    },
    {
      "epoch": 0.49104043005935716,
      "grad_norm": 1.6084866523742676,
      "learning_rate": 4.382e-05,
      "loss": 0.353,
      "step": 8769
    },
    {
      "epoch": 0.4910964273714862,
      "grad_norm": 1.383324146270752,
      "learning_rate": 4.3825e-05,
      "loss": 0.4827,
      "step": 8770
    },
    {
      "epoch": 0.4911524246836152,
      "grad_norm": 1.567399501800537,
      "learning_rate": 4.3830000000000006e-05,
      "loss": 0.4992,
      "step": 8771
    },
    {
      "epoch": 0.4912084219957442,
      "grad_norm": 1.2193971872329712,
      "learning_rate": 4.3835e-05,
      "loss": 0.382,
      "step": 8772
    },
    {
      "epoch": 0.49126441930787323,
      "grad_norm": 1.2334585189819336,
      "learning_rate": 4.384e-05,
      "loss": 0.3591,
      "step": 8773
    },
    {
      "epoch": 0.49132041662000225,
      "grad_norm": 1.1027566194534302,
      "learning_rate": 4.3845000000000005e-05,
      "loss": 0.3656,
      "step": 8774
    },
    {
      "epoch": 0.4913764139321313,
      "grad_norm": 1.3041020631790161,
      "learning_rate": 4.385e-05,
      "loss": 0.4085,
      "step": 8775
    },
    {
      "epoch": 0.4914324112442603,
      "grad_norm": 1.1698496341705322,
      "learning_rate": 4.3855e-05,
      "loss": 0.3652,
      "step": 8776
    },
    {
      "epoch": 0.4914884085563893,
      "grad_norm": 1.3277337551116943,
      "learning_rate": 4.3860000000000004e-05,
      "loss": 0.4438,
      "step": 8777
    },
    {
      "epoch": 0.49154440586851833,
      "grad_norm": 1.3195639848709106,
      "learning_rate": 4.3865e-05,
      "loss": 0.4391,
      "step": 8778
    },
    {
      "epoch": 0.49160040318064735,
      "grad_norm": 1.4638186693191528,
      "learning_rate": 4.387e-05,
      "loss": 0.449,
      "step": 8779
    },
    {
      "epoch": 0.49165640049277637,
      "grad_norm": 1.458646535873413,
      "learning_rate": 4.3875e-05,
      "loss": 0.3749,
      "step": 8780
    },
    {
      "epoch": 0.4917123978049054,
      "grad_norm": 1.220596194267273,
      "learning_rate": 4.388000000000001e-05,
      "loss": 0.4252,
      "step": 8781
    },
    {
      "epoch": 0.4917683951170344,
      "grad_norm": 1.456165075302124,
      "learning_rate": 4.3885000000000004e-05,
      "loss": 0.4328,
      "step": 8782
    },
    {
      "epoch": 0.4918243924291634,
      "grad_norm": 2.3613314628601074,
      "learning_rate": 4.389e-05,
      "loss": 0.6209,
      "step": 8783
    },
    {
      "epoch": 0.49188038974129245,
      "grad_norm": 1.0482722520828247,
      "learning_rate": 4.3895000000000006e-05,
      "loss": 0.4775,
      "step": 8784
    },
    {
      "epoch": 0.4919363870534214,
      "grad_norm": 1.2302495241165161,
      "learning_rate": 4.39e-05,
      "loss": 0.4245,
      "step": 8785
    },
    {
      "epoch": 0.49199238436555043,
      "grad_norm": 1.2513916492462158,
      "learning_rate": 4.3905e-05,
      "loss": 0.4248,
      "step": 8786
    },
    {
      "epoch": 0.49204838167767945,
      "grad_norm": 1.495124340057373,
      "learning_rate": 4.391e-05,
      "loss": 0.481,
      "step": 8787
    },
    {
      "epoch": 0.49210437898980847,
      "grad_norm": 1.1340218782424927,
      "learning_rate": 4.3915e-05,
      "loss": 0.4743,
      "step": 8788
    },
    {
      "epoch": 0.4921603763019375,
      "grad_norm": 1.4609558582305908,
      "learning_rate": 4.392e-05,
      "loss": 0.4757,
      "step": 8789
    },
    {
      "epoch": 0.4922163736140665,
      "grad_norm": 1.3570151329040527,
      "learning_rate": 4.3925e-05,
      "loss": 0.485,
      "step": 8790
    },
    {
      "epoch": 0.4922723709261955,
      "grad_norm": 1.118835687637329,
      "learning_rate": 4.393e-05,
      "loss": 0.376,
      "step": 8791
    },
    {
      "epoch": 0.49232836823832454,
      "grad_norm": 1.5721051692962646,
      "learning_rate": 4.3935000000000005e-05,
      "loss": 0.5853,
      "step": 8792
    },
    {
      "epoch": 0.49238436555045356,
      "grad_norm": 1.169965147972107,
      "learning_rate": 4.394e-05,
      "loss": 0.4243,
      "step": 8793
    },
    {
      "epoch": 0.4924403628625826,
      "grad_norm": 1.2264182567596436,
      "learning_rate": 4.3945e-05,
      "loss": 0.4102,
      "step": 8794
    },
    {
      "epoch": 0.4924963601747116,
      "grad_norm": 0.9702182412147522,
      "learning_rate": 4.3950000000000004e-05,
      "loss": 0.3123,
      "step": 8795
    },
    {
      "epoch": 0.4925523574868406,
      "grad_norm": 1.12941575050354,
      "learning_rate": 4.3955e-05,
      "loss": 0.3943,
      "step": 8796
    },
    {
      "epoch": 0.49260835479896964,
      "grad_norm": 1.5503352880477905,
      "learning_rate": 4.396e-05,
      "loss": 0.4887,
      "step": 8797
    },
    {
      "epoch": 0.49266435211109866,
      "grad_norm": 1.2890199422836304,
      "learning_rate": 4.3965000000000003e-05,
      "loss": 0.42,
      "step": 8798
    },
    {
      "epoch": 0.4927203494232277,
      "grad_norm": 1.1840254068374634,
      "learning_rate": 4.397e-05,
      "loss": 0.3709,
      "step": 8799
    },
    {
      "epoch": 0.4927763467353567,
      "grad_norm": 1.4131531715393066,
      "learning_rate": 4.3975e-05,
      "loss": 0.6136,
      "step": 8800
    },
    {
      "epoch": 0.4928323440474857,
      "grad_norm": 1.085906744003296,
      "learning_rate": 4.398e-05,
      "loss": 0.3824,
      "step": 8801
    },
    {
      "epoch": 0.49288834135961473,
      "grad_norm": 1.2747128009796143,
      "learning_rate": 4.398500000000001e-05,
      "loss": 0.3949,
      "step": 8802
    },
    {
      "epoch": 0.49294433867174375,
      "grad_norm": 1.0494940280914307,
      "learning_rate": 4.3990000000000004e-05,
      "loss": 0.3932,
      "step": 8803
    },
    {
      "epoch": 0.4930003359838728,
      "grad_norm": 1.1495386362075806,
      "learning_rate": 4.3995e-05,
      "loss": 0.3146,
      "step": 8804
    },
    {
      "epoch": 0.4930563332960018,
      "grad_norm": 1.4327843189239502,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.4102,
      "step": 8805
    },
    {
      "epoch": 0.4931123306081308,
      "grad_norm": 1.2018320560455322,
      "learning_rate": 4.4005e-05,
      "loss": 0.3846,
      "step": 8806
    },
    {
      "epoch": 0.49316832792025983,
      "grad_norm": 1.2860138416290283,
      "learning_rate": 4.401e-05,
      "loss": 0.4058,
      "step": 8807
    },
    {
      "epoch": 0.49322432523238885,
      "grad_norm": 1.254707932472229,
      "learning_rate": 4.4015e-05,
      "loss": 0.3821,
      "step": 8808
    },
    {
      "epoch": 0.49328032254451787,
      "grad_norm": 1.2617658376693726,
      "learning_rate": 4.402e-05,
      "loss": 0.4053,
      "step": 8809
    },
    {
      "epoch": 0.4933363198566469,
      "grad_norm": 1.1106071472167969,
      "learning_rate": 4.4025e-05,
      "loss": 0.319,
      "step": 8810
    },
    {
      "epoch": 0.4933923171687759,
      "grad_norm": 1.0793672800064087,
      "learning_rate": 4.4030000000000004e-05,
      "loss": 0.3452,
      "step": 8811
    },
    {
      "epoch": 0.4934483144809049,
      "grad_norm": 1.3354262113571167,
      "learning_rate": 4.4035e-05,
      "loss": 0.4187,
      "step": 8812
    },
    {
      "epoch": 0.49350431179303395,
      "grad_norm": 1.1656129360198975,
      "learning_rate": 4.4040000000000005e-05,
      "loss": 0.4096,
      "step": 8813
    },
    {
      "epoch": 0.49356030910516296,
      "grad_norm": 1.3050516843795776,
      "learning_rate": 4.4045e-05,
      "loss": 0.402,
      "step": 8814
    },
    {
      "epoch": 0.493616306417292,
      "grad_norm": 1.3620717525482178,
      "learning_rate": 4.405e-05,
      "loss": 0.4883,
      "step": 8815
    },
    {
      "epoch": 0.493672303729421,
      "grad_norm": 1.1378008127212524,
      "learning_rate": 4.4055000000000004e-05,
      "loss": 0.3795,
      "step": 8816
    },
    {
      "epoch": 0.49372830104155,
      "grad_norm": 1.1346876621246338,
      "learning_rate": 4.406e-05,
      "loss": 0.3449,
      "step": 8817
    },
    {
      "epoch": 0.49378429835367904,
      "grad_norm": 1.2142455577850342,
      "learning_rate": 4.4065e-05,
      "loss": 0.4761,
      "step": 8818
    },
    {
      "epoch": 0.49384029566580806,
      "grad_norm": 1.2473500967025757,
      "learning_rate": 4.407e-05,
      "loss": 0.5042,
      "step": 8819
    },
    {
      "epoch": 0.4938962929779371,
      "grad_norm": 1.159977912902832,
      "learning_rate": 4.4075e-05,
      "loss": 0.3856,
      "step": 8820
    },
    {
      "epoch": 0.4939522902900661,
      "grad_norm": 1.3137444257736206,
      "learning_rate": 4.4080000000000005e-05,
      "loss": 0.4382,
      "step": 8821
    },
    {
      "epoch": 0.4940082876021951,
      "grad_norm": 1.673619270324707,
      "learning_rate": 4.4085e-05,
      "loss": 0.5157,
      "step": 8822
    },
    {
      "epoch": 0.49406428491432414,
      "grad_norm": 1.3758738040924072,
      "learning_rate": 4.4090000000000006e-05,
      "loss": 0.5904,
      "step": 8823
    },
    {
      "epoch": 0.49412028222645316,
      "grad_norm": 1.015759825706482,
      "learning_rate": 4.4095000000000004e-05,
      "loss": 0.331,
      "step": 8824
    },
    {
      "epoch": 0.4941762795385822,
      "grad_norm": 1.161116600036621,
      "learning_rate": 4.41e-05,
      "loss": 0.2972,
      "step": 8825
    },
    {
      "epoch": 0.49423227685071114,
      "grad_norm": 1.3757725954055786,
      "learning_rate": 4.4105e-05,
      "loss": 0.3803,
      "step": 8826
    },
    {
      "epoch": 0.49428827416284016,
      "grad_norm": 1.5430757999420166,
      "learning_rate": 4.411e-05,
      "loss": 0.6873,
      "step": 8827
    },
    {
      "epoch": 0.4943442714749692,
      "grad_norm": 1.497576117515564,
      "learning_rate": 4.4115e-05,
      "loss": 0.5016,
      "step": 8828
    },
    {
      "epoch": 0.4944002687870982,
      "grad_norm": 1.2487151622772217,
      "learning_rate": 4.412e-05,
      "loss": 0.4236,
      "step": 8829
    },
    {
      "epoch": 0.4944562660992272,
      "grad_norm": 1.313583254814148,
      "learning_rate": 4.4125e-05,
      "loss": 0.6077,
      "step": 8830
    },
    {
      "epoch": 0.49451226341135623,
      "grad_norm": 1.2831478118896484,
      "learning_rate": 4.4130000000000006e-05,
      "loss": 0.4209,
      "step": 8831
    },
    {
      "epoch": 0.49456826072348525,
      "grad_norm": 1.4325164556503296,
      "learning_rate": 4.4135000000000003e-05,
      "loss": 0.44,
      "step": 8832
    },
    {
      "epoch": 0.4946242580356143,
      "grad_norm": 1.1198872327804565,
      "learning_rate": 4.414e-05,
      "loss": 0.451,
      "step": 8833
    },
    {
      "epoch": 0.4946802553477433,
      "grad_norm": 1.1292073726654053,
      "learning_rate": 4.4145000000000005e-05,
      "loss": 0.4119,
      "step": 8834
    },
    {
      "epoch": 0.4947362526598723,
      "grad_norm": 1.7890433073043823,
      "learning_rate": 4.415e-05,
      "loss": 0.3996,
      "step": 8835
    },
    {
      "epoch": 0.49479224997200133,
      "grad_norm": 1.0275719165802002,
      "learning_rate": 4.4155e-05,
      "loss": 0.3757,
      "step": 8836
    },
    {
      "epoch": 0.49484824728413035,
      "grad_norm": 1.1006819009780884,
      "learning_rate": 4.4160000000000004e-05,
      "loss": 0.4615,
      "step": 8837
    },
    {
      "epoch": 0.49490424459625937,
      "grad_norm": 1.203513503074646,
      "learning_rate": 4.4165e-05,
      "loss": 0.3168,
      "step": 8838
    },
    {
      "epoch": 0.4949602419083884,
      "grad_norm": 1.2412500381469727,
      "learning_rate": 4.417e-05,
      "loss": 0.4583,
      "step": 8839
    },
    {
      "epoch": 0.4950162392205174,
      "grad_norm": 1.6406822204589844,
      "learning_rate": 4.4174999999999996e-05,
      "loss": 0.5096,
      "step": 8840
    },
    {
      "epoch": 0.4950722365326464,
      "grad_norm": 1.2065621614456177,
      "learning_rate": 4.418000000000001e-05,
      "loss": 0.4388,
      "step": 8841
    },
    {
      "epoch": 0.49512823384477544,
      "grad_norm": 1.75081205368042,
      "learning_rate": 4.4185000000000005e-05,
      "loss": 0.528,
      "step": 8842
    },
    {
      "epoch": 0.49518423115690446,
      "grad_norm": 1.1402934789657593,
      "learning_rate": 4.419e-05,
      "loss": 0.389,
      "step": 8843
    },
    {
      "epoch": 0.4952402284690335,
      "grad_norm": 1.4472622871398926,
      "learning_rate": 4.4195000000000006e-05,
      "loss": 0.526,
      "step": 8844
    },
    {
      "epoch": 0.4952962257811625,
      "grad_norm": 1.3543546199798584,
      "learning_rate": 4.4200000000000004e-05,
      "loss": 0.4009,
      "step": 8845
    },
    {
      "epoch": 0.4953522230932915,
      "grad_norm": 1.2927943468093872,
      "learning_rate": 4.4205e-05,
      "loss": 0.4673,
      "step": 8846
    },
    {
      "epoch": 0.49540822040542054,
      "grad_norm": 1.1858797073364258,
      "learning_rate": 4.421e-05,
      "loss": 0.4821,
      "step": 8847
    },
    {
      "epoch": 0.49546421771754956,
      "grad_norm": 1.1288942098617554,
      "learning_rate": 4.4215e-05,
      "loss": 0.3303,
      "step": 8848
    },
    {
      "epoch": 0.4955202150296786,
      "grad_norm": 1.1077618598937988,
      "learning_rate": 4.422e-05,
      "loss": 0.3461,
      "step": 8849
    },
    {
      "epoch": 0.4955762123418076,
      "grad_norm": 0.9944883584976196,
      "learning_rate": 4.4225e-05,
      "loss": 0.3039,
      "step": 8850
    },
    {
      "epoch": 0.4956322096539366,
      "grad_norm": 1.2238086462020874,
      "learning_rate": 4.423e-05,
      "loss": 0.4426,
      "step": 8851
    },
    {
      "epoch": 0.49568820696606564,
      "grad_norm": 1.333704948425293,
      "learning_rate": 4.4235000000000006e-05,
      "loss": 0.5154,
      "step": 8852
    },
    {
      "epoch": 0.49574420427819466,
      "grad_norm": 1.2379481792449951,
      "learning_rate": 4.424e-05,
      "loss": 0.4862,
      "step": 8853
    },
    {
      "epoch": 0.4958002015903237,
      "grad_norm": 1.1883821487426758,
      "learning_rate": 4.4245e-05,
      "loss": 0.3883,
      "step": 8854
    },
    {
      "epoch": 0.4958561989024527,
      "grad_norm": 1.4488531351089478,
      "learning_rate": 4.4250000000000005e-05,
      "loss": 0.4037,
      "step": 8855
    },
    {
      "epoch": 0.4959121962145817,
      "grad_norm": 1.6411842107772827,
      "learning_rate": 4.4255e-05,
      "loss": 0.484,
      "step": 8856
    },
    {
      "epoch": 0.49596819352671073,
      "grad_norm": 1.5303682088851929,
      "learning_rate": 4.426e-05,
      "loss": 0.5324,
      "step": 8857
    },
    {
      "epoch": 0.49602419083883975,
      "grad_norm": 1.2763642072677612,
      "learning_rate": 4.4265000000000004e-05,
      "loss": 0.4157,
      "step": 8858
    },
    {
      "epoch": 0.49608018815096877,
      "grad_norm": 1.1704176664352417,
      "learning_rate": 4.427e-05,
      "loss": 0.4128,
      "step": 8859
    },
    {
      "epoch": 0.4961361854630978,
      "grad_norm": 1.387826681137085,
      "learning_rate": 4.4275e-05,
      "loss": 0.5283,
      "step": 8860
    },
    {
      "epoch": 0.4961921827752268,
      "grad_norm": 1.2293310165405273,
      "learning_rate": 4.428e-05,
      "loss": 0.548,
      "step": 8861
    },
    {
      "epoch": 0.49624818008735583,
      "grad_norm": 1.2959572076797485,
      "learning_rate": 4.428500000000001e-05,
      "loss": 0.4113,
      "step": 8862
    },
    {
      "epoch": 0.49630417739948485,
      "grad_norm": 1.3615657091140747,
      "learning_rate": 4.4290000000000005e-05,
      "loss": 0.4515,
      "step": 8863
    },
    {
      "epoch": 0.49636017471161387,
      "grad_norm": 1.145766019821167,
      "learning_rate": 4.4295e-05,
      "loss": 0.383,
      "step": 8864
    },
    {
      "epoch": 0.4964161720237429,
      "grad_norm": 1.296778678894043,
      "learning_rate": 4.43e-05,
      "loss": 0.4408,
      "step": 8865
    },
    {
      "epoch": 0.4964721693358719,
      "grad_norm": 1.1632485389709473,
      "learning_rate": 4.4305000000000004e-05,
      "loss": 0.4587,
      "step": 8866
    },
    {
      "epoch": 0.49652816664800087,
      "grad_norm": 1.2473266124725342,
      "learning_rate": 4.431e-05,
      "loss": 0.4444,
      "step": 8867
    },
    {
      "epoch": 0.4965841639601299,
      "grad_norm": 1.2199045419692993,
      "learning_rate": 4.4315e-05,
      "loss": 0.3768,
      "step": 8868
    },
    {
      "epoch": 0.4966401612722589,
      "grad_norm": 1.3166041374206543,
      "learning_rate": 4.432e-05,
      "loss": 0.4984,
      "step": 8869
    },
    {
      "epoch": 0.4966961585843879,
      "grad_norm": 1.2682785987854004,
      "learning_rate": 4.4325e-05,
      "loss": 0.461,
      "step": 8870
    },
    {
      "epoch": 0.49675215589651694,
      "grad_norm": 1.5442527532577515,
      "learning_rate": 4.4330000000000004e-05,
      "loss": 0.44,
      "step": 8871
    },
    {
      "epoch": 0.49680815320864596,
      "grad_norm": 1.106680154800415,
      "learning_rate": 4.4335e-05,
      "loss": 0.3382,
      "step": 8872
    },
    {
      "epoch": 0.496864150520775,
      "grad_norm": 1.2025519609451294,
      "learning_rate": 4.4340000000000006e-05,
      "loss": 0.4117,
      "step": 8873
    },
    {
      "epoch": 0.496920147832904,
      "grad_norm": 1.3306968212127686,
      "learning_rate": 4.4345e-05,
      "loss": 0.4045,
      "step": 8874
    },
    {
      "epoch": 0.496976145145033,
      "grad_norm": 1.4084997177124023,
      "learning_rate": 4.435e-05,
      "loss": 0.7849,
      "step": 8875
    },
    {
      "epoch": 0.49703214245716204,
      "grad_norm": 1.4674460887908936,
      "learning_rate": 4.4355000000000005e-05,
      "loss": 0.4156,
      "step": 8876
    },
    {
      "epoch": 0.49708813976929106,
      "grad_norm": 1.2347136735916138,
      "learning_rate": 4.436e-05,
      "loss": 0.3831,
      "step": 8877
    },
    {
      "epoch": 0.4971441370814201,
      "grad_norm": 1.287600040435791,
      "learning_rate": 4.4365e-05,
      "loss": 0.3978,
      "step": 8878
    },
    {
      "epoch": 0.4972001343935491,
      "grad_norm": 1.413364052772522,
      "learning_rate": 4.4370000000000004e-05,
      "loss": 0.3869,
      "step": 8879
    },
    {
      "epoch": 0.4972561317056781,
      "grad_norm": 1.6367324590682983,
      "learning_rate": 4.4375e-05,
      "loss": 0.4517,
      "step": 8880
    },
    {
      "epoch": 0.49731212901780714,
      "grad_norm": 1.4247015714645386,
      "learning_rate": 4.438e-05,
      "loss": 0.4257,
      "step": 8881
    },
    {
      "epoch": 0.49736812632993616,
      "grad_norm": 1.250550627708435,
      "learning_rate": 4.4385e-05,
      "loss": 0.5311,
      "step": 8882
    },
    {
      "epoch": 0.4974241236420652,
      "grad_norm": 1.1516586542129517,
      "learning_rate": 4.439000000000001e-05,
      "loss": 0.469,
      "step": 8883
    },
    {
      "epoch": 0.4974801209541942,
      "grad_norm": 1.2464271783828735,
      "learning_rate": 4.4395000000000004e-05,
      "loss": 0.5226,
      "step": 8884
    },
    {
      "epoch": 0.4975361182663232,
      "grad_norm": 1.1471192836761475,
      "learning_rate": 4.44e-05,
      "loss": 0.4076,
      "step": 8885
    },
    {
      "epoch": 0.49759211557845223,
      "grad_norm": 1.119562029838562,
      "learning_rate": 4.4405e-05,
      "loss": 0.4719,
      "step": 8886
    },
    {
      "epoch": 0.49764811289058125,
      "grad_norm": 1.217142105102539,
      "learning_rate": 4.4410000000000003e-05,
      "loss": 0.3851,
      "step": 8887
    },
    {
      "epoch": 0.49770411020271027,
      "grad_norm": 1.3142110109329224,
      "learning_rate": 4.4415e-05,
      "loss": 0.5367,
      "step": 8888
    },
    {
      "epoch": 0.4977601075148393,
      "grad_norm": 1.2821145057678223,
      "learning_rate": 4.442e-05,
      "loss": 0.3896,
      "step": 8889
    },
    {
      "epoch": 0.4978161048269683,
      "grad_norm": 1.1195389032363892,
      "learning_rate": 4.4425e-05,
      "loss": 0.3382,
      "step": 8890
    },
    {
      "epoch": 0.49787210213909733,
      "grad_norm": 1.240980863571167,
      "learning_rate": 4.443e-05,
      "loss": 0.5295,
      "step": 8891
    },
    {
      "epoch": 0.49792809945122635,
      "grad_norm": 1.1518019437789917,
      "learning_rate": 4.4435000000000004e-05,
      "loss": 0.3763,
      "step": 8892
    },
    {
      "epoch": 0.49798409676335537,
      "grad_norm": 1.1410083770751953,
      "learning_rate": 4.444e-05,
      "loss": 0.4042,
      "step": 8893
    },
    {
      "epoch": 0.4980400940754844,
      "grad_norm": 1.1585884094238281,
      "learning_rate": 4.4445000000000006e-05,
      "loss": 0.4042,
      "step": 8894
    },
    {
      "epoch": 0.4980960913876134,
      "grad_norm": 1.2835559844970703,
      "learning_rate": 4.445e-05,
      "loss": 0.5014,
      "step": 8895
    },
    {
      "epoch": 0.4981520886997424,
      "grad_norm": 1.4558626413345337,
      "learning_rate": 4.4455e-05,
      "loss": 0.4909,
      "step": 8896
    },
    {
      "epoch": 0.49820808601187144,
      "grad_norm": 1.0214385986328125,
      "learning_rate": 4.4460000000000005e-05,
      "loss": 0.3897,
      "step": 8897
    },
    {
      "epoch": 0.49826408332400046,
      "grad_norm": 1.1316633224487305,
      "learning_rate": 4.4465e-05,
      "loss": 0.4244,
      "step": 8898
    },
    {
      "epoch": 0.4983200806361295,
      "grad_norm": 1.2841393947601318,
      "learning_rate": 4.447e-05,
      "loss": 0.3288,
      "step": 8899
    },
    {
      "epoch": 0.4983760779482585,
      "grad_norm": 1.366647720336914,
      "learning_rate": 4.4475e-05,
      "loss": 0.4425,
      "step": 8900
    },
    {
      "epoch": 0.4984320752603875,
      "grad_norm": 1.5029202699661255,
      "learning_rate": 4.448e-05,
      "loss": 0.3958,
      "step": 8901
    },
    {
      "epoch": 0.49848807257251654,
      "grad_norm": 1.1440002918243408,
      "learning_rate": 4.4485000000000005e-05,
      "loss": 0.4056,
      "step": 8902
    },
    {
      "epoch": 0.49854406988464556,
      "grad_norm": 1.1052082777023315,
      "learning_rate": 4.449e-05,
      "loss": 0.4626,
      "step": 8903
    },
    {
      "epoch": 0.4986000671967746,
      "grad_norm": 1.1949622631072998,
      "learning_rate": 4.4495e-05,
      "loss": 0.3967,
      "step": 8904
    },
    {
      "epoch": 0.4986560645089036,
      "grad_norm": 1.3871986865997314,
      "learning_rate": 4.4500000000000004e-05,
      "loss": 0.4062,
      "step": 8905
    },
    {
      "epoch": 0.4987120618210326,
      "grad_norm": 1.0678956508636475,
      "learning_rate": 4.4505e-05,
      "loss": 0.3173,
      "step": 8906
    },
    {
      "epoch": 0.49876805913316163,
      "grad_norm": 1.2354152202606201,
      "learning_rate": 4.451e-05,
      "loss": 0.426,
      "step": 8907
    },
    {
      "epoch": 0.49882405644529065,
      "grad_norm": 1.5488899946212769,
      "learning_rate": 4.4515e-05,
      "loss": 0.5341,
      "step": 8908
    },
    {
      "epoch": 0.4988800537574196,
      "grad_norm": 1.7664490938186646,
      "learning_rate": 4.452e-05,
      "loss": 0.4924,
      "step": 8909
    },
    {
      "epoch": 0.49893605106954864,
      "grad_norm": 1.1872198581695557,
      "learning_rate": 4.4525e-05,
      "loss": 0.4802,
      "step": 8910
    },
    {
      "epoch": 0.49899204838167766,
      "grad_norm": 1.6726036071777344,
      "learning_rate": 4.453e-05,
      "loss": 0.7142,
      "step": 8911
    },
    {
      "epoch": 0.4990480456938067,
      "grad_norm": 1.430335521697998,
      "learning_rate": 4.4535000000000006e-05,
      "loss": 0.4979,
      "step": 8912
    },
    {
      "epoch": 0.4991040430059357,
      "grad_norm": 1.2542810440063477,
      "learning_rate": 4.4540000000000004e-05,
      "loss": 0.4055,
      "step": 8913
    },
    {
      "epoch": 0.4991600403180647,
      "grad_norm": 1.1580100059509277,
      "learning_rate": 4.4545e-05,
      "loss": 0.4304,
      "step": 8914
    },
    {
      "epoch": 0.49921603763019373,
      "grad_norm": 1.0954418182373047,
      "learning_rate": 4.4550000000000005e-05,
      "loss": 0.3711,
      "step": 8915
    },
    {
      "epoch": 0.49927203494232275,
      "grad_norm": 1.2141430377960205,
      "learning_rate": 4.4555e-05,
      "loss": 0.3588,
      "step": 8916
    },
    {
      "epoch": 0.49932803225445177,
      "grad_norm": 1.110926628112793,
      "learning_rate": 4.456e-05,
      "loss": 0.3741,
      "step": 8917
    },
    {
      "epoch": 0.4993840295665808,
      "grad_norm": 1.374337077140808,
      "learning_rate": 4.4565000000000004e-05,
      "loss": 0.4762,
      "step": 8918
    },
    {
      "epoch": 0.4994400268787098,
      "grad_norm": 1.0902456045150757,
      "learning_rate": 4.457e-05,
      "loss": 0.4118,
      "step": 8919
    },
    {
      "epoch": 0.4994960241908388,
      "grad_norm": 1.2432687282562256,
      "learning_rate": 4.4575e-05,
      "loss": 0.4993,
      "step": 8920
    },
    {
      "epoch": 0.49955202150296785,
      "grad_norm": 1.3850815296173096,
      "learning_rate": 4.458e-05,
      "loss": 0.5616,
      "step": 8921
    },
    {
      "epoch": 0.49960801881509687,
      "grad_norm": 1.1992584466934204,
      "learning_rate": 4.458500000000001e-05,
      "loss": 0.4051,
      "step": 8922
    },
    {
      "epoch": 0.4996640161272259,
      "grad_norm": 1.339229941368103,
      "learning_rate": 4.4590000000000005e-05,
      "loss": 0.5669,
      "step": 8923
    },
    {
      "epoch": 0.4997200134393549,
      "grad_norm": 1.2795366048812866,
      "learning_rate": 4.4595e-05,
      "loss": 0.5839,
      "step": 8924
    },
    {
      "epoch": 0.4997760107514839,
      "grad_norm": 1.505853295326233,
      "learning_rate": 4.46e-05,
      "loss": 0.4764,
      "step": 8925
    },
    {
      "epoch": 0.49983200806361294,
      "grad_norm": 1.1352922916412354,
      "learning_rate": 4.4605000000000004e-05,
      "loss": 0.4109,
      "step": 8926
    },
    {
      "epoch": 0.49988800537574196,
      "grad_norm": 1.429019570350647,
      "learning_rate": 4.461e-05,
      "loss": 0.5747,
      "step": 8927
    },
    {
      "epoch": 0.499944002687871,
      "grad_norm": 1.4581761360168457,
      "learning_rate": 4.4615e-05,
      "loss": 0.4963,
      "step": 8928
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2715257406234741,
      "learning_rate": 4.462e-05,
      "loss": 0.3885,
      "step": 8929
    },
    {
      "epoch": 0.500055997312129,
      "grad_norm": 1.158057689666748,
      "learning_rate": 4.4625e-05,
      "loss": 0.368,
      "step": 8930
    },
    {
      "epoch": 0.500111994624258,
      "grad_norm": 1.2566872835159302,
      "learning_rate": 4.463e-05,
      "loss": 0.4259,
      "step": 8931
    },
    {
      "epoch": 0.5001679919363871,
      "grad_norm": 1.050191044807434,
      "learning_rate": 4.4635e-05,
      "loss": 0.3771,
      "step": 8932
    },
    {
      "epoch": 0.5002239892485161,
      "grad_norm": 1.2480908632278442,
      "learning_rate": 4.4640000000000006e-05,
      "loss": 0.5151,
      "step": 8933
    },
    {
      "epoch": 0.5002799865606451,
      "grad_norm": 1.6038448810577393,
      "learning_rate": 4.4645000000000004e-05,
      "loss": 0.4815,
      "step": 8934
    },
    {
      "epoch": 0.5003359838727741,
      "grad_norm": 1.3310796022415161,
      "learning_rate": 4.465e-05,
      "loss": 0.4337,
      "step": 8935
    },
    {
      "epoch": 0.5003919811849031,
      "grad_norm": 1.0335931777954102,
      "learning_rate": 4.4655000000000005e-05,
      "loss": 0.3021,
      "step": 8936
    },
    {
      "epoch": 0.5004479784970322,
      "grad_norm": 1.6768550872802734,
      "learning_rate": 4.466e-05,
      "loss": 0.5165,
      "step": 8937
    },
    {
      "epoch": 0.5005039758091612,
      "grad_norm": 1.3291367292404175,
      "learning_rate": 4.4665e-05,
      "loss": 0.3469,
      "step": 8938
    },
    {
      "epoch": 0.5005599731212902,
      "grad_norm": 1.7011210918426514,
      "learning_rate": 4.467e-05,
      "loss": 0.4439,
      "step": 8939
    },
    {
      "epoch": 0.5006159704334192,
      "grad_norm": 1.314420461654663,
      "learning_rate": 4.4675e-05,
      "loss": 0.4314,
      "step": 8940
    },
    {
      "epoch": 0.5006719677455482,
      "grad_norm": 1.4272866249084473,
      "learning_rate": 4.468e-05,
      "loss": 0.6066,
      "step": 8941
    },
    {
      "epoch": 0.5007279650576772,
      "grad_norm": 1.32636559009552,
      "learning_rate": 4.4685e-05,
      "loss": 0.391,
      "step": 8942
    },
    {
      "epoch": 0.5007839623698063,
      "grad_norm": 1.1752004623413086,
      "learning_rate": 4.469e-05,
      "loss": 0.4544,
      "step": 8943
    },
    {
      "epoch": 0.5008399596819353,
      "grad_norm": 1.1247951984405518,
      "learning_rate": 4.4695000000000005e-05,
      "loss": 0.3804,
      "step": 8944
    },
    {
      "epoch": 0.5008959569940643,
      "grad_norm": 1.4395594596862793,
      "learning_rate": 4.47e-05,
      "loss": 0.3673,
      "step": 8945
    },
    {
      "epoch": 0.5009519543061933,
      "grad_norm": 1.283260464668274,
      "learning_rate": 4.4705e-05,
      "loss": 0.4168,
      "step": 8946
    },
    {
      "epoch": 0.5010079516183223,
      "grad_norm": 1.0718876123428345,
      "learning_rate": 4.4710000000000004e-05,
      "loss": 0.4354,
      "step": 8947
    },
    {
      "epoch": 0.5010639489304514,
      "grad_norm": 1.2211148738861084,
      "learning_rate": 4.4715e-05,
      "loss": 0.3799,
      "step": 8948
    },
    {
      "epoch": 0.5011199462425804,
      "grad_norm": 1.3038597106933594,
      "learning_rate": 4.472e-05,
      "loss": 0.3742,
      "step": 8949
    },
    {
      "epoch": 0.5011759435547094,
      "grad_norm": 1.9161030054092407,
      "learning_rate": 4.4725e-05,
      "loss": 0.7407,
      "step": 8950
    },
    {
      "epoch": 0.5012319408668384,
      "grad_norm": 1.223834753036499,
      "learning_rate": 4.473e-05,
      "loss": 0.3828,
      "step": 8951
    },
    {
      "epoch": 0.5012879381789674,
      "grad_norm": 1.2141587734222412,
      "learning_rate": 4.4735000000000005e-05,
      "loss": 0.3771,
      "step": 8952
    },
    {
      "epoch": 0.5013439354910965,
      "grad_norm": 1.4120736122131348,
      "learning_rate": 4.474e-05,
      "loss": 0.5211,
      "step": 8953
    },
    {
      "epoch": 0.5013999328032255,
      "grad_norm": 1.2435225248336792,
      "learning_rate": 4.4745000000000006e-05,
      "loss": 0.3964,
      "step": 8954
    },
    {
      "epoch": 0.5014559301153545,
      "grad_norm": 1.2978291511535645,
      "learning_rate": 4.4750000000000004e-05,
      "loss": 0.3875,
      "step": 8955
    },
    {
      "epoch": 0.5015119274274835,
      "grad_norm": 1.3040884733200073,
      "learning_rate": 4.4755e-05,
      "loss": 0.3999,
      "step": 8956
    },
    {
      "epoch": 0.5015679247396125,
      "grad_norm": 1.2925035953521729,
      "learning_rate": 4.4760000000000005e-05,
      "loss": 0.3641,
      "step": 8957
    },
    {
      "epoch": 0.5016239220517416,
      "grad_norm": 1.1881881952285767,
      "learning_rate": 4.4765e-05,
      "loss": 0.398,
      "step": 8958
    },
    {
      "epoch": 0.5016799193638706,
      "grad_norm": 1.1945446729660034,
      "learning_rate": 4.477e-05,
      "loss": 0.4379,
      "step": 8959
    },
    {
      "epoch": 0.5017359166759996,
      "grad_norm": 1.4754998683929443,
      "learning_rate": 4.4775e-05,
      "loss": 0.4611,
      "step": 8960
    },
    {
      "epoch": 0.5017919139881286,
      "grad_norm": 1.242138385772705,
      "learning_rate": 4.478e-05,
      "loss": 0.344,
      "step": 8961
    },
    {
      "epoch": 0.5018479113002576,
      "grad_norm": 1.179626226425171,
      "learning_rate": 4.4785000000000006e-05,
      "loss": 0.4507,
      "step": 8962
    },
    {
      "epoch": 0.5019039086123867,
      "grad_norm": 1.34430992603302,
      "learning_rate": 4.479e-05,
      "loss": 0.4302,
      "step": 8963
    },
    {
      "epoch": 0.5019599059245157,
      "grad_norm": 1.2207221984863281,
      "learning_rate": 4.4795e-05,
      "loss": 0.4214,
      "step": 8964
    },
    {
      "epoch": 0.5020159032366447,
      "grad_norm": 1.1519558429718018,
      "learning_rate": 4.4800000000000005e-05,
      "loss": 0.3989,
      "step": 8965
    },
    {
      "epoch": 0.5020719005487737,
      "grad_norm": 1.4847782850265503,
      "learning_rate": 4.4805e-05,
      "loss": 0.6974,
      "step": 8966
    },
    {
      "epoch": 0.5021278978609027,
      "grad_norm": 1.3812850713729858,
      "learning_rate": 4.481e-05,
      "loss": 0.5247,
      "step": 8967
    },
    {
      "epoch": 0.5021838951730317,
      "grad_norm": 1.2806353569030762,
      "learning_rate": 4.4815000000000004e-05,
      "loss": 0.4348,
      "step": 8968
    },
    {
      "epoch": 0.5022398924851608,
      "grad_norm": 1.4079915285110474,
      "learning_rate": 4.482e-05,
      "loss": 0.4269,
      "step": 8969
    },
    {
      "epoch": 0.5022958897972898,
      "grad_norm": 1.0850051641464233,
      "learning_rate": 4.4825e-05,
      "loss": 0.336,
      "step": 8970
    },
    {
      "epoch": 0.5023518871094187,
      "grad_norm": 1.2497671842575073,
      "learning_rate": 4.483e-05,
      "loss": 0.4074,
      "step": 8971
    },
    {
      "epoch": 0.5024078844215477,
      "grad_norm": 1.6500200033187866,
      "learning_rate": 4.483500000000001e-05,
      "loss": 0.4611,
      "step": 8972
    },
    {
      "epoch": 0.5024638817336767,
      "grad_norm": 1.4457597732543945,
      "learning_rate": 4.4840000000000004e-05,
      "loss": 0.8361,
      "step": 8973
    },
    {
      "epoch": 0.5025198790458058,
      "grad_norm": 1.1702678203582764,
      "learning_rate": 4.4845e-05,
      "loss": 0.3945,
      "step": 8974
    },
    {
      "epoch": 0.5025758763579348,
      "grad_norm": 1.2733867168426514,
      "learning_rate": 4.4850000000000006e-05,
      "loss": 0.5797,
      "step": 8975
    },
    {
      "epoch": 0.5026318736700638,
      "grad_norm": 1.2133017778396606,
      "learning_rate": 4.4855e-05,
      "loss": 0.3941,
      "step": 8976
    },
    {
      "epoch": 0.5026878709821928,
      "grad_norm": 1.2491768598556519,
      "learning_rate": 4.486e-05,
      "loss": 0.4425,
      "step": 8977
    },
    {
      "epoch": 0.5027438682943218,
      "grad_norm": 1.3059080839157104,
      "learning_rate": 4.4865e-05,
      "loss": 0.473,
      "step": 8978
    },
    {
      "epoch": 0.5027998656064508,
      "grad_norm": 1.359402060508728,
      "learning_rate": 4.487e-05,
      "loss": 0.5198,
      "step": 8979
    },
    {
      "epoch": 0.5028558629185799,
      "grad_norm": 1.8723503351211548,
      "learning_rate": 4.4875e-05,
      "loss": 0.3733,
      "step": 8980
    },
    {
      "epoch": 0.5029118602307089,
      "grad_norm": 1.2968909740447998,
      "learning_rate": 4.488e-05,
      "loss": 0.4656,
      "step": 8981
    },
    {
      "epoch": 0.5029678575428379,
      "grad_norm": 1.528132677078247,
      "learning_rate": 4.488500000000001e-05,
      "loss": 0.5171,
      "step": 8982
    },
    {
      "epoch": 0.5030238548549669,
      "grad_norm": 1.212577223777771,
      "learning_rate": 4.4890000000000006e-05,
      "loss": 0.3623,
      "step": 8983
    },
    {
      "epoch": 0.5030798521670959,
      "grad_norm": 1.7409963607788086,
      "learning_rate": 4.4895e-05,
      "loss": 0.3292,
      "step": 8984
    },
    {
      "epoch": 0.503135849479225,
      "grad_norm": 1.1049139499664307,
      "learning_rate": 4.49e-05,
      "loss": 0.358,
      "step": 8985
    },
    {
      "epoch": 0.503191846791354,
      "grad_norm": 1.239724040031433,
      "learning_rate": 4.4905000000000005e-05,
      "loss": 0.3727,
      "step": 8986
    },
    {
      "epoch": 0.503247844103483,
      "grad_norm": 1.2785840034484863,
      "learning_rate": 4.491e-05,
      "loss": 0.4344,
      "step": 8987
    },
    {
      "epoch": 0.503303841415612,
      "grad_norm": 1.2609670162200928,
      "learning_rate": 4.4915e-05,
      "loss": 0.483,
      "step": 8988
    },
    {
      "epoch": 0.503359838727741,
      "grad_norm": 1.1001750230789185,
      "learning_rate": 4.4920000000000004e-05,
      "loss": 0.3377,
      "step": 8989
    },
    {
      "epoch": 0.5034158360398701,
      "grad_norm": 1.1395809650421143,
      "learning_rate": 4.4925e-05,
      "loss": 0.4317,
      "step": 8990
    },
    {
      "epoch": 0.5034718333519991,
      "grad_norm": 1.4517383575439453,
      "learning_rate": 4.493e-05,
      "loss": 0.399,
      "step": 8991
    },
    {
      "epoch": 0.5035278306641281,
      "grad_norm": 1.731385350227356,
      "learning_rate": 4.4935e-05,
      "loss": 0.6224,
      "step": 8992
    },
    {
      "epoch": 0.5035838279762571,
      "grad_norm": 1.2057623863220215,
      "learning_rate": 4.494000000000001e-05,
      "loss": 0.478,
      "step": 8993
    },
    {
      "epoch": 0.5036398252883861,
      "grad_norm": 2.52968692779541,
      "learning_rate": 4.4945000000000004e-05,
      "loss": 0.5506,
      "step": 8994
    },
    {
      "epoch": 0.5036958226005152,
      "grad_norm": 1.3260557651519775,
      "learning_rate": 4.495e-05,
      "loss": 0.4464,
      "step": 8995
    },
    {
      "epoch": 0.5037518199126442,
      "grad_norm": 1.177143931388855,
      "learning_rate": 4.4955000000000006e-05,
      "loss": 0.405,
      "step": 8996
    },
    {
      "epoch": 0.5038078172247732,
      "grad_norm": 1.2711893320083618,
      "learning_rate": 4.496e-05,
      "loss": 0.4228,
      "step": 8997
    },
    {
      "epoch": 0.5038638145369022,
      "grad_norm": 1.2666378021240234,
      "learning_rate": 4.4965e-05,
      "loss": 0.4483,
      "step": 8998
    },
    {
      "epoch": 0.5039198118490312,
      "grad_norm": 1.1410467624664307,
      "learning_rate": 4.497e-05,
      "loss": 0.3531,
      "step": 8999
    },
    {
      "epoch": 0.5039758091611602,
      "grad_norm": 1.304840087890625,
      "learning_rate": 4.4975e-05,
      "loss": 0.4781,
      "step": 9000
    },
    {
      "epoch": 0.5040318064732893,
      "grad_norm": 1.2505258321762085,
      "learning_rate": 4.498e-05,
      "loss": 0.4548,
      "step": 9001
    },
    {
      "epoch": 0.5040878037854183,
      "grad_norm": 1.2904996871948242,
      "learning_rate": 4.4985000000000004e-05,
      "loss": 0.4078,
      "step": 9002
    },
    {
      "epoch": 0.5041438010975473,
      "grad_norm": 1.3471643924713135,
      "learning_rate": 4.499e-05,
      "loss": 0.6741,
      "step": 9003
    },
    {
      "epoch": 0.5041997984096763,
      "grad_norm": 1.1887874603271484,
      "learning_rate": 4.4995000000000005e-05,
      "loss": 0.4686,
      "step": 9004
    },
    {
      "epoch": 0.5042557957218053,
      "grad_norm": 1.4074841737747192,
      "learning_rate": 4.5e-05,
      "loss": 0.6854,
      "step": 9005
    },
    {
      "epoch": 0.5043117930339344,
      "grad_norm": 1.458162546157837,
      "learning_rate": 4.5005e-05,
      "loss": 0.4305,
      "step": 9006
    },
    {
      "epoch": 0.5043677903460634,
      "grad_norm": 1.4055298566818237,
      "learning_rate": 4.5010000000000004e-05,
      "loss": 0.3209,
      "step": 9007
    },
    {
      "epoch": 0.5044237876581924,
      "grad_norm": 1.2616369724273682,
      "learning_rate": 4.5015e-05,
      "loss": 0.4925,
      "step": 9008
    },
    {
      "epoch": 0.5044797849703214,
      "grad_norm": 1.2956013679504395,
      "learning_rate": 4.502e-05,
      "loss": 0.4253,
      "step": 9009
    },
    {
      "epoch": 0.5045357822824504,
      "grad_norm": 1.3505840301513672,
      "learning_rate": 4.5025000000000003e-05,
      "loss": 0.5589,
      "step": 9010
    },
    {
      "epoch": 0.5045917795945795,
      "grad_norm": 1.4064513444900513,
      "learning_rate": 4.503e-05,
      "loss": 0.5392,
      "step": 9011
    },
    {
      "epoch": 0.5046477769067085,
      "grad_norm": 2.8027055263519287,
      "learning_rate": 4.5035e-05,
      "loss": 0.5942,
      "step": 9012
    },
    {
      "epoch": 0.5047037742188375,
      "grad_norm": 1.185011863708496,
      "learning_rate": 4.504e-05,
      "loss": 0.4601,
      "step": 9013
    },
    {
      "epoch": 0.5047597715309665,
      "grad_norm": 1.240644931793213,
      "learning_rate": 4.504500000000001e-05,
      "loss": 0.4146,
      "step": 9014
    },
    {
      "epoch": 0.5048157688430955,
      "grad_norm": 1.5560582876205444,
      "learning_rate": 4.5050000000000004e-05,
      "loss": 0.4718,
      "step": 9015
    },
    {
      "epoch": 0.5048717661552246,
      "grad_norm": 1.1165698766708374,
      "learning_rate": 4.5055e-05,
      "loss": 0.423,
      "step": 9016
    },
    {
      "epoch": 0.5049277634673536,
      "grad_norm": 1.8999931812286377,
      "learning_rate": 4.506e-05,
      "loss": 0.3922,
      "step": 9017
    },
    {
      "epoch": 0.5049837607794826,
      "grad_norm": 1.4183663129806519,
      "learning_rate": 4.5065e-05,
      "loss": 0.4596,
      "step": 9018
    },
    {
      "epoch": 0.5050397580916116,
      "grad_norm": 1.1907249689102173,
      "learning_rate": 4.507e-05,
      "loss": 0.3696,
      "step": 9019
    },
    {
      "epoch": 0.5050957554037406,
      "grad_norm": 1.2172504663467407,
      "learning_rate": 4.5075e-05,
      "loss": 0.4613,
      "step": 9020
    },
    {
      "epoch": 0.5051517527158697,
      "grad_norm": 1.169787883758545,
      "learning_rate": 4.508e-05,
      "loss": 0.364,
      "step": 9021
    },
    {
      "epoch": 0.5052077500279987,
      "grad_norm": 1.4569367170333862,
      "learning_rate": 4.5085e-05,
      "loss": 0.4964,
      "step": 9022
    },
    {
      "epoch": 0.5052637473401277,
      "grad_norm": 1.2050981521606445,
      "learning_rate": 4.5090000000000004e-05,
      "loss": 0.3919,
      "step": 9023
    },
    {
      "epoch": 0.5053197446522567,
      "grad_norm": 1.0716274976730347,
      "learning_rate": 4.5095e-05,
      "loss": 0.3925,
      "step": 9024
    },
    {
      "epoch": 0.5053757419643857,
      "grad_norm": 1.2640608549118042,
      "learning_rate": 4.5100000000000005e-05,
      "loss": 0.397,
      "step": 9025
    },
    {
      "epoch": 0.5054317392765147,
      "grad_norm": 1.3346258401870728,
      "learning_rate": 4.5105e-05,
      "loss": 0.3929,
      "step": 9026
    },
    {
      "epoch": 0.5054877365886438,
      "grad_norm": 1.1225422620773315,
      "learning_rate": 4.511e-05,
      "loss": 0.3526,
      "step": 9027
    },
    {
      "epoch": 0.5055437339007728,
      "grad_norm": 1.5367369651794434,
      "learning_rate": 4.5115000000000004e-05,
      "loss": 0.4391,
      "step": 9028
    },
    {
      "epoch": 0.5055997312129018,
      "grad_norm": 1.249927282333374,
      "learning_rate": 4.512e-05,
      "loss": 0.4161,
      "step": 9029
    },
    {
      "epoch": 0.5056557285250308,
      "grad_norm": 1.2556788921356201,
      "learning_rate": 4.5125e-05,
      "loss": 0.3548,
      "step": 9030
    },
    {
      "epoch": 0.5057117258371598,
      "grad_norm": 1.3210538625717163,
      "learning_rate": 4.513e-05,
      "loss": 0.5145,
      "step": 9031
    },
    {
      "epoch": 0.5057677231492889,
      "grad_norm": 1.4485116004943848,
      "learning_rate": 4.5135e-05,
      "loss": 0.4993,
      "step": 9032
    },
    {
      "epoch": 0.5058237204614179,
      "grad_norm": 1.3486223220825195,
      "learning_rate": 4.5140000000000005e-05,
      "loss": 0.4924,
      "step": 9033
    },
    {
      "epoch": 0.5058797177735469,
      "grad_norm": 1.189575433731079,
      "learning_rate": 4.5145e-05,
      "loss": 0.5028,
      "step": 9034
    },
    {
      "epoch": 0.5059357150856759,
      "grad_norm": 1.2943909168243408,
      "learning_rate": 4.5150000000000006e-05,
      "loss": 0.4446,
      "step": 9035
    },
    {
      "epoch": 0.5059917123978049,
      "grad_norm": 1.1260215044021606,
      "learning_rate": 4.5155000000000004e-05,
      "loss": 0.426,
      "step": 9036
    },
    {
      "epoch": 0.506047709709934,
      "grad_norm": 1.3268543481826782,
      "learning_rate": 4.516e-05,
      "loss": 0.6069,
      "step": 9037
    },
    {
      "epoch": 0.506103707022063,
      "grad_norm": 1.2931846380233765,
      "learning_rate": 4.5165e-05,
      "loss": 0.4658,
      "step": 9038
    },
    {
      "epoch": 0.506159704334192,
      "grad_norm": 1.1556150913238525,
      "learning_rate": 4.517e-05,
      "loss": 0.3782,
      "step": 9039
    },
    {
      "epoch": 0.506215701646321,
      "grad_norm": 1.1783028841018677,
      "learning_rate": 4.5175e-05,
      "loss": 0.3293,
      "step": 9040
    },
    {
      "epoch": 0.50627169895845,
      "grad_norm": 1.607237696647644,
      "learning_rate": 4.518e-05,
      "loss": 0.4597,
      "step": 9041
    },
    {
      "epoch": 0.506327696270579,
      "grad_norm": 1.459184169769287,
      "learning_rate": 4.5185e-05,
      "loss": 0.4745,
      "step": 9042
    },
    {
      "epoch": 0.5063836935827081,
      "grad_norm": 1.2248507738113403,
      "learning_rate": 4.5190000000000006e-05,
      "loss": 0.5046,
      "step": 9043
    },
    {
      "epoch": 0.5064396908948371,
      "grad_norm": 1.4197899103164673,
      "learning_rate": 4.5195000000000004e-05,
      "loss": 0.5367,
      "step": 9044
    },
    {
      "epoch": 0.5064956882069661,
      "grad_norm": 1.3539178371429443,
      "learning_rate": 4.52e-05,
      "loss": 0.4222,
      "step": 9045
    },
    {
      "epoch": 0.5065516855190951,
      "grad_norm": 1.159487247467041,
      "learning_rate": 4.5205000000000005e-05,
      "loss": 0.4317,
      "step": 9046
    },
    {
      "epoch": 0.5066076828312241,
      "grad_norm": 1.3368951082229614,
      "learning_rate": 4.521e-05,
      "loss": 0.449,
      "step": 9047
    },
    {
      "epoch": 0.5066636801433532,
      "grad_norm": 1.304384469985962,
      "learning_rate": 4.5215e-05,
      "loss": 0.7044,
      "step": 9048
    },
    {
      "epoch": 0.5067196774554822,
      "grad_norm": 1.4194291830062866,
      "learning_rate": 4.5220000000000004e-05,
      "loss": 0.4021,
      "step": 9049
    },
    {
      "epoch": 0.5067756747676112,
      "grad_norm": 1.290582537651062,
      "learning_rate": 4.5225e-05,
      "loss": 0.5375,
      "step": 9050
    },
    {
      "epoch": 0.5068316720797402,
      "grad_norm": 2.5563111305236816,
      "learning_rate": 4.523e-05,
      "loss": 0.368,
      "step": 9051
    },
    {
      "epoch": 0.5068876693918692,
      "grad_norm": 1.0181171894073486,
      "learning_rate": 4.5234999999999996e-05,
      "loss": 0.3027,
      "step": 9052
    },
    {
      "epoch": 0.5069436667039983,
      "grad_norm": 1.1910600662231445,
      "learning_rate": 4.524000000000001e-05,
      "loss": 0.3645,
      "step": 9053
    },
    {
      "epoch": 0.5069996640161272,
      "grad_norm": 1.234055995941162,
      "learning_rate": 4.5245000000000005e-05,
      "loss": 0.4281,
      "step": 9054
    },
    {
      "epoch": 0.5070556613282562,
      "grad_norm": 1.3286703824996948,
      "learning_rate": 4.525e-05,
      "loss": 0.4125,
      "step": 9055
    },
    {
      "epoch": 0.5071116586403852,
      "grad_norm": 1.4298676252365112,
      "learning_rate": 4.5255000000000006e-05,
      "loss": 0.4882,
      "step": 9056
    },
    {
      "epoch": 0.5071676559525142,
      "grad_norm": 1.1724324226379395,
      "learning_rate": 4.5260000000000004e-05,
      "loss": 0.3975,
      "step": 9057
    },
    {
      "epoch": 0.5072236532646432,
      "grad_norm": 1.190553903579712,
      "learning_rate": 4.5265e-05,
      "loss": 0.4498,
      "step": 9058
    },
    {
      "epoch": 0.5072796505767723,
      "grad_norm": 1.2644656896591187,
      "learning_rate": 4.527e-05,
      "loss": 0.4226,
      "step": 9059
    },
    {
      "epoch": 0.5073356478889013,
      "grad_norm": 1.5083469152450562,
      "learning_rate": 4.5275e-05,
      "loss": 0.3173,
      "step": 9060
    },
    {
      "epoch": 0.5073916452010303,
      "grad_norm": 1.3417754173278809,
      "learning_rate": 4.528e-05,
      "loss": 0.3594,
      "step": 9061
    },
    {
      "epoch": 0.5074476425131593,
      "grad_norm": 1.2471858263015747,
      "learning_rate": 4.5285e-05,
      "loss": 0.3968,
      "step": 9062
    },
    {
      "epoch": 0.5075036398252883,
      "grad_norm": 1.3608206510543823,
      "learning_rate": 4.529e-05,
      "loss": 0.4765,
      "step": 9063
    },
    {
      "epoch": 0.5075596371374174,
      "grad_norm": 1.4405231475830078,
      "learning_rate": 4.5295000000000006e-05,
      "loss": 0.5328,
      "step": 9064
    },
    {
      "epoch": 0.5076156344495464,
      "grad_norm": 1.1082167625427246,
      "learning_rate": 4.53e-05,
      "loss": 0.3742,
      "step": 9065
    },
    {
      "epoch": 0.5076716317616754,
      "grad_norm": 1.232535481452942,
      "learning_rate": 4.5305e-05,
      "loss": 0.3588,
      "step": 9066
    },
    {
      "epoch": 0.5077276290738044,
      "grad_norm": 1.1461641788482666,
      "learning_rate": 4.5310000000000005e-05,
      "loss": 0.4133,
      "step": 9067
    },
    {
      "epoch": 0.5077836263859334,
      "grad_norm": 1.3012218475341797,
      "learning_rate": 4.5315e-05,
      "loss": 0.5193,
      "step": 9068
    },
    {
      "epoch": 0.5078396236980625,
      "grad_norm": 1.2144492864608765,
      "learning_rate": 4.532e-05,
      "loss": 0.3894,
      "step": 9069
    },
    {
      "epoch": 0.5078956210101915,
      "grad_norm": 1.208936333656311,
      "learning_rate": 4.5325000000000004e-05,
      "loss": 0.39,
      "step": 9070
    },
    {
      "epoch": 0.5079516183223205,
      "grad_norm": 1.1332958936691284,
      "learning_rate": 4.533e-05,
      "loss": 0.3271,
      "step": 9071
    },
    {
      "epoch": 0.5080076156344495,
      "grad_norm": 1.1151103973388672,
      "learning_rate": 4.5335e-05,
      "loss": 0.3278,
      "step": 9072
    },
    {
      "epoch": 0.5080636129465785,
      "grad_norm": 1.5224456787109375,
      "learning_rate": 4.534e-05,
      "loss": 0.485,
      "step": 9073
    },
    {
      "epoch": 0.5081196102587076,
      "grad_norm": 1.3664674758911133,
      "learning_rate": 4.534500000000001e-05,
      "loss": 0.5187,
      "step": 9074
    },
    {
      "epoch": 0.5081756075708366,
      "grad_norm": 1.2341505289077759,
      "learning_rate": 4.5350000000000005e-05,
      "loss": 0.3897,
      "step": 9075
    },
    {
      "epoch": 0.5082316048829656,
      "grad_norm": 1.1455440521240234,
      "learning_rate": 4.5355e-05,
      "loss": 0.3157,
      "step": 9076
    },
    {
      "epoch": 0.5082876021950946,
      "grad_norm": 1.1693487167358398,
      "learning_rate": 4.536e-05,
      "loss": 0.376,
      "step": 9077
    },
    {
      "epoch": 0.5083435995072236,
      "grad_norm": 1.324777364730835,
      "learning_rate": 4.5365000000000004e-05,
      "loss": 0.5683,
      "step": 9078
    },
    {
      "epoch": 0.5083995968193527,
      "grad_norm": 1.2605270147323608,
      "learning_rate": 4.537e-05,
      "loss": 0.3379,
      "step": 9079
    },
    {
      "epoch": 0.5084555941314817,
      "grad_norm": 1.3456287384033203,
      "learning_rate": 4.5375e-05,
      "loss": 0.554,
      "step": 9080
    },
    {
      "epoch": 0.5085115914436107,
      "grad_norm": 1.2584619522094727,
      "learning_rate": 4.538e-05,
      "loss": 0.5242,
      "step": 9081
    },
    {
      "epoch": 0.5085675887557397,
      "grad_norm": 1.4179725646972656,
      "learning_rate": 4.5385e-05,
      "loss": 0.4579,
      "step": 9082
    },
    {
      "epoch": 0.5086235860678687,
      "grad_norm": 1.4745380878448486,
      "learning_rate": 4.5390000000000004e-05,
      "loss": 0.428,
      "step": 9083
    },
    {
      "epoch": 0.5086795833799977,
      "grad_norm": 1.2324879169464111,
      "learning_rate": 4.5395e-05,
      "loss": 0.376,
      "step": 9084
    },
    {
      "epoch": 0.5087355806921268,
      "grad_norm": 1.1317589282989502,
      "learning_rate": 4.5400000000000006e-05,
      "loss": 0.2786,
      "step": 9085
    },
    {
      "epoch": 0.5087915780042558,
      "grad_norm": 1.1965456008911133,
      "learning_rate": 4.5405e-05,
      "loss": 0.3678,
      "step": 9086
    },
    {
      "epoch": 0.5088475753163848,
      "grad_norm": 1.6246026754379272,
      "learning_rate": 4.541e-05,
      "loss": 0.4555,
      "step": 9087
    },
    {
      "epoch": 0.5089035726285138,
      "grad_norm": 1.401979684829712,
      "learning_rate": 4.5415000000000005e-05,
      "loss": 0.5249,
      "step": 9088
    },
    {
      "epoch": 0.5089595699406428,
      "grad_norm": 1.5277830362319946,
      "learning_rate": 4.542e-05,
      "loss": 0.3628,
      "step": 9089
    },
    {
      "epoch": 0.5090155672527719,
      "grad_norm": 1.3548916578292847,
      "learning_rate": 4.5425e-05,
      "loss": 0.4446,
      "step": 9090
    },
    {
      "epoch": 0.5090715645649009,
      "grad_norm": 1.2804852724075317,
      "learning_rate": 4.543e-05,
      "loss": 0.4077,
      "step": 9091
    },
    {
      "epoch": 0.5091275618770299,
      "grad_norm": 1.2474457025527954,
      "learning_rate": 4.5435e-05,
      "loss": 0.3732,
      "step": 9092
    },
    {
      "epoch": 0.5091835591891589,
      "grad_norm": 1.3139522075653076,
      "learning_rate": 4.5440000000000005e-05,
      "loss": 0.468,
      "step": 9093
    },
    {
      "epoch": 0.5092395565012879,
      "grad_norm": 1.2315027713775635,
      "learning_rate": 4.5445e-05,
      "loss": 0.488,
      "step": 9094
    },
    {
      "epoch": 0.509295553813417,
      "grad_norm": 1.3135638236999512,
      "learning_rate": 4.545000000000001e-05,
      "loss": 0.4894,
      "step": 9095
    },
    {
      "epoch": 0.509351551125546,
      "grad_norm": 1.1322346925735474,
      "learning_rate": 4.5455000000000004e-05,
      "loss": 0.3885,
      "step": 9096
    },
    {
      "epoch": 0.509407548437675,
      "grad_norm": 1.29420804977417,
      "learning_rate": 4.546e-05,
      "loss": 0.3649,
      "step": 9097
    },
    {
      "epoch": 0.509463545749804,
      "grad_norm": 1.3108958005905151,
      "learning_rate": 4.5465e-05,
      "loss": 0.4254,
      "step": 9098
    },
    {
      "epoch": 0.509519543061933,
      "grad_norm": 1.3444880247116089,
      "learning_rate": 4.5470000000000003e-05,
      "loss": 0.4003,
      "step": 9099
    },
    {
      "epoch": 0.509575540374062,
      "grad_norm": 1.1340411901474,
      "learning_rate": 4.5475e-05,
      "loss": 0.3966,
      "step": 9100
    },
    {
      "epoch": 0.5096315376861911,
      "grad_norm": 1.425068974494934,
      "learning_rate": 4.548e-05,
      "loss": 0.4551,
      "step": 9101
    },
    {
      "epoch": 0.5096875349983201,
      "grad_norm": 1.381592869758606,
      "learning_rate": 4.5485e-05,
      "loss": 0.3084,
      "step": 9102
    },
    {
      "epoch": 0.5097435323104491,
      "grad_norm": 1.239937663078308,
      "learning_rate": 4.549000000000001e-05,
      "loss": 0.4293,
      "step": 9103
    },
    {
      "epoch": 0.5097995296225781,
      "grad_norm": 1.8166515827178955,
      "learning_rate": 4.5495000000000004e-05,
      "loss": 0.593,
      "step": 9104
    },
    {
      "epoch": 0.5098555269347071,
      "grad_norm": 1.406846046447754,
      "learning_rate": 4.55e-05,
      "loss": 0.515,
      "step": 9105
    },
    {
      "epoch": 0.5099115242468362,
      "grad_norm": 1.3348393440246582,
      "learning_rate": 4.5505000000000006e-05,
      "loss": 0.4368,
      "step": 9106
    },
    {
      "epoch": 0.5099675215589652,
      "grad_norm": 1.1971629858016968,
      "learning_rate": 4.551e-05,
      "loss": 0.5018,
      "step": 9107
    },
    {
      "epoch": 0.5100235188710942,
      "grad_norm": 1.2922112941741943,
      "learning_rate": 4.5515e-05,
      "loss": 0.4869,
      "step": 9108
    },
    {
      "epoch": 0.5100795161832232,
      "grad_norm": 4.104691982269287,
      "learning_rate": 4.5520000000000005e-05,
      "loss": 0.4964,
      "step": 9109
    },
    {
      "epoch": 0.5101355134953522,
      "grad_norm": 1.2255407571792603,
      "learning_rate": 4.5525e-05,
      "loss": 0.4425,
      "step": 9110
    },
    {
      "epoch": 0.5101915108074813,
      "grad_norm": 1.3091654777526855,
      "learning_rate": 4.553e-05,
      "loss": 0.419,
      "step": 9111
    },
    {
      "epoch": 0.5102475081196103,
      "grad_norm": 1.6149533987045288,
      "learning_rate": 4.5535e-05,
      "loss": 0.4657,
      "step": 9112
    },
    {
      "epoch": 0.5103035054317393,
      "grad_norm": 1.4710536003112793,
      "learning_rate": 4.554000000000001e-05,
      "loss": 0.4854,
      "step": 9113
    },
    {
      "epoch": 0.5103595027438683,
      "grad_norm": 1.2672430276870728,
      "learning_rate": 4.5545000000000005e-05,
      "loss": 0.5837,
      "step": 9114
    },
    {
      "epoch": 0.5104155000559973,
      "grad_norm": 1.4874082803726196,
      "learning_rate": 4.555e-05,
      "loss": 0.5358,
      "step": 9115
    },
    {
      "epoch": 0.5104714973681264,
      "grad_norm": 1.204516887664795,
      "learning_rate": 4.5555e-05,
      "loss": 0.4116,
      "step": 9116
    },
    {
      "epoch": 0.5105274946802554,
      "grad_norm": 1.1308927536010742,
      "learning_rate": 4.5560000000000004e-05,
      "loss": 0.3743,
      "step": 9117
    },
    {
      "epoch": 0.5105834919923844,
      "grad_norm": 1.3372341394424438,
      "learning_rate": 4.5565e-05,
      "loss": 0.5025,
      "step": 9118
    },
    {
      "epoch": 0.5106394893045134,
      "grad_norm": 1.2648262977600098,
      "learning_rate": 4.557e-05,
      "loss": 0.432,
      "step": 9119
    },
    {
      "epoch": 0.5106954866166424,
      "grad_norm": 1.1696102619171143,
      "learning_rate": 4.5575e-05,
      "loss": 0.4272,
      "step": 9120
    },
    {
      "epoch": 0.5107514839287715,
      "grad_norm": 1.4026856422424316,
      "learning_rate": 4.558e-05,
      "loss": 0.3541,
      "step": 9121
    },
    {
      "epoch": 0.5108074812409005,
      "grad_norm": 1.13295316696167,
      "learning_rate": 4.5585e-05,
      "loss": 0.3299,
      "step": 9122
    },
    {
      "epoch": 0.5108634785530295,
      "grad_norm": 1.5681631565093994,
      "learning_rate": 4.559e-05,
      "loss": 0.3446,
      "step": 9123
    },
    {
      "epoch": 0.5109194758651585,
      "grad_norm": 1.1067805290222168,
      "learning_rate": 4.5595000000000006e-05,
      "loss": 0.3119,
      "step": 9124
    },
    {
      "epoch": 0.5109754731772875,
      "grad_norm": 1.3130533695220947,
      "learning_rate": 4.5600000000000004e-05,
      "loss": 0.4757,
      "step": 9125
    },
    {
      "epoch": 0.5110314704894166,
      "grad_norm": 1.240734577178955,
      "learning_rate": 4.5605e-05,
      "loss": 0.389,
      "step": 9126
    },
    {
      "epoch": 0.5110874678015456,
      "grad_norm": 1.3343544006347656,
      "learning_rate": 4.5610000000000005e-05,
      "loss": 0.4339,
      "step": 9127
    },
    {
      "epoch": 0.5111434651136746,
      "grad_norm": 1.2861734628677368,
      "learning_rate": 4.5615e-05,
      "loss": 0.4505,
      "step": 9128
    },
    {
      "epoch": 0.5111994624258036,
      "grad_norm": 1.5055768489837646,
      "learning_rate": 4.562e-05,
      "loss": 0.588,
      "step": 9129
    },
    {
      "epoch": 0.5112554597379326,
      "grad_norm": 1.5687438249588013,
      "learning_rate": 4.5625e-05,
      "loss": 0.4528,
      "step": 9130
    },
    {
      "epoch": 0.5113114570500616,
      "grad_norm": 1.7964882850646973,
      "learning_rate": 4.563e-05,
      "loss": 0.5269,
      "step": 9131
    },
    {
      "epoch": 0.5113674543621907,
      "grad_norm": 1.2458393573760986,
      "learning_rate": 4.5635e-05,
      "loss": 0.5181,
      "step": 9132
    },
    {
      "epoch": 0.5114234516743197,
      "grad_norm": 1.0184763669967651,
      "learning_rate": 4.564e-05,
      "loss": 0.4279,
      "step": 9133
    },
    {
      "epoch": 0.5114794489864487,
      "grad_norm": 1.3518426418304443,
      "learning_rate": 4.564500000000001e-05,
      "loss": 0.3765,
      "step": 9134
    },
    {
      "epoch": 0.5115354462985777,
      "grad_norm": 1.2622166872024536,
      "learning_rate": 4.5650000000000005e-05,
      "loss": 0.4849,
      "step": 9135
    },
    {
      "epoch": 0.5115914436107066,
      "grad_norm": 1.3343026638031006,
      "learning_rate": 4.5655e-05,
      "loss": 0.4758,
      "step": 9136
    },
    {
      "epoch": 0.5116474409228357,
      "grad_norm": 1.2146726846694946,
      "learning_rate": 4.566e-05,
      "loss": 0.4958,
      "step": 9137
    },
    {
      "epoch": 0.5117034382349647,
      "grad_norm": 1.320116400718689,
      "learning_rate": 4.5665000000000004e-05,
      "loss": 0.4295,
      "step": 9138
    },
    {
      "epoch": 0.5117594355470937,
      "grad_norm": 1.2639986276626587,
      "learning_rate": 4.567e-05,
      "loss": 0.3921,
      "step": 9139
    },
    {
      "epoch": 0.5118154328592227,
      "grad_norm": 1.1365134716033936,
      "learning_rate": 4.5675e-05,
      "loss": 0.4634,
      "step": 9140
    },
    {
      "epoch": 0.5118714301713517,
      "grad_norm": 1.3811694383621216,
      "learning_rate": 4.568e-05,
      "loss": 0.4476,
      "step": 9141
    },
    {
      "epoch": 0.5119274274834807,
      "grad_norm": 1.1554290056228638,
      "learning_rate": 4.5685e-05,
      "loss": 0.3517,
      "step": 9142
    },
    {
      "epoch": 0.5119834247956098,
      "grad_norm": 1.1563440561294556,
      "learning_rate": 4.569e-05,
      "loss": 0.4315,
      "step": 9143
    },
    {
      "epoch": 0.5120394221077388,
      "grad_norm": 1.0936301946640015,
      "learning_rate": 4.5695e-05,
      "loss": 0.4105,
      "step": 9144
    },
    {
      "epoch": 0.5120954194198678,
      "grad_norm": 1.7846345901489258,
      "learning_rate": 4.5700000000000006e-05,
      "loss": 0.4825,
      "step": 9145
    },
    {
      "epoch": 0.5121514167319968,
      "grad_norm": 1.1873970031738281,
      "learning_rate": 4.5705000000000004e-05,
      "loss": 0.4168,
      "step": 9146
    },
    {
      "epoch": 0.5122074140441258,
      "grad_norm": 1.2181586027145386,
      "learning_rate": 4.571e-05,
      "loss": 0.4442,
      "step": 9147
    },
    {
      "epoch": 0.5122634113562549,
      "grad_norm": 1.1474303007125854,
      "learning_rate": 4.5715000000000005e-05,
      "loss": 0.449,
      "step": 9148
    },
    {
      "epoch": 0.5123194086683839,
      "grad_norm": 1.3512732982635498,
      "learning_rate": 4.572e-05,
      "loss": 0.4581,
      "step": 9149
    },
    {
      "epoch": 0.5123754059805129,
      "grad_norm": 1.2675834894180298,
      "learning_rate": 4.5725e-05,
      "loss": 0.3983,
      "step": 9150
    },
    {
      "epoch": 0.5124314032926419,
      "grad_norm": 1.3232731819152832,
      "learning_rate": 4.573e-05,
      "loss": 0.4474,
      "step": 9151
    },
    {
      "epoch": 0.5124874006047709,
      "grad_norm": 1.4368561506271362,
      "learning_rate": 4.5735e-05,
      "loss": 0.391,
      "step": 9152
    },
    {
      "epoch": 0.5125433979169,
      "grad_norm": 1.2953909635543823,
      "learning_rate": 4.574e-05,
      "loss": 0.4717,
      "step": 9153
    },
    {
      "epoch": 0.512599395229029,
      "grad_norm": 2.3071467876434326,
      "learning_rate": 4.5745e-05,
      "loss": 0.408,
      "step": 9154
    },
    {
      "epoch": 0.512655392541158,
      "grad_norm": 1.2890418767929077,
      "learning_rate": 4.575e-05,
      "loss": 0.3286,
      "step": 9155
    },
    {
      "epoch": 0.512711389853287,
      "grad_norm": 1.3090392351150513,
      "learning_rate": 4.5755000000000005e-05,
      "loss": 0.5139,
      "step": 9156
    },
    {
      "epoch": 0.512767387165416,
      "grad_norm": 1.228469967842102,
      "learning_rate": 4.576e-05,
      "loss": 0.4131,
      "step": 9157
    },
    {
      "epoch": 0.512823384477545,
      "grad_norm": 1.266729474067688,
      "learning_rate": 4.5765e-05,
      "loss": 0.4143,
      "step": 9158
    },
    {
      "epoch": 0.5128793817896741,
      "grad_norm": 1.4858880043029785,
      "learning_rate": 4.5770000000000004e-05,
      "loss": 0.4721,
      "step": 9159
    },
    {
      "epoch": 0.5129353791018031,
      "grad_norm": 1.380627155303955,
      "learning_rate": 4.5775e-05,
      "loss": 0.6853,
      "step": 9160
    },
    {
      "epoch": 0.5129913764139321,
      "grad_norm": 1.2912437915802002,
      "learning_rate": 4.578e-05,
      "loss": 0.3152,
      "step": 9161
    },
    {
      "epoch": 0.5130473737260611,
      "grad_norm": 1.0574392080307007,
      "learning_rate": 4.5785e-05,
      "loss": 0.3956,
      "step": 9162
    },
    {
      "epoch": 0.5131033710381901,
      "grad_norm": 0.939217746257782,
      "learning_rate": 4.579e-05,
      "loss": 0.2672,
      "step": 9163
    },
    {
      "epoch": 0.5131593683503192,
      "grad_norm": 1.147523045539856,
      "learning_rate": 4.5795000000000005e-05,
      "loss": 0.3168,
      "step": 9164
    },
    {
      "epoch": 0.5132153656624482,
      "grad_norm": 1.3794147968292236,
      "learning_rate": 4.58e-05,
      "loss": 0.657,
      "step": 9165
    },
    {
      "epoch": 0.5132713629745772,
      "grad_norm": 1.8935555219650269,
      "learning_rate": 4.5805000000000006e-05,
      "loss": 0.5039,
      "step": 9166
    },
    {
      "epoch": 0.5133273602867062,
      "grad_norm": 1.233540654182434,
      "learning_rate": 4.5810000000000004e-05,
      "loss": 0.4168,
      "step": 9167
    },
    {
      "epoch": 0.5133833575988352,
      "grad_norm": 1.2867122888565063,
      "learning_rate": 4.5815e-05,
      "loss": 0.4104,
      "step": 9168
    },
    {
      "epoch": 0.5134393549109643,
      "grad_norm": 1.1694250106811523,
      "learning_rate": 4.5820000000000005e-05,
      "loss": 0.425,
      "step": 9169
    },
    {
      "epoch": 0.5134953522230933,
      "grad_norm": 1.2046486139297485,
      "learning_rate": 4.5825e-05,
      "loss": 0.4514,
      "step": 9170
    },
    {
      "epoch": 0.5135513495352223,
      "grad_norm": 1.1943657398223877,
      "learning_rate": 4.583e-05,
      "loss": 0.617,
      "step": 9171
    },
    {
      "epoch": 0.5136073468473513,
      "grad_norm": 1.2283899784088135,
      "learning_rate": 4.5835e-05,
      "loss": 0.4312,
      "step": 9172
    },
    {
      "epoch": 0.5136633441594803,
      "grad_norm": 1.3342396020889282,
      "learning_rate": 4.584e-05,
      "loss": 0.4931,
      "step": 9173
    },
    {
      "epoch": 0.5137193414716094,
      "grad_norm": 1.4554526805877686,
      "learning_rate": 4.5845000000000006e-05,
      "loss": 0.4853,
      "step": 9174
    },
    {
      "epoch": 0.5137753387837384,
      "grad_norm": 1.2992995977401733,
      "learning_rate": 4.585e-05,
      "loss": 0.4486,
      "step": 9175
    },
    {
      "epoch": 0.5138313360958674,
      "grad_norm": 1.2688758373260498,
      "learning_rate": 4.5855e-05,
      "loss": 0.4145,
      "step": 9176
    },
    {
      "epoch": 0.5138873334079964,
      "grad_norm": 1.1902157068252563,
      "learning_rate": 4.5860000000000005e-05,
      "loss": 0.3951,
      "step": 9177
    },
    {
      "epoch": 0.5139433307201254,
      "grad_norm": 1.1584275960922241,
      "learning_rate": 4.5865e-05,
      "loss": 0.3811,
      "step": 9178
    },
    {
      "epoch": 0.5139993280322545,
      "grad_norm": 1.4463282823562622,
      "learning_rate": 4.587e-05,
      "loss": 0.4477,
      "step": 9179
    },
    {
      "epoch": 0.5140553253443835,
      "grad_norm": 1.225193738937378,
      "learning_rate": 4.5875000000000004e-05,
      "loss": 0.5085,
      "step": 9180
    },
    {
      "epoch": 0.5141113226565125,
      "grad_norm": 1.3751776218414307,
      "learning_rate": 4.588e-05,
      "loss": 0.4581,
      "step": 9181
    },
    {
      "epoch": 0.5141673199686415,
      "grad_norm": 1.1857109069824219,
      "learning_rate": 4.5885e-05,
      "loss": 0.4816,
      "step": 9182
    },
    {
      "epoch": 0.5142233172807705,
      "grad_norm": 1.116721510887146,
      "learning_rate": 4.589e-05,
      "loss": 0.418,
      "step": 9183
    },
    {
      "epoch": 0.5142793145928996,
      "grad_norm": 1.302652359008789,
      "learning_rate": 4.589500000000001e-05,
      "loss": 0.3992,
      "step": 9184
    },
    {
      "epoch": 0.5143353119050286,
      "grad_norm": 1.2706457376480103,
      "learning_rate": 4.5900000000000004e-05,
      "loss": 0.4317,
      "step": 9185
    },
    {
      "epoch": 0.5143913092171576,
      "grad_norm": 1.1368987560272217,
      "learning_rate": 4.5905e-05,
      "loss": 0.5304,
      "step": 9186
    },
    {
      "epoch": 0.5144473065292866,
      "grad_norm": 1.1194905042648315,
      "learning_rate": 4.5910000000000006e-05,
      "loss": 0.4262,
      "step": 9187
    },
    {
      "epoch": 0.5145033038414156,
      "grad_norm": 1.4616315364837646,
      "learning_rate": 4.5915000000000003e-05,
      "loss": 0.412,
      "step": 9188
    },
    {
      "epoch": 0.5145593011535446,
      "grad_norm": 1.2584593296051025,
      "learning_rate": 4.592e-05,
      "loss": 0.4486,
      "step": 9189
    },
    {
      "epoch": 0.5146152984656737,
      "grad_norm": 1.3513503074645996,
      "learning_rate": 4.5925e-05,
      "loss": 0.4367,
      "step": 9190
    },
    {
      "epoch": 0.5146712957778027,
      "grad_norm": 1.3395848274230957,
      "learning_rate": 4.593e-05,
      "loss": 0.4674,
      "step": 9191
    },
    {
      "epoch": 0.5147272930899317,
      "grad_norm": 1.45074462890625,
      "learning_rate": 4.5935e-05,
      "loss": 0.5375,
      "step": 9192
    },
    {
      "epoch": 0.5147832904020607,
      "grad_norm": 1.2556960582733154,
      "learning_rate": 4.594e-05,
      "loss": 0.4159,
      "step": 9193
    },
    {
      "epoch": 0.5148392877141897,
      "grad_norm": 1.1413027048110962,
      "learning_rate": 4.5945e-05,
      "loss": 0.3966,
      "step": 9194
    },
    {
      "epoch": 0.5148952850263188,
      "grad_norm": 1.5372223854064941,
      "learning_rate": 4.5950000000000006e-05,
      "loss": 0.3643,
      "step": 9195
    },
    {
      "epoch": 0.5149512823384478,
      "grad_norm": 1.4177385568618774,
      "learning_rate": 4.5955e-05,
      "loss": 0.48,
      "step": 9196
    },
    {
      "epoch": 0.5150072796505768,
      "grad_norm": 1.2636547088623047,
      "learning_rate": 4.596e-05,
      "loss": 0.4348,
      "step": 9197
    },
    {
      "epoch": 0.5150632769627058,
      "grad_norm": 1.465863823890686,
      "learning_rate": 4.5965000000000005e-05,
      "loss": 0.4749,
      "step": 9198
    },
    {
      "epoch": 0.5151192742748348,
      "grad_norm": 1.1423953771591187,
      "learning_rate": 4.597e-05,
      "loss": 0.3845,
      "step": 9199
    },
    {
      "epoch": 0.5151752715869639,
      "grad_norm": 1.100070834159851,
      "learning_rate": 4.5975e-05,
      "loss": 0.4243,
      "step": 9200
    },
    {
      "epoch": 0.5152312688990929,
      "grad_norm": 1.6481834650039673,
      "learning_rate": 4.5980000000000004e-05,
      "loss": 0.5836,
      "step": 9201
    },
    {
      "epoch": 0.5152872662112219,
      "grad_norm": 1.323297142982483,
      "learning_rate": 4.5985e-05,
      "loss": 0.4682,
      "step": 9202
    },
    {
      "epoch": 0.5153432635233509,
      "grad_norm": 1.2022228240966797,
      "learning_rate": 4.599e-05,
      "loss": 0.4252,
      "step": 9203
    },
    {
      "epoch": 0.5153992608354799,
      "grad_norm": 1.5649847984313965,
      "learning_rate": 4.5995e-05,
      "loss": 0.6934,
      "step": 9204
    },
    {
      "epoch": 0.515455258147609,
      "grad_norm": 1.4646174907684326,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.4992,
      "step": 9205
    },
    {
      "epoch": 0.515511255459738,
      "grad_norm": 1.2009707689285278,
      "learning_rate": 4.6005000000000004e-05,
      "loss": 0.4129,
      "step": 9206
    },
    {
      "epoch": 0.515567252771867,
      "grad_norm": 1.3114912509918213,
      "learning_rate": 4.601e-05,
      "loss": 0.3581,
      "step": 9207
    },
    {
      "epoch": 0.515623250083996,
      "grad_norm": 1.2285393476486206,
      "learning_rate": 4.6015000000000006e-05,
      "loss": 0.4393,
      "step": 9208
    },
    {
      "epoch": 0.515679247396125,
      "grad_norm": 1.554861307144165,
      "learning_rate": 4.602e-05,
      "loss": 0.4743,
      "step": 9209
    },
    {
      "epoch": 0.515735244708254,
      "grad_norm": 1.0693328380584717,
      "learning_rate": 4.6025e-05,
      "loss": 0.3656,
      "step": 9210
    },
    {
      "epoch": 0.5157912420203831,
      "grad_norm": 1.2261241674423218,
      "learning_rate": 4.603e-05,
      "loss": 0.4464,
      "step": 9211
    },
    {
      "epoch": 0.5158472393325121,
      "grad_norm": 1.3556733131408691,
      "learning_rate": 4.6035e-05,
      "loss": 0.4532,
      "step": 9212
    },
    {
      "epoch": 0.5159032366446411,
      "grad_norm": 1.05940580368042,
      "learning_rate": 4.604e-05,
      "loss": 0.3792,
      "step": 9213
    },
    {
      "epoch": 0.5159592339567701,
      "grad_norm": 0.9899119138717651,
      "learning_rate": 4.6045000000000004e-05,
      "loss": 0.362,
      "step": 9214
    },
    {
      "epoch": 0.5160152312688991,
      "grad_norm": 1.0863126516342163,
      "learning_rate": 4.605e-05,
      "loss": 0.4113,
      "step": 9215
    },
    {
      "epoch": 0.5160712285810282,
      "grad_norm": 1.3411493301391602,
      "learning_rate": 4.6055000000000005e-05,
      "loss": 0.4646,
      "step": 9216
    },
    {
      "epoch": 0.5161272258931572,
      "grad_norm": 1.2432260513305664,
      "learning_rate": 4.606e-05,
      "loss": 0.4235,
      "step": 9217
    },
    {
      "epoch": 0.5161832232052862,
      "grad_norm": 1.2353248596191406,
      "learning_rate": 4.6065e-05,
      "loss": 0.3575,
      "step": 9218
    },
    {
      "epoch": 0.5162392205174151,
      "grad_norm": 1.151160717010498,
      "learning_rate": 4.6070000000000004e-05,
      "loss": 0.3813,
      "step": 9219
    },
    {
      "epoch": 0.5162952178295441,
      "grad_norm": 1.0681962966918945,
      "learning_rate": 4.6075e-05,
      "loss": 0.2898,
      "step": 9220
    },
    {
      "epoch": 0.5163512151416731,
      "grad_norm": 1.4751173257827759,
      "learning_rate": 4.608e-05,
      "loss": 0.611,
      "step": 9221
    },
    {
      "epoch": 0.5164072124538022,
      "grad_norm": 1.1293771266937256,
      "learning_rate": 4.6085000000000003e-05,
      "loss": 0.4167,
      "step": 9222
    },
    {
      "epoch": 0.5164632097659312,
      "grad_norm": 1.486940622329712,
      "learning_rate": 4.609e-05,
      "loss": 0.5161,
      "step": 9223
    },
    {
      "epoch": 0.5165192070780602,
      "grad_norm": 1.2248575687408447,
      "learning_rate": 4.6095000000000005e-05,
      "loss": 0.5836,
      "step": 9224
    },
    {
      "epoch": 0.5165752043901892,
      "grad_norm": 1.1783866882324219,
      "learning_rate": 4.61e-05,
      "loss": 0.3845,
      "step": 9225
    },
    {
      "epoch": 0.5166312017023182,
      "grad_norm": 1.5090482234954834,
      "learning_rate": 4.610500000000001e-05,
      "loss": 0.397,
      "step": 9226
    },
    {
      "epoch": 0.5166871990144473,
      "grad_norm": 1.2617485523223877,
      "learning_rate": 4.6110000000000004e-05,
      "loss": 0.3639,
      "step": 9227
    },
    {
      "epoch": 0.5167431963265763,
      "grad_norm": 1.2481554746627808,
      "learning_rate": 4.6115e-05,
      "loss": 0.4247,
      "step": 9228
    },
    {
      "epoch": 0.5167991936387053,
      "grad_norm": 1.1832367181777954,
      "learning_rate": 4.612e-05,
      "loss": 0.3609,
      "step": 9229
    },
    {
      "epoch": 0.5168551909508343,
      "grad_norm": 1.161326289176941,
      "learning_rate": 4.6125e-05,
      "loss": 0.3136,
      "step": 9230
    },
    {
      "epoch": 0.5169111882629633,
      "grad_norm": 1.1889429092407227,
      "learning_rate": 4.613e-05,
      "loss": 0.4367,
      "step": 9231
    },
    {
      "epoch": 0.5169671855750924,
      "grad_norm": 1.656028389930725,
      "learning_rate": 4.6135e-05,
      "loss": 0.5329,
      "step": 9232
    },
    {
      "epoch": 0.5170231828872214,
      "grad_norm": 1.3792628049850464,
      "learning_rate": 4.614e-05,
      "loss": 0.4479,
      "step": 9233
    },
    {
      "epoch": 0.5170791801993504,
      "grad_norm": 1.2553712129592896,
      "learning_rate": 4.6145000000000006e-05,
      "loss": 0.4374,
      "step": 9234
    },
    {
      "epoch": 0.5171351775114794,
      "grad_norm": 1.0009866952896118,
      "learning_rate": 4.6150000000000004e-05,
      "loss": 0.3785,
      "step": 9235
    },
    {
      "epoch": 0.5171911748236084,
      "grad_norm": 1.2896463871002197,
      "learning_rate": 4.6155e-05,
      "loss": 0.4246,
      "step": 9236
    },
    {
      "epoch": 0.5172471721357375,
      "grad_norm": 1.2183150053024292,
      "learning_rate": 4.6160000000000005e-05,
      "loss": 0.3809,
      "step": 9237
    },
    {
      "epoch": 0.5173031694478665,
      "grad_norm": 1.196364402770996,
      "learning_rate": 4.6165e-05,
      "loss": 0.4453,
      "step": 9238
    },
    {
      "epoch": 0.5173591667599955,
      "grad_norm": 1.162322759628296,
      "learning_rate": 4.617e-05,
      "loss": 0.4103,
      "step": 9239
    },
    {
      "epoch": 0.5174151640721245,
      "grad_norm": 1.1721992492675781,
      "learning_rate": 4.6175000000000004e-05,
      "loss": 0.4286,
      "step": 9240
    },
    {
      "epoch": 0.5174711613842535,
      "grad_norm": 1.5863608121871948,
      "learning_rate": 4.618e-05,
      "loss": 0.4548,
      "step": 9241
    },
    {
      "epoch": 0.5175271586963825,
      "grad_norm": 1.2901023626327515,
      "learning_rate": 4.6185e-05,
      "loss": 0.5696,
      "step": 9242
    },
    {
      "epoch": 0.5175831560085116,
      "grad_norm": 1.0869500637054443,
      "learning_rate": 4.619e-05,
      "loss": 0.4111,
      "step": 9243
    },
    {
      "epoch": 0.5176391533206406,
      "grad_norm": 1.1780967712402344,
      "learning_rate": 4.619500000000001e-05,
      "loss": 0.3407,
      "step": 9244
    },
    {
      "epoch": 0.5176951506327696,
      "grad_norm": 1.2058454751968384,
      "learning_rate": 4.6200000000000005e-05,
      "loss": 0.4369,
      "step": 9245
    },
    {
      "epoch": 0.5177511479448986,
      "grad_norm": 1.329429268836975,
      "learning_rate": 4.6205e-05,
      "loss": 0.5259,
      "step": 9246
    },
    {
      "epoch": 0.5178071452570276,
      "grad_norm": 1.1401983499526978,
      "learning_rate": 4.6210000000000006e-05,
      "loss": 0.4004,
      "step": 9247
    },
    {
      "epoch": 0.5178631425691567,
      "grad_norm": 1.2211241722106934,
      "learning_rate": 4.6215000000000004e-05,
      "loss": 0.4571,
      "step": 9248
    },
    {
      "epoch": 0.5179191398812857,
      "grad_norm": 1.2139710187911987,
      "learning_rate": 4.622e-05,
      "loss": 0.3738,
      "step": 9249
    },
    {
      "epoch": 0.5179751371934147,
      "grad_norm": 1.2400494813919067,
      "learning_rate": 4.6225e-05,
      "loss": 0.4665,
      "step": 9250
    },
    {
      "epoch": 0.5180311345055437,
      "grad_norm": 1.2360652685165405,
      "learning_rate": 4.623e-05,
      "loss": 0.329,
      "step": 9251
    },
    {
      "epoch": 0.5180871318176727,
      "grad_norm": 1.4026408195495605,
      "learning_rate": 4.6235e-05,
      "loss": 0.5076,
      "step": 9252
    },
    {
      "epoch": 0.5181431291298018,
      "grad_norm": 1.1283280849456787,
      "learning_rate": 4.624e-05,
      "loss": 0.3409,
      "step": 9253
    },
    {
      "epoch": 0.5181991264419308,
      "grad_norm": 1.180201768875122,
      "learning_rate": 4.6245e-05,
      "loss": 0.3412,
      "step": 9254
    },
    {
      "epoch": 0.5182551237540598,
      "grad_norm": 1.263129711151123,
      "learning_rate": 4.6250000000000006e-05,
      "loss": 0.3948,
      "step": 9255
    },
    {
      "epoch": 0.5183111210661888,
      "grad_norm": 2.041626453399658,
      "learning_rate": 4.6255000000000004e-05,
      "loss": 0.4851,
      "step": 9256
    },
    {
      "epoch": 0.5183671183783178,
      "grad_norm": 1.2479692697525024,
      "learning_rate": 4.626e-05,
      "loss": 0.3156,
      "step": 9257
    },
    {
      "epoch": 0.5184231156904469,
      "grad_norm": 1.620681881904602,
      "learning_rate": 4.6265000000000005e-05,
      "loss": 0.418,
      "step": 9258
    },
    {
      "epoch": 0.5184791130025759,
      "grad_norm": 1.504738688468933,
      "learning_rate": 4.627e-05,
      "loss": 0.4166,
      "step": 9259
    },
    {
      "epoch": 0.5185351103147049,
      "grad_norm": 1.296774983406067,
      "learning_rate": 4.6275e-05,
      "loss": 0.4553,
      "step": 9260
    },
    {
      "epoch": 0.5185911076268339,
      "grad_norm": 1.2768787145614624,
      "learning_rate": 4.6280000000000004e-05,
      "loss": 0.4789,
      "step": 9261
    },
    {
      "epoch": 0.5186471049389629,
      "grad_norm": 1.1409201622009277,
      "learning_rate": 4.6285e-05,
      "loss": 0.4079,
      "step": 9262
    },
    {
      "epoch": 0.518703102251092,
      "grad_norm": 1.0574982166290283,
      "learning_rate": 4.629e-05,
      "loss": 0.3735,
      "step": 9263
    },
    {
      "epoch": 0.518759099563221,
      "grad_norm": 1.2961485385894775,
      "learning_rate": 4.6294999999999996e-05,
      "loss": 0.3911,
      "step": 9264
    },
    {
      "epoch": 0.51881509687535,
      "grad_norm": 1.4814493656158447,
      "learning_rate": 4.630000000000001e-05,
      "loss": 0.3772,
      "step": 9265
    },
    {
      "epoch": 0.518871094187479,
      "grad_norm": 1.281013011932373,
      "learning_rate": 4.6305000000000005e-05,
      "loss": 0.5341,
      "step": 9266
    },
    {
      "epoch": 0.518927091499608,
      "grad_norm": 1.3805668354034424,
      "learning_rate": 4.631e-05,
      "loss": 0.4929,
      "step": 9267
    },
    {
      "epoch": 0.518983088811737,
      "grad_norm": 1.4640450477600098,
      "learning_rate": 4.6315e-05,
      "loss": 0.4276,
      "step": 9268
    },
    {
      "epoch": 0.5190390861238661,
      "grad_norm": 1.4518136978149414,
      "learning_rate": 4.6320000000000004e-05,
      "loss": 0.5306,
      "step": 9269
    },
    {
      "epoch": 0.5190950834359951,
      "grad_norm": 1.4797710180282593,
      "learning_rate": 4.6325e-05,
      "loss": 0.3544,
      "step": 9270
    },
    {
      "epoch": 0.5191510807481241,
      "grad_norm": 1.166329026222229,
      "learning_rate": 4.633e-05,
      "loss": 0.3905,
      "step": 9271
    },
    {
      "epoch": 0.5192070780602531,
      "grad_norm": 1.1806344985961914,
      "learning_rate": 4.6335e-05,
      "loss": 0.4603,
      "step": 9272
    },
    {
      "epoch": 0.5192630753723821,
      "grad_norm": 1.1213579177856445,
      "learning_rate": 4.634e-05,
      "loss": 0.3816,
      "step": 9273
    },
    {
      "epoch": 0.5193190726845112,
      "grad_norm": 1.24141263961792,
      "learning_rate": 4.6345e-05,
      "loss": 0.4185,
      "step": 9274
    },
    {
      "epoch": 0.5193750699966402,
      "grad_norm": 1.2993707656860352,
      "learning_rate": 4.635e-05,
      "loss": 0.3636,
      "step": 9275
    },
    {
      "epoch": 0.5194310673087692,
      "grad_norm": 1.1595044136047363,
      "learning_rate": 4.6355000000000006e-05,
      "loss": 0.3716,
      "step": 9276
    },
    {
      "epoch": 0.5194870646208982,
      "grad_norm": 1.4520783424377441,
      "learning_rate": 4.636e-05,
      "loss": 0.4154,
      "step": 9277
    },
    {
      "epoch": 0.5195430619330272,
      "grad_norm": 1.390235185623169,
      "learning_rate": 4.6365e-05,
      "loss": 0.5374,
      "step": 9278
    },
    {
      "epoch": 0.5195990592451563,
      "grad_norm": 1.1509873867034912,
      "learning_rate": 4.6370000000000005e-05,
      "loss": 0.3998,
      "step": 9279
    },
    {
      "epoch": 0.5196550565572853,
      "grad_norm": 1.089005947113037,
      "learning_rate": 4.6375e-05,
      "loss": 0.3992,
      "step": 9280
    },
    {
      "epoch": 0.5197110538694143,
      "grad_norm": 1.2871884107589722,
      "learning_rate": 4.638e-05,
      "loss": 0.4491,
      "step": 9281
    },
    {
      "epoch": 0.5197670511815433,
      "grad_norm": 1.0660498142242432,
      "learning_rate": 4.6385000000000004e-05,
      "loss": 0.3495,
      "step": 9282
    },
    {
      "epoch": 0.5198230484936723,
      "grad_norm": 1.4733692407608032,
      "learning_rate": 4.639e-05,
      "loss": 0.388,
      "step": 9283
    },
    {
      "epoch": 0.5198790458058014,
      "grad_norm": 1.3952951431274414,
      "learning_rate": 4.6395e-05,
      "loss": 0.3995,
      "step": 9284
    },
    {
      "epoch": 0.5199350431179304,
      "grad_norm": 1.2849518060684204,
      "learning_rate": 4.64e-05,
      "loss": 0.3945,
      "step": 9285
    },
    {
      "epoch": 0.5199910404300594,
      "grad_norm": 1.2646955251693726,
      "learning_rate": 4.640500000000001e-05,
      "loss": 0.4605,
      "step": 9286
    },
    {
      "epoch": 0.5200470377421884,
      "grad_norm": 1.6200830936431885,
      "learning_rate": 4.6410000000000005e-05,
      "loss": 0.5205,
      "step": 9287
    },
    {
      "epoch": 0.5201030350543174,
      "grad_norm": 1.5152814388275146,
      "learning_rate": 4.6415e-05,
      "loss": 0.53,
      "step": 9288
    },
    {
      "epoch": 0.5201590323664464,
      "grad_norm": 1.5220857858657837,
      "learning_rate": 4.642e-05,
      "loss": 0.6232,
      "step": 9289
    },
    {
      "epoch": 0.5202150296785755,
      "grad_norm": 1.34884512424469,
      "learning_rate": 4.6425000000000004e-05,
      "loss": 0.4985,
      "step": 9290
    },
    {
      "epoch": 0.5202710269907045,
      "grad_norm": 1.3182399272918701,
      "learning_rate": 4.643e-05,
      "loss": 0.5307,
      "step": 9291
    },
    {
      "epoch": 0.5203270243028335,
      "grad_norm": 1.4727940559387207,
      "learning_rate": 4.6435e-05,
      "loss": 0.6653,
      "step": 9292
    },
    {
      "epoch": 0.5203830216149625,
      "grad_norm": 1.233260989189148,
      "learning_rate": 4.644e-05,
      "loss": 0.3941,
      "step": 9293
    },
    {
      "epoch": 0.5204390189270915,
      "grad_norm": 1.5735726356506348,
      "learning_rate": 4.6445e-05,
      "loss": 0.5239,
      "step": 9294
    },
    {
      "epoch": 0.5204950162392206,
      "grad_norm": 1.2462122440338135,
      "learning_rate": 4.6450000000000004e-05,
      "loss": 0.4223,
      "step": 9295
    },
    {
      "epoch": 0.5205510135513496,
      "grad_norm": 1.1779415607452393,
      "learning_rate": 4.6455e-05,
      "loss": 0.3389,
      "step": 9296
    },
    {
      "epoch": 0.5206070108634786,
      "grad_norm": 6.715813636779785,
      "learning_rate": 4.6460000000000006e-05,
      "loss": 0.5612,
      "step": 9297
    },
    {
      "epoch": 0.5206630081756076,
      "grad_norm": 1.1489617824554443,
      "learning_rate": 4.6465e-05,
      "loss": 0.3704,
      "step": 9298
    },
    {
      "epoch": 0.5207190054877366,
      "grad_norm": 1.129090666770935,
      "learning_rate": 4.647e-05,
      "loss": 0.4061,
      "step": 9299
    },
    {
      "epoch": 0.5207750027998657,
      "grad_norm": 1.7248512506484985,
      "learning_rate": 4.6475000000000005e-05,
      "loss": 0.4675,
      "step": 9300
    },
    {
      "epoch": 0.5208310001119947,
      "grad_norm": 1.4207942485809326,
      "learning_rate": 4.648e-05,
      "loss": 0.5119,
      "step": 9301
    },
    {
      "epoch": 0.5208869974241236,
      "grad_norm": 1.1348084211349487,
      "learning_rate": 4.6485e-05,
      "loss": 0.3737,
      "step": 9302
    },
    {
      "epoch": 0.5209429947362526,
      "grad_norm": 1.1250228881835938,
      "learning_rate": 4.649e-05,
      "loss": 0.4405,
      "step": 9303
    },
    {
      "epoch": 0.5209989920483816,
      "grad_norm": 1.207688808441162,
      "learning_rate": 4.6495e-05,
      "loss": 0.3961,
      "step": 9304
    },
    {
      "epoch": 0.5210549893605106,
      "grad_norm": 1.2484468221664429,
      "learning_rate": 4.6500000000000005e-05,
      "loss": 0.5452,
      "step": 9305
    },
    {
      "epoch": 0.5211109866726397,
      "grad_norm": 1.426182508468628,
      "learning_rate": 4.6505e-05,
      "loss": 0.4543,
      "step": 9306
    },
    {
      "epoch": 0.5211669839847687,
      "grad_norm": 1.3050537109375,
      "learning_rate": 4.651e-05,
      "loss": 0.3999,
      "step": 9307
    },
    {
      "epoch": 0.5212229812968977,
      "grad_norm": 1.513106346130371,
      "learning_rate": 4.6515000000000004e-05,
      "loss": 0.5753,
      "step": 9308
    },
    {
      "epoch": 0.5212789786090267,
      "grad_norm": 1.4483026266098022,
      "learning_rate": 4.652e-05,
      "loss": 0.4234,
      "step": 9309
    },
    {
      "epoch": 0.5213349759211557,
      "grad_norm": 1.6815348863601685,
      "learning_rate": 4.6525e-05,
      "loss": 0.5076,
      "step": 9310
    },
    {
      "epoch": 0.5213909732332848,
      "grad_norm": 3.2403481006622314,
      "learning_rate": 4.6530000000000003e-05,
      "loss": 0.4415,
      "step": 9311
    },
    {
      "epoch": 0.5214469705454138,
      "grad_norm": 0.9420170187950134,
      "learning_rate": 4.6535e-05,
      "loss": 0.4042,
      "step": 9312
    },
    {
      "epoch": 0.5215029678575428,
      "grad_norm": 1.3481045961380005,
      "learning_rate": 4.654e-05,
      "loss": 0.513,
      "step": 9313
    },
    {
      "epoch": 0.5215589651696718,
      "grad_norm": 1.4042633771896362,
      "learning_rate": 4.6545e-05,
      "loss": 0.4236,
      "step": 9314
    },
    {
      "epoch": 0.5216149624818008,
      "grad_norm": 1.3333860635757446,
      "learning_rate": 4.655000000000001e-05,
      "loss": 0.3404,
      "step": 9315
    },
    {
      "epoch": 0.5216709597939299,
      "grad_norm": 1.5369305610656738,
      "learning_rate": 4.6555000000000004e-05,
      "loss": 0.5177,
      "step": 9316
    },
    {
      "epoch": 0.5217269571060589,
      "grad_norm": 1.1305073499679565,
      "learning_rate": 4.656e-05,
      "loss": 0.3775,
      "step": 9317
    },
    {
      "epoch": 0.5217829544181879,
      "grad_norm": 1.3339060544967651,
      "learning_rate": 4.6565000000000006e-05,
      "loss": 0.4506,
      "step": 9318
    },
    {
      "epoch": 0.5218389517303169,
      "grad_norm": 1.486646294593811,
      "learning_rate": 4.657e-05,
      "loss": 0.4526,
      "step": 9319
    },
    {
      "epoch": 0.5218949490424459,
      "grad_norm": 2.0488927364349365,
      "learning_rate": 4.6575e-05,
      "loss": 0.3688,
      "step": 9320
    },
    {
      "epoch": 0.521950946354575,
      "grad_norm": 1.4488757848739624,
      "learning_rate": 4.6580000000000005e-05,
      "loss": 0.5925,
      "step": 9321
    },
    {
      "epoch": 0.522006943666704,
      "grad_norm": 1.2774953842163086,
      "learning_rate": 4.6585e-05,
      "loss": 0.4723,
      "step": 9322
    },
    {
      "epoch": 0.522062940978833,
      "grad_norm": 1.2316725254058838,
      "learning_rate": 4.659e-05,
      "loss": 0.4808,
      "step": 9323
    },
    {
      "epoch": 0.522118938290962,
      "grad_norm": 1.5660383701324463,
      "learning_rate": 4.6595e-05,
      "loss": 0.5014,
      "step": 9324
    },
    {
      "epoch": 0.522174935603091,
      "grad_norm": 1.5444111824035645,
      "learning_rate": 4.660000000000001e-05,
      "loss": 0.5312,
      "step": 9325
    },
    {
      "epoch": 0.52223093291522,
      "grad_norm": 1.4226216077804565,
      "learning_rate": 4.6605000000000005e-05,
      "loss": 0.5107,
      "step": 9326
    },
    {
      "epoch": 0.5222869302273491,
      "grad_norm": 1.1815792322158813,
      "learning_rate": 4.661e-05,
      "loss": 0.349,
      "step": 9327
    },
    {
      "epoch": 0.5223429275394781,
      "grad_norm": 1.321407437324524,
      "learning_rate": 4.6615e-05,
      "loss": 0.5086,
      "step": 9328
    },
    {
      "epoch": 0.5223989248516071,
      "grad_norm": 1.2084922790527344,
      "learning_rate": 4.6620000000000004e-05,
      "loss": 0.3924,
      "step": 9329
    },
    {
      "epoch": 0.5224549221637361,
      "grad_norm": 1.177322506904602,
      "learning_rate": 4.6625e-05,
      "loss": 0.3934,
      "step": 9330
    },
    {
      "epoch": 0.5225109194758651,
      "grad_norm": 1.176064133644104,
      "learning_rate": 4.663e-05,
      "loss": 0.4001,
      "step": 9331
    },
    {
      "epoch": 0.5225669167879942,
      "grad_norm": 1.2922840118408203,
      "learning_rate": 4.6635e-05,
      "loss": 0.4604,
      "step": 9332
    },
    {
      "epoch": 0.5226229141001232,
      "grad_norm": 1.0318429470062256,
      "learning_rate": 4.664e-05,
      "loss": 0.415,
      "step": 9333
    },
    {
      "epoch": 0.5226789114122522,
      "grad_norm": 1.2632776498794556,
      "learning_rate": 4.6645e-05,
      "loss": 0.4401,
      "step": 9334
    },
    {
      "epoch": 0.5227349087243812,
      "grad_norm": 1.361064076423645,
      "learning_rate": 4.665e-05,
      "loss": 0.4092,
      "step": 9335
    },
    {
      "epoch": 0.5227909060365102,
      "grad_norm": 1.4519658088684082,
      "learning_rate": 4.6655000000000006e-05,
      "loss": 0.6166,
      "step": 9336
    },
    {
      "epoch": 0.5228469033486393,
      "grad_norm": 1.2925876379013062,
      "learning_rate": 4.6660000000000004e-05,
      "loss": 0.4137,
      "step": 9337
    },
    {
      "epoch": 0.5229029006607683,
      "grad_norm": 1.1427985429763794,
      "learning_rate": 4.6665e-05,
      "loss": 0.3611,
      "step": 9338
    },
    {
      "epoch": 0.5229588979728973,
      "grad_norm": 1.4485900402069092,
      "learning_rate": 4.6670000000000005e-05,
      "loss": 0.3126,
      "step": 9339
    },
    {
      "epoch": 0.5230148952850263,
      "grad_norm": 1.188767671585083,
      "learning_rate": 4.6675e-05,
      "loss": 0.3825,
      "step": 9340
    },
    {
      "epoch": 0.5230708925971553,
      "grad_norm": 1.5355230569839478,
      "learning_rate": 4.668e-05,
      "loss": 0.3954,
      "step": 9341
    },
    {
      "epoch": 0.5231268899092844,
      "grad_norm": 1.3338741064071655,
      "learning_rate": 4.6685e-05,
      "loss": 0.4825,
      "step": 9342
    },
    {
      "epoch": 0.5231828872214134,
      "grad_norm": 1.626522183418274,
      "learning_rate": 4.669e-05,
      "loss": 0.5366,
      "step": 9343
    },
    {
      "epoch": 0.5232388845335424,
      "grad_norm": 1.5628392696380615,
      "learning_rate": 4.6695e-05,
      "loss": 0.5842,
      "step": 9344
    },
    {
      "epoch": 0.5232948818456714,
      "grad_norm": 1.1358919143676758,
      "learning_rate": 4.6700000000000003e-05,
      "loss": 0.4047,
      "step": 9345
    },
    {
      "epoch": 0.5233508791578004,
      "grad_norm": 1.2265276908874512,
      "learning_rate": 4.670500000000001e-05,
      "loss": 0.4204,
      "step": 9346
    },
    {
      "epoch": 0.5234068764699294,
      "grad_norm": 1.4317282438278198,
      "learning_rate": 4.6710000000000005e-05,
      "loss": 0.4173,
      "step": 9347
    },
    {
      "epoch": 0.5234628737820585,
      "grad_norm": 1.3782039880752563,
      "learning_rate": 4.6715e-05,
      "loss": 0.3343,
      "step": 9348
    },
    {
      "epoch": 0.5235188710941875,
      "grad_norm": 1.3642542362213135,
      "learning_rate": 4.672e-05,
      "loss": 0.4608,
      "step": 9349
    },
    {
      "epoch": 0.5235748684063165,
      "grad_norm": 1.1251015663146973,
      "learning_rate": 4.6725000000000004e-05,
      "loss": 0.3629,
      "step": 9350
    },
    {
      "epoch": 0.5236308657184455,
      "grad_norm": 1.2117772102355957,
      "learning_rate": 4.673e-05,
      "loss": 0.4681,
      "step": 9351
    },
    {
      "epoch": 0.5236868630305745,
      "grad_norm": 1.146168828010559,
      "learning_rate": 4.6735e-05,
      "loss": 0.4403,
      "step": 9352
    },
    {
      "epoch": 0.5237428603427036,
      "grad_norm": 1.1908000707626343,
      "learning_rate": 4.674e-05,
      "loss": 0.3886,
      "step": 9353
    },
    {
      "epoch": 0.5237988576548326,
      "grad_norm": 1.557708740234375,
      "learning_rate": 4.6745e-05,
      "loss": 0.513,
      "step": 9354
    },
    {
      "epoch": 0.5238548549669616,
      "grad_norm": 1.2665703296661377,
      "learning_rate": 4.6750000000000005e-05,
      "loss": 0.4626,
      "step": 9355
    },
    {
      "epoch": 0.5239108522790906,
      "grad_norm": 1.1932997703552246,
      "learning_rate": 4.6755e-05,
      "loss": 0.4521,
      "step": 9356
    },
    {
      "epoch": 0.5239668495912196,
      "grad_norm": 1.125240445137024,
      "learning_rate": 4.6760000000000006e-05,
      "loss": 0.3721,
      "step": 9357
    },
    {
      "epoch": 0.5240228469033487,
      "grad_norm": 1.2674050331115723,
      "learning_rate": 4.6765000000000004e-05,
      "loss": 0.4335,
      "step": 9358
    },
    {
      "epoch": 0.5240788442154777,
      "grad_norm": 1.3403898477554321,
      "learning_rate": 4.677e-05,
      "loss": 0.4734,
      "step": 9359
    },
    {
      "epoch": 0.5241348415276067,
      "grad_norm": 1.3714861869812012,
      "learning_rate": 4.6775000000000005e-05,
      "loss": 0.424,
      "step": 9360
    },
    {
      "epoch": 0.5241908388397357,
      "grad_norm": 1.10027277469635,
      "learning_rate": 4.678e-05,
      "loss": 0.3875,
      "step": 9361
    },
    {
      "epoch": 0.5242468361518647,
      "grad_norm": 1.2574408054351807,
      "learning_rate": 4.6785e-05,
      "loss": 0.6071,
      "step": 9362
    },
    {
      "epoch": 0.5243028334639938,
      "grad_norm": 1.5307801961898804,
      "learning_rate": 4.679e-05,
      "loss": 0.4204,
      "step": 9363
    },
    {
      "epoch": 0.5243588307761228,
      "grad_norm": 1.489932656288147,
      "learning_rate": 4.6795e-05,
      "loss": 0.5373,
      "step": 9364
    },
    {
      "epoch": 0.5244148280882518,
      "grad_norm": 1.3600836992263794,
      "learning_rate": 4.6800000000000006e-05,
      "loss": 0.4462,
      "step": 9365
    },
    {
      "epoch": 0.5244708254003808,
      "grad_norm": 1.307382345199585,
      "learning_rate": 4.6805e-05,
      "loss": 0.3492,
      "step": 9366
    },
    {
      "epoch": 0.5245268227125098,
      "grad_norm": 1.6123884916305542,
      "learning_rate": 4.681e-05,
      "loss": 0.5212,
      "step": 9367
    },
    {
      "epoch": 0.5245828200246389,
      "grad_norm": 1.2430503368377686,
      "learning_rate": 4.6815000000000005e-05,
      "loss": 0.4878,
      "step": 9368
    },
    {
      "epoch": 0.5246388173367679,
      "grad_norm": 1.4058754444122314,
      "learning_rate": 4.682e-05,
      "loss": 0.4161,
      "step": 9369
    },
    {
      "epoch": 0.5246948146488969,
      "grad_norm": 1.9148458242416382,
      "learning_rate": 4.6825e-05,
      "loss": 0.6418,
      "step": 9370
    },
    {
      "epoch": 0.5247508119610259,
      "grad_norm": 1.1850706338882446,
      "learning_rate": 4.6830000000000004e-05,
      "loss": 0.4079,
      "step": 9371
    },
    {
      "epoch": 0.5248068092731549,
      "grad_norm": 1.6921254396438599,
      "learning_rate": 4.6835e-05,
      "loss": 0.4794,
      "step": 9372
    },
    {
      "epoch": 0.524862806585284,
      "grad_norm": 1.3645633459091187,
      "learning_rate": 4.684e-05,
      "loss": 0.3625,
      "step": 9373
    },
    {
      "epoch": 0.524918803897413,
      "grad_norm": 1.3538650274276733,
      "learning_rate": 4.6845e-05,
      "loss": 0.4021,
      "step": 9374
    },
    {
      "epoch": 0.524974801209542,
      "grad_norm": 1.3050851821899414,
      "learning_rate": 4.685000000000001e-05,
      "loss": 0.5553,
      "step": 9375
    },
    {
      "epoch": 0.525030798521671,
      "grad_norm": 1.1158578395843506,
      "learning_rate": 4.6855000000000005e-05,
      "loss": 0.3695,
      "step": 9376
    },
    {
      "epoch": 0.5250867958338,
      "grad_norm": 1.3095901012420654,
      "learning_rate": 4.686e-05,
      "loss": 0.5141,
      "step": 9377
    },
    {
      "epoch": 0.525142793145929,
      "grad_norm": 1.19870126247406,
      "learning_rate": 4.6865000000000006e-05,
      "loss": 0.3386,
      "step": 9378
    },
    {
      "epoch": 0.5251987904580581,
      "grad_norm": 1.4308778047561646,
      "learning_rate": 4.6870000000000004e-05,
      "loss": 0.4569,
      "step": 9379
    },
    {
      "epoch": 0.5252547877701871,
      "grad_norm": 1.3367048501968384,
      "learning_rate": 4.6875e-05,
      "loss": 0.4415,
      "step": 9380
    },
    {
      "epoch": 0.5253107850823161,
      "grad_norm": 1.186528205871582,
      "learning_rate": 4.688e-05,
      "loss": 0.5143,
      "step": 9381
    },
    {
      "epoch": 0.5253667823944451,
      "grad_norm": 1.1099333763122559,
      "learning_rate": 4.6885e-05,
      "loss": 0.3806,
      "step": 9382
    },
    {
      "epoch": 0.5254227797065741,
      "grad_norm": 1.295485019683838,
      "learning_rate": 4.689e-05,
      "loss": 0.4786,
      "step": 9383
    },
    {
      "epoch": 0.525478777018703,
      "grad_norm": 1.3310902118682861,
      "learning_rate": 4.6895e-05,
      "loss": 0.4158,
      "step": 9384
    },
    {
      "epoch": 0.5255347743308321,
      "grad_norm": 1.4873652458190918,
      "learning_rate": 4.69e-05,
      "loss": 0.4764,
      "step": 9385
    },
    {
      "epoch": 0.5255907716429611,
      "grad_norm": 19.411251068115234,
      "learning_rate": 4.6905000000000006e-05,
      "loss": 0.5014,
      "step": 9386
    },
    {
      "epoch": 0.5256467689550901,
      "grad_norm": 1.3132100105285645,
      "learning_rate": 4.691e-05,
      "loss": 0.6241,
      "step": 9387
    },
    {
      "epoch": 0.5257027662672191,
      "grad_norm": 1.4633852243423462,
      "learning_rate": 4.6915e-05,
      "loss": 0.4219,
      "step": 9388
    },
    {
      "epoch": 0.5257587635793481,
      "grad_norm": 1.215549111366272,
      "learning_rate": 4.6920000000000005e-05,
      "loss": 0.3689,
      "step": 9389
    },
    {
      "epoch": 0.5258147608914772,
      "grad_norm": 1.663164496421814,
      "learning_rate": 4.6925e-05,
      "loss": 0.5895,
      "step": 9390
    },
    {
      "epoch": 0.5258707582036062,
      "grad_norm": 1.1354742050170898,
      "learning_rate": 4.693e-05,
      "loss": 0.4579,
      "step": 9391
    },
    {
      "epoch": 0.5259267555157352,
      "grad_norm": 1.1233223676681519,
      "learning_rate": 4.6935000000000004e-05,
      "loss": 0.5082,
      "step": 9392
    },
    {
      "epoch": 0.5259827528278642,
      "grad_norm": 1.3693325519561768,
      "learning_rate": 4.694e-05,
      "loss": 0.4704,
      "step": 9393
    },
    {
      "epoch": 0.5260387501399932,
      "grad_norm": 1.2200404405593872,
      "learning_rate": 4.6945e-05,
      "loss": 0.4775,
      "step": 9394
    },
    {
      "epoch": 0.5260947474521223,
      "grad_norm": 1.23910653591156,
      "learning_rate": 4.695e-05,
      "loss": 0.4767,
      "step": 9395
    },
    {
      "epoch": 0.5261507447642513,
      "grad_norm": 1.3311809301376343,
      "learning_rate": 4.695500000000001e-05,
      "loss": 0.2818,
      "step": 9396
    },
    {
      "epoch": 0.5262067420763803,
      "grad_norm": 1.1943596601486206,
      "learning_rate": 4.6960000000000004e-05,
      "loss": 0.4084,
      "step": 9397
    },
    {
      "epoch": 0.5262627393885093,
      "grad_norm": 1.1603953838348389,
      "learning_rate": 4.6965e-05,
      "loss": 0.3852,
      "step": 9398
    },
    {
      "epoch": 0.5263187367006383,
      "grad_norm": 1.0466657876968384,
      "learning_rate": 4.6970000000000006e-05,
      "loss": 0.4014,
      "step": 9399
    },
    {
      "epoch": 0.5263747340127674,
      "grad_norm": 1.563252568244934,
      "learning_rate": 4.6975000000000003e-05,
      "loss": 0.4187,
      "step": 9400
    },
    {
      "epoch": 0.5264307313248964,
      "grad_norm": 1.361717700958252,
      "learning_rate": 4.698e-05,
      "loss": 0.454,
      "step": 9401
    },
    {
      "epoch": 0.5264867286370254,
      "grad_norm": 1.0306333303451538,
      "learning_rate": 4.6985e-05,
      "loss": 0.3177,
      "step": 9402
    },
    {
      "epoch": 0.5265427259491544,
      "grad_norm": 1.068896770477295,
      "learning_rate": 4.699e-05,
      "loss": 0.3457,
      "step": 9403
    },
    {
      "epoch": 0.5265987232612834,
      "grad_norm": 1.3894723653793335,
      "learning_rate": 4.6995e-05,
      "loss": 0.5277,
      "step": 9404
    },
    {
      "epoch": 0.5266547205734124,
      "grad_norm": 1.188368320465088,
      "learning_rate": 4.7e-05,
      "loss": 0.3323,
      "step": 9405
    },
    {
      "epoch": 0.5267107178855415,
      "grad_norm": 1.2241084575653076,
      "learning_rate": 4.7005e-05,
      "loss": 0.4512,
      "step": 9406
    },
    {
      "epoch": 0.5267667151976705,
      "grad_norm": 1.2160418033599854,
      "learning_rate": 4.7010000000000006e-05,
      "loss": 0.4451,
      "step": 9407
    },
    {
      "epoch": 0.5268227125097995,
      "grad_norm": 1.332726001739502,
      "learning_rate": 4.7015e-05,
      "loss": 0.4397,
      "step": 9408
    },
    {
      "epoch": 0.5268787098219285,
      "grad_norm": 1.3292813301086426,
      "learning_rate": 4.702e-05,
      "loss": 0.5394,
      "step": 9409
    },
    {
      "epoch": 0.5269347071340575,
      "grad_norm": 1.5254113674163818,
      "learning_rate": 4.7025000000000005e-05,
      "loss": 0.4445,
      "step": 9410
    },
    {
      "epoch": 0.5269907044461866,
      "grad_norm": 1.443175196647644,
      "learning_rate": 4.703e-05,
      "loss": 0.4907,
      "step": 9411
    },
    {
      "epoch": 0.5270467017583156,
      "grad_norm": 1.3154882192611694,
      "learning_rate": 4.7035e-05,
      "loss": 0.3779,
      "step": 9412
    },
    {
      "epoch": 0.5271026990704446,
      "grad_norm": 1.2727384567260742,
      "learning_rate": 4.7040000000000004e-05,
      "loss": 0.4134,
      "step": 9413
    },
    {
      "epoch": 0.5271586963825736,
      "grad_norm": 1.4846564531326294,
      "learning_rate": 4.7045e-05,
      "loss": 0.4969,
      "step": 9414
    },
    {
      "epoch": 0.5272146936947026,
      "grad_norm": 1.115434169769287,
      "learning_rate": 4.705e-05,
      "loss": 0.2774,
      "step": 9415
    },
    {
      "epoch": 0.5272706910068317,
      "grad_norm": 1.1634632349014282,
      "learning_rate": 4.7055e-05,
      "loss": 0.3853,
      "step": 9416
    },
    {
      "epoch": 0.5273266883189607,
      "grad_norm": 1.208876132965088,
      "learning_rate": 4.706000000000001e-05,
      "loss": 0.5363,
      "step": 9417
    },
    {
      "epoch": 0.5273826856310897,
      "grad_norm": 1.4385795593261719,
      "learning_rate": 4.7065000000000004e-05,
      "loss": 0.4161,
      "step": 9418
    },
    {
      "epoch": 0.5274386829432187,
      "grad_norm": 1.210424542427063,
      "learning_rate": 4.707e-05,
      "loss": 0.4713,
      "step": 9419
    },
    {
      "epoch": 0.5274946802553477,
      "grad_norm": 1.1139848232269287,
      "learning_rate": 4.7075e-05,
      "loss": 0.451,
      "step": 9420
    },
    {
      "epoch": 0.5275506775674768,
      "grad_norm": 1.3247337341308594,
      "learning_rate": 4.708e-05,
      "loss": 0.4168,
      "step": 9421
    },
    {
      "epoch": 0.5276066748796058,
      "grad_norm": 1.511054515838623,
      "learning_rate": 4.7085e-05,
      "loss": 0.4623,
      "step": 9422
    },
    {
      "epoch": 0.5276626721917348,
      "grad_norm": 2.5482265949249268,
      "learning_rate": 4.709e-05,
      "loss": 0.5358,
      "step": 9423
    },
    {
      "epoch": 0.5277186695038638,
      "grad_norm": 1.2414195537567139,
      "learning_rate": 4.7095e-05,
      "loss": 0.3446,
      "step": 9424
    },
    {
      "epoch": 0.5277746668159928,
      "grad_norm": 1.1679651737213135,
      "learning_rate": 4.71e-05,
      "loss": 0.4133,
      "step": 9425
    },
    {
      "epoch": 0.5278306641281219,
      "grad_norm": 1.3512089252471924,
      "learning_rate": 4.7105000000000004e-05,
      "loss": 0.5281,
      "step": 9426
    },
    {
      "epoch": 0.5278866614402509,
      "grad_norm": 1.1578283309936523,
      "learning_rate": 4.711e-05,
      "loss": 0.3743,
      "step": 9427
    },
    {
      "epoch": 0.5279426587523799,
      "grad_norm": 1.0728938579559326,
      "learning_rate": 4.7115000000000005e-05,
      "loss": 0.3666,
      "step": 9428
    },
    {
      "epoch": 0.5279986560645089,
      "grad_norm": 1.27793288230896,
      "learning_rate": 4.712e-05,
      "loss": 0.4113,
      "step": 9429
    },
    {
      "epoch": 0.5280546533766379,
      "grad_norm": 1.2008014917373657,
      "learning_rate": 4.7125e-05,
      "loss": 0.458,
      "step": 9430
    },
    {
      "epoch": 0.528110650688767,
      "grad_norm": 1.2136887311935425,
      "learning_rate": 4.7130000000000004e-05,
      "loss": 0.382,
      "step": 9431
    },
    {
      "epoch": 0.528166648000896,
      "grad_norm": 1.2503955364227295,
      "learning_rate": 4.7135e-05,
      "loss": 0.4685,
      "step": 9432
    },
    {
      "epoch": 0.528222645313025,
      "grad_norm": 1.1596254110336304,
      "learning_rate": 4.714e-05,
      "loss": 0.4247,
      "step": 9433
    },
    {
      "epoch": 0.528278642625154,
      "grad_norm": 1.1022257804870605,
      "learning_rate": 4.7145000000000003e-05,
      "loss": 0.379,
      "step": 9434
    },
    {
      "epoch": 0.528334639937283,
      "grad_norm": 1.4563370943069458,
      "learning_rate": 4.715e-05,
      "loss": 0.4772,
      "step": 9435
    },
    {
      "epoch": 0.528390637249412,
      "grad_norm": 1.2413662672042847,
      "learning_rate": 4.7155000000000005e-05,
      "loss": 0.3936,
      "step": 9436
    },
    {
      "epoch": 0.5284466345615411,
      "grad_norm": 1.4025589227676392,
      "learning_rate": 4.716e-05,
      "loss": 0.4219,
      "step": 9437
    },
    {
      "epoch": 0.5285026318736701,
      "grad_norm": 1.4004168510437012,
      "learning_rate": 4.716500000000001e-05,
      "loss": 0.4298,
      "step": 9438
    },
    {
      "epoch": 0.5285586291857991,
      "grad_norm": 1.2061214447021484,
      "learning_rate": 4.7170000000000004e-05,
      "loss": 0.4344,
      "step": 9439
    },
    {
      "epoch": 0.5286146264979281,
      "grad_norm": 1.204013705253601,
      "learning_rate": 4.7175e-05,
      "loss": 0.389,
      "step": 9440
    },
    {
      "epoch": 0.5286706238100571,
      "grad_norm": 1.3896828889846802,
      "learning_rate": 4.718e-05,
      "loss": 0.4706,
      "step": 9441
    },
    {
      "epoch": 0.5287266211221862,
      "grad_norm": 1.1460342407226562,
      "learning_rate": 4.7185e-05,
      "loss": 0.5139,
      "step": 9442
    },
    {
      "epoch": 0.5287826184343152,
      "grad_norm": 1.0229713916778564,
      "learning_rate": 4.719e-05,
      "loss": 0.353,
      "step": 9443
    },
    {
      "epoch": 0.5288386157464442,
      "grad_norm": 1.3112131357192993,
      "learning_rate": 4.7195e-05,
      "loss": 0.6277,
      "step": 9444
    },
    {
      "epoch": 0.5288946130585732,
      "grad_norm": 1.111806869506836,
      "learning_rate": 4.72e-05,
      "loss": 0.3495,
      "step": 9445
    },
    {
      "epoch": 0.5289506103707022,
      "grad_norm": 1.380745768547058,
      "learning_rate": 4.7205000000000006e-05,
      "loss": 0.443,
      "step": 9446
    },
    {
      "epoch": 0.5290066076828313,
      "grad_norm": 1.5349440574645996,
      "learning_rate": 4.7210000000000004e-05,
      "loss": 0.7024,
      "step": 9447
    },
    {
      "epoch": 0.5290626049949603,
      "grad_norm": 1.9456593990325928,
      "learning_rate": 4.7215e-05,
      "loss": 0.6035,
      "step": 9448
    },
    {
      "epoch": 0.5291186023070893,
      "grad_norm": 1.6643338203430176,
      "learning_rate": 4.7220000000000005e-05,
      "loss": 0.6041,
      "step": 9449
    },
    {
      "epoch": 0.5291745996192183,
      "grad_norm": 1.0799925327301025,
      "learning_rate": 4.7225e-05,
      "loss": 0.4175,
      "step": 9450
    },
    {
      "epoch": 0.5292305969313473,
      "grad_norm": 1.3846999406814575,
      "learning_rate": 4.723e-05,
      "loss": 0.4605,
      "step": 9451
    },
    {
      "epoch": 0.5292865942434763,
      "grad_norm": 30.342243194580078,
      "learning_rate": 4.7235000000000004e-05,
      "loss": 0.4169,
      "step": 9452
    },
    {
      "epoch": 0.5293425915556054,
      "grad_norm": 1.0062071084976196,
      "learning_rate": 4.724e-05,
      "loss": 0.286,
      "step": 9453
    },
    {
      "epoch": 0.5293985888677344,
      "grad_norm": 1.2299600839614868,
      "learning_rate": 4.7245e-05,
      "loss": 0.6064,
      "step": 9454
    },
    {
      "epoch": 0.5294545861798634,
      "grad_norm": 1.2189158201217651,
      "learning_rate": 4.7249999999999997e-05,
      "loss": 0.4084,
      "step": 9455
    },
    {
      "epoch": 0.5295105834919924,
      "grad_norm": 1.4640798568725586,
      "learning_rate": 4.725500000000001e-05,
      "loss": 0.4591,
      "step": 9456
    },
    {
      "epoch": 0.5295665808041214,
      "grad_norm": 1.0949474573135376,
      "learning_rate": 4.7260000000000005e-05,
      "loss": 0.3065,
      "step": 9457
    },
    {
      "epoch": 0.5296225781162505,
      "grad_norm": 1.0800518989562988,
      "learning_rate": 4.7265e-05,
      "loss": 0.3289,
      "step": 9458
    },
    {
      "epoch": 0.5296785754283795,
      "grad_norm": 1.3044289350509644,
      "learning_rate": 4.7270000000000007e-05,
      "loss": 0.4441,
      "step": 9459
    },
    {
      "epoch": 0.5297345727405085,
      "grad_norm": 1.2596040964126587,
      "learning_rate": 4.7275000000000004e-05,
      "loss": 0.5076,
      "step": 9460
    },
    {
      "epoch": 0.5297905700526375,
      "grad_norm": 1.1207771301269531,
      "learning_rate": 4.728e-05,
      "loss": 0.4076,
      "step": 9461
    },
    {
      "epoch": 0.5298465673647665,
      "grad_norm": 1.3882770538330078,
      "learning_rate": 4.7285e-05,
      "loss": 0.4243,
      "step": 9462
    },
    {
      "epoch": 0.5299025646768956,
      "grad_norm": 2.043513298034668,
      "learning_rate": 4.729e-05,
      "loss": 0.5781,
      "step": 9463
    },
    {
      "epoch": 0.5299585619890246,
      "grad_norm": 1.4725967645645142,
      "learning_rate": 4.7295e-05,
      "loss": 0.6341,
      "step": 9464
    },
    {
      "epoch": 0.5300145593011536,
      "grad_norm": 1.2623803615570068,
      "learning_rate": 4.73e-05,
      "loss": 0.4125,
      "step": 9465
    },
    {
      "epoch": 0.5300705566132826,
      "grad_norm": 1.5083774328231812,
      "learning_rate": 4.7305e-05,
      "loss": 0.5131,
      "step": 9466
    },
    {
      "epoch": 0.5301265539254115,
      "grad_norm": 1.3852721452713013,
      "learning_rate": 4.7310000000000006e-05,
      "loss": 0.4347,
      "step": 9467
    },
    {
      "epoch": 0.5301825512375405,
      "grad_norm": 1.2226790189743042,
      "learning_rate": 4.7315000000000004e-05,
      "loss": 0.4741,
      "step": 9468
    },
    {
      "epoch": 0.5302385485496696,
      "grad_norm": 1.1415860652923584,
      "learning_rate": 4.732e-05,
      "loss": 0.4221,
      "step": 9469
    },
    {
      "epoch": 0.5302945458617986,
      "grad_norm": 1.0505965948104858,
      "learning_rate": 4.7325000000000005e-05,
      "loss": 0.3267,
      "step": 9470
    },
    {
      "epoch": 0.5303505431739276,
      "grad_norm": 1.618748664855957,
      "learning_rate": 4.733e-05,
      "loss": 0.4949,
      "step": 9471
    },
    {
      "epoch": 0.5304065404860566,
      "grad_norm": 1.3280977010726929,
      "learning_rate": 4.7335e-05,
      "loss": 0.3981,
      "step": 9472
    },
    {
      "epoch": 0.5304625377981856,
      "grad_norm": 1.3729336261749268,
      "learning_rate": 4.7340000000000004e-05,
      "loss": 0.4779,
      "step": 9473
    },
    {
      "epoch": 0.5305185351103147,
      "grad_norm": 1.0747077465057373,
      "learning_rate": 4.7345e-05,
      "loss": 0.4052,
      "step": 9474
    },
    {
      "epoch": 0.5305745324224437,
      "grad_norm": 1.391895055770874,
      "learning_rate": 4.735e-05,
      "loss": 0.4142,
      "step": 9475
    },
    {
      "epoch": 0.5306305297345727,
      "grad_norm": 1.2232345342636108,
      "learning_rate": 4.7355e-05,
      "loss": 0.382,
      "step": 9476
    },
    {
      "epoch": 0.5306865270467017,
      "grad_norm": 1.3728983402252197,
      "learning_rate": 4.736000000000001e-05,
      "loss": 0.6502,
      "step": 9477
    },
    {
      "epoch": 0.5307425243588307,
      "grad_norm": 1.1662178039550781,
      "learning_rate": 4.7365000000000005e-05,
      "loss": 0.4822,
      "step": 9478
    },
    {
      "epoch": 0.5307985216709598,
      "grad_norm": 1.171358346939087,
      "learning_rate": 4.737e-05,
      "loss": 0.4309,
      "step": 9479
    },
    {
      "epoch": 0.5308545189830888,
      "grad_norm": 5.396470069885254,
      "learning_rate": 4.7375e-05,
      "loss": 0.5371,
      "step": 9480
    },
    {
      "epoch": 0.5309105162952178,
      "grad_norm": 1.3976140022277832,
      "learning_rate": 4.7380000000000004e-05,
      "loss": 0.3117,
      "step": 9481
    },
    {
      "epoch": 0.5309665136073468,
      "grad_norm": 1.4935336112976074,
      "learning_rate": 4.7385e-05,
      "loss": 0.4498,
      "step": 9482
    },
    {
      "epoch": 0.5310225109194758,
      "grad_norm": 1.3390616178512573,
      "learning_rate": 4.739e-05,
      "loss": 0.4706,
      "step": 9483
    },
    {
      "epoch": 0.5310785082316049,
      "grad_norm": 1.7145739793777466,
      "learning_rate": 4.7395e-05,
      "loss": 0.4191,
      "step": 9484
    },
    {
      "epoch": 0.5311345055437339,
      "grad_norm": 1.1888912916183472,
      "learning_rate": 4.74e-05,
      "loss": 0.3363,
      "step": 9485
    },
    {
      "epoch": 0.5311905028558629,
      "grad_norm": 1.2296078205108643,
      "learning_rate": 4.7405000000000004e-05,
      "loss": 0.3789,
      "step": 9486
    },
    {
      "epoch": 0.5312465001679919,
      "grad_norm": 1.4090124368667603,
      "learning_rate": 4.741e-05,
      "loss": 0.5327,
      "step": 9487
    },
    {
      "epoch": 0.5313024974801209,
      "grad_norm": 1.6135203838348389,
      "learning_rate": 4.7415000000000006e-05,
      "loss": 0.5505,
      "step": 9488
    },
    {
      "epoch": 0.53135849479225,
      "grad_norm": 1.1662812232971191,
      "learning_rate": 4.742e-05,
      "loss": 0.3032,
      "step": 9489
    },
    {
      "epoch": 0.531414492104379,
      "grad_norm": 1.2016109228134155,
      "learning_rate": 4.7425e-05,
      "loss": 0.3916,
      "step": 9490
    },
    {
      "epoch": 0.531470489416508,
      "grad_norm": 1.1974090337753296,
      "learning_rate": 4.7430000000000005e-05,
      "loss": 0.4503,
      "step": 9491
    },
    {
      "epoch": 0.531526486728637,
      "grad_norm": 1.194698691368103,
      "learning_rate": 4.7435e-05,
      "loss": 0.3947,
      "step": 9492
    },
    {
      "epoch": 0.531582484040766,
      "grad_norm": 1.080213189125061,
      "learning_rate": 4.744e-05,
      "loss": 0.4138,
      "step": 9493
    },
    {
      "epoch": 0.531638481352895,
      "grad_norm": 1.5487487316131592,
      "learning_rate": 4.7445e-05,
      "loss": 0.398,
      "step": 9494
    },
    {
      "epoch": 0.5316944786650241,
      "grad_norm": 1.090031623840332,
      "learning_rate": 4.745e-05,
      "loss": 0.3756,
      "step": 9495
    },
    {
      "epoch": 0.5317504759771531,
      "grad_norm": 1.1094419956207275,
      "learning_rate": 4.7455000000000006e-05,
      "loss": 0.4849,
      "step": 9496
    },
    {
      "epoch": 0.5318064732892821,
      "grad_norm": 1.4495092630386353,
      "learning_rate": 4.746e-05,
      "loss": 0.3952,
      "step": 9497
    },
    {
      "epoch": 0.5318624706014111,
      "grad_norm": 1.535077452659607,
      "learning_rate": 4.746500000000001e-05,
      "loss": 0.429,
      "step": 9498
    },
    {
      "epoch": 0.5319184679135401,
      "grad_norm": 1.8276748657226562,
      "learning_rate": 4.7470000000000005e-05,
      "loss": 0.4638,
      "step": 9499
    },
    {
      "epoch": 0.5319744652256692,
      "grad_norm": 1.2453430891036987,
      "learning_rate": 4.7475e-05,
      "loss": 0.3779,
      "step": 9500
    },
    {
      "epoch": 0.5320304625377982,
      "grad_norm": 1.2442799806594849,
      "learning_rate": 4.748e-05,
      "loss": 0.6226,
      "step": 9501
    },
    {
      "epoch": 0.5320864598499272,
      "grad_norm": 1.3208845853805542,
      "learning_rate": 4.7485000000000004e-05,
      "loss": 0.3726,
      "step": 9502
    },
    {
      "epoch": 0.5321424571620562,
      "grad_norm": 1.1866168975830078,
      "learning_rate": 4.749e-05,
      "loss": 0.3711,
      "step": 9503
    },
    {
      "epoch": 0.5321984544741852,
      "grad_norm": 1.3956489562988281,
      "learning_rate": 4.7495e-05,
      "loss": 0.5071,
      "step": 9504
    },
    {
      "epoch": 0.5322544517863143,
      "grad_norm": 1.4432622194290161,
      "learning_rate": 4.75e-05,
      "loss": 0.5588,
      "step": 9505
    },
    {
      "epoch": 0.5323104490984433,
      "grad_norm": 1.4026999473571777,
      "learning_rate": 4.7505e-05,
      "loss": 0.4134,
      "step": 9506
    },
    {
      "epoch": 0.5323664464105723,
      "grad_norm": 1.4475175142288208,
      "learning_rate": 4.7510000000000004e-05,
      "loss": 0.4246,
      "step": 9507
    },
    {
      "epoch": 0.5324224437227013,
      "grad_norm": 1.3349000215530396,
      "learning_rate": 4.7515e-05,
      "loss": 0.2891,
      "step": 9508
    },
    {
      "epoch": 0.5324784410348303,
      "grad_norm": 1.7620033025741577,
      "learning_rate": 4.7520000000000006e-05,
      "loss": 0.4458,
      "step": 9509
    },
    {
      "epoch": 0.5325344383469593,
      "grad_norm": 1.14523184299469,
      "learning_rate": 4.7525e-05,
      "loss": 0.3744,
      "step": 9510
    },
    {
      "epoch": 0.5325904356590884,
      "grad_norm": 1.699660301208496,
      "learning_rate": 4.753e-05,
      "loss": 0.5277,
      "step": 9511
    },
    {
      "epoch": 0.5326464329712174,
      "grad_norm": 1.4605993032455444,
      "learning_rate": 4.7535000000000005e-05,
      "loss": 0.4717,
      "step": 9512
    },
    {
      "epoch": 0.5327024302833464,
      "grad_norm": 1.1820509433746338,
      "learning_rate": 4.754e-05,
      "loss": 0.4485,
      "step": 9513
    },
    {
      "epoch": 0.5327584275954754,
      "grad_norm": 1.1635390520095825,
      "learning_rate": 4.7545e-05,
      "loss": 0.3693,
      "step": 9514
    },
    {
      "epoch": 0.5328144249076044,
      "grad_norm": 1.3283125162124634,
      "learning_rate": 4.755e-05,
      "loss": 0.3971,
      "step": 9515
    },
    {
      "epoch": 0.5328704222197335,
      "grad_norm": 1.3234190940856934,
      "learning_rate": 4.7555e-05,
      "loss": 0.5261,
      "step": 9516
    },
    {
      "epoch": 0.5329264195318625,
      "grad_norm": 1.2321648597717285,
      "learning_rate": 4.7560000000000005e-05,
      "loss": 0.5013,
      "step": 9517
    },
    {
      "epoch": 0.5329824168439915,
      "grad_norm": 1.5043165683746338,
      "learning_rate": 4.7565e-05,
      "loss": 0.3885,
      "step": 9518
    },
    {
      "epoch": 0.5330384141561205,
      "grad_norm": 1.2735203504562378,
      "learning_rate": 4.757e-05,
      "loss": 0.5187,
      "step": 9519
    },
    {
      "epoch": 0.5330944114682495,
      "grad_norm": 1.0651447772979736,
      "learning_rate": 4.7575000000000004e-05,
      "loss": 0.3281,
      "step": 9520
    },
    {
      "epoch": 0.5331504087803786,
      "grad_norm": 1.2711256742477417,
      "learning_rate": 4.758e-05,
      "loss": 0.365,
      "step": 9521
    },
    {
      "epoch": 0.5332064060925076,
      "grad_norm": 1.3065346479415894,
      "learning_rate": 4.7585e-05,
      "loss": 0.5437,
      "step": 9522
    },
    {
      "epoch": 0.5332624034046366,
      "grad_norm": 1.5553252696990967,
      "learning_rate": 4.7590000000000003e-05,
      "loss": 0.5435,
      "step": 9523
    },
    {
      "epoch": 0.5333184007167656,
      "grad_norm": 1.0811947584152222,
      "learning_rate": 4.7595e-05,
      "loss": 0.3668,
      "step": 9524
    },
    {
      "epoch": 0.5333743980288946,
      "grad_norm": 1.2615207433700562,
      "learning_rate": 4.76e-05,
      "loss": 0.4445,
      "step": 9525
    },
    {
      "epoch": 0.5334303953410237,
      "grad_norm": 1.282523512840271,
      "learning_rate": 4.7605e-05,
      "loss": 0.3629,
      "step": 9526
    },
    {
      "epoch": 0.5334863926531527,
      "grad_norm": 1.3136628866195679,
      "learning_rate": 4.761000000000001e-05,
      "loss": 0.4487,
      "step": 9527
    },
    {
      "epoch": 0.5335423899652817,
      "grad_norm": 1.2010968923568726,
      "learning_rate": 4.7615000000000004e-05,
      "loss": 0.3197,
      "step": 9528
    },
    {
      "epoch": 0.5335983872774107,
      "grad_norm": 1.1536890268325806,
      "learning_rate": 4.762e-05,
      "loss": 0.4651,
      "step": 9529
    },
    {
      "epoch": 0.5336543845895397,
      "grad_norm": 1.1152513027191162,
      "learning_rate": 4.7625000000000006e-05,
      "loss": 0.4346,
      "step": 9530
    },
    {
      "epoch": 0.5337103819016688,
      "grad_norm": 1.399315595626831,
      "learning_rate": 4.763e-05,
      "loss": 0.4695,
      "step": 9531
    },
    {
      "epoch": 0.5337663792137978,
      "grad_norm": 1.2495238780975342,
      "learning_rate": 4.7635e-05,
      "loss": 0.352,
      "step": 9532
    },
    {
      "epoch": 0.5338223765259268,
      "grad_norm": 1.1976304054260254,
      "learning_rate": 4.7640000000000005e-05,
      "loss": 0.3945,
      "step": 9533
    },
    {
      "epoch": 0.5338783738380558,
      "grad_norm": 1.330531120300293,
      "learning_rate": 4.7645e-05,
      "loss": 0.4243,
      "step": 9534
    },
    {
      "epoch": 0.5339343711501848,
      "grad_norm": 1.3263925313949585,
      "learning_rate": 4.765e-05,
      "loss": 0.3887,
      "step": 9535
    },
    {
      "epoch": 0.5339903684623138,
      "grad_norm": 1.2777055501937866,
      "learning_rate": 4.7655e-05,
      "loss": 0.3793,
      "step": 9536
    },
    {
      "epoch": 0.5340463657744429,
      "grad_norm": 1.4256744384765625,
      "learning_rate": 4.766000000000001e-05,
      "loss": 0.5506,
      "step": 9537
    },
    {
      "epoch": 0.5341023630865719,
      "grad_norm": 1.3626021146774292,
      "learning_rate": 4.7665000000000005e-05,
      "loss": 0.4123,
      "step": 9538
    },
    {
      "epoch": 0.5341583603987009,
      "grad_norm": 1.2359644174575806,
      "learning_rate": 4.767e-05,
      "loss": 0.4575,
      "step": 9539
    },
    {
      "epoch": 0.5342143577108299,
      "grad_norm": 1.4144980907440186,
      "learning_rate": 4.7675e-05,
      "loss": 0.4501,
      "step": 9540
    },
    {
      "epoch": 0.5342703550229589,
      "grad_norm": 1.4422568082809448,
      "learning_rate": 4.7680000000000004e-05,
      "loss": 0.581,
      "step": 9541
    },
    {
      "epoch": 0.534326352335088,
      "grad_norm": Infinity,
      "learning_rate": 4.7680000000000004e-05,
      "loss": 0.4907,
      "step": 9542
    },
    {
      "epoch": 0.534382349647217,
      "grad_norm": 1.2412611246109009,
      "learning_rate": 4.7685e-05,
      "loss": 0.4355,
      "step": 9543
    },
    {
      "epoch": 0.534438346959346,
      "grad_norm": 1.3127073049545288,
      "learning_rate": 4.769e-05,
      "loss": 0.4543,
      "step": 9544
    },
    {
      "epoch": 0.534494344271475,
      "grad_norm": 1.9356169700622559,
      "learning_rate": 4.7695e-05,
      "loss": 0.5243,
      "step": 9545
    },
    {
      "epoch": 0.534550341583604,
      "grad_norm": 1.1254539489746094,
      "learning_rate": 4.77e-05,
      "loss": 0.4137,
      "step": 9546
    },
    {
      "epoch": 0.5346063388957331,
      "grad_norm": 1.1302624940872192,
      "learning_rate": 4.7705e-05,
      "loss": 0.4327,
      "step": 9547
    },
    {
      "epoch": 0.5346623362078621,
      "grad_norm": 1.411635160446167,
      "learning_rate": 4.771e-05,
      "loss": 0.5023,
      "step": 9548
    },
    {
      "epoch": 0.5347183335199911,
      "grad_norm": 1.284804344177246,
      "learning_rate": 4.7715000000000006e-05,
      "loss": 0.4215,
      "step": 9549
    },
    {
      "epoch": 0.53477433083212,
      "grad_norm": 1.2880300283432007,
      "learning_rate": 4.7720000000000004e-05,
      "loss": 0.4165,
      "step": 9550
    },
    {
      "epoch": 0.534830328144249,
      "grad_norm": 1.2266498804092407,
      "learning_rate": 4.7725e-05,
      "loss": 0.3902,
      "step": 9551
    },
    {
      "epoch": 0.534886325456378,
      "grad_norm": 1.2976826429367065,
      "learning_rate": 4.7730000000000005e-05,
      "loss": 0.4641,
      "step": 9552
    },
    {
      "epoch": 0.5349423227685071,
      "grad_norm": 1.2469403743743896,
      "learning_rate": 4.7735e-05,
      "loss": 0.3762,
      "step": 9553
    },
    {
      "epoch": 0.5349983200806361,
      "grad_norm": 1.1859091520309448,
      "learning_rate": 4.774e-05,
      "loss": 0.4301,
      "step": 9554
    },
    {
      "epoch": 0.5350543173927651,
      "grad_norm": 1.3102047443389893,
      "learning_rate": 4.7745e-05,
      "loss": 0.3257,
      "step": 9555
    },
    {
      "epoch": 0.5351103147048941,
      "grad_norm": 1.27182936668396,
      "learning_rate": 4.775e-05,
      "loss": 0.3905,
      "step": 9556
    },
    {
      "epoch": 0.5351663120170231,
      "grad_norm": 1.1418534517288208,
      "learning_rate": 4.7755e-05,
      "loss": 0.4676,
      "step": 9557
    },
    {
      "epoch": 0.5352223093291522,
      "grad_norm": 1.368029236793518,
      "learning_rate": 4.7760000000000004e-05,
      "loss": 0.4567,
      "step": 9558
    },
    {
      "epoch": 0.5352783066412812,
      "grad_norm": 1.408414602279663,
      "learning_rate": 4.7765e-05,
      "loss": 0.4692,
      "step": 9559
    },
    {
      "epoch": 0.5353343039534102,
      "grad_norm": 1.2978746891021729,
      "learning_rate": 4.7770000000000005e-05,
      "loss": 0.418,
      "step": 9560
    },
    {
      "epoch": 0.5353903012655392,
      "grad_norm": 1.1471731662750244,
      "learning_rate": 4.7775e-05,
      "loss": 0.4687,
      "step": 9561
    },
    {
      "epoch": 0.5354462985776682,
      "grad_norm": 1.217063307762146,
      "learning_rate": 4.778e-05,
      "loss": 0.3974,
      "step": 9562
    },
    {
      "epoch": 0.5355022958897973,
      "grad_norm": 1.1473416090011597,
      "learning_rate": 4.7785000000000004e-05,
      "loss": 0.4221,
      "step": 9563
    },
    {
      "epoch": 0.5355582932019263,
      "grad_norm": 1.3304030895233154,
      "learning_rate": 4.779e-05,
      "loss": 0.4207,
      "step": 9564
    },
    {
      "epoch": 0.5356142905140553,
      "grad_norm": 1.2993205785751343,
      "learning_rate": 4.7795e-05,
      "loss": 0.3556,
      "step": 9565
    },
    {
      "epoch": 0.5356702878261843,
      "grad_norm": 1.4037895202636719,
      "learning_rate": 4.78e-05,
      "loss": 0.456,
      "step": 9566
    },
    {
      "epoch": 0.5357262851383133,
      "grad_norm": 1.8493205308914185,
      "learning_rate": 4.7805e-05,
      "loss": 0.3586,
      "step": 9567
    },
    {
      "epoch": 0.5357822824504423,
      "grad_norm": 1.2599310874938965,
      "learning_rate": 4.7810000000000005e-05,
      "loss": 0.4284,
      "step": 9568
    },
    {
      "epoch": 0.5358382797625714,
      "grad_norm": 1.3864649534225464,
      "learning_rate": 4.7815e-05,
      "loss": 0.4289,
      "step": 9569
    },
    {
      "epoch": 0.5358942770747004,
      "grad_norm": 1.3962305784225464,
      "learning_rate": 4.7820000000000006e-05,
      "loss": 0.4402,
      "step": 9570
    },
    {
      "epoch": 0.5359502743868294,
      "grad_norm": 1.3587095737457275,
      "learning_rate": 4.7825000000000004e-05,
      "loss": 0.6073,
      "step": 9571
    },
    {
      "epoch": 0.5360062716989584,
      "grad_norm": 1.2020891904830933,
      "learning_rate": 4.783e-05,
      "loss": 0.3073,
      "step": 9572
    },
    {
      "epoch": 0.5360622690110874,
      "grad_norm": 1.3456652164459229,
      "learning_rate": 4.7835000000000005e-05,
      "loss": 0.5029,
      "step": 9573
    },
    {
      "epoch": 0.5361182663232165,
      "grad_norm": 1.320943832397461,
      "learning_rate": 4.784e-05,
      "loss": 0.5947,
      "step": 9574
    },
    {
      "epoch": 0.5361742636353455,
      "grad_norm": 1.4631785154342651,
      "learning_rate": 4.7845e-05,
      "loss": 0.388,
      "step": 9575
    },
    {
      "epoch": 0.5362302609474745,
      "grad_norm": 1.2012794017791748,
      "learning_rate": 4.785e-05,
      "loss": 0.4939,
      "step": 9576
    },
    {
      "epoch": 0.5362862582596035,
      "grad_norm": 1.5282450914382935,
      "learning_rate": 4.7855e-05,
      "loss": 0.4669,
      "step": 9577
    },
    {
      "epoch": 0.5363422555717325,
      "grad_norm": 1.3773553371429443,
      "learning_rate": 4.7860000000000006e-05,
      "loss": 0.5134,
      "step": 9578
    },
    {
      "epoch": 0.5363982528838616,
      "grad_norm": 1.3201590776443481,
      "learning_rate": 4.7865e-05,
      "loss": 0.3516,
      "step": 9579
    },
    {
      "epoch": 0.5364542501959906,
      "grad_norm": 1.338982343673706,
      "learning_rate": 4.787e-05,
      "loss": 0.504,
      "step": 9580
    },
    {
      "epoch": 0.5365102475081196,
      "grad_norm": 0.941386342048645,
      "learning_rate": 4.7875000000000005e-05,
      "loss": 0.2867,
      "step": 9581
    },
    {
      "epoch": 0.5365662448202486,
      "grad_norm": 1.6042498350143433,
      "learning_rate": 4.788e-05,
      "loss": 0.4469,
      "step": 9582
    },
    {
      "epoch": 0.5366222421323776,
      "grad_norm": 1.460040807723999,
      "learning_rate": 4.7885e-05,
      "loss": 0.4622,
      "step": 9583
    },
    {
      "epoch": 0.5366782394445067,
      "grad_norm": 1.2942413091659546,
      "learning_rate": 4.7890000000000004e-05,
      "loss": 0.3773,
      "step": 9584
    },
    {
      "epoch": 0.5367342367566357,
      "grad_norm": 1.3352640867233276,
      "learning_rate": 4.7895e-05,
      "loss": 0.4208,
      "step": 9585
    },
    {
      "epoch": 0.5367902340687647,
      "grad_norm": 1.1913381814956665,
      "learning_rate": 4.79e-05,
      "loss": 0.4136,
      "step": 9586
    },
    {
      "epoch": 0.5368462313808937,
      "grad_norm": 1.2104109525680542,
      "learning_rate": 4.7905e-05,
      "loss": 0.3568,
      "step": 9587
    },
    {
      "epoch": 0.5369022286930227,
      "grad_norm": 0.90604168176651,
      "learning_rate": 4.791000000000001e-05,
      "loss": 0.3124,
      "step": 9588
    },
    {
      "epoch": 0.5369582260051518,
      "grad_norm": 1.3224369287490845,
      "learning_rate": 4.7915000000000005e-05,
      "loss": 0.4234,
      "step": 9589
    },
    {
      "epoch": 0.5370142233172808,
      "grad_norm": 1.2113516330718994,
      "learning_rate": 4.792e-05,
      "loss": 0.4572,
      "step": 9590
    },
    {
      "epoch": 0.5370702206294098,
      "grad_norm": 1.168519139289856,
      "learning_rate": 4.7925000000000006e-05,
      "loss": 0.4055,
      "step": 9591
    },
    {
      "epoch": 0.5371262179415388,
      "grad_norm": 1.1018953323364258,
      "learning_rate": 4.7930000000000004e-05,
      "loss": 0.4418,
      "step": 9592
    },
    {
      "epoch": 0.5371822152536678,
      "grad_norm": 1.2094017267227173,
      "learning_rate": 4.7935e-05,
      "loss": 0.3992,
      "step": 9593
    },
    {
      "epoch": 0.5372382125657968,
      "grad_norm": 1.0364420413970947,
      "learning_rate": 4.794e-05,
      "loss": 0.3968,
      "step": 9594
    },
    {
      "epoch": 0.5372942098779259,
      "grad_norm": 1.2871984243392944,
      "learning_rate": 4.7945e-05,
      "loss": 0.4747,
      "step": 9595
    },
    {
      "epoch": 0.5373502071900549,
      "grad_norm": 1.1016348600387573,
      "learning_rate": 4.795e-05,
      "loss": 0.3754,
      "step": 9596
    },
    {
      "epoch": 0.5374062045021839,
      "grad_norm": 1.1769936084747314,
      "learning_rate": 4.7955e-05,
      "loss": 0.3593,
      "step": 9597
    },
    {
      "epoch": 0.5374622018143129,
      "grad_norm": 1.468197226524353,
      "learning_rate": 4.796e-05,
      "loss": 0.4162,
      "step": 9598
    },
    {
      "epoch": 0.5375181991264419,
      "grad_norm": 1.1717861890792847,
      "learning_rate": 4.7965000000000006e-05,
      "loss": 0.3874,
      "step": 9599
    },
    {
      "epoch": 0.537574196438571,
      "grad_norm": 1.3909096717834473,
      "learning_rate": 4.797e-05,
      "loss": 0.4744,
      "step": 9600
    },
    {
      "epoch": 0.5376301937507,
      "grad_norm": 1.302345871925354,
      "learning_rate": 4.7975e-05,
      "loss": 0.5447,
      "step": 9601
    },
    {
      "epoch": 0.537686191062829,
      "grad_norm": 1.271485447883606,
      "learning_rate": 4.7980000000000005e-05,
      "loss": 0.3747,
      "step": 9602
    },
    {
      "epoch": 0.537742188374958,
      "grad_norm": 1.1500025987625122,
      "learning_rate": 4.7985e-05,
      "loss": 0.3789,
      "step": 9603
    },
    {
      "epoch": 0.537798185687087,
      "grad_norm": 1.4302514791488647,
      "learning_rate": 4.799e-05,
      "loss": 0.5329,
      "step": 9604
    },
    {
      "epoch": 0.5378541829992161,
      "grad_norm": 1.191972017288208,
      "learning_rate": 4.7995000000000004e-05,
      "loss": 0.4231,
      "step": 9605
    },
    {
      "epoch": 0.5379101803113451,
      "grad_norm": 1.1573739051818848,
      "learning_rate": 4.8e-05,
      "loss": 0.3268,
      "step": 9606
    },
    {
      "epoch": 0.5379661776234741,
      "grad_norm": 1.227290391921997,
      "learning_rate": 4.8005e-05,
      "loss": 0.456,
      "step": 9607
    },
    {
      "epoch": 0.5380221749356031,
      "grad_norm": 8.732412338256836,
      "learning_rate": 4.801e-05,
      "loss": 0.4961,
      "step": 9608
    },
    {
      "epoch": 0.5380781722477321,
      "grad_norm": 1.2821083068847656,
      "learning_rate": 4.801500000000001e-05,
      "loss": 0.5198,
      "step": 9609
    },
    {
      "epoch": 0.5381341695598612,
      "grad_norm": 1.3780461549758911,
      "learning_rate": 4.8020000000000004e-05,
      "loss": 0.4525,
      "step": 9610
    },
    {
      "epoch": 0.5381901668719902,
      "grad_norm": 1.4324729442596436,
      "learning_rate": 4.8025e-05,
      "loss": 0.5719,
      "step": 9611
    },
    {
      "epoch": 0.5382461641841192,
      "grad_norm": 1.385636568069458,
      "learning_rate": 4.8030000000000006e-05,
      "loss": 0.5452,
      "step": 9612
    },
    {
      "epoch": 0.5383021614962482,
      "grad_norm": 1.4018421173095703,
      "learning_rate": 4.8035000000000003e-05,
      "loss": 0.4338,
      "step": 9613
    },
    {
      "epoch": 0.5383581588083772,
      "grad_norm": 1.3203582763671875,
      "learning_rate": 4.804e-05,
      "loss": 0.3865,
      "step": 9614
    },
    {
      "epoch": 0.5384141561205062,
      "grad_norm": 1.5596752166748047,
      "learning_rate": 4.8045e-05,
      "loss": 0.5408,
      "step": 9615
    },
    {
      "epoch": 0.5384701534326353,
      "grad_norm": 1.4226065874099731,
      "learning_rate": 4.805e-05,
      "loss": 0.4455,
      "step": 9616
    },
    {
      "epoch": 0.5385261507447643,
      "grad_norm": 1.3288135528564453,
      "learning_rate": 4.8055e-05,
      "loss": 0.4446,
      "step": 9617
    },
    {
      "epoch": 0.5385821480568933,
      "grad_norm": 1.5903502702713013,
      "learning_rate": 4.8060000000000004e-05,
      "loss": 0.4474,
      "step": 9618
    },
    {
      "epoch": 0.5386381453690223,
      "grad_norm": 1.1605653762817383,
      "learning_rate": 4.8065e-05,
      "loss": 0.4431,
      "step": 9619
    },
    {
      "epoch": 0.5386941426811513,
      "grad_norm": 1.0023702383041382,
      "learning_rate": 4.8070000000000006e-05,
      "loss": 0.4153,
      "step": 9620
    },
    {
      "epoch": 0.5387501399932804,
      "grad_norm": 1.7927113771438599,
      "learning_rate": 4.8075e-05,
      "loss": 0.5195,
      "step": 9621
    },
    {
      "epoch": 0.5388061373054094,
      "grad_norm": 1.159888744354248,
      "learning_rate": 4.808e-05,
      "loss": 0.4383,
      "step": 9622
    },
    {
      "epoch": 0.5388621346175384,
      "grad_norm": 1.092410922050476,
      "learning_rate": 4.8085000000000005e-05,
      "loss": 0.4009,
      "step": 9623
    },
    {
      "epoch": 0.5389181319296674,
      "grad_norm": 1.4227818250656128,
      "learning_rate": 4.809e-05,
      "loss": 0.5175,
      "step": 9624
    },
    {
      "epoch": 0.5389741292417964,
      "grad_norm": 1.3793169260025024,
      "learning_rate": 4.8095e-05,
      "loss": 0.5603,
      "step": 9625
    },
    {
      "epoch": 0.5390301265539255,
      "grad_norm": 1.1705825328826904,
      "learning_rate": 4.8100000000000004e-05,
      "loss": 0.4039,
      "step": 9626
    },
    {
      "epoch": 0.5390861238660545,
      "grad_norm": 1.0983772277832031,
      "learning_rate": 4.8105e-05,
      "loss": 0.4904,
      "step": 9627
    },
    {
      "epoch": 0.5391421211781835,
      "grad_norm": 1.1396361589431763,
      "learning_rate": 4.8110000000000005e-05,
      "loss": 0.5152,
      "step": 9628
    },
    {
      "epoch": 0.5391981184903125,
      "grad_norm": 1.3853065967559814,
      "learning_rate": 4.8115e-05,
      "loss": 0.3574,
      "step": 9629
    },
    {
      "epoch": 0.5392541158024415,
      "grad_norm": 1.1867951154708862,
      "learning_rate": 4.812000000000001e-05,
      "loss": 0.3441,
      "step": 9630
    },
    {
      "epoch": 0.5393101131145706,
      "grad_norm": 1.5658626556396484,
      "learning_rate": 4.8125000000000004e-05,
      "loss": 0.596,
      "step": 9631
    },
    {
      "epoch": 0.5393661104266995,
      "grad_norm": 1.3408464193344116,
      "learning_rate": 4.813e-05,
      "loss": 0.3396,
      "step": 9632
    },
    {
      "epoch": 0.5394221077388285,
      "grad_norm": 1.2120468616485596,
      "learning_rate": 4.8135e-05,
      "loss": 0.447,
      "step": 9633
    },
    {
      "epoch": 0.5394781050509575,
      "grad_norm": 1.206899642944336,
      "learning_rate": 4.814e-05,
      "loss": 0.4764,
      "step": 9634
    },
    {
      "epoch": 0.5395341023630865,
      "grad_norm": 1.2194875478744507,
      "learning_rate": 4.8145e-05,
      "loss": 0.4007,
      "step": 9635
    },
    {
      "epoch": 0.5395900996752155,
      "grad_norm": 1.3562273979187012,
      "learning_rate": 4.815e-05,
      "loss": 0.5112,
      "step": 9636
    },
    {
      "epoch": 0.5396460969873446,
      "grad_norm": 4.3270063400268555,
      "learning_rate": 4.8155e-05,
      "loss": 0.3975,
      "step": 9637
    },
    {
      "epoch": 0.5397020942994736,
      "grad_norm": 1.493651270866394,
      "learning_rate": 4.816e-05,
      "loss": 0.3844,
      "step": 9638
    },
    {
      "epoch": 0.5397580916116026,
      "grad_norm": 1.3843554258346558,
      "learning_rate": 4.8165000000000004e-05,
      "loss": 0.418,
      "step": 9639
    },
    {
      "epoch": 0.5398140889237316,
      "grad_norm": 1.1690887212753296,
      "learning_rate": 4.817e-05,
      "loss": 0.3975,
      "step": 9640
    },
    {
      "epoch": 0.5398700862358606,
      "grad_norm": 1.0352396965026855,
      "learning_rate": 4.8175000000000005e-05,
      "loss": 0.3196,
      "step": 9641
    },
    {
      "epoch": 0.5399260835479897,
      "grad_norm": 1.4064724445343018,
      "learning_rate": 4.818e-05,
      "loss": 0.3739,
      "step": 9642
    },
    {
      "epoch": 0.5399820808601187,
      "grad_norm": 1.2507612705230713,
      "learning_rate": 4.8185e-05,
      "loss": 0.3976,
      "step": 9643
    },
    {
      "epoch": 0.5400380781722477,
      "grad_norm": 1.3313459157943726,
      "learning_rate": 4.8190000000000004e-05,
      "loss": 0.469,
      "step": 9644
    },
    {
      "epoch": 0.5400940754843767,
      "grad_norm": 1.3563543558120728,
      "learning_rate": 4.8195e-05,
      "loss": 0.4279,
      "step": 9645
    },
    {
      "epoch": 0.5401500727965057,
      "grad_norm": 1.2791963815689087,
      "learning_rate": 4.82e-05,
      "loss": 0.4305,
      "step": 9646
    },
    {
      "epoch": 0.5402060701086348,
      "grad_norm": 1.1706186532974243,
      "learning_rate": 4.8205000000000003e-05,
      "loss": 0.3308,
      "step": 9647
    },
    {
      "epoch": 0.5402620674207638,
      "grad_norm": 1.3401410579681396,
      "learning_rate": 4.821e-05,
      "loss": 0.4504,
      "step": 9648
    },
    {
      "epoch": 0.5403180647328928,
      "grad_norm": 1.125596523284912,
      "learning_rate": 4.8215000000000005e-05,
      "loss": 0.4313,
      "step": 9649
    },
    {
      "epoch": 0.5403740620450218,
      "grad_norm": 1.2922368049621582,
      "learning_rate": 4.822e-05,
      "loss": 0.446,
      "step": 9650
    },
    {
      "epoch": 0.5404300593571508,
      "grad_norm": 1.2992316484451294,
      "learning_rate": 4.822500000000001e-05,
      "loss": 0.4018,
      "step": 9651
    },
    {
      "epoch": 0.5404860566692798,
      "grad_norm": 1.452916145324707,
      "learning_rate": 4.8230000000000004e-05,
      "loss": 0.4483,
      "step": 9652
    },
    {
      "epoch": 0.5405420539814089,
      "grad_norm": 1.273618221282959,
      "learning_rate": 4.8235e-05,
      "loss": 0.4828,
      "step": 9653
    },
    {
      "epoch": 0.5405980512935379,
      "grad_norm": 1.3946022987365723,
      "learning_rate": 4.824e-05,
      "loss": 0.4167,
      "step": 9654
    },
    {
      "epoch": 0.5406540486056669,
      "grad_norm": 1.1924132108688354,
      "learning_rate": 4.8245e-05,
      "loss": 0.4438,
      "step": 9655
    },
    {
      "epoch": 0.5407100459177959,
      "grad_norm": 1.2493716478347778,
      "learning_rate": 4.825e-05,
      "loss": 0.3493,
      "step": 9656
    },
    {
      "epoch": 0.5407660432299249,
      "grad_norm": 1.4506077766418457,
      "learning_rate": 4.8255e-05,
      "loss": 0.4415,
      "step": 9657
    },
    {
      "epoch": 0.540822040542054,
      "grad_norm": 1.2077593803405762,
      "learning_rate": 4.826e-05,
      "loss": 0.3707,
      "step": 9658
    },
    {
      "epoch": 0.540878037854183,
      "grad_norm": 1.5090138912200928,
      "learning_rate": 4.8265000000000006e-05,
      "loss": 0.4473,
      "step": 9659
    },
    {
      "epoch": 0.540934035166312,
      "grad_norm": 1.2844105958938599,
      "learning_rate": 4.8270000000000004e-05,
      "loss": 0.5055,
      "step": 9660
    },
    {
      "epoch": 0.540990032478441,
      "grad_norm": 1.228994607925415,
      "learning_rate": 4.8275e-05,
      "loss": 0.4315,
      "step": 9661
    },
    {
      "epoch": 0.54104602979057,
      "grad_norm": 1.5682957172393799,
      "learning_rate": 4.8280000000000005e-05,
      "loss": 0.4486,
      "step": 9662
    },
    {
      "epoch": 0.5411020271026991,
      "grad_norm": 1.6749250888824463,
      "learning_rate": 4.8285e-05,
      "loss": 0.5563,
      "step": 9663
    },
    {
      "epoch": 0.5411580244148281,
      "grad_norm": 1.1559009552001953,
      "learning_rate": 4.829e-05,
      "loss": 0.4678,
      "step": 9664
    },
    {
      "epoch": 0.5412140217269571,
      "grad_norm": 1.2349425554275513,
      "learning_rate": 4.8295000000000004e-05,
      "loss": 0.3568,
      "step": 9665
    },
    {
      "epoch": 0.5412700190390861,
      "grad_norm": 1.2909871339797974,
      "learning_rate": 4.83e-05,
      "loss": 0.4112,
      "step": 9666
    },
    {
      "epoch": 0.5413260163512151,
      "grad_norm": 1.3625513315200806,
      "learning_rate": 4.8305e-05,
      "loss": 0.335,
      "step": 9667
    },
    {
      "epoch": 0.5413820136633442,
      "grad_norm": 1.8493263721466064,
      "learning_rate": 4.8309999999999997e-05,
      "loss": 0.5126,
      "step": 9668
    },
    {
      "epoch": 0.5414380109754732,
      "grad_norm": 1.6567726135253906,
      "learning_rate": 4.831500000000001e-05,
      "loss": 0.486,
      "step": 9669
    },
    {
      "epoch": 0.5414940082876022,
      "grad_norm": 1.2510757446289062,
      "learning_rate": 4.8320000000000005e-05,
      "loss": 0.362,
      "step": 9670
    },
    {
      "epoch": 0.5415500055997312,
      "grad_norm": 1.2585492134094238,
      "learning_rate": 4.8325e-05,
      "loss": 0.4178,
      "step": 9671
    },
    {
      "epoch": 0.5416060029118602,
      "grad_norm": 1.2923367023468018,
      "learning_rate": 4.833e-05,
      "loss": 0.4974,
      "step": 9672
    },
    {
      "epoch": 0.5416620002239892,
      "grad_norm": 1.4480977058410645,
      "learning_rate": 4.8335000000000004e-05,
      "loss": 0.4164,
      "step": 9673
    },
    {
      "epoch": 0.5417179975361183,
      "grad_norm": 1.1718289852142334,
      "learning_rate": 4.834e-05,
      "loss": 0.3666,
      "step": 9674
    },
    {
      "epoch": 0.5417739948482473,
      "grad_norm": 1.2877403497695923,
      "learning_rate": 4.8345e-05,
      "loss": 0.4637,
      "step": 9675
    },
    {
      "epoch": 0.5418299921603763,
      "grad_norm": 1.0981522798538208,
      "learning_rate": 4.835e-05,
      "loss": 0.3544,
      "step": 9676
    },
    {
      "epoch": 0.5418859894725053,
      "grad_norm": 1.439459204673767,
      "learning_rate": 4.8355e-05,
      "loss": 0.4665,
      "step": 9677
    },
    {
      "epoch": 0.5419419867846343,
      "grad_norm": 1.380591869354248,
      "learning_rate": 4.836e-05,
      "loss": 0.3931,
      "step": 9678
    },
    {
      "epoch": 0.5419979840967634,
      "grad_norm": 1.3844058513641357,
      "learning_rate": 4.8365e-05,
      "loss": 0.3871,
      "step": 9679
    },
    {
      "epoch": 0.5420539814088924,
      "grad_norm": 1.4423999786376953,
      "learning_rate": 4.8370000000000006e-05,
      "loss": 0.5286,
      "step": 9680
    },
    {
      "epoch": 0.5421099787210214,
      "grad_norm": 1.132737159729004,
      "learning_rate": 4.8375000000000004e-05,
      "loss": 0.4016,
      "step": 9681
    },
    {
      "epoch": 0.5421659760331504,
      "grad_norm": 1.215610384941101,
      "learning_rate": 4.838e-05,
      "loss": 0.3603,
      "step": 9682
    },
    {
      "epoch": 0.5422219733452794,
      "grad_norm": 1.3401881456375122,
      "learning_rate": 4.8385000000000005e-05,
      "loss": 0.5025,
      "step": 9683
    },
    {
      "epoch": 0.5422779706574085,
      "grad_norm": 1.0976073741912842,
      "learning_rate": 4.839e-05,
      "loss": 0.3805,
      "step": 9684
    },
    {
      "epoch": 0.5423339679695375,
      "grad_norm": 1.5351399183273315,
      "learning_rate": 4.8395e-05,
      "loss": 0.5663,
      "step": 9685
    },
    {
      "epoch": 0.5423899652816665,
      "grad_norm": 1.4356317520141602,
      "learning_rate": 4.8400000000000004e-05,
      "loss": 0.4673,
      "step": 9686
    },
    {
      "epoch": 0.5424459625937955,
      "grad_norm": 1.343070387840271,
      "learning_rate": 4.8405e-05,
      "loss": 0.3795,
      "step": 9687
    },
    {
      "epoch": 0.5425019599059245,
      "grad_norm": 1.2418947219848633,
      "learning_rate": 4.841e-05,
      "loss": 0.3363,
      "step": 9688
    },
    {
      "epoch": 0.5425579572180536,
      "grad_norm": 1.2552831172943115,
      "learning_rate": 4.8415e-05,
      "loss": 0.4219,
      "step": 9689
    },
    {
      "epoch": 0.5426139545301826,
      "grad_norm": 1.0641114711761475,
      "learning_rate": 4.842000000000001e-05,
      "loss": 0.425,
      "step": 9690
    },
    {
      "epoch": 0.5426699518423116,
      "grad_norm": 1.29921293258667,
      "learning_rate": 4.8425000000000005e-05,
      "loss": 0.4904,
      "step": 9691
    },
    {
      "epoch": 0.5427259491544406,
      "grad_norm": 1.0799882411956787,
      "learning_rate": 4.843e-05,
      "loss": 0.4693,
      "step": 9692
    },
    {
      "epoch": 0.5427819464665696,
      "grad_norm": 1.4437817335128784,
      "learning_rate": 4.8435e-05,
      "loss": 0.5803,
      "step": 9693
    },
    {
      "epoch": 0.5428379437786987,
      "grad_norm": 1.2713543176651,
      "learning_rate": 4.8440000000000004e-05,
      "loss": 0.3694,
      "step": 9694
    },
    {
      "epoch": 0.5428939410908277,
      "grad_norm": 1.7143093347549438,
      "learning_rate": 4.8445e-05,
      "loss": 0.6177,
      "step": 9695
    },
    {
      "epoch": 0.5429499384029567,
      "grad_norm": 1.1647511720657349,
      "learning_rate": 4.845e-05,
      "loss": 0.3676,
      "step": 9696
    },
    {
      "epoch": 0.5430059357150857,
      "grad_norm": 1.4182149171829224,
      "learning_rate": 4.8455e-05,
      "loss": 0.3801,
      "step": 9697
    },
    {
      "epoch": 0.5430619330272147,
      "grad_norm": 1.255975604057312,
      "learning_rate": 4.846e-05,
      "loss": 0.3259,
      "step": 9698
    },
    {
      "epoch": 0.5431179303393437,
      "grad_norm": 1.003980040550232,
      "learning_rate": 4.8465000000000004e-05,
      "loss": 0.3206,
      "step": 9699
    },
    {
      "epoch": 0.5431739276514728,
      "grad_norm": 1.301956057548523,
      "learning_rate": 4.847e-05,
      "loss": 0.368,
      "step": 9700
    },
    {
      "epoch": 0.5432299249636018,
      "grad_norm": 1.4601362943649292,
      "learning_rate": 4.8475000000000006e-05,
      "loss": 0.4239,
      "step": 9701
    },
    {
      "epoch": 0.5432859222757308,
      "grad_norm": 8.5525541305542,
      "learning_rate": 4.8480000000000003e-05,
      "loss": 0.4612,
      "step": 9702
    },
    {
      "epoch": 0.5433419195878598,
      "grad_norm": 1.5122413635253906,
      "learning_rate": 4.8485e-05,
      "loss": 0.6119,
      "step": 9703
    },
    {
      "epoch": 0.5433979168999888,
      "grad_norm": 1.0617645978927612,
      "learning_rate": 4.8490000000000005e-05,
      "loss": 0.345,
      "step": 9704
    },
    {
      "epoch": 0.5434539142121179,
      "grad_norm": 1.254473328590393,
      "learning_rate": 4.8495e-05,
      "loss": 0.6635,
      "step": 9705
    },
    {
      "epoch": 0.5435099115242469,
      "grad_norm": 0.9798499345779419,
      "learning_rate": 4.85e-05,
      "loss": 0.3501,
      "step": 9706
    },
    {
      "epoch": 0.5435659088363759,
      "grad_norm": 1.405495285987854,
      "learning_rate": 4.8505e-05,
      "loss": 0.4159,
      "step": 9707
    },
    {
      "epoch": 0.5436219061485049,
      "grad_norm": 1.1959142684936523,
      "learning_rate": 4.851e-05,
      "loss": 0.5036,
      "step": 9708
    },
    {
      "epoch": 0.5436779034606339,
      "grad_norm": 1.6853902339935303,
      "learning_rate": 4.8515000000000006e-05,
      "loss": 0.4186,
      "step": 9709
    },
    {
      "epoch": 0.543733900772763,
      "grad_norm": 1.2730882167816162,
      "learning_rate": 4.852e-05,
      "loss": 0.419,
      "step": 9710
    },
    {
      "epoch": 0.543789898084892,
      "grad_norm": 1.2109779119491577,
      "learning_rate": 4.8525e-05,
      "loss": 0.4926,
      "step": 9711
    },
    {
      "epoch": 0.543845895397021,
      "grad_norm": 1.1662299633026123,
      "learning_rate": 4.8530000000000005e-05,
      "loss": 0.4484,
      "step": 9712
    },
    {
      "epoch": 0.54390189270915,
      "grad_norm": 1.0187838077545166,
      "learning_rate": 4.8535e-05,
      "loss": 0.306,
      "step": 9713
    },
    {
      "epoch": 0.543957890021279,
      "grad_norm": 1.3637886047363281,
      "learning_rate": 4.854e-05,
      "loss": 0.4984,
      "step": 9714
    },
    {
      "epoch": 0.5440138873334079,
      "grad_norm": 1.4112552404403687,
      "learning_rate": 4.8545000000000004e-05,
      "loss": 0.4013,
      "step": 9715
    },
    {
      "epoch": 0.544069884645537,
      "grad_norm": 1.2921721935272217,
      "learning_rate": 4.855e-05,
      "loss": 0.512,
      "step": 9716
    },
    {
      "epoch": 0.544125881957666,
      "grad_norm": 1.2291465997695923,
      "learning_rate": 4.8555e-05,
      "loss": 0.4627,
      "step": 9717
    },
    {
      "epoch": 0.544181879269795,
      "grad_norm": 1.3939390182495117,
      "learning_rate": 4.856e-05,
      "loss": 0.3999,
      "step": 9718
    },
    {
      "epoch": 0.544237876581924,
      "grad_norm": 1.3417540788650513,
      "learning_rate": 4.856500000000001e-05,
      "loss": 0.4685,
      "step": 9719
    },
    {
      "epoch": 0.544293873894053,
      "grad_norm": 1.4190226793289185,
      "learning_rate": 4.8570000000000004e-05,
      "loss": 0.348,
      "step": 9720
    },
    {
      "epoch": 0.5443498712061821,
      "grad_norm": 1.4172230958938599,
      "learning_rate": 4.8575e-05,
      "loss": 0.6203,
      "step": 9721
    },
    {
      "epoch": 0.5444058685183111,
      "grad_norm": 1.4824928045272827,
      "learning_rate": 4.8580000000000006e-05,
      "loss": 0.4865,
      "step": 9722
    },
    {
      "epoch": 0.5444618658304401,
      "grad_norm": 1.2636237144470215,
      "learning_rate": 4.8585e-05,
      "loss": 0.4615,
      "step": 9723
    },
    {
      "epoch": 0.5445178631425691,
      "grad_norm": 1.3195079565048218,
      "learning_rate": 4.859e-05,
      "loss": 0.4728,
      "step": 9724
    },
    {
      "epoch": 0.5445738604546981,
      "grad_norm": 1.4156670570373535,
      "learning_rate": 4.8595000000000005e-05,
      "loss": 0.5529,
      "step": 9725
    },
    {
      "epoch": 0.5446298577668272,
      "grad_norm": 1.441025733947754,
      "learning_rate": 4.86e-05,
      "loss": 0.518,
      "step": 9726
    },
    {
      "epoch": 0.5446858550789562,
      "grad_norm": 1.2869678735733032,
      "learning_rate": 4.8605e-05,
      "loss": 0.3476,
      "step": 9727
    },
    {
      "epoch": 0.5447418523910852,
      "grad_norm": 1.0636709928512573,
      "learning_rate": 4.861e-05,
      "loss": 0.4203,
      "step": 9728
    },
    {
      "epoch": 0.5447978497032142,
      "grad_norm": 1.2076289653778076,
      "learning_rate": 4.861500000000001e-05,
      "loss": 0.4452,
      "step": 9729
    },
    {
      "epoch": 0.5448538470153432,
      "grad_norm": 1.218843698501587,
      "learning_rate": 4.8620000000000005e-05,
      "loss": 0.3975,
      "step": 9730
    },
    {
      "epoch": 0.5449098443274722,
      "grad_norm": 2.140641927719116,
      "learning_rate": 4.8625e-05,
      "loss": 0.4852,
      "step": 9731
    },
    {
      "epoch": 0.5449658416396013,
      "grad_norm": 1.2406911849975586,
      "learning_rate": 4.863e-05,
      "loss": 0.4207,
      "step": 9732
    },
    {
      "epoch": 0.5450218389517303,
      "grad_norm": 1.2015538215637207,
      "learning_rate": 4.8635000000000004e-05,
      "loss": 0.4469,
      "step": 9733
    },
    {
      "epoch": 0.5450778362638593,
      "grad_norm": 1.1803382635116577,
      "learning_rate": 4.864e-05,
      "loss": 0.4405,
      "step": 9734
    },
    {
      "epoch": 0.5451338335759883,
      "grad_norm": 1.1898819208145142,
      "learning_rate": 4.8645e-05,
      "loss": 0.4249,
      "step": 9735
    },
    {
      "epoch": 0.5451898308881173,
      "grad_norm": 1.2134382724761963,
      "learning_rate": 4.8650000000000003e-05,
      "loss": 0.4567,
      "step": 9736
    },
    {
      "epoch": 0.5452458282002464,
      "grad_norm": 1.4385006427764893,
      "learning_rate": 4.8655e-05,
      "loss": 0.4381,
      "step": 9737
    },
    {
      "epoch": 0.5453018255123754,
      "grad_norm": 1.1139192581176758,
      "learning_rate": 4.866e-05,
      "loss": 0.326,
      "step": 9738
    },
    {
      "epoch": 0.5453578228245044,
      "grad_norm": 1.3936790227890015,
      "learning_rate": 4.8665e-05,
      "loss": 0.4393,
      "step": 9739
    },
    {
      "epoch": 0.5454138201366334,
      "grad_norm": 1.3649449348449707,
      "learning_rate": 4.867000000000001e-05,
      "loss": 0.5342,
      "step": 9740
    },
    {
      "epoch": 0.5454698174487624,
      "grad_norm": 1.022350549697876,
      "learning_rate": 4.8675000000000004e-05,
      "loss": 0.3274,
      "step": 9741
    },
    {
      "epoch": 0.5455258147608915,
      "grad_norm": 1.327319860458374,
      "learning_rate": 4.868e-05,
      "loss": 0.4178,
      "step": 9742
    },
    {
      "epoch": 0.5455818120730205,
      "grad_norm": 1.623950719833374,
      "learning_rate": 4.8685000000000006e-05,
      "loss": 0.4655,
      "step": 9743
    },
    {
      "epoch": 0.5456378093851495,
      "grad_norm": 1.4667751789093018,
      "learning_rate": 4.869e-05,
      "loss": 0.4312,
      "step": 9744
    },
    {
      "epoch": 0.5456938066972785,
      "grad_norm": 1.2036206722259521,
      "learning_rate": 4.8695e-05,
      "loss": 0.4892,
      "step": 9745
    },
    {
      "epoch": 0.5457498040094075,
      "grad_norm": 1.1602802276611328,
      "learning_rate": 4.87e-05,
      "loss": 0.4343,
      "step": 9746
    },
    {
      "epoch": 0.5458058013215366,
      "grad_norm": 1.2215708494186401,
      "learning_rate": 4.8705e-05,
      "loss": 0.3786,
      "step": 9747
    },
    {
      "epoch": 0.5458617986336656,
      "grad_norm": 1.2116730213165283,
      "learning_rate": 4.871e-05,
      "loss": 0.4915,
      "step": 9748
    },
    {
      "epoch": 0.5459177959457946,
      "grad_norm": 1.3704102039337158,
      "learning_rate": 4.8715000000000004e-05,
      "loss": 0.4883,
      "step": 9749
    },
    {
      "epoch": 0.5459737932579236,
      "grad_norm": 1.4366365671157837,
      "learning_rate": 4.872000000000001e-05,
      "loss": 0.437,
      "step": 9750
    },
    {
      "epoch": 0.5460297905700526,
      "grad_norm": 1.5466474294662476,
      "learning_rate": 4.8725000000000005e-05,
      "loss": 0.503,
      "step": 9751
    },
    {
      "epoch": 0.5460857878821817,
      "grad_norm": 1.1597400903701782,
      "learning_rate": 4.873e-05,
      "loss": 0.426,
      "step": 9752
    },
    {
      "epoch": 0.5461417851943107,
      "grad_norm": 1.275761604309082,
      "learning_rate": 4.8735e-05,
      "loss": 0.386,
      "step": 9753
    },
    {
      "epoch": 0.5461977825064397,
      "grad_norm": 1.024357795715332,
      "learning_rate": 4.8740000000000004e-05,
      "loss": 0.4879,
      "step": 9754
    },
    {
      "epoch": 0.5462537798185687,
      "grad_norm": 1.1839563846588135,
      "learning_rate": 4.8745e-05,
      "loss": 0.3731,
      "step": 9755
    },
    {
      "epoch": 0.5463097771306977,
      "grad_norm": 1.2781145572662354,
      "learning_rate": 4.875e-05,
      "loss": 0.3335,
      "step": 9756
    },
    {
      "epoch": 0.5463657744428267,
      "grad_norm": 1.391566514968872,
      "learning_rate": 4.8755e-05,
      "loss": 0.5612,
      "step": 9757
    },
    {
      "epoch": 0.5464217717549558,
      "grad_norm": 1.3660463094711304,
      "learning_rate": 4.876e-05,
      "loss": 0.4597,
      "step": 9758
    },
    {
      "epoch": 0.5464777690670848,
      "grad_norm": 1.1681660413742065,
      "learning_rate": 4.8765e-05,
      "loss": 0.435,
      "step": 9759
    },
    {
      "epoch": 0.5465337663792138,
      "grad_norm": 1.3454241752624512,
      "learning_rate": 4.877e-05,
      "loss": 0.4118,
      "step": 9760
    },
    {
      "epoch": 0.5465897636913428,
      "grad_norm": 1.1808377504348755,
      "learning_rate": 4.8775000000000007e-05,
      "loss": 0.3623,
      "step": 9761
    },
    {
      "epoch": 0.5466457610034718,
      "grad_norm": 1.3442800045013428,
      "learning_rate": 4.8780000000000004e-05,
      "loss": 0.5379,
      "step": 9762
    },
    {
      "epoch": 0.5467017583156009,
      "grad_norm": 1.2783876657485962,
      "learning_rate": 4.8785e-05,
      "loss": 0.4813,
      "step": 9763
    },
    {
      "epoch": 0.5467577556277299,
      "grad_norm": 1.2709224224090576,
      "learning_rate": 4.8790000000000006e-05,
      "loss": 0.3285,
      "step": 9764
    },
    {
      "epoch": 0.5468137529398589,
      "grad_norm": 1.3587431907653809,
      "learning_rate": 4.8795e-05,
      "loss": 0.3749,
      "step": 9765
    },
    {
      "epoch": 0.5468697502519879,
      "grad_norm": 1.3329561948776245,
      "learning_rate": 4.88e-05,
      "loss": 0.4517,
      "step": 9766
    },
    {
      "epoch": 0.5469257475641169,
      "grad_norm": 1.3308213949203491,
      "learning_rate": 4.8805e-05,
      "loss": 0.4159,
      "step": 9767
    },
    {
      "epoch": 0.546981744876246,
      "grad_norm": 1.2997395992279053,
      "learning_rate": 4.881e-05,
      "loss": 0.3801,
      "step": 9768
    },
    {
      "epoch": 0.547037742188375,
      "grad_norm": 1.3082472085952759,
      "learning_rate": 4.8815e-05,
      "loss": 0.5065,
      "step": 9769
    },
    {
      "epoch": 0.547093739500504,
      "grad_norm": 1.2000705003738403,
      "learning_rate": 4.8820000000000004e-05,
      "loss": 0.3443,
      "step": 9770
    },
    {
      "epoch": 0.547149736812633,
      "grad_norm": 1.4347835779190063,
      "learning_rate": 4.8825e-05,
      "loss": 0.5783,
      "step": 9771
    },
    {
      "epoch": 0.547205734124762,
      "grad_norm": 1.2902127504348755,
      "learning_rate": 4.8830000000000005e-05,
      "loss": 0.3725,
      "step": 9772
    },
    {
      "epoch": 0.547261731436891,
      "grad_norm": 1.6311655044555664,
      "learning_rate": 4.8835e-05,
      "loss": 0.4099,
      "step": 9773
    },
    {
      "epoch": 0.5473177287490201,
      "grad_norm": 1.20345938205719,
      "learning_rate": 4.884e-05,
      "loss": 0.3987,
      "step": 9774
    },
    {
      "epoch": 0.5473737260611491,
      "grad_norm": 1.1928012371063232,
      "learning_rate": 4.8845000000000004e-05,
      "loss": 0.4513,
      "step": 9775
    },
    {
      "epoch": 0.5474297233732781,
      "grad_norm": 1.2258514165878296,
      "learning_rate": 4.885e-05,
      "loss": 0.4534,
      "step": 9776
    },
    {
      "epoch": 0.5474857206854071,
      "grad_norm": 1.28345787525177,
      "learning_rate": 4.8855e-05,
      "loss": 0.429,
      "step": 9777
    },
    {
      "epoch": 0.5475417179975361,
      "grad_norm": 1.2173876762390137,
      "learning_rate": 4.886e-05,
      "loss": 0.3636,
      "step": 9778
    },
    {
      "epoch": 0.5475977153096652,
      "grad_norm": 1.2504884004592896,
      "learning_rate": 4.8865e-05,
      "loss": 0.55,
      "step": 9779
    },
    {
      "epoch": 0.5476537126217942,
      "grad_norm": 1.0849627256393433,
      "learning_rate": 4.8870000000000005e-05,
      "loss": 0.4151,
      "step": 9780
    },
    {
      "epoch": 0.5477097099339232,
      "grad_norm": 1.2154240608215332,
      "learning_rate": 4.8875e-05,
      "loss": 0.4034,
      "step": 9781
    },
    {
      "epoch": 0.5477657072460522,
      "grad_norm": 1.1576330661773682,
      "learning_rate": 4.8880000000000006e-05,
      "loss": 0.4272,
      "step": 9782
    },
    {
      "epoch": 0.5478217045581812,
      "grad_norm": 1.195458173751831,
      "learning_rate": 4.8885000000000004e-05,
      "loss": 0.4196,
      "step": 9783
    },
    {
      "epoch": 0.5478777018703103,
      "grad_norm": 1.3338898420333862,
      "learning_rate": 4.889e-05,
      "loss": 0.5115,
      "step": 9784
    },
    {
      "epoch": 0.5479336991824393,
      "grad_norm": 1.3685988187789917,
      "learning_rate": 4.8895e-05,
      "loss": 0.3735,
      "step": 9785
    },
    {
      "epoch": 0.5479896964945683,
      "grad_norm": 0.9911099076271057,
      "learning_rate": 4.89e-05,
      "loss": 0.2798,
      "step": 9786
    },
    {
      "epoch": 0.5480456938066973,
      "grad_norm": 1.5359314680099487,
      "learning_rate": 4.8905e-05,
      "loss": 0.5281,
      "step": 9787
    },
    {
      "epoch": 0.5481016911188263,
      "grad_norm": 1.1968533992767334,
      "learning_rate": 4.891e-05,
      "loss": 0.4735,
      "step": 9788
    },
    {
      "epoch": 0.5481576884309554,
      "grad_norm": 1.3393092155456543,
      "learning_rate": 4.8915e-05,
      "loss": 0.4754,
      "step": 9789
    },
    {
      "epoch": 0.5482136857430844,
      "grad_norm": 1.2218310832977295,
      "learning_rate": 4.8920000000000006e-05,
      "loss": 0.3819,
      "step": 9790
    },
    {
      "epoch": 0.5482696830552134,
      "grad_norm": 1.3259683847427368,
      "learning_rate": 4.8925e-05,
      "loss": 0.4729,
      "step": 9791
    },
    {
      "epoch": 0.5483256803673424,
      "grad_norm": 1.296262264251709,
      "learning_rate": 4.893e-05,
      "loss": 0.4812,
      "step": 9792
    },
    {
      "epoch": 0.5483816776794714,
      "grad_norm": 1.2933176755905151,
      "learning_rate": 4.8935000000000005e-05,
      "loss": 0.3868,
      "step": 9793
    },
    {
      "epoch": 0.5484376749916005,
      "grad_norm": 1.1595845222473145,
      "learning_rate": 4.894e-05,
      "loss": 0.3071,
      "step": 9794
    },
    {
      "epoch": 0.5484936723037295,
      "grad_norm": 1.2668019533157349,
      "learning_rate": 4.8945e-05,
      "loss": 0.487,
      "step": 9795
    },
    {
      "epoch": 0.5485496696158585,
      "grad_norm": 1.3164705038070679,
      "learning_rate": 4.8950000000000004e-05,
      "loss": 0.4585,
      "step": 9796
    },
    {
      "epoch": 0.5486056669279875,
      "grad_norm": 1.343824028968811,
      "learning_rate": 4.8955e-05,
      "loss": 0.4897,
      "step": 9797
    },
    {
      "epoch": 0.5486616642401164,
      "grad_norm": 1.3883174657821655,
      "learning_rate": 4.896e-05,
      "loss": 0.4444,
      "step": 9798
    },
    {
      "epoch": 0.5487176615522454,
      "grad_norm": 1.1468616724014282,
      "learning_rate": 4.8965e-05,
      "loss": 0.4213,
      "step": 9799
    },
    {
      "epoch": 0.5487736588643745,
      "grad_norm": 1.2404942512512207,
      "learning_rate": 4.897000000000001e-05,
      "loss": 0.3937,
      "step": 9800
    },
    {
      "epoch": 0.5488296561765035,
      "grad_norm": 1.168466567993164,
      "learning_rate": 4.8975000000000005e-05,
      "loss": 0.3887,
      "step": 9801
    },
    {
      "epoch": 0.5488856534886325,
      "grad_norm": 1.2331757545471191,
      "learning_rate": 4.898e-05,
      "loss": 0.4094,
      "step": 9802
    },
    {
      "epoch": 0.5489416508007615,
      "grad_norm": 1.3100107908248901,
      "learning_rate": 4.8985000000000006e-05,
      "loss": 0.515,
      "step": 9803
    },
    {
      "epoch": 0.5489976481128905,
      "grad_norm": 1.0910251140594482,
      "learning_rate": 4.8990000000000004e-05,
      "loss": 0.4082,
      "step": 9804
    },
    {
      "epoch": 0.5490536454250196,
      "grad_norm": 1.0836122035980225,
      "learning_rate": 4.8995e-05,
      "loss": 0.3982,
      "step": 9805
    },
    {
      "epoch": 0.5491096427371486,
      "grad_norm": 1.4018274545669556,
      "learning_rate": 4.9e-05,
      "loss": 0.413,
      "step": 9806
    },
    {
      "epoch": 0.5491656400492776,
      "grad_norm": 1.7344642877578735,
      "learning_rate": 4.9005e-05,
      "loss": 0.4465,
      "step": 9807
    },
    {
      "epoch": 0.5492216373614066,
      "grad_norm": 1.191381812095642,
      "learning_rate": 4.901e-05,
      "loss": 0.4133,
      "step": 9808
    },
    {
      "epoch": 0.5492776346735356,
      "grad_norm": 1.0895473957061768,
      "learning_rate": 4.9015e-05,
      "loss": 0.3796,
      "step": 9809
    },
    {
      "epoch": 0.5493336319856646,
      "grad_norm": 1.3561818599700928,
      "learning_rate": 4.902e-05,
      "loss": 0.4408,
      "step": 9810
    },
    {
      "epoch": 0.5493896292977937,
      "grad_norm": 1.2005116939544678,
      "learning_rate": 4.9025000000000006e-05,
      "loss": 0.3827,
      "step": 9811
    },
    {
      "epoch": 0.5494456266099227,
      "grad_norm": 1.3890430927276611,
      "learning_rate": 4.903e-05,
      "loss": 0.4115,
      "step": 9812
    },
    {
      "epoch": 0.5495016239220517,
      "grad_norm": 1.304787039756775,
      "learning_rate": 4.9035e-05,
      "loss": 0.3443,
      "step": 9813
    },
    {
      "epoch": 0.5495576212341807,
      "grad_norm": 1.6234766244888306,
      "learning_rate": 4.9040000000000005e-05,
      "loss": 0.5334,
      "step": 9814
    },
    {
      "epoch": 0.5496136185463097,
      "grad_norm": 1.1996712684631348,
      "learning_rate": 4.9045e-05,
      "loss": 0.487,
      "step": 9815
    },
    {
      "epoch": 0.5496696158584388,
      "grad_norm": 1.5660183429718018,
      "learning_rate": 4.905e-05,
      "loss": 0.6293,
      "step": 9816
    },
    {
      "epoch": 0.5497256131705678,
      "grad_norm": 1.3654955625534058,
      "learning_rate": 4.9055000000000004e-05,
      "loss": 0.4961,
      "step": 9817
    },
    {
      "epoch": 0.5497816104826968,
      "grad_norm": 1.6731728315353394,
      "learning_rate": 4.906e-05,
      "loss": 0.6572,
      "step": 9818
    },
    {
      "epoch": 0.5498376077948258,
      "grad_norm": 1.17780601978302,
      "learning_rate": 4.9065e-05,
      "loss": 0.4288,
      "step": 9819
    },
    {
      "epoch": 0.5498936051069548,
      "grad_norm": 1.3803881406784058,
      "learning_rate": 4.907e-05,
      "loss": 0.4575,
      "step": 9820
    },
    {
      "epoch": 0.5499496024190839,
      "grad_norm": 1.2742030620574951,
      "learning_rate": 4.907500000000001e-05,
      "loss": 0.3687,
      "step": 9821
    },
    {
      "epoch": 0.5500055997312129,
      "grad_norm": 1.4195208549499512,
      "learning_rate": 4.9080000000000004e-05,
      "loss": 0.4378,
      "step": 9822
    },
    {
      "epoch": 0.5500615970433419,
      "grad_norm": 1.21417236328125,
      "learning_rate": 4.9085e-05,
      "loss": 0.5454,
      "step": 9823
    },
    {
      "epoch": 0.5501175943554709,
      "grad_norm": 1.2277599573135376,
      "learning_rate": 4.9090000000000006e-05,
      "loss": 0.3335,
      "step": 9824
    },
    {
      "epoch": 0.5501735916675999,
      "grad_norm": 1.2298928499221802,
      "learning_rate": 4.9095000000000003e-05,
      "loss": 0.5312,
      "step": 9825
    },
    {
      "epoch": 0.550229588979729,
      "grad_norm": 1.0673483610153198,
      "learning_rate": 4.91e-05,
      "loss": 0.3275,
      "step": 9826
    },
    {
      "epoch": 0.550285586291858,
      "grad_norm": 1.1780321598052979,
      "learning_rate": 4.9105e-05,
      "loss": 0.437,
      "step": 9827
    },
    {
      "epoch": 0.550341583603987,
      "grad_norm": 1.2712862491607666,
      "learning_rate": 4.911e-05,
      "loss": 0.3853,
      "step": 9828
    },
    {
      "epoch": 0.550397580916116,
      "grad_norm": 1.6301132440567017,
      "learning_rate": 4.9115e-05,
      "loss": 0.6661,
      "step": 9829
    },
    {
      "epoch": 0.550453578228245,
      "grad_norm": 1.2521926164627075,
      "learning_rate": 4.9120000000000004e-05,
      "loss": 0.3093,
      "step": 9830
    },
    {
      "epoch": 0.550509575540374,
      "grad_norm": 1.507238507270813,
      "learning_rate": 4.9125e-05,
      "loss": 0.4127,
      "step": 9831
    },
    {
      "epoch": 0.5505655728525031,
      "grad_norm": 1.2705938816070557,
      "learning_rate": 4.9130000000000006e-05,
      "loss": 0.5905,
      "step": 9832
    },
    {
      "epoch": 0.5506215701646321,
      "grad_norm": 1.9501150846481323,
      "learning_rate": 4.9135e-05,
      "loss": 0.5766,
      "step": 9833
    },
    {
      "epoch": 0.5506775674767611,
      "grad_norm": 1.173202633857727,
      "learning_rate": 4.914e-05,
      "loss": 0.341,
      "step": 9834
    },
    {
      "epoch": 0.5507335647888901,
      "grad_norm": 1.2615176439285278,
      "learning_rate": 4.9145000000000005e-05,
      "loss": 0.4548,
      "step": 9835
    },
    {
      "epoch": 0.5507895621010191,
      "grad_norm": 1.2580984830856323,
      "learning_rate": 4.915e-05,
      "loss": 0.4681,
      "step": 9836
    },
    {
      "epoch": 0.5508455594131482,
      "grad_norm": 1.2937867641448975,
      "learning_rate": 4.9155e-05,
      "loss": 0.3072,
      "step": 9837
    },
    {
      "epoch": 0.5509015567252772,
      "grad_norm": 1.0775372982025146,
      "learning_rate": 4.9160000000000004e-05,
      "loss": 0.3315,
      "step": 9838
    },
    {
      "epoch": 0.5509575540374062,
      "grad_norm": 1.1573628187179565,
      "learning_rate": 4.9165e-05,
      "loss": 0.4074,
      "step": 9839
    },
    {
      "epoch": 0.5510135513495352,
      "grad_norm": 1.6001033782958984,
      "learning_rate": 4.9170000000000005e-05,
      "loss": 0.4437,
      "step": 9840
    },
    {
      "epoch": 0.5510695486616642,
      "grad_norm": 1.460305094718933,
      "learning_rate": 4.9175e-05,
      "loss": 0.5827,
      "step": 9841
    },
    {
      "epoch": 0.5511255459737933,
      "grad_norm": 1.392538070678711,
      "learning_rate": 4.918000000000001e-05,
      "loss": 0.4461,
      "step": 9842
    },
    {
      "epoch": 0.5511815432859223,
      "grad_norm": 1.1869699954986572,
      "learning_rate": 4.9185000000000004e-05,
      "loss": 0.3958,
      "step": 9843
    },
    {
      "epoch": 0.5512375405980513,
      "grad_norm": 1.3342792987823486,
      "learning_rate": 4.919e-05,
      "loss": 0.4169,
      "step": 9844
    },
    {
      "epoch": 0.5512935379101803,
      "grad_norm": 5.5547990798950195,
      "learning_rate": 4.9195e-05,
      "loss": 0.546,
      "step": 9845
    },
    {
      "epoch": 0.5513495352223093,
      "grad_norm": 1.1892287731170654,
      "learning_rate": 4.92e-05,
      "loss": 0.3339,
      "step": 9846
    },
    {
      "epoch": 0.5514055325344384,
      "grad_norm": 1.1834392547607422,
      "learning_rate": 4.9205e-05,
      "loss": 0.4276,
      "step": 9847
    },
    {
      "epoch": 0.5514615298465674,
      "grad_norm": 1.444663166999817,
      "learning_rate": 4.921e-05,
      "loss": 0.3811,
      "step": 9848
    },
    {
      "epoch": 0.5515175271586964,
      "grad_norm": 1.4277019500732422,
      "learning_rate": 4.9215e-05,
      "loss": 0.3793,
      "step": 9849
    },
    {
      "epoch": 0.5515735244708254,
      "grad_norm": 1.271371841430664,
      "learning_rate": 4.9220000000000006e-05,
      "loss": 0.5175,
      "step": 9850
    },
    {
      "epoch": 0.5516295217829544,
      "grad_norm": 1.8749375343322754,
      "learning_rate": 4.9225000000000004e-05,
      "loss": 0.5032,
      "step": 9851
    },
    {
      "epoch": 0.5516855190950835,
      "grad_norm": 1.6567984819412231,
      "learning_rate": 4.923e-05,
      "loss": 0.5108,
      "step": 9852
    },
    {
      "epoch": 0.5517415164072125,
      "grad_norm": 1.0445003509521484,
      "learning_rate": 4.9235000000000005e-05,
      "loss": 0.4001,
      "step": 9853
    },
    {
      "epoch": 0.5517975137193415,
      "grad_norm": 1.4645371437072754,
      "learning_rate": 4.924e-05,
      "loss": 0.4019,
      "step": 9854
    },
    {
      "epoch": 0.5518535110314705,
      "grad_norm": 1.3834123611450195,
      "learning_rate": 4.9245e-05,
      "loss": 0.5157,
      "step": 9855
    },
    {
      "epoch": 0.5519095083435995,
      "grad_norm": 1.0995264053344727,
      "learning_rate": 4.9250000000000004e-05,
      "loss": 0.3924,
      "step": 9856
    },
    {
      "epoch": 0.5519655056557285,
      "grad_norm": 1.1953632831573486,
      "learning_rate": 4.9255e-05,
      "loss": 0.5091,
      "step": 9857
    },
    {
      "epoch": 0.5520215029678576,
      "grad_norm": 1.246602177619934,
      "learning_rate": 4.926e-05,
      "loss": 0.5428,
      "step": 9858
    },
    {
      "epoch": 0.5520775002799866,
      "grad_norm": 1.1450551748275757,
      "learning_rate": 4.9265e-05,
      "loss": 0.3966,
      "step": 9859
    },
    {
      "epoch": 0.5521334975921156,
      "grad_norm": 1.177097201347351,
      "learning_rate": 4.927000000000001e-05,
      "loss": 0.3795,
      "step": 9860
    },
    {
      "epoch": 0.5521894949042446,
      "grad_norm": 1.2279947996139526,
      "learning_rate": 4.9275000000000005e-05,
      "loss": 0.4562,
      "step": 9861
    },
    {
      "epoch": 0.5522454922163736,
      "grad_norm": 1.361910104751587,
      "learning_rate": 4.928e-05,
      "loss": 0.5464,
      "step": 9862
    },
    {
      "epoch": 0.5523014895285027,
      "grad_norm": 1.1686800718307495,
      "learning_rate": 4.928500000000001e-05,
      "loss": 0.4628,
      "step": 9863
    },
    {
      "epoch": 0.5523574868406317,
      "grad_norm": 1.1429228782653809,
      "learning_rate": 4.9290000000000004e-05,
      "loss": 0.3713,
      "step": 9864
    },
    {
      "epoch": 0.5524134841527607,
      "grad_norm": 1.4145402908325195,
      "learning_rate": 4.9295e-05,
      "loss": 0.3984,
      "step": 9865
    },
    {
      "epoch": 0.5524694814648897,
      "grad_norm": 1.3317267894744873,
      "learning_rate": 4.93e-05,
      "loss": 0.5272,
      "step": 9866
    },
    {
      "epoch": 0.5525254787770187,
      "grad_norm": 1.6362569332122803,
      "learning_rate": 4.9305e-05,
      "loss": 0.5404,
      "step": 9867
    },
    {
      "epoch": 0.5525814760891478,
      "grad_norm": 1.2255516052246094,
      "learning_rate": 4.931e-05,
      "loss": 0.38,
      "step": 9868
    },
    {
      "epoch": 0.5526374734012768,
      "grad_norm": 1.2675533294677734,
      "learning_rate": 4.9315e-05,
      "loss": 0.4144,
      "step": 9869
    },
    {
      "epoch": 0.5526934707134058,
      "grad_norm": 48.612266540527344,
      "learning_rate": 4.932e-05,
      "loss": 0.3633,
      "step": 9870
    },
    {
      "epoch": 0.5527494680255348,
      "grad_norm": 1.4007257223129272,
      "learning_rate": 4.9325000000000006e-05,
      "loss": 0.3649,
      "step": 9871
    },
    {
      "epoch": 0.5528054653376638,
      "grad_norm": 1.2883875370025635,
      "learning_rate": 4.9330000000000004e-05,
      "loss": 0.455,
      "step": 9872
    },
    {
      "epoch": 0.5528614626497929,
      "grad_norm": 1.2115495204925537,
      "learning_rate": 4.9335e-05,
      "loss": 0.4835,
      "step": 9873
    },
    {
      "epoch": 0.5529174599619219,
      "grad_norm": 1.1614922285079956,
      "learning_rate": 4.9340000000000005e-05,
      "loss": 0.6316,
      "step": 9874
    },
    {
      "epoch": 0.5529734572740509,
      "grad_norm": 1.5327253341674805,
      "learning_rate": 4.9345e-05,
      "loss": 0.5755,
      "step": 9875
    },
    {
      "epoch": 0.5530294545861799,
      "grad_norm": 1.2586109638214111,
      "learning_rate": 4.935e-05,
      "loss": 0.5452,
      "step": 9876
    },
    {
      "epoch": 0.5530854518983089,
      "grad_norm": 1.5047032833099365,
      "learning_rate": 4.9355000000000004e-05,
      "loss": 0.5365,
      "step": 9877
    },
    {
      "epoch": 0.553141449210438,
      "grad_norm": 1.508359670639038,
      "learning_rate": 4.936e-05,
      "loss": 0.5892,
      "step": 9878
    },
    {
      "epoch": 0.553197446522567,
      "grad_norm": 1.305232286453247,
      "learning_rate": 4.9365e-05,
      "loss": 0.5255,
      "step": 9879
    },
    {
      "epoch": 0.5532534438346959,
      "grad_norm": 1.352198839187622,
      "learning_rate": 4.937e-05,
      "loss": 0.4257,
      "step": 9880
    },
    {
      "epoch": 0.5533094411468249,
      "grad_norm": 1.2626460790634155,
      "learning_rate": 4.937500000000001e-05,
      "loss": 0.4752,
      "step": 9881
    },
    {
      "epoch": 0.5533654384589539,
      "grad_norm": 1.0746902227401733,
      "learning_rate": 4.9380000000000005e-05,
      "loss": 0.3612,
      "step": 9882
    },
    {
      "epoch": 0.5534214357710829,
      "grad_norm": 1.0874592065811157,
      "learning_rate": 4.9385e-05,
      "loss": 0.4713,
      "step": 9883
    },
    {
      "epoch": 0.553477433083212,
      "grad_norm": 1.600517988204956,
      "learning_rate": 4.939e-05,
      "loss": 0.3904,
      "step": 9884
    },
    {
      "epoch": 0.553533430395341,
      "grad_norm": 1.0637706518173218,
      "learning_rate": 4.9395000000000004e-05,
      "loss": 0.3279,
      "step": 9885
    },
    {
      "epoch": 0.55358942770747,
      "grad_norm": 1.3719364404678345,
      "learning_rate": 4.94e-05,
      "loss": 0.5074,
      "step": 9886
    },
    {
      "epoch": 0.553645425019599,
      "grad_norm": 1.278235912322998,
      "learning_rate": 4.9405e-05,
      "loss": 0.4989,
      "step": 9887
    },
    {
      "epoch": 0.553701422331728,
      "grad_norm": 1.3625400066375732,
      "learning_rate": 4.941e-05,
      "loss": 0.4649,
      "step": 9888
    },
    {
      "epoch": 0.553757419643857,
      "grad_norm": 1.0363482236862183,
      "learning_rate": 4.9415e-05,
      "loss": 0.416,
      "step": 9889
    },
    {
      "epoch": 0.5538134169559861,
      "grad_norm": 1.31887948513031,
      "learning_rate": 4.942e-05,
      "loss": 0.4591,
      "step": 9890
    },
    {
      "epoch": 0.5538694142681151,
      "grad_norm": 1.1346615552902222,
      "learning_rate": 4.9425e-05,
      "loss": 0.3147,
      "step": 9891
    },
    {
      "epoch": 0.5539254115802441,
      "grad_norm": 1.951660394668579,
      "learning_rate": 4.9430000000000006e-05,
      "loss": 0.4978,
      "step": 9892
    },
    {
      "epoch": 0.5539814088923731,
      "grad_norm": 1.2742087841033936,
      "learning_rate": 4.9435000000000004e-05,
      "loss": 0.4216,
      "step": 9893
    },
    {
      "epoch": 0.5540374062045021,
      "grad_norm": 1.3981987237930298,
      "learning_rate": 4.944e-05,
      "loss": 0.3698,
      "step": 9894
    },
    {
      "epoch": 0.5540934035166312,
      "grad_norm": 1.213977336883545,
      "learning_rate": 4.9445000000000005e-05,
      "loss": 0.4999,
      "step": 9895
    },
    {
      "epoch": 0.5541494008287602,
      "grad_norm": 1.315979242324829,
      "learning_rate": 4.945e-05,
      "loss": 0.5324,
      "step": 9896
    },
    {
      "epoch": 0.5542053981408892,
      "grad_norm": 1.1904572248458862,
      "learning_rate": 4.9455e-05,
      "loss": 0.3628,
      "step": 9897
    },
    {
      "epoch": 0.5542613954530182,
      "grad_norm": 1.1681935787200928,
      "learning_rate": 4.946e-05,
      "loss": 0.3761,
      "step": 9898
    },
    {
      "epoch": 0.5543173927651472,
      "grad_norm": 1.256194829940796,
      "learning_rate": 4.9465e-05,
      "loss": 0.4615,
      "step": 9899
    },
    {
      "epoch": 0.5543733900772763,
      "grad_norm": 1.124131202697754,
      "learning_rate": 4.947e-05,
      "loss": 0.5317,
      "step": 9900
    },
    {
      "epoch": 0.5544293873894053,
      "grad_norm": 1.7268904447555542,
      "learning_rate": 4.9475e-05,
      "loss": 0.4717,
      "step": 9901
    },
    {
      "epoch": 0.5544853847015343,
      "grad_norm": 1.1848986148834229,
      "learning_rate": 4.948000000000001e-05,
      "loss": 0.4333,
      "step": 9902
    },
    {
      "epoch": 0.5545413820136633,
      "grad_norm": 6.534971714019775,
      "learning_rate": 4.9485000000000005e-05,
      "loss": 0.3291,
      "step": 9903
    },
    {
      "epoch": 0.5545973793257923,
      "grad_norm": 1.3286279439926147,
      "learning_rate": 4.949e-05,
      "loss": 0.4083,
      "step": 9904
    },
    {
      "epoch": 0.5546533766379214,
      "grad_norm": 1.2005430459976196,
      "learning_rate": 4.9495e-05,
      "loss": 0.4521,
      "step": 9905
    },
    {
      "epoch": 0.5547093739500504,
      "grad_norm": 1.1113883256912231,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 0.4342,
      "step": 9906
    },
    {
      "epoch": 0.5547653712621794,
      "grad_norm": 1.3328896760940552,
      "learning_rate": 4.9505e-05,
      "loss": 0.4356,
      "step": 9907
    },
    {
      "epoch": 0.5548213685743084,
      "grad_norm": 1.3005143404006958,
      "learning_rate": 4.951e-05,
      "loss": 0.5701,
      "step": 9908
    },
    {
      "epoch": 0.5548773658864374,
      "grad_norm": 1.0802335739135742,
      "learning_rate": 4.9515e-05,
      "loss": 0.4,
      "step": 9909
    },
    {
      "epoch": 0.5549333631985665,
      "grad_norm": 1.181265950202942,
      "learning_rate": 4.952e-05,
      "loss": 0.4187,
      "step": 9910
    },
    {
      "epoch": 0.5549893605106955,
      "grad_norm": 1.1380640268325806,
      "learning_rate": 4.9525000000000004e-05,
      "loss": 0.345,
      "step": 9911
    },
    {
      "epoch": 0.5550453578228245,
      "grad_norm": 1.468192458152771,
      "learning_rate": 4.953e-05,
      "loss": 0.5454,
      "step": 9912
    },
    {
      "epoch": 0.5551013551349535,
      "grad_norm": 1.2509827613830566,
      "learning_rate": 4.9535000000000006e-05,
      "loss": 0.3683,
      "step": 9913
    },
    {
      "epoch": 0.5551573524470825,
      "grad_norm": 1.237339735031128,
      "learning_rate": 4.9540000000000003e-05,
      "loss": 0.4945,
      "step": 9914
    },
    {
      "epoch": 0.5552133497592115,
      "grad_norm": 1.458763837814331,
      "learning_rate": 4.9545e-05,
      "loss": 0.4614,
      "step": 9915
    },
    {
      "epoch": 0.5552693470713406,
      "grad_norm": 1.4923385381698608,
      "learning_rate": 4.9550000000000005e-05,
      "loss": 0.4802,
      "step": 9916
    },
    {
      "epoch": 0.5553253443834696,
      "grad_norm": 1.1870142221450806,
      "learning_rate": 4.9555e-05,
      "loss": 0.3874,
      "step": 9917
    },
    {
      "epoch": 0.5553813416955986,
      "grad_norm": 1.1183218955993652,
      "learning_rate": 4.956e-05,
      "loss": 0.4729,
      "step": 9918
    },
    {
      "epoch": 0.5554373390077276,
      "grad_norm": 1.5191528797149658,
      "learning_rate": 4.9565e-05,
      "loss": 0.4362,
      "step": 9919
    },
    {
      "epoch": 0.5554933363198566,
      "grad_norm": 1.2282307147979736,
      "learning_rate": 4.957e-05,
      "loss": 0.4814,
      "step": 9920
    },
    {
      "epoch": 0.5555493336319857,
      "grad_norm": 1.3444799184799194,
      "learning_rate": 4.9575000000000006e-05,
      "loss": 0.4141,
      "step": 9921
    },
    {
      "epoch": 0.5556053309441147,
      "grad_norm": 1.3083947896957397,
      "learning_rate": 4.958e-05,
      "loss": 0.3594,
      "step": 9922
    },
    {
      "epoch": 0.5556613282562437,
      "grad_norm": 1.306146502494812,
      "learning_rate": 4.9585e-05,
      "loss": 0.3494,
      "step": 9923
    },
    {
      "epoch": 0.5557173255683727,
      "grad_norm": 1.2219531536102295,
      "learning_rate": 4.9590000000000005e-05,
      "loss": 0.3582,
      "step": 9924
    },
    {
      "epoch": 0.5557733228805017,
      "grad_norm": 1.2251049280166626,
      "learning_rate": 4.9595e-05,
      "loss": 0.3454,
      "step": 9925
    },
    {
      "epoch": 0.5558293201926308,
      "grad_norm": 1.1786918640136719,
      "learning_rate": 4.96e-05,
      "loss": 0.3274,
      "step": 9926
    },
    {
      "epoch": 0.5558853175047598,
      "grad_norm": 1.2297167778015137,
      "learning_rate": 4.9605000000000004e-05,
      "loss": 0.415,
      "step": 9927
    },
    {
      "epoch": 0.5559413148168888,
      "grad_norm": 1.1262352466583252,
      "learning_rate": 4.961e-05,
      "loss": 0.3766,
      "step": 9928
    },
    {
      "epoch": 0.5559973121290178,
      "grad_norm": 1.4815571308135986,
      "learning_rate": 4.9615e-05,
      "loss": 0.4644,
      "step": 9929
    },
    {
      "epoch": 0.5560533094411468,
      "grad_norm": 1.1427712440490723,
      "learning_rate": 4.962e-05,
      "loss": 0.3341,
      "step": 9930
    },
    {
      "epoch": 0.5561093067532759,
      "grad_norm": 1.142407774925232,
      "learning_rate": 4.962500000000001e-05,
      "loss": 0.3281,
      "step": 9931
    },
    {
      "epoch": 0.5561653040654049,
      "grad_norm": 1.2612287998199463,
      "learning_rate": 4.9630000000000004e-05,
      "loss": 0.5076,
      "step": 9932
    },
    {
      "epoch": 0.5562213013775339,
      "grad_norm": 1.4052684307098389,
      "learning_rate": 4.9635e-05,
      "loss": 0.3742,
      "step": 9933
    },
    {
      "epoch": 0.5562772986896629,
      "grad_norm": 1.068899393081665,
      "learning_rate": 4.9640000000000006e-05,
      "loss": 0.3504,
      "step": 9934
    },
    {
      "epoch": 0.5563332960017919,
      "grad_norm": 1.4374948740005493,
      "learning_rate": 4.9645e-05,
      "loss": 0.7201,
      "step": 9935
    },
    {
      "epoch": 0.556389293313921,
      "grad_norm": 1.3756743669509888,
      "learning_rate": 4.965e-05,
      "loss": 0.4793,
      "step": 9936
    },
    {
      "epoch": 0.55644529062605,
      "grad_norm": 1.3166941404342651,
      "learning_rate": 4.9655000000000005e-05,
      "loss": 0.3156,
      "step": 9937
    },
    {
      "epoch": 0.556501287938179,
      "grad_norm": 1.1218551397323608,
      "learning_rate": 4.966e-05,
      "loss": 0.4509,
      "step": 9938
    },
    {
      "epoch": 0.556557285250308,
      "grad_norm": 1.1736061573028564,
      "learning_rate": 4.9665e-05,
      "loss": 0.3869,
      "step": 9939
    },
    {
      "epoch": 0.556613282562437,
      "grad_norm": 1.6972935199737549,
      "learning_rate": 4.967e-05,
      "loss": 0.5223,
      "step": 9940
    },
    {
      "epoch": 0.556669279874566,
      "grad_norm": 1.475999116897583,
      "learning_rate": 4.967500000000001e-05,
      "loss": 0.5346,
      "step": 9941
    },
    {
      "epoch": 0.5567252771866951,
      "grad_norm": 1.411203145980835,
      "learning_rate": 4.9680000000000005e-05,
      "loss": 0.5327,
      "step": 9942
    },
    {
      "epoch": 0.5567812744988241,
      "grad_norm": 1.2205110788345337,
      "learning_rate": 4.9685e-05,
      "loss": 0.4637,
      "step": 9943
    },
    {
      "epoch": 0.5568372718109531,
      "grad_norm": 1.2715861797332764,
      "learning_rate": 4.969e-05,
      "loss": 0.5965,
      "step": 9944
    },
    {
      "epoch": 0.5568932691230821,
      "grad_norm": 1.3103289604187012,
      "learning_rate": 4.9695000000000004e-05,
      "loss": 0.4397,
      "step": 9945
    },
    {
      "epoch": 0.5569492664352111,
      "grad_norm": 1.4818700551986694,
      "learning_rate": 4.97e-05,
      "loss": 0.562,
      "step": 9946
    },
    {
      "epoch": 0.5570052637473402,
      "grad_norm": 1.9872112274169922,
      "learning_rate": 4.9705e-05,
      "loss": 0.4474,
      "step": 9947
    },
    {
      "epoch": 0.5570612610594692,
      "grad_norm": 1.3563531637191772,
      "learning_rate": 4.9710000000000003e-05,
      "loss": 0.3684,
      "step": 9948
    },
    {
      "epoch": 0.5571172583715982,
      "grad_norm": 1.163193702697754,
      "learning_rate": 4.9715e-05,
      "loss": 0.3716,
      "step": 9949
    },
    {
      "epoch": 0.5571732556837272,
      "grad_norm": 0.9077993631362915,
      "learning_rate": 4.972e-05,
      "loss": 0.3329,
      "step": 9950
    },
    {
      "epoch": 0.5572292529958562,
      "grad_norm": 1.0345548391342163,
      "learning_rate": 4.9725e-05,
      "loss": 0.3246,
      "step": 9951
    },
    {
      "epoch": 0.5572852503079853,
      "grad_norm": 1.3332639932632446,
      "learning_rate": 4.973000000000001e-05,
      "loss": 0.3712,
      "step": 9952
    },
    {
      "epoch": 0.5573412476201143,
      "grad_norm": 1.148186445236206,
      "learning_rate": 4.9735000000000004e-05,
      "loss": 0.3677,
      "step": 9953
    },
    {
      "epoch": 0.5573972449322433,
      "grad_norm": 1.2874257564544678,
      "learning_rate": 4.974e-05,
      "loss": 0.4196,
      "step": 9954
    },
    {
      "epoch": 0.5574532422443723,
      "grad_norm": 1.3812764883041382,
      "learning_rate": 4.9745000000000006e-05,
      "loss": 0.505,
      "step": 9955
    },
    {
      "epoch": 0.5575092395565013,
      "grad_norm": 1.211657166481018,
      "learning_rate": 4.975e-05,
      "loss": 0.3548,
      "step": 9956
    },
    {
      "epoch": 0.5575652368686304,
      "grad_norm": 2.642127513885498,
      "learning_rate": 4.9755e-05,
      "loss": 0.4774,
      "step": 9957
    },
    {
      "epoch": 0.5576212341807594,
      "grad_norm": 1.1937086582183838,
      "learning_rate": 4.976e-05,
      "loss": 0.3494,
      "step": 9958
    },
    {
      "epoch": 0.5576772314928884,
      "grad_norm": 1.290429711341858,
      "learning_rate": 4.9765e-05,
      "loss": 0.4232,
      "step": 9959
    },
    {
      "epoch": 0.5577332288050174,
      "grad_norm": 1.9520561695098877,
      "learning_rate": 4.977e-05,
      "loss": 0.5164,
      "step": 9960
    },
    {
      "epoch": 0.5577892261171464,
      "grad_norm": 1.2326164245605469,
      "learning_rate": 4.9775000000000004e-05,
      "loss": 0.4576,
      "step": 9961
    },
    {
      "epoch": 0.5578452234292754,
      "grad_norm": 1.1097726821899414,
      "learning_rate": 4.978e-05,
      "loss": 0.4239,
      "step": 9962
    },
    {
      "epoch": 0.5579012207414044,
      "grad_norm": 1.5771677494049072,
      "learning_rate": 4.9785000000000005e-05,
      "loss": 0.3854,
      "step": 9963
    },
    {
      "epoch": 0.5579572180535334,
      "grad_norm": 1.1490870714187622,
      "learning_rate": 4.979e-05,
      "loss": 0.3786,
      "step": 9964
    },
    {
      "epoch": 0.5580132153656624,
      "grad_norm": 1.0461891889572144,
      "learning_rate": 4.9795e-05,
      "loss": 0.3223,
      "step": 9965
    },
    {
      "epoch": 0.5580692126777914,
      "grad_norm": 1.3650190830230713,
      "learning_rate": 4.9800000000000004e-05,
      "loss": 0.492,
      "step": 9966
    },
    {
      "epoch": 0.5581252099899204,
      "grad_norm": 1.690585970878601,
      "learning_rate": 4.9805e-05,
      "loss": 0.5052,
      "step": 9967
    },
    {
      "epoch": 0.5581812073020495,
      "grad_norm": 1.3042556047439575,
      "learning_rate": 4.981e-05,
      "loss": 0.4262,
      "step": 9968
    },
    {
      "epoch": 0.5582372046141785,
      "grad_norm": 1.3416763544082642,
      "learning_rate": 4.9815e-05,
      "loss": 0.4576,
      "step": 9969
    },
    {
      "epoch": 0.5582932019263075,
      "grad_norm": 1.3321274518966675,
      "learning_rate": 4.982e-05,
      "loss": 0.4248,
      "step": 9970
    },
    {
      "epoch": 0.5583491992384365,
      "grad_norm": 1.4489244222640991,
      "learning_rate": 4.9825000000000005e-05,
      "loss": 0.5264,
      "step": 9971
    },
    {
      "epoch": 0.5584051965505655,
      "grad_norm": 1.2635912895202637,
      "learning_rate": 4.983e-05,
      "loss": 0.4988,
      "step": 9972
    },
    {
      "epoch": 0.5584611938626945,
      "grad_norm": 1.3435423374176025,
      "learning_rate": 4.9835000000000007e-05,
      "loss": 0.5256,
      "step": 9973
    },
    {
      "epoch": 0.5585171911748236,
      "grad_norm": 1.4019769430160522,
      "learning_rate": 4.9840000000000004e-05,
      "loss": 0.4462,
      "step": 9974
    },
    {
      "epoch": 0.5585731884869526,
      "grad_norm": 1.345577597618103,
      "learning_rate": 4.9845e-05,
      "loss": 0.3777,
      "step": 9975
    },
    {
      "epoch": 0.5586291857990816,
      "grad_norm": 1.290582299232483,
      "learning_rate": 4.9850000000000006e-05,
      "loss": 0.3645,
      "step": 9976
    },
    {
      "epoch": 0.5586851831112106,
      "grad_norm": 1.3481178283691406,
      "learning_rate": 4.9855e-05,
      "loss": 0.5196,
      "step": 9977
    },
    {
      "epoch": 0.5587411804233396,
      "grad_norm": 1.1358520984649658,
      "learning_rate": 4.986e-05,
      "loss": 0.383,
      "step": 9978
    },
    {
      "epoch": 0.5587971777354687,
      "grad_norm": 1.4544248580932617,
      "learning_rate": 4.9865e-05,
      "loss": 0.6048,
      "step": 9979
    },
    {
      "epoch": 0.5588531750475977,
      "grad_norm": 1.416058897972107,
      "learning_rate": 4.987e-05,
      "loss": 0.4876,
      "step": 9980
    },
    {
      "epoch": 0.5589091723597267,
      "grad_norm": 1.3810850381851196,
      "learning_rate": 4.9875000000000006e-05,
      "loss": 0.4473,
      "step": 9981
    },
    {
      "epoch": 0.5589651696718557,
      "grad_norm": 1.3577407598495483,
      "learning_rate": 4.9880000000000004e-05,
      "loss": 0.5043,
      "step": 9982
    },
    {
      "epoch": 0.5590211669839847,
      "grad_norm": 1.1616915464401245,
      "learning_rate": 4.9885e-05,
      "loss": 0.4671,
      "step": 9983
    },
    {
      "epoch": 0.5590771642961138,
      "grad_norm": 1.6595820188522339,
      "learning_rate": 4.9890000000000005e-05,
      "loss": 0.6291,
      "step": 9984
    },
    {
      "epoch": 0.5591331616082428,
      "grad_norm": 1.1977049112319946,
      "learning_rate": 4.9895e-05,
      "loss": 0.4237,
      "step": 9985
    },
    {
      "epoch": 0.5591891589203718,
      "grad_norm": 1.2889515161514282,
      "learning_rate": 4.99e-05,
      "loss": 0.4081,
      "step": 9986
    },
    {
      "epoch": 0.5592451562325008,
      "grad_norm": 1.3410112857818604,
      "learning_rate": 4.9905000000000004e-05,
      "loss": 0.414,
      "step": 9987
    },
    {
      "epoch": 0.5593011535446298,
      "grad_norm": 1.8319560289382935,
      "learning_rate": 4.991e-05,
      "loss": 0.5875,
      "step": 9988
    },
    {
      "epoch": 0.5593571508567589,
      "grad_norm": 1.1550171375274658,
      "learning_rate": 4.9915e-05,
      "loss": 0.3828,
      "step": 9989
    },
    {
      "epoch": 0.5594131481688879,
      "grad_norm": 1.2857604026794434,
      "learning_rate": 4.992e-05,
      "loss": 0.4118,
      "step": 9990
    },
    {
      "epoch": 0.5594691454810169,
      "grad_norm": 1.6735023260116577,
      "learning_rate": 4.992500000000001e-05,
      "loss": 0.5153,
      "step": 9991
    },
    {
      "epoch": 0.5595251427931459,
      "grad_norm": 1.284501075744629,
      "learning_rate": 4.9930000000000005e-05,
      "loss": 0.436,
      "step": 9992
    },
    {
      "epoch": 0.5595811401052749,
      "grad_norm": 1.1906112432479858,
      "learning_rate": 4.9935e-05,
      "loss": 0.4321,
      "step": 9993
    },
    {
      "epoch": 0.559637137417404,
      "grad_norm": 1.2857753038406372,
      "learning_rate": 4.9940000000000006e-05,
      "loss": 0.3267,
      "step": 9994
    },
    {
      "epoch": 0.559693134729533,
      "grad_norm": 1.9837729930877686,
      "learning_rate": 4.9945000000000004e-05,
      "loss": 0.4377,
      "step": 9995
    },
    {
      "epoch": 0.559749132041662,
      "grad_norm": 1.4722764492034912,
      "learning_rate": 4.995e-05,
      "loss": 0.5115,
      "step": 9996
    },
    {
      "epoch": 0.559805129353791,
      "grad_norm": 1.1309281587600708,
      "learning_rate": 4.9955e-05,
      "loss": 0.4,
      "step": 9997
    },
    {
      "epoch": 0.55986112666592,
      "grad_norm": 1.302437663078308,
      "learning_rate": 4.996e-05,
      "loss": 0.4132,
      "step": 9998
    },
    {
      "epoch": 0.559917123978049,
      "grad_norm": 1.5880764722824097,
      "learning_rate": 4.9965e-05,
      "loss": 0.6104,
      "step": 9999
    },
    {
      "epoch": 0.5599731212901781,
      "grad_norm": 1.1306363344192505,
      "learning_rate": 4.997e-05,
      "loss": 0.3663,
      "step": 10000
    },
    {
      "epoch": 0.5600291186023071,
      "grad_norm": 1.2174371480941772,
      "learning_rate": 4.9975e-05,
      "loss": 0.337,
      "step": 10001
    },
    {
      "epoch": 0.5600851159144361,
      "grad_norm": 1.394925832748413,
      "learning_rate": 4.9980000000000006e-05,
      "loss": 0.6143,
      "step": 10002
    },
    {
      "epoch": 0.5601411132265651,
      "grad_norm": 1.1916224956512451,
      "learning_rate": 4.9985e-05,
      "loss": 0.3698,
      "step": 10003
    },
    {
      "epoch": 0.5601971105386941,
      "grad_norm": 1.4514687061309814,
      "learning_rate": 4.999e-05,
      "loss": 0.4462,
      "step": 10004
    },
    {
      "epoch": 0.5602531078508232,
      "grad_norm": 1.3292044401168823,
      "learning_rate": 4.9995000000000005e-05,
      "loss": 0.4409,
      "step": 10005
    },
    {
      "epoch": 0.5603091051629522,
      "grad_norm": 1.328014850616455,
      "learning_rate": 5e-05,
      "loss": 0.4157,
      "step": 10006
    },
    {
      "epoch": 0.5603651024750812,
      "grad_norm": 1.4254311323165894,
      "learning_rate": 5.0005e-05,
      "loss": 0.5557,
      "step": 10007
    },
    {
      "epoch": 0.5604210997872102,
      "grad_norm": 1.393754482269287,
      "learning_rate": 5.0010000000000004e-05,
      "loss": 0.5511,
      "step": 10008
    },
    {
      "epoch": 0.5604770970993392,
      "grad_norm": 1.2166305780410767,
      "learning_rate": 5.0015e-05,
      "loss": 0.4034,
      "step": 10009
    },
    {
      "epoch": 0.5605330944114683,
      "grad_norm": 1.2587909698486328,
      "learning_rate": 5.002e-05,
      "loss": 0.4431,
      "step": 10010
    },
    {
      "epoch": 0.5605890917235973,
      "grad_norm": 1.1939144134521484,
      "learning_rate": 5.0025e-05,
      "loss": 0.4778,
      "step": 10011
    },
    {
      "epoch": 0.5606450890357263,
      "grad_norm": 1.3817065954208374,
      "learning_rate": 5.003e-05,
      "loss": 0.487,
      "step": 10012
    },
    {
      "epoch": 0.5607010863478553,
      "grad_norm": 1.108214020729065,
      "learning_rate": 5.0035e-05,
      "loss": 0.456,
      "step": 10013
    },
    {
      "epoch": 0.5607570836599843,
      "grad_norm": 1.1789519786834717,
      "learning_rate": 5.0039999999999995e-05,
      "loss": 0.4451,
      "step": 10014
    },
    {
      "epoch": 0.5608130809721134,
      "grad_norm": 1.1562879085540771,
      "learning_rate": 5.0045e-05,
      "loss": 0.3838,
      "step": 10015
    },
    {
      "epoch": 0.5608690782842424,
      "grad_norm": 1.337884783744812,
      "learning_rate": 5.005e-05,
      "loss": 0.6035,
      "step": 10016
    },
    {
      "epoch": 0.5609250755963714,
      "grad_norm": 1.4621641635894775,
      "learning_rate": 5.005500000000001e-05,
      "loss": 0.4793,
      "step": 10017
    },
    {
      "epoch": 0.5609810729085004,
      "grad_norm": 1.657607913017273,
      "learning_rate": 5.0060000000000005e-05,
      "loss": 0.3936,
      "step": 10018
    },
    {
      "epoch": 0.5610370702206294,
      "grad_norm": 1.2850358486175537,
      "learning_rate": 5.006500000000001e-05,
      "loss": 0.3666,
      "step": 10019
    },
    {
      "epoch": 0.5610930675327584,
      "grad_norm": 1.2727241516113281,
      "learning_rate": 5.007000000000001e-05,
      "loss": 0.5122,
      "step": 10020
    },
    {
      "epoch": 0.5611490648448875,
      "grad_norm": 1.912054181098938,
      "learning_rate": 5.0075000000000004e-05,
      "loss": 0.3845,
      "step": 10021
    },
    {
      "epoch": 0.5612050621570165,
      "grad_norm": 1.3623539209365845,
      "learning_rate": 5.008e-05,
      "loss": 0.4237,
      "step": 10022
    },
    {
      "epoch": 0.5612610594691455,
      "grad_norm": 1.307369589805603,
      "learning_rate": 5.0085000000000006e-05,
      "loss": 0.523,
      "step": 10023
    },
    {
      "epoch": 0.5613170567812745,
      "grad_norm": 1.2519283294677734,
      "learning_rate": 5.009e-05,
      "loss": 0.5827,
      "step": 10024
    },
    {
      "epoch": 0.5613730540934035,
      "grad_norm": 1.2211045026779175,
      "learning_rate": 5.0095e-05,
      "loss": 0.5273,
      "step": 10025
    },
    {
      "epoch": 0.5614290514055326,
      "grad_norm": 1.12700355052948,
      "learning_rate": 5.0100000000000005e-05,
      "loss": 0.3769,
      "step": 10026
    },
    {
      "epoch": 0.5614850487176616,
      "grad_norm": 1.1931533813476562,
      "learning_rate": 5.0105e-05,
      "loss": 0.4218,
      "step": 10027
    },
    {
      "epoch": 0.5615410460297906,
      "grad_norm": 1.2497025728225708,
      "learning_rate": 5.011e-05,
      "loss": 0.3069,
      "step": 10028
    },
    {
      "epoch": 0.5615970433419196,
      "grad_norm": 1.37238609790802,
      "learning_rate": 5.0115000000000004e-05,
      "loss": 0.4876,
      "step": 10029
    },
    {
      "epoch": 0.5616530406540486,
      "grad_norm": 1.170186996459961,
      "learning_rate": 5.012e-05,
      "loss": 0.3934,
      "step": 10030
    },
    {
      "epoch": 0.5617090379661777,
      "grad_norm": 1.1139869689941406,
      "learning_rate": 5.0125e-05,
      "loss": 0.3468,
      "step": 10031
    },
    {
      "epoch": 0.5617650352783067,
      "grad_norm": 1.4631116390228271,
      "learning_rate": 5.0129999999999996e-05,
      "loss": 0.6199,
      "step": 10032
    },
    {
      "epoch": 0.5618210325904357,
      "grad_norm": 1.2753750085830688,
      "learning_rate": 5.0135e-05,
      "loss": 0.4501,
      "step": 10033
    },
    {
      "epoch": 0.5618770299025647,
      "grad_norm": 1.632766604423523,
      "learning_rate": 5.014e-05,
      "loss": 0.4904,
      "step": 10034
    },
    {
      "epoch": 0.5619330272146937,
      "grad_norm": 1.181941270828247,
      "learning_rate": 5.0144999999999995e-05,
      "loss": 0.4378,
      "step": 10035
    },
    {
      "epoch": 0.5619890245268228,
      "grad_norm": 1.1032267808914185,
      "learning_rate": 5.015e-05,
      "loss": 0.3056,
      "step": 10036
    },
    {
      "epoch": 0.5620450218389518,
      "grad_norm": 1.1124083995819092,
      "learning_rate": 5.015500000000001e-05,
      "loss": 0.4144,
      "step": 10037
    },
    {
      "epoch": 0.5621010191510808,
      "grad_norm": 1.645929217338562,
      "learning_rate": 5.016000000000001e-05,
      "loss": 0.4506,
      "step": 10038
    },
    {
      "epoch": 0.5621570164632098,
      "grad_norm": 1.1254245042800903,
      "learning_rate": 5.0165000000000005e-05,
      "loss": 0.4107,
      "step": 10039
    },
    {
      "epoch": 0.5622130137753388,
      "grad_norm": 1.5527163743972778,
      "learning_rate": 5.017000000000001e-05,
      "loss": 0.5131,
      "step": 10040
    },
    {
      "epoch": 0.5622690110874679,
      "grad_norm": 1.2213736772537231,
      "learning_rate": 5.017500000000001e-05,
      "loss": 0.3638,
      "step": 10041
    },
    {
      "epoch": 0.5623250083995969,
      "grad_norm": 1.2467491626739502,
      "learning_rate": 5.0180000000000004e-05,
      "loss": 0.4783,
      "step": 10042
    },
    {
      "epoch": 0.5623810057117259,
      "grad_norm": 1.4593853950500488,
      "learning_rate": 5.0185e-05,
      "loss": 0.525,
      "step": 10043
    },
    {
      "epoch": 0.5624370030238549,
      "grad_norm": 1.1642142534255981,
      "learning_rate": 5.0190000000000006e-05,
      "loss": 0.4755,
      "step": 10044
    },
    {
      "epoch": 0.5624930003359839,
      "grad_norm": 1.3840899467468262,
      "learning_rate": 5.0195e-05,
      "loss": 0.4413,
      "step": 10045
    },
    {
      "epoch": 0.5625489976481128,
      "grad_norm": 1.1163687705993652,
      "learning_rate": 5.02e-05,
      "loss": 0.3448,
      "step": 10046
    },
    {
      "epoch": 0.5626049949602419,
      "grad_norm": 1.3278331756591797,
      "learning_rate": 5.0205000000000005e-05,
      "loss": 0.397,
      "step": 10047
    },
    {
      "epoch": 0.5626609922723709,
      "grad_norm": 1.5409855842590332,
      "learning_rate": 5.021e-05,
      "loss": 0.4767,
      "step": 10048
    },
    {
      "epoch": 0.5627169895844999,
      "grad_norm": 1.2589179277420044,
      "learning_rate": 5.0215e-05,
      "loss": 0.486,
      "step": 10049
    },
    {
      "epoch": 0.5627729868966289,
      "grad_norm": 1.3704774379730225,
      "learning_rate": 5.0220000000000004e-05,
      "loss": 0.514,
      "step": 10050
    },
    {
      "epoch": 0.5628289842087579,
      "grad_norm": 1.185131311416626,
      "learning_rate": 5.0225e-05,
      "loss": 0.5717,
      "step": 10051
    },
    {
      "epoch": 0.562884981520887,
      "grad_norm": 1.5050828456878662,
      "learning_rate": 5.023e-05,
      "loss": 0.4966,
      "step": 10052
    },
    {
      "epoch": 0.562940978833016,
      "grad_norm": 1.078010082244873,
      "learning_rate": 5.0234999999999996e-05,
      "loss": 0.3227,
      "step": 10053
    },
    {
      "epoch": 0.562996976145145,
      "grad_norm": 1.184657335281372,
      "learning_rate": 5.024e-05,
      "loss": 0.3177,
      "step": 10054
    },
    {
      "epoch": 0.563052973457274,
      "grad_norm": 1.322494626045227,
      "learning_rate": 5.0245e-05,
      "loss": 0.4325,
      "step": 10055
    },
    {
      "epoch": 0.563108970769403,
      "grad_norm": 1.3813997507095337,
      "learning_rate": 5.0249999999999995e-05,
      "loss": 0.586,
      "step": 10056
    },
    {
      "epoch": 0.563164968081532,
      "grad_norm": 1.1439642906188965,
      "learning_rate": 5.0255000000000006e-05,
      "loss": 0.4306,
      "step": 10057
    },
    {
      "epoch": 0.5632209653936611,
      "grad_norm": 1.2778465747833252,
      "learning_rate": 5.026000000000001e-05,
      "loss": 0.5442,
      "step": 10058
    },
    {
      "epoch": 0.5632769627057901,
      "grad_norm": 1.2710590362548828,
      "learning_rate": 5.026500000000001e-05,
      "loss": 0.5092,
      "step": 10059
    },
    {
      "epoch": 0.5633329600179191,
      "grad_norm": 1.255285382270813,
      "learning_rate": 5.0270000000000005e-05,
      "loss": 0.4325,
      "step": 10060
    },
    {
      "epoch": 0.5633889573300481,
      "grad_norm": 1.280313491821289,
      "learning_rate": 5.0275e-05,
      "loss": 0.4029,
      "step": 10061
    },
    {
      "epoch": 0.5634449546421771,
      "grad_norm": 1.267838478088379,
      "learning_rate": 5.0280000000000006e-05,
      "loss": 0.4354,
      "step": 10062
    },
    {
      "epoch": 0.5635009519543062,
      "grad_norm": 1.5349767208099365,
      "learning_rate": 5.0285000000000004e-05,
      "loss": 0.3912,
      "step": 10063
    },
    {
      "epoch": 0.5635569492664352,
      "grad_norm": 1.314571738243103,
      "learning_rate": 5.029e-05,
      "loss": 0.4128,
      "step": 10064
    },
    {
      "epoch": 0.5636129465785642,
      "grad_norm": 1.4526143074035645,
      "learning_rate": 5.0295000000000006e-05,
      "loss": 0.51,
      "step": 10065
    },
    {
      "epoch": 0.5636689438906932,
      "grad_norm": 1.4880921840667725,
      "learning_rate": 5.03e-05,
      "loss": 0.5507,
      "step": 10066
    },
    {
      "epoch": 0.5637249412028222,
      "grad_norm": 1.6427276134490967,
      "learning_rate": 5.0305e-05,
      "loss": 0.5428,
      "step": 10067
    },
    {
      "epoch": 0.5637809385149513,
      "grad_norm": 1.3415950536727905,
      "learning_rate": 5.0310000000000005e-05,
      "loss": 0.4392,
      "step": 10068
    },
    {
      "epoch": 0.5638369358270803,
      "grad_norm": 1.6944310665130615,
      "learning_rate": 5.0315e-05,
      "loss": 0.2834,
      "step": 10069
    },
    {
      "epoch": 0.5638929331392093,
      "grad_norm": 1.3977125883102417,
      "learning_rate": 5.032e-05,
      "loss": 0.4614,
      "step": 10070
    },
    {
      "epoch": 0.5639489304513383,
      "grad_norm": 1.3059650659561157,
      "learning_rate": 5.0325e-05,
      "loss": 0.4491,
      "step": 10071
    },
    {
      "epoch": 0.5640049277634673,
      "grad_norm": 1.402970790863037,
      "learning_rate": 5.033e-05,
      "loss": 0.622,
      "step": 10072
    },
    {
      "epoch": 0.5640609250755964,
      "grad_norm": 1.445788860321045,
      "learning_rate": 5.0335e-05,
      "loss": 0.558,
      "step": 10073
    },
    {
      "epoch": 0.5641169223877254,
      "grad_norm": 1.7032382488250732,
      "learning_rate": 5.0339999999999996e-05,
      "loss": 0.443,
      "step": 10074
    },
    {
      "epoch": 0.5641729196998544,
      "grad_norm": 1.2501990795135498,
      "learning_rate": 5.0345e-05,
      "loss": 0.4593,
      "step": 10075
    },
    {
      "epoch": 0.5642289170119834,
      "grad_norm": 1.5817010402679443,
      "learning_rate": 5.035e-05,
      "loss": 0.4985,
      "step": 10076
    },
    {
      "epoch": 0.5642849143241124,
      "grad_norm": 1.1269781589508057,
      "learning_rate": 5.035500000000001e-05,
      "loss": 0.4437,
      "step": 10077
    },
    {
      "epoch": 0.5643409116362414,
      "grad_norm": 1.16768479347229,
      "learning_rate": 5.0360000000000006e-05,
      "loss": 0.4073,
      "step": 10078
    },
    {
      "epoch": 0.5643969089483705,
      "grad_norm": 1.4160252809524536,
      "learning_rate": 5.036500000000001e-05,
      "loss": 0.4502,
      "step": 10079
    },
    {
      "epoch": 0.5644529062604995,
      "grad_norm": 1.240113615989685,
      "learning_rate": 5.037000000000001e-05,
      "loss": 0.4357,
      "step": 10080
    },
    {
      "epoch": 0.5645089035726285,
      "grad_norm": 1.3157459497451782,
      "learning_rate": 5.0375000000000005e-05,
      "loss": 0.4175,
      "step": 10081
    },
    {
      "epoch": 0.5645649008847575,
      "grad_norm": 1.5507278442382812,
      "learning_rate": 5.038e-05,
      "loss": 0.3846,
      "step": 10082
    },
    {
      "epoch": 0.5646208981968865,
      "grad_norm": 1.2351633310317993,
      "learning_rate": 5.0385000000000006e-05,
      "loss": 0.4688,
      "step": 10083
    },
    {
      "epoch": 0.5646768955090156,
      "grad_norm": 1.5217021703720093,
      "learning_rate": 5.0390000000000004e-05,
      "loss": 0.424,
      "step": 10084
    },
    {
      "epoch": 0.5647328928211446,
      "grad_norm": 1.1799594163894653,
      "learning_rate": 5.0395e-05,
      "loss": 0.3173,
      "step": 10085
    },
    {
      "epoch": 0.5647888901332736,
      "grad_norm": 1.218217372894287,
      "learning_rate": 5.0400000000000005e-05,
      "loss": 0.5611,
      "step": 10086
    },
    {
      "epoch": 0.5648448874454026,
      "grad_norm": 1.0080746412277222,
      "learning_rate": 5.0405e-05,
      "loss": 0.3904,
      "step": 10087
    },
    {
      "epoch": 0.5649008847575316,
      "grad_norm": 1.2529783248901367,
      "learning_rate": 5.041e-05,
      "loss": 0.3862,
      "step": 10088
    },
    {
      "epoch": 0.5649568820696607,
      "grad_norm": 1.1760334968566895,
      "learning_rate": 5.0415000000000004e-05,
      "loss": 0.4188,
      "step": 10089
    },
    {
      "epoch": 0.5650128793817897,
      "grad_norm": 1.174148440361023,
      "learning_rate": 5.042e-05,
      "loss": 0.456,
      "step": 10090
    },
    {
      "epoch": 0.5650688766939187,
      "grad_norm": 2.8847005367279053,
      "learning_rate": 5.0425e-05,
      "loss": 0.4242,
      "step": 10091
    },
    {
      "epoch": 0.5651248740060477,
      "grad_norm": 1.396600365638733,
      "learning_rate": 5.0429999999999997e-05,
      "loss": 0.3525,
      "step": 10092
    },
    {
      "epoch": 0.5651808713181767,
      "grad_norm": 2.1680355072021484,
      "learning_rate": 5.0435e-05,
      "loss": 0.4377,
      "step": 10093
    },
    {
      "epoch": 0.5652368686303058,
      "grad_norm": 1.2360775470733643,
      "learning_rate": 5.044e-05,
      "loss": 0.4383,
      "step": 10094
    },
    {
      "epoch": 0.5652928659424348,
      "grad_norm": 1.3207876682281494,
      "learning_rate": 5.0444999999999996e-05,
      "loss": 0.4131,
      "step": 10095
    },
    {
      "epoch": 0.5653488632545638,
      "grad_norm": 1.5699901580810547,
      "learning_rate": 5.045e-05,
      "loss": 0.3841,
      "step": 10096
    },
    {
      "epoch": 0.5654048605666928,
      "grad_norm": 1.070851445198059,
      "learning_rate": 5.045500000000001e-05,
      "loss": 0.2873,
      "step": 10097
    },
    {
      "epoch": 0.5654608578788218,
      "grad_norm": 1.6041005849838257,
      "learning_rate": 5.046000000000001e-05,
      "loss": 0.457,
      "step": 10098
    },
    {
      "epoch": 0.5655168551909509,
      "grad_norm": 1.3358922004699707,
      "learning_rate": 5.0465000000000006e-05,
      "loss": 0.395,
      "step": 10099
    },
    {
      "epoch": 0.5655728525030799,
      "grad_norm": 1.224945306777954,
      "learning_rate": 5.047e-05,
      "loss": 0.5468,
      "step": 10100
    },
    {
      "epoch": 0.5656288498152089,
      "grad_norm": 1.5300770998001099,
      "learning_rate": 5.047500000000001e-05,
      "loss": 0.4921,
      "step": 10101
    },
    {
      "epoch": 0.5656848471273379,
      "grad_norm": 1.1021238565444946,
      "learning_rate": 5.0480000000000005e-05,
      "loss": 0.456,
      "step": 10102
    },
    {
      "epoch": 0.5657408444394669,
      "grad_norm": 1.3550925254821777,
      "learning_rate": 5.0485e-05,
      "loss": 0.5868,
      "step": 10103
    },
    {
      "epoch": 0.565796841751596,
      "grad_norm": 1.0332305431365967,
      "learning_rate": 5.0490000000000006e-05,
      "loss": 0.3685,
      "step": 10104
    },
    {
      "epoch": 0.565852839063725,
      "grad_norm": 1.2964366674423218,
      "learning_rate": 5.0495000000000004e-05,
      "loss": 0.621,
      "step": 10105
    },
    {
      "epoch": 0.565908836375854,
      "grad_norm": 1.293464183807373,
      "learning_rate": 5.05e-05,
      "loss": 0.4659,
      "step": 10106
    },
    {
      "epoch": 0.565964833687983,
      "grad_norm": 1.5895806550979614,
      "learning_rate": 5.0505000000000005e-05,
      "loss": 0.454,
      "step": 10107
    },
    {
      "epoch": 0.566020831000112,
      "grad_norm": 1.4051517248153687,
      "learning_rate": 5.051e-05,
      "loss": 0.3849,
      "step": 10108
    },
    {
      "epoch": 0.566076828312241,
      "grad_norm": 1.385178565979004,
      "learning_rate": 5.0515e-05,
      "loss": 0.4506,
      "step": 10109
    },
    {
      "epoch": 0.5661328256243701,
      "grad_norm": 1.2554985284805298,
      "learning_rate": 5.052e-05,
      "loss": 0.524,
      "step": 10110
    },
    {
      "epoch": 0.5661888229364991,
      "grad_norm": 1.3433587551116943,
      "learning_rate": 5.0525e-05,
      "loss": 0.4234,
      "step": 10111
    },
    {
      "epoch": 0.5662448202486281,
      "grad_norm": 1.5164456367492676,
      "learning_rate": 5.053e-05,
      "loss": 0.481,
      "step": 10112
    },
    {
      "epoch": 0.5663008175607571,
      "grad_norm": 1.239671230316162,
      "learning_rate": 5.0534999999999996e-05,
      "loss": 0.4495,
      "step": 10113
    },
    {
      "epoch": 0.5663568148728861,
      "grad_norm": 1.2597676515579224,
      "learning_rate": 5.054e-05,
      "loss": 0.4129,
      "step": 10114
    },
    {
      "epoch": 0.5664128121850152,
      "grad_norm": 1.2329849004745483,
      "learning_rate": 5.0545e-05,
      "loss": 0.4358,
      "step": 10115
    },
    {
      "epoch": 0.5664688094971442,
      "grad_norm": 1.415442943572998,
      "learning_rate": 5.0549999999999995e-05,
      "loss": 0.4814,
      "step": 10116
    },
    {
      "epoch": 0.5665248068092732,
      "grad_norm": 1.3459875583648682,
      "learning_rate": 5.0555000000000006e-05,
      "loss": 0.4023,
      "step": 10117
    },
    {
      "epoch": 0.5665808041214022,
      "grad_norm": 2.1763296127319336,
      "learning_rate": 5.056000000000001e-05,
      "loss": 0.447,
      "step": 10118
    },
    {
      "epoch": 0.5666368014335312,
      "grad_norm": 1.1652395725250244,
      "learning_rate": 5.056500000000001e-05,
      "loss": 0.385,
      "step": 10119
    },
    {
      "epoch": 0.5666927987456603,
      "grad_norm": 1.3457837104797363,
      "learning_rate": 5.0570000000000005e-05,
      "loss": 0.3948,
      "step": 10120
    },
    {
      "epoch": 0.5667487960577893,
      "grad_norm": 1.684307336807251,
      "learning_rate": 5.0575e-05,
      "loss": 0.5563,
      "step": 10121
    },
    {
      "epoch": 0.5668047933699183,
      "grad_norm": 1.1551640033721924,
      "learning_rate": 5.058000000000001e-05,
      "loss": 0.528,
      "step": 10122
    },
    {
      "epoch": 0.5668607906820473,
      "grad_norm": 1.3200267553329468,
      "learning_rate": 5.0585000000000004e-05,
      "loss": 0.4706,
      "step": 10123
    },
    {
      "epoch": 0.5669167879941763,
      "grad_norm": 1.070975661277771,
      "learning_rate": 5.059e-05,
      "loss": 0.3755,
      "step": 10124
    },
    {
      "epoch": 0.5669727853063053,
      "grad_norm": 1.593381404876709,
      "learning_rate": 5.0595000000000006e-05,
      "loss": 0.3762,
      "step": 10125
    },
    {
      "epoch": 0.5670287826184344,
      "grad_norm": 1.2779661417007446,
      "learning_rate": 5.0600000000000003e-05,
      "loss": 0.4005,
      "step": 10126
    },
    {
      "epoch": 0.5670847799305634,
      "grad_norm": 1.187401533126831,
      "learning_rate": 5.0605e-05,
      "loss": 0.4789,
      "step": 10127
    },
    {
      "epoch": 0.5671407772426923,
      "grad_norm": 1.4982678890228271,
      "learning_rate": 5.0610000000000005e-05,
      "loss": 0.5009,
      "step": 10128
    },
    {
      "epoch": 0.5671967745548213,
      "grad_norm": 1.1863610744476318,
      "learning_rate": 5.0615e-05,
      "loss": 0.4679,
      "step": 10129
    },
    {
      "epoch": 0.5672527718669503,
      "grad_norm": 1.1557377576828003,
      "learning_rate": 5.062e-05,
      "loss": 0.4048,
      "step": 10130
    },
    {
      "epoch": 0.5673087691790794,
      "grad_norm": 1.194615125656128,
      "learning_rate": 5.0625e-05,
      "loss": 0.4318,
      "step": 10131
    },
    {
      "epoch": 0.5673647664912084,
      "grad_norm": 1.3305182456970215,
      "learning_rate": 5.063e-05,
      "loss": 0.483,
      "step": 10132
    },
    {
      "epoch": 0.5674207638033374,
      "grad_norm": 1.4567062854766846,
      "learning_rate": 5.0635e-05,
      "loss": 0.4314,
      "step": 10133
    },
    {
      "epoch": 0.5674767611154664,
      "grad_norm": 1.1840482950210571,
      "learning_rate": 5.0639999999999996e-05,
      "loss": 0.352,
      "step": 10134
    },
    {
      "epoch": 0.5675327584275954,
      "grad_norm": 1.8856470584869385,
      "learning_rate": 5.0645e-05,
      "loss": 0.4333,
      "step": 10135
    },
    {
      "epoch": 0.5675887557397244,
      "grad_norm": 1.2895816564559937,
      "learning_rate": 5.065e-05,
      "loss": 0.4328,
      "step": 10136
    },
    {
      "epoch": 0.5676447530518535,
      "grad_norm": 1.1949578523635864,
      "learning_rate": 5.0654999999999995e-05,
      "loss": 0.505,
      "step": 10137
    },
    {
      "epoch": 0.5677007503639825,
      "grad_norm": 1.247236728668213,
      "learning_rate": 5.0660000000000006e-05,
      "loss": 0.4934,
      "step": 10138
    },
    {
      "epoch": 0.5677567476761115,
      "grad_norm": 1.4603780508041382,
      "learning_rate": 5.0665000000000004e-05,
      "loss": 0.6489,
      "step": 10139
    },
    {
      "epoch": 0.5678127449882405,
      "grad_norm": 1.3208463191986084,
      "learning_rate": 5.067000000000001e-05,
      "loss": 0.4261,
      "step": 10140
    },
    {
      "epoch": 0.5678687423003695,
      "grad_norm": 1.293473482131958,
      "learning_rate": 5.0675000000000005e-05,
      "loss": 0.3721,
      "step": 10141
    },
    {
      "epoch": 0.5679247396124986,
      "grad_norm": 1.4508566856384277,
      "learning_rate": 5.068e-05,
      "loss": 0.3791,
      "step": 10142
    },
    {
      "epoch": 0.5679807369246276,
      "grad_norm": 1.4139388799667358,
      "learning_rate": 5.068500000000001e-05,
      "loss": 0.4025,
      "step": 10143
    },
    {
      "epoch": 0.5680367342367566,
      "grad_norm": 1.1509840488433838,
      "learning_rate": 5.0690000000000004e-05,
      "loss": 0.3744,
      "step": 10144
    },
    {
      "epoch": 0.5680927315488856,
      "grad_norm": 1.2031747102737427,
      "learning_rate": 5.0695e-05,
      "loss": 0.3816,
      "step": 10145
    },
    {
      "epoch": 0.5681487288610146,
      "grad_norm": 1.9181218147277832,
      "learning_rate": 5.0700000000000006e-05,
      "loss": 0.4964,
      "step": 10146
    },
    {
      "epoch": 0.5682047261731437,
      "grad_norm": 1.152400016784668,
      "learning_rate": 5.0705e-05,
      "loss": 0.3611,
      "step": 10147
    },
    {
      "epoch": 0.5682607234852727,
      "grad_norm": 1.409926414489746,
      "learning_rate": 5.071e-05,
      "loss": 0.3901,
      "step": 10148
    },
    {
      "epoch": 0.5683167207974017,
      "grad_norm": 1.3198037147521973,
      "learning_rate": 5.0715e-05,
      "loss": 0.4538,
      "step": 10149
    },
    {
      "epoch": 0.5683727181095307,
      "grad_norm": 1.2766097784042358,
      "learning_rate": 5.072e-05,
      "loss": 0.4074,
      "step": 10150
    },
    {
      "epoch": 0.5684287154216597,
      "grad_norm": 1.1689236164093018,
      "learning_rate": 5.0725e-05,
      "loss": 0.3714,
      "step": 10151
    },
    {
      "epoch": 0.5684847127337888,
      "grad_norm": 1.399337887763977,
      "learning_rate": 5.073e-05,
      "loss": 0.4977,
      "step": 10152
    },
    {
      "epoch": 0.5685407100459178,
      "grad_norm": 1.2349555492401123,
      "learning_rate": 5.0735e-05,
      "loss": 0.5333,
      "step": 10153
    },
    {
      "epoch": 0.5685967073580468,
      "grad_norm": 1.2456928491592407,
      "learning_rate": 5.074e-05,
      "loss": 0.4529,
      "step": 10154
    },
    {
      "epoch": 0.5686527046701758,
      "grad_norm": 1.2561348676681519,
      "learning_rate": 5.0744999999999996e-05,
      "loss": 0.449,
      "step": 10155
    },
    {
      "epoch": 0.5687087019823048,
      "grad_norm": 1.1609960794448853,
      "learning_rate": 5.075e-05,
      "loss": 0.3185,
      "step": 10156
    },
    {
      "epoch": 0.5687646992944339,
      "grad_norm": 1.2437316179275513,
      "learning_rate": 5.0755e-05,
      "loss": 0.4063,
      "step": 10157
    },
    {
      "epoch": 0.5688206966065629,
      "grad_norm": 1.2899188995361328,
      "learning_rate": 5.076000000000001e-05,
      "loss": 0.4687,
      "step": 10158
    },
    {
      "epoch": 0.5688766939186919,
      "grad_norm": 1.1509284973144531,
      "learning_rate": 5.0765000000000006e-05,
      "loss": 0.3083,
      "step": 10159
    },
    {
      "epoch": 0.5689326912308209,
      "grad_norm": 1.4670699834823608,
      "learning_rate": 5.0770000000000003e-05,
      "loss": 0.5398,
      "step": 10160
    },
    {
      "epoch": 0.5689886885429499,
      "grad_norm": 1.5057661533355713,
      "learning_rate": 5.077500000000001e-05,
      "loss": 0.45,
      "step": 10161
    },
    {
      "epoch": 0.569044685855079,
      "grad_norm": 1.4503322839736938,
      "learning_rate": 5.0780000000000005e-05,
      "loss": 0.5332,
      "step": 10162
    },
    {
      "epoch": 0.569100683167208,
      "grad_norm": 1.3776382207870483,
      "learning_rate": 5.0785e-05,
      "loss": 0.5604,
      "step": 10163
    },
    {
      "epoch": 0.569156680479337,
      "grad_norm": 1.2540619373321533,
      "learning_rate": 5.079000000000001e-05,
      "loss": 0.4003,
      "step": 10164
    },
    {
      "epoch": 0.569212677791466,
      "grad_norm": 1.4435511827468872,
      "learning_rate": 5.0795000000000004e-05,
      "loss": 0.6547,
      "step": 10165
    },
    {
      "epoch": 0.569268675103595,
      "grad_norm": 1.2559958696365356,
      "learning_rate": 5.08e-05,
      "loss": 0.4572,
      "step": 10166
    },
    {
      "epoch": 0.569324672415724,
      "grad_norm": 1.4132412672042847,
      "learning_rate": 5.0805000000000006e-05,
      "loss": 0.5495,
      "step": 10167
    },
    {
      "epoch": 0.5693806697278531,
      "grad_norm": 1.234095811843872,
      "learning_rate": 5.081e-05,
      "loss": 0.3255,
      "step": 10168
    },
    {
      "epoch": 0.5694366670399821,
      "grad_norm": 1.3978824615478516,
      "learning_rate": 5.0815e-05,
      "loss": 0.4344,
      "step": 10169
    },
    {
      "epoch": 0.5694926643521111,
      "grad_norm": 1.2753530740737915,
      "learning_rate": 5.082e-05,
      "loss": 0.3858,
      "step": 10170
    },
    {
      "epoch": 0.5695486616642401,
      "grad_norm": 1.3017603158950806,
      "learning_rate": 5.0825e-05,
      "loss": 0.4449,
      "step": 10171
    },
    {
      "epoch": 0.5696046589763691,
      "grad_norm": 1.1952513456344604,
      "learning_rate": 5.083e-05,
      "loss": 0.4657,
      "step": 10172
    },
    {
      "epoch": 0.5696606562884982,
      "grad_norm": 1.2655614614486694,
      "learning_rate": 5.0835e-05,
      "loss": 0.3996,
      "step": 10173
    },
    {
      "epoch": 0.5697166536006272,
      "grad_norm": 1.0660436153411865,
      "learning_rate": 5.084e-05,
      "loss": 0.3162,
      "step": 10174
    },
    {
      "epoch": 0.5697726509127562,
      "grad_norm": 1.3102269172668457,
      "learning_rate": 5.0845e-05,
      "loss": 0.4801,
      "step": 10175
    },
    {
      "epoch": 0.5698286482248852,
      "grad_norm": 1.2505639791488647,
      "learning_rate": 5.0849999999999996e-05,
      "loss": 0.4468,
      "step": 10176
    },
    {
      "epoch": 0.5698846455370142,
      "grad_norm": 1.0204392671585083,
      "learning_rate": 5.0855e-05,
      "loss": 0.3442,
      "step": 10177
    },
    {
      "epoch": 0.5699406428491433,
      "grad_norm": 1.2225295305252075,
      "learning_rate": 5.0860000000000004e-05,
      "loss": 0.4091,
      "step": 10178
    },
    {
      "epoch": 0.5699966401612723,
      "grad_norm": 1.420210361480713,
      "learning_rate": 5.086500000000001e-05,
      "loss": 0.5708,
      "step": 10179
    },
    {
      "epoch": 0.5700526374734013,
      "grad_norm": 1.2864094972610474,
      "learning_rate": 5.0870000000000006e-05,
      "loss": 0.4386,
      "step": 10180
    },
    {
      "epoch": 0.5701086347855303,
      "grad_norm": 1.4768576622009277,
      "learning_rate": 5.0875e-05,
      "loss": 0.3914,
      "step": 10181
    },
    {
      "epoch": 0.5701646320976593,
      "grad_norm": 1.3639792203903198,
      "learning_rate": 5.088000000000001e-05,
      "loss": 0.4886,
      "step": 10182
    },
    {
      "epoch": 0.5702206294097883,
      "grad_norm": 1.3538386821746826,
      "learning_rate": 5.0885000000000005e-05,
      "loss": 0.3957,
      "step": 10183
    },
    {
      "epoch": 0.5702766267219174,
      "grad_norm": 1.321562647819519,
      "learning_rate": 5.089e-05,
      "loss": 0.471,
      "step": 10184
    },
    {
      "epoch": 0.5703326240340464,
      "grad_norm": 1.2572190761566162,
      "learning_rate": 5.0895000000000007e-05,
      "loss": 0.4959,
      "step": 10185
    },
    {
      "epoch": 0.5703886213461754,
      "grad_norm": 1.2916969060897827,
      "learning_rate": 5.0900000000000004e-05,
      "loss": 0.4295,
      "step": 10186
    },
    {
      "epoch": 0.5704446186583044,
      "grad_norm": 1.5530734062194824,
      "learning_rate": 5.0905e-05,
      "loss": 0.5665,
      "step": 10187
    },
    {
      "epoch": 0.5705006159704334,
      "grad_norm": 1.0694494247436523,
      "learning_rate": 5.091e-05,
      "loss": 0.3661,
      "step": 10188
    },
    {
      "epoch": 0.5705566132825625,
      "grad_norm": 1.4759570360183716,
      "learning_rate": 5.0915e-05,
      "loss": 0.4288,
      "step": 10189
    },
    {
      "epoch": 0.5706126105946915,
      "grad_norm": 1.5343525409698486,
      "learning_rate": 5.092e-05,
      "loss": 0.4077,
      "step": 10190
    },
    {
      "epoch": 0.5706686079068205,
      "grad_norm": 1.1712658405303955,
      "learning_rate": 5.0925e-05,
      "loss": 0.3963,
      "step": 10191
    },
    {
      "epoch": 0.5707246052189495,
      "grad_norm": 1.7039088010787964,
      "learning_rate": 5.093e-05,
      "loss": 0.4555,
      "step": 10192
    },
    {
      "epoch": 0.5707806025310785,
      "grad_norm": 1.1426856517791748,
      "learning_rate": 5.0935e-05,
      "loss": 0.3363,
      "step": 10193
    },
    {
      "epoch": 0.5708365998432076,
      "grad_norm": 1.233491063117981,
      "learning_rate": 5.094e-05,
      "loss": 0.437,
      "step": 10194
    },
    {
      "epoch": 0.5708925971553366,
      "grad_norm": 1.1578457355499268,
      "learning_rate": 5.0945e-05,
      "loss": 0.4331,
      "step": 10195
    },
    {
      "epoch": 0.5709485944674656,
      "grad_norm": 1.2121589183807373,
      "learning_rate": 5.095e-05,
      "loss": 0.4047,
      "step": 10196
    },
    {
      "epoch": 0.5710045917795946,
      "grad_norm": 1.410102128982544,
      "learning_rate": 5.0954999999999996e-05,
      "loss": 0.4156,
      "step": 10197
    },
    {
      "epoch": 0.5710605890917236,
      "grad_norm": 1.3294005393981934,
      "learning_rate": 5.096000000000001e-05,
      "loss": 0.4113,
      "step": 10198
    },
    {
      "epoch": 0.5711165864038527,
      "grad_norm": 1.3565648794174194,
      "learning_rate": 5.0965000000000004e-05,
      "loss": 0.4551,
      "step": 10199
    },
    {
      "epoch": 0.5711725837159817,
      "grad_norm": 1.9718737602233887,
      "learning_rate": 5.097000000000001e-05,
      "loss": 0.5973,
      "step": 10200
    },
    {
      "epoch": 0.5712285810281107,
      "grad_norm": 1.5698515176773071,
      "learning_rate": 5.0975000000000006e-05,
      "loss": 0.4464,
      "step": 10201
    },
    {
      "epoch": 0.5712845783402397,
      "grad_norm": 1.5377118587493896,
      "learning_rate": 5.098e-05,
      "loss": 0.4791,
      "step": 10202
    },
    {
      "epoch": 0.5713405756523687,
      "grad_norm": 1.1959205865859985,
      "learning_rate": 5.098500000000001e-05,
      "loss": 0.4156,
      "step": 10203
    },
    {
      "epoch": 0.5713965729644978,
      "grad_norm": 1.2330917119979858,
      "learning_rate": 5.0990000000000005e-05,
      "loss": 0.4184,
      "step": 10204
    },
    {
      "epoch": 0.5714525702766268,
      "grad_norm": 1.1605263948440552,
      "learning_rate": 5.0995e-05,
      "loss": 0.3128,
      "step": 10205
    },
    {
      "epoch": 0.5715085675887558,
      "grad_norm": 1.1472564935684204,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 0.3788,
      "step": 10206
    },
    {
      "epoch": 0.5715645649008848,
      "grad_norm": 1.4026507139205933,
      "learning_rate": 5.1005000000000004e-05,
      "loss": 0.4531,
      "step": 10207
    },
    {
      "epoch": 0.5716205622130138,
      "grad_norm": 1.314774513244629,
      "learning_rate": 5.101e-05,
      "loss": 0.3528,
      "step": 10208
    },
    {
      "epoch": 0.5716765595251428,
      "grad_norm": 1.4226630926132202,
      "learning_rate": 5.1015e-05,
      "loss": 0.4123,
      "step": 10209
    },
    {
      "epoch": 0.5717325568372719,
      "grad_norm": 1.463738203048706,
      "learning_rate": 5.102e-05,
      "loss": 0.6458,
      "step": 10210
    },
    {
      "epoch": 0.5717885541494008,
      "grad_norm": 1.4039021730422974,
      "learning_rate": 5.1025e-05,
      "loss": 0.3781,
      "step": 10211
    },
    {
      "epoch": 0.5718445514615298,
      "grad_norm": 1.6951605081558228,
      "learning_rate": 5.103e-05,
      "loss": 0.5027,
      "step": 10212
    },
    {
      "epoch": 0.5719005487736588,
      "grad_norm": 1.3380426168441772,
      "learning_rate": 5.1035e-05,
      "loss": 0.5473,
      "step": 10213
    },
    {
      "epoch": 0.5719565460857878,
      "grad_norm": 1.4729443788528442,
      "learning_rate": 5.104e-05,
      "loss": 0.3846,
      "step": 10214
    },
    {
      "epoch": 0.5720125433979169,
      "grad_norm": 1.4423785209655762,
      "learning_rate": 5.1045e-05,
      "loss": 0.7526,
      "step": 10215
    },
    {
      "epoch": 0.5720685407100459,
      "grad_norm": 1.164225459098816,
      "learning_rate": 5.105e-05,
      "loss": 0.328,
      "step": 10216
    },
    {
      "epoch": 0.5721245380221749,
      "grad_norm": 1.191494107246399,
      "learning_rate": 5.1055e-05,
      "loss": 0.3963,
      "step": 10217
    },
    {
      "epoch": 0.5721805353343039,
      "grad_norm": 1.256538987159729,
      "learning_rate": 5.106000000000001e-05,
      "loss": 0.4338,
      "step": 10218
    },
    {
      "epoch": 0.5722365326464329,
      "grad_norm": 1.5546156167984009,
      "learning_rate": 5.1065000000000007e-05,
      "loss": 0.4119,
      "step": 10219
    },
    {
      "epoch": 0.572292529958562,
      "grad_norm": 1.3575255870819092,
      "learning_rate": 5.1070000000000004e-05,
      "loss": 0.3489,
      "step": 10220
    },
    {
      "epoch": 0.572348527270691,
      "grad_norm": 1.6723190546035767,
      "learning_rate": 5.107500000000001e-05,
      "loss": 0.3096,
      "step": 10221
    },
    {
      "epoch": 0.57240452458282,
      "grad_norm": 1.1107463836669922,
      "learning_rate": 5.1080000000000006e-05,
      "loss": 0.3975,
      "step": 10222
    },
    {
      "epoch": 0.572460521894949,
      "grad_norm": 1.8103303909301758,
      "learning_rate": 5.1085e-05,
      "loss": 0.36,
      "step": 10223
    },
    {
      "epoch": 0.572516519207078,
      "grad_norm": 2.5603103637695312,
      "learning_rate": 5.109000000000001e-05,
      "loss": 0.4458,
      "step": 10224
    },
    {
      "epoch": 0.572572516519207,
      "grad_norm": 1.2526216506958008,
      "learning_rate": 5.1095000000000005e-05,
      "loss": 0.3847,
      "step": 10225
    },
    {
      "epoch": 0.5726285138313361,
      "grad_norm": 1.8950400352478027,
      "learning_rate": 5.11e-05,
      "loss": 0.5301,
      "step": 10226
    },
    {
      "epoch": 0.5726845111434651,
      "grad_norm": 1.1724538803100586,
      "learning_rate": 5.1105000000000006e-05,
      "loss": 0.4366,
      "step": 10227
    },
    {
      "epoch": 0.5727405084555941,
      "grad_norm": 1.6718838214874268,
      "learning_rate": 5.1110000000000004e-05,
      "loss": 0.4654,
      "step": 10228
    },
    {
      "epoch": 0.5727965057677231,
      "grad_norm": 1.3077808618545532,
      "learning_rate": 5.1115e-05,
      "loss": 0.4893,
      "step": 10229
    },
    {
      "epoch": 0.5728525030798521,
      "grad_norm": 1.3792833089828491,
      "learning_rate": 5.112e-05,
      "loss": 0.4925,
      "step": 10230
    },
    {
      "epoch": 0.5729085003919812,
      "grad_norm": 1.2939683198928833,
      "learning_rate": 5.1125e-05,
      "loss": 0.4321,
      "step": 10231
    },
    {
      "epoch": 0.5729644977041102,
      "grad_norm": 1.9967490434646606,
      "learning_rate": 5.113e-05,
      "loss": 0.3745,
      "step": 10232
    },
    {
      "epoch": 0.5730204950162392,
      "grad_norm": 1.1713602542877197,
      "learning_rate": 5.1135e-05,
      "loss": 0.4535,
      "step": 10233
    },
    {
      "epoch": 0.5730764923283682,
      "grad_norm": 1.2153222560882568,
      "learning_rate": 5.114e-05,
      "loss": 0.3062,
      "step": 10234
    },
    {
      "epoch": 0.5731324896404972,
      "grad_norm": 1.45026695728302,
      "learning_rate": 5.1145e-05,
      "loss": 0.4027,
      "step": 10235
    },
    {
      "epoch": 0.5731884869526263,
      "grad_norm": 1.146943211555481,
      "learning_rate": 5.1149999999999996e-05,
      "loss": 0.3594,
      "step": 10236
    },
    {
      "epoch": 0.5732444842647553,
      "grad_norm": 1.2367851734161377,
      "learning_rate": 5.1155e-05,
      "loss": 0.4114,
      "step": 10237
    },
    {
      "epoch": 0.5733004815768843,
      "grad_norm": 1.1932320594787598,
      "learning_rate": 5.1160000000000005e-05,
      "loss": 0.4816,
      "step": 10238
    },
    {
      "epoch": 0.5733564788890133,
      "grad_norm": 1.2560538053512573,
      "learning_rate": 5.116500000000001e-05,
      "loss": 0.4513,
      "step": 10239
    },
    {
      "epoch": 0.5734124762011423,
      "grad_norm": 1.418629765510559,
      "learning_rate": 5.1170000000000006e-05,
      "loss": 0.4694,
      "step": 10240
    },
    {
      "epoch": 0.5734684735132713,
      "grad_norm": 1.2285903692245483,
      "learning_rate": 5.1175000000000004e-05,
      "loss": 0.4627,
      "step": 10241
    },
    {
      "epoch": 0.5735244708254004,
      "grad_norm": 1.182493805885315,
      "learning_rate": 5.118000000000001e-05,
      "loss": 0.4864,
      "step": 10242
    },
    {
      "epoch": 0.5735804681375294,
      "grad_norm": 1.2190669775009155,
      "learning_rate": 5.1185000000000005e-05,
      "loss": 0.3992,
      "step": 10243
    },
    {
      "epoch": 0.5736364654496584,
      "grad_norm": 1.2626402378082275,
      "learning_rate": 5.119e-05,
      "loss": 0.6578,
      "step": 10244
    },
    {
      "epoch": 0.5736924627617874,
      "grad_norm": 1.4835706949234009,
      "learning_rate": 5.119500000000001e-05,
      "loss": 0.3841,
      "step": 10245
    },
    {
      "epoch": 0.5737484600739164,
      "grad_norm": 1.3277225494384766,
      "learning_rate": 5.1200000000000004e-05,
      "loss": 0.4006,
      "step": 10246
    },
    {
      "epoch": 0.5738044573860455,
      "grad_norm": 1.3944958448410034,
      "learning_rate": 5.1205e-05,
      "loss": 0.56,
      "step": 10247
    },
    {
      "epoch": 0.5738604546981745,
      "grad_norm": 1.4049177169799805,
      "learning_rate": 5.121e-05,
      "loss": 0.7031,
      "step": 10248
    },
    {
      "epoch": 0.5739164520103035,
      "grad_norm": 1.1918838024139404,
      "learning_rate": 5.1215000000000003e-05,
      "loss": 0.4446,
      "step": 10249
    },
    {
      "epoch": 0.5739724493224325,
      "grad_norm": 1.4402596950531006,
      "learning_rate": 5.122e-05,
      "loss": 0.5317,
      "step": 10250
    },
    {
      "epoch": 0.5740284466345615,
      "grad_norm": 1.4664949178695679,
      "learning_rate": 5.1225e-05,
      "loss": 0.4501,
      "step": 10251
    },
    {
      "epoch": 0.5740844439466906,
      "grad_norm": 1.4004675149917603,
      "learning_rate": 5.123e-05,
      "loss": 0.4477,
      "step": 10252
    },
    {
      "epoch": 0.5741404412588196,
      "grad_norm": 1.4063783884048462,
      "learning_rate": 5.1235e-05,
      "loss": 0.4045,
      "step": 10253
    },
    {
      "epoch": 0.5741964385709486,
      "grad_norm": 1.291314959526062,
      "learning_rate": 5.124e-05,
      "loss": 0.3632,
      "step": 10254
    },
    {
      "epoch": 0.5742524358830776,
      "grad_norm": 1.3557640314102173,
      "learning_rate": 5.1245e-05,
      "loss": 0.5423,
      "step": 10255
    },
    {
      "epoch": 0.5743084331952066,
      "grad_norm": 1.3525190353393555,
      "learning_rate": 5.125e-05,
      "loss": 0.3807,
      "step": 10256
    },
    {
      "epoch": 0.5743644305073357,
      "grad_norm": 1.233000636100769,
      "learning_rate": 5.1254999999999996e-05,
      "loss": 0.3606,
      "step": 10257
    },
    {
      "epoch": 0.5744204278194647,
      "grad_norm": 2.2913849353790283,
      "learning_rate": 5.1259999999999994e-05,
      "loss": 0.5721,
      "step": 10258
    },
    {
      "epoch": 0.5744764251315937,
      "grad_norm": 1.2874081134796143,
      "learning_rate": 5.1265000000000005e-05,
      "loss": 0.4782,
      "step": 10259
    },
    {
      "epoch": 0.5745324224437227,
      "grad_norm": 1.4316374063491821,
      "learning_rate": 5.127000000000001e-05,
      "loss": 0.4022,
      "step": 10260
    },
    {
      "epoch": 0.5745884197558517,
      "grad_norm": 1.219549298286438,
      "learning_rate": 5.1275000000000006e-05,
      "loss": 0.5247,
      "step": 10261
    },
    {
      "epoch": 0.5746444170679808,
      "grad_norm": 1.2923873662948608,
      "learning_rate": 5.1280000000000004e-05,
      "loss": 0.3801,
      "step": 10262
    },
    {
      "epoch": 0.5747004143801098,
      "grad_norm": 1.3383703231811523,
      "learning_rate": 5.128500000000001e-05,
      "loss": 0.5152,
      "step": 10263
    },
    {
      "epoch": 0.5747564116922388,
      "grad_norm": 1.1318565607070923,
      "learning_rate": 5.1290000000000005e-05,
      "loss": 0.4195,
      "step": 10264
    },
    {
      "epoch": 0.5748124090043678,
      "grad_norm": 1.4272615909576416,
      "learning_rate": 5.1295e-05,
      "loss": 0.45,
      "step": 10265
    },
    {
      "epoch": 0.5748684063164968,
      "grad_norm": 1.6429742574691772,
      "learning_rate": 5.130000000000001e-05,
      "loss": 0.5195,
      "step": 10266
    },
    {
      "epoch": 0.5749244036286258,
      "grad_norm": 1.1789437532424927,
      "learning_rate": 5.1305000000000004e-05,
      "loss": 0.4308,
      "step": 10267
    },
    {
      "epoch": 0.5749804009407549,
      "grad_norm": 1.2067891359329224,
      "learning_rate": 5.131e-05,
      "loss": 0.4,
      "step": 10268
    },
    {
      "epoch": 0.5750363982528839,
      "grad_norm": 1.212969183921814,
      "learning_rate": 5.1315e-05,
      "loss": 0.5542,
      "step": 10269
    },
    {
      "epoch": 0.5750923955650129,
      "grad_norm": 1.1853617429733276,
      "learning_rate": 5.132e-05,
      "loss": 0.4596,
      "step": 10270
    },
    {
      "epoch": 0.5751483928771419,
      "grad_norm": 1.5046513080596924,
      "learning_rate": 5.1325e-05,
      "loss": 0.459,
      "step": 10271
    },
    {
      "epoch": 0.5752043901892709,
      "grad_norm": 1.5734779834747314,
      "learning_rate": 5.133e-05,
      "loss": 0.5446,
      "step": 10272
    },
    {
      "epoch": 0.5752603875014,
      "grad_norm": 1.4467272758483887,
      "learning_rate": 5.1335e-05,
      "loss": 0.517,
      "step": 10273
    },
    {
      "epoch": 0.575316384813529,
      "grad_norm": 1.5649198293685913,
      "learning_rate": 5.134e-05,
      "loss": 0.547,
      "step": 10274
    },
    {
      "epoch": 0.575372382125658,
      "grad_norm": 1.5195143222808838,
      "learning_rate": 5.1345e-05,
      "loss": 0.5048,
      "step": 10275
    },
    {
      "epoch": 0.575428379437787,
      "grad_norm": 1.2118948698043823,
      "learning_rate": 5.135e-05,
      "loss": 0.4739,
      "step": 10276
    },
    {
      "epoch": 0.575484376749916,
      "grad_norm": 1.3803354501724243,
      "learning_rate": 5.1355e-05,
      "loss": 0.4965,
      "step": 10277
    },
    {
      "epoch": 0.5755403740620451,
      "grad_norm": 1.300803303718567,
      "learning_rate": 5.1359999999999996e-05,
      "loss": 0.4283,
      "step": 10278
    },
    {
      "epoch": 0.5755963713741741,
      "grad_norm": 1.058893084526062,
      "learning_rate": 5.136500000000001e-05,
      "loss": 0.322,
      "step": 10279
    },
    {
      "epoch": 0.5756523686863031,
      "grad_norm": 1.2869795560836792,
      "learning_rate": 5.1370000000000005e-05,
      "loss": 0.3932,
      "step": 10280
    },
    {
      "epoch": 0.5757083659984321,
      "grad_norm": 1.677501916885376,
      "learning_rate": 5.137500000000001e-05,
      "loss": 0.4229,
      "step": 10281
    },
    {
      "epoch": 0.5757643633105611,
      "grad_norm": 1.824832558631897,
      "learning_rate": 5.1380000000000006e-05,
      "loss": 0.4507,
      "step": 10282
    },
    {
      "epoch": 0.5758203606226902,
      "grad_norm": 1.1936434507369995,
      "learning_rate": 5.1385000000000004e-05,
      "loss": 0.3963,
      "step": 10283
    },
    {
      "epoch": 0.5758763579348192,
      "grad_norm": 1.6029094457626343,
      "learning_rate": 5.139000000000001e-05,
      "loss": 0.5803,
      "step": 10284
    },
    {
      "epoch": 0.5759323552469482,
      "grad_norm": 1.257612943649292,
      "learning_rate": 5.1395000000000005e-05,
      "loss": 0.3776,
      "step": 10285
    },
    {
      "epoch": 0.5759883525590772,
      "grad_norm": 1.2295781373977661,
      "learning_rate": 5.14e-05,
      "loss": 0.3462,
      "step": 10286
    },
    {
      "epoch": 0.5760443498712062,
      "grad_norm": 1.4837368726730347,
      "learning_rate": 5.1405e-05,
      "loss": 0.3868,
      "step": 10287
    },
    {
      "epoch": 0.5761003471833352,
      "grad_norm": 1.1890358924865723,
      "learning_rate": 5.1410000000000004e-05,
      "loss": 0.5082,
      "step": 10288
    },
    {
      "epoch": 0.5761563444954643,
      "grad_norm": 1.2450222969055176,
      "learning_rate": 5.1415e-05,
      "loss": 0.4077,
      "step": 10289
    },
    {
      "epoch": 0.5762123418075933,
      "grad_norm": 1.1931641101837158,
      "learning_rate": 5.142e-05,
      "loss": 0.3834,
      "step": 10290
    },
    {
      "epoch": 0.5762683391197223,
      "grad_norm": 1.5010100603103638,
      "learning_rate": 5.1425e-05,
      "loss": 0.4207,
      "step": 10291
    },
    {
      "epoch": 0.5763243364318513,
      "grad_norm": 1.242691159248352,
      "learning_rate": 5.143e-05,
      "loss": 0.2911,
      "step": 10292
    },
    {
      "epoch": 0.5763803337439803,
      "grad_norm": 1.0755046606063843,
      "learning_rate": 5.1435e-05,
      "loss": 0.398,
      "step": 10293
    },
    {
      "epoch": 0.5764363310561093,
      "grad_norm": 1.1281802654266357,
      "learning_rate": 5.144e-05,
      "loss": 0.4705,
      "step": 10294
    },
    {
      "epoch": 0.5764923283682383,
      "grad_norm": 1.2325894832611084,
      "learning_rate": 5.1445e-05,
      "loss": 0.401,
      "step": 10295
    },
    {
      "epoch": 0.5765483256803673,
      "grad_norm": 1.1909339427947998,
      "learning_rate": 5.145e-05,
      "loss": 0.3308,
      "step": 10296
    },
    {
      "epoch": 0.5766043229924963,
      "grad_norm": 1.4792735576629639,
      "learning_rate": 5.1454999999999994e-05,
      "loss": 0.4901,
      "step": 10297
    },
    {
      "epoch": 0.5766603203046253,
      "grad_norm": 1.283048152923584,
      "learning_rate": 5.146e-05,
      "loss": 0.3903,
      "step": 10298
    },
    {
      "epoch": 0.5767163176167543,
      "grad_norm": 1.468291997909546,
      "learning_rate": 5.146500000000001e-05,
      "loss": 0.4658,
      "step": 10299
    },
    {
      "epoch": 0.5767723149288834,
      "grad_norm": 1.227444052696228,
      "learning_rate": 5.147000000000001e-05,
      "loss": 0.3662,
      "step": 10300
    },
    {
      "epoch": 0.5768283122410124,
      "grad_norm": 1.3208351135253906,
      "learning_rate": 5.1475000000000004e-05,
      "loss": 0.3798,
      "step": 10301
    },
    {
      "epoch": 0.5768843095531414,
      "grad_norm": 1.2385649681091309,
      "learning_rate": 5.148000000000001e-05,
      "loss": 0.3872,
      "step": 10302
    },
    {
      "epoch": 0.5769403068652704,
      "grad_norm": 1.1699784994125366,
      "learning_rate": 5.1485000000000006e-05,
      "loss": 0.3847,
      "step": 10303
    },
    {
      "epoch": 0.5769963041773994,
      "grad_norm": 1.215610146522522,
      "learning_rate": 5.149e-05,
      "loss": 0.4676,
      "step": 10304
    },
    {
      "epoch": 0.5770523014895285,
      "grad_norm": 1.3170974254608154,
      "learning_rate": 5.149500000000001e-05,
      "loss": 0.5343,
      "step": 10305
    },
    {
      "epoch": 0.5771082988016575,
      "grad_norm": 1.1466550827026367,
      "learning_rate": 5.1500000000000005e-05,
      "loss": 0.4225,
      "step": 10306
    },
    {
      "epoch": 0.5771642961137865,
      "grad_norm": 1.6524522304534912,
      "learning_rate": 5.1505e-05,
      "loss": 0.52,
      "step": 10307
    },
    {
      "epoch": 0.5772202934259155,
      "grad_norm": 1.1020444631576538,
      "learning_rate": 5.151e-05,
      "loss": 0.3842,
      "step": 10308
    },
    {
      "epoch": 0.5772762907380445,
      "grad_norm": 1.213584065437317,
      "learning_rate": 5.1515000000000004e-05,
      "loss": 0.4118,
      "step": 10309
    },
    {
      "epoch": 0.5773322880501736,
      "grad_norm": 1.2105871438980103,
      "learning_rate": 5.152e-05,
      "loss": 0.4503,
      "step": 10310
    },
    {
      "epoch": 0.5773882853623026,
      "grad_norm": 1.268365740776062,
      "learning_rate": 5.1525e-05,
      "loss": 0.3572,
      "step": 10311
    },
    {
      "epoch": 0.5774442826744316,
      "grad_norm": 1.1743181943893433,
      "learning_rate": 5.153e-05,
      "loss": 0.3623,
      "step": 10312
    },
    {
      "epoch": 0.5775002799865606,
      "grad_norm": 1.4468861818313599,
      "learning_rate": 5.1535e-05,
      "loss": 0.6882,
      "step": 10313
    },
    {
      "epoch": 0.5775562772986896,
      "grad_norm": 1.0792425870895386,
      "learning_rate": 5.154e-05,
      "loss": 0.4079,
      "step": 10314
    },
    {
      "epoch": 0.5776122746108187,
      "grad_norm": 1.282312035560608,
      "learning_rate": 5.1545e-05,
      "loss": 0.3473,
      "step": 10315
    },
    {
      "epoch": 0.5776682719229477,
      "grad_norm": 1.4069294929504395,
      "learning_rate": 5.155e-05,
      "loss": 0.4258,
      "step": 10316
    },
    {
      "epoch": 0.5777242692350767,
      "grad_norm": 1.3479154109954834,
      "learning_rate": 5.1555e-05,
      "loss": 0.4756,
      "step": 10317
    },
    {
      "epoch": 0.5777802665472057,
      "grad_norm": 1.4789682626724243,
      "learning_rate": 5.1559999999999994e-05,
      "loss": 0.5154,
      "step": 10318
    },
    {
      "epoch": 0.5778362638593347,
      "grad_norm": 1.3756661415100098,
      "learning_rate": 5.1565000000000005e-05,
      "loss": 0.5582,
      "step": 10319
    },
    {
      "epoch": 0.5778922611714638,
      "grad_norm": 1.2096366882324219,
      "learning_rate": 5.157000000000001e-05,
      "loss": 0.4084,
      "step": 10320
    },
    {
      "epoch": 0.5779482584835928,
      "grad_norm": 1.2413229942321777,
      "learning_rate": 5.157500000000001e-05,
      "loss": 0.3693,
      "step": 10321
    },
    {
      "epoch": 0.5780042557957218,
      "grad_norm": 1.2954576015472412,
      "learning_rate": 5.1580000000000004e-05,
      "loss": 0.4466,
      "step": 10322
    },
    {
      "epoch": 0.5780602531078508,
      "grad_norm": 1.0908390283584595,
      "learning_rate": 5.158500000000001e-05,
      "loss": 0.4197,
      "step": 10323
    },
    {
      "epoch": 0.5781162504199798,
      "grad_norm": 1.415282130241394,
      "learning_rate": 5.1590000000000006e-05,
      "loss": 0.429,
      "step": 10324
    },
    {
      "epoch": 0.5781722477321088,
      "grad_norm": 1.2666020393371582,
      "learning_rate": 5.1595e-05,
      "loss": 0.503,
      "step": 10325
    },
    {
      "epoch": 0.5782282450442379,
      "grad_norm": 1.303097128868103,
      "learning_rate": 5.16e-05,
      "loss": 0.398,
      "step": 10326
    },
    {
      "epoch": 0.5782842423563669,
      "grad_norm": 1.124499797821045,
      "learning_rate": 5.1605000000000005e-05,
      "loss": 0.324,
      "step": 10327
    },
    {
      "epoch": 0.5783402396684959,
      "grad_norm": 1.4660717248916626,
      "learning_rate": 5.161e-05,
      "loss": 0.3562,
      "step": 10328
    },
    {
      "epoch": 0.5783962369806249,
      "grad_norm": 1.2222203016281128,
      "learning_rate": 5.1615e-05,
      "loss": 0.3489,
      "step": 10329
    },
    {
      "epoch": 0.5784522342927539,
      "grad_norm": 2.0427260398864746,
      "learning_rate": 5.1620000000000004e-05,
      "loss": 0.339,
      "step": 10330
    },
    {
      "epoch": 0.578508231604883,
      "grad_norm": 1.3806908130645752,
      "learning_rate": 5.1625e-05,
      "loss": 0.4112,
      "step": 10331
    },
    {
      "epoch": 0.578564228917012,
      "grad_norm": 2.235337495803833,
      "learning_rate": 5.163e-05,
      "loss": 0.4836,
      "step": 10332
    },
    {
      "epoch": 0.578620226229141,
      "grad_norm": 1.1346064805984497,
      "learning_rate": 5.1635e-05,
      "loss": 0.3712,
      "step": 10333
    },
    {
      "epoch": 0.57867622354127,
      "grad_norm": 1.5988128185272217,
      "learning_rate": 5.164e-05,
      "loss": 0.4664,
      "step": 10334
    },
    {
      "epoch": 0.578732220853399,
      "grad_norm": 1.2711799144744873,
      "learning_rate": 5.1645e-05,
      "loss": 0.3467,
      "step": 10335
    },
    {
      "epoch": 0.5787882181655281,
      "grad_norm": 1.3669320344924927,
      "learning_rate": 5.1649999999999995e-05,
      "loss": 0.4762,
      "step": 10336
    },
    {
      "epoch": 0.5788442154776571,
      "grad_norm": 1.696080207824707,
      "learning_rate": 5.1655e-05,
      "loss": 0.5719,
      "step": 10337
    },
    {
      "epoch": 0.5789002127897861,
      "grad_norm": 1.1864807605743408,
      "learning_rate": 5.166e-05,
      "loss": 0.7109,
      "step": 10338
    },
    {
      "epoch": 0.5789562101019151,
      "grad_norm": 1.3186880350112915,
      "learning_rate": 5.166500000000001e-05,
      "loss": 0.3441,
      "step": 10339
    },
    {
      "epoch": 0.5790122074140441,
      "grad_norm": 1.121558666229248,
      "learning_rate": 5.1670000000000005e-05,
      "loss": 0.3171,
      "step": 10340
    },
    {
      "epoch": 0.5790682047261732,
      "grad_norm": 1.4229801893234253,
      "learning_rate": 5.167500000000001e-05,
      "loss": 0.6164,
      "step": 10341
    },
    {
      "epoch": 0.5791242020383022,
      "grad_norm": 1.4089537858963013,
      "learning_rate": 5.168000000000001e-05,
      "loss": 0.5846,
      "step": 10342
    },
    {
      "epoch": 0.5791801993504312,
      "grad_norm": 1.085536003112793,
      "learning_rate": 5.1685000000000004e-05,
      "loss": 0.3138,
      "step": 10343
    },
    {
      "epoch": 0.5792361966625602,
      "grad_norm": 1.33896803855896,
      "learning_rate": 5.169000000000001e-05,
      "loss": 0.4165,
      "step": 10344
    },
    {
      "epoch": 0.5792921939746892,
      "grad_norm": 1.3204853534698486,
      "learning_rate": 5.1695000000000006e-05,
      "loss": 0.4337,
      "step": 10345
    },
    {
      "epoch": 0.5793481912868182,
      "grad_norm": 1.2813446521759033,
      "learning_rate": 5.17e-05,
      "loss": 0.4423,
      "step": 10346
    },
    {
      "epoch": 0.5794041885989473,
      "grad_norm": 1.2630864381790161,
      "learning_rate": 5.1705e-05,
      "loss": 0.3796,
      "step": 10347
    },
    {
      "epoch": 0.5794601859110763,
      "grad_norm": 1.107636570930481,
      "learning_rate": 5.1710000000000005e-05,
      "loss": 0.343,
      "step": 10348
    },
    {
      "epoch": 0.5795161832232053,
      "grad_norm": 1.1181697845458984,
      "learning_rate": 5.1715e-05,
      "loss": 0.4129,
      "step": 10349
    },
    {
      "epoch": 0.5795721805353343,
      "grad_norm": 1.3029804229736328,
      "learning_rate": 5.172e-05,
      "loss": 0.4137,
      "step": 10350
    },
    {
      "epoch": 0.5796281778474633,
      "grad_norm": 1.280713677406311,
      "learning_rate": 5.1725000000000004e-05,
      "loss": 0.4269,
      "step": 10351
    },
    {
      "epoch": 0.5796841751595924,
      "grad_norm": 1.6649599075317383,
      "learning_rate": 5.173e-05,
      "loss": 0.5059,
      "step": 10352
    },
    {
      "epoch": 0.5797401724717214,
      "grad_norm": 1.3355119228363037,
      "learning_rate": 5.1735e-05,
      "loss": 0.4417,
      "step": 10353
    },
    {
      "epoch": 0.5797961697838504,
      "grad_norm": 1.4218158721923828,
      "learning_rate": 5.174e-05,
      "loss": 0.4339,
      "step": 10354
    },
    {
      "epoch": 0.5798521670959794,
      "grad_norm": 1.2330162525177002,
      "learning_rate": 5.1745e-05,
      "loss": 0.4606,
      "step": 10355
    },
    {
      "epoch": 0.5799081644081084,
      "grad_norm": 1.3094184398651123,
      "learning_rate": 5.175e-05,
      "loss": 0.4366,
      "step": 10356
    },
    {
      "epoch": 0.5799641617202375,
      "grad_norm": 1.5101141929626465,
      "learning_rate": 5.1754999999999995e-05,
      "loss": 0.5264,
      "step": 10357
    },
    {
      "epoch": 0.5800201590323665,
      "grad_norm": 1.387771725654602,
      "learning_rate": 5.176e-05,
      "loss": 0.5416,
      "step": 10358
    },
    {
      "epoch": 0.5800761563444955,
      "grad_norm": 1.2103217840194702,
      "learning_rate": 5.176500000000001e-05,
      "loss": 0.5158,
      "step": 10359
    },
    {
      "epoch": 0.5801321536566245,
      "grad_norm": 1.405078411102295,
      "learning_rate": 5.177000000000001e-05,
      "loss": 0.4028,
      "step": 10360
    },
    {
      "epoch": 0.5801881509687535,
      "grad_norm": 2.192035675048828,
      "learning_rate": 5.1775000000000005e-05,
      "loss": 0.4246,
      "step": 10361
    },
    {
      "epoch": 0.5802441482808826,
      "grad_norm": 1.244006872177124,
      "learning_rate": 5.178000000000001e-05,
      "loss": 0.3975,
      "step": 10362
    },
    {
      "epoch": 0.5803001455930116,
      "grad_norm": 1.2485039234161377,
      "learning_rate": 5.1785000000000006e-05,
      "loss": 0.4321,
      "step": 10363
    },
    {
      "epoch": 0.5803561429051406,
      "grad_norm": 1.3139088153839111,
      "learning_rate": 5.1790000000000004e-05,
      "loss": 0.4581,
      "step": 10364
    },
    {
      "epoch": 0.5804121402172696,
      "grad_norm": 1.3031911849975586,
      "learning_rate": 5.1795e-05,
      "loss": 0.5174,
      "step": 10365
    },
    {
      "epoch": 0.5804681375293986,
      "grad_norm": 1.196021556854248,
      "learning_rate": 5.1800000000000005e-05,
      "loss": 0.3223,
      "step": 10366
    },
    {
      "epoch": 0.5805241348415277,
      "grad_norm": 1.153767466545105,
      "learning_rate": 5.1805e-05,
      "loss": 0.411,
      "step": 10367
    },
    {
      "epoch": 0.5805801321536567,
      "grad_norm": 1.1274921894073486,
      "learning_rate": 5.181e-05,
      "loss": 0.3668,
      "step": 10368
    },
    {
      "epoch": 0.5806361294657857,
      "grad_norm": 1.3947771787643433,
      "learning_rate": 5.1815000000000005e-05,
      "loss": 0.325,
      "step": 10369
    },
    {
      "epoch": 0.5806921267779147,
      "grad_norm": 1.2500982284545898,
      "learning_rate": 5.182e-05,
      "loss": 0.4032,
      "step": 10370
    },
    {
      "epoch": 0.5807481240900437,
      "grad_norm": 1.1656206846237183,
      "learning_rate": 5.1825e-05,
      "loss": 0.4406,
      "step": 10371
    },
    {
      "epoch": 0.5808041214021727,
      "grad_norm": 1.7991421222686768,
      "learning_rate": 5.1830000000000004e-05,
      "loss": 0.5553,
      "step": 10372
    },
    {
      "epoch": 0.5808601187143018,
      "grad_norm": 1.2823164463043213,
      "learning_rate": 5.1835e-05,
      "loss": 0.3647,
      "step": 10373
    },
    {
      "epoch": 0.5809161160264308,
      "grad_norm": 1.5040351152420044,
      "learning_rate": 5.184e-05,
      "loss": 0.5526,
      "step": 10374
    },
    {
      "epoch": 0.5809721133385598,
      "grad_norm": 1.8107975721359253,
      "learning_rate": 5.1844999999999996e-05,
      "loss": 0.6746,
      "step": 10375
    },
    {
      "epoch": 0.5810281106506887,
      "grad_norm": 1.2524305582046509,
      "learning_rate": 5.185e-05,
      "loss": 0.3989,
      "step": 10376
    },
    {
      "epoch": 0.5810841079628177,
      "grad_norm": 1.2219644784927368,
      "learning_rate": 5.1855e-05,
      "loss": 0.4442,
      "step": 10377
    },
    {
      "epoch": 0.5811401052749467,
      "grad_norm": 1.2759735584259033,
      "learning_rate": 5.1859999999999995e-05,
      "loss": 0.3981,
      "step": 10378
    },
    {
      "epoch": 0.5811961025870758,
      "grad_norm": 1.234581708908081,
      "learning_rate": 5.1865000000000006e-05,
      "loss": 0.4084,
      "step": 10379
    },
    {
      "epoch": 0.5812520998992048,
      "grad_norm": 1.4750521183013916,
      "learning_rate": 5.187000000000001e-05,
      "loss": 0.501,
      "step": 10380
    },
    {
      "epoch": 0.5813080972113338,
      "grad_norm": 1.4189141988754272,
      "learning_rate": 5.187500000000001e-05,
      "loss": 0.4196,
      "step": 10381
    },
    {
      "epoch": 0.5813640945234628,
      "grad_norm": 1.362120509147644,
      "learning_rate": 5.1880000000000005e-05,
      "loss": 0.4634,
      "step": 10382
    },
    {
      "epoch": 0.5814200918355918,
      "grad_norm": 1.1404772996902466,
      "learning_rate": 5.188500000000001e-05,
      "loss": 0.3291,
      "step": 10383
    },
    {
      "epoch": 0.5814760891477209,
      "grad_norm": 1.1161601543426514,
      "learning_rate": 5.1890000000000006e-05,
      "loss": 0.4301,
      "step": 10384
    },
    {
      "epoch": 0.5815320864598499,
      "grad_norm": 1.398680329322815,
      "learning_rate": 5.1895000000000004e-05,
      "loss": 0.5729,
      "step": 10385
    },
    {
      "epoch": 0.5815880837719789,
      "grad_norm": 1.1788133382797241,
      "learning_rate": 5.19e-05,
      "loss": 0.4033,
      "step": 10386
    },
    {
      "epoch": 0.5816440810841079,
      "grad_norm": 1.075253963470459,
      "learning_rate": 5.1905000000000005e-05,
      "loss": 0.4779,
      "step": 10387
    },
    {
      "epoch": 0.5817000783962369,
      "grad_norm": 1.4342814683914185,
      "learning_rate": 5.191e-05,
      "loss": 0.56,
      "step": 10388
    },
    {
      "epoch": 0.581756075708366,
      "grad_norm": 1.4962137937545776,
      "learning_rate": 5.1915e-05,
      "loss": 0.4512,
      "step": 10389
    },
    {
      "epoch": 0.581812073020495,
      "grad_norm": 1.5346430540084839,
      "learning_rate": 5.1920000000000004e-05,
      "loss": 0.4255,
      "step": 10390
    },
    {
      "epoch": 0.581868070332624,
      "grad_norm": 1.1646647453308105,
      "learning_rate": 5.1925e-05,
      "loss": 0.3365,
      "step": 10391
    },
    {
      "epoch": 0.581924067644753,
      "grad_norm": 1.2096935510635376,
      "learning_rate": 5.193e-05,
      "loss": 0.3587,
      "step": 10392
    },
    {
      "epoch": 0.581980064956882,
      "grad_norm": 1.6582748889923096,
      "learning_rate": 5.1935e-05,
      "loss": 0.4877,
      "step": 10393
    },
    {
      "epoch": 0.5820360622690111,
      "grad_norm": 1.4083611965179443,
      "learning_rate": 5.194e-05,
      "loss": 0.474,
      "step": 10394
    },
    {
      "epoch": 0.5820920595811401,
      "grad_norm": 1.2445719242095947,
      "learning_rate": 5.1945e-05,
      "loss": 0.4816,
      "step": 10395
    },
    {
      "epoch": 0.5821480568932691,
      "grad_norm": 20.22435760498047,
      "learning_rate": 5.1949999999999996e-05,
      "loss": 0.5547,
      "step": 10396
    },
    {
      "epoch": 0.5822040542053981,
      "grad_norm": 1.334487795829773,
      "learning_rate": 5.1955e-05,
      "loss": 0.3963,
      "step": 10397
    },
    {
      "epoch": 0.5822600515175271,
      "grad_norm": 1.431045413017273,
      "learning_rate": 5.196e-05,
      "loss": 0.4289,
      "step": 10398
    },
    {
      "epoch": 0.5823160488296562,
      "grad_norm": 1.3006852865219116,
      "learning_rate": 5.1964999999999995e-05,
      "loss": 0.4283,
      "step": 10399
    },
    {
      "epoch": 0.5823720461417852,
      "grad_norm": 1.5506912469863892,
      "learning_rate": 5.1970000000000006e-05,
      "loss": 0.4408,
      "step": 10400
    },
    {
      "epoch": 0.5824280434539142,
      "grad_norm": 1.452589988708496,
      "learning_rate": 5.197500000000001e-05,
      "loss": 0.513,
      "step": 10401
    },
    {
      "epoch": 0.5824840407660432,
      "grad_norm": 1.1917253732681274,
      "learning_rate": 5.198000000000001e-05,
      "loss": 0.3804,
      "step": 10402
    },
    {
      "epoch": 0.5825400380781722,
      "grad_norm": 1.2255934476852417,
      "learning_rate": 5.1985000000000005e-05,
      "loss": 0.4037,
      "step": 10403
    },
    {
      "epoch": 0.5825960353903012,
      "grad_norm": 1.1572686433792114,
      "learning_rate": 5.199000000000001e-05,
      "loss": 0.3719,
      "step": 10404
    },
    {
      "epoch": 0.5826520327024303,
      "grad_norm": 1.2482918500900269,
      "learning_rate": 5.1995000000000006e-05,
      "loss": 0.4869,
      "step": 10405
    },
    {
      "epoch": 0.5827080300145593,
      "grad_norm": 1.2625491619110107,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 0.5219,
      "step": 10406
    },
    {
      "epoch": 0.5827640273266883,
      "grad_norm": 1.3208270072937012,
      "learning_rate": 5.2005e-05,
      "loss": 0.3986,
      "step": 10407
    },
    {
      "epoch": 0.5828200246388173,
      "grad_norm": 1.1687214374542236,
      "learning_rate": 5.2010000000000005e-05,
      "loss": 0.2842,
      "step": 10408
    },
    {
      "epoch": 0.5828760219509463,
      "grad_norm": 1.534670114517212,
      "learning_rate": 5.2015e-05,
      "loss": 0.6339,
      "step": 10409
    },
    {
      "epoch": 0.5829320192630754,
      "grad_norm": 1.1783336400985718,
      "learning_rate": 5.202e-05,
      "loss": 0.3793,
      "step": 10410
    },
    {
      "epoch": 0.5829880165752044,
      "grad_norm": 1.3272316455841064,
      "learning_rate": 5.2025000000000004e-05,
      "loss": 0.345,
      "step": 10411
    },
    {
      "epoch": 0.5830440138873334,
      "grad_norm": 1.3698999881744385,
      "learning_rate": 5.203e-05,
      "loss": 0.5388,
      "step": 10412
    },
    {
      "epoch": 0.5831000111994624,
      "grad_norm": 1.1794499158859253,
      "learning_rate": 5.2035e-05,
      "loss": 0.3658,
      "step": 10413
    },
    {
      "epoch": 0.5831560085115914,
      "grad_norm": 1.287438154220581,
      "learning_rate": 5.204e-05,
      "loss": 0.5242,
      "step": 10414
    },
    {
      "epoch": 0.5832120058237205,
      "grad_norm": 1.367053508758545,
      "learning_rate": 5.2045e-05,
      "loss": 0.5462,
      "step": 10415
    },
    {
      "epoch": 0.5832680031358495,
      "grad_norm": 1.1334296464920044,
      "learning_rate": 5.205e-05,
      "loss": 0.345,
      "step": 10416
    },
    {
      "epoch": 0.5833240004479785,
      "grad_norm": 1.7370926141738892,
      "learning_rate": 5.2054999999999995e-05,
      "loss": 0.4888,
      "step": 10417
    },
    {
      "epoch": 0.5833799977601075,
      "grad_norm": 1.3780725002288818,
      "learning_rate": 5.206e-05,
      "loss": 0.5872,
      "step": 10418
    },
    {
      "epoch": 0.5834359950722365,
      "grad_norm": 1.4774549007415771,
      "learning_rate": 5.2065e-05,
      "loss": 0.5877,
      "step": 10419
    },
    {
      "epoch": 0.5834919923843656,
      "grad_norm": 1.231362223625183,
      "learning_rate": 5.207000000000001e-05,
      "loss": 0.3213,
      "step": 10420
    },
    {
      "epoch": 0.5835479896964946,
      "grad_norm": 1.2752066850662231,
      "learning_rate": 5.2075000000000005e-05,
      "loss": 0.3385,
      "step": 10421
    },
    {
      "epoch": 0.5836039870086236,
      "grad_norm": 1.5953947305679321,
      "learning_rate": 5.208000000000001e-05,
      "loss": 0.4162,
      "step": 10422
    },
    {
      "epoch": 0.5836599843207526,
      "grad_norm": 1.1868245601654053,
      "learning_rate": 5.208500000000001e-05,
      "loss": 0.5124,
      "step": 10423
    },
    {
      "epoch": 0.5837159816328816,
      "grad_norm": 1.6609355211257935,
      "learning_rate": 5.2090000000000004e-05,
      "loss": 0.4391,
      "step": 10424
    },
    {
      "epoch": 0.5837719789450107,
      "grad_norm": 1.9022648334503174,
      "learning_rate": 5.2095e-05,
      "loss": 0.4716,
      "step": 10425
    },
    {
      "epoch": 0.5838279762571397,
      "grad_norm": 1.5023819208145142,
      "learning_rate": 5.2100000000000006e-05,
      "loss": 0.6172,
      "step": 10426
    },
    {
      "epoch": 0.5838839735692687,
      "grad_norm": 1.3575319051742554,
      "learning_rate": 5.2105000000000003e-05,
      "loss": 0.455,
      "step": 10427
    },
    {
      "epoch": 0.5839399708813977,
      "grad_norm": 1.2734533548355103,
      "learning_rate": 5.211e-05,
      "loss": 0.4149,
      "step": 10428
    },
    {
      "epoch": 0.5839959681935267,
      "grad_norm": 1.1898829936981201,
      "learning_rate": 5.2115000000000005e-05,
      "loss": 0.4806,
      "step": 10429
    },
    {
      "epoch": 0.5840519655056557,
      "grad_norm": 1.3889342546463013,
      "learning_rate": 5.212e-05,
      "loss": 0.5,
      "step": 10430
    },
    {
      "epoch": 0.5841079628177848,
      "grad_norm": 1.3933897018432617,
      "learning_rate": 5.2125e-05,
      "loss": 0.3554,
      "step": 10431
    },
    {
      "epoch": 0.5841639601299138,
      "grad_norm": 1.0557312965393066,
      "learning_rate": 5.2130000000000004e-05,
      "loss": 0.3702,
      "step": 10432
    },
    {
      "epoch": 0.5842199574420428,
      "grad_norm": 1.0459210872650146,
      "learning_rate": 5.2135e-05,
      "loss": 0.3555,
      "step": 10433
    },
    {
      "epoch": 0.5842759547541718,
      "grad_norm": 1.7518179416656494,
      "learning_rate": 5.214e-05,
      "loss": 0.4487,
      "step": 10434
    },
    {
      "epoch": 0.5843319520663008,
      "grad_norm": 1.1503556966781616,
      "learning_rate": 5.2144999999999996e-05,
      "loss": 0.4245,
      "step": 10435
    },
    {
      "epoch": 0.5843879493784299,
      "grad_norm": 1.4215766191482544,
      "learning_rate": 5.215e-05,
      "loss": 0.441,
      "step": 10436
    },
    {
      "epoch": 0.5844439466905589,
      "grad_norm": 1.3397252559661865,
      "learning_rate": 5.2155e-05,
      "loss": 0.3169,
      "step": 10437
    },
    {
      "epoch": 0.5844999440026879,
      "grad_norm": 1.2923110723495483,
      "learning_rate": 5.2159999999999995e-05,
      "loss": 0.4483,
      "step": 10438
    },
    {
      "epoch": 0.5845559413148169,
      "grad_norm": 1.4886518716812134,
      "learning_rate": 5.2165e-05,
      "loss": 0.4953,
      "step": 10439
    },
    {
      "epoch": 0.5846119386269459,
      "grad_norm": 1.6069964170455933,
      "learning_rate": 5.217000000000001e-05,
      "loss": 0.5019,
      "step": 10440
    },
    {
      "epoch": 0.584667935939075,
      "grad_norm": 1.211219072341919,
      "learning_rate": 5.217500000000001e-05,
      "loss": 0.5428,
      "step": 10441
    },
    {
      "epoch": 0.584723933251204,
      "grad_norm": 1.6830753087997437,
      "learning_rate": 5.2180000000000005e-05,
      "loss": 0.5872,
      "step": 10442
    },
    {
      "epoch": 0.584779930563333,
      "grad_norm": 1.4909956455230713,
      "learning_rate": 5.218500000000001e-05,
      "loss": 0.566,
      "step": 10443
    },
    {
      "epoch": 0.584835927875462,
      "grad_norm": 1.2251861095428467,
      "learning_rate": 5.219000000000001e-05,
      "loss": 0.4313,
      "step": 10444
    },
    {
      "epoch": 0.584891925187591,
      "grad_norm": 1.5191506147384644,
      "learning_rate": 5.2195000000000004e-05,
      "loss": 0.6065,
      "step": 10445
    },
    {
      "epoch": 0.58494792249972,
      "grad_norm": 1.486066460609436,
      "learning_rate": 5.22e-05,
      "loss": 0.4421,
      "step": 10446
    },
    {
      "epoch": 0.5850039198118491,
      "grad_norm": 1.4842216968536377,
      "learning_rate": 5.2205000000000006e-05,
      "loss": 0.4484,
      "step": 10447
    },
    {
      "epoch": 0.5850599171239781,
      "grad_norm": 1.1850941181182861,
      "learning_rate": 5.221e-05,
      "loss": 0.3796,
      "step": 10448
    },
    {
      "epoch": 0.5851159144361071,
      "grad_norm": 1.4743479490280151,
      "learning_rate": 5.2215e-05,
      "loss": 0.3642,
      "step": 10449
    },
    {
      "epoch": 0.5851719117482361,
      "grad_norm": 1.1929072141647339,
      "learning_rate": 5.2220000000000005e-05,
      "loss": 0.3456,
      "step": 10450
    },
    {
      "epoch": 0.5852279090603651,
      "grad_norm": 1.238207459449768,
      "learning_rate": 5.2225e-05,
      "loss": 0.3727,
      "step": 10451
    },
    {
      "epoch": 0.5852839063724942,
      "grad_norm": 1.2077758312225342,
      "learning_rate": 5.223e-05,
      "loss": 0.3962,
      "step": 10452
    },
    {
      "epoch": 0.5853399036846232,
      "grad_norm": 1.3768528699874878,
      "learning_rate": 5.2235000000000004e-05,
      "loss": 0.4577,
      "step": 10453
    },
    {
      "epoch": 0.5853959009967522,
      "grad_norm": 1.1925586462020874,
      "learning_rate": 5.224e-05,
      "loss": 0.54,
      "step": 10454
    },
    {
      "epoch": 0.5854518983088812,
      "grad_norm": 1.4597384929656982,
      "learning_rate": 5.2245e-05,
      "loss": 0.3021,
      "step": 10455
    },
    {
      "epoch": 0.5855078956210102,
      "grad_norm": 1.1666860580444336,
      "learning_rate": 5.2249999999999996e-05,
      "loss": 0.3434,
      "step": 10456
    },
    {
      "epoch": 0.5855638929331393,
      "grad_norm": 1.5159988403320312,
      "learning_rate": 5.2255e-05,
      "loss": 0.5155,
      "step": 10457
    },
    {
      "epoch": 0.5856198902452683,
      "grad_norm": 1.2205889225006104,
      "learning_rate": 5.226e-05,
      "loss": 0.3108,
      "step": 10458
    },
    {
      "epoch": 0.5856758875573972,
      "grad_norm": 1.1630786657333374,
      "learning_rate": 5.2264999999999995e-05,
      "loss": 0.4019,
      "step": 10459
    },
    {
      "epoch": 0.5857318848695262,
      "grad_norm": 1.2360212802886963,
      "learning_rate": 5.2270000000000006e-05,
      "loss": 0.4528,
      "step": 10460
    },
    {
      "epoch": 0.5857878821816552,
      "grad_norm": 1.3985111713409424,
      "learning_rate": 5.227500000000001e-05,
      "loss": 0.4072,
      "step": 10461
    },
    {
      "epoch": 0.5858438794937842,
      "grad_norm": 1.2422573566436768,
      "learning_rate": 5.228000000000001e-05,
      "loss": 0.4397,
      "step": 10462
    },
    {
      "epoch": 0.5858998768059133,
      "grad_norm": 1.134757161140442,
      "learning_rate": 5.2285000000000005e-05,
      "loss": 0.3392,
      "step": 10463
    },
    {
      "epoch": 0.5859558741180423,
      "grad_norm": 1.3189239501953125,
      "learning_rate": 5.229e-05,
      "loss": 0.6016,
      "step": 10464
    },
    {
      "epoch": 0.5860118714301713,
      "grad_norm": 2.8348681926727295,
      "learning_rate": 5.229500000000001e-05,
      "loss": 0.3658,
      "step": 10465
    },
    {
      "epoch": 0.5860678687423003,
      "grad_norm": 1.1461681127548218,
      "learning_rate": 5.2300000000000004e-05,
      "loss": 0.5019,
      "step": 10466
    },
    {
      "epoch": 0.5861238660544293,
      "grad_norm": 1.1149210929870605,
      "learning_rate": 5.2305e-05,
      "loss": 0.4242,
      "step": 10467
    },
    {
      "epoch": 0.5861798633665584,
      "grad_norm": 1.132940649986267,
      "learning_rate": 5.2310000000000006e-05,
      "loss": 0.4808,
      "step": 10468
    },
    {
      "epoch": 0.5862358606786874,
      "grad_norm": 1.4029748439788818,
      "learning_rate": 5.2315e-05,
      "loss": 0.4697,
      "step": 10469
    },
    {
      "epoch": 0.5862918579908164,
      "grad_norm": 1.6982300281524658,
      "learning_rate": 5.232e-05,
      "loss": 0.3914,
      "step": 10470
    },
    {
      "epoch": 0.5863478553029454,
      "grad_norm": 1.6048387289047241,
      "learning_rate": 5.2325000000000005e-05,
      "loss": 0.4111,
      "step": 10471
    },
    {
      "epoch": 0.5864038526150744,
      "grad_norm": 1.2392228841781616,
      "learning_rate": 5.233e-05,
      "loss": 0.4016,
      "step": 10472
    },
    {
      "epoch": 0.5864598499272035,
      "grad_norm": 1.248370885848999,
      "learning_rate": 5.2335e-05,
      "loss": 0.4372,
      "step": 10473
    },
    {
      "epoch": 0.5865158472393325,
      "grad_norm": 1.248731255531311,
      "learning_rate": 5.234e-05,
      "loss": 0.5049,
      "step": 10474
    },
    {
      "epoch": 0.5865718445514615,
      "grad_norm": 1.577825665473938,
      "learning_rate": 5.2345e-05,
      "loss": 0.5232,
      "step": 10475
    },
    {
      "epoch": 0.5866278418635905,
      "grad_norm": 15.490825653076172,
      "learning_rate": 5.235e-05,
      "loss": 0.6373,
      "step": 10476
    },
    {
      "epoch": 0.5866838391757195,
      "grad_norm": 1.3385379314422607,
      "learning_rate": 5.2354999999999996e-05,
      "loss": 0.405,
      "step": 10477
    },
    {
      "epoch": 0.5867398364878486,
      "grad_norm": 1.0995076894760132,
      "learning_rate": 5.236e-05,
      "loss": 0.3462,
      "step": 10478
    },
    {
      "epoch": 0.5867958337999776,
      "grad_norm": 1.135644793510437,
      "learning_rate": 5.2365e-05,
      "loss": 0.4616,
      "step": 10479
    },
    {
      "epoch": 0.5868518311121066,
      "grad_norm": 1.2388736009597778,
      "learning_rate": 5.237000000000001e-05,
      "loss": 0.3648,
      "step": 10480
    },
    {
      "epoch": 0.5869078284242356,
      "grad_norm": 1.3318978548049927,
      "learning_rate": 5.2375000000000006e-05,
      "loss": 0.5455,
      "step": 10481
    },
    {
      "epoch": 0.5869638257363646,
      "grad_norm": 1.180772066116333,
      "learning_rate": 5.238000000000001e-05,
      "loss": 0.4188,
      "step": 10482
    },
    {
      "epoch": 0.5870198230484936,
      "grad_norm": 1.0729402303695679,
      "learning_rate": 5.238500000000001e-05,
      "loss": 0.395,
      "step": 10483
    },
    {
      "epoch": 0.5870758203606227,
      "grad_norm": 1.390486717224121,
      "learning_rate": 5.2390000000000005e-05,
      "loss": 0.48,
      "step": 10484
    },
    {
      "epoch": 0.5871318176727517,
      "grad_norm": 1.126437783241272,
      "learning_rate": 5.2395e-05,
      "loss": 0.3945,
      "step": 10485
    },
    {
      "epoch": 0.5871878149848807,
      "grad_norm": 1.2294518947601318,
      "learning_rate": 5.2400000000000007e-05,
      "loss": 0.3782,
      "step": 10486
    },
    {
      "epoch": 0.5872438122970097,
      "grad_norm": 1.1480926275253296,
      "learning_rate": 5.2405000000000004e-05,
      "loss": 0.3972,
      "step": 10487
    },
    {
      "epoch": 0.5872998096091387,
      "grad_norm": 1.2376410961151123,
      "learning_rate": 5.241e-05,
      "loss": 0.5238,
      "step": 10488
    },
    {
      "epoch": 0.5873558069212678,
      "grad_norm": 1.3651033639907837,
      "learning_rate": 5.2415000000000006e-05,
      "loss": 0.4656,
      "step": 10489
    },
    {
      "epoch": 0.5874118042333968,
      "grad_norm": 1.3575665950775146,
      "learning_rate": 5.242e-05,
      "loss": 0.4168,
      "step": 10490
    },
    {
      "epoch": 0.5874678015455258,
      "grad_norm": 1.1683986186981201,
      "learning_rate": 5.2425e-05,
      "loss": 0.4651,
      "step": 10491
    },
    {
      "epoch": 0.5875237988576548,
      "grad_norm": 1.036370038986206,
      "learning_rate": 5.2430000000000005e-05,
      "loss": 0.2864,
      "step": 10492
    },
    {
      "epoch": 0.5875797961697838,
      "grad_norm": 1.2672920227050781,
      "learning_rate": 5.2435e-05,
      "loss": 0.4514,
      "step": 10493
    },
    {
      "epoch": 0.5876357934819129,
      "grad_norm": 1.1965370178222656,
      "learning_rate": 5.244e-05,
      "loss": 0.3968,
      "step": 10494
    },
    {
      "epoch": 0.5876917907940419,
      "grad_norm": 1.398776888847351,
      "learning_rate": 5.2445e-05,
      "loss": 0.4031,
      "step": 10495
    },
    {
      "epoch": 0.5877477881061709,
      "grad_norm": 1.222611904144287,
      "learning_rate": 5.245e-05,
      "loss": 0.3937,
      "step": 10496
    },
    {
      "epoch": 0.5878037854182999,
      "grad_norm": 2.0282087326049805,
      "learning_rate": 5.2455e-05,
      "loss": 0.5691,
      "step": 10497
    },
    {
      "epoch": 0.5878597827304289,
      "grad_norm": 1.4633281230926514,
      "learning_rate": 5.2459999999999996e-05,
      "loss": 0.3477,
      "step": 10498
    },
    {
      "epoch": 0.587915780042558,
      "grad_norm": 1.387283205986023,
      "learning_rate": 5.2465e-05,
      "loss": 0.4913,
      "step": 10499
    },
    {
      "epoch": 0.587971777354687,
      "grad_norm": 1.2462290525436401,
      "learning_rate": 5.247000000000001e-05,
      "loss": 0.4113,
      "step": 10500
    },
    {
      "epoch": 0.588027774666816,
      "grad_norm": 1.4916491508483887,
      "learning_rate": 5.247500000000001e-05,
      "loss": 0.3696,
      "step": 10501
    },
    {
      "epoch": 0.588083771978945,
      "grad_norm": 6.327125072479248,
      "learning_rate": 5.2480000000000006e-05,
      "loss": 0.4427,
      "step": 10502
    },
    {
      "epoch": 0.588139769291074,
      "grad_norm": 1.3494855165481567,
      "learning_rate": 5.2485e-05,
      "loss": 0.3723,
      "step": 10503
    },
    {
      "epoch": 0.588195766603203,
      "grad_norm": 1.153686285018921,
      "learning_rate": 5.249000000000001e-05,
      "loss": 0.4094,
      "step": 10504
    },
    {
      "epoch": 0.5882517639153321,
      "grad_norm": 1.3393079042434692,
      "learning_rate": 5.2495000000000005e-05,
      "loss": 0.4726,
      "step": 10505
    },
    {
      "epoch": 0.5883077612274611,
      "grad_norm": 1.1729991436004639,
      "learning_rate": 5.25e-05,
      "loss": 0.3605,
      "step": 10506
    },
    {
      "epoch": 0.5883637585395901,
      "grad_norm": 1.1829841136932373,
      "learning_rate": 5.2505000000000006e-05,
      "loss": 0.3994,
      "step": 10507
    },
    {
      "epoch": 0.5884197558517191,
      "grad_norm": 1.4506311416625977,
      "learning_rate": 5.2510000000000004e-05,
      "loss": 0.6831,
      "step": 10508
    },
    {
      "epoch": 0.5884757531638481,
      "grad_norm": 1.20071280002594,
      "learning_rate": 5.2515e-05,
      "loss": 0.3928,
      "step": 10509
    },
    {
      "epoch": 0.5885317504759772,
      "grad_norm": 1.2628368139266968,
      "learning_rate": 5.2520000000000005e-05,
      "loss": 0.4455,
      "step": 10510
    },
    {
      "epoch": 0.5885877477881062,
      "grad_norm": 1.1907291412353516,
      "learning_rate": 5.2525e-05,
      "loss": 0.4332,
      "step": 10511
    },
    {
      "epoch": 0.5886437451002352,
      "grad_norm": 1.2729754447937012,
      "learning_rate": 5.253e-05,
      "loss": 0.315,
      "step": 10512
    },
    {
      "epoch": 0.5886997424123642,
      "grad_norm": 1.0806745290756226,
      "learning_rate": 5.2535e-05,
      "loss": 0.3771,
      "step": 10513
    },
    {
      "epoch": 0.5887557397244932,
      "grad_norm": 1.6687424182891846,
      "learning_rate": 5.254e-05,
      "loss": 0.5379,
      "step": 10514
    },
    {
      "epoch": 0.5888117370366223,
      "grad_norm": 1.374032974243164,
      "learning_rate": 5.2545e-05,
      "loss": 0.4884,
      "step": 10515
    },
    {
      "epoch": 0.5888677343487513,
      "grad_norm": 1.7508906126022339,
      "learning_rate": 5.255e-05,
      "loss": 0.3592,
      "step": 10516
    },
    {
      "epoch": 0.5889237316608803,
      "grad_norm": 1.1985833644866943,
      "learning_rate": 5.2555e-05,
      "loss": 0.4139,
      "step": 10517
    },
    {
      "epoch": 0.5889797289730093,
      "grad_norm": 1.2333550453186035,
      "learning_rate": 5.256e-05,
      "loss": 0.3817,
      "step": 10518
    },
    {
      "epoch": 0.5890357262851383,
      "grad_norm": 1.2691351175308228,
      "learning_rate": 5.2564999999999996e-05,
      "loss": 0.4359,
      "step": 10519
    },
    {
      "epoch": 0.5890917235972674,
      "grad_norm": 1.2514894008636475,
      "learning_rate": 5.257e-05,
      "loss": 0.5062,
      "step": 10520
    },
    {
      "epoch": 0.5891477209093964,
      "grad_norm": 1.4081145524978638,
      "learning_rate": 5.257500000000001e-05,
      "loss": 0.3867,
      "step": 10521
    },
    {
      "epoch": 0.5892037182215254,
      "grad_norm": 1.2950271368026733,
      "learning_rate": 5.258000000000001e-05,
      "loss": 0.4013,
      "step": 10522
    },
    {
      "epoch": 0.5892597155336544,
      "grad_norm": 1.096846342086792,
      "learning_rate": 5.2585000000000006e-05,
      "loss": 0.4453,
      "step": 10523
    },
    {
      "epoch": 0.5893157128457834,
      "grad_norm": 1.0097601413726807,
      "learning_rate": 5.259e-05,
      "loss": 0.4251,
      "step": 10524
    },
    {
      "epoch": 0.5893717101579125,
      "grad_norm": 1.2172775268554688,
      "learning_rate": 5.259500000000001e-05,
      "loss": 0.3619,
      "step": 10525
    },
    {
      "epoch": 0.5894277074700415,
      "grad_norm": 1.1469647884368896,
      "learning_rate": 5.2600000000000005e-05,
      "loss": 0.4751,
      "step": 10526
    },
    {
      "epoch": 0.5894837047821705,
      "grad_norm": 1.3321958780288696,
      "learning_rate": 5.2605e-05,
      "loss": 0.3585,
      "step": 10527
    },
    {
      "epoch": 0.5895397020942995,
      "grad_norm": 1.2122538089752197,
      "learning_rate": 5.2610000000000006e-05,
      "loss": 0.3275,
      "step": 10528
    },
    {
      "epoch": 0.5895956994064285,
      "grad_norm": 1.414595603942871,
      "learning_rate": 5.2615000000000004e-05,
      "loss": 0.5268,
      "step": 10529
    },
    {
      "epoch": 0.5896516967185575,
      "grad_norm": 1.2470372915267944,
      "learning_rate": 5.262e-05,
      "loss": 0.3476,
      "step": 10530
    },
    {
      "epoch": 0.5897076940306866,
      "grad_norm": 1.3111385107040405,
      "learning_rate": 5.2625000000000005e-05,
      "loss": 0.4256,
      "step": 10531
    },
    {
      "epoch": 0.5897636913428156,
      "grad_norm": 1.2466546297073364,
      "learning_rate": 5.263e-05,
      "loss": 0.3991,
      "step": 10532
    },
    {
      "epoch": 0.5898196886549446,
      "grad_norm": 1.1770704984664917,
      "learning_rate": 5.2635e-05,
      "loss": 0.2895,
      "step": 10533
    },
    {
      "epoch": 0.5898756859670736,
      "grad_norm": 1.14566171169281,
      "learning_rate": 5.264e-05,
      "loss": 0.5139,
      "step": 10534
    },
    {
      "epoch": 0.5899316832792026,
      "grad_norm": 1.2309952974319458,
      "learning_rate": 5.2645e-05,
      "loss": 0.4874,
      "step": 10535
    },
    {
      "epoch": 0.5899876805913317,
      "grad_norm": 1.2991855144500732,
      "learning_rate": 5.265e-05,
      "loss": 0.4646,
      "step": 10536
    },
    {
      "epoch": 0.5900436779034607,
      "grad_norm": 1.1817588806152344,
      "learning_rate": 5.2654999999999996e-05,
      "loss": 0.3928,
      "step": 10537
    },
    {
      "epoch": 0.5900996752155897,
      "grad_norm": 1.4595263004302979,
      "learning_rate": 5.266e-05,
      "loss": 0.4957,
      "step": 10538
    },
    {
      "epoch": 0.5901556725277187,
      "grad_norm": 1.3468631505966187,
      "learning_rate": 5.2665e-05,
      "loss": 0.4032,
      "step": 10539
    },
    {
      "epoch": 0.5902116698398477,
      "grad_norm": 1.106804370880127,
      "learning_rate": 5.2669999999999995e-05,
      "loss": 0.495,
      "step": 10540
    },
    {
      "epoch": 0.5902676671519768,
      "grad_norm": 1.1846035718917847,
      "learning_rate": 5.2675000000000006e-05,
      "loss": 0.3707,
      "step": 10541
    },
    {
      "epoch": 0.5903236644641057,
      "grad_norm": 1.2655447721481323,
      "learning_rate": 5.2680000000000004e-05,
      "loss": 0.4262,
      "step": 10542
    },
    {
      "epoch": 0.5903796617762347,
      "grad_norm": 1.2150630950927734,
      "learning_rate": 5.268500000000001e-05,
      "loss": 0.5081,
      "step": 10543
    },
    {
      "epoch": 0.5904356590883637,
      "grad_norm": 1.4314662218093872,
      "learning_rate": 5.2690000000000005e-05,
      "loss": 0.4474,
      "step": 10544
    },
    {
      "epoch": 0.5904916564004927,
      "grad_norm": 2.045747756958008,
      "learning_rate": 5.2695e-05,
      "loss": 0.4762,
      "step": 10545
    },
    {
      "epoch": 0.5905476537126217,
      "grad_norm": 1.1559627056121826,
      "learning_rate": 5.270000000000001e-05,
      "loss": 0.3582,
      "step": 10546
    },
    {
      "epoch": 0.5906036510247508,
      "grad_norm": 1.184889793395996,
      "learning_rate": 5.2705000000000004e-05,
      "loss": 0.3376,
      "step": 10547
    },
    {
      "epoch": 0.5906596483368798,
      "grad_norm": 1.159448504447937,
      "learning_rate": 5.271e-05,
      "loss": 0.3158,
      "step": 10548
    },
    {
      "epoch": 0.5907156456490088,
      "grad_norm": 1.1259042024612427,
      "learning_rate": 5.2715000000000006e-05,
      "loss": 0.3088,
      "step": 10549
    },
    {
      "epoch": 0.5907716429611378,
      "grad_norm": 1.158154010772705,
      "learning_rate": 5.2720000000000003e-05,
      "loss": 0.3598,
      "step": 10550
    },
    {
      "epoch": 0.5908276402732668,
      "grad_norm": 1.3170642852783203,
      "learning_rate": 5.2725e-05,
      "loss": 0.3795,
      "step": 10551
    },
    {
      "epoch": 0.5908836375853959,
      "grad_norm": 1.4639804363250732,
      "learning_rate": 5.273e-05,
      "loss": 0.3567,
      "step": 10552
    },
    {
      "epoch": 0.5909396348975249,
      "grad_norm": 1.2510403394699097,
      "learning_rate": 5.2735e-05,
      "loss": 0.4231,
      "step": 10553
    },
    {
      "epoch": 0.5909956322096539,
      "grad_norm": 1.335352897644043,
      "learning_rate": 5.274e-05,
      "loss": 0.6167,
      "step": 10554
    },
    {
      "epoch": 0.5910516295217829,
      "grad_norm": 1.1949379444122314,
      "learning_rate": 5.2745e-05,
      "loss": 0.3584,
      "step": 10555
    },
    {
      "epoch": 0.5911076268339119,
      "grad_norm": 1.1833736896514893,
      "learning_rate": 5.275e-05,
      "loss": 0.3027,
      "step": 10556
    },
    {
      "epoch": 0.591163624146041,
      "grad_norm": 1.127034306526184,
      "learning_rate": 5.2755e-05,
      "loss": 0.4002,
      "step": 10557
    },
    {
      "epoch": 0.59121962145817,
      "grad_norm": 1.4355443716049194,
      "learning_rate": 5.2759999999999996e-05,
      "loss": 0.5975,
      "step": 10558
    },
    {
      "epoch": 0.591275618770299,
      "grad_norm": 1.3591961860656738,
      "learning_rate": 5.2765e-05,
      "loss": 0.3527,
      "step": 10559
    },
    {
      "epoch": 0.591331616082428,
      "grad_norm": 1.2719610929489136,
      "learning_rate": 5.277e-05,
      "loss": 0.4086,
      "step": 10560
    },
    {
      "epoch": 0.591387613394557,
      "grad_norm": 1.282096266746521,
      "learning_rate": 5.277500000000001e-05,
      "loss": 0.3531,
      "step": 10561
    },
    {
      "epoch": 0.591443610706686,
      "grad_norm": 2.146393060684204,
      "learning_rate": 5.2780000000000006e-05,
      "loss": 0.5905,
      "step": 10562
    },
    {
      "epoch": 0.5914996080188151,
      "grad_norm": 1.8344500064849854,
      "learning_rate": 5.2785000000000004e-05,
      "loss": 0.4916,
      "step": 10563
    },
    {
      "epoch": 0.5915556053309441,
      "grad_norm": 1.2841085195541382,
      "learning_rate": 5.279000000000001e-05,
      "loss": 0.4154,
      "step": 10564
    },
    {
      "epoch": 0.5916116026430731,
      "grad_norm": 1.4600125551223755,
      "learning_rate": 5.2795000000000005e-05,
      "loss": 0.5258,
      "step": 10565
    },
    {
      "epoch": 0.5916675999552021,
      "grad_norm": 1.319330096244812,
      "learning_rate": 5.28e-05,
      "loss": 0.4115,
      "step": 10566
    },
    {
      "epoch": 0.5917235972673311,
      "grad_norm": 1.253957748413086,
      "learning_rate": 5.280500000000001e-05,
      "loss": 0.4111,
      "step": 10567
    },
    {
      "epoch": 0.5917795945794602,
      "grad_norm": 2.324519634246826,
      "learning_rate": 5.2810000000000004e-05,
      "loss": 0.496,
      "step": 10568
    },
    {
      "epoch": 0.5918355918915892,
      "grad_norm": 1.4242885112762451,
      "learning_rate": 5.2815e-05,
      "loss": 0.4094,
      "step": 10569
    },
    {
      "epoch": 0.5918915892037182,
      "grad_norm": 1.340866208076477,
      "learning_rate": 5.2820000000000006e-05,
      "loss": 0.3748,
      "step": 10570
    },
    {
      "epoch": 0.5919475865158472,
      "grad_norm": 1.527490496635437,
      "learning_rate": 5.2825e-05,
      "loss": 0.5074,
      "step": 10571
    },
    {
      "epoch": 0.5920035838279762,
      "grad_norm": 1.5297170877456665,
      "learning_rate": 5.283e-05,
      "loss": 0.4071,
      "step": 10572
    },
    {
      "epoch": 0.5920595811401053,
      "grad_norm": 1.1640164852142334,
      "learning_rate": 5.2835e-05,
      "loss": 0.5012,
      "step": 10573
    },
    {
      "epoch": 0.5921155784522343,
      "grad_norm": 1.3273276090621948,
      "learning_rate": 5.284e-05,
      "loss": 0.3736,
      "step": 10574
    },
    {
      "epoch": 0.5921715757643633,
      "grad_norm": 1.29637610912323,
      "learning_rate": 5.2845e-05,
      "loss": 0.4081,
      "step": 10575
    },
    {
      "epoch": 0.5922275730764923,
      "grad_norm": 1.5078063011169434,
      "learning_rate": 5.285e-05,
      "loss": 0.8388,
      "step": 10576
    },
    {
      "epoch": 0.5922835703886213,
      "grad_norm": 1.3077116012573242,
      "learning_rate": 5.2855e-05,
      "loss": 0.5383,
      "step": 10577
    },
    {
      "epoch": 0.5923395677007504,
      "grad_norm": 1.2726671695709229,
      "learning_rate": 5.286e-05,
      "loss": 0.4237,
      "step": 10578
    },
    {
      "epoch": 0.5923955650128794,
      "grad_norm": 1.197482943534851,
      "learning_rate": 5.2864999999999996e-05,
      "loss": 0.352,
      "step": 10579
    },
    {
      "epoch": 0.5924515623250084,
      "grad_norm": 1.6226192712783813,
      "learning_rate": 5.287e-05,
      "loss": 0.6134,
      "step": 10580
    },
    {
      "epoch": 0.5925075596371374,
      "grad_norm": 1.1241241693496704,
      "learning_rate": 5.2875000000000005e-05,
      "loss": 0.5402,
      "step": 10581
    },
    {
      "epoch": 0.5925635569492664,
      "grad_norm": 1.4471862316131592,
      "learning_rate": 5.288000000000001e-05,
      "loss": 0.4068,
      "step": 10582
    },
    {
      "epoch": 0.5926195542613955,
      "grad_norm": 2.0013887882232666,
      "learning_rate": 5.2885000000000006e-05,
      "loss": 0.4744,
      "step": 10583
    },
    {
      "epoch": 0.5926755515735245,
      "grad_norm": 1.4567906856536865,
      "learning_rate": 5.2890000000000004e-05,
      "loss": 0.5836,
      "step": 10584
    },
    {
      "epoch": 0.5927315488856535,
      "grad_norm": 1.3054038286209106,
      "learning_rate": 5.289500000000001e-05,
      "loss": 0.4638,
      "step": 10585
    },
    {
      "epoch": 0.5927875461977825,
      "grad_norm": 1.1921995878219604,
      "learning_rate": 5.2900000000000005e-05,
      "loss": 0.3657,
      "step": 10586
    },
    {
      "epoch": 0.5928435435099115,
      "grad_norm": 1.283695101737976,
      "learning_rate": 5.2905e-05,
      "loss": 0.4191,
      "step": 10587
    },
    {
      "epoch": 0.5928995408220405,
      "grad_norm": 1.2014490365982056,
      "learning_rate": 5.291000000000001e-05,
      "loss": 0.3812,
      "step": 10588
    },
    {
      "epoch": 0.5929555381341696,
      "grad_norm": 1.3544204235076904,
      "learning_rate": 5.2915000000000004e-05,
      "loss": 0.3469,
      "step": 10589
    },
    {
      "epoch": 0.5930115354462986,
      "grad_norm": 1.3510054349899292,
      "learning_rate": 5.292e-05,
      "loss": 0.442,
      "step": 10590
    },
    {
      "epoch": 0.5930675327584276,
      "grad_norm": 1.387406349182129,
      "learning_rate": 5.2925000000000006e-05,
      "loss": 0.5564,
      "step": 10591
    },
    {
      "epoch": 0.5931235300705566,
      "grad_norm": 1.2237368822097778,
      "learning_rate": 5.293e-05,
      "loss": 0.4988,
      "step": 10592
    },
    {
      "epoch": 0.5931795273826856,
      "grad_norm": 1.5499701499938965,
      "learning_rate": 5.2935e-05,
      "loss": 0.4317,
      "step": 10593
    },
    {
      "epoch": 0.5932355246948147,
      "grad_norm": 6.83128547668457,
      "learning_rate": 5.294e-05,
      "loss": 0.4112,
      "step": 10594
    },
    {
      "epoch": 0.5932915220069437,
      "grad_norm": 1.2354567050933838,
      "learning_rate": 5.2945e-05,
      "loss": 0.4185,
      "step": 10595
    },
    {
      "epoch": 0.5933475193190727,
      "grad_norm": 1.5912731885910034,
      "learning_rate": 5.295e-05,
      "loss": 0.4071,
      "step": 10596
    },
    {
      "epoch": 0.5934035166312017,
      "grad_norm": 1.1895127296447754,
      "learning_rate": 5.2955e-05,
      "loss": 0.4107,
      "step": 10597
    },
    {
      "epoch": 0.5934595139433307,
      "grad_norm": 1.3144574165344238,
      "learning_rate": 5.296e-05,
      "loss": 0.3937,
      "step": 10598
    },
    {
      "epoch": 0.5935155112554598,
      "grad_norm": 1.428004503250122,
      "learning_rate": 5.2965e-05,
      "loss": 0.4642,
      "step": 10599
    },
    {
      "epoch": 0.5935715085675888,
      "grad_norm": 1.0706285238265991,
      "learning_rate": 5.2969999999999996e-05,
      "loss": 0.4732,
      "step": 10600
    },
    {
      "epoch": 0.5936275058797178,
      "grad_norm": 1.2697863578796387,
      "learning_rate": 5.297500000000001e-05,
      "loss": 0.4415,
      "step": 10601
    },
    {
      "epoch": 0.5936835031918468,
      "grad_norm": 1.2962846755981445,
      "learning_rate": 5.2980000000000004e-05,
      "loss": 0.4408,
      "step": 10602
    },
    {
      "epoch": 0.5937395005039758,
      "grad_norm": 1.3874752521514893,
      "learning_rate": 5.298500000000001e-05,
      "loss": 0.395,
      "step": 10603
    },
    {
      "epoch": 0.5937954978161049,
      "grad_norm": 1.258508563041687,
      "learning_rate": 5.2990000000000006e-05,
      "loss": 0.4574,
      "step": 10604
    },
    {
      "epoch": 0.5938514951282339,
      "grad_norm": 1.343687891960144,
      "learning_rate": 5.2995e-05,
      "loss": 0.4034,
      "step": 10605
    },
    {
      "epoch": 0.5939074924403629,
      "grad_norm": 1.2250497341156006,
      "learning_rate": 5.300000000000001e-05,
      "loss": 0.3901,
      "step": 10606
    },
    {
      "epoch": 0.5939634897524919,
      "grad_norm": 0.9959533214569092,
      "learning_rate": 5.3005000000000005e-05,
      "loss": 0.2805,
      "step": 10607
    },
    {
      "epoch": 0.5940194870646209,
      "grad_norm": 1.1464002132415771,
      "learning_rate": 5.301e-05,
      "loss": 0.3531,
      "step": 10608
    },
    {
      "epoch": 0.59407548437675,
      "grad_norm": 1.2639087438583374,
      "learning_rate": 5.3015000000000007e-05,
      "loss": 0.3548,
      "step": 10609
    },
    {
      "epoch": 0.594131481688879,
      "grad_norm": 1.5469439029693604,
      "learning_rate": 5.3020000000000004e-05,
      "loss": 0.6987,
      "step": 10610
    },
    {
      "epoch": 0.594187479001008,
      "grad_norm": 1.2237581014633179,
      "learning_rate": 5.3025e-05,
      "loss": 0.4156,
      "step": 10611
    },
    {
      "epoch": 0.594243476313137,
      "grad_norm": 1.4172309637069702,
      "learning_rate": 5.303e-05,
      "loss": 0.5293,
      "step": 10612
    },
    {
      "epoch": 0.594299473625266,
      "grad_norm": 1.323311686515808,
      "learning_rate": 5.3035e-05,
      "loss": 0.3676,
      "step": 10613
    },
    {
      "epoch": 0.594355470937395,
      "grad_norm": 1.3635905981063843,
      "learning_rate": 5.304e-05,
      "loss": 0.4182,
      "step": 10614
    },
    {
      "epoch": 0.5944114682495241,
      "grad_norm": 1.357318639755249,
      "learning_rate": 5.3045e-05,
      "loss": 0.4527,
      "step": 10615
    },
    {
      "epoch": 0.5944674655616531,
      "grad_norm": 1.4004288911819458,
      "learning_rate": 5.305e-05,
      "loss": 0.4716,
      "step": 10616
    },
    {
      "epoch": 0.5945234628737821,
      "grad_norm": 1.1210845708847046,
      "learning_rate": 5.3055e-05,
      "loss": 0.3992,
      "step": 10617
    },
    {
      "epoch": 0.5945794601859111,
      "grad_norm": 1.0251448154449463,
      "learning_rate": 5.306e-05,
      "loss": 0.354,
      "step": 10618
    },
    {
      "epoch": 0.5946354574980401,
      "grad_norm": 1.2607853412628174,
      "learning_rate": 5.3065e-05,
      "loss": 0.453,
      "step": 10619
    },
    {
      "epoch": 0.5946914548101692,
      "grad_norm": 1.6649878025054932,
      "learning_rate": 5.307e-05,
      "loss": 0.5101,
      "step": 10620
    },
    {
      "epoch": 0.5947474521222982,
      "grad_norm": 1.504562497138977,
      "learning_rate": 5.307500000000001e-05,
      "loss": 0.5634,
      "step": 10621
    },
    {
      "epoch": 0.5948034494344272,
      "grad_norm": 1.3259096145629883,
      "learning_rate": 5.308000000000001e-05,
      "loss": 0.4427,
      "step": 10622
    },
    {
      "epoch": 0.5948594467465562,
      "grad_norm": 1.3511524200439453,
      "learning_rate": 5.3085000000000004e-05,
      "loss": 0.475,
      "step": 10623
    },
    {
      "epoch": 0.5949154440586851,
      "grad_norm": 1.1541205644607544,
      "learning_rate": 5.309000000000001e-05,
      "loss": 0.4037,
      "step": 10624
    },
    {
      "epoch": 0.5949714413708141,
      "grad_norm": 1.1316460371017456,
      "learning_rate": 5.3095000000000006e-05,
      "loss": 0.3553,
      "step": 10625
    },
    {
      "epoch": 0.5950274386829432,
      "grad_norm": 1.2613017559051514,
      "learning_rate": 5.31e-05,
      "loss": 0.4714,
      "step": 10626
    },
    {
      "epoch": 0.5950834359950722,
      "grad_norm": 1.2521265745162964,
      "learning_rate": 5.310500000000001e-05,
      "loss": 0.4846,
      "step": 10627
    },
    {
      "epoch": 0.5951394333072012,
      "grad_norm": 1.2765040397644043,
      "learning_rate": 5.3110000000000005e-05,
      "loss": 0.4363,
      "step": 10628
    },
    {
      "epoch": 0.5951954306193302,
      "grad_norm": 1.2572119235992432,
      "learning_rate": 5.3115e-05,
      "loss": 0.4709,
      "step": 10629
    },
    {
      "epoch": 0.5952514279314592,
      "grad_norm": 1.1742247343063354,
      "learning_rate": 5.3120000000000006e-05,
      "loss": 0.4415,
      "step": 10630
    },
    {
      "epoch": 0.5953074252435883,
      "grad_norm": 1.5842454433441162,
      "learning_rate": 5.3125000000000004e-05,
      "loss": 0.4154,
      "step": 10631
    },
    {
      "epoch": 0.5953634225557173,
      "grad_norm": 1.1859489679336548,
      "learning_rate": 5.313e-05,
      "loss": 0.3858,
      "step": 10632
    },
    {
      "epoch": 0.5954194198678463,
      "grad_norm": 1.2519792318344116,
      "learning_rate": 5.3135e-05,
      "loss": 0.368,
      "step": 10633
    },
    {
      "epoch": 0.5954754171799753,
      "grad_norm": 1.3703736066818237,
      "learning_rate": 5.314e-05,
      "loss": 0.758,
      "step": 10634
    },
    {
      "epoch": 0.5955314144921043,
      "grad_norm": 1.2715849876403809,
      "learning_rate": 5.3145e-05,
      "loss": 0.4004,
      "step": 10635
    },
    {
      "epoch": 0.5955874118042334,
      "grad_norm": 1.769635558128357,
      "learning_rate": 5.315e-05,
      "loss": 0.4241,
      "step": 10636
    },
    {
      "epoch": 0.5956434091163624,
      "grad_norm": 1.0612351894378662,
      "learning_rate": 5.3155e-05,
      "loss": 0.3624,
      "step": 10637
    },
    {
      "epoch": 0.5956994064284914,
      "grad_norm": 1.1915764808654785,
      "learning_rate": 5.316e-05,
      "loss": 0.4014,
      "step": 10638
    },
    {
      "epoch": 0.5957554037406204,
      "grad_norm": 2.9453630447387695,
      "learning_rate": 5.3165e-05,
      "loss": 0.4261,
      "step": 10639
    },
    {
      "epoch": 0.5958114010527494,
      "grad_norm": 1.4035197496414185,
      "learning_rate": 5.317e-05,
      "loss": 0.4261,
      "step": 10640
    },
    {
      "epoch": 0.5958673983648785,
      "grad_norm": 1.150656819343567,
      "learning_rate": 5.3175e-05,
      "loss": 0.4096,
      "step": 10641
    },
    {
      "epoch": 0.5959233956770075,
      "grad_norm": 1.2750046253204346,
      "learning_rate": 5.318000000000001e-05,
      "loss": 0.4525,
      "step": 10642
    },
    {
      "epoch": 0.5959793929891365,
      "grad_norm": 1.4290735721588135,
      "learning_rate": 5.318500000000001e-05,
      "loss": 0.5292,
      "step": 10643
    },
    {
      "epoch": 0.5960353903012655,
      "grad_norm": 1.266943335533142,
      "learning_rate": 5.3190000000000004e-05,
      "loss": 0.3955,
      "step": 10644
    },
    {
      "epoch": 0.5960913876133945,
      "grad_norm": 1.522802472114563,
      "learning_rate": 5.319500000000001e-05,
      "loss": 0.7771,
      "step": 10645
    },
    {
      "epoch": 0.5961473849255235,
      "grad_norm": 1.3386750221252441,
      "learning_rate": 5.3200000000000006e-05,
      "loss": 0.4152,
      "step": 10646
    },
    {
      "epoch": 0.5962033822376526,
      "grad_norm": 1.1313281059265137,
      "learning_rate": 5.3205e-05,
      "loss": 0.4505,
      "step": 10647
    },
    {
      "epoch": 0.5962593795497816,
      "grad_norm": 1.1104485988616943,
      "learning_rate": 5.321000000000001e-05,
      "loss": 0.4002,
      "step": 10648
    },
    {
      "epoch": 0.5963153768619106,
      "grad_norm": 1.2520898580551147,
      "learning_rate": 5.3215000000000005e-05,
      "loss": 0.3537,
      "step": 10649
    },
    {
      "epoch": 0.5963713741740396,
      "grad_norm": 1.1591063737869263,
      "learning_rate": 5.322e-05,
      "loss": 0.4567,
      "step": 10650
    },
    {
      "epoch": 0.5964273714861686,
      "grad_norm": 1.056723713874817,
      "learning_rate": 5.3225e-05,
      "loss": 0.3858,
      "step": 10651
    },
    {
      "epoch": 0.5964833687982977,
      "grad_norm": 1.5650540590286255,
      "learning_rate": 5.3230000000000004e-05,
      "loss": 0.4986,
      "step": 10652
    },
    {
      "epoch": 0.5965393661104267,
      "grad_norm": 1.3419244289398193,
      "learning_rate": 5.3235e-05,
      "loss": 0.4247,
      "step": 10653
    },
    {
      "epoch": 0.5965953634225557,
      "grad_norm": 1.10451078414917,
      "learning_rate": 5.324e-05,
      "loss": 0.4241,
      "step": 10654
    },
    {
      "epoch": 0.5966513607346847,
      "grad_norm": 1.2173242568969727,
      "learning_rate": 5.3245e-05,
      "loss": 0.5483,
      "step": 10655
    },
    {
      "epoch": 0.5967073580468137,
      "grad_norm": 1.4182380437850952,
      "learning_rate": 5.325e-05,
      "loss": 0.4532,
      "step": 10656
    },
    {
      "epoch": 0.5967633553589428,
      "grad_norm": 1.1876804828643799,
      "learning_rate": 5.3255e-05,
      "loss": 0.3835,
      "step": 10657
    },
    {
      "epoch": 0.5968193526710718,
      "grad_norm": 1.3040655851364136,
      "learning_rate": 5.326e-05,
      "loss": 0.559,
      "step": 10658
    },
    {
      "epoch": 0.5968753499832008,
      "grad_norm": 1.5240356922149658,
      "learning_rate": 5.3265e-05,
      "loss": 0.456,
      "step": 10659
    },
    {
      "epoch": 0.5969313472953298,
      "grad_norm": 1.3377245664596558,
      "learning_rate": 5.3269999999999996e-05,
      "loss": 0.424,
      "step": 10660
    },
    {
      "epoch": 0.5969873446074588,
      "grad_norm": 1.2126078605651855,
      "learning_rate": 5.3274999999999994e-05,
      "loss": 0.5067,
      "step": 10661
    },
    {
      "epoch": 0.5970433419195879,
      "grad_norm": 1.19891357421875,
      "learning_rate": 5.3280000000000005e-05,
      "loss": 0.3772,
      "step": 10662
    },
    {
      "epoch": 0.5970993392317169,
      "grad_norm": 1.408841609954834,
      "learning_rate": 5.328500000000001e-05,
      "loss": 0.4489,
      "step": 10663
    },
    {
      "epoch": 0.5971553365438459,
      "grad_norm": 1.191271185874939,
      "learning_rate": 5.3290000000000006e-05,
      "loss": 0.375,
      "step": 10664
    },
    {
      "epoch": 0.5972113338559749,
      "grad_norm": 1.279780387878418,
      "learning_rate": 5.3295000000000004e-05,
      "loss": 0.38,
      "step": 10665
    },
    {
      "epoch": 0.5972673311681039,
      "grad_norm": 1.4989320039749146,
      "learning_rate": 5.330000000000001e-05,
      "loss": 0.583,
      "step": 10666
    },
    {
      "epoch": 0.597323328480233,
      "grad_norm": 1.1831014156341553,
      "learning_rate": 5.3305000000000005e-05,
      "loss": 0.3763,
      "step": 10667
    },
    {
      "epoch": 0.597379325792362,
      "grad_norm": 1.2059293985366821,
      "learning_rate": 5.331e-05,
      "loss": 0.4694,
      "step": 10668
    },
    {
      "epoch": 0.597435323104491,
      "grad_norm": 1.4759252071380615,
      "learning_rate": 5.331500000000001e-05,
      "loss": 0.5221,
      "step": 10669
    },
    {
      "epoch": 0.59749132041662,
      "grad_norm": 2.0236058235168457,
      "learning_rate": 5.3320000000000004e-05,
      "loss": 0.9048,
      "step": 10670
    },
    {
      "epoch": 0.597547317728749,
      "grad_norm": 1.2076748609542847,
      "learning_rate": 5.3325e-05,
      "loss": 0.3366,
      "step": 10671
    },
    {
      "epoch": 0.597603315040878,
      "grad_norm": 3.8239355087280273,
      "learning_rate": 5.333e-05,
      "loss": 0.3727,
      "step": 10672
    },
    {
      "epoch": 0.5976593123530071,
      "grad_norm": 1.216123104095459,
      "learning_rate": 5.3335000000000003e-05,
      "loss": 0.4195,
      "step": 10673
    },
    {
      "epoch": 0.5977153096651361,
      "grad_norm": 1.2644193172454834,
      "learning_rate": 5.334e-05,
      "loss": 0.4543,
      "step": 10674
    },
    {
      "epoch": 0.5977713069772651,
      "grad_norm": 1.345656394958496,
      "learning_rate": 5.3345e-05,
      "loss": 0.3471,
      "step": 10675
    },
    {
      "epoch": 0.5978273042893941,
      "grad_norm": 1.1759001016616821,
      "learning_rate": 5.335e-05,
      "loss": 0.4872,
      "step": 10676
    },
    {
      "epoch": 0.5978833016015231,
      "grad_norm": 1.7768365144729614,
      "learning_rate": 5.3355e-05,
      "loss": 0.5289,
      "step": 10677
    },
    {
      "epoch": 0.5979392989136522,
      "grad_norm": 1.2655423879623413,
      "learning_rate": 5.336e-05,
      "loss": 0.3933,
      "step": 10678
    },
    {
      "epoch": 0.5979952962257812,
      "grad_norm": 1.3385236263275146,
      "learning_rate": 5.3365e-05,
      "loss": 0.4937,
      "step": 10679
    },
    {
      "epoch": 0.5980512935379102,
      "grad_norm": 1.1813132762908936,
      "learning_rate": 5.337e-05,
      "loss": 0.3674,
      "step": 10680
    },
    {
      "epoch": 0.5981072908500392,
      "grad_norm": 1.2993154525756836,
      "learning_rate": 5.3374999999999996e-05,
      "loss": 0.4463,
      "step": 10681
    },
    {
      "epoch": 0.5981632881621682,
      "grad_norm": 1.3039528131484985,
      "learning_rate": 5.338000000000001e-05,
      "loss": 0.4434,
      "step": 10682
    },
    {
      "epoch": 0.5982192854742973,
      "grad_norm": 1.5547949075698853,
      "learning_rate": 5.3385000000000005e-05,
      "loss": 0.4416,
      "step": 10683
    },
    {
      "epoch": 0.5982752827864263,
      "grad_norm": 1.3940852880477905,
      "learning_rate": 5.339000000000001e-05,
      "loss": 0.4646,
      "step": 10684
    },
    {
      "epoch": 0.5983312800985553,
      "grad_norm": 1.3054916858673096,
      "learning_rate": 5.3395000000000006e-05,
      "loss": 0.3423,
      "step": 10685
    },
    {
      "epoch": 0.5983872774106843,
      "grad_norm": 1.3913575410842896,
      "learning_rate": 5.3400000000000004e-05,
      "loss": 0.4526,
      "step": 10686
    },
    {
      "epoch": 0.5984432747228133,
      "grad_norm": 1.3938846588134766,
      "learning_rate": 5.340500000000001e-05,
      "loss": 0.4867,
      "step": 10687
    },
    {
      "epoch": 0.5984992720349424,
      "grad_norm": 1.3865050077438354,
      "learning_rate": 5.3410000000000005e-05,
      "loss": 0.5223,
      "step": 10688
    },
    {
      "epoch": 0.5985552693470714,
      "grad_norm": 1.180454969406128,
      "learning_rate": 5.3415e-05,
      "loss": 0.3896,
      "step": 10689
    },
    {
      "epoch": 0.5986112666592004,
      "grad_norm": 1.3324973583221436,
      "learning_rate": 5.342e-05,
      "loss": 0.4556,
      "step": 10690
    },
    {
      "epoch": 0.5986672639713294,
      "grad_norm": 1.3338416814804077,
      "learning_rate": 5.3425000000000004e-05,
      "loss": 0.3878,
      "step": 10691
    },
    {
      "epoch": 0.5987232612834584,
      "grad_norm": 1.3270137310028076,
      "learning_rate": 5.343e-05,
      "loss": 0.478,
      "step": 10692
    },
    {
      "epoch": 0.5987792585955874,
      "grad_norm": 1.1505717039108276,
      "learning_rate": 5.3435e-05,
      "loss": 0.389,
      "step": 10693
    },
    {
      "epoch": 0.5988352559077165,
      "grad_norm": 1.434771180152893,
      "learning_rate": 5.344e-05,
      "loss": 0.3462,
      "step": 10694
    },
    {
      "epoch": 0.5988912532198455,
      "grad_norm": 1.1641889810562134,
      "learning_rate": 5.3445e-05,
      "loss": 0.424,
      "step": 10695
    },
    {
      "epoch": 0.5989472505319745,
      "grad_norm": 1.2255793809890747,
      "learning_rate": 5.345e-05,
      "loss": 0.5723,
      "step": 10696
    },
    {
      "epoch": 0.5990032478441035,
      "grad_norm": 1.0824426412582397,
      "learning_rate": 5.3455e-05,
      "loss": 0.4253,
      "step": 10697
    },
    {
      "epoch": 0.5990592451562325,
      "grad_norm": 1.2578072547912598,
      "learning_rate": 5.346e-05,
      "loss": 0.5385,
      "step": 10698
    },
    {
      "epoch": 0.5991152424683616,
      "grad_norm": 1.221637487411499,
      "learning_rate": 5.3465e-05,
      "loss": 0.4711,
      "step": 10699
    },
    {
      "epoch": 0.5991712397804906,
      "grad_norm": 1.1506383419036865,
      "learning_rate": 5.3469999999999995e-05,
      "loss": 0.3996,
      "step": 10700
    },
    {
      "epoch": 0.5992272370926196,
      "grad_norm": 1.373428463935852,
      "learning_rate": 5.3475e-05,
      "loss": 0.3844,
      "step": 10701
    },
    {
      "epoch": 0.5992832344047486,
      "grad_norm": 1.593705415725708,
      "learning_rate": 5.348000000000001e-05,
      "loss": 0.3559,
      "step": 10702
    },
    {
      "epoch": 0.5993392317168776,
      "grad_norm": 1.0618690252304077,
      "learning_rate": 5.348500000000001e-05,
      "loss": 0.4307,
      "step": 10703
    },
    {
      "epoch": 0.5993952290290067,
      "grad_norm": 1.1788705587387085,
      "learning_rate": 5.3490000000000005e-05,
      "loss": 0.4331,
      "step": 10704
    },
    {
      "epoch": 0.5994512263411357,
      "grad_norm": 1.2288025617599487,
      "learning_rate": 5.349500000000001e-05,
      "loss": 0.3987,
      "step": 10705
    },
    {
      "epoch": 0.5995072236532647,
      "grad_norm": 1.4503600597381592,
      "learning_rate": 5.3500000000000006e-05,
      "loss": 0.4142,
      "step": 10706
    },
    {
      "epoch": 0.5995632209653936,
      "grad_norm": 1.3712904453277588,
      "learning_rate": 5.3505000000000004e-05,
      "loss": 0.5967,
      "step": 10707
    },
    {
      "epoch": 0.5996192182775226,
      "grad_norm": 1.2276537418365479,
      "learning_rate": 5.351000000000001e-05,
      "loss": 0.3684,
      "step": 10708
    },
    {
      "epoch": 0.5996752155896516,
      "grad_norm": 1.0762742757797241,
      "learning_rate": 5.3515000000000005e-05,
      "loss": 0.3283,
      "step": 10709
    },
    {
      "epoch": 0.5997312129017807,
      "grad_norm": 1.382184624671936,
      "learning_rate": 5.352e-05,
      "loss": 0.6539,
      "step": 10710
    },
    {
      "epoch": 0.5997872102139097,
      "grad_norm": 1.044752836227417,
      "learning_rate": 5.3525e-05,
      "loss": 0.3505,
      "step": 10711
    },
    {
      "epoch": 0.5998432075260387,
      "grad_norm": 1.081892490386963,
      "learning_rate": 5.3530000000000004e-05,
      "loss": 0.4271,
      "step": 10712
    },
    {
      "epoch": 0.5998992048381677,
      "grad_norm": 1.3416136503219604,
      "learning_rate": 5.3535e-05,
      "loss": 0.3771,
      "step": 10713
    },
    {
      "epoch": 0.5999552021502967,
      "grad_norm": 1.536210060119629,
      "learning_rate": 5.354e-05,
      "loss": 0.6779,
      "step": 10714
    },
    {
      "epoch": 0.6000111994624258,
      "grad_norm": 1.3057374954223633,
      "learning_rate": 5.3545e-05,
      "loss": 0.4476,
      "step": 10715
    },
    {
      "epoch": 0.6000671967745548,
      "grad_norm": 1.132282018661499,
      "learning_rate": 5.355e-05,
      "loss": 0.3702,
      "step": 10716
    },
    {
      "epoch": 0.6001231940866838,
      "grad_norm": 1.3011412620544434,
      "learning_rate": 5.3555e-05,
      "loss": 0.4967,
      "step": 10717
    },
    {
      "epoch": 0.6001791913988128,
      "grad_norm": 1.2960222959518433,
      "learning_rate": 5.356e-05,
      "loss": 0.3776,
      "step": 10718
    },
    {
      "epoch": 0.6002351887109418,
      "grad_norm": 1.1946500539779663,
      "learning_rate": 5.3565e-05,
      "loss": 0.3625,
      "step": 10719
    },
    {
      "epoch": 0.6002911860230709,
      "grad_norm": 1.1649937629699707,
      "learning_rate": 5.357e-05,
      "loss": 0.4072,
      "step": 10720
    },
    {
      "epoch": 0.6003471833351999,
      "grad_norm": 1.5056583881378174,
      "learning_rate": 5.3574999999999994e-05,
      "loss": 0.6115,
      "step": 10721
    },
    {
      "epoch": 0.6004031806473289,
      "grad_norm": 1.3560359477996826,
      "learning_rate": 5.3580000000000005e-05,
      "loss": 0.4659,
      "step": 10722
    },
    {
      "epoch": 0.6004591779594579,
      "grad_norm": 1.4856892824172974,
      "learning_rate": 5.358500000000001e-05,
      "loss": 0.5304,
      "step": 10723
    },
    {
      "epoch": 0.6005151752715869,
      "grad_norm": 1.1554310321807861,
      "learning_rate": 5.359000000000001e-05,
      "loss": 0.4317,
      "step": 10724
    },
    {
      "epoch": 0.600571172583716,
      "grad_norm": 1.2974077463150024,
      "learning_rate": 5.3595000000000004e-05,
      "loss": 0.4683,
      "step": 10725
    },
    {
      "epoch": 0.600627169895845,
      "grad_norm": 1.1984494924545288,
      "learning_rate": 5.360000000000001e-05,
      "loss": 0.3942,
      "step": 10726
    },
    {
      "epoch": 0.600683167207974,
      "grad_norm": 1.267159342765808,
      "learning_rate": 5.3605000000000006e-05,
      "loss": 0.3985,
      "step": 10727
    },
    {
      "epoch": 0.600739164520103,
      "grad_norm": 1.2933393716812134,
      "learning_rate": 5.3610000000000003e-05,
      "loss": 0.521,
      "step": 10728
    },
    {
      "epoch": 0.600795161832232,
      "grad_norm": 1.145164132118225,
      "learning_rate": 5.3615e-05,
      "loss": 0.3908,
      "step": 10729
    },
    {
      "epoch": 0.600851159144361,
      "grad_norm": 1.5170440673828125,
      "learning_rate": 5.3620000000000005e-05,
      "loss": 0.4942,
      "step": 10730
    },
    {
      "epoch": 0.6009071564564901,
      "grad_norm": 1.2621502876281738,
      "learning_rate": 5.3625e-05,
      "loss": 0.445,
      "step": 10731
    },
    {
      "epoch": 0.6009631537686191,
      "grad_norm": 1.3321058750152588,
      "learning_rate": 5.363e-05,
      "loss": 0.4805,
      "step": 10732
    },
    {
      "epoch": 0.6010191510807481,
      "grad_norm": 3.6692733764648438,
      "learning_rate": 5.3635000000000004e-05,
      "loss": 0.4295,
      "step": 10733
    },
    {
      "epoch": 0.6010751483928771,
      "grad_norm": 1.2317042350769043,
      "learning_rate": 5.364e-05,
      "loss": 0.3777,
      "step": 10734
    },
    {
      "epoch": 0.6011311457050061,
      "grad_norm": 1.2057650089263916,
      "learning_rate": 5.3645e-05,
      "loss": 0.3517,
      "step": 10735
    },
    {
      "epoch": 0.6011871430171352,
      "grad_norm": 1.4327137470245361,
      "learning_rate": 5.365e-05,
      "loss": 0.4454,
      "step": 10736
    },
    {
      "epoch": 0.6012431403292642,
      "grad_norm": 1.1435508728027344,
      "learning_rate": 5.3655e-05,
      "loss": 0.5331,
      "step": 10737
    },
    {
      "epoch": 0.6012991376413932,
      "grad_norm": 1.0889784097671509,
      "learning_rate": 5.366e-05,
      "loss": 0.4379,
      "step": 10738
    },
    {
      "epoch": 0.6013551349535222,
      "grad_norm": 1.335314393043518,
      "learning_rate": 5.3664999999999995e-05,
      "loss": 0.4332,
      "step": 10739
    },
    {
      "epoch": 0.6014111322656512,
      "grad_norm": 1.3137105703353882,
      "learning_rate": 5.367e-05,
      "loss": 0.4548,
      "step": 10740
    },
    {
      "epoch": 0.6014671295777803,
      "grad_norm": 1.2829149961471558,
      "learning_rate": 5.3675e-05,
      "loss": 0.4708,
      "step": 10741
    },
    {
      "epoch": 0.6015231268899093,
      "grad_norm": 1.2563443183898926,
      "learning_rate": 5.368000000000001e-05,
      "loss": 0.5871,
      "step": 10742
    },
    {
      "epoch": 0.6015791242020383,
      "grad_norm": 1.365319013595581,
      "learning_rate": 5.3685000000000005e-05,
      "loss": 0.3273,
      "step": 10743
    },
    {
      "epoch": 0.6016351215141673,
      "grad_norm": 1.3159781694412231,
      "learning_rate": 5.369000000000001e-05,
      "loss": 0.4736,
      "step": 10744
    },
    {
      "epoch": 0.6016911188262963,
      "grad_norm": 1.2725735902786255,
      "learning_rate": 5.369500000000001e-05,
      "loss": 0.4324,
      "step": 10745
    },
    {
      "epoch": 0.6017471161384254,
      "grad_norm": 1.2729079723358154,
      "learning_rate": 5.3700000000000004e-05,
      "loss": 0.5507,
      "step": 10746
    },
    {
      "epoch": 0.6018031134505544,
      "grad_norm": 1.3266091346740723,
      "learning_rate": 5.370500000000001e-05,
      "loss": 0.3879,
      "step": 10747
    },
    {
      "epoch": 0.6018591107626834,
      "grad_norm": 1.4228885173797607,
      "learning_rate": 5.3710000000000006e-05,
      "loss": 0.4267,
      "step": 10748
    },
    {
      "epoch": 0.6019151080748124,
      "grad_norm": 1.3786468505859375,
      "learning_rate": 5.3715e-05,
      "loss": 0.6096,
      "step": 10749
    },
    {
      "epoch": 0.6019711053869414,
      "grad_norm": 1.2821747064590454,
      "learning_rate": 5.372e-05,
      "loss": 0.3711,
      "step": 10750
    },
    {
      "epoch": 0.6020271026990704,
      "grad_norm": 1.3442124128341675,
      "learning_rate": 5.3725000000000005e-05,
      "loss": 0.5343,
      "step": 10751
    },
    {
      "epoch": 0.6020831000111995,
      "grad_norm": 1.2129201889038086,
      "learning_rate": 5.373e-05,
      "loss": 0.3621,
      "step": 10752
    },
    {
      "epoch": 0.6021390973233285,
      "grad_norm": 1.8313417434692383,
      "learning_rate": 5.3735e-05,
      "loss": 0.3928,
      "step": 10753
    },
    {
      "epoch": 0.6021950946354575,
      "grad_norm": 1.3189783096313477,
      "learning_rate": 5.3740000000000004e-05,
      "loss": 0.4618,
      "step": 10754
    },
    {
      "epoch": 0.6022510919475865,
      "grad_norm": 1.542063593864441,
      "learning_rate": 5.3745e-05,
      "loss": 0.44,
      "step": 10755
    },
    {
      "epoch": 0.6023070892597155,
      "grad_norm": 1.1043028831481934,
      "learning_rate": 5.375e-05,
      "loss": 0.4144,
      "step": 10756
    },
    {
      "epoch": 0.6023630865718446,
      "grad_norm": 1.2992949485778809,
      "learning_rate": 5.3755e-05,
      "loss": 0.5384,
      "step": 10757
    },
    {
      "epoch": 0.6024190838839736,
      "grad_norm": 1.1901628971099854,
      "learning_rate": 5.376e-05,
      "loss": 0.3763,
      "step": 10758
    },
    {
      "epoch": 0.6024750811961026,
      "grad_norm": 1.3112820386886597,
      "learning_rate": 5.3765e-05,
      "loss": 0.4026,
      "step": 10759
    },
    {
      "epoch": 0.6025310785082316,
      "grad_norm": 1.519964575767517,
      "learning_rate": 5.3769999999999995e-05,
      "loss": 0.5919,
      "step": 10760
    },
    {
      "epoch": 0.6025870758203606,
      "grad_norm": 1.2642871141433716,
      "learning_rate": 5.3775e-05,
      "loss": 0.5022,
      "step": 10761
    },
    {
      "epoch": 0.6026430731324897,
      "grad_norm": 1.2007747888565063,
      "learning_rate": 5.378e-05,
      "loss": 0.5792,
      "step": 10762
    },
    {
      "epoch": 0.6026990704446187,
      "grad_norm": 1.247137427330017,
      "learning_rate": 5.378500000000001e-05,
      "loss": 0.3743,
      "step": 10763
    },
    {
      "epoch": 0.6027550677567477,
      "grad_norm": 1.2913554906845093,
      "learning_rate": 5.3790000000000005e-05,
      "loss": 0.4759,
      "step": 10764
    },
    {
      "epoch": 0.6028110650688767,
      "grad_norm": 1.1013917922973633,
      "learning_rate": 5.379500000000001e-05,
      "loss": 0.4122,
      "step": 10765
    },
    {
      "epoch": 0.6028670623810057,
      "grad_norm": 2.8226101398468018,
      "learning_rate": 5.380000000000001e-05,
      "loss": 0.3754,
      "step": 10766
    },
    {
      "epoch": 0.6029230596931348,
      "grad_norm": 1.3585474491119385,
      "learning_rate": 5.3805000000000004e-05,
      "loss": 0.3925,
      "step": 10767
    },
    {
      "epoch": 0.6029790570052638,
      "grad_norm": 0.9887531995773315,
      "learning_rate": 5.381e-05,
      "loss": 0.2893,
      "step": 10768
    },
    {
      "epoch": 0.6030350543173928,
      "grad_norm": 1.1428364515304565,
      "learning_rate": 5.3815000000000006e-05,
      "loss": 0.3991,
      "step": 10769
    },
    {
      "epoch": 0.6030910516295218,
      "grad_norm": 1.4529662132263184,
      "learning_rate": 5.382e-05,
      "loss": 0.4707,
      "step": 10770
    },
    {
      "epoch": 0.6031470489416508,
      "grad_norm": 1.2549299001693726,
      "learning_rate": 5.3825e-05,
      "loss": 0.3181,
      "step": 10771
    },
    {
      "epoch": 0.6032030462537799,
      "grad_norm": 1.3555113077163696,
      "learning_rate": 5.3830000000000005e-05,
      "loss": 0.4483,
      "step": 10772
    },
    {
      "epoch": 0.6032590435659089,
      "grad_norm": 1.4491767883300781,
      "learning_rate": 5.3835e-05,
      "loss": 0.6002,
      "step": 10773
    },
    {
      "epoch": 0.6033150408780379,
      "grad_norm": 1.3413119316101074,
      "learning_rate": 5.384e-05,
      "loss": 0.4649,
      "step": 10774
    },
    {
      "epoch": 0.6033710381901669,
      "grad_norm": 1.6401668787002563,
      "learning_rate": 5.3845000000000004e-05,
      "loss": 0.3622,
      "step": 10775
    },
    {
      "epoch": 0.6034270355022959,
      "grad_norm": 1.498465657234192,
      "learning_rate": 5.385e-05,
      "loss": 0.5619,
      "step": 10776
    },
    {
      "epoch": 0.603483032814425,
      "grad_norm": 1.2065180540084839,
      "learning_rate": 5.3855e-05,
      "loss": 0.5187,
      "step": 10777
    },
    {
      "epoch": 0.603539030126554,
      "grad_norm": 1.4181182384490967,
      "learning_rate": 5.386e-05,
      "loss": 0.4366,
      "step": 10778
    },
    {
      "epoch": 0.603595027438683,
      "grad_norm": 1.4053266048431396,
      "learning_rate": 5.3865e-05,
      "loss": 0.5172,
      "step": 10779
    },
    {
      "epoch": 0.603651024750812,
      "grad_norm": 1.3527300357818604,
      "learning_rate": 5.387e-05,
      "loss": 0.5256,
      "step": 10780
    },
    {
      "epoch": 0.603707022062941,
      "grad_norm": 1.6070576906204224,
      "learning_rate": 5.3874999999999995e-05,
      "loss": 0.5341,
      "step": 10781
    },
    {
      "epoch": 0.60376301937507,
      "grad_norm": 1.2043564319610596,
      "learning_rate": 5.388e-05,
      "loss": 0.338,
      "step": 10782
    },
    {
      "epoch": 0.6038190166871991,
      "grad_norm": 1.2934094667434692,
      "learning_rate": 5.388500000000001e-05,
      "loss": 0.3572,
      "step": 10783
    },
    {
      "epoch": 0.6038750139993281,
      "grad_norm": 1.2597966194152832,
      "learning_rate": 5.389000000000001e-05,
      "loss": 0.4674,
      "step": 10784
    },
    {
      "epoch": 0.6039310113114571,
      "grad_norm": 1.358716607093811,
      "learning_rate": 5.3895000000000005e-05,
      "loss": 0.4766,
      "step": 10785
    },
    {
      "epoch": 0.6039870086235861,
      "grad_norm": 1.2400068044662476,
      "learning_rate": 5.390000000000001e-05,
      "loss": 0.4309,
      "step": 10786
    },
    {
      "epoch": 0.6040430059357151,
      "grad_norm": 1.2446389198303223,
      "learning_rate": 5.3905000000000007e-05,
      "loss": 0.4155,
      "step": 10787
    },
    {
      "epoch": 0.6040990032478442,
      "grad_norm": 1.3253979682922363,
      "learning_rate": 5.3910000000000004e-05,
      "loss": 0.4481,
      "step": 10788
    },
    {
      "epoch": 0.6041550005599732,
      "grad_norm": 1.2350094318389893,
      "learning_rate": 5.3915e-05,
      "loss": 0.3623,
      "step": 10789
    },
    {
      "epoch": 0.6042109978721021,
      "grad_norm": 1.4209998846054077,
      "learning_rate": 5.3920000000000006e-05,
      "loss": 0.4583,
      "step": 10790
    },
    {
      "epoch": 0.6042669951842311,
      "grad_norm": 1.4345864057540894,
      "learning_rate": 5.3925e-05,
      "loss": 0.4316,
      "step": 10791
    },
    {
      "epoch": 0.6043229924963601,
      "grad_norm": 1.2540873289108276,
      "learning_rate": 5.393e-05,
      "loss": 0.4195,
      "step": 10792
    },
    {
      "epoch": 0.6043789898084891,
      "grad_norm": 1.4047833681106567,
      "learning_rate": 5.3935000000000005e-05,
      "loss": 0.4012,
      "step": 10793
    },
    {
      "epoch": 0.6044349871206182,
      "grad_norm": 1.1563358306884766,
      "learning_rate": 5.394e-05,
      "loss": 0.3616,
      "step": 10794
    },
    {
      "epoch": 0.6044909844327472,
      "grad_norm": 1.5559574365615845,
      "learning_rate": 5.3945e-05,
      "loss": 0.4022,
      "step": 10795
    },
    {
      "epoch": 0.6045469817448762,
      "grad_norm": 1.235590934753418,
      "learning_rate": 5.3950000000000004e-05,
      "loss": 0.4277,
      "step": 10796
    },
    {
      "epoch": 0.6046029790570052,
      "grad_norm": 1.1067087650299072,
      "learning_rate": 5.3955e-05,
      "loss": 0.4182,
      "step": 10797
    },
    {
      "epoch": 0.6046589763691342,
      "grad_norm": 1.5357977151870728,
      "learning_rate": 5.396e-05,
      "loss": 0.4248,
      "step": 10798
    },
    {
      "epoch": 0.6047149736812633,
      "grad_norm": 1.1177622079849243,
      "learning_rate": 5.3964999999999996e-05,
      "loss": 0.4271,
      "step": 10799
    },
    {
      "epoch": 0.6047709709933923,
      "grad_norm": 1.2507884502410889,
      "learning_rate": 5.397e-05,
      "loss": 0.4074,
      "step": 10800
    },
    {
      "epoch": 0.6048269683055213,
      "grad_norm": 1.195744276046753,
      "learning_rate": 5.3975e-05,
      "loss": 0.3231,
      "step": 10801
    },
    {
      "epoch": 0.6048829656176503,
      "grad_norm": 1.3030754327774048,
      "learning_rate": 5.3979999999999995e-05,
      "loss": 0.5291,
      "step": 10802
    },
    {
      "epoch": 0.6049389629297793,
      "grad_norm": 1.6797661781311035,
      "learning_rate": 5.3985000000000006e-05,
      "loss": 0.4854,
      "step": 10803
    },
    {
      "epoch": 0.6049949602419084,
      "grad_norm": 1.1389864683151245,
      "learning_rate": 5.399000000000001e-05,
      "loss": 0.4245,
      "step": 10804
    },
    {
      "epoch": 0.6050509575540374,
      "grad_norm": 4.239365577697754,
      "learning_rate": 5.399500000000001e-05,
      "loss": 0.4988,
      "step": 10805
    },
    {
      "epoch": 0.6051069548661664,
      "grad_norm": 1.2595585584640503,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 0.4289,
      "step": 10806
    },
    {
      "epoch": 0.6051629521782954,
      "grad_norm": 1.268800973892212,
      "learning_rate": 5.400500000000001e-05,
      "loss": 0.4455,
      "step": 10807
    },
    {
      "epoch": 0.6052189494904244,
      "grad_norm": 1.621047019958496,
      "learning_rate": 5.4010000000000006e-05,
      "loss": 0.4633,
      "step": 10808
    },
    {
      "epoch": 0.6052749468025534,
      "grad_norm": 1.2500159740447998,
      "learning_rate": 5.4015000000000004e-05,
      "loss": 0.4274,
      "step": 10809
    },
    {
      "epoch": 0.6053309441146825,
      "grad_norm": 1.5942906141281128,
      "learning_rate": 5.402e-05,
      "loss": 0.4211,
      "step": 10810
    },
    {
      "epoch": 0.6053869414268115,
      "grad_norm": 1.32368803024292,
      "learning_rate": 5.4025000000000005e-05,
      "loss": 0.5392,
      "step": 10811
    },
    {
      "epoch": 0.6054429387389405,
      "grad_norm": 1.339520812034607,
      "learning_rate": 5.403e-05,
      "loss": 0.4374,
      "step": 10812
    },
    {
      "epoch": 0.6054989360510695,
      "grad_norm": 2.1490044593811035,
      "learning_rate": 5.4035e-05,
      "loss": 0.4251,
      "step": 10813
    },
    {
      "epoch": 0.6055549333631985,
      "grad_norm": 1.3826438188552856,
      "learning_rate": 5.4040000000000004e-05,
      "loss": 0.4717,
      "step": 10814
    },
    {
      "epoch": 0.6056109306753276,
      "grad_norm": 1.2756162881851196,
      "learning_rate": 5.4045e-05,
      "loss": 0.3501,
      "step": 10815
    },
    {
      "epoch": 0.6056669279874566,
      "grad_norm": 1.0150901079177856,
      "learning_rate": 5.405e-05,
      "loss": 0.366,
      "step": 10816
    },
    {
      "epoch": 0.6057229252995856,
      "grad_norm": 1.6545933485031128,
      "learning_rate": 5.4055e-05,
      "loss": 0.4843,
      "step": 10817
    },
    {
      "epoch": 0.6057789226117146,
      "grad_norm": 1.1055387258529663,
      "learning_rate": 5.406e-05,
      "loss": 0.38,
      "step": 10818
    },
    {
      "epoch": 0.6058349199238436,
      "grad_norm": 1.436523199081421,
      "learning_rate": 5.4065e-05,
      "loss": 0.4003,
      "step": 10819
    },
    {
      "epoch": 0.6058909172359727,
      "grad_norm": 1.28406822681427,
      "learning_rate": 5.4069999999999996e-05,
      "loss": 0.5015,
      "step": 10820
    },
    {
      "epoch": 0.6059469145481017,
      "grad_norm": 1.147472620010376,
      "learning_rate": 5.4075e-05,
      "loss": 0.2851,
      "step": 10821
    },
    {
      "epoch": 0.6060029118602307,
      "grad_norm": 1.0317665338516235,
      "learning_rate": 5.408e-05,
      "loss": 0.3906,
      "step": 10822
    },
    {
      "epoch": 0.6060589091723597,
      "grad_norm": 1.2612948417663574,
      "learning_rate": 5.408500000000001e-05,
      "loss": 0.429,
      "step": 10823
    },
    {
      "epoch": 0.6061149064844887,
      "grad_norm": 1.498570203781128,
      "learning_rate": 5.4090000000000006e-05,
      "loss": 0.535,
      "step": 10824
    },
    {
      "epoch": 0.6061709037966178,
      "grad_norm": 1.2321866750717163,
      "learning_rate": 5.409500000000001e-05,
      "loss": 0.4573,
      "step": 10825
    },
    {
      "epoch": 0.6062269011087468,
      "grad_norm": 1.2394964694976807,
      "learning_rate": 5.410000000000001e-05,
      "loss": 0.4547,
      "step": 10826
    },
    {
      "epoch": 0.6062828984208758,
      "grad_norm": 1.8866934776306152,
      "learning_rate": 5.4105000000000005e-05,
      "loss": 0.4295,
      "step": 10827
    },
    {
      "epoch": 0.6063388957330048,
      "grad_norm": 1.3016738891601562,
      "learning_rate": 5.411e-05,
      "loss": 0.418,
      "step": 10828
    },
    {
      "epoch": 0.6063948930451338,
      "grad_norm": 1.2495415210723877,
      "learning_rate": 5.4115000000000006e-05,
      "loss": 0.3469,
      "step": 10829
    },
    {
      "epoch": 0.6064508903572629,
      "grad_norm": 1.394277811050415,
      "learning_rate": 5.4120000000000004e-05,
      "loss": 0.4012,
      "step": 10830
    },
    {
      "epoch": 0.6065068876693919,
      "grad_norm": 1.276740550994873,
      "learning_rate": 5.4125e-05,
      "loss": 0.4102,
      "step": 10831
    },
    {
      "epoch": 0.6065628849815209,
      "grad_norm": 1.152902364730835,
      "learning_rate": 5.4130000000000005e-05,
      "loss": 0.4301,
      "step": 10832
    },
    {
      "epoch": 0.6066188822936499,
      "grad_norm": 1.3067655563354492,
      "learning_rate": 5.4135e-05,
      "loss": 0.5145,
      "step": 10833
    },
    {
      "epoch": 0.6066748796057789,
      "grad_norm": 1.089138388633728,
      "learning_rate": 5.414e-05,
      "loss": 0.371,
      "step": 10834
    },
    {
      "epoch": 0.606730876917908,
      "grad_norm": 1.266743779182434,
      "learning_rate": 5.4145000000000004e-05,
      "loss": 0.4955,
      "step": 10835
    },
    {
      "epoch": 0.606786874230037,
      "grad_norm": 1.3165127038955688,
      "learning_rate": 5.415e-05,
      "loss": 0.4544,
      "step": 10836
    },
    {
      "epoch": 0.606842871542166,
      "grad_norm": 1.4284101724624634,
      "learning_rate": 5.4155e-05,
      "loss": 0.3966,
      "step": 10837
    },
    {
      "epoch": 0.606898868854295,
      "grad_norm": 1.1028484106063843,
      "learning_rate": 5.4159999999999996e-05,
      "loss": 0.3794,
      "step": 10838
    },
    {
      "epoch": 0.606954866166424,
      "grad_norm": 1.2298623323440552,
      "learning_rate": 5.4165e-05,
      "loss": 0.424,
      "step": 10839
    },
    {
      "epoch": 0.607010863478553,
      "grad_norm": 1.338499903678894,
      "learning_rate": 5.417e-05,
      "loss": 0.4441,
      "step": 10840
    },
    {
      "epoch": 0.6070668607906821,
      "grad_norm": 1.2862958908081055,
      "learning_rate": 5.4174999999999995e-05,
      "loss": 0.4455,
      "step": 10841
    },
    {
      "epoch": 0.6071228581028111,
      "grad_norm": 1.2241559028625488,
      "learning_rate": 5.418e-05,
      "loss": 0.421,
      "step": 10842
    },
    {
      "epoch": 0.6071788554149401,
      "grad_norm": 1.0324537754058838,
      "learning_rate": 5.418500000000001e-05,
      "loss": 0.3749,
      "step": 10843
    },
    {
      "epoch": 0.6072348527270691,
      "grad_norm": 1.3654017448425293,
      "learning_rate": 5.419000000000001e-05,
      "loss": 0.3271,
      "step": 10844
    },
    {
      "epoch": 0.6072908500391981,
      "grad_norm": 1.1245365142822266,
      "learning_rate": 5.4195000000000005e-05,
      "loss": 0.3942,
      "step": 10845
    },
    {
      "epoch": 0.6073468473513272,
      "grad_norm": 1.579898715019226,
      "learning_rate": 5.420000000000001e-05,
      "loss": 0.4232,
      "step": 10846
    },
    {
      "epoch": 0.6074028446634562,
      "grad_norm": 1.23211669921875,
      "learning_rate": 5.420500000000001e-05,
      "loss": 0.4372,
      "step": 10847
    },
    {
      "epoch": 0.6074588419755852,
      "grad_norm": 1.2611342668533325,
      "learning_rate": 5.4210000000000004e-05,
      "loss": 0.5056,
      "step": 10848
    },
    {
      "epoch": 0.6075148392877142,
      "grad_norm": 1.1216042041778564,
      "learning_rate": 5.4215e-05,
      "loss": 0.3703,
      "step": 10849
    },
    {
      "epoch": 0.6075708365998432,
      "grad_norm": 1.1612191200256348,
      "learning_rate": 5.4220000000000006e-05,
      "loss": 0.4331,
      "step": 10850
    },
    {
      "epoch": 0.6076268339119723,
      "grad_norm": 1.5577188730239868,
      "learning_rate": 5.4225000000000003e-05,
      "loss": 0.6184,
      "step": 10851
    },
    {
      "epoch": 0.6076828312241013,
      "grad_norm": 1.3442915678024292,
      "learning_rate": 5.423e-05,
      "loss": 0.3291,
      "step": 10852
    },
    {
      "epoch": 0.6077388285362303,
      "grad_norm": 1.355039358139038,
      "learning_rate": 5.4235000000000005e-05,
      "loss": 0.6839,
      "step": 10853
    },
    {
      "epoch": 0.6077948258483593,
      "grad_norm": 1.3512647151947021,
      "learning_rate": 5.424e-05,
      "loss": 0.4379,
      "step": 10854
    },
    {
      "epoch": 0.6078508231604883,
      "grad_norm": 1.8527106046676636,
      "learning_rate": 5.4245e-05,
      "loss": 0.3735,
      "step": 10855
    },
    {
      "epoch": 0.6079068204726173,
      "grad_norm": 1.3146089315414429,
      "learning_rate": 5.4250000000000004e-05,
      "loss": 0.3954,
      "step": 10856
    },
    {
      "epoch": 0.6079628177847464,
      "grad_norm": 1.2463274002075195,
      "learning_rate": 5.4255e-05,
      "loss": 0.4262,
      "step": 10857
    },
    {
      "epoch": 0.6080188150968754,
      "grad_norm": 1.1774721145629883,
      "learning_rate": 5.426e-05,
      "loss": 0.3186,
      "step": 10858
    },
    {
      "epoch": 0.6080748124090044,
      "grad_norm": 1.1638809442520142,
      "learning_rate": 5.4264999999999996e-05,
      "loss": 0.3631,
      "step": 10859
    },
    {
      "epoch": 0.6081308097211334,
      "grad_norm": 1.337945580482483,
      "learning_rate": 5.427e-05,
      "loss": 0.3337,
      "step": 10860
    },
    {
      "epoch": 0.6081868070332624,
      "grad_norm": 1.6546392440795898,
      "learning_rate": 5.4275e-05,
      "loss": 0.5638,
      "step": 10861
    },
    {
      "epoch": 0.6082428043453915,
      "grad_norm": 1.39755117893219,
      "learning_rate": 5.4279999999999995e-05,
      "loss": 0.4977,
      "step": 10862
    },
    {
      "epoch": 0.6082988016575205,
      "grad_norm": 1.1643980741500854,
      "learning_rate": 5.4285000000000006e-05,
      "loss": 0.4908,
      "step": 10863
    },
    {
      "epoch": 0.6083547989696495,
      "grad_norm": 1.3000119924545288,
      "learning_rate": 5.429000000000001e-05,
      "loss": 0.5467,
      "step": 10864
    },
    {
      "epoch": 0.6084107962817785,
      "grad_norm": 1.3411471843719482,
      "learning_rate": 5.429500000000001e-05,
      "loss": 0.6696,
      "step": 10865
    },
    {
      "epoch": 0.6084667935939075,
      "grad_norm": 1.2692595720291138,
      "learning_rate": 5.4300000000000005e-05,
      "loss": 0.4051,
      "step": 10866
    },
    {
      "epoch": 0.6085227909060366,
      "grad_norm": 1.143664836883545,
      "learning_rate": 5.4305e-05,
      "loss": 0.3524,
      "step": 10867
    },
    {
      "epoch": 0.6085787882181656,
      "grad_norm": 1.2592768669128418,
      "learning_rate": 5.431000000000001e-05,
      "loss": 0.468,
      "step": 10868
    },
    {
      "epoch": 0.6086347855302946,
      "grad_norm": 4.257694721221924,
      "learning_rate": 5.4315000000000004e-05,
      "loss": 0.375,
      "step": 10869
    },
    {
      "epoch": 0.6086907828424236,
      "grad_norm": 1.3914134502410889,
      "learning_rate": 5.432e-05,
      "loss": 0.4355,
      "step": 10870
    },
    {
      "epoch": 0.6087467801545526,
      "grad_norm": 1.4659074544906616,
      "learning_rate": 5.4325000000000006e-05,
      "loss": 0.6274,
      "step": 10871
    },
    {
      "epoch": 0.6088027774666815,
      "grad_norm": 1.2778185606002808,
      "learning_rate": 5.433e-05,
      "loss": 0.5666,
      "step": 10872
    },
    {
      "epoch": 0.6088587747788106,
      "grad_norm": 1.189117193222046,
      "learning_rate": 5.4335e-05,
      "loss": 0.4109,
      "step": 10873
    },
    {
      "epoch": 0.6089147720909396,
      "grad_norm": 1.122721791267395,
      "learning_rate": 5.4340000000000005e-05,
      "loss": 0.4035,
      "step": 10874
    },
    {
      "epoch": 0.6089707694030686,
      "grad_norm": 1.4179353713989258,
      "learning_rate": 5.4345e-05,
      "loss": 0.5543,
      "step": 10875
    },
    {
      "epoch": 0.6090267667151976,
      "grad_norm": 1.0920861959457397,
      "learning_rate": 5.435e-05,
      "loss": 0.4486,
      "step": 10876
    },
    {
      "epoch": 0.6090827640273266,
      "grad_norm": 1.174911618232727,
      "learning_rate": 5.4355e-05,
      "loss": 0.4765,
      "step": 10877
    },
    {
      "epoch": 0.6091387613394557,
      "grad_norm": 1.4525388479232788,
      "learning_rate": 5.436e-05,
      "loss": 0.4524,
      "step": 10878
    },
    {
      "epoch": 0.6091947586515847,
      "grad_norm": 1.2018752098083496,
      "learning_rate": 5.4365e-05,
      "loss": 0.3196,
      "step": 10879
    },
    {
      "epoch": 0.6092507559637137,
      "grad_norm": 1.4582374095916748,
      "learning_rate": 5.4369999999999996e-05,
      "loss": 0.4108,
      "step": 10880
    },
    {
      "epoch": 0.6093067532758427,
      "grad_norm": 1.1760350465774536,
      "learning_rate": 5.4375e-05,
      "loss": 0.3301,
      "step": 10881
    },
    {
      "epoch": 0.6093627505879717,
      "grad_norm": 1.2863696813583374,
      "learning_rate": 5.438e-05,
      "loss": 0.4477,
      "step": 10882
    },
    {
      "epoch": 0.6094187479001008,
      "grad_norm": 1.2450844049453735,
      "learning_rate": 5.4384999999999995e-05,
      "loss": 0.4449,
      "step": 10883
    },
    {
      "epoch": 0.6094747452122298,
      "grad_norm": 1.1178971529006958,
      "learning_rate": 5.4390000000000006e-05,
      "loss": 0.3883,
      "step": 10884
    },
    {
      "epoch": 0.6095307425243588,
      "grad_norm": 1.490931749343872,
      "learning_rate": 5.439500000000001e-05,
      "loss": 0.3687,
      "step": 10885
    },
    {
      "epoch": 0.6095867398364878,
      "grad_norm": 1.1268008947372437,
      "learning_rate": 5.440000000000001e-05,
      "loss": 0.4334,
      "step": 10886
    },
    {
      "epoch": 0.6096427371486168,
      "grad_norm": 1.184987187385559,
      "learning_rate": 5.4405000000000005e-05,
      "loss": 0.4247,
      "step": 10887
    },
    {
      "epoch": 0.6096987344607459,
      "grad_norm": 1.1118158102035522,
      "learning_rate": 5.441e-05,
      "loss": 0.3663,
      "step": 10888
    },
    {
      "epoch": 0.6097547317728749,
      "grad_norm": 1.3129841089248657,
      "learning_rate": 5.441500000000001e-05,
      "loss": 0.2936,
      "step": 10889
    },
    {
      "epoch": 0.6098107290850039,
      "grad_norm": 1.2024022340774536,
      "learning_rate": 5.4420000000000004e-05,
      "loss": 0.3951,
      "step": 10890
    },
    {
      "epoch": 0.6098667263971329,
      "grad_norm": 1.322929859161377,
      "learning_rate": 5.4425e-05,
      "loss": 0.3611,
      "step": 10891
    },
    {
      "epoch": 0.6099227237092619,
      "grad_norm": 1.272114872932434,
      "learning_rate": 5.4430000000000006e-05,
      "loss": 0.4719,
      "step": 10892
    },
    {
      "epoch": 0.609978721021391,
      "grad_norm": 1.081798791885376,
      "learning_rate": 5.4435e-05,
      "loss": 0.4308,
      "step": 10893
    },
    {
      "epoch": 0.61003471833352,
      "grad_norm": 1.277859091758728,
      "learning_rate": 5.444e-05,
      "loss": 0.4554,
      "step": 10894
    },
    {
      "epoch": 0.610090715645649,
      "grad_norm": 1.3497674465179443,
      "learning_rate": 5.4445000000000005e-05,
      "loss": 0.3705,
      "step": 10895
    },
    {
      "epoch": 0.610146712957778,
      "grad_norm": 1.611820936203003,
      "learning_rate": 5.445e-05,
      "loss": 0.457,
      "step": 10896
    },
    {
      "epoch": 0.610202710269907,
      "grad_norm": 1.240739107131958,
      "learning_rate": 5.4455e-05,
      "loss": 0.4434,
      "step": 10897
    },
    {
      "epoch": 0.610258707582036,
      "grad_norm": 1.3244433403015137,
      "learning_rate": 5.446e-05,
      "loss": 0.3844,
      "step": 10898
    },
    {
      "epoch": 0.6103147048941651,
      "grad_norm": 1.235936164855957,
      "learning_rate": 5.4465e-05,
      "loss": 0.363,
      "step": 10899
    },
    {
      "epoch": 0.6103707022062941,
      "grad_norm": 1.3539587259292603,
      "learning_rate": 5.447e-05,
      "loss": 0.4094,
      "step": 10900
    },
    {
      "epoch": 0.6104266995184231,
      "grad_norm": 1.5547099113464355,
      "learning_rate": 5.4474999999999996e-05,
      "loss": 0.5054,
      "step": 10901
    },
    {
      "epoch": 0.6104826968305521,
      "grad_norm": 1.3546676635742188,
      "learning_rate": 5.448e-05,
      "loss": 0.5038,
      "step": 10902
    },
    {
      "epoch": 0.6105386941426811,
      "grad_norm": 1.2518770694732666,
      "learning_rate": 5.4485e-05,
      "loss": 0.3849,
      "step": 10903
    },
    {
      "epoch": 0.6105946914548102,
      "grad_norm": 1.5321272611618042,
      "learning_rate": 5.449000000000001e-05,
      "loss": 0.4972,
      "step": 10904
    },
    {
      "epoch": 0.6106506887669392,
      "grad_norm": 1.3916819095611572,
      "learning_rate": 5.4495000000000006e-05,
      "loss": 0.4346,
      "step": 10905
    },
    {
      "epoch": 0.6107066860790682,
      "grad_norm": 1.4135522842407227,
      "learning_rate": 5.45e-05,
      "loss": 0.5123,
      "step": 10906
    },
    {
      "epoch": 0.6107626833911972,
      "grad_norm": 1.1799620389938354,
      "learning_rate": 5.450500000000001e-05,
      "loss": 0.4236,
      "step": 10907
    },
    {
      "epoch": 0.6108186807033262,
      "grad_norm": 1.6834977865219116,
      "learning_rate": 5.4510000000000005e-05,
      "loss": 0.4363,
      "step": 10908
    },
    {
      "epoch": 0.6108746780154553,
      "grad_norm": 1.3735376596450806,
      "learning_rate": 5.4515e-05,
      "loss": 0.5093,
      "step": 10909
    },
    {
      "epoch": 0.6109306753275843,
      "grad_norm": 1.518600583076477,
      "learning_rate": 5.4520000000000007e-05,
      "loss": 0.5383,
      "step": 10910
    },
    {
      "epoch": 0.6109866726397133,
      "grad_norm": 1.2262818813323975,
      "learning_rate": 5.4525000000000004e-05,
      "loss": 0.4242,
      "step": 10911
    },
    {
      "epoch": 0.6110426699518423,
      "grad_norm": 1.3563529253005981,
      "learning_rate": 5.453e-05,
      "loss": 0.5936,
      "step": 10912
    },
    {
      "epoch": 0.6110986672639713,
      "grad_norm": 1.1710436344146729,
      "learning_rate": 5.4535000000000006e-05,
      "loss": 0.4508,
      "step": 10913
    },
    {
      "epoch": 0.6111546645761003,
      "grad_norm": 1.396704077720642,
      "learning_rate": 5.454e-05,
      "loss": 0.4649,
      "step": 10914
    },
    {
      "epoch": 0.6112106618882294,
      "grad_norm": 1.7950878143310547,
      "learning_rate": 5.4545e-05,
      "loss": 0.6378,
      "step": 10915
    },
    {
      "epoch": 0.6112666592003584,
      "grad_norm": 1.2352250814437866,
      "learning_rate": 5.455e-05,
      "loss": 0.4524,
      "step": 10916
    },
    {
      "epoch": 0.6113226565124874,
      "grad_norm": 1.079251766204834,
      "learning_rate": 5.4555e-05,
      "loss": 0.3759,
      "step": 10917
    },
    {
      "epoch": 0.6113786538246164,
      "grad_norm": 1.436574101448059,
      "learning_rate": 5.456e-05,
      "loss": 0.3481,
      "step": 10918
    },
    {
      "epoch": 0.6114346511367454,
      "grad_norm": 1.346972942352295,
      "learning_rate": 5.4565e-05,
      "loss": 0.4239,
      "step": 10919
    },
    {
      "epoch": 0.6114906484488745,
      "grad_norm": 1.3086992502212524,
      "learning_rate": 5.457e-05,
      "loss": 0.4323,
      "step": 10920
    },
    {
      "epoch": 0.6115466457610035,
      "grad_norm": 1.2626492977142334,
      "learning_rate": 5.4575e-05,
      "loss": 0.4108,
      "step": 10921
    },
    {
      "epoch": 0.6116026430731325,
      "grad_norm": 1.3344029188156128,
      "learning_rate": 5.4579999999999996e-05,
      "loss": 0.4914,
      "step": 10922
    },
    {
      "epoch": 0.6116586403852615,
      "grad_norm": 1.2221949100494385,
      "learning_rate": 5.4585e-05,
      "loss": 0.3818,
      "step": 10923
    },
    {
      "epoch": 0.6117146376973905,
      "grad_norm": 1.3945127725601196,
      "learning_rate": 5.459000000000001e-05,
      "loss": 0.4025,
      "step": 10924
    },
    {
      "epoch": 0.6117706350095196,
      "grad_norm": 1.5424026250839233,
      "learning_rate": 5.459500000000001e-05,
      "loss": 0.5179,
      "step": 10925
    },
    {
      "epoch": 0.6118266323216486,
      "grad_norm": 1.215567708015442,
      "learning_rate": 5.4600000000000006e-05,
      "loss": 0.4808,
      "step": 10926
    },
    {
      "epoch": 0.6118826296337776,
      "grad_norm": 1.0344998836517334,
      "learning_rate": 5.4605e-05,
      "loss": 0.4033,
      "step": 10927
    },
    {
      "epoch": 0.6119386269459066,
      "grad_norm": 1.3918633460998535,
      "learning_rate": 5.461000000000001e-05,
      "loss": 0.4244,
      "step": 10928
    },
    {
      "epoch": 0.6119946242580356,
      "grad_norm": 1.1625525951385498,
      "learning_rate": 5.4615000000000005e-05,
      "loss": 0.4361,
      "step": 10929
    },
    {
      "epoch": 0.6120506215701647,
      "grad_norm": 1.3637739419937134,
      "learning_rate": 5.462e-05,
      "loss": 0.4096,
      "step": 10930
    },
    {
      "epoch": 0.6121066188822937,
      "grad_norm": 0.9587275385856628,
      "learning_rate": 5.4625000000000006e-05,
      "loss": 0.335,
      "step": 10931
    },
    {
      "epoch": 0.6121626161944227,
      "grad_norm": 1.3482089042663574,
      "learning_rate": 5.4630000000000004e-05,
      "loss": 0.4462,
      "step": 10932
    },
    {
      "epoch": 0.6122186135065517,
      "grad_norm": 1.388649344444275,
      "learning_rate": 5.4635e-05,
      "loss": 0.354,
      "step": 10933
    },
    {
      "epoch": 0.6122746108186807,
      "grad_norm": 1.1378109455108643,
      "learning_rate": 5.4640000000000005e-05,
      "loss": 0.316,
      "step": 10934
    },
    {
      "epoch": 0.6123306081308098,
      "grad_norm": 1.0495285987854004,
      "learning_rate": 5.4645e-05,
      "loss": 0.3512,
      "step": 10935
    },
    {
      "epoch": 0.6123866054429388,
      "grad_norm": 2.3947298526763916,
      "learning_rate": 5.465e-05,
      "loss": 0.4391,
      "step": 10936
    },
    {
      "epoch": 0.6124426027550678,
      "grad_norm": 1.2747169733047485,
      "learning_rate": 5.4655e-05,
      "loss": 0.4272,
      "step": 10937
    },
    {
      "epoch": 0.6124986000671968,
      "grad_norm": 1.2688775062561035,
      "learning_rate": 5.466e-05,
      "loss": 0.4662,
      "step": 10938
    },
    {
      "epoch": 0.6125545973793258,
      "grad_norm": 1.249367594718933,
      "learning_rate": 5.4665e-05,
      "loss": 0.3896,
      "step": 10939
    },
    {
      "epoch": 0.6126105946914548,
      "grad_norm": 1.2491068840026855,
      "learning_rate": 5.467e-05,
      "loss": 0.3809,
      "step": 10940
    },
    {
      "epoch": 0.6126665920035839,
      "grad_norm": 1.1681910753250122,
      "learning_rate": 5.4675e-05,
      "loss": 0.3824,
      "step": 10941
    },
    {
      "epoch": 0.6127225893157129,
      "grad_norm": 1.31464684009552,
      "learning_rate": 5.468e-05,
      "loss": 0.4716,
      "step": 10942
    },
    {
      "epoch": 0.6127785866278419,
      "grad_norm": 1.1723101139068604,
      "learning_rate": 5.4684999999999996e-05,
      "loss": 0.4744,
      "step": 10943
    },
    {
      "epoch": 0.6128345839399709,
      "grad_norm": 1.363568663597107,
      "learning_rate": 5.469000000000001e-05,
      "loss": 0.4266,
      "step": 10944
    },
    {
      "epoch": 0.6128905812520999,
      "grad_norm": 1.425294280052185,
      "learning_rate": 5.4695000000000004e-05,
      "loss": 0.6472,
      "step": 10945
    },
    {
      "epoch": 0.612946578564229,
      "grad_norm": 1.3231925964355469,
      "learning_rate": 5.470000000000001e-05,
      "loss": 0.372,
      "step": 10946
    },
    {
      "epoch": 0.613002575876358,
      "grad_norm": 1.4727872610092163,
      "learning_rate": 5.4705000000000006e-05,
      "loss": 0.3979,
      "step": 10947
    },
    {
      "epoch": 0.613058573188487,
      "grad_norm": 1.184914469718933,
      "learning_rate": 5.471e-05,
      "loss": 0.4542,
      "step": 10948
    },
    {
      "epoch": 0.613114570500616,
      "grad_norm": 1.3102973699569702,
      "learning_rate": 5.471500000000001e-05,
      "loss": 0.4403,
      "step": 10949
    },
    {
      "epoch": 0.613170567812745,
      "grad_norm": 1.3212430477142334,
      "learning_rate": 5.4720000000000005e-05,
      "loss": 0.4557,
      "step": 10950
    },
    {
      "epoch": 0.6132265651248741,
      "grad_norm": 1.1247193813323975,
      "learning_rate": 5.4725e-05,
      "loss": 0.3428,
      "step": 10951
    },
    {
      "epoch": 0.6132825624370031,
      "grad_norm": 2.253873825073242,
      "learning_rate": 5.4730000000000006e-05,
      "loss": 0.547,
      "step": 10952
    },
    {
      "epoch": 0.6133385597491321,
      "grad_norm": 1.2817833423614502,
      "learning_rate": 5.4735000000000004e-05,
      "loss": 0.4066,
      "step": 10953
    },
    {
      "epoch": 0.6133945570612611,
      "grad_norm": 1.4005568027496338,
      "learning_rate": 5.474e-05,
      "loss": 0.3571,
      "step": 10954
    },
    {
      "epoch": 0.61345055437339,
      "grad_norm": 1.2430723905563354,
      "learning_rate": 5.4745e-05,
      "loss": 0.4512,
      "step": 10955
    },
    {
      "epoch": 0.613506551685519,
      "grad_norm": 1.6329760551452637,
      "learning_rate": 5.475e-05,
      "loss": 0.7016,
      "step": 10956
    },
    {
      "epoch": 0.6135625489976481,
      "grad_norm": 1.369162917137146,
      "learning_rate": 5.4755e-05,
      "loss": 0.4148,
      "step": 10957
    },
    {
      "epoch": 0.6136185463097771,
      "grad_norm": 1.1662013530731201,
      "learning_rate": 5.476e-05,
      "loss": 0.4242,
      "step": 10958
    },
    {
      "epoch": 0.6136745436219061,
      "grad_norm": 1.0382499694824219,
      "learning_rate": 5.4765e-05,
      "loss": 0.3375,
      "step": 10959
    },
    {
      "epoch": 0.6137305409340351,
      "grad_norm": 1.6717263460159302,
      "learning_rate": 5.477e-05,
      "loss": 0.4685,
      "step": 10960
    },
    {
      "epoch": 0.6137865382461641,
      "grad_norm": 1.4468532800674438,
      "learning_rate": 5.4774999999999996e-05,
      "loss": 0.4467,
      "step": 10961
    },
    {
      "epoch": 0.6138425355582932,
      "grad_norm": 1.1400389671325684,
      "learning_rate": 5.478e-05,
      "loss": 0.4137,
      "step": 10962
    },
    {
      "epoch": 0.6138985328704222,
      "grad_norm": 1.5730295181274414,
      "learning_rate": 5.4785e-05,
      "loss": 0.3277,
      "step": 10963
    },
    {
      "epoch": 0.6139545301825512,
      "grad_norm": 1.0826725959777832,
      "learning_rate": 5.479000000000001e-05,
      "loss": 0.3517,
      "step": 10964
    },
    {
      "epoch": 0.6140105274946802,
      "grad_norm": 1.3723716735839844,
      "learning_rate": 5.4795000000000006e-05,
      "loss": 0.4235,
      "step": 10965
    },
    {
      "epoch": 0.6140665248068092,
      "grad_norm": 1.3577009439468384,
      "learning_rate": 5.4800000000000004e-05,
      "loss": 0.5371,
      "step": 10966
    },
    {
      "epoch": 0.6141225221189383,
      "grad_norm": 1.2552025318145752,
      "learning_rate": 5.480500000000001e-05,
      "loss": 0.4135,
      "step": 10967
    },
    {
      "epoch": 0.6141785194310673,
      "grad_norm": 1.306849479675293,
      "learning_rate": 5.4810000000000005e-05,
      "loss": 0.5673,
      "step": 10968
    },
    {
      "epoch": 0.6142345167431963,
      "grad_norm": 1.1524357795715332,
      "learning_rate": 5.4815e-05,
      "loss": 0.359,
      "step": 10969
    },
    {
      "epoch": 0.6142905140553253,
      "grad_norm": 1.3059027194976807,
      "learning_rate": 5.482000000000001e-05,
      "loss": 0.4656,
      "step": 10970
    },
    {
      "epoch": 0.6143465113674543,
      "grad_norm": 1.4936234951019287,
      "learning_rate": 5.4825000000000004e-05,
      "loss": 0.4255,
      "step": 10971
    },
    {
      "epoch": 0.6144025086795833,
      "grad_norm": 1.2601217031478882,
      "learning_rate": 5.483e-05,
      "loss": 0.6143,
      "step": 10972
    },
    {
      "epoch": 0.6144585059917124,
      "grad_norm": 1.4112039804458618,
      "learning_rate": 5.4835000000000006e-05,
      "loss": 0.5809,
      "step": 10973
    },
    {
      "epoch": 0.6145145033038414,
      "grad_norm": 1.4576923847198486,
      "learning_rate": 5.4840000000000003e-05,
      "loss": 0.3353,
      "step": 10974
    },
    {
      "epoch": 0.6145705006159704,
      "grad_norm": 1.2597851753234863,
      "learning_rate": 5.4845e-05,
      "loss": 0.5477,
      "step": 10975
    },
    {
      "epoch": 0.6146264979280994,
      "grad_norm": 1.2168810367584229,
      "learning_rate": 5.485e-05,
      "loss": 0.4293,
      "step": 10976
    },
    {
      "epoch": 0.6146824952402284,
      "grad_norm": 1.051815152168274,
      "learning_rate": 5.4855e-05,
      "loss": 0.4024,
      "step": 10977
    },
    {
      "epoch": 0.6147384925523575,
      "grad_norm": 1.2898094654083252,
      "learning_rate": 5.486e-05,
      "loss": 0.4484,
      "step": 10978
    },
    {
      "epoch": 0.6147944898644865,
      "grad_norm": 1.4077908992767334,
      "learning_rate": 5.4865e-05,
      "loss": 0.4834,
      "step": 10979
    },
    {
      "epoch": 0.6148504871766155,
      "grad_norm": 1.5623785257339478,
      "learning_rate": 5.487e-05,
      "loss": 0.625,
      "step": 10980
    },
    {
      "epoch": 0.6149064844887445,
      "grad_norm": 1.3912885189056396,
      "learning_rate": 5.4875e-05,
      "loss": 0.5837,
      "step": 10981
    },
    {
      "epoch": 0.6149624818008735,
      "grad_norm": 1.4288856983184814,
      "learning_rate": 5.4879999999999996e-05,
      "loss": 0.4109,
      "step": 10982
    },
    {
      "epoch": 0.6150184791130026,
      "grad_norm": 1.4826418161392212,
      "learning_rate": 5.4885e-05,
      "loss": 0.5164,
      "step": 10983
    },
    {
      "epoch": 0.6150744764251316,
      "grad_norm": 1.431363821029663,
      "learning_rate": 5.4890000000000005e-05,
      "loss": 0.4435,
      "step": 10984
    },
    {
      "epoch": 0.6151304737372606,
      "grad_norm": 1.2103575468063354,
      "learning_rate": 5.489500000000001e-05,
      "loss": 0.3564,
      "step": 10985
    },
    {
      "epoch": 0.6151864710493896,
      "grad_norm": 1.0259290933609009,
      "learning_rate": 5.4900000000000006e-05,
      "loss": 0.3939,
      "step": 10986
    },
    {
      "epoch": 0.6152424683615186,
      "grad_norm": 1.5478663444519043,
      "learning_rate": 5.4905000000000004e-05,
      "loss": 0.4409,
      "step": 10987
    },
    {
      "epoch": 0.6152984656736477,
      "grad_norm": 1.2921854257583618,
      "learning_rate": 5.491000000000001e-05,
      "loss": 0.4146,
      "step": 10988
    },
    {
      "epoch": 0.6153544629857767,
      "grad_norm": 1.2508203983306885,
      "learning_rate": 5.4915000000000005e-05,
      "loss": 0.4168,
      "step": 10989
    },
    {
      "epoch": 0.6154104602979057,
      "grad_norm": 1.3484240770339966,
      "learning_rate": 5.492e-05,
      "loss": 0.4035,
      "step": 10990
    },
    {
      "epoch": 0.6154664576100347,
      "grad_norm": 1.207398533821106,
      "learning_rate": 5.492500000000001e-05,
      "loss": 0.3209,
      "step": 10991
    },
    {
      "epoch": 0.6155224549221637,
      "grad_norm": 2.1470868587493896,
      "learning_rate": 5.4930000000000004e-05,
      "loss": 0.3565,
      "step": 10992
    },
    {
      "epoch": 0.6155784522342928,
      "grad_norm": 1.2360011339187622,
      "learning_rate": 5.4935e-05,
      "loss": 0.3835,
      "step": 10993
    },
    {
      "epoch": 0.6156344495464218,
      "grad_norm": 1.2802917957305908,
      "learning_rate": 5.4940000000000006e-05,
      "loss": 0.3776,
      "step": 10994
    },
    {
      "epoch": 0.6156904468585508,
      "grad_norm": 1.2238057851791382,
      "learning_rate": 5.4945e-05,
      "loss": 0.4526,
      "step": 10995
    },
    {
      "epoch": 0.6157464441706798,
      "grad_norm": 1.3067924976348877,
      "learning_rate": 5.495e-05,
      "loss": 0.4185,
      "step": 10996
    },
    {
      "epoch": 0.6158024414828088,
      "grad_norm": 1.451428771018982,
      "learning_rate": 5.4955e-05,
      "loss": 0.6839,
      "step": 10997
    },
    {
      "epoch": 0.6158584387949378,
      "grad_norm": 3.082087516784668,
      "learning_rate": 5.496e-05,
      "loss": 0.4933,
      "step": 10998
    },
    {
      "epoch": 0.6159144361070669,
      "grad_norm": 1.1497201919555664,
      "learning_rate": 5.4965e-05,
      "loss": 0.5335,
      "step": 10999
    },
    {
      "epoch": 0.6159704334191959,
      "grad_norm": 1.3007779121398926,
      "learning_rate": 5.497e-05,
      "loss": 0.4934,
      "step": 11000
    },
    {
      "epoch": 0.6160264307313249,
      "grad_norm": 1.517802357673645,
      "learning_rate": 5.4975e-05,
      "loss": 0.622,
      "step": 11001
    },
    {
      "epoch": 0.6160824280434539,
      "grad_norm": 1.2755626440048218,
      "learning_rate": 5.498e-05,
      "loss": 0.3995,
      "step": 11002
    },
    {
      "epoch": 0.6161384253555829,
      "grad_norm": 1.2727468013763428,
      "learning_rate": 5.4984999999999996e-05,
      "loss": 0.3699,
      "step": 11003
    },
    {
      "epoch": 0.616194422667712,
      "grad_norm": 1.5111430883407593,
      "learning_rate": 5.499000000000001e-05,
      "loss": 0.4512,
      "step": 11004
    },
    {
      "epoch": 0.616250419979841,
      "grad_norm": 1.1231077909469604,
      "learning_rate": 5.4995000000000005e-05,
      "loss": 0.3624,
      "step": 11005
    },
    {
      "epoch": 0.61630641729197,
      "grad_norm": 1.5678112506866455,
      "learning_rate": 5.500000000000001e-05,
      "loss": 0.4726,
      "step": 11006
    },
    {
      "epoch": 0.616362414604099,
      "grad_norm": 1.218222737312317,
      "learning_rate": 5.5005000000000006e-05,
      "loss": 0.4233,
      "step": 11007
    },
    {
      "epoch": 0.616418411916228,
      "grad_norm": 1.5231025218963623,
      "learning_rate": 5.5010000000000004e-05,
      "loss": 0.5874,
      "step": 11008
    },
    {
      "epoch": 0.6164744092283571,
      "grad_norm": 1.3632187843322754,
      "learning_rate": 5.501500000000001e-05,
      "loss": 0.5486,
      "step": 11009
    },
    {
      "epoch": 0.6165304065404861,
      "grad_norm": 1.3238469362258911,
      "learning_rate": 5.5020000000000005e-05,
      "loss": 0.5328,
      "step": 11010
    },
    {
      "epoch": 0.6165864038526151,
      "grad_norm": 1.674149751663208,
      "learning_rate": 5.5025e-05,
      "loss": 0.4168,
      "step": 11011
    },
    {
      "epoch": 0.6166424011647441,
      "grad_norm": 1.2645148038864136,
      "learning_rate": 5.503000000000001e-05,
      "loss": 0.5078,
      "step": 11012
    },
    {
      "epoch": 0.6166983984768731,
      "grad_norm": 1.3339440822601318,
      "learning_rate": 5.5035000000000004e-05,
      "loss": 0.4358,
      "step": 11013
    },
    {
      "epoch": 0.6167543957890022,
      "grad_norm": 1.254732608795166,
      "learning_rate": 5.504e-05,
      "loss": 0.5489,
      "step": 11014
    },
    {
      "epoch": 0.6168103931011312,
      "grad_norm": 1.3222465515136719,
      "learning_rate": 5.5045e-05,
      "loss": 0.6669,
      "step": 11015
    },
    {
      "epoch": 0.6168663904132602,
      "grad_norm": 1.307663083076477,
      "learning_rate": 5.505e-05,
      "loss": 0.3796,
      "step": 11016
    },
    {
      "epoch": 0.6169223877253892,
      "grad_norm": 1.1509522199630737,
      "learning_rate": 5.5055e-05,
      "loss": 0.3864,
      "step": 11017
    },
    {
      "epoch": 0.6169783850375182,
      "grad_norm": 1.4544485807418823,
      "learning_rate": 5.506e-05,
      "loss": 0.4906,
      "step": 11018
    },
    {
      "epoch": 0.6170343823496472,
      "grad_norm": 1.219419002532959,
      "learning_rate": 5.5065e-05,
      "loss": 0.493,
      "step": 11019
    },
    {
      "epoch": 0.6170903796617763,
      "grad_norm": 1.4182944297790527,
      "learning_rate": 5.507e-05,
      "loss": 0.4782,
      "step": 11020
    },
    {
      "epoch": 0.6171463769739053,
      "grad_norm": 1.3153084516525269,
      "learning_rate": 5.5075e-05,
      "loss": 0.368,
      "step": 11021
    },
    {
      "epoch": 0.6172023742860343,
      "grad_norm": 1.4359521865844727,
      "learning_rate": 5.508e-05,
      "loss": 0.6702,
      "step": 11022
    },
    {
      "epoch": 0.6172583715981633,
      "grad_norm": 1.212326169013977,
      "learning_rate": 5.5085e-05,
      "loss": 0.4304,
      "step": 11023
    },
    {
      "epoch": 0.6173143689102923,
      "grad_norm": 1.1759984493255615,
      "learning_rate": 5.5089999999999996e-05,
      "loss": 0.4048,
      "step": 11024
    },
    {
      "epoch": 0.6173703662224214,
      "grad_norm": 1.2775005102157593,
      "learning_rate": 5.509500000000001e-05,
      "loss": 0.3695,
      "step": 11025
    },
    {
      "epoch": 0.6174263635345504,
      "grad_norm": 1.2537641525268555,
      "learning_rate": 5.5100000000000004e-05,
      "loss": 0.3624,
      "step": 11026
    },
    {
      "epoch": 0.6174823608466794,
      "grad_norm": 1.345799446105957,
      "learning_rate": 5.510500000000001e-05,
      "loss": 0.6128,
      "step": 11027
    },
    {
      "epoch": 0.6175383581588084,
      "grad_norm": 1.2790555953979492,
      "learning_rate": 5.5110000000000006e-05,
      "loss": 0.4091,
      "step": 11028
    },
    {
      "epoch": 0.6175943554709374,
      "grad_norm": 1.3208950757980347,
      "learning_rate": 5.5115000000000003e-05,
      "loss": 0.4776,
      "step": 11029
    },
    {
      "epoch": 0.6176503527830665,
      "grad_norm": 1.0869702100753784,
      "learning_rate": 5.512000000000001e-05,
      "loss": 0.3377,
      "step": 11030
    },
    {
      "epoch": 0.6177063500951955,
      "grad_norm": 1.5070682764053345,
      "learning_rate": 5.5125000000000005e-05,
      "loss": 0.5009,
      "step": 11031
    },
    {
      "epoch": 0.6177623474073245,
      "grad_norm": 1.4577410221099854,
      "learning_rate": 5.513e-05,
      "loss": 0.5479,
      "step": 11032
    },
    {
      "epoch": 0.6178183447194535,
      "grad_norm": 1.3744590282440186,
      "learning_rate": 5.5135000000000007e-05,
      "loss": 0.3575,
      "step": 11033
    },
    {
      "epoch": 0.6178743420315825,
      "grad_norm": 1.0250931978225708,
      "learning_rate": 5.5140000000000004e-05,
      "loss": 0.3366,
      "step": 11034
    },
    {
      "epoch": 0.6179303393437116,
      "grad_norm": 1.3300193548202515,
      "learning_rate": 5.5145e-05,
      "loss": 0.4172,
      "step": 11035
    },
    {
      "epoch": 0.6179863366558406,
      "grad_norm": 1.2005608081817627,
      "learning_rate": 5.515e-05,
      "loss": 0.3565,
      "step": 11036
    },
    {
      "epoch": 0.6180423339679696,
      "grad_norm": 1.2163310050964355,
      "learning_rate": 5.5155e-05,
      "loss": 0.3138,
      "step": 11037
    },
    {
      "epoch": 0.6180983312800985,
      "grad_norm": 1.210322618484497,
      "learning_rate": 5.516e-05,
      "loss": 0.4677,
      "step": 11038
    },
    {
      "epoch": 0.6181543285922275,
      "grad_norm": 1.21651029586792,
      "learning_rate": 5.5165e-05,
      "loss": 0.5215,
      "step": 11039
    },
    {
      "epoch": 0.6182103259043565,
      "grad_norm": 1.3309255838394165,
      "learning_rate": 5.517e-05,
      "loss": 0.4064,
      "step": 11040
    },
    {
      "epoch": 0.6182663232164856,
      "grad_norm": 1.3202449083328247,
      "learning_rate": 5.5175e-05,
      "loss": 0.433,
      "step": 11041
    },
    {
      "epoch": 0.6183223205286146,
      "grad_norm": 1.4714305400848389,
      "learning_rate": 5.518e-05,
      "loss": 0.4863,
      "step": 11042
    },
    {
      "epoch": 0.6183783178407436,
      "grad_norm": 1.1451250314712524,
      "learning_rate": 5.5185e-05,
      "loss": 0.4105,
      "step": 11043
    },
    {
      "epoch": 0.6184343151528726,
      "grad_norm": 1.2788653373718262,
      "learning_rate": 5.519e-05,
      "loss": 0.4258,
      "step": 11044
    },
    {
      "epoch": 0.6184903124650016,
      "grad_norm": 1.2494900226593018,
      "learning_rate": 5.519500000000001e-05,
      "loss": 0.3796,
      "step": 11045
    },
    {
      "epoch": 0.6185463097771307,
      "grad_norm": 1.2078883647918701,
      "learning_rate": 5.520000000000001e-05,
      "loss": 0.4384,
      "step": 11046
    },
    {
      "epoch": 0.6186023070892597,
      "grad_norm": 1.1643199920654297,
      "learning_rate": 5.5205000000000004e-05,
      "loss": 0.3847,
      "step": 11047
    },
    {
      "epoch": 0.6186583044013887,
      "grad_norm": 2.068957805633545,
      "learning_rate": 5.521000000000001e-05,
      "loss": 0.4046,
      "step": 11048
    },
    {
      "epoch": 0.6187143017135177,
      "grad_norm": 1.1836789846420288,
      "learning_rate": 5.5215000000000006e-05,
      "loss": 0.3422,
      "step": 11049
    },
    {
      "epoch": 0.6187702990256467,
      "grad_norm": 1.6435412168502808,
      "learning_rate": 5.522e-05,
      "loss": 0.4077,
      "step": 11050
    },
    {
      "epoch": 0.6188262963377757,
      "grad_norm": 1.2401477098464966,
      "learning_rate": 5.522500000000001e-05,
      "loss": 0.4519,
      "step": 11051
    },
    {
      "epoch": 0.6188822936499048,
      "grad_norm": 1.4588103294372559,
      "learning_rate": 5.5230000000000005e-05,
      "loss": 0.5337,
      "step": 11052
    },
    {
      "epoch": 0.6189382909620338,
      "grad_norm": 1.5246227979660034,
      "learning_rate": 5.5235e-05,
      "loss": 0.5767,
      "step": 11053
    },
    {
      "epoch": 0.6189942882741628,
      "grad_norm": 3.2909200191497803,
      "learning_rate": 5.524e-05,
      "loss": 0.3592,
      "step": 11054
    },
    {
      "epoch": 0.6190502855862918,
      "grad_norm": 1.2052184343338013,
      "learning_rate": 5.5245000000000004e-05,
      "loss": 0.4188,
      "step": 11055
    },
    {
      "epoch": 0.6191062828984208,
      "grad_norm": 1.6031838655471802,
      "learning_rate": 5.525e-05,
      "loss": 0.4697,
      "step": 11056
    },
    {
      "epoch": 0.6191622802105499,
      "grad_norm": 1.0913581848144531,
      "learning_rate": 5.5255e-05,
      "loss": 0.3663,
      "step": 11057
    },
    {
      "epoch": 0.6192182775226789,
      "grad_norm": 1.2913140058517456,
      "learning_rate": 5.526e-05,
      "loss": 0.4313,
      "step": 11058
    },
    {
      "epoch": 0.6192742748348079,
      "grad_norm": 1.3415793180465698,
      "learning_rate": 5.5265e-05,
      "loss": 0.4306,
      "step": 11059
    },
    {
      "epoch": 0.6193302721469369,
      "grad_norm": 1.519203782081604,
      "learning_rate": 5.527e-05,
      "loss": 0.4767,
      "step": 11060
    },
    {
      "epoch": 0.6193862694590659,
      "grad_norm": 1.2701588869094849,
      "learning_rate": 5.5275e-05,
      "loss": 0.5083,
      "step": 11061
    },
    {
      "epoch": 0.619442266771195,
      "grad_norm": 1.0991445779800415,
      "learning_rate": 5.528e-05,
      "loss": 0.347,
      "step": 11062
    },
    {
      "epoch": 0.619498264083324,
      "grad_norm": 1.1720919609069824,
      "learning_rate": 5.5285e-05,
      "loss": 0.3684,
      "step": 11063
    },
    {
      "epoch": 0.619554261395453,
      "grad_norm": 1.3075461387634277,
      "learning_rate": 5.5289999999999994e-05,
      "loss": 0.4524,
      "step": 11064
    },
    {
      "epoch": 0.619610258707582,
      "grad_norm": 1.2739973068237305,
      "learning_rate": 5.5295000000000005e-05,
      "loss": 0.5399,
      "step": 11065
    },
    {
      "epoch": 0.619666256019711,
      "grad_norm": 1.3019740581512451,
      "learning_rate": 5.530000000000001e-05,
      "loss": 0.4455,
      "step": 11066
    },
    {
      "epoch": 0.6197222533318401,
      "grad_norm": 1.2433137893676758,
      "learning_rate": 5.530500000000001e-05,
      "loss": 0.3155,
      "step": 11067
    },
    {
      "epoch": 0.6197782506439691,
      "grad_norm": 1.2786258459091187,
      "learning_rate": 5.5310000000000004e-05,
      "loss": 0.5279,
      "step": 11068
    },
    {
      "epoch": 0.6198342479560981,
      "grad_norm": 1.3097648620605469,
      "learning_rate": 5.531500000000001e-05,
      "loss": 0.5644,
      "step": 11069
    },
    {
      "epoch": 0.6198902452682271,
      "grad_norm": 1.406480312347412,
      "learning_rate": 5.5320000000000006e-05,
      "loss": 0.5045,
      "step": 11070
    },
    {
      "epoch": 0.6199462425803561,
      "grad_norm": 1.22561776638031,
      "learning_rate": 5.5325e-05,
      "loss": 0.3946,
      "step": 11071
    },
    {
      "epoch": 0.6200022398924852,
      "grad_norm": 1.4362882375717163,
      "learning_rate": 5.533000000000001e-05,
      "loss": 0.4382,
      "step": 11072
    },
    {
      "epoch": 0.6200582372046142,
      "grad_norm": 1.520639181137085,
      "learning_rate": 5.5335000000000005e-05,
      "loss": 0.4684,
      "step": 11073
    },
    {
      "epoch": 0.6201142345167432,
      "grad_norm": 1.1990320682525635,
      "learning_rate": 5.534e-05,
      "loss": 0.4649,
      "step": 11074
    },
    {
      "epoch": 0.6201702318288722,
      "grad_norm": 1.2321856021881104,
      "learning_rate": 5.5345e-05,
      "loss": 0.3917,
      "step": 11075
    },
    {
      "epoch": 0.6202262291410012,
      "grad_norm": 1.2517001628875732,
      "learning_rate": 5.5350000000000004e-05,
      "loss": 0.3432,
      "step": 11076
    },
    {
      "epoch": 0.6202822264531302,
      "grad_norm": 1.116336703300476,
      "learning_rate": 5.5355e-05,
      "loss": 0.414,
      "step": 11077
    },
    {
      "epoch": 0.6203382237652593,
      "grad_norm": 1.4558522701263428,
      "learning_rate": 5.536e-05,
      "loss": 0.3346,
      "step": 11078
    },
    {
      "epoch": 0.6203942210773883,
      "grad_norm": 1.334283709526062,
      "learning_rate": 5.5365e-05,
      "loss": 0.4531,
      "step": 11079
    },
    {
      "epoch": 0.6204502183895173,
      "grad_norm": 1.297268033027649,
      "learning_rate": 5.537e-05,
      "loss": 0.5036,
      "step": 11080
    },
    {
      "epoch": 0.6205062157016463,
      "grad_norm": 1.2266566753387451,
      "learning_rate": 5.5375e-05,
      "loss": 0.4021,
      "step": 11081
    },
    {
      "epoch": 0.6205622130137753,
      "grad_norm": 1.2731362581253052,
      "learning_rate": 5.538e-05,
      "loss": 0.4218,
      "step": 11082
    },
    {
      "epoch": 0.6206182103259044,
      "grad_norm": 1.3133450746536255,
      "learning_rate": 5.5385e-05,
      "loss": 0.4175,
      "step": 11083
    },
    {
      "epoch": 0.6206742076380334,
      "grad_norm": 1.390400767326355,
      "learning_rate": 5.5389999999999997e-05,
      "loss": 0.3764,
      "step": 11084
    },
    {
      "epoch": 0.6207302049501624,
      "grad_norm": 1.1527926921844482,
      "learning_rate": 5.539500000000001e-05,
      "loss": 0.3754,
      "step": 11085
    },
    {
      "epoch": 0.6207862022622914,
      "grad_norm": 1.342327356338501,
      "learning_rate": 5.5400000000000005e-05,
      "loss": 0.5999,
      "step": 11086
    },
    {
      "epoch": 0.6208421995744204,
      "grad_norm": 1.1754281520843506,
      "learning_rate": 5.540500000000001e-05,
      "loss": 0.4429,
      "step": 11087
    },
    {
      "epoch": 0.6208981968865495,
      "grad_norm": 1.2577422857284546,
      "learning_rate": 5.5410000000000007e-05,
      "loss": 0.4208,
      "step": 11088
    },
    {
      "epoch": 0.6209541941986785,
      "grad_norm": 1.3542053699493408,
      "learning_rate": 5.5415000000000004e-05,
      "loss": 0.4091,
      "step": 11089
    },
    {
      "epoch": 0.6210101915108075,
      "grad_norm": 1.1279963254928589,
      "learning_rate": 5.542000000000001e-05,
      "loss": 0.3579,
      "step": 11090
    },
    {
      "epoch": 0.6210661888229365,
      "grad_norm": 0.9857292175292969,
      "learning_rate": 5.5425000000000006e-05,
      "loss": 0.2692,
      "step": 11091
    },
    {
      "epoch": 0.6211221861350655,
      "grad_norm": 1.6026922464370728,
      "learning_rate": 5.543e-05,
      "loss": 0.5099,
      "step": 11092
    },
    {
      "epoch": 0.6211781834471946,
      "grad_norm": 1.8746087551116943,
      "learning_rate": 5.5435e-05,
      "loss": 0.5977,
      "step": 11093
    },
    {
      "epoch": 0.6212341807593236,
      "grad_norm": 1.1242713928222656,
      "learning_rate": 5.5440000000000005e-05,
      "loss": 0.3337,
      "step": 11094
    },
    {
      "epoch": 0.6212901780714526,
      "grad_norm": 1.4334536790847778,
      "learning_rate": 5.5445e-05,
      "loss": 0.491,
      "step": 11095
    },
    {
      "epoch": 0.6213461753835816,
      "grad_norm": 1.4200242757797241,
      "learning_rate": 5.545e-05,
      "loss": 0.4305,
      "step": 11096
    },
    {
      "epoch": 0.6214021726957106,
      "grad_norm": 1.2854182720184326,
      "learning_rate": 5.5455000000000004e-05,
      "loss": 0.5226,
      "step": 11097
    },
    {
      "epoch": 0.6214581700078396,
      "grad_norm": 1.1912659406661987,
      "learning_rate": 5.546e-05,
      "loss": 0.4192,
      "step": 11098
    },
    {
      "epoch": 0.6215141673199687,
      "grad_norm": 1.1298226118087769,
      "learning_rate": 5.5465e-05,
      "loss": 0.3647,
      "step": 11099
    },
    {
      "epoch": 0.6215701646320977,
      "grad_norm": 1.2152447700500488,
      "learning_rate": 5.547e-05,
      "loss": 0.311,
      "step": 11100
    },
    {
      "epoch": 0.6216261619442267,
      "grad_norm": 1.1264703273773193,
      "learning_rate": 5.5475e-05,
      "loss": 0.3419,
      "step": 11101
    },
    {
      "epoch": 0.6216821592563557,
      "grad_norm": 1.2177457809448242,
      "learning_rate": 5.548e-05,
      "loss": 0.3906,
      "step": 11102
    },
    {
      "epoch": 0.6217381565684847,
      "grad_norm": 1.399300217628479,
      "learning_rate": 5.5484999999999995e-05,
      "loss": 0.4839,
      "step": 11103
    },
    {
      "epoch": 0.6217941538806138,
      "grad_norm": 1.3462135791778564,
      "learning_rate": 5.549e-05,
      "loss": 0.4196,
      "step": 11104
    },
    {
      "epoch": 0.6218501511927428,
      "grad_norm": 1.5765787363052368,
      "learning_rate": 5.549500000000001e-05,
      "loss": 0.5721,
      "step": 11105
    },
    {
      "epoch": 0.6219061485048718,
      "grad_norm": 1.2746860980987549,
      "learning_rate": 5.550000000000001e-05,
      "loss": 0.4462,
      "step": 11106
    },
    {
      "epoch": 0.6219621458170008,
      "grad_norm": 1.1315209865570068,
      "learning_rate": 5.5505000000000005e-05,
      "loss": 0.396,
      "step": 11107
    },
    {
      "epoch": 0.6220181431291298,
      "grad_norm": 1.1072025299072266,
      "learning_rate": 5.551000000000001e-05,
      "loss": 0.2609,
      "step": 11108
    },
    {
      "epoch": 0.6220741404412589,
      "grad_norm": 1.4301235675811768,
      "learning_rate": 5.5515000000000006e-05,
      "loss": 0.4324,
      "step": 11109
    },
    {
      "epoch": 0.6221301377533879,
      "grad_norm": 1.4740022420883179,
      "learning_rate": 5.5520000000000004e-05,
      "loss": 0.5201,
      "step": 11110
    },
    {
      "epoch": 0.6221861350655169,
      "grad_norm": 1.1610413789749146,
      "learning_rate": 5.552500000000001e-05,
      "loss": 0.3384,
      "step": 11111
    },
    {
      "epoch": 0.6222421323776459,
      "grad_norm": 1.0692731142044067,
      "learning_rate": 5.5530000000000005e-05,
      "loss": 0.36,
      "step": 11112
    },
    {
      "epoch": 0.6222981296897749,
      "grad_norm": 1.2527985572814941,
      "learning_rate": 5.5535e-05,
      "loss": 0.5219,
      "step": 11113
    },
    {
      "epoch": 0.622354127001904,
      "grad_norm": 1.2130718231201172,
      "learning_rate": 5.554e-05,
      "loss": 0.418,
      "step": 11114
    },
    {
      "epoch": 0.622410124314033,
      "grad_norm": 1.4109265804290771,
      "learning_rate": 5.5545000000000004e-05,
      "loss": 0.4951,
      "step": 11115
    },
    {
      "epoch": 0.622466121626162,
      "grad_norm": 1.3276960849761963,
      "learning_rate": 5.555e-05,
      "loss": 0.4697,
      "step": 11116
    },
    {
      "epoch": 0.622522118938291,
      "grad_norm": 1.4295916557312012,
      "learning_rate": 5.5555e-05,
      "loss": 0.4085,
      "step": 11117
    },
    {
      "epoch": 0.62257811625042,
      "grad_norm": 1.5808247327804565,
      "learning_rate": 5.556e-05,
      "loss": 0.4237,
      "step": 11118
    },
    {
      "epoch": 0.622634113562549,
      "grad_norm": 1.3355120420455933,
      "learning_rate": 5.5565e-05,
      "loss": 0.5678,
      "step": 11119
    },
    {
      "epoch": 0.622690110874678,
      "grad_norm": 1.452845811843872,
      "learning_rate": 5.557e-05,
      "loss": 0.5304,
      "step": 11120
    },
    {
      "epoch": 0.622746108186807,
      "grad_norm": 1.1395128965377808,
      "learning_rate": 5.5575e-05,
      "loss": 0.3455,
      "step": 11121
    },
    {
      "epoch": 0.622802105498936,
      "grad_norm": 1.3583852052688599,
      "learning_rate": 5.558e-05,
      "loss": 0.3351,
      "step": 11122
    },
    {
      "epoch": 0.622858102811065,
      "grad_norm": 1.1834583282470703,
      "learning_rate": 5.5585e-05,
      "loss": 0.4016,
      "step": 11123
    },
    {
      "epoch": 0.622914100123194,
      "grad_norm": 1.3968842029571533,
      "learning_rate": 5.5589999999999995e-05,
      "loss": 0.3693,
      "step": 11124
    },
    {
      "epoch": 0.6229700974353231,
      "grad_norm": 1.4644670486450195,
      "learning_rate": 5.5595000000000006e-05,
      "loss": 0.5647,
      "step": 11125
    },
    {
      "epoch": 0.6230260947474521,
      "grad_norm": 1.2715789079666138,
      "learning_rate": 5.560000000000001e-05,
      "loss": 0.3577,
      "step": 11126
    },
    {
      "epoch": 0.6230820920595811,
      "grad_norm": 3.06221342086792,
      "learning_rate": 5.560500000000001e-05,
      "loss": 0.5486,
      "step": 11127
    },
    {
      "epoch": 0.6231380893717101,
      "grad_norm": 1.3651798963546753,
      "learning_rate": 5.5610000000000005e-05,
      "loss": 0.4616,
      "step": 11128
    },
    {
      "epoch": 0.6231940866838391,
      "grad_norm": 1.209020972251892,
      "learning_rate": 5.561500000000001e-05,
      "loss": 0.3837,
      "step": 11129
    },
    {
      "epoch": 0.6232500839959682,
      "grad_norm": 1.3096704483032227,
      "learning_rate": 5.5620000000000006e-05,
      "loss": 0.4342,
      "step": 11130
    },
    {
      "epoch": 0.6233060813080972,
      "grad_norm": 1.4271564483642578,
      "learning_rate": 5.5625000000000004e-05,
      "loss": 0.466,
      "step": 11131
    },
    {
      "epoch": 0.6233620786202262,
      "grad_norm": 1.1934137344360352,
      "learning_rate": 5.563e-05,
      "loss": 0.4586,
      "step": 11132
    },
    {
      "epoch": 0.6234180759323552,
      "grad_norm": 1.1513158082962036,
      "learning_rate": 5.5635000000000005e-05,
      "loss": 0.341,
      "step": 11133
    },
    {
      "epoch": 0.6234740732444842,
      "grad_norm": 1.2885565757751465,
      "learning_rate": 5.564e-05,
      "loss": 0.4821,
      "step": 11134
    },
    {
      "epoch": 0.6235300705566132,
      "grad_norm": 1.2884690761566162,
      "learning_rate": 5.5645e-05,
      "loss": 0.4927,
      "step": 11135
    },
    {
      "epoch": 0.6235860678687423,
      "grad_norm": 1.2023165225982666,
      "learning_rate": 5.5650000000000004e-05,
      "loss": 0.3787,
      "step": 11136
    },
    {
      "epoch": 0.6236420651808713,
      "grad_norm": 1.1008411645889282,
      "learning_rate": 5.5655e-05,
      "loss": 0.4443,
      "step": 11137
    },
    {
      "epoch": 0.6236980624930003,
      "grad_norm": 1.222356915473938,
      "learning_rate": 5.566e-05,
      "loss": 0.4743,
      "step": 11138
    },
    {
      "epoch": 0.6237540598051293,
      "grad_norm": 1.2594475746154785,
      "learning_rate": 5.5665e-05,
      "loss": 0.5381,
      "step": 11139
    },
    {
      "epoch": 0.6238100571172583,
      "grad_norm": 1.3761041164398193,
      "learning_rate": 5.567e-05,
      "loss": 0.542,
      "step": 11140
    },
    {
      "epoch": 0.6238660544293874,
      "grad_norm": 1.3204220533370972,
      "learning_rate": 5.5675e-05,
      "loss": 0.47,
      "step": 11141
    },
    {
      "epoch": 0.6239220517415164,
      "grad_norm": 1.188469648361206,
      "learning_rate": 5.5679999999999995e-05,
      "loss": 0.3275,
      "step": 11142
    },
    {
      "epoch": 0.6239780490536454,
      "grad_norm": 1.4052780866622925,
      "learning_rate": 5.5685e-05,
      "loss": 0.4375,
      "step": 11143
    },
    {
      "epoch": 0.6240340463657744,
      "grad_norm": 1.1389778852462769,
      "learning_rate": 5.569e-05,
      "loss": 0.336,
      "step": 11144
    },
    {
      "epoch": 0.6240900436779034,
      "grad_norm": 1.3461494445800781,
      "learning_rate": 5.5694999999999994e-05,
      "loss": 0.4599,
      "step": 11145
    },
    {
      "epoch": 0.6241460409900325,
      "grad_norm": 1.2822725772857666,
      "learning_rate": 5.5700000000000005e-05,
      "loss": 0.4538,
      "step": 11146
    },
    {
      "epoch": 0.6242020383021615,
      "grad_norm": 1.4448376893997192,
      "learning_rate": 5.570500000000001e-05,
      "loss": 0.4361,
      "step": 11147
    },
    {
      "epoch": 0.6242580356142905,
      "grad_norm": 1.0926368236541748,
      "learning_rate": 5.571000000000001e-05,
      "loss": 0.3762,
      "step": 11148
    },
    {
      "epoch": 0.6243140329264195,
      "grad_norm": 1.2508419752120972,
      "learning_rate": 5.5715000000000004e-05,
      "loss": 0.2762,
      "step": 11149
    },
    {
      "epoch": 0.6243700302385485,
      "grad_norm": 1.2052196264266968,
      "learning_rate": 5.572000000000001e-05,
      "loss": 0.4175,
      "step": 11150
    },
    {
      "epoch": 0.6244260275506776,
      "grad_norm": 1.2460700273513794,
      "learning_rate": 5.5725000000000006e-05,
      "loss": 0.4135,
      "step": 11151
    },
    {
      "epoch": 0.6244820248628066,
      "grad_norm": 1.1380853652954102,
      "learning_rate": 5.5730000000000003e-05,
      "loss": 0.4291,
      "step": 11152
    },
    {
      "epoch": 0.6245380221749356,
      "grad_norm": 1.1354602575302124,
      "learning_rate": 5.5735e-05,
      "loss": 0.3984,
      "step": 11153
    },
    {
      "epoch": 0.6245940194870646,
      "grad_norm": 1.2212356328964233,
      "learning_rate": 5.5740000000000005e-05,
      "loss": 0.3591,
      "step": 11154
    },
    {
      "epoch": 0.6246500167991936,
      "grad_norm": 1.7149409055709839,
      "learning_rate": 5.5745e-05,
      "loss": 0.4434,
      "step": 11155
    },
    {
      "epoch": 0.6247060141113226,
      "grad_norm": 1.4858187437057495,
      "learning_rate": 5.575e-05,
      "loss": 0.5377,
      "step": 11156
    },
    {
      "epoch": 0.6247620114234517,
      "grad_norm": 1.2911696434020996,
      "learning_rate": 5.5755000000000004e-05,
      "loss": 0.4101,
      "step": 11157
    },
    {
      "epoch": 0.6248180087355807,
      "grad_norm": 1.328184962272644,
      "learning_rate": 5.576e-05,
      "loss": 0.3503,
      "step": 11158
    },
    {
      "epoch": 0.6248740060477097,
      "grad_norm": 1.7264142036437988,
      "learning_rate": 5.5765e-05,
      "loss": 0.4688,
      "step": 11159
    },
    {
      "epoch": 0.6249300033598387,
      "grad_norm": 1.4469679594039917,
      "learning_rate": 5.577e-05,
      "loss": 0.528,
      "step": 11160
    },
    {
      "epoch": 0.6249860006719677,
      "grad_norm": 1.1961560249328613,
      "learning_rate": 5.5775e-05,
      "loss": 0.2944,
      "step": 11161
    },
    {
      "epoch": 0.6250419979840968,
      "grad_norm": 1.1152257919311523,
      "learning_rate": 5.578e-05,
      "loss": 0.3623,
      "step": 11162
    },
    {
      "epoch": 0.6250979952962258,
      "grad_norm": 1.3937879800796509,
      "learning_rate": 5.5784999999999995e-05,
      "loss": 0.3723,
      "step": 11163
    },
    {
      "epoch": 0.6251539926083548,
      "grad_norm": 1.5608205795288086,
      "learning_rate": 5.579e-05,
      "loss": 0.5245,
      "step": 11164
    },
    {
      "epoch": 0.6252099899204838,
      "grad_norm": 1.3844794034957886,
      "learning_rate": 5.5795e-05,
      "loss": 0.4392,
      "step": 11165
    },
    {
      "epoch": 0.6252659872326128,
      "grad_norm": 1.4845026731491089,
      "learning_rate": 5.580000000000001e-05,
      "loss": 0.5709,
      "step": 11166
    },
    {
      "epoch": 0.6253219845447419,
      "grad_norm": 1.2369791269302368,
      "learning_rate": 5.5805000000000005e-05,
      "loss": 0.4965,
      "step": 11167
    },
    {
      "epoch": 0.6253779818568709,
      "grad_norm": 1.115130066871643,
      "learning_rate": 5.581000000000001e-05,
      "loss": 0.4414,
      "step": 11168
    },
    {
      "epoch": 0.6254339791689999,
      "grad_norm": 1.6319810152053833,
      "learning_rate": 5.581500000000001e-05,
      "loss": 0.5116,
      "step": 11169
    },
    {
      "epoch": 0.6254899764811289,
      "grad_norm": 1.1978349685668945,
      "learning_rate": 5.5820000000000004e-05,
      "loss": 0.4042,
      "step": 11170
    },
    {
      "epoch": 0.6255459737932579,
      "grad_norm": 1.140435814857483,
      "learning_rate": 5.582500000000001e-05,
      "loss": 0.4591,
      "step": 11171
    },
    {
      "epoch": 0.625601971105387,
      "grad_norm": 1.1942261457443237,
      "learning_rate": 5.5830000000000006e-05,
      "loss": 0.5054,
      "step": 11172
    },
    {
      "epoch": 0.625657968417516,
      "grad_norm": 1.1937553882598877,
      "learning_rate": 5.5835e-05,
      "loss": 0.4353,
      "step": 11173
    },
    {
      "epoch": 0.625713965729645,
      "grad_norm": 1.2794886827468872,
      "learning_rate": 5.584e-05,
      "loss": 0.2802,
      "step": 11174
    },
    {
      "epoch": 0.625769963041774,
      "grad_norm": 1.1961913108825684,
      "learning_rate": 5.5845000000000005e-05,
      "loss": 0.464,
      "step": 11175
    },
    {
      "epoch": 0.625825960353903,
      "grad_norm": 1.4396162033081055,
      "learning_rate": 5.585e-05,
      "loss": 0.5098,
      "step": 11176
    },
    {
      "epoch": 0.625881957666032,
      "grad_norm": 1.1948219537734985,
      "learning_rate": 5.5855e-05,
      "loss": 0.431,
      "step": 11177
    },
    {
      "epoch": 0.6259379549781611,
      "grad_norm": 2.665365695953369,
      "learning_rate": 5.5860000000000004e-05,
      "loss": 0.421,
      "step": 11178
    },
    {
      "epoch": 0.6259939522902901,
      "grad_norm": 1.2600346803665161,
      "learning_rate": 5.5865e-05,
      "loss": 0.4265,
      "step": 11179
    },
    {
      "epoch": 0.6260499496024191,
      "grad_norm": 1.7787073850631714,
      "learning_rate": 5.587e-05,
      "loss": 0.5158,
      "step": 11180
    },
    {
      "epoch": 0.6261059469145481,
      "grad_norm": 1.2258793115615845,
      "learning_rate": 5.5875e-05,
      "loss": 0.5191,
      "step": 11181
    },
    {
      "epoch": 0.6261619442266771,
      "grad_norm": 1.2070649862289429,
      "learning_rate": 5.588e-05,
      "loss": 0.451,
      "step": 11182
    },
    {
      "epoch": 0.6262179415388062,
      "grad_norm": 4.887661457061768,
      "learning_rate": 5.5885e-05,
      "loss": 0.516,
      "step": 11183
    },
    {
      "epoch": 0.6262739388509352,
      "grad_norm": 1.2355186939239502,
      "learning_rate": 5.5889999999999995e-05,
      "loss": 0.4068,
      "step": 11184
    },
    {
      "epoch": 0.6263299361630642,
      "grad_norm": 1.3576050996780396,
      "learning_rate": 5.5895e-05,
      "loss": 0.6321,
      "step": 11185
    },
    {
      "epoch": 0.6263859334751932,
      "grad_norm": 1.405781865119934,
      "learning_rate": 5.590000000000001e-05,
      "loss": 0.4014,
      "step": 11186
    },
    {
      "epoch": 0.6264419307873222,
      "grad_norm": 1.2649202346801758,
      "learning_rate": 5.590500000000001e-05,
      "loss": 0.3719,
      "step": 11187
    },
    {
      "epoch": 0.6264979280994513,
      "grad_norm": 1.101420521736145,
      "learning_rate": 5.5910000000000005e-05,
      "loss": 0.3204,
      "step": 11188
    },
    {
      "epoch": 0.6265539254115803,
      "grad_norm": 1.2335408926010132,
      "learning_rate": 5.591500000000001e-05,
      "loss": 0.4149,
      "step": 11189
    },
    {
      "epoch": 0.6266099227237093,
      "grad_norm": 1.3392070531845093,
      "learning_rate": 5.592000000000001e-05,
      "loss": 0.5609,
      "step": 11190
    },
    {
      "epoch": 0.6266659200358383,
      "grad_norm": 1.339861512184143,
      "learning_rate": 5.5925000000000004e-05,
      "loss": 0.4918,
      "step": 11191
    },
    {
      "epoch": 0.6267219173479673,
      "grad_norm": 1.6837819814682007,
      "learning_rate": 5.593e-05,
      "loss": 0.5612,
      "step": 11192
    },
    {
      "epoch": 0.6267779146600964,
      "grad_norm": 1.3606007099151611,
      "learning_rate": 5.5935000000000006e-05,
      "loss": 0.4434,
      "step": 11193
    },
    {
      "epoch": 0.6268339119722254,
      "grad_norm": 1.2954528331756592,
      "learning_rate": 5.594e-05,
      "loss": 0.5468,
      "step": 11194
    },
    {
      "epoch": 0.6268899092843544,
      "grad_norm": 6.007063865661621,
      "learning_rate": 5.5945e-05,
      "loss": 0.4323,
      "step": 11195
    },
    {
      "epoch": 0.6269459065964834,
      "grad_norm": 1.3410000801086426,
      "learning_rate": 5.5950000000000005e-05,
      "loss": 0.3654,
      "step": 11196
    },
    {
      "epoch": 0.6270019039086124,
      "grad_norm": 1.3862791061401367,
      "learning_rate": 5.5955e-05,
      "loss": 0.515,
      "step": 11197
    },
    {
      "epoch": 0.6270579012207415,
      "grad_norm": 1.2850056886672974,
      "learning_rate": 5.596e-05,
      "loss": 0.4258,
      "step": 11198
    },
    {
      "epoch": 0.6271138985328705,
      "grad_norm": 1.225555658340454,
      "learning_rate": 5.5965000000000004e-05,
      "loss": 0.4058,
      "step": 11199
    },
    {
      "epoch": 0.6271698958449995,
      "grad_norm": 1.2777115106582642,
      "learning_rate": 5.597e-05,
      "loss": 0.3095,
      "step": 11200
    },
    {
      "epoch": 0.6272258931571285,
      "grad_norm": 1.643985629081726,
      "learning_rate": 5.5975e-05,
      "loss": 0.4018,
      "step": 11201
    },
    {
      "epoch": 0.6272818904692575,
      "grad_norm": 1.3758684396743774,
      "learning_rate": 5.5979999999999996e-05,
      "loss": 0.5092,
      "step": 11202
    },
    {
      "epoch": 0.6273378877813864,
      "grad_norm": 1.4489269256591797,
      "learning_rate": 5.5985e-05,
      "loss": 0.4493,
      "step": 11203
    },
    {
      "epoch": 0.6273938850935155,
      "grad_norm": 1.2936818599700928,
      "learning_rate": 5.599e-05,
      "loss": 0.4872,
      "step": 11204
    },
    {
      "epoch": 0.6274498824056445,
      "grad_norm": 1.2241450548171997,
      "learning_rate": 5.5994999999999995e-05,
      "loss": 0.3612,
      "step": 11205
    },
    {
      "epoch": 0.6275058797177735,
      "grad_norm": 1.1195197105407715,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 0.3864,
      "step": 11206
    },
    {
      "epoch": 0.6275618770299025,
      "grad_norm": 1.2647725343704224,
      "learning_rate": 5.600500000000001e-05,
      "loss": 0.4808,
      "step": 11207
    },
    {
      "epoch": 0.6276178743420315,
      "grad_norm": 1.250089406967163,
      "learning_rate": 5.601000000000001e-05,
      "loss": 0.402,
      "step": 11208
    },
    {
      "epoch": 0.6276738716541606,
      "grad_norm": 1.9865802526474,
      "learning_rate": 5.6015000000000005e-05,
      "loss": 0.4525,
      "step": 11209
    },
    {
      "epoch": 0.6277298689662896,
      "grad_norm": 1.143970251083374,
      "learning_rate": 5.602000000000001e-05,
      "loss": 0.3206,
      "step": 11210
    },
    {
      "epoch": 0.6277858662784186,
      "grad_norm": 1.2424830198287964,
      "learning_rate": 5.6025000000000007e-05,
      "loss": 0.4686,
      "step": 11211
    },
    {
      "epoch": 0.6278418635905476,
      "grad_norm": 1.3840773105621338,
      "learning_rate": 5.6030000000000004e-05,
      "loss": 0.3806,
      "step": 11212
    },
    {
      "epoch": 0.6278978609026766,
      "grad_norm": 1.2964428663253784,
      "learning_rate": 5.6035e-05,
      "loss": 0.4664,
      "step": 11213
    },
    {
      "epoch": 0.6279538582148056,
      "grad_norm": 1.2423175573349,
      "learning_rate": 5.6040000000000006e-05,
      "loss": 0.3617,
      "step": 11214
    },
    {
      "epoch": 0.6280098555269347,
      "grad_norm": 1.3496135473251343,
      "learning_rate": 5.6045e-05,
      "loss": 0.5255,
      "step": 11215
    },
    {
      "epoch": 0.6280658528390637,
      "grad_norm": 1.2265090942382812,
      "learning_rate": 5.605e-05,
      "loss": 0.4631,
      "step": 11216
    },
    {
      "epoch": 0.6281218501511927,
      "grad_norm": 1.2823275327682495,
      "learning_rate": 5.6055000000000005e-05,
      "loss": 0.3997,
      "step": 11217
    },
    {
      "epoch": 0.6281778474633217,
      "grad_norm": 1.5288941860198975,
      "learning_rate": 5.606e-05,
      "loss": 0.5032,
      "step": 11218
    },
    {
      "epoch": 0.6282338447754507,
      "grad_norm": 1.2193326950073242,
      "learning_rate": 5.6065e-05,
      "loss": 0.401,
      "step": 11219
    },
    {
      "epoch": 0.6282898420875798,
      "grad_norm": 1.213442087173462,
      "learning_rate": 5.6070000000000004e-05,
      "loss": 0.4321,
      "step": 11220
    },
    {
      "epoch": 0.6283458393997088,
      "grad_norm": 1.3099799156188965,
      "learning_rate": 5.6075e-05,
      "loss": 0.411,
      "step": 11221
    },
    {
      "epoch": 0.6284018367118378,
      "grad_norm": 1.056294322013855,
      "learning_rate": 5.608e-05,
      "loss": 0.2766,
      "step": 11222
    },
    {
      "epoch": 0.6284578340239668,
      "grad_norm": 1.1698724031448364,
      "learning_rate": 5.6084999999999996e-05,
      "loss": 0.413,
      "step": 11223
    },
    {
      "epoch": 0.6285138313360958,
      "grad_norm": 1.145461916923523,
      "learning_rate": 5.609e-05,
      "loss": 0.2676,
      "step": 11224
    },
    {
      "epoch": 0.6285698286482249,
      "grad_norm": 1.2429014444351196,
      "learning_rate": 5.6095e-05,
      "loss": 0.4648,
      "step": 11225
    },
    {
      "epoch": 0.6286258259603539,
      "grad_norm": 1.3581485748291016,
      "learning_rate": 5.610000000000001e-05,
      "loss": 0.4733,
      "step": 11226
    },
    {
      "epoch": 0.6286818232724829,
      "grad_norm": 1.0891610383987427,
      "learning_rate": 5.6105000000000006e-05,
      "loss": 0.3796,
      "step": 11227
    },
    {
      "epoch": 0.6287378205846119,
      "grad_norm": 1.4651587009429932,
      "learning_rate": 5.611000000000001e-05,
      "loss": 0.5484,
      "step": 11228
    },
    {
      "epoch": 0.6287938178967409,
      "grad_norm": 1.1366677284240723,
      "learning_rate": 5.611500000000001e-05,
      "loss": 0.3699,
      "step": 11229
    },
    {
      "epoch": 0.62884981520887,
      "grad_norm": 1.1697146892547607,
      "learning_rate": 5.6120000000000005e-05,
      "loss": 0.4184,
      "step": 11230
    },
    {
      "epoch": 0.628905812520999,
      "grad_norm": 1.4018419981002808,
      "learning_rate": 5.6125e-05,
      "loss": 0.5213,
      "step": 11231
    },
    {
      "epoch": 0.628961809833128,
      "grad_norm": 1.3317739963531494,
      "learning_rate": 5.6130000000000006e-05,
      "loss": 0.4804,
      "step": 11232
    },
    {
      "epoch": 0.629017807145257,
      "grad_norm": 1.0773656368255615,
      "learning_rate": 5.6135000000000004e-05,
      "loss": 0.3926,
      "step": 11233
    },
    {
      "epoch": 0.629073804457386,
      "grad_norm": 2.1188793182373047,
      "learning_rate": 5.614e-05,
      "loss": 0.3593,
      "step": 11234
    },
    {
      "epoch": 0.629129801769515,
      "grad_norm": 1.1686667203903198,
      "learning_rate": 5.6145000000000005e-05,
      "loss": 0.3245,
      "step": 11235
    },
    {
      "epoch": 0.6291857990816441,
      "grad_norm": 1.5293480157852173,
      "learning_rate": 5.615e-05,
      "loss": 0.501,
      "step": 11236
    },
    {
      "epoch": 0.6292417963937731,
      "grad_norm": 1.2358907461166382,
      "learning_rate": 5.6155e-05,
      "loss": 0.3587,
      "step": 11237
    },
    {
      "epoch": 0.6292977937059021,
      "grad_norm": 6.005584239959717,
      "learning_rate": 5.6160000000000004e-05,
      "loss": 0.4272,
      "step": 11238
    },
    {
      "epoch": 0.6293537910180311,
      "grad_norm": 1.945157527923584,
      "learning_rate": 5.6165e-05,
      "loss": 0.5091,
      "step": 11239
    },
    {
      "epoch": 0.6294097883301601,
      "grad_norm": 1.199906349182129,
      "learning_rate": 5.617e-05,
      "loss": 0.4088,
      "step": 11240
    },
    {
      "epoch": 0.6294657856422892,
      "grad_norm": 1.1609842777252197,
      "learning_rate": 5.6175e-05,
      "loss": 0.37,
      "step": 11241
    },
    {
      "epoch": 0.6295217829544182,
      "grad_norm": 1.303733229637146,
      "learning_rate": 5.618e-05,
      "loss": 0.3684,
      "step": 11242
    },
    {
      "epoch": 0.6295777802665472,
      "grad_norm": 1.3500808477401733,
      "learning_rate": 5.6185e-05,
      "loss": 0.49,
      "step": 11243
    },
    {
      "epoch": 0.6296337775786762,
      "grad_norm": 1.1155012845993042,
      "learning_rate": 5.6189999999999996e-05,
      "loss": 0.3952,
      "step": 11244
    },
    {
      "epoch": 0.6296897748908052,
      "grad_norm": 1.3555607795715332,
      "learning_rate": 5.6195e-05,
      "loss": 0.4157,
      "step": 11245
    },
    {
      "epoch": 0.6297457722029343,
      "grad_norm": 1.3759735822677612,
      "learning_rate": 5.620000000000001e-05,
      "loss": 0.3974,
      "step": 11246
    },
    {
      "epoch": 0.6298017695150633,
      "grad_norm": 1.5575779676437378,
      "learning_rate": 5.620500000000001e-05,
      "loss": 0.3911,
      "step": 11247
    },
    {
      "epoch": 0.6298577668271923,
      "grad_norm": 1.4618136882781982,
      "learning_rate": 5.6210000000000006e-05,
      "loss": 0.4637,
      "step": 11248
    },
    {
      "epoch": 0.6299137641393213,
      "grad_norm": 1.4149360656738281,
      "learning_rate": 5.621500000000001e-05,
      "loss": 0.4352,
      "step": 11249
    },
    {
      "epoch": 0.6299697614514503,
      "grad_norm": 1.1021887063980103,
      "learning_rate": 5.622000000000001e-05,
      "loss": 0.2918,
      "step": 11250
    },
    {
      "epoch": 0.6300257587635794,
      "grad_norm": 1.0999575853347778,
      "learning_rate": 5.6225000000000005e-05,
      "loss": 0.4268,
      "step": 11251
    },
    {
      "epoch": 0.6300817560757084,
      "grad_norm": 1.0819861888885498,
      "learning_rate": 5.623e-05,
      "loss": 0.5071,
      "step": 11252
    },
    {
      "epoch": 0.6301377533878374,
      "grad_norm": 1.3975441455841064,
      "learning_rate": 5.6235000000000006e-05,
      "loss": 0.4135,
      "step": 11253
    },
    {
      "epoch": 0.6301937506999664,
      "grad_norm": 1.3753687143325806,
      "learning_rate": 5.6240000000000004e-05,
      "loss": 0.3901,
      "step": 11254
    },
    {
      "epoch": 0.6302497480120954,
      "grad_norm": 1.2578248977661133,
      "learning_rate": 5.6245e-05,
      "loss": 0.4152,
      "step": 11255
    },
    {
      "epoch": 0.6303057453242245,
      "grad_norm": 1.2870309352874756,
      "learning_rate": 5.6250000000000005e-05,
      "loss": 0.4242,
      "step": 11256
    },
    {
      "epoch": 0.6303617426363535,
      "grad_norm": 1.084806203842163,
      "learning_rate": 5.6255e-05,
      "loss": 0.4124,
      "step": 11257
    },
    {
      "epoch": 0.6304177399484825,
      "grad_norm": 1.4478613138198853,
      "learning_rate": 5.626e-05,
      "loss": 0.3683,
      "step": 11258
    },
    {
      "epoch": 0.6304737372606115,
      "grad_norm": 1.4411780834197998,
      "learning_rate": 5.6265000000000004e-05,
      "loss": 0.5716,
      "step": 11259
    },
    {
      "epoch": 0.6305297345727405,
      "grad_norm": 1.5875511169433594,
      "learning_rate": 5.627e-05,
      "loss": 0.4225,
      "step": 11260
    },
    {
      "epoch": 0.6305857318848695,
      "grad_norm": 1.0330952405929565,
      "learning_rate": 5.6275e-05,
      "loss": 0.4313,
      "step": 11261
    },
    {
      "epoch": 0.6306417291969986,
      "grad_norm": 1.1702361106872559,
      "learning_rate": 5.6279999999999996e-05,
      "loss": 0.4511,
      "step": 11262
    },
    {
      "epoch": 0.6306977265091276,
      "grad_norm": 1.1063364744186401,
      "learning_rate": 5.6285e-05,
      "loss": 0.3643,
      "step": 11263
    },
    {
      "epoch": 0.6307537238212566,
      "grad_norm": 1.007749080657959,
      "learning_rate": 5.629e-05,
      "loss": 0.3217,
      "step": 11264
    },
    {
      "epoch": 0.6308097211333856,
      "grad_norm": 1.3514572381973267,
      "learning_rate": 5.6294999999999995e-05,
      "loss": 0.4353,
      "step": 11265
    },
    {
      "epoch": 0.6308657184455146,
      "grad_norm": 1.3110345602035522,
      "learning_rate": 5.63e-05,
      "loss": 0.3841,
      "step": 11266
    },
    {
      "epoch": 0.6309217157576437,
      "grad_norm": 1.281907081604004,
      "learning_rate": 5.630500000000001e-05,
      "loss": 0.4366,
      "step": 11267
    },
    {
      "epoch": 0.6309777130697727,
      "grad_norm": 1.4426237344741821,
      "learning_rate": 5.631000000000001e-05,
      "loss": 0.4704,
      "step": 11268
    },
    {
      "epoch": 0.6310337103819017,
      "grad_norm": 1.2236207723617554,
      "learning_rate": 5.6315000000000005e-05,
      "loss": 0.381,
      "step": 11269
    },
    {
      "epoch": 0.6310897076940307,
      "grad_norm": 1.2905206680297852,
      "learning_rate": 5.632e-05,
      "loss": 0.4791,
      "step": 11270
    },
    {
      "epoch": 0.6311457050061597,
      "grad_norm": 1.352844476699829,
      "learning_rate": 5.632500000000001e-05,
      "loss": 0.3825,
      "step": 11271
    },
    {
      "epoch": 0.6312017023182888,
      "grad_norm": 1.462010145187378,
      "learning_rate": 5.6330000000000004e-05,
      "loss": 0.7624,
      "step": 11272
    },
    {
      "epoch": 0.6312576996304178,
      "grad_norm": 1.4606443643569946,
      "learning_rate": 5.6335e-05,
      "loss": 0.4573,
      "step": 11273
    },
    {
      "epoch": 0.6313136969425468,
      "grad_norm": 1.1596461534500122,
      "learning_rate": 5.6340000000000006e-05,
      "loss": 0.4066,
      "step": 11274
    },
    {
      "epoch": 0.6313696942546758,
      "grad_norm": 1.125241994857788,
      "learning_rate": 5.6345000000000003e-05,
      "loss": 0.3541,
      "step": 11275
    },
    {
      "epoch": 0.6314256915668048,
      "grad_norm": 1.3315197229385376,
      "learning_rate": 5.635e-05,
      "loss": 0.4487,
      "step": 11276
    },
    {
      "epoch": 0.6314816888789339,
      "grad_norm": 1.1584787368774414,
      "learning_rate": 5.6355000000000005e-05,
      "loss": 0.2897,
      "step": 11277
    },
    {
      "epoch": 0.6315376861910629,
      "grad_norm": 1.058961272239685,
      "learning_rate": 5.636e-05,
      "loss": 0.3676,
      "step": 11278
    },
    {
      "epoch": 0.6315936835031919,
      "grad_norm": 1.2803348302841187,
      "learning_rate": 5.6365e-05,
      "loss": 0.5269,
      "step": 11279
    },
    {
      "epoch": 0.6316496808153209,
      "grad_norm": 1.4021506309509277,
      "learning_rate": 5.637e-05,
      "loss": 0.3149,
      "step": 11280
    },
    {
      "epoch": 0.6317056781274499,
      "grad_norm": 1.371085286140442,
      "learning_rate": 5.6375e-05,
      "loss": 0.5523,
      "step": 11281
    },
    {
      "epoch": 0.631761675439579,
      "grad_norm": 1.7258166074752808,
      "learning_rate": 5.638e-05,
      "loss": 0.637,
      "step": 11282
    },
    {
      "epoch": 0.631817672751708,
      "grad_norm": 1.157315731048584,
      "learning_rate": 5.6384999999999996e-05,
      "loss": 0.3429,
      "step": 11283
    },
    {
      "epoch": 0.631873670063837,
      "grad_norm": 1.417036771774292,
      "learning_rate": 5.639e-05,
      "loss": 0.4629,
      "step": 11284
    },
    {
      "epoch": 0.631929667375966,
      "grad_norm": 1.314966082572937,
      "learning_rate": 5.6395e-05,
      "loss": 0.5163,
      "step": 11285
    },
    {
      "epoch": 0.6319856646880949,
      "grad_norm": 1.2751572132110596,
      "learning_rate": 5.6399999999999995e-05,
      "loss": 0.4377,
      "step": 11286
    },
    {
      "epoch": 0.6320416620002239,
      "grad_norm": 1.1246416568756104,
      "learning_rate": 5.6405000000000006e-05,
      "loss": 0.4409,
      "step": 11287
    },
    {
      "epoch": 0.632097659312353,
      "grad_norm": 1.2444161176681519,
      "learning_rate": 5.641000000000001e-05,
      "loss": 0.4737,
      "step": 11288
    },
    {
      "epoch": 0.632153656624482,
      "grad_norm": 1.0859520435333252,
      "learning_rate": 5.641500000000001e-05,
      "loss": 0.4256,
      "step": 11289
    },
    {
      "epoch": 0.632209653936611,
      "grad_norm": 1.3945276737213135,
      "learning_rate": 5.6420000000000005e-05,
      "loss": 0.3861,
      "step": 11290
    },
    {
      "epoch": 0.63226565124874,
      "grad_norm": 1.2874191999435425,
      "learning_rate": 5.6425e-05,
      "loss": 0.4776,
      "step": 11291
    },
    {
      "epoch": 0.632321648560869,
      "grad_norm": 1.2917265892028809,
      "learning_rate": 5.643000000000001e-05,
      "loss": 0.4273,
      "step": 11292
    },
    {
      "epoch": 0.632377645872998,
      "grad_norm": 1.1397157907485962,
      "learning_rate": 5.6435000000000004e-05,
      "loss": 0.3987,
      "step": 11293
    },
    {
      "epoch": 0.6324336431851271,
      "grad_norm": 1.1848894357681274,
      "learning_rate": 5.644e-05,
      "loss": 0.4868,
      "step": 11294
    },
    {
      "epoch": 0.6324896404972561,
      "grad_norm": 1.2006059885025024,
      "learning_rate": 5.6445000000000006e-05,
      "loss": 0.4025,
      "step": 11295
    },
    {
      "epoch": 0.6325456378093851,
      "grad_norm": 1.505966067314148,
      "learning_rate": 5.645e-05,
      "loss": 0.4714,
      "step": 11296
    },
    {
      "epoch": 0.6326016351215141,
      "grad_norm": 1.1986751556396484,
      "learning_rate": 5.6455e-05,
      "loss": 0.4536,
      "step": 11297
    },
    {
      "epoch": 0.6326576324336431,
      "grad_norm": 1.4272018671035767,
      "learning_rate": 5.6460000000000005e-05,
      "loss": 0.4774,
      "step": 11298
    },
    {
      "epoch": 0.6327136297457722,
      "grad_norm": 1.498241901397705,
      "learning_rate": 5.6465e-05,
      "loss": 0.5556,
      "step": 11299
    },
    {
      "epoch": 0.6327696270579012,
      "grad_norm": 6.183231830596924,
      "learning_rate": 5.647e-05,
      "loss": 0.4907,
      "step": 11300
    },
    {
      "epoch": 0.6328256243700302,
      "grad_norm": 1.3515992164611816,
      "learning_rate": 5.6475e-05,
      "loss": 0.4121,
      "step": 11301
    },
    {
      "epoch": 0.6328816216821592,
      "grad_norm": 1.2577941417694092,
      "learning_rate": 5.648e-05,
      "loss": 0.4679,
      "step": 11302
    },
    {
      "epoch": 0.6329376189942882,
      "grad_norm": 1.0943795442581177,
      "learning_rate": 5.6485e-05,
      "loss": 0.3358,
      "step": 11303
    },
    {
      "epoch": 0.6329936163064173,
      "grad_norm": 1.1649833917617798,
      "learning_rate": 5.6489999999999996e-05,
      "loss": 0.4517,
      "step": 11304
    },
    {
      "epoch": 0.6330496136185463,
      "grad_norm": 1.3486391305923462,
      "learning_rate": 5.6495e-05,
      "loss": 0.3764,
      "step": 11305
    },
    {
      "epoch": 0.6331056109306753,
      "grad_norm": 1.4467990398406982,
      "learning_rate": 5.65e-05,
      "loss": 0.4745,
      "step": 11306
    },
    {
      "epoch": 0.6331616082428043,
      "grad_norm": 1.2465481758117676,
      "learning_rate": 5.650500000000001e-05,
      "loss": 0.4743,
      "step": 11307
    },
    {
      "epoch": 0.6332176055549333,
      "grad_norm": 1.2886686325073242,
      "learning_rate": 5.6510000000000006e-05,
      "loss": 0.4211,
      "step": 11308
    },
    {
      "epoch": 0.6332736028670624,
      "grad_norm": 1.816685676574707,
      "learning_rate": 5.6515000000000004e-05,
      "loss": 0.4136,
      "step": 11309
    },
    {
      "epoch": 0.6333296001791914,
      "grad_norm": 1.1860828399658203,
      "learning_rate": 5.652000000000001e-05,
      "loss": 0.4991,
      "step": 11310
    },
    {
      "epoch": 0.6333855974913204,
      "grad_norm": 1.1487269401550293,
      "learning_rate": 5.6525000000000005e-05,
      "loss": 0.4361,
      "step": 11311
    },
    {
      "epoch": 0.6334415948034494,
      "grad_norm": 1.3433301448822021,
      "learning_rate": 5.653e-05,
      "loss": 0.3952,
      "step": 11312
    },
    {
      "epoch": 0.6334975921155784,
      "grad_norm": 1.2602077722549438,
      "learning_rate": 5.653500000000001e-05,
      "loss": 0.5292,
      "step": 11313
    },
    {
      "epoch": 0.6335535894277075,
      "grad_norm": 1.15339994430542,
      "learning_rate": 5.6540000000000004e-05,
      "loss": 0.3208,
      "step": 11314
    },
    {
      "epoch": 0.6336095867398365,
      "grad_norm": 1.2489097118377686,
      "learning_rate": 5.6545e-05,
      "loss": 0.3924,
      "step": 11315
    },
    {
      "epoch": 0.6336655840519655,
      "grad_norm": 1.2408370971679688,
      "learning_rate": 5.6550000000000006e-05,
      "loss": 0.311,
      "step": 11316
    },
    {
      "epoch": 0.6337215813640945,
      "grad_norm": 1.1563353538513184,
      "learning_rate": 5.6555e-05,
      "loss": 0.3385,
      "step": 11317
    },
    {
      "epoch": 0.6337775786762235,
      "grad_norm": 1.1346092224121094,
      "learning_rate": 5.656e-05,
      "loss": 0.2971,
      "step": 11318
    },
    {
      "epoch": 0.6338335759883525,
      "grad_norm": 1.3084992170333862,
      "learning_rate": 5.6565e-05,
      "loss": 0.4301,
      "step": 11319
    },
    {
      "epoch": 0.6338895733004816,
      "grad_norm": 1.445418357849121,
      "learning_rate": 5.657e-05,
      "loss": 0.4382,
      "step": 11320
    },
    {
      "epoch": 0.6339455706126106,
      "grad_norm": 1.1264489889144897,
      "learning_rate": 5.6575e-05,
      "loss": 0.447,
      "step": 11321
    },
    {
      "epoch": 0.6340015679247396,
      "grad_norm": 1.0811023712158203,
      "learning_rate": 5.658e-05,
      "loss": 0.3592,
      "step": 11322
    },
    {
      "epoch": 0.6340575652368686,
      "grad_norm": 1.1096742153167725,
      "learning_rate": 5.6585e-05,
      "loss": 0.3223,
      "step": 11323
    },
    {
      "epoch": 0.6341135625489976,
      "grad_norm": 1.3272207975387573,
      "learning_rate": 5.659e-05,
      "loss": 0.4864,
      "step": 11324
    },
    {
      "epoch": 0.6341695598611267,
      "grad_norm": 1.095194697380066,
      "learning_rate": 5.6594999999999996e-05,
      "loss": 0.3495,
      "step": 11325
    },
    {
      "epoch": 0.6342255571732557,
      "grad_norm": 1.1297070980072021,
      "learning_rate": 5.66e-05,
      "loss": 0.4301,
      "step": 11326
    },
    {
      "epoch": 0.6342815544853847,
      "grad_norm": 1.390947937965393,
      "learning_rate": 5.660500000000001e-05,
      "loss": 0.414,
      "step": 11327
    },
    {
      "epoch": 0.6343375517975137,
      "grad_norm": 1.2035391330718994,
      "learning_rate": 5.661000000000001e-05,
      "loss": 0.4177,
      "step": 11328
    },
    {
      "epoch": 0.6343935491096427,
      "grad_norm": 1.2064096927642822,
      "learning_rate": 5.6615000000000006e-05,
      "loss": 0.3505,
      "step": 11329
    },
    {
      "epoch": 0.6344495464217718,
      "grad_norm": 1.132487177848816,
      "learning_rate": 5.6620000000000003e-05,
      "loss": 0.4181,
      "step": 11330
    },
    {
      "epoch": 0.6345055437339008,
      "grad_norm": 1.305894136428833,
      "learning_rate": 5.662500000000001e-05,
      "loss": 0.4296,
      "step": 11331
    },
    {
      "epoch": 0.6345615410460298,
      "grad_norm": 1.579166054725647,
      "learning_rate": 5.6630000000000005e-05,
      "loss": 0.4964,
      "step": 11332
    },
    {
      "epoch": 0.6346175383581588,
      "grad_norm": 1.2170345783233643,
      "learning_rate": 5.6635e-05,
      "loss": 0.4896,
      "step": 11333
    },
    {
      "epoch": 0.6346735356702878,
      "grad_norm": 1.304287075996399,
      "learning_rate": 5.6640000000000007e-05,
      "loss": 0.5304,
      "step": 11334
    },
    {
      "epoch": 0.6347295329824169,
      "grad_norm": 1.2192530632019043,
      "learning_rate": 5.6645000000000004e-05,
      "loss": 0.3684,
      "step": 11335
    },
    {
      "epoch": 0.6347855302945459,
      "grad_norm": 1.2074230909347534,
      "learning_rate": 5.665e-05,
      "loss": 0.4713,
      "step": 11336
    },
    {
      "epoch": 0.6348415276066749,
      "grad_norm": 1.4461374282836914,
      "learning_rate": 5.6655000000000006e-05,
      "loss": 0.476,
      "step": 11337
    },
    {
      "epoch": 0.6348975249188039,
      "grad_norm": 1.287718415260315,
      "learning_rate": 5.666e-05,
      "loss": 0.4544,
      "step": 11338
    },
    {
      "epoch": 0.6349535222309329,
      "grad_norm": 1.1624178886413574,
      "learning_rate": 5.6665e-05,
      "loss": 0.3842,
      "step": 11339
    },
    {
      "epoch": 0.635009519543062,
      "grad_norm": 1.1182670593261719,
      "learning_rate": 5.667e-05,
      "loss": 0.4279,
      "step": 11340
    },
    {
      "epoch": 0.635065516855191,
      "grad_norm": 1.5850125551223755,
      "learning_rate": 5.6675e-05,
      "loss": 0.616,
      "step": 11341
    },
    {
      "epoch": 0.63512151416732,
      "grad_norm": 1.3708568811416626,
      "learning_rate": 5.668e-05,
      "loss": 0.615,
      "step": 11342
    },
    {
      "epoch": 0.635177511479449,
      "grad_norm": 1.1200449466705322,
      "learning_rate": 5.6685e-05,
      "loss": 0.35,
      "step": 11343
    },
    {
      "epoch": 0.635233508791578,
      "grad_norm": 1.2273062467575073,
      "learning_rate": 5.669e-05,
      "loss": 0.3545,
      "step": 11344
    },
    {
      "epoch": 0.635289506103707,
      "grad_norm": 1.147248387336731,
      "learning_rate": 5.6695e-05,
      "loss": 0.4342,
      "step": 11345
    },
    {
      "epoch": 0.6353455034158361,
      "grad_norm": 1.3605453968048096,
      "learning_rate": 5.6699999999999996e-05,
      "loss": 0.4196,
      "step": 11346
    },
    {
      "epoch": 0.6354015007279651,
      "grad_norm": 1.3121367692947388,
      "learning_rate": 5.670500000000001e-05,
      "loss": 0.4314,
      "step": 11347
    },
    {
      "epoch": 0.6354574980400941,
      "grad_norm": 1.303649663925171,
      "learning_rate": 5.6710000000000004e-05,
      "loss": 0.4225,
      "step": 11348
    },
    {
      "epoch": 0.6355134953522231,
      "grad_norm": 1.190080165863037,
      "learning_rate": 5.671500000000001e-05,
      "loss": 0.436,
      "step": 11349
    },
    {
      "epoch": 0.6355694926643521,
      "grad_norm": 1.2896257638931274,
      "learning_rate": 5.6720000000000006e-05,
      "loss": 0.3563,
      "step": 11350
    },
    {
      "epoch": 0.6356254899764812,
      "grad_norm": 1.3367797136306763,
      "learning_rate": 5.6725e-05,
      "loss": 0.4411,
      "step": 11351
    },
    {
      "epoch": 0.6356814872886102,
      "grad_norm": 1.5831478834152222,
      "learning_rate": 5.673000000000001e-05,
      "loss": 0.5588,
      "step": 11352
    },
    {
      "epoch": 0.6357374846007392,
      "grad_norm": 1.2519817352294922,
      "learning_rate": 5.6735000000000005e-05,
      "loss": 0.4346,
      "step": 11353
    },
    {
      "epoch": 0.6357934819128682,
      "grad_norm": 1.526275634765625,
      "learning_rate": 5.674e-05,
      "loss": 0.4632,
      "step": 11354
    },
    {
      "epoch": 0.6358494792249972,
      "grad_norm": 1.3396154642105103,
      "learning_rate": 5.6745000000000006e-05,
      "loss": 0.5087,
      "step": 11355
    },
    {
      "epoch": 0.6359054765371263,
      "grad_norm": 1.5325535535812378,
      "learning_rate": 5.6750000000000004e-05,
      "loss": 0.4743,
      "step": 11356
    },
    {
      "epoch": 0.6359614738492553,
      "grad_norm": 1.4574859142303467,
      "learning_rate": 5.6755e-05,
      "loss": 0.4546,
      "step": 11357
    },
    {
      "epoch": 0.6360174711613843,
      "grad_norm": 1.2215218544006348,
      "learning_rate": 5.6760000000000005e-05,
      "loss": 0.4006,
      "step": 11358
    },
    {
      "epoch": 0.6360734684735133,
      "grad_norm": 1.2708491086959839,
      "learning_rate": 5.6765e-05,
      "loss": 0.575,
      "step": 11359
    },
    {
      "epoch": 0.6361294657856423,
      "grad_norm": 1.4911993741989136,
      "learning_rate": 5.677e-05,
      "loss": 0.4185,
      "step": 11360
    },
    {
      "epoch": 0.6361854630977714,
      "grad_norm": 1.4460312128067017,
      "learning_rate": 5.6775e-05,
      "loss": 0.5021,
      "step": 11361
    },
    {
      "epoch": 0.6362414604099004,
      "grad_norm": 1.0688856840133667,
      "learning_rate": 5.678e-05,
      "loss": 0.3547,
      "step": 11362
    },
    {
      "epoch": 0.6362974577220294,
      "grad_norm": 1.122283935546875,
      "learning_rate": 5.6785e-05,
      "loss": 0.4026,
      "step": 11363
    },
    {
      "epoch": 0.6363534550341584,
      "grad_norm": 1.3614938259124756,
      "learning_rate": 5.679e-05,
      "loss": 0.5103,
      "step": 11364
    },
    {
      "epoch": 0.6364094523462874,
      "grad_norm": 1.4329949617385864,
      "learning_rate": 5.6795e-05,
      "loss": 0.405,
      "step": 11365
    },
    {
      "epoch": 0.6364654496584164,
      "grad_norm": 1.2382676601409912,
      "learning_rate": 5.68e-05,
      "loss": 0.4251,
      "step": 11366
    },
    {
      "epoch": 0.6365214469705455,
      "grad_norm": 1.3758963346481323,
      "learning_rate": 5.680500000000001e-05,
      "loss": 0.5959,
      "step": 11367
    },
    {
      "epoch": 0.6365774442826744,
      "grad_norm": 1.1326810121536255,
      "learning_rate": 5.681000000000001e-05,
      "loss": 0.4424,
      "step": 11368
    },
    {
      "epoch": 0.6366334415948034,
      "grad_norm": 1.1721724271774292,
      "learning_rate": 5.6815000000000004e-05,
      "loss": 0.4848,
      "step": 11369
    },
    {
      "epoch": 0.6366894389069324,
      "grad_norm": 1.471866250038147,
      "learning_rate": 5.682000000000001e-05,
      "loss": 0.3995,
      "step": 11370
    },
    {
      "epoch": 0.6367454362190614,
      "grad_norm": 1.4628483057022095,
      "learning_rate": 5.6825000000000006e-05,
      "loss": 0.4036,
      "step": 11371
    },
    {
      "epoch": 0.6368014335311905,
      "grad_norm": 1.1856060028076172,
      "learning_rate": 5.683e-05,
      "loss": 0.422,
      "step": 11372
    },
    {
      "epoch": 0.6368574308433195,
      "grad_norm": 1.3349860906600952,
      "learning_rate": 5.683500000000001e-05,
      "loss": 0.349,
      "step": 11373
    },
    {
      "epoch": 0.6369134281554485,
      "grad_norm": 1.2007198333740234,
      "learning_rate": 5.6840000000000005e-05,
      "loss": 0.3308,
      "step": 11374
    },
    {
      "epoch": 0.6369694254675775,
      "grad_norm": 1.6263080835342407,
      "learning_rate": 5.6845e-05,
      "loss": 0.4246,
      "step": 11375
    },
    {
      "epoch": 0.6370254227797065,
      "grad_norm": 1.3804060220718384,
      "learning_rate": 5.6850000000000006e-05,
      "loss": 0.5547,
      "step": 11376
    },
    {
      "epoch": 0.6370814200918355,
      "grad_norm": 1.197320580482483,
      "learning_rate": 5.6855000000000004e-05,
      "loss": 0.4101,
      "step": 11377
    },
    {
      "epoch": 0.6371374174039646,
      "grad_norm": 1.2640727758407593,
      "learning_rate": 5.686e-05,
      "loss": 0.407,
      "step": 11378
    },
    {
      "epoch": 0.6371934147160936,
      "grad_norm": 1.3745468854904175,
      "learning_rate": 5.6865e-05,
      "loss": 0.4645,
      "step": 11379
    },
    {
      "epoch": 0.6372494120282226,
      "grad_norm": 1.1405552625656128,
      "learning_rate": 5.687e-05,
      "loss": 0.3982,
      "step": 11380
    },
    {
      "epoch": 0.6373054093403516,
      "grad_norm": 1.1586295366287231,
      "learning_rate": 5.6875e-05,
      "loss": 0.3587,
      "step": 11381
    },
    {
      "epoch": 0.6373614066524806,
      "grad_norm": 1.2616404294967651,
      "learning_rate": 5.688e-05,
      "loss": 0.4723,
      "step": 11382
    },
    {
      "epoch": 0.6374174039646097,
      "grad_norm": 1.6011511087417603,
      "learning_rate": 5.6885e-05,
      "loss": 0.4597,
      "step": 11383
    },
    {
      "epoch": 0.6374734012767387,
      "grad_norm": 1.3609460592269897,
      "learning_rate": 5.689e-05,
      "loss": 0.4977,
      "step": 11384
    },
    {
      "epoch": 0.6375293985888677,
      "grad_norm": 1.7835408449172974,
      "learning_rate": 5.6894999999999997e-05,
      "loss": 0.5518,
      "step": 11385
    },
    {
      "epoch": 0.6375853959009967,
      "grad_norm": 1.487677812576294,
      "learning_rate": 5.69e-05,
      "loss": 0.5093,
      "step": 11386
    },
    {
      "epoch": 0.6376413932131257,
      "grad_norm": 1.3706413507461548,
      "learning_rate": 5.6905e-05,
      "loss": 0.5073,
      "step": 11387
    },
    {
      "epoch": 0.6376973905252548,
      "grad_norm": 1.1740338802337646,
      "learning_rate": 5.691000000000001e-05,
      "loss": 0.4306,
      "step": 11388
    },
    {
      "epoch": 0.6377533878373838,
      "grad_norm": 1.3596729040145874,
      "learning_rate": 5.6915000000000006e-05,
      "loss": 0.5845,
      "step": 11389
    },
    {
      "epoch": 0.6378093851495128,
      "grad_norm": 1.2662302255630493,
      "learning_rate": 5.6920000000000004e-05,
      "loss": 0.4194,
      "step": 11390
    },
    {
      "epoch": 0.6378653824616418,
      "grad_norm": 1.2910363674163818,
      "learning_rate": 5.692500000000001e-05,
      "loss": 0.4155,
      "step": 11391
    },
    {
      "epoch": 0.6379213797737708,
      "grad_norm": 1.5611884593963623,
      "learning_rate": 5.6930000000000006e-05,
      "loss": 0.4015,
      "step": 11392
    },
    {
      "epoch": 0.6379773770858999,
      "grad_norm": 1.6610698699951172,
      "learning_rate": 5.6935e-05,
      "loss": 0.4132,
      "step": 11393
    },
    {
      "epoch": 0.6380333743980289,
      "grad_norm": 1.3034590482711792,
      "learning_rate": 5.694000000000001e-05,
      "loss": 0.4829,
      "step": 11394
    },
    {
      "epoch": 0.6380893717101579,
      "grad_norm": 1.194251537322998,
      "learning_rate": 5.6945000000000005e-05,
      "loss": 0.3595,
      "step": 11395
    },
    {
      "epoch": 0.6381453690222869,
      "grad_norm": 1.3438389301300049,
      "learning_rate": 5.695e-05,
      "loss": 0.4172,
      "step": 11396
    },
    {
      "epoch": 0.6382013663344159,
      "grad_norm": 1.3963749408721924,
      "learning_rate": 5.6955000000000006e-05,
      "loss": 0.6238,
      "step": 11397
    },
    {
      "epoch": 0.638257363646545,
      "grad_norm": 1.1640512943267822,
      "learning_rate": 5.6960000000000004e-05,
      "loss": 0.4674,
      "step": 11398
    },
    {
      "epoch": 0.638313360958674,
      "grad_norm": 1.4330171346664429,
      "learning_rate": 5.6965e-05,
      "loss": 0.3955,
      "step": 11399
    },
    {
      "epoch": 0.638369358270803,
      "grad_norm": 1.1675355434417725,
      "learning_rate": 5.697e-05,
      "loss": 0.3712,
      "step": 11400
    },
    {
      "epoch": 0.638425355582932,
      "grad_norm": 1.525566577911377,
      "learning_rate": 5.6975e-05,
      "loss": 0.4081,
      "step": 11401
    },
    {
      "epoch": 0.638481352895061,
      "grad_norm": 1.232356309890747,
      "learning_rate": 5.698e-05,
      "loss": 0.4272,
      "step": 11402
    },
    {
      "epoch": 0.63853735020719,
      "grad_norm": 1.3237557411193848,
      "learning_rate": 5.6985e-05,
      "loss": 0.4409,
      "step": 11403
    },
    {
      "epoch": 0.6385933475193191,
      "grad_norm": 1.2331622838974,
      "learning_rate": 5.699e-05,
      "loss": 0.3428,
      "step": 11404
    },
    {
      "epoch": 0.6386493448314481,
      "grad_norm": 1.3364495038986206,
      "learning_rate": 5.6995e-05,
      "loss": 0.3592,
      "step": 11405
    },
    {
      "epoch": 0.6387053421435771,
      "grad_norm": 1.2224996089935303,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 0.462,
      "step": 11406
    },
    {
      "epoch": 0.6387613394557061,
      "grad_norm": 1.1252217292785645,
      "learning_rate": 5.7005e-05,
      "loss": 0.33,
      "step": 11407
    },
    {
      "epoch": 0.6388173367678351,
      "grad_norm": 1.2200932502746582,
      "learning_rate": 5.7010000000000005e-05,
      "loss": 0.3552,
      "step": 11408
    },
    {
      "epoch": 0.6388733340799642,
      "grad_norm": 1.2461763620376587,
      "learning_rate": 5.701500000000001e-05,
      "loss": 0.3404,
      "step": 11409
    },
    {
      "epoch": 0.6389293313920932,
      "grad_norm": 1.7498670816421509,
      "learning_rate": 5.7020000000000006e-05,
      "loss": 0.576,
      "step": 11410
    },
    {
      "epoch": 0.6389853287042222,
      "grad_norm": 1.3398674726486206,
      "learning_rate": 5.7025000000000004e-05,
      "loss": 0.4954,
      "step": 11411
    },
    {
      "epoch": 0.6390413260163512,
      "grad_norm": 1.282238245010376,
      "learning_rate": 5.703000000000001e-05,
      "loss": 0.4809,
      "step": 11412
    },
    {
      "epoch": 0.6390973233284802,
      "grad_norm": 1.2816829681396484,
      "learning_rate": 5.7035000000000005e-05,
      "loss": 0.3704,
      "step": 11413
    },
    {
      "epoch": 0.6391533206406093,
      "grad_norm": 1.433122992515564,
      "learning_rate": 5.704e-05,
      "loss": 0.5439,
      "step": 11414
    },
    {
      "epoch": 0.6392093179527383,
      "grad_norm": 3.01277494430542,
      "learning_rate": 5.704500000000001e-05,
      "loss": 0.4214,
      "step": 11415
    },
    {
      "epoch": 0.6392653152648673,
      "grad_norm": 1.63400137424469,
      "learning_rate": 5.7050000000000004e-05,
      "loss": 0.3825,
      "step": 11416
    },
    {
      "epoch": 0.6393213125769963,
      "grad_norm": 1.2651917934417725,
      "learning_rate": 5.7055e-05,
      "loss": 0.4381,
      "step": 11417
    },
    {
      "epoch": 0.6393773098891253,
      "grad_norm": 1.2086280584335327,
      "learning_rate": 5.706e-05,
      "loss": 0.4712,
      "step": 11418
    },
    {
      "epoch": 0.6394333072012544,
      "grad_norm": 1.4258328676223755,
      "learning_rate": 5.7065e-05,
      "loss": 0.4264,
      "step": 11419
    },
    {
      "epoch": 0.6394893045133834,
      "grad_norm": 1.3295502662658691,
      "learning_rate": 5.707e-05,
      "loss": 0.4893,
      "step": 11420
    },
    {
      "epoch": 0.6395453018255124,
      "grad_norm": 1.505089521408081,
      "learning_rate": 5.7075e-05,
      "loss": 0.54,
      "step": 11421
    },
    {
      "epoch": 0.6396012991376414,
      "grad_norm": 1.2171998023986816,
      "learning_rate": 5.708e-05,
      "loss": 0.5263,
      "step": 11422
    },
    {
      "epoch": 0.6396572964497704,
      "grad_norm": 1.2634024620056152,
      "learning_rate": 5.7085e-05,
      "loss": 0.5035,
      "step": 11423
    },
    {
      "epoch": 0.6397132937618994,
      "grad_norm": 1.3135502338409424,
      "learning_rate": 5.709e-05,
      "loss": 0.3592,
      "step": 11424
    },
    {
      "epoch": 0.6397692910740285,
      "grad_norm": 1.3976722955703735,
      "learning_rate": 5.7095e-05,
      "loss": 0.5523,
      "step": 11425
    },
    {
      "epoch": 0.6398252883861575,
      "grad_norm": 1.4321637153625488,
      "learning_rate": 5.71e-05,
      "loss": 0.4316,
      "step": 11426
    },
    {
      "epoch": 0.6398812856982865,
      "grad_norm": 1.2029895782470703,
      "learning_rate": 5.7104999999999996e-05,
      "loss": 0.3667,
      "step": 11427
    },
    {
      "epoch": 0.6399372830104155,
      "grad_norm": 1.2664986848831177,
      "learning_rate": 5.711000000000001e-05,
      "loss": 0.3627,
      "step": 11428
    },
    {
      "epoch": 0.6399932803225445,
      "grad_norm": 1.3680680990219116,
      "learning_rate": 5.7115000000000005e-05,
      "loss": 0.3631,
      "step": 11429
    },
    {
      "epoch": 0.6400492776346736,
      "grad_norm": 1.1282374858856201,
      "learning_rate": 5.712000000000001e-05,
      "loss": 0.399,
      "step": 11430
    },
    {
      "epoch": 0.6401052749468026,
      "grad_norm": 1.3647115230560303,
      "learning_rate": 5.7125000000000006e-05,
      "loss": 0.5315,
      "step": 11431
    },
    {
      "epoch": 0.6401612722589316,
      "grad_norm": 1.1750375032424927,
      "learning_rate": 5.7130000000000004e-05,
      "loss": 0.3252,
      "step": 11432
    },
    {
      "epoch": 0.6402172695710606,
      "grad_norm": 1.1515816450119019,
      "learning_rate": 5.713500000000001e-05,
      "loss": 0.3207,
      "step": 11433
    },
    {
      "epoch": 0.6402732668831896,
      "grad_norm": 1.2982523441314697,
      "learning_rate": 5.7140000000000005e-05,
      "loss": 0.3753,
      "step": 11434
    },
    {
      "epoch": 0.6403292641953187,
      "grad_norm": 1.3154433965682983,
      "learning_rate": 5.7145e-05,
      "loss": 0.3783,
      "step": 11435
    },
    {
      "epoch": 0.6403852615074477,
      "grad_norm": 1.3863359689712524,
      "learning_rate": 5.715000000000001e-05,
      "loss": 0.4329,
      "step": 11436
    },
    {
      "epoch": 0.6404412588195767,
      "grad_norm": 1.5635638236999512,
      "learning_rate": 5.7155000000000004e-05,
      "loss": 0.3972,
      "step": 11437
    },
    {
      "epoch": 0.6404972561317057,
      "grad_norm": 1.3954685926437378,
      "learning_rate": 5.716e-05,
      "loss": 0.4937,
      "step": 11438
    },
    {
      "epoch": 0.6405532534438347,
      "grad_norm": 1.2547791004180908,
      "learning_rate": 5.7165e-05,
      "loss": 0.4415,
      "step": 11439
    },
    {
      "epoch": 0.6406092507559638,
      "grad_norm": 1.4040956497192383,
      "learning_rate": 5.717e-05,
      "loss": 0.4813,
      "step": 11440
    },
    {
      "epoch": 0.6406652480680928,
      "grad_norm": 1.5399640798568726,
      "learning_rate": 5.7175e-05,
      "loss": 0.3528,
      "step": 11441
    },
    {
      "epoch": 0.6407212453802218,
      "grad_norm": 1.1662284135818481,
      "learning_rate": 5.718e-05,
      "loss": 0.3048,
      "step": 11442
    },
    {
      "epoch": 0.6407772426923508,
      "grad_norm": 1.4365795850753784,
      "learning_rate": 5.7185e-05,
      "loss": 0.5575,
      "step": 11443
    },
    {
      "epoch": 0.6408332400044798,
      "grad_norm": 1.2736120223999023,
      "learning_rate": 5.719e-05,
      "loss": 0.4323,
      "step": 11444
    },
    {
      "epoch": 0.6408892373166089,
      "grad_norm": 1.616166591644287,
      "learning_rate": 5.7195e-05,
      "loss": 0.4377,
      "step": 11445
    },
    {
      "epoch": 0.6409452346287379,
      "grad_norm": 1.6133275032043457,
      "learning_rate": 5.72e-05,
      "loss": 0.469,
      "step": 11446
    },
    {
      "epoch": 0.6410012319408669,
      "grad_norm": 1.2576003074645996,
      "learning_rate": 5.7205e-05,
      "loss": 0.5083,
      "step": 11447
    },
    {
      "epoch": 0.6410572292529959,
      "grad_norm": 1.3431735038757324,
      "learning_rate": 5.721000000000001e-05,
      "loss": 0.4122,
      "step": 11448
    },
    {
      "epoch": 0.6411132265651249,
      "grad_norm": 1.3998430967330933,
      "learning_rate": 5.721500000000001e-05,
      "loss": 0.563,
      "step": 11449
    },
    {
      "epoch": 0.641169223877254,
      "grad_norm": 1.4633538722991943,
      "learning_rate": 5.7220000000000004e-05,
      "loss": 0.4276,
      "step": 11450
    },
    {
      "epoch": 0.6412252211893829,
      "grad_norm": 1.1596487760543823,
      "learning_rate": 5.722500000000001e-05,
      "loss": 0.3429,
      "step": 11451
    },
    {
      "epoch": 0.6412812185015119,
      "grad_norm": 1.3280785083770752,
      "learning_rate": 5.7230000000000006e-05,
      "loss": 0.4165,
      "step": 11452
    },
    {
      "epoch": 0.6413372158136409,
      "grad_norm": 1.7387182712554932,
      "learning_rate": 5.7235000000000003e-05,
      "loss": 0.3599,
      "step": 11453
    },
    {
      "epoch": 0.6413932131257699,
      "grad_norm": 1.4629881381988525,
      "learning_rate": 5.724000000000001e-05,
      "loss": 0.3769,
      "step": 11454
    },
    {
      "epoch": 0.6414492104378989,
      "grad_norm": 1.3334035873413086,
      "learning_rate": 5.7245000000000005e-05,
      "loss": 0.335,
      "step": 11455
    },
    {
      "epoch": 0.641505207750028,
      "grad_norm": 1.2458865642547607,
      "learning_rate": 5.725e-05,
      "loss": 0.43,
      "step": 11456
    },
    {
      "epoch": 0.641561205062157,
      "grad_norm": 1.3170214891433716,
      "learning_rate": 5.7255e-05,
      "loss": 0.4686,
      "step": 11457
    },
    {
      "epoch": 0.641617202374286,
      "grad_norm": 1.4576579332351685,
      "learning_rate": 5.7260000000000004e-05,
      "loss": 0.4614,
      "step": 11458
    },
    {
      "epoch": 0.641673199686415,
      "grad_norm": 1.4827557802200317,
      "learning_rate": 5.7265e-05,
      "loss": 0.5667,
      "step": 11459
    },
    {
      "epoch": 0.641729196998544,
      "grad_norm": 1.1222044229507446,
      "learning_rate": 5.727e-05,
      "loss": 0.507,
      "step": 11460
    },
    {
      "epoch": 0.641785194310673,
      "grad_norm": 1.186562418937683,
      "learning_rate": 5.7275e-05,
      "loss": 0.3673,
      "step": 11461
    },
    {
      "epoch": 0.6418411916228021,
      "grad_norm": 1.4117891788482666,
      "learning_rate": 5.728e-05,
      "loss": 0.4402,
      "step": 11462
    },
    {
      "epoch": 0.6418971889349311,
      "grad_norm": 1.2655959129333496,
      "learning_rate": 5.7285e-05,
      "loss": 0.347,
      "step": 11463
    },
    {
      "epoch": 0.6419531862470601,
      "grad_norm": 1.325129508972168,
      "learning_rate": 5.729e-05,
      "loss": 0.5245,
      "step": 11464
    },
    {
      "epoch": 0.6420091835591891,
      "grad_norm": 1.2996641397476196,
      "learning_rate": 5.7295e-05,
      "loss": 0.4196,
      "step": 11465
    },
    {
      "epoch": 0.6420651808713181,
      "grad_norm": 1.9610968828201294,
      "learning_rate": 5.73e-05,
      "loss": 0.4092,
      "step": 11466
    },
    {
      "epoch": 0.6421211781834472,
      "grad_norm": 1.3521019220352173,
      "learning_rate": 5.7304999999999994e-05,
      "loss": 0.3975,
      "step": 11467
    },
    {
      "epoch": 0.6421771754955762,
      "grad_norm": 1.2128995656967163,
      "learning_rate": 5.7310000000000005e-05,
      "loss": 0.3544,
      "step": 11468
    },
    {
      "epoch": 0.6422331728077052,
      "grad_norm": 1.2328879833221436,
      "learning_rate": 5.731500000000001e-05,
      "loss": 0.595,
      "step": 11469
    },
    {
      "epoch": 0.6422891701198342,
      "grad_norm": 1.2000852823257446,
      "learning_rate": 5.732000000000001e-05,
      "loss": 0.325,
      "step": 11470
    },
    {
      "epoch": 0.6423451674319632,
      "grad_norm": 1.6674840450286865,
      "learning_rate": 5.7325000000000004e-05,
      "loss": 0.3719,
      "step": 11471
    },
    {
      "epoch": 0.6424011647440923,
      "grad_norm": 1.1440792083740234,
      "learning_rate": 5.733000000000001e-05,
      "loss": 0.3789,
      "step": 11472
    },
    {
      "epoch": 0.6424571620562213,
      "grad_norm": 1.2247045040130615,
      "learning_rate": 5.7335000000000006e-05,
      "loss": 0.428,
      "step": 11473
    },
    {
      "epoch": 0.6425131593683503,
      "grad_norm": 1.2148150205612183,
      "learning_rate": 5.734e-05,
      "loss": 0.4804,
      "step": 11474
    },
    {
      "epoch": 0.6425691566804793,
      "grad_norm": 1.5058478116989136,
      "learning_rate": 5.734500000000001e-05,
      "loss": 0.4637,
      "step": 11475
    },
    {
      "epoch": 0.6426251539926083,
      "grad_norm": 1.2872493267059326,
      "learning_rate": 5.7350000000000005e-05,
      "loss": 0.3831,
      "step": 11476
    },
    {
      "epoch": 0.6426811513047374,
      "grad_norm": 1.0541887283325195,
      "learning_rate": 5.7355e-05,
      "loss": 0.428,
      "step": 11477
    },
    {
      "epoch": 0.6427371486168664,
      "grad_norm": 1.1574785709381104,
      "learning_rate": 5.736e-05,
      "loss": 0.313,
      "step": 11478
    },
    {
      "epoch": 0.6427931459289954,
      "grad_norm": 1.2846903800964355,
      "learning_rate": 5.7365000000000004e-05,
      "loss": 0.4837,
      "step": 11479
    },
    {
      "epoch": 0.6428491432411244,
      "grad_norm": 1.373149037361145,
      "learning_rate": 5.737e-05,
      "loss": 0.4994,
      "step": 11480
    },
    {
      "epoch": 0.6429051405532534,
      "grad_norm": 1.2504630088806152,
      "learning_rate": 5.7375e-05,
      "loss": 0.4227,
      "step": 11481
    },
    {
      "epoch": 0.6429611378653824,
      "grad_norm": 1.0435395240783691,
      "learning_rate": 5.738e-05,
      "loss": 0.3611,
      "step": 11482
    },
    {
      "epoch": 0.6430171351775115,
      "grad_norm": 1.393623948097229,
      "learning_rate": 5.7385e-05,
      "loss": 0.4542,
      "step": 11483
    },
    {
      "epoch": 0.6430731324896405,
      "grad_norm": 1.033274531364441,
      "learning_rate": 5.739e-05,
      "loss": 0.2868,
      "step": 11484
    },
    {
      "epoch": 0.6431291298017695,
      "grad_norm": 1.6692862510681152,
      "learning_rate": 5.7395e-05,
      "loss": 0.5322,
      "step": 11485
    },
    {
      "epoch": 0.6431851271138985,
      "grad_norm": 1.2050962448120117,
      "learning_rate": 5.74e-05,
      "loss": 0.3991,
      "step": 11486
    },
    {
      "epoch": 0.6432411244260275,
      "grad_norm": 1.5836702585220337,
      "learning_rate": 5.7405e-05,
      "loss": 0.3882,
      "step": 11487
    },
    {
      "epoch": 0.6432971217381566,
      "grad_norm": 1.2420437335968018,
      "learning_rate": 5.741000000000001e-05,
      "loss": 0.3548,
      "step": 11488
    },
    {
      "epoch": 0.6433531190502856,
      "grad_norm": 1.2407389879226685,
      "learning_rate": 5.7415000000000005e-05,
      "loss": 0.4549,
      "step": 11489
    },
    {
      "epoch": 0.6434091163624146,
      "grad_norm": 1.330662488937378,
      "learning_rate": 5.742000000000001e-05,
      "loss": 0.5312,
      "step": 11490
    },
    {
      "epoch": 0.6434651136745436,
      "grad_norm": 1.766160249710083,
      "learning_rate": 5.742500000000001e-05,
      "loss": 0.5731,
      "step": 11491
    },
    {
      "epoch": 0.6435211109866726,
      "grad_norm": 1.2884317636489868,
      "learning_rate": 5.7430000000000004e-05,
      "loss": 0.5438,
      "step": 11492
    },
    {
      "epoch": 0.6435771082988017,
      "grad_norm": 1.593859314918518,
      "learning_rate": 5.743500000000001e-05,
      "loss": 0.4792,
      "step": 11493
    },
    {
      "epoch": 0.6436331056109307,
      "grad_norm": 1.1537383794784546,
      "learning_rate": 5.7440000000000006e-05,
      "loss": 0.3695,
      "step": 11494
    },
    {
      "epoch": 0.6436891029230597,
      "grad_norm": 1.2668381929397583,
      "learning_rate": 5.7445e-05,
      "loss": 0.3173,
      "step": 11495
    },
    {
      "epoch": 0.6437451002351887,
      "grad_norm": 1.3317357301712036,
      "learning_rate": 5.745e-05,
      "loss": 0.6053,
      "step": 11496
    },
    {
      "epoch": 0.6438010975473177,
      "grad_norm": 1.1135140657424927,
      "learning_rate": 5.7455000000000005e-05,
      "loss": 0.3324,
      "step": 11497
    },
    {
      "epoch": 0.6438570948594468,
      "grad_norm": 1.2193281650543213,
      "learning_rate": 5.746e-05,
      "loss": 0.378,
      "step": 11498
    },
    {
      "epoch": 0.6439130921715758,
      "grad_norm": 1.1844638586044312,
      "learning_rate": 5.7465e-05,
      "loss": 0.488,
      "step": 11499
    },
    {
      "epoch": 0.6439690894837048,
      "grad_norm": 1.4308418035507202,
      "learning_rate": 5.7470000000000004e-05,
      "loss": 0.5116,
      "step": 11500
    },
    {
      "epoch": 0.6440250867958338,
      "grad_norm": 1.4272637367248535,
      "learning_rate": 5.7475e-05,
      "loss": 0.4804,
      "step": 11501
    },
    {
      "epoch": 0.6440810841079628,
      "grad_norm": 1.3765759468078613,
      "learning_rate": 5.748e-05,
      "loss": 0.4766,
      "step": 11502
    },
    {
      "epoch": 0.6441370814200919,
      "grad_norm": 1.304327368736267,
      "learning_rate": 5.7485e-05,
      "loss": 0.5071,
      "step": 11503
    },
    {
      "epoch": 0.6441930787322209,
      "grad_norm": 1.2469215393066406,
      "learning_rate": 5.749e-05,
      "loss": 0.4449,
      "step": 11504
    },
    {
      "epoch": 0.6442490760443499,
      "grad_norm": 1.2005549669265747,
      "learning_rate": 5.7495e-05,
      "loss": 0.454,
      "step": 11505
    },
    {
      "epoch": 0.6443050733564789,
      "grad_norm": 1.0989803075790405,
      "learning_rate": 5.7499999999999995e-05,
      "loss": 0.4284,
      "step": 11506
    },
    {
      "epoch": 0.6443610706686079,
      "grad_norm": 1.1446917057037354,
      "learning_rate": 5.7505e-05,
      "loss": 0.4705,
      "step": 11507
    },
    {
      "epoch": 0.644417067980737,
      "grad_norm": 1.1719930171966553,
      "learning_rate": 5.7509999999999997e-05,
      "loss": 0.3878,
      "step": 11508
    },
    {
      "epoch": 0.644473065292866,
      "grad_norm": 1.02593195438385,
      "learning_rate": 5.751500000000001e-05,
      "loss": 0.2829,
      "step": 11509
    },
    {
      "epoch": 0.644529062604995,
      "grad_norm": 1.3297885656356812,
      "learning_rate": 5.7520000000000005e-05,
      "loss": 0.611,
      "step": 11510
    },
    {
      "epoch": 0.644585059917124,
      "grad_norm": 1.445859670639038,
      "learning_rate": 5.752500000000001e-05,
      "loss": 0.4839,
      "step": 11511
    },
    {
      "epoch": 0.644641057229253,
      "grad_norm": 1.259381890296936,
      "learning_rate": 5.7530000000000007e-05,
      "loss": 0.4401,
      "step": 11512
    },
    {
      "epoch": 0.644697054541382,
      "grad_norm": 1.2717972993850708,
      "learning_rate": 5.7535000000000004e-05,
      "loss": 0.4225,
      "step": 11513
    },
    {
      "epoch": 0.6447530518535111,
      "grad_norm": 1.3548849821090698,
      "learning_rate": 5.754000000000001e-05,
      "loss": 0.3936,
      "step": 11514
    },
    {
      "epoch": 0.6448090491656401,
      "grad_norm": 1.3008915185928345,
      "learning_rate": 5.7545000000000006e-05,
      "loss": 0.4837,
      "step": 11515
    },
    {
      "epoch": 0.6448650464777691,
      "grad_norm": 1.3620336055755615,
      "learning_rate": 5.755e-05,
      "loss": 0.8122,
      "step": 11516
    },
    {
      "epoch": 0.6449210437898981,
      "grad_norm": 1.2951780557632446,
      "learning_rate": 5.7555e-05,
      "loss": 0.4222,
      "step": 11517
    },
    {
      "epoch": 0.6449770411020271,
      "grad_norm": 1.435787558555603,
      "learning_rate": 5.7560000000000005e-05,
      "loss": 0.4281,
      "step": 11518
    },
    {
      "epoch": 0.6450330384141562,
      "grad_norm": 1.430140495300293,
      "learning_rate": 5.7565e-05,
      "loss": 0.3756,
      "step": 11519
    },
    {
      "epoch": 0.6450890357262852,
      "grad_norm": 1.2110021114349365,
      "learning_rate": 5.757e-05,
      "loss": 0.4653,
      "step": 11520
    },
    {
      "epoch": 0.6451450330384142,
      "grad_norm": 2.7886364459991455,
      "learning_rate": 5.7575000000000004e-05,
      "loss": 0.4481,
      "step": 11521
    },
    {
      "epoch": 0.6452010303505432,
      "grad_norm": 1.1463032960891724,
      "learning_rate": 5.758e-05,
      "loss": 0.3699,
      "step": 11522
    },
    {
      "epoch": 0.6452570276626722,
      "grad_norm": 1.4110069274902344,
      "learning_rate": 5.7585e-05,
      "loss": 0.4299,
      "step": 11523
    },
    {
      "epoch": 0.6453130249748013,
      "grad_norm": 1.5350940227508545,
      "learning_rate": 5.759e-05,
      "loss": 0.4105,
      "step": 11524
    },
    {
      "epoch": 0.6453690222869303,
      "grad_norm": 1.5140190124511719,
      "learning_rate": 5.7595e-05,
      "loss": 0.7324,
      "step": 11525
    },
    {
      "epoch": 0.6454250195990593,
      "grad_norm": 1.4884791374206543,
      "learning_rate": 5.76e-05,
      "loss": 0.4558,
      "step": 11526
    },
    {
      "epoch": 0.6454810169111883,
      "grad_norm": 1.242226243019104,
      "learning_rate": 5.7604999999999995e-05,
      "loss": 0.5391,
      "step": 11527
    },
    {
      "epoch": 0.6455370142233173,
      "grad_norm": 1.2266403436660767,
      "learning_rate": 5.761e-05,
      "loss": 0.4052,
      "step": 11528
    },
    {
      "epoch": 0.6455930115354463,
      "grad_norm": 1.0919597148895264,
      "learning_rate": 5.761500000000001e-05,
      "loss": 0.3521,
      "step": 11529
    },
    {
      "epoch": 0.6456490088475754,
      "grad_norm": 1.1650614738464355,
      "learning_rate": 5.762000000000001e-05,
      "loss": 0.5479,
      "step": 11530
    },
    {
      "epoch": 0.6457050061597044,
      "grad_norm": 1.1579889059066772,
      "learning_rate": 5.7625000000000005e-05,
      "loss": 0.4455,
      "step": 11531
    },
    {
      "epoch": 0.6457610034718334,
      "grad_norm": 1.603376865386963,
      "learning_rate": 5.763000000000001e-05,
      "loss": 0.4191,
      "step": 11532
    },
    {
      "epoch": 0.6458170007839624,
      "grad_norm": 1.2230231761932373,
      "learning_rate": 5.7635000000000006e-05,
      "loss": 0.4944,
      "step": 11533
    },
    {
      "epoch": 0.6458729980960913,
      "grad_norm": 1.1598926782608032,
      "learning_rate": 5.7640000000000004e-05,
      "loss": 0.374,
      "step": 11534
    },
    {
      "epoch": 0.6459289954082204,
      "grad_norm": 2.9758739471435547,
      "learning_rate": 5.7645e-05,
      "loss": 0.4522,
      "step": 11535
    },
    {
      "epoch": 0.6459849927203494,
      "grad_norm": 1.1005445718765259,
      "learning_rate": 5.7650000000000005e-05,
      "loss": 0.3054,
      "step": 11536
    },
    {
      "epoch": 0.6460409900324784,
      "grad_norm": 1.3561359643936157,
      "learning_rate": 5.7655e-05,
      "loss": 0.4538,
      "step": 11537
    },
    {
      "epoch": 0.6460969873446074,
      "grad_norm": 1.477579951286316,
      "learning_rate": 5.766e-05,
      "loss": 0.3823,
      "step": 11538
    },
    {
      "epoch": 0.6461529846567364,
      "grad_norm": 1.068372130393982,
      "learning_rate": 5.7665000000000004e-05,
      "loss": 0.2992,
      "step": 11539
    },
    {
      "epoch": 0.6462089819688654,
      "grad_norm": 1.4765825271606445,
      "learning_rate": 5.767e-05,
      "loss": 0.5325,
      "step": 11540
    },
    {
      "epoch": 0.6462649792809945,
      "grad_norm": 1.3367769718170166,
      "learning_rate": 5.7675e-05,
      "loss": 0.3413,
      "step": 11541
    },
    {
      "epoch": 0.6463209765931235,
      "grad_norm": 1.3945248126983643,
      "learning_rate": 5.7680000000000003e-05,
      "loss": 0.4743,
      "step": 11542
    },
    {
      "epoch": 0.6463769739052525,
      "grad_norm": 1.352189540863037,
      "learning_rate": 5.7685e-05,
      "loss": 0.5408,
      "step": 11543
    },
    {
      "epoch": 0.6464329712173815,
      "grad_norm": 1.473279595375061,
      "learning_rate": 5.769e-05,
      "loss": 0.5688,
      "step": 11544
    },
    {
      "epoch": 0.6464889685295105,
      "grad_norm": 1.2186814546585083,
      "learning_rate": 5.7695e-05,
      "loss": 0.3343,
      "step": 11545
    },
    {
      "epoch": 0.6465449658416396,
      "grad_norm": 1.340122103691101,
      "learning_rate": 5.77e-05,
      "loss": 0.4962,
      "step": 11546
    },
    {
      "epoch": 0.6466009631537686,
      "grad_norm": 1.6914175748825073,
      "learning_rate": 5.7705e-05,
      "loss": 0.5216,
      "step": 11547
    },
    {
      "epoch": 0.6466569604658976,
      "grad_norm": 1.2420464754104614,
      "learning_rate": 5.7709999999999995e-05,
      "loss": 0.4441,
      "step": 11548
    },
    {
      "epoch": 0.6467129577780266,
      "grad_norm": 1.2060019969940186,
      "learning_rate": 5.7715000000000006e-05,
      "loss": 0.3787,
      "step": 11549
    },
    {
      "epoch": 0.6467689550901556,
      "grad_norm": 1.010940432548523,
      "learning_rate": 5.772000000000001e-05,
      "loss": 0.3713,
      "step": 11550
    },
    {
      "epoch": 0.6468249524022847,
      "grad_norm": 1.2166849374771118,
      "learning_rate": 5.772500000000001e-05,
      "loss": 0.3891,
      "step": 11551
    },
    {
      "epoch": 0.6468809497144137,
      "grad_norm": 1.2105660438537598,
      "learning_rate": 5.7730000000000005e-05,
      "loss": 0.2847,
      "step": 11552
    },
    {
      "epoch": 0.6469369470265427,
      "grad_norm": 1.383593201637268,
      "learning_rate": 5.773500000000001e-05,
      "loss": 0.5505,
      "step": 11553
    },
    {
      "epoch": 0.6469929443386717,
      "grad_norm": 1.187862515449524,
      "learning_rate": 5.7740000000000006e-05,
      "loss": 0.3641,
      "step": 11554
    },
    {
      "epoch": 0.6470489416508007,
      "grad_norm": 1.1482435464859009,
      "learning_rate": 5.7745000000000004e-05,
      "loss": 0.5302,
      "step": 11555
    },
    {
      "epoch": 0.6471049389629298,
      "grad_norm": 1.2438193559646606,
      "learning_rate": 5.775e-05,
      "loss": 0.4347,
      "step": 11556
    },
    {
      "epoch": 0.6471609362750588,
      "grad_norm": 1.4243003129959106,
      "learning_rate": 5.7755000000000005e-05,
      "loss": 0.4023,
      "step": 11557
    },
    {
      "epoch": 0.6472169335871878,
      "grad_norm": 1.2398462295532227,
      "learning_rate": 5.776e-05,
      "loss": 0.4243,
      "step": 11558
    },
    {
      "epoch": 0.6472729308993168,
      "grad_norm": 1.0839929580688477,
      "learning_rate": 5.7765e-05,
      "loss": 0.2921,
      "step": 11559
    },
    {
      "epoch": 0.6473289282114458,
      "grad_norm": 1.2982513904571533,
      "learning_rate": 5.7770000000000004e-05,
      "loss": 0.3389,
      "step": 11560
    },
    {
      "epoch": 0.6473849255235749,
      "grad_norm": 1.5267226696014404,
      "learning_rate": 5.7775e-05,
      "loss": 0.4404,
      "step": 11561
    },
    {
      "epoch": 0.6474409228357039,
      "grad_norm": 1.4346492290496826,
      "learning_rate": 5.778e-05,
      "loss": 0.4076,
      "step": 11562
    },
    {
      "epoch": 0.6474969201478329,
      "grad_norm": 1.395491123199463,
      "learning_rate": 5.7785e-05,
      "loss": 0.4529,
      "step": 11563
    },
    {
      "epoch": 0.6475529174599619,
      "grad_norm": 1.6427137851715088,
      "learning_rate": 5.779e-05,
      "loss": 0.5369,
      "step": 11564
    },
    {
      "epoch": 0.6476089147720909,
      "grad_norm": 1.3966608047485352,
      "learning_rate": 5.7795e-05,
      "loss": 0.3765,
      "step": 11565
    },
    {
      "epoch": 0.64766491208422,
      "grad_norm": 1.2554833889007568,
      "learning_rate": 5.7799999999999995e-05,
      "loss": 0.5485,
      "step": 11566
    },
    {
      "epoch": 0.647720909396349,
      "grad_norm": 1.5577747821807861,
      "learning_rate": 5.7805e-05,
      "loss": 0.7601,
      "step": 11567
    },
    {
      "epoch": 0.647776906708478,
      "grad_norm": 1.2530993223190308,
      "learning_rate": 5.781e-05,
      "loss": 0.3431,
      "step": 11568
    },
    {
      "epoch": 0.647832904020607,
      "grad_norm": 1.155864953994751,
      "learning_rate": 5.781500000000001e-05,
      "loss": 0.468,
      "step": 11569
    },
    {
      "epoch": 0.647888901332736,
      "grad_norm": 1.0874488353729248,
      "learning_rate": 5.7820000000000005e-05,
      "loss": 0.4228,
      "step": 11570
    },
    {
      "epoch": 0.647944898644865,
      "grad_norm": 1.3322001695632935,
      "learning_rate": 5.782500000000001e-05,
      "loss": 0.5208,
      "step": 11571
    },
    {
      "epoch": 0.6480008959569941,
      "grad_norm": 1.3388378620147705,
      "learning_rate": 5.783000000000001e-05,
      "loss": 0.4283,
      "step": 11572
    },
    {
      "epoch": 0.6480568932691231,
      "grad_norm": 1.1866655349731445,
      "learning_rate": 5.7835000000000004e-05,
      "loss": 0.3104,
      "step": 11573
    },
    {
      "epoch": 0.6481128905812521,
      "grad_norm": 1.391587495803833,
      "learning_rate": 5.784000000000001e-05,
      "loss": 0.6146,
      "step": 11574
    },
    {
      "epoch": 0.6481688878933811,
      "grad_norm": 1.344538688659668,
      "learning_rate": 5.7845000000000006e-05,
      "loss": 0.3789,
      "step": 11575
    },
    {
      "epoch": 0.6482248852055101,
      "grad_norm": 1.2578099966049194,
      "learning_rate": 5.7850000000000003e-05,
      "loss": 0.5257,
      "step": 11576
    },
    {
      "epoch": 0.6482808825176392,
      "grad_norm": 1.385211706161499,
      "learning_rate": 5.7855e-05,
      "loss": 0.3098,
      "step": 11577
    },
    {
      "epoch": 0.6483368798297682,
      "grad_norm": 1.5441373586654663,
      "learning_rate": 5.7860000000000005e-05,
      "loss": 0.5056,
      "step": 11578
    },
    {
      "epoch": 0.6483928771418972,
      "grad_norm": 1.4079526662826538,
      "learning_rate": 5.7865e-05,
      "loss": 0.3969,
      "step": 11579
    },
    {
      "epoch": 0.6484488744540262,
      "grad_norm": 1.1507244110107422,
      "learning_rate": 5.787e-05,
      "loss": 0.4326,
      "step": 11580
    },
    {
      "epoch": 0.6485048717661552,
      "grad_norm": 1.1137008666992188,
      "learning_rate": 5.7875000000000004e-05,
      "loss": 0.4333,
      "step": 11581
    },
    {
      "epoch": 0.6485608690782843,
      "grad_norm": 1.256856083869934,
      "learning_rate": 5.788e-05,
      "loss": 0.495,
      "step": 11582
    },
    {
      "epoch": 0.6486168663904133,
      "grad_norm": 1.2716187238693237,
      "learning_rate": 5.7885e-05,
      "loss": 0.3649,
      "step": 11583
    },
    {
      "epoch": 0.6486728637025423,
      "grad_norm": 1.5647202730178833,
      "learning_rate": 5.789e-05,
      "loss": 0.5961,
      "step": 11584
    },
    {
      "epoch": 0.6487288610146713,
      "grad_norm": 1.144774317741394,
      "learning_rate": 5.7895e-05,
      "loss": 0.3296,
      "step": 11585
    },
    {
      "epoch": 0.6487848583268003,
      "grad_norm": 1.1031969785690308,
      "learning_rate": 5.79e-05,
      "loss": 0.3849,
      "step": 11586
    },
    {
      "epoch": 0.6488408556389293,
      "grad_norm": 1.2824819087982178,
      "learning_rate": 5.7904999999999995e-05,
      "loss": 0.3177,
      "step": 11587
    },
    {
      "epoch": 0.6488968529510584,
      "grad_norm": 1.555246353149414,
      "learning_rate": 5.791e-05,
      "loss": 0.4973,
      "step": 11588
    },
    {
      "epoch": 0.6489528502631874,
      "grad_norm": 1.6722393035888672,
      "learning_rate": 5.791500000000001e-05,
      "loss": 0.4526,
      "step": 11589
    },
    {
      "epoch": 0.6490088475753164,
      "grad_norm": 1.4631997346878052,
      "learning_rate": 5.792000000000001e-05,
      "loss": 0.4488,
      "step": 11590
    },
    {
      "epoch": 0.6490648448874454,
      "grad_norm": 1.3110997676849365,
      "learning_rate": 5.7925000000000005e-05,
      "loss": 0.434,
      "step": 11591
    },
    {
      "epoch": 0.6491208421995744,
      "grad_norm": 1.4031312465667725,
      "learning_rate": 5.793000000000001e-05,
      "loss": 0.3872,
      "step": 11592
    },
    {
      "epoch": 0.6491768395117035,
      "grad_norm": 2.0797858238220215,
      "learning_rate": 5.793500000000001e-05,
      "loss": 0.3397,
      "step": 11593
    },
    {
      "epoch": 0.6492328368238325,
      "grad_norm": 1.397301435470581,
      "learning_rate": 5.7940000000000004e-05,
      "loss": 0.4672,
      "step": 11594
    },
    {
      "epoch": 0.6492888341359615,
      "grad_norm": 1.1421433687210083,
      "learning_rate": 5.7945e-05,
      "loss": 0.3781,
      "step": 11595
    },
    {
      "epoch": 0.6493448314480905,
      "grad_norm": 1.278132438659668,
      "learning_rate": 5.7950000000000006e-05,
      "loss": 0.4038,
      "step": 11596
    },
    {
      "epoch": 0.6494008287602195,
      "grad_norm": 1.68009614944458,
      "learning_rate": 5.7955e-05,
      "loss": 0.433,
      "step": 11597
    },
    {
      "epoch": 0.6494568260723486,
      "grad_norm": 1.222383737564087,
      "learning_rate": 5.796e-05,
      "loss": 0.4574,
      "step": 11598
    },
    {
      "epoch": 0.6495128233844776,
      "grad_norm": 1.118330955505371,
      "learning_rate": 5.7965000000000005e-05,
      "loss": 0.3929,
      "step": 11599
    },
    {
      "epoch": 0.6495688206966066,
      "grad_norm": 1.2085684537887573,
      "learning_rate": 5.797e-05,
      "loss": 0.5143,
      "step": 11600
    },
    {
      "epoch": 0.6496248180087356,
      "grad_norm": 1.231821894645691,
      "learning_rate": 5.7975e-05,
      "loss": 0.4428,
      "step": 11601
    },
    {
      "epoch": 0.6496808153208646,
      "grad_norm": 1.723815679550171,
      "learning_rate": 5.7980000000000004e-05,
      "loss": 0.4408,
      "step": 11602
    },
    {
      "epoch": 0.6497368126329937,
      "grad_norm": 1.2709662914276123,
      "learning_rate": 5.7985e-05,
      "loss": 0.3881,
      "step": 11603
    },
    {
      "epoch": 0.6497928099451227,
      "grad_norm": 1.2329301834106445,
      "learning_rate": 5.799e-05,
      "loss": 0.4592,
      "step": 11604
    },
    {
      "epoch": 0.6498488072572517,
      "grad_norm": 1.4668264389038086,
      "learning_rate": 5.7994999999999996e-05,
      "loss": 0.5234,
      "step": 11605
    },
    {
      "epoch": 0.6499048045693807,
      "grad_norm": 1.1704156398773193,
      "learning_rate": 5.8e-05,
      "loss": 0.412,
      "step": 11606
    },
    {
      "epoch": 0.6499608018815097,
      "grad_norm": 1.2018678188323975,
      "learning_rate": 5.8005e-05,
      "loss": 0.4805,
      "step": 11607
    },
    {
      "epoch": 0.6500167991936388,
      "grad_norm": 1.0939933061599731,
      "learning_rate": 5.8009999999999995e-05,
      "loss": 0.4825,
      "step": 11608
    },
    {
      "epoch": 0.6500727965057678,
      "grad_norm": 1.0713413953781128,
      "learning_rate": 5.8015000000000006e-05,
      "loss": 0.3814,
      "step": 11609
    },
    {
      "epoch": 0.6501287938178968,
      "grad_norm": 1.1035113334655762,
      "learning_rate": 5.802000000000001e-05,
      "loss": 0.4044,
      "step": 11610
    },
    {
      "epoch": 0.6501847911300258,
      "grad_norm": 1.2724945545196533,
      "learning_rate": 5.802500000000001e-05,
      "loss": 0.4172,
      "step": 11611
    },
    {
      "epoch": 0.6502407884421548,
      "grad_norm": 15.259182929992676,
      "learning_rate": 5.8030000000000005e-05,
      "loss": 0.5308,
      "step": 11612
    },
    {
      "epoch": 0.6502967857542838,
      "grad_norm": 1.1987518072128296,
      "learning_rate": 5.803500000000001e-05,
      "loss": 0.5636,
      "step": 11613
    },
    {
      "epoch": 0.6503527830664129,
      "grad_norm": 1.400138258934021,
      "learning_rate": 5.804000000000001e-05,
      "loss": 0.5606,
      "step": 11614
    },
    {
      "epoch": 0.6504087803785419,
      "grad_norm": 1.3241719007492065,
      "learning_rate": 5.8045000000000004e-05,
      "loss": 0.4243,
      "step": 11615
    },
    {
      "epoch": 0.6504647776906708,
      "grad_norm": 1.204487919807434,
      "learning_rate": 5.805e-05,
      "loss": 0.4068,
      "step": 11616
    },
    {
      "epoch": 0.6505207750027998,
      "grad_norm": 1.2731162309646606,
      "learning_rate": 5.8055000000000006e-05,
      "loss": 0.404,
      "step": 11617
    },
    {
      "epoch": 0.6505767723149288,
      "grad_norm": 1.1496412754058838,
      "learning_rate": 5.806e-05,
      "loss": 0.413,
      "step": 11618
    },
    {
      "epoch": 0.6506327696270578,
      "grad_norm": 1.4973740577697754,
      "learning_rate": 5.8065e-05,
      "loss": 0.3583,
      "step": 11619
    },
    {
      "epoch": 0.6506887669391869,
      "grad_norm": 1.2994894981384277,
      "learning_rate": 5.8070000000000005e-05,
      "loss": 0.3506,
      "step": 11620
    },
    {
      "epoch": 0.6507447642513159,
      "grad_norm": 1.2027066946029663,
      "learning_rate": 5.8075e-05,
      "loss": 0.4193,
      "step": 11621
    },
    {
      "epoch": 0.6508007615634449,
      "grad_norm": 1.14454984664917,
      "learning_rate": 5.808e-05,
      "loss": 0.4292,
      "step": 11622
    },
    {
      "epoch": 0.6508567588755739,
      "grad_norm": 1.2749015092849731,
      "learning_rate": 5.8085000000000004e-05,
      "loss": 0.3946,
      "step": 11623
    },
    {
      "epoch": 0.650912756187703,
      "grad_norm": 1.0972893238067627,
      "learning_rate": 5.809e-05,
      "loss": 0.3939,
      "step": 11624
    },
    {
      "epoch": 0.650968753499832,
      "grad_norm": 1.0439338684082031,
      "learning_rate": 5.8095e-05,
      "loss": 0.4078,
      "step": 11625
    },
    {
      "epoch": 0.651024750811961,
      "grad_norm": 1.4483518600463867,
      "learning_rate": 5.8099999999999996e-05,
      "loss": 0.3705,
      "step": 11626
    },
    {
      "epoch": 0.65108074812409,
      "grad_norm": 1.3294535875320435,
      "learning_rate": 5.8105e-05,
      "loss": 0.3375,
      "step": 11627
    },
    {
      "epoch": 0.651136745436219,
      "grad_norm": 1.417929768562317,
      "learning_rate": 5.811e-05,
      "loss": 0.5627,
      "step": 11628
    },
    {
      "epoch": 0.651192742748348,
      "grad_norm": 1.3771790266036987,
      "learning_rate": 5.811500000000001e-05,
      "loss": 0.7373,
      "step": 11629
    },
    {
      "epoch": 0.6512487400604771,
      "grad_norm": 1.1689010858535767,
      "learning_rate": 5.8120000000000006e-05,
      "loss": 0.3993,
      "step": 11630
    },
    {
      "epoch": 0.6513047373726061,
      "grad_norm": 1.2407886981964111,
      "learning_rate": 5.812500000000001e-05,
      "loss": 0.4528,
      "step": 11631
    },
    {
      "epoch": 0.6513607346847351,
      "grad_norm": 1.2653611898422241,
      "learning_rate": 5.813000000000001e-05,
      "loss": 0.3802,
      "step": 11632
    },
    {
      "epoch": 0.6514167319968641,
      "grad_norm": 1.2242566347122192,
      "learning_rate": 5.8135000000000005e-05,
      "loss": 0.4715,
      "step": 11633
    },
    {
      "epoch": 0.6514727293089931,
      "grad_norm": 1.299102783203125,
      "learning_rate": 5.814e-05,
      "loss": 0.4909,
      "step": 11634
    },
    {
      "epoch": 0.6515287266211222,
      "grad_norm": 1.1339411735534668,
      "learning_rate": 5.8145000000000007e-05,
      "loss": 0.4823,
      "step": 11635
    },
    {
      "epoch": 0.6515847239332512,
      "grad_norm": 1.212303638458252,
      "learning_rate": 5.8150000000000004e-05,
      "loss": 0.4722,
      "step": 11636
    },
    {
      "epoch": 0.6516407212453802,
      "grad_norm": 1.3442906141281128,
      "learning_rate": 5.8155e-05,
      "loss": 0.4407,
      "step": 11637
    },
    {
      "epoch": 0.6516967185575092,
      "grad_norm": 5.252134323120117,
      "learning_rate": 5.8160000000000006e-05,
      "loss": 0.4684,
      "step": 11638
    },
    {
      "epoch": 0.6517527158696382,
      "grad_norm": 1.1816720962524414,
      "learning_rate": 5.8165e-05,
      "loss": 0.375,
      "step": 11639
    },
    {
      "epoch": 0.6518087131817673,
      "grad_norm": 1.3227847814559937,
      "learning_rate": 5.817e-05,
      "loss": 0.6434,
      "step": 11640
    },
    {
      "epoch": 0.6518647104938963,
      "grad_norm": 1.3233706951141357,
      "learning_rate": 5.8175000000000005e-05,
      "loss": 0.6012,
      "step": 11641
    },
    {
      "epoch": 0.6519207078060253,
      "grad_norm": 1.1149864196777344,
      "learning_rate": 5.818e-05,
      "loss": 0.4763,
      "step": 11642
    },
    {
      "epoch": 0.6519767051181543,
      "grad_norm": 1.2007625102996826,
      "learning_rate": 5.8185e-05,
      "loss": 0.4474,
      "step": 11643
    },
    {
      "epoch": 0.6520327024302833,
      "grad_norm": 1.3394529819488525,
      "learning_rate": 5.819e-05,
      "loss": 0.5211,
      "step": 11644
    },
    {
      "epoch": 0.6520886997424123,
      "grad_norm": 1.2217689752578735,
      "learning_rate": 5.8195e-05,
      "loss": 0.4672,
      "step": 11645
    },
    {
      "epoch": 0.6521446970545414,
      "grad_norm": 1.3266834020614624,
      "learning_rate": 5.82e-05,
      "loss": 0.4283,
      "step": 11646
    },
    {
      "epoch": 0.6522006943666704,
      "grad_norm": 1.2351205348968506,
      "learning_rate": 5.8204999999999996e-05,
      "loss": 0.5023,
      "step": 11647
    },
    {
      "epoch": 0.6522566916787994,
      "grad_norm": 1.3728914260864258,
      "learning_rate": 5.821e-05,
      "loss": 0.4334,
      "step": 11648
    },
    {
      "epoch": 0.6523126889909284,
      "grad_norm": 1.0195527076721191,
      "learning_rate": 5.8215e-05,
      "loss": 0.3195,
      "step": 11649
    },
    {
      "epoch": 0.6523686863030574,
      "grad_norm": 1.3558344841003418,
      "learning_rate": 5.822000000000001e-05,
      "loss": 0.3579,
      "step": 11650
    },
    {
      "epoch": 0.6524246836151865,
      "grad_norm": 1.289319634437561,
      "learning_rate": 5.8225000000000006e-05,
      "loss": 0.4538,
      "step": 11651
    },
    {
      "epoch": 0.6524806809273155,
      "grad_norm": 1.212921142578125,
      "learning_rate": 5.823000000000001e-05,
      "loss": 0.5243,
      "step": 11652
    },
    {
      "epoch": 0.6525366782394445,
      "grad_norm": 1.3482457399368286,
      "learning_rate": 5.823500000000001e-05,
      "loss": 0.4709,
      "step": 11653
    },
    {
      "epoch": 0.6525926755515735,
      "grad_norm": 1.23513662815094,
      "learning_rate": 5.8240000000000005e-05,
      "loss": 0.3948,
      "step": 11654
    },
    {
      "epoch": 0.6526486728637025,
      "grad_norm": 1.2374106645584106,
      "learning_rate": 5.8245e-05,
      "loss": 0.4464,
      "step": 11655
    },
    {
      "epoch": 0.6527046701758316,
      "grad_norm": 1.2403322458267212,
      "learning_rate": 5.8250000000000006e-05,
      "loss": 0.4029,
      "step": 11656
    },
    {
      "epoch": 0.6527606674879606,
      "grad_norm": 1.2212584018707275,
      "learning_rate": 5.8255000000000004e-05,
      "loss": 0.3104,
      "step": 11657
    },
    {
      "epoch": 0.6528166648000896,
      "grad_norm": 1.8834686279296875,
      "learning_rate": 5.826e-05,
      "loss": 0.3212,
      "step": 11658
    },
    {
      "epoch": 0.6528726621122186,
      "grad_norm": 1.3160979747772217,
      "learning_rate": 5.8265000000000005e-05,
      "loss": 0.4347,
      "step": 11659
    },
    {
      "epoch": 0.6529286594243476,
      "grad_norm": 1.1711444854736328,
      "learning_rate": 5.827e-05,
      "loss": 0.3997,
      "step": 11660
    },
    {
      "epoch": 0.6529846567364767,
      "grad_norm": 1.2330718040466309,
      "learning_rate": 5.8275e-05,
      "loss": 0.4108,
      "step": 11661
    },
    {
      "epoch": 0.6530406540486057,
      "grad_norm": 1.1606751680374146,
      "learning_rate": 5.8280000000000004e-05,
      "loss": 0.3931,
      "step": 11662
    },
    {
      "epoch": 0.6530966513607347,
      "grad_norm": 1.2084978818893433,
      "learning_rate": 5.8285e-05,
      "loss": 0.3705,
      "step": 11663
    },
    {
      "epoch": 0.6531526486728637,
      "grad_norm": 1.577621340751648,
      "learning_rate": 5.829e-05,
      "loss": 0.5828,
      "step": 11664
    },
    {
      "epoch": 0.6532086459849927,
      "grad_norm": 1.4541676044464111,
      "learning_rate": 5.8295e-05,
      "loss": 0.5594,
      "step": 11665
    },
    {
      "epoch": 0.6532646432971217,
      "grad_norm": 1.2026711702346802,
      "learning_rate": 5.83e-05,
      "loss": 0.392,
      "step": 11666
    },
    {
      "epoch": 0.6533206406092508,
      "grad_norm": 1.5014997720718384,
      "learning_rate": 5.8305e-05,
      "loss": 0.4563,
      "step": 11667
    },
    {
      "epoch": 0.6533766379213798,
      "grad_norm": 1.4592264890670776,
      "learning_rate": 5.8309999999999996e-05,
      "loss": 0.4314,
      "step": 11668
    },
    {
      "epoch": 0.6534326352335088,
      "grad_norm": 1.177983045578003,
      "learning_rate": 5.8315e-05,
      "loss": 0.4513,
      "step": 11669
    },
    {
      "epoch": 0.6534886325456378,
      "grad_norm": 1.5815566778182983,
      "learning_rate": 5.832000000000001e-05,
      "loss": 0.4594,
      "step": 11670
    },
    {
      "epoch": 0.6535446298577668,
      "grad_norm": 1.4919463396072388,
      "learning_rate": 5.832500000000001e-05,
      "loss": 0.5935,
      "step": 11671
    },
    {
      "epoch": 0.6536006271698959,
      "grad_norm": 1.1126666069030762,
      "learning_rate": 5.8330000000000006e-05,
      "loss": 0.5506,
      "step": 11672
    },
    {
      "epoch": 0.6536566244820249,
      "grad_norm": 1.260085105895996,
      "learning_rate": 5.8335e-05,
      "loss": 0.5312,
      "step": 11673
    },
    {
      "epoch": 0.6537126217941539,
      "grad_norm": 1.4133983850479126,
      "learning_rate": 5.834000000000001e-05,
      "loss": 0.459,
      "step": 11674
    },
    {
      "epoch": 0.6537686191062829,
      "grad_norm": 1.1627137660980225,
      "learning_rate": 5.8345000000000005e-05,
      "loss": 0.4538,
      "step": 11675
    },
    {
      "epoch": 0.6538246164184119,
      "grad_norm": 1.4787089824676514,
      "learning_rate": 5.835e-05,
      "loss": 0.4798,
      "step": 11676
    },
    {
      "epoch": 0.653880613730541,
      "grad_norm": 1.3909518718719482,
      "learning_rate": 5.8355000000000006e-05,
      "loss": 0.4618,
      "step": 11677
    },
    {
      "epoch": 0.65393661104267,
      "grad_norm": 1.359830617904663,
      "learning_rate": 5.8360000000000004e-05,
      "loss": 0.4567,
      "step": 11678
    },
    {
      "epoch": 0.653992608354799,
      "grad_norm": 1.7665683031082153,
      "learning_rate": 5.8365e-05,
      "loss": 0.5119,
      "step": 11679
    },
    {
      "epoch": 0.654048605666928,
      "grad_norm": 1.4605189561843872,
      "learning_rate": 5.8370000000000005e-05,
      "loss": 0.6055,
      "step": 11680
    },
    {
      "epoch": 0.654104602979057,
      "grad_norm": 1.4210529327392578,
      "learning_rate": 5.8375e-05,
      "loss": 0.3897,
      "step": 11681
    },
    {
      "epoch": 0.6541606002911861,
      "grad_norm": 0.9742177128791809,
      "learning_rate": 5.838e-05,
      "loss": 0.2776,
      "step": 11682
    },
    {
      "epoch": 0.6542165976033151,
      "grad_norm": 1.2380565404891968,
      "learning_rate": 5.8385e-05,
      "loss": 0.4494,
      "step": 11683
    },
    {
      "epoch": 0.6542725949154441,
      "grad_norm": 1.3207238912582397,
      "learning_rate": 5.839e-05,
      "loss": 0.5514,
      "step": 11684
    },
    {
      "epoch": 0.6543285922275731,
      "grad_norm": 1.5492300987243652,
      "learning_rate": 5.8395e-05,
      "loss": 0.3392,
      "step": 11685
    },
    {
      "epoch": 0.6543845895397021,
      "grad_norm": 1.3901665210723877,
      "learning_rate": 5.8399999999999997e-05,
      "loss": 0.4178,
      "step": 11686
    },
    {
      "epoch": 0.6544405868518312,
      "grad_norm": 1.3766168355941772,
      "learning_rate": 5.8405e-05,
      "loss": 0.4283,
      "step": 11687
    },
    {
      "epoch": 0.6544965841639602,
      "grad_norm": 1.496849536895752,
      "learning_rate": 5.841e-05,
      "loss": 0.5418,
      "step": 11688
    },
    {
      "epoch": 0.6545525814760892,
      "grad_norm": 1.4288110733032227,
      "learning_rate": 5.8414999999999996e-05,
      "loss": 0.4658,
      "step": 11689
    },
    {
      "epoch": 0.6546085787882182,
      "grad_norm": 1.2487083673477173,
      "learning_rate": 5.8420000000000006e-05,
      "loss": 0.4451,
      "step": 11690
    },
    {
      "epoch": 0.6546645761003472,
      "grad_norm": 1.3596370220184326,
      "learning_rate": 5.842500000000001e-05,
      "loss": 0.3989,
      "step": 11691
    },
    {
      "epoch": 0.6547205734124762,
      "grad_norm": 1.147562026977539,
      "learning_rate": 5.843000000000001e-05,
      "loss": 0.4082,
      "step": 11692
    },
    {
      "epoch": 0.6547765707246053,
      "grad_norm": 1.269890308380127,
      "learning_rate": 5.8435000000000005e-05,
      "loss": 0.3645,
      "step": 11693
    },
    {
      "epoch": 0.6548325680367343,
      "grad_norm": 1.1032383441925049,
      "learning_rate": 5.844e-05,
      "loss": 0.3172,
      "step": 11694
    },
    {
      "epoch": 0.6548885653488633,
      "grad_norm": 1.4504663944244385,
      "learning_rate": 5.844500000000001e-05,
      "loss": 0.4008,
      "step": 11695
    },
    {
      "epoch": 0.6549445626609923,
      "grad_norm": 1.4644497632980347,
      "learning_rate": 5.8450000000000005e-05,
      "loss": 0.465,
      "step": 11696
    },
    {
      "epoch": 0.6550005599731213,
      "grad_norm": 1.419569492340088,
      "learning_rate": 5.8455e-05,
      "loss": 0.4715,
      "step": 11697
    },
    {
      "epoch": 0.6550565572852504,
      "grad_norm": 1.3105918169021606,
      "learning_rate": 5.8460000000000006e-05,
      "loss": 0.412,
      "step": 11698
    },
    {
      "epoch": 0.6551125545973793,
      "grad_norm": 1.656390905380249,
      "learning_rate": 5.8465000000000004e-05,
      "loss": 0.6538,
      "step": 11699
    },
    {
      "epoch": 0.6551685519095083,
      "grad_norm": 1.3274263143539429,
      "learning_rate": 5.847e-05,
      "loss": 0.4924,
      "step": 11700
    },
    {
      "epoch": 0.6552245492216373,
      "grad_norm": 1.4678207635879517,
      "learning_rate": 5.8475000000000005e-05,
      "loss": 0.6668,
      "step": 11701
    },
    {
      "epoch": 0.6552805465337663,
      "grad_norm": 1.424195647239685,
      "learning_rate": 5.848e-05,
      "loss": 0.562,
      "step": 11702
    },
    {
      "epoch": 0.6553365438458953,
      "grad_norm": 1.2749199867248535,
      "learning_rate": 5.8485e-05,
      "loss": 0.4165,
      "step": 11703
    },
    {
      "epoch": 0.6553925411580244,
      "grad_norm": 1.5304670333862305,
      "learning_rate": 5.849e-05,
      "loss": 0.5117,
      "step": 11704
    },
    {
      "epoch": 0.6554485384701534,
      "grad_norm": 1.12393319606781,
      "learning_rate": 5.8495e-05,
      "loss": 0.4185,
      "step": 11705
    },
    {
      "epoch": 0.6555045357822824,
      "grad_norm": 1.443090796470642,
      "learning_rate": 5.85e-05,
      "loss": 0.3894,
      "step": 11706
    },
    {
      "epoch": 0.6555605330944114,
      "grad_norm": 1.3098422288894653,
      "learning_rate": 5.8504999999999996e-05,
      "loss": 0.499,
      "step": 11707
    },
    {
      "epoch": 0.6556165304065404,
      "grad_norm": 1.2031232118606567,
      "learning_rate": 5.851e-05,
      "loss": 0.3625,
      "step": 11708
    },
    {
      "epoch": 0.6556725277186695,
      "grad_norm": 1.471382737159729,
      "learning_rate": 5.8515e-05,
      "loss": 0.3973,
      "step": 11709
    },
    {
      "epoch": 0.6557285250307985,
      "grad_norm": 1.3180073499679565,
      "learning_rate": 5.852000000000001e-05,
      "loss": 0.4428,
      "step": 11710
    },
    {
      "epoch": 0.6557845223429275,
      "grad_norm": 1.5244274139404297,
      "learning_rate": 5.8525000000000006e-05,
      "loss": 0.5903,
      "step": 11711
    },
    {
      "epoch": 0.6558405196550565,
      "grad_norm": 1.1666597127914429,
      "learning_rate": 5.8530000000000004e-05,
      "loss": 0.363,
      "step": 11712
    },
    {
      "epoch": 0.6558965169671855,
      "grad_norm": 1.1757006645202637,
      "learning_rate": 5.853500000000001e-05,
      "loss": 0.41,
      "step": 11713
    },
    {
      "epoch": 0.6559525142793146,
      "grad_norm": 1.4187425374984741,
      "learning_rate": 5.8540000000000005e-05,
      "loss": 0.3809,
      "step": 11714
    },
    {
      "epoch": 0.6560085115914436,
      "grad_norm": 1.2364821434020996,
      "learning_rate": 5.8545e-05,
      "loss": 0.3692,
      "step": 11715
    },
    {
      "epoch": 0.6560645089035726,
      "grad_norm": 1.5956910848617554,
      "learning_rate": 5.855000000000001e-05,
      "loss": 0.4042,
      "step": 11716
    },
    {
      "epoch": 0.6561205062157016,
      "grad_norm": 1.5447666645050049,
      "learning_rate": 5.8555000000000004e-05,
      "loss": 0.5662,
      "step": 11717
    },
    {
      "epoch": 0.6561765035278306,
      "grad_norm": 1.403235912322998,
      "learning_rate": 5.856e-05,
      "loss": 0.5386,
      "step": 11718
    },
    {
      "epoch": 0.6562325008399597,
      "grad_norm": 1.3280143737792969,
      "learning_rate": 5.8565000000000006e-05,
      "loss": 0.3795,
      "step": 11719
    },
    {
      "epoch": 0.6562884981520887,
      "grad_norm": 1.3676313161849976,
      "learning_rate": 5.857e-05,
      "loss": 0.4279,
      "step": 11720
    },
    {
      "epoch": 0.6563444954642177,
      "grad_norm": 1.336786150932312,
      "learning_rate": 5.8575e-05,
      "loss": 0.4123,
      "step": 11721
    },
    {
      "epoch": 0.6564004927763467,
      "grad_norm": 1.216882348060608,
      "learning_rate": 5.858e-05,
      "loss": 0.4441,
      "step": 11722
    },
    {
      "epoch": 0.6564564900884757,
      "grad_norm": 12.355265617370605,
      "learning_rate": 5.8585e-05,
      "loss": 0.4049,
      "step": 11723
    },
    {
      "epoch": 0.6565124874006047,
      "grad_norm": 1.1024394035339355,
      "learning_rate": 5.859e-05,
      "loss": 0.4176,
      "step": 11724
    },
    {
      "epoch": 0.6565684847127338,
      "grad_norm": 1.2378270626068115,
      "learning_rate": 5.8595e-05,
      "loss": 0.3146,
      "step": 11725
    },
    {
      "epoch": 0.6566244820248628,
      "grad_norm": 1.1719168424606323,
      "learning_rate": 5.86e-05,
      "loss": 0.3656,
      "step": 11726
    },
    {
      "epoch": 0.6566804793369918,
      "grad_norm": 1.1853828430175781,
      "learning_rate": 5.8605e-05,
      "loss": 0.3465,
      "step": 11727
    },
    {
      "epoch": 0.6567364766491208,
      "grad_norm": 1.3977810144424438,
      "learning_rate": 5.8609999999999996e-05,
      "loss": 0.436,
      "step": 11728
    },
    {
      "epoch": 0.6567924739612498,
      "grad_norm": 1.3636177778244019,
      "learning_rate": 5.8615e-05,
      "loss": 0.4208,
      "step": 11729
    },
    {
      "epoch": 0.6568484712733789,
      "grad_norm": 1.3164868354797363,
      "learning_rate": 5.862000000000001e-05,
      "loss": 0.4447,
      "step": 11730
    },
    {
      "epoch": 0.6569044685855079,
      "grad_norm": 1.3972855806350708,
      "learning_rate": 5.862500000000001e-05,
      "loss": 0.4463,
      "step": 11731
    },
    {
      "epoch": 0.6569604658976369,
      "grad_norm": 1.3859697580337524,
      "learning_rate": 5.8630000000000006e-05,
      "loss": 0.5974,
      "step": 11732
    },
    {
      "epoch": 0.6570164632097659,
      "grad_norm": 1.2899953126907349,
      "learning_rate": 5.8635000000000004e-05,
      "loss": 0.3652,
      "step": 11733
    },
    {
      "epoch": 0.6570724605218949,
      "grad_norm": 1.3799649477005005,
      "learning_rate": 5.864000000000001e-05,
      "loss": 0.4831,
      "step": 11734
    },
    {
      "epoch": 0.657128457834024,
      "grad_norm": 1.995896339416504,
      "learning_rate": 5.8645000000000005e-05,
      "loss": 0.5063,
      "step": 11735
    },
    {
      "epoch": 0.657184455146153,
      "grad_norm": 1.4469842910766602,
      "learning_rate": 5.865e-05,
      "loss": 0.4502,
      "step": 11736
    },
    {
      "epoch": 0.657240452458282,
      "grad_norm": 1.168785572052002,
      "learning_rate": 5.865500000000001e-05,
      "loss": 0.3616,
      "step": 11737
    },
    {
      "epoch": 0.657296449770411,
      "grad_norm": 1.3240330219268799,
      "learning_rate": 5.8660000000000004e-05,
      "loss": 0.4383,
      "step": 11738
    },
    {
      "epoch": 0.65735244708254,
      "grad_norm": 1.1602659225463867,
      "learning_rate": 5.8665e-05,
      "loss": 0.3743,
      "step": 11739
    },
    {
      "epoch": 0.6574084443946691,
      "grad_norm": 1.331207513809204,
      "learning_rate": 5.8670000000000006e-05,
      "loss": 0.3482,
      "step": 11740
    },
    {
      "epoch": 0.6574644417067981,
      "grad_norm": 1.1242907047271729,
      "learning_rate": 5.8675e-05,
      "loss": 0.4016,
      "step": 11741
    },
    {
      "epoch": 0.6575204390189271,
      "grad_norm": 1.4794058799743652,
      "learning_rate": 5.868e-05,
      "loss": 0.6036,
      "step": 11742
    },
    {
      "epoch": 0.6575764363310561,
      "grad_norm": 1.4201059341430664,
      "learning_rate": 5.8685e-05,
      "loss": 0.5238,
      "step": 11743
    },
    {
      "epoch": 0.6576324336431851,
      "grad_norm": 1.3712326288223267,
      "learning_rate": 5.869e-05,
      "loss": 0.4391,
      "step": 11744
    },
    {
      "epoch": 0.6576884309553142,
      "grad_norm": 1.6325640678405762,
      "learning_rate": 5.8695e-05,
      "loss": 0.4827,
      "step": 11745
    },
    {
      "epoch": 0.6577444282674432,
      "grad_norm": 1.094815969467163,
      "learning_rate": 5.87e-05,
      "loss": 0.4478,
      "step": 11746
    },
    {
      "epoch": 0.6578004255795722,
      "grad_norm": 1.4926972389221191,
      "learning_rate": 5.8705e-05,
      "loss": 0.4445,
      "step": 11747
    },
    {
      "epoch": 0.6578564228917012,
      "grad_norm": 1.0443830490112305,
      "learning_rate": 5.871e-05,
      "loss": 0.3351,
      "step": 11748
    },
    {
      "epoch": 0.6579124202038302,
      "grad_norm": 1.129218339920044,
      "learning_rate": 5.8714999999999996e-05,
      "loss": 0.3688,
      "step": 11749
    },
    {
      "epoch": 0.6579684175159592,
      "grad_norm": 1.281672477722168,
      "learning_rate": 5.872000000000001e-05,
      "loss": 0.5401,
      "step": 11750
    },
    {
      "epoch": 0.6580244148280883,
      "grad_norm": 1.2422080039978027,
      "learning_rate": 5.8725000000000004e-05,
      "loss": 0.3815,
      "step": 11751
    },
    {
      "epoch": 0.6580804121402173,
      "grad_norm": 1.2020775079727173,
      "learning_rate": 5.873000000000001e-05,
      "loss": 0.376,
      "step": 11752
    },
    {
      "epoch": 0.6581364094523463,
      "grad_norm": 1.1493568420410156,
      "learning_rate": 5.8735000000000006e-05,
      "loss": 0.4852,
      "step": 11753
    },
    {
      "epoch": 0.6581924067644753,
      "grad_norm": 1.532774806022644,
      "learning_rate": 5.8740000000000003e-05,
      "loss": 0.3621,
      "step": 11754
    },
    {
      "epoch": 0.6582484040766043,
      "grad_norm": 1.2717229127883911,
      "learning_rate": 5.874500000000001e-05,
      "loss": 0.4511,
      "step": 11755
    },
    {
      "epoch": 0.6583044013887334,
      "grad_norm": 1.4544858932495117,
      "learning_rate": 5.8750000000000005e-05,
      "loss": 0.5239,
      "step": 11756
    },
    {
      "epoch": 0.6583603987008624,
      "grad_norm": 1.1929380893707275,
      "learning_rate": 5.8755e-05,
      "loss": 0.4486,
      "step": 11757
    },
    {
      "epoch": 0.6584163960129914,
      "grad_norm": 1.1866029500961304,
      "learning_rate": 5.876000000000001e-05,
      "loss": 0.3347,
      "step": 11758
    },
    {
      "epoch": 0.6584723933251204,
      "grad_norm": 1.4016965627670288,
      "learning_rate": 5.8765000000000004e-05,
      "loss": 0.5889,
      "step": 11759
    },
    {
      "epoch": 0.6585283906372494,
      "grad_norm": 1.2263785600662231,
      "learning_rate": 5.877e-05,
      "loss": 0.3747,
      "step": 11760
    },
    {
      "epoch": 0.6585843879493785,
      "grad_norm": 1.1574546098709106,
      "learning_rate": 5.8775000000000006e-05,
      "loss": 0.4193,
      "step": 11761
    },
    {
      "epoch": 0.6586403852615075,
      "grad_norm": 1.2843056917190552,
      "learning_rate": 5.878e-05,
      "loss": 0.4564,
      "step": 11762
    },
    {
      "epoch": 0.6586963825736365,
      "grad_norm": 1.2758581638336182,
      "learning_rate": 5.8785e-05,
      "loss": 0.3833,
      "step": 11763
    },
    {
      "epoch": 0.6587523798857655,
      "grad_norm": 1.386577844619751,
      "learning_rate": 5.879e-05,
      "loss": 0.481,
      "step": 11764
    },
    {
      "epoch": 0.6588083771978945,
      "grad_norm": 1.3712282180786133,
      "learning_rate": 5.8795e-05,
      "loss": 0.5142,
      "step": 11765
    },
    {
      "epoch": 0.6588643745100236,
      "grad_norm": 1.4404501914978027,
      "learning_rate": 5.88e-05,
      "loss": 0.467,
      "step": 11766
    },
    {
      "epoch": 0.6589203718221526,
      "grad_norm": 1.213527798652649,
      "learning_rate": 5.8805e-05,
      "loss": 0.3757,
      "step": 11767
    },
    {
      "epoch": 0.6589763691342816,
      "grad_norm": 1.4828687906265259,
      "learning_rate": 5.881e-05,
      "loss": 0.4814,
      "step": 11768
    },
    {
      "epoch": 0.6590323664464106,
      "grad_norm": 1.2644912004470825,
      "learning_rate": 5.8815e-05,
      "loss": 0.4162,
      "step": 11769
    },
    {
      "epoch": 0.6590883637585396,
      "grad_norm": 1.4774798154830933,
      "learning_rate": 5.8819999999999996e-05,
      "loss": 0.6192,
      "step": 11770
    },
    {
      "epoch": 0.6591443610706686,
      "grad_norm": 2.14587140083313,
      "learning_rate": 5.882500000000001e-05,
      "loss": 0.479,
      "step": 11771
    },
    {
      "epoch": 0.6592003583827977,
      "grad_norm": 1.6047451496124268,
      "learning_rate": 5.8830000000000004e-05,
      "loss": 0.4926,
      "step": 11772
    },
    {
      "epoch": 0.6592563556949267,
      "grad_norm": 1.5371648073196411,
      "learning_rate": 5.883500000000001e-05,
      "loss": 0.3083,
      "step": 11773
    },
    {
      "epoch": 0.6593123530070557,
      "grad_norm": 1.403256893157959,
      "learning_rate": 5.8840000000000006e-05,
      "loss": 0.5472,
      "step": 11774
    },
    {
      "epoch": 0.6593683503191847,
      "grad_norm": 1.0998344421386719,
      "learning_rate": 5.8845e-05,
      "loss": 0.3273,
      "step": 11775
    },
    {
      "epoch": 0.6594243476313137,
      "grad_norm": 1.3494468927383423,
      "learning_rate": 5.885000000000001e-05,
      "loss": 0.706,
      "step": 11776
    },
    {
      "epoch": 0.6594803449434428,
      "grad_norm": 1.1570298671722412,
      "learning_rate": 5.8855000000000005e-05,
      "loss": 0.3141,
      "step": 11777
    },
    {
      "epoch": 0.6595363422555718,
      "grad_norm": 1.5827627182006836,
      "learning_rate": 5.886e-05,
      "loss": 0.5053,
      "step": 11778
    },
    {
      "epoch": 0.6595923395677008,
      "grad_norm": 1.2346322536468506,
      "learning_rate": 5.8865000000000006e-05,
      "loss": 0.3815,
      "step": 11779
    },
    {
      "epoch": 0.6596483368798298,
      "grad_norm": 1.3112822771072388,
      "learning_rate": 5.8870000000000004e-05,
      "loss": 0.4709,
      "step": 11780
    },
    {
      "epoch": 0.6597043341919588,
      "grad_norm": 1.2978674173355103,
      "learning_rate": 5.8875e-05,
      "loss": 0.3935,
      "step": 11781
    },
    {
      "epoch": 0.6597603315040877,
      "grad_norm": 1.329642653465271,
      "learning_rate": 5.888e-05,
      "loss": 0.4252,
      "step": 11782
    },
    {
      "epoch": 0.6598163288162168,
      "grad_norm": 1.0972132682800293,
      "learning_rate": 5.8885e-05,
      "loss": 0.3462,
      "step": 11783
    },
    {
      "epoch": 0.6598723261283458,
      "grad_norm": 1.3023643493652344,
      "learning_rate": 5.889e-05,
      "loss": 0.4253,
      "step": 11784
    },
    {
      "epoch": 0.6599283234404748,
      "grad_norm": 1.5814366340637207,
      "learning_rate": 5.8895e-05,
      "loss": 0.4563,
      "step": 11785
    },
    {
      "epoch": 0.6599843207526038,
      "grad_norm": 1.316964864730835,
      "learning_rate": 5.89e-05,
      "loss": 0.4273,
      "step": 11786
    },
    {
      "epoch": 0.6600403180647328,
      "grad_norm": 1.2883378267288208,
      "learning_rate": 5.8905e-05,
      "loss": 0.5418,
      "step": 11787
    },
    {
      "epoch": 0.6600963153768619,
      "grad_norm": 1.5526007413864136,
      "learning_rate": 5.891e-05,
      "loss": 0.4849,
      "step": 11788
    },
    {
      "epoch": 0.6601523126889909,
      "grad_norm": 12.64220905303955,
      "learning_rate": 5.8915e-05,
      "loss": 0.4976,
      "step": 11789
    },
    {
      "epoch": 0.6602083100011199,
      "grad_norm": 1.3078986406326294,
      "learning_rate": 5.892e-05,
      "loss": 0.5009,
      "step": 11790
    },
    {
      "epoch": 0.6602643073132489,
      "grad_norm": 1.2736436128616333,
      "learning_rate": 5.892500000000001e-05,
      "loss": 0.4243,
      "step": 11791
    },
    {
      "epoch": 0.6603203046253779,
      "grad_norm": 1.6053931713104248,
      "learning_rate": 5.893000000000001e-05,
      "loss": 0.5212,
      "step": 11792
    },
    {
      "epoch": 0.660376301937507,
      "grad_norm": 1.125852108001709,
      "learning_rate": 5.8935000000000004e-05,
      "loss": 0.3243,
      "step": 11793
    },
    {
      "epoch": 0.660432299249636,
      "grad_norm": 1.2537152767181396,
      "learning_rate": 5.894000000000001e-05,
      "loss": 0.3947,
      "step": 11794
    },
    {
      "epoch": 0.660488296561765,
      "grad_norm": 1.3903148174285889,
      "learning_rate": 5.8945000000000006e-05,
      "loss": 0.3863,
      "step": 11795
    },
    {
      "epoch": 0.660544293873894,
      "grad_norm": 1.2561471462249756,
      "learning_rate": 5.895e-05,
      "loss": 0.4033,
      "step": 11796
    },
    {
      "epoch": 0.660600291186023,
      "grad_norm": 1.3503178358078003,
      "learning_rate": 5.895500000000001e-05,
      "loss": 0.5024,
      "step": 11797
    },
    {
      "epoch": 0.6606562884981521,
      "grad_norm": 1.8672709465026855,
      "learning_rate": 5.8960000000000005e-05,
      "loss": 0.3454,
      "step": 11798
    },
    {
      "epoch": 0.6607122858102811,
      "grad_norm": 1.1492935419082642,
      "learning_rate": 5.8965e-05,
      "loss": 0.3286,
      "step": 11799
    },
    {
      "epoch": 0.6607682831224101,
      "grad_norm": 1.6859663724899292,
      "learning_rate": 5.8970000000000006e-05,
      "loss": 0.4703,
      "step": 11800
    },
    {
      "epoch": 0.6608242804345391,
      "grad_norm": 1.3863939046859741,
      "learning_rate": 5.8975000000000004e-05,
      "loss": 0.3802,
      "step": 11801
    },
    {
      "epoch": 0.6608802777466681,
      "grad_norm": 1.2479121685028076,
      "learning_rate": 5.898e-05,
      "loss": 0.428,
      "step": 11802
    },
    {
      "epoch": 0.6609362750587972,
      "grad_norm": 1.3120057582855225,
      "learning_rate": 5.8985e-05,
      "loss": 0.5067,
      "step": 11803
    },
    {
      "epoch": 0.6609922723709262,
      "grad_norm": 1.1948027610778809,
      "learning_rate": 5.899e-05,
      "loss": 0.3471,
      "step": 11804
    },
    {
      "epoch": 0.6610482696830552,
      "grad_norm": 1.3826274871826172,
      "learning_rate": 5.8995e-05,
      "loss": 0.5424,
      "step": 11805
    },
    {
      "epoch": 0.6611042669951842,
      "grad_norm": 1.0666022300720215,
      "learning_rate": 5.9e-05,
      "loss": 0.3621,
      "step": 11806
    },
    {
      "epoch": 0.6611602643073132,
      "grad_norm": 1.216614007949829,
      "learning_rate": 5.9005e-05,
      "loss": 0.3731,
      "step": 11807
    },
    {
      "epoch": 0.6612162616194422,
      "grad_norm": 1.5249807834625244,
      "learning_rate": 5.901e-05,
      "loss": 0.5525,
      "step": 11808
    },
    {
      "epoch": 0.6612722589315713,
      "grad_norm": 1.3568562269210815,
      "learning_rate": 5.9014999999999997e-05,
      "loss": 0.435,
      "step": 11809
    },
    {
      "epoch": 0.6613282562437003,
      "grad_norm": 1.2439298629760742,
      "learning_rate": 5.902e-05,
      "loss": 0.423,
      "step": 11810
    },
    {
      "epoch": 0.6613842535558293,
      "grad_norm": 1.246422290802002,
      "learning_rate": 5.9025000000000005e-05,
      "loss": 0.321,
      "step": 11811
    },
    {
      "epoch": 0.6614402508679583,
      "grad_norm": 1.2262554168701172,
      "learning_rate": 5.903000000000001e-05,
      "loss": 0.4811,
      "step": 11812
    },
    {
      "epoch": 0.6614962481800873,
      "grad_norm": 1.3477610349655151,
      "learning_rate": 5.9035000000000007e-05,
      "loss": 0.4632,
      "step": 11813
    },
    {
      "epoch": 0.6615522454922164,
      "grad_norm": 1.1747660636901855,
      "learning_rate": 5.9040000000000004e-05,
      "loss": 0.374,
      "step": 11814
    },
    {
      "epoch": 0.6616082428043454,
      "grad_norm": 1.1574561595916748,
      "learning_rate": 5.904500000000001e-05,
      "loss": 0.317,
      "step": 11815
    },
    {
      "epoch": 0.6616642401164744,
      "grad_norm": 1.3460285663604736,
      "learning_rate": 5.9050000000000006e-05,
      "loss": 0.5394,
      "step": 11816
    },
    {
      "epoch": 0.6617202374286034,
      "grad_norm": 1.1897482872009277,
      "learning_rate": 5.9055e-05,
      "loss": 0.353,
      "step": 11817
    },
    {
      "epoch": 0.6617762347407324,
      "grad_norm": 1.5638235807418823,
      "learning_rate": 5.906000000000001e-05,
      "loss": 0.4266,
      "step": 11818
    },
    {
      "epoch": 0.6618322320528615,
      "grad_norm": 1.418049931526184,
      "learning_rate": 5.9065000000000005e-05,
      "loss": 0.5869,
      "step": 11819
    },
    {
      "epoch": 0.6618882293649905,
      "grad_norm": 1.3842191696166992,
      "learning_rate": 5.907e-05,
      "loss": 0.4286,
      "step": 11820
    },
    {
      "epoch": 0.6619442266771195,
      "grad_norm": 1.5180695056915283,
      "learning_rate": 5.9075e-05,
      "loss": 0.5938,
      "step": 11821
    },
    {
      "epoch": 0.6620002239892485,
      "grad_norm": 1.3481963872909546,
      "learning_rate": 5.9080000000000004e-05,
      "loss": 0.4343,
      "step": 11822
    },
    {
      "epoch": 0.6620562213013775,
      "grad_norm": 1.3385549783706665,
      "learning_rate": 5.9085e-05,
      "loss": 0.3347,
      "step": 11823
    },
    {
      "epoch": 0.6621122186135066,
      "grad_norm": 1.5572330951690674,
      "learning_rate": 5.909e-05,
      "loss": 0.4192,
      "step": 11824
    },
    {
      "epoch": 0.6621682159256356,
      "grad_norm": 1.4639410972595215,
      "learning_rate": 5.9095e-05,
      "loss": 0.5353,
      "step": 11825
    },
    {
      "epoch": 0.6622242132377646,
      "grad_norm": 1.4320757389068604,
      "learning_rate": 5.91e-05,
      "loss": 0.473,
      "step": 11826
    },
    {
      "epoch": 0.6622802105498936,
      "grad_norm": 1.3239295482635498,
      "learning_rate": 5.9105e-05,
      "loss": 0.4496,
      "step": 11827
    },
    {
      "epoch": 0.6623362078620226,
      "grad_norm": 1.1210360527038574,
      "learning_rate": 5.911e-05,
      "loss": 0.418,
      "step": 11828
    },
    {
      "epoch": 0.6623922051741516,
      "grad_norm": 2.27600359916687,
      "learning_rate": 5.9115e-05,
      "loss": 0.4041,
      "step": 11829
    },
    {
      "epoch": 0.6624482024862807,
      "grad_norm": 1.2792423963546753,
      "learning_rate": 5.9119999999999996e-05,
      "loss": 0.36,
      "step": 11830
    },
    {
      "epoch": 0.6625041997984097,
      "grad_norm": 1.1539493799209595,
      "learning_rate": 5.912500000000001e-05,
      "loss": 0.3898,
      "step": 11831
    },
    {
      "epoch": 0.6625601971105387,
      "grad_norm": 1.31495201587677,
      "learning_rate": 5.9130000000000005e-05,
      "loss": 0.4026,
      "step": 11832
    },
    {
      "epoch": 0.6626161944226677,
      "grad_norm": 1.0842198133468628,
      "learning_rate": 5.913500000000001e-05,
      "loss": 0.3748,
      "step": 11833
    },
    {
      "epoch": 0.6626721917347967,
      "grad_norm": 1.3249589204788208,
      "learning_rate": 5.9140000000000006e-05,
      "loss": 0.5073,
      "step": 11834
    },
    {
      "epoch": 0.6627281890469258,
      "grad_norm": 1.3649876117706299,
      "learning_rate": 5.9145000000000004e-05,
      "loss": 0.43,
      "step": 11835
    },
    {
      "epoch": 0.6627841863590548,
      "grad_norm": 1.2828794717788696,
      "learning_rate": 5.915000000000001e-05,
      "loss": 0.417,
      "step": 11836
    },
    {
      "epoch": 0.6628401836711838,
      "grad_norm": 1.9186866283416748,
      "learning_rate": 5.9155000000000005e-05,
      "loss": 0.4914,
      "step": 11837
    },
    {
      "epoch": 0.6628961809833128,
      "grad_norm": 1.1696540117263794,
      "learning_rate": 5.916e-05,
      "loss": 0.3673,
      "step": 11838
    },
    {
      "epoch": 0.6629521782954418,
      "grad_norm": 1.321453332901001,
      "learning_rate": 5.916500000000001e-05,
      "loss": 0.4073,
      "step": 11839
    },
    {
      "epoch": 0.6630081756075709,
      "grad_norm": 1.1427853107452393,
      "learning_rate": 5.9170000000000004e-05,
      "loss": 0.4659,
      "step": 11840
    },
    {
      "epoch": 0.6630641729196999,
      "grad_norm": 1.3836218118667603,
      "learning_rate": 5.9175e-05,
      "loss": 0.3893,
      "step": 11841
    },
    {
      "epoch": 0.6631201702318289,
      "grad_norm": 1.653493881225586,
      "learning_rate": 5.918e-05,
      "loss": 0.431,
      "step": 11842
    },
    {
      "epoch": 0.6631761675439579,
      "grad_norm": 1.120078444480896,
      "learning_rate": 5.9185000000000003e-05,
      "loss": 0.3408,
      "step": 11843
    },
    {
      "epoch": 0.6632321648560869,
      "grad_norm": 1.3832314014434814,
      "learning_rate": 5.919e-05,
      "loss": 0.4999,
      "step": 11844
    },
    {
      "epoch": 0.663288162168216,
      "grad_norm": 1.2807661294937134,
      "learning_rate": 5.9195e-05,
      "loss": 0.3549,
      "step": 11845
    },
    {
      "epoch": 0.663344159480345,
      "grad_norm": 1.1888970136642456,
      "learning_rate": 5.92e-05,
      "loss": 0.3814,
      "step": 11846
    },
    {
      "epoch": 0.663400156792474,
      "grad_norm": 1.893478274345398,
      "learning_rate": 5.9205e-05,
      "loss": 0.4226,
      "step": 11847
    },
    {
      "epoch": 0.663456154104603,
      "grad_norm": 1.1474511623382568,
      "learning_rate": 5.921e-05,
      "loss": 0.3579,
      "step": 11848
    },
    {
      "epoch": 0.663512151416732,
      "grad_norm": 1.5900508165359497,
      "learning_rate": 5.9215e-05,
      "loss": 0.6149,
      "step": 11849
    },
    {
      "epoch": 0.663568148728861,
      "grad_norm": 1.3894587755203247,
      "learning_rate": 5.922e-05,
      "loss": 0.3685,
      "step": 11850
    },
    {
      "epoch": 0.6636241460409901,
      "grad_norm": 1.6272608041763306,
      "learning_rate": 5.922500000000001e-05,
      "loss": 0.5183,
      "step": 11851
    },
    {
      "epoch": 0.6636801433531191,
      "grad_norm": 1.393485188484192,
      "learning_rate": 5.923000000000001e-05,
      "loss": 0.5789,
      "step": 11852
    },
    {
      "epoch": 0.6637361406652481,
      "grad_norm": 1.3365730047225952,
      "learning_rate": 5.9235000000000005e-05,
      "loss": 0.3492,
      "step": 11853
    },
    {
      "epoch": 0.6637921379773771,
      "grad_norm": 1.1424380540847778,
      "learning_rate": 5.924000000000001e-05,
      "loss": 0.314,
      "step": 11854
    },
    {
      "epoch": 0.6638481352895061,
      "grad_norm": 1.5570404529571533,
      "learning_rate": 5.9245000000000006e-05,
      "loss": 0.4278,
      "step": 11855
    },
    {
      "epoch": 0.6639041326016352,
      "grad_norm": 1.0318195819854736,
      "learning_rate": 5.9250000000000004e-05,
      "loss": 0.3839,
      "step": 11856
    },
    {
      "epoch": 0.6639601299137642,
      "grad_norm": 1.310918927192688,
      "learning_rate": 5.925500000000001e-05,
      "loss": 0.4464,
      "step": 11857
    },
    {
      "epoch": 0.6640161272258932,
      "grad_norm": 1.6875386238098145,
      "learning_rate": 5.9260000000000005e-05,
      "loss": 0.557,
      "step": 11858
    },
    {
      "epoch": 0.6640721245380222,
      "grad_norm": 1.0900417566299438,
      "learning_rate": 5.9265e-05,
      "loss": 0.3614,
      "step": 11859
    },
    {
      "epoch": 0.6641281218501512,
      "grad_norm": 1.054628849029541,
      "learning_rate": 5.927e-05,
      "loss": 0.3527,
      "step": 11860
    },
    {
      "epoch": 0.6641841191622803,
      "grad_norm": 1.3811562061309814,
      "learning_rate": 5.9275000000000004e-05,
      "loss": 0.4464,
      "step": 11861
    },
    {
      "epoch": 0.6642401164744093,
      "grad_norm": 1.394370675086975,
      "learning_rate": 5.928e-05,
      "loss": 0.6263,
      "step": 11862
    },
    {
      "epoch": 0.6642961137865383,
      "grad_norm": 1.579593539237976,
      "learning_rate": 5.9285e-05,
      "loss": 0.5119,
      "step": 11863
    },
    {
      "epoch": 0.6643521110986672,
      "grad_norm": 1.3653138875961304,
      "learning_rate": 5.929e-05,
      "loss": 0.4603,
      "step": 11864
    },
    {
      "epoch": 0.6644081084107962,
      "grad_norm": 1.4746373891830444,
      "learning_rate": 5.9295e-05,
      "loss": 0.4037,
      "step": 11865
    },
    {
      "epoch": 0.6644641057229252,
      "grad_norm": 1.2131935358047485,
      "learning_rate": 5.93e-05,
      "loss": 0.3738,
      "step": 11866
    },
    {
      "epoch": 0.6645201030350543,
      "grad_norm": 1.1263573169708252,
      "learning_rate": 5.9305e-05,
      "loss": 0.3515,
      "step": 11867
    },
    {
      "epoch": 0.6645761003471833,
      "grad_norm": 1.3177452087402344,
      "learning_rate": 5.931e-05,
      "loss": 0.3774,
      "step": 11868
    },
    {
      "epoch": 0.6646320976593123,
      "grad_norm": 1.270390272140503,
      "learning_rate": 5.9315e-05,
      "loss": 0.4753,
      "step": 11869
    },
    {
      "epoch": 0.6646880949714413,
      "grad_norm": 1.494642734527588,
      "learning_rate": 5.9319999999999994e-05,
      "loss": 0.3979,
      "step": 11870
    },
    {
      "epoch": 0.6647440922835703,
      "grad_norm": 1.6839993000030518,
      "learning_rate": 5.9325000000000005e-05,
      "loss": 0.5764,
      "step": 11871
    },
    {
      "epoch": 0.6648000895956994,
      "grad_norm": 1.3141297101974487,
      "learning_rate": 5.933000000000001e-05,
      "loss": 0.3978,
      "step": 11872
    },
    {
      "epoch": 0.6648560869078284,
      "grad_norm": 1.3168694972991943,
      "learning_rate": 5.933500000000001e-05,
      "loss": 0.469,
      "step": 11873
    },
    {
      "epoch": 0.6649120842199574,
      "grad_norm": 1.372299313545227,
      "learning_rate": 5.9340000000000004e-05,
      "loss": 0.4825,
      "step": 11874
    },
    {
      "epoch": 0.6649680815320864,
      "grad_norm": 2.188124895095825,
      "learning_rate": 5.934500000000001e-05,
      "loss": 0.5163,
      "step": 11875
    },
    {
      "epoch": 0.6650240788442154,
      "grad_norm": 1.4128270149230957,
      "learning_rate": 5.9350000000000006e-05,
      "loss": 0.4864,
      "step": 11876
    },
    {
      "epoch": 0.6650800761563445,
      "grad_norm": 1.202697992324829,
      "learning_rate": 5.9355000000000003e-05,
      "loss": 0.4417,
      "step": 11877
    },
    {
      "epoch": 0.6651360734684735,
      "grad_norm": 1.3172637224197388,
      "learning_rate": 5.936000000000001e-05,
      "loss": 0.4383,
      "step": 11878
    },
    {
      "epoch": 0.6651920707806025,
      "grad_norm": 1.202485203742981,
      "learning_rate": 5.9365000000000005e-05,
      "loss": 0.3311,
      "step": 11879
    },
    {
      "epoch": 0.6652480680927315,
      "grad_norm": 1.5256835222244263,
      "learning_rate": 5.937e-05,
      "loss": 0.3707,
      "step": 11880
    },
    {
      "epoch": 0.6653040654048605,
      "grad_norm": 1.5215630531311035,
      "learning_rate": 5.9375e-05,
      "loss": 0.4309,
      "step": 11881
    },
    {
      "epoch": 0.6653600627169896,
      "grad_norm": 1.437851071357727,
      "learning_rate": 5.9380000000000004e-05,
      "loss": 0.3928,
      "step": 11882
    },
    {
      "epoch": 0.6654160600291186,
      "grad_norm": 1.1500213146209717,
      "learning_rate": 5.9385e-05,
      "loss": 0.3762,
      "step": 11883
    },
    {
      "epoch": 0.6654720573412476,
      "grad_norm": 1.196663737297058,
      "learning_rate": 5.939e-05,
      "loss": 0.3458,
      "step": 11884
    },
    {
      "epoch": 0.6655280546533766,
      "grad_norm": 1.365906000137329,
      "learning_rate": 5.9395e-05,
      "loss": 0.4867,
      "step": 11885
    },
    {
      "epoch": 0.6655840519655056,
      "grad_norm": 1.3021512031555176,
      "learning_rate": 5.94e-05,
      "loss": 0.45,
      "step": 11886
    },
    {
      "epoch": 0.6656400492776346,
      "grad_norm": 1.5049903392791748,
      "learning_rate": 5.9405e-05,
      "loss": 0.4254,
      "step": 11887
    },
    {
      "epoch": 0.6656960465897637,
      "grad_norm": 1.4015634059906006,
      "learning_rate": 5.941e-05,
      "loss": 0.4781,
      "step": 11888
    },
    {
      "epoch": 0.6657520439018927,
      "grad_norm": 1.026158332824707,
      "learning_rate": 5.9415e-05,
      "loss": 0.3566,
      "step": 11889
    },
    {
      "epoch": 0.6658080412140217,
      "grad_norm": 1.1147710084915161,
      "learning_rate": 5.942e-05,
      "loss": 0.4144,
      "step": 11890
    },
    {
      "epoch": 0.6658640385261507,
      "grad_norm": 1.2159868478775024,
      "learning_rate": 5.9424999999999994e-05,
      "loss": 0.4443,
      "step": 11891
    },
    {
      "epoch": 0.6659200358382797,
      "grad_norm": 1.2928882837295532,
      "learning_rate": 5.9430000000000005e-05,
      "loss": 0.4125,
      "step": 11892
    },
    {
      "epoch": 0.6659760331504088,
      "grad_norm": 1.3846945762634277,
      "learning_rate": 5.943500000000001e-05,
      "loss": 0.4197,
      "step": 11893
    },
    {
      "epoch": 0.6660320304625378,
      "grad_norm": 1.3825461864471436,
      "learning_rate": 5.944000000000001e-05,
      "loss": 0.5739,
      "step": 11894
    },
    {
      "epoch": 0.6660880277746668,
      "grad_norm": 1.2266974449157715,
      "learning_rate": 5.9445000000000004e-05,
      "loss": 0.375,
      "step": 11895
    },
    {
      "epoch": 0.6661440250867958,
      "grad_norm": 1.3153057098388672,
      "learning_rate": 5.945000000000001e-05,
      "loss": 0.4659,
      "step": 11896
    },
    {
      "epoch": 0.6662000223989248,
      "grad_norm": 1.5280402898788452,
      "learning_rate": 5.9455000000000006e-05,
      "loss": 0.4301,
      "step": 11897
    },
    {
      "epoch": 0.6662560197110539,
      "grad_norm": 1.302307367324829,
      "learning_rate": 5.946e-05,
      "loss": 0.5529,
      "step": 11898
    },
    {
      "epoch": 0.6663120170231829,
      "grad_norm": 1.386218547821045,
      "learning_rate": 5.9465e-05,
      "loss": 0.4544,
      "step": 11899
    },
    {
      "epoch": 0.6663680143353119,
      "grad_norm": 1.0509182214736938,
      "learning_rate": 5.9470000000000005e-05,
      "loss": 0.3286,
      "step": 11900
    },
    {
      "epoch": 0.6664240116474409,
      "grad_norm": 1.423601508140564,
      "learning_rate": 5.9475e-05,
      "loss": 0.4119,
      "step": 11901
    },
    {
      "epoch": 0.6664800089595699,
      "grad_norm": 1.2592089176177979,
      "learning_rate": 5.948e-05,
      "loss": 0.6676,
      "step": 11902
    },
    {
      "epoch": 0.666536006271699,
      "grad_norm": 1.3460267782211304,
      "learning_rate": 5.9485000000000004e-05,
      "loss": 0.4699,
      "step": 11903
    },
    {
      "epoch": 0.666592003583828,
      "grad_norm": 1.334141731262207,
      "learning_rate": 5.949e-05,
      "loss": 0.5719,
      "step": 11904
    },
    {
      "epoch": 0.666648000895957,
      "grad_norm": 1.3642629384994507,
      "learning_rate": 5.9495e-05,
      "loss": 0.5086,
      "step": 11905
    },
    {
      "epoch": 0.666703998208086,
      "grad_norm": 1.1799554824829102,
      "learning_rate": 5.95e-05,
      "loss": 0.4032,
      "step": 11906
    },
    {
      "epoch": 0.666759995520215,
      "grad_norm": 1.4249459505081177,
      "learning_rate": 5.9505e-05,
      "loss": 0.4316,
      "step": 11907
    },
    {
      "epoch": 0.666815992832344,
      "grad_norm": 1.4522744417190552,
      "learning_rate": 5.951e-05,
      "loss": 0.4173,
      "step": 11908
    },
    {
      "epoch": 0.6668719901444731,
      "grad_norm": 1.1182037591934204,
      "learning_rate": 5.9514999999999995e-05,
      "loss": 0.3313,
      "step": 11909
    },
    {
      "epoch": 0.6669279874566021,
      "grad_norm": 1.2711259126663208,
      "learning_rate": 5.952e-05,
      "loss": 0.3685,
      "step": 11910
    },
    {
      "epoch": 0.6669839847687311,
      "grad_norm": 5.938239574432373,
      "learning_rate": 5.9525e-05,
      "loss": 0.4984,
      "step": 11911
    },
    {
      "epoch": 0.6670399820808601,
      "grad_norm": 1.4067604541778564,
      "learning_rate": 5.953000000000001e-05,
      "loss": 0.4658,
      "step": 11912
    },
    {
      "epoch": 0.6670959793929891,
      "grad_norm": 1.7628065347671509,
      "learning_rate": 5.9535000000000005e-05,
      "loss": 0.3645,
      "step": 11913
    },
    {
      "epoch": 0.6671519767051182,
      "grad_norm": 1.2518030405044556,
      "learning_rate": 5.954000000000001e-05,
      "loss": 0.4882,
      "step": 11914
    },
    {
      "epoch": 0.6672079740172472,
      "grad_norm": 1.567389965057373,
      "learning_rate": 5.954500000000001e-05,
      "loss": 0.4946,
      "step": 11915
    },
    {
      "epoch": 0.6672639713293762,
      "grad_norm": 1.301295280456543,
      "learning_rate": 5.9550000000000004e-05,
      "loss": 0.3501,
      "step": 11916
    },
    {
      "epoch": 0.6673199686415052,
      "grad_norm": 1.160940170288086,
      "learning_rate": 5.955500000000001e-05,
      "loss": 0.347,
      "step": 11917
    },
    {
      "epoch": 0.6673759659536342,
      "grad_norm": 1.2055591344833374,
      "learning_rate": 5.9560000000000006e-05,
      "loss": 0.3748,
      "step": 11918
    },
    {
      "epoch": 0.6674319632657633,
      "grad_norm": 1.1648057699203491,
      "learning_rate": 5.9565e-05,
      "loss": 0.3992,
      "step": 11919
    },
    {
      "epoch": 0.6674879605778923,
      "grad_norm": 1.2689573764801025,
      "learning_rate": 5.957e-05,
      "loss": 0.3886,
      "step": 11920
    },
    {
      "epoch": 0.6675439578900213,
      "grad_norm": 1.2126526832580566,
      "learning_rate": 5.9575000000000005e-05,
      "loss": 0.3961,
      "step": 11921
    },
    {
      "epoch": 0.6675999552021503,
      "grad_norm": 1.1302536725997925,
      "learning_rate": 5.958e-05,
      "loss": 0.3949,
      "step": 11922
    },
    {
      "epoch": 0.6676559525142793,
      "grad_norm": 1.5190685987472534,
      "learning_rate": 5.9585e-05,
      "loss": 0.63,
      "step": 11923
    },
    {
      "epoch": 0.6677119498264084,
      "grad_norm": 1.2824280261993408,
      "learning_rate": 5.9590000000000004e-05,
      "loss": 0.434,
      "step": 11924
    },
    {
      "epoch": 0.6677679471385374,
      "grad_norm": 1.348188042640686,
      "learning_rate": 5.9595e-05,
      "loss": 0.368,
      "step": 11925
    },
    {
      "epoch": 0.6678239444506664,
      "grad_norm": 1.1867979764938354,
      "learning_rate": 5.96e-05,
      "loss": 0.4094,
      "step": 11926
    },
    {
      "epoch": 0.6678799417627954,
      "grad_norm": 1.3989386558532715,
      "learning_rate": 5.9605e-05,
      "loss": 0.5733,
      "step": 11927
    },
    {
      "epoch": 0.6679359390749244,
      "grad_norm": 1.3534163236618042,
      "learning_rate": 5.961e-05,
      "loss": 0.4177,
      "step": 11928
    },
    {
      "epoch": 0.6679919363870535,
      "grad_norm": 1.5746853351593018,
      "learning_rate": 5.9615e-05,
      "loss": 0.4174,
      "step": 11929
    },
    {
      "epoch": 0.6680479336991825,
      "grad_norm": 1.2192744016647339,
      "learning_rate": 5.9619999999999995e-05,
      "loss": 0.4422,
      "step": 11930
    },
    {
      "epoch": 0.6681039310113115,
      "grad_norm": 2.8761403560638428,
      "learning_rate": 5.9625e-05,
      "loss": 0.5961,
      "step": 11931
    },
    {
      "epoch": 0.6681599283234405,
      "grad_norm": 1.376819372177124,
      "learning_rate": 5.963000000000001e-05,
      "loss": 0.5999,
      "step": 11932
    },
    {
      "epoch": 0.6682159256355695,
      "grad_norm": 1.293157696723938,
      "learning_rate": 5.963500000000001e-05,
      "loss": 0.3914,
      "step": 11933
    },
    {
      "epoch": 0.6682719229476985,
      "grad_norm": 1.537792444229126,
      "learning_rate": 5.9640000000000005e-05,
      "loss": 0.5856,
      "step": 11934
    },
    {
      "epoch": 0.6683279202598276,
      "grad_norm": 2.8495757579803467,
      "learning_rate": 5.964500000000001e-05,
      "loss": 0.788,
      "step": 11935
    },
    {
      "epoch": 0.6683839175719566,
      "grad_norm": 1.2567057609558105,
      "learning_rate": 5.9650000000000007e-05,
      "loss": 0.4586,
      "step": 11936
    },
    {
      "epoch": 0.6684399148840856,
      "grad_norm": 1.1390223503112793,
      "learning_rate": 5.9655000000000004e-05,
      "loss": 0.3774,
      "step": 11937
    },
    {
      "epoch": 0.6684959121962146,
      "grad_norm": 1.2039321660995483,
      "learning_rate": 5.966000000000001e-05,
      "loss": 0.3559,
      "step": 11938
    },
    {
      "epoch": 0.6685519095083436,
      "grad_norm": 1.622532606124878,
      "learning_rate": 5.9665000000000006e-05,
      "loss": 0.3729,
      "step": 11939
    },
    {
      "epoch": 0.6686079068204727,
      "grad_norm": 1.2902191877365112,
      "learning_rate": 5.967e-05,
      "loss": 0.4087,
      "step": 11940
    },
    {
      "epoch": 0.6686639041326017,
      "grad_norm": 1.5999029874801636,
      "learning_rate": 5.9675e-05,
      "loss": 0.4653,
      "step": 11941
    },
    {
      "epoch": 0.6687199014447307,
      "grad_norm": 1.2554268836975098,
      "learning_rate": 5.9680000000000005e-05,
      "loss": 0.4276,
      "step": 11942
    },
    {
      "epoch": 0.6687758987568597,
      "grad_norm": 1.4497487545013428,
      "learning_rate": 5.9685e-05,
      "loss": 0.4258,
      "step": 11943
    },
    {
      "epoch": 0.6688318960689887,
      "grad_norm": 1.1641937494277954,
      "learning_rate": 5.969e-05,
      "loss": 0.3416,
      "step": 11944
    },
    {
      "epoch": 0.6688878933811178,
      "grad_norm": 1.5143848657608032,
      "learning_rate": 5.9695000000000004e-05,
      "loss": 0.5738,
      "step": 11945
    },
    {
      "epoch": 0.6689438906932468,
      "grad_norm": 1.2914464473724365,
      "learning_rate": 5.97e-05,
      "loss": 0.3678,
      "step": 11946
    },
    {
      "epoch": 0.6689998880053757,
      "grad_norm": 1.261762022972107,
      "learning_rate": 5.9705e-05,
      "loss": 0.5169,
      "step": 11947
    },
    {
      "epoch": 0.6690558853175047,
      "grad_norm": 1.398228406906128,
      "learning_rate": 5.971e-05,
      "loss": 0.4794,
      "step": 11948
    },
    {
      "epoch": 0.6691118826296337,
      "grad_norm": 2.0928797721862793,
      "learning_rate": 5.9715e-05,
      "loss": 0.4342,
      "step": 11949
    },
    {
      "epoch": 0.6691678799417627,
      "grad_norm": 1.274298071861267,
      "learning_rate": 5.972e-05,
      "loss": 0.3518,
      "step": 11950
    },
    {
      "epoch": 0.6692238772538918,
      "grad_norm": 1.3640156984329224,
      "learning_rate": 5.9724999999999995e-05,
      "loss": 0.4479,
      "step": 11951
    },
    {
      "epoch": 0.6692798745660208,
      "grad_norm": 1.148178219795227,
      "learning_rate": 5.9730000000000006e-05,
      "loss": 0.3329,
      "step": 11952
    },
    {
      "epoch": 0.6693358718781498,
      "grad_norm": 1.2451313734054565,
      "learning_rate": 5.973500000000001e-05,
      "loss": 0.3687,
      "step": 11953
    },
    {
      "epoch": 0.6693918691902788,
      "grad_norm": 1.5776304006576538,
      "learning_rate": 5.974000000000001e-05,
      "loss": 0.3967,
      "step": 11954
    },
    {
      "epoch": 0.6694478665024078,
      "grad_norm": 1.358848214149475,
      "learning_rate": 5.9745000000000005e-05,
      "loss": 0.416,
      "step": 11955
    },
    {
      "epoch": 0.6695038638145369,
      "grad_norm": 1.2372437715530396,
      "learning_rate": 5.975000000000001e-05,
      "loss": 0.4196,
      "step": 11956
    },
    {
      "epoch": 0.6695598611266659,
      "grad_norm": 1.1792287826538086,
      "learning_rate": 5.9755000000000006e-05,
      "loss": 0.4522,
      "step": 11957
    },
    {
      "epoch": 0.6696158584387949,
      "grad_norm": 1.178227186203003,
      "learning_rate": 5.9760000000000004e-05,
      "loss": 0.3618,
      "step": 11958
    },
    {
      "epoch": 0.6696718557509239,
      "grad_norm": 1.2957943677902222,
      "learning_rate": 5.9765e-05,
      "loss": 0.5151,
      "step": 11959
    },
    {
      "epoch": 0.6697278530630529,
      "grad_norm": 1.2186528444290161,
      "learning_rate": 5.9770000000000005e-05,
      "loss": 0.5495,
      "step": 11960
    },
    {
      "epoch": 0.669783850375182,
      "grad_norm": 1.2566514015197754,
      "learning_rate": 5.9775e-05,
      "loss": 0.4227,
      "step": 11961
    },
    {
      "epoch": 0.669839847687311,
      "grad_norm": 1.2219884395599365,
      "learning_rate": 5.978e-05,
      "loss": 0.319,
      "step": 11962
    },
    {
      "epoch": 0.66989584499944,
      "grad_norm": 1.4038097858428955,
      "learning_rate": 5.9785000000000004e-05,
      "loss": 0.4801,
      "step": 11963
    },
    {
      "epoch": 0.669951842311569,
      "grad_norm": 1.5065792798995972,
      "learning_rate": 5.979e-05,
      "loss": 0.421,
      "step": 11964
    },
    {
      "epoch": 0.670007839623698,
      "grad_norm": 1.3518203496932983,
      "learning_rate": 5.9795e-05,
      "loss": 0.3793,
      "step": 11965
    },
    {
      "epoch": 0.670063836935827,
      "grad_norm": 1.4339953660964966,
      "learning_rate": 5.9800000000000003e-05,
      "loss": 0.4307,
      "step": 11966
    },
    {
      "epoch": 0.6701198342479561,
      "grad_norm": 1.2827121019363403,
      "learning_rate": 5.9805e-05,
      "loss": 0.3806,
      "step": 11967
    },
    {
      "epoch": 0.6701758315600851,
      "grad_norm": NaN,
      "learning_rate": 5.9805e-05,
      "loss": 0.4681,
      "step": 11968
    },
    {
      "epoch": 0.6702318288722141,
      "grad_norm": 1.5028961896896362,
      "learning_rate": 5.981e-05,
      "loss": 0.4211,
      "step": 11969
    },
    {
      "epoch": 0.6702878261843431,
      "grad_norm": 1.3298301696777344,
      "learning_rate": 5.9814999999999996e-05,
      "loss": 0.303,
      "step": 11970
    },
    {
      "epoch": 0.6703438234964721,
      "grad_norm": 1.2868156433105469,
      "learning_rate": 5.982e-05,
      "loss": 0.5618,
      "step": 11971
    },
    {
      "epoch": 0.6703998208086012,
      "grad_norm": 1.4557560682296753,
      "learning_rate": 5.9825e-05,
      "loss": 0.5228,
      "step": 11972
    },
    {
      "epoch": 0.6704558181207302,
      "grad_norm": 1.225768804550171,
      "learning_rate": 5.983000000000001e-05,
      "loss": 0.4383,
      "step": 11973
    },
    {
      "epoch": 0.6705118154328592,
      "grad_norm": 1.3572560548782349,
      "learning_rate": 5.9835000000000006e-05,
      "loss": 0.4428,
      "step": 11974
    },
    {
      "epoch": 0.6705678127449882,
      "grad_norm": 2.91058349609375,
      "learning_rate": 5.984000000000001e-05,
      "loss": 0.3326,
      "step": 11975
    },
    {
      "epoch": 0.6706238100571172,
      "grad_norm": 1.3695838451385498,
      "learning_rate": 5.984500000000001e-05,
      "loss": 0.3944,
      "step": 11976
    },
    {
      "epoch": 0.6706798073692463,
      "grad_norm": 1.2205719947814941,
      "learning_rate": 5.9850000000000005e-05,
      "loss": 0.3951,
      "step": 11977
    },
    {
      "epoch": 0.6707358046813753,
      "grad_norm": 1.076446294784546,
      "learning_rate": 5.985500000000001e-05,
      "loss": 0.3704,
      "step": 11978
    },
    {
      "epoch": 0.6707918019935043,
      "grad_norm": 1.4320250749588013,
      "learning_rate": 5.9860000000000006e-05,
      "loss": 0.5801,
      "step": 11979
    },
    {
      "epoch": 0.6708477993056333,
      "grad_norm": 1.3300652503967285,
      "learning_rate": 5.9865000000000004e-05,
      "loss": 0.3682,
      "step": 11980
    },
    {
      "epoch": 0.6709037966177623,
      "grad_norm": 1.949277400970459,
      "learning_rate": 5.987e-05,
      "loss": 0.4824,
      "step": 11981
    },
    {
      "epoch": 0.6709597939298914,
      "grad_norm": 1.1415592432022095,
      "learning_rate": 5.9875000000000005e-05,
      "loss": 0.4576,
      "step": 11982
    },
    {
      "epoch": 0.6710157912420204,
      "grad_norm": 1.2905715703964233,
      "learning_rate": 5.988e-05,
      "loss": 0.425,
      "step": 11983
    },
    {
      "epoch": 0.6710717885541494,
      "grad_norm": 1.4143505096435547,
      "learning_rate": 5.9885e-05,
      "loss": 0.4279,
      "step": 11984
    },
    {
      "epoch": 0.6711277858662784,
      "grad_norm": 1.546417474746704,
      "learning_rate": 5.9890000000000004e-05,
      "loss": 0.4881,
      "step": 11985
    },
    {
      "epoch": 0.6711837831784074,
      "grad_norm": 1.5658057928085327,
      "learning_rate": 5.9895e-05,
      "loss": 0.4668,
      "step": 11986
    },
    {
      "epoch": 0.6712397804905365,
      "grad_norm": 1.1546103954315186,
      "learning_rate": 5.99e-05,
      "loss": 0.4428,
      "step": 11987
    },
    {
      "epoch": 0.6712957778026655,
      "grad_norm": 1.3035434484481812,
      "learning_rate": 5.9905e-05,
      "loss": 0.4807,
      "step": 11988
    },
    {
      "epoch": 0.6713517751147945,
      "grad_norm": 1.1266698837280273,
      "learning_rate": 5.991e-05,
      "loss": 0.4117,
      "step": 11989
    },
    {
      "epoch": 0.6714077724269235,
      "grad_norm": 1.3537970781326294,
      "learning_rate": 5.9915e-05,
      "loss": 0.4936,
      "step": 11990
    },
    {
      "epoch": 0.6714637697390525,
      "grad_norm": 1.3205552101135254,
      "learning_rate": 5.9919999999999996e-05,
      "loss": 0.4109,
      "step": 11991
    },
    {
      "epoch": 0.6715197670511815,
      "grad_norm": 1.755386471748352,
      "learning_rate": 5.9925e-05,
      "loss": 0.5293,
      "step": 11992
    },
    {
      "epoch": 0.6715757643633106,
      "grad_norm": 1.450015902519226,
      "learning_rate": 5.993000000000001e-05,
      "loss": 0.4599,
      "step": 11993
    },
    {
      "epoch": 0.6716317616754396,
      "grad_norm": 1.3256274461746216,
      "learning_rate": 5.993500000000001e-05,
      "loss": 0.4843,
      "step": 11994
    },
    {
      "epoch": 0.6716877589875686,
      "grad_norm": 1.287933349609375,
      "learning_rate": 5.9940000000000005e-05,
      "loss": 0.36,
      "step": 11995
    },
    {
      "epoch": 0.6717437562996976,
      "grad_norm": 1.2362840175628662,
      "learning_rate": 5.994500000000001e-05,
      "loss": 0.5034,
      "step": 11996
    },
    {
      "epoch": 0.6717997536118266,
      "grad_norm": 1.6679575443267822,
      "learning_rate": 5.995000000000001e-05,
      "loss": 0.4632,
      "step": 11997
    },
    {
      "epoch": 0.6718557509239557,
      "grad_norm": 1.4457893371582031,
      "learning_rate": 5.9955000000000004e-05,
      "loss": 0.6138,
      "step": 11998
    },
    {
      "epoch": 0.6719117482360847,
      "grad_norm": 1.1589275598526,
      "learning_rate": 5.996e-05,
      "loss": 0.3804,
      "step": 11999
    },
    {
      "epoch": 0.6719677455482137,
      "grad_norm": 1.1779927015304565,
      "learning_rate": 5.9965000000000006e-05,
      "loss": 0.4292,
      "step": 12000
    },
    {
      "epoch": 0.6720237428603427,
      "grad_norm": 1.5580350160598755,
      "learning_rate": 5.9970000000000004e-05,
      "loss": 0.3699,
      "step": 12001
    },
    {
      "epoch": 0.6720797401724717,
      "grad_norm": 1.4929412603378296,
      "learning_rate": 5.9975e-05,
      "loss": 0.4857,
      "step": 12002
    },
    {
      "epoch": 0.6721357374846008,
      "grad_norm": 1.3599839210510254,
      "learning_rate": 5.9980000000000005e-05,
      "loss": 0.5874,
      "step": 12003
    },
    {
      "epoch": 0.6721917347967298,
      "grad_norm": 1.2680851221084595,
      "learning_rate": 5.9985e-05,
      "loss": 0.4742,
      "step": 12004
    },
    {
      "epoch": 0.6722477321088588,
      "grad_norm": 1.1076149940490723,
      "learning_rate": 5.999e-05,
      "loss": 0.3154,
      "step": 12005
    },
    {
      "epoch": 0.6723037294209878,
      "grad_norm": 1.2600069046020508,
      "learning_rate": 5.9995000000000004e-05,
      "loss": 0.342,
      "step": 12006
    },
    {
      "epoch": 0.6723597267331168,
      "grad_norm": 1.2424794435501099,
      "learning_rate": 6e-05,
      "loss": 0.4882,
      "step": 12007
    },
    {
      "epoch": 0.6724157240452459,
      "grad_norm": 1.4509917497634888,
      "learning_rate": 6.0005e-05,
      "loss": 0.3048,
      "step": 12008
    },
    {
      "epoch": 0.6724717213573749,
      "grad_norm": 1.324674367904663,
      "learning_rate": 6.0009999999999996e-05,
      "loss": 0.395,
      "step": 12009
    },
    {
      "epoch": 0.6725277186695039,
      "grad_norm": 1.259440541267395,
      "learning_rate": 6.0015e-05,
      "loss": 0.4889,
      "step": 12010
    },
    {
      "epoch": 0.6725837159816329,
      "grad_norm": 1.3536385297775269,
      "learning_rate": 6.002e-05,
      "loss": 0.498,
      "step": 12011
    },
    {
      "epoch": 0.6726397132937619,
      "grad_norm": 1.317745566368103,
      "learning_rate": 6.0024999999999995e-05,
      "loss": 0.4705,
      "step": 12012
    },
    {
      "epoch": 0.672695710605891,
      "grad_norm": 1.2811329364776611,
      "learning_rate": 6.003e-05,
      "loss": 0.4175,
      "step": 12013
    },
    {
      "epoch": 0.67275170791802,
      "grad_norm": 1.3161181211471558,
      "learning_rate": 6.003500000000001e-05,
      "loss": 0.5964,
      "step": 12014
    },
    {
      "epoch": 0.672807705230149,
      "grad_norm": 1.2640495300292969,
      "learning_rate": 6.004000000000001e-05,
      "loss": 0.4612,
      "step": 12015
    },
    {
      "epoch": 0.672863702542278,
      "grad_norm": 1.3181904554367065,
      "learning_rate": 6.0045000000000005e-05,
      "loss": 0.3851,
      "step": 12016
    },
    {
      "epoch": 0.672919699854407,
      "grad_norm": 1.3061329126358032,
      "learning_rate": 6.005000000000001e-05,
      "loss": 0.5407,
      "step": 12017
    },
    {
      "epoch": 0.672975697166536,
      "grad_norm": 1.2655748128890991,
      "learning_rate": 6.005500000000001e-05,
      "loss": 0.3327,
      "step": 12018
    },
    {
      "epoch": 0.6730316944786651,
      "grad_norm": 1.140716552734375,
      "learning_rate": 6.0060000000000004e-05,
      "loss": 0.4395,
      "step": 12019
    },
    {
      "epoch": 0.6730876917907941,
      "grad_norm": 1.422457218170166,
      "learning_rate": 6.0065e-05,
      "loss": 0.6095,
      "step": 12020
    },
    {
      "epoch": 0.6731436891029231,
      "grad_norm": 1.1209276914596558,
      "learning_rate": 6.0070000000000006e-05,
      "loss": 0.3438,
      "step": 12021
    },
    {
      "epoch": 0.6731996864150521,
      "grad_norm": 1.7475590705871582,
      "learning_rate": 6.0075e-05,
      "loss": 0.5939,
      "step": 12022
    },
    {
      "epoch": 0.6732556837271811,
      "grad_norm": 1.2284224033355713,
      "learning_rate": 6.008e-05,
      "loss": 0.3576,
      "step": 12023
    },
    {
      "epoch": 0.6733116810393102,
      "grad_norm": 1.2152773141860962,
      "learning_rate": 6.0085000000000005e-05,
      "loss": 0.432,
      "step": 12024
    },
    {
      "epoch": 0.6733676783514392,
      "grad_norm": 1.2505340576171875,
      "learning_rate": 6.009e-05,
      "loss": 0.3708,
      "step": 12025
    },
    {
      "epoch": 0.6734236756635682,
      "grad_norm": 1.2982544898986816,
      "learning_rate": 6.0095e-05,
      "loss": 0.5278,
      "step": 12026
    },
    {
      "epoch": 0.6734796729756972,
      "grad_norm": 1.1628079414367676,
      "learning_rate": 6.0100000000000004e-05,
      "loss": 0.3659,
      "step": 12027
    },
    {
      "epoch": 0.6735356702878262,
      "grad_norm": 1.8002924919128418,
      "learning_rate": 6.0105e-05,
      "loss": 0.8369,
      "step": 12028
    },
    {
      "epoch": 0.6735916675999553,
      "grad_norm": 1.223206877708435,
      "learning_rate": 6.011e-05,
      "loss": 0.5043,
      "step": 12029
    },
    {
      "epoch": 0.6736476649120842,
      "grad_norm": 1.3584004640579224,
      "learning_rate": 6.0114999999999996e-05,
      "loss": 0.4154,
      "step": 12030
    },
    {
      "epoch": 0.6737036622242132,
      "grad_norm": 1.5717308521270752,
      "learning_rate": 6.012e-05,
      "loss": 0.4392,
      "step": 12031
    },
    {
      "epoch": 0.6737596595363422,
      "grad_norm": 1.321874737739563,
      "learning_rate": 6.0125e-05,
      "loss": 0.3455,
      "step": 12032
    },
    {
      "epoch": 0.6738156568484712,
      "grad_norm": 1.0951192378997803,
      "learning_rate": 6.0129999999999995e-05,
      "loss": 0.3107,
      "step": 12033
    },
    {
      "epoch": 0.6738716541606002,
      "grad_norm": 1.5254805088043213,
      "learning_rate": 6.0135000000000006e-05,
      "loss": 0.3985,
      "step": 12034
    },
    {
      "epoch": 0.6739276514727293,
      "grad_norm": 1.4667788743972778,
      "learning_rate": 6.014000000000001e-05,
      "loss": 0.4162,
      "step": 12035
    },
    {
      "epoch": 0.6739836487848583,
      "grad_norm": 1.4369940757751465,
      "learning_rate": 6.014500000000001e-05,
      "loss": 0.6691,
      "step": 12036
    },
    {
      "epoch": 0.6740396460969873,
      "grad_norm": 1.646856427192688,
      "learning_rate": 6.0150000000000005e-05,
      "loss": 0.392,
      "step": 12037
    },
    {
      "epoch": 0.6740956434091163,
      "grad_norm": 1.3220666646957397,
      "learning_rate": 6.0155e-05,
      "loss": 0.4313,
      "step": 12038
    },
    {
      "epoch": 0.6741516407212453,
      "grad_norm": 1.4972738027572632,
      "learning_rate": 6.016000000000001e-05,
      "loss": 0.492,
      "step": 12039
    },
    {
      "epoch": 0.6742076380333744,
      "grad_norm": 1.316117763519287,
      "learning_rate": 6.0165000000000004e-05,
      "loss": 0.5395,
      "step": 12040
    },
    {
      "epoch": 0.6742636353455034,
      "grad_norm": 1.4296866655349731,
      "learning_rate": 6.017e-05,
      "loss": 0.4829,
      "step": 12041
    },
    {
      "epoch": 0.6743196326576324,
      "grad_norm": 1.4425750970840454,
      "learning_rate": 6.0175000000000006e-05,
      "loss": 0.3922,
      "step": 12042
    },
    {
      "epoch": 0.6743756299697614,
      "grad_norm": 1.1359617710113525,
      "learning_rate": 6.018e-05,
      "loss": 0.4163,
      "step": 12043
    },
    {
      "epoch": 0.6744316272818904,
      "grad_norm": 1.3443559408187866,
      "learning_rate": 6.0185e-05,
      "loss": 0.4645,
      "step": 12044
    },
    {
      "epoch": 0.6744876245940195,
      "grad_norm": 1.6530473232269287,
      "learning_rate": 6.0190000000000005e-05,
      "loss": 0.4853,
      "step": 12045
    },
    {
      "epoch": 0.6745436219061485,
      "grad_norm": 1.2223633527755737,
      "learning_rate": 6.0195e-05,
      "loss": 0.4603,
      "step": 12046
    },
    {
      "epoch": 0.6745996192182775,
      "grad_norm": 1.292128086090088,
      "learning_rate": 6.02e-05,
      "loss": 0.4506,
      "step": 12047
    },
    {
      "epoch": 0.6746556165304065,
      "grad_norm": 1.9163436889648438,
      "learning_rate": 6.0205e-05,
      "loss": 0.6832,
      "step": 12048
    },
    {
      "epoch": 0.6747116138425355,
      "grad_norm": 1.1957800388336182,
      "learning_rate": 6.021e-05,
      "loss": 0.327,
      "step": 12049
    },
    {
      "epoch": 0.6747676111546645,
      "grad_norm": 1.2962968349456787,
      "learning_rate": 6.0215e-05,
      "loss": 0.5793,
      "step": 12050
    },
    {
      "epoch": 0.6748236084667936,
      "grad_norm": 1.2703187465667725,
      "learning_rate": 6.0219999999999996e-05,
      "loss": 0.3853,
      "step": 12051
    },
    {
      "epoch": 0.6748796057789226,
      "grad_norm": 1.4209612607955933,
      "learning_rate": 6.0225e-05,
      "loss": 0.5524,
      "step": 12052
    },
    {
      "epoch": 0.6749356030910516,
      "grad_norm": 1.231265902519226,
      "learning_rate": 6.023e-05,
      "loss": 0.5274,
      "step": 12053
    },
    {
      "epoch": 0.6749916004031806,
      "grad_norm": 1.2571219205856323,
      "learning_rate": 6.023500000000001e-05,
      "loss": 0.422,
      "step": 12054
    },
    {
      "epoch": 0.6750475977153096,
      "grad_norm": 1.330564022064209,
      "learning_rate": 6.0240000000000006e-05,
      "loss": 0.4988,
      "step": 12055
    },
    {
      "epoch": 0.6751035950274387,
      "grad_norm": 1.2424752712249756,
      "learning_rate": 6.024500000000001e-05,
      "loss": 0.5119,
      "step": 12056
    },
    {
      "epoch": 0.6751595923395677,
      "grad_norm": 1.1252355575561523,
      "learning_rate": 6.025000000000001e-05,
      "loss": 0.3737,
      "step": 12057
    },
    {
      "epoch": 0.6752155896516967,
      "grad_norm": 1.6888601779937744,
      "learning_rate": 6.0255000000000005e-05,
      "loss": 0.5231,
      "step": 12058
    },
    {
      "epoch": 0.6752715869638257,
      "grad_norm": 1.2207615375518799,
      "learning_rate": 6.026e-05,
      "loss": 0.3587,
      "step": 12059
    },
    {
      "epoch": 0.6753275842759547,
      "grad_norm": 1.3332748413085938,
      "learning_rate": 6.0265000000000007e-05,
      "loss": 0.4438,
      "step": 12060
    },
    {
      "epoch": 0.6753835815880838,
      "grad_norm": 1.7057472467422485,
      "learning_rate": 6.0270000000000004e-05,
      "loss": 0.6373,
      "step": 12061
    },
    {
      "epoch": 0.6754395789002128,
      "grad_norm": 1.2983251810073853,
      "learning_rate": 6.0275e-05,
      "loss": 0.5587,
      "step": 12062
    },
    {
      "epoch": 0.6754955762123418,
      "grad_norm": 1.5316787958145142,
      "learning_rate": 6.0280000000000006e-05,
      "loss": 0.4676,
      "step": 12063
    },
    {
      "epoch": 0.6755515735244708,
      "grad_norm": 1.215063214302063,
      "learning_rate": 6.0285e-05,
      "loss": 0.4541,
      "step": 12064
    },
    {
      "epoch": 0.6756075708365998,
      "grad_norm": 1.2608975172042847,
      "learning_rate": 6.029e-05,
      "loss": 0.5397,
      "step": 12065
    },
    {
      "epoch": 0.6756635681487289,
      "grad_norm": 1.618356704711914,
      "learning_rate": 6.0295000000000005e-05,
      "loss": 0.5113,
      "step": 12066
    },
    {
      "epoch": 0.6757195654608579,
      "grad_norm": 1.4438884258270264,
      "learning_rate": 6.03e-05,
      "loss": 0.4424,
      "step": 12067
    },
    {
      "epoch": 0.6757755627729869,
      "grad_norm": 1.1770647764205933,
      "learning_rate": 6.0305e-05,
      "loss": 0.3894,
      "step": 12068
    },
    {
      "epoch": 0.6758315600851159,
      "grad_norm": 1.245314121246338,
      "learning_rate": 6.031e-05,
      "loss": 0.4122,
      "step": 12069
    },
    {
      "epoch": 0.6758875573972449,
      "grad_norm": 1.1871123313903809,
      "learning_rate": 6.0315e-05,
      "loss": 0.4421,
      "step": 12070
    },
    {
      "epoch": 0.675943554709374,
      "grad_norm": 1.7059030532836914,
      "learning_rate": 6.032e-05,
      "loss": 0.4171,
      "step": 12071
    },
    {
      "epoch": 0.675999552021503,
      "grad_norm": 1.3483079671859741,
      "learning_rate": 6.0324999999999996e-05,
      "loss": 0.4056,
      "step": 12072
    },
    {
      "epoch": 0.676055549333632,
      "grad_norm": 1.364949345588684,
      "learning_rate": 6.033e-05,
      "loss": 0.3944,
      "step": 12073
    },
    {
      "epoch": 0.676111546645761,
      "grad_norm": 1.3981837034225464,
      "learning_rate": 6.033500000000001e-05,
      "loss": 0.398,
      "step": 12074
    },
    {
      "epoch": 0.67616754395789,
      "grad_norm": 1.3125524520874023,
      "learning_rate": 6.034000000000001e-05,
      "loss": 0.4499,
      "step": 12075
    },
    {
      "epoch": 0.676223541270019,
      "grad_norm": 1.201357364654541,
      "learning_rate": 6.0345000000000006e-05,
      "loss": 0.4868,
      "step": 12076
    },
    {
      "epoch": 0.6762795385821481,
      "grad_norm": 1.6828097105026245,
      "learning_rate": 6.035e-05,
      "loss": 0.5972,
      "step": 12077
    },
    {
      "epoch": 0.6763355358942771,
      "grad_norm": 1.5219569206237793,
      "learning_rate": 6.035500000000001e-05,
      "loss": 0.4371,
      "step": 12078
    },
    {
      "epoch": 0.6763915332064061,
      "grad_norm": 1.4877394437789917,
      "learning_rate": 6.0360000000000005e-05,
      "loss": 0.4413,
      "step": 12079
    },
    {
      "epoch": 0.6764475305185351,
      "grad_norm": 1.6040446758270264,
      "learning_rate": 6.0365e-05,
      "loss": 0.4621,
      "step": 12080
    },
    {
      "epoch": 0.6765035278306641,
      "grad_norm": 1.199930191040039,
      "learning_rate": 6.0370000000000006e-05,
      "loss": 0.4529,
      "step": 12081
    },
    {
      "epoch": 0.6765595251427932,
      "grad_norm": 1.2376455068588257,
      "learning_rate": 6.0375000000000004e-05,
      "loss": 0.4784,
      "step": 12082
    },
    {
      "epoch": 0.6766155224549222,
      "grad_norm": 1.2274621725082397,
      "learning_rate": 6.038e-05,
      "loss": 0.3986,
      "step": 12083
    },
    {
      "epoch": 0.6766715197670512,
      "grad_norm": 1.4890775680541992,
      "learning_rate": 6.0385000000000005e-05,
      "loss": 0.5813,
      "step": 12084
    },
    {
      "epoch": 0.6767275170791802,
      "grad_norm": 1.232393741607666,
      "learning_rate": 6.039e-05,
      "loss": 0.4071,
      "step": 12085
    },
    {
      "epoch": 0.6767835143913092,
      "grad_norm": 1.3618180751800537,
      "learning_rate": 6.0395e-05,
      "loss": 0.4631,
      "step": 12086
    },
    {
      "epoch": 0.6768395117034383,
      "grad_norm": 1.5815465450286865,
      "learning_rate": 6.04e-05,
      "loss": 0.6354,
      "step": 12087
    },
    {
      "epoch": 0.6768955090155673,
      "grad_norm": 1.2645756006240845,
      "learning_rate": 6.0405e-05,
      "loss": 0.427,
      "step": 12088
    },
    {
      "epoch": 0.6769515063276963,
      "grad_norm": 1.0445808172225952,
      "learning_rate": 6.041e-05,
      "loss": 0.332,
      "step": 12089
    },
    {
      "epoch": 0.6770075036398253,
      "grad_norm": 1.3043770790100098,
      "learning_rate": 6.0415e-05,
      "loss": 0.6544,
      "step": 12090
    },
    {
      "epoch": 0.6770635009519543,
      "grad_norm": 1.4303696155548096,
      "learning_rate": 6.042e-05,
      "loss": 0.4022,
      "step": 12091
    },
    {
      "epoch": 0.6771194982640834,
      "grad_norm": 1.2758795022964478,
      "learning_rate": 6.0425e-05,
      "loss": 0.4598,
      "step": 12092
    },
    {
      "epoch": 0.6771754955762124,
      "grad_norm": 1.3888673782348633,
      "learning_rate": 6.0429999999999996e-05,
      "loss": 0.4979,
      "step": 12093
    },
    {
      "epoch": 0.6772314928883414,
      "grad_norm": 1.3244187831878662,
      "learning_rate": 6.043500000000001e-05,
      "loss": 0.4616,
      "step": 12094
    },
    {
      "epoch": 0.6772874902004704,
      "grad_norm": 1.1622129678726196,
      "learning_rate": 6.044000000000001e-05,
      "loss": 0.4378,
      "step": 12095
    },
    {
      "epoch": 0.6773434875125994,
      "grad_norm": 1.3151376247406006,
      "learning_rate": 6.044500000000001e-05,
      "loss": 0.3957,
      "step": 12096
    },
    {
      "epoch": 0.6773994848247284,
      "grad_norm": 4.065365314483643,
      "learning_rate": 6.0450000000000006e-05,
      "loss": 0.4165,
      "step": 12097
    },
    {
      "epoch": 0.6774554821368575,
      "grad_norm": 1.9255039691925049,
      "learning_rate": 6.0455e-05,
      "loss": 0.5765,
      "step": 12098
    },
    {
      "epoch": 0.6775114794489865,
      "grad_norm": 1.1353352069854736,
      "learning_rate": 6.046000000000001e-05,
      "loss": 0.507,
      "step": 12099
    },
    {
      "epoch": 0.6775674767611155,
      "grad_norm": 1.3141318559646606,
      "learning_rate": 6.0465000000000005e-05,
      "loss": 0.4241,
      "step": 12100
    },
    {
      "epoch": 0.6776234740732445,
      "grad_norm": 1.657446265220642,
      "learning_rate": 6.047e-05,
      "loss": 0.4933,
      "step": 12101
    },
    {
      "epoch": 0.6776794713853735,
      "grad_norm": 1.1255770921707153,
      "learning_rate": 6.0475000000000006e-05,
      "loss": 0.3442,
      "step": 12102
    },
    {
      "epoch": 0.6777354686975026,
      "grad_norm": 1.3445061445236206,
      "learning_rate": 6.0480000000000004e-05,
      "loss": 0.4012,
      "step": 12103
    },
    {
      "epoch": 0.6777914660096316,
      "grad_norm": 1.5071191787719727,
      "learning_rate": 6.0485e-05,
      "loss": 0.431,
      "step": 12104
    },
    {
      "epoch": 0.6778474633217606,
      "grad_norm": 1.302628517150879,
      "learning_rate": 6.0490000000000005e-05,
      "loss": 0.3635,
      "step": 12105
    },
    {
      "epoch": 0.6779034606338896,
      "grad_norm": 1.3854845762252808,
      "learning_rate": 6.0495e-05,
      "loss": 0.5189,
      "step": 12106
    },
    {
      "epoch": 0.6779594579460186,
      "grad_norm": 1.2401306629180908,
      "learning_rate": 6.05e-05,
      "loss": 0.4829,
      "step": 12107
    },
    {
      "epoch": 0.6780154552581477,
      "grad_norm": 1.5215063095092773,
      "learning_rate": 6.0505e-05,
      "loss": 0.5669,
      "step": 12108
    },
    {
      "epoch": 0.6780714525702767,
      "grad_norm": 1.374367594718933,
      "learning_rate": 6.051e-05,
      "loss": 0.5544,
      "step": 12109
    },
    {
      "epoch": 0.6781274498824057,
      "grad_norm": 1.4461313486099243,
      "learning_rate": 6.0515e-05,
      "loss": 0.5488,
      "step": 12110
    },
    {
      "epoch": 0.6781834471945347,
      "grad_norm": 1.4633358716964722,
      "learning_rate": 6.0519999999999997e-05,
      "loss": 0.594,
      "step": 12111
    },
    {
      "epoch": 0.6782394445066637,
      "grad_norm": 1.371083378791809,
      "learning_rate": 6.0525e-05,
      "loss": 0.4453,
      "step": 12112
    },
    {
      "epoch": 0.6782954418187926,
      "grad_norm": 1.2095295190811157,
      "learning_rate": 6.053e-05,
      "loss": 0.34,
      "step": 12113
    },
    {
      "epoch": 0.6783514391309217,
      "grad_norm": 1.2679212093353271,
      "learning_rate": 6.053500000000001e-05,
      "loss": 0.4288,
      "step": 12114
    },
    {
      "epoch": 0.6784074364430507,
      "grad_norm": 1.5355103015899658,
      "learning_rate": 6.0540000000000007e-05,
      "loss": 0.5017,
      "step": 12115
    },
    {
      "epoch": 0.6784634337551797,
      "grad_norm": 1.0712206363677979,
      "learning_rate": 6.0545000000000004e-05,
      "loss": 0.3312,
      "step": 12116
    },
    {
      "epoch": 0.6785194310673087,
      "grad_norm": 1.1691625118255615,
      "learning_rate": 6.055000000000001e-05,
      "loss": 0.4556,
      "step": 12117
    },
    {
      "epoch": 0.6785754283794377,
      "grad_norm": 1.2647243738174438,
      "learning_rate": 6.0555000000000006e-05,
      "loss": 0.3252,
      "step": 12118
    },
    {
      "epoch": 0.6786314256915668,
      "grad_norm": 1.303964614868164,
      "learning_rate": 6.056e-05,
      "loss": 0.5992,
      "step": 12119
    },
    {
      "epoch": 0.6786874230036958,
      "grad_norm": 1.539599895477295,
      "learning_rate": 6.056500000000001e-05,
      "loss": 0.5647,
      "step": 12120
    },
    {
      "epoch": 0.6787434203158248,
      "grad_norm": 1.6583836078643799,
      "learning_rate": 6.0570000000000005e-05,
      "loss": 0.5323,
      "step": 12121
    },
    {
      "epoch": 0.6787994176279538,
      "grad_norm": 1.4732065200805664,
      "learning_rate": 6.0575e-05,
      "loss": 0.4887,
      "step": 12122
    },
    {
      "epoch": 0.6788554149400828,
      "grad_norm": 1.8859001398086548,
      "learning_rate": 6.0580000000000006e-05,
      "loss": 0.6046,
      "step": 12123
    },
    {
      "epoch": 0.6789114122522119,
      "grad_norm": 1.1522440910339355,
      "learning_rate": 6.0585000000000004e-05,
      "loss": 0.385,
      "step": 12124
    },
    {
      "epoch": 0.6789674095643409,
      "grad_norm": 1.0091975927352905,
      "learning_rate": 6.059e-05,
      "loss": 0.3164,
      "step": 12125
    },
    {
      "epoch": 0.6790234068764699,
      "grad_norm": 1.124294638633728,
      "learning_rate": 6.0595000000000005e-05,
      "loss": 0.3647,
      "step": 12126
    },
    {
      "epoch": 0.6790794041885989,
      "grad_norm": 1.23323392868042,
      "learning_rate": 6.06e-05,
      "loss": 0.4144,
      "step": 12127
    },
    {
      "epoch": 0.6791354015007279,
      "grad_norm": 1.3646094799041748,
      "learning_rate": 6.0605e-05,
      "loss": 0.7289,
      "step": 12128
    },
    {
      "epoch": 0.679191398812857,
      "grad_norm": 1.2699626684188843,
      "learning_rate": 6.061e-05,
      "loss": 0.429,
      "step": 12129
    },
    {
      "epoch": 0.679247396124986,
      "grad_norm": 1.2049400806427002,
      "learning_rate": 6.0615e-05,
      "loss": 0.4621,
      "step": 12130
    },
    {
      "epoch": 0.679303393437115,
      "grad_norm": 1.487858533859253,
      "learning_rate": 6.062e-05,
      "loss": 0.4598,
      "step": 12131
    },
    {
      "epoch": 0.679359390749244,
      "grad_norm": 1.4971121549606323,
      "learning_rate": 6.0624999999999996e-05,
      "loss": 0.508,
      "step": 12132
    },
    {
      "epoch": 0.679415388061373,
      "grad_norm": 1.1997946500778198,
      "learning_rate": 6.063e-05,
      "loss": 0.4751,
      "step": 12133
    },
    {
      "epoch": 0.679471385373502,
      "grad_norm": 1.2344303131103516,
      "learning_rate": 6.0635e-05,
      "loss": 0.4005,
      "step": 12134
    },
    {
      "epoch": 0.6795273826856311,
      "grad_norm": 1.1131056547164917,
      "learning_rate": 6.064000000000001e-05,
      "loss": 0.3827,
      "step": 12135
    },
    {
      "epoch": 0.6795833799977601,
      "grad_norm": 1.1705071926116943,
      "learning_rate": 6.0645000000000006e-05,
      "loss": 0.3733,
      "step": 12136
    },
    {
      "epoch": 0.6796393773098891,
      "grad_norm": 1.2471826076507568,
      "learning_rate": 6.0650000000000004e-05,
      "loss": 0.4472,
      "step": 12137
    },
    {
      "epoch": 0.6796953746220181,
      "grad_norm": 1.3577433824539185,
      "learning_rate": 6.065500000000001e-05,
      "loss": 0.4631,
      "step": 12138
    },
    {
      "epoch": 0.6797513719341471,
      "grad_norm": 1.1097514629364014,
      "learning_rate": 6.0660000000000005e-05,
      "loss": 0.3606,
      "step": 12139
    },
    {
      "epoch": 0.6798073692462762,
      "grad_norm": 1.6002696752548218,
      "learning_rate": 6.0665e-05,
      "loss": 0.4334,
      "step": 12140
    },
    {
      "epoch": 0.6798633665584052,
      "grad_norm": 1.2866092920303345,
      "learning_rate": 6.067000000000001e-05,
      "loss": 0.4569,
      "step": 12141
    },
    {
      "epoch": 0.6799193638705342,
      "grad_norm": 1.1357052326202393,
      "learning_rate": 6.0675000000000004e-05,
      "loss": 0.4529,
      "step": 12142
    },
    {
      "epoch": 0.6799753611826632,
      "grad_norm": 1.271215796470642,
      "learning_rate": 6.068e-05,
      "loss": 0.3355,
      "step": 12143
    },
    {
      "epoch": 0.6800313584947922,
      "grad_norm": 1.1626436710357666,
      "learning_rate": 6.0685000000000006e-05,
      "loss": 0.4175,
      "step": 12144
    },
    {
      "epoch": 0.6800873558069213,
      "grad_norm": 1.1848807334899902,
      "learning_rate": 6.069e-05,
      "loss": 0.4359,
      "step": 12145
    },
    {
      "epoch": 0.6801433531190503,
      "grad_norm": 1.495087742805481,
      "learning_rate": 6.0695e-05,
      "loss": 0.4236,
      "step": 12146
    },
    {
      "epoch": 0.6801993504311793,
      "grad_norm": 1.203644037246704,
      "learning_rate": 6.07e-05,
      "loss": 0.4306,
      "step": 12147
    },
    {
      "epoch": 0.6802553477433083,
      "grad_norm": 1.1470236778259277,
      "learning_rate": 6.0705e-05,
      "loss": 0.4808,
      "step": 12148
    },
    {
      "epoch": 0.6803113450554373,
      "grad_norm": 1.2754034996032715,
      "learning_rate": 6.071e-05,
      "loss": 0.496,
      "step": 12149
    },
    {
      "epoch": 0.6803673423675664,
      "grad_norm": 1.3479679822921753,
      "learning_rate": 6.0715e-05,
      "loss": 0.6041,
      "step": 12150
    },
    {
      "epoch": 0.6804233396796954,
      "grad_norm": 1.2230275869369507,
      "learning_rate": 6.072e-05,
      "loss": 0.3986,
      "step": 12151
    },
    {
      "epoch": 0.6804793369918244,
      "grad_norm": 1.4950847625732422,
      "learning_rate": 6.0725e-05,
      "loss": 0.4485,
      "step": 12152
    },
    {
      "epoch": 0.6805353343039534,
      "grad_norm": 1.4482049942016602,
      "learning_rate": 6.0729999999999996e-05,
      "loss": 0.4365,
      "step": 12153
    },
    {
      "epoch": 0.6805913316160824,
      "grad_norm": 1.2511510848999023,
      "learning_rate": 6.0735e-05,
      "loss": 0.5161,
      "step": 12154
    },
    {
      "epoch": 0.6806473289282114,
      "grad_norm": 1.4707467555999756,
      "learning_rate": 6.074000000000001e-05,
      "loss": 0.4518,
      "step": 12155
    },
    {
      "epoch": 0.6807033262403405,
      "grad_norm": 1.0941394567489624,
      "learning_rate": 6.074500000000001e-05,
      "loss": 0.3309,
      "step": 12156
    },
    {
      "epoch": 0.6807593235524695,
      "grad_norm": 1.1759692430496216,
      "learning_rate": 6.0750000000000006e-05,
      "loss": 0.3841,
      "step": 12157
    },
    {
      "epoch": 0.6808153208645985,
      "grad_norm": 1.0564043521881104,
      "learning_rate": 6.0755000000000004e-05,
      "loss": 0.4067,
      "step": 12158
    },
    {
      "epoch": 0.6808713181767275,
      "grad_norm": 1.4290287494659424,
      "learning_rate": 6.076000000000001e-05,
      "loss": 0.4944,
      "step": 12159
    },
    {
      "epoch": 0.6809273154888565,
      "grad_norm": 1.222448468208313,
      "learning_rate": 6.0765000000000005e-05,
      "loss": 0.3964,
      "step": 12160
    },
    {
      "epoch": 0.6809833128009856,
      "grad_norm": 1.0129196643829346,
      "learning_rate": 6.077e-05,
      "loss": 0.4556,
      "step": 12161
    },
    {
      "epoch": 0.6810393101131146,
      "grad_norm": 1.4554996490478516,
      "learning_rate": 6.077500000000001e-05,
      "loss": 0.3385,
      "step": 12162
    },
    {
      "epoch": 0.6810953074252436,
      "grad_norm": 1.183143973350525,
      "learning_rate": 6.0780000000000004e-05,
      "loss": 0.4086,
      "step": 12163
    },
    {
      "epoch": 0.6811513047373726,
      "grad_norm": 1.2586287260055542,
      "learning_rate": 6.0785e-05,
      "loss": 0.3804,
      "step": 12164
    },
    {
      "epoch": 0.6812073020495016,
      "grad_norm": 1.4544435739517212,
      "learning_rate": 6.0790000000000006e-05,
      "loss": 0.3895,
      "step": 12165
    },
    {
      "epoch": 0.6812632993616307,
      "grad_norm": 1.2856827974319458,
      "learning_rate": 6.0795e-05,
      "loss": 0.4406,
      "step": 12166
    },
    {
      "epoch": 0.6813192966737597,
      "grad_norm": 1.5319968461990356,
      "learning_rate": 6.08e-05,
      "loss": 0.4159,
      "step": 12167
    },
    {
      "epoch": 0.6813752939858887,
      "grad_norm": 1.0890932083129883,
      "learning_rate": 6.0805e-05,
      "loss": 0.3229,
      "step": 12168
    },
    {
      "epoch": 0.6814312912980177,
      "grad_norm": 1.1322203874588013,
      "learning_rate": 6.081e-05,
      "loss": 0.3951,
      "step": 12169
    },
    {
      "epoch": 0.6814872886101467,
      "grad_norm": 1.456730604171753,
      "learning_rate": 6.0815e-05,
      "loss": 0.459,
      "step": 12170
    },
    {
      "epoch": 0.6815432859222758,
      "grad_norm": 1.2333000898361206,
      "learning_rate": 6.082e-05,
      "loss": 0.3485,
      "step": 12171
    },
    {
      "epoch": 0.6815992832344048,
      "grad_norm": 1.2088539600372314,
      "learning_rate": 6.0825e-05,
      "loss": 0.4149,
      "step": 12172
    },
    {
      "epoch": 0.6816552805465338,
      "grad_norm": 1.328800916671753,
      "learning_rate": 6.083e-05,
      "loss": 0.5339,
      "step": 12173
    },
    {
      "epoch": 0.6817112778586628,
      "grad_norm": 1.2763824462890625,
      "learning_rate": 6.0834999999999996e-05,
      "loss": 0.4658,
      "step": 12174
    },
    {
      "epoch": 0.6817672751707918,
      "grad_norm": 1.3026294708251953,
      "learning_rate": 6.084000000000001e-05,
      "loss": 0.4155,
      "step": 12175
    },
    {
      "epoch": 0.6818232724829209,
      "grad_norm": 1.4607728719711304,
      "learning_rate": 6.0845000000000004e-05,
      "loss": 0.5532,
      "step": 12176
    },
    {
      "epoch": 0.6818792697950499,
      "grad_norm": 1.214138150215149,
      "learning_rate": 6.085000000000001e-05,
      "loss": 0.4987,
      "step": 12177
    },
    {
      "epoch": 0.6819352671071789,
      "grad_norm": 1.393033504486084,
      "learning_rate": 6.0855000000000006e-05,
      "loss": 0.3502,
      "step": 12178
    },
    {
      "epoch": 0.6819912644193079,
      "grad_norm": 1.378004550933838,
      "learning_rate": 6.0860000000000003e-05,
      "loss": 0.4423,
      "step": 12179
    },
    {
      "epoch": 0.6820472617314369,
      "grad_norm": 1.2123223543167114,
      "learning_rate": 6.086500000000001e-05,
      "loss": 0.3355,
      "step": 12180
    },
    {
      "epoch": 0.682103259043566,
      "grad_norm": 1.2966917753219604,
      "learning_rate": 6.0870000000000005e-05,
      "loss": 0.4241,
      "step": 12181
    },
    {
      "epoch": 0.682159256355695,
      "grad_norm": 1.3020904064178467,
      "learning_rate": 6.0875e-05,
      "loss": 0.4779,
      "step": 12182
    },
    {
      "epoch": 0.682215253667824,
      "grad_norm": 1.3528094291687012,
      "learning_rate": 6.088000000000001e-05,
      "loss": 0.427,
      "step": 12183
    },
    {
      "epoch": 0.682271250979953,
      "grad_norm": 1.0275715589523315,
      "learning_rate": 6.0885000000000004e-05,
      "loss": 0.3687,
      "step": 12184
    },
    {
      "epoch": 0.682327248292082,
      "grad_norm": 1.3559116125106812,
      "learning_rate": 6.089e-05,
      "loss": 0.4377,
      "step": 12185
    },
    {
      "epoch": 0.682383245604211,
      "grad_norm": 2.2175986766815186,
      "learning_rate": 6.0895e-05,
      "loss": 0.4603,
      "step": 12186
    },
    {
      "epoch": 0.6824392429163401,
      "grad_norm": 1.3152580261230469,
      "learning_rate": 6.09e-05,
      "loss": 0.4857,
      "step": 12187
    },
    {
      "epoch": 0.6824952402284691,
      "grad_norm": 1.3260953426361084,
      "learning_rate": 6.0905e-05,
      "loss": 0.4299,
      "step": 12188
    },
    {
      "epoch": 0.6825512375405981,
      "grad_norm": 1.5903900861740112,
      "learning_rate": 6.091e-05,
      "loss": 0.6495,
      "step": 12189
    },
    {
      "epoch": 0.6826072348527271,
      "grad_norm": 1.2022602558135986,
      "learning_rate": 6.0915e-05,
      "loss": 0.3198,
      "step": 12190
    },
    {
      "epoch": 0.6826632321648561,
      "grad_norm": 1.2887011766433716,
      "learning_rate": 6.092e-05,
      "loss": 0.409,
      "step": 12191
    },
    {
      "epoch": 0.6827192294769852,
      "grad_norm": 1.371610164642334,
      "learning_rate": 6.0925e-05,
      "loss": 0.4759,
      "step": 12192
    },
    {
      "epoch": 0.6827752267891142,
      "grad_norm": 1.2783170938491821,
      "learning_rate": 6.093e-05,
      "loss": 0.4126,
      "step": 12193
    },
    {
      "epoch": 0.6828312241012432,
      "grad_norm": 1.5165284872055054,
      "learning_rate": 6.0935e-05,
      "loss": 0.5952,
      "step": 12194
    },
    {
      "epoch": 0.6828872214133721,
      "grad_norm": 1.315489411354065,
      "learning_rate": 6.094000000000001e-05,
      "loss": 0.4228,
      "step": 12195
    },
    {
      "epoch": 0.6829432187255011,
      "grad_norm": 1.4239274263381958,
      "learning_rate": 6.094500000000001e-05,
      "loss": 0.5519,
      "step": 12196
    },
    {
      "epoch": 0.6829992160376301,
      "grad_norm": 1.2774584293365479,
      "learning_rate": 6.0950000000000004e-05,
      "loss": 0.3841,
      "step": 12197
    },
    {
      "epoch": 0.6830552133497592,
      "grad_norm": 1.2198424339294434,
      "learning_rate": 6.095500000000001e-05,
      "loss": 0.4392,
      "step": 12198
    },
    {
      "epoch": 0.6831112106618882,
      "grad_norm": 1.5404493808746338,
      "learning_rate": 6.0960000000000006e-05,
      "loss": 0.5669,
      "step": 12199
    },
    {
      "epoch": 0.6831672079740172,
      "grad_norm": 1.1638553142547607,
      "learning_rate": 6.0965e-05,
      "loss": 0.3687,
      "step": 12200
    },
    {
      "epoch": 0.6832232052861462,
      "grad_norm": 1.316155195236206,
      "learning_rate": 6.097000000000001e-05,
      "loss": 0.5432,
      "step": 12201
    },
    {
      "epoch": 0.6832792025982752,
      "grad_norm": 1.062378168106079,
      "learning_rate": 6.0975000000000005e-05,
      "loss": 0.3574,
      "step": 12202
    },
    {
      "epoch": 0.6833351999104043,
      "grad_norm": 1.0928003787994385,
      "learning_rate": 6.098e-05,
      "loss": 0.3067,
      "step": 12203
    },
    {
      "epoch": 0.6833911972225333,
      "grad_norm": 1.4202882051467896,
      "learning_rate": 6.0985000000000006e-05,
      "loss": 0.4275,
      "step": 12204
    },
    {
      "epoch": 0.6834471945346623,
      "grad_norm": 1.2420756816864014,
      "learning_rate": 6.0990000000000004e-05,
      "loss": 0.407,
      "step": 12205
    },
    {
      "epoch": 0.6835031918467913,
      "grad_norm": 1.6382828950881958,
      "learning_rate": 6.0995e-05,
      "loss": 0.8044,
      "step": 12206
    },
    {
      "epoch": 0.6835591891589203,
      "grad_norm": 1.3947349786758423,
      "learning_rate": 6.1e-05,
      "loss": 0.4024,
      "step": 12207
    },
    {
      "epoch": 0.6836151864710494,
      "grad_norm": 1.2955342531204224,
      "learning_rate": 6.1005e-05,
      "loss": 0.5735,
      "step": 12208
    },
    {
      "epoch": 0.6836711837831784,
      "grad_norm": 1.34746515750885,
      "learning_rate": 6.101e-05,
      "loss": 0.3859,
      "step": 12209
    },
    {
      "epoch": 0.6837271810953074,
      "grad_norm": 1.415920376777649,
      "learning_rate": 6.1015e-05,
      "loss": 0.5651,
      "step": 12210
    },
    {
      "epoch": 0.6837831784074364,
      "grad_norm": 1.3192853927612305,
      "learning_rate": 6.102e-05,
      "loss": 0.4725,
      "step": 12211
    },
    {
      "epoch": 0.6838391757195654,
      "grad_norm": 1.2164310216903687,
      "learning_rate": 6.1025e-05,
      "loss": 0.4347,
      "step": 12212
    },
    {
      "epoch": 0.6838951730316944,
      "grad_norm": 1.2607762813568115,
      "learning_rate": 6.103e-05,
      "loss": 0.4457,
      "step": 12213
    },
    {
      "epoch": 0.6839511703438235,
      "grad_norm": 1.4990837574005127,
      "learning_rate": 6.1035e-05,
      "loss": 0.5186,
      "step": 12214
    },
    {
      "epoch": 0.6840071676559525,
      "grad_norm": 1.424031376838684,
      "learning_rate": 6.104000000000001e-05,
      "loss": 0.4896,
      "step": 12215
    },
    {
      "epoch": 0.6840631649680815,
      "grad_norm": 1.4935369491577148,
      "learning_rate": 6.104500000000001e-05,
      "loss": 0.4456,
      "step": 12216
    },
    {
      "epoch": 0.6841191622802105,
      "grad_norm": 1.1517411470413208,
      "learning_rate": 6.105e-05,
      "loss": 0.4245,
      "step": 12217
    },
    {
      "epoch": 0.6841751595923395,
      "grad_norm": 1.3826349973678589,
      "learning_rate": 6.1055e-05,
      "loss": 0.5045,
      "step": 12218
    },
    {
      "epoch": 0.6842311569044686,
      "grad_norm": 2.0401902198791504,
      "learning_rate": 6.106e-05,
      "loss": 0.5802,
      "step": 12219
    },
    {
      "epoch": 0.6842871542165976,
      "grad_norm": 1.386125922203064,
      "learning_rate": 6.1065e-05,
      "loss": 0.3533,
      "step": 12220
    },
    {
      "epoch": 0.6843431515287266,
      "grad_norm": 1.3331139087677002,
      "learning_rate": 6.107000000000001e-05,
      "loss": 0.5166,
      "step": 12221
    },
    {
      "epoch": 0.6843991488408556,
      "grad_norm": 1.130191683769226,
      "learning_rate": 6.107500000000001e-05,
      "loss": 0.4323,
      "step": 12222
    },
    {
      "epoch": 0.6844551461529846,
      "grad_norm": 1.1950807571411133,
      "learning_rate": 6.108e-05,
      "loss": 0.4142,
      "step": 12223
    },
    {
      "epoch": 0.6845111434651137,
      "grad_norm": 1.4747769832611084,
      "learning_rate": 6.1085e-05,
      "loss": 0.505,
      "step": 12224
    },
    {
      "epoch": 0.6845671407772427,
      "grad_norm": 2.184282064437866,
      "learning_rate": 6.109e-05,
      "loss": 0.426,
      "step": 12225
    },
    {
      "epoch": 0.6846231380893717,
      "grad_norm": 1.0928298234939575,
      "learning_rate": 6.1095e-05,
      "loss": 0.3717,
      "step": 12226
    },
    {
      "epoch": 0.6846791354015007,
      "grad_norm": 1.3561042547225952,
      "learning_rate": 6.110000000000001e-05,
      "loss": 0.4217,
      "step": 12227
    },
    {
      "epoch": 0.6847351327136297,
      "grad_norm": 1.2690937519073486,
      "learning_rate": 6.1105e-05,
      "loss": 0.5044,
      "step": 12228
    },
    {
      "epoch": 0.6847911300257588,
      "grad_norm": 1.5557212829589844,
      "learning_rate": 6.111e-05,
      "loss": 0.4117,
      "step": 12229
    },
    {
      "epoch": 0.6848471273378878,
      "grad_norm": 1.4569201469421387,
      "learning_rate": 6.1115e-05,
      "loss": 0.5277,
      "step": 12230
    },
    {
      "epoch": 0.6849031246500168,
      "grad_norm": 1.5207160711288452,
      "learning_rate": 6.112e-05,
      "loss": 0.5777,
      "step": 12231
    },
    {
      "epoch": 0.6849591219621458,
      "grad_norm": 1.2555395364761353,
      "learning_rate": 6.1125e-05,
      "loss": 0.3153,
      "step": 12232
    },
    {
      "epoch": 0.6850151192742748,
      "grad_norm": 1.2674345970153809,
      "learning_rate": 6.112999999999999e-05,
      "loss": 0.4369,
      "step": 12233
    },
    {
      "epoch": 0.6850711165864038,
      "grad_norm": 1.3939169645309448,
      "learning_rate": 6.1135e-05,
      "loss": 0.5536,
      "step": 12234
    },
    {
      "epoch": 0.6851271138985329,
      "grad_norm": 1.5421286821365356,
      "learning_rate": 6.114000000000001e-05,
      "loss": 0.6619,
      "step": 12235
    },
    {
      "epoch": 0.6851831112106619,
      "grad_norm": 1.2183438539505005,
      "learning_rate": 6.114500000000001e-05,
      "loss": 0.412,
      "step": 12236
    },
    {
      "epoch": 0.6852391085227909,
      "grad_norm": 1.5787529945373535,
      "learning_rate": 6.115000000000001e-05,
      "loss": 0.4965,
      "step": 12237
    },
    {
      "epoch": 0.6852951058349199,
      "grad_norm": 1.439435362815857,
      "learning_rate": 6.1155e-05,
      "loss": 0.546,
      "step": 12238
    },
    {
      "epoch": 0.685351103147049,
      "grad_norm": 1.4464318752288818,
      "learning_rate": 6.116e-05,
      "loss": 0.4975,
      "step": 12239
    },
    {
      "epoch": 0.685407100459178,
      "grad_norm": 1.4630446434020996,
      "learning_rate": 6.1165e-05,
      "loss": 0.4319,
      "step": 12240
    },
    {
      "epoch": 0.685463097771307,
      "grad_norm": 1.3636311292648315,
      "learning_rate": 6.117e-05,
      "loss": 0.3872,
      "step": 12241
    },
    {
      "epoch": 0.685519095083436,
      "grad_norm": 1.2092517614364624,
      "learning_rate": 6.117500000000001e-05,
      "loss": 0.4466,
      "step": 12242
    },
    {
      "epoch": 0.685575092395565,
      "grad_norm": 1.2505483627319336,
      "learning_rate": 6.118000000000001e-05,
      "loss": 0.463,
      "step": 12243
    },
    {
      "epoch": 0.685631089707694,
      "grad_norm": 1.0488157272338867,
      "learning_rate": 6.1185e-05,
      "loss": 0.4061,
      "step": 12244
    },
    {
      "epoch": 0.6856870870198231,
      "grad_norm": 1.2378517389297485,
      "learning_rate": 6.119e-05,
      "loss": 0.4247,
      "step": 12245
    },
    {
      "epoch": 0.6857430843319521,
      "grad_norm": 1.3138020038604736,
      "learning_rate": 6.1195e-05,
      "loss": 0.4243,
      "step": 12246
    },
    {
      "epoch": 0.6857990816440811,
      "grad_norm": 1.1253302097320557,
      "learning_rate": 6.12e-05,
      "loss": 0.449,
      "step": 12247
    },
    {
      "epoch": 0.6858550789562101,
      "grad_norm": 1.4977877140045166,
      "learning_rate": 6.120500000000001e-05,
      "loss": 0.5941,
      "step": 12248
    },
    {
      "epoch": 0.6859110762683391,
      "grad_norm": 1.7441450357437134,
      "learning_rate": 6.121e-05,
      "loss": 0.4848,
      "step": 12249
    },
    {
      "epoch": 0.6859670735804682,
      "grad_norm": 1.2781704664230347,
      "learning_rate": 6.1215e-05,
      "loss": 0.5405,
      "step": 12250
    },
    {
      "epoch": 0.6860230708925972,
      "grad_norm": 1.3380956649780273,
      "learning_rate": 6.122e-05,
      "loss": 0.446,
      "step": 12251
    },
    {
      "epoch": 0.6860790682047262,
      "grad_norm": 1.4156084060668945,
      "learning_rate": 6.1225e-05,
      "loss": 0.6244,
      "step": 12252
    },
    {
      "epoch": 0.6861350655168552,
      "grad_norm": 1.3929370641708374,
      "learning_rate": 6.123e-05,
      "loss": 0.3996,
      "step": 12253
    },
    {
      "epoch": 0.6861910628289842,
      "grad_norm": 1.4578877687454224,
      "learning_rate": 6.123499999999999e-05,
      "loss": 0.5644,
      "step": 12254
    },
    {
      "epoch": 0.6862470601411133,
      "grad_norm": 1.2098445892333984,
      "learning_rate": 6.124e-05,
      "loss": 0.4117,
      "step": 12255
    },
    {
      "epoch": 0.6863030574532423,
      "grad_norm": 0.9289770722389221,
      "learning_rate": 6.124500000000001e-05,
      "loss": 0.2874,
      "step": 12256
    },
    {
      "epoch": 0.6863590547653713,
      "grad_norm": 1.2058159112930298,
      "learning_rate": 6.125000000000001e-05,
      "loss": 0.4046,
      "step": 12257
    },
    {
      "epoch": 0.6864150520775003,
      "grad_norm": 1.042585015296936,
      "learning_rate": 6.125500000000001e-05,
      "loss": 0.4132,
      "step": 12258
    },
    {
      "epoch": 0.6864710493896293,
      "grad_norm": 1.7090892791748047,
      "learning_rate": 6.126e-05,
      "loss": 0.4091,
      "step": 12259
    },
    {
      "epoch": 0.6865270467017583,
      "grad_norm": 1.4218131303787231,
      "learning_rate": 6.1265e-05,
      "loss": 0.4224,
      "step": 12260
    },
    {
      "epoch": 0.6865830440138874,
      "grad_norm": 1.2966798543930054,
      "learning_rate": 6.127e-05,
      "loss": 0.3678,
      "step": 12261
    },
    {
      "epoch": 0.6866390413260164,
      "grad_norm": 1.028228759765625,
      "learning_rate": 6.1275e-05,
      "loss": 0.3057,
      "step": 12262
    },
    {
      "epoch": 0.6866950386381454,
      "grad_norm": 1.1287555694580078,
      "learning_rate": 6.128000000000001e-05,
      "loss": 0.3543,
      "step": 12263
    },
    {
      "epoch": 0.6867510359502744,
      "grad_norm": 1.1865333318710327,
      "learning_rate": 6.128500000000001e-05,
      "loss": 0.5076,
      "step": 12264
    },
    {
      "epoch": 0.6868070332624034,
      "grad_norm": 1.2991281747817993,
      "learning_rate": 6.129e-05,
      "loss": 0.3995,
      "step": 12265
    },
    {
      "epoch": 0.6868630305745325,
      "grad_norm": 1.4616847038269043,
      "learning_rate": 6.1295e-05,
      "loss": 0.4717,
      "step": 12266
    },
    {
      "epoch": 0.6869190278866615,
      "grad_norm": 1.5509908199310303,
      "learning_rate": 6.13e-05,
      "loss": 0.4697,
      "step": 12267
    },
    {
      "epoch": 0.6869750251987905,
      "grad_norm": 1.568916916847229,
      "learning_rate": 6.1305e-05,
      "loss": 0.6279,
      "step": 12268
    },
    {
      "epoch": 0.6870310225109195,
      "grad_norm": 1.1783875226974487,
      "learning_rate": 6.131e-05,
      "loss": 0.3793,
      "step": 12269
    },
    {
      "epoch": 0.6870870198230485,
      "grad_norm": 1.2726467847824097,
      "learning_rate": 6.1315e-05,
      "loss": 0.5254,
      "step": 12270
    },
    {
      "epoch": 0.6871430171351776,
      "grad_norm": 1.444631576538086,
      "learning_rate": 6.132e-05,
      "loss": 0.4486,
      "step": 12271
    },
    {
      "epoch": 0.6871990144473066,
      "grad_norm": 32.72178649902344,
      "learning_rate": 6.1325e-05,
      "loss": 0.5434,
      "step": 12272
    },
    {
      "epoch": 0.6872550117594356,
      "grad_norm": 1.4250762462615967,
      "learning_rate": 6.133e-05,
      "loss": 0.4798,
      "step": 12273
    },
    {
      "epoch": 0.6873110090715646,
      "grad_norm": 1.2289494276046753,
      "learning_rate": 6.1335e-05,
      "loss": 0.4453,
      "step": 12274
    },
    {
      "epoch": 0.6873670063836936,
      "grad_norm": 1.4113556146621704,
      "learning_rate": 6.133999999999999e-05,
      "loss": 0.4163,
      "step": 12275
    },
    {
      "epoch": 0.6874230036958227,
      "grad_norm": 1.1996521949768066,
      "learning_rate": 6.1345e-05,
      "loss": 0.3363,
      "step": 12276
    },
    {
      "epoch": 0.6874790010079517,
      "grad_norm": 1.3179504871368408,
      "learning_rate": 6.135000000000001e-05,
      "loss": 0.4021,
      "step": 12277
    },
    {
      "epoch": 0.6875349983200806,
      "grad_norm": 1.270889163017273,
      "learning_rate": 6.135500000000001e-05,
      "loss": 0.4603,
      "step": 12278
    },
    {
      "epoch": 0.6875909956322096,
      "grad_norm": 1.40614652633667,
      "learning_rate": 6.136000000000001e-05,
      "loss": 0.5777,
      "step": 12279
    },
    {
      "epoch": 0.6876469929443386,
      "grad_norm": 1.3910115957260132,
      "learning_rate": 6.1365e-05,
      "loss": 0.4963,
      "step": 12280
    },
    {
      "epoch": 0.6877029902564676,
      "grad_norm": 1.5937877893447876,
      "learning_rate": 6.137e-05,
      "loss": 0.4653,
      "step": 12281
    },
    {
      "epoch": 0.6877589875685967,
      "grad_norm": 1.1973016262054443,
      "learning_rate": 6.1375e-05,
      "loss": 0.4001,
      "step": 12282
    },
    {
      "epoch": 0.6878149848807257,
      "grad_norm": 1.4405215978622437,
      "learning_rate": 6.138e-05,
      "loss": 0.5029,
      "step": 12283
    },
    {
      "epoch": 0.6878709821928547,
      "grad_norm": 1.4870195388793945,
      "learning_rate": 6.138500000000001e-05,
      "loss": 0.5531,
      "step": 12284
    },
    {
      "epoch": 0.6879269795049837,
      "grad_norm": 1.3453369140625,
      "learning_rate": 6.139000000000001e-05,
      "loss": 0.4839,
      "step": 12285
    },
    {
      "epoch": 0.6879829768171127,
      "grad_norm": 1.1019184589385986,
      "learning_rate": 6.1395e-05,
      "loss": 0.338,
      "step": 12286
    },
    {
      "epoch": 0.6880389741292418,
      "grad_norm": 1.1243878602981567,
      "learning_rate": 6.14e-05,
      "loss": 0.3211,
      "step": 12287
    },
    {
      "epoch": 0.6880949714413708,
      "grad_norm": 1.2816708087921143,
      "learning_rate": 6.1405e-05,
      "loss": 0.468,
      "step": 12288
    },
    {
      "epoch": 0.6881509687534998,
      "grad_norm": 1.57627534866333,
      "learning_rate": 6.141e-05,
      "loss": 0.6315,
      "step": 12289
    },
    {
      "epoch": 0.6882069660656288,
      "grad_norm": 1.2261550426483154,
      "learning_rate": 6.1415e-05,
      "loss": 0.3907,
      "step": 12290
    },
    {
      "epoch": 0.6882629633777578,
      "grad_norm": 1.2663525342941284,
      "learning_rate": 6.142e-05,
      "loss": 0.3551,
      "step": 12291
    },
    {
      "epoch": 0.6883189606898868,
      "grad_norm": 1.3640536069869995,
      "learning_rate": 6.1425e-05,
      "loss": 0.4819,
      "step": 12292
    },
    {
      "epoch": 0.6883749580020159,
      "grad_norm": 1.5169423818588257,
      "learning_rate": 6.143e-05,
      "loss": 0.4374,
      "step": 12293
    },
    {
      "epoch": 0.6884309553141449,
      "grad_norm": 1.093767523765564,
      "learning_rate": 6.1435e-05,
      "loss": 0.4196,
      "step": 12294
    },
    {
      "epoch": 0.6884869526262739,
      "grad_norm": 2.810868501663208,
      "learning_rate": 6.144e-05,
      "loss": 0.3948,
      "step": 12295
    },
    {
      "epoch": 0.6885429499384029,
      "grad_norm": 1.4540945291519165,
      "learning_rate": 6.1445e-05,
      "loss": 0.4799,
      "step": 12296
    },
    {
      "epoch": 0.688598947250532,
      "grad_norm": 1.3927223682403564,
      "learning_rate": 6.145e-05,
      "loss": 0.3329,
      "step": 12297
    },
    {
      "epoch": 0.688654944562661,
      "grad_norm": 1.3568576574325562,
      "learning_rate": 6.1455e-05,
      "loss": 0.42,
      "step": 12298
    },
    {
      "epoch": 0.68871094187479,
      "grad_norm": 1.440532922744751,
      "learning_rate": 6.146000000000001e-05,
      "loss": 0.6115,
      "step": 12299
    },
    {
      "epoch": 0.688766939186919,
      "grad_norm": 1.3961313962936401,
      "learning_rate": 6.146500000000001e-05,
      "loss": 0.4438,
      "step": 12300
    },
    {
      "epoch": 0.688822936499048,
      "grad_norm": 1.181216835975647,
      "learning_rate": 6.147e-05,
      "loss": 0.4751,
      "step": 12301
    },
    {
      "epoch": 0.688878933811177,
      "grad_norm": 1.2675639390945435,
      "learning_rate": 6.1475e-05,
      "loss": 0.3858,
      "step": 12302
    },
    {
      "epoch": 0.6889349311233061,
      "grad_norm": 1.2192527055740356,
      "learning_rate": 6.148e-05,
      "loss": 0.375,
      "step": 12303
    },
    {
      "epoch": 0.6889909284354351,
      "grad_norm": 1.0635168552398682,
      "learning_rate": 6.1485e-05,
      "loss": 0.4258,
      "step": 12304
    },
    {
      "epoch": 0.6890469257475641,
      "grad_norm": 1.2519820928573608,
      "learning_rate": 6.149000000000001e-05,
      "loss": 0.414,
      "step": 12305
    },
    {
      "epoch": 0.6891029230596931,
      "grad_norm": 1.3190950155258179,
      "learning_rate": 6.1495e-05,
      "loss": 0.386,
      "step": 12306
    },
    {
      "epoch": 0.6891589203718221,
      "grad_norm": 1.805222988128662,
      "learning_rate": 6.15e-05,
      "loss": 0.5352,
      "step": 12307
    },
    {
      "epoch": 0.6892149176839512,
      "grad_norm": 1.282267451286316,
      "learning_rate": 6.1505e-05,
      "loss": 0.3987,
      "step": 12308
    },
    {
      "epoch": 0.6892709149960802,
      "grad_norm": 1.27260160446167,
      "learning_rate": 6.151e-05,
      "loss": 0.5293,
      "step": 12309
    },
    {
      "epoch": 0.6893269123082092,
      "grad_norm": 1.2897119522094727,
      "learning_rate": 6.1515e-05,
      "loss": 0.3931,
      "step": 12310
    },
    {
      "epoch": 0.6893829096203382,
      "grad_norm": 1.5478432178497314,
      "learning_rate": 6.152e-05,
      "loss": 0.5605,
      "step": 12311
    },
    {
      "epoch": 0.6894389069324672,
      "grad_norm": 1.4531878232955933,
      "learning_rate": 6.1525e-05,
      "loss": 0.4144,
      "step": 12312
    },
    {
      "epoch": 0.6894949042445963,
      "grad_norm": 1.2213512659072876,
      "learning_rate": 6.153e-05,
      "loss": 0.3277,
      "step": 12313
    },
    {
      "epoch": 0.6895509015567253,
      "grad_norm": 1.3131664991378784,
      "learning_rate": 6.1535e-05,
      "loss": 0.4082,
      "step": 12314
    },
    {
      "epoch": 0.6896068988688543,
      "grad_norm": 1.3242067098617554,
      "learning_rate": 6.154e-05,
      "loss": 0.3955,
      "step": 12315
    },
    {
      "epoch": 0.6896628961809833,
      "grad_norm": 1.5959453582763672,
      "learning_rate": 6.154500000000001e-05,
      "loss": 0.5431,
      "step": 12316
    },
    {
      "epoch": 0.6897188934931123,
      "grad_norm": 1.1596336364746094,
      "learning_rate": 6.155e-05,
      "loss": 0.5232,
      "step": 12317
    },
    {
      "epoch": 0.6897748908052413,
      "grad_norm": 1.3719199895858765,
      "learning_rate": 6.1555e-05,
      "loss": 0.5622,
      "step": 12318
    },
    {
      "epoch": 0.6898308881173704,
      "grad_norm": 1.2654657363891602,
      "learning_rate": 6.156e-05,
      "loss": 0.4272,
      "step": 12319
    },
    {
      "epoch": 0.6898868854294994,
      "grad_norm": 1.224621295928955,
      "learning_rate": 6.156500000000001e-05,
      "loss": 0.3971,
      "step": 12320
    },
    {
      "epoch": 0.6899428827416284,
      "grad_norm": 1.293892502784729,
      "learning_rate": 6.157000000000001e-05,
      "loss": 0.5648,
      "step": 12321
    },
    {
      "epoch": 0.6899988800537574,
      "grad_norm": 1.255736231803894,
      "learning_rate": 6.1575e-05,
      "loss": 0.4034,
      "step": 12322
    },
    {
      "epoch": 0.6900548773658864,
      "grad_norm": 1.3798484802246094,
      "learning_rate": 6.158e-05,
      "loss": 0.3675,
      "step": 12323
    },
    {
      "epoch": 0.6901108746780155,
      "grad_norm": 1.1086690425872803,
      "learning_rate": 6.1585e-05,
      "loss": 0.3527,
      "step": 12324
    },
    {
      "epoch": 0.6901668719901445,
      "grad_norm": 1.802785873413086,
      "learning_rate": 6.159e-05,
      "loss": 0.3634,
      "step": 12325
    },
    {
      "epoch": 0.6902228693022735,
      "grad_norm": 1.3808271884918213,
      "learning_rate": 6.159500000000001e-05,
      "loss": 0.4957,
      "step": 12326
    },
    {
      "epoch": 0.6902788666144025,
      "grad_norm": 1.555474877357483,
      "learning_rate": 6.16e-05,
      "loss": 0.5661,
      "step": 12327
    },
    {
      "epoch": 0.6903348639265315,
      "grad_norm": 7.169306755065918,
      "learning_rate": 6.1605e-05,
      "loss": 0.4269,
      "step": 12328
    },
    {
      "epoch": 0.6903908612386606,
      "grad_norm": 1.2660406827926636,
      "learning_rate": 6.161e-05,
      "loss": 0.4906,
      "step": 12329
    },
    {
      "epoch": 0.6904468585507896,
      "grad_norm": 1.4176205396652222,
      "learning_rate": 6.1615e-05,
      "loss": 0.5224,
      "step": 12330
    },
    {
      "epoch": 0.6905028558629186,
      "grad_norm": 1.3572801351547241,
      "learning_rate": 6.162e-05,
      "loss": 0.483,
      "step": 12331
    },
    {
      "epoch": 0.6905588531750476,
      "grad_norm": 1.1732454299926758,
      "learning_rate": 6.1625e-05,
      "loss": 0.3916,
      "step": 12332
    },
    {
      "epoch": 0.6906148504871766,
      "grad_norm": 1.292445182800293,
      "learning_rate": 6.163e-05,
      "loss": 0.4665,
      "step": 12333
    },
    {
      "epoch": 0.6906708477993057,
      "grad_norm": 1.5300066471099854,
      "learning_rate": 6.1635e-05,
      "loss": 0.607,
      "step": 12334
    },
    {
      "epoch": 0.6907268451114347,
      "grad_norm": 1.125780701637268,
      "learning_rate": 6.164e-05,
      "loss": 0.3816,
      "step": 12335
    },
    {
      "epoch": 0.6907828424235637,
      "grad_norm": 1.141910433769226,
      "learning_rate": 6.164500000000001e-05,
      "loss": 0.4099,
      "step": 12336
    },
    {
      "epoch": 0.6908388397356927,
      "grad_norm": 1.2341314554214478,
      "learning_rate": 6.165000000000001e-05,
      "loss": 0.4981,
      "step": 12337
    },
    {
      "epoch": 0.6908948370478217,
      "grad_norm": 1.4774041175842285,
      "learning_rate": 6.1655e-05,
      "loss": 0.5521,
      "step": 12338
    },
    {
      "epoch": 0.6909508343599507,
      "grad_norm": 1.1159309148788452,
      "learning_rate": 6.166e-05,
      "loss": 0.3608,
      "step": 12339
    },
    {
      "epoch": 0.6910068316720798,
      "grad_norm": 1.3648525476455688,
      "learning_rate": 6.1665e-05,
      "loss": 0.4344,
      "step": 12340
    },
    {
      "epoch": 0.6910628289842088,
      "grad_norm": 1.148815631866455,
      "learning_rate": 6.167000000000001e-05,
      "loss": 0.4819,
      "step": 12341
    },
    {
      "epoch": 0.6911188262963378,
      "grad_norm": 1.500679612159729,
      "learning_rate": 6.167500000000001e-05,
      "loss": 0.6123,
      "step": 12342
    },
    {
      "epoch": 0.6911748236084668,
      "grad_norm": 1.464145541191101,
      "learning_rate": 6.168e-05,
      "loss": 0.5013,
      "step": 12343
    },
    {
      "epoch": 0.6912308209205958,
      "grad_norm": 1.5020862817764282,
      "learning_rate": 6.1685e-05,
      "loss": 0.4593,
      "step": 12344
    },
    {
      "epoch": 0.6912868182327249,
      "grad_norm": 1.1385457515716553,
      "learning_rate": 6.169e-05,
      "loss": 0.387,
      "step": 12345
    },
    {
      "epoch": 0.6913428155448539,
      "grad_norm": 1.4682669639587402,
      "learning_rate": 6.1695e-05,
      "loss": 0.432,
      "step": 12346
    },
    {
      "epoch": 0.6913988128569829,
      "grad_norm": 1.1639906167984009,
      "learning_rate": 6.170000000000001e-05,
      "loss": 0.3979,
      "step": 12347
    },
    {
      "epoch": 0.6914548101691119,
      "grad_norm": 2.324922561645508,
      "learning_rate": 6.1705e-05,
      "loss": 0.4675,
      "step": 12348
    },
    {
      "epoch": 0.6915108074812409,
      "grad_norm": 1.420125126838684,
      "learning_rate": 6.171e-05,
      "loss": 0.5791,
      "step": 12349
    },
    {
      "epoch": 0.69156680479337,
      "grad_norm": 1.4984275102615356,
      "learning_rate": 6.1715e-05,
      "loss": 0.5217,
      "step": 12350
    },
    {
      "epoch": 0.691622802105499,
      "grad_norm": 1.2690491676330566,
      "learning_rate": 6.172e-05,
      "loss": 0.4017,
      "step": 12351
    },
    {
      "epoch": 0.691678799417628,
      "grad_norm": 1.1910409927368164,
      "learning_rate": 6.1725e-05,
      "loss": 0.3263,
      "step": 12352
    },
    {
      "epoch": 0.691734796729757,
      "grad_norm": 3.2725179195404053,
      "learning_rate": 6.173e-05,
      "loss": 0.6993,
      "step": 12353
    },
    {
      "epoch": 0.691790794041886,
      "grad_norm": 1.2835434675216675,
      "learning_rate": 6.1735e-05,
      "loss": 0.467,
      "step": 12354
    },
    {
      "epoch": 0.6918467913540151,
      "grad_norm": 1.5669128894805908,
      "learning_rate": 6.174e-05,
      "loss": 0.4122,
      "step": 12355
    },
    {
      "epoch": 0.6919027886661441,
      "grad_norm": 1.442619800567627,
      "learning_rate": 6.174500000000001e-05,
      "loss": 0.4024,
      "step": 12356
    },
    {
      "epoch": 0.6919587859782731,
      "grad_norm": 1.4724701642990112,
      "learning_rate": 6.175000000000001e-05,
      "loss": 0.4798,
      "step": 12357
    },
    {
      "epoch": 0.6920147832904021,
      "grad_norm": 1.193954348564148,
      "learning_rate": 6.175500000000001e-05,
      "loss": 0.3632,
      "step": 12358
    },
    {
      "epoch": 0.6920707806025311,
      "grad_norm": 1.225625991821289,
      "learning_rate": 6.176e-05,
      "loss": 0.3221,
      "step": 12359
    },
    {
      "epoch": 0.6921267779146602,
      "grad_norm": 1.286931037902832,
      "learning_rate": 6.1765e-05,
      "loss": 0.417,
      "step": 12360
    },
    {
      "epoch": 0.6921827752267891,
      "grad_norm": 1.1550440788269043,
      "learning_rate": 6.177e-05,
      "loss": 0.3877,
      "step": 12361
    },
    {
      "epoch": 0.6922387725389181,
      "grad_norm": 1.173991084098816,
      "learning_rate": 6.177500000000001e-05,
      "loss": 0.4283,
      "step": 12362
    },
    {
      "epoch": 0.6922947698510471,
      "grad_norm": 1.178902268409729,
      "learning_rate": 6.178000000000001e-05,
      "loss": 0.4929,
      "step": 12363
    },
    {
      "epoch": 0.6923507671631761,
      "grad_norm": 1.4234992265701294,
      "learning_rate": 6.1785e-05,
      "loss": 0.4473,
      "step": 12364
    },
    {
      "epoch": 0.6924067644753051,
      "grad_norm": 1.3138865232467651,
      "learning_rate": 6.179e-05,
      "loss": 0.4019,
      "step": 12365
    },
    {
      "epoch": 0.6924627617874342,
      "grad_norm": 1.3836005926132202,
      "learning_rate": 6.1795e-05,
      "loss": 0.4197,
      "step": 12366
    },
    {
      "epoch": 0.6925187590995632,
      "grad_norm": 1.3653817176818848,
      "learning_rate": 6.18e-05,
      "loss": 0.422,
      "step": 12367
    },
    {
      "epoch": 0.6925747564116922,
      "grad_norm": 5.243404865264893,
      "learning_rate": 6.1805e-05,
      "loss": 0.462,
      "step": 12368
    },
    {
      "epoch": 0.6926307537238212,
      "grad_norm": 1.4292103052139282,
      "learning_rate": 6.181e-05,
      "loss": 0.5952,
      "step": 12369
    },
    {
      "epoch": 0.6926867510359502,
      "grad_norm": 1.1659621000289917,
      "learning_rate": 6.1815e-05,
      "loss": 0.4945,
      "step": 12370
    },
    {
      "epoch": 0.6927427483480793,
      "grad_norm": 1.1917778253555298,
      "learning_rate": 6.182e-05,
      "loss": 0.3642,
      "step": 12371
    },
    {
      "epoch": 0.6927987456602083,
      "grad_norm": 0.9862884283065796,
      "learning_rate": 6.1825e-05,
      "loss": 0.3606,
      "step": 12372
    },
    {
      "epoch": 0.6928547429723373,
      "grad_norm": 1.9535349607467651,
      "learning_rate": 6.183e-05,
      "loss": 0.4091,
      "step": 12373
    },
    {
      "epoch": 0.6929107402844663,
      "grad_norm": 1.199156403541565,
      "learning_rate": 6.1835e-05,
      "loss": 0.3657,
      "step": 12374
    },
    {
      "epoch": 0.6929667375965953,
      "grad_norm": 1.1949660778045654,
      "learning_rate": 6.184e-05,
      "loss": 0.4283,
      "step": 12375
    },
    {
      "epoch": 0.6930227349087243,
      "grad_norm": 1.4052292108535767,
      "learning_rate": 6.184500000000002e-05,
      "loss": 0.3788,
      "step": 12376
    },
    {
      "epoch": 0.6930787322208534,
      "grad_norm": 1.4408352375030518,
      "learning_rate": 6.185000000000001e-05,
      "loss": 0.4129,
      "step": 12377
    },
    {
      "epoch": 0.6931347295329824,
      "grad_norm": 1.9941606521606445,
      "learning_rate": 6.185500000000001e-05,
      "loss": 0.5218,
      "step": 12378
    },
    {
      "epoch": 0.6931907268451114,
      "grad_norm": 1.2464509010314941,
      "learning_rate": 6.186000000000001e-05,
      "loss": 0.4011,
      "step": 12379
    },
    {
      "epoch": 0.6932467241572404,
      "grad_norm": 1.2264740467071533,
      "learning_rate": 6.1865e-05,
      "loss": 0.3415,
      "step": 12380
    },
    {
      "epoch": 0.6933027214693694,
      "grad_norm": 1.244165062904358,
      "learning_rate": 6.187e-05,
      "loss": 0.4134,
      "step": 12381
    },
    {
      "epoch": 0.6933587187814985,
      "grad_norm": 1.3536137342453003,
      "learning_rate": 6.1875e-05,
      "loss": 0.5228,
      "step": 12382
    },
    {
      "epoch": 0.6934147160936275,
      "grad_norm": 1.158705472946167,
      "learning_rate": 6.188000000000001e-05,
      "loss": 0.3152,
      "step": 12383
    },
    {
      "epoch": 0.6934707134057565,
      "grad_norm": 1.3614567518234253,
      "learning_rate": 6.188500000000001e-05,
      "loss": 0.4177,
      "step": 12384
    },
    {
      "epoch": 0.6935267107178855,
      "grad_norm": 1.2631393671035767,
      "learning_rate": 6.189e-05,
      "loss": 0.3932,
      "step": 12385
    },
    {
      "epoch": 0.6935827080300145,
      "grad_norm": 1.2913925647735596,
      "learning_rate": 6.1895e-05,
      "loss": 0.4944,
      "step": 12386
    },
    {
      "epoch": 0.6936387053421436,
      "grad_norm": 1.3278769254684448,
      "learning_rate": 6.19e-05,
      "loss": 0.4233,
      "step": 12387
    },
    {
      "epoch": 0.6936947026542726,
      "grad_norm": 1.3033205270767212,
      "learning_rate": 6.1905e-05,
      "loss": 0.4485,
      "step": 12388
    },
    {
      "epoch": 0.6937506999664016,
      "grad_norm": 1.3875821828842163,
      "learning_rate": 6.191e-05,
      "loss": 0.5035,
      "step": 12389
    },
    {
      "epoch": 0.6938066972785306,
      "grad_norm": 1.3573859930038452,
      "learning_rate": 6.1915e-05,
      "loss": 0.4555,
      "step": 12390
    },
    {
      "epoch": 0.6938626945906596,
      "grad_norm": 1.2488489151000977,
      "learning_rate": 6.192e-05,
      "loss": 0.4228,
      "step": 12391
    },
    {
      "epoch": 0.6939186919027887,
      "grad_norm": 1.43985915184021,
      "learning_rate": 6.1925e-05,
      "loss": 0.5246,
      "step": 12392
    },
    {
      "epoch": 0.6939746892149177,
      "grad_norm": 1.0981749296188354,
      "learning_rate": 6.193e-05,
      "loss": 0.377,
      "step": 12393
    },
    {
      "epoch": 0.6940306865270467,
      "grad_norm": 1.3188191652297974,
      "learning_rate": 6.1935e-05,
      "loss": 0.3934,
      "step": 12394
    },
    {
      "epoch": 0.6940866838391757,
      "grad_norm": 1.3046042919158936,
      "learning_rate": 6.193999999999999e-05,
      "loss": 0.5209,
      "step": 12395
    },
    {
      "epoch": 0.6941426811513047,
      "grad_norm": 1.3684697151184082,
      "learning_rate": 6.1945e-05,
      "loss": 0.529,
      "step": 12396
    },
    {
      "epoch": 0.6941986784634337,
      "grad_norm": 49.712005615234375,
      "learning_rate": 6.195e-05,
      "loss": 0.5506,
      "step": 12397
    },
    {
      "epoch": 0.6942546757755628,
      "grad_norm": 1.3110227584838867,
      "learning_rate": 6.195500000000001e-05,
      "loss": 0.4521,
      "step": 12398
    },
    {
      "epoch": 0.6943106730876918,
      "grad_norm": 1.4124492406845093,
      "learning_rate": 6.196000000000001e-05,
      "loss": 0.6361,
      "step": 12399
    },
    {
      "epoch": 0.6943666703998208,
      "grad_norm": 1.3802233934402466,
      "learning_rate": 6.196500000000001e-05,
      "loss": 0.5113,
      "step": 12400
    },
    {
      "epoch": 0.6944226677119498,
      "grad_norm": 1.4428071975708008,
      "learning_rate": 6.197e-05,
      "loss": 0.5623,
      "step": 12401
    },
    {
      "epoch": 0.6944786650240788,
      "grad_norm": 1.30404531955719,
      "learning_rate": 6.1975e-05,
      "loss": 0.5918,
      "step": 12402
    },
    {
      "epoch": 0.6945346623362079,
      "grad_norm": 1.4471380710601807,
      "learning_rate": 6.198e-05,
      "loss": 0.4736,
      "step": 12403
    },
    {
      "epoch": 0.6945906596483369,
      "grad_norm": 1.5136996507644653,
      "learning_rate": 6.198500000000001e-05,
      "loss": 0.5243,
      "step": 12404
    },
    {
      "epoch": 0.6946466569604659,
      "grad_norm": 1.5524667501449585,
      "learning_rate": 6.199000000000001e-05,
      "loss": 0.4637,
      "step": 12405
    },
    {
      "epoch": 0.6947026542725949,
      "grad_norm": 10.976086616516113,
      "learning_rate": 6.1995e-05,
      "loss": 0.3784,
      "step": 12406
    },
    {
      "epoch": 0.6947586515847239,
      "grad_norm": 1.4833440780639648,
      "learning_rate": 6.2e-05,
      "loss": 0.6505,
      "step": 12407
    },
    {
      "epoch": 0.694814648896853,
      "grad_norm": 1.3871524333953857,
      "learning_rate": 6.2005e-05,
      "loss": 0.4584,
      "step": 12408
    },
    {
      "epoch": 0.694870646208982,
      "grad_norm": 1.3627489805221558,
      "learning_rate": 6.201e-05,
      "loss": 0.3724,
      "step": 12409
    },
    {
      "epoch": 0.694926643521111,
      "grad_norm": 1.328230381011963,
      "learning_rate": 6.2015e-05,
      "loss": 0.3918,
      "step": 12410
    },
    {
      "epoch": 0.69498264083324,
      "grad_norm": 1.4259014129638672,
      "learning_rate": 6.202e-05,
      "loss": 0.3789,
      "step": 12411
    },
    {
      "epoch": 0.695038638145369,
      "grad_norm": 1.0605638027191162,
      "learning_rate": 6.2025e-05,
      "loss": 0.3887,
      "step": 12412
    },
    {
      "epoch": 0.6950946354574981,
      "grad_norm": 1.4853936433792114,
      "learning_rate": 6.203e-05,
      "loss": 0.5887,
      "step": 12413
    },
    {
      "epoch": 0.6951506327696271,
      "grad_norm": 1.384651780128479,
      "learning_rate": 6.2035e-05,
      "loss": 0.6094,
      "step": 12414
    },
    {
      "epoch": 0.6952066300817561,
      "grad_norm": 1.2497036457061768,
      "learning_rate": 6.204e-05,
      "loss": 0.3074,
      "step": 12415
    },
    {
      "epoch": 0.6952626273938851,
      "grad_norm": 1.0918927192687988,
      "learning_rate": 6.204499999999999e-05,
      "loss": 0.3625,
      "step": 12416
    },
    {
      "epoch": 0.6953186247060141,
      "grad_norm": 1.1047202348709106,
      "learning_rate": 6.205e-05,
      "loss": 0.3787,
      "step": 12417
    },
    {
      "epoch": 0.6953746220181432,
      "grad_norm": 1.1791718006134033,
      "learning_rate": 6.2055e-05,
      "loss": 0.4311,
      "step": 12418
    },
    {
      "epoch": 0.6954306193302722,
      "grad_norm": 1.9835759401321411,
      "learning_rate": 6.206000000000001e-05,
      "loss": 0.5259,
      "step": 12419
    },
    {
      "epoch": 0.6954866166424012,
      "grad_norm": 1.4982942342758179,
      "learning_rate": 6.206500000000001e-05,
      "loss": 0.4543,
      "step": 12420
    },
    {
      "epoch": 0.6955426139545302,
      "grad_norm": 1.2310798168182373,
      "learning_rate": 6.207000000000001e-05,
      "loss": 0.4151,
      "step": 12421
    },
    {
      "epoch": 0.6955986112666592,
      "grad_norm": 1.3420618772506714,
      "learning_rate": 6.2075e-05,
      "loss": 0.4868,
      "step": 12422
    },
    {
      "epoch": 0.6956546085787882,
      "grad_norm": 1.1320631504058838,
      "learning_rate": 6.208e-05,
      "loss": 0.3478,
      "step": 12423
    },
    {
      "epoch": 0.6957106058909173,
      "grad_norm": 1.4887796640396118,
      "learning_rate": 6.2085e-05,
      "loss": 0.613,
      "step": 12424
    },
    {
      "epoch": 0.6957666032030463,
      "grad_norm": 1.5807126760482788,
      "learning_rate": 6.209000000000001e-05,
      "loss": 0.4946,
      "step": 12425
    },
    {
      "epoch": 0.6958226005151753,
      "grad_norm": 1.3650822639465332,
      "learning_rate": 6.209500000000001e-05,
      "loss": 0.4914,
      "step": 12426
    },
    {
      "epoch": 0.6958785978273043,
      "grad_norm": 1.8704893589019775,
      "learning_rate": 6.21e-05,
      "loss": 0.3987,
      "step": 12427
    },
    {
      "epoch": 0.6959345951394333,
      "grad_norm": 1.4728385210037231,
      "learning_rate": 6.2105e-05,
      "loss": 0.5204,
      "step": 12428
    },
    {
      "epoch": 0.6959905924515624,
      "grad_norm": 1.3263381719589233,
      "learning_rate": 6.211e-05,
      "loss": 0.4777,
      "step": 12429
    },
    {
      "epoch": 0.6960465897636914,
      "grad_norm": 1.2977200746536255,
      "learning_rate": 6.2115e-05,
      "loss": 0.3269,
      "step": 12430
    },
    {
      "epoch": 0.6961025870758204,
      "grad_norm": 1.1716684103012085,
      "learning_rate": 6.212e-05,
      "loss": 0.4632,
      "step": 12431
    },
    {
      "epoch": 0.6961585843879494,
      "grad_norm": 1.213047981262207,
      "learning_rate": 6.2125e-05,
      "loss": 0.5351,
      "step": 12432
    },
    {
      "epoch": 0.6962145817000784,
      "grad_norm": 1.2444779872894287,
      "learning_rate": 6.213e-05,
      "loss": 0.3439,
      "step": 12433
    },
    {
      "epoch": 0.6962705790122075,
      "grad_norm": 1.1619069576263428,
      "learning_rate": 6.2135e-05,
      "loss": 0.4513,
      "step": 12434
    },
    {
      "epoch": 0.6963265763243365,
      "grad_norm": 1.0866138935089111,
      "learning_rate": 6.214e-05,
      "loss": 0.3634,
      "step": 12435
    },
    {
      "epoch": 0.6963825736364655,
      "grad_norm": 1.5795514583587646,
      "learning_rate": 6.2145e-05,
      "loss": 0.6332,
      "step": 12436
    },
    {
      "epoch": 0.6964385709485945,
      "grad_norm": 1.434275507926941,
      "learning_rate": 6.215e-05,
      "loss": 0.3181,
      "step": 12437
    },
    {
      "epoch": 0.6964945682607235,
      "grad_norm": 1.4398802518844604,
      "learning_rate": 6.2155e-05,
      "loss": 0.4559,
      "step": 12438
    },
    {
      "epoch": 0.6965505655728526,
      "grad_norm": 1.1475560665130615,
      "learning_rate": 6.216e-05,
      "loss": 0.3817,
      "step": 12439
    },
    {
      "epoch": 0.6966065628849816,
      "grad_norm": 1.1969971656799316,
      "learning_rate": 6.216500000000001e-05,
      "loss": 0.4641,
      "step": 12440
    },
    {
      "epoch": 0.6966625601971106,
      "grad_norm": 1.2024345397949219,
      "learning_rate": 6.217000000000001e-05,
      "loss": 0.4557,
      "step": 12441
    },
    {
      "epoch": 0.6967185575092396,
      "grad_norm": 1.4936630725860596,
      "learning_rate": 6.217500000000001e-05,
      "loss": 0.4337,
      "step": 12442
    },
    {
      "epoch": 0.6967745548213685,
      "grad_norm": 1.4987143278121948,
      "learning_rate": 6.218e-05,
      "loss": 0.4671,
      "step": 12443
    },
    {
      "epoch": 0.6968305521334975,
      "grad_norm": 1.2767128944396973,
      "learning_rate": 6.2185e-05,
      "loss": 0.4376,
      "step": 12444
    },
    {
      "epoch": 0.6968865494456266,
      "grad_norm": 1.389953851699829,
      "learning_rate": 6.219e-05,
      "loss": 0.3997,
      "step": 12445
    },
    {
      "epoch": 0.6969425467577556,
      "grad_norm": 1.530099868774414,
      "learning_rate": 6.2195e-05,
      "loss": 0.3824,
      "step": 12446
    },
    {
      "epoch": 0.6969985440698846,
      "grad_norm": 1.2455320358276367,
      "learning_rate": 6.220000000000001e-05,
      "loss": 0.5118,
      "step": 12447
    },
    {
      "epoch": 0.6970545413820136,
      "grad_norm": 1.564921259880066,
      "learning_rate": 6.2205e-05,
      "loss": 0.5814,
      "step": 12448
    },
    {
      "epoch": 0.6971105386941426,
      "grad_norm": 1.4247167110443115,
      "learning_rate": 6.221e-05,
      "loss": 0.35,
      "step": 12449
    },
    {
      "epoch": 0.6971665360062717,
      "grad_norm": 1.2676891088485718,
      "learning_rate": 6.2215e-05,
      "loss": 0.2909,
      "step": 12450
    },
    {
      "epoch": 0.6972225333184007,
      "grad_norm": 1.6524531841278076,
      "learning_rate": 6.222e-05,
      "loss": 0.4867,
      "step": 12451
    },
    {
      "epoch": 0.6972785306305297,
      "grad_norm": 1.3316035270690918,
      "learning_rate": 6.2225e-05,
      "loss": 0.383,
      "step": 12452
    },
    {
      "epoch": 0.6973345279426587,
      "grad_norm": 1.362134337425232,
      "learning_rate": 6.223e-05,
      "loss": 0.4233,
      "step": 12453
    },
    {
      "epoch": 0.6973905252547877,
      "grad_norm": 1.4702951908111572,
      "learning_rate": 6.2235e-05,
      "loss": 0.4932,
      "step": 12454
    },
    {
      "epoch": 0.6974465225669167,
      "grad_norm": 1.5142104625701904,
      "learning_rate": 6.224e-05,
      "loss": 0.4368,
      "step": 12455
    },
    {
      "epoch": 0.6975025198790458,
      "grad_norm": 1.3793469667434692,
      "learning_rate": 6.2245e-05,
      "loss": 0.3769,
      "step": 12456
    },
    {
      "epoch": 0.6975585171911748,
      "grad_norm": 1.3909077644348145,
      "learning_rate": 6.225000000000001e-05,
      "loss": 0.4614,
      "step": 12457
    },
    {
      "epoch": 0.6976145145033038,
      "grad_norm": 1.0924532413482666,
      "learning_rate": 6.2255e-05,
      "loss": 0.4464,
      "step": 12458
    },
    {
      "epoch": 0.6976705118154328,
      "grad_norm": 1.3595489263534546,
      "learning_rate": 6.226e-05,
      "loss": 0.3865,
      "step": 12459
    },
    {
      "epoch": 0.6977265091275618,
      "grad_norm": 1.1982531547546387,
      "learning_rate": 6.2265e-05,
      "loss": 0.3582,
      "step": 12460
    },
    {
      "epoch": 0.6977825064396909,
      "grad_norm": 1.8145815134048462,
      "learning_rate": 6.227000000000001e-05,
      "loss": 0.5622,
      "step": 12461
    },
    {
      "epoch": 0.6978385037518199,
      "grad_norm": 1.2581560611724854,
      "learning_rate": 6.227500000000001e-05,
      "loss": 0.3973,
      "step": 12462
    },
    {
      "epoch": 0.6978945010639489,
      "grad_norm": 1.8066720962524414,
      "learning_rate": 6.228000000000001e-05,
      "loss": 0.4065,
      "step": 12463
    },
    {
      "epoch": 0.6979504983760779,
      "grad_norm": 1.3089772462844849,
      "learning_rate": 6.2285e-05,
      "loss": 0.5027,
      "step": 12464
    },
    {
      "epoch": 0.6980064956882069,
      "grad_norm": 1.2001739740371704,
      "learning_rate": 6.229e-05,
      "loss": 0.4057,
      "step": 12465
    },
    {
      "epoch": 0.698062493000336,
      "grad_norm": 1.4888426065444946,
      "learning_rate": 6.2295e-05,
      "loss": 0.5365,
      "step": 12466
    },
    {
      "epoch": 0.698118490312465,
      "grad_norm": 1.1244386434555054,
      "learning_rate": 6.23e-05,
      "loss": 0.4614,
      "step": 12467
    },
    {
      "epoch": 0.698174487624594,
      "grad_norm": 1.2285563945770264,
      "learning_rate": 6.230500000000001e-05,
      "loss": 0.4636,
      "step": 12468
    },
    {
      "epoch": 0.698230484936723,
      "grad_norm": 1.3267616033554077,
      "learning_rate": 6.231e-05,
      "loss": 0.4245,
      "step": 12469
    },
    {
      "epoch": 0.698286482248852,
      "grad_norm": 1.3063656091690063,
      "learning_rate": 6.2315e-05,
      "loss": 0.3789,
      "step": 12470
    },
    {
      "epoch": 0.698342479560981,
      "grad_norm": 1.3477351665496826,
      "learning_rate": 6.232e-05,
      "loss": 0.5748,
      "step": 12471
    },
    {
      "epoch": 0.6983984768731101,
      "grad_norm": 1.2163971662521362,
      "learning_rate": 6.2325e-05,
      "loss": 0.4104,
      "step": 12472
    },
    {
      "epoch": 0.6984544741852391,
      "grad_norm": 1.1639045476913452,
      "learning_rate": 6.233e-05,
      "loss": 0.4309,
      "step": 12473
    },
    {
      "epoch": 0.6985104714973681,
      "grad_norm": 1.8191872835159302,
      "learning_rate": 6.2335e-05,
      "loss": 0.478,
      "step": 12474
    },
    {
      "epoch": 0.6985664688094971,
      "grad_norm": 1.1949455738067627,
      "learning_rate": 6.234e-05,
      "loss": 0.3786,
      "step": 12475
    },
    {
      "epoch": 0.6986224661216262,
      "grad_norm": 1.2190887928009033,
      "learning_rate": 6.2345e-05,
      "loss": 0.5706,
      "step": 12476
    },
    {
      "epoch": 0.6986784634337552,
      "grad_norm": 1.1594059467315674,
      "learning_rate": 6.235000000000001e-05,
      "loss": 0.4136,
      "step": 12477
    },
    {
      "epoch": 0.6987344607458842,
      "grad_norm": 1.2436951398849487,
      "learning_rate": 6.235500000000001e-05,
      "loss": 0.4618,
      "step": 12478
    },
    {
      "epoch": 0.6987904580580132,
      "grad_norm": 1.2693654298782349,
      "learning_rate": 6.236e-05,
      "loss": 0.4969,
      "step": 12479
    },
    {
      "epoch": 0.6988464553701422,
      "grad_norm": 1.348768949508667,
      "learning_rate": 6.2365e-05,
      "loss": 0.518,
      "step": 12480
    },
    {
      "epoch": 0.6989024526822712,
      "grad_norm": 1.1391533613204956,
      "learning_rate": 6.237e-05,
      "loss": 0.4162,
      "step": 12481
    },
    {
      "epoch": 0.6989584499944003,
      "grad_norm": 1.248640537261963,
      "learning_rate": 6.237500000000001e-05,
      "loss": 0.3715,
      "step": 12482
    },
    {
      "epoch": 0.6990144473065293,
      "grad_norm": 1.3822144269943237,
      "learning_rate": 6.238000000000001e-05,
      "loss": 0.4397,
      "step": 12483
    },
    {
      "epoch": 0.6990704446186583,
      "grad_norm": 1.16531503200531,
      "learning_rate": 6.2385e-05,
      "loss": 0.4137,
      "step": 12484
    },
    {
      "epoch": 0.6991264419307873,
      "grad_norm": 1.2543193101882935,
      "learning_rate": 6.239e-05,
      "loss": 0.5052,
      "step": 12485
    },
    {
      "epoch": 0.6991824392429163,
      "grad_norm": 2.1988441944122314,
      "learning_rate": 6.2395e-05,
      "loss": 0.6602,
      "step": 12486
    },
    {
      "epoch": 0.6992384365550454,
      "grad_norm": 1.329607367515564,
      "learning_rate": 6.24e-05,
      "loss": 0.3931,
      "step": 12487
    },
    {
      "epoch": 0.6992944338671744,
      "grad_norm": 1.3688476085662842,
      "learning_rate": 6.2405e-05,
      "loss": 0.4576,
      "step": 12488
    },
    {
      "epoch": 0.6993504311793034,
      "grad_norm": 1.3245147466659546,
      "learning_rate": 6.241000000000001e-05,
      "loss": 0.5143,
      "step": 12489
    },
    {
      "epoch": 0.6994064284914324,
      "grad_norm": 1.143288016319275,
      "learning_rate": 6.2415e-05,
      "loss": 0.5049,
      "step": 12490
    },
    {
      "epoch": 0.6994624258035614,
      "grad_norm": 1.5038130283355713,
      "learning_rate": 6.242e-05,
      "loss": 0.6721,
      "step": 12491
    },
    {
      "epoch": 0.6995184231156905,
      "grad_norm": 1.218773365020752,
      "learning_rate": 6.2425e-05,
      "loss": 0.475,
      "step": 12492
    },
    {
      "epoch": 0.6995744204278195,
      "grad_norm": 1.2821155786514282,
      "learning_rate": 6.243e-05,
      "loss": 0.396,
      "step": 12493
    },
    {
      "epoch": 0.6996304177399485,
      "grad_norm": 1.1763544082641602,
      "learning_rate": 6.2435e-05,
      "loss": 0.4884,
      "step": 12494
    },
    {
      "epoch": 0.6996864150520775,
      "grad_norm": 1.1693071126937866,
      "learning_rate": 6.244e-05,
      "loss": 0.3576,
      "step": 12495
    },
    {
      "epoch": 0.6997424123642065,
      "grad_norm": 1.5130113363265991,
      "learning_rate": 6.2445e-05,
      "loss": 0.4434,
      "step": 12496
    },
    {
      "epoch": 0.6997984096763356,
      "grad_norm": 1.4724360704421997,
      "learning_rate": 6.245000000000001e-05,
      "loss": 0.4186,
      "step": 12497
    },
    {
      "epoch": 0.6998544069884646,
      "grad_norm": 1.420949935913086,
      "learning_rate": 6.245500000000001e-05,
      "loss": 0.4128,
      "step": 12498
    },
    {
      "epoch": 0.6999104043005936,
      "grad_norm": 1.2526235580444336,
      "learning_rate": 6.246000000000001e-05,
      "loss": 0.3879,
      "step": 12499
    },
    {
      "epoch": 0.6999664016127226,
      "grad_norm": 1.2274563312530518,
      "learning_rate": 6.2465e-05,
      "loss": 0.3422,
      "step": 12500
    },
    {
      "epoch": 0.7000223989248516,
      "grad_norm": 1.237396240234375,
      "learning_rate": 6.247e-05,
      "loss": 0.3423,
      "step": 12501
    },
    {
      "epoch": 0.7000783962369806,
      "grad_norm": 1.1403456926345825,
      "learning_rate": 6.2475e-05,
      "loss": 0.4302,
      "step": 12502
    },
    {
      "epoch": 0.7001343935491097,
      "grad_norm": 1.3087342977523804,
      "learning_rate": 6.248000000000001e-05,
      "loss": 0.3942,
      "step": 12503
    },
    {
      "epoch": 0.7001903908612387,
      "grad_norm": 1.552927851676941,
      "learning_rate": 6.248500000000001e-05,
      "loss": 0.6173,
      "step": 12504
    },
    {
      "epoch": 0.7002463881733677,
      "grad_norm": 1.3391687870025635,
      "learning_rate": 6.249e-05,
      "loss": 0.5829,
      "step": 12505
    },
    {
      "epoch": 0.7003023854854967,
      "grad_norm": 1.5082244873046875,
      "learning_rate": 6.2495e-05,
      "loss": 0.4968,
      "step": 12506
    },
    {
      "epoch": 0.7003583827976257,
      "grad_norm": 1.4176688194274902,
      "learning_rate": 6.25e-05,
      "loss": 0.4616,
      "step": 12507
    },
    {
      "epoch": 0.7004143801097548,
      "grad_norm": 1.3099764585494995,
      "learning_rate": 6.2505e-05,
      "loss": 0.437,
      "step": 12508
    },
    {
      "epoch": 0.7004703774218838,
      "grad_norm": 1.3941704034805298,
      "learning_rate": 6.251e-05,
      "loss": 0.3963,
      "step": 12509
    },
    {
      "epoch": 0.7005263747340128,
      "grad_norm": 1.2004023790359497,
      "learning_rate": 6.251500000000001e-05,
      "loss": 0.3655,
      "step": 12510
    },
    {
      "epoch": 0.7005823720461418,
      "grad_norm": 1.379607081413269,
      "learning_rate": 6.252e-05,
      "loss": 0.5685,
      "step": 12511
    },
    {
      "epoch": 0.7006383693582708,
      "grad_norm": 1.3818296194076538,
      "learning_rate": 6.2525e-05,
      "loss": 0.3594,
      "step": 12512
    },
    {
      "epoch": 0.7006943666703999,
      "grad_norm": 1.1807204484939575,
      "learning_rate": 6.253e-05,
      "loss": 0.3541,
      "step": 12513
    },
    {
      "epoch": 0.7007503639825289,
      "grad_norm": 1.0599114894866943,
      "learning_rate": 6.2535e-05,
      "loss": 0.3048,
      "step": 12514
    },
    {
      "epoch": 0.7008063612946579,
      "grad_norm": 1.2263143062591553,
      "learning_rate": 6.254e-05,
      "loss": 0.4669,
      "step": 12515
    },
    {
      "epoch": 0.7008623586067869,
      "grad_norm": 2.299558639526367,
      "learning_rate": 6.254499999999999e-05,
      "loss": 0.5666,
      "step": 12516
    },
    {
      "epoch": 0.7009183559189159,
      "grad_norm": 1.2690045833587646,
      "learning_rate": 6.255e-05,
      "loss": 0.3804,
      "step": 12517
    },
    {
      "epoch": 0.700974353231045,
      "grad_norm": 1.2532697916030884,
      "learning_rate": 6.255500000000001e-05,
      "loss": 0.498,
      "step": 12518
    },
    {
      "epoch": 0.701030350543174,
      "grad_norm": 1.5119946002960205,
      "learning_rate": 6.256000000000001e-05,
      "loss": 0.5361,
      "step": 12519
    },
    {
      "epoch": 0.701086347855303,
      "grad_norm": 1.0479772090911865,
      "learning_rate": 6.256500000000001e-05,
      "loss": 0.3354,
      "step": 12520
    },
    {
      "epoch": 0.701142345167432,
      "grad_norm": 1.426035761833191,
      "learning_rate": 6.257e-05,
      "loss": 0.329,
      "step": 12521
    },
    {
      "epoch": 0.701198342479561,
      "grad_norm": 1.311549425125122,
      "learning_rate": 6.2575e-05,
      "loss": 0.4817,
      "step": 12522
    },
    {
      "epoch": 0.70125433979169,
      "grad_norm": 1.2965736389160156,
      "learning_rate": 6.258e-05,
      "loss": 0.6261,
      "step": 12523
    },
    {
      "epoch": 0.7013103371038191,
      "grad_norm": 1.3294415473937988,
      "learning_rate": 6.258500000000001e-05,
      "loss": 0.4701,
      "step": 12524
    },
    {
      "epoch": 0.7013663344159481,
      "grad_norm": 1.2690424919128418,
      "learning_rate": 6.259000000000001e-05,
      "loss": 0.3445,
      "step": 12525
    },
    {
      "epoch": 0.701422331728077,
      "grad_norm": 1.2250972986221313,
      "learning_rate": 6.2595e-05,
      "loss": 0.3364,
      "step": 12526
    },
    {
      "epoch": 0.701478329040206,
      "grad_norm": 1.245665431022644,
      "learning_rate": 6.26e-05,
      "loss": 0.4073,
      "step": 12527
    },
    {
      "epoch": 0.701534326352335,
      "grad_norm": 2.446558713912964,
      "learning_rate": 6.2605e-05,
      "loss": 0.4662,
      "step": 12528
    },
    {
      "epoch": 0.701590323664464,
      "grad_norm": 1.2567524909973145,
      "learning_rate": 6.261e-05,
      "loss": 0.3178,
      "step": 12529
    },
    {
      "epoch": 0.7016463209765931,
      "grad_norm": 1.420933485031128,
      "learning_rate": 6.2615e-05,
      "loss": 0.387,
      "step": 12530
    },
    {
      "epoch": 0.7017023182887221,
      "grad_norm": 1.4116603136062622,
      "learning_rate": 6.262000000000001e-05,
      "loss": 0.4641,
      "step": 12531
    },
    {
      "epoch": 0.7017583156008511,
      "grad_norm": 1.1638530492782593,
      "learning_rate": 6.2625e-05,
      "loss": 0.4309,
      "step": 12532
    },
    {
      "epoch": 0.7018143129129801,
      "grad_norm": 1.426966667175293,
      "learning_rate": 6.263e-05,
      "loss": 0.4635,
      "step": 12533
    },
    {
      "epoch": 0.7018703102251092,
      "grad_norm": 1.309417963027954,
      "learning_rate": 6.2635e-05,
      "loss": 0.452,
      "step": 12534
    },
    {
      "epoch": 0.7019263075372382,
      "grad_norm": 1.2854201793670654,
      "learning_rate": 6.264e-05,
      "loss": 0.4823,
      "step": 12535
    },
    {
      "epoch": 0.7019823048493672,
      "grad_norm": 1.50555419921875,
      "learning_rate": 6.2645e-05,
      "loss": 0.4936,
      "step": 12536
    },
    {
      "epoch": 0.7020383021614962,
      "grad_norm": 1.2499947547912598,
      "learning_rate": 6.264999999999999e-05,
      "loss": 0.4279,
      "step": 12537
    },
    {
      "epoch": 0.7020942994736252,
      "grad_norm": 1.2926799058914185,
      "learning_rate": 6.2655e-05,
      "loss": 0.2997,
      "step": 12538
    },
    {
      "epoch": 0.7021502967857542,
      "grad_norm": 1.2959471940994263,
      "learning_rate": 6.266000000000001e-05,
      "loss": 0.3984,
      "step": 12539
    },
    {
      "epoch": 0.7022062940978833,
      "grad_norm": 1.2488656044006348,
      "learning_rate": 6.266500000000001e-05,
      "loss": 0.4622,
      "step": 12540
    },
    {
      "epoch": 0.7022622914100123,
      "grad_norm": 1.310998797416687,
      "learning_rate": 6.267000000000001e-05,
      "loss": 0.4812,
      "step": 12541
    },
    {
      "epoch": 0.7023182887221413,
      "grad_norm": 1.280368685722351,
      "learning_rate": 6.2675e-05,
      "loss": 0.4793,
      "step": 12542
    },
    {
      "epoch": 0.7023742860342703,
      "grad_norm": 1.4046286344528198,
      "learning_rate": 6.268e-05,
      "loss": 0.3768,
      "step": 12543
    },
    {
      "epoch": 0.7024302833463993,
      "grad_norm": 1.514461636543274,
      "learning_rate": 6.2685e-05,
      "loss": 0.3981,
      "step": 12544
    },
    {
      "epoch": 0.7024862806585284,
      "grad_norm": 1.2564232349395752,
      "learning_rate": 6.269e-05,
      "loss": 0.4399,
      "step": 12545
    },
    {
      "epoch": 0.7025422779706574,
      "grad_norm": 1.1517951488494873,
      "learning_rate": 6.269500000000001e-05,
      "loss": 0.424,
      "step": 12546
    },
    {
      "epoch": 0.7025982752827864,
      "grad_norm": 1.3670238256454468,
      "learning_rate": 6.27e-05,
      "loss": 0.5629,
      "step": 12547
    },
    {
      "epoch": 0.7026542725949154,
      "grad_norm": 1.3139759302139282,
      "learning_rate": 6.2705e-05,
      "loss": 0.4085,
      "step": 12548
    },
    {
      "epoch": 0.7027102699070444,
      "grad_norm": 1.2447679042816162,
      "learning_rate": 6.271e-05,
      "loss": 0.5478,
      "step": 12549
    },
    {
      "epoch": 0.7027662672191735,
      "grad_norm": 1.2819429636001587,
      "learning_rate": 6.2715e-05,
      "loss": 0.3848,
      "step": 12550
    },
    {
      "epoch": 0.7028222645313025,
      "grad_norm": 3.032064437866211,
      "learning_rate": 6.272e-05,
      "loss": 0.4426,
      "step": 12551
    },
    {
      "epoch": 0.7028782618434315,
      "grad_norm": 1.4964185953140259,
      "learning_rate": 6.2725e-05,
      "loss": 0.516,
      "step": 12552
    },
    {
      "epoch": 0.7029342591555605,
      "grad_norm": 1.7251659631729126,
      "learning_rate": 6.273e-05,
      "loss": 0.4446,
      "step": 12553
    },
    {
      "epoch": 0.7029902564676895,
      "grad_norm": 1.5610597133636475,
      "learning_rate": 6.2735e-05,
      "loss": 0.5465,
      "step": 12554
    },
    {
      "epoch": 0.7030462537798186,
      "grad_norm": 1.3260618448257446,
      "learning_rate": 6.274e-05,
      "loss": 0.49,
      "step": 12555
    },
    {
      "epoch": 0.7031022510919476,
      "grad_norm": 1.0873066186904907,
      "learning_rate": 6.2745e-05,
      "loss": 0.3162,
      "step": 12556
    },
    {
      "epoch": 0.7031582484040766,
      "grad_norm": 1.1887412071228027,
      "learning_rate": 6.275e-05,
      "loss": 0.3976,
      "step": 12557
    },
    {
      "epoch": 0.7032142457162056,
      "grad_norm": 1.362644076347351,
      "learning_rate": 6.2755e-05,
      "loss": 0.4904,
      "step": 12558
    },
    {
      "epoch": 0.7032702430283346,
      "grad_norm": 1.2175319194793701,
      "learning_rate": 6.276e-05,
      "loss": 0.4208,
      "step": 12559
    },
    {
      "epoch": 0.7033262403404636,
      "grad_norm": 1.2674423456192017,
      "learning_rate": 6.276500000000001e-05,
      "loss": 0.3864,
      "step": 12560
    },
    {
      "epoch": 0.7033822376525927,
      "grad_norm": 1.1710255146026611,
      "learning_rate": 6.277000000000001e-05,
      "loss": 0.4219,
      "step": 12561
    },
    {
      "epoch": 0.7034382349647217,
      "grad_norm": 1.1940264701843262,
      "learning_rate": 6.277500000000001e-05,
      "loss": 0.3946,
      "step": 12562
    },
    {
      "epoch": 0.7034942322768507,
      "grad_norm": 1.3122938871383667,
      "learning_rate": 6.278e-05,
      "loss": 0.3898,
      "step": 12563
    },
    {
      "epoch": 0.7035502295889797,
      "grad_norm": 1.266348958015442,
      "learning_rate": 6.2785e-05,
      "loss": 0.5555,
      "step": 12564
    },
    {
      "epoch": 0.7036062269011087,
      "grad_norm": 1.386495590209961,
      "learning_rate": 6.279e-05,
      "loss": 0.4076,
      "step": 12565
    },
    {
      "epoch": 0.7036622242132378,
      "grad_norm": 1.2832168340682983,
      "learning_rate": 6.2795e-05,
      "loss": 0.4575,
      "step": 12566
    },
    {
      "epoch": 0.7037182215253668,
      "grad_norm": 1.2073702812194824,
      "learning_rate": 6.280000000000001e-05,
      "loss": 0.3843,
      "step": 12567
    },
    {
      "epoch": 0.7037742188374958,
      "grad_norm": 1.2366464138031006,
      "learning_rate": 6.2805e-05,
      "loss": 0.4584,
      "step": 12568
    },
    {
      "epoch": 0.7038302161496248,
      "grad_norm": 1.237715482711792,
      "learning_rate": 6.281e-05,
      "loss": 0.446,
      "step": 12569
    },
    {
      "epoch": 0.7038862134617538,
      "grad_norm": 1.3560377359390259,
      "learning_rate": 6.2815e-05,
      "loss": 0.376,
      "step": 12570
    },
    {
      "epoch": 0.7039422107738829,
      "grad_norm": 1.4787386655807495,
      "learning_rate": 6.282e-05,
      "loss": 0.4276,
      "step": 12571
    },
    {
      "epoch": 0.7039982080860119,
      "grad_norm": 1.650567889213562,
      "learning_rate": 6.2825e-05,
      "loss": 0.5232,
      "step": 12572
    },
    {
      "epoch": 0.7040542053981409,
      "grad_norm": 1.2303379774093628,
      "learning_rate": 6.283e-05,
      "loss": 0.3249,
      "step": 12573
    },
    {
      "epoch": 0.7041102027102699,
      "grad_norm": 1.3153318166732788,
      "learning_rate": 6.2835e-05,
      "loss": 0.5754,
      "step": 12574
    },
    {
      "epoch": 0.7041662000223989,
      "grad_norm": 1.261511206626892,
      "learning_rate": 6.284e-05,
      "loss": 0.3882,
      "step": 12575
    },
    {
      "epoch": 0.704222197334528,
      "grad_norm": 1.7436730861663818,
      "learning_rate": 6.2845e-05,
      "loss": 0.5699,
      "step": 12576
    },
    {
      "epoch": 0.704278194646657,
      "grad_norm": 1.4731782674789429,
      "learning_rate": 6.285e-05,
      "loss": 0.4778,
      "step": 12577
    },
    {
      "epoch": 0.704334191958786,
      "grad_norm": 1.2351558208465576,
      "learning_rate": 6.285500000000001e-05,
      "loss": 0.4492,
      "step": 12578
    },
    {
      "epoch": 0.704390189270915,
      "grad_norm": 1.2724463939666748,
      "learning_rate": 6.286e-05,
      "loss": 0.3817,
      "step": 12579
    },
    {
      "epoch": 0.704446186583044,
      "grad_norm": 1.1775709390640259,
      "learning_rate": 6.2865e-05,
      "loss": 0.4378,
      "step": 12580
    },
    {
      "epoch": 0.704502183895173,
      "grad_norm": 1.4919737577438354,
      "learning_rate": 6.287000000000001e-05,
      "loss": 0.4648,
      "step": 12581
    },
    {
      "epoch": 0.7045581812073021,
      "grad_norm": 1.245802402496338,
      "learning_rate": 6.287500000000001e-05,
      "loss": 0.373,
      "step": 12582
    },
    {
      "epoch": 0.7046141785194311,
      "grad_norm": 1.3589507341384888,
      "learning_rate": 6.288000000000001e-05,
      "loss": 0.5279,
      "step": 12583
    },
    {
      "epoch": 0.7046701758315601,
      "grad_norm": 1.2483769655227661,
      "learning_rate": 6.2885e-05,
      "loss": 0.3568,
      "step": 12584
    },
    {
      "epoch": 0.7047261731436891,
      "grad_norm": 1.0464344024658203,
      "learning_rate": 6.289e-05,
      "loss": 0.4032,
      "step": 12585
    },
    {
      "epoch": 0.7047821704558181,
      "grad_norm": 1.4985077381134033,
      "learning_rate": 6.2895e-05,
      "loss": 0.4843,
      "step": 12586
    },
    {
      "epoch": 0.7048381677679472,
      "grad_norm": 1.618809461593628,
      "learning_rate": 6.29e-05,
      "loss": 0.4632,
      "step": 12587
    },
    {
      "epoch": 0.7048941650800762,
      "grad_norm": 1.4173755645751953,
      "learning_rate": 6.290500000000001e-05,
      "loss": 0.4738,
      "step": 12588
    },
    {
      "epoch": 0.7049501623922052,
      "grad_norm": 1.209714412689209,
      "learning_rate": 6.291e-05,
      "loss": 0.3928,
      "step": 12589
    },
    {
      "epoch": 0.7050061597043342,
      "grad_norm": 1.331560730934143,
      "learning_rate": 6.2915e-05,
      "loss": 0.4719,
      "step": 12590
    },
    {
      "epoch": 0.7050621570164632,
      "grad_norm": 1.475605845451355,
      "learning_rate": 6.292e-05,
      "loss": 0.5298,
      "step": 12591
    },
    {
      "epoch": 0.7051181543285923,
      "grad_norm": 1.38764488697052,
      "learning_rate": 6.2925e-05,
      "loss": 0.4586,
      "step": 12592
    },
    {
      "epoch": 0.7051741516407213,
      "grad_norm": 1.2886463403701782,
      "learning_rate": 6.293e-05,
      "loss": 0.4298,
      "step": 12593
    },
    {
      "epoch": 0.7052301489528503,
      "grad_norm": 1.1384798288345337,
      "learning_rate": 6.293499999999999e-05,
      "loss": 0.4246,
      "step": 12594
    },
    {
      "epoch": 0.7052861462649793,
      "grad_norm": 1.3754823207855225,
      "learning_rate": 6.294e-05,
      "loss": 0.4228,
      "step": 12595
    },
    {
      "epoch": 0.7053421435771083,
      "grad_norm": 1.4134275913238525,
      "learning_rate": 6.2945e-05,
      "loss": 0.597,
      "step": 12596
    },
    {
      "epoch": 0.7053981408892374,
      "grad_norm": 1.2388256788253784,
      "learning_rate": 6.295e-05,
      "loss": 0.4314,
      "step": 12597
    },
    {
      "epoch": 0.7054541382013664,
      "grad_norm": 1.4481347799301147,
      "learning_rate": 6.295500000000001e-05,
      "loss": 0.4934,
      "step": 12598
    },
    {
      "epoch": 0.7055101355134954,
      "grad_norm": 1.091850757598877,
      "learning_rate": 6.296000000000001e-05,
      "loss": 0.3545,
      "step": 12599
    },
    {
      "epoch": 0.7055661328256244,
      "grad_norm": 1.3457605838775635,
      "learning_rate": 6.2965e-05,
      "loss": 0.5096,
      "step": 12600
    },
    {
      "epoch": 0.7056221301377534,
      "grad_norm": 1.4634277820587158,
      "learning_rate": 6.297e-05,
      "loss": 0.4189,
      "step": 12601
    },
    {
      "epoch": 0.7056781274498825,
      "grad_norm": 1.5778405666351318,
      "learning_rate": 6.297500000000001e-05,
      "loss": 0.492,
      "step": 12602
    },
    {
      "epoch": 0.7057341247620115,
      "grad_norm": 1.2554309368133545,
      "learning_rate": 6.298000000000001e-05,
      "loss": 0.3473,
      "step": 12603
    },
    {
      "epoch": 0.7057901220741405,
      "grad_norm": 1.4002500772476196,
      "learning_rate": 6.298500000000001e-05,
      "loss": 0.5078,
      "step": 12604
    },
    {
      "epoch": 0.7058461193862695,
      "grad_norm": 1.460457444190979,
      "learning_rate": 6.299e-05,
      "loss": 0.3477,
      "step": 12605
    },
    {
      "epoch": 0.7059021166983985,
      "grad_norm": 1.3724795579910278,
      "learning_rate": 6.2995e-05,
      "loss": 0.4048,
      "step": 12606
    },
    {
      "epoch": 0.7059581140105275,
      "grad_norm": 1.3621898889541626,
      "learning_rate": 6.3e-05,
      "loss": 0.5058,
      "step": 12607
    },
    {
      "epoch": 0.7060141113226566,
      "grad_norm": 1.2200628519058228,
      "learning_rate": 6.3005e-05,
      "loss": 0.4472,
      "step": 12608
    },
    {
      "epoch": 0.7060701086347855,
      "grad_norm": 1.37726628780365,
      "learning_rate": 6.301000000000001e-05,
      "loss": 0.4492,
      "step": 12609
    },
    {
      "epoch": 0.7061261059469145,
      "grad_norm": 1.4450502395629883,
      "learning_rate": 6.3015e-05,
      "loss": 0.4134,
      "step": 12610
    },
    {
      "epoch": 0.7061821032590435,
      "grad_norm": 1.2746467590332031,
      "learning_rate": 6.302e-05,
      "loss": 0.4434,
      "step": 12611
    },
    {
      "epoch": 0.7062381005711725,
      "grad_norm": 1.3004050254821777,
      "learning_rate": 6.3025e-05,
      "loss": 0.4678,
      "step": 12612
    },
    {
      "epoch": 0.7062940978833016,
      "grad_norm": 1.4135793447494507,
      "learning_rate": 6.303e-05,
      "loss": 0.505,
      "step": 12613
    },
    {
      "epoch": 0.7063500951954306,
      "grad_norm": 1.072414755821228,
      "learning_rate": 6.3035e-05,
      "loss": 0.3845,
      "step": 12614
    },
    {
      "epoch": 0.7064060925075596,
      "grad_norm": 1.2727646827697754,
      "learning_rate": 6.303999999999999e-05,
      "loss": 0.5119,
      "step": 12615
    },
    {
      "epoch": 0.7064620898196886,
      "grad_norm": 1.1109548807144165,
      "learning_rate": 6.3045e-05,
      "loss": 0.3195,
      "step": 12616
    },
    {
      "epoch": 0.7065180871318176,
      "grad_norm": 1.3027209043502808,
      "learning_rate": 6.305e-05,
      "loss": 0.5456,
      "step": 12617
    },
    {
      "epoch": 0.7065740844439466,
      "grad_norm": 1.3179601430892944,
      "learning_rate": 6.305500000000001e-05,
      "loss": 0.4321,
      "step": 12618
    },
    {
      "epoch": 0.7066300817560757,
      "grad_norm": 1.138843059539795,
      "learning_rate": 6.306000000000001e-05,
      "loss": 0.4206,
      "step": 12619
    },
    {
      "epoch": 0.7066860790682047,
      "grad_norm": 1.6393864154815674,
      "learning_rate": 6.306500000000001e-05,
      "loss": 0.6926,
      "step": 12620
    },
    {
      "epoch": 0.7067420763803337,
      "grad_norm": 1.7525949478149414,
      "learning_rate": 6.307e-05,
      "loss": 0.4402,
      "step": 12621
    },
    {
      "epoch": 0.7067980736924627,
      "grad_norm": 1.4333950281143188,
      "learning_rate": 6.3075e-05,
      "loss": 0.4057,
      "step": 12622
    },
    {
      "epoch": 0.7068540710045917,
      "grad_norm": 1.2789214849472046,
      "learning_rate": 6.308e-05,
      "loss": 0.3135,
      "step": 12623
    },
    {
      "epoch": 0.7069100683167208,
      "grad_norm": 1.5111428499221802,
      "learning_rate": 6.308500000000001e-05,
      "loss": 0.3997,
      "step": 12624
    },
    {
      "epoch": 0.7069660656288498,
      "grad_norm": 1.579714298248291,
      "learning_rate": 6.309000000000001e-05,
      "loss": 0.4526,
      "step": 12625
    },
    {
      "epoch": 0.7070220629409788,
      "grad_norm": 1.551098346710205,
      "learning_rate": 6.3095e-05,
      "loss": 0.4695,
      "step": 12626
    },
    {
      "epoch": 0.7070780602531078,
      "grad_norm": 1.2486560344696045,
      "learning_rate": 6.31e-05,
      "loss": 0.4226,
      "step": 12627
    },
    {
      "epoch": 0.7071340575652368,
      "grad_norm": 1.126120924949646,
      "learning_rate": 6.3105e-05,
      "loss": 0.3612,
      "step": 12628
    },
    {
      "epoch": 0.7071900548773659,
      "grad_norm": 1.2583171129226685,
      "learning_rate": 6.311e-05,
      "loss": 0.3793,
      "step": 12629
    },
    {
      "epoch": 0.7072460521894949,
      "grad_norm": 1.390459656715393,
      "learning_rate": 6.311500000000001e-05,
      "loss": 0.4242,
      "step": 12630
    },
    {
      "epoch": 0.7073020495016239,
      "grad_norm": 1.1530574560165405,
      "learning_rate": 6.312e-05,
      "loss": 0.4057,
      "step": 12631
    },
    {
      "epoch": 0.7073580468137529,
      "grad_norm": 1.2216347455978394,
      "learning_rate": 6.3125e-05,
      "loss": 0.3678,
      "step": 12632
    },
    {
      "epoch": 0.7074140441258819,
      "grad_norm": 1.5343226194381714,
      "learning_rate": 6.313e-05,
      "loss": 0.5646,
      "step": 12633
    },
    {
      "epoch": 0.707470041438011,
      "grad_norm": 1.2741063833236694,
      "learning_rate": 6.3135e-05,
      "loss": 0.4095,
      "step": 12634
    },
    {
      "epoch": 0.70752603875014,
      "grad_norm": 1.3892006874084473,
      "learning_rate": 6.314e-05,
      "loss": 0.4059,
      "step": 12635
    },
    {
      "epoch": 0.707582036062269,
      "grad_norm": 1.618849754333496,
      "learning_rate": 6.314499999999999e-05,
      "loss": 0.4831,
      "step": 12636
    },
    {
      "epoch": 0.707638033374398,
      "grad_norm": 1.3106848001480103,
      "learning_rate": 6.315e-05,
      "loss": 0.4332,
      "step": 12637
    },
    {
      "epoch": 0.707694030686527,
      "grad_norm": 1.281575083732605,
      "learning_rate": 6.3155e-05,
      "loss": 0.4828,
      "step": 12638
    },
    {
      "epoch": 0.707750027998656,
      "grad_norm": 1.260123610496521,
      "learning_rate": 6.316000000000001e-05,
      "loss": 0.4523,
      "step": 12639
    },
    {
      "epoch": 0.7078060253107851,
      "grad_norm": 1.2649827003479004,
      "learning_rate": 6.316500000000001e-05,
      "loss": 0.5126,
      "step": 12640
    },
    {
      "epoch": 0.7078620226229141,
      "grad_norm": 1.6030091047286987,
      "learning_rate": 6.317e-05,
      "loss": 0.4465,
      "step": 12641
    },
    {
      "epoch": 0.7079180199350431,
      "grad_norm": 1.3819293975830078,
      "learning_rate": 6.3175e-05,
      "loss": 0.4599,
      "step": 12642
    },
    {
      "epoch": 0.7079740172471721,
      "grad_norm": 1.3676986694335938,
      "learning_rate": 6.318e-05,
      "loss": 0.4164,
      "step": 12643
    },
    {
      "epoch": 0.7080300145593011,
      "grad_norm": 1.3112050294876099,
      "learning_rate": 6.3185e-05,
      "loss": 0.4768,
      "step": 12644
    },
    {
      "epoch": 0.7080860118714302,
      "grad_norm": 1.3456107378005981,
      "learning_rate": 6.319000000000001e-05,
      "loss": 0.4973,
      "step": 12645
    },
    {
      "epoch": 0.7081420091835592,
      "grad_norm": 1.1713309288024902,
      "learning_rate": 6.319500000000001e-05,
      "loss": 0.3567,
      "step": 12646
    },
    {
      "epoch": 0.7081980064956882,
      "grad_norm": 1.4542640447616577,
      "learning_rate": 6.32e-05,
      "loss": 0.5801,
      "step": 12647
    },
    {
      "epoch": 0.7082540038078172,
      "grad_norm": 1.784436583518982,
      "learning_rate": 6.3205e-05,
      "loss": 0.5084,
      "step": 12648
    },
    {
      "epoch": 0.7083100011199462,
      "grad_norm": 1.2475916147232056,
      "learning_rate": 6.321e-05,
      "loss": 0.4811,
      "step": 12649
    },
    {
      "epoch": 0.7083659984320753,
      "grad_norm": 1.2328693866729736,
      "learning_rate": 6.3215e-05,
      "loss": 0.4592,
      "step": 12650
    },
    {
      "epoch": 0.7084219957442043,
      "grad_norm": 1.145531177520752,
      "learning_rate": 6.322000000000001e-05,
      "loss": 0.424,
      "step": 12651
    },
    {
      "epoch": 0.7084779930563333,
      "grad_norm": 1.4359947443008423,
      "learning_rate": 6.3225e-05,
      "loss": 0.4709,
      "step": 12652
    },
    {
      "epoch": 0.7085339903684623,
      "grad_norm": 1.411778211593628,
      "learning_rate": 6.323e-05,
      "loss": 0.4565,
      "step": 12653
    },
    {
      "epoch": 0.7085899876805913,
      "grad_norm": 1.3843156099319458,
      "learning_rate": 6.3235e-05,
      "loss": 0.485,
      "step": 12654
    },
    {
      "epoch": 0.7086459849927204,
      "grad_norm": 1.4193692207336426,
      "learning_rate": 6.324e-05,
      "loss": 0.3932,
      "step": 12655
    },
    {
      "epoch": 0.7087019823048494,
      "grad_norm": 1.2704010009765625,
      "learning_rate": 6.3245e-05,
      "loss": 0.4651,
      "step": 12656
    },
    {
      "epoch": 0.7087579796169784,
      "grad_norm": 1.1232733726501465,
      "learning_rate": 6.324999999999999e-05,
      "loss": 0.3719,
      "step": 12657
    },
    {
      "epoch": 0.7088139769291074,
      "grad_norm": 1.3058933019638062,
      "learning_rate": 6.3255e-05,
      "loss": 0.3575,
      "step": 12658
    },
    {
      "epoch": 0.7088699742412364,
      "grad_norm": 1.5541898012161255,
      "learning_rate": 6.326000000000001e-05,
      "loss": 0.6517,
      "step": 12659
    },
    {
      "epoch": 0.7089259715533655,
      "grad_norm": 1.325561761856079,
      "learning_rate": 6.326500000000001e-05,
      "loss": 0.3889,
      "step": 12660
    },
    {
      "epoch": 0.7089819688654945,
      "grad_norm": 1.5662405490875244,
      "learning_rate": 6.327000000000001e-05,
      "loss": 0.405,
      "step": 12661
    },
    {
      "epoch": 0.7090379661776235,
      "grad_norm": 1.0807271003723145,
      "learning_rate": 6.3275e-05,
      "loss": 0.3512,
      "step": 12662
    },
    {
      "epoch": 0.7090939634897525,
      "grad_norm": 1.5273305177688599,
      "learning_rate": 6.328e-05,
      "loss": 0.6008,
      "step": 12663
    },
    {
      "epoch": 0.7091499608018815,
      "grad_norm": 1.6548724174499512,
      "learning_rate": 6.3285e-05,
      "loss": 0.5224,
      "step": 12664
    },
    {
      "epoch": 0.7092059581140105,
      "grad_norm": 1.264316201210022,
      "learning_rate": 6.329e-05,
      "loss": 0.5297,
      "step": 12665
    },
    {
      "epoch": 0.7092619554261396,
      "grad_norm": 1.1656309366226196,
      "learning_rate": 6.329500000000001e-05,
      "loss": 0.3994,
      "step": 12666
    },
    {
      "epoch": 0.7093179527382686,
      "grad_norm": 2.9174137115478516,
      "learning_rate": 6.330000000000001e-05,
      "loss": 0.4231,
      "step": 12667
    },
    {
      "epoch": 0.7093739500503976,
      "grad_norm": 1.241214632987976,
      "learning_rate": 6.3305e-05,
      "loss": 0.4536,
      "step": 12668
    },
    {
      "epoch": 0.7094299473625266,
      "grad_norm": 1.5645548105239868,
      "learning_rate": 6.331e-05,
      "loss": 0.5548,
      "step": 12669
    },
    {
      "epoch": 0.7094859446746556,
      "grad_norm": 1.4205975532531738,
      "learning_rate": 6.3315e-05,
      "loss": 0.6061,
      "step": 12670
    },
    {
      "epoch": 0.7095419419867847,
      "grad_norm": 1.774970531463623,
      "learning_rate": 6.332e-05,
      "loss": 0.4486,
      "step": 12671
    },
    {
      "epoch": 0.7095979392989137,
      "grad_norm": 1.1116825342178345,
      "learning_rate": 6.3325e-05,
      "loss": 0.3824,
      "step": 12672
    },
    {
      "epoch": 0.7096539366110427,
      "grad_norm": 1.1717495918273926,
      "learning_rate": 6.333e-05,
      "loss": 0.5045,
      "step": 12673
    },
    {
      "epoch": 0.7097099339231717,
      "grad_norm": 1.1295884847640991,
      "learning_rate": 6.3335e-05,
      "loss": 0.3551,
      "step": 12674
    },
    {
      "epoch": 0.7097659312353007,
      "grad_norm": 1.3893578052520752,
      "learning_rate": 6.334e-05,
      "loss": 0.5575,
      "step": 12675
    },
    {
      "epoch": 0.7098219285474298,
      "grad_norm": 1.4093233346939087,
      "learning_rate": 6.3345e-05,
      "loss": 0.4796,
      "step": 12676
    },
    {
      "epoch": 0.7098779258595588,
      "grad_norm": 1.3685671091079712,
      "learning_rate": 6.335e-05,
      "loss": 0.4079,
      "step": 12677
    },
    {
      "epoch": 0.7099339231716878,
      "grad_norm": 1.190311074256897,
      "learning_rate": 6.335499999999999e-05,
      "loss": 0.4145,
      "step": 12678
    },
    {
      "epoch": 0.7099899204838168,
      "grad_norm": 1.2564928531646729,
      "learning_rate": 6.336e-05,
      "loss": 0.4877,
      "step": 12679
    },
    {
      "epoch": 0.7100459177959458,
      "grad_norm": 1.264681100845337,
      "learning_rate": 6.336500000000001e-05,
      "loss": 0.4781,
      "step": 12680
    },
    {
      "epoch": 0.7101019151080749,
      "grad_norm": 1.2790048122406006,
      "learning_rate": 6.337000000000001e-05,
      "loss": 0.4392,
      "step": 12681
    },
    {
      "epoch": 0.7101579124202039,
      "grad_norm": 1.1290431022644043,
      "learning_rate": 6.337500000000001e-05,
      "loss": 0.4491,
      "step": 12682
    },
    {
      "epoch": 0.7102139097323329,
      "grad_norm": 1.465829610824585,
      "learning_rate": 6.338e-05,
      "loss": 0.6806,
      "step": 12683
    },
    {
      "epoch": 0.7102699070444619,
      "grad_norm": 1.6494287252426147,
      "learning_rate": 6.3385e-05,
      "loss": 0.486,
      "step": 12684
    },
    {
      "epoch": 0.7103259043565909,
      "grad_norm": 1.357857584953308,
      "learning_rate": 6.339e-05,
      "loss": 0.4848,
      "step": 12685
    },
    {
      "epoch": 0.71038190166872,
      "grad_norm": 1.3813881874084473,
      "learning_rate": 6.3395e-05,
      "loss": 0.4979,
      "step": 12686
    },
    {
      "epoch": 0.710437898980849,
      "grad_norm": 1.551567792892456,
      "learning_rate": 6.340000000000001e-05,
      "loss": 0.3673,
      "step": 12687
    },
    {
      "epoch": 0.710493896292978,
      "grad_norm": 1.2936947345733643,
      "learning_rate": 6.340500000000001e-05,
      "loss": 0.4462,
      "step": 12688
    },
    {
      "epoch": 0.710549893605107,
      "grad_norm": 1.3695658445358276,
      "learning_rate": 6.341e-05,
      "loss": 0.4734,
      "step": 12689
    },
    {
      "epoch": 0.710605890917236,
      "grad_norm": 1.2387371063232422,
      "learning_rate": 6.3415e-05,
      "loss": 0.3892,
      "step": 12690
    },
    {
      "epoch": 0.7106618882293649,
      "grad_norm": 1.2142603397369385,
      "learning_rate": 6.342e-05,
      "loss": 0.4328,
      "step": 12691
    },
    {
      "epoch": 0.710717885541494,
      "grad_norm": 1.2077988386154175,
      "learning_rate": 6.3425e-05,
      "loss": 0.415,
      "step": 12692
    },
    {
      "epoch": 0.710773882853623,
      "grad_norm": 1.2667608261108398,
      "learning_rate": 6.343e-05,
      "loss": 0.4881,
      "step": 12693
    },
    {
      "epoch": 0.710829880165752,
      "grad_norm": 1.4840319156646729,
      "learning_rate": 6.3435e-05,
      "loss": 0.48,
      "step": 12694
    },
    {
      "epoch": 0.710885877477881,
      "grad_norm": 1.4201076030731201,
      "learning_rate": 6.344e-05,
      "loss": 0.5295,
      "step": 12695
    },
    {
      "epoch": 0.71094187479001,
      "grad_norm": 1.192383885383606,
      "learning_rate": 6.3445e-05,
      "loss": 0.3654,
      "step": 12696
    },
    {
      "epoch": 0.710997872102139,
      "grad_norm": 1.3051652908325195,
      "learning_rate": 6.345e-05,
      "loss": 0.4585,
      "step": 12697
    },
    {
      "epoch": 0.7110538694142681,
      "grad_norm": 1.7367881536483765,
      "learning_rate": 6.3455e-05,
      "loss": 0.4431,
      "step": 12698
    },
    {
      "epoch": 0.7111098667263971,
      "grad_norm": 1.2035497426986694,
      "learning_rate": 6.346e-05,
      "loss": 0.4178,
      "step": 12699
    },
    {
      "epoch": 0.7111658640385261,
      "grad_norm": 1.2499315738677979,
      "learning_rate": 6.3465e-05,
      "loss": 0.4182,
      "step": 12700
    },
    {
      "epoch": 0.7112218613506551,
      "grad_norm": 1.4131886959075928,
      "learning_rate": 6.347e-05,
      "loss": 0.4649,
      "step": 12701
    },
    {
      "epoch": 0.7112778586627841,
      "grad_norm": 1.190421223640442,
      "learning_rate": 6.347500000000001e-05,
      "loss": 0.4147,
      "step": 12702
    },
    {
      "epoch": 0.7113338559749132,
      "grad_norm": 1.193487286567688,
      "learning_rate": 6.348000000000001e-05,
      "loss": 0.4261,
      "step": 12703
    },
    {
      "epoch": 0.7113898532870422,
      "grad_norm": 1.1812646389007568,
      "learning_rate": 6.3485e-05,
      "loss": 0.4381,
      "step": 12704
    },
    {
      "epoch": 0.7114458505991712,
      "grad_norm": 1.2294243574142456,
      "learning_rate": 6.349e-05,
      "loss": 0.3583,
      "step": 12705
    },
    {
      "epoch": 0.7115018479113002,
      "grad_norm": 2.127265691757202,
      "learning_rate": 6.3495e-05,
      "loss": 0.5528,
      "step": 12706
    },
    {
      "epoch": 0.7115578452234292,
      "grad_norm": 1.3275443315505981,
      "learning_rate": 6.35e-05,
      "loss": 0.3748,
      "step": 12707
    },
    {
      "epoch": 0.7116138425355583,
      "grad_norm": 1.2263245582580566,
      "learning_rate": 6.350500000000001e-05,
      "loss": 0.4958,
      "step": 12708
    },
    {
      "epoch": 0.7116698398476873,
      "grad_norm": 1.5293418169021606,
      "learning_rate": 6.351000000000001e-05,
      "loss": 0.3799,
      "step": 12709
    },
    {
      "epoch": 0.7117258371598163,
      "grad_norm": 1.5596578121185303,
      "learning_rate": 6.3515e-05,
      "loss": 0.4192,
      "step": 12710
    },
    {
      "epoch": 0.7117818344719453,
      "grad_norm": 1.153295874595642,
      "learning_rate": 6.352e-05,
      "loss": 0.4549,
      "step": 12711
    },
    {
      "epoch": 0.7118378317840743,
      "grad_norm": 1.3420997858047485,
      "learning_rate": 6.3525e-05,
      "loss": 0.4939,
      "step": 12712
    },
    {
      "epoch": 0.7118938290962034,
      "grad_norm": 1.3926844596862793,
      "learning_rate": 6.353e-05,
      "loss": 0.3718,
      "step": 12713
    },
    {
      "epoch": 0.7119498264083324,
      "grad_norm": 1.3157209157943726,
      "learning_rate": 6.3535e-05,
      "loss": 0.5368,
      "step": 12714
    },
    {
      "epoch": 0.7120058237204614,
      "grad_norm": 1.580293893814087,
      "learning_rate": 6.354e-05,
      "loss": 0.5469,
      "step": 12715
    },
    {
      "epoch": 0.7120618210325904,
      "grad_norm": 1.5111911296844482,
      "learning_rate": 6.3545e-05,
      "loss": 0.3721,
      "step": 12716
    },
    {
      "epoch": 0.7121178183447194,
      "grad_norm": 1.2916908264160156,
      "learning_rate": 6.355e-05,
      "loss": 0.3981,
      "step": 12717
    },
    {
      "epoch": 0.7121738156568485,
      "grad_norm": 1.370994210243225,
      "learning_rate": 6.3555e-05,
      "loss": 0.3765,
      "step": 12718
    },
    {
      "epoch": 0.7122298129689775,
      "grad_norm": 1.1901644468307495,
      "learning_rate": 6.356000000000001e-05,
      "loss": 0.3658,
      "step": 12719
    },
    {
      "epoch": 0.7122858102811065,
      "grad_norm": 1.3271005153656006,
      "learning_rate": 6.3565e-05,
      "loss": 0.5493,
      "step": 12720
    },
    {
      "epoch": 0.7123418075932355,
      "grad_norm": 1.1498123407363892,
      "learning_rate": 6.357e-05,
      "loss": 0.4468,
      "step": 12721
    },
    {
      "epoch": 0.7123978049053645,
      "grad_norm": 1.1788270473480225,
      "learning_rate": 6.3575e-05,
      "loss": 0.3985,
      "step": 12722
    },
    {
      "epoch": 0.7124538022174935,
      "grad_norm": 1.8106350898742676,
      "learning_rate": 6.358000000000001e-05,
      "loss": 0.6286,
      "step": 12723
    },
    {
      "epoch": 0.7125097995296226,
      "grad_norm": 1.4803918600082397,
      "learning_rate": 6.358500000000001e-05,
      "loss": 0.4002,
      "step": 12724
    },
    {
      "epoch": 0.7125657968417516,
      "grad_norm": 1.188391923904419,
      "learning_rate": 6.359e-05,
      "loss": 0.3289,
      "step": 12725
    },
    {
      "epoch": 0.7126217941538806,
      "grad_norm": 1.2638044357299805,
      "learning_rate": 6.3595e-05,
      "loss": 0.3745,
      "step": 12726
    },
    {
      "epoch": 0.7126777914660096,
      "grad_norm": 1.017531156539917,
      "learning_rate": 6.36e-05,
      "loss": 0.337,
      "step": 12727
    },
    {
      "epoch": 0.7127337887781386,
      "grad_norm": 1.3558027744293213,
      "learning_rate": 6.3605e-05,
      "loss": 0.3277,
      "step": 12728
    },
    {
      "epoch": 0.7127897860902677,
      "grad_norm": 1.1347521543502808,
      "learning_rate": 6.361000000000001e-05,
      "loss": 0.4376,
      "step": 12729
    },
    {
      "epoch": 0.7128457834023967,
      "grad_norm": 1.4346020221710205,
      "learning_rate": 6.3615e-05,
      "loss": 0.5466,
      "step": 12730
    },
    {
      "epoch": 0.7129017807145257,
      "grad_norm": 1.54099702835083,
      "learning_rate": 6.362e-05,
      "loss": 0.517,
      "step": 12731
    },
    {
      "epoch": 0.7129577780266547,
      "grad_norm": 1.1725080013275146,
      "learning_rate": 6.3625e-05,
      "loss": 0.3622,
      "step": 12732
    },
    {
      "epoch": 0.7130137753387837,
      "grad_norm": 1.2479534149169922,
      "learning_rate": 6.363e-05,
      "loss": 0.3899,
      "step": 12733
    },
    {
      "epoch": 0.7130697726509128,
      "grad_norm": 1.2837066650390625,
      "learning_rate": 6.3635e-05,
      "loss": 0.3856,
      "step": 12734
    },
    {
      "epoch": 0.7131257699630418,
      "grad_norm": 2.2546517848968506,
      "learning_rate": 6.364e-05,
      "loss": 0.5325,
      "step": 12735
    },
    {
      "epoch": 0.7131817672751708,
      "grad_norm": 1.216312289237976,
      "learning_rate": 6.3645e-05,
      "loss": 0.4332,
      "step": 12736
    },
    {
      "epoch": 0.7132377645872998,
      "grad_norm": 1.3184088468551636,
      "learning_rate": 6.365e-05,
      "loss": 0.4355,
      "step": 12737
    },
    {
      "epoch": 0.7132937618994288,
      "grad_norm": 1.1263552904129028,
      "learning_rate": 6.3655e-05,
      "loss": 0.3072,
      "step": 12738
    },
    {
      "epoch": 0.7133497592115579,
      "grad_norm": 1.165421962738037,
      "learning_rate": 6.366000000000001e-05,
      "loss": 0.4261,
      "step": 12739
    },
    {
      "epoch": 0.7134057565236869,
      "grad_norm": 1.4257819652557373,
      "learning_rate": 6.366500000000001e-05,
      "loss": 0.5089,
      "step": 12740
    },
    {
      "epoch": 0.7134617538358159,
      "grad_norm": 1.3485167026519775,
      "learning_rate": 6.367e-05,
      "loss": 0.4479,
      "step": 12741
    },
    {
      "epoch": 0.7135177511479449,
      "grad_norm": 1.4713619947433472,
      "learning_rate": 6.3675e-05,
      "loss": 0.4504,
      "step": 12742
    },
    {
      "epoch": 0.7135737484600739,
      "grad_norm": 1.085688591003418,
      "learning_rate": 6.368e-05,
      "loss": 0.4326,
      "step": 12743
    },
    {
      "epoch": 0.713629745772203,
      "grad_norm": 1.2095509767532349,
      "learning_rate": 6.368500000000001e-05,
      "loss": 0.3957,
      "step": 12744
    },
    {
      "epoch": 0.713685743084332,
      "grad_norm": 1.5142602920532227,
      "learning_rate": 6.369000000000001e-05,
      "loss": 0.3968,
      "step": 12745
    },
    {
      "epoch": 0.713741740396461,
      "grad_norm": 1.1609373092651367,
      "learning_rate": 6.3695e-05,
      "loss": 0.5362,
      "step": 12746
    },
    {
      "epoch": 0.71379773770859,
      "grad_norm": 1.6774296760559082,
      "learning_rate": 6.37e-05,
      "loss": 0.5318,
      "step": 12747
    },
    {
      "epoch": 0.713853735020719,
      "grad_norm": 1.3412060737609863,
      "learning_rate": 6.3705e-05,
      "loss": 0.3999,
      "step": 12748
    },
    {
      "epoch": 0.713909732332848,
      "grad_norm": 1.1753593683242798,
      "learning_rate": 6.371e-05,
      "loss": 0.3468,
      "step": 12749
    },
    {
      "epoch": 0.7139657296449771,
      "grad_norm": 1.4292224645614624,
      "learning_rate": 6.371500000000001e-05,
      "loss": 0.5395,
      "step": 12750
    },
    {
      "epoch": 0.7140217269571061,
      "grad_norm": 1.2235126495361328,
      "learning_rate": 6.372e-05,
      "loss": 0.3857,
      "step": 12751
    },
    {
      "epoch": 0.7140777242692351,
      "grad_norm": 1.4244022369384766,
      "learning_rate": 6.3725e-05,
      "loss": 0.4213,
      "step": 12752
    },
    {
      "epoch": 0.7141337215813641,
      "grad_norm": 1.2730166912078857,
      "learning_rate": 6.373e-05,
      "loss": 0.4058,
      "step": 12753
    },
    {
      "epoch": 0.7141897188934931,
      "grad_norm": 1.4534602165222168,
      "learning_rate": 6.3735e-05,
      "loss": 0.4449,
      "step": 12754
    },
    {
      "epoch": 0.7142457162056222,
      "grad_norm": 1.2470502853393555,
      "learning_rate": 6.374e-05,
      "loss": 0.5016,
      "step": 12755
    },
    {
      "epoch": 0.7143017135177512,
      "grad_norm": 1.146397352218628,
      "learning_rate": 6.3745e-05,
      "loss": 0.3615,
      "step": 12756
    },
    {
      "epoch": 0.7143577108298802,
      "grad_norm": 1.190064549446106,
      "learning_rate": 6.375e-05,
      "loss": 0.4055,
      "step": 12757
    },
    {
      "epoch": 0.7144137081420092,
      "grad_norm": 1.2497453689575195,
      "learning_rate": 6.3755e-05,
      "loss": 0.4397,
      "step": 12758
    },
    {
      "epoch": 0.7144697054541382,
      "grad_norm": 2.066960096359253,
      "learning_rate": 6.376e-05,
      "loss": 0.357,
      "step": 12759
    },
    {
      "epoch": 0.7145257027662673,
      "grad_norm": 1.6281381845474243,
      "learning_rate": 6.376500000000001e-05,
      "loss": 0.5837,
      "step": 12760
    },
    {
      "epoch": 0.7145817000783963,
      "grad_norm": 1.5029911994934082,
      "learning_rate": 6.377000000000001e-05,
      "loss": 0.4694,
      "step": 12761
    },
    {
      "epoch": 0.7146376973905253,
      "grad_norm": 1.3037540912628174,
      "learning_rate": 6.3775e-05,
      "loss": 0.3326,
      "step": 12762
    },
    {
      "epoch": 0.7146936947026543,
      "grad_norm": 1.1242607831954956,
      "learning_rate": 6.378e-05,
      "loss": 0.4072,
      "step": 12763
    },
    {
      "epoch": 0.7147496920147833,
      "grad_norm": 1.3720523118972778,
      "learning_rate": 6.3785e-05,
      "loss": 0.4946,
      "step": 12764
    },
    {
      "epoch": 0.7148056893269124,
      "grad_norm": 1.3609472513198853,
      "learning_rate": 6.379000000000001e-05,
      "loss": 0.4443,
      "step": 12765
    },
    {
      "epoch": 0.7148616866390414,
      "grad_norm": 1.4723418951034546,
      "learning_rate": 6.379500000000001e-05,
      "loss": 0.4779,
      "step": 12766
    },
    {
      "epoch": 0.7149176839511704,
      "grad_norm": 1.2363005876541138,
      "learning_rate": 6.38e-05,
      "loss": 0.3756,
      "step": 12767
    },
    {
      "epoch": 0.7149736812632994,
      "grad_norm": 1.1192009449005127,
      "learning_rate": 6.3805e-05,
      "loss": 0.4275,
      "step": 12768
    },
    {
      "epoch": 0.7150296785754284,
      "grad_norm": 1.8555961847305298,
      "learning_rate": 6.381e-05,
      "loss": 0.5432,
      "step": 12769
    },
    {
      "epoch": 0.7150856758875574,
      "grad_norm": 1.238851547241211,
      "learning_rate": 6.3815e-05,
      "loss": 0.364,
      "step": 12770
    },
    {
      "epoch": 0.7151416731996865,
      "grad_norm": 1.2626633644104004,
      "learning_rate": 6.382e-05,
      "loss": 0.4722,
      "step": 12771
    },
    {
      "epoch": 0.7151976705118155,
      "grad_norm": 2.2175941467285156,
      "learning_rate": 6.3825e-05,
      "loss": 0.3666,
      "step": 12772
    },
    {
      "epoch": 0.7152536678239445,
      "grad_norm": 1.3716325759887695,
      "learning_rate": 6.383e-05,
      "loss": 0.383,
      "step": 12773
    },
    {
      "epoch": 0.7153096651360734,
      "grad_norm": 1.627503752708435,
      "learning_rate": 6.3835e-05,
      "loss": 0.4719,
      "step": 12774
    },
    {
      "epoch": 0.7153656624482024,
      "grad_norm": 1.6553013324737549,
      "learning_rate": 6.384e-05,
      "loss": 0.4163,
      "step": 12775
    },
    {
      "epoch": 0.7154216597603315,
      "grad_norm": 1.3761839866638184,
      "learning_rate": 6.3845e-05,
      "loss": 0.5103,
      "step": 12776
    },
    {
      "epoch": 0.7154776570724605,
      "grad_norm": 1.385082483291626,
      "learning_rate": 6.385e-05,
      "loss": 0.5036,
      "step": 12777
    },
    {
      "epoch": 0.7155336543845895,
      "grad_norm": 1.3864158391952515,
      "learning_rate": 6.3855e-05,
      "loss": 0.5604,
      "step": 12778
    },
    {
      "epoch": 0.7155896516967185,
      "grad_norm": 1.4270308017730713,
      "learning_rate": 6.386e-05,
      "loss": 0.5372,
      "step": 12779
    },
    {
      "epoch": 0.7156456490088475,
      "grad_norm": 1.5776902437210083,
      "learning_rate": 6.386500000000001e-05,
      "loss": 0.6883,
      "step": 12780
    },
    {
      "epoch": 0.7157016463209765,
      "grad_norm": 1.3400368690490723,
      "learning_rate": 6.387000000000001e-05,
      "loss": 0.4332,
      "step": 12781
    },
    {
      "epoch": 0.7157576436331056,
      "grad_norm": 1.2608451843261719,
      "learning_rate": 6.387500000000001e-05,
      "loss": 0.3786,
      "step": 12782
    },
    {
      "epoch": 0.7158136409452346,
      "grad_norm": 1.2663267850875854,
      "learning_rate": 6.388e-05,
      "loss": 0.4651,
      "step": 12783
    },
    {
      "epoch": 0.7158696382573636,
      "grad_norm": 1.5927826166152954,
      "learning_rate": 6.3885e-05,
      "loss": 0.4652,
      "step": 12784
    },
    {
      "epoch": 0.7159256355694926,
      "grad_norm": 1.1670840978622437,
      "learning_rate": 6.389e-05,
      "loss": 0.4082,
      "step": 12785
    },
    {
      "epoch": 0.7159816328816216,
      "grad_norm": 1.3903788328170776,
      "learning_rate": 6.389500000000001e-05,
      "loss": 0.5638,
      "step": 12786
    },
    {
      "epoch": 0.7160376301937507,
      "grad_norm": 1.4020166397094727,
      "learning_rate": 6.390000000000001e-05,
      "loss": 0.4935,
      "step": 12787
    },
    {
      "epoch": 0.7160936275058797,
      "grad_norm": 1.471501111984253,
      "learning_rate": 6.3905e-05,
      "loss": 0.4349,
      "step": 12788
    },
    {
      "epoch": 0.7161496248180087,
      "grad_norm": 1.266884207725525,
      "learning_rate": 6.391e-05,
      "loss": 0.3633,
      "step": 12789
    },
    {
      "epoch": 0.7162056221301377,
      "grad_norm": 1.3894003629684448,
      "learning_rate": 6.3915e-05,
      "loss": 0.4857,
      "step": 12790
    },
    {
      "epoch": 0.7162616194422667,
      "grad_norm": 1.3955460786819458,
      "learning_rate": 6.392e-05,
      "loss": 0.4931,
      "step": 12791
    },
    {
      "epoch": 0.7163176167543958,
      "grad_norm": 1.221116065979004,
      "learning_rate": 6.3925e-05,
      "loss": 0.3948,
      "step": 12792
    },
    {
      "epoch": 0.7163736140665248,
      "grad_norm": 3.43851900100708,
      "learning_rate": 6.393e-05,
      "loss": 0.4225,
      "step": 12793
    },
    {
      "epoch": 0.7164296113786538,
      "grad_norm": 1.3714956045150757,
      "learning_rate": 6.3935e-05,
      "loss": 0.4307,
      "step": 12794
    },
    {
      "epoch": 0.7164856086907828,
      "grad_norm": 1.4571897983551025,
      "learning_rate": 6.394e-05,
      "loss": 0.4694,
      "step": 12795
    },
    {
      "epoch": 0.7165416060029118,
      "grad_norm": 1.2747544050216675,
      "learning_rate": 6.3945e-05,
      "loss": 0.4313,
      "step": 12796
    },
    {
      "epoch": 0.7165976033150409,
      "grad_norm": 1.3819502592086792,
      "learning_rate": 6.395e-05,
      "loss": 0.4512,
      "step": 12797
    },
    {
      "epoch": 0.7166536006271699,
      "grad_norm": 1.1545897722244263,
      "learning_rate": 6.3955e-05,
      "loss": 0.3672,
      "step": 12798
    },
    {
      "epoch": 0.7167095979392989,
      "grad_norm": 1.3497068881988525,
      "learning_rate": 6.396e-05,
      "loss": 0.4123,
      "step": 12799
    },
    {
      "epoch": 0.7167655952514279,
      "grad_norm": 1.1514780521392822,
      "learning_rate": 6.3965e-05,
      "loss": 0.422,
      "step": 12800
    },
    {
      "epoch": 0.7168215925635569,
      "grad_norm": 1.310577154159546,
      "learning_rate": 6.397000000000001e-05,
      "loss": 0.3955,
      "step": 12801
    },
    {
      "epoch": 0.716877589875686,
      "grad_norm": 1.2854505777359009,
      "learning_rate": 6.397500000000001e-05,
      "loss": 0.521,
      "step": 12802
    },
    {
      "epoch": 0.716933587187815,
      "grad_norm": 1.0573272705078125,
      "learning_rate": 6.398000000000001e-05,
      "loss": 0.3111,
      "step": 12803
    },
    {
      "epoch": 0.716989584499944,
      "grad_norm": 12.162367820739746,
      "learning_rate": 6.3985e-05,
      "loss": 0.6242,
      "step": 12804
    },
    {
      "epoch": 0.717045581812073,
      "grad_norm": 1.1641361713409424,
      "learning_rate": 6.399e-05,
      "loss": 0.4572,
      "step": 12805
    },
    {
      "epoch": 0.717101579124202,
      "grad_norm": 1.853898525238037,
      "learning_rate": 6.3995e-05,
      "loss": 0.4984,
      "step": 12806
    },
    {
      "epoch": 0.717157576436331,
      "grad_norm": 1.2919056415557861,
      "learning_rate": 6.400000000000001e-05,
      "loss": 0.4071,
      "step": 12807
    },
    {
      "epoch": 0.7172135737484601,
      "grad_norm": 1.2785226106643677,
      "learning_rate": 6.400500000000001e-05,
      "loss": 0.35,
      "step": 12808
    },
    {
      "epoch": 0.7172695710605891,
      "grad_norm": 1.0631603002548218,
      "learning_rate": 6.401e-05,
      "loss": 0.389,
      "step": 12809
    },
    {
      "epoch": 0.7173255683727181,
      "grad_norm": 1.295305848121643,
      "learning_rate": 6.4015e-05,
      "loss": 0.4316,
      "step": 12810
    },
    {
      "epoch": 0.7173815656848471,
      "grad_norm": 1.3592380285263062,
      "learning_rate": 6.402e-05,
      "loss": 0.4869,
      "step": 12811
    },
    {
      "epoch": 0.7174375629969761,
      "grad_norm": 1.337204933166504,
      "learning_rate": 6.4025e-05,
      "loss": 0.3931,
      "step": 12812
    },
    {
      "epoch": 0.7174935603091052,
      "grad_norm": 1.4174797534942627,
      "learning_rate": 6.403e-05,
      "loss": 0.3998,
      "step": 12813
    },
    {
      "epoch": 0.7175495576212342,
      "grad_norm": 1.4384292364120483,
      "learning_rate": 6.4035e-05,
      "loss": 0.442,
      "step": 12814
    },
    {
      "epoch": 0.7176055549333632,
      "grad_norm": 1.1505624055862427,
      "learning_rate": 6.404e-05,
      "loss": 0.5024,
      "step": 12815
    },
    {
      "epoch": 0.7176615522454922,
      "grad_norm": 1.3162777423858643,
      "learning_rate": 6.4045e-05,
      "loss": 0.5222,
      "step": 12816
    },
    {
      "epoch": 0.7177175495576212,
      "grad_norm": 1.2605162858963013,
      "learning_rate": 6.405e-05,
      "loss": 0.2991,
      "step": 12817
    },
    {
      "epoch": 0.7177735468697503,
      "grad_norm": 1.4509625434875488,
      "learning_rate": 6.4055e-05,
      "loss": 0.4539,
      "step": 12818
    },
    {
      "epoch": 0.7178295441818793,
      "grad_norm": 1.4207115173339844,
      "learning_rate": 6.405999999999999e-05,
      "loss": 0.4775,
      "step": 12819
    },
    {
      "epoch": 0.7178855414940083,
      "grad_norm": 1.4416894912719727,
      "learning_rate": 6.4065e-05,
      "loss": 0.4061,
      "step": 12820
    },
    {
      "epoch": 0.7179415388061373,
      "grad_norm": 1.3681820631027222,
      "learning_rate": 6.407e-05,
      "loss": 0.5323,
      "step": 12821
    },
    {
      "epoch": 0.7179975361182663,
      "grad_norm": 1.2461649179458618,
      "learning_rate": 6.407500000000001e-05,
      "loss": 0.4439,
      "step": 12822
    },
    {
      "epoch": 0.7180535334303954,
      "grad_norm": 1.3961012363433838,
      "learning_rate": 6.408000000000001e-05,
      "loss": 0.5067,
      "step": 12823
    },
    {
      "epoch": 0.7181095307425244,
      "grad_norm": 1.3075144290924072,
      "learning_rate": 6.408500000000001e-05,
      "loss": 0.35,
      "step": 12824
    },
    {
      "epoch": 0.7181655280546534,
      "grad_norm": 1.199385166168213,
      "learning_rate": 6.409e-05,
      "loss": 0.5217,
      "step": 12825
    },
    {
      "epoch": 0.7182215253667824,
      "grad_norm": 1.7142282724380493,
      "learning_rate": 6.4095e-05,
      "loss": 0.5228,
      "step": 12826
    },
    {
      "epoch": 0.7182775226789114,
      "grad_norm": 1.2804813385009766,
      "learning_rate": 6.41e-05,
      "loss": 0.4752,
      "step": 12827
    },
    {
      "epoch": 0.7183335199910404,
      "grad_norm": 1.6355879306793213,
      "learning_rate": 6.410500000000001e-05,
      "loss": 0.3663,
      "step": 12828
    },
    {
      "epoch": 0.7183895173031695,
      "grad_norm": 2.1395740509033203,
      "learning_rate": 6.411000000000001e-05,
      "loss": 0.4131,
      "step": 12829
    },
    {
      "epoch": 0.7184455146152985,
      "grad_norm": 1.2016159296035767,
      "learning_rate": 6.4115e-05,
      "loss": 0.4172,
      "step": 12830
    },
    {
      "epoch": 0.7185015119274275,
      "grad_norm": 4.299350261688232,
      "learning_rate": 6.412e-05,
      "loss": 0.5124,
      "step": 12831
    },
    {
      "epoch": 0.7185575092395565,
      "grad_norm": 1.3119045495986938,
      "learning_rate": 6.4125e-05,
      "loss": 0.5367,
      "step": 12832
    },
    {
      "epoch": 0.7186135065516855,
      "grad_norm": 1.2507914304733276,
      "learning_rate": 6.413e-05,
      "loss": 0.4462,
      "step": 12833
    },
    {
      "epoch": 0.7186695038638146,
      "grad_norm": 1.6498754024505615,
      "learning_rate": 6.4135e-05,
      "loss": 0.4533,
      "step": 12834
    },
    {
      "epoch": 0.7187255011759436,
      "grad_norm": 1.2412898540496826,
      "learning_rate": 6.414e-05,
      "loss": 0.4036,
      "step": 12835
    },
    {
      "epoch": 0.7187814984880726,
      "grad_norm": 1.275437593460083,
      "learning_rate": 6.4145e-05,
      "loss": 0.5085,
      "step": 12836
    },
    {
      "epoch": 0.7188374958002016,
      "grad_norm": 1.2589842081069946,
      "learning_rate": 6.415e-05,
      "loss": 0.4849,
      "step": 12837
    },
    {
      "epoch": 0.7188934931123306,
      "grad_norm": 1.271705985069275,
      "learning_rate": 6.4155e-05,
      "loss": 0.5544,
      "step": 12838
    },
    {
      "epoch": 0.7189494904244597,
      "grad_norm": 1.4079164266586304,
      "learning_rate": 6.416e-05,
      "loss": 0.5601,
      "step": 12839
    },
    {
      "epoch": 0.7190054877365887,
      "grad_norm": 1.3210337162017822,
      "learning_rate": 6.4165e-05,
      "loss": 0.5112,
      "step": 12840
    },
    {
      "epoch": 0.7190614850487177,
      "grad_norm": 1.2597122192382812,
      "learning_rate": 6.417e-05,
      "loss": 0.4337,
      "step": 12841
    },
    {
      "epoch": 0.7191174823608467,
      "grad_norm": 1.3668289184570312,
      "learning_rate": 6.4175e-05,
      "loss": 0.443,
      "step": 12842
    },
    {
      "epoch": 0.7191734796729757,
      "grad_norm": 1.4604369401931763,
      "learning_rate": 6.418000000000001e-05,
      "loss": 0.4761,
      "step": 12843
    },
    {
      "epoch": 0.7192294769851048,
      "grad_norm": 1.1284401416778564,
      "learning_rate": 6.418500000000001e-05,
      "loss": 0.3613,
      "step": 12844
    },
    {
      "epoch": 0.7192854742972338,
      "grad_norm": 1.336661458015442,
      "learning_rate": 6.419000000000001e-05,
      "loss": 0.3995,
      "step": 12845
    },
    {
      "epoch": 0.7193414716093628,
      "grad_norm": 1.4011123180389404,
      "learning_rate": 6.4195e-05,
      "loss": 0.4359,
      "step": 12846
    },
    {
      "epoch": 0.7193974689214918,
      "grad_norm": 1.2431459426879883,
      "learning_rate": 6.42e-05,
      "loss": 0.4012,
      "step": 12847
    },
    {
      "epoch": 0.7194534662336208,
      "grad_norm": 1.1655406951904297,
      "learning_rate": 6.4205e-05,
      "loss": 0.4497,
      "step": 12848
    },
    {
      "epoch": 0.7195094635457498,
      "grad_norm": 1.292120099067688,
      "learning_rate": 6.421e-05,
      "loss": 0.5142,
      "step": 12849
    },
    {
      "epoch": 0.7195654608578789,
      "grad_norm": 1.2695152759552002,
      "learning_rate": 6.421500000000001e-05,
      "loss": 0.4447,
      "step": 12850
    },
    {
      "epoch": 0.7196214581700079,
      "grad_norm": 1.2664965391159058,
      "learning_rate": 6.422e-05,
      "loss": 0.4946,
      "step": 12851
    },
    {
      "epoch": 0.7196774554821369,
      "grad_norm": 1.2135577201843262,
      "learning_rate": 6.4225e-05,
      "loss": 0.3936,
      "step": 12852
    },
    {
      "epoch": 0.7197334527942659,
      "grad_norm": 1.182833194732666,
      "learning_rate": 6.423e-05,
      "loss": 0.3167,
      "step": 12853
    },
    {
      "epoch": 0.719789450106395,
      "grad_norm": 1.412680983543396,
      "learning_rate": 6.4235e-05,
      "loss": 0.4906,
      "step": 12854
    },
    {
      "epoch": 0.719845447418524,
      "grad_norm": 1.304890751838684,
      "learning_rate": 6.424e-05,
      "loss": 0.3311,
      "step": 12855
    },
    {
      "epoch": 0.719901444730653,
      "grad_norm": 1.1928796768188477,
      "learning_rate": 6.4245e-05,
      "loss": 0.4449,
      "step": 12856
    },
    {
      "epoch": 0.7199574420427819,
      "grad_norm": 2.066993236541748,
      "learning_rate": 6.425e-05,
      "loss": 0.3949,
      "step": 12857
    },
    {
      "epoch": 0.7200134393549109,
      "grad_norm": 1.359500527381897,
      "learning_rate": 6.4255e-05,
      "loss": 0.4686,
      "step": 12858
    },
    {
      "epoch": 0.7200694366670399,
      "grad_norm": 1.3475192785263062,
      "learning_rate": 6.426e-05,
      "loss": 0.482,
      "step": 12859
    },
    {
      "epoch": 0.720125433979169,
      "grad_norm": 1.0215314626693726,
      "learning_rate": 6.426500000000001e-05,
      "loss": 0.3383,
      "step": 12860
    },
    {
      "epoch": 0.720181431291298,
      "grad_norm": 1.4091228246688843,
      "learning_rate": 6.427e-05,
      "loss": 0.4618,
      "step": 12861
    },
    {
      "epoch": 0.720237428603427,
      "grad_norm": 1.5251009464263916,
      "learning_rate": 6.4275e-05,
      "loss": 0.4077,
      "step": 12862
    },
    {
      "epoch": 0.720293425915556,
      "grad_norm": 1.6396993398666382,
      "learning_rate": 6.428e-05,
      "loss": 0.5588,
      "step": 12863
    },
    {
      "epoch": 0.720349423227685,
      "grad_norm": 3.3280935287475586,
      "learning_rate": 6.428500000000001e-05,
      "loss": 0.5493,
      "step": 12864
    },
    {
      "epoch": 0.720405420539814,
      "grad_norm": 1.4691766500473022,
      "learning_rate": 6.429000000000001e-05,
      "loss": 0.5274,
      "step": 12865
    },
    {
      "epoch": 0.7204614178519431,
      "grad_norm": 1.2549264430999756,
      "learning_rate": 6.429500000000001e-05,
      "loss": 0.5649,
      "step": 12866
    },
    {
      "epoch": 0.7205174151640721,
      "grad_norm": 1.3790887594223022,
      "learning_rate": 6.43e-05,
      "loss": 0.4306,
      "step": 12867
    },
    {
      "epoch": 0.7205734124762011,
      "grad_norm": 1.192781925201416,
      "learning_rate": 6.4305e-05,
      "loss": 0.44,
      "step": 12868
    },
    {
      "epoch": 0.7206294097883301,
      "grad_norm": 1.3899825811386108,
      "learning_rate": 6.431e-05,
      "loss": 0.4697,
      "step": 12869
    },
    {
      "epoch": 0.7206854071004591,
      "grad_norm": 1.4914478063583374,
      "learning_rate": 6.4315e-05,
      "loss": 0.3433,
      "step": 12870
    },
    {
      "epoch": 0.7207414044125882,
      "grad_norm": 1.2485928535461426,
      "learning_rate": 6.432000000000001e-05,
      "loss": 0.3903,
      "step": 12871
    },
    {
      "epoch": 0.7207974017247172,
      "grad_norm": 1.469435691833496,
      "learning_rate": 6.4325e-05,
      "loss": 0.4088,
      "step": 12872
    },
    {
      "epoch": 0.7208533990368462,
      "grad_norm": 1.2373088598251343,
      "learning_rate": 6.433e-05,
      "loss": 0.3797,
      "step": 12873
    },
    {
      "epoch": 0.7209093963489752,
      "grad_norm": 1.1302622556686401,
      "learning_rate": 6.4335e-05,
      "loss": 0.4827,
      "step": 12874
    },
    {
      "epoch": 0.7209653936611042,
      "grad_norm": 1.3318990468978882,
      "learning_rate": 6.434e-05,
      "loss": 0.3497,
      "step": 12875
    },
    {
      "epoch": 0.7210213909732333,
      "grad_norm": 1.1196736097335815,
      "learning_rate": 6.4345e-05,
      "loss": 0.4114,
      "step": 12876
    },
    {
      "epoch": 0.7210773882853623,
      "grad_norm": 1.4482390880584717,
      "learning_rate": 6.435e-05,
      "loss": 0.4194,
      "step": 12877
    },
    {
      "epoch": 0.7211333855974913,
      "grad_norm": 1.2796387672424316,
      "learning_rate": 6.4355e-05,
      "loss": 0.4891,
      "step": 12878
    },
    {
      "epoch": 0.7211893829096203,
      "grad_norm": 1.1463838815689087,
      "learning_rate": 6.436e-05,
      "loss": 0.4692,
      "step": 12879
    },
    {
      "epoch": 0.7212453802217493,
      "grad_norm": 6.655930995941162,
      "learning_rate": 6.436500000000001e-05,
      "loss": 0.4129,
      "step": 12880
    },
    {
      "epoch": 0.7213013775338784,
      "grad_norm": 3.030787706375122,
      "learning_rate": 6.437000000000001e-05,
      "loss": 0.3954,
      "step": 12881
    },
    {
      "epoch": 0.7213573748460074,
      "grad_norm": 1.4530386924743652,
      "learning_rate": 6.4375e-05,
      "loss": 0.612,
      "step": 12882
    },
    {
      "epoch": 0.7214133721581364,
      "grad_norm": 1.3220127820968628,
      "learning_rate": 6.438e-05,
      "loss": 0.4964,
      "step": 12883
    },
    {
      "epoch": 0.7214693694702654,
      "grad_norm": 1.7192901372909546,
      "learning_rate": 6.4385e-05,
      "loss": 0.4847,
      "step": 12884
    },
    {
      "epoch": 0.7215253667823944,
      "grad_norm": 1.3402737379074097,
      "learning_rate": 6.439000000000001e-05,
      "loss": 0.4283,
      "step": 12885
    },
    {
      "epoch": 0.7215813640945234,
      "grad_norm": 1.1397950649261475,
      "learning_rate": 6.439500000000001e-05,
      "loss": 0.3931,
      "step": 12886
    },
    {
      "epoch": 0.7216373614066525,
      "grad_norm": 1.5136781930923462,
      "learning_rate": 6.440000000000001e-05,
      "loss": 0.581,
      "step": 12887
    },
    {
      "epoch": 0.7216933587187815,
      "grad_norm": 1.2671502828598022,
      "learning_rate": 6.4405e-05,
      "loss": 0.537,
      "step": 12888
    },
    {
      "epoch": 0.7217493560309105,
      "grad_norm": 1.1760938167572021,
      "learning_rate": 6.441e-05,
      "loss": 0.4935,
      "step": 12889
    },
    {
      "epoch": 0.7218053533430395,
      "grad_norm": 1.2217832803726196,
      "learning_rate": 6.4415e-05,
      "loss": 0.3571,
      "step": 12890
    },
    {
      "epoch": 0.7218613506551685,
      "grad_norm": 1.1983466148376465,
      "learning_rate": 6.442e-05,
      "loss": 0.3651,
      "step": 12891
    },
    {
      "epoch": 0.7219173479672976,
      "grad_norm": 1.4146442413330078,
      "learning_rate": 6.442500000000001e-05,
      "loss": 0.4483,
      "step": 12892
    },
    {
      "epoch": 0.7219733452794266,
      "grad_norm": 1.398146152496338,
      "learning_rate": 6.443e-05,
      "loss": 0.6141,
      "step": 12893
    },
    {
      "epoch": 0.7220293425915556,
      "grad_norm": 1.4189565181732178,
      "learning_rate": 6.4435e-05,
      "loss": 0.3798,
      "step": 12894
    },
    {
      "epoch": 0.7220853399036846,
      "grad_norm": 1.2826626300811768,
      "learning_rate": 6.444e-05,
      "loss": 0.5178,
      "step": 12895
    },
    {
      "epoch": 0.7221413372158136,
      "grad_norm": 1.3823177814483643,
      "learning_rate": 6.4445e-05,
      "loss": 0.4631,
      "step": 12896
    },
    {
      "epoch": 0.7221973345279427,
      "grad_norm": 1.345367670059204,
      "learning_rate": 6.445e-05,
      "loss": 0.437,
      "step": 12897
    },
    {
      "epoch": 0.7222533318400717,
      "grad_norm": 1.2119885683059692,
      "learning_rate": 6.4455e-05,
      "loss": 0.4164,
      "step": 12898
    },
    {
      "epoch": 0.7223093291522007,
      "grad_norm": 1.1886770725250244,
      "learning_rate": 6.446e-05,
      "loss": 0.355,
      "step": 12899
    },
    {
      "epoch": 0.7223653264643297,
      "grad_norm": 1.188043475151062,
      "learning_rate": 6.4465e-05,
      "loss": 0.4286,
      "step": 12900
    },
    {
      "epoch": 0.7224213237764587,
      "grad_norm": 1.4560166597366333,
      "learning_rate": 6.447000000000001e-05,
      "loss": 0.4311,
      "step": 12901
    },
    {
      "epoch": 0.7224773210885878,
      "grad_norm": 1.4028239250183105,
      "learning_rate": 6.447500000000001e-05,
      "loss": 0.4554,
      "step": 12902
    },
    {
      "epoch": 0.7225333184007168,
      "grad_norm": 1.1268011331558228,
      "learning_rate": 6.448e-05,
      "loss": 0.4178,
      "step": 12903
    },
    {
      "epoch": 0.7225893157128458,
      "grad_norm": 1.1890791654586792,
      "learning_rate": 6.4485e-05,
      "loss": 0.3923,
      "step": 12904
    },
    {
      "epoch": 0.7226453130249748,
      "grad_norm": 1.1738241910934448,
      "learning_rate": 6.449e-05,
      "loss": 0.3756,
      "step": 12905
    },
    {
      "epoch": 0.7227013103371038,
      "grad_norm": 1.6105811595916748,
      "learning_rate": 6.449500000000001e-05,
      "loss": 0.3918,
      "step": 12906
    },
    {
      "epoch": 0.7227573076492328,
      "grad_norm": 1.286133050918579,
      "learning_rate": 6.450000000000001e-05,
      "loss": 0.3927,
      "step": 12907
    },
    {
      "epoch": 0.7228133049613619,
      "grad_norm": 1.4088801145553589,
      "learning_rate": 6.4505e-05,
      "loss": 0.6302,
      "step": 12908
    },
    {
      "epoch": 0.7228693022734909,
      "grad_norm": 1.468022108078003,
      "learning_rate": 6.451e-05,
      "loss": 0.4253,
      "step": 12909
    },
    {
      "epoch": 0.7229252995856199,
      "grad_norm": 1.3817226886749268,
      "learning_rate": 6.4515e-05,
      "loss": 0.5141,
      "step": 12910
    },
    {
      "epoch": 0.7229812968977489,
      "grad_norm": 1.8205647468566895,
      "learning_rate": 6.452e-05,
      "loss": 0.4257,
      "step": 12911
    },
    {
      "epoch": 0.723037294209878,
      "grad_norm": 1.2383532524108887,
      "learning_rate": 6.4525e-05,
      "loss": 0.3925,
      "step": 12912
    },
    {
      "epoch": 0.723093291522007,
      "grad_norm": 1.3751835823059082,
      "learning_rate": 6.453000000000001e-05,
      "loss": 0.5789,
      "step": 12913
    },
    {
      "epoch": 0.723149288834136,
      "grad_norm": 1.4135364294052124,
      "learning_rate": 6.4535e-05,
      "loss": 0.402,
      "step": 12914
    },
    {
      "epoch": 0.723205286146265,
      "grad_norm": 1.4469807147979736,
      "learning_rate": 6.454e-05,
      "loss": 0.4442,
      "step": 12915
    },
    {
      "epoch": 0.723261283458394,
      "grad_norm": 1.4926486015319824,
      "learning_rate": 6.4545e-05,
      "loss": 0.5421,
      "step": 12916
    },
    {
      "epoch": 0.723317280770523,
      "grad_norm": 4.189614772796631,
      "learning_rate": 6.455e-05,
      "loss": 0.4641,
      "step": 12917
    },
    {
      "epoch": 0.7233732780826521,
      "grad_norm": 1.3739595413208008,
      "learning_rate": 6.4555e-05,
      "loss": 0.3941,
      "step": 12918
    },
    {
      "epoch": 0.7234292753947811,
      "grad_norm": 1.3635294437408447,
      "learning_rate": 6.455999999999999e-05,
      "loss": 0.5162,
      "step": 12919
    },
    {
      "epoch": 0.7234852727069101,
      "grad_norm": 1.2362215518951416,
      "learning_rate": 6.4565e-05,
      "loss": 0.3528,
      "step": 12920
    },
    {
      "epoch": 0.7235412700190391,
      "grad_norm": 1.1968994140625,
      "learning_rate": 6.457000000000001e-05,
      "loss": 0.4326,
      "step": 12921
    },
    {
      "epoch": 0.7235972673311681,
      "grad_norm": 1.122292160987854,
      "learning_rate": 6.457500000000001e-05,
      "loss": 0.401,
      "step": 12922
    },
    {
      "epoch": 0.7236532646432972,
      "grad_norm": 1.3467565774917603,
      "learning_rate": 6.458000000000001e-05,
      "loss": 0.4402,
      "step": 12923
    },
    {
      "epoch": 0.7237092619554262,
      "grad_norm": 1.0921175479888916,
      "learning_rate": 6.4585e-05,
      "loss": 0.392,
      "step": 12924
    },
    {
      "epoch": 0.7237652592675552,
      "grad_norm": 1.3415946960449219,
      "learning_rate": 6.459e-05,
      "loss": 0.3967,
      "step": 12925
    },
    {
      "epoch": 0.7238212565796842,
      "grad_norm": 2.375540018081665,
      "learning_rate": 6.4595e-05,
      "loss": 0.4013,
      "step": 12926
    },
    {
      "epoch": 0.7238772538918132,
      "grad_norm": 1.3974196910858154,
      "learning_rate": 6.460000000000001e-05,
      "loss": 0.5083,
      "step": 12927
    },
    {
      "epoch": 0.7239332512039423,
      "grad_norm": 1.1457161903381348,
      "learning_rate": 6.460500000000001e-05,
      "loss": 0.351,
      "step": 12928
    },
    {
      "epoch": 0.7239892485160713,
      "grad_norm": 1.063173532485962,
      "learning_rate": 6.461e-05,
      "loss": 0.3628,
      "step": 12929
    },
    {
      "epoch": 0.7240452458282003,
      "grad_norm": 1.3780083656311035,
      "learning_rate": 6.4615e-05,
      "loss": 0.4839,
      "step": 12930
    },
    {
      "epoch": 0.7241012431403293,
      "grad_norm": 1.3097578287124634,
      "learning_rate": 6.462e-05,
      "loss": 0.4928,
      "step": 12931
    },
    {
      "epoch": 0.7241572404524583,
      "grad_norm": 1.202901840209961,
      "learning_rate": 6.4625e-05,
      "loss": 0.4396,
      "step": 12932
    },
    {
      "epoch": 0.7242132377645873,
      "grad_norm": 1.4199342727661133,
      "learning_rate": 6.463e-05,
      "loss": 0.4244,
      "step": 12933
    },
    {
      "epoch": 0.7242692350767164,
      "grad_norm": 1.2170206308364868,
      "learning_rate": 6.463500000000001e-05,
      "loss": 0.421,
      "step": 12934
    },
    {
      "epoch": 0.7243252323888454,
      "grad_norm": 1.0460795164108276,
      "learning_rate": 6.464e-05,
      "loss": 0.3263,
      "step": 12935
    },
    {
      "epoch": 0.7243812297009744,
      "grad_norm": 1.2562450170516968,
      "learning_rate": 6.4645e-05,
      "loss": 0.3932,
      "step": 12936
    },
    {
      "epoch": 0.7244372270131034,
      "grad_norm": 1.3742763996124268,
      "learning_rate": 6.465e-05,
      "loss": 0.444,
      "step": 12937
    },
    {
      "epoch": 0.7244932243252324,
      "grad_norm": 1.135968804359436,
      "learning_rate": 6.4655e-05,
      "loss": 0.3864,
      "step": 12938
    },
    {
      "epoch": 0.7245492216373614,
      "grad_norm": 1.221862554550171,
      "learning_rate": 6.466e-05,
      "loss": 0.3532,
      "step": 12939
    },
    {
      "epoch": 0.7246052189494904,
      "grad_norm": 1.5764732360839844,
      "learning_rate": 6.466499999999999e-05,
      "loss": 0.4656,
      "step": 12940
    },
    {
      "epoch": 0.7246612162616194,
      "grad_norm": 1.252284288406372,
      "learning_rate": 6.467e-05,
      "loss": 0.4488,
      "step": 12941
    },
    {
      "epoch": 0.7247172135737484,
      "grad_norm": 1.1536979675292969,
      "learning_rate": 6.467500000000001e-05,
      "loss": 0.4659,
      "step": 12942
    },
    {
      "epoch": 0.7247732108858774,
      "grad_norm": 1.3067823648452759,
      "learning_rate": 6.468000000000001e-05,
      "loss": 0.514,
      "step": 12943
    },
    {
      "epoch": 0.7248292081980064,
      "grad_norm": 1.3132990598678589,
      "learning_rate": 6.468500000000001e-05,
      "loss": 0.4565,
      "step": 12944
    },
    {
      "epoch": 0.7248852055101355,
      "grad_norm": 1.3057440519332886,
      "learning_rate": 6.469e-05,
      "loss": 0.5572,
      "step": 12945
    },
    {
      "epoch": 0.7249412028222645,
      "grad_norm": 1.1999316215515137,
      "learning_rate": 6.4695e-05,
      "loss": 0.375,
      "step": 12946
    },
    {
      "epoch": 0.7249972001343935,
      "grad_norm": 1.336259365081787,
      "learning_rate": 6.47e-05,
      "loss": 0.5179,
      "step": 12947
    },
    {
      "epoch": 0.7250531974465225,
      "grad_norm": 1.6089376211166382,
      "learning_rate": 6.4705e-05,
      "loss": 0.4512,
      "step": 12948
    },
    {
      "epoch": 0.7251091947586515,
      "grad_norm": 1.5541924238204956,
      "learning_rate": 6.471000000000001e-05,
      "loss": 0.5098,
      "step": 12949
    },
    {
      "epoch": 0.7251651920707806,
      "grad_norm": 1.271735429763794,
      "learning_rate": 6.4715e-05,
      "loss": 0.5747,
      "step": 12950
    },
    {
      "epoch": 0.7252211893829096,
      "grad_norm": 1.406443476676941,
      "learning_rate": 6.472e-05,
      "loss": 0.3573,
      "step": 12951
    },
    {
      "epoch": 0.7252771866950386,
      "grad_norm": 1.480695366859436,
      "learning_rate": 6.4725e-05,
      "loss": 0.5508,
      "step": 12952
    },
    {
      "epoch": 0.7253331840071676,
      "grad_norm": 1.1610902547836304,
      "learning_rate": 6.473e-05,
      "loss": 0.4652,
      "step": 12953
    },
    {
      "epoch": 0.7253891813192966,
      "grad_norm": 1.2197450399398804,
      "learning_rate": 6.4735e-05,
      "loss": 0.3855,
      "step": 12954
    },
    {
      "epoch": 0.7254451786314257,
      "grad_norm": 1.2905586957931519,
      "learning_rate": 6.474000000000001e-05,
      "loss": 0.5056,
      "step": 12955
    },
    {
      "epoch": 0.7255011759435547,
      "grad_norm": 1.1793146133422852,
      "learning_rate": 6.4745e-05,
      "loss": 0.3365,
      "step": 12956
    },
    {
      "epoch": 0.7255571732556837,
      "grad_norm": 1.4936169385910034,
      "learning_rate": 6.475e-05,
      "loss": 0.4332,
      "step": 12957
    },
    {
      "epoch": 0.7256131705678127,
      "grad_norm": 1.0383176803588867,
      "learning_rate": 6.4755e-05,
      "loss": 0.2899,
      "step": 12958
    },
    {
      "epoch": 0.7256691678799417,
      "grad_norm": 1.3028063774108887,
      "learning_rate": 6.476e-05,
      "loss": 0.4425,
      "step": 12959
    },
    {
      "epoch": 0.7257251651920708,
      "grad_norm": 1.3707900047302246,
      "learning_rate": 6.4765e-05,
      "loss": 0.4871,
      "step": 12960
    },
    {
      "epoch": 0.7257811625041998,
      "grad_norm": 1.2507411241531372,
      "learning_rate": 6.477e-05,
      "loss": 0.4744,
      "step": 12961
    },
    {
      "epoch": 0.7258371598163288,
      "grad_norm": 1.4732862710952759,
      "learning_rate": 6.4775e-05,
      "loss": 0.5197,
      "step": 12962
    },
    {
      "epoch": 0.7258931571284578,
      "grad_norm": 1.2263737916946411,
      "learning_rate": 6.478000000000001e-05,
      "loss": 0.4142,
      "step": 12963
    },
    {
      "epoch": 0.7259491544405868,
      "grad_norm": 1.123755693435669,
      "learning_rate": 6.478500000000001e-05,
      "loss": 0.4705,
      "step": 12964
    },
    {
      "epoch": 0.7260051517527158,
      "grad_norm": 1.5642553567886353,
      "learning_rate": 6.479000000000001e-05,
      "loss": 0.3635,
      "step": 12965
    },
    {
      "epoch": 0.7260611490648449,
      "grad_norm": 3.230133295059204,
      "learning_rate": 6.4795e-05,
      "loss": 0.5285,
      "step": 12966
    },
    {
      "epoch": 0.7261171463769739,
      "grad_norm": 1.2619363069534302,
      "learning_rate": 6.48e-05,
      "loss": 0.4288,
      "step": 12967
    },
    {
      "epoch": 0.7261731436891029,
      "grad_norm": 1.3217151165008545,
      "learning_rate": 6.4805e-05,
      "loss": 0.3963,
      "step": 12968
    },
    {
      "epoch": 0.7262291410012319,
      "grad_norm": 1.244097113609314,
      "learning_rate": 6.481e-05,
      "loss": 0.4191,
      "step": 12969
    },
    {
      "epoch": 0.726285138313361,
      "grad_norm": 1.190354824066162,
      "learning_rate": 6.481500000000001e-05,
      "loss": 0.4705,
      "step": 12970
    },
    {
      "epoch": 0.72634113562549,
      "grad_norm": 1.2711552381515503,
      "learning_rate": 6.482e-05,
      "loss": 0.3871,
      "step": 12971
    },
    {
      "epoch": 0.726397132937619,
      "grad_norm": 1.2860972881317139,
      "learning_rate": 6.4825e-05,
      "loss": 0.5483,
      "step": 12972
    },
    {
      "epoch": 0.726453130249748,
      "grad_norm": 1.2053567171096802,
      "learning_rate": 6.483e-05,
      "loss": 0.3243,
      "step": 12973
    },
    {
      "epoch": 0.726509127561877,
      "grad_norm": 1.2540823221206665,
      "learning_rate": 6.4835e-05,
      "loss": 0.4828,
      "step": 12974
    },
    {
      "epoch": 0.726565124874006,
      "grad_norm": 1.240599274635315,
      "learning_rate": 6.484e-05,
      "loss": 0.466,
      "step": 12975
    },
    {
      "epoch": 0.7266211221861351,
      "grad_norm": 1.5479391813278198,
      "learning_rate": 6.484500000000001e-05,
      "loss": 0.5901,
      "step": 12976
    },
    {
      "epoch": 0.7266771194982641,
      "grad_norm": 1.4428088665008545,
      "learning_rate": 6.485e-05,
      "loss": 0.4771,
      "step": 12977
    },
    {
      "epoch": 0.7267331168103931,
      "grad_norm": 1.6157454252243042,
      "learning_rate": 6.4855e-05,
      "loss": 0.5049,
      "step": 12978
    },
    {
      "epoch": 0.7267891141225221,
      "grad_norm": 1.34217369556427,
      "learning_rate": 6.486e-05,
      "loss": 0.4706,
      "step": 12979
    },
    {
      "epoch": 0.7268451114346511,
      "grad_norm": 1.2550715208053589,
      "learning_rate": 6.4865e-05,
      "loss": 0.3991,
      "step": 12980
    },
    {
      "epoch": 0.7269011087467802,
      "grad_norm": 1.307175874710083,
      "learning_rate": 6.487000000000001e-05,
      "loss": 0.5009,
      "step": 12981
    },
    {
      "epoch": 0.7269571060589092,
      "grad_norm": 1.1628906726837158,
      "learning_rate": 6.4875e-05,
      "loss": 0.3632,
      "step": 12982
    },
    {
      "epoch": 0.7270131033710382,
      "grad_norm": 1.2792617082595825,
      "learning_rate": 6.488e-05,
      "loss": 0.4168,
      "step": 12983
    },
    {
      "epoch": 0.7270691006831672,
      "grad_norm": 1.4946825504302979,
      "learning_rate": 6.488500000000001e-05,
      "loss": 0.486,
      "step": 12984
    },
    {
      "epoch": 0.7271250979952962,
      "grad_norm": 1.4570118188858032,
      "learning_rate": 6.489000000000001e-05,
      "loss": 0.5857,
      "step": 12985
    },
    {
      "epoch": 0.7271810953074253,
      "grad_norm": 1.5580264329910278,
      "learning_rate": 6.489500000000001e-05,
      "loss": 0.353,
      "step": 12986
    },
    {
      "epoch": 0.7272370926195543,
      "grad_norm": 1.3818445205688477,
      "learning_rate": 6.49e-05,
      "loss": 0.3925,
      "step": 12987
    },
    {
      "epoch": 0.7272930899316833,
      "grad_norm": 1.2081820964813232,
      "learning_rate": 6.4905e-05,
      "loss": 0.4304,
      "step": 12988
    },
    {
      "epoch": 0.7273490872438123,
      "grad_norm": 1.2454428672790527,
      "learning_rate": 6.491e-05,
      "loss": 0.477,
      "step": 12989
    },
    {
      "epoch": 0.7274050845559413,
      "grad_norm": 1.2278811931610107,
      "learning_rate": 6.4915e-05,
      "loss": 0.3489,
      "step": 12990
    },
    {
      "epoch": 0.7274610818680703,
      "grad_norm": 1.3734474182128906,
      "learning_rate": 6.492000000000001e-05,
      "loss": 0.448,
      "step": 12991
    },
    {
      "epoch": 0.7275170791801994,
      "grad_norm": 1.1877390146255493,
      "learning_rate": 6.4925e-05,
      "loss": 0.3798,
      "step": 12992
    },
    {
      "epoch": 0.7275730764923284,
      "grad_norm": 1.2094286680221558,
      "learning_rate": 6.493e-05,
      "loss": 0.4001,
      "step": 12993
    },
    {
      "epoch": 0.7276290738044574,
      "grad_norm": 1.2303513288497925,
      "learning_rate": 6.4935e-05,
      "loss": 0.3729,
      "step": 12994
    },
    {
      "epoch": 0.7276850711165864,
      "grad_norm": 1.6889305114746094,
      "learning_rate": 6.494e-05,
      "loss": 0.4747,
      "step": 12995
    },
    {
      "epoch": 0.7277410684287154,
      "grad_norm": 1.3299304246902466,
      "learning_rate": 6.4945e-05,
      "loss": 0.5057,
      "step": 12996
    },
    {
      "epoch": 0.7277970657408445,
      "grad_norm": 1.2950594425201416,
      "learning_rate": 6.494999999999999e-05,
      "loss": 0.4396,
      "step": 12997
    },
    {
      "epoch": 0.7278530630529735,
      "grad_norm": 1.956496238708496,
      "learning_rate": 6.4955e-05,
      "loss": 0.5345,
      "step": 12998
    },
    {
      "epoch": 0.7279090603651025,
      "grad_norm": 1.3873339891433716,
      "learning_rate": 6.496e-05,
      "loss": 0.405,
      "step": 12999
    },
    {
      "epoch": 0.7279650576772315,
      "grad_norm": 1.2748873233795166,
      "learning_rate": 6.4965e-05,
      "loss": 0.5211,
      "step": 13000
    },
    {
      "epoch": 0.7280210549893605,
      "grad_norm": 1.6707043647766113,
      "learning_rate": 6.497000000000001e-05,
      "loss": 0.5523,
      "step": 13001
    },
    {
      "epoch": 0.7280770523014896,
      "grad_norm": 1.3483299016952515,
      "learning_rate": 6.497500000000001e-05,
      "loss": 0.4982,
      "step": 13002
    },
    {
      "epoch": 0.7281330496136186,
      "grad_norm": 1.3101677894592285,
      "learning_rate": 6.498e-05,
      "loss": 0.4722,
      "step": 13003
    },
    {
      "epoch": 0.7281890469257476,
      "grad_norm": 1.118424892425537,
      "learning_rate": 6.4985e-05,
      "loss": 0.3566,
      "step": 13004
    },
    {
      "epoch": 0.7282450442378766,
      "grad_norm": 1.1747689247131348,
      "learning_rate": 6.499000000000001e-05,
      "loss": 0.4689,
      "step": 13005
    },
    {
      "epoch": 0.7283010415500056,
      "grad_norm": 1.495073676109314,
      "learning_rate": 6.499500000000001e-05,
      "loss": 0.45,
      "step": 13006
    },
    {
      "epoch": 0.7283570388621347,
      "grad_norm": 1.3751918077468872,
      "learning_rate": 6.500000000000001e-05,
      "loss": 0.4043,
      "step": 13007
    },
    {
      "epoch": 0.7284130361742637,
      "grad_norm": 1.296090841293335,
      "learning_rate": 6.5005e-05,
      "loss": 0.4572,
      "step": 13008
    },
    {
      "epoch": 0.7284690334863927,
      "grad_norm": 1.2352558374404907,
      "learning_rate": 6.501e-05,
      "loss": 0.3579,
      "step": 13009
    },
    {
      "epoch": 0.7285250307985217,
      "grad_norm": 1.1969058513641357,
      "learning_rate": 6.5015e-05,
      "loss": 0.4123,
      "step": 13010
    },
    {
      "epoch": 0.7285810281106507,
      "grad_norm": 1.451724648475647,
      "learning_rate": 6.502e-05,
      "loss": 0.5631,
      "step": 13011
    },
    {
      "epoch": 0.7286370254227797,
      "grad_norm": 1.25950288772583,
      "learning_rate": 6.502500000000001e-05,
      "loss": 0.4782,
      "step": 13012
    },
    {
      "epoch": 0.7286930227349088,
      "grad_norm": 1.3958215713500977,
      "learning_rate": 6.503e-05,
      "loss": 0.5441,
      "step": 13013
    },
    {
      "epoch": 0.7287490200470378,
      "grad_norm": 1.6094831228256226,
      "learning_rate": 6.5035e-05,
      "loss": 0.5775,
      "step": 13014
    },
    {
      "epoch": 0.7288050173591668,
      "grad_norm": 1.5766825675964355,
      "learning_rate": 6.504e-05,
      "loss": 0.5992,
      "step": 13015
    },
    {
      "epoch": 0.7288610146712958,
      "grad_norm": 1.4682761430740356,
      "learning_rate": 6.5045e-05,
      "loss": 0.5773,
      "step": 13016
    },
    {
      "epoch": 0.7289170119834248,
      "grad_norm": 1.3241996765136719,
      "learning_rate": 6.505e-05,
      "loss": 0.4616,
      "step": 13017
    },
    {
      "epoch": 0.7289730092955539,
      "grad_norm": 1.1775118112564087,
      "learning_rate": 6.505499999999999e-05,
      "loss": 0.3069,
      "step": 13018
    },
    {
      "epoch": 0.7290290066076829,
      "grad_norm": 1.3832868337631226,
      "learning_rate": 6.506e-05,
      "loss": 0.5166,
      "step": 13019
    },
    {
      "epoch": 0.7290850039198119,
      "grad_norm": 1.269210934638977,
      "learning_rate": 6.5065e-05,
      "loss": 0.426,
      "step": 13020
    },
    {
      "epoch": 0.7291410012319409,
      "grad_norm": 1.6523929834365845,
      "learning_rate": 6.507e-05,
      "loss": 0.5015,
      "step": 13021
    },
    {
      "epoch": 0.7291969985440698,
      "grad_norm": 1.3177670240402222,
      "learning_rate": 6.507500000000001e-05,
      "loss": 0.4026,
      "step": 13022
    },
    {
      "epoch": 0.7292529958561988,
      "grad_norm": 1.1591671705245972,
      "learning_rate": 6.508000000000001e-05,
      "loss": 0.3467,
      "step": 13023
    },
    {
      "epoch": 0.7293089931683279,
      "grad_norm": 1.3172328472137451,
      "learning_rate": 6.5085e-05,
      "loss": 0.4742,
      "step": 13024
    },
    {
      "epoch": 0.7293649904804569,
      "grad_norm": 1.0297353267669678,
      "learning_rate": 6.509e-05,
      "loss": 0.3457,
      "step": 13025
    },
    {
      "epoch": 0.7294209877925859,
      "grad_norm": 1.4599758386611938,
      "learning_rate": 6.5095e-05,
      "loss": 0.5066,
      "step": 13026
    },
    {
      "epoch": 0.7294769851047149,
      "grad_norm": 1.2901194095611572,
      "learning_rate": 6.510000000000001e-05,
      "loss": 0.4009,
      "step": 13027
    },
    {
      "epoch": 0.729532982416844,
      "grad_norm": 1.7665644884109497,
      "learning_rate": 6.510500000000001e-05,
      "loss": 0.3786,
      "step": 13028
    },
    {
      "epoch": 0.729588979728973,
      "grad_norm": 1.3884735107421875,
      "learning_rate": 6.511e-05,
      "loss": 0.4049,
      "step": 13029
    },
    {
      "epoch": 0.729644977041102,
      "grad_norm": 1.4180721044540405,
      "learning_rate": 6.5115e-05,
      "loss": 0.4338,
      "step": 13030
    },
    {
      "epoch": 0.729700974353231,
      "grad_norm": 1.2886587381362915,
      "learning_rate": 6.512e-05,
      "loss": 0.4717,
      "step": 13031
    },
    {
      "epoch": 0.72975697166536,
      "grad_norm": 1.4126027822494507,
      "learning_rate": 6.5125e-05,
      "loss": 0.5387,
      "step": 13032
    },
    {
      "epoch": 0.729812968977489,
      "grad_norm": 1.5100982189178467,
      "learning_rate": 6.513000000000001e-05,
      "loss": 0.6658,
      "step": 13033
    },
    {
      "epoch": 0.7298689662896181,
      "grad_norm": 5.311023235321045,
      "learning_rate": 6.5135e-05,
      "loss": 0.4693,
      "step": 13034
    },
    {
      "epoch": 0.7299249636017471,
      "grad_norm": 1.2274329662322998,
      "learning_rate": 6.514e-05,
      "loss": 0.4763,
      "step": 13035
    },
    {
      "epoch": 0.7299809609138761,
      "grad_norm": 1.3793895244598389,
      "learning_rate": 6.5145e-05,
      "loss": 0.3569,
      "step": 13036
    },
    {
      "epoch": 0.7300369582260051,
      "grad_norm": 1.3775551319122314,
      "learning_rate": 6.515e-05,
      "loss": 0.6268,
      "step": 13037
    },
    {
      "epoch": 0.7300929555381341,
      "grad_norm": 1.4914031028747559,
      "learning_rate": 6.5155e-05,
      "loss": 0.6025,
      "step": 13038
    },
    {
      "epoch": 0.7301489528502632,
      "grad_norm": 1.4676573276519775,
      "learning_rate": 6.515999999999999e-05,
      "loss": 0.4987,
      "step": 13039
    },
    {
      "epoch": 0.7302049501623922,
      "grad_norm": 1.2466171979904175,
      "learning_rate": 6.5165e-05,
      "loss": 0.3876,
      "step": 13040
    },
    {
      "epoch": 0.7302609474745212,
      "grad_norm": 1.5200568437576294,
      "learning_rate": 6.517e-05,
      "loss": 0.4826,
      "step": 13041
    },
    {
      "epoch": 0.7303169447866502,
      "grad_norm": 1.2348226308822632,
      "learning_rate": 6.517500000000001e-05,
      "loss": 0.3823,
      "step": 13042
    },
    {
      "epoch": 0.7303729420987792,
      "grad_norm": 1.1773552894592285,
      "learning_rate": 6.518000000000001e-05,
      "loss": 0.4015,
      "step": 13043
    },
    {
      "epoch": 0.7304289394109083,
      "grad_norm": 1.558884620666504,
      "learning_rate": 6.518500000000001e-05,
      "loss": 0.4081,
      "step": 13044
    },
    {
      "epoch": 0.7304849367230373,
      "grad_norm": 1.3640022277832031,
      "learning_rate": 6.519e-05,
      "loss": 0.4403,
      "step": 13045
    },
    {
      "epoch": 0.7305409340351663,
      "grad_norm": 1.130803108215332,
      "learning_rate": 6.5195e-05,
      "loss": 0.3317,
      "step": 13046
    },
    {
      "epoch": 0.7305969313472953,
      "grad_norm": 1.2308452129364014,
      "learning_rate": 6.52e-05,
      "loss": 0.4061,
      "step": 13047
    },
    {
      "epoch": 0.7306529286594243,
      "grad_norm": 1.3676975965499878,
      "learning_rate": 6.520500000000001e-05,
      "loss": 0.4609,
      "step": 13048
    },
    {
      "epoch": 0.7307089259715533,
      "grad_norm": 1.168544054031372,
      "learning_rate": 6.521000000000001e-05,
      "loss": 0.374,
      "step": 13049
    },
    {
      "epoch": 0.7307649232836824,
      "grad_norm": 1.2402279376983643,
      "learning_rate": 6.5215e-05,
      "loss": 0.4881,
      "step": 13050
    },
    {
      "epoch": 0.7308209205958114,
      "grad_norm": 1.1424239873886108,
      "learning_rate": 6.522e-05,
      "loss": 0.3579,
      "step": 13051
    },
    {
      "epoch": 0.7308769179079404,
      "grad_norm": 1.6257461309432983,
      "learning_rate": 6.5225e-05,
      "loss": 0.5411,
      "step": 13052
    },
    {
      "epoch": 0.7309329152200694,
      "grad_norm": 1.138053059577942,
      "learning_rate": 6.523e-05,
      "loss": 0.3891,
      "step": 13053
    },
    {
      "epoch": 0.7309889125321984,
      "grad_norm": 1.118177056312561,
      "learning_rate": 6.523500000000001e-05,
      "loss": 0.3239,
      "step": 13054
    },
    {
      "epoch": 0.7310449098443275,
      "grad_norm": 1.5924105644226074,
      "learning_rate": 6.524e-05,
      "loss": 0.5878,
      "step": 13055
    },
    {
      "epoch": 0.7311009071564565,
      "grad_norm": 1.3991609811782837,
      "learning_rate": 6.5245e-05,
      "loss": 0.3695,
      "step": 13056
    },
    {
      "epoch": 0.7311569044685855,
      "grad_norm": 1.188379168510437,
      "learning_rate": 6.525e-05,
      "loss": 0.3401,
      "step": 13057
    },
    {
      "epoch": 0.7312129017807145,
      "grad_norm": 1.364189863204956,
      "learning_rate": 6.5255e-05,
      "loss": 0.351,
      "step": 13058
    },
    {
      "epoch": 0.7312688990928435,
      "grad_norm": 1.245988368988037,
      "learning_rate": 6.526e-05,
      "loss": 0.4417,
      "step": 13059
    },
    {
      "epoch": 0.7313248964049726,
      "grad_norm": 1.2236649990081787,
      "learning_rate": 6.526499999999999e-05,
      "loss": 0.3758,
      "step": 13060
    },
    {
      "epoch": 0.7313808937171016,
      "grad_norm": 1.0870695114135742,
      "learning_rate": 6.527e-05,
      "loss": 0.2935,
      "step": 13061
    },
    {
      "epoch": 0.7314368910292306,
      "grad_norm": 1.4593501091003418,
      "learning_rate": 6.527500000000001e-05,
      "loss": 0.4677,
      "step": 13062
    },
    {
      "epoch": 0.7314928883413596,
      "grad_norm": 1.3815888166427612,
      "learning_rate": 6.528000000000001e-05,
      "loss": 0.3849,
      "step": 13063
    },
    {
      "epoch": 0.7315488856534886,
      "grad_norm": 1.5010366439819336,
      "learning_rate": 6.528500000000001e-05,
      "loss": 0.416,
      "step": 13064
    },
    {
      "epoch": 0.7316048829656177,
      "grad_norm": 1.7675964832305908,
      "learning_rate": 6.529e-05,
      "loss": 0.3649,
      "step": 13065
    },
    {
      "epoch": 0.7316608802777467,
      "grad_norm": 1.5239472389221191,
      "learning_rate": 6.5295e-05,
      "loss": 0.4546,
      "step": 13066
    },
    {
      "epoch": 0.7317168775898757,
      "grad_norm": 1.4678200483322144,
      "learning_rate": 6.53e-05,
      "loss": 0.4545,
      "step": 13067
    },
    {
      "epoch": 0.7317728749020047,
      "grad_norm": 1.3831298351287842,
      "learning_rate": 6.5305e-05,
      "loss": 0.4052,
      "step": 13068
    },
    {
      "epoch": 0.7318288722141337,
      "grad_norm": 1.2077752351760864,
      "learning_rate": 6.531000000000001e-05,
      "loss": 0.4256,
      "step": 13069
    },
    {
      "epoch": 0.7318848695262627,
      "grad_norm": 1.3360260725021362,
      "learning_rate": 6.531500000000001e-05,
      "loss": 0.4347,
      "step": 13070
    },
    {
      "epoch": 0.7319408668383918,
      "grad_norm": 1.3099371194839478,
      "learning_rate": 6.532e-05,
      "loss": 0.4831,
      "step": 13071
    },
    {
      "epoch": 0.7319968641505208,
      "grad_norm": 1.2227438688278198,
      "learning_rate": 6.5325e-05,
      "loss": 0.4765,
      "step": 13072
    },
    {
      "epoch": 0.7320528614626498,
      "grad_norm": 1.2485935688018799,
      "learning_rate": 6.533e-05,
      "loss": 0.3406,
      "step": 13073
    },
    {
      "epoch": 0.7321088587747788,
      "grad_norm": 1.3343572616577148,
      "learning_rate": 6.5335e-05,
      "loss": 0.4625,
      "step": 13074
    },
    {
      "epoch": 0.7321648560869078,
      "grad_norm": 1.1251615285873413,
      "learning_rate": 6.534e-05,
      "loss": 0.3521,
      "step": 13075
    },
    {
      "epoch": 0.7322208533990369,
      "grad_norm": 1.5101428031921387,
      "learning_rate": 6.5345e-05,
      "loss": 0.5716,
      "step": 13076
    },
    {
      "epoch": 0.7322768507111659,
      "grad_norm": 1.3580188751220703,
      "learning_rate": 6.535e-05,
      "loss": 0.4463,
      "step": 13077
    },
    {
      "epoch": 0.7323328480232949,
      "grad_norm": 1.3270171880722046,
      "learning_rate": 6.5355e-05,
      "loss": 0.5816,
      "step": 13078
    },
    {
      "epoch": 0.7323888453354239,
      "grad_norm": 1.1969938278198242,
      "learning_rate": 6.536e-05,
      "loss": 0.354,
      "step": 13079
    },
    {
      "epoch": 0.7324448426475529,
      "grad_norm": 1.433491587638855,
      "learning_rate": 6.5365e-05,
      "loss": 0.3844,
      "step": 13080
    },
    {
      "epoch": 0.732500839959682,
      "grad_norm": 1.302200198173523,
      "learning_rate": 6.536999999999999e-05,
      "loss": 0.4395,
      "step": 13081
    },
    {
      "epoch": 0.732556837271811,
      "grad_norm": 1.4608463048934937,
      "learning_rate": 6.5375e-05,
      "loss": 0.4483,
      "step": 13082
    },
    {
      "epoch": 0.73261283458394,
      "grad_norm": 1.3989903926849365,
      "learning_rate": 6.538000000000001e-05,
      "loss": 0.4813,
      "step": 13083
    },
    {
      "epoch": 0.732668831896069,
      "grad_norm": 1.2609761953353882,
      "learning_rate": 6.538500000000001e-05,
      "loss": 0.39,
      "step": 13084
    },
    {
      "epoch": 0.732724829208198,
      "grad_norm": 1.1653172969818115,
      "learning_rate": 6.539000000000001e-05,
      "loss": 0.4204,
      "step": 13085
    },
    {
      "epoch": 0.732780826520327,
      "grad_norm": 1.3216698169708252,
      "learning_rate": 6.5395e-05,
      "loss": 0.4472,
      "step": 13086
    },
    {
      "epoch": 0.7328368238324561,
      "grad_norm": 1.3972324132919312,
      "learning_rate": 6.54e-05,
      "loss": 0.5031,
      "step": 13087
    },
    {
      "epoch": 0.7328928211445851,
      "grad_norm": 1.205294132232666,
      "learning_rate": 6.5405e-05,
      "loss": 0.4253,
      "step": 13088
    },
    {
      "epoch": 0.7329488184567141,
      "grad_norm": 1.3671629428863525,
      "learning_rate": 6.541e-05,
      "loss": 0.4748,
      "step": 13089
    },
    {
      "epoch": 0.7330048157688431,
      "grad_norm": 1.2196563482284546,
      "learning_rate": 6.541500000000001e-05,
      "loss": 0.4939,
      "step": 13090
    },
    {
      "epoch": 0.7330608130809722,
      "grad_norm": 1.2559672594070435,
      "learning_rate": 6.542000000000001e-05,
      "loss": 0.3385,
      "step": 13091
    },
    {
      "epoch": 0.7331168103931012,
      "grad_norm": 1.2398109436035156,
      "learning_rate": 6.5425e-05,
      "loss": 0.4241,
      "step": 13092
    },
    {
      "epoch": 0.7331728077052302,
      "grad_norm": 1.8966413736343384,
      "learning_rate": 6.543e-05,
      "loss": 0.5796,
      "step": 13093
    },
    {
      "epoch": 0.7332288050173592,
      "grad_norm": 1.4565420150756836,
      "learning_rate": 6.5435e-05,
      "loss": 0.4427,
      "step": 13094
    },
    {
      "epoch": 0.7332848023294882,
      "grad_norm": 1.4604127407073975,
      "learning_rate": 6.544e-05,
      "loss": 0.5573,
      "step": 13095
    },
    {
      "epoch": 0.7333407996416172,
      "grad_norm": 1.456725001335144,
      "learning_rate": 6.5445e-05,
      "loss": 0.459,
      "step": 13096
    },
    {
      "epoch": 0.7333967969537463,
      "grad_norm": 1.8771783113479614,
      "learning_rate": 6.545e-05,
      "loss": 0.4563,
      "step": 13097
    },
    {
      "epoch": 0.7334527942658753,
      "grad_norm": 1.2887972593307495,
      "learning_rate": 6.5455e-05,
      "loss": 0.5581,
      "step": 13098
    },
    {
      "epoch": 0.7335087915780043,
      "grad_norm": 1.593860387802124,
      "learning_rate": 6.546e-05,
      "loss": 0.4394,
      "step": 13099
    },
    {
      "epoch": 0.7335647888901333,
      "grad_norm": 1.411238431930542,
      "learning_rate": 6.5465e-05,
      "loss": 0.4793,
      "step": 13100
    },
    {
      "epoch": 0.7336207862022623,
      "grad_norm": 1.213128924369812,
      "learning_rate": 6.547e-05,
      "loss": 0.4316,
      "step": 13101
    },
    {
      "epoch": 0.7336767835143914,
      "grad_norm": 1.4428856372833252,
      "learning_rate": 6.5475e-05,
      "loss": 0.4737,
      "step": 13102
    },
    {
      "epoch": 0.7337327808265204,
      "grad_norm": 1.2088133096694946,
      "learning_rate": 6.548e-05,
      "loss": 0.4506,
      "step": 13103
    },
    {
      "epoch": 0.7337887781386494,
      "grad_norm": 1.3256062269210815,
      "learning_rate": 6.548500000000001e-05,
      "loss": 0.5224,
      "step": 13104
    },
    {
      "epoch": 0.7338447754507783,
      "grad_norm": 1.1555911302566528,
      "learning_rate": 6.549000000000001e-05,
      "loss": 0.3475,
      "step": 13105
    },
    {
      "epoch": 0.7339007727629073,
      "grad_norm": 1.248432993888855,
      "learning_rate": 6.549500000000001e-05,
      "loss": 0.3861,
      "step": 13106
    },
    {
      "epoch": 0.7339567700750363,
      "grad_norm": 1.095834732055664,
      "learning_rate": 6.55e-05,
      "loss": 0.3302,
      "step": 13107
    },
    {
      "epoch": 0.7340127673871654,
      "grad_norm": 0.9473075866699219,
      "learning_rate": 6.5505e-05,
      "loss": 0.2724,
      "step": 13108
    },
    {
      "epoch": 0.7340687646992944,
      "grad_norm": Infinity,
      "learning_rate": 6.5505e-05,
      "loss": 0.4543,
      "step": 13109
    },
    {
      "epoch": 0.7341247620114234,
      "grad_norm": 1.1974743604660034,
      "learning_rate": 6.551e-05,
      "loss": 0.3362,
      "step": 13110
    },
    {
      "epoch": 0.7341807593235524,
      "grad_norm": 1.695594072341919,
      "learning_rate": 6.5515e-05,
      "loss": 0.4449,
      "step": 13111
    },
    {
      "epoch": 0.7342367566356814,
      "grad_norm": 1.3305492401123047,
      "learning_rate": 6.552000000000001e-05,
      "loss": 0.3226,
      "step": 13112
    },
    {
      "epoch": 0.7342927539478105,
      "grad_norm": 1.1246436834335327,
      "learning_rate": 6.552500000000001e-05,
      "loss": 0.3647,
      "step": 13113
    },
    {
      "epoch": 0.7343487512599395,
      "grad_norm": 1.2378196716308594,
      "learning_rate": 6.553e-05,
      "loss": 0.4981,
      "step": 13114
    },
    {
      "epoch": 0.7344047485720685,
      "grad_norm": 2.289897918701172,
      "learning_rate": 6.5535e-05,
      "loss": 0.5156,
      "step": 13115
    },
    {
      "epoch": 0.7344607458841975,
      "grad_norm": 1.1310378313064575,
      "learning_rate": 6.554e-05,
      "loss": 0.4595,
      "step": 13116
    },
    {
      "epoch": 0.7345167431963265,
      "grad_norm": 1.167104959487915,
      "learning_rate": 6.5545e-05,
      "loss": 0.3827,
      "step": 13117
    },
    {
      "epoch": 0.7345727405084556,
      "grad_norm": 1.4493050575256348,
      "learning_rate": 6.555e-05,
      "loss": 0.4118,
      "step": 13118
    },
    {
      "epoch": 0.7346287378205846,
      "grad_norm": 1.5088547468185425,
      "learning_rate": 6.5555e-05,
      "loss": 0.4831,
      "step": 13119
    },
    {
      "epoch": 0.7346847351327136,
      "grad_norm": 1.1988805532455444,
      "learning_rate": 6.556e-05,
      "loss": 0.419,
      "step": 13120
    },
    {
      "epoch": 0.7347407324448426,
      "grad_norm": 1.561507225036621,
      "learning_rate": 6.5565e-05,
      "loss": 0.5169,
      "step": 13121
    },
    {
      "epoch": 0.7347967297569716,
      "grad_norm": 1.3324174880981445,
      "learning_rate": 6.557e-05,
      "loss": 0.4567,
      "step": 13122
    },
    {
      "epoch": 0.7348527270691007,
      "grad_norm": 1.2111668586730957,
      "learning_rate": 6.557500000000001e-05,
      "loss": 0.4117,
      "step": 13123
    },
    {
      "epoch": 0.7349087243812297,
      "grad_norm": 1.8720570802688599,
      "learning_rate": 6.558e-05,
      "loss": 0.5808,
      "step": 13124
    },
    {
      "epoch": 0.7349647216933587,
      "grad_norm": 1.7627100944519043,
      "learning_rate": 6.5585e-05,
      "loss": 0.4777,
      "step": 13125
    },
    {
      "epoch": 0.7350207190054877,
      "grad_norm": 1.185268759727478,
      "learning_rate": 6.559e-05,
      "loss": 0.3688,
      "step": 13126
    },
    {
      "epoch": 0.7350767163176167,
      "grad_norm": 1.5869851112365723,
      "learning_rate": 6.559500000000001e-05,
      "loss": 0.5782,
      "step": 13127
    },
    {
      "epoch": 0.7351327136297457,
      "grad_norm": 1.3997811079025269,
      "learning_rate": 6.560000000000001e-05,
      "loss": 0.3659,
      "step": 13128
    },
    {
      "epoch": 0.7351887109418748,
      "grad_norm": 1.2208737134933472,
      "learning_rate": 6.5605e-05,
      "loss": 0.4605,
      "step": 13129
    },
    {
      "epoch": 0.7352447082540038,
      "grad_norm": 1.28058922290802,
      "learning_rate": 6.561e-05,
      "loss": 0.537,
      "step": 13130
    },
    {
      "epoch": 0.7353007055661328,
      "grad_norm": 1.0700160264968872,
      "learning_rate": 6.5615e-05,
      "loss": 0.4376,
      "step": 13131
    },
    {
      "epoch": 0.7353567028782618,
      "grad_norm": 1.3895633220672607,
      "learning_rate": 6.562e-05,
      "loss": 0.4032,
      "step": 13132
    },
    {
      "epoch": 0.7354127001903908,
      "grad_norm": 1.2850769758224487,
      "learning_rate": 6.562500000000001e-05,
      "loss": 0.5439,
      "step": 13133
    },
    {
      "epoch": 0.7354686975025199,
      "grad_norm": 1.3515318632125854,
      "learning_rate": 6.563000000000001e-05,
      "loss": 0.5074,
      "step": 13134
    },
    {
      "epoch": 0.7355246948146489,
      "grad_norm": 1.4849662780761719,
      "learning_rate": 6.5635e-05,
      "loss": 0.6059,
      "step": 13135
    },
    {
      "epoch": 0.7355806921267779,
      "grad_norm": 1.2736537456512451,
      "learning_rate": 6.564e-05,
      "loss": 0.5221,
      "step": 13136
    },
    {
      "epoch": 0.7356366894389069,
      "grad_norm": 1.1367994546890259,
      "learning_rate": 6.5645e-05,
      "loss": 0.4085,
      "step": 13137
    },
    {
      "epoch": 0.7356926867510359,
      "grad_norm": 1.3408769369125366,
      "learning_rate": 6.565e-05,
      "loss": 0.4669,
      "step": 13138
    },
    {
      "epoch": 0.735748684063165,
      "grad_norm": 1.2693918943405151,
      "learning_rate": 6.5655e-05,
      "loss": 0.4661,
      "step": 13139
    },
    {
      "epoch": 0.735804681375294,
      "grad_norm": 1.6538734436035156,
      "learning_rate": 6.566e-05,
      "loss": 0.4861,
      "step": 13140
    },
    {
      "epoch": 0.735860678687423,
      "grad_norm": 1.1471478939056396,
      "learning_rate": 6.5665e-05,
      "loss": 0.4333,
      "step": 13141
    },
    {
      "epoch": 0.735916675999552,
      "grad_norm": 1.5495682954788208,
      "learning_rate": 6.567e-05,
      "loss": 0.4726,
      "step": 13142
    },
    {
      "epoch": 0.735972673311681,
      "grad_norm": 1.4408315420150757,
      "learning_rate": 6.5675e-05,
      "loss": 0.5122,
      "step": 13143
    },
    {
      "epoch": 0.73602867062381,
      "grad_norm": 1.115729808807373,
      "learning_rate": 6.568000000000001e-05,
      "loss": 0.3786,
      "step": 13144
    },
    {
      "epoch": 0.7360846679359391,
      "grad_norm": 1.2041523456573486,
      "learning_rate": 6.5685e-05,
      "loss": 0.3107,
      "step": 13145
    },
    {
      "epoch": 0.7361406652480681,
      "grad_norm": 1.2265079021453857,
      "learning_rate": 6.569e-05,
      "loss": 0.3852,
      "step": 13146
    },
    {
      "epoch": 0.7361966625601971,
      "grad_norm": 1.2132033109664917,
      "learning_rate": 6.5695e-05,
      "loss": 0.475,
      "step": 13147
    },
    {
      "epoch": 0.7362526598723261,
      "grad_norm": 1.082369089126587,
      "learning_rate": 6.570000000000001e-05,
      "loss": 0.3402,
      "step": 13148
    },
    {
      "epoch": 0.7363086571844552,
      "grad_norm": 1.333266019821167,
      "learning_rate": 6.570500000000001e-05,
      "loss": 0.4125,
      "step": 13149
    },
    {
      "epoch": 0.7363646544965842,
      "grad_norm": 1.2038555145263672,
      "learning_rate": 6.571e-05,
      "loss": 0.3986,
      "step": 13150
    },
    {
      "epoch": 0.7364206518087132,
      "grad_norm": 1.3272336721420288,
      "learning_rate": 6.5715e-05,
      "loss": 0.4539,
      "step": 13151
    },
    {
      "epoch": 0.7364766491208422,
      "grad_norm": 1.3587881326675415,
      "learning_rate": 6.572e-05,
      "loss": 0.4755,
      "step": 13152
    },
    {
      "epoch": 0.7365326464329712,
      "grad_norm": 1.3513455390930176,
      "learning_rate": 6.5725e-05,
      "loss": 0.4184,
      "step": 13153
    },
    {
      "epoch": 0.7365886437451002,
      "grad_norm": 1.403626561164856,
      "learning_rate": 6.573000000000001e-05,
      "loss": 0.5434,
      "step": 13154
    },
    {
      "epoch": 0.7366446410572293,
      "grad_norm": 1.3098336458206177,
      "learning_rate": 6.5735e-05,
      "loss": 0.4434,
      "step": 13155
    },
    {
      "epoch": 0.7367006383693583,
      "grad_norm": 1.129221796989441,
      "learning_rate": 6.574e-05,
      "loss": 0.3664,
      "step": 13156
    },
    {
      "epoch": 0.7367566356814873,
      "grad_norm": 1.4928340911865234,
      "learning_rate": 6.5745e-05,
      "loss": 0.4269,
      "step": 13157
    },
    {
      "epoch": 0.7368126329936163,
      "grad_norm": 1.458243727684021,
      "learning_rate": 6.575e-05,
      "loss": 0.44,
      "step": 13158
    },
    {
      "epoch": 0.7368686303057453,
      "grad_norm": 1.3369060754776,
      "learning_rate": 6.5755e-05,
      "loss": 0.5347,
      "step": 13159
    },
    {
      "epoch": 0.7369246276178744,
      "grad_norm": 1.359553337097168,
      "learning_rate": 6.576e-05,
      "loss": 0.3842,
      "step": 13160
    },
    {
      "epoch": 0.7369806249300034,
      "grad_norm": 1.1033194065093994,
      "learning_rate": 6.5765e-05,
      "loss": 0.3966,
      "step": 13161
    },
    {
      "epoch": 0.7370366222421324,
      "grad_norm": 1.2944777011871338,
      "learning_rate": 6.577e-05,
      "loss": 0.5503,
      "step": 13162
    },
    {
      "epoch": 0.7370926195542614,
      "grad_norm": 1.1972578763961792,
      "learning_rate": 6.5775e-05,
      "loss": 0.4032,
      "step": 13163
    },
    {
      "epoch": 0.7371486168663904,
      "grad_norm": 1.368035078048706,
      "learning_rate": 6.578000000000001e-05,
      "loss": 0.4314,
      "step": 13164
    },
    {
      "epoch": 0.7372046141785195,
      "grad_norm": 1.3839082717895508,
      "learning_rate": 6.578500000000001e-05,
      "loss": 0.3676,
      "step": 13165
    },
    {
      "epoch": 0.7372606114906485,
      "grad_norm": 1.2854747772216797,
      "learning_rate": 6.579e-05,
      "loss": 0.4638,
      "step": 13166
    },
    {
      "epoch": 0.7373166088027775,
      "grad_norm": 1.4696745872497559,
      "learning_rate": 6.5795e-05,
      "loss": 0.4465,
      "step": 13167
    },
    {
      "epoch": 0.7373726061149065,
      "grad_norm": 1.368017315864563,
      "learning_rate": 6.58e-05,
      "loss": 0.4276,
      "step": 13168
    },
    {
      "epoch": 0.7374286034270355,
      "grad_norm": 1.272612452507019,
      "learning_rate": 6.580500000000001e-05,
      "loss": 0.4243,
      "step": 13169
    },
    {
      "epoch": 0.7374846007391646,
      "grad_norm": 1.2280429601669312,
      "learning_rate": 6.581000000000001e-05,
      "loss": 0.5592,
      "step": 13170
    },
    {
      "epoch": 0.7375405980512936,
      "grad_norm": 1.3454099893569946,
      "learning_rate": 6.5815e-05,
      "loss": 0.4519,
      "step": 13171
    },
    {
      "epoch": 0.7375965953634226,
      "grad_norm": 1.195507526397705,
      "learning_rate": 6.582e-05,
      "loss": 0.384,
      "step": 13172
    },
    {
      "epoch": 0.7376525926755516,
      "grad_norm": 0.985457718372345,
      "learning_rate": 6.5825e-05,
      "loss": 0.3247,
      "step": 13173
    },
    {
      "epoch": 0.7377085899876806,
      "grad_norm": 1.602728247642517,
      "learning_rate": 6.583e-05,
      "loss": 0.4885,
      "step": 13174
    },
    {
      "epoch": 0.7377645872998096,
      "grad_norm": 1.2869107723236084,
      "learning_rate": 6.5835e-05,
      "loss": 0.3641,
      "step": 13175
    },
    {
      "epoch": 0.7378205846119387,
      "grad_norm": 1.3212437629699707,
      "learning_rate": 6.584e-05,
      "loss": 0.4423,
      "step": 13176
    },
    {
      "epoch": 0.7378765819240677,
      "grad_norm": 2.345132350921631,
      "learning_rate": 6.5845e-05,
      "loss": 0.3675,
      "step": 13177
    },
    {
      "epoch": 0.7379325792361967,
      "grad_norm": 1.5399130582809448,
      "learning_rate": 6.585e-05,
      "loss": 0.6307,
      "step": 13178
    },
    {
      "epoch": 0.7379885765483257,
      "grad_norm": 1.4886460304260254,
      "learning_rate": 6.5855e-05,
      "loss": 0.3838,
      "step": 13179
    },
    {
      "epoch": 0.7380445738604547,
      "grad_norm": 1.2662973403930664,
      "learning_rate": 6.586e-05,
      "loss": 0.4451,
      "step": 13180
    },
    {
      "epoch": 0.7381005711725838,
      "grad_norm": 1.113808035850525,
      "learning_rate": 6.5865e-05,
      "loss": 0.4122,
      "step": 13181
    },
    {
      "epoch": 0.7381565684847128,
      "grad_norm": 1.5719285011291504,
      "learning_rate": 6.587e-05,
      "loss": 0.4248,
      "step": 13182
    },
    {
      "epoch": 0.7382125657968418,
      "grad_norm": 1.3040238618850708,
      "learning_rate": 6.5875e-05,
      "loss": 0.4513,
      "step": 13183
    },
    {
      "epoch": 0.7382685631089708,
      "grad_norm": 1.3562145233154297,
      "learning_rate": 6.588000000000001e-05,
      "loss": 0.4359,
      "step": 13184
    },
    {
      "epoch": 0.7383245604210998,
      "grad_norm": 1.2630589008331299,
      "learning_rate": 6.588500000000001e-05,
      "loss": 0.4191,
      "step": 13185
    },
    {
      "epoch": 0.7383805577332289,
      "grad_norm": 1.2360169887542725,
      "learning_rate": 6.589000000000001e-05,
      "loss": 0.4591,
      "step": 13186
    },
    {
      "epoch": 0.7384365550453578,
      "grad_norm": 1.296000361442566,
      "learning_rate": 6.5895e-05,
      "loss": 0.4327,
      "step": 13187
    },
    {
      "epoch": 0.7384925523574868,
      "grad_norm": 1.523987054824829,
      "learning_rate": 6.59e-05,
      "loss": 0.456,
      "step": 13188
    },
    {
      "epoch": 0.7385485496696158,
      "grad_norm": 1.7142254114151,
      "learning_rate": 6.5905e-05,
      "loss": 0.5521,
      "step": 13189
    },
    {
      "epoch": 0.7386045469817448,
      "grad_norm": 1.0339454412460327,
      "learning_rate": 6.591000000000001e-05,
      "loss": 0.3673,
      "step": 13190
    },
    {
      "epoch": 0.7386605442938738,
      "grad_norm": 1.3455902338027954,
      "learning_rate": 6.591500000000001e-05,
      "loss": 0.4764,
      "step": 13191
    },
    {
      "epoch": 0.7387165416060029,
      "grad_norm": 1.2935000658035278,
      "learning_rate": 6.592e-05,
      "loss": 0.4256,
      "step": 13192
    },
    {
      "epoch": 0.7387725389181319,
      "grad_norm": 1.2083535194396973,
      "learning_rate": 6.5925e-05,
      "loss": 0.3604,
      "step": 13193
    },
    {
      "epoch": 0.7388285362302609,
      "grad_norm": 1.4803552627563477,
      "learning_rate": 6.593e-05,
      "loss": 0.4738,
      "step": 13194
    },
    {
      "epoch": 0.7388845335423899,
      "grad_norm": 1.3232859373092651,
      "learning_rate": 6.5935e-05,
      "loss": 0.4574,
      "step": 13195
    },
    {
      "epoch": 0.7389405308545189,
      "grad_norm": 1.2311434745788574,
      "learning_rate": 6.594e-05,
      "loss": 0.5001,
      "step": 13196
    },
    {
      "epoch": 0.738996528166648,
      "grad_norm": 1.3587087392807007,
      "learning_rate": 6.5945e-05,
      "loss": 0.5302,
      "step": 13197
    },
    {
      "epoch": 0.739052525478777,
      "grad_norm": 1.3923612833023071,
      "learning_rate": 6.595e-05,
      "loss": 0.3617,
      "step": 13198
    },
    {
      "epoch": 0.739108522790906,
      "grad_norm": 1.3061606884002686,
      "learning_rate": 6.5955e-05,
      "loss": 0.5975,
      "step": 13199
    },
    {
      "epoch": 0.739164520103035,
      "grad_norm": 1.2932555675506592,
      "learning_rate": 6.596e-05,
      "loss": 0.4979,
      "step": 13200
    },
    {
      "epoch": 0.739220517415164,
      "grad_norm": 2.5977580547332764,
      "learning_rate": 6.5965e-05,
      "loss": 0.3575,
      "step": 13201
    },
    {
      "epoch": 0.739276514727293,
      "grad_norm": 1.2210155725479126,
      "learning_rate": 6.597e-05,
      "loss": 0.5616,
      "step": 13202
    },
    {
      "epoch": 0.7393325120394221,
      "grad_norm": 1.3751977682113647,
      "learning_rate": 6.5975e-05,
      "loss": 0.4162,
      "step": 13203
    },
    {
      "epoch": 0.7393885093515511,
      "grad_norm": 1.316551685333252,
      "learning_rate": 6.598e-05,
      "loss": 0.5675,
      "step": 13204
    },
    {
      "epoch": 0.7394445066636801,
      "grad_norm": 1.4882951974868774,
      "learning_rate": 6.598500000000001e-05,
      "loss": 0.5544,
      "step": 13205
    },
    {
      "epoch": 0.7395005039758091,
      "grad_norm": 1.2412019968032837,
      "learning_rate": 6.599000000000001e-05,
      "loss": 0.3857,
      "step": 13206
    },
    {
      "epoch": 0.7395565012879382,
      "grad_norm": 1.1741154193878174,
      "learning_rate": 6.599500000000001e-05,
      "loss": 0.4972,
      "step": 13207
    },
    {
      "epoch": 0.7396124986000672,
      "grad_norm": 1.460400938987732,
      "learning_rate": 6.6e-05,
      "loss": 0.4688,
      "step": 13208
    },
    {
      "epoch": 0.7396684959121962,
      "grad_norm": 1.3373973369598389,
      "learning_rate": 6.6005e-05,
      "loss": 0.3796,
      "step": 13209
    },
    {
      "epoch": 0.7397244932243252,
      "grad_norm": 1.4087990522384644,
      "learning_rate": 6.601e-05,
      "loss": 0.7111,
      "step": 13210
    },
    {
      "epoch": 0.7397804905364542,
      "grad_norm": 1.1778781414031982,
      "learning_rate": 6.601500000000001e-05,
      "loss": 0.2816,
      "step": 13211
    },
    {
      "epoch": 0.7398364878485832,
      "grad_norm": 1.573437213897705,
      "learning_rate": 6.602000000000001e-05,
      "loss": 0.5817,
      "step": 13212
    },
    {
      "epoch": 0.7398924851607123,
      "grad_norm": 1.4049535989761353,
      "learning_rate": 6.6025e-05,
      "loss": 0.3513,
      "step": 13213
    },
    {
      "epoch": 0.7399484824728413,
      "grad_norm": 1.3372673988342285,
      "learning_rate": 6.603e-05,
      "loss": 0.523,
      "step": 13214
    },
    {
      "epoch": 0.7400044797849703,
      "grad_norm": 1.2721821069717407,
      "learning_rate": 6.6035e-05,
      "loss": 0.5244,
      "step": 13215
    },
    {
      "epoch": 0.7400604770970993,
      "grad_norm": 1.2308804988861084,
      "learning_rate": 6.604e-05,
      "loss": 0.3764,
      "step": 13216
    },
    {
      "epoch": 0.7401164744092283,
      "grad_norm": 1.28029465675354,
      "learning_rate": 6.6045e-05,
      "loss": 0.4245,
      "step": 13217
    },
    {
      "epoch": 0.7401724717213574,
      "grad_norm": 1.2959840297698975,
      "learning_rate": 6.605e-05,
      "loss": 0.5164,
      "step": 13218
    },
    {
      "epoch": 0.7402284690334864,
      "grad_norm": 1.1691176891326904,
      "learning_rate": 6.6055e-05,
      "loss": 0.4248,
      "step": 13219
    },
    {
      "epoch": 0.7402844663456154,
      "grad_norm": 1.4101749658584595,
      "learning_rate": 6.606e-05,
      "loss": 0.4106,
      "step": 13220
    },
    {
      "epoch": 0.7403404636577444,
      "grad_norm": 1.2446681261062622,
      "learning_rate": 6.6065e-05,
      "loss": 0.5375,
      "step": 13221
    },
    {
      "epoch": 0.7403964609698734,
      "grad_norm": 1.48183274269104,
      "learning_rate": 6.607e-05,
      "loss": 0.543,
      "step": 13222
    },
    {
      "epoch": 0.7404524582820025,
      "grad_norm": 1.2818819284439087,
      "learning_rate": 6.6075e-05,
      "loss": 0.3969,
      "step": 13223
    },
    {
      "epoch": 0.7405084555941315,
      "grad_norm": 1.3402365446090698,
      "learning_rate": 6.608e-05,
      "loss": 0.3758,
      "step": 13224
    },
    {
      "epoch": 0.7405644529062605,
      "grad_norm": 1.1508859395980835,
      "learning_rate": 6.6085e-05,
      "loss": 0.4746,
      "step": 13225
    },
    {
      "epoch": 0.7406204502183895,
      "grad_norm": 1.318395733833313,
      "learning_rate": 6.609000000000001e-05,
      "loss": 0.4383,
      "step": 13226
    },
    {
      "epoch": 0.7406764475305185,
      "grad_norm": 1.2404429912567139,
      "learning_rate": 6.609500000000001e-05,
      "loss": 0.4597,
      "step": 13227
    },
    {
      "epoch": 0.7407324448426476,
      "grad_norm": 1.0450571775436401,
      "learning_rate": 6.610000000000001e-05,
      "loss": 0.3557,
      "step": 13228
    },
    {
      "epoch": 0.7407884421547766,
      "grad_norm": 1.453265905380249,
      "learning_rate": 6.6105e-05,
      "loss": 0.4221,
      "step": 13229
    },
    {
      "epoch": 0.7408444394669056,
      "grad_norm": 1.6890655755996704,
      "learning_rate": 6.611e-05,
      "loss": 0.4914,
      "step": 13230
    },
    {
      "epoch": 0.7409004367790346,
      "grad_norm": 1.4456104040145874,
      "learning_rate": 6.6115e-05,
      "loss": 0.343,
      "step": 13231
    },
    {
      "epoch": 0.7409564340911636,
      "grad_norm": 1.3714569807052612,
      "learning_rate": 6.612000000000001e-05,
      "loss": 0.365,
      "step": 13232
    },
    {
      "epoch": 0.7410124314032926,
      "grad_norm": 1.425466537475586,
      "learning_rate": 6.612500000000001e-05,
      "loss": 0.431,
      "step": 13233
    },
    {
      "epoch": 0.7410684287154217,
      "grad_norm": 1.1059119701385498,
      "learning_rate": 6.613e-05,
      "loss": 0.4517,
      "step": 13234
    },
    {
      "epoch": 0.7411244260275507,
      "grad_norm": 1.4817248582839966,
      "learning_rate": 6.6135e-05,
      "loss": 0.456,
      "step": 13235
    },
    {
      "epoch": 0.7411804233396797,
      "grad_norm": 1.1598478555679321,
      "learning_rate": 6.614e-05,
      "loss": 0.3313,
      "step": 13236
    },
    {
      "epoch": 0.7412364206518087,
      "grad_norm": 1.3717542886734009,
      "learning_rate": 6.6145e-05,
      "loss": 0.5052,
      "step": 13237
    },
    {
      "epoch": 0.7412924179639377,
      "grad_norm": 1.4055230617523193,
      "learning_rate": 6.615e-05,
      "loss": 0.4718,
      "step": 13238
    },
    {
      "epoch": 0.7413484152760668,
      "grad_norm": 2.4605820178985596,
      "learning_rate": 6.6155e-05,
      "loss": 0.4463,
      "step": 13239
    },
    {
      "epoch": 0.7414044125881958,
      "grad_norm": 1.2312734127044678,
      "learning_rate": 6.616e-05,
      "loss": 0.4432,
      "step": 13240
    },
    {
      "epoch": 0.7414604099003248,
      "grad_norm": 1.5527117252349854,
      "learning_rate": 6.6165e-05,
      "loss": 0.7414,
      "step": 13241
    },
    {
      "epoch": 0.7415164072124538,
      "grad_norm": 1.4419389963150024,
      "learning_rate": 6.617e-05,
      "loss": 0.4566,
      "step": 13242
    },
    {
      "epoch": 0.7415724045245828,
      "grad_norm": 1.1105155944824219,
      "learning_rate": 6.6175e-05,
      "loss": 0.3224,
      "step": 13243
    },
    {
      "epoch": 0.7416284018367119,
      "grad_norm": 1.355978012084961,
      "learning_rate": 6.618e-05,
      "loss": 0.5827,
      "step": 13244
    },
    {
      "epoch": 0.7416843991488409,
      "grad_norm": 1.4871820211410522,
      "learning_rate": 6.6185e-05,
      "loss": 0.5677,
      "step": 13245
    },
    {
      "epoch": 0.7417403964609699,
      "grad_norm": 1.2730991840362549,
      "learning_rate": 6.619e-05,
      "loss": 0.4863,
      "step": 13246
    },
    {
      "epoch": 0.7417963937730989,
      "grad_norm": 6.5206217765808105,
      "learning_rate": 6.619500000000001e-05,
      "loss": 0.4476,
      "step": 13247
    },
    {
      "epoch": 0.7418523910852279,
      "grad_norm": 1.3759405612945557,
      "learning_rate": 6.620000000000001e-05,
      "loss": 0.4373,
      "step": 13248
    },
    {
      "epoch": 0.741908388397357,
      "grad_norm": 1.4291212558746338,
      "learning_rate": 6.620500000000001e-05,
      "loss": 0.4196,
      "step": 13249
    },
    {
      "epoch": 0.741964385709486,
      "grad_norm": 1.3058536052703857,
      "learning_rate": 6.621e-05,
      "loss": 0.4366,
      "step": 13250
    },
    {
      "epoch": 0.742020383021615,
      "grad_norm": 1.2478581666946411,
      "learning_rate": 6.6215e-05,
      "loss": 0.3752,
      "step": 13251
    },
    {
      "epoch": 0.742076380333744,
      "grad_norm": 1.4263951778411865,
      "learning_rate": 6.622e-05,
      "loss": 0.5217,
      "step": 13252
    },
    {
      "epoch": 0.742132377645873,
      "grad_norm": 1.5010628700256348,
      "learning_rate": 6.6225e-05,
      "loss": 0.5169,
      "step": 13253
    },
    {
      "epoch": 0.742188374958002,
      "grad_norm": 1.0944013595581055,
      "learning_rate": 6.623000000000001e-05,
      "loss": 0.3406,
      "step": 13254
    },
    {
      "epoch": 0.7422443722701311,
      "grad_norm": 1.07964289188385,
      "learning_rate": 6.6235e-05,
      "loss": 0.3515,
      "step": 13255
    },
    {
      "epoch": 0.7423003695822601,
      "grad_norm": 1.139193058013916,
      "learning_rate": 6.624e-05,
      "loss": 0.4078,
      "step": 13256
    },
    {
      "epoch": 0.7423563668943891,
      "grad_norm": 1.1736921072006226,
      "learning_rate": 6.6245e-05,
      "loss": 0.359,
      "step": 13257
    },
    {
      "epoch": 0.7424123642065181,
      "grad_norm": 1.059420108795166,
      "learning_rate": 6.625e-05,
      "loss": 0.4044,
      "step": 13258
    },
    {
      "epoch": 0.7424683615186471,
      "grad_norm": 1.1916396617889404,
      "learning_rate": 6.6255e-05,
      "loss": 0.3729,
      "step": 13259
    },
    {
      "epoch": 0.7425243588307762,
      "grad_norm": 1.566210150718689,
      "learning_rate": 6.626e-05,
      "loss": 0.4848,
      "step": 13260
    },
    {
      "epoch": 0.7425803561429052,
      "grad_norm": 1.3561042547225952,
      "learning_rate": 6.6265e-05,
      "loss": 0.437,
      "step": 13261
    },
    {
      "epoch": 0.7426363534550342,
      "grad_norm": 1.1792902946472168,
      "learning_rate": 6.627e-05,
      "loss": 0.3632,
      "step": 13262
    },
    {
      "epoch": 0.7426923507671632,
      "grad_norm": 1.3329291343688965,
      "learning_rate": 6.6275e-05,
      "loss": 0.4487,
      "step": 13263
    },
    {
      "epoch": 0.7427483480792922,
      "grad_norm": 1.4227898120880127,
      "learning_rate": 6.628e-05,
      "loss": 0.5119,
      "step": 13264
    },
    {
      "epoch": 0.7428043453914213,
      "grad_norm": 1.191625714302063,
      "learning_rate": 6.6285e-05,
      "loss": 0.4121,
      "step": 13265
    },
    {
      "epoch": 0.7428603427035503,
      "grad_norm": 1.4005130529403687,
      "learning_rate": 6.629e-05,
      "loss": 0.5107,
      "step": 13266
    },
    {
      "epoch": 0.7429163400156793,
      "grad_norm": 1.187508225440979,
      "learning_rate": 6.6295e-05,
      "loss": 0.4321,
      "step": 13267
    },
    {
      "epoch": 0.7429723373278083,
      "grad_norm": 1.2492766380310059,
      "learning_rate": 6.630000000000001e-05,
      "loss": 0.4224,
      "step": 13268
    },
    {
      "epoch": 0.7430283346399373,
      "grad_norm": 1.3098195791244507,
      "learning_rate": 6.630500000000001e-05,
      "loss": 0.3745,
      "step": 13269
    },
    {
      "epoch": 0.7430843319520662,
      "grad_norm": 1.9010365009307861,
      "learning_rate": 6.631000000000001e-05,
      "loss": 0.5458,
      "step": 13270
    },
    {
      "epoch": 0.7431403292641953,
      "grad_norm": 1.2563695907592773,
      "learning_rate": 6.6315e-05,
      "loss": 0.3183,
      "step": 13271
    },
    {
      "epoch": 0.7431963265763243,
      "grad_norm": 1.824275255203247,
      "learning_rate": 6.632e-05,
      "loss": 0.5482,
      "step": 13272
    },
    {
      "epoch": 0.7432523238884533,
      "grad_norm": 1.6714365482330322,
      "learning_rate": 6.6325e-05,
      "loss": 0.392,
      "step": 13273
    },
    {
      "epoch": 0.7433083212005823,
      "grad_norm": 1.7867680788040161,
      "learning_rate": 6.633e-05,
      "loss": 0.5196,
      "step": 13274
    },
    {
      "epoch": 0.7433643185127113,
      "grad_norm": 1.5641555786132812,
      "learning_rate": 6.633500000000001e-05,
      "loss": 0.5168,
      "step": 13275
    },
    {
      "epoch": 0.7434203158248404,
      "grad_norm": 1.3071430921554565,
      "learning_rate": 6.634e-05,
      "loss": 0.6255,
      "step": 13276
    },
    {
      "epoch": 0.7434763131369694,
      "grad_norm": 1.3810094594955444,
      "learning_rate": 6.6345e-05,
      "loss": 0.5977,
      "step": 13277
    },
    {
      "epoch": 0.7435323104490984,
      "grad_norm": 1.140381932258606,
      "learning_rate": 6.635e-05,
      "loss": 0.3518,
      "step": 13278
    },
    {
      "epoch": 0.7435883077612274,
      "grad_norm": 1.3058286905288696,
      "learning_rate": 6.6355e-05,
      "loss": 0.5564,
      "step": 13279
    },
    {
      "epoch": 0.7436443050733564,
      "grad_norm": 1.2812680006027222,
      "learning_rate": 6.636e-05,
      "loss": 0.391,
      "step": 13280
    },
    {
      "epoch": 0.7437003023854855,
      "grad_norm": 1.4396165609359741,
      "learning_rate": 6.6365e-05,
      "loss": 0.4724,
      "step": 13281
    },
    {
      "epoch": 0.7437562996976145,
      "grad_norm": 1.5894254446029663,
      "learning_rate": 6.637e-05,
      "loss": 0.4654,
      "step": 13282
    },
    {
      "epoch": 0.7438122970097435,
      "grad_norm": 1.2542455196380615,
      "learning_rate": 6.6375e-05,
      "loss": 0.3891,
      "step": 13283
    },
    {
      "epoch": 0.7438682943218725,
      "grad_norm": 1.3622913360595703,
      "learning_rate": 6.638e-05,
      "loss": 0.4856,
      "step": 13284
    },
    {
      "epoch": 0.7439242916340015,
      "grad_norm": 1.3187345266342163,
      "learning_rate": 6.638500000000001e-05,
      "loss": 0.4348,
      "step": 13285
    },
    {
      "epoch": 0.7439802889461306,
      "grad_norm": 1.3023624420166016,
      "learning_rate": 6.639e-05,
      "loss": 0.548,
      "step": 13286
    },
    {
      "epoch": 0.7440362862582596,
      "grad_norm": 1.7420628070831299,
      "learning_rate": 6.6395e-05,
      "loss": 0.5877,
      "step": 13287
    },
    {
      "epoch": 0.7440922835703886,
      "grad_norm": 1.394727349281311,
      "learning_rate": 6.64e-05,
      "loss": 0.4225,
      "step": 13288
    },
    {
      "epoch": 0.7441482808825176,
      "grad_norm": 1.368857741355896,
      "learning_rate": 6.640500000000001e-05,
      "loss": 0.3973,
      "step": 13289
    },
    {
      "epoch": 0.7442042781946466,
      "grad_norm": 1.4397590160369873,
      "learning_rate": 6.641000000000001e-05,
      "loss": 0.4232,
      "step": 13290
    },
    {
      "epoch": 0.7442602755067756,
      "grad_norm": 1.462835669517517,
      "learning_rate": 6.641500000000001e-05,
      "loss": 0.408,
      "step": 13291
    },
    {
      "epoch": 0.7443162728189047,
      "grad_norm": 1.2053711414337158,
      "learning_rate": 6.642e-05,
      "loss": 0.3666,
      "step": 13292
    },
    {
      "epoch": 0.7443722701310337,
      "grad_norm": 1.3954273462295532,
      "learning_rate": 6.6425e-05,
      "loss": 0.4062,
      "step": 13293
    },
    {
      "epoch": 0.7444282674431627,
      "grad_norm": 1.2463713884353638,
      "learning_rate": 6.643e-05,
      "loss": 0.4379,
      "step": 13294
    },
    {
      "epoch": 0.7444842647552917,
      "grad_norm": 1.265592098236084,
      "learning_rate": 6.6435e-05,
      "loss": 0.4643,
      "step": 13295
    },
    {
      "epoch": 0.7445402620674207,
      "grad_norm": 1.3765982389450073,
      "learning_rate": 6.644000000000001e-05,
      "loss": 0.4802,
      "step": 13296
    },
    {
      "epoch": 0.7445962593795498,
      "grad_norm": 1.4257162809371948,
      "learning_rate": 6.6445e-05,
      "loss": 0.4158,
      "step": 13297
    },
    {
      "epoch": 0.7446522566916788,
      "grad_norm": 1.2696559429168701,
      "learning_rate": 6.645e-05,
      "loss": 0.4479,
      "step": 13298
    },
    {
      "epoch": 0.7447082540038078,
      "grad_norm": 1.3374682664871216,
      "learning_rate": 6.6455e-05,
      "loss": 0.3887,
      "step": 13299
    },
    {
      "epoch": 0.7447642513159368,
      "grad_norm": 1.3666870594024658,
      "learning_rate": 6.646e-05,
      "loss": 0.4426,
      "step": 13300
    },
    {
      "epoch": 0.7448202486280658,
      "grad_norm": 1.2497977018356323,
      "learning_rate": 6.6465e-05,
      "loss": 0.5023,
      "step": 13301
    },
    {
      "epoch": 0.7448762459401949,
      "grad_norm": 1.2170549631118774,
      "learning_rate": 6.647e-05,
      "loss": 0.4167,
      "step": 13302
    },
    {
      "epoch": 0.7449322432523239,
      "grad_norm": 1.234202265739441,
      "learning_rate": 6.6475e-05,
      "loss": 0.4098,
      "step": 13303
    },
    {
      "epoch": 0.7449882405644529,
      "grad_norm": 1.3242830038070679,
      "learning_rate": 6.648e-05,
      "loss": 0.4357,
      "step": 13304
    },
    {
      "epoch": 0.7450442378765819,
      "grad_norm": 1.7305229902267456,
      "learning_rate": 6.648500000000001e-05,
      "loss": 0.4228,
      "step": 13305
    },
    {
      "epoch": 0.7451002351887109,
      "grad_norm": 1.6236753463745117,
      "learning_rate": 6.649000000000001e-05,
      "loss": 0.5207,
      "step": 13306
    },
    {
      "epoch": 0.74515623250084,
      "grad_norm": 1.654104471206665,
      "learning_rate": 6.6495e-05,
      "loss": 0.5178,
      "step": 13307
    },
    {
      "epoch": 0.745212229812969,
      "grad_norm": 1.4475064277648926,
      "learning_rate": 6.65e-05,
      "loss": 0.5265,
      "step": 13308
    },
    {
      "epoch": 0.745268227125098,
      "grad_norm": 1.9489260911941528,
      "learning_rate": 6.6505e-05,
      "loss": 0.4266,
      "step": 13309
    },
    {
      "epoch": 0.745324224437227,
      "grad_norm": 1.3361420631408691,
      "learning_rate": 6.651000000000001e-05,
      "loss": 0.3858,
      "step": 13310
    },
    {
      "epoch": 0.745380221749356,
      "grad_norm": 1.3716928958892822,
      "learning_rate": 6.651500000000001e-05,
      "loss": 0.4142,
      "step": 13311
    },
    {
      "epoch": 0.745436219061485,
      "grad_norm": 1.0638846158981323,
      "learning_rate": 6.652000000000001e-05,
      "loss": 0.3401,
      "step": 13312
    },
    {
      "epoch": 0.7454922163736141,
      "grad_norm": 1.6768540143966675,
      "learning_rate": 6.6525e-05,
      "loss": 0.4647,
      "step": 13313
    },
    {
      "epoch": 0.7455482136857431,
      "grad_norm": 1.8939396142959595,
      "learning_rate": 6.653e-05,
      "loss": 0.4594,
      "step": 13314
    },
    {
      "epoch": 0.7456042109978721,
      "grad_norm": 1.3508588075637817,
      "learning_rate": 6.6535e-05,
      "loss": 0.345,
      "step": 13315
    },
    {
      "epoch": 0.7456602083100011,
      "grad_norm": 1.226591944694519,
      "learning_rate": 6.654e-05,
      "loss": 0.3969,
      "step": 13316
    },
    {
      "epoch": 0.7457162056221301,
      "grad_norm": 1.1925842761993408,
      "learning_rate": 6.654500000000001e-05,
      "loss": 0.5444,
      "step": 13317
    },
    {
      "epoch": 0.7457722029342592,
      "grad_norm": 1.6634101867675781,
      "learning_rate": 6.655e-05,
      "loss": 0.5147,
      "step": 13318
    },
    {
      "epoch": 0.7458282002463882,
      "grad_norm": 1.3468554019927979,
      "learning_rate": 6.6555e-05,
      "loss": 0.4423,
      "step": 13319
    },
    {
      "epoch": 0.7458841975585172,
      "grad_norm": 1.2828065156936646,
      "learning_rate": 6.656e-05,
      "loss": 0.3592,
      "step": 13320
    },
    {
      "epoch": 0.7459401948706462,
      "grad_norm": 1.5055537223815918,
      "learning_rate": 6.6565e-05,
      "loss": 0.445,
      "step": 13321
    },
    {
      "epoch": 0.7459961921827752,
      "grad_norm": 1.1748046875,
      "learning_rate": 6.657e-05,
      "loss": 0.3662,
      "step": 13322
    },
    {
      "epoch": 0.7460521894949043,
      "grad_norm": 1.1813485622406006,
      "learning_rate": 6.657499999999999e-05,
      "loss": 0.3375,
      "step": 13323
    },
    {
      "epoch": 0.7461081868070333,
      "grad_norm": 1.186800241470337,
      "learning_rate": 6.658e-05,
      "loss": 0.4368,
      "step": 13324
    },
    {
      "epoch": 0.7461641841191623,
      "grad_norm": 1.5023611783981323,
      "learning_rate": 6.658500000000001e-05,
      "loss": 0.4546,
      "step": 13325
    },
    {
      "epoch": 0.7462201814312913,
      "grad_norm": 1.4181703329086304,
      "learning_rate": 6.659000000000001e-05,
      "loss": 0.4254,
      "step": 13326
    },
    {
      "epoch": 0.7462761787434203,
      "grad_norm": 1.2328901290893555,
      "learning_rate": 6.659500000000001e-05,
      "loss": 0.37,
      "step": 13327
    },
    {
      "epoch": 0.7463321760555494,
      "grad_norm": 2.5455312728881836,
      "learning_rate": 6.66e-05,
      "loss": 0.5461,
      "step": 13328
    },
    {
      "epoch": 0.7463881733676784,
      "grad_norm": 1.4440556764602661,
      "learning_rate": 6.6605e-05,
      "loss": 0.5065,
      "step": 13329
    },
    {
      "epoch": 0.7464441706798074,
      "grad_norm": 1.3674674034118652,
      "learning_rate": 6.661e-05,
      "loss": 0.4871,
      "step": 13330
    },
    {
      "epoch": 0.7465001679919364,
      "grad_norm": 1.402139663696289,
      "learning_rate": 6.661500000000001e-05,
      "loss": 0.4866,
      "step": 13331
    },
    {
      "epoch": 0.7465561653040654,
      "grad_norm": 1.5644489526748657,
      "learning_rate": 6.662000000000001e-05,
      "loss": 0.4805,
      "step": 13332
    },
    {
      "epoch": 0.7466121626161945,
      "grad_norm": 1.9279708862304688,
      "learning_rate": 6.6625e-05,
      "loss": 0.6546,
      "step": 13333
    },
    {
      "epoch": 0.7466681599283235,
      "grad_norm": 1.2155046463012695,
      "learning_rate": 6.663e-05,
      "loss": 0.3641,
      "step": 13334
    },
    {
      "epoch": 0.7467241572404525,
      "grad_norm": 1.3035589456558228,
      "learning_rate": 6.6635e-05,
      "loss": 0.3426,
      "step": 13335
    },
    {
      "epoch": 0.7467801545525815,
      "grad_norm": 1.1661807298660278,
      "learning_rate": 6.664e-05,
      "loss": 0.4449,
      "step": 13336
    },
    {
      "epoch": 0.7468361518647105,
      "grad_norm": 1.3552536964416504,
      "learning_rate": 6.6645e-05,
      "loss": 0.3868,
      "step": 13337
    },
    {
      "epoch": 0.7468921491768395,
      "grad_norm": 1.4642640352249146,
      "learning_rate": 6.665000000000001e-05,
      "loss": 0.3162,
      "step": 13338
    },
    {
      "epoch": 0.7469481464889686,
      "grad_norm": 1.3704559803009033,
      "learning_rate": 6.6655e-05,
      "loss": 0.5632,
      "step": 13339
    },
    {
      "epoch": 0.7470041438010976,
      "grad_norm": 1.122069239616394,
      "learning_rate": 6.666e-05,
      "loss": 0.5401,
      "step": 13340
    },
    {
      "epoch": 0.7470601411132266,
      "grad_norm": 1.4768750667572021,
      "learning_rate": 6.6665e-05,
      "loss": 0.782,
      "step": 13341
    },
    {
      "epoch": 0.7471161384253556,
      "grad_norm": 1.275857925415039,
      "learning_rate": 6.667e-05,
      "loss": 0.4492,
      "step": 13342
    },
    {
      "epoch": 0.7471721357374846,
      "grad_norm": 1.7961218357086182,
      "learning_rate": 6.6675e-05,
      "loss": 0.4196,
      "step": 13343
    },
    {
      "epoch": 0.7472281330496137,
      "grad_norm": 1.3244459629058838,
      "learning_rate": 6.667999999999999e-05,
      "loss": 0.3435,
      "step": 13344
    },
    {
      "epoch": 0.7472841303617427,
      "grad_norm": 1.2896100282669067,
      "learning_rate": 6.6685e-05,
      "loss": 0.4248,
      "step": 13345
    },
    {
      "epoch": 0.7473401276738717,
      "grad_norm": 1.4216582775115967,
      "learning_rate": 6.669000000000001e-05,
      "loss": 0.4556,
      "step": 13346
    },
    {
      "epoch": 0.7473961249860007,
      "grad_norm": 1.4091801643371582,
      "learning_rate": 6.669500000000001e-05,
      "loss": 0.6151,
      "step": 13347
    },
    {
      "epoch": 0.7474521222981297,
      "grad_norm": 1.1799017190933228,
      "learning_rate": 6.670000000000001e-05,
      "loss": 0.4962,
      "step": 13348
    },
    {
      "epoch": 0.7475081196102588,
      "grad_norm": 1.501054286956787,
      "learning_rate": 6.6705e-05,
      "loss": 0.6032,
      "step": 13349
    },
    {
      "epoch": 0.7475641169223878,
      "grad_norm": 1.2588993310928345,
      "learning_rate": 6.671e-05,
      "loss": 0.3253,
      "step": 13350
    },
    {
      "epoch": 0.7476201142345168,
      "grad_norm": 1.1493582725524902,
      "learning_rate": 6.6715e-05,
      "loss": 0.3031,
      "step": 13351
    },
    {
      "epoch": 0.7476761115466458,
      "grad_norm": 1.250685691833496,
      "learning_rate": 6.672e-05,
      "loss": 0.5526,
      "step": 13352
    },
    {
      "epoch": 0.7477321088587747,
      "grad_norm": 1.6470725536346436,
      "learning_rate": 6.672500000000001e-05,
      "loss": 0.6043,
      "step": 13353
    },
    {
      "epoch": 0.7477881061709037,
      "grad_norm": 1.3246068954467773,
      "learning_rate": 6.673e-05,
      "loss": 0.3791,
      "step": 13354
    },
    {
      "epoch": 0.7478441034830328,
      "grad_norm": 1.1623932123184204,
      "learning_rate": 6.6735e-05,
      "loss": 0.4429,
      "step": 13355
    },
    {
      "epoch": 0.7479001007951618,
      "grad_norm": 1.8283212184906006,
      "learning_rate": 6.674e-05,
      "loss": 0.7647,
      "step": 13356
    },
    {
      "epoch": 0.7479560981072908,
      "grad_norm": 1.244064211845398,
      "learning_rate": 6.6745e-05,
      "loss": 0.5447,
      "step": 13357
    },
    {
      "epoch": 0.7480120954194198,
      "grad_norm": 1.3712365627288818,
      "learning_rate": 6.675e-05,
      "loss": 0.5122,
      "step": 13358
    },
    {
      "epoch": 0.7480680927315488,
      "grad_norm": 1.1770358085632324,
      "learning_rate": 6.675500000000001e-05,
      "loss": 0.3591,
      "step": 13359
    },
    {
      "epoch": 0.7481240900436779,
      "grad_norm": 1.14165198802948,
      "learning_rate": 6.676e-05,
      "loss": 0.3627,
      "step": 13360
    },
    {
      "epoch": 0.7481800873558069,
      "grad_norm": 1.3313946723937988,
      "learning_rate": 6.6765e-05,
      "loss": 0.4086,
      "step": 13361
    },
    {
      "epoch": 0.7482360846679359,
      "grad_norm": 1.3255826234817505,
      "learning_rate": 6.677e-05,
      "loss": 0.4672,
      "step": 13362
    },
    {
      "epoch": 0.7482920819800649,
      "grad_norm": 1.2458029985427856,
      "learning_rate": 6.6775e-05,
      "loss": 0.4414,
      "step": 13363
    },
    {
      "epoch": 0.7483480792921939,
      "grad_norm": 1.2485015392303467,
      "learning_rate": 6.678e-05,
      "loss": 0.4208,
      "step": 13364
    },
    {
      "epoch": 0.748404076604323,
      "grad_norm": 1.3235641717910767,
      "learning_rate": 6.6785e-05,
      "loss": 0.5394,
      "step": 13365
    },
    {
      "epoch": 0.748460073916452,
      "grad_norm": 1.4212559461593628,
      "learning_rate": 6.679e-05,
      "loss": 0.4528,
      "step": 13366
    },
    {
      "epoch": 0.748516071228581,
      "grad_norm": 1.4086921215057373,
      "learning_rate": 6.679500000000001e-05,
      "loss": 0.524,
      "step": 13367
    },
    {
      "epoch": 0.74857206854071,
      "grad_norm": 0.9677045941352844,
      "learning_rate": 6.680000000000001e-05,
      "loss": 0.3381,
      "step": 13368
    },
    {
      "epoch": 0.748628065852839,
      "grad_norm": 1.6913973093032837,
      "learning_rate": 6.680500000000001e-05,
      "loss": 0.4431,
      "step": 13369
    },
    {
      "epoch": 0.748684063164968,
      "grad_norm": 1.1492834091186523,
      "learning_rate": 6.681e-05,
      "loss": 0.3583,
      "step": 13370
    },
    {
      "epoch": 0.7487400604770971,
      "grad_norm": 1.4888323545455933,
      "learning_rate": 6.6815e-05,
      "loss": 0.5411,
      "step": 13371
    },
    {
      "epoch": 0.7487960577892261,
      "grad_norm": 1.3886725902557373,
      "learning_rate": 6.682e-05,
      "loss": 0.6303,
      "step": 13372
    },
    {
      "epoch": 0.7488520551013551,
      "grad_norm": 1.3541219234466553,
      "learning_rate": 6.6825e-05,
      "loss": 0.348,
      "step": 13373
    },
    {
      "epoch": 0.7489080524134841,
      "grad_norm": 1.4529330730438232,
      "learning_rate": 6.683000000000001e-05,
      "loss": 0.4437,
      "step": 13374
    },
    {
      "epoch": 0.7489640497256131,
      "grad_norm": 2.124105930328369,
      "learning_rate": 6.6835e-05,
      "loss": 0.5812,
      "step": 13375
    },
    {
      "epoch": 0.7490200470377422,
      "grad_norm": 1.2463494539260864,
      "learning_rate": 6.684e-05,
      "loss": 0.4733,
      "step": 13376
    },
    {
      "epoch": 0.7490760443498712,
      "grad_norm": 1.0593266487121582,
      "learning_rate": 6.6845e-05,
      "loss": 0.3661,
      "step": 13377
    },
    {
      "epoch": 0.7491320416620002,
      "grad_norm": 1.3837532997131348,
      "learning_rate": 6.685e-05,
      "loss": 0.4814,
      "step": 13378
    },
    {
      "epoch": 0.7491880389741292,
      "grad_norm": 2.01131010055542,
      "learning_rate": 6.6855e-05,
      "loss": 0.4869,
      "step": 13379
    },
    {
      "epoch": 0.7492440362862582,
      "grad_norm": 1.162909746170044,
      "learning_rate": 6.686000000000001e-05,
      "loss": 0.4443,
      "step": 13380
    },
    {
      "epoch": 0.7493000335983873,
      "grad_norm": 1.2567288875579834,
      "learning_rate": 6.6865e-05,
      "loss": 0.4941,
      "step": 13381
    },
    {
      "epoch": 0.7493560309105163,
      "grad_norm": 1.3693965673446655,
      "learning_rate": 6.687e-05,
      "loss": 0.5171,
      "step": 13382
    },
    {
      "epoch": 0.7494120282226453,
      "grad_norm": 1.385277509689331,
      "learning_rate": 6.6875e-05,
      "loss": 0.36,
      "step": 13383
    },
    {
      "epoch": 0.7494680255347743,
      "grad_norm": 1.4657427072525024,
      "learning_rate": 6.688e-05,
      "loss": 0.3572,
      "step": 13384
    },
    {
      "epoch": 0.7495240228469033,
      "grad_norm": 1.1794122457504272,
      "learning_rate": 6.6885e-05,
      "loss": 0.4101,
      "step": 13385
    },
    {
      "epoch": 0.7495800201590324,
      "grad_norm": 1.3354967832565308,
      "learning_rate": 6.689e-05,
      "loss": 0.4576,
      "step": 13386
    },
    {
      "epoch": 0.7496360174711614,
      "grad_norm": 1.1651242971420288,
      "learning_rate": 6.6895e-05,
      "loss": 0.3619,
      "step": 13387
    },
    {
      "epoch": 0.7496920147832904,
      "grad_norm": 1.4015265703201294,
      "learning_rate": 6.690000000000001e-05,
      "loss": 0.4246,
      "step": 13388
    },
    {
      "epoch": 0.7497480120954194,
      "grad_norm": 1.1027058362960815,
      "learning_rate": 6.690500000000001e-05,
      "loss": 0.2979,
      "step": 13389
    },
    {
      "epoch": 0.7498040094075484,
      "grad_norm": 1.2594717741012573,
      "learning_rate": 6.691000000000001e-05,
      "loss": 0.4395,
      "step": 13390
    },
    {
      "epoch": 0.7498600067196775,
      "grad_norm": 1.2497899532318115,
      "learning_rate": 6.6915e-05,
      "loss": 0.4734,
      "step": 13391
    },
    {
      "epoch": 0.7499160040318065,
      "grad_norm": 1.5366617441177368,
      "learning_rate": 6.692e-05,
      "loss": 0.4336,
      "step": 13392
    },
    {
      "epoch": 0.7499720013439355,
      "grad_norm": 1.5138517618179321,
      "learning_rate": 6.6925e-05,
      "loss": 0.6008,
      "step": 13393
    },
    {
      "epoch": 0.7500279986560645,
      "grad_norm": 1.1237903833389282,
      "learning_rate": 6.693e-05,
      "loss": 0.4135,
      "step": 13394
    },
    {
      "epoch": 0.7500839959681935,
      "grad_norm": 1.1958673000335693,
      "learning_rate": 6.693500000000001e-05,
      "loss": 0.3744,
      "step": 13395
    },
    {
      "epoch": 0.7501399932803225,
      "grad_norm": 1.316476821899414,
      "learning_rate": 6.694e-05,
      "loss": 0.4043,
      "step": 13396
    },
    {
      "epoch": 0.7501959905924516,
      "grad_norm": 1.3259350061416626,
      "learning_rate": 6.6945e-05,
      "loss": 0.4395,
      "step": 13397
    },
    {
      "epoch": 0.7502519879045806,
      "grad_norm": 1.565314769744873,
      "learning_rate": 6.695e-05,
      "loss": 0.4311,
      "step": 13398
    },
    {
      "epoch": 0.7503079852167096,
      "grad_norm": 1.2138341665267944,
      "learning_rate": 6.6955e-05,
      "loss": 0.3549,
      "step": 13399
    },
    {
      "epoch": 0.7503639825288386,
      "grad_norm": 1.5309661626815796,
      "learning_rate": 6.696e-05,
      "loss": 0.6146,
      "step": 13400
    },
    {
      "epoch": 0.7504199798409676,
      "grad_norm": 1.2558754682540894,
      "learning_rate": 6.6965e-05,
      "loss": 0.4875,
      "step": 13401
    },
    {
      "epoch": 0.7504759771530967,
      "grad_norm": 1.4507876634597778,
      "learning_rate": 6.697e-05,
      "loss": 0.6295,
      "step": 13402
    },
    {
      "epoch": 0.7505319744652257,
      "grad_norm": 1.652795433998108,
      "learning_rate": 6.6975e-05,
      "loss": 0.3718,
      "step": 13403
    },
    {
      "epoch": 0.7505879717773547,
      "grad_norm": 1.3381702899932861,
      "learning_rate": 6.698e-05,
      "loss": 0.4106,
      "step": 13404
    },
    {
      "epoch": 0.7506439690894837,
      "grad_norm": 1.7910661697387695,
      "learning_rate": 6.6985e-05,
      "loss": 0.4129,
      "step": 13405
    },
    {
      "epoch": 0.7506999664016127,
      "grad_norm": 1.3057547807693481,
      "learning_rate": 6.699000000000001e-05,
      "loss": 0.3802,
      "step": 13406
    },
    {
      "epoch": 0.7507559637137418,
      "grad_norm": 1.3384629487991333,
      "learning_rate": 6.6995e-05,
      "loss": 0.5628,
      "step": 13407
    },
    {
      "epoch": 0.7508119610258708,
      "grad_norm": 1.34722101688385,
      "learning_rate": 6.7e-05,
      "loss": 0.4655,
      "step": 13408
    },
    {
      "epoch": 0.7508679583379998,
      "grad_norm": 1.3082317113876343,
      "learning_rate": 6.700500000000001e-05,
      "loss": 0.4847,
      "step": 13409
    },
    {
      "epoch": 0.7509239556501288,
      "grad_norm": 1.2084414958953857,
      "learning_rate": 6.701000000000001e-05,
      "loss": 0.5166,
      "step": 13410
    },
    {
      "epoch": 0.7509799529622578,
      "grad_norm": 1.056401252746582,
      "learning_rate": 6.701500000000001e-05,
      "loss": 0.4287,
      "step": 13411
    },
    {
      "epoch": 0.7510359502743869,
      "grad_norm": 1.2451685667037964,
      "learning_rate": 6.702e-05,
      "loss": 0.3245,
      "step": 13412
    },
    {
      "epoch": 0.7510919475865159,
      "grad_norm": 1.2550543546676636,
      "learning_rate": 6.7025e-05,
      "loss": 0.415,
      "step": 13413
    },
    {
      "epoch": 0.7511479448986449,
      "grad_norm": 1.3785005807876587,
      "learning_rate": 6.703e-05,
      "loss": 0.352,
      "step": 13414
    },
    {
      "epoch": 0.7512039422107739,
      "grad_norm": 1.3353605270385742,
      "learning_rate": 6.7035e-05,
      "loss": 0.445,
      "step": 13415
    },
    {
      "epoch": 0.7512599395229029,
      "grad_norm": 1.0877524614334106,
      "learning_rate": 6.704000000000001e-05,
      "loss": 0.3922,
      "step": 13416
    },
    {
      "epoch": 0.751315936835032,
      "grad_norm": 1.2942429780960083,
      "learning_rate": 6.7045e-05,
      "loss": 0.3311,
      "step": 13417
    },
    {
      "epoch": 0.751371934147161,
      "grad_norm": 1.3884506225585938,
      "learning_rate": 6.705e-05,
      "loss": 0.4471,
      "step": 13418
    },
    {
      "epoch": 0.75142793145929,
      "grad_norm": 1.247389554977417,
      "learning_rate": 6.7055e-05,
      "loss": 0.3752,
      "step": 13419
    },
    {
      "epoch": 0.751483928771419,
      "grad_norm": 1.2123414278030396,
      "learning_rate": 6.706e-05,
      "loss": 0.4752,
      "step": 13420
    },
    {
      "epoch": 0.751539926083548,
      "grad_norm": 1.7403959035873413,
      "learning_rate": 6.7065e-05,
      "loss": 0.5192,
      "step": 13421
    },
    {
      "epoch": 0.751595923395677,
      "grad_norm": 1.4864706993103027,
      "learning_rate": 6.706999999999999e-05,
      "loss": 0.3862,
      "step": 13422
    },
    {
      "epoch": 0.7516519207078061,
      "grad_norm": 1.4361765384674072,
      "learning_rate": 6.7075e-05,
      "loss": 0.4101,
      "step": 13423
    },
    {
      "epoch": 0.7517079180199351,
      "grad_norm": 1.5567173957824707,
      "learning_rate": 6.708e-05,
      "loss": 0.5029,
      "step": 13424
    },
    {
      "epoch": 0.7517639153320641,
      "grad_norm": 1.390423059463501,
      "learning_rate": 6.7085e-05,
      "loss": 0.3613,
      "step": 13425
    },
    {
      "epoch": 0.7518199126441931,
      "grad_norm": 1.4552497863769531,
      "learning_rate": 6.709000000000001e-05,
      "loss": 0.7664,
      "step": 13426
    },
    {
      "epoch": 0.7518759099563221,
      "grad_norm": 1.305458426475525,
      "learning_rate": 6.709500000000001e-05,
      "loss": 0.5114,
      "step": 13427
    },
    {
      "epoch": 0.7519319072684512,
      "grad_norm": 1.2647517919540405,
      "learning_rate": 6.71e-05,
      "loss": 0.3717,
      "step": 13428
    },
    {
      "epoch": 0.7519879045805802,
      "grad_norm": 1.1407110691070557,
      "learning_rate": 6.7105e-05,
      "loss": 0.3953,
      "step": 13429
    },
    {
      "epoch": 0.7520439018927092,
      "grad_norm": 1.303435206413269,
      "learning_rate": 6.711e-05,
      "loss": 0.4151,
      "step": 13430
    },
    {
      "epoch": 0.7520998992048382,
      "grad_norm": 1.5761874914169312,
      "learning_rate": 6.711500000000001e-05,
      "loss": 0.3881,
      "step": 13431
    },
    {
      "epoch": 0.7521558965169672,
      "grad_norm": 1.368018388748169,
      "learning_rate": 6.712000000000001e-05,
      "loss": 0.3724,
      "step": 13432
    },
    {
      "epoch": 0.7522118938290963,
      "grad_norm": 1.412675142288208,
      "learning_rate": 6.7125e-05,
      "loss": 0.5171,
      "step": 13433
    },
    {
      "epoch": 0.7522678911412253,
      "grad_norm": 1.142570972442627,
      "learning_rate": 6.713e-05,
      "loss": 0.3583,
      "step": 13434
    },
    {
      "epoch": 0.7523238884533542,
      "grad_norm": 1.1345243453979492,
      "learning_rate": 6.7135e-05,
      "loss": 0.3614,
      "step": 13435
    },
    {
      "epoch": 0.7523798857654832,
      "grad_norm": 1.4500483274459839,
      "learning_rate": 6.714e-05,
      "loss": 0.4051,
      "step": 13436
    },
    {
      "epoch": 0.7524358830776122,
      "grad_norm": 1.1398577690124512,
      "learning_rate": 6.714500000000001e-05,
      "loss": 0.3196,
      "step": 13437
    },
    {
      "epoch": 0.7524918803897412,
      "grad_norm": 1.4100042581558228,
      "learning_rate": 6.715e-05,
      "loss": 0.4972,
      "step": 13438
    },
    {
      "epoch": 0.7525478777018703,
      "grad_norm": 1.1147397756576538,
      "learning_rate": 6.7155e-05,
      "loss": 0.4089,
      "step": 13439
    },
    {
      "epoch": 0.7526038750139993,
      "grad_norm": 1.1418505907058716,
      "learning_rate": 6.716e-05,
      "loss": 0.3055,
      "step": 13440
    },
    {
      "epoch": 0.7526598723261283,
      "grad_norm": 1.6036012172698975,
      "learning_rate": 6.7165e-05,
      "loss": 0.476,
      "step": 13441
    },
    {
      "epoch": 0.7527158696382573,
      "grad_norm": 1.084713101387024,
      "learning_rate": 6.717e-05,
      "loss": 0.3097,
      "step": 13442
    },
    {
      "epoch": 0.7527718669503863,
      "grad_norm": 1.3821686506271362,
      "learning_rate": 6.717499999999999e-05,
      "loss": 0.5434,
      "step": 13443
    },
    {
      "epoch": 0.7528278642625154,
      "grad_norm": 1.393151879310608,
      "learning_rate": 6.718e-05,
      "loss": 0.4664,
      "step": 13444
    },
    {
      "epoch": 0.7528838615746444,
      "grad_norm": 1.1680833101272583,
      "learning_rate": 6.7185e-05,
      "loss": 0.4461,
      "step": 13445
    },
    {
      "epoch": 0.7529398588867734,
      "grad_norm": 1.2255024909973145,
      "learning_rate": 6.719000000000001e-05,
      "loss": 0.4013,
      "step": 13446
    },
    {
      "epoch": 0.7529958561989024,
      "grad_norm": 1.1132417917251587,
      "learning_rate": 6.719500000000001e-05,
      "loss": 0.3616,
      "step": 13447
    },
    {
      "epoch": 0.7530518535110314,
      "grad_norm": 1.598866581916809,
      "learning_rate": 6.720000000000001e-05,
      "loss": 0.4688,
      "step": 13448
    },
    {
      "epoch": 0.7531078508231605,
      "grad_norm": 1.1186206340789795,
      "learning_rate": 6.7205e-05,
      "loss": 0.3437,
      "step": 13449
    },
    {
      "epoch": 0.7531638481352895,
      "grad_norm": 1.6304622888565063,
      "learning_rate": 6.721e-05,
      "loss": 0.6735,
      "step": 13450
    },
    {
      "epoch": 0.7532198454474185,
      "grad_norm": 1.4465316534042358,
      "learning_rate": 6.7215e-05,
      "loss": 0.4097,
      "step": 13451
    },
    {
      "epoch": 0.7532758427595475,
      "grad_norm": 1.664799451828003,
      "learning_rate": 6.722000000000001e-05,
      "loss": 0.4707,
      "step": 13452
    },
    {
      "epoch": 0.7533318400716765,
      "grad_norm": 1.2107914686203003,
      "learning_rate": 6.722500000000001e-05,
      "loss": 0.3472,
      "step": 13453
    },
    {
      "epoch": 0.7533878373838055,
      "grad_norm": 1.3441057205200195,
      "learning_rate": 6.723e-05,
      "loss": 0.4359,
      "step": 13454
    },
    {
      "epoch": 0.7534438346959346,
      "grad_norm": 1.2599084377288818,
      "learning_rate": 6.7235e-05,
      "loss": 0.4591,
      "step": 13455
    },
    {
      "epoch": 0.7534998320080636,
      "grad_norm": 1.185228705406189,
      "learning_rate": 6.724e-05,
      "loss": 0.3497,
      "step": 13456
    },
    {
      "epoch": 0.7535558293201926,
      "grad_norm": 1.2232575416564941,
      "learning_rate": 6.7245e-05,
      "loss": 0.4625,
      "step": 13457
    },
    {
      "epoch": 0.7536118266323216,
      "grad_norm": 1.4489551782608032,
      "learning_rate": 6.725000000000001e-05,
      "loss": 0.5915,
      "step": 13458
    },
    {
      "epoch": 0.7536678239444506,
      "grad_norm": 1.467835783958435,
      "learning_rate": 6.7255e-05,
      "loss": 0.6399,
      "step": 13459
    },
    {
      "epoch": 0.7537238212565797,
      "grad_norm": 1.36167311668396,
      "learning_rate": 6.726e-05,
      "loss": 0.4153,
      "step": 13460
    },
    {
      "epoch": 0.7537798185687087,
      "grad_norm": 1.2901496887207031,
      "learning_rate": 6.7265e-05,
      "loss": 0.3864,
      "step": 13461
    },
    {
      "epoch": 0.7538358158808377,
      "grad_norm": 1.549117922782898,
      "learning_rate": 6.727e-05,
      "loss": 0.4436,
      "step": 13462
    },
    {
      "epoch": 0.7538918131929667,
      "grad_norm": 1.1801596879959106,
      "learning_rate": 6.7275e-05,
      "loss": 0.3789,
      "step": 13463
    },
    {
      "epoch": 0.7539478105050957,
      "grad_norm": 1.5706390142440796,
      "learning_rate": 6.727999999999999e-05,
      "loss": 0.3809,
      "step": 13464
    },
    {
      "epoch": 0.7540038078172248,
      "grad_norm": 1.2169467210769653,
      "learning_rate": 6.7285e-05,
      "loss": 0.4053,
      "step": 13465
    },
    {
      "epoch": 0.7540598051293538,
      "grad_norm": 1.3880095481872559,
      "learning_rate": 6.729000000000001e-05,
      "loss": 0.5238,
      "step": 13466
    },
    {
      "epoch": 0.7541158024414828,
      "grad_norm": 1.2271162271499634,
      "learning_rate": 6.729500000000001e-05,
      "loss": 0.386,
      "step": 13467
    },
    {
      "epoch": 0.7541717997536118,
      "grad_norm": 1.463864803314209,
      "learning_rate": 6.730000000000001e-05,
      "loss": 0.4024,
      "step": 13468
    },
    {
      "epoch": 0.7542277970657408,
      "grad_norm": 1.4349846839904785,
      "learning_rate": 6.730500000000001e-05,
      "loss": 0.5111,
      "step": 13469
    },
    {
      "epoch": 0.7542837943778699,
      "grad_norm": 1.2246228456497192,
      "learning_rate": 6.731e-05,
      "loss": 0.3641,
      "step": 13470
    },
    {
      "epoch": 0.7543397916899989,
      "grad_norm": 1.2935744524002075,
      "learning_rate": 6.7315e-05,
      "loss": 0.5459,
      "step": 13471
    },
    {
      "epoch": 0.7543957890021279,
      "grad_norm": 1.280429482460022,
      "learning_rate": 6.732e-05,
      "loss": 0.4411,
      "step": 13472
    },
    {
      "epoch": 0.7544517863142569,
      "grad_norm": 1.3651620149612427,
      "learning_rate": 6.732500000000001e-05,
      "loss": 0.4289,
      "step": 13473
    },
    {
      "epoch": 0.7545077836263859,
      "grad_norm": 1.0948057174682617,
      "learning_rate": 6.733000000000001e-05,
      "loss": 0.3642,
      "step": 13474
    },
    {
      "epoch": 0.754563780938515,
      "grad_norm": 1.3790465593338013,
      "learning_rate": 6.7335e-05,
      "loss": 0.4769,
      "step": 13475
    },
    {
      "epoch": 0.754619778250644,
      "grad_norm": 1.553052544593811,
      "learning_rate": 6.734e-05,
      "loss": 0.5529,
      "step": 13476
    },
    {
      "epoch": 0.754675775562773,
      "grad_norm": 1.7630292177200317,
      "learning_rate": 6.7345e-05,
      "loss": 0.6346,
      "step": 13477
    },
    {
      "epoch": 0.754731772874902,
      "grad_norm": 1.2925057411193848,
      "learning_rate": 6.735e-05,
      "loss": 0.4822,
      "step": 13478
    },
    {
      "epoch": 0.754787770187031,
      "grad_norm": 1.9060975313186646,
      "learning_rate": 6.735500000000001e-05,
      "loss": 0.4757,
      "step": 13479
    },
    {
      "epoch": 0.75484376749916,
      "grad_norm": 1.1911603212356567,
      "learning_rate": 6.736e-05,
      "loss": 0.4558,
      "step": 13480
    },
    {
      "epoch": 0.7548997648112891,
      "grad_norm": 1.3044967651367188,
      "learning_rate": 6.7365e-05,
      "loss": 0.4002,
      "step": 13481
    },
    {
      "epoch": 0.7549557621234181,
      "grad_norm": 1.0621124505996704,
      "learning_rate": 6.737e-05,
      "loss": 0.3953,
      "step": 13482
    },
    {
      "epoch": 0.7550117594355471,
      "grad_norm": 1.3665282726287842,
      "learning_rate": 6.7375e-05,
      "loss": 0.4164,
      "step": 13483
    },
    {
      "epoch": 0.7550677567476761,
      "grad_norm": 1.1974210739135742,
      "learning_rate": 6.738e-05,
      "loss": 0.4412,
      "step": 13484
    },
    {
      "epoch": 0.7551237540598051,
      "grad_norm": 1.2153085470199585,
      "learning_rate": 6.738499999999999e-05,
      "loss": 0.3828,
      "step": 13485
    },
    {
      "epoch": 0.7551797513719342,
      "grad_norm": 1.622483730316162,
      "learning_rate": 6.739e-05,
      "loss": 0.4713,
      "step": 13486
    },
    {
      "epoch": 0.7552357486840632,
      "grad_norm": 1.2181439399719238,
      "learning_rate": 6.739500000000001e-05,
      "loss": 0.4743,
      "step": 13487
    },
    {
      "epoch": 0.7552917459961922,
      "grad_norm": 1.6612908840179443,
      "learning_rate": 6.740000000000001e-05,
      "loss": 0.3973,
      "step": 13488
    },
    {
      "epoch": 0.7553477433083212,
      "grad_norm": 1.333847165107727,
      "learning_rate": 6.740500000000001e-05,
      "loss": 0.3932,
      "step": 13489
    },
    {
      "epoch": 0.7554037406204502,
      "grad_norm": 1.3070459365844727,
      "learning_rate": 6.741000000000001e-05,
      "loss": 0.431,
      "step": 13490
    },
    {
      "epoch": 0.7554597379325793,
      "grad_norm": 1.1635133028030396,
      "learning_rate": 6.7415e-05,
      "loss": 0.3468,
      "step": 13491
    },
    {
      "epoch": 0.7555157352447083,
      "grad_norm": 1.4309470653533936,
      "learning_rate": 6.742e-05,
      "loss": 0.4128,
      "step": 13492
    },
    {
      "epoch": 0.7555717325568373,
      "grad_norm": 1.3312102556228638,
      "learning_rate": 6.7425e-05,
      "loss": 0.4404,
      "step": 13493
    },
    {
      "epoch": 0.7556277298689663,
      "grad_norm": 1.4434672594070435,
      "learning_rate": 6.743000000000001e-05,
      "loss": 0.5623,
      "step": 13494
    },
    {
      "epoch": 0.7556837271810953,
      "grad_norm": 1.1383618116378784,
      "learning_rate": 6.743500000000001e-05,
      "loss": 0.3618,
      "step": 13495
    },
    {
      "epoch": 0.7557397244932244,
      "grad_norm": 1.1406939029693604,
      "learning_rate": 6.744e-05,
      "loss": 0.3949,
      "step": 13496
    },
    {
      "epoch": 0.7557957218053534,
      "grad_norm": 1.3334400653839111,
      "learning_rate": 6.7445e-05,
      "loss": 0.4681,
      "step": 13497
    },
    {
      "epoch": 0.7558517191174824,
      "grad_norm": 1.2865227460861206,
      "learning_rate": 6.745e-05,
      "loss": 0.4014,
      "step": 13498
    },
    {
      "epoch": 0.7559077164296114,
      "grad_norm": 1.1794874668121338,
      "learning_rate": 6.7455e-05,
      "loss": 0.4435,
      "step": 13499
    },
    {
      "epoch": 0.7559637137417404,
      "grad_norm": 1.5550849437713623,
      "learning_rate": 6.746e-05,
      "loss": 0.6375,
      "step": 13500
    },
    {
      "epoch": 0.7560197110538694,
      "grad_norm": 1.6652644872665405,
      "learning_rate": 6.7465e-05,
      "loss": 0.5682,
      "step": 13501
    },
    {
      "epoch": 0.7560757083659985,
      "grad_norm": 1.117836594581604,
      "learning_rate": 6.747e-05,
      "loss": 0.3993,
      "step": 13502
    },
    {
      "epoch": 0.7561317056781275,
      "grad_norm": 1.3415014743804932,
      "learning_rate": 6.7475e-05,
      "loss": 0.4238,
      "step": 13503
    },
    {
      "epoch": 0.7561877029902565,
      "grad_norm": 1.3984713554382324,
      "learning_rate": 6.748e-05,
      "loss": 0.4271,
      "step": 13504
    },
    {
      "epoch": 0.7562437003023855,
      "grad_norm": 1.6960017681121826,
      "learning_rate": 6.7485e-05,
      "loss": 0.3231,
      "step": 13505
    },
    {
      "epoch": 0.7562996976145145,
      "grad_norm": 1.6180530786514282,
      "learning_rate": 6.749e-05,
      "loss": 0.477,
      "step": 13506
    },
    {
      "epoch": 0.7563556949266436,
      "grad_norm": 1.4171092510223389,
      "learning_rate": 6.7495e-05,
      "loss": 0.4831,
      "step": 13507
    },
    {
      "epoch": 0.7564116922387726,
      "grad_norm": 1.1733931303024292,
      "learning_rate": 6.750000000000001e-05,
      "loss": 0.4208,
      "step": 13508
    },
    {
      "epoch": 0.7564676895509016,
      "grad_norm": 1.4384032487869263,
      "learning_rate": 6.750500000000001e-05,
      "loss": 0.5576,
      "step": 13509
    },
    {
      "epoch": 0.7565236868630306,
      "grad_norm": 1.3238472938537598,
      "learning_rate": 6.751000000000001e-05,
      "loss": 0.4113,
      "step": 13510
    },
    {
      "epoch": 0.7565796841751596,
      "grad_norm": 1.176927924156189,
      "learning_rate": 6.7515e-05,
      "loss": 0.3882,
      "step": 13511
    },
    {
      "epoch": 0.7566356814872887,
      "grad_norm": 1.1933873891830444,
      "learning_rate": 6.752e-05,
      "loss": 0.496,
      "step": 13512
    },
    {
      "epoch": 0.7566916787994177,
      "grad_norm": 1.1905460357666016,
      "learning_rate": 6.7525e-05,
      "loss": 0.3585,
      "step": 13513
    },
    {
      "epoch": 0.7567476761115467,
      "grad_norm": 1.154370665550232,
      "learning_rate": 6.753e-05,
      "loss": 0.3804,
      "step": 13514
    },
    {
      "epoch": 0.7568036734236757,
      "grad_norm": 1.5908572673797607,
      "learning_rate": 6.753500000000001e-05,
      "loss": 0.5003,
      "step": 13515
    },
    {
      "epoch": 0.7568596707358047,
      "grad_norm": 2.966088056564331,
      "learning_rate": 6.754000000000001e-05,
      "loss": 0.4471,
      "step": 13516
    },
    {
      "epoch": 0.7569156680479338,
      "grad_norm": 1.2605220079421997,
      "learning_rate": 6.7545e-05,
      "loss": 0.3171,
      "step": 13517
    },
    {
      "epoch": 0.7569716653600627,
      "grad_norm": 1.172694444656372,
      "learning_rate": 6.755e-05,
      "loss": 0.4792,
      "step": 13518
    },
    {
      "epoch": 0.7570276626721917,
      "grad_norm": 1.4007148742675781,
      "learning_rate": 6.7555e-05,
      "loss": 0.5274,
      "step": 13519
    },
    {
      "epoch": 0.7570836599843207,
      "grad_norm": 1.4705430269241333,
      "learning_rate": 6.756e-05,
      "loss": 0.5258,
      "step": 13520
    },
    {
      "epoch": 0.7571396572964497,
      "grad_norm": 1.2992222309112549,
      "learning_rate": 6.7565e-05,
      "loss": 0.4533,
      "step": 13521
    },
    {
      "epoch": 0.7571956546085787,
      "grad_norm": 1.5344949960708618,
      "learning_rate": 6.757e-05,
      "loss": 0.5269,
      "step": 13522
    },
    {
      "epoch": 0.7572516519207078,
      "grad_norm": 1.204139232635498,
      "learning_rate": 6.7575e-05,
      "loss": 0.5333,
      "step": 13523
    },
    {
      "epoch": 0.7573076492328368,
      "grad_norm": 1.3827623128890991,
      "learning_rate": 6.758e-05,
      "loss": 0.548,
      "step": 13524
    },
    {
      "epoch": 0.7573636465449658,
      "grad_norm": 1.5382299423217773,
      "learning_rate": 6.7585e-05,
      "loss": 0.5404,
      "step": 13525
    },
    {
      "epoch": 0.7574196438570948,
      "grad_norm": 1.6641407012939453,
      "learning_rate": 6.759e-05,
      "loss": 0.6407,
      "step": 13526
    },
    {
      "epoch": 0.7574756411692238,
      "grad_norm": 1.459222674369812,
      "learning_rate": 6.7595e-05,
      "loss": 0.4695,
      "step": 13527
    },
    {
      "epoch": 0.7575316384813529,
      "grad_norm": 1.2386016845703125,
      "learning_rate": 6.76e-05,
      "loss": 0.5066,
      "step": 13528
    },
    {
      "epoch": 0.7575876357934819,
      "grad_norm": 1.3396717309951782,
      "learning_rate": 6.7605e-05,
      "loss": 0.3974,
      "step": 13529
    },
    {
      "epoch": 0.7576436331056109,
      "grad_norm": 3.966275691986084,
      "learning_rate": 6.761000000000001e-05,
      "loss": 0.4284,
      "step": 13530
    },
    {
      "epoch": 0.7576996304177399,
      "grad_norm": 1.2423146963119507,
      "learning_rate": 6.761500000000001e-05,
      "loss": 0.4615,
      "step": 13531
    },
    {
      "epoch": 0.7577556277298689,
      "grad_norm": 1.0985716581344604,
      "learning_rate": 6.762e-05,
      "loss": 0.3827,
      "step": 13532
    },
    {
      "epoch": 0.757811625041998,
      "grad_norm": 1.3467158079147339,
      "learning_rate": 6.7625e-05,
      "loss": 0.4836,
      "step": 13533
    },
    {
      "epoch": 0.757867622354127,
      "grad_norm": 1.2831975221633911,
      "learning_rate": 6.763e-05,
      "loss": 0.4462,
      "step": 13534
    },
    {
      "epoch": 0.757923619666256,
      "grad_norm": 1.3927476406097412,
      "learning_rate": 6.7635e-05,
      "loss": 0.4237,
      "step": 13535
    },
    {
      "epoch": 0.757979616978385,
      "grad_norm": 1.3788520097732544,
      "learning_rate": 6.764000000000001e-05,
      "loss": 0.4037,
      "step": 13536
    },
    {
      "epoch": 0.758035614290514,
      "grad_norm": 1.3232665061950684,
      "learning_rate": 6.764500000000001e-05,
      "loss": 0.5126,
      "step": 13537
    },
    {
      "epoch": 0.758091611602643,
      "grad_norm": 1.4086594581604004,
      "learning_rate": 6.765e-05,
      "loss": 0.4718,
      "step": 13538
    },
    {
      "epoch": 0.7581476089147721,
      "grad_norm": 1.207565188407898,
      "learning_rate": 6.7655e-05,
      "loss": 0.3503,
      "step": 13539
    },
    {
      "epoch": 0.7582036062269011,
      "grad_norm": 1.3686875104904175,
      "learning_rate": 6.766e-05,
      "loss": 0.3985,
      "step": 13540
    },
    {
      "epoch": 0.7582596035390301,
      "grad_norm": 1.418041706085205,
      "learning_rate": 6.7665e-05,
      "loss": 0.4116,
      "step": 13541
    },
    {
      "epoch": 0.7583156008511591,
      "grad_norm": 1.3729579448699951,
      "learning_rate": 6.767e-05,
      "loss": 0.5298,
      "step": 13542
    },
    {
      "epoch": 0.7583715981632881,
      "grad_norm": 1.4995180368423462,
      "learning_rate": 6.7675e-05,
      "loss": 0.5458,
      "step": 13543
    },
    {
      "epoch": 0.7584275954754172,
      "grad_norm": 1.378974199295044,
      "learning_rate": 6.768e-05,
      "loss": 0.3962,
      "step": 13544
    },
    {
      "epoch": 0.7584835927875462,
      "grad_norm": 1.2189967632293701,
      "learning_rate": 6.7685e-05,
      "loss": 0.5196,
      "step": 13545
    },
    {
      "epoch": 0.7585395900996752,
      "grad_norm": 1.2721692323684692,
      "learning_rate": 6.769e-05,
      "loss": 0.38,
      "step": 13546
    },
    {
      "epoch": 0.7585955874118042,
      "grad_norm": 1.3236334323883057,
      "learning_rate": 6.769500000000001e-05,
      "loss": 0.4855,
      "step": 13547
    },
    {
      "epoch": 0.7586515847239332,
      "grad_norm": 1.087719202041626,
      "learning_rate": 6.77e-05,
      "loss": 0.3678,
      "step": 13548
    },
    {
      "epoch": 0.7587075820360623,
      "grad_norm": 1.3718489408493042,
      "learning_rate": 6.7705e-05,
      "loss": 0.4305,
      "step": 13549
    },
    {
      "epoch": 0.7587635793481913,
      "grad_norm": 1.2739298343658447,
      "learning_rate": 6.771e-05,
      "loss": 0.4258,
      "step": 13550
    },
    {
      "epoch": 0.7588195766603203,
      "grad_norm": 1.6008694171905518,
      "learning_rate": 6.771500000000001e-05,
      "loss": 0.5363,
      "step": 13551
    },
    {
      "epoch": 0.7588755739724493,
      "grad_norm": 1.3075045347213745,
      "learning_rate": 6.772000000000001e-05,
      "loss": 0.4086,
      "step": 13552
    },
    {
      "epoch": 0.7589315712845783,
      "grad_norm": 1.3609967231750488,
      "learning_rate": 6.7725e-05,
      "loss": 0.4109,
      "step": 13553
    },
    {
      "epoch": 0.7589875685967074,
      "grad_norm": 1.360772967338562,
      "learning_rate": 6.773e-05,
      "loss": 0.4327,
      "step": 13554
    },
    {
      "epoch": 0.7590435659088364,
      "grad_norm": 1.3756797313690186,
      "learning_rate": 6.7735e-05,
      "loss": 0.4837,
      "step": 13555
    },
    {
      "epoch": 0.7590995632209654,
      "grad_norm": 1.610328197479248,
      "learning_rate": 6.774e-05,
      "loss": 0.6,
      "step": 13556
    },
    {
      "epoch": 0.7591555605330944,
      "grad_norm": 1.6325621604919434,
      "learning_rate": 6.774500000000001e-05,
      "loss": 0.5773,
      "step": 13557
    },
    {
      "epoch": 0.7592115578452234,
      "grad_norm": 1.2873172760009766,
      "learning_rate": 6.775000000000001e-05,
      "loss": 0.4575,
      "step": 13558
    },
    {
      "epoch": 0.7592675551573524,
      "grad_norm": 1.4128810167312622,
      "learning_rate": 6.7755e-05,
      "loss": 0.3474,
      "step": 13559
    },
    {
      "epoch": 0.7593235524694815,
      "grad_norm": 1.3517918586730957,
      "learning_rate": 6.776e-05,
      "loss": 0.4743,
      "step": 13560
    },
    {
      "epoch": 0.7593795497816105,
      "grad_norm": 1.365537166595459,
      "learning_rate": 6.7765e-05,
      "loss": 0.5071,
      "step": 13561
    },
    {
      "epoch": 0.7594355470937395,
      "grad_norm": 1.3406610488891602,
      "learning_rate": 6.777e-05,
      "loss": 0.3936,
      "step": 13562
    },
    {
      "epoch": 0.7594915444058685,
      "grad_norm": 1.4251781702041626,
      "learning_rate": 6.7775e-05,
      "loss": 0.5897,
      "step": 13563
    },
    {
      "epoch": 0.7595475417179975,
      "grad_norm": 1.5039125680923462,
      "learning_rate": 6.778e-05,
      "loss": 0.5221,
      "step": 13564
    },
    {
      "epoch": 0.7596035390301266,
      "grad_norm": 1.3651732206344604,
      "learning_rate": 6.7785e-05,
      "loss": 0.4698,
      "step": 13565
    },
    {
      "epoch": 0.7596595363422556,
      "grad_norm": 1.6269944906234741,
      "learning_rate": 6.779e-05,
      "loss": 0.5133,
      "step": 13566
    },
    {
      "epoch": 0.7597155336543846,
      "grad_norm": 1.255621075630188,
      "learning_rate": 6.779500000000001e-05,
      "loss": 0.4111,
      "step": 13567
    },
    {
      "epoch": 0.7597715309665136,
      "grad_norm": 1.2345309257507324,
      "learning_rate": 6.780000000000001e-05,
      "loss": 0.3922,
      "step": 13568
    },
    {
      "epoch": 0.7598275282786426,
      "grad_norm": 1.3083750009536743,
      "learning_rate": 6.7805e-05,
      "loss": 0.421,
      "step": 13569
    },
    {
      "epoch": 0.7598835255907717,
      "grad_norm": 1.4425714015960693,
      "learning_rate": 6.781e-05,
      "loss": 0.7085,
      "step": 13570
    },
    {
      "epoch": 0.7599395229029007,
      "grad_norm": 1.5089733600616455,
      "learning_rate": 6.7815e-05,
      "loss": 0.4205,
      "step": 13571
    },
    {
      "epoch": 0.7599955202150297,
      "grad_norm": 1.9460489749908447,
      "learning_rate": 6.782000000000001e-05,
      "loss": 0.5807,
      "step": 13572
    },
    {
      "epoch": 0.7600515175271587,
      "grad_norm": 1.3572713136672974,
      "learning_rate": 6.782500000000001e-05,
      "loss": 0.551,
      "step": 13573
    },
    {
      "epoch": 0.7601075148392877,
      "grad_norm": 1.2764432430267334,
      "learning_rate": 6.783e-05,
      "loss": 0.5627,
      "step": 13574
    },
    {
      "epoch": 0.7601635121514168,
      "grad_norm": 1.253567099571228,
      "learning_rate": 6.7835e-05,
      "loss": 0.3933,
      "step": 13575
    },
    {
      "epoch": 0.7602195094635458,
      "grad_norm": 1.4845340251922607,
      "learning_rate": 6.784e-05,
      "loss": 0.4841,
      "step": 13576
    },
    {
      "epoch": 0.7602755067756748,
      "grad_norm": 1.1086807250976562,
      "learning_rate": 6.7845e-05,
      "loss": 0.3112,
      "step": 13577
    },
    {
      "epoch": 0.7603315040878038,
      "grad_norm": 1.3579661846160889,
      "learning_rate": 6.785e-05,
      "loss": 0.4462,
      "step": 13578
    },
    {
      "epoch": 0.7603875013999328,
      "grad_norm": 1.42841374874115,
      "learning_rate": 6.785500000000001e-05,
      "loss": 0.4167,
      "step": 13579
    },
    {
      "epoch": 0.7604434987120618,
      "grad_norm": 1.4315341711044312,
      "learning_rate": 6.786e-05,
      "loss": 0.4883,
      "step": 13580
    },
    {
      "epoch": 0.7604994960241909,
      "grad_norm": 1.3810499906539917,
      "learning_rate": 6.7865e-05,
      "loss": 0.5526,
      "step": 13581
    },
    {
      "epoch": 0.7605554933363199,
      "grad_norm": 1.5452520847320557,
      "learning_rate": 6.787e-05,
      "loss": 0.294,
      "step": 13582
    },
    {
      "epoch": 0.7606114906484489,
      "grad_norm": 1.2834396362304688,
      "learning_rate": 6.7875e-05,
      "loss": 0.5102,
      "step": 13583
    },
    {
      "epoch": 0.7606674879605779,
      "grad_norm": 1.6737674474716187,
      "learning_rate": 6.788e-05,
      "loss": 0.6717,
      "step": 13584
    },
    {
      "epoch": 0.760723485272707,
      "grad_norm": 1.3482942581176758,
      "learning_rate": 6.7885e-05,
      "loss": 0.4505,
      "step": 13585
    },
    {
      "epoch": 0.760779482584836,
      "grad_norm": 1.315356731414795,
      "learning_rate": 6.789e-05,
      "loss": 0.345,
      "step": 13586
    },
    {
      "epoch": 0.760835479896965,
      "grad_norm": 1.2866826057434082,
      "learning_rate": 6.789500000000001e-05,
      "loss": 0.3925,
      "step": 13587
    },
    {
      "epoch": 0.760891477209094,
      "grad_norm": 1.3313685655593872,
      "learning_rate": 6.790000000000001e-05,
      "loss": 0.4995,
      "step": 13588
    },
    {
      "epoch": 0.760947474521223,
      "grad_norm": 1.3239092826843262,
      "learning_rate": 6.790500000000001e-05,
      "loss": 0.5177,
      "step": 13589
    },
    {
      "epoch": 0.761003471833352,
      "grad_norm": 1.577412724494934,
      "learning_rate": 6.791e-05,
      "loss": 0.6218,
      "step": 13590
    },
    {
      "epoch": 0.7610594691454811,
      "grad_norm": 1.3489168882369995,
      "learning_rate": 6.7915e-05,
      "loss": 0.5704,
      "step": 13591
    },
    {
      "epoch": 0.7611154664576101,
      "grad_norm": 1.1966787576675415,
      "learning_rate": 6.792e-05,
      "loss": 0.4038,
      "step": 13592
    },
    {
      "epoch": 0.7611714637697391,
      "grad_norm": 1.236101746559143,
      "learning_rate": 6.792500000000001e-05,
      "loss": 0.392,
      "step": 13593
    },
    {
      "epoch": 0.7612274610818681,
      "grad_norm": 1.2937026023864746,
      "learning_rate": 6.793000000000001e-05,
      "loss": 0.39,
      "step": 13594
    },
    {
      "epoch": 0.7612834583939971,
      "grad_norm": 1.1674257516860962,
      "learning_rate": 6.7935e-05,
      "loss": 0.3888,
      "step": 13595
    },
    {
      "epoch": 0.7613394557061262,
      "grad_norm": 1.514306664466858,
      "learning_rate": 6.794e-05,
      "loss": 0.6035,
      "step": 13596
    },
    {
      "epoch": 0.7613954530182552,
      "grad_norm": 1.2262016534805298,
      "learning_rate": 6.7945e-05,
      "loss": 0.3332,
      "step": 13597
    },
    {
      "epoch": 0.7614514503303842,
      "grad_norm": 1.750225305557251,
      "learning_rate": 6.795e-05,
      "loss": 0.4764,
      "step": 13598
    },
    {
      "epoch": 0.7615074476425132,
      "grad_norm": 1.3644682168960571,
      "learning_rate": 6.7955e-05,
      "loss": 0.5144,
      "step": 13599
    },
    {
      "epoch": 0.7615634449546422,
      "grad_norm": 1.456863522529602,
      "learning_rate": 6.796e-05,
      "loss": 0.5256,
      "step": 13600
    },
    {
      "epoch": 0.7616194422667711,
      "grad_norm": 0.9988186359405518,
      "learning_rate": 6.7965e-05,
      "loss": 0.2737,
      "step": 13601
    },
    {
      "epoch": 0.7616754395789002,
      "grad_norm": 1.4243215322494507,
      "learning_rate": 6.797e-05,
      "loss": 0.4492,
      "step": 13602
    },
    {
      "epoch": 0.7617314368910292,
      "grad_norm": 1.331834316253662,
      "learning_rate": 6.7975e-05,
      "loss": 0.3962,
      "step": 13603
    },
    {
      "epoch": 0.7617874342031582,
      "grad_norm": 1.409032940864563,
      "learning_rate": 6.798e-05,
      "loss": 0.4875,
      "step": 13604
    },
    {
      "epoch": 0.7618434315152872,
      "grad_norm": 1.417380452156067,
      "learning_rate": 6.7985e-05,
      "loss": 0.4,
      "step": 13605
    },
    {
      "epoch": 0.7618994288274162,
      "grad_norm": 1.4346762895584106,
      "learning_rate": 6.799e-05,
      "loss": 0.3644,
      "step": 13606
    },
    {
      "epoch": 0.7619554261395453,
      "grad_norm": 1.5617364645004272,
      "learning_rate": 6.7995e-05,
      "loss": 0.6251,
      "step": 13607
    },
    {
      "epoch": 0.7620114234516743,
      "grad_norm": 1.664028525352478,
      "learning_rate": 6.800000000000001e-05,
      "loss": 0.4658,
      "step": 13608
    },
    {
      "epoch": 0.7620674207638033,
      "grad_norm": 1.2303657531738281,
      "learning_rate": 6.800500000000001e-05,
      "loss": 0.3691,
      "step": 13609
    },
    {
      "epoch": 0.7621234180759323,
      "grad_norm": 1.4195746183395386,
      "learning_rate": 6.801000000000001e-05,
      "loss": 0.4241,
      "step": 13610
    },
    {
      "epoch": 0.7621794153880613,
      "grad_norm": 1.0718209743499756,
      "learning_rate": 6.8015e-05,
      "loss": 0.4468,
      "step": 13611
    },
    {
      "epoch": 0.7622354127001904,
      "grad_norm": 1.1220693588256836,
      "learning_rate": 6.802e-05,
      "loss": 0.3926,
      "step": 13612
    },
    {
      "epoch": 0.7622914100123194,
      "grad_norm": 1.277579426765442,
      "learning_rate": 6.8025e-05,
      "loss": 0.3934,
      "step": 13613
    },
    {
      "epoch": 0.7623474073244484,
      "grad_norm": 1.3952373266220093,
      "learning_rate": 6.803000000000001e-05,
      "loss": 0.5871,
      "step": 13614
    },
    {
      "epoch": 0.7624034046365774,
      "grad_norm": 1.304663062095642,
      "learning_rate": 6.803500000000001e-05,
      "loss": 0.4321,
      "step": 13615
    },
    {
      "epoch": 0.7624594019487064,
      "grad_norm": 1.3302100896835327,
      "learning_rate": 6.804e-05,
      "loss": 0.4849,
      "step": 13616
    },
    {
      "epoch": 0.7625153992608354,
      "grad_norm": 1.1536314487457275,
      "learning_rate": 6.8045e-05,
      "loss": 0.4198,
      "step": 13617
    },
    {
      "epoch": 0.7625713965729645,
      "grad_norm": 1.2717252969741821,
      "learning_rate": 6.805e-05,
      "loss": 0.4946,
      "step": 13618
    },
    {
      "epoch": 0.7626273938850935,
      "grad_norm": 1.3787095546722412,
      "learning_rate": 6.8055e-05,
      "loss": 0.5078,
      "step": 13619
    },
    {
      "epoch": 0.7626833911972225,
      "grad_norm": 1.3335047960281372,
      "learning_rate": 6.806e-05,
      "loss": 0.5409,
      "step": 13620
    },
    {
      "epoch": 0.7627393885093515,
      "grad_norm": 1.2705175876617432,
      "learning_rate": 6.8065e-05,
      "loss": 0.4512,
      "step": 13621
    },
    {
      "epoch": 0.7627953858214805,
      "grad_norm": 1.4200292825698853,
      "learning_rate": 6.807e-05,
      "loss": 0.3421,
      "step": 13622
    },
    {
      "epoch": 0.7628513831336096,
      "grad_norm": 1.419069766998291,
      "learning_rate": 6.8075e-05,
      "loss": 0.4204,
      "step": 13623
    },
    {
      "epoch": 0.7629073804457386,
      "grad_norm": 1.2540982961654663,
      "learning_rate": 6.808e-05,
      "loss": 0.5944,
      "step": 13624
    },
    {
      "epoch": 0.7629633777578676,
      "grad_norm": 1.294551134109497,
      "learning_rate": 6.8085e-05,
      "loss": 0.4136,
      "step": 13625
    },
    {
      "epoch": 0.7630193750699966,
      "grad_norm": 1.2086056470870972,
      "learning_rate": 6.809e-05,
      "loss": 0.4379,
      "step": 13626
    },
    {
      "epoch": 0.7630753723821256,
      "grad_norm": 1.7807788848876953,
      "learning_rate": 6.8095e-05,
      "loss": 0.5198,
      "step": 13627
    },
    {
      "epoch": 0.7631313696942547,
      "grad_norm": 1.272684931755066,
      "learning_rate": 6.81e-05,
      "loss": 0.4241,
      "step": 13628
    },
    {
      "epoch": 0.7631873670063837,
      "grad_norm": 1.271483063697815,
      "learning_rate": 6.810500000000001e-05,
      "loss": 0.3404,
      "step": 13629
    },
    {
      "epoch": 0.7632433643185127,
      "grad_norm": 1.1285747289657593,
      "learning_rate": 6.811000000000001e-05,
      "loss": 0.4228,
      "step": 13630
    },
    {
      "epoch": 0.7632993616306417,
      "grad_norm": 1.5059736967086792,
      "learning_rate": 6.811500000000001e-05,
      "loss": 0.5092,
      "step": 13631
    },
    {
      "epoch": 0.7633553589427707,
      "grad_norm": 1.2974300384521484,
      "learning_rate": 6.812e-05,
      "loss": 0.462,
      "step": 13632
    },
    {
      "epoch": 0.7634113562548998,
      "grad_norm": 1.4447135925292969,
      "learning_rate": 6.8125e-05,
      "loss": 0.3697,
      "step": 13633
    },
    {
      "epoch": 0.7634673535670288,
      "grad_norm": 1.396356225013733,
      "learning_rate": 6.813e-05,
      "loss": 0.454,
      "step": 13634
    },
    {
      "epoch": 0.7635233508791578,
      "grad_norm": 1.2440048456192017,
      "learning_rate": 6.813500000000001e-05,
      "loss": 0.4406,
      "step": 13635
    },
    {
      "epoch": 0.7635793481912868,
      "grad_norm": 1.2469977140426636,
      "learning_rate": 6.814000000000001e-05,
      "loss": 0.4061,
      "step": 13636
    },
    {
      "epoch": 0.7636353455034158,
      "grad_norm": 1.319535255432129,
      "learning_rate": 6.8145e-05,
      "loss": 0.4064,
      "step": 13637
    },
    {
      "epoch": 0.7636913428155448,
      "grad_norm": 1.4810253381729126,
      "learning_rate": 6.815e-05,
      "loss": 0.4953,
      "step": 13638
    },
    {
      "epoch": 0.7637473401276739,
      "grad_norm": 1.311285376548767,
      "learning_rate": 6.8155e-05,
      "loss": 0.395,
      "step": 13639
    },
    {
      "epoch": 0.7638033374398029,
      "grad_norm": 1.3299849033355713,
      "learning_rate": 6.816e-05,
      "loss": 0.476,
      "step": 13640
    },
    {
      "epoch": 0.7638593347519319,
      "grad_norm": 1.3407683372497559,
      "learning_rate": 6.8165e-05,
      "loss": 0.386,
      "step": 13641
    },
    {
      "epoch": 0.7639153320640609,
      "grad_norm": 1.4075672626495361,
      "learning_rate": 6.817e-05,
      "loss": 0.4974,
      "step": 13642
    },
    {
      "epoch": 0.76397132937619,
      "grad_norm": 1.3753952980041504,
      "learning_rate": 6.8175e-05,
      "loss": 0.4195,
      "step": 13643
    },
    {
      "epoch": 0.764027326688319,
      "grad_norm": 1.2342287302017212,
      "learning_rate": 6.818e-05,
      "loss": 0.4695,
      "step": 13644
    },
    {
      "epoch": 0.764083324000448,
      "grad_norm": 1.3846280574798584,
      "learning_rate": 6.8185e-05,
      "loss": 0.3887,
      "step": 13645
    },
    {
      "epoch": 0.764139321312577,
      "grad_norm": 1.123100996017456,
      "learning_rate": 6.819e-05,
      "loss": 0.452,
      "step": 13646
    },
    {
      "epoch": 0.764195318624706,
      "grad_norm": 1.3150691986083984,
      "learning_rate": 6.8195e-05,
      "loss": 0.5618,
      "step": 13647
    },
    {
      "epoch": 0.764251315936835,
      "grad_norm": 1.6028003692626953,
      "learning_rate": 6.82e-05,
      "loss": 0.5873,
      "step": 13648
    },
    {
      "epoch": 0.7643073132489641,
      "grad_norm": 1.3623460531234741,
      "learning_rate": 6.8205e-05,
      "loss": 0.4849,
      "step": 13649
    },
    {
      "epoch": 0.7643633105610931,
      "grad_norm": 1.2319543361663818,
      "learning_rate": 6.821000000000001e-05,
      "loss": 0.4245,
      "step": 13650
    },
    {
      "epoch": 0.7644193078732221,
      "grad_norm": 1.5644632577896118,
      "learning_rate": 6.821500000000001e-05,
      "loss": 0.6767,
      "step": 13651
    },
    {
      "epoch": 0.7644753051853511,
      "grad_norm": 1.3389331102371216,
      "learning_rate": 6.822000000000001e-05,
      "loss": 0.4127,
      "step": 13652
    },
    {
      "epoch": 0.7645313024974801,
      "grad_norm": 1.673305630683899,
      "learning_rate": 6.8225e-05,
      "loss": 0.4908,
      "step": 13653
    },
    {
      "epoch": 0.7645872998096092,
      "grad_norm": 1.2320618629455566,
      "learning_rate": 6.823e-05,
      "loss": 0.4999,
      "step": 13654
    },
    {
      "epoch": 0.7646432971217382,
      "grad_norm": 1.3539533615112305,
      "learning_rate": 6.8235e-05,
      "loss": 0.4486,
      "step": 13655
    },
    {
      "epoch": 0.7646992944338672,
      "grad_norm": 1.133371114730835,
      "learning_rate": 6.824e-05,
      "loss": 0.329,
      "step": 13656
    },
    {
      "epoch": 0.7647552917459962,
      "grad_norm": 1.3948217630386353,
      "learning_rate": 6.824500000000001e-05,
      "loss": 0.4525,
      "step": 13657
    },
    {
      "epoch": 0.7648112890581252,
      "grad_norm": 1.3602687120437622,
      "learning_rate": 6.825e-05,
      "loss": 0.4447,
      "step": 13658
    },
    {
      "epoch": 0.7648672863702543,
      "grad_norm": 1.3339569568634033,
      "learning_rate": 6.8255e-05,
      "loss": 0.5464,
      "step": 13659
    },
    {
      "epoch": 0.7649232836823833,
      "grad_norm": 1.3197356462478638,
      "learning_rate": 6.826e-05,
      "loss": 0.3726,
      "step": 13660
    },
    {
      "epoch": 0.7649792809945123,
      "grad_norm": 1.3283066749572754,
      "learning_rate": 6.8265e-05,
      "loss": 0.5509,
      "step": 13661
    },
    {
      "epoch": 0.7650352783066413,
      "grad_norm": 1.8376636505126953,
      "learning_rate": 6.827e-05,
      "loss": 0.6377,
      "step": 13662
    },
    {
      "epoch": 0.7650912756187703,
      "grad_norm": 1.1652458906173706,
      "learning_rate": 6.8275e-05,
      "loss": 0.4488,
      "step": 13663
    },
    {
      "epoch": 0.7651472729308993,
      "grad_norm": 1.3074110746383667,
      "learning_rate": 6.828e-05,
      "loss": 0.4817,
      "step": 13664
    },
    {
      "epoch": 0.7652032702430284,
      "grad_norm": 1.4740490913391113,
      "learning_rate": 6.8285e-05,
      "loss": 0.4235,
      "step": 13665
    },
    {
      "epoch": 0.7652592675551574,
      "grad_norm": 1.4304113388061523,
      "learning_rate": 6.829e-05,
      "loss": 0.5427,
      "step": 13666
    },
    {
      "epoch": 0.7653152648672864,
      "grad_norm": 1.463375210762024,
      "learning_rate": 6.8295e-05,
      "loss": 0.4777,
      "step": 13667
    },
    {
      "epoch": 0.7653712621794154,
      "grad_norm": 1.1754728555679321,
      "learning_rate": 6.83e-05,
      "loss": 0.394,
      "step": 13668
    },
    {
      "epoch": 0.7654272594915444,
      "grad_norm": 1.3912222385406494,
      "learning_rate": 6.8305e-05,
      "loss": 0.5556,
      "step": 13669
    },
    {
      "epoch": 0.7654832568036735,
      "grad_norm": 1.3535871505737305,
      "learning_rate": 6.831e-05,
      "loss": 0.5625,
      "step": 13670
    },
    {
      "epoch": 0.7655392541158025,
      "grad_norm": 1.1886454820632935,
      "learning_rate": 6.831500000000001e-05,
      "loss": 0.4734,
      "step": 13671
    },
    {
      "epoch": 0.7655952514279315,
      "grad_norm": 2.200488567352295,
      "learning_rate": 6.832000000000001e-05,
      "loss": 0.5001,
      "step": 13672
    },
    {
      "epoch": 0.7656512487400605,
      "grad_norm": 1.2043800354003906,
      "learning_rate": 6.832500000000001e-05,
      "loss": 0.4501,
      "step": 13673
    },
    {
      "epoch": 0.7657072460521895,
      "grad_norm": 1.228277325630188,
      "learning_rate": 6.833e-05,
      "loss": 0.4027,
      "step": 13674
    },
    {
      "epoch": 0.7657632433643186,
      "grad_norm": 1.2629377841949463,
      "learning_rate": 6.8335e-05,
      "loss": 0.4552,
      "step": 13675
    },
    {
      "epoch": 0.7658192406764476,
      "grad_norm": 1.2324979305267334,
      "learning_rate": 6.834e-05,
      "loss": 0.3709,
      "step": 13676
    },
    {
      "epoch": 0.7658752379885766,
      "grad_norm": 1.163761854171753,
      "learning_rate": 6.8345e-05,
      "loss": 0.5439,
      "step": 13677
    },
    {
      "epoch": 0.7659312353007056,
      "grad_norm": 1.2088426351547241,
      "learning_rate": 6.835000000000001e-05,
      "loss": 0.4621,
      "step": 13678
    },
    {
      "epoch": 0.7659872326128346,
      "grad_norm": 1.2792140245437622,
      "learning_rate": 6.8355e-05,
      "loss": 0.4059,
      "step": 13679
    },
    {
      "epoch": 0.7660432299249637,
      "grad_norm": 1.3498033285140991,
      "learning_rate": 6.836e-05,
      "loss": 0.4016,
      "step": 13680
    },
    {
      "epoch": 0.7660992272370927,
      "grad_norm": 1.7404643297195435,
      "learning_rate": 6.8365e-05,
      "loss": 0.5288,
      "step": 13681
    },
    {
      "epoch": 0.7661552245492217,
      "grad_norm": 1.2052669525146484,
      "learning_rate": 6.837e-05,
      "loss": 0.4172,
      "step": 13682
    },
    {
      "epoch": 0.7662112218613506,
      "grad_norm": 1.1906694173812866,
      "learning_rate": 6.8375e-05,
      "loss": 0.3374,
      "step": 13683
    },
    {
      "epoch": 0.7662672191734796,
      "grad_norm": 1.4606155157089233,
      "learning_rate": 6.838e-05,
      "loss": 0.4478,
      "step": 13684
    },
    {
      "epoch": 0.7663232164856086,
      "grad_norm": 1.5571273565292358,
      "learning_rate": 6.8385e-05,
      "loss": 0.4997,
      "step": 13685
    },
    {
      "epoch": 0.7663792137977377,
      "grad_norm": 1.4292577505111694,
      "learning_rate": 6.839e-05,
      "loss": 0.4201,
      "step": 13686
    },
    {
      "epoch": 0.7664352111098667,
      "grad_norm": 1.3577619791030884,
      "learning_rate": 6.8395e-05,
      "loss": 0.393,
      "step": 13687
    },
    {
      "epoch": 0.7664912084219957,
      "grad_norm": 1.3015681505203247,
      "learning_rate": 6.840000000000001e-05,
      "loss": 0.3905,
      "step": 13688
    },
    {
      "epoch": 0.7665472057341247,
      "grad_norm": 1.588666558265686,
      "learning_rate": 6.8405e-05,
      "loss": 0.4982,
      "step": 13689
    },
    {
      "epoch": 0.7666032030462537,
      "grad_norm": 1.0863832235336304,
      "learning_rate": 6.841e-05,
      "loss": 0.3979,
      "step": 13690
    },
    {
      "epoch": 0.7666592003583828,
      "grad_norm": 1.5667450428009033,
      "learning_rate": 6.8415e-05,
      "loss": 0.6567,
      "step": 13691
    },
    {
      "epoch": 0.7667151976705118,
      "grad_norm": 1.1819483041763306,
      "learning_rate": 6.842000000000001e-05,
      "loss": 0.5536,
      "step": 13692
    },
    {
      "epoch": 0.7667711949826408,
      "grad_norm": 1.4729474782943726,
      "learning_rate": 6.842500000000001e-05,
      "loss": 0.5742,
      "step": 13693
    },
    {
      "epoch": 0.7668271922947698,
      "grad_norm": 1.5023722648620605,
      "learning_rate": 6.843000000000001e-05,
      "loss": 0.4246,
      "step": 13694
    },
    {
      "epoch": 0.7668831896068988,
      "grad_norm": 1.6079078912734985,
      "learning_rate": 6.8435e-05,
      "loss": 0.4576,
      "step": 13695
    },
    {
      "epoch": 0.7669391869190278,
      "grad_norm": 1.3467698097229004,
      "learning_rate": 6.844e-05,
      "loss": 0.4981,
      "step": 13696
    },
    {
      "epoch": 0.7669951842311569,
      "grad_norm": 1.4072273969650269,
      "learning_rate": 6.8445e-05,
      "loss": 0.3155,
      "step": 13697
    },
    {
      "epoch": 0.7670511815432859,
      "grad_norm": 1.3815951347351074,
      "learning_rate": 6.845e-05,
      "loss": 0.376,
      "step": 13698
    },
    {
      "epoch": 0.7671071788554149,
      "grad_norm": 1.4208966493606567,
      "learning_rate": 6.845500000000001e-05,
      "loss": 0.5671,
      "step": 13699
    },
    {
      "epoch": 0.7671631761675439,
      "grad_norm": 1.2100359201431274,
      "learning_rate": 6.846e-05,
      "loss": 0.4106,
      "step": 13700
    },
    {
      "epoch": 0.767219173479673,
      "grad_norm": 1.3779208660125732,
      "learning_rate": 6.8465e-05,
      "loss": 0.5216,
      "step": 13701
    },
    {
      "epoch": 0.767275170791802,
      "grad_norm": 1.2398920059204102,
      "learning_rate": 6.847e-05,
      "loss": 0.3448,
      "step": 13702
    },
    {
      "epoch": 0.767331168103931,
      "grad_norm": 1.2889721393585205,
      "learning_rate": 6.8475e-05,
      "loss": 0.4105,
      "step": 13703
    },
    {
      "epoch": 0.76738716541606,
      "grad_norm": 1.327928066253662,
      "learning_rate": 6.848e-05,
      "loss": 0.5157,
      "step": 13704
    },
    {
      "epoch": 0.767443162728189,
      "grad_norm": 1.3076730966567993,
      "learning_rate": 6.8485e-05,
      "loss": 0.3998,
      "step": 13705
    },
    {
      "epoch": 0.767499160040318,
      "grad_norm": 1.2470765113830566,
      "learning_rate": 6.849e-05,
      "loss": 0.3305,
      "step": 13706
    },
    {
      "epoch": 0.7675551573524471,
      "grad_norm": 1.3026156425476074,
      "learning_rate": 6.8495e-05,
      "loss": 0.4167,
      "step": 13707
    },
    {
      "epoch": 0.7676111546645761,
      "grad_norm": 1.1424435377120972,
      "learning_rate": 6.850000000000001e-05,
      "loss": 0.4479,
      "step": 13708
    },
    {
      "epoch": 0.7676671519767051,
      "grad_norm": 1.2309578657150269,
      "learning_rate": 6.850500000000001e-05,
      "loss": 0.4619,
      "step": 13709
    },
    {
      "epoch": 0.7677231492888341,
      "grad_norm": 1.623602032661438,
      "learning_rate": 6.851e-05,
      "loss": 0.4734,
      "step": 13710
    },
    {
      "epoch": 0.7677791466009631,
      "grad_norm": 1.4080479145050049,
      "learning_rate": 6.8515e-05,
      "loss": 0.421,
      "step": 13711
    },
    {
      "epoch": 0.7678351439130922,
      "grad_norm": 1.8655375242233276,
      "learning_rate": 6.852e-05,
      "loss": 0.3954,
      "step": 13712
    },
    {
      "epoch": 0.7678911412252212,
      "grad_norm": 1.8153020143508911,
      "learning_rate": 6.852500000000001e-05,
      "loss": 0.5158,
      "step": 13713
    },
    {
      "epoch": 0.7679471385373502,
      "grad_norm": 1.4256774187088013,
      "learning_rate": 6.853000000000001e-05,
      "loss": 0.423,
      "step": 13714
    },
    {
      "epoch": 0.7680031358494792,
      "grad_norm": 1.6449263095855713,
      "learning_rate": 6.853500000000001e-05,
      "loss": 0.5707,
      "step": 13715
    },
    {
      "epoch": 0.7680591331616082,
      "grad_norm": 1.398479700088501,
      "learning_rate": 6.854e-05,
      "loss": 0.4827,
      "step": 13716
    },
    {
      "epoch": 0.7681151304737373,
      "grad_norm": 1.153760313987732,
      "learning_rate": 6.8545e-05,
      "loss": 0.4057,
      "step": 13717
    },
    {
      "epoch": 0.7681711277858663,
      "grad_norm": 1.3056362867355347,
      "learning_rate": 6.855e-05,
      "loss": 0.4192,
      "step": 13718
    },
    {
      "epoch": 0.7682271250979953,
      "grad_norm": 1.4091055393218994,
      "learning_rate": 6.8555e-05,
      "loss": 0.5524,
      "step": 13719
    },
    {
      "epoch": 0.7682831224101243,
      "grad_norm": 1.2711598873138428,
      "learning_rate": 6.856000000000001e-05,
      "loss": 0.4829,
      "step": 13720
    },
    {
      "epoch": 0.7683391197222533,
      "grad_norm": 1.2869224548339844,
      "learning_rate": 6.8565e-05,
      "loss": 0.6806,
      "step": 13721
    },
    {
      "epoch": 0.7683951170343823,
      "grad_norm": 1.2931668758392334,
      "learning_rate": 6.857e-05,
      "loss": 0.4188,
      "step": 13722
    },
    {
      "epoch": 0.7684511143465114,
      "grad_norm": 1.210431456565857,
      "learning_rate": 6.8575e-05,
      "loss": 0.3685,
      "step": 13723
    },
    {
      "epoch": 0.7685071116586404,
      "grad_norm": 1.2958704233169556,
      "learning_rate": 6.858e-05,
      "loss": 0.4263,
      "step": 13724
    },
    {
      "epoch": 0.7685631089707694,
      "grad_norm": 1.4035395383834839,
      "learning_rate": 6.8585e-05,
      "loss": 0.4423,
      "step": 13725
    },
    {
      "epoch": 0.7686191062828984,
      "grad_norm": 1.3132297992706299,
      "learning_rate": 6.858999999999999e-05,
      "loss": 0.4727,
      "step": 13726
    },
    {
      "epoch": 0.7686751035950274,
      "grad_norm": 1.3928653001785278,
      "learning_rate": 6.8595e-05,
      "loss": 0.3781,
      "step": 13727
    },
    {
      "epoch": 0.7687311009071565,
      "grad_norm": 1.354992389678955,
      "learning_rate": 6.860000000000001e-05,
      "loss": 0.4253,
      "step": 13728
    },
    {
      "epoch": 0.7687870982192855,
      "grad_norm": 1.6718236207962036,
      "learning_rate": 6.860500000000001e-05,
      "loss": 0.6554,
      "step": 13729
    },
    {
      "epoch": 0.7688430955314145,
      "grad_norm": 1.3217116594314575,
      "learning_rate": 6.861000000000001e-05,
      "loss": 0.4524,
      "step": 13730
    },
    {
      "epoch": 0.7688990928435435,
      "grad_norm": 1.318081259727478,
      "learning_rate": 6.8615e-05,
      "loss": 0.4761,
      "step": 13731
    },
    {
      "epoch": 0.7689550901556725,
      "grad_norm": 0.9450418949127197,
      "learning_rate": 6.862e-05,
      "loss": 0.3411,
      "step": 13732
    },
    {
      "epoch": 0.7690110874678016,
      "grad_norm": 1.217575192451477,
      "learning_rate": 6.8625e-05,
      "loss": 0.4743,
      "step": 13733
    },
    {
      "epoch": 0.7690670847799306,
      "grad_norm": 1.4697747230529785,
      "learning_rate": 6.863000000000001e-05,
      "loss": 0.573,
      "step": 13734
    },
    {
      "epoch": 0.7691230820920596,
      "grad_norm": 1.4503259658813477,
      "learning_rate": 6.863500000000001e-05,
      "loss": 0.4563,
      "step": 13735
    },
    {
      "epoch": 0.7691790794041886,
      "grad_norm": 1.291875958442688,
      "learning_rate": 6.864000000000001e-05,
      "loss": 0.5397,
      "step": 13736
    },
    {
      "epoch": 0.7692350767163176,
      "grad_norm": 1.412310242652893,
      "learning_rate": 6.8645e-05,
      "loss": 0.5399,
      "step": 13737
    },
    {
      "epoch": 0.7692910740284467,
      "grad_norm": 1.4714233875274658,
      "learning_rate": 6.865e-05,
      "loss": 0.6858,
      "step": 13738
    },
    {
      "epoch": 0.7693470713405757,
      "grad_norm": 1.0596532821655273,
      "learning_rate": 6.8655e-05,
      "loss": 0.2782,
      "step": 13739
    },
    {
      "epoch": 0.7694030686527047,
      "grad_norm": 1.4776426553726196,
      "learning_rate": 6.866e-05,
      "loss": 0.4573,
      "step": 13740
    },
    {
      "epoch": 0.7694590659648337,
      "grad_norm": 1.279836893081665,
      "learning_rate": 6.866500000000001e-05,
      "loss": 0.3918,
      "step": 13741
    },
    {
      "epoch": 0.7695150632769627,
      "grad_norm": 1.4288512468338013,
      "learning_rate": 6.867e-05,
      "loss": 0.3375,
      "step": 13742
    },
    {
      "epoch": 0.7695710605890917,
      "grad_norm": 1.5765928030014038,
      "learning_rate": 6.8675e-05,
      "loss": 0.4134,
      "step": 13743
    },
    {
      "epoch": 0.7696270579012208,
      "grad_norm": 1.4100160598754883,
      "learning_rate": 6.868e-05,
      "loss": 0.4122,
      "step": 13744
    },
    {
      "epoch": 0.7696830552133498,
      "grad_norm": 1.3476622104644775,
      "learning_rate": 6.8685e-05,
      "loss": 0.4397,
      "step": 13745
    },
    {
      "epoch": 0.7697390525254788,
      "grad_norm": 1.3707756996154785,
      "learning_rate": 6.869e-05,
      "loss": 0.4645,
      "step": 13746
    },
    {
      "epoch": 0.7697950498376078,
      "grad_norm": 1.6237640380859375,
      "learning_rate": 6.869499999999999e-05,
      "loss": 0.5476,
      "step": 13747
    },
    {
      "epoch": 0.7698510471497368,
      "grad_norm": 1.3181239366531372,
      "learning_rate": 6.87e-05,
      "loss": 0.3942,
      "step": 13748
    },
    {
      "epoch": 0.7699070444618659,
      "grad_norm": 1.353237509727478,
      "learning_rate": 6.870500000000001e-05,
      "loss": 0.4023,
      "step": 13749
    },
    {
      "epoch": 0.7699630417739949,
      "grad_norm": 1.4873833656311035,
      "learning_rate": 6.871000000000001e-05,
      "loss": 0.4376,
      "step": 13750
    },
    {
      "epoch": 0.7700190390861239,
      "grad_norm": 1.360853672027588,
      "learning_rate": 6.871500000000001e-05,
      "loss": 0.4512,
      "step": 13751
    },
    {
      "epoch": 0.7700750363982529,
      "grad_norm": 1.18434476852417,
      "learning_rate": 6.872e-05,
      "loss": 0.3869,
      "step": 13752
    },
    {
      "epoch": 0.7701310337103819,
      "grad_norm": 1.1423625946044922,
      "learning_rate": 6.8725e-05,
      "loss": 0.3605,
      "step": 13753
    },
    {
      "epoch": 0.770187031022511,
      "grad_norm": 1.325973391532898,
      "learning_rate": 6.873e-05,
      "loss": 0.2949,
      "step": 13754
    },
    {
      "epoch": 0.77024302833464,
      "grad_norm": 1.4096765518188477,
      "learning_rate": 6.8735e-05,
      "loss": 0.3872,
      "step": 13755
    },
    {
      "epoch": 0.770299025646769,
      "grad_norm": 1.3526920080184937,
      "learning_rate": 6.874000000000001e-05,
      "loss": 0.4204,
      "step": 13756
    },
    {
      "epoch": 0.770355022958898,
      "grad_norm": 1.1253491640090942,
      "learning_rate": 6.8745e-05,
      "loss": 0.3993,
      "step": 13757
    },
    {
      "epoch": 0.770411020271027,
      "grad_norm": 1.1818180084228516,
      "learning_rate": 6.875e-05,
      "loss": 0.4489,
      "step": 13758
    },
    {
      "epoch": 0.770467017583156,
      "grad_norm": 1.7405974864959717,
      "learning_rate": 6.8755e-05,
      "loss": 0.5501,
      "step": 13759
    },
    {
      "epoch": 0.7705230148952851,
      "grad_norm": 1.3856478929519653,
      "learning_rate": 6.876e-05,
      "loss": 0.6071,
      "step": 13760
    },
    {
      "epoch": 0.7705790122074141,
      "grad_norm": 1.6659204959869385,
      "learning_rate": 6.8765e-05,
      "loss": 0.6156,
      "step": 13761
    },
    {
      "epoch": 0.7706350095195431,
      "grad_norm": 1.35427987575531,
      "learning_rate": 6.877000000000001e-05,
      "loss": 0.4552,
      "step": 13762
    },
    {
      "epoch": 0.7706910068316721,
      "grad_norm": 1.2575783729553223,
      "learning_rate": 6.8775e-05,
      "loss": 0.423,
      "step": 13763
    },
    {
      "epoch": 0.7707470041438012,
      "grad_norm": 1.2999074459075928,
      "learning_rate": 6.878e-05,
      "loss": 0.5113,
      "step": 13764
    },
    {
      "epoch": 0.7708030014559302,
      "grad_norm": 1.3737356662750244,
      "learning_rate": 6.8785e-05,
      "loss": 0.4653,
      "step": 13765
    },
    {
      "epoch": 0.7708589987680591,
      "grad_norm": 1.4483639001846313,
      "learning_rate": 6.879e-05,
      "loss": 0.5,
      "step": 13766
    },
    {
      "epoch": 0.7709149960801881,
      "grad_norm": 1.2880616188049316,
      "learning_rate": 6.8795e-05,
      "loss": 0.3674,
      "step": 13767
    },
    {
      "epoch": 0.7709709933923171,
      "grad_norm": 1.4078574180603027,
      "learning_rate": 6.879999999999999e-05,
      "loss": 0.5425,
      "step": 13768
    },
    {
      "epoch": 0.7710269907044461,
      "grad_norm": 1.2571057081222534,
      "learning_rate": 6.8805e-05,
      "loss": 0.3848,
      "step": 13769
    },
    {
      "epoch": 0.7710829880165752,
      "grad_norm": 1.589552640914917,
      "learning_rate": 6.881000000000001e-05,
      "loss": 0.6412,
      "step": 13770
    },
    {
      "epoch": 0.7711389853287042,
      "grad_norm": 1.9987590312957764,
      "learning_rate": 6.881500000000001e-05,
      "loss": 0.5261,
      "step": 13771
    },
    {
      "epoch": 0.7711949826408332,
      "grad_norm": 1.2986613512039185,
      "learning_rate": 6.882000000000001e-05,
      "loss": 0.4911,
      "step": 13772
    },
    {
      "epoch": 0.7712509799529622,
      "grad_norm": 1.2847800254821777,
      "learning_rate": 6.8825e-05,
      "loss": 0.4441,
      "step": 13773
    },
    {
      "epoch": 0.7713069772650912,
      "grad_norm": 1.485494613647461,
      "learning_rate": 6.883e-05,
      "loss": 0.6333,
      "step": 13774
    },
    {
      "epoch": 0.7713629745772203,
      "grad_norm": 1.34844970703125,
      "learning_rate": 6.8835e-05,
      "loss": 0.3288,
      "step": 13775
    },
    {
      "epoch": 0.7714189718893493,
      "grad_norm": 1.1112004518508911,
      "learning_rate": 6.884e-05,
      "loss": 0.3564,
      "step": 13776
    },
    {
      "epoch": 0.7714749692014783,
      "grad_norm": 1.1809488534927368,
      "learning_rate": 6.884500000000001e-05,
      "loss": 0.3969,
      "step": 13777
    },
    {
      "epoch": 0.7715309665136073,
      "grad_norm": 1.8307031393051147,
      "learning_rate": 6.885e-05,
      "loss": 0.4789,
      "step": 13778
    },
    {
      "epoch": 0.7715869638257363,
      "grad_norm": 1.2297064065933228,
      "learning_rate": 6.8855e-05,
      "loss": 0.4582,
      "step": 13779
    },
    {
      "epoch": 0.7716429611378653,
      "grad_norm": 1.400884747505188,
      "learning_rate": 6.886e-05,
      "loss": 0.4653,
      "step": 13780
    },
    {
      "epoch": 0.7716989584499944,
      "grad_norm": 1.321446180343628,
      "learning_rate": 6.8865e-05,
      "loss": 0.3972,
      "step": 13781
    },
    {
      "epoch": 0.7717549557621234,
      "grad_norm": 1.7086085081100464,
      "learning_rate": 6.887e-05,
      "loss": 0.6464,
      "step": 13782
    },
    {
      "epoch": 0.7718109530742524,
      "grad_norm": 1.319392204284668,
      "learning_rate": 6.887500000000001e-05,
      "loss": 0.3759,
      "step": 13783
    },
    {
      "epoch": 0.7718669503863814,
      "grad_norm": 1.6760494709014893,
      "learning_rate": 6.888e-05,
      "loss": 0.5043,
      "step": 13784
    },
    {
      "epoch": 0.7719229476985104,
      "grad_norm": 1.1474812030792236,
      "learning_rate": 6.8885e-05,
      "loss": 0.3287,
      "step": 13785
    },
    {
      "epoch": 0.7719789450106395,
      "grad_norm": 1.2620575428009033,
      "learning_rate": 6.889e-05,
      "loss": 0.4179,
      "step": 13786
    },
    {
      "epoch": 0.7720349423227685,
      "grad_norm": 1.1926451921463013,
      "learning_rate": 6.8895e-05,
      "loss": 0.3944,
      "step": 13787
    },
    {
      "epoch": 0.7720909396348975,
      "grad_norm": 1.2984777688980103,
      "learning_rate": 6.89e-05,
      "loss": 0.4094,
      "step": 13788
    },
    {
      "epoch": 0.7721469369470265,
      "grad_norm": 1.5784839391708374,
      "learning_rate": 6.8905e-05,
      "loss": 0.5894,
      "step": 13789
    },
    {
      "epoch": 0.7722029342591555,
      "grad_norm": 1.2304500341415405,
      "learning_rate": 6.891e-05,
      "loss": 0.3704,
      "step": 13790
    },
    {
      "epoch": 0.7722589315712846,
      "grad_norm": 1.6105252504348755,
      "learning_rate": 6.891500000000001e-05,
      "loss": 0.5236,
      "step": 13791
    },
    {
      "epoch": 0.7723149288834136,
      "grad_norm": 1.457632303237915,
      "learning_rate": 6.892000000000001e-05,
      "loss": 0.4524,
      "step": 13792
    },
    {
      "epoch": 0.7723709261955426,
      "grad_norm": 1.252092719078064,
      "learning_rate": 6.892500000000001e-05,
      "loss": 0.382,
      "step": 13793
    },
    {
      "epoch": 0.7724269235076716,
      "grad_norm": 1.1379681825637817,
      "learning_rate": 6.893e-05,
      "loss": 0.3817,
      "step": 13794
    },
    {
      "epoch": 0.7724829208198006,
      "grad_norm": 1.3736088275909424,
      "learning_rate": 6.8935e-05,
      "loss": 0.4011,
      "step": 13795
    },
    {
      "epoch": 0.7725389181319297,
      "grad_norm": 1.4902105331420898,
      "learning_rate": 6.894e-05,
      "loss": 0.4807,
      "step": 13796
    },
    {
      "epoch": 0.7725949154440587,
      "grad_norm": 1.2289371490478516,
      "learning_rate": 6.8945e-05,
      "loss": 0.3919,
      "step": 13797
    },
    {
      "epoch": 0.7726509127561877,
      "grad_norm": 1.515181541442871,
      "learning_rate": 6.895000000000001e-05,
      "loss": 0.4261,
      "step": 13798
    },
    {
      "epoch": 0.7727069100683167,
      "grad_norm": 1.2557302713394165,
      "learning_rate": 6.8955e-05,
      "loss": 0.518,
      "step": 13799
    },
    {
      "epoch": 0.7727629073804457,
      "grad_norm": 1.485304832458496,
      "learning_rate": 6.896e-05,
      "loss": 0.4222,
      "step": 13800
    },
    {
      "epoch": 0.7728189046925747,
      "grad_norm": 1.253737211227417,
      "learning_rate": 6.8965e-05,
      "loss": 0.4721,
      "step": 13801
    },
    {
      "epoch": 0.7728749020047038,
      "grad_norm": 1.5144121646881104,
      "learning_rate": 6.897e-05,
      "loss": 0.4277,
      "step": 13802
    },
    {
      "epoch": 0.7729308993168328,
      "grad_norm": 1.363663673400879,
      "learning_rate": 6.8975e-05,
      "loss": 0.3993,
      "step": 13803
    },
    {
      "epoch": 0.7729868966289618,
      "grad_norm": 1.209930181503296,
      "learning_rate": 6.898e-05,
      "loss": 0.3794,
      "step": 13804
    },
    {
      "epoch": 0.7730428939410908,
      "grad_norm": 1.2857134342193604,
      "learning_rate": 6.8985e-05,
      "loss": 0.4527,
      "step": 13805
    },
    {
      "epoch": 0.7730988912532198,
      "grad_norm": 2.3742313385009766,
      "learning_rate": 6.899e-05,
      "loss": 0.5023,
      "step": 13806
    },
    {
      "epoch": 0.7731548885653489,
      "grad_norm": 1.337531328201294,
      "learning_rate": 6.8995e-05,
      "loss": 0.4598,
      "step": 13807
    },
    {
      "epoch": 0.7732108858774779,
      "grad_norm": 1.3381874561309814,
      "learning_rate": 6.9e-05,
      "loss": 0.452,
      "step": 13808
    },
    {
      "epoch": 0.7732668831896069,
      "grad_norm": 1.3037018775939941,
      "learning_rate": 6.900500000000001e-05,
      "loss": 0.3577,
      "step": 13809
    },
    {
      "epoch": 0.7733228805017359,
      "grad_norm": 1.2436840534210205,
      "learning_rate": 6.901e-05,
      "loss": 0.3838,
      "step": 13810
    },
    {
      "epoch": 0.7733788778138649,
      "grad_norm": 1.198750376701355,
      "learning_rate": 6.9015e-05,
      "loss": 0.3625,
      "step": 13811
    },
    {
      "epoch": 0.773434875125994,
      "grad_norm": 1.8446670770645142,
      "learning_rate": 6.902000000000001e-05,
      "loss": 0.4405,
      "step": 13812
    },
    {
      "epoch": 0.773490872438123,
      "grad_norm": 1.359084963798523,
      "learning_rate": 6.902500000000001e-05,
      "loss": 0.4634,
      "step": 13813
    },
    {
      "epoch": 0.773546869750252,
      "grad_norm": 1.36271333694458,
      "learning_rate": 6.903000000000001e-05,
      "loss": 0.5752,
      "step": 13814
    },
    {
      "epoch": 0.773602867062381,
      "grad_norm": 1.6614081859588623,
      "learning_rate": 6.9035e-05,
      "loss": 0.51,
      "step": 13815
    },
    {
      "epoch": 0.77365886437451,
      "grad_norm": 1.3628937005996704,
      "learning_rate": 6.904e-05,
      "loss": 0.4572,
      "step": 13816
    },
    {
      "epoch": 0.773714861686639,
      "grad_norm": 1.6391777992248535,
      "learning_rate": 6.9045e-05,
      "loss": 0.6078,
      "step": 13817
    },
    {
      "epoch": 0.7737708589987681,
      "grad_norm": 1.3059842586517334,
      "learning_rate": 6.905e-05,
      "loss": 0.3356,
      "step": 13818
    },
    {
      "epoch": 0.7738268563108971,
      "grad_norm": 1.2347959280014038,
      "learning_rate": 6.905500000000001e-05,
      "loss": 0.3631,
      "step": 13819
    },
    {
      "epoch": 0.7738828536230261,
      "grad_norm": 1.3058514595031738,
      "learning_rate": 6.906e-05,
      "loss": 0.4657,
      "step": 13820
    },
    {
      "epoch": 0.7739388509351551,
      "grad_norm": 1.529215693473816,
      "learning_rate": 6.9065e-05,
      "loss": 0.4599,
      "step": 13821
    },
    {
      "epoch": 0.7739948482472842,
      "grad_norm": 1.2192203998565674,
      "learning_rate": 6.907e-05,
      "loss": 0.4931,
      "step": 13822
    },
    {
      "epoch": 0.7740508455594132,
      "grad_norm": 1.276862621307373,
      "learning_rate": 6.9075e-05,
      "loss": 0.5643,
      "step": 13823
    },
    {
      "epoch": 0.7741068428715422,
      "grad_norm": 1.3896090984344482,
      "learning_rate": 6.908e-05,
      "loss": 0.4071,
      "step": 13824
    },
    {
      "epoch": 0.7741628401836712,
      "grad_norm": 1.2592624425888062,
      "learning_rate": 6.9085e-05,
      "loss": 0.4726,
      "step": 13825
    },
    {
      "epoch": 0.7742188374958002,
      "grad_norm": 1.6384234428405762,
      "learning_rate": 6.909e-05,
      "loss": 0.4709,
      "step": 13826
    },
    {
      "epoch": 0.7742748348079292,
      "grad_norm": 1.4587829113006592,
      "learning_rate": 6.9095e-05,
      "loss": 0.3398,
      "step": 13827
    },
    {
      "epoch": 0.7743308321200583,
      "grad_norm": 1.1166255474090576,
      "learning_rate": 6.91e-05,
      "loss": 0.402,
      "step": 13828
    },
    {
      "epoch": 0.7743868294321873,
      "grad_norm": 1.5124624967575073,
      "learning_rate": 6.910500000000001e-05,
      "loss": 0.4448,
      "step": 13829
    },
    {
      "epoch": 0.7744428267443163,
      "grad_norm": 1.1770051717758179,
      "learning_rate": 6.911000000000001e-05,
      "loss": 0.3488,
      "step": 13830
    },
    {
      "epoch": 0.7744988240564453,
      "grad_norm": 1.3130956888198853,
      "learning_rate": 6.9115e-05,
      "loss": 0.4856,
      "step": 13831
    },
    {
      "epoch": 0.7745548213685743,
      "grad_norm": 1.3943486213684082,
      "learning_rate": 6.912e-05,
      "loss": 0.4011,
      "step": 13832
    },
    {
      "epoch": 0.7746108186807034,
      "grad_norm": 1.4471852779388428,
      "learning_rate": 6.9125e-05,
      "loss": 0.3976,
      "step": 13833
    },
    {
      "epoch": 0.7746668159928324,
      "grad_norm": 1.2625128030776978,
      "learning_rate": 6.913000000000001e-05,
      "loss": 0.4263,
      "step": 13834
    },
    {
      "epoch": 0.7747228133049614,
      "grad_norm": 1.5544123649597168,
      "learning_rate": 6.913500000000001e-05,
      "loss": 0.4449,
      "step": 13835
    },
    {
      "epoch": 0.7747788106170904,
      "grad_norm": 1.125077247619629,
      "learning_rate": 6.914e-05,
      "loss": 0.4424,
      "step": 13836
    },
    {
      "epoch": 0.7748348079292194,
      "grad_norm": 1.201973557472229,
      "learning_rate": 6.9145e-05,
      "loss": 0.4998,
      "step": 13837
    },
    {
      "epoch": 0.7748908052413485,
      "grad_norm": 1.315532922744751,
      "learning_rate": 6.915e-05,
      "loss": 0.3769,
      "step": 13838
    },
    {
      "epoch": 0.7749468025534775,
      "grad_norm": 2.161365032196045,
      "learning_rate": 6.9155e-05,
      "loss": 0.4246,
      "step": 13839
    },
    {
      "epoch": 0.7750027998656065,
      "grad_norm": 1.1771349906921387,
      "learning_rate": 6.916000000000001e-05,
      "loss": 0.4635,
      "step": 13840
    },
    {
      "epoch": 0.7750587971777355,
      "grad_norm": 1.7514574527740479,
      "learning_rate": 6.9165e-05,
      "loss": 0.5846,
      "step": 13841
    },
    {
      "epoch": 0.7751147944898645,
      "grad_norm": 1.6608920097351074,
      "learning_rate": 6.917e-05,
      "loss": 0.5855,
      "step": 13842
    },
    {
      "epoch": 0.7751707918019936,
      "grad_norm": 1.6512091159820557,
      "learning_rate": 6.9175e-05,
      "loss": 0.5509,
      "step": 13843
    },
    {
      "epoch": 0.7752267891141226,
      "grad_norm": 1.455511450767517,
      "learning_rate": 6.918e-05,
      "loss": 0.4465,
      "step": 13844
    },
    {
      "epoch": 0.7752827864262516,
      "grad_norm": 1.0359824895858765,
      "learning_rate": 6.9185e-05,
      "loss": 0.3096,
      "step": 13845
    },
    {
      "epoch": 0.7753387837383806,
      "grad_norm": 1.3309375047683716,
      "learning_rate": 6.918999999999999e-05,
      "loss": 0.4897,
      "step": 13846
    },
    {
      "epoch": 0.7753947810505096,
      "grad_norm": 1.2151447534561157,
      "learning_rate": 6.9195e-05,
      "loss": 0.4117,
      "step": 13847
    },
    {
      "epoch": 0.7754507783626386,
      "grad_norm": 1.316588282585144,
      "learning_rate": 6.92e-05,
      "loss": 0.3778,
      "step": 13848
    },
    {
      "epoch": 0.7755067756747676,
      "grad_norm": 1.5426836013793945,
      "learning_rate": 6.920500000000001e-05,
      "loss": 0.5271,
      "step": 13849
    },
    {
      "epoch": 0.7755627729868966,
      "grad_norm": 1.4366511106491089,
      "learning_rate": 6.921000000000001e-05,
      "loss": 0.4839,
      "step": 13850
    },
    {
      "epoch": 0.7756187702990256,
      "grad_norm": 1.266235113143921,
      "learning_rate": 6.921500000000001e-05,
      "loss": 0.3653,
      "step": 13851
    },
    {
      "epoch": 0.7756747676111546,
      "grad_norm": 1.1202263832092285,
      "learning_rate": 6.922e-05,
      "loss": 0.3637,
      "step": 13852
    },
    {
      "epoch": 0.7757307649232836,
      "grad_norm": 1.158772349357605,
      "learning_rate": 6.9225e-05,
      "loss": 0.3914,
      "step": 13853
    },
    {
      "epoch": 0.7757867622354127,
      "grad_norm": 1.7575604915618896,
      "learning_rate": 6.923e-05,
      "loss": 0.4897,
      "step": 13854
    },
    {
      "epoch": 0.7758427595475417,
      "grad_norm": 1.211358904838562,
      "learning_rate": 6.923500000000001e-05,
      "loss": 0.4437,
      "step": 13855
    },
    {
      "epoch": 0.7758987568596707,
      "grad_norm": 1.5842890739440918,
      "learning_rate": 6.924000000000001e-05,
      "loss": 0.6314,
      "step": 13856
    },
    {
      "epoch": 0.7759547541717997,
      "grad_norm": 1.2781901359558105,
      "learning_rate": 6.9245e-05,
      "loss": 0.4226,
      "step": 13857
    },
    {
      "epoch": 0.7760107514839287,
      "grad_norm": 1.6134580373764038,
      "learning_rate": 6.925e-05,
      "loss": 0.4942,
      "step": 13858
    },
    {
      "epoch": 0.7760667487960577,
      "grad_norm": 1.2906938791275024,
      "learning_rate": 6.9255e-05,
      "loss": 0.3801,
      "step": 13859
    },
    {
      "epoch": 0.7761227461081868,
      "grad_norm": 1.284714937210083,
      "learning_rate": 6.926e-05,
      "loss": 0.5016,
      "step": 13860
    },
    {
      "epoch": 0.7761787434203158,
      "grad_norm": 1.2326958179473877,
      "learning_rate": 6.926500000000001e-05,
      "loss": 0.4594,
      "step": 13861
    },
    {
      "epoch": 0.7762347407324448,
      "grad_norm": 1.385058045387268,
      "learning_rate": 6.927e-05,
      "loss": 0.4893,
      "step": 13862
    },
    {
      "epoch": 0.7762907380445738,
      "grad_norm": 1.1791129112243652,
      "learning_rate": 6.9275e-05,
      "loss": 0.4252,
      "step": 13863
    },
    {
      "epoch": 0.7763467353567028,
      "grad_norm": 1.4034109115600586,
      "learning_rate": 6.928e-05,
      "loss": 0.4095,
      "step": 13864
    },
    {
      "epoch": 0.7764027326688319,
      "grad_norm": 1.4608080387115479,
      "learning_rate": 6.9285e-05,
      "loss": 0.6999,
      "step": 13865
    },
    {
      "epoch": 0.7764587299809609,
      "grad_norm": 1.1830004453659058,
      "learning_rate": 6.929e-05,
      "loss": 0.5181,
      "step": 13866
    },
    {
      "epoch": 0.7765147272930899,
      "grad_norm": 1.135304570198059,
      "learning_rate": 6.929499999999999e-05,
      "loss": 0.4598,
      "step": 13867
    },
    {
      "epoch": 0.7765707246052189,
      "grad_norm": 1.295250415802002,
      "learning_rate": 6.93e-05,
      "loss": 0.3736,
      "step": 13868
    },
    {
      "epoch": 0.7766267219173479,
      "grad_norm": 1.9412102699279785,
      "learning_rate": 6.930500000000001e-05,
      "loss": 0.5801,
      "step": 13869
    },
    {
      "epoch": 0.776682719229477,
      "grad_norm": 1.6821688413619995,
      "learning_rate": 6.931000000000001e-05,
      "loss": 0.4209,
      "step": 13870
    },
    {
      "epoch": 0.776738716541606,
      "grad_norm": 1.1794018745422363,
      "learning_rate": 6.931500000000001e-05,
      "loss": 0.3961,
      "step": 13871
    },
    {
      "epoch": 0.776794713853735,
      "grad_norm": 1.363509178161621,
      "learning_rate": 6.932000000000001e-05,
      "loss": 0.4314,
      "step": 13872
    },
    {
      "epoch": 0.776850711165864,
      "grad_norm": 1.2927194833755493,
      "learning_rate": 6.9325e-05,
      "loss": 0.3952,
      "step": 13873
    },
    {
      "epoch": 0.776906708477993,
      "grad_norm": 1.4138803482055664,
      "learning_rate": 6.933e-05,
      "loss": 0.4139,
      "step": 13874
    },
    {
      "epoch": 0.776962705790122,
      "grad_norm": 1.3323150873184204,
      "learning_rate": 6.9335e-05,
      "loss": 0.4429,
      "step": 13875
    },
    {
      "epoch": 0.7770187031022511,
      "grad_norm": 1.705682635307312,
      "learning_rate": 6.934000000000001e-05,
      "loss": 0.4408,
      "step": 13876
    },
    {
      "epoch": 0.7770747004143801,
      "grad_norm": 1.4556496143341064,
      "learning_rate": 6.934500000000001e-05,
      "loss": 0.3778,
      "step": 13877
    },
    {
      "epoch": 0.7771306977265091,
      "grad_norm": 1.2827378511428833,
      "learning_rate": 6.935e-05,
      "loss": 0.4104,
      "step": 13878
    },
    {
      "epoch": 0.7771866950386381,
      "grad_norm": 1.4275895357131958,
      "learning_rate": 6.9355e-05,
      "loss": 0.4003,
      "step": 13879
    },
    {
      "epoch": 0.7772426923507672,
      "grad_norm": 1.1352436542510986,
      "learning_rate": 6.936e-05,
      "loss": 0.3701,
      "step": 13880
    },
    {
      "epoch": 0.7772986896628962,
      "grad_norm": 1.2699083089828491,
      "learning_rate": 6.9365e-05,
      "loss": 0.4275,
      "step": 13881
    },
    {
      "epoch": 0.7773546869750252,
      "grad_norm": 1.5980900526046753,
      "learning_rate": 6.937000000000001e-05,
      "loss": 0.3992,
      "step": 13882
    },
    {
      "epoch": 0.7774106842871542,
      "grad_norm": 1.665728211402893,
      "learning_rate": 6.9375e-05,
      "loss": 0.4682,
      "step": 13883
    },
    {
      "epoch": 0.7774666815992832,
      "grad_norm": 1.3462138175964355,
      "learning_rate": 6.938e-05,
      "loss": 0.4193,
      "step": 13884
    },
    {
      "epoch": 0.7775226789114122,
      "grad_norm": 0.9974067211151123,
      "learning_rate": 6.9385e-05,
      "loss": 0.3148,
      "step": 13885
    },
    {
      "epoch": 0.7775786762235413,
      "grad_norm": 1.3148261308670044,
      "learning_rate": 6.939e-05,
      "loss": 0.4683,
      "step": 13886
    },
    {
      "epoch": 0.7776346735356703,
      "grad_norm": 1.506017804145813,
      "learning_rate": 6.9395e-05,
      "loss": 0.5044,
      "step": 13887
    },
    {
      "epoch": 0.7776906708477993,
      "grad_norm": 1.2350032329559326,
      "learning_rate": 6.939999999999999e-05,
      "loss": 0.4377,
      "step": 13888
    },
    {
      "epoch": 0.7777466681599283,
      "grad_norm": 1.1359614133834839,
      "learning_rate": 6.9405e-05,
      "loss": 0.4154,
      "step": 13889
    },
    {
      "epoch": 0.7778026654720573,
      "grad_norm": 1.454512119293213,
      "learning_rate": 6.941000000000001e-05,
      "loss": 0.673,
      "step": 13890
    },
    {
      "epoch": 0.7778586627841864,
      "grad_norm": 1.3806427717208862,
      "learning_rate": 6.941500000000001e-05,
      "loss": 0.4112,
      "step": 13891
    },
    {
      "epoch": 0.7779146600963154,
      "grad_norm": 1.360846757888794,
      "learning_rate": 6.942000000000001e-05,
      "loss": 0.3939,
      "step": 13892
    },
    {
      "epoch": 0.7779706574084444,
      "grad_norm": 1.238250970840454,
      "learning_rate": 6.942500000000001e-05,
      "loss": 0.3914,
      "step": 13893
    },
    {
      "epoch": 0.7780266547205734,
      "grad_norm": 4.463929176330566,
      "learning_rate": 6.943e-05,
      "loss": 0.4305,
      "step": 13894
    },
    {
      "epoch": 0.7780826520327024,
      "grad_norm": 1.5247230529785156,
      "learning_rate": 6.9435e-05,
      "loss": 0.5265,
      "step": 13895
    },
    {
      "epoch": 0.7781386493448315,
      "grad_norm": 1.4591853618621826,
      "learning_rate": 6.944e-05,
      "loss": 0.4177,
      "step": 13896
    },
    {
      "epoch": 0.7781946466569605,
      "grad_norm": 1.3938111066818237,
      "learning_rate": 6.944500000000001e-05,
      "loss": 0.3971,
      "step": 13897
    },
    {
      "epoch": 0.7782506439690895,
      "grad_norm": 1.2990190982818604,
      "learning_rate": 6.945000000000001e-05,
      "loss": 0.4432,
      "step": 13898
    },
    {
      "epoch": 0.7783066412812185,
      "grad_norm": 1.4695848226547241,
      "learning_rate": 6.9455e-05,
      "loss": 0.4576,
      "step": 13899
    },
    {
      "epoch": 0.7783626385933475,
      "grad_norm": 1.2041194438934326,
      "learning_rate": 6.946e-05,
      "loss": 0.4222,
      "step": 13900
    },
    {
      "epoch": 0.7784186359054766,
      "grad_norm": 1.2224990129470825,
      "learning_rate": 6.9465e-05,
      "loss": 0.5282,
      "step": 13901
    },
    {
      "epoch": 0.7784746332176056,
      "grad_norm": 1.1354583501815796,
      "learning_rate": 6.947e-05,
      "loss": 0.3981,
      "step": 13902
    },
    {
      "epoch": 0.7785306305297346,
      "grad_norm": 1.738417387008667,
      "learning_rate": 6.9475e-05,
      "loss": 0.6282,
      "step": 13903
    },
    {
      "epoch": 0.7785866278418636,
      "grad_norm": 1.2167282104492188,
      "learning_rate": 6.948e-05,
      "loss": 0.3819,
      "step": 13904
    },
    {
      "epoch": 0.7786426251539926,
      "grad_norm": 1.361893653869629,
      "learning_rate": 6.9485e-05,
      "loss": 0.4069,
      "step": 13905
    },
    {
      "epoch": 0.7786986224661216,
      "grad_norm": 1.4380555152893066,
      "learning_rate": 6.949e-05,
      "loss": 0.4445,
      "step": 13906
    },
    {
      "epoch": 0.7787546197782507,
      "grad_norm": 1.24867582321167,
      "learning_rate": 6.9495e-05,
      "loss": 0.4167,
      "step": 13907
    },
    {
      "epoch": 0.7788106170903797,
      "grad_norm": 1.3571693897247314,
      "learning_rate": 6.95e-05,
      "loss": 0.4471,
      "step": 13908
    },
    {
      "epoch": 0.7788666144025087,
      "grad_norm": 1.3829060792922974,
      "learning_rate": 6.950499999999999e-05,
      "loss": 0.3551,
      "step": 13909
    },
    {
      "epoch": 0.7789226117146377,
      "grad_norm": 1.2228182554244995,
      "learning_rate": 6.951e-05,
      "loss": 0.5215,
      "step": 13910
    },
    {
      "epoch": 0.7789786090267667,
      "grad_norm": 1.6797186136245728,
      "learning_rate": 6.951500000000001e-05,
      "loss": 0.4942,
      "step": 13911
    },
    {
      "epoch": 0.7790346063388958,
      "grad_norm": 1.561423420906067,
      "learning_rate": 6.952000000000001e-05,
      "loss": 0.5206,
      "step": 13912
    },
    {
      "epoch": 0.7790906036510248,
      "grad_norm": 1.3722482919692993,
      "learning_rate": 6.952500000000001e-05,
      "loss": 0.4192,
      "step": 13913
    },
    {
      "epoch": 0.7791466009631538,
      "grad_norm": 1.2754452228546143,
      "learning_rate": 6.953000000000001e-05,
      "loss": 0.5925,
      "step": 13914
    },
    {
      "epoch": 0.7792025982752828,
      "grad_norm": 1.2323298454284668,
      "learning_rate": 6.9535e-05,
      "loss": 0.4066,
      "step": 13915
    },
    {
      "epoch": 0.7792585955874118,
      "grad_norm": 1.2997041940689087,
      "learning_rate": 6.954e-05,
      "loss": 0.4002,
      "step": 13916
    },
    {
      "epoch": 0.7793145928995409,
      "grad_norm": 1.1204379796981812,
      "learning_rate": 6.9545e-05,
      "loss": 0.3701,
      "step": 13917
    },
    {
      "epoch": 0.7793705902116699,
      "grad_norm": 1.3768155574798584,
      "learning_rate": 6.955000000000001e-05,
      "loss": 0.3913,
      "step": 13918
    },
    {
      "epoch": 0.7794265875237989,
      "grad_norm": 1.28208327293396,
      "learning_rate": 6.955500000000001e-05,
      "loss": 0.4146,
      "step": 13919
    },
    {
      "epoch": 0.7794825848359279,
      "grad_norm": 1.2160152196884155,
      "learning_rate": 6.956e-05,
      "loss": 0.397,
      "step": 13920
    },
    {
      "epoch": 0.7795385821480569,
      "grad_norm": 1.4384526014328003,
      "learning_rate": 6.9565e-05,
      "loss": 0.5679,
      "step": 13921
    },
    {
      "epoch": 0.779594579460186,
      "grad_norm": 1.3922152519226074,
      "learning_rate": 6.957e-05,
      "loss": 0.4086,
      "step": 13922
    },
    {
      "epoch": 0.779650576772315,
      "grad_norm": 1.5517241954803467,
      "learning_rate": 6.9575e-05,
      "loss": 0.4071,
      "step": 13923
    },
    {
      "epoch": 0.779706574084444,
      "grad_norm": 1.2017390727996826,
      "learning_rate": 6.958e-05,
      "loss": 0.4027,
      "step": 13924
    },
    {
      "epoch": 0.779762571396573,
      "grad_norm": 1.3809328079223633,
      "learning_rate": 6.9585e-05,
      "loss": 0.4484,
      "step": 13925
    },
    {
      "epoch": 0.779818568708702,
      "grad_norm": 1.2521413564682007,
      "learning_rate": 6.959e-05,
      "loss": 0.3402,
      "step": 13926
    },
    {
      "epoch": 0.779874566020831,
      "grad_norm": 1.704215407371521,
      "learning_rate": 6.9595e-05,
      "loss": 0.4422,
      "step": 13927
    },
    {
      "epoch": 0.7799305633329601,
      "grad_norm": 1.3313181400299072,
      "learning_rate": 6.96e-05,
      "loss": 0.4553,
      "step": 13928
    },
    {
      "epoch": 0.7799865606450891,
      "grad_norm": 1.4314274787902832,
      "learning_rate": 6.9605e-05,
      "loss": 0.3606,
      "step": 13929
    },
    {
      "epoch": 0.7800425579572181,
      "grad_norm": 1.0212676525115967,
      "learning_rate": 6.961e-05,
      "loss": 0.3565,
      "step": 13930
    },
    {
      "epoch": 0.780098555269347,
      "grad_norm": 1.203959345817566,
      "learning_rate": 6.9615e-05,
      "loss": 0.4249,
      "step": 13931
    },
    {
      "epoch": 0.780154552581476,
      "grad_norm": 1.0592491626739502,
      "learning_rate": 6.962e-05,
      "loss": 0.423,
      "step": 13932
    },
    {
      "epoch": 0.780210549893605,
      "grad_norm": 1.5868914127349854,
      "learning_rate": 6.962500000000001e-05,
      "loss": 0.348,
      "step": 13933
    },
    {
      "epoch": 0.7802665472057341,
      "grad_norm": 5.164170742034912,
      "learning_rate": 6.963000000000001e-05,
      "loss": 0.4902,
      "step": 13934
    },
    {
      "epoch": 0.7803225445178631,
      "grad_norm": 1.4760080575942993,
      "learning_rate": 6.9635e-05,
      "loss": 0.5056,
      "step": 13935
    },
    {
      "epoch": 0.7803785418299921,
      "grad_norm": 1.3616795539855957,
      "learning_rate": 6.964e-05,
      "loss": 0.372,
      "step": 13936
    },
    {
      "epoch": 0.7804345391421211,
      "grad_norm": 1.3553009033203125,
      "learning_rate": 6.9645e-05,
      "loss": 0.4697,
      "step": 13937
    },
    {
      "epoch": 0.7804905364542502,
      "grad_norm": 1.2548644542694092,
      "learning_rate": 6.965e-05,
      "loss": 0.4852,
      "step": 13938
    },
    {
      "epoch": 0.7805465337663792,
      "grad_norm": 1.1524333953857422,
      "learning_rate": 6.965500000000001e-05,
      "loss": 0.4164,
      "step": 13939
    },
    {
      "epoch": 0.7806025310785082,
      "grad_norm": 1.3469394445419312,
      "learning_rate": 6.966000000000001e-05,
      "loss": 0.3957,
      "step": 13940
    },
    {
      "epoch": 0.7806585283906372,
      "grad_norm": 1.2270981073379517,
      "learning_rate": 6.9665e-05,
      "loss": 0.4294,
      "step": 13941
    },
    {
      "epoch": 0.7807145257027662,
      "grad_norm": 1.1579169034957886,
      "learning_rate": 6.967e-05,
      "loss": 0.3281,
      "step": 13942
    },
    {
      "epoch": 0.7807705230148952,
      "grad_norm": 1.1557217836380005,
      "learning_rate": 6.9675e-05,
      "loss": 0.3436,
      "step": 13943
    },
    {
      "epoch": 0.7808265203270243,
      "grad_norm": 1.1874650716781616,
      "learning_rate": 6.968e-05,
      "loss": 0.3193,
      "step": 13944
    },
    {
      "epoch": 0.7808825176391533,
      "grad_norm": 1.2706297636032104,
      "learning_rate": 6.9685e-05,
      "loss": 0.5212,
      "step": 13945
    },
    {
      "epoch": 0.7809385149512823,
      "grad_norm": 1.314117431640625,
      "learning_rate": 6.969e-05,
      "loss": 0.4,
      "step": 13946
    },
    {
      "epoch": 0.7809945122634113,
      "grad_norm": 1.414960265159607,
      "learning_rate": 6.9695e-05,
      "loss": 0.4314,
      "step": 13947
    },
    {
      "epoch": 0.7810505095755403,
      "grad_norm": 1.3312770128250122,
      "learning_rate": 6.97e-05,
      "loss": 0.3932,
      "step": 13948
    },
    {
      "epoch": 0.7811065068876694,
      "grad_norm": 1.2104490995407104,
      "learning_rate": 6.9705e-05,
      "loss": 0.371,
      "step": 13949
    },
    {
      "epoch": 0.7811625041997984,
      "grad_norm": 1.268545389175415,
      "learning_rate": 6.971000000000001e-05,
      "loss": 0.4738,
      "step": 13950
    },
    {
      "epoch": 0.7812185015119274,
      "grad_norm": 1.1525225639343262,
      "learning_rate": 6.9715e-05,
      "loss": 0.4031,
      "step": 13951
    },
    {
      "epoch": 0.7812744988240564,
      "grad_norm": 1.3071669340133667,
      "learning_rate": 6.972e-05,
      "loss": 0.401,
      "step": 13952
    },
    {
      "epoch": 0.7813304961361854,
      "grad_norm": 1.194106936454773,
      "learning_rate": 6.9725e-05,
      "loss": 0.3375,
      "step": 13953
    },
    {
      "epoch": 0.7813864934483145,
      "grad_norm": 1.4018372297286987,
      "learning_rate": 6.973000000000001e-05,
      "loss": 0.4572,
      "step": 13954
    },
    {
      "epoch": 0.7814424907604435,
      "grad_norm": 1.109297275543213,
      "learning_rate": 6.973500000000001e-05,
      "loss": 0.4132,
      "step": 13955
    },
    {
      "epoch": 0.7814984880725725,
      "grad_norm": 1.2089378833770752,
      "learning_rate": 6.974e-05,
      "loss": 0.5602,
      "step": 13956
    },
    {
      "epoch": 0.7815544853847015,
      "grad_norm": 1.5532406568527222,
      "learning_rate": 6.9745e-05,
      "loss": 0.471,
      "step": 13957
    },
    {
      "epoch": 0.7816104826968305,
      "grad_norm": 1.2478727102279663,
      "learning_rate": 6.975e-05,
      "loss": 0.4428,
      "step": 13958
    },
    {
      "epoch": 0.7816664800089596,
      "grad_norm": 1.18569815158844,
      "learning_rate": 6.9755e-05,
      "loss": 0.3875,
      "step": 13959
    },
    {
      "epoch": 0.7817224773210886,
      "grad_norm": 1.3408461809158325,
      "learning_rate": 6.976000000000001e-05,
      "loss": 0.5038,
      "step": 13960
    },
    {
      "epoch": 0.7817784746332176,
      "grad_norm": 1.3232312202453613,
      "learning_rate": 6.976500000000001e-05,
      "loss": 0.4234,
      "step": 13961
    },
    {
      "epoch": 0.7818344719453466,
      "grad_norm": 1.516841173171997,
      "learning_rate": 6.977e-05,
      "loss": 0.6285,
      "step": 13962
    },
    {
      "epoch": 0.7818904692574756,
      "grad_norm": 1.223027229309082,
      "learning_rate": 6.9775e-05,
      "loss": 0.4015,
      "step": 13963
    },
    {
      "epoch": 0.7819464665696046,
      "grad_norm": 1.454139232635498,
      "learning_rate": 6.978e-05,
      "loss": 0.5237,
      "step": 13964
    },
    {
      "epoch": 0.7820024638817337,
      "grad_norm": 1.4948914051055908,
      "learning_rate": 6.9785e-05,
      "loss": 0.4537,
      "step": 13965
    },
    {
      "epoch": 0.7820584611938627,
      "grad_norm": 1.4777477979660034,
      "learning_rate": 6.979e-05,
      "loss": 0.4717,
      "step": 13966
    },
    {
      "epoch": 0.7821144585059917,
      "grad_norm": 1.4510791301727295,
      "learning_rate": 6.9795e-05,
      "loss": 0.5227,
      "step": 13967
    },
    {
      "epoch": 0.7821704558181207,
      "grad_norm": 1.7112095355987549,
      "learning_rate": 6.98e-05,
      "loss": 0.4873,
      "step": 13968
    },
    {
      "epoch": 0.7822264531302497,
      "grad_norm": 1.3013123273849487,
      "learning_rate": 6.9805e-05,
      "loss": 0.352,
      "step": 13969
    },
    {
      "epoch": 0.7822824504423788,
      "grad_norm": 1.1731970310211182,
      "learning_rate": 6.981000000000001e-05,
      "loss": 0.4815,
      "step": 13970
    },
    {
      "epoch": 0.7823384477545078,
      "grad_norm": 1.0791093111038208,
      "learning_rate": 6.981500000000001e-05,
      "loss": 0.3778,
      "step": 13971
    },
    {
      "epoch": 0.7823944450666368,
      "grad_norm": 1.2413989305496216,
      "learning_rate": 6.982e-05,
      "loss": 0.3655,
      "step": 13972
    },
    {
      "epoch": 0.7824504423787658,
      "grad_norm": 1.4981213808059692,
      "learning_rate": 6.9825e-05,
      "loss": 0.3967,
      "step": 13973
    },
    {
      "epoch": 0.7825064396908948,
      "grad_norm": 1.2649314403533936,
      "learning_rate": 6.983e-05,
      "loss": 0.4468,
      "step": 13974
    },
    {
      "epoch": 0.7825624370030239,
      "grad_norm": 1.3142908811569214,
      "learning_rate": 6.983500000000001e-05,
      "loss": 0.4962,
      "step": 13975
    },
    {
      "epoch": 0.7826184343151529,
      "grad_norm": 1.1367021799087524,
      "learning_rate": 6.984000000000001e-05,
      "loss": 0.3033,
      "step": 13976
    },
    {
      "epoch": 0.7826744316272819,
      "grad_norm": 1.2410575151443481,
      "learning_rate": 6.9845e-05,
      "loss": 0.4894,
      "step": 13977
    },
    {
      "epoch": 0.7827304289394109,
      "grad_norm": 1.424251675605774,
      "learning_rate": 6.985e-05,
      "loss": 0.365,
      "step": 13978
    },
    {
      "epoch": 0.7827864262515399,
      "grad_norm": 1.9061392545700073,
      "learning_rate": 6.9855e-05,
      "loss": 0.5065,
      "step": 13979
    },
    {
      "epoch": 0.782842423563669,
      "grad_norm": 1.4891753196716309,
      "learning_rate": 6.986e-05,
      "loss": 0.4166,
      "step": 13980
    },
    {
      "epoch": 0.782898420875798,
      "grad_norm": 1.5439541339874268,
      "learning_rate": 6.9865e-05,
      "loss": 0.4611,
      "step": 13981
    },
    {
      "epoch": 0.782954418187927,
      "grad_norm": 1.2894482612609863,
      "learning_rate": 6.987000000000001e-05,
      "loss": 0.4427,
      "step": 13982
    },
    {
      "epoch": 0.783010415500056,
      "grad_norm": 1.278134822845459,
      "learning_rate": 6.9875e-05,
      "loss": 0.5133,
      "step": 13983
    },
    {
      "epoch": 0.783066412812185,
      "grad_norm": 1.510360598564148,
      "learning_rate": 6.988e-05,
      "loss": 0.4293,
      "step": 13984
    },
    {
      "epoch": 0.783122410124314,
      "grad_norm": 1.4213038682937622,
      "learning_rate": 6.9885e-05,
      "loss": 0.4431,
      "step": 13985
    },
    {
      "epoch": 0.7831784074364431,
      "grad_norm": 1.2273017168045044,
      "learning_rate": 6.989e-05,
      "loss": 0.4495,
      "step": 13986
    },
    {
      "epoch": 0.7832344047485721,
      "grad_norm": 1.3446868658065796,
      "learning_rate": 6.9895e-05,
      "loss": 0.4086,
      "step": 13987
    },
    {
      "epoch": 0.7832904020607011,
      "grad_norm": 1.4655638933181763,
      "learning_rate": 6.99e-05,
      "loss": 0.4838,
      "step": 13988
    },
    {
      "epoch": 0.7833463993728301,
      "grad_norm": 1.3987774848937988,
      "learning_rate": 6.9905e-05,
      "loss": 0.493,
      "step": 13989
    },
    {
      "epoch": 0.7834023966849591,
      "grad_norm": 1.6490641832351685,
      "learning_rate": 6.991000000000001e-05,
      "loss": 0.5507,
      "step": 13990
    },
    {
      "epoch": 0.7834583939970882,
      "grad_norm": 1.5870490074157715,
      "learning_rate": 6.991500000000001e-05,
      "loss": 0.5309,
      "step": 13991
    },
    {
      "epoch": 0.7835143913092172,
      "grad_norm": 1.0920445919036865,
      "learning_rate": 6.992000000000001e-05,
      "loss": 0.4331,
      "step": 13992
    },
    {
      "epoch": 0.7835703886213462,
      "grad_norm": 1.8054163455963135,
      "learning_rate": 6.9925e-05,
      "loss": 0.3573,
      "step": 13993
    },
    {
      "epoch": 0.7836263859334752,
      "grad_norm": 1.3015691041946411,
      "learning_rate": 6.993e-05,
      "loss": 0.5052,
      "step": 13994
    },
    {
      "epoch": 0.7836823832456042,
      "grad_norm": 1.448464035987854,
      "learning_rate": 6.9935e-05,
      "loss": 0.3574,
      "step": 13995
    },
    {
      "epoch": 0.7837383805577333,
      "grad_norm": 1.2324483394622803,
      "learning_rate": 6.994000000000001e-05,
      "loss": 0.4016,
      "step": 13996
    },
    {
      "epoch": 0.7837943778698623,
      "grad_norm": 1.290958046913147,
      "learning_rate": 6.994500000000001e-05,
      "loss": 0.4793,
      "step": 13997
    },
    {
      "epoch": 0.7838503751819913,
      "grad_norm": 1.1485828161239624,
      "learning_rate": 6.995e-05,
      "loss": 0.3982,
      "step": 13998
    },
    {
      "epoch": 0.7839063724941203,
      "grad_norm": 1.4403445720672607,
      "learning_rate": 6.9955e-05,
      "loss": 0.3792,
      "step": 13999
    },
    {
      "epoch": 0.7839623698062493,
      "grad_norm": 1.1863340139389038,
      "learning_rate": 6.996e-05,
      "loss": 0.4099,
      "step": 14000
    },
    {
      "epoch": 0.7840183671183784,
      "grad_norm": 1.294132113456726,
      "learning_rate": 6.9965e-05,
      "loss": 0.4635,
      "step": 14001
    },
    {
      "epoch": 0.7840743644305074,
      "grad_norm": 1.403907060623169,
      "learning_rate": 6.997e-05,
      "loss": 0.4105,
      "step": 14002
    },
    {
      "epoch": 0.7841303617426364,
      "grad_norm": 1.434434413909912,
      "learning_rate": 6.997500000000001e-05,
      "loss": 0.4805,
      "step": 14003
    },
    {
      "epoch": 0.7841863590547654,
      "grad_norm": 1.1685349941253662,
      "learning_rate": 6.998e-05,
      "loss": 0.5337,
      "step": 14004
    },
    {
      "epoch": 0.7842423563668944,
      "grad_norm": 2.138056516647339,
      "learning_rate": 6.9985e-05,
      "loss": 0.4324,
      "step": 14005
    },
    {
      "epoch": 0.7842983536790235,
      "grad_norm": 1.5927270650863647,
      "learning_rate": 6.999e-05,
      "loss": 0.4646,
      "step": 14006
    },
    {
      "epoch": 0.7843543509911525,
      "grad_norm": 2.062490463256836,
      "learning_rate": 6.9995e-05,
      "loss": 0.6349,
      "step": 14007
    },
    {
      "epoch": 0.7844103483032815,
      "grad_norm": 1.1932083368301392,
      "learning_rate": 7e-05,
      "loss": 0.3989,
      "step": 14008
    },
    {
      "epoch": 0.7844663456154105,
      "grad_norm": 1.3754913806915283,
      "learning_rate": 7.0005e-05,
      "loss": 0.4343,
      "step": 14009
    },
    {
      "epoch": 0.7845223429275395,
      "grad_norm": 1.5349290370941162,
      "learning_rate": 7.001e-05,
      "loss": 0.4089,
      "step": 14010
    },
    {
      "epoch": 0.7845783402396685,
      "grad_norm": 1.2927325963974,
      "learning_rate": 7.001500000000001e-05,
      "loss": 0.5192,
      "step": 14011
    },
    {
      "epoch": 0.7846343375517976,
      "grad_norm": 2.542348861694336,
      "learning_rate": 7.002000000000001e-05,
      "loss": 0.5494,
      "step": 14012
    },
    {
      "epoch": 0.7846903348639266,
      "grad_norm": 1.211527943611145,
      "learning_rate": 7.002500000000001e-05,
      "loss": 0.3993,
      "step": 14013
    },
    {
      "epoch": 0.7847463321760555,
      "grad_norm": 1.3985072374343872,
      "learning_rate": 7.003e-05,
      "loss": 0.4081,
      "step": 14014
    },
    {
      "epoch": 0.7848023294881845,
      "grad_norm": 1.479193091392517,
      "learning_rate": 7.0035e-05,
      "loss": 0.569,
      "step": 14015
    },
    {
      "epoch": 0.7848583268003135,
      "grad_norm": 1.100593090057373,
      "learning_rate": 7.004e-05,
      "loss": 0.3533,
      "step": 14016
    },
    {
      "epoch": 0.7849143241124426,
      "grad_norm": 1.3223495483398438,
      "learning_rate": 7.004500000000001e-05,
      "loss": 0.3749,
      "step": 14017
    },
    {
      "epoch": 0.7849703214245716,
      "grad_norm": 1.162502408027649,
      "learning_rate": 7.005000000000001e-05,
      "loss": 0.4615,
      "step": 14018
    },
    {
      "epoch": 0.7850263187367006,
      "grad_norm": 1.402768611907959,
      "learning_rate": 7.0055e-05,
      "loss": 0.5245,
      "step": 14019
    },
    {
      "epoch": 0.7850823160488296,
      "grad_norm": 1.5840893983840942,
      "learning_rate": 7.006e-05,
      "loss": 0.5788,
      "step": 14020
    },
    {
      "epoch": 0.7851383133609586,
      "grad_norm": 1.3359031677246094,
      "learning_rate": 7.0065e-05,
      "loss": 0.4687,
      "step": 14021
    },
    {
      "epoch": 0.7851943106730876,
      "grad_norm": 1.5367978811264038,
      "learning_rate": 7.007e-05,
      "loss": 0.5541,
      "step": 14022
    },
    {
      "epoch": 0.7852503079852167,
      "grad_norm": 1.2812166213989258,
      "learning_rate": 7.0075e-05,
      "loss": 0.4377,
      "step": 14023
    },
    {
      "epoch": 0.7853063052973457,
      "grad_norm": 1.3251447677612305,
      "learning_rate": 7.008e-05,
      "loss": 0.405,
      "step": 14024
    },
    {
      "epoch": 0.7853623026094747,
      "grad_norm": 1.500702142715454,
      "learning_rate": 7.0085e-05,
      "loss": 0.5783,
      "step": 14025
    },
    {
      "epoch": 0.7854182999216037,
      "grad_norm": 1.5428801774978638,
      "learning_rate": 7.009e-05,
      "loss": 0.588,
      "step": 14026
    },
    {
      "epoch": 0.7854742972337327,
      "grad_norm": 1.34929358959198,
      "learning_rate": 7.0095e-05,
      "loss": 0.5225,
      "step": 14027
    },
    {
      "epoch": 0.7855302945458618,
      "grad_norm": 1.4423037767410278,
      "learning_rate": 7.01e-05,
      "loss": 0.522,
      "step": 14028
    },
    {
      "epoch": 0.7855862918579908,
      "grad_norm": 1.3482623100280762,
      "learning_rate": 7.0105e-05,
      "loss": 0.5341,
      "step": 14029
    },
    {
      "epoch": 0.7856422891701198,
      "grad_norm": 1.7001186609268188,
      "learning_rate": 7.011e-05,
      "loss": 0.5013,
      "step": 14030
    },
    {
      "epoch": 0.7856982864822488,
      "grad_norm": 1.2402490377426147,
      "learning_rate": 7.0115e-05,
      "loss": 0.5603,
      "step": 14031
    },
    {
      "epoch": 0.7857542837943778,
      "grad_norm": 1.2932721376419067,
      "learning_rate": 7.012000000000001e-05,
      "loss": 0.4235,
      "step": 14032
    },
    {
      "epoch": 0.7858102811065069,
      "grad_norm": 1.446073293685913,
      "learning_rate": 7.012500000000001e-05,
      "loss": 0.4654,
      "step": 14033
    },
    {
      "epoch": 0.7858662784186359,
      "grad_norm": 1.7081900835037231,
      "learning_rate": 7.013000000000001e-05,
      "loss": 0.4237,
      "step": 14034
    },
    {
      "epoch": 0.7859222757307649,
      "grad_norm": 1.3166462182998657,
      "learning_rate": 7.0135e-05,
      "loss": 0.4698,
      "step": 14035
    },
    {
      "epoch": 0.7859782730428939,
      "grad_norm": 1.2614250183105469,
      "learning_rate": 7.014e-05,
      "loss": 0.4302,
      "step": 14036
    },
    {
      "epoch": 0.7860342703550229,
      "grad_norm": 1.2707799673080444,
      "learning_rate": 7.0145e-05,
      "loss": 0.4416,
      "step": 14037
    },
    {
      "epoch": 0.786090267667152,
      "grad_norm": 1.3705872297286987,
      "learning_rate": 7.015000000000001e-05,
      "loss": 0.429,
      "step": 14038
    },
    {
      "epoch": 0.786146264979281,
      "grad_norm": 1.3415343761444092,
      "learning_rate": 7.015500000000001e-05,
      "loss": 0.5315,
      "step": 14039
    },
    {
      "epoch": 0.78620226229141,
      "grad_norm": 1.2904845476150513,
      "learning_rate": 7.016e-05,
      "loss": 0.4053,
      "step": 14040
    },
    {
      "epoch": 0.786258259603539,
      "grad_norm": 1.431028127670288,
      "learning_rate": 7.0165e-05,
      "loss": 0.5252,
      "step": 14041
    },
    {
      "epoch": 0.786314256915668,
      "grad_norm": 9.211777687072754,
      "learning_rate": 7.017e-05,
      "loss": 0.4757,
      "step": 14042
    },
    {
      "epoch": 0.786370254227797,
      "grad_norm": 1.9496930837631226,
      "learning_rate": 7.0175e-05,
      "loss": 0.6311,
      "step": 14043
    },
    {
      "epoch": 0.7864262515399261,
      "grad_norm": 1.450932502746582,
      "learning_rate": 7.018e-05,
      "loss": 0.712,
      "step": 14044
    },
    {
      "epoch": 0.7864822488520551,
      "grad_norm": 1.3744438886642456,
      "learning_rate": 7.0185e-05,
      "loss": 0.4701,
      "step": 14045
    },
    {
      "epoch": 0.7865382461641841,
      "grad_norm": 1.468982219696045,
      "learning_rate": 7.019e-05,
      "loss": 0.508,
      "step": 14046
    },
    {
      "epoch": 0.7865942434763131,
      "grad_norm": 1.406665563583374,
      "learning_rate": 7.0195e-05,
      "loss": 0.4218,
      "step": 14047
    },
    {
      "epoch": 0.7866502407884421,
      "grad_norm": 1.1718058586120605,
      "learning_rate": 7.02e-05,
      "loss": 0.3639,
      "step": 14048
    },
    {
      "epoch": 0.7867062381005712,
      "grad_norm": 1.7525266408920288,
      "learning_rate": 7.0205e-05,
      "loss": 0.4124,
      "step": 14049
    },
    {
      "epoch": 0.7867622354127002,
      "grad_norm": 1.2831170558929443,
      "learning_rate": 7.021e-05,
      "loss": 0.4299,
      "step": 14050
    },
    {
      "epoch": 0.7868182327248292,
      "grad_norm": 1.143329381942749,
      "learning_rate": 7.0215e-05,
      "loss": 0.386,
      "step": 14051
    },
    {
      "epoch": 0.7868742300369582,
      "grad_norm": 1.5027421712875366,
      "learning_rate": 7.022e-05,
      "loss": 0.4731,
      "step": 14052
    },
    {
      "epoch": 0.7869302273490872,
      "grad_norm": 1.4526931047439575,
      "learning_rate": 7.022500000000001e-05,
      "loss": 0.5402,
      "step": 14053
    },
    {
      "epoch": 0.7869862246612163,
      "grad_norm": 1.1389609575271606,
      "learning_rate": 7.023000000000001e-05,
      "loss": 0.3371,
      "step": 14054
    },
    {
      "epoch": 0.7870422219733453,
      "grad_norm": 1.5279737710952759,
      "learning_rate": 7.023500000000001e-05,
      "loss": 0.408,
      "step": 14055
    },
    {
      "epoch": 0.7870982192854743,
      "grad_norm": 1.5773791074752808,
      "learning_rate": 7.024e-05,
      "loss": 0.5432,
      "step": 14056
    },
    {
      "epoch": 0.7871542165976033,
      "grad_norm": 2.4261014461517334,
      "learning_rate": 7.0245e-05,
      "loss": 0.471,
      "step": 14057
    },
    {
      "epoch": 0.7872102139097323,
      "grad_norm": 1.2080491781234741,
      "learning_rate": 7.025e-05,
      "loss": 0.3309,
      "step": 14058
    },
    {
      "epoch": 0.7872662112218614,
      "grad_norm": 1.195871353149414,
      "learning_rate": 7.025500000000001e-05,
      "loss": 0.4542,
      "step": 14059
    },
    {
      "epoch": 0.7873222085339904,
      "grad_norm": 1.3698365688323975,
      "learning_rate": 7.026000000000001e-05,
      "loss": 0.5356,
      "step": 14060
    },
    {
      "epoch": 0.7873782058461194,
      "grad_norm": 1.4013510942459106,
      "learning_rate": 7.0265e-05,
      "loss": 0.5224,
      "step": 14061
    },
    {
      "epoch": 0.7874342031582484,
      "grad_norm": 1.2595869302749634,
      "learning_rate": 7.027e-05,
      "loss": 0.5053,
      "step": 14062
    },
    {
      "epoch": 0.7874902004703774,
      "grad_norm": 1.0256685018539429,
      "learning_rate": 7.0275e-05,
      "loss": 0.4091,
      "step": 14063
    },
    {
      "epoch": 0.7875461977825065,
      "grad_norm": 1.2182501554489136,
      "learning_rate": 7.028e-05,
      "loss": 0.4386,
      "step": 14064
    },
    {
      "epoch": 0.7876021950946355,
      "grad_norm": 1.3985791206359863,
      "learning_rate": 7.0285e-05,
      "loss": 0.4712,
      "step": 14065
    },
    {
      "epoch": 0.7876581924067645,
      "grad_norm": 1.2584521770477295,
      "learning_rate": 7.029e-05,
      "loss": 0.4653,
      "step": 14066
    },
    {
      "epoch": 0.7877141897188935,
      "grad_norm": 1.3298203945159912,
      "learning_rate": 7.0295e-05,
      "loss": 0.576,
      "step": 14067
    },
    {
      "epoch": 0.7877701870310225,
      "grad_norm": 1.344087839126587,
      "learning_rate": 7.03e-05,
      "loss": 0.442,
      "step": 14068
    },
    {
      "epoch": 0.7878261843431515,
      "grad_norm": 1.3564358949661255,
      "learning_rate": 7.0305e-05,
      "loss": 0.449,
      "step": 14069
    },
    {
      "epoch": 0.7878821816552806,
      "grad_norm": 1.4164124727249146,
      "learning_rate": 7.031e-05,
      "loss": 0.4238,
      "step": 14070
    },
    {
      "epoch": 0.7879381789674096,
      "grad_norm": 1.408128261566162,
      "learning_rate": 7.031500000000001e-05,
      "loss": 0.4914,
      "step": 14071
    },
    {
      "epoch": 0.7879941762795386,
      "grad_norm": 1.1966450214385986,
      "learning_rate": 7.032e-05,
      "loss": 0.3067,
      "step": 14072
    },
    {
      "epoch": 0.7880501735916676,
      "grad_norm": 1.1928439140319824,
      "learning_rate": 7.0325e-05,
      "loss": 0.3834,
      "step": 14073
    },
    {
      "epoch": 0.7881061709037966,
      "grad_norm": 1.3703701496124268,
      "learning_rate": 7.033000000000001e-05,
      "loss": 0.5714,
      "step": 14074
    },
    {
      "epoch": 0.7881621682159257,
      "grad_norm": 1.4749839305877686,
      "learning_rate": 7.033500000000001e-05,
      "loss": 0.4225,
      "step": 14075
    },
    {
      "epoch": 0.7882181655280547,
      "grad_norm": 1.267511010169983,
      "learning_rate": 7.034000000000001e-05,
      "loss": 0.511,
      "step": 14076
    },
    {
      "epoch": 0.7882741628401837,
      "grad_norm": 1.5715607404708862,
      "learning_rate": 7.0345e-05,
      "loss": 0.4602,
      "step": 14077
    },
    {
      "epoch": 0.7883301601523127,
      "grad_norm": 1.459873080253601,
      "learning_rate": 7.035e-05,
      "loss": 0.578,
      "step": 14078
    },
    {
      "epoch": 0.7883861574644417,
      "grad_norm": 1.2043129205703735,
      "learning_rate": 7.0355e-05,
      "loss": 0.3964,
      "step": 14079
    },
    {
      "epoch": 0.7884421547765708,
      "grad_norm": 1.171628713607788,
      "learning_rate": 7.036e-05,
      "loss": 0.4532,
      "step": 14080
    },
    {
      "epoch": 0.7884981520886998,
      "grad_norm": 1.5888665914535522,
      "learning_rate": 7.036500000000001e-05,
      "loss": 0.4802,
      "step": 14081
    },
    {
      "epoch": 0.7885541494008288,
      "grad_norm": 1.1802852153778076,
      "learning_rate": 7.037e-05,
      "loss": 0.3877,
      "step": 14082
    },
    {
      "epoch": 0.7886101467129578,
      "grad_norm": 1.3492436408996582,
      "learning_rate": 7.0375e-05,
      "loss": 0.4503,
      "step": 14083
    },
    {
      "epoch": 0.7886661440250868,
      "grad_norm": 3.586104393005371,
      "learning_rate": 7.038e-05,
      "loss": 0.3421,
      "step": 14084
    },
    {
      "epoch": 0.7887221413372159,
      "grad_norm": 1.3017324209213257,
      "learning_rate": 7.0385e-05,
      "loss": 0.367,
      "step": 14085
    },
    {
      "epoch": 0.7887781386493449,
      "grad_norm": 1.4720596075057983,
      "learning_rate": 7.039e-05,
      "loss": 0.47,
      "step": 14086
    },
    {
      "epoch": 0.7888341359614739,
      "grad_norm": 1.2844620943069458,
      "learning_rate": 7.0395e-05,
      "loss": 0.3905,
      "step": 14087
    },
    {
      "epoch": 0.7888901332736029,
      "grad_norm": 1.5005204677581787,
      "learning_rate": 7.04e-05,
      "loss": 0.6698,
      "step": 14088
    },
    {
      "epoch": 0.7889461305857319,
      "grad_norm": 1.1908706426620483,
      "learning_rate": 7.0405e-05,
      "loss": 0.4621,
      "step": 14089
    },
    {
      "epoch": 0.789002127897861,
      "grad_norm": 1.265588641166687,
      "learning_rate": 7.041e-05,
      "loss": 0.422,
      "step": 14090
    },
    {
      "epoch": 0.78905812520999,
      "grad_norm": 1.56885826587677,
      "learning_rate": 7.041500000000001e-05,
      "loss": 0.4852,
      "step": 14091
    },
    {
      "epoch": 0.789114122522119,
      "grad_norm": 1.1783219575881958,
      "learning_rate": 7.042000000000001e-05,
      "loss": 0.4621,
      "step": 14092
    },
    {
      "epoch": 0.789170119834248,
      "grad_norm": 1.3888453245162964,
      "learning_rate": 7.0425e-05,
      "loss": 0.3956,
      "step": 14093
    },
    {
      "epoch": 0.789226117146377,
      "grad_norm": 1.3621768951416016,
      "learning_rate": 7.043e-05,
      "loss": 0.6413,
      "step": 14094
    },
    {
      "epoch": 0.789282114458506,
      "grad_norm": 1.1520731449127197,
      "learning_rate": 7.043500000000001e-05,
      "loss": 0.353,
      "step": 14095
    },
    {
      "epoch": 0.7893381117706351,
      "grad_norm": 1.5501999855041504,
      "learning_rate": 7.044000000000001e-05,
      "loss": 0.5938,
      "step": 14096
    },
    {
      "epoch": 0.789394109082764,
      "grad_norm": 1.583893895149231,
      "learning_rate": 7.044500000000001e-05,
      "loss": 0.3894,
      "step": 14097
    },
    {
      "epoch": 0.789450106394893,
      "grad_norm": 1.5419281721115112,
      "learning_rate": 7.045e-05,
      "loss": 0.6311,
      "step": 14098
    },
    {
      "epoch": 0.789506103707022,
      "grad_norm": 1.2712949514389038,
      "learning_rate": 7.0455e-05,
      "loss": 0.5394,
      "step": 14099
    },
    {
      "epoch": 0.789562101019151,
      "grad_norm": 1.4368669986724854,
      "learning_rate": 7.046e-05,
      "loss": 0.4303,
      "step": 14100
    },
    {
      "epoch": 0.78961809833128,
      "grad_norm": 1.144955039024353,
      "learning_rate": 7.0465e-05,
      "loss": 0.4121,
      "step": 14101
    },
    {
      "epoch": 0.7896740956434091,
      "grad_norm": 7.278684139251709,
      "learning_rate": 7.047000000000001e-05,
      "loss": 0.3975,
      "step": 14102
    },
    {
      "epoch": 0.7897300929555381,
      "grad_norm": 1.2687106132507324,
      "learning_rate": 7.0475e-05,
      "loss": 0.3977,
      "step": 14103
    },
    {
      "epoch": 0.7897860902676671,
      "grad_norm": 1.5099725723266602,
      "learning_rate": 7.048e-05,
      "loss": 0.4205,
      "step": 14104
    },
    {
      "epoch": 0.7898420875797961,
      "grad_norm": 1.4072988033294678,
      "learning_rate": 7.0485e-05,
      "loss": 0.406,
      "step": 14105
    },
    {
      "epoch": 0.7898980848919251,
      "grad_norm": 1.2100664377212524,
      "learning_rate": 7.049e-05,
      "loss": 0.318,
      "step": 14106
    },
    {
      "epoch": 0.7899540822040542,
      "grad_norm": 1.2987401485443115,
      "learning_rate": 7.0495e-05,
      "loss": 0.4051,
      "step": 14107
    },
    {
      "epoch": 0.7900100795161832,
      "grad_norm": 1.3227951526641846,
      "learning_rate": 7.05e-05,
      "loss": 0.4309,
      "step": 14108
    },
    {
      "epoch": 0.7900660768283122,
      "grad_norm": 1.254512071609497,
      "learning_rate": 7.0505e-05,
      "loss": 0.4531,
      "step": 14109
    },
    {
      "epoch": 0.7901220741404412,
      "grad_norm": 1.2160489559173584,
      "learning_rate": 7.051e-05,
      "loss": 0.5015,
      "step": 14110
    },
    {
      "epoch": 0.7901780714525702,
      "grad_norm": 1.265352725982666,
      "learning_rate": 7.051500000000001e-05,
      "loss": 0.3967,
      "step": 14111
    },
    {
      "epoch": 0.7902340687646993,
      "grad_norm": 1.3767404556274414,
      "learning_rate": 7.052000000000001e-05,
      "loss": 0.4449,
      "step": 14112
    },
    {
      "epoch": 0.7902900660768283,
      "grad_norm": 1.5246636867523193,
      "learning_rate": 7.0525e-05,
      "loss": 0.5072,
      "step": 14113
    },
    {
      "epoch": 0.7903460633889573,
      "grad_norm": 1.4679063558578491,
      "learning_rate": 7.053e-05,
      "loss": 0.5503,
      "step": 14114
    },
    {
      "epoch": 0.7904020607010863,
      "grad_norm": 1.1133171319961548,
      "learning_rate": 7.0535e-05,
      "loss": 0.3749,
      "step": 14115
    },
    {
      "epoch": 0.7904580580132153,
      "grad_norm": 1.1903791427612305,
      "learning_rate": 7.054000000000001e-05,
      "loss": 0.4349,
      "step": 14116
    },
    {
      "epoch": 0.7905140553253444,
      "grad_norm": 1.286184310913086,
      "learning_rate": 7.054500000000001e-05,
      "loss": 0.3713,
      "step": 14117
    },
    {
      "epoch": 0.7905700526374734,
      "grad_norm": 1.1068525314331055,
      "learning_rate": 7.055000000000001e-05,
      "loss": 0.3936,
      "step": 14118
    },
    {
      "epoch": 0.7906260499496024,
      "grad_norm": 1.2849111557006836,
      "learning_rate": 7.0555e-05,
      "loss": 0.4914,
      "step": 14119
    },
    {
      "epoch": 0.7906820472617314,
      "grad_norm": 1.794370174407959,
      "learning_rate": 7.056e-05,
      "loss": 0.5837,
      "step": 14120
    },
    {
      "epoch": 0.7907380445738604,
      "grad_norm": 1.2527025938034058,
      "learning_rate": 7.0565e-05,
      "loss": 0.3589,
      "step": 14121
    },
    {
      "epoch": 0.7907940418859895,
      "grad_norm": 1.3123115301132202,
      "learning_rate": 7.057e-05,
      "loss": 0.4757,
      "step": 14122
    },
    {
      "epoch": 0.7908500391981185,
      "grad_norm": 1.499796986579895,
      "learning_rate": 7.057500000000001e-05,
      "loss": 0.5368,
      "step": 14123
    },
    {
      "epoch": 0.7909060365102475,
      "grad_norm": 1.1614887714385986,
      "learning_rate": 7.058e-05,
      "loss": 0.5316,
      "step": 14124
    },
    {
      "epoch": 0.7909620338223765,
      "grad_norm": 1.4116251468658447,
      "learning_rate": 7.0585e-05,
      "loss": 0.5434,
      "step": 14125
    },
    {
      "epoch": 0.7910180311345055,
      "grad_norm": 1.3411842584609985,
      "learning_rate": 7.059e-05,
      "loss": 0.407,
      "step": 14126
    },
    {
      "epoch": 0.7910740284466345,
      "grad_norm": 1.2646327018737793,
      "learning_rate": 7.0595e-05,
      "loss": 0.4557,
      "step": 14127
    },
    {
      "epoch": 0.7911300257587636,
      "grad_norm": 1.333034634590149,
      "learning_rate": 7.06e-05,
      "loss": 0.3658,
      "step": 14128
    },
    {
      "epoch": 0.7911860230708926,
      "grad_norm": 1.3084169626235962,
      "learning_rate": 7.060499999999999e-05,
      "loss": 0.4472,
      "step": 14129
    },
    {
      "epoch": 0.7912420203830216,
      "grad_norm": 1.3331811428070068,
      "learning_rate": 7.061e-05,
      "loss": 0.4248,
      "step": 14130
    },
    {
      "epoch": 0.7912980176951506,
      "grad_norm": 1.2928749322891235,
      "learning_rate": 7.061500000000001e-05,
      "loss": 0.3972,
      "step": 14131
    },
    {
      "epoch": 0.7913540150072796,
      "grad_norm": 1.325481653213501,
      "learning_rate": 7.062000000000001e-05,
      "loss": 0.3963,
      "step": 14132
    },
    {
      "epoch": 0.7914100123194087,
      "grad_norm": 1.6766866445541382,
      "learning_rate": 7.062500000000001e-05,
      "loss": 0.5129,
      "step": 14133
    },
    {
      "epoch": 0.7914660096315377,
      "grad_norm": 1.099776268005371,
      "learning_rate": 7.063e-05,
      "loss": 0.4562,
      "step": 14134
    },
    {
      "epoch": 0.7915220069436667,
      "grad_norm": 1.5759435892105103,
      "learning_rate": 7.0635e-05,
      "loss": 0.4211,
      "step": 14135
    },
    {
      "epoch": 0.7915780042557957,
      "grad_norm": 1.4366590976715088,
      "learning_rate": 7.064e-05,
      "loss": 0.5245,
      "step": 14136
    },
    {
      "epoch": 0.7916340015679247,
      "grad_norm": 1.5051610469818115,
      "learning_rate": 7.064500000000001e-05,
      "loss": 0.6648,
      "step": 14137
    },
    {
      "epoch": 0.7916899988800538,
      "grad_norm": 1.4565407037734985,
      "learning_rate": 7.065000000000001e-05,
      "loss": 0.4574,
      "step": 14138
    },
    {
      "epoch": 0.7917459961921828,
      "grad_norm": 1.1879507303237915,
      "learning_rate": 7.065500000000001e-05,
      "loss": 0.4324,
      "step": 14139
    },
    {
      "epoch": 0.7918019935043118,
      "grad_norm": 1.3431316614151,
      "learning_rate": 7.066e-05,
      "loss": 0.5131,
      "step": 14140
    },
    {
      "epoch": 0.7918579908164408,
      "grad_norm": 1.2967579364776611,
      "learning_rate": 7.0665e-05,
      "loss": 0.4321,
      "step": 14141
    },
    {
      "epoch": 0.7919139881285698,
      "grad_norm": 1.5854921340942383,
      "learning_rate": 7.067e-05,
      "loss": 0.5023,
      "step": 14142
    },
    {
      "epoch": 0.7919699854406989,
      "grad_norm": 1.1517387628555298,
      "learning_rate": 7.0675e-05,
      "loss": 0.3641,
      "step": 14143
    },
    {
      "epoch": 0.7920259827528279,
      "grad_norm": 1.1762385368347168,
      "learning_rate": 7.068000000000001e-05,
      "loss": 0.4396,
      "step": 14144
    },
    {
      "epoch": 0.7920819800649569,
      "grad_norm": 1.2819037437438965,
      "learning_rate": 7.0685e-05,
      "loss": 0.4133,
      "step": 14145
    },
    {
      "epoch": 0.7921379773770859,
      "grad_norm": 1.4618840217590332,
      "learning_rate": 7.069e-05,
      "loss": 0.3764,
      "step": 14146
    },
    {
      "epoch": 0.7921939746892149,
      "grad_norm": 1.475507140159607,
      "learning_rate": 7.0695e-05,
      "loss": 0.4936,
      "step": 14147
    },
    {
      "epoch": 0.792249972001344,
      "grad_norm": 1.2078789472579956,
      "learning_rate": 7.07e-05,
      "loss": 0.3813,
      "step": 14148
    },
    {
      "epoch": 0.792305969313473,
      "grad_norm": 1.1872726678848267,
      "learning_rate": 7.0705e-05,
      "loss": 0.3415,
      "step": 14149
    },
    {
      "epoch": 0.792361966625602,
      "grad_norm": 1.275092601776123,
      "learning_rate": 7.070999999999999e-05,
      "loss": 0.5457,
      "step": 14150
    },
    {
      "epoch": 0.792417963937731,
      "grad_norm": 1.3667881488800049,
      "learning_rate": 7.0715e-05,
      "loss": 0.5445,
      "step": 14151
    },
    {
      "epoch": 0.79247396124986,
      "grad_norm": 1.5850286483764648,
      "learning_rate": 7.072000000000001e-05,
      "loss": 0.7943,
      "step": 14152
    },
    {
      "epoch": 0.792529958561989,
      "grad_norm": 1.4312115907669067,
      "learning_rate": 7.072500000000001e-05,
      "loss": 0.4663,
      "step": 14153
    },
    {
      "epoch": 0.7925859558741181,
      "grad_norm": 1.304882526397705,
      "learning_rate": 7.073000000000001e-05,
      "loss": 0.4723,
      "step": 14154
    },
    {
      "epoch": 0.7926419531862471,
      "grad_norm": 1.721720576286316,
      "learning_rate": 7.0735e-05,
      "loss": 0.4387,
      "step": 14155
    },
    {
      "epoch": 0.7926979504983761,
      "grad_norm": 1.597560167312622,
      "learning_rate": 7.074e-05,
      "loss": 0.4141,
      "step": 14156
    },
    {
      "epoch": 0.7927539478105051,
      "grad_norm": 1.3952064514160156,
      "learning_rate": 7.0745e-05,
      "loss": 0.4149,
      "step": 14157
    },
    {
      "epoch": 0.7928099451226341,
      "grad_norm": 1.1094516515731812,
      "learning_rate": 7.075e-05,
      "loss": 0.318,
      "step": 14158
    },
    {
      "epoch": 0.7928659424347632,
      "grad_norm": 1.298769235610962,
      "learning_rate": 7.075500000000001e-05,
      "loss": 0.3279,
      "step": 14159
    },
    {
      "epoch": 0.7929219397468922,
      "grad_norm": 1.2508598566055298,
      "learning_rate": 7.076000000000001e-05,
      "loss": 0.401,
      "step": 14160
    },
    {
      "epoch": 0.7929779370590212,
      "grad_norm": 1.1670008897781372,
      "learning_rate": 7.0765e-05,
      "loss": 0.355,
      "step": 14161
    },
    {
      "epoch": 0.7930339343711502,
      "grad_norm": 1.7155007123947144,
      "learning_rate": 7.077e-05,
      "loss": 0.3922,
      "step": 14162
    },
    {
      "epoch": 0.7930899316832792,
      "grad_norm": 1.4616706371307373,
      "learning_rate": 7.0775e-05,
      "loss": 0.4847,
      "step": 14163
    },
    {
      "epoch": 0.7931459289954083,
      "grad_norm": 1.3160438537597656,
      "learning_rate": 7.078e-05,
      "loss": 0.4121,
      "step": 14164
    },
    {
      "epoch": 0.7932019263075373,
      "grad_norm": 1.2417327165603638,
      "learning_rate": 7.078500000000001e-05,
      "loss": 0.5756,
      "step": 14165
    },
    {
      "epoch": 0.7932579236196663,
      "grad_norm": 1.224461555480957,
      "learning_rate": 7.079e-05,
      "loss": 0.451,
      "step": 14166
    },
    {
      "epoch": 0.7933139209317953,
      "grad_norm": 1.3468503952026367,
      "learning_rate": 7.0795e-05,
      "loss": 0.473,
      "step": 14167
    },
    {
      "epoch": 0.7933699182439243,
      "grad_norm": 1.8887944221496582,
      "learning_rate": 7.08e-05,
      "loss": 0.4912,
      "step": 14168
    },
    {
      "epoch": 0.7934259155560534,
      "grad_norm": 1.1624226570129395,
      "learning_rate": 7.0805e-05,
      "loss": 0.4854,
      "step": 14169
    },
    {
      "epoch": 0.7934819128681824,
      "grad_norm": 1.4265356063842773,
      "learning_rate": 7.081e-05,
      "loss": 0.4913,
      "step": 14170
    },
    {
      "epoch": 0.7935379101803114,
      "grad_norm": 1.379022240638733,
      "learning_rate": 7.081499999999999e-05,
      "loss": 0.5303,
      "step": 14171
    },
    {
      "epoch": 0.7935939074924404,
      "grad_norm": 1.1742662191390991,
      "learning_rate": 7.082e-05,
      "loss": 0.3983,
      "step": 14172
    },
    {
      "epoch": 0.7936499048045694,
      "grad_norm": 1.4051592350006104,
      "learning_rate": 7.082500000000001e-05,
      "loss": 0.3314,
      "step": 14173
    },
    {
      "epoch": 0.7937059021166984,
      "grad_norm": 1.246253252029419,
      "learning_rate": 7.083000000000001e-05,
      "loss": 0.3705,
      "step": 14174
    },
    {
      "epoch": 0.7937618994288275,
      "grad_norm": 1.3388421535491943,
      "learning_rate": 7.083500000000001e-05,
      "loss": 0.5268,
      "step": 14175
    },
    {
      "epoch": 0.7938178967409565,
      "grad_norm": 1.4494574069976807,
      "learning_rate": 7.084e-05,
      "loss": 0.5604,
      "step": 14176
    },
    {
      "epoch": 0.7938738940530855,
      "grad_norm": 1.3175182342529297,
      "learning_rate": 7.0845e-05,
      "loss": 0.4475,
      "step": 14177
    },
    {
      "epoch": 0.7939298913652145,
      "grad_norm": 1.9641207456588745,
      "learning_rate": 7.085e-05,
      "loss": 0.391,
      "step": 14178
    },
    {
      "epoch": 0.7939858886773434,
      "grad_norm": 1.4529749155044556,
      "learning_rate": 7.0855e-05,
      "loss": 0.3616,
      "step": 14179
    },
    {
      "epoch": 0.7940418859894725,
      "grad_norm": 1.5091392993927002,
      "learning_rate": 7.086000000000001e-05,
      "loss": 0.6272,
      "step": 14180
    },
    {
      "epoch": 0.7940978833016015,
      "grad_norm": 1.480791687965393,
      "learning_rate": 7.0865e-05,
      "loss": 0.5521,
      "step": 14181
    },
    {
      "epoch": 0.7941538806137305,
      "grad_norm": 1.2192174196243286,
      "learning_rate": 7.087e-05,
      "loss": 0.4446,
      "step": 14182
    },
    {
      "epoch": 0.7942098779258595,
      "grad_norm": 1.1119846105575562,
      "learning_rate": 7.0875e-05,
      "loss": 0.432,
      "step": 14183
    },
    {
      "epoch": 0.7942658752379885,
      "grad_norm": 1.2087913751602173,
      "learning_rate": 7.088e-05,
      "loss": 0.4761,
      "step": 14184
    },
    {
      "epoch": 0.7943218725501175,
      "grad_norm": 1.1879780292510986,
      "learning_rate": 7.0885e-05,
      "loss": 0.4461,
      "step": 14185
    },
    {
      "epoch": 0.7943778698622466,
      "grad_norm": 2.624023914337158,
      "learning_rate": 7.089000000000001e-05,
      "loss": 0.5263,
      "step": 14186
    },
    {
      "epoch": 0.7944338671743756,
      "grad_norm": 1.4163124561309814,
      "learning_rate": 7.0895e-05,
      "loss": 0.4879,
      "step": 14187
    },
    {
      "epoch": 0.7944898644865046,
      "grad_norm": 1.2269207239151,
      "learning_rate": 7.09e-05,
      "loss": 0.4756,
      "step": 14188
    },
    {
      "epoch": 0.7945458617986336,
      "grad_norm": 1.2559641599655151,
      "learning_rate": 7.0905e-05,
      "loss": 0.6148,
      "step": 14189
    },
    {
      "epoch": 0.7946018591107626,
      "grad_norm": 1.4977549314498901,
      "learning_rate": 7.091e-05,
      "loss": 0.594,
      "step": 14190
    },
    {
      "epoch": 0.7946578564228917,
      "grad_norm": 1.6066761016845703,
      "learning_rate": 7.0915e-05,
      "loss": 0.3459,
      "step": 14191
    },
    {
      "epoch": 0.7947138537350207,
      "grad_norm": 1.1700197458267212,
      "learning_rate": 7.092e-05,
      "loss": 0.4241,
      "step": 14192
    },
    {
      "epoch": 0.7947698510471497,
      "grad_norm": 1.5863022804260254,
      "learning_rate": 7.0925e-05,
      "loss": 0.7132,
      "step": 14193
    },
    {
      "epoch": 0.7948258483592787,
      "grad_norm": 1.3558669090270996,
      "learning_rate": 7.093000000000001e-05,
      "loss": 0.3699,
      "step": 14194
    },
    {
      "epoch": 0.7948818456714077,
      "grad_norm": 1.8490558862686157,
      "learning_rate": 7.093500000000001e-05,
      "loss": 0.6204,
      "step": 14195
    },
    {
      "epoch": 0.7949378429835368,
      "grad_norm": 1.3741960525512695,
      "learning_rate": 7.094000000000001e-05,
      "loss": 0.4074,
      "step": 14196
    },
    {
      "epoch": 0.7949938402956658,
      "grad_norm": 1.4264682531356812,
      "learning_rate": 7.0945e-05,
      "loss": 0.4239,
      "step": 14197
    },
    {
      "epoch": 0.7950498376077948,
      "grad_norm": 1.9793999195098877,
      "learning_rate": 7.095e-05,
      "loss": 0.4577,
      "step": 14198
    },
    {
      "epoch": 0.7951058349199238,
      "grad_norm": 1.2810734510421753,
      "learning_rate": 7.0955e-05,
      "loss": 0.4092,
      "step": 14199
    },
    {
      "epoch": 0.7951618322320528,
      "grad_norm": 1.3208116292953491,
      "learning_rate": 7.096e-05,
      "loss": 0.4284,
      "step": 14200
    },
    {
      "epoch": 0.7952178295441819,
      "grad_norm": 1.63396418094635,
      "learning_rate": 7.096500000000001e-05,
      "loss": 0.5604,
      "step": 14201
    },
    {
      "epoch": 0.7952738268563109,
      "grad_norm": 1.3054780960083008,
      "learning_rate": 7.097e-05,
      "loss": 0.4291,
      "step": 14202
    },
    {
      "epoch": 0.7953298241684399,
      "grad_norm": 1.2806179523468018,
      "learning_rate": 7.0975e-05,
      "loss": 0.4277,
      "step": 14203
    },
    {
      "epoch": 0.7953858214805689,
      "grad_norm": 1.4290262460708618,
      "learning_rate": 7.098e-05,
      "loss": 0.5616,
      "step": 14204
    },
    {
      "epoch": 0.7954418187926979,
      "grad_norm": 1.2974903583526611,
      "learning_rate": 7.0985e-05,
      "loss": 0.3513,
      "step": 14205
    },
    {
      "epoch": 0.795497816104827,
      "grad_norm": 1.2196741104125977,
      "learning_rate": 7.099e-05,
      "loss": 0.4768,
      "step": 14206
    },
    {
      "epoch": 0.795553813416956,
      "grad_norm": 2.9551949501037598,
      "learning_rate": 7.0995e-05,
      "loss": 0.6393,
      "step": 14207
    },
    {
      "epoch": 0.795609810729085,
      "grad_norm": 1.1885533332824707,
      "learning_rate": 7.1e-05,
      "loss": 0.4344,
      "step": 14208
    },
    {
      "epoch": 0.795665808041214,
      "grad_norm": 1.4281381368637085,
      "learning_rate": 7.1005e-05,
      "loss": 0.4298,
      "step": 14209
    },
    {
      "epoch": 0.795721805353343,
      "grad_norm": 1.1432418823242188,
      "learning_rate": 7.101e-05,
      "loss": 0.4367,
      "step": 14210
    },
    {
      "epoch": 0.795777802665472,
      "grad_norm": 1.2060902118682861,
      "learning_rate": 7.1015e-05,
      "loss": 0.4498,
      "step": 14211
    },
    {
      "epoch": 0.7958337999776011,
      "grad_norm": 1.2184255123138428,
      "learning_rate": 7.102000000000001e-05,
      "loss": 0.4082,
      "step": 14212
    },
    {
      "epoch": 0.7958897972897301,
      "grad_norm": 1.3108534812927246,
      "learning_rate": 7.1025e-05,
      "loss": 0.4099,
      "step": 14213
    },
    {
      "epoch": 0.7959457946018591,
      "grad_norm": 1.232239007949829,
      "learning_rate": 7.103e-05,
      "loss": 0.6333,
      "step": 14214
    },
    {
      "epoch": 0.7960017919139881,
      "grad_norm": 1.3570442199707031,
      "learning_rate": 7.103500000000001e-05,
      "loss": 0.5427,
      "step": 14215
    },
    {
      "epoch": 0.7960577892261171,
      "grad_norm": 1.299333930015564,
      "learning_rate": 7.104000000000001e-05,
      "loss": 0.4296,
      "step": 14216
    },
    {
      "epoch": 0.7961137865382462,
      "grad_norm": 1.7934577465057373,
      "learning_rate": 7.104500000000001e-05,
      "loss": 0.6227,
      "step": 14217
    },
    {
      "epoch": 0.7961697838503752,
      "grad_norm": 1.5457813739776611,
      "learning_rate": 7.105e-05,
      "loss": 0.4957,
      "step": 14218
    },
    {
      "epoch": 0.7962257811625042,
      "grad_norm": 1.2532612085342407,
      "learning_rate": 7.1055e-05,
      "loss": 0.3346,
      "step": 14219
    },
    {
      "epoch": 0.7962817784746332,
      "grad_norm": 2.023634910583496,
      "learning_rate": 7.106e-05,
      "loss": 0.44,
      "step": 14220
    },
    {
      "epoch": 0.7963377757867622,
      "grad_norm": 6.002133369445801,
      "learning_rate": 7.1065e-05,
      "loss": 0.4516,
      "step": 14221
    },
    {
      "epoch": 0.7963937730988913,
      "grad_norm": 1.5097990036010742,
      "learning_rate": 7.107000000000001e-05,
      "loss": 0.5253,
      "step": 14222
    },
    {
      "epoch": 0.7964497704110203,
      "grad_norm": 1.0850706100463867,
      "learning_rate": 7.1075e-05,
      "loss": 0.3218,
      "step": 14223
    },
    {
      "epoch": 0.7965057677231493,
      "grad_norm": 1.238458514213562,
      "learning_rate": 7.108e-05,
      "loss": 0.4333,
      "step": 14224
    },
    {
      "epoch": 0.7965617650352783,
      "grad_norm": 1.1130365133285522,
      "learning_rate": 7.1085e-05,
      "loss": 0.4513,
      "step": 14225
    },
    {
      "epoch": 0.7966177623474073,
      "grad_norm": 1.3274425268173218,
      "learning_rate": 7.109e-05,
      "loss": 0.4381,
      "step": 14226
    },
    {
      "epoch": 0.7966737596595364,
      "grad_norm": 1.3055750131607056,
      "learning_rate": 7.1095e-05,
      "loss": 0.4583,
      "step": 14227
    },
    {
      "epoch": 0.7967297569716654,
      "grad_norm": 1.180130958557129,
      "learning_rate": 7.11e-05,
      "loss": 0.4189,
      "step": 14228
    },
    {
      "epoch": 0.7967857542837944,
      "grad_norm": 1.506287932395935,
      "learning_rate": 7.1105e-05,
      "loss": 0.5207,
      "step": 14229
    },
    {
      "epoch": 0.7968417515959234,
      "grad_norm": 1.1442244052886963,
      "learning_rate": 7.111e-05,
      "loss": 0.5126,
      "step": 14230
    },
    {
      "epoch": 0.7968977489080524,
      "grad_norm": 1.3099751472473145,
      "learning_rate": 7.1115e-05,
      "loss": 0.566,
      "step": 14231
    },
    {
      "epoch": 0.7969537462201814,
      "grad_norm": 1.828741192817688,
      "learning_rate": 7.112000000000001e-05,
      "loss": 0.3673,
      "step": 14232
    },
    {
      "epoch": 0.7970097435323105,
      "grad_norm": 1.4499508142471313,
      "learning_rate": 7.112500000000001e-05,
      "loss": 0.5663,
      "step": 14233
    },
    {
      "epoch": 0.7970657408444395,
      "grad_norm": 1.4727813005447388,
      "learning_rate": 7.113e-05,
      "loss": 0.5223,
      "step": 14234
    },
    {
      "epoch": 0.7971217381565685,
      "grad_norm": 1.4278827905654907,
      "learning_rate": 7.1135e-05,
      "loss": 0.5051,
      "step": 14235
    },
    {
      "epoch": 0.7971777354686975,
      "grad_norm": 2.509204387664795,
      "learning_rate": 7.114e-05,
      "loss": 0.5877,
      "step": 14236
    },
    {
      "epoch": 0.7972337327808265,
      "grad_norm": 1.3308311700820923,
      "learning_rate": 7.114500000000001e-05,
      "loss": 0.4522,
      "step": 14237
    },
    {
      "epoch": 0.7972897300929556,
      "grad_norm": 1.5752297639846802,
      "learning_rate": 7.115000000000001e-05,
      "loss": 0.43,
      "step": 14238
    },
    {
      "epoch": 0.7973457274050846,
      "grad_norm": 2.0403895378112793,
      "learning_rate": 7.1155e-05,
      "loss": 0.5288,
      "step": 14239
    },
    {
      "epoch": 0.7974017247172136,
      "grad_norm": 1.2725965976715088,
      "learning_rate": 7.116e-05,
      "loss": 0.5799,
      "step": 14240
    },
    {
      "epoch": 0.7974577220293426,
      "grad_norm": 1.2783291339874268,
      "learning_rate": 7.1165e-05,
      "loss": 0.4713,
      "step": 14241
    },
    {
      "epoch": 0.7975137193414716,
      "grad_norm": 1.2267341613769531,
      "learning_rate": 7.117e-05,
      "loss": 0.3125,
      "step": 14242
    },
    {
      "epoch": 0.7975697166536007,
      "grad_norm": 1.5969126224517822,
      "learning_rate": 7.117500000000001e-05,
      "loss": 0.4726,
      "step": 14243
    },
    {
      "epoch": 0.7976257139657297,
      "grad_norm": 1.3979538679122925,
      "learning_rate": 7.118e-05,
      "loss": 0.3592,
      "step": 14244
    },
    {
      "epoch": 0.7976817112778587,
      "grad_norm": 1.6626211404800415,
      "learning_rate": 7.1185e-05,
      "loss": 0.4393,
      "step": 14245
    },
    {
      "epoch": 0.7977377085899877,
      "grad_norm": 1.2637313604354858,
      "learning_rate": 7.119e-05,
      "loss": 0.3928,
      "step": 14246
    },
    {
      "epoch": 0.7977937059021167,
      "grad_norm": 1.3991094827651978,
      "learning_rate": 7.1195e-05,
      "loss": 0.452,
      "step": 14247
    },
    {
      "epoch": 0.7978497032142458,
      "grad_norm": 1.4132726192474365,
      "learning_rate": 7.12e-05,
      "loss": 0.4239,
      "step": 14248
    },
    {
      "epoch": 0.7979057005263748,
      "grad_norm": 1.7055772542953491,
      "learning_rate": 7.1205e-05,
      "loss": 0.6003,
      "step": 14249
    },
    {
      "epoch": 0.7979616978385038,
      "grad_norm": 1.3924280405044556,
      "learning_rate": 7.121e-05,
      "loss": 0.5031,
      "step": 14250
    },
    {
      "epoch": 0.7980176951506328,
      "grad_norm": 1.4206241369247437,
      "learning_rate": 7.1215e-05,
      "loss": 0.4005,
      "step": 14251
    },
    {
      "epoch": 0.7980736924627618,
      "grad_norm": 1.3712095022201538,
      "learning_rate": 7.122000000000001e-05,
      "loss": 0.4737,
      "step": 14252
    },
    {
      "epoch": 0.7981296897748908,
      "grad_norm": 1.3633575439453125,
      "learning_rate": 7.122500000000001e-05,
      "loss": 0.6029,
      "step": 14253
    },
    {
      "epoch": 0.7981856870870199,
      "grad_norm": 1.168293833732605,
      "learning_rate": 7.123000000000001e-05,
      "loss": 0.5213,
      "step": 14254
    },
    {
      "epoch": 0.7982416843991489,
      "grad_norm": 1.3674473762512207,
      "learning_rate": 7.1235e-05,
      "loss": 0.4504,
      "step": 14255
    },
    {
      "epoch": 0.7982976817112779,
      "grad_norm": 1.1504228115081787,
      "learning_rate": 7.124e-05,
      "loss": 0.431,
      "step": 14256
    },
    {
      "epoch": 0.7983536790234069,
      "grad_norm": 1.3997844457626343,
      "learning_rate": 7.1245e-05,
      "loss": 0.5008,
      "step": 14257
    },
    {
      "epoch": 0.798409676335536,
      "grad_norm": 1.2385599613189697,
      "learning_rate": 7.125000000000001e-05,
      "loss": 0.3936,
      "step": 14258
    },
    {
      "epoch": 0.798465673647665,
      "grad_norm": 1.5361371040344238,
      "learning_rate": 7.125500000000001e-05,
      "loss": 0.3703,
      "step": 14259
    },
    {
      "epoch": 0.798521670959794,
      "grad_norm": 1.2596443891525269,
      "learning_rate": 7.126e-05,
      "loss": 0.4015,
      "step": 14260
    },
    {
      "epoch": 0.798577668271923,
      "grad_norm": 1.484503149986267,
      "learning_rate": 7.1265e-05,
      "loss": 0.4397,
      "step": 14261
    },
    {
      "epoch": 0.7986336655840519,
      "grad_norm": 1.2304725646972656,
      "learning_rate": 7.127e-05,
      "loss": 0.3993,
      "step": 14262
    },
    {
      "epoch": 0.7986896628961809,
      "grad_norm": 1.341511607170105,
      "learning_rate": 7.1275e-05,
      "loss": 0.4711,
      "step": 14263
    },
    {
      "epoch": 0.79874566020831,
      "grad_norm": 1.4946208000183105,
      "learning_rate": 7.128000000000001e-05,
      "loss": 0.475,
      "step": 14264
    },
    {
      "epoch": 0.798801657520439,
      "grad_norm": 1.28714919090271,
      "learning_rate": 7.1285e-05,
      "loss": 0.4342,
      "step": 14265
    },
    {
      "epoch": 0.798857654832568,
      "grad_norm": 1.2437433004379272,
      "learning_rate": 7.129e-05,
      "loss": 0.5012,
      "step": 14266
    },
    {
      "epoch": 0.798913652144697,
      "grad_norm": 1.260019302368164,
      "learning_rate": 7.1295e-05,
      "loss": 0.3945,
      "step": 14267
    },
    {
      "epoch": 0.798969649456826,
      "grad_norm": 1.257361888885498,
      "learning_rate": 7.13e-05,
      "loss": 0.4328,
      "step": 14268
    },
    {
      "epoch": 0.799025646768955,
      "grad_norm": 1.4107636213302612,
      "learning_rate": 7.1305e-05,
      "loss": 0.5142,
      "step": 14269
    },
    {
      "epoch": 0.7990816440810841,
      "grad_norm": 1.4554567337036133,
      "learning_rate": 7.130999999999999e-05,
      "loss": 0.4655,
      "step": 14270
    },
    {
      "epoch": 0.7991376413932131,
      "grad_norm": 1.271363615989685,
      "learning_rate": 7.1315e-05,
      "loss": 0.4211,
      "step": 14271
    },
    {
      "epoch": 0.7991936387053421,
      "grad_norm": 1.533981204032898,
      "learning_rate": 7.132e-05,
      "loss": 0.3774,
      "step": 14272
    },
    {
      "epoch": 0.7992496360174711,
      "grad_norm": 1.338004231452942,
      "learning_rate": 7.132500000000001e-05,
      "loss": 0.4101,
      "step": 14273
    },
    {
      "epoch": 0.7993056333296001,
      "grad_norm": 1.4239612817764282,
      "learning_rate": 7.133000000000001e-05,
      "loss": 0.4688,
      "step": 14274
    },
    {
      "epoch": 0.7993616306417292,
      "grad_norm": 1.2443686723709106,
      "learning_rate": 7.133500000000001e-05,
      "loss": 0.452,
      "step": 14275
    },
    {
      "epoch": 0.7994176279538582,
      "grad_norm": 2.020554780960083,
      "learning_rate": 7.134e-05,
      "loss": 0.5314,
      "step": 14276
    },
    {
      "epoch": 0.7994736252659872,
      "grad_norm": 1.0896626710891724,
      "learning_rate": 7.1345e-05,
      "loss": 0.4325,
      "step": 14277
    },
    {
      "epoch": 0.7995296225781162,
      "grad_norm": 1.2125056982040405,
      "learning_rate": 7.135e-05,
      "loss": 0.3837,
      "step": 14278
    },
    {
      "epoch": 0.7995856198902452,
      "grad_norm": 1.316235065460205,
      "learning_rate": 7.135500000000001e-05,
      "loss": 0.5725,
      "step": 14279
    },
    {
      "epoch": 0.7996416172023743,
      "grad_norm": 1.1320780515670776,
      "learning_rate": 7.136000000000001e-05,
      "loss": 0.3444,
      "step": 14280
    },
    {
      "epoch": 0.7996976145145033,
      "grad_norm": 1.5037559270858765,
      "learning_rate": 7.1365e-05,
      "loss": 0.6402,
      "step": 14281
    },
    {
      "epoch": 0.7997536118266323,
      "grad_norm": 1.3142647743225098,
      "learning_rate": 7.137e-05,
      "loss": 0.4549,
      "step": 14282
    },
    {
      "epoch": 0.7998096091387613,
      "grad_norm": 1.2375869750976562,
      "learning_rate": 7.1375e-05,
      "loss": 0.3238,
      "step": 14283
    },
    {
      "epoch": 0.7998656064508903,
      "grad_norm": 1.1469448804855347,
      "learning_rate": 7.138e-05,
      "loss": 0.418,
      "step": 14284
    },
    {
      "epoch": 0.7999216037630194,
      "grad_norm": 1.3749374151229858,
      "learning_rate": 7.138500000000001e-05,
      "loss": 0.5568,
      "step": 14285
    },
    {
      "epoch": 0.7999776010751484,
      "grad_norm": 1.19504976272583,
      "learning_rate": 7.139e-05,
      "loss": 0.3862,
      "step": 14286
    },
    {
      "epoch": 0.8000335983872774,
      "grad_norm": 5.354552745819092,
      "learning_rate": 7.1395e-05,
      "loss": 0.4797,
      "step": 14287
    },
    {
      "epoch": 0.8000895956994064,
      "grad_norm": 1.2926965951919556,
      "learning_rate": 7.14e-05,
      "loss": 0.3903,
      "step": 14288
    },
    {
      "epoch": 0.8001455930115354,
      "grad_norm": 1.509556770324707,
      "learning_rate": 7.1405e-05,
      "loss": 0.4057,
      "step": 14289
    },
    {
      "epoch": 0.8002015903236644,
      "grad_norm": 1.476244330406189,
      "learning_rate": 7.141e-05,
      "loss": 0.5082,
      "step": 14290
    },
    {
      "epoch": 0.8002575876357935,
      "grad_norm": 1.268317699432373,
      "learning_rate": 7.141499999999999e-05,
      "loss": 0.4666,
      "step": 14291
    },
    {
      "epoch": 0.8003135849479225,
      "grad_norm": 1.8188954591751099,
      "learning_rate": 7.142e-05,
      "loss": 0.4773,
      "step": 14292
    },
    {
      "epoch": 0.8003695822600515,
      "grad_norm": 1.49734628200531,
      "learning_rate": 7.142500000000001e-05,
      "loss": 0.4786,
      "step": 14293
    },
    {
      "epoch": 0.8004255795721805,
      "grad_norm": 2.688525438308716,
      "learning_rate": 7.143000000000001e-05,
      "loss": 0.4933,
      "step": 14294
    },
    {
      "epoch": 0.8004815768843095,
      "grad_norm": 1.2296005487442017,
      "learning_rate": 7.143500000000001e-05,
      "loss": 0.3708,
      "step": 14295
    },
    {
      "epoch": 0.8005375741964386,
      "grad_norm": 1.2410680055618286,
      "learning_rate": 7.144000000000001e-05,
      "loss": 0.4211,
      "step": 14296
    },
    {
      "epoch": 0.8005935715085676,
      "grad_norm": 1.3349018096923828,
      "learning_rate": 7.1445e-05,
      "loss": 0.5248,
      "step": 14297
    },
    {
      "epoch": 0.8006495688206966,
      "grad_norm": 1.2120205163955688,
      "learning_rate": 7.145e-05,
      "loss": 0.4427,
      "step": 14298
    },
    {
      "epoch": 0.8007055661328256,
      "grad_norm": 1.176427960395813,
      "learning_rate": 7.1455e-05,
      "loss": 0.5057,
      "step": 14299
    },
    {
      "epoch": 0.8007615634449546,
      "grad_norm": 1.1894642114639282,
      "learning_rate": 7.146000000000001e-05,
      "loss": 0.3401,
      "step": 14300
    },
    {
      "epoch": 0.8008175607570837,
      "grad_norm": 1.3587576150894165,
      "learning_rate": 7.146500000000001e-05,
      "loss": 0.5311,
      "step": 14301
    },
    {
      "epoch": 0.8008735580692127,
      "grad_norm": 1.4819519519805908,
      "learning_rate": 7.147e-05,
      "loss": 0.6611,
      "step": 14302
    },
    {
      "epoch": 0.8009295553813417,
      "grad_norm": 1.2229000329971313,
      "learning_rate": 7.1475e-05,
      "loss": 0.42,
      "step": 14303
    },
    {
      "epoch": 0.8009855526934707,
      "grad_norm": 1.2849124670028687,
      "learning_rate": 7.148e-05,
      "loss": 0.4447,
      "step": 14304
    },
    {
      "epoch": 0.8010415500055997,
      "grad_norm": 1.5820670127868652,
      "learning_rate": 7.1485e-05,
      "loss": 0.5244,
      "step": 14305
    },
    {
      "epoch": 0.8010975473177288,
      "grad_norm": 1.4715663194656372,
      "learning_rate": 7.149e-05,
      "loss": 0.4639,
      "step": 14306
    },
    {
      "epoch": 0.8011535446298578,
      "grad_norm": 1.313481330871582,
      "learning_rate": 7.1495e-05,
      "loss": 0.341,
      "step": 14307
    },
    {
      "epoch": 0.8012095419419868,
      "grad_norm": 3.2087135314941406,
      "learning_rate": 7.15e-05,
      "loss": 0.544,
      "step": 14308
    },
    {
      "epoch": 0.8012655392541158,
      "grad_norm": 1.0600281953811646,
      "learning_rate": 7.1505e-05,
      "loss": 0.3455,
      "step": 14309
    },
    {
      "epoch": 0.8013215365662448,
      "grad_norm": 1.1321611404418945,
      "learning_rate": 7.151e-05,
      "loss": 0.4154,
      "step": 14310
    },
    {
      "epoch": 0.8013775338783738,
      "grad_norm": 1.4016467332839966,
      "learning_rate": 7.1515e-05,
      "loss": 0.4927,
      "step": 14311
    },
    {
      "epoch": 0.8014335311905029,
      "grad_norm": 1.1130542755126953,
      "learning_rate": 7.151999999999999e-05,
      "loss": 0.3972,
      "step": 14312
    },
    {
      "epoch": 0.8014895285026319,
      "grad_norm": 1.4507575035095215,
      "learning_rate": 7.1525e-05,
      "loss": 0.399,
      "step": 14313
    },
    {
      "epoch": 0.8015455258147609,
      "grad_norm": 1.2351462841033936,
      "learning_rate": 7.153000000000001e-05,
      "loss": 0.3804,
      "step": 14314
    },
    {
      "epoch": 0.8016015231268899,
      "grad_norm": 1.241011381149292,
      "learning_rate": 7.153500000000001e-05,
      "loss": 0.4423,
      "step": 14315
    },
    {
      "epoch": 0.801657520439019,
      "grad_norm": 1.063472032546997,
      "learning_rate": 7.154000000000001e-05,
      "loss": 0.2882,
      "step": 14316
    },
    {
      "epoch": 0.801713517751148,
      "grad_norm": 1.2055721282958984,
      "learning_rate": 7.154500000000001e-05,
      "loss": 0.4239,
      "step": 14317
    },
    {
      "epoch": 0.801769515063277,
      "grad_norm": 1.5081279277801514,
      "learning_rate": 7.155e-05,
      "loss": 0.413,
      "step": 14318
    },
    {
      "epoch": 0.801825512375406,
      "grad_norm": 1.3589757680892944,
      "learning_rate": 7.1555e-05,
      "loss": 0.407,
      "step": 14319
    },
    {
      "epoch": 0.801881509687535,
      "grad_norm": 1.2902450561523438,
      "learning_rate": 7.156e-05,
      "loss": 0.3748,
      "step": 14320
    },
    {
      "epoch": 0.801937506999664,
      "grad_norm": 1.3496792316436768,
      "learning_rate": 7.156500000000001e-05,
      "loss": 0.3717,
      "step": 14321
    },
    {
      "epoch": 0.8019935043117931,
      "grad_norm": 1.3816170692443848,
      "learning_rate": 7.157000000000001e-05,
      "loss": 0.3834,
      "step": 14322
    },
    {
      "epoch": 0.8020495016239221,
      "grad_norm": 1.2812403440475464,
      "learning_rate": 7.1575e-05,
      "loss": 0.4117,
      "step": 14323
    },
    {
      "epoch": 0.8021054989360511,
      "grad_norm": 1.4883240461349487,
      "learning_rate": 7.158e-05,
      "loss": 0.4894,
      "step": 14324
    },
    {
      "epoch": 0.8021614962481801,
      "grad_norm": 1.299131989479065,
      "learning_rate": 7.1585e-05,
      "loss": 0.5771,
      "step": 14325
    },
    {
      "epoch": 0.8022174935603091,
      "grad_norm": 1.1620566844940186,
      "learning_rate": 7.159e-05,
      "loss": 0.5,
      "step": 14326
    },
    {
      "epoch": 0.8022734908724382,
      "grad_norm": 1.3025226593017578,
      "learning_rate": 7.1595e-05,
      "loss": 0.4366,
      "step": 14327
    },
    {
      "epoch": 0.8023294881845672,
      "grad_norm": 1.404446005821228,
      "learning_rate": 7.16e-05,
      "loss": 0.7016,
      "step": 14328
    },
    {
      "epoch": 0.8023854854966962,
      "grad_norm": 1.3371330499649048,
      "learning_rate": 7.1605e-05,
      "loss": 0.5923,
      "step": 14329
    },
    {
      "epoch": 0.8024414828088252,
      "grad_norm": 1.2328046560287476,
      "learning_rate": 7.161e-05,
      "loss": 0.4832,
      "step": 14330
    },
    {
      "epoch": 0.8024974801209542,
      "grad_norm": 1.2135400772094727,
      "learning_rate": 7.1615e-05,
      "loss": 0.39,
      "step": 14331
    },
    {
      "epoch": 0.8025534774330833,
      "grad_norm": 1.1489354372024536,
      "learning_rate": 7.162e-05,
      "loss": 0.4436,
      "step": 14332
    },
    {
      "epoch": 0.8026094747452123,
      "grad_norm": 1.3100652694702148,
      "learning_rate": 7.1625e-05,
      "loss": 0.6064,
      "step": 14333
    },
    {
      "epoch": 0.8026654720573413,
      "grad_norm": 1.2295804023742676,
      "learning_rate": 7.163e-05,
      "loss": 0.3946,
      "step": 14334
    },
    {
      "epoch": 0.8027214693694703,
      "grad_norm": 3.268707513809204,
      "learning_rate": 7.1635e-05,
      "loss": 0.4619,
      "step": 14335
    },
    {
      "epoch": 0.8027774666815993,
      "grad_norm": 1.4066771268844604,
      "learning_rate": 7.164000000000001e-05,
      "loss": 0.5977,
      "step": 14336
    },
    {
      "epoch": 0.8028334639937283,
      "grad_norm": 1.3116049766540527,
      "learning_rate": 7.164500000000001e-05,
      "loss": 0.4666,
      "step": 14337
    },
    {
      "epoch": 0.8028894613058574,
      "grad_norm": 1.5711578130722046,
      "learning_rate": 7.165000000000001e-05,
      "loss": 0.4854,
      "step": 14338
    },
    {
      "epoch": 0.8029454586179864,
      "grad_norm": 1.2447172403335571,
      "learning_rate": 7.1655e-05,
      "loss": 0.4169,
      "step": 14339
    },
    {
      "epoch": 0.8030014559301154,
      "grad_norm": 1.405240535736084,
      "learning_rate": 7.166e-05,
      "loss": 0.7066,
      "step": 14340
    },
    {
      "epoch": 0.8030574532422444,
      "grad_norm": 1.3452883958816528,
      "learning_rate": 7.1665e-05,
      "loss": 0.4707,
      "step": 14341
    },
    {
      "epoch": 0.8031134505543734,
      "grad_norm": 1.3635317087173462,
      "learning_rate": 7.167000000000001e-05,
      "loss": 0.448,
      "step": 14342
    },
    {
      "epoch": 0.8031694478665025,
      "grad_norm": 1.2283295392990112,
      "learning_rate": 7.167500000000001e-05,
      "loss": 0.4203,
      "step": 14343
    },
    {
      "epoch": 0.8032254451786315,
      "grad_norm": 1.4867265224456787,
      "learning_rate": 7.168e-05,
      "loss": 0.4884,
      "step": 14344
    },
    {
      "epoch": 0.8032814424907604,
      "grad_norm": 1.251662015914917,
      "learning_rate": 7.1685e-05,
      "loss": 0.4322,
      "step": 14345
    },
    {
      "epoch": 0.8033374398028894,
      "grad_norm": 1.4250129461288452,
      "learning_rate": 7.169e-05,
      "loss": 0.5191,
      "step": 14346
    },
    {
      "epoch": 0.8033934371150184,
      "grad_norm": 1.3150770664215088,
      "learning_rate": 7.1695e-05,
      "loss": 0.4644,
      "step": 14347
    },
    {
      "epoch": 0.8034494344271474,
      "grad_norm": 1.645158052444458,
      "learning_rate": 7.17e-05,
      "loss": 0.4428,
      "step": 14348
    },
    {
      "epoch": 0.8035054317392765,
      "grad_norm": 1.4586005210876465,
      "learning_rate": 7.1705e-05,
      "loss": 0.5313,
      "step": 14349
    },
    {
      "epoch": 0.8035614290514055,
      "grad_norm": 1.369107961654663,
      "learning_rate": 7.171e-05,
      "loss": 0.4005,
      "step": 14350
    },
    {
      "epoch": 0.8036174263635345,
      "grad_norm": 1.3012665510177612,
      "learning_rate": 7.1715e-05,
      "loss": 0.5529,
      "step": 14351
    },
    {
      "epoch": 0.8036734236756635,
      "grad_norm": 1.325852632522583,
      "learning_rate": 7.172e-05,
      "loss": 0.3732,
      "step": 14352
    },
    {
      "epoch": 0.8037294209877925,
      "grad_norm": 1.499667763710022,
      "learning_rate": 7.172500000000001e-05,
      "loss": 0.4682,
      "step": 14353
    },
    {
      "epoch": 0.8037854182999216,
      "grad_norm": 1.3996325731277466,
      "learning_rate": 7.173e-05,
      "loss": 0.4142,
      "step": 14354
    },
    {
      "epoch": 0.8038414156120506,
      "grad_norm": 1.36122465133667,
      "learning_rate": 7.1735e-05,
      "loss": 0.4767,
      "step": 14355
    },
    {
      "epoch": 0.8038974129241796,
      "grad_norm": 1.3188210725784302,
      "learning_rate": 7.174e-05,
      "loss": 0.5409,
      "step": 14356
    },
    {
      "epoch": 0.8039534102363086,
      "grad_norm": 1.2389452457427979,
      "learning_rate": 7.174500000000001e-05,
      "loss": 0.4215,
      "step": 14357
    },
    {
      "epoch": 0.8040094075484376,
      "grad_norm": 1.310027003288269,
      "learning_rate": 7.175000000000001e-05,
      "loss": 0.4265,
      "step": 14358
    },
    {
      "epoch": 0.8040654048605667,
      "grad_norm": 1.260506272315979,
      "learning_rate": 7.1755e-05,
      "loss": 0.398,
      "step": 14359
    },
    {
      "epoch": 0.8041214021726957,
      "grad_norm": 1.1169949769973755,
      "learning_rate": 7.176e-05,
      "loss": 0.4912,
      "step": 14360
    },
    {
      "epoch": 0.8041773994848247,
      "grad_norm": 1.4652414321899414,
      "learning_rate": 7.1765e-05,
      "loss": 0.4644,
      "step": 14361
    },
    {
      "epoch": 0.8042333967969537,
      "grad_norm": 1.5234915018081665,
      "learning_rate": 7.177e-05,
      "loss": 0.5104,
      "step": 14362
    },
    {
      "epoch": 0.8042893941090827,
      "grad_norm": 1.2871246337890625,
      "learning_rate": 7.177500000000001e-05,
      "loss": 0.3849,
      "step": 14363
    },
    {
      "epoch": 0.8043453914212118,
      "grad_norm": 1.1576374769210815,
      "learning_rate": 7.178000000000001e-05,
      "loss": 0.4857,
      "step": 14364
    },
    {
      "epoch": 0.8044013887333408,
      "grad_norm": 1.2182807922363281,
      "learning_rate": 7.1785e-05,
      "loss": 0.4108,
      "step": 14365
    },
    {
      "epoch": 0.8044573860454698,
      "grad_norm": 1.3278043270111084,
      "learning_rate": 7.179e-05,
      "loss": 0.379,
      "step": 14366
    },
    {
      "epoch": 0.8045133833575988,
      "grad_norm": 1.2018606662750244,
      "learning_rate": 7.1795e-05,
      "loss": 0.4943,
      "step": 14367
    },
    {
      "epoch": 0.8045693806697278,
      "grad_norm": 1.2104520797729492,
      "learning_rate": 7.18e-05,
      "loss": 0.4814,
      "step": 14368
    },
    {
      "epoch": 0.8046253779818568,
      "grad_norm": 1.061835765838623,
      "learning_rate": 7.1805e-05,
      "loss": 0.3731,
      "step": 14369
    },
    {
      "epoch": 0.8046813752939859,
      "grad_norm": 1.4871950149536133,
      "learning_rate": 7.181e-05,
      "loss": 0.5726,
      "step": 14370
    },
    {
      "epoch": 0.8047373726061149,
      "grad_norm": 1.4260612726211548,
      "learning_rate": 7.1815e-05,
      "loss": 0.4471,
      "step": 14371
    },
    {
      "epoch": 0.8047933699182439,
      "grad_norm": 1.37614107131958,
      "learning_rate": 7.182e-05,
      "loss": 0.4413,
      "step": 14372
    },
    {
      "epoch": 0.8048493672303729,
      "grad_norm": 1.2057019472122192,
      "learning_rate": 7.182500000000001e-05,
      "loss": 0.4537,
      "step": 14373
    },
    {
      "epoch": 0.8049053645425019,
      "grad_norm": 1.2680689096450806,
      "learning_rate": 7.183000000000001e-05,
      "loss": 0.5552,
      "step": 14374
    },
    {
      "epoch": 0.804961361854631,
      "grad_norm": 1.451594591140747,
      "learning_rate": 7.1835e-05,
      "loss": 0.5571,
      "step": 14375
    },
    {
      "epoch": 0.80501735916676,
      "grad_norm": 1.197312355041504,
      "learning_rate": 7.184e-05,
      "loss": 0.3676,
      "step": 14376
    },
    {
      "epoch": 0.805073356478889,
      "grad_norm": 1.3863509893417358,
      "learning_rate": 7.1845e-05,
      "loss": 0.4757,
      "step": 14377
    },
    {
      "epoch": 0.805129353791018,
      "grad_norm": 1.2120007276535034,
      "learning_rate": 7.185000000000001e-05,
      "loss": 0.4002,
      "step": 14378
    },
    {
      "epoch": 0.805185351103147,
      "grad_norm": 1.3701173067092896,
      "learning_rate": 7.185500000000001e-05,
      "loss": 0.4783,
      "step": 14379
    },
    {
      "epoch": 0.8052413484152761,
      "grad_norm": 1.5542365312576294,
      "learning_rate": 7.186e-05,
      "loss": 0.3459,
      "step": 14380
    },
    {
      "epoch": 0.8052973457274051,
      "grad_norm": 1.4818055629730225,
      "learning_rate": 7.1865e-05,
      "loss": 0.5042,
      "step": 14381
    },
    {
      "epoch": 0.8053533430395341,
      "grad_norm": 1.4172481298446655,
      "learning_rate": 7.187e-05,
      "loss": 0.4043,
      "step": 14382
    },
    {
      "epoch": 0.8054093403516631,
      "grad_norm": 1.235385775566101,
      "learning_rate": 7.1875e-05,
      "loss": 0.3711,
      "step": 14383
    },
    {
      "epoch": 0.8054653376637921,
      "grad_norm": 1.370848536491394,
      "learning_rate": 7.188e-05,
      "loss": 0.3752,
      "step": 14384
    },
    {
      "epoch": 0.8055213349759212,
      "grad_norm": 1.4222896099090576,
      "learning_rate": 7.188500000000001e-05,
      "loss": 0.4127,
      "step": 14385
    },
    {
      "epoch": 0.8055773322880502,
      "grad_norm": 1.5399459600448608,
      "learning_rate": 7.189e-05,
      "loss": 0.4703,
      "step": 14386
    },
    {
      "epoch": 0.8056333296001792,
      "grad_norm": 1.208827018737793,
      "learning_rate": 7.1895e-05,
      "loss": 0.4682,
      "step": 14387
    },
    {
      "epoch": 0.8056893269123082,
      "grad_norm": 1.3453835248947144,
      "learning_rate": 7.19e-05,
      "loss": 0.4047,
      "step": 14388
    },
    {
      "epoch": 0.8057453242244372,
      "grad_norm": 1.356747031211853,
      "learning_rate": 7.1905e-05,
      "loss": 0.6252,
      "step": 14389
    },
    {
      "epoch": 0.8058013215365663,
      "grad_norm": 1.1763570308685303,
      "learning_rate": 7.191e-05,
      "loss": 0.4417,
      "step": 14390
    },
    {
      "epoch": 0.8058573188486953,
      "grad_norm": 1.4643850326538086,
      "learning_rate": 7.1915e-05,
      "loss": 0.6276,
      "step": 14391
    },
    {
      "epoch": 0.8059133161608243,
      "grad_norm": 1.474872350692749,
      "learning_rate": 7.192e-05,
      "loss": 0.4014,
      "step": 14392
    },
    {
      "epoch": 0.8059693134729533,
      "grad_norm": 1.3869831562042236,
      "learning_rate": 7.1925e-05,
      "loss": 0.4205,
      "step": 14393
    },
    {
      "epoch": 0.8060253107850823,
      "grad_norm": 1.3573460578918457,
      "learning_rate": 7.193000000000001e-05,
      "loss": 0.4208,
      "step": 14394
    },
    {
      "epoch": 0.8060813080972113,
      "grad_norm": 1.432698130607605,
      "learning_rate": 7.193500000000001e-05,
      "loss": 0.4238,
      "step": 14395
    },
    {
      "epoch": 0.8061373054093404,
      "grad_norm": 1.7275192737579346,
      "learning_rate": 7.194e-05,
      "loss": 0.4974,
      "step": 14396
    },
    {
      "epoch": 0.8061933027214694,
      "grad_norm": 1.3129981756210327,
      "learning_rate": 7.1945e-05,
      "loss": 0.4118,
      "step": 14397
    },
    {
      "epoch": 0.8062493000335984,
      "grad_norm": 1.3502404689788818,
      "learning_rate": 7.195e-05,
      "loss": 0.4085,
      "step": 14398
    },
    {
      "epoch": 0.8063052973457274,
      "grad_norm": 1.3459618091583252,
      "learning_rate": 7.195500000000001e-05,
      "loss": 0.4328,
      "step": 14399
    },
    {
      "epoch": 0.8063612946578564,
      "grad_norm": 1.3390995264053345,
      "learning_rate": 7.196000000000001e-05,
      "loss": 0.3985,
      "step": 14400
    },
    {
      "epoch": 0.8064172919699855,
      "grad_norm": 1.316535472869873,
      "learning_rate": 7.1965e-05,
      "loss": 0.4415,
      "step": 14401
    },
    {
      "epoch": 0.8064732892821145,
      "grad_norm": 1.365905523300171,
      "learning_rate": 7.197e-05,
      "loss": 0.4783,
      "step": 14402
    },
    {
      "epoch": 0.8065292865942435,
      "grad_norm": 1.3806086778640747,
      "learning_rate": 7.1975e-05,
      "loss": 0.4347,
      "step": 14403
    },
    {
      "epoch": 0.8065852839063725,
      "grad_norm": 1.2446452379226685,
      "learning_rate": 7.198e-05,
      "loss": 0.5143,
      "step": 14404
    },
    {
      "epoch": 0.8066412812185015,
      "grad_norm": 1.1861881017684937,
      "learning_rate": 7.1985e-05,
      "loss": 0.3499,
      "step": 14405
    },
    {
      "epoch": 0.8066972785306306,
      "grad_norm": 1.2290480136871338,
      "learning_rate": 7.199000000000001e-05,
      "loss": 0.397,
      "step": 14406
    },
    {
      "epoch": 0.8067532758427596,
      "grad_norm": 1.3085788488388062,
      "learning_rate": 7.1995e-05,
      "loss": 0.3467,
      "step": 14407
    },
    {
      "epoch": 0.8068092731548886,
      "grad_norm": 1.5729981660842896,
      "learning_rate": 7.2e-05,
      "loss": 0.4784,
      "step": 14408
    },
    {
      "epoch": 0.8068652704670176,
      "grad_norm": 1.4126801490783691,
      "learning_rate": 7.2005e-05,
      "loss": 0.4939,
      "step": 14409
    },
    {
      "epoch": 0.8069212677791466,
      "grad_norm": 1.1600823402404785,
      "learning_rate": 7.201e-05,
      "loss": 0.4275,
      "step": 14410
    },
    {
      "epoch": 0.8069772650912757,
      "grad_norm": 1.2501877546310425,
      "learning_rate": 7.2015e-05,
      "loss": 0.3635,
      "step": 14411
    },
    {
      "epoch": 0.8070332624034047,
      "grad_norm": 1.3730149269104004,
      "learning_rate": 7.202e-05,
      "loss": 0.4545,
      "step": 14412
    },
    {
      "epoch": 0.8070892597155337,
      "grad_norm": 1.0997185707092285,
      "learning_rate": 7.2025e-05,
      "loss": 0.3805,
      "step": 14413
    },
    {
      "epoch": 0.8071452570276627,
      "grad_norm": 1.1816157102584839,
      "learning_rate": 7.203000000000001e-05,
      "loss": 0.4126,
      "step": 14414
    },
    {
      "epoch": 0.8072012543397917,
      "grad_norm": 1.4850237369537354,
      "learning_rate": 7.203500000000001e-05,
      "loss": 0.4057,
      "step": 14415
    },
    {
      "epoch": 0.8072572516519207,
      "grad_norm": 1.332733154296875,
      "learning_rate": 7.204000000000001e-05,
      "loss": 0.6524,
      "step": 14416
    },
    {
      "epoch": 0.8073132489640498,
      "grad_norm": 1.2679888010025024,
      "learning_rate": 7.2045e-05,
      "loss": 0.6952,
      "step": 14417
    },
    {
      "epoch": 0.8073692462761788,
      "grad_norm": 1.5113410949707031,
      "learning_rate": 7.205e-05,
      "loss": 0.5025,
      "step": 14418
    },
    {
      "epoch": 0.8074252435883078,
      "grad_norm": 1.4489210844039917,
      "learning_rate": 7.2055e-05,
      "loss": 0.5082,
      "step": 14419
    },
    {
      "epoch": 0.8074812409004368,
      "grad_norm": 2.1691977977752686,
      "learning_rate": 7.206000000000001e-05,
      "loss": 0.6638,
      "step": 14420
    },
    {
      "epoch": 0.8075372382125658,
      "grad_norm": 1.3792576789855957,
      "learning_rate": 7.206500000000001e-05,
      "loss": 0.4289,
      "step": 14421
    },
    {
      "epoch": 0.8075932355246949,
      "grad_norm": 3.897068500518799,
      "learning_rate": 7.207e-05,
      "loss": 0.417,
      "step": 14422
    },
    {
      "epoch": 0.8076492328368239,
      "grad_norm": 1.4487427473068237,
      "learning_rate": 7.2075e-05,
      "loss": 0.5249,
      "step": 14423
    },
    {
      "epoch": 0.8077052301489529,
      "grad_norm": 1.3798766136169434,
      "learning_rate": 7.208e-05,
      "loss": 0.4358,
      "step": 14424
    },
    {
      "epoch": 0.8077612274610819,
      "grad_norm": 1.737931251525879,
      "learning_rate": 7.2085e-05,
      "loss": 0.4121,
      "step": 14425
    },
    {
      "epoch": 0.8078172247732109,
      "grad_norm": 1.1489999294281006,
      "learning_rate": 7.209e-05,
      "loss": 0.4085,
      "step": 14426
    },
    {
      "epoch": 0.8078732220853398,
      "grad_norm": 1.190069556236267,
      "learning_rate": 7.209500000000001e-05,
      "loss": 0.5369,
      "step": 14427
    },
    {
      "epoch": 0.8079292193974689,
      "grad_norm": 1.5137462615966797,
      "learning_rate": 7.21e-05,
      "loss": 0.7515,
      "step": 14428
    },
    {
      "epoch": 0.8079852167095979,
      "grad_norm": 1.2904857397079468,
      "learning_rate": 7.2105e-05,
      "loss": 0.4236,
      "step": 14429
    },
    {
      "epoch": 0.8080412140217269,
      "grad_norm": 1.3198213577270508,
      "learning_rate": 7.211e-05,
      "loss": 0.4046,
      "step": 14430
    },
    {
      "epoch": 0.8080972113338559,
      "grad_norm": 1.3352689743041992,
      "learning_rate": 7.2115e-05,
      "loss": 0.4601,
      "step": 14431
    },
    {
      "epoch": 0.8081532086459849,
      "grad_norm": 1.275355577468872,
      "learning_rate": 7.212e-05,
      "loss": 0.4128,
      "step": 14432
    },
    {
      "epoch": 0.808209205958114,
      "grad_norm": 1.2645851373672485,
      "learning_rate": 7.2125e-05,
      "loss": 0.403,
      "step": 14433
    },
    {
      "epoch": 0.808265203270243,
      "grad_norm": 1.3488813638687134,
      "learning_rate": 7.213e-05,
      "loss": 0.5088,
      "step": 14434
    },
    {
      "epoch": 0.808321200582372,
      "grad_norm": 1.4230356216430664,
      "learning_rate": 7.213500000000001e-05,
      "loss": 0.6918,
      "step": 14435
    },
    {
      "epoch": 0.808377197894501,
      "grad_norm": 1.0565516948699951,
      "learning_rate": 7.214000000000001e-05,
      "loss": 0.3467,
      "step": 14436
    },
    {
      "epoch": 0.80843319520663,
      "grad_norm": 1.523375153541565,
      "learning_rate": 7.214500000000001e-05,
      "loss": 0.3884,
      "step": 14437
    },
    {
      "epoch": 0.8084891925187591,
      "grad_norm": 1.429703950881958,
      "learning_rate": 7.215e-05,
      "loss": 0.4894,
      "step": 14438
    },
    {
      "epoch": 0.8085451898308881,
      "grad_norm": 1.278806447982788,
      "learning_rate": 7.2155e-05,
      "loss": 0.4793,
      "step": 14439
    },
    {
      "epoch": 0.8086011871430171,
      "grad_norm": 1.5889099836349487,
      "learning_rate": 7.216e-05,
      "loss": 0.5262,
      "step": 14440
    },
    {
      "epoch": 0.8086571844551461,
      "grad_norm": 1.4965639114379883,
      "learning_rate": 7.216500000000001e-05,
      "loss": 0.4079,
      "step": 14441
    },
    {
      "epoch": 0.8087131817672751,
      "grad_norm": 1.2740325927734375,
      "learning_rate": 7.217000000000001e-05,
      "loss": 0.3079,
      "step": 14442
    },
    {
      "epoch": 0.8087691790794042,
      "grad_norm": 1.7930535078048706,
      "learning_rate": 7.2175e-05,
      "loss": 0.4116,
      "step": 14443
    },
    {
      "epoch": 0.8088251763915332,
      "grad_norm": 1.1276098489761353,
      "learning_rate": 7.218e-05,
      "loss": 0.3954,
      "step": 14444
    },
    {
      "epoch": 0.8088811737036622,
      "grad_norm": 1.5948141813278198,
      "learning_rate": 7.2185e-05,
      "loss": 0.5563,
      "step": 14445
    },
    {
      "epoch": 0.8089371710157912,
      "grad_norm": 2.2157845497131348,
      "learning_rate": 7.219e-05,
      "loss": 0.5056,
      "step": 14446
    },
    {
      "epoch": 0.8089931683279202,
      "grad_norm": 1.2125484943389893,
      "learning_rate": 7.2195e-05,
      "loss": 0.3818,
      "step": 14447
    },
    {
      "epoch": 0.8090491656400493,
      "grad_norm": 1.572281837463379,
      "learning_rate": 7.22e-05,
      "loss": 0.4074,
      "step": 14448
    },
    {
      "epoch": 0.8091051629521783,
      "grad_norm": 1.3786903619766235,
      "learning_rate": 7.2205e-05,
      "loss": 0.4561,
      "step": 14449
    },
    {
      "epoch": 0.8091611602643073,
      "grad_norm": 1.3767876625061035,
      "learning_rate": 7.221e-05,
      "loss": 0.4368,
      "step": 14450
    },
    {
      "epoch": 0.8092171575764363,
      "grad_norm": 1.33573579788208,
      "learning_rate": 7.2215e-05,
      "loss": 0.3953,
      "step": 14451
    },
    {
      "epoch": 0.8092731548885653,
      "grad_norm": 1.474442720413208,
      "learning_rate": 7.222e-05,
      "loss": 0.6192,
      "step": 14452
    },
    {
      "epoch": 0.8093291522006943,
      "grad_norm": 1.7762633562088013,
      "learning_rate": 7.2225e-05,
      "loss": 0.3859,
      "step": 14453
    },
    {
      "epoch": 0.8093851495128234,
      "grad_norm": 1.1925138235092163,
      "learning_rate": 7.223e-05,
      "loss": 0.4591,
      "step": 14454
    },
    {
      "epoch": 0.8094411468249524,
      "grad_norm": 1.2057689428329468,
      "learning_rate": 7.2235e-05,
      "loss": 0.4817,
      "step": 14455
    },
    {
      "epoch": 0.8094971441370814,
      "grad_norm": 1.4963434934616089,
      "learning_rate": 7.224000000000001e-05,
      "loss": 0.4412,
      "step": 14456
    },
    {
      "epoch": 0.8095531414492104,
      "grad_norm": 1.4457064867019653,
      "learning_rate": 7.224500000000001e-05,
      "loss": 0.3906,
      "step": 14457
    },
    {
      "epoch": 0.8096091387613394,
      "grad_norm": 1.3031740188598633,
      "learning_rate": 7.225000000000001e-05,
      "loss": 0.3942,
      "step": 14458
    },
    {
      "epoch": 0.8096651360734685,
      "grad_norm": 1.4185705184936523,
      "learning_rate": 7.2255e-05,
      "loss": 0.5671,
      "step": 14459
    },
    {
      "epoch": 0.8097211333855975,
      "grad_norm": 1.244755744934082,
      "learning_rate": 7.226e-05,
      "loss": 0.5442,
      "step": 14460
    },
    {
      "epoch": 0.8097771306977265,
      "grad_norm": 2.351407766342163,
      "learning_rate": 7.2265e-05,
      "loss": 0.414,
      "step": 14461
    },
    {
      "epoch": 0.8098331280098555,
      "grad_norm": 1.2181895971298218,
      "learning_rate": 7.227000000000001e-05,
      "loss": 0.4345,
      "step": 14462
    },
    {
      "epoch": 0.8098891253219845,
      "grad_norm": 1.3296558856964111,
      "learning_rate": 7.227500000000001e-05,
      "loss": 0.4265,
      "step": 14463
    },
    {
      "epoch": 0.8099451226341136,
      "grad_norm": 1.4543119668960571,
      "learning_rate": 7.228e-05,
      "loss": 0.4575,
      "step": 14464
    },
    {
      "epoch": 0.8100011199462426,
      "grad_norm": 1.1617560386657715,
      "learning_rate": 7.2285e-05,
      "loss": 0.4275,
      "step": 14465
    },
    {
      "epoch": 0.8100571172583716,
      "grad_norm": 1.3625268936157227,
      "learning_rate": 7.229e-05,
      "loss": 0.5049,
      "step": 14466
    },
    {
      "epoch": 0.8101131145705006,
      "grad_norm": 1.1242477893829346,
      "learning_rate": 7.2295e-05,
      "loss": 0.3465,
      "step": 14467
    },
    {
      "epoch": 0.8101691118826296,
      "grad_norm": 1.2985645532608032,
      "learning_rate": 7.23e-05,
      "loss": 0.5279,
      "step": 14468
    },
    {
      "epoch": 0.8102251091947587,
      "grad_norm": 1.2515453100204468,
      "learning_rate": 7.2305e-05,
      "loss": 0.4025,
      "step": 14469
    },
    {
      "epoch": 0.8102811065068877,
      "grad_norm": 1.2598460912704468,
      "learning_rate": 7.231e-05,
      "loss": 0.4909,
      "step": 14470
    },
    {
      "epoch": 0.8103371038190167,
      "grad_norm": 1.250108003616333,
      "learning_rate": 7.2315e-05,
      "loss": 0.4534,
      "step": 14471
    },
    {
      "epoch": 0.8103931011311457,
      "grad_norm": 1.2872648239135742,
      "learning_rate": 7.232e-05,
      "loss": 0.3846,
      "step": 14472
    },
    {
      "epoch": 0.8104490984432747,
      "grad_norm": 2.0495991706848145,
      "learning_rate": 7.2325e-05,
      "loss": 0.462,
      "step": 14473
    },
    {
      "epoch": 0.8105050957554037,
      "grad_norm": 1.2422711849212646,
      "learning_rate": 7.233000000000001e-05,
      "loss": 0.3957,
      "step": 14474
    },
    {
      "epoch": 0.8105610930675328,
      "grad_norm": 1.7333513498306274,
      "learning_rate": 7.2335e-05,
      "loss": 0.4894,
      "step": 14475
    },
    {
      "epoch": 0.8106170903796618,
      "grad_norm": 1.6531480550765991,
      "learning_rate": 7.234e-05,
      "loss": 0.5644,
      "step": 14476
    },
    {
      "epoch": 0.8106730876917908,
      "grad_norm": 1.3454701900482178,
      "learning_rate": 7.234500000000001e-05,
      "loss": 0.4615,
      "step": 14477
    },
    {
      "epoch": 0.8107290850039198,
      "grad_norm": 1.1700000762939453,
      "learning_rate": 7.235000000000001e-05,
      "loss": 0.303,
      "step": 14478
    },
    {
      "epoch": 0.8107850823160488,
      "grad_norm": 1.2846060991287231,
      "learning_rate": 7.235500000000001e-05,
      "loss": 0.5067,
      "step": 14479
    },
    {
      "epoch": 0.8108410796281779,
      "grad_norm": 1.861580729484558,
      "learning_rate": 7.236e-05,
      "loss": 0.4014,
      "step": 14480
    },
    {
      "epoch": 0.8108970769403069,
      "grad_norm": 1.2220993041992188,
      "learning_rate": 7.2365e-05,
      "loss": 0.4058,
      "step": 14481
    },
    {
      "epoch": 0.8109530742524359,
      "grad_norm": 1.3710349798202515,
      "learning_rate": 7.237e-05,
      "loss": 0.4855,
      "step": 14482
    },
    {
      "epoch": 0.8110090715645649,
      "grad_norm": 2.1424448490142822,
      "learning_rate": 7.2375e-05,
      "loss": 0.3967,
      "step": 14483
    },
    {
      "epoch": 0.8110650688766939,
      "grad_norm": 1.3352257013320923,
      "learning_rate": 7.238000000000001e-05,
      "loss": 0.3667,
      "step": 14484
    },
    {
      "epoch": 0.811121066188823,
      "grad_norm": 1.3583341836929321,
      "learning_rate": 7.2385e-05,
      "loss": 0.3491,
      "step": 14485
    },
    {
      "epoch": 0.811177063500952,
      "grad_norm": 1.2811883687973022,
      "learning_rate": 7.239e-05,
      "loss": 0.3825,
      "step": 14486
    },
    {
      "epoch": 0.811233060813081,
      "grad_norm": 1.3344007730484009,
      "learning_rate": 7.2395e-05,
      "loss": 0.4904,
      "step": 14487
    },
    {
      "epoch": 0.81128905812521,
      "grad_norm": 1.446425437927246,
      "learning_rate": 7.24e-05,
      "loss": 0.5377,
      "step": 14488
    },
    {
      "epoch": 0.811345055437339,
      "grad_norm": 1.0479023456573486,
      "learning_rate": 7.2405e-05,
      "loss": 0.3216,
      "step": 14489
    },
    {
      "epoch": 0.811401052749468,
      "grad_norm": 1.2680141925811768,
      "learning_rate": 7.241e-05,
      "loss": 0.4654,
      "step": 14490
    },
    {
      "epoch": 0.8114570500615971,
      "grad_norm": 1.3281927108764648,
      "learning_rate": 7.2415e-05,
      "loss": 0.4584,
      "step": 14491
    },
    {
      "epoch": 0.8115130473737261,
      "grad_norm": 1.3772964477539062,
      "learning_rate": 7.242e-05,
      "loss": 0.375,
      "step": 14492
    },
    {
      "epoch": 0.8115690446858551,
      "grad_norm": 1.623201847076416,
      "learning_rate": 7.2425e-05,
      "loss": 0.4131,
      "step": 14493
    },
    {
      "epoch": 0.8116250419979841,
      "grad_norm": 1.2621368169784546,
      "learning_rate": 7.243000000000001e-05,
      "loss": 0.3265,
      "step": 14494
    },
    {
      "epoch": 0.8116810393101132,
      "grad_norm": 1.3126702308654785,
      "learning_rate": 7.243500000000001e-05,
      "loss": 0.3985,
      "step": 14495
    },
    {
      "epoch": 0.8117370366222422,
      "grad_norm": 1.4472118616104126,
      "learning_rate": 7.244e-05,
      "loss": 0.3782,
      "step": 14496
    },
    {
      "epoch": 0.8117930339343712,
      "grad_norm": 1.2617502212524414,
      "learning_rate": 7.2445e-05,
      "loss": 0.3656,
      "step": 14497
    },
    {
      "epoch": 0.8118490312465002,
      "grad_norm": 1.3361226320266724,
      "learning_rate": 7.245000000000001e-05,
      "loss": 0.4149,
      "step": 14498
    },
    {
      "epoch": 0.8119050285586292,
      "grad_norm": 1.5286263227462769,
      "learning_rate": 7.245500000000001e-05,
      "loss": 0.5432,
      "step": 14499
    },
    {
      "epoch": 0.8119610258707582,
      "grad_norm": 1.36464262008667,
      "learning_rate": 7.246000000000001e-05,
      "loss": 0.3893,
      "step": 14500
    },
    {
      "epoch": 0.8120170231828873,
      "grad_norm": 1.3603463172912598,
      "learning_rate": 7.2465e-05,
      "loss": 0.3698,
      "step": 14501
    },
    {
      "epoch": 0.8120730204950163,
      "grad_norm": 1.6051450967788696,
      "learning_rate": 7.247e-05,
      "loss": 0.5028,
      "step": 14502
    },
    {
      "epoch": 0.8121290178071453,
      "grad_norm": 1.5369080305099487,
      "learning_rate": 7.2475e-05,
      "loss": 0.4999,
      "step": 14503
    },
    {
      "epoch": 0.8121850151192743,
      "grad_norm": 1.2656768560409546,
      "learning_rate": 7.248e-05,
      "loss": 0.3857,
      "step": 14504
    },
    {
      "epoch": 0.8122410124314033,
      "grad_norm": 1.2804858684539795,
      "learning_rate": 7.248500000000001e-05,
      "loss": 0.3913,
      "step": 14505
    },
    {
      "epoch": 0.8122970097435324,
      "grad_norm": 1.3018038272857666,
      "learning_rate": 7.249e-05,
      "loss": 0.3833,
      "step": 14506
    },
    {
      "epoch": 0.8123530070556614,
      "grad_norm": 1.3546369075775146,
      "learning_rate": 7.2495e-05,
      "loss": 0.4575,
      "step": 14507
    },
    {
      "epoch": 0.8124090043677904,
      "grad_norm": 1.5813671350479126,
      "learning_rate": 7.25e-05,
      "loss": 0.4787,
      "step": 14508
    },
    {
      "epoch": 0.8124650016799194,
      "grad_norm": 1.3506122827529907,
      "learning_rate": 7.2505e-05,
      "loss": 0.4252,
      "step": 14509
    },
    {
      "epoch": 0.8125209989920483,
      "grad_norm": 1.5459997653961182,
      "learning_rate": 7.251e-05,
      "loss": 0.479,
      "step": 14510
    },
    {
      "epoch": 0.8125769963041773,
      "grad_norm": 1.3185906410217285,
      "learning_rate": 7.2515e-05,
      "loss": 0.3693,
      "step": 14511
    },
    {
      "epoch": 0.8126329936163064,
      "grad_norm": 1.6713619232177734,
      "learning_rate": 7.252e-05,
      "loss": 0.4934,
      "step": 14512
    },
    {
      "epoch": 0.8126889909284354,
      "grad_norm": 1.160532832145691,
      "learning_rate": 7.2525e-05,
      "loss": 0.4034,
      "step": 14513
    },
    {
      "epoch": 0.8127449882405644,
      "grad_norm": 1.6146104335784912,
      "learning_rate": 7.253e-05,
      "loss": 0.3794,
      "step": 14514
    },
    {
      "epoch": 0.8128009855526934,
      "grad_norm": 1.3655707836151123,
      "learning_rate": 7.253500000000001e-05,
      "loss": 0.5995,
      "step": 14515
    },
    {
      "epoch": 0.8128569828648224,
      "grad_norm": 1.1747153997421265,
      "learning_rate": 7.254000000000001e-05,
      "loss": 0.4118,
      "step": 14516
    },
    {
      "epoch": 0.8129129801769515,
      "grad_norm": 1.23712158203125,
      "learning_rate": 7.2545e-05,
      "loss": 0.2912,
      "step": 14517
    },
    {
      "epoch": 0.8129689774890805,
      "grad_norm": 1.6816219091415405,
      "learning_rate": 7.255e-05,
      "loss": 0.4653,
      "step": 14518
    },
    {
      "epoch": 0.8130249748012095,
      "grad_norm": 1.5153015851974487,
      "learning_rate": 7.255500000000001e-05,
      "loss": 0.4508,
      "step": 14519
    },
    {
      "epoch": 0.8130809721133385,
      "grad_norm": 1.2474896907806396,
      "learning_rate": 7.256000000000001e-05,
      "loss": 0.4264,
      "step": 14520
    },
    {
      "epoch": 0.8131369694254675,
      "grad_norm": 1.249538540840149,
      "learning_rate": 7.256500000000001e-05,
      "loss": 0.4027,
      "step": 14521
    },
    {
      "epoch": 0.8131929667375966,
      "grad_norm": 1.1467176675796509,
      "learning_rate": 7.257e-05,
      "loss": 0.3656,
      "step": 14522
    },
    {
      "epoch": 0.8132489640497256,
      "grad_norm": 1.2952277660369873,
      "learning_rate": 7.2575e-05,
      "loss": 0.3393,
      "step": 14523
    },
    {
      "epoch": 0.8133049613618546,
      "grad_norm": 1.2148288488388062,
      "learning_rate": 7.258e-05,
      "loss": 0.4045,
      "step": 14524
    },
    {
      "epoch": 0.8133609586739836,
      "grad_norm": 1.3868563175201416,
      "learning_rate": 7.2585e-05,
      "loss": 0.5312,
      "step": 14525
    },
    {
      "epoch": 0.8134169559861126,
      "grad_norm": 1.3878198862075806,
      "learning_rate": 7.259000000000001e-05,
      "loss": 0.5242,
      "step": 14526
    },
    {
      "epoch": 0.8134729532982417,
      "grad_norm": 1.3271074295043945,
      "learning_rate": 7.2595e-05,
      "loss": 0.3862,
      "step": 14527
    },
    {
      "epoch": 0.8135289506103707,
      "grad_norm": 1.4036977291107178,
      "learning_rate": 7.26e-05,
      "loss": 0.5575,
      "step": 14528
    },
    {
      "epoch": 0.8135849479224997,
      "grad_norm": 1.1951898336410522,
      "learning_rate": 7.2605e-05,
      "loss": 0.3614,
      "step": 14529
    },
    {
      "epoch": 0.8136409452346287,
      "grad_norm": 1.2342334985733032,
      "learning_rate": 7.261e-05,
      "loss": 0.4671,
      "step": 14530
    },
    {
      "epoch": 0.8136969425467577,
      "grad_norm": 1.5239843130111694,
      "learning_rate": 7.2615e-05,
      "loss": 0.5482,
      "step": 14531
    },
    {
      "epoch": 0.8137529398588867,
      "grad_norm": 1.4657853841781616,
      "learning_rate": 7.261999999999999e-05,
      "loss": 0.4421,
      "step": 14532
    },
    {
      "epoch": 0.8138089371710158,
      "grad_norm": 1.2028380632400513,
      "learning_rate": 7.2625e-05,
      "loss": 0.4187,
      "step": 14533
    },
    {
      "epoch": 0.8138649344831448,
      "grad_norm": 1.6086699962615967,
      "learning_rate": 7.263e-05,
      "loss": 0.5446,
      "step": 14534
    },
    {
      "epoch": 0.8139209317952738,
      "grad_norm": 1.3743810653686523,
      "learning_rate": 7.263500000000001e-05,
      "loss": 0.4005,
      "step": 14535
    },
    {
      "epoch": 0.8139769291074028,
      "grad_norm": 1.5481847524642944,
      "learning_rate": 7.264000000000001e-05,
      "loss": 0.5506,
      "step": 14536
    },
    {
      "epoch": 0.8140329264195318,
      "grad_norm": 1.3572007417678833,
      "learning_rate": 7.2645e-05,
      "loss": 0.3101,
      "step": 14537
    },
    {
      "epoch": 0.8140889237316609,
      "grad_norm": 1.2318710088729858,
      "learning_rate": 7.265e-05,
      "loss": 0.3469,
      "step": 14538
    },
    {
      "epoch": 0.8141449210437899,
      "grad_norm": 1.2858757972717285,
      "learning_rate": 7.2655e-05,
      "loss": 0.3661,
      "step": 14539
    },
    {
      "epoch": 0.8142009183559189,
      "grad_norm": 1.3333518505096436,
      "learning_rate": 7.266000000000001e-05,
      "loss": 0.4861,
      "step": 14540
    },
    {
      "epoch": 0.8142569156680479,
      "grad_norm": 1.3598148822784424,
      "learning_rate": 7.266500000000001e-05,
      "loss": 0.4595,
      "step": 14541
    },
    {
      "epoch": 0.8143129129801769,
      "grad_norm": 1.4633821249008179,
      "learning_rate": 7.267000000000001e-05,
      "loss": 0.4354,
      "step": 14542
    },
    {
      "epoch": 0.814368910292306,
      "grad_norm": 1.4942946434020996,
      "learning_rate": 7.2675e-05,
      "loss": 0.5181,
      "step": 14543
    },
    {
      "epoch": 0.814424907604435,
      "grad_norm": 1.449614405632019,
      "learning_rate": 7.268e-05,
      "loss": 0.4668,
      "step": 14544
    },
    {
      "epoch": 0.814480904916564,
      "grad_norm": 1.1244574785232544,
      "learning_rate": 7.2685e-05,
      "loss": 0.4041,
      "step": 14545
    },
    {
      "epoch": 0.814536902228693,
      "grad_norm": 1.3382452726364136,
      "learning_rate": 7.269e-05,
      "loss": 0.4061,
      "step": 14546
    },
    {
      "epoch": 0.814592899540822,
      "grad_norm": 1.4895119667053223,
      "learning_rate": 7.269500000000001e-05,
      "loss": 0.4328,
      "step": 14547
    },
    {
      "epoch": 0.814648896852951,
      "grad_norm": 1.4116780757904053,
      "learning_rate": 7.27e-05,
      "loss": 0.5754,
      "step": 14548
    },
    {
      "epoch": 0.8147048941650801,
      "grad_norm": 1.4497257471084595,
      "learning_rate": 7.2705e-05,
      "loss": 0.5019,
      "step": 14549
    },
    {
      "epoch": 0.8147608914772091,
      "grad_norm": 1.269838571548462,
      "learning_rate": 7.271e-05,
      "loss": 0.4255,
      "step": 14550
    },
    {
      "epoch": 0.8148168887893381,
      "grad_norm": 1.9430867433547974,
      "learning_rate": 7.2715e-05,
      "loss": 0.5007,
      "step": 14551
    },
    {
      "epoch": 0.8148728861014671,
      "grad_norm": 1.705452561378479,
      "learning_rate": 7.272e-05,
      "loss": 0.5779,
      "step": 14552
    },
    {
      "epoch": 0.8149288834135962,
      "grad_norm": 1.1430672407150269,
      "learning_rate": 7.272499999999999e-05,
      "loss": 0.4957,
      "step": 14553
    },
    {
      "epoch": 0.8149848807257252,
      "grad_norm": 1.5716427564620972,
      "learning_rate": 7.273e-05,
      "loss": 0.5149,
      "step": 14554
    },
    {
      "epoch": 0.8150408780378542,
      "grad_norm": 1.5491863489151,
      "learning_rate": 7.273500000000001e-05,
      "loss": 0.4984,
      "step": 14555
    },
    {
      "epoch": 0.8150968753499832,
      "grad_norm": 1.7216907739639282,
      "learning_rate": 7.274000000000001e-05,
      "loss": 0.4223,
      "step": 14556
    },
    {
      "epoch": 0.8151528726621122,
      "grad_norm": 1.058103322982788,
      "learning_rate": 7.274500000000001e-05,
      "loss": 0.3283,
      "step": 14557
    },
    {
      "epoch": 0.8152088699742412,
      "grad_norm": 1.297997236251831,
      "learning_rate": 7.275e-05,
      "loss": 0.4349,
      "step": 14558
    },
    {
      "epoch": 0.8152648672863703,
      "grad_norm": 1.323997139930725,
      "learning_rate": 7.2755e-05,
      "loss": 0.4357,
      "step": 14559
    },
    {
      "epoch": 0.8153208645984993,
      "grad_norm": 1.1607613563537598,
      "learning_rate": 7.276e-05,
      "loss": 0.3407,
      "step": 14560
    },
    {
      "epoch": 0.8153768619106283,
      "grad_norm": 2.034667730331421,
      "learning_rate": 7.2765e-05,
      "loss": 0.5754,
      "step": 14561
    },
    {
      "epoch": 0.8154328592227573,
      "grad_norm": 1.323005199432373,
      "learning_rate": 7.277000000000001e-05,
      "loss": 0.5629,
      "step": 14562
    },
    {
      "epoch": 0.8154888565348863,
      "grad_norm": 1.2635260820388794,
      "learning_rate": 7.277500000000001e-05,
      "loss": 0.4229,
      "step": 14563
    },
    {
      "epoch": 0.8155448538470154,
      "grad_norm": 1.1667816638946533,
      "learning_rate": 7.278e-05,
      "loss": 0.3357,
      "step": 14564
    },
    {
      "epoch": 0.8156008511591444,
      "grad_norm": 1.0956239700317383,
      "learning_rate": 7.2785e-05,
      "loss": 0.3747,
      "step": 14565
    },
    {
      "epoch": 0.8156568484712734,
      "grad_norm": 3.721742868423462,
      "learning_rate": 7.279e-05,
      "loss": 0.452,
      "step": 14566
    },
    {
      "epoch": 0.8157128457834024,
      "grad_norm": 1.572638750076294,
      "learning_rate": 7.2795e-05,
      "loss": 0.5118,
      "step": 14567
    },
    {
      "epoch": 0.8157688430955314,
      "grad_norm": 1.690908432006836,
      "learning_rate": 7.280000000000001e-05,
      "loss": 0.6523,
      "step": 14568
    },
    {
      "epoch": 0.8158248404076605,
      "grad_norm": 1.463591456413269,
      "learning_rate": 7.2805e-05,
      "loss": 0.5766,
      "step": 14569
    },
    {
      "epoch": 0.8158808377197895,
      "grad_norm": 1.3467861413955688,
      "learning_rate": 7.281e-05,
      "loss": 0.4124,
      "step": 14570
    },
    {
      "epoch": 0.8159368350319185,
      "grad_norm": 1.3160570859909058,
      "learning_rate": 7.2815e-05,
      "loss": 0.4452,
      "step": 14571
    },
    {
      "epoch": 0.8159928323440475,
      "grad_norm": 1.3122376203536987,
      "learning_rate": 7.282e-05,
      "loss": 0.4212,
      "step": 14572
    },
    {
      "epoch": 0.8160488296561765,
      "grad_norm": 1.2728044986724854,
      "learning_rate": 7.2825e-05,
      "loss": 0.4936,
      "step": 14573
    },
    {
      "epoch": 0.8161048269683056,
      "grad_norm": 1.4045751094818115,
      "learning_rate": 7.282999999999999e-05,
      "loss": 0.4088,
      "step": 14574
    },
    {
      "epoch": 0.8161608242804346,
      "grad_norm": 1.1426284313201904,
      "learning_rate": 7.2835e-05,
      "loss": 0.3795,
      "step": 14575
    },
    {
      "epoch": 0.8162168215925636,
      "grad_norm": 1.2386523485183716,
      "learning_rate": 7.284000000000001e-05,
      "loss": 0.4168,
      "step": 14576
    },
    {
      "epoch": 0.8162728189046926,
      "grad_norm": 1.2142409086227417,
      "learning_rate": 7.284500000000001e-05,
      "loss": 0.398,
      "step": 14577
    },
    {
      "epoch": 0.8163288162168216,
      "grad_norm": 1.250333547592163,
      "learning_rate": 7.285000000000001e-05,
      "loss": 0.324,
      "step": 14578
    },
    {
      "epoch": 0.8163848135289506,
      "grad_norm": 1.2744383811950684,
      "learning_rate": 7.2855e-05,
      "loss": 0.4283,
      "step": 14579
    },
    {
      "epoch": 0.8164408108410797,
      "grad_norm": 1.6865350008010864,
      "learning_rate": 7.286e-05,
      "loss": 0.4068,
      "step": 14580
    },
    {
      "epoch": 0.8164968081532087,
      "grad_norm": 1.2743353843688965,
      "learning_rate": 7.2865e-05,
      "loss": 0.3827,
      "step": 14581
    },
    {
      "epoch": 0.8165528054653377,
      "grad_norm": 1.450026035308838,
      "learning_rate": 7.287e-05,
      "loss": 0.5665,
      "step": 14582
    },
    {
      "epoch": 0.8166088027774667,
      "grad_norm": 1.5260260105133057,
      "learning_rate": 7.287500000000001e-05,
      "loss": 0.3995,
      "step": 14583
    },
    {
      "epoch": 0.8166648000895957,
      "grad_norm": 2.230234146118164,
      "learning_rate": 7.288000000000001e-05,
      "loss": 0.5908,
      "step": 14584
    },
    {
      "epoch": 0.8167207974017248,
      "grad_norm": 1.2641435861587524,
      "learning_rate": 7.2885e-05,
      "loss": 0.3928,
      "step": 14585
    },
    {
      "epoch": 0.8167767947138538,
      "grad_norm": 1.340026617050171,
      "learning_rate": 7.289e-05,
      "loss": 0.4319,
      "step": 14586
    },
    {
      "epoch": 0.8168327920259828,
      "grad_norm": 1.2149425745010376,
      "learning_rate": 7.2895e-05,
      "loss": 0.3706,
      "step": 14587
    },
    {
      "epoch": 0.8168887893381118,
      "grad_norm": 1.1634408235549927,
      "learning_rate": 7.29e-05,
      "loss": 0.5095,
      "step": 14588
    },
    {
      "epoch": 0.8169447866502408,
      "grad_norm": 1.4276528358459473,
      "learning_rate": 7.290500000000001e-05,
      "loss": 0.4456,
      "step": 14589
    },
    {
      "epoch": 0.8170007839623699,
      "grad_norm": 1.5272024869918823,
      "learning_rate": 7.291e-05,
      "loss": 0.4431,
      "step": 14590
    },
    {
      "epoch": 0.8170567812744989,
      "grad_norm": 1.952351450920105,
      "learning_rate": 7.2915e-05,
      "loss": 0.611,
      "step": 14591
    },
    {
      "epoch": 0.8171127785866279,
      "grad_norm": 1.099005103111267,
      "learning_rate": 7.292e-05,
      "loss": 0.3382,
      "step": 14592
    },
    {
      "epoch": 0.8171687758987568,
      "grad_norm": 1.6788164377212524,
      "learning_rate": 7.2925e-05,
      "loss": 0.5231,
      "step": 14593
    },
    {
      "epoch": 0.8172247732108858,
      "grad_norm": 1.1630085706710815,
      "learning_rate": 7.293e-05,
      "loss": 0.3644,
      "step": 14594
    },
    {
      "epoch": 0.8172807705230148,
      "grad_norm": 1.2604320049285889,
      "learning_rate": 7.2935e-05,
      "loss": 0.519,
      "step": 14595
    },
    {
      "epoch": 0.8173367678351439,
      "grad_norm": 5.7034831047058105,
      "learning_rate": 7.294e-05,
      "loss": 0.4025,
      "step": 14596
    },
    {
      "epoch": 0.8173927651472729,
      "grad_norm": 1.554646372795105,
      "learning_rate": 7.294500000000001e-05,
      "loss": 0.4582,
      "step": 14597
    },
    {
      "epoch": 0.8174487624594019,
      "grad_norm": 1.5256388187408447,
      "learning_rate": 7.295000000000001e-05,
      "loss": 0.4781,
      "step": 14598
    },
    {
      "epoch": 0.8175047597715309,
      "grad_norm": 1.668744683265686,
      "learning_rate": 7.295500000000001e-05,
      "loss": 0.7531,
      "step": 14599
    },
    {
      "epoch": 0.8175607570836599,
      "grad_norm": 1.4327677488327026,
      "learning_rate": 7.296e-05,
      "loss": 0.7833,
      "step": 14600
    },
    {
      "epoch": 0.817616754395789,
      "grad_norm": 1.1870002746582031,
      "learning_rate": 7.2965e-05,
      "loss": 0.4161,
      "step": 14601
    },
    {
      "epoch": 0.817672751707918,
      "grad_norm": 1.5035710334777832,
      "learning_rate": 7.297e-05,
      "loss": 0.3995,
      "step": 14602
    },
    {
      "epoch": 0.817728749020047,
      "grad_norm": 1.594335913658142,
      "learning_rate": 7.2975e-05,
      "loss": 0.52,
      "step": 14603
    },
    {
      "epoch": 0.817784746332176,
      "grad_norm": 1.2316128015518188,
      "learning_rate": 7.298000000000001e-05,
      "loss": 0.424,
      "step": 14604
    },
    {
      "epoch": 0.817840743644305,
      "grad_norm": 1.5011049509048462,
      "learning_rate": 7.298500000000001e-05,
      "loss": 0.4467,
      "step": 14605
    },
    {
      "epoch": 0.817896740956434,
      "grad_norm": 1.3760048151016235,
      "learning_rate": 7.299e-05,
      "loss": 0.3854,
      "step": 14606
    },
    {
      "epoch": 0.8179527382685631,
      "grad_norm": 1.1981438398361206,
      "learning_rate": 7.2995e-05,
      "loss": 0.5182,
      "step": 14607
    },
    {
      "epoch": 0.8180087355806921,
      "grad_norm": 1.446954369544983,
      "learning_rate": 7.3e-05,
      "loss": 0.5173,
      "step": 14608
    },
    {
      "epoch": 0.8180647328928211,
      "grad_norm": 1.686732530593872,
      "learning_rate": 7.3005e-05,
      "loss": 0.56,
      "step": 14609
    },
    {
      "epoch": 0.8181207302049501,
      "grad_norm": 1.051320195198059,
      "learning_rate": 7.301e-05,
      "loss": 0.3697,
      "step": 14610
    },
    {
      "epoch": 0.8181767275170791,
      "grad_norm": 1.1930115222930908,
      "learning_rate": 7.3015e-05,
      "loss": 0.4837,
      "step": 14611
    },
    {
      "epoch": 0.8182327248292082,
      "grad_norm": 1.5659046173095703,
      "learning_rate": 7.302e-05,
      "loss": 0.4421,
      "step": 14612
    },
    {
      "epoch": 0.8182887221413372,
      "grad_norm": 1.5599424839019775,
      "learning_rate": 7.3025e-05,
      "loss": 0.4769,
      "step": 14613
    },
    {
      "epoch": 0.8183447194534662,
      "grad_norm": 1.2555923461914062,
      "learning_rate": 7.303e-05,
      "loss": 0.4743,
      "step": 14614
    },
    {
      "epoch": 0.8184007167655952,
      "grad_norm": 1.4767690896987915,
      "learning_rate": 7.303500000000001e-05,
      "loss": 0.4749,
      "step": 14615
    },
    {
      "epoch": 0.8184567140777242,
      "grad_norm": 1.3402944803237915,
      "learning_rate": 7.304e-05,
      "loss": 0.5187,
      "step": 14616
    },
    {
      "epoch": 0.8185127113898533,
      "grad_norm": 1.3602509498596191,
      "learning_rate": 7.3045e-05,
      "loss": 0.4312,
      "step": 14617
    },
    {
      "epoch": 0.8185687087019823,
      "grad_norm": 1.3979710340499878,
      "learning_rate": 7.305000000000001e-05,
      "loss": 0.4078,
      "step": 14618
    },
    {
      "epoch": 0.8186247060141113,
      "grad_norm": 1.2349612712860107,
      "learning_rate": 7.305500000000001e-05,
      "loss": 0.3946,
      "step": 14619
    },
    {
      "epoch": 0.8186807033262403,
      "grad_norm": 1.2803958654403687,
      "learning_rate": 7.306000000000001e-05,
      "loss": 0.3911,
      "step": 14620
    },
    {
      "epoch": 0.8187367006383693,
      "grad_norm": 1.5141361951828003,
      "learning_rate": 7.3065e-05,
      "loss": 0.4704,
      "step": 14621
    },
    {
      "epoch": 0.8187926979504984,
      "grad_norm": 1.7877403497695923,
      "learning_rate": 7.307e-05,
      "loss": 0.5888,
      "step": 14622
    },
    {
      "epoch": 0.8188486952626274,
      "grad_norm": 1.3140099048614502,
      "learning_rate": 7.3075e-05,
      "loss": 0.3834,
      "step": 14623
    },
    {
      "epoch": 0.8189046925747564,
      "grad_norm": 1.3405500650405884,
      "learning_rate": 7.308e-05,
      "loss": 0.4656,
      "step": 14624
    },
    {
      "epoch": 0.8189606898868854,
      "grad_norm": 1.3182367086410522,
      "learning_rate": 7.308500000000001e-05,
      "loss": 0.3299,
      "step": 14625
    },
    {
      "epoch": 0.8190166871990144,
      "grad_norm": 1.8528672456741333,
      "learning_rate": 7.309e-05,
      "loss": 0.5492,
      "step": 14626
    },
    {
      "epoch": 0.8190726845111435,
      "grad_norm": 1.2025448083877563,
      "learning_rate": 7.3095e-05,
      "loss": 0.4631,
      "step": 14627
    },
    {
      "epoch": 0.8191286818232725,
      "grad_norm": 1.1996395587921143,
      "learning_rate": 7.31e-05,
      "loss": 0.4278,
      "step": 14628
    },
    {
      "epoch": 0.8191846791354015,
      "grad_norm": 1.288961410522461,
      "learning_rate": 7.3105e-05,
      "loss": 0.4867,
      "step": 14629
    },
    {
      "epoch": 0.8192406764475305,
      "grad_norm": 1.4086191654205322,
      "learning_rate": 7.311e-05,
      "loss": 0.463,
      "step": 14630
    },
    {
      "epoch": 0.8192966737596595,
      "grad_norm": 1.3368200063705444,
      "learning_rate": 7.3115e-05,
      "loss": 0.4777,
      "step": 14631
    },
    {
      "epoch": 0.8193526710717886,
      "grad_norm": 0.9781476855278015,
      "learning_rate": 7.312e-05,
      "loss": 0.3705,
      "step": 14632
    },
    {
      "epoch": 0.8194086683839176,
      "grad_norm": 1.185136318206787,
      "learning_rate": 7.3125e-05,
      "loss": 0.5119,
      "step": 14633
    },
    {
      "epoch": 0.8194646656960466,
      "grad_norm": 1.1368544101715088,
      "learning_rate": 7.313e-05,
      "loss": 0.4276,
      "step": 14634
    },
    {
      "epoch": 0.8195206630081756,
      "grad_norm": 1.0748132467269897,
      "learning_rate": 7.3135e-05,
      "loss": 0.3838,
      "step": 14635
    },
    {
      "epoch": 0.8195766603203046,
      "grad_norm": 1.304819941520691,
      "learning_rate": 7.314000000000001e-05,
      "loss": 0.4756,
      "step": 14636
    },
    {
      "epoch": 0.8196326576324336,
      "grad_norm": 1.168513298034668,
      "learning_rate": 7.3145e-05,
      "loss": 0.3447,
      "step": 14637
    },
    {
      "epoch": 0.8196886549445627,
      "grad_norm": 1.4905389547348022,
      "learning_rate": 7.315e-05,
      "loss": 0.5992,
      "step": 14638
    },
    {
      "epoch": 0.8197446522566917,
      "grad_norm": 1.1134607791900635,
      "learning_rate": 7.315500000000001e-05,
      "loss": 0.4236,
      "step": 14639
    },
    {
      "epoch": 0.8198006495688207,
      "grad_norm": 1.1270670890808105,
      "learning_rate": 7.316000000000001e-05,
      "loss": 0.354,
      "step": 14640
    },
    {
      "epoch": 0.8198566468809497,
      "grad_norm": 1.2572729587554932,
      "learning_rate": 7.316500000000001e-05,
      "loss": 0.3315,
      "step": 14641
    },
    {
      "epoch": 0.8199126441930787,
      "grad_norm": 1.289846420288086,
      "learning_rate": 7.317e-05,
      "loss": 0.3378,
      "step": 14642
    },
    {
      "epoch": 0.8199686415052078,
      "grad_norm": 1.543160080909729,
      "learning_rate": 7.3175e-05,
      "loss": 0.5779,
      "step": 14643
    },
    {
      "epoch": 0.8200246388173368,
      "grad_norm": 1.4106050729751587,
      "learning_rate": 7.318e-05,
      "loss": 0.4014,
      "step": 14644
    },
    {
      "epoch": 0.8200806361294658,
      "grad_norm": 1.1911691427230835,
      "learning_rate": 7.3185e-05,
      "loss": 0.4199,
      "step": 14645
    },
    {
      "epoch": 0.8201366334415948,
      "grad_norm": 1.0628832578659058,
      "learning_rate": 7.319000000000001e-05,
      "loss": 0.4342,
      "step": 14646
    },
    {
      "epoch": 0.8201926307537238,
      "grad_norm": 1.2558344602584839,
      "learning_rate": 7.3195e-05,
      "loss": 0.4735,
      "step": 14647
    },
    {
      "epoch": 0.8202486280658529,
      "grad_norm": 1.5138523578643799,
      "learning_rate": 7.32e-05,
      "loss": 0.6924,
      "step": 14648
    },
    {
      "epoch": 0.8203046253779819,
      "grad_norm": 1.438155174255371,
      "learning_rate": 7.3205e-05,
      "loss": 0.4288,
      "step": 14649
    },
    {
      "epoch": 0.8203606226901109,
      "grad_norm": 1.302852749824524,
      "learning_rate": 7.321e-05,
      "loss": 0.4153,
      "step": 14650
    },
    {
      "epoch": 0.8204166200022399,
      "grad_norm": 1.169129729270935,
      "learning_rate": 7.3215e-05,
      "loss": 0.3858,
      "step": 14651
    },
    {
      "epoch": 0.8204726173143689,
      "grad_norm": 1.8877336978912354,
      "learning_rate": 7.322e-05,
      "loss": 0.5608,
      "step": 14652
    },
    {
      "epoch": 0.820528614626498,
      "grad_norm": 1.320983648300171,
      "learning_rate": 7.3225e-05,
      "loss": 0.4959,
      "step": 14653
    },
    {
      "epoch": 0.820584611938627,
      "grad_norm": 1.4541881084442139,
      "learning_rate": 7.323e-05,
      "loss": 0.5543,
      "step": 14654
    },
    {
      "epoch": 0.820640609250756,
      "grad_norm": 1.3759751319885254,
      "learning_rate": 7.3235e-05,
      "loss": 0.4852,
      "step": 14655
    },
    {
      "epoch": 0.820696606562885,
      "grad_norm": 1.4528175592422485,
      "learning_rate": 7.324000000000001e-05,
      "loss": 0.478,
      "step": 14656
    },
    {
      "epoch": 0.820752603875014,
      "grad_norm": 1.3313637971878052,
      "learning_rate": 7.324500000000001e-05,
      "loss": 0.4429,
      "step": 14657
    },
    {
      "epoch": 0.820808601187143,
      "grad_norm": 1.0332139730453491,
      "learning_rate": 7.325e-05,
      "loss": 0.2853,
      "step": 14658
    },
    {
      "epoch": 0.8208645984992721,
      "grad_norm": 1.4055781364440918,
      "learning_rate": 7.3255e-05,
      "loss": 0.4461,
      "step": 14659
    },
    {
      "epoch": 0.8209205958114011,
      "grad_norm": 1.1871622800827026,
      "learning_rate": 7.326e-05,
      "loss": 0.5183,
      "step": 14660
    },
    {
      "epoch": 0.8209765931235301,
      "grad_norm": 1.436928391456604,
      "learning_rate": 7.326500000000001e-05,
      "loss": 0.4143,
      "step": 14661
    },
    {
      "epoch": 0.8210325904356591,
      "grad_norm": 1.2425044775009155,
      "learning_rate": 7.327000000000001e-05,
      "loss": 0.4609,
      "step": 14662
    },
    {
      "epoch": 0.8210885877477881,
      "grad_norm": 1.4125325679779053,
      "learning_rate": 7.3275e-05,
      "loss": 0.4595,
      "step": 14663
    },
    {
      "epoch": 0.8211445850599172,
      "grad_norm": 1.0556050539016724,
      "learning_rate": 7.328e-05,
      "loss": 0.3873,
      "step": 14664
    },
    {
      "epoch": 0.8212005823720462,
      "grad_norm": 1.2359877824783325,
      "learning_rate": 7.3285e-05,
      "loss": 0.3886,
      "step": 14665
    },
    {
      "epoch": 0.8212565796841752,
      "grad_norm": 1.4855868816375732,
      "learning_rate": 7.329e-05,
      "loss": 0.5341,
      "step": 14666
    },
    {
      "epoch": 0.8213125769963042,
      "grad_norm": 3.715852975845337,
      "learning_rate": 7.329500000000001e-05,
      "loss": 0.4834,
      "step": 14667
    },
    {
      "epoch": 0.8213685743084332,
      "grad_norm": 1.4016079902648926,
      "learning_rate": 7.33e-05,
      "loss": 0.502,
      "step": 14668
    },
    {
      "epoch": 0.8214245716205623,
      "grad_norm": 1.375775694847107,
      "learning_rate": 7.3305e-05,
      "loss": 0.3553,
      "step": 14669
    },
    {
      "epoch": 0.8214805689326913,
      "grad_norm": 1.2052714824676514,
      "learning_rate": 7.331e-05,
      "loss": 0.3535,
      "step": 14670
    },
    {
      "epoch": 0.8215365662448203,
      "grad_norm": 1.445199966430664,
      "learning_rate": 7.3315e-05,
      "loss": 0.5214,
      "step": 14671
    },
    {
      "epoch": 0.8215925635569493,
      "grad_norm": 1.3662270307540894,
      "learning_rate": 7.332e-05,
      "loss": 0.4388,
      "step": 14672
    },
    {
      "epoch": 0.8216485608690783,
      "grad_norm": 1.2618768215179443,
      "learning_rate": 7.3325e-05,
      "loss": 0.6232,
      "step": 14673
    },
    {
      "epoch": 0.8217045581812074,
      "grad_norm": 1.0770246982574463,
      "learning_rate": 7.333e-05,
      "loss": 0.3527,
      "step": 14674
    },
    {
      "epoch": 0.8217605554933363,
      "grad_norm": 1.4090774059295654,
      "learning_rate": 7.3335e-05,
      "loss": 0.4901,
      "step": 14675
    },
    {
      "epoch": 0.8218165528054653,
      "grad_norm": 1.4112480878829956,
      "learning_rate": 7.334000000000001e-05,
      "loss": 0.4592,
      "step": 14676
    },
    {
      "epoch": 0.8218725501175943,
      "grad_norm": 1.2244633436203003,
      "learning_rate": 7.334500000000001e-05,
      "loss": 0.3995,
      "step": 14677
    },
    {
      "epoch": 0.8219285474297233,
      "grad_norm": 1.3836400508880615,
      "learning_rate": 7.335000000000001e-05,
      "loss": 0.4217,
      "step": 14678
    },
    {
      "epoch": 0.8219845447418523,
      "grad_norm": 1.394639015197754,
      "learning_rate": 7.3355e-05,
      "loss": 0.3731,
      "step": 14679
    },
    {
      "epoch": 0.8220405420539814,
      "grad_norm": 1.2811203002929688,
      "learning_rate": 7.336e-05,
      "loss": 0.4087,
      "step": 14680
    },
    {
      "epoch": 0.8220965393661104,
      "grad_norm": 1.214834213256836,
      "learning_rate": 7.3365e-05,
      "loss": 0.467,
      "step": 14681
    },
    {
      "epoch": 0.8221525366782394,
      "grad_norm": 1.3365193605422974,
      "learning_rate": 7.337000000000001e-05,
      "loss": 0.6025,
      "step": 14682
    },
    {
      "epoch": 0.8222085339903684,
      "grad_norm": 1.3410736322402954,
      "learning_rate": 7.337500000000001e-05,
      "loss": 0.4645,
      "step": 14683
    },
    {
      "epoch": 0.8222645313024974,
      "grad_norm": 1.3538296222686768,
      "learning_rate": 7.338e-05,
      "loss": 0.4959,
      "step": 14684
    },
    {
      "epoch": 0.8223205286146265,
      "grad_norm": 1.3033719062805176,
      "learning_rate": 7.3385e-05,
      "loss": 0.4946,
      "step": 14685
    },
    {
      "epoch": 0.8223765259267555,
      "grad_norm": 1.2510207891464233,
      "learning_rate": 7.339e-05,
      "loss": 0.5141,
      "step": 14686
    },
    {
      "epoch": 0.8224325232388845,
      "grad_norm": 1.329649567604065,
      "learning_rate": 7.3395e-05,
      "loss": 0.4308,
      "step": 14687
    },
    {
      "epoch": 0.8224885205510135,
      "grad_norm": 1.3454447984695435,
      "learning_rate": 7.340000000000001e-05,
      "loss": 0.4015,
      "step": 14688
    },
    {
      "epoch": 0.8225445178631425,
      "grad_norm": 1.414605975151062,
      "learning_rate": 7.3405e-05,
      "loss": 0.4734,
      "step": 14689
    },
    {
      "epoch": 0.8226005151752716,
      "grad_norm": 1.2809250354766846,
      "learning_rate": 7.341e-05,
      "loss": 0.2809,
      "step": 14690
    },
    {
      "epoch": 0.8226565124874006,
      "grad_norm": 1.1174598932266235,
      "learning_rate": 7.3415e-05,
      "loss": 0.402,
      "step": 14691
    },
    {
      "epoch": 0.8227125097995296,
      "grad_norm": 1.1026337146759033,
      "learning_rate": 7.342e-05,
      "loss": 0.5006,
      "step": 14692
    },
    {
      "epoch": 0.8227685071116586,
      "grad_norm": 1.4404120445251465,
      "learning_rate": 7.3425e-05,
      "loss": 0.5915,
      "step": 14693
    },
    {
      "epoch": 0.8228245044237876,
      "grad_norm": 1.2131141424179077,
      "learning_rate": 7.342999999999999e-05,
      "loss": 0.4807,
      "step": 14694
    },
    {
      "epoch": 0.8228805017359166,
      "grad_norm": 2.460108757019043,
      "learning_rate": 7.3435e-05,
      "loss": 0.4563,
      "step": 14695
    },
    {
      "epoch": 0.8229364990480457,
      "grad_norm": 1.672230839729309,
      "learning_rate": 7.344000000000002e-05,
      "loss": 0.5001,
      "step": 14696
    },
    {
      "epoch": 0.8229924963601747,
      "grad_norm": 1.2148809432983398,
      "learning_rate": 7.344500000000001e-05,
      "loss": 0.4161,
      "step": 14697
    },
    {
      "epoch": 0.8230484936723037,
      "grad_norm": 1.3984878063201904,
      "learning_rate": 7.345000000000001e-05,
      "loss": 0.4249,
      "step": 14698
    },
    {
      "epoch": 0.8231044909844327,
      "grad_norm": 1.5426331758499146,
      "learning_rate": 7.345500000000001e-05,
      "loss": 0.6712,
      "step": 14699
    },
    {
      "epoch": 0.8231604882965617,
      "grad_norm": 1.246315836906433,
      "learning_rate": 7.346e-05,
      "loss": 0.5127,
      "step": 14700
    },
    {
      "epoch": 0.8232164856086908,
      "grad_norm": 1.3528317213058472,
      "learning_rate": 7.3465e-05,
      "loss": 0.4185,
      "step": 14701
    },
    {
      "epoch": 0.8232724829208198,
      "grad_norm": 2.795318841934204,
      "learning_rate": 7.347e-05,
      "loss": 0.3799,
      "step": 14702
    },
    {
      "epoch": 0.8233284802329488,
      "grad_norm": 1.1334058046340942,
      "learning_rate": 7.347500000000001e-05,
      "loss": 0.4266,
      "step": 14703
    },
    {
      "epoch": 0.8233844775450778,
      "grad_norm": 1.4261080026626587,
      "learning_rate": 7.348000000000001e-05,
      "loss": 0.4662,
      "step": 14704
    },
    {
      "epoch": 0.8234404748572068,
      "grad_norm": 1.3351155519485474,
      "learning_rate": 7.3485e-05,
      "loss": 0.4606,
      "step": 14705
    },
    {
      "epoch": 0.8234964721693359,
      "grad_norm": 1.4264384508132935,
      "learning_rate": 7.349e-05,
      "loss": 0.4275,
      "step": 14706
    },
    {
      "epoch": 0.8235524694814649,
      "grad_norm": 1.2684037685394287,
      "learning_rate": 7.3495e-05,
      "loss": 0.4041,
      "step": 14707
    },
    {
      "epoch": 0.8236084667935939,
      "grad_norm": 1.1517019271850586,
      "learning_rate": 7.35e-05,
      "loss": 0.3418,
      "step": 14708
    },
    {
      "epoch": 0.8236644641057229,
      "grad_norm": 1.2840877771377563,
      "learning_rate": 7.3505e-05,
      "loss": 0.4765,
      "step": 14709
    },
    {
      "epoch": 0.8237204614178519,
      "grad_norm": 1.1933214664459229,
      "learning_rate": 7.351e-05,
      "loss": 0.4337,
      "step": 14710
    },
    {
      "epoch": 0.823776458729981,
      "grad_norm": 1.3279502391815186,
      "learning_rate": 7.3515e-05,
      "loss": 0.3734,
      "step": 14711
    },
    {
      "epoch": 0.82383245604211,
      "grad_norm": 1.504241943359375,
      "learning_rate": 7.352e-05,
      "loss": 0.3383,
      "step": 14712
    },
    {
      "epoch": 0.823888453354239,
      "grad_norm": 1.501292109489441,
      "learning_rate": 7.3525e-05,
      "loss": 0.4715,
      "step": 14713
    },
    {
      "epoch": 0.823944450666368,
      "grad_norm": 1.2501533031463623,
      "learning_rate": 7.353e-05,
      "loss": 0.3866,
      "step": 14714
    },
    {
      "epoch": 0.824000447978497,
      "grad_norm": 1.207091212272644,
      "learning_rate": 7.353499999999999e-05,
      "loss": 0.4442,
      "step": 14715
    },
    {
      "epoch": 0.824056445290626,
      "grad_norm": 1.4631602764129639,
      "learning_rate": 7.354e-05,
      "loss": 0.445,
      "step": 14716
    },
    {
      "epoch": 0.8241124426027551,
      "grad_norm": 1.2972570657730103,
      "learning_rate": 7.354500000000001e-05,
      "loss": 0.4775,
      "step": 14717
    },
    {
      "epoch": 0.8241684399148841,
      "grad_norm": 1.363891363143921,
      "learning_rate": 7.355000000000001e-05,
      "loss": 0.3694,
      "step": 14718
    },
    {
      "epoch": 0.8242244372270131,
      "grad_norm": 1.2534159421920776,
      "learning_rate": 7.355500000000001e-05,
      "loss": 0.3984,
      "step": 14719
    },
    {
      "epoch": 0.8242804345391421,
      "grad_norm": 2.2425355911254883,
      "learning_rate": 7.356000000000001e-05,
      "loss": 0.4672,
      "step": 14720
    },
    {
      "epoch": 0.8243364318512711,
      "grad_norm": 1.1252158880233765,
      "learning_rate": 7.3565e-05,
      "loss": 0.3602,
      "step": 14721
    },
    {
      "epoch": 0.8243924291634002,
      "grad_norm": 1.5778850317001343,
      "learning_rate": 7.357e-05,
      "loss": 0.5639,
      "step": 14722
    },
    {
      "epoch": 0.8244484264755292,
      "grad_norm": 1.4036636352539062,
      "learning_rate": 7.3575e-05,
      "loss": 0.5913,
      "step": 14723
    },
    {
      "epoch": 0.8245044237876582,
      "grad_norm": 1.3537404537200928,
      "learning_rate": 7.358000000000001e-05,
      "loss": 0.4174,
      "step": 14724
    },
    {
      "epoch": 0.8245604210997872,
      "grad_norm": 1.240222692489624,
      "learning_rate": 7.358500000000001e-05,
      "loss": 0.5879,
      "step": 14725
    },
    {
      "epoch": 0.8246164184119162,
      "grad_norm": 1.2327338457107544,
      "learning_rate": 7.359e-05,
      "loss": 0.4506,
      "step": 14726
    },
    {
      "epoch": 0.8246724157240453,
      "grad_norm": 1.5453565120697021,
      "learning_rate": 7.3595e-05,
      "loss": 0.4951,
      "step": 14727
    },
    {
      "epoch": 0.8247284130361743,
      "grad_norm": 1.184607744216919,
      "learning_rate": 7.36e-05,
      "loss": 0.3828,
      "step": 14728
    },
    {
      "epoch": 0.8247844103483033,
      "grad_norm": 1.2312238216400146,
      "learning_rate": 7.3605e-05,
      "loss": 0.4146,
      "step": 14729
    },
    {
      "epoch": 0.8248404076604323,
      "grad_norm": 1.4124467372894287,
      "learning_rate": 7.361e-05,
      "loss": 0.5357,
      "step": 14730
    },
    {
      "epoch": 0.8248964049725613,
      "grad_norm": 1.378571629524231,
      "learning_rate": 7.3615e-05,
      "loss": 0.4415,
      "step": 14731
    },
    {
      "epoch": 0.8249524022846904,
      "grad_norm": 1.1218804121017456,
      "learning_rate": 7.362e-05,
      "loss": 0.3804,
      "step": 14732
    },
    {
      "epoch": 0.8250083995968194,
      "grad_norm": 1.1442683935165405,
      "learning_rate": 7.3625e-05,
      "loss": 0.3825,
      "step": 14733
    },
    {
      "epoch": 0.8250643969089484,
      "grad_norm": 1.2137649059295654,
      "learning_rate": 7.363e-05,
      "loss": 0.3702,
      "step": 14734
    },
    {
      "epoch": 0.8251203942210774,
      "grad_norm": 1.3204543590545654,
      "learning_rate": 7.3635e-05,
      "loss": 0.4778,
      "step": 14735
    },
    {
      "epoch": 0.8251763915332064,
      "grad_norm": 1.361220359802246,
      "learning_rate": 7.364e-05,
      "loss": 0.3699,
      "step": 14736
    },
    {
      "epoch": 0.8252323888453355,
      "grad_norm": 1.2736396789550781,
      "learning_rate": 7.3645e-05,
      "loss": 0.4897,
      "step": 14737
    },
    {
      "epoch": 0.8252883861574645,
      "grad_norm": 1.2590088844299316,
      "learning_rate": 7.365e-05,
      "loss": 0.4012,
      "step": 14738
    },
    {
      "epoch": 0.8253443834695935,
      "grad_norm": 1.4426013231277466,
      "learning_rate": 7.365500000000001e-05,
      "loss": 0.5248,
      "step": 14739
    },
    {
      "epoch": 0.8254003807817225,
      "grad_norm": 1.2324867248535156,
      "learning_rate": 7.366000000000001e-05,
      "loss": 0.3447,
      "step": 14740
    },
    {
      "epoch": 0.8254563780938515,
      "grad_norm": 1.4885910749435425,
      "learning_rate": 7.366500000000001e-05,
      "loss": 0.3961,
      "step": 14741
    },
    {
      "epoch": 0.8255123754059805,
      "grad_norm": 1.3118256330490112,
      "learning_rate": 7.367e-05,
      "loss": 0.428,
      "step": 14742
    },
    {
      "epoch": 0.8255683727181096,
      "grad_norm": 1.4559918642044067,
      "learning_rate": 7.3675e-05,
      "loss": 0.5029,
      "step": 14743
    },
    {
      "epoch": 0.8256243700302386,
      "grad_norm": 1.3392595052719116,
      "learning_rate": 7.368e-05,
      "loss": 0.4989,
      "step": 14744
    },
    {
      "epoch": 0.8256803673423676,
      "grad_norm": 1.6044344902038574,
      "learning_rate": 7.368500000000001e-05,
      "loss": 0.5751,
      "step": 14745
    },
    {
      "epoch": 0.8257363646544966,
      "grad_norm": 1.5788519382476807,
      "learning_rate": 7.369000000000001e-05,
      "loss": 0.4325,
      "step": 14746
    },
    {
      "epoch": 0.8257923619666256,
      "grad_norm": 1.4509316682815552,
      "learning_rate": 7.3695e-05,
      "loss": 0.5399,
      "step": 14747
    },
    {
      "epoch": 0.8258483592787547,
      "grad_norm": 1.5302283763885498,
      "learning_rate": 7.37e-05,
      "loss": 0.4131,
      "step": 14748
    },
    {
      "epoch": 0.8259043565908837,
      "grad_norm": 1.2519524097442627,
      "learning_rate": 7.3705e-05,
      "loss": 0.523,
      "step": 14749
    },
    {
      "epoch": 0.8259603539030127,
      "grad_norm": 0.9893085360527039,
      "learning_rate": 7.371e-05,
      "loss": 0.3256,
      "step": 14750
    },
    {
      "epoch": 0.8260163512151417,
      "grad_norm": 1.1434038877487183,
      "learning_rate": 7.3715e-05,
      "loss": 0.318,
      "step": 14751
    },
    {
      "epoch": 0.8260723485272707,
      "grad_norm": 1.723732352256775,
      "learning_rate": 7.372e-05,
      "loss": 0.5284,
      "step": 14752
    },
    {
      "epoch": 0.8261283458393998,
      "grad_norm": 1.2246155738830566,
      "learning_rate": 7.3725e-05,
      "loss": 0.4872,
      "step": 14753
    },
    {
      "epoch": 0.8261843431515288,
      "grad_norm": 1.3805774450302124,
      "learning_rate": 7.373e-05,
      "loss": 0.4494,
      "step": 14754
    },
    {
      "epoch": 0.8262403404636578,
      "grad_norm": 1.4114630222320557,
      "learning_rate": 7.3735e-05,
      "loss": 0.484,
      "step": 14755
    },
    {
      "epoch": 0.8262963377757868,
      "grad_norm": 1.6827677488327026,
      "learning_rate": 7.374000000000001e-05,
      "loss": 0.5552,
      "step": 14756
    },
    {
      "epoch": 0.8263523350879158,
      "grad_norm": 1.3255912065505981,
      "learning_rate": 7.3745e-05,
      "loss": 0.4747,
      "step": 14757
    },
    {
      "epoch": 0.8264083324000447,
      "grad_norm": 1.1039899587631226,
      "learning_rate": 7.375e-05,
      "loss": 0.4505,
      "step": 14758
    },
    {
      "epoch": 0.8264643297121738,
      "grad_norm": 1.1913259029388428,
      "learning_rate": 7.3755e-05,
      "loss": 0.4721,
      "step": 14759
    },
    {
      "epoch": 0.8265203270243028,
      "grad_norm": 1.2081841230392456,
      "learning_rate": 7.376000000000001e-05,
      "loss": 0.3813,
      "step": 14760
    },
    {
      "epoch": 0.8265763243364318,
      "grad_norm": 1.3080998659133911,
      "learning_rate": 7.376500000000001e-05,
      "loss": 0.4688,
      "step": 14761
    },
    {
      "epoch": 0.8266323216485608,
      "grad_norm": 1.1383557319641113,
      "learning_rate": 7.377000000000001e-05,
      "loss": 0.3989,
      "step": 14762
    },
    {
      "epoch": 0.8266883189606898,
      "grad_norm": 1.406874179840088,
      "learning_rate": 7.3775e-05,
      "loss": 0.5117,
      "step": 14763
    },
    {
      "epoch": 0.8267443162728189,
      "grad_norm": 1.470577597618103,
      "learning_rate": 7.378e-05,
      "loss": 0.4923,
      "step": 14764
    },
    {
      "epoch": 0.8268003135849479,
      "grad_norm": 1.519362449645996,
      "learning_rate": 7.3785e-05,
      "loss": 0.4208,
      "step": 14765
    },
    {
      "epoch": 0.8268563108970769,
      "grad_norm": 1.3467808961868286,
      "learning_rate": 7.379000000000001e-05,
      "loss": 0.4447,
      "step": 14766
    },
    {
      "epoch": 0.8269123082092059,
      "grad_norm": 1.2084959745407104,
      "learning_rate": 7.379500000000001e-05,
      "loss": 0.4171,
      "step": 14767
    },
    {
      "epoch": 0.8269683055213349,
      "grad_norm": 1.4759771823883057,
      "learning_rate": 7.38e-05,
      "loss": 0.3606,
      "step": 14768
    },
    {
      "epoch": 0.827024302833464,
      "grad_norm": 1.2212929725646973,
      "learning_rate": 7.3805e-05,
      "loss": 0.4007,
      "step": 14769
    },
    {
      "epoch": 0.827080300145593,
      "grad_norm": 1.5924476385116577,
      "learning_rate": 7.381e-05,
      "loss": 0.3969,
      "step": 14770
    },
    {
      "epoch": 0.827136297457722,
      "grad_norm": 1.227035641670227,
      "learning_rate": 7.3815e-05,
      "loss": 0.5113,
      "step": 14771
    },
    {
      "epoch": 0.827192294769851,
      "grad_norm": 1.3617016077041626,
      "learning_rate": 7.382e-05,
      "loss": 0.4139,
      "step": 14772
    },
    {
      "epoch": 0.82724829208198,
      "grad_norm": 1.984702706336975,
      "learning_rate": 7.3825e-05,
      "loss": 0.5963,
      "step": 14773
    },
    {
      "epoch": 0.827304289394109,
      "grad_norm": 1.2086858749389648,
      "learning_rate": 7.383e-05,
      "loss": 0.4192,
      "step": 14774
    },
    {
      "epoch": 0.8273602867062381,
      "grad_norm": 1.24480140209198,
      "learning_rate": 7.3835e-05,
      "loss": 0.4102,
      "step": 14775
    },
    {
      "epoch": 0.8274162840183671,
      "grad_norm": 1.2117775678634644,
      "learning_rate": 7.384e-05,
      "loss": 0.5034,
      "step": 14776
    },
    {
      "epoch": 0.8274722813304961,
      "grad_norm": 1.4940330982208252,
      "learning_rate": 7.384500000000001e-05,
      "loss": 0.4258,
      "step": 14777
    },
    {
      "epoch": 0.8275282786426251,
      "grad_norm": 1.0816054344177246,
      "learning_rate": 7.385e-05,
      "loss": 0.3952,
      "step": 14778
    },
    {
      "epoch": 0.8275842759547541,
      "grad_norm": 1.1663109064102173,
      "learning_rate": 7.3855e-05,
      "loss": 0.3236,
      "step": 14779
    },
    {
      "epoch": 0.8276402732668832,
      "grad_norm": 1.3746048212051392,
      "learning_rate": 7.386e-05,
      "loss": 0.3799,
      "step": 14780
    },
    {
      "epoch": 0.8276962705790122,
      "grad_norm": 1.4794203042984009,
      "learning_rate": 7.386500000000001e-05,
      "loss": 0.4392,
      "step": 14781
    },
    {
      "epoch": 0.8277522678911412,
      "grad_norm": 1.4070956707000732,
      "learning_rate": 7.387000000000001e-05,
      "loss": 0.4524,
      "step": 14782
    },
    {
      "epoch": 0.8278082652032702,
      "grad_norm": 1.380784511566162,
      "learning_rate": 7.3875e-05,
      "loss": 0.3083,
      "step": 14783
    },
    {
      "epoch": 0.8278642625153992,
      "grad_norm": 1.4081639051437378,
      "learning_rate": 7.388e-05,
      "loss": 0.5004,
      "step": 14784
    },
    {
      "epoch": 0.8279202598275283,
      "grad_norm": 1.2172077894210815,
      "learning_rate": 7.3885e-05,
      "loss": 0.4313,
      "step": 14785
    },
    {
      "epoch": 0.8279762571396573,
      "grad_norm": 1.530921459197998,
      "learning_rate": 7.389e-05,
      "loss": 0.6821,
      "step": 14786
    },
    {
      "epoch": 0.8280322544517863,
      "grad_norm": 1.1566896438598633,
      "learning_rate": 7.3895e-05,
      "loss": 0.4739,
      "step": 14787
    },
    {
      "epoch": 0.8280882517639153,
      "grad_norm": 1.351818561553955,
      "learning_rate": 7.390000000000001e-05,
      "loss": 0.4297,
      "step": 14788
    },
    {
      "epoch": 0.8281442490760443,
      "grad_norm": 1.5123695135116577,
      "learning_rate": 7.3905e-05,
      "loss": 0.6097,
      "step": 14789
    },
    {
      "epoch": 0.8282002463881734,
      "grad_norm": 1.398011565208435,
      "learning_rate": 7.391e-05,
      "loss": 0.4429,
      "step": 14790
    },
    {
      "epoch": 0.8282562437003024,
      "grad_norm": 1.2961584329605103,
      "learning_rate": 7.3915e-05,
      "loss": 0.4447,
      "step": 14791
    },
    {
      "epoch": 0.8283122410124314,
      "grad_norm": 1.4222519397735596,
      "learning_rate": 7.392e-05,
      "loss": 0.5819,
      "step": 14792
    },
    {
      "epoch": 0.8283682383245604,
      "grad_norm": 1.1378669738769531,
      "learning_rate": 7.3925e-05,
      "loss": 0.3585,
      "step": 14793
    },
    {
      "epoch": 0.8284242356366894,
      "grad_norm": 1.3640563488006592,
      "learning_rate": 7.393e-05,
      "loss": 0.3788,
      "step": 14794
    },
    {
      "epoch": 0.8284802329488185,
      "grad_norm": 1.566734790802002,
      "learning_rate": 7.3935e-05,
      "loss": 0.3884,
      "step": 14795
    },
    {
      "epoch": 0.8285362302609475,
      "grad_norm": 1.1821701526641846,
      "learning_rate": 7.394e-05,
      "loss": 0.4079,
      "step": 14796
    },
    {
      "epoch": 0.8285922275730765,
      "grad_norm": 1.4555513858795166,
      "learning_rate": 7.394500000000001e-05,
      "loss": 0.4253,
      "step": 14797
    },
    {
      "epoch": 0.8286482248852055,
      "grad_norm": 1.151355266571045,
      "learning_rate": 7.395000000000001e-05,
      "loss": 0.5326,
      "step": 14798
    },
    {
      "epoch": 0.8287042221973345,
      "grad_norm": 1.3573161363601685,
      "learning_rate": 7.3955e-05,
      "loss": 0.4826,
      "step": 14799
    },
    {
      "epoch": 0.8287602195094635,
      "grad_norm": 1.3706074953079224,
      "learning_rate": 7.396e-05,
      "loss": 0.4708,
      "step": 14800
    },
    {
      "epoch": 0.8288162168215926,
      "grad_norm": 1.7695225477218628,
      "learning_rate": 7.3965e-05,
      "loss": 0.5706,
      "step": 14801
    },
    {
      "epoch": 0.8288722141337216,
      "grad_norm": 1.3132703304290771,
      "learning_rate": 7.397000000000001e-05,
      "loss": 0.5201,
      "step": 14802
    },
    {
      "epoch": 0.8289282114458506,
      "grad_norm": 1.317012071609497,
      "learning_rate": 7.397500000000001e-05,
      "loss": 0.3876,
      "step": 14803
    },
    {
      "epoch": 0.8289842087579796,
      "grad_norm": 1.2215791940689087,
      "learning_rate": 7.398e-05,
      "loss": 0.4146,
      "step": 14804
    },
    {
      "epoch": 0.8290402060701086,
      "grad_norm": 1.2587159872055054,
      "learning_rate": 7.3985e-05,
      "loss": 0.5045,
      "step": 14805
    },
    {
      "epoch": 0.8290962033822377,
      "grad_norm": 16.493318557739258,
      "learning_rate": 7.399e-05,
      "loss": 0.3607,
      "step": 14806
    },
    {
      "epoch": 0.8291522006943667,
      "grad_norm": 1.1195287704467773,
      "learning_rate": 7.3995e-05,
      "loss": 0.3974,
      "step": 14807
    },
    {
      "epoch": 0.8292081980064957,
      "grad_norm": 1.345656394958496,
      "learning_rate": 7.4e-05,
      "loss": 0.4637,
      "step": 14808
    },
    {
      "epoch": 0.8292641953186247,
      "grad_norm": 1.370466947555542,
      "learning_rate": 7.400500000000001e-05,
      "loss": 0.4364,
      "step": 14809
    },
    {
      "epoch": 0.8293201926307537,
      "grad_norm": 1.2453478574752808,
      "learning_rate": 7.401e-05,
      "loss": 0.3906,
      "step": 14810
    },
    {
      "epoch": 0.8293761899428828,
      "grad_norm": 1.512919306755066,
      "learning_rate": 7.4015e-05,
      "loss": 0.4683,
      "step": 14811
    },
    {
      "epoch": 0.8294321872550118,
      "grad_norm": 1.2530479431152344,
      "learning_rate": 7.402e-05,
      "loss": 0.5021,
      "step": 14812
    },
    {
      "epoch": 0.8294881845671408,
      "grad_norm": 1.4404563903808594,
      "learning_rate": 7.4025e-05,
      "loss": 0.4307,
      "step": 14813
    },
    {
      "epoch": 0.8295441818792698,
      "grad_norm": 1.1528865098953247,
      "learning_rate": 7.403e-05,
      "loss": 0.4487,
      "step": 14814
    },
    {
      "epoch": 0.8296001791913988,
      "grad_norm": 1.3025373220443726,
      "learning_rate": 7.4035e-05,
      "loss": 0.4264,
      "step": 14815
    },
    {
      "epoch": 0.8296561765035279,
      "grad_norm": 1.4294639825820923,
      "learning_rate": 7.404e-05,
      "loss": 0.4792,
      "step": 14816
    },
    {
      "epoch": 0.8297121738156569,
      "grad_norm": 1.3184431791305542,
      "learning_rate": 7.404500000000001e-05,
      "loss": 0.4518,
      "step": 14817
    },
    {
      "epoch": 0.8297681711277859,
      "grad_norm": 1.3912808895111084,
      "learning_rate": 7.405000000000001e-05,
      "loss": 0.4099,
      "step": 14818
    },
    {
      "epoch": 0.8298241684399149,
      "grad_norm": 1.3266398906707764,
      "learning_rate": 7.405500000000001e-05,
      "loss": 0.4376,
      "step": 14819
    },
    {
      "epoch": 0.8298801657520439,
      "grad_norm": 1.2280153036117554,
      "learning_rate": 7.406e-05,
      "loss": 0.4315,
      "step": 14820
    },
    {
      "epoch": 0.829936163064173,
      "grad_norm": 1.4548487663269043,
      "learning_rate": 7.4065e-05,
      "loss": 0.5355,
      "step": 14821
    },
    {
      "epoch": 0.829992160376302,
      "grad_norm": 1.5744315385818481,
      "learning_rate": 7.407e-05,
      "loss": 0.4781,
      "step": 14822
    },
    {
      "epoch": 0.830048157688431,
      "grad_norm": 1.4761401414871216,
      "learning_rate": 7.407500000000001e-05,
      "loss": 0.4804,
      "step": 14823
    },
    {
      "epoch": 0.83010415500056,
      "grad_norm": 1.4220203161239624,
      "learning_rate": 7.408000000000001e-05,
      "loss": 0.4612,
      "step": 14824
    },
    {
      "epoch": 0.830160152312689,
      "grad_norm": 1.3385324478149414,
      "learning_rate": 7.4085e-05,
      "loss": 0.4546,
      "step": 14825
    },
    {
      "epoch": 0.830216149624818,
      "grad_norm": 1.2253103256225586,
      "learning_rate": 7.409e-05,
      "loss": 0.3805,
      "step": 14826
    },
    {
      "epoch": 0.8302721469369471,
      "grad_norm": 1.189700961112976,
      "learning_rate": 7.4095e-05,
      "loss": 0.3957,
      "step": 14827
    },
    {
      "epoch": 0.8303281442490761,
      "grad_norm": 1.3408607244491577,
      "learning_rate": 7.41e-05,
      "loss": 0.4735,
      "step": 14828
    },
    {
      "epoch": 0.8303841415612051,
      "grad_norm": 1.4462240934371948,
      "learning_rate": 7.4105e-05,
      "loss": 0.4593,
      "step": 14829
    },
    {
      "epoch": 0.8304401388733341,
      "grad_norm": 1.2908709049224854,
      "learning_rate": 7.411000000000001e-05,
      "loss": 0.4502,
      "step": 14830
    },
    {
      "epoch": 0.8304961361854631,
      "grad_norm": 1.2498122453689575,
      "learning_rate": 7.4115e-05,
      "loss": 0.3867,
      "step": 14831
    },
    {
      "epoch": 0.8305521334975922,
      "grad_norm": 1.355921983718872,
      "learning_rate": 7.412e-05,
      "loss": 0.4423,
      "step": 14832
    },
    {
      "epoch": 0.8306081308097212,
      "grad_norm": 1.0706628561019897,
      "learning_rate": 7.4125e-05,
      "loss": 0.3468,
      "step": 14833
    },
    {
      "epoch": 0.8306641281218502,
      "grad_norm": 1.3038891553878784,
      "learning_rate": 7.413e-05,
      "loss": 0.3944,
      "step": 14834
    },
    {
      "epoch": 0.8307201254339792,
      "grad_norm": 1.3403329849243164,
      "learning_rate": 7.4135e-05,
      "loss": 0.4666,
      "step": 14835
    },
    {
      "epoch": 0.8307761227461082,
      "grad_norm": 11.348949432373047,
      "learning_rate": 7.414e-05,
      "loss": 0.502,
      "step": 14836
    },
    {
      "epoch": 0.8308321200582373,
      "grad_norm": 2.9683735370635986,
      "learning_rate": 7.4145e-05,
      "loss": 0.3808,
      "step": 14837
    },
    {
      "epoch": 0.8308881173703663,
      "grad_norm": 1.7411690950393677,
      "learning_rate": 7.415000000000001e-05,
      "loss": 0.6224,
      "step": 14838
    },
    {
      "epoch": 0.8309441146824953,
      "grad_norm": 1.753604769706726,
      "learning_rate": 7.415500000000001e-05,
      "loss": 0.5623,
      "step": 14839
    },
    {
      "epoch": 0.8310001119946243,
      "grad_norm": 1.29805588722229,
      "learning_rate": 7.416000000000001e-05,
      "loss": 0.4957,
      "step": 14840
    },
    {
      "epoch": 0.8310561093067532,
      "grad_norm": 1.355239987373352,
      "learning_rate": 7.4165e-05,
      "loss": 0.4066,
      "step": 14841
    },
    {
      "epoch": 0.8311121066188822,
      "grad_norm": 1.1854850053787231,
      "learning_rate": 7.417e-05,
      "loss": 0.4089,
      "step": 14842
    },
    {
      "epoch": 0.8311681039310113,
      "grad_norm": 1.3505926132202148,
      "learning_rate": 7.4175e-05,
      "loss": 0.4427,
      "step": 14843
    },
    {
      "epoch": 0.8312241012431403,
      "grad_norm": 1.7071194648742676,
      "learning_rate": 7.418000000000001e-05,
      "loss": 0.5477,
      "step": 14844
    },
    {
      "epoch": 0.8312800985552693,
      "grad_norm": 1.307734727859497,
      "learning_rate": 7.418500000000001e-05,
      "loss": 0.3538,
      "step": 14845
    },
    {
      "epoch": 0.8313360958673983,
      "grad_norm": 1.3063827753067017,
      "learning_rate": 7.419e-05,
      "loss": 0.4007,
      "step": 14846
    },
    {
      "epoch": 0.8313920931795273,
      "grad_norm": 1.40926194190979,
      "learning_rate": 7.4195e-05,
      "loss": 0.4593,
      "step": 14847
    },
    {
      "epoch": 0.8314480904916564,
      "grad_norm": 1.2591607570648193,
      "learning_rate": 7.42e-05,
      "loss": 0.4995,
      "step": 14848
    },
    {
      "epoch": 0.8315040878037854,
      "grad_norm": 1.5662548542022705,
      "learning_rate": 7.4205e-05,
      "loss": 0.5592,
      "step": 14849
    },
    {
      "epoch": 0.8315600851159144,
      "grad_norm": 1.2944042682647705,
      "learning_rate": 7.421e-05,
      "loss": 0.4518,
      "step": 14850
    },
    {
      "epoch": 0.8316160824280434,
      "grad_norm": 2.1192898750305176,
      "learning_rate": 7.421500000000001e-05,
      "loss": 0.4295,
      "step": 14851
    },
    {
      "epoch": 0.8316720797401724,
      "grad_norm": 1.2583346366882324,
      "learning_rate": 7.422e-05,
      "loss": 0.4781,
      "step": 14852
    },
    {
      "epoch": 0.8317280770523015,
      "grad_norm": 1.570609450340271,
      "learning_rate": 7.4225e-05,
      "loss": 0.6466,
      "step": 14853
    },
    {
      "epoch": 0.8317840743644305,
      "grad_norm": 1.2126221656799316,
      "learning_rate": 7.423e-05,
      "loss": 0.4425,
      "step": 14854
    },
    {
      "epoch": 0.8318400716765595,
      "grad_norm": 1.2975512742996216,
      "learning_rate": 7.4235e-05,
      "loss": 0.3918,
      "step": 14855
    },
    {
      "epoch": 0.8318960689886885,
      "grad_norm": 1.341474175453186,
      "learning_rate": 7.424e-05,
      "loss": 0.4896,
      "step": 14856
    },
    {
      "epoch": 0.8319520663008175,
      "grad_norm": 1.5290112495422363,
      "learning_rate": 7.4245e-05,
      "loss": 0.4209,
      "step": 14857
    },
    {
      "epoch": 0.8320080636129465,
      "grad_norm": 1.3061316013336182,
      "learning_rate": 7.425e-05,
      "loss": 0.4874,
      "step": 14858
    },
    {
      "epoch": 0.8320640609250756,
      "grad_norm": 1.4568842649459839,
      "learning_rate": 7.425500000000001e-05,
      "loss": 0.4203,
      "step": 14859
    },
    {
      "epoch": 0.8321200582372046,
      "grad_norm": 1.3135074377059937,
      "learning_rate": 7.426000000000001e-05,
      "loss": 0.4105,
      "step": 14860
    },
    {
      "epoch": 0.8321760555493336,
      "grad_norm": 1.4716944694519043,
      "learning_rate": 7.426500000000001e-05,
      "loss": 0.5895,
      "step": 14861
    },
    {
      "epoch": 0.8322320528614626,
      "grad_norm": 1.6510661840438843,
      "learning_rate": 7.427e-05,
      "loss": 0.476,
      "step": 14862
    },
    {
      "epoch": 0.8322880501735916,
      "grad_norm": 1.1166248321533203,
      "learning_rate": 7.4275e-05,
      "loss": 0.3829,
      "step": 14863
    },
    {
      "epoch": 0.8323440474857207,
      "grad_norm": 1.3078899383544922,
      "learning_rate": 7.428e-05,
      "loss": 0.419,
      "step": 14864
    },
    {
      "epoch": 0.8324000447978497,
      "grad_norm": 1.235997200012207,
      "learning_rate": 7.428500000000001e-05,
      "loss": 0.4324,
      "step": 14865
    },
    {
      "epoch": 0.8324560421099787,
      "grad_norm": 1.4174054861068726,
      "learning_rate": 7.429000000000001e-05,
      "loss": 0.4429,
      "step": 14866
    },
    {
      "epoch": 0.8325120394221077,
      "grad_norm": 1.2512991428375244,
      "learning_rate": 7.4295e-05,
      "loss": 0.5797,
      "step": 14867
    },
    {
      "epoch": 0.8325680367342367,
      "grad_norm": 1.4484134912490845,
      "learning_rate": 7.43e-05,
      "loss": 0.4633,
      "step": 14868
    },
    {
      "epoch": 0.8326240340463658,
      "grad_norm": 1.1724870204925537,
      "learning_rate": 7.4305e-05,
      "loss": 0.4174,
      "step": 14869
    },
    {
      "epoch": 0.8326800313584948,
      "grad_norm": 1.6425628662109375,
      "learning_rate": 7.431e-05,
      "loss": 0.4574,
      "step": 14870
    },
    {
      "epoch": 0.8327360286706238,
      "grad_norm": 1.5310993194580078,
      "learning_rate": 7.4315e-05,
      "loss": 0.4455,
      "step": 14871
    },
    {
      "epoch": 0.8327920259827528,
      "grad_norm": 1.3307411670684814,
      "learning_rate": 7.432e-05,
      "loss": 0.3753,
      "step": 14872
    },
    {
      "epoch": 0.8328480232948818,
      "grad_norm": 1.6938661336898804,
      "learning_rate": 7.4325e-05,
      "loss": 0.3857,
      "step": 14873
    },
    {
      "epoch": 0.8329040206070109,
      "grad_norm": 1.1644052267074585,
      "learning_rate": 7.433e-05,
      "loss": 0.37,
      "step": 14874
    },
    {
      "epoch": 0.8329600179191399,
      "grad_norm": 1.5414255857467651,
      "learning_rate": 7.4335e-05,
      "loss": 0.7109,
      "step": 14875
    },
    {
      "epoch": 0.8330160152312689,
      "grad_norm": 1.226073145866394,
      "learning_rate": 7.434e-05,
      "loss": 0.4286,
      "step": 14876
    },
    {
      "epoch": 0.8330720125433979,
      "grad_norm": 1.476332187652588,
      "learning_rate": 7.434500000000001e-05,
      "loss": 0.5368,
      "step": 14877
    },
    {
      "epoch": 0.8331280098555269,
      "grad_norm": 1.1078234910964966,
      "learning_rate": 7.435e-05,
      "loss": 0.3977,
      "step": 14878
    },
    {
      "epoch": 0.833184007167656,
      "grad_norm": 1.0875986814498901,
      "learning_rate": 7.4355e-05,
      "loss": 0.4219,
      "step": 14879
    },
    {
      "epoch": 0.833240004479785,
      "grad_norm": 1.0852160453796387,
      "learning_rate": 7.436000000000001e-05,
      "loss": 0.3652,
      "step": 14880
    },
    {
      "epoch": 0.833296001791914,
      "grad_norm": 1.3678065538406372,
      "learning_rate": 7.436500000000001e-05,
      "loss": 0.4264,
      "step": 14881
    },
    {
      "epoch": 0.833351999104043,
      "grad_norm": 1.3367291688919067,
      "learning_rate": 7.437000000000001e-05,
      "loss": 0.4543,
      "step": 14882
    },
    {
      "epoch": 0.833407996416172,
      "grad_norm": 1.0837889909744263,
      "learning_rate": 7.4375e-05,
      "loss": 0.3629,
      "step": 14883
    },
    {
      "epoch": 0.833463993728301,
      "grad_norm": 1.1942131519317627,
      "learning_rate": 7.438e-05,
      "loss": 0.4346,
      "step": 14884
    },
    {
      "epoch": 0.8335199910404301,
      "grad_norm": 1.2060596942901611,
      "learning_rate": 7.4385e-05,
      "loss": 0.4946,
      "step": 14885
    },
    {
      "epoch": 0.8335759883525591,
      "grad_norm": 1.5555088520050049,
      "learning_rate": 7.439e-05,
      "loss": 0.4262,
      "step": 14886
    },
    {
      "epoch": 0.8336319856646881,
      "grad_norm": 1.3655695915222168,
      "learning_rate": 7.439500000000001e-05,
      "loss": 0.4072,
      "step": 14887
    },
    {
      "epoch": 0.8336879829768171,
      "grad_norm": 1.4177491664886475,
      "learning_rate": 7.44e-05,
      "loss": 0.3854,
      "step": 14888
    },
    {
      "epoch": 0.8337439802889461,
      "grad_norm": 1.4612762928009033,
      "learning_rate": 7.4405e-05,
      "loss": 0.5113,
      "step": 14889
    },
    {
      "epoch": 0.8337999776010752,
      "grad_norm": 2.1552562713623047,
      "learning_rate": 7.441e-05,
      "loss": 0.4914,
      "step": 14890
    },
    {
      "epoch": 0.8338559749132042,
      "grad_norm": 1.421797752380371,
      "learning_rate": 7.4415e-05,
      "loss": 0.4435,
      "step": 14891
    },
    {
      "epoch": 0.8339119722253332,
      "grad_norm": 28.51599884033203,
      "learning_rate": 7.442e-05,
      "loss": 0.4078,
      "step": 14892
    },
    {
      "epoch": 0.8339679695374622,
      "grad_norm": 1.7614316940307617,
      "learning_rate": 7.4425e-05,
      "loss": 0.4645,
      "step": 14893
    },
    {
      "epoch": 0.8340239668495912,
      "grad_norm": 1.2875900268554688,
      "learning_rate": 7.443e-05,
      "loss": 0.3867,
      "step": 14894
    },
    {
      "epoch": 0.8340799641617203,
      "grad_norm": 1.457829236984253,
      "learning_rate": 7.4435e-05,
      "loss": 0.5773,
      "step": 14895
    },
    {
      "epoch": 0.8341359614738493,
      "grad_norm": 1.4923008680343628,
      "learning_rate": 7.444e-05,
      "loss": 0.5008,
      "step": 14896
    },
    {
      "epoch": 0.8341919587859783,
      "grad_norm": 1.265728235244751,
      "learning_rate": 7.4445e-05,
      "loss": 0.4255,
      "step": 14897
    },
    {
      "epoch": 0.8342479560981073,
      "grad_norm": 1.3114256858825684,
      "learning_rate": 7.445000000000001e-05,
      "loss": 0.4625,
      "step": 14898
    },
    {
      "epoch": 0.8343039534102363,
      "grad_norm": 1.353049874305725,
      "learning_rate": 7.4455e-05,
      "loss": 0.4348,
      "step": 14899
    },
    {
      "epoch": 0.8343599507223654,
      "grad_norm": 1.4858341217041016,
      "learning_rate": 7.446e-05,
      "loss": 0.5839,
      "step": 14900
    },
    {
      "epoch": 0.8344159480344944,
      "grad_norm": 1.201106071472168,
      "learning_rate": 7.446500000000001e-05,
      "loss": 0.5097,
      "step": 14901
    },
    {
      "epoch": 0.8344719453466234,
      "grad_norm": 1.4960027933120728,
      "learning_rate": 7.447000000000001e-05,
      "loss": 0.3893,
      "step": 14902
    },
    {
      "epoch": 0.8345279426587524,
      "grad_norm": 1.461047887802124,
      "learning_rate": 7.447500000000001e-05,
      "loss": 0.5777,
      "step": 14903
    },
    {
      "epoch": 0.8345839399708814,
      "grad_norm": 1.317950963973999,
      "learning_rate": 7.448e-05,
      "loss": 0.3558,
      "step": 14904
    },
    {
      "epoch": 0.8346399372830104,
      "grad_norm": 1.1693061590194702,
      "learning_rate": 7.4485e-05,
      "loss": 0.3988,
      "step": 14905
    },
    {
      "epoch": 0.8346959345951395,
      "grad_norm": 1.4293357133865356,
      "learning_rate": 7.449e-05,
      "loss": 0.3711,
      "step": 14906
    },
    {
      "epoch": 0.8347519319072685,
      "grad_norm": 1.3089927434921265,
      "learning_rate": 7.4495e-05,
      "loss": 0.4544,
      "step": 14907
    },
    {
      "epoch": 0.8348079292193975,
      "grad_norm": 1.3952184915542603,
      "learning_rate": 7.450000000000001e-05,
      "loss": 0.5343,
      "step": 14908
    },
    {
      "epoch": 0.8348639265315265,
      "grad_norm": 1.7103863954544067,
      "learning_rate": 7.4505e-05,
      "loss": 0.4547,
      "step": 14909
    },
    {
      "epoch": 0.8349199238436555,
      "grad_norm": 1.2795830965042114,
      "learning_rate": 7.451e-05,
      "loss": 0.4001,
      "step": 14910
    },
    {
      "epoch": 0.8349759211557846,
      "grad_norm": 1.4509633779525757,
      "learning_rate": 7.4515e-05,
      "loss": 0.5354,
      "step": 14911
    },
    {
      "epoch": 0.8350319184679136,
      "grad_norm": 1.3074486255645752,
      "learning_rate": 7.452e-05,
      "loss": 0.3619,
      "step": 14912
    },
    {
      "epoch": 0.8350879157800426,
      "grad_norm": 1.2817742824554443,
      "learning_rate": 7.4525e-05,
      "loss": 0.4242,
      "step": 14913
    },
    {
      "epoch": 0.8351439130921716,
      "grad_norm": 1.4056373834609985,
      "learning_rate": 7.453e-05,
      "loss": 0.546,
      "step": 14914
    },
    {
      "epoch": 0.8351999104043006,
      "grad_norm": 1.1839416027069092,
      "learning_rate": 7.4535e-05,
      "loss": 0.4322,
      "step": 14915
    },
    {
      "epoch": 0.8352559077164297,
      "grad_norm": 1.1238529682159424,
      "learning_rate": 7.454e-05,
      "loss": 0.3937,
      "step": 14916
    },
    {
      "epoch": 0.8353119050285587,
      "grad_norm": 1.6225190162658691,
      "learning_rate": 7.4545e-05,
      "loss": 0.5391,
      "step": 14917
    },
    {
      "epoch": 0.8353679023406877,
      "grad_norm": 1.2570643424987793,
      "learning_rate": 7.455000000000001e-05,
      "loss": 0.5262,
      "step": 14918
    },
    {
      "epoch": 0.8354238996528167,
      "grad_norm": 1.3023627996444702,
      "learning_rate": 7.455500000000001e-05,
      "loss": 0.5817,
      "step": 14919
    },
    {
      "epoch": 0.8354798969649457,
      "grad_norm": 1.1538527011871338,
      "learning_rate": 7.456e-05,
      "loss": 0.3635,
      "step": 14920
    },
    {
      "epoch": 0.8355358942770748,
      "grad_norm": 1.3623279333114624,
      "learning_rate": 7.4565e-05,
      "loss": 0.45,
      "step": 14921
    },
    {
      "epoch": 0.8355918915892038,
      "grad_norm": 1.2881258726119995,
      "learning_rate": 7.457000000000001e-05,
      "loss": 0.4108,
      "step": 14922
    },
    {
      "epoch": 0.8356478889013328,
      "grad_norm": 1.0918225049972534,
      "learning_rate": 7.457500000000001e-05,
      "loss": 0.3451,
      "step": 14923
    },
    {
      "epoch": 0.8357038862134617,
      "grad_norm": 1.4319071769714355,
      "learning_rate": 7.458000000000001e-05,
      "loss": 0.5995,
      "step": 14924
    },
    {
      "epoch": 0.8357598835255907,
      "grad_norm": 1.2443747520446777,
      "learning_rate": 7.4585e-05,
      "loss": 0.4229,
      "step": 14925
    },
    {
      "epoch": 0.8358158808377197,
      "grad_norm": 1.3476805686950684,
      "learning_rate": 7.459e-05,
      "loss": 0.3255,
      "step": 14926
    },
    {
      "epoch": 0.8358718781498488,
      "grad_norm": 1.546398401260376,
      "learning_rate": 7.4595e-05,
      "loss": 0.4513,
      "step": 14927
    },
    {
      "epoch": 0.8359278754619778,
      "grad_norm": 1.3001435995101929,
      "learning_rate": 7.46e-05,
      "loss": 0.4581,
      "step": 14928
    },
    {
      "epoch": 0.8359838727741068,
      "grad_norm": 1.1330454349517822,
      "learning_rate": 7.460500000000001e-05,
      "loss": 0.3801,
      "step": 14929
    },
    {
      "epoch": 0.8360398700862358,
      "grad_norm": 1.2715420722961426,
      "learning_rate": 7.461e-05,
      "loss": 0.3954,
      "step": 14930
    },
    {
      "epoch": 0.8360958673983648,
      "grad_norm": 1.3216328620910645,
      "learning_rate": 7.4615e-05,
      "loss": 0.4547,
      "step": 14931
    },
    {
      "epoch": 0.8361518647104939,
      "grad_norm": 1.295122742652893,
      "learning_rate": 7.462e-05,
      "loss": 0.5081,
      "step": 14932
    },
    {
      "epoch": 0.8362078620226229,
      "grad_norm": 1.2410364151000977,
      "learning_rate": 7.4625e-05,
      "loss": 0.3546,
      "step": 14933
    },
    {
      "epoch": 0.8362638593347519,
      "grad_norm": 1.381272554397583,
      "learning_rate": 7.463e-05,
      "loss": 0.4826,
      "step": 14934
    },
    {
      "epoch": 0.8363198566468809,
      "grad_norm": 1.2284221649169922,
      "learning_rate": 7.463499999999999e-05,
      "loss": 0.4441,
      "step": 14935
    },
    {
      "epoch": 0.8363758539590099,
      "grad_norm": 1.4489960670471191,
      "learning_rate": 7.464e-05,
      "loss": 0.4896,
      "step": 14936
    },
    {
      "epoch": 0.836431851271139,
      "grad_norm": 1.1005908250808716,
      "learning_rate": 7.4645e-05,
      "loss": 0.3141,
      "step": 14937
    },
    {
      "epoch": 0.836487848583268,
      "grad_norm": 1.5796070098876953,
      "learning_rate": 7.465000000000001e-05,
      "loss": 0.5244,
      "step": 14938
    },
    {
      "epoch": 0.836543845895397,
      "grad_norm": 1.3035825490951538,
      "learning_rate": 7.465500000000001e-05,
      "loss": 0.3867,
      "step": 14939
    },
    {
      "epoch": 0.836599843207526,
      "grad_norm": 1.3744882345199585,
      "learning_rate": 7.466000000000001e-05,
      "loss": 0.4519,
      "step": 14940
    },
    {
      "epoch": 0.836655840519655,
      "grad_norm": 1.293918490409851,
      "learning_rate": 7.4665e-05,
      "loss": 0.4301,
      "step": 14941
    },
    {
      "epoch": 0.836711837831784,
      "grad_norm": 1.397953987121582,
      "learning_rate": 7.467e-05,
      "loss": 0.3475,
      "step": 14942
    },
    {
      "epoch": 0.8367678351439131,
      "grad_norm": 1.3367972373962402,
      "learning_rate": 7.467500000000001e-05,
      "loss": 0.4344,
      "step": 14943
    },
    {
      "epoch": 0.8368238324560421,
      "grad_norm": 1.3484013080596924,
      "learning_rate": 7.468000000000001e-05,
      "loss": 0.4412,
      "step": 14944
    },
    {
      "epoch": 0.8368798297681711,
      "grad_norm": 1.146432638168335,
      "learning_rate": 7.468500000000001e-05,
      "loss": 0.3475,
      "step": 14945
    },
    {
      "epoch": 0.8369358270803001,
      "grad_norm": 1.3615920543670654,
      "learning_rate": 7.469e-05,
      "loss": 0.4965,
      "step": 14946
    },
    {
      "epoch": 0.8369918243924291,
      "grad_norm": 1.2948979139328003,
      "learning_rate": 7.4695e-05,
      "loss": 0.4753,
      "step": 14947
    },
    {
      "epoch": 0.8370478217045582,
      "grad_norm": 1.672621726989746,
      "learning_rate": 7.47e-05,
      "loss": 0.3665,
      "step": 14948
    },
    {
      "epoch": 0.8371038190166872,
      "grad_norm": 1.4697469472885132,
      "learning_rate": 7.4705e-05,
      "loss": 0.5146,
      "step": 14949
    },
    {
      "epoch": 0.8371598163288162,
      "grad_norm": 1.4369267225265503,
      "learning_rate": 7.471000000000001e-05,
      "loss": 0.4534,
      "step": 14950
    },
    {
      "epoch": 0.8372158136409452,
      "grad_norm": 1.263033390045166,
      "learning_rate": 7.4715e-05,
      "loss": 0.3747,
      "step": 14951
    },
    {
      "epoch": 0.8372718109530742,
      "grad_norm": 1.2970012426376343,
      "learning_rate": 7.472e-05,
      "loss": 0.4123,
      "step": 14952
    },
    {
      "epoch": 0.8373278082652033,
      "grad_norm": 1.3430755138397217,
      "learning_rate": 7.4725e-05,
      "loss": 0.5154,
      "step": 14953
    },
    {
      "epoch": 0.8373838055773323,
      "grad_norm": 1.4384161233901978,
      "learning_rate": 7.473e-05,
      "loss": 0.4757,
      "step": 14954
    },
    {
      "epoch": 0.8374398028894613,
      "grad_norm": 1.372983694076538,
      "learning_rate": 7.4735e-05,
      "loss": 0.5302,
      "step": 14955
    },
    {
      "epoch": 0.8374958002015903,
      "grad_norm": 1.2524195909500122,
      "learning_rate": 7.473999999999999e-05,
      "loss": 0.3936,
      "step": 14956
    },
    {
      "epoch": 0.8375517975137193,
      "grad_norm": 1.3363986015319824,
      "learning_rate": 7.4745e-05,
      "loss": 0.5856,
      "step": 14957
    },
    {
      "epoch": 0.8376077948258484,
      "grad_norm": 1.2847176790237427,
      "learning_rate": 7.475000000000001e-05,
      "loss": 0.4496,
      "step": 14958
    },
    {
      "epoch": 0.8376637921379774,
      "grad_norm": 1.5691107511520386,
      "learning_rate": 7.475500000000001e-05,
      "loss": 0.8147,
      "step": 14959
    },
    {
      "epoch": 0.8377197894501064,
      "grad_norm": 1.3995745182037354,
      "learning_rate": 7.476000000000001e-05,
      "loss": 0.4678,
      "step": 14960
    },
    {
      "epoch": 0.8377757867622354,
      "grad_norm": 1.3471348285675049,
      "learning_rate": 7.4765e-05,
      "loss": 0.4194,
      "step": 14961
    },
    {
      "epoch": 0.8378317840743644,
      "grad_norm": 1.2455815076828003,
      "learning_rate": 7.477e-05,
      "loss": 0.3867,
      "step": 14962
    },
    {
      "epoch": 0.8378877813864934,
      "grad_norm": 1.4678806066513062,
      "learning_rate": 7.4775e-05,
      "loss": 0.4391,
      "step": 14963
    },
    {
      "epoch": 0.8379437786986225,
      "grad_norm": 1.3188345432281494,
      "learning_rate": 7.478e-05,
      "loss": 0.4947,
      "step": 14964
    },
    {
      "epoch": 0.8379997760107515,
      "grad_norm": 1.3664520978927612,
      "learning_rate": 7.478500000000001e-05,
      "loss": 0.4572,
      "step": 14965
    },
    {
      "epoch": 0.8380557733228805,
      "grad_norm": 1.267961025238037,
      "learning_rate": 7.479000000000001e-05,
      "loss": 0.3737,
      "step": 14966
    },
    {
      "epoch": 0.8381117706350095,
      "grad_norm": 1.320089340209961,
      "learning_rate": 7.4795e-05,
      "loss": 0.4796,
      "step": 14967
    },
    {
      "epoch": 0.8381677679471385,
      "grad_norm": 1.3131184577941895,
      "learning_rate": 7.48e-05,
      "loss": 0.4372,
      "step": 14968
    },
    {
      "epoch": 0.8382237652592676,
      "grad_norm": 1.4215002059936523,
      "learning_rate": 7.4805e-05,
      "loss": 0.4411,
      "step": 14969
    },
    {
      "epoch": 0.8382797625713966,
      "grad_norm": 1.3251869678497314,
      "learning_rate": 7.481e-05,
      "loss": 0.4582,
      "step": 14970
    },
    {
      "epoch": 0.8383357598835256,
      "grad_norm": 1.3846755027770996,
      "learning_rate": 7.481500000000001e-05,
      "loss": 0.5226,
      "step": 14971
    },
    {
      "epoch": 0.8383917571956546,
      "grad_norm": 1.3473392724990845,
      "learning_rate": 7.482e-05,
      "loss": 0.5661,
      "step": 14972
    },
    {
      "epoch": 0.8384477545077836,
      "grad_norm": 1.5047401189804077,
      "learning_rate": 7.4825e-05,
      "loss": 0.5873,
      "step": 14973
    },
    {
      "epoch": 0.8385037518199127,
      "grad_norm": 1.4454703330993652,
      "learning_rate": 7.483e-05,
      "loss": 0.6128,
      "step": 14974
    },
    {
      "epoch": 0.8385597491320417,
      "grad_norm": 1.2447439432144165,
      "learning_rate": 7.4835e-05,
      "loss": 0.3712,
      "step": 14975
    },
    {
      "epoch": 0.8386157464441707,
      "grad_norm": 1.2866668701171875,
      "learning_rate": 7.484e-05,
      "loss": 0.3689,
      "step": 14976
    },
    {
      "epoch": 0.8386717437562997,
      "grad_norm": 1.4072551727294922,
      "learning_rate": 7.484499999999999e-05,
      "loss": 0.4984,
      "step": 14977
    },
    {
      "epoch": 0.8387277410684287,
      "grad_norm": 1.4915257692337036,
      "learning_rate": 7.485e-05,
      "loss": 0.3813,
      "step": 14978
    },
    {
      "epoch": 0.8387837383805578,
      "grad_norm": 1.3246511220932007,
      "learning_rate": 7.485500000000001e-05,
      "loss": 0.4371,
      "step": 14979
    },
    {
      "epoch": 0.8388397356926868,
      "grad_norm": 1.0489832162857056,
      "learning_rate": 7.486000000000001e-05,
      "loss": 0.3958,
      "step": 14980
    },
    {
      "epoch": 0.8388957330048158,
      "grad_norm": 1.1089919805526733,
      "learning_rate": 7.486500000000001e-05,
      "loss": 0.3411,
      "step": 14981
    },
    {
      "epoch": 0.8389517303169448,
      "grad_norm": 1.6093395948410034,
      "learning_rate": 7.487e-05,
      "loss": 0.4069,
      "step": 14982
    },
    {
      "epoch": 0.8390077276290738,
      "grad_norm": 1.4661673307418823,
      "learning_rate": 7.4875e-05,
      "loss": 0.5336,
      "step": 14983
    },
    {
      "epoch": 0.8390637249412028,
      "grad_norm": 1.3630739450454712,
      "learning_rate": 7.488e-05,
      "loss": 0.4706,
      "step": 14984
    },
    {
      "epoch": 0.8391197222533319,
      "grad_norm": 1.1160573959350586,
      "learning_rate": 7.4885e-05,
      "loss": 0.3812,
      "step": 14985
    },
    {
      "epoch": 0.8391757195654609,
      "grad_norm": 1.460625410079956,
      "learning_rate": 7.489000000000001e-05,
      "loss": 0.615,
      "step": 14986
    },
    {
      "epoch": 0.8392317168775899,
      "grad_norm": 1.4919368028640747,
      "learning_rate": 7.489500000000001e-05,
      "loss": 0.6785,
      "step": 14987
    },
    {
      "epoch": 0.8392877141897189,
      "grad_norm": 1.2921801805496216,
      "learning_rate": 7.49e-05,
      "loss": 0.3843,
      "step": 14988
    },
    {
      "epoch": 0.8393437115018479,
      "grad_norm": 1.3943148851394653,
      "learning_rate": 7.4905e-05,
      "loss": 0.5812,
      "step": 14989
    },
    {
      "epoch": 0.839399708813977,
      "grad_norm": 1.2905611991882324,
      "learning_rate": 7.491e-05,
      "loss": 0.4547,
      "step": 14990
    },
    {
      "epoch": 0.839455706126106,
      "grad_norm": 1.599746823310852,
      "learning_rate": 7.4915e-05,
      "loss": 0.4198,
      "step": 14991
    },
    {
      "epoch": 0.839511703438235,
      "grad_norm": 1.3307349681854248,
      "learning_rate": 7.492000000000001e-05,
      "loss": 0.4406,
      "step": 14992
    },
    {
      "epoch": 0.839567700750364,
      "grad_norm": 1.2734228372573853,
      "learning_rate": 7.4925e-05,
      "loss": 0.5028,
      "step": 14993
    },
    {
      "epoch": 0.839623698062493,
      "grad_norm": 1.4966728687286377,
      "learning_rate": 7.493e-05,
      "loss": 0.4875,
      "step": 14994
    },
    {
      "epoch": 0.8396796953746221,
      "grad_norm": 1.447335958480835,
      "learning_rate": 7.4935e-05,
      "loss": 0.4251,
      "step": 14995
    },
    {
      "epoch": 0.8397356926867511,
      "grad_norm": 1.4132452011108398,
      "learning_rate": 7.494e-05,
      "loss": 0.5464,
      "step": 14996
    },
    {
      "epoch": 0.8397916899988801,
      "grad_norm": 1.608242154121399,
      "learning_rate": 7.4945e-05,
      "loss": 0.3826,
      "step": 14997
    },
    {
      "epoch": 0.8398476873110091,
      "grad_norm": 1.1616274118423462,
      "learning_rate": 7.495e-05,
      "loss": 0.3667,
      "step": 14998
    },
    {
      "epoch": 0.8399036846231381,
      "grad_norm": 1.2818100452423096,
      "learning_rate": 7.4955e-05,
      "loss": 0.4147,
      "step": 14999
    },
    {
      "epoch": 0.8399596819352672,
      "grad_norm": 1.1615766286849976,
      "learning_rate": 7.496000000000001e-05,
      "loss": 0.3461,
      "step": 15000
    },
    {
      "epoch": 0.8400156792473962,
      "grad_norm": 1.5612907409667969,
      "learning_rate": 7.496500000000001e-05,
      "loss": 0.4007,
      "step": 15001
    },
    {
      "epoch": 0.8400716765595252,
      "grad_norm": 1.4862791299819946,
      "learning_rate": 7.497000000000001e-05,
      "loss": 0.4315,
      "step": 15002
    },
    {
      "epoch": 0.8401276738716542,
      "grad_norm": 1.3636221885681152,
      "learning_rate": 7.4975e-05,
      "loss": 0.5253,
      "step": 15003
    },
    {
      "epoch": 0.8401836711837832,
      "grad_norm": 1.2589805126190186,
      "learning_rate": 7.498e-05,
      "loss": 0.3413,
      "step": 15004
    },
    {
      "epoch": 0.8402396684959123,
      "grad_norm": 1.2923073768615723,
      "learning_rate": 7.4985e-05,
      "loss": 0.3613,
      "step": 15005
    },
    {
      "epoch": 0.8402956658080412,
      "grad_norm": 1.5071004629135132,
      "learning_rate": 7.499e-05,
      "loss": 0.385,
      "step": 15006
    },
    {
      "epoch": 0.8403516631201702,
      "grad_norm": 1.355750560760498,
      "learning_rate": 7.499500000000001e-05,
      "loss": 0.3953,
      "step": 15007
    },
    {
      "epoch": 0.8404076604322992,
      "grad_norm": 1.3409547805786133,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.3501,
      "step": 15008
    },
    {
      "epoch": 0.8404636577444282,
      "grad_norm": 1.5930057764053345,
      "learning_rate": 7.5005e-05,
      "loss": 0.4344,
      "step": 15009
    },
    {
      "epoch": 0.8405196550565572,
      "grad_norm": 1.585240364074707,
      "learning_rate": 7.501e-05,
      "loss": 0.7247,
      "step": 15010
    },
    {
      "epoch": 0.8405756523686863,
      "grad_norm": 1.1724711656570435,
      "learning_rate": 7.5015e-05,
      "loss": 0.4556,
      "step": 15011
    },
    {
      "epoch": 0.8406316496808153,
      "grad_norm": 1.4890203475952148,
      "learning_rate": 7.502e-05,
      "loss": 0.5924,
      "step": 15012
    },
    {
      "epoch": 0.8406876469929443,
      "grad_norm": 1.615662932395935,
      "learning_rate": 7.502500000000001e-05,
      "loss": 0.4121,
      "step": 15013
    },
    {
      "epoch": 0.8407436443050733,
      "grad_norm": 1.2852932214736938,
      "learning_rate": 7.503e-05,
      "loss": 0.393,
      "step": 15014
    },
    {
      "epoch": 0.8407996416172023,
      "grad_norm": 1.6026692390441895,
      "learning_rate": 7.5035e-05,
      "loss": 0.4125,
      "step": 15015
    },
    {
      "epoch": 0.8408556389293314,
      "grad_norm": 1.4187570810317993,
      "learning_rate": 7.504e-05,
      "loss": 0.4448,
      "step": 15016
    },
    {
      "epoch": 0.8409116362414604,
      "grad_norm": 1.3563053607940674,
      "learning_rate": 7.5045e-05,
      "loss": 0.4576,
      "step": 15017
    },
    {
      "epoch": 0.8409676335535894,
      "grad_norm": 1.2016352415084839,
      "learning_rate": 7.505e-05,
      "loss": 0.4305,
      "step": 15018
    },
    {
      "epoch": 0.8410236308657184,
      "grad_norm": 1.245765209197998,
      "learning_rate": 7.5055e-05,
      "loss": 0.3862,
      "step": 15019
    },
    {
      "epoch": 0.8410796281778474,
      "grad_norm": 2.2996387481689453,
      "learning_rate": 7.506e-05,
      "loss": 0.5382,
      "step": 15020
    },
    {
      "epoch": 0.8411356254899764,
      "grad_norm": 1.327736735343933,
      "learning_rate": 7.506500000000001e-05,
      "loss": 0.5347,
      "step": 15021
    },
    {
      "epoch": 0.8411916228021055,
      "grad_norm": 1.5099338293075562,
      "learning_rate": 7.507000000000001e-05,
      "loss": 0.5358,
      "step": 15022
    },
    {
      "epoch": 0.8412476201142345,
      "grad_norm": 1.208007574081421,
      "learning_rate": 7.507500000000001e-05,
      "loss": 0.5175,
      "step": 15023
    },
    {
      "epoch": 0.8413036174263635,
      "grad_norm": 1.517074465751648,
      "learning_rate": 7.508e-05,
      "loss": 0.5982,
      "step": 15024
    },
    {
      "epoch": 0.8413596147384925,
      "grad_norm": 1.3720316886901855,
      "learning_rate": 7.5085e-05,
      "loss": 0.474,
      "step": 15025
    },
    {
      "epoch": 0.8414156120506215,
      "grad_norm": 1.3069417476654053,
      "learning_rate": 7.509e-05,
      "loss": 0.4922,
      "step": 15026
    },
    {
      "epoch": 0.8414716093627506,
      "grad_norm": 1.3628120422363281,
      "learning_rate": 7.5095e-05,
      "loss": 0.4216,
      "step": 15027
    },
    {
      "epoch": 0.8415276066748796,
      "grad_norm": 1.2434688806533813,
      "learning_rate": 7.510000000000001e-05,
      "loss": 0.629,
      "step": 15028
    },
    {
      "epoch": 0.8415836039870086,
      "grad_norm": 1.5129505395889282,
      "learning_rate": 7.510500000000001e-05,
      "loss": 0.422,
      "step": 15029
    },
    {
      "epoch": 0.8416396012991376,
      "grad_norm": 1.1953586339950562,
      "learning_rate": 7.511e-05,
      "loss": 0.4075,
      "step": 15030
    },
    {
      "epoch": 0.8416955986112666,
      "grad_norm": 2.293485403060913,
      "learning_rate": 7.5115e-05,
      "loss": 0.6196,
      "step": 15031
    },
    {
      "epoch": 0.8417515959233957,
      "grad_norm": 1.3731416463851929,
      "learning_rate": 7.512e-05,
      "loss": 0.4824,
      "step": 15032
    },
    {
      "epoch": 0.8418075932355247,
      "grad_norm": 1.2921339273452759,
      "learning_rate": 7.5125e-05,
      "loss": 0.4374,
      "step": 15033
    },
    {
      "epoch": 0.8418635905476537,
      "grad_norm": 1.2573421001434326,
      "learning_rate": 7.513e-05,
      "loss": 0.3117,
      "step": 15034
    },
    {
      "epoch": 0.8419195878597827,
      "grad_norm": 1.2236018180847168,
      "learning_rate": 7.5135e-05,
      "loss": 0.4238,
      "step": 15035
    },
    {
      "epoch": 0.8419755851719117,
      "grad_norm": 1.2526265382766724,
      "learning_rate": 7.514e-05,
      "loss": 0.3968,
      "step": 15036
    },
    {
      "epoch": 0.8420315824840408,
      "grad_norm": 1.4436498880386353,
      "learning_rate": 7.5145e-05,
      "loss": 0.4659,
      "step": 15037
    },
    {
      "epoch": 0.8420875797961698,
      "grad_norm": 1.447879433631897,
      "learning_rate": 7.515e-05,
      "loss": 0.4036,
      "step": 15038
    },
    {
      "epoch": 0.8421435771082988,
      "grad_norm": 1.3573596477508545,
      "learning_rate": 7.515500000000001e-05,
      "loss": 0.3824,
      "step": 15039
    },
    {
      "epoch": 0.8421995744204278,
      "grad_norm": 1.569237470626831,
      "learning_rate": 7.516e-05,
      "loss": 0.395,
      "step": 15040
    },
    {
      "epoch": 0.8422555717325568,
      "grad_norm": 1.4944884777069092,
      "learning_rate": 7.5165e-05,
      "loss": 0.4338,
      "step": 15041
    },
    {
      "epoch": 0.8423115690446858,
      "grad_norm": 1.386260747909546,
      "learning_rate": 7.517000000000001e-05,
      "loss": 0.5996,
      "step": 15042
    },
    {
      "epoch": 0.8423675663568149,
      "grad_norm": 2.475252389907837,
      "learning_rate": 7.517500000000001e-05,
      "loss": 0.5768,
      "step": 15043
    },
    {
      "epoch": 0.8424235636689439,
      "grad_norm": 1.4581243991851807,
      "learning_rate": 7.518000000000001e-05,
      "loss": 0.3723,
      "step": 15044
    },
    {
      "epoch": 0.8424795609810729,
      "grad_norm": 1.2736772298812866,
      "learning_rate": 7.5185e-05,
      "loss": 0.5126,
      "step": 15045
    },
    {
      "epoch": 0.8425355582932019,
      "grad_norm": 1.3051806688308716,
      "learning_rate": 7.519e-05,
      "loss": 0.4082,
      "step": 15046
    },
    {
      "epoch": 0.8425915556053309,
      "grad_norm": 1.5884802341461182,
      "learning_rate": 7.5195e-05,
      "loss": 0.4608,
      "step": 15047
    },
    {
      "epoch": 0.84264755291746,
      "grad_norm": 1.7127121686935425,
      "learning_rate": 7.52e-05,
      "loss": 0.5517,
      "step": 15048
    },
    {
      "epoch": 0.842703550229589,
      "grad_norm": 1.2679857015609741,
      "learning_rate": 7.520500000000001e-05,
      "loss": 0.4228,
      "step": 15049
    },
    {
      "epoch": 0.842759547541718,
      "grad_norm": 1.5982797145843506,
      "learning_rate": 7.521e-05,
      "loss": 0.5136,
      "step": 15050
    },
    {
      "epoch": 0.842815544853847,
      "grad_norm": 1.4622713327407837,
      "learning_rate": 7.5215e-05,
      "loss": 0.4572,
      "step": 15051
    },
    {
      "epoch": 0.842871542165976,
      "grad_norm": 1.3255765438079834,
      "learning_rate": 7.522e-05,
      "loss": 0.4086,
      "step": 15052
    },
    {
      "epoch": 0.8429275394781051,
      "grad_norm": 1.319079041481018,
      "learning_rate": 7.5225e-05,
      "loss": 0.4433,
      "step": 15053
    },
    {
      "epoch": 0.8429835367902341,
      "grad_norm": 1.2004789113998413,
      "learning_rate": 7.523e-05,
      "loss": 0.3553,
      "step": 15054
    },
    {
      "epoch": 0.8430395341023631,
      "grad_norm": 1.1800658702850342,
      "learning_rate": 7.5235e-05,
      "loss": 0.5275,
      "step": 15055
    },
    {
      "epoch": 0.8430955314144921,
      "grad_norm": 1.158860206604004,
      "learning_rate": 7.524e-05,
      "loss": 0.3951,
      "step": 15056
    },
    {
      "epoch": 0.8431515287266211,
      "grad_norm": 1.3727558851242065,
      "learning_rate": 7.5245e-05,
      "loss": 0.5408,
      "step": 15057
    },
    {
      "epoch": 0.8432075260387502,
      "grad_norm": 1.2245267629623413,
      "learning_rate": 7.525e-05,
      "loss": 0.4581,
      "step": 15058
    },
    {
      "epoch": 0.8432635233508792,
      "grad_norm": 1.5283687114715576,
      "learning_rate": 7.525500000000001e-05,
      "loss": 0.6485,
      "step": 15059
    },
    {
      "epoch": 0.8433195206630082,
      "grad_norm": 1.5196453332901,
      "learning_rate": 7.526000000000001e-05,
      "loss": 0.5316,
      "step": 15060
    },
    {
      "epoch": 0.8433755179751372,
      "grad_norm": 1.3572794198989868,
      "learning_rate": 7.5265e-05,
      "loss": 0.4793,
      "step": 15061
    },
    {
      "epoch": 0.8434315152872662,
      "grad_norm": 2.0908303260803223,
      "learning_rate": 7.527e-05,
      "loss": 0.432,
      "step": 15062
    },
    {
      "epoch": 0.8434875125993953,
      "grad_norm": 1.378475546836853,
      "learning_rate": 7.5275e-05,
      "loss": 0.4168,
      "step": 15063
    },
    {
      "epoch": 0.8435435099115243,
      "grad_norm": 1.2672103643417358,
      "learning_rate": 7.528000000000001e-05,
      "loss": 0.4288,
      "step": 15064
    },
    {
      "epoch": 0.8435995072236533,
      "grad_norm": 1.2985295057296753,
      "learning_rate": 7.528500000000001e-05,
      "loss": 0.4394,
      "step": 15065
    },
    {
      "epoch": 0.8436555045357823,
      "grad_norm": 1.3514337539672852,
      "learning_rate": 7.529e-05,
      "loss": 0.5229,
      "step": 15066
    },
    {
      "epoch": 0.8437115018479113,
      "grad_norm": 1.2348365783691406,
      "learning_rate": 7.5295e-05,
      "loss": 0.6591,
      "step": 15067
    },
    {
      "epoch": 0.8437674991600403,
      "grad_norm": 1.2063206434249878,
      "learning_rate": 7.53e-05,
      "loss": 0.4903,
      "step": 15068
    },
    {
      "epoch": 0.8438234964721694,
      "grad_norm": 1.1621588468551636,
      "learning_rate": 7.5305e-05,
      "loss": 0.5143,
      "step": 15069
    },
    {
      "epoch": 0.8438794937842984,
      "grad_norm": 1.2593555450439453,
      "learning_rate": 7.531000000000001e-05,
      "loss": 0.4102,
      "step": 15070
    },
    {
      "epoch": 0.8439354910964274,
      "grad_norm": 1.4822927713394165,
      "learning_rate": 7.5315e-05,
      "loss": 0.5137,
      "step": 15071
    },
    {
      "epoch": 0.8439914884085564,
      "grad_norm": 1.429641604423523,
      "learning_rate": 7.532e-05,
      "loss": 0.5827,
      "step": 15072
    },
    {
      "epoch": 0.8440474857206854,
      "grad_norm": 1.4930412769317627,
      "learning_rate": 7.5325e-05,
      "loss": 0.5212,
      "step": 15073
    },
    {
      "epoch": 0.8441034830328145,
      "grad_norm": 1.2361456155776978,
      "learning_rate": 7.533e-05,
      "loss": 0.4542,
      "step": 15074
    },
    {
      "epoch": 0.8441594803449435,
      "grad_norm": 1.2064471244812012,
      "learning_rate": 7.5335e-05,
      "loss": 0.3553,
      "step": 15075
    },
    {
      "epoch": 0.8442154776570725,
      "grad_norm": 1.3847750425338745,
      "learning_rate": 7.534e-05,
      "loss": 0.5201,
      "step": 15076
    },
    {
      "epoch": 0.8442714749692015,
      "grad_norm": 1.3404045104980469,
      "learning_rate": 7.5345e-05,
      "loss": 0.4421,
      "step": 15077
    },
    {
      "epoch": 0.8443274722813305,
      "grad_norm": 1.3262113332748413,
      "learning_rate": 7.535e-05,
      "loss": 0.5306,
      "step": 15078
    },
    {
      "epoch": 0.8443834695934596,
      "grad_norm": 1.6203056573867798,
      "learning_rate": 7.535500000000001e-05,
      "loss": 0.4103,
      "step": 15079
    },
    {
      "epoch": 0.8444394669055886,
      "grad_norm": 1.3159303665161133,
      "learning_rate": 7.536000000000001e-05,
      "loss": 0.4167,
      "step": 15080
    },
    {
      "epoch": 0.8444954642177176,
      "grad_norm": 1.398086428642273,
      "learning_rate": 7.536500000000001e-05,
      "loss": 0.514,
      "step": 15081
    },
    {
      "epoch": 0.8445514615298466,
      "grad_norm": 1.2203478813171387,
      "learning_rate": 7.537e-05,
      "loss": 0.3646,
      "step": 15082
    },
    {
      "epoch": 0.8446074588419756,
      "grad_norm": 1.266960859298706,
      "learning_rate": 7.5375e-05,
      "loss": 0.4621,
      "step": 15083
    },
    {
      "epoch": 0.8446634561541047,
      "grad_norm": 1.3884551525115967,
      "learning_rate": 7.538e-05,
      "loss": 0.4161,
      "step": 15084
    },
    {
      "epoch": 0.8447194534662337,
      "grad_norm": 1.5007433891296387,
      "learning_rate": 7.538500000000001e-05,
      "loss": 0.5216,
      "step": 15085
    },
    {
      "epoch": 0.8447754507783627,
      "grad_norm": 3.044949531555176,
      "learning_rate": 7.539000000000001e-05,
      "loss": 0.406,
      "step": 15086
    },
    {
      "epoch": 0.8448314480904917,
      "grad_norm": 1.3742420673370361,
      "learning_rate": 7.5395e-05,
      "loss": 0.5471,
      "step": 15087
    },
    {
      "epoch": 0.8448874454026207,
      "grad_norm": 1.5436530113220215,
      "learning_rate": 7.54e-05,
      "loss": 0.516,
      "step": 15088
    },
    {
      "epoch": 0.8449434427147496,
      "grad_norm": 1.3303571939468384,
      "learning_rate": 7.5405e-05,
      "loss": 0.4711,
      "step": 15089
    },
    {
      "epoch": 0.8449994400268787,
      "grad_norm": 1.3449771404266357,
      "learning_rate": 7.541e-05,
      "loss": 0.3937,
      "step": 15090
    },
    {
      "epoch": 0.8450554373390077,
      "grad_norm": 1.3330497741699219,
      "learning_rate": 7.541500000000001e-05,
      "loss": 0.3399,
      "step": 15091
    },
    {
      "epoch": 0.8451114346511367,
      "grad_norm": 1.2649072408676147,
      "learning_rate": 7.542e-05,
      "loss": 0.4364,
      "step": 15092
    },
    {
      "epoch": 0.8451674319632657,
      "grad_norm": 1.2320575714111328,
      "learning_rate": 7.5425e-05,
      "loss": 0.4069,
      "step": 15093
    },
    {
      "epoch": 0.8452234292753947,
      "grad_norm": 1.6531530618667603,
      "learning_rate": 7.543e-05,
      "loss": 0.4329,
      "step": 15094
    },
    {
      "epoch": 0.8452794265875238,
      "grad_norm": 1.2341358661651611,
      "learning_rate": 7.5435e-05,
      "loss": 0.3878,
      "step": 15095
    },
    {
      "epoch": 0.8453354238996528,
      "grad_norm": 2.1931910514831543,
      "learning_rate": 7.544e-05,
      "loss": 0.4141,
      "step": 15096
    },
    {
      "epoch": 0.8453914212117818,
      "grad_norm": 1.1421024799346924,
      "learning_rate": 7.5445e-05,
      "loss": 0.4015,
      "step": 15097
    },
    {
      "epoch": 0.8454474185239108,
      "grad_norm": 1.6613743305206299,
      "learning_rate": 7.545e-05,
      "loss": 0.4879,
      "step": 15098
    },
    {
      "epoch": 0.8455034158360398,
      "grad_norm": 1.3891104459762573,
      "learning_rate": 7.545500000000002e-05,
      "loss": 0.3483,
      "step": 15099
    },
    {
      "epoch": 0.8455594131481688,
      "grad_norm": 1.3898457288742065,
      "learning_rate": 7.546000000000001e-05,
      "loss": 0.6164,
      "step": 15100
    },
    {
      "epoch": 0.8456154104602979,
      "grad_norm": 1.2472468614578247,
      "learning_rate": 7.546500000000001e-05,
      "loss": 0.4381,
      "step": 15101
    },
    {
      "epoch": 0.8456714077724269,
      "grad_norm": 1.5568019151687622,
      "learning_rate": 7.547000000000001e-05,
      "loss": 0.5354,
      "step": 15102
    },
    {
      "epoch": 0.8457274050845559,
      "grad_norm": 1.1311805248260498,
      "learning_rate": 7.5475e-05,
      "loss": 0.4659,
      "step": 15103
    },
    {
      "epoch": 0.8457834023966849,
      "grad_norm": 1.4916313886642456,
      "learning_rate": 7.548e-05,
      "loss": 0.5287,
      "step": 15104
    },
    {
      "epoch": 0.8458393997088139,
      "grad_norm": 1.3321473598480225,
      "learning_rate": 7.5485e-05,
      "loss": 0.4549,
      "step": 15105
    },
    {
      "epoch": 0.845895397020943,
      "grad_norm": 1.2652349472045898,
      "learning_rate": 7.549000000000001e-05,
      "loss": 0.4562,
      "step": 15106
    },
    {
      "epoch": 0.845951394333072,
      "grad_norm": 1.383254885673523,
      "learning_rate": 7.549500000000001e-05,
      "loss": 0.3811,
      "step": 15107
    },
    {
      "epoch": 0.846007391645201,
      "grad_norm": 1.3633276224136353,
      "learning_rate": 7.55e-05,
      "loss": 0.4664,
      "step": 15108
    },
    {
      "epoch": 0.84606338895733,
      "grad_norm": 1.2604163885116577,
      "learning_rate": 7.5505e-05,
      "loss": 0.4612,
      "step": 15109
    },
    {
      "epoch": 0.846119386269459,
      "grad_norm": 1.3349034786224365,
      "learning_rate": 7.551e-05,
      "loss": 0.4329,
      "step": 15110
    },
    {
      "epoch": 0.8461753835815881,
      "grad_norm": 1.3420207500457764,
      "learning_rate": 7.5515e-05,
      "loss": 0.3875,
      "step": 15111
    },
    {
      "epoch": 0.8462313808937171,
      "grad_norm": 1.4629493951797485,
      "learning_rate": 7.552e-05,
      "loss": 0.4527,
      "step": 15112
    },
    {
      "epoch": 0.8462873782058461,
      "grad_norm": 1.209176778793335,
      "learning_rate": 7.5525e-05,
      "loss": 0.4839,
      "step": 15113
    },
    {
      "epoch": 0.8463433755179751,
      "grad_norm": 1.5998519659042358,
      "learning_rate": 7.553e-05,
      "loss": 0.4067,
      "step": 15114
    },
    {
      "epoch": 0.8463993728301041,
      "grad_norm": 1.3559772968292236,
      "learning_rate": 7.5535e-05,
      "loss": 0.3687,
      "step": 15115
    },
    {
      "epoch": 0.8464553701422332,
      "grad_norm": 3.905630350112915,
      "learning_rate": 7.554e-05,
      "loss": 0.3848,
      "step": 15116
    },
    {
      "epoch": 0.8465113674543622,
      "grad_norm": 1.3527567386627197,
      "learning_rate": 7.5545e-05,
      "loss": 0.4556,
      "step": 15117
    },
    {
      "epoch": 0.8465673647664912,
      "grad_norm": 1.3242403268814087,
      "learning_rate": 7.555e-05,
      "loss": 0.4031,
      "step": 15118
    },
    {
      "epoch": 0.8466233620786202,
      "grad_norm": 1.3410563468933105,
      "learning_rate": 7.5555e-05,
      "loss": 0.5529,
      "step": 15119
    },
    {
      "epoch": 0.8466793593907492,
      "grad_norm": 1.413779854774475,
      "learning_rate": 7.556000000000002e-05,
      "loss": 0.444,
      "step": 15120
    },
    {
      "epoch": 0.8467353567028783,
      "grad_norm": 1.4130003452301025,
      "learning_rate": 7.556500000000001e-05,
      "loss": 0.4252,
      "step": 15121
    },
    {
      "epoch": 0.8467913540150073,
      "grad_norm": 1.1051877737045288,
      "learning_rate": 7.557000000000001e-05,
      "loss": 0.3686,
      "step": 15122
    },
    {
      "epoch": 0.8468473513271363,
      "grad_norm": 1.4178258180618286,
      "learning_rate": 7.557500000000001e-05,
      "loss": 0.374,
      "step": 15123
    },
    {
      "epoch": 0.8469033486392653,
      "grad_norm": 1.5272403955459595,
      "learning_rate": 7.558e-05,
      "loss": 0.5934,
      "step": 15124
    },
    {
      "epoch": 0.8469593459513943,
      "grad_norm": 1.3880999088287354,
      "learning_rate": 7.5585e-05,
      "loss": 0.4118,
      "step": 15125
    },
    {
      "epoch": 0.8470153432635233,
      "grad_norm": 1.6361058950424194,
      "learning_rate": 7.559e-05,
      "loss": 0.5314,
      "step": 15126
    },
    {
      "epoch": 0.8470713405756524,
      "grad_norm": 1.244228720664978,
      "learning_rate": 7.559500000000001e-05,
      "loss": 0.4525,
      "step": 15127
    },
    {
      "epoch": 0.8471273378877814,
      "grad_norm": 2.738976001739502,
      "learning_rate": 7.560000000000001e-05,
      "loss": 0.4815,
      "step": 15128
    },
    {
      "epoch": 0.8471833351999104,
      "grad_norm": 1.1920626163482666,
      "learning_rate": 7.5605e-05,
      "loss": 0.4106,
      "step": 15129
    },
    {
      "epoch": 0.8472393325120394,
      "grad_norm": 1.376732587814331,
      "learning_rate": 7.561e-05,
      "loss": 0.4497,
      "step": 15130
    },
    {
      "epoch": 0.8472953298241684,
      "grad_norm": 1.3465228080749512,
      "learning_rate": 7.5615e-05,
      "loss": 0.4248,
      "step": 15131
    },
    {
      "epoch": 0.8473513271362975,
      "grad_norm": 1.2644684314727783,
      "learning_rate": 7.562e-05,
      "loss": 0.5093,
      "step": 15132
    },
    {
      "epoch": 0.8474073244484265,
      "grad_norm": 1.762965440750122,
      "learning_rate": 7.5625e-05,
      "loss": 0.446,
      "step": 15133
    },
    {
      "epoch": 0.8474633217605555,
      "grad_norm": 1.3486056327819824,
      "learning_rate": 7.563e-05,
      "loss": 0.4694,
      "step": 15134
    },
    {
      "epoch": 0.8475193190726845,
      "grad_norm": 1.1375880241394043,
      "learning_rate": 7.5635e-05,
      "loss": 0.3783,
      "step": 15135
    },
    {
      "epoch": 0.8475753163848135,
      "grad_norm": 1.2902443408966064,
      "learning_rate": 7.564e-05,
      "loss": 0.4339,
      "step": 15136
    },
    {
      "epoch": 0.8476313136969426,
      "grad_norm": 1.3332070112228394,
      "learning_rate": 7.5645e-05,
      "loss": 0.4528,
      "step": 15137
    },
    {
      "epoch": 0.8476873110090716,
      "grad_norm": 1.3735326528549194,
      "learning_rate": 7.565e-05,
      "loss": 0.421,
      "step": 15138
    },
    {
      "epoch": 0.8477433083212006,
      "grad_norm": 1.3245290517807007,
      "learning_rate": 7.565499999999999e-05,
      "loss": 0.4636,
      "step": 15139
    },
    {
      "epoch": 0.8477993056333296,
      "grad_norm": 1.122619867324829,
      "learning_rate": 7.566e-05,
      "loss": 0.3773,
      "step": 15140
    },
    {
      "epoch": 0.8478553029454586,
      "grad_norm": 1.3315730094909668,
      "learning_rate": 7.5665e-05,
      "loss": 0.378,
      "step": 15141
    },
    {
      "epoch": 0.8479113002575877,
      "grad_norm": 1.272456169128418,
      "learning_rate": 7.567000000000001e-05,
      "loss": 0.4726,
      "step": 15142
    },
    {
      "epoch": 0.8479672975697167,
      "grad_norm": 1.2605549097061157,
      "learning_rate": 7.567500000000001e-05,
      "loss": 0.4612,
      "step": 15143
    },
    {
      "epoch": 0.8480232948818457,
      "grad_norm": 1.3905287981033325,
      "learning_rate": 7.568000000000001e-05,
      "loss": 0.4521,
      "step": 15144
    },
    {
      "epoch": 0.8480792921939747,
      "grad_norm": 1.5169981718063354,
      "learning_rate": 7.5685e-05,
      "loss": 0.405,
      "step": 15145
    },
    {
      "epoch": 0.8481352895061037,
      "grad_norm": 1.1962016820907593,
      "learning_rate": 7.569e-05,
      "loss": 0.3827,
      "step": 15146
    },
    {
      "epoch": 0.8481912868182327,
      "grad_norm": 1.2600289583206177,
      "learning_rate": 7.5695e-05,
      "loss": 0.3677,
      "step": 15147
    },
    {
      "epoch": 0.8482472841303618,
      "grad_norm": 1.3277665376663208,
      "learning_rate": 7.570000000000001e-05,
      "loss": 0.5631,
      "step": 15148
    },
    {
      "epoch": 0.8483032814424908,
      "grad_norm": 1.6062098741531372,
      "learning_rate": 7.570500000000001e-05,
      "loss": 0.5264,
      "step": 15149
    },
    {
      "epoch": 0.8483592787546198,
      "grad_norm": 1.3468025922775269,
      "learning_rate": 7.571e-05,
      "loss": 0.4625,
      "step": 15150
    },
    {
      "epoch": 0.8484152760667488,
      "grad_norm": 1.3188852071762085,
      "learning_rate": 7.5715e-05,
      "loss": 0.4286,
      "step": 15151
    },
    {
      "epoch": 0.8484712733788778,
      "grad_norm": 1.5755048990249634,
      "learning_rate": 7.572e-05,
      "loss": 0.5279,
      "step": 15152
    },
    {
      "epoch": 0.8485272706910069,
      "grad_norm": 1.7589013576507568,
      "learning_rate": 7.5725e-05,
      "loss": 0.3309,
      "step": 15153
    },
    {
      "epoch": 0.8485832680031359,
      "grad_norm": 1.310774564743042,
      "learning_rate": 7.573e-05,
      "loss": 0.3594,
      "step": 15154
    },
    {
      "epoch": 0.8486392653152649,
      "grad_norm": 1.4295710325241089,
      "learning_rate": 7.5735e-05,
      "loss": 0.4235,
      "step": 15155
    },
    {
      "epoch": 0.8486952626273939,
      "grad_norm": 1.4306899309158325,
      "learning_rate": 7.574e-05,
      "loss": 0.4259,
      "step": 15156
    },
    {
      "epoch": 0.8487512599395229,
      "grad_norm": 1.4165624380111694,
      "learning_rate": 7.5745e-05,
      "loss": 0.5563,
      "step": 15157
    },
    {
      "epoch": 0.848807257251652,
      "grad_norm": 1.251570701599121,
      "learning_rate": 7.575e-05,
      "loss": 0.4512,
      "step": 15158
    },
    {
      "epoch": 0.848863254563781,
      "grad_norm": 1.3429430723190308,
      "learning_rate": 7.5755e-05,
      "loss": 0.4967,
      "step": 15159
    },
    {
      "epoch": 0.84891925187591,
      "grad_norm": 1.4190011024475098,
      "learning_rate": 7.576e-05,
      "loss": 0.4926,
      "step": 15160
    },
    {
      "epoch": 0.848975249188039,
      "grad_norm": 1.4082642793655396,
      "learning_rate": 7.5765e-05,
      "loss": 0.4795,
      "step": 15161
    },
    {
      "epoch": 0.849031246500168,
      "grad_norm": 1.2854467630386353,
      "learning_rate": 7.577e-05,
      "loss": 0.4818,
      "step": 15162
    },
    {
      "epoch": 0.849087243812297,
      "grad_norm": 1.1384862661361694,
      "learning_rate": 7.577500000000001e-05,
      "loss": 0.3906,
      "step": 15163
    },
    {
      "epoch": 0.8491432411244261,
      "grad_norm": 1.175520658493042,
      "learning_rate": 7.578000000000001e-05,
      "loss": 0.3959,
      "step": 15164
    },
    {
      "epoch": 0.8491992384365551,
      "grad_norm": 1.3986434936523438,
      "learning_rate": 7.578500000000001e-05,
      "loss": 0.6178,
      "step": 15165
    },
    {
      "epoch": 0.8492552357486841,
      "grad_norm": 1.4496233463287354,
      "learning_rate": 7.579e-05,
      "loss": 0.4642,
      "step": 15166
    },
    {
      "epoch": 0.8493112330608131,
      "grad_norm": 1.6438530683517456,
      "learning_rate": 7.5795e-05,
      "loss": 0.607,
      "step": 15167
    },
    {
      "epoch": 0.8493672303729422,
      "grad_norm": 1.8109285831451416,
      "learning_rate": 7.58e-05,
      "loss": 0.4636,
      "step": 15168
    },
    {
      "epoch": 0.8494232276850712,
      "grad_norm": 1.416939377784729,
      "learning_rate": 7.580500000000001e-05,
      "loss": 0.3799,
      "step": 15169
    },
    {
      "epoch": 0.8494792249972002,
      "grad_norm": 1.3079711198806763,
      "learning_rate": 7.581000000000001e-05,
      "loss": 0.4225,
      "step": 15170
    },
    {
      "epoch": 0.8495352223093292,
      "grad_norm": 1.6238187551498413,
      "learning_rate": 7.5815e-05,
      "loss": 0.5868,
      "step": 15171
    },
    {
      "epoch": 0.8495912196214581,
      "grad_norm": 1.2737096548080444,
      "learning_rate": 7.582e-05,
      "loss": 0.328,
      "step": 15172
    },
    {
      "epoch": 0.8496472169335871,
      "grad_norm": 1.239679217338562,
      "learning_rate": 7.5825e-05,
      "loss": 0.3602,
      "step": 15173
    },
    {
      "epoch": 0.8497032142457162,
      "grad_norm": 1.4801266193389893,
      "learning_rate": 7.583e-05,
      "loss": 0.3883,
      "step": 15174
    },
    {
      "epoch": 0.8497592115578452,
      "grad_norm": 1.7090065479278564,
      "learning_rate": 7.5835e-05,
      "loss": 0.5019,
      "step": 15175
    },
    {
      "epoch": 0.8498152088699742,
      "grad_norm": 1.245547890663147,
      "learning_rate": 7.584e-05,
      "loss": 0.3724,
      "step": 15176
    },
    {
      "epoch": 0.8498712061821032,
      "grad_norm": 1.361806869506836,
      "learning_rate": 7.5845e-05,
      "loss": 0.5009,
      "step": 15177
    },
    {
      "epoch": 0.8499272034942322,
      "grad_norm": 1.7137019634246826,
      "learning_rate": 7.585e-05,
      "loss": 0.6438,
      "step": 15178
    },
    {
      "epoch": 0.8499832008063612,
      "grad_norm": 1.3690975904464722,
      "learning_rate": 7.5855e-05,
      "loss": 0.4153,
      "step": 15179
    },
    {
      "epoch": 0.8500391981184903,
      "grad_norm": 1.3201725482940674,
      "learning_rate": 7.586000000000001e-05,
      "loss": 0.4907,
      "step": 15180
    },
    {
      "epoch": 0.8500951954306193,
      "grad_norm": 1.3027015924453735,
      "learning_rate": 7.5865e-05,
      "loss": 0.475,
      "step": 15181
    },
    {
      "epoch": 0.8501511927427483,
      "grad_norm": 1.2826457023620605,
      "learning_rate": 7.587e-05,
      "loss": 0.456,
      "step": 15182
    },
    {
      "epoch": 0.8502071900548773,
      "grad_norm": 1.317867398262024,
      "learning_rate": 7.5875e-05,
      "loss": 0.5059,
      "step": 15183
    },
    {
      "epoch": 0.8502631873670063,
      "grad_norm": 2.2673516273498535,
      "learning_rate": 7.588000000000001e-05,
      "loss": 0.6258,
      "step": 15184
    },
    {
      "epoch": 0.8503191846791354,
      "grad_norm": 1.1475238800048828,
      "learning_rate": 7.588500000000001e-05,
      "loss": 0.3576,
      "step": 15185
    },
    {
      "epoch": 0.8503751819912644,
      "grad_norm": 1.4983441829681396,
      "learning_rate": 7.589000000000001e-05,
      "loss": 0.5368,
      "step": 15186
    },
    {
      "epoch": 0.8504311793033934,
      "grad_norm": 1.2127751111984253,
      "learning_rate": 7.5895e-05,
      "loss": 0.4064,
      "step": 15187
    },
    {
      "epoch": 0.8504871766155224,
      "grad_norm": 1.471183180809021,
      "learning_rate": 7.59e-05,
      "loss": 0.4226,
      "step": 15188
    },
    {
      "epoch": 0.8505431739276514,
      "grad_norm": 1.3438286781311035,
      "learning_rate": 7.5905e-05,
      "loss": 0.4563,
      "step": 15189
    },
    {
      "epoch": 0.8505991712397805,
      "grad_norm": 1.1882891654968262,
      "learning_rate": 7.591e-05,
      "loss": 0.521,
      "step": 15190
    },
    {
      "epoch": 0.8506551685519095,
      "grad_norm": 1.0482243299484253,
      "learning_rate": 7.591500000000001e-05,
      "loss": 0.4479,
      "step": 15191
    },
    {
      "epoch": 0.8507111658640385,
      "grad_norm": 1.3335113525390625,
      "learning_rate": 7.592e-05,
      "loss": 0.3779,
      "step": 15192
    },
    {
      "epoch": 0.8507671631761675,
      "grad_norm": 1.2946522235870361,
      "learning_rate": 7.5925e-05,
      "loss": 0.4433,
      "step": 15193
    },
    {
      "epoch": 0.8508231604882965,
      "grad_norm": 1.3130769729614258,
      "learning_rate": 7.593e-05,
      "loss": 0.5303,
      "step": 15194
    },
    {
      "epoch": 0.8508791578004256,
      "grad_norm": 1.0719996690750122,
      "learning_rate": 7.5935e-05,
      "loss": 0.4205,
      "step": 15195
    },
    {
      "epoch": 0.8509351551125546,
      "grad_norm": 1.2909539937973022,
      "learning_rate": 7.594e-05,
      "loss": 0.4152,
      "step": 15196
    },
    {
      "epoch": 0.8509911524246836,
      "grad_norm": 1.605582594871521,
      "learning_rate": 7.5945e-05,
      "loss": 0.5042,
      "step": 15197
    },
    {
      "epoch": 0.8510471497368126,
      "grad_norm": 1.7889682054519653,
      "learning_rate": 7.595e-05,
      "loss": 0.4765,
      "step": 15198
    },
    {
      "epoch": 0.8511031470489416,
      "grad_norm": 1.495755910873413,
      "learning_rate": 7.5955e-05,
      "loss": 0.4569,
      "step": 15199
    },
    {
      "epoch": 0.8511591443610707,
      "grad_norm": 1.4306620359420776,
      "learning_rate": 7.596000000000001e-05,
      "loss": 0.5663,
      "step": 15200
    },
    {
      "epoch": 0.8512151416731997,
      "grad_norm": 1.3635423183441162,
      "learning_rate": 7.596500000000001e-05,
      "loss": 0.4369,
      "step": 15201
    },
    {
      "epoch": 0.8512711389853287,
      "grad_norm": 1.4563435316085815,
      "learning_rate": 7.597e-05,
      "loss": 0.4164,
      "step": 15202
    },
    {
      "epoch": 0.8513271362974577,
      "grad_norm": 1.192082166671753,
      "learning_rate": 7.5975e-05,
      "loss": 0.3142,
      "step": 15203
    },
    {
      "epoch": 0.8513831336095867,
      "grad_norm": 1.3374476432800293,
      "learning_rate": 7.598e-05,
      "loss": 0.3901,
      "step": 15204
    },
    {
      "epoch": 0.8514391309217157,
      "grad_norm": 1.3649131059646606,
      "learning_rate": 7.598500000000001e-05,
      "loss": 0.4348,
      "step": 15205
    },
    {
      "epoch": 0.8514951282338448,
      "grad_norm": 84.07178497314453,
      "learning_rate": 7.599000000000001e-05,
      "loss": 0.5072,
      "step": 15206
    },
    {
      "epoch": 0.8515511255459738,
      "grad_norm": 1.9043371677398682,
      "learning_rate": 7.5995e-05,
      "loss": 0.5572,
      "step": 15207
    },
    {
      "epoch": 0.8516071228581028,
      "grad_norm": 1.4176578521728516,
      "learning_rate": 7.6e-05,
      "loss": 0.5742,
      "step": 15208
    },
    {
      "epoch": 0.8516631201702318,
      "grad_norm": 1.2849376201629639,
      "learning_rate": 7.6005e-05,
      "loss": 0.437,
      "step": 15209
    },
    {
      "epoch": 0.8517191174823608,
      "grad_norm": 1.14579439163208,
      "learning_rate": 7.601e-05,
      "loss": 0.3665,
      "step": 15210
    },
    {
      "epoch": 0.8517751147944899,
      "grad_norm": 1.326216459274292,
      "learning_rate": 7.6015e-05,
      "loss": 0.4595,
      "step": 15211
    },
    {
      "epoch": 0.8518311121066189,
      "grad_norm": 1.5373824834823608,
      "learning_rate": 7.602000000000001e-05,
      "loss": 0.4353,
      "step": 15212
    },
    {
      "epoch": 0.8518871094187479,
      "grad_norm": 1.2989906072616577,
      "learning_rate": 7.6025e-05,
      "loss": 0.3503,
      "step": 15213
    },
    {
      "epoch": 0.8519431067308769,
      "grad_norm": 1.641305685043335,
      "learning_rate": 7.603e-05,
      "loss": 0.4397,
      "step": 15214
    },
    {
      "epoch": 0.8519991040430059,
      "grad_norm": 1.5231834650039673,
      "learning_rate": 7.6035e-05,
      "loss": 0.4273,
      "step": 15215
    },
    {
      "epoch": 0.852055101355135,
      "grad_norm": 6.989379405975342,
      "learning_rate": 7.604e-05,
      "loss": 0.3557,
      "step": 15216
    },
    {
      "epoch": 0.852111098667264,
      "grad_norm": 1.4232429265975952,
      "learning_rate": 7.6045e-05,
      "loss": 0.4825,
      "step": 15217
    },
    {
      "epoch": 0.852167095979393,
      "grad_norm": 8.40113353729248,
      "learning_rate": 7.605e-05,
      "loss": 0.3653,
      "step": 15218
    },
    {
      "epoch": 0.852223093291522,
      "grad_norm": 1.3674389123916626,
      "learning_rate": 7.6055e-05,
      "loss": 0.4705,
      "step": 15219
    },
    {
      "epoch": 0.852279090603651,
      "grad_norm": 1.4717456102371216,
      "learning_rate": 7.606000000000001e-05,
      "loss": 0.483,
      "step": 15220
    },
    {
      "epoch": 0.85233508791578,
      "grad_norm": 1.2045587301254272,
      "learning_rate": 7.606500000000001e-05,
      "loss": 0.3697,
      "step": 15221
    },
    {
      "epoch": 0.8523910852279091,
      "grad_norm": 1.6508337259292603,
      "learning_rate": 7.607000000000001e-05,
      "loss": 0.4677,
      "step": 15222
    },
    {
      "epoch": 0.8524470825400381,
      "grad_norm": 1.5064278841018677,
      "learning_rate": 7.6075e-05,
      "loss": 0.4107,
      "step": 15223
    },
    {
      "epoch": 0.8525030798521671,
      "grad_norm": 1.36526620388031,
      "learning_rate": 7.608e-05,
      "loss": 0.5868,
      "step": 15224
    },
    {
      "epoch": 0.8525590771642961,
      "grad_norm": 1.386756181716919,
      "learning_rate": 7.6085e-05,
      "loss": 0.4697,
      "step": 15225
    },
    {
      "epoch": 0.8526150744764251,
      "grad_norm": 1.6597256660461426,
      "learning_rate": 7.609000000000001e-05,
      "loss": 0.5056,
      "step": 15226
    },
    {
      "epoch": 0.8526710717885542,
      "grad_norm": 1.4845725297927856,
      "learning_rate": 7.609500000000001e-05,
      "loss": 0.4151,
      "step": 15227
    },
    {
      "epoch": 0.8527270691006832,
      "grad_norm": 1.6620419025421143,
      "learning_rate": 7.61e-05,
      "loss": 0.4783,
      "step": 15228
    },
    {
      "epoch": 0.8527830664128122,
      "grad_norm": 1.298387050628662,
      "learning_rate": 7.6105e-05,
      "loss": 0.4438,
      "step": 15229
    },
    {
      "epoch": 0.8528390637249412,
      "grad_norm": 1.2491012811660767,
      "learning_rate": 7.611e-05,
      "loss": 0.3297,
      "step": 15230
    },
    {
      "epoch": 0.8528950610370702,
      "grad_norm": 1.2535948753356934,
      "learning_rate": 7.6115e-05,
      "loss": 0.4548,
      "step": 15231
    },
    {
      "epoch": 0.8529510583491993,
      "grad_norm": 1.3734540939331055,
      "learning_rate": 7.612e-05,
      "loss": 0.3724,
      "step": 15232
    },
    {
      "epoch": 0.8530070556613283,
      "grad_norm": 1.4256091117858887,
      "learning_rate": 7.612500000000001e-05,
      "loss": 0.5826,
      "step": 15233
    },
    {
      "epoch": 0.8530630529734573,
      "grad_norm": 1.5222828388214111,
      "learning_rate": 7.613e-05,
      "loss": 0.5786,
      "step": 15234
    },
    {
      "epoch": 0.8531190502855863,
      "grad_norm": 1.5543458461761475,
      "learning_rate": 7.6135e-05,
      "loss": 0.5745,
      "step": 15235
    },
    {
      "epoch": 0.8531750475977153,
      "grad_norm": 1.5437970161437988,
      "learning_rate": 7.614e-05,
      "loss": 0.4744,
      "step": 15236
    },
    {
      "epoch": 0.8532310449098444,
      "grad_norm": 1.250725507736206,
      "learning_rate": 7.6145e-05,
      "loss": 0.4594,
      "step": 15237
    },
    {
      "epoch": 0.8532870422219734,
      "grad_norm": 1.443010926246643,
      "learning_rate": 7.615e-05,
      "loss": 0.3214,
      "step": 15238
    },
    {
      "epoch": 0.8533430395341024,
      "grad_norm": 1.4554890394210815,
      "learning_rate": 7.6155e-05,
      "loss": 0.5692,
      "step": 15239
    },
    {
      "epoch": 0.8533990368462314,
      "grad_norm": 1.387817621231079,
      "learning_rate": 7.616e-05,
      "loss": 0.3915,
      "step": 15240
    },
    {
      "epoch": 0.8534550341583604,
      "grad_norm": 1.4982044696807861,
      "learning_rate": 7.616500000000001e-05,
      "loss": 0.4794,
      "step": 15241
    },
    {
      "epoch": 0.8535110314704895,
      "grad_norm": 1.5566614866256714,
      "learning_rate": 7.617000000000001e-05,
      "loss": 0.5169,
      "step": 15242
    },
    {
      "epoch": 0.8535670287826185,
      "grad_norm": 1.342090368270874,
      "learning_rate": 7.617500000000001e-05,
      "loss": 0.5644,
      "step": 15243
    },
    {
      "epoch": 0.8536230260947475,
      "grad_norm": 1.3355064392089844,
      "learning_rate": 7.618e-05,
      "loss": 0.4928,
      "step": 15244
    },
    {
      "epoch": 0.8536790234068765,
      "grad_norm": 1.677080512046814,
      "learning_rate": 7.6185e-05,
      "loss": 0.4778,
      "step": 15245
    },
    {
      "epoch": 0.8537350207190055,
      "grad_norm": 1.5180878639221191,
      "learning_rate": 7.619e-05,
      "loss": 0.5091,
      "step": 15246
    },
    {
      "epoch": 0.8537910180311346,
      "grad_norm": 1.383988857269287,
      "learning_rate": 7.619500000000001e-05,
      "loss": 0.4522,
      "step": 15247
    },
    {
      "epoch": 0.8538470153432636,
      "grad_norm": 1.5268155336380005,
      "learning_rate": 7.620000000000001e-05,
      "loss": 0.4806,
      "step": 15248
    },
    {
      "epoch": 0.8539030126553926,
      "grad_norm": 1.1285520792007446,
      "learning_rate": 7.6205e-05,
      "loss": 0.4177,
      "step": 15249
    },
    {
      "epoch": 0.8539590099675216,
      "grad_norm": 1.116123914718628,
      "learning_rate": 7.621e-05,
      "loss": 0.3905,
      "step": 15250
    },
    {
      "epoch": 0.8540150072796506,
      "grad_norm": 1.492278814315796,
      "learning_rate": 7.6215e-05,
      "loss": 0.544,
      "step": 15251
    },
    {
      "epoch": 0.8540710045917796,
      "grad_norm": 1.5980716943740845,
      "learning_rate": 7.622e-05,
      "loss": 0.4494,
      "step": 15252
    },
    {
      "epoch": 0.8541270019039087,
      "grad_norm": 1.2674134969711304,
      "learning_rate": 7.6225e-05,
      "loss": 0.4129,
      "step": 15253
    },
    {
      "epoch": 0.8541829992160376,
      "grad_norm": 1.328264832496643,
      "learning_rate": 7.623000000000001e-05,
      "loss": 0.3812,
      "step": 15254
    },
    {
      "epoch": 0.8542389965281666,
      "grad_norm": 1.4931414127349854,
      "learning_rate": 7.6235e-05,
      "loss": 0.5149,
      "step": 15255
    },
    {
      "epoch": 0.8542949938402956,
      "grad_norm": 1.322326898574829,
      "learning_rate": 7.624e-05,
      "loss": 0.4716,
      "step": 15256
    },
    {
      "epoch": 0.8543509911524246,
      "grad_norm": 1.2993319034576416,
      "learning_rate": 7.6245e-05,
      "loss": 0.3725,
      "step": 15257
    },
    {
      "epoch": 0.8544069884645537,
      "grad_norm": 1.541629433631897,
      "learning_rate": 7.625e-05,
      "loss": 0.5011,
      "step": 15258
    },
    {
      "epoch": 0.8544629857766827,
      "grad_norm": 1.1747040748596191,
      "learning_rate": 7.6255e-05,
      "loss": 0.489,
      "step": 15259
    },
    {
      "epoch": 0.8545189830888117,
      "grad_norm": 1.365580677986145,
      "learning_rate": 7.625999999999999e-05,
      "loss": 0.4291,
      "step": 15260
    },
    {
      "epoch": 0.8545749804009407,
      "grad_norm": 1.0650475025177002,
      "learning_rate": 7.6265e-05,
      "loss": 0.3304,
      "step": 15261
    },
    {
      "epoch": 0.8546309777130697,
      "grad_norm": 1.2407748699188232,
      "learning_rate": 7.627000000000001e-05,
      "loss": 0.394,
      "step": 15262
    },
    {
      "epoch": 0.8546869750251987,
      "grad_norm": 1.5234675407409668,
      "learning_rate": 7.627500000000001e-05,
      "loss": 0.5103,
      "step": 15263
    },
    {
      "epoch": 0.8547429723373278,
      "grad_norm": 1.4129886627197266,
      "learning_rate": 7.628000000000001e-05,
      "loss": 0.5637,
      "step": 15264
    },
    {
      "epoch": 0.8547989696494568,
      "grad_norm": 1.5237212181091309,
      "learning_rate": 7.6285e-05,
      "loss": 0.4244,
      "step": 15265
    },
    {
      "epoch": 0.8548549669615858,
      "grad_norm": 1.416741132736206,
      "learning_rate": 7.629e-05,
      "loss": 0.5953,
      "step": 15266
    },
    {
      "epoch": 0.8549109642737148,
      "grad_norm": 1.2391144037246704,
      "learning_rate": 7.6295e-05,
      "loss": 0.4641,
      "step": 15267
    },
    {
      "epoch": 0.8549669615858438,
      "grad_norm": 1.3451381921768188,
      "learning_rate": 7.630000000000001e-05,
      "loss": 0.4012,
      "step": 15268
    },
    {
      "epoch": 0.8550229588979729,
      "grad_norm": 1.365315318107605,
      "learning_rate": 7.630500000000001e-05,
      "loss": 0.5176,
      "step": 15269
    },
    {
      "epoch": 0.8550789562101019,
      "grad_norm": 1.6232542991638184,
      "learning_rate": 7.631e-05,
      "loss": 0.4692,
      "step": 15270
    },
    {
      "epoch": 0.8551349535222309,
      "grad_norm": 1.7078979015350342,
      "learning_rate": 7.6315e-05,
      "loss": 0.5395,
      "step": 15271
    },
    {
      "epoch": 0.8551909508343599,
      "grad_norm": 1.5448113679885864,
      "learning_rate": 7.632e-05,
      "loss": 0.5784,
      "step": 15272
    },
    {
      "epoch": 0.8552469481464889,
      "grad_norm": 1.3233789205551147,
      "learning_rate": 7.6325e-05,
      "loss": 0.4248,
      "step": 15273
    },
    {
      "epoch": 0.855302945458618,
      "grad_norm": 1.5929086208343506,
      "learning_rate": 7.633e-05,
      "loss": 0.4113,
      "step": 15274
    },
    {
      "epoch": 0.855358942770747,
      "grad_norm": 1.4089661836624146,
      "learning_rate": 7.633500000000001e-05,
      "loss": 0.4067,
      "step": 15275
    },
    {
      "epoch": 0.855414940082876,
      "grad_norm": 1.22316575050354,
      "learning_rate": 7.634e-05,
      "loss": 0.4686,
      "step": 15276
    },
    {
      "epoch": 0.855470937395005,
      "grad_norm": 1.2601726055145264,
      "learning_rate": 7.6345e-05,
      "loss": 0.3796,
      "step": 15277
    },
    {
      "epoch": 0.855526934707134,
      "grad_norm": 1.2563194036483765,
      "learning_rate": 7.635e-05,
      "loss": 0.5263,
      "step": 15278
    },
    {
      "epoch": 0.855582932019263,
      "grad_norm": 1.494057059288025,
      "learning_rate": 7.6355e-05,
      "loss": 0.4191,
      "step": 15279
    },
    {
      "epoch": 0.8556389293313921,
      "grad_norm": 1.328596830368042,
      "learning_rate": 7.636e-05,
      "loss": 0.4457,
      "step": 15280
    },
    {
      "epoch": 0.8556949266435211,
      "grad_norm": 1.3479852676391602,
      "learning_rate": 7.6365e-05,
      "loss": 0.4119,
      "step": 15281
    },
    {
      "epoch": 0.8557509239556501,
      "grad_norm": 1.1821396350860596,
      "learning_rate": 7.637e-05,
      "loss": 0.3704,
      "step": 15282
    },
    {
      "epoch": 0.8558069212677791,
      "grad_norm": 1.317995309829712,
      "learning_rate": 7.637500000000001e-05,
      "loss": 0.6254,
      "step": 15283
    },
    {
      "epoch": 0.8558629185799081,
      "grad_norm": 1.3521196842193604,
      "learning_rate": 7.638000000000001e-05,
      "loss": 0.4053,
      "step": 15284
    },
    {
      "epoch": 0.8559189158920372,
      "grad_norm": 1.6777294874191284,
      "learning_rate": 7.638500000000001e-05,
      "loss": 0.6206,
      "step": 15285
    },
    {
      "epoch": 0.8559749132041662,
      "grad_norm": 1.53383469581604,
      "learning_rate": 7.639e-05,
      "loss": 0.5084,
      "step": 15286
    },
    {
      "epoch": 0.8560309105162952,
      "grad_norm": 1.3479366302490234,
      "learning_rate": 7.6395e-05,
      "loss": 0.5463,
      "step": 15287
    },
    {
      "epoch": 0.8560869078284242,
      "grad_norm": 1.2493224143981934,
      "learning_rate": 7.64e-05,
      "loss": 0.3651,
      "step": 15288
    },
    {
      "epoch": 0.8561429051405532,
      "grad_norm": 1.287906527519226,
      "learning_rate": 7.6405e-05,
      "loss": 0.4258,
      "step": 15289
    },
    {
      "epoch": 0.8561989024526823,
      "grad_norm": 1.672400712966919,
      "learning_rate": 7.641000000000001e-05,
      "loss": 0.5002,
      "step": 15290
    },
    {
      "epoch": 0.8562548997648113,
      "grad_norm": 1.20218026638031,
      "learning_rate": 7.6415e-05,
      "loss": 0.5053,
      "step": 15291
    },
    {
      "epoch": 0.8563108970769403,
      "grad_norm": 1.4538565874099731,
      "learning_rate": 7.642e-05,
      "loss": 0.4315,
      "step": 15292
    },
    {
      "epoch": 0.8563668943890693,
      "grad_norm": 1.3052338361740112,
      "learning_rate": 7.6425e-05,
      "loss": 0.4419,
      "step": 15293
    },
    {
      "epoch": 0.8564228917011983,
      "grad_norm": 1.3309745788574219,
      "learning_rate": 7.643e-05,
      "loss": 0.5276,
      "step": 15294
    },
    {
      "epoch": 0.8564788890133274,
      "grad_norm": 1.4033682346343994,
      "learning_rate": 7.6435e-05,
      "loss": 0.4153,
      "step": 15295
    },
    {
      "epoch": 0.8565348863254564,
      "grad_norm": 1.365225911140442,
      "learning_rate": 7.644e-05,
      "loss": 0.5216,
      "step": 15296
    },
    {
      "epoch": 0.8565908836375854,
      "grad_norm": 1.053643822669983,
      "learning_rate": 7.6445e-05,
      "loss": 0.3019,
      "step": 15297
    },
    {
      "epoch": 0.8566468809497144,
      "grad_norm": 1.4441500902175903,
      "learning_rate": 7.645e-05,
      "loss": 0.5822,
      "step": 15298
    },
    {
      "epoch": 0.8567028782618434,
      "grad_norm": 1.664686679840088,
      "learning_rate": 7.6455e-05,
      "loss": 0.5443,
      "step": 15299
    },
    {
      "epoch": 0.8567588755739725,
      "grad_norm": 1.169580101966858,
      "learning_rate": 7.646e-05,
      "loss": 0.5139,
      "step": 15300
    },
    {
      "epoch": 0.8568148728861015,
      "grad_norm": 1.4870423078536987,
      "learning_rate": 7.646500000000001e-05,
      "loss": 0.4219,
      "step": 15301
    },
    {
      "epoch": 0.8568708701982305,
      "grad_norm": 1.3848826885223389,
      "learning_rate": 7.647e-05,
      "loss": 0.5761,
      "step": 15302
    },
    {
      "epoch": 0.8569268675103595,
      "grad_norm": 1.1729873418807983,
      "learning_rate": 7.6475e-05,
      "loss": 0.3544,
      "step": 15303
    },
    {
      "epoch": 0.8569828648224885,
      "grad_norm": 3.545872926712036,
      "learning_rate": 7.648000000000001e-05,
      "loss": 0.5091,
      "step": 15304
    },
    {
      "epoch": 0.8570388621346176,
      "grad_norm": 1.288378119468689,
      "learning_rate": 7.648500000000001e-05,
      "loss": 0.4387,
      "step": 15305
    },
    {
      "epoch": 0.8570948594467466,
      "grad_norm": 1.4308961629867554,
      "learning_rate": 7.649000000000001e-05,
      "loss": 0.4135,
      "step": 15306
    },
    {
      "epoch": 0.8571508567588756,
      "grad_norm": 1.287602424621582,
      "learning_rate": 7.6495e-05,
      "loss": 0.4679,
      "step": 15307
    },
    {
      "epoch": 0.8572068540710046,
      "grad_norm": 1.4903717041015625,
      "learning_rate": 7.65e-05,
      "loss": 0.5455,
      "step": 15308
    },
    {
      "epoch": 0.8572628513831336,
      "grad_norm": 1.3370749950408936,
      "learning_rate": 7.6505e-05,
      "loss": 0.5237,
      "step": 15309
    },
    {
      "epoch": 0.8573188486952626,
      "grad_norm": 1.2127388715744019,
      "learning_rate": 7.651e-05,
      "loss": 0.4822,
      "step": 15310
    },
    {
      "epoch": 0.8573748460073917,
      "grad_norm": 1.4833921194076538,
      "learning_rate": 7.651500000000001e-05,
      "loss": 0.4984,
      "step": 15311
    },
    {
      "epoch": 0.8574308433195207,
      "grad_norm": 2.3310141563415527,
      "learning_rate": 7.652e-05,
      "loss": 0.7687,
      "step": 15312
    },
    {
      "epoch": 0.8574868406316497,
      "grad_norm": 1.4999405145645142,
      "learning_rate": 7.6525e-05,
      "loss": 0.5486,
      "step": 15313
    },
    {
      "epoch": 0.8575428379437787,
      "grad_norm": 1.5353020429611206,
      "learning_rate": 7.653e-05,
      "loss": 0.515,
      "step": 15314
    },
    {
      "epoch": 0.8575988352559077,
      "grad_norm": 1.1975631713867188,
      "learning_rate": 7.6535e-05,
      "loss": 0.3858,
      "step": 15315
    },
    {
      "epoch": 0.8576548325680368,
      "grad_norm": 1.2844549417495728,
      "learning_rate": 7.654e-05,
      "loss": 0.5304,
      "step": 15316
    },
    {
      "epoch": 0.8577108298801658,
      "grad_norm": 1.4918015003204346,
      "learning_rate": 7.6545e-05,
      "loss": 0.4314,
      "step": 15317
    },
    {
      "epoch": 0.8577668271922948,
      "grad_norm": 1.3247824907302856,
      "learning_rate": 7.655e-05,
      "loss": 0.4832,
      "step": 15318
    },
    {
      "epoch": 0.8578228245044238,
      "grad_norm": 1.0839030742645264,
      "learning_rate": 7.6555e-05,
      "loss": 0.3746,
      "step": 15319
    },
    {
      "epoch": 0.8578788218165528,
      "grad_norm": 1.2216871976852417,
      "learning_rate": 7.656e-05,
      "loss": 0.378,
      "step": 15320
    },
    {
      "epoch": 0.8579348191286819,
      "grad_norm": 1.0671459436416626,
      "learning_rate": 7.656500000000001e-05,
      "loss": 0.3347,
      "step": 15321
    },
    {
      "epoch": 0.8579908164408109,
      "grad_norm": 1.2441599369049072,
      "learning_rate": 7.657000000000001e-05,
      "loss": 0.3581,
      "step": 15322
    },
    {
      "epoch": 0.8580468137529399,
      "grad_norm": 1.1995137929916382,
      "learning_rate": 7.6575e-05,
      "loss": 0.3563,
      "step": 15323
    },
    {
      "epoch": 0.8581028110650689,
      "grad_norm": 1.481691837310791,
      "learning_rate": 7.658e-05,
      "loss": 0.4686,
      "step": 15324
    },
    {
      "epoch": 0.8581588083771979,
      "grad_norm": 1.2402544021606445,
      "learning_rate": 7.658500000000001e-05,
      "loss": 0.4165,
      "step": 15325
    },
    {
      "epoch": 0.858214805689327,
      "grad_norm": 1.3916577100753784,
      "learning_rate": 7.659000000000001e-05,
      "loss": 0.3222,
      "step": 15326
    },
    {
      "epoch": 0.858270803001456,
      "grad_norm": 1.6940088272094727,
      "learning_rate": 7.659500000000001e-05,
      "loss": 0.468,
      "step": 15327
    },
    {
      "epoch": 0.858326800313585,
      "grad_norm": 1.287348747253418,
      "learning_rate": 7.66e-05,
      "loss": 0.3956,
      "step": 15328
    },
    {
      "epoch": 0.858382797625714,
      "grad_norm": 3.0163638591766357,
      "learning_rate": 7.6605e-05,
      "loss": 0.7145,
      "step": 15329
    },
    {
      "epoch": 0.858438794937843,
      "grad_norm": 1.291735053062439,
      "learning_rate": 7.661e-05,
      "loss": 0.4072,
      "step": 15330
    },
    {
      "epoch": 0.858494792249972,
      "grad_norm": 1.3850016593933105,
      "learning_rate": 7.6615e-05,
      "loss": 0.4995,
      "step": 15331
    },
    {
      "epoch": 0.8585507895621011,
      "grad_norm": 1.3612180948257446,
      "learning_rate": 7.662000000000001e-05,
      "loss": 0.3674,
      "step": 15332
    },
    {
      "epoch": 0.8586067868742301,
      "grad_norm": 1.2428797483444214,
      "learning_rate": 7.6625e-05,
      "loss": 0.4511,
      "step": 15333
    },
    {
      "epoch": 0.8586627841863591,
      "grad_norm": 1.3600984811782837,
      "learning_rate": 7.663e-05,
      "loss": 0.4131,
      "step": 15334
    },
    {
      "epoch": 0.8587187814984881,
      "grad_norm": 1.4094133377075195,
      "learning_rate": 7.6635e-05,
      "loss": 0.4748,
      "step": 15335
    },
    {
      "epoch": 0.8587747788106171,
      "grad_norm": 2.0889029502868652,
      "learning_rate": 7.664e-05,
      "loss": 0.3707,
      "step": 15336
    },
    {
      "epoch": 0.858830776122746,
      "grad_norm": 1.264032006263733,
      "learning_rate": 7.6645e-05,
      "loss": 0.3294,
      "step": 15337
    },
    {
      "epoch": 0.8588867734348751,
      "grad_norm": 1.217186689376831,
      "learning_rate": 7.664999999999999e-05,
      "loss": 0.3299,
      "step": 15338
    },
    {
      "epoch": 0.8589427707470041,
      "grad_norm": 2.11641001701355,
      "learning_rate": 7.6655e-05,
      "loss": 0.4554,
      "step": 15339
    },
    {
      "epoch": 0.8589987680591331,
      "grad_norm": 1.3534648418426514,
      "learning_rate": 7.666e-05,
      "loss": 0.4348,
      "step": 15340
    },
    {
      "epoch": 0.8590547653712621,
      "grad_norm": 0.9543183445930481,
      "learning_rate": 7.666500000000001e-05,
      "loss": 0.333,
      "step": 15341
    },
    {
      "epoch": 0.8591107626833911,
      "grad_norm": 1.2656171321868896,
      "learning_rate": 7.667000000000001e-05,
      "loss": 0.4672,
      "step": 15342
    },
    {
      "epoch": 0.8591667599955202,
      "grad_norm": 2.572000741958618,
      "learning_rate": 7.667500000000001e-05,
      "loss": 0.5044,
      "step": 15343
    },
    {
      "epoch": 0.8592227573076492,
      "grad_norm": 1.4788674116134644,
      "learning_rate": 7.668e-05,
      "loss": 0.3862,
      "step": 15344
    },
    {
      "epoch": 0.8592787546197782,
      "grad_norm": 1.557162880897522,
      "learning_rate": 7.6685e-05,
      "loss": 0.3956,
      "step": 15345
    },
    {
      "epoch": 0.8593347519319072,
      "grad_norm": 1.3017799854278564,
      "learning_rate": 7.669000000000001e-05,
      "loss": 0.3904,
      "step": 15346
    },
    {
      "epoch": 0.8593907492440362,
      "grad_norm": 2.0657570362091064,
      "learning_rate": 7.669500000000001e-05,
      "loss": 0.46,
      "step": 15347
    },
    {
      "epoch": 0.8594467465561653,
      "grad_norm": 1.1990329027175903,
      "learning_rate": 7.670000000000001e-05,
      "loss": 0.3548,
      "step": 15348
    },
    {
      "epoch": 0.8595027438682943,
      "grad_norm": 2.280388116836548,
      "learning_rate": 7.6705e-05,
      "loss": 0.4669,
      "step": 15349
    },
    {
      "epoch": 0.8595587411804233,
      "grad_norm": 1.4168832302093506,
      "learning_rate": 7.671e-05,
      "loss": 0.4556,
      "step": 15350
    },
    {
      "epoch": 0.8596147384925523,
      "grad_norm": 1.4790887832641602,
      "learning_rate": 7.6715e-05,
      "loss": 0.3881,
      "step": 15351
    },
    {
      "epoch": 0.8596707358046813,
      "grad_norm": 1.7048856019973755,
      "learning_rate": 7.672e-05,
      "loss": 0.4468,
      "step": 15352
    },
    {
      "epoch": 0.8597267331168104,
      "grad_norm": 1.3507384061813354,
      "learning_rate": 7.672500000000001e-05,
      "loss": 0.4268,
      "step": 15353
    },
    {
      "epoch": 0.8597827304289394,
      "grad_norm": 1.2550270557403564,
      "learning_rate": 7.673e-05,
      "loss": 0.4574,
      "step": 15354
    },
    {
      "epoch": 0.8598387277410684,
      "grad_norm": 1.3710527420043945,
      "learning_rate": 7.6735e-05,
      "loss": 0.422,
      "step": 15355
    },
    {
      "epoch": 0.8598947250531974,
      "grad_norm": 1.1961610317230225,
      "learning_rate": 7.674e-05,
      "loss": 0.3967,
      "step": 15356
    },
    {
      "epoch": 0.8599507223653264,
      "grad_norm": 1.5828375816345215,
      "learning_rate": 7.6745e-05,
      "loss": 0.4511,
      "step": 15357
    },
    {
      "epoch": 0.8600067196774555,
      "grad_norm": 1.3332316875457764,
      "learning_rate": 7.675e-05,
      "loss": 0.4498,
      "step": 15358
    },
    {
      "epoch": 0.8600627169895845,
      "grad_norm": 1.31644868850708,
      "learning_rate": 7.675499999999999e-05,
      "loss": 0.4281,
      "step": 15359
    },
    {
      "epoch": 0.8601187143017135,
      "grad_norm": 1.2825725078582764,
      "learning_rate": 7.676e-05,
      "loss": 0.4345,
      "step": 15360
    },
    {
      "epoch": 0.8601747116138425,
      "grad_norm": 1.291054368019104,
      "learning_rate": 7.676500000000001e-05,
      "loss": 0.3941,
      "step": 15361
    },
    {
      "epoch": 0.8602307089259715,
      "grad_norm": 1.2731778621673584,
      "learning_rate": 7.677000000000001e-05,
      "loss": 0.438,
      "step": 15362
    },
    {
      "epoch": 0.8602867062381006,
      "grad_norm": 1.3587656021118164,
      "learning_rate": 7.677500000000001e-05,
      "loss": 0.414,
      "step": 15363
    },
    {
      "epoch": 0.8603427035502296,
      "grad_norm": 1.2961088418960571,
      "learning_rate": 7.678000000000001e-05,
      "loss": 0.4495,
      "step": 15364
    },
    {
      "epoch": 0.8603987008623586,
      "grad_norm": 1.5349632501602173,
      "learning_rate": 7.6785e-05,
      "loss": 0.6048,
      "step": 15365
    },
    {
      "epoch": 0.8604546981744876,
      "grad_norm": 1.4466733932495117,
      "learning_rate": 7.679e-05,
      "loss": 0.392,
      "step": 15366
    },
    {
      "epoch": 0.8605106954866166,
      "grad_norm": 1.298704981803894,
      "learning_rate": 7.6795e-05,
      "loss": 0.4187,
      "step": 15367
    },
    {
      "epoch": 0.8605666927987456,
      "grad_norm": 1.2818284034729004,
      "learning_rate": 7.680000000000001e-05,
      "loss": 0.4395,
      "step": 15368
    },
    {
      "epoch": 0.8606226901108747,
      "grad_norm": 1.5905777215957642,
      "learning_rate": 7.680500000000001e-05,
      "loss": 0.4923,
      "step": 15369
    },
    {
      "epoch": 0.8606786874230037,
      "grad_norm": 1.232959508895874,
      "learning_rate": 7.681e-05,
      "loss": 0.3201,
      "step": 15370
    },
    {
      "epoch": 0.8607346847351327,
      "grad_norm": 1.4469730854034424,
      "learning_rate": 7.6815e-05,
      "loss": 0.4906,
      "step": 15371
    },
    {
      "epoch": 0.8607906820472617,
      "grad_norm": 1.1165945529937744,
      "learning_rate": 7.682e-05,
      "loss": 0.4475,
      "step": 15372
    },
    {
      "epoch": 0.8608466793593907,
      "grad_norm": 1.3485347032546997,
      "learning_rate": 7.6825e-05,
      "loss": 0.4444,
      "step": 15373
    },
    {
      "epoch": 0.8609026766715198,
      "grad_norm": 1.6563959121704102,
      "learning_rate": 7.683000000000001e-05,
      "loss": 0.4601,
      "step": 15374
    },
    {
      "epoch": 0.8609586739836488,
      "grad_norm": 1.7581827640533447,
      "learning_rate": 7.6835e-05,
      "loss": 0.4572,
      "step": 15375
    },
    {
      "epoch": 0.8610146712957778,
      "grad_norm": 1.1497478485107422,
      "learning_rate": 7.684e-05,
      "loss": 0.3935,
      "step": 15376
    },
    {
      "epoch": 0.8610706686079068,
      "grad_norm": 1.2260955572128296,
      "learning_rate": 7.6845e-05,
      "loss": 0.3778,
      "step": 15377
    },
    {
      "epoch": 0.8611266659200358,
      "grad_norm": 1.366969108581543,
      "learning_rate": 7.685e-05,
      "loss": 0.4371,
      "step": 15378
    },
    {
      "epoch": 0.8611826632321649,
      "grad_norm": 1.4714306592941284,
      "learning_rate": 7.6855e-05,
      "loss": 0.4312,
      "step": 15379
    },
    {
      "epoch": 0.8612386605442939,
      "grad_norm": 1.4076884984970093,
      "learning_rate": 7.685999999999999e-05,
      "loss": 0.466,
      "step": 15380
    },
    {
      "epoch": 0.8612946578564229,
      "grad_norm": 1.2253717184066772,
      "learning_rate": 7.6865e-05,
      "loss": 0.4248,
      "step": 15381
    },
    {
      "epoch": 0.8613506551685519,
      "grad_norm": 1.476537823677063,
      "learning_rate": 7.687000000000001e-05,
      "loss": 0.3281,
      "step": 15382
    },
    {
      "epoch": 0.8614066524806809,
      "grad_norm": 1.4724180698394775,
      "learning_rate": 7.687500000000001e-05,
      "loss": 0.4495,
      "step": 15383
    },
    {
      "epoch": 0.86146264979281,
      "grad_norm": 1.313353180885315,
      "learning_rate": 7.688000000000001e-05,
      "loss": 0.4246,
      "step": 15384
    },
    {
      "epoch": 0.861518647104939,
      "grad_norm": 1.5207279920578003,
      "learning_rate": 7.6885e-05,
      "loss": 0.5077,
      "step": 15385
    },
    {
      "epoch": 0.861574644417068,
      "grad_norm": 1.2960842847824097,
      "learning_rate": 7.689e-05,
      "loss": 0.4312,
      "step": 15386
    },
    {
      "epoch": 0.861630641729197,
      "grad_norm": 1.3373053073883057,
      "learning_rate": 7.6895e-05,
      "loss": 0.5328,
      "step": 15387
    },
    {
      "epoch": 0.861686639041326,
      "grad_norm": 1.7143656015396118,
      "learning_rate": 7.69e-05,
      "loss": 0.5065,
      "step": 15388
    },
    {
      "epoch": 0.861742636353455,
      "grad_norm": 1.6221736669540405,
      "learning_rate": 7.690500000000001e-05,
      "loss": 0.4077,
      "step": 15389
    },
    {
      "epoch": 0.8617986336655841,
      "grad_norm": 1.740646243095398,
      "learning_rate": 7.691000000000001e-05,
      "loss": 0.4296,
      "step": 15390
    },
    {
      "epoch": 0.8618546309777131,
      "grad_norm": 1.6884620189666748,
      "learning_rate": 7.6915e-05,
      "loss": 0.6193,
      "step": 15391
    },
    {
      "epoch": 0.8619106282898421,
      "grad_norm": 1.6343050003051758,
      "learning_rate": 7.692e-05,
      "loss": 0.6356,
      "step": 15392
    },
    {
      "epoch": 0.8619666256019711,
      "grad_norm": 1.3310794830322266,
      "learning_rate": 7.6925e-05,
      "loss": 0.4613,
      "step": 15393
    },
    {
      "epoch": 0.8620226229141001,
      "grad_norm": 1.2125768661499023,
      "learning_rate": 7.693e-05,
      "loss": 0.3971,
      "step": 15394
    },
    {
      "epoch": 0.8620786202262292,
      "grad_norm": 1.165642499923706,
      "learning_rate": 7.693500000000001e-05,
      "loss": 0.4871,
      "step": 15395
    },
    {
      "epoch": 0.8621346175383582,
      "grad_norm": 1.8752145767211914,
      "learning_rate": 7.694e-05,
      "loss": 0.4689,
      "step": 15396
    },
    {
      "epoch": 0.8621906148504872,
      "grad_norm": 1.4456344842910767,
      "learning_rate": 7.6945e-05,
      "loss": 0.4598,
      "step": 15397
    },
    {
      "epoch": 0.8622466121626162,
      "grad_norm": 1.1199686527252197,
      "learning_rate": 7.695e-05,
      "loss": 0.371,
      "step": 15398
    },
    {
      "epoch": 0.8623026094747452,
      "grad_norm": 1.0419329404830933,
      "learning_rate": 7.6955e-05,
      "loss": 0.3077,
      "step": 15399
    },
    {
      "epoch": 0.8623586067868743,
      "grad_norm": 1.233064889907837,
      "learning_rate": 7.696e-05,
      "loss": 0.3843,
      "step": 15400
    },
    {
      "epoch": 0.8624146040990033,
      "grad_norm": 1.3966174125671387,
      "learning_rate": 7.696499999999999e-05,
      "loss": 0.5682,
      "step": 15401
    },
    {
      "epoch": 0.8624706014111323,
      "grad_norm": 2.0899782180786133,
      "learning_rate": 7.697e-05,
      "loss": 0.5866,
      "step": 15402
    },
    {
      "epoch": 0.8625265987232613,
      "grad_norm": 1.5108164548873901,
      "learning_rate": 7.697500000000001e-05,
      "loss": 0.4806,
      "step": 15403
    },
    {
      "epoch": 0.8625825960353903,
      "grad_norm": 1.3947193622589111,
      "learning_rate": 7.698000000000001e-05,
      "loss": 0.4549,
      "step": 15404
    },
    {
      "epoch": 0.8626385933475194,
      "grad_norm": 1.3674523830413818,
      "learning_rate": 7.698500000000001e-05,
      "loss": 0.5378,
      "step": 15405
    },
    {
      "epoch": 0.8626945906596484,
      "grad_norm": 1.3286527395248413,
      "learning_rate": 7.699e-05,
      "loss": 0.604,
      "step": 15406
    },
    {
      "epoch": 0.8627505879717774,
      "grad_norm": 1.4222450256347656,
      "learning_rate": 7.6995e-05,
      "loss": 0.4108,
      "step": 15407
    },
    {
      "epoch": 0.8628065852839064,
      "grad_norm": 1.323704481124878,
      "learning_rate": 7.7e-05,
      "loss": 0.5305,
      "step": 15408
    },
    {
      "epoch": 0.8628625825960354,
      "grad_norm": 1.5488990545272827,
      "learning_rate": 7.7005e-05,
      "loss": 0.6045,
      "step": 15409
    },
    {
      "epoch": 0.8629185799081645,
      "grad_norm": 1.2930275201797485,
      "learning_rate": 7.701000000000001e-05,
      "loss": 0.5057,
      "step": 15410
    },
    {
      "epoch": 0.8629745772202935,
      "grad_norm": 1.2379000186920166,
      "learning_rate": 7.701500000000001e-05,
      "loss": 0.4044,
      "step": 15411
    },
    {
      "epoch": 0.8630305745324225,
      "grad_norm": 1.2456347942352295,
      "learning_rate": 7.702e-05,
      "loss": 0.4925,
      "step": 15412
    },
    {
      "epoch": 0.8630865718445515,
      "grad_norm": 1.2400336265563965,
      "learning_rate": 7.7025e-05,
      "loss": 0.3953,
      "step": 15413
    },
    {
      "epoch": 0.8631425691566805,
      "grad_norm": 1.3767483234405518,
      "learning_rate": 7.703e-05,
      "loss": 0.3056,
      "step": 15414
    },
    {
      "epoch": 0.8631985664688095,
      "grad_norm": 1.6071982383728027,
      "learning_rate": 7.7035e-05,
      "loss": 0.7268,
      "step": 15415
    },
    {
      "epoch": 0.8632545637809386,
      "grad_norm": 1.1186975240707397,
      "learning_rate": 7.704000000000001e-05,
      "loss": 0.3486,
      "step": 15416
    },
    {
      "epoch": 0.8633105610930676,
      "grad_norm": 1.5963722467422485,
      "learning_rate": 7.7045e-05,
      "loss": 0.4169,
      "step": 15417
    },
    {
      "epoch": 0.8633665584051966,
      "grad_norm": 1.4779984951019287,
      "learning_rate": 7.705e-05,
      "loss": 0.5113,
      "step": 15418
    },
    {
      "epoch": 0.8634225557173256,
      "grad_norm": 1.7189549207687378,
      "learning_rate": 7.7055e-05,
      "loss": 0.6054,
      "step": 15419
    },
    {
      "epoch": 0.8634785530294545,
      "grad_norm": 1.1392791271209717,
      "learning_rate": 7.706e-05,
      "loss": 0.3658,
      "step": 15420
    },
    {
      "epoch": 0.8635345503415836,
      "grad_norm": 1.5345208644866943,
      "learning_rate": 7.7065e-05,
      "loss": 0.5942,
      "step": 15421
    },
    {
      "epoch": 0.8635905476537126,
      "grad_norm": 1.671932339668274,
      "learning_rate": 7.707e-05,
      "loss": 0.4105,
      "step": 15422
    },
    {
      "epoch": 0.8636465449658416,
      "grad_norm": 1.54603111743927,
      "learning_rate": 7.7075e-05,
      "loss": 0.4105,
      "step": 15423
    },
    {
      "epoch": 0.8637025422779706,
      "grad_norm": 1.509049892425537,
      "learning_rate": 7.708000000000001e-05,
      "loss": 0.4501,
      "step": 15424
    },
    {
      "epoch": 0.8637585395900996,
      "grad_norm": 1.3530577421188354,
      "learning_rate": 7.708500000000001e-05,
      "loss": 0.5009,
      "step": 15425
    },
    {
      "epoch": 0.8638145369022286,
      "grad_norm": 1.230531930923462,
      "learning_rate": 7.709000000000001e-05,
      "loss": 0.3958,
      "step": 15426
    },
    {
      "epoch": 0.8638705342143577,
      "grad_norm": 1.2434871196746826,
      "learning_rate": 7.7095e-05,
      "loss": 0.5073,
      "step": 15427
    },
    {
      "epoch": 0.8639265315264867,
      "grad_norm": 1.2971305847167969,
      "learning_rate": 7.71e-05,
      "loss": 0.4372,
      "step": 15428
    },
    {
      "epoch": 0.8639825288386157,
      "grad_norm": 1.4570695161819458,
      "learning_rate": 7.7105e-05,
      "loss": 0.4053,
      "step": 15429
    },
    {
      "epoch": 0.8640385261507447,
      "grad_norm": 1.3198705911636353,
      "learning_rate": 7.711e-05,
      "loss": 0.381,
      "step": 15430
    },
    {
      "epoch": 0.8640945234628737,
      "grad_norm": 1.161210536956787,
      "learning_rate": 7.711500000000001e-05,
      "loss": 0.4301,
      "step": 15431
    },
    {
      "epoch": 0.8641505207750028,
      "grad_norm": 1.0020915269851685,
      "learning_rate": 7.712000000000001e-05,
      "loss": 0.3555,
      "step": 15432
    },
    {
      "epoch": 0.8642065180871318,
      "grad_norm": 1.200257420539856,
      "learning_rate": 7.7125e-05,
      "loss": 0.4197,
      "step": 15433
    },
    {
      "epoch": 0.8642625153992608,
      "grad_norm": 1.7010304927825928,
      "learning_rate": 7.713e-05,
      "loss": 0.5349,
      "step": 15434
    },
    {
      "epoch": 0.8643185127113898,
      "grad_norm": 1.8268228769302368,
      "learning_rate": 7.7135e-05,
      "loss": 0.5794,
      "step": 15435
    },
    {
      "epoch": 0.8643745100235188,
      "grad_norm": 1.5086874961853027,
      "learning_rate": 7.714e-05,
      "loss": 0.4878,
      "step": 15436
    },
    {
      "epoch": 0.8644305073356479,
      "grad_norm": 1.0217642784118652,
      "learning_rate": 7.7145e-05,
      "loss": 0.3234,
      "step": 15437
    },
    {
      "epoch": 0.8644865046477769,
      "grad_norm": 5.203007698059082,
      "learning_rate": 7.715e-05,
      "loss": 0.511,
      "step": 15438
    },
    {
      "epoch": 0.8645425019599059,
      "grad_norm": 1.233720302581787,
      "learning_rate": 7.7155e-05,
      "loss": 0.4566,
      "step": 15439
    },
    {
      "epoch": 0.8645984992720349,
      "grad_norm": 1.3341400623321533,
      "learning_rate": 7.716e-05,
      "loss": 0.404,
      "step": 15440
    },
    {
      "epoch": 0.8646544965841639,
      "grad_norm": 1.4108461141586304,
      "learning_rate": 7.7165e-05,
      "loss": 0.4925,
      "step": 15441
    },
    {
      "epoch": 0.864710493896293,
      "grad_norm": 1.3659836053848267,
      "learning_rate": 7.717000000000001e-05,
      "loss": 0.4156,
      "step": 15442
    },
    {
      "epoch": 0.864766491208422,
      "grad_norm": 1.1967355012893677,
      "learning_rate": 7.7175e-05,
      "loss": 0.386,
      "step": 15443
    },
    {
      "epoch": 0.864822488520551,
      "grad_norm": 1.2768745422363281,
      "learning_rate": 7.718e-05,
      "loss": 0.4394,
      "step": 15444
    },
    {
      "epoch": 0.86487848583268,
      "grad_norm": 1.3104383945465088,
      "learning_rate": 7.718500000000001e-05,
      "loss": 0.3974,
      "step": 15445
    },
    {
      "epoch": 0.864934483144809,
      "grad_norm": 1.3468825817108154,
      "learning_rate": 7.719000000000001e-05,
      "loss": 0.4178,
      "step": 15446
    },
    {
      "epoch": 0.864990480456938,
      "grad_norm": 1.2500470876693726,
      "learning_rate": 7.719500000000001e-05,
      "loss": 0.4704,
      "step": 15447
    },
    {
      "epoch": 0.8650464777690671,
      "grad_norm": 1.2548214197158813,
      "learning_rate": 7.72e-05,
      "loss": 0.3746,
      "step": 15448
    },
    {
      "epoch": 0.8651024750811961,
      "grad_norm": 1.4752436876296997,
      "learning_rate": 7.7205e-05,
      "loss": 0.5735,
      "step": 15449
    },
    {
      "epoch": 0.8651584723933251,
      "grad_norm": 1.3975766897201538,
      "learning_rate": 7.721e-05,
      "loss": 0.4493,
      "step": 15450
    },
    {
      "epoch": 0.8652144697054541,
      "grad_norm": 6.8397417068481445,
      "learning_rate": 7.7215e-05,
      "loss": 0.4978,
      "step": 15451
    },
    {
      "epoch": 0.8652704670175831,
      "grad_norm": 1.2908713817596436,
      "learning_rate": 7.722000000000001e-05,
      "loss": 0.4509,
      "step": 15452
    },
    {
      "epoch": 0.8653264643297122,
      "grad_norm": 1.1639848947525024,
      "learning_rate": 7.722500000000001e-05,
      "loss": 0.5475,
      "step": 15453
    },
    {
      "epoch": 0.8653824616418412,
      "grad_norm": 1.2309406995773315,
      "learning_rate": 7.723e-05,
      "loss": 0.4391,
      "step": 15454
    },
    {
      "epoch": 0.8654384589539702,
      "grad_norm": 1.3136157989501953,
      "learning_rate": 7.7235e-05,
      "loss": 0.4677,
      "step": 15455
    },
    {
      "epoch": 0.8654944562660992,
      "grad_norm": 1.334875226020813,
      "learning_rate": 7.724e-05,
      "loss": 0.3689,
      "step": 15456
    },
    {
      "epoch": 0.8655504535782282,
      "grad_norm": 1.2034419775009155,
      "learning_rate": 7.7245e-05,
      "loss": 0.4591,
      "step": 15457
    },
    {
      "epoch": 0.8656064508903573,
      "grad_norm": 1.3061434030532837,
      "learning_rate": 7.725e-05,
      "loss": 0.4413,
      "step": 15458
    },
    {
      "epoch": 0.8656624482024863,
      "grad_norm": 1.155679702758789,
      "learning_rate": 7.7255e-05,
      "loss": 0.386,
      "step": 15459
    },
    {
      "epoch": 0.8657184455146153,
      "grad_norm": 1.4922846555709839,
      "learning_rate": 7.726e-05,
      "loss": 0.5156,
      "step": 15460
    },
    {
      "epoch": 0.8657744428267443,
      "grad_norm": 1.457108497619629,
      "learning_rate": 7.7265e-05,
      "loss": 0.5872,
      "step": 15461
    },
    {
      "epoch": 0.8658304401388733,
      "grad_norm": 1.2564916610717773,
      "learning_rate": 7.727000000000001e-05,
      "loss": 0.4891,
      "step": 15462
    },
    {
      "epoch": 0.8658864374510024,
      "grad_norm": 1.2155927419662476,
      "learning_rate": 7.727500000000001e-05,
      "loss": 0.4745,
      "step": 15463
    },
    {
      "epoch": 0.8659424347631314,
      "grad_norm": 1.302109956741333,
      "learning_rate": 7.728e-05,
      "loss": 0.4322,
      "step": 15464
    },
    {
      "epoch": 0.8659984320752604,
      "grad_norm": 1.4536240100860596,
      "learning_rate": 7.7285e-05,
      "loss": 0.4242,
      "step": 15465
    },
    {
      "epoch": 0.8660544293873894,
      "grad_norm": 1.3679760694503784,
      "learning_rate": 7.729e-05,
      "loss": 0.4916,
      "step": 15466
    },
    {
      "epoch": 0.8661104266995184,
      "grad_norm": 1.3587265014648438,
      "learning_rate": 7.729500000000001e-05,
      "loss": 0.4235,
      "step": 15467
    },
    {
      "epoch": 0.8661664240116475,
      "grad_norm": 1.4559144973754883,
      "learning_rate": 7.730000000000001e-05,
      "loss": 0.4911,
      "step": 15468
    },
    {
      "epoch": 0.8662224213237765,
      "grad_norm": 1.8773778676986694,
      "learning_rate": 7.7305e-05,
      "loss": 0.6553,
      "step": 15469
    },
    {
      "epoch": 0.8662784186359055,
      "grad_norm": 1.2014228105545044,
      "learning_rate": 7.731e-05,
      "loss": 0.3914,
      "step": 15470
    },
    {
      "epoch": 0.8663344159480345,
      "grad_norm": 1.4217487573623657,
      "learning_rate": 7.7315e-05,
      "loss": 0.3885,
      "step": 15471
    },
    {
      "epoch": 0.8663904132601635,
      "grad_norm": 1.4292007684707642,
      "learning_rate": 7.732e-05,
      "loss": 0.4123,
      "step": 15472
    },
    {
      "epoch": 0.8664464105722925,
      "grad_norm": 1.520644187927246,
      "learning_rate": 7.732500000000001e-05,
      "loss": 0.6345,
      "step": 15473
    },
    {
      "epoch": 0.8665024078844216,
      "grad_norm": 1.1639766693115234,
      "learning_rate": 7.733e-05,
      "loss": 0.4798,
      "step": 15474
    },
    {
      "epoch": 0.8665584051965506,
      "grad_norm": 1.337589144706726,
      "learning_rate": 7.7335e-05,
      "loss": 0.3851,
      "step": 15475
    },
    {
      "epoch": 0.8666144025086796,
      "grad_norm": 1.5339282751083374,
      "learning_rate": 7.734e-05,
      "loss": 0.4823,
      "step": 15476
    },
    {
      "epoch": 0.8666703998208086,
      "grad_norm": 1.498666524887085,
      "learning_rate": 7.7345e-05,
      "loss": 0.5346,
      "step": 15477
    },
    {
      "epoch": 0.8667263971329376,
      "grad_norm": 1.5173438787460327,
      "learning_rate": 7.735e-05,
      "loss": 0.4974,
      "step": 15478
    },
    {
      "epoch": 0.8667823944450667,
      "grad_norm": 1.3957185745239258,
      "learning_rate": 7.7355e-05,
      "loss": 0.5111,
      "step": 15479
    },
    {
      "epoch": 0.8668383917571957,
      "grad_norm": 1.2964214086532593,
      "learning_rate": 7.736e-05,
      "loss": 0.4033,
      "step": 15480
    },
    {
      "epoch": 0.8668943890693247,
      "grad_norm": 1.4217289686203003,
      "learning_rate": 7.7365e-05,
      "loss": 0.5553,
      "step": 15481
    },
    {
      "epoch": 0.8669503863814537,
      "grad_norm": 1.6912500858306885,
      "learning_rate": 7.737000000000001e-05,
      "loss": 0.4515,
      "step": 15482
    },
    {
      "epoch": 0.8670063836935827,
      "grad_norm": 1.4011566638946533,
      "learning_rate": 7.737500000000001e-05,
      "loss": 0.467,
      "step": 15483
    },
    {
      "epoch": 0.8670623810057118,
      "grad_norm": 1.3058698177337646,
      "learning_rate": 7.738000000000001e-05,
      "loss": 0.4093,
      "step": 15484
    },
    {
      "epoch": 0.8671183783178408,
      "grad_norm": 1.1876654624938965,
      "learning_rate": 7.7385e-05,
      "loss": 0.5413,
      "step": 15485
    },
    {
      "epoch": 0.8671743756299698,
      "grad_norm": 1.3856619596481323,
      "learning_rate": 7.739e-05,
      "loss": 0.5422,
      "step": 15486
    },
    {
      "epoch": 0.8672303729420988,
      "grad_norm": 1.1427805423736572,
      "learning_rate": 7.7395e-05,
      "loss": 0.3884,
      "step": 15487
    },
    {
      "epoch": 0.8672863702542278,
      "grad_norm": 1.272996425628662,
      "learning_rate": 7.740000000000001e-05,
      "loss": 0.6489,
      "step": 15488
    },
    {
      "epoch": 0.8673423675663569,
      "grad_norm": 1.6561285257339478,
      "learning_rate": 7.740500000000001e-05,
      "loss": 0.5978,
      "step": 15489
    },
    {
      "epoch": 0.8673983648784859,
      "grad_norm": 1.341091513633728,
      "learning_rate": 7.741e-05,
      "loss": 0.4276,
      "step": 15490
    },
    {
      "epoch": 0.8674543621906149,
      "grad_norm": 1.3549751043319702,
      "learning_rate": 7.7415e-05,
      "loss": 0.4541,
      "step": 15491
    },
    {
      "epoch": 0.8675103595027439,
      "grad_norm": 1.370042085647583,
      "learning_rate": 7.742e-05,
      "loss": 0.379,
      "step": 15492
    },
    {
      "epoch": 0.8675663568148729,
      "grad_norm": 1.3413833379745483,
      "learning_rate": 7.7425e-05,
      "loss": 0.4339,
      "step": 15493
    },
    {
      "epoch": 0.867622354127002,
      "grad_norm": 1.3769487142562866,
      "learning_rate": 7.743000000000001e-05,
      "loss": 0.4813,
      "step": 15494
    },
    {
      "epoch": 0.867678351439131,
      "grad_norm": 1.2802037000656128,
      "learning_rate": 7.7435e-05,
      "loss": 0.492,
      "step": 15495
    },
    {
      "epoch": 0.86773434875126,
      "grad_norm": 1.201843500137329,
      "learning_rate": 7.744e-05,
      "loss": 0.3885,
      "step": 15496
    },
    {
      "epoch": 0.867790346063389,
      "grad_norm": 1.21647310256958,
      "learning_rate": 7.7445e-05,
      "loss": 0.3994,
      "step": 15497
    },
    {
      "epoch": 0.867846343375518,
      "grad_norm": 1.3415783643722534,
      "learning_rate": 7.745e-05,
      "loss": 0.3511,
      "step": 15498
    },
    {
      "epoch": 0.867902340687647,
      "grad_norm": 1.1152540445327759,
      "learning_rate": 7.7455e-05,
      "loss": 0.406,
      "step": 15499
    },
    {
      "epoch": 0.8679583379997761,
      "grad_norm": 1.1693726778030396,
      "learning_rate": 7.746e-05,
      "loss": 0.4369,
      "step": 15500
    },
    {
      "epoch": 0.8680143353119051,
      "grad_norm": 1.5938900709152222,
      "learning_rate": 7.7465e-05,
      "loss": 0.5564,
      "step": 15501
    },
    {
      "epoch": 0.868070332624034,
      "grad_norm": 1.3347742557525635,
      "learning_rate": 7.747000000000002e-05,
      "loss": 0.481,
      "step": 15502
    },
    {
      "epoch": 0.868126329936163,
      "grad_norm": 1.2332980632781982,
      "learning_rate": 7.747500000000001e-05,
      "loss": 0.4362,
      "step": 15503
    },
    {
      "epoch": 0.868182327248292,
      "grad_norm": 1.2018414735794067,
      "learning_rate": 7.748000000000001e-05,
      "loss": 0.4191,
      "step": 15504
    },
    {
      "epoch": 0.868238324560421,
      "grad_norm": 1.2105592489242554,
      "learning_rate": 7.748500000000001e-05,
      "loss": 0.517,
      "step": 15505
    },
    {
      "epoch": 0.8682943218725501,
      "grad_norm": 1.053380012512207,
      "learning_rate": 7.749e-05,
      "loss": 0.4104,
      "step": 15506
    },
    {
      "epoch": 0.8683503191846791,
      "grad_norm": 1.3285924196243286,
      "learning_rate": 7.7495e-05,
      "loss": 0.4991,
      "step": 15507
    },
    {
      "epoch": 0.8684063164968081,
      "grad_norm": 1.1642602682113647,
      "learning_rate": 7.75e-05,
      "loss": 0.3911,
      "step": 15508
    },
    {
      "epoch": 0.8684623138089371,
      "grad_norm": 1.2761082649230957,
      "learning_rate": 7.750500000000001e-05,
      "loss": 0.4166,
      "step": 15509
    },
    {
      "epoch": 0.8685183111210661,
      "grad_norm": 1.2622443437576294,
      "learning_rate": 7.751000000000001e-05,
      "loss": 0.4354,
      "step": 15510
    },
    {
      "epoch": 0.8685743084331952,
      "grad_norm": 1.3382610082626343,
      "learning_rate": 7.7515e-05,
      "loss": 0.3583,
      "step": 15511
    },
    {
      "epoch": 0.8686303057453242,
      "grad_norm": 1.1973257064819336,
      "learning_rate": 7.752e-05,
      "loss": 0.3613,
      "step": 15512
    },
    {
      "epoch": 0.8686863030574532,
      "grad_norm": 1.2233986854553223,
      "learning_rate": 7.7525e-05,
      "loss": 0.4168,
      "step": 15513
    },
    {
      "epoch": 0.8687423003695822,
      "grad_norm": 1.197396159172058,
      "learning_rate": 7.753e-05,
      "loss": 0.344,
      "step": 15514
    },
    {
      "epoch": 0.8687982976817112,
      "grad_norm": 1.5905447006225586,
      "learning_rate": 7.7535e-05,
      "loss": 0.6329,
      "step": 15515
    },
    {
      "epoch": 0.8688542949938403,
      "grad_norm": 1.442287802696228,
      "learning_rate": 7.754e-05,
      "loss": 0.3922,
      "step": 15516
    },
    {
      "epoch": 0.8689102923059693,
      "grad_norm": 1.1113076210021973,
      "learning_rate": 7.7545e-05,
      "loss": 0.4077,
      "step": 15517
    },
    {
      "epoch": 0.8689662896180983,
      "grad_norm": 1.2457160949707031,
      "learning_rate": 7.755e-05,
      "loss": 0.471,
      "step": 15518
    },
    {
      "epoch": 0.8690222869302273,
      "grad_norm": 1.5347809791564941,
      "learning_rate": 7.7555e-05,
      "loss": 0.3644,
      "step": 15519
    },
    {
      "epoch": 0.8690782842423563,
      "grad_norm": 1.2472450733184814,
      "learning_rate": 7.756e-05,
      "loss": 0.4156,
      "step": 15520
    },
    {
      "epoch": 0.8691342815544854,
      "grad_norm": 1.208138108253479,
      "learning_rate": 7.7565e-05,
      "loss": 0.468,
      "step": 15521
    },
    {
      "epoch": 0.8691902788666144,
      "grad_norm": 1.2988399267196655,
      "learning_rate": 7.757e-05,
      "loss": 0.4566,
      "step": 15522
    },
    {
      "epoch": 0.8692462761787434,
      "grad_norm": 1.5431101322174072,
      "learning_rate": 7.757500000000002e-05,
      "loss": 0.4289,
      "step": 15523
    },
    {
      "epoch": 0.8693022734908724,
      "grad_norm": 1.3862583637237549,
      "learning_rate": 7.758000000000001e-05,
      "loss": 0.5599,
      "step": 15524
    },
    {
      "epoch": 0.8693582708030014,
      "grad_norm": 1.4205782413482666,
      "learning_rate": 7.758500000000001e-05,
      "loss": 0.4798,
      "step": 15525
    },
    {
      "epoch": 0.8694142681151305,
      "grad_norm": 1.4005579948425293,
      "learning_rate": 7.759000000000001e-05,
      "loss": 0.3991,
      "step": 15526
    },
    {
      "epoch": 0.8694702654272595,
      "grad_norm": 1.517439603805542,
      "learning_rate": 7.7595e-05,
      "loss": 0.4242,
      "step": 15527
    },
    {
      "epoch": 0.8695262627393885,
      "grad_norm": 1.5895737409591675,
      "learning_rate": 7.76e-05,
      "loss": 0.5689,
      "step": 15528
    },
    {
      "epoch": 0.8695822600515175,
      "grad_norm": 1.289066195487976,
      "learning_rate": 7.7605e-05,
      "loss": 0.4296,
      "step": 15529
    },
    {
      "epoch": 0.8696382573636465,
      "grad_norm": 1.2449251413345337,
      "learning_rate": 7.761000000000001e-05,
      "loss": 0.4266,
      "step": 15530
    },
    {
      "epoch": 0.8696942546757755,
      "grad_norm": 1.6113264560699463,
      "learning_rate": 7.761500000000001e-05,
      "loss": 0.5132,
      "step": 15531
    },
    {
      "epoch": 0.8697502519879046,
      "grad_norm": 1.1368285417556763,
      "learning_rate": 7.762e-05,
      "loss": 0.439,
      "step": 15532
    },
    {
      "epoch": 0.8698062493000336,
      "grad_norm": 1.2893623113632202,
      "learning_rate": 7.7625e-05,
      "loss": 0.4703,
      "step": 15533
    },
    {
      "epoch": 0.8698622466121626,
      "grad_norm": 1.5299174785614014,
      "learning_rate": 7.763e-05,
      "loss": 0.4481,
      "step": 15534
    },
    {
      "epoch": 0.8699182439242916,
      "grad_norm": 1.1706514358520508,
      "learning_rate": 7.7635e-05,
      "loss": 0.3218,
      "step": 15535
    },
    {
      "epoch": 0.8699742412364206,
      "grad_norm": 1.2985281944274902,
      "learning_rate": 7.764e-05,
      "loss": 0.3237,
      "step": 15536
    },
    {
      "epoch": 0.8700302385485497,
      "grad_norm": 1.3335989713668823,
      "learning_rate": 7.7645e-05,
      "loss": 0.4881,
      "step": 15537
    },
    {
      "epoch": 0.8700862358606787,
      "grad_norm": 1.3737127780914307,
      "learning_rate": 7.765e-05,
      "loss": 0.3775,
      "step": 15538
    },
    {
      "epoch": 0.8701422331728077,
      "grad_norm": 1.4452928304672241,
      "learning_rate": 7.7655e-05,
      "loss": 0.3665,
      "step": 15539
    },
    {
      "epoch": 0.8701982304849367,
      "grad_norm": 1.1835672855377197,
      "learning_rate": 7.766e-05,
      "loss": 0.3957,
      "step": 15540
    },
    {
      "epoch": 0.8702542277970657,
      "grad_norm": 1.1987941265106201,
      "learning_rate": 7.7665e-05,
      "loss": 0.4849,
      "step": 15541
    },
    {
      "epoch": 0.8703102251091948,
      "grad_norm": 1.0974875688552856,
      "learning_rate": 7.767e-05,
      "loss": 0.4018,
      "step": 15542
    },
    {
      "epoch": 0.8703662224213238,
      "grad_norm": 1.0797059535980225,
      "learning_rate": 7.7675e-05,
      "loss": 0.3083,
      "step": 15543
    },
    {
      "epoch": 0.8704222197334528,
      "grad_norm": 1.196189284324646,
      "learning_rate": 7.768e-05,
      "loss": 0.3826,
      "step": 15544
    },
    {
      "epoch": 0.8704782170455818,
      "grad_norm": 1.3408902883529663,
      "learning_rate": 7.768500000000001e-05,
      "loss": 0.4565,
      "step": 15545
    },
    {
      "epoch": 0.8705342143577108,
      "grad_norm": 1.1600315570831299,
      "learning_rate": 7.769000000000001e-05,
      "loss": 0.3222,
      "step": 15546
    },
    {
      "epoch": 0.8705902116698399,
      "grad_norm": 1.8778573274612427,
      "learning_rate": 7.769500000000001e-05,
      "loss": 0.5961,
      "step": 15547
    },
    {
      "epoch": 0.8706462089819689,
      "grad_norm": 1.2116731405258179,
      "learning_rate": 7.77e-05,
      "loss": 0.4234,
      "step": 15548
    },
    {
      "epoch": 0.8707022062940979,
      "grad_norm": 1.3063853979110718,
      "learning_rate": 7.7705e-05,
      "loss": 0.38,
      "step": 15549
    },
    {
      "epoch": 0.8707582036062269,
      "grad_norm": 1.646710753440857,
      "learning_rate": 7.771e-05,
      "loss": 0.433,
      "step": 15550
    },
    {
      "epoch": 0.8708142009183559,
      "grad_norm": 1.3051859140396118,
      "learning_rate": 7.771500000000001e-05,
      "loss": 0.4738,
      "step": 15551
    },
    {
      "epoch": 0.870870198230485,
      "grad_norm": 1.4985663890838623,
      "learning_rate": 7.772000000000001e-05,
      "loss": 0.4587,
      "step": 15552
    },
    {
      "epoch": 0.870926195542614,
      "grad_norm": 1.1227905750274658,
      "learning_rate": 7.7725e-05,
      "loss": 0.3363,
      "step": 15553
    },
    {
      "epoch": 0.870982192854743,
      "grad_norm": 1.4132035970687866,
      "learning_rate": 7.773e-05,
      "loss": 0.5337,
      "step": 15554
    },
    {
      "epoch": 0.871038190166872,
      "grad_norm": 1.3326305150985718,
      "learning_rate": 7.7735e-05,
      "loss": 0.4576,
      "step": 15555
    },
    {
      "epoch": 0.871094187479001,
      "grad_norm": 1.2112631797790527,
      "learning_rate": 7.774e-05,
      "loss": 0.405,
      "step": 15556
    },
    {
      "epoch": 0.87115018479113,
      "grad_norm": 1.383243203163147,
      "learning_rate": 7.7745e-05,
      "loss": 0.6684,
      "step": 15557
    },
    {
      "epoch": 0.8712061821032591,
      "grad_norm": 1.0421710014343262,
      "learning_rate": 7.775e-05,
      "loss": 0.3893,
      "step": 15558
    },
    {
      "epoch": 0.8712621794153881,
      "grad_norm": 1.2611445188522339,
      "learning_rate": 7.7755e-05,
      "loss": 0.5054,
      "step": 15559
    },
    {
      "epoch": 0.8713181767275171,
      "grad_norm": 1.14493727684021,
      "learning_rate": 7.776e-05,
      "loss": 0.352,
      "step": 15560
    },
    {
      "epoch": 0.8713741740396461,
      "grad_norm": 1.6791402101516724,
      "learning_rate": 7.7765e-05,
      "loss": 0.5136,
      "step": 15561
    },
    {
      "epoch": 0.8714301713517751,
      "grad_norm": 1.4859578609466553,
      "learning_rate": 7.777e-05,
      "loss": 0.3403,
      "step": 15562
    },
    {
      "epoch": 0.8714861686639042,
      "grad_norm": 1.1438791751861572,
      "learning_rate": 7.7775e-05,
      "loss": 0.3434,
      "step": 15563
    },
    {
      "epoch": 0.8715421659760332,
      "grad_norm": 1.1297763586044312,
      "learning_rate": 7.778e-05,
      "loss": 0.4069,
      "step": 15564
    },
    {
      "epoch": 0.8715981632881622,
      "grad_norm": 1.4422338008880615,
      "learning_rate": 7.7785e-05,
      "loss": 0.4356,
      "step": 15565
    },
    {
      "epoch": 0.8716541606002912,
      "grad_norm": 1.3220010995864868,
      "learning_rate": 7.779000000000001e-05,
      "loss": 0.4501,
      "step": 15566
    },
    {
      "epoch": 0.8717101579124202,
      "grad_norm": 1.1621001958847046,
      "learning_rate": 7.779500000000001e-05,
      "loss": 0.4435,
      "step": 15567
    },
    {
      "epoch": 0.8717661552245493,
      "grad_norm": 1.2144877910614014,
      "learning_rate": 7.780000000000001e-05,
      "loss": 0.4044,
      "step": 15568
    },
    {
      "epoch": 0.8718221525366783,
      "grad_norm": 1.7398021221160889,
      "learning_rate": 7.7805e-05,
      "loss": 0.7119,
      "step": 15569
    },
    {
      "epoch": 0.8718781498488073,
      "grad_norm": 1.3021275997161865,
      "learning_rate": 7.781e-05,
      "loss": 0.5303,
      "step": 15570
    },
    {
      "epoch": 0.8719341471609363,
      "grad_norm": 1.234926700592041,
      "learning_rate": 7.7815e-05,
      "loss": 0.4424,
      "step": 15571
    },
    {
      "epoch": 0.8719901444730653,
      "grad_norm": 1.5147863626480103,
      "learning_rate": 7.782000000000001e-05,
      "loss": 0.4176,
      "step": 15572
    },
    {
      "epoch": 0.8720461417851944,
      "grad_norm": 1.1744723320007324,
      "learning_rate": 7.782500000000001e-05,
      "loss": 0.3972,
      "step": 15573
    },
    {
      "epoch": 0.8721021390973234,
      "grad_norm": 1.2651808261871338,
      "learning_rate": 7.783e-05,
      "loss": 0.3941,
      "step": 15574
    },
    {
      "epoch": 0.8721581364094524,
      "grad_norm": 1.311531901359558,
      "learning_rate": 7.7835e-05,
      "loss": 0.4906,
      "step": 15575
    },
    {
      "epoch": 0.8722141337215814,
      "grad_norm": 1.4220561981201172,
      "learning_rate": 7.784e-05,
      "loss": 0.4883,
      "step": 15576
    },
    {
      "epoch": 0.8722701310337104,
      "grad_norm": 1.3571114540100098,
      "learning_rate": 7.7845e-05,
      "loss": 0.6236,
      "step": 15577
    },
    {
      "epoch": 0.8723261283458394,
      "grad_norm": 1.26026451587677,
      "learning_rate": 7.785e-05,
      "loss": 0.5265,
      "step": 15578
    },
    {
      "epoch": 0.8723821256579685,
      "grad_norm": 1.316271185874939,
      "learning_rate": 7.7855e-05,
      "loss": 0.4858,
      "step": 15579
    },
    {
      "epoch": 0.8724381229700975,
      "grad_norm": 1.3338090181350708,
      "learning_rate": 7.786e-05,
      "loss": 0.4271,
      "step": 15580
    },
    {
      "epoch": 0.8724941202822265,
      "grad_norm": 1.221513271331787,
      "learning_rate": 7.7865e-05,
      "loss": 0.3846,
      "step": 15581
    },
    {
      "epoch": 0.8725501175943555,
      "grad_norm": 1.2444828748703003,
      "learning_rate": 7.787e-05,
      "loss": 0.4285,
      "step": 15582
    },
    {
      "epoch": 0.8726061149064845,
      "grad_norm": 1.3883877992630005,
      "learning_rate": 7.787500000000001e-05,
      "loss": 0.3942,
      "step": 15583
    },
    {
      "epoch": 0.8726621122186136,
      "grad_norm": 1.7320107221603394,
      "learning_rate": 7.788e-05,
      "loss": 0.4515,
      "step": 15584
    },
    {
      "epoch": 0.8727181095307425,
      "grad_norm": 1.1985026597976685,
      "learning_rate": 7.7885e-05,
      "loss": 0.4076,
      "step": 15585
    },
    {
      "epoch": 0.8727741068428715,
      "grad_norm": 1.548317551612854,
      "learning_rate": 7.789e-05,
      "loss": 0.3293,
      "step": 15586
    },
    {
      "epoch": 0.8728301041550005,
      "grad_norm": 1.435633897781372,
      "learning_rate": 7.789500000000001e-05,
      "loss": 0.4572,
      "step": 15587
    },
    {
      "epoch": 0.8728861014671295,
      "grad_norm": 1.9437674283981323,
      "learning_rate": 7.790000000000001e-05,
      "loss": 0.508,
      "step": 15588
    },
    {
      "epoch": 0.8729420987792585,
      "grad_norm": 5.298740863800049,
      "learning_rate": 7.790500000000001e-05,
      "loss": 0.5213,
      "step": 15589
    },
    {
      "epoch": 0.8729980960913876,
      "grad_norm": 1.305383324623108,
      "learning_rate": 7.791e-05,
      "loss": 0.4151,
      "step": 15590
    },
    {
      "epoch": 0.8730540934035166,
      "grad_norm": 1.3252203464508057,
      "learning_rate": 7.7915e-05,
      "loss": 0.4882,
      "step": 15591
    },
    {
      "epoch": 0.8731100907156456,
      "grad_norm": 1.3112610578536987,
      "learning_rate": 7.792e-05,
      "loss": 0.3805,
      "step": 15592
    },
    {
      "epoch": 0.8731660880277746,
      "grad_norm": 1.2379817962646484,
      "learning_rate": 7.792500000000001e-05,
      "loss": 0.4596,
      "step": 15593
    },
    {
      "epoch": 0.8732220853399036,
      "grad_norm": 1.6146538257598877,
      "learning_rate": 7.793000000000001e-05,
      "loss": 0.4251,
      "step": 15594
    },
    {
      "epoch": 0.8732780826520327,
      "grad_norm": 1.3385467529296875,
      "learning_rate": 7.7935e-05,
      "loss": 0.4956,
      "step": 15595
    },
    {
      "epoch": 0.8733340799641617,
      "grad_norm": 1.4091204404830933,
      "learning_rate": 7.794e-05,
      "loss": 0.4343,
      "step": 15596
    },
    {
      "epoch": 0.8733900772762907,
      "grad_norm": 1.2925491333007812,
      "learning_rate": 7.7945e-05,
      "loss": 0.4075,
      "step": 15597
    },
    {
      "epoch": 0.8734460745884197,
      "grad_norm": 1.2249665260314941,
      "learning_rate": 7.795e-05,
      "loss": 0.4046,
      "step": 15598
    },
    {
      "epoch": 0.8735020719005487,
      "grad_norm": 1.1581363677978516,
      "learning_rate": 7.7955e-05,
      "loss": 0.3682,
      "step": 15599
    },
    {
      "epoch": 0.8735580692126778,
      "grad_norm": 1.7862924337387085,
      "learning_rate": 7.796e-05,
      "loss": 0.526,
      "step": 15600
    },
    {
      "epoch": 0.8736140665248068,
      "grad_norm": 1.3446723222732544,
      "learning_rate": 7.7965e-05,
      "loss": 0.4079,
      "step": 15601
    },
    {
      "epoch": 0.8736700638369358,
      "grad_norm": 1.334972620010376,
      "learning_rate": 7.797e-05,
      "loss": 0.4373,
      "step": 15602
    },
    {
      "epoch": 0.8737260611490648,
      "grad_norm": 1.2533605098724365,
      "learning_rate": 7.797500000000001e-05,
      "loss": 0.3595,
      "step": 15603
    },
    {
      "epoch": 0.8737820584611938,
      "grad_norm": 1.8424146175384521,
      "learning_rate": 7.798000000000001e-05,
      "loss": 0.566,
      "step": 15604
    },
    {
      "epoch": 0.8738380557733229,
      "grad_norm": 1.4950603246688843,
      "learning_rate": 7.7985e-05,
      "loss": 0.378,
      "step": 15605
    },
    {
      "epoch": 0.8738940530854519,
      "grad_norm": 1.2845555543899536,
      "learning_rate": 7.799e-05,
      "loss": 0.42,
      "step": 15606
    },
    {
      "epoch": 0.8739500503975809,
      "grad_norm": 1.5531342029571533,
      "learning_rate": 7.7995e-05,
      "loss": 0.515,
      "step": 15607
    },
    {
      "epoch": 0.8740060477097099,
      "grad_norm": 1.1322746276855469,
      "learning_rate": 7.800000000000001e-05,
      "loss": 0.354,
      "step": 15608
    },
    {
      "epoch": 0.8740620450218389,
      "grad_norm": 1.6688734292984009,
      "learning_rate": 7.800500000000001e-05,
      "loss": 0.4178,
      "step": 15609
    },
    {
      "epoch": 0.874118042333968,
      "grad_norm": 1.2702504396438599,
      "learning_rate": 7.801000000000001e-05,
      "loss": 0.5135,
      "step": 15610
    },
    {
      "epoch": 0.874174039646097,
      "grad_norm": 1.1688611507415771,
      "learning_rate": 7.8015e-05,
      "loss": 0.3942,
      "step": 15611
    },
    {
      "epoch": 0.874230036958226,
      "grad_norm": 1.407023549079895,
      "learning_rate": 7.802e-05,
      "loss": 0.4159,
      "step": 15612
    },
    {
      "epoch": 0.874286034270355,
      "grad_norm": 1.5352704524993896,
      "learning_rate": 7.8025e-05,
      "loss": 0.4463,
      "step": 15613
    },
    {
      "epoch": 0.874342031582484,
      "grad_norm": 1.5974230766296387,
      "learning_rate": 7.803e-05,
      "loss": 0.4942,
      "step": 15614
    },
    {
      "epoch": 0.874398028894613,
      "grad_norm": 1.779258131980896,
      "learning_rate": 7.803500000000001e-05,
      "loss": 0.515,
      "step": 15615
    },
    {
      "epoch": 0.8744540262067421,
      "grad_norm": 1.6793160438537598,
      "learning_rate": 7.804e-05,
      "loss": 0.4746,
      "step": 15616
    },
    {
      "epoch": 0.8745100235188711,
      "grad_norm": 1.193279504776001,
      "learning_rate": 7.8045e-05,
      "loss": 0.4896,
      "step": 15617
    },
    {
      "epoch": 0.8745660208310001,
      "grad_norm": 1.3777776956558228,
      "learning_rate": 7.805e-05,
      "loss": 0.5491,
      "step": 15618
    },
    {
      "epoch": 0.8746220181431291,
      "grad_norm": 1.351875901222229,
      "learning_rate": 7.8055e-05,
      "loss": 0.4955,
      "step": 15619
    },
    {
      "epoch": 0.8746780154552581,
      "grad_norm": 1.1543796062469482,
      "learning_rate": 7.806e-05,
      "loss": 0.3552,
      "step": 15620
    },
    {
      "epoch": 0.8747340127673872,
      "grad_norm": 1.2773261070251465,
      "learning_rate": 7.8065e-05,
      "loss": 0.4395,
      "step": 15621
    },
    {
      "epoch": 0.8747900100795162,
      "grad_norm": 1.5070065259933472,
      "learning_rate": 7.807e-05,
      "loss": 0.6016,
      "step": 15622
    },
    {
      "epoch": 0.8748460073916452,
      "grad_norm": 1.2362004518508911,
      "learning_rate": 7.807500000000001e-05,
      "loss": 0.4658,
      "step": 15623
    },
    {
      "epoch": 0.8749020047037742,
      "grad_norm": 1.290040135383606,
      "learning_rate": 7.808000000000001e-05,
      "loss": 0.5284,
      "step": 15624
    },
    {
      "epoch": 0.8749580020159032,
      "grad_norm": 5.069690704345703,
      "learning_rate": 7.808500000000001e-05,
      "loss": 0.57,
      "step": 15625
    },
    {
      "epoch": 0.8750139993280323,
      "grad_norm": 1.8353949785232544,
      "learning_rate": 7.809e-05,
      "loss": 0.6024,
      "step": 15626
    },
    {
      "epoch": 0.8750699966401613,
      "grad_norm": 1.3642324209213257,
      "learning_rate": 7.8095e-05,
      "loss": 0.3545,
      "step": 15627
    },
    {
      "epoch": 0.8751259939522903,
      "grad_norm": 1.3173775672912598,
      "learning_rate": 7.81e-05,
      "loss": 0.4785,
      "step": 15628
    },
    {
      "epoch": 0.8751819912644193,
      "grad_norm": 1.51857590675354,
      "learning_rate": 7.810500000000001e-05,
      "loss": 0.39,
      "step": 15629
    },
    {
      "epoch": 0.8752379885765483,
      "grad_norm": 1.6354382038116455,
      "learning_rate": 7.811000000000001e-05,
      "loss": 0.4176,
      "step": 15630
    },
    {
      "epoch": 0.8752939858886774,
      "grad_norm": 1.3690122365951538,
      "learning_rate": 7.811500000000001e-05,
      "loss": 0.4327,
      "step": 15631
    },
    {
      "epoch": 0.8753499832008064,
      "grad_norm": 1.4135335683822632,
      "learning_rate": 7.812e-05,
      "loss": 0.5097,
      "step": 15632
    },
    {
      "epoch": 0.8754059805129354,
      "grad_norm": 1.1350250244140625,
      "learning_rate": 7.8125e-05,
      "loss": 0.4193,
      "step": 15633
    },
    {
      "epoch": 0.8754619778250644,
      "grad_norm": 1.281648874282837,
      "learning_rate": 7.813e-05,
      "loss": 0.37,
      "step": 15634
    },
    {
      "epoch": 0.8755179751371934,
      "grad_norm": 1.2919368743896484,
      "learning_rate": 7.8135e-05,
      "loss": 0.4193,
      "step": 15635
    },
    {
      "epoch": 0.8755739724493224,
      "grad_norm": 1.283706784248352,
      "learning_rate": 7.814000000000001e-05,
      "loss": 0.4654,
      "step": 15636
    },
    {
      "epoch": 0.8756299697614515,
      "grad_norm": 1.662026047706604,
      "learning_rate": 7.8145e-05,
      "loss": 0.3832,
      "step": 15637
    },
    {
      "epoch": 0.8756859670735805,
      "grad_norm": 1.2785576581954956,
      "learning_rate": 7.815e-05,
      "loss": 0.3598,
      "step": 15638
    },
    {
      "epoch": 0.8757419643857095,
      "grad_norm": 1.264336109161377,
      "learning_rate": 7.8155e-05,
      "loss": 0.4208,
      "step": 15639
    },
    {
      "epoch": 0.8757979616978385,
      "grad_norm": 1.2264667749404907,
      "learning_rate": 7.816e-05,
      "loss": 0.4188,
      "step": 15640
    },
    {
      "epoch": 0.8758539590099675,
      "grad_norm": 1.6351864337921143,
      "learning_rate": 7.8165e-05,
      "loss": 0.5137,
      "step": 15641
    },
    {
      "epoch": 0.8759099563220966,
      "grad_norm": 1.2195281982421875,
      "learning_rate": 7.817e-05,
      "loss": 0.4903,
      "step": 15642
    },
    {
      "epoch": 0.8759659536342256,
      "grad_norm": 1.5257734060287476,
      "learning_rate": 7.8175e-05,
      "loss": 0.4517,
      "step": 15643
    },
    {
      "epoch": 0.8760219509463546,
      "grad_norm": 1.3661330938339233,
      "learning_rate": 7.818000000000001e-05,
      "loss": 0.4797,
      "step": 15644
    },
    {
      "epoch": 0.8760779482584836,
      "grad_norm": 1.2897977828979492,
      "learning_rate": 7.818500000000001e-05,
      "loss": 0.5837,
      "step": 15645
    },
    {
      "epoch": 0.8761339455706126,
      "grad_norm": 1.2548011541366577,
      "learning_rate": 7.819000000000001e-05,
      "loss": 0.5132,
      "step": 15646
    },
    {
      "epoch": 0.8761899428827417,
      "grad_norm": 1.4747647047042847,
      "learning_rate": 7.8195e-05,
      "loss": 0.5048,
      "step": 15647
    },
    {
      "epoch": 0.8762459401948707,
      "grad_norm": 1.3684711456298828,
      "learning_rate": 7.82e-05,
      "loss": 0.4655,
      "step": 15648
    },
    {
      "epoch": 0.8763019375069997,
      "grad_norm": 1.2558674812316895,
      "learning_rate": 7.8205e-05,
      "loss": 0.3309,
      "step": 15649
    },
    {
      "epoch": 0.8763579348191287,
      "grad_norm": 1.528770089149475,
      "learning_rate": 7.821000000000001e-05,
      "loss": 0.502,
      "step": 15650
    },
    {
      "epoch": 0.8764139321312577,
      "grad_norm": 1.3130013942718506,
      "learning_rate": 7.821500000000001e-05,
      "loss": 0.4539,
      "step": 15651
    },
    {
      "epoch": 0.8764699294433868,
      "grad_norm": 3.9340603351593018,
      "learning_rate": 7.822e-05,
      "loss": 0.4584,
      "step": 15652
    },
    {
      "epoch": 0.8765259267555158,
      "grad_norm": 1.3083142042160034,
      "learning_rate": 7.8225e-05,
      "loss": 0.3608,
      "step": 15653
    },
    {
      "epoch": 0.8765819240676448,
      "grad_norm": 1.1950018405914307,
      "learning_rate": 7.823e-05,
      "loss": 0.4288,
      "step": 15654
    },
    {
      "epoch": 0.8766379213797738,
      "grad_norm": 1.6428717374801636,
      "learning_rate": 7.8235e-05,
      "loss": 0.6502,
      "step": 15655
    },
    {
      "epoch": 0.8766939186919028,
      "grad_norm": 1.3999062776565552,
      "learning_rate": 7.824e-05,
      "loss": 0.5594,
      "step": 15656
    },
    {
      "epoch": 0.8767499160040318,
      "grad_norm": 1.3282313346862793,
      "learning_rate": 7.824500000000001e-05,
      "loss": 0.4249,
      "step": 15657
    },
    {
      "epoch": 0.8768059133161609,
      "grad_norm": 1.4244219064712524,
      "learning_rate": 7.825e-05,
      "loss": 0.3722,
      "step": 15658
    },
    {
      "epoch": 0.8768619106282899,
      "grad_norm": 1.458259105682373,
      "learning_rate": 7.8255e-05,
      "loss": 0.3885,
      "step": 15659
    },
    {
      "epoch": 0.8769179079404189,
      "grad_norm": 1.1726912260055542,
      "learning_rate": 7.826e-05,
      "loss": 0.4037,
      "step": 15660
    },
    {
      "epoch": 0.8769739052525479,
      "grad_norm": 1.3612245321273804,
      "learning_rate": 7.8265e-05,
      "loss": 0.4036,
      "step": 15661
    },
    {
      "epoch": 0.8770299025646769,
      "grad_norm": 1.4680825471878052,
      "learning_rate": 7.827e-05,
      "loss": 0.3367,
      "step": 15662
    },
    {
      "epoch": 0.877085899876806,
      "grad_norm": 2.020721435546875,
      "learning_rate": 7.827499999999999e-05,
      "loss": 0.5772,
      "step": 15663
    },
    {
      "epoch": 0.877141897188935,
      "grad_norm": 1.4212771654129028,
      "learning_rate": 7.828e-05,
      "loss": 0.6183,
      "step": 15664
    },
    {
      "epoch": 0.877197894501064,
      "grad_norm": 1.1750584840774536,
      "learning_rate": 7.828500000000001e-05,
      "loss": 0.3967,
      "step": 15665
    },
    {
      "epoch": 0.877253891813193,
      "grad_norm": 1.2902772426605225,
      "learning_rate": 7.829000000000001e-05,
      "loss": 0.4072,
      "step": 15666
    },
    {
      "epoch": 0.877309889125322,
      "grad_norm": 1.3469256162643433,
      "learning_rate": 7.829500000000001e-05,
      "loss": 0.3589,
      "step": 15667
    },
    {
      "epoch": 0.877365886437451,
      "grad_norm": 1.5020853281021118,
      "learning_rate": 7.83e-05,
      "loss": 0.6214,
      "step": 15668
    },
    {
      "epoch": 0.87742188374958,
      "grad_norm": 1.1948949098587036,
      "learning_rate": 7.8305e-05,
      "loss": 0.4081,
      "step": 15669
    },
    {
      "epoch": 0.877477881061709,
      "grad_norm": 1.28453528881073,
      "learning_rate": 7.831e-05,
      "loss": 0.3609,
      "step": 15670
    },
    {
      "epoch": 0.877533878373838,
      "grad_norm": 1.2847203016281128,
      "learning_rate": 7.831500000000001e-05,
      "loss": 0.4036,
      "step": 15671
    },
    {
      "epoch": 0.877589875685967,
      "grad_norm": 1.308562994003296,
      "learning_rate": 7.832000000000001e-05,
      "loss": 0.3373,
      "step": 15672
    },
    {
      "epoch": 0.877645872998096,
      "grad_norm": 1.245770812034607,
      "learning_rate": 7.8325e-05,
      "loss": 0.4617,
      "step": 15673
    },
    {
      "epoch": 0.8777018703102251,
      "grad_norm": 1.3665430545806885,
      "learning_rate": 7.833e-05,
      "loss": 0.622,
      "step": 15674
    },
    {
      "epoch": 0.8777578676223541,
      "grad_norm": 1.1770883798599243,
      "learning_rate": 7.8335e-05,
      "loss": 0.3847,
      "step": 15675
    },
    {
      "epoch": 0.8778138649344831,
      "grad_norm": 1.5456660985946655,
      "learning_rate": 7.834e-05,
      "loss": 0.3619,
      "step": 15676
    },
    {
      "epoch": 0.8778698622466121,
      "grad_norm": 1.3694406747817993,
      "learning_rate": 7.8345e-05,
      "loss": 0.3838,
      "step": 15677
    },
    {
      "epoch": 0.8779258595587411,
      "grad_norm": 1.4143106937408447,
      "learning_rate": 7.835000000000001e-05,
      "loss": 0.5196,
      "step": 15678
    },
    {
      "epoch": 0.8779818568708702,
      "grad_norm": 1.2907812595367432,
      "learning_rate": 7.8355e-05,
      "loss": 0.3555,
      "step": 15679
    },
    {
      "epoch": 0.8780378541829992,
      "grad_norm": 1.3018662929534912,
      "learning_rate": 7.836e-05,
      "loss": 0.4758,
      "step": 15680
    },
    {
      "epoch": 0.8780938514951282,
      "grad_norm": 1.2399076223373413,
      "learning_rate": 7.8365e-05,
      "loss": 0.32,
      "step": 15681
    },
    {
      "epoch": 0.8781498488072572,
      "grad_norm": 1.7134053707122803,
      "learning_rate": 7.837e-05,
      "loss": 0.3929,
      "step": 15682
    },
    {
      "epoch": 0.8782058461193862,
      "grad_norm": 1.286105990409851,
      "learning_rate": 7.8375e-05,
      "loss": 0.4124,
      "step": 15683
    },
    {
      "epoch": 0.8782618434315153,
      "grad_norm": 1.7660576105117798,
      "learning_rate": 7.838e-05,
      "loss": 0.4636,
      "step": 15684
    },
    {
      "epoch": 0.8783178407436443,
      "grad_norm": 1.1781138181686401,
      "learning_rate": 7.8385e-05,
      "loss": 0.5726,
      "step": 15685
    },
    {
      "epoch": 0.8783738380557733,
      "grad_norm": 1.2534667253494263,
      "learning_rate": 7.839000000000001e-05,
      "loss": 0.3376,
      "step": 15686
    },
    {
      "epoch": 0.8784298353679023,
      "grad_norm": 1.2173240184783936,
      "learning_rate": 7.839500000000001e-05,
      "loss": 0.4651,
      "step": 15687
    },
    {
      "epoch": 0.8784858326800313,
      "grad_norm": 1.2226678133010864,
      "learning_rate": 7.840000000000001e-05,
      "loss": 0.5286,
      "step": 15688
    },
    {
      "epoch": 0.8785418299921604,
      "grad_norm": 1.1224699020385742,
      "learning_rate": 7.8405e-05,
      "loss": 0.5105,
      "step": 15689
    },
    {
      "epoch": 0.8785978273042894,
      "grad_norm": 1.8798848390579224,
      "learning_rate": 7.841e-05,
      "loss": 0.5881,
      "step": 15690
    },
    {
      "epoch": 0.8786538246164184,
      "grad_norm": 1.2383431196212769,
      "learning_rate": 7.8415e-05,
      "loss": 0.3522,
      "step": 15691
    },
    {
      "epoch": 0.8787098219285474,
      "grad_norm": 1.2047781944274902,
      "learning_rate": 7.842e-05,
      "loss": 0.3768,
      "step": 15692
    },
    {
      "epoch": 0.8787658192406764,
      "grad_norm": 1.237811803817749,
      "learning_rate": 7.842500000000001e-05,
      "loss": 0.3701,
      "step": 15693
    },
    {
      "epoch": 0.8788218165528054,
      "grad_norm": 1.475827693939209,
      "learning_rate": 7.843e-05,
      "loss": 0.4622,
      "step": 15694
    },
    {
      "epoch": 0.8788778138649345,
      "grad_norm": 1.2367972135543823,
      "learning_rate": 7.8435e-05,
      "loss": 0.4071,
      "step": 15695
    },
    {
      "epoch": 0.8789338111770635,
      "grad_norm": 1.360946774482727,
      "learning_rate": 7.844e-05,
      "loss": 0.5247,
      "step": 15696
    },
    {
      "epoch": 0.8789898084891925,
      "grad_norm": 1.1919349431991577,
      "learning_rate": 7.8445e-05,
      "loss": 0.3536,
      "step": 15697
    },
    {
      "epoch": 0.8790458058013215,
      "grad_norm": 1.7997097969055176,
      "learning_rate": 7.845e-05,
      "loss": 0.6277,
      "step": 15698
    },
    {
      "epoch": 0.8791018031134505,
      "grad_norm": 1.5822625160217285,
      "learning_rate": 7.845500000000001e-05,
      "loss": 0.6634,
      "step": 15699
    },
    {
      "epoch": 0.8791578004255796,
      "grad_norm": 1.187794804573059,
      "learning_rate": 7.846e-05,
      "loss": 0.3938,
      "step": 15700
    },
    {
      "epoch": 0.8792137977377086,
      "grad_norm": 1.319350242614746,
      "learning_rate": 7.8465e-05,
      "loss": 0.3277,
      "step": 15701
    },
    {
      "epoch": 0.8792697950498376,
      "grad_norm": 1.2361481189727783,
      "learning_rate": 7.847e-05,
      "loss": 0.4237,
      "step": 15702
    },
    {
      "epoch": 0.8793257923619666,
      "grad_norm": 1.2398425340652466,
      "learning_rate": 7.8475e-05,
      "loss": 0.4074,
      "step": 15703
    },
    {
      "epoch": 0.8793817896740956,
      "grad_norm": 1.4108473062515259,
      "learning_rate": 7.848000000000001e-05,
      "loss": 0.5114,
      "step": 15704
    },
    {
      "epoch": 0.8794377869862247,
      "grad_norm": 1.2769901752471924,
      "learning_rate": 7.8485e-05,
      "loss": 0.5534,
      "step": 15705
    },
    {
      "epoch": 0.8794937842983537,
      "grad_norm": 1.167389154434204,
      "learning_rate": 7.849e-05,
      "loss": 0.3794,
      "step": 15706
    },
    {
      "epoch": 0.8795497816104827,
      "grad_norm": 1.3811312913894653,
      "learning_rate": 7.849500000000001e-05,
      "loss": 0.4559,
      "step": 15707
    },
    {
      "epoch": 0.8796057789226117,
      "grad_norm": 1.0768953561782837,
      "learning_rate": 7.850000000000001e-05,
      "loss": 0.3338,
      "step": 15708
    },
    {
      "epoch": 0.8796617762347407,
      "grad_norm": 1.198590874671936,
      "learning_rate": 7.850500000000001e-05,
      "loss": 0.3763,
      "step": 15709
    },
    {
      "epoch": 0.8797177735468698,
      "grad_norm": 1.430693507194519,
      "learning_rate": 7.851e-05,
      "loss": 0.4116,
      "step": 15710
    },
    {
      "epoch": 0.8797737708589988,
      "grad_norm": 1.407012939453125,
      "learning_rate": 7.8515e-05,
      "loss": 0.4667,
      "step": 15711
    },
    {
      "epoch": 0.8798297681711278,
      "grad_norm": 1.4793479442596436,
      "learning_rate": 7.852e-05,
      "loss": 0.3616,
      "step": 15712
    },
    {
      "epoch": 0.8798857654832568,
      "grad_norm": 1.2668476104736328,
      "learning_rate": 7.8525e-05,
      "loss": 0.58,
      "step": 15713
    },
    {
      "epoch": 0.8799417627953858,
      "grad_norm": 1.2861369848251343,
      "learning_rate": 7.853000000000001e-05,
      "loss": 0.5646,
      "step": 15714
    },
    {
      "epoch": 0.8799977601075148,
      "grad_norm": 1.1182711124420166,
      "learning_rate": 7.8535e-05,
      "loss": 0.385,
      "step": 15715
    },
    {
      "epoch": 0.8800537574196439,
      "grad_norm": 1.5841948986053467,
      "learning_rate": 7.854e-05,
      "loss": 0.4527,
      "step": 15716
    },
    {
      "epoch": 0.8801097547317729,
      "grad_norm": 1.2600511312484741,
      "learning_rate": 7.8545e-05,
      "loss": 0.4581,
      "step": 15717
    },
    {
      "epoch": 0.8801657520439019,
      "grad_norm": 1.081499695777893,
      "learning_rate": 7.855e-05,
      "loss": 0.363,
      "step": 15718
    },
    {
      "epoch": 0.8802217493560309,
      "grad_norm": 1.4237945079803467,
      "learning_rate": 7.8555e-05,
      "loss": 0.5863,
      "step": 15719
    },
    {
      "epoch": 0.8802777466681599,
      "grad_norm": 1.3789386749267578,
      "learning_rate": 7.856000000000001e-05,
      "loss": 0.3903,
      "step": 15720
    },
    {
      "epoch": 0.880333743980289,
      "grad_norm": 1.7766565084457397,
      "learning_rate": 7.8565e-05,
      "loss": 0.5534,
      "step": 15721
    },
    {
      "epoch": 0.880389741292418,
      "grad_norm": 1.305375576019287,
      "learning_rate": 7.857e-05,
      "loss": 0.4963,
      "step": 15722
    },
    {
      "epoch": 0.880445738604547,
      "grad_norm": 1.1422768831253052,
      "learning_rate": 7.8575e-05,
      "loss": 0.3225,
      "step": 15723
    },
    {
      "epoch": 0.880501735916676,
      "grad_norm": 1.2995232343673706,
      "learning_rate": 7.858000000000001e-05,
      "loss": 0.3413,
      "step": 15724
    },
    {
      "epoch": 0.880557733228805,
      "grad_norm": 1.1238868236541748,
      "learning_rate": 7.858500000000001e-05,
      "loss": 0.3428,
      "step": 15725
    },
    {
      "epoch": 0.8806137305409341,
      "grad_norm": 2.277445077896118,
      "learning_rate": 7.859e-05,
      "loss": 0.4319,
      "step": 15726
    },
    {
      "epoch": 0.8806697278530631,
      "grad_norm": 1.423354983329773,
      "learning_rate": 7.8595e-05,
      "loss": 0.5173,
      "step": 15727
    },
    {
      "epoch": 0.8807257251651921,
      "grad_norm": 1.0879390239715576,
      "learning_rate": 7.860000000000001e-05,
      "loss": 0.4703,
      "step": 15728
    },
    {
      "epoch": 0.8807817224773211,
      "grad_norm": 1.377094030380249,
      "learning_rate": 7.860500000000001e-05,
      "loss": 0.459,
      "step": 15729
    },
    {
      "epoch": 0.8808377197894501,
      "grad_norm": 1.2914865016937256,
      "learning_rate": 7.861000000000001e-05,
      "loss": 0.4543,
      "step": 15730
    },
    {
      "epoch": 0.8808937171015792,
      "grad_norm": 1.1602896451950073,
      "learning_rate": 7.8615e-05,
      "loss": 0.3261,
      "step": 15731
    },
    {
      "epoch": 0.8809497144137082,
      "grad_norm": 1.5694911479949951,
      "learning_rate": 7.862e-05,
      "loss": 0.4,
      "step": 15732
    },
    {
      "epoch": 0.8810057117258372,
      "grad_norm": 1.3546191453933716,
      "learning_rate": 7.8625e-05,
      "loss": 0.3636,
      "step": 15733
    },
    {
      "epoch": 0.8810617090379662,
      "grad_norm": 1.2792226076126099,
      "learning_rate": 7.863e-05,
      "loss": 0.4535,
      "step": 15734
    },
    {
      "epoch": 0.8811177063500952,
      "grad_norm": 1.7178654670715332,
      "learning_rate": 7.863500000000001e-05,
      "loss": 0.3848,
      "step": 15735
    },
    {
      "epoch": 0.8811737036622243,
      "grad_norm": 1.3578240871429443,
      "learning_rate": 7.864e-05,
      "loss": 0.4088,
      "step": 15736
    },
    {
      "epoch": 0.8812297009743533,
      "grad_norm": 1.5478277206420898,
      "learning_rate": 7.8645e-05,
      "loss": 0.4679,
      "step": 15737
    },
    {
      "epoch": 0.8812856982864823,
      "grad_norm": 1.336410403251648,
      "learning_rate": 7.865e-05,
      "loss": 0.4307,
      "step": 15738
    },
    {
      "epoch": 0.8813416955986113,
      "grad_norm": 1.3684492111206055,
      "learning_rate": 7.8655e-05,
      "loss": 0.4762,
      "step": 15739
    },
    {
      "epoch": 0.8813976929107403,
      "grad_norm": 1.430402398109436,
      "learning_rate": 7.866e-05,
      "loss": 0.6,
      "step": 15740
    },
    {
      "epoch": 0.8814536902228693,
      "grad_norm": 1.1227734088897705,
      "learning_rate": 7.866499999999999e-05,
      "loss": 0.3133,
      "step": 15741
    },
    {
      "epoch": 0.8815096875349984,
      "grad_norm": 1.2948923110961914,
      "learning_rate": 7.867e-05,
      "loss": 0.4961,
      "step": 15742
    },
    {
      "epoch": 0.8815656848471274,
      "grad_norm": 1.7092138528823853,
      "learning_rate": 7.8675e-05,
      "loss": 0.5236,
      "step": 15743
    },
    {
      "epoch": 0.8816216821592564,
      "grad_norm": 1.4647992849349976,
      "learning_rate": 7.868000000000001e-05,
      "loss": 0.5177,
      "step": 15744
    },
    {
      "epoch": 0.8816776794713854,
      "grad_norm": 1.3417359590530396,
      "learning_rate": 7.868500000000001e-05,
      "loss": 0.4036,
      "step": 15745
    },
    {
      "epoch": 0.8817336767835144,
      "grad_norm": 1.4374005794525146,
      "learning_rate": 7.869000000000001e-05,
      "loss": 0.4864,
      "step": 15746
    },
    {
      "epoch": 0.8817896740956435,
      "grad_norm": 1.2635575532913208,
      "learning_rate": 7.8695e-05,
      "loss": 0.4443,
      "step": 15747
    },
    {
      "epoch": 0.8818456714077725,
      "grad_norm": 1.4097075462341309,
      "learning_rate": 7.87e-05,
      "loss": 0.4626,
      "step": 15748
    },
    {
      "epoch": 0.8819016687199015,
      "grad_norm": 1.5398057699203491,
      "learning_rate": 7.870500000000001e-05,
      "loss": 0.6255,
      "step": 15749
    },
    {
      "epoch": 0.8819576660320304,
      "grad_norm": 1.2846860885620117,
      "learning_rate": 7.871000000000001e-05,
      "loss": 0.422,
      "step": 15750
    },
    {
      "epoch": 0.8820136633441594,
      "grad_norm": 1.2987191677093506,
      "learning_rate": 7.871500000000001e-05,
      "loss": 0.4639,
      "step": 15751
    },
    {
      "epoch": 0.8820696606562884,
      "grad_norm": 1.2411824464797974,
      "learning_rate": 7.872e-05,
      "loss": 0.4454,
      "step": 15752
    },
    {
      "epoch": 0.8821256579684175,
      "grad_norm": 1.18653404712677,
      "learning_rate": 7.8725e-05,
      "loss": 0.4113,
      "step": 15753
    },
    {
      "epoch": 0.8821816552805465,
      "grad_norm": 1.386753797531128,
      "learning_rate": 7.873e-05,
      "loss": 0.5086,
      "step": 15754
    },
    {
      "epoch": 0.8822376525926755,
      "grad_norm": 1.205653429031372,
      "learning_rate": 7.8735e-05,
      "loss": 0.4265,
      "step": 15755
    },
    {
      "epoch": 0.8822936499048045,
      "grad_norm": 1.688432216644287,
      "learning_rate": 7.874000000000001e-05,
      "loss": 0.531,
      "step": 15756
    },
    {
      "epoch": 0.8823496472169335,
      "grad_norm": 1.3085002899169922,
      "learning_rate": 7.8745e-05,
      "loss": 0.4349,
      "step": 15757
    },
    {
      "epoch": 0.8824056445290626,
      "grad_norm": 1.0334367752075195,
      "learning_rate": 7.875e-05,
      "loss": 0.4277,
      "step": 15758
    },
    {
      "epoch": 0.8824616418411916,
      "grad_norm": 1.3466542959213257,
      "learning_rate": 7.8755e-05,
      "loss": 0.4581,
      "step": 15759
    },
    {
      "epoch": 0.8825176391533206,
      "grad_norm": 1.5849618911743164,
      "learning_rate": 7.876e-05,
      "loss": 0.5612,
      "step": 15760
    },
    {
      "epoch": 0.8825736364654496,
      "grad_norm": 1.3699510097503662,
      "learning_rate": 7.8765e-05,
      "loss": 0.4414,
      "step": 15761
    },
    {
      "epoch": 0.8826296337775786,
      "grad_norm": 1.2993191480636597,
      "learning_rate": 7.876999999999999e-05,
      "loss": 0.4806,
      "step": 15762
    },
    {
      "epoch": 0.8826856310897077,
      "grad_norm": 1.302834153175354,
      "learning_rate": 7.8775e-05,
      "loss": 0.3442,
      "step": 15763
    },
    {
      "epoch": 0.8827416284018367,
      "grad_norm": 1.1885366439819336,
      "learning_rate": 7.878e-05,
      "loss": 0.4186,
      "step": 15764
    },
    {
      "epoch": 0.8827976257139657,
      "grad_norm": 2.4305038452148438,
      "learning_rate": 7.878500000000001e-05,
      "loss": 0.786,
      "step": 15765
    },
    {
      "epoch": 0.8828536230260947,
      "grad_norm": 1.4897726774215698,
      "learning_rate": 7.879000000000001e-05,
      "loss": 0.5225,
      "step": 15766
    },
    {
      "epoch": 0.8829096203382237,
      "grad_norm": 1.3332767486572266,
      "learning_rate": 7.879500000000001e-05,
      "loss": 0.4555,
      "step": 15767
    },
    {
      "epoch": 0.8829656176503528,
      "grad_norm": 1.3837586641311646,
      "learning_rate": 7.88e-05,
      "loss": 0.5187,
      "step": 15768
    },
    {
      "epoch": 0.8830216149624818,
      "grad_norm": 1.3469719886779785,
      "learning_rate": 7.8805e-05,
      "loss": 0.3342,
      "step": 15769
    },
    {
      "epoch": 0.8830776122746108,
      "grad_norm": 1.5540140867233276,
      "learning_rate": 7.881e-05,
      "loss": 0.4922,
      "step": 15770
    },
    {
      "epoch": 0.8831336095867398,
      "grad_norm": 1.3087104558944702,
      "learning_rate": 7.881500000000001e-05,
      "loss": 0.4845,
      "step": 15771
    },
    {
      "epoch": 0.8831896068988688,
      "grad_norm": 1.4179688692092896,
      "learning_rate": 7.882000000000001e-05,
      "loss": 0.422,
      "step": 15772
    },
    {
      "epoch": 0.8832456042109978,
      "grad_norm": 1.3407858610153198,
      "learning_rate": 7.8825e-05,
      "loss": 0.5993,
      "step": 15773
    },
    {
      "epoch": 0.8833016015231269,
      "grad_norm": 1.298976182937622,
      "learning_rate": 7.883e-05,
      "loss": 0.6,
      "step": 15774
    },
    {
      "epoch": 0.8833575988352559,
      "grad_norm": 2.641345739364624,
      "learning_rate": 7.8835e-05,
      "loss": 0.4339,
      "step": 15775
    },
    {
      "epoch": 0.8834135961473849,
      "grad_norm": 1.4696484804153442,
      "learning_rate": 7.884e-05,
      "loss": 0.7233,
      "step": 15776
    },
    {
      "epoch": 0.8834695934595139,
      "grad_norm": 1.3720401525497437,
      "learning_rate": 7.884500000000001e-05,
      "loss": 0.4915,
      "step": 15777
    },
    {
      "epoch": 0.8835255907716429,
      "grad_norm": 1.4552339315414429,
      "learning_rate": 7.885e-05,
      "loss": 0.5732,
      "step": 15778
    },
    {
      "epoch": 0.883581588083772,
      "grad_norm": 1.0989445447921753,
      "learning_rate": 7.8855e-05,
      "loss": 0.4422,
      "step": 15779
    },
    {
      "epoch": 0.883637585395901,
      "grad_norm": 1.638524055480957,
      "learning_rate": 7.886e-05,
      "loss": 0.6128,
      "step": 15780
    },
    {
      "epoch": 0.88369358270803,
      "grad_norm": 1.354801893234253,
      "learning_rate": 7.8865e-05,
      "loss": 0.4369,
      "step": 15781
    },
    {
      "epoch": 0.883749580020159,
      "grad_norm": 1.3780988454818726,
      "learning_rate": 7.887e-05,
      "loss": 0.5277,
      "step": 15782
    },
    {
      "epoch": 0.883805577332288,
      "grad_norm": 2.5515425205230713,
      "learning_rate": 7.887499999999999e-05,
      "loss": 0.4847,
      "step": 15783
    },
    {
      "epoch": 0.8838615746444171,
      "grad_norm": 1.3097612857818604,
      "learning_rate": 7.888e-05,
      "loss": 0.489,
      "step": 15784
    },
    {
      "epoch": 0.8839175719565461,
      "grad_norm": 1.5259464979171753,
      "learning_rate": 7.888500000000001e-05,
      "loss": 0.6229,
      "step": 15785
    },
    {
      "epoch": 0.8839735692686751,
      "grad_norm": 1.48321533203125,
      "learning_rate": 7.889000000000001e-05,
      "loss": 0.4613,
      "step": 15786
    },
    {
      "epoch": 0.8840295665808041,
      "grad_norm": 1.4632923603057861,
      "learning_rate": 7.889500000000001e-05,
      "loss": 0.5773,
      "step": 15787
    },
    {
      "epoch": 0.8840855638929331,
      "grad_norm": 1.2408758401870728,
      "learning_rate": 7.890000000000001e-05,
      "loss": 0.4215,
      "step": 15788
    },
    {
      "epoch": 0.8841415612050622,
      "grad_norm": 1.5239912271499634,
      "learning_rate": 7.8905e-05,
      "loss": 0.404,
      "step": 15789
    },
    {
      "epoch": 0.8841975585171912,
      "grad_norm": 1.1883434057235718,
      "learning_rate": 7.891e-05,
      "loss": 0.4913,
      "step": 15790
    },
    {
      "epoch": 0.8842535558293202,
      "grad_norm": 1.1972849369049072,
      "learning_rate": 7.8915e-05,
      "loss": 0.3678,
      "step": 15791
    },
    {
      "epoch": 0.8843095531414492,
      "grad_norm": 1.5056698322296143,
      "learning_rate": 7.892000000000001e-05,
      "loss": 0.4903,
      "step": 15792
    },
    {
      "epoch": 0.8843655504535782,
      "grad_norm": 1.1644717454910278,
      "learning_rate": 7.892500000000001e-05,
      "loss": 0.3522,
      "step": 15793
    },
    {
      "epoch": 0.8844215477657072,
      "grad_norm": 1.3911675214767456,
      "learning_rate": 7.893e-05,
      "loss": 0.4528,
      "step": 15794
    },
    {
      "epoch": 0.8844775450778363,
      "grad_norm": 1.78517484664917,
      "learning_rate": 7.8935e-05,
      "loss": 0.6006,
      "step": 15795
    },
    {
      "epoch": 0.8845335423899653,
      "grad_norm": 1.4201488494873047,
      "learning_rate": 7.894e-05,
      "loss": 0.4007,
      "step": 15796
    },
    {
      "epoch": 0.8845895397020943,
      "grad_norm": 1.1176270246505737,
      "learning_rate": 7.8945e-05,
      "loss": 0.3761,
      "step": 15797
    },
    {
      "epoch": 0.8846455370142233,
      "grad_norm": 1.5850352048873901,
      "learning_rate": 7.895000000000001e-05,
      "loss": 0.5135,
      "step": 15798
    },
    {
      "epoch": 0.8847015343263523,
      "grad_norm": 1.4305088520050049,
      "learning_rate": 7.8955e-05,
      "loss": 0.5224,
      "step": 15799
    },
    {
      "epoch": 0.8847575316384814,
      "grad_norm": 1.3241698741912842,
      "learning_rate": 7.896e-05,
      "loss": 0.4852,
      "step": 15800
    },
    {
      "epoch": 0.8848135289506104,
      "grad_norm": 1.5602823495864868,
      "learning_rate": 7.8965e-05,
      "loss": 0.5188,
      "step": 15801
    },
    {
      "epoch": 0.8848695262627394,
      "grad_norm": 1.3524249792099,
      "learning_rate": 7.897e-05,
      "loss": 0.462,
      "step": 15802
    },
    {
      "epoch": 0.8849255235748684,
      "grad_norm": 1.0575517416000366,
      "learning_rate": 7.8975e-05,
      "loss": 0.3493,
      "step": 15803
    },
    {
      "epoch": 0.8849815208869974,
      "grad_norm": 1.414480447769165,
      "learning_rate": 7.897999999999999e-05,
      "loss": 0.4526,
      "step": 15804
    },
    {
      "epoch": 0.8850375181991265,
      "grad_norm": 1.4470492601394653,
      "learning_rate": 7.8985e-05,
      "loss": 0.3683,
      "step": 15805
    },
    {
      "epoch": 0.8850935155112555,
      "grad_norm": 1.4290904998779297,
      "learning_rate": 7.899000000000001e-05,
      "loss": 0.5473,
      "step": 15806
    },
    {
      "epoch": 0.8851495128233845,
      "grad_norm": 1.4914425611495972,
      "learning_rate": 7.899500000000001e-05,
      "loss": 0.5157,
      "step": 15807
    },
    {
      "epoch": 0.8852055101355135,
      "grad_norm": 2.0020601749420166,
      "learning_rate": 7.900000000000001e-05,
      "loss": 0.5264,
      "step": 15808
    },
    {
      "epoch": 0.8852615074476425,
      "grad_norm": 1.4715309143066406,
      "learning_rate": 7.9005e-05,
      "loss": 0.5628,
      "step": 15809
    },
    {
      "epoch": 0.8853175047597716,
      "grad_norm": 1.3492498397827148,
      "learning_rate": 7.901e-05,
      "loss": 0.5557,
      "step": 15810
    },
    {
      "epoch": 0.8853735020719006,
      "grad_norm": 1.3634693622589111,
      "learning_rate": 7.9015e-05,
      "loss": 0.4594,
      "step": 15811
    },
    {
      "epoch": 0.8854294993840296,
      "grad_norm": 1.4548414945602417,
      "learning_rate": 7.902e-05,
      "loss": 0.3877,
      "step": 15812
    },
    {
      "epoch": 0.8854854966961586,
      "grad_norm": 1.302975058555603,
      "learning_rate": 7.902500000000001e-05,
      "loss": 0.4062,
      "step": 15813
    },
    {
      "epoch": 0.8855414940082876,
      "grad_norm": 1.6429636478424072,
      "learning_rate": 7.903000000000001e-05,
      "loss": 0.644,
      "step": 15814
    },
    {
      "epoch": 0.8855974913204167,
      "grad_norm": 1.5866872072219849,
      "learning_rate": 7.9035e-05,
      "loss": 0.5736,
      "step": 15815
    },
    {
      "epoch": 0.8856534886325457,
      "grad_norm": 1.7080777883529663,
      "learning_rate": 7.904e-05,
      "loss": 0.529,
      "step": 15816
    },
    {
      "epoch": 0.8857094859446747,
      "grad_norm": 1.3886007070541382,
      "learning_rate": 7.9045e-05,
      "loss": 0.3237,
      "step": 15817
    },
    {
      "epoch": 0.8857654832568037,
      "grad_norm": 1.5394459962844849,
      "learning_rate": 7.905e-05,
      "loss": 0.4803,
      "step": 15818
    },
    {
      "epoch": 0.8858214805689327,
      "grad_norm": 1.3017991781234741,
      "learning_rate": 7.905500000000001e-05,
      "loss": 0.4171,
      "step": 15819
    },
    {
      "epoch": 0.8858774778810617,
      "grad_norm": 1.491532802581787,
      "learning_rate": 7.906e-05,
      "loss": 0.5411,
      "step": 15820
    },
    {
      "epoch": 0.8859334751931908,
      "grad_norm": 1.353472113609314,
      "learning_rate": 7.9065e-05,
      "loss": 0.4454,
      "step": 15821
    },
    {
      "epoch": 0.8859894725053198,
      "grad_norm": 18.59945297241211,
      "learning_rate": 7.907e-05,
      "loss": 0.472,
      "step": 15822
    },
    {
      "epoch": 0.8860454698174488,
      "grad_norm": 1.55799400806427,
      "learning_rate": 7.9075e-05,
      "loss": 0.514,
      "step": 15823
    },
    {
      "epoch": 0.8861014671295778,
      "grad_norm": 1.2203803062438965,
      "learning_rate": 7.908e-05,
      "loss": 0.4197,
      "step": 15824
    },
    {
      "epoch": 0.8861574644417068,
      "grad_norm": 1.3588980436325073,
      "learning_rate": 7.9085e-05,
      "loss": 0.3994,
      "step": 15825
    },
    {
      "epoch": 0.8862134617538359,
      "grad_norm": 1.366632342338562,
      "learning_rate": 7.909e-05,
      "loss": 0.4418,
      "step": 15826
    },
    {
      "epoch": 0.8862694590659649,
      "grad_norm": 1.4658832550048828,
      "learning_rate": 7.909500000000001e-05,
      "loss": 0.5261,
      "step": 15827
    },
    {
      "epoch": 0.8863254563780939,
      "grad_norm": 1.6701146364212036,
      "learning_rate": 7.910000000000001e-05,
      "loss": 0.4231,
      "step": 15828
    },
    {
      "epoch": 0.8863814536902229,
      "grad_norm": 1.3779981136322021,
      "learning_rate": 7.910500000000001e-05,
      "loss": 0.5005,
      "step": 15829
    },
    {
      "epoch": 0.8864374510023519,
      "grad_norm": 1.3617491722106934,
      "learning_rate": 7.911e-05,
      "loss": 0.4605,
      "step": 15830
    },
    {
      "epoch": 0.886493448314481,
      "grad_norm": 1.4876869916915894,
      "learning_rate": 7.9115e-05,
      "loss": 0.5499,
      "step": 15831
    },
    {
      "epoch": 0.88654944562661,
      "grad_norm": 1.2605429887771606,
      "learning_rate": 7.912e-05,
      "loss": 0.3691,
      "step": 15832
    },
    {
      "epoch": 0.8866054429387389,
      "grad_norm": 1.374754548072815,
      "learning_rate": 7.9125e-05,
      "loss": 0.395,
      "step": 15833
    },
    {
      "epoch": 0.8866614402508679,
      "grad_norm": 1.4355651140213013,
      "learning_rate": 7.913000000000001e-05,
      "loss": 0.4694,
      "step": 15834
    },
    {
      "epoch": 0.8867174375629969,
      "grad_norm": 2.087085723876953,
      "learning_rate": 7.913500000000001e-05,
      "loss": 0.4419,
      "step": 15835
    },
    {
      "epoch": 0.8867734348751259,
      "grad_norm": 1.4105737209320068,
      "learning_rate": 7.914e-05,
      "loss": 0.4611,
      "step": 15836
    },
    {
      "epoch": 0.886829432187255,
      "grad_norm": 1.5528287887573242,
      "learning_rate": 7.9145e-05,
      "loss": 0.4813,
      "step": 15837
    },
    {
      "epoch": 0.886885429499384,
      "grad_norm": 1.5225160121917725,
      "learning_rate": 7.915e-05,
      "loss": 0.3797,
      "step": 15838
    },
    {
      "epoch": 0.886941426811513,
      "grad_norm": 1.4016565084457397,
      "learning_rate": 7.9155e-05,
      "loss": 0.436,
      "step": 15839
    },
    {
      "epoch": 0.886997424123642,
      "grad_norm": 1.6190400123596191,
      "learning_rate": 7.916e-05,
      "loss": 0.6496,
      "step": 15840
    },
    {
      "epoch": 0.887053421435771,
      "grad_norm": 1.536169171333313,
      "learning_rate": 7.9165e-05,
      "loss": 0.491,
      "step": 15841
    },
    {
      "epoch": 0.8871094187479001,
      "grad_norm": 1.1836466789245605,
      "learning_rate": 7.917e-05,
      "loss": 0.3229,
      "step": 15842
    },
    {
      "epoch": 0.8871654160600291,
      "grad_norm": 1.2826460599899292,
      "learning_rate": 7.9175e-05,
      "loss": 0.3797,
      "step": 15843
    },
    {
      "epoch": 0.8872214133721581,
      "grad_norm": 1.3410890102386475,
      "learning_rate": 7.918e-05,
      "loss": 0.4594,
      "step": 15844
    },
    {
      "epoch": 0.8872774106842871,
      "grad_norm": 1.5151267051696777,
      "learning_rate": 7.918500000000001e-05,
      "loss": 0.6094,
      "step": 15845
    },
    {
      "epoch": 0.8873334079964161,
      "grad_norm": 1.3532662391662598,
      "learning_rate": 7.919e-05,
      "loss": 0.552,
      "step": 15846
    },
    {
      "epoch": 0.8873894053085452,
      "grad_norm": 1.1911065578460693,
      "learning_rate": 7.9195e-05,
      "loss": 0.3841,
      "step": 15847
    },
    {
      "epoch": 0.8874454026206742,
      "grad_norm": 1.689738392829895,
      "learning_rate": 7.920000000000001e-05,
      "loss": 0.4415,
      "step": 15848
    },
    {
      "epoch": 0.8875013999328032,
      "grad_norm": 1.3184703588485718,
      "learning_rate": 7.920500000000001e-05,
      "loss": 0.4592,
      "step": 15849
    },
    {
      "epoch": 0.8875573972449322,
      "grad_norm": 1.3035863637924194,
      "learning_rate": 7.921000000000001e-05,
      "loss": 0.3915,
      "step": 15850
    },
    {
      "epoch": 0.8876133945570612,
      "grad_norm": 1.5418294668197632,
      "learning_rate": 7.9215e-05,
      "loss": 0.4651,
      "step": 15851
    },
    {
      "epoch": 0.8876693918691902,
      "grad_norm": 1.2926342487335205,
      "learning_rate": 7.922e-05,
      "loss": 0.3957,
      "step": 15852
    },
    {
      "epoch": 0.8877253891813193,
      "grad_norm": 1.4467759132385254,
      "learning_rate": 7.9225e-05,
      "loss": 0.4977,
      "step": 15853
    },
    {
      "epoch": 0.8877813864934483,
      "grad_norm": 1.3778395652770996,
      "learning_rate": 7.923e-05,
      "loss": 0.4759,
      "step": 15854
    },
    {
      "epoch": 0.8878373838055773,
      "grad_norm": 1.3714171648025513,
      "learning_rate": 7.923500000000001e-05,
      "loss": 0.3592,
      "step": 15855
    },
    {
      "epoch": 0.8878933811177063,
      "grad_norm": 1.3562978506088257,
      "learning_rate": 7.924000000000001e-05,
      "loss": 0.4602,
      "step": 15856
    },
    {
      "epoch": 0.8879493784298353,
      "grad_norm": 1.3002963066101074,
      "learning_rate": 7.9245e-05,
      "loss": 0.4336,
      "step": 15857
    },
    {
      "epoch": 0.8880053757419644,
      "grad_norm": 1.0840893983840942,
      "learning_rate": 7.925e-05,
      "loss": 0.4134,
      "step": 15858
    },
    {
      "epoch": 0.8880613730540934,
      "grad_norm": 1.65232515335083,
      "learning_rate": 7.9255e-05,
      "loss": 0.6223,
      "step": 15859
    },
    {
      "epoch": 0.8881173703662224,
      "grad_norm": 1.4145983457565308,
      "learning_rate": 7.926e-05,
      "loss": 0.5579,
      "step": 15860
    },
    {
      "epoch": 0.8881733676783514,
      "grad_norm": 1.2491377592086792,
      "learning_rate": 7.9265e-05,
      "loss": 0.4983,
      "step": 15861
    },
    {
      "epoch": 0.8882293649904804,
      "grad_norm": 1.2925677299499512,
      "learning_rate": 7.927e-05,
      "loss": 0.4141,
      "step": 15862
    },
    {
      "epoch": 0.8882853623026095,
      "grad_norm": 1.2815871238708496,
      "learning_rate": 7.9275e-05,
      "loss": 0.3611,
      "step": 15863
    },
    {
      "epoch": 0.8883413596147385,
      "grad_norm": 1.209580421447754,
      "learning_rate": 7.928e-05,
      "loss": 0.3394,
      "step": 15864
    },
    {
      "epoch": 0.8883973569268675,
      "grad_norm": 1.3125176429748535,
      "learning_rate": 7.928500000000001e-05,
      "loss": 0.4889,
      "step": 15865
    },
    {
      "epoch": 0.8884533542389965,
      "grad_norm": 1.1995253562927246,
      "learning_rate": 7.929000000000001e-05,
      "loss": 0.4111,
      "step": 15866
    },
    {
      "epoch": 0.8885093515511255,
      "grad_norm": 1.5066097974777222,
      "learning_rate": 7.9295e-05,
      "loss": 0.3786,
      "step": 15867
    },
    {
      "epoch": 0.8885653488632546,
      "grad_norm": 1.250152349472046,
      "learning_rate": 7.93e-05,
      "loss": 0.3288,
      "step": 15868
    },
    {
      "epoch": 0.8886213461753836,
      "grad_norm": 1.4152107238769531,
      "learning_rate": 7.9305e-05,
      "loss": 0.5992,
      "step": 15869
    },
    {
      "epoch": 0.8886773434875126,
      "grad_norm": 2.087266683578491,
      "learning_rate": 7.931000000000001e-05,
      "loss": 0.4672,
      "step": 15870
    },
    {
      "epoch": 0.8887333407996416,
      "grad_norm": 1.2819197177886963,
      "learning_rate": 7.931500000000001e-05,
      "loss": 0.3629,
      "step": 15871
    },
    {
      "epoch": 0.8887893381117706,
      "grad_norm": 1.2254847288131714,
      "learning_rate": 7.932e-05,
      "loss": 0.4562,
      "step": 15872
    },
    {
      "epoch": 0.8888453354238997,
      "grad_norm": 1.9349656105041504,
      "learning_rate": 7.9325e-05,
      "loss": 0.7662,
      "step": 15873
    },
    {
      "epoch": 0.8889013327360287,
      "grad_norm": 3.4781572818756104,
      "learning_rate": 7.933e-05,
      "loss": 0.4305,
      "step": 15874
    },
    {
      "epoch": 0.8889573300481577,
      "grad_norm": 1.4061557054519653,
      "learning_rate": 7.9335e-05,
      "loss": 0.5388,
      "step": 15875
    },
    {
      "epoch": 0.8890133273602867,
      "grad_norm": 1.6819357872009277,
      "learning_rate": 7.934000000000001e-05,
      "loss": 0.5555,
      "step": 15876
    },
    {
      "epoch": 0.8890693246724157,
      "grad_norm": 1.3523751497268677,
      "learning_rate": 7.934500000000001e-05,
      "loss": 0.3901,
      "step": 15877
    },
    {
      "epoch": 0.8891253219845447,
      "grad_norm": 1.488936185836792,
      "learning_rate": 7.935e-05,
      "loss": 0.6156,
      "step": 15878
    },
    {
      "epoch": 0.8891813192966738,
      "grad_norm": 1.2007721662521362,
      "learning_rate": 7.9355e-05,
      "loss": 0.3833,
      "step": 15879
    },
    {
      "epoch": 0.8892373166088028,
      "grad_norm": 1.336026668548584,
      "learning_rate": 7.936e-05,
      "loss": 0.5105,
      "step": 15880
    },
    {
      "epoch": 0.8892933139209318,
      "grad_norm": 1.1636326313018799,
      "learning_rate": 7.9365e-05,
      "loss": 0.4252,
      "step": 15881
    },
    {
      "epoch": 0.8893493112330608,
      "grad_norm": 1.8537273406982422,
      "learning_rate": 7.937e-05,
      "loss": 0.4205,
      "step": 15882
    },
    {
      "epoch": 0.8894053085451898,
      "grad_norm": 1.2801110744476318,
      "learning_rate": 7.9375e-05,
      "loss": 0.6122,
      "step": 15883
    },
    {
      "epoch": 0.8894613058573189,
      "grad_norm": 1.5170996189117432,
      "learning_rate": 7.938e-05,
      "loss": 0.439,
      "step": 15884
    },
    {
      "epoch": 0.8895173031694479,
      "grad_norm": 1.350251317024231,
      "learning_rate": 7.9385e-05,
      "loss": 0.4352,
      "step": 15885
    },
    {
      "epoch": 0.8895733004815769,
      "grad_norm": 1.1521936655044556,
      "learning_rate": 7.939000000000001e-05,
      "loss": 0.4359,
      "step": 15886
    },
    {
      "epoch": 0.8896292977937059,
      "grad_norm": 1.2937862873077393,
      "learning_rate": 7.939500000000001e-05,
      "loss": 0.3774,
      "step": 15887
    },
    {
      "epoch": 0.8896852951058349,
      "grad_norm": 1.3156973123550415,
      "learning_rate": 7.94e-05,
      "loss": 0.4469,
      "step": 15888
    },
    {
      "epoch": 0.889741292417964,
      "grad_norm": 1.879453420639038,
      "learning_rate": 7.9405e-05,
      "loss": 0.5854,
      "step": 15889
    },
    {
      "epoch": 0.889797289730093,
      "grad_norm": 18.626432418823242,
      "learning_rate": 7.941e-05,
      "loss": 0.4712,
      "step": 15890
    },
    {
      "epoch": 0.889853287042222,
      "grad_norm": 1.63935387134552,
      "learning_rate": 7.941500000000001e-05,
      "loss": 0.435,
      "step": 15891
    },
    {
      "epoch": 0.889909284354351,
      "grad_norm": 1.7789877653121948,
      "learning_rate": 7.942000000000001e-05,
      "loss": 0.4656,
      "step": 15892
    },
    {
      "epoch": 0.88996528166648,
      "grad_norm": 1.30344557762146,
      "learning_rate": 7.9425e-05,
      "loss": 0.4466,
      "step": 15893
    },
    {
      "epoch": 0.890021278978609,
      "grad_norm": 1.3301328420639038,
      "learning_rate": 7.943e-05,
      "loss": 0.4277,
      "step": 15894
    },
    {
      "epoch": 0.8900772762907381,
      "grad_norm": 1.4361047744750977,
      "learning_rate": 7.9435e-05,
      "loss": 0.4576,
      "step": 15895
    },
    {
      "epoch": 0.8901332736028671,
      "grad_norm": 1.6864228248596191,
      "learning_rate": 7.944e-05,
      "loss": 0.3797,
      "step": 15896
    },
    {
      "epoch": 0.8901892709149961,
      "grad_norm": 1.6416587829589844,
      "learning_rate": 7.944500000000001e-05,
      "loss": 0.4188,
      "step": 15897
    },
    {
      "epoch": 0.8902452682271251,
      "grad_norm": 1.5706303119659424,
      "learning_rate": 7.945e-05,
      "loss": 0.4691,
      "step": 15898
    },
    {
      "epoch": 0.8903012655392541,
      "grad_norm": 1.7088900804519653,
      "learning_rate": 7.9455e-05,
      "loss": 0.6489,
      "step": 15899
    },
    {
      "epoch": 0.8903572628513832,
      "grad_norm": 1.3953651189804077,
      "learning_rate": 7.946e-05,
      "loss": 0.4973,
      "step": 15900
    },
    {
      "epoch": 0.8904132601635122,
      "grad_norm": 1.3055979013442993,
      "learning_rate": 7.9465e-05,
      "loss": 0.3984,
      "step": 15901
    },
    {
      "epoch": 0.8904692574756412,
      "grad_norm": 1.4346344470977783,
      "learning_rate": 7.947e-05,
      "loss": 0.5526,
      "step": 15902
    },
    {
      "epoch": 0.8905252547877702,
      "grad_norm": 1.500628113746643,
      "learning_rate": 7.9475e-05,
      "loss": 0.4486,
      "step": 15903
    },
    {
      "epoch": 0.8905812520998992,
      "grad_norm": 1.1303764581680298,
      "learning_rate": 7.948e-05,
      "loss": 0.3409,
      "step": 15904
    },
    {
      "epoch": 0.8906372494120283,
      "grad_norm": 1.253481149673462,
      "learning_rate": 7.9485e-05,
      "loss": 0.4313,
      "step": 15905
    },
    {
      "epoch": 0.8906932467241573,
      "grad_norm": 1.2080345153808594,
      "learning_rate": 7.949000000000001e-05,
      "loss": 0.4027,
      "step": 15906
    },
    {
      "epoch": 0.8907492440362863,
      "grad_norm": 1.382400393486023,
      "learning_rate": 7.949500000000001e-05,
      "loss": 0.4287,
      "step": 15907
    },
    {
      "epoch": 0.8908052413484153,
      "grad_norm": 1.2246397733688354,
      "learning_rate": 7.950000000000001e-05,
      "loss": 0.4408,
      "step": 15908
    },
    {
      "epoch": 0.8908612386605443,
      "grad_norm": 1.305453896522522,
      "learning_rate": 7.9505e-05,
      "loss": 0.4523,
      "step": 15909
    },
    {
      "epoch": 0.8909172359726734,
      "grad_norm": 1.3066540956497192,
      "learning_rate": 7.951e-05,
      "loss": 0.3914,
      "step": 15910
    },
    {
      "epoch": 0.8909732332848024,
      "grad_norm": 1.8098177909851074,
      "learning_rate": 7.9515e-05,
      "loss": 0.5402,
      "step": 15911
    },
    {
      "epoch": 0.8910292305969314,
      "grad_norm": 1.3305491209030151,
      "learning_rate": 7.952000000000001e-05,
      "loss": 0.4655,
      "step": 15912
    },
    {
      "epoch": 0.8910852279090604,
      "grad_norm": 1.63957941532135,
      "learning_rate": 7.952500000000001e-05,
      "loss": 0.3788,
      "step": 15913
    },
    {
      "epoch": 0.8911412252211894,
      "grad_norm": 1.2962210178375244,
      "learning_rate": 7.953e-05,
      "loss": 0.5002,
      "step": 15914
    },
    {
      "epoch": 0.8911972225333185,
      "grad_norm": 1.2458339929580688,
      "learning_rate": 7.9535e-05,
      "loss": 0.3197,
      "step": 15915
    },
    {
      "epoch": 0.8912532198454474,
      "grad_norm": 1.050899863243103,
      "learning_rate": 7.954e-05,
      "loss": 0.3315,
      "step": 15916
    },
    {
      "epoch": 0.8913092171575764,
      "grad_norm": 1.336843490600586,
      "learning_rate": 7.9545e-05,
      "loss": 0.4845,
      "step": 15917
    },
    {
      "epoch": 0.8913652144697054,
      "grad_norm": 1.130553126335144,
      "learning_rate": 7.955e-05,
      "loss": 0.3586,
      "step": 15918
    },
    {
      "epoch": 0.8914212117818344,
      "grad_norm": 1.126340389251709,
      "learning_rate": 7.9555e-05,
      "loss": 0.3741,
      "step": 15919
    },
    {
      "epoch": 0.8914772090939634,
      "grad_norm": 1.239076018333435,
      "learning_rate": 7.956e-05,
      "loss": 0.4854,
      "step": 15920
    },
    {
      "epoch": 0.8915332064060925,
      "grad_norm": 29.588964462280273,
      "learning_rate": 7.9565e-05,
      "loss": 0.4257,
      "step": 15921
    },
    {
      "epoch": 0.8915892037182215,
      "grad_norm": 1.1070128679275513,
      "learning_rate": 7.957e-05,
      "loss": 0.3545,
      "step": 15922
    },
    {
      "epoch": 0.8916452010303505,
      "grad_norm": 1.6583384275436401,
      "learning_rate": 7.9575e-05,
      "loss": 0.4977,
      "step": 15923
    },
    {
      "epoch": 0.8917011983424795,
      "grad_norm": 1.3087083101272583,
      "learning_rate": 7.958e-05,
      "loss": 0.5173,
      "step": 15924
    },
    {
      "epoch": 0.8917571956546085,
      "grad_norm": 1.3332713842391968,
      "learning_rate": 7.9585e-05,
      "loss": 0.5326,
      "step": 15925
    },
    {
      "epoch": 0.8918131929667376,
      "grad_norm": 1.3929810523986816,
      "learning_rate": 7.959000000000002e-05,
      "loss": 0.4613,
      "step": 15926
    },
    {
      "epoch": 0.8918691902788666,
      "grad_norm": 1.5982316732406616,
      "learning_rate": 7.959500000000001e-05,
      "loss": 0.4301,
      "step": 15927
    },
    {
      "epoch": 0.8919251875909956,
      "grad_norm": 1.7531770467758179,
      "learning_rate": 7.960000000000001e-05,
      "loss": 0.4654,
      "step": 15928
    },
    {
      "epoch": 0.8919811849031246,
      "grad_norm": 1.43354070186615,
      "learning_rate": 7.960500000000001e-05,
      "loss": 0.4953,
      "step": 15929
    },
    {
      "epoch": 0.8920371822152536,
      "grad_norm": 1.5353807210922241,
      "learning_rate": 7.961e-05,
      "loss": 0.4647,
      "step": 15930
    },
    {
      "epoch": 0.8920931795273827,
      "grad_norm": 1.599859356880188,
      "learning_rate": 7.9615e-05,
      "loss": 0.4708,
      "step": 15931
    },
    {
      "epoch": 0.8921491768395117,
      "grad_norm": 1.3043773174285889,
      "learning_rate": 7.962e-05,
      "loss": 0.4345,
      "step": 15932
    },
    {
      "epoch": 0.8922051741516407,
      "grad_norm": 1.63169527053833,
      "learning_rate": 7.962500000000001e-05,
      "loss": 0.4663,
      "step": 15933
    },
    {
      "epoch": 0.8922611714637697,
      "grad_norm": 2.183722734451294,
      "learning_rate": 7.963000000000001e-05,
      "loss": 0.4604,
      "step": 15934
    },
    {
      "epoch": 0.8923171687758987,
      "grad_norm": 1.592522144317627,
      "learning_rate": 7.9635e-05,
      "loss": 0.5142,
      "step": 15935
    },
    {
      "epoch": 0.8923731660880277,
      "grad_norm": 1.4822566509246826,
      "learning_rate": 7.964e-05,
      "loss": 0.5215,
      "step": 15936
    },
    {
      "epoch": 0.8924291634001568,
      "grad_norm": 1.3690813779830933,
      "learning_rate": 7.9645e-05,
      "loss": 0.5041,
      "step": 15937
    },
    {
      "epoch": 0.8924851607122858,
      "grad_norm": 1.424587607383728,
      "learning_rate": 7.965e-05,
      "loss": 0.4089,
      "step": 15938
    },
    {
      "epoch": 0.8925411580244148,
      "grad_norm": 1.2622556686401367,
      "learning_rate": 7.9655e-05,
      "loss": 0.3578,
      "step": 15939
    },
    {
      "epoch": 0.8925971553365438,
      "grad_norm": 1.181199550628662,
      "learning_rate": 7.966e-05,
      "loss": 0.3952,
      "step": 15940
    },
    {
      "epoch": 0.8926531526486728,
      "grad_norm": 1.1826465129852295,
      "learning_rate": 7.9665e-05,
      "loss": 0.4911,
      "step": 15941
    },
    {
      "epoch": 0.8927091499608019,
      "grad_norm": 1.2569098472595215,
      "learning_rate": 7.967e-05,
      "loss": 0.4143,
      "step": 15942
    },
    {
      "epoch": 0.8927651472729309,
      "grad_norm": 1.435709834098816,
      "learning_rate": 7.9675e-05,
      "loss": 0.4349,
      "step": 15943
    },
    {
      "epoch": 0.8928211445850599,
      "grad_norm": 1.2740654945373535,
      "learning_rate": 7.968e-05,
      "loss": 0.425,
      "step": 15944
    },
    {
      "epoch": 0.8928771418971889,
      "grad_norm": 1.3047778606414795,
      "learning_rate": 7.9685e-05,
      "loss": 0.3786,
      "step": 15945
    },
    {
      "epoch": 0.8929331392093179,
      "grad_norm": 1.2718031406402588,
      "learning_rate": 7.969e-05,
      "loss": 0.3694,
      "step": 15946
    },
    {
      "epoch": 0.892989136521447,
      "grad_norm": 1.4117116928100586,
      "learning_rate": 7.9695e-05,
      "loss": 0.5094,
      "step": 15947
    },
    {
      "epoch": 0.893045133833576,
      "grad_norm": 1.591858983039856,
      "learning_rate": 7.970000000000001e-05,
      "loss": 0.5699,
      "step": 15948
    },
    {
      "epoch": 0.893101131145705,
      "grad_norm": 1.3295998573303223,
      "learning_rate": 7.970500000000001e-05,
      "loss": 0.3798,
      "step": 15949
    },
    {
      "epoch": 0.893157128457834,
      "grad_norm": 1.3497673273086548,
      "learning_rate": 7.971000000000001e-05,
      "loss": 0.3794,
      "step": 15950
    },
    {
      "epoch": 0.893213125769963,
      "grad_norm": 1.523391604423523,
      "learning_rate": 7.9715e-05,
      "loss": 0.5808,
      "step": 15951
    },
    {
      "epoch": 0.893269123082092,
      "grad_norm": 1.3123940229415894,
      "learning_rate": 7.972e-05,
      "loss": 0.4704,
      "step": 15952
    },
    {
      "epoch": 0.8933251203942211,
      "grad_norm": 1.4037657976150513,
      "learning_rate": 7.9725e-05,
      "loss": 0.4478,
      "step": 15953
    },
    {
      "epoch": 0.8933811177063501,
      "grad_norm": 21.667104721069336,
      "learning_rate": 7.973000000000001e-05,
      "loss": 0.3895,
      "step": 15954
    },
    {
      "epoch": 0.8934371150184791,
      "grad_norm": 1.4874041080474854,
      "learning_rate": 7.973500000000001e-05,
      "loss": 0.4544,
      "step": 15955
    },
    {
      "epoch": 0.8934931123306081,
      "grad_norm": 1.2745970487594604,
      "learning_rate": 7.974e-05,
      "loss": 0.3299,
      "step": 15956
    },
    {
      "epoch": 0.8935491096427371,
      "grad_norm": 1.364397644996643,
      "learning_rate": 7.9745e-05,
      "loss": 0.5139,
      "step": 15957
    },
    {
      "epoch": 0.8936051069548662,
      "grad_norm": 1.5993179082870483,
      "learning_rate": 7.975e-05,
      "loss": 0.4982,
      "step": 15958
    },
    {
      "epoch": 0.8936611042669952,
      "grad_norm": 1.1834697723388672,
      "learning_rate": 7.9755e-05,
      "loss": 0.3477,
      "step": 15959
    },
    {
      "epoch": 0.8937171015791242,
      "grad_norm": 1.264442801475525,
      "learning_rate": 7.976e-05,
      "loss": 0.3819,
      "step": 15960
    },
    {
      "epoch": 0.8937730988912532,
      "grad_norm": 1.2564032077789307,
      "learning_rate": 7.9765e-05,
      "loss": 0.3424,
      "step": 15961
    },
    {
      "epoch": 0.8938290962033822,
      "grad_norm": 1.2035491466522217,
      "learning_rate": 7.977e-05,
      "loss": 0.361,
      "step": 15962
    },
    {
      "epoch": 0.8938850935155113,
      "grad_norm": 1.5151845216751099,
      "learning_rate": 7.9775e-05,
      "loss": 0.3303,
      "step": 15963
    },
    {
      "epoch": 0.8939410908276403,
      "grad_norm": 1.3209103345870972,
      "learning_rate": 7.978e-05,
      "loss": 0.468,
      "step": 15964
    },
    {
      "epoch": 0.8939970881397693,
      "grad_norm": 1.5421687364578247,
      "learning_rate": 7.9785e-05,
      "loss": 0.5283,
      "step": 15965
    },
    {
      "epoch": 0.8940530854518983,
      "grad_norm": 1.4798215627670288,
      "learning_rate": 7.979000000000001e-05,
      "loss": 0.4598,
      "step": 15966
    },
    {
      "epoch": 0.8941090827640273,
      "grad_norm": 1.294493556022644,
      "learning_rate": 7.9795e-05,
      "loss": 0.5248,
      "step": 15967
    },
    {
      "epoch": 0.8941650800761564,
      "grad_norm": 1.600492000579834,
      "learning_rate": 7.98e-05,
      "loss": 0.5206,
      "step": 15968
    },
    {
      "epoch": 0.8942210773882854,
      "grad_norm": 1.415737509727478,
      "learning_rate": 7.980500000000001e-05,
      "loss": 0.4547,
      "step": 15969
    },
    {
      "epoch": 0.8942770747004144,
      "grad_norm": 1.2604701519012451,
      "learning_rate": 7.981000000000001e-05,
      "loss": 0.4808,
      "step": 15970
    },
    {
      "epoch": 0.8943330720125434,
      "grad_norm": 1.2768656015396118,
      "learning_rate": 7.981500000000001e-05,
      "loss": 0.3808,
      "step": 15971
    },
    {
      "epoch": 0.8943890693246724,
      "grad_norm": 1.1699621677398682,
      "learning_rate": 7.982e-05,
      "loss": 0.366,
      "step": 15972
    },
    {
      "epoch": 0.8944450666368015,
      "grad_norm": 1.4129713773727417,
      "learning_rate": 7.9825e-05,
      "loss": 0.4452,
      "step": 15973
    },
    {
      "epoch": 0.8945010639489305,
      "grad_norm": 1.3117674589157104,
      "learning_rate": 7.983e-05,
      "loss": 0.4753,
      "step": 15974
    },
    {
      "epoch": 0.8945570612610595,
      "grad_norm": 1.3465110063552856,
      "learning_rate": 7.983500000000001e-05,
      "loss": 0.468,
      "step": 15975
    },
    {
      "epoch": 0.8946130585731885,
      "grad_norm": 1.24776291847229,
      "learning_rate": 7.984000000000001e-05,
      "loss": 0.437,
      "step": 15976
    },
    {
      "epoch": 0.8946690558853175,
      "grad_norm": 1.4752442836761475,
      "learning_rate": 7.9845e-05,
      "loss": 0.3954,
      "step": 15977
    },
    {
      "epoch": 0.8947250531974466,
      "grad_norm": 1.5850889682769775,
      "learning_rate": 7.985e-05,
      "loss": 0.4348,
      "step": 15978
    },
    {
      "epoch": 0.8947810505095756,
      "grad_norm": 1.7630277872085571,
      "learning_rate": 7.9855e-05,
      "loss": 0.3952,
      "step": 15979
    },
    {
      "epoch": 0.8948370478217046,
      "grad_norm": 1.4778448343276978,
      "learning_rate": 7.986e-05,
      "loss": 0.6347,
      "step": 15980
    },
    {
      "epoch": 0.8948930451338336,
      "grad_norm": 1.1664304733276367,
      "learning_rate": 7.9865e-05,
      "loss": 0.3586,
      "step": 15981
    },
    {
      "epoch": 0.8949490424459626,
      "grad_norm": 1.3973556756973267,
      "learning_rate": 7.987e-05,
      "loss": 0.4527,
      "step": 15982
    },
    {
      "epoch": 0.8950050397580916,
      "grad_norm": 1.5247793197631836,
      "learning_rate": 7.9875e-05,
      "loss": 0.4154,
      "step": 15983
    },
    {
      "epoch": 0.8950610370702207,
      "grad_norm": 1.149987816810608,
      "learning_rate": 7.988e-05,
      "loss": 0.3322,
      "step": 15984
    },
    {
      "epoch": 0.8951170343823497,
      "grad_norm": 1.3923696279525757,
      "learning_rate": 7.9885e-05,
      "loss": 0.4424,
      "step": 15985
    },
    {
      "epoch": 0.8951730316944787,
      "grad_norm": 1.550440788269043,
      "learning_rate": 7.989000000000001e-05,
      "loss": 0.3841,
      "step": 15986
    },
    {
      "epoch": 0.8952290290066077,
      "grad_norm": 1.2467762231826782,
      "learning_rate": 7.9895e-05,
      "loss": 0.5213,
      "step": 15987
    },
    {
      "epoch": 0.8952850263187367,
      "grad_norm": 1.3135510683059692,
      "learning_rate": 7.99e-05,
      "loss": 0.4367,
      "step": 15988
    },
    {
      "epoch": 0.8953410236308658,
      "grad_norm": 1.4266338348388672,
      "learning_rate": 7.9905e-05,
      "loss": 0.5519,
      "step": 15989
    },
    {
      "epoch": 0.8953970209429948,
      "grad_norm": 1.6020103693008423,
      "learning_rate": 7.991000000000001e-05,
      "loss": 0.4317,
      "step": 15990
    },
    {
      "epoch": 0.8954530182551238,
      "grad_norm": 1.1448312997817993,
      "learning_rate": 7.991500000000001e-05,
      "loss": 0.3692,
      "step": 15991
    },
    {
      "epoch": 0.8955090155672528,
      "grad_norm": 1.4078099727630615,
      "learning_rate": 7.992000000000001e-05,
      "loss": 0.5585,
      "step": 15992
    },
    {
      "epoch": 0.8955650128793818,
      "grad_norm": 1.2579874992370605,
      "learning_rate": 7.9925e-05,
      "loss": 0.4945,
      "step": 15993
    },
    {
      "epoch": 0.8956210101915109,
      "grad_norm": 1.5178617238998413,
      "learning_rate": 7.993e-05,
      "loss": 0.524,
      "step": 15994
    },
    {
      "epoch": 0.8956770075036399,
      "grad_norm": 1.4215368032455444,
      "learning_rate": 7.9935e-05,
      "loss": 0.4524,
      "step": 15995
    },
    {
      "epoch": 0.8957330048157689,
      "grad_norm": 1.0902271270751953,
      "learning_rate": 7.994000000000001e-05,
      "loss": 0.3212,
      "step": 15996
    },
    {
      "epoch": 0.8957890021278979,
      "grad_norm": 1.4094494581222534,
      "learning_rate": 7.994500000000001e-05,
      "loss": 0.4249,
      "step": 15997
    },
    {
      "epoch": 0.8958449994400268,
      "grad_norm": 1.415575623512268,
      "learning_rate": 7.995e-05,
      "loss": 0.4705,
      "step": 15998
    },
    {
      "epoch": 0.8959009967521558,
      "grad_norm": 1.1373982429504395,
      "learning_rate": 7.9955e-05,
      "loss": 0.3789,
      "step": 15999
    },
    {
      "epoch": 0.8959569940642849,
      "grad_norm": 1.6810451745986938,
      "learning_rate": 7.996e-05,
      "loss": 0.7872,
      "step": 16000
    },
    {
      "epoch": 0.8960129913764139,
      "grad_norm": 1.4576246738433838,
      "learning_rate": 7.9965e-05,
      "loss": 0.361,
      "step": 16001
    },
    {
      "epoch": 0.8960689886885429,
      "grad_norm": 3.6870405673980713,
      "learning_rate": 7.997e-05,
      "loss": 0.4053,
      "step": 16002
    },
    {
      "epoch": 0.8961249860006719,
      "grad_norm": 1.5046871900558472,
      "learning_rate": 7.9975e-05,
      "loss": 0.4212,
      "step": 16003
    },
    {
      "epoch": 0.8961809833128009,
      "grad_norm": 1.788367509841919,
      "learning_rate": 7.998e-05,
      "loss": 0.5489,
      "step": 16004
    },
    {
      "epoch": 0.89623698062493,
      "grad_norm": 1.2205194234848022,
      "learning_rate": 7.9985e-05,
      "loss": 0.3749,
      "step": 16005
    },
    {
      "epoch": 0.896292977937059,
      "grad_norm": 1.5499311685562134,
      "learning_rate": 7.999000000000001e-05,
      "loss": 0.4994,
      "step": 16006
    },
    {
      "epoch": 0.896348975249188,
      "grad_norm": 1.4454174041748047,
      "learning_rate": 7.999500000000001e-05,
      "loss": 0.588,
      "step": 16007
    },
    {
      "epoch": 0.896404972561317,
      "grad_norm": 1.3178184032440186,
      "learning_rate": 8e-05,
      "loss": 0.4279,
      "step": 16008
    },
    {
      "epoch": 0.896460969873446,
      "grad_norm": 1.215785264968872,
      "learning_rate": 8.0005e-05,
      "loss": 0.5083,
      "step": 16009
    },
    {
      "epoch": 0.896516967185575,
      "grad_norm": 1.4268912076950073,
      "learning_rate": 8.001e-05,
      "loss": 0.4572,
      "step": 16010
    },
    {
      "epoch": 0.8965729644977041,
      "grad_norm": 1.2209932804107666,
      "learning_rate": 8.001500000000001e-05,
      "loss": 0.3982,
      "step": 16011
    },
    {
      "epoch": 0.8966289618098331,
      "grad_norm": 1.4563535451889038,
      "learning_rate": 8.002000000000001e-05,
      "loss": 0.4739,
      "step": 16012
    },
    {
      "epoch": 0.8966849591219621,
      "grad_norm": 1.3663113117218018,
      "learning_rate": 8.002500000000001e-05,
      "loss": 0.4169,
      "step": 16013
    },
    {
      "epoch": 0.8967409564340911,
      "grad_norm": 1.5242372751235962,
      "learning_rate": 8.003e-05,
      "loss": 0.4739,
      "step": 16014
    },
    {
      "epoch": 0.8967969537462201,
      "grad_norm": 1.2546685934066772,
      "learning_rate": 8.0035e-05,
      "loss": 0.423,
      "step": 16015
    },
    {
      "epoch": 0.8968529510583492,
      "grad_norm": 1.2493762969970703,
      "learning_rate": 8.004e-05,
      "loss": 0.4325,
      "step": 16016
    },
    {
      "epoch": 0.8969089483704782,
      "grad_norm": 1.390856146812439,
      "learning_rate": 8.0045e-05,
      "loss": 0.5716,
      "step": 16017
    },
    {
      "epoch": 0.8969649456826072,
      "grad_norm": 1.460411787033081,
      "learning_rate": 8.005000000000001e-05,
      "loss": 0.4483,
      "step": 16018
    },
    {
      "epoch": 0.8970209429947362,
      "grad_norm": 1.5322096347808838,
      "learning_rate": 8.0055e-05,
      "loss": 0.4411,
      "step": 16019
    },
    {
      "epoch": 0.8970769403068652,
      "grad_norm": 1.2709147930145264,
      "learning_rate": 8.006e-05,
      "loss": 0.5024,
      "step": 16020
    },
    {
      "epoch": 0.8971329376189943,
      "grad_norm": 1.3853907585144043,
      "learning_rate": 8.0065e-05,
      "loss": 0.4838,
      "step": 16021
    },
    {
      "epoch": 0.8971889349311233,
      "grad_norm": 1.5160138607025146,
      "learning_rate": 8.007e-05,
      "loss": 0.4062,
      "step": 16022
    },
    {
      "epoch": 0.8972449322432523,
      "grad_norm": 1.3848271369934082,
      "learning_rate": 8.0075e-05,
      "loss": 0.4543,
      "step": 16023
    },
    {
      "epoch": 0.8973009295553813,
      "grad_norm": 1.159733772277832,
      "learning_rate": 8.008e-05,
      "loss": 0.4621,
      "step": 16024
    },
    {
      "epoch": 0.8973569268675103,
      "grad_norm": 1.551933765411377,
      "learning_rate": 8.0085e-05,
      "loss": 0.5854,
      "step": 16025
    },
    {
      "epoch": 0.8974129241796394,
      "grad_norm": 1.224943995475769,
      "learning_rate": 8.009e-05,
      "loss": 0.4938,
      "step": 16026
    },
    {
      "epoch": 0.8974689214917684,
      "grad_norm": 1.6137648820877075,
      "learning_rate": 8.009500000000001e-05,
      "loss": 0.5849,
      "step": 16027
    },
    {
      "epoch": 0.8975249188038974,
      "grad_norm": 1.6982605457305908,
      "learning_rate": 8.010000000000001e-05,
      "loss": 0.5129,
      "step": 16028
    },
    {
      "epoch": 0.8975809161160264,
      "grad_norm": 1.256736397743225,
      "learning_rate": 8.0105e-05,
      "loss": 0.5327,
      "step": 16029
    },
    {
      "epoch": 0.8976369134281554,
      "grad_norm": 1.533632755279541,
      "learning_rate": 8.011e-05,
      "loss": 0.5459,
      "step": 16030
    },
    {
      "epoch": 0.8976929107402845,
      "grad_norm": 1.3175910711288452,
      "learning_rate": 8.0115e-05,
      "loss": 0.5595,
      "step": 16031
    },
    {
      "epoch": 0.8977489080524135,
      "grad_norm": 1.1984654664993286,
      "learning_rate": 8.012000000000001e-05,
      "loss": 0.3818,
      "step": 16032
    },
    {
      "epoch": 0.8978049053645425,
      "grad_norm": 1.3063366413116455,
      "learning_rate": 8.012500000000001e-05,
      "loss": 0.3868,
      "step": 16033
    },
    {
      "epoch": 0.8978609026766715,
      "grad_norm": 1.4270581007003784,
      "learning_rate": 8.013000000000001e-05,
      "loss": 0.4259,
      "step": 16034
    },
    {
      "epoch": 0.8979168999888005,
      "grad_norm": 1.5793156623840332,
      "learning_rate": 8.0135e-05,
      "loss": 0.4375,
      "step": 16035
    },
    {
      "epoch": 0.8979728973009296,
      "grad_norm": 1.4425928592681885,
      "learning_rate": 8.014e-05,
      "loss": 0.5081,
      "step": 16036
    },
    {
      "epoch": 0.8980288946130586,
      "grad_norm": 1.380345344543457,
      "learning_rate": 8.0145e-05,
      "loss": 0.5136,
      "step": 16037
    },
    {
      "epoch": 0.8980848919251876,
      "grad_norm": 1.4106237888336182,
      "learning_rate": 8.015e-05,
      "loss": 0.4984,
      "step": 16038
    },
    {
      "epoch": 0.8981408892373166,
      "grad_norm": 1.3798160552978516,
      "learning_rate": 8.015500000000001e-05,
      "loss": 0.4104,
      "step": 16039
    },
    {
      "epoch": 0.8981968865494456,
      "grad_norm": 1.4829031229019165,
      "learning_rate": 8.016e-05,
      "loss": 0.4247,
      "step": 16040
    },
    {
      "epoch": 0.8982528838615746,
      "grad_norm": 1.0722377300262451,
      "learning_rate": 8.0165e-05,
      "loss": 0.3371,
      "step": 16041
    },
    {
      "epoch": 0.8983088811737037,
      "grad_norm": 1.2194339036941528,
      "learning_rate": 8.017e-05,
      "loss": 0.4685,
      "step": 16042
    },
    {
      "epoch": 0.8983648784858327,
      "grad_norm": 1.2947598695755005,
      "learning_rate": 8.0175e-05,
      "loss": 0.3746,
      "step": 16043
    },
    {
      "epoch": 0.8984208757979617,
      "grad_norm": 1.1339155435562134,
      "learning_rate": 8.018e-05,
      "loss": 0.4044,
      "step": 16044
    },
    {
      "epoch": 0.8984768731100907,
      "grad_norm": 1.5063127279281616,
      "learning_rate": 8.0185e-05,
      "loss": 0.4086,
      "step": 16045
    },
    {
      "epoch": 0.8985328704222197,
      "grad_norm": 1.4349595308303833,
      "learning_rate": 8.019e-05,
      "loss": 0.5806,
      "step": 16046
    },
    {
      "epoch": 0.8985888677343488,
      "grad_norm": 1.2671962976455688,
      "learning_rate": 8.019500000000001e-05,
      "loss": 0.3964,
      "step": 16047
    },
    {
      "epoch": 0.8986448650464778,
      "grad_norm": 2.124438524246216,
      "learning_rate": 8.020000000000001e-05,
      "loss": 0.4742,
      "step": 16048
    },
    {
      "epoch": 0.8987008623586068,
      "grad_norm": 1.3917243480682373,
      "learning_rate": 8.020500000000001e-05,
      "loss": 0.5335,
      "step": 16049
    },
    {
      "epoch": 0.8987568596707358,
      "grad_norm": 1.321627140045166,
      "learning_rate": 8.021e-05,
      "loss": 0.4422,
      "step": 16050
    },
    {
      "epoch": 0.8988128569828648,
      "grad_norm": 1.4380525350570679,
      "learning_rate": 8.0215e-05,
      "loss": 0.4308,
      "step": 16051
    },
    {
      "epoch": 0.8988688542949939,
      "grad_norm": 1.2992310523986816,
      "learning_rate": 8.022e-05,
      "loss": 0.4118,
      "step": 16052
    },
    {
      "epoch": 0.8989248516071229,
      "grad_norm": 1.675029993057251,
      "learning_rate": 8.022500000000001e-05,
      "loss": 0.6676,
      "step": 16053
    },
    {
      "epoch": 0.8989808489192519,
      "grad_norm": 1.4144927263259888,
      "learning_rate": 8.023000000000001e-05,
      "loss": 0.3992,
      "step": 16054
    },
    {
      "epoch": 0.8990368462313809,
      "grad_norm": 1.5401591062545776,
      "learning_rate": 8.023500000000001e-05,
      "loss": 0.4584,
      "step": 16055
    },
    {
      "epoch": 0.8990928435435099,
      "grad_norm": 1.4692164659500122,
      "learning_rate": 8.024e-05,
      "loss": 0.5731,
      "step": 16056
    },
    {
      "epoch": 0.899148840855639,
      "grad_norm": 1.1132906675338745,
      "learning_rate": 8.0245e-05,
      "loss": 0.3796,
      "step": 16057
    },
    {
      "epoch": 0.899204838167768,
      "grad_norm": 1.2593834400177002,
      "learning_rate": 8.025e-05,
      "loss": 0.4325,
      "step": 16058
    },
    {
      "epoch": 0.899260835479897,
      "grad_norm": 1.2930701971054077,
      "learning_rate": 8.0255e-05,
      "loss": 0.4755,
      "step": 16059
    },
    {
      "epoch": 0.899316832792026,
      "grad_norm": 1.2894867658615112,
      "learning_rate": 8.026000000000001e-05,
      "loss": 0.6381,
      "step": 16060
    },
    {
      "epoch": 0.899372830104155,
      "grad_norm": 1.3451348543167114,
      "learning_rate": 8.0265e-05,
      "loss": 0.4582,
      "step": 16061
    },
    {
      "epoch": 0.899428827416284,
      "grad_norm": 1.569953441619873,
      "learning_rate": 8.027e-05,
      "loss": 0.4933,
      "step": 16062
    },
    {
      "epoch": 0.8994848247284131,
      "grad_norm": 1.1205308437347412,
      "learning_rate": 8.0275e-05,
      "loss": 0.329,
      "step": 16063
    },
    {
      "epoch": 0.8995408220405421,
      "grad_norm": 1.2865349054336548,
      "learning_rate": 8.028e-05,
      "loss": 0.4729,
      "step": 16064
    },
    {
      "epoch": 0.8995968193526711,
      "grad_norm": 1.3408750295639038,
      "learning_rate": 8.0285e-05,
      "loss": 0.4201,
      "step": 16065
    },
    {
      "epoch": 0.8996528166648001,
      "grad_norm": 1.244759440422058,
      "learning_rate": 8.028999999999999e-05,
      "loss": 0.4149,
      "step": 16066
    },
    {
      "epoch": 0.8997088139769291,
      "grad_norm": 1.2448692321777344,
      "learning_rate": 8.0295e-05,
      "loss": 0.3632,
      "step": 16067
    },
    {
      "epoch": 0.8997648112890582,
      "grad_norm": 1.2463021278381348,
      "learning_rate": 8.030000000000001e-05,
      "loss": 0.4785,
      "step": 16068
    },
    {
      "epoch": 0.8998208086011872,
      "grad_norm": 1.367929458618164,
      "learning_rate": 8.030500000000001e-05,
      "loss": 0.4811,
      "step": 16069
    },
    {
      "epoch": 0.8998768059133162,
      "grad_norm": 1.4838855266571045,
      "learning_rate": 8.031000000000001e-05,
      "loss": 0.425,
      "step": 16070
    },
    {
      "epoch": 0.8999328032254452,
      "grad_norm": 1.6777338981628418,
      "learning_rate": 8.0315e-05,
      "loss": 0.5884,
      "step": 16071
    },
    {
      "epoch": 0.8999888005375742,
      "grad_norm": 1.2199866771697998,
      "learning_rate": 8.032e-05,
      "loss": 0.4392,
      "step": 16072
    },
    {
      "epoch": 0.9000447978497033,
      "grad_norm": 1.167232632637024,
      "learning_rate": 8.0325e-05,
      "loss": 0.4452,
      "step": 16073
    },
    {
      "epoch": 0.9001007951618323,
      "grad_norm": 1.5272061824798584,
      "learning_rate": 8.033000000000001e-05,
      "loss": 0.4721,
      "step": 16074
    },
    {
      "epoch": 0.9001567924739613,
      "grad_norm": 1.4772456884384155,
      "learning_rate": 8.033500000000001e-05,
      "loss": 0.5618,
      "step": 16075
    },
    {
      "epoch": 0.9002127897860903,
      "grad_norm": 1.3673405647277832,
      "learning_rate": 8.034e-05,
      "loss": 0.4879,
      "step": 16076
    },
    {
      "epoch": 0.9002687870982193,
      "grad_norm": 1.3085155487060547,
      "learning_rate": 8.0345e-05,
      "loss": 0.3939,
      "step": 16077
    },
    {
      "epoch": 0.9003247844103484,
      "grad_norm": 1.544012188911438,
      "learning_rate": 8.035e-05,
      "loss": 0.3759,
      "step": 16078
    },
    {
      "epoch": 0.9003807817224774,
      "grad_norm": 1.4712990522384644,
      "learning_rate": 8.0355e-05,
      "loss": 0.5326,
      "step": 16079
    },
    {
      "epoch": 0.9004367790346064,
      "grad_norm": 1.9467556476593018,
      "learning_rate": 8.036e-05,
      "loss": 0.3925,
      "step": 16080
    },
    {
      "epoch": 0.9004927763467353,
      "grad_norm": 1.3120015859603882,
      "learning_rate": 8.036500000000001e-05,
      "loss": 0.3204,
      "step": 16081
    },
    {
      "epoch": 0.9005487736588643,
      "grad_norm": Infinity,
      "learning_rate": 8.036500000000001e-05,
      "loss": 0.4318,
      "step": 16082
    },
    {
      "epoch": 0.9006047709709933,
      "grad_norm": 1.3546849489212036,
      "learning_rate": 8.037e-05,
      "loss": 0.3974,
      "step": 16083
    },
    {
      "epoch": 0.9006607682831224,
      "grad_norm": 1.37135910987854,
      "learning_rate": 8.0375e-05,
      "loss": 0.5232,
      "step": 16084
    },
    {
      "epoch": 0.9007167655952514,
      "grad_norm": 1.3977696895599365,
      "learning_rate": 8.038e-05,
      "loss": 0.4193,
      "step": 16085
    },
    {
      "epoch": 0.9007727629073804,
      "grad_norm": 1.3352080583572388,
      "learning_rate": 8.0385e-05,
      "loss": 0.472,
      "step": 16086
    },
    {
      "epoch": 0.9008287602195094,
      "grad_norm": 1.2196705341339111,
      "learning_rate": 8.039e-05,
      "loss": 0.356,
      "step": 16087
    },
    {
      "epoch": 0.9008847575316384,
      "grad_norm": 1.4114445447921753,
      "learning_rate": 8.0395e-05,
      "loss": 0.4789,
      "step": 16088
    },
    {
      "epoch": 0.9009407548437675,
      "grad_norm": 1.3660703897476196,
      "learning_rate": 8.04e-05,
      "loss": 0.4888,
      "step": 16089
    },
    {
      "epoch": 0.9009967521558965,
      "grad_norm": 1.3786678314208984,
      "learning_rate": 8.040500000000001e-05,
      "loss": 0.4506,
      "step": 16090
    },
    {
      "epoch": 0.9010527494680255,
      "grad_norm": 1.1101665496826172,
      "learning_rate": 8.041000000000001e-05,
      "loss": 0.3506,
      "step": 16091
    },
    {
      "epoch": 0.9011087467801545,
      "grad_norm": 1.319434404373169,
      "learning_rate": 8.041500000000001e-05,
      "loss": 0.3757,
      "step": 16092
    },
    {
      "epoch": 0.9011647440922835,
      "grad_norm": 1.656463861465454,
      "learning_rate": 8.042e-05,
      "loss": 0.5559,
      "step": 16093
    },
    {
      "epoch": 0.9012207414044126,
      "grad_norm": 1.1022865772247314,
      "learning_rate": 8.0425e-05,
      "loss": 0.3704,
      "step": 16094
    },
    {
      "epoch": 0.9012767387165416,
      "grad_norm": 1.38589346408844,
      "learning_rate": 8.043e-05,
      "loss": 0.4217,
      "step": 16095
    },
    {
      "epoch": 0.9013327360286706,
      "grad_norm": 1.690219521522522,
      "learning_rate": 8.0435e-05,
      "loss": 0.5664,
      "step": 16096
    },
    {
      "epoch": 0.9013887333407996,
      "grad_norm": 1.1399365663528442,
      "learning_rate": 8.044000000000001e-05,
      "loss": 0.4389,
      "step": 16097
    },
    {
      "epoch": 0.9014447306529286,
      "grad_norm": 1.3624540567398071,
      "learning_rate": 8.0445e-05,
      "loss": 0.4216,
      "step": 16098
    },
    {
      "epoch": 0.9015007279650576,
      "grad_norm": 1.3338139057159424,
      "learning_rate": 8.045e-05,
      "loss": 0.4295,
      "step": 16099
    },
    {
      "epoch": 0.9015567252771867,
      "grad_norm": 1.6596754789352417,
      "learning_rate": 8.0455e-05,
      "loss": 0.4851,
      "step": 16100
    },
    {
      "epoch": 0.9016127225893157,
      "grad_norm": 1.45106840133667,
      "learning_rate": 8.046e-05,
      "loss": 0.4712,
      "step": 16101
    },
    {
      "epoch": 0.9016687199014447,
      "grad_norm": 1.7162147760391235,
      "learning_rate": 8.0465e-05,
      "loss": 0.5216,
      "step": 16102
    },
    {
      "epoch": 0.9017247172135737,
      "grad_norm": 1.3345248699188232,
      "learning_rate": 8.047000000000001e-05,
      "loss": 0.4811,
      "step": 16103
    },
    {
      "epoch": 0.9017807145257027,
      "grad_norm": 1.5572048425674438,
      "learning_rate": 8.0475e-05,
      "loss": 0.5564,
      "step": 16104
    },
    {
      "epoch": 0.9018367118378318,
      "grad_norm": 1.1450647115707397,
      "learning_rate": 8.048e-05,
      "loss": 0.4413,
      "step": 16105
    },
    {
      "epoch": 0.9018927091499608,
      "grad_norm": 1.807526707649231,
      "learning_rate": 8.0485e-05,
      "loss": 0.6043,
      "step": 16106
    },
    {
      "epoch": 0.9019487064620898,
      "grad_norm": 1.389998435974121,
      "learning_rate": 8.049e-05,
      "loss": 0.497,
      "step": 16107
    },
    {
      "epoch": 0.9020047037742188,
      "grad_norm": 1.0663912296295166,
      "learning_rate": 8.049500000000001e-05,
      "loss": 0.3759,
      "step": 16108
    },
    {
      "epoch": 0.9020607010863478,
      "grad_norm": 1.4233379364013672,
      "learning_rate": 8.05e-05,
      "loss": 0.4,
      "step": 16109
    },
    {
      "epoch": 0.9021166983984769,
      "grad_norm": 1.518378496170044,
      "learning_rate": 8.0505e-05,
      "loss": 0.4115,
      "step": 16110
    },
    {
      "epoch": 0.9021726957106059,
      "grad_norm": 1.3768906593322754,
      "learning_rate": 8.051000000000001e-05,
      "loss": 0.5397,
      "step": 16111
    },
    {
      "epoch": 0.9022286930227349,
      "grad_norm": 1.376766562461853,
      "learning_rate": 8.051500000000001e-05,
      "loss": 0.4534,
      "step": 16112
    },
    {
      "epoch": 0.9022846903348639,
      "grad_norm": 1.288089394569397,
      "learning_rate": 8.052000000000001e-05,
      "loss": 0.4622,
      "step": 16113
    },
    {
      "epoch": 0.9023406876469929,
      "grad_norm": 1.2125897407531738,
      "learning_rate": 8.0525e-05,
      "loss": 0.4396,
      "step": 16114
    },
    {
      "epoch": 0.902396684959122,
      "grad_norm": 2.0193521976470947,
      "learning_rate": 8.053e-05,
      "loss": 0.6333,
      "step": 16115
    },
    {
      "epoch": 0.902452682271251,
      "grad_norm": 1.7033817768096924,
      "learning_rate": 8.0535e-05,
      "loss": 0.5278,
      "step": 16116
    },
    {
      "epoch": 0.90250867958338,
      "grad_norm": 1.4649571180343628,
      "learning_rate": 8.054e-05,
      "loss": 0.5143,
      "step": 16117
    },
    {
      "epoch": 0.902564676895509,
      "grad_norm": 1.4818642139434814,
      "learning_rate": 8.054500000000001e-05,
      "loss": 0.4286,
      "step": 16118
    },
    {
      "epoch": 0.902620674207638,
      "grad_norm": 1.3658850193023682,
      "learning_rate": 8.055e-05,
      "loss": 0.551,
      "step": 16119
    },
    {
      "epoch": 0.902676671519767,
      "grad_norm": 1.3626432418823242,
      "learning_rate": 8.0555e-05,
      "loss": 0.4361,
      "step": 16120
    },
    {
      "epoch": 0.9027326688318961,
      "grad_norm": 1.1653423309326172,
      "learning_rate": 8.056e-05,
      "loss": 0.3681,
      "step": 16121
    },
    {
      "epoch": 0.9027886661440251,
      "grad_norm": 1.0749971866607666,
      "learning_rate": 8.0565e-05,
      "loss": 0.3763,
      "step": 16122
    },
    {
      "epoch": 0.9028446634561541,
      "grad_norm": 1.35155189037323,
      "learning_rate": 8.057e-05,
      "loss": 0.393,
      "step": 16123
    },
    {
      "epoch": 0.9029006607682831,
      "grad_norm": 1.3368000984191895,
      "learning_rate": 8.057500000000001e-05,
      "loss": 0.3983,
      "step": 16124
    },
    {
      "epoch": 0.9029566580804121,
      "grad_norm": 1.3507204055786133,
      "learning_rate": 8.058e-05,
      "loss": 0.4552,
      "step": 16125
    },
    {
      "epoch": 0.9030126553925412,
      "grad_norm": 1.1446025371551514,
      "learning_rate": 8.0585e-05,
      "loss": 0.3386,
      "step": 16126
    },
    {
      "epoch": 0.9030686527046702,
      "grad_norm": 1.4161683320999146,
      "learning_rate": 8.059e-05,
      "loss": 0.4751,
      "step": 16127
    },
    {
      "epoch": 0.9031246500167992,
      "grad_norm": 1.2866870164871216,
      "learning_rate": 8.059500000000001e-05,
      "loss": 0.45,
      "step": 16128
    },
    {
      "epoch": 0.9031806473289282,
      "grad_norm": 1.401611089706421,
      "learning_rate": 8.060000000000001e-05,
      "loss": 0.5611,
      "step": 16129
    },
    {
      "epoch": 0.9032366446410572,
      "grad_norm": 1.4381974935531616,
      "learning_rate": 8.0605e-05,
      "loss": 0.6571,
      "step": 16130
    },
    {
      "epoch": 0.9032926419531863,
      "grad_norm": 1.134797215461731,
      "learning_rate": 8.061e-05,
      "loss": 0.3239,
      "step": 16131
    },
    {
      "epoch": 0.9033486392653153,
      "grad_norm": 1.5025545358657837,
      "learning_rate": 8.061500000000001e-05,
      "loss": 0.4626,
      "step": 16132
    },
    {
      "epoch": 0.9034046365774443,
      "grad_norm": 1.3810558319091797,
      "learning_rate": 8.062000000000001e-05,
      "loss": 0.5958,
      "step": 16133
    },
    {
      "epoch": 0.9034606338895733,
      "grad_norm": 1.4801636934280396,
      "learning_rate": 8.062500000000001e-05,
      "loss": 0.4656,
      "step": 16134
    },
    {
      "epoch": 0.9035166312017023,
      "grad_norm": 1.2599828243255615,
      "learning_rate": 8.063e-05,
      "loss": 0.4976,
      "step": 16135
    },
    {
      "epoch": 0.9035726285138314,
      "grad_norm": 1.2133804559707642,
      "learning_rate": 8.0635e-05,
      "loss": 0.4485,
      "step": 16136
    },
    {
      "epoch": 0.9036286258259604,
      "grad_norm": 1.4889907836914062,
      "learning_rate": 8.064e-05,
      "loss": 0.5225,
      "step": 16137
    },
    {
      "epoch": 0.9036846231380894,
      "grad_norm": 1.5218982696533203,
      "learning_rate": 8.0645e-05,
      "loss": 0.4227,
      "step": 16138
    },
    {
      "epoch": 0.9037406204502184,
      "grad_norm": 1.1445194482803345,
      "learning_rate": 8.065000000000001e-05,
      "loss": 0.3879,
      "step": 16139
    },
    {
      "epoch": 0.9037966177623474,
      "grad_norm": 1.486558198928833,
      "learning_rate": 8.0655e-05,
      "loss": 0.5492,
      "step": 16140
    },
    {
      "epoch": 0.9038526150744765,
      "grad_norm": 1.294519305229187,
      "learning_rate": 8.066e-05,
      "loss": 0.5715,
      "step": 16141
    },
    {
      "epoch": 0.9039086123866055,
      "grad_norm": 1.384840726852417,
      "learning_rate": 8.0665e-05,
      "loss": 0.5428,
      "step": 16142
    },
    {
      "epoch": 0.9039646096987345,
      "grad_norm": 1.2878063917160034,
      "learning_rate": 8.067e-05,
      "loss": 0.4181,
      "step": 16143
    },
    {
      "epoch": 0.9040206070108635,
      "grad_norm": 1.132713794708252,
      "learning_rate": 8.0675e-05,
      "loss": 0.3261,
      "step": 16144
    },
    {
      "epoch": 0.9040766043229925,
      "grad_norm": 1.303546667098999,
      "learning_rate": 8.068e-05,
      "loss": 0.3057,
      "step": 16145
    },
    {
      "epoch": 0.9041326016351215,
      "grad_norm": 1.181791067123413,
      "learning_rate": 8.0685e-05,
      "loss": 0.3682,
      "step": 16146
    },
    {
      "epoch": 0.9041885989472506,
      "grad_norm": 1.2825514078140259,
      "learning_rate": 8.069e-05,
      "loss": 0.4431,
      "step": 16147
    },
    {
      "epoch": 0.9042445962593796,
      "grad_norm": 1.2796927690505981,
      "learning_rate": 8.0695e-05,
      "loss": 0.496,
      "step": 16148
    },
    {
      "epoch": 0.9043005935715086,
      "grad_norm": 1.2571866512298584,
      "learning_rate": 8.070000000000001e-05,
      "loss": 0.4749,
      "step": 16149
    },
    {
      "epoch": 0.9043565908836376,
      "grad_norm": 1.2255233526229858,
      "learning_rate": 8.070500000000001e-05,
      "loss": 0.4372,
      "step": 16150
    },
    {
      "epoch": 0.9044125881957666,
      "grad_norm": 1.5090402364730835,
      "learning_rate": 8.071e-05,
      "loss": 0.7129,
      "step": 16151
    },
    {
      "epoch": 0.9044685855078957,
      "grad_norm": 1.2965089082717896,
      "learning_rate": 8.0715e-05,
      "loss": 0.4358,
      "step": 16152
    },
    {
      "epoch": 0.9045245828200247,
      "grad_norm": 1.4195002317428589,
      "learning_rate": 8.072000000000001e-05,
      "loss": 0.4692,
      "step": 16153
    },
    {
      "epoch": 0.9045805801321537,
      "grad_norm": 1.3298155069351196,
      "learning_rate": 8.072500000000001e-05,
      "loss": 0.4046,
      "step": 16154
    },
    {
      "epoch": 0.9046365774442827,
      "grad_norm": 1.4429786205291748,
      "learning_rate": 8.073000000000001e-05,
      "loss": 0.5544,
      "step": 16155
    },
    {
      "epoch": 0.9046925747564117,
      "grad_norm": 1.4150007963180542,
      "learning_rate": 8.0735e-05,
      "loss": 0.4725,
      "step": 16156
    },
    {
      "epoch": 0.9047485720685408,
      "grad_norm": 1.5700832605361938,
      "learning_rate": 8.074e-05,
      "loss": 0.6968,
      "step": 16157
    },
    {
      "epoch": 0.9048045693806698,
      "grad_norm": 1.4435895681381226,
      "learning_rate": 8.0745e-05,
      "loss": 0.6842,
      "step": 16158
    },
    {
      "epoch": 0.9048605666927988,
      "grad_norm": 18.472076416015625,
      "learning_rate": 8.075e-05,
      "loss": 0.7216,
      "step": 16159
    },
    {
      "epoch": 0.9049165640049278,
      "grad_norm": 1.087300181388855,
      "learning_rate": 8.075500000000001e-05,
      "loss": 0.4338,
      "step": 16160
    },
    {
      "epoch": 0.9049725613170568,
      "grad_norm": 1.6640185117721558,
      "learning_rate": 8.076e-05,
      "loss": 0.8026,
      "step": 16161
    },
    {
      "epoch": 0.9050285586291859,
      "grad_norm": 1.339402675628662,
      "learning_rate": 8.0765e-05,
      "loss": 0.424,
      "step": 16162
    },
    {
      "epoch": 0.9050845559413149,
      "grad_norm": 1.7281773090362549,
      "learning_rate": 8.077e-05,
      "loss": 0.7183,
      "step": 16163
    },
    {
      "epoch": 0.9051405532534438,
      "grad_norm": 1.195365309715271,
      "learning_rate": 8.0775e-05,
      "loss": 0.4868,
      "step": 16164
    },
    {
      "epoch": 0.9051965505655728,
      "grad_norm": 1.2876406908035278,
      "learning_rate": 8.078e-05,
      "loss": 0.5234,
      "step": 16165
    },
    {
      "epoch": 0.9052525478777018,
      "grad_norm": 1.2299528121948242,
      "learning_rate": 8.078499999999999e-05,
      "loss": 0.4793,
      "step": 16166
    },
    {
      "epoch": 0.9053085451898308,
      "grad_norm": 1.245281457901001,
      "learning_rate": 8.079e-05,
      "loss": 0.4979,
      "step": 16167
    },
    {
      "epoch": 0.9053645425019599,
      "grad_norm": 1.3213170766830444,
      "learning_rate": 8.0795e-05,
      "loss": 0.5371,
      "step": 16168
    },
    {
      "epoch": 0.9054205398140889,
      "grad_norm": 1.53616201877594,
      "learning_rate": 8.080000000000001e-05,
      "loss": 0.6766,
      "step": 16169
    },
    {
      "epoch": 0.9054765371262179,
      "grad_norm": 1.4785538911819458,
      "learning_rate": 8.080500000000001e-05,
      "loss": 0.4185,
      "step": 16170
    },
    {
      "epoch": 0.9055325344383469,
      "grad_norm": 1.1670349836349487,
      "learning_rate": 8.081000000000001e-05,
      "loss": 0.3507,
      "step": 16171
    },
    {
      "epoch": 0.9055885317504759,
      "grad_norm": 1.273755431175232,
      "learning_rate": 8.0815e-05,
      "loss": 0.4314,
      "step": 16172
    },
    {
      "epoch": 0.905644529062605,
      "grad_norm": 1.750573992729187,
      "learning_rate": 8.082e-05,
      "loss": 0.4869,
      "step": 16173
    },
    {
      "epoch": 0.905700526374734,
      "grad_norm": 1.0891785621643066,
      "learning_rate": 8.082500000000001e-05,
      "loss": 0.3164,
      "step": 16174
    },
    {
      "epoch": 0.905756523686863,
      "grad_norm": 1.1787875890731812,
      "learning_rate": 8.083000000000001e-05,
      "loss": 0.3236,
      "step": 16175
    },
    {
      "epoch": 0.905812520998992,
      "grad_norm": 1.3453223705291748,
      "learning_rate": 8.083500000000001e-05,
      "loss": 0.3288,
      "step": 16176
    },
    {
      "epoch": 0.905868518311121,
      "grad_norm": 1.140633463859558,
      "learning_rate": 8.084e-05,
      "loss": 0.3574,
      "step": 16177
    },
    {
      "epoch": 0.90592451562325,
      "grad_norm": 1.3836084604263306,
      "learning_rate": 8.0845e-05,
      "loss": 0.4461,
      "step": 16178
    },
    {
      "epoch": 0.9059805129353791,
      "grad_norm": 3.173196792602539,
      "learning_rate": 8.085e-05,
      "loss": 0.3756,
      "step": 16179
    },
    {
      "epoch": 0.9060365102475081,
      "grad_norm": 1.382127046585083,
      "learning_rate": 8.0855e-05,
      "loss": 0.5083,
      "step": 16180
    },
    {
      "epoch": 0.9060925075596371,
      "grad_norm": 1.0873388051986694,
      "learning_rate": 8.086000000000001e-05,
      "loss": 0.4796,
      "step": 16181
    },
    {
      "epoch": 0.9061485048717661,
      "grad_norm": 1.4298620223999023,
      "learning_rate": 8.0865e-05,
      "loss": 0.4708,
      "step": 16182
    },
    {
      "epoch": 0.9062045021838951,
      "grad_norm": 1.3999799489974976,
      "learning_rate": 8.087e-05,
      "loss": 0.6052,
      "step": 16183
    },
    {
      "epoch": 0.9062604994960242,
      "grad_norm": 1.205410361289978,
      "learning_rate": 8.0875e-05,
      "loss": 0.4133,
      "step": 16184
    },
    {
      "epoch": 0.9063164968081532,
      "grad_norm": 1.6735310554504395,
      "learning_rate": 8.088e-05,
      "loss": 0.4577,
      "step": 16185
    },
    {
      "epoch": 0.9063724941202822,
      "grad_norm": 1.5331871509552002,
      "learning_rate": 8.0885e-05,
      "loss": 0.4554,
      "step": 16186
    },
    {
      "epoch": 0.9064284914324112,
      "grad_norm": 1.5027611255645752,
      "learning_rate": 8.088999999999999e-05,
      "loss": 0.57,
      "step": 16187
    },
    {
      "epoch": 0.9064844887445402,
      "grad_norm": 1.3087024688720703,
      "learning_rate": 8.0895e-05,
      "loss": 0.3807,
      "step": 16188
    },
    {
      "epoch": 0.9065404860566693,
      "grad_norm": 1.2069069147109985,
      "learning_rate": 8.090000000000001e-05,
      "loss": 0.344,
      "step": 16189
    },
    {
      "epoch": 0.9065964833687983,
      "grad_norm": 1.1890544891357422,
      "learning_rate": 8.090500000000001e-05,
      "loss": 0.4263,
      "step": 16190
    },
    {
      "epoch": 0.9066524806809273,
      "grad_norm": 1.5398627519607544,
      "learning_rate": 8.091000000000001e-05,
      "loss": 0.4006,
      "step": 16191
    },
    {
      "epoch": 0.9067084779930563,
      "grad_norm": 1.345105528831482,
      "learning_rate": 8.091500000000001e-05,
      "loss": 0.5528,
      "step": 16192
    },
    {
      "epoch": 0.9067644753051853,
      "grad_norm": 1.2979389429092407,
      "learning_rate": 8.092e-05,
      "loss": 0.3823,
      "step": 16193
    },
    {
      "epoch": 0.9068204726173144,
      "grad_norm": 1.7370052337646484,
      "learning_rate": 8.0925e-05,
      "loss": 0.7235,
      "step": 16194
    },
    {
      "epoch": 0.9068764699294434,
      "grad_norm": 1.5376818180084229,
      "learning_rate": 8.093e-05,
      "loss": 0.4005,
      "step": 16195
    },
    {
      "epoch": 0.9069324672415724,
      "grad_norm": 1.472191572189331,
      "learning_rate": 8.093500000000001e-05,
      "loss": 0.399,
      "step": 16196
    },
    {
      "epoch": 0.9069884645537014,
      "grad_norm": 1.2682433128356934,
      "learning_rate": 8.094000000000001e-05,
      "loss": 0.4769,
      "step": 16197
    },
    {
      "epoch": 0.9070444618658304,
      "grad_norm": 1.5884227752685547,
      "learning_rate": 8.0945e-05,
      "loss": 0.5694,
      "step": 16198
    },
    {
      "epoch": 0.9071004591779595,
      "grad_norm": 1.2543954849243164,
      "learning_rate": 8.095e-05,
      "loss": 0.5715,
      "step": 16199
    },
    {
      "epoch": 0.9071564564900885,
      "grad_norm": 1.4897866249084473,
      "learning_rate": 8.0955e-05,
      "loss": 0.4489,
      "step": 16200
    },
    {
      "epoch": 0.9072124538022175,
      "grad_norm": 1.388708233833313,
      "learning_rate": 8.096e-05,
      "loss": 0.5487,
      "step": 16201
    },
    {
      "epoch": 0.9072684511143465,
      "grad_norm": 1.382103443145752,
      "learning_rate": 8.096500000000001e-05,
      "loss": 0.458,
      "step": 16202
    },
    {
      "epoch": 0.9073244484264755,
      "grad_norm": 1.297139286994934,
      "learning_rate": 8.097e-05,
      "loss": 0.4633,
      "step": 16203
    },
    {
      "epoch": 0.9073804457386045,
      "grad_norm": 1.2164654731750488,
      "learning_rate": 8.0975e-05,
      "loss": 0.3735,
      "step": 16204
    },
    {
      "epoch": 0.9074364430507336,
      "grad_norm": 1.4001030921936035,
      "learning_rate": 8.098e-05,
      "loss": 0.4218,
      "step": 16205
    },
    {
      "epoch": 0.9074924403628626,
      "grad_norm": 1.3910261392593384,
      "learning_rate": 8.0985e-05,
      "loss": 0.5399,
      "step": 16206
    },
    {
      "epoch": 0.9075484376749916,
      "grad_norm": 1.693255066871643,
      "learning_rate": 8.099e-05,
      "loss": 0.5294,
      "step": 16207
    },
    {
      "epoch": 0.9076044349871206,
      "grad_norm": 1.3666377067565918,
      "learning_rate": 8.099499999999999e-05,
      "loss": 0.4751,
      "step": 16208
    },
    {
      "epoch": 0.9076604322992496,
      "grad_norm": 9.84433650970459,
      "learning_rate": 8.1e-05,
      "loss": 0.512,
      "step": 16209
    },
    {
      "epoch": 0.9077164296113787,
      "grad_norm": 1.3591711521148682,
      "learning_rate": 8.100500000000001e-05,
      "loss": 0.4619,
      "step": 16210
    },
    {
      "epoch": 0.9077724269235077,
      "grad_norm": 1.3341845273971558,
      "learning_rate": 8.101000000000001e-05,
      "loss": 0.4369,
      "step": 16211
    },
    {
      "epoch": 0.9078284242356367,
      "grad_norm": 1.8678618669509888,
      "learning_rate": 8.101500000000001e-05,
      "loss": 0.4691,
      "step": 16212
    },
    {
      "epoch": 0.9078844215477657,
      "grad_norm": 1.4997316598892212,
      "learning_rate": 8.102000000000001e-05,
      "loss": 0.5044,
      "step": 16213
    },
    {
      "epoch": 0.9079404188598947,
      "grad_norm": 1.2690948247909546,
      "learning_rate": 8.1025e-05,
      "loss": 0.3845,
      "step": 16214
    },
    {
      "epoch": 0.9079964161720238,
      "grad_norm": 1.580435037612915,
      "learning_rate": 8.103e-05,
      "loss": 0.5397,
      "step": 16215
    },
    {
      "epoch": 0.9080524134841528,
      "grad_norm": 1.1663674116134644,
      "learning_rate": 8.1035e-05,
      "loss": 0.4878,
      "step": 16216
    },
    {
      "epoch": 0.9081084107962818,
      "grad_norm": 1.1223293542861938,
      "learning_rate": 8.104000000000001e-05,
      "loss": 0.3977,
      "step": 16217
    },
    {
      "epoch": 0.9081644081084108,
      "grad_norm": 1.6381984949111938,
      "learning_rate": 8.104500000000001e-05,
      "loss": 0.5847,
      "step": 16218
    },
    {
      "epoch": 0.9082204054205398,
      "grad_norm": 1.454047679901123,
      "learning_rate": 8.105e-05,
      "loss": 0.4213,
      "step": 16219
    },
    {
      "epoch": 0.9082764027326689,
      "grad_norm": 2.117156982421875,
      "learning_rate": 8.1055e-05,
      "loss": 0.4909,
      "step": 16220
    },
    {
      "epoch": 0.9083324000447979,
      "grad_norm": 1.3888218402862549,
      "learning_rate": 8.106e-05,
      "loss": 0.3746,
      "step": 16221
    },
    {
      "epoch": 0.9083883973569269,
      "grad_norm": 1.1721341609954834,
      "learning_rate": 8.1065e-05,
      "loss": 0.3608,
      "step": 16222
    },
    {
      "epoch": 0.9084443946690559,
      "grad_norm": 1.574838399887085,
      "learning_rate": 8.107000000000001e-05,
      "loss": 0.5261,
      "step": 16223
    },
    {
      "epoch": 0.9085003919811849,
      "grad_norm": 1.385015845298767,
      "learning_rate": 8.1075e-05,
      "loss": 0.4958,
      "step": 16224
    },
    {
      "epoch": 0.908556389293314,
      "grad_norm": 1.306646704673767,
      "learning_rate": 8.108e-05,
      "loss": 0.4316,
      "step": 16225
    },
    {
      "epoch": 0.908612386605443,
      "grad_norm": 1.5103237628936768,
      "learning_rate": 8.1085e-05,
      "loss": 0.6305,
      "step": 16226
    },
    {
      "epoch": 0.908668383917572,
      "grad_norm": 1.5798276662826538,
      "learning_rate": 8.109e-05,
      "loss": 0.5126,
      "step": 16227
    },
    {
      "epoch": 0.908724381229701,
      "grad_norm": 1.7766233682632446,
      "learning_rate": 8.1095e-05,
      "loss": 0.6594,
      "step": 16228
    },
    {
      "epoch": 0.90878037854183,
      "grad_norm": 1.152177333831787,
      "learning_rate": 8.11e-05,
      "loss": 0.3107,
      "step": 16229
    },
    {
      "epoch": 0.908836375853959,
      "grad_norm": 1.5890731811523438,
      "learning_rate": 8.1105e-05,
      "loss": 0.439,
      "step": 16230
    },
    {
      "epoch": 0.9088923731660881,
      "grad_norm": 1.157092809677124,
      "learning_rate": 8.111000000000001e-05,
      "loss": 0.4707,
      "step": 16231
    },
    {
      "epoch": 0.9089483704782171,
      "grad_norm": 1.3347845077514648,
      "learning_rate": 8.111500000000001e-05,
      "loss": 0.4352,
      "step": 16232
    },
    {
      "epoch": 0.9090043677903461,
      "grad_norm": 1.407378911972046,
      "learning_rate": 8.112000000000001e-05,
      "loss": 0.3694,
      "step": 16233
    },
    {
      "epoch": 0.9090603651024751,
      "grad_norm": 1.2295944690704346,
      "learning_rate": 8.112500000000001e-05,
      "loss": 0.3802,
      "step": 16234
    },
    {
      "epoch": 0.9091163624146041,
      "grad_norm": 1.3493835926055908,
      "learning_rate": 8.113e-05,
      "loss": 0.4413,
      "step": 16235
    },
    {
      "epoch": 0.9091723597267332,
      "grad_norm": 1.043889045715332,
      "learning_rate": 8.1135e-05,
      "loss": 0.4428,
      "step": 16236
    },
    {
      "epoch": 0.9092283570388622,
      "grad_norm": 1.533015489578247,
      "learning_rate": 8.114e-05,
      "loss": 0.5359,
      "step": 16237
    },
    {
      "epoch": 0.9092843543509912,
      "grad_norm": 1.045485258102417,
      "learning_rate": 8.114500000000001e-05,
      "loss": 0.3455,
      "step": 16238
    },
    {
      "epoch": 0.9093403516631202,
      "grad_norm": 1.2184590101242065,
      "learning_rate": 8.115000000000001e-05,
      "loss": 0.3954,
      "step": 16239
    },
    {
      "epoch": 0.9093963489752492,
      "grad_norm": 1.220358967781067,
      "learning_rate": 8.1155e-05,
      "loss": 0.3347,
      "step": 16240
    },
    {
      "epoch": 0.9094523462873783,
      "grad_norm": 1.1589940786361694,
      "learning_rate": 8.116e-05,
      "loss": 0.5209,
      "step": 16241
    },
    {
      "epoch": 0.9095083435995073,
      "grad_norm": 1.555674433708191,
      "learning_rate": 8.1165e-05,
      "loss": 0.5367,
      "step": 16242
    },
    {
      "epoch": 0.9095643409116363,
      "grad_norm": 1.1752678155899048,
      "learning_rate": 8.117e-05,
      "loss": 0.4052,
      "step": 16243
    },
    {
      "epoch": 0.9096203382237653,
      "grad_norm": 1.2515757083892822,
      "learning_rate": 8.1175e-05,
      "loss": 0.4064,
      "step": 16244
    },
    {
      "epoch": 0.9096763355358943,
      "grad_norm": 1.4909850358963013,
      "learning_rate": 8.118e-05,
      "loss": 0.4566,
      "step": 16245
    },
    {
      "epoch": 0.9097323328480232,
      "grad_norm": 1.183108925819397,
      "learning_rate": 8.1185e-05,
      "loss": 0.43,
      "step": 16246
    },
    {
      "epoch": 0.9097883301601523,
      "grad_norm": 1.329343318939209,
      "learning_rate": 8.119e-05,
      "loss": 0.4417,
      "step": 16247
    },
    {
      "epoch": 0.9098443274722813,
      "grad_norm": 1.2194442749023438,
      "learning_rate": 8.1195e-05,
      "loss": 0.403,
      "step": 16248
    },
    {
      "epoch": 0.9099003247844103,
      "grad_norm": 1.368298888206482,
      "learning_rate": 8.120000000000001e-05,
      "loss": 0.4122,
      "step": 16249
    },
    {
      "epoch": 0.9099563220965393,
      "grad_norm": 1.3849939107894897,
      "learning_rate": 8.1205e-05,
      "loss": 0.3577,
      "step": 16250
    },
    {
      "epoch": 0.9100123194086683,
      "grad_norm": 1.4291472434997559,
      "learning_rate": 8.121e-05,
      "loss": 0.5803,
      "step": 16251
    },
    {
      "epoch": 0.9100683167207974,
      "grad_norm": 1.1422356367111206,
      "learning_rate": 8.121500000000001e-05,
      "loss": 0.3736,
      "step": 16252
    },
    {
      "epoch": 0.9101243140329264,
      "grad_norm": 1.247540831565857,
      "learning_rate": 8.122000000000001e-05,
      "loss": 0.4379,
      "step": 16253
    },
    {
      "epoch": 0.9101803113450554,
      "grad_norm": 1.3040695190429688,
      "learning_rate": 8.122500000000001e-05,
      "loss": 0.506,
      "step": 16254
    },
    {
      "epoch": 0.9102363086571844,
      "grad_norm": 1.257175087928772,
      "learning_rate": 8.123e-05,
      "loss": 0.3589,
      "step": 16255
    },
    {
      "epoch": 0.9102923059693134,
      "grad_norm": 1.5271435976028442,
      "learning_rate": 8.1235e-05,
      "loss": 0.4279,
      "step": 16256
    },
    {
      "epoch": 0.9103483032814425,
      "grad_norm": 1.2693156003952026,
      "learning_rate": 8.124e-05,
      "loss": 0.4413,
      "step": 16257
    },
    {
      "epoch": 0.9104043005935715,
      "grad_norm": 1.3279603719711304,
      "learning_rate": 8.1245e-05,
      "loss": 0.4359,
      "step": 16258
    },
    {
      "epoch": 0.9104602979057005,
      "grad_norm": 1.4256255626678467,
      "learning_rate": 8.125000000000001e-05,
      "loss": 0.4211,
      "step": 16259
    },
    {
      "epoch": 0.9105162952178295,
      "grad_norm": 1.497703194618225,
      "learning_rate": 8.125500000000001e-05,
      "loss": 0.4449,
      "step": 16260
    },
    {
      "epoch": 0.9105722925299585,
      "grad_norm": 1.348031997680664,
      "learning_rate": 8.126e-05,
      "loss": 0.3811,
      "step": 16261
    },
    {
      "epoch": 0.9106282898420875,
      "grad_norm": 1.5301815271377563,
      "learning_rate": 8.1265e-05,
      "loss": 0.6615,
      "step": 16262
    },
    {
      "epoch": 0.9106842871542166,
      "grad_norm": 1.6751320362091064,
      "learning_rate": 8.127e-05,
      "loss": 0.5039,
      "step": 16263
    },
    {
      "epoch": 0.9107402844663456,
      "grad_norm": 1.3533034324645996,
      "learning_rate": 8.1275e-05,
      "loss": 0.379,
      "step": 16264
    },
    {
      "epoch": 0.9107962817784746,
      "grad_norm": 1.2872544527053833,
      "learning_rate": 8.128e-05,
      "loss": 0.4944,
      "step": 16265
    },
    {
      "epoch": 0.9108522790906036,
      "grad_norm": 1.281751036643982,
      "learning_rate": 8.1285e-05,
      "loss": 0.4542,
      "step": 16266
    },
    {
      "epoch": 0.9109082764027326,
      "grad_norm": 1.4266624450683594,
      "learning_rate": 8.129e-05,
      "loss": 0.4478,
      "step": 16267
    },
    {
      "epoch": 0.9109642737148617,
      "grad_norm": 1.7636620998382568,
      "learning_rate": 8.1295e-05,
      "loss": 0.6262,
      "step": 16268
    },
    {
      "epoch": 0.9110202710269907,
      "grad_norm": 1.3292865753173828,
      "learning_rate": 8.13e-05,
      "loss": 0.456,
      "step": 16269
    },
    {
      "epoch": 0.9110762683391197,
      "grad_norm": 1.4947370290756226,
      "learning_rate": 8.130500000000001e-05,
      "loss": 0.5257,
      "step": 16270
    },
    {
      "epoch": 0.9111322656512487,
      "grad_norm": 1.2445459365844727,
      "learning_rate": 8.131e-05,
      "loss": 0.4555,
      "step": 16271
    },
    {
      "epoch": 0.9111882629633777,
      "grad_norm": 1.5176284313201904,
      "learning_rate": 8.1315e-05,
      "loss": 0.6189,
      "step": 16272
    },
    {
      "epoch": 0.9112442602755068,
      "grad_norm": 1.2148562669754028,
      "learning_rate": 8.132e-05,
      "loss": 0.4785,
      "step": 16273
    },
    {
      "epoch": 0.9113002575876358,
      "grad_norm": 1.1544235944747925,
      "learning_rate": 8.132500000000001e-05,
      "loss": 0.4687,
      "step": 16274
    },
    {
      "epoch": 0.9113562548997648,
      "grad_norm": 1.3045356273651123,
      "learning_rate": 8.133000000000001e-05,
      "loss": 0.5035,
      "step": 16275
    },
    {
      "epoch": 0.9114122522118938,
      "grad_norm": 1.4231603145599365,
      "learning_rate": 8.1335e-05,
      "loss": 0.4949,
      "step": 16276
    },
    {
      "epoch": 0.9114682495240228,
      "grad_norm": 1.5219913721084595,
      "learning_rate": 8.134e-05,
      "loss": 0.5055,
      "step": 16277
    },
    {
      "epoch": 0.9115242468361519,
      "grad_norm": 1.6135090589523315,
      "learning_rate": 8.1345e-05,
      "loss": 0.5868,
      "step": 16278
    },
    {
      "epoch": 0.9115802441482809,
      "grad_norm": 1.9982261657714844,
      "learning_rate": 8.135e-05,
      "loss": 0.7034,
      "step": 16279
    },
    {
      "epoch": 0.9116362414604099,
      "grad_norm": 1.4176361560821533,
      "learning_rate": 8.135500000000001e-05,
      "loss": 0.3944,
      "step": 16280
    },
    {
      "epoch": 0.9116922387725389,
      "grad_norm": 1.4432766437530518,
      "learning_rate": 8.136000000000001e-05,
      "loss": 0.4427,
      "step": 16281
    },
    {
      "epoch": 0.9117482360846679,
      "grad_norm": 1.5466388463974,
      "learning_rate": 8.1365e-05,
      "loss": 0.4778,
      "step": 16282
    },
    {
      "epoch": 0.911804233396797,
      "grad_norm": 1.649107575416565,
      "learning_rate": 8.137e-05,
      "loss": 0.546,
      "step": 16283
    },
    {
      "epoch": 0.911860230708926,
      "grad_norm": 1.3462942838668823,
      "learning_rate": 8.1375e-05,
      "loss": 0.38,
      "step": 16284
    },
    {
      "epoch": 0.911916228021055,
      "grad_norm": 1.2513823509216309,
      "learning_rate": 8.138e-05,
      "loss": 0.571,
      "step": 16285
    },
    {
      "epoch": 0.911972225333184,
      "grad_norm": 1.1332285404205322,
      "learning_rate": 8.1385e-05,
      "loss": 0.3993,
      "step": 16286
    },
    {
      "epoch": 0.912028222645313,
      "grad_norm": 1.279436707496643,
      "learning_rate": 8.139e-05,
      "loss": 0.5434,
      "step": 16287
    },
    {
      "epoch": 0.912084219957442,
      "grad_norm": 1.4637783765792847,
      "learning_rate": 8.1395e-05,
      "loss": 0.4162,
      "step": 16288
    },
    {
      "epoch": 0.9121402172695711,
      "grad_norm": 1.2769397497177124,
      "learning_rate": 8.14e-05,
      "loss": 0.4708,
      "step": 16289
    },
    {
      "epoch": 0.9121962145817001,
      "grad_norm": 1.3884077072143555,
      "learning_rate": 8.140500000000001e-05,
      "loss": 0.4197,
      "step": 16290
    },
    {
      "epoch": 0.9122522118938291,
      "grad_norm": 1.5181313753128052,
      "learning_rate": 8.141000000000001e-05,
      "loss": 0.4158,
      "step": 16291
    },
    {
      "epoch": 0.9123082092059581,
      "grad_norm": 1.2915698289871216,
      "learning_rate": 8.1415e-05,
      "loss": 0.4737,
      "step": 16292
    },
    {
      "epoch": 0.9123642065180871,
      "grad_norm": 1.173693299293518,
      "learning_rate": 8.142e-05,
      "loss": 0.3346,
      "step": 16293
    },
    {
      "epoch": 0.9124202038302162,
      "grad_norm": 1.1621513366699219,
      "learning_rate": 8.1425e-05,
      "loss": 0.4959,
      "step": 16294
    },
    {
      "epoch": 0.9124762011423452,
      "grad_norm": 1.2622287273406982,
      "learning_rate": 8.143000000000001e-05,
      "loss": 0.3461,
      "step": 16295
    },
    {
      "epoch": 0.9125321984544742,
      "grad_norm": 1.4559603929519653,
      "learning_rate": 8.143500000000001e-05,
      "loss": 0.5704,
      "step": 16296
    },
    {
      "epoch": 0.9125881957666032,
      "grad_norm": 1.3613815307617188,
      "learning_rate": 8.144e-05,
      "loss": 0.4548,
      "step": 16297
    },
    {
      "epoch": 0.9126441930787322,
      "grad_norm": 1.4421823024749756,
      "learning_rate": 8.1445e-05,
      "loss": 0.4864,
      "step": 16298
    },
    {
      "epoch": 0.9127001903908613,
      "grad_norm": 1.381778359413147,
      "learning_rate": 8.145e-05,
      "loss": 0.5673,
      "step": 16299
    },
    {
      "epoch": 0.9127561877029903,
      "grad_norm": 1.7580300569534302,
      "learning_rate": 8.1455e-05,
      "loss": 0.4055,
      "step": 16300
    },
    {
      "epoch": 0.9128121850151193,
      "grad_norm": 1.568524956703186,
      "learning_rate": 8.146000000000001e-05,
      "loss": 0.597,
      "step": 16301
    },
    {
      "epoch": 0.9128681823272483,
      "grad_norm": 1.4807724952697754,
      "learning_rate": 8.146500000000001e-05,
      "loss": 0.6351,
      "step": 16302
    },
    {
      "epoch": 0.9129241796393773,
      "grad_norm": 1.2681360244750977,
      "learning_rate": 8.147e-05,
      "loss": 0.3429,
      "step": 16303
    },
    {
      "epoch": 0.9129801769515064,
      "grad_norm": 1.101849913597107,
      "learning_rate": 8.1475e-05,
      "loss": 0.3718,
      "step": 16304
    },
    {
      "epoch": 0.9130361742636354,
      "grad_norm": 1.3666799068450928,
      "learning_rate": 8.148e-05,
      "loss": 0.3794,
      "step": 16305
    },
    {
      "epoch": 0.9130921715757644,
      "grad_norm": 1.3681020736694336,
      "learning_rate": 8.1485e-05,
      "loss": 0.5026,
      "step": 16306
    },
    {
      "epoch": 0.9131481688878934,
      "grad_norm": 1.272249698638916,
      "learning_rate": 8.149e-05,
      "loss": 0.3594,
      "step": 16307
    },
    {
      "epoch": 0.9132041662000224,
      "grad_norm": 1.4186328649520874,
      "learning_rate": 8.1495e-05,
      "loss": 0.5804,
      "step": 16308
    },
    {
      "epoch": 0.9132601635121514,
      "grad_norm": 1.6269071102142334,
      "learning_rate": 8.15e-05,
      "loss": 0.4691,
      "step": 16309
    },
    {
      "epoch": 0.9133161608242805,
      "grad_norm": 1.2336229085922241,
      "learning_rate": 8.150500000000001e-05,
      "loss": 0.5393,
      "step": 16310
    },
    {
      "epoch": 0.9133721581364095,
      "grad_norm": 1.4158024787902832,
      "learning_rate": 8.151000000000001e-05,
      "loss": 0.4173,
      "step": 16311
    },
    {
      "epoch": 0.9134281554485385,
      "grad_norm": 1.4624764919281006,
      "learning_rate": 8.151500000000001e-05,
      "loss": 0.4952,
      "step": 16312
    },
    {
      "epoch": 0.9134841527606675,
      "grad_norm": 1.443229079246521,
      "learning_rate": 8.152e-05,
      "loss": 0.6761,
      "step": 16313
    },
    {
      "epoch": 0.9135401500727965,
      "grad_norm": 1.2824628353118896,
      "learning_rate": 8.1525e-05,
      "loss": 0.6185,
      "step": 16314
    },
    {
      "epoch": 0.9135961473849256,
      "grad_norm": 1.2834609746932983,
      "learning_rate": 8.153e-05,
      "loss": 0.4322,
      "step": 16315
    },
    {
      "epoch": 0.9136521446970546,
      "grad_norm": 1.5608112812042236,
      "learning_rate": 8.153500000000001e-05,
      "loss": 0.5059,
      "step": 16316
    },
    {
      "epoch": 0.9137081420091836,
      "grad_norm": 1.5314184427261353,
      "learning_rate": 8.154000000000001e-05,
      "loss": 0.415,
      "step": 16317
    },
    {
      "epoch": 0.9137641393213126,
      "grad_norm": 1.286604881286621,
      "learning_rate": 8.1545e-05,
      "loss": 0.5446,
      "step": 16318
    },
    {
      "epoch": 0.9138201366334416,
      "grad_norm": 1.4087798595428467,
      "learning_rate": 8.155e-05,
      "loss": 0.4123,
      "step": 16319
    },
    {
      "epoch": 0.9138761339455707,
      "grad_norm": 1.2750279903411865,
      "learning_rate": 8.1555e-05,
      "loss": 0.5283,
      "step": 16320
    },
    {
      "epoch": 0.9139321312576997,
      "grad_norm": 1.311269760131836,
      "learning_rate": 8.156e-05,
      "loss": 0.6315,
      "step": 16321
    },
    {
      "epoch": 0.9139881285698287,
      "grad_norm": 1.63417387008667,
      "learning_rate": 8.1565e-05,
      "loss": 0.4848,
      "step": 16322
    },
    {
      "epoch": 0.9140441258819577,
      "grad_norm": 1.385966181755066,
      "learning_rate": 8.157e-05,
      "loss": 0.4425,
      "step": 16323
    },
    {
      "epoch": 0.9141001231940867,
      "grad_norm": 1.4401918649673462,
      "learning_rate": 8.1575e-05,
      "loss": 0.4885,
      "step": 16324
    },
    {
      "epoch": 0.9141561205062158,
      "grad_norm": 1.2513331174850464,
      "learning_rate": 8.158e-05,
      "loss": 0.4367,
      "step": 16325
    },
    {
      "epoch": 0.9142121178183448,
      "grad_norm": 1.186728596687317,
      "learning_rate": 8.1585e-05,
      "loss": 0.5121,
      "step": 16326
    },
    {
      "epoch": 0.9142681151304738,
      "grad_norm": 1.219286322593689,
      "learning_rate": 8.159e-05,
      "loss": 0.3755,
      "step": 16327
    },
    {
      "epoch": 0.9143241124426028,
      "grad_norm": 1.8178082704544067,
      "learning_rate": 8.1595e-05,
      "loss": 0.3789,
      "step": 16328
    },
    {
      "epoch": 0.9143801097547317,
      "grad_norm": 1.340780258178711,
      "learning_rate": 8.16e-05,
      "loss": 0.4837,
      "step": 16329
    },
    {
      "epoch": 0.9144361070668607,
      "grad_norm": 1.390148401260376,
      "learning_rate": 8.160500000000002e-05,
      "loss": 0.4514,
      "step": 16330
    },
    {
      "epoch": 0.9144921043789898,
      "grad_norm": 1.273046612739563,
      "learning_rate": 8.161000000000001e-05,
      "loss": 0.4115,
      "step": 16331
    },
    {
      "epoch": 0.9145481016911188,
      "grad_norm": 1.2908871173858643,
      "learning_rate": 8.161500000000001e-05,
      "loss": 0.4347,
      "step": 16332
    },
    {
      "epoch": 0.9146040990032478,
      "grad_norm": 1.2404327392578125,
      "learning_rate": 8.162000000000001e-05,
      "loss": 0.3864,
      "step": 16333
    },
    {
      "epoch": 0.9146600963153768,
      "grad_norm": 1.2438278198242188,
      "learning_rate": 8.1625e-05,
      "loss": 0.4607,
      "step": 16334
    },
    {
      "epoch": 0.9147160936275058,
      "grad_norm": 1.335219383239746,
      "learning_rate": 8.163e-05,
      "loss": 0.4362,
      "step": 16335
    },
    {
      "epoch": 0.9147720909396349,
      "grad_norm": 1.4740371704101562,
      "learning_rate": 8.1635e-05,
      "loss": 0.4317,
      "step": 16336
    },
    {
      "epoch": 0.9148280882517639,
      "grad_norm": 1.3565070629119873,
      "learning_rate": 8.164000000000001e-05,
      "loss": 0.4093,
      "step": 16337
    },
    {
      "epoch": 0.9148840855638929,
      "grad_norm": 1.1543887853622437,
      "learning_rate": 8.164500000000001e-05,
      "loss": 0.4376,
      "step": 16338
    },
    {
      "epoch": 0.9149400828760219,
      "grad_norm": 1.3757209777832031,
      "learning_rate": 8.165e-05,
      "loss": 0.356,
      "step": 16339
    },
    {
      "epoch": 0.9149960801881509,
      "grad_norm": 1.2366504669189453,
      "learning_rate": 8.1655e-05,
      "loss": 0.3948,
      "step": 16340
    },
    {
      "epoch": 0.91505207750028,
      "grad_norm": 1.3505820035934448,
      "learning_rate": 8.166e-05,
      "loss": 0.4726,
      "step": 16341
    },
    {
      "epoch": 0.915108074812409,
      "grad_norm": 1.359368920326233,
      "learning_rate": 8.1665e-05,
      "loss": 0.4518,
      "step": 16342
    },
    {
      "epoch": 0.915164072124538,
      "grad_norm": 1.362165093421936,
      "learning_rate": 8.167e-05,
      "loss": 0.5659,
      "step": 16343
    },
    {
      "epoch": 0.915220069436667,
      "grad_norm": 1.3214900493621826,
      "learning_rate": 8.1675e-05,
      "loss": 0.4294,
      "step": 16344
    },
    {
      "epoch": 0.915276066748796,
      "grad_norm": 1.562992811203003,
      "learning_rate": 8.168e-05,
      "loss": 0.4319,
      "step": 16345
    },
    {
      "epoch": 0.915332064060925,
      "grad_norm": 1.7184698581695557,
      "learning_rate": 8.1685e-05,
      "loss": 0.5534,
      "step": 16346
    },
    {
      "epoch": 0.9153880613730541,
      "grad_norm": 1.6614658832550049,
      "learning_rate": 8.169e-05,
      "loss": 0.5746,
      "step": 16347
    },
    {
      "epoch": 0.9154440586851831,
      "grad_norm": 1.4695688486099243,
      "learning_rate": 8.1695e-05,
      "loss": 0.5482,
      "step": 16348
    },
    {
      "epoch": 0.9155000559973121,
      "grad_norm": 1.4225592613220215,
      "learning_rate": 8.17e-05,
      "loss": 0.5707,
      "step": 16349
    },
    {
      "epoch": 0.9155560533094411,
      "grad_norm": 1.2848695516586304,
      "learning_rate": 8.1705e-05,
      "loss": 0.344,
      "step": 16350
    },
    {
      "epoch": 0.9156120506215701,
      "grad_norm": 1.327746868133545,
      "learning_rate": 8.171e-05,
      "loss": 0.5246,
      "step": 16351
    },
    {
      "epoch": 0.9156680479336992,
      "grad_norm": 1.410145878791809,
      "learning_rate": 8.171500000000001e-05,
      "loss": 0.5809,
      "step": 16352
    },
    {
      "epoch": 0.9157240452458282,
      "grad_norm": 1.1476210355758667,
      "learning_rate": 8.172000000000001e-05,
      "loss": 0.3379,
      "step": 16353
    },
    {
      "epoch": 0.9157800425579572,
      "grad_norm": 1.2684223651885986,
      "learning_rate": 8.172500000000001e-05,
      "loss": 0.4586,
      "step": 16354
    },
    {
      "epoch": 0.9158360398700862,
      "grad_norm": 1.221681833267212,
      "learning_rate": 8.173e-05,
      "loss": 0.4008,
      "step": 16355
    },
    {
      "epoch": 0.9158920371822152,
      "grad_norm": 1.29313325881958,
      "learning_rate": 8.1735e-05,
      "loss": 0.3409,
      "step": 16356
    },
    {
      "epoch": 0.9159480344943443,
      "grad_norm": 1.157299518585205,
      "learning_rate": 8.174e-05,
      "loss": 0.4208,
      "step": 16357
    },
    {
      "epoch": 0.9160040318064733,
      "grad_norm": 1.4694918394088745,
      "learning_rate": 8.174500000000001e-05,
      "loss": 0.6095,
      "step": 16358
    },
    {
      "epoch": 0.9160600291186023,
      "grad_norm": 1.538213849067688,
      "learning_rate": 8.175000000000001e-05,
      "loss": 0.4755,
      "step": 16359
    },
    {
      "epoch": 0.9161160264307313,
      "grad_norm": 1.3335171937942505,
      "learning_rate": 8.1755e-05,
      "loss": 0.4496,
      "step": 16360
    },
    {
      "epoch": 0.9161720237428603,
      "grad_norm": 1.4281189441680908,
      "learning_rate": 8.176e-05,
      "loss": 0.4759,
      "step": 16361
    },
    {
      "epoch": 0.9162280210549893,
      "grad_norm": 1.1768161058425903,
      "learning_rate": 8.1765e-05,
      "loss": 0.3872,
      "step": 16362
    },
    {
      "epoch": 0.9162840183671184,
      "grad_norm": 1.4018878936767578,
      "learning_rate": 8.177e-05,
      "loss": 0.4976,
      "step": 16363
    },
    {
      "epoch": 0.9163400156792474,
      "grad_norm": 1.1655353307724,
      "learning_rate": 8.1775e-05,
      "loss": 0.4554,
      "step": 16364
    },
    {
      "epoch": 0.9163960129913764,
      "grad_norm": 1.2682573795318604,
      "learning_rate": 8.178e-05,
      "loss": 0.3436,
      "step": 16365
    },
    {
      "epoch": 0.9164520103035054,
      "grad_norm": 1.6166516542434692,
      "learning_rate": 8.1785e-05,
      "loss": 0.4677,
      "step": 16366
    },
    {
      "epoch": 0.9165080076156344,
      "grad_norm": 1.3583062887191772,
      "learning_rate": 8.179e-05,
      "loss": 0.4421,
      "step": 16367
    },
    {
      "epoch": 0.9165640049277635,
      "grad_norm": 1.5479578971862793,
      "learning_rate": 8.1795e-05,
      "loss": 0.6007,
      "step": 16368
    },
    {
      "epoch": 0.9166200022398925,
      "grad_norm": 1.1865326166152954,
      "learning_rate": 8.18e-05,
      "loss": 0.4101,
      "step": 16369
    },
    {
      "epoch": 0.9166759995520215,
      "grad_norm": 1.5260560512542725,
      "learning_rate": 8.180500000000001e-05,
      "loss": 0.5892,
      "step": 16370
    },
    {
      "epoch": 0.9167319968641505,
      "grad_norm": 1.3571499586105347,
      "learning_rate": 8.181e-05,
      "loss": 0.3783,
      "step": 16371
    },
    {
      "epoch": 0.9167879941762795,
      "grad_norm": 1.2588247060775757,
      "learning_rate": 8.1815e-05,
      "loss": 0.4172,
      "step": 16372
    },
    {
      "epoch": 0.9168439914884086,
      "grad_norm": 1.2398922443389893,
      "learning_rate": 8.182000000000001e-05,
      "loss": 0.4312,
      "step": 16373
    },
    {
      "epoch": 0.9168999888005376,
      "grad_norm": 1.4239126443862915,
      "learning_rate": 8.182500000000001e-05,
      "loss": 0.428,
      "step": 16374
    },
    {
      "epoch": 0.9169559861126666,
      "grad_norm": 1.1611047983169556,
      "learning_rate": 8.183000000000001e-05,
      "loss": 0.3953,
      "step": 16375
    },
    {
      "epoch": 0.9170119834247956,
      "grad_norm": 1.212927222251892,
      "learning_rate": 8.1835e-05,
      "loss": 0.4476,
      "step": 16376
    },
    {
      "epoch": 0.9170679807369246,
      "grad_norm": 1.2236628532409668,
      "learning_rate": 8.184e-05,
      "loss": 0.4802,
      "step": 16377
    },
    {
      "epoch": 0.9171239780490537,
      "grad_norm": 1.3253746032714844,
      "learning_rate": 8.1845e-05,
      "loss": 0.5255,
      "step": 16378
    },
    {
      "epoch": 0.9171799753611827,
      "grad_norm": 1.1778866052627563,
      "learning_rate": 8.185000000000001e-05,
      "loss": 0.464,
      "step": 16379
    },
    {
      "epoch": 0.9172359726733117,
      "grad_norm": 1.138843059539795,
      "learning_rate": 8.185500000000001e-05,
      "loss": 0.331,
      "step": 16380
    },
    {
      "epoch": 0.9172919699854407,
      "grad_norm": 1.3189870119094849,
      "learning_rate": 8.186e-05,
      "loss": 0.397,
      "step": 16381
    },
    {
      "epoch": 0.9173479672975697,
      "grad_norm": 1.7547982931137085,
      "learning_rate": 8.1865e-05,
      "loss": 0.5269,
      "step": 16382
    },
    {
      "epoch": 0.9174039646096988,
      "grad_norm": 1.3204976320266724,
      "learning_rate": 8.187e-05,
      "loss": 0.3563,
      "step": 16383
    },
    {
      "epoch": 0.9174599619218278,
      "grad_norm": 1.531217098236084,
      "learning_rate": 8.1875e-05,
      "loss": 0.6081,
      "step": 16384
    },
    {
      "epoch": 0.9175159592339568,
      "grad_norm": 1.2082085609436035,
      "learning_rate": 8.188e-05,
      "loss": 0.3193,
      "step": 16385
    },
    {
      "epoch": 0.9175719565460858,
      "grad_norm": 1.1349927186965942,
      "learning_rate": 8.1885e-05,
      "loss": 0.3463,
      "step": 16386
    },
    {
      "epoch": 0.9176279538582148,
      "grad_norm": 1.5495916604995728,
      "learning_rate": 8.189e-05,
      "loss": 0.4939,
      "step": 16387
    },
    {
      "epoch": 0.9176839511703438,
      "grad_norm": 1.2847042083740234,
      "learning_rate": 8.1895e-05,
      "loss": 0.5711,
      "step": 16388
    },
    {
      "epoch": 0.9177399484824729,
      "grad_norm": 1.3056803941726685,
      "learning_rate": 8.19e-05,
      "loss": 0.3994,
      "step": 16389
    },
    {
      "epoch": 0.9177959457946019,
      "grad_norm": 1.2236343622207642,
      "learning_rate": 8.1905e-05,
      "loss": 0.3643,
      "step": 16390
    },
    {
      "epoch": 0.9178519431067309,
      "grad_norm": 1.272200584411621,
      "learning_rate": 8.191000000000001e-05,
      "loss": 0.4945,
      "step": 16391
    },
    {
      "epoch": 0.9179079404188599,
      "grad_norm": 1.504387378692627,
      "learning_rate": 8.1915e-05,
      "loss": 0.4055,
      "step": 16392
    },
    {
      "epoch": 0.9179639377309889,
      "grad_norm": 1.4934040307998657,
      "learning_rate": 8.192e-05,
      "loss": 0.4389,
      "step": 16393
    },
    {
      "epoch": 0.918019935043118,
      "grad_norm": 1.3719617128372192,
      "learning_rate": 8.192500000000001e-05,
      "loss": 0.4529,
      "step": 16394
    },
    {
      "epoch": 0.918075932355247,
      "grad_norm": 1.6728545427322388,
      "learning_rate": 8.193000000000001e-05,
      "loss": 0.4228,
      "step": 16395
    },
    {
      "epoch": 0.918131929667376,
      "grad_norm": 1.1509687900543213,
      "learning_rate": 8.193500000000001e-05,
      "loss": 0.341,
      "step": 16396
    },
    {
      "epoch": 0.918187926979505,
      "grad_norm": 1.6074631214141846,
      "learning_rate": 8.194e-05,
      "loss": 0.5055,
      "step": 16397
    },
    {
      "epoch": 0.918243924291634,
      "grad_norm": 1.5949623584747314,
      "learning_rate": 8.1945e-05,
      "loss": 0.4419,
      "step": 16398
    },
    {
      "epoch": 0.9182999216037631,
      "grad_norm": 1.4974658489227295,
      "learning_rate": 8.195e-05,
      "loss": 0.4307,
      "step": 16399
    },
    {
      "epoch": 0.9183559189158921,
      "grad_norm": 1.205846905708313,
      "learning_rate": 8.195500000000001e-05,
      "loss": 0.4287,
      "step": 16400
    },
    {
      "epoch": 0.9184119162280211,
      "grad_norm": 1.346282958984375,
      "learning_rate": 8.196000000000001e-05,
      "loss": 0.4806,
      "step": 16401
    },
    {
      "epoch": 0.9184679135401501,
      "grad_norm": 1.577929139137268,
      "learning_rate": 8.1965e-05,
      "loss": 0.5559,
      "step": 16402
    },
    {
      "epoch": 0.9185239108522791,
      "grad_norm": 1.194756269454956,
      "learning_rate": 8.197e-05,
      "loss": 0.42,
      "step": 16403
    },
    {
      "epoch": 0.9185799081644082,
      "grad_norm": 1.3071774244308472,
      "learning_rate": 8.1975e-05,
      "loss": 0.4382,
      "step": 16404
    },
    {
      "epoch": 0.9186359054765372,
      "grad_norm": 1.2871538400650024,
      "learning_rate": 8.198e-05,
      "loss": 0.5422,
      "step": 16405
    },
    {
      "epoch": 0.9186919027886662,
      "grad_norm": 1.2749103307724,
      "learning_rate": 8.1985e-05,
      "loss": 0.5609,
      "step": 16406
    },
    {
      "epoch": 0.9187479001007952,
      "grad_norm": 1.362830638885498,
      "learning_rate": 8.199e-05,
      "loss": 0.5608,
      "step": 16407
    },
    {
      "epoch": 0.9188038974129242,
      "grad_norm": 1.5428667068481445,
      "learning_rate": 8.1995e-05,
      "loss": 0.4483,
      "step": 16408
    },
    {
      "epoch": 0.9188598947250533,
      "grad_norm": 1.3721121549606323,
      "learning_rate": 8.2e-05,
      "loss": 0.4578,
      "step": 16409
    },
    {
      "epoch": 0.9189158920371823,
      "grad_norm": 1.4253727197647095,
      "learning_rate": 8.2005e-05,
      "loss": 0.4589,
      "step": 16410
    },
    {
      "epoch": 0.9189718893493113,
      "grad_norm": 1.0003160238265991,
      "learning_rate": 8.201000000000001e-05,
      "loss": 0.3628,
      "step": 16411
    },
    {
      "epoch": 0.9190278866614402,
      "grad_norm": 1.298473834991455,
      "learning_rate": 8.2015e-05,
      "loss": 0.3898,
      "step": 16412
    },
    {
      "epoch": 0.9190838839735692,
      "grad_norm": 1.5291424989700317,
      "learning_rate": 8.202e-05,
      "loss": 0.5072,
      "step": 16413
    },
    {
      "epoch": 0.9191398812856982,
      "grad_norm": 1.2101683616638184,
      "learning_rate": 8.2025e-05,
      "loss": 0.4064,
      "step": 16414
    },
    {
      "epoch": 0.9191958785978273,
      "grad_norm": 1.035366415977478,
      "learning_rate": 8.203000000000001e-05,
      "loss": 0.3954,
      "step": 16415
    },
    {
      "epoch": 0.9192518759099563,
      "grad_norm": 1.2232056856155396,
      "learning_rate": 8.203500000000001e-05,
      "loss": 0.3527,
      "step": 16416
    },
    {
      "epoch": 0.9193078732220853,
      "grad_norm": 1.4133915901184082,
      "learning_rate": 8.204000000000001e-05,
      "loss": 0.4413,
      "step": 16417
    },
    {
      "epoch": 0.9193638705342143,
      "grad_norm": 1.235101580619812,
      "learning_rate": 8.2045e-05,
      "loss": 0.3635,
      "step": 16418
    },
    {
      "epoch": 0.9194198678463433,
      "grad_norm": 1.292309045791626,
      "learning_rate": 8.205e-05,
      "loss": 0.5364,
      "step": 16419
    },
    {
      "epoch": 0.9194758651584723,
      "grad_norm": 1.4182237386703491,
      "learning_rate": 8.2055e-05,
      "loss": 0.3983,
      "step": 16420
    },
    {
      "epoch": 0.9195318624706014,
      "grad_norm": 1.3632667064666748,
      "learning_rate": 8.206e-05,
      "loss": 0.4713,
      "step": 16421
    },
    {
      "epoch": 0.9195878597827304,
      "grad_norm": 1.3614498376846313,
      "learning_rate": 8.206500000000001e-05,
      "loss": 0.4299,
      "step": 16422
    },
    {
      "epoch": 0.9196438570948594,
      "grad_norm": 0.9320577383041382,
      "learning_rate": 8.207e-05,
      "loss": 0.2702,
      "step": 16423
    },
    {
      "epoch": 0.9196998544069884,
      "grad_norm": 1.6794301271438599,
      "learning_rate": 8.2075e-05,
      "loss": 0.4165,
      "step": 16424
    },
    {
      "epoch": 0.9197558517191174,
      "grad_norm": 1.3629662990570068,
      "learning_rate": 8.208e-05,
      "loss": 0.39,
      "step": 16425
    },
    {
      "epoch": 0.9198118490312465,
      "grad_norm": 1.3306039571762085,
      "learning_rate": 8.2085e-05,
      "loss": 0.3896,
      "step": 16426
    },
    {
      "epoch": 0.9198678463433755,
      "grad_norm": 1.297487497329712,
      "learning_rate": 8.209e-05,
      "loss": 0.4326,
      "step": 16427
    },
    {
      "epoch": 0.9199238436555045,
      "grad_norm": 1.2556846141815186,
      "learning_rate": 8.2095e-05,
      "loss": 0.4173,
      "step": 16428
    },
    {
      "epoch": 0.9199798409676335,
      "grad_norm": 1.4238240718841553,
      "learning_rate": 8.21e-05,
      "loss": 0.4721,
      "step": 16429
    },
    {
      "epoch": 0.9200358382797625,
      "grad_norm": 1.0150232315063477,
      "learning_rate": 8.2105e-05,
      "loss": 0.3388,
      "step": 16430
    },
    {
      "epoch": 0.9200918355918916,
      "grad_norm": 1.9842538833618164,
      "learning_rate": 8.211000000000001e-05,
      "loss": 0.5868,
      "step": 16431
    },
    {
      "epoch": 0.9201478329040206,
      "grad_norm": 1.5507006645202637,
      "learning_rate": 8.211500000000001e-05,
      "loss": 0.5215,
      "step": 16432
    },
    {
      "epoch": 0.9202038302161496,
      "grad_norm": 1.2344601154327393,
      "learning_rate": 8.212e-05,
      "loss": 0.5651,
      "step": 16433
    },
    {
      "epoch": 0.9202598275282786,
      "grad_norm": 2.3398501873016357,
      "learning_rate": 8.2125e-05,
      "loss": 0.5618,
      "step": 16434
    },
    {
      "epoch": 0.9203158248404076,
      "grad_norm": 1.16732919216156,
      "learning_rate": 8.213e-05,
      "loss": 0.5229,
      "step": 16435
    },
    {
      "epoch": 0.9203718221525367,
      "grad_norm": 1.2803624868392944,
      "learning_rate": 8.213500000000001e-05,
      "loss": 0.4516,
      "step": 16436
    },
    {
      "epoch": 0.9204278194646657,
      "grad_norm": 1.6683622598648071,
      "learning_rate": 8.214000000000001e-05,
      "loss": 0.5961,
      "step": 16437
    },
    {
      "epoch": 0.9204838167767947,
      "grad_norm": 1.0868639945983887,
      "learning_rate": 8.214500000000001e-05,
      "loss": 0.4601,
      "step": 16438
    },
    {
      "epoch": 0.9205398140889237,
      "grad_norm": 1.460677981376648,
      "learning_rate": 8.215e-05,
      "loss": 0.5205,
      "step": 16439
    },
    {
      "epoch": 0.9205958114010527,
      "grad_norm": 1.5984394550323486,
      "learning_rate": 8.2155e-05,
      "loss": 0.4798,
      "step": 16440
    },
    {
      "epoch": 0.9206518087131818,
      "grad_norm": 1.2280601263046265,
      "learning_rate": 8.216e-05,
      "loss": 0.5228,
      "step": 16441
    },
    {
      "epoch": 0.9207078060253108,
      "grad_norm": 1.8393840789794922,
      "learning_rate": 8.2165e-05,
      "loss": 0.4127,
      "step": 16442
    },
    {
      "epoch": 0.9207638033374398,
      "grad_norm": 1.3817163705825806,
      "learning_rate": 8.217000000000001e-05,
      "loss": 0.4368,
      "step": 16443
    },
    {
      "epoch": 0.9208198006495688,
      "grad_norm": 2.4562225341796875,
      "learning_rate": 8.2175e-05,
      "loss": 0.6111,
      "step": 16444
    },
    {
      "epoch": 0.9208757979616978,
      "grad_norm": 1.3783890008926392,
      "learning_rate": 8.218e-05,
      "loss": 0.3892,
      "step": 16445
    },
    {
      "epoch": 0.9209317952738268,
      "grad_norm": 1.3932751417160034,
      "learning_rate": 8.2185e-05,
      "loss": 0.4043,
      "step": 16446
    },
    {
      "epoch": 0.9209877925859559,
      "grad_norm": 1.3778951168060303,
      "learning_rate": 8.219e-05,
      "loss": 0.4582,
      "step": 16447
    },
    {
      "epoch": 0.9210437898980849,
      "grad_norm": 1.257495641708374,
      "learning_rate": 8.2195e-05,
      "loss": 0.4668,
      "step": 16448
    },
    {
      "epoch": 0.9210997872102139,
      "grad_norm": 1.1436564922332764,
      "learning_rate": 8.22e-05,
      "loss": 0.3684,
      "step": 16449
    },
    {
      "epoch": 0.9211557845223429,
      "grad_norm": 1.3091379404067993,
      "learning_rate": 8.2205e-05,
      "loss": 0.398,
      "step": 16450
    },
    {
      "epoch": 0.9212117818344719,
      "grad_norm": 1.2673567533493042,
      "learning_rate": 8.221000000000001e-05,
      "loss": 0.4482,
      "step": 16451
    },
    {
      "epoch": 0.921267779146601,
      "grad_norm": 1.4292402267456055,
      "learning_rate": 8.221500000000001e-05,
      "loss": 0.5005,
      "step": 16452
    },
    {
      "epoch": 0.92132377645873,
      "grad_norm": 1.2949180603027344,
      "learning_rate": 8.222000000000001e-05,
      "loss": 0.3964,
      "step": 16453
    },
    {
      "epoch": 0.921379773770859,
      "grad_norm": 1.2452545166015625,
      "learning_rate": 8.2225e-05,
      "loss": 0.4522,
      "step": 16454
    },
    {
      "epoch": 0.921435771082988,
      "grad_norm": 1.1206986904144287,
      "learning_rate": 8.223e-05,
      "loss": 0.3496,
      "step": 16455
    },
    {
      "epoch": 0.921491768395117,
      "grad_norm": 1.346388578414917,
      "learning_rate": 8.2235e-05,
      "loss": 0.405,
      "step": 16456
    },
    {
      "epoch": 0.9215477657072461,
      "grad_norm": 1.2031829357147217,
      "learning_rate": 8.224000000000001e-05,
      "loss": 0.3709,
      "step": 16457
    },
    {
      "epoch": 0.9216037630193751,
      "grad_norm": 1.5863611698150635,
      "learning_rate": 8.224500000000001e-05,
      "loss": 0.556,
      "step": 16458
    },
    {
      "epoch": 0.9216597603315041,
      "grad_norm": 1.2165085077285767,
      "learning_rate": 8.225000000000001e-05,
      "loss": 0.3987,
      "step": 16459
    },
    {
      "epoch": 0.9217157576436331,
      "grad_norm": 1.18646240234375,
      "learning_rate": 8.2255e-05,
      "loss": 0.4075,
      "step": 16460
    },
    {
      "epoch": 0.9217717549557621,
      "grad_norm": 1.255533218383789,
      "learning_rate": 8.226e-05,
      "loss": 0.3696,
      "step": 16461
    },
    {
      "epoch": 0.9218277522678912,
      "grad_norm": 1.4369428157806396,
      "learning_rate": 8.2265e-05,
      "loss": 0.5192,
      "step": 16462
    },
    {
      "epoch": 0.9218837495800202,
      "grad_norm": 1.2948023080825806,
      "learning_rate": 8.227e-05,
      "loss": 0.4968,
      "step": 16463
    },
    {
      "epoch": 0.9219397468921492,
      "grad_norm": 1.2731658220291138,
      "learning_rate": 8.227500000000001e-05,
      "loss": 0.455,
      "step": 16464
    },
    {
      "epoch": 0.9219957442042782,
      "grad_norm": 1.2603859901428223,
      "learning_rate": 8.228e-05,
      "loss": 0.3907,
      "step": 16465
    },
    {
      "epoch": 0.9220517415164072,
      "grad_norm": 1.4726279973983765,
      "learning_rate": 8.2285e-05,
      "loss": 0.4225,
      "step": 16466
    },
    {
      "epoch": 0.9221077388285362,
      "grad_norm": 1.221868634223938,
      "learning_rate": 8.229e-05,
      "loss": 0.5006,
      "step": 16467
    },
    {
      "epoch": 0.9221637361406653,
      "grad_norm": 1.2886098623275757,
      "learning_rate": 8.2295e-05,
      "loss": 0.507,
      "step": 16468
    },
    {
      "epoch": 0.9222197334527943,
      "grad_norm": 1.2557547092437744,
      "learning_rate": 8.23e-05,
      "loss": 0.4825,
      "step": 16469
    },
    {
      "epoch": 0.9222757307649233,
      "grad_norm": 1.2773897647857666,
      "learning_rate": 8.230499999999999e-05,
      "loss": 0.3528,
      "step": 16470
    },
    {
      "epoch": 0.9223317280770523,
      "grad_norm": 1.585194706916809,
      "learning_rate": 8.231e-05,
      "loss": 0.4966,
      "step": 16471
    },
    {
      "epoch": 0.9223877253891813,
      "grad_norm": 1.293116807937622,
      "learning_rate": 8.231500000000001e-05,
      "loss": 0.653,
      "step": 16472
    },
    {
      "epoch": 0.9224437227013104,
      "grad_norm": 1.1291966438293457,
      "learning_rate": 8.232000000000001e-05,
      "loss": 0.4101,
      "step": 16473
    },
    {
      "epoch": 0.9224997200134394,
      "grad_norm": 1.5031859874725342,
      "learning_rate": 8.232500000000001e-05,
      "loss": 0.6122,
      "step": 16474
    },
    {
      "epoch": 0.9225557173255684,
      "grad_norm": 1.3225125074386597,
      "learning_rate": 8.233e-05,
      "loss": 0.346,
      "step": 16475
    },
    {
      "epoch": 0.9226117146376974,
      "grad_norm": 1.2840086221694946,
      "learning_rate": 8.2335e-05,
      "loss": 0.395,
      "step": 16476
    },
    {
      "epoch": 0.9226677119498264,
      "grad_norm": 1.2742666006088257,
      "learning_rate": 8.234e-05,
      "loss": 0.4647,
      "step": 16477
    },
    {
      "epoch": 0.9227237092619555,
      "grad_norm": 1.4567025899887085,
      "learning_rate": 8.234500000000001e-05,
      "loss": 0.4579,
      "step": 16478
    },
    {
      "epoch": 0.9227797065740845,
      "grad_norm": 1.147629976272583,
      "learning_rate": 8.235000000000001e-05,
      "loss": 0.4129,
      "step": 16479
    },
    {
      "epoch": 0.9228357038862135,
      "grad_norm": 1.72379469871521,
      "learning_rate": 8.235500000000001e-05,
      "loss": 0.5756,
      "step": 16480
    },
    {
      "epoch": 0.9228917011983425,
      "grad_norm": 1.4656895399093628,
      "learning_rate": 8.236e-05,
      "loss": 0.4804,
      "step": 16481
    },
    {
      "epoch": 0.9229476985104715,
      "grad_norm": 1.4475468397140503,
      "learning_rate": 8.2365e-05,
      "loss": 0.4392,
      "step": 16482
    },
    {
      "epoch": 0.9230036958226006,
      "grad_norm": 1.415730595588684,
      "learning_rate": 8.237e-05,
      "loss": 0.451,
      "step": 16483
    },
    {
      "epoch": 0.9230596931347296,
      "grad_norm": 1.2868967056274414,
      "learning_rate": 8.2375e-05,
      "loss": 0.5013,
      "step": 16484
    },
    {
      "epoch": 0.9231156904468586,
      "grad_norm": 1.1820437908172607,
      "learning_rate": 8.238000000000001e-05,
      "loss": 0.4294,
      "step": 16485
    },
    {
      "epoch": 0.9231716877589876,
      "grad_norm": 1.4844865798950195,
      "learning_rate": 8.2385e-05,
      "loss": 0.5896,
      "step": 16486
    },
    {
      "epoch": 0.9232276850711166,
      "grad_norm": 1.2885370254516602,
      "learning_rate": 8.239e-05,
      "loss": 0.443,
      "step": 16487
    },
    {
      "epoch": 0.9232836823832457,
      "grad_norm": 1.252520203590393,
      "learning_rate": 8.2395e-05,
      "loss": 0.3869,
      "step": 16488
    },
    {
      "epoch": 0.9233396796953747,
      "grad_norm": 1.2477056980133057,
      "learning_rate": 8.24e-05,
      "loss": 0.5054,
      "step": 16489
    },
    {
      "epoch": 0.9233956770075037,
      "grad_norm": 1.156029462814331,
      "learning_rate": 8.2405e-05,
      "loss": 0.4791,
      "step": 16490
    },
    {
      "epoch": 0.9234516743196327,
      "grad_norm": 1.2756543159484863,
      "learning_rate": 8.241e-05,
      "loss": 0.4116,
      "step": 16491
    },
    {
      "epoch": 0.9235076716317617,
      "grad_norm": 1.3437877893447876,
      "learning_rate": 8.2415e-05,
      "loss": 0.4112,
      "step": 16492
    },
    {
      "epoch": 0.9235636689438907,
      "grad_norm": 4.821994304656982,
      "learning_rate": 8.242000000000001e-05,
      "loss": 0.4393,
      "step": 16493
    },
    {
      "epoch": 0.9236196662560197,
      "grad_norm": 4.62672758102417,
      "learning_rate": 8.242500000000001e-05,
      "loss": 0.4004,
      "step": 16494
    },
    {
      "epoch": 0.9236756635681487,
      "grad_norm": 1.2945533990859985,
      "learning_rate": 8.243000000000001e-05,
      "loss": 0.3963,
      "step": 16495
    },
    {
      "epoch": 0.9237316608802777,
      "grad_norm": 1.3269444704055786,
      "learning_rate": 8.2435e-05,
      "loss": 0.343,
      "step": 16496
    },
    {
      "epoch": 0.9237876581924067,
      "grad_norm": 1.295454740524292,
      "learning_rate": 8.244e-05,
      "loss": 0.4811,
      "step": 16497
    },
    {
      "epoch": 0.9238436555045357,
      "grad_norm": 1.302964448928833,
      "learning_rate": 8.2445e-05,
      "loss": 0.5828,
      "step": 16498
    },
    {
      "epoch": 0.9238996528166648,
      "grad_norm": 1.400241494178772,
      "learning_rate": 8.245e-05,
      "loss": 0.5116,
      "step": 16499
    },
    {
      "epoch": 0.9239556501287938,
      "grad_norm": 1.2892831563949585,
      "learning_rate": 8.245500000000001e-05,
      "loss": 0.4536,
      "step": 16500
    },
    {
      "epoch": 0.9240116474409228,
      "grad_norm": 1.1384472846984863,
      "learning_rate": 8.246e-05,
      "loss": 0.3839,
      "step": 16501
    },
    {
      "epoch": 0.9240676447530518,
      "grad_norm": 1.221259355545044,
      "learning_rate": 8.2465e-05,
      "loss": 0.3242,
      "step": 16502
    },
    {
      "epoch": 0.9241236420651808,
      "grad_norm": 1.291494607925415,
      "learning_rate": 8.247e-05,
      "loss": 0.4998,
      "step": 16503
    },
    {
      "epoch": 0.9241796393773098,
      "grad_norm": 1.2921441793441772,
      "learning_rate": 8.2475e-05,
      "loss": 0.4613,
      "step": 16504
    },
    {
      "epoch": 0.9242356366894389,
      "grad_norm": 1.4710813760757446,
      "learning_rate": 8.248e-05,
      "loss": 0.369,
      "step": 16505
    },
    {
      "epoch": 0.9242916340015679,
      "grad_norm": 1.471215009689331,
      "learning_rate": 8.248500000000001e-05,
      "loss": 0.4822,
      "step": 16506
    },
    {
      "epoch": 0.9243476313136969,
      "grad_norm": 1.4474198818206787,
      "learning_rate": 8.249e-05,
      "loss": 0.5187,
      "step": 16507
    },
    {
      "epoch": 0.9244036286258259,
      "grad_norm": 1.4380496740341187,
      "learning_rate": 8.2495e-05,
      "loss": 0.5645,
      "step": 16508
    },
    {
      "epoch": 0.9244596259379549,
      "grad_norm": 1.4628536701202393,
      "learning_rate": 8.25e-05,
      "loss": 0.4177,
      "step": 16509
    },
    {
      "epoch": 0.924515623250084,
      "grad_norm": 1.2975531816482544,
      "learning_rate": 8.2505e-05,
      "loss": 0.513,
      "step": 16510
    },
    {
      "epoch": 0.924571620562213,
      "grad_norm": 1.196489930152893,
      "learning_rate": 8.251e-05,
      "loss": 0.4424,
      "step": 16511
    },
    {
      "epoch": 0.924627617874342,
      "grad_norm": 1.557321548461914,
      "learning_rate": 8.2515e-05,
      "loss": 0.5826,
      "step": 16512
    },
    {
      "epoch": 0.924683615186471,
      "grad_norm": 1.4021459817886353,
      "learning_rate": 8.252e-05,
      "loss": 0.681,
      "step": 16513
    },
    {
      "epoch": 0.9247396124986,
      "grad_norm": 1.1830254793167114,
      "learning_rate": 8.252500000000001e-05,
      "loss": 0.3778,
      "step": 16514
    },
    {
      "epoch": 0.9247956098107291,
      "grad_norm": 1.2348469495773315,
      "learning_rate": 8.253000000000001e-05,
      "loss": 0.393,
      "step": 16515
    },
    {
      "epoch": 0.9248516071228581,
      "grad_norm": 1.3843134641647339,
      "learning_rate": 8.253500000000001e-05,
      "loss": 0.4101,
      "step": 16516
    },
    {
      "epoch": 0.9249076044349871,
      "grad_norm": 1.307335376739502,
      "learning_rate": 8.254e-05,
      "loss": 0.3418,
      "step": 16517
    },
    {
      "epoch": 0.9249636017471161,
      "grad_norm": 1.4313379526138306,
      "learning_rate": 8.2545e-05,
      "loss": 0.4476,
      "step": 16518
    },
    {
      "epoch": 0.9250195990592451,
      "grad_norm": 1.4691137075424194,
      "learning_rate": 8.255e-05,
      "loss": 0.457,
      "step": 16519
    },
    {
      "epoch": 0.9250755963713742,
      "grad_norm": 3.8030176162719727,
      "learning_rate": 8.2555e-05,
      "loss": 0.4801,
      "step": 16520
    },
    {
      "epoch": 0.9251315936835032,
      "grad_norm": 1.2685890197753906,
      "learning_rate": 8.256000000000001e-05,
      "loss": 0.4545,
      "step": 16521
    },
    {
      "epoch": 0.9251875909956322,
      "grad_norm": 1.1432857513427734,
      "learning_rate": 8.2565e-05,
      "loss": 0.3794,
      "step": 16522
    },
    {
      "epoch": 0.9252435883077612,
      "grad_norm": 1.3015574216842651,
      "learning_rate": 8.257e-05,
      "loss": 0.3819,
      "step": 16523
    },
    {
      "epoch": 0.9252995856198902,
      "grad_norm": 1.5656224489212036,
      "learning_rate": 8.2575e-05,
      "loss": 0.5617,
      "step": 16524
    },
    {
      "epoch": 0.9253555829320192,
      "grad_norm": 1.4016870260238647,
      "learning_rate": 8.258e-05,
      "loss": 0.609,
      "step": 16525
    },
    {
      "epoch": 0.9254115802441483,
      "grad_norm": 1.5549887418746948,
      "learning_rate": 8.2585e-05,
      "loss": 0.4389,
      "step": 16526
    },
    {
      "epoch": 0.9254675775562773,
      "grad_norm": 1.205660343170166,
      "learning_rate": 8.259000000000001e-05,
      "loss": 0.4418,
      "step": 16527
    },
    {
      "epoch": 0.9255235748684063,
      "grad_norm": 1.5776665210723877,
      "learning_rate": 8.2595e-05,
      "loss": 0.5697,
      "step": 16528
    },
    {
      "epoch": 0.9255795721805353,
      "grad_norm": 1.3645782470703125,
      "learning_rate": 8.26e-05,
      "loss": 0.4751,
      "step": 16529
    },
    {
      "epoch": 0.9256355694926643,
      "grad_norm": 1.3706555366516113,
      "learning_rate": 8.2605e-05,
      "loss": 0.4326,
      "step": 16530
    },
    {
      "epoch": 0.9256915668047934,
      "grad_norm": 1.230024814605713,
      "learning_rate": 8.261e-05,
      "loss": 0.3903,
      "step": 16531
    },
    {
      "epoch": 0.9257475641169224,
      "grad_norm": 1.4342620372772217,
      "learning_rate": 8.261500000000001e-05,
      "loss": 0.7193,
      "step": 16532
    },
    {
      "epoch": 0.9258035614290514,
      "grad_norm": 1.3708159923553467,
      "learning_rate": 8.262e-05,
      "loss": 0.4468,
      "step": 16533
    },
    {
      "epoch": 0.9258595587411804,
      "grad_norm": 1.3850219249725342,
      "learning_rate": 8.2625e-05,
      "loss": 0.5445,
      "step": 16534
    },
    {
      "epoch": 0.9259155560533094,
      "grad_norm": 1.1610344648361206,
      "learning_rate": 8.263000000000001e-05,
      "loss": 0.3923,
      "step": 16535
    },
    {
      "epoch": 0.9259715533654385,
      "grad_norm": 1.4086459875106812,
      "learning_rate": 8.263500000000001e-05,
      "loss": 0.5273,
      "step": 16536
    },
    {
      "epoch": 0.9260275506775675,
      "grad_norm": 1.4805762767791748,
      "learning_rate": 8.264000000000001e-05,
      "loss": 0.4365,
      "step": 16537
    },
    {
      "epoch": 0.9260835479896965,
      "grad_norm": 1.6059681177139282,
      "learning_rate": 8.2645e-05,
      "loss": 0.5667,
      "step": 16538
    },
    {
      "epoch": 0.9261395453018255,
      "grad_norm": 1.3339136838912964,
      "learning_rate": 8.265e-05,
      "loss": 0.3483,
      "step": 16539
    },
    {
      "epoch": 0.9261955426139545,
      "grad_norm": 1.4039710760116577,
      "learning_rate": 8.2655e-05,
      "loss": 0.3865,
      "step": 16540
    },
    {
      "epoch": 0.9262515399260836,
      "grad_norm": 1.1990165710449219,
      "learning_rate": 8.266e-05,
      "loss": 0.3419,
      "step": 16541
    },
    {
      "epoch": 0.9263075372382126,
      "grad_norm": 1.5688766241073608,
      "learning_rate": 8.266500000000001e-05,
      "loss": 0.5469,
      "step": 16542
    },
    {
      "epoch": 0.9263635345503416,
      "grad_norm": 1.3112396001815796,
      "learning_rate": 8.267e-05,
      "loss": 0.4534,
      "step": 16543
    },
    {
      "epoch": 0.9264195318624706,
      "grad_norm": 1.5226678848266602,
      "learning_rate": 8.2675e-05,
      "loss": 0.5196,
      "step": 16544
    },
    {
      "epoch": 0.9264755291745996,
      "grad_norm": 1.2306163311004639,
      "learning_rate": 8.268e-05,
      "loss": 0.4405,
      "step": 16545
    },
    {
      "epoch": 0.9265315264867287,
      "grad_norm": 1.2804595232009888,
      "learning_rate": 8.2685e-05,
      "loss": 0.4288,
      "step": 16546
    },
    {
      "epoch": 0.9265875237988577,
      "grad_norm": 1.1443030834197998,
      "learning_rate": 8.269e-05,
      "loss": 0.4405,
      "step": 16547
    },
    {
      "epoch": 0.9266435211109867,
      "grad_norm": 1.2803187370300293,
      "learning_rate": 8.269500000000001e-05,
      "loss": 0.4594,
      "step": 16548
    },
    {
      "epoch": 0.9266995184231157,
      "grad_norm": 1.303547978401184,
      "learning_rate": 8.27e-05,
      "loss": 0.3836,
      "step": 16549
    },
    {
      "epoch": 0.9267555157352447,
      "grad_norm": 1.2507973909378052,
      "learning_rate": 8.2705e-05,
      "loss": 0.5085,
      "step": 16550
    },
    {
      "epoch": 0.9268115130473737,
      "grad_norm": 1.3687591552734375,
      "learning_rate": 8.271e-05,
      "loss": 0.539,
      "step": 16551
    },
    {
      "epoch": 0.9268675103595028,
      "grad_norm": 1.1975895166397095,
      "learning_rate": 8.271500000000001e-05,
      "loss": 0.4621,
      "step": 16552
    },
    {
      "epoch": 0.9269235076716318,
      "grad_norm": 1.4076459407806396,
      "learning_rate": 8.272000000000001e-05,
      "loss": 0.4201,
      "step": 16553
    },
    {
      "epoch": 0.9269795049837608,
      "grad_norm": 1.4311432838439941,
      "learning_rate": 8.2725e-05,
      "loss": 0.5233,
      "step": 16554
    },
    {
      "epoch": 0.9270355022958898,
      "grad_norm": 1.3195104598999023,
      "learning_rate": 8.273e-05,
      "loss": 0.4051,
      "step": 16555
    },
    {
      "epoch": 0.9270914996080188,
      "grad_norm": 1.2419683933258057,
      "learning_rate": 8.273500000000001e-05,
      "loss": 0.4731,
      "step": 16556
    },
    {
      "epoch": 0.9271474969201479,
      "grad_norm": 1.3806172609329224,
      "learning_rate": 8.274000000000001e-05,
      "loss": 0.5336,
      "step": 16557
    },
    {
      "epoch": 0.9272034942322769,
      "grad_norm": 1.577425241470337,
      "learning_rate": 8.274500000000001e-05,
      "loss": 0.43,
      "step": 16558
    },
    {
      "epoch": 0.9272594915444059,
      "grad_norm": 1.5444074869155884,
      "learning_rate": 8.275e-05,
      "loss": 0.4589,
      "step": 16559
    },
    {
      "epoch": 0.9273154888565349,
      "grad_norm": 1.5758018493652344,
      "learning_rate": 8.2755e-05,
      "loss": 0.5047,
      "step": 16560
    },
    {
      "epoch": 0.9273714861686639,
      "grad_norm": 1.2698769569396973,
      "learning_rate": 8.276e-05,
      "loss": 0.4111,
      "step": 16561
    },
    {
      "epoch": 0.927427483480793,
      "grad_norm": 1.5544837713241577,
      "learning_rate": 8.2765e-05,
      "loss": 0.4808,
      "step": 16562
    },
    {
      "epoch": 0.927483480792922,
      "grad_norm": 1.2893900871276855,
      "learning_rate": 8.277000000000001e-05,
      "loss": 0.4719,
      "step": 16563
    },
    {
      "epoch": 0.927539478105051,
      "grad_norm": 1.3675788640975952,
      "learning_rate": 8.2775e-05,
      "loss": 0.403,
      "step": 16564
    },
    {
      "epoch": 0.92759547541718,
      "grad_norm": 1.1784521341323853,
      "learning_rate": 8.278e-05,
      "loss": 0.4722,
      "step": 16565
    },
    {
      "epoch": 0.927651472729309,
      "grad_norm": 1.1031605005264282,
      "learning_rate": 8.2785e-05,
      "loss": 0.4225,
      "step": 16566
    },
    {
      "epoch": 0.927707470041438,
      "grad_norm": 1.4057633876800537,
      "learning_rate": 8.279e-05,
      "loss": 0.4327,
      "step": 16567
    },
    {
      "epoch": 0.9277634673535671,
      "grad_norm": 1.7156649827957153,
      "learning_rate": 8.2795e-05,
      "loss": 0.7462,
      "step": 16568
    },
    {
      "epoch": 0.9278194646656961,
      "grad_norm": 1.4485633373260498,
      "learning_rate": 8.28e-05,
      "loss": 0.4922,
      "step": 16569
    },
    {
      "epoch": 0.9278754619778251,
      "grad_norm": 1.3397701978683472,
      "learning_rate": 8.2805e-05,
      "loss": 0.499,
      "step": 16570
    },
    {
      "epoch": 0.9279314592899541,
      "grad_norm": 1.3101890087127686,
      "learning_rate": 8.281e-05,
      "loss": 0.4668,
      "step": 16571
    },
    {
      "epoch": 0.9279874566020831,
      "grad_norm": 1.1900054216384888,
      "learning_rate": 8.281500000000001e-05,
      "loss": 0.4898,
      "step": 16572
    },
    {
      "epoch": 0.9280434539142122,
      "grad_norm": 1.3926546573638916,
      "learning_rate": 8.282000000000001e-05,
      "loss": 0.5736,
      "step": 16573
    },
    {
      "epoch": 0.9280994512263412,
      "grad_norm": 1.7720388174057007,
      "learning_rate": 8.282500000000001e-05,
      "loss": 0.4774,
      "step": 16574
    },
    {
      "epoch": 0.9281554485384702,
      "grad_norm": 1.2900577783584595,
      "learning_rate": 8.283e-05,
      "loss": 0.3573,
      "step": 16575
    },
    {
      "epoch": 0.9282114458505992,
      "grad_norm": 1.13142991065979,
      "learning_rate": 8.2835e-05,
      "loss": 0.3229,
      "step": 16576
    },
    {
      "epoch": 0.9282674431627281,
      "grad_norm": 1.5441309213638306,
      "learning_rate": 8.284000000000001e-05,
      "loss": 0.5427,
      "step": 16577
    },
    {
      "epoch": 0.9283234404748572,
      "grad_norm": 1.2781037092208862,
      "learning_rate": 8.284500000000001e-05,
      "loss": 0.408,
      "step": 16578
    },
    {
      "epoch": 0.9283794377869862,
      "grad_norm": 1.2270132303237915,
      "learning_rate": 8.285000000000001e-05,
      "loss": 0.3863,
      "step": 16579
    },
    {
      "epoch": 0.9284354350991152,
      "grad_norm": 1.265037178993225,
      "learning_rate": 8.2855e-05,
      "loss": 0.4642,
      "step": 16580
    },
    {
      "epoch": 0.9284914324112442,
      "grad_norm": 1.491784930229187,
      "learning_rate": 8.286e-05,
      "loss": 0.5653,
      "step": 16581
    },
    {
      "epoch": 0.9285474297233732,
      "grad_norm": 1.302323341369629,
      "learning_rate": 8.2865e-05,
      "loss": 0.3473,
      "step": 16582
    },
    {
      "epoch": 0.9286034270355022,
      "grad_norm": 1.445154070854187,
      "learning_rate": 8.287e-05,
      "loss": 0.4246,
      "step": 16583
    },
    {
      "epoch": 0.9286594243476313,
      "grad_norm": 1.2335448265075684,
      "learning_rate": 8.287500000000001e-05,
      "loss": 0.4162,
      "step": 16584
    },
    {
      "epoch": 0.9287154216597603,
      "grad_norm": 1.3359366655349731,
      "learning_rate": 8.288e-05,
      "loss": 0.448,
      "step": 16585
    },
    {
      "epoch": 0.9287714189718893,
      "grad_norm": 1.3777626752853394,
      "learning_rate": 8.2885e-05,
      "loss": 0.3824,
      "step": 16586
    },
    {
      "epoch": 0.9288274162840183,
      "grad_norm": 1.393613338470459,
      "learning_rate": 8.289e-05,
      "loss": 0.5083,
      "step": 16587
    },
    {
      "epoch": 0.9288834135961473,
      "grad_norm": 1.359660029411316,
      "learning_rate": 8.2895e-05,
      "loss": 0.4547,
      "step": 16588
    },
    {
      "epoch": 0.9289394109082764,
      "grad_norm": 1.5443047285079956,
      "learning_rate": 8.29e-05,
      "loss": 0.3964,
      "step": 16589
    },
    {
      "epoch": 0.9289954082204054,
      "grad_norm": 1.789296269416809,
      "learning_rate": 8.290499999999999e-05,
      "loss": 0.5635,
      "step": 16590
    },
    {
      "epoch": 0.9290514055325344,
      "grad_norm": 1.4631083011627197,
      "learning_rate": 8.291e-05,
      "loss": 0.359,
      "step": 16591
    },
    {
      "epoch": 0.9291074028446634,
      "grad_norm": 1.3464879989624023,
      "learning_rate": 8.291500000000002e-05,
      "loss": 0.5024,
      "step": 16592
    },
    {
      "epoch": 0.9291634001567924,
      "grad_norm": 1.7417012453079224,
      "learning_rate": 8.292000000000001e-05,
      "loss": 0.5396,
      "step": 16593
    },
    {
      "epoch": 0.9292193974689215,
      "grad_norm": 1.5067992210388184,
      "learning_rate": 8.292500000000001e-05,
      "loss": 0.4364,
      "step": 16594
    },
    {
      "epoch": 0.9292753947810505,
      "grad_norm": 1.6404314041137695,
      "learning_rate": 8.293000000000001e-05,
      "loss": 0.5575,
      "step": 16595
    },
    {
      "epoch": 0.9293313920931795,
      "grad_norm": 1.6139005422592163,
      "learning_rate": 8.2935e-05,
      "loss": 0.6856,
      "step": 16596
    },
    {
      "epoch": 0.9293873894053085,
      "grad_norm": 3.3078603744506836,
      "learning_rate": 8.294e-05,
      "loss": 0.4122,
      "step": 16597
    },
    {
      "epoch": 0.9294433867174375,
      "grad_norm": 2.09658145904541,
      "learning_rate": 8.2945e-05,
      "loss": 0.5927,
      "step": 16598
    },
    {
      "epoch": 0.9294993840295666,
      "grad_norm": 1.4565671682357788,
      "learning_rate": 8.295000000000001e-05,
      "loss": 0.5263,
      "step": 16599
    },
    {
      "epoch": 0.9295553813416956,
      "grad_norm": 1.625476360321045,
      "learning_rate": 8.295500000000001e-05,
      "loss": 0.6797,
      "step": 16600
    },
    {
      "epoch": 0.9296113786538246,
      "grad_norm": 1.2399108409881592,
      "learning_rate": 8.296e-05,
      "loss": 0.4101,
      "step": 16601
    },
    {
      "epoch": 0.9296673759659536,
      "grad_norm": 1.4441423416137695,
      "learning_rate": 8.2965e-05,
      "loss": 0.4181,
      "step": 16602
    },
    {
      "epoch": 0.9297233732780826,
      "grad_norm": 1.3000669479370117,
      "learning_rate": 8.297e-05,
      "loss": 0.3524,
      "step": 16603
    },
    {
      "epoch": 0.9297793705902117,
      "grad_norm": 1.5105751752853394,
      "learning_rate": 8.2975e-05,
      "loss": 0.5565,
      "step": 16604
    },
    {
      "epoch": 0.9298353679023407,
      "grad_norm": 1.344003677368164,
      "learning_rate": 8.298000000000001e-05,
      "loss": 0.4283,
      "step": 16605
    },
    {
      "epoch": 0.9298913652144697,
      "grad_norm": 1.0473260879516602,
      "learning_rate": 8.2985e-05,
      "loss": 0.3195,
      "step": 16606
    },
    {
      "epoch": 0.9299473625265987,
      "grad_norm": 8.235751152038574,
      "learning_rate": 8.299e-05,
      "loss": 0.3389,
      "step": 16607
    },
    {
      "epoch": 0.9300033598387277,
      "grad_norm": 1.305876612663269,
      "learning_rate": 8.2995e-05,
      "loss": 0.4715,
      "step": 16608
    },
    {
      "epoch": 0.9300593571508567,
      "grad_norm": 1.1514440774917603,
      "learning_rate": 8.3e-05,
      "loss": 0.4138,
      "step": 16609
    },
    {
      "epoch": 0.9301153544629858,
      "grad_norm": 1.5276670455932617,
      "learning_rate": 8.3005e-05,
      "loss": 0.4012,
      "step": 16610
    },
    {
      "epoch": 0.9301713517751148,
      "grad_norm": 1.2899521589279175,
      "learning_rate": 8.300999999999999e-05,
      "loss": 0.4227,
      "step": 16611
    },
    {
      "epoch": 0.9302273490872438,
      "grad_norm": 1.303055763244629,
      "learning_rate": 8.3015e-05,
      "loss": 0.5667,
      "step": 16612
    },
    {
      "epoch": 0.9302833463993728,
      "grad_norm": 1.1715697050094604,
      "learning_rate": 8.302000000000001e-05,
      "loss": 0.4125,
      "step": 16613
    },
    {
      "epoch": 0.9303393437115018,
      "grad_norm": 1.5779167413711548,
      "learning_rate": 8.302500000000001e-05,
      "loss": 0.5931,
      "step": 16614
    },
    {
      "epoch": 0.9303953410236309,
      "grad_norm": 1.1428412199020386,
      "learning_rate": 8.303000000000001e-05,
      "loss": 0.3914,
      "step": 16615
    },
    {
      "epoch": 0.9304513383357599,
      "grad_norm": 1.3851007223129272,
      "learning_rate": 8.303500000000001e-05,
      "loss": 0.567,
      "step": 16616
    },
    {
      "epoch": 0.9305073356478889,
      "grad_norm": 1.3200896978378296,
      "learning_rate": 8.304e-05,
      "loss": 0.3061,
      "step": 16617
    },
    {
      "epoch": 0.9305633329600179,
      "grad_norm": 1.2225103378295898,
      "learning_rate": 8.3045e-05,
      "loss": 0.2826,
      "step": 16618
    },
    {
      "epoch": 0.9306193302721469,
      "grad_norm": 1.1365717649459839,
      "learning_rate": 8.305e-05,
      "loss": 0.3784,
      "step": 16619
    },
    {
      "epoch": 0.930675327584276,
      "grad_norm": 1.8045048713684082,
      "learning_rate": 8.305500000000001e-05,
      "loss": 0.6675,
      "step": 16620
    },
    {
      "epoch": 0.930731324896405,
      "grad_norm": 1.4891531467437744,
      "learning_rate": 8.306000000000001e-05,
      "loss": 0.4465,
      "step": 16621
    },
    {
      "epoch": 0.930787322208534,
      "grad_norm": 1.1786224842071533,
      "learning_rate": 8.3065e-05,
      "loss": 0.4102,
      "step": 16622
    },
    {
      "epoch": 0.930843319520663,
      "grad_norm": 1.4224238395690918,
      "learning_rate": 8.307e-05,
      "loss": 0.4362,
      "step": 16623
    },
    {
      "epoch": 0.930899316832792,
      "grad_norm": 1.1878151893615723,
      "learning_rate": 8.3075e-05,
      "loss": 0.4253,
      "step": 16624
    },
    {
      "epoch": 0.930955314144921,
      "grad_norm": 1.362742304801941,
      "learning_rate": 8.308e-05,
      "loss": 0.4014,
      "step": 16625
    },
    {
      "epoch": 0.9310113114570501,
      "grad_norm": 1.4134865999221802,
      "learning_rate": 8.308500000000001e-05,
      "loss": 0.5373,
      "step": 16626
    },
    {
      "epoch": 0.9310673087691791,
      "grad_norm": 1.3262542486190796,
      "learning_rate": 8.309e-05,
      "loss": 0.3624,
      "step": 16627
    },
    {
      "epoch": 0.9311233060813081,
      "grad_norm": 1.2941330671310425,
      "learning_rate": 8.3095e-05,
      "loss": 0.4494,
      "step": 16628
    },
    {
      "epoch": 0.9311793033934371,
      "grad_norm": 1.194333553314209,
      "learning_rate": 8.31e-05,
      "loss": 0.4017,
      "step": 16629
    },
    {
      "epoch": 0.9312353007055661,
      "grad_norm": 1.6164854764938354,
      "learning_rate": 8.3105e-05,
      "loss": 0.5246,
      "step": 16630
    },
    {
      "epoch": 0.9312912980176952,
      "grad_norm": 1.2223023176193237,
      "learning_rate": 8.311e-05,
      "loss": 0.4499,
      "step": 16631
    },
    {
      "epoch": 0.9313472953298242,
      "grad_norm": 1.225380539894104,
      "learning_rate": 8.3115e-05,
      "loss": 0.446,
      "step": 16632
    },
    {
      "epoch": 0.9314032926419532,
      "grad_norm": 1.4562822580337524,
      "learning_rate": 8.312e-05,
      "loss": 0.537,
      "step": 16633
    },
    {
      "epoch": 0.9314592899540822,
      "grad_norm": 1.2595573663711548,
      "learning_rate": 8.312500000000001e-05,
      "loss": 0.4092,
      "step": 16634
    },
    {
      "epoch": 0.9315152872662112,
      "grad_norm": 1.570261001586914,
      "learning_rate": 8.313000000000001e-05,
      "loss": 0.4921,
      "step": 16635
    },
    {
      "epoch": 0.9315712845783403,
      "grad_norm": 1.8107719421386719,
      "learning_rate": 8.313500000000001e-05,
      "loss": 0.4667,
      "step": 16636
    },
    {
      "epoch": 0.9316272818904693,
      "grad_norm": 1.3004285097122192,
      "learning_rate": 8.314000000000001e-05,
      "loss": 0.61,
      "step": 16637
    },
    {
      "epoch": 0.9316832792025983,
      "grad_norm": 1.2579715251922607,
      "learning_rate": 8.3145e-05,
      "loss": 0.3779,
      "step": 16638
    },
    {
      "epoch": 0.9317392765147273,
      "grad_norm": 1.3704969882965088,
      "learning_rate": 8.315e-05,
      "loss": 0.4827,
      "step": 16639
    },
    {
      "epoch": 0.9317952738268563,
      "grad_norm": 1.4027427434921265,
      "learning_rate": 8.3155e-05,
      "loss": 0.3904,
      "step": 16640
    },
    {
      "epoch": 0.9318512711389854,
      "grad_norm": 1.5530561208724976,
      "learning_rate": 8.316000000000001e-05,
      "loss": 0.6129,
      "step": 16641
    },
    {
      "epoch": 0.9319072684511144,
      "grad_norm": 1.5409584045410156,
      "learning_rate": 8.316500000000001e-05,
      "loss": 0.575,
      "step": 16642
    },
    {
      "epoch": 0.9319632657632434,
      "grad_norm": 1.448671817779541,
      "learning_rate": 8.317e-05,
      "loss": 0.4654,
      "step": 16643
    },
    {
      "epoch": 0.9320192630753724,
      "grad_norm": 1.2616862058639526,
      "learning_rate": 8.3175e-05,
      "loss": 0.5618,
      "step": 16644
    },
    {
      "epoch": 0.9320752603875014,
      "grad_norm": 1.637648344039917,
      "learning_rate": 8.318e-05,
      "loss": 0.473,
      "step": 16645
    },
    {
      "epoch": 0.9321312576996305,
      "grad_norm": 1.5618937015533447,
      "learning_rate": 8.3185e-05,
      "loss": 0.4767,
      "step": 16646
    },
    {
      "epoch": 0.9321872550117595,
      "grad_norm": 1.2212520837783813,
      "learning_rate": 8.319e-05,
      "loss": 0.4776,
      "step": 16647
    },
    {
      "epoch": 0.9322432523238885,
      "grad_norm": 1.1408857107162476,
      "learning_rate": 8.3195e-05,
      "loss": 0.3799,
      "step": 16648
    },
    {
      "epoch": 0.9322992496360175,
      "grad_norm": 1.4778286218643188,
      "learning_rate": 8.32e-05,
      "loss": 0.38,
      "step": 16649
    },
    {
      "epoch": 0.9323552469481465,
      "grad_norm": 1.2437379360198975,
      "learning_rate": 8.3205e-05,
      "loss": 0.4526,
      "step": 16650
    },
    {
      "epoch": 0.9324112442602756,
      "grad_norm": 5.390810012817383,
      "learning_rate": 8.321e-05,
      "loss": 0.3792,
      "step": 16651
    },
    {
      "epoch": 0.9324672415724046,
      "grad_norm": 1.3441483974456787,
      "learning_rate": 8.3215e-05,
      "loss": 0.4469,
      "step": 16652
    },
    {
      "epoch": 0.9325232388845336,
      "grad_norm": 1.81027090549469,
      "learning_rate": 8.322e-05,
      "loss": 0.4594,
      "step": 16653
    },
    {
      "epoch": 0.9325792361966626,
      "grad_norm": 1.5646641254425049,
      "learning_rate": 8.3225e-05,
      "loss": 0.5921,
      "step": 16654
    },
    {
      "epoch": 0.9326352335087916,
      "grad_norm": 1.2193459272384644,
      "learning_rate": 8.323000000000001e-05,
      "loss": 0.3902,
      "step": 16655
    },
    {
      "epoch": 0.9326912308209206,
      "grad_norm": 1.1926816701889038,
      "learning_rate": 8.323500000000001e-05,
      "loss": 0.4753,
      "step": 16656
    },
    {
      "epoch": 0.9327472281330497,
      "grad_norm": 1.4411406517028809,
      "learning_rate": 8.324000000000001e-05,
      "loss": 0.4647,
      "step": 16657
    },
    {
      "epoch": 0.9328032254451787,
      "grad_norm": 1.874061107635498,
      "learning_rate": 8.324500000000001e-05,
      "loss": 0.5395,
      "step": 16658
    },
    {
      "epoch": 0.9328592227573077,
      "grad_norm": 1.3584909439086914,
      "learning_rate": 8.325e-05,
      "loss": 0.4404,
      "step": 16659
    },
    {
      "epoch": 0.9329152200694366,
      "grad_norm": 1.2228422164916992,
      "learning_rate": 8.3255e-05,
      "loss": 0.2974,
      "step": 16660
    },
    {
      "epoch": 0.9329712173815656,
      "grad_norm": 1.4384647607803345,
      "learning_rate": 8.326e-05,
      "loss": 0.4368,
      "step": 16661
    },
    {
      "epoch": 0.9330272146936947,
      "grad_norm": 1.2427054643630981,
      "learning_rate": 8.326500000000001e-05,
      "loss": 0.4453,
      "step": 16662
    },
    {
      "epoch": 0.9330832120058237,
      "grad_norm": 1.569198489189148,
      "learning_rate": 8.327000000000001e-05,
      "loss": 0.5041,
      "step": 16663
    },
    {
      "epoch": 0.9331392093179527,
      "grad_norm": 1.2273691892623901,
      "learning_rate": 8.3275e-05,
      "loss": 0.4874,
      "step": 16664
    },
    {
      "epoch": 0.9331952066300817,
      "grad_norm": 1.4037173986434937,
      "learning_rate": 8.328e-05,
      "loss": 0.5313,
      "step": 16665
    },
    {
      "epoch": 0.9332512039422107,
      "grad_norm": 1.7003096342086792,
      "learning_rate": 8.3285e-05,
      "loss": 0.4736,
      "step": 16666
    },
    {
      "epoch": 0.9333072012543397,
      "grad_norm": 1.7694745063781738,
      "learning_rate": 8.329e-05,
      "loss": 0.5357,
      "step": 16667
    },
    {
      "epoch": 0.9333631985664688,
      "grad_norm": 1.3882302045822144,
      "learning_rate": 8.3295e-05,
      "loss": 0.3981,
      "step": 16668
    },
    {
      "epoch": 0.9334191958785978,
      "grad_norm": 1.2965500354766846,
      "learning_rate": 8.33e-05,
      "loss": 0.4082,
      "step": 16669
    },
    {
      "epoch": 0.9334751931907268,
      "grad_norm": 1.3104054927825928,
      "learning_rate": 8.3305e-05,
      "loss": 0.4348,
      "step": 16670
    },
    {
      "epoch": 0.9335311905028558,
      "grad_norm": 1.2672455310821533,
      "learning_rate": 8.331e-05,
      "loss": 0.3781,
      "step": 16671
    },
    {
      "epoch": 0.9335871878149848,
      "grad_norm": 1.2638646364212036,
      "learning_rate": 8.3315e-05,
      "loss": 0.5076,
      "step": 16672
    },
    {
      "epoch": 0.9336431851271139,
      "grad_norm": 1.0242950916290283,
      "learning_rate": 8.332000000000001e-05,
      "loss": 0.5022,
      "step": 16673
    },
    {
      "epoch": 0.9336991824392429,
      "grad_norm": 1.3578529357910156,
      "learning_rate": 8.3325e-05,
      "loss": 0.5606,
      "step": 16674
    },
    {
      "epoch": 0.9337551797513719,
      "grad_norm": 1.5048909187316895,
      "learning_rate": 8.333e-05,
      "loss": 0.368,
      "step": 16675
    },
    {
      "epoch": 0.9338111770635009,
      "grad_norm": 1.1758440732955933,
      "learning_rate": 8.3335e-05,
      "loss": 0.372,
      "step": 16676
    },
    {
      "epoch": 0.9338671743756299,
      "grad_norm": 1.7584550380706787,
      "learning_rate": 8.334000000000001e-05,
      "loss": 0.6781,
      "step": 16677
    },
    {
      "epoch": 0.933923171687759,
      "grad_norm": 1.4949352741241455,
      "learning_rate": 8.334500000000001e-05,
      "loss": 0.5093,
      "step": 16678
    },
    {
      "epoch": 0.933979168999888,
      "grad_norm": 1.2756378650665283,
      "learning_rate": 8.335e-05,
      "loss": 0.4561,
      "step": 16679
    },
    {
      "epoch": 0.934035166312017,
      "grad_norm": 1.7087390422821045,
      "learning_rate": 8.3355e-05,
      "loss": 0.3433,
      "step": 16680
    },
    {
      "epoch": 0.934091163624146,
      "grad_norm": 1.3154563903808594,
      "learning_rate": 8.336e-05,
      "loss": 0.5667,
      "step": 16681
    },
    {
      "epoch": 0.934147160936275,
      "grad_norm": 1.68277108669281,
      "learning_rate": 8.3365e-05,
      "loss": 0.4285,
      "step": 16682
    },
    {
      "epoch": 0.934203158248404,
      "grad_norm": 1.7122166156768799,
      "learning_rate": 8.337000000000001e-05,
      "loss": 0.5155,
      "step": 16683
    },
    {
      "epoch": 0.9342591555605331,
      "grad_norm": 1.2633670568466187,
      "learning_rate": 8.337500000000001e-05,
      "loss": 0.434,
      "step": 16684
    },
    {
      "epoch": 0.9343151528726621,
      "grad_norm": 1.2328050136566162,
      "learning_rate": 8.338e-05,
      "loss": 0.3519,
      "step": 16685
    },
    {
      "epoch": 0.9343711501847911,
      "grad_norm": 1.4168370962142944,
      "learning_rate": 8.3385e-05,
      "loss": 0.4409,
      "step": 16686
    },
    {
      "epoch": 0.9344271474969201,
      "grad_norm": 1.3801827430725098,
      "learning_rate": 8.339e-05,
      "loss": 0.4707,
      "step": 16687
    },
    {
      "epoch": 0.9344831448090491,
      "grad_norm": 1.3615226745605469,
      "learning_rate": 8.3395e-05,
      "loss": 0.5223,
      "step": 16688
    },
    {
      "epoch": 0.9345391421211782,
      "grad_norm": 1.6777215003967285,
      "learning_rate": 8.34e-05,
      "loss": 0.6625,
      "step": 16689
    },
    {
      "epoch": 0.9345951394333072,
      "grad_norm": 1.1215518712997437,
      "learning_rate": 8.3405e-05,
      "loss": 0.4153,
      "step": 16690
    },
    {
      "epoch": 0.9346511367454362,
      "grad_norm": 1.509817361831665,
      "learning_rate": 8.341e-05,
      "loss": 0.5476,
      "step": 16691
    },
    {
      "epoch": 0.9347071340575652,
      "grad_norm": 1.4509403705596924,
      "learning_rate": 8.3415e-05,
      "loss": 0.44,
      "step": 16692
    },
    {
      "epoch": 0.9347631313696942,
      "grad_norm": 1.4264498949050903,
      "learning_rate": 8.342000000000001e-05,
      "loss": 0.5042,
      "step": 16693
    },
    {
      "epoch": 0.9348191286818233,
      "grad_norm": 1.282489538192749,
      "learning_rate": 8.342500000000001e-05,
      "loss": 0.4117,
      "step": 16694
    },
    {
      "epoch": 0.9348751259939523,
      "grad_norm": 1.2684929370880127,
      "learning_rate": 8.343e-05,
      "loss": 0.528,
      "step": 16695
    },
    {
      "epoch": 0.9349311233060813,
      "grad_norm": 1.458742380142212,
      "learning_rate": 8.3435e-05,
      "loss": 0.409,
      "step": 16696
    },
    {
      "epoch": 0.9349871206182103,
      "grad_norm": 1.156498908996582,
      "learning_rate": 8.344e-05,
      "loss": 0.563,
      "step": 16697
    },
    {
      "epoch": 0.9350431179303393,
      "grad_norm": 1.2255337238311768,
      "learning_rate": 8.344500000000001e-05,
      "loss": 0.4638,
      "step": 16698
    },
    {
      "epoch": 0.9350991152424684,
      "grad_norm": 1.8082275390625,
      "learning_rate": 8.345000000000001e-05,
      "loss": 0.5849,
      "step": 16699
    },
    {
      "epoch": 0.9351551125545974,
      "grad_norm": 1.4461427927017212,
      "learning_rate": 8.3455e-05,
      "loss": 0.4325,
      "step": 16700
    },
    {
      "epoch": 0.9352111098667264,
      "grad_norm": 1.1311047077178955,
      "learning_rate": 8.346e-05,
      "loss": 0.3848,
      "step": 16701
    },
    {
      "epoch": 0.9352671071788554,
      "grad_norm": 1.0894145965576172,
      "learning_rate": 8.3465e-05,
      "loss": 0.3834,
      "step": 16702
    },
    {
      "epoch": 0.9353231044909844,
      "grad_norm": 3.1649255752563477,
      "learning_rate": 8.347e-05,
      "loss": 0.3808,
      "step": 16703
    },
    {
      "epoch": 0.9353791018031135,
      "grad_norm": 1.191104769706726,
      "learning_rate": 8.347500000000001e-05,
      "loss": 0.3527,
      "step": 16704
    },
    {
      "epoch": 0.9354350991152425,
      "grad_norm": 1.2181001901626587,
      "learning_rate": 8.348000000000001e-05,
      "loss": 0.3814,
      "step": 16705
    },
    {
      "epoch": 0.9354910964273715,
      "grad_norm": 1.340221881866455,
      "learning_rate": 8.3485e-05,
      "loss": 0.3855,
      "step": 16706
    },
    {
      "epoch": 0.9355470937395005,
      "grad_norm": 1.5500530004501343,
      "learning_rate": 8.349e-05,
      "loss": 0.4846,
      "step": 16707
    },
    {
      "epoch": 0.9356030910516295,
      "grad_norm": 1.3324302434921265,
      "learning_rate": 8.3495e-05,
      "loss": 0.4714,
      "step": 16708
    },
    {
      "epoch": 0.9356590883637586,
      "grad_norm": 1.3207728862762451,
      "learning_rate": 8.35e-05,
      "loss": 0.5499,
      "step": 16709
    },
    {
      "epoch": 0.9357150856758876,
      "grad_norm": 1.2513694763183594,
      "learning_rate": 8.3505e-05,
      "loss": 0.3918,
      "step": 16710
    },
    {
      "epoch": 0.9357710829880166,
      "grad_norm": 1.3769958019256592,
      "learning_rate": 8.351e-05,
      "loss": 0.4827,
      "step": 16711
    },
    {
      "epoch": 0.9358270803001456,
      "grad_norm": 1.632336139678955,
      "learning_rate": 8.3515e-05,
      "loss": 0.5865,
      "step": 16712
    },
    {
      "epoch": 0.9358830776122746,
      "grad_norm": 1.361846923828125,
      "learning_rate": 8.352000000000001e-05,
      "loss": 0.5407,
      "step": 16713
    },
    {
      "epoch": 0.9359390749244036,
      "grad_norm": 1.3150957822799683,
      "learning_rate": 8.352500000000001e-05,
      "loss": 0.5172,
      "step": 16714
    },
    {
      "epoch": 0.9359950722365327,
      "grad_norm": 1.3028266429901123,
      "learning_rate": 8.353000000000001e-05,
      "loss": 0.4127,
      "step": 16715
    },
    {
      "epoch": 0.9360510695486617,
      "grad_norm": 1.2794989347457886,
      "learning_rate": 8.3535e-05,
      "loss": 0.4052,
      "step": 16716
    },
    {
      "epoch": 0.9361070668607907,
      "grad_norm": 1.2054892778396606,
      "learning_rate": 8.354e-05,
      "loss": 0.3915,
      "step": 16717
    },
    {
      "epoch": 0.9361630641729197,
      "grad_norm": 1.4225001335144043,
      "learning_rate": 8.3545e-05,
      "loss": 0.4819,
      "step": 16718
    },
    {
      "epoch": 0.9362190614850487,
      "grad_norm": 1.4767318964004517,
      "learning_rate": 8.355000000000001e-05,
      "loss": 0.3541,
      "step": 16719
    },
    {
      "epoch": 0.9362750587971778,
      "grad_norm": 1.5410804748535156,
      "learning_rate": 8.355500000000001e-05,
      "loss": 0.4682,
      "step": 16720
    },
    {
      "epoch": 0.9363310561093068,
      "grad_norm": 1.2901997566223145,
      "learning_rate": 8.356e-05,
      "loss": 0.4525,
      "step": 16721
    },
    {
      "epoch": 0.9363870534214358,
      "grad_norm": 1.4899600744247437,
      "learning_rate": 8.3565e-05,
      "loss": 0.5563,
      "step": 16722
    },
    {
      "epoch": 0.9364430507335648,
      "grad_norm": 1.5106761455535889,
      "learning_rate": 8.357e-05,
      "loss": 0.4132,
      "step": 16723
    },
    {
      "epoch": 0.9364990480456938,
      "grad_norm": 1.3609869480133057,
      "learning_rate": 8.3575e-05,
      "loss": 0.4436,
      "step": 16724
    },
    {
      "epoch": 0.9365550453578229,
      "grad_norm": 1.3122451305389404,
      "learning_rate": 8.358e-05,
      "loss": 0.4533,
      "step": 16725
    },
    {
      "epoch": 0.9366110426699519,
      "grad_norm": 1.235106110572815,
      "learning_rate": 8.358500000000001e-05,
      "loss": 0.4028,
      "step": 16726
    },
    {
      "epoch": 0.9366670399820809,
      "grad_norm": 1.664569616317749,
      "learning_rate": 8.359e-05,
      "loss": 0.4744,
      "step": 16727
    },
    {
      "epoch": 0.9367230372942099,
      "grad_norm": 1.2943017482757568,
      "learning_rate": 8.3595e-05,
      "loss": 0.5061,
      "step": 16728
    },
    {
      "epoch": 0.9367790346063389,
      "grad_norm": 2.0248494148254395,
      "learning_rate": 8.36e-05,
      "loss": 0.5164,
      "step": 16729
    },
    {
      "epoch": 0.936835031918468,
      "grad_norm": 1.248183250427246,
      "learning_rate": 8.3605e-05,
      "loss": 0.4729,
      "step": 16730
    },
    {
      "epoch": 0.936891029230597,
      "grad_norm": 1.6285631656646729,
      "learning_rate": 8.361e-05,
      "loss": 0.4802,
      "step": 16731
    },
    {
      "epoch": 0.936947026542726,
      "grad_norm": 1.0719648599624634,
      "learning_rate": 8.3615e-05,
      "loss": 0.4056,
      "step": 16732
    },
    {
      "epoch": 0.937003023854855,
      "grad_norm": 1.513091802597046,
      "learning_rate": 8.362000000000002e-05,
      "loss": 0.6647,
      "step": 16733
    },
    {
      "epoch": 0.937059021166984,
      "grad_norm": 1.4053661823272705,
      "learning_rate": 8.362500000000001e-05,
      "loss": 0.4836,
      "step": 16734
    },
    {
      "epoch": 0.937115018479113,
      "grad_norm": 1.584080696105957,
      "learning_rate": 8.363000000000001e-05,
      "loss": 0.527,
      "step": 16735
    },
    {
      "epoch": 0.9371710157912421,
      "grad_norm": 1.1603384017944336,
      "learning_rate": 8.363500000000001e-05,
      "loss": 0.3775,
      "step": 16736
    },
    {
      "epoch": 0.9372270131033711,
      "grad_norm": 1.2846858501434326,
      "learning_rate": 8.364e-05,
      "loss": 0.5528,
      "step": 16737
    },
    {
      "epoch": 0.9372830104155001,
      "grad_norm": 1.8479763269424438,
      "learning_rate": 8.3645e-05,
      "loss": 0.5682,
      "step": 16738
    },
    {
      "epoch": 0.9373390077276291,
      "grad_norm": 1.514147400856018,
      "learning_rate": 8.365e-05,
      "loss": 0.3887,
      "step": 16739
    },
    {
      "epoch": 0.9373950050397581,
      "grad_norm": 1.4268624782562256,
      "learning_rate": 8.365500000000001e-05,
      "loss": 0.57,
      "step": 16740
    },
    {
      "epoch": 0.9374510023518872,
      "grad_norm": 1.175947904586792,
      "learning_rate": 8.366000000000001e-05,
      "loss": 0.4704,
      "step": 16741
    },
    {
      "epoch": 0.9375069996640161,
      "grad_norm": 1.1671470403671265,
      "learning_rate": 8.3665e-05,
      "loss": 0.3743,
      "step": 16742
    },
    {
      "epoch": 0.9375629969761451,
      "grad_norm": 1.4053637981414795,
      "learning_rate": 8.367e-05,
      "loss": 0.4791,
      "step": 16743
    },
    {
      "epoch": 0.9376189942882741,
      "grad_norm": 1.3294618129730225,
      "learning_rate": 8.3675e-05,
      "loss": 0.4177,
      "step": 16744
    },
    {
      "epoch": 0.9376749916004031,
      "grad_norm": 1.4835890531539917,
      "learning_rate": 8.368e-05,
      "loss": 0.6239,
      "step": 16745
    },
    {
      "epoch": 0.9377309889125321,
      "grad_norm": 1.3424429893493652,
      "learning_rate": 8.3685e-05,
      "loss": 0.4418,
      "step": 16746
    },
    {
      "epoch": 0.9377869862246612,
      "grad_norm": 1.3203662633895874,
      "learning_rate": 8.369000000000001e-05,
      "loss": 0.4023,
      "step": 16747
    },
    {
      "epoch": 0.9378429835367902,
      "grad_norm": 1.531296968460083,
      "learning_rate": 8.3695e-05,
      "loss": 0.4345,
      "step": 16748
    },
    {
      "epoch": 0.9378989808489192,
      "grad_norm": 1.4427509307861328,
      "learning_rate": 8.37e-05,
      "loss": 0.4551,
      "step": 16749
    },
    {
      "epoch": 0.9379549781610482,
      "grad_norm": 1.5274933576583862,
      "learning_rate": 8.3705e-05,
      "loss": 0.6143,
      "step": 16750
    },
    {
      "epoch": 0.9380109754731772,
      "grad_norm": 1.2396273612976074,
      "learning_rate": 8.371e-05,
      "loss": 0.3572,
      "step": 16751
    },
    {
      "epoch": 0.9380669727853063,
      "grad_norm": 1.183813214302063,
      "learning_rate": 8.3715e-05,
      "loss": 0.403,
      "step": 16752
    },
    {
      "epoch": 0.9381229700974353,
      "grad_norm": 1.4945588111877441,
      "learning_rate": 8.372e-05,
      "loss": 0.4316,
      "step": 16753
    },
    {
      "epoch": 0.9381789674095643,
      "grad_norm": 1.4189181327819824,
      "learning_rate": 8.3725e-05,
      "loss": 0.4808,
      "step": 16754
    },
    {
      "epoch": 0.9382349647216933,
      "grad_norm": 1.2214527130126953,
      "learning_rate": 8.373000000000001e-05,
      "loss": 0.467,
      "step": 16755
    },
    {
      "epoch": 0.9382909620338223,
      "grad_norm": 1.5569618940353394,
      "learning_rate": 8.373500000000001e-05,
      "loss": 0.458,
      "step": 16756
    },
    {
      "epoch": 0.9383469593459514,
      "grad_norm": 1.3493481874465942,
      "learning_rate": 8.374000000000001e-05,
      "loss": 0.4004,
      "step": 16757
    },
    {
      "epoch": 0.9384029566580804,
      "grad_norm": 1.4711847305297852,
      "learning_rate": 8.3745e-05,
      "loss": 0.4783,
      "step": 16758
    },
    {
      "epoch": 0.9384589539702094,
      "grad_norm": 1.4054988622665405,
      "learning_rate": 8.375e-05,
      "loss": 0.5384,
      "step": 16759
    },
    {
      "epoch": 0.9385149512823384,
      "grad_norm": 1.324813723564148,
      "learning_rate": 8.3755e-05,
      "loss": 0.3708,
      "step": 16760
    },
    {
      "epoch": 0.9385709485944674,
      "grad_norm": 1.344856858253479,
      "learning_rate": 8.376000000000001e-05,
      "loss": 0.5535,
      "step": 16761
    },
    {
      "epoch": 0.9386269459065965,
      "grad_norm": 1.283390760421753,
      "learning_rate": 8.376500000000001e-05,
      "loss": 0.6027,
      "step": 16762
    },
    {
      "epoch": 0.9386829432187255,
      "grad_norm": 1.4102137088775635,
      "learning_rate": 8.377e-05,
      "loss": 0.7469,
      "step": 16763
    },
    {
      "epoch": 0.9387389405308545,
      "grad_norm": 1.2193641662597656,
      "learning_rate": 8.3775e-05,
      "loss": 0.4421,
      "step": 16764
    },
    {
      "epoch": 0.9387949378429835,
      "grad_norm": 1.1807458400726318,
      "learning_rate": 8.378e-05,
      "loss": 0.5354,
      "step": 16765
    },
    {
      "epoch": 0.9388509351551125,
      "grad_norm": 1.3945485353469849,
      "learning_rate": 8.3785e-05,
      "loss": 0.3762,
      "step": 16766
    },
    {
      "epoch": 0.9389069324672416,
      "grad_norm": 1.2609732151031494,
      "learning_rate": 8.379e-05,
      "loss": 0.3866,
      "step": 16767
    },
    {
      "epoch": 0.9389629297793706,
      "grad_norm": 1.4377027750015259,
      "learning_rate": 8.3795e-05,
      "loss": 0.5106,
      "step": 16768
    },
    {
      "epoch": 0.9390189270914996,
      "grad_norm": 1.3723695278167725,
      "learning_rate": 8.38e-05,
      "loss": 0.4197,
      "step": 16769
    },
    {
      "epoch": 0.9390749244036286,
      "grad_norm": 1.7024558782577515,
      "learning_rate": 8.3805e-05,
      "loss": 0.533,
      "step": 16770
    },
    {
      "epoch": 0.9391309217157576,
      "grad_norm": 1.4607455730438232,
      "learning_rate": 8.381e-05,
      "loss": 0.3979,
      "step": 16771
    },
    {
      "epoch": 0.9391869190278866,
      "grad_norm": 1.6503533124923706,
      "learning_rate": 8.3815e-05,
      "loss": 0.5965,
      "step": 16772
    },
    {
      "epoch": 0.9392429163400157,
      "grad_norm": 1.5065014362335205,
      "learning_rate": 8.382e-05,
      "loss": 0.712,
      "step": 16773
    },
    {
      "epoch": 0.9392989136521447,
      "grad_norm": 1.4836997985839844,
      "learning_rate": 8.3825e-05,
      "loss": 0.5417,
      "step": 16774
    },
    {
      "epoch": 0.9393549109642737,
      "grad_norm": 1.4093387126922607,
      "learning_rate": 8.383e-05,
      "loss": 0.4704,
      "step": 16775
    },
    {
      "epoch": 0.9394109082764027,
      "grad_norm": 1.4547462463378906,
      "learning_rate": 8.383500000000001e-05,
      "loss": 0.4811,
      "step": 16776
    },
    {
      "epoch": 0.9394669055885317,
      "grad_norm": 1.3972203731536865,
      "learning_rate": 8.384000000000001e-05,
      "loss": 0.4777,
      "step": 16777
    },
    {
      "epoch": 0.9395229029006608,
      "grad_norm": 1.5986653566360474,
      "learning_rate": 8.384500000000001e-05,
      "loss": 0.4574,
      "step": 16778
    },
    {
      "epoch": 0.9395789002127898,
      "grad_norm": 1.2717307806015015,
      "learning_rate": 8.385e-05,
      "loss": 0.3281,
      "step": 16779
    },
    {
      "epoch": 0.9396348975249188,
      "grad_norm": 1.1803277730941772,
      "learning_rate": 8.3855e-05,
      "loss": 0.3698,
      "step": 16780
    },
    {
      "epoch": 0.9396908948370478,
      "grad_norm": 1.5105961561203003,
      "learning_rate": 8.386e-05,
      "loss": 0.4268,
      "step": 16781
    },
    {
      "epoch": 0.9397468921491768,
      "grad_norm": 1.2344133853912354,
      "learning_rate": 8.386500000000001e-05,
      "loss": 0.4531,
      "step": 16782
    },
    {
      "epoch": 0.9398028894613059,
      "grad_norm": 1.1643778085708618,
      "learning_rate": 8.387000000000001e-05,
      "loss": 0.3649,
      "step": 16783
    },
    {
      "epoch": 0.9398588867734349,
      "grad_norm": 1.440381646156311,
      "learning_rate": 8.3875e-05,
      "loss": 0.5847,
      "step": 16784
    },
    {
      "epoch": 0.9399148840855639,
      "grad_norm": 1.1849454641342163,
      "learning_rate": 8.388e-05,
      "loss": 0.3915,
      "step": 16785
    },
    {
      "epoch": 0.9399708813976929,
      "grad_norm": 1.1867284774780273,
      "learning_rate": 8.3885e-05,
      "loss": 0.4209,
      "step": 16786
    },
    {
      "epoch": 0.9400268787098219,
      "grad_norm": 1.245200276374817,
      "learning_rate": 8.389e-05,
      "loss": 0.4678,
      "step": 16787
    },
    {
      "epoch": 0.940082876021951,
      "grad_norm": 1.2511872053146362,
      "learning_rate": 8.3895e-05,
      "loss": 0.4868,
      "step": 16788
    },
    {
      "epoch": 0.94013887333408,
      "grad_norm": 2.0155186653137207,
      "learning_rate": 8.39e-05,
      "loss": 0.4851,
      "step": 16789
    },
    {
      "epoch": 0.940194870646209,
      "grad_norm": 1.2310574054718018,
      "learning_rate": 8.3905e-05,
      "loss": 0.4153,
      "step": 16790
    },
    {
      "epoch": 0.940250867958338,
      "grad_norm": 1.2299989461898804,
      "learning_rate": 8.391e-05,
      "loss": 0.4983,
      "step": 16791
    },
    {
      "epoch": 0.940306865270467,
      "grad_norm": 1.5917248725891113,
      "learning_rate": 8.3915e-05,
      "loss": 0.4836,
      "step": 16792
    },
    {
      "epoch": 0.940362862582596,
      "grad_norm": 1.1603355407714844,
      "learning_rate": 8.392e-05,
      "loss": 0.4125,
      "step": 16793
    },
    {
      "epoch": 0.9404188598947251,
      "grad_norm": 1.0655100345611572,
      "learning_rate": 8.392500000000001e-05,
      "loss": 0.3189,
      "step": 16794
    },
    {
      "epoch": 0.9404748572068541,
      "grad_norm": 1.6109964847564697,
      "learning_rate": 8.393e-05,
      "loss": 0.52,
      "step": 16795
    },
    {
      "epoch": 0.9405308545189831,
      "grad_norm": 1.5058531761169434,
      "learning_rate": 8.3935e-05,
      "loss": 0.4478,
      "step": 16796
    },
    {
      "epoch": 0.9405868518311121,
      "grad_norm": 1.4384710788726807,
      "learning_rate": 8.394000000000001e-05,
      "loss": 0.4512,
      "step": 16797
    },
    {
      "epoch": 0.9406428491432411,
      "grad_norm": 1.2084736824035645,
      "learning_rate": 8.394500000000001e-05,
      "loss": 0.3652,
      "step": 16798
    },
    {
      "epoch": 0.9406988464553702,
      "grad_norm": 1.5859885215759277,
      "learning_rate": 8.395000000000001e-05,
      "loss": 0.4277,
      "step": 16799
    },
    {
      "epoch": 0.9407548437674992,
      "grad_norm": 1.2028608322143555,
      "learning_rate": 8.3955e-05,
      "loss": 0.4893,
      "step": 16800
    },
    {
      "epoch": 0.9408108410796282,
      "grad_norm": 1.3841071128845215,
      "learning_rate": 8.396e-05,
      "loss": 0.439,
      "step": 16801
    },
    {
      "epoch": 0.9408668383917572,
      "grad_norm": 1.3934539556503296,
      "learning_rate": 8.3965e-05,
      "loss": 0.408,
      "step": 16802
    },
    {
      "epoch": 0.9409228357038862,
      "grad_norm": 1.1731603145599365,
      "learning_rate": 8.397000000000001e-05,
      "loss": 0.3979,
      "step": 16803
    },
    {
      "epoch": 0.9409788330160153,
      "grad_norm": 1.3989534378051758,
      "learning_rate": 8.397500000000001e-05,
      "loss": 0.432,
      "step": 16804
    },
    {
      "epoch": 0.9410348303281443,
      "grad_norm": 1.3040473461151123,
      "learning_rate": 8.398e-05,
      "loss": 0.4161,
      "step": 16805
    },
    {
      "epoch": 0.9410908276402733,
      "grad_norm": 1.228775978088379,
      "learning_rate": 8.3985e-05,
      "loss": 0.3721,
      "step": 16806
    },
    {
      "epoch": 0.9411468249524023,
      "grad_norm": 1.2703782320022583,
      "learning_rate": 8.399e-05,
      "loss": 0.5118,
      "step": 16807
    },
    {
      "epoch": 0.9412028222645313,
      "grad_norm": 1.5137567520141602,
      "learning_rate": 8.3995e-05,
      "loss": 0.4539,
      "step": 16808
    },
    {
      "epoch": 0.9412588195766604,
      "grad_norm": 1.775665283203125,
      "learning_rate": 8.4e-05,
      "loss": 0.583,
      "step": 16809
    },
    {
      "epoch": 0.9413148168887894,
      "grad_norm": 1.314042091369629,
      "learning_rate": 8.4005e-05,
      "loss": 0.426,
      "step": 16810
    },
    {
      "epoch": 0.9413708142009184,
      "grad_norm": 1.167601466178894,
      "learning_rate": 8.401e-05,
      "loss": 0.5671,
      "step": 16811
    },
    {
      "epoch": 0.9414268115130474,
      "grad_norm": 1.4067963361740112,
      "learning_rate": 8.4015e-05,
      "loss": 0.4454,
      "step": 16812
    },
    {
      "epoch": 0.9414828088251764,
      "grad_norm": 2.502490282058716,
      "learning_rate": 8.402e-05,
      "loss": 0.4565,
      "step": 16813
    },
    {
      "epoch": 0.9415388061373055,
      "grad_norm": 2.032898426055908,
      "learning_rate": 8.402500000000001e-05,
      "loss": 0.7062,
      "step": 16814
    },
    {
      "epoch": 0.9415948034494345,
      "grad_norm": 1.5732282400131226,
      "learning_rate": 8.403000000000001e-05,
      "loss": 0.4778,
      "step": 16815
    },
    {
      "epoch": 0.9416508007615635,
      "grad_norm": 1.5170363187789917,
      "learning_rate": 8.4035e-05,
      "loss": 0.5445,
      "step": 16816
    },
    {
      "epoch": 0.9417067980736925,
      "grad_norm": 1.4666023254394531,
      "learning_rate": 8.404e-05,
      "loss": 0.433,
      "step": 16817
    },
    {
      "epoch": 0.9417627953858215,
      "grad_norm": 1.441066026687622,
      "learning_rate": 8.404500000000001e-05,
      "loss": 0.6679,
      "step": 16818
    },
    {
      "epoch": 0.9418187926979505,
      "grad_norm": 1.2354779243469238,
      "learning_rate": 8.405000000000001e-05,
      "loss": 0.4044,
      "step": 16819
    },
    {
      "epoch": 0.9418747900100796,
      "grad_norm": 1.2528882026672363,
      "learning_rate": 8.405500000000001e-05,
      "loss": 0.4331,
      "step": 16820
    },
    {
      "epoch": 0.9419307873222086,
      "grad_norm": 1.175680160522461,
      "learning_rate": 8.406e-05,
      "loss": 0.4404,
      "step": 16821
    },
    {
      "epoch": 0.9419867846343376,
      "grad_norm": 2.039055347442627,
      "learning_rate": 8.4065e-05,
      "loss": 0.4098,
      "step": 16822
    },
    {
      "epoch": 0.9420427819464666,
      "grad_norm": 1.3916031122207642,
      "learning_rate": 8.407e-05,
      "loss": 0.4205,
      "step": 16823
    },
    {
      "epoch": 0.9420987792585956,
      "grad_norm": 1.3693307638168335,
      "learning_rate": 8.4075e-05,
      "loss": 0.5207,
      "step": 16824
    },
    {
      "epoch": 0.9421547765707246,
      "grad_norm": 1.4005661010742188,
      "learning_rate": 8.408000000000001e-05,
      "loss": 0.524,
      "step": 16825
    },
    {
      "epoch": 0.9422107738828536,
      "grad_norm": 1.458531141281128,
      "learning_rate": 8.4085e-05,
      "loss": 0.5229,
      "step": 16826
    },
    {
      "epoch": 0.9422667711949826,
      "grad_norm": 1.1541227102279663,
      "learning_rate": 8.409e-05,
      "loss": 0.3867,
      "step": 16827
    },
    {
      "epoch": 0.9423227685071116,
      "grad_norm": 1.5439306497573853,
      "learning_rate": 8.4095e-05,
      "loss": 0.531,
      "step": 16828
    },
    {
      "epoch": 0.9423787658192406,
      "grad_norm": 1.3004478216171265,
      "learning_rate": 8.41e-05,
      "loss": 0.4577,
      "step": 16829
    },
    {
      "epoch": 0.9424347631313696,
      "grad_norm": 1.5048997402191162,
      "learning_rate": 8.4105e-05,
      "loss": 0.5917,
      "step": 16830
    },
    {
      "epoch": 0.9424907604434987,
      "grad_norm": 1.692022681236267,
      "learning_rate": 8.411e-05,
      "loss": 0.4814,
      "step": 16831
    },
    {
      "epoch": 0.9425467577556277,
      "grad_norm": 1.2445706129074097,
      "learning_rate": 8.4115e-05,
      "loss": 0.5822,
      "step": 16832
    },
    {
      "epoch": 0.9426027550677567,
      "grad_norm": 1.32802152633667,
      "learning_rate": 8.412e-05,
      "loss": 0.499,
      "step": 16833
    },
    {
      "epoch": 0.9426587523798857,
      "grad_norm": 1.4759972095489502,
      "learning_rate": 8.412500000000001e-05,
      "loss": 0.5567,
      "step": 16834
    },
    {
      "epoch": 0.9427147496920147,
      "grad_norm": 1.3070430755615234,
      "learning_rate": 8.413000000000001e-05,
      "loss": 0.361,
      "step": 16835
    },
    {
      "epoch": 0.9427707470041438,
      "grad_norm": 1.3818913698196411,
      "learning_rate": 8.4135e-05,
      "loss": 0.4857,
      "step": 16836
    },
    {
      "epoch": 0.9428267443162728,
      "grad_norm": 1.1885535717010498,
      "learning_rate": 8.414e-05,
      "loss": 0.4102,
      "step": 16837
    },
    {
      "epoch": 0.9428827416284018,
      "grad_norm": 1.4455609321594238,
      "learning_rate": 8.4145e-05,
      "loss": 0.3973,
      "step": 16838
    },
    {
      "epoch": 0.9429387389405308,
      "grad_norm": 1.3386549949645996,
      "learning_rate": 8.415000000000001e-05,
      "loss": 0.5643,
      "step": 16839
    },
    {
      "epoch": 0.9429947362526598,
      "grad_norm": 1.6348649263381958,
      "learning_rate": 8.415500000000001e-05,
      "loss": 0.4125,
      "step": 16840
    },
    {
      "epoch": 0.9430507335647889,
      "grad_norm": 1.3806694746017456,
      "learning_rate": 8.416000000000001e-05,
      "loss": 0.4462,
      "step": 16841
    },
    {
      "epoch": 0.9431067308769179,
      "grad_norm": 1.826649785041809,
      "learning_rate": 8.4165e-05,
      "loss": 0.4605,
      "step": 16842
    },
    {
      "epoch": 0.9431627281890469,
      "grad_norm": 1.4051538705825806,
      "learning_rate": 8.417e-05,
      "loss": 0.5421,
      "step": 16843
    },
    {
      "epoch": 0.9432187255011759,
      "grad_norm": 1.68427312374115,
      "learning_rate": 8.4175e-05,
      "loss": 0.5481,
      "step": 16844
    },
    {
      "epoch": 0.9432747228133049,
      "grad_norm": 1.3187127113342285,
      "learning_rate": 8.418e-05,
      "loss": 0.405,
      "step": 16845
    },
    {
      "epoch": 0.943330720125434,
      "grad_norm": 1.2708790302276611,
      "learning_rate": 8.418500000000001e-05,
      "loss": 0.6116,
      "step": 16846
    },
    {
      "epoch": 0.943386717437563,
      "grad_norm": 1.282198190689087,
      "learning_rate": 8.419e-05,
      "loss": 0.3726,
      "step": 16847
    },
    {
      "epoch": 0.943442714749692,
      "grad_norm": 1.1929875612258911,
      "learning_rate": 8.4195e-05,
      "loss": 0.4536,
      "step": 16848
    },
    {
      "epoch": 0.943498712061821,
      "grad_norm": 1.3808897733688354,
      "learning_rate": 8.42e-05,
      "loss": 0.4514,
      "step": 16849
    },
    {
      "epoch": 0.94355470937395,
      "grad_norm": 1.4703110456466675,
      "learning_rate": 8.4205e-05,
      "loss": 0.4597,
      "step": 16850
    },
    {
      "epoch": 0.943610706686079,
      "grad_norm": 1.349044680595398,
      "learning_rate": 8.421e-05,
      "loss": 0.6292,
      "step": 16851
    },
    {
      "epoch": 0.9436667039982081,
      "grad_norm": 1.147835373878479,
      "learning_rate": 8.4215e-05,
      "loss": 0.4185,
      "step": 16852
    },
    {
      "epoch": 0.9437227013103371,
      "grad_norm": 1.3355592489242554,
      "learning_rate": 8.422e-05,
      "loss": 0.5128,
      "step": 16853
    },
    {
      "epoch": 0.9437786986224661,
      "grad_norm": 1.4008398056030273,
      "learning_rate": 8.422500000000001e-05,
      "loss": 0.5362,
      "step": 16854
    },
    {
      "epoch": 0.9438346959345951,
      "grad_norm": 1.6245315074920654,
      "learning_rate": 8.423000000000001e-05,
      "loss": 0.6372,
      "step": 16855
    },
    {
      "epoch": 0.9438906932467241,
      "grad_norm": 2.5869712829589844,
      "learning_rate": 8.423500000000001e-05,
      "loss": 0.5831,
      "step": 16856
    },
    {
      "epoch": 0.9439466905588532,
      "grad_norm": 1.4545773267745972,
      "learning_rate": 8.424e-05,
      "loss": 0.5986,
      "step": 16857
    },
    {
      "epoch": 0.9440026878709822,
      "grad_norm": 1.1646499633789062,
      "learning_rate": 8.4245e-05,
      "loss": 0.475,
      "step": 16858
    },
    {
      "epoch": 0.9440586851831112,
      "grad_norm": 1.5359723567962646,
      "learning_rate": 8.425e-05,
      "loss": 0.3858,
      "step": 16859
    },
    {
      "epoch": 0.9441146824952402,
      "grad_norm": 1.070334553718567,
      "learning_rate": 8.425500000000001e-05,
      "loss": 0.3192,
      "step": 16860
    },
    {
      "epoch": 0.9441706798073692,
      "grad_norm": 1.770363211631775,
      "learning_rate": 8.426000000000001e-05,
      "loss": 0.4904,
      "step": 16861
    },
    {
      "epoch": 0.9442266771194983,
      "grad_norm": 1.579593300819397,
      "learning_rate": 8.426500000000001e-05,
      "loss": 0.586,
      "step": 16862
    },
    {
      "epoch": 0.9442826744316273,
      "grad_norm": 1.3765833377838135,
      "learning_rate": 8.427e-05,
      "loss": 0.5836,
      "step": 16863
    },
    {
      "epoch": 0.9443386717437563,
      "grad_norm": 1.5055797100067139,
      "learning_rate": 8.4275e-05,
      "loss": 0.4513,
      "step": 16864
    },
    {
      "epoch": 0.9443946690558853,
      "grad_norm": 1.366385579109192,
      "learning_rate": 8.428e-05,
      "loss": 0.4448,
      "step": 16865
    },
    {
      "epoch": 0.9444506663680143,
      "grad_norm": 1.3146216869354248,
      "learning_rate": 8.4285e-05,
      "loss": 0.4196,
      "step": 16866
    },
    {
      "epoch": 0.9445066636801434,
      "grad_norm": 1.6299651861190796,
      "learning_rate": 8.429000000000001e-05,
      "loss": 0.517,
      "step": 16867
    },
    {
      "epoch": 0.9445626609922724,
      "grad_norm": 1.3505542278289795,
      "learning_rate": 8.4295e-05,
      "loss": 0.4939,
      "step": 16868
    },
    {
      "epoch": 0.9446186583044014,
      "grad_norm": 1.7189971208572388,
      "learning_rate": 8.43e-05,
      "loss": 0.491,
      "step": 16869
    },
    {
      "epoch": 0.9446746556165304,
      "grad_norm": 1.3334531784057617,
      "learning_rate": 8.4305e-05,
      "loss": 0.4253,
      "step": 16870
    },
    {
      "epoch": 0.9447306529286594,
      "grad_norm": 1.3857353925704956,
      "learning_rate": 8.431e-05,
      "loss": 0.4126,
      "step": 16871
    },
    {
      "epoch": 0.9447866502407885,
      "grad_norm": 1.5548064708709717,
      "learning_rate": 8.4315e-05,
      "loss": 0.5157,
      "step": 16872
    },
    {
      "epoch": 0.9448426475529175,
      "grad_norm": 1.399116039276123,
      "learning_rate": 8.431999999999999e-05,
      "loss": 0.413,
      "step": 16873
    },
    {
      "epoch": 0.9448986448650465,
      "grad_norm": 1.8226274251937866,
      "learning_rate": 8.4325e-05,
      "loss": 0.4868,
      "step": 16874
    },
    {
      "epoch": 0.9449546421771755,
      "grad_norm": 1.6294915676116943,
      "learning_rate": 8.433000000000001e-05,
      "loss": 0.511,
      "step": 16875
    },
    {
      "epoch": 0.9450106394893045,
      "grad_norm": 1.645871877670288,
      "learning_rate": 8.433500000000001e-05,
      "loss": 0.3838,
      "step": 16876
    },
    {
      "epoch": 0.9450666368014335,
      "grad_norm": 1.2966974973678589,
      "learning_rate": 8.434000000000001e-05,
      "loss": 0.4449,
      "step": 16877
    },
    {
      "epoch": 0.9451226341135626,
      "grad_norm": 1.1157574653625488,
      "learning_rate": 8.4345e-05,
      "loss": 0.3294,
      "step": 16878
    },
    {
      "epoch": 0.9451786314256916,
      "grad_norm": 1.3588204383850098,
      "learning_rate": 8.435e-05,
      "loss": 0.5036,
      "step": 16879
    },
    {
      "epoch": 0.9452346287378206,
      "grad_norm": 1.3987594842910767,
      "learning_rate": 8.4355e-05,
      "loss": 0.5028,
      "step": 16880
    },
    {
      "epoch": 0.9452906260499496,
      "grad_norm": 1.1793015003204346,
      "learning_rate": 8.436000000000001e-05,
      "loss": 0.3366,
      "step": 16881
    },
    {
      "epoch": 0.9453466233620786,
      "grad_norm": 1.2872474193572998,
      "learning_rate": 8.436500000000001e-05,
      "loss": 0.4089,
      "step": 16882
    },
    {
      "epoch": 0.9454026206742077,
      "grad_norm": 1.6301637887954712,
      "learning_rate": 8.437000000000001e-05,
      "loss": 0.4551,
      "step": 16883
    },
    {
      "epoch": 0.9454586179863367,
      "grad_norm": 1.6496950387954712,
      "learning_rate": 8.4375e-05,
      "loss": 0.5353,
      "step": 16884
    },
    {
      "epoch": 0.9455146152984657,
      "grad_norm": 1.4343160390853882,
      "learning_rate": 8.438e-05,
      "loss": 0.4322,
      "step": 16885
    },
    {
      "epoch": 0.9455706126105947,
      "grad_norm": 1.707632303237915,
      "learning_rate": 8.4385e-05,
      "loss": 0.392,
      "step": 16886
    },
    {
      "epoch": 0.9456266099227237,
      "grad_norm": 1.0887588262557983,
      "learning_rate": 8.439e-05,
      "loss": 0.3682,
      "step": 16887
    },
    {
      "epoch": 0.9456826072348528,
      "grad_norm": 1.4382503032684326,
      "learning_rate": 8.439500000000001e-05,
      "loss": 0.4761,
      "step": 16888
    },
    {
      "epoch": 0.9457386045469818,
      "grad_norm": 1.2576593160629272,
      "learning_rate": 8.44e-05,
      "loss": 0.4996,
      "step": 16889
    },
    {
      "epoch": 0.9457946018591108,
      "grad_norm": 1.357924461364746,
      "learning_rate": 8.4405e-05,
      "loss": 0.4564,
      "step": 16890
    },
    {
      "epoch": 0.9458505991712398,
      "grad_norm": 1.4293816089630127,
      "learning_rate": 8.441e-05,
      "loss": 0.4203,
      "step": 16891
    },
    {
      "epoch": 0.9459065964833688,
      "grad_norm": 1.5781736373901367,
      "learning_rate": 8.4415e-05,
      "loss": 0.4303,
      "step": 16892
    },
    {
      "epoch": 0.9459625937954979,
      "grad_norm": 1.687091588973999,
      "learning_rate": 8.442e-05,
      "loss": 0.5184,
      "step": 16893
    },
    {
      "epoch": 0.9460185911076269,
      "grad_norm": 1.4222277402877808,
      "learning_rate": 8.442499999999999e-05,
      "loss": 0.508,
      "step": 16894
    },
    {
      "epoch": 0.9460745884197559,
      "grad_norm": 1.134165644645691,
      "learning_rate": 8.443e-05,
      "loss": 0.4689,
      "step": 16895
    },
    {
      "epoch": 0.9461305857318849,
      "grad_norm": 1.3414344787597656,
      "learning_rate": 8.443500000000001e-05,
      "loss": 0.4429,
      "step": 16896
    },
    {
      "epoch": 0.9461865830440139,
      "grad_norm": 1.2850489616394043,
      "learning_rate": 8.444000000000001e-05,
      "loss": 0.4379,
      "step": 16897
    },
    {
      "epoch": 0.946242580356143,
      "grad_norm": 1.5533628463745117,
      "learning_rate": 8.444500000000001e-05,
      "loss": 0.3544,
      "step": 16898
    },
    {
      "epoch": 0.946298577668272,
      "grad_norm": 1.3350168466567993,
      "learning_rate": 8.445e-05,
      "loss": 0.4305,
      "step": 16899
    },
    {
      "epoch": 0.946354574980401,
      "grad_norm": 1.4601842164993286,
      "learning_rate": 8.4455e-05,
      "loss": 0.4564,
      "step": 16900
    },
    {
      "epoch": 0.94641057229253,
      "grad_norm": 1.246171474456787,
      "learning_rate": 8.446e-05,
      "loss": 0.3276,
      "step": 16901
    },
    {
      "epoch": 0.946466569604659,
      "grad_norm": 1.3086272478103638,
      "learning_rate": 8.4465e-05,
      "loss": 0.4221,
      "step": 16902
    },
    {
      "epoch": 0.946522566916788,
      "grad_norm": 1.4926385879516602,
      "learning_rate": 8.447000000000001e-05,
      "loss": 0.5245,
      "step": 16903
    },
    {
      "epoch": 0.9465785642289171,
      "grad_norm": 1.481048822402954,
      "learning_rate": 8.447500000000001e-05,
      "loss": 0.395,
      "step": 16904
    },
    {
      "epoch": 0.9466345615410461,
      "grad_norm": 1.3172568082809448,
      "learning_rate": 8.448e-05,
      "loss": 0.4195,
      "step": 16905
    },
    {
      "epoch": 0.9466905588531751,
      "grad_norm": 1.406826376914978,
      "learning_rate": 8.4485e-05,
      "loss": 0.4089,
      "step": 16906
    },
    {
      "epoch": 0.9467465561653041,
      "grad_norm": 1.5490084886550903,
      "learning_rate": 8.449e-05,
      "loss": 0.4135,
      "step": 16907
    },
    {
      "epoch": 0.946802553477433,
      "grad_norm": 1.8110581636428833,
      "learning_rate": 8.4495e-05,
      "loss": 0.6009,
      "step": 16908
    },
    {
      "epoch": 0.946858550789562,
      "grad_norm": 1.189125895500183,
      "learning_rate": 8.450000000000001e-05,
      "loss": 0.5155,
      "step": 16909
    },
    {
      "epoch": 0.9469145481016911,
      "grad_norm": 1.2542072534561157,
      "learning_rate": 8.4505e-05,
      "loss": 0.4491,
      "step": 16910
    },
    {
      "epoch": 0.9469705454138201,
      "grad_norm": 1.4154157638549805,
      "learning_rate": 8.451e-05,
      "loss": 0.5446,
      "step": 16911
    },
    {
      "epoch": 0.9470265427259491,
      "grad_norm": 1.1997610330581665,
      "learning_rate": 8.4515e-05,
      "loss": 0.5155,
      "step": 16912
    },
    {
      "epoch": 0.9470825400380781,
      "grad_norm": 1.0751558542251587,
      "learning_rate": 8.452e-05,
      "loss": 0.3765,
      "step": 16913
    },
    {
      "epoch": 0.9471385373502071,
      "grad_norm": 1.3078124523162842,
      "learning_rate": 8.4525e-05,
      "loss": 0.5072,
      "step": 16914
    },
    {
      "epoch": 0.9471945346623362,
      "grad_norm": 1.3146898746490479,
      "learning_rate": 8.453e-05,
      "loss": 0.4415,
      "step": 16915
    },
    {
      "epoch": 0.9472505319744652,
      "grad_norm": 1.3087681531906128,
      "learning_rate": 8.4535e-05,
      "loss": 0.3962,
      "step": 16916
    },
    {
      "epoch": 0.9473065292865942,
      "grad_norm": 1.1553888320922852,
      "learning_rate": 8.454000000000001e-05,
      "loss": 0.4969,
      "step": 16917
    },
    {
      "epoch": 0.9473625265987232,
      "grad_norm": 1.4546152353286743,
      "learning_rate": 8.454500000000001e-05,
      "loss": 0.5268,
      "step": 16918
    },
    {
      "epoch": 0.9474185239108522,
      "grad_norm": 1.5179592370986938,
      "learning_rate": 8.455000000000001e-05,
      "loss": 0.4177,
      "step": 16919
    },
    {
      "epoch": 0.9474745212229813,
      "grad_norm": 1.39425790309906,
      "learning_rate": 8.4555e-05,
      "loss": 0.568,
      "step": 16920
    },
    {
      "epoch": 0.9475305185351103,
      "grad_norm": 1.3313343524932861,
      "learning_rate": 8.456e-05,
      "loss": 0.419,
      "step": 16921
    },
    {
      "epoch": 0.9475865158472393,
      "grad_norm": 1.3817088603973389,
      "learning_rate": 8.4565e-05,
      "loss": 0.5545,
      "step": 16922
    },
    {
      "epoch": 0.9476425131593683,
      "grad_norm": 1.2225130796432495,
      "learning_rate": 8.457e-05,
      "loss": 0.4863,
      "step": 16923
    },
    {
      "epoch": 0.9476985104714973,
      "grad_norm": 1.3912243843078613,
      "learning_rate": 8.457500000000001e-05,
      "loss": 0.5267,
      "step": 16924
    },
    {
      "epoch": 0.9477545077836264,
      "grad_norm": 1.3930424451828003,
      "learning_rate": 8.458e-05,
      "loss": 0.3551,
      "step": 16925
    },
    {
      "epoch": 0.9478105050957554,
      "grad_norm": 1.426986575126648,
      "learning_rate": 8.4585e-05,
      "loss": 0.4643,
      "step": 16926
    },
    {
      "epoch": 0.9478665024078844,
      "grad_norm": 1.3757551908493042,
      "learning_rate": 8.459e-05,
      "loss": 0.4426,
      "step": 16927
    },
    {
      "epoch": 0.9479224997200134,
      "grad_norm": 1.4466967582702637,
      "learning_rate": 8.4595e-05,
      "loss": 0.4384,
      "step": 16928
    },
    {
      "epoch": 0.9479784970321424,
      "grad_norm": 1.2996398210525513,
      "learning_rate": 8.46e-05,
      "loss": 0.3729,
      "step": 16929
    },
    {
      "epoch": 0.9480344943442715,
      "grad_norm": 1.6008819341659546,
      "learning_rate": 8.460500000000001e-05,
      "loss": 0.4521,
      "step": 16930
    },
    {
      "epoch": 0.9480904916564005,
      "grad_norm": 1.7591370344161987,
      "learning_rate": 8.461e-05,
      "loss": 0.561,
      "step": 16931
    },
    {
      "epoch": 0.9481464889685295,
      "grad_norm": 1.2559250593185425,
      "learning_rate": 8.4615e-05,
      "loss": 0.3675,
      "step": 16932
    },
    {
      "epoch": 0.9482024862806585,
      "grad_norm": 1.5842159986495972,
      "learning_rate": 8.462e-05,
      "loss": 0.6105,
      "step": 16933
    },
    {
      "epoch": 0.9482584835927875,
      "grad_norm": 1.522939920425415,
      "learning_rate": 8.4625e-05,
      "loss": 0.4102,
      "step": 16934
    },
    {
      "epoch": 0.9483144809049165,
      "grad_norm": 1.306370496749878,
      "learning_rate": 8.463000000000001e-05,
      "loss": 0.515,
      "step": 16935
    },
    {
      "epoch": 0.9483704782170456,
      "grad_norm": 1.504814624786377,
      "learning_rate": 8.4635e-05,
      "loss": 0.5157,
      "step": 16936
    },
    {
      "epoch": 0.9484264755291746,
      "grad_norm": 1.1809239387512207,
      "learning_rate": 8.464e-05,
      "loss": 0.4107,
      "step": 16937
    },
    {
      "epoch": 0.9484824728413036,
      "grad_norm": 1.5091184377670288,
      "learning_rate": 8.464500000000001e-05,
      "loss": 0.4104,
      "step": 16938
    },
    {
      "epoch": 0.9485384701534326,
      "grad_norm": 1.0965172052383423,
      "learning_rate": 8.465000000000001e-05,
      "loss": 0.2987,
      "step": 16939
    },
    {
      "epoch": 0.9485944674655616,
      "grad_norm": 1.2307863235473633,
      "learning_rate": 8.465500000000001e-05,
      "loss": 0.535,
      "step": 16940
    },
    {
      "epoch": 0.9486504647776907,
      "grad_norm": 1.4812849760055542,
      "learning_rate": 8.466e-05,
      "loss": 0.4636,
      "step": 16941
    },
    {
      "epoch": 0.9487064620898197,
      "grad_norm": 1.4107130765914917,
      "learning_rate": 8.4665e-05,
      "loss": 0.622,
      "step": 16942
    },
    {
      "epoch": 0.9487624594019487,
      "grad_norm": 1.3292529582977295,
      "learning_rate": 8.467e-05,
      "loss": 0.4732,
      "step": 16943
    },
    {
      "epoch": 0.9488184567140777,
      "grad_norm": 1.2414759397506714,
      "learning_rate": 8.4675e-05,
      "loss": 0.3468,
      "step": 16944
    },
    {
      "epoch": 0.9488744540262067,
      "grad_norm": 1.1517809629440308,
      "learning_rate": 8.468000000000001e-05,
      "loss": 0.3438,
      "step": 16945
    },
    {
      "epoch": 0.9489304513383358,
      "grad_norm": 1.5465229749679565,
      "learning_rate": 8.4685e-05,
      "loss": 0.4394,
      "step": 16946
    },
    {
      "epoch": 0.9489864486504648,
      "grad_norm": 1.4621566534042358,
      "learning_rate": 8.469e-05,
      "loss": 0.3958,
      "step": 16947
    },
    {
      "epoch": 0.9490424459625938,
      "grad_norm": 1.2637304067611694,
      "learning_rate": 8.4695e-05,
      "loss": 0.3675,
      "step": 16948
    },
    {
      "epoch": 0.9490984432747228,
      "grad_norm": 1.4915884733200073,
      "learning_rate": 8.47e-05,
      "loss": 0.4249,
      "step": 16949
    },
    {
      "epoch": 0.9491544405868518,
      "grad_norm": 1.419655203819275,
      "learning_rate": 8.4705e-05,
      "loss": 0.6283,
      "step": 16950
    },
    {
      "epoch": 0.9492104378989809,
      "grad_norm": 1.1955921649932861,
      "learning_rate": 8.471000000000001e-05,
      "loss": 0.48,
      "step": 16951
    },
    {
      "epoch": 0.9492664352111099,
      "grad_norm": 1.340035319328308,
      "learning_rate": 8.4715e-05,
      "loss": 0.6691,
      "step": 16952
    },
    {
      "epoch": 0.9493224325232389,
      "grad_norm": 1.3427761793136597,
      "learning_rate": 8.472e-05,
      "loss": 0.503,
      "step": 16953
    },
    {
      "epoch": 0.9493784298353679,
      "grad_norm": 1.4067109823226929,
      "learning_rate": 8.4725e-05,
      "loss": 0.6549,
      "step": 16954
    },
    {
      "epoch": 0.9494344271474969,
      "grad_norm": 1.3630707263946533,
      "learning_rate": 8.473000000000001e-05,
      "loss": 0.4791,
      "step": 16955
    },
    {
      "epoch": 0.949490424459626,
      "grad_norm": 1.6496031284332275,
      "learning_rate": 8.473500000000001e-05,
      "loss": 0.6371,
      "step": 16956
    },
    {
      "epoch": 0.949546421771755,
      "grad_norm": 1.4741486310958862,
      "learning_rate": 8.474e-05,
      "loss": 0.3872,
      "step": 16957
    },
    {
      "epoch": 0.949602419083884,
      "grad_norm": 1.1507982015609741,
      "learning_rate": 8.4745e-05,
      "loss": 0.4622,
      "step": 16958
    },
    {
      "epoch": 0.949658416396013,
      "grad_norm": 1.512468695640564,
      "learning_rate": 8.475000000000001e-05,
      "loss": 0.6612,
      "step": 16959
    },
    {
      "epoch": 0.949714413708142,
      "grad_norm": 1.4250510931015015,
      "learning_rate": 8.475500000000001e-05,
      "loss": 0.4561,
      "step": 16960
    },
    {
      "epoch": 0.949770411020271,
      "grad_norm": 1.3143805265426636,
      "learning_rate": 8.476000000000001e-05,
      "loss": 0.4001,
      "step": 16961
    },
    {
      "epoch": 0.9498264083324001,
      "grad_norm": 1.4907798767089844,
      "learning_rate": 8.4765e-05,
      "loss": 0.5428,
      "step": 16962
    },
    {
      "epoch": 0.9498824056445291,
      "grad_norm": 1.3315340280532837,
      "learning_rate": 8.477e-05,
      "loss": 0.4319,
      "step": 16963
    },
    {
      "epoch": 0.9499384029566581,
      "grad_norm": 1.20774507522583,
      "learning_rate": 8.4775e-05,
      "loss": 0.4153,
      "step": 16964
    },
    {
      "epoch": 0.9499944002687871,
      "grad_norm": 1.1649962663650513,
      "learning_rate": 8.478e-05,
      "loss": 0.4366,
      "step": 16965
    },
    {
      "epoch": 0.9500503975809161,
      "grad_norm": 1.3137903213500977,
      "learning_rate": 8.478500000000001e-05,
      "loss": 0.4325,
      "step": 16966
    },
    {
      "epoch": 0.9501063948930452,
      "grad_norm": 1.3243414163589478,
      "learning_rate": 8.479e-05,
      "loss": 0.3953,
      "step": 16967
    },
    {
      "epoch": 0.9501623922051742,
      "grad_norm": 1.4575765132904053,
      "learning_rate": 8.4795e-05,
      "loss": 0.4123,
      "step": 16968
    },
    {
      "epoch": 0.9502183895173032,
      "grad_norm": 1.3398209810256958,
      "learning_rate": 8.48e-05,
      "loss": 0.4228,
      "step": 16969
    },
    {
      "epoch": 0.9502743868294322,
      "grad_norm": 1.8145989179611206,
      "learning_rate": 8.4805e-05,
      "loss": 0.5517,
      "step": 16970
    },
    {
      "epoch": 0.9503303841415612,
      "grad_norm": 1.2563401460647583,
      "learning_rate": 8.481e-05,
      "loss": 0.4388,
      "step": 16971
    },
    {
      "epoch": 0.9503863814536903,
      "grad_norm": 1.535429835319519,
      "learning_rate": 8.4815e-05,
      "loss": 0.4084,
      "step": 16972
    },
    {
      "epoch": 0.9504423787658193,
      "grad_norm": 1.1452304124832153,
      "learning_rate": 8.482e-05,
      "loss": 0.3845,
      "step": 16973
    },
    {
      "epoch": 0.9504983760779483,
      "grad_norm": 1.621180772781372,
      "learning_rate": 8.4825e-05,
      "loss": 0.4521,
      "step": 16974
    },
    {
      "epoch": 0.9505543733900773,
      "grad_norm": 1.3514586687088013,
      "learning_rate": 8.483000000000001e-05,
      "loss": 0.6353,
      "step": 16975
    },
    {
      "epoch": 0.9506103707022063,
      "grad_norm": 1.2678074836730957,
      "learning_rate": 8.483500000000001e-05,
      "loss": 0.4231,
      "step": 16976
    },
    {
      "epoch": 0.9506663680143354,
      "grad_norm": 2.2612082958221436,
      "learning_rate": 8.484000000000001e-05,
      "loss": 0.4763,
      "step": 16977
    },
    {
      "epoch": 0.9507223653264644,
      "grad_norm": 1.457580327987671,
      "learning_rate": 8.4845e-05,
      "loss": 0.5832,
      "step": 16978
    },
    {
      "epoch": 0.9507783626385934,
      "grad_norm": 3.180603265762329,
      "learning_rate": 8.485e-05,
      "loss": 0.4155,
      "step": 16979
    },
    {
      "epoch": 0.9508343599507224,
      "grad_norm": 1.9469780921936035,
      "learning_rate": 8.485500000000001e-05,
      "loss": 0.3662,
      "step": 16980
    },
    {
      "epoch": 0.9508903572628514,
      "grad_norm": 1.345381498336792,
      "learning_rate": 8.486000000000001e-05,
      "loss": 0.4756,
      "step": 16981
    },
    {
      "epoch": 0.9509463545749804,
      "grad_norm": 1.5418356657028198,
      "learning_rate": 8.486500000000001e-05,
      "loss": 0.4512,
      "step": 16982
    },
    {
      "epoch": 0.9510023518871095,
      "grad_norm": 1.2618006467819214,
      "learning_rate": 8.487e-05,
      "loss": 0.5224,
      "step": 16983
    },
    {
      "epoch": 0.9510583491992385,
      "grad_norm": 1.2593483924865723,
      "learning_rate": 8.4875e-05,
      "loss": 0.4596,
      "step": 16984
    },
    {
      "epoch": 0.9511143465113675,
      "grad_norm": 1.267127513885498,
      "learning_rate": 8.488e-05,
      "loss": 0.5895,
      "step": 16985
    },
    {
      "epoch": 0.9511703438234965,
      "grad_norm": 1.5504367351531982,
      "learning_rate": 8.4885e-05,
      "loss": 0.5344,
      "step": 16986
    },
    {
      "epoch": 0.9512263411356255,
      "grad_norm": 1.2736339569091797,
      "learning_rate": 8.489000000000001e-05,
      "loss": 0.4524,
      "step": 16987
    },
    {
      "epoch": 0.9512823384477546,
      "grad_norm": 1.1661581993103027,
      "learning_rate": 8.4895e-05,
      "loss": 0.343,
      "step": 16988
    },
    {
      "epoch": 0.9513383357598836,
      "grad_norm": 1.334139347076416,
      "learning_rate": 8.49e-05,
      "loss": 0.5219,
      "step": 16989
    },
    {
      "epoch": 0.9513943330720125,
      "grad_norm": 1.4772909879684448,
      "learning_rate": 8.4905e-05,
      "loss": 0.555,
      "step": 16990
    },
    {
      "epoch": 0.9514503303841415,
      "grad_norm": 1.1652812957763672,
      "learning_rate": 8.491e-05,
      "loss": 0.3417,
      "step": 16991
    },
    {
      "epoch": 0.9515063276962705,
      "grad_norm": 1.2831329107284546,
      "learning_rate": 8.4915e-05,
      "loss": 0.4816,
      "step": 16992
    },
    {
      "epoch": 0.9515623250083995,
      "grad_norm": 1.5837223529815674,
      "learning_rate": 8.492e-05,
      "loss": 0.4307,
      "step": 16993
    },
    {
      "epoch": 0.9516183223205286,
      "grad_norm": 1.555816888809204,
      "learning_rate": 8.4925e-05,
      "loss": 0.4459,
      "step": 16994
    },
    {
      "epoch": 0.9516743196326576,
      "grad_norm": 1.3275021314620972,
      "learning_rate": 8.493000000000002e-05,
      "loss": 0.4356,
      "step": 16995
    },
    {
      "epoch": 0.9517303169447866,
      "grad_norm": 1.6596986055374146,
      "learning_rate": 8.493500000000001e-05,
      "loss": 0.4259,
      "step": 16996
    },
    {
      "epoch": 0.9517863142569156,
      "grad_norm": 1.3667863607406616,
      "learning_rate": 8.494000000000001e-05,
      "loss": 0.5015,
      "step": 16997
    },
    {
      "epoch": 0.9518423115690446,
      "grad_norm": 1.5858535766601562,
      "learning_rate": 8.494500000000001e-05,
      "loss": 0.3896,
      "step": 16998
    },
    {
      "epoch": 0.9518983088811737,
      "grad_norm": 1.3232752084732056,
      "learning_rate": 8.495e-05,
      "loss": 0.3544,
      "step": 16999
    },
    {
      "epoch": 0.9519543061933027,
      "grad_norm": 1.571776270866394,
      "learning_rate": 8.4955e-05,
      "loss": 0.4382,
      "step": 17000
    },
    {
      "epoch": 0.9520103035054317,
      "grad_norm": 1.4432038068771362,
      "learning_rate": 8.496e-05,
      "loss": 0.4218,
      "step": 17001
    },
    {
      "epoch": 0.9520663008175607,
      "grad_norm": 1.3933186531066895,
      "learning_rate": 8.496500000000001e-05,
      "loss": 0.4635,
      "step": 17002
    },
    {
      "epoch": 0.9521222981296897,
      "grad_norm": 1.338154911994934,
      "learning_rate": 8.497000000000001e-05,
      "loss": 0.4775,
      "step": 17003
    },
    {
      "epoch": 0.9521782954418188,
      "grad_norm": 1.3936713933944702,
      "learning_rate": 8.4975e-05,
      "loss": 0.5565,
      "step": 17004
    },
    {
      "epoch": 0.9522342927539478,
      "grad_norm": 1.5405932664871216,
      "learning_rate": 8.498e-05,
      "loss": 0.5839,
      "step": 17005
    },
    {
      "epoch": 0.9522902900660768,
      "grad_norm": 1.1904425621032715,
      "learning_rate": 8.4985e-05,
      "loss": 0.3904,
      "step": 17006
    },
    {
      "epoch": 0.9523462873782058,
      "grad_norm": 1.8471018075942993,
      "learning_rate": 8.499e-05,
      "loss": 0.4705,
      "step": 17007
    },
    {
      "epoch": 0.9524022846903348,
      "grad_norm": 1.429958462715149,
      "learning_rate": 8.499500000000001e-05,
      "loss": 0.4797,
      "step": 17008
    },
    {
      "epoch": 0.9524582820024639,
      "grad_norm": 1.300558090209961,
      "learning_rate": 8.5e-05,
      "loss": 0.3766,
      "step": 17009
    },
    {
      "epoch": 0.9525142793145929,
      "grad_norm": 1.1773066520690918,
      "learning_rate": 8.5005e-05,
      "loss": 0.4107,
      "step": 17010
    },
    {
      "epoch": 0.9525702766267219,
      "grad_norm": 1.355825662612915,
      "learning_rate": 8.501e-05,
      "loss": 0.6292,
      "step": 17011
    },
    {
      "epoch": 0.9526262739388509,
      "grad_norm": 1.293701171875,
      "learning_rate": 8.5015e-05,
      "loss": 0.4225,
      "step": 17012
    },
    {
      "epoch": 0.9526822712509799,
      "grad_norm": 1.2822965383529663,
      "learning_rate": 8.502e-05,
      "loss": 0.5179,
      "step": 17013
    },
    {
      "epoch": 0.952738268563109,
      "grad_norm": 1.3549543619155884,
      "learning_rate": 8.502499999999999e-05,
      "loss": 0.4233,
      "step": 17014
    },
    {
      "epoch": 0.952794265875238,
      "grad_norm": 1.1742892265319824,
      "learning_rate": 8.503e-05,
      "loss": 0.3484,
      "step": 17015
    },
    {
      "epoch": 0.952850263187367,
      "grad_norm": 1.4947805404663086,
      "learning_rate": 8.503500000000002e-05,
      "loss": 0.4144,
      "step": 17016
    },
    {
      "epoch": 0.952906260499496,
      "grad_norm": 2.0162227153778076,
      "learning_rate": 8.504000000000001e-05,
      "loss": 0.5194,
      "step": 17017
    },
    {
      "epoch": 0.952962257811625,
      "grad_norm": 1.4323337078094482,
      "learning_rate": 8.504500000000001e-05,
      "loss": 0.5476,
      "step": 17018
    },
    {
      "epoch": 0.953018255123754,
      "grad_norm": 1.0337882041931152,
      "learning_rate": 8.505000000000001e-05,
      "loss": 0.3104,
      "step": 17019
    },
    {
      "epoch": 0.9530742524358831,
      "grad_norm": 1.3247967958450317,
      "learning_rate": 8.5055e-05,
      "loss": 0.4969,
      "step": 17020
    },
    {
      "epoch": 0.9531302497480121,
      "grad_norm": 1.4347736835479736,
      "learning_rate": 8.506e-05,
      "loss": 0.4299,
      "step": 17021
    },
    {
      "epoch": 0.9531862470601411,
      "grad_norm": 1.526903510093689,
      "learning_rate": 8.5065e-05,
      "loss": 0.3932,
      "step": 17022
    },
    {
      "epoch": 0.9532422443722701,
      "grad_norm": 1.2358578443527222,
      "learning_rate": 8.507000000000001e-05,
      "loss": 0.3121,
      "step": 17023
    },
    {
      "epoch": 0.9532982416843991,
      "grad_norm": 2.889700174331665,
      "learning_rate": 8.507500000000001e-05,
      "loss": 0.7682,
      "step": 17024
    },
    {
      "epoch": 0.9533542389965282,
      "grad_norm": 1.4642601013183594,
      "learning_rate": 8.508e-05,
      "loss": 0.3698,
      "step": 17025
    },
    {
      "epoch": 0.9534102363086572,
      "grad_norm": 1.27633798122406,
      "learning_rate": 8.5085e-05,
      "loss": 0.4462,
      "step": 17026
    },
    {
      "epoch": 0.9534662336207862,
      "grad_norm": 1.7518889904022217,
      "learning_rate": 8.509e-05,
      "loss": 0.5751,
      "step": 17027
    },
    {
      "epoch": 0.9535222309329152,
      "grad_norm": 1.443544864654541,
      "learning_rate": 8.5095e-05,
      "loss": 0.378,
      "step": 17028
    },
    {
      "epoch": 0.9535782282450442,
      "grad_norm": 1.3720136880874634,
      "learning_rate": 8.510000000000001e-05,
      "loss": 0.4845,
      "step": 17029
    },
    {
      "epoch": 0.9536342255571733,
      "grad_norm": 1.889662265777588,
      "learning_rate": 8.5105e-05,
      "loss": 0.4681,
      "step": 17030
    },
    {
      "epoch": 0.9536902228693023,
      "grad_norm": 1.6645822525024414,
      "learning_rate": 8.511e-05,
      "loss": 0.3319,
      "step": 17031
    },
    {
      "epoch": 0.9537462201814313,
      "grad_norm": 1.5410239696502686,
      "learning_rate": 8.5115e-05,
      "loss": 0.527,
      "step": 17032
    },
    {
      "epoch": 0.9538022174935603,
      "grad_norm": 1.3207294940948486,
      "learning_rate": 8.512e-05,
      "loss": 0.4837,
      "step": 17033
    },
    {
      "epoch": 0.9538582148056893,
      "grad_norm": 1.1162564754486084,
      "learning_rate": 8.5125e-05,
      "loss": 0.3487,
      "step": 17034
    },
    {
      "epoch": 0.9539142121178183,
      "grad_norm": 1.2722265720367432,
      "learning_rate": 8.512999999999999e-05,
      "loss": 0.4862,
      "step": 17035
    },
    {
      "epoch": 0.9539702094299474,
      "grad_norm": 1.4618171453475952,
      "learning_rate": 8.5135e-05,
      "loss": 0.4082,
      "step": 17036
    },
    {
      "epoch": 0.9540262067420764,
      "grad_norm": 1.6251577138900757,
      "learning_rate": 8.514000000000001e-05,
      "loss": 0.4704,
      "step": 17037
    },
    {
      "epoch": 0.9540822040542054,
      "grad_norm": 1.2648488283157349,
      "learning_rate": 8.514500000000001e-05,
      "loss": 0.3907,
      "step": 17038
    },
    {
      "epoch": 0.9541382013663344,
      "grad_norm": 1.171286702156067,
      "learning_rate": 8.515000000000001e-05,
      "loss": 0.36,
      "step": 17039
    },
    {
      "epoch": 0.9541941986784634,
      "grad_norm": 1.8883570432662964,
      "learning_rate": 8.515500000000001e-05,
      "loss": 0.4311,
      "step": 17040
    },
    {
      "epoch": 0.9542501959905925,
      "grad_norm": 1.5154889822006226,
      "learning_rate": 8.516e-05,
      "loss": 0.6413,
      "step": 17041
    },
    {
      "epoch": 0.9543061933027215,
      "grad_norm": 1.3099116086959839,
      "learning_rate": 8.5165e-05,
      "loss": 0.4196,
      "step": 17042
    },
    {
      "epoch": 0.9543621906148505,
      "grad_norm": 1.5377181768417358,
      "learning_rate": 8.517e-05,
      "loss": 0.5744,
      "step": 17043
    },
    {
      "epoch": 0.9544181879269795,
      "grad_norm": 1.5144118070602417,
      "learning_rate": 8.517500000000001e-05,
      "loss": 0.3586,
      "step": 17044
    },
    {
      "epoch": 0.9544741852391085,
      "grad_norm": 1.6528390645980835,
      "learning_rate": 8.518000000000001e-05,
      "loss": 0.4618,
      "step": 17045
    },
    {
      "epoch": 0.9545301825512376,
      "grad_norm": 2.705566883087158,
      "learning_rate": 8.5185e-05,
      "loss": 0.3898,
      "step": 17046
    },
    {
      "epoch": 0.9545861798633666,
      "grad_norm": 1.237256646156311,
      "learning_rate": 8.519e-05,
      "loss": 0.4158,
      "step": 17047
    },
    {
      "epoch": 0.9546421771754956,
      "grad_norm": 1.4359909296035767,
      "learning_rate": 8.5195e-05,
      "loss": 0.5538,
      "step": 17048
    },
    {
      "epoch": 0.9546981744876246,
      "grad_norm": 1.3809127807617188,
      "learning_rate": 8.52e-05,
      "loss": 0.4524,
      "step": 17049
    },
    {
      "epoch": 0.9547541717997536,
      "grad_norm": 1.2590758800506592,
      "learning_rate": 8.5205e-05,
      "loss": 0.4239,
      "step": 17050
    },
    {
      "epoch": 0.9548101691118827,
      "grad_norm": 1.2650855779647827,
      "learning_rate": 8.521e-05,
      "loss": 0.5073,
      "step": 17051
    },
    {
      "epoch": 0.9548661664240117,
      "grad_norm": 1.4202791452407837,
      "learning_rate": 8.5215e-05,
      "loss": 0.4295,
      "step": 17052
    },
    {
      "epoch": 0.9549221637361407,
      "grad_norm": 1.4298592805862427,
      "learning_rate": 8.522e-05,
      "loss": 0.645,
      "step": 17053
    },
    {
      "epoch": 0.9549781610482697,
      "grad_norm": 1.6812329292297363,
      "learning_rate": 8.5225e-05,
      "loss": 0.4883,
      "step": 17054
    },
    {
      "epoch": 0.9550341583603987,
      "grad_norm": 1.1203745603561401,
      "learning_rate": 8.523e-05,
      "loss": 0.4202,
      "step": 17055
    },
    {
      "epoch": 0.9550901556725278,
      "grad_norm": 1.4017579555511475,
      "learning_rate": 8.5235e-05,
      "loss": 0.3603,
      "step": 17056
    },
    {
      "epoch": 0.9551461529846568,
      "grad_norm": 1.3865840435028076,
      "learning_rate": 8.524e-05,
      "loss": 0.436,
      "step": 17057
    },
    {
      "epoch": 0.9552021502967858,
      "grad_norm": 1.513572096824646,
      "learning_rate": 8.524500000000001e-05,
      "loss": 0.4439,
      "step": 17058
    },
    {
      "epoch": 0.9552581476089148,
      "grad_norm": 1.321823000907898,
      "learning_rate": 8.525000000000001e-05,
      "loss": 0.4888,
      "step": 17059
    },
    {
      "epoch": 0.9553141449210438,
      "grad_norm": 1.6148601770401,
      "learning_rate": 8.525500000000001e-05,
      "loss": 0.4308,
      "step": 17060
    },
    {
      "epoch": 0.9553701422331728,
      "grad_norm": 1.5410012006759644,
      "learning_rate": 8.526000000000001e-05,
      "loss": 0.5282,
      "step": 17061
    },
    {
      "epoch": 0.9554261395453019,
      "grad_norm": 1.0875070095062256,
      "learning_rate": 8.5265e-05,
      "loss": 0.3218,
      "step": 17062
    },
    {
      "epoch": 0.9554821368574309,
      "grad_norm": 1.3766624927520752,
      "learning_rate": 8.527e-05,
      "loss": 0.4481,
      "step": 17063
    },
    {
      "epoch": 0.9555381341695599,
      "grad_norm": 1.3956143856048584,
      "learning_rate": 8.5275e-05,
      "loss": 0.4638,
      "step": 17064
    },
    {
      "epoch": 0.9555941314816889,
      "grad_norm": 1.3810175657272339,
      "learning_rate": 8.528000000000001e-05,
      "loss": 0.579,
      "step": 17065
    },
    {
      "epoch": 0.9556501287938179,
      "grad_norm": 1.3077311515808105,
      "learning_rate": 8.528500000000001e-05,
      "loss": 0.3388,
      "step": 17066
    },
    {
      "epoch": 0.955706126105947,
      "grad_norm": 2.1643025875091553,
      "learning_rate": 8.529e-05,
      "loss": 0.4967,
      "step": 17067
    },
    {
      "epoch": 0.955762123418076,
      "grad_norm": 1.6373200416564941,
      "learning_rate": 8.5295e-05,
      "loss": 0.5301,
      "step": 17068
    },
    {
      "epoch": 0.955818120730205,
      "grad_norm": 1.2569293975830078,
      "learning_rate": 8.53e-05,
      "loss": 0.3281,
      "step": 17069
    },
    {
      "epoch": 0.955874118042334,
      "grad_norm": 1.2961771488189697,
      "learning_rate": 8.5305e-05,
      "loss": 0.5222,
      "step": 17070
    },
    {
      "epoch": 0.955930115354463,
      "grad_norm": 1.2537667751312256,
      "learning_rate": 8.531e-05,
      "loss": 0.5211,
      "step": 17071
    },
    {
      "epoch": 0.9559861126665921,
      "grad_norm": 1.3391064405441284,
      "learning_rate": 8.5315e-05,
      "loss": 0.4964,
      "step": 17072
    },
    {
      "epoch": 0.956042109978721,
      "grad_norm": 1.6892999410629272,
      "learning_rate": 8.532e-05,
      "loss": 0.7413,
      "step": 17073
    },
    {
      "epoch": 0.95609810729085,
      "grad_norm": 1.190972924232483,
      "learning_rate": 8.5325e-05,
      "loss": 0.4132,
      "step": 17074
    },
    {
      "epoch": 0.956154104602979,
      "grad_norm": 1.376509666442871,
      "learning_rate": 8.533e-05,
      "loss": 0.4137,
      "step": 17075
    },
    {
      "epoch": 0.956210101915108,
      "grad_norm": 1.2693827152252197,
      "learning_rate": 8.533500000000001e-05,
      "loss": 0.3201,
      "step": 17076
    },
    {
      "epoch": 0.956266099227237,
      "grad_norm": 1.4007991552352905,
      "learning_rate": 8.534e-05,
      "loss": 0.4881,
      "step": 17077
    },
    {
      "epoch": 0.9563220965393661,
      "grad_norm": 1.4203362464904785,
      "learning_rate": 8.5345e-05,
      "loss": 0.5946,
      "step": 17078
    },
    {
      "epoch": 0.9563780938514951,
      "grad_norm": 1.3453959226608276,
      "learning_rate": 8.535e-05,
      "loss": 0.4659,
      "step": 17079
    },
    {
      "epoch": 0.9564340911636241,
      "grad_norm": 1.370445966720581,
      "learning_rate": 8.535500000000001e-05,
      "loss": 0.3657,
      "step": 17080
    },
    {
      "epoch": 0.9564900884757531,
      "grad_norm": 1.295682430267334,
      "learning_rate": 8.536000000000001e-05,
      "loss": 0.2921,
      "step": 17081
    },
    {
      "epoch": 0.9565460857878821,
      "grad_norm": 1.332443356513977,
      "learning_rate": 8.536500000000001e-05,
      "loss": 0.3777,
      "step": 17082
    },
    {
      "epoch": 0.9566020831000112,
      "grad_norm": 1.307560682296753,
      "learning_rate": 8.537e-05,
      "loss": 0.5362,
      "step": 17083
    },
    {
      "epoch": 0.9566580804121402,
      "grad_norm": 1.2536908388137817,
      "learning_rate": 8.5375e-05,
      "loss": 0.4425,
      "step": 17084
    },
    {
      "epoch": 0.9567140777242692,
      "grad_norm": 1.1129144430160522,
      "learning_rate": 8.538e-05,
      "loss": 0.3942,
      "step": 17085
    },
    {
      "epoch": 0.9567700750363982,
      "grad_norm": 1.0530842542648315,
      "learning_rate": 8.538500000000001e-05,
      "loss": 0.4956,
      "step": 17086
    },
    {
      "epoch": 0.9568260723485272,
      "grad_norm": 1.2498310804367065,
      "learning_rate": 8.539000000000001e-05,
      "loss": 0.4406,
      "step": 17087
    },
    {
      "epoch": 0.9568820696606563,
      "grad_norm": 1.247690200805664,
      "learning_rate": 8.5395e-05,
      "loss": 0.3611,
      "step": 17088
    },
    {
      "epoch": 0.9569380669727853,
      "grad_norm": 1.384846568107605,
      "learning_rate": 8.54e-05,
      "loss": 0.5635,
      "step": 17089
    },
    {
      "epoch": 0.9569940642849143,
      "grad_norm": 1.3289271593093872,
      "learning_rate": 8.5405e-05,
      "loss": 0.5475,
      "step": 17090
    },
    {
      "epoch": 0.9570500615970433,
      "grad_norm": 1.2485624551773071,
      "learning_rate": 8.541e-05,
      "loss": 0.4813,
      "step": 17091
    },
    {
      "epoch": 0.9571060589091723,
      "grad_norm": 1.3284590244293213,
      "learning_rate": 8.5415e-05,
      "loss": 0.4166,
      "step": 17092
    },
    {
      "epoch": 0.9571620562213013,
      "grad_norm": 1.539612054824829,
      "learning_rate": 8.542e-05,
      "loss": 0.3568,
      "step": 17093
    },
    {
      "epoch": 0.9572180535334304,
      "grad_norm": 1.638933777809143,
      "learning_rate": 8.5425e-05,
      "loss": 0.4398,
      "step": 17094
    },
    {
      "epoch": 0.9572740508455594,
      "grad_norm": 1.4553638696670532,
      "learning_rate": 8.543e-05,
      "loss": 0.5311,
      "step": 17095
    },
    {
      "epoch": 0.9573300481576884,
      "grad_norm": 1.381110668182373,
      "learning_rate": 8.543500000000001e-05,
      "loss": 0.5229,
      "step": 17096
    },
    {
      "epoch": 0.9573860454698174,
      "grad_norm": 1.2284245491027832,
      "learning_rate": 8.544000000000001e-05,
      "loss": 0.4791,
      "step": 17097
    },
    {
      "epoch": 0.9574420427819464,
      "grad_norm": 1.2967196702957153,
      "learning_rate": 8.5445e-05,
      "loss": 0.3985,
      "step": 17098
    },
    {
      "epoch": 0.9574980400940755,
      "grad_norm": 1.2861863374710083,
      "learning_rate": 8.545e-05,
      "loss": 0.4384,
      "step": 17099
    },
    {
      "epoch": 0.9575540374062045,
      "grad_norm": 1.610902190208435,
      "learning_rate": 8.5455e-05,
      "loss": 0.5805,
      "step": 17100
    },
    {
      "epoch": 0.9576100347183335,
      "grad_norm": 1.3004497289657593,
      "learning_rate": 8.546000000000001e-05,
      "loss": 0.4394,
      "step": 17101
    },
    {
      "epoch": 0.9576660320304625,
      "grad_norm": 1.4884735345840454,
      "learning_rate": 8.546500000000001e-05,
      "loss": 0.6698,
      "step": 17102
    },
    {
      "epoch": 0.9577220293425915,
      "grad_norm": 1.7032983303070068,
      "learning_rate": 8.547e-05,
      "loss": 0.6283,
      "step": 17103
    },
    {
      "epoch": 0.9577780266547206,
      "grad_norm": 1.5157755613327026,
      "learning_rate": 8.5475e-05,
      "loss": 0.5819,
      "step": 17104
    },
    {
      "epoch": 0.9578340239668496,
      "grad_norm": 1.2883331775665283,
      "learning_rate": 8.548e-05,
      "loss": 0.4398,
      "step": 17105
    },
    {
      "epoch": 0.9578900212789786,
      "grad_norm": 1.4675449132919312,
      "learning_rate": 8.5485e-05,
      "loss": 0.385,
      "step": 17106
    },
    {
      "epoch": 0.9579460185911076,
      "grad_norm": 1.3032020330429077,
      "learning_rate": 8.549000000000001e-05,
      "loss": 0.4696,
      "step": 17107
    },
    {
      "epoch": 0.9580020159032366,
      "grad_norm": 1.1724374294281006,
      "learning_rate": 8.549500000000001e-05,
      "loss": 0.369,
      "step": 17108
    },
    {
      "epoch": 0.9580580132153657,
      "grad_norm": 1.5858454704284668,
      "learning_rate": 8.55e-05,
      "loss": 0.5147,
      "step": 17109
    },
    {
      "epoch": 0.9581140105274947,
      "grad_norm": 1.1673816442489624,
      "learning_rate": 8.5505e-05,
      "loss": 0.3849,
      "step": 17110
    },
    {
      "epoch": 0.9581700078396237,
      "grad_norm": 1.6120600700378418,
      "learning_rate": 8.551e-05,
      "loss": 0.3709,
      "step": 17111
    },
    {
      "epoch": 0.9582260051517527,
      "grad_norm": 1.4594321250915527,
      "learning_rate": 8.5515e-05,
      "loss": 0.5787,
      "step": 17112
    },
    {
      "epoch": 0.9582820024638817,
      "grad_norm": 1.3463350534439087,
      "learning_rate": 8.552e-05,
      "loss": 0.4482,
      "step": 17113
    },
    {
      "epoch": 0.9583379997760108,
      "grad_norm": 1.336738109588623,
      "learning_rate": 8.5525e-05,
      "loss": 0.4269,
      "step": 17114
    },
    {
      "epoch": 0.9583939970881398,
      "grad_norm": 1.1716670989990234,
      "learning_rate": 8.553e-05,
      "loss": 0.4297,
      "step": 17115
    },
    {
      "epoch": 0.9584499944002688,
      "grad_norm": 1.1067936420440674,
      "learning_rate": 8.553500000000001e-05,
      "loss": 0.3488,
      "step": 17116
    },
    {
      "epoch": 0.9585059917123978,
      "grad_norm": 1.4455829858779907,
      "learning_rate": 8.554000000000001e-05,
      "loss": 0.4391,
      "step": 17117
    },
    {
      "epoch": 0.9585619890245268,
      "grad_norm": 1.3242000341415405,
      "learning_rate": 8.554500000000001e-05,
      "loss": 0.4274,
      "step": 17118
    },
    {
      "epoch": 0.9586179863366558,
      "grad_norm": 1.7039105892181396,
      "learning_rate": 8.555e-05,
      "loss": 0.4254,
      "step": 17119
    },
    {
      "epoch": 0.9586739836487849,
      "grad_norm": 1.2343722581863403,
      "learning_rate": 8.5555e-05,
      "loss": 0.4459,
      "step": 17120
    },
    {
      "epoch": 0.9587299809609139,
      "grad_norm": 1.8182319402694702,
      "learning_rate": 8.556e-05,
      "loss": 0.4805,
      "step": 17121
    },
    {
      "epoch": 0.9587859782730429,
      "grad_norm": 1.6502269506454468,
      "learning_rate": 8.556500000000001e-05,
      "loss": 0.5003,
      "step": 17122
    },
    {
      "epoch": 0.9588419755851719,
      "grad_norm": 1.1833775043487549,
      "learning_rate": 8.557000000000001e-05,
      "loss": 0.4476,
      "step": 17123
    },
    {
      "epoch": 0.9588979728973009,
      "grad_norm": 1.6448407173156738,
      "learning_rate": 8.5575e-05,
      "loss": 0.3923,
      "step": 17124
    },
    {
      "epoch": 0.95895397020943,
      "grad_norm": 1.1200000047683716,
      "learning_rate": 8.558e-05,
      "loss": 0.3327,
      "step": 17125
    },
    {
      "epoch": 0.959009967521559,
      "grad_norm": 1.3580704927444458,
      "learning_rate": 8.5585e-05,
      "loss": 0.5893,
      "step": 17126
    },
    {
      "epoch": 0.959065964833688,
      "grad_norm": 1.5426150560379028,
      "learning_rate": 8.559e-05,
      "loss": 0.5055,
      "step": 17127
    },
    {
      "epoch": 0.959121962145817,
      "grad_norm": 1.2711679935455322,
      "learning_rate": 8.559500000000001e-05,
      "loss": 0.4388,
      "step": 17128
    },
    {
      "epoch": 0.959177959457946,
      "grad_norm": 1.3287243843078613,
      "learning_rate": 8.560000000000001e-05,
      "loss": 0.4177,
      "step": 17129
    },
    {
      "epoch": 0.9592339567700751,
      "grad_norm": 1.3373806476593018,
      "learning_rate": 8.5605e-05,
      "loss": 0.4425,
      "step": 17130
    },
    {
      "epoch": 0.9592899540822041,
      "grad_norm": 1.3446122407913208,
      "learning_rate": 8.561e-05,
      "loss": 0.4756,
      "step": 17131
    },
    {
      "epoch": 0.9593459513943331,
      "grad_norm": 1.3530346155166626,
      "learning_rate": 8.5615e-05,
      "loss": 0.365,
      "step": 17132
    },
    {
      "epoch": 0.9594019487064621,
      "grad_norm": 1.2455790042877197,
      "learning_rate": 8.562e-05,
      "loss": 0.5752,
      "step": 17133
    },
    {
      "epoch": 0.9594579460185911,
      "grad_norm": 1.099302053451538,
      "learning_rate": 8.5625e-05,
      "loss": 0.3418,
      "step": 17134
    },
    {
      "epoch": 0.9595139433307202,
      "grad_norm": 1.6025974750518799,
      "learning_rate": 8.563e-05,
      "loss": 0.4204,
      "step": 17135
    },
    {
      "epoch": 0.9595699406428492,
      "grad_norm": 1.3592160940170288,
      "learning_rate": 8.5635e-05,
      "loss": 0.524,
      "step": 17136
    },
    {
      "epoch": 0.9596259379549782,
      "grad_norm": 1.4095449447631836,
      "learning_rate": 8.564000000000001e-05,
      "loss": 0.4715,
      "step": 17137
    },
    {
      "epoch": 0.9596819352671072,
      "grad_norm": 1.1044148206710815,
      "learning_rate": 8.564500000000001e-05,
      "loss": 0.5169,
      "step": 17138
    },
    {
      "epoch": 0.9597379325792362,
      "grad_norm": 1.1242592334747314,
      "learning_rate": 8.565000000000001e-05,
      "loss": 0.4757,
      "step": 17139
    },
    {
      "epoch": 0.9597939298913652,
      "grad_norm": 1.244508147239685,
      "learning_rate": 8.5655e-05,
      "loss": 0.3963,
      "step": 17140
    },
    {
      "epoch": 0.9598499272034943,
      "grad_norm": 1.54185152053833,
      "learning_rate": 8.566e-05,
      "loss": 0.4904,
      "step": 17141
    },
    {
      "epoch": 0.9599059245156233,
      "grad_norm": 1.4076370000839233,
      "learning_rate": 8.5665e-05,
      "loss": 0.4381,
      "step": 17142
    },
    {
      "epoch": 0.9599619218277523,
      "grad_norm": 1.2301522493362427,
      "learning_rate": 8.567000000000001e-05,
      "loss": 0.4052,
      "step": 17143
    },
    {
      "epoch": 0.9600179191398813,
      "grad_norm": 1.486867070198059,
      "learning_rate": 8.567500000000001e-05,
      "loss": 0.4805,
      "step": 17144
    },
    {
      "epoch": 0.9600739164520103,
      "grad_norm": 1.4362130165100098,
      "learning_rate": 8.568e-05,
      "loss": 0.3875,
      "step": 17145
    },
    {
      "epoch": 0.9601299137641394,
      "grad_norm": 1.2532141208648682,
      "learning_rate": 8.5685e-05,
      "loss": 0.4779,
      "step": 17146
    },
    {
      "epoch": 0.9601859110762684,
      "grad_norm": 1.3376730680465698,
      "learning_rate": 8.569e-05,
      "loss": 0.4556,
      "step": 17147
    },
    {
      "epoch": 0.9602419083883974,
      "grad_norm": 1.3041802644729614,
      "learning_rate": 8.5695e-05,
      "loss": 0.6248,
      "step": 17148
    },
    {
      "epoch": 0.9602979057005264,
      "grad_norm": 3.7895519733428955,
      "learning_rate": 8.57e-05,
      "loss": 0.3966,
      "step": 17149
    },
    {
      "epoch": 0.9603539030126554,
      "grad_norm": 1.2806825637817383,
      "learning_rate": 8.570500000000001e-05,
      "loss": 0.5187,
      "step": 17150
    },
    {
      "epoch": 0.9604099003247845,
      "grad_norm": 1.4069244861602783,
      "learning_rate": 8.571e-05,
      "loss": 0.5099,
      "step": 17151
    },
    {
      "epoch": 0.9604658976369135,
      "grad_norm": 1.3413307666778564,
      "learning_rate": 8.5715e-05,
      "loss": 0.3637,
      "step": 17152
    },
    {
      "epoch": 0.9605218949490425,
      "grad_norm": 1.2224640846252441,
      "learning_rate": 8.572e-05,
      "loss": 0.4022,
      "step": 17153
    },
    {
      "epoch": 0.9605778922611715,
      "grad_norm": 1.2687058448791504,
      "learning_rate": 8.5725e-05,
      "loss": 0.532,
      "step": 17154
    },
    {
      "epoch": 0.9606338895733005,
      "grad_norm": 1.3773020505905151,
      "learning_rate": 8.573e-05,
      "loss": 0.4421,
      "step": 17155
    },
    {
      "epoch": 0.9606898868854294,
      "grad_norm": 1.491414189338684,
      "learning_rate": 8.5735e-05,
      "loss": 0.5085,
      "step": 17156
    },
    {
      "epoch": 0.9607458841975585,
      "grad_norm": 1.4372575283050537,
      "learning_rate": 8.574000000000002e-05,
      "loss": 0.4531,
      "step": 17157
    },
    {
      "epoch": 0.9608018815096875,
      "grad_norm": 1.0840134620666504,
      "learning_rate": 8.574500000000001e-05,
      "loss": 0.3929,
      "step": 17158
    },
    {
      "epoch": 0.9608578788218165,
      "grad_norm": 1.4550862312316895,
      "learning_rate": 8.575000000000001e-05,
      "loss": 0.4343,
      "step": 17159
    },
    {
      "epoch": 0.9609138761339455,
      "grad_norm": 1.675118327140808,
      "learning_rate": 8.575500000000001e-05,
      "loss": 0.4982,
      "step": 17160
    },
    {
      "epoch": 0.9609698734460745,
      "grad_norm": 1.2108522653579712,
      "learning_rate": 8.576e-05,
      "loss": 0.3745,
      "step": 17161
    },
    {
      "epoch": 0.9610258707582036,
      "grad_norm": 1.415676474571228,
      "learning_rate": 8.5765e-05,
      "loss": 0.5187,
      "step": 17162
    },
    {
      "epoch": 0.9610818680703326,
      "grad_norm": 1.544421911239624,
      "learning_rate": 8.577e-05,
      "loss": 0.4471,
      "step": 17163
    },
    {
      "epoch": 0.9611378653824616,
      "grad_norm": 1.3246055841445923,
      "learning_rate": 8.577500000000001e-05,
      "loss": 0.4553,
      "step": 17164
    },
    {
      "epoch": 0.9611938626945906,
      "grad_norm": 1.423762321472168,
      "learning_rate": 8.578000000000001e-05,
      "loss": 0.5542,
      "step": 17165
    },
    {
      "epoch": 0.9612498600067196,
      "grad_norm": 1.6242868900299072,
      "learning_rate": 8.5785e-05,
      "loss": 0.4095,
      "step": 17166
    },
    {
      "epoch": 0.9613058573188487,
      "grad_norm": 1.2840968370437622,
      "learning_rate": 8.579e-05,
      "loss": 0.5696,
      "step": 17167
    },
    {
      "epoch": 0.9613618546309777,
      "grad_norm": 1.1831138134002686,
      "learning_rate": 8.5795e-05,
      "loss": 0.3876,
      "step": 17168
    },
    {
      "epoch": 0.9614178519431067,
      "grad_norm": 1.7165281772613525,
      "learning_rate": 8.58e-05,
      "loss": 0.4842,
      "step": 17169
    },
    {
      "epoch": 0.9614738492552357,
      "grad_norm": 1.2815358638763428,
      "learning_rate": 8.5805e-05,
      "loss": 0.46,
      "step": 17170
    },
    {
      "epoch": 0.9615298465673647,
      "grad_norm": 1.4860469102859497,
      "learning_rate": 8.581000000000001e-05,
      "loss": 0.4817,
      "step": 17171
    },
    {
      "epoch": 0.9615858438794938,
      "grad_norm": 1.146111011505127,
      "learning_rate": 8.5815e-05,
      "loss": 0.4267,
      "step": 17172
    },
    {
      "epoch": 0.9616418411916228,
      "grad_norm": 1.5226686000823975,
      "learning_rate": 8.582e-05,
      "loss": 0.3614,
      "step": 17173
    },
    {
      "epoch": 0.9616978385037518,
      "grad_norm": 1.1706711053848267,
      "learning_rate": 8.5825e-05,
      "loss": 0.3884,
      "step": 17174
    },
    {
      "epoch": 0.9617538358158808,
      "grad_norm": 1.2649924755096436,
      "learning_rate": 8.583e-05,
      "loss": 0.317,
      "step": 17175
    },
    {
      "epoch": 0.9618098331280098,
      "grad_norm": 1.4722168445587158,
      "learning_rate": 8.5835e-05,
      "loss": 0.4549,
      "step": 17176
    },
    {
      "epoch": 0.9618658304401388,
      "grad_norm": 1.4678112268447876,
      "learning_rate": 8.584e-05,
      "loss": 0.452,
      "step": 17177
    },
    {
      "epoch": 0.9619218277522679,
      "grad_norm": 1.369017243385315,
      "learning_rate": 8.5845e-05,
      "loss": 0.4349,
      "step": 17178
    },
    {
      "epoch": 0.9619778250643969,
      "grad_norm": 1.6606252193450928,
      "learning_rate": 8.585000000000001e-05,
      "loss": 0.4326,
      "step": 17179
    },
    {
      "epoch": 0.9620338223765259,
      "grad_norm": 1.2879390716552734,
      "learning_rate": 8.585500000000001e-05,
      "loss": 0.4629,
      "step": 17180
    },
    {
      "epoch": 0.9620898196886549,
      "grad_norm": 1.4565268754959106,
      "learning_rate": 8.586000000000001e-05,
      "loss": 0.4547,
      "step": 17181
    },
    {
      "epoch": 0.9621458170007839,
      "grad_norm": 1.0399436950683594,
      "learning_rate": 8.5865e-05,
      "loss": 0.3099,
      "step": 17182
    },
    {
      "epoch": 0.962201814312913,
      "grad_norm": 1.2132089138031006,
      "learning_rate": 8.587e-05,
      "loss": 0.4141,
      "step": 17183
    },
    {
      "epoch": 0.962257811625042,
      "grad_norm": 1.2838457822799683,
      "learning_rate": 8.5875e-05,
      "loss": 0.3578,
      "step": 17184
    },
    {
      "epoch": 0.962313808937171,
      "grad_norm": 1.459041953086853,
      "learning_rate": 8.588000000000001e-05,
      "loss": 0.444,
      "step": 17185
    },
    {
      "epoch": 0.9623698062493,
      "grad_norm": 1.8635562658309937,
      "learning_rate": 8.588500000000001e-05,
      "loss": 0.4549,
      "step": 17186
    },
    {
      "epoch": 0.962425803561429,
      "grad_norm": 1.378082036972046,
      "learning_rate": 8.589e-05,
      "loss": 0.4109,
      "step": 17187
    },
    {
      "epoch": 0.9624818008735581,
      "grad_norm": 1.1600459814071655,
      "learning_rate": 8.5895e-05,
      "loss": 0.3097,
      "step": 17188
    },
    {
      "epoch": 0.9625377981856871,
      "grad_norm": 1.5207675695419312,
      "learning_rate": 8.59e-05,
      "loss": 0.4211,
      "step": 17189
    },
    {
      "epoch": 0.9625937954978161,
      "grad_norm": 1.275919795036316,
      "learning_rate": 8.5905e-05,
      "loss": 0.5067,
      "step": 17190
    },
    {
      "epoch": 0.9626497928099451,
      "grad_norm": 1.1985430717468262,
      "learning_rate": 8.591e-05,
      "loss": 0.4454,
      "step": 17191
    },
    {
      "epoch": 0.9627057901220741,
      "grad_norm": 1.5325583219528198,
      "learning_rate": 8.5915e-05,
      "loss": 0.5317,
      "step": 17192
    },
    {
      "epoch": 0.9627617874342032,
      "grad_norm": 1.1471023559570312,
      "learning_rate": 8.592e-05,
      "loss": 0.388,
      "step": 17193
    },
    {
      "epoch": 0.9628177847463322,
      "grad_norm": 1.355360507965088,
      "learning_rate": 8.5925e-05,
      "loss": 0.5246,
      "step": 17194
    },
    {
      "epoch": 0.9628737820584612,
      "grad_norm": 1.3444854021072388,
      "learning_rate": 8.593e-05,
      "loss": 0.5281,
      "step": 17195
    },
    {
      "epoch": 0.9629297793705902,
      "grad_norm": 1.592160701751709,
      "learning_rate": 8.5935e-05,
      "loss": 0.4489,
      "step": 17196
    },
    {
      "epoch": 0.9629857766827192,
      "grad_norm": 1.5497907400131226,
      "learning_rate": 8.594000000000001e-05,
      "loss": 0.5357,
      "step": 17197
    },
    {
      "epoch": 0.9630417739948482,
      "grad_norm": 1.4236985445022583,
      "learning_rate": 8.5945e-05,
      "loss": 0.4859,
      "step": 17198
    },
    {
      "epoch": 0.9630977713069773,
      "grad_norm": 1.5251559019088745,
      "learning_rate": 8.595e-05,
      "loss": 0.5002,
      "step": 17199
    },
    {
      "epoch": 0.9631537686191063,
      "grad_norm": 1.4532705545425415,
      "learning_rate": 8.595500000000001e-05,
      "loss": 0.5314,
      "step": 17200
    },
    {
      "epoch": 0.9632097659312353,
      "grad_norm": 1.423785924911499,
      "learning_rate": 8.596000000000001e-05,
      "loss": 0.582,
      "step": 17201
    },
    {
      "epoch": 0.9632657632433643,
      "grad_norm": 1.3997337818145752,
      "learning_rate": 8.596500000000001e-05,
      "loss": 0.4636,
      "step": 17202
    },
    {
      "epoch": 0.9633217605554933,
      "grad_norm": 1.3916946649551392,
      "learning_rate": 8.597e-05,
      "loss": 0.4764,
      "step": 17203
    },
    {
      "epoch": 0.9633777578676224,
      "grad_norm": 1.5298632383346558,
      "learning_rate": 8.5975e-05,
      "loss": 0.4478,
      "step": 17204
    },
    {
      "epoch": 0.9634337551797514,
      "grad_norm": 1.5225152969360352,
      "learning_rate": 8.598e-05,
      "loss": 0.5898,
      "step": 17205
    },
    {
      "epoch": 0.9634897524918804,
      "grad_norm": 1.1563955545425415,
      "learning_rate": 8.598500000000001e-05,
      "loss": 0.539,
      "step": 17206
    },
    {
      "epoch": 0.9635457498040094,
      "grad_norm": 1.3186798095703125,
      "learning_rate": 8.599000000000001e-05,
      "loss": 0.472,
      "step": 17207
    },
    {
      "epoch": 0.9636017471161384,
      "grad_norm": 1.4567782878875732,
      "learning_rate": 8.5995e-05,
      "loss": 0.5397,
      "step": 17208
    },
    {
      "epoch": 0.9636577444282675,
      "grad_norm": 1.0257200002670288,
      "learning_rate": 8.6e-05,
      "loss": 0.3696,
      "step": 17209
    },
    {
      "epoch": 0.9637137417403965,
      "grad_norm": 1.7328156232833862,
      "learning_rate": 8.6005e-05,
      "loss": 0.3124,
      "step": 17210
    },
    {
      "epoch": 0.9637697390525255,
      "grad_norm": 1.5354329347610474,
      "learning_rate": 8.601e-05,
      "loss": 0.483,
      "step": 17211
    },
    {
      "epoch": 0.9638257363646545,
      "grad_norm": 1.1250163316726685,
      "learning_rate": 8.6015e-05,
      "loss": 0.4669,
      "step": 17212
    },
    {
      "epoch": 0.9638817336767835,
      "grad_norm": 1.317323088645935,
      "learning_rate": 8.602e-05,
      "loss": 0.3965,
      "step": 17213
    },
    {
      "epoch": 0.9639377309889126,
      "grad_norm": 1.4232598543167114,
      "learning_rate": 8.6025e-05,
      "loss": 0.4235,
      "step": 17214
    },
    {
      "epoch": 0.9639937283010416,
      "grad_norm": 1.5435523986816406,
      "learning_rate": 8.603e-05,
      "loss": 0.4698,
      "step": 17215
    },
    {
      "epoch": 0.9640497256131706,
      "grad_norm": 1.3362163305282593,
      "learning_rate": 8.6035e-05,
      "loss": 0.3945,
      "step": 17216
    },
    {
      "epoch": 0.9641057229252996,
      "grad_norm": 1.4581693410873413,
      "learning_rate": 8.604000000000001e-05,
      "loss": 0.5974,
      "step": 17217
    },
    {
      "epoch": 0.9641617202374286,
      "grad_norm": 1.4092007875442505,
      "learning_rate": 8.604500000000001e-05,
      "loss": 0.5306,
      "step": 17218
    },
    {
      "epoch": 0.9642177175495577,
      "grad_norm": 1.7647417783737183,
      "learning_rate": 8.605e-05,
      "loss": 0.5479,
      "step": 17219
    },
    {
      "epoch": 0.9642737148616867,
      "grad_norm": 1.4271140098571777,
      "learning_rate": 8.6055e-05,
      "loss": 0.5704,
      "step": 17220
    },
    {
      "epoch": 0.9643297121738157,
      "grad_norm": 1.4965245723724365,
      "learning_rate": 8.606000000000001e-05,
      "loss": 0.5712,
      "step": 17221
    },
    {
      "epoch": 0.9643857094859447,
      "grad_norm": 1.320960283279419,
      "learning_rate": 8.606500000000001e-05,
      "loss": 0.5041,
      "step": 17222
    },
    {
      "epoch": 0.9644417067980737,
      "grad_norm": 1.5413872003555298,
      "learning_rate": 8.607000000000001e-05,
      "loss": 0.5382,
      "step": 17223
    },
    {
      "epoch": 0.9644977041102027,
      "grad_norm": 1.509645938873291,
      "learning_rate": 8.6075e-05,
      "loss": 0.3631,
      "step": 17224
    },
    {
      "epoch": 0.9645537014223318,
      "grad_norm": 1.2352864742279053,
      "learning_rate": 8.608e-05,
      "loss": 0.3771,
      "step": 17225
    },
    {
      "epoch": 0.9646096987344608,
      "grad_norm": 1.4004476070404053,
      "learning_rate": 8.6085e-05,
      "loss": 0.4592,
      "step": 17226
    },
    {
      "epoch": 0.9646656960465898,
      "grad_norm": 1.2401401996612549,
      "learning_rate": 8.609e-05,
      "loss": 0.4409,
      "step": 17227
    },
    {
      "epoch": 0.9647216933587188,
      "grad_norm": 1.4171693325042725,
      "learning_rate": 8.609500000000001e-05,
      "loss": 0.4446,
      "step": 17228
    },
    {
      "epoch": 0.9647776906708478,
      "grad_norm": 1.1623423099517822,
      "learning_rate": 8.61e-05,
      "loss": 0.5162,
      "step": 17229
    },
    {
      "epoch": 0.9648336879829769,
      "grad_norm": 1.4732013940811157,
      "learning_rate": 8.6105e-05,
      "loss": 0.411,
      "step": 17230
    },
    {
      "epoch": 0.9648896852951059,
      "grad_norm": 1.3330683708190918,
      "learning_rate": 8.611e-05,
      "loss": 0.463,
      "step": 17231
    },
    {
      "epoch": 0.9649456826072349,
      "grad_norm": 1.6553144454956055,
      "learning_rate": 8.6115e-05,
      "loss": 0.4626,
      "step": 17232
    },
    {
      "epoch": 0.9650016799193639,
      "grad_norm": 1.2183603048324585,
      "learning_rate": 8.612e-05,
      "loss": 0.4028,
      "step": 17233
    },
    {
      "epoch": 0.9650576772314929,
      "grad_norm": 1.3921277523040771,
      "learning_rate": 8.6125e-05,
      "loss": 0.5236,
      "step": 17234
    },
    {
      "epoch": 0.965113674543622,
      "grad_norm": 1.318587303161621,
      "learning_rate": 8.613e-05,
      "loss": 0.3898,
      "step": 17235
    },
    {
      "epoch": 0.965169671855751,
      "grad_norm": 1.4311283826828003,
      "learning_rate": 8.6135e-05,
      "loss": 0.3776,
      "step": 17236
    },
    {
      "epoch": 0.96522566916788,
      "grad_norm": 2.413759231567383,
      "learning_rate": 8.614000000000001e-05,
      "loss": 0.602,
      "step": 17237
    },
    {
      "epoch": 0.9652816664800089,
      "grad_norm": 2.402284860610962,
      "learning_rate": 8.614500000000001e-05,
      "loss": 0.4754,
      "step": 17238
    },
    {
      "epoch": 0.9653376637921379,
      "grad_norm": 1.4711947441101074,
      "learning_rate": 8.615000000000001e-05,
      "loss": 0.5307,
      "step": 17239
    },
    {
      "epoch": 0.9653936611042669,
      "grad_norm": 1.5300489664077759,
      "learning_rate": 8.6155e-05,
      "loss": 0.5064,
      "step": 17240
    },
    {
      "epoch": 0.965449658416396,
      "grad_norm": 1.4783146381378174,
      "learning_rate": 8.616e-05,
      "loss": 0.3733,
      "step": 17241
    },
    {
      "epoch": 0.965505655728525,
      "grad_norm": 1.3357300758361816,
      "learning_rate": 8.616500000000001e-05,
      "loss": 0.4655,
      "step": 17242
    },
    {
      "epoch": 0.965561653040654,
      "grad_norm": 1.3752564191818237,
      "learning_rate": 8.617000000000001e-05,
      "loss": 0.5256,
      "step": 17243
    },
    {
      "epoch": 0.965617650352783,
      "grad_norm": 1.4365041255950928,
      "learning_rate": 8.617500000000001e-05,
      "loss": 0.4837,
      "step": 17244
    },
    {
      "epoch": 0.965673647664912,
      "grad_norm": 1.1746888160705566,
      "learning_rate": 8.618e-05,
      "loss": 0.464,
      "step": 17245
    },
    {
      "epoch": 0.9657296449770411,
      "grad_norm": 1.3307267427444458,
      "learning_rate": 8.6185e-05,
      "loss": 0.4821,
      "step": 17246
    },
    {
      "epoch": 0.9657856422891701,
      "grad_norm": 1.5634570121765137,
      "learning_rate": 8.619e-05,
      "loss": 0.5751,
      "step": 17247
    },
    {
      "epoch": 0.9658416396012991,
      "grad_norm": 1.3047250509262085,
      "learning_rate": 8.6195e-05,
      "loss": 0.4271,
      "step": 17248
    },
    {
      "epoch": 0.9658976369134281,
      "grad_norm": 1.172465443611145,
      "learning_rate": 8.620000000000001e-05,
      "loss": 0.4076,
      "step": 17249
    },
    {
      "epoch": 0.9659536342255571,
      "grad_norm": 1.4473350048065186,
      "learning_rate": 8.6205e-05,
      "loss": 0.5583,
      "step": 17250
    },
    {
      "epoch": 0.9660096315376862,
      "grad_norm": 1.2676153182983398,
      "learning_rate": 8.621e-05,
      "loss": 0.3651,
      "step": 17251
    },
    {
      "epoch": 0.9660656288498152,
      "grad_norm": 1.218522071838379,
      "learning_rate": 8.6215e-05,
      "loss": 0.4182,
      "step": 17252
    },
    {
      "epoch": 0.9661216261619442,
      "grad_norm": 1.5967514514923096,
      "learning_rate": 8.622e-05,
      "loss": 0.5798,
      "step": 17253
    },
    {
      "epoch": 0.9661776234740732,
      "grad_norm": 1.5131630897521973,
      "learning_rate": 8.6225e-05,
      "loss": 0.6109,
      "step": 17254
    },
    {
      "epoch": 0.9662336207862022,
      "grad_norm": 1.3282312154769897,
      "learning_rate": 8.623e-05,
      "loss": 0.3214,
      "step": 17255
    },
    {
      "epoch": 0.9662896180983312,
      "grad_norm": 1.2723884582519531,
      "learning_rate": 8.6235e-05,
      "loss": 0.3748,
      "step": 17256
    },
    {
      "epoch": 0.9663456154104603,
      "grad_norm": 4.182846546173096,
      "learning_rate": 8.624000000000001e-05,
      "loss": 0.4868,
      "step": 17257
    },
    {
      "epoch": 0.9664016127225893,
      "grad_norm": 1.388085961341858,
      "learning_rate": 8.624500000000001e-05,
      "loss": 0.3495,
      "step": 17258
    },
    {
      "epoch": 0.9664576100347183,
      "grad_norm": 1.6867340803146362,
      "learning_rate": 8.625000000000001e-05,
      "loss": 0.3838,
      "step": 17259
    },
    {
      "epoch": 0.9665136073468473,
      "grad_norm": 1.412258505821228,
      "learning_rate": 8.625500000000001e-05,
      "loss": 0.322,
      "step": 17260
    },
    {
      "epoch": 0.9665696046589763,
      "grad_norm": 1.4775829315185547,
      "learning_rate": 8.626e-05,
      "loss": 0.6763,
      "step": 17261
    },
    {
      "epoch": 0.9666256019711054,
      "grad_norm": 1.2024606466293335,
      "learning_rate": 8.6265e-05,
      "loss": 0.3556,
      "step": 17262
    },
    {
      "epoch": 0.9666815992832344,
      "grad_norm": 1.649871587753296,
      "learning_rate": 8.627000000000001e-05,
      "loss": 0.5428,
      "step": 17263
    },
    {
      "epoch": 0.9667375965953634,
      "grad_norm": 1.4487943649291992,
      "learning_rate": 8.627500000000001e-05,
      "loss": 0.514,
      "step": 17264
    },
    {
      "epoch": 0.9667935939074924,
      "grad_norm": 1.4263319969177246,
      "learning_rate": 8.628000000000001e-05,
      "loss": 0.422,
      "step": 17265
    },
    {
      "epoch": 0.9668495912196214,
      "grad_norm": 1.298326015472412,
      "learning_rate": 8.6285e-05,
      "loss": 0.4818,
      "step": 17266
    },
    {
      "epoch": 0.9669055885317505,
      "grad_norm": 1.5739593505859375,
      "learning_rate": 8.629e-05,
      "loss": 0.547,
      "step": 17267
    },
    {
      "epoch": 0.9669615858438795,
      "grad_norm": 1.1422499418258667,
      "learning_rate": 8.6295e-05,
      "loss": 0.327,
      "step": 17268
    },
    {
      "epoch": 0.9670175831560085,
      "grad_norm": 1.441144585609436,
      "learning_rate": 8.63e-05,
      "loss": 0.5159,
      "step": 17269
    },
    {
      "epoch": 0.9670735804681375,
      "grad_norm": 1.6179550886154175,
      "learning_rate": 8.630500000000001e-05,
      "loss": 0.5163,
      "step": 17270
    },
    {
      "epoch": 0.9671295777802665,
      "grad_norm": 1.4002225399017334,
      "learning_rate": 8.631e-05,
      "loss": 0.4987,
      "step": 17271
    },
    {
      "epoch": 0.9671855750923956,
      "grad_norm": 1.424810528755188,
      "learning_rate": 8.6315e-05,
      "loss": 0.4542,
      "step": 17272
    },
    {
      "epoch": 0.9672415724045246,
      "grad_norm": 1.5833181142807007,
      "learning_rate": 8.632e-05,
      "loss": 0.6777,
      "step": 17273
    },
    {
      "epoch": 0.9672975697166536,
      "grad_norm": 1.6691884994506836,
      "learning_rate": 8.6325e-05,
      "loss": 0.4705,
      "step": 17274
    },
    {
      "epoch": 0.9673535670287826,
      "grad_norm": 1.240027666091919,
      "learning_rate": 8.633e-05,
      "loss": 0.5124,
      "step": 17275
    },
    {
      "epoch": 0.9674095643409116,
      "grad_norm": 1.2922923564910889,
      "learning_rate": 8.633499999999999e-05,
      "loss": 0.4151,
      "step": 17276
    },
    {
      "epoch": 0.9674655616530407,
      "grad_norm": 1.161521077156067,
      "learning_rate": 8.634e-05,
      "loss": 0.3309,
      "step": 17277
    },
    {
      "epoch": 0.9675215589651697,
      "grad_norm": 1.3965418338775635,
      "learning_rate": 8.634500000000001e-05,
      "loss": 0.4749,
      "step": 17278
    },
    {
      "epoch": 0.9675775562772987,
      "grad_norm": 1.2737889289855957,
      "learning_rate": 8.635000000000001e-05,
      "loss": 0.4331,
      "step": 17279
    },
    {
      "epoch": 0.9676335535894277,
      "grad_norm": 1.2584105730056763,
      "learning_rate": 8.635500000000001e-05,
      "loss": 0.4195,
      "step": 17280
    },
    {
      "epoch": 0.9676895509015567,
      "grad_norm": 1.2243163585662842,
      "learning_rate": 8.636e-05,
      "loss": 0.429,
      "step": 17281
    },
    {
      "epoch": 0.9677455482136857,
      "grad_norm": 1.387087345123291,
      "learning_rate": 8.6365e-05,
      "loss": 0.5707,
      "step": 17282
    },
    {
      "epoch": 0.9678015455258148,
      "grad_norm": 1.210732340812683,
      "learning_rate": 8.637e-05,
      "loss": 0.5474,
      "step": 17283
    },
    {
      "epoch": 0.9678575428379438,
      "grad_norm": 1.1186721324920654,
      "learning_rate": 8.637500000000001e-05,
      "loss": 0.3673,
      "step": 17284
    },
    {
      "epoch": 0.9679135401500728,
      "grad_norm": 1.5173031091690063,
      "learning_rate": 8.638000000000001e-05,
      "loss": 0.4944,
      "step": 17285
    },
    {
      "epoch": 0.9679695374622018,
      "grad_norm": 1.1591531038284302,
      "learning_rate": 8.638500000000001e-05,
      "loss": 0.4421,
      "step": 17286
    },
    {
      "epoch": 0.9680255347743308,
      "grad_norm": 1.3933157920837402,
      "learning_rate": 8.639e-05,
      "loss": 0.443,
      "step": 17287
    },
    {
      "epoch": 0.9680815320864599,
      "grad_norm": 1.1293524503707886,
      "learning_rate": 8.6395e-05,
      "loss": 0.3323,
      "step": 17288
    },
    {
      "epoch": 0.9681375293985889,
      "grad_norm": 1.3659924268722534,
      "learning_rate": 8.64e-05,
      "loss": 0.3717,
      "step": 17289
    },
    {
      "epoch": 0.9681935267107179,
      "grad_norm": 1.6097309589385986,
      "learning_rate": 8.6405e-05,
      "loss": 0.4152,
      "step": 17290
    },
    {
      "epoch": 0.9682495240228469,
      "grad_norm": 1.181439995765686,
      "learning_rate": 8.641000000000001e-05,
      "loss": 0.404,
      "step": 17291
    },
    {
      "epoch": 0.9683055213349759,
      "grad_norm": 1.6815600395202637,
      "learning_rate": 8.6415e-05,
      "loss": 0.6659,
      "step": 17292
    },
    {
      "epoch": 0.968361518647105,
      "grad_norm": 1.570206642150879,
      "learning_rate": 8.642e-05,
      "loss": 0.4618,
      "step": 17293
    },
    {
      "epoch": 0.968417515959234,
      "grad_norm": 2.163844347000122,
      "learning_rate": 8.6425e-05,
      "loss": 0.643,
      "step": 17294
    },
    {
      "epoch": 0.968473513271363,
      "grad_norm": 1.467824101448059,
      "learning_rate": 8.643e-05,
      "loss": 0.5083,
      "step": 17295
    },
    {
      "epoch": 0.968529510583492,
      "grad_norm": 1.1489348411560059,
      "learning_rate": 8.6435e-05,
      "loss": 0.3732,
      "step": 17296
    },
    {
      "epoch": 0.968585507895621,
      "grad_norm": 1.3897266387939453,
      "learning_rate": 8.643999999999999e-05,
      "loss": 0.4392,
      "step": 17297
    },
    {
      "epoch": 0.96864150520775,
      "grad_norm": 1.2860050201416016,
      "learning_rate": 8.6445e-05,
      "loss": 0.4827,
      "step": 17298
    },
    {
      "epoch": 0.9686975025198791,
      "grad_norm": 1.547238826751709,
      "learning_rate": 8.645000000000001e-05,
      "loss": 0.4105,
      "step": 17299
    },
    {
      "epoch": 0.9687534998320081,
      "grad_norm": 1.42068350315094,
      "learning_rate": 8.645500000000001e-05,
      "loss": 0.4412,
      "step": 17300
    },
    {
      "epoch": 0.9688094971441371,
      "grad_norm": 1.328653335571289,
      "learning_rate": 8.646000000000001e-05,
      "loss": 0.4863,
      "step": 17301
    },
    {
      "epoch": 0.9688654944562661,
      "grad_norm": 1.281003713607788,
      "learning_rate": 8.6465e-05,
      "loss": 0.4571,
      "step": 17302
    },
    {
      "epoch": 0.9689214917683951,
      "grad_norm": 1.40518057346344,
      "learning_rate": 8.647e-05,
      "loss": 0.3568,
      "step": 17303
    },
    {
      "epoch": 0.9689774890805242,
      "grad_norm": 1.4284865856170654,
      "learning_rate": 8.6475e-05,
      "loss": 0.5044,
      "step": 17304
    },
    {
      "epoch": 0.9690334863926532,
      "grad_norm": 1.402702808380127,
      "learning_rate": 8.648e-05,
      "loss": 0.5023,
      "step": 17305
    },
    {
      "epoch": 0.9690894837047822,
      "grad_norm": 1.465108871459961,
      "learning_rate": 8.648500000000001e-05,
      "loss": 0.5385,
      "step": 17306
    },
    {
      "epoch": 0.9691454810169112,
      "grad_norm": 1.6162699460983276,
      "learning_rate": 8.649000000000001e-05,
      "loss": 0.5153,
      "step": 17307
    },
    {
      "epoch": 0.9692014783290402,
      "grad_norm": 1.3178231716156006,
      "learning_rate": 8.6495e-05,
      "loss": 0.4102,
      "step": 17308
    },
    {
      "epoch": 0.9692574756411693,
      "grad_norm": 1.271968960762024,
      "learning_rate": 8.65e-05,
      "loss": 0.3879,
      "step": 17309
    },
    {
      "epoch": 0.9693134729532983,
      "grad_norm": 1.488120436668396,
      "learning_rate": 8.6505e-05,
      "loss": 0.4599,
      "step": 17310
    },
    {
      "epoch": 0.9693694702654273,
      "grad_norm": 1.620308518409729,
      "learning_rate": 8.651e-05,
      "loss": 0.665,
      "step": 17311
    },
    {
      "epoch": 0.9694254675775563,
      "grad_norm": 1.3775193691253662,
      "learning_rate": 8.651500000000001e-05,
      "loss": 0.4418,
      "step": 17312
    },
    {
      "epoch": 0.9694814648896853,
      "grad_norm": 1.3837276697158813,
      "learning_rate": 8.652e-05,
      "loss": 0.5351,
      "step": 17313
    },
    {
      "epoch": 0.9695374622018144,
      "grad_norm": 1.4161573648452759,
      "learning_rate": 8.6525e-05,
      "loss": 0.4661,
      "step": 17314
    },
    {
      "epoch": 0.9695934595139434,
      "grad_norm": 1.5309301614761353,
      "learning_rate": 8.653e-05,
      "loss": 0.5751,
      "step": 17315
    },
    {
      "epoch": 0.9696494568260724,
      "grad_norm": 1.2019197940826416,
      "learning_rate": 8.6535e-05,
      "loss": 0.4031,
      "step": 17316
    },
    {
      "epoch": 0.9697054541382014,
      "grad_norm": 1.1792582273483276,
      "learning_rate": 8.654e-05,
      "loss": 0.4181,
      "step": 17317
    },
    {
      "epoch": 0.9697614514503304,
      "grad_norm": 1.3851470947265625,
      "learning_rate": 8.6545e-05,
      "loss": 0.5365,
      "step": 17318
    },
    {
      "epoch": 0.9698174487624595,
      "grad_norm": 1.6818479299545288,
      "learning_rate": 8.655e-05,
      "loss": 0.4797,
      "step": 17319
    },
    {
      "epoch": 0.9698734460745885,
      "grad_norm": 1.4252855777740479,
      "learning_rate": 8.655500000000001e-05,
      "loss": 0.4359,
      "step": 17320
    },
    {
      "epoch": 0.9699294433867174,
      "grad_norm": 1.3790574073791504,
      "learning_rate": 8.656000000000001e-05,
      "loss": 0.7042,
      "step": 17321
    },
    {
      "epoch": 0.9699854406988464,
      "grad_norm": 1.2539957761764526,
      "learning_rate": 8.656500000000001e-05,
      "loss": 0.4094,
      "step": 17322
    },
    {
      "epoch": 0.9700414380109754,
      "grad_norm": 1.3838392496109009,
      "learning_rate": 8.657e-05,
      "loss": 0.4915,
      "step": 17323
    },
    {
      "epoch": 0.9700974353231044,
      "grad_norm": 1.2799766063690186,
      "learning_rate": 8.6575e-05,
      "loss": 0.4512,
      "step": 17324
    },
    {
      "epoch": 0.9701534326352335,
      "grad_norm": 1.4863401651382446,
      "learning_rate": 8.658e-05,
      "loss": 0.4828,
      "step": 17325
    },
    {
      "epoch": 0.9702094299473625,
      "grad_norm": 1.1538506746292114,
      "learning_rate": 8.6585e-05,
      "loss": 0.3684,
      "step": 17326
    },
    {
      "epoch": 0.9702654272594915,
      "grad_norm": 1.3310805559158325,
      "learning_rate": 8.659000000000001e-05,
      "loss": 0.3406,
      "step": 17327
    },
    {
      "epoch": 0.9703214245716205,
      "grad_norm": 1.559973120689392,
      "learning_rate": 8.659500000000001e-05,
      "loss": 0.3151,
      "step": 17328
    },
    {
      "epoch": 0.9703774218837495,
      "grad_norm": 1.297188401222229,
      "learning_rate": 8.66e-05,
      "loss": 0.3679,
      "step": 17329
    },
    {
      "epoch": 0.9704334191958786,
      "grad_norm": 1.1994104385375977,
      "learning_rate": 8.6605e-05,
      "loss": 0.3181,
      "step": 17330
    },
    {
      "epoch": 0.9704894165080076,
      "grad_norm": 1.5522327423095703,
      "learning_rate": 8.661e-05,
      "loss": 0.4533,
      "step": 17331
    },
    {
      "epoch": 0.9705454138201366,
      "grad_norm": 1.4846142530441284,
      "learning_rate": 8.6615e-05,
      "loss": 0.497,
      "step": 17332
    },
    {
      "epoch": 0.9706014111322656,
      "grad_norm": 4.936802864074707,
      "learning_rate": 8.662000000000001e-05,
      "loss": 0.4634,
      "step": 17333
    },
    {
      "epoch": 0.9706574084443946,
      "grad_norm": 1.461236834526062,
      "learning_rate": 8.6625e-05,
      "loss": 0.4195,
      "step": 17334
    },
    {
      "epoch": 0.9707134057565237,
      "grad_norm": 1.3690143823623657,
      "learning_rate": 8.663e-05,
      "loss": 0.3715,
      "step": 17335
    },
    {
      "epoch": 0.9707694030686527,
      "grad_norm": 1.4367308616638184,
      "learning_rate": 8.6635e-05,
      "loss": 0.4884,
      "step": 17336
    },
    {
      "epoch": 0.9708254003807817,
      "grad_norm": 1.3397955894470215,
      "learning_rate": 8.664e-05,
      "loss": 0.3518,
      "step": 17337
    },
    {
      "epoch": 0.9708813976929107,
      "grad_norm": 1.7689563035964966,
      "learning_rate": 8.664500000000001e-05,
      "loss": 0.5787,
      "step": 17338
    },
    {
      "epoch": 0.9709373950050397,
      "grad_norm": 1.5664008855819702,
      "learning_rate": 8.665e-05,
      "loss": 0.3727,
      "step": 17339
    },
    {
      "epoch": 0.9709933923171687,
      "grad_norm": 1.1659214496612549,
      "learning_rate": 8.6655e-05,
      "loss": 0.4004,
      "step": 17340
    },
    {
      "epoch": 0.9710493896292978,
      "grad_norm": 1.307281255722046,
      "learning_rate": 8.666000000000001e-05,
      "loss": 0.3862,
      "step": 17341
    },
    {
      "epoch": 0.9711053869414268,
      "grad_norm": 1.6505941152572632,
      "learning_rate": 8.666500000000001e-05,
      "loss": 0.4579,
      "step": 17342
    },
    {
      "epoch": 0.9711613842535558,
      "grad_norm": 1.163074254989624,
      "learning_rate": 8.667000000000001e-05,
      "loss": 0.4259,
      "step": 17343
    },
    {
      "epoch": 0.9712173815656848,
      "grad_norm": 1.2999498844146729,
      "learning_rate": 8.6675e-05,
      "loss": 0.4441,
      "step": 17344
    },
    {
      "epoch": 0.9712733788778138,
      "grad_norm": 2.1191458702087402,
      "learning_rate": 8.668e-05,
      "loss": 0.426,
      "step": 17345
    },
    {
      "epoch": 0.9713293761899429,
      "grad_norm": 1.8392921686172485,
      "learning_rate": 8.6685e-05,
      "loss": 0.4483,
      "step": 17346
    },
    {
      "epoch": 0.9713853735020719,
      "grad_norm": 1.2343018054962158,
      "learning_rate": 8.669e-05,
      "loss": 0.5183,
      "step": 17347
    },
    {
      "epoch": 0.9714413708142009,
      "grad_norm": 1.1489077806472778,
      "learning_rate": 8.669500000000001e-05,
      "loss": 0.3422,
      "step": 17348
    },
    {
      "epoch": 0.9714973681263299,
      "grad_norm": 1.1706421375274658,
      "learning_rate": 8.67e-05,
      "loss": 0.3887,
      "step": 17349
    },
    {
      "epoch": 0.9715533654384589,
      "grad_norm": 1.3635989427566528,
      "learning_rate": 8.6705e-05,
      "loss": 0.5152,
      "step": 17350
    },
    {
      "epoch": 0.971609362750588,
      "grad_norm": 2.037436008453369,
      "learning_rate": 8.671e-05,
      "loss": 0.3749,
      "step": 17351
    },
    {
      "epoch": 0.971665360062717,
      "grad_norm": 1.750329852104187,
      "learning_rate": 8.6715e-05,
      "loss": 0.6067,
      "step": 17352
    },
    {
      "epoch": 0.971721357374846,
      "grad_norm": 1.270072340965271,
      "learning_rate": 8.672e-05,
      "loss": 0.4039,
      "step": 17353
    },
    {
      "epoch": 0.971777354686975,
      "grad_norm": 1.7762908935546875,
      "learning_rate": 8.672500000000001e-05,
      "loss": 0.5357,
      "step": 17354
    },
    {
      "epoch": 0.971833351999104,
      "grad_norm": 1.4279133081436157,
      "learning_rate": 8.673e-05,
      "loss": 0.4171,
      "step": 17355
    },
    {
      "epoch": 0.971889349311233,
      "grad_norm": 1.1249388456344604,
      "learning_rate": 8.6735e-05,
      "loss": 0.3568,
      "step": 17356
    },
    {
      "epoch": 0.9719453466233621,
      "grad_norm": 1.4502819776535034,
      "learning_rate": 8.674e-05,
      "loss": 0.4598,
      "step": 17357
    },
    {
      "epoch": 0.9720013439354911,
      "grad_norm": 1.2537726163864136,
      "learning_rate": 8.674500000000001e-05,
      "loss": 0.3844,
      "step": 17358
    },
    {
      "epoch": 0.9720573412476201,
      "grad_norm": 1.2995517253875732,
      "learning_rate": 8.675000000000001e-05,
      "loss": 0.5004,
      "step": 17359
    },
    {
      "epoch": 0.9721133385597491,
      "grad_norm": 1.4475985765457153,
      "learning_rate": 8.6755e-05,
      "loss": 0.4624,
      "step": 17360
    },
    {
      "epoch": 0.9721693358718781,
      "grad_norm": 1.2398110628128052,
      "learning_rate": 8.676e-05,
      "loss": 0.3665,
      "step": 17361
    },
    {
      "epoch": 0.9722253331840072,
      "grad_norm": 1.439234972000122,
      "learning_rate": 8.676500000000001e-05,
      "loss": 0.4756,
      "step": 17362
    },
    {
      "epoch": 0.9722813304961362,
      "grad_norm": 1.49754798412323,
      "learning_rate": 8.677000000000001e-05,
      "loss": 0.455,
      "step": 17363
    },
    {
      "epoch": 0.9723373278082652,
      "grad_norm": 1.302474021911621,
      "learning_rate": 8.677500000000001e-05,
      "loss": 0.4373,
      "step": 17364
    },
    {
      "epoch": 0.9723933251203942,
      "grad_norm": 1.0640931129455566,
      "learning_rate": 8.678e-05,
      "loss": 0.3671,
      "step": 17365
    },
    {
      "epoch": 0.9724493224325232,
      "grad_norm": 1.2272655963897705,
      "learning_rate": 8.6785e-05,
      "loss": 0.4005,
      "step": 17366
    },
    {
      "epoch": 0.9725053197446523,
      "grad_norm": 1.508810043334961,
      "learning_rate": 8.679e-05,
      "loss": 0.4766,
      "step": 17367
    },
    {
      "epoch": 0.9725613170567813,
      "grad_norm": 1.224143147468567,
      "learning_rate": 8.6795e-05,
      "loss": 0.4397,
      "step": 17368
    },
    {
      "epoch": 0.9726173143689103,
      "grad_norm": 1.2601372003555298,
      "learning_rate": 8.680000000000001e-05,
      "loss": 0.4508,
      "step": 17369
    },
    {
      "epoch": 0.9726733116810393,
      "grad_norm": 1.5283092260360718,
      "learning_rate": 8.6805e-05,
      "loss": 0.3897,
      "step": 17370
    },
    {
      "epoch": 0.9727293089931683,
      "grad_norm": 1.430024266242981,
      "learning_rate": 8.681e-05,
      "loss": 0.5453,
      "step": 17371
    },
    {
      "epoch": 0.9727853063052974,
      "grad_norm": 1.3398207426071167,
      "learning_rate": 8.6815e-05,
      "loss": 0.4539,
      "step": 17372
    },
    {
      "epoch": 0.9728413036174264,
      "grad_norm": 1.4060691595077515,
      "learning_rate": 8.682e-05,
      "loss": 0.4397,
      "step": 17373
    },
    {
      "epoch": 0.9728973009295554,
      "grad_norm": 1.418944001197815,
      "learning_rate": 8.6825e-05,
      "loss": 0.4962,
      "step": 17374
    },
    {
      "epoch": 0.9729532982416844,
      "grad_norm": 1.4422852993011475,
      "learning_rate": 8.683e-05,
      "loss": 0.5276,
      "step": 17375
    },
    {
      "epoch": 0.9730092955538134,
      "grad_norm": 1.3235855102539062,
      "learning_rate": 8.6835e-05,
      "loss": 0.4473,
      "step": 17376
    },
    {
      "epoch": 0.9730652928659425,
      "grad_norm": 1.2004975080490112,
      "learning_rate": 8.684e-05,
      "loss": 0.418,
      "step": 17377
    },
    {
      "epoch": 0.9731212901780715,
      "grad_norm": 1.4729422330856323,
      "learning_rate": 8.684500000000001e-05,
      "loss": 0.455,
      "step": 17378
    },
    {
      "epoch": 0.9731772874902005,
      "grad_norm": 1.4228463172912598,
      "learning_rate": 8.685000000000001e-05,
      "loss": 0.5516,
      "step": 17379
    },
    {
      "epoch": 0.9732332848023295,
      "grad_norm": 1.2448140382766724,
      "learning_rate": 8.685500000000001e-05,
      "loss": 0.4987,
      "step": 17380
    },
    {
      "epoch": 0.9732892821144585,
      "grad_norm": 1.2316290140151978,
      "learning_rate": 8.686e-05,
      "loss": 0.4023,
      "step": 17381
    },
    {
      "epoch": 0.9733452794265876,
      "grad_norm": 1.3800770044326782,
      "learning_rate": 8.6865e-05,
      "loss": 0.5761,
      "step": 17382
    },
    {
      "epoch": 0.9734012767387166,
      "grad_norm": 1.0687819719314575,
      "learning_rate": 8.687000000000001e-05,
      "loss": 0.4053,
      "step": 17383
    },
    {
      "epoch": 0.9734572740508456,
      "grad_norm": 1.395995855331421,
      "learning_rate": 8.687500000000001e-05,
      "loss": 0.4674,
      "step": 17384
    },
    {
      "epoch": 0.9735132713629746,
      "grad_norm": 1.3109594583511353,
      "learning_rate": 8.688000000000001e-05,
      "loss": 0.4991,
      "step": 17385
    },
    {
      "epoch": 0.9735692686751036,
      "grad_norm": 1.2651557922363281,
      "learning_rate": 8.6885e-05,
      "loss": 0.4577,
      "step": 17386
    },
    {
      "epoch": 0.9736252659872326,
      "grad_norm": 1.2861329317092896,
      "learning_rate": 8.689e-05,
      "loss": 0.4672,
      "step": 17387
    },
    {
      "epoch": 0.9736812632993617,
      "grad_norm": 1.2998154163360596,
      "learning_rate": 8.6895e-05,
      "loss": 0.3578,
      "step": 17388
    },
    {
      "epoch": 0.9737372606114907,
      "grad_norm": 1.3852159976959229,
      "learning_rate": 8.69e-05,
      "loss": 0.4885,
      "step": 17389
    },
    {
      "epoch": 0.9737932579236197,
      "grad_norm": 1.4287030696868896,
      "learning_rate": 8.690500000000001e-05,
      "loss": 0.4675,
      "step": 17390
    },
    {
      "epoch": 0.9738492552357487,
      "grad_norm": 1.1838635206222534,
      "learning_rate": 8.691e-05,
      "loss": 0.4898,
      "step": 17391
    },
    {
      "epoch": 0.9739052525478777,
      "grad_norm": 1.3776044845581055,
      "learning_rate": 8.6915e-05,
      "loss": 0.4377,
      "step": 17392
    },
    {
      "epoch": 0.9739612498600068,
      "grad_norm": 1.543620228767395,
      "learning_rate": 8.692e-05,
      "loss": 0.5551,
      "step": 17393
    },
    {
      "epoch": 0.9740172471721358,
      "grad_norm": 1.5677738189697266,
      "learning_rate": 8.6925e-05,
      "loss": 0.5707,
      "step": 17394
    },
    {
      "epoch": 0.9740732444842648,
      "grad_norm": 1.1838555335998535,
      "learning_rate": 8.693e-05,
      "loss": 0.4575,
      "step": 17395
    },
    {
      "epoch": 0.9741292417963938,
      "grad_norm": 1.5340828895568848,
      "learning_rate": 8.6935e-05,
      "loss": 0.5805,
      "step": 17396
    },
    {
      "epoch": 0.9741852391085228,
      "grad_norm": 1.3245877027511597,
      "learning_rate": 8.694e-05,
      "loss": 0.5607,
      "step": 17397
    },
    {
      "epoch": 0.9742412364206519,
      "grad_norm": 1.3191213607788086,
      "learning_rate": 8.6945e-05,
      "loss": 0.5458,
      "step": 17398
    },
    {
      "epoch": 0.9742972337327809,
      "grad_norm": 1.0440481901168823,
      "learning_rate": 8.695000000000001e-05,
      "loss": 0.3783,
      "step": 17399
    },
    {
      "epoch": 0.9743532310449099,
      "grad_norm": 1.5507785081863403,
      "learning_rate": 8.695500000000001e-05,
      "loss": 0.443,
      "step": 17400
    },
    {
      "epoch": 0.9744092283570389,
      "grad_norm": 1.5314637422561646,
      "learning_rate": 8.696000000000001e-05,
      "loss": 0.5015,
      "step": 17401
    },
    {
      "epoch": 0.9744652256691679,
      "grad_norm": 1.307962417602539,
      "learning_rate": 8.6965e-05,
      "loss": 0.4961,
      "step": 17402
    },
    {
      "epoch": 0.974521222981297,
      "grad_norm": 1.4043619632720947,
      "learning_rate": 8.697e-05,
      "loss": 0.5223,
      "step": 17403
    },
    {
      "epoch": 0.9745772202934259,
      "grad_norm": 1.1751432418823242,
      "learning_rate": 8.6975e-05,
      "loss": 0.3579,
      "step": 17404
    },
    {
      "epoch": 0.9746332176055549,
      "grad_norm": 1.4457268714904785,
      "learning_rate": 8.698000000000001e-05,
      "loss": 0.431,
      "step": 17405
    },
    {
      "epoch": 0.9746892149176839,
      "grad_norm": 1.4647364616394043,
      "learning_rate": 8.698500000000001e-05,
      "loss": 0.5258,
      "step": 17406
    },
    {
      "epoch": 0.9747452122298129,
      "grad_norm": 1.5925337076187134,
      "learning_rate": 8.699e-05,
      "loss": 0.3815,
      "step": 17407
    },
    {
      "epoch": 0.9748012095419419,
      "grad_norm": 1.443375587463379,
      "learning_rate": 8.6995e-05,
      "loss": 0.4828,
      "step": 17408
    },
    {
      "epoch": 0.974857206854071,
      "grad_norm": 1.2334703207015991,
      "learning_rate": 8.7e-05,
      "loss": 0.4202,
      "step": 17409
    },
    {
      "epoch": 0.9749132041662,
      "grad_norm": 1.5328794717788696,
      "learning_rate": 8.7005e-05,
      "loss": 0.3994,
      "step": 17410
    },
    {
      "epoch": 0.974969201478329,
      "grad_norm": 1.2115094661712646,
      "learning_rate": 8.701000000000001e-05,
      "loss": 0.3725,
      "step": 17411
    },
    {
      "epoch": 0.975025198790458,
      "grad_norm": 1.4653737545013428,
      "learning_rate": 8.7015e-05,
      "loss": 0.5088,
      "step": 17412
    },
    {
      "epoch": 0.975081196102587,
      "grad_norm": 1.5247368812561035,
      "learning_rate": 8.702e-05,
      "loss": 0.5465,
      "step": 17413
    },
    {
      "epoch": 0.975137193414716,
      "grad_norm": 1.2807185649871826,
      "learning_rate": 8.7025e-05,
      "loss": 0.3851,
      "step": 17414
    },
    {
      "epoch": 0.9751931907268451,
      "grad_norm": 1.2373749017715454,
      "learning_rate": 8.703e-05,
      "loss": 0.343,
      "step": 17415
    },
    {
      "epoch": 0.9752491880389741,
      "grad_norm": 1.3598071336746216,
      "learning_rate": 8.7035e-05,
      "loss": 0.6644,
      "step": 17416
    },
    {
      "epoch": 0.9753051853511031,
      "grad_norm": 1.1725260019302368,
      "learning_rate": 8.704e-05,
      "loss": 0.4689,
      "step": 17417
    },
    {
      "epoch": 0.9753611826632321,
      "grad_norm": 1.242869257926941,
      "learning_rate": 8.7045e-05,
      "loss": 0.4059,
      "step": 17418
    },
    {
      "epoch": 0.9754171799753611,
      "grad_norm": 1.1475459337234497,
      "learning_rate": 8.705000000000002e-05,
      "loss": 0.4537,
      "step": 17419
    },
    {
      "epoch": 0.9754731772874902,
      "grad_norm": 1.1499022245407104,
      "learning_rate": 8.705500000000001e-05,
      "loss": 0.3748,
      "step": 17420
    },
    {
      "epoch": 0.9755291745996192,
      "grad_norm": 1.149317979812622,
      "learning_rate": 8.706000000000001e-05,
      "loss": 0.4822,
      "step": 17421
    },
    {
      "epoch": 0.9755851719117482,
      "grad_norm": 1.4652608633041382,
      "learning_rate": 8.706500000000001e-05,
      "loss": 0.4387,
      "step": 17422
    },
    {
      "epoch": 0.9756411692238772,
      "grad_norm": 1.5358394384384155,
      "learning_rate": 8.707e-05,
      "loss": 0.5631,
      "step": 17423
    },
    {
      "epoch": 0.9756971665360062,
      "grad_norm": 1.3846830129623413,
      "learning_rate": 8.7075e-05,
      "loss": 0.4613,
      "step": 17424
    },
    {
      "epoch": 0.9757531638481353,
      "grad_norm": 1.6016181707382202,
      "learning_rate": 8.708e-05,
      "loss": 0.5172,
      "step": 17425
    },
    {
      "epoch": 0.9758091611602643,
      "grad_norm": 1.251336693763733,
      "learning_rate": 8.708500000000001e-05,
      "loss": 0.4126,
      "step": 17426
    },
    {
      "epoch": 0.9758651584723933,
      "grad_norm": 1.4441380500793457,
      "learning_rate": 8.709000000000001e-05,
      "loss": 0.3877,
      "step": 17427
    },
    {
      "epoch": 0.9759211557845223,
      "grad_norm": 1.1034600734710693,
      "learning_rate": 8.7095e-05,
      "loss": 0.4298,
      "step": 17428
    },
    {
      "epoch": 0.9759771530966513,
      "grad_norm": 1.1990065574645996,
      "learning_rate": 8.71e-05,
      "loss": 0.4104,
      "step": 17429
    },
    {
      "epoch": 0.9760331504087804,
      "grad_norm": 1.752121090888977,
      "learning_rate": 8.7105e-05,
      "loss": 0.664,
      "step": 17430
    },
    {
      "epoch": 0.9760891477209094,
      "grad_norm": 1.2689826488494873,
      "learning_rate": 8.711e-05,
      "loss": 0.517,
      "step": 17431
    },
    {
      "epoch": 0.9761451450330384,
      "grad_norm": 1.443172812461853,
      "learning_rate": 8.711500000000001e-05,
      "loss": 0.4914,
      "step": 17432
    },
    {
      "epoch": 0.9762011423451674,
      "grad_norm": 1.3688075542449951,
      "learning_rate": 8.712e-05,
      "loss": 0.4629,
      "step": 17433
    },
    {
      "epoch": 0.9762571396572964,
      "grad_norm": 1.4039663076400757,
      "learning_rate": 8.7125e-05,
      "loss": 0.4362,
      "step": 17434
    },
    {
      "epoch": 0.9763131369694255,
      "grad_norm": 1.5527126789093018,
      "learning_rate": 8.713e-05,
      "loss": 0.5913,
      "step": 17435
    },
    {
      "epoch": 0.9763691342815545,
      "grad_norm": 1.6251251697540283,
      "learning_rate": 8.7135e-05,
      "loss": 0.4114,
      "step": 17436
    },
    {
      "epoch": 0.9764251315936835,
      "grad_norm": 1.7148405313491821,
      "learning_rate": 8.714e-05,
      "loss": 0.531,
      "step": 17437
    },
    {
      "epoch": 0.9764811289058125,
      "grad_norm": 1.2098383903503418,
      "learning_rate": 8.714499999999999e-05,
      "loss": 0.3567,
      "step": 17438
    },
    {
      "epoch": 0.9765371262179415,
      "grad_norm": 1.2557868957519531,
      "learning_rate": 8.715e-05,
      "loss": 0.3424,
      "step": 17439
    },
    {
      "epoch": 0.9765931235300706,
      "grad_norm": 1.4591310024261475,
      "learning_rate": 8.715500000000002e-05,
      "loss": 0.5535,
      "step": 17440
    },
    {
      "epoch": 0.9766491208421996,
      "grad_norm": 1.4864718914031982,
      "learning_rate": 8.716000000000001e-05,
      "loss": 0.4163,
      "step": 17441
    },
    {
      "epoch": 0.9767051181543286,
      "grad_norm": 1.585402488708496,
      "learning_rate": 8.716500000000001e-05,
      "loss": 0.4101,
      "step": 17442
    },
    {
      "epoch": 0.9767611154664576,
      "grad_norm": 1.4526938199996948,
      "learning_rate": 8.717000000000001e-05,
      "loss": 0.5043,
      "step": 17443
    },
    {
      "epoch": 0.9768171127785866,
      "grad_norm": 1.1431570053100586,
      "learning_rate": 8.7175e-05,
      "loss": 0.3464,
      "step": 17444
    },
    {
      "epoch": 0.9768731100907156,
      "grad_norm": 1.8126105070114136,
      "learning_rate": 8.718e-05,
      "loss": 0.656,
      "step": 17445
    },
    {
      "epoch": 0.9769291074028447,
      "grad_norm": 1.3256500959396362,
      "learning_rate": 8.7185e-05,
      "loss": 0.4084,
      "step": 17446
    },
    {
      "epoch": 0.9769851047149737,
      "grad_norm": 1.631635069847107,
      "learning_rate": 8.719000000000001e-05,
      "loss": 0.5274,
      "step": 17447
    },
    {
      "epoch": 0.9770411020271027,
      "grad_norm": 1.4243981838226318,
      "learning_rate": 8.719500000000001e-05,
      "loss": 0.4474,
      "step": 17448
    },
    {
      "epoch": 0.9770970993392317,
      "grad_norm": 1.605637788772583,
      "learning_rate": 8.72e-05,
      "loss": 0.4262,
      "step": 17449
    },
    {
      "epoch": 0.9771530966513607,
      "grad_norm": 1.4703837633132935,
      "learning_rate": 8.7205e-05,
      "loss": 0.5151,
      "step": 17450
    },
    {
      "epoch": 0.9772090939634898,
      "grad_norm": 1.3215372562408447,
      "learning_rate": 8.721e-05,
      "loss": 0.3395,
      "step": 17451
    },
    {
      "epoch": 0.9772650912756188,
      "grad_norm": 1.419661283493042,
      "learning_rate": 8.7215e-05,
      "loss": 0.5087,
      "step": 17452
    },
    {
      "epoch": 0.9773210885877478,
      "grad_norm": 1.253587245941162,
      "learning_rate": 8.722e-05,
      "loss": 0.4516,
      "step": 17453
    },
    {
      "epoch": 0.9773770858998768,
      "grad_norm": 1.3379052877426147,
      "learning_rate": 8.7225e-05,
      "loss": 0.3923,
      "step": 17454
    },
    {
      "epoch": 0.9774330832120058,
      "grad_norm": 1.2216262817382812,
      "learning_rate": 8.723e-05,
      "loss": 0.4176,
      "step": 17455
    },
    {
      "epoch": 0.9774890805241349,
      "grad_norm": 1.610276460647583,
      "learning_rate": 8.7235e-05,
      "loss": 0.5338,
      "step": 17456
    },
    {
      "epoch": 0.9775450778362639,
      "grad_norm": 1.5220662355422974,
      "learning_rate": 8.724e-05,
      "loss": 0.7544,
      "step": 17457
    },
    {
      "epoch": 0.9776010751483929,
      "grad_norm": 1.516425371170044,
      "learning_rate": 8.7245e-05,
      "loss": 0.4472,
      "step": 17458
    },
    {
      "epoch": 0.9776570724605219,
      "grad_norm": 1.4756171703338623,
      "learning_rate": 8.725e-05,
      "loss": 0.4746,
      "step": 17459
    },
    {
      "epoch": 0.9777130697726509,
      "grad_norm": 1.1557035446166992,
      "learning_rate": 8.7255e-05,
      "loss": 0.4096,
      "step": 17460
    },
    {
      "epoch": 0.97776906708478,
      "grad_norm": 4.008048057556152,
      "learning_rate": 8.726000000000001e-05,
      "loss": 0.5853,
      "step": 17461
    },
    {
      "epoch": 0.977825064396909,
      "grad_norm": 1.3212580680847168,
      "learning_rate": 8.726500000000001e-05,
      "loss": 0.4024,
      "step": 17462
    },
    {
      "epoch": 0.977881061709038,
      "grad_norm": 1.3305388689041138,
      "learning_rate": 8.727000000000001e-05,
      "loss": 0.3946,
      "step": 17463
    },
    {
      "epoch": 0.977937059021167,
      "grad_norm": 1.2800099849700928,
      "learning_rate": 8.727500000000001e-05,
      "loss": 0.4303,
      "step": 17464
    },
    {
      "epoch": 0.977993056333296,
      "grad_norm": 1.755967140197754,
      "learning_rate": 8.728e-05,
      "loss": 0.6422,
      "step": 17465
    },
    {
      "epoch": 0.978049053645425,
      "grad_norm": 1.3023176193237305,
      "learning_rate": 8.7285e-05,
      "loss": 0.4761,
      "step": 17466
    },
    {
      "epoch": 0.9781050509575541,
      "grad_norm": 1.6691690683364868,
      "learning_rate": 8.729e-05,
      "loss": 0.4489,
      "step": 17467
    },
    {
      "epoch": 0.9781610482696831,
      "grad_norm": 1.2901387214660645,
      "learning_rate": 8.729500000000001e-05,
      "loss": 0.3703,
      "step": 17468
    },
    {
      "epoch": 0.9782170455818121,
      "grad_norm": 1.5257415771484375,
      "learning_rate": 8.730000000000001e-05,
      "loss": 0.5251,
      "step": 17469
    },
    {
      "epoch": 0.9782730428939411,
      "grad_norm": 1.2863874435424805,
      "learning_rate": 8.7305e-05,
      "loss": 0.5034,
      "step": 17470
    },
    {
      "epoch": 0.9783290402060701,
      "grad_norm": 1.2762781381607056,
      "learning_rate": 8.731e-05,
      "loss": 0.6128,
      "step": 17471
    },
    {
      "epoch": 0.9783850375181992,
      "grad_norm": 1.3336706161499023,
      "learning_rate": 8.7315e-05,
      "loss": 0.5472,
      "step": 17472
    },
    {
      "epoch": 0.9784410348303282,
      "grad_norm": 1.3449336290359497,
      "learning_rate": 8.732e-05,
      "loss": 0.4165,
      "step": 17473
    },
    {
      "epoch": 0.9784970321424572,
      "grad_norm": 1.3002980947494507,
      "learning_rate": 8.7325e-05,
      "loss": 0.4092,
      "step": 17474
    },
    {
      "epoch": 0.9785530294545862,
      "grad_norm": 1.4328842163085938,
      "learning_rate": 8.733e-05,
      "loss": 0.3841,
      "step": 17475
    },
    {
      "epoch": 0.9786090267667152,
      "grad_norm": 1.2100359201431274,
      "learning_rate": 8.7335e-05,
      "loss": 0.3087,
      "step": 17476
    },
    {
      "epoch": 0.9786650240788443,
      "grad_norm": 1.296263337135315,
      "learning_rate": 8.734e-05,
      "loss": 0.4679,
      "step": 17477
    },
    {
      "epoch": 0.9787210213909733,
      "grad_norm": 1.3720515966415405,
      "learning_rate": 8.7345e-05,
      "loss": 0.5072,
      "step": 17478
    },
    {
      "epoch": 0.9787770187031023,
      "grad_norm": 1.2821012735366821,
      "learning_rate": 8.735000000000001e-05,
      "loss": 0.4174,
      "step": 17479
    },
    {
      "epoch": 0.9788330160152313,
      "grad_norm": 1.033850073814392,
      "learning_rate": 8.7355e-05,
      "loss": 0.429,
      "step": 17480
    },
    {
      "epoch": 0.9788890133273603,
      "grad_norm": 1.2118123769760132,
      "learning_rate": 8.736e-05,
      "loss": 0.3842,
      "step": 17481
    },
    {
      "epoch": 0.9789450106394894,
      "grad_norm": 1.198876142501831,
      "learning_rate": 8.7365e-05,
      "loss": 0.4736,
      "step": 17482
    },
    {
      "epoch": 0.9790010079516184,
      "grad_norm": 2.4156970977783203,
      "learning_rate": 8.737000000000001e-05,
      "loss": 0.5503,
      "step": 17483
    },
    {
      "epoch": 0.9790570052637474,
      "grad_norm": 1.1911977529525757,
      "learning_rate": 8.737500000000001e-05,
      "loss": 0.4204,
      "step": 17484
    },
    {
      "epoch": 0.9791130025758764,
      "grad_norm": 1.6712185144424438,
      "learning_rate": 8.738000000000001e-05,
      "loss": 0.4103,
      "step": 17485
    },
    {
      "epoch": 0.9791689998880053,
      "grad_norm": 1.511715292930603,
      "learning_rate": 8.7385e-05,
      "loss": 0.4508,
      "step": 17486
    },
    {
      "epoch": 0.9792249972001343,
      "grad_norm": 1.4734218120574951,
      "learning_rate": 8.739e-05,
      "loss": 0.4137,
      "step": 17487
    },
    {
      "epoch": 0.9792809945122634,
      "grad_norm": 1.347332239151001,
      "learning_rate": 8.7395e-05,
      "loss": 0.4178,
      "step": 17488
    },
    {
      "epoch": 0.9793369918243924,
      "grad_norm": 1.2072690725326538,
      "learning_rate": 8.740000000000001e-05,
      "loss": 0.4265,
      "step": 17489
    },
    {
      "epoch": 0.9793929891365214,
      "grad_norm": 1.4905452728271484,
      "learning_rate": 8.740500000000001e-05,
      "loss": 0.5137,
      "step": 17490
    },
    {
      "epoch": 0.9794489864486504,
      "grad_norm": 1.2617844343185425,
      "learning_rate": 8.741e-05,
      "loss": 0.303,
      "step": 17491
    },
    {
      "epoch": 0.9795049837607794,
      "grad_norm": 1.215404748916626,
      "learning_rate": 8.7415e-05,
      "loss": 0.4278,
      "step": 17492
    },
    {
      "epoch": 0.9795609810729085,
      "grad_norm": 1.5266979932785034,
      "learning_rate": 8.742e-05,
      "loss": 0.435,
      "step": 17493
    },
    {
      "epoch": 0.9796169783850375,
      "grad_norm": 1.3276944160461426,
      "learning_rate": 8.7425e-05,
      "loss": 0.5681,
      "step": 17494
    },
    {
      "epoch": 0.9796729756971665,
      "grad_norm": 1.4566097259521484,
      "learning_rate": 8.743e-05,
      "loss": 0.5974,
      "step": 17495
    },
    {
      "epoch": 0.9797289730092955,
      "grad_norm": 1.4847010374069214,
      "learning_rate": 8.7435e-05,
      "loss": 0.4646,
      "step": 17496
    },
    {
      "epoch": 0.9797849703214245,
      "grad_norm": 1.2967168092727661,
      "learning_rate": 8.744e-05,
      "loss": 0.4152,
      "step": 17497
    },
    {
      "epoch": 0.9798409676335536,
      "grad_norm": 1.37638521194458,
      "learning_rate": 8.7445e-05,
      "loss": 0.3974,
      "step": 17498
    },
    {
      "epoch": 0.9798969649456826,
      "grad_norm": 1.160549283027649,
      "learning_rate": 8.745000000000001e-05,
      "loss": 0.4809,
      "step": 17499
    },
    {
      "epoch": 0.9799529622578116,
      "grad_norm": 1.4034357070922852,
      "learning_rate": 8.745500000000001e-05,
      "loss": 0.4828,
      "step": 17500
    },
    {
      "epoch": 0.9800089595699406,
      "grad_norm": 1.6993513107299805,
      "learning_rate": 8.746e-05,
      "loss": 0.403,
      "step": 17501
    },
    {
      "epoch": 0.9800649568820696,
      "grad_norm": 1.4433549642562866,
      "learning_rate": 8.7465e-05,
      "loss": 0.5207,
      "step": 17502
    },
    {
      "epoch": 0.9801209541941986,
      "grad_norm": 1.9773705005645752,
      "learning_rate": 8.747e-05,
      "loss": 0.6333,
      "step": 17503
    },
    {
      "epoch": 0.9801769515063277,
      "grad_norm": 1.4764324426651,
      "learning_rate": 8.747500000000001e-05,
      "loss": 0.5438,
      "step": 17504
    },
    {
      "epoch": 0.9802329488184567,
      "grad_norm": 1.410233736038208,
      "learning_rate": 8.748000000000001e-05,
      "loss": 0.5832,
      "step": 17505
    },
    {
      "epoch": 0.9802889461305857,
      "grad_norm": 1.5553776025772095,
      "learning_rate": 8.748500000000001e-05,
      "loss": 0.5932,
      "step": 17506
    },
    {
      "epoch": 0.9803449434427147,
      "grad_norm": 1.5294225215911865,
      "learning_rate": 8.749e-05,
      "loss": 0.6163,
      "step": 17507
    },
    {
      "epoch": 0.9804009407548437,
      "grad_norm": 1.3601484298706055,
      "learning_rate": 8.7495e-05,
      "loss": 0.5146,
      "step": 17508
    },
    {
      "epoch": 0.9804569380669728,
      "grad_norm": 1.3084996938705444,
      "learning_rate": 8.75e-05,
      "loss": 0.4682,
      "step": 17509
    },
    {
      "epoch": 0.9805129353791018,
      "grad_norm": 1.2972066402435303,
      "learning_rate": 8.750500000000001e-05,
      "loss": 0.4591,
      "step": 17510
    },
    {
      "epoch": 0.9805689326912308,
      "grad_norm": 1.3795164823532104,
      "learning_rate": 8.751000000000001e-05,
      "loss": 0.406,
      "step": 17511
    },
    {
      "epoch": 0.9806249300033598,
      "grad_norm": 1.280336618423462,
      "learning_rate": 8.7515e-05,
      "loss": 0.4563,
      "step": 17512
    },
    {
      "epoch": 0.9806809273154888,
      "grad_norm": 1.3981839418411255,
      "learning_rate": 8.752e-05,
      "loss": 0.4962,
      "step": 17513
    },
    {
      "epoch": 0.9807369246276179,
      "grad_norm": 1.3525123596191406,
      "learning_rate": 8.7525e-05,
      "loss": 0.3947,
      "step": 17514
    },
    {
      "epoch": 0.9807929219397469,
      "grad_norm": 1.4933942556381226,
      "learning_rate": 8.753e-05,
      "loss": 0.4733,
      "step": 17515
    },
    {
      "epoch": 0.9808489192518759,
      "grad_norm": 1.405686378479004,
      "learning_rate": 8.7535e-05,
      "loss": 0.467,
      "step": 17516
    },
    {
      "epoch": 0.9809049165640049,
      "grad_norm": 1.1117249727249146,
      "learning_rate": 8.754e-05,
      "loss": 0.3743,
      "step": 17517
    },
    {
      "epoch": 0.9809609138761339,
      "grad_norm": 1.4071615934371948,
      "learning_rate": 8.7545e-05,
      "loss": 0.3418,
      "step": 17518
    },
    {
      "epoch": 0.981016911188263,
      "grad_norm": 1.627893090248108,
      "learning_rate": 8.755e-05,
      "loss": 0.6502,
      "step": 17519
    },
    {
      "epoch": 0.981072908500392,
      "grad_norm": 1.286058783531189,
      "learning_rate": 8.755500000000001e-05,
      "loss": 0.4648,
      "step": 17520
    },
    {
      "epoch": 0.981128905812521,
      "grad_norm": 1.3681645393371582,
      "learning_rate": 8.756000000000001e-05,
      "loss": 0.4158,
      "step": 17521
    },
    {
      "epoch": 0.98118490312465,
      "grad_norm": 1.4059162139892578,
      "learning_rate": 8.7565e-05,
      "loss": 0.4763,
      "step": 17522
    },
    {
      "epoch": 0.981240900436779,
      "grad_norm": 126.30540466308594,
      "learning_rate": 8.757e-05,
      "loss": 0.4891,
      "step": 17523
    },
    {
      "epoch": 0.981296897748908,
      "grad_norm": 1.467044472694397,
      "learning_rate": 8.7575e-05,
      "loss": 0.576,
      "step": 17524
    },
    {
      "epoch": 0.9813528950610371,
      "grad_norm": 1.4908976554870605,
      "learning_rate": 8.758000000000001e-05,
      "loss": 0.506,
      "step": 17525
    },
    {
      "epoch": 0.9814088923731661,
      "grad_norm": 1.2298582792282104,
      "learning_rate": 8.758500000000001e-05,
      "loss": 0.4315,
      "step": 17526
    },
    {
      "epoch": 0.9814648896852951,
      "grad_norm": 1.304078221321106,
      "learning_rate": 8.759e-05,
      "loss": 0.42,
      "step": 17527
    },
    {
      "epoch": 0.9815208869974241,
      "grad_norm": 1.5696616172790527,
      "learning_rate": 8.7595e-05,
      "loss": 0.5689,
      "step": 17528
    },
    {
      "epoch": 0.9815768843095531,
      "grad_norm": 1.8061628341674805,
      "learning_rate": 8.76e-05,
      "loss": 0.5197,
      "step": 17529
    },
    {
      "epoch": 0.9816328816216822,
      "grad_norm": 2.1518683433532715,
      "learning_rate": 8.7605e-05,
      "loss": 0.597,
      "step": 17530
    },
    {
      "epoch": 0.9816888789338112,
      "grad_norm": 1.4769752025604248,
      "learning_rate": 8.761000000000001e-05,
      "loss": 0.5969,
      "step": 17531
    },
    {
      "epoch": 0.9817448762459402,
      "grad_norm": 1.3505018949508667,
      "learning_rate": 8.761500000000001e-05,
      "loss": 0.4329,
      "step": 17532
    },
    {
      "epoch": 0.9818008735580692,
      "grad_norm": 1.226112723350525,
      "learning_rate": 8.762e-05,
      "loss": 0.4346,
      "step": 17533
    },
    {
      "epoch": 0.9818568708701982,
      "grad_norm": 1.3486183881759644,
      "learning_rate": 8.7625e-05,
      "loss": 0.409,
      "step": 17534
    },
    {
      "epoch": 0.9819128681823273,
      "grad_norm": 1.6185193061828613,
      "learning_rate": 8.763e-05,
      "loss": 0.4879,
      "step": 17535
    },
    {
      "epoch": 0.9819688654944563,
      "grad_norm": 1.1983236074447632,
      "learning_rate": 8.7635e-05,
      "loss": 0.2968,
      "step": 17536
    },
    {
      "epoch": 0.9820248628065853,
      "grad_norm": 1.2322864532470703,
      "learning_rate": 8.764e-05,
      "loss": 0.4065,
      "step": 17537
    },
    {
      "epoch": 0.9820808601187143,
      "grad_norm": 1.4110888242721558,
      "learning_rate": 8.7645e-05,
      "loss": 0.4467,
      "step": 17538
    },
    {
      "epoch": 0.9821368574308433,
      "grad_norm": 1.2793995141983032,
      "learning_rate": 8.765e-05,
      "loss": 0.4855,
      "step": 17539
    },
    {
      "epoch": 0.9821928547429724,
      "grad_norm": 1.2857661247253418,
      "learning_rate": 8.765500000000001e-05,
      "loss": 0.4428,
      "step": 17540
    },
    {
      "epoch": 0.9822488520551014,
      "grad_norm": 1.4148931503295898,
      "learning_rate": 8.766000000000001e-05,
      "loss": 0.3957,
      "step": 17541
    },
    {
      "epoch": 0.9823048493672304,
      "grad_norm": 1.4639850854873657,
      "learning_rate": 8.766500000000001e-05,
      "loss": 0.5383,
      "step": 17542
    },
    {
      "epoch": 0.9823608466793594,
      "grad_norm": 1.7329463958740234,
      "learning_rate": 8.767e-05,
      "loss": 0.3988,
      "step": 17543
    },
    {
      "epoch": 0.9824168439914884,
      "grad_norm": 1.2866063117980957,
      "learning_rate": 8.7675e-05,
      "loss": 0.4273,
      "step": 17544
    },
    {
      "epoch": 0.9824728413036175,
      "grad_norm": 1.3373000621795654,
      "learning_rate": 8.768e-05,
      "loss": 0.5587,
      "step": 17545
    },
    {
      "epoch": 0.9825288386157465,
      "grad_norm": 1.371846318244934,
      "learning_rate": 8.768500000000001e-05,
      "loss": 0.451,
      "step": 17546
    },
    {
      "epoch": 0.9825848359278755,
      "grad_norm": 2.8197245597839355,
      "learning_rate": 8.769000000000001e-05,
      "loss": 0.4816,
      "step": 17547
    },
    {
      "epoch": 0.9826408332400045,
      "grad_norm": 3.393686532974243,
      "learning_rate": 8.7695e-05,
      "loss": 0.4298,
      "step": 17548
    },
    {
      "epoch": 0.9826968305521335,
      "grad_norm": 1.4075634479522705,
      "learning_rate": 8.77e-05,
      "loss": 0.4598,
      "step": 17549
    },
    {
      "epoch": 0.9827528278642625,
      "grad_norm": 1.9391776323318481,
      "learning_rate": 8.7705e-05,
      "loss": 0.5691,
      "step": 17550
    },
    {
      "epoch": 0.9828088251763916,
      "grad_norm": 1.6095472574234009,
      "learning_rate": 8.771e-05,
      "loss": 0.7115,
      "step": 17551
    },
    {
      "epoch": 0.9828648224885206,
      "grad_norm": 1.1371383666992188,
      "learning_rate": 8.7715e-05,
      "loss": 0.3423,
      "step": 17552
    },
    {
      "epoch": 0.9829208198006496,
      "grad_norm": 1.347645878791809,
      "learning_rate": 8.772000000000001e-05,
      "loss": 0.463,
      "step": 17553
    },
    {
      "epoch": 0.9829768171127786,
      "grad_norm": 1.7110610008239746,
      "learning_rate": 8.7725e-05,
      "loss": 0.5672,
      "step": 17554
    },
    {
      "epoch": 0.9830328144249076,
      "grad_norm": 1.2346199750900269,
      "learning_rate": 8.773e-05,
      "loss": 0.4478,
      "step": 17555
    },
    {
      "epoch": 0.9830888117370367,
      "grad_norm": 1.3220741748809814,
      "learning_rate": 8.7735e-05,
      "loss": 0.4089,
      "step": 17556
    },
    {
      "epoch": 0.9831448090491657,
      "grad_norm": 1.3453768491744995,
      "learning_rate": 8.774e-05,
      "loss": 0.4418,
      "step": 17557
    },
    {
      "epoch": 0.9832008063612947,
      "grad_norm": 1.3212966918945312,
      "learning_rate": 8.7745e-05,
      "loss": 0.383,
      "step": 17558
    },
    {
      "epoch": 0.9832568036734237,
      "grad_norm": 1.2123596668243408,
      "learning_rate": 8.775e-05,
      "loss": 0.4264,
      "step": 17559
    },
    {
      "epoch": 0.9833128009855527,
      "grad_norm": 1.3323320150375366,
      "learning_rate": 8.775500000000002e-05,
      "loss": 0.4671,
      "step": 17560
    },
    {
      "epoch": 0.9833687982976818,
      "grad_norm": 1.3458378314971924,
      "learning_rate": 8.776000000000001e-05,
      "loss": 0.407,
      "step": 17561
    },
    {
      "epoch": 0.9834247956098108,
      "grad_norm": 1.4096357822418213,
      "learning_rate": 8.776500000000001e-05,
      "loss": 0.4173,
      "step": 17562
    },
    {
      "epoch": 0.9834807929219398,
      "grad_norm": 1.3322290182113647,
      "learning_rate": 8.777000000000001e-05,
      "loss": 0.3692,
      "step": 17563
    },
    {
      "epoch": 0.9835367902340688,
      "grad_norm": 1.205859661102295,
      "learning_rate": 8.7775e-05,
      "loss": 0.4507,
      "step": 17564
    },
    {
      "epoch": 0.9835927875461978,
      "grad_norm": 1.7954727411270142,
      "learning_rate": 8.778e-05,
      "loss": 0.4743,
      "step": 17565
    },
    {
      "epoch": 0.9836487848583269,
      "grad_norm": 1.2893177270889282,
      "learning_rate": 8.7785e-05,
      "loss": 0.379,
      "step": 17566
    },
    {
      "epoch": 0.9837047821704559,
      "grad_norm": 1.3160830736160278,
      "learning_rate": 8.779000000000001e-05,
      "loss": 0.4332,
      "step": 17567
    },
    {
      "epoch": 0.9837607794825849,
      "grad_norm": 1.1817042827606201,
      "learning_rate": 8.779500000000001e-05,
      "loss": 0.5421,
      "step": 17568
    },
    {
      "epoch": 0.9838167767947138,
      "grad_norm": 1.2465851306915283,
      "learning_rate": 8.78e-05,
      "loss": 0.4129,
      "step": 17569
    },
    {
      "epoch": 0.9838727741068428,
      "grad_norm": 1.3811452388763428,
      "learning_rate": 8.7805e-05,
      "loss": 0.4752,
      "step": 17570
    },
    {
      "epoch": 0.9839287714189718,
      "grad_norm": 1.6648184061050415,
      "learning_rate": 8.781e-05,
      "loss": 0.4578,
      "step": 17571
    },
    {
      "epoch": 0.9839847687311009,
      "grad_norm": 1.3105189800262451,
      "learning_rate": 8.7815e-05,
      "loss": 0.5465,
      "step": 17572
    },
    {
      "epoch": 0.9840407660432299,
      "grad_norm": 1.455780029296875,
      "learning_rate": 8.782e-05,
      "loss": 0.4989,
      "step": 17573
    },
    {
      "epoch": 0.9840967633553589,
      "grad_norm": 1.006490707397461,
      "learning_rate": 8.782500000000001e-05,
      "loss": 0.4122,
      "step": 17574
    },
    {
      "epoch": 0.9841527606674879,
      "grad_norm": 1.4180700778961182,
      "learning_rate": 8.783e-05,
      "loss": 0.5415,
      "step": 17575
    },
    {
      "epoch": 0.9842087579796169,
      "grad_norm": 1.6900314092636108,
      "learning_rate": 8.7835e-05,
      "loss": 0.5968,
      "step": 17576
    },
    {
      "epoch": 0.984264755291746,
      "grad_norm": 1.7504936456680298,
      "learning_rate": 8.784e-05,
      "loss": 0.5195,
      "step": 17577
    },
    {
      "epoch": 0.984320752603875,
      "grad_norm": 1.4187743663787842,
      "learning_rate": 8.7845e-05,
      "loss": 0.45,
      "step": 17578
    },
    {
      "epoch": 0.984376749916004,
      "grad_norm": 1.480238914489746,
      "learning_rate": 8.785e-05,
      "loss": 0.5978,
      "step": 17579
    },
    {
      "epoch": 0.984432747228133,
      "grad_norm": 1.2080800533294678,
      "learning_rate": 8.7855e-05,
      "loss": 0.4829,
      "step": 17580
    },
    {
      "epoch": 0.984488744540262,
      "grad_norm": 1.6299505233764648,
      "learning_rate": 8.786e-05,
      "loss": 0.501,
      "step": 17581
    },
    {
      "epoch": 0.984544741852391,
      "grad_norm": 1.3159761428833008,
      "learning_rate": 8.786500000000001e-05,
      "loss": 0.4247,
      "step": 17582
    },
    {
      "epoch": 0.9846007391645201,
      "grad_norm": 1.294948935508728,
      "learning_rate": 8.787000000000001e-05,
      "loss": 0.3801,
      "step": 17583
    },
    {
      "epoch": 0.9846567364766491,
      "grad_norm": 1.212965488433838,
      "learning_rate": 8.787500000000001e-05,
      "loss": 0.3939,
      "step": 17584
    },
    {
      "epoch": 0.9847127337887781,
      "grad_norm": 1.37157142162323,
      "learning_rate": 8.788e-05,
      "loss": 0.634,
      "step": 17585
    },
    {
      "epoch": 0.9847687311009071,
      "grad_norm": 2.70932936668396,
      "learning_rate": 8.7885e-05,
      "loss": 0.5367,
      "step": 17586
    },
    {
      "epoch": 0.9848247284130361,
      "grad_norm": 1.6919304132461548,
      "learning_rate": 8.789e-05,
      "loss": 0.5444,
      "step": 17587
    },
    {
      "epoch": 0.9848807257251652,
      "grad_norm": 1.4852466583251953,
      "learning_rate": 8.789500000000001e-05,
      "loss": 0.4759,
      "step": 17588
    },
    {
      "epoch": 0.9849367230372942,
      "grad_norm": 1.3218944072723389,
      "learning_rate": 8.790000000000001e-05,
      "loss": 0.4551,
      "step": 17589
    },
    {
      "epoch": 0.9849927203494232,
      "grad_norm": 1.262381672859192,
      "learning_rate": 8.7905e-05,
      "loss": 0.3755,
      "step": 17590
    },
    {
      "epoch": 0.9850487176615522,
      "grad_norm": 1.518549919128418,
      "learning_rate": 8.791e-05,
      "loss": 0.4507,
      "step": 17591
    },
    {
      "epoch": 0.9851047149736812,
      "grad_norm": 1.26132333278656,
      "learning_rate": 8.7915e-05,
      "loss": 0.5181,
      "step": 17592
    },
    {
      "epoch": 0.9851607122858103,
      "grad_norm": 1.578444480895996,
      "learning_rate": 8.792e-05,
      "loss": 0.4636,
      "step": 17593
    },
    {
      "epoch": 0.9852167095979393,
      "grad_norm": 1.3818082809448242,
      "learning_rate": 8.7925e-05,
      "loss": 0.5473,
      "step": 17594
    },
    {
      "epoch": 0.9852727069100683,
      "grad_norm": 1.6644206047058105,
      "learning_rate": 8.793000000000001e-05,
      "loss": 0.4045,
      "step": 17595
    },
    {
      "epoch": 0.9853287042221973,
      "grad_norm": 1.2437721490859985,
      "learning_rate": 8.7935e-05,
      "loss": 0.5042,
      "step": 17596
    },
    {
      "epoch": 0.9853847015343263,
      "grad_norm": 1.3465697765350342,
      "learning_rate": 8.794e-05,
      "loss": 0.423,
      "step": 17597
    },
    {
      "epoch": 0.9854406988464554,
      "grad_norm": 1.3346095085144043,
      "learning_rate": 8.7945e-05,
      "loss": 0.4192,
      "step": 17598
    },
    {
      "epoch": 0.9854966961585844,
      "grad_norm": 1.2999358177185059,
      "learning_rate": 8.795e-05,
      "loss": 0.461,
      "step": 17599
    },
    {
      "epoch": 0.9855526934707134,
      "grad_norm": 1.4455510377883911,
      "learning_rate": 8.795500000000001e-05,
      "loss": 0.5385,
      "step": 17600
    },
    {
      "epoch": 0.9856086907828424,
      "grad_norm": 1.5050334930419922,
      "learning_rate": 8.796e-05,
      "loss": 0.5009,
      "step": 17601
    },
    {
      "epoch": 0.9856646880949714,
      "grad_norm": 1.2636960744857788,
      "learning_rate": 8.7965e-05,
      "loss": 0.4595,
      "step": 17602
    },
    {
      "epoch": 0.9857206854071004,
      "grad_norm": 1.377002477645874,
      "learning_rate": 8.797000000000001e-05,
      "loss": 0.4101,
      "step": 17603
    },
    {
      "epoch": 0.9857766827192295,
      "grad_norm": 1.3277605772018433,
      "learning_rate": 8.797500000000001e-05,
      "loss": 0.4982,
      "step": 17604
    },
    {
      "epoch": 0.9858326800313585,
      "grad_norm": 1.494991660118103,
      "learning_rate": 8.798000000000001e-05,
      "loss": 0.4443,
      "step": 17605
    },
    {
      "epoch": 0.9858886773434875,
      "grad_norm": 1.3216522932052612,
      "learning_rate": 8.7985e-05,
      "loss": 0.4982,
      "step": 17606
    },
    {
      "epoch": 0.9859446746556165,
      "grad_norm": 1.3640851974487305,
      "learning_rate": 8.799e-05,
      "loss": 0.417,
      "step": 17607
    },
    {
      "epoch": 0.9860006719677455,
      "grad_norm": 1.2586225271224976,
      "learning_rate": 8.7995e-05,
      "loss": 0.499,
      "step": 17608
    },
    {
      "epoch": 0.9860566692798746,
      "grad_norm": 1.2728242874145508,
      "learning_rate": 8.800000000000001e-05,
      "loss": 0.3669,
      "step": 17609
    },
    {
      "epoch": 0.9861126665920036,
      "grad_norm": 1.3577446937561035,
      "learning_rate": 8.800500000000001e-05,
      "loss": 0.6406,
      "step": 17610
    },
    {
      "epoch": 0.9861686639041326,
      "grad_norm": 1.105636715888977,
      "learning_rate": 8.801e-05,
      "loss": 0.3837,
      "step": 17611
    },
    {
      "epoch": 0.9862246612162616,
      "grad_norm": 1.332719087600708,
      "learning_rate": 8.8015e-05,
      "loss": 0.5369,
      "step": 17612
    },
    {
      "epoch": 0.9862806585283906,
      "grad_norm": 1.1900907754898071,
      "learning_rate": 8.802e-05,
      "loss": 0.4035,
      "step": 17613
    },
    {
      "epoch": 0.9863366558405197,
      "grad_norm": 1.3678020238876343,
      "learning_rate": 8.8025e-05,
      "loss": 0.4893,
      "step": 17614
    },
    {
      "epoch": 0.9863926531526487,
      "grad_norm": 1.2033085823059082,
      "learning_rate": 8.803e-05,
      "loss": 0.4647,
      "step": 17615
    },
    {
      "epoch": 0.9864486504647777,
      "grad_norm": 1.5403302907943726,
      "learning_rate": 8.8035e-05,
      "loss": 0.4371,
      "step": 17616
    },
    {
      "epoch": 0.9865046477769067,
      "grad_norm": 1.3274661302566528,
      "learning_rate": 8.804e-05,
      "loss": 0.446,
      "step": 17617
    },
    {
      "epoch": 0.9865606450890357,
      "grad_norm": 1.631810188293457,
      "learning_rate": 8.8045e-05,
      "loss": 0.4998,
      "step": 17618
    },
    {
      "epoch": 0.9866166424011648,
      "grad_norm": 11.381399154663086,
      "learning_rate": 8.805e-05,
      "loss": 0.4679,
      "step": 17619
    },
    {
      "epoch": 0.9866726397132938,
      "grad_norm": 1.4126825332641602,
      "learning_rate": 8.805500000000001e-05,
      "loss": 0.4284,
      "step": 17620
    },
    {
      "epoch": 0.9867286370254228,
      "grad_norm": 1.444777488708496,
      "learning_rate": 8.806000000000001e-05,
      "loss": 0.4622,
      "step": 17621
    },
    {
      "epoch": 0.9867846343375518,
      "grad_norm": 1.8333213329315186,
      "learning_rate": 8.8065e-05,
      "loss": 0.6327,
      "step": 17622
    },
    {
      "epoch": 0.9868406316496808,
      "grad_norm": 1.1897823810577393,
      "learning_rate": 8.807e-05,
      "loss": 0.4171,
      "step": 17623
    },
    {
      "epoch": 0.9868966289618099,
      "grad_norm": 1.3115020990371704,
      "learning_rate": 8.807500000000001e-05,
      "loss": 0.3187,
      "step": 17624
    },
    {
      "epoch": 0.9869526262739389,
      "grad_norm": 1.45095956325531,
      "learning_rate": 8.808000000000001e-05,
      "loss": 0.4641,
      "step": 17625
    },
    {
      "epoch": 0.9870086235860679,
      "grad_norm": 1.350498914718628,
      "learning_rate": 8.808500000000001e-05,
      "loss": 0.4694,
      "step": 17626
    },
    {
      "epoch": 0.9870646208981969,
      "grad_norm": 1.168498158454895,
      "learning_rate": 8.809e-05,
      "loss": 0.4279,
      "step": 17627
    },
    {
      "epoch": 0.9871206182103259,
      "grad_norm": 1.2061278820037842,
      "learning_rate": 8.8095e-05,
      "loss": 0.5163,
      "step": 17628
    },
    {
      "epoch": 0.987176615522455,
      "grad_norm": 1.3452043533325195,
      "learning_rate": 8.81e-05,
      "loss": 0.5216,
      "step": 17629
    },
    {
      "epoch": 0.987232612834584,
      "grad_norm": 1.3370094299316406,
      "learning_rate": 8.8105e-05,
      "loss": 0.4456,
      "step": 17630
    },
    {
      "epoch": 0.987288610146713,
      "grad_norm": 1.491627812385559,
      "learning_rate": 8.811000000000001e-05,
      "loss": 0.5166,
      "step": 17631
    },
    {
      "epoch": 0.987344607458842,
      "grad_norm": 1.3679686784744263,
      "learning_rate": 8.8115e-05,
      "loss": 0.3736,
      "step": 17632
    },
    {
      "epoch": 0.987400604770971,
      "grad_norm": 1.4824379682540894,
      "learning_rate": 8.812e-05,
      "loss": 0.4467,
      "step": 17633
    },
    {
      "epoch": 0.9874566020831,
      "grad_norm": 0.9939557909965515,
      "learning_rate": 8.8125e-05,
      "loss": 0.4024,
      "step": 17634
    },
    {
      "epoch": 0.9875125993952291,
      "grad_norm": 1.2251510620117188,
      "learning_rate": 8.813e-05,
      "loss": 0.4063,
      "step": 17635
    },
    {
      "epoch": 0.9875685967073581,
      "grad_norm": 1.4843429327011108,
      "learning_rate": 8.8135e-05,
      "loss": 0.5691,
      "step": 17636
    },
    {
      "epoch": 0.9876245940194871,
      "grad_norm": 1.5110747814178467,
      "learning_rate": 8.814e-05,
      "loss": 0.7143,
      "step": 17637
    },
    {
      "epoch": 0.9876805913316161,
      "grad_norm": 1.2455945014953613,
      "learning_rate": 8.8145e-05,
      "loss": 0.3352,
      "step": 17638
    },
    {
      "epoch": 0.9877365886437451,
      "grad_norm": 1.3711780309677124,
      "learning_rate": 8.815e-05,
      "loss": 0.5051,
      "step": 17639
    },
    {
      "epoch": 0.9877925859558742,
      "grad_norm": 1.326949954032898,
      "learning_rate": 8.8155e-05,
      "loss": 0.5136,
      "step": 17640
    },
    {
      "epoch": 0.9878485832680032,
      "grad_norm": 1.5293852090835571,
      "learning_rate": 8.816000000000001e-05,
      "loss": 0.5172,
      "step": 17641
    },
    {
      "epoch": 0.9879045805801322,
      "grad_norm": 1.548806071281433,
      "learning_rate": 8.816500000000001e-05,
      "loss": 0.4185,
      "step": 17642
    },
    {
      "epoch": 0.9879605778922612,
      "grad_norm": 1.25412917137146,
      "learning_rate": 8.817e-05,
      "loss": 0.495,
      "step": 17643
    },
    {
      "epoch": 0.9880165752043902,
      "grad_norm": 1.476805329322815,
      "learning_rate": 8.8175e-05,
      "loss": 0.4058,
      "step": 17644
    },
    {
      "epoch": 0.9880725725165193,
      "grad_norm": 1.4195153713226318,
      "learning_rate": 8.818000000000001e-05,
      "loss": 0.4833,
      "step": 17645
    },
    {
      "epoch": 0.9881285698286483,
      "grad_norm": 1.246290683746338,
      "learning_rate": 8.818500000000001e-05,
      "loss": 0.3048,
      "step": 17646
    },
    {
      "epoch": 0.9881845671407773,
      "grad_norm": 1.3022139072418213,
      "learning_rate": 8.819000000000001e-05,
      "loss": 0.437,
      "step": 17647
    },
    {
      "epoch": 0.9882405644529063,
      "grad_norm": 1.6030832529067993,
      "learning_rate": 8.8195e-05,
      "loss": 0.3815,
      "step": 17648
    },
    {
      "epoch": 0.9882965617650353,
      "grad_norm": 1.3837579488754272,
      "learning_rate": 8.82e-05,
      "loss": 0.5043,
      "step": 17649
    },
    {
      "epoch": 0.9883525590771643,
      "grad_norm": 1.3769114017486572,
      "learning_rate": 8.8205e-05,
      "loss": 0.4242,
      "step": 17650
    },
    {
      "epoch": 0.9884085563892934,
      "grad_norm": 1.2110369205474854,
      "learning_rate": 8.821e-05,
      "loss": 0.5027,
      "step": 17651
    },
    {
      "epoch": 0.9884645537014223,
      "grad_norm": 1.4913884401321411,
      "learning_rate": 8.821500000000001e-05,
      "loss": 0.4245,
      "step": 17652
    },
    {
      "epoch": 0.9885205510135513,
      "grad_norm": 1.4101661443710327,
      "learning_rate": 8.822e-05,
      "loss": 0.4611,
      "step": 17653
    },
    {
      "epoch": 0.9885765483256803,
      "grad_norm": 1.5386029481887817,
      "learning_rate": 8.8225e-05,
      "loss": 0.7085,
      "step": 17654
    },
    {
      "epoch": 0.9886325456378093,
      "grad_norm": 1.4647287130355835,
      "learning_rate": 8.823e-05,
      "loss": 0.5565,
      "step": 17655
    },
    {
      "epoch": 0.9886885429499384,
      "grad_norm": 1.3486679792404175,
      "learning_rate": 8.8235e-05,
      "loss": 0.4155,
      "step": 17656
    },
    {
      "epoch": 0.9887445402620674,
      "grad_norm": 1.1760742664337158,
      "learning_rate": 8.824e-05,
      "loss": 0.42,
      "step": 17657
    },
    {
      "epoch": 0.9888005375741964,
      "grad_norm": 10.632444381713867,
      "learning_rate": 8.8245e-05,
      "loss": 0.4341,
      "step": 17658
    },
    {
      "epoch": 0.9888565348863254,
      "grad_norm": 1.3989155292510986,
      "learning_rate": 8.825e-05,
      "loss": 0.491,
      "step": 17659
    },
    {
      "epoch": 0.9889125321984544,
      "grad_norm": 1.2893259525299072,
      "learning_rate": 8.8255e-05,
      "loss": 0.4318,
      "step": 17660
    },
    {
      "epoch": 0.9889685295105834,
      "grad_norm": 1.8009469509124756,
      "learning_rate": 8.826000000000001e-05,
      "loss": 0.5432,
      "step": 17661
    },
    {
      "epoch": 0.9890245268227125,
      "grad_norm": 1.3617624044418335,
      "learning_rate": 8.826500000000001e-05,
      "loss": 0.4691,
      "step": 17662
    },
    {
      "epoch": 0.9890805241348415,
      "grad_norm": 1.30439293384552,
      "learning_rate": 8.827000000000001e-05,
      "loss": 0.4716,
      "step": 17663
    },
    {
      "epoch": 0.9891365214469705,
      "grad_norm": 1.1191641092300415,
      "learning_rate": 8.8275e-05,
      "loss": 0.3351,
      "step": 17664
    },
    {
      "epoch": 0.9891925187590995,
      "grad_norm": 1.3394598960876465,
      "learning_rate": 8.828e-05,
      "loss": 0.4725,
      "step": 17665
    },
    {
      "epoch": 0.9892485160712285,
      "grad_norm": 1.580841302871704,
      "learning_rate": 8.828500000000001e-05,
      "loss": 0.6894,
      "step": 17666
    },
    {
      "epoch": 0.9893045133833576,
      "grad_norm": 1.1006057262420654,
      "learning_rate": 8.829000000000001e-05,
      "loss": 0.3338,
      "step": 17667
    },
    {
      "epoch": 0.9893605106954866,
      "grad_norm": 1.2439556121826172,
      "learning_rate": 8.829500000000001e-05,
      "loss": 0.5283,
      "step": 17668
    },
    {
      "epoch": 0.9894165080076156,
      "grad_norm": 1.3054133653640747,
      "learning_rate": 8.83e-05,
      "loss": 0.4204,
      "step": 17669
    },
    {
      "epoch": 0.9894725053197446,
      "grad_norm": 1.3397842645645142,
      "learning_rate": 8.8305e-05,
      "loss": 0.39,
      "step": 17670
    },
    {
      "epoch": 0.9895285026318736,
      "grad_norm": 1.3987393379211426,
      "learning_rate": 8.831e-05,
      "loss": 0.5048,
      "step": 17671
    },
    {
      "epoch": 0.9895844999440027,
      "grad_norm": 1.7089719772338867,
      "learning_rate": 8.8315e-05,
      "loss": 0.4487,
      "step": 17672
    },
    {
      "epoch": 0.9896404972561317,
      "grad_norm": 1.145412802696228,
      "learning_rate": 8.832000000000001e-05,
      "loss": 0.452,
      "step": 17673
    },
    {
      "epoch": 0.9896964945682607,
      "grad_norm": 1.261096715927124,
      "learning_rate": 8.8325e-05,
      "loss": 0.3973,
      "step": 17674
    },
    {
      "epoch": 0.9897524918803897,
      "grad_norm": 1.509187936782837,
      "learning_rate": 8.833e-05,
      "loss": 0.5809,
      "step": 17675
    },
    {
      "epoch": 0.9898084891925187,
      "grad_norm": 1.092221975326538,
      "learning_rate": 8.8335e-05,
      "loss": 0.3629,
      "step": 17676
    },
    {
      "epoch": 0.9898644865046478,
      "grad_norm": 1.2134952545166016,
      "learning_rate": 8.834e-05,
      "loss": 0.4333,
      "step": 17677
    },
    {
      "epoch": 0.9899204838167768,
      "grad_norm": 1.1407698392868042,
      "learning_rate": 8.8345e-05,
      "loss": 0.3594,
      "step": 17678
    },
    {
      "epoch": 0.9899764811289058,
      "grad_norm": 1.4732252359390259,
      "learning_rate": 8.834999999999999e-05,
      "loss": 0.3671,
      "step": 17679
    },
    {
      "epoch": 0.9900324784410348,
      "grad_norm": 1.455226182937622,
      "learning_rate": 8.8355e-05,
      "loss": 0.448,
      "step": 17680
    },
    {
      "epoch": 0.9900884757531638,
      "grad_norm": 1.2765408754348755,
      "learning_rate": 8.836000000000001e-05,
      "loss": 0.464,
      "step": 17681
    },
    {
      "epoch": 0.9901444730652929,
      "grad_norm": 1.222954511642456,
      "learning_rate": 8.836500000000001e-05,
      "loss": 0.4208,
      "step": 17682
    },
    {
      "epoch": 0.9902004703774219,
      "grad_norm": 2.3051717281341553,
      "learning_rate": 8.837000000000001e-05,
      "loss": 0.647,
      "step": 17683
    },
    {
      "epoch": 0.9902564676895509,
      "grad_norm": 1.7088819742202759,
      "learning_rate": 8.837500000000001e-05,
      "loss": 0.4154,
      "step": 17684
    },
    {
      "epoch": 0.9903124650016799,
      "grad_norm": 1.267587661743164,
      "learning_rate": 8.838e-05,
      "loss": 0.404,
      "step": 17685
    },
    {
      "epoch": 0.9903684623138089,
      "grad_norm": 1.8990023136138916,
      "learning_rate": 8.8385e-05,
      "loss": 0.4948,
      "step": 17686
    },
    {
      "epoch": 0.990424459625938,
      "grad_norm": 1.133603572845459,
      "learning_rate": 8.839000000000001e-05,
      "loss": 0.3743,
      "step": 17687
    },
    {
      "epoch": 0.990480456938067,
      "grad_norm": 1.6615586280822754,
      "learning_rate": 8.839500000000001e-05,
      "loss": 0.5648,
      "step": 17688
    },
    {
      "epoch": 0.990536454250196,
      "grad_norm": 1.5147998332977295,
      "learning_rate": 8.840000000000001e-05,
      "loss": 0.4682,
      "step": 17689
    },
    {
      "epoch": 0.990592451562325,
      "grad_norm": 1.6206152439117432,
      "learning_rate": 8.8405e-05,
      "loss": 0.47,
      "step": 17690
    },
    {
      "epoch": 0.990648448874454,
      "grad_norm": 1.3913646936416626,
      "learning_rate": 8.841e-05,
      "loss": 0.4006,
      "step": 17691
    },
    {
      "epoch": 0.990704446186583,
      "grad_norm": 1.4173760414123535,
      "learning_rate": 8.8415e-05,
      "loss": 0.4517,
      "step": 17692
    },
    {
      "epoch": 0.9907604434987121,
      "grad_norm": 1.551391839981079,
      "learning_rate": 8.842e-05,
      "loss": 0.5697,
      "step": 17693
    },
    {
      "epoch": 0.9908164408108411,
      "grad_norm": 1.2823461294174194,
      "learning_rate": 8.842500000000001e-05,
      "loss": 0.3911,
      "step": 17694
    },
    {
      "epoch": 0.9908724381229701,
      "grad_norm": 1.2825812101364136,
      "learning_rate": 8.843e-05,
      "loss": 0.5166,
      "step": 17695
    },
    {
      "epoch": 0.9909284354350991,
      "grad_norm": 1.237261176109314,
      "learning_rate": 8.8435e-05,
      "loss": 0.4583,
      "step": 17696
    },
    {
      "epoch": 0.9909844327472281,
      "grad_norm": 1.2546817064285278,
      "learning_rate": 8.844e-05,
      "loss": 0.4322,
      "step": 17697
    },
    {
      "epoch": 0.9910404300593572,
      "grad_norm": 1.210726261138916,
      "learning_rate": 8.8445e-05,
      "loss": 0.4521,
      "step": 17698
    },
    {
      "epoch": 0.9910964273714862,
      "grad_norm": 1.7791892290115356,
      "learning_rate": 8.845e-05,
      "loss": 0.4756,
      "step": 17699
    },
    {
      "epoch": 0.9911524246836152,
      "grad_norm": 1.4492498636245728,
      "learning_rate": 8.845499999999999e-05,
      "loss": 0.4875,
      "step": 17700
    },
    {
      "epoch": 0.9912084219957442,
      "grad_norm": 1.4883558750152588,
      "learning_rate": 8.846e-05,
      "loss": 0.4393,
      "step": 17701
    },
    {
      "epoch": 0.9912644193078732,
      "grad_norm": 1.568536639213562,
      "learning_rate": 8.846500000000001e-05,
      "loss": 0.5384,
      "step": 17702
    },
    {
      "epoch": 0.9913204166200023,
      "grad_norm": 1.1386370658874512,
      "learning_rate": 8.847000000000001e-05,
      "loss": 0.4197,
      "step": 17703
    },
    {
      "epoch": 0.9913764139321313,
      "grad_norm": 1.130841851234436,
      "learning_rate": 8.847500000000001e-05,
      "loss": 0.553,
      "step": 17704
    },
    {
      "epoch": 0.9914324112442603,
      "grad_norm": 1.9779821634292603,
      "learning_rate": 8.848e-05,
      "loss": 0.4041,
      "step": 17705
    },
    {
      "epoch": 0.9914884085563893,
      "grad_norm": 1.2233911752700806,
      "learning_rate": 8.8485e-05,
      "loss": 0.4139,
      "step": 17706
    },
    {
      "epoch": 0.9915444058685183,
      "grad_norm": 1.1328318119049072,
      "learning_rate": 8.849e-05,
      "loss": 0.3593,
      "step": 17707
    },
    {
      "epoch": 0.9916004031806473,
      "grad_norm": 1.307611107826233,
      "learning_rate": 8.849500000000001e-05,
      "loss": 0.4044,
      "step": 17708
    },
    {
      "epoch": 0.9916564004927764,
      "grad_norm": 1.3479758501052856,
      "learning_rate": 8.850000000000001e-05,
      "loss": 0.4071,
      "step": 17709
    },
    {
      "epoch": 0.9917123978049054,
      "grad_norm": 1.8989241123199463,
      "learning_rate": 8.850500000000001e-05,
      "loss": 0.5629,
      "step": 17710
    },
    {
      "epoch": 0.9917683951170344,
      "grad_norm": 1.6310759782791138,
      "learning_rate": 8.851e-05,
      "loss": 0.3843,
      "step": 17711
    },
    {
      "epoch": 0.9918243924291634,
      "grad_norm": 1.456925392150879,
      "learning_rate": 8.8515e-05,
      "loss": 0.416,
      "step": 17712
    },
    {
      "epoch": 0.9918803897412924,
      "grad_norm": 1.4533954858779907,
      "learning_rate": 8.852e-05,
      "loss": 0.5127,
      "step": 17713
    },
    {
      "epoch": 0.9919363870534215,
      "grad_norm": 1.5278843641281128,
      "learning_rate": 8.8525e-05,
      "loss": 0.5731,
      "step": 17714
    },
    {
      "epoch": 0.9919923843655505,
      "grad_norm": 1.6043784618377686,
      "learning_rate": 8.853000000000001e-05,
      "loss": 0.4541,
      "step": 17715
    },
    {
      "epoch": 0.9920483816776795,
      "grad_norm": 1.2234134674072266,
      "learning_rate": 8.8535e-05,
      "loss": 0.4573,
      "step": 17716
    },
    {
      "epoch": 0.9921043789898085,
      "grad_norm": 1.176305890083313,
      "learning_rate": 8.854e-05,
      "loss": 0.3466,
      "step": 17717
    },
    {
      "epoch": 0.9921603763019375,
      "grad_norm": 1.4414507150650024,
      "learning_rate": 8.8545e-05,
      "loss": 0.4582,
      "step": 17718
    },
    {
      "epoch": 0.9922163736140666,
      "grad_norm": 1.346954345703125,
      "learning_rate": 8.855e-05,
      "loss": 0.4941,
      "step": 17719
    },
    {
      "epoch": 0.9922723709261956,
      "grad_norm": 1.308688998222351,
      "learning_rate": 8.8555e-05,
      "loss": 0.5205,
      "step": 17720
    },
    {
      "epoch": 0.9923283682383246,
      "grad_norm": 1.1533242464065552,
      "learning_rate": 8.856e-05,
      "loss": 0.5297,
      "step": 17721
    },
    {
      "epoch": 0.9923843655504536,
      "grad_norm": 1.1984797716140747,
      "learning_rate": 8.8565e-05,
      "loss": 0.43,
      "step": 17722
    },
    {
      "epoch": 0.9924403628625826,
      "grad_norm": 1.5509682893753052,
      "learning_rate": 8.857000000000001e-05,
      "loss": 0.4958,
      "step": 17723
    },
    {
      "epoch": 0.9924963601747117,
      "grad_norm": 1.3433796167373657,
      "learning_rate": 8.857500000000001e-05,
      "loss": 0.3858,
      "step": 17724
    },
    {
      "epoch": 0.9925523574868407,
      "grad_norm": 1.4598404169082642,
      "learning_rate": 8.858000000000001e-05,
      "loss": 0.4344,
      "step": 17725
    },
    {
      "epoch": 0.9926083547989697,
      "grad_norm": 1.2984787225723267,
      "learning_rate": 8.8585e-05,
      "loss": 0.4054,
      "step": 17726
    },
    {
      "epoch": 0.9926643521110987,
      "grad_norm": 1.8481416702270508,
      "learning_rate": 8.859e-05,
      "loss": 0.3646,
      "step": 17727
    },
    {
      "epoch": 0.9927203494232277,
      "grad_norm": 1.2608082294464111,
      "learning_rate": 8.8595e-05,
      "loss": 0.3594,
      "step": 17728
    },
    {
      "epoch": 0.9927763467353568,
      "grad_norm": 1.4416066408157349,
      "learning_rate": 8.86e-05,
      "loss": 0.5569,
      "step": 17729
    },
    {
      "epoch": 0.9928323440474858,
      "grad_norm": 1.2480077743530273,
      "learning_rate": 8.860500000000001e-05,
      "loss": 0.3737,
      "step": 17730
    },
    {
      "epoch": 0.9928883413596148,
      "grad_norm": 1.1432256698608398,
      "learning_rate": 8.861000000000001e-05,
      "loss": 0.4381,
      "step": 17731
    },
    {
      "epoch": 0.9929443386717438,
      "grad_norm": 1.6371777057647705,
      "learning_rate": 8.8615e-05,
      "loss": 0.5299,
      "step": 17732
    },
    {
      "epoch": 0.9930003359838728,
      "grad_norm": 1.286447525024414,
      "learning_rate": 8.862e-05,
      "loss": 0.4627,
      "step": 17733
    },
    {
      "epoch": 0.9930563332960017,
      "grad_norm": 1.3441230058670044,
      "learning_rate": 8.8625e-05,
      "loss": 0.38,
      "step": 17734
    },
    {
      "epoch": 0.9931123306081308,
      "grad_norm": 1.1864452362060547,
      "learning_rate": 8.863e-05,
      "loss": 0.4335,
      "step": 17735
    },
    {
      "epoch": 0.9931683279202598,
      "grad_norm": 1.2555714845657349,
      "learning_rate": 8.863500000000001e-05,
      "loss": 0.4425,
      "step": 17736
    },
    {
      "epoch": 0.9932243252323888,
      "grad_norm": 1.1101124286651611,
      "learning_rate": 8.864e-05,
      "loss": 0.3574,
      "step": 17737
    },
    {
      "epoch": 0.9932803225445178,
      "grad_norm": 1.2899377346038818,
      "learning_rate": 8.8645e-05,
      "loss": 0.3888,
      "step": 17738
    },
    {
      "epoch": 0.9933363198566468,
      "grad_norm": 1.324385643005371,
      "learning_rate": 8.865e-05,
      "loss": 0.4137,
      "step": 17739
    },
    {
      "epoch": 0.9933923171687759,
      "grad_norm": 1.7079662084579468,
      "learning_rate": 8.8655e-05,
      "loss": 0.6243,
      "step": 17740
    },
    {
      "epoch": 0.9934483144809049,
      "grad_norm": 1.450404405593872,
      "learning_rate": 8.866000000000001e-05,
      "loss": 0.5835,
      "step": 17741
    },
    {
      "epoch": 0.9935043117930339,
      "grad_norm": 1.3371881246566772,
      "learning_rate": 8.8665e-05,
      "loss": 0.4234,
      "step": 17742
    },
    {
      "epoch": 0.9935603091051629,
      "grad_norm": 1.562181830406189,
      "learning_rate": 8.867e-05,
      "loss": 0.4009,
      "step": 17743
    },
    {
      "epoch": 0.9936163064172919,
      "grad_norm": 1.3748550415039062,
      "learning_rate": 8.867500000000001e-05,
      "loss": 0.4405,
      "step": 17744
    },
    {
      "epoch": 0.993672303729421,
      "grad_norm": 1.4513928890228271,
      "learning_rate": 8.868000000000001e-05,
      "loss": 0.425,
      "step": 17745
    },
    {
      "epoch": 0.99372830104155,
      "grad_norm": 1.3562967777252197,
      "learning_rate": 8.868500000000001e-05,
      "loss": 0.3928,
      "step": 17746
    },
    {
      "epoch": 0.993784298353679,
      "grad_norm": 1.7254388332366943,
      "learning_rate": 8.869e-05,
      "loss": 0.502,
      "step": 17747
    },
    {
      "epoch": 0.993840295665808,
      "grad_norm": 1.351060152053833,
      "learning_rate": 8.8695e-05,
      "loss": 0.3679,
      "step": 17748
    },
    {
      "epoch": 0.993896292977937,
      "grad_norm": 1.322018027305603,
      "learning_rate": 8.87e-05,
      "loss": 0.461,
      "step": 17749
    },
    {
      "epoch": 0.993952290290066,
      "grad_norm": 1.4136666059494019,
      "learning_rate": 8.8705e-05,
      "loss": 0.4954,
      "step": 17750
    },
    {
      "epoch": 0.9940082876021951,
      "grad_norm": 1.2768642902374268,
      "learning_rate": 8.871000000000001e-05,
      "loss": 0.4126,
      "step": 17751
    },
    {
      "epoch": 0.9940642849143241,
      "grad_norm": 1.3570771217346191,
      "learning_rate": 8.871500000000001e-05,
      "loss": 0.4779,
      "step": 17752
    },
    {
      "epoch": 0.9941202822264531,
      "grad_norm": 1.1728743314743042,
      "learning_rate": 8.872e-05,
      "loss": 0.5258,
      "step": 17753
    },
    {
      "epoch": 0.9941762795385821,
      "grad_norm": 1.4993699789047241,
      "learning_rate": 8.8725e-05,
      "loss": 0.4644,
      "step": 17754
    },
    {
      "epoch": 0.9942322768507111,
      "grad_norm": 1.169426441192627,
      "learning_rate": 8.873e-05,
      "loss": 0.4376,
      "step": 17755
    },
    {
      "epoch": 0.9942882741628402,
      "grad_norm": 1.6827619075775146,
      "learning_rate": 8.8735e-05,
      "loss": 0.5448,
      "step": 17756
    },
    {
      "epoch": 0.9943442714749692,
      "grad_norm": 1.4702149629592896,
      "learning_rate": 8.874000000000001e-05,
      "loss": 0.4164,
      "step": 17757
    },
    {
      "epoch": 0.9944002687870982,
      "grad_norm": 1.4925825595855713,
      "learning_rate": 8.8745e-05,
      "loss": 0.4294,
      "step": 17758
    },
    {
      "epoch": 0.9944562660992272,
      "grad_norm": 1.1866360902786255,
      "learning_rate": 8.875e-05,
      "loss": 0.3493,
      "step": 17759
    },
    {
      "epoch": 0.9945122634113562,
      "grad_norm": 1.4568305015563965,
      "learning_rate": 8.8755e-05,
      "loss": 0.4478,
      "step": 17760
    },
    {
      "epoch": 0.9945682607234853,
      "grad_norm": 1.296949863433838,
      "learning_rate": 8.876e-05,
      "loss": 0.3953,
      "step": 17761
    },
    {
      "epoch": 0.9946242580356143,
      "grad_norm": 1.423387050628662,
      "learning_rate": 8.876500000000001e-05,
      "loss": 0.5378,
      "step": 17762
    },
    {
      "epoch": 0.9946802553477433,
      "grad_norm": 1.4243385791778564,
      "learning_rate": 8.877e-05,
      "loss": 0.4648,
      "step": 17763
    },
    {
      "epoch": 0.9947362526598723,
      "grad_norm": 1.4275507926940918,
      "learning_rate": 8.8775e-05,
      "loss": 0.5134,
      "step": 17764
    },
    {
      "epoch": 0.9947922499720013,
      "grad_norm": 1.2002226114273071,
      "learning_rate": 8.878000000000001e-05,
      "loss": 0.3779,
      "step": 17765
    },
    {
      "epoch": 0.9948482472841303,
      "grad_norm": 1.3369922637939453,
      "learning_rate": 8.878500000000001e-05,
      "loss": 0.4891,
      "step": 17766
    },
    {
      "epoch": 0.9949042445962594,
      "grad_norm": 2.1659765243530273,
      "learning_rate": 8.879000000000001e-05,
      "loss": 0.4689,
      "step": 17767
    },
    {
      "epoch": 0.9949602419083884,
      "grad_norm": 1.449819803237915,
      "learning_rate": 8.8795e-05,
      "loss": 0.4796,
      "step": 17768
    },
    {
      "epoch": 0.9950162392205174,
      "grad_norm": 1.4384304285049438,
      "learning_rate": 8.88e-05,
      "loss": 0.5106,
      "step": 17769
    },
    {
      "epoch": 0.9950722365326464,
      "grad_norm": 1.7336833477020264,
      "learning_rate": 8.8805e-05,
      "loss": 0.5333,
      "step": 17770
    },
    {
      "epoch": 0.9951282338447754,
      "grad_norm": 1.467752456665039,
      "learning_rate": 8.881e-05,
      "loss": 0.5923,
      "step": 17771
    },
    {
      "epoch": 0.9951842311569045,
      "grad_norm": 1.5609163045883179,
      "learning_rate": 8.881500000000001e-05,
      "loss": 0.5225,
      "step": 17772
    },
    {
      "epoch": 0.9952402284690335,
      "grad_norm": 1.3803917169570923,
      "learning_rate": 8.882000000000001e-05,
      "loss": 0.4381,
      "step": 17773
    },
    {
      "epoch": 0.9952962257811625,
      "grad_norm": 1.325348138809204,
      "learning_rate": 8.8825e-05,
      "loss": 0.5434,
      "step": 17774
    },
    {
      "epoch": 0.9953522230932915,
      "grad_norm": 1.1791011095046997,
      "learning_rate": 8.883e-05,
      "loss": 0.3529,
      "step": 17775
    },
    {
      "epoch": 0.9954082204054205,
      "grad_norm": 1.1796578168869019,
      "learning_rate": 8.8835e-05,
      "loss": 0.4318,
      "step": 17776
    },
    {
      "epoch": 0.9954642177175496,
      "grad_norm": 1.7437174320220947,
      "learning_rate": 8.884e-05,
      "loss": 0.462,
      "step": 17777
    },
    {
      "epoch": 0.9955202150296786,
      "grad_norm": 1.3527717590332031,
      "learning_rate": 8.8845e-05,
      "loss": 0.5139,
      "step": 17778
    },
    {
      "epoch": 0.9955762123418076,
      "grad_norm": 1.0909597873687744,
      "learning_rate": 8.885e-05,
      "loss": 0.3406,
      "step": 17779
    },
    {
      "epoch": 0.9956322096539366,
      "grad_norm": 15.612887382507324,
      "learning_rate": 8.8855e-05,
      "loss": 0.5652,
      "step": 17780
    },
    {
      "epoch": 0.9956882069660656,
      "grad_norm": 1.3589320182800293,
      "learning_rate": 8.886e-05,
      "loss": 0.4209,
      "step": 17781
    },
    {
      "epoch": 0.9957442042781947,
      "grad_norm": 1.6181668043136597,
      "learning_rate": 8.886500000000001e-05,
      "loss": 0.4392,
      "step": 17782
    },
    {
      "epoch": 0.9958002015903237,
      "grad_norm": 1.5727357864379883,
      "learning_rate": 8.887000000000001e-05,
      "loss": 0.5498,
      "step": 17783
    },
    {
      "epoch": 0.9958561989024527,
      "grad_norm": 1.449902892112732,
      "learning_rate": 8.8875e-05,
      "loss": 0.3859,
      "step": 17784
    },
    {
      "epoch": 0.9959121962145817,
      "grad_norm": 1.5982789993286133,
      "learning_rate": 8.888e-05,
      "loss": 0.4078,
      "step": 17785
    },
    {
      "epoch": 0.9959681935267107,
      "grad_norm": 1.2767308950424194,
      "learning_rate": 8.888500000000001e-05,
      "loss": 0.4659,
      "step": 17786
    },
    {
      "epoch": 0.9960241908388398,
      "grad_norm": 1.3726425170898438,
      "learning_rate": 8.889000000000001e-05,
      "loss": 0.4481,
      "step": 17787
    },
    {
      "epoch": 0.9960801881509688,
      "grad_norm": 1.338931918144226,
      "learning_rate": 8.889500000000001e-05,
      "loss": 0.4569,
      "step": 17788
    },
    {
      "epoch": 0.9961361854630978,
      "grad_norm": 1.3055534362792969,
      "learning_rate": 8.89e-05,
      "loss": 0.3763,
      "step": 17789
    },
    {
      "epoch": 0.9961921827752268,
      "grad_norm": 1.2787420749664307,
      "learning_rate": 8.8905e-05,
      "loss": 0.4308,
      "step": 17790
    },
    {
      "epoch": 0.9962481800873558,
      "grad_norm": 1.5374587774276733,
      "learning_rate": 8.891e-05,
      "loss": 0.5375,
      "step": 17791
    },
    {
      "epoch": 0.9963041773994848,
      "grad_norm": 1.3534173965454102,
      "learning_rate": 8.8915e-05,
      "loss": 0.4976,
      "step": 17792
    },
    {
      "epoch": 0.9963601747116139,
      "grad_norm": 1.5223244428634644,
      "learning_rate": 8.892000000000001e-05,
      "loss": 0.4936,
      "step": 17793
    },
    {
      "epoch": 0.9964161720237429,
      "grad_norm": 1.176383137702942,
      "learning_rate": 8.8925e-05,
      "loss": 0.373,
      "step": 17794
    },
    {
      "epoch": 0.9964721693358719,
      "grad_norm": 1.2590463161468506,
      "learning_rate": 8.893e-05,
      "loss": 0.4684,
      "step": 17795
    },
    {
      "epoch": 0.9965281666480009,
      "grad_norm": 1.581672191619873,
      "learning_rate": 8.8935e-05,
      "loss": 0.5669,
      "step": 17796
    },
    {
      "epoch": 0.9965841639601299,
      "grad_norm": 2.492479085922241,
      "learning_rate": 8.894e-05,
      "loss": 0.5244,
      "step": 17797
    },
    {
      "epoch": 0.996640161272259,
      "grad_norm": 1.3283013105392456,
      "learning_rate": 8.8945e-05,
      "loss": 0.55,
      "step": 17798
    },
    {
      "epoch": 0.996696158584388,
      "grad_norm": 1.2206213474273682,
      "learning_rate": 8.895e-05,
      "loss": 0.3917,
      "step": 17799
    },
    {
      "epoch": 0.996752155896517,
      "grad_norm": 1.5171903371810913,
      "learning_rate": 8.8955e-05,
      "loss": 0.5488,
      "step": 17800
    },
    {
      "epoch": 0.996808153208646,
      "grad_norm": 1.507001280784607,
      "learning_rate": 8.896e-05,
      "loss": 0.5199,
      "step": 17801
    },
    {
      "epoch": 0.996864150520775,
      "grad_norm": 1.26813542842865,
      "learning_rate": 8.896500000000001e-05,
      "loss": 0.3859,
      "step": 17802
    },
    {
      "epoch": 0.9969201478329041,
      "grad_norm": 2.1569724082946777,
      "learning_rate": 8.897000000000001e-05,
      "loss": 0.5384,
      "step": 17803
    },
    {
      "epoch": 0.9969761451450331,
      "grad_norm": 2.1141278743743896,
      "learning_rate": 8.897500000000001e-05,
      "loss": 0.4599,
      "step": 17804
    },
    {
      "epoch": 0.9970321424571621,
      "grad_norm": 1.3248432874679565,
      "learning_rate": 8.898e-05,
      "loss": 0.4108,
      "step": 17805
    },
    {
      "epoch": 0.9970881397692911,
      "grad_norm": 1.2197864055633545,
      "learning_rate": 8.8985e-05,
      "loss": 0.4157,
      "step": 17806
    },
    {
      "epoch": 0.9971441370814201,
      "grad_norm": 1.5786558389663696,
      "learning_rate": 8.899e-05,
      "loss": 0.5085,
      "step": 17807
    },
    {
      "epoch": 0.9972001343935492,
      "grad_norm": 1.2038183212280273,
      "learning_rate": 8.899500000000001e-05,
      "loss": 0.3443,
      "step": 17808
    },
    {
      "epoch": 0.9972561317056782,
      "grad_norm": 1.4003777503967285,
      "learning_rate": 8.900000000000001e-05,
      "loss": 0.4955,
      "step": 17809
    },
    {
      "epoch": 0.9973121290178072,
      "grad_norm": 1.3404408693313599,
      "learning_rate": 8.9005e-05,
      "loss": 0.424,
      "step": 17810
    },
    {
      "epoch": 0.9973681263299362,
      "grad_norm": 1.5213918685913086,
      "learning_rate": 8.901e-05,
      "loss": 0.4937,
      "step": 17811
    },
    {
      "epoch": 0.9974241236420652,
      "grad_norm": 1.4018992185592651,
      "learning_rate": 8.9015e-05,
      "loss": 0.4621,
      "step": 17812
    },
    {
      "epoch": 0.9974801209541942,
      "grad_norm": 1.1799334287643433,
      "learning_rate": 8.902e-05,
      "loss": 0.3614,
      "step": 17813
    },
    {
      "epoch": 0.9975361182663233,
      "grad_norm": 1.24405038356781,
      "learning_rate": 8.902500000000001e-05,
      "loss": 0.4992,
      "step": 17814
    },
    {
      "epoch": 0.9975921155784523,
      "grad_norm": 1.2421824932098389,
      "learning_rate": 8.903e-05,
      "loss": 0.3443,
      "step": 17815
    },
    {
      "epoch": 0.9976481128905813,
      "grad_norm": 1.294106125831604,
      "learning_rate": 8.9035e-05,
      "loss": 0.5638,
      "step": 17816
    },
    {
      "epoch": 0.9977041102027102,
      "grad_norm": 1.2034837007522583,
      "learning_rate": 8.904e-05,
      "loss": 0.4029,
      "step": 17817
    },
    {
      "epoch": 0.9977601075148392,
      "grad_norm": 1.393774151802063,
      "learning_rate": 8.9045e-05,
      "loss": 0.5215,
      "step": 17818
    },
    {
      "epoch": 0.9978161048269683,
      "grad_norm": 5.649298667907715,
      "learning_rate": 8.905e-05,
      "loss": 0.4517,
      "step": 17819
    },
    {
      "epoch": 0.9978721021390973,
      "grad_norm": 1.5890209674835205,
      "learning_rate": 8.9055e-05,
      "loss": 0.5306,
      "step": 17820
    },
    {
      "epoch": 0.9979280994512263,
      "grad_norm": 1.5135754346847534,
      "learning_rate": 8.906e-05,
      "loss": 0.4627,
      "step": 17821
    },
    {
      "epoch": 0.9979840967633553,
      "grad_norm": 1.3312042951583862,
      "learning_rate": 8.906500000000002e-05,
      "loss": 0.4954,
      "step": 17822
    },
    {
      "epoch": 0.9980400940754843,
      "grad_norm": 1.1777615547180176,
      "learning_rate": 8.907000000000001e-05,
      "loss": 0.3609,
      "step": 17823
    },
    {
      "epoch": 0.9980960913876133,
      "grad_norm": 1.5533791780471802,
      "learning_rate": 8.907500000000001e-05,
      "loss": 0.4959,
      "step": 17824
    },
    {
      "epoch": 0.9981520886997424,
      "grad_norm": 1.2568227052688599,
      "learning_rate": 8.908000000000001e-05,
      "loss": 0.4182,
      "step": 17825
    },
    {
      "epoch": 0.9982080860118714,
      "grad_norm": 1.339856743812561,
      "learning_rate": 8.9085e-05,
      "loss": 0.4881,
      "step": 17826
    },
    {
      "epoch": 0.9982640833240004,
      "grad_norm": 1.3573658466339111,
      "learning_rate": 8.909e-05,
      "loss": 0.5631,
      "step": 17827
    },
    {
      "epoch": 0.9983200806361294,
      "grad_norm": 1.392850637435913,
      "learning_rate": 8.9095e-05,
      "loss": 0.3904,
      "step": 17828
    },
    {
      "epoch": 0.9983760779482584,
      "grad_norm": 1.140561580657959,
      "learning_rate": 8.910000000000001e-05,
      "loss": 0.3946,
      "step": 17829
    },
    {
      "epoch": 0.9984320752603875,
      "grad_norm": 1.7466707229614258,
      "learning_rate": 8.910500000000001e-05,
      "loss": 0.6854,
      "step": 17830
    },
    {
      "epoch": 0.9984880725725165,
      "grad_norm": 1.1528984308242798,
      "learning_rate": 8.911e-05,
      "loss": 0.4211,
      "step": 17831
    },
    {
      "epoch": 0.9985440698846455,
      "grad_norm": 1.7609009742736816,
      "learning_rate": 8.9115e-05,
      "loss": 0.4003,
      "step": 17832
    },
    {
      "epoch": 0.9986000671967745,
      "grad_norm": 1.2808367013931274,
      "learning_rate": 8.912e-05,
      "loss": 0.4723,
      "step": 17833
    },
    {
      "epoch": 0.9986560645089035,
      "grad_norm": 1.5797677040100098,
      "learning_rate": 8.9125e-05,
      "loss": 0.4325,
      "step": 17834
    },
    {
      "epoch": 0.9987120618210326,
      "grad_norm": 1.264971375465393,
      "learning_rate": 8.913000000000001e-05,
      "loss": 0.4001,
      "step": 17835
    },
    {
      "epoch": 0.9987680591331616,
      "grad_norm": 1.4042302370071411,
      "learning_rate": 8.9135e-05,
      "loss": 0.3977,
      "step": 17836
    },
    {
      "epoch": 0.9988240564452906,
      "grad_norm": 1.3492649793624878,
      "learning_rate": 8.914e-05,
      "loss": 0.4877,
      "step": 17837
    },
    {
      "epoch": 0.9988800537574196,
      "grad_norm": 1.1757004261016846,
      "learning_rate": 8.9145e-05,
      "loss": 0.4391,
      "step": 17838
    },
    {
      "epoch": 0.9989360510695486,
      "grad_norm": 1.409857153892517,
      "learning_rate": 8.915e-05,
      "loss": 0.4832,
      "step": 17839
    },
    {
      "epoch": 0.9989920483816777,
      "grad_norm": 1.3181517124176025,
      "learning_rate": 8.9155e-05,
      "loss": 0.3506,
      "step": 17840
    },
    {
      "epoch": 0.9990480456938067,
      "grad_norm": 1.33633553981781,
      "learning_rate": 8.916e-05,
      "loss": 0.5155,
      "step": 17841
    },
    {
      "epoch": 0.9991040430059357,
      "grad_norm": 1.723892092704773,
      "learning_rate": 8.9165e-05,
      "loss": 0.5702,
      "step": 17842
    },
    {
      "epoch": 0.9991600403180647,
      "grad_norm": 1.2972443103790283,
      "learning_rate": 8.917000000000002e-05,
      "loss": 0.4963,
      "step": 17843
    },
    {
      "epoch": 0.9992160376301937,
      "grad_norm": 1.8608943223953247,
      "learning_rate": 8.917500000000001e-05,
      "loss": 0.5375,
      "step": 17844
    },
    {
      "epoch": 0.9992720349423228,
      "grad_norm": 1.3446626663208008,
      "learning_rate": 8.918000000000001e-05,
      "loss": 0.5504,
      "step": 17845
    },
    {
      "epoch": 0.9993280322544518,
      "grad_norm": 1.269340991973877,
      "learning_rate": 8.918500000000001e-05,
      "loss": 0.4592,
      "step": 17846
    },
    {
      "epoch": 0.9993840295665808,
      "grad_norm": 1.349295735359192,
      "learning_rate": 8.919e-05,
      "loss": 0.4749,
      "step": 17847
    },
    {
      "epoch": 0.9994400268787098,
      "grad_norm": 1.5455998182296753,
      "learning_rate": 8.9195e-05,
      "loss": 0.5692,
      "step": 17848
    },
    {
      "epoch": 0.9994960241908388,
      "grad_norm": 1.5913318395614624,
      "learning_rate": 8.92e-05,
      "loss": 0.3864,
      "step": 17849
    },
    {
      "epoch": 0.9995520215029678,
      "grad_norm": 1.3120813369750977,
      "learning_rate": 8.920500000000001e-05,
      "loss": 0.4912,
      "step": 17850
    },
    {
      "epoch": 0.9996080188150969,
      "grad_norm": 1.6308808326721191,
      "learning_rate": 8.921000000000001e-05,
      "loss": 0.5462,
      "step": 17851
    },
    {
      "epoch": 0.9996640161272259,
      "grad_norm": 1.414174199104309,
      "learning_rate": 8.9215e-05,
      "loss": 0.5364,
      "step": 17852
    },
    {
      "epoch": 0.9997200134393549,
      "grad_norm": 1.3153877258300781,
      "learning_rate": 8.922e-05,
      "loss": 0.4214,
      "step": 17853
    },
    {
      "epoch": 0.9997760107514839,
      "grad_norm": 1.3772268295288086,
      "learning_rate": 8.9225e-05,
      "loss": 0.5077,
      "step": 17854
    },
    {
      "epoch": 0.9998320080636129,
      "grad_norm": 1.1395400762557983,
      "learning_rate": 8.923e-05,
      "loss": 0.4192,
      "step": 17855
    },
    {
      "epoch": 0.999888005375742,
      "grad_norm": 1.4553617238998413,
      "learning_rate": 8.9235e-05,
      "loss": 0.5172,
      "step": 17856
    },
    {
      "epoch": 0.999944002687871,
      "grad_norm": 1.2411185503005981,
      "learning_rate": 8.924e-05,
      "loss": 0.3903,
      "step": 17857
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.379507541656494,
      "learning_rate": 8.9245e-05,
      "loss": 0.2554,
      "step": 17858
    },
    {
      "epoch": 1.000055997312129,
      "grad_norm": 1.1951563358306885,
      "learning_rate": 8.925e-05,
      "loss": 0.3561,
      "step": 17859
    },
    {
      "epoch": 1.000111994624258,
      "grad_norm": 1.2772942781448364,
      "learning_rate": 8.9255e-05,
      "loss": 0.4737,
      "step": 17860
    },
    {
      "epoch": 1.000167991936387,
      "grad_norm": 1.2272050380706787,
      "learning_rate": 8.926e-05,
      "loss": 0.3712,
      "step": 17861
    },
    {
      "epoch": 1.000223989248516,
      "grad_norm": 1.2424622774124146,
      "learning_rate": 8.9265e-05,
      "loss": 0.44,
      "step": 17862
    },
    {
      "epoch": 1.000279986560645,
      "grad_norm": 1.1765683889389038,
      "learning_rate": 8.927e-05,
      "loss": 0.4108,
      "step": 17863
    },
    {
      "epoch": 1.0003359838727741,
      "grad_norm": 1.3974347114562988,
      "learning_rate": 8.927500000000002e-05,
      "loss": 0.4063,
      "step": 17864
    },
    {
      "epoch": 1.0003919811849031,
      "grad_norm": 1.3876872062683105,
      "learning_rate": 8.928000000000001e-05,
      "loss": 0.3698,
      "step": 17865
    },
    {
      "epoch": 1.0004479784970322,
      "grad_norm": 1.5575116872787476,
      "learning_rate": 8.928500000000001e-05,
      "loss": 0.3858,
      "step": 17866
    },
    {
      "epoch": 1.0005039758091612,
      "grad_norm": 1.4825502634048462,
      "learning_rate": 8.929000000000001e-05,
      "loss": 0.4251,
      "step": 17867
    },
    {
      "epoch": 1.0005599731212902,
      "grad_norm": 2.1418943405151367,
      "learning_rate": 8.9295e-05,
      "loss": 0.4481,
      "step": 17868
    },
    {
      "epoch": 1.0006159704334192,
      "grad_norm": 1.5139989852905273,
      "learning_rate": 8.93e-05,
      "loss": 0.4508,
      "step": 17869
    },
    {
      "epoch": 1.0006719677455482,
      "grad_norm": 1.6377043724060059,
      "learning_rate": 8.9305e-05,
      "loss": 0.42,
      "step": 17870
    },
    {
      "epoch": 1.0007279650576772,
      "grad_norm": 1.472362995147705,
      "learning_rate": 8.931000000000001e-05,
      "loss": 0.5715,
      "step": 17871
    },
    {
      "epoch": 1.0007839623698063,
      "grad_norm": 1.52193284034729,
      "learning_rate": 8.931500000000001e-05,
      "loss": 0.517,
      "step": 17872
    },
    {
      "epoch": 1.0008399596819353,
      "grad_norm": 2.265660047531128,
      "learning_rate": 8.932e-05,
      "loss": 0.4281,
      "step": 17873
    },
    {
      "epoch": 1.0008959569940643,
      "grad_norm": 1.5788400173187256,
      "learning_rate": 8.9325e-05,
      "loss": 0.4571,
      "step": 17874
    },
    {
      "epoch": 1.0009519543061933,
      "grad_norm": 1.347461223602295,
      "learning_rate": 8.933e-05,
      "loss": 0.3671,
      "step": 17875
    },
    {
      "epoch": 1.0010079516183223,
      "grad_norm": 1.3150992393493652,
      "learning_rate": 8.9335e-05,
      "loss": 0.4848,
      "step": 17876
    },
    {
      "epoch": 1.0010639489304514,
      "grad_norm": 1.1444026231765747,
      "learning_rate": 8.934e-05,
      "loss": 0.3704,
      "step": 17877
    },
    {
      "epoch": 1.0011199462425804,
      "grad_norm": 1.5898854732513428,
      "learning_rate": 8.9345e-05,
      "loss": 0.4398,
      "step": 17878
    },
    {
      "epoch": 1.0011759435547094,
      "grad_norm": 1.4028594493865967,
      "learning_rate": 8.935e-05,
      "loss": 0.3731,
      "step": 17879
    },
    {
      "epoch": 1.0012319408668384,
      "grad_norm": 1.3735319375991821,
      "learning_rate": 8.9355e-05,
      "loss": 0.4651,
      "step": 17880
    },
    {
      "epoch": 1.0012879381789674,
      "grad_norm": 1.1216611862182617,
      "learning_rate": 8.936e-05,
      "loss": 0.4876,
      "step": 17881
    },
    {
      "epoch": 1.0013439354910965,
      "grad_norm": 1.6765284538269043,
      "learning_rate": 8.936500000000001e-05,
      "loss": 0.6061,
      "step": 17882
    },
    {
      "epoch": 1.0013999328032255,
      "grad_norm": 1.298638939857483,
      "learning_rate": 8.937e-05,
      "loss": 0.3629,
      "step": 17883
    },
    {
      "epoch": 1.0014559301153545,
      "grad_norm": 1.259941577911377,
      "learning_rate": 8.9375e-05,
      "loss": 0.2662,
      "step": 17884
    },
    {
      "epoch": 1.0015119274274835,
      "grad_norm": 1.2943047285079956,
      "learning_rate": 8.938e-05,
      "loss": 0.4017,
      "step": 17885
    },
    {
      "epoch": 1.0015679247396125,
      "grad_norm": 1.2645395994186401,
      "learning_rate": 8.938500000000001e-05,
      "loss": 0.4045,
      "step": 17886
    },
    {
      "epoch": 1.0016239220517416,
      "grad_norm": 1.259494662284851,
      "learning_rate": 8.939000000000001e-05,
      "loss": 0.3124,
      "step": 17887
    },
    {
      "epoch": 1.0016799193638706,
      "grad_norm": 1.5386265516281128,
      "learning_rate": 8.939500000000001e-05,
      "loss": 0.5339,
      "step": 17888
    },
    {
      "epoch": 1.0017359166759996,
      "grad_norm": 2.078693389892578,
      "learning_rate": 8.94e-05,
      "loss": 0.414,
      "step": 17889
    },
    {
      "epoch": 1.0017919139881286,
      "grad_norm": 1.1553360223770142,
      "learning_rate": 8.9405e-05,
      "loss": 0.478,
      "step": 17890
    },
    {
      "epoch": 1.0018479113002576,
      "grad_norm": 1.3309661149978638,
      "learning_rate": 8.941e-05,
      "loss": 0.4211,
      "step": 17891
    },
    {
      "epoch": 1.0019039086123867,
      "grad_norm": 1.216221809387207,
      "learning_rate": 8.941500000000001e-05,
      "loss": 0.5105,
      "step": 17892
    },
    {
      "epoch": 1.0019599059245157,
      "grad_norm": 1.3124853372573853,
      "learning_rate": 8.942000000000001e-05,
      "loss": 0.5729,
      "step": 17893
    },
    {
      "epoch": 1.0020159032366447,
      "grad_norm": 1.3122268915176392,
      "learning_rate": 8.9425e-05,
      "loss": 0.4651,
      "step": 17894
    },
    {
      "epoch": 1.0020719005487737,
      "grad_norm": 1.1882063150405884,
      "learning_rate": 8.943e-05,
      "loss": 0.3396,
      "step": 17895
    },
    {
      "epoch": 1.0021278978609027,
      "grad_norm": 1.3756940364837646,
      "learning_rate": 8.9435e-05,
      "loss": 0.3477,
      "step": 17896
    },
    {
      "epoch": 1.0021838951730317,
      "grad_norm": 1.2090256214141846,
      "learning_rate": 8.944e-05,
      "loss": 0.4787,
      "step": 17897
    },
    {
      "epoch": 1.0022398924851608,
      "grad_norm": 1.4074785709381104,
      "learning_rate": 8.9445e-05,
      "loss": 0.5455,
      "step": 17898
    },
    {
      "epoch": 1.0022958897972898,
      "grad_norm": 1.383531928062439,
      "learning_rate": 8.945e-05,
      "loss": 0.4804,
      "step": 17899
    },
    {
      "epoch": 1.0023518871094188,
      "grad_norm": 1.4040472507476807,
      "learning_rate": 8.9455e-05,
      "loss": 0.5287,
      "step": 17900
    },
    {
      "epoch": 1.0024078844215478,
      "grad_norm": 1.351027488708496,
      "learning_rate": 8.946e-05,
      "loss": 0.3817,
      "step": 17901
    },
    {
      "epoch": 1.0024638817336768,
      "grad_norm": 1.2312889099121094,
      "learning_rate": 8.9465e-05,
      "loss": 0.4973,
      "step": 17902
    },
    {
      "epoch": 1.0025198790458059,
      "grad_norm": 1.5979794263839722,
      "learning_rate": 8.947000000000001e-05,
      "loss": 0.5239,
      "step": 17903
    },
    {
      "epoch": 1.0025758763579349,
      "grad_norm": 1.2218527793884277,
      "learning_rate": 8.9475e-05,
      "loss": 0.5034,
      "step": 17904
    },
    {
      "epoch": 1.002631873670064,
      "grad_norm": 1.7417958974838257,
      "learning_rate": 8.948e-05,
      "loss": 0.5016,
      "step": 17905
    },
    {
      "epoch": 1.002687870982193,
      "grad_norm": 1.3673913478851318,
      "learning_rate": 8.9485e-05,
      "loss": 0.3737,
      "step": 17906
    },
    {
      "epoch": 1.002743868294322,
      "grad_norm": 1.341664433479309,
      "learning_rate": 8.949000000000001e-05,
      "loss": 0.4445,
      "step": 17907
    },
    {
      "epoch": 1.002799865606451,
      "grad_norm": 1.3660603761672974,
      "learning_rate": 8.949500000000001e-05,
      "loss": 0.4294,
      "step": 17908
    },
    {
      "epoch": 1.00285586291858,
      "grad_norm": 1.244755744934082,
      "learning_rate": 8.950000000000001e-05,
      "loss": 0.5295,
      "step": 17909
    },
    {
      "epoch": 1.002911860230709,
      "grad_norm": 1.594760775566101,
      "learning_rate": 8.9505e-05,
      "loss": 0.3796,
      "step": 17910
    },
    {
      "epoch": 1.002967857542838,
      "grad_norm": 1.0852961540222168,
      "learning_rate": 8.951e-05,
      "loss": 0.3523,
      "step": 17911
    },
    {
      "epoch": 1.003023854854967,
      "grad_norm": 1.42945396900177,
      "learning_rate": 8.9515e-05,
      "loss": 0.4033,
      "step": 17912
    },
    {
      "epoch": 1.003079852167096,
      "grad_norm": 1.4561173915863037,
      "learning_rate": 8.952000000000001e-05,
      "loss": 0.3425,
      "step": 17913
    },
    {
      "epoch": 1.003135849479225,
      "grad_norm": 1.454376459121704,
      "learning_rate": 8.952500000000001e-05,
      "loss": 0.368,
      "step": 17914
    },
    {
      "epoch": 1.003191846791354,
      "grad_norm": 1.29525887966156,
      "learning_rate": 8.953e-05,
      "loss": 0.3543,
      "step": 17915
    },
    {
      "epoch": 1.003247844103483,
      "grad_norm": 1.374119520187378,
      "learning_rate": 8.9535e-05,
      "loss": 0.486,
      "step": 17916
    },
    {
      "epoch": 1.0033038414156121,
      "grad_norm": 1.313292384147644,
      "learning_rate": 8.954e-05,
      "loss": 0.3136,
      "step": 17917
    },
    {
      "epoch": 1.0033598387277411,
      "grad_norm": 1.3198938369750977,
      "learning_rate": 8.9545e-05,
      "loss": 0.3961,
      "step": 17918
    },
    {
      "epoch": 1.0034158360398702,
      "grad_norm": 1.3486634492874146,
      "learning_rate": 8.955e-05,
      "loss": 0.3769,
      "step": 17919
    },
    {
      "epoch": 1.0034718333519992,
      "grad_norm": 1.2486048936843872,
      "learning_rate": 8.9555e-05,
      "loss": 0.3641,
      "step": 17920
    },
    {
      "epoch": 1.0035278306641282,
      "grad_norm": 1.2347756624221802,
      "learning_rate": 8.956e-05,
      "loss": 0.4007,
      "step": 17921
    },
    {
      "epoch": 1.0035838279762572,
      "grad_norm": 1.5285927057266235,
      "learning_rate": 8.9565e-05,
      "loss": 0.5858,
      "step": 17922
    },
    {
      "epoch": 1.0036398252883862,
      "grad_norm": 1.3821576833724976,
      "learning_rate": 8.957000000000001e-05,
      "loss": 0.3995,
      "step": 17923
    },
    {
      "epoch": 1.0036958226005153,
      "grad_norm": 1.405597448348999,
      "learning_rate": 8.957500000000001e-05,
      "loss": 0.423,
      "step": 17924
    },
    {
      "epoch": 1.0037518199126443,
      "grad_norm": 1.061468243598938,
      "learning_rate": 8.958e-05,
      "loss": 0.3636,
      "step": 17925
    },
    {
      "epoch": 1.0038078172247733,
      "grad_norm": 1.2281023263931274,
      "learning_rate": 8.9585e-05,
      "loss": 0.3992,
      "step": 17926
    },
    {
      "epoch": 1.0038638145369023,
      "grad_norm": 1.4296306371688843,
      "learning_rate": 8.959e-05,
      "loss": 0.4056,
      "step": 17927
    },
    {
      "epoch": 1.0039198118490313,
      "grad_norm": 1.3915297985076904,
      "learning_rate": 8.959500000000001e-05,
      "loss": 0.4944,
      "step": 17928
    },
    {
      "epoch": 1.0039758091611604,
      "grad_norm": 1.2303251028060913,
      "learning_rate": 8.960000000000001e-05,
      "loss": 0.4113,
      "step": 17929
    },
    {
      "epoch": 1.0040318064732894,
      "grad_norm": 1.4397661685943604,
      "learning_rate": 8.960500000000001e-05,
      "loss": 0.4788,
      "step": 17930
    },
    {
      "epoch": 1.0040878037854184,
      "grad_norm": 1.4920543432235718,
      "learning_rate": 8.961e-05,
      "loss": 0.4156,
      "step": 17931
    },
    {
      "epoch": 1.0041438010975474,
      "grad_norm": 1.8774982690811157,
      "learning_rate": 8.9615e-05,
      "loss": 0.4992,
      "step": 17932
    },
    {
      "epoch": 1.0041997984096764,
      "grad_norm": 1.4324315786361694,
      "learning_rate": 8.962e-05,
      "loss": 0.3682,
      "step": 17933
    },
    {
      "epoch": 1.0042557957218055,
      "grad_norm": 1.3314342498779297,
      "learning_rate": 8.962500000000001e-05,
      "loss": 0.4155,
      "step": 17934
    },
    {
      "epoch": 1.0043117930339345,
      "grad_norm": 1.4088034629821777,
      "learning_rate": 8.963000000000001e-05,
      "loss": 0.497,
      "step": 17935
    },
    {
      "epoch": 1.0043677903460635,
      "grad_norm": 1.3769229650497437,
      "learning_rate": 8.9635e-05,
      "loss": 0.4401,
      "step": 17936
    },
    {
      "epoch": 1.0044237876581925,
      "grad_norm": 1.2523490190505981,
      "learning_rate": 8.964e-05,
      "loss": 0.3812,
      "step": 17937
    },
    {
      "epoch": 1.0044797849703215,
      "grad_norm": 1.2642509937286377,
      "learning_rate": 8.9645e-05,
      "loss": 0.4266,
      "step": 17938
    },
    {
      "epoch": 1.0045357822824506,
      "grad_norm": 1.5512166023254395,
      "learning_rate": 8.965e-05,
      "loss": 0.4254,
      "step": 17939
    },
    {
      "epoch": 1.0045917795945796,
      "grad_norm": 1.2381703853607178,
      "learning_rate": 8.9655e-05,
      "loss": 0.3338,
      "step": 17940
    },
    {
      "epoch": 1.0046477769067084,
      "grad_norm": 1.2523268461227417,
      "learning_rate": 8.966e-05,
      "loss": 0.4377,
      "step": 17941
    },
    {
      "epoch": 1.0047037742188374,
      "grad_norm": 1.6331632137298584,
      "learning_rate": 8.9665e-05,
      "loss": 0.5569,
      "step": 17942
    },
    {
      "epoch": 1.0047597715309664,
      "grad_norm": 1.2084189653396606,
      "learning_rate": 8.967000000000001e-05,
      "loss": 0.5117,
      "step": 17943
    },
    {
      "epoch": 1.0048157688430954,
      "grad_norm": 1.5852941274642944,
      "learning_rate": 8.967500000000001e-05,
      "loss": 0.4931,
      "step": 17944
    },
    {
      "epoch": 1.0048717661552244,
      "grad_norm": 1.2750952243804932,
      "learning_rate": 8.968000000000001e-05,
      "loss": 0.4169,
      "step": 17945
    },
    {
      "epoch": 1.0049277634673535,
      "grad_norm": 1.368571162223816,
      "learning_rate": 8.9685e-05,
      "loss": 0.4283,
      "step": 17946
    },
    {
      "epoch": 1.0049837607794825,
      "grad_norm": 1.2909531593322754,
      "learning_rate": 8.969e-05,
      "loss": 0.4161,
      "step": 17947
    },
    {
      "epoch": 1.0050397580916115,
      "grad_norm": 1.4677278995513916,
      "learning_rate": 8.9695e-05,
      "loss": 0.3967,
      "step": 17948
    },
    {
      "epoch": 1.0050957554037405,
      "grad_norm": 1.840155839920044,
      "learning_rate": 8.970000000000001e-05,
      "loss": 0.4164,
      "step": 17949
    },
    {
      "epoch": 1.0051517527158695,
      "grad_norm": 1.129811406135559,
      "learning_rate": 8.970500000000001e-05,
      "loss": 0.3269,
      "step": 17950
    },
    {
      "epoch": 1.0052077500279986,
      "grad_norm": 1.4732576608657837,
      "learning_rate": 8.971e-05,
      "loss": 0.4069,
      "step": 17951
    },
    {
      "epoch": 1.0052637473401276,
      "grad_norm": 1.497945785522461,
      "learning_rate": 8.9715e-05,
      "loss": 0.4268,
      "step": 17952
    },
    {
      "epoch": 1.0053197446522566,
      "grad_norm": 1.3697718381881714,
      "learning_rate": 8.972e-05,
      "loss": 0.467,
      "step": 17953
    },
    {
      "epoch": 1.0053757419643856,
      "grad_norm": 1.3891284465789795,
      "learning_rate": 8.9725e-05,
      "loss": 0.3637,
      "step": 17954
    },
    {
      "epoch": 1.0054317392765146,
      "grad_norm": 1.3554348945617676,
      "learning_rate": 8.973e-05,
      "loss": 0.5177,
      "step": 17955
    },
    {
      "epoch": 1.0054877365886437,
      "grad_norm": 1.5390160083770752,
      "learning_rate": 8.973500000000001e-05,
      "loss": 0.4884,
      "step": 17956
    },
    {
      "epoch": 1.0055437339007727,
      "grad_norm": 1.1982821226119995,
      "learning_rate": 8.974e-05,
      "loss": 0.4356,
      "step": 17957
    },
    {
      "epoch": 1.0055997312129017,
      "grad_norm": 1.4162977933883667,
      "learning_rate": 8.9745e-05,
      "loss": 0.4273,
      "step": 17958
    },
    {
      "epoch": 1.0056557285250307,
      "grad_norm": 1.566547155380249,
      "learning_rate": 8.975e-05,
      "loss": 0.5884,
      "step": 17959
    },
    {
      "epoch": 1.0057117258371597,
      "grad_norm": 1.2957085371017456,
      "learning_rate": 8.9755e-05,
      "loss": 0.3804,
      "step": 17960
    },
    {
      "epoch": 1.0057677231492888,
      "grad_norm": 1.5151442289352417,
      "learning_rate": 8.976e-05,
      "loss": 0.5292,
      "step": 17961
    },
    {
      "epoch": 1.0058237204614178,
      "grad_norm": 1.4708261489868164,
      "learning_rate": 8.9765e-05,
      "loss": 0.4407,
      "step": 17962
    },
    {
      "epoch": 1.0058797177735468,
      "grad_norm": 1.3044770956039429,
      "learning_rate": 8.977000000000002e-05,
      "loss": 0.6537,
      "step": 17963
    },
    {
      "epoch": 1.0059357150856758,
      "grad_norm": 1.1265450716018677,
      "learning_rate": 8.977500000000001e-05,
      "loss": 0.4736,
      "step": 17964
    },
    {
      "epoch": 1.0059917123978048,
      "grad_norm": 1.3219960927963257,
      "learning_rate": 8.978000000000001e-05,
      "loss": 0.3908,
      "step": 17965
    },
    {
      "epoch": 1.0060477097099338,
      "grad_norm": 1.1987253427505493,
      "learning_rate": 8.978500000000001e-05,
      "loss": 0.4455,
      "step": 17966
    },
    {
      "epoch": 1.0061037070220629,
      "grad_norm": 1.7745444774627686,
      "learning_rate": 8.979e-05,
      "loss": 0.5543,
      "step": 17967
    },
    {
      "epoch": 1.0061597043341919,
      "grad_norm": 1.3833746910095215,
      "learning_rate": 8.9795e-05,
      "loss": 0.5038,
      "step": 17968
    },
    {
      "epoch": 1.006215701646321,
      "grad_norm": 1.6064419746398926,
      "learning_rate": 8.98e-05,
      "loss": 0.4707,
      "step": 17969
    },
    {
      "epoch": 1.00627169895845,
      "grad_norm": 1.8970528841018677,
      "learning_rate": 8.980500000000001e-05,
      "loss": 0.532,
      "step": 17970
    },
    {
      "epoch": 1.006327696270579,
      "grad_norm": 1.2309457063674927,
      "learning_rate": 8.981000000000001e-05,
      "loss": 0.375,
      "step": 17971
    },
    {
      "epoch": 1.006383693582708,
      "grad_norm": 1.2881358861923218,
      "learning_rate": 8.9815e-05,
      "loss": 0.4442,
      "step": 17972
    },
    {
      "epoch": 1.006439690894837,
      "grad_norm": 2.184856653213501,
      "learning_rate": 8.982e-05,
      "loss": 0.3165,
      "step": 17973
    },
    {
      "epoch": 1.006495688206966,
      "grad_norm": 1.2457536458969116,
      "learning_rate": 8.9825e-05,
      "loss": 0.3519,
      "step": 17974
    },
    {
      "epoch": 1.006551685519095,
      "grad_norm": 1.2922468185424805,
      "learning_rate": 8.983e-05,
      "loss": 0.361,
      "step": 17975
    },
    {
      "epoch": 1.006607682831224,
      "grad_norm": 1.3228814601898193,
      "learning_rate": 8.9835e-05,
      "loss": 0.377,
      "step": 17976
    },
    {
      "epoch": 1.006663680143353,
      "grad_norm": 1.4710068702697754,
      "learning_rate": 8.984000000000001e-05,
      "loss": 0.4726,
      "step": 17977
    },
    {
      "epoch": 1.006719677455482,
      "grad_norm": 1.1677091121673584,
      "learning_rate": 8.9845e-05,
      "loss": 0.5768,
      "step": 17978
    },
    {
      "epoch": 1.006775674767611,
      "grad_norm": 1.220658302307129,
      "learning_rate": 8.985e-05,
      "loss": 0.4234,
      "step": 17979
    },
    {
      "epoch": 1.0068316720797401,
      "grad_norm": 1.3482164144515991,
      "learning_rate": 8.9855e-05,
      "loss": 0.3233,
      "step": 17980
    },
    {
      "epoch": 1.0068876693918691,
      "grad_norm": 1.226611614227295,
      "learning_rate": 8.986e-05,
      "loss": 0.4663,
      "step": 17981
    },
    {
      "epoch": 1.0069436667039982,
      "grad_norm": 2.298543691635132,
      "learning_rate": 8.9865e-05,
      "loss": 0.4426,
      "step": 17982
    },
    {
      "epoch": 1.0069996640161272,
      "grad_norm": 1.257027268409729,
      "learning_rate": 8.987e-05,
      "loss": 0.454,
      "step": 17983
    },
    {
      "epoch": 1.0070556613282562,
      "grad_norm": 1.2130361795425415,
      "learning_rate": 8.9875e-05,
      "loss": 0.5599,
      "step": 17984
    },
    {
      "epoch": 1.0071116586403852,
      "grad_norm": 1.4071629047393799,
      "learning_rate": 8.988000000000001e-05,
      "loss": 0.4986,
      "step": 17985
    },
    {
      "epoch": 1.0071676559525142,
      "grad_norm": 1.417580485343933,
      "learning_rate": 8.988500000000001e-05,
      "loss": 0.426,
      "step": 17986
    },
    {
      "epoch": 1.0072236532646432,
      "grad_norm": 1.2887449264526367,
      "learning_rate": 8.989000000000001e-05,
      "loss": 0.3997,
      "step": 17987
    },
    {
      "epoch": 1.0072796505767723,
      "grad_norm": 0.9595540165901184,
      "learning_rate": 8.9895e-05,
      "loss": 0.2766,
      "step": 17988
    },
    {
      "epoch": 1.0073356478889013,
      "grad_norm": 1.271773099899292,
      "learning_rate": 8.99e-05,
      "loss": 0.4387,
      "step": 17989
    },
    {
      "epoch": 1.0073916452010303,
      "grad_norm": 1.2614787817001343,
      "learning_rate": 8.9905e-05,
      "loss": 0.3617,
      "step": 17990
    },
    {
      "epoch": 1.0074476425131593,
      "grad_norm": 1.2842191457748413,
      "learning_rate": 8.991000000000001e-05,
      "loss": 0.3945,
      "step": 17991
    },
    {
      "epoch": 1.0075036398252883,
      "grad_norm": 1.3162118196487427,
      "learning_rate": 8.991500000000001e-05,
      "loss": 0.402,
      "step": 17992
    },
    {
      "epoch": 1.0075596371374174,
      "grad_norm": 1.1518367528915405,
      "learning_rate": 8.992e-05,
      "loss": 0.453,
      "step": 17993
    },
    {
      "epoch": 1.0076156344495464,
      "grad_norm": 1.467724323272705,
      "learning_rate": 8.9925e-05,
      "loss": 0.3989,
      "step": 17994
    },
    {
      "epoch": 1.0076716317616754,
      "grad_norm": 1.5824946165084839,
      "learning_rate": 8.993e-05,
      "loss": 0.4059,
      "step": 17995
    },
    {
      "epoch": 1.0077276290738044,
      "grad_norm": 1.2596561908721924,
      "learning_rate": 8.9935e-05,
      "loss": 0.3671,
      "step": 17996
    },
    {
      "epoch": 1.0077836263859334,
      "grad_norm": 1.2339483499526978,
      "learning_rate": 8.994e-05,
      "loss": 0.359,
      "step": 17997
    },
    {
      "epoch": 1.0078396236980625,
      "grad_norm": 1.5595121383666992,
      "learning_rate": 8.994500000000001e-05,
      "loss": 0.4069,
      "step": 17998
    },
    {
      "epoch": 1.0078956210101915,
      "grad_norm": 1.2758089303970337,
      "learning_rate": 8.995e-05,
      "loss": 0.5216,
      "step": 17999
    },
    {
      "epoch": 1.0079516183223205,
      "grad_norm": 1.5260242223739624,
      "learning_rate": 8.9955e-05,
      "loss": 0.49,
      "step": 18000
    },
    {
      "epoch": 1.0080076156344495,
      "grad_norm": 1.3660736083984375,
      "learning_rate": 8.996e-05,
      "loss": 0.3987,
      "step": 18001
    },
    {
      "epoch": 1.0080636129465785,
      "grad_norm": 1.0238418579101562,
      "learning_rate": 8.9965e-05,
      "loss": 0.3057,
      "step": 18002
    },
    {
      "epoch": 1.0081196102587076,
      "grad_norm": 1.3377869129180908,
      "learning_rate": 8.997000000000001e-05,
      "loss": 0.5516,
      "step": 18003
    },
    {
      "epoch": 1.0081756075708366,
      "grad_norm": 1.399433970451355,
      "learning_rate": 8.9975e-05,
      "loss": 0.433,
      "step": 18004
    },
    {
      "epoch": 1.0082316048829656,
      "grad_norm": 1.4594414234161377,
      "learning_rate": 8.998e-05,
      "loss": 0.3464,
      "step": 18005
    },
    {
      "epoch": 1.0082876021950946,
      "grad_norm": 1.1530085802078247,
      "learning_rate": 8.998500000000001e-05,
      "loss": 0.3205,
      "step": 18006
    },
    {
      "epoch": 1.0083435995072236,
      "grad_norm": 1.2954506874084473,
      "learning_rate": 8.999000000000001e-05,
      "loss": 0.3969,
      "step": 18007
    },
    {
      "epoch": 1.0083995968193527,
      "grad_norm": 1.3576916456222534,
      "learning_rate": 8.999500000000001e-05,
      "loss": 0.4051,
      "step": 18008
    },
    {
      "epoch": 1.0084555941314817,
      "grad_norm": 1.2132610082626343,
      "learning_rate": 9e-05,
      "loss": 0.4627,
      "step": 18009
    },
    {
      "epoch": 1.0085115914436107,
      "grad_norm": 1.4459677934646606,
      "learning_rate": 9.0005e-05,
      "loss": 0.4382,
      "step": 18010
    },
    {
      "epoch": 1.0085675887557397,
      "grad_norm": 1.4301097393035889,
      "learning_rate": 9.001e-05,
      "loss": 0.477,
      "step": 18011
    },
    {
      "epoch": 1.0086235860678687,
      "grad_norm": 1.1272304058074951,
      "learning_rate": 9.001500000000001e-05,
      "loss": 0.3235,
      "step": 18012
    },
    {
      "epoch": 1.0086795833799977,
      "grad_norm": 1.4270473718643188,
      "learning_rate": 9.002000000000001e-05,
      "loss": 0.4742,
      "step": 18013
    },
    {
      "epoch": 1.0087355806921268,
      "grad_norm": 1.3131418228149414,
      "learning_rate": 9.0025e-05,
      "loss": 0.3297,
      "step": 18014
    },
    {
      "epoch": 1.0087915780042558,
      "grad_norm": 1.2703737020492554,
      "learning_rate": 9.003e-05,
      "loss": 0.3442,
      "step": 18015
    },
    {
      "epoch": 1.0088475753163848,
      "grad_norm": 1.1593832969665527,
      "learning_rate": 9.0035e-05,
      "loss": 0.3833,
      "step": 18016
    },
    {
      "epoch": 1.0089035726285138,
      "grad_norm": 1.2208229303359985,
      "learning_rate": 9.004e-05,
      "loss": 0.2724,
      "step": 18017
    },
    {
      "epoch": 1.0089595699406428,
      "grad_norm": 1.154589056968689,
      "learning_rate": 9.0045e-05,
      "loss": 0.3801,
      "step": 18018
    },
    {
      "epoch": 1.0090155672527719,
      "grad_norm": 1.2062815427780151,
      "learning_rate": 9.005000000000001e-05,
      "loss": 0.3698,
      "step": 18019
    },
    {
      "epoch": 1.0090715645649009,
      "grad_norm": 1.1477242708206177,
      "learning_rate": 9.0055e-05,
      "loss": 0.3661,
      "step": 18020
    },
    {
      "epoch": 1.00912756187703,
      "grad_norm": 1.3812172412872314,
      "learning_rate": 9.006e-05,
      "loss": 0.4077,
      "step": 18021
    },
    {
      "epoch": 1.009183559189159,
      "grad_norm": 1.115505337715149,
      "learning_rate": 9.0065e-05,
      "loss": 0.4108,
      "step": 18022
    },
    {
      "epoch": 1.009239556501288,
      "grad_norm": 1.1311782598495483,
      "learning_rate": 9.007e-05,
      "loss": 0.3722,
      "step": 18023
    },
    {
      "epoch": 1.009295553813417,
      "grad_norm": 1.1441736221313477,
      "learning_rate": 9.007500000000001e-05,
      "loss": 0.3694,
      "step": 18024
    },
    {
      "epoch": 1.009351551125546,
      "grad_norm": 1.6131408214569092,
      "learning_rate": 9.008e-05,
      "loss": 0.4297,
      "step": 18025
    },
    {
      "epoch": 1.009407548437675,
      "grad_norm": 1.1815921068191528,
      "learning_rate": 9.0085e-05,
      "loss": 0.4238,
      "step": 18026
    },
    {
      "epoch": 1.009463545749804,
      "grad_norm": 1.164927363395691,
      "learning_rate": 9.009000000000001e-05,
      "loss": 0.4517,
      "step": 18027
    },
    {
      "epoch": 1.009519543061933,
      "grad_norm": 1.2917762994766235,
      "learning_rate": 9.009500000000001e-05,
      "loss": 0.3382,
      "step": 18028
    },
    {
      "epoch": 1.009575540374062,
      "grad_norm": 1.2980742454528809,
      "learning_rate": 9.010000000000001e-05,
      "loss": 0.4674,
      "step": 18029
    },
    {
      "epoch": 1.009631537686191,
      "grad_norm": 1.3152427673339844,
      "learning_rate": 9.0105e-05,
      "loss": 0.5338,
      "step": 18030
    },
    {
      "epoch": 1.00968753499832,
      "grad_norm": 1.1749911308288574,
      "learning_rate": 9.011e-05,
      "loss": 0.5392,
      "step": 18031
    },
    {
      "epoch": 1.009743532310449,
      "grad_norm": 1.306301236152649,
      "learning_rate": 9.0115e-05,
      "loss": 0.4575,
      "step": 18032
    },
    {
      "epoch": 1.0097995296225781,
      "grad_norm": 1.0710113048553467,
      "learning_rate": 9.012e-05,
      "loss": 0.3521,
      "step": 18033
    },
    {
      "epoch": 1.0098555269347071,
      "grad_norm": 1.607362985610962,
      "learning_rate": 9.012500000000001e-05,
      "loss": 0.4986,
      "step": 18034
    },
    {
      "epoch": 1.0099115242468362,
      "grad_norm": 1.424028992652893,
      "learning_rate": 9.013e-05,
      "loss": 0.4623,
      "step": 18035
    },
    {
      "epoch": 1.0099675215589652,
      "grad_norm": 1.2317618131637573,
      "learning_rate": 9.0135e-05,
      "loss": 0.3437,
      "step": 18036
    },
    {
      "epoch": 1.0100235188710942,
      "grad_norm": 1.0915732383728027,
      "learning_rate": 9.014e-05,
      "loss": 0.3834,
      "step": 18037
    },
    {
      "epoch": 1.0100795161832232,
      "grad_norm": 1.811660647392273,
      "learning_rate": 9.0145e-05,
      "loss": 0.3943,
      "step": 18038
    },
    {
      "epoch": 1.0101355134953522,
      "grad_norm": 1.3503577709197998,
      "learning_rate": 9.015e-05,
      "loss": 0.4379,
      "step": 18039
    },
    {
      "epoch": 1.0101915108074813,
      "grad_norm": 1.38282310962677,
      "learning_rate": 9.0155e-05,
      "loss": 0.5986,
      "step": 18040
    },
    {
      "epoch": 1.0102475081196103,
      "grad_norm": 1.1291764974594116,
      "learning_rate": 9.016e-05,
      "loss": 0.3217,
      "step": 18041
    },
    {
      "epoch": 1.0103035054317393,
      "grad_norm": 1.450632929801941,
      "learning_rate": 9.0165e-05,
      "loss": 0.3424,
      "step": 18042
    },
    {
      "epoch": 1.0103595027438683,
      "grad_norm": 1.3379839658737183,
      "learning_rate": 9.017e-05,
      "loss": 0.3377,
      "step": 18043
    },
    {
      "epoch": 1.0104155000559973,
      "grad_norm": 1.598143458366394,
      "learning_rate": 9.017500000000001e-05,
      "loss": 0.4294,
      "step": 18044
    },
    {
      "epoch": 1.0104714973681264,
      "grad_norm": 1.1232203245162964,
      "learning_rate": 9.018000000000001e-05,
      "loss": 0.3088,
      "step": 18045
    },
    {
      "epoch": 1.0105274946802554,
      "grad_norm": 1.3722296953201294,
      "learning_rate": 9.0185e-05,
      "loss": 0.466,
      "step": 18046
    },
    {
      "epoch": 1.0105834919923844,
      "grad_norm": 2.807011127471924,
      "learning_rate": 9.019e-05,
      "loss": 0.5928,
      "step": 18047
    },
    {
      "epoch": 1.0106394893045134,
      "grad_norm": 1.2643423080444336,
      "learning_rate": 9.019500000000001e-05,
      "loss": 0.4106,
      "step": 18048
    },
    {
      "epoch": 1.0106954866166424,
      "grad_norm": 1.3486216068267822,
      "learning_rate": 9.020000000000001e-05,
      "loss": 0.4426,
      "step": 18049
    },
    {
      "epoch": 1.0107514839287715,
      "grad_norm": 1.446391224861145,
      "learning_rate": 9.020500000000001e-05,
      "loss": 0.3916,
      "step": 18050
    },
    {
      "epoch": 1.0108074812409005,
      "grad_norm": 1.3402318954467773,
      "learning_rate": 9.021e-05,
      "loss": 0.3619,
      "step": 18051
    },
    {
      "epoch": 1.0108634785530295,
      "grad_norm": 1.302295446395874,
      "learning_rate": 9.0215e-05,
      "loss": 0.2946,
      "step": 18052
    },
    {
      "epoch": 1.0109194758651585,
      "grad_norm": 1.367607831954956,
      "learning_rate": 9.022e-05,
      "loss": 0.4433,
      "step": 18053
    },
    {
      "epoch": 1.0109754731772875,
      "grad_norm": 1.4880040884017944,
      "learning_rate": 9.0225e-05,
      "loss": 0.4124,
      "step": 18054
    },
    {
      "epoch": 1.0110314704894166,
      "grad_norm": 1.5772989988327026,
      "learning_rate": 9.023000000000001e-05,
      "loss": 0.5283,
      "step": 18055
    },
    {
      "epoch": 1.0110874678015456,
      "grad_norm": 1.2875374555587769,
      "learning_rate": 9.0235e-05,
      "loss": 0.5414,
      "step": 18056
    },
    {
      "epoch": 1.0111434651136746,
      "grad_norm": 1.363318920135498,
      "learning_rate": 9.024e-05,
      "loss": 0.484,
      "step": 18057
    },
    {
      "epoch": 1.0111994624258036,
      "grad_norm": 1.4215304851531982,
      "learning_rate": 9.0245e-05,
      "loss": 0.3954,
      "step": 18058
    },
    {
      "epoch": 1.0112554597379326,
      "grad_norm": 1.273647665977478,
      "learning_rate": 9.025e-05,
      "loss": 0.3881,
      "step": 18059
    },
    {
      "epoch": 1.0113114570500616,
      "grad_norm": 1.1219748258590698,
      "learning_rate": 9.0255e-05,
      "loss": 0.432,
      "step": 18060
    },
    {
      "epoch": 1.0113674543621907,
      "grad_norm": 1.283856749534607,
      "learning_rate": 9.026e-05,
      "loss": 0.3797,
      "step": 18061
    },
    {
      "epoch": 1.0114234516743197,
      "grad_norm": 1.2342506647109985,
      "learning_rate": 9.0265e-05,
      "loss": 0.3694,
      "step": 18062
    },
    {
      "epoch": 1.0114794489864487,
      "grad_norm": 1.2998380661010742,
      "learning_rate": 9.027e-05,
      "loss": 0.3719,
      "step": 18063
    },
    {
      "epoch": 1.0115354462985777,
      "grad_norm": 1.250680685043335,
      "learning_rate": 9.027500000000001e-05,
      "loss": 0.4127,
      "step": 18064
    },
    {
      "epoch": 1.0115914436107067,
      "grad_norm": 1.2504264116287231,
      "learning_rate": 9.028000000000001e-05,
      "loss": 0.4795,
      "step": 18065
    },
    {
      "epoch": 1.0116474409228358,
      "grad_norm": 1.4863789081573486,
      "learning_rate": 9.028500000000001e-05,
      "loss": 0.4474,
      "step": 18066
    },
    {
      "epoch": 1.0117034382349648,
      "grad_norm": 1.2361491918563843,
      "learning_rate": 9.029e-05,
      "loss": 0.4698,
      "step": 18067
    },
    {
      "epoch": 1.0117594355470938,
      "grad_norm": 1.6292223930358887,
      "learning_rate": 9.0295e-05,
      "loss": 0.4569,
      "step": 18068
    },
    {
      "epoch": 1.0118154328592228,
      "grad_norm": 1.5858267545700073,
      "learning_rate": 9.030000000000001e-05,
      "loss": 0.3559,
      "step": 18069
    },
    {
      "epoch": 1.0118714301713518,
      "grad_norm": 1.1331160068511963,
      "learning_rate": 9.030500000000001e-05,
      "loss": 0.295,
      "step": 18070
    },
    {
      "epoch": 1.0119274274834809,
      "grad_norm": 1.5241893529891968,
      "learning_rate": 9.031000000000001e-05,
      "loss": 0.4411,
      "step": 18071
    },
    {
      "epoch": 1.0119834247956099,
      "grad_norm": 1.3934721946716309,
      "learning_rate": 9.0315e-05,
      "loss": 0.5948,
      "step": 18072
    },
    {
      "epoch": 1.012039422107739,
      "grad_norm": 1.6788913011550903,
      "learning_rate": 9.032e-05,
      "loss": 0.3659,
      "step": 18073
    },
    {
      "epoch": 1.012095419419868,
      "grad_norm": 1.335060715675354,
      "learning_rate": 9.0325e-05,
      "loss": 0.31,
      "step": 18074
    },
    {
      "epoch": 1.012151416731997,
      "grad_norm": 1.4532748460769653,
      "learning_rate": 9.033e-05,
      "loss": 0.524,
      "step": 18075
    },
    {
      "epoch": 1.012207414044126,
      "grad_norm": 1.3463493585586548,
      "learning_rate": 9.033500000000001e-05,
      "loss": 0.3406,
      "step": 18076
    },
    {
      "epoch": 1.012263411356255,
      "grad_norm": 1.3908874988555908,
      "learning_rate": 9.034e-05,
      "loss": 0.3856,
      "step": 18077
    },
    {
      "epoch": 1.012319408668384,
      "grad_norm": 1.3213330507278442,
      "learning_rate": 9.0345e-05,
      "loss": 0.3887,
      "step": 18078
    },
    {
      "epoch": 1.012375405980513,
      "grad_norm": 1.2592127323150635,
      "learning_rate": 9.035e-05,
      "loss": 0.3366,
      "step": 18079
    },
    {
      "epoch": 1.012431403292642,
      "grad_norm": 1.0710480213165283,
      "learning_rate": 9.0355e-05,
      "loss": 0.3704,
      "step": 18080
    },
    {
      "epoch": 1.012487400604771,
      "grad_norm": 1.4461910724639893,
      "learning_rate": 9.036e-05,
      "loss": 0.5877,
      "step": 18081
    },
    {
      "epoch": 1.0125433979169,
      "grad_norm": 1.5060564279556274,
      "learning_rate": 9.0365e-05,
      "loss": 0.5551,
      "step": 18082
    },
    {
      "epoch": 1.012599395229029,
      "grad_norm": 1.3501626253128052,
      "learning_rate": 9.037e-05,
      "loss": 0.4159,
      "step": 18083
    },
    {
      "epoch": 1.012655392541158,
      "grad_norm": 1.2449951171875,
      "learning_rate": 9.037500000000001e-05,
      "loss": 0.4655,
      "step": 18084
    },
    {
      "epoch": 1.0127113898532871,
      "grad_norm": 1.2275044918060303,
      "learning_rate": 9.038000000000001e-05,
      "loss": 0.3742,
      "step": 18085
    },
    {
      "epoch": 1.0127673871654161,
      "grad_norm": 1.4492491483688354,
      "learning_rate": 9.038500000000001e-05,
      "loss": 0.4606,
      "step": 18086
    },
    {
      "epoch": 1.0128233844775452,
      "grad_norm": 1.2193833589553833,
      "learning_rate": 9.039000000000001e-05,
      "loss": 0.447,
      "step": 18087
    },
    {
      "epoch": 1.0128793817896742,
      "grad_norm": 1.247789978981018,
      "learning_rate": 9.0395e-05,
      "loss": 0.446,
      "step": 18088
    },
    {
      "epoch": 1.0129353791018032,
      "grad_norm": 1.3559398651123047,
      "learning_rate": 9.04e-05,
      "loss": 0.5124,
      "step": 18089
    },
    {
      "epoch": 1.0129913764139322,
      "grad_norm": 1.4070031642913818,
      "learning_rate": 9.040500000000001e-05,
      "loss": 0.39,
      "step": 18090
    },
    {
      "epoch": 1.0130473737260612,
      "grad_norm": 1.2927995920181274,
      "learning_rate": 9.041000000000001e-05,
      "loss": 0.3433,
      "step": 18091
    },
    {
      "epoch": 1.0131033710381903,
      "grad_norm": 1.4199851751327515,
      "learning_rate": 9.041500000000001e-05,
      "loss": 0.3848,
      "step": 18092
    },
    {
      "epoch": 1.0131593683503193,
      "grad_norm": 1.090008020401001,
      "learning_rate": 9.042e-05,
      "loss": 0.3827,
      "step": 18093
    },
    {
      "epoch": 1.0132153656624483,
      "grad_norm": 1.307547926902771,
      "learning_rate": 9.0425e-05,
      "loss": 0.4308,
      "step": 18094
    },
    {
      "epoch": 1.0132713629745773,
      "grad_norm": 1.404300332069397,
      "learning_rate": 9.043e-05,
      "loss": 0.5085,
      "step": 18095
    },
    {
      "epoch": 1.0133273602867063,
      "grad_norm": 1.1837518215179443,
      "learning_rate": 9.0435e-05,
      "loss": 0.404,
      "step": 18096
    },
    {
      "epoch": 1.0133833575988354,
      "grad_norm": 1.3322854042053223,
      "learning_rate": 9.044000000000001e-05,
      "loss": 0.4325,
      "step": 18097
    },
    {
      "epoch": 1.0134393549109644,
      "grad_norm": 1.3566287755966187,
      "learning_rate": 9.0445e-05,
      "loss": 0.5156,
      "step": 18098
    },
    {
      "epoch": 1.0134953522230934,
      "grad_norm": 1.5615484714508057,
      "learning_rate": 9.045e-05,
      "loss": 0.3841,
      "step": 18099
    },
    {
      "epoch": 1.0135513495352224,
      "grad_norm": 3.3497726917266846,
      "learning_rate": 9.0455e-05,
      "loss": 0.4027,
      "step": 18100
    },
    {
      "epoch": 1.0136073468473514,
      "grad_norm": 1.3943313360214233,
      "learning_rate": 9.046e-05,
      "loss": 0.3069,
      "step": 18101
    },
    {
      "epoch": 1.0136633441594805,
      "grad_norm": 1.2476478815078735,
      "learning_rate": 9.0465e-05,
      "loss": 0.5064,
      "step": 18102
    },
    {
      "epoch": 1.0137193414716095,
      "grad_norm": 1.4256521463394165,
      "learning_rate": 9.046999999999999e-05,
      "loss": 0.4229,
      "step": 18103
    },
    {
      "epoch": 1.0137753387837385,
      "grad_norm": 1.2707245349884033,
      "learning_rate": 9.0475e-05,
      "loss": 0.5447,
      "step": 18104
    },
    {
      "epoch": 1.0138313360958675,
      "grad_norm": 1.1792269945144653,
      "learning_rate": 9.048000000000001e-05,
      "loss": 0.4014,
      "step": 18105
    },
    {
      "epoch": 1.0138873334079963,
      "grad_norm": 1.4103285074234009,
      "learning_rate": 9.048500000000001e-05,
      "loss": 0.5126,
      "step": 18106
    },
    {
      "epoch": 1.0139433307201253,
      "grad_norm": 1.3352609872817993,
      "learning_rate": 9.049000000000001e-05,
      "loss": 0.3986,
      "step": 18107
    },
    {
      "epoch": 1.0139993280322543,
      "grad_norm": 1.3411178588867188,
      "learning_rate": 9.049500000000001e-05,
      "loss": 0.32,
      "step": 18108
    },
    {
      "epoch": 1.0140553253443834,
      "grad_norm": 1.361093282699585,
      "learning_rate": 9.05e-05,
      "loss": 0.4446,
      "step": 18109
    },
    {
      "epoch": 1.0141113226565124,
      "grad_norm": 1.238129734992981,
      "learning_rate": 9.0505e-05,
      "loss": 0.3947,
      "step": 18110
    },
    {
      "epoch": 1.0141673199686414,
      "grad_norm": 1.7928906679153442,
      "learning_rate": 9.051000000000001e-05,
      "loss": 0.3852,
      "step": 18111
    },
    {
      "epoch": 1.0142233172807704,
      "grad_norm": 1.4398564100265503,
      "learning_rate": 9.051500000000001e-05,
      "loss": 0.4593,
      "step": 18112
    },
    {
      "epoch": 1.0142793145928994,
      "grad_norm": 1.3197253942489624,
      "learning_rate": 9.052000000000001e-05,
      "loss": 0.3959,
      "step": 18113
    },
    {
      "epoch": 1.0143353119050285,
      "grad_norm": 1.3324925899505615,
      "learning_rate": 9.0525e-05,
      "loss": 0.3748,
      "step": 18114
    },
    {
      "epoch": 1.0143913092171575,
      "grad_norm": 1.260272741317749,
      "learning_rate": 9.053e-05,
      "loss": 0.3949,
      "step": 18115
    },
    {
      "epoch": 1.0144473065292865,
      "grad_norm": 1.5691372156143188,
      "learning_rate": 9.0535e-05,
      "loss": 0.3,
      "step": 18116
    },
    {
      "epoch": 1.0145033038414155,
      "grad_norm": 1.142928123474121,
      "learning_rate": 9.054e-05,
      "loss": 0.3668,
      "step": 18117
    },
    {
      "epoch": 1.0145593011535445,
      "grad_norm": 1.4724918603897095,
      "learning_rate": 9.054500000000001e-05,
      "loss": 0.4484,
      "step": 18118
    },
    {
      "epoch": 1.0146152984656736,
      "grad_norm": 1.3410335779190063,
      "learning_rate": 9.055e-05,
      "loss": 0.4898,
      "step": 18119
    },
    {
      "epoch": 1.0146712957778026,
      "grad_norm": 1.319427728652954,
      "learning_rate": 9.0555e-05,
      "loss": 0.4391,
      "step": 18120
    },
    {
      "epoch": 1.0147272930899316,
      "grad_norm": 1.2103686332702637,
      "learning_rate": 9.056e-05,
      "loss": 0.3887,
      "step": 18121
    },
    {
      "epoch": 1.0147832904020606,
      "grad_norm": 1.0283188819885254,
      "learning_rate": 9.0565e-05,
      "loss": 0.3699,
      "step": 18122
    },
    {
      "epoch": 1.0148392877141896,
      "grad_norm": 1.727451205253601,
      "learning_rate": 9.057e-05,
      "loss": 0.6659,
      "step": 18123
    },
    {
      "epoch": 1.0148952850263186,
      "grad_norm": 1.2413078546524048,
      "learning_rate": 9.0575e-05,
      "loss": 0.4249,
      "step": 18124
    },
    {
      "epoch": 1.0149512823384477,
      "grad_norm": 2.3689627647399902,
      "learning_rate": 9.058e-05,
      "loss": 0.4267,
      "step": 18125
    },
    {
      "epoch": 1.0150072796505767,
      "grad_norm": 1.4286799430847168,
      "learning_rate": 9.058500000000001e-05,
      "loss": 0.3831,
      "step": 18126
    },
    {
      "epoch": 1.0150632769627057,
      "grad_norm": 1.223140835762024,
      "learning_rate": 9.059000000000001e-05,
      "loss": 0.4263,
      "step": 18127
    },
    {
      "epoch": 1.0151192742748347,
      "grad_norm": 1.2765775918960571,
      "learning_rate": 9.059500000000001e-05,
      "loss": 0.4055,
      "step": 18128
    },
    {
      "epoch": 1.0151752715869637,
      "grad_norm": 1.4111768007278442,
      "learning_rate": 9.06e-05,
      "loss": 0.4041,
      "step": 18129
    },
    {
      "epoch": 1.0152312688990928,
      "grad_norm": 1.454330325126648,
      "learning_rate": 9.0605e-05,
      "loss": 0.3906,
      "step": 18130
    },
    {
      "epoch": 1.0152872662112218,
      "grad_norm": 1.5828416347503662,
      "learning_rate": 9.061e-05,
      "loss": 0.4863,
      "step": 18131
    },
    {
      "epoch": 1.0153432635233508,
      "grad_norm": 1.7145265340805054,
      "learning_rate": 9.0615e-05,
      "loss": 0.5591,
      "step": 18132
    },
    {
      "epoch": 1.0153992608354798,
      "grad_norm": 1.3797694444656372,
      "learning_rate": 9.062000000000001e-05,
      "loss": 0.4024,
      "step": 18133
    },
    {
      "epoch": 1.0154552581476088,
      "grad_norm": 1.3734862804412842,
      "learning_rate": 9.062500000000001e-05,
      "loss": 0.3615,
      "step": 18134
    },
    {
      "epoch": 1.0155112554597379,
      "grad_norm": 1.5914740562438965,
      "learning_rate": 9.063e-05,
      "loss": 0.3757,
      "step": 18135
    },
    {
      "epoch": 1.0155672527718669,
      "grad_norm": 1.196204423904419,
      "learning_rate": 9.0635e-05,
      "loss": 0.3574,
      "step": 18136
    },
    {
      "epoch": 1.015623250083996,
      "grad_norm": 1.4615166187286377,
      "learning_rate": 9.064e-05,
      "loss": 0.4394,
      "step": 18137
    },
    {
      "epoch": 1.015679247396125,
      "grad_norm": 1.5073596239089966,
      "learning_rate": 9.0645e-05,
      "loss": 0.3384,
      "step": 18138
    },
    {
      "epoch": 1.015735244708254,
      "grad_norm": 1.6935042142868042,
      "learning_rate": 9.065000000000001e-05,
      "loss": 0.6493,
      "step": 18139
    },
    {
      "epoch": 1.015791242020383,
      "grad_norm": 1.2316350936889648,
      "learning_rate": 9.0655e-05,
      "loss": 0.4317,
      "step": 18140
    },
    {
      "epoch": 1.015847239332512,
      "grad_norm": 1.423403024673462,
      "learning_rate": 9.066e-05,
      "loss": 0.4692,
      "step": 18141
    },
    {
      "epoch": 1.015903236644641,
      "grad_norm": 1.4836894273757935,
      "learning_rate": 9.0665e-05,
      "loss": 0.4839,
      "step": 18142
    },
    {
      "epoch": 1.01595923395677,
      "grad_norm": 1.208901047706604,
      "learning_rate": 9.067e-05,
      "loss": 0.2979,
      "step": 18143
    },
    {
      "epoch": 1.016015231268899,
      "grad_norm": 1.1822084188461304,
      "learning_rate": 9.0675e-05,
      "loss": 0.3664,
      "step": 18144
    },
    {
      "epoch": 1.016071228581028,
      "grad_norm": 1.3010916709899902,
      "learning_rate": 9.068e-05,
      "loss": 0.4969,
      "step": 18145
    },
    {
      "epoch": 1.016127225893157,
      "grad_norm": 1.1672778129577637,
      "learning_rate": 9.0685e-05,
      "loss": 0.4616,
      "step": 18146
    },
    {
      "epoch": 1.016183223205286,
      "grad_norm": 1.3962794542312622,
      "learning_rate": 9.069000000000001e-05,
      "loss": 0.3049,
      "step": 18147
    },
    {
      "epoch": 1.016239220517415,
      "grad_norm": 1.4669468402862549,
      "learning_rate": 9.069500000000001e-05,
      "loss": 0.4411,
      "step": 18148
    },
    {
      "epoch": 1.0162952178295441,
      "grad_norm": 1.3549784421920776,
      "learning_rate": 9.070000000000001e-05,
      "loss": 0.3852,
      "step": 18149
    },
    {
      "epoch": 1.0163512151416731,
      "grad_norm": 1.1801265478134155,
      "learning_rate": 9.0705e-05,
      "loss": 0.3111,
      "step": 18150
    },
    {
      "epoch": 1.0164072124538022,
      "grad_norm": 1.7732163667678833,
      "learning_rate": 9.071e-05,
      "loss": 0.4371,
      "step": 18151
    },
    {
      "epoch": 1.0164632097659312,
      "grad_norm": 1.2452828884124756,
      "learning_rate": 9.0715e-05,
      "loss": 0.4395,
      "step": 18152
    },
    {
      "epoch": 1.0165192070780602,
      "grad_norm": 1.396883487701416,
      "learning_rate": 9.072e-05,
      "loss": 0.4847,
      "step": 18153
    },
    {
      "epoch": 1.0165752043901892,
      "grad_norm": 1.3976457118988037,
      "learning_rate": 9.072500000000001e-05,
      "loss": 0.4056,
      "step": 18154
    },
    {
      "epoch": 1.0166312017023182,
      "grad_norm": 1.2956854104995728,
      "learning_rate": 9.073000000000001e-05,
      "loss": 0.402,
      "step": 18155
    },
    {
      "epoch": 1.0166871990144473,
      "grad_norm": 1.400383472442627,
      "learning_rate": 9.0735e-05,
      "loss": 0.3289,
      "step": 18156
    },
    {
      "epoch": 1.0167431963265763,
      "grad_norm": 1.3283774852752686,
      "learning_rate": 9.074e-05,
      "loss": 0.3727,
      "step": 18157
    },
    {
      "epoch": 1.0167991936387053,
      "grad_norm": 1.4578040838241577,
      "learning_rate": 9.0745e-05,
      "loss": 0.4059,
      "step": 18158
    },
    {
      "epoch": 1.0168551909508343,
      "grad_norm": 1.25513756275177,
      "learning_rate": 9.075e-05,
      "loss": 0.4249,
      "step": 18159
    },
    {
      "epoch": 1.0169111882629633,
      "grad_norm": 2.642794370651245,
      "learning_rate": 9.075500000000001e-05,
      "loss": 0.5193,
      "step": 18160
    },
    {
      "epoch": 1.0169671855750924,
      "grad_norm": 1.7181202173233032,
      "learning_rate": 9.076e-05,
      "loss": 0.4596,
      "step": 18161
    },
    {
      "epoch": 1.0170231828872214,
      "grad_norm": 1.3589255809783936,
      "learning_rate": 9.0765e-05,
      "loss": 0.615,
      "step": 18162
    },
    {
      "epoch": 1.0170791801993504,
      "grad_norm": 1.5083166360855103,
      "learning_rate": 9.077e-05,
      "loss": 0.4295,
      "step": 18163
    },
    {
      "epoch": 1.0171351775114794,
      "grad_norm": 2.4867968559265137,
      "learning_rate": 9.0775e-05,
      "loss": 0.4186,
      "step": 18164
    },
    {
      "epoch": 1.0171911748236084,
      "grad_norm": 1.244420051574707,
      "learning_rate": 9.078000000000001e-05,
      "loss": 0.399,
      "step": 18165
    },
    {
      "epoch": 1.0172471721357375,
      "grad_norm": 1.086611270904541,
      "learning_rate": 9.0785e-05,
      "loss": 0.4294,
      "step": 18166
    },
    {
      "epoch": 1.0173031694478665,
      "grad_norm": 1.6594682931900024,
      "learning_rate": 9.079e-05,
      "loss": 0.6585,
      "step": 18167
    },
    {
      "epoch": 1.0173591667599955,
      "grad_norm": Infinity,
      "learning_rate": 9.079e-05,
      "loss": 0.4973,
      "step": 18168
    },
    {
      "epoch": 1.0174151640721245,
      "grad_norm": 1.4147491455078125,
      "learning_rate": 9.079500000000001e-05,
      "loss": 0.3646,
      "step": 18169
    },
    {
      "epoch": 1.0174711613842535,
      "grad_norm": 1.4743132591247559,
      "learning_rate": 9.080000000000001e-05,
      "loss": 0.4226,
      "step": 18170
    },
    {
      "epoch": 1.0175271586963825,
      "grad_norm": 1.4382675886154175,
      "learning_rate": 9.080500000000001e-05,
      "loss": 0.4246,
      "step": 18171
    },
    {
      "epoch": 1.0175831560085116,
      "grad_norm": 1.3491721153259277,
      "learning_rate": 9.081e-05,
      "loss": 0.5039,
      "step": 18172
    },
    {
      "epoch": 1.0176391533206406,
      "grad_norm": 1.3085360527038574,
      "learning_rate": 9.0815e-05,
      "loss": 0.4194,
      "step": 18173
    },
    {
      "epoch": 1.0176951506327696,
      "grad_norm": 1.3242313861846924,
      "learning_rate": 9.082e-05,
      "loss": 0.4244,
      "step": 18174
    },
    {
      "epoch": 1.0177511479448986,
      "grad_norm": 1.317463755607605,
      "learning_rate": 9.0825e-05,
      "loss": 0.4964,
      "step": 18175
    },
    {
      "epoch": 1.0178071452570276,
      "grad_norm": 1.1602150201797485,
      "learning_rate": 9.083000000000001e-05,
      "loss": 0.341,
      "step": 18176
    },
    {
      "epoch": 1.0178631425691567,
      "grad_norm": 1.1702351570129395,
      "learning_rate": 9.083500000000001e-05,
      "loss": 0.3677,
      "step": 18177
    },
    {
      "epoch": 1.0179191398812857,
      "grad_norm": 1.5528980493545532,
      "learning_rate": 9.084e-05,
      "loss": 0.4968,
      "step": 18178
    },
    {
      "epoch": 1.0179751371934147,
      "grad_norm": 1.2101691961288452,
      "learning_rate": 9.0845e-05,
      "loss": 0.3585,
      "step": 18179
    },
    {
      "epoch": 1.0180311345055437,
      "grad_norm": 1.367945671081543,
      "learning_rate": 9.085e-05,
      "loss": 0.3716,
      "step": 18180
    },
    {
      "epoch": 1.0180871318176727,
      "grad_norm": 1.5212262868881226,
      "learning_rate": 9.0855e-05,
      "loss": 0.4099,
      "step": 18181
    },
    {
      "epoch": 1.0181431291298018,
      "grad_norm": 1.1195323467254639,
      "learning_rate": 9.086e-05,
      "loss": 0.3555,
      "step": 18182
    },
    {
      "epoch": 1.0181991264419308,
      "grad_norm": 1.7062853574752808,
      "learning_rate": 9.0865e-05,
      "loss": 0.4578,
      "step": 18183
    },
    {
      "epoch": 1.0182551237540598,
      "grad_norm": 1.607496976852417,
      "learning_rate": 9.087e-05,
      "loss": 0.5234,
      "step": 18184
    },
    {
      "epoch": 1.0183111210661888,
      "grad_norm": 1.255609393119812,
      "learning_rate": 9.0875e-05,
      "loss": 0.5072,
      "step": 18185
    },
    {
      "epoch": 1.0183671183783178,
      "grad_norm": 1.1043859720230103,
      "learning_rate": 9.088000000000001e-05,
      "loss": 0.4505,
      "step": 18186
    },
    {
      "epoch": 1.0184231156904469,
      "grad_norm": 1.301988959312439,
      "learning_rate": 9.088500000000001e-05,
      "loss": 0.4329,
      "step": 18187
    },
    {
      "epoch": 1.0184791130025759,
      "grad_norm": 1.3064353466033936,
      "learning_rate": 9.089e-05,
      "loss": 0.3995,
      "step": 18188
    },
    {
      "epoch": 1.018535110314705,
      "grad_norm": 1.32597815990448,
      "learning_rate": 9.0895e-05,
      "loss": 0.4248,
      "step": 18189
    },
    {
      "epoch": 1.018591107626834,
      "grad_norm": 1.489014983177185,
      "learning_rate": 9.090000000000001e-05,
      "loss": 0.4669,
      "step": 18190
    },
    {
      "epoch": 1.018647104938963,
      "grad_norm": 1.5068203210830688,
      "learning_rate": 9.090500000000001e-05,
      "loss": 0.4461,
      "step": 18191
    },
    {
      "epoch": 1.018703102251092,
      "grad_norm": 1.1500208377838135,
      "learning_rate": 9.091000000000001e-05,
      "loss": 0.3835,
      "step": 18192
    },
    {
      "epoch": 1.018759099563221,
      "grad_norm": 1.5500932931900024,
      "learning_rate": 9.0915e-05,
      "loss": 0.4162,
      "step": 18193
    },
    {
      "epoch": 1.01881509687535,
      "grad_norm": 1.2251145839691162,
      "learning_rate": 9.092e-05,
      "loss": 0.3043,
      "step": 18194
    },
    {
      "epoch": 1.018871094187479,
      "grad_norm": 1.2725141048431396,
      "learning_rate": 9.0925e-05,
      "loss": 0.4343,
      "step": 18195
    },
    {
      "epoch": 1.018927091499608,
      "grad_norm": 1.1705259084701538,
      "learning_rate": 9.093e-05,
      "loss": 0.3781,
      "step": 18196
    },
    {
      "epoch": 1.018983088811737,
      "grad_norm": 1.4429186582565308,
      "learning_rate": 9.093500000000001e-05,
      "loss": 0.5007,
      "step": 18197
    },
    {
      "epoch": 1.019039086123866,
      "grad_norm": 1.3703272342681885,
      "learning_rate": 9.094000000000001e-05,
      "loss": 0.54,
      "step": 18198
    },
    {
      "epoch": 1.019095083435995,
      "grad_norm": 1.161527395248413,
      "learning_rate": 9.0945e-05,
      "loss": 0.3372,
      "step": 18199
    },
    {
      "epoch": 1.019151080748124,
      "grad_norm": 1.3180170059204102,
      "learning_rate": 9.095e-05,
      "loss": 0.4064,
      "step": 18200
    },
    {
      "epoch": 1.0192070780602531,
      "grad_norm": 1.463131070137024,
      "learning_rate": 9.0955e-05,
      "loss": 0.5625,
      "step": 18201
    },
    {
      "epoch": 1.0192630753723821,
      "grad_norm": 1.2182964086532593,
      "learning_rate": 9.096e-05,
      "loss": 0.3779,
      "step": 18202
    },
    {
      "epoch": 1.0193190726845112,
      "grad_norm": 1.4279230833053589,
      "learning_rate": 9.0965e-05,
      "loss": 0.4652,
      "step": 18203
    },
    {
      "epoch": 1.0193750699966402,
      "grad_norm": 1.2451354265213013,
      "learning_rate": 9.097e-05,
      "loss": 0.4531,
      "step": 18204
    },
    {
      "epoch": 1.0194310673087692,
      "grad_norm": 1.2272124290466309,
      "learning_rate": 9.0975e-05,
      "loss": 0.3787,
      "step": 18205
    },
    {
      "epoch": 1.0194870646208982,
      "grad_norm": 1.1763030290603638,
      "learning_rate": 9.098000000000001e-05,
      "loss": 0.4008,
      "step": 18206
    },
    {
      "epoch": 1.0195430619330272,
      "grad_norm": 1.3578581809997559,
      "learning_rate": 9.098500000000001e-05,
      "loss": 0.4206,
      "step": 18207
    },
    {
      "epoch": 1.0195990592451563,
      "grad_norm": 1.4249365329742432,
      "learning_rate": 9.099000000000001e-05,
      "loss": 0.4374,
      "step": 18208
    },
    {
      "epoch": 1.0196550565572853,
      "grad_norm": 1.328666090965271,
      "learning_rate": 9.0995e-05,
      "loss": 0.4919,
      "step": 18209
    },
    {
      "epoch": 1.0197110538694143,
      "grad_norm": 1.2589588165283203,
      "learning_rate": 9.1e-05,
      "loss": 0.5186,
      "step": 18210
    },
    {
      "epoch": 1.0197670511815433,
      "grad_norm": 1.3986668586730957,
      "learning_rate": 9.1005e-05,
      "loss": 0.3669,
      "step": 18211
    },
    {
      "epoch": 1.0198230484936723,
      "grad_norm": 1.1582605838775635,
      "learning_rate": 9.101000000000001e-05,
      "loss": 0.4514,
      "step": 18212
    },
    {
      "epoch": 1.0198790458058014,
      "grad_norm": 1.40287184715271,
      "learning_rate": 9.101500000000001e-05,
      "loss": 0.5558,
      "step": 18213
    },
    {
      "epoch": 1.0199350431179304,
      "grad_norm": 1.3524171113967896,
      "learning_rate": 9.102e-05,
      "loss": 0.4443,
      "step": 18214
    },
    {
      "epoch": 1.0199910404300594,
      "grad_norm": 1.5427931547164917,
      "learning_rate": 9.1025e-05,
      "loss": 0.4846,
      "step": 18215
    },
    {
      "epoch": 1.0200470377421884,
      "grad_norm": 1.319544792175293,
      "learning_rate": 9.103e-05,
      "loss": 0.4859,
      "step": 18216
    },
    {
      "epoch": 1.0201030350543174,
      "grad_norm": 1.5592738389968872,
      "learning_rate": 9.1035e-05,
      "loss": 0.4762,
      "step": 18217
    },
    {
      "epoch": 1.0201590323664464,
      "grad_norm": 1.3388237953186035,
      "learning_rate": 9.104000000000001e-05,
      "loss": 0.578,
      "step": 18218
    },
    {
      "epoch": 1.0202150296785755,
      "grad_norm": 1.6388988494873047,
      "learning_rate": 9.1045e-05,
      "loss": 0.4988,
      "step": 18219
    },
    {
      "epoch": 1.0202710269907045,
      "grad_norm": 1.4908218383789062,
      "learning_rate": 9.105e-05,
      "loss": 0.3613,
      "step": 18220
    },
    {
      "epoch": 1.0203270243028335,
      "grad_norm": 1.2812426090240479,
      "learning_rate": 9.1055e-05,
      "loss": 0.2655,
      "step": 18221
    },
    {
      "epoch": 1.0203830216149625,
      "grad_norm": 1.3546313047409058,
      "learning_rate": 9.106e-05,
      "loss": 0.3827,
      "step": 18222
    },
    {
      "epoch": 1.0204390189270915,
      "grad_norm": 1.247186541557312,
      "learning_rate": 9.1065e-05,
      "loss": 0.4516,
      "step": 18223
    },
    {
      "epoch": 1.0204950162392206,
      "grad_norm": 1.2498414516448975,
      "learning_rate": 9.107e-05,
      "loss": 0.4024,
      "step": 18224
    },
    {
      "epoch": 1.0205510135513496,
      "grad_norm": 1.2576067447662354,
      "learning_rate": 9.1075e-05,
      "loss": 0.4612,
      "step": 18225
    },
    {
      "epoch": 1.0206070108634786,
      "grad_norm": 1.5343748331069946,
      "learning_rate": 9.108000000000002e-05,
      "loss": 0.4571,
      "step": 18226
    },
    {
      "epoch": 1.0206630081756076,
      "grad_norm": 1.9127240180969238,
      "learning_rate": 9.108500000000001e-05,
      "loss": 0.369,
      "step": 18227
    },
    {
      "epoch": 1.0207190054877366,
      "grad_norm": 1.4005156755447388,
      "learning_rate": 9.109000000000001e-05,
      "loss": 0.508,
      "step": 18228
    },
    {
      "epoch": 1.0207750027998657,
      "grad_norm": 1.408104658126831,
      "learning_rate": 9.109500000000001e-05,
      "loss": 0.468,
      "step": 18229
    },
    {
      "epoch": 1.0208310001119947,
      "grad_norm": 1.2827550172805786,
      "learning_rate": 9.11e-05,
      "loss": 0.5104,
      "step": 18230
    },
    {
      "epoch": 1.0208869974241237,
      "grad_norm": 1.292176604270935,
      "learning_rate": 9.1105e-05,
      "loss": 0.4796,
      "step": 18231
    },
    {
      "epoch": 1.0209429947362527,
      "grad_norm": 1.4443269968032837,
      "learning_rate": 9.111e-05,
      "loss": 0.3355,
      "step": 18232
    },
    {
      "epoch": 1.0209989920483817,
      "grad_norm": 1.3418875932693481,
      "learning_rate": 9.111500000000001e-05,
      "loss": 0.5003,
      "step": 18233
    },
    {
      "epoch": 1.0210549893605108,
      "grad_norm": 1.4856966733932495,
      "learning_rate": 9.112000000000001e-05,
      "loss": 0.3811,
      "step": 18234
    },
    {
      "epoch": 1.0211109866726398,
      "grad_norm": 1.056575059890747,
      "learning_rate": 9.1125e-05,
      "loss": 0.2906,
      "step": 18235
    },
    {
      "epoch": 1.0211669839847688,
      "grad_norm": 1.321616291999817,
      "learning_rate": 9.113e-05,
      "loss": 0.3858,
      "step": 18236
    },
    {
      "epoch": 1.0212229812968978,
      "grad_norm": 1.1759700775146484,
      "learning_rate": 9.1135e-05,
      "loss": 0.3259,
      "step": 18237
    },
    {
      "epoch": 1.0212789786090268,
      "grad_norm": 1.299067735671997,
      "learning_rate": 9.114e-05,
      "loss": 0.4186,
      "step": 18238
    },
    {
      "epoch": 1.0213349759211559,
      "grad_norm": 1.4101417064666748,
      "learning_rate": 9.114500000000001e-05,
      "loss": 0.4097,
      "step": 18239
    },
    {
      "epoch": 1.0213909732332849,
      "grad_norm": 1.4865729808807373,
      "learning_rate": 9.115e-05,
      "loss": 0.4895,
      "step": 18240
    },
    {
      "epoch": 1.021446970545414,
      "grad_norm": 1.4383394718170166,
      "learning_rate": 9.1155e-05,
      "loss": 0.4237,
      "step": 18241
    },
    {
      "epoch": 1.021502967857543,
      "grad_norm": 1.3762346506118774,
      "learning_rate": 9.116e-05,
      "loss": 0.4098,
      "step": 18242
    },
    {
      "epoch": 1.021558965169672,
      "grad_norm": 1.2840474843978882,
      "learning_rate": 9.1165e-05,
      "loss": 0.396,
      "step": 18243
    },
    {
      "epoch": 1.021614962481801,
      "grad_norm": 1.7581106424331665,
      "learning_rate": 9.117e-05,
      "loss": 0.3805,
      "step": 18244
    },
    {
      "epoch": 1.02167095979393,
      "grad_norm": 1.226104497909546,
      "learning_rate": 9.1175e-05,
      "loss": 0.3699,
      "step": 18245
    },
    {
      "epoch": 1.021726957106059,
      "grad_norm": 1.236308217048645,
      "learning_rate": 9.118e-05,
      "loss": 0.4676,
      "step": 18246
    },
    {
      "epoch": 1.021782954418188,
      "grad_norm": 1.1668379306793213,
      "learning_rate": 9.118500000000002e-05,
      "loss": 0.3804,
      "step": 18247
    },
    {
      "epoch": 1.021838951730317,
      "grad_norm": 1.8728046417236328,
      "learning_rate": 9.119000000000001e-05,
      "loss": 0.5154,
      "step": 18248
    },
    {
      "epoch": 1.021894949042446,
      "grad_norm": 1.1699916124343872,
      "learning_rate": 9.119500000000001e-05,
      "loss": 0.4142,
      "step": 18249
    },
    {
      "epoch": 1.021950946354575,
      "grad_norm": 1.4368023872375488,
      "learning_rate": 9.120000000000001e-05,
      "loss": 0.5335,
      "step": 18250
    },
    {
      "epoch": 1.022006943666704,
      "grad_norm": 1.3821005821228027,
      "learning_rate": 9.1205e-05,
      "loss": 0.417,
      "step": 18251
    },
    {
      "epoch": 1.022062940978833,
      "grad_norm": 1.9109565019607544,
      "learning_rate": 9.121e-05,
      "loss": 0.5542,
      "step": 18252
    },
    {
      "epoch": 1.0221189382909621,
      "grad_norm": 1.257783055305481,
      "learning_rate": 9.1215e-05,
      "loss": 0.3066,
      "step": 18253
    },
    {
      "epoch": 1.0221749356030911,
      "grad_norm": 1.4408705234527588,
      "learning_rate": 9.122000000000001e-05,
      "loss": 0.4391,
      "step": 18254
    },
    {
      "epoch": 1.0222309329152202,
      "grad_norm": 1.3678799867630005,
      "learning_rate": 9.122500000000001e-05,
      "loss": 0.3597,
      "step": 18255
    },
    {
      "epoch": 1.0222869302273492,
      "grad_norm": 1.3051496744155884,
      "learning_rate": 9.123e-05,
      "loss": 0.4144,
      "step": 18256
    },
    {
      "epoch": 1.0223429275394782,
      "grad_norm": 1.2545008659362793,
      "learning_rate": 9.1235e-05,
      "loss": 0.4483,
      "step": 18257
    },
    {
      "epoch": 1.0223989248516072,
      "grad_norm": 1.4460846185684204,
      "learning_rate": 9.124e-05,
      "loss": 0.3909,
      "step": 18258
    },
    {
      "epoch": 1.0224549221637362,
      "grad_norm": 1.2341158390045166,
      "learning_rate": 9.1245e-05,
      "loss": 0.387,
      "step": 18259
    },
    {
      "epoch": 1.0225109194758653,
      "grad_norm": 1.3547271490097046,
      "learning_rate": 9.125e-05,
      "loss": 0.4874,
      "step": 18260
    },
    {
      "epoch": 1.0225669167879943,
      "grad_norm": 1.2583049535751343,
      "learning_rate": 9.1255e-05,
      "loss": 0.3336,
      "step": 18261
    },
    {
      "epoch": 1.0226229141001233,
      "grad_norm": 1.4579052925109863,
      "learning_rate": 9.126e-05,
      "loss": 0.4597,
      "step": 18262
    },
    {
      "epoch": 1.0226789114122523,
      "grad_norm": 1.7459715604782104,
      "learning_rate": 9.1265e-05,
      "loss": 0.4076,
      "step": 18263
    },
    {
      "epoch": 1.0227349087243813,
      "grad_norm": 1.2544258832931519,
      "learning_rate": 9.127e-05,
      "loss": 0.338,
      "step": 18264
    },
    {
      "epoch": 1.0227909060365103,
      "grad_norm": 2.3433997631073,
      "learning_rate": 9.1275e-05,
      "loss": 0.5613,
      "step": 18265
    },
    {
      "epoch": 1.0228469033486394,
      "grad_norm": 1.5074681043624878,
      "learning_rate": 9.128e-05,
      "loss": 0.4918,
      "step": 18266
    },
    {
      "epoch": 1.0229029006607684,
      "grad_norm": 1.2791783809661865,
      "learning_rate": 9.1285e-05,
      "loss": 0.4341,
      "step": 18267
    },
    {
      "epoch": 1.0229588979728974,
      "grad_norm": 1.2252360582351685,
      "learning_rate": 9.129000000000002e-05,
      "loss": 0.3463,
      "step": 18268
    },
    {
      "epoch": 1.0230148952850264,
      "grad_norm": 1.2673494815826416,
      "learning_rate": 9.129500000000001e-05,
      "loss": 0.3853,
      "step": 18269
    },
    {
      "epoch": 1.0230708925971554,
      "grad_norm": 1.4124587774276733,
      "learning_rate": 9.130000000000001e-05,
      "loss": 0.3272,
      "step": 18270
    },
    {
      "epoch": 1.0231268899092845,
      "grad_norm": 1.2844294309616089,
      "learning_rate": 9.130500000000001e-05,
      "loss": 0.3963,
      "step": 18271
    },
    {
      "epoch": 1.0231828872214135,
      "grad_norm": 1.359400749206543,
      "learning_rate": 9.131e-05,
      "loss": 0.5973,
      "step": 18272
    },
    {
      "epoch": 1.0232388845335423,
      "grad_norm": 1.4733316898345947,
      "learning_rate": 9.1315e-05,
      "loss": 0.5078,
      "step": 18273
    },
    {
      "epoch": 1.0232948818456713,
      "grad_norm": 1.5604074001312256,
      "learning_rate": 9.132e-05,
      "loss": 0.6136,
      "step": 18274
    },
    {
      "epoch": 1.0233508791578003,
      "grad_norm": 1.272163987159729,
      "learning_rate": 9.132500000000001e-05,
      "loss": 0.3756,
      "step": 18275
    },
    {
      "epoch": 1.0234068764699293,
      "grad_norm": 1.4646979570388794,
      "learning_rate": 9.133000000000001e-05,
      "loss": 0.5123,
      "step": 18276
    },
    {
      "epoch": 1.0234628737820584,
      "grad_norm": 1.6235682964324951,
      "learning_rate": 9.1335e-05,
      "loss": 0.4594,
      "step": 18277
    },
    {
      "epoch": 1.0235188710941874,
      "grad_norm": 1.4288482666015625,
      "learning_rate": 9.134e-05,
      "loss": 0.4221,
      "step": 18278
    },
    {
      "epoch": 1.0235748684063164,
      "grad_norm": 1.1828737258911133,
      "learning_rate": 9.1345e-05,
      "loss": 0.3298,
      "step": 18279
    },
    {
      "epoch": 1.0236308657184454,
      "grad_norm": 1.3354418277740479,
      "learning_rate": 9.135e-05,
      "loss": 0.4301,
      "step": 18280
    },
    {
      "epoch": 1.0236868630305744,
      "grad_norm": 1.3682681322097778,
      "learning_rate": 9.1355e-05,
      "loss": 0.5212,
      "step": 18281
    },
    {
      "epoch": 1.0237428603427035,
      "grad_norm": 1.308532953262329,
      "learning_rate": 9.136e-05,
      "loss": 0.4403,
      "step": 18282
    },
    {
      "epoch": 1.0237988576548325,
      "grad_norm": 1.3649508953094482,
      "learning_rate": 9.1365e-05,
      "loss": 0.3935,
      "step": 18283
    },
    {
      "epoch": 1.0238548549669615,
      "grad_norm": 1.3734993934631348,
      "learning_rate": 9.137e-05,
      "loss": 0.4426,
      "step": 18284
    },
    {
      "epoch": 1.0239108522790905,
      "grad_norm": 1.473650336265564,
      "learning_rate": 9.1375e-05,
      "loss": 0.5127,
      "step": 18285
    },
    {
      "epoch": 1.0239668495912195,
      "grad_norm": 1.4458955526351929,
      "learning_rate": 9.138e-05,
      "loss": 0.4166,
      "step": 18286
    },
    {
      "epoch": 1.0240228469033485,
      "grad_norm": 1.4608911275863647,
      "learning_rate": 9.138500000000001e-05,
      "loss": 0.6066,
      "step": 18287
    },
    {
      "epoch": 1.0240788442154776,
      "grad_norm": 1.498123049736023,
      "learning_rate": 9.139e-05,
      "loss": 0.5515,
      "step": 18288
    },
    {
      "epoch": 1.0241348415276066,
      "grad_norm": 1.2164201736450195,
      "learning_rate": 9.1395e-05,
      "loss": 0.4468,
      "step": 18289
    },
    {
      "epoch": 1.0241908388397356,
      "grad_norm": 1.4614959955215454,
      "learning_rate": 9.140000000000001e-05,
      "loss": 0.4695,
      "step": 18290
    },
    {
      "epoch": 1.0242468361518646,
      "grad_norm": 1.2819448709487915,
      "learning_rate": 9.140500000000001e-05,
      "loss": 0.302,
      "step": 18291
    },
    {
      "epoch": 1.0243028334639936,
      "grad_norm": 1.484433650970459,
      "learning_rate": 9.141000000000001e-05,
      "loss": 0.418,
      "step": 18292
    },
    {
      "epoch": 1.0243588307761227,
      "grad_norm": 1.1295392513275146,
      "learning_rate": 9.1415e-05,
      "loss": 0.377,
      "step": 18293
    },
    {
      "epoch": 1.0244148280882517,
      "grad_norm": 1.4290921688079834,
      "learning_rate": 9.142e-05,
      "loss": 0.4408,
      "step": 18294
    },
    {
      "epoch": 1.0244708254003807,
      "grad_norm": 1.4190647602081299,
      "learning_rate": 9.1425e-05,
      "loss": 0.4337,
      "step": 18295
    },
    {
      "epoch": 1.0245268227125097,
      "grad_norm": 1.1803679466247559,
      "learning_rate": 9.143000000000001e-05,
      "loss": 0.4562,
      "step": 18296
    },
    {
      "epoch": 1.0245828200246387,
      "grad_norm": 1.1149933338165283,
      "learning_rate": 9.143500000000001e-05,
      "loss": 0.3951,
      "step": 18297
    },
    {
      "epoch": 1.0246388173367678,
      "grad_norm": 1.1982587575912476,
      "learning_rate": 9.144e-05,
      "loss": 0.3147,
      "step": 18298
    },
    {
      "epoch": 1.0246948146488968,
      "grad_norm": 1.5093990564346313,
      "learning_rate": 9.1445e-05,
      "loss": 0.65,
      "step": 18299
    },
    {
      "epoch": 1.0247508119610258,
      "grad_norm": 1.3103488683700562,
      "learning_rate": 9.145e-05,
      "loss": 0.3724,
      "step": 18300
    },
    {
      "epoch": 1.0248068092731548,
      "grad_norm": 1.3719797134399414,
      "learning_rate": 9.1455e-05,
      "loss": 0.3564,
      "step": 18301
    },
    {
      "epoch": 1.0248628065852838,
      "grad_norm": 1.3001863956451416,
      "learning_rate": 9.146e-05,
      "loss": 0.5071,
      "step": 18302
    },
    {
      "epoch": 1.0249188038974129,
      "grad_norm": 1.3157297372817993,
      "learning_rate": 9.1465e-05,
      "loss": 0.4178,
      "step": 18303
    },
    {
      "epoch": 1.0249748012095419,
      "grad_norm": 1.3867977857589722,
      "learning_rate": 9.147e-05,
      "loss": 0.4625,
      "step": 18304
    },
    {
      "epoch": 1.025030798521671,
      "grad_norm": 1.1817615032196045,
      "learning_rate": 9.1475e-05,
      "loss": 0.397,
      "step": 18305
    },
    {
      "epoch": 1.0250867958338,
      "grad_norm": 1.3469206094741821,
      "learning_rate": 9.148e-05,
      "loss": 0.4254,
      "step": 18306
    },
    {
      "epoch": 1.025142793145929,
      "grad_norm": 1.2162150144577026,
      "learning_rate": 9.148500000000001e-05,
      "loss": 0.4049,
      "step": 18307
    },
    {
      "epoch": 1.025198790458058,
      "grad_norm": 1.859230399131775,
      "learning_rate": 9.149e-05,
      "loss": 0.6715,
      "step": 18308
    },
    {
      "epoch": 1.025254787770187,
      "grad_norm": 1.5233372449874878,
      "learning_rate": 9.1495e-05,
      "loss": 0.5154,
      "step": 18309
    },
    {
      "epoch": 1.025310785082316,
      "grad_norm": 1.410709023475647,
      "learning_rate": 9.15e-05,
      "loss": 0.4479,
      "step": 18310
    },
    {
      "epoch": 1.025366782394445,
      "grad_norm": 1.3257741928100586,
      "learning_rate": 9.150500000000001e-05,
      "loss": 0.5305,
      "step": 18311
    },
    {
      "epoch": 1.025422779706574,
      "grad_norm": 1.2185670137405396,
      "learning_rate": 9.151000000000001e-05,
      "loss": 0.4023,
      "step": 18312
    },
    {
      "epoch": 1.025478777018703,
      "grad_norm": 1.3654381036758423,
      "learning_rate": 9.151500000000001e-05,
      "loss": 0.4937,
      "step": 18313
    },
    {
      "epoch": 1.025534774330832,
      "grad_norm": 1.2604769468307495,
      "learning_rate": 9.152e-05,
      "loss": 0.3389,
      "step": 18314
    },
    {
      "epoch": 1.025590771642961,
      "grad_norm": 1.3107566833496094,
      "learning_rate": 9.1525e-05,
      "loss": 0.399,
      "step": 18315
    },
    {
      "epoch": 1.02564676895509,
      "grad_norm": 1.2344932556152344,
      "learning_rate": 9.153e-05,
      "loss": 0.388,
      "step": 18316
    },
    {
      "epoch": 1.0257027662672191,
      "grad_norm": 1.4607726335525513,
      "learning_rate": 9.153500000000001e-05,
      "loss": 0.4296,
      "step": 18317
    },
    {
      "epoch": 1.0257587635793481,
      "grad_norm": 1.532738208770752,
      "learning_rate": 9.154000000000001e-05,
      "loss": 0.3561,
      "step": 18318
    },
    {
      "epoch": 1.0258147608914772,
      "grad_norm": 1.3724843263626099,
      "learning_rate": 9.1545e-05,
      "loss": 0.3672,
      "step": 18319
    },
    {
      "epoch": 1.0258707582036062,
      "grad_norm": 1.2287768125534058,
      "learning_rate": 9.155e-05,
      "loss": 0.4009,
      "step": 18320
    },
    {
      "epoch": 1.0259267555157352,
      "grad_norm": 1.1697520017623901,
      "learning_rate": 9.1555e-05,
      "loss": 0.3992,
      "step": 18321
    },
    {
      "epoch": 1.0259827528278642,
      "grad_norm": 1.450138807296753,
      "learning_rate": 9.156e-05,
      "loss": 0.5989,
      "step": 18322
    },
    {
      "epoch": 1.0260387501399932,
      "grad_norm": 1.4806028604507446,
      "learning_rate": 9.1565e-05,
      "loss": 0.4256,
      "step": 18323
    },
    {
      "epoch": 1.0260947474521223,
      "grad_norm": 1.3607051372528076,
      "learning_rate": 9.157e-05,
      "loss": 0.4582,
      "step": 18324
    },
    {
      "epoch": 1.0261507447642513,
      "grad_norm": 1.2938125133514404,
      "learning_rate": 9.1575e-05,
      "loss": 0.4895,
      "step": 18325
    },
    {
      "epoch": 1.0262067420763803,
      "grad_norm": 1.1564923524856567,
      "learning_rate": 9.158e-05,
      "loss": 0.347,
      "step": 18326
    },
    {
      "epoch": 1.0262627393885093,
      "grad_norm": 1.0445786714553833,
      "learning_rate": 9.158500000000001e-05,
      "loss": 0.3474,
      "step": 18327
    },
    {
      "epoch": 1.0263187367006383,
      "grad_norm": 1.220381259918213,
      "learning_rate": 9.159000000000001e-05,
      "loss": 0.3853,
      "step": 18328
    },
    {
      "epoch": 1.0263747340127674,
      "grad_norm": 1.3459187746047974,
      "learning_rate": 9.1595e-05,
      "loss": 0.3873,
      "step": 18329
    },
    {
      "epoch": 1.0264307313248964,
      "grad_norm": 1.1996020078659058,
      "learning_rate": 9.16e-05,
      "loss": 0.3603,
      "step": 18330
    },
    {
      "epoch": 1.0264867286370254,
      "grad_norm": 1.67203688621521,
      "learning_rate": 9.1605e-05,
      "loss": 0.5885,
      "step": 18331
    },
    {
      "epoch": 1.0265427259491544,
      "grad_norm": 1.098457932472229,
      "learning_rate": 9.161000000000001e-05,
      "loss": 0.3171,
      "step": 18332
    },
    {
      "epoch": 1.0265987232612834,
      "grad_norm": 1.4290902614593506,
      "learning_rate": 9.161500000000001e-05,
      "loss": 0.4719,
      "step": 18333
    },
    {
      "epoch": 1.0266547205734124,
      "grad_norm": 1.4965934753417969,
      "learning_rate": 9.162000000000001e-05,
      "loss": 0.43,
      "step": 18334
    },
    {
      "epoch": 1.0267107178855415,
      "grad_norm": 1.3025615215301514,
      "learning_rate": 9.1625e-05,
      "loss": 0.353,
      "step": 18335
    },
    {
      "epoch": 1.0267667151976705,
      "grad_norm": 1.5858601331710815,
      "learning_rate": 9.163e-05,
      "loss": 0.5798,
      "step": 18336
    },
    {
      "epoch": 1.0268227125097995,
      "grad_norm": 1.6514794826507568,
      "learning_rate": 9.1635e-05,
      "loss": 0.5493,
      "step": 18337
    },
    {
      "epoch": 1.0268787098219285,
      "grad_norm": 1.5772913694381714,
      "learning_rate": 9.164000000000001e-05,
      "loss": 0.4204,
      "step": 18338
    },
    {
      "epoch": 1.0269347071340575,
      "grad_norm": 1.5542999505996704,
      "learning_rate": 9.164500000000001e-05,
      "loss": 0.4284,
      "step": 18339
    },
    {
      "epoch": 1.0269907044461866,
      "grad_norm": 1.5654528141021729,
      "learning_rate": 9.165e-05,
      "loss": 0.3435,
      "step": 18340
    },
    {
      "epoch": 1.0270467017583156,
      "grad_norm": 1.1904045343399048,
      "learning_rate": 9.1655e-05,
      "loss": 0.4665,
      "step": 18341
    },
    {
      "epoch": 1.0271026990704446,
      "grad_norm": 1.2764575481414795,
      "learning_rate": 9.166e-05,
      "loss": 0.3937,
      "step": 18342
    },
    {
      "epoch": 1.0271586963825736,
      "grad_norm": 1.2397255897521973,
      "learning_rate": 9.1665e-05,
      "loss": 0.4035,
      "step": 18343
    },
    {
      "epoch": 1.0272146936947026,
      "grad_norm": 1.5256158113479614,
      "learning_rate": 9.167e-05,
      "loss": 0.4795,
      "step": 18344
    },
    {
      "epoch": 1.0272706910068317,
      "grad_norm": 1.3618723154067993,
      "learning_rate": 9.1675e-05,
      "loss": 0.5174,
      "step": 18345
    },
    {
      "epoch": 1.0273266883189607,
      "grad_norm": 1.2115530967712402,
      "learning_rate": 9.168e-05,
      "loss": 0.338,
      "step": 18346
    },
    {
      "epoch": 1.0273826856310897,
      "grad_norm": 1.500746250152588,
      "learning_rate": 9.168500000000001e-05,
      "loss": 0.4163,
      "step": 18347
    },
    {
      "epoch": 1.0274386829432187,
      "grad_norm": 1.482271432876587,
      "learning_rate": 9.169000000000001e-05,
      "loss": 0.4528,
      "step": 18348
    },
    {
      "epoch": 1.0274946802553477,
      "grad_norm": 1.9786421060562134,
      "learning_rate": 9.169500000000001e-05,
      "loss": 0.451,
      "step": 18349
    },
    {
      "epoch": 1.0275506775674768,
      "grad_norm": 1.3467090129852295,
      "learning_rate": 9.17e-05,
      "loss": 0.3977,
      "step": 18350
    },
    {
      "epoch": 1.0276066748796058,
      "grad_norm": 1.366949439048767,
      "learning_rate": 9.1705e-05,
      "loss": 0.5047,
      "step": 18351
    },
    {
      "epoch": 1.0276626721917348,
      "grad_norm": 1.2228416204452515,
      "learning_rate": 9.171e-05,
      "loss": 0.3472,
      "step": 18352
    },
    {
      "epoch": 1.0277186695038638,
      "grad_norm": 1.6578097343444824,
      "learning_rate": 9.171500000000001e-05,
      "loss": 0.4835,
      "step": 18353
    },
    {
      "epoch": 1.0277746668159928,
      "grad_norm": 1.2654086351394653,
      "learning_rate": 9.172000000000001e-05,
      "loss": 0.3836,
      "step": 18354
    },
    {
      "epoch": 1.0278306641281219,
      "grad_norm": 1.4109214544296265,
      "learning_rate": 9.172500000000001e-05,
      "loss": 0.487,
      "step": 18355
    },
    {
      "epoch": 1.0278866614402509,
      "grad_norm": 1.3260384798049927,
      "learning_rate": 9.173e-05,
      "loss": 0.4185,
      "step": 18356
    },
    {
      "epoch": 1.02794265875238,
      "grad_norm": 1.336680293083191,
      "learning_rate": 9.1735e-05,
      "loss": 0.3993,
      "step": 18357
    },
    {
      "epoch": 1.027998656064509,
      "grad_norm": 1.0807775259017944,
      "learning_rate": 9.174e-05,
      "loss": 0.3301,
      "step": 18358
    },
    {
      "epoch": 1.028054653376638,
      "grad_norm": 1.4308130741119385,
      "learning_rate": 9.1745e-05,
      "loss": 0.5478,
      "step": 18359
    },
    {
      "epoch": 1.028110650688767,
      "grad_norm": 1.6996850967407227,
      "learning_rate": 9.175000000000001e-05,
      "loss": 0.5778,
      "step": 18360
    },
    {
      "epoch": 1.028166648000896,
      "grad_norm": 1.3647295236587524,
      "learning_rate": 9.1755e-05,
      "loss": 0.4445,
      "step": 18361
    },
    {
      "epoch": 1.028222645313025,
      "grad_norm": 1.6652171611785889,
      "learning_rate": 9.176e-05,
      "loss": 0.556,
      "step": 18362
    },
    {
      "epoch": 1.028278642625154,
      "grad_norm": 1.6018441915512085,
      "learning_rate": 9.1765e-05,
      "loss": 0.5107,
      "step": 18363
    },
    {
      "epoch": 1.028334639937283,
      "grad_norm": 1.3393938541412354,
      "learning_rate": 9.177e-05,
      "loss": 0.4032,
      "step": 18364
    },
    {
      "epoch": 1.028390637249412,
      "grad_norm": 1.1796034574508667,
      "learning_rate": 9.1775e-05,
      "loss": 0.4536,
      "step": 18365
    },
    {
      "epoch": 1.028446634561541,
      "grad_norm": 1.275696039199829,
      "learning_rate": 9.178e-05,
      "loss": 0.4154,
      "step": 18366
    },
    {
      "epoch": 1.02850263187367,
      "grad_norm": 3.7940006256103516,
      "learning_rate": 9.178500000000002e-05,
      "loss": 0.5143,
      "step": 18367
    },
    {
      "epoch": 1.028558629185799,
      "grad_norm": 1.4529520273208618,
      "learning_rate": 9.179000000000001e-05,
      "loss": 0.5135,
      "step": 18368
    },
    {
      "epoch": 1.0286146264979281,
      "grad_norm": 1.36226487159729,
      "learning_rate": 9.179500000000001e-05,
      "loss": 0.4201,
      "step": 18369
    },
    {
      "epoch": 1.0286706238100571,
      "grad_norm": 1.2332403659820557,
      "learning_rate": 9.180000000000001e-05,
      "loss": 0.3876,
      "step": 18370
    },
    {
      "epoch": 1.0287266211221862,
      "grad_norm": 1.2992181777954102,
      "learning_rate": 9.1805e-05,
      "loss": 0.6808,
      "step": 18371
    },
    {
      "epoch": 1.0287826184343152,
      "grad_norm": 1.4992225170135498,
      "learning_rate": 9.181e-05,
      "loss": 0.3968,
      "step": 18372
    },
    {
      "epoch": 1.0288386157464442,
      "grad_norm": 1.1045231819152832,
      "learning_rate": 9.1815e-05,
      "loss": 0.3763,
      "step": 18373
    },
    {
      "epoch": 1.0288946130585732,
      "grad_norm": 1.3332529067993164,
      "learning_rate": 9.182000000000001e-05,
      "loss": 0.3534,
      "step": 18374
    },
    {
      "epoch": 1.0289506103707022,
      "grad_norm": 1.446587085723877,
      "learning_rate": 9.182500000000001e-05,
      "loss": 0.3957,
      "step": 18375
    },
    {
      "epoch": 1.0290066076828313,
      "grad_norm": 1.5129408836364746,
      "learning_rate": 9.183000000000001e-05,
      "loss": 0.5899,
      "step": 18376
    },
    {
      "epoch": 1.0290626049949603,
      "grad_norm": 1.2569142580032349,
      "learning_rate": 9.1835e-05,
      "loss": 0.3884,
      "step": 18377
    },
    {
      "epoch": 1.0291186023070893,
      "grad_norm": 1.3488404750823975,
      "learning_rate": 9.184e-05,
      "loss": 0.4009,
      "step": 18378
    },
    {
      "epoch": 1.0291745996192183,
      "grad_norm": 1.46926748752594,
      "learning_rate": 9.1845e-05,
      "loss": 0.4344,
      "step": 18379
    },
    {
      "epoch": 1.0292305969313473,
      "grad_norm": 1.3317769765853882,
      "learning_rate": 9.185e-05,
      "loss": 0.4538,
      "step": 18380
    },
    {
      "epoch": 1.0292865942434763,
      "grad_norm": 1.6697031259536743,
      "learning_rate": 9.185500000000001e-05,
      "loss": 0.3738,
      "step": 18381
    },
    {
      "epoch": 1.0293425915556054,
      "grad_norm": 1.295625925064087,
      "learning_rate": 9.186e-05,
      "loss": 0.5232,
      "step": 18382
    },
    {
      "epoch": 1.0293985888677344,
      "grad_norm": 1.269540786743164,
      "learning_rate": 9.1865e-05,
      "loss": 0.4541,
      "step": 18383
    },
    {
      "epoch": 1.0294545861798634,
      "grad_norm": 1.8306798934936523,
      "learning_rate": 9.187e-05,
      "loss": 0.3955,
      "step": 18384
    },
    {
      "epoch": 1.0295105834919924,
      "grad_norm": 1.1614495515823364,
      "learning_rate": 9.1875e-05,
      "loss": 0.3892,
      "step": 18385
    },
    {
      "epoch": 1.0295665808041214,
      "grad_norm": 1.5643037557601929,
      "learning_rate": 9.188e-05,
      "loss": 0.398,
      "step": 18386
    },
    {
      "epoch": 1.0296225781162505,
      "grad_norm": 1.1431219577789307,
      "learning_rate": 9.1885e-05,
      "loss": 0.4726,
      "step": 18387
    },
    {
      "epoch": 1.0296785754283795,
      "grad_norm": 1.1050772666931152,
      "learning_rate": 9.189e-05,
      "loss": 0.3813,
      "step": 18388
    },
    {
      "epoch": 1.0297345727405085,
      "grad_norm": 3.193002223968506,
      "learning_rate": 9.189500000000001e-05,
      "loss": 0.524,
      "step": 18389
    },
    {
      "epoch": 1.0297905700526375,
      "grad_norm": 1.2781938314437866,
      "learning_rate": 9.190000000000001e-05,
      "loss": 0.3412,
      "step": 18390
    },
    {
      "epoch": 1.0298465673647665,
      "grad_norm": 1.3508274555206299,
      "learning_rate": 9.190500000000001e-05,
      "loss": 0.5309,
      "step": 18391
    },
    {
      "epoch": 1.0299025646768956,
      "grad_norm": 1.6212399005889893,
      "learning_rate": 9.191e-05,
      "loss": 0.5822,
      "step": 18392
    },
    {
      "epoch": 1.0299585619890246,
      "grad_norm": 1.4462803602218628,
      "learning_rate": 9.1915e-05,
      "loss": 0.2874,
      "step": 18393
    },
    {
      "epoch": 1.0300145593011536,
      "grad_norm": 1.1816555261611938,
      "learning_rate": 9.192e-05,
      "loss": 0.4075,
      "step": 18394
    },
    {
      "epoch": 1.0300705566132826,
      "grad_norm": 1.5166535377502441,
      "learning_rate": 9.192500000000001e-05,
      "loss": 0.389,
      "step": 18395
    },
    {
      "epoch": 1.0301265539254116,
      "grad_norm": 1.4264452457427979,
      "learning_rate": 9.193000000000001e-05,
      "loss": 0.4658,
      "step": 18396
    },
    {
      "epoch": 1.0301825512375407,
      "grad_norm": 1.33519446849823,
      "learning_rate": 9.1935e-05,
      "loss": 0.5087,
      "step": 18397
    },
    {
      "epoch": 1.0302385485496697,
      "grad_norm": 1.2397849559783936,
      "learning_rate": 9.194e-05,
      "loss": 0.4526,
      "step": 18398
    },
    {
      "epoch": 1.0302945458617987,
      "grad_norm": 1.933770775794983,
      "learning_rate": 9.1945e-05,
      "loss": 0.3647,
      "step": 18399
    },
    {
      "epoch": 1.0303505431739277,
      "grad_norm": 1.5117770433425903,
      "learning_rate": 9.195e-05,
      "loss": 0.4367,
      "step": 18400
    },
    {
      "epoch": 1.0304065404860567,
      "grad_norm": 1.671377182006836,
      "learning_rate": 9.1955e-05,
      "loss": 0.3343,
      "step": 18401
    },
    {
      "epoch": 1.0304625377981858,
      "grad_norm": 1.203366756439209,
      "learning_rate": 9.196000000000001e-05,
      "loss": 0.3819,
      "step": 18402
    },
    {
      "epoch": 1.0305185351103148,
      "grad_norm": 1.5717040300369263,
      "learning_rate": 9.1965e-05,
      "loss": 0.4543,
      "step": 18403
    },
    {
      "epoch": 1.0305745324224438,
      "grad_norm": 1.1201183795928955,
      "learning_rate": 9.197e-05,
      "loss": 0.2794,
      "step": 18404
    },
    {
      "epoch": 1.0306305297345728,
      "grad_norm": 1.88822603225708,
      "learning_rate": 9.1975e-05,
      "loss": 0.4396,
      "step": 18405
    },
    {
      "epoch": 1.0306865270467018,
      "grad_norm": 1.2081702947616577,
      "learning_rate": 9.198e-05,
      "loss": 0.3072,
      "step": 18406
    },
    {
      "epoch": 1.0307425243588308,
      "grad_norm": 1.5484460592269897,
      "learning_rate": 9.1985e-05,
      "loss": 0.5021,
      "step": 18407
    },
    {
      "epoch": 1.0307985216709599,
      "grad_norm": 1.197572946548462,
      "learning_rate": 9.199e-05,
      "loss": 0.3615,
      "step": 18408
    },
    {
      "epoch": 1.0308545189830889,
      "grad_norm": 1.2413281202316284,
      "learning_rate": 9.1995e-05,
      "loss": 0.4234,
      "step": 18409
    },
    {
      "epoch": 1.030910516295218,
      "grad_norm": 1.3042575120925903,
      "learning_rate": 9.200000000000001e-05,
      "loss": 0.4667,
      "step": 18410
    },
    {
      "epoch": 1.030966513607347,
      "grad_norm": 1.9163047075271606,
      "learning_rate": 9.200500000000001e-05,
      "loss": 0.2956,
      "step": 18411
    },
    {
      "epoch": 1.031022510919476,
      "grad_norm": 1.655187726020813,
      "learning_rate": 9.201000000000001e-05,
      "loss": 0.3838,
      "step": 18412
    },
    {
      "epoch": 1.031078508231605,
      "grad_norm": 1.4335438013076782,
      "learning_rate": 9.2015e-05,
      "loss": 0.4645,
      "step": 18413
    },
    {
      "epoch": 1.031134505543734,
      "grad_norm": 1.4914640188217163,
      "learning_rate": 9.202e-05,
      "loss": 0.403,
      "step": 18414
    },
    {
      "epoch": 1.031190502855863,
      "grad_norm": 1.4218559265136719,
      "learning_rate": 9.2025e-05,
      "loss": 0.3799,
      "step": 18415
    },
    {
      "epoch": 1.031246500167992,
      "grad_norm": 1.4433684349060059,
      "learning_rate": 9.203000000000001e-05,
      "loss": 0.4391,
      "step": 18416
    },
    {
      "epoch": 1.031302497480121,
      "grad_norm": 1.1900036334991455,
      "learning_rate": 9.203500000000001e-05,
      "loss": 0.5269,
      "step": 18417
    },
    {
      "epoch": 1.03135849479225,
      "grad_norm": 1.30314040184021,
      "learning_rate": 9.204e-05,
      "loss": 0.3646,
      "step": 18418
    },
    {
      "epoch": 1.031414492104379,
      "grad_norm": 1.6013752222061157,
      "learning_rate": 9.2045e-05,
      "loss": 0.3689,
      "step": 18419
    },
    {
      "epoch": 1.031470489416508,
      "grad_norm": 1.0355218648910522,
      "learning_rate": 9.205e-05,
      "loss": 0.3397,
      "step": 18420
    },
    {
      "epoch": 1.0315264867286371,
      "grad_norm": 1.4996814727783203,
      "learning_rate": 9.2055e-05,
      "loss": 0.3549,
      "step": 18421
    },
    {
      "epoch": 1.0315824840407661,
      "grad_norm": 1.2774845361709595,
      "learning_rate": 9.206e-05,
      "loss": 0.4775,
      "step": 18422
    },
    {
      "epoch": 1.0316384813528952,
      "grad_norm": 1.398349642753601,
      "learning_rate": 9.206500000000001e-05,
      "loss": 0.3908,
      "step": 18423
    },
    {
      "epoch": 1.0316944786650242,
      "grad_norm": 1.645747423171997,
      "learning_rate": 9.207e-05,
      "loss": 0.5009,
      "step": 18424
    },
    {
      "epoch": 1.0317504759771532,
      "grad_norm": 1.55572509765625,
      "learning_rate": 9.2075e-05,
      "loss": 0.5306,
      "step": 18425
    },
    {
      "epoch": 1.0318064732892822,
      "grad_norm": 1.4952770471572876,
      "learning_rate": 9.208e-05,
      "loss": 0.4225,
      "step": 18426
    },
    {
      "epoch": 1.0318624706014112,
      "grad_norm": 1.4356719255447388,
      "learning_rate": 9.2085e-05,
      "loss": 0.5911,
      "step": 18427
    },
    {
      "epoch": 1.0319184679135402,
      "grad_norm": 1.3004816770553589,
      "learning_rate": 9.209000000000001e-05,
      "loss": 0.4836,
      "step": 18428
    },
    {
      "epoch": 1.0319744652256693,
      "grad_norm": 1.3479303121566772,
      "learning_rate": 9.2095e-05,
      "loss": 0.4258,
      "step": 18429
    },
    {
      "epoch": 1.0320304625377983,
      "grad_norm": 1.1808431148529053,
      "learning_rate": 9.21e-05,
      "loss": 0.4005,
      "step": 18430
    },
    {
      "epoch": 1.0320864598499273,
      "grad_norm": 1.135550618171692,
      "learning_rate": 9.210500000000001e-05,
      "loss": 0.3771,
      "step": 18431
    },
    {
      "epoch": 1.0321424571620563,
      "grad_norm": 1.316703200340271,
      "learning_rate": 9.211000000000001e-05,
      "loss": 0.4245,
      "step": 18432
    },
    {
      "epoch": 1.0321984544741853,
      "grad_norm": 1.4228503704071045,
      "learning_rate": 9.211500000000001e-05,
      "loss": 0.4636,
      "step": 18433
    },
    {
      "epoch": 1.0322544517863144,
      "grad_norm": 1.084437370300293,
      "learning_rate": 9.212e-05,
      "loss": 0.3343,
      "step": 18434
    },
    {
      "epoch": 1.0323104490984432,
      "grad_norm": 1.2737746238708496,
      "learning_rate": 9.2125e-05,
      "loss": 0.4865,
      "step": 18435
    },
    {
      "epoch": 1.0323664464105722,
      "grad_norm": 1.5287773609161377,
      "learning_rate": 9.213e-05,
      "loss": 0.5212,
      "step": 18436
    },
    {
      "epoch": 1.0324224437227012,
      "grad_norm": 1.218571662902832,
      "learning_rate": 9.2135e-05,
      "loss": 0.3727,
      "step": 18437
    },
    {
      "epoch": 1.0324784410348302,
      "grad_norm": 1.2497469186782837,
      "learning_rate": 9.214000000000001e-05,
      "loss": 0.4716,
      "step": 18438
    },
    {
      "epoch": 1.0325344383469592,
      "grad_norm": 1.2556761503219604,
      "learning_rate": 9.2145e-05,
      "loss": 0.5003,
      "step": 18439
    },
    {
      "epoch": 1.0325904356590883,
      "grad_norm": 1.5302544832229614,
      "learning_rate": 9.215e-05,
      "loss": 0.353,
      "step": 18440
    },
    {
      "epoch": 1.0326464329712173,
      "grad_norm": 1.4618993997573853,
      "learning_rate": 9.2155e-05,
      "loss": 0.5285,
      "step": 18441
    },
    {
      "epoch": 1.0327024302833463,
      "grad_norm": 1.3802485466003418,
      "learning_rate": 9.216e-05,
      "loss": 0.4933,
      "step": 18442
    },
    {
      "epoch": 1.0327584275954753,
      "grad_norm": 1.2149531841278076,
      "learning_rate": 9.2165e-05,
      "loss": 0.4026,
      "step": 18443
    },
    {
      "epoch": 1.0328144249076043,
      "grad_norm": 1.3150397539138794,
      "learning_rate": 9.217000000000001e-05,
      "loss": 0.5103,
      "step": 18444
    },
    {
      "epoch": 1.0328704222197334,
      "grad_norm": 1.1010187864303589,
      "learning_rate": 9.2175e-05,
      "loss": 0.4004,
      "step": 18445
    },
    {
      "epoch": 1.0329264195318624,
      "grad_norm": 1.3528146743774414,
      "learning_rate": 9.218e-05,
      "loss": 0.4566,
      "step": 18446
    },
    {
      "epoch": 1.0329824168439914,
      "grad_norm": 1.2506998777389526,
      "learning_rate": 9.2185e-05,
      "loss": 0.378,
      "step": 18447
    },
    {
      "epoch": 1.0330384141561204,
      "grad_norm": 1.246854543685913,
      "learning_rate": 9.219000000000001e-05,
      "loss": 0.351,
      "step": 18448
    },
    {
      "epoch": 1.0330944114682494,
      "grad_norm": 1.083481788635254,
      "learning_rate": 9.219500000000001e-05,
      "loss": 0.3952,
      "step": 18449
    },
    {
      "epoch": 1.0331504087803784,
      "grad_norm": 1.1672245264053345,
      "learning_rate": 9.22e-05,
      "loss": 0.3209,
      "step": 18450
    },
    {
      "epoch": 1.0332064060925075,
      "grad_norm": 1.3521337509155273,
      "learning_rate": 9.2205e-05,
      "loss": 0.4215,
      "step": 18451
    },
    {
      "epoch": 1.0332624034046365,
      "grad_norm": 2.070510149002075,
      "learning_rate": 9.221000000000001e-05,
      "loss": 0.4247,
      "step": 18452
    },
    {
      "epoch": 1.0333184007167655,
      "grad_norm": 1.2192219495773315,
      "learning_rate": 9.221500000000001e-05,
      "loss": 0.4238,
      "step": 18453
    },
    {
      "epoch": 1.0333743980288945,
      "grad_norm": 1.336152195930481,
      "learning_rate": 9.222000000000001e-05,
      "loss": 0.3947,
      "step": 18454
    },
    {
      "epoch": 1.0334303953410235,
      "grad_norm": 1.402063250541687,
      "learning_rate": 9.2225e-05,
      "loss": 0.4521,
      "step": 18455
    },
    {
      "epoch": 1.0334863926531526,
      "grad_norm": 1.3092293739318848,
      "learning_rate": 9.223e-05,
      "loss": 0.4352,
      "step": 18456
    },
    {
      "epoch": 1.0335423899652816,
      "grad_norm": 1.3286820650100708,
      "learning_rate": 9.2235e-05,
      "loss": 0.4113,
      "step": 18457
    },
    {
      "epoch": 1.0335983872774106,
      "grad_norm": 2.4885976314544678,
      "learning_rate": 9.224e-05,
      "loss": 0.3486,
      "step": 18458
    },
    {
      "epoch": 1.0336543845895396,
      "grad_norm": 1.1975092887878418,
      "learning_rate": 9.224500000000001e-05,
      "loss": 0.3241,
      "step": 18459
    },
    {
      "epoch": 1.0337103819016686,
      "grad_norm": 1.226528525352478,
      "learning_rate": 9.225e-05,
      "loss": 0.4842,
      "step": 18460
    },
    {
      "epoch": 1.0337663792137977,
      "grad_norm": 1.5238783359527588,
      "learning_rate": 9.2255e-05,
      "loss": 0.4655,
      "step": 18461
    },
    {
      "epoch": 1.0338223765259267,
      "grad_norm": 1.4806811809539795,
      "learning_rate": 9.226e-05,
      "loss": 0.4675,
      "step": 18462
    },
    {
      "epoch": 1.0338783738380557,
      "grad_norm": 1.292662501335144,
      "learning_rate": 9.2265e-05,
      "loss": 0.369,
      "step": 18463
    },
    {
      "epoch": 1.0339343711501847,
      "grad_norm": 1.2743792533874512,
      "learning_rate": 9.227e-05,
      "loss": 0.4125,
      "step": 18464
    },
    {
      "epoch": 1.0339903684623137,
      "grad_norm": 1.2489051818847656,
      "learning_rate": 9.2275e-05,
      "loss": 0.3937,
      "step": 18465
    },
    {
      "epoch": 1.0340463657744428,
      "grad_norm": 1.2261844873428345,
      "learning_rate": 9.228e-05,
      "loss": 0.3851,
      "step": 18466
    },
    {
      "epoch": 1.0341023630865718,
      "grad_norm": 1.483903169631958,
      "learning_rate": 9.2285e-05,
      "loss": 0.5393,
      "step": 18467
    },
    {
      "epoch": 1.0341583603987008,
      "grad_norm": 1.15414297580719,
      "learning_rate": 9.229000000000001e-05,
      "loss": 0.4449,
      "step": 18468
    },
    {
      "epoch": 1.0342143577108298,
      "grad_norm": 1.0663937330245972,
      "learning_rate": 9.229500000000001e-05,
      "loss": 0.3002,
      "step": 18469
    },
    {
      "epoch": 1.0342703550229588,
      "grad_norm": 1.4299031496047974,
      "learning_rate": 9.230000000000001e-05,
      "loss": 0.4362,
      "step": 18470
    },
    {
      "epoch": 1.0343263523350879,
      "grad_norm": 1.37778902053833,
      "learning_rate": 9.2305e-05,
      "loss": 0.4495,
      "step": 18471
    },
    {
      "epoch": 1.0343823496472169,
      "grad_norm": 1.7153373956680298,
      "learning_rate": 9.231e-05,
      "loss": 0.4534,
      "step": 18472
    },
    {
      "epoch": 1.034438346959346,
      "grad_norm": 1.1465803384780884,
      "learning_rate": 9.231500000000001e-05,
      "loss": 0.3172,
      "step": 18473
    },
    {
      "epoch": 1.034494344271475,
      "grad_norm": 1.223066806793213,
      "learning_rate": 9.232000000000001e-05,
      "loss": 0.5094,
      "step": 18474
    },
    {
      "epoch": 1.034550341583604,
      "grad_norm": 1.3934199810028076,
      "learning_rate": 9.232500000000001e-05,
      "loss": 0.3479,
      "step": 18475
    },
    {
      "epoch": 1.034606338895733,
      "grad_norm": 1.3378269672393799,
      "learning_rate": 9.233e-05,
      "loss": 0.496,
      "step": 18476
    },
    {
      "epoch": 1.034662336207862,
      "grad_norm": 1.2652432918548584,
      "learning_rate": 9.2335e-05,
      "loss": 0.461,
      "step": 18477
    },
    {
      "epoch": 1.034718333519991,
      "grad_norm": 1.155977725982666,
      "learning_rate": 9.234e-05,
      "loss": 0.2798,
      "step": 18478
    },
    {
      "epoch": 1.03477433083212,
      "grad_norm": 1.3027223348617554,
      "learning_rate": 9.2345e-05,
      "loss": 0.4,
      "step": 18479
    },
    {
      "epoch": 1.034830328144249,
      "grad_norm": 1.2647991180419922,
      "learning_rate": 9.235000000000001e-05,
      "loss": 0.4596,
      "step": 18480
    },
    {
      "epoch": 1.034886325456378,
      "grad_norm": 1.5008424520492554,
      "learning_rate": 9.2355e-05,
      "loss": 0.4096,
      "step": 18481
    },
    {
      "epoch": 1.034942322768507,
      "grad_norm": 1.7912589311599731,
      "learning_rate": 9.236e-05,
      "loss": 0.4737,
      "step": 18482
    },
    {
      "epoch": 1.034998320080636,
      "grad_norm": 1.201886534690857,
      "learning_rate": 9.2365e-05,
      "loss": 0.5048,
      "step": 18483
    },
    {
      "epoch": 1.035054317392765,
      "grad_norm": 1.1117593050003052,
      "learning_rate": 9.237e-05,
      "loss": 0.3959,
      "step": 18484
    },
    {
      "epoch": 1.0351103147048941,
      "grad_norm": 1.2172096967697144,
      "learning_rate": 9.2375e-05,
      "loss": 0.4206,
      "step": 18485
    },
    {
      "epoch": 1.0351663120170231,
      "grad_norm": 1.2553060054779053,
      "learning_rate": 9.238e-05,
      "loss": 0.5313,
      "step": 18486
    },
    {
      "epoch": 1.0352223093291522,
      "grad_norm": 1.4407403469085693,
      "learning_rate": 9.2385e-05,
      "loss": 0.4232,
      "step": 18487
    },
    {
      "epoch": 1.0352783066412812,
      "grad_norm": 1.471834659576416,
      "learning_rate": 9.239000000000001e-05,
      "loss": 0.383,
      "step": 18488
    },
    {
      "epoch": 1.0353343039534102,
      "grad_norm": 1.1300737857818604,
      "learning_rate": 9.239500000000001e-05,
      "loss": 0.3147,
      "step": 18489
    },
    {
      "epoch": 1.0353903012655392,
      "grad_norm": 1.2885539531707764,
      "learning_rate": 9.240000000000001e-05,
      "loss": 0.3815,
      "step": 18490
    },
    {
      "epoch": 1.0354462985776682,
      "grad_norm": 1.3831018209457397,
      "learning_rate": 9.240500000000001e-05,
      "loss": 0.4184,
      "step": 18491
    },
    {
      "epoch": 1.0355022958897973,
      "grad_norm": 1.3933441638946533,
      "learning_rate": 9.241e-05,
      "loss": 0.501,
      "step": 18492
    },
    {
      "epoch": 1.0355582932019263,
      "grad_norm": 1.4546265602111816,
      "learning_rate": 9.2415e-05,
      "loss": 0.3986,
      "step": 18493
    },
    {
      "epoch": 1.0356142905140553,
      "grad_norm": 1.6683248281478882,
      "learning_rate": 9.242000000000001e-05,
      "loss": 0.42,
      "step": 18494
    },
    {
      "epoch": 1.0356702878261843,
      "grad_norm": 1.5890135765075684,
      "learning_rate": 9.242500000000001e-05,
      "loss": 0.3606,
      "step": 18495
    },
    {
      "epoch": 1.0357262851383133,
      "grad_norm": 1.2694878578186035,
      "learning_rate": 9.243000000000001e-05,
      "loss": 0.3669,
      "step": 18496
    },
    {
      "epoch": 1.0357822824504423,
      "grad_norm": 1.2919161319732666,
      "learning_rate": 9.2435e-05,
      "loss": 0.3652,
      "step": 18497
    },
    {
      "epoch": 1.0358382797625714,
      "grad_norm": 1.3796987533569336,
      "learning_rate": 9.244e-05,
      "loss": 0.6024,
      "step": 18498
    },
    {
      "epoch": 1.0358942770747004,
      "grad_norm": 1.7107839584350586,
      "learning_rate": 9.2445e-05,
      "loss": 0.6868,
      "step": 18499
    },
    {
      "epoch": 1.0359502743868294,
      "grad_norm": 1.4201178550720215,
      "learning_rate": 9.245e-05,
      "loss": 0.3733,
      "step": 18500
    },
    {
      "epoch": 1.0360062716989584,
      "grad_norm": 1.321673035621643,
      "learning_rate": 9.245500000000001e-05,
      "loss": 0.3997,
      "step": 18501
    },
    {
      "epoch": 1.0360622690110874,
      "grad_norm": 1.5087130069732666,
      "learning_rate": 9.246e-05,
      "loss": 0.422,
      "step": 18502
    },
    {
      "epoch": 1.0361182663232165,
      "grad_norm": 1.3422858715057373,
      "learning_rate": 9.2465e-05,
      "loss": 0.4653,
      "step": 18503
    },
    {
      "epoch": 1.0361742636353455,
      "grad_norm": 1.1116987466812134,
      "learning_rate": 9.247e-05,
      "loss": 0.2964,
      "step": 18504
    },
    {
      "epoch": 1.0362302609474745,
      "grad_norm": 1.3748716115951538,
      "learning_rate": 9.2475e-05,
      "loss": 0.4254,
      "step": 18505
    },
    {
      "epoch": 1.0362862582596035,
      "grad_norm": 1.7915778160095215,
      "learning_rate": 9.248e-05,
      "loss": 0.4567,
      "step": 18506
    },
    {
      "epoch": 1.0363422555717325,
      "grad_norm": 1.5386706590652466,
      "learning_rate": 9.248499999999999e-05,
      "loss": 0.474,
      "step": 18507
    },
    {
      "epoch": 1.0363982528838616,
      "grad_norm": 1.4944480657577515,
      "learning_rate": 9.249e-05,
      "loss": 0.351,
      "step": 18508
    },
    {
      "epoch": 1.0364542501959906,
      "grad_norm": 1.542141318321228,
      "learning_rate": 9.249500000000001e-05,
      "loss": 0.4844,
      "step": 18509
    },
    {
      "epoch": 1.0365102475081196,
      "grad_norm": 1.530866265296936,
      "learning_rate": 9.250000000000001e-05,
      "loss": 0.4474,
      "step": 18510
    },
    {
      "epoch": 1.0365662448202486,
      "grad_norm": 1.4048144817352295,
      "learning_rate": 9.250500000000001e-05,
      "loss": 0.4107,
      "step": 18511
    },
    {
      "epoch": 1.0366222421323776,
      "grad_norm": 1.313151478767395,
      "learning_rate": 9.251000000000001e-05,
      "loss": 0.5219,
      "step": 18512
    },
    {
      "epoch": 1.0366782394445067,
      "grad_norm": 1.5241057872772217,
      "learning_rate": 9.2515e-05,
      "loss": 0.4295,
      "step": 18513
    },
    {
      "epoch": 1.0367342367566357,
      "grad_norm": 1.1829602718353271,
      "learning_rate": 9.252e-05,
      "loss": 0.3259,
      "step": 18514
    },
    {
      "epoch": 1.0367902340687647,
      "grad_norm": 1.3312814235687256,
      "learning_rate": 9.252500000000001e-05,
      "loss": 0.3562,
      "step": 18515
    },
    {
      "epoch": 1.0368462313808937,
      "grad_norm": 1.2442947626113892,
      "learning_rate": 9.253000000000001e-05,
      "loss": 0.4495,
      "step": 18516
    },
    {
      "epoch": 1.0369022286930227,
      "grad_norm": 1.3141032457351685,
      "learning_rate": 9.253500000000001e-05,
      "loss": 0.4944,
      "step": 18517
    },
    {
      "epoch": 1.0369582260051518,
      "grad_norm": 2.926708459854126,
      "learning_rate": 9.254e-05,
      "loss": 0.42,
      "step": 18518
    },
    {
      "epoch": 1.0370142233172808,
      "grad_norm": 1.6537495851516724,
      "learning_rate": 9.2545e-05,
      "loss": 0.3126,
      "step": 18519
    },
    {
      "epoch": 1.0370702206294098,
      "grad_norm": 1.189548134803772,
      "learning_rate": 9.255e-05,
      "loss": 0.4293,
      "step": 18520
    },
    {
      "epoch": 1.0371262179415388,
      "grad_norm": 1.8494853973388672,
      "learning_rate": 9.2555e-05,
      "loss": 0.4212,
      "step": 18521
    },
    {
      "epoch": 1.0371822152536678,
      "grad_norm": 1.5236074924468994,
      "learning_rate": 9.256000000000001e-05,
      "loss": 0.484,
      "step": 18522
    },
    {
      "epoch": 1.0372382125657968,
      "grad_norm": 1.5900028944015503,
      "learning_rate": 9.2565e-05,
      "loss": 0.4683,
      "step": 18523
    },
    {
      "epoch": 1.0372942098779259,
      "grad_norm": 1.4063224792480469,
      "learning_rate": 9.257e-05,
      "loss": 0.4586,
      "step": 18524
    },
    {
      "epoch": 1.0373502071900549,
      "grad_norm": 1.189079999923706,
      "learning_rate": 9.2575e-05,
      "loss": 0.389,
      "step": 18525
    },
    {
      "epoch": 1.037406204502184,
      "grad_norm": 1.4907859563827515,
      "learning_rate": 9.258e-05,
      "loss": 0.4019,
      "step": 18526
    },
    {
      "epoch": 1.037462201814313,
      "grad_norm": 1.4381401538848877,
      "learning_rate": 9.2585e-05,
      "loss": 0.5002,
      "step": 18527
    },
    {
      "epoch": 1.037518199126442,
      "grad_norm": 1.2284777164459229,
      "learning_rate": 9.258999999999999e-05,
      "loss": 0.4127,
      "step": 18528
    },
    {
      "epoch": 1.037574196438571,
      "grad_norm": 1.3698266744613647,
      "learning_rate": 9.2595e-05,
      "loss": 0.4886,
      "step": 18529
    },
    {
      "epoch": 1.0376301937507,
      "grad_norm": 1.2363755702972412,
      "learning_rate": 9.260000000000001e-05,
      "loss": 0.4111,
      "step": 18530
    },
    {
      "epoch": 1.037686191062829,
      "grad_norm": 1.3494611978530884,
      "learning_rate": 9.260500000000001e-05,
      "loss": 0.4135,
      "step": 18531
    },
    {
      "epoch": 1.037742188374958,
      "grad_norm": 1.5822937488555908,
      "learning_rate": 9.261000000000001e-05,
      "loss": 0.5227,
      "step": 18532
    },
    {
      "epoch": 1.037798185687087,
      "grad_norm": 1.212691307067871,
      "learning_rate": 9.261500000000001e-05,
      "loss": 0.3829,
      "step": 18533
    },
    {
      "epoch": 1.037854182999216,
      "grad_norm": 1.24557626247406,
      "learning_rate": 9.262e-05,
      "loss": 0.3104,
      "step": 18534
    },
    {
      "epoch": 1.037910180311345,
      "grad_norm": 1.5108773708343506,
      "learning_rate": 9.2625e-05,
      "loss": 0.5821,
      "step": 18535
    },
    {
      "epoch": 1.037966177623474,
      "grad_norm": 1.411641240119934,
      "learning_rate": 9.263e-05,
      "loss": 0.4712,
      "step": 18536
    },
    {
      "epoch": 1.0380221749356031,
      "grad_norm": 1.4698002338409424,
      "learning_rate": 9.263500000000001e-05,
      "loss": 0.4311,
      "step": 18537
    },
    {
      "epoch": 1.0380781722477321,
      "grad_norm": 1.2794344425201416,
      "learning_rate": 9.264000000000001e-05,
      "loss": 0.5081,
      "step": 18538
    },
    {
      "epoch": 1.0381341695598612,
      "grad_norm": 1.553958773612976,
      "learning_rate": 9.2645e-05,
      "loss": 0.3672,
      "step": 18539
    },
    {
      "epoch": 1.0381901668719902,
      "grad_norm": 1.2659039497375488,
      "learning_rate": 9.265e-05,
      "loss": 0.4383,
      "step": 18540
    },
    {
      "epoch": 1.0382461641841192,
      "grad_norm": 1.397771954536438,
      "learning_rate": 9.2655e-05,
      "loss": 0.4917,
      "step": 18541
    },
    {
      "epoch": 1.0383021614962482,
      "grad_norm": 1.0613263845443726,
      "learning_rate": 9.266e-05,
      "loss": 0.303,
      "step": 18542
    },
    {
      "epoch": 1.0383581588083772,
      "grad_norm": 1.3095909357070923,
      "learning_rate": 9.266500000000001e-05,
      "loss": 0.3641,
      "step": 18543
    },
    {
      "epoch": 1.0384141561205062,
      "grad_norm": 1.1477950811386108,
      "learning_rate": 9.267e-05,
      "loss": 0.4294,
      "step": 18544
    },
    {
      "epoch": 1.0384701534326353,
      "grad_norm": 1.4217934608459473,
      "learning_rate": 9.2675e-05,
      "loss": 0.3987,
      "step": 18545
    },
    {
      "epoch": 1.0385261507447643,
      "grad_norm": 1.280185580253601,
      "learning_rate": 9.268e-05,
      "loss": 0.3985,
      "step": 18546
    },
    {
      "epoch": 1.0385821480568933,
      "grad_norm": 1.384057641029358,
      "learning_rate": 9.2685e-05,
      "loss": 0.6648,
      "step": 18547
    },
    {
      "epoch": 1.0386381453690223,
      "grad_norm": 1.3243428468704224,
      "learning_rate": 9.269e-05,
      "loss": 0.4763,
      "step": 18548
    },
    {
      "epoch": 1.0386941426811513,
      "grad_norm": 1.3901629447937012,
      "learning_rate": 9.2695e-05,
      "loss": 0.5321,
      "step": 18549
    },
    {
      "epoch": 1.0387501399932804,
      "grad_norm": 1.280961513519287,
      "learning_rate": 9.27e-05,
      "loss": 0.4644,
      "step": 18550
    },
    {
      "epoch": 1.0388061373054094,
      "grad_norm": 1.2251585721969604,
      "learning_rate": 9.270500000000001e-05,
      "loss": 0.4093,
      "step": 18551
    },
    {
      "epoch": 1.0388621346175384,
      "grad_norm": 1.186393141746521,
      "learning_rate": 9.271000000000001e-05,
      "loss": 0.3122,
      "step": 18552
    },
    {
      "epoch": 1.0389181319296674,
      "grad_norm": 1.3896955251693726,
      "learning_rate": 9.271500000000001e-05,
      "loss": 0.4324,
      "step": 18553
    },
    {
      "epoch": 1.0389741292417964,
      "grad_norm": 1.251495361328125,
      "learning_rate": 9.272e-05,
      "loss": 0.3037,
      "step": 18554
    },
    {
      "epoch": 1.0390301265539255,
      "grad_norm": 1.3746366500854492,
      "learning_rate": 9.2725e-05,
      "loss": 0.4114,
      "step": 18555
    },
    {
      "epoch": 1.0390861238660545,
      "grad_norm": 1.6332991123199463,
      "learning_rate": 9.273e-05,
      "loss": 0.4307,
      "step": 18556
    },
    {
      "epoch": 1.0391421211781835,
      "grad_norm": 1.337018609046936,
      "learning_rate": 9.2735e-05,
      "loss": 0.4085,
      "step": 18557
    },
    {
      "epoch": 1.0391981184903125,
      "grad_norm": 1.35471510887146,
      "learning_rate": 9.274000000000001e-05,
      "loss": 0.5234,
      "step": 18558
    },
    {
      "epoch": 1.0392541158024415,
      "grad_norm": 1.4117554426193237,
      "learning_rate": 9.274500000000001e-05,
      "loss": 0.4672,
      "step": 18559
    },
    {
      "epoch": 1.0393101131145706,
      "grad_norm": 1.4028817415237427,
      "learning_rate": 9.275e-05,
      "loss": 0.4823,
      "step": 18560
    },
    {
      "epoch": 1.0393661104266996,
      "grad_norm": 1.233064889907837,
      "learning_rate": 9.2755e-05,
      "loss": 0.5555,
      "step": 18561
    },
    {
      "epoch": 1.0394221077388286,
      "grad_norm": 1.1610888242721558,
      "learning_rate": 9.276e-05,
      "loss": 0.3154,
      "step": 18562
    },
    {
      "epoch": 1.0394781050509576,
      "grad_norm": 1.180823564529419,
      "learning_rate": 9.2765e-05,
      "loss": 0.3393,
      "step": 18563
    },
    {
      "epoch": 1.0395341023630866,
      "grad_norm": 1.627644419670105,
      "learning_rate": 9.277000000000001e-05,
      "loss": 0.5276,
      "step": 18564
    },
    {
      "epoch": 1.0395900996752157,
      "grad_norm": 1.388357162475586,
      "learning_rate": 9.2775e-05,
      "loss": 0.4377,
      "step": 18565
    },
    {
      "epoch": 1.0396460969873447,
      "grad_norm": 1.2008401155471802,
      "learning_rate": 9.278e-05,
      "loss": 0.4566,
      "step": 18566
    },
    {
      "epoch": 1.0397020942994737,
      "grad_norm": 1.2332335710525513,
      "learning_rate": 9.2785e-05,
      "loss": 0.4352,
      "step": 18567
    },
    {
      "epoch": 1.0397580916116027,
      "grad_norm": 1.3955116271972656,
      "learning_rate": 9.279e-05,
      "loss": 0.3749,
      "step": 18568
    },
    {
      "epoch": 1.0398140889237317,
      "grad_norm": 1.4083069562911987,
      "learning_rate": 9.279500000000001e-05,
      "loss": 0.4986,
      "step": 18569
    },
    {
      "epoch": 1.0398700862358607,
      "grad_norm": 1.2220817804336548,
      "learning_rate": 9.28e-05,
      "loss": 0.3388,
      "step": 18570
    },
    {
      "epoch": 1.0399260835479898,
      "grad_norm": 3.772998571395874,
      "learning_rate": 9.2805e-05,
      "loss": 0.4862,
      "step": 18571
    },
    {
      "epoch": 1.0399820808601188,
      "grad_norm": 1.30063796043396,
      "learning_rate": 9.281000000000001e-05,
      "loss": 0.5021,
      "step": 18572
    },
    {
      "epoch": 1.0400380781722478,
      "grad_norm": 1.389096736907959,
      "learning_rate": 9.281500000000001e-05,
      "loss": 0.4203,
      "step": 18573
    },
    {
      "epoch": 1.0400940754843768,
      "grad_norm": 1.1973382234573364,
      "learning_rate": 9.282000000000001e-05,
      "loss": 0.46,
      "step": 18574
    },
    {
      "epoch": 1.0401500727965058,
      "grad_norm": 1.3328988552093506,
      "learning_rate": 9.2825e-05,
      "loss": 0.4274,
      "step": 18575
    },
    {
      "epoch": 1.0402060701086349,
      "grad_norm": 1.541388988494873,
      "learning_rate": 9.283e-05,
      "loss": 0.5745,
      "step": 18576
    },
    {
      "epoch": 1.0402620674207639,
      "grad_norm": 1.273723840713501,
      "learning_rate": 9.2835e-05,
      "loss": 0.3905,
      "step": 18577
    },
    {
      "epoch": 1.040318064732893,
      "grad_norm": 1.2285714149475098,
      "learning_rate": 9.284e-05,
      "loss": 0.3658,
      "step": 18578
    },
    {
      "epoch": 1.040374062045022,
      "grad_norm": 1.3290058374404907,
      "learning_rate": 9.284500000000001e-05,
      "loss": 0.3227,
      "step": 18579
    },
    {
      "epoch": 1.040430059357151,
      "grad_norm": 1.489951491355896,
      "learning_rate": 9.285000000000001e-05,
      "loss": 0.4602,
      "step": 18580
    },
    {
      "epoch": 1.04048605666928,
      "grad_norm": 1.428558111190796,
      "learning_rate": 9.2855e-05,
      "loss": 0.4313,
      "step": 18581
    },
    {
      "epoch": 1.040542053981409,
      "grad_norm": 1.4195235967636108,
      "learning_rate": 9.286e-05,
      "loss": 0.4642,
      "step": 18582
    },
    {
      "epoch": 1.040598051293538,
      "grad_norm": 1.3619436025619507,
      "learning_rate": 9.2865e-05,
      "loss": 0.4122,
      "step": 18583
    },
    {
      "epoch": 1.040654048605667,
      "grad_norm": 1.297425627708435,
      "learning_rate": 9.287e-05,
      "loss": 0.4537,
      "step": 18584
    },
    {
      "epoch": 1.040710045917796,
      "grad_norm": 1.6776831150054932,
      "learning_rate": 9.2875e-05,
      "loss": 0.5702,
      "step": 18585
    },
    {
      "epoch": 1.040766043229925,
      "grad_norm": 1.1916131973266602,
      "learning_rate": 9.288e-05,
      "loss": 0.3784,
      "step": 18586
    },
    {
      "epoch": 1.040822040542054,
      "grad_norm": 1.7893197536468506,
      "learning_rate": 9.2885e-05,
      "loss": 0.4912,
      "step": 18587
    },
    {
      "epoch": 1.040878037854183,
      "grad_norm": 1.0918089151382446,
      "learning_rate": 9.289e-05,
      "loss": 0.3608,
      "step": 18588
    },
    {
      "epoch": 1.040934035166312,
      "grad_norm": 1.0850043296813965,
      "learning_rate": 9.289500000000001e-05,
      "loss": 0.321,
      "step": 18589
    },
    {
      "epoch": 1.0409900324784411,
      "grad_norm": 1.3966138362884521,
      "learning_rate": 9.290000000000001e-05,
      "loss": 0.4968,
      "step": 18590
    },
    {
      "epoch": 1.0410460297905701,
      "grad_norm": 1.399404525756836,
      "learning_rate": 9.2905e-05,
      "loss": 0.4167,
      "step": 18591
    },
    {
      "epoch": 1.0411020271026992,
      "grad_norm": 1.5733047723770142,
      "learning_rate": 9.291e-05,
      "loss": 0.399,
      "step": 18592
    },
    {
      "epoch": 1.0411580244148282,
      "grad_norm": 1.2910435199737549,
      "learning_rate": 9.291500000000001e-05,
      "loss": 0.5577,
      "step": 18593
    },
    {
      "epoch": 1.0412140217269572,
      "grad_norm": 1.4125996828079224,
      "learning_rate": 9.292000000000001e-05,
      "loss": 0.4281,
      "step": 18594
    },
    {
      "epoch": 1.0412700190390862,
      "grad_norm": 1.448743224143982,
      "learning_rate": 9.292500000000001e-05,
      "loss": 0.373,
      "step": 18595
    },
    {
      "epoch": 1.0413260163512152,
      "grad_norm": 1.3208740949630737,
      "learning_rate": 9.293e-05,
      "loss": 0.4838,
      "step": 18596
    },
    {
      "epoch": 1.0413820136633443,
      "grad_norm": 2.6943812370300293,
      "learning_rate": 9.2935e-05,
      "loss": 0.3669,
      "step": 18597
    },
    {
      "epoch": 1.0414380109754733,
      "grad_norm": 1.666571021080017,
      "learning_rate": 9.294e-05,
      "loss": 0.4782,
      "step": 18598
    },
    {
      "epoch": 1.0414940082876023,
      "grad_norm": 1.346853256225586,
      "learning_rate": 9.2945e-05,
      "loss": 0.4829,
      "step": 18599
    },
    {
      "epoch": 1.0415500055997313,
      "grad_norm": 1.322637915611267,
      "learning_rate": 9.295000000000001e-05,
      "loss": 0.3847,
      "step": 18600
    },
    {
      "epoch": 1.0416060029118603,
      "grad_norm": 1.4859142303466797,
      "learning_rate": 9.295500000000001e-05,
      "loss": 0.4769,
      "step": 18601
    },
    {
      "epoch": 1.0416620002239894,
      "grad_norm": 1.3755728006362915,
      "learning_rate": 9.296e-05,
      "loss": 0.4521,
      "step": 18602
    },
    {
      "epoch": 1.0417179975361184,
      "grad_norm": 1.2686610221862793,
      "learning_rate": 9.2965e-05,
      "loss": 0.3776,
      "step": 18603
    },
    {
      "epoch": 1.0417739948482474,
      "grad_norm": 1.4579969644546509,
      "learning_rate": 9.297e-05,
      "loss": 0.5432,
      "step": 18604
    },
    {
      "epoch": 1.0418299921603762,
      "grad_norm": 1.4644408226013184,
      "learning_rate": 9.2975e-05,
      "loss": 0.4334,
      "step": 18605
    },
    {
      "epoch": 1.0418859894725052,
      "grad_norm": 1.4493407011032104,
      "learning_rate": 9.298e-05,
      "loss": 0.4742,
      "step": 18606
    },
    {
      "epoch": 1.0419419867846342,
      "grad_norm": 1.270177960395813,
      "learning_rate": 9.2985e-05,
      "loss": 0.3793,
      "step": 18607
    },
    {
      "epoch": 1.0419979840967633,
      "grad_norm": 1.3753674030303955,
      "learning_rate": 9.299e-05,
      "loss": 0.4165,
      "step": 18608
    },
    {
      "epoch": 1.0420539814088923,
      "grad_norm": 1.2743052244186401,
      "learning_rate": 9.299500000000001e-05,
      "loss": 0.443,
      "step": 18609
    },
    {
      "epoch": 1.0421099787210213,
      "grad_norm": 1.2452137470245361,
      "learning_rate": 9.300000000000001e-05,
      "loss": 0.3657,
      "step": 18610
    },
    {
      "epoch": 1.0421659760331503,
      "grad_norm": 1.2093377113342285,
      "learning_rate": 9.300500000000001e-05,
      "loss": 0.4707,
      "step": 18611
    },
    {
      "epoch": 1.0422219733452793,
      "grad_norm": 1.4324147701263428,
      "learning_rate": 9.301e-05,
      "loss": 0.392,
      "step": 18612
    },
    {
      "epoch": 1.0422779706574083,
      "grad_norm": 1.2902770042419434,
      "learning_rate": 9.3015e-05,
      "loss": 0.4421,
      "step": 18613
    },
    {
      "epoch": 1.0423339679695374,
      "grad_norm": 1.3648157119750977,
      "learning_rate": 9.302e-05,
      "loss": 0.4302,
      "step": 18614
    },
    {
      "epoch": 1.0423899652816664,
      "grad_norm": 1.7705236673355103,
      "learning_rate": 9.302500000000001e-05,
      "loss": 0.4566,
      "step": 18615
    },
    {
      "epoch": 1.0424459625937954,
      "grad_norm": 1.3672006130218506,
      "learning_rate": 9.303000000000001e-05,
      "loss": 0.4822,
      "step": 18616
    },
    {
      "epoch": 1.0425019599059244,
      "grad_norm": 1.1181634664535522,
      "learning_rate": 9.3035e-05,
      "loss": 0.3502,
      "step": 18617
    },
    {
      "epoch": 1.0425579572180534,
      "grad_norm": 1.2181497812271118,
      "learning_rate": 9.304e-05,
      "loss": 0.3622,
      "step": 18618
    },
    {
      "epoch": 1.0426139545301825,
      "grad_norm": 1.4582486152648926,
      "learning_rate": 9.3045e-05,
      "loss": 0.3453,
      "step": 18619
    },
    {
      "epoch": 1.0426699518423115,
      "grad_norm": 1.049888253211975,
      "learning_rate": 9.305e-05,
      "loss": 0.385,
      "step": 18620
    },
    {
      "epoch": 1.0427259491544405,
      "grad_norm": 1.425157904624939,
      "learning_rate": 9.305500000000001e-05,
      "loss": 0.4258,
      "step": 18621
    },
    {
      "epoch": 1.0427819464665695,
      "grad_norm": 1.3080180883407593,
      "learning_rate": 9.306000000000001e-05,
      "loss": 0.4685,
      "step": 18622
    },
    {
      "epoch": 1.0428379437786985,
      "grad_norm": 1.5657936334609985,
      "learning_rate": 9.3065e-05,
      "loss": 0.6739,
      "step": 18623
    },
    {
      "epoch": 1.0428939410908276,
      "grad_norm": 1.6061358451843262,
      "learning_rate": 9.307e-05,
      "loss": 0.4863,
      "step": 18624
    },
    {
      "epoch": 1.0429499384029566,
      "grad_norm": 1.7165899276733398,
      "learning_rate": 9.3075e-05,
      "loss": 0.6032,
      "step": 18625
    },
    {
      "epoch": 1.0430059357150856,
      "grad_norm": 1.1328483819961548,
      "learning_rate": 9.308e-05,
      "loss": 0.3495,
      "step": 18626
    },
    {
      "epoch": 1.0430619330272146,
      "grad_norm": 1.3610625267028809,
      "learning_rate": 9.3085e-05,
      "loss": 0.4514,
      "step": 18627
    },
    {
      "epoch": 1.0431179303393436,
      "grad_norm": 1.328114628791809,
      "learning_rate": 9.309e-05,
      "loss": 0.4111,
      "step": 18628
    },
    {
      "epoch": 1.0431739276514727,
      "grad_norm": 1.3617134094238281,
      "learning_rate": 9.309500000000002e-05,
      "loss": 0.4748,
      "step": 18629
    },
    {
      "epoch": 1.0432299249636017,
      "grad_norm": 4.176542282104492,
      "learning_rate": 9.310000000000001e-05,
      "loss": 0.4475,
      "step": 18630
    },
    {
      "epoch": 1.0432859222757307,
      "grad_norm": 1.5613341331481934,
      "learning_rate": 9.310500000000001e-05,
      "loss": 0.472,
      "step": 18631
    },
    {
      "epoch": 1.0433419195878597,
      "grad_norm": 1.3893775939941406,
      "learning_rate": 9.311000000000001e-05,
      "loss": 0.4798,
      "step": 18632
    },
    {
      "epoch": 1.0433979168999887,
      "grad_norm": 1.4824298620224,
      "learning_rate": 9.3115e-05,
      "loss": 0.4829,
      "step": 18633
    },
    {
      "epoch": 1.0434539142121178,
      "grad_norm": 1.209849238395691,
      "learning_rate": 9.312e-05,
      "loss": 0.36,
      "step": 18634
    },
    {
      "epoch": 1.0435099115242468,
      "grad_norm": 1.2994004487991333,
      "learning_rate": 9.3125e-05,
      "loss": 0.5197,
      "step": 18635
    },
    {
      "epoch": 1.0435659088363758,
      "grad_norm": 1.2604016065597534,
      "learning_rate": 9.313000000000001e-05,
      "loss": 0.4214,
      "step": 18636
    },
    {
      "epoch": 1.0436219061485048,
      "grad_norm": 1.1662832498550415,
      "learning_rate": 9.313500000000001e-05,
      "loss": 0.3889,
      "step": 18637
    },
    {
      "epoch": 1.0436779034606338,
      "grad_norm": 1.609951376914978,
      "learning_rate": 9.314e-05,
      "loss": 0.3799,
      "step": 18638
    },
    {
      "epoch": 1.0437339007727628,
      "grad_norm": 1.2422561645507812,
      "learning_rate": 9.3145e-05,
      "loss": 0.4896,
      "step": 18639
    },
    {
      "epoch": 1.0437898980848919,
      "grad_norm": 1.5622755289077759,
      "learning_rate": 9.315e-05,
      "loss": 0.4953,
      "step": 18640
    },
    {
      "epoch": 1.0438458953970209,
      "grad_norm": 1.173612356185913,
      "learning_rate": 9.3155e-05,
      "loss": 0.4147,
      "step": 18641
    },
    {
      "epoch": 1.04390189270915,
      "grad_norm": 1.1745073795318604,
      "learning_rate": 9.316000000000001e-05,
      "loss": 0.366,
      "step": 18642
    },
    {
      "epoch": 1.043957890021279,
      "grad_norm": 1.1597683429718018,
      "learning_rate": 9.3165e-05,
      "loss": 0.3775,
      "step": 18643
    },
    {
      "epoch": 1.044013887333408,
      "grad_norm": 1.605271339416504,
      "learning_rate": 9.317e-05,
      "loss": 0.3384,
      "step": 18644
    },
    {
      "epoch": 1.044069884645537,
      "grad_norm": 1.1824640035629272,
      "learning_rate": 9.3175e-05,
      "loss": 0.5947,
      "step": 18645
    },
    {
      "epoch": 1.044125881957666,
      "grad_norm": 1.3627582788467407,
      "learning_rate": 9.318e-05,
      "loss": 0.4234,
      "step": 18646
    },
    {
      "epoch": 1.044181879269795,
      "grad_norm": 1.2604939937591553,
      "learning_rate": 9.3185e-05,
      "loss": 0.4672,
      "step": 18647
    },
    {
      "epoch": 1.044237876581924,
      "grad_norm": 1.429374098777771,
      "learning_rate": 9.319e-05,
      "loss": 0.47,
      "step": 18648
    },
    {
      "epoch": 1.044293873894053,
      "grad_norm": 1.1948559284210205,
      "learning_rate": 9.3195e-05,
      "loss": 0.3278,
      "step": 18649
    },
    {
      "epoch": 1.044349871206182,
      "grad_norm": 1.5485488176345825,
      "learning_rate": 9.320000000000002e-05,
      "loss": 0.6168,
      "step": 18650
    },
    {
      "epoch": 1.044405868518311,
      "grad_norm": 1.5370079278945923,
      "learning_rate": 9.320500000000001e-05,
      "loss": 0.3781,
      "step": 18651
    },
    {
      "epoch": 1.04446186583044,
      "grad_norm": 1.2994308471679688,
      "learning_rate": 9.321000000000001e-05,
      "loss": 0.437,
      "step": 18652
    },
    {
      "epoch": 1.0445178631425691,
      "grad_norm": 1.4331862926483154,
      "learning_rate": 9.321500000000001e-05,
      "loss": 0.5217,
      "step": 18653
    },
    {
      "epoch": 1.0445738604546981,
      "grad_norm": 1.451013445854187,
      "learning_rate": 9.322e-05,
      "loss": 0.5015,
      "step": 18654
    },
    {
      "epoch": 1.0446298577668272,
      "grad_norm": 1.409241795539856,
      "learning_rate": 9.3225e-05,
      "loss": 0.5115,
      "step": 18655
    },
    {
      "epoch": 1.0446858550789562,
      "grad_norm": 1.174866795539856,
      "learning_rate": 9.323e-05,
      "loss": 0.4376,
      "step": 18656
    },
    {
      "epoch": 1.0447418523910852,
      "grad_norm": 1.2477177381515503,
      "learning_rate": 9.323500000000001e-05,
      "loss": 0.3533,
      "step": 18657
    },
    {
      "epoch": 1.0447978497032142,
      "grad_norm": 1.4175711870193481,
      "learning_rate": 9.324000000000001e-05,
      "loss": 0.3326,
      "step": 18658
    },
    {
      "epoch": 1.0448538470153432,
      "grad_norm": 1.3794015645980835,
      "learning_rate": 9.3245e-05,
      "loss": 0.3185,
      "step": 18659
    },
    {
      "epoch": 1.0449098443274722,
      "grad_norm": 1.3675583600997925,
      "learning_rate": 9.325e-05,
      "loss": 0.3837,
      "step": 18660
    },
    {
      "epoch": 1.0449658416396013,
      "grad_norm": 1.3625034093856812,
      "learning_rate": 9.3255e-05,
      "loss": 0.5808,
      "step": 18661
    },
    {
      "epoch": 1.0450218389517303,
      "grad_norm": 2.7807505130767822,
      "learning_rate": 9.326e-05,
      "loss": 0.4191,
      "step": 18662
    },
    {
      "epoch": 1.0450778362638593,
      "grad_norm": 1.6819387674331665,
      "learning_rate": 9.326500000000001e-05,
      "loss": 0.5341,
      "step": 18663
    },
    {
      "epoch": 1.0451338335759883,
      "grad_norm": 1.4379454851150513,
      "learning_rate": 9.327e-05,
      "loss": 0.4973,
      "step": 18664
    },
    {
      "epoch": 1.0451898308881173,
      "grad_norm": 1.2826933860778809,
      "learning_rate": 9.3275e-05,
      "loss": 0.4619,
      "step": 18665
    },
    {
      "epoch": 1.0452458282002464,
      "grad_norm": 1.3095221519470215,
      "learning_rate": 9.328e-05,
      "loss": 0.451,
      "step": 18666
    },
    {
      "epoch": 1.0453018255123754,
      "grad_norm": 1.3317927122116089,
      "learning_rate": 9.3285e-05,
      "loss": 0.4667,
      "step": 18667
    },
    {
      "epoch": 1.0453578228245044,
      "grad_norm": 1.3291807174682617,
      "learning_rate": 9.329e-05,
      "loss": 0.353,
      "step": 18668
    },
    {
      "epoch": 1.0454138201366334,
      "grad_norm": 1.4625712633132935,
      "learning_rate": 9.3295e-05,
      "loss": 0.4523,
      "step": 18669
    },
    {
      "epoch": 1.0454698174487624,
      "grad_norm": 1.2807096242904663,
      "learning_rate": 9.33e-05,
      "loss": 0.451,
      "step": 18670
    },
    {
      "epoch": 1.0455258147608915,
      "grad_norm": 1.2990058660507202,
      "learning_rate": 9.330500000000002e-05,
      "loss": 0.4871,
      "step": 18671
    },
    {
      "epoch": 1.0455818120730205,
      "grad_norm": 1.5651144981384277,
      "learning_rate": 9.331000000000001e-05,
      "loss": 0.3684,
      "step": 18672
    },
    {
      "epoch": 1.0456378093851495,
      "grad_norm": 3.8242337703704834,
      "learning_rate": 9.331500000000001e-05,
      "loss": 0.3723,
      "step": 18673
    },
    {
      "epoch": 1.0456938066972785,
      "grad_norm": 1.4493603706359863,
      "learning_rate": 9.332000000000001e-05,
      "loss": 0.4018,
      "step": 18674
    },
    {
      "epoch": 1.0457498040094075,
      "grad_norm": 1.471799612045288,
      "learning_rate": 9.3325e-05,
      "loss": 0.5327,
      "step": 18675
    },
    {
      "epoch": 1.0458058013215366,
      "grad_norm": 1.222642183303833,
      "learning_rate": 9.333e-05,
      "loss": 0.3335,
      "step": 18676
    },
    {
      "epoch": 1.0458617986336656,
      "grad_norm": 1.3658721446990967,
      "learning_rate": 9.3335e-05,
      "loss": 0.4231,
      "step": 18677
    },
    {
      "epoch": 1.0459177959457946,
      "grad_norm": 1.2885686159133911,
      "learning_rate": 9.334000000000001e-05,
      "loss": 0.3942,
      "step": 18678
    },
    {
      "epoch": 1.0459737932579236,
      "grad_norm": 1.3054735660552979,
      "learning_rate": 9.334500000000001e-05,
      "loss": 0.3332,
      "step": 18679
    },
    {
      "epoch": 1.0460297905700526,
      "grad_norm": 1.490426778793335,
      "learning_rate": 9.335e-05,
      "loss": 0.4277,
      "step": 18680
    },
    {
      "epoch": 1.0460857878821817,
      "grad_norm": 1.3818128108978271,
      "learning_rate": 9.3355e-05,
      "loss": 0.3945,
      "step": 18681
    },
    {
      "epoch": 1.0461417851943107,
      "grad_norm": 1.3044003248214722,
      "learning_rate": 9.336e-05,
      "loss": 0.4266,
      "step": 18682
    },
    {
      "epoch": 1.0461977825064397,
      "grad_norm": 1.2530335187911987,
      "learning_rate": 9.3365e-05,
      "loss": 0.3952,
      "step": 18683
    },
    {
      "epoch": 1.0462537798185687,
      "grad_norm": 1.3521382808685303,
      "learning_rate": 9.337e-05,
      "loss": 0.4088,
      "step": 18684
    },
    {
      "epoch": 1.0463097771306977,
      "grad_norm": 1.6739455461502075,
      "learning_rate": 9.3375e-05,
      "loss": 0.4374,
      "step": 18685
    },
    {
      "epoch": 1.0463657744428267,
      "grad_norm": 1.3913389444351196,
      "learning_rate": 9.338e-05,
      "loss": 0.4193,
      "step": 18686
    },
    {
      "epoch": 1.0464217717549558,
      "grad_norm": 1.3583130836486816,
      "learning_rate": 9.3385e-05,
      "loss": 0.5621,
      "step": 18687
    },
    {
      "epoch": 1.0464777690670848,
      "grad_norm": 2.3243002891540527,
      "learning_rate": 9.339e-05,
      "loss": 0.4014,
      "step": 18688
    },
    {
      "epoch": 1.0465337663792138,
      "grad_norm": 1.4661026000976562,
      "learning_rate": 9.3395e-05,
      "loss": 0.4266,
      "step": 18689
    },
    {
      "epoch": 1.0465897636913428,
      "grad_norm": 1.2348803281784058,
      "learning_rate": 9.340000000000001e-05,
      "loss": 0.3395,
      "step": 18690
    },
    {
      "epoch": 1.0466457610034718,
      "grad_norm": 1.4044138193130493,
      "learning_rate": 9.3405e-05,
      "loss": 0.4421,
      "step": 18691
    },
    {
      "epoch": 1.0467017583156009,
      "grad_norm": 1.2582522630691528,
      "learning_rate": 9.341000000000002e-05,
      "loss": 0.3806,
      "step": 18692
    },
    {
      "epoch": 1.0467577556277299,
      "grad_norm": 1.372855544090271,
      "learning_rate": 9.341500000000001e-05,
      "loss": 0.4049,
      "step": 18693
    },
    {
      "epoch": 1.046813752939859,
      "grad_norm": 1.436766505241394,
      "learning_rate": 9.342000000000001e-05,
      "loss": 0.4663,
      "step": 18694
    },
    {
      "epoch": 1.046869750251988,
      "grad_norm": 1.523422360420227,
      "learning_rate": 9.342500000000001e-05,
      "loss": 0.4943,
      "step": 18695
    },
    {
      "epoch": 1.046925747564117,
      "grad_norm": 1.4499166011810303,
      "learning_rate": 9.343e-05,
      "loss": 0.5359,
      "step": 18696
    },
    {
      "epoch": 1.046981744876246,
      "grad_norm": 1.202656626701355,
      "learning_rate": 9.3435e-05,
      "loss": 0.4509,
      "step": 18697
    },
    {
      "epoch": 1.047037742188375,
      "grad_norm": 1.2682124376296997,
      "learning_rate": 9.344e-05,
      "loss": 0.4227,
      "step": 18698
    },
    {
      "epoch": 1.047093739500504,
      "grad_norm": 1.3726247549057007,
      "learning_rate": 9.344500000000001e-05,
      "loss": 0.4103,
      "step": 18699
    },
    {
      "epoch": 1.047149736812633,
      "grad_norm": 1.339306116104126,
      "learning_rate": 9.345000000000001e-05,
      "loss": 0.585,
      "step": 18700
    },
    {
      "epoch": 1.047205734124762,
      "grad_norm": 2.1495375633239746,
      "learning_rate": 9.3455e-05,
      "loss": 0.4186,
      "step": 18701
    },
    {
      "epoch": 1.047261731436891,
      "grad_norm": 1.4257663488388062,
      "learning_rate": 9.346e-05,
      "loss": 0.4874,
      "step": 18702
    },
    {
      "epoch": 1.04731772874902,
      "grad_norm": 1.301926851272583,
      "learning_rate": 9.3465e-05,
      "loss": 0.581,
      "step": 18703
    },
    {
      "epoch": 1.047373726061149,
      "grad_norm": 1.357356071472168,
      "learning_rate": 9.347e-05,
      "loss": 0.4752,
      "step": 18704
    },
    {
      "epoch": 1.047429723373278,
      "grad_norm": 1.335082769393921,
      "learning_rate": 9.3475e-05,
      "loss": 0.5403,
      "step": 18705
    },
    {
      "epoch": 1.0474857206854071,
      "grad_norm": 1.386541724205017,
      "learning_rate": 9.348e-05,
      "loss": 0.3512,
      "step": 18706
    },
    {
      "epoch": 1.0475417179975361,
      "grad_norm": 1.3624595403671265,
      "learning_rate": 9.3485e-05,
      "loss": 0.5465,
      "step": 18707
    },
    {
      "epoch": 1.0475977153096652,
      "grad_norm": 1.5704821348190308,
      "learning_rate": 9.349e-05,
      "loss": 0.4433,
      "step": 18708
    },
    {
      "epoch": 1.0476537126217942,
      "grad_norm": 1.186461091041565,
      "learning_rate": 9.3495e-05,
      "loss": 0.3858,
      "step": 18709
    },
    {
      "epoch": 1.0477097099339232,
      "grad_norm": 1.1894114017486572,
      "learning_rate": 9.350000000000001e-05,
      "loss": 0.3432,
      "step": 18710
    },
    {
      "epoch": 1.0477657072460522,
      "grad_norm": 1.700202226638794,
      "learning_rate": 9.350500000000001e-05,
      "loss": 0.482,
      "step": 18711
    },
    {
      "epoch": 1.0478217045581812,
      "grad_norm": 1.487416386604309,
      "learning_rate": 9.351e-05,
      "loss": 0.4917,
      "step": 18712
    },
    {
      "epoch": 1.0478777018703103,
      "grad_norm": 1.6582293510437012,
      "learning_rate": 9.3515e-05,
      "loss": 0.4482,
      "step": 18713
    },
    {
      "epoch": 1.0479336991824393,
      "grad_norm": 1.4676433801651,
      "learning_rate": 9.352000000000001e-05,
      "loss": 0.4053,
      "step": 18714
    },
    {
      "epoch": 1.0479896964945683,
      "grad_norm": 1.5263688564300537,
      "learning_rate": 9.352500000000001e-05,
      "loss": 0.5713,
      "step": 18715
    },
    {
      "epoch": 1.0480456938066973,
      "grad_norm": 1.2526209354400635,
      "learning_rate": 9.353000000000001e-05,
      "loss": 0.6906,
      "step": 18716
    },
    {
      "epoch": 1.0481016911188263,
      "grad_norm": 1.3255534172058105,
      "learning_rate": 9.3535e-05,
      "loss": 0.4322,
      "step": 18717
    },
    {
      "epoch": 1.0481576884309554,
      "grad_norm": 1.4175266027450562,
      "learning_rate": 9.354e-05,
      "loss": 0.3491,
      "step": 18718
    },
    {
      "epoch": 1.0482136857430844,
      "grad_norm": 1.4038658142089844,
      "learning_rate": 9.3545e-05,
      "loss": 0.3943,
      "step": 18719
    },
    {
      "epoch": 1.0482696830552134,
      "grad_norm": 1.6249943971633911,
      "learning_rate": 9.355000000000001e-05,
      "loss": 0.4448,
      "step": 18720
    },
    {
      "epoch": 1.0483256803673424,
      "grad_norm": 1.6626101732254028,
      "learning_rate": 9.355500000000001e-05,
      "loss": 0.4816,
      "step": 18721
    },
    {
      "epoch": 1.0483816776794714,
      "grad_norm": 1.1830973625183105,
      "learning_rate": 9.356e-05,
      "loss": 0.4251,
      "step": 18722
    },
    {
      "epoch": 1.0484376749916005,
      "grad_norm": 1.3155674934387207,
      "learning_rate": 9.3565e-05,
      "loss": 0.4556,
      "step": 18723
    },
    {
      "epoch": 1.0484936723037295,
      "grad_norm": 1.5291848182678223,
      "learning_rate": 9.357e-05,
      "loss": 0.5577,
      "step": 18724
    },
    {
      "epoch": 1.0485496696158585,
      "grad_norm": 10.447802543640137,
      "learning_rate": 9.3575e-05,
      "loss": 0.4634,
      "step": 18725
    },
    {
      "epoch": 1.0486056669279875,
      "grad_norm": 1.464624047279358,
      "learning_rate": 9.358e-05,
      "loss": 0.5386,
      "step": 18726
    },
    {
      "epoch": 1.0486616642401165,
      "grad_norm": 1.2470518350601196,
      "learning_rate": 9.3585e-05,
      "loss": 0.4673,
      "step": 18727
    },
    {
      "epoch": 1.0487176615522456,
      "grad_norm": 1.7545257806777954,
      "learning_rate": 9.359e-05,
      "loss": 0.4946,
      "step": 18728
    },
    {
      "epoch": 1.0487736588643746,
      "grad_norm": 1.2152029275894165,
      "learning_rate": 9.3595e-05,
      "loss": 0.3201,
      "step": 18729
    },
    {
      "epoch": 1.0488296561765036,
      "grad_norm": 1.2757127285003662,
      "learning_rate": 9.360000000000001e-05,
      "loss": 0.4564,
      "step": 18730
    },
    {
      "epoch": 1.0488856534886326,
      "grad_norm": 1.3454784154891968,
      "learning_rate": 9.360500000000001e-05,
      "loss": 0.4314,
      "step": 18731
    },
    {
      "epoch": 1.0489416508007616,
      "grad_norm": 1.4702589511871338,
      "learning_rate": 9.361e-05,
      "loss": 0.4204,
      "step": 18732
    },
    {
      "epoch": 1.0489976481128906,
      "grad_norm": 1.1373652219772339,
      "learning_rate": 9.3615e-05,
      "loss": 0.3671,
      "step": 18733
    },
    {
      "epoch": 1.0490536454250197,
      "grad_norm": 1.4210302829742432,
      "learning_rate": 9.362e-05,
      "loss": 0.419,
      "step": 18734
    },
    {
      "epoch": 1.0491096427371487,
      "grad_norm": 1.38583505153656,
      "learning_rate": 9.362500000000001e-05,
      "loss": 0.4431,
      "step": 18735
    },
    {
      "epoch": 1.0491656400492777,
      "grad_norm": 1.4215130805969238,
      "learning_rate": 9.363000000000001e-05,
      "loss": 0.5827,
      "step": 18736
    },
    {
      "epoch": 1.0492216373614067,
      "grad_norm": 1.4129101037979126,
      "learning_rate": 9.363500000000001e-05,
      "loss": 0.4041,
      "step": 18737
    },
    {
      "epoch": 1.0492776346735357,
      "grad_norm": 1.5683997869491577,
      "learning_rate": 9.364e-05,
      "loss": 0.5229,
      "step": 18738
    },
    {
      "epoch": 1.0493336319856648,
      "grad_norm": 1.083092451095581,
      "learning_rate": 9.3645e-05,
      "loss": 0.3352,
      "step": 18739
    },
    {
      "epoch": 1.0493896292977938,
      "grad_norm": 1.3357940912246704,
      "learning_rate": 9.365e-05,
      "loss": 0.5065,
      "step": 18740
    },
    {
      "epoch": 1.0494456266099228,
      "grad_norm": 1.3490136861801147,
      "learning_rate": 9.365500000000001e-05,
      "loss": 0.4825,
      "step": 18741
    },
    {
      "epoch": 1.0495016239220518,
      "grad_norm": 1.1608885526657104,
      "learning_rate": 9.366000000000001e-05,
      "loss": 0.3584,
      "step": 18742
    },
    {
      "epoch": 1.0495576212341808,
      "grad_norm": 3.1415653228759766,
      "learning_rate": 9.3665e-05,
      "loss": 0.4131,
      "step": 18743
    },
    {
      "epoch": 1.0496136185463099,
      "grad_norm": 1.2254018783569336,
      "learning_rate": 9.367e-05,
      "loss": 0.3774,
      "step": 18744
    },
    {
      "epoch": 1.0496696158584389,
      "grad_norm": 1.1179746389389038,
      "learning_rate": 9.3675e-05,
      "loss": 0.3453,
      "step": 18745
    },
    {
      "epoch": 1.049725613170568,
      "grad_norm": 1.2203619480133057,
      "learning_rate": 9.368e-05,
      "loss": 0.5304,
      "step": 18746
    },
    {
      "epoch": 1.049781610482697,
      "grad_norm": 1.4112677574157715,
      "learning_rate": 9.3685e-05,
      "loss": 0.7577,
      "step": 18747
    },
    {
      "epoch": 1.049837607794826,
      "grad_norm": 1.3277983665466309,
      "learning_rate": 9.369e-05,
      "loss": 0.3917,
      "step": 18748
    },
    {
      "epoch": 1.049893605106955,
      "grad_norm": 1.2955080270767212,
      "learning_rate": 9.3695e-05,
      "loss": 0.3897,
      "step": 18749
    },
    {
      "epoch": 1.049949602419084,
      "grad_norm": 1.3035756349563599,
      "learning_rate": 9.370000000000001e-05,
      "loss": 0.3826,
      "step": 18750
    },
    {
      "epoch": 1.050005599731213,
      "grad_norm": 1.3221250772476196,
      "learning_rate": 9.370500000000001e-05,
      "loss": 0.3784,
      "step": 18751
    },
    {
      "epoch": 1.050061597043342,
      "grad_norm": 1.2354234457015991,
      "learning_rate": 9.371000000000001e-05,
      "loss": 0.3577,
      "step": 18752
    },
    {
      "epoch": 1.050117594355471,
      "grad_norm": 1.1802183389663696,
      "learning_rate": 9.3715e-05,
      "loss": 0.4117,
      "step": 18753
    },
    {
      "epoch": 1.0501735916676,
      "grad_norm": 1.2856926918029785,
      "learning_rate": 9.372e-05,
      "loss": 0.4059,
      "step": 18754
    },
    {
      "epoch": 1.050229588979729,
      "grad_norm": 1.5393387079238892,
      "learning_rate": 9.3725e-05,
      "loss": 0.3178,
      "step": 18755
    },
    {
      "epoch": 1.050285586291858,
      "grad_norm": 1.3233131170272827,
      "learning_rate": 9.373000000000001e-05,
      "loss": 0.3248,
      "step": 18756
    },
    {
      "epoch": 1.050341583603987,
      "grad_norm": 1.6761584281921387,
      "learning_rate": 9.373500000000001e-05,
      "loss": 0.4015,
      "step": 18757
    },
    {
      "epoch": 1.0503975809161161,
      "grad_norm": 1.2219070196151733,
      "learning_rate": 9.374000000000001e-05,
      "loss": 0.6149,
      "step": 18758
    },
    {
      "epoch": 1.0504535782282451,
      "grad_norm": 1.2063491344451904,
      "learning_rate": 9.3745e-05,
      "loss": 0.3681,
      "step": 18759
    },
    {
      "epoch": 1.0505095755403742,
      "grad_norm": 1.8176213502883911,
      "learning_rate": 9.375e-05,
      "loss": 0.5237,
      "step": 18760
    },
    {
      "epoch": 1.0505655728525032,
      "grad_norm": 1.5886811017990112,
      "learning_rate": 9.3755e-05,
      "loss": 0.461,
      "step": 18761
    },
    {
      "epoch": 1.0506215701646322,
      "grad_norm": 1.32243812084198,
      "learning_rate": 9.376e-05,
      "loss": 0.4032,
      "step": 18762
    },
    {
      "epoch": 1.0506775674767612,
      "grad_norm": 1.169313669204712,
      "learning_rate": 9.376500000000001e-05,
      "loss": 0.4166,
      "step": 18763
    },
    {
      "epoch": 1.0507335647888902,
      "grad_norm": 1.393068790435791,
      "learning_rate": 9.377e-05,
      "loss": 0.4246,
      "step": 18764
    },
    {
      "epoch": 1.050789562101019,
      "grad_norm": 1.259882926940918,
      "learning_rate": 9.3775e-05,
      "loss": 0.4142,
      "step": 18765
    },
    {
      "epoch": 1.050845559413148,
      "grad_norm": 1.1903698444366455,
      "learning_rate": 9.378e-05,
      "loss": 0.3495,
      "step": 18766
    },
    {
      "epoch": 1.050901556725277,
      "grad_norm": 1.4354101419448853,
      "learning_rate": 9.3785e-05,
      "loss": 0.3947,
      "step": 18767
    },
    {
      "epoch": 1.050957554037406,
      "grad_norm": 1.1266363859176636,
      "learning_rate": 9.379e-05,
      "loss": 0.3524,
      "step": 18768
    },
    {
      "epoch": 1.0510135513495351,
      "grad_norm": 1.3983913660049438,
      "learning_rate": 9.3795e-05,
      "loss": 0.4935,
      "step": 18769
    },
    {
      "epoch": 1.0510695486616641,
      "grad_norm": 1.611491084098816,
      "learning_rate": 9.38e-05,
      "loss": 0.5854,
      "step": 18770
    },
    {
      "epoch": 1.0511255459737932,
      "grad_norm": 1.1961493492126465,
      "learning_rate": 9.380500000000001e-05,
      "loss": 0.403,
      "step": 18771
    },
    {
      "epoch": 1.0511815432859222,
      "grad_norm": 1.6286276578903198,
      "learning_rate": 9.381000000000001e-05,
      "loss": 0.3887,
      "step": 18772
    },
    {
      "epoch": 1.0512375405980512,
      "grad_norm": 1.3481632471084595,
      "learning_rate": 9.381500000000001e-05,
      "loss": 0.3905,
      "step": 18773
    },
    {
      "epoch": 1.0512935379101802,
      "grad_norm": 1.800219178199768,
      "learning_rate": 9.382e-05,
      "loss": 0.6068,
      "step": 18774
    },
    {
      "epoch": 1.0513495352223092,
      "grad_norm": 1.3384664058685303,
      "learning_rate": 9.3825e-05,
      "loss": 0.3587,
      "step": 18775
    },
    {
      "epoch": 1.0514055325344382,
      "grad_norm": 1.3007813692092896,
      "learning_rate": 9.383e-05,
      "loss": 0.3219,
      "step": 18776
    },
    {
      "epoch": 1.0514615298465673,
      "grad_norm": 1.9404252767562866,
      "learning_rate": 9.383500000000001e-05,
      "loss": 0.535,
      "step": 18777
    },
    {
      "epoch": 1.0515175271586963,
      "grad_norm": 1.2272975444793701,
      "learning_rate": 9.384000000000001e-05,
      "loss": 0.432,
      "step": 18778
    },
    {
      "epoch": 1.0515735244708253,
      "grad_norm": 1.2873376607894897,
      "learning_rate": 9.384500000000001e-05,
      "loss": 0.3672,
      "step": 18779
    },
    {
      "epoch": 1.0516295217829543,
      "grad_norm": 1.1425484418869019,
      "learning_rate": 9.385e-05,
      "loss": 0.3483,
      "step": 18780
    },
    {
      "epoch": 1.0516855190950833,
      "grad_norm": 1.2005585432052612,
      "learning_rate": 9.3855e-05,
      "loss": 0.4503,
      "step": 18781
    },
    {
      "epoch": 1.0517415164072124,
      "grad_norm": 1.5329253673553467,
      "learning_rate": 9.386e-05,
      "loss": 0.565,
      "step": 18782
    },
    {
      "epoch": 1.0517975137193414,
      "grad_norm": 1.3012638092041016,
      "learning_rate": 9.3865e-05,
      "loss": 0.4552,
      "step": 18783
    },
    {
      "epoch": 1.0518535110314704,
      "grad_norm": 1.4215017557144165,
      "learning_rate": 9.387000000000001e-05,
      "loss": 0.385,
      "step": 18784
    },
    {
      "epoch": 1.0519095083435994,
      "grad_norm": 1.4076894521713257,
      "learning_rate": 9.3875e-05,
      "loss": 0.4185,
      "step": 18785
    },
    {
      "epoch": 1.0519655056557284,
      "grad_norm": 1.6196259260177612,
      "learning_rate": 9.388e-05,
      "loss": 0.4357,
      "step": 18786
    },
    {
      "epoch": 1.0520215029678575,
      "grad_norm": 1.3829759359359741,
      "learning_rate": 9.3885e-05,
      "loss": 0.5274,
      "step": 18787
    },
    {
      "epoch": 1.0520775002799865,
      "grad_norm": 1.3827085494995117,
      "learning_rate": 9.389e-05,
      "loss": 0.3359,
      "step": 18788
    },
    {
      "epoch": 1.0521334975921155,
      "grad_norm": 1.3541009426116943,
      "learning_rate": 9.3895e-05,
      "loss": 0.3799,
      "step": 18789
    },
    {
      "epoch": 1.0521894949042445,
      "grad_norm": 7.985986232757568,
      "learning_rate": 9.39e-05,
      "loss": 0.478,
      "step": 18790
    },
    {
      "epoch": 1.0522454922163735,
      "grad_norm": 1.4385887384414673,
      "learning_rate": 9.3905e-05,
      "loss": 0.495,
      "step": 18791
    },
    {
      "epoch": 1.0523014895285026,
      "grad_norm": 1.43264639377594,
      "learning_rate": 9.391000000000001e-05,
      "loss": 0.492,
      "step": 18792
    },
    {
      "epoch": 1.0523574868406316,
      "grad_norm": 1.3586641550064087,
      "learning_rate": 9.391500000000001e-05,
      "loss": 0.4638,
      "step": 18793
    },
    {
      "epoch": 1.0524134841527606,
      "grad_norm": 1.3236429691314697,
      "learning_rate": 9.392000000000001e-05,
      "loss": 0.4102,
      "step": 18794
    },
    {
      "epoch": 1.0524694814648896,
      "grad_norm": 1.5813895463943481,
      "learning_rate": 9.3925e-05,
      "loss": 0.7058,
      "step": 18795
    },
    {
      "epoch": 1.0525254787770186,
      "grad_norm": 1.8021186590194702,
      "learning_rate": 9.393e-05,
      "loss": 0.3773,
      "step": 18796
    },
    {
      "epoch": 1.0525814760891476,
      "grad_norm": 1.561477780342102,
      "learning_rate": 9.3935e-05,
      "loss": 0.4236,
      "step": 18797
    },
    {
      "epoch": 1.0526374734012767,
      "grad_norm": 1.3825465440750122,
      "learning_rate": 9.394000000000001e-05,
      "loss": 0.4629,
      "step": 18798
    },
    {
      "epoch": 1.0526934707134057,
      "grad_norm": 2.319859504699707,
      "learning_rate": 9.394500000000001e-05,
      "loss": 0.3467,
      "step": 18799
    },
    {
      "epoch": 1.0527494680255347,
      "grad_norm": 1.4497575759887695,
      "learning_rate": 9.395000000000001e-05,
      "loss": 0.3631,
      "step": 18800
    },
    {
      "epoch": 1.0528054653376637,
      "grad_norm": 1.0492933988571167,
      "learning_rate": 9.3955e-05,
      "loss": 0.2956,
      "step": 18801
    },
    {
      "epoch": 1.0528614626497927,
      "grad_norm": 1.265881061553955,
      "learning_rate": 9.396e-05,
      "loss": 0.463,
      "step": 18802
    },
    {
      "epoch": 1.0529174599619218,
      "grad_norm": 1.474715232849121,
      "learning_rate": 9.3965e-05,
      "loss": 0.5241,
      "step": 18803
    },
    {
      "epoch": 1.0529734572740508,
      "grad_norm": 1.3244526386260986,
      "learning_rate": 9.397e-05,
      "loss": 0.3758,
      "step": 18804
    },
    {
      "epoch": 1.0530294545861798,
      "grad_norm": 1.293935775756836,
      "learning_rate": 9.397500000000001e-05,
      "loss": 0.3872,
      "step": 18805
    },
    {
      "epoch": 1.0530854518983088,
      "grad_norm": 1.3484982252120972,
      "learning_rate": 9.398e-05,
      "loss": 0.4226,
      "step": 18806
    },
    {
      "epoch": 1.0531414492104378,
      "grad_norm": 1.3209012746810913,
      "learning_rate": 9.3985e-05,
      "loss": 0.4594,
      "step": 18807
    },
    {
      "epoch": 1.0531974465225669,
      "grad_norm": 1.4305565357208252,
      "learning_rate": 9.399e-05,
      "loss": 0.4233,
      "step": 18808
    },
    {
      "epoch": 1.0532534438346959,
      "grad_norm": 1.4356799125671387,
      "learning_rate": 9.3995e-05,
      "loss": 0.4133,
      "step": 18809
    },
    {
      "epoch": 1.053309441146825,
      "grad_norm": 1.3230232000350952,
      "learning_rate": 9.4e-05,
      "loss": 0.4562,
      "step": 18810
    },
    {
      "epoch": 1.053365438458954,
      "grad_norm": 1.3221404552459717,
      "learning_rate": 9.4005e-05,
      "loss": 0.3312,
      "step": 18811
    },
    {
      "epoch": 1.053421435771083,
      "grad_norm": 1.6068238019943237,
      "learning_rate": 9.401e-05,
      "loss": 0.5273,
      "step": 18812
    },
    {
      "epoch": 1.053477433083212,
      "grad_norm": 1.308494210243225,
      "learning_rate": 9.401500000000001e-05,
      "loss": 0.5055,
      "step": 18813
    },
    {
      "epoch": 1.053533430395341,
      "grad_norm": 1.3897660970687866,
      "learning_rate": 9.402000000000001e-05,
      "loss": 0.4352,
      "step": 18814
    },
    {
      "epoch": 1.05358942770747,
      "grad_norm": 1.3756804466247559,
      "learning_rate": 9.402500000000001e-05,
      "loss": 0.5146,
      "step": 18815
    },
    {
      "epoch": 1.053645425019599,
      "grad_norm": 1.3395336866378784,
      "learning_rate": 9.403e-05,
      "loss": 0.5114,
      "step": 18816
    },
    {
      "epoch": 1.053701422331728,
      "grad_norm": 1.4863812923431396,
      "learning_rate": 9.4035e-05,
      "loss": 0.4387,
      "step": 18817
    },
    {
      "epoch": 1.053757419643857,
      "grad_norm": 1.3136694431304932,
      "learning_rate": 9.404e-05,
      "loss": 0.5058,
      "step": 18818
    },
    {
      "epoch": 1.053813416955986,
      "grad_norm": 1.463541865348816,
      "learning_rate": 9.404500000000001e-05,
      "loss": 0.372,
      "step": 18819
    },
    {
      "epoch": 1.053869414268115,
      "grad_norm": 1.429172158241272,
      "learning_rate": 9.405000000000001e-05,
      "loss": 0.5003,
      "step": 18820
    },
    {
      "epoch": 1.053925411580244,
      "grad_norm": 1.1957285404205322,
      "learning_rate": 9.4055e-05,
      "loss": 0.4128,
      "step": 18821
    },
    {
      "epoch": 1.0539814088923731,
      "grad_norm": 1.3483213186264038,
      "learning_rate": 9.406e-05,
      "loss": 0.4783,
      "step": 18822
    },
    {
      "epoch": 1.0540374062045021,
      "grad_norm": 1.742538332939148,
      "learning_rate": 9.4065e-05,
      "loss": 0.4073,
      "step": 18823
    },
    {
      "epoch": 1.0540934035166312,
      "grad_norm": 1.7641878128051758,
      "learning_rate": 9.407e-05,
      "loss": 0.4633,
      "step": 18824
    },
    {
      "epoch": 1.0541494008287602,
      "grad_norm": 1.4936950206756592,
      "learning_rate": 9.4075e-05,
      "loss": 0.5027,
      "step": 18825
    },
    {
      "epoch": 1.0542053981408892,
      "grad_norm": 1.5363181829452515,
      "learning_rate": 9.408000000000001e-05,
      "loss": 0.3967,
      "step": 18826
    },
    {
      "epoch": 1.0542613954530182,
      "grad_norm": 1.4092928171157837,
      "learning_rate": 9.4085e-05,
      "loss": 0.4312,
      "step": 18827
    },
    {
      "epoch": 1.0543173927651472,
      "grad_norm": 1.1858136653900146,
      "learning_rate": 9.409e-05,
      "loss": 0.5055,
      "step": 18828
    },
    {
      "epoch": 1.0543733900772763,
      "grad_norm": 1.7902591228485107,
      "learning_rate": 9.4095e-05,
      "loss": 0.479,
      "step": 18829
    },
    {
      "epoch": 1.0544293873894053,
      "grad_norm": 1.6157180070877075,
      "learning_rate": 9.41e-05,
      "loss": 0.369,
      "step": 18830
    },
    {
      "epoch": 1.0544853847015343,
      "grad_norm": 1.1800177097320557,
      "learning_rate": 9.410500000000001e-05,
      "loss": 0.332,
      "step": 18831
    },
    {
      "epoch": 1.0545413820136633,
      "grad_norm": 1.1747642755508423,
      "learning_rate": 9.411e-05,
      "loss": 0.4682,
      "step": 18832
    },
    {
      "epoch": 1.0545973793257923,
      "grad_norm": 1.514975905418396,
      "learning_rate": 9.4115e-05,
      "loss": 0.4393,
      "step": 18833
    },
    {
      "epoch": 1.0546533766379214,
      "grad_norm": 1.2441729307174683,
      "learning_rate": 9.412000000000001e-05,
      "loss": 0.4483,
      "step": 18834
    },
    {
      "epoch": 1.0547093739500504,
      "grad_norm": 1.2949442863464355,
      "learning_rate": 9.412500000000001e-05,
      "loss": 0.4996,
      "step": 18835
    },
    {
      "epoch": 1.0547653712621794,
      "grad_norm": 1.818418264389038,
      "learning_rate": 9.413000000000001e-05,
      "loss": 0.5126,
      "step": 18836
    },
    {
      "epoch": 1.0548213685743084,
      "grad_norm": 1.7463511228561401,
      "learning_rate": 9.4135e-05,
      "loss": 0.5482,
      "step": 18837
    },
    {
      "epoch": 1.0548773658864374,
      "grad_norm": 1.280035376548767,
      "learning_rate": 9.414e-05,
      "loss": 0.4009,
      "step": 18838
    },
    {
      "epoch": 1.0549333631985665,
      "grad_norm": 1.5921680927276611,
      "learning_rate": 9.4145e-05,
      "loss": 0.6184,
      "step": 18839
    },
    {
      "epoch": 1.0549893605106955,
      "grad_norm": 1.3982776403427124,
      "learning_rate": 9.415e-05,
      "loss": 0.3379,
      "step": 18840
    },
    {
      "epoch": 1.0550453578228245,
      "grad_norm": 1.1702890396118164,
      "learning_rate": 9.415500000000001e-05,
      "loss": 0.3119,
      "step": 18841
    },
    {
      "epoch": 1.0551013551349535,
      "grad_norm": 1.1738208532333374,
      "learning_rate": 9.416e-05,
      "loss": 0.3463,
      "step": 18842
    },
    {
      "epoch": 1.0551573524470825,
      "grad_norm": 1.2477097511291504,
      "learning_rate": 9.4165e-05,
      "loss": 0.4123,
      "step": 18843
    },
    {
      "epoch": 1.0552133497592115,
      "grad_norm": 1.3173540830612183,
      "learning_rate": 9.417e-05,
      "loss": 0.4082,
      "step": 18844
    },
    {
      "epoch": 1.0552693470713406,
      "grad_norm": 1.4024620056152344,
      "learning_rate": 9.4175e-05,
      "loss": 0.498,
      "step": 18845
    },
    {
      "epoch": 1.0553253443834696,
      "grad_norm": 1.6407623291015625,
      "learning_rate": 9.418e-05,
      "loss": 0.5054,
      "step": 18846
    },
    {
      "epoch": 1.0553813416955986,
      "grad_norm": 1.3435959815979004,
      "learning_rate": 9.418500000000001e-05,
      "loss": 0.5827,
      "step": 18847
    },
    {
      "epoch": 1.0554373390077276,
      "grad_norm": 1.2907919883728027,
      "learning_rate": 9.419e-05,
      "loss": 0.4695,
      "step": 18848
    },
    {
      "epoch": 1.0554933363198566,
      "grad_norm": 1.2824134826660156,
      "learning_rate": 9.4195e-05,
      "loss": 0.4491,
      "step": 18849
    },
    {
      "epoch": 1.0555493336319857,
      "grad_norm": 1.4175795316696167,
      "learning_rate": 9.42e-05,
      "loss": 0.4447,
      "step": 18850
    },
    {
      "epoch": 1.0556053309441147,
      "grad_norm": 2.4582059383392334,
      "learning_rate": 9.420500000000001e-05,
      "loss": 0.4851,
      "step": 18851
    },
    {
      "epoch": 1.0556613282562437,
      "grad_norm": 1.510203242301941,
      "learning_rate": 9.421000000000001e-05,
      "loss": 0.4234,
      "step": 18852
    },
    {
      "epoch": 1.0557173255683727,
      "grad_norm": 1.3534705638885498,
      "learning_rate": 9.4215e-05,
      "loss": 0.4249,
      "step": 18853
    },
    {
      "epoch": 1.0557733228805017,
      "grad_norm": 1.524954915046692,
      "learning_rate": 9.422e-05,
      "loss": 0.5413,
      "step": 18854
    },
    {
      "epoch": 1.0558293201926308,
      "grad_norm": 1.2934017181396484,
      "learning_rate": 9.422500000000001e-05,
      "loss": 0.4422,
      "step": 18855
    },
    {
      "epoch": 1.0558853175047598,
      "grad_norm": 1.3412871360778809,
      "learning_rate": 9.423000000000001e-05,
      "loss": 0.4235,
      "step": 18856
    },
    {
      "epoch": 1.0559413148168888,
      "grad_norm": 1.5189380645751953,
      "learning_rate": 9.423500000000001e-05,
      "loss": 0.4656,
      "step": 18857
    },
    {
      "epoch": 1.0559973121290178,
      "grad_norm": 1.3191028833389282,
      "learning_rate": 9.424e-05,
      "loss": 0.4504,
      "step": 18858
    },
    {
      "epoch": 1.0560533094411468,
      "grad_norm": 1.3765827417373657,
      "learning_rate": 9.4245e-05,
      "loss": 0.4747,
      "step": 18859
    },
    {
      "epoch": 1.0561093067532759,
      "grad_norm": 1.6227412223815918,
      "learning_rate": 9.425e-05,
      "loss": 0.488,
      "step": 18860
    },
    {
      "epoch": 1.0561653040654049,
      "grad_norm": 2.1638755798339844,
      "learning_rate": 9.4255e-05,
      "loss": 0.4988,
      "step": 18861
    },
    {
      "epoch": 1.056221301377534,
      "grad_norm": 1.5395900011062622,
      "learning_rate": 9.426000000000001e-05,
      "loss": 0.594,
      "step": 18862
    },
    {
      "epoch": 1.056277298689663,
      "grad_norm": 1.609810709953308,
      "learning_rate": 9.4265e-05,
      "loss": 0.4513,
      "step": 18863
    },
    {
      "epoch": 1.056333296001792,
      "grad_norm": 1.3589756488800049,
      "learning_rate": 9.427e-05,
      "loss": 0.4122,
      "step": 18864
    },
    {
      "epoch": 1.056389293313921,
      "grad_norm": 1.2109688520431519,
      "learning_rate": 9.4275e-05,
      "loss": 0.3998,
      "step": 18865
    },
    {
      "epoch": 1.05644529062605,
      "grad_norm": 1.3328644037246704,
      "learning_rate": 9.428e-05,
      "loss": 0.447,
      "step": 18866
    },
    {
      "epoch": 1.056501287938179,
      "grad_norm": 1.28719961643219,
      "learning_rate": 9.4285e-05,
      "loss": 0.4122,
      "step": 18867
    },
    {
      "epoch": 1.056557285250308,
      "grad_norm": 1.3460605144500732,
      "learning_rate": 9.429000000000001e-05,
      "loss": 0.4158,
      "step": 18868
    },
    {
      "epoch": 1.056613282562437,
      "grad_norm": 1.3511147499084473,
      "learning_rate": 9.4295e-05,
      "loss": 0.4722,
      "step": 18869
    },
    {
      "epoch": 1.056669279874566,
      "grad_norm": 1.66860830783844,
      "learning_rate": 9.43e-05,
      "loss": 0.4749,
      "step": 18870
    },
    {
      "epoch": 1.056725277186695,
      "grad_norm": 1.2137311697006226,
      "learning_rate": 9.430500000000001e-05,
      "loss": 0.3863,
      "step": 18871
    },
    {
      "epoch": 1.056781274498824,
      "grad_norm": 1.441867709159851,
      "learning_rate": 9.431000000000001e-05,
      "loss": 0.5332,
      "step": 18872
    },
    {
      "epoch": 1.056837271810953,
      "grad_norm": 1.3843780755996704,
      "learning_rate": 9.431500000000001e-05,
      "loss": 0.3801,
      "step": 18873
    },
    {
      "epoch": 1.0568932691230821,
      "grad_norm": 1.5861767530441284,
      "learning_rate": 9.432e-05,
      "loss": 0.478,
      "step": 18874
    },
    {
      "epoch": 1.0569492664352111,
      "grad_norm": 1.8049677610397339,
      "learning_rate": 9.4325e-05,
      "loss": 0.5031,
      "step": 18875
    },
    {
      "epoch": 1.0570052637473402,
      "grad_norm": 1.302350640296936,
      "learning_rate": 9.433000000000001e-05,
      "loss": 0.5332,
      "step": 18876
    },
    {
      "epoch": 1.0570612610594692,
      "grad_norm": 1.555769920349121,
      "learning_rate": 9.433500000000001e-05,
      "loss": 0.4586,
      "step": 18877
    },
    {
      "epoch": 1.0571172583715982,
      "grad_norm": 1.334062933921814,
      "learning_rate": 9.434000000000001e-05,
      "loss": 0.4274,
      "step": 18878
    },
    {
      "epoch": 1.0571732556837272,
      "grad_norm": 1.4169632196426392,
      "learning_rate": 9.4345e-05,
      "loss": 0.4329,
      "step": 18879
    },
    {
      "epoch": 1.0572292529958562,
      "grad_norm": 1.3843140602111816,
      "learning_rate": 9.435e-05,
      "loss": 0.4196,
      "step": 18880
    },
    {
      "epoch": 1.0572852503079853,
      "grad_norm": 1.5343679189682007,
      "learning_rate": 9.4355e-05,
      "loss": 0.434,
      "step": 18881
    },
    {
      "epoch": 1.0573412476201143,
      "grad_norm": 1.3447123765945435,
      "learning_rate": 9.436e-05,
      "loss": 0.4374,
      "step": 18882
    },
    {
      "epoch": 1.0573972449322433,
      "grad_norm": 1.2755141258239746,
      "learning_rate": 9.436500000000001e-05,
      "loss": 0.497,
      "step": 18883
    },
    {
      "epoch": 1.0574532422443723,
      "grad_norm": 1.4663509130477905,
      "learning_rate": 9.437e-05,
      "loss": 0.449,
      "step": 18884
    },
    {
      "epoch": 1.0575092395565013,
      "grad_norm": 1.2889498472213745,
      "learning_rate": 9.4375e-05,
      "loss": 0.4059,
      "step": 18885
    },
    {
      "epoch": 1.0575652368686304,
      "grad_norm": 1.7589125633239746,
      "learning_rate": 9.438e-05,
      "loss": 0.3877,
      "step": 18886
    },
    {
      "epoch": 1.0576212341807594,
      "grad_norm": 1.254963994026184,
      "learning_rate": 9.4385e-05,
      "loss": 0.3096,
      "step": 18887
    },
    {
      "epoch": 1.0576772314928884,
      "grad_norm": 1.3054019212722778,
      "learning_rate": 9.439e-05,
      "loss": 0.3835,
      "step": 18888
    },
    {
      "epoch": 1.0577332288050174,
      "grad_norm": 1.5005621910095215,
      "learning_rate": 9.439500000000001e-05,
      "loss": 0.4646,
      "step": 18889
    },
    {
      "epoch": 1.0577892261171464,
      "grad_norm": 1.1724367141723633,
      "learning_rate": 9.44e-05,
      "loss": 0.4213,
      "step": 18890
    },
    {
      "epoch": 1.0578452234292754,
      "grad_norm": 1.463555097579956,
      "learning_rate": 9.4405e-05,
      "loss": 0.4219,
      "step": 18891
    },
    {
      "epoch": 1.0579012207414045,
      "grad_norm": 1.779472827911377,
      "learning_rate": 9.441000000000001e-05,
      "loss": 0.3379,
      "step": 18892
    },
    {
      "epoch": 1.0579572180535335,
      "grad_norm": 1.2690653800964355,
      "learning_rate": 9.441500000000001e-05,
      "loss": 0.3729,
      "step": 18893
    },
    {
      "epoch": 1.0580132153656625,
      "grad_norm": 1.3225834369659424,
      "learning_rate": 9.442000000000001e-05,
      "loss": 0.4082,
      "step": 18894
    },
    {
      "epoch": 1.0580692126777915,
      "grad_norm": 1.572891354560852,
      "learning_rate": 9.4425e-05,
      "loss": 0.3969,
      "step": 18895
    },
    {
      "epoch": 1.0581252099899205,
      "grad_norm": 1.2776237726211548,
      "learning_rate": 9.443e-05,
      "loss": 0.4541,
      "step": 18896
    },
    {
      "epoch": 1.0581812073020496,
      "grad_norm": 1.635740041732788,
      "learning_rate": 9.443500000000001e-05,
      "loss": 0.4054,
      "step": 18897
    },
    {
      "epoch": 1.0582372046141786,
      "grad_norm": 1.3353666067123413,
      "learning_rate": 9.444000000000001e-05,
      "loss": 0.4354,
      "step": 18898
    },
    {
      "epoch": 1.0582932019263076,
      "grad_norm": 1.283207654953003,
      "learning_rate": 9.444500000000001e-05,
      "loss": 0.4254,
      "step": 18899
    },
    {
      "epoch": 1.0583491992384366,
      "grad_norm": 1.3528674840927124,
      "learning_rate": 9.445e-05,
      "loss": 0.3926,
      "step": 18900
    },
    {
      "epoch": 1.0584051965505656,
      "grad_norm": 1.2042877674102783,
      "learning_rate": 9.4455e-05,
      "loss": 0.4369,
      "step": 18901
    },
    {
      "epoch": 1.0584611938626947,
      "grad_norm": 2.829955816268921,
      "learning_rate": 9.446e-05,
      "loss": 0.4321,
      "step": 18902
    },
    {
      "epoch": 1.0585171911748237,
      "grad_norm": 1.424607515335083,
      "learning_rate": 9.4465e-05,
      "loss": 0.3693,
      "step": 18903
    },
    {
      "epoch": 1.0585731884869527,
      "grad_norm": 1.0847461223602295,
      "learning_rate": 9.447000000000001e-05,
      "loss": 0.3607,
      "step": 18904
    },
    {
      "epoch": 1.0586291857990817,
      "grad_norm": 1.390052080154419,
      "learning_rate": 9.4475e-05,
      "loss": 0.4118,
      "step": 18905
    },
    {
      "epoch": 1.0586851831112107,
      "grad_norm": 1.8173706531524658,
      "learning_rate": 9.448e-05,
      "loss": 0.5275,
      "step": 18906
    },
    {
      "epoch": 1.0587411804233398,
      "grad_norm": 1.606350302696228,
      "learning_rate": 9.4485e-05,
      "loss": 0.3748,
      "step": 18907
    },
    {
      "epoch": 1.0587971777354688,
      "grad_norm": 1.3614684343338013,
      "learning_rate": 9.449e-05,
      "loss": 0.384,
      "step": 18908
    },
    {
      "epoch": 1.0588531750475978,
      "grad_norm": 1.325899600982666,
      "learning_rate": 9.4495e-05,
      "loss": 0.3822,
      "step": 18909
    },
    {
      "epoch": 1.0589091723597268,
      "grad_norm": 1.5942211151123047,
      "learning_rate": 9.449999999999999e-05,
      "loss": 0.3482,
      "step": 18910
    },
    {
      "epoch": 1.0589651696718558,
      "grad_norm": 1.237679362297058,
      "learning_rate": 9.4505e-05,
      "loss": 0.3954,
      "step": 18911
    },
    {
      "epoch": 1.0590211669839849,
      "grad_norm": 1.4813698530197144,
      "learning_rate": 9.451000000000002e-05,
      "loss": 0.4544,
      "step": 18912
    },
    {
      "epoch": 1.0590771642961139,
      "grad_norm": 1.36385977268219,
      "learning_rate": 9.451500000000001e-05,
      "loss": 0.4298,
      "step": 18913
    },
    {
      "epoch": 1.059133161608243,
      "grad_norm": 1.14071786403656,
      "learning_rate": 9.452000000000001e-05,
      "loss": 0.4289,
      "step": 18914
    },
    {
      "epoch": 1.059189158920372,
      "grad_norm": 1.6904393434524536,
      "learning_rate": 9.452500000000001e-05,
      "loss": 0.4813,
      "step": 18915
    },
    {
      "epoch": 1.059245156232501,
      "grad_norm": 1.3985432386398315,
      "learning_rate": 9.453e-05,
      "loss": 0.3883,
      "step": 18916
    },
    {
      "epoch": 1.05930115354463,
      "grad_norm": 1.5503175258636475,
      "learning_rate": 9.4535e-05,
      "loss": 0.3758,
      "step": 18917
    },
    {
      "epoch": 1.059357150856759,
      "grad_norm": 1.3641659021377563,
      "learning_rate": 9.454000000000001e-05,
      "loss": 0.4202,
      "step": 18918
    },
    {
      "epoch": 1.059413148168888,
      "grad_norm": 1.790743350982666,
      "learning_rate": 9.454500000000001e-05,
      "loss": 0.3753,
      "step": 18919
    },
    {
      "epoch": 1.059469145481017,
      "grad_norm": 1.3057456016540527,
      "learning_rate": 9.455000000000001e-05,
      "loss": 0.4931,
      "step": 18920
    },
    {
      "epoch": 1.059525142793146,
      "grad_norm": 1.4196068048477173,
      "learning_rate": 9.4555e-05,
      "loss": 0.399,
      "step": 18921
    },
    {
      "epoch": 1.059581140105275,
      "grad_norm": 1.8194833993911743,
      "learning_rate": 9.456e-05,
      "loss": 0.4895,
      "step": 18922
    },
    {
      "epoch": 1.059637137417404,
      "grad_norm": 1.3155800104141235,
      "learning_rate": 9.4565e-05,
      "loss": 0.4106,
      "step": 18923
    },
    {
      "epoch": 1.059693134729533,
      "grad_norm": 1.8820222616195679,
      "learning_rate": 9.457e-05,
      "loss": 0.4736,
      "step": 18924
    },
    {
      "epoch": 1.059749132041662,
      "grad_norm": 1.1942620277404785,
      "learning_rate": 9.457500000000001e-05,
      "loss": 0.3874,
      "step": 18925
    },
    {
      "epoch": 1.0598051293537911,
      "grad_norm": 1.4331386089324951,
      "learning_rate": 9.458e-05,
      "loss": 0.5039,
      "step": 18926
    },
    {
      "epoch": 1.0598611266659201,
      "grad_norm": 1.2756425142288208,
      "learning_rate": 9.4585e-05,
      "loss": 0.5379,
      "step": 18927
    },
    {
      "epoch": 1.0599171239780492,
      "grad_norm": 1.3890256881713867,
      "learning_rate": 9.459e-05,
      "loss": 0.3719,
      "step": 18928
    },
    {
      "epoch": 1.0599731212901782,
      "grad_norm": 1.3086320161819458,
      "learning_rate": 9.4595e-05,
      "loss": 0.5043,
      "step": 18929
    },
    {
      "epoch": 1.0600291186023072,
      "grad_norm": 1.5591317415237427,
      "learning_rate": 9.46e-05,
      "loss": 0.3195,
      "step": 18930
    },
    {
      "epoch": 1.0600851159144362,
      "grad_norm": 1.288242220878601,
      "learning_rate": 9.460499999999999e-05,
      "loss": 0.3377,
      "step": 18931
    },
    {
      "epoch": 1.0601411132265652,
      "grad_norm": 1.243438720703125,
      "learning_rate": 9.461e-05,
      "loss": 0.3815,
      "step": 18932
    },
    {
      "epoch": 1.0601971105386943,
      "grad_norm": 1.3783605098724365,
      "learning_rate": 9.461500000000001e-05,
      "loss": 0.3725,
      "step": 18933
    },
    {
      "epoch": 1.0602531078508233,
      "grad_norm": 1.2204675674438477,
      "learning_rate": 9.462000000000001e-05,
      "loss": 0.4538,
      "step": 18934
    },
    {
      "epoch": 1.0603091051629523,
      "grad_norm": 1.437117099761963,
      "learning_rate": 9.462500000000001e-05,
      "loss": 0.4731,
      "step": 18935
    },
    {
      "epoch": 1.060365102475081,
      "grad_norm": 1.4627983570098877,
      "learning_rate": 9.463000000000001e-05,
      "loss": 0.4642,
      "step": 18936
    },
    {
      "epoch": 1.06042109978721,
      "grad_norm": 1.3718695640563965,
      "learning_rate": 9.4635e-05,
      "loss": 0.5397,
      "step": 18937
    },
    {
      "epoch": 1.0604770970993391,
      "grad_norm": 1.5296236276626587,
      "learning_rate": 9.464e-05,
      "loss": 0.4356,
      "step": 18938
    },
    {
      "epoch": 1.0605330944114681,
      "grad_norm": 1.440040111541748,
      "learning_rate": 9.4645e-05,
      "loss": 0.5243,
      "step": 18939
    },
    {
      "epoch": 1.0605890917235972,
      "grad_norm": 21.09760284423828,
      "learning_rate": 9.465000000000001e-05,
      "loss": 0.3917,
      "step": 18940
    },
    {
      "epoch": 1.0606450890357262,
      "grad_norm": 1.457410216331482,
      "learning_rate": 9.465500000000001e-05,
      "loss": 0.3693,
      "step": 18941
    },
    {
      "epoch": 1.0607010863478552,
      "grad_norm": 1.5546989440917969,
      "learning_rate": 9.466e-05,
      "loss": 0.4964,
      "step": 18942
    },
    {
      "epoch": 1.0607570836599842,
      "grad_norm": 1.2661175727844238,
      "learning_rate": 9.4665e-05,
      "loss": 0.4675,
      "step": 18943
    },
    {
      "epoch": 1.0608130809721132,
      "grad_norm": 1.2626347541809082,
      "learning_rate": 9.467e-05,
      "loss": 0.3433,
      "step": 18944
    },
    {
      "epoch": 1.0608690782842423,
      "grad_norm": 1.317460298538208,
      "learning_rate": 9.4675e-05,
      "loss": 0.35,
      "step": 18945
    },
    {
      "epoch": 1.0609250755963713,
      "grad_norm": 1.5489500761032104,
      "learning_rate": 9.468000000000001e-05,
      "loss": 0.4688,
      "step": 18946
    },
    {
      "epoch": 1.0609810729085003,
      "grad_norm": 1.44960618019104,
      "learning_rate": 9.4685e-05,
      "loss": 0.4104,
      "step": 18947
    },
    {
      "epoch": 1.0610370702206293,
      "grad_norm": 1.649051308631897,
      "learning_rate": 9.469e-05,
      "loss": 0.3846,
      "step": 18948
    },
    {
      "epoch": 1.0610930675327583,
      "grad_norm": 1.4297840595245361,
      "learning_rate": 9.4695e-05,
      "loss": 0.4522,
      "step": 18949
    },
    {
      "epoch": 1.0611490648448874,
      "grad_norm": 1.7101448774337769,
      "learning_rate": 9.47e-05,
      "loss": 0.4883,
      "step": 18950
    },
    {
      "epoch": 1.0612050621570164,
      "grad_norm": 1.4615607261657715,
      "learning_rate": 9.4705e-05,
      "loss": 0.3946,
      "step": 18951
    },
    {
      "epoch": 1.0612610594691454,
      "grad_norm": 1.315509557723999,
      "learning_rate": 9.471e-05,
      "loss": 0.491,
      "step": 18952
    },
    {
      "epoch": 1.0613170567812744,
      "grad_norm": 1.490499496459961,
      "learning_rate": 9.4715e-05,
      "loss": 0.4898,
      "step": 18953
    },
    {
      "epoch": 1.0613730540934034,
      "grad_norm": 1.748432993888855,
      "learning_rate": 9.472000000000001e-05,
      "loss": 0.5113,
      "step": 18954
    },
    {
      "epoch": 1.0614290514055325,
      "grad_norm": 1.2763144969940186,
      "learning_rate": 9.472500000000001e-05,
      "loss": 0.4567,
      "step": 18955
    },
    {
      "epoch": 1.0614850487176615,
      "grad_norm": 1.5092469453811646,
      "learning_rate": 9.473000000000001e-05,
      "loss": 0.4869,
      "step": 18956
    },
    {
      "epoch": 1.0615410460297905,
      "grad_norm": 1.5785478353500366,
      "learning_rate": 9.473500000000001e-05,
      "loss": 0.5633,
      "step": 18957
    },
    {
      "epoch": 1.0615970433419195,
      "grad_norm": 1.1675353050231934,
      "learning_rate": 9.474e-05,
      "loss": 0.42,
      "step": 18958
    },
    {
      "epoch": 1.0616530406540485,
      "grad_norm": 1.4426264762878418,
      "learning_rate": 9.4745e-05,
      "loss": 0.5753,
      "step": 18959
    },
    {
      "epoch": 1.0617090379661775,
      "grad_norm": 2.0994317531585693,
      "learning_rate": 9.475e-05,
      "loss": 0.4554,
      "step": 18960
    },
    {
      "epoch": 1.0617650352783066,
      "grad_norm": 1.3798431158065796,
      "learning_rate": 9.475500000000001e-05,
      "loss": 0.3845,
      "step": 18961
    },
    {
      "epoch": 1.0618210325904356,
      "grad_norm": 1.6509004831314087,
      "learning_rate": 9.476000000000001e-05,
      "loss": 0.5474,
      "step": 18962
    },
    {
      "epoch": 1.0618770299025646,
      "grad_norm": 1.4672154188156128,
      "learning_rate": 9.4765e-05,
      "loss": 0.5159,
      "step": 18963
    },
    {
      "epoch": 1.0619330272146936,
      "grad_norm": 1.2632747888565063,
      "learning_rate": 9.477e-05,
      "loss": 0.4353,
      "step": 18964
    },
    {
      "epoch": 1.0619890245268226,
      "grad_norm": 1.3404821157455444,
      "learning_rate": 9.4775e-05,
      "loss": 0.3383,
      "step": 18965
    },
    {
      "epoch": 1.0620450218389517,
      "grad_norm": 1.3777574300765991,
      "learning_rate": 9.478e-05,
      "loss": 0.5561,
      "step": 18966
    },
    {
      "epoch": 1.0621010191510807,
      "grad_norm": 1.2554987668991089,
      "learning_rate": 9.478500000000001e-05,
      "loss": 0.3572,
      "step": 18967
    },
    {
      "epoch": 1.0621570164632097,
      "grad_norm": 1.453681468963623,
      "learning_rate": 9.479e-05,
      "loss": 0.3301,
      "step": 18968
    },
    {
      "epoch": 1.0622130137753387,
      "grad_norm": 1.1490294933319092,
      "learning_rate": 9.4795e-05,
      "loss": 0.3352,
      "step": 18969
    },
    {
      "epoch": 1.0622690110874677,
      "grad_norm": 2.009647846221924,
      "learning_rate": 9.48e-05,
      "loss": 0.6631,
      "step": 18970
    },
    {
      "epoch": 1.0623250083995968,
      "grad_norm": 1.4922118186950684,
      "learning_rate": 9.4805e-05,
      "loss": 0.5982,
      "step": 18971
    },
    {
      "epoch": 1.0623810057117258,
      "grad_norm": 1.34416663646698,
      "learning_rate": 9.481000000000001e-05,
      "loss": 0.4867,
      "step": 18972
    },
    {
      "epoch": 1.0624370030238548,
      "grad_norm": 1.2582811117172241,
      "learning_rate": 9.4815e-05,
      "loss": 0.3392,
      "step": 18973
    },
    {
      "epoch": 1.0624930003359838,
      "grad_norm": 1.1551052331924438,
      "learning_rate": 9.482e-05,
      "loss": 0.3629,
      "step": 18974
    },
    {
      "epoch": 1.0625489976481128,
      "grad_norm": 1.3015782833099365,
      "learning_rate": 9.482500000000001e-05,
      "loss": 0.3792,
      "step": 18975
    },
    {
      "epoch": 1.0626049949602419,
      "grad_norm": 1.3310070037841797,
      "learning_rate": 9.483000000000001e-05,
      "loss": 0.3958,
      "step": 18976
    },
    {
      "epoch": 1.0626609922723709,
      "grad_norm": 1.3582322597503662,
      "learning_rate": 9.483500000000001e-05,
      "loss": 0.3922,
      "step": 18977
    },
    {
      "epoch": 1.0627169895845,
      "grad_norm": 1.480780005455017,
      "learning_rate": 9.484e-05,
      "loss": 0.3405,
      "step": 18978
    },
    {
      "epoch": 1.062772986896629,
      "grad_norm": 1.2930588722229004,
      "learning_rate": 9.4845e-05,
      "loss": 0.4571,
      "step": 18979
    },
    {
      "epoch": 1.062828984208758,
      "grad_norm": 1.5896929502487183,
      "learning_rate": 9.485e-05,
      "loss": 0.5357,
      "step": 18980
    },
    {
      "epoch": 1.062884981520887,
      "grad_norm": 1.4487299919128418,
      "learning_rate": 9.4855e-05,
      "loss": 0.5018,
      "step": 18981
    },
    {
      "epoch": 1.062940978833016,
      "grad_norm": 1.6484873294830322,
      "learning_rate": 9.486000000000001e-05,
      "loss": 0.6373,
      "step": 18982
    },
    {
      "epoch": 1.062996976145145,
      "grad_norm": 1.294859766960144,
      "learning_rate": 9.486500000000001e-05,
      "loss": 0.3981,
      "step": 18983
    },
    {
      "epoch": 1.063052973457274,
      "grad_norm": 1.2828603982925415,
      "learning_rate": 9.487e-05,
      "loss": 0.5343,
      "step": 18984
    },
    {
      "epoch": 1.063108970769403,
      "grad_norm": 1.579521656036377,
      "learning_rate": 9.4875e-05,
      "loss": 0.4794,
      "step": 18985
    },
    {
      "epoch": 1.063164968081532,
      "grad_norm": 5.587820529937744,
      "learning_rate": 9.488e-05,
      "loss": 0.3993,
      "step": 18986
    },
    {
      "epoch": 1.063220965393661,
      "grad_norm": 1.2712122201919556,
      "learning_rate": 9.4885e-05,
      "loss": 0.3959,
      "step": 18987
    },
    {
      "epoch": 1.06327696270579,
      "grad_norm": 1.1356114149093628,
      "learning_rate": 9.489e-05,
      "loss": 0.4589,
      "step": 18988
    },
    {
      "epoch": 1.063332960017919,
      "grad_norm": 2.2960779666900635,
      "learning_rate": 9.4895e-05,
      "loss": 0.2882,
      "step": 18989
    },
    {
      "epoch": 1.0633889573300481,
      "grad_norm": 1.2106236219406128,
      "learning_rate": 9.49e-05,
      "loss": 0.3994,
      "step": 18990
    },
    {
      "epoch": 1.0634449546421771,
      "grad_norm": 1.1267436742782593,
      "learning_rate": 9.4905e-05,
      "loss": 0.3675,
      "step": 18991
    },
    {
      "epoch": 1.0635009519543062,
      "grad_norm": 1.2601475715637207,
      "learning_rate": 9.491000000000001e-05,
      "loss": 0.3452,
      "step": 18992
    },
    {
      "epoch": 1.0635569492664352,
      "grad_norm": 1.4580140113830566,
      "learning_rate": 9.491500000000001e-05,
      "loss": 0.4447,
      "step": 18993
    },
    {
      "epoch": 1.0636129465785642,
      "grad_norm": 1.2755945920944214,
      "learning_rate": 9.492e-05,
      "loss": 0.4647,
      "step": 18994
    },
    {
      "epoch": 1.0636689438906932,
      "grad_norm": 1.2138370275497437,
      "learning_rate": 9.4925e-05,
      "loss": 0.387,
      "step": 18995
    },
    {
      "epoch": 1.0637249412028222,
      "grad_norm": 1.2418153285980225,
      "learning_rate": 9.493000000000001e-05,
      "loss": 0.3565,
      "step": 18996
    },
    {
      "epoch": 1.0637809385149513,
      "grad_norm": 1.2355155944824219,
      "learning_rate": 9.493500000000001e-05,
      "loss": 0.4235,
      "step": 18997
    },
    {
      "epoch": 1.0638369358270803,
      "grad_norm": 1.5285604000091553,
      "learning_rate": 9.494000000000001e-05,
      "loss": 0.3818,
      "step": 18998
    },
    {
      "epoch": 1.0638929331392093,
      "grad_norm": 1.3427233695983887,
      "learning_rate": 9.4945e-05,
      "loss": 0.4629,
      "step": 18999
    },
    {
      "epoch": 1.0639489304513383,
      "grad_norm": 1.4263298511505127,
      "learning_rate": 9.495e-05,
      "loss": 0.5107,
      "step": 19000
    },
    {
      "epoch": 1.0640049277634673,
      "grad_norm": 1.3277918100357056,
      "learning_rate": 9.4955e-05,
      "loss": 0.4914,
      "step": 19001
    },
    {
      "epoch": 1.0640609250755964,
      "grad_norm": 1.5316247940063477,
      "learning_rate": 9.496e-05,
      "loss": 0.5536,
      "step": 19002
    },
    {
      "epoch": 1.0641169223877254,
      "grad_norm": 1.4550143480300903,
      "learning_rate": 9.496500000000001e-05,
      "loss": 0.3668,
      "step": 19003
    },
    {
      "epoch": 1.0641729196998544,
      "grad_norm": 1.1481560468673706,
      "learning_rate": 9.497000000000001e-05,
      "loss": 0.3922,
      "step": 19004
    },
    {
      "epoch": 1.0642289170119834,
      "grad_norm": 1.3690632581710815,
      "learning_rate": 9.4975e-05,
      "loss": 0.5181,
      "step": 19005
    },
    {
      "epoch": 1.0642849143241124,
      "grad_norm": 1.583183765411377,
      "learning_rate": 9.498e-05,
      "loss": 0.4482,
      "step": 19006
    },
    {
      "epoch": 1.0643409116362414,
      "grad_norm": 1.3478230237960815,
      "learning_rate": 9.4985e-05,
      "loss": 0.3658,
      "step": 19007
    },
    {
      "epoch": 1.0643969089483705,
      "grad_norm": 3.4860141277313232,
      "learning_rate": 9.499e-05,
      "loss": 0.4589,
      "step": 19008
    },
    {
      "epoch": 1.0644529062604995,
      "grad_norm": 1.381873369216919,
      "learning_rate": 9.4995e-05,
      "loss": 0.4322,
      "step": 19009
    },
    {
      "epoch": 1.0645089035726285,
      "grad_norm": 1.5568053722381592,
      "learning_rate": 9.5e-05,
      "loss": 0.3266,
      "step": 19010
    },
    {
      "epoch": 1.0645649008847575,
      "grad_norm": 1.5077698230743408,
      "learning_rate": 9.5005e-05,
      "loss": 0.3767,
      "step": 19011
    },
    {
      "epoch": 1.0646208981968865,
      "grad_norm": 1.1655912399291992,
      "learning_rate": 9.501e-05,
      "loss": 0.3115,
      "step": 19012
    },
    {
      "epoch": 1.0646768955090156,
      "grad_norm": 1.681504487991333,
      "learning_rate": 9.501500000000001e-05,
      "loss": 0.4994,
      "step": 19013
    },
    {
      "epoch": 1.0647328928211446,
      "grad_norm": 1.292641520500183,
      "learning_rate": 9.502000000000001e-05,
      "loss": 0.4135,
      "step": 19014
    },
    {
      "epoch": 1.0647888901332736,
      "grad_norm": 1.6553596258163452,
      "learning_rate": 9.5025e-05,
      "loss": 0.4628,
      "step": 19015
    },
    {
      "epoch": 1.0648448874454026,
      "grad_norm": 1.3552294969558716,
      "learning_rate": 9.503e-05,
      "loss": 0.4012,
      "step": 19016
    },
    {
      "epoch": 1.0649008847575316,
      "grad_norm": 1.2220262289047241,
      "learning_rate": 9.5035e-05,
      "loss": 0.4394,
      "step": 19017
    },
    {
      "epoch": 1.0649568820696607,
      "grad_norm": 1.2594716548919678,
      "learning_rate": 9.504000000000001e-05,
      "loss": 0.3832,
      "step": 19018
    },
    {
      "epoch": 1.0650128793817897,
      "grad_norm": 1.3809083700180054,
      "learning_rate": 9.504500000000001e-05,
      "loss": 0.5147,
      "step": 19019
    },
    {
      "epoch": 1.0650688766939187,
      "grad_norm": 1.7652674913406372,
      "learning_rate": 9.505e-05,
      "loss": 0.3956,
      "step": 19020
    },
    {
      "epoch": 1.0651248740060477,
      "grad_norm": 1.3767143487930298,
      "learning_rate": 9.5055e-05,
      "loss": 0.4599,
      "step": 19021
    },
    {
      "epoch": 1.0651808713181767,
      "grad_norm": 1.350810170173645,
      "learning_rate": 9.506e-05,
      "loss": 0.4033,
      "step": 19022
    },
    {
      "epoch": 1.0652368686303058,
      "grad_norm": 2.0616953372955322,
      "learning_rate": 9.5065e-05,
      "loss": 0.4707,
      "step": 19023
    },
    {
      "epoch": 1.0652928659424348,
      "grad_norm": 1.2971410751342773,
      "learning_rate": 9.507000000000001e-05,
      "loss": 0.4084,
      "step": 19024
    },
    {
      "epoch": 1.0653488632545638,
      "grad_norm": 1.3095693588256836,
      "learning_rate": 9.507500000000001e-05,
      "loss": 0.3893,
      "step": 19025
    },
    {
      "epoch": 1.0654048605666928,
      "grad_norm": 1.9630202054977417,
      "learning_rate": 9.508e-05,
      "loss": 0.4432,
      "step": 19026
    },
    {
      "epoch": 1.0654608578788218,
      "grad_norm": 1.2579559087753296,
      "learning_rate": 9.5085e-05,
      "loss": 0.5054,
      "step": 19027
    },
    {
      "epoch": 1.0655168551909509,
      "grad_norm": 1.3713284730911255,
      "learning_rate": 9.509e-05,
      "loss": 0.4316,
      "step": 19028
    },
    {
      "epoch": 1.0655728525030799,
      "grad_norm": 1.4795838594436646,
      "learning_rate": 9.5095e-05,
      "loss": 0.4322,
      "step": 19029
    },
    {
      "epoch": 1.065628849815209,
      "grad_norm": 1.1935054063796997,
      "learning_rate": 9.51e-05,
      "loss": 0.3195,
      "step": 19030
    },
    {
      "epoch": 1.065684847127338,
      "grad_norm": 1.5417760610580444,
      "learning_rate": 9.5105e-05,
      "loss": 0.4727,
      "step": 19031
    },
    {
      "epoch": 1.065740844439467,
      "grad_norm": 1.6064913272857666,
      "learning_rate": 9.511e-05,
      "loss": 0.4643,
      "step": 19032
    },
    {
      "epoch": 1.065796841751596,
      "grad_norm": 1.329726219177246,
      "learning_rate": 9.511500000000001e-05,
      "loss": 0.4171,
      "step": 19033
    },
    {
      "epoch": 1.065852839063725,
      "grad_norm": 1.7679561376571655,
      "learning_rate": 9.512000000000001e-05,
      "loss": 0.5562,
      "step": 19034
    },
    {
      "epoch": 1.065908836375854,
      "grad_norm": 1.3980681896209717,
      "learning_rate": 9.512500000000001e-05,
      "loss": 0.5964,
      "step": 19035
    },
    {
      "epoch": 1.065964833687983,
      "grad_norm": 1.4354372024536133,
      "learning_rate": 9.513e-05,
      "loss": 0.5044,
      "step": 19036
    },
    {
      "epoch": 1.066020831000112,
      "grad_norm": 1.3081278800964355,
      "learning_rate": 9.5135e-05,
      "loss": 0.3527,
      "step": 19037
    },
    {
      "epoch": 1.066076828312241,
      "grad_norm": 1.4948159456253052,
      "learning_rate": 9.514e-05,
      "loss": 0.4191,
      "step": 19038
    },
    {
      "epoch": 1.06613282562437,
      "grad_norm": 1.452414631843567,
      "learning_rate": 9.514500000000001e-05,
      "loss": 0.5055,
      "step": 19039
    },
    {
      "epoch": 1.066188822936499,
      "grad_norm": 1.4004466533660889,
      "learning_rate": 9.515000000000001e-05,
      "loss": 0.3828,
      "step": 19040
    },
    {
      "epoch": 1.066244820248628,
      "grad_norm": 1.7150640487670898,
      "learning_rate": 9.5155e-05,
      "loss": 0.3618,
      "step": 19041
    },
    {
      "epoch": 1.0663008175607571,
      "grad_norm": 1.6523611545562744,
      "learning_rate": 9.516e-05,
      "loss": 0.4781,
      "step": 19042
    },
    {
      "epoch": 1.0663568148728861,
      "grad_norm": 1.2870184183120728,
      "learning_rate": 9.5165e-05,
      "loss": 0.4917,
      "step": 19043
    },
    {
      "epoch": 1.0664128121850152,
      "grad_norm": 1.2794761657714844,
      "learning_rate": 9.517e-05,
      "loss": 0.5292,
      "step": 19044
    },
    {
      "epoch": 1.0664688094971442,
      "grad_norm": 1.593009352684021,
      "learning_rate": 9.517500000000001e-05,
      "loss": 0.4965,
      "step": 19045
    },
    {
      "epoch": 1.0665248068092732,
      "grad_norm": 1.3673381805419922,
      "learning_rate": 9.518000000000001e-05,
      "loss": 0.4164,
      "step": 19046
    },
    {
      "epoch": 1.0665808041214022,
      "grad_norm": 2.2797043323516846,
      "learning_rate": 9.5185e-05,
      "loss": 0.49,
      "step": 19047
    },
    {
      "epoch": 1.0666368014335312,
      "grad_norm": 1.3565945625305176,
      "learning_rate": 9.519e-05,
      "loss": 0.464,
      "step": 19048
    },
    {
      "epoch": 1.0666927987456603,
      "grad_norm": 1.221509575843811,
      "learning_rate": 9.5195e-05,
      "loss": 0.3296,
      "step": 19049
    },
    {
      "epoch": 1.0667487960577893,
      "grad_norm": 1.562795639038086,
      "learning_rate": 9.52e-05,
      "loss": 0.3887,
      "step": 19050
    },
    {
      "epoch": 1.0668047933699183,
      "grad_norm": 1.1983307600021362,
      "learning_rate": 9.5205e-05,
      "loss": 0.4013,
      "step": 19051
    },
    {
      "epoch": 1.0668607906820473,
      "grad_norm": 1.505826473236084,
      "learning_rate": 9.521e-05,
      "loss": 0.4318,
      "step": 19052
    },
    {
      "epoch": 1.0669167879941763,
      "grad_norm": 1.3908984661102295,
      "learning_rate": 9.521500000000002e-05,
      "loss": 0.4813,
      "step": 19053
    },
    {
      "epoch": 1.0669727853063053,
      "grad_norm": 1.2964426279067993,
      "learning_rate": 9.522000000000001e-05,
      "loss": 0.3802,
      "step": 19054
    },
    {
      "epoch": 1.0670287826184344,
      "grad_norm": 1.3526910543441772,
      "learning_rate": 9.522500000000001e-05,
      "loss": 0.4145,
      "step": 19055
    },
    {
      "epoch": 1.0670847799305634,
      "grad_norm": 1.2798247337341309,
      "learning_rate": 9.523000000000001e-05,
      "loss": 0.4685,
      "step": 19056
    },
    {
      "epoch": 1.0671407772426924,
      "grad_norm": 1.429029941558838,
      "learning_rate": 9.5235e-05,
      "loss": 0.382,
      "step": 19057
    },
    {
      "epoch": 1.0671967745548214,
      "grad_norm": 1.4787951707839966,
      "learning_rate": 9.524e-05,
      "loss": 0.4634,
      "step": 19058
    },
    {
      "epoch": 1.0672527718669504,
      "grad_norm": 1.306098461151123,
      "learning_rate": 9.5245e-05,
      "loss": 0.5,
      "step": 19059
    },
    {
      "epoch": 1.0673087691790795,
      "grad_norm": 1.5572845935821533,
      "learning_rate": 9.525000000000001e-05,
      "loss": 0.4532,
      "step": 19060
    },
    {
      "epoch": 1.0673647664912085,
      "grad_norm": 1.5002474784851074,
      "learning_rate": 9.525500000000001e-05,
      "loss": 0.4291,
      "step": 19061
    },
    {
      "epoch": 1.0674207638033375,
      "grad_norm": 1.4859400987625122,
      "learning_rate": 9.526e-05,
      "loss": 0.4613,
      "step": 19062
    },
    {
      "epoch": 1.0674767611154665,
      "grad_norm": 1.4809750318527222,
      "learning_rate": 9.5265e-05,
      "loss": 0.439,
      "step": 19063
    },
    {
      "epoch": 1.0675327584275955,
      "grad_norm": 1.3810079097747803,
      "learning_rate": 9.527e-05,
      "loss": 0.3838,
      "step": 19064
    },
    {
      "epoch": 1.0675887557397246,
      "grad_norm": 1.2954858541488647,
      "learning_rate": 9.5275e-05,
      "loss": 0.3757,
      "step": 19065
    },
    {
      "epoch": 1.0676447530518536,
      "grad_norm": 1.359278917312622,
      "learning_rate": 9.528000000000001e-05,
      "loss": 0.4049,
      "step": 19066
    },
    {
      "epoch": 1.0677007503639826,
      "grad_norm": 1.168003797531128,
      "learning_rate": 9.5285e-05,
      "loss": 0.392,
      "step": 19067
    },
    {
      "epoch": 1.0677567476761116,
      "grad_norm": 1.284511923789978,
      "learning_rate": 9.529e-05,
      "loss": 0.3769,
      "step": 19068
    },
    {
      "epoch": 1.0678127449882406,
      "grad_norm": 1.1729944944381714,
      "learning_rate": 9.5295e-05,
      "loss": 0.3494,
      "step": 19069
    },
    {
      "epoch": 1.0678687423003697,
      "grad_norm": 1.1762340068817139,
      "learning_rate": 9.53e-05,
      "loss": 0.3331,
      "step": 19070
    },
    {
      "epoch": 1.0679247396124987,
      "grad_norm": 1.6314793825149536,
      "learning_rate": 9.5305e-05,
      "loss": 0.5729,
      "step": 19071
    },
    {
      "epoch": 1.0679807369246277,
      "grad_norm": 1.3847637176513672,
      "learning_rate": 9.531e-05,
      "loss": 0.4225,
      "step": 19072
    },
    {
      "epoch": 1.0680367342367567,
      "grad_norm": 1.3760457038879395,
      "learning_rate": 9.5315e-05,
      "loss": 0.4248,
      "step": 19073
    },
    {
      "epoch": 1.0680927315488857,
      "grad_norm": 1.4888118505477905,
      "learning_rate": 9.532000000000002e-05,
      "loss": 0.5156,
      "step": 19074
    },
    {
      "epoch": 1.0681487288610148,
      "grad_norm": 1.3431413173675537,
      "learning_rate": 9.532500000000001e-05,
      "loss": 0.3934,
      "step": 19075
    },
    {
      "epoch": 1.0682047261731438,
      "grad_norm": 1.3805855512619019,
      "learning_rate": 9.533000000000001e-05,
      "loss": 0.5146,
      "step": 19076
    },
    {
      "epoch": 1.0682607234852728,
      "grad_norm": 1.291238784790039,
      "learning_rate": 9.533500000000001e-05,
      "loss": 0.3844,
      "step": 19077
    },
    {
      "epoch": 1.0683167207974018,
      "grad_norm": 1.8542561531066895,
      "learning_rate": 9.534e-05,
      "loss": 0.4608,
      "step": 19078
    },
    {
      "epoch": 1.0683727181095308,
      "grad_norm": 1.3636932373046875,
      "learning_rate": 9.5345e-05,
      "loss": 0.4417,
      "step": 19079
    },
    {
      "epoch": 1.0684287154216598,
      "grad_norm": 1.1776975393295288,
      "learning_rate": 9.535e-05,
      "loss": 0.3979,
      "step": 19080
    },
    {
      "epoch": 1.0684847127337889,
      "grad_norm": 1.3916935920715332,
      "learning_rate": 9.535500000000001e-05,
      "loss": 0.4552,
      "step": 19081
    },
    {
      "epoch": 1.0685407100459179,
      "grad_norm": 1.281245470046997,
      "learning_rate": 9.536000000000001e-05,
      "loss": 0.4894,
      "step": 19082
    },
    {
      "epoch": 1.068596707358047,
      "grad_norm": 1.235292911529541,
      "learning_rate": 9.5365e-05,
      "loss": 0.5557,
      "step": 19083
    },
    {
      "epoch": 1.068652704670176,
      "grad_norm": 1.126031756401062,
      "learning_rate": 9.537e-05,
      "loss": 0.3541,
      "step": 19084
    },
    {
      "epoch": 1.068708701982305,
      "grad_norm": 1.4543650150299072,
      "learning_rate": 9.5375e-05,
      "loss": 0.3107,
      "step": 19085
    },
    {
      "epoch": 1.068764699294434,
      "grad_norm": 1.264474868774414,
      "learning_rate": 9.538e-05,
      "loss": 0.5231,
      "step": 19086
    },
    {
      "epoch": 1.068820696606563,
      "grad_norm": 1.4919121265411377,
      "learning_rate": 9.5385e-05,
      "loss": 0.4949,
      "step": 19087
    },
    {
      "epoch": 1.068876693918692,
      "grad_norm": 1.2485785484313965,
      "learning_rate": 9.539e-05,
      "loss": 0.3303,
      "step": 19088
    },
    {
      "epoch": 1.068932691230821,
      "grad_norm": 1.2648602724075317,
      "learning_rate": 9.5395e-05,
      "loss": 0.3607,
      "step": 19089
    },
    {
      "epoch": 1.06898868854295,
      "grad_norm": 1.0592108964920044,
      "learning_rate": 9.54e-05,
      "loss": 0.2496,
      "step": 19090
    },
    {
      "epoch": 1.069044685855079,
      "grad_norm": 1.2441421747207642,
      "learning_rate": 9.5405e-05,
      "loss": 0.4313,
      "step": 19091
    },
    {
      "epoch": 1.069100683167208,
      "grad_norm": 1.3437831401824951,
      "learning_rate": 9.541e-05,
      "loss": 0.4893,
      "step": 19092
    },
    {
      "epoch": 1.0691566804793369,
      "grad_norm": 1.3966460227966309,
      "learning_rate": 9.541500000000001e-05,
      "loss": 0.4479,
      "step": 19093
    },
    {
      "epoch": 1.069212677791466,
      "grad_norm": 1.3548359870910645,
      "learning_rate": 9.542e-05,
      "loss": 0.4455,
      "step": 19094
    },
    {
      "epoch": 1.069268675103595,
      "grad_norm": 2.854562520980835,
      "learning_rate": 9.542500000000002e-05,
      "loss": 0.3788,
      "step": 19095
    },
    {
      "epoch": 1.069324672415724,
      "grad_norm": 1.4204295873641968,
      "learning_rate": 9.543000000000001e-05,
      "loss": 0.5879,
      "step": 19096
    },
    {
      "epoch": 1.069380669727853,
      "grad_norm": 1.1766201257705688,
      "learning_rate": 9.543500000000001e-05,
      "loss": 0.3723,
      "step": 19097
    },
    {
      "epoch": 1.069436667039982,
      "grad_norm": 1.2578608989715576,
      "learning_rate": 9.544000000000001e-05,
      "loss": 0.4654,
      "step": 19098
    },
    {
      "epoch": 1.069492664352111,
      "grad_norm": 1.2997828722000122,
      "learning_rate": 9.5445e-05,
      "loss": 0.478,
      "step": 19099
    },
    {
      "epoch": 1.06954866166424,
      "grad_norm": 1.2996234893798828,
      "learning_rate": 9.545e-05,
      "loss": 0.5364,
      "step": 19100
    },
    {
      "epoch": 1.069604658976369,
      "grad_norm": 1.490668535232544,
      "learning_rate": 9.5455e-05,
      "loss": 0.5021,
      "step": 19101
    },
    {
      "epoch": 1.069660656288498,
      "grad_norm": 1.256386160850525,
      "learning_rate": 9.546000000000001e-05,
      "loss": 0.5062,
      "step": 19102
    },
    {
      "epoch": 1.069716653600627,
      "grad_norm": 1.200835943222046,
      "learning_rate": 9.546500000000001e-05,
      "loss": 0.3749,
      "step": 19103
    },
    {
      "epoch": 1.069772650912756,
      "grad_norm": 1.45115065574646,
      "learning_rate": 9.547e-05,
      "loss": 0.5045,
      "step": 19104
    },
    {
      "epoch": 1.069828648224885,
      "grad_norm": 1.0986417531967163,
      "learning_rate": 9.5475e-05,
      "loss": 0.4245,
      "step": 19105
    },
    {
      "epoch": 1.0698846455370141,
      "grad_norm": 1.5335986614227295,
      "learning_rate": 9.548e-05,
      "loss": 0.3657,
      "step": 19106
    },
    {
      "epoch": 1.0699406428491431,
      "grad_norm": 1.6153874397277832,
      "learning_rate": 9.5485e-05,
      "loss": 0.386,
      "step": 19107
    },
    {
      "epoch": 1.0699966401612722,
      "grad_norm": 1.272781252861023,
      "learning_rate": 9.549e-05,
      "loss": 0.4127,
      "step": 19108
    },
    {
      "epoch": 1.0700526374734012,
      "grad_norm": 1.1290336847305298,
      "learning_rate": 9.5495e-05,
      "loss": 0.33,
      "step": 19109
    },
    {
      "epoch": 1.0701086347855302,
      "grad_norm": 1.4628112316131592,
      "learning_rate": 9.55e-05,
      "loss": 0.3112,
      "step": 19110
    },
    {
      "epoch": 1.0701646320976592,
      "grad_norm": 2.051401138305664,
      "learning_rate": 9.5505e-05,
      "loss": 0.4451,
      "step": 19111
    },
    {
      "epoch": 1.0702206294097882,
      "grad_norm": 1.1548988819122314,
      "learning_rate": 9.551e-05,
      "loss": 0.3769,
      "step": 19112
    },
    {
      "epoch": 1.0702766267219173,
      "grad_norm": 1.3833144903182983,
      "learning_rate": 9.551500000000001e-05,
      "loss": 0.3899,
      "step": 19113
    },
    {
      "epoch": 1.0703326240340463,
      "grad_norm": 1.3498610258102417,
      "learning_rate": 9.552000000000001e-05,
      "loss": 0.4589,
      "step": 19114
    },
    {
      "epoch": 1.0703886213461753,
      "grad_norm": 1.2717036008834839,
      "learning_rate": 9.5525e-05,
      "loss": 0.3871,
      "step": 19115
    },
    {
      "epoch": 1.0704446186583043,
      "grad_norm": 1.3312551975250244,
      "learning_rate": 9.553e-05,
      "loss": 0.4529,
      "step": 19116
    },
    {
      "epoch": 1.0705006159704333,
      "grad_norm": 1.3813267946243286,
      "learning_rate": 9.553500000000001e-05,
      "loss": 0.4138,
      "step": 19117
    },
    {
      "epoch": 1.0705566132825624,
      "grad_norm": 1.2773473262786865,
      "learning_rate": 9.554000000000001e-05,
      "loss": 0.4857,
      "step": 19118
    },
    {
      "epoch": 1.0706126105946914,
      "grad_norm": 1.2579201459884644,
      "learning_rate": 9.554500000000001e-05,
      "loss": 0.4754,
      "step": 19119
    },
    {
      "epoch": 1.0706686079068204,
      "grad_norm": 1.289850115776062,
      "learning_rate": 9.555e-05,
      "loss": 0.4207,
      "step": 19120
    },
    {
      "epoch": 1.0707246052189494,
      "grad_norm": 1.1401543617248535,
      "learning_rate": 9.5555e-05,
      "loss": 0.4304,
      "step": 19121
    },
    {
      "epoch": 1.0707806025310784,
      "grad_norm": 1.5424062013626099,
      "learning_rate": 9.556e-05,
      "loss": 0.4428,
      "step": 19122
    },
    {
      "epoch": 1.0708365998432074,
      "grad_norm": 1.9160706996917725,
      "learning_rate": 9.556500000000001e-05,
      "loss": 0.6462,
      "step": 19123
    },
    {
      "epoch": 1.0708925971553365,
      "grad_norm": 1.756384253501892,
      "learning_rate": 9.557000000000001e-05,
      "loss": 0.4831,
      "step": 19124
    },
    {
      "epoch": 1.0709485944674655,
      "grad_norm": 1.5352463722229004,
      "learning_rate": 9.5575e-05,
      "loss": 0.4154,
      "step": 19125
    },
    {
      "epoch": 1.0710045917795945,
      "grad_norm": 1.402227520942688,
      "learning_rate": 9.558e-05,
      "loss": 0.4776,
      "step": 19126
    },
    {
      "epoch": 1.0710605890917235,
      "grad_norm": 1.381800889968872,
      "learning_rate": 9.5585e-05,
      "loss": 0.3569,
      "step": 19127
    },
    {
      "epoch": 1.0711165864038525,
      "grad_norm": 1.2688425779342651,
      "learning_rate": 9.559e-05,
      "loss": 0.3758,
      "step": 19128
    },
    {
      "epoch": 1.0711725837159816,
      "grad_norm": 1.3217486143112183,
      "learning_rate": 9.5595e-05,
      "loss": 0.5328,
      "step": 19129
    },
    {
      "epoch": 1.0712285810281106,
      "grad_norm": 1.3088239431381226,
      "learning_rate": 9.56e-05,
      "loss": 0.4262,
      "step": 19130
    },
    {
      "epoch": 1.0712845783402396,
      "grad_norm": 1.3540040254592896,
      "learning_rate": 9.5605e-05,
      "loss": 0.4453,
      "step": 19131
    },
    {
      "epoch": 1.0713405756523686,
      "grad_norm": 1.3389081954956055,
      "learning_rate": 9.561e-05,
      "loss": 0.3616,
      "step": 19132
    },
    {
      "epoch": 1.0713965729644976,
      "grad_norm": 1.123694658279419,
      "learning_rate": 9.561500000000001e-05,
      "loss": 0.4807,
      "step": 19133
    },
    {
      "epoch": 1.0714525702766267,
      "grad_norm": 1.4012280702590942,
      "learning_rate": 9.562000000000001e-05,
      "loss": 0.5071,
      "step": 19134
    },
    {
      "epoch": 1.0715085675887557,
      "grad_norm": 1.4097554683685303,
      "learning_rate": 9.562500000000001e-05,
      "loss": 0.4201,
      "step": 19135
    },
    {
      "epoch": 1.0715645649008847,
      "grad_norm": 1.218183994293213,
      "learning_rate": 9.563e-05,
      "loss": 0.3332,
      "step": 19136
    },
    {
      "epoch": 1.0716205622130137,
      "grad_norm": 1.2299362421035767,
      "learning_rate": 9.5635e-05,
      "loss": 0.3997,
      "step": 19137
    },
    {
      "epoch": 1.0716765595251427,
      "grad_norm": 1.709261178970337,
      "learning_rate": 9.564000000000001e-05,
      "loss": 0.459,
      "step": 19138
    },
    {
      "epoch": 1.0717325568372718,
      "grad_norm": 1.3686734437942505,
      "learning_rate": 9.564500000000001e-05,
      "loss": 0.4295,
      "step": 19139
    },
    {
      "epoch": 1.0717885541494008,
      "grad_norm": 1.5993621349334717,
      "learning_rate": 9.565000000000001e-05,
      "loss": 0.4597,
      "step": 19140
    },
    {
      "epoch": 1.0718445514615298,
      "grad_norm": 1.6603903770446777,
      "learning_rate": 9.5655e-05,
      "loss": 0.5369,
      "step": 19141
    },
    {
      "epoch": 1.0719005487736588,
      "grad_norm": 1.1640290021896362,
      "learning_rate": 9.566e-05,
      "loss": 0.4805,
      "step": 19142
    },
    {
      "epoch": 1.0719565460857878,
      "grad_norm": 1.2247138023376465,
      "learning_rate": 9.5665e-05,
      "loss": 0.3195,
      "step": 19143
    },
    {
      "epoch": 1.0720125433979169,
      "grad_norm": 1.2630603313446045,
      "learning_rate": 9.567000000000001e-05,
      "loss": 0.3868,
      "step": 19144
    },
    {
      "epoch": 1.0720685407100459,
      "grad_norm": 1.3521305322647095,
      "learning_rate": 9.567500000000001e-05,
      "loss": 0.4709,
      "step": 19145
    },
    {
      "epoch": 1.0721245380221749,
      "grad_norm": 1.1388635635375977,
      "learning_rate": 9.568e-05,
      "loss": 0.2951,
      "step": 19146
    },
    {
      "epoch": 1.072180535334304,
      "grad_norm": 1.536655306816101,
      "learning_rate": 9.5685e-05,
      "loss": 0.3331,
      "step": 19147
    },
    {
      "epoch": 1.072236532646433,
      "grad_norm": 1.6921892166137695,
      "learning_rate": 9.569e-05,
      "loss": 0.4127,
      "step": 19148
    },
    {
      "epoch": 1.072292529958562,
      "grad_norm": 1.5893220901489258,
      "learning_rate": 9.5695e-05,
      "loss": 0.396,
      "step": 19149
    },
    {
      "epoch": 1.072348527270691,
      "grad_norm": 2.503692865371704,
      "learning_rate": 9.57e-05,
      "loss": 0.4211,
      "step": 19150
    },
    {
      "epoch": 1.07240452458282,
      "grad_norm": 1.5990415811538696,
      "learning_rate": 9.5705e-05,
      "loss": 0.491,
      "step": 19151
    },
    {
      "epoch": 1.072460521894949,
      "grad_norm": 1.5531582832336426,
      "learning_rate": 9.571e-05,
      "loss": 0.5286,
      "step": 19152
    },
    {
      "epoch": 1.072516519207078,
      "grad_norm": 1.3615837097167969,
      "learning_rate": 9.5715e-05,
      "loss": 0.436,
      "step": 19153
    },
    {
      "epoch": 1.072572516519207,
      "grad_norm": 1.5618793964385986,
      "learning_rate": 9.572000000000001e-05,
      "loss": 0.4309,
      "step": 19154
    },
    {
      "epoch": 1.072628513831336,
      "grad_norm": 1.3727819919586182,
      "learning_rate": 9.572500000000001e-05,
      "loss": 0.5035,
      "step": 19155
    },
    {
      "epoch": 1.072684511143465,
      "grad_norm": 1.286844253540039,
      "learning_rate": 9.573e-05,
      "loss": 0.3753,
      "step": 19156
    },
    {
      "epoch": 1.072740508455594,
      "grad_norm": 1.4722813367843628,
      "learning_rate": 9.5735e-05,
      "loss": 0.3495,
      "step": 19157
    },
    {
      "epoch": 1.0727965057677231,
      "grad_norm": 1.631972074508667,
      "learning_rate": 9.574e-05,
      "loss": 0.4748,
      "step": 19158
    },
    {
      "epoch": 1.0728525030798521,
      "grad_norm": 1.638663411140442,
      "learning_rate": 9.574500000000001e-05,
      "loss": 0.4136,
      "step": 19159
    },
    {
      "epoch": 1.0729085003919812,
      "grad_norm": 1.4947776794433594,
      "learning_rate": 9.575000000000001e-05,
      "loss": 0.5904,
      "step": 19160
    },
    {
      "epoch": 1.0729644977041102,
      "grad_norm": 2.0101702213287354,
      "learning_rate": 9.575500000000001e-05,
      "loss": 0.5917,
      "step": 19161
    },
    {
      "epoch": 1.0730204950162392,
      "grad_norm": 1.1375631093978882,
      "learning_rate": 9.576e-05,
      "loss": 0.3687,
      "step": 19162
    },
    {
      "epoch": 1.0730764923283682,
      "grad_norm": 1.2634828090667725,
      "learning_rate": 9.5765e-05,
      "loss": 0.3762,
      "step": 19163
    },
    {
      "epoch": 1.0731324896404972,
      "grad_norm": 1.2566579580307007,
      "learning_rate": 9.577e-05,
      "loss": 0.4033,
      "step": 19164
    },
    {
      "epoch": 1.0731884869526263,
      "grad_norm": 1.6312873363494873,
      "learning_rate": 9.5775e-05,
      "loss": 0.5984,
      "step": 19165
    },
    {
      "epoch": 1.0732444842647553,
      "grad_norm": 1.2147444486618042,
      "learning_rate": 9.578000000000001e-05,
      "loss": 0.4717,
      "step": 19166
    },
    {
      "epoch": 1.0733004815768843,
      "grad_norm": 1.5377756357192993,
      "learning_rate": 9.5785e-05,
      "loss": 0.4054,
      "step": 19167
    },
    {
      "epoch": 1.0733564788890133,
      "grad_norm": 1.4068466424942017,
      "learning_rate": 9.579e-05,
      "loss": 0.3693,
      "step": 19168
    },
    {
      "epoch": 1.0734124762011423,
      "grad_norm": 1.3765770196914673,
      "learning_rate": 9.5795e-05,
      "loss": 0.5127,
      "step": 19169
    },
    {
      "epoch": 1.0734684735132713,
      "grad_norm": 1.680324673652649,
      "learning_rate": 9.58e-05,
      "loss": 0.5666,
      "step": 19170
    },
    {
      "epoch": 1.0735244708254004,
      "grad_norm": 1.2332838773727417,
      "learning_rate": 9.5805e-05,
      "loss": 0.4255,
      "step": 19171
    },
    {
      "epoch": 1.0735804681375294,
      "grad_norm": 1.4358876943588257,
      "learning_rate": 9.581e-05,
      "loss": 0.3233,
      "step": 19172
    },
    {
      "epoch": 1.0736364654496584,
      "grad_norm": 1.2500696182250977,
      "learning_rate": 9.5815e-05,
      "loss": 0.4533,
      "step": 19173
    },
    {
      "epoch": 1.0736924627617874,
      "grad_norm": 1.4486896991729736,
      "learning_rate": 9.582000000000001e-05,
      "loss": 0.4665,
      "step": 19174
    },
    {
      "epoch": 1.0737484600739164,
      "grad_norm": 1.3838056325912476,
      "learning_rate": 9.582500000000001e-05,
      "loss": 0.4326,
      "step": 19175
    },
    {
      "epoch": 1.0738044573860455,
      "grad_norm": 1.5006517171859741,
      "learning_rate": 9.583000000000001e-05,
      "loss": 0.5085,
      "step": 19176
    },
    {
      "epoch": 1.0738604546981745,
      "grad_norm": 1.25216543674469,
      "learning_rate": 9.5835e-05,
      "loss": 0.4017,
      "step": 19177
    },
    {
      "epoch": 1.0739164520103035,
      "grad_norm": 1.4105465412139893,
      "learning_rate": 9.584e-05,
      "loss": 0.482,
      "step": 19178
    },
    {
      "epoch": 1.0739724493224325,
      "grad_norm": 1.3745070695877075,
      "learning_rate": 9.5845e-05,
      "loss": 0.4596,
      "step": 19179
    },
    {
      "epoch": 1.0740284466345615,
      "grad_norm": 1.3403490781784058,
      "learning_rate": 9.585000000000001e-05,
      "loss": 0.4411,
      "step": 19180
    },
    {
      "epoch": 1.0740844439466906,
      "grad_norm": 1.0515056848526,
      "learning_rate": 9.585500000000001e-05,
      "loss": 0.3387,
      "step": 19181
    },
    {
      "epoch": 1.0741404412588196,
      "grad_norm": 1.3254003524780273,
      "learning_rate": 9.586000000000001e-05,
      "loss": 0.3439,
      "step": 19182
    },
    {
      "epoch": 1.0741964385709486,
      "grad_norm": 1.3216105699539185,
      "learning_rate": 9.5865e-05,
      "loss": 0.3528,
      "step": 19183
    },
    {
      "epoch": 1.0742524358830776,
      "grad_norm": 1.2448352575302124,
      "learning_rate": 9.587e-05,
      "loss": 0.3025,
      "step": 19184
    },
    {
      "epoch": 1.0743084331952066,
      "grad_norm": 1.499056339263916,
      "learning_rate": 9.5875e-05,
      "loss": 0.4477,
      "step": 19185
    },
    {
      "epoch": 1.0743644305073357,
      "grad_norm": 1.3132444620132446,
      "learning_rate": 9.588e-05,
      "loss": 0.5061,
      "step": 19186
    },
    {
      "epoch": 1.0744204278194647,
      "grad_norm": 1.2185949087142944,
      "learning_rate": 9.588500000000001e-05,
      "loss": 0.2891,
      "step": 19187
    },
    {
      "epoch": 1.0744764251315937,
      "grad_norm": 1.2195074558258057,
      "learning_rate": 9.589e-05,
      "loss": 0.326,
      "step": 19188
    },
    {
      "epoch": 1.0745324224437227,
      "grad_norm": 1.6875252723693848,
      "learning_rate": 9.5895e-05,
      "loss": 0.5185,
      "step": 19189
    },
    {
      "epoch": 1.0745884197558517,
      "grad_norm": 1.3300776481628418,
      "learning_rate": 9.59e-05,
      "loss": 0.3762,
      "step": 19190
    },
    {
      "epoch": 1.0746444170679808,
      "grad_norm": 1.4575303792953491,
      "learning_rate": 9.5905e-05,
      "loss": 0.4495,
      "step": 19191
    },
    {
      "epoch": 1.0747004143801098,
      "grad_norm": 1.3437873125076294,
      "learning_rate": 9.591e-05,
      "loss": 0.5067,
      "step": 19192
    },
    {
      "epoch": 1.0747564116922388,
      "grad_norm": 1.2452030181884766,
      "learning_rate": 9.5915e-05,
      "loss": 0.3846,
      "step": 19193
    },
    {
      "epoch": 1.0748124090043678,
      "grad_norm": 1.2425785064697266,
      "learning_rate": 9.592e-05,
      "loss": 0.3747,
      "step": 19194
    },
    {
      "epoch": 1.0748684063164968,
      "grad_norm": 1.5638777017593384,
      "learning_rate": 9.592500000000001e-05,
      "loss": 0.4093,
      "step": 19195
    },
    {
      "epoch": 1.0749244036286258,
      "grad_norm": 1.5321054458618164,
      "learning_rate": 9.593000000000001e-05,
      "loss": 0.4748,
      "step": 19196
    },
    {
      "epoch": 1.0749804009407549,
      "grad_norm": 1.5617907047271729,
      "learning_rate": 9.593500000000001e-05,
      "loss": 0.4677,
      "step": 19197
    },
    {
      "epoch": 1.0750363982528839,
      "grad_norm": 2.0262441635131836,
      "learning_rate": 9.594e-05,
      "loss": 0.6669,
      "step": 19198
    },
    {
      "epoch": 1.075092395565013,
      "grad_norm": 1.2591593265533447,
      "learning_rate": 9.5945e-05,
      "loss": 0.5268,
      "step": 19199
    },
    {
      "epoch": 1.075148392877142,
      "grad_norm": 1.3188189268112183,
      "learning_rate": 9.595e-05,
      "loss": 0.4543,
      "step": 19200
    },
    {
      "epoch": 1.075204390189271,
      "grad_norm": 1.450954794883728,
      "learning_rate": 9.595500000000001e-05,
      "loss": 0.4581,
      "step": 19201
    },
    {
      "epoch": 1.0752603875014,
      "grad_norm": 1.6723936796188354,
      "learning_rate": 9.596000000000001e-05,
      "loss": 0.5799,
      "step": 19202
    },
    {
      "epoch": 1.075316384813529,
      "grad_norm": 1.4126452207565308,
      "learning_rate": 9.596500000000001e-05,
      "loss": 0.4893,
      "step": 19203
    },
    {
      "epoch": 1.075372382125658,
      "grad_norm": 1.315180778503418,
      "learning_rate": 9.597e-05,
      "loss": 0.4201,
      "step": 19204
    },
    {
      "epoch": 1.075428379437787,
      "grad_norm": 1.350109577178955,
      "learning_rate": 9.5975e-05,
      "loss": 0.4438,
      "step": 19205
    },
    {
      "epoch": 1.075484376749916,
      "grad_norm": 1.246303677558899,
      "learning_rate": 9.598e-05,
      "loss": 0.3991,
      "step": 19206
    },
    {
      "epoch": 1.075540374062045,
      "grad_norm": 1.2813001871109009,
      "learning_rate": 9.5985e-05,
      "loss": 0.42,
      "step": 19207
    },
    {
      "epoch": 1.075596371374174,
      "grad_norm": 1.320867657661438,
      "learning_rate": 9.599000000000001e-05,
      "loss": 0.4516,
      "step": 19208
    },
    {
      "epoch": 1.075652368686303,
      "grad_norm": 1.4397691488265991,
      "learning_rate": 9.5995e-05,
      "loss": 0.4549,
      "step": 19209
    },
    {
      "epoch": 1.0757083659984321,
      "grad_norm": 1.1421473026275635,
      "learning_rate": 9.6e-05,
      "loss": 0.3526,
      "step": 19210
    },
    {
      "epoch": 1.0757643633105611,
      "grad_norm": 1.3536735773086548,
      "learning_rate": 9.6005e-05,
      "loss": 0.3772,
      "step": 19211
    },
    {
      "epoch": 1.0758203606226902,
      "grad_norm": 1.417104959487915,
      "learning_rate": 9.601e-05,
      "loss": 0.4627,
      "step": 19212
    },
    {
      "epoch": 1.0758763579348192,
      "grad_norm": 1.2568718194961548,
      "learning_rate": 9.6015e-05,
      "loss": 0.3361,
      "step": 19213
    },
    {
      "epoch": 1.0759323552469482,
      "grad_norm": 1.1949340105056763,
      "learning_rate": 9.602e-05,
      "loss": 0.4164,
      "step": 19214
    },
    {
      "epoch": 1.0759883525590772,
      "grad_norm": 1.399598479270935,
      "learning_rate": 9.6025e-05,
      "loss": 0.4094,
      "step": 19215
    },
    {
      "epoch": 1.0760443498712062,
      "grad_norm": 1.5406945943832397,
      "learning_rate": 9.603000000000001e-05,
      "loss": 0.4701,
      "step": 19216
    },
    {
      "epoch": 1.0761003471833352,
      "grad_norm": 1.3081169128417969,
      "learning_rate": 9.603500000000001e-05,
      "loss": 0.4717,
      "step": 19217
    },
    {
      "epoch": 1.0761563444954643,
      "grad_norm": 1.2680391073226929,
      "learning_rate": 9.604000000000001e-05,
      "loss": 0.4676,
      "step": 19218
    },
    {
      "epoch": 1.0762123418075933,
      "grad_norm": 1.303810954093933,
      "learning_rate": 9.6045e-05,
      "loss": 0.41,
      "step": 19219
    },
    {
      "epoch": 1.0762683391197223,
      "grad_norm": 1.2404943704605103,
      "learning_rate": 9.605e-05,
      "loss": 0.445,
      "step": 19220
    },
    {
      "epoch": 1.0763243364318513,
      "grad_norm": 1.246006727218628,
      "learning_rate": 9.6055e-05,
      "loss": 0.458,
      "step": 19221
    },
    {
      "epoch": 1.0763803337439803,
      "grad_norm": 1.2949378490447998,
      "learning_rate": 9.606000000000001e-05,
      "loss": 0.4365,
      "step": 19222
    },
    {
      "epoch": 1.0764363310561094,
      "grad_norm": 1.7878241539001465,
      "learning_rate": 9.606500000000001e-05,
      "loss": 0.4829,
      "step": 19223
    },
    {
      "epoch": 1.0764923283682384,
      "grad_norm": 1.2644611597061157,
      "learning_rate": 9.607000000000001e-05,
      "loss": 0.455,
      "step": 19224
    },
    {
      "epoch": 1.0765483256803674,
      "grad_norm": 1.3939383029937744,
      "learning_rate": 9.6075e-05,
      "loss": 0.4113,
      "step": 19225
    },
    {
      "epoch": 1.0766043229924964,
      "grad_norm": 1.207818627357483,
      "learning_rate": 9.608e-05,
      "loss": 0.3622,
      "step": 19226
    },
    {
      "epoch": 1.0766603203046254,
      "grad_norm": 1.1098518371582031,
      "learning_rate": 9.6085e-05,
      "loss": 0.2977,
      "step": 19227
    },
    {
      "epoch": 1.0767163176167545,
      "grad_norm": 1.2676883935928345,
      "learning_rate": 9.609e-05,
      "loss": 0.3818,
      "step": 19228
    },
    {
      "epoch": 1.0767723149288835,
      "grad_norm": 1.369973063468933,
      "learning_rate": 9.609500000000001e-05,
      "loss": 0.4579,
      "step": 19229
    },
    {
      "epoch": 1.0768283122410125,
      "grad_norm": 1.4226983785629272,
      "learning_rate": 9.61e-05,
      "loss": 0.3708,
      "step": 19230
    },
    {
      "epoch": 1.0768843095531415,
      "grad_norm": 1.5574685335159302,
      "learning_rate": 9.6105e-05,
      "loss": 0.5837,
      "step": 19231
    },
    {
      "epoch": 1.0769403068652705,
      "grad_norm": 1.4205504655838013,
      "learning_rate": 9.611e-05,
      "loss": 0.5139,
      "step": 19232
    },
    {
      "epoch": 1.0769963041773996,
      "grad_norm": 1.4705474376678467,
      "learning_rate": 9.6115e-05,
      "loss": 0.5394,
      "step": 19233
    },
    {
      "epoch": 1.0770523014895286,
      "grad_norm": 1.4634571075439453,
      "learning_rate": 9.612000000000001e-05,
      "loss": 0.448,
      "step": 19234
    },
    {
      "epoch": 1.0771082988016576,
      "grad_norm": 1.6852033138275146,
      "learning_rate": 9.6125e-05,
      "loss": 0.5067,
      "step": 19235
    },
    {
      "epoch": 1.0771642961137866,
      "grad_norm": 1.2660261392593384,
      "learning_rate": 9.613e-05,
      "loss": 0.3121,
      "step": 19236
    },
    {
      "epoch": 1.0772202934259156,
      "grad_norm": 1.2085484266281128,
      "learning_rate": 9.613500000000001e-05,
      "loss": 0.4393,
      "step": 19237
    },
    {
      "epoch": 1.0772762907380447,
      "grad_norm": 1.3156460523605347,
      "learning_rate": 9.614000000000001e-05,
      "loss": 0.5263,
      "step": 19238
    },
    {
      "epoch": 1.0773322880501737,
      "grad_norm": 1.5212336778640747,
      "learning_rate": 9.614500000000001e-05,
      "loss": 0.4745,
      "step": 19239
    },
    {
      "epoch": 1.0773882853623027,
      "grad_norm": 1.2039270401000977,
      "learning_rate": 9.615e-05,
      "loss": 0.3301,
      "step": 19240
    },
    {
      "epoch": 1.0774442826744317,
      "grad_norm": 1.3416954278945923,
      "learning_rate": 9.6155e-05,
      "loss": 0.2978,
      "step": 19241
    },
    {
      "epoch": 1.0775002799865607,
      "grad_norm": 1.633178472518921,
      "learning_rate": 9.616e-05,
      "loss": 0.3377,
      "step": 19242
    },
    {
      "epoch": 1.0775562772986897,
      "grad_norm": 1.3483723402023315,
      "learning_rate": 9.616500000000001e-05,
      "loss": 0.4082,
      "step": 19243
    },
    {
      "epoch": 1.0776122746108188,
      "grad_norm": 1.492039680480957,
      "learning_rate": 9.617000000000001e-05,
      "loss": 0.5695,
      "step": 19244
    },
    {
      "epoch": 1.0776682719229478,
      "grad_norm": 1.0585488080978394,
      "learning_rate": 9.6175e-05,
      "loss": 0.3845,
      "step": 19245
    },
    {
      "epoch": 1.0777242692350768,
      "grad_norm": 1.646497368812561,
      "learning_rate": 9.618e-05,
      "loss": 0.4469,
      "step": 19246
    },
    {
      "epoch": 1.0777802665472058,
      "grad_norm": 1.4915268421173096,
      "learning_rate": 9.6185e-05,
      "loss": 0.4839,
      "step": 19247
    },
    {
      "epoch": 1.0778362638593348,
      "grad_norm": 1.1967825889587402,
      "learning_rate": 9.619e-05,
      "loss": 0.3524,
      "step": 19248
    },
    {
      "epoch": 1.0778922611714639,
      "grad_norm": 1.4445247650146484,
      "learning_rate": 9.6195e-05,
      "loss": 0.3228,
      "step": 19249
    },
    {
      "epoch": 1.0779482584835929,
      "grad_norm": 1.1791483163833618,
      "learning_rate": 9.620000000000001e-05,
      "loss": 0.3623,
      "step": 19250
    },
    {
      "epoch": 1.078004255795722,
      "grad_norm": 1.2073333263397217,
      "learning_rate": 9.6205e-05,
      "loss": 0.3775,
      "step": 19251
    },
    {
      "epoch": 1.078060253107851,
      "grad_norm": 1.238425612449646,
      "learning_rate": 9.621e-05,
      "loss": 0.3242,
      "step": 19252
    },
    {
      "epoch": 1.07811625041998,
      "grad_norm": 1.1087088584899902,
      "learning_rate": 9.6215e-05,
      "loss": 0.3238,
      "step": 19253
    },
    {
      "epoch": 1.078172247732109,
      "grad_norm": 1.262492060661316,
      "learning_rate": 9.622000000000001e-05,
      "loss": 0.3786,
      "step": 19254
    },
    {
      "epoch": 1.078228245044238,
      "grad_norm": 1.1630018949508667,
      "learning_rate": 9.622500000000001e-05,
      "loss": 0.342,
      "step": 19255
    },
    {
      "epoch": 1.078284242356367,
      "grad_norm": 1.9120094776153564,
      "learning_rate": 9.623e-05,
      "loss": 0.6781,
      "step": 19256
    },
    {
      "epoch": 1.078340239668496,
      "grad_norm": 1.4458720684051514,
      "learning_rate": 9.6235e-05,
      "loss": 0.4242,
      "step": 19257
    },
    {
      "epoch": 1.078396236980625,
      "grad_norm": 1.3422973155975342,
      "learning_rate": 9.624000000000001e-05,
      "loss": 0.3918,
      "step": 19258
    },
    {
      "epoch": 1.078452234292754,
      "grad_norm": 1.4883275032043457,
      "learning_rate": 9.624500000000001e-05,
      "loss": 0.6011,
      "step": 19259
    },
    {
      "epoch": 1.078508231604883,
      "grad_norm": 1.652848720550537,
      "learning_rate": 9.625000000000001e-05,
      "loss": 0.4489,
      "step": 19260
    },
    {
      "epoch": 1.078564228917012,
      "grad_norm": 1.555734395980835,
      "learning_rate": 9.6255e-05,
      "loss": 0.3759,
      "step": 19261
    },
    {
      "epoch": 1.078620226229141,
      "grad_norm": 1.3970298767089844,
      "learning_rate": 9.626e-05,
      "loss": 0.4004,
      "step": 19262
    },
    {
      "epoch": 1.0786762235412701,
      "grad_norm": 1.3191635608673096,
      "learning_rate": 9.6265e-05,
      "loss": 0.3588,
      "step": 19263
    },
    {
      "epoch": 1.0787322208533991,
      "grad_norm": 1.6098352670669556,
      "learning_rate": 9.627e-05,
      "loss": 0.4608,
      "step": 19264
    },
    {
      "epoch": 1.0787882181655282,
      "grad_norm": 1.4320333003997803,
      "learning_rate": 9.627500000000001e-05,
      "loss": 0.4395,
      "step": 19265
    },
    {
      "epoch": 1.0788442154776572,
      "grad_norm": 1.2062251567840576,
      "learning_rate": 9.628e-05,
      "loss": 0.3891,
      "step": 19266
    },
    {
      "epoch": 1.0789002127897862,
      "grad_norm": 1.2579712867736816,
      "learning_rate": 9.6285e-05,
      "loss": 0.4037,
      "step": 19267
    },
    {
      "epoch": 1.0789562101019152,
      "grad_norm": 1.2883987426757812,
      "learning_rate": 9.629e-05,
      "loss": 0.4493,
      "step": 19268
    },
    {
      "epoch": 1.079012207414044,
      "grad_norm": 1.0617892742156982,
      "learning_rate": 9.6295e-05,
      "loss": 0.3791,
      "step": 19269
    },
    {
      "epoch": 1.079068204726173,
      "grad_norm": 1.2531203031539917,
      "learning_rate": 9.63e-05,
      "loss": 0.3705,
      "step": 19270
    },
    {
      "epoch": 1.079124202038302,
      "grad_norm": 1.6580188274383545,
      "learning_rate": 9.630500000000001e-05,
      "loss": 0.5208,
      "step": 19271
    },
    {
      "epoch": 1.079180199350431,
      "grad_norm": 1.5703518390655518,
      "learning_rate": 9.631e-05,
      "loss": 0.5463,
      "step": 19272
    },
    {
      "epoch": 1.07923619666256,
      "grad_norm": 1.3446650505065918,
      "learning_rate": 9.6315e-05,
      "loss": 0.4433,
      "step": 19273
    },
    {
      "epoch": 1.0792921939746891,
      "grad_norm": 1.2784892320632935,
      "learning_rate": 9.632e-05,
      "loss": 0.3716,
      "step": 19274
    },
    {
      "epoch": 1.0793481912868181,
      "grad_norm": 1.4385497570037842,
      "learning_rate": 9.632500000000001e-05,
      "loss": 0.4583,
      "step": 19275
    },
    {
      "epoch": 1.0794041885989472,
      "grad_norm": 1.7008953094482422,
      "learning_rate": 9.633000000000001e-05,
      "loss": 0.6018,
      "step": 19276
    },
    {
      "epoch": 1.0794601859110762,
      "grad_norm": 1.5577605962753296,
      "learning_rate": 9.6335e-05,
      "loss": 0.4435,
      "step": 19277
    },
    {
      "epoch": 1.0795161832232052,
      "grad_norm": 1.368937373161316,
      "learning_rate": 9.634e-05,
      "loss": 0.3812,
      "step": 19278
    },
    {
      "epoch": 1.0795721805353342,
      "grad_norm": 1.4653809070587158,
      "learning_rate": 9.634500000000001e-05,
      "loss": 0.4979,
      "step": 19279
    },
    {
      "epoch": 1.0796281778474632,
      "grad_norm": 1.2361409664154053,
      "learning_rate": 9.635000000000001e-05,
      "loss": 0.4288,
      "step": 19280
    },
    {
      "epoch": 1.0796841751595923,
      "grad_norm": 1.3378677368164062,
      "learning_rate": 9.635500000000001e-05,
      "loss": 0.4036,
      "step": 19281
    },
    {
      "epoch": 1.0797401724717213,
      "grad_norm": 1.4238026142120361,
      "learning_rate": 9.636e-05,
      "loss": 0.4819,
      "step": 19282
    },
    {
      "epoch": 1.0797961697838503,
      "grad_norm": 1.2204433679580688,
      "learning_rate": 9.6365e-05,
      "loss": 0.4173,
      "step": 19283
    },
    {
      "epoch": 1.0798521670959793,
      "grad_norm": 1.1371172666549683,
      "learning_rate": 9.637e-05,
      "loss": 0.3358,
      "step": 19284
    },
    {
      "epoch": 1.0799081644081083,
      "grad_norm": 1.2987414598464966,
      "learning_rate": 9.6375e-05,
      "loss": 0.4411,
      "step": 19285
    },
    {
      "epoch": 1.0799641617202373,
      "grad_norm": 1.4364862442016602,
      "learning_rate": 9.638000000000001e-05,
      "loss": 0.3644,
      "step": 19286
    },
    {
      "epoch": 1.0800201590323664,
      "grad_norm": 1.258409857749939,
      "learning_rate": 9.6385e-05,
      "loss": 0.4312,
      "step": 19287
    },
    {
      "epoch": 1.0800761563444954,
      "grad_norm": 1.504977822303772,
      "learning_rate": 9.639e-05,
      "loss": 0.426,
      "step": 19288
    },
    {
      "epoch": 1.0801321536566244,
      "grad_norm": 1.5396018028259277,
      "learning_rate": 9.6395e-05,
      "loss": 0.5399,
      "step": 19289
    },
    {
      "epoch": 1.0801881509687534,
      "grad_norm": 1.4223546981811523,
      "learning_rate": 9.64e-05,
      "loss": 0.5389,
      "step": 19290
    },
    {
      "epoch": 1.0802441482808824,
      "grad_norm": 1.4549106359481812,
      "learning_rate": 9.6405e-05,
      "loss": 0.5428,
      "step": 19291
    },
    {
      "epoch": 1.0803001455930115,
      "grad_norm": 1.9123473167419434,
      "learning_rate": 9.641000000000001e-05,
      "loss": 0.4298,
      "step": 19292
    },
    {
      "epoch": 1.0803561429051405,
      "grad_norm": 1.4224637746810913,
      "learning_rate": 9.6415e-05,
      "loss": 0.3685,
      "step": 19293
    },
    {
      "epoch": 1.0804121402172695,
      "grad_norm": 1.3219951391220093,
      "learning_rate": 9.642e-05,
      "loss": 0.4869,
      "step": 19294
    },
    {
      "epoch": 1.0804681375293985,
      "grad_norm": 1.3343758583068848,
      "learning_rate": 9.642500000000001e-05,
      "loss": 0.492,
      "step": 19295
    },
    {
      "epoch": 1.0805241348415275,
      "grad_norm": 1.5717768669128418,
      "learning_rate": 9.643000000000001e-05,
      "loss": 0.6136,
      "step": 19296
    },
    {
      "epoch": 1.0805801321536566,
      "grad_norm": 1.5233100652694702,
      "learning_rate": 9.643500000000001e-05,
      "loss": 0.4301,
      "step": 19297
    },
    {
      "epoch": 1.0806361294657856,
      "grad_norm": 1.520534634590149,
      "learning_rate": 9.644e-05,
      "loss": 0.482,
      "step": 19298
    },
    {
      "epoch": 1.0806921267779146,
      "grad_norm": 1.3004443645477295,
      "learning_rate": 9.6445e-05,
      "loss": 0.3732,
      "step": 19299
    },
    {
      "epoch": 1.0807481240900436,
      "grad_norm": 1.151798963546753,
      "learning_rate": 9.645000000000001e-05,
      "loss": 0.298,
      "step": 19300
    },
    {
      "epoch": 1.0808041214021726,
      "grad_norm": 1.3476203680038452,
      "learning_rate": 9.645500000000001e-05,
      "loss": 0.4183,
      "step": 19301
    },
    {
      "epoch": 1.0808601187143017,
      "grad_norm": 1.2625181674957275,
      "learning_rate": 9.646000000000001e-05,
      "loss": 0.4778,
      "step": 19302
    },
    {
      "epoch": 1.0809161160264307,
      "grad_norm": 1.2647287845611572,
      "learning_rate": 9.6465e-05,
      "loss": 0.4746,
      "step": 19303
    },
    {
      "epoch": 1.0809721133385597,
      "grad_norm": 1.4775749444961548,
      "learning_rate": 9.647e-05,
      "loss": 0.4654,
      "step": 19304
    },
    {
      "epoch": 1.0810281106506887,
      "grad_norm": 1.6221728324890137,
      "learning_rate": 9.6475e-05,
      "loss": 0.387,
      "step": 19305
    },
    {
      "epoch": 1.0810841079628177,
      "grad_norm": 1.1883046627044678,
      "learning_rate": 9.648e-05,
      "loss": 0.3965,
      "step": 19306
    },
    {
      "epoch": 1.0811401052749467,
      "grad_norm": 1.416805624961853,
      "learning_rate": 9.648500000000001e-05,
      "loss": 0.4267,
      "step": 19307
    },
    {
      "epoch": 1.0811961025870758,
      "grad_norm": 1.3891547918319702,
      "learning_rate": 9.649e-05,
      "loss": 0.4943,
      "step": 19308
    },
    {
      "epoch": 1.0812520998992048,
      "grad_norm": 1.1932092905044556,
      "learning_rate": 9.6495e-05,
      "loss": 0.4577,
      "step": 19309
    },
    {
      "epoch": 1.0813080972113338,
      "grad_norm": 1.4168490171432495,
      "learning_rate": 9.65e-05,
      "loss": 0.499,
      "step": 19310
    },
    {
      "epoch": 1.0813640945234628,
      "grad_norm": 1.3673006296157837,
      "learning_rate": 9.6505e-05,
      "loss": 0.4831,
      "step": 19311
    },
    {
      "epoch": 1.0814200918355918,
      "grad_norm": 1.3717122077941895,
      "learning_rate": 9.651e-05,
      "loss": 0.4299,
      "step": 19312
    },
    {
      "epoch": 1.0814760891477209,
      "grad_norm": 1.4127898216247559,
      "learning_rate": 9.6515e-05,
      "loss": 0.462,
      "step": 19313
    },
    {
      "epoch": 1.0815320864598499,
      "grad_norm": 1.3625534772872925,
      "learning_rate": 9.652e-05,
      "loss": 0.3846,
      "step": 19314
    },
    {
      "epoch": 1.081588083771979,
      "grad_norm": 1.3918216228485107,
      "learning_rate": 9.652500000000002e-05,
      "loss": 0.3508,
      "step": 19315
    },
    {
      "epoch": 1.081644081084108,
      "grad_norm": 1.4035346508026123,
      "learning_rate": 9.653000000000001e-05,
      "loss": 0.6266,
      "step": 19316
    },
    {
      "epoch": 1.081700078396237,
      "grad_norm": 1.3695547580718994,
      "learning_rate": 9.653500000000001e-05,
      "loss": 0.3242,
      "step": 19317
    },
    {
      "epoch": 1.081756075708366,
      "grad_norm": 1.5471813678741455,
      "learning_rate": 9.654000000000001e-05,
      "loss": 0.4938,
      "step": 19318
    },
    {
      "epoch": 1.081812073020495,
      "grad_norm": 1.4240856170654297,
      "learning_rate": 9.6545e-05,
      "loss": 0.3783,
      "step": 19319
    },
    {
      "epoch": 1.081868070332624,
      "grad_norm": 1.166264533996582,
      "learning_rate": 9.655e-05,
      "loss": 0.3181,
      "step": 19320
    },
    {
      "epoch": 1.081924067644753,
      "grad_norm": 1.5024768114089966,
      "learning_rate": 9.655500000000001e-05,
      "loss": 0.606,
      "step": 19321
    },
    {
      "epoch": 1.081980064956882,
      "grad_norm": 1.1460301876068115,
      "learning_rate": 9.656000000000001e-05,
      "loss": 0.436,
      "step": 19322
    },
    {
      "epoch": 1.082036062269011,
      "grad_norm": 1.4128327369689941,
      "learning_rate": 9.656500000000001e-05,
      "loss": 0.4741,
      "step": 19323
    },
    {
      "epoch": 1.08209205958114,
      "grad_norm": 1.320252776145935,
      "learning_rate": 9.657e-05,
      "loss": 0.4336,
      "step": 19324
    },
    {
      "epoch": 1.082148056893269,
      "grad_norm": 1.1598433256149292,
      "learning_rate": 9.6575e-05,
      "loss": 0.3989,
      "step": 19325
    },
    {
      "epoch": 1.0822040542053981,
      "grad_norm": 1.2715833187103271,
      "learning_rate": 9.658e-05,
      "loss": 0.3745,
      "step": 19326
    },
    {
      "epoch": 1.0822600515175271,
      "grad_norm": 1.4066200256347656,
      "learning_rate": 9.6585e-05,
      "loss": 0.4034,
      "step": 19327
    },
    {
      "epoch": 1.0823160488296562,
      "grad_norm": 1.2118477821350098,
      "learning_rate": 9.659000000000001e-05,
      "loss": 0.4449,
      "step": 19328
    },
    {
      "epoch": 1.0823720461417852,
      "grad_norm": 1.5100220441818237,
      "learning_rate": 9.6595e-05,
      "loss": 0.5064,
      "step": 19329
    },
    {
      "epoch": 1.0824280434539142,
      "grad_norm": 1.2827850580215454,
      "learning_rate": 9.66e-05,
      "loss": 0.4138,
      "step": 19330
    },
    {
      "epoch": 1.0824840407660432,
      "grad_norm": 1.5457252264022827,
      "learning_rate": 9.6605e-05,
      "loss": 0.5519,
      "step": 19331
    },
    {
      "epoch": 1.0825400380781722,
      "grad_norm": 1.2562763690948486,
      "learning_rate": 9.661e-05,
      "loss": 0.4989,
      "step": 19332
    },
    {
      "epoch": 1.0825960353903012,
      "grad_norm": 1.4032789468765259,
      "learning_rate": 9.6615e-05,
      "loss": 0.4255,
      "step": 19333
    },
    {
      "epoch": 1.0826520327024303,
      "grad_norm": 1.6235026121139526,
      "learning_rate": 9.661999999999999e-05,
      "loss": 0.4823,
      "step": 19334
    },
    {
      "epoch": 1.0827080300145593,
      "grad_norm": 1.156258225440979,
      "learning_rate": 9.6625e-05,
      "loss": 0.3425,
      "step": 19335
    },
    {
      "epoch": 1.0827640273266883,
      "grad_norm": 1.303497552871704,
      "learning_rate": 9.663000000000002e-05,
      "loss": 0.4756,
      "step": 19336
    },
    {
      "epoch": 1.0828200246388173,
      "grad_norm": 1.3825955390930176,
      "learning_rate": 9.663500000000001e-05,
      "loss": 0.3229,
      "step": 19337
    },
    {
      "epoch": 1.0828760219509463,
      "grad_norm": 1.3050435781478882,
      "learning_rate": 9.664000000000001e-05,
      "loss": 0.4469,
      "step": 19338
    },
    {
      "epoch": 1.0829320192630754,
      "grad_norm": 1.271113395690918,
      "learning_rate": 9.664500000000001e-05,
      "loss": 0.385,
      "step": 19339
    },
    {
      "epoch": 1.0829880165752044,
      "grad_norm": 1.216892957687378,
      "learning_rate": 9.665e-05,
      "loss": 0.3499,
      "step": 19340
    },
    {
      "epoch": 1.0830440138873334,
      "grad_norm": 1.3005635738372803,
      "learning_rate": 9.6655e-05,
      "loss": 0.4114,
      "step": 19341
    },
    {
      "epoch": 1.0831000111994624,
      "grad_norm": 1.6403224468231201,
      "learning_rate": 9.666e-05,
      "loss": 0.4046,
      "step": 19342
    },
    {
      "epoch": 1.0831560085115914,
      "grad_norm": 1.3765748739242554,
      "learning_rate": 9.666500000000001e-05,
      "loss": 0.5173,
      "step": 19343
    },
    {
      "epoch": 1.0832120058237205,
      "grad_norm": 1.6299631595611572,
      "learning_rate": 9.667000000000001e-05,
      "loss": 0.3675,
      "step": 19344
    },
    {
      "epoch": 1.0832680031358495,
      "grad_norm": 1.4607957601547241,
      "learning_rate": 9.6675e-05,
      "loss": 0.4178,
      "step": 19345
    },
    {
      "epoch": 1.0833240004479785,
      "grad_norm": 1.4849029779434204,
      "learning_rate": 9.668e-05,
      "loss": 0.5477,
      "step": 19346
    },
    {
      "epoch": 1.0833799977601075,
      "grad_norm": 1.2126410007476807,
      "learning_rate": 9.6685e-05,
      "loss": 0.3774,
      "step": 19347
    },
    {
      "epoch": 1.0834359950722365,
      "grad_norm": 1.4253648519515991,
      "learning_rate": 9.669e-05,
      "loss": 0.3828,
      "step": 19348
    },
    {
      "epoch": 1.0834919923843656,
      "grad_norm": 1.1440986394882202,
      "learning_rate": 9.669500000000001e-05,
      "loss": 0.2512,
      "step": 19349
    },
    {
      "epoch": 1.0835479896964946,
      "grad_norm": 1.4302659034729004,
      "learning_rate": 9.67e-05,
      "loss": 0.3265,
      "step": 19350
    },
    {
      "epoch": 1.0836039870086236,
      "grad_norm": 1.5102487802505493,
      "learning_rate": 9.6705e-05,
      "loss": 0.3999,
      "step": 19351
    },
    {
      "epoch": 1.0836599843207526,
      "grad_norm": 1.5125442743301392,
      "learning_rate": 9.671e-05,
      "loss": 0.5036,
      "step": 19352
    },
    {
      "epoch": 1.0837159816328816,
      "grad_norm": 1.411964774131775,
      "learning_rate": 9.6715e-05,
      "loss": 0.6515,
      "step": 19353
    },
    {
      "epoch": 1.0837719789450107,
      "grad_norm": 1.6002410650253296,
      "learning_rate": 9.672e-05,
      "loss": 0.6385,
      "step": 19354
    },
    {
      "epoch": 1.0838279762571397,
      "grad_norm": 1.8580633401870728,
      "learning_rate": 9.6725e-05,
      "loss": 0.605,
      "step": 19355
    },
    {
      "epoch": 1.0838839735692687,
      "grad_norm": 1.1607344150543213,
      "learning_rate": 9.673e-05,
      "loss": 0.4321,
      "step": 19356
    },
    {
      "epoch": 1.0839399708813977,
      "grad_norm": 1.320904016494751,
      "learning_rate": 9.673500000000001e-05,
      "loss": 0.4001,
      "step": 19357
    },
    {
      "epoch": 1.0839959681935267,
      "grad_norm": 1.255415439605713,
      "learning_rate": 9.674000000000001e-05,
      "loss": 0.4552,
      "step": 19358
    },
    {
      "epoch": 1.0840519655056557,
      "grad_norm": 1.2322289943695068,
      "learning_rate": 9.674500000000001e-05,
      "loss": 0.4472,
      "step": 19359
    },
    {
      "epoch": 1.0841079628177848,
      "grad_norm": 1.4008407592773438,
      "learning_rate": 9.675000000000001e-05,
      "loss": 0.5539,
      "step": 19360
    },
    {
      "epoch": 1.0841639601299138,
      "grad_norm": 1.384464144706726,
      "learning_rate": 9.6755e-05,
      "loss": 0.3652,
      "step": 19361
    },
    {
      "epoch": 1.0842199574420428,
      "grad_norm": 1.4799270629882812,
      "learning_rate": 9.676e-05,
      "loss": 0.4102,
      "step": 19362
    },
    {
      "epoch": 1.0842759547541718,
      "grad_norm": 1.5826606750488281,
      "learning_rate": 9.6765e-05,
      "loss": 0.5595,
      "step": 19363
    },
    {
      "epoch": 1.0843319520663008,
      "grad_norm": 1.3750200271606445,
      "learning_rate": 9.677000000000001e-05,
      "loss": 0.4728,
      "step": 19364
    },
    {
      "epoch": 1.0843879493784299,
      "grad_norm": 1.9382109642028809,
      "learning_rate": 9.677500000000001e-05,
      "loss": 0.5068,
      "step": 19365
    },
    {
      "epoch": 1.0844439466905589,
      "grad_norm": 1.368626356124878,
      "learning_rate": 9.678e-05,
      "loss": 0.3895,
      "step": 19366
    },
    {
      "epoch": 1.084499944002688,
      "grad_norm": 1.2477309703826904,
      "learning_rate": 9.6785e-05,
      "loss": 0.3411,
      "step": 19367
    },
    {
      "epoch": 1.084555941314817,
      "grad_norm": 1.4445922374725342,
      "learning_rate": 9.679e-05,
      "loss": 0.3236,
      "step": 19368
    },
    {
      "epoch": 1.084611938626946,
      "grad_norm": 1.6062517166137695,
      "learning_rate": 9.6795e-05,
      "loss": 0.4774,
      "step": 19369
    },
    {
      "epoch": 1.084667935939075,
      "grad_norm": 1.1364208459854126,
      "learning_rate": 9.680000000000001e-05,
      "loss": 0.4174,
      "step": 19370
    },
    {
      "epoch": 1.084723933251204,
      "grad_norm": 1.2720904350280762,
      "learning_rate": 9.6805e-05,
      "loss": 0.3245,
      "step": 19371
    },
    {
      "epoch": 1.084779930563333,
      "grad_norm": 1.477702260017395,
      "learning_rate": 9.681e-05,
      "loss": 0.5477,
      "step": 19372
    },
    {
      "epoch": 1.084835927875462,
      "grad_norm": 1.1432007551193237,
      "learning_rate": 9.6815e-05,
      "loss": 0.4232,
      "step": 19373
    },
    {
      "epoch": 1.084891925187591,
      "grad_norm": 1.412406086921692,
      "learning_rate": 9.682e-05,
      "loss": 0.4568,
      "step": 19374
    },
    {
      "epoch": 1.08494792249972,
      "grad_norm": 1.2835264205932617,
      "learning_rate": 9.682500000000001e-05,
      "loss": 0.377,
      "step": 19375
    },
    {
      "epoch": 1.085003919811849,
      "grad_norm": 1.3556221723556519,
      "learning_rate": 9.683e-05,
      "loss": 0.3866,
      "step": 19376
    },
    {
      "epoch": 1.085059917123978,
      "grad_norm": 1.238083004951477,
      "learning_rate": 9.6835e-05,
      "loss": 0.54,
      "step": 19377
    },
    {
      "epoch": 1.085115914436107,
      "grad_norm": 1.3918308019638062,
      "learning_rate": 9.684000000000001e-05,
      "loss": 0.4972,
      "step": 19378
    },
    {
      "epoch": 1.0851719117482361,
      "grad_norm": 1.1738877296447754,
      "learning_rate": 9.684500000000001e-05,
      "loss": 0.3268,
      "step": 19379
    },
    {
      "epoch": 1.0852279090603651,
      "grad_norm": 1.3730249404907227,
      "learning_rate": 9.685000000000001e-05,
      "loss": 0.4502,
      "step": 19380
    },
    {
      "epoch": 1.0852839063724942,
      "grad_norm": 1.495382308959961,
      "learning_rate": 9.685500000000001e-05,
      "loss": 0.428,
      "step": 19381
    },
    {
      "epoch": 1.0853399036846232,
      "grad_norm": 1.2371774911880493,
      "learning_rate": 9.686e-05,
      "loss": 0.3295,
      "step": 19382
    },
    {
      "epoch": 1.0853959009967522,
      "grad_norm": 1.256272792816162,
      "learning_rate": 9.6865e-05,
      "loss": 0.5049,
      "step": 19383
    },
    {
      "epoch": 1.0854518983088812,
      "grad_norm": 1.5617402791976929,
      "learning_rate": 9.687e-05,
      "loss": 0.3444,
      "step": 19384
    },
    {
      "epoch": 1.0855078956210102,
      "grad_norm": 1.2238343954086304,
      "learning_rate": 9.687500000000001e-05,
      "loss": 0.4621,
      "step": 19385
    },
    {
      "epoch": 1.0855638929331393,
      "grad_norm": 1.7954620122909546,
      "learning_rate": 9.688000000000001e-05,
      "loss": 0.536,
      "step": 19386
    },
    {
      "epoch": 1.0856198902452683,
      "grad_norm": 1.3573501110076904,
      "learning_rate": 9.6885e-05,
      "loss": 0.4388,
      "step": 19387
    },
    {
      "epoch": 1.0856758875573973,
      "grad_norm": 1.399285078048706,
      "learning_rate": 9.689e-05,
      "loss": 0.5818,
      "step": 19388
    },
    {
      "epoch": 1.0857318848695263,
      "grad_norm": 1.4428678750991821,
      "learning_rate": 9.6895e-05,
      "loss": 0.4234,
      "step": 19389
    },
    {
      "epoch": 1.0857878821816553,
      "grad_norm": 1.4424530267715454,
      "learning_rate": 9.69e-05,
      "loss": 0.4421,
      "step": 19390
    },
    {
      "epoch": 1.0858438794937844,
      "grad_norm": 1.754023790359497,
      "learning_rate": 9.6905e-05,
      "loss": 0.4428,
      "step": 19391
    },
    {
      "epoch": 1.0858998768059134,
      "grad_norm": 1.343117594718933,
      "learning_rate": 9.691e-05,
      "loss": 0.312,
      "step": 19392
    },
    {
      "epoch": 1.0859558741180424,
      "grad_norm": 1.1642510890960693,
      "learning_rate": 9.6915e-05,
      "loss": 0.3357,
      "step": 19393
    },
    {
      "epoch": 1.0860118714301714,
      "grad_norm": 1.524693250656128,
      "learning_rate": 9.692e-05,
      "loss": 0.375,
      "step": 19394
    },
    {
      "epoch": 1.0860678687423004,
      "grad_norm": 1.7570533752441406,
      "learning_rate": 9.6925e-05,
      "loss": 0.5246,
      "step": 19395
    },
    {
      "epoch": 1.0861238660544295,
      "grad_norm": 1.2837694883346558,
      "learning_rate": 9.693000000000001e-05,
      "loss": 0.3471,
      "step": 19396
    },
    {
      "epoch": 1.0861798633665585,
      "grad_norm": 1.1543211936950684,
      "learning_rate": 9.6935e-05,
      "loss": 0.3616,
      "step": 19397
    },
    {
      "epoch": 1.0862358606786875,
      "grad_norm": 1.3171840906143188,
      "learning_rate": 9.694e-05,
      "loss": 0.4241,
      "step": 19398
    },
    {
      "epoch": 1.0862918579908165,
      "grad_norm": 1.5649409294128418,
      "learning_rate": 9.694500000000001e-05,
      "loss": 0.4944,
      "step": 19399
    },
    {
      "epoch": 1.0863478553029455,
      "grad_norm": 1.2931212186813354,
      "learning_rate": 9.695000000000001e-05,
      "loss": 0.5089,
      "step": 19400
    },
    {
      "epoch": 1.0864038526150746,
      "grad_norm": 1.3104225397109985,
      "learning_rate": 9.695500000000001e-05,
      "loss": 0.4478,
      "step": 19401
    },
    {
      "epoch": 1.0864598499272036,
      "grad_norm": 1.1763811111450195,
      "learning_rate": 9.696000000000001e-05,
      "loss": 0.4075,
      "step": 19402
    },
    {
      "epoch": 1.0865158472393326,
      "grad_norm": 1.2341376543045044,
      "learning_rate": 9.6965e-05,
      "loss": 0.5711,
      "step": 19403
    },
    {
      "epoch": 1.0865718445514616,
      "grad_norm": 1.2019206285476685,
      "learning_rate": 9.697e-05,
      "loss": 0.3148,
      "step": 19404
    },
    {
      "epoch": 1.0866278418635906,
      "grad_norm": 1.6193609237670898,
      "learning_rate": 9.6975e-05,
      "loss": 0.4713,
      "step": 19405
    },
    {
      "epoch": 1.0866838391757196,
      "grad_norm": 1.4534461498260498,
      "learning_rate": 9.698000000000001e-05,
      "loss": 0.4004,
      "step": 19406
    },
    {
      "epoch": 1.0867398364878487,
      "grad_norm": 1.1196978092193604,
      "learning_rate": 9.698500000000001e-05,
      "loss": 0.3419,
      "step": 19407
    },
    {
      "epoch": 1.0867958337999777,
      "grad_norm": 1.272823452949524,
      "learning_rate": 9.699e-05,
      "loss": 0.4797,
      "step": 19408
    },
    {
      "epoch": 1.0868518311121067,
      "grad_norm": 6.032217979431152,
      "learning_rate": 9.6995e-05,
      "loss": 0.5645,
      "step": 19409
    },
    {
      "epoch": 1.0869078284242357,
      "grad_norm": 1.2282921075820923,
      "learning_rate": 9.7e-05,
      "loss": 0.494,
      "step": 19410
    },
    {
      "epoch": 1.0869638257363647,
      "grad_norm": 1.432591199874878,
      "learning_rate": 9.7005e-05,
      "loss": 0.4727,
      "step": 19411
    },
    {
      "epoch": 1.0870198230484938,
      "grad_norm": 1.428158164024353,
      "learning_rate": 9.701e-05,
      "loss": 0.3715,
      "step": 19412
    },
    {
      "epoch": 1.0870758203606228,
      "grad_norm": 1.33553147315979,
      "learning_rate": 9.7015e-05,
      "loss": 0.4723,
      "step": 19413
    },
    {
      "epoch": 1.0871318176727518,
      "grad_norm": 1.2606146335601807,
      "learning_rate": 9.702e-05,
      "loss": 0.4094,
      "step": 19414
    },
    {
      "epoch": 1.0871878149848808,
      "grad_norm": 1.3955553770065308,
      "learning_rate": 9.7025e-05,
      "loss": 0.3692,
      "step": 19415
    },
    {
      "epoch": 1.0872438122970098,
      "grad_norm": 1.202028751373291,
      "learning_rate": 9.703000000000001e-05,
      "loss": 0.4275,
      "step": 19416
    },
    {
      "epoch": 1.0872998096091389,
      "grad_norm": 1.497829556465149,
      "learning_rate": 9.703500000000001e-05,
      "loss": 0.6001,
      "step": 19417
    },
    {
      "epoch": 1.0873558069212679,
      "grad_norm": 1.1924201250076294,
      "learning_rate": 9.704e-05,
      "loss": 0.3614,
      "step": 19418
    },
    {
      "epoch": 1.087411804233397,
      "grad_norm": 1.5991381406784058,
      "learning_rate": 9.7045e-05,
      "loss": 0.5625,
      "step": 19419
    },
    {
      "epoch": 1.087467801545526,
      "grad_norm": 1.3351833820343018,
      "learning_rate": 9.705e-05,
      "loss": 0.5205,
      "step": 19420
    },
    {
      "epoch": 1.087523798857655,
      "grad_norm": 1.2472409009933472,
      "learning_rate": 9.705500000000001e-05,
      "loss": 0.3638,
      "step": 19421
    },
    {
      "epoch": 1.087579796169784,
      "grad_norm": 1.1877214908599854,
      "learning_rate": 9.706000000000001e-05,
      "loss": 0.4038,
      "step": 19422
    },
    {
      "epoch": 1.087635793481913,
      "grad_norm": 1.28447425365448,
      "learning_rate": 9.7065e-05,
      "loss": 0.4491,
      "step": 19423
    },
    {
      "epoch": 1.0876917907940418,
      "grad_norm": 1.2487019300460815,
      "learning_rate": 9.707e-05,
      "loss": 0.5416,
      "step": 19424
    },
    {
      "epoch": 1.0877477881061708,
      "grad_norm": 1.412729024887085,
      "learning_rate": 9.7075e-05,
      "loss": 0.4928,
      "step": 19425
    },
    {
      "epoch": 1.0878037854182998,
      "grad_norm": 1.8077267408370972,
      "learning_rate": 9.708e-05,
      "loss": 0.4177,
      "step": 19426
    },
    {
      "epoch": 1.0878597827304288,
      "grad_norm": 1.2820024490356445,
      "learning_rate": 9.708500000000001e-05,
      "loss": 0.4856,
      "step": 19427
    },
    {
      "epoch": 1.0879157800425578,
      "grad_norm": 1.74324369430542,
      "learning_rate": 9.709000000000001e-05,
      "loss": 0.3792,
      "step": 19428
    },
    {
      "epoch": 1.0879717773546869,
      "grad_norm": 1.2279449701309204,
      "learning_rate": 9.7095e-05,
      "loss": 0.341,
      "step": 19429
    },
    {
      "epoch": 1.0880277746668159,
      "grad_norm": 1.4273266792297363,
      "learning_rate": 9.71e-05,
      "loss": 0.4763,
      "step": 19430
    },
    {
      "epoch": 1.088083771978945,
      "grad_norm": 1.2334628105163574,
      "learning_rate": 9.7105e-05,
      "loss": 0.3511,
      "step": 19431
    },
    {
      "epoch": 1.088139769291074,
      "grad_norm": 1.3678807020187378,
      "learning_rate": 9.711e-05,
      "loss": 0.3458,
      "step": 19432
    },
    {
      "epoch": 1.088195766603203,
      "grad_norm": 1.2043136358261108,
      "learning_rate": 9.7115e-05,
      "loss": 0.348,
      "step": 19433
    },
    {
      "epoch": 1.088251763915332,
      "grad_norm": 1.4696546792984009,
      "learning_rate": 9.712e-05,
      "loss": 0.5889,
      "step": 19434
    },
    {
      "epoch": 1.088307761227461,
      "grad_norm": 1.3678463697433472,
      "learning_rate": 9.7125e-05,
      "loss": 0.5672,
      "step": 19435
    },
    {
      "epoch": 1.08836375853959,
      "grad_norm": 1.7388309240341187,
      "learning_rate": 9.713000000000001e-05,
      "loss": 0.3944,
      "step": 19436
    },
    {
      "epoch": 1.088419755851719,
      "grad_norm": 1.696213722229004,
      "learning_rate": 9.713500000000001e-05,
      "loss": 0.3871,
      "step": 19437
    },
    {
      "epoch": 1.088475753163848,
      "grad_norm": 1.5011556148529053,
      "learning_rate": 9.714000000000001e-05,
      "loss": 0.4874,
      "step": 19438
    },
    {
      "epoch": 1.088531750475977,
      "grad_norm": 1.4647469520568848,
      "learning_rate": 9.7145e-05,
      "loss": 0.3355,
      "step": 19439
    },
    {
      "epoch": 1.088587747788106,
      "grad_norm": 1.3213003873825073,
      "learning_rate": 9.715e-05,
      "loss": 0.33,
      "step": 19440
    },
    {
      "epoch": 1.088643745100235,
      "grad_norm": 1.250311255455017,
      "learning_rate": 9.7155e-05,
      "loss": 0.4597,
      "step": 19441
    },
    {
      "epoch": 1.0886997424123641,
      "grad_norm": 1.4594535827636719,
      "learning_rate": 9.716000000000001e-05,
      "loss": 0.4793,
      "step": 19442
    },
    {
      "epoch": 1.0887557397244931,
      "grad_norm": 1.3642311096191406,
      "learning_rate": 9.716500000000001e-05,
      "loss": 0.4608,
      "step": 19443
    },
    {
      "epoch": 1.0888117370366222,
      "grad_norm": 2.3334622383117676,
      "learning_rate": 9.717e-05,
      "loss": 0.5047,
      "step": 19444
    },
    {
      "epoch": 1.0888677343487512,
      "grad_norm": 1.3684736490249634,
      "learning_rate": 9.7175e-05,
      "loss": 0.3814,
      "step": 19445
    },
    {
      "epoch": 1.0889237316608802,
      "grad_norm": 1.1973755359649658,
      "learning_rate": 9.718e-05,
      "loss": 0.4501,
      "step": 19446
    },
    {
      "epoch": 1.0889797289730092,
      "grad_norm": 1.4940571784973145,
      "learning_rate": 9.7185e-05,
      "loss": 0.6219,
      "step": 19447
    },
    {
      "epoch": 1.0890357262851382,
      "grad_norm": 1.252989649772644,
      "learning_rate": 9.719000000000001e-05,
      "loss": 0.41,
      "step": 19448
    },
    {
      "epoch": 1.0890917235972672,
      "grad_norm": 1.185983419418335,
      "learning_rate": 9.719500000000001e-05,
      "loss": 0.4355,
      "step": 19449
    },
    {
      "epoch": 1.0891477209093963,
      "grad_norm": 1.1518192291259766,
      "learning_rate": 9.72e-05,
      "loss": 0.3613,
      "step": 19450
    },
    {
      "epoch": 1.0892037182215253,
      "grad_norm": 1.5924357175827026,
      "learning_rate": 9.7205e-05,
      "loss": 0.4771,
      "step": 19451
    },
    {
      "epoch": 1.0892597155336543,
      "grad_norm": 1.1367547512054443,
      "learning_rate": 9.721e-05,
      "loss": 0.4243,
      "step": 19452
    },
    {
      "epoch": 1.0893157128457833,
      "grad_norm": 1.4977604150772095,
      "learning_rate": 9.7215e-05,
      "loss": 0.3691,
      "step": 19453
    },
    {
      "epoch": 1.0893717101579123,
      "grad_norm": 1.2454397678375244,
      "learning_rate": 9.722e-05,
      "loss": 0.5785,
      "step": 19454
    },
    {
      "epoch": 1.0894277074700414,
      "grad_norm": 1.5086647272109985,
      "learning_rate": 9.7225e-05,
      "loss": 0.3714,
      "step": 19455
    },
    {
      "epoch": 1.0894837047821704,
      "grad_norm": 1.3603461980819702,
      "learning_rate": 9.723000000000002e-05,
      "loss": 0.5453,
      "step": 19456
    },
    {
      "epoch": 1.0895397020942994,
      "grad_norm": 1.7683531045913696,
      "learning_rate": 9.723500000000001e-05,
      "loss": 0.7335,
      "step": 19457
    },
    {
      "epoch": 1.0895956994064284,
      "grad_norm": 2.0208988189697266,
      "learning_rate": 9.724000000000001e-05,
      "loss": 0.4242,
      "step": 19458
    },
    {
      "epoch": 1.0896516967185574,
      "grad_norm": 1.403065800666809,
      "learning_rate": 9.724500000000001e-05,
      "loss": 0.4319,
      "step": 19459
    },
    {
      "epoch": 1.0897076940306865,
      "grad_norm": 1.7681505680084229,
      "learning_rate": 9.725e-05,
      "loss": 0.5649,
      "step": 19460
    },
    {
      "epoch": 1.0897636913428155,
      "grad_norm": 5.0404133796691895,
      "learning_rate": 9.7255e-05,
      "loss": 0.3528,
      "step": 19461
    },
    {
      "epoch": 1.0898196886549445,
      "grad_norm": 1.498591661453247,
      "learning_rate": 9.726e-05,
      "loss": 0.5161,
      "step": 19462
    },
    {
      "epoch": 1.0898756859670735,
      "grad_norm": 1.8758131265640259,
      "learning_rate": 9.726500000000001e-05,
      "loss": 0.5007,
      "step": 19463
    },
    {
      "epoch": 1.0899316832792025,
      "grad_norm": 1.260574221611023,
      "learning_rate": 9.727000000000001e-05,
      "loss": 0.3745,
      "step": 19464
    },
    {
      "epoch": 1.0899876805913316,
      "grad_norm": 1.312031865119934,
      "learning_rate": 9.7275e-05,
      "loss": 0.4584,
      "step": 19465
    },
    {
      "epoch": 1.0900436779034606,
      "grad_norm": 1.5896401405334473,
      "learning_rate": 9.728e-05,
      "loss": 0.47,
      "step": 19466
    },
    {
      "epoch": 1.0900996752155896,
      "grad_norm": 1.0249112844467163,
      "learning_rate": 9.7285e-05,
      "loss": 0.3238,
      "step": 19467
    },
    {
      "epoch": 1.0901556725277186,
      "grad_norm": 1.251260757446289,
      "learning_rate": 9.729e-05,
      "loss": 0.3825,
      "step": 19468
    },
    {
      "epoch": 1.0902116698398476,
      "grad_norm": 1.4393593072891235,
      "learning_rate": 9.729500000000001e-05,
      "loss": 0.4745,
      "step": 19469
    },
    {
      "epoch": 1.0902676671519766,
      "grad_norm": 1.361510157585144,
      "learning_rate": 9.730000000000001e-05,
      "loss": 0.419,
      "step": 19470
    },
    {
      "epoch": 1.0903236644641057,
      "grad_norm": 1.2182210683822632,
      "learning_rate": 9.7305e-05,
      "loss": 0.4122,
      "step": 19471
    },
    {
      "epoch": 1.0903796617762347,
      "grad_norm": 1.2895756959915161,
      "learning_rate": 9.731e-05,
      "loss": 0.406,
      "step": 19472
    },
    {
      "epoch": 1.0904356590883637,
      "grad_norm": 1.3974794149398804,
      "learning_rate": 9.7315e-05,
      "loss": 0.5104,
      "step": 19473
    },
    {
      "epoch": 1.0904916564004927,
      "grad_norm": 1.6768306493759155,
      "learning_rate": 9.732e-05,
      "loss": 0.4613,
      "step": 19474
    },
    {
      "epoch": 1.0905476537126217,
      "grad_norm": 1.3968340158462524,
      "learning_rate": 9.7325e-05,
      "loss": 0.4519,
      "step": 19475
    },
    {
      "epoch": 1.0906036510247508,
      "grad_norm": 1.7102560997009277,
      "learning_rate": 9.733e-05,
      "loss": 0.5399,
      "step": 19476
    },
    {
      "epoch": 1.0906596483368798,
      "grad_norm": 1.3042027950286865,
      "learning_rate": 9.733500000000002e-05,
      "loss": 0.4404,
      "step": 19477
    },
    {
      "epoch": 1.0907156456490088,
      "grad_norm": 1.2797521352767944,
      "learning_rate": 9.734000000000001e-05,
      "loss": 0.4302,
      "step": 19478
    },
    {
      "epoch": 1.0907716429611378,
      "grad_norm": 1.3735034465789795,
      "learning_rate": 9.734500000000001e-05,
      "loss": 0.4131,
      "step": 19479
    },
    {
      "epoch": 1.0908276402732668,
      "grad_norm": 1.4389700889587402,
      "learning_rate": 9.735000000000001e-05,
      "loss": 0.5966,
      "step": 19480
    },
    {
      "epoch": 1.0908836375853959,
      "grad_norm": 1.6658045053482056,
      "learning_rate": 9.7355e-05,
      "loss": 0.5419,
      "step": 19481
    },
    {
      "epoch": 1.0909396348975249,
      "grad_norm": 1.3026015758514404,
      "learning_rate": 9.736e-05,
      "loss": 0.5054,
      "step": 19482
    },
    {
      "epoch": 1.090995632209654,
      "grad_norm": 1.3731954097747803,
      "learning_rate": 9.7365e-05,
      "loss": 0.4267,
      "step": 19483
    },
    {
      "epoch": 1.091051629521783,
      "grad_norm": 1.4135109186172485,
      "learning_rate": 9.737000000000001e-05,
      "loss": 0.451,
      "step": 19484
    },
    {
      "epoch": 1.091107626833912,
      "grad_norm": 1.2790257930755615,
      "learning_rate": 9.737500000000001e-05,
      "loss": 0.3954,
      "step": 19485
    },
    {
      "epoch": 1.091163624146041,
      "grad_norm": 1.4345476627349854,
      "learning_rate": 9.738e-05,
      "loss": 0.4735,
      "step": 19486
    },
    {
      "epoch": 1.09121962145817,
      "grad_norm": 1.2349601984024048,
      "learning_rate": 9.7385e-05,
      "loss": 0.3137,
      "step": 19487
    },
    {
      "epoch": 1.091275618770299,
      "grad_norm": 1.20396089553833,
      "learning_rate": 9.739e-05,
      "loss": 0.4095,
      "step": 19488
    },
    {
      "epoch": 1.091331616082428,
      "grad_norm": 1.3240493535995483,
      "learning_rate": 9.7395e-05,
      "loss": 0.4282,
      "step": 19489
    },
    {
      "epoch": 1.091387613394557,
      "grad_norm": 1.3587307929992676,
      "learning_rate": 9.74e-05,
      "loss": 0.5745,
      "step": 19490
    },
    {
      "epoch": 1.091443610706686,
      "grad_norm": 1.4722732305526733,
      "learning_rate": 9.7405e-05,
      "loss": 0.4488,
      "step": 19491
    },
    {
      "epoch": 1.091499608018815,
      "grad_norm": 1.6319332122802734,
      "learning_rate": 9.741e-05,
      "loss": 0.4726,
      "step": 19492
    },
    {
      "epoch": 1.091555605330944,
      "grad_norm": 1.2616441249847412,
      "learning_rate": 9.7415e-05,
      "loss": 0.4085,
      "step": 19493
    },
    {
      "epoch": 1.091611602643073,
      "grad_norm": 1.4479644298553467,
      "learning_rate": 9.742e-05,
      "loss": 0.4895,
      "step": 19494
    },
    {
      "epoch": 1.0916675999552021,
      "grad_norm": 1.1906683444976807,
      "learning_rate": 9.7425e-05,
      "loss": 0.3664,
      "step": 19495
    },
    {
      "epoch": 1.0917235972673311,
      "grad_norm": 1.1659542322158813,
      "learning_rate": 9.743000000000001e-05,
      "loss": 0.455,
      "step": 19496
    },
    {
      "epoch": 1.0917795945794602,
      "grad_norm": 1.4838449954986572,
      "learning_rate": 9.7435e-05,
      "loss": 0.4508,
      "step": 19497
    },
    {
      "epoch": 1.0918355918915892,
      "grad_norm": 1.3363466262817383,
      "learning_rate": 9.744000000000002e-05,
      "loss": 0.3449,
      "step": 19498
    },
    {
      "epoch": 1.0918915892037182,
      "grad_norm": 1.0771715641021729,
      "learning_rate": 9.744500000000001e-05,
      "loss": 0.3355,
      "step": 19499
    },
    {
      "epoch": 1.0919475865158472,
      "grad_norm": 1.1941698789596558,
      "learning_rate": 9.745000000000001e-05,
      "loss": 0.4237,
      "step": 19500
    },
    {
      "epoch": 1.0920035838279762,
      "grad_norm": 1.3635395765304565,
      "learning_rate": 9.745500000000001e-05,
      "loss": 0.5534,
      "step": 19501
    },
    {
      "epoch": 1.0920595811401053,
      "grad_norm": 1.4608535766601562,
      "learning_rate": 9.746e-05,
      "loss": 0.3937,
      "step": 19502
    },
    {
      "epoch": 1.0921155784522343,
      "grad_norm": 1.342836856842041,
      "learning_rate": 9.7465e-05,
      "loss": 0.4632,
      "step": 19503
    },
    {
      "epoch": 1.0921715757643633,
      "grad_norm": 1.4179961681365967,
      "learning_rate": 9.747e-05,
      "loss": 0.5522,
      "step": 19504
    },
    {
      "epoch": 1.0922275730764923,
      "grad_norm": 1.4744871854782104,
      "learning_rate": 9.747500000000001e-05,
      "loss": 0.4017,
      "step": 19505
    },
    {
      "epoch": 1.0922835703886213,
      "grad_norm": 1.4139132499694824,
      "learning_rate": 9.748000000000001e-05,
      "loss": 0.4151,
      "step": 19506
    },
    {
      "epoch": 1.0923395677007504,
      "grad_norm": 1.4782044887542725,
      "learning_rate": 9.7485e-05,
      "loss": 0.603,
      "step": 19507
    },
    {
      "epoch": 1.0923955650128794,
      "grad_norm": 1.4270756244659424,
      "learning_rate": 9.749e-05,
      "loss": 0.4025,
      "step": 19508
    },
    {
      "epoch": 1.0924515623250084,
      "grad_norm": 1.2587467432022095,
      "learning_rate": 9.7495e-05,
      "loss": 0.3908,
      "step": 19509
    },
    {
      "epoch": 1.0925075596371374,
      "grad_norm": 1.3345550298690796,
      "learning_rate": 9.75e-05,
      "loss": 0.3895,
      "step": 19510
    },
    {
      "epoch": 1.0925635569492664,
      "grad_norm": 1.5031352043151855,
      "learning_rate": 9.7505e-05,
      "loss": 0.4788,
      "step": 19511
    },
    {
      "epoch": 1.0926195542613955,
      "grad_norm": 1.3319941759109497,
      "learning_rate": 9.751e-05,
      "loss": 0.5293,
      "step": 19512
    },
    {
      "epoch": 1.0926755515735245,
      "grad_norm": 1.1762086153030396,
      "learning_rate": 9.7515e-05,
      "loss": 0.4477,
      "step": 19513
    },
    {
      "epoch": 1.0927315488856535,
      "grad_norm": 1.3730016946792603,
      "learning_rate": 9.752e-05,
      "loss": 0.4041,
      "step": 19514
    },
    {
      "epoch": 1.0927875461977825,
      "grad_norm": 1.591992735862732,
      "learning_rate": 9.7525e-05,
      "loss": 0.481,
      "step": 19515
    },
    {
      "epoch": 1.0928435435099115,
      "grad_norm": 1.3490495681762695,
      "learning_rate": 9.753e-05,
      "loss": 0.3612,
      "step": 19516
    },
    {
      "epoch": 1.0928995408220405,
      "grad_norm": 1.2400485277175903,
      "learning_rate": 9.753500000000001e-05,
      "loss": 0.4447,
      "step": 19517
    },
    {
      "epoch": 1.0929555381341696,
      "grad_norm": 1.4158413410186768,
      "learning_rate": 9.754e-05,
      "loss": 0.5537,
      "step": 19518
    },
    {
      "epoch": 1.0930115354462986,
      "grad_norm": 1.2359211444854736,
      "learning_rate": 9.7545e-05,
      "loss": 0.5455,
      "step": 19519
    },
    {
      "epoch": 1.0930675327584276,
      "grad_norm": 1.2567752599716187,
      "learning_rate": 9.755000000000001e-05,
      "loss": 0.364,
      "step": 19520
    },
    {
      "epoch": 1.0931235300705566,
      "grad_norm": 1.3764561414718628,
      "learning_rate": 9.755500000000001e-05,
      "loss": 0.4302,
      "step": 19521
    },
    {
      "epoch": 1.0931795273826856,
      "grad_norm": 1.321619987487793,
      "learning_rate": 9.756000000000001e-05,
      "loss": 0.4692,
      "step": 19522
    },
    {
      "epoch": 1.0932355246948147,
      "grad_norm": 1.33164381980896,
      "learning_rate": 9.7565e-05,
      "loss": 0.3829,
      "step": 19523
    },
    {
      "epoch": 1.0932915220069437,
      "grad_norm": 1.5573786497116089,
      "learning_rate": 9.757e-05,
      "loss": 0.4894,
      "step": 19524
    },
    {
      "epoch": 1.0933475193190727,
      "grad_norm": 1.4197397232055664,
      "learning_rate": 9.7575e-05,
      "loss": 0.3993,
      "step": 19525
    },
    {
      "epoch": 1.0934035166312017,
      "grad_norm": 1.4187829494476318,
      "learning_rate": 9.758000000000001e-05,
      "loss": 0.4583,
      "step": 19526
    },
    {
      "epoch": 1.0934595139433307,
      "grad_norm": 1.202857255935669,
      "learning_rate": 9.758500000000001e-05,
      "loss": 0.3423,
      "step": 19527
    },
    {
      "epoch": 1.0935155112554598,
      "grad_norm": 1.4020442962646484,
      "learning_rate": 9.759e-05,
      "loss": 0.4366,
      "step": 19528
    },
    {
      "epoch": 1.0935715085675888,
      "grad_norm": 1.3176480531692505,
      "learning_rate": 9.7595e-05,
      "loss": 0.4086,
      "step": 19529
    },
    {
      "epoch": 1.0936275058797178,
      "grad_norm": 1.1130186319351196,
      "learning_rate": 9.76e-05,
      "loss": 0.3393,
      "step": 19530
    },
    {
      "epoch": 1.0936835031918468,
      "grad_norm": 1.2685071229934692,
      "learning_rate": 9.7605e-05,
      "loss": 0.4529,
      "step": 19531
    },
    {
      "epoch": 1.0937395005039758,
      "grad_norm": 1.6565825939178467,
      "learning_rate": 9.761e-05,
      "loss": 0.4969,
      "step": 19532
    },
    {
      "epoch": 1.0937954978161049,
      "grad_norm": 1.3428248167037964,
      "learning_rate": 9.7615e-05,
      "loss": 0.384,
      "step": 19533
    },
    {
      "epoch": 1.0938514951282339,
      "grad_norm": 1.4593104124069214,
      "learning_rate": 9.762e-05,
      "loss": 0.4969,
      "step": 19534
    },
    {
      "epoch": 1.093907492440363,
      "grad_norm": 1.5722178220748901,
      "learning_rate": 9.7625e-05,
      "loss": 0.5102,
      "step": 19535
    },
    {
      "epoch": 1.093963489752492,
      "grad_norm": 1.4600510597229004,
      "learning_rate": 9.763e-05,
      "loss": 0.3806,
      "step": 19536
    },
    {
      "epoch": 1.094019487064621,
      "grad_norm": 1.3687782287597656,
      "learning_rate": 9.763500000000001e-05,
      "loss": 0.5298,
      "step": 19537
    },
    {
      "epoch": 1.09407548437675,
      "grad_norm": 1.763283610343933,
      "learning_rate": 9.764000000000001e-05,
      "loss": 0.4613,
      "step": 19538
    },
    {
      "epoch": 1.094131481688879,
      "grad_norm": 1.4355372190475464,
      "learning_rate": 9.7645e-05,
      "loss": 0.4203,
      "step": 19539
    },
    {
      "epoch": 1.094187479001008,
      "grad_norm": 1.6046817302703857,
      "learning_rate": 9.765e-05,
      "loss": 0.3555,
      "step": 19540
    },
    {
      "epoch": 1.094243476313137,
      "grad_norm": 1.6429340839385986,
      "learning_rate": 9.765500000000001e-05,
      "loss": 0.4662,
      "step": 19541
    },
    {
      "epoch": 1.094299473625266,
      "grad_norm": 1.4519562721252441,
      "learning_rate": 9.766000000000001e-05,
      "loss": 0.4883,
      "step": 19542
    },
    {
      "epoch": 1.094355470937395,
      "grad_norm": 1.3624401092529297,
      "learning_rate": 9.766500000000001e-05,
      "loss": 0.4762,
      "step": 19543
    },
    {
      "epoch": 1.094411468249524,
      "grad_norm": 1.355677843093872,
      "learning_rate": 9.767e-05,
      "loss": 0.3493,
      "step": 19544
    },
    {
      "epoch": 1.094467465561653,
      "grad_norm": 6.697079658508301,
      "learning_rate": 9.7675e-05,
      "loss": 0.4235,
      "step": 19545
    },
    {
      "epoch": 1.094523462873782,
      "grad_norm": 1.138710379600525,
      "learning_rate": 9.768e-05,
      "loss": 0.3862,
      "step": 19546
    },
    {
      "epoch": 1.0945794601859111,
      "grad_norm": 1.5354340076446533,
      "learning_rate": 9.768500000000001e-05,
      "loss": 0.4287,
      "step": 19547
    },
    {
      "epoch": 1.0946354574980401,
      "grad_norm": 1.3873952627182007,
      "learning_rate": 9.769000000000001e-05,
      "loss": 0.3725,
      "step": 19548
    },
    {
      "epoch": 1.0946914548101692,
      "grad_norm": 1.2548458576202393,
      "learning_rate": 9.7695e-05,
      "loss": 0.3202,
      "step": 19549
    },
    {
      "epoch": 1.0947474521222982,
      "grad_norm": 1.577109932899475,
      "learning_rate": 9.77e-05,
      "loss": 0.587,
      "step": 19550
    },
    {
      "epoch": 1.0948034494344272,
      "grad_norm": 1.3174759149551392,
      "learning_rate": 9.7705e-05,
      "loss": 0.4666,
      "step": 19551
    },
    {
      "epoch": 1.0948594467465562,
      "grad_norm": 1.461652159690857,
      "learning_rate": 9.771e-05,
      "loss": 0.4711,
      "step": 19552
    },
    {
      "epoch": 1.0949154440586852,
      "grad_norm": 1.3167462348937988,
      "learning_rate": 9.7715e-05,
      "loss": 0.4441,
      "step": 19553
    },
    {
      "epoch": 1.0949714413708143,
      "grad_norm": 1.1816465854644775,
      "learning_rate": 9.772e-05,
      "loss": 0.3607,
      "step": 19554
    },
    {
      "epoch": 1.0950274386829433,
      "grad_norm": 1.3829306364059448,
      "learning_rate": 9.7725e-05,
      "loss": 0.4564,
      "step": 19555
    },
    {
      "epoch": 1.0950834359950723,
      "grad_norm": 1.3023033142089844,
      "learning_rate": 9.773e-05,
      "loss": 0.3954,
      "step": 19556
    },
    {
      "epoch": 1.0951394333072013,
      "grad_norm": 1.0449341535568237,
      "learning_rate": 9.773500000000001e-05,
      "loss": 0.3799,
      "step": 19557
    },
    {
      "epoch": 1.0951954306193303,
      "grad_norm": 1.3946932554244995,
      "learning_rate": 9.774000000000001e-05,
      "loss": 0.428,
      "step": 19558
    },
    {
      "epoch": 1.0952514279314594,
      "grad_norm": 1.5530643463134766,
      "learning_rate": 9.774500000000001e-05,
      "loss": 0.565,
      "step": 19559
    },
    {
      "epoch": 1.0953074252435884,
      "grad_norm": 1.2083131074905396,
      "learning_rate": 9.775e-05,
      "loss": 0.2811,
      "step": 19560
    },
    {
      "epoch": 1.0953634225557174,
      "grad_norm": 1.2603507041931152,
      "learning_rate": 9.7755e-05,
      "loss": 0.596,
      "step": 19561
    },
    {
      "epoch": 1.0954194198678464,
      "grad_norm": 1.5251781940460205,
      "learning_rate": 9.776000000000001e-05,
      "loss": 0.4129,
      "step": 19562
    },
    {
      "epoch": 1.0954754171799754,
      "grad_norm": 1.2989113330841064,
      "learning_rate": 9.776500000000001e-05,
      "loss": 0.4234,
      "step": 19563
    },
    {
      "epoch": 1.0955314144921044,
      "grad_norm": 1.4201045036315918,
      "learning_rate": 9.777000000000001e-05,
      "loss": 0.4128,
      "step": 19564
    },
    {
      "epoch": 1.0955874118042335,
      "grad_norm": 1.3607844114303589,
      "learning_rate": 9.7775e-05,
      "loss": 0.6465,
      "step": 19565
    },
    {
      "epoch": 1.0956434091163625,
      "grad_norm": 1.6310099363327026,
      "learning_rate": 9.778e-05,
      "loss": 0.4283,
      "step": 19566
    },
    {
      "epoch": 1.0956994064284915,
      "grad_norm": 1.1884732246398926,
      "learning_rate": 9.7785e-05,
      "loss": 0.3746,
      "step": 19567
    },
    {
      "epoch": 1.0957554037406205,
      "grad_norm": 1.3570232391357422,
      "learning_rate": 9.779e-05,
      "loss": 0.4903,
      "step": 19568
    },
    {
      "epoch": 1.0958114010527495,
      "grad_norm": 1.44740629196167,
      "learning_rate": 9.779500000000001e-05,
      "loss": 0.4056,
      "step": 19569
    },
    {
      "epoch": 1.0958673983648786,
      "grad_norm": 1.1972736120224,
      "learning_rate": 9.78e-05,
      "loss": 0.3,
      "step": 19570
    },
    {
      "epoch": 1.0959233956770076,
      "grad_norm": 1.7060189247131348,
      "learning_rate": 9.7805e-05,
      "loss": 0.4381,
      "step": 19571
    },
    {
      "epoch": 1.0959793929891366,
      "grad_norm": 1.6647682189941406,
      "learning_rate": 9.781e-05,
      "loss": 0.4836,
      "step": 19572
    },
    {
      "epoch": 1.0960353903012656,
      "grad_norm": 1.5249102115631104,
      "learning_rate": 9.7815e-05,
      "loss": 0.4272,
      "step": 19573
    },
    {
      "epoch": 1.0960913876133946,
      "grad_norm": 1.2867717742919922,
      "learning_rate": 9.782e-05,
      "loss": 0.5391,
      "step": 19574
    },
    {
      "epoch": 1.0961473849255237,
      "grad_norm": 1.3553507328033447,
      "learning_rate": 9.7825e-05,
      "loss": 0.4903,
      "step": 19575
    },
    {
      "epoch": 1.0962033822376527,
      "grad_norm": 1.1969085931777954,
      "learning_rate": 9.783e-05,
      "loss": 0.4377,
      "step": 19576
    },
    {
      "epoch": 1.0962593795497817,
      "grad_norm": 1.2163243293762207,
      "learning_rate": 9.783500000000001e-05,
      "loss": 0.3771,
      "step": 19577
    },
    {
      "epoch": 1.0963153768619107,
      "grad_norm": 2.052074432373047,
      "learning_rate": 9.784000000000001e-05,
      "loss": 0.3332,
      "step": 19578
    },
    {
      "epoch": 1.0963713741740397,
      "grad_norm": 1.3686301708221436,
      "learning_rate": 9.784500000000001e-05,
      "loss": 0.5045,
      "step": 19579
    },
    {
      "epoch": 1.0964273714861688,
      "grad_norm": 1.6917316913604736,
      "learning_rate": 9.785e-05,
      "loss": 0.4673,
      "step": 19580
    },
    {
      "epoch": 1.0964833687982978,
      "grad_norm": 1.682552456855774,
      "learning_rate": 9.7855e-05,
      "loss": 0.4664,
      "step": 19581
    },
    {
      "epoch": 1.0965393661104268,
      "grad_norm": 1.4329495429992676,
      "learning_rate": 9.786e-05,
      "loss": 0.4861,
      "step": 19582
    },
    {
      "epoch": 1.0965953634225558,
      "grad_norm": 1.289271354675293,
      "learning_rate": 9.786500000000001e-05,
      "loss": 0.3885,
      "step": 19583
    },
    {
      "epoch": 1.0966513607346848,
      "grad_norm": 1.4165812730789185,
      "learning_rate": 9.787000000000001e-05,
      "loss": 0.4461,
      "step": 19584
    },
    {
      "epoch": 1.0967073580468139,
      "grad_norm": 1.2960909605026245,
      "learning_rate": 9.787500000000001e-05,
      "loss": 0.5017,
      "step": 19585
    },
    {
      "epoch": 1.0967633553589429,
      "grad_norm": 1.6284939050674438,
      "learning_rate": 9.788e-05,
      "loss": 0.453,
      "step": 19586
    },
    {
      "epoch": 1.096819352671072,
      "grad_norm": 1.2743760347366333,
      "learning_rate": 9.7885e-05,
      "loss": 0.2907,
      "step": 19587
    },
    {
      "epoch": 1.096875349983201,
      "grad_norm": 1.3700569868087769,
      "learning_rate": 9.789e-05,
      "loss": 0.4638,
      "step": 19588
    },
    {
      "epoch": 1.09693134729533,
      "grad_norm": 1.3631771802902222,
      "learning_rate": 9.7895e-05,
      "loss": 0.3999,
      "step": 19589
    },
    {
      "epoch": 1.096987344607459,
      "grad_norm": 1.553858757019043,
      "learning_rate": 9.790000000000001e-05,
      "loss": 0.514,
      "step": 19590
    },
    {
      "epoch": 1.097043341919588,
      "grad_norm": 1.1983616352081299,
      "learning_rate": 9.7905e-05,
      "loss": 0.3723,
      "step": 19591
    },
    {
      "epoch": 1.097099339231717,
      "grad_norm": 1.3355696201324463,
      "learning_rate": 9.791e-05,
      "loss": 0.3802,
      "step": 19592
    },
    {
      "epoch": 1.097155336543846,
      "grad_norm": 1.3248353004455566,
      "learning_rate": 9.7915e-05,
      "loss": 0.5072,
      "step": 19593
    },
    {
      "epoch": 1.097211333855975,
      "grad_norm": 1.1041831970214844,
      "learning_rate": 9.792e-05,
      "loss": 0.3715,
      "step": 19594
    },
    {
      "epoch": 1.097267331168104,
      "grad_norm": 1.2678688764572144,
      "learning_rate": 9.7925e-05,
      "loss": 0.4019,
      "step": 19595
    },
    {
      "epoch": 1.097323328480233,
      "grad_norm": 1.381998896598816,
      "learning_rate": 9.793e-05,
      "loss": 0.5244,
      "step": 19596
    },
    {
      "epoch": 1.097379325792362,
      "grad_norm": 1.5407527685165405,
      "learning_rate": 9.7935e-05,
      "loss": 0.4575,
      "step": 19597
    },
    {
      "epoch": 1.097435323104491,
      "grad_norm": 1.1095999479293823,
      "learning_rate": 9.794000000000001e-05,
      "loss": 0.3277,
      "step": 19598
    },
    {
      "epoch": 1.0974913204166201,
      "grad_norm": 1.348344326019287,
      "learning_rate": 9.794500000000001e-05,
      "loss": 0.4471,
      "step": 19599
    },
    {
      "epoch": 1.097547317728749,
      "grad_norm": 1.2373028993606567,
      "learning_rate": 9.795000000000001e-05,
      "loss": 0.5059,
      "step": 19600
    },
    {
      "epoch": 1.097603315040878,
      "grad_norm": 1.3573837280273438,
      "learning_rate": 9.7955e-05,
      "loss": 0.4558,
      "step": 19601
    },
    {
      "epoch": 1.097659312353007,
      "grad_norm": 1.5080405473709106,
      "learning_rate": 9.796e-05,
      "loss": 0.5021,
      "step": 19602
    },
    {
      "epoch": 1.097715309665136,
      "grad_norm": 1.1538140773773193,
      "learning_rate": 9.7965e-05,
      "loss": 0.365,
      "step": 19603
    },
    {
      "epoch": 1.097771306977265,
      "grad_norm": 1.2251487970352173,
      "learning_rate": 9.797000000000001e-05,
      "loss": 0.3905,
      "step": 19604
    },
    {
      "epoch": 1.097827304289394,
      "grad_norm": 1.2731026411056519,
      "learning_rate": 9.797500000000001e-05,
      "loss": 0.3414,
      "step": 19605
    },
    {
      "epoch": 1.097883301601523,
      "grad_norm": 1.5197020769119263,
      "learning_rate": 9.798000000000001e-05,
      "loss": 0.5708,
      "step": 19606
    },
    {
      "epoch": 1.097939298913652,
      "grad_norm": 1.5866721868515015,
      "learning_rate": 9.7985e-05,
      "loss": 0.5346,
      "step": 19607
    },
    {
      "epoch": 1.097995296225781,
      "grad_norm": 1.284306526184082,
      "learning_rate": 9.799e-05,
      "loss": 0.4819,
      "step": 19608
    },
    {
      "epoch": 1.09805129353791,
      "grad_norm": 1.3480887413024902,
      "learning_rate": 9.7995e-05,
      "loss": 0.4383,
      "step": 19609
    },
    {
      "epoch": 1.098107290850039,
      "grad_norm": 1.904942274093628,
      "learning_rate": 9.8e-05,
      "loss": 0.433,
      "step": 19610
    },
    {
      "epoch": 1.0981632881621681,
      "grad_norm": 1.3913054466247559,
      "learning_rate": 9.800500000000001e-05,
      "loss": 0.4088,
      "step": 19611
    },
    {
      "epoch": 1.0982192854742971,
      "grad_norm": 1.3774702548980713,
      "learning_rate": 9.801e-05,
      "loss": 0.3585,
      "step": 19612
    },
    {
      "epoch": 1.0982752827864262,
      "grad_norm": 1.439720630645752,
      "learning_rate": 9.8015e-05,
      "loss": 0.4768,
      "step": 19613
    },
    {
      "epoch": 1.0983312800985552,
      "grad_norm": 1.4266635179519653,
      "learning_rate": 9.802e-05,
      "loss": 0.4101,
      "step": 19614
    },
    {
      "epoch": 1.0983872774106842,
      "grad_norm": 1.3650580644607544,
      "learning_rate": 9.8025e-05,
      "loss": 0.326,
      "step": 19615
    },
    {
      "epoch": 1.0984432747228132,
      "grad_norm": 1.794957160949707,
      "learning_rate": 9.803e-05,
      "loss": 0.5428,
      "step": 19616
    },
    {
      "epoch": 1.0984992720349422,
      "grad_norm": 1.2034342288970947,
      "learning_rate": 9.8035e-05,
      "loss": 0.361,
      "step": 19617
    },
    {
      "epoch": 1.0985552693470713,
      "grad_norm": 1.705862045288086,
      "learning_rate": 9.804e-05,
      "loss": 0.5285,
      "step": 19618
    },
    {
      "epoch": 1.0986112666592003,
      "grad_norm": 1.4839178323745728,
      "learning_rate": 9.804500000000001e-05,
      "loss": 0.5371,
      "step": 19619
    },
    {
      "epoch": 1.0986672639713293,
      "grad_norm": 1.4856160879135132,
      "learning_rate": 9.805000000000001e-05,
      "loss": 0.4007,
      "step": 19620
    },
    {
      "epoch": 1.0987232612834583,
      "grad_norm": 1.6990814208984375,
      "learning_rate": 9.805500000000001e-05,
      "loss": 0.4334,
      "step": 19621
    },
    {
      "epoch": 1.0987792585955873,
      "grad_norm": 1.3838716745376587,
      "learning_rate": 9.806e-05,
      "loss": 0.5617,
      "step": 19622
    },
    {
      "epoch": 1.0988352559077164,
      "grad_norm": 1.3111406564712524,
      "learning_rate": 9.8065e-05,
      "loss": 0.5367,
      "step": 19623
    },
    {
      "epoch": 1.0988912532198454,
      "grad_norm": 1.1596437692642212,
      "learning_rate": 9.807e-05,
      "loss": 0.3757,
      "step": 19624
    },
    {
      "epoch": 1.0989472505319744,
      "grad_norm": 1.465855360031128,
      "learning_rate": 9.807500000000001e-05,
      "loss": 0.4408,
      "step": 19625
    },
    {
      "epoch": 1.0990032478441034,
      "grad_norm": 1.2543396949768066,
      "learning_rate": 9.808000000000001e-05,
      "loss": 0.4489,
      "step": 19626
    },
    {
      "epoch": 1.0990592451562324,
      "grad_norm": 1.525762677192688,
      "learning_rate": 9.808500000000001e-05,
      "loss": 0.5929,
      "step": 19627
    },
    {
      "epoch": 1.0991152424683615,
      "grad_norm": 1.179938554763794,
      "learning_rate": 9.809e-05,
      "loss": 0.4061,
      "step": 19628
    },
    {
      "epoch": 1.0991712397804905,
      "grad_norm": 1.2849570512771606,
      "learning_rate": 9.8095e-05,
      "loss": 0.4987,
      "step": 19629
    },
    {
      "epoch": 1.0992272370926195,
      "grad_norm": 1.1673990488052368,
      "learning_rate": 9.81e-05,
      "loss": 0.3156,
      "step": 19630
    },
    {
      "epoch": 1.0992832344047485,
      "grad_norm": 1.4120322465896606,
      "learning_rate": 9.8105e-05,
      "loss": 0.4933,
      "step": 19631
    },
    {
      "epoch": 1.0993392317168775,
      "grad_norm": 1.819898009300232,
      "learning_rate": 9.811000000000001e-05,
      "loss": 0.4087,
      "step": 19632
    },
    {
      "epoch": 1.0993952290290065,
      "grad_norm": 1.1839394569396973,
      "learning_rate": 9.8115e-05,
      "loss": 0.4765,
      "step": 19633
    },
    {
      "epoch": 1.0994512263411356,
      "grad_norm": 1.347901463508606,
      "learning_rate": 9.812e-05,
      "loss": 0.417,
      "step": 19634
    },
    {
      "epoch": 1.0995072236532646,
      "grad_norm": 1.4624589681625366,
      "learning_rate": 9.8125e-05,
      "loss": 0.4186,
      "step": 19635
    },
    {
      "epoch": 1.0995632209653936,
      "grad_norm": 1.3179152011871338,
      "learning_rate": 9.813e-05,
      "loss": 0.3827,
      "step": 19636
    },
    {
      "epoch": 1.0996192182775226,
      "grad_norm": 1.1995933055877686,
      "learning_rate": 9.8135e-05,
      "loss": 0.3866,
      "step": 19637
    },
    {
      "epoch": 1.0996752155896516,
      "grad_norm": 1.2979129552841187,
      "learning_rate": 9.814e-05,
      "loss": 0.4895,
      "step": 19638
    },
    {
      "epoch": 1.0997312129017807,
      "grad_norm": 1.2624669075012207,
      "learning_rate": 9.8145e-05,
      "loss": 0.4745,
      "step": 19639
    },
    {
      "epoch": 1.0997872102139097,
      "grad_norm": 1.2604899406433105,
      "learning_rate": 9.815000000000001e-05,
      "loss": 0.4489,
      "step": 19640
    },
    {
      "epoch": 1.0998432075260387,
      "grad_norm": 1.3490601778030396,
      "learning_rate": 9.815500000000001e-05,
      "loss": 0.414,
      "step": 19641
    },
    {
      "epoch": 1.0998992048381677,
      "grad_norm": 1.4333354234695435,
      "learning_rate": 9.816000000000001e-05,
      "loss": 0.43,
      "step": 19642
    },
    {
      "epoch": 1.0999552021502967,
      "grad_norm": 1.474410057067871,
      "learning_rate": 9.8165e-05,
      "loss": 0.5682,
      "step": 19643
    },
    {
      "epoch": 1.1000111994624258,
      "grad_norm": 1.4290974140167236,
      "learning_rate": 9.817e-05,
      "loss": 0.4719,
      "step": 19644
    },
    {
      "epoch": 1.1000671967745548,
      "grad_norm": 1.4126131534576416,
      "learning_rate": 9.8175e-05,
      "loss": 0.4099,
      "step": 19645
    },
    {
      "epoch": 1.1001231940866838,
      "grad_norm": 1.2462862730026245,
      "learning_rate": 9.818000000000001e-05,
      "loss": 0.3927,
      "step": 19646
    },
    {
      "epoch": 1.1001791913988128,
      "grad_norm": 1.3805636167526245,
      "learning_rate": 9.818500000000001e-05,
      "loss": 0.3539,
      "step": 19647
    },
    {
      "epoch": 1.1002351887109418,
      "grad_norm": 1.6017307043075562,
      "learning_rate": 9.819000000000001e-05,
      "loss": 0.434,
      "step": 19648
    },
    {
      "epoch": 1.1002911860230709,
      "grad_norm": 1.4915167093276978,
      "learning_rate": 9.8195e-05,
      "loss": 0.2893,
      "step": 19649
    },
    {
      "epoch": 1.1003471833351999,
      "grad_norm": 1.3202970027923584,
      "learning_rate": 9.82e-05,
      "loss": 0.4855,
      "step": 19650
    },
    {
      "epoch": 1.100403180647329,
      "grad_norm": 1.255157232284546,
      "learning_rate": 9.8205e-05,
      "loss": 0.383,
      "step": 19651
    },
    {
      "epoch": 1.100459177959458,
      "grad_norm": 1.5192474126815796,
      "learning_rate": 9.821e-05,
      "loss": 0.5071,
      "step": 19652
    },
    {
      "epoch": 1.100515175271587,
      "grad_norm": 1.2128627300262451,
      "learning_rate": 9.821500000000001e-05,
      "loss": 0.352,
      "step": 19653
    },
    {
      "epoch": 1.100571172583716,
      "grad_norm": 1.457595944404602,
      "learning_rate": 9.822e-05,
      "loss": 0.3849,
      "step": 19654
    },
    {
      "epoch": 1.100627169895845,
      "grad_norm": 1.1318013668060303,
      "learning_rate": 9.8225e-05,
      "loss": 0.3474,
      "step": 19655
    },
    {
      "epoch": 1.100683167207974,
      "grad_norm": 1.130577564239502,
      "learning_rate": 9.823e-05,
      "loss": 0.4295,
      "step": 19656
    },
    {
      "epoch": 1.100739164520103,
      "grad_norm": 1.4926098585128784,
      "learning_rate": 9.8235e-05,
      "loss": 0.3817,
      "step": 19657
    },
    {
      "epoch": 1.100795161832232,
      "grad_norm": 1.287928819656372,
      "learning_rate": 9.824000000000001e-05,
      "loss": 0.3621,
      "step": 19658
    },
    {
      "epoch": 1.100851159144361,
      "grad_norm": 1.5619772672653198,
      "learning_rate": 9.8245e-05,
      "loss": 0.4055,
      "step": 19659
    },
    {
      "epoch": 1.10090715645649,
      "grad_norm": 1.1738126277923584,
      "learning_rate": 9.825e-05,
      "loss": 0.301,
      "step": 19660
    },
    {
      "epoch": 1.100963153768619,
      "grad_norm": 1.506620168685913,
      "learning_rate": 9.825500000000001e-05,
      "loss": 0.4636,
      "step": 19661
    },
    {
      "epoch": 1.101019151080748,
      "grad_norm": 1.37851881980896,
      "learning_rate": 9.826000000000001e-05,
      "loss": 0.4036,
      "step": 19662
    },
    {
      "epoch": 1.1010751483928771,
      "grad_norm": 1.145391821861267,
      "learning_rate": 9.826500000000001e-05,
      "loss": 0.4037,
      "step": 19663
    },
    {
      "epoch": 1.1011311457050061,
      "grad_norm": 1.5921082496643066,
      "learning_rate": 9.827e-05,
      "loss": 0.4837,
      "step": 19664
    },
    {
      "epoch": 1.1011871430171352,
      "grad_norm": 1.2360310554504395,
      "learning_rate": 9.8275e-05,
      "loss": 0.3909,
      "step": 19665
    },
    {
      "epoch": 1.1012431403292642,
      "grad_norm": 1.3479084968566895,
      "learning_rate": 9.828e-05,
      "loss": 0.3891,
      "step": 19666
    },
    {
      "epoch": 1.1012991376413932,
      "grad_norm": 1.4180879592895508,
      "learning_rate": 9.8285e-05,
      "loss": 0.4824,
      "step": 19667
    },
    {
      "epoch": 1.1013551349535222,
      "grad_norm": 1.416598916053772,
      "learning_rate": 9.829000000000001e-05,
      "loss": 0.4931,
      "step": 19668
    },
    {
      "epoch": 1.1014111322656512,
      "grad_norm": 1.3986800909042358,
      "learning_rate": 9.8295e-05,
      "loss": 0.4377,
      "step": 19669
    },
    {
      "epoch": 1.1014671295777803,
      "grad_norm": 1.4680436849594116,
      "learning_rate": 9.83e-05,
      "loss": 0.4207,
      "step": 19670
    },
    {
      "epoch": 1.1015231268899093,
      "grad_norm": 1.2534722089767456,
      "learning_rate": 9.8305e-05,
      "loss": 0.4465,
      "step": 19671
    },
    {
      "epoch": 1.1015791242020383,
      "grad_norm": 1.4206799268722534,
      "learning_rate": 9.831e-05,
      "loss": 0.3915,
      "step": 19672
    },
    {
      "epoch": 1.1016351215141673,
      "grad_norm": 1.613774299621582,
      "learning_rate": 9.8315e-05,
      "loss": 0.5944,
      "step": 19673
    },
    {
      "epoch": 1.1016911188262963,
      "grad_norm": 1.422844409942627,
      "learning_rate": 9.832000000000001e-05,
      "loss": 0.4587,
      "step": 19674
    },
    {
      "epoch": 1.1017471161384254,
      "grad_norm": 1.4094058275222778,
      "learning_rate": 9.8325e-05,
      "loss": 0.4965,
      "step": 19675
    },
    {
      "epoch": 1.1018031134505544,
      "grad_norm": 1.4469799995422363,
      "learning_rate": 9.833e-05,
      "loss": 0.4653,
      "step": 19676
    },
    {
      "epoch": 1.1018591107626834,
      "grad_norm": 1.2660422325134277,
      "learning_rate": 9.8335e-05,
      "loss": 0.4329,
      "step": 19677
    },
    {
      "epoch": 1.1019151080748124,
      "grad_norm": 1.3411662578582764,
      "learning_rate": 9.834000000000001e-05,
      "loss": 0.4491,
      "step": 19678
    },
    {
      "epoch": 1.1019711053869414,
      "grad_norm": 1.233513593673706,
      "learning_rate": 9.834500000000001e-05,
      "loss": 0.5012,
      "step": 19679
    },
    {
      "epoch": 1.1020271026990704,
      "grad_norm": 1.741809368133545,
      "learning_rate": 9.835e-05,
      "loss": 0.4552,
      "step": 19680
    },
    {
      "epoch": 1.1020831000111995,
      "grad_norm": 1.3911317586898804,
      "learning_rate": 9.8355e-05,
      "loss": 0.4226,
      "step": 19681
    },
    {
      "epoch": 1.1021390973233285,
      "grad_norm": 1.219521403312683,
      "learning_rate": 9.836000000000001e-05,
      "loss": 0.3876,
      "step": 19682
    },
    {
      "epoch": 1.1021950946354575,
      "grad_norm": 1.204106092453003,
      "learning_rate": 9.836500000000001e-05,
      "loss": 0.3628,
      "step": 19683
    },
    {
      "epoch": 1.1022510919475865,
      "grad_norm": 1.6147651672363281,
      "learning_rate": 9.837000000000001e-05,
      "loss": 0.4808,
      "step": 19684
    },
    {
      "epoch": 1.1023070892597155,
      "grad_norm": 1.4001786708831787,
      "learning_rate": 9.8375e-05,
      "loss": 0.4303,
      "step": 19685
    },
    {
      "epoch": 1.1023630865718446,
      "grad_norm": 1.4113482236862183,
      "learning_rate": 9.838e-05,
      "loss": 0.5507,
      "step": 19686
    },
    {
      "epoch": 1.1024190838839736,
      "grad_norm": 1.3065470457077026,
      "learning_rate": 9.8385e-05,
      "loss": 0.4914,
      "step": 19687
    },
    {
      "epoch": 1.1024750811961026,
      "grad_norm": 1.4770466089248657,
      "learning_rate": 9.839e-05,
      "loss": 0.4447,
      "step": 19688
    },
    {
      "epoch": 1.1025310785082316,
      "grad_norm": 1.696991205215454,
      "learning_rate": 9.839500000000001e-05,
      "loss": 0.6721,
      "step": 19689
    },
    {
      "epoch": 1.1025870758203606,
      "grad_norm": 1.3368000984191895,
      "learning_rate": 9.84e-05,
      "loss": 0.3748,
      "step": 19690
    },
    {
      "epoch": 1.1026430731324897,
      "grad_norm": 1.4557217359542847,
      "learning_rate": 9.8405e-05,
      "loss": 0.3309,
      "step": 19691
    },
    {
      "epoch": 1.1026990704446187,
      "grad_norm": 1.1468091011047363,
      "learning_rate": 9.841e-05,
      "loss": 0.323,
      "step": 19692
    },
    {
      "epoch": 1.1027550677567477,
      "grad_norm": 1.4091640710830688,
      "learning_rate": 9.8415e-05,
      "loss": 0.4994,
      "step": 19693
    },
    {
      "epoch": 1.1028110650688767,
      "grad_norm": 1.1214548349380493,
      "learning_rate": 9.842e-05,
      "loss": 0.4106,
      "step": 19694
    },
    {
      "epoch": 1.1028670623810057,
      "grad_norm": 1.9984763860702515,
      "learning_rate": 9.842500000000001e-05,
      "loss": 0.5373,
      "step": 19695
    },
    {
      "epoch": 1.1029230596931348,
      "grad_norm": 1.2726832628250122,
      "learning_rate": 9.843e-05,
      "loss": 0.4828,
      "step": 19696
    },
    {
      "epoch": 1.1029790570052638,
      "grad_norm": 1.4429244995117188,
      "learning_rate": 9.8435e-05,
      "loss": 0.4125,
      "step": 19697
    },
    {
      "epoch": 1.1030350543173928,
      "grad_norm": 1.4349812269210815,
      "learning_rate": 9.844000000000001e-05,
      "loss": 0.3782,
      "step": 19698
    },
    {
      "epoch": 1.1030910516295218,
      "grad_norm": 1.1398873329162598,
      "learning_rate": 9.844500000000001e-05,
      "loss": 0.3644,
      "step": 19699
    },
    {
      "epoch": 1.1031470489416508,
      "grad_norm": 1.466614842414856,
      "learning_rate": 9.845000000000001e-05,
      "loss": 0.4185,
      "step": 19700
    },
    {
      "epoch": 1.1032030462537799,
      "grad_norm": 1.4406763315200806,
      "learning_rate": 9.8455e-05,
      "loss": 0.5374,
      "step": 19701
    },
    {
      "epoch": 1.1032590435659089,
      "grad_norm": 1.4726649522781372,
      "learning_rate": 9.846e-05,
      "loss": 0.4873,
      "step": 19702
    },
    {
      "epoch": 1.103315040878038,
      "grad_norm": 1.8432420492172241,
      "learning_rate": 9.846500000000001e-05,
      "loss": 0.5391,
      "step": 19703
    },
    {
      "epoch": 1.103371038190167,
      "grad_norm": 1.6330406665802002,
      "learning_rate": 9.847000000000001e-05,
      "loss": 0.4411,
      "step": 19704
    },
    {
      "epoch": 1.103427035502296,
      "grad_norm": 1.1099885702133179,
      "learning_rate": 9.847500000000001e-05,
      "loss": 0.358,
      "step": 19705
    },
    {
      "epoch": 1.103483032814425,
      "grad_norm": 1.2998239994049072,
      "learning_rate": 9.848e-05,
      "loss": 0.6747,
      "step": 19706
    },
    {
      "epoch": 1.103539030126554,
      "grad_norm": 1.2724019289016724,
      "learning_rate": 9.8485e-05,
      "loss": 0.3967,
      "step": 19707
    },
    {
      "epoch": 1.103595027438683,
      "grad_norm": 1.0637582540512085,
      "learning_rate": 9.849e-05,
      "loss": 0.5166,
      "step": 19708
    },
    {
      "epoch": 1.103651024750812,
      "grad_norm": 1.5849372148513794,
      "learning_rate": 9.8495e-05,
      "loss": 0.5009,
      "step": 19709
    },
    {
      "epoch": 1.103707022062941,
      "grad_norm": 1.1329911947250366,
      "learning_rate": 9.850000000000001e-05,
      "loss": 0.4318,
      "step": 19710
    },
    {
      "epoch": 1.10376301937507,
      "grad_norm": 1.33184814453125,
      "learning_rate": 9.8505e-05,
      "loss": 0.3949,
      "step": 19711
    },
    {
      "epoch": 1.103819016687199,
      "grad_norm": 1.7777830362319946,
      "learning_rate": 9.851e-05,
      "loss": 0.2906,
      "step": 19712
    },
    {
      "epoch": 1.103875013999328,
      "grad_norm": 1.3954609632492065,
      "learning_rate": 9.8515e-05,
      "loss": 0.4586,
      "step": 19713
    },
    {
      "epoch": 1.103931011311457,
      "grad_norm": 1.442158818244934,
      "learning_rate": 9.852e-05,
      "loss": 0.4672,
      "step": 19714
    },
    {
      "epoch": 1.1039870086235861,
      "grad_norm": 1.2966798543930054,
      "learning_rate": 9.8525e-05,
      "loss": 0.3713,
      "step": 19715
    },
    {
      "epoch": 1.1040430059357151,
      "grad_norm": 1.3441522121429443,
      "learning_rate": 9.853e-05,
      "loss": 0.5063,
      "step": 19716
    },
    {
      "epoch": 1.1040990032478442,
      "grad_norm": 1.5825473070144653,
      "learning_rate": 9.8535e-05,
      "loss": 0.4611,
      "step": 19717
    },
    {
      "epoch": 1.1041550005599732,
      "grad_norm": 1.2520579099655151,
      "learning_rate": 9.854000000000002e-05,
      "loss": 0.4068,
      "step": 19718
    },
    {
      "epoch": 1.1042109978721022,
      "grad_norm": 1.5830234289169312,
      "learning_rate": 9.854500000000001e-05,
      "loss": 0.544,
      "step": 19719
    },
    {
      "epoch": 1.1042669951842312,
      "grad_norm": 1.355677604675293,
      "learning_rate": 9.855000000000001e-05,
      "loss": 0.4227,
      "step": 19720
    },
    {
      "epoch": 1.1043229924963602,
      "grad_norm": 1.1585386991500854,
      "learning_rate": 9.855500000000001e-05,
      "loss": 0.412,
      "step": 19721
    },
    {
      "epoch": 1.1043789898084893,
      "grad_norm": 1.249297857284546,
      "learning_rate": 9.856e-05,
      "loss": 0.4418,
      "step": 19722
    },
    {
      "epoch": 1.1044349871206183,
      "grad_norm": 1.499921202659607,
      "learning_rate": 9.8565e-05,
      "loss": 0.4302,
      "step": 19723
    },
    {
      "epoch": 1.1044909844327473,
      "grad_norm": 1.3647310733795166,
      "learning_rate": 9.857000000000001e-05,
      "loss": 0.3631,
      "step": 19724
    },
    {
      "epoch": 1.1045469817448763,
      "grad_norm": 1.7964094877243042,
      "learning_rate": 9.857500000000001e-05,
      "loss": 0.5309,
      "step": 19725
    },
    {
      "epoch": 1.1046029790570053,
      "grad_norm": 1.1209633350372314,
      "learning_rate": 9.858000000000001e-05,
      "loss": 0.3896,
      "step": 19726
    },
    {
      "epoch": 1.1046589763691343,
      "grad_norm": 1.3948016166687012,
      "learning_rate": 9.8585e-05,
      "loss": 0.3297,
      "step": 19727
    },
    {
      "epoch": 1.1047149736812634,
      "grad_norm": 1.5150035619735718,
      "learning_rate": 9.859e-05,
      "loss": 0.419,
      "step": 19728
    },
    {
      "epoch": 1.1047709709933924,
      "grad_norm": 1.659343957901001,
      "learning_rate": 9.8595e-05,
      "loss": 0.6311,
      "step": 19729
    },
    {
      "epoch": 1.1048269683055214,
      "grad_norm": 1.4386720657348633,
      "learning_rate": 9.86e-05,
      "loss": 0.4453,
      "step": 19730
    },
    {
      "epoch": 1.1048829656176504,
      "grad_norm": 1.4890894889831543,
      "learning_rate": 9.860500000000001e-05,
      "loss": 0.518,
      "step": 19731
    },
    {
      "epoch": 1.1049389629297794,
      "grad_norm": 1.4367374181747437,
      "learning_rate": 9.861e-05,
      "loss": 0.5873,
      "step": 19732
    },
    {
      "epoch": 1.1049949602419085,
      "grad_norm": 1.6929882764816284,
      "learning_rate": 9.8615e-05,
      "loss": 0.4449,
      "step": 19733
    },
    {
      "epoch": 1.1050509575540375,
      "grad_norm": 1.3177821636199951,
      "learning_rate": 9.862e-05,
      "loss": 0.3507,
      "step": 19734
    },
    {
      "epoch": 1.1051069548661665,
      "grad_norm": 1.3555843830108643,
      "learning_rate": 9.8625e-05,
      "loss": 0.4354,
      "step": 19735
    },
    {
      "epoch": 1.1051629521782955,
      "grad_norm": 1.4369921684265137,
      "learning_rate": 9.863e-05,
      "loss": 0.4411,
      "step": 19736
    },
    {
      "epoch": 1.1052189494904245,
      "grad_norm": 1.2860368490219116,
      "learning_rate": 9.8635e-05,
      "loss": 0.3695,
      "step": 19737
    },
    {
      "epoch": 1.1052749468025536,
      "grad_norm": 1.36696457862854,
      "learning_rate": 9.864e-05,
      "loss": 0.4118,
      "step": 19738
    },
    {
      "epoch": 1.1053309441146826,
      "grad_norm": 1.2955561876296997,
      "learning_rate": 9.864500000000002e-05,
      "loss": 0.3879,
      "step": 19739
    },
    {
      "epoch": 1.1053869414268116,
      "grad_norm": 1.7504791021347046,
      "learning_rate": 9.865000000000001e-05,
      "loss": 0.4655,
      "step": 19740
    },
    {
      "epoch": 1.1054429387389406,
      "grad_norm": 1.4993438720703125,
      "learning_rate": 9.865500000000001e-05,
      "loss": 0.4497,
      "step": 19741
    },
    {
      "epoch": 1.1054989360510696,
      "grad_norm": 1.411922574043274,
      "learning_rate": 9.866000000000001e-05,
      "loss": 0.5643,
      "step": 19742
    },
    {
      "epoch": 1.1055549333631987,
      "grad_norm": 1.3271328210830688,
      "learning_rate": 9.8665e-05,
      "loss": 0.4518,
      "step": 19743
    },
    {
      "epoch": 1.1056109306753277,
      "grad_norm": 1.5522270202636719,
      "learning_rate": 9.867e-05,
      "loss": 0.4669,
      "step": 19744
    },
    {
      "epoch": 1.1056669279874567,
      "grad_norm": 1.6213088035583496,
      "learning_rate": 9.8675e-05,
      "loss": 0.39,
      "step": 19745
    },
    {
      "epoch": 1.1057229252995857,
      "grad_norm": 1.1774853467941284,
      "learning_rate": 9.868000000000001e-05,
      "loss": 0.4392,
      "step": 19746
    },
    {
      "epoch": 1.1057789226117147,
      "grad_norm": 1.3302592039108276,
      "learning_rate": 9.868500000000001e-05,
      "loss": 0.4107,
      "step": 19747
    },
    {
      "epoch": 1.1058349199238438,
      "grad_norm": 1.5091776847839355,
      "learning_rate": 9.869e-05,
      "loss": 0.4309,
      "step": 19748
    },
    {
      "epoch": 1.1058909172359728,
      "grad_norm": 1.4652016162872314,
      "learning_rate": 9.8695e-05,
      "loss": 0.4422,
      "step": 19749
    },
    {
      "epoch": 1.1059469145481018,
      "grad_norm": 1.6019277572631836,
      "learning_rate": 9.87e-05,
      "loss": 0.4204,
      "step": 19750
    },
    {
      "epoch": 1.1060029118602308,
      "grad_norm": 1.4288212060928345,
      "learning_rate": 9.8705e-05,
      "loss": 0.4841,
      "step": 19751
    },
    {
      "epoch": 1.1060589091723598,
      "grad_norm": 1.8394572734832764,
      "learning_rate": 9.871000000000001e-05,
      "loss": 0.4747,
      "step": 19752
    },
    {
      "epoch": 1.1061149064844888,
      "grad_norm": 1.221510648727417,
      "learning_rate": 9.8715e-05,
      "loss": 0.463,
      "step": 19753
    },
    {
      "epoch": 1.1061709037966179,
      "grad_norm": 1.2516493797302246,
      "learning_rate": 9.872e-05,
      "loss": 0.3969,
      "step": 19754
    },
    {
      "epoch": 1.1062269011087467,
      "grad_norm": 1.5816762447357178,
      "learning_rate": 9.8725e-05,
      "loss": 0.449,
      "step": 19755
    },
    {
      "epoch": 1.1062828984208757,
      "grad_norm": 1.6973276138305664,
      "learning_rate": 9.873e-05,
      "loss": 0.4978,
      "step": 19756
    },
    {
      "epoch": 1.1063388957330047,
      "grad_norm": 1.1648826599121094,
      "learning_rate": 9.8735e-05,
      "loss": 0.3786,
      "step": 19757
    },
    {
      "epoch": 1.1063948930451337,
      "grad_norm": 1.2789723873138428,
      "learning_rate": 9.874e-05,
      "loss": 0.3556,
      "step": 19758
    },
    {
      "epoch": 1.1064508903572627,
      "grad_norm": 1.9684841632843018,
      "learning_rate": 9.8745e-05,
      "loss": 0.4716,
      "step": 19759
    },
    {
      "epoch": 1.1065068876693918,
      "grad_norm": 1.2908488512039185,
      "learning_rate": 9.875000000000002e-05,
      "loss": 0.4018,
      "step": 19760
    },
    {
      "epoch": 1.1065628849815208,
      "grad_norm": 1.6126617193222046,
      "learning_rate": 9.875500000000001e-05,
      "loss": 0.4275,
      "step": 19761
    },
    {
      "epoch": 1.1066188822936498,
      "grad_norm": 1.5051411390304565,
      "learning_rate": 9.876000000000001e-05,
      "loss": 0.4489,
      "step": 19762
    },
    {
      "epoch": 1.1066748796057788,
      "grad_norm": 1.4963515996932983,
      "learning_rate": 9.876500000000001e-05,
      "loss": 0.4229,
      "step": 19763
    },
    {
      "epoch": 1.1067308769179078,
      "grad_norm": 1.419785499572754,
      "learning_rate": 9.877e-05,
      "loss": 0.4226,
      "step": 19764
    },
    {
      "epoch": 1.1067868742300369,
      "grad_norm": 1.2918188571929932,
      "learning_rate": 9.8775e-05,
      "loss": 0.4028,
      "step": 19765
    },
    {
      "epoch": 1.1068428715421659,
      "grad_norm": 1.3430591821670532,
      "learning_rate": 9.878e-05,
      "loss": 0.4244,
      "step": 19766
    },
    {
      "epoch": 1.106898868854295,
      "grad_norm": 1.2693862915039062,
      "learning_rate": 9.878500000000001e-05,
      "loss": 0.4661,
      "step": 19767
    },
    {
      "epoch": 1.106954866166424,
      "grad_norm": 1.434861421585083,
      "learning_rate": 9.879000000000001e-05,
      "loss": 0.5121,
      "step": 19768
    },
    {
      "epoch": 1.107010863478553,
      "grad_norm": 1.250809669494629,
      "learning_rate": 9.8795e-05,
      "loss": 0.3467,
      "step": 19769
    },
    {
      "epoch": 1.107066860790682,
      "grad_norm": 1.286453127861023,
      "learning_rate": 9.88e-05,
      "loss": 0.5006,
      "step": 19770
    },
    {
      "epoch": 1.107122858102811,
      "grad_norm": 1.2133138179779053,
      "learning_rate": 9.8805e-05,
      "loss": 0.4423,
      "step": 19771
    },
    {
      "epoch": 1.10717885541494,
      "grad_norm": 1.489060401916504,
      "learning_rate": 9.881e-05,
      "loss": 0.3541,
      "step": 19772
    },
    {
      "epoch": 1.107234852727069,
      "grad_norm": 1.31699800491333,
      "learning_rate": 9.881500000000001e-05,
      "loss": 0.3656,
      "step": 19773
    },
    {
      "epoch": 1.107290850039198,
      "grad_norm": 1.2022706270217896,
      "learning_rate": 9.882e-05,
      "loss": 0.4431,
      "step": 19774
    },
    {
      "epoch": 1.107346847351327,
      "grad_norm": 1.1489864587783813,
      "learning_rate": 9.8825e-05,
      "loss": 0.3205,
      "step": 19775
    },
    {
      "epoch": 1.107402844663456,
      "grad_norm": 1.2494655847549438,
      "learning_rate": 9.883e-05,
      "loss": 0.3814,
      "step": 19776
    },
    {
      "epoch": 1.107458841975585,
      "grad_norm": 1.5846800804138184,
      "learning_rate": 9.8835e-05,
      "loss": 0.4948,
      "step": 19777
    },
    {
      "epoch": 1.107514839287714,
      "grad_norm": 1.374254584312439,
      "learning_rate": 9.884e-05,
      "loss": 0.5802,
      "step": 19778
    },
    {
      "epoch": 1.1075708365998431,
      "grad_norm": 1.5100247859954834,
      "learning_rate": 9.8845e-05,
      "loss": 0.4543,
      "step": 19779
    },
    {
      "epoch": 1.1076268339119721,
      "grad_norm": 1.4424753189086914,
      "learning_rate": 9.885e-05,
      "loss": 0.4236,
      "step": 19780
    },
    {
      "epoch": 1.1076828312241012,
      "grad_norm": 1.4685970544815063,
      "learning_rate": 9.885500000000001e-05,
      "loss": 0.4548,
      "step": 19781
    },
    {
      "epoch": 1.1077388285362302,
      "grad_norm": 1.7806181907653809,
      "learning_rate": 9.886000000000001e-05,
      "loss": 0.3814,
      "step": 19782
    },
    {
      "epoch": 1.1077948258483592,
      "grad_norm": 1.264974594116211,
      "learning_rate": 9.886500000000001e-05,
      "loss": 0.3284,
      "step": 19783
    },
    {
      "epoch": 1.1078508231604882,
      "grad_norm": 1.2328931093215942,
      "learning_rate": 9.887000000000001e-05,
      "loss": 0.5104,
      "step": 19784
    },
    {
      "epoch": 1.1079068204726172,
      "grad_norm": 1.28622567653656,
      "learning_rate": 9.8875e-05,
      "loss": 0.4541,
      "step": 19785
    },
    {
      "epoch": 1.1079628177847463,
      "grad_norm": 1.5155245065689087,
      "learning_rate": 9.888e-05,
      "loss": 0.4498,
      "step": 19786
    },
    {
      "epoch": 1.1080188150968753,
      "grad_norm": 1.269075632095337,
      "learning_rate": 9.8885e-05,
      "loss": 0.2918,
      "step": 19787
    },
    {
      "epoch": 1.1080748124090043,
      "grad_norm": 2.4623186588287354,
      "learning_rate": 9.889000000000001e-05,
      "loss": 0.4697,
      "step": 19788
    },
    {
      "epoch": 1.1081308097211333,
      "grad_norm": 1.508831262588501,
      "learning_rate": 9.889500000000001e-05,
      "loss": 0.629,
      "step": 19789
    },
    {
      "epoch": 1.1081868070332623,
      "grad_norm": 1.5116913318634033,
      "learning_rate": 9.89e-05,
      "loss": 0.5501,
      "step": 19790
    },
    {
      "epoch": 1.1082428043453914,
      "grad_norm": 1.5507465600967407,
      "learning_rate": 9.8905e-05,
      "loss": 0.443,
      "step": 19791
    },
    {
      "epoch": 1.1082988016575204,
      "grad_norm": 1.4017212390899658,
      "learning_rate": 9.891e-05,
      "loss": 0.4627,
      "step": 19792
    },
    {
      "epoch": 1.1083547989696494,
      "grad_norm": 1.7601009607315063,
      "learning_rate": 9.8915e-05,
      "loss": 0.323,
      "step": 19793
    },
    {
      "epoch": 1.1084107962817784,
      "grad_norm": 1.5531072616577148,
      "learning_rate": 9.892e-05,
      "loss": 0.4752,
      "step": 19794
    },
    {
      "epoch": 1.1084667935939074,
      "grad_norm": 1.5044505596160889,
      "learning_rate": 9.8925e-05,
      "loss": 0.5064,
      "step": 19795
    },
    {
      "epoch": 1.1085227909060364,
      "grad_norm": 1.2802830934524536,
      "learning_rate": 9.893e-05,
      "loss": 0.5113,
      "step": 19796
    },
    {
      "epoch": 1.1085787882181655,
      "grad_norm": 1.742136001586914,
      "learning_rate": 9.8935e-05,
      "loss": 0.7409,
      "step": 19797
    },
    {
      "epoch": 1.1086347855302945,
      "grad_norm": 1.4399957656860352,
      "learning_rate": 9.894e-05,
      "loss": 0.5201,
      "step": 19798
    },
    {
      "epoch": 1.1086907828424235,
      "grad_norm": 1.4919958114624023,
      "learning_rate": 9.894500000000001e-05,
      "loss": 0.4162,
      "step": 19799
    },
    {
      "epoch": 1.1087467801545525,
      "grad_norm": 1.313231348991394,
      "learning_rate": 9.895e-05,
      "loss": 0.377,
      "step": 19800
    },
    {
      "epoch": 1.1088027774666815,
      "grad_norm": 1.7701061964035034,
      "learning_rate": 9.8955e-05,
      "loss": 0.4284,
      "step": 19801
    },
    {
      "epoch": 1.1088587747788106,
      "grad_norm": 1.6026102304458618,
      "learning_rate": 9.896000000000001e-05,
      "loss": 0.4224,
      "step": 19802
    },
    {
      "epoch": 1.1089147720909396,
      "grad_norm": 1.1849709749221802,
      "learning_rate": 9.896500000000001e-05,
      "loss": 0.4069,
      "step": 19803
    },
    {
      "epoch": 1.1089707694030686,
      "grad_norm": 1.3921117782592773,
      "learning_rate": 9.897000000000001e-05,
      "loss": 0.4451,
      "step": 19804
    },
    {
      "epoch": 1.1090267667151976,
      "grad_norm": 1.1940969228744507,
      "learning_rate": 9.897500000000001e-05,
      "loss": 0.4032,
      "step": 19805
    },
    {
      "epoch": 1.1090827640273266,
      "grad_norm": 1.337113857269287,
      "learning_rate": 9.898e-05,
      "loss": 0.4439,
      "step": 19806
    },
    {
      "epoch": 1.1091387613394557,
      "grad_norm": 1.444793701171875,
      "learning_rate": 9.8985e-05,
      "loss": 0.494,
      "step": 19807
    },
    {
      "epoch": 1.1091947586515847,
      "grad_norm": 1.5827836990356445,
      "learning_rate": 9.899e-05,
      "loss": 0.4808,
      "step": 19808
    },
    {
      "epoch": 1.1092507559637137,
      "grad_norm": 1.4604209661483765,
      "learning_rate": 9.899500000000001e-05,
      "loss": 0.4782,
      "step": 19809
    },
    {
      "epoch": 1.1093067532758427,
      "grad_norm": 1.7103921175003052,
      "learning_rate": 9.900000000000001e-05,
      "loss": 0.7892,
      "step": 19810
    },
    {
      "epoch": 1.1093627505879717,
      "grad_norm": 1.7148162126541138,
      "learning_rate": 9.9005e-05,
      "loss": 0.3339,
      "step": 19811
    },
    {
      "epoch": 1.1094187479001008,
      "grad_norm": 1.3999277353286743,
      "learning_rate": 9.901e-05,
      "loss": 0.5298,
      "step": 19812
    },
    {
      "epoch": 1.1094747452122298,
      "grad_norm": 1.165982723236084,
      "learning_rate": 9.9015e-05,
      "loss": 0.4168,
      "step": 19813
    },
    {
      "epoch": 1.1095307425243588,
      "grad_norm": 1.240397334098816,
      "learning_rate": 9.902e-05,
      "loss": 0.3886,
      "step": 19814
    },
    {
      "epoch": 1.1095867398364878,
      "grad_norm": 1.2412090301513672,
      "learning_rate": 9.9025e-05,
      "loss": 0.32,
      "step": 19815
    },
    {
      "epoch": 1.1096427371486168,
      "grad_norm": 1.3912166357040405,
      "learning_rate": 9.903e-05,
      "loss": 0.4508,
      "step": 19816
    },
    {
      "epoch": 1.1096987344607459,
      "grad_norm": 1.2844611406326294,
      "learning_rate": 9.9035e-05,
      "loss": 0.4375,
      "step": 19817
    },
    {
      "epoch": 1.1097547317728749,
      "grad_norm": 2.117445707321167,
      "learning_rate": 9.904e-05,
      "loss": 0.601,
      "step": 19818
    },
    {
      "epoch": 1.1098107290850039,
      "grad_norm": 1.6978082656860352,
      "learning_rate": 9.904500000000001e-05,
      "loss": 0.512,
      "step": 19819
    },
    {
      "epoch": 1.109866726397133,
      "grad_norm": 1.3439422845840454,
      "learning_rate": 9.905000000000001e-05,
      "loss": 0.473,
      "step": 19820
    },
    {
      "epoch": 1.109922723709262,
      "grad_norm": 1.410443663597107,
      "learning_rate": 9.9055e-05,
      "loss": 0.401,
      "step": 19821
    },
    {
      "epoch": 1.109978721021391,
      "grad_norm": 1.4224430322647095,
      "learning_rate": 9.906e-05,
      "loss": 0.4945,
      "step": 19822
    },
    {
      "epoch": 1.11003471833352,
      "grad_norm": 1.2218562364578247,
      "learning_rate": 9.9065e-05,
      "loss": 0.4759,
      "step": 19823
    },
    {
      "epoch": 1.110090715645649,
      "grad_norm": 1.3283394575119019,
      "learning_rate": 9.907000000000001e-05,
      "loss": 0.4862,
      "step": 19824
    },
    {
      "epoch": 1.110146712957778,
      "grad_norm": 1.2634313106536865,
      "learning_rate": 9.907500000000001e-05,
      "loss": 0.4187,
      "step": 19825
    },
    {
      "epoch": 1.110202710269907,
      "grad_norm": 1.288685917854309,
      "learning_rate": 9.908000000000001e-05,
      "loss": 0.3703,
      "step": 19826
    },
    {
      "epoch": 1.110258707582036,
      "grad_norm": 1.2056894302368164,
      "learning_rate": 9.9085e-05,
      "loss": 0.4127,
      "step": 19827
    },
    {
      "epoch": 1.110314704894165,
      "grad_norm": 1.2373862266540527,
      "learning_rate": 9.909e-05,
      "loss": 0.4087,
      "step": 19828
    },
    {
      "epoch": 1.110370702206294,
      "grad_norm": 1.2632445096969604,
      "learning_rate": 9.9095e-05,
      "loss": 0.4093,
      "step": 19829
    },
    {
      "epoch": 1.110426699518423,
      "grad_norm": 1.4656373262405396,
      "learning_rate": 9.910000000000001e-05,
      "loss": 0.5283,
      "step": 19830
    },
    {
      "epoch": 1.1104826968305521,
      "grad_norm": 1.6199040412902832,
      "learning_rate": 9.910500000000001e-05,
      "loss": 0.4204,
      "step": 19831
    },
    {
      "epoch": 1.1105386941426811,
      "grad_norm": 1.6431338787078857,
      "learning_rate": 9.911e-05,
      "loss": 0.4061,
      "step": 19832
    },
    {
      "epoch": 1.1105946914548102,
      "grad_norm": 1.3754061460494995,
      "learning_rate": 9.9115e-05,
      "loss": 0.3841,
      "step": 19833
    },
    {
      "epoch": 1.1106506887669392,
      "grad_norm": 1.3536350727081299,
      "learning_rate": 9.912e-05,
      "loss": 0.4043,
      "step": 19834
    },
    {
      "epoch": 1.1107066860790682,
      "grad_norm": 1.239001750946045,
      "learning_rate": 9.9125e-05,
      "loss": 0.4993,
      "step": 19835
    },
    {
      "epoch": 1.1107626833911972,
      "grad_norm": 1.9410868883132935,
      "learning_rate": 9.913e-05,
      "loss": 0.5781,
      "step": 19836
    },
    {
      "epoch": 1.1108186807033262,
      "grad_norm": 1.3264325857162476,
      "learning_rate": 9.9135e-05,
      "loss": 0.3521,
      "step": 19837
    },
    {
      "epoch": 1.1108746780154553,
      "grad_norm": 1.7010470628738403,
      "learning_rate": 9.914e-05,
      "loss": 0.4823,
      "step": 19838
    },
    {
      "epoch": 1.1109306753275843,
      "grad_norm": 1.9255300760269165,
      "learning_rate": 9.914500000000001e-05,
      "loss": 0.5195,
      "step": 19839
    },
    {
      "epoch": 1.1109866726397133,
      "grad_norm": 1.4976234436035156,
      "learning_rate": 9.915000000000001e-05,
      "loss": 0.4206,
      "step": 19840
    },
    {
      "epoch": 1.1110426699518423,
      "grad_norm": 1.4784563779830933,
      "learning_rate": 9.915500000000001e-05,
      "loss": 0.3892,
      "step": 19841
    },
    {
      "epoch": 1.1110986672639713,
      "grad_norm": 1.702567458152771,
      "learning_rate": 9.916e-05,
      "loss": 0.467,
      "step": 19842
    },
    {
      "epoch": 1.1111546645761003,
      "grad_norm": 1.2280988693237305,
      "learning_rate": 9.9165e-05,
      "loss": 0.4062,
      "step": 19843
    },
    {
      "epoch": 1.1112106618882294,
      "grad_norm": 1.4703607559204102,
      "learning_rate": 9.917e-05,
      "loss": 0.5394,
      "step": 19844
    },
    {
      "epoch": 1.1112666592003584,
      "grad_norm": 1.282779335975647,
      "learning_rate": 9.917500000000001e-05,
      "loss": 0.3558,
      "step": 19845
    },
    {
      "epoch": 1.1113226565124874,
      "grad_norm": 1.4535640478134155,
      "learning_rate": 9.918000000000001e-05,
      "loss": 0.4454,
      "step": 19846
    },
    {
      "epoch": 1.1113786538246164,
      "grad_norm": 1.484246850013733,
      "learning_rate": 9.9185e-05,
      "loss": 0.4246,
      "step": 19847
    },
    {
      "epoch": 1.1114346511367454,
      "grad_norm": 1.4134314060211182,
      "learning_rate": 9.919e-05,
      "loss": 0.3681,
      "step": 19848
    },
    {
      "epoch": 1.1114906484488745,
      "grad_norm": 1.7610067129135132,
      "learning_rate": 9.9195e-05,
      "loss": 0.5749,
      "step": 19849
    },
    {
      "epoch": 1.1115466457610035,
      "grad_norm": 1.642107605934143,
      "learning_rate": 9.92e-05,
      "loss": 0.48,
      "step": 19850
    },
    {
      "epoch": 1.1116026430731325,
      "grad_norm": 1.2779661417007446,
      "learning_rate": 9.920500000000001e-05,
      "loss": 0.3769,
      "step": 19851
    },
    {
      "epoch": 1.1116586403852615,
      "grad_norm": 1.3027851581573486,
      "learning_rate": 9.921000000000001e-05,
      "loss": 0.4104,
      "step": 19852
    },
    {
      "epoch": 1.1117146376973905,
      "grad_norm": 1.3508186340332031,
      "learning_rate": 9.9215e-05,
      "loss": 0.4403,
      "step": 19853
    },
    {
      "epoch": 1.1117706350095196,
      "grad_norm": 1.39118230342865,
      "learning_rate": 9.922e-05,
      "loss": 0.4971,
      "step": 19854
    },
    {
      "epoch": 1.1118266323216486,
      "grad_norm": 1.4822081327438354,
      "learning_rate": 9.9225e-05,
      "loss": 0.4282,
      "step": 19855
    },
    {
      "epoch": 1.1118826296337776,
      "grad_norm": 1.4425870180130005,
      "learning_rate": 9.923e-05,
      "loss": 0.3867,
      "step": 19856
    },
    {
      "epoch": 1.1119386269459066,
      "grad_norm": 1.5058995485305786,
      "learning_rate": 9.9235e-05,
      "loss": 0.3943,
      "step": 19857
    },
    {
      "epoch": 1.1119946242580356,
      "grad_norm": 1.1833412647247314,
      "learning_rate": 9.924e-05,
      "loss": 0.384,
      "step": 19858
    },
    {
      "epoch": 1.1120506215701647,
      "grad_norm": 1.3850737810134888,
      "learning_rate": 9.924500000000002e-05,
      "loss": 0.4203,
      "step": 19859
    },
    {
      "epoch": 1.1121066188822937,
      "grad_norm": 1.5803650617599487,
      "learning_rate": 9.925000000000001e-05,
      "loss": 0.4912,
      "step": 19860
    },
    {
      "epoch": 1.1121626161944227,
      "grad_norm": 1.3808683156967163,
      "learning_rate": 9.925500000000001e-05,
      "loss": 0.4308,
      "step": 19861
    },
    {
      "epoch": 1.1122186135065517,
      "grad_norm": 1.4314091205596924,
      "learning_rate": 9.926000000000001e-05,
      "loss": 0.5991,
      "step": 19862
    },
    {
      "epoch": 1.1122746108186807,
      "grad_norm": 1.2249228954315186,
      "learning_rate": 9.9265e-05,
      "loss": 0.5417,
      "step": 19863
    },
    {
      "epoch": 1.1123306081308098,
      "grad_norm": 1.2787408828735352,
      "learning_rate": 9.927e-05,
      "loss": 0.3841,
      "step": 19864
    },
    {
      "epoch": 1.1123866054429388,
      "grad_norm": 1.3893275260925293,
      "learning_rate": 9.9275e-05,
      "loss": 0.467,
      "step": 19865
    },
    {
      "epoch": 1.1124426027550678,
      "grad_norm": 1.149018406867981,
      "learning_rate": 9.928000000000001e-05,
      "loss": 0.39,
      "step": 19866
    },
    {
      "epoch": 1.1124986000671968,
      "grad_norm": 1.4450526237487793,
      "learning_rate": 9.928500000000001e-05,
      "loss": 0.371,
      "step": 19867
    },
    {
      "epoch": 1.1125545973793258,
      "grad_norm": 1.4963886737823486,
      "learning_rate": 9.929e-05,
      "loss": 0.4029,
      "step": 19868
    },
    {
      "epoch": 1.1126105946914548,
      "grad_norm": 1.2389373779296875,
      "learning_rate": 9.9295e-05,
      "loss": 0.4354,
      "step": 19869
    },
    {
      "epoch": 1.1126665920035839,
      "grad_norm": 1.412307858467102,
      "learning_rate": 9.93e-05,
      "loss": 0.4996,
      "step": 19870
    },
    {
      "epoch": 1.1127225893157129,
      "grad_norm": 1.8511191606521606,
      "learning_rate": 9.9305e-05,
      "loss": 0.4936,
      "step": 19871
    },
    {
      "epoch": 1.112778586627842,
      "grad_norm": 1.1849406957626343,
      "learning_rate": 9.931000000000001e-05,
      "loss": 0.3702,
      "step": 19872
    },
    {
      "epoch": 1.112834583939971,
      "grad_norm": 1.2159291505813599,
      "learning_rate": 9.931500000000001e-05,
      "loss": 0.414,
      "step": 19873
    },
    {
      "epoch": 1.1128905812521,
      "grad_norm": 1.3139448165893555,
      "learning_rate": 9.932e-05,
      "loss": 0.3703,
      "step": 19874
    },
    {
      "epoch": 1.112946578564229,
      "grad_norm": 1.3350764513015747,
      "learning_rate": 9.9325e-05,
      "loss": 0.6695,
      "step": 19875
    },
    {
      "epoch": 1.113002575876358,
      "grad_norm": 1.1906070709228516,
      "learning_rate": 9.933e-05,
      "loss": 0.4074,
      "step": 19876
    },
    {
      "epoch": 1.113058573188487,
      "grad_norm": 1.3370988368988037,
      "learning_rate": 9.9335e-05,
      "loss": 0.553,
      "step": 19877
    },
    {
      "epoch": 1.113114570500616,
      "grad_norm": 1.351703405380249,
      "learning_rate": 9.934e-05,
      "loss": 0.4098,
      "step": 19878
    },
    {
      "epoch": 1.113170567812745,
      "grad_norm": 1.4156376123428345,
      "learning_rate": 9.9345e-05,
      "loss": 0.4852,
      "step": 19879
    },
    {
      "epoch": 1.113226565124874,
      "grad_norm": 1.2463830709457397,
      "learning_rate": 9.935000000000002e-05,
      "loss": 0.3755,
      "step": 19880
    },
    {
      "epoch": 1.113282562437003,
      "grad_norm": 1.4752771854400635,
      "learning_rate": 9.935500000000001e-05,
      "loss": 0.4217,
      "step": 19881
    },
    {
      "epoch": 1.113338559749132,
      "grad_norm": 1.4596432447433472,
      "learning_rate": 9.936000000000001e-05,
      "loss": 0.4529,
      "step": 19882
    },
    {
      "epoch": 1.1133945570612611,
      "grad_norm": 1.6576263904571533,
      "learning_rate": 9.936500000000001e-05,
      "loss": 0.5571,
      "step": 19883
    },
    {
      "epoch": 1.1134505543733901,
      "grad_norm": 1.398348331451416,
      "learning_rate": 9.937e-05,
      "loss": 0.4008,
      "step": 19884
    },
    {
      "epoch": 1.1135065516855192,
      "grad_norm": 1.4356671571731567,
      "learning_rate": 9.9375e-05,
      "loss": 0.4391,
      "step": 19885
    },
    {
      "epoch": 1.1135625489976482,
      "grad_norm": 1.416871190071106,
      "learning_rate": 9.938e-05,
      "loss": 0.5502,
      "step": 19886
    },
    {
      "epoch": 1.1136185463097772,
      "grad_norm": 1.1983762979507446,
      "learning_rate": 9.938500000000001e-05,
      "loss": 0.4317,
      "step": 19887
    },
    {
      "epoch": 1.1136745436219062,
      "grad_norm": 1.3804419040679932,
      "learning_rate": 9.939000000000001e-05,
      "loss": 0.4029,
      "step": 19888
    },
    {
      "epoch": 1.1137305409340352,
      "grad_norm": 1.318977952003479,
      "learning_rate": 9.9395e-05,
      "loss": 0.5044,
      "step": 19889
    },
    {
      "epoch": 1.1137865382461642,
      "grad_norm": 1.2622560262680054,
      "learning_rate": 9.94e-05,
      "loss": 0.3935,
      "step": 19890
    },
    {
      "epoch": 1.1138425355582933,
      "grad_norm": 1.322582483291626,
      "learning_rate": 9.9405e-05,
      "loss": 0.4329,
      "step": 19891
    },
    {
      "epoch": 1.1138985328704223,
      "grad_norm": 1.4327465295791626,
      "learning_rate": 9.941e-05,
      "loss": 0.4451,
      "step": 19892
    },
    {
      "epoch": 1.1139545301825513,
      "grad_norm": 1.2685604095458984,
      "learning_rate": 9.9415e-05,
      "loss": 0.3814,
      "step": 19893
    },
    {
      "epoch": 1.1140105274946803,
      "grad_norm": 1.1945087909698486,
      "learning_rate": 9.942000000000001e-05,
      "loss": 0.4192,
      "step": 19894
    },
    {
      "epoch": 1.1140665248068093,
      "grad_norm": 1.2283843755722046,
      "learning_rate": 9.9425e-05,
      "loss": 0.5873,
      "step": 19895
    },
    {
      "epoch": 1.1141225221189384,
      "grad_norm": 1.350475788116455,
      "learning_rate": 9.943e-05,
      "loss": 0.3933,
      "step": 19896
    },
    {
      "epoch": 1.1141785194310674,
      "grad_norm": 1.5221939086914062,
      "learning_rate": 9.9435e-05,
      "loss": 0.4215,
      "step": 19897
    },
    {
      "epoch": 1.1142345167431964,
      "grad_norm": 1.5168911218643188,
      "learning_rate": 9.944e-05,
      "loss": 0.4496,
      "step": 19898
    },
    {
      "epoch": 1.1142905140553254,
      "grad_norm": 1.6667182445526123,
      "learning_rate": 9.9445e-05,
      "loss": 0.4945,
      "step": 19899
    },
    {
      "epoch": 1.1143465113674544,
      "grad_norm": 1.4806978702545166,
      "learning_rate": 9.945e-05,
      "loss": 0.3982,
      "step": 19900
    },
    {
      "epoch": 1.1144025086795835,
      "grad_norm": 1.5881752967834473,
      "learning_rate": 9.945500000000002e-05,
      "loss": 0.4349,
      "step": 19901
    },
    {
      "epoch": 1.1144585059917125,
      "grad_norm": 1.4468464851379395,
      "learning_rate": 9.946000000000001e-05,
      "loss": 0.5028,
      "step": 19902
    },
    {
      "epoch": 1.1145145033038415,
      "grad_norm": 1.4290721416473389,
      "learning_rate": 9.946500000000001e-05,
      "loss": 0.5494,
      "step": 19903
    },
    {
      "epoch": 1.1145705006159705,
      "grad_norm": 1.4594035148620605,
      "learning_rate": 9.947000000000001e-05,
      "loss": 0.4004,
      "step": 19904
    },
    {
      "epoch": 1.1146264979280995,
      "grad_norm": 1.3635236024856567,
      "learning_rate": 9.9475e-05,
      "loss": 0.5512,
      "step": 19905
    },
    {
      "epoch": 1.1146824952402286,
      "grad_norm": 1.4220784902572632,
      "learning_rate": 9.948e-05,
      "loss": 0.3574,
      "step": 19906
    },
    {
      "epoch": 1.1147384925523576,
      "grad_norm": 1.4105898141860962,
      "learning_rate": 9.9485e-05,
      "loss": 0.5975,
      "step": 19907
    },
    {
      "epoch": 1.1147944898644866,
      "grad_norm": 1.3315365314483643,
      "learning_rate": 9.949000000000001e-05,
      "loss": 0.4772,
      "step": 19908
    },
    {
      "epoch": 1.1148504871766156,
      "grad_norm": 1.4663732051849365,
      "learning_rate": 9.949500000000001e-05,
      "loss": 0.4859,
      "step": 19909
    },
    {
      "epoch": 1.1149064844887446,
      "grad_norm": 1.4048585891723633,
      "learning_rate": 9.95e-05,
      "loss": 0.4054,
      "step": 19910
    },
    {
      "epoch": 1.1149624818008737,
      "grad_norm": 1.4173774719238281,
      "learning_rate": 9.9505e-05,
      "loss": 0.395,
      "step": 19911
    },
    {
      "epoch": 1.1150184791130027,
      "grad_norm": 1.297888994216919,
      "learning_rate": 9.951e-05,
      "loss": 0.3928,
      "step": 19912
    },
    {
      "epoch": 1.1150744764251317,
      "grad_norm": 1.144591212272644,
      "learning_rate": 9.9515e-05,
      "loss": 0.3601,
      "step": 19913
    },
    {
      "epoch": 1.1151304737372607,
      "grad_norm": 1.7306883335113525,
      "learning_rate": 9.952e-05,
      "loss": 0.4475,
      "step": 19914
    },
    {
      "epoch": 1.1151864710493897,
      "grad_norm": 1.7416670322418213,
      "learning_rate": 9.952500000000001e-05,
      "loss": 0.5564,
      "step": 19915
    },
    {
      "epoch": 1.1152424683615187,
      "grad_norm": 1.2478415966033936,
      "learning_rate": 9.953e-05,
      "loss": 0.5177,
      "step": 19916
    },
    {
      "epoch": 1.1152984656736478,
      "grad_norm": 1.2330882549285889,
      "learning_rate": 9.9535e-05,
      "loss": 0.4486,
      "step": 19917
    },
    {
      "epoch": 1.1153544629857768,
      "grad_norm": 1.2023470401763916,
      "learning_rate": 9.954e-05,
      "loss": 0.5095,
      "step": 19918
    },
    {
      "epoch": 1.1154104602979058,
      "grad_norm": 1.6836299896240234,
      "learning_rate": 9.9545e-05,
      "loss": 0.5091,
      "step": 19919
    },
    {
      "epoch": 1.1154664576100348,
      "grad_norm": 1.297433614730835,
      "learning_rate": 9.955000000000001e-05,
      "loss": 0.3423,
      "step": 19920
    },
    {
      "epoch": 1.1155224549221638,
      "grad_norm": 1.5119662284851074,
      "learning_rate": 9.9555e-05,
      "loss": 0.4965,
      "step": 19921
    },
    {
      "epoch": 1.1155784522342929,
      "grad_norm": 1.4331485033035278,
      "learning_rate": 9.956e-05,
      "loss": 0.4122,
      "step": 19922
    },
    {
      "epoch": 1.1156344495464219,
      "grad_norm": 1.4502153396606445,
      "learning_rate": 9.956500000000001e-05,
      "loss": 0.4129,
      "step": 19923
    },
    {
      "epoch": 1.115690446858551,
      "grad_norm": 1.3752232789993286,
      "learning_rate": 9.957000000000001e-05,
      "loss": 0.408,
      "step": 19924
    },
    {
      "epoch": 1.11574644417068,
      "grad_norm": 1.5951974391937256,
      "learning_rate": 9.957500000000001e-05,
      "loss": 0.5723,
      "step": 19925
    },
    {
      "epoch": 1.115802441482809,
      "grad_norm": 1.2176891565322876,
      "learning_rate": 9.958e-05,
      "loss": 0.4047,
      "step": 19926
    },
    {
      "epoch": 1.115858438794938,
      "grad_norm": 1.3480055332183838,
      "learning_rate": 9.9585e-05,
      "loss": 0.4002,
      "step": 19927
    },
    {
      "epoch": 1.115914436107067,
      "grad_norm": 1.6599169969558716,
      "learning_rate": 9.959e-05,
      "loss": 0.6165,
      "step": 19928
    },
    {
      "epoch": 1.115970433419196,
      "grad_norm": 1.3681401014328003,
      "learning_rate": 9.959500000000001e-05,
      "loss": 0.7238,
      "step": 19929
    },
    {
      "epoch": 1.116026430731325,
      "grad_norm": 1.1669130325317383,
      "learning_rate": 9.960000000000001e-05,
      "loss": 0.3322,
      "step": 19930
    },
    {
      "epoch": 1.1160824280434538,
      "grad_norm": 1.4666856527328491,
      "learning_rate": 9.9605e-05,
      "loss": 0.5052,
      "step": 19931
    },
    {
      "epoch": 1.1161384253555828,
      "grad_norm": 1.4829007387161255,
      "learning_rate": 9.961e-05,
      "loss": 0.3775,
      "step": 19932
    },
    {
      "epoch": 1.1161944226677118,
      "grad_norm": 1.5757139921188354,
      "learning_rate": 9.9615e-05,
      "loss": 0.3391,
      "step": 19933
    },
    {
      "epoch": 1.1162504199798409,
      "grad_norm": 1.6869962215423584,
      "learning_rate": 9.962e-05,
      "loss": 0.6753,
      "step": 19934
    },
    {
      "epoch": 1.1163064172919699,
      "grad_norm": 1.349286675453186,
      "learning_rate": 9.9625e-05,
      "loss": 0.4492,
      "step": 19935
    },
    {
      "epoch": 1.116362414604099,
      "grad_norm": 1.3248045444488525,
      "learning_rate": 9.963e-05,
      "loss": 0.3965,
      "step": 19936
    },
    {
      "epoch": 1.116418411916228,
      "grad_norm": 1.3681490421295166,
      "learning_rate": 9.9635e-05,
      "loss": 0.4602,
      "step": 19937
    },
    {
      "epoch": 1.116474409228357,
      "grad_norm": 1.4884835481643677,
      "learning_rate": 9.964e-05,
      "loss": 0.4888,
      "step": 19938
    },
    {
      "epoch": 1.116530406540486,
      "grad_norm": 1.4044023752212524,
      "learning_rate": 9.9645e-05,
      "loss": 0.6349,
      "step": 19939
    },
    {
      "epoch": 1.116586403852615,
      "grad_norm": 1.4194635152816772,
      "learning_rate": 9.965000000000001e-05,
      "loss": 0.5979,
      "step": 19940
    },
    {
      "epoch": 1.116642401164744,
      "grad_norm": 1.5223853588104248,
      "learning_rate": 9.965500000000001e-05,
      "loss": 0.4812,
      "step": 19941
    },
    {
      "epoch": 1.116698398476873,
      "grad_norm": 1.4153226613998413,
      "learning_rate": 9.966e-05,
      "loss": 0.4602,
      "step": 19942
    },
    {
      "epoch": 1.116754395789002,
      "grad_norm": 1.2923203706741333,
      "learning_rate": 9.9665e-05,
      "loss": 0.4366,
      "step": 19943
    },
    {
      "epoch": 1.116810393101131,
      "grad_norm": 1.382320523262024,
      "learning_rate": 9.967000000000001e-05,
      "loss": 0.4938,
      "step": 19944
    },
    {
      "epoch": 1.11686639041326,
      "grad_norm": 1.4697731733322144,
      "learning_rate": 9.967500000000001e-05,
      "loss": 0.4776,
      "step": 19945
    },
    {
      "epoch": 1.116922387725389,
      "grad_norm": 1.3057854175567627,
      "learning_rate": 9.968000000000001e-05,
      "loss": 0.489,
      "step": 19946
    },
    {
      "epoch": 1.1169783850375181,
      "grad_norm": 1.5513603687286377,
      "learning_rate": 9.9685e-05,
      "loss": 0.4308,
      "step": 19947
    },
    {
      "epoch": 1.1170343823496471,
      "grad_norm": 1.413629174232483,
      "learning_rate": 9.969e-05,
      "loss": 0.413,
      "step": 19948
    },
    {
      "epoch": 1.1170903796617762,
      "grad_norm": 1.5906438827514648,
      "learning_rate": 9.9695e-05,
      "loss": 0.3332,
      "step": 19949
    },
    {
      "epoch": 1.1171463769739052,
      "grad_norm": 1.3415582180023193,
      "learning_rate": 9.970000000000001e-05,
      "loss": 0.5089,
      "step": 19950
    },
    {
      "epoch": 1.1172023742860342,
      "grad_norm": 1.3672477006912231,
      "learning_rate": 9.970500000000001e-05,
      "loss": 0.4277,
      "step": 19951
    },
    {
      "epoch": 1.1172583715981632,
      "grad_norm": 1.3722107410430908,
      "learning_rate": 9.971e-05,
      "loss": 0.4536,
      "step": 19952
    },
    {
      "epoch": 1.1173143689102922,
      "grad_norm": 1.2795250415802002,
      "learning_rate": 9.9715e-05,
      "loss": 0.4078,
      "step": 19953
    },
    {
      "epoch": 1.1173703662224213,
      "grad_norm": 1.5060938596725464,
      "learning_rate": 9.972e-05,
      "loss": 0.4411,
      "step": 19954
    },
    {
      "epoch": 1.1174263635345503,
      "grad_norm": 1.2826478481292725,
      "learning_rate": 9.9725e-05,
      "loss": 0.4602,
      "step": 19955
    },
    {
      "epoch": 1.1174823608466793,
      "grad_norm": 1.434885859489441,
      "learning_rate": 9.973e-05,
      "loss": 0.453,
      "step": 19956
    },
    {
      "epoch": 1.1175383581588083,
      "grad_norm": 1.210551381111145,
      "learning_rate": 9.9735e-05,
      "loss": 0.4582,
      "step": 19957
    },
    {
      "epoch": 1.1175943554709373,
      "grad_norm": 1.3893661499023438,
      "learning_rate": 9.974e-05,
      "loss": 0.4822,
      "step": 19958
    },
    {
      "epoch": 1.1176503527830663,
      "grad_norm": 1.260366678237915,
      "learning_rate": 9.9745e-05,
      "loss": 0.3964,
      "step": 19959
    },
    {
      "epoch": 1.1177063500951954,
      "grad_norm": 1.0513705015182495,
      "learning_rate": 9.975000000000001e-05,
      "loss": 0.3309,
      "step": 19960
    },
    {
      "epoch": 1.1177623474073244,
      "grad_norm": 4.4490790367126465,
      "learning_rate": 9.975500000000001e-05,
      "loss": 0.5086,
      "step": 19961
    },
    {
      "epoch": 1.1178183447194534,
      "grad_norm": 1.1814807653427124,
      "learning_rate": 9.976000000000001e-05,
      "loss": 0.4661,
      "step": 19962
    },
    {
      "epoch": 1.1178743420315824,
      "grad_norm": 1.4177594184875488,
      "learning_rate": 9.9765e-05,
      "loss": 0.454,
      "step": 19963
    },
    {
      "epoch": 1.1179303393437114,
      "grad_norm": 1.2462059259414673,
      "learning_rate": 9.977e-05,
      "loss": 0.5212,
      "step": 19964
    },
    {
      "epoch": 1.1179863366558405,
      "grad_norm": 1.2088459730148315,
      "learning_rate": 9.977500000000001e-05,
      "loss": 0.3805,
      "step": 19965
    },
    {
      "epoch": 1.1180423339679695,
      "grad_norm": 1.2562636137008667,
      "learning_rate": 9.978000000000001e-05,
      "loss": 0.4543,
      "step": 19966
    },
    {
      "epoch": 1.1180983312800985,
      "grad_norm": 1.3372293710708618,
      "learning_rate": 9.978500000000001e-05,
      "loss": 0.639,
      "step": 19967
    },
    {
      "epoch": 1.1181543285922275,
      "grad_norm": 1.5341663360595703,
      "learning_rate": 9.979e-05,
      "loss": 0.4539,
      "step": 19968
    },
    {
      "epoch": 1.1182103259043565,
      "grad_norm": 1.4747451543807983,
      "learning_rate": 9.9795e-05,
      "loss": 0.4447,
      "step": 19969
    },
    {
      "epoch": 1.1182663232164856,
      "grad_norm": 1.4931836128234863,
      "learning_rate": 9.98e-05,
      "loss": 0.4791,
      "step": 19970
    },
    {
      "epoch": 1.1183223205286146,
      "grad_norm": 1.6403145790100098,
      "learning_rate": 9.9805e-05,
      "loss": 0.4906,
      "step": 19971
    },
    {
      "epoch": 1.1183783178407436,
      "grad_norm": 1.296626091003418,
      "learning_rate": 9.981000000000001e-05,
      "loss": 0.4241,
      "step": 19972
    },
    {
      "epoch": 1.1184343151528726,
      "grad_norm": 1.3703420162200928,
      "learning_rate": 9.9815e-05,
      "loss": 0.3721,
      "step": 19973
    },
    {
      "epoch": 1.1184903124650016,
      "grad_norm": 1.6404469013214111,
      "learning_rate": 9.982e-05,
      "loss": 0.4843,
      "step": 19974
    },
    {
      "epoch": 1.1185463097771307,
      "grad_norm": 1.3929030895233154,
      "learning_rate": 9.9825e-05,
      "loss": 0.518,
      "step": 19975
    },
    {
      "epoch": 1.1186023070892597,
      "grad_norm": 1.3429993391036987,
      "learning_rate": 9.983e-05,
      "loss": 0.3587,
      "step": 19976
    },
    {
      "epoch": 1.1186583044013887,
      "grad_norm": 1.367019534111023,
      "learning_rate": 9.9835e-05,
      "loss": 0.6241,
      "step": 19977
    },
    {
      "epoch": 1.1187143017135177,
      "grad_norm": 1.4297236204147339,
      "learning_rate": 9.984e-05,
      "loss": 0.557,
      "step": 19978
    },
    {
      "epoch": 1.1187702990256467,
      "grad_norm": 1.4048442840576172,
      "learning_rate": 9.9845e-05,
      "loss": 0.4619,
      "step": 19979
    },
    {
      "epoch": 1.1188262963377757,
      "grad_norm": 1.6790038347244263,
      "learning_rate": 9.985000000000001e-05,
      "loss": 0.5104,
      "step": 19980
    },
    {
      "epoch": 1.1188822936499048,
      "grad_norm": 1.2666529417037964,
      "learning_rate": 9.985500000000001e-05,
      "loss": 0.4194,
      "step": 19981
    },
    {
      "epoch": 1.1189382909620338,
      "grad_norm": 1.4653370380401611,
      "learning_rate": 9.986000000000001e-05,
      "loss": 0.3837,
      "step": 19982
    },
    {
      "epoch": 1.1189942882741628,
      "grad_norm": 1.3848682641983032,
      "learning_rate": 9.986500000000001e-05,
      "loss": 0.5892,
      "step": 19983
    },
    {
      "epoch": 1.1190502855862918,
      "grad_norm": 1.2159497737884521,
      "learning_rate": 9.987e-05,
      "loss": 0.4306,
      "step": 19984
    },
    {
      "epoch": 1.1191062828984208,
      "grad_norm": 1.0413435697555542,
      "learning_rate": 9.9875e-05,
      "loss": 0.3403,
      "step": 19985
    },
    {
      "epoch": 1.1191622802105499,
      "grad_norm": 1.1079907417297363,
      "learning_rate": 9.988000000000001e-05,
      "loss": 0.3271,
      "step": 19986
    },
    {
      "epoch": 1.1192182775226789,
      "grad_norm": 1.1294519901275635,
      "learning_rate": 9.988500000000001e-05,
      "loss": 0.3808,
      "step": 19987
    },
    {
      "epoch": 1.119274274834808,
      "grad_norm": 1.372883677482605,
      "learning_rate": 9.989000000000001e-05,
      "loss": 0.4385,
      "step": 19988
    },
    {
      "epoch": 1.119330272146937,
      "grad_norm": 1.23002028465271,
      "learning_rate": 9.9895e-05,
      "loss": 0.4712,
      "step": 19989
    },
    {
      "epoch": 1.119386269459066,
      "grad_norm": 1.3323321342468262,
      "learning_rate": 9.99e-05,
      "loss": 0.3743,
      "step": 19990
    },
    {
      "epoch": 1.119442266771195,
      "grad_norm": 1.47197425365448,
      "learning_rate": 9.9905e-05,
      "loss": 0.4424,
      "step": 19991
    },
    {
      "epoch": 1.119498264083324,
      "grad_norm": 1.7073791027069092,
      "learning_rate": 9.991e-05,
      "loss": 0.4088,
      "step": 19992
    },
    {
      "epoch": 1.119554261395453,
      "grad_norm": 1.2094861268997192,
      "learning_rate": 9.991500000000001e-05,
      "loss": 0.3556,
      "step": 19993
    },
    {
      "epoch": 1.119610258707582,
      "grad_norm": 1.3336609601974487,
      "learning_rate": 9.992e-05,
      "loss": 0.4196,
      "step": 19994
    },
    {
      "epoch": 1.119666256019711,
      "grad_norm": 1.3385021686553955,
      "learning_rate": 9.9925e-05,
      "loss": 0.4877,
      "step": 19995
    },
    {
      "epoch": 1.11972225333184,
      "grad_norm": 1.3331232070922852,
      "learning_rate": 9.993e-05,
      "loss": 0.429,
      "step": 19996
    },
    {
      "epoch": 1.119778250643969,
      "grad_norm": 1.5061585903167725,
      "learning_rate": 9.9935e-05,
      "loss": 0.4839,
      "step": 19997
    },
    {
      "epoch": 1.119834247956098,
      "grad_norm": 1.2313766479492188,
      "learning_rate": 9.994e-05,
      "loss": 0.4743,
      "step": 19998
    },
    {
      "epoch": 1.1198902452682271,
      "grad_norm": 1.3281816244125366,
      "learning_rate": 9.9945e-05,
      "loss": 0.4664,
      "step": 19999
    },
    {
      "epoch": 1.1199462425803561,
      "grad_norm": 1.4939744472503662,
      "learning_rate": 9.995e-05,
      "loss": 0.6259,
      "step": 20000
    },
    {
      "epoch": 1.1200022398924852,
      "grad_norm": 1.670356035232544,
      "learning_rate": 9.995500000000001e-05,
      "loss": 0.3992,
      "step": 20001
    },
    {
      "epoch": 1.1200582372046142,
      "grad_norm": 1.2830830812454224,
      "learning_rate": 9.996000000000001e-05,
      "loss": 0.4945,
      "step": 20002
    },
    {
      "epoch": 1.1201142345167432,
      "grad_norm": 1.5486961603164673,
      "learning_rate": 9.996500000000001e-05,
      "loss": 0.3299,
      "step": 20003
    },
    {
      "epoch": 1.1201702318288722,
      "grad_norm": 1.5304434299468994,
      "learning_rate": 9.997e-05,
      "loss": 0.5126,
      "step": 20004
    },
    {
      "epoch": 1.1202262291410012,
      "grad_norm": 1.4435508251190186,
      "learning_rate": 9.9975e-05,
      "loss": 0.4213,
      "step": 20005
    },
    {
      "epoch": 1.1202822264531302,
      "grad_norm": 1.157757043838501,
      "learning_rate": 9.998e-05,
      "loss": 0.34,
      "step": 20006
    },
    {
      "epoch": 1.1203382237652593,
      "grad_norm": 1.3835796117782593,
      "learning_rate": 9.998500000000001e-05,
      "loss": 0.5336,
      "step": 20007
    },
    {
      "epoch": 1.1203942210773883,
      "grad_norm": 1.747640609741211,
      "learning_rate": 9.999000000000001e-05,
      "loss": 0.4485,
      "step": 20008
    },
    {
      "epoch": 1.1204502183895173,
      "grad_norm": 1.5142403841018677,
      "learning_rate": 9.999500000000001e-05,
      "loss": 0.6584,
      "step": 20009
    },
    {
      "epoch": 1.1205062157016463,
      "grad_norm": 1.684420108795166,
      "learning_rate": 0.0001,
      "loss": 0.5178,
      "step": 20010
    },
    {
      "epoch": 1.1205622130137753,
      "grad_norm": 1.2592800855636597,
      "learning_rate": 9.999973684210526e-05,
      "loss": 0.4609,
      "step": 20011
    },
    {
      "epoch": 1.1206182103259044,
      "grad_norm": 1.562971830368042,
      "learning_rate": 9.999947368421054e-05,
      "loss": 0.5383,
      "step": 20012
    },
    {
      "epoch": 1.1206742076380334,
      "grad_norm": 1.318636417388916,
      "learning_rate": 9.999921052631578e-05,
      "loss": 0.3895,
      "step": 20013
    },
    {
      "epoch": 1.1207302049501624,
      "grad_norm": 1.1930203437805176,
      "learning_rate": 9.999894736842106e-05,
      "loss": 0.3013,
      "step": 20014
    },
    {
      "epoch": 1.1207862022622914,
      "grad_norm": 1.4527639150619507,
      "learning_rate": 9.999868421052632e-05,
      "loss": 0.3946,
      "step": 20015
    },
    {
      "epoch": 1.1208421995744204,
      "grad_norm": 1.5577555894851685,
      "learning_rate": 9.999842105263159e-05,
      "loss": 0.5031,
      "step": 20016
    },
    {
      "epoch": 1.1208981968865495,
      "grad_norm": 1.3654688596725464,
      "learning_rate": 9.999815789473685e-05,
      "loss": 0.4687,
      "step": 20017
    },
    {
      "epoch": 1.1209541941986785,
      "grad_norm": 1.3246850967407227,
      "learning_rate": 9.999789473684211e-05,
      "loss": 0.4485,
      "step": 20018
    },
    {
      "epoch": 1.1210101915108075,
      "grad_norm": 1.4335442781448364,
      "learning_rate": 9.999763157894737e-05,
      "loss": 0.4369,
      "step": 20019
    },
    {
      "epoch": 1.1210661888229365,
      "grad_norm": 1.2487568855285645,
      "learning_rate": 9.999736842105264e-05,
      "loss": 0.4527,
      "step": 20020
    },
    {
      "epoch": 1.1211221861350655,
      "grad_norm": 1.457713007926941,
      "learning_rate": 9.99971052631579e-05,
      "loss": 0.4996,
      "step": 20021
    },
    {
      "epoch": 1.1211781834471946,
      "grad_norm": 1.7885032892227173,
      "learning_rate": 9.999684210526316e-05,
      "loss": 0.456,
      "step": 20022
    },
    {
      "epoch": 1.1212341807593236,
      "grad_norm": 1.3303115367889404,
      "learning_rate": 9.999657894736842e-05,
      "loss": 0.3477,
      "step": 20023
    },
    {
      "epoch": 1.1212901780714526,
      "grad_norm": 1.3550374507904053,
      "learning_rate": 9.99963157894737e-05,
      "loss": 0.3813,
      "step": 20024
    },
    {
      "epoch": 1.1213461753835816,
      "grad_norm": 1.4085867404937744,
      "learning_rate": 9.999605263157895e-05,
      "loss": 0.4964,
      "step": 20025
    },
    {
      "epoch": 1.1214021726957106,
      "grad_norm": 1.7398312091827393,
      "learning_rate": 9.999578947368421e-05,
      "loss": 0.4673,
      "step": 20026
    },
    {
      "epoch": 1.1214581700078396,
      "grad_norm": 1.3218722343444824,
      "learning_rate": 9.999552631578947e-05,
      "loss": 0.3901,
      "step": 20027
    },
    {
      "epoch": 1.1215141673199687,
      "grad_norm": 1.3908706903457642,
      "learning_rate": 9.999526315789473e-05,
      "loss": 0.5142,
      "step": 20028
    },
    {
      "epoch": 1.1215701646320977,
      "grad_norm": 1.4564788341522217,
      "learning_rate": 9.999500000000001e-05,
      "loss": 0.4245,
      "step": 20029
    },
    {
      "epoch": 1.1216261619442267,
      "grad_norm": 1.5256010293960571,
      "learning_rate": 9.999473684210527e-05,
      "loss": 0.3972,
      "step": 20030
    },
    {
      "epoch": 1.1216821592563557,
      "grad_norm": 1.153856873512268,
      "learning_rate": 9.999447368421053e-05,
      "loss": 0.3373,
      "step": 20031
    },
    {
      "epoch": 1.1217381565684847,
      "grad_norm": 1.2020797729492188,
      "learning_rate": 9.999421052631579e-05,
      "loss": 0.3982,
      "step": 20032
    },
    {
      "epoch": 1.1217941538806138,
      "grad_norm": 1.2495990991592407,
      "learning_rate": 9.999394736842106e-05,
      "loss": 0.2986,
      "step": 20033
    },
    {
      "epoch": 1.1218501511927428,
      "grad_norm": 1.400131106376648,
      "learning_rate": 9.999368421052632e-05,
      "loss": 0.3613,
      "step": 20034
    },
    {
      "epoch": 1.1219061485048718,
      "grad_norm": 1.214959979057312,
      "learning_rate": 9.999342105263159e-05,
      "loss": 0.44,
      "step": 20035
    },
    {
      "epoch": 1.1219621458170008,
      "grad_norm": 1.2081224918365479,
      "learning_rate": 9.999315789473684e-05,
      "loss": 0.3817,
      "step": 20036
    },
    {
      "epoch": 1.1220181431291298,
      "grad_norm": 1.3629376888275146,
      "learning_rate": 9.999289473684211e-05,
      "loss": 0.5077,
      "step": 20037
    },
    {
      "epoch": 1.1220741404412589,
      "grad_norm": 1.4008897542953491,
      "learning_rate": 9.999263157894737e-05,
      "loss": 0.47,
      "step": 20038
    },
    {
      "epoch": 1.1221301377533879,
      "grad_norm": 1.4684243202209473,
      "learning_rate": 9.999236842105265e-05,
      "loss": 0.3975,
      "step": 20039
    },
    {
      "epoch": 1.122186135065517,
      "grad_norm": 1.8027886152267456,
      "learning_rate": 9.999210526315789e-05,
      "loss": 0.4434,
      "step": 20040
    },
    {
      "epoch": 1.122242132377646,
      "grad_norm": 1.8867990970611572,
      "learning_rate": 9.999184210526316e-05,
      "loss": 0.362,
      "step": 20041
    },
    {
      "epoch": 1.122298129689775,
      "grad_norm": 1.4636904001235962,
      "learning_rate": 9.999157894736842e-05,
      "loss": 0.4318,
      "step": 20042
    },
    {
      "epoch": 1.122354127001904,
      "grad_norm": 1.3516290187835693,
      "learning_rate": 9.99913157894737e-05,
      "loss": 0.5048,
      "step": 20043
    },
    {
      "epoch": 1.122410124314033,
      "grad_norm": 1.6282988786697388,
      "learning_rate": 9.999105263157896e-05,
      "loss": 0.643,
      "step": 20044
    },
    {
      "epoch": 1.122466121626162,
      "grad_norm": 1.0735915899276733,
      "learning_rate": 9.99907894736842e-05,
      "loss": 0.3689,
      "step": 20045
    },
    {
      "epoch": 1.122522118938291,
      "grad_norm": 1.4593976736068726,
      "learning_rate": 9.999052631578948e-05,
      "loss": 0.3651,
      "step": 20046
    },
    {
      "epoch": 1.12257811625042,
      "grad_norm": 1.146026849746704,
      "learning_rate": 9.999026315789474e-05,
      "loss": 0.3888,
      "step": 20047
    },
    {
      "epoch": 1.122634113562549,
      "grad_norm": 1.2889912128448486,
      "learning_rate": 9.999000000000001e-05,
      "loss": 0.3881,
      "step": 20048
    },
    {
      "epoch": 1.122690110874678,
      "grad_norm": 1.4708292484283447,
      "learning_rate": 9.998973684210527e-05,
      "loss": 0.4643,
      "step": 20049
    },
    {
      "epoch": 1.122746108186807,
      "grad_norm": 1.4523096084594727,
      "learning_rate": 9.998947368421053e-05,
      "loss": 0.5212,
      "step": 20050
    },
    {
      "epoch": 1.122802105498936,
      "grad_norm": 1.3572546243667603,
      "learning_rate": 9.998921052631579e-05,
      "loss": 0.485,
      "step": 20051
    },
    {
      "epoch": 1.1228581028110651,
      "grad_norm": 1.9971601963043213,
      "learning_rate": 9.998894736842106e-05,
      "loss": 0.3975,
      "step": 20052
    },
    {
      "epoch": 1.1229141001231941,
      "grad_norm": 1.4978028535842896,
      "learning_rate": 9.998868421052632e-05,
      "loss": 0.4827,
      "step": 20053
    },
    {
      "epoch": 1.1229700974353232,
      "grad_norm": 1.2745866775512695,
      "learning_rate": 9.998842105263158e-05,
      "loss": 0.3795,
      "step": 20054
    },
    {
      "epoch": 1.1230260947474522,
      "grad_norm": 1.484288215637207,
      "learning_rate": 9.998815789473684e-05,
      "loss": 0.3018,
      "step": 20055
    },
    {
      "epoch": 1.1230820920595812,
      "grad_norm": 1.2436482906341553,
      "learning_rate": 9.998789473684211e-05,
      "loss": 0.3835,
      "step": 20056
    },
    {
      "epoch": 1.1231380893717102,
      "grad_norm": 1.1885489225387573,
      "learning_rate": 9.998763157894737e-05,
      "loss": 0.3402,
      "step": 20057
    },
    {
      "epoch": 1.1231940866838392,
      "grad_norm": 1.3961933851242065,
      "learning_rate": 9.998736842105263e-05,
      "loss": 0.5356,
      "step": 20058
    },
    {
      "epoch": 1.1232500839959683,
      "grad_norm": 1.2171860933303833,
      "learning_rate": 9.99871052631579e-05,
      "loss": 0.4313,
      "step": 20059
    },
    {
      "epoch": 1.1233060813080973,
      "grad_norm": 1.3281238079071045,
      "learning_rate": 9.998684210526317e-05,
      "loss": 0.5049,
      "step": 20060
    },
    {
      "epoch": 1.1233620786202263,
      "grad_norm": 1.1421741247177124,
      "learning_rate": 9.998657894736843e-05,
      "loss": 0.3011,
      "step": 20061
    },
    {
      "epoch": 1.1234180759323553,
      "grad_norm": 1.3167433738708496,
      "learning_rate": 9.998631578947369e-05,
      "loss": 0.6157,
      "step": 20062
    },
    {
      "epoch": 1.1234740732444843,
      "grad_norm": 1.498759150505066,
      "learning_rate": 9.998605263157895e-05,
      "loss": 0.3853,
      "step": 20063
    },
    {
      "epoch": 1.1235300705566134,
      "grad_norm": 1.7297914028167725,
      "learning_rate": 9.99857894736842e-05,
      "loss": 0.4637,
      "step": 20064
    },
    {
      "epoch": 1.1235860678687424,
      "grad_norm": 1.525895357131958,
      "learning_rate": 9.998552631578948e-05,
      "loss": 0.5171,
      "step": 20065
    },
    {
      "epoch": 1.1236420651808714,
      "grad_norm": 1.3542238473892212,
      "learning_rate": 9.998526315789474e-05,
      "loss": 0.3932,
      "step": 20066
    },
    {
      "epoch": 1.1236980624930004,
      "grad_norm": 1.4197075366973877,
      "learning_rate": 9.998500000000001e-05,
      "loss": 0.406,
      "step": 20067
    },
    {
      "epoch": 1.1237540598051294,
      "grad_norm": 1.2412537336349487,
      "learning_rate": 9.998473684210526e-05,
      "loss": 0.3571,
      "step": 20068
    },
    {
      "epoch": 1.1238100571172585,
      "grad_norm": 1.256371021270752,
      "learning_rate": 9.998447368421053e-05,
      "loss": 0.3646,
      "step": 20069
    },
    {
      "epoch": 1.1238660544293875,
      "grad_norm": 1.223979115486145,
      "learning_rate": 9.998421052631579e-05,
      "loss": 0.4346,
      "step": 20070
    },
    {
      "epoch": 1.1239220517415165,
      "grad_norm": 1.237862229347229,
      "learning_rate": 9.998394736842107e-05,
      "loss": 0.4622,
      "step": 20071
    },
    {
      "epoch": 1.1239780490536455,
      "grad_norm": 1.4763144254684448,
      "learning_rate": 9.998368421052632e-05,
      "loss": 0.4447,
      "step": 20072
    },
    {
      "epoch": 1.1240340463657745,
      "grad_norm": 1.2453192472457886,
      "learning_rate": 9.998342105263158e-05,
      "loss": 0.4321,
      "step": 20073
    },
    {
      "epoch": 1.1240900436779035,
      "grad_norm": 1.6521111726760864,
      "learning_rate": 9.998315789473684e-05,
      "loss": 0.4248,
      "step": 20074
    },
    {
      "epoch": 1.1241460409900326,
      "grad_norm": 1.4103846549987793,
      "learning_rate": 9.998289473684212e-05,
      "loss": 0.5108,
      "step": 20075
    },
    {
      "epoch": 1.1242020383021616,
      "grad_norm": 1.3992470502853394,
      "learning_rate": 9.998263157894738e-05,
      "loss": 0.432,
      "step": 20076
    },
    {
      "epoch": 1.1242580356142906,
      "grad_norm": 1.2738618850708008,
      "learning_rate": 9.998236842105264e-05,
      "loss": 0.4344,
      "step": 20077
    },
    {
      "epoch": 1.1243140329264196,
      "grad_norm": 1.4086074829101562,
      "learning_rate": 9.99821052631579e-05,
      "loss": 0.4177,
      "step": 20078
    },
    {
      "epoch": 1.1243700302385486,
      "grad_norm": 1.2592560052871704,
      "learning_rate": 9.998184210526317e-05,
      "loss": 0.3631,
      "step": 20079
    },
    {
      "epoch": 1.1244260275506777,
      "grad_norm": 1.4146602153778076,
      "learning_rate": 9.998157894736843e-05,
      "loss": 0.4133,
      "step": 20080
    },
    {
      "epoch": 1.1244820248628067,
      "grad_norm": 1.422480821609497,
      "learning_rate": 9.998131578947369e-05,
      "loss": 0.4386,
      "step": 20081
    },
    {
      "epoch": 1.1245380221749357,
      "grad_norm": 1.299598217010498,
      "learning_rate": 9.998105263157895e-05,
      "loss": 0.4574,
      "step": 20082
    },
    {
      "epoch": 1.1245940194870647,
      "grad_norm": 1.1920071840286255,
      "learning_rate": 9.998078947368421e-05,
      "loss": 0.4934,
      "step": 20083
    },
    {
      "epoch": 1.1246500167991937,
      "grad_norm": 1.2635859251022339,
      "learning_rate": 9.998052631578948e-05,
      "loss": 0.4196,
      "step": 20084
    },
    {
      "epoch": 1.1247060141113228,
      "grad_norm": 1.3033627271652222,
      "learning_rate": 9.998026315789474e-05,
      "loss": 0.3937,
      "step": 20085
    },
    {
      "epoch": 1.1247620114234516,
      "grad_norm": 1.5347754955291748,
      "learning_rate": 9.998e-05,
      "loss": 0.5463,
      "step": 20086
    },
    {
      "epoch": 1.1248180087355806,
      "grad_norm": 1.1636805534362793,
      "learning_rate": 9.997973684210526e-05,
      "loss": 0.3372,
      "step": 20087
    },
    {
      "epoch": 1.1248740060477096,
      "grad_norm": 1.404335379600525,
      "learning_rate": 9.997947368421053e-05,
      "loss": 0.3842,
      "step": 20088
    },
    {
      "epoch": 1.1249300033598386,
      "grad_norm": 1.210408329963684,
      "learning_rate": 9.99792105263158e-05,
      "loss": 0.4594,
      "step": 20089
    },
    {
      "epoch": 1.1249860006719676,
      "grad_norm": 1.3475546836853027,
      "learning_rate": 9.997894736842107e-05,
      "loss": 0.4462,
      "step": 20090
    },
    {
      "epoch": 1.1250419979840967,
      "grad_norm": 1.3855321407318115,
      "learning_rate": 9.997868421052631e-05,
      "loss": 0.4473,
      "step": 20091
    },
    {
      "epoch": 1.1250979952962257,
      "grad_norm": 1.3210816383361816,
      "learning_rate": 9.997842105263159e-05,
      "loss": 0.4109,
      "step": 20092
    },
    {
      "epoch": 1.1251539926083547,
      "grad_norm": 1.2670806646347046,
      "learning_rate": 9.997815789473685e-05,
      "loss": 0.3964,
      "step": 20093
    },
    {
      "epoch": 1.1252099899204837,
      "grad_norm": 1.292494297027588,
      "learning_rate": 9.997789473684212e-05,
      "loss": 0.3616,
      "step": 20094
    },
    {
      "epoch": 1.1252659872326127,
      "grad_norm": 1.4878965616226196,
      "learning_rate": 9.997763157894737e-05,
      "loss": 0.4488,
      "step": 20095
    },
    {
      "epoch": 1.1253219845447417,
      "grad_norm": 1.2568440437316895,
      "learning_rate": 9.997736842105264e-05,
      "loss": 0.425,
      "step": 20096
    },
    {
      "epoch": 1.1253779818568708,
      "grad_norm": 1.6238255500793457,
      "learning_rate": 9.99771052631579e-05,
      "loss": 0.4368,
      "step": 20097
    },
    {
      "epoch": 1.1254339791689998,
      "grad_norm": 1.3254094123840332,
      "learning_rate": 9.997684210526316e-05,
      "loss": 0.3848,
      "step": 20098
    },
    {
      "epoch": 1.1254899764811288,
      "grad_norm": 1.2998703718185425,
      "learning_rate": 9.997657894736843e-05,
      "loss": 0.3981,
      "step": 20099
    },
    {
      "epoch": 1.1255459737932578,
      "grad_norm": 1.4188164472579956,
      "learning_rate": 9.997631578947368e-05,
      "loss": 0.569,
      "step": 20100
    },
    {
      "epoch": 1.1256019711053868,
      "grad_norm": 1.50592839717865,
      "learning_rate": 9.997605263157895e-05,
      "loss": 0.665,
      "step": 20101
    },
    {
      "epoch": 1.1256579684175159,
      "grad_norm": 1.3922353982925415,
      "learning_rate": 9.997578947368421e-05,
      "loss": 0.5353,
      "step": 20102
    },
    {
      "epoch": 1.1257139657296449,
      "grad_norm": 1.3044599294662476,
      "learning_rate": 9.997552631578948e-05,
      "loss": 0.439,
      "step": 20103
    },
    {
      "epoch": 1.125769963041774,
      "grad_norm": 1.473437786102295,
      "learning_rate": 9.997526315789474e-05,
      "loss": 0.5626,
      "step": 20104
    },
    {
      "epoch": 1.125825960353903,
      "grad_norm": 1.2697563171386719,
      "learning_rate": 9.9975e-05,
      "loss": 0.501,
      "step": 20105
    },
    {
      "epoch": 1.125881957666032,
      "grad_norm": 1.4197789430618286,
      "learning_rate": 9.997473684210526e-05,
      "loss": 0.5738,
      "step": 20106
    },
    {
      "epoch": 1.125937954978161,
      "grad_norm": 1.2517338991165161,
      "learning_rate": 9.997447368421054e-05,
      "loss": 0.434,
      "step": 20107
    },
    {
      "epoch": 1.12599395229029,
      "grad_norm": 1.3196479082107544,
      "learning_rate": 9.99742105263158e-05,
      "loss": 0.5725,
      "step": 20108
    },
    {
      "epoch": 1.126049949602419,
      "grad_norm": 1.2089532613754272,
      "learning_rate": 9.997394736842106e-05,
      "loss": 0.4245,
      "step": 20109
    },
    {
      "epoch": 1.126105946914548,
      "grad_norm": 1.2290751934051514,
      "learning_rate": 9.997368421052632e-05,
      "loss": 0.5232,
      "step": 20110
    },
    {
      "epoch": 1.126161944226677,
      "grad_norm": 1.5362193584442139,
      "learning_rate": 9.997342105263159e-05,
      "loss": 0.5124,
      "step": 20111
    },
    {
      "epoch": 1.126217941538806,
      "grad_norm": 1.1183627843856812,
      "learning_rate": 9.997315789473685e-05,
      "loss": 0.3339,
      "step": 20112
    },
    {
      "epoch": 1.126273938850935,
      "grad_norm": 1.3534480333328247,
      "learning_rate": 9.997289473684211e-05,
      "loss": 0.37,
      "step": 20113
    },
    {
      "epoch": 1.126329936163064,
      "grad_norm": 1.407397747039795,
      "learning_rate": 9.997263157894737e-05,
      "loss": 0.3819,
      "step": 20114
    },
    {
      "epoch": 1.1263859334751931,
      "grad_norm": 1.2968662977218628,
      "learning_rate": 9.997236842105263e-05,
      "loss": 0.4112,
      "step": 20115
    },
    {
      "epoch": 1.1264419307873221,
      "grad_norm": 1.645027756690979,
      "learning_rate": 9.99721052631579e-05,
      "loss": 0.5773,
      "step": 20116
    },
    {
      "epoch": 1.1264979280994512,
      "grad_norm": 1.0771293640136719,
      "learning_rate": 9.997184210526316e-05,
      "loss": 0.3627,
      "step": 20117
    },
    {
      "epoch": 1.1265539254115802,
      "grad_norm": 1.7107199430465698,
      "learning_rate": 9.997157894736842e-05,
      "loss": 0.6184,
      "step": 20118
    },
    {
      "epoch": 1.1266099227237092,
      "grad_norm": 1.67527174949646,
      "learning_rate": 9.997131578947368e-05,
      "loss": 0.6594,
      "step": 20119
    },
    {
      "epoch": 1.1266659200358382,
      "grad_norm": 1.3696786165237427,
      "learning_rate": 9.997105263157895e-05,
      "loss": 0.3945,
      "step": 20120
    },
    {
      "epoch": 1.1267219173479672,
      "grad_norm": 1.4254775047302246,
      "learning_rate": 9.997078947368421e-05,
      "loss": 0.5711,
      "step": 20121
    },
    {
      "epoch": 1.1267779146600962,
      "grad_norm": 1.559033989906311,
      "learning_rate": 9.997052631578949e-05,
      "loss": 0.5888,
      "step": 20122
    },
    {
      "epoch": 1.1268339119722253,
      "grad_norm": 1.3969683647155762,
      "learning_rate": 9.997026315789473e-05,
      "loss": 0.4524,
      "step": 20123
    },
    {
      "epoch": 1.1268899092843543,
      "grad_norm": 1.4289509057998657,
      "learning_rate": 9.997e-05,
      "loss": 0.4408,
      "step": 20124
    },
    {
      "epoch": 1.1269459065964833,
      "grad_norm": 1.3135063648223877,
      "learning_rate": 9.996973684210527e-05,
      "loss": 0.5331,
      "step": 20125
    },
    {
      "epoch": 1.1270019039086123,
      "grad_norm": 1.2597249746322632,
      "learning_rate": 9.996947368421054e-05,
      "loss": 0.3825,
      "step": 20126
    },
    {
      "epoch": 1.1270579012207413,
      "grad_norm": 1.3031622171401978,
      "learning_rate": 9.99692105263158e-05,
      "loss": 0.5424,
      "step": 20127
    },
    {
      "epoch": 1.1271138985328704,
      "grad_norm": 1.5318948030471802,
      "learning_rate": 9.996894736842106e-05,
      "loss": 0.4662,
      "step": 20128
    },
    {
      "epoch": 1.1271698958449994,
      "grad_norm": 1.6408461332321167,
      "learning_rate": 9.996868421052632e-05,
      "loss": 0.4568,
      "step": 20129
    },
    {
      "epoch": 1.1272258931571284,
      "grad_norm": 1.4029548168182373,
      "learning_rate": 9.996842105263159e-05,
      "loss": 0.5072,
      "step": 20130
    },
    {
      "epoch": 1.1272818904692574,
      "grad_norm": 1.8679447174072266,
      "learning_rate": 9.996815789473685e-05,
      "loss": 0.6695,
      "step": 20131
    },
    {
      "epoch": 1.1273378877813864,
      "grad_norm": 1.3182162046432495,
      "learning_rate": 9.99678947368421e-05,
      "loss": 0.477,
      "step": 20132
    },
    {
      "epoch": 1.1273938850935155,
      "grad_norm": 1.8410577774047852,
      "learning_rate": 9.996763157894737e-05,
      "loss": 0.5332,
      "step": 20133
    },
    {
      "epoch": 1.1274498824056445,
      "grad_norm": 1.2833086252212524,
      "learning_rate": 9.996736842105263e-05,
      "loss": 0.4246,
      "step": 20134
    },
    {
      "epoch": 1.1275058797177735,
      "grad_norm": 1.196503758430481,
      "learning_rate": 9.99671052631579e-05,
      "loss": 0.4582,
      "step": 20135
    },
    {
      "epoch": 1.1275618770299025,
      "grad_norm": 1.2653039693832397,
      "learning_rate": 9.996684210526316e-05,
      "loss": 0.4371,
      "step": 20136
    },
    {
      "epoch": 1.1276178743420315,
      "grad_norm": 1.2870479822158813,
      "learning_rate": 9.996657894736842e-05,
      "loss": 0.3413,
      "step": 20137
    },
    {
      "epoch": 1.1276738716541606,
      "grad_norm": 1.4446899890899658,
      "learning_rate": 9.996631578947368e-05,
      "loss": 0.3806,
      "step": 20138
    },
    {
      "epoch": 1.1277298689662896,
      "grad_norm": 1.706510066986084,
      "learning_rate": 9.996605263157896e-05,
      "loss": 0.6093,
      "step": 20139
    },
    {
      "epoch": 1.1277858662784186,
      "grad_norm": 1.2224639654159546,
      "learning_rate": 9.996578947368422e-05,
      "loss": 0.4163,
      "step": 20140
    },
    {
      "epoch": 1.1278418635905476,
      "grad_norm": 1.2918223142623901,
      "learning_rate": 9.996552631578948e-05,
      "loss": 0.4171,
      "step": 20141
    },
    {
      "epoch": 1.1278978609026766,
      "grad_norm": 1.895620346069336,
      "learning_rate": 9.996526315789474e-05,
      "loss": 0.434,
      "step": 20142
    },
    {
      "epoch": 1.1279538582148056,
      "grad_norm": 1.5262978076934814,
      "learning_rate": 9.996500000000001e-05,
      "loss": 0.4186,
      "step": 20143
    },
    {
      "epoch": 1.1280098555269347,
      "grad_norm": 1.4467686414718628,
      "learning_rate": 9.996473684210527e-05,
      "loss": 0.4421,
      "step": 20144
    },
    {
      "epoch": 1.1280658528390637,
      "grad_norm": 1.2918657064437866,
      "learning_rate": 9.996447368421054e-05,
      "loss": 0.4428,
      "step": 20145
    },
    {
      "epoch": 1.1281218501511927,
      "grad_norm": 1.3140455484390259,
      "learning_rate": 9.996421052631579e-05,
      "loss": 0.4878,
      "step": 20146
    },
    {
      "epoch": 1.1281778474633217,
      "grad_norm": 1.4031810760498047,
      "learning_rate": 9.996394736842106e-05,
      "loss": 0.5236,
      "step": 20147
    },
    {
      "epoch": 1.1282338447754507,
      "grad_norm": 1.2455710172653198,
      "learning_rate": 9.996368421052632e-05,
      "loss": 0.4509,
      "step": 20148
    },
    {
      "epoch": 1.1282898420875798,
      "grad_norm": 1.3842425346374512,
      "learning_rate": 9.996342105263158e-05,
      "loss": 0.4766,
      "step": 20149
    },
    {
      "epoch": 1.1283458393997088,
      "grad_norm": 1.1807523965835571,
      "learning_rate": 9.996315789473684e-05,
      "loss": 0.401,
      "step": 20150
    },
    {
      "epoch": 1.1284018367118378,
      "grad_norm": 1.1878485679626465,
      "learning_rate": 9.99628947368421e-05,
      "loss": 0.455,
      "step": 20151
    },
    {
      "epoch": 1.1284578340239668,
      "grad_norm": 1.2592793703079224,
      "learning_rate": 9.996263157894737e-05,
      "loss": 0.4084,
      "step": 20152
    },
    {
      "epoch": 1.1285138313360958,
      "grad_norm": 1.6455912590026855,
      "learning_rate": 9.996236842105263e-05,
      "loss": 0.4427,
      "step": 20153
    },
    {
      "epoch": 1.1285698286482249,
      "grad_norm": 1.357524037361145,
      "learning_rate": 9.996210526315791e-05,
      "loss": 0.3725,
      "step": 20154
    },
    {
      "epoch": 1.1286258259603539,
      "grad_norm": 1.5846613645553589,
      "learning_rate": 9.996184210526315e-05,
      "loss": 0.4286,
      "step": 20155
    },
    {
      "epoch": 1.128681823272483,
      "grad_norm": 1.5454031229019165,
      "learning_rate": 9.996157894736843e-05,
      "loss": 0.5492,
      "step": 20156
    },
    {
      "epoch": 1.128737820584612,
      "grad_norm": 1.4770495891571045,
      "learning_rate": 9.996131578947369e-05,
      "loss": 0.4957,
      "step": 20157
    },
    {
      "epoch": 1.128793817896741,
      "grad_norm": 1.6208133697509766,
      "learning_rate": 9.996105263157896e-05,
      "loss": 0.4944,
      "step": 20158
    },
    {
      "epoch": 1.12884981520887,
      "grad_norm": 1.5294944047927856,
      "learning_rate": 9.996078947368422e-05,
      "loss": 0.5993,
      "step": 20159
    },
    {
      "epoch": 1.128905812520999,
      "grad_norm": 1.2093846797943115,
      "learning_rate": 9.996052631578948e-05,
      "loss": 0.3749,
      "step": 20160
    },
    {
      "epoch": 1.128961809833128,
      "grad_norm": 1.3696269989013672,
      "learning_rate": 9.996026315789474e-05,
      "loss": 0.4462,
      "step": 20161
    },
    {
      "epoch": 1.129017807145257,
      "grad_norm": 1.315756916999817,
      "learning_rate": 9.996000000000001e-05,
      "loss": 0.4758,
      "step": 20162
    },
    {
      "epoch": 1.129073804457386,
      "grad_norm": 1.4041450023651123,
      "learning_rate": 9.995973684210527e-05,
      "loss": 0.3999,
      "step": 20163
    },
    {
      "epoch": 1.129129801769515,
      "grad_norm": 1.455556035041809,
      "learning_rate": 9.995947368421053e-05,
      "loss": 0.4594,
      "step": 20164
    },
    {
      "epoch": 1.129185799081644,
      "grad_norm": 1.2051758766174316,
      "learning_rate": 9.995921052631579e-05,
      "loss": 0.3883,
      "step": 20165
    },
    {
      "epoch": 1.129241796393773,
      "grad_norm": 1.2870639562606812,
      "learning_rate": 9.995894736842105e-05,
      "loss": 0.5278,
      "step": 20166
    },
    {
      "epoch": 1.129297793705902,
      "grad_norm": 1.8627378940582275,
      "learning_rate": 9.995868421052632e-05,
      "loss": 0.4947,
      "step": 20167
    },
    {
      "epoch": 1.1293537910180311,
      "grad_norm": 1.3675086498260498,
      "learning_rate": 9.995842105263158e-05,
      "loss": 0.3386,
      "step": 20168
    },
    {
      "epoch": 1.1294097883301601,
      "grad_norm": 1.2064725160598755,
      "learning_rate": 9.995815789473684e-05,
      "loss": 0.4058,
      "step": 20169
    },
    {
      "epoch": 1.1294657856422892,
      "grad_norm": 1.3493707180023193,
      "learning_rate": 9.99578947368421e-05,
      "loss": 0.4858,
      "step": 20170
    },
    {
      "epoch": 1.1295217829544182,
      "grad_norm": 1.31057870388031,
      "learning_rate": 9.995763157894738e-05,
      "loss": 0.3552,
      "step": 20171
    },
    {
      "epoch": 1.1295777802665472,
      "grad_norm": 1.3686400651931763,
      "learning_rate": 9.995736842105264e-05,
      "loss": 0.4329,
      "step": 20172
    },
    {
      "epoch": 1.1296337775786762,
      "grad_norm": 1.170465350151062,
      "learning_rate": 9.99571052631579e-05,
      "loss": 0.3591,
      "step": 20173
    },
    {
      "epoch": 1.1296897748908052,
      "grad_norm": 1.2574832439422607,
      "learning_rate": 9.995684210526316e-05,
      "loss": 0.4382,
      "step": 20174
    },
    {
      "epoch": 1.1297457722029343,
      "grad_norm": 1.652482509613037,
      "learning_rate": 9.995657894736843e-05,
      "loss": 0.4608,
      "step": 20175
    },
    {
      "epoch": 1.1298017695150633,
      "grad_norm": 1.4721181392669678,
      "learning_rate": 9.995631578947369e-05,
      "loss": 0.4936,
      "step": 20176
    },
    {
      "epoch": 1.1298577668271923,
      "grad_norm": 1.2871087789535522,
      "learning_rate": 9.995605263157896e-05,
      "loss": 0.3359,
      "step": 20177
    },
    {
      "epoch": 1.1299137641393213,
      "grad_norm": 1.4253180027008057,
      "learning_rate": 9.995578947368421e-05,
      "loss": 0.4732,
      "step": 20178
    },
    {
      "epoch": 1.1299697614514503,
      "grad_norm": 1.4310779571533203,
      "learning_rate": 9.995552631578948e-05,
      "loss": 0.45,
      "step": 20179
    },
    {
      "epoch": 1.1300257587635794,
      "grad_norm": 1.3896969556808472,
      "learning_rate": 9.995526315789474e-05,
      "loss": 0.4549,
      "step": 20180
    },
    {
      "epoch": 1.1300817560757084,
      "grad_norm": 4.925571441650391,
      "learning_rate": 9.995500000000001e-05,
      "loss": 0.4441,
      "step": 20181
    },
    {
      "epoch": 1.1301377533878374,
      "grad_norm": 1.9883801937103271,
      "learning_rate": 9.995473684210527e-05,
      "loss": 0.431,
      "step": 20182
    },
    {
      "epoch": 1.1301937506999664,
      "grad_norm": 1.221846580505371,
      "learning_rate": 9.995447368421052e-05,
      "loss": 0.3792,
      "step": 20183
    },
    {
      "epoch": 1.1302497480120954,
      "grad_norm": 1.3770896196365356,
      "learning_rate": 9.99542105263158e-05,
      "loss": 0.5133,
      "step": 20184
    },
    {
      "epoch": 1.1303057453242245,
      "grad_norm": 1.4019097089767456,
      "learning_rate": 9.995394736842105e-05,
      "loss": 0.5023,
      "step": 20185
    },
    {
      "epoch": 1.1303617426363535,
      "grad_norm": 1.2293190956115723,
      "learning_rate": 9.995368421052633e-05,
      "loss": 0.4406,
      "step": 20186
    },
    {
      "epoch": 1.1304177399484825,
      "grad_norm": 1.328400731086731,
      "learning_rate": 9.995342105263157e-05,
      "loss": 0.4856,
      "step": 20187
    },
    {
      "epoch": 1.1304737372606115,
      "grad_norm": 1.5837477445602417,
      "learning_rate": 9.995315789473685e-05,
      "loss": 0.4598,
      "step": 20188
    },
    {
      "epoch": 1.1305297345727405,
      "grad_norm": 1.52598237991333,
      "learning_rate": 9.99528947368421e-05,
      "loss": 0.4943,
      "step": 20189
    },
    {
      "epoch": 1.1305857318848695,
      "grad_norm": 1.2836613655090332,
      "learning_rate": 9.995263157894738e-05,
      "loss": 0.431,
      "step": 20190
    },
    {
      "epoch": 1.1306417291969986,
      "grad_norm": 2.122936487197876,
      "learning_rate": 9.995236842105264e-05,
      "loss": 0.5085,
      "step": 20191
    },
    {
      "epoch": 1.1306977265091276,
      "grad_norm": 1.7128816843032837,
      "learning_rate": 9.99521052631579e-05,
      "loss": 0.4088,
      "step": 20192
    },
    {
      "epoch": 1.1307537238212566,
      "grad_norm": 2.03788685798645,
      "learning_rate": 9.995184210526316e-05,
      "loss": 0.6117,
      "step": 20193
    },
    {
      "epoch": 1.1308097211333856,
      "grad_norm": 1.322121024131775,
      "learning_rate": 9.995157894736843e-05,
      "loss": 0.323,
      "step": 20194
    },
    {
      "epoch": 1.1308657184455146,
      "grad_norm": 1.515406847000122,
      "learning_rate": 9.995131578947369e-05,
      "loss": 0.4084,
      "step": 20195
    },
    {
      "epoch": 1.1309217157576437,
      "grad_norm": 1.3654356002807617,
      "learning_rate": 9.995105263157895e-05,
      "loss": 0.3751,
      "step": 20196
    },
    {
      "epoch": 1.1309777130697727,
      "grad_norm": 1.6777520179748535,
      "learning_rate": 9.995078947368421e-05,
      "loss": 0.5004,
      "step": 20197
    },
    {
      "epoch": 1.1310337103819017,
      "grad_norm": 1.3237303495407104,
      "learning_rate": 9.995052631578948e-05,
      "loss": 0.4237,
      "step": 20198
    },
    {
      "epoch": 1.1310897076940307,
      "grad_norm": 1.4779011011123657,
      "learning_rate": 9.995026315789474e-05,
      "loss": 0.3831,
      "step": 20199
    },
    {
      "epoch": 1.1311457050061597,
      "grad_norm": 1.2313202619552612,
      "learning_rate": 9.995e-05,
      "loss": 0.3035,
      "step": 20200
    },
    {
      "epoch": 1.1312017023182888,
      "grad_norm": 1.7041383981704712,
      "learning_rate": 9.994973684210526e-05,
      "loss": 0.4756,
      "step": 20201
    },
    {
      "epoch": 1.1312576996304178,
      "grad_norm": 1.3493001461029053,
      "learning_rate": 9.994947368421052e-05,
      "loss": 0.6141,
      "step": 20202
    },
    {
      "epoch": 1.1313136969425468,
      "grad_norm": 1.3731756210327148,
      "learning_rate": 9.99492105263158e-05,
      "loss": 0.3948,
      "step": 20203
    },
    {
      "epoch": 1.1313696942546758,
      "grad_norm": 1.9412846565246582,
      "learning_rate": 9.994894736842106e-05,
      "loss": 0.5277,
      "step": 20204
    },
    {
      "epoch": 1.1314256915668048,
      "grad_norm": 1.3892247676849365,
      "learning_rate": 9.994868421052632e-05,
      "loss": 0.5429,
      "step": 20205
    },
    {
      "epoch": 1.1314816888789339,
      "grad_norm": 1.6253856420516968,
      "learning_rate": 9.994842105263158e-05,
      "loss": 0.611,
      "step": 20206
    },
    {
      "epoch": 1.1315376861910629,
      "grad_norm": 1.4898595809936523,
      "learning_rate": 9.994815789473685e-05,
      "loss": 0.3932,
      "step": 20207
    },
    {
      "epoch": 1.131593683503192,
      "grad_norm": 1.2663226127624512,
      "learning_rate": 9.994789473684211e-05,
      "loss": 0.5623,
      "step": 20208
    },
    {
      "epoch": 1.131649680815321,
      "grad_norm": 1.4919707775115967,
      "learning_rate": 9.994763157894738e-05,
      "loss": 0.4961,
      "step": 20209
    },
    {
      "epoch": 1.13170567812745,
      "grad_norm": 1.314507007598877,
      "learning_rate": 9.994736842105263e-05,
      "loss": 0.503,
      "step": 20210
    },
    {
      "epoch": 1.131761675439579,
      "grad_norm": 1.7011282444000244,
      "learning_rate": 9.99471052631579e-05,
      "loss": 0.4651,
      "step": 20211
    },
    {
      "epoch": 1.131817672751708,
      "grad_norm": 1.2699919939041138,
      "learning_rate": 9.994684210526316e-05,
      "loss": 0.493,
      "step": 20212
    },
    {
      "epoch": 1.131873670063837,
      "grad_norm": 1.4917173385620117,
      "learning_rate": 9.994657894736843e-05,
      "loss": 0.4717,
      "step": 20213
    },
    {
      "epoch": 1.131929667375966,
      "grad_norm": 1.2997092008590698,
      "learning_rate": 9.99463157894737e-05,
      "loss": 0.5401,
      "step": 20214
    },
    {
      "epoch": 1.131985664688095,
      "grad_norm": 2.0021791458129883,
      "learning_rate": 9.994605263157895e-05,
      "loss": 0.6804,
      "step": 20215
    },
    {
      "epoch": 1.132041662000224,
      "grad_norm": 1.6106153726577759,
      "learning_rate": 9.994578947368421e-05,
      "loss": 0.3436,
      "step": 20216
    },
    {
      "epoch": 1.132097659312353,
      "grad_norm": 1.16148841381073,
      "learning_rate": 9.994552631578949e-05,
      "loss": 0.4427,
      "step": 20217
    },
    {
      "epoch": 1.132153656624482,
      "grad_norm": 1.8368303775787354,
      "learning_rate": 9.994526315789475e-05,
      "loss": 0.5691,
      "step": 20218
    },
    {
      "epoch": 1.132209653936611,
      "grad_norm": 1.274310827255249,
      "learning_rate": 9.9945e-05,
      "loss": 0.4706,
      "step": 20219
    },
    {
      "epoch": 1.1322656512487401,
      "grad_norm": 1.1377419233322144,
      "learning_rate": 9.994473684210527e-05,
      "loss": 0.4518,
      "step": 20220
    },
    {
      "epoch": 1.1323216485608691,
      "grad_norm": 1.1848833560943604,
      "learning_rate": 9.994447368421053e-05,
      "loss": 0.3248,
      "step": 20221
    },
    {
      "epoch": 1.1323776458729982,
      "grad_norm": 1.4867243766784668,
      "learning_rate": 9.99442105263158e-05,
      "loss": 0.4576,
      "step": 20222
    },
    {
      "epoch": 1.1324336431851272,
      "grad_norm": 1.7923411130905151,
      "learning_rate": 9.994394736842106e-05,
      "loss": 0.4153,
      "step": 20223
    },
    {
      "epoch": 1.1324896404972562,
      "grad_norm": 1.1907631158828735,
      "learning_rate": 9.994368421052632e-05,
      "loss": 0.3663,
      "step": 20224
    },
    {
      "epoch": 1.1325456378093852,
      "grad_norm": 1.2464754581451416,
      "learning_rate": 9.994342105263158e-05,
      "loss": 0.4706,
      "step": 20225
    },
    {
      "epoch": 1.1326016351215142,
      "grad_norm": 1.1599808931350708,
      "learning_rate": 9.994315789473685e-05,
      "loss": 0.4196,
      "step": 20226
    },
    {
      "epoch": 1.1326576324336433,
      "grad_norm": 1.3001772165298462,
      "learning_rate": 9.994289473684211e-05,
      "loss": 0.4365,
      "step": 20227
    },
    {
      "epoch": 1.1327136297457723,
      "grad_norm": 1.241241455078125,
      "learning_rate": 9.994263157894737e-05,
      "loss": 0.3607,
      "step": 20228
    },
    {
      "epoch": 1.1327696270579013,
      "grad_norm": 1.176579475402832,
      "learning_rate": 9.994236842105263e-05,
      "loss": 0.2961,
      "step": 20229
    },
    {
      "epoch": 1.1328256243700303,
      "grad_norm": 1.525162696838379,
      "learning_rate": 9.99421052631579e-05,
      "loss": 0.5938,
      "step": 20230
    },
    {
      "epoch": 1.1328816216821593,
      "grad_norm": 1.3000282049179077,
      "learning_rate": 9.994184210526316e-05,
      "loss": 0.4874,
      "step": 20231
    },
    {
      "epoch": 1.1329376189942884,
      "grad_norm": 1.2870616912841797,
      "learning_rate": 9.994157894736844e-05,
      "loss": 0.4187,
      "step": 20232
    },
    {
      "epoch": 1.1329936163064174,
      "grad_norm": 1.3188973665237427,
      "learning_rate": 9.994131578947368e-05,
      "loss": 0.4117,
      "step": 20233
    },
    {
      "epoch": 1.1330496136185464,
      "grad_norm": 1.2505483627319336,
      "learning_rate": 9.994105263157896e-05,
      "loss": 0.302,
      "step": 20234
    },
    {
      "epoch": 1.1331056109306754,
      "grad_norm": 1.1003673076629639,
      "learning_rate": 9.994078947368422e-05,
      "loss": 0.383,
      "step": 20235
    },
    {
      "epoch": 1.1331616082428044,
      "grad_norm": 1.0437648296356201,
      "learning_rate": 9.994052631578948e-05,
      "loss": 0.4162,
      "step": 20236
    },
    {
      "epoch": 1.1332176055549334,
      "grad_norm": 1.2069073915481567,
      "learning_rate": 9.994026315789475e-05,
      "loss": 0.3115,
      "step": 20237
    },
    {
      "epoch": 1.1332736028670625,
      "grad_norm": 1.2497919797897339,
      "learning_rate": 9.994e-05,
      "loss": 0.4764,
      "step": 20238
    },
    {
      "epoch": 1.1333296001791915,
      "grad_norm": 1.428944706916809,
      "learning_rate": 9.993973684210527e-05,
      "loss": 0.5107,
      "step": 20239
    },
    {
      "epoch": 1.1333855974913205,
      "grad_norm": 1.3723713159561157,
      "learning_rate": 9.993947368421053e-05,
      "loss": 0.5085,
      "step": 20240
    },
    {
      "epoch": 1.1334415948034495,
      "grad_norm": 1.3960775136947632,
      "learning_rate": 9.99392105263158e-05,
      "loss": 0.5389,
      "step": 20241
    },
    {
      "epoch": 1.1334975921155785,
      "grad_norm": 1.1459434032440186,
      "learning_rate": 9.993894736842105e-05,
      "loss": 0.3664,
      "step": 20242
    },
    {
      "epoch": 1.1335535894277076,
      "grad_norm": 1.4901725053787231,
      "learning_rate": 9.993868421052632e-05,
      "loss": 0.5682,
      "step": 20243
    },
    {
      "epoch": 1.1336095867398366,
      "grad_norm": 1.2028594017028809,
      "learning_rate": 9.993842105263158e-05,
      "loss": 0.2462,
      "step": 20244
    },
    {
      "epoch": 1.1336655840519656,
      "grad_norm": 1.3778281211853027,
      "learning_rate": 9.993815789473685e-05,
      "loss": 0.4629,
      "step": 20245
    },
    {
      "epoch": 1.1337215813640946,
      "grad_norm": 1.487456202507019,
      "learning_rate": 9.993789473684211e-05,
      "loss": 0.4372,
      "step": 20246
    },
    {
      "epoch": 1.1337775786762236,
      "grad_norm": 2.2966628074645996,
      "learning_rate": 9.993763157894737e-05,
      "loss": 0.6028,
      "step": 20247
    },
    {
      "epoch": 1.1338335759883527,
      "grad_norm": 1.343872308731079,
      "learning_rate": 9.993736842105263e-05,
      "loss": 0.3569,
      "step": 20248
    },
    {
      "epoch": 1.1338895733004817,
      "grad_norm": 1.3263475894927979,
      "learning_rate": 9.99371052631579e-05,
      "loss": 0.4684,
      "step": 20249
    },
    {
      "epoch": 1.1339455706126107,
      "grad_norm": 1.538833737373352,
      "learning_rate": 9.993684210526317e-05,
      "loss": 0.5163,
      "step": 20250
    },
    {
      "epoch": 1.1340015679247397,
      "grad_norm": 1.429449200630188,
      "learning_rate": 9.993657894736843e-05,
      "loss": 0.454,
      "step": 20251
    },
    {
      "epoch": 1.1340575652368687,
      "grad_norm": 1.3572133779525757,
      "learning_rate": 9.993631578947369e-05,
      "loss": 0.6305,
      "step": 20252
    },
    {
      "epoch": 1.1341135625489978,
      "grad_norm": 1.3160983324050903,
      "learning_rate": 9.993605263157895e-05,
      "loss": 0.3628,
      "step": 20253
    },
    {
      "epoch": 1.1341695598611268,
      "grad_norm": 1.3186285495758057,
      "learning_rate": 9.993578947368422e-05,
      "loss": 0.467,
      "step": 20254
    },
    {
      "epoch": 1.1342255571732558,
      "grad_norm": 1.1341521739959717,
      "learning_rate": 9.993552631578948e-05,
      "loss": 0.3895,
      "step": 20255
    },
    {
      "epoch": 1.1342815544853848,
      "grad_norm": 1.5238111019134521,
      "learning_rate": 9.993526315789474e-05,
      "loss": 0.3964,
      "step": 20256
    },
    {
      "epoch": 1.1343375517975138,
      "grad_norm": 1.2447681427001953,
      "learning_rate": 9.9935e-05,
      "loss": 0.4356,
      "step": 20257
    },
    {
      "epoch": 1.1343935491096429,
      "grad_norm": 1.2781203985214233,
      "learning_rate": 9.993473684210527e-05,
      "loss": 0.3634,
      "step": 20258
    },
    {
      "epoch": 1.1344495464217719,
      "grad_norm": 1.5016010999679565,
      "learning_rate": 9.993447368421053e-05,
      "loss": 0.4863,
      "step": 20259
    },
    {
      "epoch": 1.134505543733901,
      "grad_norm": 1.2829502820968628,
      "learning_rate": 9.993421052631579e-05,
      "loss": 0.465,
      "step": 20260
    },
    {
      "epoch": 1.13456154104603,
      "grad_norm": 1.2866705656051636,
      "learning_rate": 9.993394736842105e-05,
      "loss": 0.3207,
      "step": 20261
    },
    {
      "epoch": 1.134617538358159,
      "grad_norm": 1.5260947942733765,
      "learning_rate": 9.993368421052632e-05,
      "loss": 0.4029,
      "step": 20262
    },
    {
      "epoch": 1.134673535670288,
      "grad_norm": 1.3858051300048828,
      "learning_rate": 9.993342105263158e-05,
      "loss": 0.3886,
      "step": 20263
    },
    {
      "epoch": 1.134729532982417,
      "grad_norm": 1.4056178331375122,
      "learning_rate": 9.993315789473686e-05,
      "loss": 0.6166,
      "step": 20264
    },
    {
      "epoch": 1.134785530294546,
      "grad_norm": 2.4712114334106445,
      "learning_rate": 9.99328947368421e-05,
      "loss": 0.4633,
      "step": 20265
    },
    {
      "epoch": 1.1348415276066748,
      "grad_norm": 1.2079639434814453,
      "learning_rate": 9.993263157894738e-05,
      "loss": 0.385,
      "step": 20266
    },
    {
      "epoch": 1.1348975249188038,
      "grad_norm": 1.3643262386322021,
      "learning_rate": 9.993236842105264e-05,
      "loss": 0.4943,
      "step": 20267
    },
    {
      "epoch": 1.1349535222309328,
      "grad_norm": 1.1451470851898193,
      "learning_rate": 9.993210526315791e-05,
      "loss": 0.3125,
      "step": 20268
    },
    {
      "epoch": 1.1350095195430618,
      "grad_norm": 1.769522786140442,
      "learning_rate": 9.993184210526317e-05,
      "loss": 0.4586,
      "step": 20269
    },
    {
      "epoch": 1.1350655168551909,
      "grad_norm": 9.267542839050293,
      "learning_rate": 9.993157894736841e-05,
      "loss": 0.4237,
      "step": 20270
    },
    {
      "epoch": 1.1351215141673199,
      "grad_norm": 3.177318572998047,
      "learning_rate": 9.993131578947369e-05,
      "loss": 0.4726,
      "step": 20271
    },
    {
      "epoch": 1.135177511479449,
      "grad_norm": 1.3887708187103271,
      "learning_rate": 9.993105263157895e-05,
      "loss": 0.4423,
      "step": 20272
    },
    {
      "epoch": 1.135233508791578,
      "grad_norm": 1.4719115495681763,
      "learning_rate": 9.993078947368422e-05,
      "loss": 0.449,
      "step": 20273
    },
    {
      "epoch": 1.135289506103707,
      "grad_norm": 1.4405137300491333,
      "learning_rate": 9.993052631578948e-05,
      "loss": 0.4776,
      "step": 20274
    },
    {
      "epoch": 1.135345503415836,
      "grad_norm": 1.2587157487869263,
      "learning_rate": 9.993026315789474e-05,
      "loss": 0.4237,
      "step": 20275
    },
    {
      "epoch": 1.135401500727965,
      "grad_norm": 1.2803078889846802,
      "learning_rate": 9.993e-05,
      "loss": 0.4338,
      "step": 20276
    },
    {
      "epoch": 1.135457498040094,
      "grad_norm": 1.1684249639511108,
      "learning_rate": 9.992973684210527e-05,
      "loss": 0.3643,
      "step": 20277
    },
    {
      "epoch": 1.135513495352223,
      "grad_norm": 1.356095314025879,
      "learning_rate": 9.992947368421053e-05,
      "loss": 0.5039,
      "step": 20278
    },
    {
      "epoch": 1.135569492664352,
      "grad_norm": 1.1681362390518188,
      "learning_rate": 9.992921052631579e-05,
      "loss": 0.4442,
      "step": 20279
    },
    {
      "epoch": 1.135625489976481,
      "grad_norm": 1.3019719123840332,
      "learning_rate": 9.992894736842105e-05,
      "loss": 0.4194,
      "step": 20280
    },
    {
      "epoch": 1.13568148728861,
      "grad_norm": 1.1020950078964233,
      "learning_rate": 9.992868421052633e-05,
      "loss": 0.3683,
      "step": 20281
    },
    {
      "epoch": 1.135737484600739,
      "grad_norm": 1.5293177366256714,
      "learning_rate": 9.992842105263159e-05,
      "loss": 0.494,
      "step": 20282
    },
    {
      "epoch": 1.135793481912868,
      "grad_norm": 1.2927979230880737,
      "learning_rate": 9.992815789473685e-05,
      "loss": 0.4483,
      "step": 20283
    },
    {
      "epoch": 1.1358494792249971,
      "grad_norm": 1.9020200967788696,
      "learning_rate": 9.99278947368421e-05,
      "loss": 0.5442,
      "step": 20284
    },
    {
      "epoch": 1.1359054765371261,
      "grad_norm": 1.4373677968978882,
      "learning_rate": 9.992763157894738e-05,
      "loss": 0.4161,
      "step": 20285
    },
    {
      "epoch": 1.1359614738492552,
      "grad_norm": 1.4182528257369995,
      "learning_rate": 9.992736842105264e-05,
      "loss": 0.4502,
      "step": 20286
    },
    {
      "epoch": 1.1360174711613842,
      "grad_norm": 1.112560510635376,
      "learning_rate": 9.99271052631579e-05,
      "loss": 0.3414,
      "step": 20287
    },
    {
      "epoch": 1.1360734684735132,
      "grad_norm": 1.262067198753357,
      "learning_rate": 9.992684210526316e-05,
      "loss": 0.4059,
      "step": 20288
    },
    {
      "epoch": 1.1361294657856422,
      "grad_norm": 1.3725446462631226,
      "learning_rate": 9.992657894736842e-05,
      "loss": 0.44,
      "step": 20289
    },
    {
      "epoch": 1.1361854630977712,
      "grad_norm": 1.2827565670013428,
      "learning_rate": 9.992631578947369e-05,
      "loss": 0.4307,
      "step": 20290
    },
    {
      "epoch": 1.1362414604099003,
      "grad_norm": 1.291059136390686,
      "learning_rate": 9.992605263157895e-05,
      "loss": 0.4441,
      "step": 20291
    },
    {
      "epoch": 1.1362974577220293,
      "grad_norm": 1.5054551362991333,
      "learning_rate": 9.992578947368422e-05,
      "loss": 0.477,
      "step": 20292
    },
    {
      "epoch": 1.1363534550341583,
      "grad_norm": 1.8364559412002563,
      "learning_rate": 9.992552631578947e-05,
      "loss": 0.4822,
      "step": 20293
    },
    {
      "epoch": 1.1364094523462873,
      "grad_norm": 1.191460371017456,
      "learning_rate": 9.992526315789474e-05,
      "loss": 0.4269,
      "step": 20294
    },
    {
      "epoch": 1.1364654496584163,
      "grad_norm": 3.196669578552246,
      "learning_rate": 9.9925e-05,
      "loss": 0.5738,
      "step": 20295
    },
    {
      "epoch": 1.1365214469705454,
      "grad_norm": 1.3577088117599487,
      "learning_rate": 9.992473684210528e-05,
      "loss": 0.4304,
      "step": 20296
    },
    {
      "epoch": 1.1365774442826744,
      "grad_norm": 1.5136973857879639,
      "learning_rate": 9.992447368421052e-05,
      "loss": 0.4075,
      "step": 20297
    },
    {
      "epoch": 1.1366334415948034,
      "grad_norm": 1.4799716472625732,
      "learning_rate": 9.99242105263158e-05,
      "loss": 0.4673,
      "step": 20298
    },
    {
      "epoch": 1.1366894389069324,
      "grad_norm": 1.289697527885437,
      "learning_rate": 9.992394736842106e-05,
      "loss": 0.3812,
      "step": 20299
    },
    {
      "epoch": 1.1367454362190614,
      "grad_norm": 1.0879672765731812,
      "learning_rate": 9.992368421052633e-05,
      "loss": 0.3818,
      "step": 20300
    },
    {
      "epoch": 1.1368014335311905,
      "grad_norm": 1.3830888271331787,
      "learning_rate": 9.992342105263159e-05,
      "loss": 0.4774,
      "step": 20301
    },
    {
      "epoch": 1.1368574308433195,
      "grad_norm": 1.3574031591415405,
      "learning_rate": 9.992315789473685e-05,
      "loss": 0.6394,
      "step": 20302
    },
    {
      "epoch": 1.1369134281554485,
      "grad_norm": 1.266954779624939,
      "learning_rate": 9.992289473684211e-05,
      "loss": 0.4814,
      "step": 20303
    },
    {
      "epoch": 1.1369694254675775,
      "grad_norm": 1.6034536361694336,
      "learning_rate": 9.992263157894737e-05,
      "loss": 0.5257,
      "step": 20304
    },
    {
      "epoch": 1.1370254227797065,
      "grad_norm": 1.2043932676315308,
      "learning_rate": 9.992236842105264e-05,
      "loss": 0.4585,
      "step": 20305
    },
    {
      "epoch": 1.1370814200918355,
      "grad_norm": 1.4146957397460938,
      "learning_rate": 9.99221052631579e-05,
      "loss": 0.4716,
      "step": 20306
    },
    {
      "epoch": 1.1371374174039646,
      "grad_norm": 1.373159646987915,
      "learning_rate": 9.992184210526316e-05,
      "loss": 0.3746,
      "step": 20307
    },
    {
      "epoch": 1.1371934147160936,
      "grad_norm": 1.1979814767837524,
      "learning_rate": 9.992157894736842e-05,
      "loss": 0.3212,
      "step": 20308
    },
    {
      "epoch": 1.1372494120282226,
      "grad_norm": 1.7314233779907227,
      "learning_rate": 9.99213157894737e-05,
      "loss": 0.3733,
      "step": 20309
    },
    {
      "epoch": 1.1373054093403516,
      "grad_norm": 1.1924324035644531,
      "learning_rate": 9.992105263157895e-05,
      "loss": 0.4887,
      "step": 20310
    },
    {
      "epoch": 1.1373614066524806,
      "grad_norm": 1.3776707649230957,
      "learning_rate": 9.992078947368421e-05,
      "loss": 0.3593,
      "step": 20311
    },
    {
      "epoch": 1.1374174039646097,
      "grad_norm": 1.463600754737854,
      "learning_rate": 9.992052631578947e-05,
      "loss": 0.4736,
      "step": 20312
    },
    {
      "epoch": 1.1374734012767387,
      "grad_norm": 1.376007080078125,
      "learning_rate": 9.992026315789475e-05,
      "loss": 0.4454,
      "step": 20313
    },
    {
      "epoch": 1.1375293985888677,
      "grad_norm": 1.5093474388122559,
      "learning_rate": 9.992e-05,
      "loss": 0.4242,
      "step": 20314
    },
    {
      "epoch": 1.1375853959009967,
      "grad_norm": 1.3137797117233276,
      "learning_rate": 9.991973684210527e-05,
      "loss": 0.4289,
      "step": 20315
    },
    {
      "epoch": 1.1376413932131257,
      "grad_norm": 1.4704359769821167,
      "learning_rate": 9.991947368421053e-05,
      "loss": 0.4248,
      "step": 20316
    },
    {
      "epoch": 1.1376973905252548,
      "grad_norm": 1.3900424242019653,
      "learning_rate": 9.99192105263158e-05,
      "loss": 0.412,
      "step": 20317
    },
    {
      "epoch": 1.1377533878373838,
      "grad_norm": 1.325015664100647,
      "learning_rate": 9.991894736842106e-05,
      "loss": 0.5808,
      "step": 20318
    },
    {
      "epoch": 1.1378093851495128,
      "grad_norm": 1.2525914907455444,
      "learning_rate": 9.991868421052633e-05,
      "loss": 0.5415,
      "step": 20319
    },
    {
      "epoch": 1.1378653824616418,
      "grad_norm": 1.4956260919570923,
      "learning_rate": 9.991842105263158e-05,
      "loss": 0.4928,
      "step": 20320
    },
    {
      "epoch": 1.1379213797737708,
      "grad_norm": 1.1873290538787842,
      "learning_rate": 9.991815789473685e-05,
      "loss": 0.3795,
      "step": 20321
    },
    {
      "epoch": 1.1379773770858999,
      "grad_norm": 1.2872254848480225,
      "learning_rate": 9.991789473684211e-05,
      "loss": 0.4509,
      "step": 20322
    },
    {
      "epoch": 1.1380333743980289,
      "grad_norm": 1.3473589420318604,
      "learning_rate": 9.991763157894737e-05,
      "loss": 0.343,
      "step": 20323
    },
    {
      "epoch": 1.138089371710158,
      "grad_norm": 1.6654990911483765,
      "learning_rate": 9.991736842105264e-05,
      "loss": 0.5751,
      "step": 20324
    },
    {
      "epoch": 1.138145369022287,
      "grad_norm": 1.8152382373809814,
      "learning_rate": 9.991710526315789e-05,
      "loss": 0.3431,
      "step": 20325
    },
    {
      "epoch": 1.138201366334416,
      "grad_norm": 1.5003809928894043,
      "learning_rate": 9.991684210526316e-05,
      "loss": 0.4206,
      "step": 20326
    },
    {
      "epoch": 1.138257363646545,
      "grad_norm": 1.2405860424041748,
      "learning_rate": 9.991657894736842e-05,
      "loss": 0.3798,
      "step": 20327
    },
    {
      "epoch": 1.138313360958674,
      "grad_norm": 1.306026577949524,
      "learning_rate": 9.99163157894737e-05,
      "loss": 0.4388,
      "step": 20328
    },
    {
      "epoch": 1.138369358270803,
      "grad_norm": 1.517681360244751,
      "learning_rate": 9.991605263157896e-05,
      "loss": 0.5016,
      "step": 20329
    },
    {
      "epoch": 1.138425355582932,
      "grad_norm": 2.5119831562042236,
      "learning_rate": 9.991578947368422e-05,
      "loss": 0.4232,
      "step": 20330
    },
    {
      "epoch": 1.138481352895061,
      "grad_norm": 1.238312005996704,
      "learning_rate": 9.991552631578948e-05,
      "loss": 0.3336,
      "step": 20331
    },
    {
      "epoch": 1.13853735020719,
      "grad_norm": 1.3370847702026367,
      "learning_rate": 9.991526315789475e-05,
      "loss": 0.474,
      "step": 20332
    },
    {
      "epoch": 1.138593347519319,
      "grad_norm": 3.627495527267456,
      "learning_rate": 9.991500000000001e-05,
      "loss": 0.5222,
      "step": 20333
    },
    {
      "epoch": 1.138649344831448,
      "grad_norm": 1.314353585243225,
      "learning_rate": 9.991473684210527e-05,
      "loss": 0.4736,
      "step": 20334
    },
    {
      "epoch": 1.138705342143577,
      "grad_norm": 1.693156361579895,
      "learning_rate": 9.991447368421053e-05,
      "loss": 0.5348,
      "step": 20335
    },
    {
      "epoch": 1.1387613394557061,
      "grad_norm": 5.299622535705566,
      "learning_rate": 9.99142105263158e-05,
      "loss": 0.4973,
      "step": 20336
    },
    {
      "epoch": 1.1388173367678351,
      "grad_norm": 1.2720335721969604,
      "learning_rate": 9.991394736842106e-05,
      "loss": 0.3379,
      "step": 20337
    },
    {
      "epoch": 1.1388733340799642,
      "grad_norm": 1.6211950778961182,
      "learning_rate": 9.991368421052632e-05,
      "loss": 0.4027,
      "step": 20338
    },
    {
      "epoch": 1.1389293313920932,
      "grad_norm": 1.5061919689178467,
      "learning_rate": 9.991342105263158e-05,
      "loss": 0.6308,
      "step": 20339
    },
    {
      "epoch": 1.1389853287042222,
      "grad_norm": 1.3513188362121582,
      "learning_rate": 9.991315789473684e-05,
      "loss": 0.4841,
      "step": 20340
    },
    {
      "epoch": 1.1390413260163512,
      "grad_norm": 1.410709023475647,
      "learning_rate": 9.991289473684211e-05,
      "loss": 0.3373,
      "step": 20341
    },
    {
      "epoch": 1.1390973233284802,
      "grad_norm": 1.2556793689727783,
      "learning_rate": 9.991263157894737e-05,
      "loss": 0.4272,
      "step": 20342
    },
    {
      "epoch": 1.1391533206406093,
      "grad_norm": 1.5008658170700073,
      "learning_rate": 9.991236842105263e-05,
      "loss": 0.5107,
      "step": 20343
    },
    {
      "epoch": 1.1392093179527383,
      "grad_norm": 1.5143998861312866,
      "learning_rate": 9.991210526315789e-05,
      "loss": 0.6184,
      "step": 20344
    },
    {
      "epoch": 1.1392653152648673,
      "grad_norm": 1.224623441696167,
      "learning_rate": 9.991184210526317e-05,
      "loss": 0.4284,
      "step": 20345
    },
    {
      "epoch": 1.1393213125769963,
      "grad_norm": 1.4190151691436768,
      "learning_rate": 9.991157894736843e-05,
      "loss": 0.4862,
      "step": 20346
    },
    {
      "epoch": 1.1393773098891253,
      "grad_norm": 1.2998725175857544,
      "learning_rate": 9.991131578947368e-05,
      "loss": 0.4061,
      "step": 20347
    },
    {
      "epoch": 1.1394333072012544,
      "grad_norm": 1.2712948322296143,
      "learning_rate": 9.991105263157894e-05,
      "loss": 0.4702,
      "step": 20348
    },
    {
      "epoch": 1.1394893045133834,
      "grad_norm": 1.305171012878418,
      "learning_rate": 9.991078947368422e-05,
      "loss": 0.3515,
      "step": 20349
    },
    {
      "epoch": 1.1395453018255124,
      "grad_norm": 1.5488324165344238,
      "learning_rate": 9.991052631578948e-05,
      "loss": 0.4834,
      "step": 20350
    },
    {
      "epoch": 1.1396012991376414,
      "grad_norm": 1.2901452779769897,
      "learning_rate": 9.991026315789475e-05,
      "loss": 0.4051,
      "step": 20351
    },
    {
      "epoch": 1.1396572964497704,
      "grad_norm": 1.3085746765136719,
      "learning_rate": 9.991e-05,
      "loss": 0.4558,
      "step": 20352
    },
    {
      "epoch": 1.1397132937618994,
      "grad_norm": 1.3028777837753296,
      "learning_rate": 9.990973684210527e-05,
      "loss": 0.5685,
      "step": 20353
    },
    {
      "epoch": 1.1397692910740285,
      "grad_norm": 1.3169550895690918,
      "learning_rate": 9.990947368421053e-05,
      "loss": 0.3991,
      "step": 20354
    },
    {
      "epoch": 1.1398252883861575,
      "grad_norm": 1.3387531042099,
      "learning_rate": 9.99092105263158e-05,
      "loss": 0.3692,
      "step": 20355
    },
    {
      "epoch": 1.1398812856982865,
      "grad_norm": 1.3168439865112305,
      "learning_rate": 9.990894736842106e-05,
      "loss": 0.3662,
      "step": 20356
    },
    {
      "epoch": 1.1399372830104155,
      "grad_norm": 1.3575280904769897,
      "learning_rate": 9.990868421052631e-05,
      "loss": 0.4071,
      "step": 20357
    },
    {
      "epoch": 1.1399932803225445,
      "grad_norm": 1.6750175952911377,
      "learning_rate": 9.990842105263158e-05,
      "loss": 0.6067,
      "step": 20358
    },
    {
      "epoch": 1.1400492776346736,
      "grad_norm": 1.2230628728866577,
      "learning_rate": 9.990815789473684e-05,
      "loss": 0.4325,
      "step": 20359
    },
    {
      "epoch": 1.1401052749468026,
      "grad_norm": 1.4564143419265747,
      "learning_rate": 9.990789473684212e-05,
      "loss": 0.3646,
      "step": 20360
    },
    {
      "epoch": 1.1401612722589316,
      "grad_norm": 1.7707886695861816,
      "learning_rate": 9.990763157894738e-05,
      "loss": 0.3448,
      "step": 20361
    },
    {
      "epoch": 1.1402172695710606,
      "grad_norm": 1.2103395462036133,
      "learning_rate": 9.990736842105264e-05,
      "loss": 0.4444,
      "step": 20362
    },
    {
      "epoch": 1.1402732668831896,
      "grad_norm": 1.2842963933944702,
      "learning_rate": 9.99071052631579e-05,
      "loss": 0.5412,
      "step": 20363
    },
    {
      "epoch": 1.1403292641953187,
      "grad_norm": 1.1807801723480225,
      "learning_rate": 9.990684210526317e-05,
      "loss": 0.4317,
      "step": 20364
    },
    {
      "epoch": 1.1403852615074477,
      "grad_norm": 1.2823046445846558,
      "learning_rate": 9.990657894736843e-05,
      "loss": 0.5272,
      "step": 20365
    },
    {
      "epoch": 1.1404412588195767,
      "grad_norm": 1.4724619388580322,
      "learning_rate": 9.990631578947369e-05,
      "loss": 0.4194,
      "step": 20366
    },
    {
      "epoch": 1.1404972561317057,
      "grad_norm": 1.0696430206298828,
      "learning_rate": 9.990605263157895e-05,
      "loss": 0.311,
      "step": 20367
    },
    {
      "epoch": 1.1405532534438347,
      "grad_norm": 1.2921104431152344,
      "learning_rate": 9.990578947368422e-05,
      "loss": 0.434,
      "step": 20368
    },
    {
      "epoch": 1.1406092507559638,
      "grad_norm": 1.2019782066345215,
      "learning_rate": 9.990552631578948e-05,
      "loss": 0.3851,
      "step": 20369
    },
    {
      "epoch": 1.1406652480680928,
      "grad_norm": 1.3826487064361572,
      "learning_rate": 9.990526315789474e-05,
      "loss": 0.5371,
      "step": 20370
    },
    {
      "epoch": 1.1407212453802218,
      "grad_norm": 1.5002570152282715,
      "learning_rate": 9.9905e-05,
      "loss": 0.357,
      "step": 20371
    },
    {
      "epoch": 1.1407772426923508,
      "grad_norm": 1.426370620727539,
      "learning_rate": 9.990473684210527e-05,
      "loss": 0.5019,
      "step": 20372
    },
    {
      "epoch": 1.1408332400044798,
      "grad_norm": 1.5056532621383667,
      "learning_rate": 9.990447368421053e-05,
      "loss": 0.5796,
      "step": 20373
    },
    {
      "epoch": 1.1408892373166089,
      "grad_norm": 1.2421802282333374,
      "learning_rate": 9.990421052631579e-05,
      "loss": 0.3594,
      "step": 20374
    },
    {
      "epoch": 1.1409452346287379,
      "grad_norm": 1.8825669288635254,
      "learning_rate": 9.990394736842105e-05,
      "loss": 0.4201,
      "step": 20375
    },
    {
      "epoch": 1.141001231940867,
      "grad_norm": 1.3323009014129639,
      "learning_rate": 9.990368421052631e-05,
      "loss": 0.4136,
      "step": 20376
    },
    {
      "epoch": 1.141057229252996,
      "grad_norm": 1.405066728591919,
      "learning_rate": 9.990342105263159e-05,
      "loss": 0.4292,
      "step": 20377
    },
    {
      "epoch": 1.141113226565125,
      "grad_norm": 1.2221298217773438,
      "learning_rate": 9.990315789473684e-05,
      "loss": 0.3495,
      "step": 20378
    },
    {
      "epoch": 1.141169223877254,
      "grad_norm": 1.2865514755249023,
      "learning_rate": 9.990289473684212e-05,
      "loss": 0.3648,
      "step": 20379
    },
    {
      "epoch": 1.141225221189383,
      "grad_norm": 1.406097173690796,
      "learning_rate": 9.990263157894736e-05,
      "loss": 0.428,
      "step": 20380
    },
    {
      "epoch": 1.141281218501512,
      "grad_norm": 2.4247918128967285,
      "learning_rate": 9.990236842105264e-05,
      "loss": 0.4855,
      "step": 20381
    },
    {
      "epoch": 1.141337215813641,
      "grad_norm": 1.4736703634262085,
      "learning_rate": 9.99021052631579e-05,
      "loss": 0.4213,
      "step": 20382
    },
    {
      "epoch": 1.14139321312577,
      "grad_norm": 1.3291274309158325,
      "learning_rate": 9.990184210526317e-05,
      "loss": 0.3984,
      "step": 20383
    },
    {
      "epoch": 1.141449210437899,
      "grad_norm": 1.3679965734481812,
      "learning_rate": 9.990157894736843e-05,
      "loss": 0.4346,
      "step": 20384
    },
    {
      "epoch": 1.141505207750028,
      "grad_norm": 1.220797061920166,
      "learning_rate": 9.990131578947369e-05,
      "loss": 0.4462,
      "step": 20385
    },
    {
      "epoch": 1.141561205062157,
      "grad_norm": 1.4401963949203491,
      "learning_rate": 9.990105263157895e-05,
      "loss": 0.357,
      "step": 20386
    },
    {
      "epoch": 1.141617202374286,
      "grad_norm": 1.3059804439544678,
      "learning_rate": 9.990078947368422e-05,
      "loss": 0.4317,
      "step": 20387
    },
    {
      "epoch": 1.1416731996864151,
      "grad_norm": 2.1954119205474854,
      "learning_rate": 9.990052631578948e-05,
      "loss": 0.576,
      "step": 20388
    },
    {
      "epoch": 1.1417291969985441,
      "grad_norm": 1.312447190284729,
      "learning_rate": 9.990026315789474e-05,
      "loss": 0.4316,
      "step": 20389
    },
    {
      "epoch": 1.1417851943106732,
      "grad_norm": 1.3905044794082642,
      "learning_rate": 9.99e-05,
      "loss": 0.6226,
      "step": 20390
    },
    {
      "epoch": 1.1418411916228022,
      "grad_norm": 1.4395277500152588,
      "learning_rate": 9.989973684210526e-05,
      "loss": 0.6674,
      "step": 20391
    },
    {
      "epoch": 1.1418971889349312,
      "grad_norm": 1.7190442085266113,
      "learning_rate": 9.989947368421054e-05,
      "loss": 0.4215,
      "step": 20392
    },
    {
      "epoch": 1.1419531862470602,
      "grad_norm": 1.292906641960144,
      "learning_rate": 9.98992105263158e-05,
      "loss": 0.4206,
      "step": 20393
    },
    {
      "epoch": 1.1420091835591892,
      "grad_norm": 1.3469399213790894,
      "learning_rate": 9.989894736842105e-05,
      "loss": 0.3768,
      "step": 20394
    },
    {
      "epoch": 1.1420651808713183,
      "grad_norm": 2.9747273921966553,
      "learning_rate": 9.989868421052631e-05,
      "loss": 0.4534,
      "step": 20395
    },
    {
      "epoch": 1.1421211781834473,
      "grad_norm": 1.2159113883972168,
      "learning_rate": 9.989842105263159e-05,
      "loss": 0.4651,
      "step": 20396
    },
    {
      "epoch": 1.1421771754955763,
      "grad_norm": 1.318577527999878,
      "learning_rate": 9.989815789473685e-05,
      "loss": 0.412,
      "step": 20397
    },
    {
      "epoch": 1.1422331728077053,
      "grad_norm": 1.2219117879867554,
      "learning_rate": 9.989789473684211e-05,
      "loss": 0.4052,
      "step": 20398
    },
    {
      "epoch": 1.1422891701198343,
      "grad_norm": 1.5057100057601929,
      "learning_rate": 9.989763157894737e-05,
      "loss": 0.4254,
      "step": 20399
    },
    {
      "epoch": 1.1423451674319633,
      "grad_norm": 1.3558597564697266,
      "learning_rate": 9.989736842105264e-05,
      "loss": 0.4651,
      "step": 20400
    },
    {
      "epoch": 1.1424011647440924,
      "grad_norm": 9.762716293334961,
      "learning_rate": 9.98971052631579e-05,
      "loss": 0.4503,
      "step": 20401
    },
    {
      "epoch": 1.1424571620562214,
      "grad_norm": 1.5793808698654175,
      "learning_rate": 9.989684210526316e-05,
      "loss": 0.4069,
      "step": 20402
    },
    {
      "epoch": 1.1425131593683504,
      "grad_norm": 1.2100074291229248,
      "learning_rate": 9.989657894736842e-05,
      "loss": 0.33,
      "step": 20403
    },
    {
      "epoch": 1.1425691566804794,
      "grad_norm": 1.594765543937683,
      "learning_rate": 9.989631578947369e-05,
      "loss": 0.5182,
      "step": 20404
    },
    {
      "epoch": 1.1426251539926084,
      "grad_norm": 1.6750227212905884,
      "learning_rate": 9.989605263157895e-05,
      "loss": 0.4615,
      "step": 20405
    },
    {
      "epoch": 1.1426811513047375,
      "grad_norm": 1.3448501825332642,
      "learning_rate": 9.989578947368423e-05,
      "loss": 0.512,
      "step": 20406
    },
    {
      "epoch": 1.1427371486168665,
      "grad_norm": 1.508351445198059,
      "learning_rate": 9.989552631578947e-05,
      "loss": 0.5387,
      "step": 20407
    },
    {
      "epoch": 1.1427931459289955,
      "grad_norm": 1.4007755517959595,
      "learning_rate": 9.989526315789473e-05,
      "loss": 0.4336,
      "step": 20408
    },
    {
      "epoch": 1.1428491432411245,
      "grad_norm": 1.5024051666259766,
      "learning_rate": 9.9895e-05,
      "loss": 0.3761,
      "step": 20409
    },
    {
      "epoch": 1.1429051405532535,
      "grad_norm": 1.227728009223938,
      "learning_rate": 9.989473684210526e-05,
      "loss": 0.5032,
      "step": 20410
    },
    {
      "epoch": 1.1429611378653823,
      "grad_norm": 1.4767719507217407,
      "learning_rate": 9.989447368421054e-05,
      "loss": 0.4621,
      "step": 20411
    },
    {
      "epoch": 1.1430171351775114,
      "grad_norm": 1.2244586944580078,
      "learning_rate": 9.989421052631578e-05,
      "loss": 0.4291,
      "step": 20412
    },
    {
      "epoch": 1.1430731324896404,
      "grad_norm": 1.3493458032608032,
      "learning_rate": 9.989394736842106e-05,
      "loss": 0.3902,
      "step": 20413
    },
    {
      "epoch": 1.1431291298017694,
      "grad_norm": 1.400938868522644,
      "learning_rate": 9.989368421052632e-05,
      "loss": 0.421,
      "step": 20414
    },
    {
      "epoch": 1.1431851271138984,
      "grad_norm": 1.3430699110031128,
      "learning_rate": 9.989342105263159e-05,
      "loss": 0.429,
      "step": 20415
    },
    {
      "epoch": 1.1432411244260274,
      "grad_norm": 1.5283825397491455,
      "learning_rate": 9.989315789473685e-05,
      "loss": 0.5953,
      "step": 20416
    },
    {
      "epoch": 1.1432971217381565,
      "grad_norm": 1.3737848997116089,
      "learning_rate": 9.989289473684211e-05,
      "loss": 0.4791,
      "step": 20417
    },
    {
      "epoch": 1.1433531190502855,
      "grad_norm": 1.2857928276062012,
      "learning_rate": 9.989263157894737e-05,
      "loss": 0.5354,
      "step": 20418
    },
    {
      "epoch": 1.1434091163624145,
      "grad_norm": 1.208808422088623,
      "learning_rate": 9.989236842105264e-05,
      "loss": 0.4551,
      "step": 20419
    },
    {
      "epoch": 1.1434651136745435,
      "grad_norm": 1.372432827949524,
      "learning_rate": 9.98921052631579e-05,
      "loss": 0.3604,
      "step": 20420
    },
    {
      "epoch": 1.1435211109866725,
      "grad_norm": 1.3018593788146973,
      "learning_rate": 9.989184210526316e-05,
      "loss": 0.3871,
      "step": 20421
    },
    {
      "epoch": 1.1435771082988015,
      "grad_norm": 1.326655387878418,
      "learning_rate": 9.989157894736842e-05,
      "loss": 0.3472,
      "step": 20422
    },
    {
      "epoch": 1.1436331056109306,
      "grad_norm": 1.8529084920883179,
      "learning_rate": 9.98913157894737e-05,
      "loss": 0.5448,
      "step": 20423
    },
    {
      "epoch": 1.1436891029230596,
      "grad_norm": 1.4630001783370972,
      "learning_rate": 9.989105263157896e-05,
      "loss": 0.4667,
      "step": 20424
    },
    {
      "epoch": 1.1437451002351886,
      "grad_norm": 1.5406907796859741,
      "learning_rate": 9.989078947368421e-05,
      "loss": 0.4948,
      "step": 20425
    },
    {
      "epoch": 1.1438010975473176,
      "grad_norm": 1.7303768396377563,
      "learning_rate": 9.989052631578947e-05,
      "loss": 0.4362,
      "step": 20426
    },
    {
      "epoch": 1.1438570948594466,
      "grad_norm": 1.3066902160644531,
      "learning_rate": 9.989026315789473e-05,
      "loss": 0.4374,
      "step": 20427
    },
    {
      "epoch": 1.1439130921715757,
      "grad_norm": 1.3579692840576172,
      "learning_rate": 9.989000000000001e-05,
      "loss": 0.4402,
      "step": 20428
    },
    {
      "epoch": 1.1439690894837047,
      "grad_norm": 1.7817784547805786,
      "learning_rate": 9.988973684210527e-05,
      "loss": 0.5121,
      "step": 20429
    },
    {
      "epoch": 1.1440250867958337,
      "grad_norm": 1.3389939069747925,
      "learning_rate": 9.988947368421053e-05,
      "loss": 0.4115,
      "step": 20430
    },
    {
      "epoch": 1.1440810841079627,
      "grad_norm": 1.320111632347107,
      "learning_rate": 9.988921052631579e-05,
      "loss": 0.4343,
      "step": 20431
    },
    {
      "epoch": 1.1441370814200917,
      "grad_norm": 1.242687463760376,
      "learning_rate": 9.988894736842106e-05,
      "loss": 0.3491,
      "step": 20432
    },
    {
      "epoch": 1.1441930787322208,
      "grad_norm": 1.3836536407470703,
      "learning_rate": 9.988868421052632e-05,
      "loss": 0.3486,
      "step": 20433
    },
    {
      "epoch": 1.1442490760443498,
      "grad_norm": 1.419258952140808,
      "learning_rate": 9.988842105263159e-05,
      "loss": 0.4177,
      "step": 20434
    },
    {
      "epoch": 1.1443050733564788,
      "grad_norm": 1.140465259552002,
      "learning_rate": 9.988815789473684e-05,
      "loss": 0.4119,
      "step": 20435
    },
    {
      "epoch": 1.1443610706686078,
      "grad_norm": 1.4027020931243896,
      "learning_rate": 9.988789473684211e-05,
      "loss": 0.3921,
      "step": 20436
    },
    {
      "epoch": 1.1444170679807368,
      "grad_norm": 1.3744571208953857,
      "learning_rate": 9.988763157894737e-05,
      "loss": 0.4064,
      "step": 20437
    },
    {
      "epoch": 1.1444730652928659,
      "grad_norm": 1.6248918771743774,
      "learning_rate": 9.988736842105265e-05,
      "loss": 0.5104,
      "step": 20438
    },
    {
      "epoch": 1.1445290626049949,
      "grad_norm": 1.2133638858795166,
      "learning_rate": 9.98871052631579e-05,
      "loss": 0.3985,
      "step": 20439
    },
    {
      "epoch": 1.144585059917124,
      "grad_norm": 1.2253282070159912,
      "learning_rate": 9.988684210526316e-05,
      "loss": 0.4253,
      "step": 20440
    },
    {
      "epoch": 1.144641057229253,
      "grad_norm": 1.3476512432098389,
      "learning_rate": 9.988657894736842e-05,
      "loss": 0.389,
      "step": 20441
    },
    {
      "epoch": 1.144697054541382,
      "grad_norm": 1.5335700511932373,
      "learning_rate": 9.98863157894737e-05,
      "loss": 0.3119,
      "step": 20442
    },
    {
      "epoch": 1.144753051853511,
      "grad_norm": 1.3555032014846802,
      "learning_rate": 9.988605263157896e-05,
      "loss": 0.5063,
      "step": 20443
    },
    {
      "epoch": 1.14480904916564,
      "grad_norm": 1.432375431060791,
      "learning_rate": 9.98857894736842e-05,
      "loss": 0.5124,
      "step": 20444
    },
    {
      "epoch": 1.144865046477769,
      "grad_norm": 1.4518640041351318,
      "learning_rate": 9.988552631578948e-05,
      "loss": 0.426,
      "step": 20445
    },
    {
      "epoch": 1.144921043789898,
      "grad_norm": 1.2916922569274902,
      "learning_rate": 9.988526315789474e-05,
      "loss": 0.4223,
      "step": 20446
    },
    {
      "epoch": 1.144977041102027,
      "grad_norm": 1.2688654661178589,
      "learning_rate": 9.988500000000001e-05,
      "loss": 0.3681,
      "step": 20447
    },
    {
      "epoch": 1.145033038414156,
      "grad_norm": 2.6093363761901855,
      "learning_rate": 9.988473684210527e-05,
      "loss": 0.4633,
      "step": 20448
    },
    {
      "epoch": 1.145089035726285,
      "grad_norm": 1.2063078880310059,
      "learning_rate": 9.988447368421053e-05,
      "loss": 0.3911,
      "step": 20449
    },
    {
      "epoch": 1.145145033038414,
      "grad_norm": 1.247310757637024,
      "learning_rate": 9.988421052631579e-05,
      "loss": 0.3752,
      "step": 20450
    },
    {
      "epoch": 1.145201030350543,
      "grad_norm": 1.5782006978988647,
      "learning_rate": 9.988394736842106e-05,
      "loss": 0.4034,
      "step": 20451
    },
    {
      "epoch": 1.1452570276626721,
      "grad_norm": 1.9347234964370728,
      "learning_rate": 9.988368421052632e-05,
      "loss": 0.5993,
      "step": 20452
    },
    {
      "epoch": 1.1453130249748011,
      "grad_norm": 1.2036774158477783,
      "learning_rate": 9.988342105263158e-05,
      "loss": 0.4467,
      "step": 20453
    },
    {
      "epoch": 1.1453690222869302,
      "grad_norm": 1.3685601949691772,
      "learning_rate": 9.988315789473684e-05,
      "loss": 0.4881,
      "step": 20454
    },
    {
      "epoch": 1.1454250195990592,
      "grad_norm": 2.390435218811035,
      "learning_rate": 9.988289473684212e-05,
      "loss": 0.4933,
      "step": 20455
    },
    {
      "epoch": 1.1454810169111882,
      "grad_norm": 1.2845882177352905,
      "learning_rate": 9.988263157894737e-05,
      "loss": 0.4687,
      "step": 20456
    },
    {
      "epoch": 1.1455370142233172,
      "grad_norm": 1.82760751247406,
      "learning_rate": 9.988236842105263e-05,
      "loss": 0.4768,
      "step": 20457
    },
    {
      "epoch": 1.1455930115354462,
      "grad_norm": 1.1389309167861938,
      "learning_rate": 9.98821052631579e-05,
      "loss": 0.4183,
      "step": 20458
    },
    {
      "epoch": 1.1456490088475753,
      "grad_norm": 1.4520819187164307,
      "learning_rate": 9.988184210526317e-05,
      "loss": 0.3848,
      "step": 20459
    },
    {
      "epoch": 1.1457050061597043,
      "grad_norm": 1.26947021484375,
      "learning_rate": 9.988157894736843e-05,
      "loss": 0.4503,
      "step": 20460
    },
    {
      "epoch": 1.1457610034718333,
      "grad_norm": 1.2558846473693848,
      "learning_rate": 9.988131578947369e-05,
      "loss": 0.3364,
      "step": 20461
    },
    {
      "epoch": 1.1458170007839623,
      "grad_norm": 1.8046354055404663,
      "learning_rate": 9.988105263157895e-05,
      "loss": 0.6396,
      "step": 20462
    },
    {
      "epoch": 1.1458729980960913,
      "grad_norm": 1.3201208114624023,
      "learning_rate": 9.98807894736842e-05,
      "loss": 0.4427,
      "step": 20463
    },
    {
      "epoch": 1.1459289954082204,
      "grad_norm": 1.6858083009719849,
      "learning_rate": 9.988052631578948e-05,
      "loss": 0.5229,
      "step": 20464
    },
    {
      "epoch": 1.1459849927203494,
      "grad_norm": 1.1688151359558105,
      "learning_rate": 9.988026315789474e-05,
      "loss": 0.43,
      "step": 20465
    },
    {
      "epoch": 1.1460409900324784,
      "grad_norm": 1.6450614929199219,
      "learning_rate": 9.988000000000001e-05,
      "loss": 0.4932,
      "step": 20466
    },
    {
      "epoch": 1.1460969873446074,
      "grad_norm": 1.4939018487930298,
      "learning_rate": 9.987973684210526e-05,
      "loss": 0.4432,
      "step": 20467
    },
    {
      "epoch": 1.1461529846567364,
      "grad_norm": 1.5850712060928345,
      "learning_rate": 9.987947368421053e-05,
      "loss": 0.4831,
      "step": 20468
    },
    {
      "epoch": 1.1462089819688654,
      "grad_norm": 1.2346611022949219,
      "learning_rate": 9.987921052631579e-05,
      "loss": 0.4935,
      "step": 20469
    },
    {
      "epoch": 1.1462649792809945,
      "grad_norm": 1.5039423704147339,
      "learning_rate": 9.987894736842107e-05,
      "loss": 0.4986,
      "step": 20470
    },
    {
      "epoch": 1.1463209765931235,
      "grad_norm": 1.4204386472702026,
      "learning_rate": 9.987868421052632e-05,
      "loss": 0.3778,
      "step": 20471
    },
    {
      "epoch": 1.1463769739052525,
      "grad_norm": 1.4285898208618164,
      "learning_rate": 9.987842105263158e-05,
      "loss": 0.4692,
      "step": 20472
    },
    {
      "epoch": 1.1464329712173815,
      "grad_norm": 1.6953669786453247,
      "learning_rate": 9.987815789473684e-05,
      "loss": 0.3309,
      "step": 20473
    },
    {
      "epoch": 1.1464889685295105,
      "grad_norm": 1.5123811960220337,
      "learning_rate": 9.987789473684212e-05,
      "loss": 0.4395,
      "step": 20474
    },
    {
      "epoch": 1.1465449658416396,
      "grad_norm": 1.2125428915023804,
      "learning_rate": 9.987763157894738e-05,
      "loss": 0.3578,
      "step": 20475
    },
    {
      "epoch": 1.1466009631537686,
      "grad_norm": 1.2554445266723633,
      "learning_rate": 9.987736842105264e-05,
      "loss": 0.41,
      "step": 20476
    },
    {
      "epoch": 1.1466569604658976,
      "grad_norm": 1.1073169708251953,
      "learning_rate": 9.98771052631579e-05,
      "loss": 0.3199,
      "step": 20477
    },
    {
      "epoch": 1.1467129577780266,
      "grad_norm": 1.5877275466918945,
      "learning_rate": 9.987684210526316e-05,
      "loss": 0.4738,
      "step": 20478
    },
    {
      "epoch": 1.1467689550901556,
      "grad_norm": 1.2053771018981934,
      "learning_rate": 9.987657894736843e-05,
      "loss": 0.4658,
      "step": 20479
    },
    {
      "epoch": 1.1468249524022847,
      "grad_norm": 1.2012308835983276,
      "learning_rate": 9.987631578947369e-05,
      "loss": 0.4932,
      "step": 20480
    },
    {
      "epoch": 1.1468809497144137,
      "grad_norm": 1.4577291011810303,
      "learning_rate": 9.987605263157895e-05,
      "loss": 0.4037,
      "step": 20481
    },
    {
      "epoch": 1.1469369470265427,
      "grad_norm": 1.980048656463623,
      "learning_rate": 9.987578947368421e-05,
      "loss": 0.5397,
      "step": 20482
    },
    {
      "epoch": 1.1469929443386717,
      "grad_norm": 1.3027416467666626,
      "learning_rate": 9.987552631578948e-05,
      "loss": 0.4341,
      "step": 20483
    },
    {
      "epoch": 1.1470489416508007,
      "grad_norm": 1.3045016527175903,
      "learning_rate": 9.987526315789474e-05,
      "loss": 0.4104,
      "step": 20484
    },
    {
      "epoch": 1.1471049389629298,
      "grad_norm": 1.537992000579834,
      "learning_rate": 9.9875e-05,
      "loss": 0.554,
      "step": 20485
    },
    {
      "epoch": 1.1471609362750588,
      "grad_norm": 1.3715176582336426,
      "learning_rate": 9.987473684210526e-05,
      "loss": 0.4554,
      "step": 20486
    },
    {
      "epoch": 1.1472169335871878,
      "grad_norm": 1.564469337463379,
      "learning_rate": 9.987447368421053e-05,
      "loss": 0.4798,
      "step": 20487
    },
    {
      "epoch": 1.1472729308993168,
      "grad_norm": 1.3704088926315308,
      "learning_rate": 9.98742105263158e-05,
      "loss": 0.4266,
      "step": 20488
    },
    {
      "epoch": 1.1473289282114458,
      "grad_norm": 1.655909538269043,
      "learning_rate": 9.987394736842107e-05,
      "loss": 0.5037,
      "step": 20489
    },
    {
      "epoch": 1.1473849255235749,
      "grad_norm": 1.3380017280578613,
      "learning_rate": 9.987368421052631e-05,
      "loss": 0.5295,
      "step": 20490
    },
    {
      "epoch": 1.1474409228357039,
      "grad_norm": 1.4491338729858398,
      "learning_rate": 9.987342105263159e-05,
      "loss": 0.3792,
      "step": 20491
    },
    {
      "epoch": 1.1474969201478329,
      "grad_norm": 1.2327592372894287,
      "learning_rate": 9.987315789473685e-05,
      "loss": 0.3807,
      "step": 20492
    },
    {
      "epoch": 1.147552917459962,
      "grad_norm": 1.2314577102661133,
      "learning_rate": 9.987289473684212e-05,
      "loss": 0.3485,
      "step": 20493
    },
    {
      "epoch": 1.147608914772091,
      "grad_norm": 1.2795847654342651,
      "learning_rate": 9.987263157894738e-05,
      "loss": 0.4447,
      "step": 20494
    },
    {
      "epoch": 1.14766491208422,
      "grad_norm": 1.2632372379302979,
      "learning_rate": 9.987236842105263e-05,
      "loss": 0.4008,
      "step": 20495
    },
    {
      "epoch": 1.147720909396349,
      "grad_norm": 1.2883623838424683,
      "learning_rate": 9.98721052631579e-05,
      "loss": 0.4505,
      "step": 20496
    },
    {
      "epoch": 1.147776906708478,
      "grad_norm": 1.2738240957260132,
      "learning_rate": 9.987184210526316e-05,
      "loss": 0.3776,
      "step": 20497
    },
    {
      "epoch": 1.147832904020607,
      "grad_norm": 1.2178534269332886,
      "learning_rate": 9.987157894736843e-05,
      "loss": 0.3194,
      "step": 20498
    },
    {
      "epoch": 1.147888901332736,
      "grad_norm": 1.4835726022720337,
      "learning_rate": 9.987131578947368e-05,
      "loss": 0.383,
      "step": 20499
    },
    {
      "epoch": 1.147944898644865,
      "grad_norm": 1.263297438621521,
      "learning_rate": 9.987105263157895e-05,
      "loss": 0.3836,
      "step": 20500
    },
    {
      "epoch": 1.148000895956994,
      "grad_norm": 1.3925824165344238,
      "learning_rate": 9.987078947368421e-05,
      "loss": 0.4921,
      "step": 20501
    },
    {
      "epoch": 1.148056893269123,
      "grad_norm": 1.2069382667541504,
      "learning_rate": 9.987052631578948e-05,
      "loss": 0.4491,
      "step": 20502
    },
    {
      "epoch": 1.148112890581252,
      "grad_norm": 1.632684588432312,
      "learning_rate": 9.987026315789474e-05,
      "loss": 0.5105,
      "step": 20503
    },
    {
      "epoch": 1.1481688878933811,
      "grad_norm": 1.347937822341919,
      "learning_rate": 9.987e-05,
      "loss": 0.4321,
      "step": 20504
    },
    {
      "epoch": 1.1482248852055101,
      "grad_norm": 1.5664825439453125,
      "learning_rate": 9.986973684210526e-05,
      "loss": 0.4648,
      "step": 20505
    },
    {
      "epoch": 1.1482808825176392,
      "grad_norm": 1.4892882108688354,
      "learning_rate": 9.986947368421054e-05,
      "loss": 0.4631,
      "step": 20506
    },
    {
      "epoch": 1.1483368798297682,
      "grad_norm": 1.3442851305007935,
      "learning_rate": 9.98692105263158e-05,
      "loss": 0.5604,
      "step": 20507
    },
    {
      "epoch": 1.1483928771418972,
      "grad_norm": 1.3856630325317383,
      "learning_rate": 9.986894736842106e-05,
      "loss": 0.3672,
      "step": 20508
    },
    {
      "epoch": 1.1484488744540262,
      "grad_norm": 1.2508807182312012,
      "learning_rate": 9.986868421052632e-05,
      "loss": 0.4337,
      "step": 20509
    },
    {
      "epoch": 1.1485048717661552,
      "grad_norm": 1.9127460718154907,
      "learning_rate": 9.986842105263159e-05,
      "loss": 0.4016,
      "step": 20510
    },
    {
      "epoch": 1.1485608690782843,
      "grad_norm": 1.5735092163085938,
      "learning_rate": 9.986815789473685e-05,
      "loss": 0.5388,
      "step": 20511
    },
    {
      "epoch": 1.1486168663904133,
      "grad_norm": 1.1594423055648804,
      "learning_rate": 9.986789473684211e-05,
      "loss": 0.3916,
      "step": 20512
    },
    {
      "epoch": 1.1486728637025423,
      "grad_norm": 1.6421160697937012,
      "learning_rate": 9.986763157894737e-05,
      "loss": 0.5958,
      "step": 20513
    },
    {
      "epoch": 1.1487288610146713,
      "grad_norm": 1.5584092140197754,
      "learning_rate": 9.986736842105263e-05,
      "loss": 0.4277,
      "step": 20514
    },
    {
      "epoch": 1.1487848583268003,
      "grad_norm": 1.321160078048706,
      "learning_rate": 9.98671052631579e-05,
      "loss": 0.4582,
      "step": 20515
    },
    {
      "epoch": 1.1488408556389293,
      "grad_norm": 2.9948270320892334,
      "learning_rate": 9.986684210526316e-05,
      "loss": 0.4916,
      "step": 20516
    },
    {
      "epoch": 1.1488968529510584,
      "grad_norm": 1.7630518674850464,
      "learning_rate": 9.986657894736842e-05,
      "loss": 0.4982,
      "step": 20517
    },
    {
      "epoch": 1.1489528502631874,
      "grad_norm": 1.5841964483261108,
      "learning_rate": 9.986631578947368e-05,
      "loss": 0.4953,
      "step": 20518
    },
    {
      "epoch": 1.1490088475753164,
      "grad_norm": 1.4199410676956177,
      "learning_rate": 9.986605263157895e-05,
      "loss": 0.3478,
      "step": 20519
    },
    {
      "epoch": 1.1490648448874454,
      "grad_norm": 1.2151843309402466,
      "learning_rate": 9.986578947368421e-05,
      "loss": 0.3648,
      "step": 20520
    },
    {
      "epoch": 1.1491208421995744,
      "grad_norm": 1.4104688167572021,
      "learning_rate": 9.986552631578949e-05,
      "loss": 0.4732,
      "step": 20521
    },
    {
      "epoch": 1.1491768395117035,
      "grad_norm": 1.6134860515594482,
      "learning_rate": 9.986526315789473e-05,
      "loss": 0.4832,
      "step": 20522
    },
    {
      "epoch": 1.1492328368238325,
      "grad_norm": 1.2508913278579712,
      "learning_rate": 9.986500000000001e-05,
      "loss": 0.4126,
      "step": 20523
    },
    {
      "epoch": 1.1492888341359615,
      "grad_norm": 7.151427268981934,
      "learning_rate": 9.986473684210527e-05,
      "loss": 0.5626,
      "step": 20524
    },
    {
      "epoch": 1.1493448314480905,
      "grad_norm": 1.5135068893432617,
      "learning_rate": 9.986447368421054e-05,
      "loss": 0.5287,
      "step": 20525
    },
    {
      "epoch": 1.1494008287602195,
      "grad_norm": 1.3105688095092773,
      "learning_rate": 9.98642105263158e-05,
      "loss": 0.453,
      "step": 20526
    },
    {
      "epoch": 1.1494568260723486,
      "grad_norm": 1.422707438468933,
      "learning_rate": 9.986394736842106e-05,
      "loss": 0.4645,
      "step": 20527
    },
    {
      "epoch": 1.1495128233844776,
      "grad_norm": 1.54299795627594,
      "learning_rate": 9.986368421052632e-05,
      "loss": 0.497,
      "step": 20528
    },
    {
      "epoch": 1.1495688206966066,
      "grad_norm": 1.470493197441101,
      "learning_rate": 9.986342105263158e-05,
      "loss": 0.4095,
      "step": 20529
    },
    {
      "epoch": 1.1496248180087356,
      "grad_norm": 1.6525027751922607,
      "learning_rate": 9.986315789473685e-05,
      "loss": 0.4128,
      "step": 20530
    },
    {
      "epoch": 1.1496808153208646,
      "grad_norm": 1.5500452518463135,
      "learning_rate": 9.986289473684211e-05,
      "loss": 0.3531,
      "step": 20531
    },
    {
      "epoch": 1.1497368126329937,
      "grad_norm": 1.5278432369232178,
      "learning_rate": 9.986263157894737e-05,
      "loss": 0.6385,
      "step": 20532
    },
    {
      "epoch": 1.1497928099451227,
      "grad_norm": 1.5010216236114502,
      "learning_rate": 9.986236842105263e-05,
      "loss": 0.4497,
      "step": 20533
    },
    {
      "epoch": 1.1498488072572517,
      "grad_norm": 1.275917649269104,
      "learning_rate": 9.98621052631579e-05,
      "loss": 0.4798,
      "step": 20534
    },
    {
      "epoch": 1.1499048045693807,
      "grad_norm": 1.313026785850525,
      "learning_rate": 9.986184210526316e-05,
      "loss": 0.4179,
      "step": 20535
    },
    {
      "epoch": 1.1499608018815097,
      "grad_norm": 1.8314224481582642,
      "learning_rate": 9.986157894736842e-05,
      "loss": 0.3713,
      "step": 20536
    },
    {
      "epoch": 1.1500167991936388,
      "grad_norm": 1.3292884826660156,
      "learning_rate": 9.986131578947368e-05,
      "loss": 0.4248,
      "step": 20537
    },
    {
      "epoch": 1.1500727965057678,
      "grad_norm": 1.2726421356201172,
      "learning_rate": 9.986105263157896e-05,
      "loss": 0.3801,
      "step": 20538
    },
    {
      "epoch": 1.1501287938178968,
      "grad_norm": 1.2855806350708008,
      "learning_rate": 9.986078947368422e-05,
      "loss": 0.4511,
      "step": 20539
    },
    {
      "epoch": 1.1501847911300258,
      "grad_norm": 1.3955669403076172,
      "learning_rate": 9.986052631578948e-05,
      "loss": 0.5053,
      "step": 20540
    },
    {
      "epoch": 1.1502407884421548,
      "grad_norm": 1.2548118829727173,
      "learning_rate": 9.986026315789474e-05,
      "loss": 0.3955,
      "step": 20541
    },
    {
      "epoch": 1.1502967857542838,
      "grad_norm": 1.2880767583847046,
      "learning_rate": 9.986000000000001e-05,
      "loss": 0.3904,
      "step": 20542
    },
    {
      "epoch": 1.1503527830664129,
      "grad_norm": 1.262155294418335,
      "learning_rate": 9.985973684210527e-05,
      "loss": 0.4347,
      "step": 20543
    },
    {
      "epoch": 1.1504087803785419,
      "grad_norm": 1.4613949060440063,
      "learning_rate": 9.985947368421054e-05,
      "loss": 0.47,
      "step": 20544
    },
    {
      "epoch": 1.150464777690671,
      "grad_norm": 1.3043702840805054,
      "learning_rate": 9.985921052631579e-05,
      "loss": 0.3711,
      "step": 20545
    },
    {
      "epoch": 1.1505207750028,
      "grad_norm": 1.5589770078659058,
      "learning_rate": 9.985894736842105e-05,
      "loss": 0.3877,
      "step": 20546
    },
    {
      "epoch": 1.150576772314929,
      "grad_norm": 1.381598711013794,
      "learning_rate": 9.985868421052632e-05,
      "loss": 0.4932,
      "step": 20547
    },
    {
      "epoch": 1.150632769627058,
      "grad_norm": 1.4977377653121948,
      "learning_rate": 9.985842105263158e-05,
      "loss": 0.5111,
      "step": 20548
    },
    {
      "epoch": 1.150688766939187,
      "grad_norm": 2.3996832370758057,
      "learning_rate": 9.985815789473684e-05,
      "loss": 0.4615,
      "step": 20549
    },
    {
      "epoch": 1.150744764251316,
      "grad_norm": 1.307060956954956,
      "learning_rate": 9.98578947368421e-05,
      "loss": 0.5004,
      "step": 20550
    },
    {
      "epoch": 1.150800761563445,
      "grad_norm": 1.3533759117126465,
      "learning_rate": 9.985763157894737e-05,
      "loss": 0.3433,
      "step": 20551
    },
    {
      "epoch": 1.150856758875574,
      "grad_norm": 1.4080413579940796,
      "learning_rate": 9.985736842105263e-05,
      "loss": 0.3815,
      "step": 20552
    },
    {
      "epoch": 1.150912756187703,
      "grad_norm": 1.3012210130691528,
      "learning_rate": 9.985710526315791e-05,
      "loss": 0.529,
      "step": 20553
    },
    {
      "epoch": 1.150968753499832,
      "grad_norm": 1.5659222602844238,
      "learning_rate": 9.985684210526315e-05,
      "loss": 0.4356,
      "step": 20554
    },
    {
      "epoch": 1.151024750811961,
      "grad_norm": 1.316042423248291,
      "learning_rate": 9.985657894736843e-05,
      "loss": 0.4283,
      "step": 20555
    },
    {
      "epoch": 1.1510807481240901,
      "grad_norm": 1.5393595695495605,
      "learning_rate": 9.985631578947369e-05,
      "loss": 0.4821,
      "step": 20556
    },
    {
      "epoch": 1.1511367454362191,
      "grad_norm": 1.5968881845474243,
      "learning_rate": 9.985605263157896e-05,
      "loss": 0.5475,
      "step": 20557
    },
    {
      "epoch": 1.1511927427483482,
      "grad_norm": 1.4276247024536133,
      "learning_rate": 9.985578947368422e-05,
      "loss": 0.4981,
      "step": 20558
    },
    {
      "epoch": 1.1512487400604772,
      "grad_norm": 1.418346881866455,
      "learning_rate": 9.985552631578948e-05,
      "loss": 0.4105,
      "step": 20559
    },
    {
      "epoch": 1.1513047373726062,
      "grad_norm": 1.4498636722564697,
      "learning_rate": 9.985526315789474e-05,
      "loss": 0.4128,
      "step": 20560
    },
    {
      "epoch": 1.1513607346847352,
      "grad_norm": 2.0232760906219482,
      "learning_rate": 9.985500000000001e-05,
      "loss": 0.6095,
      "step": 20561
    },
    {
      "epoch": 1.1514167319968642,
      "grad_norm": 1.7056657075881958,
      "learning_rate": 9.985473684210527e-05,
      "loss": 0.3965,
      "step": 20562
    },
    {
      "epoch": 1.1514727293089932,
      "grad_norm": 1.3510469198226929,
      "learning_rate": 9.985447368421053e-05,
      "loss": 0.3811,
      "step": 20563
    },
    {
      "epoch": 1.1515287266211223,
      "grad_norm": 1.7105083465576172,
      "learning_rate": 9.985421052631579e-05,
      "loss": 0.4305,
      "step": 20564
    },
    {
      "epoch": 1.1515847239332513,
      "grad_norm": 1.1439239978790283,
      "learning_rate": 9.985394736842105e-05,
      "loss": 0.3628,
      "step": 20565
    },
    {
      "epoch": 1.1516407212453803,
      "grad_norm": 1.3918046951293945,
      "learning_rate": 9.985368421052632e-05,
      "loss": 0.4256,
      "step": 20566
    },
    {
      "epoch": 1.1516967185575093,
      "grad_norm": 1.1805944442749023,
      "learning_rate": 9.985342105263158e-05,
      "loss": 0.4436,
      "step": 20567
    },
    {
      "epoch": 1.1517527158696383,
      "grad_norm": 1.303356409072876,
      "learning_rate": 9.985315789473684e-05,
      "loss": 0.363,
      "step": 20568
    },
    {
      "epoch": 1.1518087131817674,
      "grad_norm": 9.980767250061035,
      "learning_rate": 9.98528947368421e-05,
      "loss": 0.4461,
      "step": 20569
    },
    {
      "epoch": 1.1518647104938964,
      "grad_norm": 1.2744040489196777,
      "learning_rate": 9.985263157894738e-05,
      "loss": 0.4167,
      "step": 20570
    },
    {
      "epoch": 1.1519207078060254,
      "grad_norm": 1.841589331626892,
      "learning_rate": 9.985236842105264e-05,
      "loss": 0.3982,
      "step": 20571
    },
    {
      "epoch": 1.1519767051181544,
      "grad_norm": 1.4686000347137451,
      "learning_rate": 9.98521052631579e-05,
      "loss": 0.4736,
      "step": 20572
    },
    {
      "epoch": 1.1520327024302834,
      "grad_norm": 2.5662789344787598,
      "learning_rate": 9.985184210526316e-05,
      "loss": 0.3899,
      "step": 20573
    },
    {
      "epoch": 1.1520886997424125,
      "grad_norm": 3.3797099590301514,
      "learning_rate": 9.985157894736843e-05,
      "loss": 0.661,
      "step": 20574
    },
    {
      "epoch": 1.1521446970545415,
      "grad_norm": 1.292691946029663,
      "learning_rate": 9.985131578947369e-05,
      "loss": 0.5272,
      "step": 20575
    },
    {
      "epoch": 1.1522006943666705,
      "grad_norm": 1.3117496967315674,
      "learning_rate": 9.985105263157896e-05,
      "loss": 0.42,
      "step": 20576
    },
    {
      "epoch": 1.1522566916787995,
      "grad_norm": 1.5081813335418701,
      "learning_rate": 9.985078947368421e-05,
      "loss": 0.6265,
      "step": 20577
    },
    {
      "epoch": 1.1523126889909285,
      "grad_norm": 1.4189059734344482,
      "learning_rate": 9.985052631578948e-05,
      "loss": 0.5678,
      "step": 20578
    },
    {
      "epoch": 1.1523686863030576,
      "grad_norm": 1.3685775995254517,
      "learning_rate": 9.985026315789474e-05,
      "loss": 0.449,
      "step": 20579
    },
    {
      "epoch": 1.1524246836151866,
      "grad_norm": 1.1283231973648071,
      "learning_rate": 9.985000000000001e-05,
      "loss": 0.3494,
      "step": 20580
    },
    {
      "epoch": 1.1524806809273156,
      "grad_norm": 1.373646855354309,
      "learning_rate": 9.984973684210527e-05,
      "loss": 0.5035,
      "step": 20581
    },
    {
      "epoch": 1.1525366782394446,
      "grad_norm": 1.3119111061096191,
      "learning_rate": 9.984947368421052e-05,
      "loss": 0.4533,
      "step": 20582
    },
    {
      "epoch": 1.1525926755515736,
      "grad_norm": 1.592946171760559,
      "learning_rate": 9.98492105263158e-05,
      "loss": 0.4196,
      "step": 20583
    },
    {
      "epoch": 1.1526486728637027,
      "grad_norm": 1.5488710403442383,
      "learning_rate": 9.984894736842105e-05,
      "loss": 0.5623,
      "step": 20584
    },
    {
      "epoch": 1.1527046701758317,
      "grad_norm": 1.075756549835205,
      "learning_rate": 9.984868421052633e-05,
      "loss": 0.295,
      "step": 20585
    },
    {
      "epoch": 1.1527606674879607,
      "grad_norm": 1.412971019744873,
      "learning_rate": 9.984842105263159e-05,
      "loss": 0.3412,
      "step": 20586
    },
    {
      "epoch": 1.1528166648000897,
      "grad_norm": 1.327127456665039,
      "learning_rate": 9.984815789473685e-05,
      "loss": 0.3978,
      "step": 20587
    },
    {
      "epoch": 1.1528726621122187,
      "grad_norm": 1.4128966331481934,
      "learning_rate": 9.98478947368421e-05,
      "loss": 0.513,
      "step": 20588
    },
    {
      "epoch": 1.1529286594243477,
      "grad_norm": 1.3311222791671753,
      "learning_rate": 9.984763157894738e-05,
      "loss": 0.3648,
      "step": 20589
    },
    {
      "epoch": 1.1529846567364768,
      "grad_norm": 1.5850287675857544,
      "learning_rate": 9.984736842105264e-05,
      "loss": 0.6009,
      "step": 20590
    },
    {
      "epoch": 1.1530406540486058,
      "grad_norm": 1.4272699356079102,
      "learning_rate": 9.98471052631579e-05,
      "loss": 0.4599,
      "step": 20591
    },
    {
      "epoch": 1.1530966513607348,
      "grad_norm": 1.2315454483032227,
      "learning_rate": 9.984684210526316e-05,
      "loss": 0.4101,
      "step": 20592
    },
    {
      "epoch": 1.1531526486728638,
      "grad_norm": 1.3531090021133423,
      "learning_rate": 9.984657894736843e-05,
      "loss": 0.5647,
      "step": 20593
    },
    {
      "epoch": 1.1532086459849928,
      "grad_norm": 1.4263765811920166,
      "learning_rate": 9.984631578947369e-05,
      "loss": 0.4509,
      "step": 20594
    },
    {
      "epoch": 1.1532646432971219,
      "grad_norm": 1.655713438987732,
      "learning_rate": 9.984605263157895e-05,
      "loss": 0.4888,
      "step": 20595
    },
    {
      "epoch": 1.1533206406092509,
      "grad_norm": 1.4959297180175781,
      "learning_rate": 9.984578947368421e-05,
      "loss": 0.4943,
      "step": 20596
    },
    {
      "epoch": 1.1533766379213797,
      "grad_norm": 1.14749276638031,
      "learning_rate": 9.984552631578948e-05,
      "loss": 0.3827,
      "step": 20597
    },
    {
      "epoch": 1.1534326352335087,
      "grad_norm": 1.2836191654205322,
      "learning_rate": 9.984526315789474e-05,
      "loss": 0.4814,
      "step": 20598
    },
    {
      "epoch": 1.1534886325456377,
      "grad_norm": 1.748325228691101,
      "learning_rate": 9.9845e-05,
      "loss": 0.4994,
      "step": 20599
    },
    {
      "epoch": 1.1535446298577667,
      "grad_norm": 1.253800392150879,
      "learning_rate": 9.984473684210526e-05,
      "loss": 0.3932,
      "step": 20600
    },
    {
      "epoch": 1.1536006271698958,
      "grad_norm": 1.2215180397033691,
      "learning_rate": 9.984447368421052e-05,
      "loss": 0.4599,
      "step": 20601
    },
    {
      "epoch": 1.1536566244820248,
      "grad_norm": 1.1674331426620483,
      "learning_rate": 9.98442105263158e-05,
      "loss": 0.3978,
      "step": 20602
    },
    {
      "epoch": 1.1537126217941538,
      "grad_norm": 1.3800455331802368,
      "learning_rate": 9.984394736842106e-05,
      "loss": 0.4135,
      "step": 20603
    },
    {
      "epoch": 1.1537686191062828,
      "grad_norm": 1.2307895421981812,
      "learning_rate": 9.984368421052632e-05,
      "loss": 0.4272,
      "step": 20604
    },
    {
      "epoch": 1.1538246164184118,
      "grad_norm": 1.4166638851165771,
      "learning_rate": 9.984342105263158e-05,
      "loss": 0.4664,
      "step": 20605
    },
    {
      "epoch": 1.1538806137305408,
      "grad_norm": 1.3545113801956177,
      "learning_rate": 9.984315789473685e-05,
      "loss": 0.4026,
      "step": 20606
    },
    {
      "epoch": 1.1539366110426699,
      "grad_norm": 1.3365097045898438,
      "learning_rate": 9.984289473684211e-05,
      "loss": 0.4322,
      "step": 20607
    },
    {
      "epoch": 1.1539926083547989,
      "grad_norm": 1.1657264232635498,
      "learning_rate": 9.984263157894738e-05,
      "loss": 0.3561,
      "step": 20608
    },
    {
      "epoch": 1.154048605666928,
      "grad_norm": 1.6107335090637207,
      "learning_rate": 9.984236842105263e-05,
      "loss": 0.4779,
      "step": 20609
    },
    {
      "epoch": 1.154104602979057,
      "grad_norm": 1.2185475826263428,
      "learning_rate": 9.98421052631579e-05,
      "loss": 0.4163,
      "step": 20610
    },
    {
      "epoch": 1.154160600291186,
      "grad_norm": 1.1416774988174438,
      "learning_rate": 9.984184210526316e-05,
      "loss": 0.3986,
      "step": 20611
    },
    {
      "epoch": 1.154216597603315,
      "grad_norm": 1.601789951324463,
      "learning_rate": 9.984157894736843e-05,
      "loss": 0.5496,
      "step": 20612
    },
    {
      "epoch": 1.154272594915444,
      "grad_norm": 1.754992127418518,
      "learning_rate": 9.98413157894737e-05,
      "loss": 0.4872,
      "step": 20613
    },
    {
      "epoch": 1.154328592227573,
      "grad_norm": 1.2180742025375366,
      "learning_rate": 9.984105263157895e-05,
      "loss": 0.5108,
      "step": 20614
    },
    {
      "epoch": 1.154384589539702,
      "grad_norm": 1.204809546470642,
      "learning_rate": 9.984078947368421e-05,
      "loss": 0.4099,
      "step": 20615
    },
    {
      "epoch": 1.154440586851831,
      "grad_norm": 1.3774548768997192,
      "learning_rate": 9.984052631578947e-05,
      "loss": 0.5555,
      "step": 20616
    },
    {
      "epoch": 1.15449658416396,
      "grad_norm": 1.548224925994873,
      "learning_rate": 9.984026315789475e-05,
      "loss": 0.4562,
      "step": 20617
    },
    {
      "epoch": 1.154552581476089,
      "grad_norm": 1.4371669292449951,
      "learning_rate": 9.984e-05,
      "loss": 0.4748,
      "step": 20618
    },
    {
      "epoch": 1.154608578788218,
      "grad_norm": 1.6385565996170044,
      "learning_rate": 9.983973684210527e-05,
      "loss": 0.5372,
      "step": 20619
    },
    {
      "epoch": 1.1546645761003471,
      "grad_norm": 1.3957754373550415,
      "learning_rate": 9.983947368421053e-05,
      "loss": 0.3931,
      "step": 20620
    },
    {
      "epoch": 1.1547205734124761,
      "grad_norm": 1.3508726358413696,
      "learning_rate": 9.98392105263158e-05,
      "loss": 0.591,
      "step": 20621
    },
    {
      "epoch": 1.1547765707246052,
      "grad_norm": 1.2779326438903809,
      "learning_rate": 9.983894736842106e-05,
      "loss": 0.4228,
      "step": 20622
    },
    {
      "epoch": 1.1548325680367342,
      "grad_norm": 1.6317874193191528,
      "learning_rate": 9.983868421052632e-05,
      "loss": 0.7641,
      "step": 20623
    },
    {
      "epoch": 1.1548885653488632,
      "grad_norm": 1.2334160804748535,
      "learning_rate": 9.983842105263158e-05,
      "loss": 0.3859,
      "step": 20624
    },
    {
      "epoch": 1.1549445626609922,
      "grad_norm": 1.157307744026184,
      "learning_rate": 9.983815789473685e-05,
      "loss": 0.3748,
      "step": 20625
    },
    {
      "epoch": 1.1550005599731212,
      "grad_norm": 1.2137049436569214,
      "learning_rate": 9.983789473684211e-05,
      "loss": 0.3819,
      "step": 20626
    },
    {
      "epoch": 1.1550565572852503,
      "grad_norm": 1.1553492546081543,
      "learning_rate": 9.983763157894737e-05,
      "loss": 0.3774,
      "step": 20627
    },
    {
      "epoch": 1.1551125545973793,
      "grad_norm": 1.3425161838531494,
      "learning_rate": 9.983736842105263e-05,
      "loss": 0.359,
      "step": 20628
    },
    {
      "epoch": 1.1551685519095083,
      "grad_norm": 1.3786944150924683,
      "learning_rate": 9.98371052631579e-05,
      "loss": 0.4054,
      "step": 20629
    },
    {
      "epoch": 1.1552245492216373,
      "grad_norm": 1.623252511024475,
      "learning_rate": 9.983684210526316e-05,
      "loss": 0.3692,
      "step": 20630
    },
    {
      "epoch": 1.1552805465337663,
      "grad_norm": 1.4085582494735718,
      "learning_rate": 9.983657894736844e-05,
      "loss": 0.4427,
      "step": 20631
    },
    {
      "epoch": 1.1553365438458953,
      "grad_norm": 1.2663112878799438,
      "learning_rate": 9.983631578947368e-05,
      "loss": 0.3932,
      "step": 20632
    },
    {
      "epoch": 1.1553925411580244,
      "grad_norm": 1.4995399713516235,
      "learning_rate": 9.983605263157894e-05,
      "loss": 0.5565,
      "step": 20633
    },
    {
      "epoch": 1.1554485384701534,
      "grad_norm": 1.4257159233093262,
      "learning_rate": 9.983578947368422e-05,
      "loss": 0.5338,
      "step": 20634
    },
    {
      "epoch": 1.1555045357822824,
      "grad_norm": 1.286906361579895,
      "learning_rate": 9.983552631578948e-05,
      "loss": 0.4541,
      "step": 20635
    },
    {
      "epoch": 1.1555605330944114,
      "grad_norm": 1.3742042779922485,
      "learning_rate": 9.983526315789475e-05,
      "loss": 0.4161,
      "step": 20636
    },
    {
      "epoch": 1.1556165304065404,
      "grad_norm": 1.2167344093322754,
      "learning_rate": 9.9835e-05,
      "loss": 0.3699,
      "step": 20637
    },
    {
      "epoch": 1.1556725277186695,
      "grad_norm": 2.8422434329986572,
      "learning_rate": 9.983473684210527e-05,
      "loss": 0.4744,
      "step": 20638
    },
    {
      "epoch": 1.1557285250307985,
      "grad_norm": 1.3396075963974,
      "learning_rate": 9.983447368421053e-05,
      "loss": 0.4409,
      "step": 20639
    },
    {
      "epoch": 1.1557845223429275,
      "grad_norm": 1.3033193349838257,
      "learning_rate": 9.98342105263158e-05,
      "loss": 0.3278,
      "step": 20640
    },
    {
      "epoch": 1.1558405196550565,
      "grad_norm": 1.7807316780090332,
      "learning_rate": 9.983394736842106e-05,
      "loss": 0.407,
      "step": 20641
    },
    {
      "epoch": 1.1558965169671855,
      "grad_norm": 1.249908208847046,
      "learning_rate": 9.983368421052632e-05,
      "loss": 0.386,
      "step": 20642
    },
    {
      "epoch": 1.1559525142793146,
      "grad_norm": 1.384415864944458,
      "learning_rate": 9.983342105263158e-05,
      "loss": 0.4447,
      "step": 20643
    },
    {
      "epoch": 1.1560085115914436,
      "grad_norm": 1.1966683864593506,
      "learning_rate": 9.983315789473685e-05,
      "loss": 0.3824,
      "step": 20644
    },
    {
      "epoch": 1.1560645089035726,
      "grad_norm": 1.3831819295883179,
      "learning_rate": 9.983289473684211e-05,
      "loss": 0.4426,
      "step": 20645
    },
    {
      "epoch": 1.1561205062157016,
      "grad_norm": 1.4994510412216187,
      "learning_rate": 9.983263157894737e-05,
      "loss": 0.4747,
      "step": 20646
    },
    {
      "epoch": 1.1561765035278306,
      "grad_norm": 1.6412901878356934,
      "learning_rate": 9.983236842105263e-05,
      "loss": 0.4636,
      "step": 20647
    },
    {
      "epoch": 1.1562325008399597,
      "grad_norm": 1.3096891641616821,
      "learning_rate": 9.98321052631579e-05,
      "loss": 0.5052,
      "step": 20648
    },
    {
      "epoch": 1.1562884981520887,
      "grad_norm": 1.3484374284744263,
      "learning_rate": 9.983184210526317e-05,
      "loss": 0.453,
      "step": 20649
    },
    {
      "epoch": 1.1563444954642177,
      "grad_norm": 1.302478551864624,
      "learning_rate": 9.983157894736843e-05,
      "loss": 0.3929,
      "step": 20650
    },
    {
      "epoch": 1.1564004927763467,
      "grad_norm": 1.4896066188812256,
      "learning_rate": 9.983131578947369e-05,
      "loss": 0.5056,
      "step": 20651
    },
    {
      "epoch": 1.1564564900884757,
      "grad_norm": 1.3241114616394043,
      "learning_rate": 9.983105263157895e-05,
      "loss": 0.3247,
      "step": 20652
    },
    {
      "epoch": 1.1565124874006047,
      "grad_norm": 1.3620420694351196,
      "learning_rate": 9.983078947368422e-05,
      "loss": 0.4644,
      "step": 20653
    },
    {
      "epoch": 1.1565684847127338,
      "grad_norm": 1.547397255897522,
      "learning_rate": 9.983052631578948e-05,
      "loss": 0.346,
      "step": 20654
    },
    {
      "epoch": 1.1566244820248628,
      "grad_norm": 1.4957448244094849,
      "learning_rate": 9.983026315789474e-05,
      "loss": 0.5126,
      "step": 20655
    },
    {
      "epoch": 1.1566804793369918,
      "grad_norm": 1.281531810760498,
      "learning_rate": 9.983e-05,
      "loss": 0.3975,
      "step": 20656
    },
    {
      "epoch": 1.1567364766491208,
      "grad_norm": 1.444127082824707,
      "learning_rate": 9.982973684210527e-05,
      "loss": 0.4188,
      "step": 20657
    },
    {
      "epoch": 1.1567924739612498,
      "grad_norm": 1.3772447109222412,
      "learning_rate": 9.982947368421053e-05,
      "loss": 0.3654,
      "step": 20658
    },
    {
      "epoch": 1.1568484712733789,
      "grad_norm": 1.1936354637145996,
      "learning_rate": 9.982921052631579e-05,
      "loss": 0.3566,
      "step": 20659
    },
    {
      "epoch": 1.1569044685855079,
      "grad_norm": 1.5570268630981445,
      "learning_rate": 9.982894736842105e-05,
      "loss": 0.3684,
      "step": 20660
    },
    {
      "epoch": 1.156960465897637,
      "grad_norm": 1.8814806938171387,
      "learning_rate": 9.982868421052632e-05,
      "loss": 0.3629,
      "step": 20661
    },
    {
      "epoch": 1.157016463209766,
      "grad_norm": 1.587518334388733,
      "learning_rate": 9.982842105263158e-05,
      "loss": 0.4118,
      "step": 20662
    },
    {
      "epoch": 1.157072460521895,
      "grad_norm": 1.2009578943252563,
      "learning_rate": 9.982815789473686e-05,
      "loss": 0.359,
      "step": 20663
    },
    {
      "epoch": 1.157128457834024,
      "grad_norm": 1.4607332944869995,
      "learning_rate": 9.98278947368421e-05,
      "loss": 0.4445,
      "step": 20664
    },
    {
      "epoch": 1.157184455146153,
      "grad_norm": 2.0582501888275146,
      "learning_rate": 9.982763157894738e-05,
      "loss": 0.4486,
      "step": 20665
    },
    {
      "epoch": 1.157240452458282,
      "grad_norm": 1.9901195764541626,
      "learning_rate": 9.982736842105264e-05,
      "loss": 0.4279,
      "step": 20666
    },
    {
      "epoch": 1.157296449770411,
      "grad_norm": 2.280203342437744,
      "learning_rate": 9.98271052631579e-05,
      "loss": 0.5867,
      "step": 20667
    },
    {
      "epoch": 1.15735244708254,
      "grad_norm": 1.4749000072479248,
      "learning_rate": 9.982684210526317e-05,
      "loss": 0.4305,
      "step": 20668
    },
    {
      "epoch": 1.157408444394669,
      "grad_norm": 1.3941112756729126,
      "learning_rate": 9.982657894736842e-05,
      "loss": 0.5329,
      "step": 20669
    },
    {
      "epoch": 1.157464441706798,
      "grad_norm": 1.2100077867507935,
      "learning_rate": 9.982631578947369e-05,
      "loss": 0.4537,
      "step": 20670
    },
    {
      "epoch": 1.157520439018927,
      "grad_norm": 1.497073769569397,
      "learning_rate": 9.982605263157895e-05,
      "loss": 0.4575,
      "step": 20671
    },
    {
      "epoch": 1.1575764363310561,
      "grad_norm": 1.3474293947219849,
      "learning_rate": 9.982578947368422e-05,
      "loss": 0.4577,
      "step": 20672
    },
    {
      "epoch": 1.1576324336431851,
      "grad_norm": 1.2917795181274414,
      "learning_rate": 9.982552631578948e-05,
      "loss": 0.3241,
      "step": 20673
    },
    {
      "epoch": 1.1576884309553142,
      "grad_norm": 1.353052020072937,
      "learning_rate": 9.982526315789474e-05,
      "loss": 0.5881,
      "step": 20674
    },
    {
      "epoch": 1.1577444282674432,
      "grad_norm": 1.6958577632904053,
      "learning_rate": 9.9825e-05,
      "loss": 0.7388,
      "step": 20675
    },
    {
      "epoch": 1.1578004255795722,
      "grad_norm": 1.5526212453842163,
      "learning_rate": 9.982473684210527e-05,
      "loss": 0.4998,
      "step": 20676
    },
    {
      "epoch": 1.1578564228917012,
      "grad_norm": 1.227476716041565,
      "learning_rate": 9.982447368421053e-05,
      "loss": 0.4287,
      "step": 20677
    },
    {
      "epoch": 1.1579124202038302,
      "grad_norm": 1.4508161544799805,
      "learning_rate": 9.98242105263158e-05,
      "loss": 0.4634,
      "step": 20678
    },
    {
      "epoch": 1.1579684175159592,
      "grad_norm": 1.591030240058899,
      "learning_rate": 9.982394736842105e-05,
      "loss": 0.5235,
      "step": 20679
    },
    {
      "epoch": 1.1580244148280883,
      "grad_norm": 1.2589104175567627,
      "learning_rate": 9.982368421052633e-05,
      "loss": 0.4743,
      "step": 20680
    },
    {
      "epoch": 1.1580804121402173,
      "grad_norm": 1.3506782054901123,
      "learning_rate": 9.982342105263159e-05,
      "loss": 0.4832,
      "step": 20681
    },
    {
      "epoch": 1.1581364094523463,
      "grad_norm": 2.7089169025421143,
      "learning_rate": 9.982315789473685e-05,
      "loss": 0.3896,
      "step": 20682
    },
    {
      "epoch": 1.1581924067644753,
      "grad_norm": 1.3554126024246216,
      "learning_rate": 9.98228947368421e-05,
      "loss": 0.4197,
      "step": 20683
    },
    {
      "epoch": 1.1582484040766043,
      "grad_norm": 1.5139764547348022,
      "learning_rate": 9.982263157894738e-05,
      "loss": 0.4632,
      "step": 20684
    },
    {
      "epoch": 1.1583044013887334,
      "grad_norm": 1.5954136848449707,
      "learning_rate": 9.982236842105264e-05,
      "loss": 0.5863,
      "step": 20685
    },
    {
      "epoch": 1.1583603987008624,
      "grad_norm": 1.5175155401229858,
      "learning_rate": 9.98221052631579e-05,
      "loss": 0.3828,
      "step": 20686
    },
    {
      "epoch": 1.1584163960129914,
      "grad_norm": 1.2836077213287354,
      "learning_rate": 9.982184210526316e-05,
      "loss": 0.4393,
      "step": 20687
    },
    {
      "epoch": 1.1584723933251204,
      "grad_norm": 1.2348936796188354,
      "learning_rate": 9.982157894736842e-05,
      "loss": 0.4897,
      "step": 20688
    },
    {
      "epoch": 1.1585283906372494,
      "grad_norm": 1.135157823562622,
      "learning_rate": 9.982131578947369e-05,
      "loss": 0.375,
      "step": 20689
    },
    {
      "epoch": 1.1585843879493785,
      "grad_norm": 1.1472995281219482,
      "learning_rate": 9.982105263157895e-05,
      "loss": 0.4163,
      "step": 20690
    },
    {
      "epoch": 1.1586403852615075,
      "grad_norm": 1.7743520736694336,
      "learning_rate": 9.982078947368422e-05,
      "loss": 0.5812,
      "step": 20691
    },
    {
      "epoch": 1.1586963825736365,
      "grad_norm": 1.519546627998352,
      "learning_rate": 9.982052631578947e-05,
      "loss": 0.4854,
      "step": 20692
    },
    {
      "epoch": 1.1587523798857655,
      "grad_norm": 1.6588484048843384,
      "learning_rate": 9.982026315789474e-05,
      "loss": 0.4428,
      "step": 20693
    },
    {
      "epoch": 1.1588083771978945,
      "grad_norm": 1.4660861492156982,
      "learning_rate": 9.982e-05,
      "loss": 0.5322,
      "step": 20694
    },
    {
      "epoch": 1.1588643745100236,
      "grad_norm": 1.6412279605865479,
      "learning_rate": 9.981973684210528e-05,
      "loss": 0.3968,
      "step": 20695
    },
    {
      "epoch": 1.1589203718221526,
      "grad_norm": 1.2905727624893188,
      "learning_rate": 9.981947368421054e-05,
      "loss": 0.3578,
      "step": 20696
    },
    {
      "epoch": 1.1589763691342816,
      "grad_norm": 1.4095821380615234,
      "learning_rate": 9.98192105263158e-05,
      "loss": 0.5055,
      "step": 20697
    },
    {
      "epoch": 1.1590323664464106,
      "grad_norm": 1.3190431594848633,
      "learning_rate": 9.981894736842106e-05,
      "loss": 0.4439,
      "step": 20698
    },
    {
      "epoch": 1.1590883637585396,
      "grad_norm": 1.1977996826171875,
      "learning_rate": 9.981868421052633e-05,
      "loss": 0.3916,
      "step": 20699
    },
    {
      "epoch": 1.1591443610706686,
      "grad_norm": 1.425007700920105,
      "learning_rate": 9.981842105263159e-05,
      "loss": 0.3826,
      "step": 20700
    },
    {
      "epoch": 1.1592003583827977,
      "grad_norm": 1.4595378637313843,
      "learning_rate": 9.981815789473685e-05,
      "loss": 0.4243,
      "step": 20701
    },
    {
      "epoch": 1.1592563556949267,
      "grad_norm": 1.5491729974746704,
      "learning_rate": 9.981789473684211e-05,
      "loss": 0.5941,
      "step": 20702
    },
    {
      "epoch": 1.1593123530070557,
      "grad_norm": 1.4100964069366455,
      "learning_rate": 9.981763157894737e-05,
      "loss": 0.5043,
      "step": 20703
    },
    {
      "epoch": 1.1593683503191847,
      "grad_norm": 1.2770140171051025,
      "learning_rate": 9.981736842105264e-05,
      "loss": 0.4207,
      "step": 20704
    },
    {
      "epoch": 1.1594243476313137,
      "grad_norm": 2.4358294010162354,
      "learning_rate": 9.98171052631579e-05,
      "loss": 0.3678,
      "step": 20705
    },
    {
      "epoch": 1.1594803449434428,
      "grad_norm": 1.441752314567566,
      "learning_rate": 9.981684210526316e-05,
      "loss": 0.5353,
      "step": 20706
    },
    {
      "epoch": 1.1595363422555718,
      "grad_norm": 1.4326294660568237,
      "learning_rate": 9.981657894736842e-05,
      "loss": 0.4606,
      "step": 20707
    },
    {
      "epoch": 1.1595923395677008,
      "grad_norm": 1.4000003337860107,
      "learning_rate": 9.98163157894737e-05,
      "loss": 0.442,
      "step": 20708
    },
    {
      "epoch": 1.1596483368798298,
      "grad_norm": 1.3965154886245728,
      "learning_rate": 9.981605263157895e-05,
      "loss": 0.4879,
      "step": 20709
    },
    {
      "epoch": 1.1597043341919588,
      "grad_norm": 1.3141099214553833,
      "learning_rate": 9.981578947368421e-05,
      "loss": 0.3911,
      "step": 20710
    },
    {
      "epoch": 1.1597603315040879,
      "grad_norm": 1.27518630027771,
      "learning_rate": 9.981552631578947e-05,
      "loss": 0.3508,
      "step": 20711
    },
    {
      "epoch": 1.1598163288162169,
      "grad_norm": 1.3997254371643066,
      "learning_rate": 9.981526315789475e-05,
      "loss": 0.3492,
      "step": 20712
    },
    {
      "epoch": 1.159872326128346,
      "grad_norm": 1.1966441869735718,
      "learning_rate": 9.9815e-05,
      "loss": 0.4952,
      "step": 20713
    },
    {
      "epoch": 1.159928323440475,
      "grad_norm": 1.563307762145996,
      "learning_rate": 9.981473684210527e-05,
      "loss": 0.4078,
      "step": 20714
    },
    {
      "epoch": 1.159984320752604,
      "grad_norm": 1.7178142070770264,
      "learning_rate": 9.981447368421053e-05,
      "loss": 0.4521,
      "step": 20715
    },
    {
      "epoch": 1.160040318064733,
      "grad_norm": 1.086503028869629,
      "learning_rate": 9.98142105263158e-05,
      "loss": 0.3867,
      "step": 20716
    },
    {
      "epoch": 1.160096315376862,
      "grad_norm": 1.123307466506958,
      "learning_rate": 9.981394736842106e-05,
      "loss": 0.346,
      "step": 20717
    },
    {
      "epoch": 1.160152312688991,
      "grad_norm": 1.4775654077529907,
      "learning_rate": 9.981368421052633e-05,
      "loss": 0.5772,
      "step": 20718
    },
    {
      "epoch": 1.16020831000112,
      "grad_norm": 1.3776569366455078,
      "learning_rate": 9.981342105263158e-05,
      "loss": 0.441,
      "step": 20719
    },
    {
      "epoch": 1.160264307313249,
      "grad_norm": 1.5174689292907715,
      "learning_rate": 9.981315789473684e-05,
      "loss": 0.4709,
      "step": 20720
    },
    {
      "epoch": 1.160320304625378,
      "grad_norm": 1.0853722095489502,
      "learning_rate": 9.981289473684211e-05,
      "loss": 0.4545,
      "step": 20721
    },
    {
      "epoch": 1.160376301937507,
      "grad_norm": 1.148170828819275,
      "learning_rate": 9.981263157894737e-05,
      "loss": 0.368,
      "step": 20722
    },
    {
      "epoch": 1.160432299249636,
      "grad_norm": 1.2531620264053345,
      "learning_rate": 9.981236842105264e-05,
      "loss": 0.395,
      "step": 20723
    },
    {
      "epoch": 1.160488296561765,
      "grad_norm": 1.188633680343628,
      "learning_rate": 9.981210526315789e-05,
      "loss": 0.3,
      "step": 20724
    },
    {
      "epoch": 1.1605442938738941,
      "grad_norm": 1.3251160383224487,
      "learning_rate": 9.981184210526316e-05,
      "loss": 0.4416,
      "step": 20725
    },
    {
      "epoch": 1.1606002911860231,
      "grad_norm": 1.6037322282791138,
      "learning_rate": 9.981157894736842e-05,
      "loss": 0.4821,
      "step": 20726
    },
    {
      "epoch": 1.1606562884981522,
      "grad_norm": 1.3819395303726196,
      "learning_rate": 9.98113157894737e-05,
      "loss": 0.4847,
      "step": 20727
    },
    {
      "epoch": 1.1607122858102812,
      "grad_norm": 1.2603572607040405,
      "learning_rate": 9.981105263157896e-05,
      "loss": 0.401,
      "step": 20728
    },
    {
      "epoch": 1.1607682831224102,
      "grad_norm": 1.2100732326507568,
      "learning_rate": 9.981078947368422e-05,
      "loss": 0.5267,
      "step": 20729
    },
    {
      "epoch": 1.1608242804345392,
      "grad_norm": 2.2614517211914062,
      "learning_rate": 9.981052631578948e-05,
      "loss": 0.4493,
      "step": 20730
    },
    {
      "epoch": 1.1608802777466682,
      "grad_norm": 1.3757879734039307,
      "learning_rate": 9.981026315789475e-05,
      "loss": 0.4935,
      "step": 20731
    },
    {
      "epoch": 1.1609362750587973,
      "grad_norm": 1.8969056606292725,
      "learning_rate": 9.981000000000001e-05,
      "loss": 0.3687,
      "step": 20732
    },
    {
      "epoch": 1.1609922723709263,
      "grad_norm": 1.234013319015503,
      "learning_rate": 9.980973684210527e-05,
      "loss": 0.3627,
      "step": 20733
    },
    {
      "epoch": 1.1610482696830553,
      "grad_norm": 1.4058195352554321,
      "learning_rate": 9.980947368421053e-05,
      "loss": 0.4655,
      "step": 20734
    },
    {
      "epoch": 1.1611042669951843,
      "grad_norm": 1.357609748840332,
      "learning_rate": 9.98092105263158e-05,
      "loss": 0.4541,
      "step": 20735
    },
    {
      "epoch": 1.1611602643073133,
      "grad_norm": 1.443774938583374,
      "learning_rate": 9.980894736842106e-05,
      "loss": 0.4308,
      "step": 20736
    },
    {
      "epoch": 1.1612162616194424,
      "grad_norm": 1.4773916006088257,
      "learning_rate": 9.980868421052632e-05,
      "loss": 0.4524,
      "step": 20737
    },
    {
      "epoch": 1.1612722589315714,
      "grad_norm": 1.1345545053482056,
      "learning_rate": 9.980842105263158e-05,
      "loss": 0.489,
      "step": 20738
    },
    {
      "epoch": 1.1613282562437004,
      "grad_norm": 1.2463173866271973,
      "learning_rate": 9.980815789473684e-05,
      "loss": 0.3472,
      "step": 20739
    },
    {
      "epoch": 1.1613842535558294,
      "grad_norm": 1.7290700674057007,
      "learning_rate": 9.980789473684211e-05,
      "loss": 0.424,
      "step": 20740
    },
    {
      "epoch": 1.1614402508679584,
      "grad_norm": 1.5474966764450073,
      "learning_rate": 9.980763157894737e-05,
      "loss": 0.415,
      "step": 20741
    },
    {
      "epoch": 1.1614962481800872,
      "grad_norm": 1.5076913833618164,
      "learning_rate": 9.980736842105263e-05,
      "loss": 0.4521,
      "step": 20742
    },
    {
      "epoch": 1.1615522454922163,
      "grad_norm": 1.579359531402588,
      "learning_rate": 9.980710526315789e-05,
      "loss": 0.5414,
      "step": 20743
    },
    {
      "epoch": 1.1616082428043453,
      "grad_norm": 1.6798981428146362,
      "learning_rate": 9.980684210526317e-05,
      "loss": 0.3569,
      "step": 20744
    },
    {
      "epoch": 1.1616642401164743,
      "grad_norm": 1.8302743434906006,
      "learning_rate": 9.980657894736843e-05,
      "loss": 0.4196,
      "step": 20745
    },
    {
      "epoch": 1.1617202374286033,
      "grad_norm": 1.3661141395568848,
      "learning_rate": 9.98063157894737e-05,
      "loss": 0.4603,
      "step": 20746
    },
    {
      "epoch": 1.1617762347407323,
      "grad_norm": 1.3627281188964844,
      "learning_rate": 9.980605263157894e-05,
      "loss": 0.3453,
      "step": 20747
    },
    {
      "epoch": 1.1618322320528613,
      "grad_norm": 1.2853434085845947,
      "learning_rate": 9.980578947368422e-05,
      "loss": 0.3704,
      "step": 20748
    },
    {
      "epoch": 1.1618882293649904,
      "grad_norm": 1.2826396226882935,
      "learning_rate": 9.980552631578948e-05,
      "loss": 0.4098,
      "step": 20749
    },
    {
      "epoch": 1.1619442266771194,
      "grad_norm": 1.340430736541748,
      "learning_rate": 9.980526315789475e-05,
      "loss": 0.5137,
      "step": 20750
    },
    {
      "epoch": 1.1620002239892484,
      "grad_norm": 1.456680417060852,
      "learning_rate": 9.9805e-05,
      "loss": 0.4012,
      "step": 20751
    },
    {
      "epoch": 1.1620562213013774,
      "grad_norm": 1.3966798782348633,
      "learning_rate": 9.980473684210527e-05,
      "loss": 0.4541,
      "step": 20752
    },
    {
      "epoch": 1.1621122186135064,
      "grad_norm": 1.2240643501281738,
      "learning_rate": 9.980447368421053e-05,
      "loss": 0.51,
      "step": 20753
    },
    {
      "epoch": 1.1621682159256355,
      "grad_norm": 1.1132729053497314,
      "learning_rate": 9.980421052631579e-05,
      "loss": 0.3773,
      "step": 20754
    },
    {
      "epoch": 1.1622242132377645,
      "grad_norm": 1.4580998420715332,
      "learning_rate": 9.980394736842106e-05,
      "loss": 0.4872,
      "step": 20755
    },
    {
      "epoch": 1.1622802105498935,
      "grad_norm": 1.3250617980957031,
      "learning_rate": 9.980368421052631e-05,
      "loss": 0.3912,
      "step": 20756
    },
    {
      "epoch": 1.1623362078620225,
      "grad_norm": 1.5409741401672363,
      "learning_rate": 9.980342105263158e-05,
      "loss": 0.3819,
      "step": 20757
    },
    {
      "epoch": 1.1623922051741515,
      "grad_norm": 1.1932024955749512,
      "learning_rate": 9.980315789473684e-05,
      "loss": 0.4859,
      "step": 20758
    },
    {
      "epoch": 1.1624482024862806,
      "grad_norm": 1.4459530115127563,
      "learning_rate": 9.980289473684212e-05,
      "loss": 0.4208,
      "step": 20759
    },
    {
      "epoch": 1.1625041997984096,
      "grad_norm": 1.2691309452056885,
      "learning_rate": 9.980263157894738e-05,
      "loss": 0.4219,
      "step": 20760
    },
    {
      "epoch": 1.1625601971105386,
      "grad_norm": 1.4836251735687256,
      "learning_rate": 9.980236842105264e-05,
      "loss": 0.3515,
      "step": 20761
    },
    {
      "epoch": 1.1626161944226676,
      "grad_norm": 1.3650662899017334,
      "learning_rate": 9.98021052631579e-05,
      "loss": 0.4355,
      "step": 20762
    },
    {
      "epoch": 1.1626721917347966,
      "grad_norm": 1.3967193365097046,
      "learning_rate": 9.980184210526317e-05,
      "loss": 0.4403,
      "step": 20763
    },
    {
      "epoch": 1.1627281890469257,
      "grad_norm": 1.680420160293579,
      "learning_rate": 9.980157894736843e-05,
      "loss": 0.6033,
      "step": 20764
    },
    {
      "epoch": 1.1627841863590547,
      "grad_norm": 1.2812414169311523,
      "learning_rate": 9.980131578947369e-05,
      "loss": 0.3003,
      "step": 20765
    },
    {
      "epoch": 1.1628401836711837,
      "grad_norm": 1.4538753032684326,
      "learning_rate": 9.980105263157895e-05,
      "loss": 0.4283,
      "step": 20766
    },
    {
      "epoch": 1.1628961809833127,
      "grad_norm": 1.4691946506500244,
      "learning_rate": 9.980078947368422e-05,
      "loss": 0.4283,
      "step": 20767
    },
    {
      "epoch": 1.1629521782954417,
      "grad_norm": 1.609779953956604,
      "learning_rate": 9.980052631578948e-05,
      "loss": 0.4358,
      "step": 20768
    },
    {
      "epoch": 1.1630081756075707,
      "grad_norm": 1.4415971040725708,
      "learning_rate": 9.980026315789474e-05,
      "loss": 0.5102,
      "step": 20769
    },
    {
      "epoch": 1.1630641729196998,
      "grad_norm": 1.41593337059021,
      "learning_rate": 9.98e-05,
      "loss": 0.4327,
      "step": 20770
    },
    {
      "epoch": 1.1631201702318288,
      "grad_norm": 5.720890045166016,
      "learning_rate": 9.979973684210526e-05,
      "loss": 0.5614,
      "step": 20771
    },
    {
      "epoch": 1.1631761675439578,
      "grad_norm": 1.302211880683899,
      "learning_rate": 9.979947368421053e-05,
      "loss": 0.4625,
      "step": 20772
    },
    {
      "epoch": 1.1632321648560868,
      "grad_norm": 1.50058114528656,
      "learning_rate": 9.979921052631579e-05,
      "loss": 0.4799,
      "step": 20773
    },
    {
      "epoch": 1.1632881621682158,
      "grad_norm": 1.2870792150497437,
      "learning_rate": 9.979894736842105e-05,
      "loss": 0.4244,
      "step": 20774
    },
    {
      "epoch": 1.1633441594803449,
      "grad_norm": 1.3122467994689941,
      "learning_rate": 9.979868421052631e-05,
      "loss": 0.3458,
      "step": 20775
    },
    {
      "epoch": 1.1634001567924739,
      "grad_norm": 2.0894978046417236,
      "learning_rate": 9.979842105263159e-05,
      "loss": 0.5191,
      "step": 20776
    },
    {
      "epoch": 1.163456154104603,
      "grad_norm": 1.4296149015426636,
      "learning_rate": 9.979815789473685e-05,
      "loss": 0.4258,
      "step": 20777
    },
    {
      "epoch": 1.163512151416732,
      "grad_norm": 1.4373489618301392,
      "learning_rate": 9.979789473684212e-05,
      "loss": 0.4261,
      "step": 20778
    },
    {
      "epoch": 1.163568148728861,
      "grad_norm": 1.5073524713516235,
      "learning_rate": 9.979763157894736e-05,
      "loss": 0.4356,
      "step": 20779
    },
    {
      "epoch": 1.16362414604099,
      "grad_norm": 1.6112574338912964,
      "learning_rate": 9.979736842105264e-05,
      "loss": 0.606,
      "step": 20780
    },
    {
      "epoch": 1.163680143353119,
      "grad_norm": 1.4365694522857666,
      "learning_rate": 9.97971052631579e-05,
      "loss": 0.454,
      "step": 20781
    },
    {
      "epoch": 1.163736140665248,
      "grad_norm": 1.2466692924499512,
      "learning_rate": 9.979684210526317e-05,
      "loss": 0.3782,
      "step": 20782
    },
    {
      "epoch": 1.163792137977377,
      "grad_norm": 1.4135067462921143,
      "learning_rate": 9.979657894736843e-05,
      "loss": 0.5012,
      "step": 20783
    },
    {
      "epoch": 1.163848135289506,
      "grad_norm": 1.2118281126022339,
      "learning_rate": 9.979631578947369e-05,
      "loss": 0.4247,
      "step": 20784
    },
    {
      "epoch": 1.163904132601635,
      "grad_norm": 1.41605544090271,
      "learning_rate": 9.979605263157895e-05,
      "loss": 0.4478,
      "step": 20785
    },
    {
      "epoch": 1.163960129913764,
      "grad_norm": 1.5420175790786743,
      "learning_rate": 9.979578947368422e-05,
      "loss": 0.4679,
      "step": 20786
    },
    {
      "epoch": 1.164016127225893,
      "grad_norm": 1.2664000988006592,
      "learning_rate": 9.979552631578948e-05,
      "loss": 0.4349,
      "step": 20787
    },
    {
      "epoch": 1.1640721245380221,
      "grad_norm": 1.2067443132400513,
      "learning_rate": 9.979526315789474e-05,
      "loss": 0.4742,
      "step": 20788
    },
    {
      "epoch": 1.1641281218501511,
      "grad_norm": 1.4331861734390259,
      "learning_rate": 9.9795e-05,
      "loss": 0.4219,
      "step": 20789
    },
    {
      "epoch": 1.1641841191622802,
      "grad_norm": 2.042576789855957,
      "learning_rate": 9.979473684210526e-05,
      "loss": 0.4347,
      "step": 20790
    },
    {
      "epoch": 1.1642401164744092,
      "grad_norm": 1.2423555850982666,
      "learning_rate": 9.979447368421054e-05,
      "loss": 0.4861,
      "step": 20791
    },
    {
      "epoch": 1.1642961137865382,
      "grad_norm": 1.7242858409881592,
      "learning_rate": 9.97942105263158e-05,
      "loss": 0.6151,
      "step": 20792
    },
    {
      "epoch": 1.1643521110986672,
      "grad_norm": 1.4775879383087158,
      "learning_rate": 9.979394736842105e-05,
      "loss": 0.4475,
      "step": 20793
    },
    {
      "epoch": 1.1644081084107962,
      "grad_norm": 1.215835452079773,
      "learning_rate": 9.979368421052631e-05,
      "loss": 0.4152,
      "step": 20794
    },
    {
      "epoch": 1.1644641057229252,
      "grad_norm": 1.7902518510818481,
      "learning_rate": 9.979342105263159e-05,
      "loss": 0.5051,
      "step": 20795
    },
    {
      "epoch": 1.1645201030350543,
      "grad_norm": 1.3555628061294556,
      "learning_rate": 9.979315789473685e-05,
      "loss": 0.4597,
      "step": 20796
    },
    {
      "epoch": 1.1645761003471833,
      "grad_norm": 1.3614206314086914,
      "learning_rate": 9.979289473684211e-05,
      "loss": 0.4464,
      "step": 20797
    },
    {
      "epoch": 1.1646320976593123,
      "grad_norm": 1.5785999298095703,
      "learning_rate": 9.979263157894737e-05,
      "loss": 0.494,
      "step": 20798
    },
    {
      "epoch": 1.1646880949714413,
      "grad_norm": 1.6326556205749512,
      "learning_rate": 9.979236842105264e-05,
      "loss": 0.5103,
      "step": 20799
    },
    {
      "epoch": 1.1647440922835703,
      "grad_norm": 1.3406559228897095,
      "learning_rate": 9.97921052631579e-05,
      "loss": 0.4389,
      "step": 20800
    },
    {
      "epoch": 1.1648000895956994,
      "grad_norm": 1.7493377923965454,
      "learning_rate": 9.979184210526317e-05,
      "loss": 0.5527,
      "step": 20801
    },
    {
      "epoch": 1.1648560869078284,
      "grad_norm": 1.2731801271438599,
      "learning_rate": 9.979157894736842e-05,
      "loss": 0.4271,
      "step": 20802
    },
    {
      "epoch": 1.1649120842199574,
      "grad_norm": 1.9941129684448242,
      "learning_rate": 9.979131578947369e-05,
      "loss": 0.4659,
      "step": 20803
    },
    {
      "epoch": 1.1649680815320864,
      "grad_norm": 1.557101845741272,
      "learning_rate": 9.979105263157895e-05,
      "loss": 0.5432,
      "step": 20804
    },
    {
      "epoch": 1.1650240788442154,
      "grad_norm": 1.1992847919464111,
      "learning_rate": 9.979078947368423e-05,
      "loss": 0.4119,
      "step": 20805
    },
    {
      "epoch": 1.1650800761563445,
      "grad_norm": 1.6048921346664429,
      "learning_rate": 9.979052631578947e-05,
      "loss": 0.5382,
      "step": 20806
    },
    {
      "epoch": 1.1651360734684735,
      "grad_norm": 1.4330283403396606,
      "learning_rate": 9.979026315789473e-05,
      "loss": 0.498,
      "step": 20807
    },
    {
      "epoch": 1.1651920707806025,
      "grad_norm": 1.2458750009536743,
      "learning_rate": 9.979e-05,
      "loss": 0.3971,
      "step": 20808
    },
    {
      "epoch": 1.1652480680927315,
      "grad_norm": 1.2125712633132935,
      "learning_rate": 9.978973684210526e-05,
      "loss": 0.3569,
      "step": 20809
    },
    {
      "epoch": 1.1653040654048605,
      "grad_norm": 1.4834809303283691,
      "learning_rate": 9.978947368421054e-05,
      "loss": 0.3777,
      "step": 20810
    },
    {
      "epoch": 1.1653600627169896,
      "grad_norm": 1.3419028520584106,
      "learning_rate": 9.978921052631578e-05,
      "loss": 0.4946,
      "step": 20811
    },
    {
      "epoch": 1.1654160600291186,
      "grad_norm": 1.4450846910476685,
      "learning_rate": 9.978894736842106e-05,
      "loss": 0.4897,
      "step": 20812
    },
    {
      "epoch": 1.1654720573412476,
      "grad_norm": 1.1625169515609741,
      "learning_rate": 9.978868421052632e-05,
      "loss": 0.3192,
      "step": 20813
    },
    {
      "epoch": 1.1655280546533766,
      "grad_norm": 1.422334909439087,
      "learning_rate": 9.978842105263159e-05,
      "loss": 0.413,
      "step": 20814
    },
    {
      "epoch": 1.1655840519655056,
      "grad_norm": 1.3449368476867676,
      "learning_rate": 9.978815789473685e-05,
      "loss": 0.477,
      "step": 20815
    },
    {
      "epoch": 1.1656400492776346,
      "grad_norm": 1.451908826828003,
      "learning_rate": 9.978789473684211e-05,
      "loss": 0.4372,
      "step": 20816
    },
    {
      "epoch": 1.1656960465897637,
      "grad_norm": 1.3651716709136963,
      "learning_rate": 9.978763157894737e-05,
      "loss": 0.4729,
      "step": 20817
    },
    {
      "epoch": 1.1657520439018927,
      "grad_norm": 1.708986759185791,
      "learning_rate": 9.978736842105264e-05,
      "loss": 0.4606,
      "step": 20818
    },
    {
      "epoch": 1.1658080412140217,
      "grad_norm": 1.6787852048873901,
      "learning_rate": 9.97871052631579e-05,
      "loss": 0.496,
      "step": 20819
    },
    {
      "epoch": 1.1658640385261507,
      "grad_norm": 1.1393404006958008,
      "learning_rate": 9.978684210526316e-05,
      "loss": 0.3712,
      "step": 20820
    },
    {
      "epoch": 1.1659200358382797,
      "grad_norm": 1.606389045715332,
      "learning_rate": 9.978657894736842e-05,
      "loss": 0.468,
      "step": 20821
    },
    {
      "epoch": 1.1659760331504088,
      "grad_norm": 1.4294898509979248,
      "learning_rate": 9.97863157894737e-05,
      "loss": 0.4755,
      "step": 20822
    },
    {
      "epoch": 1.1660320304625378,
      "grad_norm": 1.9825870990753174,
      "learning_rate": 9.978605263157896e-05,
      "loss": 0.4481,
      "step": 20823
    },
    {
      "epoch": 1.1660880277746668,
      "grad_norm": 1.1417534351348877,
      "learning_rate": 9.978578947368421e-05,
      "loss": 0.3835,
      "step": 20824
    },
    {
      "epoch": 1.1661440250867958,
      "grad_norm": 1.6894428730010986,
      "learning_rate": 9.978552631578947e-05,
      "loss": 0.6405,
      "step": 20825
    },
    {
      "epoch": 1.1662000223989248,
      "grad_norm": 1.5436031818389893,
      "learning_rate": 9.978526315789473e-05,
      "loss": 0.5195,
      "step": 20826
    },
    {
      "epoch": 1.1662560197110539,
      "grad_norm": 1.8321113586425781,
      "learning_rate": 9.978500000000001e-05,
      "loss": 0.4712,
      "step": 20827
    },
    {
      "epoch": 1.1663120170231829,
      "grad_norm": 1.4048885107040405,
      "learning_rate": 9.978473684210527e-05,
      "loss": 0.4554,
      "step": 20828
    },
    {
      "epoch": 1.166368014335312,
      "grad_norm": 1.2561533451080322,
      "learning_rate": 9.978447368421053e-05,
      "loss": 0.4446,
      "step": 20829
    },
    {
      "epoch": 1.166424011647441,
      "grad_norm": 1.5617958307266235,
      "learning_rate": 9.978421052631579e-05,
      "loss": 0.4572,
      "step": 20830
    },
    {
      "epoch": 1.16648000895957,
      "grad_norm": 2.4134650230407715,
      "learning_rate": 9.978394736842106e-05,
      "loss": 0.3244,
      "step": 20831
    },
    {
      "epoch": 1.166536006271699,
      "grad_norm": 2.2136783599853516,
      "learning_rate": 9.978368421052632e-05,
      "loss": 0.5679,
      "step": 20832
    },
    {
      "epoch": 1.166592003583828,
      "grad_norm": 1.5566372871398926,
      "learning_rate": 9.978342105263159e-05,
      "loss": 0.3798,
      "step": 20833
    },
    {
      "epoch": 1.166648000895957,
      "grad_norm": 1.5471076965332031,
      "learning_rate": 9.978315789473684e-05,
      "loss": 0.2983,
      "step": 20834
    },
    {
      "epoch": 1.166703998208086,
      "grad_norm": 1.1122899055480957,
      "learning_rate": 9.978289473684211e-05,
      "loss": 0.3883,
      "step": 20835
    },
    {
      "epoch": 1.166759995520215,
      "grad_norm": 1.3770607709884644,
      "learning_rate": 9.978263157894737e-05,
      "loss": 0.414,
      "step": 20836
    },
    {
      "epoch": 1.166815992832344,
      "grad_norm": 1.6381101608276367,
      "learning_rate": 9.978236842105265e-05,
      "loss": 0.4451,
      "step": 20837
    },
    {
      "epoch": 1.166871990144473,
      "grad_norm": 1.3374508619308472,
      "learning_rate": 9.97821052631579e-05,
      "loss": 0.6068,
      "step": 20838
    },
    {
      "epoch": 1.166927987456602,
      "grad_norm": 1.1922324895858765,
      "learning_rate": 9.978184210526317e-05,
      "loss": 0.4348,
      "step": 20839
    },
    {
      "epoch": 1.166983984768731,
      "grad_norm": 1.3648905754089355,
      "learning_rate": 9.978157894736842e-05,
      "loss": 0.5869,
      "step": 20840
    },
    {
      "epoch": 1.1670399820808601,
      "grad_norm": 1.3370119333267212,
      "learning_rate": 9.978131578947368e-05,
      "loss": 0.3223,
      "step": 20841
    },
    {
      "epoch": 1.1670959793929891,
      "grad_norm": 1.3336783647537231,
      "learning_rate": 9.978105263157896e-05,
      "loss": 0.3605,
      "step": 20842
    },
    {
      "epoch": 1.1671519767051182,
      "grad_norm": 1.1952753067016602,
      "learning_rate": 9.978078947368422e-05,
      "loss": 0.409,
      "step": 20843
    },
    {
      "epoch": 1.1672079740172472,
      "grad_norm": 1.3847600221633911,
      "learning_rate": 9.978052631578948e-05,
      "loss": 0.473,
      "step": 20844
    },
    {
      "epoch": 1.1672639713293762,
      "grad_norm": 1.280964970588684,
      "learning_rate": 9.978026315789474e-05,
      "loss": 0.4478,
      "step": 20845
    },
    {
      "epoch": 1.1673199686415052,
      "grad_norm": 1.550473928451538,
      "learning_rate": 9.978000000000001e-05,
      "loss": 0.3973,
      "step": 20846
    },
    {
      "epoch": 1.1673759659536342,
      "grad_norm": 1.2537569999694824,
      "learning_rate": 9.977973684210527e-05,
      "loss": 0.4855,
      "step": 20847
    },
    {
      "epoch": 1.1674319632657633,
      "grad_norm": 1.225132703781128,
      "learning_rate": 9.977947368421053e-05,
      "loss": 0.3958,
      "step": 20848
    },
    {
      "epoch": 1.1674879605778923,
      "grad_norm": 1.2652547359466553,
      "learning_rate": 9.977921052631579e-05,
      "loss": 0.4369,
      "step": 20849
    },
    {
      "epoch": 1.1675439578900213,
      "grad_norm": 1.440535306930542,
      "learning_rate": 9.977894736842106e-05,
      "loss": 0.5253,
      "step": 20850
    },
    {
      "epoch": 1.1675999552021503,
      "grad_norm": 1.3133693933486938,
      "learning_rate": 9.977868421052632e-05,
      "loss": 0.3711,
      "step": 20851
    },
    {
      "epoch": 1.1676559525142793,
      "grad_norm": 1.2892177104949951,
      "learning_rate": 9.977842105263158e-05,
      "loss": 0.4003,
      "step": 20852
    },
    {
      "epoch": 1.1677119498264084,
      "grad_norm": 1.1579002141952515,
      "learning_rate": 9.977815789473684e-05,
      "loss": 0.4041,
      "step": 20853
    },
    {
      "epoch": 1.1677679471385374,
      "grad_norm": 1.5492885112762451,
      "learning_rate": 9.977789473684212e-05,
      "loss": 0.4028,
      "step": 20854
    },
    {
      "epoch": 1.1678239444506664,
      "grad_norm": 1.4084981679916382,
      "learning_rate": 9.977763157894737e-05,
      "loss": 0.5682,
      "step": 20855
    },
    {
      "epoch": 1.1678799417627954,
      "grad_norm": 1.1969753503799438,
      "learning_rate": 9.977736842105265e-05,
      "loss": 0.4436,
      "step": 20856
    },
    {
      "epoch": 1.1679359390749244,
      "grad_norm": 1.3001867532730103,
      "learning_rate": 9.97771052631579e-05,
      "loss": 0.3737,
      "step": 20857
    },
    {
      "epoch": 1.1679919363870535,
      "grad_norm": 1.2896686792373657,
      "learning_rate": 9.977684210526315e-05,
      "loss": 0.3692,
      "step": 20858
    },
    {
      "epoch": 1.1680479336991825,
      "grad_norm": 1.4139043092727661,
      "learning_rate": 9.977657894736843e-05,
      "loss": 0.4735,
      "step": 20859
    },
    {
      "epoch": 1.1681039310113115,
      "grad_norm": 1.5470494031906128,
      "learning_rate": 9.977631578947369e-05,
      "loss": 0.5146,
      "step": 20860
    },
    {
      "epoch": 1.1681599283234405,
      "grad_norm": 1.162338376045227,
      "learning_rate": 9.977605263157895e-05,
      "loss": 0.3699,
      "step": 20861
    },
    {
      "epoch": 1.1682159256355695,
      "grad_norm": 1.4471659660339355,
      "learning_rate": 9.97757894736842e-05,
      "loss": 0.571,
      "step": 20862
    },
    {
      "epoch": 1.1682719229476985,
      "grad_norm": 1.1782408952713013,
      "learning_rate": 9.977552631578948e-05,
      "loss": 0.4531,
      "step": 20863
    },
    {
      "epoch": 1.1683279202598276,
      "grad_norm": 1.2059881687164307,
      "learning_rate": 9.977526315789474e-05,
      "loss": 0.4108,
      "step": 20864
    },
    {
      "epoch": 1.1683839175719566,
      "grad_norm": 1.3328951597213745,
      "learning_rate": 9.977500000000001e-05,
      "loss": 0.5086,
      "step": 20865
    },
    {
      "epoch": 1.1684399148840856,
      "grad_norm": 1.190344214439392,
      "learning_rate": 9.977473684210526e-05,
      "loss": 0.4355,
      "step": 20866
    },
    {
      "epoch": 1.1684959121962146,
      "grad_norm": 2.1711649894714355,
      "learning_rate": 9.977447368421053e-05,
      "loss": 0.404,
      "step": 20867
    },
    {
      "epoch": 1.1685519095083436,
      "grad_norm": 1.2574708461761475,
      "learning_rate": 9.977421052631579e-05,
      "loss": 0.3654,
      "step": 20868
    },
    {
      "epoch": 1.1686079068204727,
      "grad_norm": 1.353613018989563,
      "learning_rate": 9.977394736842107e-05,
      "loss": 0.396,
      "step": 20869
    },
    {
      "epoch": 1.1686639041326017,
      "grad_norm": 1.3925970792770386,
      "learning_rate": 9.977368421052633e-05,
      "loss": 0.5186,
      "step": 20870
    },
    {
      "epoch": 1.1687199014447307,
      "grad_norm": 1.6180205345153809,
      "learning_rate": 9.977342105263158e-05,
      "loss": 0.4202,
      "step": 20871
    },
    {
      "epoch": 1.1687758987568597,
      "grad_norm": 1.2593497037887573,
      "learning_rate": 9.977315789473684e-05,
      "loss": 0.4175,
      "step": 20872
    },
    {
      "epoch": 1.1688318960689887,
      "grad_norm": 1.4117871522903442,
      "learning_rate": 9.977289473684212e-05,
      "loss": 0.3986,
      "step": 20873
    },
    {
      "epoch": 1.1688878933811178,
      "grad_norm": 1.1125867366790771,
      "learning_rate": 9.977263157894738e-05,
      "loss": 0.296,
      "step": 20874
    },
    {
      "epoch": 1.1689438906932468,
      "grad_norm": 1.1166563034057617,
      "learning_rate": 9.977236842105264e-05,
      "loss": 0.3429,
      "step": 20875
    },
    {
      "epoch": 1.1689998880053758,
      "grad_norm": 1.1332788467407227,
      "learning_rate": 9.97721052631579e-05,
      "loss": 0.3374,
      "step": 20876
    },
    {
      "epoch": 1.1690558853175048,
      "grad_norm": 1.4194594621658325,
      "learning_rate": 9.977184210526316e-05,
      "loss": 0.3917,
      "step": 20877
    },
    {
      "epoch": 1.1691118826296338,
      "grad_norm": 1.3464772701263428,
      "learning_rate": 9.977157894736843e-05,
      "loss": 0.5524,
      "step": 20878
    },
    {
      "epoch": 1.1691678799417629,
      "grad_norm": 1.4248733520507812,
      "learning_rate": 9.977131578947369e-05,
      "loss": 0.6119,
      "step": 20879
    },
    {
      "epoch": 1.1692238772538919,
      "grad_norm": 1.4053601026535034,
      "learning_rate": 9.977105263157895e-05,
      "loss": 0.4178,
      "step": 20880
    },
    {
      "epoch": 1.169279874566021,
      "grad_norm": 1.308698058128357,
      "learning_rate": 9.977078947368421e-05,
      "loss": 0.4344,
      "step": 20881
    },
    {
      "epoch": 1.16933587187815,
      "grad_norm": 1.3528685569763184,
      "learning_rate": 9.977052631578948e-05,
      "loss": 0.5386,
      "step": 20882
    },
    {
      "epoch": 1.169391869190279,
      "grad_norm": 1.090651273727417,
      "learning_rate": 9.977026315789474e-05,
      "loss": 0.4617,
      "step": 20883
    },
    {
      "epoch": 1.169447866502408,
      "grad_norm": 2.5601022243499756,
      "learning_rate": 9.977e-05,
      "loss": 0.4759,
      "step": 20884
    },
    {
      "epoch": 1.169503863814537,
      "grad_norm": 1.2338849306106567,
      "learning_rate": 9.976973684210526e-05,
      "loss": 0.3925,
      "step": 20885
    },
    {
      "epoch": 1.169559861126666,
      "grad_norm": 1.5448415279388428,
      "learning_rate": 9.976947368421053e-05,
      "loss": 0.6681,
      "step": 20886
    },
    {
      "epoch": 1.169615858438795,
      "grad_norm": 1.7976655960083008,
      "learning_rate": 9.97692105263158e-05,
      "loss": 0.4577,
      "step": 20887
    },
    {
      "epoch": 1.169671855750924,
      "grad_norm": 1.1426849365234375,
      "learning_rate": 9.976894736842107e-05,
      "loss": 0.4016,
      "step": 20888
    },
    {
      "epoch": 1.169727853063053,
      "grad_norm": 1.3843919038772583,
      "learning_rate": 9.976868421052631e-05,
      "loss": 0.5273,
      "step": 20889
    },
    {
      "epoch": 1.169783850375182,
      "grad_norm": 1.341097116470337,
      "learning_rate": 9.976842105263159e-05,
      "loss": 0.4748,
      "step": 20890
    },
    {
      "epoch": 1.169839847687311,
      "grad_norm": 1.3938333988189697,
      "learning_rate": 9.976815789473685e-05,
      "loss": 0.586,
      "step": 20891
    },
    {
      "epoch": 1.16989584499944,
      "grad_norm": 1.2222189903259277,
      "learning_rate": 9.976789473684211e-05,
      "loss": 0.4304,
      "step": 20892
    },
    {
      "epoch": 1.1699518423115691,
      "grad_norm": 1.2287477254867554,
      "learning_rate": 9.976763157894738e-05,
      "loss": 0.3932,
      "step": 20893
    },
    {
      "epoch": 1.1700078396236981,
      "grad_norm": 1.4622381925582886,
      "learning_rate": 9.976736842105263e-05,
      "loss": 0.4012,
      "step": 20894
    },
    {
      "epoch": 1.1700638369358272,
      "grad_norm": 1.5046672821044922,
      "learning_rate": 9.97671052631579e-05,
      "loss": 0.5457,
      "step": 20895
    },
    {
      "epoch": 1.1701198342479562,
      "grad_norm": 1.8150392770767212,
      "learning_rate": 9.976684210526316e-05,
      "loss": 0.6839,
      "step": 20896
    },
    {
      "epoch": 1.1701758315600852,
      "grad_norm": 1.2193825244903564,
      "learning_rate": 9.976657894736843e-05,
      "loss": 0.4728,
      "step": 20897
    },
    {
      "epoch": 1.1702318288722142,
      "grad_norm": 1.1769912242889404,
      "learning_rate": 9.976631578947368e-05,
      "loss": 0.3759,
      "step": 20898
    },
    {
      "epoch": 1.1702878261843432,
      "grad_norm": 1.4072461128234863,
      "learning_rate": 9.976605263157895e-05,
      "loss": 0.3223,
      "step": 20899
    },
    {
      "epoch": 1.1703438234964723,
      "grad_norm": 1.457567572593689,
      "learning_rate": 9.976578947368421e-05,
      "loss": 0.5848,
      "step": 20900
    },
    {
      "epoch": 1.1703998208086013,
      "grad_norm": 1.1242045164108276,
      "learning_rate": 9.976552631578949e-05,
      "loss": 0.3526,
      "step": 20901
    },
    {
      "epoch": 1.1704558181207303,
      "grad_norm": 5.152798175811768,
      "learning_rate": 9.976526315789474e-05,
      "loss": 0.5271,
      "step": 20902
    },
    {
      "epoch": 1.1705118154328593,
      "grad_norm": 1.4303909540176392,
      "learning_rate": 9.9765e-05,
      "loss": 0.4677,
      "step": 20903
    },
    {
      "epoch": 1.1705678127449883,
      "grad_norm": 1.189496397972107,
      "learning_rate": 9.976473684210526e-05,
      "loss": 0.3779,
      "step": 20904
    },
    {
      "epoch": 1.1706238100571174,
      "grad_norm": 1.820021629333496,
      "learning_rate": 9.976447368421054e-05,
      "loss": 0.5416,
      "step": 20905
    },
    {
      "epoch": 1.1706798073692464,
      "grad_norm": 2.0237650871276855,
      "learning_rate": 9.97642105263158e-05,
      "loss": 0.6019,
      "step": 20906
    },
    {
      "epoch": 1.1707358046813754,
      "grad_norm": 1.4961506128311157,
      "learning_rate": 9.976394736842106e-05,
      "loss": 0.4167,
      "step": 20907
    },
    {
      "epoch": 1.1707918019935044,
      "grad_norm": 1.4235213994979858,
      "learning_rate": 9.976368421052632e-05,
      "loss": 0.4684,
      "step": 20908
    },
    {
      "epoch": 1.1708477993056334,
      "grad_norm": 1.7053412199020386,
      "learning_rate": 9.976342105263158e-05,
      "loss": 0.404,
      "step": 20909
    },
    {
      "epoch": 1.1709037966177624,
      "grad_norm": 1.4406343698501587,
      "learning_rate": 9.976315789473685e-05,
      "loss": 0.4105,
      "step": 20910
    },
    {
      "epoch": 1.1709597939298915,
      "grad_norm": 1.739396572113037,
      "learning_rate": 9.976289473684211e-05,
      "loss": 0.551,
      "step": 20911
    },
    {
      "epoch": 1.1710157912420205,
      "grad_norm": 1.2890076637268066,
      "learning_rate": 9.976263157894737e-05,
      "loss": 0.4523,
      "step": 20912
    },
    {
      "epoch": 1.1710717885541495,
      "grad_norm": 1.667888879776001,
      "learning_rate": 9.976236842105263e-05,
      "loss": 0.3999,
      "step": 20913
    },
    {
      "epoch": 1.1711277858662785,
      "grad_norm": 1.4856075048446655,
      "learning_rate": 9.97621052631579e-05,
      "loss": 0.5666,
      "step": 20914
    },
    {
      "epoch": 1.1711837831784075,
      "grad_norm": 1.2786290645599365,
      "learning_rate": 9.976184210526316e-05,
      "loss": 0.4664,
      "step": 20915
    },
    {
      "epoch": 1.1712397804905366,
      "grad_norm": 1.2476434707641602,
      "learning_rate": 9.976157894736842e-05,
      "loss": 0.4253,
      "step": 20916
    },
    {
      "epoch": 1.1712957778026656,
      "grad_norm": 1.4082708358764648,
      "learning_rate": 9.976131578947368e-05,
      "loss": 0.4569,
      "step": 20917
    },
    {
      "epoch": 1.1713517751147946,
      "grad_norm": 4.578914642333984,
      "learning_rate": 9.976105263157895e-05,
      "loss": 0.3723,
      "step": 20918
    },
    {
      "epoch": 1.1714077724269236,
      "grad_norm": 1.360572338104248,
      "learning_rate": 9.976078947368421e-05,
      "loss": 0.4925,
      "step": 20919
    },
    {
      "epoch": 1.1714637697390526,
      "grad_norm": 1.3907374143600464,
      "learning_rate": 9.976052631578949e-05,
      "loss": 0.5032,
      "step": 20920
    },
    {
      "epoch": 1.1715197670511817,
      "grad_norm": 1.2744126319885254,
      "learning_rate": 9.976026315789473e-05,
      "loss": 0.3994,
      "step": 20921
    },
    {
      "epoch": 1.1715757643633107,
      "grad_norm": 1.6579738855361938,
      "learning_rate": 9.976000000000001e-05,
      "loss": 0.5932,
      "step": 20922
    },
    {
      "epoch": 1.1716317616754397,
      "grad_norm": 1.3585381507873535,
      "learning_rate": 9.975973684210527e-05,
      "loss": 0.4995,
      "step": 20923
    },
    {
      "epoch": 1.1716877589875687,
      "grad_norm": 1.6113086938858032,
      "learning_rate": 9.975947368421054e-05,
      "loss": 0.3571,
      "step": 20924
    },
    {
      "epoch": 1.1717437562996977,
      "grad_norm": 1.8635375499725342,
      "learning_rate": 9.97592105263158e-05,
      "loss": 0.6094,
      "step": 20925
    },
    {
      "epoch": 1.1717997536118268,
      "grad_norm": 1.3696166276931763,
      "learning_rate": 9.975894736842106e-05,
      "loss": 0.5199,
      "step": 20926
    },
    {
      "epoch": 1.1718557509239558,
      "grad_norm": 1.7059210538864136,
      "learning_rate": 9.975868421052632e-05,
      "loss": 0.6103,
      "step": 20927
    },
    {
      "epoch": 1.1719117482360846,
      "grad_norm": 1.4262621402740479,
      "learning_rate": 9.975842105263158e-05,
      "loss": 0.5573,
      "step": 20928
    },
    {
      "epoch": 1.1719677455482136,
      "grad_norm": 1.6460988521575928,
      "learning_rate": 9.975815789473685e-05,
      "loss": 0.6694,
      "step": 20929
    },
    {
      "epoch": 1.1720237428603426,
      "grad_norm": 1.3685556650161743,
      "learning_rate": 9.975789473684211e-05,
      "loss": 0.4285,
      "step": 20930
    },
    {
      "epoch": 1.1720797401724716,
      "grad_norm": 1.4430515766143799,
      "learning_rate": 9.975763157894737e-05,
      "loss": 0.4029,
      "step": 20931
    },
    {
      "epoch": 1.1721357374846006,
      "grad_norm": 1.3129043579101562,
      "learning_rate": 9.975736842105263e-05,
      "loss": 0.4033,
      "step": 20932
    },
    {
      "epoch": 1.1721917347967297,
      "grad_norm": 1.3137214183807373,
      "learning_rate": 9.97571052631579e-05,
      "loss": 0.3447,
      "step": 20933
    },
    {
      "epoch": 1.1722477321088587,
      "grad_norm": 1.2377527952194214,
      "learning_rate": 9.975684210526316e-05,
      "loss": 0.4837,
      "step": 20934
    },
    {
      "epoch": 1.1723037294209877,
      "grad_norm": 1.4682475328445435,
      "learning_rate": 9.975657894736842e-05,
      "loss": 0.3254,
      "step": 20935
    },
    {
      "epoch": 1.1723597267331167,
      "grad_norm": 1.4442527294158936,
      "learning_rate": 9.975631578947368e-05,
      "loss": 0.5485,
      "step": 20936
    },
    {
      "epoch": 1.1724157240452457,
      "grad_norm": 1.2747081518173218,
      "learning_rate": 9.975605263157896e-05,
      "loss": 0.3679,
      "step": 20937
    },
    {
      "epoch": 1.1724717213573748,
      "grad_norm": 1.3651938438415527,
      "learning_rate": 9.975578947368422e-05,
      "loss": 0.5902,
      "step": 20938
    },
    {
      "epoch": 1.1725277186695038,
      "grad_norm": 1.3734140396118164,
      "learning_rate": 9.975552631578948e-05,
      "loss": 0.4437,
      "step": 20939
    },
    {
      "epoch": 1.1725837159816328,
      "grad_norm": 3.9022696018218994,
      "learning_rate": 9.975526315789474e-05,
      "loss": 0.4237,
      "step": 20940
    },
    {
      "epoch": 1.1726397132937618,
      "grad_norm": 1.5584402084350586,
      "learning_rate": 9.975500000000001e-05,
      "loss": 0.496,
      "step": 20941
    },
    {
      "epoch": 1.1726957106058908,
      "grad_norm": 1.225854516029358,
      "learning_rate": 9.975473684210527e-05,
      "loss": 0.4197,
      "step": 20942
    },
    {
      "epoch": 1.1727517079180199,
      "grad_norm": 1.3228957653045654,
      "learning_rate": 9.975447368421054e-05,
      "loss": 0.4086,
      "step": 20943
    },
    {
      "epoch": 1.1728077052301489,
      "grad_norm": 1.4241491556167603,
      "learning_rate": 9.975421052631579e-05,
      "loss": 0.5034,
      "step": 20944
    },
    {
      "epoch": 1.172863702542278,
      "grad_norm": 1.351749300956726,
      "learning_rate": 9.975394736842105e-05,
      "loss": 0.4078,
      "step": 20945
    },
    {
      "epoch": 1.172919699854407,
      "grad_norm": 1.266481637954712,
      "learning_rate": 9.975368421052632e-05,
      "loss": 0.4345,
      "step": 20946
    },
    {
      "epoch": 1.172975697166536,
      "grad_norm": 1.4176220893859863,
      "learning_rate": 9.975342105263158e-05,
      "loss": 0.5739,
      "step": 20947
    },
    {
      "epoch": 1.173031694478665,
      "grad_norm": 1.3308656215667725,
      "learning_rate": 9.975315789473685e-05,
      "loss": 0.4177,
      "step": 20948
    },
    {
      "epoch": 1.173087691790794,
      "grad_norm": 1.5689510107040405,
      "learning_rate": 9.97528947368421e-05,
      "loss": 0.49,
      "step": 20949
    },
    {
      "epoch": 1.173143689102923,
      "grad_norm": 1.4418233633041382,
      "learning_rate": 9.975263157894737e-05,
      "loss": 0.4703,
      "step": 20950
    },
    {
      "epoch": 1.173199686415052,
      "grad_norm": 1.4436392784118652,
      "learning_rate": 9.975236842105263e-05,
      "loss": 0.4242,
      "step": 20951
    },
    {
      "epoch": 1.173255683727181,
      "grad_norm": 1.423810362815857,
      "learning_rate": 9.975210526315791e-05,
      "loss": 0.3812,
      "step": 20952
    },
    {
      "epoch": 1.17331168103931,
      "grad_norm": 1.2401375770568848,
      "learning_rate": 9.975184210526315e-05,
      "loss": 0.4388,
      "step": 20953
    },
    {
      "epoch": 1.173367678351439,
      "grad_norm": 1.3293055295944214,
      "learning_rate": 9.975157894736843e-05,
      "loss": 0.4964,
      "step": 20954
    },
    {
      "epoch": 1.173423675663568,
      "grad_norm": 1.3621879816055298,
      "learning_rate": 9.975131578947369e-05,
      "loss": 0.4066,
      "step": 20955
    },
    {
      "epoch": 1.173479672975697,
      "grad_norm": 1.1478829383850098,
      "learning_rate": 9.975105263157896e-05,
      "loss": 0.3469,
      "step": 20956
    },
    {
      "epoch": 1.1735356702878261,
      "grad_norm": 1.63117253780365,
      "learning_rate": 9.975078947368422e-05,
      "loss": 0.6342,
      "step": 20957
    },
    {
      "epoch": 1.1735916675999551,
      "grad_norm": 1.3149467706680298,
      "learning_rate": 9.975052631578948e-05,
      "loss": 0.3948,
      "step": 20958
    },
    {
      "epoch": 1.1736476649120842,
      "grad_norm": 1.46638822555542,
      "learning_rate": 9.975026315789474e-05,
      "loss": 0.4828,
      "step": 20959
    },
    {
      "epoch": 1.1737036622242132,
      "grad_norm": 1.3919765949249268,
      "learning_rate": 9.975000000000001e-05,
      "loss": 0.4549,
      "step": 20960
    },
    {
      "epoch": 1.1737596595363422,
      "grad_norm": 1.3568085432052612,
      "learning_rate": 9.974973684210527e-05,
      "loss": 0.4909,
      "step": 20961
    },
    {
      "epoch": 1.1738156568484712,
      "grad_norm": 1.5416512489318848,
      "learning_rate": 9.974947368421053e-05,
      "loss": 0.6284,
      "step": 20962
    },
    {
      "epoch": 1.1738716541606002,
      "grad_norm": 1.1301103830337524,
      "learning_rate": 9.974921052631579e-05,
      "loss": 0.4017,
      "step": 20963
    },
    {
      "epoch": 1.1739276514727293,
      "grad_norm": 1.399775505065918,
      "learning_rate": 9.974894736842105e-05,
      "loss": 0.49,
      "step": 20964
    },
    {
      "epoch": 1.1739836487848583,
      "grad_norm": 1.6052110195159912,
      "learning_rate": 9.974868421052632e-05,
      "loss": 0.5099,
      "step": 20965
    },
    {
      "epoch": 1.1740396460969873,
      "grad_norm": 1.214698314666748,
      "learning_rate": 9.974842105263158e-05,
      "loss": 0.4041,
      "step": 20966
    },
    {
      "epoch": 1.1740956434091163,
      "grad_norm": 1.6586273908615112,
      "learning_rate": 9.974815789473684e-05,
      "loss": 0.4254,
      "step": 20967
    },
    {
      "epoch": 1.1741516407212453,
      "grad_norm": 1.307340383529663,
      "learning_rate": 9.97478947368421e-05,
      "loss": 0.4385,
      "step": 20968
    },
    {
      "epoch": 1.1742076380333744,
      "grad_norm": 1.3171969652175903,
      "learning_rate": 9.974763157894738e-05,
      "loss": 0.5378,
      "step": 20969
    },
    {
      "epoch": 1.1742636353455034,
      "grad_norm": 1.87435781955719,
      "learning_rate": 9.974736842105264e-05,
      "loss": 0.5037,
      "step": 20970
    },
    {
      "epoch": 1.1743196326576324,
      "grad_norm": 1.3954530954360962,
      "learning_rate": 9.97471052631579e-05,
      "loss": 0.5401,
      "step": 20971
    },
    {
      "epoch": 1.1743756299697614,
      "grad_norm": 1.3167064189910889,
      "learning_rate": 9.974684210526316e-05,
      "loss": 0.4602,
      "step": 20972
    },
    {
      "epoch": 1.1744316272818904,
      "grad_norm": 1.5326517820358276,
      "learning_rate": 9.974657894736843e-05,
      "loss": 0.4996,
      "step": 20973
    },
    {
      "epoch": 1.1744876245940195,
      "grad_norm": 1.471760869026184,
      "learning_rate": 9.974631578947369e-05,
      "loss": 0.3486,
      "step": 20974
    },
    {
      "epoch": 1.1745436219061485,
      "grad_norm": 1.2557377815246582,
      "learning_rate": 9.974605263157896e-05,
      "loss": 0.4018,
      "step": 20975
    },
    {
      "epoch": 1.1745996192182775,
      "grad_norm": 1.3838822841644287,
      "learning_rate": 9.974578947368421e-05,
      "loss": 0.4234,
      "step": 20976
    },
    {
      "epoch": 1.1746556165304065,
      "grad_norm": 1.2633038759231567,
      "learning_rate": 9.974552631578948e-05,
      "loss": 0.4,
      "step": 20977
    },
    {
      "epoch": 1.1747116138425355,
      "grad_norm": 1.2640166282653809,
      "learning_rate": 9.974526315789474e-05,
      "loss": 0.3977,
      "step": 20978
    },
    {
      "epoch": 1.1747676111546645,
      "grad_norm": 1.1059231758117676,
      "learning_rate": 9.9745e-05,
      "loss": 0.3712,
      "step": 20979
    },
    {
      "epoch": 1.1748236084667936,
      "grad_norm": 1.2637633085250854,
      "learning_rate": 9.974473684210527e-05,
      "loss": 0.5684,
      "step": 20980
    },
    {
      "epoch": 1.1748796057789226,
      "grad_norm": 1.5723506212234497,
      "learning_rate": 9.974447368421052e-05,
      "loss": 0.4166,
      "step": 20981
    },
    {
      "epoch": 1.1749356030910516,
      "grad_norm": 1.3939402103424072,
      "learning_rate": 9.97442105263158e-05,
      "loss": 0.4191,
      "step": 20982
    },
    {
      "epoch": 1.1749916004031806,
      "grad_norm": 1.2307891845703125,
      "learning_rate": 9.974394736842105e-05,
      "loss": 0.3587,
      "step": 20983
    },
    {
      "epoch": 1.1750475977153096,
      "grad_norm": 1.5806177854537964,
      "learning_rate": 9.974368421052633e-05,
      "loss": 0.5389,
      "step": 20984
    },
    {
      "epoch": 1.1751035950274387,
      "grad_norm": 1.2895103693008423,
      "learning_rate": 9.974342105263159e-05,
      "loss": 0.469,
      "step": 20985
    },
    {
      "epoch": 1.1751595923395677,
      "grad_norm": 1.590126395225525,
      "learning_rate": 9.974315789473685e-05,
      "loss": 0.5257,
      "step": 20986
    },
    {
      "epoch": 1.1752155896516967,
      "grad_norm": 1.3139359951019287,
      "learning_rate": 9.97428947368421e-05,
      "loss": 0.4209,
      "step": 20987
    },
    {
      "epoch": 1.1752715869638257,
      "grad_norm": 1.2274155616760254,
      "learning_rate": 9.974263157894738e-05,
      "loss": 0.4931,
      "step": 20988
    },
    {
      "epoch": 1.1753275842759547,
      "grad_norm": 1.2153725624084473,
      "learning_rate": 9.974236842105264e-05,
      "loss": 0.5214,
      "step": 20989
    },
    {
      "epoch": 1.1753835815880838,
      "grad_norm": 1.3513826131820679,
      "learning_rate": 9.97421052631579e-05,
      "loss": 0.4725,
      "step": 20990
    },
    {
      "epoch": 1.1754395789002128,
      "grad_norm": 1.201668381690979,
      "learning_rate": 9.974184210526316e-05,
      "loss": 0.5036,
      "step": 20991
    },
    {
      "epoch": 1.1754955762123418,
      "grad_norm": 1.4883041381835938,
      "learning_rate": 9.974157894736843e-05,
      "loss": 0.4364,
      "step": 20992
    },
    {
      "epoch": 1.1755515735244708,
      "grad_norm": 1.4872490167617798,
      "learning_rate": 9.974131578947369e-05,
      "loss": 0.446,
      "step": 20993
    },
    {
      "epoch": 1.1756075708365998,
      "grad_norm": 1.2378628253936768,
      "learning_rate": 9.974105263157895e-05,
      "loss": 0.4013,
      "step": 20994
    },
    {
      "epoch": 1.1756635681487289,
      "grad_norm": 1.235908031463623,
      "learning_rate": 9.974078947368421e-05,
      "loss": 0.3845,
      "step": 20995
    },
    {
      "epoch": 1.1757195654608579,
      "grad_norm": 1.1609922647476196,
      "learning_rate": 9.974052631578947e-05,
      "loss": 0.5331,
      "step": 20996
    },
    {
      "epoch": 1.175775562772987,
      "grad_norm": 1.4406769275665283,
      "learning_rate": 9.974026315789474e-05,
      "loss": 0.4935,
      "step": 20997
    },
    {
      "epoch": 1.175831560085116,
      "grad_norm": 1.3535735607147217,
      "learning_rate": 9.974e-05,
      "loss": 0.3864,
      "step": 20998
    },
    {
      "epoch": 1.175887557397245,
      "grad_norm": 1.3318134546279907,
      "learning_rate": 9.973973684210526e-05,
      "loss": 0.4113,
      "step": 20999
    },
    {
      "epoch": 1.175943554709374,
      "grad_norm": 1.5921144485473633,
      "learning_rate": 9.973947368421052e-05,
      "loss": 0.3982,
      "step": 21000
    },
    {
      "epoch": 1.175999552021503,
      "grad_norm": 1.7537912130355835,
      "learning_rate": 9.97392105263158e-05,
      "loss": 0.4266,
      "step": 21001
    },
    {
      "epoch": 1.176055549333632,
      "grad_norm": 1.2887009382247925,
      "learning_rate": 9.973894736842106e-05,
      "loss": 0.4111,
      "step": 21002
    },
    {
      "epoch": 1.176111546645761,
      "grad_norm": 1.4059815406799316,
      "learning_rate": 9.973868421052633e-05,
      "loss": 0.3887,
      "step": 21003
    },
    {
      "epoch": 1.17616754395789,
      "grad_norm": 1.3824046850204468,
      "learning_rate": 9.973842105263158e-05,
      "loss": 0.4825,
      "step": 21004
    },
    {
      "epoch": 1.176223541270019,
      "grad_norm": 2.390695571899414,
      "learning_rate": 9.973815789473685e-05,
      "loss": 0.5852,
      "step": 21005
    },
    {
      "epoch": 1.176279538582148,
      "grad_norm": 1.2309800386428833,
      "learning_rate": 9.973789473684211e-05,
      "loss": 0.3685,
      "step": 21006
    },
    {
      "epoch": 1.176335535894277,
      "grad_norm": 1.358850121498108,
      "learning_rate": 9.973763157894738e-05,
      "loss": 0.3702,
      "step": 21007
    },
    {
      "epoch": 1.176391533206406,
      "grad_norm": 2.115166187286377,
      "learning_rate": 9.973736842105263e-05,
      "loss": 0.5503,
      "step": 21008
    },
    {
      "epoch": 1.1764475305185351,
      "grad_norm": 1.2767928838729858,
      "learning_rate": 9.97371052631579e-05,
      "loss": 0.4729,
      "step": 21009
    },
    {
      "epoch": 1.1765035278306641,
      "grad_norm": 1.5280072689056396,
      "learning_rate": 9.973684210526316e-05,
      "loss": 0.3253,
      "step": 21010
    },
    {
      "epoch": 1.1765595251427932,
      "grad_norm": 1.5275877714157104,
      "learning_rate": 9.973657894736843e-05,
      "loss": 0.4214,
      "step": 21011
    },
    {
      "epoch": 1.1766155224549222,
      "grad_norm": 1.173607587814331,
      "learning_rate": 9.97363157894737e-05,
      "loss": 0.3094,
      "step": 21012
    },
    {
      "epoch": 1.1766715197670512,
      "grad_norm": 1.2253994941711426,
      "learning_rate": 9.973605263157894e-05,
      "loss": 0.4247,
      "step": 21013
    },
    {
      "epoch": 1.1767275170791802,
      "grad_norm": 1.1790021657943726,
      "learning_rate": 9.973578947368421e-05,
      "loss": 0.353,
      "step": 21014
    },
    {
      "epoch": 1.1767835143913092,
      "grad_norm": 1.2789949178695679,
      "learning_rate": 9.973552631578947e-05,
      "loss": 0.3371,
      "step": 21015
    },
    {
      "epoch": 1.1768395117034383,
      "grad_norm": 1.2025184631347656,
      "learning_rate": 9.973526315789475e-05,
      "loss": 0.474,
      "step": 21016
    },
    {
      "epoch": 1.1768955090155673,
      "grad_norm": 1.3859963417053223,
      "learning_rate": 9.9735e-05,
      "loss": 0.5285,
      "step": 21017
    },
    {
      "epoch": 1.1769515063276963,
      "grad_norm": 1.4734511375427246,
      "learning_rate": 9.973473684210527e-05,
      "loss": 0.4577,
      "step": 21018
    },
    {
      "epoch": 1.1770075036398253,
      "grad_norm": 1.6155064105987549,
      "learning_rate": 9.973447368421053e-05,
      "loss": 0.4629,
      "step": 21019
    },
    {
      "epoch": 1.1770635009519543,
      "grad_norm": 1.3375678062438965,
      "learning_rate": 9.97342105263158e-05,
      "loss": 0.3894,
      "step": 21020
    },
    {
      "epoch": 1.1771194982640834,
      "grad_norm": 1.2337805032730103,
      "learning_rate": 9.973394736842106e-05,
      "loss": 0.3927,
      "step": 21021
    },
    {
      "epoch": 1.1771754955762124,
      "grad_norm": 1.4827440977096558,
      "learning_rate": 9.973368421052632e-05,
      "loss": 0.4476,
      "step": 21022
    },
    {
      "epoch": 1.1772314928883414,
      "grad_norm": 1.2658737897872925,
      "learning_rate": 9.973342105263158e-05,
      "loss": 0.4027,
      "step": 21023
    },
    {
      "epoch": 1.1772874902004704,
      "grad_norm": 1.3625433444976807,
      "learning_rate": 9.973315789473685e-05,
      "loss": 0.4178,
      "step": 21024
    },
    {
      "epoch": 1.1773434875125994,
      "grad_norm": 1.4328525066375732,
      "learning_rate": 9.973289473684211e-05,
      "loss": 0.4612,
      "step": 21025
    },
    {
      "epoch": 1.1773994848247284,
      "grad_norm": 1.4753912687301636,
      "learning_rate": 9.973263157894737e-05,
      "loss": 0.4367,
      "step": 21026
    },
    {
      "epoch": 1.1774554821368575,
      "grad_norm": 1.2543789148330688,
      "learning_rate": 9.973236842105263e-05,
      "loss": 0.4802,
      "step": 21027
    },
    {
      "epoch": 1.1775114794489865,
      "grad_norm": 1.206464409828186,
      "learning_rate": 9.97321052631579e-05,
      "loss": 0.4139,
      "step": 21028
    },
    {
      "epoch": 1.1775674767611155,
      "grad_norm": 1.302475094795227,
      "learning_rate": 9.973184210526316e-05,
      "loss": 0.4461,
      "step": 21029
    },
    {
      "epoch": 1.1776234740732445,
      "grad_norm": 1.0936827659606934,
      "learning_rate": 9.973157894736842e-05,
      "loss": 0.4318,
      "step": 21030
    },
    {
      "epoch": 1.1776794713853735,
      "grad_norm": 1.2795578241348267,
      "learning_rate": 9.973131578947368e-05,
      "loss": 0.3422,
      "step": 21031
    },
    {
      "epoch": 1.1777354686975026,
      "grad_norm": 1.5770725011825562,
      "learning_rate": 9.973105263157894e-05,
      "loss": 0.4887,
      "step": 21032
    },
    {
      "epoch": 1.1777914660096316,
      "grad_norm": 1.2700650691986084,
      "learning_rate": 9.973078947368422e-05,
      "loss": 0.5678,
      "step": 21033
    },
    {
      "epoch": 1.1778474633217606,
      "grad_norm": 1.371291160583496,
      "learning_rate": 9.973052631578948e-05,
      "loss": 0.4998,
      "step": 21034
    },
    {
      "epoch": 1.1779034606338896,
      "grad_norm": 1.370296597480774,
      "learning_rate": 9.973026315789475e-05,
      "loss": 0.4893,
      "step": 21035
    },
    {
      "epoch": 1.1779594579460186,
      "grad_norm": 1.5093014240264893,
      "learning_rate": 9.973e-05,
      "loss": 0.4992,
      "step": 21036
    },
    {
      "epoch": 1.1780154552581477,
      "grad_norm": 1.3816587924957275,
      "learning_rate": 9.972973684210527e-05,
      "loss": 0.3892,
      "step": 21037
    },
    {
      "epoch": 1.1780714525702767,
      "grad_norm": 1.5106549263000488,
      "learning_rate": 9.972947368421053e-05,
      "loss": 0.4294,
      "step": 21038
    },
    {
      "epoch": 1.1781274498824057,
      "grad_norm": 1.2390880584716797,
      "learning_rate": 9.97292105263158e-05,
      "loss": 0.3547,
      "step": 21039
    },
    {
      "epoch": 1.1781834471945347,
      "grad_norm": 1.52642822265625,
      "learning_rate": 9.972894736842106e-05,
      "loss": 0.5257,
      "step": 21040
    },
    {
      "epoch": 1.1782394445066637,
      "grad_norm": 1.2675113677978516,
      "learning_rate": 9.972868421052632e-05,
      "loss": 0.4144,
      "step": 21041
    },
    {
      "epoch": 1.1782954418187928,
      "grad_norm": 1.4749233722686768,
      "learning_rate": 9.972842105263158e-05,
      "loss": 0.4099,
      "step": 21042
    },
    {
      "epoch": 1.1783514391309218,
      "grad_norm": 1.2252509593963623,
      "learning_rate": 9.972815789473685e-05,
      "loss": 0.4446,
      "step": 21043
    },
    {
      "epoch": 1.1784074364430508,
      "grad_norm": 1.230692982673645,
      "learning_rate": 9.972789473684211e-05,
      "loss": 0.4159,
      "step": 21044
    },
    {
      "epoch": 1.1784634337551798,
      "grad_norm": 1.4128847122192383,
      "learning_rate": 9.972763157894737e-05,
      "loss": 0.4486,
      "step": 21045
    },
    {
      "epoch": 1.1785194310673088,
      "grad_norm": 1.276092290878296,
      "learning_rate": 9.972736842105263e-05,
      "loss": 0.4334,
      "step": 21046
    },
    {
      "epoch": 1.1785754283794379,
      "grad_norm": 1.8333954811096191,
      "learning_rate": 9.97271052631579e-05,
      "loss": 0.5329,
      "step": 21047
    },
    {
      "epoch": 1.1786314256915669,
      "grad_norm": 1.220476746559143,
      "learning_rate": 9.972684210526317e-05,
      "loss": 0.425,
      "step": 21048
    },
    {
      "epoch": 1.1786874230036959,
      "grad_norm": 1.54896879196167,
      "learning_rate": 9.972657894736843e-05,
      "loss": 0.5269,
      "step": 21049
    },
    {
      "epoch": 1.178743420315825,
      "grad_norm": 1.5744624137878418,
      "learning_rate": 9.972631578947369e-05,
      "loss": 0.3989,
      "step": 21050
    },
    {
      "epoch": 1.178799417627954,
      "grad_norm": 1.5536669492721558,
      "learning_rate": 9.972605263157895e-05,
      "loss": 0.4169,
      "step": 21051
    },
    {
      "epoch": 1.178855414940083,
      "grad_norm": 1.5690683126449585,
      "learning_rate": 9.972578947368422e-05,
      "loss": 0.4538,
      "step": 21052
    },
    {
      "epoch": 1.178911412252212,
      "grad_norm": 1.5268265008926392,
      "learning_rate": 9.972552631578948e-05,
      "loss": 0.3558,
      "step": 21053
    },
    {
      "epoch": 1.178967409564341,
      "grad_norm": 1.5727897882461548,
      "learning_rate": 9.972526315789474e-05,
      "loss": 0.6762,
      "step": 21054
    },
    {
      "epoch": 1.17902340687647,
      "grad_norm": 1.347290277481079,
      "learning_rate": 9.9725e-05,
      "loss": 0.4515,
      "step": 21055
    },
    {
      "epoch": 1.179079404188599,
      "grad_norm": 1.4351496696472168,
      "learning_rate": 9.972473684210527e-05,
      "loss": 0.5595,
      "step": 21056
    },
    {
      "epoch": 1.179135401500728,
      "grad_norm": 1.2999906539916992,
      "learning_rate": 9.972447368421053e-05,
      "loss": 0.478,
      "step": 21057
    },
    {
      "epoch": 1.179191398812857,
      "grad_norm": 1.2803971767425537,
      "learning_rate": 9.97242105263158e-05,
      "loss": 0.503,
      "step": 21058
    },
    {
      "epoch": 1.179247396124986,
      "grad_norm": 2.6762804985046387,
      "learning_rate": 9.972394736842105e-05,
      "loss": 0.5161,
      "step": 21059
    },
    {
      "epoch": 1.179303393437115,
      "grad_norm": 1.2523714303970337,
      "learning_rate": 9.972368421052632e-05,
      "loss": 0.3979,
      "step": 21060
    },
    {
      "epoch": 1.1793593907492441,
      "grad_norm": 1.3335241079330444,
      "learning_rate": 9.972342105263158e-05,
      "loss": 0.424,
      "step": 21061
    },
    {
      "epoch": 1.1794153880613731,
      "grad_norm": 1.3811438083648682,
      "learning_rate": 9.972315789473686e-05,
      "loss": 0.4412,
      "step": 21062
    },
    {
      "epoch": 1.1794713853735022,
      "grad_norm": 1.3896058797836304,
      "learning_rate": 9.97228947368421e-05,
      "loss": 0.5693,
      "step": 21063
    },
    {
      "epoch": 1.1795273826856312,
      "grad_norm": 1.2304675579071045,
      "learning_rate": 9.972263157894738e-05,
      "loss": 0.3832,
      "step": 21064
    },
    {
      "epoch": 1.1795833799977602,
      "grad_norm": 1.9563837051391602,
      "learning_rate": 9.972236842105264e-05,
      "loss": 0.4691,
      "step": 21065
    },
    {
      "epoch": 1.1796393773098892,
      "grad_norm": 1.3286216259002686,
      "learning_rate": 9.97221052631579e-05,
      "loss": 0.446,
      "step": 21066
    },
    {
      "epoch": 1.1796953746220182,
      "grad_norm": 1.3801134824752808,
      "learning_rate": 9.972184210526317e-05,
      "loss": 0.5168,
      "step": 21067
    },
    {
      "epoch": 1.1797513719341473,
      "grad_norm": 1.3077819347381592,
      "learning_rate": 9.972157894736842e-05,
      "loss": 0.4792,
      "step": 21068
    },
    {
      "epoch": 1.1798073692462763,
      "grad_norm": 1.4988596439361572,
      "learning_rate": 9.972131578947369e-05,
      "loss": 0.5545,
      "step": 21069
    },
    {
      "epoch": 1.1798633665584053,
      "grad_norm": 1.228618860244751,
      "learning_rate": 9.972105263157895e-05,
      "loss": 0.407,
      "step": 21070
    },
    {
      "epoch": 1.1799193638705343,
      "grad_norm": 1.342016339302063,
      "learning_rate": 9.972078947368422e-05,
      "loss": 0.4125,
      "step": 21071
    },
    {
      "epoch": 1.1799753611826633,
      "grad_norm": 1.7181639671325684,
      "learning_rate": 9.972052631578948e-05,
      "loss": 0.3964,
      "step": 21072
    },
    {
      "epoch": 1.1800313584947921,
      "grad_norm": 1.7656135559082031,
      "learning_rate": 9.972026315789474e-05,
      "loss": 0.532,
      "step": 21073
    },
    {
      "epoch": 1.1800873558069211,
      "grad_norm": 1.4386123418807983,
      "learning_rate": 9.972e-05,
      "loss": 0.4891,
      "step": 21074
    },
    {
      "epoch": 1.1801433531190502,
      "grad_norm": 1.3285212516784668,
      "learning_rate": 9.971973684210527e-05,
      "loss": 0.4108,
      "step": 21075
    },
    {
      "epoch": 1.1801993504311792,
      "grad_norm": 1.3960325717926025,
      "learning_rate": 9.971947368421053e-05,
      "loss": 0.3639,
      "step": 21076
    },
    {
      "epoch": 1.1802553477433082,
      "grad_norm": 1.492347002029419,
      "learning_rate": 9.97192105263158e-05,
      "loss": 0.5306,
      "step": 21077
    },
    {
      "epoch": 1.1803113450554372,
      "grad_norm": 1.1340129375457764,
      "learning_rate": 9.971894736842105e-05,
      "loss": 0.3658,
      "step": 21078
    },
    {
      "epoch": 1.1803673423675662,
      "grad_norm": 1.175807237625122,
      "learning_rate": 9.971868421052633e-05,
      "loss": 0.344,
      "step": 21079
    },
    {
      "epoch": 1.1804233396796953,
      "grad_norm": 1.6824955940246582,
      "learning_rate": 9.971842105263159e-05,
      "loss": 0.5288,
      "step": 21080
    },
    {
      "epoch": 1.1804793369918243,
      "grad_norm": 1.2158433198928833,
      "learning_rate": 9.971815789473685e-05,
      "loss": 0.3972,
      "step": 21081
    },
    {
      "epoch": 1.1805353343039533,
      "grad_norm": 1.6033775806427002,
      "learning_rate": 9.97178947368421e-05,
      "loss": 0.4921,
      "step": 21082
    },
    {
      "epoch": 1.1805913316160823,
      "grad_norm": 1.1867610216140747,
      "learning_rate": 9.971763157894737e-05,
      "loss": 0.3832,
      "step": 21083
    },
    {
      "epoch": 1.1806473289282113,
      "grad_norm": 1.357165813446045,
      "learning_rate": 9.971736842105264e-05,
      "loss": 0.4266,
      "step": 21084
    },
    {
      "epoch": 1.1807033262403404,
      "grad_norm": 1.2187163829803467,
      "learning_rate": 9.97171052631579e-05,
      "loss": 0.4727,
      "step": 21085
    },
    {
      "epoch": 1.1807593235524694,
      "grad_norm": 1.299239993095398,
      "learning_rate": 9.971684210526316e-05,
      "loss": 0.4482,
      "step": 21086
    },
    {
      "epoch": 1.1808153208645984,
      "grad_norm": 1.2567377090454102,
      "learning_rate": 9.971657894736842e-05,
      "loss": 0.445,
      "step": 21087
    },
    {
      "epoch": 1.1808713181767274,
      "grad_norm": 1.2322558164596558,
      "learning_rate": 9.971631578947369e-05,
      "loss": 0.3458,
      "step": 21088
    },
    {
      "epoch": 1.1809273154888564,
      "grad_norm": 1.7484732866287231,
      "learning_rate": 9.971605263157895e-05,
      "loss": 0.4452,
      "step": 21089
    },
    {
      "epoch": 1.1809833128009855,
      "grad_norm": 1.2590219974517822,
      "learning_rate": 9.971578947368422e-05,
      "loss": 0.4281,
      "step": 21090
    },
    {
      "epoch": 1.1810393101131145,
      "grad_norm": 1.654700756072998,
      "learning_rate": 9.971552631578947e-05,
      "loss": 0.5242,
      "step": 21091
    },
    {
      "epoch": 1.1810953074252435,
      "grad_norm": 1.3631558418273926,
      "learning_rate": 9.971526315789474e-05,
      "loss": 0.4196,
      "step": 21092
    },
    {
      "epoch": 1.1811513047373725,
      "grad_norm": 1.3610966205596924,
      "learning_rate": 9.9715e-05,
      "loss": 0.4374,
      "step": 21093
    },
    {
      "epoch": 1.1812073020495015,
      "grad_norm": 1.3292502164840698,
      "learning_rate": 9.971473684210528e-05,
      "loss": 0.5613,
      "step": 21094
    },
    {
      "epoch": 1.1812632993616305,
      "grad_norm": 1.3234503269195557,
      "learning_rate": 9.971447368421054e-05,
      "loss": 0.4301,
      "step": 21095
    },
    {
      "epoch": 1.1813192966737596,
      "grad_norm": 1.2691986560821533,
      "learning_rate": 9.97142105263158e-05,
      "loss": 0.3603,
      "step": 21096
    },
    {
      "epoch": 1.1813752939858886,
      "grad_norm": 1.2768298387527466,
      "learning_rate": 9.971394736842106e-05,
      "loss": 0.3793,
      "step": 21097
    },
    {
      "epoch": 1.1814312912980176,
      "grad_norm": 1.067457914352417,
      "learning_rate": 9.971368421052633e-05,
      "loss": 0.5021,
      "step": 21098
    },
    {
      "epoch": 1.1814872886101466,
      "grad_norm": 1.585533857345581,
      "learning_rate": 9.971342105263159e-05,
      "loss": 0.7983,
      "step": 21099
    },
    {
      "epoch": 1.1815432859222756,
      "grad_norm": 1.301196575164795,
      "learning_rate": 9.971315789473683e-05,
      "loss": 0.4705,
      "step": 21100
    },
    {
      "epoch": 1.1815992832344047,
      "grad_norm": 1.3239070177078247,
      "learning_rate": 9.971289473684211e-05,
      "loss": 0.4195,
      "step": 21101
    },
    {
      "epoch": 1.1816552805465337,
      "grad_norm": 1.5636088848114014,
      "learning_rate": 9.971263157894737e-05,
      "loss": 0.544,
      "step": 21102
    },
    {
      "epoch": 1.1817112778586627,
      "grad_norm": 1.20651113986969,
      "learning_rate": 9.971236842105264e-05,
      "loss": 0.4805,
      "step": 21103
    },
    {
      "epoch": 1.1817672751707917,
      "grad_norm": 1.2414714097976685,
      "learning_rate": 9.97121052631579e-05,
      "loss": 0.3596,
      "step": 21104
    },
    {
      "epoch": 1.1818232724829207,
      "grad_norm": 1.429947853088379,
      "learning_rate": 9.971184210526316e-05,
      "loss": 0.4822,
      "step": 21105
    },
    {
      "epoch": 1.1818792697950498,
      "grad_norm": 1.3667641878128052,
      "learning_rate": 9.971157894736842e-05,
      "loss": 0.3992,
      "step": 21106
    },
    {
      "epoch": 1.1819352671071788,
      "grad_norm": 1.1207997798919678,
      "learning_rate": 9.97113157894737e-05,
      "loss": 0.3619,
      "step": 21107
    },
    {
      "epoch": 1.1819912644193078,
      "grad_norm": 1.050496220588684,
      "learning_rate": 9.971105263157895e-05,
      "loss": 0.3202,
      "step": 21108
    },
    {
      "epoch": 1.1820472617314368,
      "grad_norm": 1.368922472000122,
      "learning_rate": 9.971078947368421e-05,
      "loss": 0.585,
      "step": 21109
    },
    {
      "epoch": 1.1821032590435658,
      "grad_norm": 1.4292868375778198,
      "learning_rate": 9.971052631578947e-05,
      "loss": 0.3907,
      "step": 21110
    },
    {
      "epoch": 1.1821592563556949,
      "grad_norm": 1.2813223600387573,
      "learning_rate": 9.971026315789475e-05,
      "loss": 0.3623,
      "step": 21111
    },
    {
      "epoch": 1.1822152536678239,
      "grad_norm": 1.4284342527389526,
      "learning_rate": 9.971e-05,
      "loss": 0.4354,
      "step": 21112
    },
    {
      "epoch": 1.182271250979953,
      "grad_norm": 1.565657615661621,
      "learning_rate": 9.970973684210528e-05,
      "loss": 0.4779,
      "step": 21113
    },
    {
      "epoch": 1.182327248292082,
      "grad_norm": 1.3310754299163818,
      "learning_rate": 9.970947368421053e-05,
      "loss": 0.4205,
      "step": 21114
    },
    {
      "epoch": 1.182383245604211,
      "grad_norm": 1.2147763967514038,
      "learning_rate": 9.97092105263158e-05,
      "loss": 0.4249,
      "step": 21115
    },
    {
      "epoch": 1.18243924291634,
      "grad_norm": 1.3342591524124146,
      "learning_rate": 9.970894736842106e-05,
      "loss": 0.4208,
      "step": 21116
    },
    {
      "epoch": 1.182495240228469,
      "grad_norm": 1.507914662361145,
      "learning_rate": 9.970868421052632e-05,
      "loss": 0.6189,
      "step": 21117
    },
    {
      "epoch": 1.182551237540598,
      "grad_norm": 1.5464972257614136,
      "learning_rate": 9.970842105263158e-05,
      "loss": 0.5017,
      "step": 21118
    },
    {
      "epoch": 1.182607234852727,
      "grad_norm": 1.3173556327819824,
      "learning_rate": 9.970815789473684e-05,
      "loss": 0.4078,
      "step": 21119
    },
    {
      "epoch": 1.182663232164856,
      "grad_norm": 1.3701661825180054,
      "learning_rate": 9.970789473684211e-05,
      "loss": 0.5615,
      "step": 21120
    },
    {
      "epoch": 1.182719229476985,
      "grad_norm": 1.4368195533752441,
      "learning_rate": 9.970763157894737e-05,
      "loss": 0.4509,
      "step": 21121
    },
    {
      "epoch": 1.182775226789114,
      "grad_norm": 1.326619029045105,
      "learning_rate": 9.970736842105264e-05,
      "loss": 0.4387,
      "step": 21122
    },
    {
      "epoch": 1.182831224101243,
      "grad_norm": 1.341138243675232,
      "learning_rate": 9.970710526315789e-05,
      "loss": 0.4683,
      "step": 21123
    },
    {
      "epoch": 1.182887221413372,
      "grad_norm": 1.3173452615737915,
      "learning_rate": 9.970684210526316e-05,
      "loss": 0.4314,
      "step": 21124
    },
    {
      "epoch": 1.1829432187255011,
      "grad_norm": 1.32731032371521,
      "learning_rate": 9.970657894736842e-05,
      "loss": 0.431,
      "step": 21125
    },
    {
      "epoch": 1.1829992160376301,
      "grad_norm": 1.4874069690704346,
      "learning_rate": 9.97063157894737e-05,
      "loss": 0.5038,
      "step": 21126
    },
    {
      "epoch": 1.1830552133497592,
      "grad_norm": 1.3662054538726807,
      "learning_rate": 9.970605263157896e-05,
      "loss": 0.4337,
      "step": 21127
    },
    {
      "epoch": 1.1831112106618882,
      "grad_norm": 1.2005529403686523,
      "learning_rate": 9.970578947368422e-05,
      "loss": 0.3957,
      "step": 21128
    },
    {
      "epoch": 1.1831672079740172,
      "grad_norm": 1.4747321605682373,
      "learning_rate": 9.970552631578948e-05,
      "loss": 0.4642,
      "step": 21129
    },
    {
      "epoch": 1.1832232052861462,
      "grad_norm": 1.3802555799484253,
      "learning_rate": 9.970526315789475e-05,
      "loss": 0.4484,
      "step": 21130
    },
    {
      "epoch": 1.1832792025982752,
      "grad_norm": 1.442726731300354,
      "learning_rate": 9.970500000000001e-05,
      "loss": 0.4193,
      "step": 21131
    },
    {
      "epoch": 1.1833351999104043,
      "grad_norm": 1.507625937461853,
      "learning_rate": 9.970473684210527e-05,
      "loss": 0.5281,
      "step": 21132
    },
    {
      "epoch": 1.1833911972225333,
      "grad_norm": 1.5044076442718506,
      "learning_rate": 9.970447368421053e-05,
      "loss": 0.4702,
      "step": 21133
    },
    {
      "epoch": 1.1834471945346623,
      "grad_norm": 2.6730031967163086,
      "learning_rate": 9.970421052631579e-05,
      "loss": 0.338,
      "step": 21134
    },
    {
      "epoch": 1.1835031918467913,
      "grad_norm": 1.2748754024505615,
      "learning_rate": 9.970394736842106e-05,
      "loss": 0.5369,
      "step": 21135
    },
    {
      "epoch": 1.1835591891589203,
      "grad_norm": 1.3605180978775024,
      "learning_rate": 9.970368421052632e-05,
      "loss": 0.6243,
      "step": 21136
    },
    {
      "epoch": 1.1836151864710494,
      "grad_norm": 1.4342129230499268,
      "learning_rate": 9.970342105263158e-05,
      "loss": 0.3888,
      "step": 21137
    },
    {
      "epoch": 1.1836711837831784,
      "grad_norm": 1.2869508266448975,
      "learning_rate": 9.970315789473684e-05,
      "loss": 0.5126,
      "step": 21138
    },
    {
      "epoch": 1.1837271810953074,
      "grad_norm": 1.6351805925369263,
      "learning_rate": 9.970289473684211e-05,
      "loss": 0.5752,
      "step": 21139
    },
    {
      "epoch": 1.1837831784074364,
      "grad_norm": 1.45833158493042,
      "learning_rate": 9.970263157894737e-05,
      "loss": 0.3881,
      "step": 21140
    },
    {
      "epoch": 1.1838391757195654,
      "grad_norm": 1.2941596508026123,
      "learning_rate": 9.970236842105263e-05,
      "loss": 0.5191,
      "step": 21141
    },
    {
      "epoch": 1.1838951730316944,
      "grad_norm": 1.3377100229263306,
      "learning_rate": 9.970210526315789e-05,
      "loss": 0.3507,
      "step": 21142
    },
    {
      "epoch": 1.1839511703438235,
      "grad_norm": 1.2845380306243896,
      "learning_rate": 9.970184210526317e-05,
      "loss": 0.4726,
      "step": 21143
    },
    {
      "epoch": 1.1840071676559525,
      "grad_norm": 1.2385921478271484,
      "learning_rate": 9.970157894736843e-05,
      "loss": 0.4044,
      "step": 21144
    },
    {
      "epoch": 1.1840631649680815,
      "grad_norm": 1.1426398754119873,
      "learning_rate": 9.97013157894737e-05,
      "loss": 0.4968,
      "step": 21145
    },
    {
      "epoch": 1.1841191622802105,
      "grad_norm": 1.7909942865371704,
      "learning_rate": 9.970105263157895e-05,
      "loss": 0.3966,
      "step": 21146
    },
    {
      "epoch": 1.1841751595923395,
      "grad_norm": 2.319905996322632,
      "learning_rate": 9.970078947368422e-05,
      "loss": 0.5612,
      "step": 21147
    },
    {
      "epoch": 1.1842311569044686,
      "grad_norm": 1.4069215059280396,
      "learning_rate": 9.970052631578948e-05,
      "loss": 0.5357,
      "step": 21148
    },
    {
      "epoch": 1.1842871542165976,
      "grad_norm": 1.180469036102295,
      "learning_rate": 9.970026315789475e-05,
      "loss": 0.3794,
      "step": 21149
    },
    {
      "epoch": 1.1843431515287266,
      "grad_norm": 1.1504889726638794,
      "learning_rate": 9.970000000000001e-05,
      "loss": 0.3408,
      "step": 21150
    },
    {
      "epoch": 1.1843991488408556,
      "grad_norm": 1.4989124536514282,
      "learning_rate": 9.969973684210526e-05,
      "loss": 0.5002,
      "step": 21151
    },
    {
      "epoch": 1.1844551461529846,
      "grad_norm": 1.8440818786621094,
      "learning_rate": 9.969947368421053e-05,
      "loss": 0.6285,
      "step": 21152
    },
    {
      "epoch": 1.1845111434651137,
      "grad_norm": 7.958072662353516,
      "learning_rate": 9.969921052631579e-05,
      "loss": 0.3883,
      "step": 21153
    },
    {
      "epoch": 1.1845671407772427,
      "grad_norm": 1.494621992111206,
      "learning_rate": 9.969894736842106e-05,
      "loss": 0.5401,
      "step": 21154
    },
    {
      "epoch": 1.1846231380893717,
      "grad_norm": 1.2599050998687744,
      "learning_rate": 9.969868421052631e-05,
      "loss": 0.4195,
      "step": 21155
    },
    {
      "epoch": 1.1846791354015007,
      "grad_norm": 1.1688024997711182,
      "learning_rate": 9.969842105263158e-05,
      "loss": 0.3738,
      "step": 21156
    },
    {
      "epoch": 1.1847351327136297,
      "grad_norm": 1.5886930227279663,
      "learning_rate": 9.969815789473684e-05,
      "loss": 0.4661,
      "step": 21157
    },
    {
      "epoch": 1.1847911300257588,
      "grad_norm": 1.2828972339630127,
      "learning_rate": 9.969789473684212e-05,
      "loss": 0.3685,
      "step": 21158
    },
    {
      "epoch": 1.1848471273378878,
      "grad_norm": 1.2016386985778809,
      "learning_rate": 9.969763157894738e-05,
      "loss": 0.3928,
      "step": 21159
    },
    {
      "epoch": 1.1849031246500168,
      "grad_norm": 1.6821802854537964,
      "learning_rate": 9.969736842105264e-05,
      "loss": 0.4608,
      "step": 21160
    },
    {
      "epoch": 1.1849591219621458,
      "grad_norm": 1.7325512170791626,
      "learning_rate": 9.96971052631579e-05,
      "loss": 0.376,
      "step": 21161
    },
    {
      "epoch": 1.1850151192742748,
      "grad_norm": 1.6614699363708496,
      "learning_rate": 9.969684210526317e-05,
      "loss": 0.3121,
      "step": 21162
    },
    {
      "epoch": 1.1850711165864038,
      "grad_norm": 1.2217885255813599,
      "learning_rate": 9.969657894736843e-05,
      "loss": 0.3949,
      "step": 21163
    },
    {
      "epoch": 1.1851271138985329,
      "grad_norm": 1.9034477472305298,
      "learning_rate": 9.969631578947369e-05,
      "loss": 0.6209,
      "step": 21164
    },
    {
      "epoch": 1.1851831112106619,
      "grad_norm": 1.560486078262329,
      "learning_rate": 9.969605263157895e-05,
      "loss": 0.5011,
      "step": 21165
    },
    {
      "epoch": 1.185239108522791,
      "grad_norm": 1.3662017583847046,
      "learning_rate": 9.969578947368422e-05,
      "loss": 0.4632,
      "step": 21166
    },
    {
      "epoch": 1.18529510583492,
      "grad_norm": 1.2816357612609863,
      "learning_rate": 9.969552631578948e-05,
      "loss": 0.371,
      "step": 21167
    },
    {
      "epoch": 1.185351103147049,
      "grad_norm": 1.4631057977676392,
      "learning_rate": 9.969526315789474e-05,
      "loss": 0.491,
      "step": 21168
    },
    {
      "epoch": 1.185407100459178,
      "grad_norm": 1.1913096904754639,
      "learning_rate": 9.9695e-05,
      "loss": 0.4349,
      "step": 21169
    },
    {
      "epoch": 1.185463097771307,
      "grad_norm": 1.2965353727340698,
      "learning_rate": 9.969473684210526e-05,
      "loss": 0.4607,
      "step": 21170
    },
    {
      "epoch": 1.185519095083436,
      "grad_norm": 1.4136563539505005,
      "learning_rate": 9.969447368421053e-05,
      "loss": 0.4314,
      "step": 21171
    },
    {
      "epoch": 1.185575092395565,
      "grad_norm": 1.3343579769134521,
      "learning_rate": 9.969421052631579e-05,
      "loss": 0.4353,
      "step": 21172
    },
    {
      "epoch": 1.185631089707694,
      "grad_norm": 1.5607622861862183,
      "learning_rate": 9.969394736842105e-05,
      "loss": 0.4919,
      "step": 21173
    },
    {
      "epoch": 1.185687087019823,
      "grad_norm": 1.2649791240692139,
      "learning_rate": 9.969368421052631e-05,
      "loss": 0.5258,
      "step": 21174
    },
    {
      "epoch": 1.185743084331952,
      "grad_norm": 1.422687292098999,
      "learning_rate": 9.969342105263159e-05,
      "loss": 0.556,
      "step": 21175
    },
    {
      "epoch": 1.185799081644081,
      "grad_norm": 1.6056827306747437,
      "learning_rate": 9.969315789473685e-05,
      "loss": 0.439,
      "step": 21176
    },
    {
      "epoch": 1.1858550789562101,
      "grad_norm": 1.2976654767990112,
      "learning_rate": 9.969289473684212e-05,
      "loss": 0.4113,
      "step": 21177
    },
    {
      "epoch": 1.1859110762683391,
      "grad_norm": 1.3225038051605225,
      "learning_rate": 9.969263157894736e-05,
      "loss": 0.4313,
      "step": 21178
    },
    {
      "epoch": 1.1859670735804682,
      "grad_norm": 1.2887746095657349,
      "learning_rate": 9.969236842105264e-05,
      "loss": 0.4083,
      "step": 21179
    },
    {
      "epoch": 1.1860230708925972,
      "grad_norm": 1.6592974662780762,
      "learning_rate": 9.96921052631579e-05,
      "loss": 0.4067,
      "step": 21180
    },
    {
      "epoch": 1.1860790682047262,
      "grad_norm": 1.1698036193847656,
      "learning_rate": 9.969184210526317e-05,
      "loss": 0.4029,
      "step": 21181
    },
    {
      "epoch": 1.1861350655168552,
      "grad_norm": 1.4141706228256226,
      "learning_rate": 9.969157894736843e-05,
      "loss": 0.4824,
      "step": 21182
    },
    {
      "epoch": 1.1861910628289842,
      "grad_norm": 1.3743826150894165,
      "learning_rate": 9.969131578947369e-05,
      "loss": 0.4966,
      "step": 21183
    },
    {
      "epoch": 1.1862470601411133,
      "grad_norm": 1.381693720817566,
      "learning_rate": 9.969105263157895e-05,
      "loss": 0.5144,
      "step": 21184
    },
    {
      "epoch": 1.1863030574532423,
      "grad_norm": 1.380395770072937,
      "learning_rate": 9.969078947368422e-05,
      "loss": 0.3171,
      "step": 21185
    },
    {
      "epoch": 1.1863590547653713,
      "grad_norm": 1.3432729244232178,
      "learning_rate": 9.969052631578948e-05,
      "loss": 0.4076,
      "step": 21186
    },
    {
      "epoch": 1.1864150520775003,
      "grad_norm": 1.2328461408615112,
      "learning_rate": 9.969026315789474e-05,
      "loss": 0.3554,
      "step": 21187
    },
    {
      "epoch": 1.1864710493896293,
      "grad_norm": 1.296217918395996,
      "learning_rate": 9.969e-05,
      "loss": 0.5135,
      "step": 21188
    },
    {
      "epoch": 1.1865270467017583,
      "grad_norm": 35.66089630126953,
      "learning_rate": 9.968973684210526e-05,
      "loss": 0.6095,
      "step": 21189
    },
    {
      "epoch": 1.1865830440138874,
      "grad_norm": 1.3872572183609009,
      "learning_rate": 9.968947368421054e-05,
      "loss": 0.4508,
      "step": 21190
    },
    {
      "epoch": 1.1866390413260164,
      "grad_norm": 2.5166049003601074,
      "learning_rate": 9.96892105263158e-05,
      "loss": 0.6397,
      "step": 21191
    },
    {
      "epoch": 1.1866950386381454,
      "grad_norm": 1.4581425189971924,
      "learning_rate": 9.968894736842106e-05,
      "loss": 0.4259,
      "step": 21192
    },
    {
      "epoch": 1.1867510359502744,
      "grad_norm": 2.6536355018615723,
      "learning_rate": 9.968868421052631e-05,
      "loss": 0.5135,
      "step": 21193
    },
    {
      "epoch": 1.1868070332624034,
      "grad_norm": 1.2793033123016357,
      "learning_rate": 9.968842105263159e-05,
      "loss": 0.3964,
      "step": 21194
    },
    {
      "epoch": 1.1868630305745325,
      "grad_norm": 1.4136197566986084,
      "learning_rate": 9.968815789473685e-05,
      "loss": 0.5737,
      "step": 21195
    },
    {
      "epoch": 1.1869190278866615,
      "grad_norm": 1.2036255598068237,
      "learning_rate": 9.968789473684211e-05,
      "loss": 0.359,
      "step": 21196
    },
    {
      "epoch": 1.1869750251987905,
      "grad_norm": 1.5432047843933105,
      "learning_rate": 9.968763157894737e-05,
      "loss": 0.4279,
      "step": 21197
    },
    {
      "epoch": 1.1870310225109195,
      "grad_norm": 1.3293156623840332,
      "learning_rate": 9.968736842105264e-05,
      "loss": 0.3859,
      "step": 21198
    },
    {
      "epoch": 1.1870870198230485,
      "grad_norm": 1.7360200881958008,
      "learning_rate": 9.96871052631579e-05,
      "loss": 0.4776,
      "step": 21199
    },
    {
      "epoch": 1.1871430171351776,
      "grad_norm": 2.022270917892456,
      "learning_rate": 9.968684210526317e-05,
      "loss": 0.3391,
      "step": 21200
    },
    {
      "epoch": 1.1871990144473066,
      "grad_norm": 1.334336280822754,
      "learning_rate": 9.968657894736842e-05,
      "loss": 0.5049,
      "step": 21201
    },
    {
      "epoch": 1.1872550117594356,
      "grad_norm": 1.9071433544158936,
      "learning_rate": 9.968631578947369e-05,
      "loss": 0.3772,
      "step": 21202
    },
    {
      "epoch": 1.1873110090715646,
      "grad_norm": 1.2202825546264648,
      "learning_rate": 9.968605263157895e-05,
      "loss": 0.4446,
      "step": 21203
    },
    {
      "epoch": 1.1873670063836936,
      "grad_norm": 1.5728347301483154,
      "learning_rate": 9.968578947368421e-05,
      "loss": 0.5384,
      "step": 21204
    },
    {
      "epoch": 1.1874230036958227,
      "grad_norm": 1.3353360891342163,
      "learning_rate": 9.968552631578949e-05,
      "loss": 0.3864,
      "step": 21205
    },
    {
      "epoch": 1.1874790010079517,
      "grad_norm": 1.5640684366226196,
      "learning_rate": 9.968526315789473e-05,
      "loss": 0.439,
      "step": 21206
    },
    {
      "epoch": 1.1875349983200807,
      "grad_norm": 1.3092772960662842,
      "learning_rate": 9.9685e-05,
      "loss": 0.4114,
      "step": 21207
    },
    {
      "epoch": 1.1875909956322097,
      "grad_norm": 1.6550471782684326,
      "learning_rate": 9.968473684210526e-05,
      "loss": 0.4563,
      "step": 21208
    },
    {
      "epoch": 1.1876469929443387,
      "grad_norm": 1.2956535816192627,
      "learning_rate": 9.968447368421054e-05,
      "loss": 0.4703,
      "step": 21209
    },
    {
      "epoch": 1.1877029902564677,
      "grad_norm": 1.3555691242218018,
      "learning_rate": 9.968421052631578e-05,
      "loss": 0.5481,
      "step": 21210
    },
    {
      "epoch": 1.1877589875685968,
      "grad_norm": 1.2641892433166504,
      "learning_rate": 9.968394736842106e-05,
      "loss": 0.4387,
      "step": 21211
    },
    {
      "epoch": 1.1878149848807258,
      "grad_norm": 1.3119490146636963,
      "learning_rate": 9.968368421052632e-05,
      "loss": 0.3682,
      "step": 21212
    },
    {
      "epoch": 1.1878709821928548,
      "grad_norm": 1.429458737373352,
      "learning_rate": 9.968342105263159e-05,
      "loss": 0.4616,
      "step": 21213
    },
    {
      "epoch": 1.1879269795049838,
      "grad_norm": 1.39130437374115,
      "learning_rate": 9.968315789473685e-05,
      "loss": 0.4945,
      "step": 21214
    },
    {
      "epoch": 1.1879829768171128,
      "grad_norm": 1.6254644393920898,
      "learning_rate": 9.968289473684211e-05,
      "loss": 0.5361,
      "step": 21215
    },
    {
      "epoch": 1.1880389741292419,
      "grad_norm": 1.241886854171753,
      "learning_rate": 9.968263157894737e-05,
      "loss": 0.4756,
      "step": 21216
    },
    {
      "epoch": 1.1880949714413709,
      "grad_norm": 1.2532511949539185,
      "learning_rate": 9.968236842105264e-05,
      "loss": 0.4426,
      "step": 21217
    },
    {
      "epoch": 1.1881509687535,
      "grad_norm": 1.3908519744873047,
      "learning_rate": 9.96821052631579e-05,
      "loss": 0.5752,
      "step": 21218
    },
    {
      "epoch": 1.188206966065629,
      "grad_norm": 1.4657741785049438,
      "learning_rate": 9.968184210526316e-05,
      "loss": 0.4851,
      "step": 21219
    },
    {
      "epoch": 1.188262963377758,
      "grad_norm": 1.380887746810913,
      "learning_rate": 9.968157894736842e-05,
      "loss": 0.4347,
      "step": 21220
    },
    {
      "epoch": 1.188318960689887,
      "grad_norm": 1.3606361150741577,
      "learning_rate": 9.968131578947368e-05,
      "loss": 0.4537,
      "step": 21221
    },
    {
      "epoch": 1.188374958002016,
      "grad_norm": 1.447104573249817,
      "learning_rate": 9.968105263157896e-05,
      "loss": 0.4163,
      "step": 21222
    },
    {
      "epoch": 1.188430955314145,
      "grad_norm": 8.842304229736328,
      "learning_rate": 9.968078947368422e-05,
      "loss": 0.4515,
      "step": 21223
    },
    {
      "epoch": 1.188486952626274,
      "grad_norm": 1.6093858480453491,
      "learning_rate": 9.968052631578947e-05,
      "loss": 0.5563,
      "step": 21224
    },
    {
      "epoch": 1.188542949938403,
      "grad_norm": 1.502869963645935,
      "learning_rate": 9.968026315789473e-05,
      "loss": 0.4019,
      "step": 21225
    },
    {
      "epoch": 1.188598947250532,
      "grad_norm": 1.334153413772583,
      "learning_rate": 9.968000000000001e-05,
      "loss": 0.5447,
      "step": 21226
    },
    {
      "epoch": 1.188654944562661,
      "grad_norm": 1.394298791885376,
      "learning_rate": 9.967973684210527e-05,
      "loss": 0.4455,
      "step": 21227
    },
    {
      "epoch": 1.18871094187479,
      "grad_norm": 1.8458468914031982,
      "learning_rate": 9.967947368421053e-05,
      "loss": 0.4829,
      "step": 21228
    },
    {
      "epoch": 1.1887669391869191,
      "grad_norm": 1.3631985187530518,
      "learning_rate": 9.967921052631579e-05,
      "loss": 0.4603,
      "step": 21229
    },
    {
      "epoch": 1.1888229364990481,
      "grad_norm": 1.7145485877990723,
      "learning_rate": 9.967894736842106e-05,
      "loss": 0.4807,
      "step": 21230
    },
    {
      "epoch": 1.1888789338111772,
      "grad_norm": 1.5248175859451294,
      "learning_rate": 9.967868421052632e-05,
      "loss": 0.5424,
      "step": 21231
    },
    {
      "epoch": 1.1889349311233062,
      "grad_norm": 1.4054481983184814,
      "learning_rate": 9.96784210526316e-05,
      "loss": 0.3944,
      "step": 21232
    },
    {
      "epoch": 1.1889909284354352,
      "grad_norm": 1.3018110990524292,
      "learning_rate": 9.967815789473684e-05,
      "loss": 0.5388,
      "step": 21233
    },
    {
      "epoch": 1.1890469257475642,
      "grad_norm": 1.7539788484573364,
      "learning_rate": 9.967789473684211e-05,
      "loss": 0.5211,
      "step": 21234
    },
    {
      "epoch": 1.1891029230596932,
      "grad_norm": 1.2612608671188354,
      "learning_rate": 9.967763157894737e-05,
      "loss": 0.3915,
      "step": 21235
    },
    {
      "epoch": 1.1891589203718222,
      "grad_norm": 1.3768184185028076,
      "learning_rate": 9.967736842105265e-05,
      "loss": 0.4413,
      "step": 21236
    },
    {
      "epoch": 1.1892149176839513,
      "grad_norm": 1.557417392730713,
      "learning_rate": 9.96771052631579e-05,
      "loss": 0.5665,
      "step": 21237
    },
    {
      "epoch": 1.1892709149960803,
      "grad_norm": 1.3070082664489746,
      "learning_rate": 9.967684210526315e-05,
      "loss": 0.4983,
      "step": 21238
    },
    {
      "epoch": 1.1893269123082093,
      "grad_norm": 1.38764226436615,
      "learning_rate": 9.967657894736842e-05,
      "loss": 0.4311,
      "step": 21239
    },
    {
      "epoch": 1.1893829096203383,
      "grad_norm": 1.4594128131866455,
      "learning_rate": 9.967631578947368e-05,
      "loss": 0.5161,
      "step": 21240
    },
    {
      "epoch": 1.1894389069324673,
      "grad_norm": 1.1985621452331543,
      "learning_rate": 9.967605263157896e-05,
      "loss": 0.403,
      "step": 21241
    },
    {
      "epoch": 1.1894949042445964,
      "grad_norm": 1.1932530403137207,
      "learning_rate": 9.967578947368422e-05,
      "loss": 0.4045,
      "step": 21242
    },
    {
      "epoch": 1.1895509015567254,
      "grad_norm": 1.3379814624786377,
      "learning_rate": 9.967552631578948e-05,
      "loss": 0.42,
      "step": 21243
    },
    {
      "epoch": 1.1896068988688544,
      "grad_norm": 1.2524373531341553,
      "learning_rate": 9.967526315789474e-05,
      "loss": 0.3603,
      "step": 21244
    },
    {
      "epoch": 1.1896628961809834,
      "grad_norm": 1.932081699371338,
      "learning_rate": 9.967500000000001e-05,
      "loss": 0.5632,
      "step": 21245
    },
    {
      "epoch": 1.1897188934931124,
      "grad_norm": 1.4900124073028564,
      "learning_rate": 9.967473684210527e-05,
      "loss": 0.6733,
      "step": 21246
    },
    {
      "epoch": 1.1897748908052415,
      "grad_norm": 1.4772049188613892,
      "learning_rate": 9.967447368421053e-05,
      "loss": 0.5363,
      "step": 21247
    },
    {
      "epoch": 1.1898308881173705,
      "grad_norm": 1.3865453004837036,
      "learning_rate": 9.967421052631579e-05,
      "loss": 0.5008,
      "step": 21248
    },
    {
      "epoch": 1.1898868854294995,
      "grad_norm": 1.3593593835830688,
      "learning_rate": 9.967394736842106e-05,
      "loss": 0.439,
      "step": 21249
    },
    {
      "epoch": 1.1899428827416285,
      "grad_norm": 1.2697149515151978,
      "learning_rate": 9.967368421052632e-05,
      "loss": 0.391,
      "step": 21250
    },
    {
      "epoch": 1.1899988800537575,
      "grad_norm": 1.3371034860610962,
      "learning_rate": 9.967342105263158e-05,
      "loss": 0.447,
      "step": 21251
    },
    {
      "epoch": 1.1900548773658866,
      "grad_norm": 1.3540842533111572,
      "learning_rate": 9.967315789473684e-05,
      "loss": 0.466,
      "step": 21252
    },
    {
      "epoch": 1.1901108746780156,
      "grad_norm": 1.2161946296691895,
      "learning_rate": 9.967289473684212e-05,
      "loss": 0.4371,
      "step": 21253
    },
    {
      "epoch": 1.1901668719901446,
      "grad_norm": 1.0837620496749878,
      "learning_rate": 9.967263157894738e-05,
      "loss": 0.406,
      "step": 21254
    },
    {
      "epoch": 1.1902228693022736,
      "grad_norm": 1.256539225578308,
      "learning_rate": 9.967236842105263e-05,
      "loss": 0.4907,
      "step": 21255
    },
    {
      "epoch": 1.1902788666144026,
      "grad_norm": 1.5777192115783691,
      "learning_rate": 9.96721052631579e-05,
      "loss": 0.3872,
      "step": 21256
    },
    {
      "epoch": 1.1903348639265316,
      "grad_norm": 1.4282094240188599,
      "learning_rate": 9.967184210526315e-05,
      "loss": 0.5761,
      "step": 21257
    },
    {
      "epoch": 1.1903908612386607,
      "grad_norm": 1.4186253547668457,
      "learning_rate": 9.967157894736843e-05,
      "loss": 0.4186,
      "step": 21258
    },
    {
      "epoch": 1.1904468585507895,
      "grad_norm": 1.409191370010376,
      "learning_rate": 9.967131578947369e-05,
      "loss": 0.4498,
      "step": 21259
    },
    {
      "epoch": 1.1905028558629185,
      "grad_norm": 1.305984377861023,
      "learning_rate": 9.967105263157896e-05,
      "loss": 0.4489,
      "step": 21260
    },
    {
      "epoch": 1.1905588531750475,
      "grad_norm": 1.7411454916000366,
      "learning_rate": 9.96707894736842e-05,
      "loss": 0.4493,
      "step": 21261
    },
    {
      "epoch": 1.1906148504871765,
      "grad_norm": 1.277187466621399,
      "learning_rate": 9.967052631578948e-05,
      "loss": 0.4765,
      "step": 21262
    },
    {
      "epoch": 1.1906708477993055,
      "grad_norm": 1.2747749090194702,
      "learning_rate": 9.967026315789474e-05,
      "loss": 0.4391,
      "step": 21263
    },
    {
      "epoch": 1.1907268451114346,
      "grad_norm": 1.30585515499115,
      "learning_rate": 9.967000000000001e-05,
      "loss": 0.4503,
      "step": 21264
    },
    {
      "epoch": 1.1907828424235636,
      "grad_norm": 1.7725613117218018,
      "learning_rate": 9.966973684210526e-05,
      "loss": 0.6786,
      "step": 21265
    },
    {
      "epoch": 1.1908388397356926,
      "grad_norm": 1.4894185066223145,
      "learning_rate": 9.966947368421053e-05,
      "loss": 0.4221,
      "step": 21266
    },
    {
      "epoch": 1.1908948370478216,
      "grad_norm": 1.205344319343567,
      "learning_rate": 9.966921052631579e-05,
      "loss": 0.4327,
      "step": 21267
    },
    {
      "epoch": 1.1909508343599506,
      "grad_norm": 1.4172561168670654,
      "learning_rate": 9.966894736842107e-05,
      "loss": 0.4166,
      "step": 21268
    },
    {
      "epoch": 1.1910068316720797,
      "grad_norm": 1.1164103746414185,
      "learning_rate": 9.966868421052633e-05,
      "loss": 0.316,
      "step": 21269
    },
    {
      "epoch": 1.1910628289842087,
      "grad_norm": 1.3036106824874878,
      "learning_rate": 9.966842105263158e-05,
      "loss": 0.4415,
      "step": 21270
    },
    {
      "epoch": 1.1911188262963377,
      "grad_norm": 1.145293951034546,
      "learning_rate": 9.966815789473684e-05,
      "loss": 0.335,
      "step": 21271
    },
    {
      "epoch": 1.1911748236084667,
      "grad_norm": 1.4188669919967651,
      "learning_rate": 9.96678947368421e-05,
      "loss": 0.5391,
      "step": 21272
    },
    {
      "epoch": 1.1912308209205957,
      "grad_norm": 1.3428798913955688,
      "learning_rate": 9.966763157894738e-05,
      "loss": 0.4446,
      "step": 21273
    },
    {
      "epoch": 1.1912868182327248,
      "grad_norm": 1.8040522336959839,
      "learning_rate": 9.966736842105264e-05,
      "loss": 0.5103,
      "step": 21274
    },
    {
      "epoch": 1.1913428155448538,
      "grad_norm": 1.2655612230300903,
      "learning_rate": 9.96671052631579e-05,
      "loss": 0.4005,
      "step": 21275
    },
    {
      "epoch": 1.1913988128569828,
      "grad_norm": 1.1684056520462036,
      "learning_rate": 9.966684210526316e-05,
      "loss": 0.2977,
      "step": 21276
    },
    {
      "epoch": 1.1914548101691118,
      "grad_norm": 1.3208248615264893,
      "learning_rate": 9.966657894736843e-05,
      "loss": 0.5333,
      "step": 21277
    },
    {
      "epoch": 1.1915108074812408,
      "grad_norm": 1.7194039821624756,
      "learning_rate": 9.966631578947369e-05,
      "loss": 0.4584,
      "step": 21278
    },
    {
      "epoch": 1.1915668047933698,
      "grad_norm": 1.319769263267517,
      "learning_rate": 9.966605263157895e-05,
      "loss": 0.5436,
      "step": 21279
    },
    {
      "epoch": 1.1916228021054989,
      "grad_norm": 1.57979416847229,
      "learning_rate": 9.966578947368421e-05,
      "loss": 0.3921,
      "step": 21280
    },
    {
      "epoch": 1.1916787994176279,
      "grad_norm": 1.232108235359192,
      "learning_rate": 9.966552631578948e-05,
      "loss": 0.3542,
      "step": 21281
    },
    {
      "epoch": 1.191734796729757,
      "grad_norm": 1.295562505722046,
      "learning_rate": 9.966526315789474e-05,
      "loss": 0.4108,
      "step": 21282
    },
    {
      "epoch": 1.191790794041886,
      "grad_norm": 1.457898736000061,
      "learning_rate": 9.9665e-05,
      "loss": 0.4532,
      "step": 21283
    },
    {
      "epoch": 1.191846791354015,
      "grad_norm": 1.6455161571502686,
      "learning_rate": 9.966473684210526e-05,
      "loss": 0.5247,
      "step": 21284
    },
    {
      "epoch": 1.191902788666144,
      "grad_norm": 1.4352648258209229,
      "learning_rate": 9.966447368421054e-05,
      "loss": 0.4688,
      "step": 21285
    },
    {
      "epoch": 1.191958785978273,
      "grad_norm": 42.5954475402832,
      "learning_rate": 9.96642105263158e-05,
      "loss": 0.5555,
      "step": 21286
    },
    {
      "epoch": 1.192014783290402,
      "grad_norm": 1.550161361694336,
      "learning_rate": 9.966394736842107e-05,
      "loss": 0.493,
      "step": 21287
    },
    {
      "epoch": 1.192070780602531,
      "grad_norm": 1.437830924987793,
      "learning_rate": 9.966368421052631e-05,
      "loss": 0.4395,
      "step": 21288
    },
    {
      "epoch": 1.19212677791466,
      "grad_norm": 1.3734545707702637,
      "learning_rate": 9.966342105263159e-05,
      "loss": 0.432,
      "step": 21289
    },
    {
      "epoch": 1.192182775226789,
      "grad_norm": 1.2367222309112549,
      "learning_rate": 9.966315789473685e-05,
      "loss": 0.3989,
      "step": 21290
    },
    {
      "epoch": 1.192238772538918,
      "grad_norm": 1.257411003112793,
      "learning_rate": 9.966289473684211e-05,
      "loss": 0.4927,
      "step": 21291
    },
    {
      "epoch": 1.192294769851047,
      "grad_norm": 1.535960078239441,
      "learning_rate": 9.966263157894738e-05,
      "loss": 0.3534,
      "step": 21292
    },
    {
      "epoch": 1.1923507671631761,
      "grad_norm": 1.664744257926941,
      "learning_rate": 9.966236842105263e-05,
      "loss": 0.4783,
      "step": 21293
    },
    {
      "epoch": 1.1924067644753051,
      "grad_norm": 1.3406223058700562,
      "learning_rate": 9.96621052631579e-05,
      "loss": 0.5224,
      "step": 21294
    },
    {
      "epoch": 1.1924627617874342,
      "grad_norm": 1.4745821952819824,
      "learning_rate": 9.966184210526316e-05,
      "loss": 0.3627,
      "step": 21295
    },
    {
      "epoch": 1.1925187590995632,
      "grad_norm": 1.4468576908111572,
      "learning_rate": 9.966157894736843e-05,
      "loss": 0.4381,
      "step": 21296
    },
    {
      "epoch": 1.1925747564116922,
      "grad_norm": 1.48208749294281,
      "learning_rate": 9.966131578947369e-05,
      "loss": 0.4918,
      "step": 21297
    },
    {
      "epoch": 1.1926307537238212,
      "grad_norm": 1.2047662734985352,
      "learning_rate": 9.966105263157895e-05,
      "loss": 0.5098,
      "step": 21298
    },
    {
      "epoch": 1.1926867510359502,
      "grad_norm": 1.2078361511230469,
      "learning_rate": 9.966078947368421e-05,
      "loss": 0.4489,
      "step": 21299
    },
    {
      "epoch": 1.1927427483480793,
      "grad_norm": 1.2705317735671997,
      "learning_rate": 9.966052631578949e-05,
      "loss": 0.3713,
      "step": 21300
    },
    {
      "epoch": 1.1927987456602083,
      "grad_norm": 1.3581575155258179,
      "learning_rate": 9.966026315789474e-05,
      "loss": 0.376,
      "step": 21301
    },
    {
      "epoch": 1.1928547429723373,
      "grad_norm": 1.405099630355835,
      "learning_rate": 9.966e-05,
      "loss": 0.4952,
      "step": 21302
    },
    {
      "epoch": 1.1929107402844663,
      "grad_norm": 1.1661505699157715,
      "learning_rate": 9.965973684210526e-05,
      "loss": 0.3296,
      "step": 21303
    },
    {
      "epoch": 1.1929667375965953,
      "grad_norm": 1.4038792848587036,
      "learning_rate": 9.965947368421054e-05,
      "loss": 0.3139,
      "step": 21304
    },
    {
      "epoch": 1.1930227349087243,
      "grad_norm": 1.3428869247436523,
      "learning_rate": 9.96592105263158e-05,
      "loss": 0.4924,
      "step": 21305
    },
    {
      "epoch": 1.1930787322208534,
      "grad_norm": 1.3574388027191162,
      "learning_rate": 9.965894736842106e-05,
      "loss": 0.5391,
      "step": 21306
    },
    {
      "epoch": 1.1931347295329824,
      "grad_norm": 1.272050142288208,
      "learning_rate": 9.965868421052632e-05,
      "loss": 0.401,
      "step": 21307
    },
    {
      "epoch": 1.1931907268451114,
      "grad_norm": 1.4629734754562378,
      "learning_rate": 9.965842105263158e-05,
      "loss": 0.4211,
      "step": 21308
    },
    {
      "epoch": 1.1932467241572404,
      "grad_norm": 1.3369213342666626,
      "learning_rate": 9.965815789473685e-05,
      "loss": 0.3077,
      "step": 21309
    },
    {
      "epoch": 1.1933027214693694,
      "grad_norm": 1.5254815816879272,
      "learning_rate": 9.965789473684211e-05,
      "loss": 0.6437,
      "step": 21310
    },
    {
      "epoch": 1.1933587187814985,
      "grad_norm": 1.391893744468689,
      "learning_rate": 9.965763157894737e-05,
      "loss": 0.4797,
      "step": 21311
    },
    {
      "epoch": 1.1934147160936275,
      "grad_norm": 1.3041632175445557,
      "learning_rate": 9.965736842105263e-05,
      "loss": 0.4406,
      "step": 21312
    },
    {
      "epoch": 1.1934707134057565,
      "grad_norm": 1.2722002267837524,
      "learning_rate": 9.96571052631579e-05,
      "loss": 0.3874,
      "step": 21313
    },
    {
      "epoch": 1.1935267107178855,
      "grad_norm": 1.2253620624542236,
      "learning_rate": 9.965684210526316e-05,
      "loss": 0.4311,
      "step": 21314
    },
    {
      "epoch": 1.1935827080300145,
      "grad_norm": 1.3320913314819336,
      "learning_rate": 9.965657894736844e-05,
      "loss": 0.3835,
      "step": 21315
    },
    {
      "epoch": 1.1936387053421436,
      "grad_norm": 1.3502252101898193,
      "learning_rate": 9.965631578947368e-05,
      "loss": 0.4198,
      "step": 21316
    },
    {
      "epoch": 1.1936947026542726,
      "grad_norm": 1.4054888486862183,
      "learning_rate": 9.965605263157895e-05,
      "loss": 0.4118,
      "step": 21317
    },
    {
      "epoch": 1.1937506999664016,
      "grad_norm": 1.4755635261535645,
      "learning_rate": 9.965578947368421e-05,
      "loss": 0.3879,
      "step": 21318
    },
    {
      "epoch": 1.1938066972785306,
      "grad_norm": 1.1939069032669067,
      "learning_rate": 9.965552631578949e-05,
      "loss": 0.4434,
      "step": 21319
    },
    {
      "epoch": 1.1938626945906596,
      "grad_norm": 1.3770779371261597,
      "learning_rate": 9.965526315789473e-05,
      "loss": 0.4295,
      "step": 21320
    },
    {
      "epoch": 1.1939186919027887,
      "grad_norm": 1.2577886581420898,
      "learning_rate": 9.965500000000001e-05,
      "loss": 0.3482,
      "step": 21321
    },
    {
      "epoch": 1.1939746892149177,
      "grad_norm": 1.3705942630767822,
      "learning_rate": 9.965473684210527e-05,
      "loss": 0.4651,
      "step": 21322
    },
    {
      "epoch": 1.1940306865270467,
      "grad_norm": 1.2908939123153687,
      "learning_rate": 9.965447368421054e-05,
      "loss": 0.3609,
      "step": 21323
    },
    {
      "epoch": 1.1940866838391757,
      "grad_norm": 1.2212295532226562,
      "learning_rate": 9.96542105263158e-05,
      "loss": 0.4819,
      "step": 21324
    },
    {
      "epoch": 1.1941426811513047,
      "grad_norm": 2.922239303588867,
      "learning_rate": 9.965394736842105e-05,
      "loss": 0.3876,
      "step": 21325
    },
    {
      "epoch": 1.1941986784634337,
      "grad_norm": 1.2339801788330078,
      "learning_rate": 9.965368421052632e-05,
      "loss": 0.3714,
      "step": 21326
    },
    {
      "epoch": 1.1942546757755628,
      "grad_norm": 1.2695056200027466,
      "learning_rate": 9.965342105263158e-05,
      "loss": 0.4169,
      "step": 21327
    },
    {
      "epoch": 1.1943106730876918,
      "grad_norm": 1.566595196723938,
      "learning_rate": 9.965315789473685e-05,
      "loss": 0.4286,
      "step": 21328
    },
    {
      "epoch": 1.1943666703998208,
      "grad_norm": 1.6339783668518066,
      "learning_rate": 9.965289473684211e-05,
      "loss": 0.5141,
      "step": 21329
    },
    {
      "epoch": 1.1944226677119498,
      "grad_norm": 1.6444056034088135,
      "learning_rate": 9.965263157894737e-05,
      "loss": 0.4442,
      "step": 21330
    },
    {
      "epoch": 1.1944786650240788,
      "grad_norm": 9.891744613647461,
      "learning_rate": 9.965236842105263e-05,
      "loss": 0.6255,
      "step": 21331
    },
    {
      "epoch": 1.1945346623362079,
      "grad_norm": 1.4509602785110474,
      "learning_rate": 9.96521052631579e-05,
      "loss": 0.5928,
      "step": 21332
    },
    {
      "epoch": 1.1945906596483369,
      "grad_norm": 1.338374376296997,
      "learning_rate": 9.965184210526316e-05,
      "loss": 0.4418,
      "step": 21333
    },
    {
      "epoch": 1.194646656960466,
      "grad_norm": 2.459693670272827,
      "learning_rate": 9.965157894736842e-05,
      "loss": 0.4032,
      "step": 21334
    },
    {
      "epoch": 1.194702654272595,
      "grad_norm": 1.3252931833267212,
      "learning_rate": 9.965131578947368e-05,
      "loss": 0.3934,
      "step": 21335
    },
    {
      "epoch": 1.194758651584724,
      "grad_norm": 1.5309665203094482,
      "learning_rate": 9.965105263157896e-05,
      "loss": 0.5197,
      "step": 21336
    },
    {
      "epoch": 1.194814648896853,
      "grad_norm": 1.3036658763885498,
      "learning_rate": 9.965078947368422e-05,
      "loss": 0.4037,
      "step": 21337
    },
    {
      "epoch": 1.194870646208982,
      "grad_norm": 1.296344518661499,
      "learning_rate": 9.965052631578948e-05,
      "loss": 0.5048,
      "step": 21338
    },
    {
      "epoch": 1.194926643521111,
      "grad_norm": 1.0616509914398193,
      "learning_rate": 9.965026315789474e-05,
      "loss": 0.339,
      "step": 21339
    },
    {
      "epoch": 1.19498264083324,
      "grad_norm": 1.7376258373260498,
      "learning_rate": 9.965000000000001e-05,
      "loss": 0.5438,
      "step": 21340
    },
    {
      "epoch": 1.195038638145369,
      "grad_norm": 1.2771480083465576,
      "learning_rate": 9.964973684210527e-05,
      "loss": 0.4788,
      "step": 21341
    },
    {
      "epoch": 1.195094635457498,
      "grad_norm": 1.6966956853866577,
      "learning_rate": 9.964947368421053e-05,
      "loss": 0.477,
      "step": 21342
    },
    {
      "epoch": 1.195150632769627,
      "grad_norm": 1.2666939496994019,
      "learning_rate": 9.964921052631579e-05,
      "loss": 0.6467,
      "step": 21343
    },
    {
      "epoch": 1.195206630081756,
      "grad_norm": 1.2245177030563354,
      "learning_rate": 9.964894736842105e-05,
      "loss": 0.4658,
      "step": 21344
    },
    {
      "epoch": 1.1952626273938851,
      "grad_norm": 1.3075501918792725,
      "learning_rate": 9.964868421052632e-05,
      "loss": 0.4255,
      "step": 21345
    },
    {
      "epoch": 1.1953186247060141,
      "grad_norm": 1.3158923387527466,
      "learning_rate": 9.964842105263158e-05,
      "loss": 0.4028,
      "step": 21346
    },
    {
      "epoch": 1.1953746220181432,
      "grad_norm": 1.310836911201477,
      "learning_rate": 9.964815789473686e-05,
      "loss": 0.4359,
      "step": 21347
    },
    {
      "epoch": 1.1954306193302722,
      "grad_norm": 1.4936611652374268,
      "learning_rate": 9.96478947368421e-05,
      "loss": 0.5623,
      "step": 21348
    },
    {
      "epoch": 1.1954866166424012,
      "grad_norm": 1.1484125852584839,
      "learning_rate": 9.964763157894737e-05,
      "loss": 0.3725,
      "step": 21349
    },
    {
      "epoch": 1.1955426139545302,
      "grad_norm": 1.3295791149139404,
      "learning_rate": 9.964736842105263e-05,
      "loss": 0.5219,
      "step": 21350
    },
    {
      "epoch": 1.1955986112666592,
      "grad_norm": 1.4058802127838135,
      "learning_rate": 9.964710526315791e-05,
      "loss": 0.4106,
      "step": 21351
    },
    {
      "epoch": 1.1956546085787882,
      "grad_norm": 1.1733869314193726,
      "learning_rate": 9.964684210526317e-05,
      "loss": 0.3727,
      "step": 21352
    },
    {
      "epoch": 1.1957106058909173,
      "grad_norm": 1.3899941444396973,
      "learning_rate": 9.964657894736843e-05,
      "loss": 0.4666,
      "step": 21353
    },
    {
      "epoch": 1.1957666032030463,
      "grad_norm": 1.5182164907455444,
      "learning_rate": 9.964631578947369e-05,
      "loss": 0.6035,
      "step": 21354
    },
    {
      "epoch": 1.1958226005151753,
      "grad_norm": 1.3368122577667236,
      "learning_rate": 9.964605263157896e-05,
      "loss": 0.4835,
      "step": 21355
    },
    {
      "epoch": 1.1958785978273043,
      "grad_norm": 1.2818577289581299,
      "learning_rate": 9.964578947368422e-05,
      "loss": 0.5112,
      "step": 21356
    },
    {
      "epoch": 1.1959345951394333,
      "grad_norm": 1.1480575799942017,
      "learning_rate": 9.964552631578948e-05,
      "loss": 0.4568,
      "step": 21357
    },
    {
      "epoch": 1.1959905924515624,
      "grad_norm": 1.4644219875335693,
      "learning_rate": 9.964526315789474e-05,
      "loss": 0.4418,
      "step": 21358
    },
    {
      "epoch": 1.1960465897636914,
      "grad_norm": 1.3618824481964111,
      "learning_rate": 9.9645e-05,
      "loss": 0.4163,
      "step": 21359
    },
    {
      "epoch": 1.1961025870758204,
      "grad_norm": 1.3760063648223877,
      "learning_rate": 9.964473684210527e-05,
      "loss": 0.4738,
      "step": 21360
    },
    {
      "epoch": 1.1961585843879494,
      "grad_norm": 1.3598028421401978,
      "learning_rate": 9.964447368421053e-05,
      "loss": 0.3665,
      "step": 21361
    },
    {
      "epoch": 1.1962145817000784,
      "grad_norm": 1.2918466329574585,
      "learning_rate": 9.964421052631579e-05,
      "loss": 0.4522,
      "step": 21362
    },
    {
      "epoch": 1.1962705790122075,
      "grad_norm": 1.4884247779846191,
      "learning_rate": 9.964394736842105e-05,
      "loss": 0.4283,
      "step": 21363
    },
    {
      "epoch": 1.1963265763243365,
      "grad_norm": 1.1714822053909302,
      "learning_rate": 9.964368421052632e-05,
      "loss": 0.4868,
      "step": 21364
    },
    {
      "epoch": 1.1963825736364655,
      "grad_norm": 1.141059160232544,
      "learning_rate": 9.964342105263158e-05,
      "loss": 0.3575,
      "step": 21365
    },
    {
      "epoch": 1.1964385709485945,
      "grad_norm": 1.2665141820907593,
      "learning_rate": 9.964315789473684e-05,
      "loss": 0.4476,
      "step": 21366
    },
    {
      "epoch": 1.1964945682607235,
      "grad_norm": 1.3462717533111572,
      "learning_rate": 9.96428947368421e-05,
      "loss": 0.3294,
      "step": 21367
    },
    {
      "epoch": 1.1965505655728526,
      "grad_norm": 1.5221977233886719,
      "learning_rate": 9.964263157894738e-05,
      "loss": 0.4968,
      "step": 21368
    },
    {
      "epoch": 1.1966065628849816,
      "grad_norm": 2.3991973400115967,
      "learning_rate": 9.964236842105264e-05,
      "loss": 0.555,
      "step": 21369
    },
    {
      "epoch": 1.1966625601971106,
      "grad_norm": 1.454770565032959,
      "learning_rate": 9.96421052631579e-05,
      "loss": 0.4786,
      "step": 21370
    },
    {
      "epoch": 1.1967185575092396,
      "grad_norm": 1.3468774557113647,
      "learning_rate": 9.964184210526316e-05,
      "loss": 0.4394,
      "step": 21371
    },
    {
      "epoch": 1.1967745548213686,
      "grad_norm": 1.3988748788833618,
      "learning_rate": 9.964157894736843e-05,
      "loss": 0.4111,
      "step": 21372
    },
    {
      "epoch": 1.1968305521334976,
      "grad_norm": 1.39087975025177,
      "learning_rate": 9.964131578947369e-05,
      "loss": 0.4374,
      "step": 21373
    },
    {
      "epoch": 1.1968865494456267,
      "grad_norm": 1.1370114088058472,
      "learning_rate": 9.964105263157896e-05,
      "loss": 0.4175,
      "step": 21374
    },
    {
      "epoch": 1.1969425467577557,
      "grad_norm": 1.2697925567626953,
      "learning_rate": 9.964078947368421e-05,
      "loss": 0.3974,
      "step": 21375
    },
    {
      "epoch": 1.1969985440698847,
      "grad_norm": 1.4969922304153442,
      "learning_rate": 9.964052631578947e-05,
      "loss": 0.6954,
      "step": 21376
    },
    {
      "epoch": 1.1970545413820137,
      "grad_norm": 1.474172592163086,
      "learning_rate": 9.964026315789474e-05,
      "loss": 0.4591,
      "step": 21377
    },
    {
      "epoch": 1.1971105386941427,
      "grad_norm": 1.2626137733459473,
      "learning_rate": 9.964e-05,
      "loss": 0.4082,
      "step": 21378
    },
    {
      "epoch": 1.1971665360062718,
      "grad_norm": 1.0900942087173462,
      "learning_rate": 9.963973684210527e-05,
      "loss": 0.3125,
      "step": 21379
    },
    {
      "epoch": 1.1972225333184008,
      "grad_norm": 1.2060264348983765,
      "learning_rate": 9.963947368421052e-05,
      "loss": 0.561,
      "step": 21380
    },
    {
      "epoch": 1.1972785306305298,
      "grad_norm": 1.5008060932159424,
      "learning_rate": 9.96392105263158e-05,
      "loss": 0.4794,
      "step": 21381
    },
    {
      "epoch": 1.1973345279426588,
      "grad_norm": 1.6069817543029785,
      "learning_rate": 9.963894736842105e-05,
      "loss": 0.7367,
      "step": 21382
    },
    {
      "epoch": 1.1973905252547878,
      "grad_norm": 1.6059280633926392,
      "learning_rate": 9.963868421052633e-05,
      "loss": 0.4607,
      "step": 21383
    },
    {
      "epoch": 1.1974465225669169,
      "grad_norm": 1.3171371221542358,
      "learning_rate": 9.963842105263159e-05,
      "loss": 0.4136,
      "step": 21384
    },
    {
      "epoch": 1.1975025198790459,
      "grad_norm": 1.3216816186904907,
      "learning_rate": 9.963815789473685e-05,
      "loss": 0.3876,
      "step": 21385
    },
    {
      "epoch": 1.197558517191175,
      "grad_norm": 1.4280669689178467,
      "learning_rate": 9.96378947368421e-05,
      "loss": 0.3501,
      "step": 21386
    },
    {
      "epoch": 1.197614514503304,
      "grad_norm": 1.2969985008239746,
      "learning_rate": 9.963763157894738e-05,
      "loss": 0.4646,
      "step": 21387
    },
    {
      "epoch": 1.197670511815433,
      "grad_norm": 1.5294933319091797,
      "learning_rate": 9.963736842105264e-05,
      "loss": 0.4396,
      "step": 21388
    },
    {
      "epoch": 1.197726509127562,
      "grad_norm": 1.4797515869140625,
      "learning_rate": 9.96371052631579e-05,
      "loss": 0.5789,
      "step": 21389
    },
    {
      "epoch": 1.197782506439691,
      "grad_norm": 1.872753381729126,
      "learning_rate": 9.963684210526316e-05,
      "loss": 0.4848,
      "step": 21390
    },
    {
      "epoch": 1.19783850375182,
      "grad_norm": 1.5574126243591309,
      "learning_rate": 9.963657894736843e-05,
      "loss": 0.4941,
      "step": 21391
    },
    {
      "epoch": 1.197894501063949,
      "grad_norm": 12.988001823425293,
      "learning_rate": 9.963631578947369e-05,
      "loss": 0.5403,
      "step": 21392
    },
    {
      "epoch": 1.197950498376078,
      "grad_norm": 1.631577730178833,
      "learning_rate": 9.963605263157895e-05,
      "loss": 0.4207,
      "step": 21393
    },
    {
      "epoch": 1.198006495688207,
      "grad_norm": 1.3914836645126343,
      "learning_rate": 9.963578947368421e-05,
      "loss": 0.4603,
      "step": 21394
    },
    {
      "epoch": 1.198062493000336,
      "grad_norm": 1.272395133972168,
      "learning_rate": 9.963552631578947e-05,
      "loss": 0.3924,
      "step": 21395
    },
    {
      "epoch": 1.198118490312465,
      "grad_norm": 1.0716838836669922,
      "learning_rate": 9.963526315789474e-05,
      "loss": 0.3835,
      "step": 21396
    },
    {
      "epoch": 1.198174487624594,
      "grad_norm": 1.6219170093536377,
      "learning_rate": 9.9635e-05,
      "loss": 0.6136,
      "step": 21397
    },
    {
      "epoch": 1.1982304849367231,
      "grad_norm": 1.3890399932861328,
      "learning_rate": 9.963473684210526e-05,
      "loss": 0.4818,
      "step": 21398
    },
    {
      "epoch": 1.1982864822488521,
      "grad_norm": 1.3197038173675537,
      "learning_rate": 9.963447368421052e-05,
      "loss": 0.3899,
      "step": 21399
    },
    {
      "epoch": 1.1983424795609812,
      "grad_norm": 1.3574610948562622,
      "learning_rate": 9.96342105263158e-05,
      "loss": 0.4244,
      "step": 21400
    },
    {
      "epoch": 1.1983984768731102,
      "grad_norm": 1.9468427896499634,
      "learning_rate": 9.963394736842106e-05,
      "loss": 0.4961,
      "step": 21401
    },
    {
      "epoch": 1.1984544741852392,
      "grad_norm": 1.6920266151428223,
      "learning_rate": 9.963368421052633e-05,
      "loss": 0.446,
      "step": 21402
    },
    {
      "epoch": 1.198510471497368,
      "grad_norm": 1.560967206954956,
      "learning_rate": 9.963342105263158e-05,
      "loss": 0.4072,
      "step": 21403
    },
    {
      "epoch": 1.198566468809497,
      "grad_norm": 1.432548999786377,
      "learning_rate": 9.963315789473685e-05,
      "loss": 0.4393,
      "step": 21404
    },
    {
      "epoch": 1.198622466121626,
      "grad_norm": 1.4270511865615845,
      "learning_rate": 9.963289473684211e-05,
      "loss": 0.3981,
      "step": 21405
    },
    {
      "epoch": 1.198678463433755,
      "grad_norm": 1.488874912261963,
      "learning_rate": 9.963263157894738e-05,
      "loss": 0.4374,
      "step": 21406
    },
    {
      "epoch": 1.198734460745884,
      "grad_norm": 1.3100860118865967,
      "learning_rate": 9.963236842105264e-05,
      "loss": 0.5362,
      "step": 21407
    },
    {
      "epoch": 1.198790458058013,
      "grad_norm": 1.2270326614379883,
      "learning_rate": 9.96321052631579e-05,
      "loss": 0.3289,
      "step": 21408
    },
    {
      "epoch": 1.1988464553701421,
      "grad_norm": 1.1580663919448853,
      "learning_rate": 9.963184210526316e-05,
      "loss": 0.4817,
      "step": 21409
    },
    {
      "epoch": 1.1989024526822711,
      "grad_norm": 1.5381953716278076,
      "learning_rate": 9.963157894736843e-05,
      "loss": 0.4355,
      "step": 21410
    },
    {
      "epoch": 1.1989584499944002,
      "grad_norm": 1.052895188331604,
      "learning_rate": 9.96313157894737e-05,
      "loss": 0.3743,
      "step": 21411
    },
    {
      "epoch": 1.1990144473065292,
      "grad_norm": 1.3846203088760376,
      "learning_rate": 9.963105263157894e-05,
      "loss": 0.4153,
      "step": 21412
    },
    {
      "epoch": 1.1990704446186582,
      "grad_norm": 1.3448079824447632,
      "learning_rate": 9.963078947368421e-05,
      "loss": 0.3799,
      "step": 21413
    },
    {
      "epoch": 1.1991264419307872,
      "grad_norm": 1.3910526037216187,
      "learning_rate": 9.963052631578947e-05,
      "loss": 0.3322,
      "step": 21414
    },
    {
      "epoch": 1.1991824392429162,
      "grad_norm": 1.280060052871704,
      "learning_rate": 9.963026315789475e-05,
      "loss": 0.4396,
      "step": 21415
    },
    {
      "epoch": 1.1992384365550453,
      "grad_norm": 1.3302288055419922,
      "learning_rate": 9.963e-05,
      "loss": 0.3988,
      "step": 21416
    },
    {
      "epoch": 1.1992944338671743,
      "grad_norm": 1.322451114654541,
      "learning_rate": 9.962973684210527e-05,
      "loss": 0.3296,
      "step": 21417
    },
    {
      "epoch": 1.1993504311793033,
      "grad_norm": 1.6919171810150146,
      "learning_rate": 9.962947368421053e-05,
      "loss": 0.4415,
      "step": 21418
    },
    {
      "epoch": 1.1994064284914323,
      "grad_norm": 1.2231345176696777,
      "learning_rate": 9.96292105263158e-05,
      "loss": 0.3932,
      "step": 21419
    },
    {
      "epoch": 1.1994624258035613,
      "grad_norm": 1.4488109350204468,
      "learning_rate": 9.962894736842106e-05,
      "loss": 0.4924,
      "step": 21420
    },
    {
      "epoch": 1.1995184231156903,
      "grad_norm": 1.2579987049102783,
      "learning_rate": 9.962868421052632e-05,
      "loss": 0.4286,
      "step": 21421
    },
    {
      "epoch": 1.1995744204278194,
      "grad_norm": 1.4669923782348633,
      "learning_rate": 9.962842105263158e-05,
      "loss": 0.4557,
      "step": 21422
    },
    {
      "epoch": 1.1996304177399484,
      "grad_norm": 1.3621535301208496,
      "learning_rate": 9.962815789473685e-05,
      "loss": 0.454,
      "step": 21423
    },
    {
      "epoch": 1.1996864150520774,
      "grad_norm": 1.6935299634933472,
      "learning_rate": 9.962789473684211e-05,
      "loss": 0.4028,
      "step": 21424
    },
    {
      "epoch": 1.1997424123642064,
      "grad_norm": 1.788367509841919,
      "learning_rate": 9.962763157894737e-05,
      "loss": 0.4214,
      "step": 21425
    },
    {
      "epoch": 1.1997984096763354,
      "grad_norm": 1.615911841392517,
      "learning_rate": 9.962736842105263e-05,
      "loss": 0.4857,
      "step": 21426
    },
    {
      "epoch": 1.1998544069884645,
      "grad_norm": 1.5796549320220947,
      "learning_rate": 9.96271052631579e-05,
      "loss": 0.4454,
      "step": 21427
    },
    {
      "epoch": 1.1999104043005935,
      "grad_norm": 1.3333300352096558,
      "learning_rate": 9.962684210526316e-05,
      "loss": 0.4156,
      "step": 21428
    },
    {
      "epoch": 1.1999664016127225,
      "grad_norm": 1.4734907150268555,
      "learning_rate": 9.962657894736842e-05,
      "loss": 0.5741,
      "step": 21429
    },
    {
      "epoch": 1.2000223989248515,
      "grad_norm": 1.3020298480987549,
      "learning_rate": 9.962631578947368e-05,
      "loss": 0.4521,
      "step": 21430
    },
    {
      "epoch": 1.2000783962369805,
      "grad_norm": 1.2103228569030762,
      "learning_rate": 9.962605263157894e-05,
      "loss": 0.3992,
      "step": 21431
    },
    {
      "epoch": 1.2001343935491096,
      "grad_norm": 1.1992607116699219,
      "learning_rate": 9.962578947368422e-05,
      "loss": 0.3216,
      "step": 21432
    },
    {
      "epoch": 1.2001903908612386,
      "grad_norm": 1.7995073795318604,
      "learning_rate": 9.962552631578948e-05,
      "loss": 0.5035,
      "step": 21433
    },
    {
      "epoch": 1.2002463881733676,
      "grad_norm": 1.3206636905670166,
      "learning_rate": 9.962526315789475e-05,
      "loss": 0.3423,
      "step": 21434
    },
    {
      "epoch": 1.2003023854854966,
      "grad_norm": 1.1745672225952148,
      "learning_rate": 9.9625e-05,
      "loss": 0.4342,
      "step": 21435
    },
    {
      "epoch": 1.2003583827976256,
      "grad_norm": 1.5004053115844727,
      "learning_rate": 9.962473684210527e-05,
      "loss": 0.5099,
      "step": 21436
    },
    {
      "epoch": 1.2004143801097547,
      "grad_norm": 1.3555021286010742,
      "learning_rate": 9.962447368421053e-05,
      "loss": 0.4008,
      "step": 21437
    },
    {
      "epoch": 1.2004703774218837,
      "grad_norm": 1.3807787895202637,
      "learning_rate": 9.96242105263158e-05,
      "loss": 0.4379,
      "step": 21438
    },
    {
      "epoch": 1.2005263747340127,
      "grad_norm": 1.484658122062683,
      "learning_rate": 9.962394736842106e-05,
      "loss": 0.5916,
      "step": 21439
    },
    {
      "epoch": 1.2005823720461417,
      "grad_norm": 1.4990845918655396,
      "learning_rate": 9.962368421052632e-05,
      "loss": 0.3455,
      "step": 21440
    },
    {
      "epoch": 1.2006383693582707,
      "grad_norm": 1.7396429777145386,
      "learning_rate": 9.962342105263158e-05,
      "loss": 0.4284,
      "step": 21441
    },
    {
      "epoch": 1.2006943666703997,
      "grad_norm": 1.1902375221252441,
      "learning_rate": 9.962315789473685e-05,
      "loss": 0.3986,
      "step": 21442
    },
    {
      "epoch": 1.2007503639825288,
      "grad_norm": 1.521207571029663,
      "learning_rate": 9.962289473684211e-05,
      "loss": 0.4777,
      "step": 21443
    },
    {
      "epoch": 1.2008063612946578,
      "grad_norm": 1.5875146389007568,
      "learning_rate": 9.962263157894737e-05,
      "loss": 0.6852,
      "step": 21444
    },
    {
      "epoch": 1.2008623586067868,
      "grad_norm": 1.0622169971466064,
      "learning_rate": 9.962236842105263e-05,
      "loss": 0.3953,
      "step": 21445
    },
    {
      "epoch": 1.2009183559189158,
      "grad_norm": 1.5218700170516968,
      "learning_rate": 9.96221052631579e-05,
      "loss": 0.5035,
      "step": 21446
    },
    {
      "epoch": 1.2009743532310448,
      "grad_norm": 1.2983635663986206,
      "learning_rate": 9.962184210526317e-05,
      "loss": 0.3688,
      "step": 21447
    },
    {
      "epoch": 1.2010303505431739,
      "grad_norm": 1.4676125049591064,
      "learning_rate": 9.962157894736843e-05,
      "loss": 0.4088,
      "step": 21448
    },
    {
      "epoch": 1.2010863478553029,
      "grad_norm": 1.672722578048706,
      "learning_rate": 9.962131578947369e-05,
      "loss": 0.5554,
      "step": 21449
    },
    {
      "epoch": 1.201142345167432,
      "grad_norm": 1.4020755290985107,
      "learning_rate": 9.962105263157895e-05,
      "loss": 0.3693,
      "step": 21450
    },
    {
      "epoch": 1.201198342479561,
      "grad_norm": 2.088850498199463,
      "learning_rate": 9.962078947368422e-05,
      "loss": 0.3458,
      "step": 21451
    },
    {
      "epoch": 1.20125433979169,
      "grad_norm": 1.5886541604995728,
      "learning_rate": 9.962052631578948e-05,
      "loss": 0.5027,
      "step": 21452
    },
    {
      "epoch": 1.201310337103819,
      "grad_norm": 1.2782281637191772,
      "learning_rate": 9.962026315789474e-05,
      "loss": 0.4541,
      "step": 21453
    },
    {
      "epoch": 1.201366334415948,
      "grad_norm": 1.2737568616867065,
      "learning_rate": 9.962e-05,
      "loss": 0.4547,
      "step": 21454
    },
    {
      "epoch": 1.201422331728077,
      "grad_norm": 1.5377733707427979,
      "learning_rate": 9.961973684210527e-05,
      "loss": 0.5331,
      "step": 21455
    },
    {
      "epoch": 1.201478329040206,
      "grad_norm": 1.5774884223937988,
      "learning_rate": 9.961947368421053e-05,
      "loss": 0.4701,
      "step": 21456
    },
    {
      "epoch": 1.201534326352335,
      "grad_norm": 1.1816591024398804,
      "learning_rate": 9.96192105263158e-05,
      "loss": 0.3861,
      "step": 21457
    },
    {
      "epoch": 1.201590323664464,
      "grad_norm": 1.193567156791687,
      "learning_rate": 9.961894736842105e-05,
      "loss": 0.408,
      "step": 21458
    },
    {
      "epoch": 1.201646320976593,
      "grad_norm": 1.3032221794128418,
      "learning_rate": 9.961868421052632e-05,
      "loss": 0.4111,
      "step": 21459
    },
    {
      "epoch": 1.201702318288722,
      "grad_norm": 1.4719740152359009,
      "learning_rate": 9.961842105263158e-05,
      "loss": 0.5088,
      "step": 21460
    },
    {
      "epoch": 1.2017583156008511,
      "grad_norm": 1.2805112600326538,
      "learning_rate": 9.961815789473686e-05,
      "loss": 0.381,
      "step": 21461
    },
    {
      "epoch": 1.2018143129129801,
      "grad_norm": 1.557629108428955,
      "learning_rate": 9.961789473684212e-05,
      "loss": 0.4849,
      "step": 21462
    },
    {
      "epoch": 1.2018703102251092,
      "grad_norm": 1.222686767578125,
      "learning_rate": 9.961763157894736e-05,
      "loss": 0.4187,
      "step": 21463
    },
    {
      "epoch": 1.2019263075372382,
      "grad_norm": 1.6230337619781494,
      "learning_rate": 9.961736842105264e-05,
      "loss": 0.5211,
      "step": 21464
    },
    {
      "epoch": 1.2019823048493672,
      "grad_norm": 1.3267515897750854,
      "learning_rate": 9.96171052631579e-05,
      "loss": 0.4405,
      "step": 21465
    },
    {
      "epoch": 1.2020383021614962,
      "grad_norm": 1.1997919082641602,
      "learning_rate": 9.961684210526317e-05,
      "loss": 0.307,
      "step": 21466
    },
    {
      "epoch": 1.2020942994736252,
      "grad_norm": 1.6164220571517944,
      "learning_rate": 9.961657894736842e-05,
      "loss": 0.4621,
      "step": 21467
    },
    {
      "epoch": 1.2021502967857542,
      "grad_norm": 1.169376254081726,
      "learning_rate": 9.961631578947369e-05,
      "loss": 0.366,
      "step": 21468
    },
    {
      "epoch": 1.2022062940978833,
      "grad_norm": 1.418872594833374,
      "learning_rate": 9.961605263157895e-05,
      "loss": 0.3817,
      "step": 21469
    },
    {
      "epoch": 1.2022622914100123,
      "grad_norm": 1.3538204431533813,
      "learning_rate": 9.961578947368422e-05,
      "loss": 0.4116,
      "step": 21470
    },
    {
      "epoch": 1.2023182887221413,
      "grad_norm": 1.2791597843170166,
      "learning_rate": 9.961552631578948e-05,
      "loss": 0.4707,
      "step": 21471
    },
    {
      "epoch": 1.2023742860342703,
      "grad_norm": 1.051046371459961,
      "learning_rate": 9.961526315789474e-05,
      "loss": 0.3611,
      "step": 21472
    },
    {
      "epoch": 1.2024302833463993,
      "grad_norm": 1.4842638969421387,
      "learning_rate": 9.9615e-05,
      "loss": 0.4924,
      "step": 21473
    },
    {
      "epoch": 1.2024862806585284,
      "grad_norm": 1.563080072402954,
      "learning_rate": 9.961473684210527e-05,
      "loss": 0.5495,
      "step": 21474
    },
    {
      "epoch": 1.2025422779706574,
      "grad_norm": 1.3472084999084473,
      "learning_rate": 9.961447368421053e-05,
      "loss": 0.3709,
      "step": 21475
    },
    {
      "epoch": 1.2025982752827864,
      "grad_norm": 1.3370013236999512,
      "learning_rate": 9.96142105263158e-05,
      "loss": 0.3595,
      "step": 21476
    },
    {
      "epoch": 1.2026542725949154,
      "grad_norm": 1.3887287378311157,
      "learning_rate": 9.961394736842105e-05,
      "loss": 0.4461,
      "step": 21477
    },
    {
      "epoch": 1.2027102699070444,
      "grad_norm": 1.4351294040679932,
      "learning_rate": 9.961368421052633e-05,
      "loss": 0.4641,
      "step": 21478
    },
    {
      "epoch": 1.2027662672191735,
      "grad_norm": 1.1647791862487793,
      "learning_rate": 9.961342105263159e-05,
      "loss": 0.3386,
      "step": 21479
    },
    {
      "epoch": 1.2028222645313025,
      "grad_norm": 1.3226819038391113,
      "learning_rate": 9.961315789473685e-05,
      "loss": 0.4429,
      "step": 21480
    },
    {
      "epoch": 1.2028782618434315,
      "grad_norm": 2.2665774822235107,
      "learning_rate": 9.96128947368421e-05,
      "loss": 0.441,
      "step": 21481
    },
    {
      "epoch": 1.2029342591555605,
      "grad_norm": 1.6629362106323242,
      "learning_rate": 9.961263157894737e-05,
      "loss": 0.3264,
      "step": 21482
    },
    {
      "epoch": 1.2029902564676895,
      "grad_norm": 1.1915266513824463,
      "learning_rate": 9.961236842105264e-05,
      "loss": 0.323,
      "step": 21483
    },
    {
      "epoch": 1.2030462537798186,
      "grad_norm": 1.7877452373504639,
      "learning_rate": 9.96121052631579e-05,
      "loss": 0.4892,
      "step": 21484
    },
    {
      "epoch": 1.2031022510919476,
      "grad_norm": 1.3100731372833252,
      "learning_rate": 9.961184210526316e-05,
      "loss": 0.4047,
      "step": 21485
    },
    {
      "epoch": 1.2031582484040766,
      "grad_norm": 1.5466324090957642,
      "learning_rate": 9.961157894736842e-05,
      "loss": 0.4025,
      "step": 21486
    },
    {
      "epoch": 1.2032142457162056,
      "grad_norm": 1.2910239696502686,
      "learning_rate": 9.961131578947369e-05,
      "loss": 0.4358,
      "step": 21487
    },
    {
      "epoch": 1.2032702430283346,
      "grad_norm": 1.1227089166641235,
      "learning_rate": 9.961105263157895e-05,
      "loss": 0.3257,
      "step": 21488
    },
    {
      "epoch": 1.2033262403404636,
      "grad_norm": 3.1409835815429688,
      "learning_rate": 9.961078947368422e-05,
      "loss": 0.4212,
      "step": 21489
    },
    {
      "epoch": 1.2033822376525927,
      "grad_norm": 1.5216270685195923,
      "learning_rate": 9.961052631578947e-05,
      "loss": 0.6025,
      "step": 21490
    },
    {
      "epoch": 1.2034382349647217,
      "grad_norm": 1.5169122219085693,
      "learning_rate": 9.961026315789474e-05,
      "loss": 0.6018,
      "step": 21491
    },
    {
      "epoch": 1.2034942322768507,
      "grad_norm": 1.2840487957000732,
      "learning_rate": 9.961e-05,
      "loss": 0.3487,
      "step": 21492
    },
    {
      "epoch": 1.2035502295889797,
      "grad_norm": 1.481377363204956,
      "learning_rate": 9.960973684210528e-05,
      "loss": 0.5206,
      "step": 21493
    },
    {
      "epoch": 1.2036062269011087,
      "grad_norm": 1.334641456604004,
      "learning_rate": 9.960947368421054e-05,
      "loss": 0.3421,
      "step": 21494
    },
    {
      "epoch": 1.2036622242132378,
      "grad_norm": 1.4304063320159912,
      "learning_rate": 9.96092105263158e-05,
      "loss": 0.5398,
      "step": 21495
    },
    {
      "epoch": 1.2037182215253668,
      "grad_norm": 1.308811902999878,
      "learning_rate": 9.960894736842106e-05,
      "loss": 0.4299,
      "step": 21496
    },
    {
      "epoch": 1.2037742188374958,
      "grad_norm": 1.3681529760360718,
      "learning_rate": 9.960868421052632e-05,
      "loss": 0.5409,
      "step": 21497
    },
    {
      "epoch": 1.2038302161496248,
      "grad_norm": 1.987178921699524,
      "learning_rate": 9.960842105263159e-05,
      "loss": 0.5736,
      "step": 21498
    },
    {
      "epoch": 1.2038862134617538,
      "grad_norm": 1.3422547578811646,
      "learning_rate": 9.960815789473685e-05,
      "loss": 0.4158,
      "step": 21499
    },
    {
      "epoch": 1.2039422107738829,
      "grad_norm": 1.2379730939865112,
      "learning_rate": 9.960789473684211e-05,
      "loss": 0.3255,
      "step": 21500
    },
    {
      "epoch": 1.2039982080860119,
      "grad_norm": 1.7198050022125244,
      "learning_rate": 9.960763157894737e-05,
      "loss": 0.561,
      "step": 21501
    },
    {
      "epoch": 1.204054205398141,
      "grad_norm": 1.3510574102401733,
      "learning_rate": 9.960736842105264e-05,
      "loss": 0.4318,
      "step": 21502
    },
    {
      "epoch": 1.20411020271027,
      "grad_norm": 1.4553757905960083,
      "learning_rate": 9.96071052631579e-05,
      "loss": 0.5253,
      "step": 21503
    },
    {
      "epoch": 1.204166200022399,
      "grad_norm": 1.3594117164611816,
      "learning_rate": 9.960684210526316e-05,
      "loss": 0.3898,
      "step": 21504
    },
    {
      "epoch": 1.204222197334528,
      "grad_norm": 1.6740353107452393,
      "learning_rate": 9.960657894736842e-05,
      "loss": 0.4418,
      "step": 21505
    },
    {
      "epoch": 1.204278194646657,
      "grad_norm": 1.3242052793502808,
      "learning_rate": 9.96063157894737e-05,
      "loss": 0.4278,
      "step": 21506
    },
    {
      "epoch": 1.204334191958786,
      "grad_norm": 1.1198610067367554,
      "learning_rate": 9.960605263157895e-05,
      "loss": 0.3511,
      "step": 21507
    },
    {
      "epoch": 1.204390189270915,
      "grad_norm": 1.5841436386108398,
      "learning_rate": 9.960578947368421e-05,
      "loss": 0.4644,
      "step": 21508
    },
    {
      "epoch": 1.204446186583044,
      "grad_norm": 1.480082392692566,
      "learning_rate": 9.960552631578947e-05,
      "loss": 0.4577,
      "step": 21509
    },
    {
      "epoch": 1.204502183895173,
      "grad_norm": 1.6126313209533691,
      "learning_rate": 9.960526315789475e-05,
      "loss": 0.5515,
      "step": 21510
    },
    {
      "epoch": 1.204558181207302,
      "grad_norm": 1.250475287437439,
      "learning_rate": 9.9605e-05,
      "loss": 0.4454,
      "step": 21511
    },
    {
      "epoch": 1.204614178519431,
      "grad_norm": 1.662462830543518,
      "learning_rate": 9.960473684210528e-05,
      "loss": 0.5816,
      "step": 21512
    },
    {
      "epoch": 1.20467017583156,
      "grad_norm": 1.361572504043579,
      "learning_rate": 9.960447368421053e-05,
      "loss": 0.4464,
      "step": 21513
    },
    {
      "epoch": 1.2047261731436891,
      "grad_norm": 1.2813810110092163,
      "learning_rate": 9.960421052631579e-05,
      "loss": 0.4901,
      "step": 21514
    },
    {
      "epoch": 1.2047821704558181,
      "grad_norm": 1.5056607723236084,
      "learning_rate": 9.960394736842106e-05,
      "loss": 0.5114,
      "step": 21515
    },
    {
      "epoch": 1.2048381677679472,
      "grad_norm": 1.4565441608428955,
      "learning_rate": 9.960368421052632e-05,
      "loss": 0.5469,
      "step": 21516
    },
    {
      "epoch": 1.2048941650800762,
      "grad_norm": 1.654646635055542,
      "learning_rate": 9.960342105263159e-05,
      "loss": 0.4382,
      "step": 21517
    },
    {
      "epoch": 1.2049501623922052,
      "grad_norm": 1.1464526653289795,
      "learning_rate": 9.960315789473684e-05,
      "loss": 0.381,
      "step": 21518
    },
    {
      "epoch": 1.2050061597043342,
      "grad_norm": 2.7664570808410645,
      "learning_rate": 9.960289473684211e-05,
      "loss": 0.6341,
      "step": 21519
    },
    {
      "epoch": 1.2050621570164632,
      "grad_norm": 1.6704425811767578,
      "learning_rate": 9.960263157894737e-05,
      "loss": 0.4385,
      "step": 21520
    },
    {
      "epoch": 1.2051181543285923,
      "grad_norm": 1.5679821968078613,
      "learning_rate": 9.960236842105264e-05,
      "loss": 0.4312,
      "step": 21521
    },
    {
      "epoch": 1.2051741516407213,
      "grad_norm": 1.2296874523162842,
      "learning_rate": 9.960210526315789e-05,
      "loss": 0.4045,
      "step": 21522
    },
    {
      "epoch": 1.2052301489528503,
      "grad_norm": 1.1888378858566284,
      "learning_rate": 9.960184210526316e-05,
      "loss": 0.4166,
      "step": 21523
    },
    {
      "epoch": 1.2052861462649793,
      "grad_norm": 1.4093694686889648,
      "learning_rate": 9.960157894736842e-05,
      "loss": 0.4597,
      "step": 21524
    },
    {
      "epoch": 1.2053421435771083,
      "grad_norm": 1.5005062818527222,
      "learning_rate": 9.96013157894737e-05,
      "loss": 0.4724,
      "step": 21525
    },
    {
      "epoch": 1.2053981408892374,
      "grad_norm": 1.2124828100204468,
      "learning_rate": 9.960105263157896e-05,
      "loss": 0.3277,
      "step": 21526
    },
    {
      "epoch": 1.2054541382013664,
      "grad_norm": 1.685762882232666,
      "learning_rate": 9.960078947368422e-05,
      "loss": 0.3987,
      "step": 21527
    },
    {
      "epoch": 1.2055101355134954,
      "grad_norm": 1.3396176099777222,
      "learning_rate": 9.960052631578948e-05,
      "loss": 0.511,
      "step": 21528
    },
    {
      "epoch": 1.2055661328256244,
      "grad_norm": 1.2800570726394653,
      "learning_rate": 9.960026315789475e-05,
      "loss": 0.3879,
      "step": 21529
    },
    {
      "epoch": 1.2056221301377534,
      "grad_norm": 1.1974067687988281,
      "learning_rate": 9.960000000000001e-05,
      "loss": 0.3899,
      "step": 21530
    },
    {
      "epoch": 1.2056781274498825,
      "grad_norm": 1.2989071607589722,
      "learning_rate": 9.959973684210527e-05,
      "loss": 0.4172,
      "step": 21531
    },
    {
      "epoch": 1.2057341247620115,
      "grad_norm": 1.431694507598877,
      "learning_rate": 9.959947368421053e-05,
      "loss": 0.4452,
      "step": 21532
    },
    {
      "epoch": 1.2057901220741405,
      "grad_norm": 1.1873540878295898,
      "learning_rate": 9.959921052631579e-05,
      "loss": 0.4166,
      "step": 21533
    },
    {
      "epoch": 1.2058461193862695,
      "grad_norm": 1.2419899702072144,
      "learning_rate": 9.959894736842106e-05,
      "loss": 0.4983,
      "step": 21534
    },
    {
      "epoch": 1.2059021166983985,
      "grad_norm": 1.3938467502593994,
      "learning_rate": 9.959868421052632e-05,
      "loss": 0.5308,
      "step": 21535
    },
    {
      "epoch": 1.2059581140105275,
      "grad_norm": 1.1994540691375732,
      "learning_rate": 9.959842105263158e-05,
      "loss": 0.4284,
      "step": 21536
    },
    {
      "epoch": 1.2060141113226566,
      "grad_norm": 1.2388249635696411,
      "learning_rate": 9.959815789473684e-05,
      "loss": 0.3959,
      "step": 21537
    },
    {
      "epoch": 1.2060701086347856,
      "grad_norm": 1.3360904455184937,
      "learning_rate": 9.959789473684211e-05,
      "loss": 0.6121,
      "step": 21538
    },
    {
      "epoch": 1.2061261059469146,
      "grad_norm": 1.4434248208999634,
      "learning_rate": 9.959763157894737e-05,
      "loss": 0.4103,
      "step": 21539
    },
    {
      "epoch": 1.2061821032590436,
      "grad_norm": 1.458721399307251,
      "learning_rate": 9.959736842105263e-05,
      "loss": 0.444,
      "step": 21540
    },
    {
      "epoch": 1.2062381005711726,
      "grad_norm": 1.3926796913146973,
      "learning_rate": 9.959710526315789e-05,
      "loss": 0.5739,
      "step": 21541
    },
    {
      "epoch": 1.2062940978833017,
      "grad_norm": 1.5073621273040771,
      "learning_rate": 9.959684210526317e-05,
      "loss": 0.4406,
      "step": 21542
    },
    {
      "epoch": 1.2063500951954307,
      "grad_norm": 1.4399853944778442,
      "learning_rate": 9.959657894736843e-05,
      "loss": 0.3999,
      "step": 21543
    },
    {
      "epoch": 1.2064060925075597,
      "grad_norm": 1.4274659156799316,
      "learning_rate": 9.95963157894737e-05,
      "loss": 0.4341,
      "step": 21544
    },
    {
      "epoch": 1.2064620898196887,
      "grad_norm": 1.4583518505096436,
      "learning_rate": 9.959605263157895e-05,
      "loss": 0.4265,
      "step": 21545
    },
    {
      "epoch": 1.2065180871318177,
      "grad_norm": 1.3247216939926147,
      "learning_rate": 9.959578947368422e-05,
      "loss": 0.4782,
      "step": 21546
    },
    {
      "epoch": 1.2065740844439468,
      "grad_norm": 1.3564420938491821,
      "learning_rate": 9.959552631578948e-05,
      "loss": 0.5076,
      "step": 21547
    },
    {
      "epoch": 1.2066300817560758,
      "grad_norm": 1.6099863052368164,
      "learning_rate": 9.959526315789475e-05,
      "loss": 0.5902,
      "step": 21548
    },
    {
      "epoch": 1.2066860790682048,
      "grad_norm": 1.083328127861023,
      "learning_rate": 9.959500000000001e-05,
      "loss": 0.3665,
      "step": 21549
    },
    {
      "epoch": 1.2067420763803338,
      "grad_norm": 1.2699973583221436,
      "learning_rate": 9.959473684210526e-05,
      "loss": 0.4768,
      "step": 21550
    },
    {
      "epoch": 1.2067980736924628,
      "grad_norm": 1.5290131568908691,
      "learning_rate": 9.959447368421053e-05,
      "loss": 0.5987,
      "step": 21551
    },
    {
      "epoch": 1.2068540710045919,
      "grad_norm": 1.3370295763015747,
      "learning_rate": 9.959421052631579e-05,
      "loss": 0.3452,
      "step": 21552
    },
    {
      "epoch": 1.2069100683167209,
      "grad_norm": 1.2521337270736694,
      "learning_rate": 9.959394736842106e-05,
      "loss": 0.3931,
      "step": 21553
    },
    {
      "epoch": 1.20696606562885,
      "grad_norm": 1.3776090145111084,
      "learning_rate": 9.959368421052632e-05,
      "loss": 0.4932,
      "step": 21554
    },
    {
      "epoch": 1.207022062940979,
      "grad_norm": 1.7571406364440918,
      "learning_rate": 9.959342105263158e-05,
      "loss": 0.3998,
      "step": 21555
    },
    {
      "epoch": 1.207078060253108,
      "grad_norm": 1.3792510032653809,
      "learning_rate": 9.959315789473684e-05,
      "loss": 0.4431,
      "step": 21556
    },
    {
      "epoch": 1.207134057565237,
      "grad_norm": 1.34672212600708,
      "learning_rate": 9.959289473684212e-05,
      "loss": 0.3362,
      "step": 21557
    },
    {
      "epoch": 1.207190054877366,
      "grad_norm": 1.3030239343643188,
      "learning_rate": 9.959263157894738e-05,
      "loss": 0.4635,
      "step": 21558
    },
    {
      "epoch": 1.207246052189495,
      "grad_norm": 1.6461795568466187,
      "learning_rate": 9.959236842105264e-05,
      "loss": 0.4938,
      "step": 21559
    },
    {
      "epoch": 1.207302049501624,
      "grad_norm": 1.3412219285964966,
      "learning_rate": 9.95921052631579e-05,
      "loss": 0.4992,
      "step": 21560
    },
    {
      "epoch": 1.207358046813753,
      "grad_norm": 1.4476515054702759,
      "learning_rate": 9.959184210526317e-05,
      "loss": 0.4607,
      "step": 21561
    },
    {
      "epoch": 1.207414044125882,
      "grad_norm": 1.3024920225143433,
      "learning_rate": 9.959157894736843e-05,
      "loss": 0.447,
      "step": 21562
    },
    {
      "epoch": 1.207470041438011,
      "grad_norm": 1.2919872999191284,
      "learning_rate": 9.959131578947369e-05,
      "loss": 0.5535,
      "step": 21563
    },
    {
      "epoch": 1.20752603875014,
      "grad_norm": 1.3789916038513184,
      "learning_rate": 9.959105263157895e-05,
      "loss": 0.4159,
      "step": 21564
    },
    {
      "epoch": 1.207582036062269,
      "grad_norm": 1.3930158615112305,
      "learning_rate": 9.959078947368422e-05,
      "loss": 0.3977,
      "step": 21565
    },
    {
      "epoch": 1.2076380333743981,
      "grad_norm": 1.4170221090316772,
      "learning_rate": 9.959052631578948e-05,
      "loss": 0.5159,
      "step": 21566
    },
    {
      "epoch": 1.2076940306865271,
      "grad_norm": 1.5150773525238037,
      "learning_rate": 9.959026315789474e-05,
      "loss": 0.429,
      "step": 21567
    },
    {
      "epoch": 1.2077500279986562,
      "grad_norm": 6.028048992156982,
      "learning_rate": 9.959e-05,
      "loss": 0.5761,
      "step": 21568
    },
    {
      "epoch": 1.2078060253107852,
      "grad_norm": 1.591076135635376,
      "learning_rate": 9.958973684210526e-05,
      "loss": 0.4305,
      "step": 21569
    },
    {
      "epoch": 1.2078620226229142,
      "grad_norm": 1.4415366649627686,
      "learning_rate": 9.958947368421053e-05,
      "loss": 0.3811,
      "step": 21570
    },
    {
      "epoch": 1.2079180199350432,
      "grad_norm": 1.2981065511703491,
      "learning_rate": 9.958921052631579e-05,
      "loss": 0.4176,
      "step": 21571
    },
    {
      "epoch": 1.2079740172471722,
      "grad_norm": 1.2798056602478027,
      "learning_rate": 9.958894736842105e-05,
      "loss": 0.4618,
      "step": 21572
    },
    {
      "epoch": 1.2080300145593013,
      "grad_norm": 1.2739033699035645,
      "learning_rate": 9.958868421052631e-05,
      "loss": 0.4072,
      "step": 21573
    },
    {
      "epoch": 1.2080860118714303,
      "grad_norm": 1.510717749595642,
      "learning_rate": 9.958842105263159e-05,
      "loss": 0.6019,
      "step": 21574
    },
    {
      "epoch": 1.2081420091835593,
      "grad_norm": 1.3855395317077637,
      "learning_rate": 9.958815789473685e-05,
      "loss": 0.5853,
      "step": 21575
    },
    {
      "epoch": 1.2081980064956883,
      "grad_norm": 1.42533540725708,
      "learning_rate": 9.958789473684212e-05,
      "loss": 0.4465,
      "step": 21576
    },
    {
      "epoch": 1.2082540038078173,
      "grad_norm": 1.3876395225524902,
      "learning_rate": 9.958763157894736e-05,
      "loss": 0.4108,
      "step": 21577
    },
    {
      "epoch": 1.2083100011199464,
      "grad_norm": 1.4974334239959717,
      "learning_rate": 9.958736842105264e-05,
      "loss": 0.6935,
      "step": 21578
    },
    {
      "epoch": 1.2083659984320754,
      "grad_norm": 1.5912235975265503,
      "learning_rate": 9.95871052631579e-05,
      "loss": 0.5145,
      "step": 21579
    },
    {
      "epoch": 1.2084219957442044,
      "grad_norm": 1.3313535451889038,
      "learning_rate": 9.958684210526317e-05,
      "loss": 0.4336,
      "step": 21580
    },
    {
      "epoch": 1.2084779930563334,
      "grad_norm": 1.5140681266784668,
      "learning_rate": 9.958657894736843e-05,
      "loss": 0.4945,
      "step": 21581
    },
    {
      "epoch": 1.2085339903684624,
      "grad_norm": 1.1587097644805908,
      "learning_rate": 9.958631578947369e-05,
      "loss": 0.3736,
      "step": 21582
    },
    {
      "epoch": 1.2085899876805914,
      "grad_norm": 1.2712211608886719,
      "learning_rate": 9.958605263157895e-05,
      "loss": 0.3106,
      "step": 21583
    },
    {
      "epoch": 1.2086459849927205,
      "grad_norm": 1.2500139474868774,
      "learning_rate": 9.958578947368421e-05,
      "loss": 0.4155,
      "step": 21584
    },
    {
      "epoch": 1.2087019823048495,
      "grad_norm": 1.386947751045227,
      "learning_rate": 9.958552631578948e-05,
      "loss": 0.4118,
      "step": 21585
    },
    {
      "epoch": 1.2087579796169785,
      "grad_norm": 1.6933709383010864,
      "learning_rate": 9.958526315789474e-05,
      "loss": 0.4991,
      "step": 21586
    },
    {
      "epoch": 1.2088139769291075,
      "grad_norm": 1.3202474117279053,
      "learning_rate": 9.9585e-05,
      "loss": 0.4367,
      "step": 21587
    },
    {
      "epoch": 1.2088699742412365,
      "grad_norm": 1.3380085229873657,
      "learning_rate": 9.958473684210526e-05,
      "loss": 0.4499,
      "step": 21588
    },
    {
      "epoch": 1.2089259715533653,
      "grad_norm": 1.1275161504745483,
      "learning_rate": 9.958447368421054e-05,
      "loss": 0.3378,
      "step": 21589
    },
    {
      "epoch": 1.2089819688654944,
      "grad_norm": 1.6810859441757202,
      "learning_rate": 9.95842105263158e-05,
      "loss": 0.4943,
      "step": 21590
    },
    {
      "epoch": 1.2090379661776234,
      "grad_norm": 1.4383162260055542,
      "learning_rate": 9.958394736842106e-05,
      "loss": 0.541,
      "step": 21591
    },
    {
      "epoch": 1.2090939634897524,
      "grad_norm": 1.5488122701644897,
      "learning_rate": 9.958368421052632e-05,
      "loss": 0.661,
      "step": 21592
    },
    {
      "epoch": 1.2091499608018814,
      "grad_norm": 1.4834942817687988,
      "learning_rate": 9.958342105263159e-05,
      "loss": 0.4147,
      "step": 21593
    },
    {
      "epoch": 1.2092059581140104,
      "grad_norm": 1.313153624534607,
      "learning_rate": 9.958315789473685e-05,
      "loss": 0.3932,
      "step": 21594
    },
    {
      "epoch": 1.2092619554261395,
      "grad_norm": 1.1389399766921997,
      "learning_rate": 9.958289473684211e-05,
      "loss": 0.3893,
      "step": 21595
    },
    {
      "epoch": 1.2093179527382685,
      "grad_norm": 1.2442681789398193,
      "learning_rate": 9.958263157894737e-05,
      "loss": 0.4184,
      "step": 21596
    },
    {
      "epoch": 1.2093739500503975,
      "grad_norm": 1.3375062942504883,
      "learning_rate": 9.958236842105264e-05,
      "loss": 0.4146,
      "step": 21597
    },
    {
      "epoch": 1.2094299473625265,
      "grad_norm": 1.2993109226226807,
      "learning_rate": 9.95821052631579e-05,
      "loss": 0.4562,
      "step": 21598
    },
    {
      "epoch": 1.2094859446746555,
      "grad_norm": 1.506517767906189,
      "learning_rate": 9.958184210526317e-05,
      "loss": 0.5449,
      "step": 21599
    },
    {
      "epoch": 1.2095419419867846,
      "grad_norm": 1.4779539108276367,
      "learning_rate": 9.958157894736842e-05,
      "loss": 0.401,
      "step": 21600
    },
    {
      "epoch": 1.2095979392989136,
      "grad_norm": 1.145774483680725,
      "learning_rate": 9.958131578947368e-05,
      "loss": 0.446,
      "step": 21601
    },
    {
      "epoch": 1.2096539366110426,
      "grad_norm": 1.577304482460022,
      "learning_rate": 9.958105263157895e-05,
      "loss": 0.4408,
      "step": 21602
    },
    {
      "epoch": 1.2097099339231716,
      "grad_norm": 1.4342930316925049,
      "learning_rate": 9.958078947368421e-05,
      "loss": 0.4376,
      "step": 21603
    },
    {
      "epoch": 1.2097659312353006,
      "grad_norm": 1.370580792427063,
      "learning_rate": 9.958052631578949e-05,
      "loss": 0.3853,
      "step": 21604
    },
    {
      "epoch": 1.2098219285474296,
      "grad_norm": 1.2701548337936401,
      "learning_rate": 9.958026315789473e-05,
      "loss": 0.5224,
      "step": 21605
    },
    {
      "epoch": 1.2098779258595587,
      "grad_norm": 1.2481516599655151,
      "learning_rate": 9.958e-05,
      "loss": 0.3483,
      "step": 21606
    },
    {
      "epoch": 1.2099339231716877,
      "grad_norm": 1.2778574228286743,
      "learning_rate": 9.957973684210527e-05,
      "loss": 0.4317,
      "step": 21607
    },
    {
      "epoch": 1.2099899204838167,
      "grad_norm": 1.218406081199646,
      "learning_rate": 9.957947368421054e-05,
      "loss": 0.432,
      "step": 21608
    },
    {
      "epoch": 1.2100459177959457,
      "grad_norm": 1.25705885887146,
      "learning_rate": 9.95792105263158e-05,
      "loss": 0.4008,
      "step": 21609
    },
    {
      "epoch": 1.2101019151080747,
      "grad_norm": 1.8609923124313354,
      "learning_rate": 9.957894736842106e-05,
      "loss": 0.5195,
      "step": 21610
    },
    {
      "epoch": 1.2101579124202038,
      "grad_norm": 1.0980687141418457,
      "learning_rate": 9.957868421052632e-05,
      "loss": 0.3524,
      "step": 21611
    },
    {
      "epoch": 1.2102139097323328,
      "grad_norm": 1.4224040508270264,
      "learning_rate": 9.957842105263159e-05,
      "loss": 0.4607,
      "step": 21612
    },
    {
      "epoch": 1.2102699070444618,
      "grad_norm": 1.319006323814392,
      "learning_rate": 9.957815789473685e-05,
      "loss": 0.5207,
      "step": 21613
    },
    {
      "epoch": 1.2103259043565908,
      "grad_norm": 1.6345844268798828,
      "learning_rate": 9.957789473684211e-05,
      "loss": 0.5688,
      "step": 21614
    },
    {
      "epoch": 1.2103819016687198,
      "grad_norm": 1.1791088581085205,
      "learning_rate": 9.957763157894737e-05,
      "loss": 0.3316,
      "step": 21615
    },
    {
      "epoch": 1.2104378989808489,
      "grad_norm": 1.996338129043579,
      "learning_rate": 9.957736842105264e-05,
      "loss": 0.3288,
      "step": 21616
    },
    {
      "epoch": 1.2104938962929779,
      "grad_norm": 1.5343947410583496,
      "learning_rate": 9.95771052631579e-05,
      "loss": 0.3911,
      "step": 21617
    },
    {
      "epoch": 1.210549893605107,
      "grad_norm": 1.213384747505188,
      "learning_rate": 9.957684210526316e-05,
      "loss": 0.4806,
      "step": 21618
    },
    {
      "epoch": 1.210605890917236,
      "grad_norm": 1.6705191135406494,
      "learning_rate": 9.957657894736842e-05,
      "loss": 0.4419,
      "step": 21619
    },
    {
      "epoch": 1.210661888229365,
      "grad_norm": 1.6513222455978394,
      "learning_rate": 9.957631578947368e-05,
      "loss": 0.4492,
      "step": 21620
    },
    {
      "epoch": 1.210717885541494,
      "grad_norm": 1.4181777238845825,
      "learning_rate": 9.957605263157896e-05,
      "loss": 0.4674,
      "step": 21621
    },
    {
      "epoch": 1.210773882853623,
      "grad_norm": 4.974144458770752,
      "learning_rate": 9.957578947368422e-05,
      "loss": 0.5406,
      "step": 21622
    },
    {
      "epoch": 1.210829880165752,
      "grad_norm": 1.1779018640518188,
      "learning_rate": 9.957552631578948e-05,
      "loss": 0.4889,
      "step": 21623
    },
    {
      "epoch": 1.210885877477881,
      "grad_norm": 1.3417086601257324,
      "learning_rate": 9.957526315789473e-05,
      "loss": 0.3843,
      "step": 21624
    },
    {
      "epoch": 1.21094187479001,
      "grad_norm": 1.226450800895691,
      "learning_rate": 9.957500000000001e-05,
      "loss": 0.3566,
      "step": 21625
    },
    {
      "epoch": 1.210997872102139,
      "grad_norm": 1.1275731325149536,
      "learning_rate": 9.957473684210527e-05,
      "loss": 0.3272,
      "step": 21626
    },
    {
      "epoch": 1.211053869414268,
      "grad_norm": 1.4575562477111816,
      "learning_rate": 9.957447368421053e-05,
      "loss": 0.5366,
      "step": 21627
    },
    {
      "epoch": 1.211109866726397,
      "grad_norm": 1.3971645832061768,
      "learning_rate": 9.957421052631579e-05,
      "loss": 0.442,
      "step": 21628
    },
    {
      "epoch": 1.211165864038526,
      "grad_norm": 1.5425329208374023,
      "learning_rate": 9.957394736842106e-05,
      "loss": 0.5183,
      "step": 21629
    },
    {
      "epoch": 1.2112218613506551,
      "grad_norm": 1.0971078872680664,
      "learning_rate": 9.957368421052632e-05,
      "loss": 0.3486,
      "step": 21630
    },
    {
      "epoch": 1.2112778586627841,
      "grad_norm": 1.5646895170211792,
      "learning_rate": 9.95734210526316e-05,
      "loss": 0.4559,
      "step": 21631
    },
    {
      "epoch": 1.2113338559749132,
      "grad_norm": 1.5675113201141357,
      "learning_rate": 9.957315789473684e-05,
      "loss": 0.423,
      "step": 21632
    },
    {
      "epoch": 1.2113898532870422,
      "grad_norm": 1.320599913597107,
      "learning_rate": 9.957289473684211e-05,
      "loss": 0.3928,
      "step": 21633
    },
    {
      "epoch": 1.2114458505991712,
      "grad_norm": 1.4620963335037231,
      "learning_rate": 9.957263157894737e-05,
      "loss": 0.4002,
      "step": 21634
    },
    {
      "epoch": 1.2115018479113002,
      "grad_norm": 1.4255797863006592,
      "learning_rate": 9.957236842105263e-05,
      "loss": 0.4461,
      "step": 21635
    },
    {
      "epoch": 1.2115578452234292,
      "grad_norm": 1.2193968296051025,
      "learning_rate": 9.95721052631579e-05,
      "loss": 0.4334,
      "step": 21636
    },
    {
      "epoch": 1.2116138425355583,
      "grad_norm": 1.4193572998046875,
      "learning_rate": 9.957184210526315e-05,
      "loss": 0.5235,
      "step": 21637
    },
    {
      "epoch": 1.2116698398476873,
      "grad_norm": 1.3480080366134644,
      "learning_rate": 9.957157894736843e-05,
      "loss": 0.4982,
      "step": 21638
    },
    {
      "epoch": 1.2117258371598163,
      "grad_norm": 1.2971700429916382,
      "learning_rate": 9.957131578947368e-05,
      "loss": 0.5002,
      "step": 21639
    },
    {
      "epoch": 1.2117818344719453,
      "grad_norm": 4.556559085845947,
      "learning_rate": 9.957105263157896e-05,
      "loss": 0.4989,
      "step": 21640
    },
    {
      "epoch": 1.2118378317840743,
      "grad_norm": 1.9515577554702759,
      "learning_rate": 9.957078947368422e-05,
      "loss": 0.4755,
      "step": 21641
    },
    {
      "epoch": 1.2118938290962034,
      "grad_norm": 1.8058103322982788,
      "learning_rate": 9.957052631578948e-05,
      "loss": 0.4307,
      "step": 21642
    },
    {
      "epoch": 1.2119498264083324,
      "grad_norm": 1.3275461196899414,
      "learning_rate": 9.957026315789474e-05,
      "loss": 0.43,
      "step": 21643
    },
    {
      "epoch": 1.2120058237204614,
      "grad_norm": 3.464611530303955,
      "learning_rate": 9.957000000000001e-05,
      "loss": 0.4449,
      "step": 21644
    },
    {
      "epoch": 1.2120618210325904,
      "grad_norm": 1.3496110439300537,
      "learning_rate": 9.956973684210527e-05,
      "loss": 0.396,
      "step": 21645
    },
    {
      "epoch": 1.2121178183447194,
      "grad_norm": 1.2601943016052246,
      "learning_rate": 9.956947368421053e-05,
      "loss": 0.3952,
      "step": 21646
    },
    {
      "epoch": 1.2121738156568485,
      "grad_norm": 1.7452715635299683,
      "learning_rate": 9.956921052631579e-05,
      "loss": 0.3525,
      "step": 21647
    },
    {
      "epoch": 1.2122298129689775,
      "grad_norm": 1.5159343481063843,
      "learning_rate": 9.956894736842106e-05,
      "loss": 0.504,
      "step": 21648
    },
    {
      "epoch": 1.2122858102811065,
      "grad_norm": 1.2624573707580566,
      "learning_rate": 9.956868421052632e-05,
      "loss": 0.3595,
      "step": 21649
    },
    {
      "epoch": 1.2123418075932355,
      "grad_norm": 1.3922098875045776,
      "learning_rate": 9.956842105263158e-05,
      "loss": 0.2768,
      "step": 21650
    },
    {
      "epoch": 1.2123978049053645,
      "grad_norm": 1.1745513677597046,
      "learning_rate": 9.956815789473684e-05,
      "loss": 0.327,
      "step": 21651
    },
    {
      "epoch": 1.2124538022174935,
      "grad_norm": 1.3081787824630737,
      "learning_rate": 9.956789473684212e-05,
      "loss": 0.3648,
      "step": 21652
    },
    {
      "epoch": 1.2125097995296226,
      "grad_norm": 1.6047451496124268,
      "learning_rate": 9.956763157894738e-05,
      "loss": 0.3854,
      "step": 21653
    },
    {
      "epoch": 1.2125657968417516,
      "grad_norm": 1.5055924654006958,
      "learning_rate": 9.956736842105263e-05,
      "loss": 0.4896,
      "step": 21654
    },
    {
      "epoch": 1.2126217941538806,
      "grad_norm": 1.1641325950622559,
      "learning_rate": 9.95671052631579e-05,
      "loss": 0.3244,
      "step": 21655
    },
    {
      "epoch": 1.2126777914660096,
      "grad_norm": 1.4093161821365356,
      "learning_rate": 9.956684210526315e-05,
      "loss": 0.6209,
      "step": 21656
    },
    {
      "epoch": 1.2127337887781386,
      "grad_norm": 1.447231411933899,
      "learning_rate": 9.956657894736843e-05,
      "loss": 0.5382,
      "step": 21657
    },
    {
      "epoch": 1.2127897860902677,
      "grad_norm": 1.467686653137207,
      "learning_rate": 9.956631578947369e-05,
      "loss": 0.5004,
      "step": 21658
    },
    {
      "epoch": 1.2128457834023967,
      "grad_norm": 1.2182942628860474,
      "learning_rate": 9.956605263157896e-05,
      "loss": 0.3318,
      "step": 21659
    },
    {
      "epoch": 1.2129017807145257,
      "grad_norm": 1.369750738143921,
      "learning_rate": 9.956578947368421e-05,
      "loss": 0.4887,
      "step": 21660
    },
    {
      "epoch": 1.2129577780266547,
      "grad_norm": 1.306819200515747,
      "learning_rate": 9.956552631578948e-05,
      "loss": 0.4644,
      "step": 21661
    },
    {
      "epoch": 1.2130137753387837,
      "grad_norm": 1.2664053440093994,
      "learning_rate": 9.956526315789474e-05,
      "loss": 0.3991,
      "step": 21662
    },
    {
      "epoch": 1.2130697726509128,
      "grad_norm": 1.3251110315322876,
      "learning_rate": 9.956500000000001e-05,
      "loss": 0.4963,
      "step": 21663
    },
    {
      "epoch": 1.2131257699630418,
      "grad_norm": 1.359578013420105,
      "learning_rate": 9.956473684210527e-05,
      "loss": 0.4374,
      "step": 21664
    },
    {
      "epoch": 1.2131817672751708,
      "grad_norm": 1.6653258800506592,
      "learning_rate": 9.956447368421053e-05,
      "loss": 0.4607,
      "step": 21665
    },
    {
      "epoch": 1.2132377645872998,
      "grad_norm": 1.531223177909851,
      "learning_rate": 9.956421052631579e-05,
      "loss": 0.4853,
      "step": 21666
    },
    {
      "epoch": 1.2132937618994288,
      "grad_norm": 1.707360863685608,
      "learning_rate": 9.956394736842107e-05,
      "loss": 0.5206,
      "step": 21667
    },
    {
      "epoch": 1.2133497592115579,
      "grad_norm": 1.4159029722213745,
      "learning_rate": 9.956368421052633e-05,
      "loss": 0.4003,
      "step": 21668
    },
    {
      "epoch": 1.2134057565236869,
      "grad_norm": 1.3171870708465576,
      "learning_rate": 9.956342105263159e-05,
      "loss": 0.4966,
      "step": 21669
    },
    {
      "epoch": 1.213461753835816,
      "grad_norm": 1.4929555654525757,
      "learning_rate": 9.956315789473684e-05,
      "loss": 0.4954,
      "step": 21670
    },
    {
      "epoch": 1.213517751147945,
      "grad_norm": 1.3464287519454956,
      "learning_rate": 9.95628947368421e-05,
      "loss": 0.3948,
      "step": 21671
    },
    {
      "epoch": 1.213573748460074,
      "grad_norm": 1.375730037689209,
      "learning_rate": 9.956263157894738e-05,
      "loss": 0.4981,
      "step": 21672
    },
    {
      "epoch": 1.213629745772203,
      "grad_norm": 1.1861927509307861,
      "learning_rate": 9.956236842105264e-05,
      "loss": 0.3404,
      "step": 21673
    },
    {
      "epoch": 1.213685743084332,
      "grad_norm": 1.2284128665924072,
      "learning_rate": 9.95621052631579e-05,
      "loss": 0.4488,
      "step": 21674
    },
    {
      "epoch": 1.213741740396461,
      "grad_norm": 1.2582566738128662,
      "learning_rate": 9.956184210526316e-05,
      "loss": 0.3798,
      "step": 21675
    },
    {
      "epoch": 1.21379773770859,
      "grad_norm": 1.329533576965332,
      "learning_rate": 9.956157894736843e-05,
      "loss": 0.5515,
      "step": 21676
    },
    {
      "epoch": 1.213853735020719,
      "grad_norm": 1.1948121786117554,
      "learning_rate": 9.956131578947369e-05,
      "loss": 0.4594,
      "step": 21677
    },
    {
      "epoch": 1.213909732332848,
      "grad_norm": 1.0382745265960693,
      "learning_rate": 9.956105263157895e-05,
      "loss": 0.3061,
      "step": 21678
    },
    {
      "epoch": 1.213965729644977,
      "grad_norm": 1.280571699142456,
      "learning_rate": 9.956078947368421e-05,
      "loss": 0.5167,
      "step": 21679
    },
    {
      "epoch": 1.214021726957106,
      "grad_norm": 1.3734811544418335,
      "learning_rate": 9.956052631578948e-05,
      "loss": 0.5118,
      "step": 21680
    },
    {
      "epoch": 1.214077724269235,
      "grad_norm": 1.4955744743347168,
      "learning_rate": 9.956026315789474e-05,
      "loss": 0.3749,
      "step": 21681
    },
    {
      "epoch": 1.2141337215813641,
      "grad_norm": 1.4537034034729004,
      "learning_rate": 9.956e-05,
      "loss": 0.4924,
      "step": 21682
    },
    {
      "epoch": 1.2141897188934931,
      "grad_norm": 1.1773545742034912,
      "learning_rate": 9.955973684210526e-05,
      "loss": 0.3595,
      "step": 21683
    },
    {
      "epoch": 1.2142457162056222,
      "grad_norm": 1.5288422107696533,
      "learning_rate": 9.955947368421054e-05,
      "loss": 0.4485,
      "step": 21684
    },
    {
      "epoch": 1.2143017135177512,
      "grad_norm": 1.2528376579284668,
      "learning_rate": 9.95592105263158e-05,
      "loss": 0.4384,
      "step": 21685
    },
    {
      "epoch": 1.2143577108298802,
      "grad_norm": 1.4522837400436401,
      "learning_rate": 9.955894736842107e-05,
      "loss": 0.4189,
      "step": 21686
    },
    {
      "epoch": 1.2144137081420092,
      "grad_norm": 2.0988106727600098,
      "learning_rate": 9.955868421052631e-05,
      "loss": 0.5293,
      "step": 21687
    },
    {
      "epoch": 1.2144697054541382,
      "grad_norm": 1.4099358320236206,
      "learning_rate": 9.955842105263157e-05,
      "loss": 0.4423,
      "step": 21688
    },
    {
      "epoch": 1.2145257027662673,
      "grad_norm": 1.4312374591827393,
      "learning_rate": 9.955815789473685e-05,
      "loss": 0.5194,
      "step": 21689
    },
    {
      "epoch": 1.2145817000783963,
      "grad_norm": 1.713837742805481,
      "learning_rate": 9.955789473684211e-05,
      "loss": 0.3867,
      "step": 21690
    },
    {
      "epoch": 1.2146376973905253,
      "grad_norm": 1.3737242221832275,
      "learning_rate": 9.955763157894738e-05,
      "loss": 0.4443,
      "step": 21691
    },
    {
      "epoch": 1.2146936947026543,
      "grad_norm": 1.7588526010513306,
      "learning_rate": 9.955736842105263e-05,
      "loss": 0.7413,
      "step": 21692
    },
    {
      "epoch": 1.2147496920147833,
      "grad_norm": 1.221935749053955,
      "learning_rate": 9.95571052631579e-05,
      "loss": 0.4119,
      "step": 21693
    },
    {
      "epoch": 1.2148056893269124,
      "grad_norm": 1.4457955360412598,
      "learning_rate": 9.955684210526316e-05,
      "loss": 0.4465,
      "step": 21694
    },
    {
      "epoch": 1.2148616866390414,
      "grad_norm": 1.3118256330490112,
      "learning_rate": 9.955657894736843e-05,
      "loss": 0.4991,
      "step": 21695
    },
    {
      "epoch": 1.2149176839511704,
      "grad_norm": 1.287116527557373,
      "learning_rate": 9.955631578947369e-05,
      "loss": 0.5206,
      "step": 21696
    },
    {
      "epoch": 1.2149736812632994,
      "grad_norm": 1.5391192436218262,
      "learning_rate": 9.955605263157895e-05,
      "loss": 0.4731,
      "step": 21697
    },
    {
      "epoch": 1.2150296785754284,
      "grad_norm": 1.3973290920257568,
      "learning_rate": 9.955578947368421e-05,
      "loss": 0.4245,
      "step": 21698
    },
    {
      "epoch": 1.2150856758875574,
      "grad_norm": 1.5534621477127075,
      "learning_rate": 9.955552631578949e-05,
      "loss": 0.5042,
      "step": 21699
    },
    {
      "epoch": 1.2151416731996865,
      "grad_norm": 1.383452296257019,
      "learning_rate": 9.955526315789475e-05,
      "loss": 0.4167,
      "step": 21700
    },
    {
      "epoch": 1.2151976705118155,
      "grad_norm": 1.4800420999526978,
      "learning_rate": 9.9555e-05,
      "loss": 0.4413,
      "step": 21701
    },
    {
      "epoch": 1.2152536678239445,
      "grad_norm": 1.3420456647872925,
      "learning_rate": 9.955473684210526e-05,
      "loss": 0.5102,
      "step": 21702
    },
    {
      "epoch": 1.2153096651360735,
      "grad_norm": 1.287355661392212,
      "learning_rate": 9.955447368421054e-05,
      "loss": 0.3622,
      "step": 21703
    },
    {
      "epoch": 1.2153656624482025,
      "grad_norm": 1.442916750907898,
      "learning_rate": 9.95542105263158e-05,
      "loss": 0.443,
      "step": 21704
    },
    {
      "epoch": 1.2154216597603316,
      "grad_norm": 1.3073047399520874,
      "learning_rate": 9.955394736842106e-05,
      "loss": 0.4629,
      "step": 21705
    },
    {
      "epoch": 1.2154776570724606,
      "grad_norm": 1.3350088596343994,
      "learning_rate": 9.955368421052632e-05,
      "loss": 0.4702,
      "step": 21706
    },
    {
      "epoch": 1.2155336543845896,
      "grad_norm": 1.529811143875122,
      "learning_rate": 9.955342105263158e-05,
      "loss": 0.4841,
      "step": 21707
    },
    {
      "epoch": 1.2155896516967186,
      "grad_norm": 1.3264755010604858,
      "learning_rate": 9.955315789473685e-05,
      "loss": 0.4314,
      "step": 21708
    },
    {
      "epoch": 1.2156456490088476,
      "grad_norm": 1.4241228103637695,
      "learning_rate": 9.955289473684211e-05,
      "loss": 0.5245,
      "step": 21709
    },
    {
      "epoch": 1.2157016463209767,
      "grad_norm": 3.6906840801239014,
      "learning_rate": 9.955263157894737e-05,
      "loss": 0.4509,
      "step": 21710
    },
    {
      "epoch": 1.2157576436331057,
      "grad_norm": 1.7241957187652588,
      "learning_rate": 9.955236842105263e-05,
      "loss": 0.42,
      "step": 21711
    },
    {
      "epoch": 1.2158136409452347,
      "grad_norm": 1.3581733703613281,
      "learning_rate": 9.95521052631579e-05,
      "loss": 0.4015,
      "step": 21712
    },
    {
      "epoch": 1.2158696382573637,
      "grad_norm": 1.3444708585739136,
      "learning_rate": 9.955184210526316e-05,
      "loss": 0.5752,
      "step": 21713
    },
    {
      "epoch": 1.2159256355694927,
      "grad_norm": 1.5004737377166748,
      "learning_rate": 9.955157894736844e-05,
      "loss": 0.4007,
      "step": 21714
    },
    {
      "epoch": 1.2159816328816218,
      "grad_norm": 1.3935128450393677,
      "learning_rate": 9.955131578947368e-05,
      "loss": 0.4082,
      "step": 21715
    },
    {
      "epoch": 1.2160376301937508,
      "grad_norm": 1.5477619171142578,
      "learning_rate": 9.955105263157895e-05,
      "loss": 0.6332,
      "step": 21716
    },
    {
      "epoch": 1.2160936275058798,
      "grad_norm": 1.4518890380859375,
      "learning_rate": 9.955078947368421e-05,
      "loss": 0.3699,
      "step": 21717
    },
    {
      "epoch": 1.2161496248180088,
      "grad_norm": 1.3490943908691406,
      "learning_rate": 9.955052631578949e-05,
      "loss": 0.4337,
      "step": 21718
    },
    {
      "epoch": 1.2162056221301378,
      "grad_norm": 1.44427490234375,
      "learning_rate": 9.955026315789473e-05,
      "loss": 0.4376,
      "step": 21719
    },
    {
      "epoch": 1.2162616194422669,
      "grad_norm": 1.369771122932434,
      "learning_rate": 9.955000000000001e-05,
      "loss": 0.389,
      "step": 21720
    },
    {
      "epoch": 1.2163176167543959,
      "grad_norm": 1.5623319149017334,
      "learning_rate": 9.954973684210527e-05,
      "loss": 0.4553,
      "step": 21721
    },
    {
      "epoch": 1.2163736140665249,
      "grad_norm": 1.4721201658248901,
      "learning_rate": 9.954947368421053e-05,
      "loss": 0.5529,
      "step": 21722
    },
    {
      "epoch": 1.216429611378654,
      "grad_norm": 1.42538583278656,
      "learning_rate": 9.95492105263158e-05,
      "loss": 0.4611,
      "step": 21723
    },
    {
      "epoch": 1.216485608690783,
      "grad_norm": 1.4518930912017822,
      "learning_rate": 9.954894736842105e-05,
      "loss": 0.4414,
      "step": 21724
    },
    {
      "epoch": 1.216541606002912,
      "grad_norm": 1.5610616207122803,
      "learning_rate": 9.954868421052632e-05,
      "loss": 0.3666,
      "step": 21725
    },
    {
      "epoch": 1.216597603315041,
      "grad_norm": 1.145919680595398,
      "learning_rate": 9.954842105263158e-05,
      "loss": 0.444,
      "step": 21726
    },
    {
      "epoch": 1.21665360062717,
      "grad_norm": 1.3703927993774414,
      "learning_rate": 9.954815789473685e-05,
      "loss": 0.4152,
      "step": 21727
    },
    {
      "epoch": 1.216709597939299,
      "grad_norm": 1.3518468141555786,
      "learning_rate": 9.954789473684211e-05,
      "loss": 0.4267,
      "step": 21728
    },
    {
      "epoch": 1.216765595251428,
      "grad_norm": 1.1545908451080322,
      "learning_rate": 9.954763157894737e-05,
      "loss": 0.4011,
      "step": 21729
    },
    {
      "epoch": 1.216821592563557,
      "grad_norm": 1.2211787700653076,
      "learning_rate": 9.954736842105263e-05,
      "loss": 0.3923,
      "step": 21730
    },
    {
      "epoch": 1.216877589875686,
      "grad_norm": 1.3977153301239014,
      "learning_rate": 9.95471052631579e-05,
      "loss": 0.5393,
      "step": 21731
    },
    {
      "epoch": 1.216933587187815,
      "grad_norm": 1.1466578245162964,
      "learning_rate": 9.954684210526316e-05,
      "loss": 0.5073,
      "step": 21732
    },
    {
      "epoch": 1.216989584499944,
      "grad_norm": 1.2234030961990356,
      "learning_rate": 9.954657894736842e-05,
      "loss": 0.3789,
      "step": 21733
    },
    {
      "epoch": 1.217045581812073,
      "grad_norm": 1.222891926765442,
      "learning_rate": 9.954631578947368e-05,
      "loss": 0.4936,
      "step": 21734
    },
    {
      "epoch": 1.217101579124202,
      "grad_norm": 1.2851600646972656,
      "learning_rate": 9.954605263157896e-05,
      "loss": 0.5131,
      "step": 21735
    },
    {
      "epoch": 1.217157576436331,
      "grad_norm": 1.6744606494903564,
      "learning_rate": 9.954578947368422e-05,
      "loss": 0.5656,
      "step": 21736
    },
    {
      "epoch": 1.21721357374846,
      "grad_norm": 1.4397356510162354,
      "learning_rate": 9.954552631578948e-05,
      "loss": 0.4175,
      "step": 21737
    },
    {
      "epoch": 1.217269571060589,
      "grad_norm": 1.3127284049987793,
      "learning_rate": 9.954526315789474e-05,
      "loss": 0.6474,
      "step": 21738
    },
    {
      "epoch": 1.217325568372718,
      "grad_norm": 1.4042620658874512,
      "learning_rate": 9.9545e-05,
      "loss": 0.4045,
      "step": 21739
    },
    {
      "epoch": 1.217381565684847,
      "grad_norm": 1.4031119346618652,
      "learning_rate": 9.954473684210527e-05,
      "loss": 0.4204,
      "step": 21740
    },
    {
      "epoch": 1.217437562996976,
      "grad_norm": 1.441322684288025,
      "learning_rate": 9.954447368421053e-05,
      "loss": 0.3801,
      "step": 21741
    },
    {
      "epoch": 1.217493560309105,
      "grad_norm": 1.4328351020812988,
      "learning_rate": 9.954421052631579e-05,
      "loss": 0.4234,
      "step": 21742
    },
    {
      "epoch": 1.217549557621234,
      "grad_norm": 1.3555734157562256,
      "learning_rate": 9.954394736842105e-05,
      "loss": 0.418,
      "step": 21743
    },
    {
      "epoch": 1.217605554933363,
      "grad_norm": 1.1974542140960693,
      "learning_rate": 9.954368421052632e-05,
      "loss": 0.4094,
      "step": 21744
    },
    {
      "epoch": 1.217661552245492,
      "grad_norm": 1.3164422512054443,
      "learning_rate": 9.954342105263158e-05,
      "loss": 0.5754,
      "step": 21745
    },
    {
      "epoch": 1.2177175495576211,
      "grad_norm": 1.6550158262252808,
      "learning_rate": 9.954315789473686e-05,
      "loss": 0.4881,
      "step": 21746
    },
    {
      "epoch": 1.2177735468697501,
      "grad_norm": 1.514971375465393,
      "learning_rate": 9.95428947368421e-05,
      "loss": 0.6,
      "step": 21747
    },
    {
      "epoch": 1.2178295441818792,
      "grad_norm": 1.312474012374878,
      "learning_rate": 9.954263157894737e-05,
      "loss": 0.3611,
      "step": 21748
    },
    {
      "epoch": 1.2178855414940082,
      "grad_norm": 1.4859702587127686,
      "learning_rate": 9.954236842105263e-05,
      "loss": 0.4517,
      "step": 21749
    },
    {
      "epoch": 1.2179415388061372,
      "grad_norm": 1.599524736404419,
      "learning_rate": 9.954210526315791e-05,
      "loss": 0.486,
      "step": 21750
    },
    {
      "epoch": 1.2179975361182662,
      "grad_norm": 1.16690993309021,
      "learning_rate": 9.954184210526317e-05,
      "loss": 0.33,
      "step": 21751
    },
    {
      "epoch": 1.2180535334303952,
      "grad_norm": 1.520194172859192,
      "learning_rate": 9.954157894736843e-05,
      "loss": 0.4749,
      "step": 21752
    },
    {
      "epoch": 1.2181095307425243,
      "grad_norm": 1.2532674074172974,
      "learning_rate": 9.954131578947369e-05,
      "loss": 0.375,
      "step": 21753
    },
    {
      "epoch": 1.2181655280546533,
      "grad_norm": 1.128656029701233,
      "learning_rate": 9.954105263157896e-05,
      "loss": 0.359,
      "step": 21754
    },
    {
      "epoch": 1.2182215253667823,
      "grad_norm": 1.3241829872131348,
      "learning_rate": 9.954078947368422e-05,
      "loss": 0.4322,
      "step": 21755
    },
    {
      "epoch": 1.2182775226789113,
      "grad_norm": 1.3982354402542114,
      "learning_rate": 9.954052631578948e-05,
      "loss": 0.6009,
      "step": 21756
    },
    {
      "epoch": 1.2183335199910403,
      "grad_norm": 1.492138147354126,
      "learning_rate": 9.954026315789474e-05,
      "loss": 0.4982,
      "step": 21757
    },
    {
      "epoch": 1.2183895173031694,
      "grad_norm": 1.5098954439163208,
      "learning_rate": 9.954e-05,
      "loss": 0.5411,
      "step": 21758
    },
    {
      "epoch": 1.2184455146152984,
      "grad_norm": 1.4105461835861206,
      "learning_rate": 9.953973684210527e-05,
      "loss": 0.4803,
      "step": 21759
    },
    {
      "epoch": 1.2185015119274274,
      "grad_norm": 1.2827019691467285,
      "learning_rate": 9.953947368421053e-05,
      "loss": 0.416,
      "step": 21760
    },
    {
      "epoch": 1.2185575092395564,
      "grad_norm": 1.4231181144714355,
      "learning_rate": 9.953921052631579e-05,
      "loss": 0.4757,
      "step": 21761
    },
    {
      "epoch": 1.2186135065516854,
      "grad_norm": 1.319778323173523,
      "learning_rate": 9.953894736842105e-05,
      "loss": 0.4043,
      "step": 21762
    },
    {
      "epoch": 1.2186695038638145,
      "grad_norm": 1.1663310527801514,
      "learning_rate": 9.953868421052632e-05,
      "loss": 0.3814,
      "step": 21763
    },
    {
      "epoch": 1.2187255011759435,
      "grad_norm": 1.4952927827835083,
      "learning_rate": 9.953842105263158e-05,
      "loss": 0.4432,
      "step": 21764
    },
    {
      "epoch": 1.2187814984880725,
      "grad_norm": 1.2660924196243286,
      "learning_rate": 9.953815789473684e-05,
      "loss": 0.6356,
      "step": 21765
    },
    {
      "epoch": 1.2188374958002015,
      "grad_norm": 1.4471004009246826,
      "learning_rate": 9.95378947368421e-05,
      "loss": 0.4191,
      "step": 21766
    },
    {
      "epoch": 1.2188934931123305,
      "grad_norm": 1.4613450765609741,
      "learning_rate": 9.953763157894738e-05,
      "loss": 0.487,
      "step": 21767
    },
    {
      "epoch": 1.2189494904244595,
      "grad_norm": 2.181941032409668,
      "learning_rate": 9.953736842105264e-05,
      "loss": 0.3376,
      "step": 21768
    },
    {
      "epoch": 1.2190054877365886,
      "grad_norm": 1.267209529876709,
      "learning_rate": 9.953710526315791e-05,
      "loss": 0.3515,
      "step": 21769
    },
    {
      "epoch": 1.2190614850487176,
      "grad_norm": 1.8140164613723755,
      "learning_rate": 9.953684210526316e-05,
      "loss": 0.5538,
      "step": 21770
    },
    {
      "epoch": 1.2191174823608466,
      "grad_norm": 1.7960492372512817,
      "learning_rate": 9.953657894736843e-05,
      "loss": 0.5585,
      "step": 21771
    },
    {
      "epoch": 1.2191734796729756,
      "grad_norm": 1.3258659839630127,
      "learning_rate": 9.953631578947369e-05,
      "loss": 0.4138,
      "step": 21772
    },
    {
      "epoch": 1.2192294769851046,
      "grad_norm": 1.4085932970046997,
      "learning_rate": 9.953605263157896e-05,
      "loss": 0.5124,
      "step": 21773
    },
    {
      "epoch": 1.2192854742972337,
      "grad_norm": 1.3527405261993408,
      "learning_rate": 9.953578947368421e-05,
      "loss": 0.3983,
      "step": 21774
    },
    {
      "epoch": 1.2193414716093627,
      "grad_norm": 1.5936415195465088,
      "learning_rate": 9.953552631578947e-05,
      "loss": 0.4506,
      "step": 21775
    },
    {
      "epoch": 1.2193974689214917,
      "grad_norm": 1.4233524799346924,
      "learning_rate": 9.953526315789474e-05,
      "loss": 0.5028,
      "step": 21776
    },
    {
      "epoch": 1.2194534662336207,
      "grad_norm": 1.5624884366989136,
      "learning_rate": 9.9535e-05,
      "loss": 0.5677,
      "step": 21777
    },
    {
      "epoch": 1.2195094635457497,
      "grad_norm": 1.445286750793457,
      "learning_rate": 9.953473684210527e-05,
      "loss": 0.4837,
      "step": 21778
    },
    {
      "epoch": 1.2195654608578788,
      "grad_norm": 1.3730077743530273,
      "learning_rate": 9.953447368421052e-05,
      "loss": 0.4522,
      "step": 21779
    },
    {
      "epoch": 1.2196214581700078,
      "grad_norm": 1.4525657892227173,
      "learning_rate": 9.95342105263158e-05,
      "loss": 0.5774,
      "step": 21780
    },
    {
      "epoch": 1.2196774554821368,
      "grad_norm": 1.1443114280700684,
      "learning_rate": 9.953394736842105e-05,
      "loss": 0.3527,
      "step": 21781
    },
    {
      "epoch": 1.2197334527942658,
      "grad_norm": 1.6584999561309814,
      "learning_rate": 9.953368421052633e-05,
      "loss": 0.3506,
      "step": 21782
    },
    {
      "epoch": 1.2197894501063948,
      "grad_norm": 1.3194817304611206,
      "learning_rate": 9.953342105263159e-05,
      "loss": 0.3406,
      "step": 21783
    },
    {
      "epoch": 1.2198454474185239,
      "grad_norm": 1.5065053701400757,
      "learning_rate": 9.953315789473685e-05,
      "loss": 0.4697,
      "step": 21784
    },
    {
      "epoch": 1.2199014447306529,
      "grad_norm": 1.257745623588562,
      "learning_rate": 9.95328947368421e-05,
      "loss": 0.3553,
      "step": 21785
    },
    {
      "epoch": 1.219957442042782,
      "grad_norm": 1.4275087118148804,
      "learning_rate": 9.953263157894738e-05,
      "loss": 0.5639,
      "step": 21786
    },
    {
      "epoch": 1.220013439354911,
      "grad_norm": 1.3018772602081299,
      "learning_rate": 9.953236842105264e-05,
      "loss": 0.5343,
      "step": 21787
    },
    {
      "epoch": 1.22006943666704,
      "grad_norm": 1.5407583713531494,
      "learning_rate": 9.95321052631579e-05,
      "loss": 0.5871,
      "step": 21788
    },
    {
      "epoch": 1.220125433979169,
      "grad_norm": 1.4225414991378784,
      "learning_rate": 9.953184210526316e-05,
      "loss": 0.4552,
      "step": 21789
    },
    {
      "epoch": 1.220181431291298,
      "grad_norm": 1.134092926979065,
      "learning_rate": 9.953157894736843e-05,
      "loss": 0.4514,
      "step": 21790
    },
    {
      "epoch": 1.220237428603427,
      "grad_norm": 1.4558898210525513,
      "learning_rate": 9.953131578947369e-05,
      "loss": 0.4356,
      "step": 21791
    },
    {
      "epoch": 1.220293425915556,
      "grad_norm": 1.0426757335662842,
      "learning_rate": 9.953105263157895e-05,
      "loss": 0.4332,
      "step": 21792
    },
    {
      "epoch": 1.220349423227685,
      "grad_norm": 1.164726972579956,
      "learning_rate": 9.953078947368421e-05,
      "loss": 0.3746,
      "step": 21793
    },
    {
      "epoch": 1.220405420539814,
      "grad_norm": 1.221907138824463,
      "learning_rate": 9.953052631578947e-05,
      "loss": 0.5337,
      "step": 21794
    },
    {
      "epoch": 1.220461417851943,
      "grad_norm": 1.24656343460083,
      "learning_rate": 9.953026315789474e-05,
      "loss": 0.3993,
      "step": 21795
    },
    {
      "epoch": 1.220517415164072,
      "grad_norm": 1.378005862236023,
      "learning_rate": 9.953e-05,
      "loss": 0.4511,
      "step": 21796
    },
    {
      "epoch": 1.220573412476201,
      "grad_norm": 1.2552727460861206,
      "learning_rate": 9.952973684210526e-05,
      "loss": 0.3287,
      "step": 21797
    },
    {
      "epoch": 1.2206294097883301,
      "grad_norm": 1.2627837657928467,
      "learning_rate": 9.952947368421052e-05,
      "loss": 0.4697,
      "step": 21798
    },
    {
      "epoch": 1.2206854071004591,
      "grad_norm": 1.59735906124115,
      "learning_rate": 9.95292105263158e-05,
      "loss": 0.6023,
      "step": 21799
    },
    {
      "epoch": 1.2207414044125882,
      "grad_norm": 1.2051788568496704,
      "learning_rate": 9.952894736842106e-05,
      "loss": 0.4292,
      "step": 21800
    },
    {
      "epoch": 1.2207974017247172,
      "grad_norm": 1.4525799751281738,
      "learning_rate": 9.952868421052633e-05,
      "loss": 0.4569,
      "step": 21801
    },
    {
      "epoch": 1.2208533990368462,
      "grad_norm": 1.2709426879882812,
      "learning_rate": 9.952842105263158e-05,
      "loss": 0.34,
      "step": 21802
    },
    {
      "epoch": 1.2209093963489752,
      "grad_norm": 1.4722206592559814,
      "learning_rate": 9.952815789473685e-05,
      "loss": 0.4586,
      "step": 21803
    },
    {
      "epoch": 1.2209653936611042,
      "grad_norm": 1.3889596462249756,
      "learning_rate": 9.952789473684211e-05,
      "loss": 0.4202,
      "step": 21804
    },
    {
      "epoch": 1.2210213909732333,
      "grad_norm": 1.5910038948059082,
      "learning_rate": 9.952763157894738e-05,
      "loss": 0.4474,
      "step": 21805
    },
    {
      "epoch": 1.2210773882853623,
      "grad_norm": 1.3301708698272705,
      "learning_rate": 9.952736842105264e-05,
      "loss": 0.467,
      "step": 21806
    },
    {
      "epoch": 1.2211333855974913,
      "grad_norm": 1.9921144247055054,
      "learning_rate": 9.95271052631579e-05,
      "loss": 0.3968,
      "step": 21807
    },
    {
      "epoch": 1.2211893829096203,
      "grad_norm": 1.4369726181030273,
      "learning_rate": 9.952684210526316e-05,
      "loss": 0.4611,
      "step": 21808
    },
    {
      "epoch": 1.2212453802217493,
      "grad_norm": 1.7992271184921265,
      "learning_rate": 9.952657894736842e-05,
      "loss": 0.5521,
      "step": 21809
    },
    {
      "epoch": 1.2213013775338784,
      "grad_norm": 1.1561758518218994,
      "learning_rate": 9.95263157894737e-05,
      "loss": 0.3971,
      "step": 21810
    },
    {
      "epoch": 1.2213573748460074,
      "grad_norm": 1.366234540939331,
      "learning_rate": 9.952605263157895e-05,
      "loss": 0.5476,
      "step": 21811
    },
    {
      "epoch": 1.2214133721581364,
      "grad_norm": 1.1349620819091797,
      "learning_rate": 9.952578947368421e-05,
      "loss": 0.4166,
      "step": 21812
    },
    {
      "epoch": 1.2214693694702654,
      "grad_norm": 1.4356046915054321,
      "learning_rate": 9.952552631578947e-05,
      "loss": 0.4494,
      "step": 21813
    },
    {
      "epoch": 1.2215253667823944,
      "grad_norm": 1.5290894508361816,
      "learning_rate": 9.952526315789475e-05,
      "loss": 0.4659,
      "step": 21814
    },
    {
      "epoch": 1.2215813640945234,
      "grad_norm": 2.1164944171905518,
      "learning_rate": 9.952500000000001e-05,
      "loss": 0.4189,
      "step": 21815
    },
    {
      "epoch": 1.2216373614066525,
      "grad_norm": 1.3442543745040894,
      "learning_rate": 9.952473684210527e-05,
      "loss": 0.4612,
      "step": 21816
    },
    {
      "epoch": 1.2216933587187815,
      "grad_norm": 1.2771326303482056,
      "learning_rate": 9.952447368421053e-05,
      "loss": 0.4856,
      "step": 21817
    },
    {
      "epoch": 1.2217493560309105,
      "grad_norm": 1.2080665826797485,
      "learning_rate": 9.95242105263158e-05,
      "loss": 0.32,
      "step": 21818
    },
    {
      "epoch": 1.2218053533430395,
      "grad_norm": 1.4783048629760742,
      "learning_rate": 9.952394736842106e-05,
      "loss": 0.4008,
      "step": 21819
    },
    {
      "epoch": 1.2218613506551685,
      "grad_norm": 1.331005573272705,
      "learning_rate": 9.952368421052632e-05,
      "loss": 0.4148,
      "step": 21820
    },
    {
      "epoch": 1.2219173479672976,
      "grad_norm": 1.3131736516952515,
      "learning_rate": 9.952342105263158e-05,
      "loss": 0.5158,
      "step": 21821
    },
    {
      "epoch": 1.2219733452794266,
      "grad_norm": 1.3083288669586182,
      "learning_rate": 9.952315789473685e-05,
      "loss": 0.4283,
      "step": 21822
    },
    {
      "epoch": 1.2220293425915556,
      "grad_norm": 1.2380086183547974,
      "learning_rate": 9.952289473684211e-05,
      "loss": 0.505,
      "step": 21823
    },
    {
      "epoch": 1.2220853399036846,
      "grad_norm": 1.2146620750427246,
      "learning_rate": 9.952263157894739e-05,
      "loss": 0.4575,
      "step": 21824
    },
    {
      "epoch": 1.2221413372158136,
      "grad_norm": 1.1539934873580933,
      "learning_rate": 9.952236842105263e-05,
      "loss": 0.4541,
      "step": 21825
    },
    {
      "epoch": 1.2221973345279427,
      "grad_norm": 1.4234150648117065,
      "learning_rate": 9.952210526315789e-05,
      "loss": 0.3875,
      "step": 21826
    },
    {
      "epoch": 1.2222533318400717,
      "grad_norm": 1.3258788585662842,
      "learning_rate": 9.952184210526316e-05,
      "loss": 0.3533,
      "step": 21827
    },
    {
      "epoch": 1.2223093291522007,
      "grad_norm": 1.063740611076355,
      "learning_rate": 9.952157894736842e-05,
      "loss": 0.3701,
      "step": 21828
    },
    {
      "epoch": 1.2223653264643297,
      "grad_norm": 1.3509206771850586,
      "learning_rate": 9.952131578947368e-05,
      "loss": 0.5831,
      "step": 21829
    },
    {
      "epoch": 1.2224213237764587,
      "grad_norm": 1.35112464427948,
      "learning_rate": 9.952105263157894e-05,
      "loss": 0.3951,
      "step": 21830
    },
    {
      "epoch": 1.2224773210885878,
      "grad_norm": 1.0923048257827759,
      "learning_rate": 9.952078947368422e-05,
      "loss": 0.2773,
      "step": 21831
    },
    {
      "epoch": 1.2225333184007168,
      "grad_norm": 1.3539552688598633,
      "learning_rate": 9.952052631578948e-05,
      "loss": 0.4607,
      "step": 21832
    },
    {
      "epoch": 1.2225893157128458,
      "grad_norm": 2.1112303733825684,
      "learning_rate": 9.952026315789475e-05,
      "loss": 0.4197,
      "step": 21833
    },
    {
      "epoch": 1.2226453130249748,
      "grad_norm": 1.36155366897583,
      "learning_rate": 9.952e-05,
      "loss": 0.4435,
      "step": 21834
    },
    {
      "epoch": 1.2227013103371038,
      "grad_norm": 1.6987247467041016,
      "learning_rate": 9.951973684210527e-05,
      "loss": 0.6736,
      "step": 21835
    },
    {
      "epoch": 1.2227573076492328,
      "grad_norm": 1.2438112497329712,
      "learning_rate": 9.951947368421053e-05,
      "loss": 0.3778,
      "step": 21836
    },
    {
      "epoch": 1.2228133049613619,
      "grad_norm": 1.5742439031600952,
      "learning_rate": 9.95192105263158e-05,
      "loss": 0.4305,
      "step": 21837
    },
    {
      "epoch": 1.2228693022734909,
      "grad_norm": 1.335503101348877,
      "learning_rate": 9.951894736842106e-05,
      "loss": 0.4913,
      "step": 21838
    },
    {
      "epoch": 1.22292529958562,
      "grad_norm": 1.3608275651931763,
      "learning_rate": 9.951868421052632e-05,
      "loss": 0.3757,
      "step": 21839
    },
    {
      "epoch": 1.222981296897749,
      "grad_norm": 1.3223425149917603,
      "learning_rate": 9.951842105263158e-05,
      "loss": 0.3729,
      "step": 21840
    },
    {
      "epoch": 1.223037294209878,
      "grad_norm": 1.2835404872894287,
      "learning_rate": 9.951815789473685e-05,
      "loss": 0.3632,
      "step": 21841
    },
    {
      "epoch": 1.223093291522007,
      "grad_norm": 1.4302679300308228,
      "learning_rate": 9.951789473684211e-05,
      "loss": 0.496,
      "step": 21842
    },
    {
      "epoch": 1.223149288834136,
      "grad_norm": 2.2417383193969727,
      "learning_rate": 9.951763157894737e-05,
      "loss": 0.4393,
      "step": 21843
    },
    {
      "epoch": 1.223205286146265,
      "grad_norm": 1.3884906768798828,
      "learning_rate": 9.951736842105263e-05,
      "loss": 0.5005,
      "step": 21844
    },
    {
      "epoch": 1.223261283458394,
      "grad_norm": 1.4113117456436157,
      "learning_rate": 9.95171052631579e-05,
      "loss": 0.4553,
      "step": 21845
    },
    {
      "epoch": 1.223317280770523,
      "grad_norm": 1.2487066984176636,
      "learning_rate": 9.951684210526317e-05,
      "loss": 0.3879,
      "step": 21846
    },
    {
      "epoch": 1.223373278082652,
      "grad_norm": 1.2708920240402222,
      "learning_rate": 9.951657894736843e-05,
      "loss": 0.4038,
      "step": 21847
    },
    {
      "epoch": 1.223429275394781,
      "grad_norm": 1.1820656061172485,
      "learning_rate": 9.951631578947369e-05,
      "loss": 0.6154,
      "step": 21848
    },
    {
      "epoch": 1.22348527270691,
      "grad_norm": 1.363383173942566,
      "learning_rate": 9.951605263157895e-05,
      "loss": 0.4523,
      "step": 21849
    },
    {
      "epoch": 1.2235412700190391,
      "grad_norm": 1.3615005016326904,
      "learning_rate": 9.951578947368422e-05,
      "loss": 0.4351,
      "step": 21850
    },
    {
      "epoch": 1.2235972673311681,
      "grad_norm": 1.5357478857040405,
      "learning_rate": 9.951552631578948e-05,
      "loss": 0.3839,
      "step": 21851
    },
    {
      "epoch": 1.2236532646432972,
      "grad_norm": 1.2987799644470215,
      "learning_rate": 9.951526315789474e-05,
      "loss": 0.4115,
      "step": 21852
    },
    {
      "epoch": 1.2237092619554262,
      "grad_norm": 1.168209195137024,
      "learning_rate": 9.9515e-05,
      "loss": 0.3717,
      "step": 21853
    },
    {
      "epoch": 1.2237652592675552,
      "grad_norm": 2.50730562210083,
      "learning_rate": 9.951473684210527e-05,
      "loss": 0.4838,
      "step": 21854
    },
    {
      "epoch": 1.2238212565796842,
      "grad_norm": 1.7884670495986938,
      "learning_rate": 9.951447368421053e-05,
      "loss": 0.5113,
      "step": 21855
    },
    {
      "epoch": 1.2238772538918132,
      "grad_norm": 1.4803868532180786,
      "learning_rate": 9.95142105263158e-05,
      "loss": 0.4637,
      "step": 21856
    },
    {
      "epoch": 1.2239332512039423,
      "grad_norm": 1.1885650157928467,
      "learning_rate": 9.951394736842105e-05,
      "loss": 0.4261,
      "step": 21857
    },
    {
      "epoch": 1.2239892485160713,
      "grad_norm": 1.30832839012146,
      "learning_rate": 9.951368421052632e-05,
      "loss": 0.4534,
      "step": 21858
    },
    {
      "epoch": 1.2240452458282003,
      "grad_norm": 1.4795804023742676,
      "learning_rate": 9.951342105263158e-05,
      "loss": 0.4565,
      "step": 21859
    },
    {
      "epoch": 1.2241012431403293,
      "grad_norm": 1.636049509048462,
      "learning_rate": 9.951315789473684e-05,
      "loss": 0.5965,
      "step": 21860
    },
    {
      "epoch": 1.2241572404524583,
      "grad_norm": 1.4008960723876953,
      "learning_rate": 9.951289473684212e-05,
      "loss": 0.4041,
      "step": 21861
    },
    {
      "epoch": 1.2242132377645873,
      "grad_norm": 1.2367534637451172,
      "learning_rate": 9.951263157894736e-05,
      "loss": 0.4344,
      "step": 21862
    },
    {
      "epoch": 1.2242692350767164,
      "grad_norm": 1.3793079853057861,
      "learning_rate": 9.951236842105264e-05,
      "loss": 0.367,
      "step": 21863
    },
    {
      "epoch": 1.2243252323888454,
      "grad_norm": 1.6336742639541626,
      "learning_rate": 9.95121052631579e-05,
      "loss": 0.4368,
      "step": 21864
    },
    {
      "epoch": 1.2243812297009744,
      "grad_norm": 1.6996160745620728,
      "learning_rate": 9.951184210526317e-05,
      "loss": 0.4245,
      "step": 21865
    },
    {
      "epoch": 1.2244372270131034,
      "grad_norm": 1.3252843618392944,
      "learning_rate": 9.951157894736843e-05,
      "loss": 0.3982,
      "step": 21866
    },
    {
      "epoch": 1.2244932243252324,
      "grad_norm": 1.583376407623291,
      "learning_rate": 9.951131578947369e-05,
      "loss": 0.4504,
      "step": 21867
    },
    {
      "epoch": 1.2245492216373615,
      "grad_norm": 1.4512195587158203,
      "learning_rate": 9.951105263157895e-05,
      "loss": 0.481,
      "step": 21868
    },
    {
      "epoch": 1.2246052189494905,
      "grad_norm": 1.6560012102127075,
      "learning_rate": 9.951078947368422e-05,
      "loss": 0.4535,
      "step": 21869
    },
    {
      "epoch": 1.2246612162616195,
      "grad_norm": 3.5533268451690674,
      "learning_rate": 9.951052631578948e-05,
      "loss": 0.4402,
      "step": 21870
    },
    {
      "epoch": 1.2247172135737485,
      "grad_norm": 1.4186865091323853,
      "learning_rate": 9.951026315789474e-05,
      "loss": 0.5834,
      "step": 21871
    },
    {
      "epoch": 1.2247732108858775,
      "grad_norm": 1.329943299293518,
      "learning_rate": 9.951e-05,
      "loss": 0.4864,
      "step": 21872
    },
    {
      "epoch": 1.2248292081980066,
      "grad_norm": 1.423318862915039,
      "learning_rate": 9.950973684210527e-05,
      "loss": 0.4489,
      "step": 21873
    },
    {
      "epoch": 1.2248852055101356,
      "grad_norm": 1.3300644159317017,
      "learning_rate": 9.950947368421053e-05,
      "loss": 0.4682,
      "step": 21874
    },
    {
      "epoch": 1.2249412028222646,
      "grad_norm": 1.6161000728607178,
      "learning_rate": 9.95092105263158e-05,
      "loss": 0.4761,
      "step": 21875
    },
    {
      "epoch": 1.2249972001343936,
      "grad_norm": 1.3090307712554932,
      "learning_rate": 9.950894736842105e-05,
      "loss": 0.3507,
      "step": 21876
    },
    {
      "epoch": 1.2250531974465226,
      "grad_norm": 1.590484380722046,
      "learning_rate": 9.950868421052631e-05,
      "loss": 0.5039,
      "step": 21877
    },
    {
      "epoch": 1.2251091947586517,
      "grad_norm": 1.5652788877487183,
      "learning_rate": 9.950842105263159e-05,
      "loss": 0.409,
      "step": 21878
    },
    {
      "epoch": 1.2251651920707807,
      "grad_norm": 1.3590055704116821,
      "learning_rate": 9.950815789473685e-05,
      "loss": 0.4055,
      "step": 21879
    },
    {
      "epoch": 1.2252211893829097,
      "grad_norm": 7.608685493469238,
      "learning_rate": 9.95078947368421e-05,
      "loss": 0.3679,
      "step": 21880
    },
    {
      "epoch": 1.2252771866950387,
      "grad_norm": 1.1881346702575684,
      "learning_rate": 9.950763157894737e-05,
      "loss": 0.4288,
      "step": 21881
    },
    {
      "epoch": 1.2253331840071677,
      "grad_norm": 1.3057074546813965,
      "learning_rate": 9.950736842105264e-05,
      "loss": 0.519,
      "step": 21882
    },
    {
      "epoch": 1.2253891813192967,
      "grad_norm": 1.5817879438400269,
      "learning_rate": 9.95071052631579e-05,
      "loss": 0.6551,
      "step": 21883
    },
    {
      "epoch": 1.2254451786314258,
      "grad_norm": 1.1580766439437866,
      "learning_rate": 9.950684210526316e-05,
      "loss": 0.4142,
      "step": 21884
    },
    {
      "epoch": 1.2255011759435548,
      "grad_norm": 1.2933225631713867,
      "learning_rate": 9.950657894736842e-05,
      "loss": 0.3733,
      "step": 21885
    },
    {
      "epoch": 1.2255571732556838,
      "grad_norm": 1.273256540298462,
      "learning_rate": 9.950631578947369e-05,
      "loss": 0.346,
      "step": 21886
    },
    {
      "epoch": 1.2256131705678128,
      "grad_norm": 1.346725583076477,
      "learning_rate": 9.950605263157895e-05,
      "loss": 0.4856,
      "step": 21887
    },
    {
      "epoch": 1.2256691678799418,
      "grad_norm": 1.4436297416687012,
      "learning_rate": 9.950578947368422e-05,
      "loss": 0.3798,
      "step": 21888
    },
    {
      "epoch": 1.2257251651920709,
      "grad_norm": 1.3356685638427734,
      "learning_rate": 9.950552631578947e-05,
      "loss": 0.4504,
      "step": 21889
    },
    {
      "epoch": 1.2257811625041999,
      "grad_norm": 1.2018934488296509,
      "learning_rate": 9.950526315789474e-05,
      "loss": 0.5068,
      "step": 21890
    },
    {
      "epoch": 1.225837159816329,
      "grad_norm": 1.6328685283660889,
      "learning_rate": 9.9505e-05,
      "loss": 0.4376,
      "step": 21891
    },
    {
      "epoch": 1.225893157128458,
      "grad_norm": 1.0057116746902466,
      "learning_rate": 9.950473684210528e-05,
      "loss": 0.3189,
      "step": 21892
    },
    {
      "epoch": 1.225949154440587,
      "grad_norm": 1.5768277645111084,
      "learning_rate": 9.950447368421054e-05,
      "loss": 0.4603,
      "step": 21893
    },
    {
      "epoch": 1.226005151752716,
      "grad_norm": 1.7225438356399536,
      "learning_rate": 9.95042105263158e-05,
      "loss": 0.4968,
      "step": 21894
    },
    {
      "epoch": 1.226061149064845,
      "grad_norm": 1.4130609035491943,
      "learning_rate": 9.950394736842106e-05,
      "loss": 0.4008,
      "step": 21895
    },
    {
      "epoch": 1.226117146376974,
      "grad_norm": 1.2787760496139526,
      "learning_rate": 9.950368421052632e-05,
      "loss": 0.4211,
      "step": 21896
    },
    {
      "epoch": 1.226173143689103,
      "grad_norm": 1.7041151523590088,
      "learning_rate": 9.950342105263159e-05,
      "loss": 0.4104,
      "step": 21897
    },
    {
      "epoch": 1.226229141001232,
      "grad_norm": 1.1386144161224365,
      "learning_rate": 9.950315789473685e-05,
      "loss": 0.3393,
      "step": 21898
    },
    {
      "epoch": 1.226285138313361,
      "grad_norm": 1.2944810390472412,
      "learning_rate": 9.950289473684211e-05,
      "loss": 0.3709,
      "step": 21899
    },
    {
      "epoch": 1.22634113562549,
      "grad_norm": 1.467772126197815,
      "learning_rate": 9.950263157894737e-05,
      "loss": 0.5673,
      "step": 21900
    },
    {
      "epoch": 1.226397132937619,
      "grad_norm": 1.4492416381835938,
      "learning_rate": 9.950236842105264e-05,
      "loss": 0.5022,
      "step": 21901
    },
    {
      "epoch": 1.2264531302497481,
      "grad_norm": 1.3122985363006592,
      "learning_rate": 9.95021052631579e-05,
      "loss": 0.38,
      "step": 21902
    },
    {
      "epoch": 1.2265091275618771,
      "grad_norm": 1.4372916221618652,
      "learning_rate": 9.950184210526316e-05,
      "loss": 0.4658,
      "step": 21903
    },
    {
      "epoch": 1.2265651248740062,
      "grad_norm": 1.2543593645095825,
      "learning_rate": 9.950157894736842e-05,
      "loss": 0.4077,
      "step": 21904
    },
    {
      "epoch": 1.2266211221861352,
      "grad_norm": 1.4239617586135864,
      "learning_rate": 9.95013157894737e-05,
      "loss": 0.4194,
      "step": 21905
    },
    {
      "epoch": 1.2266771194982642,
      "grad_norm": 1.496111273765564,
      "learning_rate": 9.950105263157895e-05,
      "loss": 0.3906,
      "step": 21906
    },
    {
      "epoch": 1.2267331168103932,
      "grad_norm": 1.2875492572784424,
      "learning_rate": 9.950078947368421e-05,
      "loss": 0.4025,
      "step": 21907
    },
    {
      "epoch": 1.2267891141225222,
      "grad_norm": 1.4934632778167725,
      "learning_rate": 9.950052631578947e-05,
      "loss": 0.4507,
      "step": 21908
    },
    {
      "epoch": 1.2268451114346512,
      "grad_norm": 1.660032033920288,
      "learning_rate": 9.950026315789475e-05,
      "loss": 0.5312,
      "step": 21909
    },
    {
      "epoch": 1.2269011087467803,
      "grad_norm": 1.4773995876312256,
      "learning_rate": 9.95e-05,
      "loss": 0.498,
      "step": 21910
    },
    {
      "epoch": 1.2269571060589093,
      "grad_norm": 1.4536614418029785,
      "learning_rate": 9.949973684210528e-05,
      "loss": 0.4295,
      "step": 21911
    },
    {
      "epoch": 1.2270131033710383,
      "grad_norm": 1.2980177402496338,
      "learning_rate": 9.949947368421053e-05,
      "loss": 0.3971,
      "step": 21912
    },
    {
      "epoch": 1.2270691006831673,
      "grad_norm": 1.3747931718826294,
      "learning_rate": 9.949921052631579e-05,
      "loss": 0.458,
      "step": 21913
    },
    {
      "epoch": 1.2271250979952963,
      "grad_norm": 1.1824984550476074,
      "learning_rate": 9.949894736842106e-05,
      "loss": 0.4842,
      "step": 21914
    },
    {
      "epoch": 1.2271810953074254,
      "grad_norm": 1.315635323524475,
      "learning_rate": 9.949868421052632e-05,
      "loss": 0.4605,
      "step": 21915
    },
    {
      "epoch": 1.2272370926195544,
      "grad_norm": 1.3485517501831055,
      "learning_rate": 9.949842105263159e-05,
      "loss": 0.4314,
      "step": 21916
    },
    {
      "epoch": 1.2272930899316834,
      "grad_norm": 1.3250218629837036,
      "learning_rate": 9.949815789473684e-05,
      "loss": 0.3913,
      "step": 21917
    },
    {
      "epoch": 1.2273490872438124,
      "grad_norm": 1.4313737154006958,
      "learning_rate": 9.949789473684211e-05,
      "loss": 0.4219,
      "step": 21918
    },
    {
      "epoch": 1.2274050845559414,
      "grad_norm": 1.1929755210876465,
      "learning_rate": 9.949763157894737e-05,
      "loss": 0.4118,
      "step": 21919
    },
    {
      "epoch": 1.2274610818680702,
      "grad_norm": 1.24820876121521,
      "learning_rate": 9.949736842105264e-05,
      "loss": 0.363,
      "step": 21920
    },
    {
      "epoch": 1.2275170791801993,
      "grad_norm": 1.482637643814087,
      "learning_rate": 9.949710526315789e-05,
      "loss": 0.413,
      "step": 21921
    },
    {
      "epoch": 1.2275730764923283,
      "grad_norm": 1.4789204597473145,
      "learning_rate": 9.949684210526316e-05,
      "loss": 0.4624,
      "step": 21922
    },
    {
      "epoch": 1.2276290738044573,
      "grad_norm": 1.3046389818191528,
      "learning_rate": 9.949657894736842e-05,
      "loss": 0.385,
      "step": 21923
    },
    {
      "epoch": 1.2276850711165863,
      "grad_norm": 1.3254011869430542,
      "learning_rate": 9.94963157894737e-05,
      "loss": 0.4519,
      "step": 21924
    },
    {
      "epoch": 1.2277410684287153,
      "grad_norm": 1.443664789199829,
      "learning_rate": 9.949605263157896e-05,
      "loss": 0.5582,
      "step": 21925
    },
    {
      "epoch": 1.2277970657408444,
      "grad_norm": 1.4166078567504883,
      "learning_rate": 9.949578947368422e-05,
      "loss": 0.5071,
      "step": 21926
    },
    {
      "epoch": 1.2278530630529734,
      "grad_norm": 1.342435598373413,
      "learning_rate": 9.949552631578948e-05,
      "loss": 0.4589,
      "step": 21927
    },
    {
      "epoch": 1.2279090603651024,
      "grad_norm": 1.318955898284912,
      "learning_rate": 9.949526315789475e-05,
      "loss": 0.3732,
      "step": 21928
    },
    {
      "epoch": 1.2279650576772314,
      "grad_norm": 1.3455950021743774,
      "learning_rate": 9.949500000000001e-05,
      "loss": 0.3828,
      "step": 21929
    },
    {
      "epoch": 1.2280210549893604,
      "grad_norm": 1.5169085264205933,
      "learning_rate": 9.949473684210527e-05,
      "loss": 0.3917,
      "step": 21930
    },
    {
      "epoch": 1.2280770523014894,
      "grad_norm": 1.2915533781051636,
      "learning_rate": 9.949447368421053e-05,
      "loss": 0.396,
      "step": 21931
    },
    {
      "epoch": 1.2281330496136185,
      "grad_norm": 1.4210155010223389,
      "learning_rate": 9.949421052631579e-05,
      "loss": 0.5223,
      "step": 21932
    },
    {
      "epoch": 1.2281890469257475,
      "grad_norm": 1.20391845703125,
      "learning_rate": 9.949394736842106e-05,
      "loss": 0.3596,
      "step": 21933
    },
    {
      "epoch": 1.2282450442378765,
      "grad_norm": 1.4098650217056274,
      "learning_rate": 9.949368421052632e-05,
      "loss": 0.6011,
      "step": 21934
    },
    {
      "epoch": 1.2283010415500055,
      "grad_norm": 1.37461256980896,
      "learning_rate": 9.949342105263158e-05,
      "loss": 0.495,
      "step": 21935
    },
    {
      "epoch": 1.2283570388621345,
      "grad_norm": 1.2328859567642212,
      "learning_rate": 9.949315789473684e-05,
      "loss": 0.4217,
      "step": 21936
    },
    {
      "epoch": 1.2284130361742636,
      "grad_norm": 1.310872197151184,
      "learning_rate": 9.949289473684211e-05,
      "loss": 0.3175,
      "step": 21937
    },
    {
      "epoch": 1.2284690334863926,
      "grad_norm": 1.284334421157837,
      "learning_rate": 9.949263157894737e-05,
      "loss": 0.4362,
      "step": 21938
    },
    {
      "epoch": 1.2285250307985216,
      "grad_norm": 1.304630160331726,
      "learning_rate": 9.949236842105263e-05,
      "loss": 0.3955,
      "step": 21939
    },
    {
      "epoch": 1.2285810281106506,
      "grad_norm": 1.5838603973388672,
      "learning_rate": 9.949210526315789e-05,
      "loss": 0.5161,
      "step": 21940
    },
    {
      "epoch": 1.2286370254227796,
      "grad_norm": 1.4570293426513672,
      "learning_rate": 9.949184210526317e-05,
      "loss": 0.4528,
      "step": 21941
    },
    {
      "epoch": 1.2286930227349087,
      "grad_norm": 1.5105780363082886,
      "learning_rate": 9.949157894736843e-05,
      "loss": 0.3747,
      "step": 21942
    },
    {
      "epoch": 1.2287490200470377,
      "grad_norm": 1.4986460208892822,
      "learning_rate": 9.94913157894737e-05,
      "loss": 0.4594,
      "step": 21943
    },
    {
      "epoch": 1.2288050173591667,
      "grad_norm": 1.5359302759170532,
      "learning_rate": 9.949105263157895e-05,
      "loss": 0.4651,
      "step": 21944
    },
    {
      "epoch": 1.2288610146712957,
      "grad_norm": 1.283356785774231,
      "learning_rate": 9.949078947368422e-05,
      "loss": 0.416,
      "step": 21945
    },
    {
      "epoch": 1.2289170119834247,
      "grad_norm": 1.3096617460250854,
      "learning_rate": 9.949052631578948e-05,
      "loss": 0.3662,
      "step": 21946
    },
    {
      "epoch": 1.2289730092955538,
      "grad_norm": 1.2936625480651855,
      "learning_rate": 9.949026315789474e-05,
      "loss": 0.4686,
      "step": 21947
    },
    {
      "epoch": 1.2290290066076828,
      "grad_norm": 1.3283549547195435,
      "learning_rate": 9.949000000000001e-05,
      "loss": 0.4289,
      "step": 21948
    },
    {
      "epoch": 1.2290850039198118,
      "grad_norm": 1.4151219129562378,
      "learning_rate": 9.948973684210526e-05,
      "loss": 0.62,
      "step": 21949
    },
    {
      "epoch": 1.2291410012319408,
      "grad_norm": 1.4237349033355713,
      "learning_rate": 9.948947368421053e-05,
      "loss": 0.5798,
      "step": 21950
    },
    {
      "epoch": 1.2291969985440698,
      "grad_norm": 1.4631147384643555,
      "learning_rate": 9.948921052631579e-05,
      "loss": 0.485,
      "step": 21951
    },
    {
      "epoch": 1.2292529958561988,
      "grad_norm": 1.5169850587844849,
      "learning_rate": 9.948894736842106e-05,
      "loss": 0.5413,
      "step": 21952
    },
    {
      "epoch": 1.2293089931683279,
      "grad_norm": 1.290627121925354,
      "learning_rate": 9.948868421052632e-05,
      "loss": 0.3879,
      "step": 21953
    },
    {
      "epoch": 1.2293649904804569,
      "grad_norm": 1.3381317853927612,
      "learning_rate": 9.948842105263158e-05,
      "loss": 0.4044,
      "step": 21954
    },
    {
      "epoch": 1.229420987792586,
      "grad_norm": 1.0008268356323242,
      "learning_rate": 9.948815789473684e-05,
      "loss": 0.3143,
      "step": 21955
    },
    {
      "epoch": 1.229476985104715,
      "grad_norm": 2.7601494789123535,
      "learning_rate": 9.948789473684212e-05,
      "loss": 0.4273,
      "step": 21956
    },
    {
      "epoch": 1.229532982416844,
      "grad_norm": 1.354504108428955,
      "learning_rate": 9.948763157894738e-05,
      "loss": 0.3963,
      "step": 21957
    },
    {
      "epoch": 1.229588979728973,
      "grad_norm": 1.7259458303451538,
      "learning_rate": 9.948736842105264e-05,
      "loss": 0.5035,
      "step": 21958
    },
    {
      "epoch": 1.229644977041102,
      "grad_norm": 1.5698692798614502,
      "learning_rate": 9.94871052631579e-05,
      "loss": 0.4932,
      "step": 21959
    },
    {
      "epoch": 1.229700974353231,
      "grad_norm": 1.169247031211853,
      "learning_rate": 9.948684210526317e-05,
      "loss": 0.3707,
      "step": 21960
    },
    {
      "epoch": 1.22975697166536,
      "grad_norm": 1.2915327548980713,
      "learning_rate": 9.948657894736843e-05,
      "loss": 0.5059,
      "step": 21961
    },
    {
      "epoch": 1.229812968977489,
      "grad_norm": 1.2802354097366333,
      "learning_rate": 9.948631578947369e-05,
      "loss": 0.4147,
      "step": 21962
    },
    {
      "epoch": 1.229868966289618,
      "grad_norm": 1.6147280931472778,
      "learning_rate": 9.948605263157895e-05,
      "loss": 0.4679,
      "step": 21963
    },
    {
      "epoch": 1.229924963601747,
      "grad_norm": 1.5009760856628418,
      "learning_rate": 9.948578947368421e-05,
      "loss": 0.4806,
      "step": 21964
    },
    {
      "epoch": 1.229980960913876,
      "grad_norm": 1.4112187623977661,
      "learning_rate": 9.948552631578948e-05,
      "loss": 0.5722,
      "step": 21965
    },
    {
      "epoch": 1.2300369582260051,
      "grad_norm": 1.3325297832489014,
      "learning_rate": 9.948526315789474e-05,
      "loss": 0.4073,
      "step": 21966
    },
    {
      "epoch": 1.2300929555381341,
      "grad_norm": 1.1587578058242798,
      "learning_rate": 9.9485e-05,
      "loss": 0.3948,
      "step": 21967
    },
    {
      "epoch": 1.2301489528502632,
      "grad_norm": 1.315110683441162,
      "learning_rate": 9.948473684210526e-05,
      "loss": 0.3932,
      "step": 21968
    },
    {
      "epoch": 1.2302049501623922,
      "grad_norm": 1.4653820991516113,
      "learning_rate": 9.948447368421053e-05,
      "loss": 0.4289,
      "step": 21969
    },
    {
      "epoch": 1.2302609474745212,
      "grad_norm": 1.9477462768554688,
      "learning_rate": 9.948421052631579e-05,
      "loss": 0.7315,
      "step": 21970
    },
    {
      "epoch": 1.2303169447866502,
      "grad_norm": 1.3053503036499023,
      "learning_rate": 9.948394736842107e-05,
      "loss": 0.5627,
      "step": 21971
    },
    {
      "epoch": 1.2303729420987792,
      "grad_norm": 1.2114213705062866,
      "learning_rate": 9.948368421052631e-05,
      "loss": 0.4151,
      "step": 21972
    },
    {
      "epoch": 1.2304289394109083,
      "grad_norm": 1.4587870836257935,
      "learning_rate": 9.948342105263159e-05,
      "loss": 0.4482,
      "step": 21973
    },
    {
      "epoch": 1.2304849367230373,
      "grad_norm": 1.4069160223007202,
      "learning_rate": 9.948315789473685e-05,
      "loss": 0.4085,
      "step": 21974
    },
    {
      "epoch": 1.2305409340351663,
      "grad_norm": 1.241890549659729,
      "learning_rate": 9.948289473684212e-05,
      "loss": 0.4834,
      "step": 21975
    },
    {
      "epoch": 1.2305969313472953,
      "grad_norm": 1.5539815425872803,
      "learning_rate": 9.948263157894737e-05,
      "loss": 0.6285,
      "step": 21976
    },
    {
      "epoch": 1.2306529286594243,
      "grad_norm": 1.247409462928772,
      "learning_rate": 9.948236842105264e-05,
      "loss": 0.4226,
      "step": 21977
    },
    {
      "epoch": 1.2307089259715533,
      "grad_norm": 1.646337866783142,
      "learning_rate": 9.94821052631579e-05,
      "loss": 0.4629,
      "step": 21978
    },
    {
      "epoch": 1.2307649232836824,
      "grad_norm": 1.3517980575561523,
      "learning_rate": 9.948184210526317e-05,
      "loss": 0.3765,
      "step": 21979
    },
    {
      "epoch": 1.2308209205958114,
      "grad_norm": 1.4111981391906738,
      "learning_rate": 9.948157894736843e-05,
      "loss": 0.4171,
      "step": 21980
    },
    {
      "epoch": 1.2308769179079404,
      "grad_norm": 1.6455378532409668,
      "learning_rate": 9.948131578947368e-05,
      "loss": 0.5553,
      "step": 21981
    },
    {
      "epoch": 1.2309329152200694,
      "grad_norm": 1.2768588066101074,
      "learning_rate": 9.948105263157895e-05,
      "loss": 0.432,
      "step": 21982
    },
    {
      "epoch": 1.2309889125321984,
      "grad_norm": 1.9461045265197754,
      "learning_rate": 9.948078947368421e-05,
      "loss": 0.5182,
      "step": 21983
    },
    {
      "epoch": 1.2310449098443275,
      "grad_norm": 1.5526859760284424,
      "learning_rate": 9.948052631578948e-05,
      "loss": 0.5356,
      "step": 21984
    },
    {
      "epoch": 1.2311009071564565,
      "grad_norm": 1.4912222623825073,
      "learning_rate": 9.948026315789474e-05,
      "loss": 0.5068,
      "step": 21985
    },
    {
      "epoch": 1.2311569044685855,
      "grad_norm": 1.3449803590774536,
      "learning_rate": 9.948e-05,
      "loss": 0.4846,
      "step": 21986
    },
    {
      "epoch": 1.2312129017807145,
      "grad_norm": 1.4552032947540283,
      "learning_rate": 9.947973684210526e-05,
      "loss": 0.4275,
      "step": 21987
    },
    {
      "epoch": 1.2312688990928435,
      "grad_norm": 1.3703938722610474,
      "learning_rate": 9.947947368421054e-05,
      "loss": 0.5228,
      "step": 21988
    },
    {
      "epoch": 1.2313248964049726,
      "grad_norm": 2.451890230178833,
      "learning_rate": 9.94792105263158e-05,
      "loss": 0.3579,
      "step": 21989
    },
    {
      "epoch": 1.2313808937171016,
      "grad_norm": 1.6656391620635986,
      "learning_rate": 9.947894736842106e-05,
      "loss": 0.5006,
      "step": 21990
    },
    {
      "epoch": 1.2314368910292306,
      "grad_norm": 1.4365131855010986,
      "learning_rate": 9.947868421052632e-05,
      "loss": 0.3193,
      "step": 21991
    },
    {
      "epoch": 1.2314928883413596,
      "grad_norm": 1.176457405090332,
      "learning_rate": 9.947842105263159e-05,
      "loss": 0.335,
      "step": 21992
    },
    {
      "epoch": 1.2315488856534886,
      "grad_norm": 1.615265130996704,
      "learning_rate": 9.947815789473685e-05,
      "loss": 0.4727,
      "step": 21993
    },
    {
      "epoch": 1.2316048829656177,
      "grad_norm": 1.2442609071731567,
      "learning_rate": 9.947789473684211e-05,
      "loss": 0.5069,
      "step": 21994
    },
    {
      "epoch": 1.2316608802777467,
      "grad_norm": 1.1496793031692505,
      "learning_rate": 9.947763157894737e-05,
      "loss": 0.4064,
      "step": 21995
    },
    {
      "epoch": 1.2317168775898757,
      "grad_norm": 1.6059843301773071,
      "learning_rate": 9.947736842105264e-05,
      "loss": 0.385,
      "step": 21996
    },
    {
      "epoch": 1.2317728749020047,
      "grad_norm": 1.0907304286956787,
      "learning_rate": 9.94771052631579e-05,
      "loss": 0.3182,
      "step": 21997
    },
    {
      "epoch": 1.2318288722141337,
      "grad_norm": 1.3871437311172485,
      "learning_rate": 9.947684210526316e-05,
      "loss": 0.432,
      "step": 21998
    },
    {
      "epoch": 1.2318848695262627,
      "grad_norm": 1.5002268552780151,
      "learning_rate": 9.947657894736842e-05,
      "loss": 0.4025,
      "step": 21999
    },
    {
      "epoch": 1.2319408668383918,
      "grad_norm": 1.7191928625106812,
      "learning_rate": 9.947631578947368e-05,
      "loss": 0.459,
      "step": 22000
    },
    {
      "epoch": 1.2319968641505208,
      "grad_norm": 1.2988009452819824,
      "learning_rate": 9.947605263157895e-05,
      "loss": 0.3909,
      "step": 22001
    },
    {
      "epoch": 1.2320528614626498,
      "grad_norm": 1.4330321550369263,
      "learning_rate": 9.947578947368421e-05,
      "loss": 0.4795,
      "step": 22002
    },
    {
      "epoch": 1.2321088587747788,
      "grad_norm": 1.443952202796936,
      "learning_rate": 9.947552631578949e-05,
      "loss": 0.4142,
      "step": 22003
    },
    {
      "epoch": 1.2321648560869078,
      "grad_norm": 1.4141242504119873,
      "learning_rate": 9.947526315789473e-05,
      "loss": 0.4104,
      "step": 22004
    },
    {
      "epoch": 1.2322208533990369,
      "grad_norm": 1.3222407102584839,
      "learning_rate": 9.9475e-05,
      "loss": 0.5119,
      "step": 22005
    },
    {
      "epoch": 1.2322768507111659,
      "grad_norm": 1.3565367460250854,
      "learning_rate": 9.947473684210527e-05,
      "loss": 0.537,
      "step": 22006
    },
    {
      "epoch": 1.232332848023295,
      "grad_norm": 1.2684520483016968,
      "learning_rate": 9.947447368421054e-05,
      "loss": 0.5206,
      "step": 22007
    },
    {
      "epoch": 1.232388845335424,
      "grad_norm": 2.818842887878418,
      "learning_rate": 9.94742105263158e-05,
      "loss": 0.43,
      "step": 22008
    },
    {
      "epoch": 1.232444842647553,
      "grad_norm": 1.4474869966506958,
      "learning_rate": 9.947394736842106e-05,
      "loss": 0.5481,
      "step": 22009
    },
    {
      "epoch": 1.232500839959682,
      "grad_norm": 1.3364468812942505,
      "learning_rate": 9.947368421052632e-05,
      "loss": 0.417,
      "step": 22010
    },
    {
      "epoch": 1.232556837271811,
      "grad_norm": 1.1697131395339966,
      "learning_rate": 9.947342105263159e-05,
      "loss": 0.392,
      "step": 22011
    },
    {
      "epoch": 1.23261283458394,
      "grad_norm": 1.129098653793335,
      "learning_rate": 9.947315789473685e-05,
      "loss": 0.3236,
      "step": 22012
    },
    {
      "epoch": 1.232668831896069,
      "grad_norm": 1.2740800380706787,
      "learning_rate": 9.947289473684211e-05,
      "loss": 0.4697,
      "step": 22013
    },
    {
      "epoch": 1.232724829208198,
      "grad_norm": 1.2584391832351685,
      "learning_rate": 9.947263157894737e-05,
      "loss": 0.4366,
      "step": 22014
    },
    {
      "epoch": 1.232780826520327,
      "grad_norm": 1.5128530263900757,
      "learning_rate": 9.947236842105264e-05,
      "loss": 0.4795,
      "step": 22015
    },
    {
      "epoch": 1.232836823832456,
      "grad_norm": 1.2469381093978882,
      "learning_rate": 9.94721052631579e-05,
      "loss": 0.3744,
      "step": 22016
    },
    {
      "epoch": 1.232892821144585,
      "grad_norm": 1.313928246498108,
      "learning_rate": 9.947184210526316e-05,
      "loss": 0.5378,
      "step": 22017
    },
    {
      "epoch": 1.2329488184567141,
      "grad_norm": 1.3469971418380737,
      "learning_rate": 9.947157894736842e-05,
      "loss": 0.3226,
      "step": 22018
    },
    {
      "epoch": 1.2330048157688431,
      "grad_norm": 1.7495861053466797,
      "learning_rate": 9.947131578947368e-05,
      "loss": 0.4469,
      "step": 22019
    },
    {
      "epoch": 1.2330608130809722,
      "grad_norm": 1.542587399482727,
      "learning_rate": 9.947105263157896e-05,
      "loss": 0.4056,
      "step": 22020
    },
    {
      "epoch": 1.2331168103931012,
      "grad_norm": 1.398809552192688,
      "learning_rate": 9.947078947368422e-05,
      "loss": 0.5727,
      "step": 22021
    },
    {
      "epoch": 1.2331728077052302,
      "grad_norm": 1.3981959819793701,
      "learning_rate": 9.947052631578948e-05,
      "loss": 0.4962,
      "step": 22022
    },
    {
      "epoch": 1.2332288050173592,
      "grad_norm": 1.6644326448440552,
      "learning_rate": 9.947026315789473e-05,
      "loss": 0.7883,
      "step": 22023
    },
    {
      "epoch": 1.2332848023294882,
      "grad_norm": 1.3118822574615479,
      "learning_rate": 9.947000000000001e-05,
      "loss": 0.4914,
      "step": 22024
    },
    {
      "epoch": 1.2333407996416172,
      "grad_norm": 1.5341873168945312,
      "learning_rate": 9.946973684210527e-05,
      "loss": 0.478,
      "step": 22025
    },
    {
      "epoch": 1.2333967969537463,
      "grad_norm": 1.3558776378631592,
      "learning_rate": 9.946947368421054e-05,
      "loss": 0.4463,
      "step": 22026
    },
    {
      "epoch": 1.2334527942658753,
      "grad_norm": 1.348998785018921,
      "learning_rate": 9.946921052631579e-05,
      "loss": 0.4356,
      "step": 22027
    },
    {
      "epoch": 1.2335087915780043,
      "grad_norm": 1.3207428455352783,
      "learning_rate": 9.946894736842106e-05,
      "loss": 0.5011,
      "step": 22028
    },
    {
      "epoch": 1.2335647888901333,
      "grad_norm": 1.0505967140197754,
      "learning_rate": 9.946868421052632e-05,
      "loss": 0.3547,
      "step": 22029
    },
    {
      "epoch": 1.2336207862022623,
      "grad_norm": 1.4877405166625977,
      "learning_rate": 9.94684210526316e-05,
      "loss": 0.5126,
      "step": 22030
    },
    {
      "epoch": 1.2336767835143914,
      "grad_norm": 1.4617226123809814,
      "learning_rate": 9.946815789473684e-05,
      "loss": 0.457,
      "step": 22031
    },
    {
      "epoch": 1.2337327808265204,
      "grad_norm": 1.4387129545211792,
      "learning_rate": 9.946789473684211e-05,
      "loss": 0.5297,
      "step": 22032
    },
    {
      "epoch": 1.2337887781386494,
      "grad_norm": 1.2057104110717773,
      "learning_rate": 9.946763157894737e-05,
      "loss": 0.44,
      "step": 22033
    },
    {
      "epoch": 1.2338447754507784,
      "grad_norm": 1.140297532081604,
      "learning_rate": 9.946736842105263e-05,
      "loss": 0.4564,
      "step": 22034
    },
    {
      "epoch": 1.2339007727629074,
      "grad_norm": 1.2437033653259277,
      "learning_rate": 9.94671052631579e-05,
      "loss": 0.4692,
      "step": 22035
    },
    {
      "epoch": 1.2339567700750365,
      "grad_norm": 1.368577003479004,
      "learning_rate": 9.946684210526315e-05,
      "loss": 0.4024,
      "step": 22036
    },
    {
      "epoch": 1.2340127673871655,
      "grad_norm": 1.985155463218689,
      "learning_rate": 9.946657894736843e-05,
      "loss": 0.5369,
      "step": 22037
    },
    {
      "epoch": 1.2340687646992945,
      "grad_norm": 1.2214103937149048,
      "learning_rate": 9.946631578947369e-05,
      "loss": 0.3905,
      "step": 22038
    },
    {
      "epoch": 1.2341247620114235,
      "grad_norm": 1.4224534034729004,
      "learning_rate": 9.946605263157896e-05,
      "loss": 0.4445,
      "step": 22039
    },
    {
      "epoch": 1.2341807593235525,
      "grad_norm": 1.625382423400879,
      "learning_rate": 9.946578947368422e-05,
      "loss": 0.4275,
      "step": 22040
    },
    {
      "epoch": 1.2342367566356816,
      "grad_norm": 1.3813602924346924,
      "learning_rate": 9.946552631578948e-05,
      "loss": 0.4503,
      "step": 22041
    },
    {
      "epoch": 1.2342927539478106,
      "grad_norm": 1.353810429573059,
      "learning_rate": 9.946526315789474e-05,
      "loss": 0.5219,
      "step": 22042
    },
    {
      "epoch": 1.2343487512599396,
      "grad_norm": 1.115501880645752,
      "learning_rate": 9.946500000000001e-05,
      "loss": 0.2978,
      "step": 22043
    },
    {
      "epoch": 1.2344047485720686,
      "grad_norm": 1.4730665683746338,
      "learning_rate": 9.946473684210527e-05,
      "loss": 0.3858,
      "step": 22044
    },
    {
      "epoch": 1.2344607458841976,
      "grad_norm": 5.164679527282715,
      "learning_rate": 9.946447368421053e-05,
      "loss": 0.3937,
      "step": 22045
    },
    {
      "epoch": 1.2345167431963266,
      "grad_norm": 1.3251657485961914,
      "learning_rate": 9.946421052631579e-05,
      "loss": 0.4165,
      "step": 22046
    },
    {
      "epoch": 1.2345727405084557,
      "grad_norm": 2.141411781311035,
      "learning_rate": 9.946394736842106e-05,
      "loss": 0.593,
      "step": 22047
    },
    {
      "epoch": 1.2346287378205847,
      "grad_norm": 1.614847183227539,
      "learning_rate": 9.946368421052632e-05,
      "loss": 0.3216,
      "step": 22048
    },
    {
      "epoch": 1.2346847351327137,
      "grad_norm": 1.5325751304626465,
      "learning_rate": 9.946342105263158e-05,
      "loss": 0.5522,
      "step": 22049
    },
    {
      "epoch": 1.2347407324448427,
      "grad_norm": 1.457053542137146,
      "learning_rate": 9.946315789473684e-05,
      "loss": 0.5598,
      "step": 22050
    },
    {
      "epoch": 1.2347967297569717,
      "grad_norm": 1.3280045986175537,
      "learning_rate": 9.94628947368421e-05,
      "loss": 0.4834,
      "step": 22051
    },
    {
      "epoch": 1.2348527270691008,
      "grad_norm": 1.3620203733444214,
      "learning_rate": 9.946263157894738e-05,
      "loss": 0.4212,
      "step": 22052
    },
    {
      "epoch": 1.2349087243812298,
      "grad_norm": 1.6541097164154053,
      "learning_rate": 9.946236842105264e-05,
      "loss": 0.5099,
      "step": 22053
    },
    {
      "epoch": 1.2349647216933588,
      "grad_norm": 1.9816346168518066,
      "learning_rate": 9.94621052631579e-05,
      "loss": 0.6487,
      "step": 22054
    },
    {
      "epoch": 1.2350207190054878,
      "grad_norm": 1.4045215845108032,
      "learning_rate": 9.946184210526315e-05,
      "loss": 0.4277,
      "step": 22055
    },
    {
      "epoch": 1.2350767163176168,
      "grad_norm": 1.0655875205993652,
      "learning_rate": 9.946157894736843e-05,
      "loss": 0.3683,
      "step": 22056
    },
    {
      "epoch": 1.2351327136297459,
      "grad_norm": 1.6678571701049805,
      "learning_rate": 9.946131578947369e-05,
      "loss": 0.5642,
      "step": 22057
    },
    {
      "epoch": 1.2351887109418749,
      "grad_norm": 1.317575216293335,
      "learning_rate": 9.946105263157896e-05,
      "loss": 0.3587,
      "step": 22058
    },
    {
      "epoch": 1.235244708254004,
      "grad_norm": 1.7448569536209106,
      "learning_rate": 9.946078947368421e-05,
      "loss": 0.5373,
      "step": 22059
    },
    {
      "epoch": 1.235300705566133,
      "grad_norm": 1.504167914390564,
      "learning_rate": 9.946052631578948e-05,
      "loss": 0.4347,
      "step": 22060
    },
    {
      "epoch": 1.235356702878262,
      "grad_norm": 1.193869709968567,
      "learning_rate": 9.946026315789474e-05,
      "loss": 0.3772,
      "step": 22061
    },
    {
      "epoch": 1.235412700190391,
      "grad_norm": 1.2586314678192139,
      "learning_rate": 9.946000000000001e-05,
      "loss": 0.4539,
      "step": 22062
    },
    {
      "epoch": 1.23546869750252,
      "grad_norm": 1.4514977931976318,
      "learning_rate": 9.945973684210527e-05,
      "loss": 0.3762,
      "step": 22063
    },
    {
      "epoch": 1.235524694814649,
      "grad_norm": 2.0244147777557373,
      "learning_rate": 9.945947368421053e-05,
      "loss": 0.5464,
      "step": 22064
    },
    {
      "epoch": 1.2355806921267778,
      "grad_norm": 1.2562710046768188,
      "learning_rate": 9.945921052631579e-05,
      "loss": 0.3524,
      "step": 22065
    },
    {
      "epoch": 1.2356366894389068,
      "grad_norm": 1.542137861251831,
      "learning_rate": 9.945894736842107e-05,
      "loss": 0.552,
      "step": 22066
    },
    {
      "epoch": 1.2356926867510358,
      "grad_norm": 1.2267441749572754,
      "learning_rate": 9.945868421052633e-05,
      "loss": 0.4243,
      "step": 22067
    },
    {
      "epoch": 1.2357486840631648,
      "grad_norm": 1.5006121397018433,
      "learning_rate": 9.945842105263159e-05,
      "loss": 0.4032,
      "step": 22068
    },
    {
      "epoch": 1.2358046813752939,
      "grad_norm": 2.746145725250244,
      "learning_rate": 9.945815789473684e-05,
      "loss": 0.4965,
      "step": 22069
    },
    {
      "epoch": 1.2358606786874229,
      "grad_norm": 1.4521000385284424,
      "learning_rate": 9.94578947368421e-05,
      "loss": 0.5064,
      "step": 22070
    },
    {
      "epoch": 1.235916675999552,
      "grad_norm": 1.5078973770141602,
      "learning_rate": 9.945763157894738e-05,
      "loss": 0.4564,
      "step": 22071
    },
    {
      "epoch": 1.235972673311681,
      "grad_norm": 1.340214490890503,
      "learning_rate": 9.945736842105264e-05,
      "loss": 0.3385,
      "step": 22072
    },
    {
      "epoch": 1.23602867062381,
      "grad_norm": 2.683298110961914,
      "learning_rate": 9.94571052631579e-05,
      "loss": 0.6248,
      "step": 22073
    },
    {
      "epoch": 1.236084667935939,
      "grad_norm": 2.3832719326019287,
      "learning_rate": 9.945684210526316e-05,
      "loss": 0.3919,
      "step": 22074
    },
    {
      "epoch": 1.236140665248068,
      "grad_norm": 1.521113634109497,
      "learning_rate": 9.945657894736843e-05,
      "loss": 0.4484,
      "step": 22075
    },
    {
      "epoch": 1.236196662560197,
      "grad_norm": 1.3267568349838257,
      "learning_rate": 9.945631578947369e-05,
      "loss": 0.4635,
      "step": 22076
    },
    {
      "epoch": 1.236252659872326,
      "grad_norm": 1.3356620073318481,
      "learning_rate": 9.945605263157895e-05,
      "loss": 0.438,
      "step": 22077
    },
    {
      "epoch": 1.236308657184455,
      "grad_norm": 1.4089046716690063,
      "learning_rate": 9.945578947368421e-05,
      "loss": 0.6273,
      "step": 22078
    },
    {
      "epoch": 1.236364654496584,
      "grad_norm": 1.6194113492965698,
      "learning_rate": 9.945552631578948e-05,
      "loss": 0.4205,
      "step": 22079
    },
    {
      "epoch": 1.236420651808713,
      "grad_norm": 1.4121274948120117,
      "learning_rate": 9.945526315789474e-05,
      "loss": 0.453,
      "step": 22080
    },
    {
      "epoch": 1.236476649120842,
      "grad_norm": 1.3679141998291016,
      "learning_rate": 9.945500000000002e-05,
      "loss": 0.484,
      "step": 22081
    },
    {
      "epoch": 1.2365326464329711,
      "grad_norm": 1.5797394514083862,
      "learning_rate": 9.945473684210526e-05,
      "loss": 0.5339,
      "step": 22082
    },
    {
      "epoch": 1.2365886437451001,
      "grad_norm": 1.3631484508514404,
      "learning_rate": 9.945447368421054e-05,
      "loss": 0.3056,
      "step": 22083
    },
    {
      "epoch": 1.2366446410572292,
      "grad_norm": 2.1892387866973877,
      "learning_rate": 9.94542105263158e-05,
      "loss": 0.4002,
      "step": 22084
    },
    {
      "epoch": 1.2367006383693582,
      "grad_norm": 1.414676547050476,
      "learning_rate": 9.945394736842105e-05,
      "loss": 0.5185,
      "step": 22085
    },
    {
      "epoch": 1.2367566356814872,
      "grad_norm": 1.4846559762954712,
      "learning_rate": 9.945368421052631e-05,
      "loss": 0.5629,
      "step": 22086
    },
    {
      "epoch": 1.2368126329936162,
      "grad_norm": 1.285040259361267,
      "learning_rate": 9.945342105263157e-05,
      "loss": 0.42,
      "step": 22087
    },
    {
      "epoch": 1.2368686303057452,
      "grad_norm": 1.1735033988952637,
      "learning_rate": 9.945315789473685e-05,
      "loss": 0.4051,
      "step": 22088
    },
    {
      "epoch": 1.2369246276178743,
      "grad_norm": 1.2226616144180298,
      "learning_rate": 9.945289473684211e-05,
      "loss": 0.3276,
      "step": 22089
    },
    {
      "epoch": 1.2369806249300033,
      "grad_norm": 1.4109970331192017,
      "learning_rate": 9.945263157894738e-05,
      "loss": 0.4347,
      "step": 22090
    },
    {
      "epoch": 1.2370366222421323,
      "grad_norm": 1.3883955478668213,
      "learning_rate": 9.945236842105263e-05,
      "loss": 0.3999,
      "step": 22091
    },
    {
      "epoch": 1.2370926195542613,
      "grad_norm": 1.646531105041504,
      "learning_rate": 9.94521052631579e-05,
      "loss": 0.5099,
      "step": 22092
    },
    {
      "epoch": 1.2371486168663903,
      "grad_norm": 1.2651643753051758,
      "learning_rate": 9.945184210526316e-05,
      "loss": 0.3708,
      "step": 22093
    },
    {
      "epoch": 1.2372046141785193,
      "grad_norm": 1.265564203262329,
      "learning_rate": 9.945157894736843e-05,
      "loss": 0.398,
      "step": 22094
    },
    {
      "epoch": 1.2372606114906484,
      "grad_norm": 1.3503236770629883,
      "learning_rate": 9.945131578947369e-05,
      "loss": 0.3481,
      "step": 22095
    },
    {
      "epoch": 1.2373166088027774,
      "grad_norm": 1.5110692977905273,
      "learning_rate": 9.945105263157895e-05,
      "loss": 0.4045,
      "step": 22096
    },
    {
      "epoch": 1.2373726061149064,
      "grad_norm": 1.2529535293579102,
      "learning_rate": 9.945078947368421e-05,
      "loss": 0.4833,
      "step": 22097
    },
    {
      "epoch": 1.2374286034270354,
      "grad_norm": 1.2186033725738525,
      "learning_rate": 9.945052631578949e-05,
      "loss": 0.408,
      "step": 22098
    },
    {
      "epoch": 1.2374846007391644,
      "grad_norm": 1.1480300426483154,
      "learning_rate": 9.945026315789475e-05,
      "loss": 0.4217,
      "step": 22099
    },
    {
      "epoch": 1.2375405980512935,
      "grad_norm": 1.8707743883132935,
      "learning_rate": 9.945e-05,
      "loss": 0.4986,
      "step": 22100
    },
    {
      "epoch": 1.2375965953634225,
      "grad_norm": 1.02902090549469,
      "learning_rate": 9.944973684210526e-05,
      "loss": 0.3439,
      "step": 22101
    },
    {
      "epoch": 1.2376525926755515,
      "grad_norm": 1.3396210670471191,
      "learning_rate": 9.944947368421052e-05,
      "loss": 0.4738,
      "step": 22102
    },
    {
      "epoch": 1.2377085899876805,
      "grad_norm": 1.529415488243103,
      "learning_rate": 9.94492105263158e-05,
      "loss": 0.562,
      "step": 22103
    },
    {
      "epoch": 1.2377645872998095,
      "grad_norm": 1.4021307229995728,
      "learning_rate": 9.944894736842106e-05,
      "loss": 0.5112,
      "step": 22104
    },
    {
      "epoch": 1.2378205846119386,
      "grad_norm": 1.6247341632843018,
      "learning_rate": 9.944868421052632e-05,
      "loss": 0.4893,
      "step": 22105
    },
    {
      "epoch": 1.2378765819240676,
      "grad_norm": 2.6440608501434326,
      "learning_rate": 9.944842105263158e-05,
      "loss": 0.3807,
      "step": 22106
    },
    {
      "epoch": 1.2379325792361966,
      "grad_norm": 1.268375277519226,
      "learning_rate": 9.944815789473685e-05,
      "loss": 0.3197,
      "step": 22107
    },
    {
      "epoch": 1.2379885765483256,
      "grad_norm": 1.6202696561813354,
      "learning_rate": 9.944789473684211e-05,
      "loss": 0.5831,
      "step": 22108
    },
    {
      "epoch": 1.2380445738604546,
      "grad_norm": 1.4476566314697266,
      "learning_rate": 9.944763157894737e-05,
      "loss": 0.5444,
      "step": 22109
    },
    {
      "epoch": 1.2381005711725837,
      "grad_norm": 1.3280572891235352,
      "learning_rate": 9.944736842105263e-05,
      "loss": 0.3884,
      "step": 22110
    },
    {
      "epoch": 1.2381565684847127,
      "grad_norm": 1.6256791353225708,
      "learning_rate": 9.94471052631579e-05,
      "loss": 0.496,
      "step": 22111
    },
    {
      "epoch": 1.2382125657968417,
      "grad_norm": 1.2308534383773804,
      "learning_rate": 9.944684210526316e-05,
      "loss": 0.498,
      "step": 22112
    },
    {
      "epoch": 1.2382685631089707,
      "grad_norm": 1.2840787172317505,
      "learning_rate": 9.944657894736844e-05,
      "loss": 0.478,
      "step": 22113
    },
    {
      "epoch": 1.2383245604210997,
      "grad_norm": 1.9302207231521606,
      "learning_rate": 9.944631578947368e-05,
      "loss": 0.3156,
      "step": 22114
    },
    {
      "epoch": 1.2383805577332287,
      "grad_norm": 1.466124415397644,
      "learning_rate": 9.944605263157896e-05,
      "loss": 0.5097,
      "step": 22115
    },
    {
      "epoch": 1.2384365550453578,
      "grad_norm": 1.376786470413208,
      "learning_rate": 9.944578947368421e-05,
      "loss": 0.5116,
      "step": 22116
    },
    {
      "epoch": 1.2384925523574868,
      "grad_norm": 1.3685163259506226,
      "learning_rate": 9.944552631578949e-05,
      "loss": 0.5624,
      "step": 22117
    },
    {
      "epoch": 1.2385485496696158,
      "grad_norm": 1.5305886268615723,
      "learning_rate": 9.944526315789475e-05,
      "loss": 0.4855,
      "step": 22118
    },
    {
      "epoch": 1.2386045469817448,
      "grad_norm": 1.2085416316986084,
      "learning_rate": 9.9445e-05,
      "loss": 0.382,
      "step": 22119
    },
    {
      "epoch": 1.2386605442938738,
      "grad_norm": 1.3168268203735352,
      "learning_rate": 9.944473684210527e-05,
      "loss": 0.5213,
      "step": 22120
    },
    {
      "epoch": 1.2387165416060029,
      "grad_norm": 1.3097447156906128,
      "learning_rate": 9.944447368421053e-05,
      "loss": 0.4341,
      "step": 22121
    },
    {
      "epoch": 1.2387725389181319,
      "grad_norm": 1.3421961069107056,
      "learning_rate": 9.94442105263158e-05,
      "loss": 0.3893,
      "step": 22122
    },
    {
      "epoch": 1.238828536230261,
      "grad_norm": 1.2957892417907715,
      "learning_rate": 9.944394736842105e-05,
      "loss": 0.4352,
      "step": 22123
    },
    {
      "epoch": 1.23888453354239,
      "grad_norm": 1.2423800230026245,
      "learning_rate": 9.944368421052632e-05,
      "loss": 0.454,
      "step": 22124
    },
    {
      "epoch": 1.238940530854519,
      "grad_norm": 1.5311256647109985,
      "learning_rate": 9.944342105263158e-05,
      "loss": 0.5097,
      "step": 22125
    },
    {
      "epoch": 1.238996528166648,
      "grad_norm": 1.586444616317749,
      "learning_rate": 9.944315789473685e-05,
      "loss": 0.4933,
      "step": 22126
    },
    {
      "epoch": 1.239052525478777,
      "grad_norm": 1.3216956853866577,
      "learning_rate": 9.944289473684211e-05,
      "loss": 0.3512,
      "step": 22127
    },
    {
      "epoch": 1.239108522790906,
      "grad_norm": 1.4504178762435913,
      "learning_rate": 9.944263157894737e-05,
      "loss": 0.4363,
      "step": 22128
    },
    {
      "epoch": 1.239164520103035,
      "grad_norm": 1.3706461191177368,
      "learning_rate": 9.944236842105263e-05,
      "loss": 0.3395,
      "step": 22129
    },
    {
      "epoch": 1.239220517415164,
      "grad_norm": 1.4141967296600342,
      "learning_rate": 9.94421052631579e-05,
      "loss": 0.3825,
      "step": 22130
    },
    {
      "epoch": 1.239276514727293,
      "grad_norm": 1.3471068143844604,
      "learning_rate": 9.944184210526316e-05,
      "loss": 0.4097,
      "step": 22131
    },
    {
      "epoch": 1.239332512039422,
      "grad_norm": 1.5354331731796265,
      "learning_rate": 9.944157894736842e-05,
      "loss": 0.4644,
      "step": 22132
    },
    {
      "epoch": 1.239388509351551,
      "grad_norm": 1.3198585510253906,
      "learning_rate": 9.944131578947368e-05,
      "loss": 0.4278,
      "step": 22133
    },
    {
      "epoch": 1.2394445066636801,
      "grad_norm": 1.4136360883712769,
      "learning_rate": 9.944105263157896e-05,
      "loss": 0.3762,
      "step": 22134
    },
    {
      "epoch": 1.2395005039758091,
      "grad_norm": 1.4282978773117065,
      "learning_rate": 9.944078947368422e-05,
      "loss": 0.4303,
      "step": 22135
    },
    {
      "epoch": 1.2395565012879382,
      "grad_norm": 1.3022985458374023,
      "learning_rate": 9.944052631578949e-05,
      "loss": 0.4176,
      "step": 22136
    },
    {
      "epoch": 1.2396124986000672,
      "grad_norm": 1.4591301679611206,
      "learning_rate": 9.944026315789474e-05,
      "loss": 0.4183,
      "step": 22137
    },
    {
      "epoch": 1.2396684959121962,
      "grad_norm": 1.5927734375,
      "learning_rate": 9.944e-05,
      "loss": 0.5347,
      "step": 22138
    },
    {
      "epoch": 1.2397244932243252,
      "grad_norm": 1.2869540452957153,
      "learning_rate": 9.943973684210527e-05,
      "loss": 0.4606,
      "step": 22139
    },
    {
      "epoch": 1.2397804905364542,
      "grad_norm": 1.3185064792633057,
      "learning_rate": 9.943947368421053e-05,
      "loss": 0.6397,
      "step": 22140
    },
    {
      "epoch": 1.2398364878485832,
      "grad_norm": 1.3768068552017212,
      "learning_rate": 9.943921052631579e-05,
      "loss": 0.3936,
      "step": 22141
    },
    {
      "epoch": 1.2398924851607123,
      "grad_norm": 1.537149429321289,
      "learning_rate": 9.943894736842105e-05,
      "loss": 0.4421,
      "step": 22142
    },
    {
      "epoch": 1.2399484824728413,
      "grad_norm": 1.7025221586227417,
      "learning_rate": 9.943868421052632e-05,
      "loss": 0.6484,
      "step": 22143
    },
    {
      "epoch": 1.2400044797849703,
      "grad_norm": 1.379907488822937,
      "learning_rate": 9.943842105263158e-05,
      "loss": 0.4589,
      "step": 22144
    },
    {
      "epoch": 1.2400604770970993,
      "grad_norm": 1.6071739196777344,
      "learning_rate": 9.943815789473686e-05,
      "loss": 0.5187,
      "step": 22145
    },
    {
      "epoch": 1.2401164744092283,
      "grad_norm": 1.2255569696426392,
      "learning_rate": 9.94378947368421e-05,
      "loss": 0.5301,
      "step": 22146
    },
    {
      "epoch": 1.2401724717213574,
      "grad_norm": 1.5059415102005005,
      "learning_rate": 9.943763157894737e-05,
      "loss": 0.5854,
      "step": 22147
    },
    {
      "epoch": 1.2402284690334864,
      "grad_norm": 1.4591748714447021,
      "learning_rate": 9.943736842105263e-05,
      "loss": 0.3683,
      "step": 22148
    },
    {
      "epoch": 1.2402844663456154,
      "grad_norm": 1.818745493888855,
      "learning_rate": 9.943710526315791e-05,
      "loss": 0.5289,
      "step": 22149
    },
    {
      "epoch": 1.2403404636577444,
      "grad_norm": 1.3020381927490234,
      "learning_rate": 9.943684210526317e-05,
      "loss": 0.4695,
      "step": 22150
    },
    {
      "epoch": 1.2403964609698734,
      "grad_norm": 1.2086421251296997,
      "learning_rate": 9.943657894736843e-05,
      "loss": 0.3829,
      "step": 22151
    },
    {
      "epoch": 1.2404524582820025,
      "grad_norm": 1.4827592372894287,
      "learning_rate": 9.943631578947369e-05,
      "loss": 0.6491,
      "step": 22152
    },
    {
      "epoch": 1.2405084555941315,
      "grad_norm": 1.2764463424682617,
      "learning_rate": 9.943605263157896e-05,
      "loss": 0.4088,
      "step": 22153
    },
    {
      "epoch": 1.2405644529062605,
      "grad_norm": 1.2118462324142456,
      "learning_rate": 9.943578947368422e-05,
      "loss": 0.3283,
      "step": 22154
    },
    {
      "epoch": 1.2406204502183895,
      "grad_norm": 1.3163617849349976,
      "learning_rate": 9.943552631578948e-05,
      "loss": 0.454,
      "step": 22155
    },
    {
      "epoch": 1.2406764475305185,
      "grad_norm": 1.503927230834961,
      "learning_rate": 9.943526315789474e-05,
      "loss": 0.5271,
      "step": 22156
    },
    {
      "epoch": 1.2407324448426476,
      "grad_norm": 1.8561662435531616,
      "learning_rate": 9.9435e-05,
      "loss": 0.5902,
      "step": 22157
    },
    {
      "epoch": 1.2407884421547766,
      "grad_norm": 1.3484762907028198,
      "learning_rate": 9.943473684210527e-05,
      "loss": 0.44,
      "step": 22158
    },
    {
      "epoch": 1.2408444394669056,
      "grad_norm": 1.370915412902832,
      "learning_rate": 9.943447368421053e-05,
      "loss": 0.5732,
      "step": 22159
    },
    {
      "epoch": 1.2409004367790346,
      "grad_norm": 1.4066773653030396,
      "learning_rate": 9.943421052631579e-05,
      "loss": 0.45,
      "step": 22160
    },
    {
      "epoch": 1.2409564340911636,
      "grad_norm": 1.4867023229599,
      "learning_rate": 9.943394736842105e-05,
      "loss": 0.4574,
      "step": 22161
    },
    {
      "epoch": 1.2410124314032926,
      "grad_norm": 1.3713905811309814,
      "learning_rate": 9.943368421052632e-05,
      "loss": 0.4768,
      "step": 22162
    },
    {
      "epoch": 1.2410684287154217,
      "grad_norm": 1.5865164995193481,
      "learning_rate": 9.943342105263158e-05,
      "loss": 0.4932,
      "step": 22163
    },
    {
      "epoch": 1.2411244260275507,
      "grad_norm": 1.3177303075790405,
      "learning_rate": 9.943315789473684e-05,
      "loss": 0.4736,
      "step": 22164
    },
    {
      "epoch": 1.2411804233396797,
      "grad_norm": 1.3435002565383911,
      "learning_rate": 9.94328947368421e-05,
      "loss": 0.5665,
      "step": 22165
    },
    {
      "epoch": 1.2412364206518087,
      "grad_norm": 2.1346640586853027,
      "learning_rate": 9.943263157894738e-05,
      "loss": 0.4935,
      "step": 22166
    },
    {
      "epoch": 1.2412924179639377,
      "grad_norm": 1.5035032033920288,
      "learning_rate": 9.943236842105264e-05,
      "loss": 0.4118,
      "step": 22167
    },
    {
      "epoch": 1.2413484152760668,
      "grad_norm": 1.4672106504440308,
      "learning_rate": 9.943210526315791e-05,
      "loss": 0.3559,
      "step": 22168
    },
    {
      "epoch": 1.2414044125881958,
      "grad_norm": 1.371329426765442,
      "learning_rate": 9.943184210526316e-05,
      "loss": 0.4685,
      "step": 22169
    },
    {
      "epoch": 1.2414604099003248,
      "grad_norm": 1.3162622451782227,
      "learning_rate": 9.943157894736843e-05,
      "loss": 0.4055,
      "step": 22170
    },
    {
      "epoch": 1.2415164072124538,
      "grad_norm": 1.2451874017715454,
      "learning_rate": 9.943131578947369e-05,
      "loss": 0.5376,
      "step": 22171
    },
    {
      "epoch": 1.2415724045245828,
      "grad_norm": 1.2837245464324951,
      "learning_rate": 9.943105263157895e-05,
      "loss": 0.473,
      "step": 22172
    },
    {
      "epoch": 1.2416284018367119,
      "grad_norm": 1.3146480321884155,
      "learning_rate": 9.943078947368422e-05,
      "loss": 0.3934,
      "step": 22173
    },
    {
      "epoch": 1.2416843991488409,
      "grad_norm": 1.25737464427948,
      "learning_rate": 9.943052631578947e-05,
      "loss": 0.3653,
      "step": 22174
    },
    {
      "epoch": 1.24174039646097,
      "grad_norm": 1.4885339736938477,
      "learning_rate": 9.943026315789474e-05,
      "loss": 0.3836,
      "step": 22175
    },
    {
      "epoch": 1.241796393773099,
      "grad_norm": 1.474367380142212,
      "learning_rate": 9.943e-05,
      "loss": 0.5566,
      "step": 22176
    },
    {
      "epoch": 1.241852391085228,
      "grad_norm": 1.4683201313018799,
      "learning_rate": 9.942973684210528e-05,
      "loss": 0.4191,
      "step": 22177
    },
    {
      "epoch": 1.241908388397357,
      "grad_norm": 1.3633314371109009,
      "learning_rate": 9.942947368421052e-05,
      "loss": 0.4509,
      "step": 22178
    },
    {
      "epoch": 1.241964385709486,
      "grad_norm": 1.3684865236282349,
      "learning_rate": 9.94292105263158e-05,
      "loss": 0.4024,
      "step": 22179
    },
    {
      "epoch": 1.242020383021615,
      "grad_norm": 1.5431170463562012,
      "learning_rate": 9.942894736842105e-05,
      "loss": 0.5027,
      "step": 22180
    },
    {
      "epoch": 1.242076380333744,
      "grad_norm": 1.3154023885726929,
      "learning_rate": 9.942868421052633e-05,
      "loss": 0.5557,
      "step": 22181
    },
    {
      "epoch": 1.242132377645873,
      "grad_norm": 1.136061429977417,
      "learning_rate": 9.942842105263159e-05,
      "loss": 0.4273,
      "step": 22182
    },
    {
      "epoch": 1.242188374958002,
      "grad_norm": 1.4089622497558594,
      "learning_rate": 9.942815789473685e-05,
      "loss": 0.4442,
      "step": 22183
    },
    {
      "epoch": 1.242244372270131,
      "grad_norm": 1.4251272678375244,
      "learning_rate": 9.94278947368421e-05,
      "loss": 0.4511,
      "step": 22184
    },
    {
      "epoch": 1.24230036958226,
      "grad_norm": 1.441521406173706,
      "learning_rate": 9.942763157894738e-05,
      "loss": 0.4659,
      "step": 22185
    },
    {
      "epoch": 1.242356366894389,
      "grad_norm": 1.539623737335205,
      "learning_rate": 9.942736842105264e-05,
      "loss": 0.5026,
      "step": 22186
    },
    {
      "epoch": 1.2424123642065181,
      "grad_norm": 1.4089666604995728,
      "learning_rate": 9.94271052631579e-05,
      "loss": 0.3992,
      "step": 22187
    },
    {
      "epoch": 1.2424683615186471,
      "grad_norm": 1.3999193906784058,
      "learning_rate": 9.942684210526316e-05,
      "loss": 0.6125,
      "step": 22188
    },
    {
      "epoch": 1.2425243588307762,
      "grad_norm": 1.5073438882827759,
      "learning_rate": 9.942657894736842e-05,
      "loss": 0.4619,
      "step": 22189
    },
    {
      "epoch": 1.2425803561429052,
      "grad_norm": 1.6267902851104736,
      "learning_rate": 9.942631578947369e-05,
      "loss": 0.5663,
      "step": 22190
    },
    {
      "epoch": 1.2426363534550342,
      "grad_norm": 1.5895700454711914,
      "learning_rate": 9.942605263157895e-05,
      "loss": 0.4105,
      "step": 22191
    },
    {
      "epoch": 1.2426923507671632,
      "grad_norm": 1.6286300420761108,
      "learning_rate": 9.942578947368421e-05,
      "loss": 0.5225,
      "step": 22192
    },
    {
      "epoch": 1.2427483480792922,
      "grad_norm": 1.1963218450546265,
      "learning_rate": 9.942552631578947e-05,
      "loss": 0.3773,
      "step": 22193
    },
    {
      "epoch": 1.2428043453914213,
      "grad_norm": 1.268042802810669,
      "learning_rate": 9.942526315789474e-05,
      "loss": 0.4605,
      "step": 22194
    },
    {
      "epoch": 1.2428603427035503,
      "grad_norm": 1.5324139595031738,
      "learning_rate": 9.9425e-05,
      "loss": 0.4588,
      "step": 22195
    },
    {
      "epoch": 1.2429163400156793,
      "grad_norm": 1.2663476467132568,
      "learning_rate": 9.942473684210526e-05,
      "loss": 0.4409,
      "step": 22196
    },
    {
      "epoch": 1.2429723373278083,
      "grad_norm": 1.4255861043930054,
      "learning_rate": 9.942447368421052e-05,
      "loss": 0.4213,
      "step": 22197
    },
    {
      "epoch": 1.2430283346399373,
      "grad_norm": 1.4221341609954834,
      "learning_rate": 9.94242105263158e-05,
      "loss": 0.4848,
      "step": 22198
    },
    {
      "epoch": 1.2430843319520664,
      "grad_norm": 1.505860686302185,
      "learning_rate": 9.942394736842106e-05,
      "loss": 0.6597,
      "step": 22199
    },
    {
      "epoch": 1.2431403292641954,
      "grad_norm": 1.4762026071548462,
      "learning_rate": 9.942368421052633e-05,
      "loss": 0.4869,
      "step": 22200
    },
    {
      "epoch": 1.2431963265763244,
      "grad_norm": 1.6010220050811768,
      "learning_rate": 9.942342105263158e-05,
      "loss": 0.5208,
      "step": 22201
    },
    {
      "epoch": 1.2432523238884534,
      "grad_norm": 1.171999216079712,
      "learning_rate": 9.942315789473685e-05,
      "loss": 0.3945,
      "step": 22202
    },
    {
      "epoch": 1.2433083212005824,
      "grad_norm": 1.299700379371643,
      "learning_rate": 9.942289473684211e-05,
      "loss": 0.4803,
      "step": 22203
    },
    {
      "epoch": 1.2433643185127115,
      "grad_norm": 1.5711313486099243,
      "learning_rate": 9.942263157894738e-05,
      "loss": 0.4983,
      "step": 22204
    },
    {
      "epoch": 1.2434203158248405,
      "grad_norm": 1.389089822769165,
      "learning_rate": 9.942236842105264e-05,
      "loss": 0.4457,
      "step": 22205
    },
    {
      "epoch": 1.2434763131369695,
      "grad_norm": 1.341274380683899,
      "learning_rate": 9.942210526315789e-05,
      "loss": 0.3424,
      "step": 22206
    },
    {
      "epoch": 1.2435323104490985,
      "grad_norm": 1.1177542209625244,
      "learning_rate": 9.942184210526316e-05,
      "loss": 0.3997,
      "step": 22207
    },
    {
      "epoch": 1.2435883077612275,
      "grad_norm": 1.4258272647857666,
      "learning_rate": 9.942157894736842e-05,
      "loss": 0.4808,
      "step": 22208
    },
    {
      "epoch": 1.2436443050733565,
      "grad_norm": 1.605284571647644,
      "learning_rate": 9.94213157894737e-05,
      "loss": 0.4504,
      "step": 22209
    },
    {
      "epoch": 1.2437003023854856,
      "grad_norm": 1.4095648527145386,
      "learning_rate": 9.942105263157895e-05,
      "loss": 0.3706,
      "step": 22210
    },
    {
      "epoch": 1.2437562996976146,
      "grad_norm": 1.2311546802520752,
      "learning_rate": 9.942078947368421e-05,
      "loss": 0.3636,
      "step": 22211
    },
    {
      "epoch": 1.2438122970097436,
      "grad_norm": 1.4106169939041138,
      "learning_rate": 9.942052631578947e-05,
      "loss": 0.5184,
      "step": 22212
    },
    {
      "epoch": 1.2438682943218726,
      "grad_norm": 1.5023138523101807,
      "learning_rate": 9.942026315789475e-05,
      "loss": 0.3936,
      "step": 22213
    },
    {
      "epoch": 1.2439242916340016,
      "grad_norm": 1.2082806825637817,
      "learning_rate": 9.942000000000001e-05,
      "loss": 0.5197,
      "step": 22214
    },
    {
      "epoch": 1.2439802889461307,
      "grad_norm": 1.411442756652832,
      "learning_rate": 9.941973684210527e-05,
      "loss": 0.4805,
      "step": 22215
    },
    {
      "epoch": 1.2440362862582597,
      "grad_norm": 1.3924330472946167,
      "learning_rate": 9.941947368421053e-05,
      "loss": 0.3279,
      "step": 22216
    },
    {
      "epoch": 1.2440922835703887,
      "grad_norm": 1.2783757448196411,
      "learning_rate": 9.94192105263158e-05,
      "loss": 0.4446,
      "step": 22217
    },
    {
      "epoch": 1.2441482808825177,
      "grad_norm": 1.6165719032287598,
      "learning_rate": 9.941894736842106e-05,
      "loss": 0.5575,
      "step": 22218
    },
    {
      "epoch": 1.2442042781946467,
      "grad_norm": 1.1490000486373901,
      "learning_rate": 9.941868421052632e-05,
      "loss": 0.4069,
      "step": 22219
    },
    {
      "epoch": 1.2442602755067758,
      "grad_norm": 1.1328785419464111,
      "learning_rate": 9.941842105263158e-05,
      "loss": 0.3499,
      "step": 22220
    },
    {
      "epoch": 1.2443162728189048,
      "grad_norm": 1.3411613702774048,
      "learning_rate": 9.941815789473685e-05,
      "loss": 0.4329,
      "step": 22221
    },
    {
      "epoch": 1.2443722701310338,
      "grad_norm": 1.3682115077972412,
      "learning_rate": 9.941789473684211e-05,
      "loss": 0.4656,
      "step": 22222
    },
    {
      "epoch": 1.2444282674431628,
      "grad_norm": 1.3170279264450073,
      "learning_rate": 9.941763157894737e-05,
      "loss": 0.5161,
      "step": 22223
    },
    {
      "epoch": 1.2444842647552918,
      "grad_norm": 1.417770504951477,
      "learning_rate": 9.941736842105263e-05,
      "loss": 0.3697,
      "step": 22224
    },
    {
      "epoch": 1.2445402620674209,
      "grad_norm": 1.6053637266159058,
      "learning_rate": 9.941710526315789e-05,
      "loss": 0.4575,
      "step": 22225
    },
    {
      "epoch": 1.2445962593795499,
      "grad_norm": 1.4814941883087158,
      "learning_rate": 9.941684210526316e-05,
      "loss": 0.3615,
      "step": 22226
    },
    {
      "epoch": 1.244652256691679,
      "grad_norm": 1.3396278619766235,
      "learning_rate": 9.941657894736842e-05,
      "loss": 0.4416,
      "step": 22227
    },
    {
      "epoch": 1.244708254003808,
      "grad_norm": 1.1928372383117676,
      "learning_rate": 9.94163157894737e-05,
      "loss": 0.4007,
      "step": 22228
    },
    {
      "epoch": 1.244764251315937,
      "grad_norm": 1.2904865741729736,
      "learning_rate": 9.941605263157894e-05,
      "loss": 0.7117,
      "step": 22229
    },
    {
      "epoch": 1.244820248628066,
      "grad_norm": 1.3465505838394165,
      "learning_rate": 9.941578947368422e-05,
      "loss": 0.4399,
      "step": 22230
    },
    {
      "epoch": 1.244876245940195,
      "grad_norm": 1.3889199495315552,
      "learning_rate": 9.941552631578948e-05,
      "loss": 0.4516,
      "step": 22231
    },
    {
      "epoch": 1.244932243252324,
      "grad_norm": 1.256568193435669,
      "learning_rate": 9.941526315789475e-05,
      "loss": 0.4759,
      "step": 22232
    },
    {
      "epoch": 1.244988240564453,
      "grad_norm": 1.5486410856246948,
      "learning_rate": 9.9415e-05,
      "loss": 0.4454,
      "step": 22233
    },
    {
      "epoch": 1.245044237876582,
      "grad_norm": 1.2451177835464478,
      "learning_rate": 9.941473684210527e-05,
      "loss": 0.3403,
      "step": 22234
    },
    {
      "epoch": 1.245100235188711,
      "grad_norm": 1.2428570985794067,
      "learning_rate": 9.941447368421053e-05,
      "loss": 0.377,
      "step": 22235
    },
    {
      "epoch": 1.24515623250084,
      "grad_norm": 1.336918830871582,
      "learning_rate": 9.94142105263158e-05,
      "loss": 0.4067,
      "step": 22236
    },
    {
      "epoch": 1.245212229812969,
      "grad_norm": 1.4773956537246704,
      "learning_rate": 9.941394736842106e-05,
      "loss": 0.4488,
      "step": 22237
    },
    {
      "epoch": 1.245268227125098,
      "grad_norm": 1.2210487127304077,
      "learning_rate": 9.941368421052632e-05,
      "loss": 0.4032,
      "step": 22238
    },
    {
      "epoch": 1.2453242244372271,
      "grad_norm": 1.3869651556015015,
      "learning_rate": 9.941342105263158e-05,
      "loss": 0.4829,
      "step": 22239
    },
    {
      "epoch": 1.2453802217493561,
      "grad_norm": 1.6309747695922852,
      "learning_rate": 9.941315789473684e-05,
      "loss": 0.4308,
      "step": 22240
    },
    {
      "epoch": 1.2454362190614852,
      "grad_norm": 1.2051997184753418,
      "learning_rate": 9.941289473684211e-05,
      "loss": 0.5023,
      "step": 22241
    },
    {
      "epoch": 1.2454922163736142,
      "grad_norm": 1.7952145338058472,
      "learning_rate": 9.941263157894737e-05,
      "loss": 0.3756,
      "step": 22242
    },
    {
      "epoch": 1.2455482136857432,
      "grad_norm": 1.40574049949646,
      "learning_rate": 9.941236842105263e-05,
      "loss": 0.4318,
      "step": 22243
    },
    {
      "epoch": 1.2456042109978722,
      "grad_norm": 1.3252264261245728,
      "learning_rate": 9.94121052631579e-05,
      "loss": 0.6048,
      "step": 22244
    },
    {
      "epoch": 1.2456602083100012,
      "grad_norm": 1.380267858505249,
      "learning_rate": 9.941184210526317e-05,
      "loss": 0.4548,
      "step": 22245
    },
    {
      "epoch": 1.2457162056221303,
      "grad_norm": 1.1676914691925049,
      "learning_rate": 9.941157894736843e-05,
      "loss": 0.4655,
      "step": 22246
    },
    {
      "epoch": 1.2457722029342593,
      "grad_norm": 1.3717881441116333,
      "learning_rate": 9.941131578947369e-05,
      "loss": 0.3673,
      "step": 22247
    },
    {
      "epoch": 1.2458282002463883,
      "grad_norm": 1.1353402137756348,
      "learning_rate": 9.941105263157895e-05,
      "loss": 0.355,
      "step": 22248
    },
    {
      "epoch": 1.2458841975585173,
      "grad_norm": 1.4019495248794556,
      "learning_rate": 9.941078947368422e-05,
      "loss": 0.3895,
      "step": 22249
    },
    {
      "epoch": 1.2459401948706463,
      "grad_norm": 1.2970120906829834,
      "learning_rate": 9.941052631578948e-05,
      "loss": 0.3942,
      "step": 22250
    },
    {
      "epoch": 1.2459961921827751,
      "grad_norm": 1.249554991722107,
      "learning_rate": 9.941026315789474e-05,
      "loss": 0.3587,
      "step": 22251
    },
    {
      "epoch": 1.2460521894949041,
      "grad_norm": 1.4984300136566162,
      "learning_rate": 9.941e-05,
      "loss": 0.5534,
      "step": 22252
    },
    {
      "epoch": 1.2461081868070332,
      "grad_norm": 1.3894104957580566,
      "learning_rate": 9.940973684210527e-05,
      "loss": 0.4597,
      "step": 22253
    },
    {
      "epoch": 1.2461641841191622,
      "grad_norm": 1.3068922758102417,
      "learning_rate": 9.940947368421053e-05,
      "loss": 0.3798,
      "step": 22254
    },
    {
      "epoch": 1.2462201814312912,
      "grad_norm": 1.3988882303237915,
      "learning_rate": 9.94092105263158e-05,
      "loss": 0.4954,
      "step": 22255
    },
    {
      "epoch": 1.2462761787434202,
      "grad_norm": 1.2204664945602417,
      "learning_rate": 9.940894736842105e-05,
      "loss": 0.4786,
      "step": 22256
    },
    {
      "epoch": 1.2463321760555492,
      "grad_norm": 1.2899380922317505,
      "learning_rate": 9.940868421052632e-05,
      "loss": 0.4025,
      "step": 22257
    },
    {
      "epoch": 1.2463881733676783,
      "grad_norm": 1.3493857383728027,
      "learning_rate": 9.940842105263158e-05,
      "loss": 0.5435,
      "step": 22258
    },
    {
      "epoch": 1.2464441706798073,
      "grad_norm": 1.234354853630066,
      "learning_rate": 9.940815789473684e-05,
      "loss": 0.3658,
      "step": 22259
    },
    {
      "epoch": 1.2465001679919363,
      "grad_norm": 1.7639458179473877,
      "learning_rate": 9.940789473684212e-05,
      "loss": 0.4752,
      "step": 22260
    },
    {
      "epoch": 1.2465561653040653,
      "grad_norm": 1.5952787399291992,
      "learning_rate": 9.940763157894736e-05,
      "loss": 0.508,
      "step": 22261
    },
    {
      "epoch": 1.2466121626161943,
      "grad_norm": 1.4668351411819458,
      "learning_rate": 9.940736842105264e-05,
      "loss": 0.4437,
      "step": 22262
    },
    {
      "epoch": 1.2466681599283234,
      "grad_norm": 1.3444815874099731,
      "learning_rate": 9.94071052631579e-05,
      "loss": 0.3824,
      "step": 22263
    },
    {
      "epoch": 1.2467241572404524,
      "grad_norm": 1.4767779111862183,
      "learning_rate": 9.940684210526317e-05,
      "loss": 0.4853,
      "step": 22264
    },
    {
      "epoch": 1.2467801545525814,
      "grad_norm": 1.5499085187911987,
      "learning_rate": 9.940657894736843e-05,
      "loss": 0.4013,
      "step": 22265
    },
    {
      "epoch": 1.2468361518647104,
      "grad_norm": 1.2688138484954834,
      "learning_rate": 9.940631578947369e-05,
      "loss": 0.4274,
      "step": 22266
    },
    {
      "epoch": 1.2468921491768394,
      "grad_norm": 1.3768786191940308,
      "learning_rate": 9.940605263157895e-05,
      "loss": 0.4536,
      "step": 22267
    },
    {
      "epoch": 1.2469481464889685,
      "grad_norm": 1.2985590696334839,
      "learning_rate": 9.940578947368422e-05,
      "loss": 0.4155,
      "step": 22268
    },
    {
      "epoch": 1.2470041438010975,
      "grad_norm": 1.2233731746673584,
      "learning_rate": 9.940552631578948e-05,
      "loss": 0.3946,
      "step": 22269
    },
    {
      "epoch": 1.2470601411132265,
      "grad_norm": 1.357564091682434,
      "learning_rate": 9.940526315789474e-05,
      "loss": 0.3937,
      "step": 22270
    },
    {
      "epoch": 1.2471161384253555,
      "grad_norm": 1.3891246318817139,
      "learning_rate": 9.9405e-05,
      "loss": 0.4024,
      "step": 22271
    },
    {
      "epoch": 1.2471721357374845,
      "grad_norm": 1.4539968967437744,
      "learning_rate": 9.940473684210527e-05,
      "loss": 0.4038,
      "step": 22272
    },
    {
      "epoch": 1.2472281330496136,
      "grad_norm": 1.7275545597076416,
      "learning_rate": 9.940447368421053e-05,
      "loss": 0.5758,
      "step": 22273
    },
    {
      "epoch": 1.2472841303617426,
      "grad_norm": 1.1803683042526245,
      "learning_rate": 9.94042105263158e-05,
      "loss": 0.4849,
      "step": 22274
    },
    {
      "epoch": 1.2473401276738716,
      "grad_norm": 1.2474204301834106,
      "learning_rate": 9.940394736842105e-05,
      "loss": 0.4421,
      "step": 22275
    },
    {
      "epoch": 1.2473961249860006,
      "grad_norm": 1.5943682193756104,
      "learning_rate": 9.940368421052631e-05,
      "loss": 0.5201,
      "step": 22276
    },
    {
      "epoch": 1.2474521222981296,
      "grad_norm": 1.2650312185287476,
      "learning_rate": 9.940342105263159e-05,
      "loss": 0.3833,
      "step": 22277
    },
    {
      "epoch": 1.2475081196102586,
      "grad_norm": 4.03690242767334,
      "learning_rate": 9.940315789473685e-05,
      "loss": 0.4693,
      "step": 22278
    },
    {
      "epoch": 1.2475641169223877,
      "grad_norm": 1.3714172840118408,
      "learning_rate": 9.94028947368421e-05,
      "loss": 0.5724,
      "step": 22279
    },
    {
      "epoch": 1.2476201142345167,
      "grad_norm": 1.2636523246765137,
      "learning_rate": 9.940263157894737e-05,
      "loss": 0.4871,
      "step": 22280
    },
    {
      "epoch": 1.2476761115466457,
      "grad_norm": 1.3796244859695435,
      "learning_rate": 9.940236842105264e-05,
      "loss": 0.4291,
      "step": 22281
    },
    {
      "epoch": 1.2477321088587747,
      "grad_norm": 1.2901058197021484,
      "learning_rate": 9.94021052631579e-05,
      "loss": 0.5145,
      "step": 22282
    },
    {
      "epoch": 1.2477881061709037,
      "grad_norm": 1.2571921348571777,
      "learning_rate": 9.940184210526317e-05,
      "loss": 0.3346,
      "step": 22283
    },
    {
      "epoch": 1.2478441034830328,
      "grad_norm": 1.4464619159698486,
      "learning_rate": 9.940157894736842e-05,
      "loss": 0.5793,
      "step": 22284
    },
    {
      "epoch": 1.2479001007951618,
      "grad_norm": 1.5002973079681396,
      "learning_rate": 9.940131578947369e-05,
      "loss": 0.4873,
      "step": 22285
    },
    {
      "epoch": 1.2479560981072908,
      "grad_norm": 1.4570118188858032,
      "learning_rate": 9.940105263157895e-05,
      "loss": 0.3904,
      "step": 22286
    },
    {
      "epoch": 1.2480120954194198,
      "grad_norm": 1.5554147958755493,
      "learning_rate": 9.940078947368422e-05,
      "loss": 0.4778,
      "step": 22287
    },
    {
      "epoch": 1.2480680927315488,
      "grad_norm": 8.259346961975098,
      "learning_rate": 9.940052631578947e-05,
      "loss": 0.4362,
      "step": 22288
    },
    {
      "epoch": 1.2481240900436779,
      "grad_norm": 1.4304662942886353,
      "learning_rate": 9.940026315789474e-05,
      "loss": 0.5186,
      "step": 22289
    },
    {
      "epoch": 1.2481800873558069,
      "grad_norm": 1.2556445598602295,
      "learning_rate": 9.94e-05,
      "loss": 0.3436,
      "step": 22290
    },
    {
      "epoch": 1.248236084667936,
      "grad_norm": 1.1076174974441528,
      "learning_rate": 9.939973684210528e-05,
      "loss": 0.3614,
      "step": 22291
    },
    {
      "epoch": 1.248292081980065,
      "grad_norm": 1.23257315158844,
      "learning_rate": 9.939947368421054e-05,
      "loss": 0.4392,
      "step": 22292
    },
    {
      "epoch": 1.248348079292194,
      "grad_norm": 1.283825397491455,
      "learning_rate": 9.939921052631578e-05,
      "loss": 0.4014,
      "step": 22293
    },
    {
      "epoch": 1.248404076604323,
      "grad_norm": 1.176995038986206,
      "learning_rate": 9.939894736842106e-05,
      "loss": 0.3408,
      "step": 22294
    },
    {
      "epoch": 1.248460073916452,
      "grad_norm": 1.4385688304901123,
      "learning_rate": 9.939868421052632e-05,
      "loss": 0.3907,
      "step": 22295
    },
    {
      "epoch": 1.248516071228581,
      "grad_norm": 1.8741458654403687,
      "learning_rate": 9.939842105263159e-05,
      "loss": 0.8512,
      "step": 22296
    },
    {
      "epoch": 1.24857206854071,
      "grad_norm": 1.3005434274673462,
      "learning_rate": 9.939815789473685e-05,
      "loss": 0.5882,
      "step": 22297
    },
    {
      "epoch": 1.248628065852839,
      "grad_norm": 1.5760449171066284,
      "learning_rate": 9.939789473684211e-05,
      "loss": 0.3411,
      "step": 22298
    },
    {
      "epoch": 1.248684063164968,
      "grad_norm": 1.1352293491363525,
      "learning_rate": 9.939763157894737e-05,
      "loss": 0.4053,
      "step": 22299
    },
    {
      "epoch": 1.248740060477097,
      "grad_norm": 1.546398639678955,
      "learning_rate": 9.939736842105264e-05,
      "loss": 0.4613,
      "step": 22300
    },
    {
      "epoch": 1.248796057789226,
      "grad_norm": 1.702652096748352,
      "learning_rate": 9.93971052631579e-05,
      "loss": 0.5096,
      "step": 22301
    },
    {
      "epoch": 1.248852055101355,
      "grad_norm": 1.2364565134048462,
      "learning_rate": 9.939684210526316e-05,
      "loss": 0.3482,
      "step": 22302
    },
    {
      "epoch": 1.2489080524134841,
      "grad_norm": 1.5198919773101807,
      "learning_rate": 9.939657894736842e-05,
      "loss": 0.585,
      "step": 22303
    },
    {
      "epoch": 1.2489640497256131,
      "grad_norm": 1.2503507137298584,
      "learning_rate": 9.93963157894737e-05,
      "loss": 0.3867,
      "step": 22304
    },
    {
      "epoch": 1.2490200470377422,
      "grad_norm": 1.4735126495361328,
      "learning_rate": 9.939605263157895e-05,
      "loss": 0.4261,
      "step": 22305
    },
    {
      "epoch": 1.2490760443498712,
      "grad_norm": 1.3438857793807983,
      "learning_rate": 9.939578947368421e-05,
      "loss": 0.4776,
      "step": 22306
    },
    {
      "epoch": 1.2491320416620002,
      "grad_norm": 1.3483028411865234,
      "learning_rate": 9.939552631578947e-05,
      "loss": 0.5217,
      "step": 22307
    },
    {
      "epoch": 1.2491880389741292,
      "grad_norm": 1.2666211128234863,
      "learning_rate": 9.939526315789475e-05,
      "loss": 0.4004,
      "step": 22308
    },
    {
      "epoch": 1.2492440362862582,
      "grad_norm": 1.336466670036316,
      "learning_rate": 9.9395e-05,
      "loss": 0.4599,
      "step": 22309
    },
    {
      "epoch": 1.2493000335983873,
      "grad_norm": 1.5537548065185547,
      "learning_rate": 9.939473684210527e-05,
      "loss": 0.6118,
      "step": 22310
    },
    {
      "epoch": 1.2493560309105163,
      "grad_norm": 1.4111347198486328,
      "learning_rate": 9.939447368421053e-05,
      "loss": 0.4152,
      "step": 22311
    },
    {
      "epoch": 1.2494120282226453,
      "grad_norm": 1.4069594144821167,
      "learning_rate": 9.939421052631579e-05,
      "loss": 0.4386,
      "step": 22312
    },
    {
      "epoch": 1.2494680255347743,
      "grad_norm": 1.263898253440857,
      "learning_rate": 9.939394736842106e-05,
      "loss": 0.4336,
      "step": 22313
    },
    {
      "epoch": 1.2495240228469033,
      "grad_norm": 1.617743730545044,
      "learning_rate": 9.939368421052632e-05,
      "loss": 0.6002,
      "step": 22314
    },
    {
      "epoch": 1.2495800201590324,
      "grad_norm": 1.2375640869140625,
      "learning_rate": 9.939342105263159e-05,
      "loss": 0.4006,
      "step": 22315
    },
    {
      "epoch": 1.2496360174711614,
      "grad_norm": 1.4769809246063232,
      "learning_rate": 9.939315789473684e-05,
      "loss": 0.4202,
      "step": 22316
    },
    {
      "epoch": 1.2496920147832904,
      "grad_norm": 1.7477962970733643,
      "learning_rate": 9.939289473684211e-05,
      "loss": 0.5736,
      "step": 22317
    },
    {
      "epoch": 1.2497480120954194,
      "grad_norm": 1.5742456912994385,
      "learning_rate": 9.939263157894737e-05,
      "loss": 0.4195,
      "step": 22318
    },
    {
      "epoch": 1.2498040094075484,
      "grad_norm": 1.5923107862472534,
      "learning_rate": 9.939236842105264e-05,
      "loss": 0.4855,
      "step": 22319
    },
    {
      "epoch": 1.2498600067196775,
      "grad_norm": 1.0994890928268433,
      "learning_rate": 9.93921052631579e-05,
      "loss": 0.2878,
      "step": 22320
    },
    {
      "epoch": 1.2499160040318065,
      "grad_norm": 1.2776751518249512,
      "learning_rate": 9.939184210526316e-05,
      "loss": 0.5888,
      "step": 22321
    },
    {
      "epoch": 1.2499720013439355,
      "grad_norm": 1.275792121887207,
      "learning_rate": 9.939157894736842e-05,
      "loss": 0.4022,
      "step": 22322
    },
    {
      "epoch": 1.2500279986560645,
      "grad_norm": 1.2924615144729614,
      "learning_rate": 9.93913157894737e-05,
      "loss": 0.4109,
      "step": 22323
    },
    {
      "epoch": 1.2500839959681935,
      "grad_norm": 1.181273341178894,
      "learning_rate": 9.939105263157896e-05,
      "loss": 0.4517,
      "step": 22324
    },
    {
      "epoch": 1.2501399932803225,
      "grad_norm": 1.4977375268936157,
      "learning_rate": 9.939078947368422e-05,
      "loss": 0.4974,
      "step": 22325
    },
    {
      "epoch": 1.2501959905924516,
      "grad_norm": 1.384088158607483,
      "learning_rate": 9.939052631578948e-05,
      "loss": 0.3408,
      "step": 22326
    },
    {
      "epoch": 1.2502519879045806,
      "grad_norm": 1.5096538066864014,
      "learning_rate": 9.939026315789474e-05,
      "loss": 0.5548,
      "step": 22327
    },
    {
      "epoch": 1.2503079852167096,
      "grad_norm": 1.2352845668792725,
      "learning_rate": 9.939000000000001e-05,
      "loss": 0.3673,
      "step": 22328
    },
    {
      "epoch": 1.2503639825288386,
      "grad_norm": 1.3208589553833008,
      "learning_rate": 9.938973684210527e-05,
      "loss": 0.6064,
      "step": 22329
    },
    {
      "epoch": 1.2504199798409676,
      "grad_norm": 2.8895554542541504,
      "learning_rate": 9.938947368421053e-05,
      "loss": 0.4623,
      "step": 22330
    },
    {
      "epoch": 1.2504759771530967,
      "grad_norm": 1.7574106454849243,
      "learning_rate": 9.938921052631579e-05,
      "loss": 0.5314,
      "step": 22331
    },
    {
      "epoch": 1.2505319744652257,
      "grad_norm": 1.6027345657348633,
      "learning_rate": 9.938894736842106e-05,
      "loss": 0.442,
      "step": 22332
    },
    {
      "epoch": 1.2505879717773547,
      "grad_norm": 2.04962158203125,
      "learning_rate": 9.938868421052632e-05,
      "loss": 0.5385,
      "step": 22333
    },
    {
      "epoch": 1.2506439690894837,
      "grad_norm": 1.3174147605895996,
      "learning_rate": 9.938842105263158e-05,
      "loss": 0.569,
      "step": 22334
    },
    {
      "epoch": 1.2506999664016127,
      "grad_norm": 1.7815295457839966,
      "learning_rate": 9.938815789473684e-05,
      "loss": 0.5231,
      "step": 22335
    },
    {
      "epoch": 1.2507559637137418,
      "grad_norm": 1.4336323738098145,
      "learning_rate": 9.938789473684211e-05,
      "loss": 0.6225,
      "step": 22336
    },
    {
      "epoch": 1.2508119610258708,
      "grad_norm": 1.342552661895752,
      "learning_rate": 9.938763157894737e-05,
      "loss": 0.4109,
      "step": 22337
    },
    {
      "epoch": 1.2508679583379998,
      "grad_norm": 1.2027740478515625,
      "learning_rate": 9.938736842105265e-05,
      "loss": 0.3825,
      "step": 22338
    },
    {
      "epoch": 1.2509239556501288,
      "grad_norm": 1.3601100444793701,
      "learning_rate": 9.938710526315789e-05,
      "loss": 0.448,
      "step": 22339
    },
    {
      "epoch": 1.2509799529622578,
      "grad_norm": 1.4747097492218018,
      "learning_rate": 9.938684210526317e-05,
      "loss": 0.4084,
      "step": 22340
    },
    {
      "epoch": 1.2510359502743869,
      "grad_norm": 1.3954651355743408,
      "learning_rate": 9.938657894736843e-05,
      "loss": 0.4549,
      "step": 22341
    },
    {
      "epoch": 1.2510919475865159,
      "grad_norm": 1.2875089645385742,
      "learning_rate": 9.93863157894737e-05,
      "loss": 0.3957,
      "step": 22342
    },
    {
      "epoch": 1.251147944898645,
      "grad_norm": 1.372676968574524,
      "learning_rate": 9.938605263157895e-05,
      "loss": 0.4569,
      "step": 22343
    },
    {
      "epoch": 1.251203942210774,
      "grad_norm": 1.1782981157302856,
      "learning_rate": 9.93857894736842e-05,
      "loss": 0.4145,
      "step": 22344
    },
    {
      "epoch": 1.251259939522903,
      "grad_norm": 1.5319457054138184,
      "learning_rate": 9.938552631578948e-05,
      "loss": 0.5579,
      "step": 22345
    },
    {
      "epoch": 1.251315936835032,
      "grad_norm": 1.3733348846435547,
      "learning_rate": 9.938526315789474e-05,
      "loss": 0.4205,
      "step": 22346
    },
    {
      "epoch": 1.251371934147161,
      "grad_norm": 1.3196403980255127,
      "learning_rate": 9.938500000000001e-05,
      "loss": 0.3781,
      "step": 22347
    },
    {
      "epoch": 1.25142793145929,
      "grad_norm": 1.1703699827194214,
      "learning_rate": 9.938473684210526e-05,
      "loss": 0.3795,
      "step": 22348
    },
    {
      "epoch": 1.251483928771419,
      "grad_norm": 1.5440356731414795,
      "learning_rate": 9.938447368421053e-05,
      "loss": 0.4559,
      "step": 22349
    },
    {
      "epoch": 1.251539926083548,
      "grad_norm": 1.190883755683899,
      "learning_rate": 9.938421052631579e-05,
      "loss": 0.3914,
      "step": 22350
    },
    {
      "epoch": 1.251595923395677,
      "grad_norm": 1.065455436706543,
      "learning_rate": 9.938394736842106e-05,
      "loss": 0.3576,
      "step": 22351
    },
    {
      "epoch": 1.251651920707806,
      "grad_norm": 1.5451136827468872,
      "learning_rate": 9.938368421052632e-05,
      "loss": 0.5983,
      "step": 22352
    },
    {
      "epoch": 1.251707918019935,
      "grad_norm": 1.3333629369735718,
      "learning_rate": 9.938342105263158e-05,
      "loss": 0.4733,
      "step": 22353
    },
    {
      "epoch": 1.251763915332064,
      "grad_norm": 1.1256550550460815,
      "learning_rate": 9.938315789473684e-05,
      "loss": 0.3686,
      "step": 22354
    },
    {
      "epoch": 1.2518199126441931,
      "grad_norm": 1.2385841608047485,
      "learning_rate": 9.938289473684212e-05,
      "loss": 0.4191,
      "step": 22355
    },
    {
      "epoch": 1.2518759099563221,
      "grad_norm": 1.2452980279922485,
      "learning_rate": 9.938263157894738e-05,
      "loss": 0.4756,
      "step": 22356
    },
    {
      "epoch": 1.2519319072684512,
      "grad_norm": 1.3275479078292847,
      "learning_rate": 9.938236842105264e-05,
      "loss": 0.3779,
      "step": 22357
    },
    {
      "epoch": 1.2519879045805802,
      "grad_norm": 1.409300446510315,
      "learning_rate": 9.93821052631579e-05,
      "loss": 0.3782,
      "step": 22358
    },
    {
      "epoch": 1.2520439018927092,
      "grad_norm": 1.6561707258224487,
      "learning_rate": 9.938184210526317e-05,
      "loss": 0.5491,
      "step": 22359
    },
    {
      "epoch": 1.2520998992048382,
      "grad_norm": 1.3989508152008057,
      "learning_rate": 9.938157894736843e-05,
      "loss": 0.4917,
      "step": 22360
    },
    {
      "epoch": 1.2521558965169672,
      "grad_norm": 1.2363028526306152,
      "learning_rate": 9.938131578947369e-05,
      "loss": 0.4661,
      "step": 22361
    },
    {
      "epoch": 1.2522118938290963,
      "grad_norm": 1.3518849611282349,
      "learning_rate": 9.938105263157895e-05,
      "loss": 0.4335,
      "step": 22362
    },
    {
      "epoch": 1.2522678911412253,
      "grad_norm": 1.5282328128814697,
      "learning_rate": 9.938078947368421e-05,
      "loss": 0.5125,
      "step": 22363
    },
    {
      "epoch": 1.2523238884533543,
      "grad_norm": 1.4124184846878052,
      "learning_rate": 9.938052631578948e-05,
      "loss": 0.5615,
      "step": 22364
    },
    {
      "epoch": 1.2523798857654833,
      "grad_norm": 1.2635563611984253,
      "learning_rate": 9.938026315789474e-05,
      "loss": 0.403,
      "step": 22365
    },
    {
      "epoch": 1.2524358830776123,
      "grad_norm": 1.281112551689148,
      "learning_rate": 9.938e-05,
      "loss": 0.3814,
      "step": 22366
    },
    {
      "epoch": 1.2524918803897414,
      "grad_norm": 1.242078423500061,
      "learning_rate": 9.937973684210526e-05,
      "loss": 0.4,
      "step": 22367
    },
    {
      "epoch": 1.2525478777018704,
      "grad_norm": 1.2379521131515503,
      "learning_rate": 9.937947368421053e-05,
      "loss": 0.4875,
      "step": 22368
    },
    {
      "epoch": 1.2526038750139994,
      "grad_norm": 1.6423168182373047,
      "learning_rate": 9.93792105263158e-05,
      "loss": 0.4695,
      "step": 22369
    },
    {
      "epoch": 1.2526598723261284,
      "grad_norm": 1.422355055809021,
      "learning_rate": 9.937894736842107e-05,
      "loss": 0.5009,
      "step": 22370
    },
    {
      "epoch": 1.2527158696382574,
      "grad_norm": 1.112548828125,
      "learning_rate": 9.937868421052631e-05,
      "loss": 0.3949,
      "step": 22371
    },
    {
      "epoch": 1.2527718669503864,
      "grad_norm": 1.3380110263824463,
      "learning_rate": 9.937842105263159e-05,
      "loss": 0.458,
      "step": 22372
    },
    {
      "epoch": 1.2528278642625155,
      "grad_norm": 1.386528730392456,
      "learning_rate": 9.937815789473685e-05,
      "loss": 0.4417,
      "step": 22373
    },
    {
      "epoch": 1.2528838615746445,
      "grad_norm": 1.596781849861145,
      "learning_rate": 9.937789473684212e-05,
      "loss": 0.6139,
      "step": 22374
    },
    {
      "epoch": 1.2529398588867735,
      "grad_norm": 1.2813609838485718,
      "learning_rate": 9.937763157894738e-05,
      "loss": 0.41,
      "step": 22375
    },
    {
      "epoch": 1.2529958561989025,
      "grad_norm": 2.168900728225708,
      "learning_rate": 9.937736842105264e-05,
      "loss": 0.6009,
      "step": 22376
    },
    {
      "epoch": 1.2530518535110315,
      "grad_norm": 1.4229382276535034,
      "learning_rate": 9.93771052631579e-05,
      "loss": 0.4844,
      "step": 22377
    },
    {
      "epoch": 1.2531078508231606,
      "grad_norm": 1.288794755935669,
      "learning_rate": 9.937684210526317e-05,
      "loss": 0.4613,
      "step": 22378
    },
    {
      "epoch": 1.2531638481352896,
      "grad_norm": 1.322526216506958,
      "learning_rate": 9.937657894736843e-05,
      "loss": 0.4081,
      "step": 22379
    },
    {
      "epoch": 1.2532198454474186,
      "grad_norm": 1.375086784362793,
      "learning_rate": 9.937631578947368e-05,
      "loss": 0.3796,
      "step": 22380
    },
    {
      "epoch": 1.2532758427595476,
      "grad_norm": 1.1509244441986084,
      "learning_rate": 9.937605263157895e-05,
      "loss": 0.4033,
      "step": 22381
    },
    {
      "epoch": 1.2533318400716766,
      "grad_norm": 1.2860116958618164,
      "learning_rate": 9.937578947368421e-05,
      "loss": 0.462,
      "step": 22382
    },
    {
      "epoch": 1.2533878373838057,
      "grad_norm": 1.4602913856506348,
      "learning_rate": 9.937552631578948e-05,
      "loss": 0.4564,
      "step": 22383
    },
    {
      "epoch": 1.2534438346959347,
      "grad_norm": 1.9868360757827759,
      "learning_rate": 9.937526315789474e-05,
      "loss": 0.4496,
      "step": 22384
    },
    {
      "epoch": 1.2534998320080635,
      "grad_norm": 1.5726895332336426,
      "learning_rate": 9.9375e-05,
      "loss": 0.4239,
      "step": 22385
    },
    {
      "epoch": 1.2535558293201925,
      "grad_norm": 2.0822389125823975,
      "learning_rate": 9.937473684210526e-05,
      "loss": 0.4286,
      "step": 22386
    },
    {
      "epoch": 1.2536118266323215,
      "grad_norm": 1.3133800029754639,
      "learning_rate": 9.937447368421054e-05,
      "loss": 0.5122,
      "step": 22387
    },
    {
      "epoch": 1.2536678239444505,
      "grad_norm": 1.394006371498108,
      "learning_rate": 9.93742105263158e-05,
      "loss": 0.4558,
      "step": 22388
    },
    {
      "epoch": 1.2537238212565796,
      "grad_norm": 1.356560230255127,
      "learning_rate": 9.937394736842106e-05,
      "loss": 0.424,
      "step": 22389
    },
    {
      "epoch": 1.2537798185687086,
      "grad_norm": 1.328401803970337,
      "learning_rate": 9.937368421052632e-05,
      "loss": 0.5114,
      "step": 22390
    },
    {
      "epoch": 1.2538358158808376,
      "grad_norm": 1.2721163034439087,
      "learning_rate": 9.937342105263159e-05,
      "loss": 0.4165,
      "step": 22391
    },
    {
      "epoch": 1.2538918131929666,
      "grad_norm": 3.9038705825805664,
      "learning_rate": 9.937315789473685e-05,
      "loss": 0.3836,
      "step": 22392
    },
    {
      "epoch": 1.2539478105050956,
      "grad_norm": 1.5038775205612183,
      "learning_rate": 9.937289473684211e-05,
      "loss": 0.6409,
      "step": 22393
    },
    {
      "epoch": 1.2540038078172246,
      "grad_norm": 1.7684061527252197,
      "learning_rate": 9.937263157894737e-05,
      "loss": 0.4112,
      "step": 22394
    },
    {
      "epoch": 1.2540598051293537,
      "grad_norm": 1.331947922706604,
      "learning_rate": 9.937236842105264e-05,
      "loss": 0.4065,
      "step": 22395
    },
    {
      "epoch": 1.2541158024414827,
      "grad_norm": 1.4933711290359497,
      "learning_rate": 9.93721052631579e-05,
      "loss": 0.384,
      "step": 22396
    },
    {
      "epoch": 1.2541717997536117,
      "grad_norm": 1.411307454109192,
      "learning_rate": 9.937184210526316e-05,
      "loss": 0.4413,
      "step": 22397
    },
    {
      "epoch": 1.2542277970657407,
      "grad_norm": 1.5018677711486816,
      "learning_rate": 9.937157894736842e-05,
      "loss": 0.4437,
      "step": 22398
    },
    {
      "epoch": 1.2542837943778697,
      "grad_norm": 1.2441569566726685,
      "learning_rate": 9.937131578947368e-05,
      "loss": 0.404,
      "step": 22399
    },
    {
      "epoch": 1.2543397916899988,
      "grad_norm": 1.35121488571167,
      "learning_rate": 9.937105263157895e-05,
      "loss": 0.4505,
      "step": 22400
    },
    {
      "epoch": 1.2543957890021278,
      "grad_norm": 1.5654253959655762,
      "learning_rate": 9.937078947368421e-05,
      "loss": 0.4104,
      "step": 22401
    },
    {
      "epoch": 1.2544517863142568,
      "grad_norm": 1.331819772720337,
      "learning_rate": 9.937052631578949e-05,
      "loss": 0.3457,
      "step": 22402
    },
    {
      "epoch": 1.2545077836263858,
      "grad_norm": 1.3496921062469482,
      "learning_rate": 9.937026315789473e-05,
      "loss": 0.4175,
      "step": 22403
    },
    {
      "epoch": 1.2545637809385148,
      "grad_norm": 1.254142165184021,
      "learning_rate": 9.937e-05,
      "loss": 0.5591,
      "step": 22404
    },
    {
      "epoch": 1.2546197782506439,
      "grad_norm": 1.2177461385726929,
      "learning_rate": 9.936973684210527e-05,
      "loss": 0.3637,
      "step": 22405
    },
    {
      "epoch": 1.2546757755627729,
      "grad_norm": 1.4695017337799072,
      "learning_rate": 9.936947368421054e-05,
      "loss": 0.4761,
      "step": 22406
    },
    {
      "epoch": 1.254731772874902,
      "grad_norm": 1.5132089853286743,
      "learning_rate": 9.93692105263158e-05,
      "loss": 0.634,
      "step": 22407
    },
    {
      "epoch": 1.254787770187031,
      "grad_norm": 1.3467780351638794,
      "learning_rate": 9.936894736842106e-05,
      "loss": 0.6429,
      "step": 22408
    },
    {
      "epoch": 1.25484376749916,
      "grad_norm": 1.2279789447784424,
      "learning_rate": 9.936868421052632e-05,
      "loss": 0.4997,
      "step": 22409
    },
    {
      "epoch": 1.254899764811289,
      "grad_norm": 1.2986432313919067,
      "learning_rate": 9.936842105263159e-05,
      "loss": 0.4413,
      "step": 22410
    },
    {
      "epoch": 1.254955762123418,
      "grad_norm": 1.3174965381622314,
      "learning_rate": 9.936815789473685e-05,
      "loss": 0.3556,
      "step": 22411
    },
    {
      "epoch": 1.255011759435547,
      "grad_norm": 1.1024069786071777,
      "learning_rate": 9.936789473684211e-05,
      "loss": 0.3554,
      "step": 22412
    },
    {
      "epoch": 1.255067756747676,
      "grad_norm": 13.55655574798584,
      "learning_rate": 9.936763157894737e-05,
      "loss": 0.4215,
      "step": 22413
    },
    {
      "epoch": 1.255123754059805,
      "grad_norm": 1.4097927808761597,
      "learning_rate": 9.936736842105263e-05,
      "loss": 0.441,
      "step": 22414
    },
    {
      "epoch": 1.255179751371934,
      "grad_norm": 1.3753725290298462,
      "learning_rate": 9.93671052631579e-05,
      "loss": 0.5994,
      "step": 22415
    },
    {
      "epoch": 1.255235748684063,
      "grad_norm": 1.4475756883621216,
      "learning_rate": 9.936684210526316e-05,
      "loss": 0.3789,
      "step": 22416
    },
    {
      "epoch": 1.255291745996192,
      "grad_norm": 2.690183401107788,
      "learning_rate": 9.936657894736842e-05,
      "loss": 0.445,
      "step": 22417
    },
    {
      "epoch": 1.255347743308321,
      "grad_norm": 1.5758112668991089,
      "learning_rate": 9.936631578947368e-05,
      "loss": 0.4321,
      "step": 22418
    },
    {
      "epoch": 1.2554037406204501,
      "grad_norm": 1.377387285232544,
      "learning_rate": 9.936605263157896e-05,
      "loss": 0.5935,
      "step": 22419
    },
    {
      "epoch": 1.2554597379325791,
      "grad_norm": 1.4579378366470337,
      "learning_rate": 9.936578947368422e-05,
      "loss": 0.3929,
      "step": 22420
    },
    {
      "epoch": 1.2555157352447082,
      "grad_norm": 1.174188494682312,
      "learning_rate": 9.936552631578948e-05,
      "loss": 0.3638,
      "step": 22421
    },
    {
      "epoch": 1.2555717325568372,
      "grad_norm": 1.19265615940094,
      "learning_rate": 9.936526315789474e-05,
      "loss": 0.3975,
      "step": 22422
    },
    {
      "epoch": 1.2556277298689662,
      "grad_norm": 1.4435083866119385,
      "learning_rate": 9.936500000000001e-05,
      "loss": 0.4415,
      "step": 22423
    },
    {
      "epoch": 1.2556837271810952,
      "grad_norm": 1.4970040321350098,
      "learning_rate": 9.936473684210527e-05,
      "loss": 0.5084,
      "step": 22424
    },
    {
      "epoch": 1.2557397244932242,
      "grad_norm": 1.7829281091690063,
      "learning_rate": 9.936447368421054e-05,
      "loss": 0.4626,
      "step": 22425
    },
    {
      "epoch": 1.2557957218053533,
      "grad_norm": 1.4064826965332031,
      "learning_rate": 9.936421052631579e-05,
      "loss": 0.4295,
      "step": 22426
    },
    {
      "epoch": 1.2558517191174823,
      "grad_norm": 1.2395046949386597,
      "learning_rate": 9.936394736842106e-05,
      "loss": 0.5363,
      "step": 22427
    },
    {
      "epoch": 1.2559077164296113,
      "grad_norm": 1.4189103841781616,
      "learning_rate": 9.936368421052632e-05,
      "loss": 0.346,
      "step": 22428
    },
    {
      "epoch": 1.2559637137417403,
      "grad_norm": 1.219213843345642,
      "learning_rate": 9.93634210526316e-05,
      "loss": 0.4594,
      "step": 22429
    },
    {
      "epoch": 1.2560197110538693,
      "grad_norm": 1.280315637588501,
      "learning_rate": 9.936315789473685e-05,
      "loss": 0.3594,
      "step": 22430
    },
    {
      "epoch": 1.2560757083659984,
      "grad_norm": 1.197350025177002,
      "learning_rate": 9.93628947368421e-05,
      "loss": 0.4468,
      "step": 22431
    },
    {
      "epoch": 1.2561317056781274,
      "grad_norm": 1.1670315265655518,
      "learning_rate": 9.936263157894737e-05,
      "loss": 0.3612,
      "step": 22432
    },
    {
      "epoch": 1.2561877029902564,
      "grad_norm": 1.382169485092163,
      "learning_rate": 9.936236842105263e-05,
      "loss": 0.4715,
      "step": 22433
    },
    {
      "epoch": 1.2562437003023854,
      "grad_norm": 1.1867884397506714,
      "learning_rate": 9.93621052631579e-05,
      "loss": 0.3865,
      "step": 22434
    },
    {
      "epoch": 1.2562996976145144,
      "grad_norm": 1.3531599044799805,
      "learning_rate": 9.936184210526315e-05,
      "loss": 0.4135,
      "step": 22435
    },
    {
      "epoch": 1.2563556949266435,
      "grad_norm": 1.5227432250976562,
      "learning_rate": 9.936157894736843e-05,
      "loss": 0.4965,
      "step": 22436
    },
    {
      "epoch": 1.2564116922387725,
      "grad_norm": 1.5554322004318237,
      "learning_rate": 9.936131578947369e-05,
      "loss": 0.63,
      "step": 22437
    },
    {
      "epoch": 1.2564676895509015,
      "grad_norm": 1.5383269786834717,
      "learning_rate": 9.936105263157896e-05,
      "loss": 0.5737,
      "step": 22438
    },
    {
      "epoch": 1.2565236868630305,
      "grad_norm": 1.520043134689331,
      "learning_rate": 9.936078947368422e-05,
      "loss": 0.5204,
      "step": 22439
    },
    {
      "epoch": 1.2565796841751595,
      "grad_norm": 1.1973000764846802,
      "learning_rate": 9.936052631578948e-05,
      "loss": 0.3516,
      "step": 22440
    },
    {
      "epoch": 1.2566356814872885,
      "grad_norm": 1.0638693571090698,
      "learning_rate": 9.936026315789474e-05,
      "loss": 0.3651,
      "step": 22441
    },
    {
      "epoch": 1.2566916787994176,
      "grad_norm": 3.457150936126709,
      "learning_rate": 9.936000000000001e-05,
      "loss": 0.4624,
      "step": 22442
    },
    {
      "epoch": 1.2567476761115466,
      "grad_norm": 1.3856191635131836,
      "learning_rate": 9.935973684210527e-05,
      "loss": 0.4131,
      "step": 22443
    },
    {
      "epoch": 1.2568036734236756,
      "grad_norm": 1.7095839977264404,
      "learning_rate": 9.935947368421053e-05,
      "loss": 0.4786,
      "step": 22444
    },
    {
      "epoch": 1.2568596707358046,
      "grad_norm": 1.5006550550460815,
      "learning_rate": 9.935921052631579e-05,
      "loss": 0.4271,
      "step": 22445
    },
    {
      "epoch": 1.2569156680479336,
      "grad_norm": 1.334295392036438,
      "learning_rate": 9.935894736842106e-05,
      "loss": 0.4931,
      "step": 22446
    },
    {
      "epoch": 1.2569716653600627,
      "grad_norm": 1.4089369773864746,
      "learning_rate": 9.935868421052632e-05,
      "loss": 0.5179,
      "step": 22447
    },
    {
      "epoch": 1.2570276626721917,
      "grad_norm": 1.551820993423462,
      "learning_rate": 9.935842105263158e-05,
      "loss": 0.6057,
      "step": 22448
    },
    {
      "epoch": 1.2570836599843207,
      "grad_norm": 1.658945083618164,
      "learning_rate": 9.935815789473684e-05,
      "loss": 0.3865,
      "step": 22449
    },
    {
      "epoch": 1.2571396572964497,
      "grad_norm": 1.4844506978988647,
      "learning_rate": 9.93578947368421e-05,
      "loss": 0.4183,
      "step": 22450
    },
    {
      "epoch": 1.2571956546085787,
      "grad_norm": 1.2960132360458374,
      "learning_rate": 9.935763157894738e-05,
      "loss": 0.4867,
      "step": 22451
    },
    {
      "epoch": 1.2572516519207078,
      "grad_norm": 1.2157917022705078,
      "learning_rate": 9.935736842105264e-05,
      "loss": 0.4369,
      "step": 22452
    },
    {
      "epoch": 1.2573076492328368,
      "grad_norm": 1.3133894205093384,
      "learning_rate": 9.93571052631579e-05,
      "loss": 0.3787,
      "step": 22453
    },
    {
      "epoch": 1.2573636465449658,
      "grad_norm": 1.1887586116790771,
      "learning_rate": 9.935684210526315e-05,
      "loss": 0.3378,
      "step": 22454
    },
    {
      "epoch": 1.2574196438570948,
      "grad_norm": 1.3167790174484253,
      "learning_rate": 9.935657894736843e-05,
      "loss": 0.3459,
      "step": 22455
    },
    {
      "epoch": 1.2574756411692238,
      "grad_norm": 1.3035658597946167,
      "learning_rate": 9.935631578947369e-05,
      "loss": 0.4213,
      "step": 22456
    },
    {
      "epoch": 1.2575316384813529,
      "grad_norm": 1.229737401008606,
      "learning_rate": 9.935605263157896e-05,
      "loss": 0.2927,
      "step": 22457
    },
    {
      "epoch": 1.2575876357934819,
      "grad_norm": 1.2345038652420044,
      "learning_rate": 9.935578947368421e-05,
      "loss": 0.4091,
      "step": 22458
    },
    {
      "epoch": 1.257643633105611,
      "grad_norm": 1.5438814163208008,
      "learning_rate": 9.935552631578948e-05,
      "loss": 0.5458,
      "step": 22459
    },
    {
      "epoch": 1.25769963041774,
      "grad_norm": 1.3399029970169067,
      "learning_rate": 9.935526315789474e-05,
      "loss": 0.4387,
      "step": 22460
    },
    {
      "epoch": 1.257755627729869,
      "grad_norm": 1.5059328079223633,
      "learning_rate": 9.935500000000001e-05,
      "loss": 0.5205,
      "step": 22461
    },
    {
      "epoch": 1.257811625041998,
      "grad_norm": 1.3784549236297607,
      "learning_rate": 9.935473684210527e-05,
      "loss": 0.4553,
      "step": 22462
    },
    {
      "epoch": 1.257867622354127,
      "grad_norm": 1.2881489992141724,
      "learning_rate": 9.935447368421053e-05,
      "loss": 0.4272,
      "step": 22463
    },
    {
      "epoch": 1.257923619666256,
      "grad_norm": 1.2905099391937256,
      "learning_rate": 9.935421052631579e-05,
      "loss": 0.4378,
      "step": 22464
    },
    {
      "epoch": 1.257979616978385,
      "grad_norm": 1.4473720788955688,
      "learning_rate": 9.935394736842105e-05,
      "loss": 0.4647,
      "step": 22465
    },
    {
      "epoch": 1.258035614290514,
      "grad_norm": 1.42154860496521,
      "learning_rate": 9.935368421052633e-05,
      "loss": 0.3559,
      "step": 22466
    },
    {
      "epoch": 1.258091611602643,
      "grad_norm": 1.2381058931350708,
      "learning_rate": 9.935342105263159e-05,
      "loss": 0.4903,
      "step": 22467
    },
    {
      "epoch": 1.258147608914772,
      "grad_norm": 1.1547553539276123,
      "learning_rate": 9.935315789473685e-05,
      "loss": 0.2718,
      "step": 22468
    },
    {
      "epoch": 1.258203606226901,
      "grad_norm": 1.392749547958374,
      "learning_rate": 9.93528947368421e-05,
      "loss": 0.5243,
      "step": 22469
    },
    {
      "epoch": 1.25825960353903,
      "grad_norm": 1.1658364534378052,
      "learning_rate": 9.935263157894738e-05,
      "loss": 0.4561,
      "step": 22470
    },
    {
      "epoch": 1.2583156008511591,
      "grad_norm": 1.4107457399368286,
      "learning_rate": 9.935236842105264e-05,
      "loss": 0.3249,
      "step": 22471
    },
    {
      "epoch": 1.2583715981632881,
      "grad_norm": 1.2678806781768799,
      "learning_rate": 9.93521052631579e-05,
      "loss": 0.3788,
      "step": 22472
    },
    {
      "epoch": 1.2584275954754172,
      "grad_norm": 1.3851932287216187,
      "learning_rate": 9.935184210526316e-05,
      "loss": 0.4094,
      "step": 22473
    },
    {
      "epoch": 1.2584835927875462,
      "grad_norm": 1.5530339479446411,
      "learning_rate": 9.935157894736843e-05,
      "loss": 0.6043,
      "step": 22474
    },
    {
      "epoch": 1.2585395900996752,
      "grad_norm": 1.711771845817566,
      "learning_rate": 9.935131578947369e-05,
      "loss": 0.6201,
      "step": 22475
    },
    {
      "epoch": 1.2585955874118042,
      "grad_norm": 1.614958643913269,
      "learning_rate": 9.935105263157895e-05,
      "loss": 0.5384,
      "step": 22476
    },
    {
      "epoch": 1.2586515847239332,
      "grad_norm": 1.2900781631469727,
      "learning_rate": 9.935078947368421e-05,
      "loss": 0.3236,
      "step": 22477
    },
    {
      "epoch": 1.2587075820360623,
      "grad_norm": 1.4119040966033936,
      "learning_rate": 9.935052631578948e-05,
      "loss": 0.4346,
      "step": 22478
    },
    {
      "epoch": 1.2587635793481913,
      "grad_norm": 1.2123664617538452,
      "learning_rate": 9.935026315789474e-05,
      "loss": 0.3862,
      "step": 22479
    },
    {
      "epoch": 1.2588195766603203,
      "grad_norm": 1.2141798734664917,
      "learning_rate": 9.935000000000002e-05,
      "loss": 0.4515,
      "step": 22480
    },
    {
      "epoch": 1.2588755739724493,
      "grad_norm": 1.382216453552246,
      "learning_rate": 9.934973684210526e-05,
      "loss": 0.4536,
      "step": 22481
    },
    {
      "epoch": 1.2589315712845783,
      "grad_norm": 1.378426432609558,
      "learning_rate": 9.934947368421052e-05,
      "loss": 0.581,
      "step": 22482
    },
    {
      "epoch": 1.2589875685967074,
      "grad_norm": 1.417020320892334,
      "learning_rate": 9.93492105263158e-05,
      "loss": 0.3848,
      "step": 22483
    },
    {
      "epoch": 1.2590435659088364,
      "grad_norm": 1.6204237937927246,
      "learning_rate": 9.934894736842106e-05,
      "loss": 0.5406,
      "step": 22484
    },
    {
      "epoch": 1.2590995632209654,
      "grad_norm": 1.4758626222610474,
      "learning_rate": 9.934868421052633e-05,
      "loss": 0.5592,
      "step": 22485
    },
    {
      "epoch": 1.2591555605330944,
      "grad_norm": 3.563157320022583,
      "learning_rate": 9.934842105263157e-05,
      "loss": 0.3563,
      "step": 22486
    },
    {
      "epoch": 1.2592115578452234,
      "grad_norm": 1.2028464078903198,
      "learning_rate": 9.934815789473685e-05,
      "loss": 0.3935,
      "step": 22487
    },
    {
      "epoch": 1.2592675551573524,
      "grad_norm": 1.3662872314453125,
      "learning_rate": 9.934789473684211e-05,
      "loss": 0.3752,
      "step": 22488
    },
    {
      "epoch": 1.2593235524694815,
      "grad_norm": 1.7939461469650269,
      "learning_rate": 9.934763157894738e-05,
      "loss": 0.4497,
      "step": 22489
    },
    {
      "epoch": 1.2593795497816105,
      "grad_norm": 1.3583811521530151,
      "learning_rate": 9.934736842105263e-05,
      "loss": 0.4819,
      "step": 22490
    },
    {
      "epoch": 1.2594355470937395,
      "grad_norm": 1.4784542322158813,
      "learning_rate": 9.93471052631579e-05,
      "loss": 0.512,
      "step": 22491
    },
    {
      "epoch": 1.2594915444058685,
      "grad_norm": 1.7761197090148926,
      "learning_rate": 9.934684210526316e-05,
      "loss": 0.4361,
      "step": 22492
    },
    {
      "epoch": 1.2595475417179975,
      "grad_norm": 1.5552124977111816,
      "learning_rate": 9.934657894736843e-05,
      "loss": 0.4968,
      "step": 22493
    },
    {
      "epoch": 1.2596035390301266,
      "grad_norm": 1.419356107711792,
      "learning_rate": 9.934631578947369e-05,
      "loss": 0.4412,
      "step": 22494
    },
    {
      "epoch": 1.2596595363422556,
      "grad_norm": 1.3594200611114502,
      "learning_rate": 9.934605263157895e-05,
      "loss": 0.4545,
      "step": 22495
    },
    {
      "epoch": 1.2597155336543846,
      "grad_norm": 1.2832539081573486,
      "learning_rate": 9.934578947368421e-05,
      "loss": 0.4305,
      "step": 22496
    },
    {
      "epoch": 1.2597715309665136,
      "grad_norm": 1.2215479612350464,
      "learning_rate": 9.934552631578949e-05,
      "loss": 0.4875,
      "step": 22497
    },
    {
      "epoch": 1.2598275282786426,
      "grad_norm": 1.4554733037948608,
      "learning_rate": 9.934526315789475e-05,
      "loss": 0.4148,
      "step": 22498
    },
    {
      "epoch": 1.2598835255907717,
      "grad_norm": 1.5905406475067139,
      "learning_rate": 9.9345e-05,
      "loss": 0.4645,
      "step": 22499
    },
    {
      "epoch": 1.2599395229029007,
      "grad_norm": 1.4548741579055786,
      "learning_rate": 9.934473684210526e-05,
      "loss": 0.4613,
      "step": 22500
    },
    {
      "epoch": 1.2599955202150297,
      "grad_norm": 1.1292834281921387,
      "learning_rate": 9.934447368421052e-05,
      "loss": 0.3886,
      "step": 22501
    },
    {
      "epoch": 1.2600515175271587,
      "grad_norm": 1.3880661725997925,
      "learning_rate": 9.93442105263158e-05,
      "loss": 0.4223,
      "step": 22502
    },
    {
      "epoch": 1.2601075148392877,
      "grad_norm": 1.3671998977661133,
      "learning_rate": 9.934394736842106e-05,
      "loss": 0.4806,
      "step": 22503
    },
    {
      "epoch": 1.2601635121514168,
      "grad_norm": 4.363045692443848,
      "learning_rate": 9.934368421052632e-05,
      "loss": 0.4813,
      "step": 22504
    },
    {
      "epoch": 1.2602195094635458,
      "grad_norm": 1.4290465116500854,
      "learning_rate": 9.934342105263158e-05,
      "loss": 0.5236,
      "step": 22505
    },
    {
      "epoch": 1.2602755067756748,
      "grad_norm": 2.196024179458618,
      "learning_rate": 9.934315789473685e-05,
      "loss": 0.3131,
      "step": 22506
    },
    {
      "epoch": 1.2603315040878038,
      "grad_norm": 1.4151439666748047,
      "learning_rate": 9.934289473684211e-05,
      "loss": 0.4163,
      "step": 22507
    },
    {
      "epoch": 1.2603875013999328,
      "grad_norm": 1.2434852123260498,
      "learning_rate": 9.934263157894737e-05,
      "loss": 0.4072,
      "step": 22508
    },
    {
      "epoch": 1.2604434987120618,
      "grad_norm": 1.2777451276779175,
      "learning_rate": 9.934236842105263e-05,
      "loss": 0.4306,
      "step": 22509
    },
    {
      "epoch": 1.2604994960241909,
      "grad_norm": 1.3524788618087769,
      "learning_rate": 9.93421052631579e-05,
      "loss": 0.4388,
      "step": 22510
    },
    {
      "epoch": 1.2605554933363199,
      "grad_norm": 1.5618058443069458,
      "learning_rate": 9.934184210526316e-05,
      "loss": 0.6227,
      "step": 22511
    },
    {
      "epoch": 1.260611490648449,
      "grad_norm": 1.243468165397644,
      "learning_rate": 9.934157894736844e-05,
      "loss": 0.4222,
      "step": 22512
    },
    {
      "epoch": 1.260667487960578,
      "grad_norm": 2.1506240367889404,
      "learning_rate": 9.934131578947368e-05,
      "loss": 0.4433,
      "step": 22513
    },
    {
      "epoch": 1.260723485272707,
      "grad_norm": 3.119558572769165,
      "learning_rate": 9.934105263157896e-05,
      "loss": 0.4319,
      "step": 22514
    },
    {
      "epoch": 1.260779482584836,
      "grad_norm": 1.3329764604568481,
      "learning_rate": 9.934078947368421e-05,
      "loss": 0.4438,
      "step": 22515
    },
    {
      "epoch": 1.260835479896965,
      "grad_norm": 1.4162479639053345,
      "learning_rate": 9.934052631578949e-05,
      "loss": 0.4017,
      "step": 22516
    },
    {
      "epoch": 1.260891477209094,
      "grad_norm": 1.6753971576690674,
      "learning_rate": 9.934026315789475e-05,
      "loss": 0.5185,
      "step": 22517
    },
    {
      "epoch": 1.260947474521223,
      "grad_norm": 1.3797465562820435,
      "learning_rate": 9.934e-05,
      "loss": 0.3933,
      "step": 22518
    },
    {
      "epoch": 1.261003471833352,
      "grad_norm": 1.4132909774780273,
      "learning_rate": 9.933973684210527e-05,
      "loss": 0.4236,
      "step": 22519
    },
    {
      "epoch": 1.261059469145481,
      "grad_norm": 1.2999058961868286,
      "learning_rate": 9.933947368421053e-05,
      "loss": 0.3285,
      "step": 22520
    },
    {
      "epoch": 1.26111546645761,
      "grad_norm": 1.4819581508636475,
      "learning_rate": 9.93392105263158e-05,
      "loss": 0.3924,
      "step": 22521
    },
    {
      "epoch": 1.261171463769739,
      "grad_norm": 2.1245462894439697,
      "learning_rate": 9.933894736842106e-05,
      "loss": 0.4404,
      "step": 22522
    },
    {
      "epoch": 1.2612274610818681,
      "grad_norm": 1.2867658138275146,
      "learning_rate": 9.933868421052632e-05,
      "loss": 0.5039,
      "step": 22523
    },
    {
      "epoch": 1.2612834583939971,
      "grad_norm": 1.4297866821289062,
      "learning_rate": 9.933842105263158e-05,
      "loss": 0.4432,
      "step": 22524
    },
    {
      "epoch": 1.2613394557061262,
      "grad_norm": 1.3429068326950073,
      "learning_rate": 9.933815789473685e-05,
      "loss": 0.5242,
      "step": 22525
    },
    {
      "epoch": 1.2613954530182552,
      "grad_norm": 2.5976576805114746,
      "learning_rate": 9.933789473684211e-05,
      "loss": 0.5616,
      "step": 22526
    },
    {
      "epoch": 1.2614514503303842,
      "grad_norm": 1.0975341796875,
      "learning_rate": 9.933763157894737e-05,
      "loss": 0.3919,
      "step": 22527
    },
    {
      "epoch": 1.2615074476425132,
      "grad_norm": 1.465174913406372,
      "learning_rate": 9.933736842105263e-05,
      "loss": 0.5706,
      "step": 22528
    },
    {
      "epoch": 1.2615634449546422,
      "grad_norm": 1.300743579864502,
      "learning_rate": 9.93371052631579e-05,
      "loss": 0.4728,
      "step": 22529
    },
    {
      "epoch": 1.2616194422667713,
      "grad_norm": 2.122889757156372,
      "learning_rate": 9.933684210526317e-05,
      "loss": 0.3431,
      "step": 22530
    },
    {
      "epoch": 1.2616754395789003,
      "grad_norm": 1.3574353456497192,
      "learning_rate": 9.933657894736842e-05,
      "loss": 0.4763,
      "step": 22531
    },
    {
      "epoch": 1.2617314368910293,
      "grad_norm": 1.1337097883224487,
      "learning_rate": 9.933631578947368e-05,
      "loss": 0.3846,
      "step": 22532
    },
    {
      "epoch": 1.2617874342031583,
      "grad_norm": 1.2418605089187622,
      "learning_rate": 9.933605263157896e-05,
      "loss": 0.4468,
      "step": 22533
    },
    {
      "epoch": 1.2618434315152873,
      "grad_norm": 1.3232368230819702,
      "learning_rate": 9.933578947368422e-05,
      "loss": 0.5621,
      "step": 22534
    },
    {
      "epoch": 1.2618994288274163,
      "grad_norm": 1.1776726245880127,
      "learning_rate": 9.933552631578948e-05,
      "loss": 0.4511,
      "step": 22535
    },
    {
      "epoch": 1.2619554261395454,
      "grad_norm": 1.4379807710647583,
      "learning_rate": 9.933526315789474e-05,
      "loss": 0.4569,
      "step": 22536
    },
    {
      "epoch": 1.2620114234516744,
      "grad_norm": 1.3012895584106445,
      "learning_rate": 9.9335e-05,
      "loss": 0.3947,
      "step": 22537
    },
    {
      "epoch": 1.2620674207638034,
      "grad_norm": 1.8353028297424316,
      "learning_rate": 9.933473684210527e-05,
      "loss": 0.3924,
      "step": 22538
    },
    {
      "epoch": 1.2621234180759324,
      "grad_norm": 1.2095950841903687,
      "learning_rate": 9.933447368421053e-05,
      "loss": 0.4401,
      "step": 22539
    },
    {
      "epoch": 1.2621794153880614,
      "grad_norm": 1.2847932577133179,
      "learning_rate": 9.93342105263158e-05,
      "loss": 0.4456,
      "step": 22540
    },
    {
      "epoch": 1.2622354127001905,
      "grad_norm": 1.5020012855529785,
      "learning_rate": 9.933394736842105e-05,
      "loss": 0.5991,
      "step": 22541
    },
    {
      "epoch": 1.2622914100123195,
      "grad_norm": 1.4112101793289185,
      "learning_rate": 9.933368421052632e-05,
      "loss": 0.4483,
      "step": 22542
    },
    {
      "epoch": 1.2623474073244485,
      "grad_norm": 3.058119058609009,
      "learning_rate": 9.933342105263158e-05,
      "loss": 0.3954,
      "step": 22543
    },
    {
      "epoch": 1.2624034046365775,
      "grad_norm": 1.421589732170105,
      "learning_rate": 9.933315789473686e-05,
      "loss": 0.4271,
      "step": 22544
    },
    {
      "epoch": 1.2624594019487065,
      "grad_norm": 1.2064660787582397,
      "learning_rate": 9.93328947368421e-05,
      "loss": 0.3404,
      "step": 22545
    },
    {
      "epoch": 1.2625153992608356,
      "grad_norm": 1.434069037437439,
      "learning_rate": 9.933263157894737e-05,
      "loss": 0.4127,
      "step": 22546
    },
    {
      "epoch": 1.2625713965729646,
      "grad_norm": 1.3939950466156006,
      "learning_rate": 9.933236842105263e-05,
      "loss": 0.4086,
      "step": 22547
    },
    {
      "epoch": 1.2626273938850936,
      "grad_norm": 1.2280699014663696,
      "learning_rate": 9.933210526315791e-05,
      "loss": 0.5022,
      "step": 22548
    },
    {
      "epoch": 1.2626833911972226,
      "grad_norm": 1.229111909866333,
      "learning_rate": 9.933184210526317e-05,
      "loss": 0.4686,
      "step": 22549
    },
    {
      "epoch": 1.2627393885093516,
      "grad_norm": 1.3211842775344849,
      "learning_rate": 9.933157894736843e-05,
      "loss": 0.3912,
      "step": 22550
    },
    {
      "epoch": 1.2627953858214807,
      "grad_norm": 1.461344838142395,
      "learning_rate": 9.933131578947369e-05,
      "loss": 0.5204,
      "step": 22551
    },
    {
      "epoch": 1.2628513831336097,
      "grad_norm": 1.4679924249649048,
      "learning_rate": 9.933105263157895e-05,
      "loss": 0.4477,
      "step": 22552
    },
    {
      "epoch": 1.2629073804457387,
      "grad_norm": 1.2944622039794922,
      "learning_rate": 9.933078947368422e-05,
      "loss": 0.3761,
      "step": 22553
    },
    {
      "epoch": 1.2629633777578677,
      "grad_norm": 1.4670464992523193,
      "learning_rate": 9.933052631578948e-05,
      "loss": 0.4294,
      "step": 22554
    },
    {
      "epoch": 1.2630193750699967,
      "grad_norm": 1.3800796270370483,
      "learning_rate": 9.933026315789474e-05,
      "loss": 0.4694,
      "step": 22555
    },
    {
      "epoch": 1.2630753723821257,
      "grad_norm": 1.2836834192276,
      "learning_rate": 9.933e-05,
      "loss": 0.437,
      "step": 22556
    },
    {
      "epoch": 1.2631313696942548,
      "grad_norm": 1.2517718076705933,
      "learning_rate": 9.932973684210527e-05,
      "loss": 0.446,
      "step": 22557
    },
    {
      "epoch": 1.2631873670063838,
      "grad_norm": 1.3426647186279297,
      "learning_rate": 9.932947368421053e-05,
      "loss": 0.4189,
      "step": 22558
    },
    {
      "epoch": 1.2632433643185128,
      "grad_norm": 1.2995389699935913,
      "learning_rate": 9.932921052631579e-05,
      "loss": 0.3016,
      "step": 22559
    },
    {
      "epoch": 1.2632993616306418,
      "grad_norm": 1.4668916463851929,
      "learning_rate": 9.932894736842105e-05,
      "loss": 0.3776,
      "step": 22560
    },
    {
      "epoch": 1.2633553589427708,
      "grad_norm": 1.3852781057357788,
      "learning_rate": 9.932868421052633e-05,
      "loss": 0.5454,
      "step": 22561
    },
    {
      "epoch": 1.2634113562548999,
      "grad_norm": 1.3087972402572632,
      "learning_rate": 9.932842105263158e-05,
      "loss": 0.4569,
      "step": 22562
    },
    {
      "epoch": 1.2634673535670289,
      "grad_norm": 1.356679916381836,
      "learning_rate": 9.932815789473684e-05,
      "loss": 0.4116,
      "step": 22563
    },
    {
      "epoch": 1.263523350879158,
      "grad_norm": 1.1689960956573486,
      "learning_rate": 9.93278947368421e-05,
      "loss": 0.3984,
      "step": 22564
    },
    {
      "epoch": 1.263579348191287,
      "grad_norm": 1.408976674079895,
      "learning_rate": 9.932763157894738e-05,
      "loss": 0.5703,
      "step": 22565
    },
    {
      "epoch": 1.263635345503416,
      "grad_norm": 1.362504243850708,
      "learning_rate": 9.932736842105264e-05,
      "loss": 0.3978,
      "step": 22566
    },
    {
      "epoch": 1.263691342815545,
      "grad_norm": 1.2659173011779785,
      "learning_rate": 9.932710526315791e-05,
      "loss": 0.4344,
      "step": 22567
    },
    {
      "epoch": 1.263747340127674,
      "grad_norm": 1.2167366743087769,
      "learning_rate": 9.932684210526316e-05,
      "loss": 0.4085,
      "step": 22568
    },
    {
      "epoch": 1.263803337439803,
      "grad_norm": 1.1932754516601562,
      "learning_rate": 9.932657894736842e-05,
      "loss": 0.3856,
      "step": 22569
    },
    {
      "epoch": 1.263859334751932,
      "grad_norm": 1.2302398681640625,
      "learning_rate": 9.932631578947369e-05,
      "loss": 0.3514,
      "step": 22570
    },
    {
      "epoch": 1.263915332064061,
      "grad_norm": 1.4113627672195435,
      "learning_rate": 9.932605263157895e-05,
      "loss": 0.562,
      "step": 22571
    },
    {
      "epoch": 1.26397132937619,
      "grad_norm": 1.2027363777160645,
      "learning_rate": 9.932578947368422e-05,
      "loss": 0.4555,
      "step": 22572
    },
    {
      "epoch": 1.264027326688319,
      "grad_norm": 1.3833410739898682,
      "learning_rate": 9.932552631578947e-05,
      "loss": 0.4095,
      "step": 22573
    },
    {
      "epoch": 1.264083324000448,
      "grad_norm": 1.2301329374313354,
      "learning_rate": 9.932526315789474e-05,
      "loss": 0.3541,
      "step": 22574
    },
    {
      "epoch": 1.2641393213125771,
      "grad_norm": 1.1803874969482422,
      "learning_rate": 9.9325e-05,
      "loss": 0.3284,
      "step": 22575
    },
    {
      "epoch": 1.2641953186247061,
      "grad_norm": 1.8832403421401978,
      "learning_rate": 9.932473684210528e-05,
      "loss": 0.4084,
      "step": 22576
    },
    {
      "epoch": 1.2642513159368352,
      "grad_norm": 1.1612879037857056,
      "learning_rate": 9.932447368421053e-05,
      "loss": 0.3429,
      "step": 22577
    },
    {
      "epoch": 1.2643073132489642,
      "grad_norm": 1.7041854858398438,
      "learning_rate": 9.93242105263158e-05,
      "loss": 0.3767,
      "step": 22578
    },
    {
      "epoch": 1.2643633105610932,
      "grad_norm": 1.5245976448059082,
      "learning_rate": 9.932394736842105e-05,
      "loss": 0.4273,
      "step": 22579
    },
    {
      "epoch": 1.2644193078732222,
      "grad_norm": 1.4128202199935913,
      "learning_rate": 9.932368421052633e-05,
      "loss": 0.3816,
      "step": 22580
    },
    {
      "epoch": 1.2644753051853512,
      "grad_norm": 1.1555976867675781,
      "learning_rate": 9.932342105263159e-05,
      "loss": 0.4031,
      "step": 22581
    },
    {
      "epoch": 1.2645313024974802,
      "grad_norm": 1.5287786722183228,
      "learning_rate": 9.932315789473685e-05,
      "loss": 0.4134,
      "step": 22582
    },
    {
      "epoch": 1.2645872998096093,
      "grad_norm": 1.2848385572433472,
      "learning_rate": 9.932289473684211e-05,
      "loss": 0.3846,
      "step": 22583
    },
    {
      "epoch": 1.2646432971217383,
      "grad_norm": 4.332828521728516,
      "learning_rate": 9.932263157894738e-05,
      "loss": 0.5469,
      "step": 22584
    },
    {
      "epoch": 1.2646992944338673,
      "grad_norm": 1.4228366613388062,
      "learning_rate": 9.932236842105264e-05,
      "loss": 0.6276,
      "step": 22585
    },
    {
      "epoch": 1.2647552917459963,
      "grad_norm": 1.3125123977661133,
      "learning_rate": 9.93221052631579e-05,
      "loss": 0.5502,
      "step": 22586
    },
    {
      "epoch": 1.2648112890581253,
      "grad_norm": 1.3094737529754639,
      "learning_rate": 9.932184210526316e-05,
      "loss": 0.4707,
      "step": 22587
    },
    {
      "epoch": 1.2648672863702544,
      "grad_norm": 1.190169334411621,
      "learning_rate": 9.932157894736842e-05,
      "loss": 0.4817,
      "step": 22588
    },
    {
      "epoch": 1.2649232836823834,
      "grad_norm": 1.2978438138961792,
      "learning_rate": 9.932131578947369e-05,
      "loss": 0.4921,
      "step": 22589
    },
    {
      "epoch": 1.2649792809945124,
      "grad_norm": 1.3994370698928833,
      "learning_rate": 9.932105263157895e-05,
      "loss": 0.8326,
      "step": 22590
    },
    {
      "epoch": 1.2650352783066414,
      "grad_norm": Infinity,
      "learning_rate": 9.932105263157895e-05,
      "loss": 0.5344,
      "step": 22591
    },
    {
      "epoch": 1.2650912756187702,
      "grad_norm": 1.2328864336013794,
      "learning_rate": 9.932078947368421e-05,
      "loss": 0.3464,
      "step": 22592
    },
    {
      "epoch": 1.2651472729308992,
      "grad_norm": 1.366552710533142,
      "learning_rate": 9.932052631578947e-05,
      "loss": 0.481,
      "step": 22593
    },
    {
      "epoch": 1.2652032702430283,
      "grad_norm": 1.4918768405914307,
      "learning_rate": 9.932026315789474e-05,
      "loss": 0.5796,
      "step": 22594
    },
    {
      "epoch": 1.2652592675551573,
      "grad_norm": 1.2238956689834595,
      "learning_rate": 9.932e-05,
      "loss": 0.4425,
      "step": 22595
    },
    {
      "epoch": 1.2653152648672863,
      "grad_norm": 1.2184131145477295,
      "learning_rate": 9.931973684210526e-05,
      "loss": 0.3699,
      "step": 22596
    },
    {
      "epoch": 1.2653712621794153,
      "grad_norm": 1.5443089008331299,
      "learning_rate": 9.931947368421052e-05,
      "loss": 0.4648,
      "step": 22597
    },
    {
      "epoch": 1.2654272594915443,
      "grad_norm": 1.5369009971618652,
      "learning_rate": 9.93192105263158e-05,
      "loss": 0.4838,
      "step": 22598
    },
    {
      "epoch": 1.2654832568036734,
      "grad_norm": 1.3724949359893799,
      "learning_rate": 9.931894736842106e-05,
      "loss": 0.5331,
      "step": 22599
    },
    {
      "epoch": 1.2655392541158024,
      "grad_norm": 1.2850879430770874,
      "learning_rate": 9.931868421052633e-05,
      "loss": 0.4102,
      "step": 22600
    },
    {
      "epoch": 1.2655952514279314,
      "grad_norm": 1.2060558795928955,
      "learning_rate": 9.931842105263158e-05,
      "loss": 0.4479,
      "step": 22601
    },
    {
      "epoch": 1.2656512487400604,
      "grad_norm": 1.307312250137329,
      "learning_rate": 9.931815789473685e-05,
      "loss": 0.4925,
      "step": 22602
    },
    {
      "epoch": 1.2657072460521894,
      "grad_norm": 1.4637165069580078,
      "learning_rate": 9.931789473684211e-05,
      "loss": 0.4648,
      "step": 22603
    },
    {
      "epoch": 1.2657632433643184,
      "grad_norm": 1.487190842628479,
      "learning_rate": 9.931763157894737e-05,
      "loss": 0.4919,
      "step": 22604
    },
    {
      "epoch": 1.2658192406764475,
      "grad_norm": 1.3126341104507446,
      "learning_rate": 9.931736842105264e-05,
      "loss": 0.3671,
      "step": 22605
    },
    {
      "epoch": 1.2658752379885765,
      "grad_norm": 1.2969690561294556,
      "learning_rate": 9.931710526315789e-05,
      "loss": 0.5605,
      "step": 22606
    },
    {
      "epoch": 1.2659312353007055,
      "grad_norm": 1.6308763027191162,
      "learning_rate": 9.931684210526316e-05,
      "loss": 0.4663,
      "step": 22607
    },
    {
      "epoch": 1.2659872326128345,
      "grad_norm": 1.4421902894973755,
      "learning_rate": 9.931657894736842e-05,
      "loss": 0.5375,
      "step": 22608
    },
    {
      "epoch": 1.2660432299249635,
      "grad_norm": 1.2509255409240723,
      "learning_rate": 9.93163157894737e-05,
      "loss": 0.4908,
      "step": 22609
    },
    {
      "epoch": 1.2660992272370926,
      "grad_norm": 1.2548978328704834,
      "learning_rate": 9.931605263157895e-05,
      "loss": 0.4617,
      "step": 22610
    },
    {
      "epoch": 1.2661552245492216,
      "grad_norm": 1.403206467628479,
      "learning_rate": 9.931578947368421e-05,
      "loss": 0.4799,
      "step": 22611
    },
    {
      "epoch": 1.2662112218613506,
      "grad_norm": 1.8052750825881958,
      "learning_rate": 9.931552631578947e-05,
      "loss": 0.5925,
      "step": 22612
    },
    {
      "epoch": 1.2662672191734796,
      "grad_norm": 6.041443824768066,
      "learning_rate": 9.931526315789475e-05,
      "loss": 0.4942,
      "step": 22613
    },
    {
      "epoch": 1.2663232164856086,
      "grad_norm": 1.2583448886871338,
      "learning_rate": 9.931500000000001e-05,
      "loss": 0.456,
      "step": 22614
    },
    {
      "epoch": 1.2663792137977377,
      "grad_norm": 1.2603918313980103,
      "learning_rate": 9.931473684210527e-05,
      "loss": 0.3992,
      "step": 22615
    },
    {
      "epoch": 1.2664352111098667,
      "grad_norm": 1.594143271446228,
      "learning_rate": 9.931447368421053e-05,
      "loss": 0.6076,
      "step": 22616
    },
    {
      "epoch": 1.2664912084219957,
      "grad_norm": 1.2620702981948853,
      "learning_rate": 9.93142105263158e-05,
      "loss": 0.3765,
      "step": 22617
    },
    {
      "epoch": 1.2665472057341247,
      "grad_norm": 1.5117707252502441,
      "learning_rate": 9.931394736842106e-05,
      "loss": 0.4807,
      "step": 22618
    },
    {
      "epoch": 1.2666032030462537,
      "grad_norm": 1.297575831413269,
      "learning_rate": 9.931368421052632e-05,
      "loss": 0.4377,
      "step": 22619
    },
    {
      "epoch": 1.2666592003583828,
      "grad_norm": 1.1439862251281738,
      "learning_rate": 9.931342105263158e-05,
      "loss": 0.2711,
      "step": 22620
    },
    {
      "epoch": 1.2667151976705118,
      "grad_norm": 1.2498003244400024,
      "learning_rate": 9.931315789473685e-05,
      "loss": 0.4216,
      "step": 22621
    },
    {
      "epoch": 1.2667711949826408,
      "grad_norm": 1.4508490562438965,
      "learning_rate": 9.931289473684211e-05,
      "loss": 0.5733,
      "step": 22622
    },
    {
      "epoch": 1.2668271922947698,
      "grad_norm": 1.2860407829284668,
      "learning_rate": 9.931263157894737e-05,
      "loss": 0.4151,
      "step": 22623
    },
    {
      "epoch": 1.2668831896068988,
      "grad_norm": 1.059950590133667,
      "learning_rate": 9.931236842105263e-05,
      "loss": 0.4224,
      "step": 22624
    },
    {
      "epoch": 1.2669391869190278,
      "grad_norm": 1.1142617464065552,
      "learning_rate": 9.931210526315789e-05,
      "loss": 0.348,
      "step": 22625
    },
    {
      "epoch": 1.2669951842311569,
      "grad_norm": 2.086233615875244,
      "learning_rate": 9.931184210526316e-05,
      "loss": 0.5684,
      "step": 22626
    },
    {
      "epoch": 1.2670511815432859,
      "grad_norm": 1.5305190086364746,
      "learning_rate": 9.931157894736842e-05,
      "loss": 0.4595,
      "step": 22627
    },
    {
      "epoch": 1.267107178855415,
      "grad_norm": 1.4111287593841553,
      "learning_rate": 9.93113157894737e-05,
      "loss": 0.5267,
      "step": 22628
    },
    {
      "epoch": 1.267163176167544,
      "grad_norm": 2.5089962482452393,
      "learning_rate": 9.931105263157894e-05,
      "loss": 0.4802,
      "step": 22629
    },
    {
      "epoch": 1.267219173479673,
      "grad_norm": 1.3890100717544556,
      "learning_rate": 9.931078947368422e-05,
      "loss": 0.4647,
      "step": 22630
    },
    {
      "epoch": 1.267275170791802,
      "grad_norm": 1.2288728952407837,
      "learning_rate": 9.931052631578948e-05,
      "loss": 0.4682,
      "step": 22631
    },
    {
      "epoch": 1.267331168103931,
      "grad_norm": 1.0214381217956543,
      "learning_rate": 9.931026315789475e-05,
      "loss": 0.3127,
      "step": 22632
    },
    {
      "epoch": 1.26738716541606,
      "grad_norm": 1.4180487394332886,
      "learning_rate": 9.931000000000001e-05,
      "loss": 0.3874,
      "step": 22633
    },
    {
      "epoch": 1.267443162728189,
      "grad_norm": 1.4683669805526733,
      "learning_rate": 9.930973684210527e-05,
      "loss": 0.424,
      "step": 22634
    },
    {
      "epoch": 1.267499160040318,
      "grad_norm": 1.7716941833496094,
      "learning_rate": 9.930947368421053e-05,
      "loss": 0.5721,
      "step": 22635
    },
    {
      "epoch": 1.267555157352447,
      "grad_norm": 1.7148070335388184,
      "learning_rate": 9.93092105263158e-05,
      "loss": 0.5187,
      "step": 22636
    },
    {
      "epoch": 1.267611154664576,
      "grad_norm": 1.155671238899231,
      "learning_rate": 9.930894736842106e-05,
      "loss": 0.2932,
      "step": 22637
    },
    {
      "epoch": 1.267667151976705,
      "grad_norm": 1.2717502117156982,
      "learning_rate": 9.930868421052632e-05,
      "loss": 0.3633,
      "step": 22638
    },
    {
      "epoch": 1.2677231492888341,
      "grad_norm": 1.1481229066848755,
      "learning_rate": 9.930842105263158e-05,
      "loss": 0.3724,
      "step": 22639
    },
    {
      "epoch": 1.2677791466009631,
      "grad_norm": 1.5281686782836914,
      "learning_rate": 9.930815789473684e-05,
      "loss": 0.4342,
      "step": 22640
    },
    {
      "epoch": 1.2678351439130922,
      "grad_norm": 1.3560229539871216,
      "learning_rate": 9.930789473684211e-05,
      "loss": 0.3925,
      "step": 22641
    },
    {
      "epoch": 1.2678911412252212,
      "grad_norm": 1.318763017654419,
      "learning_rate": 9.930763157894737e-05,
      "loss": 0.3761,
      "step": 22642
    },
    {
      "epoch": 1.2679471385373502,
      "grad_norm": 1.2733261585235596,
      "learning_rate": 9.930736842105263e-05,
      "loss": 0.5068,
      "step": 22643
    },
    {
      "epoch": 1.2680031358494792,
      "grad_norm": 1.5524920225143433,
      "learning_rate": 9.93071052631579e-05,
      "loss": 0.4662,
      "step": 22644
    },
    {
      "epoch": 1.2680591331616082,
      "grad_norm": 1.2721418142318726,
      "learning_rate": 9.930684210526317e-05,
      "loss": 0.4295,
      "step": 22645
    },
    {
      "epoch": 1.2681151304737373,
      "grad_norm": 1.371688723564148,
      "learning_rate": 9.930657894736843e-05,
      "loss": 0.4775,
      "step": 22646
    },
    {
      "epoch": 1.2681711277858663,
      "grad_norm": 1.32282292842865,
      "learning_rate": 9.930631578947369e-05,
      "loss": 0.4748,
      "step": 22647
    },
    {
      "epoch": 1.2682271250979953,
      "grad_norm": 1.218231439590454,
      "learning_rate": 9.930605263157895e-05,
      "loss": 0.4281,
      "step": 22648
    },
    {
      "epoch": 1.2682831224101243,
      "grad_norm": 1.451771855354309,
      "learning_rate": 9.930578947368422e-05,
      "loss": 0.4842,
      "step": 22649
    },
    {
      "epoch": 1.2683391197222533,
      "grad_norm": 3.515385150909424,
      "learning_rate": 9.930552631578948e-05,
      "loss": 0.5745,
      "step": 22650
    },
    {
      "epoch": 1.2683951170343823,
      "grad_norm": 1.4690711498260498,
      "learning_rate": 9.930526315789474e-05,
      "loss": 0.4232,
      "step": 22651
    },
    {
      "epoch": 1.2684511143465114,
      "grad_norm": 2.0906107425689697,
      "learning_rate": 9.9305e-05,
      "loss": 0.3972,
      "step": 22652
    },
    {
      "epoch": 1.2685071116586404,
      "grad_norm": 1.2995021343231201,
      "learning_rate": 9.930473684210527e-05,
      "loss": 0.5252,
      "step": 22653
    },
    {
      "epoch": 1.2685631089707694,
      "grad_norm": 1.2416428327560425,
      "learning_rate": 9.930447368421053e-05,
      "loss": 0.4176,
      "step": 22654
    },
    {
      "epoch": 1.2686191062828984,
      "grad_norm": 1.553397536277771,
      "learning_rate": 9.93042105263158e-05,
      "loss": 0.4392,
      "step": 22655
    },
    {
      "epoch": 1.2686751035950274,
      "grad_norm": 1.4556041955947876,
      "learning_rate": 9.930394736842105e-05,
      "loss": 0.4612,
      "step": 22656
    },
    {
      "epoch": 1.2687311009071565,
      "grad_norm": 1.3560270071029663,
      "learning_rate": 9.930368421052631e-05,
      "loss": 0.5722,
      "step": 22657
    },
    {
      "epoch": 1.2687870982192855,
      "grad_norm": 1.7231885194778442,
      "learning_rate": 9.930342105263158e-05,
      "loss": 0.5763,
      "step": 22658
    },
    {
      "epoch": 1.2688430955314145,
      "grad_norm": 1.4385744333267212,
      "learning_rate": 9.930315789473684e-05,
      "loss": 0.3599,
      "step": 22659
    },
    {
      "epoch": 1.2688990928435435,
      "grad_norm": 2.524521589279175,
      "learning_rate": 9.930289473684212e-05,
      "loss": 0.6094,
      "step": 22660
    },
    {
      "epoch": 1.2689550901556725,
      "grad_norm": 3.413239002227783,
      "learning_rate": 9.930263157894736e-05,
      "loss": 0.6705,
      "step": 22661
    },
    {
      "epoch": 1.2690110874678016,
      "grad_norm": 1.4177204370498657,
      "learning_rate": 9.930236842105264e-05,
      "loss": 0.4917,
      "step": 22662
    },
    {
      "epoch": 1.2690670847799306,
      "grad_norm": 1.4034504890441895,
      "learning_rate": 9.93021052631579e-05,
      "loss": 0.432,
      "step": 22663
    },
    {
      "epoch": 1.2691230820920596,
      "grad_norm": 1.50716233253479,
      "learning_rate": 9.930184210526317e-05,
      "loss": 0.5209,
      "step": 22664
    },
    {
      "epoch": 1.2691790794041886,
      "grad_norm": 1.3623316287994385,
      "learning_rate": 9.930157894736843e-05,
      "loss": 0.4581,
      "step": 22665
    },
    {
      "epoch": 1.2692350767163176,
      "grad_norm": 1.5193994045257568,
      "learning_rate": 9.930131578947369e-05,
      "loss": 0.3974,
      "step": 22666
    },
    {
      "epoch": 1.2692910740284467,
      "grad_norm": 1.293651819229126,
      "learning_rate": 9.930105263157895e-05,
      "loss": 0.3696,
      "step": 22667
    },
    {
      "epoch": 1.2693470713405757,
      "grad_norm": 1.8996182680130005,
      "learning_rate": 9.930078947368422e-05,
      "loss": 0.4571,
      "step": 22668
    },
    {
      "epoch": 1.2694030686527047,
      "grad_norm": 1.2406213283538818,
      "learning_rate": 9.930052631578948e-05,
      "loss": 0.4494,
      "step": 22669
    },
    {
      "epoch": 1.2694590659648337,
      "grad_norm": 4.635396957397461,
      "learning_rate": 9.930026315789474e-05,
      "loss": 0.5383,
      "step": 22670
    },
    {
      "epoch": 1.2695150632769627,
      "grad_norm": 1.3629764318466187,
      "learning_rate": 9.93e-05,
      "loss": 0.3442,
      "step": 22671
    },
    {
      "epoch": 1.2695710605890917,
      "grad_norm": 1.5809091329574585,
      "learning_rate": 9.929973684210527e-05,
      "loss": 0.4282,
      "step": 22672
    },
    {
      "epoch": 1.2696270579012208,
      "grad_norm": 1.1668684482574463,
      "learning_rate": 9.929947368421053e-05,
      "loss": 0.3434,
      "step": 22673
    },
    {
      "epoch": 1.2696830552133498,
      "grad_norm": 1.1767786741256714,
      "learning_rate": 9.92992105263158e-05,
      "loss": 0.3571,
      "step": 22674
    },
    {
      "epoch": 1.2697390525254788,
      "grad_norm": 1.4719984531402588,
      "learning_rate": 9.929894736842105e-05,
      "loss": 0.4222,
      "step": 22675
    },
    {
      "epoch": 1.2697950498376078,
      "grad_norm": 1.181861162185669,
      "learning_rate": 9.929868421052631e-05,
      "loss": 0.4426,
      "step": 22676
    },
    {
      "epoch": 1.2698510471497368,
      "grad_norm": 1.6003828048706055,
      "learning_rate": 9.929842105263159e-05,
      "loss": 0.4858,
      "step": 22677
    },
    {
      "epoch": 1.2699070444618659,
      "grad_norm": 2.3208487033843994,
      "learning_rate": 9.929815789473685e-05,
      "loss": 0.506,
      "step": 22678
    },
    {
      "epoch": 1.2699630417739949,
      "grad_norm": 1.3948818445205688,
      "learning_rate": 9.92978947368421e-05,
      "loss": 0.3287,
      "step": 22679
    },
    {
      "epoch": 1.270019039086124,
      "grad_norm": 1.5072851181030273,
      "learning_rate": 9.929763157894737e-05,
      "loss": 0.4279,
      "step": 22680
    },
    {
      "epoch": 1.270075036398253,
      "grad_norm": 1.5477532148361206,
      "learning_rate": 9.929736842105264e-05,
      "loss": 0.5258,
      "step": 22681
    },
    {
      "epoch": 1.270131033710382,
      "grad_norm": 1.3021478652954102,
      "learning_rate": 9.92971052631579e-05,
      "loss": 0.4106,
      "step": 22682
    },
    {
      "epoch": 1.270187031022511,
      "grad_norm": 12.972148895263672,
      "learning_rate": 9.929684210526317e-05,
      "loss": 0.4896,
      "step": 22683
    },
    {
      "epoch": 1.27024302833464,
      "grad_norm": 1.3581346273422241,
      "learning_rate": 9.929657894736842e-05,
      "loss": 0.5091,
      "step": 22684
    },
    {
      "epoch": 1.270299025646769,
      "grad_norm": 2.3846018314361572,
      "learning_rate": 9.929631578947369e-05,
      "loss": 0.5777,
      "step": 22685
    },
    {
      "epoch": 1.270355022958898,
      "grad_norm": 1.26251220703125,
      "learning_rate": 9.929605263157895e-05,
      "loss": 0.4046,
      "step": 22686
    },
    {
      "epoch": 1.270411020271027,
      "grad_norm": 1.7669390439987183,
      "learning_rate": 9.929578947368422e-05,
      "loss": 0.3931,
      "step": 22687
    },
    {
      "epoch": 1.270467017583156,
      "grad_norm": 1.5544970035552979,
      "learning_rate": 9.929552631578948e-05,
      "loss": 0.5815,
      "step": 22688
    },
    {
      "epoch": 1.270523014895285,
      "grad_norm": 1.0569502115249634,
      "learning_rate": 9.929526315789474e-05,
      "loss": 0.3724,
      "step": 22689
    },
    {
      "epoch": 1.270579012207414,
      "grad_norm": 1.8381023406982422,
      "learning_rate": 9.9295e-05,
      "loss": 0.5538,
      "step": 22690
    },
    {
      "epoch": 1.2706350095195431,
      "grad_norm": 1.1855922937393188,
      "learning_rate": 9.929473684210526e-05,
      "loss": 0.4204,
      "step": 22691
    },
    {
      "epoch": 1.2706910068316721,
      "grad_norm": 1.2583905458450317,
      "learning_rate": 9.929447368421054e-05,
      "loss": 0.4074,
      "step": 22692
    },
    {
      "epoch": 1.2707470041438012,
      "grad_norm": 1.2125978469848633,
      "learning_rate": 9.929421052631578e-05,
      "loss": 0.4162,
      "step": 22693
    },
    {
      "epoch": 1.2708030014559302,
      "grad_norm": 1.5250681638717651,
      "learning_rate": 9.929394736842106e-05,
      "loss": 0.4806,
      "step": 22694
    },
    {
      "epoch": 1.2708589987680592,
      "grad_norm": 1.3331366777420044,
      "learning_rate": 9.929368421052632e-05,
      "loss": 0.3697,
      "step": 22695
    },
    {
      "epoch": 1.2709149960801882,
      "grad_norm": 1.1396459341049194,
      "learning_rate": 9.929342105263159e-05,
      "loss": 0.4164,
      "step": 22696
    },
    {
      "epoch": 1.2709709933923172,
      "grad_norm": 1.2768718004226685,
      "learning_rate": 9.929315789473685e-05,
      "loss": 0.3931,
      "step": 22697
    },
    {
      "epoch": 1.2710269907044462,
      "grad_norm": 1.8912655115127563,
      "learning_rate": 9.929289473684211e-05,
      "loss": 0.5287,
      "step": 22698
    },
    {
      "epoch": 1.2710829880165753,
      "grad_norm": 1.3284953832626343,
      "learning_rate": 9.929263157894737e-05,
      "loss": 0.4845,
      "step": 22699
    },
    {
      "epoch": 1.2711389853287043,
      "grad_norm": 1.2338652610778809,
      "learning_rate": 9.929236842105264e-05,
      "loss": 0.4182,
      "step": 22700
    },
    {
      "epoch": 1.2711949826408333,
      "grad_norm": 1.208163857460022,
      "learning_rate": 9.92921052631579e-05,
      "loss": 0.4332,
      "step": 22701
    },
    {
      "epoch": 1.2712509799529623,
      "grad_norm": 1.2945038080215454,
      "learning_rate": 9.929184210526316e-05,
      "loss": 0.4089,
      "step": 22702
    },
    {
      "epoch": 1.2713069772650913,
      "grad_norm": 1.4350838661193848,
      "learning_rate": 9.929157894736842e-05,
      "loss": 0.5331,
      "step": 22703
    },
    {
      "epoch": 1.2713629745772204,
      "grad_norm": 1.4823426008224487,
      "learning_rate": 9.92913157894737e-05,
      "loss": 0.4397,
      "step": 22704
    },
    {
      "epoch": 1.2714189718893494,
      "grad_norm": 1.3634966611862183,
      "learning_rate": 9.929105263157895e-05,
      "loss": 0.4962,
      "step": 22705
    },
    {
      "epoch": 1.2714749692014784,
      "grad_norm": 1.4452133178710938,
      "learning_rate": 9.929078947368421e-05,
      "loss": 0.5064,
      "step": 22706
    },
    {
      "epoch": 1.2715309665136074,
      "grad_norm": 1.3377217054367065,
      "learning_rate": 9.929052631578947e-05,
      "loss": 0.4391,
      "step": 22707
    },
    {
      "epoch": 1.2715869638257364,
      "grad_norm": 1.923106074333191,
      "learning_rate": 9.929026315789473e-05,
      "loss": 0.5312,
      "step": 22708
    },
    {
      "epoch": 1.2716429611378655,
      "grad_norm": 1.1288068294525146,
      "learning_rate": 9.929e-05,
      "loss": 0.4602,
      "step": 22709
    },
    {
      "epoch": 1.2716989584499945,
      "grad_norm": 1.2240934371948242,
      "learning_rate": 9.928973684210527e-05,
      "loss": 0.4053,
      "step": 22710
    },
    {
      "epoch": 1.2717549557621235,
      "grad_norm": 1.4392454624176025,
      "learning_rate": 9.928947368421053e-05,
      "loss": 0.4006,
      "step": 22711
    },
    {
      "epoch": 1.2718109530742525,
      "grad_norm": 1.578526496887207,
      "learning_rate": 9.928921052631579e-05,
      "loss": 0.5425,
      "step": 22712
    },
    {
      "epoch": 1.2718669503863815,
      "grad_norm": 2.0550968647003174,
      "learning_rate": 9.928894736842106e-05,
      "loss": 0.5106,
      "step": 22713
    },
    {
      "epoch": 1.2719229476985106,
      "grad_norm": 1.3066192865371704,
      "learning_rate": 9.928868421052632e-05,
      "loss": 0.3744,
      "step": 22714
    },
    {
      "epoch": 1.2719789450106396,
      "grad_norm": 1.4003432989120483,
      "learning_rate": 9.928842105263159e-05,
      "loss": 0.4796,
      "step": 22715
    },
    {
      "epoch": 1.2720349423227684,
      "grad_norm": 1.2575757503509521,
      "learning_rate": 9.928815789473684e-05,
      "loss": 0.3904,
      "step": 22716
    },
    {
      "epoch": 1.2720909396348974,
      "grad_norm": 1.1841113567352295,
      "learning_rate": 9.928789473684211e-05,
      "loss": 0.3849,
      "step": 22717
    },
    {
      "epoch": 1.2721469369470264,
      "grad_norm": 1.306808352470398,
      "learning_rate": 9.928763157894737e-05,
      "loss": 0.434,
      "step": 22718
    },
    {
      "epoch": 1.2722029342591554,
      "grad_norm": 1.1899160146713257,
      "learning_rate": 9.928736842105264e-05,
      "loss": 0.4461,
      "step": 22719
    },
    {
      "epoch": 1.2722589315712844,
      "grad_norm": 1.0495973825454712,
      "learning_rate": 9.92871052631579e-05,
      "loss": 0.4398,
      "step": 22720
    },
    {
      "epoch": 1.2723149288834135,
      "grad_norm": 1.5488747358322144,
      "learning_rate": 9.928684210526316e-05,
      "loss": 0.5062,
      "step": 22721
    },
    {
      "epoch": 1.2723709261955425,
      "grad_norm": 1.3600432872772217,
      "learning_rate": 9.928657894736842e-05,
      "loss": 0.4004,
      "step": 22722
    },
    {
      "epoch": 1.2724269235076715,
      "grad_norm": 1.1288678646087646,
      "learning_rate": 9.92863157894737e-05,
      "loss": 0.4184,
      "step": 22723
    },
    {
      "epoch": 1.2724829208198005,
      "grad_norm": 1.2847695350646973,
      "learning_rate": 9.928605263157896e-05,
      "loss": 0.434,
      "step": 22724
    },
    {
      "epoch": 1.2725389181319295,
      "grad_norm": 1.3045952320098877,
      "learning_rate": 9.928578947368422e-05,
      "loss": 0.4622,
      "step": 22725
    },
    {
      "epoch": 1.2725949154440586,
      "grad_norm": 1.2724335193634033,
      "learning_rate": 9.928552631578948e-05,
      "loss": 0.3853,
      "step": 22726
    },
    {
      "epoch": 1.2726509127561876,
      "grad_norm": 1.4003771543502808,
      "learning_rate": 9.928526315789474e-05,
      "loss": 0.4671,
      "step": 22727
    },
    {
      "epoch": 1.2727069100683166,
      "grad_norm": 1.9080029726028442,
      "learning_rate": 9.928500000000001e-05,
      "loss": 0.6663,
      "step": 22728
    },
    {
      "epoch": 1.2727629073804456,
      "grad_norm": 1.3403178453445435,
      "learning_rate": 9.928473684210527e-05,
      "loss": 0.4044,
      "step": 22729
    },
    {
      "epoch": 1.2728189046925746,
      "grad_norm": 1.050033450126648,
      "learning_rate": 9.928447368421053e-05,
      "loss": 0.3496,
      "step": 22730
    },
    {
      "epoch": 1.2728749020047037,
      "grad_norm": 1.0832682847976685,
      "learning_rate": 9.928421052631579e-05,
      "loss": 0.4166,
      "step": 22731
    },
    {
      "epoch": 1.2729308993168327,
      "grad_norm": 1.425012469291687,
      "learning_rate": 9.928394736842106e-05,
      "loss": 0.4182,
      "step": 22732
    },
    {
      "epoch": 1.2729868966289617,
      "grad_norm": 1.3718820810317993,
      "learning_rate": 9.928368421052632e-05,
      "loss": 0.5987,
      "step": 22733
    },
    {
      "epoch": 1.2730428939410907,
      "grad_norm": 1.3492567539215088,
      "learning_rate": 9.928342105263158e-05,
      "loss": 0.4202,
      "step": 22734
    },
    {
      "epoch": 1.2730988912532197,
      "grad_norm": 9.869452476501465,
      "learning_rate": 9.928315789473684e-05,
      "loss": 0.5806,
      "step": 22735
    },
    {
      "epoch": 1.2731548885653488,
      "grad_norm": 1.481376051902771,
      "learning_rate": 9.928289473684211e-05,
      "loss": 0.4489,
      "step": 22736
    },
    {
      "epoch": 1.2732108858774778,
      "grad_norm": 1.1709439754486084,
      "learning_rate": 9.928263157894737e-05,
      "loss": 0.3878,
      "step": 22737
    },
    {
      "epoch": 1.2732668831896068,
      "grad_norm": 4.314883708953857,
      "learning_rate": 9.928236842105265e-05,
      "loss": 0.5497,
      "step": 22738
    },
    {
      "epoch": 1.2733228805017358,
      "grad_norm": 1.635369062423706,
      "learning_rate": 9.928210526315789e-05,
      "loss": 0.4089,
      "step": 22739
    },
    {
      "epoch": 1.2733788778138648,
      "grad_norm": 1.622637391090393,
      "learning_rate": 9.928184210526317e-05,
      "loss": 0.4684,
      "step": 22740
    },
    {
      "epoch": 1.2734348751259938,
      "grad_norm": 1.3720473051071167,
      "learning_rate": 9.928157894736843e-05,
      "loss": 0.3942,
      "step": 22741
    },
    {
      "epoch": 1.2734908724381229,
      "grad_norm": 1.2264175415039062,
      "learning_rate": 9.92813157894737e-05,
      "loss": 0.3322,
      "step": 22742
    },
    {
      "epoch": 1.2735468697502519,
      "grad_norm": 1.4567515850067139,
      "learning_rate": 9.928105263157895e-05,
      "loss": 0.4325,
      "step": 22743
    },
    {
      "epoch": 1.273602867062381,
      "grad_norm": 1.5054537057876587,
      "learning_rate": 9.92807894736842e-05,
      "loss": 0.4869,
      "step": 22744
    },
    {
      "epoch": 1.27365886437451,
      "grad_norm": 1.3365561962127686,
      "learning_rate": 9.928052631578948e-05,
      "loss": 0.4938,
      "step": 22745
    },
    {
      "epoch": 1.273714861686639,
      "grad_norm": 1.5146760940551758,
      "learning_rate": 9.928026315789474e-05,
      "loss": 0.5028,
      "step": 22746
    },
    {
      "epoch": 1.273770858998768,
      "grad_norm": 1.0109838247299194,
      "learning_rate": 9.928000000000001e-05,
      "loss": 0.3685,
      "step": 22747
    },
    {
      "epoch": 1.273826856310897,
      "grad_norm": 1.3436310291290283,
      "learning_rate": 9.927973684210526e-05,
      "loss": 0.5354,
      "step": 22748
    },
    {
      "epoch": 1.273882853623026,
      "grad_norm": 1.4745495319366455,
      "learning_rate": 9.927947368421053e-05,
      "loss": 0.4171,
      "step": 22749
    },
    {
      "epoch": 1.273938850935155,
      "grad_norm": 1.4597543478012085,
      "learning_rate": 9.927921052631579e-05,
      "loss": 0.3829,
      "step": 22750
    },
    {
      "epoch": 1.273994848247284,
      "grad_norm": 1.4160975217819214,
      "learning_rate": 9.927894736842106e-05,
      "loss": 0.4748,
      "step": 22751
    },
    {
      "epoch": 1.274050845559413,
      "grad_norm": 1.3686274290084839,
      "learning_rate": 9.927868421052632e-05,
      "loss": 0.4098,
      "step": 22752
    },
    {
      "epoch": 1.274106842871542,
      "grad_norm": 1.3850740194320679,
      "learning_rate": 9.927842105263158e-05,
      "loss": 0.4831,
      "step": 22753
    },
    {
      "epoch": 1.274162840183671,
      "grad_norm": 2.055142402648926,
      "learning_rate": 9.927815789473684e-05,
      "loss": 0.4452,
      "step": 22754
    },
    {
      "epoch": 1.2742188374958001,
      "grad_norm": 1.2674225568771362,
      "learning_rate": 9.927789473684212e-05,
      "loss": 0.3733,
      "step": 22755
    },
    {
      "epoch": 1.2742748348079291,
      "grad_norm": 1.5821294784545898,
      "learning_rate": 9.927763157894738e-05,
      "loss": 0.4717,
      "step": 22756
    },
    {
      "epoch": 1.2743308321200582,
      "grad_norm": 1.2697193622589111,
      "learning_rate": 9.927736842105264e-05,
      "loss": 0.3312,
      "step": 22757
    },
    {
      "epoch": 1.2743868294321872,
      "grad_norm": 1.1778618097305298,
      "learning_rate": 9.92771052631579e-05,
      "loss": 0.3977,
      "step": 22758
    },
    {
      "epoch": 1.2744428267443162,
      "grad_norm": 1.5188071727752686,
      "learning_rate": 9.927684210526317e-05,
      "loss": 0.5024,
      "step": 22759
    },
    {
      "epoch": 1.2744988240564452,
      "grad_norm": 2.8332552909851074,
      "learning_rate": 9.927657894736843e-05,
      "loss": 0.4515,
      "step": 22760
    },
    {
      "epoch": 1.2745548213685742,
      "grad_norm": 0.9822652339935303,
      "learning_rate": 9.927631578947369e-05,
      "loss": 0.2588,
      "step": 22761
    },
    {
      "epoch": 1.2746108186807033,
      "grad_norm": 1.1252400875091553,
      "learning_rate": 9.927605263157895e-05,
      "loss": 0.4655,
      "step": 22762
    },
    {
      "epoch": 1.2746668159928323,
      "grad_norm": 1.4667878150939941,
      "learning_rate": 9.927578947368421e-05,
      "loss": 0.5684,
      "step": 22763
    },
    {
      "epoch": 1.2747228133049613,
      "grad_norm": 1.3001575469970703,
      "learning_rate": 9.927552631578948e-05,
      "loss": 0.3713,
      "step": 22764
    },
    {
      "epoch": 1.2747788106170903,
      "grad_norm": 1.185944676399231,
      "learning_rate": 9.927526315789474e-05,
      "loss": 0.3744,
      "step": 22765
    },
    {
      "epoch": 1.2748348079292193,
      "grad_norm": 1.4352209568023682,
      "learning_rate": 9.9275e-05,
      "loss": 0.5192,
      "step": 22766
    },
    {
      "epoch": 1.2748908052413483,
      "grad_norm": 1.1593495607376099,
      "learning_rate": 9.927473684210526e-05,
      "loss": 0.3618,
      "step": 22767
    },
    {
      "epoch": 1.2749468025534774,
      "grad_norm": 1.3951646089553833,
      "learning_rate": 9.927447368421053e-05,
      "loss": 0.4428,
      "step": 22768
    },
    {
      "epoch": 1.2750027998656064,
      "grad_norm": 1.2441003322601318,
      "learning_rate": 9.92742105263158e-05,
      "loss": 0.37,
      "step": 22769
    },
    {
      "epoch": 1.2750587971777354,
      "grad_norm": 1.2693815231323242,
      "learning_rate": 9.927394736842107e-05,
      "loss": 0.3531,
      "step": 22770
    },
    {
      "epoch": 1.2751147944898644,
      "grad_norm": 1.3525625467300415,
      "learning_rate": 9.927368421052631e-05,
      "loss": 0.3369,
      "step": 22771
    },
    {
      "epoch": 1.2751707918019934,
      "grad_norm": 1.563995122909546,
      "learning_rate": 9.927342105263159e-05,
      "loss": 0.4396,
      "step": 22772
    },
    {
      "epoch": 1.2752267891141225,
      "grad_norm": 1.6154165267944336,
      "learning_rate": 9.927315789473685e-05,
      "loss": 0.4805,
      "step": 22773
    },
    {
      "epoch": 1.2752827864262515,
      "grad_norm": 1.5421653985977173,
      "learning_rate": 9.927289473684212e-05,
      "loss": 0.4318,
      "step": 22774
    },
    {
      "epoch": 1.2753387837383805,
      "grad_norm": 1.5196174383163452,
      "learning_rate": 9.927263157894738e-05,
      "loss": 0.4372,
      "step": 22775
    },
    {
      "epoch": 1.2753947810505095,
      "grad_norm": 1.498060703277588,
      "learning_rate": 9.927236842105264e-05,
      "loss": 0.5817,
      "step": 22776
    },
    {
      "epoch": 1.2754507783626385,
      "grad_norm": 1.1516247987747192,
      "learning_rate": 9.92721052631579e-05,
      "loss": 0.4153,
      "step": 22777
    },
    {
      "epoch": 1.2755067756747676,
      "grad_norm": 1.5450985431671143,
      "learning_rate": 9.927184210526316e-05,
      "loss": 0.3987,
      "step": 22778
    },
    {
      "epoch": 1.2755627729868966,
      "grad_norm": 1.2992115020751953,
      "learning_rate": 9.927157894736843e-05,
      "loss": 0.4102,
      "step": 22779
    },
    {
      "epoch": 1.2756187702990256,
      "grad_norm": 1.2836987972259521,
      "learning_rate": 9.927131578947369e-05,
      "loss": 0.4377,
      "step": 22780
    },
    {
      "epoch": 1.2756747676111546,
      "grad_norm": 1.2355024814605713,
      "learning_rate": 9.927105263157895e-05,
      "loss": 0.3656,
      "step": 22781
    },
    {
      "epoch": 1.2757307649232836,
      "grad_norm": 1.5421550273895264,
      "learning_rate": 9.927078947368421e-05,
      "loss": 0.395,
      "step": 22782
    },
    {
      "epoch": 1.2757867622354127,
      "grad_norm": 1.3215744495391846,
      "learning_rate": 9.927052631578948e-05,
      "loss": 0.3083,
      "step": 22783
    },
    {
      "epoch": 1.2758427595475417,
      "grad_norm": 1.0425972938537598,
      "learning_rate": 9.927026315789474e-05,
      "loss": 0.3134,
      "step": 22784
    },
    {
      "epoch": 1.2758987568596707,
      "grad_norm": 1.2993683815002441,
      "learning_rate": 9.927e-05,
      "loss": 0.4841,
      "step": 22785
    },
    {
      "epoch": 1.2759547541717997,
      "grad_norm": 1.6722605228424072,
      "learning_rate": 9.926973684210526e-05,
      "loss": 0.6632,
      "step": 22786
    },
    {
      "epoch": 1.2760107514839287,
      "grad_norm": 1.634016513824463,
      "learning_rate": 9.926947368421054e-05,
      "loss": 0.4314,
      "step": 22787
    },
    {
      "epoch": 1.2760667487960577,
      "grad_norm": 1.4416072368621826,
      "learning_rate": 9.92692105263158e-05,
      "loss": 0.4843,
      "step": 22788
    },
    {
      "epoch": 1.2761227461081868,
      "grad_norm": 1.86212158203125,
      "learning_rate": 9.926894736842106e-05,
      "loss": 0.3136,
      "step": 22789
    },
    {
      "epoch": 1.2761787434203158,
      "grad_norm": 1.419284701347351,
      "learning_rate": 9.926868421052632e-05,
      "loss": 0.4174,
      "step": 22790
    },
    {
      "epoch": 1.2762347407324448,
      "grad_norm": 1.5763509273529053,
      "learning_rate": 9.926842105263159e-05,
      "loss": 0.4896,
      "step": 22791
    },
    {
      "epoch": 1.2762907380445738,
      "grad_norm": 1.4166450500488281,
      "learning_rate": 9.926815789473685e-05,
      "loss": 0.5477,
      "step": 22792
    },
    {
      "epoch": 1.2763467353567028,
      "grad_norm": 1.4179977178573608,
      "learning_rate": 9.926789473684212e-05,
      "loss": 0.5523,
      "step": 22793
    },
    {
      "epoch": 1.2764027326688319,
      "grad_norm": 1.2980576753616333,
      "learning_rate": 9.926763157894737e-05,
      "loss": 0.4405,
      "step": 22794
    },
    {
      "epoch": 1.2764587299809609,
      "grad_norm": 1.8971774578094482,
      "learning_rate": 9.926736842105263e-05,
      "loss": 0.3989,
      "step": 22795
    },
    {
      "epoch": 1.27651472729309,
      "grad_norm": 1.6122095584869385,
      "learning_rate": 9.92671052631579e-05,
      "loss": 0.4271,
      "step": 22796
    },
    {
      "epoch": 1.276570724605219,
      "grad_norm": 1.5946292877197266,
      "learning_rate": 9.926684210526316e-05,
      "loss": 0.6605,
      "step": 22797
    },
    {
      "epoch": 1.276626721917348,
      "grad_norm": 1.4192845821380615,
      "learning_rate": 9.926657894736842e-05,
      "loss": 0.4882,
      "step": 22798
    },
    {
      "epoch": 1.276682719229477,
      "grad_norm": 1.4047622680664062,
      "learning_rate": 9.926631578947368e-05,
      "loss": 0.5338,
      "step": 22799
    },
    {
      "epoch": 1.276738716541606,
      "grad_norm": 2.6966567039489746,
      "learning_rate": 9.926605263157895e-05,
      "loss": 0.4442,
      "step": 22800
    },
    {
      "epoch": 1.276794713853735,
      "grad_norm": 1.3931602239608765,
      "learning_rate": 9.926578947368421e-05,
      "loss": 0.3877,
      "step": 22801
    },
    {
      "epoch": 1.276850711165864,
      "grad_norm": 1.3101481199264526,
      "learning_rate": 9.926552631578949e-05,
      "loss": 0.426,
      "step": 22802
    },
    {
      "epoch": 1.276906708477993,
      "grad_norm": 1.3449358940124512,
      "learning_rate": 9.926526315789473e-05,
      "loss": 0.3264,
      "step": 22803
    },
    {
      "epoch": 1.276962705790122,
      "grad_norm": 1.4406251907348633,
      "learning_rate": 9.9265e-05,
      "loss": 0.4213,
      "step": 22804
    },
    {
      "epoch": 1.277018703102251,
      "grad_norm": 1.2217750549316406,
      "learning_rate": 9.926473684210527e-05,
      "loss": 0.527,
      "step": 22805
    },
    {
      "epoch": 1.27707470041438,
      "grad_norm": 1.2291377782821655,
      "learning_rate": 9.926447368421054e-05,
      "loss": 0.5167,
      "step": 22806
    },
    {
      "epoch": 1.2771306977265091,
      "grad_norm": 1.5686665773391724,
      "learning_rate": 9.92642105263158e-05,
      "loss": 0.4669,
      "step": 22807
    },
    {
      "epoch": 1.2771866950386381,
      "grad_norm": 1.1758569478988647,
      "learning_rate": 9.926394736842106e-05,
      "loss": 0.3745,
      "step": 22808
    },
    {
      "epoch": 1.2772426923507672,
      "grad_norm": 1.3279191255569458,
      "learning_rate": 9.926368421052632e-05,
      "loss": 0.5415,
      "step": 22809
    },
    {
      "epoch": 1.2772986896628962,
      "grad_norm": 1.224311351776123,
      "learning_rate": 9.926342105263159e-05,
      "loss": 0.4282,
      "step": 22810
    },
    {
      "epoch": 1.2773546869750252,
      "grad_norm": 1.3506956100463867,
      "learning_rate": 9.926315789473685e-05,
      "loss": 0.4217,
      "step": 22811
    },
    {
      "epoch": 1.2774106842871542,
      "grad_norm": 1.8883819580078125,
      "learning_rate": 9.926289473684211e-05,
      "loss": 0.4152,
      "step": 22812
    },
    {
      "epoch": 1.2774666815992832,
      "grad_norm": 1.3244603872299194,
      "learning_rate": 9.926263157894737e-05,
      "loss": 0.3734,
      "step": 22813
    },
    {
      "epoch": 1.2775226789114122,
      "grad_norm": 1.224166989326477,
      "learning_rate": 9.926236842105263e-05,
      "loss": 0.3873,
      "step": 22814
    },
    {
      "epoch": 1.2775786762235413,
      "grad_norm": 1.5324288606643677,
      "learning_rate": 9.92621052631579e-05,
      "loss": 0.5001,
      "step": 22815
    },
    {
      "epoch": 1.2776346735356703,
      "grad_norm": 1.5295687913894653,
      "learning_rate": 9.926184210526316e-05,
      "loss": 0.4734,
      "step": 22816
    },
    {
      "epoch": 1.2776906708477993,
      "grad_norm": 1.729920744895935,
      "learning_rate": 9.926157894736842e-05,
      "loss": 0.5655,
      "step": 22817
    },
    {
      "epoch": 1.2777466681599283,
      "grad_norm": 1.1620714664459229,
      "learning_rate": 9.926131578947368e-05,
      "loss": 0.3914,
      "step": 22818
    },
    {
      "epoch": 1.2778026654720573,
      "grad_norm": 1.2352561950683594,
      "learning_rate": 9.926105263157896e-05,
      "loss": 0.4323,
      "step": 22819
    },
    {
      "epoch": 1.2778586627841864,
      "grad_norm": 1.185273289680481,
      "learning_rate": 9.926078947368422e-05,
      "loss": 0.3784,
      "step": 22820
    },
    {
      "epoch": 1.2779146600963154,
      "grad_norm": 1.7075848579406738,
      "learning_rate": 9.926052631578948e-05,
      "loss": 0.4938,
      "step": 22821
    },
    {
      "epoch": 1.2779706574084444,
      "grad_norm": 1.936911940574646,
      "learning_rate": 9.926026315789474e-05,
      "loss": 0.5181,
      "step": 22822
    },
    {
      "epoch": 1.2780266547205734,
      "grad_norm": 1.4712456464767456,
      "learning_rate": 9.926000000000001e-05,
      "loss": 0.482,
      "step": 22823
    },
    {
      "epoch": 1.2780826520327024,
      "grad_norm": 1.2620044946670532,
      "learning_rate": 9.925973684210527e-05,
      "loss": 0.3792,
      "step": 22824
    },
    {
      "epoch": 1.2781386493448315,
      "grad_norm": 1.149962067604065,
      "learning_rate": 9.925947368421054e-05,
      "loss": 0.413,
      "step": 22825
    },
    {
      "epoch": 1.2781946466569605,
      "grad_norm": 1.4623929262161255,
      "learning_rate": 9.925921052631579e-05,
      "loss": 0.4461,
      "step": 22826
    },
    {
      "epoch": 1.2782506439690895,
      "grad_norm": 1.2333474159240723,
      "learning_rate": 9.925894736842106e-05,
      "loss": 0.5105,
      "step": 22827
    },
    {
      "epoch": 1.2783066412812185,
      "grad_norm": 1.1953965425491333,
      "learning_rate": 9.925868421052632e-05,
      "loss": 0.3357,
      "step": 22828
    },
    {
      "epoch": 1.2783626385933475,
      "grad_norm": 1.5779125690460205,
      "learning_rate": 9.925842105263158e-05,
      "loss": 0.4918,
      "step": 22829
    },
    {
      "epoch": 1.2784186359054766,
      "grad_norm": 1.3370437622070312,
      "learning_rate": 9.925815789473685e-05,
      "loss": 0.441,
      "step": 22830
    },
    {
      "epoch": 1.2784746332176056,
      "grad_norm": 1.671709418296814,
      "learning_rate": 9.92578947368421e-05,
      "loss": 0.5659,
      "step": 22831
    },
    {
      "epoch": 1.2785306305297346,
      "grad_norm": 1.4015928506851196,
      "learning_rate": 9.925763157894737e-05,
      "loss": 0.3417,
      "step": 22832
    },
    {
      "epoch": 1.2785866278418636,
      "grad_norm": 1.3909051418304443,
      "learning_rate": 9.925736842105263e-05,
      "loss": 0.4817,
      "step": 22833
    },
    {
      "epoch": 1.2786426251539926,
      "grad_norm": 1.496991515159607,
      "learning_rate": 9.92571052631579e-05,
      "loss": 0.4239,
      "step": 22834
    },
    {
      "epoch": 1.2786986224661216,
      "grad_norm": 1.3109850883483887,
      "learning_rate": 9.925684210526317e-05,
      "loss": 0.4174,
      "step": 22835
    },
    {
      "epoch": 1.2787546197782507,
      "grad_norm": 1.5020232200622559,
      "learning_rate": 9.925657894736843e-05,
      "loss": 0.3756,
      "step": 22836
    },
    {
      "epoch": 1.2788106170903797,
      "grad_norm": 1.3662246465682983,
      "learning_rate": 9.925631578947369e-05,
      "loss": 0.3375,
      "step": 22837
    },
    {
      "epoch": 1.2788666144025087,
      "grad_norm": 1.225379228591919,
      "learning_rate": 9.925605263157896e-05,
      "loss": 0.3597,
      "step": 22838
    },
    {
      "epoch": 1.2789226117146377,
      "grad_norm": 1.8974720239639282,
      "learning_rate": 9.925578947368422e-05,
      "loss": 0.455,
      "step": 22839
    },
    {
      "epoch": 1.2789786090267667,
      "grad_norm": 1.445088267326355,
      "learning_rate": 9.925552631578948e-05,
      "loss": 0.3239,
      "step": 22840
    },
    {
      "epoch": 1.2790346063388958,
      "grad_norm": 1.2902582883834839,
      "learning_rate": 9.925526315789474e-05,
      "loss": 0.595,
      "step": 22841
    },
    {
      "epoch": 1.2790906036510248,
      "grad_norm": 1.2275384664535522,
      "learning_rate": 9.925500000000001e-05,
      "loss": 0.3344,
      "step": 22842
    },
    {
      "epoch": 1.2791466009631538,
      "grad_norm": 1.4865615367889404,
      "learning_rate": 9.925473684210527e-05,
      "loss": 0.5262,
      "step": 22843
    },
    {
      "epoch": 1.2792025982752828,
      "grad_norm": 1.4789400100708008,
      "learning_rate": 9.925447368421053e-05,
      "loss": 0.4692,
      "step": 22844
    },
    {
      "epoch": 1.2792585955874118,
      "grad_norm": 1.9096077680587769,
      "learning_rate": 9.925421052631579e-05,
      "loss": 0.4197,
      "step": 22845
    },
    {
      "epoch": 1.2793145928995409,
      "grad_norm": 9.500982284545898,
      "learning_rate": 9.925394736842105e-05,
      "loss": 0.3817,
      "step": 22846
    },
    {
      "epoch": 1.2793705902116699,
      "grad_norm": 1.4734511375427246,
      "learning_rate": 9.925368421052632e-05,
      "loss": 0.4654,
      "step": 22847
    },
    {
      "epoch": 1.279426587523799,
      "grad_norm": 1.6119033098220825,
      "learning_rate": 9.925342105263158e-05,
      "loss": 0.4628,
      "step": 22848
    },
    {
      "epoch": 1.279482584835928,
      "grad_norm": 1.4294317960739136,
      "learning_rate": 9.925315789473684e-05,
      "loss": 0.4772,
      "step": 22849
    },
    {
      "epoch": 1.279538582148057,
      "grad_norm": 1.3593608140945435,
      "learning_rate": 9.92528947368421e-05,
      "loss": 0.478,
      "step": 22850
    },
    {
      "epoch": 1.279594579460186,
      "grad_norm": 1.2838557958602905,
      "learning_rate": 9.925263157894738e-05,
      "loss": 0.4266,
      "step": 22851
    },
    {
      "epoch": 1.279650576772315,
      "grad_norm": 1.3299751281738281,
      "learning_rate": 9.925236842105264e-05,
      "loss": 0.4981,
      "step": 22852
    },
    {
      "epoch": 1.279706574084444,
      "grad_norm": 1.2452366352081299,
      "learning_rate": 9.92521052631579e-05,
      "loss": 0.5585,
      "step": 22853
    },
    {
      "epoch": 1.279762571396573,
      "grad_norm": 1.288326621055603,
      "learning_rate": 9.925184210526315e-05,
      "loss": 0.4661,
      "step": 22854
    },
    {
      "epoch": 1.279818568708702,
      "grad_norm": 1.3746986389160156,
      "learning_rate": 9.925157894736843e-05,
      "loss": 0.4094,
      "step": 22855
    },
    {
      "epoch": 1.279874566020831,
      "grad_norm": 1.115032434463501,
      "learning_rate": 9.925131578947369e-05,
      "loss": 0.4007,
      "step": 22856
    },
    {
      "epoch": 1.27993056333296,
      "grad_norm": 1.2488641738891602,
      "learning_rate": 9.925105263157896e-05,
      "loss": 0.4994,
      "step": 22857
    },
    {
      "epoch": 1.279986560645089,
      "grad_norm": 1.425959825515747,
      "learning_rate": 9.925078947368421e-05,
      "loss": 0.4653,
      "step": 22858
    },
    {
      "epoch": 1.280042557957218,
      "grad_norm": 1.4689654111862183,
      "learning_rate": 9.925052631578948e-05,
      "loss": 0.5289,
      "step": 22859
    },
    {
      "epoch": 1.2800985552693471,
      "grad_norm": 1.4871768951416016,
      "learning_rate": 9.925026315789474e-05,
      "loss": 0.4693,
      "step": 22860
    },
    {
      "epoch": 1.2801545525814761,
      "grad_norm": 1.5057939291000366,
      "learning_rate": 9.925000000000001e-05,
      "loss": 0.5627,
      "step": 22861
    },
    {
      "epoch": 1.2802105498936052,
      "grad_norm": 1.2794889211654663,
      "learning_rate": 9.924973684210527e-05,
      "loss": 0.4061,
      "step": 22862
    },
    {
      "epoch": 1.2802665472057342,
      "grad_norm": 1.3119163513183594,
      "learning_rate": 9.924947368421053e-05,
      "loss": 0.3685,
      "step": 22863
    },
    {
      "epoch": 1.2803225445178632,
      "grad_norm": 1.499916672706604,
      "learning_rate": 9.924921052631579e-05,
      "loss": 0.4594,
      "step": 22864
    },
    {
      "epoch": 1.2803785418299922,
      "grad_norm": 1.4060181379318237,
      "learning_rate": 9.924894736842105e-05,
      "loss": 0.3679,
      "step": 22865
    },
    {
      "epoch": 1.2804345391421212,
      "grad_norm": 1.4560726881027222,
      "learning_rate": 9.924868421052633e-05,
      "loss": 0.4411,
      "step": 22866
    },
    {
      "epoch": 1.2804905364542503,
      "grad_norm": 1.4118001461029053,
      "learning_rate": 9.924842105263159e-05,
      "loss": 0.4929,
      "step": 22867
    },
    {
      "epoch": 1.2805465337663793,
      "grad_norm": 1.429483413696289,
      "learning_rate": 9.924815789473685e-05,
      "loss": 0.6687,
      "step": 22868
    },
    {
      "epoch": 1.2806025310785083,
      "grad_norm": 1.3398168087005615,
      "learning_rate": 9.92478947368421e-05,
      "loss": 0.4995,
      "step": 22869
    },
    {
      "epoch": 1.2806585283906373,
      "grad_norm": 1.3888211250305176,
      "learning_rate": 9.924763157894738e-05,
      "loss": 0.4382,
      "step": 22870
    },
    {
      "epoch": 1.2807145257027663,
      "grad_norm": 1.420980453491211,
      "learning_rate": 9.924736842105264e-05,
      "loss": 0.5059,
      "step": 22871
    },
    {
      "epoch": 1.2807705230148954,
      "grad_norm": 1.3897972106933594,
      "learning_rate": 9.92471052631579e-05,
      "loss": 0.5511,
      "step": 22872
    },
    {
      "epoch": 1.2808265203270244,
      "grad_norm": 1.4518322944641113,
      "learning_rate": 9.924684210526316e-05,
      "loss": 0.5084,
      "step": 22873
    },
    {
      "epoch": 1.2808825176391534,
      "grad_norm": 1.3902095556259155,
      "learning_rate": 9.924657894736843e-05,
      "loss": 0.7499,
      "step": 22874
    },
    {
      "epoch": 1.2809385149512824,
      "grad_norm": 1.311393141746521,
      "learning_rate": 9.924631578947369e-05,
      "loss": 0.4051,
      "step": 22875
    },
    {
      "epoch": 1.2809945122634114,
      "grad_norm": 1.9083179235458374,
      "learning_rate": 9.924605263157895e-05,
      "loss": 0.4966,
      "step": 22876
    },
    {
      "epoch": 1.2810505095755405,
      "grad_norm": 1.2876859903335571,
      "learning_rate": 9.924578947368421e-05,
      "loss": 0.4092,
      "step": 22877
    },
    {
      "epoch": 1.2811065068876695,
      "grad_norm": 1.4630753993988037,
      "learning_rate": 9.924552631578948e-05,
      "loss": 0.5745,
      "step": 22878
    },
    {
      "epoch": 1.2811625041997985,
      "grad_norm": 1.289911150932312,
      "learning_rate": 9.924526315789474e-05,
      "loss": 0.4048,
      "step": 22879
    },
    {
      "epoch": 1.2812185015119275,
      "grad_norm": 1.2508763074874878,
      "learning_rate": 9.924500000000002e-05,
      "loss": 0.4841,
      "step": 22880
    },
    {
      "epoch": 1.2812744988240565,
      "grad_norm": 1.3895246982574463,
      "learning_rate": 9.924473684210526e-05,
      "loss": 0.4195,
      "step": 22881
    },
    {
      "epoch": 1.2813304961361855,
      "grad_norm": 1.732861876487732,
      "learning_rate": 9.924447368421052e-05,
      "loss": 0.4791,
      "step": 22882
    },
    {
      "epoch": 1.2813864934483146,
      "grad_norm": 1.433110237121582,
      "learning_rate": 9.92442105263158e-05,
      "loss": 0.4858,
      "step": 22883
    },
    {
      "epoch": 1.2814424907604436,
      "grad_norm": 1.2530598640441895,
      "learning_rate": 9.924394736842106e-05,
      "loss": 0.4106,
      "step": 22884
    },
    {
      "epoch": 1.2814984880725726,
      "grad_norm": 1.45210599899292,
      "learning_rate": 9.924368421052633e-05,
      "loss": 0.4895,
      "step": 22885
    },
    {
      "epoch": 1.2815544853847016,
      "grad_norm": 1.2971900701522827,
      "learning_rate": 9.924342105263157e-05,
      "loss": 0.4203,
      "step": 22886
    },
    {
      "epoch": 1.2816104826968306,
      "grad_norm": 1.337934136390686,
      "learning_rate": 9.924315789473685e-05,
      "loss": 0.438,
      "step": 22887
    },
    {
      "epoch": 1.2816664800089597,
      "grad_norm": 1.5100983381271362,
      "learning_rate": 9.924289473684211e-05,
      "loss": 0.4976,
      "step": 22888
    },
    {
      "epoch": 1.2817224773210887,
      "grad_norm": 1.305716872215271,
      "learning_rate": 9.924263157894738e-05,
      "loss": 0.5341,
      "step": 22889
    },
    {
      "epoch": 1.2817784746332177,
      "grad_norm": 1.6484627723693848,
      "learning_rate": 9.924236842105264e-05,
      "loss": 0.5838,
      "step": 22890
    },
    {
      "epoch": 1.2818344719453467,
      "grad_norm": 1.4973548650741577,
      "learning_rate": 9.92421052631579e-05,
      "loss": 0.4746,
      "step": 22891
    },
    {
      "epoch": 1.2818904692574757,
      "grad_norm": 1.2990798950195312,
      "learning_rate": 9.924184210526316e-05,
      "loss": 0.3971,
      "step": 22892
    },
    {
      "epoch": 1.2819464665696048,
      "grad_norm": 1.5169073343276978,
      "learning_rate": 9.924157894736843e-05,
      "loss": 0.5484,
      "step": 22893
    },
    {
      "epoch": 1.2820024638817338,
      "grad_norm": 1.4381048679351807,
      "learning_rate": 9.924131578947369e-05,
      "loss": 0.4447,
      "step": 22894
    },
    {
      "epoch": 1.2820584611938628,
      "grad_norm": 1.9256312847137451,
      "learning_rate": 9.924105263157895e-05,
      "loss": 0.3603,
      "step": 22895
    },
    {
      "epoch": 1.2821144585059918,
      "grad_norm": 1.35012686252594,
      "learning_rate": 9.924078947368421e-05,
      "loss": 0.3953,
      "step": 22896
    },
    {
      "epoch": 1.2821704558181208,
      "grad_norm": 1.3643207550048828,
      "learning_rate": 9.924052631578949e-05,
      "loss": 0.4825,
      "step": 22897
    },
    {
      "epoch": 1.2822264531302499,
      "grad_norm": 1.6131207942962646,
      "learning_rate": 9.924026315789475e-05,
      "loss": 0.5158,
      "step": 22898
    },
    {
      "epoch": 1.2822824504423789,
      "grad_norm": 1.4574543237686157,
      "learning_rate": 9.924e-05,
      "loss": 0.485,
      "step": 22899
    },
    {
      "epoch": 1.282338447754508,
      "grad_norm": 1.2956045866012573,
      "learning_rate": 9.923973684210527e-05,
      "loss": 0.4301,
      "step": 22900
    },
    {
      "epoch": 1.282394445066637,
      "grad_norm": 1.2448629140853882,
      "learning_rate": 9.923947368421052e-05,
      "loss": 0.4139,
      "step": 22901
    },
    {
      "epoch": 1.282450442378766,
      "grad_norm": 1.3458913564682007,
      "learning_rate": 9.92392105263158e-05,
      "loss": 0.5027,
      "step": 22902
    },
    {
      "epoch": 1.282506439690895,
      "grad_norm": 1.2780959606170654,
      "learning_rate": 9.923894736842106e-05,
      "loss": 0.348,
      "step": 22903
    },
    {
      "epoch": 1.282562437003024,
      "grad_norm": 1.369726300239563,
      "learning_rate": 9.923868421052632e-05,
      "loss": 0.4936,
      "step": 22904
    },
    {
      "epoch": 1.282618434315153,
      "grad_norm": 1.415688395500183,
      "learning_rate": 9.923842105263158e-05,
      "loss": 0.4766,
      "step": 22905
    },
    {
      "epoch": 1.282674431627282,
      "grad_norm": 1.3793411254882812,
      "learning_rate": 9.923815789473685e-05,
      "loss": 0.419,
      "step": 22906
    },
    {
      "epoch": 1.282730428939411,
      "grad_norm": 1.4151756763458252,
      "learning_rate": 9.923789473684211e-05,
      "loss": 0.5377,
      "step": 22907
    },
    {
      "epoch": 1.28278642625154,
      "grad_norm": 1.4490402936935425,
      "learning_rate": 9.923763157894737e-05,
      "loss": 0.4964,
      "step": 22908
    },
    {
      "epoch": 1.282842423563669,
      "grad_norm": 1.4372507333755493,
      "learning_rate": 9.923736842105263e-05,
      "loss": 0.455,
      "step": 22909
    },
    {
      "epoch": 1.282898420875798,
      "grad_norm": 1.2414966821670532,
      "learning_rate": 9.92371052631579e-05,
      "loss": 0.438,
      "step": 22910
    },
    {
      "epoch": 1.282954418187927,
      "grad_norm": 1.3037598133087158,
      "learning_rate": 9.923684210526316e-05,
      "loss": 0.4289,
      "step": 22911
    },
    {
      "epoch": 1.2830104155000561,
      "grad_norm": 2.441774368286133,
      "learning_rate": 9.923657894736844e-05,
      "loss": 0.486,
      "step": 22912
    },
    {
      "epoch": 1.2830664128121851,
      "grad_norm": 1.678240418434143,
      "learning_rate": 9.923631578947368e-05,
      "loss": 0.4303,
      "step": 22913
    },
    {
      "epoch": 1.2831224101243142,
      "grad_norm": 1.2907674312591553,
      "learning_rate": 9.923605263157896e-05,
      "loss": 0.4251,
      "step": 22914
    },
    {
      "epoch": 1.2831784074364432,
      "grad_norm": 1.5444412231445312,
      "learning_rate": 9.923578947368422e-05,
      "loss": 0.5689,
      "step": 22915
    },
    {
      "epoch": 1.2832344047485722,
      "grad_norm": 1.678706169128418,
      "learning_rate": 9.923552631578947e-05,
      "loss": 0.5201,
      "step": 22916
    },
    {
      "epoch": 1.2832904020607012,
      "grad_norm": 1.3194148540496826,
      "learning_rate": 9.923526315789475e-05,
      "loss": 0.3818,
      "step": 22917
    },
    {
      "epoch": 1.2833463993728302,
      "grad_norm": 1.344393014907837,
      "learning_rate": 9.9235e-05,
      "loss": 0.4487,
      "step": 22918
    },
    {
      "epoch": 1.2834023966849593,
      "grad_norm": 1.3960899114608765,
      "learning_rate": 9.923473684210527e-05,
      "loss": 0.4525,
      "step": 22919
    },
    {
      "epoch": 1.2834583939970883,
      "grad_norm": 1.4669640064239502,
      "learning_rate": 9.923447368421053e-05,
      "loss": 0.3751,
      "step": 22920
    },
    {
      "epoch": 1.2835143913092173,
      "grad_norm": 1.2237739562988281,
      "learning_rate": 9.92342105263158e-05,
      "loss": 0.4054,
      "step": 22921
    },
    {
      "epoch": 1.283570388621346,
      "grad_norm": 1.3809826374053955,
      "learning_rate": 9.923394736842106e-05,
      "loss": 0.405,
      "step": 22922
    },
    {
      "epoch": 1.2836263859334751,
      "grad_norm": 1.3056340217590332,
      "learning_rate": 9.923368421052632e-05,
      "loss": 0.4087,
      "step": 22923
    },
    {
      "epoch": 1.2836823832456041,
      "grad_norm": 1.3003950119018555,
      "learning_rate": 9.923342105263158e-05,
      "loss": 0.5366,
      "step": 22924
    },
    {
      "epoch": 1.2837383805577331,
      "grad_norm": 1.3883391618728638,
      "learning_rate": 9.923315789473685e-05,
      "loss": 0.4015,
      "step": 22925
    },
    {
      "epoch": 1.2837943778698622,
      "grad_norm": 1.3687617778778076,
      "learning_rate": 9.923289473684211e-05,
      "loss": 0.5419,
      "step": 22926
    },
    {
      "epoch": 1.2838503751819912,
      "grad_norm": 1.247031569480896,
      "learning_rate": 9.923263157894737e-05,
      "loss": 0.4375,
      "step": 22927
    },
    {
      "epoch": 1.2839063724941202,
      "grad_norm": 1.4073357582092285,
      "learning_rate": 9.923236842105263e-05,
      "loss": 0.4831,
      "step": 22928
    },
    {
      "epoch": 1.2839623698062492,
      "grad_norm": 1.2339050769805908,
      "learning_rate": 9.92321052631579e-05,
      "loss": 0.4407,
      "step": 22929
    },
    {
      "epoch": 1.2840183671183782,
      "grad_norm": 1.2139928340911865,
      "learning_rate": 9.923184210526317e-05,
      "loss": 0.4936,
      "step": 22930
    },
    {
      "epoch": 1.2840743644305073,
      "grad_norm": 1.5380933284759521,
      "learning_rate": 9.923157894736842e-05,
      "loss": 0.4634,
      "step": 22931
    },
    {
      "epoch": 1.2841303617426363,
      "grad_norm": 1.1507809162139893,
      "learning_rate": 9.923131578947368e-05,
      "loss": 0.434,
      "step": 22932
    },
    {
      "epoch": 1.2841863590547653,
      "grad_norm": 1.1982436180114746,
      "learning_rate": 9.923105263157894e-05,
      "loss": 0.3755,
      "step": 22933
    },
    {
      "epoch": 1.2842423563668943,
      "grad_norm": 1.8436267375946045,
      "learning_rate": 9.923078947368422e-05,
      "loss": 0.4652,
      "step": 22934
    },
    {
      "epoch": 1.2842983536790233,
      "grad_norm": 2.4296042919158936,
      "learning_rate": 9.923052631578948e-05,
      "loss": 0.7161,
      "step": 22935
    },
    {
      "epoch": 1.2843543509911524,
      "grad_norm": 1.224182367324829,
      "learning_rate": 9.923026315789474e-05,
      "loss": 0.5098,
      "step": 22936
    },
    {
      "epoch": 1.2844103483032814,
      "grad_norm": 1.4148139953613281,
      "learning_rate": 9.923e-05,
      "loss": 0.4536,
      "step": 22937
    },
    {
      "epoch": 1.2844663456154104,
      "grad_norm": 1.2621914148330688,
      "learning_rate": 9.922973684210527e-05,
      "loss": 0.4855,
      "step": 22938
    },
    {
      "epoch": 1.2845223429275394,
      "grad_norm": 1.5679540634155273,
      "learning_rate": 9.922947368421053e-05,
      "loss": 0.5845,
      "step": 22939
    },
    {
      "epoch": 1.2845783402396684,
      "grad_norm": 1.6275293827056885,
      "learning_rate": 9.92292105263158e-05,
      "loss": 0.4481,
      "step": 22940
    },
    {
      "epoch": 1.2846343375517975,
      "grad_norm": 1.2808688879013062,
      "learning_rate": 9.922894736842105e-05,
      "loss": 0.4123,
      "step": 22941
    },
    {
      "epoch": 1.2846903348639265,
      "grad_norm": 1.4030125141143799,
      "learning_rate": 9.922868421052632e-05,
      "loss": 0.5084,
      "step": 22942
    },
    {
      "epoch": 1.2847463321760555,
      "grad_norm": 1.191109299659729,
      "learning_rate": 9.922842105263158e-05,
      "loss": 0.4447,
      "step": 22943
    },
    {
      "epoch": 1.2848023294881845,
      "grad_norm": 1.2105380296707153,
      "learning_rate": 9.922815789473686e-05,
      "loss": 0.4081,
      "step": 22944
    },
    {
      "epoch": 1.2848583268003135,
      "grad_norm": 1.7374064922332764,
      "learning_rate": 9.92278947368421e-05,
      "loss": 0.4894,
      "step": 22945
    },
    {
      "epoch": 1.2849143241124426,
      "grad_norm": 1.5357162952423096,
      "learning_rate": 9.922763157894738e-05,
      "loss": 0.4604,
      "step": 22946
    },
    {
      "epoch": 1.2849703214245716,
      "grad_norm": 1.4711189270019531,
      "learning_rate": 9.922736842105263e-05,
      "loss": 0.523,
      "step": 22947
    },
    {
      "epoch": 1.2850263187367006,
      "grad_norm": 1.3166229724884033,
      "learning_rate": 9.922710526315791e-05,
      "loss": 0.4023,
      "step": 22948
    },
    {
      "epoch": 1.2850823160488296,
      "grad_norm": 1.3460315465927124,
      "learning_rate": 9.922684210526317e-05,
      "loss": 0.4786,
      "step": 22949
    },
    {
      "epoch": 1.2851383133609586,
      "grad_norm": 1.4328230619430542,
      "learning_rate": 9.922657894736841e-05,
      "loss": 0.6095,
      "step": 22950
    },
    {
      "epoch": 1.2851943106730876,
      "grad_norm": 1.4445209503173828,
      "learning_rate": 9.922631578947369e-05,
      "loss": 0.4819,
      "step": 22951
    },
    {
      "epoch": 1.2852503079852167,
      "grad_norm": 1.5904972553253174,
      "learning_rate": 9.922605263157895e-05,
      "loss": 0.4077,
      "step": 22952
    },
    {
      "epoch": 1.2853063052973457,
      "grad_norm": 1.2449522018432617,
      "learning_rate": 9.922578947368422e-05,
      "loss": 0.4679,
      "step": 22953
    },
    {
      "epoch": 1.2853623026094747,
      "grad_norm": 1.2776936292648315,
      "learning_rate": 9.922552631578948e-05,
      "loss": 0.4167,
      "step": 22954
    },
    {
      "epoch": 1.2854182999216037,
      "grad_norm": 5.033515930175781,
      "learning_rate": 9.922526315789474e-05,
      "loss": 0.4469,
      "step": 22955
    },
    {
      "epoch": 1.2854742972337327,
      "grad_norm": 1.3768417835235596,
      "learning_rate": 9.9225e-05,
      "loss": 0.4248,
      "step": 22956
    },
    {
      "epoch": 1.2855302945458618,
      "grad_norm": 1.1071889400482178,
      "learning_rate": 9.922473684210527e-05,
      "loss": 0.3615,
      "step": 22957
    },
    {
      "epoch": 1.2855862918579908,
      "grad_norm": 1.2394598722457886,
      "learning_rate": 9.922447368421053e-05,
      "loss": 0.4888,
      "step": 22958
    },
    {
      "epoch": 1.2856422891701198,
      "grad_norm": 1.2708501815795898,
      "learning_rate": 9.922421052631579e-05,
      "loss": 0.4206,
      "step": 22959
    },
    {
      "epoch": 1.2856982864822488,
      "grad_norm": 1.3693835735321045,
      "learning_rate": 9.922394736842105e-05,
      "loss": 0.4744,
      "step": 22960
    },
    {
      "epoch": 1.2857542837943778,
      "grad_norm": 1.2162554264068604,
      "learning_rate": 9.922368421052633e-05,
      "loss": 0.317,
      "step": 22961
    },
    {
      "epoch": 1.2858102811065069,
      "grad_norm": 1.262914776802063,
      "learning_rate": 9.922342105263158e-05,
      "loss": 0.4,
      "step": 22962
    },
    {
      "epoch": 1.2858662784186359,
      "grad_norm": 1.3579424619674683,
      "learning_rate": 9.922315789473684e-05,
      "loss": 0.4434,
      "step": 22963
    },
    {
      "epoch": 1.285922275730765,
      "grad_norm": 1.2783117294311523,
      "learning_rate": 9.92228947368421e-05,
      "loss": 0.398,
      "step": 22964
    },
    {
      "epoch": 1.285978273042894,
      "grad_norm": 1.3735289573669434,
      "learning_rate": 9.922263157894738e-05,
      "loss": 0.4491,
      "step": 22965
    },
    {
      "epoch": 1.286034270355023,
      "grad_norm": 1.716067910194397,
      "learning_rate": 9.922236842105264e-05,
      "loss": 0.5226,
      "step": 22966
    },
    {
      "epoch": 1.286090267667152,
      "grad_norm": 1.2945685386657715,
      "learning_rate": 9.92221052631579e-05,
      "loss": 0.4089,
      "step": 22967
    },
    {
      "epoch": 1.286146264979281,
      "grad_norm": 1.5746315717697144,
      "learning_rate": 9.922184210526316e-05,
      "loss": 0.4545,
      "step": 22968
    },
    {
      "epoch": 1.28620226229141,
      "grad_norm": 1.5032798051834106,
      "learning_rate": 9.922157894736842e-05,
      "loss": 0.4913,
      "step": 22969
    },
    {
      "epoch": 1.286258259603539,
      "grad_norm": 1.5064482688903809,
      "learning_rate": 9.922131578947369e-05,
      "loss": 0.4879,
      "step": 22970
    },
    {
      "epoch": 1.286314256915668,
      "grad_norm": 1.6576777696609497,
      "learning_rate": 9.922105263157895e-05,
      "loss": 0.445,
      "step": 22971
    },
    {
      "epoch": 1.286370254227797,
      "grad_norm": 1.1466107368469238,
      "learning_rate": 9.922078947368422e-05,
      "loss": 0.3226,
      "step": 22972
    },
    {
      "epoch": 1.286426251539926,
      "grad_norm": 1.3217601776123047,
      "learning_rate": 9.922052631578947e-05,
      "loss": 0.4444,
      "step": 22973
    },
    {
      "epoch": 1.286482248852055,
      "grad_norm": 1.2319965362548828,
      "learning_rate": 9.922026315789474e-05,
      "loss": 0.4418,
      "step": 22974
    },
    {
      "epoch": 1.286538246164184,
      "grad_norm": 1.4229353666305542,
      "learning_rate": 9.922e-05,
      "loss": 0.52,
      "step": 22975
    },
    {
      "epoch": 1.2865942434763131,
      "grad_norm": 1.6926286220550537,
      "learning_rate": 9.921973684210528e-05,
      "loss": 0.4617,
      "step": 22976
    },
    {
      "epoch": 1.2866502407884421,
      "grad_norm": 1.4075435400009155,
      "learning_rate": 9.921947368421054e-05,
      "loss": 0.4135,
      "step": 22977
    },
    {
      "epoch": 1.2867062381005712,
      "grad_norm": 1.431416630744934,
      "learning_rate": 9.92192105263158e-05,
      "loss": 0.5034,
      "step": 22978
    },
    {
      "epoch": 1.2867622354127002,
      "grad_norm": 1.2957077026367188,
      "learning_rate": 9.921894736842105e-05,
      "loss": 0.4524,
      "step": 22979
    },
    {
      "epoch": 1.2868182327248292,
      "grad_norm": 1.4212028980255127,
      "learning_rate": 9.921868421052633e-05,
      "loss": 0.4904,
      "step": 22980
    },
    {
      "epoch": 1.2868742300369582,
      "grad_norm": 1.1704585552215576,
      "learning_rate": 9.921842105263159e-05,
      "loss": 0.483,
      "step": 22981
    },
    {
      "epoch": 1.2869302273490872,
      "grad_norm": 1.2489521503448486,
      "learning_rate": 9.921815789473685e-05,
      "loss": 0.428,
      "step": 22982
    },
    {
      "epoch": 1.2869862246612163,
      "grad_norm": 1.6500221490859985,
      "learning_rate": 9.921789473684211e-05,
      "loss": 0.5857,
      "step": 22983
    },
    {
      "epoch": 1.2870422219733453,
      "grad_norm": 1.25044584274292,
      "learning_rate": 9.921763157894738e-05,
      "loss": 0.4966,
      "step": 22984
    },
    {
      "epoch": 1.2870982192854743,
      "grad_norm": 1.3077419996261597,
      "learning_rate": 9.921736842105264e-05,
      "loss": 0.4304,
      "step": 22985
    },
    {
      "epoch": 1.2871542165976033,
      "grad_norm": 1.6721961498260498,
      "learning_rate": 9.92171052631579e-05,
      "loss": 0.5613,
      "step": 22986
    },
    {
      "epoch": 1.2872102139097323,
      "grad_norm": 1.2416599988937378,
      "learning_rate": 9.921684210526316e-05,
      "loss": 0.5033,
      "step": 22987
    },
    {
      "epoch": 1.2872662112218614,
      "grad_norm": 1.3946799039840698,
      "learning_rate": 9.921657894736842e-05,
      "loss": 0.4012,
      "step": 22988
    },
    {
      "epoch": 1.2873222085339904,
      "grad_norm": 1.3088847398757935,
      "learning_rate": 9.921631578947369e-05,
      "loss": 0.3774,
      "step": 22989
    },
    {
      "epoch": 1.2873782058461194,
      "grad_norm": 1.2221100330352783,
      "learning_rate": 9.921605263157895e-05,
      "loss": 0.4021,
      "step": 22990
    },
    {
      "epoch": 1.2874342031582484,
      "grad_norm": 1.2882754802703857,
      "learning_rate": 9.921578947368421e-05,
      "loss": 0.4187,
      "step": 22991
    },
    {
      "epoch": 1.2874902004703774,
      "grad_norm": 1.244855523109436,
      "learning_rate": 9.921552631578947e-05,
      "loss": 0.4645,
      "step": 22992
    },
    {
      "epoch": 1.2875461977825065,
      "grad_norm": 1.4222966432571411,
      "learning_rate": 9.921526315789474e-05,
      "loss": 0.4661,
      "step": 22993
    },
    {
      "epoch": 1.2876021950946355,
      "grad_norm": 1.1522328853607178,
      "learning_rate": 9.9215e-05,
      "loss": 0.35,
      "step": 22994
    },
    {
      "epoch": 1.2876581924067645,
      "grad_norm": 1.5621757507324219,
      "learning_rate": 9.921473684210528e-05,
      "loss": 0.3748,
      "step": 22995
    },
    {
      "epoch": 1.2877141897188935,
      "grad_norm": 1.5290151834487915,
      "learning_rate": 9.921447368421052e-05,
      "loss": 0.5444,
      "step": 22996
    },
    {
      "epoch": 1.2877701870310225,
      "grad_norm": 1.228925347328186,
      "learning_rate": 9.92142105263158e-05,
      "loss": 0.4329,
      "step": 22997
    },
    {
      "epoch": 1.2878261843431515,
      "grad_norm": 1.1519784927368164,
      "learning_rate": 9.921394736842106e-05,
      "loss": 0.3533,
      "step": 22998
    },
    {
      "epoch": 1.2878821816552806,
      "grad_norm": 1.2597635984420776,
      "learning_rate": 9.921368421052633e-05,
      "loss": 0.5011,
      "step": 22999
    },
    {
      "epoch": 1.2879381789674096,
      "grad_norm": 1.2440520524978638,
      "learning_rate": 9.921342105263158e-05,
      "loss": 0.5224,
      "step": 23000
    },
    {
      "epoch": 1.2879941762795386,
      "grad_norm": 1.4682767391204834,
      "learning_rate": 9.921315789473685e-05,
      "loss": 0.3505,
      "step": 23001
    },
    {
      "epoch": 1.2880501735916676,
      "grad_norm": 1.5257654190063477,
      "learning_rate": 9.921289473684211e-05,
      "loss": 0.5408,
      "step": 23002
    },
    {
      "epoch": 1.2881061709037966,
      "grad_norm": 1.4215542078018188,
      "learning_rate": 9.921263157894737e-05,
      "loss": 0.466,
      "step": 23003
    },
    {
      "epoch": 1.2881621682159257,
      "grad_norm": 1.2798796892166138,
      "learning_rate": 9.921236842105264e-05,
      "loss": 0.4863,
      "step": 23004
    },
    {
      "epoch": 1.2882181655280547,
      "grad_norm": 1.1246417760849,
      "learning_rate": 9.921210526315789e-05,
      "loss": 0.4249,
      "step": 23005
    },
    {
      "epoch": 1.2882741628401837,
      "grad_norm": 1.7751320600509644,
      "learning_rate": 9.921184210526316e-05,
      "loss": 0.4136,
      "step": 23006
    },
    {
      "epoch": 1.2883301601523127,
      "grad_norm": 1.4135159254074097,
      "learning_rate": 9.921157894736842e-05,
      "loss": 0.5479,
      "step": 23007
    },
    {
      "epoch": 1.2883861574644417,
      "grad_norm": 1.7594817876815796,
      "learning_rate": 9.92113157894737e-05,
      "loss": 0.4854,
      "step": 23008
    },
    {
      "epoch": 1.2884421547765708,
      "grad_norm": 1.517059564590454,
      "learning_rate": 9.921105263157895e-05,
      "loss": 0.8628,
      "step": 23009
    },
    {
      "epoch": 1.2884981520886998,
      "grad_norm": 1.2950935363769531,
      "learning_rate": 9.921078947368421e-05,
      "loss": 0.3913,
      "step": 23010
    },
    {
      "epoch": 1.2885541494008288,
      "grad_norm": 1.353660225868225,
      "learning_rate": 9.921052631578947e-05,
      "loss": 0.4868,
      "step": 23011
    },
    {
      "epoch": 1.2886101467129578,
      "grad_norm": 0.9766411185264587,
      "learning_rate": 9.921026315789475e-05,
      "loss": 0.407,
      "step": 23012
    },
    {
      "epoch": 1.2886661440250868,
      "grad_norm": 1.3170545101165771,
      "learning_rate": 9.921000000000001e-05,
      "loss": 0.5691,
      "step": 23013
    },
    {
      "epoch": 1.2887221413372159,
      "grad_norm": 5.148563385009766,
      "learning_rate": 9.920973684210527e-05,
      "loss": 0.3936,
      "step": 23014
    },
    {
      "epoch": 1.2887781386493449,
      "grad_norm": 1.4028126001358032,
      "learning_rate": 9.920947368421053e-05,
      "loss": 0.4737,
      "step": 23015
    },
    {
      "epoch": 1.288834135961474,
      "grad_norm": 1.4964874982833862,
      "learning_rate": 9.92092105263158e-05,
      "loss": 0.5049,
      "step": 23016
    },
    {
      "epoch": 1.288890133273603,
      "grad_norm": 2.170239210128784,
      "learning_rate": 9.920894736842106e-05,
      "loss": 0.4979,
      "step": 23017
    },
    {
      "epoch": 1.288946130585732,
      "grad_norm": 1.45847487449646,
      "learning_rate": 9.920868421052632e-05,
      "loss": 0.4045,
      "step": 23018
    },
    {
      "epoch": 1.289002127897861,
      "grad_norm": 1.2943224906921387,
      "learning_rate": 9.920842105263158e-05,
      "loss": 0.3783,
      "step": 23019
    },
    {
      "epoch": 1.28905812520999,
      "grad_norm": 1.3317760229110718,
      "learning_rate": 9.920815789473684e-05,
      "loss": 0.4034,
      "step": 23020
    },
    {
      "epoch": 1.289114122522119,
      "grad_norm": 1.1763025522232056,
      "learning_rate": 9.920789473684211e-05,
      "loss": 0.4745,
      "step": 23021
    },
    {
      "epoch": 1.289170119834248,
      "grad_norm": 1.17929208278656,
      "learning_rate": 9.920763157894737e-05,
      "loss": 0.4017,
      "step": 23022
    },
    {
      "epoch": 1.289226117146377,
      "grad_norm": 1.2791979312896729,
      "learning_rate": 9.920736842105263e-05,
      "loss": 0.4489,
      "step": 23023
    },
    {
      "epoch": 1.289282114458506,
      "grad_norm": 4.220419406890869,
      "learning_rate": 9.920710526315789e-05,
      "loss": 0.3953,
      "step": 23024
    },
    {
      "epoch": 1.289338111770635,
      "grad_norm": 1.5760184526443481,
      "learning_rate": 9.920684210526316e-05,
      "loss": 0.6107,
      "step": 23025
    },
    {
      "epoch": 1.289394109082764,
      "grad_norm": 1.4542875289916992,
      "learning_rate": 9.920657894736842e-05,
      "loss": 0.5295,
      "step": 23026
    },
    {
      "epoch": 1.289450106394893,
      "grad_norm": 1.0801390409469604,
      "learning_rate": 9.92063157894737e-05,
      "loss": 0.4279,
      "step": 23027
    },
    {
      "epoch": 1.2895061037070221,
      "grad_norm": 1.3517122268676758,
      "learning_rate": 9.920605263157894e-05,
      "loss": 0.6221,
      "step": 23028
    },
    {
      "epoch": 1.2895621010191511,
      "grad_norm": 1.0612396001815796,
      "learning_rate": 9.920578947368422e-05,
      "loss": 0.379,
      "step": 23029
    },
    {
      "epoch": 1.2896180983312802,
      "grad_norm": 1.3833634853363037,
      "learning_rate": 9.920552631578948e-05,
      "loss": 0.5444,
      "step": 23030
    },
    {
      "epoch": 1.2896740956434092,
      "grad_norm": 1.5076874494552612,
      "learning_rate": 9.920526315789475e-05,
      "loss": 0.4223,
      "step": 23031
    },
    {
      "epoch": 1.2897300929555382,
      "grad_norm": 1.2255945205688477,
      "learning_rate": 9.920500000000001e-05,
      "loss": 0.4352,
      "step": 23032
    },
    {
      "epoch": 1.2897860902676672,
      "grad_norm": 16.56756591796875,
      "learning_rate": 9.920473684210527e-05,
      "loss": 0.4737,
      "step": 23033
    },
    {
      "epoch": 1.2898420875797962,
      "grad_norm": 1.5384169816970825,
      "learning_rate": 9.920447368421053e-05,
      "loss": 0.5292,
      "step": 23034
    },
    {
      "epoch": 1.2898980848919253,
      "grad_norm": 1.4383050203323364,
      "learning_rate": 9.92042105263158e-05,
      "loss": 0.5717,
      "step": 23035
    },
    {
      "epoch": 1.2899540822040543,
      "grad_norm": 1.4422905445098877,
      "learning_rate": 9.920394736842106e-05,
      "loss": 0.5326,
      "step": 23036
    },
    {
      "epoch": 1.2900100795161833,
      "grad_norm": 1.395366907119751,
      "learning_rate": 9.920368421052632e-05,
      "loss": 0.5032,
      "step": 23037
    },
    {
      "epoch": 1.2900660768283123,
      "grad_norm": 1.51004159450531,
      "learning_rate": 9.920342105263158e-05,
      "loss": 0.584,
      "step": 23038
    },
    {
      "epoch": 1.2901220741404413,
      "grad_norm": 12.25399112701416,
      "learning_rate": 9.920315789473684e-05,
      "loss": 0.4347,
      "step": 23039
    },
    {
      "epoch": 1.2901780714525704,
      "grad_norm": 1.4509413242340088,
      "learning_rate": 9.920289473684211e-05,
      "loss": 0.4863,
      "step": 23040
    },
    {
      "epoch": 1.2902340687646994,
      "grad_norm": 1.5943634510040283,
      "learning_rate": 9.920263157894737e-05,
      "loss": 0.47,
      "step": 23041
    },
    {
      "epoch": 1.2902900660768284,
      "grad_norm": 1.487524390220642,
      "learning_rate": 9.920236842105263e-05,
      "loss": 0.4692,
      "step": 23042
    },
    {
      "epoch": 1.2903460633889574,
      "grad_norm": 1.4222885370254517,
      "learning_rate": 9.92021052631579e-05,
      "loss": 0.4326,
      "step": 23043
    },
    {
      "epoch": 1.2904020607010864,
      "grad_norm": 1.3612477779388428,
      "learning_rate": 9.920184210526317e-05,
      "loss": 0.4119,
      "step": 23044
    },
    {
      "epoch": 1.2904580580132154,
      "grad_norm": 1.4671094417572021,
      "learning_rate": 9.920157894736843e-05,
      "loss": 0.4623,
      "step": 23045
    },
    {
      "epoch": 1.2905140553253442,
      "grad_norm": 1.35136079788208,
      "learning_rate": 9.920131578947369e-05,
      "loss": 0.4039,
      "step": 23046
    },
    {
      "epoch": 1.2905700526374733,
      "grad_norm": 1.349650502204895,
      "learning_rate": 9.920105263157895e-05,
      "loss": 0.4961,
      "step": 23047
    },
    {
      "epoch": 1.2906260499496023,
      "grad_norm": 1.610949158668518,
      "learning_rate": 9.920078947368422e-05,
      "loss": 0.4722,
      "step": 23048
    },
    {
      "epoch": 1.2906820472617313,
      "grad_norm": 1.4275761842727661,
      "learning_rate": 9.920052631578948e-05,
      "loss": 0.4182,
      "step": 23049
    },
    {
      "epoch": 1.2907380445738603,
      "grad_norm": 1.6514980792999268,
      "learning_rate": 9.920026315789475e-05,
      "loss": 0.5518,
      "step": 23050
    },
    {
      "epoch": 1.2907940418859893,
      "grad_norm": 1.4580165147781372,
      "learning_rate": 9.92e-05,
      "loss": 0.3794,
      "step": 23051
    },
    {
      "epoch": 1.2908500391981184,
      "grad_norm": 1.6363708972930908,
      "learning_rate": 9.919973684210527e-05,
      "loss": 0.4578,
      "step": 23052
    },
    {
      "epoch": 1.2909060365102474,
      "grad_norm": 1.6622533798217773,
      "learning_rate": 9.919947368421053e-05,
      "loss": 0.7171,
      "step": 23053
    },
    {
      "epoch": 1.2909620338223764,
      "grad_norm": 1.8035329580307007,
      "learning_rate": 9.919921052631579e-05,
      "loss": 0.4881,
      "step": 23054
    },
    {
      "epoch": 1.2910180311345054,
      "grad_norm": 1.3802659511566162,
      "learning_rate": 9.919894736842105e-05,
      "loss": 0.4492,
      "step": 23055
    },
    {
      "epoch": 1.2910740284466344,
      "grad_norm": 2.016417980194092,
      "learning_rate": 9.919868421052631e-05,
      "loss": 0.5521,
      "step": 23056
    },
    {
      "epoch": 1.2911300257587635,
      "grad_norm": 1.479732632637024,
      "learning_rate": 9.919842105263158e-05,
      "loss": 0.4334,
      "step": 23057
    },
    {
      "epoch": 1.2911860230708925,
      "grad_norm": 1.2081102132797241,
      "learning_rate": 9.919815789473684e-05,
      "loss": 0.4467,
      "step": 23058
    },
    {
      "epoch": 1.2912420203830215,
      "grad_norm": 1.1472679376602173,
      "learning_rate": 9.919789473684212e-05,
      "loss": 0.3787,
      "step": 23059
    },
    {
      "epoch": 1.2912980176951505,
      "grad_norm": 1.3147450685501099,
      "learning_rate": 9.919763157894736e-05,
      "loss": 0.3942,
      "step": 23060
    },
    {
      "epoch": 1.2913540150072795,
      "grad_norm": 1.5052769184112549,
      "learning_rate": 9.919736842105264e-05,
      "loss": 0.4573,
      "step": 23061
    },
    {
      "epoch": 1.2914100123194086,
      "grad_norm": 1.358510136604309,
      "learning_rate": 9.91971052631579e-05,
      "loss": 0.5059,
      "step": 23062
    },
    {
      "epoch": 1.2914660096315376,
      "grad_norm": 1.321942925453186,
      "learning_rate": 9.919684210526317e-05,
      "loss": 0.5401,
      "step": 23063
    },
    {
      "epoch": 1.2915220069436666,
      "grad_norm": 1.363074779510498,
      "learning_rate": 9.919657894736843e-05,
      "loss": 0.4345,
      "step": 23064
    },
    {
      "epoch": 1.2915780042557956,
      "grad_norm": 1.3865578174591064,
      "learning_rate": 9.919631578947369e-05,
      "loss": 0.5602,
      "step": 23065
    },
    {
      "epoch": 1.2916340015679246,
      "grad_norm": 1.3393489122390747,
      "learning_rate": 9.919605263157895e-05,
      "loss": 0.3869,
      "step": 23066
    },
    {
      "epoch": 1.2916899988800536,
      "grad_norm": 1.5695127248764038,
      "learning_rate": 9.919578947368422e-05,
      "loss": 0.5572,
      "step": 23067
    },
    {
      "epoch": 1.2917459961921827,
      "grad_norm": 1.1903185844421387,
      "learning_rate": 9.919552631578948e-05,
      "loss": 0.4136,
      "step": 23068
    },
    {
      "epoch": 1.2918019935043117,
      "grad_norm": 1.7370045185089111,
      "learning_rate": 9.919526315789474e-05,
      "loss": 0.6811,
      "step": 23069
    },
    {
      "epoch": 1.2918579908164407,
      "grad_norm": 1.3773322105407715,
      "learning_rate": 9.9195e-05,
      "loss": 0.4945,
      "step": 23070
    },
    {
      "epoch": 1.2919139881285697,
      "grad_norm": 1.3980776071548462,
      "learning_rate": 9.919473684210526e-05,
      "loss": 0.4945,
      "step": 23071
    },
    {
      "epoch": 1.2919699854406987,
      "grad_norm": 1.73771071434021,
      "learning_rate": 9.919447368421053e-05,
      "loss": 0.4702,
      "step": 23072
    },
    {
      "epoch": 1.2920259827528278,
      "grad_norm": 1.491405963897705,
      "learning_rate": 9.91942105263158e-05,
      "loss": 0.5025,
      "step": 23073
    },
    {
      "epoch": 1.2920819800649568,
      "grad_norm": 1.2875639200210571,
      "learning_rate": 9.919394736842105e-05,
      "loss": 0.3839,
      "step": 23074
    },
    {
      "epoch": 1.2921379773770858,
      "grad_norm": 1.4439479112625122,
      "learning_rate": 9.919368421052631e-05,
      "loss": 0.4741,
      "step": 23075
    },
    {
      "epoch": 1.2921939746892148,
      "grad_norm": 1.5555182695388794,
      "learning_rate": 9.919342105263159e-05,
      "loss": 0.547,
      "step": 23076
    },
    {
      "epoch": 1.2922499720013438,
      "grad_norm": 1.3721479177474976,
      "learning_rate": 9.919315789473685e-05,
      "loss": 0.3663,
      "step": 23077
    },
    {
      "epoch": 1.2923059693134729,
      "grad_norm": 4.716894626617432,
      "learning_rate": 9.91928947368421e-05,
      "loss": 0.4126,
      "step": 23078
    },
    {
      "epoch": 1.2923619666256019,
      "grad_norm": 1.2825888395309448,
      "learning_rate": 9.919263157894737e-05,
      "loss": 0.4151,
      "step": 23079
    },
    {
      "epoch": 1.292417963937731,
      "grad_norm": 1.594100832939148,
      "learning_rate": 9.919236842105264e-05,
      "loss": 0.5598,
      "step": 23080
    },
    {
      "epoch": 1.29247396124986,
      "grad_norm": 1.3940268754959106,
      "learning_rate": 9.91921052631579e-05,
      "loss": 0.4255,
      "step": 23081
    },
    {
      "epoch": 1.292529958561989,
      "grad_norm": 1.435741901397705,
      "learning_rate": 9.919184210526317e-05,
      "loss": 0.5045,
      "step": 23082
    },
    {
      "epoch": 1.292585955874118,
      "grad_norm": 1.2281544208526611,
      "learning_rate": 9.919157894736842e-05,
      "loss": 0.4039,
      "step": 23083
    },
    {
      "epoch": 1.292641953186247,
      "grad_norm": 1.472573161125183,
      "learning_rate": 9.919131578947369e-05,
      "loss": 0.5481,
      "step": 23084
    },
    {
      "epoch": 1.292697950498376,
      "grad_norm": 1.5225231647491455,
      "learning_rate": 9.919105263157895e-05,
      "loss": 0.5484,
      "step": 23085
    },
    {
      "epoch": 1.292753947810505,
      "grad_norm": 1.4332185983657837,
      "learning_rate": 9.919078947368422e-05,
      "loss": 0.5739,
      "step": 23086
    },
    {
      "epoch": 1.292809945122634,
      "grad_norm": 1.1747983694076538,
      "learning_rate": 9.919052631578948e-05,
      "loss": 0.3832,
      "step": 23087
    },
    {
      "epoch": 1.292865942434763,
      "grad_norm": 1.521575927734375,
      "learning_rate": 9.919026315789473e-05,
      "loss": 0.4869,
      "step": 23088
    },
    {
      "epoch": 1.292921939746892,
      "grad_norm": 1.1403270959854126,
      "learning_rate": 9.919e-05,
      "loss": 0.3319,
      "step": 23089
    },
    {
      "epoch": 1.292977937059021,
      "grad_norm": 2.0384562015533447,
      "learning_rate": 9.918973684210526e-05,
      "loss": 0.4807,
      "step": 23090
    },
    {
      "epoch": 1.29303393437115,
      "grad_norm": 1.4302958250045776,
      "learning_rate": 9.918947368421054e-05,
      "loss": 0.5648,
      "step": 23091
    },
    {
      "epoch": 1.2930899316832791,
      "grad_norm": 1.2359665632247925,
      "learning_rate": 9.918921052631578e-05,
      "loss": 0.5447,
      "step": 23092
    },
    {
      "epoch": 1.2931459289954081,
      "grad_norm": 1.3534879684448242,
      "learning_rate": 9.918894736842106e-05,
      "loss": 0.6683,
      "step": 23093
    },
    {
      "epoch": 1.2932019263075372,
      "grad_norm": 1.5503642559051514,
      "learning_rate": 9.918868421052632e-05,
      "loss": 0.5193,
      "step": 23094
    },
    {
      "epoch": 1.2932579236196662,
      "grad_norm": 1.4643770456314087,
      "learning_rate": 9.918842105263159e-05,
      "loss": 0.4131,
      "step": 23095
    },
    {
      "epoch": 1.2933139209317952,
      "grad_norm": 2.064157247543335,
      "learning_rate": 9.918815789473685e-05,
      "loss": 0.4761,
      "step": 23096
    },
    {
      "epoch": 1.2933699182439242,
      "grad_norm": 1.501900553703308,
      "learning_rate": 9.918789473684211e-05,
      "loss": 0.4655,
      "step": 23097
    },
    {
      "epoch": 1.2934259155560532,
      "grad_norm": 1.5062413215637207,
      "learning_rate": 9.918763157894737e-05,
      "loss": 0.4824,
      "step": 23098
    },
    {
      "epoch": 1.2934819128681823,
      "grad_norm": 2.126584053039551,
      "learning_rate": 9.918736842105264e-05,
      "loss": 0.4974,
      "step": 23099
    },
    {
      "epoch": 1.2935379101803113,
      "grad_norm": 1.206376075744629,
      "learning_rate": 9.91871052631579e-05,
      "loss": 0.4751,
      "step": 23100
    },
    {
      "epoch": 1.2935939074924403,
      "grad_norm": 1.2563661336898804,
      "learning_rate": 9.918684210526316e-05,
      "loss": 0.4306,
      "step": 23101
    },
    {
      "epoch": 1.2936499048045693,
      "grad_norm": 1.3478997945785522,
      "learning_rate": 9.918657894736842e-05,
      "loss": 0.4101,
      "step": 23102
    },
    {
      "epoch": 1.2937059021166983,
      "grad_norm": 1.3861345052719116,
      "learning_rate": 9.91863157894737e-05,
      "loss": 0.4732,
      "step": 23103
    },
    {
      "epoch": 1.2937618994288274,
      "grad_norm": 1.8398720026016235,
      "learning_rate": 9.918605263157895e-05,
      "loss": 0.5285,
      "step": 23104
    },
    {
      "epoch": 1.2938178967409564,
      "grad_norm": 1.252652645111084,
      "learning_rate": 9.918578947368423e-05,
      "loss": 0.4509,
      "step": 23105
    },
    {
      "epoch": 1.2938738940530854,
      "grad_norm": 1.211471676826477,
      "learning_rate": 9.918552631578947e-05,
      "loss": 0.4291,
      "step": 23106
    },
    {
      "epoch": 1.2939298913652144,
      "grad_norm": 1.5865445137023926,
      "learning_rate": 9.918526315789473e-05,
      "loss": 0.6073,
      "step": 23107
    },
    {
      "epoch": 1.2939858886773434,
      "grad_norm": 1.4167582988739014,
      "learning_rate": 9.9185e-05,
      "loss": 0.4617,
      "step": 23108
    },
    {
      "epoch": 1.2940418859894725,
      "grad_norm": 1.1195567846298218,
      "learning_rate": 9.918473684210527e-05,
      "loss": 0.4275,
      "step": 23109
    },
    {
      "epoch": 1.2940978833016015,
      "grad_norm": 1.261151909828186,
      "learning_rate": 9.918447368421053e-05,
      "loss": 0.4719,
      "step": 23110
    },
    {
      "epoch": 1.2941538806137305,
      "grad_norm": 1.4415532350540161,
      "learning_rate": 9.918421052631579e-05,
      "loss": 0.4583,
      "step": 23111
    },
    {
      "epoch": 1.2942098779258595,
      "grad_norm": 1.394088864326477,
      "learning_rate": 9.918394736842106e-05,
      "loss": 0.488,
      "step": 23112
    },
    {
      "epoch": 1.2942658752379885,
      "grad_norm": 1.8100812435150146,
      "learning_rate": 9.918368421052632e-05,
      "loss": 0.6291,
      "step": 23113
    },
    {
      "epoch": 1.2943218725501175,
      "grad_norm": 1.3433568477630615,
      "learning_rate": 9.918342105263159e-05,
      "loss": 0.5986,
      "step": 23114
    },
    {
      "epoch": 1.2943778698622466,
      "grad_norm": 1.442548155784607,
      "learning_rate": 9.918315789473684e-05,
      "loss": 0.412,
      "step": 23115
    },
    {
      "epoch": 1.2944338671743756,
      "grad_norm": 1.5105018615722656,
      "learning_rate": 9.918289473684211e-05,
      "loss": 0.3668,
      "step": 23116
    },
    {
      "epoch": 1.2944898644865046,
      "grad_norm": 1.1323000192642212,
      "learning_rate": 9.918263157894737e-05,
      "loss": 0.4888,
      "step": 23117
    },
    {
      "epoch": 1.2945458617986336,
      "grad_norm": 1.354244589805603,
      "learning_rate": 9.918236842105264e-05,
      "loss": 0.4052,
      "step": 23118
    },
    {
      "epoch": 1.2946018591107626,
      "grad_norm": 1.3402349948883057,
      "learning_rate": 9.91821052631579e-05,
      "loss": 0.3926,
      "step": 23119
    },
    {
      "epoch": 1.2946578564228917,
      "grad_norm": 1.3419550657272339,
      "learning_rate": 9.918184210526316e-05,
      "loss": 0.437,
      "step": 23120
    },
    {
      "epoch": 1.2947138537350207,
      "grad_norm": 2.1664059162139893,
      "learning_rate": 9.918157894736842e-05,
      "loss": 0.6513,
      "step": 23121
    },
    {
      "epoch": 1.2947698510471497,
      "grad_norm": 1.4331457614898682,
      "learning_rate": 9.91813157894737e-05,
      "loss": 0.4336,
      "step": 23122
    },
    {
      "epoch": 1.2948258483592787,
      "grad_norm": 1.2470271587371826,
      "learning_rate": 9.918105263157896e-05,
      "loss": 0.4391,
      "step": 23123
    },
    {
      "epoch": 1.2948818456714077,
      "grad_norm": 1.4891743659973145,
      "learning_rate": 9.918078947368422e-05,
      "loss": 0.4758,
      "step": 23124
    },
    {
      "epoch": 1.2949378429835368,
      "grad_norm": 1.537827491760254,
      "learning_rate": 9.918052631578948e-05,
      "loss": 0.5983,
      "step": 23125
    },
    {
      "epoch": 1.2949938402956658,
      "grad_norm": 1.628988265991211,
      "learning_rate": 9.918026315789474e-05,
      "loss": 0.4892,
      "step": 23126
    },
    {
      "epoch": 1.2950498376077948,
      "grad_norm": 1.4828397035598755,
      "learning_rate": 9.918000000000001e-05,
      "loss": 0.4178,
      "step": 23127
    },
    {
      "epoch": 1.2951058349199238,
      "grad_norm": 1.4297609329223633,
      "learning_rate": 9.917973684210527e-05,
      "loss": 0.6309,
      "step": 23128
    },
    {
      "epoch": 1.2951618322320528,
      "grad_norm": 1.074594497680664,
      "learning_rate": 9.917947368421053e-05,
      "loss": 0.3676,
      "step": 23129
    },
    {
      "epoch": 1.2952178295441819,
      "grad_norm": 1.2312418222427368,
      "learning_rate": 9.917921052631579e-05,
      "loss": 0.3847,
      "step": 23130
    },
    {
      "epoch": 1.2952738268563109,
      "grad_norm": 1.3568131923675537,
      "learning_rate": 9.917894736842106e-05,
      "loss": 0.5376,
      "step": 23131
    },
    {
      "epoch": 1.29532982416844,
      "grad_norm": 1.3103629350662231,
      "learning_rate": 9.917868421052632e-05,
      "loss": 0.5244,
      "step": 23132
    },
    {
      "epoch": 1.295385821480569,
      "grad_norm": 1.4107483625411987,
      "learning_rate": 9.917842105263158e-05,
      "loss": 0.462,
      "step": 23133
    },
    {
      "epoch": 1.295441818792698,
      "grad_norm": 1.229911208152771,
      "learning_rate": 9.917815789473684e-05,
      "loss": 0.303,
      "step": 23134
    },
    {
      "epoch": 1.295497816104827,
      "grad_norm": 2.078338623046875,
      "learning_rate": 9.917789473684211e-05,
      "loss": 0.4735,
      "step": 23135
    },
    {
      "epoch": 1.295553813416956,
      "grad_norm": 1.6818737983703613,
      "learning_rate": 9.917763157894737e-05,
      "loss": 0.4912,
      "step": 23136
    },
    {
      "epoch": 1.295609810729085,
      "grad_norm": 1.308260202407837,
      "learning_rate": 9.917736842105265e-05,
      "loss": 0.5235,
      "step": 23137
    },
    {
      "epoch": 1.295665808041214,
      "grad_norm": 1.2334599494934082,
      "learning_rate": 9.91771052631579e-05,
      "loss": 0.3955,
      "step": 23138
    },
    {
      "epoch": 1.295721805353343,
      "grad_norm": 1.2334140539169312,
      "learning_rate": 9.917684210526317e-05,
      "loss": 0.3996,
      "step": 23139
    },
    {
      "epoch": 1.295777802665472,
      "grad_norm": 1.3915631771087646,
      "learning_rate": 9.917657894736843e-05,
      "loss": 0.4741,
      "step": 23140
    },
    {
      "epoch": 1.295833799977601,
      "grad_norm": 1.263722538948059,
      "learning_rate": 9.917631578947369e-05,
      "loss": 0.5301,
      "step": 23141
    },
    {
      "epoch": 1.29588979728973,
      "grad_norm": 1.5782407522201538,
      "learning_rate": 9.917605263157896e-05,
      "loss": 0.5274,
      "step": 23142
    },
    {
      "epoch": 1.295945794601859,
      "grad_norm": 1.4976212978363037,
      "learning_rate": 9.91757894736842e-05,
      "loss": 0.4581,
      "step": 23143
    },
    {
      "epoch": 1.2960017919139881,
      "grad_norm": 1.3090317249298096,
      "learning_rate": 9.917552631578948e-05,
      "loss": 0.4289,
      "step": 23144
    },
    {
      "epoch": 1.2960577892261171,
      "grad_norm": 1.5705372095108032,
      "learning_rate": 9.917526315789474e-05,
      "loss": 0.4602,
      "step": 23145
    },
    {
      "epoch": 1.2961137865382462,
      "grad_norm": 1.349718451499939,
      "learning_rate": 9.917500000000001e-05,
      "loss": 0.5322,
      "step": 23146
    },
    {
      "epoch": 1.2961697838503752,
      "grad_norm": 1.5553501844406128,
      "learning_rate": 9.917473684210526e-05,
      "loss": 0.5053,
      "step": 23147
    },
    {
      "epoch": 1.2962257811625042,
      "grad_norm": 1.4202433824539185,
      "learning_rate": 9.917447368421053e-05,
      "loss": 0.4849,
      "step": 23148
    },
    {
      "epoch": 1.2962817784746332,
      "grad_norm": 10.982044219970703,
      "learning_rate": 9.917421052631579e-05,
      "loss": 0.4594,
      "step": 23149
    },
    {
      "epoch": 1.2963377757867622,
      "grad_norm": 2.137157917022705,
      "learning_rate": 9.917394736842106e-05,
      "loss": 0.4211,
      "step": 23150
    },
    {
      "epoch": 1.2963937730988913,
      "grad_norm": 1.470205545425415,
      "learning_rate": 9.917368421052632e-05,
      "loss": 0.5466,
      "step": 23151
    },
    {
      "epoch": 1.2964497704110203,
      "grad_norm": 1.2946199178695679,
      "learning_rate": 9.917342105263158e-05,
      "loss": 0.3932,
      "step": 23152
    },
    {
      "epoch": 1.2965057677231493,
      "grad_norm": 1.8522518873214722,
      "learning_rate": 9.917315789473684e-05,
      "loss": 0.5165,
      "step": 23153
    },
    {
      "epoch": 1.2965617650352783,
      "grad_norm": 1.4067450761795044,
      "learning_rate": 9.917289473684212e-05,
      "loss": 0.4807,
      "step": 23154
    },
    {
      "epoch": 1.2966177623474073,
      "grad_norm": 1.2604808807373047,
      "learning_rate": 9.917263157894738e-05,
      "loss": 0.4159,
      "step": 23155
    },
    {
      "epoch": 1.2966737596595364,
      "grad_norm": 1.3013297319412231,
      "learning_rate": 9.917236842105264e-05,
      "loss": 0.4577,
      "step": 23156
    },
    {
      "epoch": 1.2967297569716654,
      "grad_norm": 1.9174307584762573,
      "learning_rate": 9.91721052631579e-05,
      "loss": 0.5487,
      "step": 23157
    },
    {
      "epoch": 1.2967857542837944,
      "grad_norm": 1.2069087028503418,
      "learning_rate": 9.917184210526316e-05,
      "loss": 0.4228,
      "step": 23158
    },
    {
      "epoch": 1.2968417515959234,
      "grad_norm": 1.5480200052261353,
      "learning_rate": 9.917157894736843e-05,
      "loss": 0.5662,
      "step": 23159
    },
    {
      "epoch": 1.2968977489080524,
      "grad_norm": 1.130193829536438,
      "learning_rate": 9.917131578947369e-05,
      "loss": 0.4401,
      "step": 23160
    },
    {
      "epoch": 1.2969537462201814,
      "grad_norm": 1.2581520080566406,
      "learning_rate": 9.917105263157895e-05,
      "loss": 0.3823,
      "step": 23161
    },
    {
      "epoch": 1.2970097435323105,
      "grad_norm": 1.2995219230651855,
      "learning_rate": 9.917078947368421e-05,
      "loss": 0.39,
      "step": 23162
    },
    {
      "epoch": 1.2970657408444395,
      "grad_norm": 1.1968741416931152,
      "learning_rate": 9.917052631578948e-05,
      "loss": 0.371,
      "step": 23163
    },
    {
      "epoch": 1.2971217381565685,
      "grad_norm": 1.3204034566879272,
      "learning_rate": 9.917026315789474e-05,
      "loss": 0.4821,
      "step": 23164
    },
    {
      "epoch": 1.2971777354686975,
      "grad_norm": 1.4278463125228882,
      "learning_rate": 9.917e-05,
      "loss": 0.4841,
      "step": 23165
    },
    {
      "epoch": 1.2972337327808265,
      "grad_norm": 1.7300227880477905,
      "learning_rate": 9.916973684210526e-05,
      "loss": 0.4935,
      "step": 23166
    },
    {
      "epoch": 1.2972897300929556,
      "grad_norm": 1.3219255208969116,
      "learning_rate": 9.916947368421053e-05,
      "loss": 0.4569,
      "step": 23167
    },
    {
      "epoch": 1.2973457274050846,
      "grad_norm": 1.4268701076507568,
      "learning_rate": 9.91692105263158e-05,
      "loss": 0.5079,
      "step": 23168
    },
    {
      "epoch": 1.2974017247172136,
      "grad_norm": 1.519562005996704,
      "learning_rate": 9.916894736842107e-05,
      "loss": 0.4115,
      "step": 23169
    },
    {
      "epoch": 1.2974577220293426,
      "grad_norm": 1.5569610595703125,
      "learning_rate": 9.916868421052631e-05,
      "loss": 0.4973,
      "step": 23170
    },
    {
      "epoch": 1.2975137193414716,
      "grad_norm": 1.2861106395721436,
      "learning_rate": 9.916842105263159e-05,
      "loss": 0.4534,
      "step": 23171
    },
    {
      "epoch": 1.2975697166536007,
      "grad_norm": 1.3012754917144775,
      "learning_rate": 9.916815789473685e-05,
      "loss": 0.5036,
      "step": 23172
    },
    {
      "epoch": 1.2976257139657297,
      "grad_norm": 1.5721306800842285,
      "learning_rate": 9.916789473684212e-05,
      "loss": 0.5698,
      "step": 23173
    },
    {
      "epoch": 1.2976817112778587,
      "grad_norm": 1.498849868774414,
      "learning_rate": 9.916763157894738e-05,
      "loss": 0.5996,
      "step": 23174
    },
    {
      "epoch": 1.2977377085899877,
      "grad_norm": 1.4276922941207886,
      "learning_rate": 9.916736842105263e-05,
      "loss": 0.4091,
      "step": 23175
    },
    {
      "epoch": 1.2977937059021167,
      "grad_norm": 1.5499815940856934,
      "learning_rate": 9.91671052631579e-05,
      "loss": 0.3802,
      "step": 23176
    },
    {
      "epoch": 1.2978497032142458,
      "grad_norm": 1.2192739248275757,
      "learning_rate": 9.916684210526316e-05,
      "loss": 0.5103,
      "step": 23177
    },
    {
      "epoch": 1.2979057005263748,
      "grad_norm": 1.1659903526306152,
      "learning_rate": 9.916657894736843e-05,
      "loss": 0.4654,
      "step": 23178
    },
    {
      "epoch": 1.2979616978385038,
      "grad_norm": 1.4987658262252808,
      "learning_rate": 9.916631578947369e-05,
      "loss": 0.4804,
      "step": 23179
    },
    {
      "epoch": 1.2980176951506328,
      "grad_norm": 2.007099151611328,
      "learning_rate": 9.916605263157895e-05,
      "loss": 0.5708,
      "step": 23180
    },
    {
      "epoch": 1.2980736924627618,
      "grad_norm": 1.3453506231307983,
      "learning_rate": 9.916578947368421e-05,
      "loss": 0.4868,
      "step": 23181
    },
    {
      "epoch": 1.2981296897748908,
      "grad_norm": 1.3088680505752563,
      "learning_rate": 9.916552631578948e-05,
      "loss": 0.4905,
      "step": 23182
    },
    {
      "epoch": 1.2981856870870199,
      "grad_norm": 1.4027286767959595,
      "learning_rate": 9.916526315789474e-05,
      "loss": 0.6571,
      "step": 23183
    },
    {
      "epoch": 1.2982416843991489,
      "grad_norm": 1.422242522239685,
      "learning_rate": 9.9165e-05,
      "loss": 0.4173,
      "step": 23184
    },
    {
      "epoch": 1.298297681711278,
      "grad_norm": 1.295256495475769,
      "learning_rate": 9.916473684210526e-05,
      "loss": 0.4624,
      "step": 23185
    },
    {
      "epoch": 1.298353679023407,
      "grad_norm": 1.3469479084014893,
      "learning_rate": 9.916447368421054e-05,
      "loss": 0.6286,
      "step": 23186
    },
    {
      "epoch": 1.298409676335536,
      "grad_norm": 1.4076720476150513,
      "learning_rate": 9.91642105263158e-05,
      "loss": 0.4399,
      "step": 23187
    },
    {
      "epoch": 1.298465673647665,
      "grad_norm": 1.4321870803833008,
      "learning_rate": 9.916394736842106e-05,
      "loss": 0.4432,
      "step": 23188
    },
    {
      "epoch": 1.298521670959794,
      "grad_norm": 1.5671515464782715,
      "learning_rate": 9.916368421052632e-05,
      "loss": 0.4282,
      "step": 23189
    },
    {
      "epoch": 1.298577668271923,
      "grad_norm": 1.2739958763122559,
      "learning_rate": 9.916342105263159e-05,
      "loss": 0.3489,
      "step": 23190
    },
    {
      "epoch": 1.298633665584052,
      "grad_norm": 1.3998924493789673,
      "learning_rate": 9.916315789473685e-05,
      "loss": 0.4127,
      "step": 23191
    },
    {
      "epoch": 1.298689662896181,
      "grad_norm": 1.3785351514816284,
      "learning_rate": 9.916289473684211e-05,
      "loss": 0.452,
      "step": 23192
    },
    {
      "epoch": 1.29874566020831,
      "grad_norm": 1.5077903270721436,
      "learning_rate": 9.916263157894737e-05,
      "loss": 0.4126,
      "step": 23193
    },
    {
      "epoch": 1.298801657520439,
      "grad_norm": 1.72370445728302,
      "learning_rate": 9.916236842105263e-05,
      "loss": 0.513,
      "step": 23194
    },
    {
      "epoch": 1.298857654832568,
      "grad_norm": 1.2757314443588257,
      "learning_rate": 9.91621052631579e-05,
      "loss": 0.4414,
      "step": 23195
    },
    {
      "epoch": 1.2989136521446971,
      "grad_norm": 6.817765712738037,
      "learning_rate": 9.916184210526316e-05,
      "loss": 0.4587,
      "step": 23196
    },
    {
      "epoch": 1.2989696494568261,
      "grad_norm": 1.236328363418579,
      "learning_rate": 9.916157894736843e-05,
      "loss": 0.5042,
      "step": 23197
    },
    {
      "epoch": 1.2990256467689552,
      "grad_norm": 1.5084718465805054,
      "learning_rate": 9.916131578947368e-05,
      "loss": 0.6613,
      "step": 23198
    },
    {
      "epoch": 1.2990816440810842,
      "grad_norm": 1.5099937915802002,
      "learning_rate": 9.916105263157895e-05,
      "loss": 0.4576,
      "step": 23199
    },
    {
      "epoch": 1.2991376413932132,
      "grad_norm": 1.9682420492172241,
      "learning_rate": 9.916078947368421e-05,
      "loss": 0.508,
      "step": 23200
    },
    {
      "epoch": 1.2991936387053422,
      "grad_norm": 1.282922625541687,
      "learning_rate": 9.916052631578949e-05,
      "loss": 0.5232,
      "step": 23201
    },
    {
      "epoch": 1.2992496360174712,
      "grad_norm": 1.4363484382629395,
      "learning_rate": 9.916026315789473e-05,
      "loss": 0.5648,
      "step": 23202
    },
    {
      "epoch": 1.2993056333296003,
      "grad_norm": 1.1488429307937622,
      "learning_rate": 9.916e-05,
      "loss": 0.5025,
      "step": 23203
    },
    {
      "epoch": 1.2993616306417293,
      "grad_norm": 1.7818005084991455,
      "learning_rate": 9.915973684210527e-05,
      "loss": 0.7077,
      "step": 23204
    },
    {
      "epoch": 1.2994176279538583,
      "grad_norm": 1.6766014099121094,
      "learning_rate": 9.915947368421054e-05,
      "loss": 0.4379,
      "step": 23205
    },
    {
      "epoch": 1.2994736252659873,
      "grad_norm": 1.5596418380737305,
      "learning_rate": 9.91592105263158e-05,
      "loss": 0.4459,
      "step": 23206
    },
    {
      "epoch": 1.2995296225781163,
      "grad_norm": 1.570223093032837,
      "learning_rate": 9.915894736842106e-05,
      "loss": 0.5508,
      "step": 23207
    },
    {
      "epoch": 1.2995856198902453,
      "grad_norm": 1.3504315614700317,
      "learning_rate": 9.915868421052632e-05,
      "loss": 0.4035,
      "step": 23208
    },
    {
      "epoch": 1.2996416172023744,
      "grad_norm": 1.6046736240386963,
      "learning_rate": 9.915842105263158e-05,
      "loss": 0.3972,
      "step": 23209
    },
    {
      "epoch": 1.2996976145145034,
      "grad_norm": 1.485824465751648,
      "learning_rate": 9.915815789473685e-05,
      "loss": 0.4442,
      "step": 23210
    },
    {
      "epoch": 1.2997536118266324,
      "grad_norm": 1.156082272529602,
      "learning_rate": 9.915789473684211e-05,
      "loss": 0.4251,
      "step": 23211
    },
    {
      "epoch": 1.2998096091387614,
      "grad_norm": 1.5914150476455688,
      "learning_rate": 9.915763157894737e-05,
      "loss": 0.5364,
      "step": 23212
    },
    {
      "epoch": 1.2998656064508904,
      "grad_norm": 1.5587588548660278,
      "learning_rate": 9.915736842105263e-05,
      "loss": 0.5468,
      "step": 23213
    },
    {
      "epoch": 1.2999216037630195,
      "grad_norm": 1.4740647077560425,
      "learning_rate": 9.91571052631579e-05,
      "loss": 0.4464,
      "step": 23214
    },
    {
      "epoch": 1.2999776010751485,
      "grad_norm": 1.7286256551742554,
      "learning_rate": 9.915684210526316e-05,
      "loss": 0.5414,
      "step": 23215
    },
    {
      "epoch": 1.3000335983872775,
      "grad_norm": 1.2766107320785522,
      "learning_rate": 9.915657894736842e-05,
      "loss": 0.4173,
      "step": 23216
    },
    {
      "epoch": 1.3000895956994065,
      "grad_norm": 6.632284641265869,
      "learning_rate": 9.915631578947368e-05,
      "loss": 0.5423,
      "step": 23217
    },
    {
      "epoch": 1.3001455930115355,
      "grad_norm": 1.4115962982177734,
      "learning_rate": 9.915605263157896e-05,
      "loss": 0.5377,
      "step": 23218
    },
    {
      "epoch": 1.3002015903236646,
      "grad_norm": 1.3499791622161865,
      "learning_rate": 9.915578947368422e-05,
      "loss": 0.4549,
      "step": 23219
    },
    {
      "epoch": 1.3002575876357936,
      "grad_norm": 1.404026746749878,
      "learning_rate": 9.915552631578948e-05,
      "loss": 0.5127,
      "step": 23220
    },
    {
      "epoch": 1.3003135849479226,
      "grad_norm": 8.091444969177246,
      "learning_rate": 9.915526315789474e-05,
      "loss": 0.3913,
      "step": 23221
    },
    {
      "epoch": 1.3003695822600516,
      "grad_norm": 1.43260657787323,
      "learning_rate": 9.915500000000001e-05,
      "loss": 0.4801,
      "step": 23222
    },
    {
      "epoch": 1.3004255795721806,
      "grad_norm": 1.3025494813919067,
      "learning_rate": 9.915473684210527e-05,
      "loss": 0.3282,
      "step": 23223
    },
    {
      "epoch": 1.3004815768843097,
      "grad_norm": 1.3490900993347168,
      "learning_rate": 9.915447368421054e-05,
      "loss": 0.4077,
      "step": 23224
    },
    {
      "epoch": 1.3005375741964387,
      "grad_norm": 1.1420457363128662,
      "learning_rate": 9.915421052631579e-05,
      "loss": 0.3791,
      "step": 23225
    },
    {
      "epoch": 1.3005935715085677,
      "grad_norm": 1.450870156288147,
      "learning_rate": 9.915394736842106e-05,
      "loss": 0.3893,
      "step": 23226
    },
    {
      "epoch": 1.3006495688206967,
      "grad_norm": 1.3151649236679077,
      "learning_rate": 9.915368421052632e-05,
      "loss": 0.4269,
      "step": 23227
    },
    {
      "epoch": 1.3007055661328257,
      "grad_norm": 1.3947206735610962,
      "learning_rate": 9.915342105263158e-05,
      "loss": 0.5421,
      "step": 23228
    },
    {
      "epoch": 1.3007615634449547,
      "grad_norm": 1.2942920923233032,
      "learning_rate": 9.915315789473685e-05,
      "loss": 0.4889,
      "step": 23229
    },
    {
      "epoch": 1.3008175607570838,
      "grad_norm": 1.3855652809143066,
      "learning_rate": 9.91528947368421e-05,
      "loss": 0.4664,
      "step": 23230
    },
    {
      "epoch": 1.3008735580692128,
      "grad_norm": 1.5210803747177124,
      "learning_rate": 9.915263157894737e-05,
      "loss": 0.447,
      "step": 23231
    },
    {
      "epoch": 1.3009295553813418,
      "grad_norm": 1.2672810554504395,
      "learning_rate": 9.915236842105263e-05,
      "loss": 0.4659,
      "step": 23232
    },
    {
      "epoch": 1.3009855526934708,
      "grad_norm": 1.268456220626831,
      "learning_rate": 9.91521052631579e-05,
      "loss": 0.3986,
      "step": 23233
    },
    {
      "epoch": 1.3010415500055998,
      "grad_norm": 1.3206121921539307,
      "learning_rate": 9.915184210526317e-05,
      "loss": 0.4101,
      "step": 23234
    },
    {
      "epoch": 1.3010975473177289,
      "grad_norm": 1.2786929607391357,
      "learning_rate": 9.915157894736843e-05,
      "loss": 0.412,
      "step": 23235
    },
    {
      "epoch": 1.3011535446298579,
      "grad_norm": 1.324182152748108,
      "learning_rate": 9.915131578947369e-05,
      "loss": 0.5193,
      "step": 23236
    },
    {
      "epoch": 1.301209541941987,
      "grad_norm": 1.262039303779602,
      "learning_rate": 9.915105263157896e-05,
      "loss": 0.3655,
      "step": 23237
    },
    {
      "epoch": 1.301265539254116,
      "grad_norm": 1.12849760055542,
      "learning_rate": 9.915078947368422e-05,
      "loss": 0.371,
      "step": 23238
    },
    {
      "epoch": 1.301321536566245,
      "grad_norm": 1.2479850053787231,
      "learning_rate": 9.915052631578948e-05,
      "loss": 0.4755,
      "step": 23239
    },
    {
      "epoch": 1.301377533878374,
      "grad_norm": 1.3527796268463135,
      "learning_rate": 9.915026315789474e-05,
      "loss": 0.5213,
      "step": 23240
    },
    {
      "epoch": 1.301433531190503,
      "grad_norm": 1.407912015914917,
      "learning_rate": 9.915000000000001e-05,
      "loss": 0.4465,
      "step": 23241
    },
    {
      "epoch": 1.301489528502632,
      "grad_norm": 1.2366207838058472,
      "learning_rate": 9.914973684210527e-05,
      "loss": 0.3874,
      "step": 23242
    },
    {
      "epoch": 1.301545525814761,
      "grad_norm": 1.6261301040649414,
      "learning_rate": 9.914947368421053e-05,
      "loss": 0.4496,
      "step": 23243
    },
    {
      "epoch": 1.30160152312689,
      "grad_norm": 2.06445574760437,
      "learning_rate": 9.914921052631579e-05,
      "loss": 0.7028,
      "step": 23244
    },
    {
      "epoch": 1.301657520439019,
      "grad_norm": 1.4373807907104492,
      "learning_rate": 9.914894736842105e-05,
      "loss": 0.4088,
      "step": 23245
    },
    {
      "epoch": 1.301713517751148,
      "grad_norm": 1.5163663625717163,
      "learning_rate": 9.914868421052632e-05,
      "loss": 0.4876,
      "step": 23246
    },
    {
      "epoch": 1.301769515063277,
      "grad_norm": 1.2399814128875732,
      "learning_rate": 9.914842105263158e-05,
      "loss": 0.4709,
      "step": 23247
    },
    {
      "epoch": 1.3018255123754061,
      "grad_norm": 1.331455945968628,
      "learning_rate": 9.914815789473684e-05,
      "loss": 0.4659,
      "step": 23248
    },
    {
      "epoch": 1.3018815096875351,
      "grad_norm": 1.4218822717666626,
      "learning_rate": 9.91478947368421e-05,
      "loss": 0.4753,
      "step": 23249
    },
    {
      "epoch": 1.3019375069996642,
      "grad_norm": 1.5947976112365723,
      "learning_rate": 9.914763157894738e-05,
      "loss": 0.5404,
      "step": 23250
    },
    {
      "epoch": 1.3019935043117932,
      "grad_norm": 1.4025335311889648,
      "learning_rate": 9.914736842105264e-05,
      "loss": 0.4188,
      "step": 23251
    },
    {
      "epoch": 1.3020495016239222,
      "grad_norm": 1.6437119245529175,
      "learning_rate": 9.914710526315791e-05,
      "loss": 0.5713,
      "step": 23252
    },
    {
      "epoch": 1.302105498936051,
      "grad_norm": 1.9255142211914062,
      "learning_rate": 9.914684210526316e-05,
      "loss": 0.6225,
      "step": 23253
    },
    {
      "epoch": 1.30216149624818,
      "grad_norm": 1.7305785417556763,
      "learning_rate": 9.914657894736843e-05,
      "loss": 0.4303,
      "step": 23254
    },
    {
      "epoch": 1.302217493560309,
      "grad_norm": 1.2466832399368286,
      "learning_rate": 9.914631578947369e-05,
      "loss": 0.4593,
      "step": 23255
    },
    {
      "epoch": 1.302273490872438,
      "grad_norm": 1.3282427787780762,
      "learning_rate": 9.914605263157896e-05,
      "loss": 0.4284,
      "step": 23256
    },
    {
      "epoch": 1.302329488184567,
      "grad_norm": 1.3135789632797241,
      "learning_rate": 9.914578947368421e-05,
      "loss": 0.5275,
      "step": 23257
    },
    {
      "epoch": 1.302385485496696,
      "grad_norm": 1.4246985912322998,
      "learning_rate": 9.914552631578948e-05,
      "loss": 0.4657,
      "step": 23258
    },
    {
      "epoch": 1.302441482808825,
      "grad_norm": 1.2614291906356812,
      "learning_rate": 9.914526315789474e-05,
      "loss": 0.4454,
      "step": 23259
    },
    {
      "epoch": 1.3024974801209541,
      "grad_norm": 1.3540866374969482,
      "learning_rate": 9.914500000000001e-05,
      "loss": 0.5905,
      "step": 23260
    },
    {
      "epoch": 1.3025534774330831,
      "grad_norm": 1.4114477634429932,
      "learning_rate": 9.914473684210527e-05,
      "loss": 0.4183,
      "step": 23261
    },
    {
      "epoch": 1.3026094747452122,
      "grad_norm": 1.3629343509674072,
      "learning_rate": 9.914447368421052e-05,
      "loss": 0.5109,
      "step": 23262
    },
    {
      "epoch": 1.3026654720573412,
      "grad_norm": 1.3114348649978638,
      "learning_rate": 9.914421052631579e-05,
      "loss": 0.404,
      "step": 23263
    },
    {
      "epoch": 1.3027214693694702,
      "grad_norm": 1.1841384172439575,
      "learning_rate": 9.914394736842105e-05,
      "loss": 0.4458,
      "step": 23264
    },
    {
      "epoch": 1.3027774666815992,
      "grad_norm": 1.4203022718429565,
      "learning_rate": 9.914368421052633e-05,
      "loss": 0.5124,
      "step": 23265
    },
    {
      "epoch": 1.3028334639937282,
      "grad_norm": 1.2591602802276611,
      "learning_rate": 9.914342105263159e-05,
      "loss": 0.3617,
      "step": 23266
    },
    {
      "epoch": 1.3028894613058573,
      "grad_norm": 1.407633900642395,
      "learning_rate": 9.914315789473685e-05,
      "loss": 0.5957,
      "step": 23267
    },
    {
      "epoch": 1.3029454586179863,
      "grad_norm": 1.4555846452713013,
      "learning_rate": 9.91428947368421e-05,
      "loss": 0.5717,
      "step": 23268
    },
    {
      "epoch": 1.3030014559301153,
      "grad_norm": 1.4491373300552368,
      "learning_rate": 9.914263157894738e-05,
      "loss": 0.5847,
      "step": 23269
    },
    {
      "epoch": 1.3030574532422443,
      "grad_norm": 1.156622052192688,
      "learning_rate": 9.914236842105264e-05,
      "loss": 0.4499,
      "step": 23270
    },
    {
      "epoch": 1.3031134505543733,
      "grad_norm": 1.415259599685669,
      "learning_rate": 9.91421052631579e-05,
      "loss": 0.4797,
      "step": 23271
    },
    {
      "epoch": 1.3031694478665024,
      "grad_norm": 1.3011866807937622,
      "learning_rate": 9.914184210526316e-05,
      "loss": 0.5682,
      "step": 23272
    },
    {
      "epoch": 1.3032254451786314,
      "grad_norm": 1.2020739316940308,
      "learning_rate": 9.914157894736843e-05,
      "loss": 0.4119,
      "step": 23273
    },
    {
      "epoch": 1.3032814424907604,
      "grad_norm": 1.252511739730835,
      "learning_rate": 9.914131578947369e-05,
      "loss": 0.4318,
      "step": 23274
    },
    {
      "epoch": 1.3033374398028894,
      "grad_norm": 1.3260678052902222,
      "learning_rate": 9.914105263157895e-05,
      "loss": 0.456,
      "step": 23275
    },
    {
      "epoch": 1.3033934371150184,
      "grad_norm": 1.4194862842559814,
      "learning_rate": 9.914078947368421e-05,
      "loss": 0.3703,
      "step": 23276
    },
    {
      "epoch": 1.3034494344271474,
      "grad_norm": 4.396236419677734,
      "learning_rate": 9.914052631578948e-05,
      "loss": 0.5069,
      "step": 23277
    },
    {
      "epoch": 1.3035054317392765,
      "grad_norm": 1.5681681632995605,
      "learning_rate": 9.914026315789474e-05,
      "loss": 0.3922,
      "step": 23278
    },
    {
      "epoch": 1.3035614290514055,
      "grad_norm": 1.3979411125183105,
      "learning_rate": 9.914e-05,
      "loss": 0.4714,
      "step": 23279
    },
    {
      "epoch": 1.3036174263635345,
      "grad_norm": 1.3500561714172363,
      "learning_rate": 9.913973684210526e-05,
      "loss": 0.4894,
      "step": 23280
    },
    {
      "epoch": 1.3036734236756635,
      "grad_norm": 1.3158615827560425,
      "learning_rate": 9.913947368421052e-05,
      "loss": 0.4118,
      "step": 23281
    },
    {
      "epoch": 1.3037294209877925,
      "grad_norm": 1.5024524927139282,
      "learning_rate": 9.91392105263158e-05,
      "loss": 0.5543,
      "step": 23282
    },
    {
      "epoch": 1.3037854182999216,
      "grad_norm": 1.415050745010376,
      "learning_rate": 9.913894736842106e-05,
      "loss": 0.4649,
      "step": 23283
    },
    {
      "epoch": 1.3038414156120506,
      "grad_norm": 1.1837997436523438,
      "learning_rate": 9.913868421052633e-05,
      "loss": 0.3367,
      "step": 23284
    },
    {
      "epoch": 1.3038974129241796,
      "grad_norm": 1.5360347032546997,
      "learning_rate": 9.913842105263157e-05,
      "loss": 0.4749,
      "step": 23285
    },
    {
      "epoch": 1.3039534102363086,
      "grad_norm": 1.4858312606811523,
      "learning_rate": 9.913815789473685e-05,
      "loss": 0.5332,
      "step": 23286
    },
    {
      "epoch": 1.3040094075484376,
      "grad_norm": 1.4720791578292847,
      "learning_rate": 9.913789473684211e-05,
      "loss": 0.4504,
      "step": 23287
    },
    {
      "epoch": 1.3040654048605667,
      "grad_norm": 1.3621578216552734,
      "learning_rate": 9.913763157894738e-05,
      "loss": 0.4723,
      "step": 23288
    },
    {
      "epoch": 1.3041214021726957,
      "grad_norm": 1.3581650257110596,
      "learning_rate": 9.913736842105264e-05,
      "loss": 0.4798,
      "step": 23289
    },
    {
      "epoch": 1.3041773994848247,
      "grad_norm": 1.228894591331482,
      "learning_rate": 9.91371052631579e-05,
      "loss": 0.4839,
      "step": 23290
    },
    {
      "epoch": 1.3042333967969537,
      "grad_norm": 1.6072814464569092,
      "learning_rate": 9.913684210526316e-05,
      "loss": 0.5408,
      "step": 23291
    },
    {
      "epoch": 1.3042893941090827,
      "grad_norm": 1.314273476600647,
      "learning_rate": 9.913657894736843e-05,
      "loss": 0.428,
      "step": 23292
    },
    {
      "epoch": 1.3043453914212118,
      "grad_norm": 1.6539702415466309,
      "learning_rate": 9.91363157894737e-05,
      "loss": 0.4635,
      "step": 23293
    },
    {
      "epoch": 1.3044013887333408,
      "grad_norm": 1.5021799802780151,
      "learning_rate": 9.913605263157895e-05,
      "loss": 0.4711,
      "step": 23294
    },
    {
      "epoch": 1.3044573860454698,
      "grad_norm": 1.4261759519577026,
      "learning_rate": 9.913578947368421e-05,
      "loss": 0.3836,
      "step": 23295
    },
    {
      "epoch": 1.3045133833575988,
      "grad_norm": 1.695491909980774,
      "learning_rate": 9.913552631578947e-05,
      "loss": 0.43,
      "step": 23296
    },
    {
      "epoch": 1.3045693806697278,
      "grad_norm": 1.4659150838851929,
      "learning_rate": 9.913526315789475e-05,
      "loss": 0.4848,
      "step": 23297
    },
    {
      "epoch": 1.3046253779818568,
      "grad_norm": 1.2196158170700073,
      "learning_rate": 9.9135e-05,
      "loss": 0.479,
      "step": 23298
    },
    {
      "epoch": 1.3046813752939859,
      "grad_norm": 1.092777967453003,
      "learning_rate": 9.913473684210527e-05,
      "loss": 0.3985,
      "step": 23299
    },
    {
      "epoch": 1.3047373726061149,
      "grad_norm": 1.4441514015197754,
      "learning_rate": 9.913447368421052e-05,
      "loss": 0.5424,
      "step": 23300
    },
    {
      "epoch": 1.304793369918244,
      "grad_norm": 1.565144658088684,
      "learning_rate": 9.91342105263158e-05,
      "loss": 0.4983,
      "step": 23301
    },
    {
      "epoch": 1.304849367230373,
      "grad_norm": 1.3440790176391602,
      "learning_rate": 9.913394736842106e-05,
      "loss": 0.3815,
      "step": 23302
    },
    {
      "epoch": 1.304905364542502,
      "grad_norm": 1.4669320583343506,
      "learning_rate": 9.913368421052632e-05,
      "loss": 0.5202,
      "step": 23303
    },
    {
      "epoch": 1.304961361854631,
      "grad_norm": 1.23827064037323,
      "learning_rate": 9.913342105263158e-05,
      "loss": 0.4864,
      "step": 23304
    },
    {
      "epoch": 1.30501735916676,
      "grad_norm": 1.6211961507797241,
      "learning_rate": 9.913315789473685e-05,
      "loss": 0.6918,
      "step": 23305
    },
    {
      "epoch": 1.305073356478889,
      "grad_norm": 1.5507512092590332,
      "learning_rate": 9.913289473684211e-05,
      "loss": 0.4171,
      "step": 23306
    },
    {
      "epoch": 1.305129353791018,
      "grad_norm": 1.3851318359375,
      "learning_rate": 9.913263157894738e-05,
      "loss": 0.59,
      "step": 23307
    },
    {
      "epoch": 1.305185351103147,
      "grad_norm": 1.2288877964019775,
      "learning_rate": 9.913236842105263e-05,
      "loss": 0.574,
      "step": 23308
    },
    {
      "epoch": 1.305241348415276,
      "grad_norm": 1.1446152925491333,
      "learning_rate": 9.91321052631579e-05,
      "loss": 0.4483,
      "step": 23309
    },
    {
      "epoch": 1.305297345727405,
      "grad_norm": 1.4239884614944458,
      "learning_rate": 9.913184210526316e-05,
      "loss": 0.5982,
      "step": 23310
    },
    {
      "epoch": 1.305353343039534,
      "grad_norm": 1.4093072414398193,
      "learning_rate": 9.913157894736844e-05,
      "loss": 0.5805,
      "step": 23311
    },
    {
      "epoch": 1.3054093403516631,
      "grad_norm": 1.4945576190948486,
      "learning_rate": 9.913131578947368e-05,
      "loss": 0.4839,
      "step": 23312
    },
    {
      "epoch": 1.3054653376637921,
      "grad_norm": 1.3525145053863525,
      "learning_rate": 9.913105263157894e-05,
      "loss": 0.4899,
      "step": 23313
    },
    {
      "epoch": 1.3055213349759212,
      "grad_norm": 3.119349241256714,
      "learning_rate": 9.913078947368422e-05,
      "loss": 0.5881,
      "step": 23314
    },
    {
      "epoch": 1.3055773322880502,
      "grad_norm": 1.3717396259307861,
      "learning_rate": 9.913052631578948e-05,
      "loss": 0.4957,
      "step": 23315
    },
    {
      "epoch": 1.3056333296001792,
      "grad_norm": 1.430946707725525,
      "learning_rate": 9.913026315789475e-05,
      "loss": 0.5012,
      "step": 23316
    },
    {
      "epoch": 1.3056893269123082,
      "grad_norm": 1.260886788368225,
      "learning_rate": 9.913e-05,
      "loss": 0.365,
      "step": 23317
    },
    {
      "epoch": 1.3057453242244372,
      "grad_norm": 1.356167197227478,
      "learning_rate": 9.912973684210527e-05,
      "loss": 0.4189,
      "step": 23318
    },
    {
      "epoch": 1.3058013215365663,
      "grad_norm": 1.2889000177383423,
      "learning_rate": 9.912947368421053e-05,
      "loss": 0.4207,
      "step": 23319
    },
    {
      "epoch": 1.3058573188486953,
      "grad_norm": 1.3411356210708618,
      "learning_rate": 9.91292105263158e-05,
      "loss": 0.5744,
      "step": 23320
    },
    {
      "epoch": 1.3059133161608243,
      "grad_norm": 1.5483306646347046,
      "learning_rate": 9.912894736842106e-05,
      "loss": 0.4928,
      "step": 23321
    },
    {
      "epoch": 1.3059693134729533,
      "grad_norm": 1.1878876686096191,
      "learning_rate": 9.912868421052632e-05,
      "loss": 0.4499,
      "step": 23322
    },
    {
      "epoch": 1.3060253107850823,
      "grad_norm": 1.2884507179260254,
      "learning_rate": 9.912842105263158e-05,
      "loss": 0.4424,
      "step": 23323
    },
    {
      "epoch": 1.3060813080972113,
      "grad_norm": 1.4340940713882446,
      "learning_rate": 9.912815789473685e-05,
      "loss": 0.3464,
      "step": 23324
    },
    {
      "epoch": 1.3061373054093404,
      "grad_norm": 1.3988069295883179,
      "learning_rate": 9.912789473684211e-05,
      "loss": 0.4857,
      "step": 23325
    },
    {
      "epoch": 1.3061933027214694,
      "grad_norm": 1.6385242938995361,
      "learning_rate": 9.912763157894737e-05,
      "loss": 0.4739,
      "step": 23326
    },
    {
      "epoch": 1.3062493000335984,
      "grad_norm": 1.5921891927719116,
      "learning_rate": 9.912736842105263e-05,
      "loss": 0.4359,
      "step": 23327
    },
    {
      "epoch": 1.3063052973457274,
      "grad_norm": 1.787548303604126,
      "learning_rate": 9.91271052631579e-05,
      "loss": 0.6647,
      "step": 23328
    },
    {
      "epoch": 1.3063612946578564,
      "grad_norm": 1.3793574571609497,
      "learning_rate": 9.912684210526317e-05,
      "loss": 0.5143,
      "step": 23329
    },
    {
      "epoch": 1.3064172919699855,
      "grad_norm": 1.6733791828155518,
      "learning_rate": 9.912657894736843e-05,
      "loss": 0.3727,
      "step": 23330
    },
    {
      "epoch": 1.3064732892821145,
      "grad_norm": 1.2901532649993896,
      "learning_rate": 9.912631578947368e-05,
      "loss": 0.3639,
      "step": 23331
    },
    {
      "epoch": 1.3065292865942435,
      "grad_norm": 1.5952303409576416,
      "learning_rate": 9.912605263157894e-05,
      "loss": 0.4403,
      "step": 23332
    },
    {
      "epoch": 1.3065852839063725,
      "grad_norm": 1.2913167476654053,
      "learning_rate": 9.912578947368422e-05,
      "loss": 0.4291,
      "step": 23333
    },
    {
      "epoch": 1.3066412812185015,
      "grad_norm": 1.5069353580474854,
      "learning_rate": 9.912552631578948e-05,
      "loss": 0.5659,
      "step": 23334
    },
    {
      "epoch": 1.3066972785306306,
      "grad_norm": 1.4820927381515503,
      "learning_rate": 9.912526315789474e-05,
      "loss": 0.6137,
      "step": 23335
    },
    {
      "epoch": 1.3067532758427596,
      "grad_norm": 1.560227870941162,
      "learning_rate": 9.9125e-05,
      "loss": 0.5614,
      "step": 23336
    },
    {
      "epoch": 1.3068092731548886,
      "grad_norm": 1.298736572265625,
      "learning_rate": 9.912473684210527e-05,
      "loss": 0.6134,
      "step": 23337
    },
    {
      "epoch": 1.3068652704670176,
      "grad_norm": 1.2767888307571411,
      "learning_rate": 9.912447368421053e-05,
      "loss": 0.4978,
      "step": 23338
    },
    {
      "epoch": 1.3069212677791466,
      "grad_norm": 1.344883680343628,
      "learning_rate": 9.91242105263158e-05,
      "loss": 0.4717,
      "step": 23339
    },
    {
      "epoch": 1.3069772650912757,
      "grad_norm": 1.1105804443359375,
      "learning_rate": 9.912394736842105e-05,
      "loss": 0.4326,
      "step": 23340
    },
    {
      "epoch": 1.3070332624034047,
      "grad_norm": 1.3717206716537476,
      "learning_rate": 9.912368421052632e-05,
      "loss": 0.4515,
      "step": 23341
    },
    {
      "epoch": 1.3070892597155337,
      "grad_norm": 1.207067847251892,
      "learning_rate": 9.912342105263158e-05,
      "loss": 0.3183,
      "step": 23342
    },
    {
      "epoch": 1.3071452570276627,
      "grad_norm": 1.5931912660598755,
      "learning_rate": 9.912315789473686e-05,
      "loss": 0.5249,
      "step": 23343
    },
    {
      "epoch": 1.3072012543397917,
      "grad_norm": 1.3497254848480225,
      "learning_rate": 9.912289473684212e-05,
      "loss": 0.4493,
      "step": 23344
    },
    {
      "epoch": 1.3072572516519207,
      "grad_norm": 1.2425850629806519,
      "learning_rate": 9.912263157894738e-05,
      "loss": 0.5717,
      "step": 23345
    },
    {
      "epoch": 1.3073132489640498,
      "grad_norm": 1.7057499885559082,
      "learning_rate": 9.912236842105264e-05,
      "loss": 0.6663,
      "step": 23346
    },
    {
      "epoch": 1.3073692462761788,
      "grad_norm": 1.1396676301956177,
      "learning_rate": 9.912210526315791e-05,
      "loss": 0.4206,
      "step": 23347
    },
    {
      "epoch": 1.3074252435883078,
      "grad_norm": 1.438910961151123,
      "learning_rate": 9.912184210526317e-05,
      "loss": 0.521,
      "step": 23348
    },
    {
      "epoch": 1.3074812409004368,
      "grad_norm": 1.5462173223495483,
      "learning_rate": 9.912157894736841e-05,
      "loss": 0.5551,
      "step": 23349
    },
    {
      "epoch": 1.3075372382125658,
      "grad_norm": 1.5258138179779053,
      "learning_rate": 9.912131578947369e-05,
      "loss": 0.4648,
      "step": 23350
    },
    {
      "epoch": 1.3075932355246949,
      "grad_norm": 1.5073356628417969,
      "learning_rate": 9.912105263157895e-05,
      "loss": 0.4733,
      "step": 23351
    },
    {
      "epoch": 1.3076492328368239,
      "grad_norm": 1.3298369646072388,
      "learning_rate": 9.912078947368422e-05,
      "loss": 0.5597,
      "step": 23352
    },
    {
      "epoch": 1.307705230148953,
      "grad_norm": 1.5825093984603882,
      "learning_rate": 9.912052631578948e-05,
      "loss": 0.5574,
      "step": 23353
    },
    {
      "epoch": 1.307761227461082,
      "grad_norm": 1.3182618618011475,
      "learning_rate": 9.912026315789474e-05,
      "loss": 0.5032,
      "step": 23354
    },
    {
      "epoch": 1.307817224773211,
      "grad_norm": 1.4643890857696533,
      "learning_rate": 9.912e-05,
      "loss": 0.6005,
      "step": 23355
    },
    {
      "epoch": 1.30787322208534,
      "grad_norm": 1.3440163135528564,
      "learning_rate": 9.911973684210527e-05,
      "loss": 0.6323,
      "step": 23356
    },
    {
      "epoch": 1.307929219397469,
      "grad_norm": 1.2643388509750366,
      "learning_rate": 9.911947368421053e-05,
      "loss": 0.3923,
      "step": 23357
    },
    {
      "epoch": 1.307985216709598,
      "grad_norm": 1.3173118829727173,
      "learning_rate": 9.911921052631579e-05,
      "loss": 0.5506,
      "step": 23358
    },
    {
      "epoch": 1.308041214021727,
      "grad_norm": 1.325744867324829,
      "learning_rate": 9.911894736842105e-05,
      "loss": 0.3308,
      "step": 23359
    },
    {
      "epoch": 1.308097211333856,
      "grad_norm": 1.4909019470214844,
      "learning_rate": 9.911868421052633e-05,
      "loss": 0.4128,
      "step": 23360
    },
    {
      "epoch": 1.308153208645985,
      "grad_norm": 1.180491328239441,
      "learning_rate": 9.911842105263159e-05,
      "loss": 0.5487,
      "step": 23361
    },
    {
      "epoch": 1.308209205958114,
      "grad_norm": 1.4190880060195923,
      "learning_rate": 9.911815789473686e-05,
      "loss": 0.3859,
      "step": 23362
    },
    {
      "epoch": 1.308265203270243,
      "grad_norm": 1.1913870573043823,
      "learning_rate": 9.91178947368421e-05,
      "loss": 0.4741,
      "step": 23363
    },
    {
      "epoch": 1.3083212005823721,
      "grad_norm": 1.2166470289230347,
      "learning_rate": 9.911763157894738e-05,
      "loss": 0.3879,
      "step": 23364
    },
    {
      "epoch": 1.3083771978945011,
      "grad_norm": 1.200237512588501,
      "learning_rate": 9.911736842105264e-05,
      "loss": 0.4206,
      "step": 23365
    },
    {
      "epoch": 1.3084331952066302,
      "grad_norm": 1.4765000343322754,
      "learning_rate": 9.91171052631579e-05,
      "loss": 0.4388,
      "step": 23366
    },
    {
      "epoch": 1.3084891925187592,
      "grad_norm": 1.7343289852142334,
      "learning_rate": 9.911684210526316e-05,
      "loss": 0.5143,
      "step": 23367
    },
    {
      "epoch": 1.3085451898308882,
      "grad_norm": 1.2377667427062988,
      "learning_rate": 9.911657894736842e-05,
      "loss": 0.4297,
      "step": 23368
    },
    {
      "epoch": 1.3086011871430172,
      "grad_norm": 1.5643452405929565,
      "learning_rate": 9.911631578947369e-05,
      "loss": 0.4753,
      "step": 23369
    },
    {
      "epoch": 1.3086571844551462,
      "grad_norm": 1.6922138929367065,
      "learning_rate": 9.911605263157895e-05,
      "loss": 0.4719,
      "step": 23370
    },
    {
      "epoch": 1.3087131817672752,
      "grad_norm": 1.2513607740402222,
      "learning_rate": 9.911578947368422e-05,
      "loss": 0.401,
      "step": 23371
    },
    {
      "epoch": 1.3087691790794043,
      "grad_norm": 1.266663908958435,
      "learning_rate": 9.911552631578947e-05,
      "loss": 0.4367,
      "step": 23372
    },
    {
      "epoch": 1.3088251763915333,
      "grad_norm": 1.1690346002578735,
      "learning_rate": 9.911526315789474e-05,
      "loss": 0.3916,
      "step": 23373
    },
    {
      "epoch": 1.3088811737036623,
      "grad_norm": 1.3350450992584229,
      "learning_rate": 9.9115e-05,
      "loss": 0.4463,
      "step": 23374
    },
    {
      "epoch": 1.3089371710157913,
      "grad_norm": 1.3662041425704956,
      "learning_rate": 9.911473684210528e-05,
      "loss": 0.4163,
      "step": 23375
    },
    {
      "epoch": 1.3089931683279203,
      "grad_norm": 1.4900022745132446,
      "learning_rate": 9.911447368421054e-05,
      "loss": 0.4521,
      "step": 23376
    },
    {
      "epoch": 1.3090491656400491,
      "grad_norm": 1.3184181451797485,
      "learning_rate": 9.91142105263158e-05,
      "loss": 0.4926,
      "step": 23377
    },
    {
      "epoch": 1.3091051629521782,
      "grad_norm": 1.287375807762146,
      "learning_rate": 9.911394736842105e-05,
      "loss": 0.4877,
      "step": 23378
    },
    {
      "epoch": 1.3091611602643072,
      "grad_norm": 1.5189950466156006,
      "learning_rate": 9.911368421052633e-05,
      "loss": 0.5017,
      "step": 23379
    },
    {
      "epoch": 1.3092171575764362,
      "grad_norm": 1.565103530883789,
      "learning_rate": 9.911342105263159e-05,
      "loss": 0.6216,
      "step": 23380
    },
    {
      "epoch": 1.3092731548885652,
      "grad_norm": 1.495690941810608,
      "learning_rate": 9.911315789473685e-05,
      "loss": 0.5092,
      "step": 23381
    },
    {
      "epoch": 1.3093291522006942,
      "grad_norm": 1.5382283926010132,
      "learning_rate": 9.911289473684211e-05,
      "loss": 0.7018,
      "step": 23382
    },
    {
      "epoch": 1.3093851495128233,
      "grad_norm": 1.5031582117080688,
      "learning_rate": 9.911263157894737e-05,
      "loss": 0.5123,
      "step": 23383
    },
    {
      "epoch": 1.3094411468249523,
      "grad_norm": 1.1811401844024658,
      "learning_rate": 9.911236842105264e-05,
      "loss": 0.5173,
      "step": 23384
    },
    {
      "epoch": 1.3094971441370813,
      "grad_norm": 1.1787773370742798,
      "learning_rate": 9.91121052631579e-05,
      "loss": 0.4346,
      "step": 23385
    },
    {
      "epoch": 1.3095531414492103,
      "grad_norm": 1.3113309144973755,
      "learning_rate": 9.911184210526316e-05,
      "loss": 0.3488,
      "step": 23386
    },
    {
      "epoch": 1.3096091387613393,
      "grad_norm": 1.4525781869888306,
      "learning_rate": 9.911157894736842e-05,
      "loss": 0.4528,
      "step": 23387
    },
    {
      "epoch": 1.3096651360734684,
      "grad_norm": 1.2588186264038086,
      "learning_rate": 9.911131578947369e-05,
      "loss": 0.4124,
      "step": 23388
    },
    {
      "epoch": 1.3097211333855974,
      "grad_norm": 1.700707197189331,
      "learning_rate": 9.911105263157895e-05,
      "loss": 0.4042,
      "step": 23389
    },
    {
      "epoch": 1.3097771306977264,
      "grad_norm": 2.0704498291015625,
      "learning_rate": 9.911078947368421e-05,
      "loss": 0.5692,
      "step": 23390
    },
    {
      "epoch": 1.3098331280098554,
      "grad_norm": 1.415603756904602,
      "learning_rate": 9.911052631578947e-05,
      "loss": 0.376,
      "step": 23391
    },
    {
      "epoch": 1.3098891253219844,
      "grad_norm": 1.505025029182434,
      "learning_rate": 9.911026315789475e-05,
      "loss": 0.4533,
      "step": 23392
    },
    {
      "epoch": 1.3099451226341134,
      "grad_norm": 1.4242792129516602,
      "learning_rate": 9.911e-05,
      "loss": 0.4603,
      "step": 23393
    },
    {
      "epoch": 1.3100011199462425,
      "grad_norm": 1.1377774477005005,
      "learning_rate": 9.910973684210528e-05,
      "loss": 0.3962,
      "step": 23394
    },
    {
      "epoch": 1.3100571172583715,
      "grad_norm": 1.831237554550171,
      "learning_rate": 9.910947368421052e-05,
      "loss": 0.4864,
      "step": 23395
    },
    {
      "epoch": 1.3101131145705005,
      "grad_norm": 1.5969734191894531,
      "learning_rate": 9.91092105263158e-05,
      "loss": 0.4875,
      "step": 23396
    },
    {
      "epoch": 1.3101691118826295,
      "grad_norm": 1.4190731048583984,
      "learning_rate": 9.910894736842106e-05,
      "loss": 0.4684,
      "step": 23397
    },
    {
      "epoch": 1.3102251091947585,
      "grad_norm": 1.3128427267074585,
      "learning_rate": 9.910868421052633e-05,
      "loss": 0.5009,
      "step": 23398
    },
    {
      "epoch": 1.3102811065068876,
      "grad_norm": 1.2901273965835571,
      "learning_rate": 9.910842105263159e-05,
      "loss": 0.5086,
      "step": 23399
    },
    {
      "epoch": 1.3103371038190166,
      "grad_norm": 1.7611383199691772,
      "learning_rate": 9.910815789473684e-05,
      "loss": 0.461,
      "step": 23400
    },
    {
      "epoch": 1.3103931011311456,
      "grad_norm": 1.4367409944534302,
      "learning_rate": 9.910789473684211e-05,
      "loss": 0.6223,
      "step": 23401
    },
    {
      "epoch": 1.3104490984432746,
      "grad_norm": 1.338951587677002,
      "learning_rate": 9.910763157894737e-05,
      "loss": 0.4048,
      "step": 23402
    },
    {
      "epoch": 1.3105050957554036,
      "grad_norm": 1.4415680170059204,
      "learning_rate": 9.910736842105264e-05,
      "loss": 0.5566,
      "step": 23403
    },
    {
      "epoch": 1.3105610930675327,
      "grad_norm": 1.296931266784668,
      "learning_rate": 9.910710526315789e-05,
      "loss": 0.5373,
      "step": 23404
    },
    {
      "epoch": 1.3106170903796617,
      "grad_norm": 1.7117139101028442,
      "learning_rate": 9.910684210526316e-05,
      "loss": 0.4762,
      "step": 23405
    },
    {
      "epoch": 1.3106730876917907,
      "grad_norm": 1.2670321464538574,
      "learning_rate": 9.910657894736842e-05,
      "loss": 0.4376,
      "step": 23406
    },
    {
      "epoch": 1.3107290850039197,
      "grad_norm": 1.2396507263183594,
      "learning_rate": 9.91063157894737e-05,
      "loss": 0.5328,
      "step": 23407
    },
    {
      "epoch": 1.3107850823160487,
      "grad_norm": 1.573288917541504,
      "learning_rate": 9.910605263157895e-05,
      "loss": 0.7282,
      "step": 23408
    },
    {
      "epoch": 1.3108410796281778,
      "grad_norm": 1.1891690492630005,
      "learning_rate": 9.910578947368421e-05,
      "loss": 0.4956,
      "step": 23409
    },
    {
      "epoch": 1.3108970769403068,
      "grad_norm": 1.2904771566390991,
      "learning_rate": 9.910552631578947e-05,
      "loss": 0.5151,
      "step": 23410
    },
    {
      "epoch": 1.3109530742524358,
      "grad_norm": 1.278806209564209,
      "learning_rate": 9.910526315789475e-05,
      "loss": 0.5051,
      "step": 23411
    },
    {
      "epoch": 1.3110090715645648,
      "grad_norm": 1.4368467330932617,
      "learning_rate": 9.910500000000001e-05,
      "loss": 0.447,
      "step": 23412
    },
    {
      "epoch": 1.3110650688766938,
      "grad_norm": 1.5844480991363525,
      "learning_rate": 9.910473684210527e-05,
      "loss": 0.6849,
      "step": 23413
    },
    {
      "epoch": 1.3111210661888228,
      "grad_norm": 1.3631752729415894,
      "learning_rate": 9.910447368421053e-05,
      "loss": 0.5,
      "step": 23414
    },
    {
      "epoch": 1.3111770635009519,
      "grad_norm": 1.7622642517089844,
      "learning_rate": 9.91042105263158e-05,
      "loss": 0.7695,
      "step": 23415
    },
    {
      "epoch": 1.3112330608130809,
      "grad_norm": 1.1892203092575073,
      "learning_rate": 9.910394736842106e-05,
      "loss": 0.3501,
      "step": 23416
    },
    {
      "epoch": 1.31128905812521,
      "grad_norm": 1.4493550062179565,
      "learning_rate": 9.910368421052632e-05,
      "loss": 0.5148,
      "step": 23417
    },
    {
      "epoch": 1.311345055437339,
      "grad_norm": 2.0674352645874023,
      "learning_rate": 9.910342105263158e-05,
      "loss": 0.3919,
      "step": 23418
    },
    {
      "epoch": 1.311401052749468,
      "grad_norm": 1.2321887016296387,
      "learning_rate": 9.910315789473684e-05,
      "loss": 0.4374,
      "step": 23419
    },
    {
      "epoch": 1.311457050061597,
      "grad_norm": 1.5014137029647827,
      "learning_rate": 9.910289473684211e-05,
      "loss": 0.5516,
      "step": 23420
    },
    {
      "epoch": 1.311513047373726,
      "grad_norm": 1.337637186050415,
      "learning_rate": 9.910263157894737e-05,
      "loss": 0.4281,
      "step": 23421
    },
    {
      "epoch": 1.311569044685855,
      "grad_norm": 1.747450351715088,
      "learning_rate": 9.910236842105263e-05,
      "loss": 0.594,
      "step": 23422
    },
    {
      "epoch": 1.311625041997984,
      "grad_norm": 1.2766363620758057,
      "learning_rate": 9.910210526315789e-05,
      "loss": 0.4447,
      "step": 23423
    },
    {
      "epoch": 1.311681039310113,
      "grad_norm": 1.3487766981124878,
      "learning_rate": 9.910184210526316e-05,
      "loss": 0.5729,
      "step": 23424
    },
    {
      "epoch": 1.311737036622242,
      "grad_norm": 1.3309218883514404,
      "learning_rate": 9.910157894736842e-05,
      "loss": 0.5082,
      "step": 23425
    },
    {
      "epoch": 1.311793033934371,
      "grad_norm": 1.2421057224273682,
      "learning_rate": 9.91013157894737e-05,
      "loss": 0.423,
      "step": 23426
    },
    {
      "epoch": 1.3118490312465,
      "grad_norm": 1.2512723207473755,
      "learning_rate": 9.910105263157894e-05,
      "loss": 0.3813,
      "step": 23427
    },
    {
      "epoch": 1.3119050285586291,
      "grad_norm": 1.492981195449829,
      "learning_rate": 9.910078947368422e-05,
      "loss": 0.5883,
      "step": 23428
    },
    {
      "epoch": 1.3119610258707581,
      "grad_norm": 1.3836313486099243,
      "learning_rate": 9.910052631578948e-05,
      "loss": 0.4316,
      "step": 23429
    },
    {
      "epoch": 1.3120170231828872,
      "grad_norm": 1.6830849647521973,
      "learning_rate": 9.910026315789475e-05,
      "loss": 0.5704,
      "step": 23430
    },
    {
      "epoch": 1.3120730204950162,
      "grad_norm": 1.978153944015503,
      "learning_rate": 9.910000000000001e-05,
      "loss": 0.5258,
      "step": 23431
    },
    {
      "epoch": 1.3121290178071452,
      "grad_norm": 1.2363853454589844,
      "learning_rate": 9.909973684210527e-05,
      "loss": 0.4237,
      "step": 23432
    },
    {
      "epoch": 1.3121850151192742,
      "grad_norm": 1.2318207025527954,
      "learning_rate": 9.909947368421053e-05,
      "loss": 0.4274,
      "step": 23433
    },
    {
      "epoch": 1.3122410124314032,
      "grad_norm": 1.1699894666671753,
      "learning_rate": 9.909921052631579e-05,
      "loss": 0.4445,
      "step": 23434
    },
    {
      "epoch": 1.3122970097435323,
      "grad_norm": 1.2759298086166382,
      "learning_rate": 9.909894736842106e-05,
      "loss": 0.4688,
      "step": 23435
    },
    {
      "epoch": 1.3123530070556613,
      "grad_norm": 1.3733446598052979,
      "learning_rate": 9.909868421052632e-05,
      "loss": 0.4503,
      "step": 23436
    },
    {
      "epoch": 1.3124090043677903,
      "grad_norm": 1.1985161304473877,
      "learning_rate": 9.909842105263158e-05,
      "loss": 0.469,
      "step": 23437
    },
    {
      "epoch": 1.3124650016799193,
      "grad_norm": 1.4328134059906006,
      "learning_rate": 9.909815789473684e-05,
      "loss": 0.4739,
      "step": 23438
    },
    {
      "epoch": 1.3125209989920483,
      "grad_norm": 1.2340866327285767,
      "learning_rate": 9.909789473684211e-05,
      "loss": 0.415,
      "step": 23439
    },
    {
      "epoch": 1.3125769963041773,
      "grad_norm": 1.4035005569458008,
      "learning_rate": 9.909763157894737e-05,
      "loss": 0.3763,
      "step": 23440
    },
    {
      "epoch": 1.3126329936163064,
      "grad_norm": 1.3108558654785156,
      "learning_rate": 9.909736842105263e-05,
      "loss": 0.371,
      "step": 23441
    },
    {
      "epoch": 1.3126889909284354,
      "grad_norm": 1.7617404460906982,
      "learning_rate": 9.90971052631579e-05,
      "loss": 0.5191,
      "step": 23442
    },
    {
      "epoch": 1.3127449882405644,
      "grad_norm": 1.3765901327133179,
      "learning_rate": 9.909684210526317e-05,
      "loss": 0.6761,
      "step": 23443
    },
    {
      "epoch": 1.3128009855526934,
      "grad_norm": 1.166473150253296,
      "learning_rate": 9.909657894736843e-05,
      "loss": 0.4358,
      "step": 23444
    },
    {
      "epoch": 1.3128569828648224,
      "grad_norm": 1.3755375146865845,
      "learning_rate": 9.909631578947369e-05,
      "loss": 0.4297,
      "step": 23445
    },
    {
      "epoch": 1.3129129801769515,
      "grad_norm": 1.3300050497055054,
      "learning_rate": 9.909605263157895e-05,
      "loss": 0.4262,
      "step": 23446
    },
    {
      "epoch": 1.3129689774890805,
      "grad_norm": 1.282021403312683,
      "learning_rate": 9.909578947368422e-05,
      "loss": 0.4449,
      "step": 23447
    },
    {
      "epoch": 1.3130249748012095,
      "grad_norm": 1.2514032125473022,
      "learning_rate": 9.909552631578948e-05,
      "loss": 0.5067,
      "step": 23448
    },
    {
      "epoch": 1.3130809721133385,
      "grad_norm": 1.432243824005127,
      "learning_rate": 9.909526315789475e-05,
      "loss": 0.4014,
      "step": 23449
    },
    {
      "epoch": 1.3131369694254675,
      "grad_norm": 1.4824650287628174,
      "learning_rate": 9.9095e-05,
      "loss": 0.4567,
      "step": 23450
    },
    {
      "epoch": 1.3131929667375966,
      "grad_norm": 2.4755280017852783,
      "learning_rate": 9.909473684210526e-05,
      "loss": 0.5068,
      "step": 23451
    },
    {
      "epoch": 1.3132489640497256,
      "grad_norm": 1.7152857780456543,
      "learning_rate": 9.909447368421053e-05,
      "loss": 0.4997,
      "step": 23452
    },
    {
      "epoch": 1.3133049613618546,
      "grad_norm": 1.4264469146728516,
      "learning_rate": 9.909421052631579e-05,
      "loss": 0.475,
      "step": 23453
    },
    {
      "epoch": 1.3133609586739836,
      "grad_norm": 1.1713494062423706,
      "learning_rate": 9.909394736842107e-05,
      "loss": 0.317,
      "step": 23454
    },
    {
      "epoch": 1.3134169559861126,
      "grad_norm": 1.0777666568756104,
      "learning_rate": 9.909368421052631e-05,
      "loss": 0.3438,
      "step": 23455
    },
    {
      "epoch": 1.3134729532982417,
      "grad_norm": 1.5504034757614136,
      "learning_rate": 9.909342105263158e-05,
      "loss": 0.7497,
      "step": 23456
    },
    {
      "epoch": 1.3135289506103707,
      "grad_norm": 1.3663454055786133,
      "learning_rate": 9.909315789473684e-05,
      "loss": 0.4087,
      "step": 23457
    },
    {
      "epoch": 1.3135849479224997,
      "grad_norm": 1.5986995697021484,
      "learning_rate": 9.909289473684212e-05,
      "loss": 0.5057,
      "step": 23458
    },
    {
      "epoch": 1.3136409452346287,
      "grad_norm": 1.3435286283493042,
      "learning_rate": 9.909263157894736e-05,
      "loss": 0.3877,
      "step": 23459
    },
    {
      "epoch": 1.3136969425467577,
      "grad_norm": 1.3977755308151245,
      "learning_rate": 9.909236842105264e-05,
      "loss": 0.5025,
      "step": 23460
    },
    {
      "epoch": 1.3137529398588867,
      "grad_norm": 1.5699728727340698,
      "learning_rate": 9.90921052631579e-05,
      "loss": 0.4892,
      "step": 23461
    },
    {
      "epoch": 1.3138089371710158,
      "grad_norm": 1.2305196523666382,
      "learning_rate": 9.909184210526317e-05,
      "loss": 0.4396,
      "step": 23462
    },
    {
      "epoch": 1.3138649344831448,
      "grad_norm": 1.4703571796417236,
      "learning_rate": 9.909157894736843e-05,
      "loss": 0.4488,
      "step": 23463
    },
    {
      "epoch": 1.3139209317952738,
      "grad_norm": 2.4165940284729004,
      "learning_rate": 9.909131578947369e-05,
      "loss": 0.4089,
      "step": 23464
    },
    {
      "epoch": 1.3139769291074028,
      "grad_norm": 1.211911916732788,
      "learning_rate": 9.909105263157895e-05,
      "loss": 0.4259,
      "step": 23465
    },
    {
      "epoch": 1.3140329264195318,
      "grad_norm": 1.1673394441604614,
      "learning_rate": 9.909078947368422e-05,
      "loss": 0.3949,
      "step": 23466
    },
    {
      "epoch": 1.3140889237316609,
      "grad_norm": 1.3991273641586304,
      "learning_rate": 9.909052631578948e-05,
      "loss": 0.4416,
      "step": 23467
    },
    {
      "epoch": 1.3141449210437899,
      "grad_norm": 9.993056297302246,
      "learning_rate": 9.909026315789474e-05,
      "loss": 0.5116,
      "step": 23468
    },
    {
      "epoch": 1.314200918355919,
      "grad_norm": 1.3935984373092651,
      "learning_rate": 9.909e-05,
      "loss": 0.4229,
      "step": 23469
    },
    {
      "epoch": 1.314256915668048,
      "grad_norm": 1.576490044593811,
      "learning_rate": 9.908973684210526e-05,
      "loss": 0.5644,
      "step": 23470
    },
    {
      "epoch": 1.314312912980177,
      "grad_norm": 1.3270258903503418,
      "learning_rate": 9.908947368421053e-05,
      "loss": 0.4396,
      "step": 23471
    },
    {
      "epoch": 1.314368910292306,
      "grad_norm": 1.487135410308838,
      "learning_rate": 9.90892105263158e-05,
      "loss": 0.6049,
      "step": 23472
    },
    {
      "epoch": 1.314424907604435,
      "grad_norm": 1.3678137063980103,
      "learning_rate": 9.908894736842105e-05,
      "loss": 0.5267,
      "step": 23473
    },
    {
      "epoch": 1.314480904916564,
      "grad_norm": 2.1125478744506836,
      "learning_rate": 9.908868421052631e-05,
      "loss": 0.4868,
      "step": 23474
    },
    {
      "epoch": 1.314536902228693,
      "grad_norm": 1.335037112236023,
      "learning_rate": 9.908842105263159e-05,
      "loss": 0.5048,
      "step": 23475
    },
    {
      "epoch": 1.314592899540822,
      "grad_norm": 1.3628175258636475,
      "learning_rate": 9.908815789473685e-05,
      "loss": 0.4408,
      "step": 23476
    },
    {
      "epoch": 1.314648896852951,
      "grad_norm": 1.4586461782455444,
      "learning_rate": 9.90878947368421e-05,
      "loss": 0.5362,
      "step": 23477
    },
    {
      "epoch": 1.31470489416508,
      "grad_norm": 1.2521742582321167,
      "learning_rate": 9.908763157894737e-05,
      "loss": 0.4385,
      "step": 23478
    },
    {
      "epoch": 1.314760891477209,
      "grad_norm": 1.889302134513855,
      "learning_rate": 9.908736842105264e-05,
      "loss": 0.7015,
      "step": 23479
    },
    {
      "epoch": 1.3148168887893381,
      "grad_norm": 1.3009964227676392,
      "learning_rate": 9.90871052631579e-05,
      "loss": 0.4561,
      "step": 23480
    },
    {
      "epoch": 1.3148728861014671,
      "grad_norm": 1.9913792610168457,
      "learning_rate": 9.908684210526317e-05,
      "loss": 0.4249,
      "step": 23481
    },
    {
      "epoch": 1.3149288834135962,
      "grad_norm": 1.304996132850647,
      "learning_rate": 9.908657894736842e-05,
      "loss": 0.5047,
      "step": 23482
    },
    {
      "epoch": 1.3149848807257252,
      "grad_norm": 1.3131805658340454,
      "learning_rate": 9.908631578947369e-05,
      "loss": 0.4517,
      "step": 23483
    },
    {
      "epoch": 1.3150408780378542,
      "grad_norm": 1.6353704929351807,
      "learning_rate": 9.908605263157895e-05,
      "loss": 0.396,
      "step": 23484
    },
    {
      "epoch": 1.3150968753499832,
      "grad_norm": 1.500420093536377,
      "learning_rate": 9.908578947368423e-05,
      "loss": 0.5526,
      "step": 23485
    },
    {
      "epoch": 1.3151528726621122,
      "grad_norm": 1.3532007932662964,
      "learning_rate": 9.908552631578948e-05,
      "loss": 0.4128,
      "step": 23486
    },
    {
      "epoch": 1.3152088699742412,
      "grad_norm": 1.5647958517074585,
      "learning_rate": 9.908526315789473e-05,
      "loss": 0.5633,
      "step": 23487
    },
    {
      "epoch": 1.3152648672863703,
      "grad_norm": 1.302091121673584,
      "learning_rate": 9.9085e-05,
      "loss": 0.4586,
      "step": 23488
    },
    {
      "epoch": 1.3153208645984993,
      "grad_norm": 1.4732964038848877,
      "learning_rate": 9.908473684210526e-05,
      "loss": 0.5431,
      "step": 23489
    },
    {
      "epoch": 1.3153768619106283,
      "grad_norm": 1.4219179153442383,
      "learning_rate": 9.908447368421054e-05,
      "loss": 0.5483,
      "step": 23490
    },
    {
      "epoch": 1.3154328592227573,
      "grad_norm": 1.3076339960098267,
      "learning_rate": 9.90842105263158e-05,
      "loss": 0.4174,
      "step": 23491
    },
    {
      "epoch": 1.3154888565348863,
      "grad_norm": 1.2017974853515625,
      "learning_rate": 9.908394736842106e-05,
      "loss": 0.4726,
      "step": 23492
    },
    {
      "epoch": 1.3155448538470154,
      "grad_norm": 1.4847337007522583,
      "learning_rate": 9.908368421052632e-05,
      "loss": 0.5449,
      "step": 23493
    },
    {
      "epoch": 1.3156008511591444,
      "grad_norm": 1.6240051984786987,
      "learning_rate": 9.908342105263159e-05,
      "loss": 0.4613,
      "step": 23494
    },
    {
      "epoch": 1.3156568484712734,
      "grad_norm": 1.3387105464935303,
      "learning_rate": 9.908315789473685e-05,
      "loss": 0.5762,
      "step": 23495
    },
    {
      "epoch": 1.3157128457834024,
      "grad_norm": 1.5709962844848633,
      "learning_rate": 9.908289473684211e-05,
      "loss": 0.4868,
      "step": 23496
    },
    {
      "epoch": 1.3157688430955314,
      "grad_norm": 1.442383885383606,
      "learning_rate": 9.908263157894737e-05,
      "loss": 0.516,
      "step": 23497
    },
    {
      "epoch": 1.3158248404076605,
      "grad_norm": 1.1911067962646484,
      "learning_rate": 9.908236842105264e-05,
      "loss": 0.4622,
      "step": 23498
    },
    {
      "epoch": 1.3158808377197895,
      "grad_norm": 1.2875454425811768,
      "learning_rate": 9.90821052631579e-05,
      "loss": 0.3478,
      "step": 23499
    },
    {
      "epoch": 1.3159368350319185,
      "grad_norm": 1.450999140739441,
      "learning_rate": 9.908184210526316e-05,
      "loss": 0.3776,
      "step": 23500
    },
    {
      "epoch": 1.3159928323440475,
      "grad_norm": 1.3420605659484863,
      "learning_rate": 9.908157894736842e-05,
      "loss": 0.3914,
      "step": 23501
    },
    {
      "epoch": 1.3160488296561765,
      "grad_norm": 1.3781684637069702,
      "learning_rate": 9.90813157894737e-05,
      "loss": 0.4826,
      "step": 23502
    },
    {
      "epoch": 1.3161048269683056,
      "grad_norm": 1.4783337116241455,
      "learning_rate": 9.908105263157895e-05,
      "loss": 0.4982,
      "step": 23503
    },
    {
      "epoch": 1.3161608242804346,
      "grad_norm": 1.4523431062698364,
      "learning_rate": 9.908078947368421e-05,
      "loss": 0.4266,
      "step": 23504
    },
    {
      "epoch": 1.3162168215925636,
      "grad_norm": 1.678612470626831,
      "learning_rate": 9.908052631578947e-05,
      "loss": 0.5663,
      "step": 23505
    },
    {
      "epoch": 1.3162728189046926,
      "grad_norm": 1.7157851457595825,
      "learning_rate": 9.908026315789473e-05,
      "loss": 0.4389,
      "step": 23506
    },
    {
      "epoch": 1.3163288162168216,
      "grad_norm": 1.1345454454421997,
      "learning_rate": 9.908000000000001e-05,
      "loss": 0.3387,
      "step": 23507
    },
    {
      "epoch": 1.3163848135289506,
      "grad_norm": 1.5600314140319824,
      "learning_rate": 9.907973684210527e-05,
      "loss": 0.4027,
      "step": 23508
    },
    {
      "epoch": 1.3164408108410797,
      "grad_norm": 1.4446918964385986,
      "learning_rate": 9.907947368421054e-05,
      "loss": 0.3934,
      "step": 23509
    },
    {
      "epoch": 1.3164968081532087,
      "grad_norm": 1.4340838193893433,
      "learning_rate": 9.907921052631579e-05,
      "loss": 0.4871,
      "step": 23510
    },
    {
      "epoch": 1.3165528054653377,
      "grad_norm": 1.360880732536316,
      "learning_rate": 9.907894736842106e-05,
      "loss": 0.4362,
      "step": 23511
    },
    {
      "epoch": 1.3166088027774667,
      "grad_norm": 1.3731484413146973,
      "learning_rate": 9.907868421052632e-05,
      "loss": 0.3982,
      "step": 23512
    },
    {
      "epoch": 1.3166648000895957,
      "grad_norm": 1.2852195501327515,
      "learning_rate": 9.907842105263159e-05,
      "loss": 0.4248,
      "step": 23513
    },
    {
      "epoch": 1.3167207974017248,
      "grad_norm": 1.6905219554901123,
      "learning_rate": 9.907815789473684e-05,
      "loss": 0.466,
      "step": 23514
    },
    {
      "epoch": 1.3167767947138538,
      "grad_norm": 1.1920872926712036,
      "learning_rate": 9.907789473684211e-05,
      "loss": 0.3411,
      "step": 23515
    },
    {
      "epoch": 1.3168327920259828,
      "grad_norm": 1.4498956203460693,
      "learning_rate": 9.907763157894737e-05,
      "loss": 0.4507,
      "step": 23516
    },
    {
      "epoch": 1.3168887893381118,
      "grad_norm": 1.4717071056365967,
      "learning_rate": 9.907736842105264e-05,
      "loss": 0.635,
      "step": 23517
    },
    {
      "epoch": 1.3169447866502408,
      "grad_norm": 1.6787028312683105,
      "learning_rate": 9.90771052631579e-05,
      "loss": 0.4732,
      "step": 23518
    },
    {
      "epoch": 1.3170007839623699,
      "grad_norm": 1.3689689636230469,
      "learning_rate": 9.907684210526316e-05,
      "loss": 0.3653,
      "step": 23519
    },
    {
      "epoch": 1.3170567812744989,
      "grad_norm": 1.5186647176742554,
      "learning_rate": 9.907657894736842e-05,
      "loss": 0.5114,
      "step": 23520
    },
    {
      "epoch": 1.317112778586628,
      "grad_norm": 1.6663926839828491,
      "learning_rate": 9.907631578947368e-05,
      "loss": 0.5142,
      "step": 23521
    },
    {
      "epoch": 1.317168775898757,
      "grad_norm": 1.1840616464614868,
      "learning_rate": 9.907605263157896e-05,
      "loss": 0.3912,
      "step": 23522
    },
    {
      "epoch": 1.317224773210886,
      "grad_norm": 1.6792411804199219,
      "learning_rate": 9.907578947368422e-05,
      "loss": 0.4961,
      "step": 23523
    },
    {
      "epoch": 1.317280770523015,
      "grad_norm": 1.4093555212020874,
      "learning_rate": 9.907552631578948e-05,
      "loss": 0.4001,
      "step": 23524
    },
    {
      "epoch": 1.317336767835144,
      "grad_norm": 1.30805242061615,
      "learning_rate": 9.907526315789474e-05,
      "loss": 0.4575,
      "step": 23525
    },
    {
      "epoch": 1.317392765147273,
      "grad_norm": 1.1401379108428955,
      "learning_rate": 9.907500000000001e-05,
      "loss": 0.2999,
      "step": 23526
    },
    {
      "epoch": 1.317448762459402,
      "grad_norm": 1.3849704265594482,
      "learning_rate": 9.907473684210527e-05,
      "loss": 0.419,
      "step": 23527
    },
    {
      "epoch": 1.317504759771531,
      "grad_norm": 1.422619104385376,
      "learning_rate": 9.907447368421053e-05,
      "loss": 0.4429,
      "step": 23528
    },
    {
      "epoch": 1.31756075708366,
      "grad_norm": 2.4414379596710205,
      "learning_rate": 9.907421052631579e-05,
      "loss": 0.5035,
      "step": 23529
    },
    {
      "epoch": 1.317616754395789,
      "grad_norm": 1.6046645641326904,
      "learning_rate": 9.907394736842106e-05,
      "loss": 0.4716,
      "step": 23530
    },
    {
      "epoch": 1.317672751707918,
      "grad_norm": 1.6596109867095947,
      "learning_rate": 9.907368421052632e-05,
      "loss": 0.6331,
      "step": 23531
    },
    {
      "epoch": 1.317728749020047,
      "grad_norm": 1.4059385061264038,
      "learning_rate": 9.907342105263158e-05,
      "loss": 0.6395,
      "step": 23532
    },
    {
      "epoch": 1.3177847463321761,
      "grad_norm": 1.3787490129470825,
      "learning_rate": 9.907315789473684e-05,
      "loss": 0.4719,
      "step": 23533
    },
    {
      "epoch": 1.3178407436443051,
      "grad_norm": 1.3036671876907349,
      "learning_rate": 9.907289473684211e-05,
      "loss": 0.5903,
      "step": 23534
    },
    {
      "epoch": 1.3178967409564342,
      "grad_norm": 1.2987549304962158,
      "learning_rate": 9.907263157894737e-05,
      "loss": 0.4609,
      "step": 23535
    },
    {
      "epoch": 1.3179527382685632,
      "grad_norm": 1.3021806478500366,
      "learning_rate": 9.907236842105265e-05,
      "loss": 0.4061,
      "step": 23536
    },
    {
      "epoch": 1.3180087355806922,
      "grad_norm": 1.3273011445999146,
      "learning_rate": 9.90721052631579e-05,
      "loss": 0.5937,
      "step": 23537
    },
    {
      "epoch": 1.3180647328928212,
      "grad_norm": 1.6703534126281738,
      "learning_rate": 9.907184210526315e-05,
      "loss": 0.6635,
      "step": 23538
    },
    {
      "epoch": 1.3181207302049502,
      "grad_norm": 1.2274681329727173,
      "learning_rate": 9.907157894736843e-05,
      "loss": 0.418,
      "step": 23539
    },
    {
      "epoch": 1.3181767275170793,
      "grad_norm": 1.3198310136795044,
      "learning_rate": 9.907131578947369e-05,
      "loss": 0.516,
      "step": 23540
    },
    {
      "epoch": 1.3182327248292083,
      "grad_norm": 0.9746121168136597,
      "learning_rate": 9.907105263157896e-05,
      "loss": 0.3675,
      "step": 23541
    },
    {
      "epoch": 1.3182887221413373,
      "grad_norm": 1.48617684841156,
      "learning_rate": 9.90707894736842e-05,
      "loss": 0.4091,
      "step": 23542
    },
    {
      "epoch": 1.3183447194534663,
      "grad_norm": 1.5356407165527344,
      "learning_rate": 9.907052631578948e-05,
      "loss": 0.5064,
      "step": 23543
    },
    {
      "epoch": 1.3184007167655953,
      "grad_norm": 1.353983998298645,
      "learning_rate": 9.907026315789474e-05,
      "loss": 0.5867,
      "step": 23544
    },
    {
      "epoch": 1.3184567140777244,
      "grad_norm": 1.3710651397705078,
      "learning_rate": 9.907000000000001e-05,
      "loss": 0.515,
      "step": 23545
    },
    {
      "epoch": 1.3185127113898534,
      "grad_norm": 1.3370361328125,
      "learning_rate": 9.906973684210527e-05,
      "loss": 0.3541,
      "step": 23546
    },
    {
      "epoch": 1.3185687087019824,
      "grad_norm": 1.3643617630004883,
      "learning_rate": 9.906947368421053e-05,
      "loss": 0.3977,
      "step": 23547
    },
    {
      "epoch": 1.3186247060141114,
      "grad_norm": 1.2992652654647827,
      "learning_rate": 9.906921052631579e-05,
      "loss": 0.5181,
      "step": 23548
    },
    {
      "epoch": 1.3186807033262404,
      "grad_norm": 2.7935121059417725,
      "learning_rate": 9.906894736842106e-05,
      "loss": 0.4438,
      "step": 23549
    },
    {
      "epoch": 1.3187367006383695,
      "grad_norm": 1.092849612236023,
      "learning_rate": 9.906868421052632e-05,
      "loss": 0.3636,
      "step": 23550
    },
    {
      "epoch": 1.3187926979504985,
      "grad_norm": 1.3572300672531128,
      "learning_rate": 9.906842105263158e-05,
      "loss": 0.3727,
      "step": 23551
    },
    {
      "epoch": 1.3188486952626275,
      "grad_norm": 1.3177443742752075,
      "learning_rate": 9.906815789473684e-05,
      "loss": 0.4337,
      "step": 23552
    },
    {
      "epoch": 1.3189046925747565,
      "grad_norm": 1.4217324256896973,
      "learning_rate": 9.906789473684212e-05,
      "loss": 0.4441,
      "step": 23553
    },
    {
      "epoch": 1.3189606898868855,
      "grad_norm": 1.5611952543258667,
      "learning_rate": 9.906763157894738e-05,
      "loss": 0.4666,
      "step": 23554
    },
    {
      "epoch": 1.3190166871990145,
      "grad_norm": 1.3093900680541992,
      "learning_rate": 9.906736842105264e-05,
      "loss": 0.4245,
      "step": 23555
    },
    {
      "epoch": 1.3190726845111436,
      "grad_norm": 1.3968415260314941,
      "learning_rate": 9.90671052631579e-05,
      "loss": 0.4779,
      "step": 23556
    },
    {
      "epoch": 1.3191286818232726,
      "grad_norm": 1.3176441192626953,
      "learning_rate": 9.906684210526316e-05,
      "loss": 0.5966,
      "step": 23557
    },
    {
      "epoch": 1.3191846791354016,
      "grad_norm": 1.1914488077163696,
      "learning_rate": 9.906657894736843e-05,
      "loss": 0.3269,
      "step": 23558
    },
    {
      "epoch": 1.3192406764475306,
      "grad_norm": 1.5402307510375977,
      "learning_rate": 9.906631578947369e-05,
      "loss": 0.4766,
      "step": 23559
    },
    {
      "epoch": 1.3192966737596596,
      "grad_norm": 1.4681423902511597,
      "learning_rate": 9.906605263157895e-05,
      "loss": 0.6561,
      "step": 23560
    },
    {
      "epoch": 1.3193526710717887,
      "grad_norm": 1.2147117853164673,
      "learning_rate": 9.906578947368421e-05,
      "loss": 0.3518,
      "step": 23561
    },
    {
      "epoch": 1.3194086683839177,
      "grad_norm": 1.1922838687896729,
      "learning_rate": 9.906552631578948e-05,
      "loss": 0.4168,
      "step": 23562
    },
    {
      "epoch": 1.3194646656960467,
      "grad_norm": 1.4813134670257568,
      "learning_rate": 9.906526315789474e-05,
      "loss": 0.4051,
      "step": 23563
    },
    {
      "epoch": 1.3195206630081757,
      "grad_norm": 1.6255486011505127,
      "learning_rate": 9.9065e-05,
      "loss": 0.6109,
      "step": 23564
    },
    {
      "epoch": 1.3195766603203047,
      "grad_norm": 1.5070154666900635,
      "learning_rate": 9.906473684210526e-05,
      "loss": 0.5359,
      "step": 23565
    },
    {
      "epoch": 1.3196326576324338,
      "grad_norm": 1.2078232765197754,
      "learning_rate": 9.906447368421053e-05,
      "loss": 0.5,
      "step": 23566
    },
    {
      "epoch": 1.3196886549445628,
      "grad_norm": 1.221826195716858,
      "learning_rate": 9.90642105263158e-05,
      "loss": 0.4715,
      "step": 23567
    },
    {
      "epoch": 1.3197446522566918,
      "grad_norm": 1.1665562391281128,
      "learning_rate": 9.906394736842107e-05,
      "loss": 0.4743,
      "step": 23568
    },
    {
      "epoch": 1.3198006495688208,
      "grad_norm": 1.7769436836242676,
      "learning_rate": 9.906368421052631e-05,
      "loss": 0.651,
      "step": 23569
    },
    {
      "epoch": 1.3198566468809498,
      "grad_norm": 1.6282001733779907,
      "learning_rate": 9.906342105263159e-05,
      "loss": 0.5087,
      "step": 23570
    },
    {
      "epoch": 1.3199126441930789,
      "grad_norm": 1.4367485046386719,
      "learning_rate": 9.906315789473685e-05,
      "loss": 0.5163,
      "step": 23571
    },
    {
      "epoch": 1.3199686415052079,
      "grad_norm": 1.1818881034851074,
      "learning_rate": 9.90628947368421e-05,
      "loss": 0.4329,
      "step": 23572
    },
    {
      "epoch": 1.320024638817337,
      "grad_norm": 1.3425483703613281,
      "learning_rate": 9.906263157894738e-05,
      "loss": 0.4593,
      "step": 23573
    },
    {
      "epoch": 1.320080636129466,
      "grad_norm": 1.2086644172668457,
      "learning_rate": 9.906236842105263e-05,
      "loss": 0.497,
      "step": 23574
    },
    {
      "epoch": 1.320136633441595,
      "grad_norm": 1.4390406608581543,
      "learning_rate": 9.90621052631579e-05,
      "loss": 0.607,
      "step": 23575
    },
    {
      "epoch": 1.320192630753724,
      "grad_norm": 1.5852943658828735,
      "learning_rate": 9.906184210526316e-05,
      "loss": 0.4744,
      "step": 23576
    },
    {
      "epoch": 1.320248628065853,
      "grad_norm": 1.4295347929000854,
      "learning_rate": 9.906157894736843e-05,
      "loss": 0.3604,
      "step": 23577
    },
    {
      "epoch": 1.320304625377982,
      "grad_norm": 1.4102352857589722,
      "learning_rate": 9.906131578947369e-05,
      "loss": 0.4666,
      "step": 23578
    },
    {
      "epoch": 1.320360622690111,
      "grad_norm": 1.2955905199050903,
      "learning_rate": 9.906105263157895e-05,
      "loss": 0.3701,
      "step": 23579
    },
    {
      "epoch": 1.32041662000224,
      "grad_norm": 1.466480016708374,
      "learning_rate": 9.906078947368421e-05,
      "loss": 0.7739,
      "step": 23580
    },
    {
      "epoch": 1.320472617314369,
      "grad_norm": 1.4667768478393555,
      "learning_rate": 9.906052631578948e-05,
      "loss": 0.6104,
      "step": 23581
    },
    {
      "epoch": 1.320528614626498,
      "grad_norm": 1.45987868309021,
      "learning_rate": 9.906026315789474e-05,
      "loss": 0.5134,
      "step": 23582
    },
    {
      "epoch": 1.320584611938627,
      "grad_norm": 1.2193151712417603,
      "learning_rate": 9.906e-05,
      "loss": 0.5586,
      "step": 23583
    },
    {
      "epoch": 1.3206406092507559,
      "grad_norm": 1.5995523929595947,
      "learning_rate": 9.905973684210526e-05,
      "loss": 0.4621,
      "step": 23584
    },
    {
      "epoch": 1.320696606562885,
      "grad_norm": 1.2353379726409912,
      "learning_rate": 9.905947368421054e-05,
      "loss": 0.514,
      "step": 23585
    },
    {
      "epoch": 1.320752603875014,
      "grad_norm": 1.2061514854431152,
      "learning_rate": 9.90592105263158e-05,
      "loss": 0.4795,
      "step": 23586
    },
    {
      "epoch": 1.320808601187143,
      "grad_norm": 1.2943384647369385,
      "learning_rate": 9.905894736842106e-05,
      "loss": 0.3866,
      "step": 23587
    },
    {
      "epoch": 1.320864598499272,
      "grad_norm": 1.351973056793213,
      "learning_rate": 9.905868421052632e-05,
      "loss": 0.4919,
      "step": 23588
    },
    {
      "epoch": 1.320920595811401,
      "grad_norm": 1.2900298833847046,
      "learning_rate": 9.905842105263159e-05,
      "loss": 0.3786,
      "step": 23589
    },
    {
      "epoch": 1.32097659312353,
      "grad_norm": 1.6199102401733398,
      "learning_rate": 9.905815789473685e-05,
      "loss": 0.523,
      "step": 23590
    },
    {
      "epoch": 1.321032590435659,
      "grad_norm": 1.2421436309814453,
      "learning_rate": 9.905789473684211e-05,
      "loss": 0.5326,
      "step": 23591
    },
    {
      "epoch": 1.321088587747788,
      "grad_norm": 1.2802703380584717,
      "learning_rate": 9.905763157894737e-05,
      "loss": 0.4736,
      "step": 23592
    },
    {
      "epoch": 1.321144585059917,
      "grad_norm": 1.3789219856262207,
      "learning_rate": 9.905736842105263e-05,
      "loss": 0.5246,
      "step": 23593
    },
    {
      "epoch": 1.321200582372046,
      "grad_norm": 2.457754373550415,
      "learning_rate": 9.90571052631579e-05,
      "loss": 0.6164,
      "step": 23594
    },
    {
      "epoch": 1.321256579684175,
      "grad_norm": 1.5327216386795044,
      "learning_rate": 9.905684210526316e-05,
      "loss": 0.4155,
      "step": 23595
    },
    {
      "epoch": 1.3213125769963041,
      "grad_norm": 1.3117456436157227,
      "learning_rate": 9.905657894736843e-05,
      "loss": 0.502,
      "step": 23596
    },
    {
      "epoch": 1.3213685743084331,
      "grad_norm": 1.27993905544281,
      "learning_rate": 9.905631578947368e-05,
      "loss": 0.4094,
      "step": 23597
    },
    {
      "epoch": 1.3214245716205621,
      "grad_norm": 1.54668128490448,
      "learning_rate": 9.905605263157895e-05,
      "loss": 0.3735,
      "step": 23598
    },
    {
      "epoch": 1.3214805689326912,
      "grad_norm": 2.8389892578125,
      "learning_rate": 9.905578947368421e-05,
      "loss": 0.5307,
      "step": 23599
    },
    {
      "epoch": 1.3215365662448202,
      "grad_norm": 1.3819185495376587,
      "learning_rate": 9.905552631578949e-05,
      "loss": 0.5652,
      "step": 23600
    },
    {
      "epoch": 1.3215925635569492,
      "grad_norm": 1.5571867227554321,
      "learning_rate": 9.905526315789475e-05,
      "loss": 0.5315,
      "step": 23601
    },
    {
      "epoch": 1.3216485608690782,
      "grad_norm": 1.5231329202651978,
      "learning_rate": 9.9055e-05,
      "loss": 0.4304,
      "step": 23602
    },
    {
      "epoch": 1.3217045581812072,
      "grad_norm": 1.5173362493515015,
      "learning_rate": 9.905473684210527e-05,
      "loss": 0.3471,
      "step": 23603
    },
    {
      "epoch": 1.3217605554933363,
      "grad_norm": 1.2694761753082275,
      "learning_rate": 9.905447368421054e-05,
      "loss": 0.5271,
      "step": 23604
    },
    {
      "epoch": 1.3218165528054653,
      "grad_norm": 1.2435005903244019,
      "learning_rate": 9.90542105263158e-05,
      "loss": 0.3912,
      "step": 23605
    },
    {
      "epoch": 1.3218725501175943,
      "grad_norm": 1.2682331800460815,
      "learning_rate": 9.905394736842106e-05,
      "loss": 0.4828,
      "step": 23606
    },
    {
      "epoch": 1.3219285474297233,
      "grad_norm": 1.5103245973587036,
      "learning_rate": 9.905368421052632e-05,
      "loss": 0.4391,
      "step": 23607
    },
    {
      "epoch": 1.3219845447418523,
      "grad_norm": 1.295146107673645,
      "learning_rate": 9.905342105263158e-05,
      "loss": 0.3748,
      "step": 23608
    },
    {
      "epoch": 1.3220405420539814,
      "grad_norm": 1.4384539127349854,
      "learning_rate": 9.905315789473685e-05,
      "loss": 0.5867,
      "step": 23609
    },
    {
      "epoch": 1.3220965393661104,
      "grad_norm": 1.448047399520874,
      "learning_rate": 9.905289473684211e-05,
      "loss": 0.4441,
      "step": 23610
    },
    {
      "epoch": 1.3221525366782394,
      "grad_norm": 1.8027104139328003,
      "learning_rate": 9.905263157894737e-05,
      "loss": 0.5361,
      "step": 23611
    },
    {
      "epoch": 1.3222085339903684,
      "grad_norm": 1.6681878566741943,
      "learning_rate": 9.905236842105263e-05,
      "loss": 0.5267,
      "step": 23612
    },
    {
      "epoch": 1.3222645313024974,
      "grad_norm": 1.4393134117126465,
      "learning_rate": 9.90521052631579e-05,
      "loss": 0.7328,
      "step": 23613
    },
    {
      "epoch": 1.3223205286146265,
      "grad_norm": 1.2529213428497314,
      "learning_rate": 9.905184210526316e-05,
      "loss": 0.4681,
      "step": 23614
    },
    {
      "epoch": 1.3223765259267555,
      "grad_norm": 1.8222781419754028,
      "learning_rate": 9.905157894736842e-05,
      "loss": 0.5243,
      "step": 23615
    },
    {
      "epoch": 1.3224325232388845,
      "grad_norm": 1.4331270456314087,
      "learning_rate": 9.905131578947368e-05,
      "loss": 0.4813,
      "step": 23616
    },
    {
      "epoch": 1.3224885205510135,
      "grad_norm": 1.3592449426651,
      "learning_rate": 9.905105263157896e-05,
      "loss": 0.5771,
      "step": 23617
    },
    {
      "epoch": 1.3225445178631425,
      "grad_norm": 1.7799861431121826,
      "learning_rate": 9.905078947368422e-05,
      "loss": 0.6256,
      "step": 23618
    },
    {
      "epoch": 1.3226005151752716,
      "grad_norm": 1.8847559690475464,
      "learning_rate": 9.905052631578948e-05,
      "loss": 0.6124,
      "step": 23619
    },
    {
      "epoch": 1.3226565124874006,
      "grad_norm": 1.9002000093460083,
      "learning_rate": 9.905026315789474e-05,
      "loss": 0.5131,
      "step": 23620
    },
    {
      "epoch": 1.3227125097995296,
      "grad_norm": 1.2628037929534912,
      "learning_rate": 9.905000000000001e-05,
      "loss": 0.5107,
      "step": 23621
    },
    {
      "epoch": 1.3227685071116586,
      "grad_norm": 1.39218270778656,
      "learning_rate": 9.904973684210527e-05,
      "loss": 0.4912,
      "step": 23622
    },
    {
      "epoch": 1.3228245044237876,
      "grad_norm": 1.3301101922988892,
      "learning_rate": 9.904947368421054e-05,
      "loss": 0.4268,
      "step": 23623
    },
    {
      "epoch": 1.3228805017359166,
      "grad_norm": 1.2446775436401367,
      "learning_rate": 9.904921052631579e-05,
      "loss": 0.3518,
      "step": 23624
    },
    {
      "epoch": 1.3229364990480457,
      "grad_norm": 1.6048548221588135,
      "learning_rate": 9.904894736842105e-05,
      "loss": 0.5149,
      "step": 23625
    },
    {
      "epoch": 1.3229924963601747,
      "grad_norm": 1.354109764099121,
      "learning_rate": 9.904868421052632e-05,
      "loss": 0.4675,
      "step": 23626
    },
    {
      "epoch": 1.3230484936723037,
      "grad_norm": 1.8846253156661987,
      "learning_rate": 9.904842105263158e-05,
      "loss": 0.48,
      "step": 23627
    },
    {
      "epoch": 1.3231044909844327,
      "grad_norm": 1.7262190580368042,
      "learning_rate": 9.904815789473685e-05,
      "loss": 0.6207,
      "step": 23628
    },
    {
      "epoch": 1.3231604882965617,
      "grad_norm": 1.422507405281067,
      "learning_rate": 9.90478947368421e-05,
      "loss": 0.4041,
      "step": 23629
    },
    {
      "epoch": 1.3232164856086908,
      "grad_norm": 1.5060955286026,
      "learning_rate": 9.904763157894737e-05,
      "loss": 0.454,
      "step": 23630
    },
    {
      "epoch": 1.3232724829208198,
      "grad_norm": 1.2696168422698975,
      "learning_rate": 9.904736842105263e-05,
      "loss": 0.4178,
      "step": 23631
    },
    {
      "epoch": 1.3233284802329488,
      "grad_norm": 18.032329559326172,
      "learning_rate": 9.90471052631579e-05,
      "loss": 0.4343,
      "step": 23632
    },
    {
      "epoch": 1.3233844775450778,
      "grad_norm": 1.286111831665039,
      "learning_rate": 9.904684210526317e-05,
      "loss": 0.5,
      "step": 23633
    },
    {
      "epoch": 1.3234404748572068,
      "grad_norm": 1.2645292282104492,
      "learning_rate": 9.904657894736843e-05,
      "loss": 0.3458,
      "step": 23634
    },
    {
      "epoch": 1.3234964721693359,
      "grad_norm": 1.7569935321807861,
      "learning_rate": 9.904631578947369e-05,
      "loss": 0.4716,
      "step": 23635
    },
    {
      "epoch": 1.3235524694814649,
      "grad_norm": 5.75446891784668,
      "learning_rate": 9.904605263157896e-05,
      "loss": 0.4636,
      "step": 23636
    },
    {
      "epoch": 1.323608466793594,
      "grad_norm": 1.3120683431625366,
      "learning_rate": 9.904578947368422e-05,
      "loss": 0.5186,
      "step": 23637
    },
    {
      "epoch": 1.323664464105723,
      "grad_norm": 1.477587103843689,
      "learning_rate": 9.904552631578948e-05,
      "loss": 0.3833,
      "step": 23638
    },
    {
      "epoch": 1.323720461417852,
      "grad_norm": 2.1010634899139404,
      "learning_rate": 9.904526315789474e-05,
      "loss": 0.5562,
      "step": 23639
    },
    {
      "epoch": 1.323776458729981,
      "grad_norm": 1.5339431762695312,
      "learning_rate": 9.904500000000001e-05,
      "loss": 0.4913,
      "step": 23640
    },
    {
      "epoch": 1.32383245604211,
      "grad_norm": 1.4985063076019287,
      "learning_rate": 9.904473684210527e-05,
      "loss": 0.4716,
      "step": 23641
    },
    {
      "epoch": 1.323888453354239,
      "grad_norm": 1.2650494575500488,
      "learning_rate": 9.904447368421053e-05,
      "loss": 0.4502,
      "step": 23642
    },
    {
      "epoch": 1.323944450666368,
      "grad_norm": 1.399677038192749,
      "learning_rate": 9.904421052631579e-05,
      "loss": 0.453,
      "step": 23643
    },
    {
      "epoch": 1.324000447978497,
      "grad_norm": 1.3583924770355225,
      "learning_rate": 9.904394736842105e-05,
      "loss": 0.4261,
      "step": 23644
    },
    {
      "epoch": 1.324056445290626,
      "grad_norm": 1.2457411289215088,
      "learning_rate": 9.904368421052632e-05,
      "loss": 0.4543,
      "step": 23645
    },
    {
      "epoch": 1.324112442602755,
      "grad_norm": 1.3103523254394531,
      "learning_rate": 9.904342105263158e-05,
      "loss": 0.4961,
      "step": 23646
    },
    {
      "epoch": 1.324168439914884,
      "grad_norm": 2.0466365814208984,
      "learning_rate": 9.904315789473684e-05,
      "loss": 0.5445,
      "step": 23647
    },
    {
      "epoch": 1.324224437227013,
      "grad_norm": 1.5205439329147339,
      "learning_rate": 9.90428947368421e-05,
      "loss": 0.4668,
      "step": 23648
    },
    {
      "epoch": 1.3242804345391421,
      "grad_norm": 1.236509919166565,
      "learning_rate": 9.904263157894738e-05,
      "loss": 0.428,
      "step": 23649
    },
    {
      "epoch": 1.3243364318512711,
      "grad_norm": 1.7488242387771606,
      "learning_rate": 9.904236842105264e-05,
      "loss": 0.523,
      "step": 23650
    },
    {
      "epoch": 1.3243924291634002,
      "grad_norm": 1.2344638109207153,
      "learning_rate": 9.904210526315791e-05,
      "loss": 0.6077,
      "step": 23651
    },
    {
      "epoch": 1.3244484264755292,
      "grad_norm": 1.4946460723876953,
      "learning_rate": 9.904184210526316e-05,
      "loss": 0.4272,
      "step": 23652
    },
    {
      "epoch": 1.3245044237876582,
      "grad_norm": 1.2750486135482788,
      "learning_rate": 9.904157894736843e-05,
      "loss": 0.3809,
      "step": 23653
    },
    {
      "epoch": 1.3245604210997872,
      "grad_norm": 1.2810460329055786,
      "learning_rate": 9.904131578947369e-05,
      "loss": 0.45,
      "step": 23654
    },
    {
      "epoch": 1.3246164184119162,
      "grad_norm": 1.2462742328643799,
      "learning_rate": 9.904105263157896e-05,
      "loss": 0.452,
      "step": 23655
    },
    {
      "epoch": 1.3246724157240453,
      "grad_norm": 1.3404079675674438,
      "learning_rate": 9.904078947368422e-05,
      "loss": 0.426,
      "step": 23656
    },
    {
      "epoch": 1.3247284130361743,
      "grad_norm": 2.8471829891204834,
      "learning_rate": 9.904052631578948e-05,
      "loss": 0.4118,
      "step": 23657
    },
    {
      "epoch": 1.3247844103483033,
      "grad_norm": 1.2869397401809692,
      "learning_rate": 9.904026315789474e-05,
      "loss": 0.3979,
      "step": 23658
    },
    {
      "epoch": 1.3248404076604323,
      "grad_norm": 1.3661129474639893,
      "learning_rate": 9.904e-05,
      "loss": 0.5137,
      "step": 23659
    },
    {
      "epoch": 1.3248964049725613,
      "grad_norm": 1.7654578685760498,
      "learning_rate": 9.903973684210527e-05,
      "loss": 0.5692,
      "step": 23660
    },
    {
      "epoch": 1.3249524022846904,
      "grad_norm": 1.4775665998458862,
      "learning_rate": 9.903947368421052e-05,
      "loss": 0.4675,
      "step": 23661
    },
    {
      "epoch": 1.3250083995968194,
      "grad_norm": 1.3470083475112915,
      "learning_rate": 9.903921052631579e-05,
      "loss": 0.461,
      "step": 23662
    },
    {
      "epoch": 1.3250643969089484,
      "grad_norm": 1.2037103176116943,
      "learning_rate": 9.903894736842105e-05,
      "loss": 0.3816,
      "step": 23663
    },
    {
      "epoch": 1.3251203942210774,
      "grad_norm": 1.2058035135269165,
      "learning_rate": 9.903868421052633e-05,
      "loss": 0.372,
      "step": 23664
    },
    {
      "epoch": 1.3251763915332064,
      "grad_norm": 1.8195207118988037,
      "learning_rate": 9.903842105263159e-05,
      "loss": 0.5475,
      "step": 23665
    },
    {
      "epoch": 1.3252323888453355,
      "grad_norm": 1.3020286560058594,
      "learning_rate": 9.903815789473685e-05,
      "loss": 0.4242,
      "step": 23666
    },
    {
      "epoch": 1.3252883861574645,
      "grad_norm": 1.6143763065338135,
      "learning_rate": 9.90378947368421e-05,
      "loss": 0.4134,
      "step": 23667
    },
    {
      "epoch": 1.3253443834695935,
      "grad_norm": 1.2927172183990479,
      "learning_rate": 9.903763157894738e-05,
      "loss": 0.3848,
      "step": 23668
    },
    {
      "epoch": 1.3254003807817225,
      "grad_norm": 1.150124430656433,
      "learning_rate": 9.903736842105264e-05,
      "loss": 0.44,
      "step": 23669
    },
    {
      "epoch": 1.3254563780938515,
      "grad_norm": 1.4438631534576416,
      "learning_rate": 9.90371052631579e-05,
      "loss": 0.4656,
      "step": 23670
    },
    {
      "epoch": 1.3255123754059805,
      "grad_norm": 1.6583176851272583,
      "learning_rate": 9.903684210526316e-05,
      "loss": 0.5393,
      "step": 23671
    },
    {
      "epoch": 1.3255683727181096,
      "grad_norm": 1.572582721710205,
      "learning_rate": 9.903657894736843e-05,
      "loss": 0.4246,
      "step": 23672
    },
    {
      "epoch": 1.3256243700302386,
      "grad_norm": 1.8211469650268555,
      "learning_rate": 9.903631578947369e-05,
      "loss": 0.5976,
      "step": 23673
    },
    {
      "epoch": 1.3256803673423676,
      "grad_norm": 1.1383578777313232,
      "learning_rate": 9.903605263157895e-05,
      "loss": 0.3202,
      "step": 23674
    },
    {
      "epoch": 1.3257363646544966,
      "grad_norm": 1.207336664199829,
      "learning_rate": 9.903578947368421e-05,
      "loss": 0.4443,
      "step": 23675
    },
    {
      "epoch": 1.3257923619666256,
      "grad_norm": 1.387158751487732,
      "learning_rate": 9.903552631578947e-05,
      "loss": 0.5514,
      "step": 23676
    },
    {
      "epoch": 1.3258483592787547,
      "grad_norm": 1.4295538663864136,
      "learning_rate": 9.903526315789474e-05,
      "loss": 0.4565,
      "step": 23677
    },
    {
      "epoch": 1.3259043565908837,
      "grad_norm": 1.262218952178955,
      "learning_rate": 9.9035e-05,
      "loss": 0.3952,
      "step": 23678
    },
    {
      "epoch": 1.3259603539030127,
      "grad_norm": 2.4699628353118896,
      "learning_rate": 9.903473684210526e-05,
      "loss": 0.571,
      "step": 23679
    },
    {
      "epoch": 1.3260163512151417,
      "grad_norm": 1.5111093521118164,
      "learning_rate": 9.903447368421052e-05,
      "loss": 0.4995,
      "step": 23680
    },
    {
      "epoch": 1.3260723485272707,
      "grad_norm": 1.4468804597854614,
      "learning_rate": 9.90342105263158e-05,
      "loss": 0.4995,
      "step": 23681
    },
    {
      "epoch": 1.3261283458393998,
      "grad_norm": 1.4177590608596802,
      "learning_rate": 9.903394736842106e-05,
      "loss": 0.4481,
      "step": 23682
    },
    {
      "epoch": 1.3261843431515288,
      "grad_norm": 1.2426563501358032,
      "learning_rate": 9.903368421052633e-05,
      "loss": 0.6114,
      "step": 23683
    },
    {
      "epoch": 1.3262403404636578,
      "grad_norm": 1.3937419652938843,
      "learning_rate": 9.903342105263157e-05,
      "loss": 0.5069,
      "step": 23684
    },
    {
      "epoch": 1.3262963377757868,
      "grad_norm": 1.3842486143112183,
      "learning_rate": 9.903315789473685e-05,
      "loss": 0.5955,
      "step": 23685
    },
    {
      "epoch": 1.3263523350879158,
      "grad_norm": 1.3546769618988037,
      "learning_rate": 9.903289473684211e-05,
      "loss": 0.4685,
      "step": 23686
    },
    {
      "epoch": 1.3264083324000449,
      "grad_norm": 1.254425287246704,
      "learning_rate": 9.903263157894738e-05,
      "loss": 0.5517,
      "step": 23687
    },
    {
      "epoch": 1.3264643297121739,
      "grad_norm": 1.3651771545410156,
      "learning_rate": 9.903236842105264e-05,
      "loss": 0.4507,
      "step": 23688
    },
    {
      "epoch": 1.326520327024303,
      "grad_norm": 1.5061709880828857,
      "learning_rate": 9.90321052631579e-05,
      "loss": 0.5048,
      "step": 23689
    },
    {
      "epoch": 1.326576324336432,
      "grad_norm": 1.3575242757797241,
      "learning_rate": 9.903184210526316e-05,
      "loss": 0.5553,
      "step": 23690
    },
    {
      "epoch": 1.326632321648561,
      "grad_norm": 1.389123558998108,
      "learning_rate": 9.903157894736843e-05,
      "loss": 0.5087,
      "step": 23691
    },
    {
      "epoch": 1.32668831896069,
      "grad_norm": 1.5058927536010742,
      "learning_rate": 9.90313157894737e-05,
      "loss": 0.4397,
      "step": 23692
    },
    {
      "epoch": 1.326744316272819,
      "grad_norm": 1.1221836805343628,
      "learning_rate": 9.903105263157895e-05,
      "loss": 0.4602,
      "step": 23693
    },
    {
      "epoch": 1.326800313584948,
      "grad_norm": 1.5292891263961792,
      "learning_rate": 9.903078947368421e-05,
      "loss": 0.6224,
      "step": 23694
    },
    {
      "epoch": 1.326856310897077,
      "grad_norm": 1.6511558294296265,
      "learning_rate": 9.903052631578947e-05,
      "loss": 0.4988,
      "step": 23695
    },
    {
      "epoch": 1.326912308209206,
      "grad_norm": 1.4084031581878662,
      "learning_rate": 9.903026315789475e-05,
      "loss": 0.4657,
      "step": 23696
    },
    {
      "epoch": 1.326968305521335,
      "grad_norm": 1.500809669494629,
      "learning_rate": 9.903e-05,
      "loss": 0.5624,
      "step": 23697
    },
    {
      "epoch": 1.327024302833464,
      "grad_norm": 1.824904441833496,
      "learning_rate": 9.902973684210527e-05,
      "loss": 0.5121,
      "step": 23698
    },
    {
      "epoch": 1.327080300145593,
      "grad_norm": 1.3979873657226562,
      "learning_rate": 9.902947368421053e-05,
      "loss": 0.3647,
      "step": 23699
    },
    {
      "epoch": 1.327136297457722,
      "grad_norm": 1.4668703079223633,
      "learning_rate": 9.90292105263158e-05,
      "loss": 0.5255,
      "step": 23700
    },
    {
      "epoch": 1.3271922947698511,
      "grad_norm": 1.708409070968628,
      "learning_rate": 9.902894736842106e-05,
      "loss": 0.5224,
      "step": 23701
    },
    {
      "epoch": 1.3272482920819801,
      "grad_norm": 1.5218987464904785,
      "learning_rate": 9.902868421052632e-05,
      "loss": 0.373,
      "step": 23702
    },
    {
      "epoch": 1.3273042893941092,
      "grad_norm": 1.4584457874298096,
      "learning_rate": 9.902842105263158e-05,
      "loss": 0.4978,
      "step": 23703
    },
    {
      "epoch": 1.3273602867062382,
      "grad_norm": 1.9057399034500122,
      "learning_rate": 9.902815789473685e-05,
      "loss": 0.4955,
      "step": 23704
    },
    {
      "epoch": 1.3274162840183672,
      "grad_norm": 1.1189838647842407,
      "learning_rate": 9.902789473684211e-05,
      "loss": 0.3878,
      "step": 23705
    },
    {
      "epoch": 1.3274722813304962,
      "grad_norm": 1.8672751188278198,
      "learning_rate": 9.902763157894738e-05,
      "loss": 0.4425,
      "step": 23706
    },
    {
      "epoch": 1.3275282786426252,
      "grad_norm": 1.2922204732894897,
      "learning_rate": 9.902736842105263e-05,
      "loss": 0.4279,
      "step": 23707
    },
    {
      "epoch": 1.327584275954754,
      "grad_norm": 1.1615028381347656,
      "learning_rate": 9.90271052631579e-05,
      "loss": 0.3908,
      "step": 23708
    },
    {
      "epoch": 1.327640273266883,
      "grad_norm": 1.163578748703003,
      "learning_rate": 9.902684210526316e-05,
      "loss": 0.4818,
      "step": 23709
    },
    {
      "epoch": 1.327696270579012,
      "grad_norm": 1.2902050018310547,
      "learning_rate": 9.902657894736844e-05,
      "loss": 0.4775,
      "step": 23710
    },
    {
      "epoch": 1.327752267891141,
      "grad_norm": 1.1960757970809937,
      "learning_rate": 9.90263157894737e-05,
      "loss": 0.3787,
      "step": 23711
    },
    {
      "epoch": 1.32780826520327,
      "grad_norm": 1.5162867307662964,
      "learning_rate": 9.902605263157894e-05,
      "loss": 0.5849,
      "step": 23712
    },
    {
      "epoch": 1.3278642625153991,
      "grad_norm": 1.2826370000839233,
      "learning_rate": 9.902578947368422e-05,
      "loss": 0.4882,
      "step": 23713
    },
    {
      "epoch": 1.3279202598275281,
      "grad_norm": 1.2803281545639038,
      "learning_rate": 9.902552631578948e-05,
      "loss": 0.4433,
      "step": 23714
    },
    {
      "epoch": 1.3279762571396572,
      "grad_norm": 2.0065600872039795,
      "learning_rate": 9.902526315789475e-05,
      "loss": 0.5693,
      "step": 23715
    },
    {
      "epoch": 1.3280322544517862,
      "grad_norm": 1.1913907527923584,
      "learning_rate": 9.9025e-05,
      "loss": 0.4042,
      "step": 23716
    },
    {
      "epoch": 1.3280882517639152,
      "grad_norm": 1.5494247674942017,
      "learning_rate": 9.902473684210527e-05,
      "loss": 0.4673,
      "step": 23717
    },
    {
      "epoch": 1.3281442490760442,
      "grad_norm": 1.230475664138794,
      "learning_rate": 9.902447368421053e-05,
      "loss": 0.4799,
      "step": 23718
    },
    {
      "epoch": 1.3282002463881732,
      "grad_norm": 1.455629587173462,
      "learning_rate": 9.90242105263158e-05,
      "loss": 0.5006,
      "step": 23719
    },
    {
      "epoch": 1.3282562437003023,
      "grad_norm": 1.6144088506698608,
      "learning_rate": 9.902394736842106e-05,
      "loss": 0.4582,
      "step": 23720
    },
    {
      "epoch": 1.3283122410124313,
      "grad_norm": 1.2861231565475464,
      "learning_rate": 9.902368421052632e-05,
      "loss": 0.4438,
      "step": 23721
    },
    {
      "epoch": 1.3283682383245603,
      "grad_norm": 1.314123272895813,
      "learning_rate": 9.902342105263158e-05,
      "loss": 0.4057,
      "step": 23722
    },
    {
      "epoch": 1.3284242356366893,
      "grad_norm": 1.452102780342102,
      "learning_rate": 9.902315789473685e-05,
      "loss": 0.4223,
      "step": 23723
    },
    {
      "epoch": 1.3284802329488183,
      "grad_norm": 1.2253562211990356,
      "learning_rate": 9.902289473684211e-05,
      "loss": 0.501,
      "step": 23724
    },
    {
      "epoch": 1.3285362302609474,
      "grad_norm": 1.557931661605835,
      "learning_rate": 9.902263157894737e-05,
      "loss": 0.4189,
      "step": 23725
    },
    {
      "epoch": 1.3285922275730764,
      "grad_norm": 1.4173012971878052,
      "learning_rate": 9.902236842105263e-05,
      "loss": 0.4374,
      "step": 23726
    },
    {
      "epoch": 1.3286482248852054,
      "grad_norm": 1.6278053522109985,
      "learning_rate": 9.90221052631579e-05,
      "loss": 0.4733,
      "step": 23727
    },
    {
      "epoch": 1.3287042221973344,
      "grad_norm": 2.0490424633026123,
      "learning_rate": 9.902184210526317e-05,
      "loss": 0.4737,
      "step": 23728
    },
    {
      "epoch": 1.3287602195094634,
      "grad_norm": 1.7847868204116821,
      "learning_rate": 9.902157894736843e-05,
      "loss": 0.4162,
      "step": 23729
    },
    {
      "epoch": 1.3288162168215925,
      "grad_norm": 1.4404356479644775,
      "learning_rate": 9.902131578947369e-05,
      "loss": 0.3731,
      "step": 23730
    },
    {
      "epoch": 1.3288722141337215,
      "grad_norm": 1.6273163557052612,
      "learning_rate": 9.902105263157894e-05,
      "loss": 0.5964,
      "step": 23731
    },
    {
      "epoch": 1.3289282114458505,
      "grad_norm": 1.505690097808838,
      "learning_rate": 9.902078947368422e-05,
      "loss": 0.5087,
      "step": 23732
    },
    {
      "epoch": 1.3289842087579795,
      "grad_norm": 1.3277822732925415,
      "learning_rate": 9.902052631578948e-05,
      "loss": 0.6378,
      "step": 23733
    },
    {
      "epoch": 1.3290402060701085,
      "grad_norm": 1.4372085332870483,
      "learning_rate": 9.902026315789474e-05,
      "loss": 0.6282,
      "step": 23734
    },
    {
      "epoch": 1.3290962033822376,
      "grad_norm": 1.8132154941558838,
      "learning_rate": 9.902e-05,
      "loss": 0.5666,
      "step": 23735
    },
    {
      "epoch": 1.3291522006943666,
      "grad_norm": 1.291549563407898,
      "learning_rate": 9.901973684210527e-05,
      "loss": 0.3794,
      "step": 23736
    },
    {
      "epoch": 1.3292081980064956,
      "grad_norm": 1.598890781402588,
      "learning_rate": 9.901947368421053e-05,
      "loss": 0.5991,
      "step": 23737
    },
    {
      "epoch": 1.3292641953186246,
      "grad_norm": 1.4591436386108398,
      "learning_rate": 9.90192105263158e-05,
      "loss": 0.4907,
      "step": 23738
    },
    {
      "epoch": 1.3293201926307536,
      "grad_norm": 1.3021881580352783,
      "learning_rate": 9.901894736842105e-05,
      "loss": 0.458,
      "step": 23739
    },
    {
      "epoch": 1.3293761899428826,
      "grad_norm": 1.9970439672470093,
      "learning_rate": 9.901868421052632e-05,
      "loss": 0.5475,
      "step": 23740
    },
    {
      "epoch": 1.3294321872550117,
      "grad_norm": 1.3215672969818115,
      "learning_rate": 9.901842105263158e-05,
      "loss": 0.4985,
      "step": 23741
    },
    {
      "epoch": 1.3294881845671407,
      "grad_norm": 1.1670761108398438,
      "learning_rate": 9.901815789473686e-05,
      "loss": 0.4236,
      "step": 23742
    },
    {
      "epoch": 1.3295441818792697,
      "grad_norm": 1.6619517803192139,
      "learning_rate": 9.901789473684212e-05,
      "loss": 0.5797,
      "step": 23743
    },
    {
      "epoch": 1.3296001791913987,
      "grad_norm": 1.4194841384887695,
      "learning_rate": 9.901763157894738e-05,
      "loss": 0.4307,
      "step": 23744
    },
    {
      "epoch": 1.3296561765035277,
      "grad_norm": 1.455501914024353,
      "learning_rate": 9.901736842105264e-05,
      "loss": 0.4563,
      "step": 23745
    },
    {
      "epoch": 1.3297121738156568,
      "grad_norm": 1.743912696838379,
      "learning_rate": 9.90171052631579e-05,
      "loss": 0.5251,
      "step": 23746
    },
    {
      "epoch": 1.3297681711277858,
      "grad_norm": 1.3530521392822266,
      "learning_rate": 9.901684210526317e-05,
      "loss": 0.412,
      "step": 23747
    },
    {
      "epoch": 1.3298241684399148,
      "grad_norm": 1.2381137609481812,
      "learning_rate": 9.901657894736843e-05,
      "loss": 0.4692,
      "step": 23748
    },
    {
      "epoch": 1.3298801657520438,
      "grad_norm": 1.2860863208770752,
      "learning_rate": 9.901631578947369e-05,
      "loss": 0.396,
      "step": 23749
    },
    {
      "epoch": 1.3299361630641728,
      "grad_norm": 1.500382423400879,
      "learning_rate": 9.901605263157895e-05,
      "loss": 0.6196,
      "step": 23750
    },
    {
      "epoch": 1.3299921603763019,
      "grad_norm": 1.6573114395141602,
      "learning_rate": 9.901578947368422e-05,
      "loss": 0.4758,
      "step": 23751
    },
    {
      "epoch": 1.3300481576884309,
      "grad_norm": 1.65608811378479,
      "learning_rate": 9.901552631578948e-05,
      "loss": 0.5383,
      "step": 23752
    },
    {
      "epoch": 1.33010415500056,
      "grad_norm": 1.4041887521743774,
      "learning_rate": 9.901526315789474e-05,
      "loss": 0.5479,
      "step": 23753
    },
    {
      "epoch": 1.330160152312689,
      "grad_norm": 1.302161455154419,
      "learning_rate": 9.9015e-05,
      "loss": 0.4635,
      "step": 23754
    },
    {
      "epoch": 1.330216149624818,
      "grad_norm": 1.4354580640792847,
      "learning_rate": 9.901473684210527e-05,
      "loss": 0.4113,
      "step": 23755
    },
    {
      "epoch": 1.330272146936947,
      "grad_norm": 1.409231424331665,
      "learning_rate": 9.901447368421053e-05,
      "loss": 0.4626,
      "step": 23756
    },
    {
      "epoch": 1.330328144249076,
      "grad_norm": 1.9290640354156494,
      "learning_rate": 9.901421052631579e-05,
      "loss": 0.6919,
      "step": 23757
    },
    {
      "epoch": 1.330384141561205,
      "grad_norm": 1.1375243663787842,
      "learning_rate": 9.901394736842105e-05,
      "loss": 0.4853,
      "step": 23758
    },
    {
      "epoch": 1.330440138873334,
      "grad_norm": 1.466916561126709,
      "learning_rate": 9.901368421052633e-05,
      "loss": 0.4337,
      "step": 23759
    },
    {
      "epoch": 1.330496136185463,
      "grad_norm": 1.4821162223815918,
      "learning_rate": 9.901342105263159e-05,
      "loss": 0.4767,
      "step": 23760
    },
    {
      "epoch": 1.330552133497592,
      "grad_norm": 2.9241535663604736,
      "learning_rate": 9.901315789473686e-05,
      "loss": 0.4131,
      "step": 23761
    },
    {
      "epoch": 1.330608130809721,
      "grad_norm": 1.6557778120040894,
      "learning_rate": 9.90128947368421e-05,
      "loss": 0.4093,
      "step": 23762
    },
    {
      "epoch": 1.33066412812185,
      "grad_norm": 1.8381612300872803,
      "learning_rate": 9.901263157894736e-05,
      "loss": 1.0502,
      "step": 23763
    },
    {
      "epoch": 1.330720125433979,
      "grad_norm": 1.4598925113677979,
      "learning_rate": 9.901236842105264e-05,
      "loss": 0.5094,
      "step": 23764
    },
    {
      "epoch": 1.3307761227461081,
      "grad_norm": 1.5245411396026611,
      "learning_rate": 9.90121052631579e-05,
      "loss": 0.4248,
      "step": 23765
    },
    {
      "epoch": 1.3308321200582371,
      "grad_norm": 1.3863434791564941,
      "learning_rate": 9.901184210526316e-05,
      "loss": 0.4848,
      "step": 23766
    },
    {
      "epoch": 1.3308881173703662,
      "grad_norm": 1.2266355752944946,
      "learning_rate": 9.901157894736842e-05,
      "loss": 0.5014,
      "step": 23767
    },
    {
      "epoch": 1.3309441146824952,
      "grad_norm": 1.5366392135620117,
      "learning_rate": 9.901131578947369e-05,
      "loss": 0.4406,
      "step": 23768
    },
    {
      "epoch": 1.3310001119946242,
      "grad_norm": 1.3355239629745483,
      "learning_rate": 9.901105263157895e-05,
      "loss": 0.5142,
      "step": 23769
    },
    {
      "epoch": 1.3310561093067532,
      "grad_norm": 1.3259963989257812,
      "learning_rate": 9.901078947368422e-05,
      "loss": 0.4757,
      "step": 23770
    },
    {
      "epoch": 1.3311121066188822,
      "grad_norm": 1.2686805725097656,
      "learning_rate": 9.901052631578947e-05,
      "loss": 0.4744,
      "step": 23771
    },
    {
      "epoch": 1.3311681039310113,
      "grad_norm": 1.6363102197647095,
      "learning_rate": 9.901026315789474e-05,
      "loss": 0.5022,
      "step": 23772
    },
    {
      "epoch": 1.3312241012431403,
      "grad_norm": 1.3323304653167725,
      "learning_rate": 9.901e-05,
      "loss": 0.4078,
      "step": 23773
    },
    {
      "epoch": 1.3312800985552693,
      "grad_norm": 1.2837271690368652,
      "learning_rate": 9.900973684210528e-05,
      "loss": 0.4101,
      "step": 23774
    },
    {
      "epoch": 1.3313360958673983,
      "grad_norm": 1.3309253454208374,
      "learning_rate": 9.900947368421054e-05,
      "loss": 0.5137,
      "step": 23775
    },
    {
      "epoch": 1.3313920931795273,
      "grad_norm": 1.3191558122634888,
      "learning_rate": 9.90092105263158e-05,
      "loss": 0.4689,
      "step": 23776
    },
    {
      "epoch": 1.3314480904916564,
      "grad_norm": 1.7781141996383667,
      "learning_rate": 9.900894736842105e-05,
      "loss": 0.6415,
      "step": 23777
    },
    {
      "epoch": 1.3315040878037854,
      "grad_norm": 1.3525689840316772,
      "learning_rate": 9.900868421052633e-05,
      "loss": 0.4296,
      "step": 23778
    },
    {
      "epoch": 1.3315600851159144,
      "grad_norm": 1.1907037496566772,
      "learning_rate": 9.900842105263159e-05,
      "loss": 0.3696,
      "step": 23779
    },
    {
      "epoch": 1.3316160824280434,
      "grad_norm": 1.3596187829971313,
      "learning_rate": 9.900815789473685e-05,
      "loss": 0.451,
      "step": 23780
    },
    {
      "epoch": 1.3316720797401724,
      "grad_norm": 1.2794840335845947,
      "learning_rate": 9.900789473684211e-05,
      "loss": 0.5895,
      "step": 23781
    },
    {
      "epoch": 1.3317280770523015,
      "grad_norm": 1.1582374572753906,
      "learning_rate": 9.900763157894737e-05,
      "loss": 0.4681,
      "step": 23782
    },
    {
      "epoch": 1.3317840743644305,
      "grad_norm": 1.742173433303833,
      "learning_rate": 9.900736842105264e-05,
      "loss": 0.5168,
      "step": 23783
    },
    {
      "epoch": 1.3318400716765595,
      "grad_norm": 1.382517695426941,
      "learning_rate": 9.90071052631579e-05,
      "loss": 0.4484,
      "step": 23784
    },
    {
      "epoch": 1.3318960689886885,
      "grad_norm": 1.4867075681686401,
      "learning_rate": 9.900684210526316e-05,
      "loss": 0.5246,
      "step": 23785
    },
    {
      "epoch": 1.3319520663008175,
      "grad_norm": 1.6939916610717773,
      "learning_rate": 9.900657894736842e-05,
      "loss": 0.5886,
      "step": 23786
    },
    {
      "epoch": 1.3320080636129465,
      "grad_norm": 1.1807746887207031,
      "learning_rate": 9.900631578947369e-05,
      "loss": 0.3459,
      "step": 23787
    },
    {
      "epoch": 1.3320640609250756,
      "grad_norm": 1.394364595413208,
      "learning_rate": 9.900605263157895e-05,
      "loss": 0.4881,
      "step": 23788
    },
    {
      "epoch": 1.3321200582372046,
      "grad_norm": 1.2194969654083252,
      "learning_rate": 9.900578947368421e-05,
      "loss": 0.4091,
      "step": 23789
    },
    {
      "epoch": 1.3321760555493336,
      "grad_norm": 1.711050033569336,
      "learning_rate": 9.900552631578947e-05,
      "loss": 0.5224,
      "step": 23790
    },
    {
      "epoch": 1.3322320528614626,
      "grad_norm": 1.814692497253418,
      "learning_rate": 9.900526315789475e-05,
      "loss": 0.4065,
      "step": 23791
    },
    {
      "epoch": 1.3322880501735916,
      "grad_norm": 1.7624183893203735,
      "learning_rate": 9.9005e-05,
      "loss": 0.4169,
      "step": 23792
    },
    {
      "epoch": 1.3323440474857207,
      "grad_norm": 1.291270136833191,
      "learning_rate": 9.900473684210528e-05,
      "loss": 0.4397,
      "step": 23793
    },
    {
      "epoch": 1.3324000447978497,
      "grad_norm": 2.6617846488952637,
      "learning_rate": 9.900447368421052e-05,
      "loss": 0.681,
      "step": 23794
    },
    {
      "epoch": 1.3324560421099787,
      "grad_norm": 1.305954933166504,
      "learning_rate": 9.90042105263158e-05,
      "loss": 0.5125,
      "step": 23795
    },
    {
      "epoch": 1.3325120394221077,
      "grad_norm": 1.3788312673568726,
      "learning_rate": 9.900394736842106e-05,
      "loss": 0.4138,
      "step": 23796
    },
    {
      "epoch": 1.3325680367342367,
      "grad_norm": 1.4607813358306885,
      "learning_rate": 9.900368421052632e-05,
      "loss": 0.4552,
      "step": 23797
    },
    {
      "epoch": 1.3326240340463658,
      "grad_norm": 1.2364460229873657,
      "learning_rate": 9.900342105263159e-05,
      "loss": 0.3507,
      "step": 23798
    },
    {
      "epoch": 1.3326800313584948,
      "grad_norm": 1.564463496208191,
      "learning_rate": 9.900315789473684e-05,
      "loss": 0.4571,
      "step": 23799
    },
    {
      "epoch": 1.3327360286706238,
      "grad_norm": 1.3653931617736816,
      "learning_rate": 9.900289473684211e-05,
      "loss": 0.4893,
      "step": 23800
    },
    {
      "epoch": 1.3327920259827528,
      "grad_norm": 1.6454015970230103,
      "learning_rate": 9.900263157894737e-05,
      "loss": 0.7647,
      "step": 23801
    },
    {
      "epoch": 1.3328480232948818,
      "grad_norm": 2.1272964477539062,
      "learning_rate": 9.900236842105264e-05,
      "loss": 0.7298,
      "step": 23802
    },
    {
      "epoch": 1.3329040206070109,
      "grad_norm": 1.3992019891738892,
      "learning_rate": 9.90021052631579e-05,
      "loss": 0.4638,
      "step": 23803
    },
    {
      "epoch": 1.3329600179191399,
      "grad_norm": 1.2260433435440063,
      "learning_rate": 9.900184210526316e-05,
      "loss": 0.457,
      "step": 23804
    },
    {
      "epoch": 1.333016015231269,
      "grad_norm": 1.2614973783493042,
      "learning_rate": 9.900157894736842e-05,
      "loss": 0.5395,
      "step": 23805
    },
    {
      "epoch": 1.333072012543398,
      "grad_norm": 1.2293323278427124,
      "learning_rate": 9.90013157894737e-05,
      "loss": 0.4591,
      "step": 23806
    },
    {
      "epoch": 1.333128009855527,
      "grad_norm": 1.2137826681137085,
      "learning_rate": 9.900105263157896e-05,
      "loss": 0.4168,
      "step": 23807
    },
    {
      "epoch": 1.333184007167656,
      "grad_norm": 1.2654191255569458,
      "learning_rate": 9.900078947368421e-05,
      "loss": 0.4625,
      "step": 23808
    },
    {
      "epoch": 1.333240004479785,
      "grad_norm": 1.5558730363845825,
      "learning_rate": 9.900052631578947e-05,
      "loss": 0.5058,
      "step": 23809
    },
    {
      "epoch": 1.333296001791914,
      "grad_norm": 1.408207893371582,
      "learning_rate": 9.900026315789475e-05,
      "loss": 0.466,
      "step": 23810
    },
    {
      "epoch": 1.333351999104043,
      "grad_norm": 1.3477212190628052,
      "learning_rate": 9.900000000000001e-05,
      "loss": 0.4401,
      "step": 23811
    },
    {
      "epoch": 1.333407996416172,
      "grad_norm": 1.4396218061447144,
      "learning_rate": 9.899973684210527e-05,
      "loss": 0.4345,
      "step": 23812
    },
    {
      "epoch": 1.333463993728301,
      "grad_norm": 1.7757333517074585,
      "learning_rate": 9.899947368421053e-05,
      "loss": 0.4976,
      "step": 23813
    },
    {
      "epoch": 1.33351999104043,
      "grad_norm": 1.4348074197769165,
      "learning_rate": 9.899921052631579e-05,
      "loss": 0.5476,
      "step": 23814
    },
    {
      "epoch": 1.333575988352559,
      "grad_norm": 1.3225101232528687,
      "learning_rate": 9.899894736842106e-05,
      "loss": 0.371,
      "step": 23815
    },
    {
      "epoch": 1.333631985664688,
      "grad_norm": 1.611476182937622,
      "learning_rate": 9.899868421052632e-05,
      "loss": 0.6114,
      "step": 23816
    },
    {
      "epoch": 1.3336879829768171,
      "grad_norm": 1.424462080001831,
      "learning_rate": 9.899842105263158e-05,
      "loss": 0.3367,
      "step": 23817
    },
    {
      "epoch": 1.3337439802889461,
      "grad_norm": 1.14649498462677,
      "learning_rate": 9.899815789473684e-05,
      "loss": 0.3709,
      "step": 23818
    },
    {
      "epoch": 1.3337999776010752,
      "grad_norm": 1.7224383354187012,
      "learning_rate": 9.899789473684211e-05,
      "loss": 0.6526,
      "step": 23819
    },
    {
      "epoch": 1.3338559749132042,
      "grad_norm": 1.3974765539169312,
      "learning_rate": 9.899763157894737e-05,
      "loss": 0.6072,
      "step": 23820
    },
    {
      "epoch": 1.3339119722253332,
      "grad_norm": 1.3082847595214844,
      "learning_rate": 9.899736842105263e-05,
      "loss": 0.4481,
      "step": 23821
    },
    {
      "epoch": 1.3339679695374622,
      "grad_norm": 1.349003791809082,
      "learning_rate": 9.899710526315789e-05,
      "loss": 0.5265,
      "step": 23822
    },
    {
      "epoch": 1.3340239668495912,
      "grad_norm": 1.996715784072876,
      "learning_rate": 9.899684210526316e-05,
      "loss": 0.5606,
      "step": 23823
    },
    {
      "epoch": 1.3340799641617203,
      "grad_norm": 1.4019711017608643,
      "learning_rate": 9.899657894736842e-05,
      "loss": 0.4384,
      "step": 23824
    },
    {
      "epoch": 1.3341359614738493,
      "grad_norm": 1.392748236656189,
      "learning_rate": 9.89963157894737e-05,
      "loss": 0.5603,
      "step": 23825
    },
    {
      "epoch": 1.3341919587859783,
      "grad_norm": 1.4841468334197998,
      "learning_rate": 9.899605263157894e-05,
      "loss": 0.4993,
      "step": 23826
    },
    {
      "epoch": 1.3342479560981073,
      "grad_norm": 1.6232941150665283,
      "learning_rate": 9.899578947368422e-05,
      "loss": 0.5431,
      "step": 23827
    },
    {
      "epoch": 1.3343039534102363,
      "grad_norm": 1.5400526523590088,
      "learning_rate": 9.899552631578948e-05,
      "loss": 0.436,
      "step": 23828
    },
    {
      "epoch": 1.3343599507223654,
      "grad_norm": 1.1462382078170776,
      "learning_rate": 9.899526315789475e-05,
      "loss": 0.4438,
      "step": 23829
    },
    {
      "epoch": 1.3344159480344944,
      "grad_norm": 1.3616477251052856,
      "learning_rate": 9.899500000000001e-05,
      "loss": 0.3233,
      "step": 23830
    },
    {
      "epoch": 1.3344719453466234,
      "grad_norm": 2.2361109256744385,
      "learning_rate": 9.899473684210527e-05,
      "loss": 0.7127,
      "step": 23831
    },
    {
      "epoch": 1.3345279426587524,
      "grad_norm": 1.9489185810089111,
      "learning_rate": 9.899447368421053e-05,
      "loss": 0.4546,
      "step": 23832
    },
    {
      "epoch": 1.3345839399708814,
      "grad_norm": 7.203779220581055,
      "learning_rate": 9.899421052631579e-05,
      "loss": 0.5088,
      "step": 23833
    },
    {
      "epoch": 1.3346399372830104,
      "grad_norm": 1.516080617904663,
      "learning_rate": 9.899394736842106e-05,
      "loss": 0.4153,
      "step": 23834
    },
    {
      "epoch": 1.3346959345951395,
      "grad_norm": 1.2757153511047363,
      "learning_rate": 9.899368421052632e-05,
      "loss": 0.4722,
      "step": 23835
    },
    {
      "epoch": 1.3347519319072685,
      "grad_norm": 1.1644984483718872,
      "learning_rate": 9.899342105263158e-05,
      "loss": 0.4474,
      "step": 23836
    },
    {
      "epoch": 1.3348079292193975,
      "grad_norm": 1.3059358596801758,
      "learning_rate": 9.899315789473684e-05,
      "loss": 0.5506,
      "step": 23837
    },
    {
      "epoch": 1.3348639265315265,
      "grad_norm": 1.5706517696380615,
      "learning_rate": 9.899289473684212e-05,
      "loss": 0.4251,
      "step": 23838
    },
    {
      "epoch": 1.3349199238436555,
      "grad_norm": 1.316448450088501,
      "learning_rate": 9.899263157894737e-05,
      "loss": 0.4417,
      "step": 23839
    },
    {
      "epoch": 1.3349759211557846,
      "grad_norm": 1.5210957527160645,
      "learning_rate": 9.899236842105263e-05,
      "loss": 0.4646,
      "step": 23840
    },
    {
      "epoch": 1.3350319184679136,
      "grad_norm": 1.292188286781311,
      "learning_rate": 9.89921052631579e-05,
      "loss": 0.4646,
      "step": 23841
    },
    {
      "epoch": 1.3350879157800426,
      "grad_norm": 1.5652449131011963,
      "learning_rate": 9.899184210526317e-05,
      "loss": 0.3949,
      "step": 23842
    },
    {
      "epoch": 1.3351439130921716,
      "grad_norm": 2.441153049468994,
      "learning_rate": 9.899157894736843e-05,
      "loss": 0.4277,
      "step": 23843
    },
    {
      "epoch": 1.3351999104043006,
      "grad_norm": 1.1771906614303589,
      "learning_rate": 9.899131578947369e-05,
      "loss": 0.4003,
      "step": 23844
    },
    {
      "epoch": 1.3352559077164297,
      "grad_norm": 1.5029371976852417,
      "learning_rate": 9.899105263157895e-05,
      "loss": 0.5011,
      "step": 23845
    },
    {
      "epoch": 1.3353119050285587,
      "grad_norm": 1.25845468044281,
      "learning_rate": 9.899078947368422e-05,
      "loss": 0.4785,
      "step": 23846
    },
    {
      "epoch": 1.3353679023406877,
      "grad_norm": 1.39874267578125,
      "learning_rate": 9.899052631578948e-05,
      "loss": 0.4808,
      "step": 23847
    },
    {
      "epoch": 1.3354238996528167,
      "grad_norm": 1.8373452425003052,
      "learning_rate": 9.899026315789475e-05,
      "loss": 0.5285,
      "step": 23848
    },
    {
      "epoch": 1.3354798969649457,
      "grad_norm": 1.3263756036758423,
      "learning_rate": 9.899e-05,
      "loss": 0.5416,
      "step": 23849
    },
    {
      "epoch": 1.3355358942770748,
      "grad_norm": 1.2071919441223145,
      "learning_rate": 9.898973684210526e-05,
      "loss": 0.4166,
      "step": 23850
    },
    {
      "epoch": 1.3355918915892038,
      "grad_norm": 1.3608685731887817,
      "learning_rate": 9.898947368421053e-05,
      "loss": 0.4098,
      "step": 23851
    },
    {
      "epoch": 1.3356478889013328,
      "grad_norm": 1.263219952583313,
      "learning_rate": 9.898921052631579e-05,
      "loss": 0.4067,
      "step": 23852
    },
    {
      "epoch": 1.3357038862134618,
      "grad_norm": 1.5391923189163208,
      "learning_rate": 9.898894736842107e-05,
      "loss": 0.51,
      "step": 23853
    },
    {
      "epoch": 1.3357598835255908,
      "grad_norm": 1.4992666244506836,
      "learning_rate": 9.898868421052631e-05,
      "loss": 0.5237,
      "step": 23854
    },
    {
      "epoch": 1.3358158808377198,
      "grad_norm": 1.4221733808517456,
      "learning_rate": 9.898842105263158e-05,
      "loss": 0.473,
      "step": 23855
    },
    {
      "epoch": 1.3358718781498489,
      "grad_norm": 1.4912809133529663,
      "learning_rate": 9.898815789473684e-05,
      "loss": 0.4861,
      "step": 23856
    },
    {
      "epoch": 1.3359278754619779,
      "grad_norm": 1.3533376455307007,
      "learning_rate": 9.898789473684212e-05,
      "loss": 0.5411,
      "step": 23857
    },
    {
      "epoch": 1.335983872774107,
      "grad_norm": 1.2698400020599365,
      "learning_rate": 9.898763157894738e-05,
      "loss": 0.4958,
      "step": 23858
    },
    {
      "epoch": 1.336039870086236,
      "grad_norm": 1.6540385484695435,
      "learning_rate": 9.898736842105264e-05,
      "loss": 0.5944,
      "step": 23859
    },
    {
      "epoch": 1.336095867398365,
      "grad_norm": 1.4382905960083008,
      "learning_rate": 9.89871052631579e-05,
      "loss": 0.4067,
      "step": 23860
    },
    {
      "epoch": 1.336151864710494,
      "grad_norm": 1.4954227209091187,
      "learning_rate": 9.898684210526317e-05,
      "loss": 0.5319,
      "step": 23861
    },
    {
      "epoch": 1.336207862022623,
      "grad_norm": 1.2915523052215576,
      "learning_rate": 9.898657894736843e-05,
      "loss": 0.4074,
      "step": 23862
    },
    {
      "epoch": 1.336263859334752,
      "grad_norm": 1.2406599521636963,
      "learning_rate": 9.898631578947369e-05,
      "loss": 0.4717,
      "step": 23863
    },
    {
      "epoch": 1.336319856646881,
      "grad_norm": 1.2790749073028564,
      "learning_rate": 9.898605263157895e-05,
      "loss": 0.5137,
      "step": 23864
    },
    {
      "epoch": 1.33637585395901,
      "grad_norm": 1.512863039970398,
      "learning_rate": 9.898578947368422e-05,
      "loss": 0.4512,
      "step": 23865
    },
    {
      "epoch": 1.336431851271139,
      "grad_norm": 1.4381901025772095,
      "learning_rate": 9.898552631578948e-05,
      "loss": 0.5711,
      "step": 23866
    },
    {
      "epoch": 1.336487848583268,
      "grad_norm": 1.1895484924316406,
      "learning_rate": 9.898526315789474e-05,
      "loss": 0.4086,
      "step": 23867
    },
    {
      "epoch": 1.336543845895397,
      "grad_norm": 1.587947964668274,
      "learning_rate": 9.8985e-05,
      "loss": 0.46,
      "step": 23868
    },
    {
      "epoch": 1.3365998432075261,
      "grad_norm": 1.3747241497039795,
      "learning_rate": 9.898473684210526e-05,
      "loss": 0.549,
      "step": 23869
    },
    {
      "epoch": 1.3366558405196551,
      "grad_norm": 1.4006222486495972,
      "learning_rate": 9.898447368421053e-05,
      "loss": 0.4107,
      "step": 23870
    },
    {
      "epoch": 1.3367118378317842,
      "grad_norm": 1.5141626596450806,
      "learning_rate": 9.89842105263158e-05,
      "loss": 0.5305,
      "step": 23871
    },
    {
      "epoch": 1.3367678351439132,
      "grad_norm": 1.4598748683929443,
      "learning_rate": 9.898394736842105e-05,
      "loss": 0.4646,
      "step": 23872
    },
    {
      "epoch": 1.3368238324560422,
      "grad_norm": 1.4114829301834106,
      "learning_rate": 9.898368421052631e-05,
      "loss": 0.4784,
      "step": 23873
    },
    {
      "epoch": 1.3368798297681712,
      "grad_norm": 1.3672314882278442,
      "learning_rate": 9.898342105263159e-05,
      "loss": 0.4479,
      "step": 23874
    },
    {
      "epoch": 1.3369358270803002,
      "grad_norm": 1.234108567237854,
      "learning_rate": 9.898315789473685e-05,
      "loss": 0.4793,
      "step": 23875
    },
    {
      "epoch": 1.3369918243924293,
      "grad_norm": 1.3132939338684082,
      "learning_rate": 9.89828947368421e-05,
      "loss": 0.4638,
      "step": 23876
    },
    {
      "epoch": 1.3370478217045583,
      "grad_norm": 1.4558367729187012,
      "learning_rate": 9.898263157894737e-05,
      "loss": 0.566,
      "step": 23877
    },
    {
      "epoch": 1.3371038190166873,
      "grad_norm": 1.4092020988464355,
      "learning_rate": 9.898236842105264e-05,
      "loss": 0.3604,
      "step": 23878
    },
    {
      "epoch": 1.3371598163288163,
      "grad_norm": 1.476643443107605,
      "learning_rate": 9.89821052631579e-05,
      "loss": 0.5447,
      "step": 23879
    },
    {
      "epoch": 1.3372158136409453,
      "grad_norm": 1.3778104782104492,
      "learning_rate": 9.898184210526317e-05,
      "loss": 0.4364,
      "step": 23880
    },
    {
      "epoch": 1.3372718109530743,
      "grad_norm": 1.4713430404663086,
      "learning_rate": 9.898157894736842e-05,
      "loss": 0.516,
      "step": 23881
    },
    {
      "epoch": 1.3373278082652034,
      "grad_norm": 1.4517395496368408,
      "learning_rate": 9.898131578947369e-05,
      "loss": 0.4876,
      "step": 23882
    },
    {
      "epoch": 1.3373838055773324,
      "grad_norm": 1.3174575567245483,
      "learning_rate": 9.898105263157895e-05,
      "loss": 0.4194,
      "step": 23883
    },
    {
      "epoch": 1.3374398028894614,
      "grad_norm": 1.451480507850647,
      "learning_rate": 9.898078947368421e-05,
      "loss": 0.5662,
      "step": 23884
    },
    {
      "epoch": 1.3374958002015904,
      "grad_norm": 1.321994662284851,
      "learning_rate": 9.898052631578948e-05,
      "loss": 0.3988,
      "step": 23885
    },
    {
      "epoch": 1.3375517975137194,
      "grad_norm": 1.323826789855957,
      "learning_rate": 9.898026315789473e-05,
      "loss": 0.4865,
      "step": 23886
    },
    {
      "epoch": 1.3376077948258485,
      "grad_norm": 1.6025311946868896,
      "learning_rate": 9.898e-05,
      "loss": 0.5381,
      "step": 23887
    },
    {
      "epoch": 1.3376637921379775,
      "grad_norm": 1.4213122129440308,
      "learning_rate": 9.897973684210526e-05,
      "loss": 0.5454,
      "step": 23888
    },
    {
      "epoch": 1.3377197894501065,
      "grad_norm": 1.103776216506958,
      "learning_rate": 9.897947368421054e-05,
      "loss": 0.4856,
      "step": 23889
    },
    {
      "epoch": 1.3377757867622355,
      "grad_norm": 1.3291431665420532,
      "learning_rate": 9.89792105263158e-05,
      "loss": 0.4835,
      "step": 23890
    },
    {
      "epoch": 1.3378317840743645,
      "grad_norm": 1.5485930442810059,
      "learning_rate": 9.897894736842106e-05,
      "loss": 0.6487,
      "step": 23891
    },
    {
      "epoch": 1.3378877813864936,
      "grad_norm": 1.4208502769470215,
      "learning_rate": 9.897868421052632e-05,
      "loss": 0.4842,
      "step": 23892
    },
    {
      "epoch": 1.3379437786986226,
      "grad_norm": 1.3566259145736694,
      "learning_rate": 9.897842105263159e-05,
      "loss": 0.3162,
      "step": 23893
    },
    {
      "epoch": 1.3379997760107516,
      "grad_norm": 1.3701115846633911,
      "learning_rate": 9.897815789473685e-05,
      "loss": 0.4999,
      "step": 23894
    },
    {
      "epoch": 1.3380557733228806,
      "grad_norm": 1.1461424827575684,
      "learning_rate": 9.897789473684211e-05,
      "loss": 0.4746,
      "step": 23895
    },
    {
      "epoch": 1.3381117706350096,
      "grad_norm": 1.3429561853408813,
      "learning_rate": 9.897763157894737e-05,
      "loss": 0.4234,
      "step": 23896
    },
    {
      "epoch": 1.3381677679471387,
      "grad_norm": 1.2542699575424194,
      "learning_rate": 9.897736842105264e-05,
      "loss": 0.4548,
      "step": 23897
    },
    {
      "epoch": 1.3382237652592677,
      "grad_norm": 1.7728514671325684,
      "learning_rate": 9.89771052631579e-05,
      "loss": 0.5006,
      "step": 23898
    },
    {
      "epoch": 1.3382797625713967,
      "grad_norm": 2.911759376525879,
      "learning_rate": 9.897684210526316e-05,
      "loss": 0.4549,
      "step": 23899
    },
    {
      "epoch": 1.3383357598835257,
      "grad_norm": 2.126786708831787,
      "learning_rate": 9.897657894736842e-05,
      "loss": 0.5309,
      "step": 23900
    },
    {
      "epoch": 1.3383917571956547,
      "grad_norm": 1.303514003753662,
      "learning_rate": 9.897631578947368e-05,
      "loss": 0.4588,
      "step": 23901
    },
    {
      "epoch": 1.3384477545077837,
      "grad_norm": 1.3816440105438232,
      "learning_rate": 9.897605263157895e-05,
      "loss": 0.4875,
      "step": 23902
    },
    {
      "epoch": 1.3385037518199128,
      "grad_norm": 1.1241973638534546,
      "learning_rate": 9.897578947368421e-05,
      "loss": 0.3535,
      "step": 23903
    },
    {
      "epoch": 1.3385597491320418,
      "grad_norm": 1.3934160470962524,
      "learning_rate": 9.897552631578947e-05,
      "loss": 0.4369,
      "step": 23904
    },
    {
      "epoch": 1.3386157464441708,
      "grad_norm": 1.5238735675811768,
      "learning_rate": 9.897526315789473e-05,
      "loss": 0.4324,
      "step": 23905
    },
    {
      "epoch": 1.3386717437562998,
      "grad_norm": 1.5582380294799805,
      "learning_rate": 9.897500000000001e-05,
      "loss": 0.5663,
      "step": 23906
    },
    {
      "epoch": 1.3387277410684288,
      "grad_norm": 1.082220435142517,
      "learning_rate": 9.897473684210527e-05,
      "loss": 0.4133,
      "step": 23907
    },
    {
      "epoch": 1.3387837383805579,
      "grad_norm": 1.3235070705413818,
      "learning_rate": 9.897447368421054e-05,
      "loss": 0.4129,
      "step": 23908
    },
    {
      "epoch": 1.3388397356926869,
      "grad_norm": 1.3532748222351074,
      "learning_rate": 9.897421052631579e-05,
      "loss": 0.5135,
      "step": 23909
    },
    {
      "epoch": 1.338895733004816,
      "grad_norm": 1.4414713382720947,
      "learning_rate": 9.897394736842106e-05,
      "loss": 0.3811,
      "step": 23910
    },
    {
      "epoch": 1.338951730316945,
      "grad_norm": 1.3468427658081055,
      "learning_rate": 9.897368421052632e-05,
      "loss": 0.4218,
      "step": 23911
    },
    {
      "epoch": 1.339007727629074,
      "grad_norm": 1.3484735488891602,
      "learning_rate": 9.897342105263159e-05,
      "loss": 0.4637,
      "step": 23912
    },
    {
      "epoch": 1.339063724941203,
      "grad_norm": 2.199549436569214,
      "learning_rate": 9.897315789473685e-05,
      "loss": 0.5371,
      "step": 23913
    },
    {
      "epoch": 1.339119722253332,
      "grad_norm": 1.2956798076629639,
      "learning_rate": 9.897289473684211e-05,
      "loss": 0.4634,
      "step": 23914
    },
    {
      "epoch": 1.3391757195654608,
      "grad_norm": 1.543900728225708,
      "learning_rate": 9.897263157894737e-05,
      "loss": 0.5257,
      "step": 23915
    },
    {
      "epoch": 1.3392317168775898,
      "grad_norm": 1.3237605094909668,
      "learning_rate": 9.897236842105264e-05,
      "loss": 0.4308,
      "step": 23916
    },
    {
      "epoch": 1.3392877141897188,
      "grad_norm": 1.7733278274536133,
      "learning_rate": 9.89721052631579e-05,
      "loss": 0.4651,
      "step": 23917
    },
    {
      "epoch": 1.3393437115018478,
      "grad_norm": 1.2714046239852905,
      "learning_rate": 9.897184210526315e-05,
      "loss": 0.4149,
      "step": 23918
    },
    {
      "epoch": 1.3393997088139769,
      "grad_norm": 1.4521942138671875,
      "learning_rate": 9.897157894736842e-05,
      "loss": 0.3297,
      "step": 23919
    },
    {
      "epoch": 1.3394557061261059,
      "grad_norm": 1.4575254917144775,
      "learning_rate": 9.897131578947368e-05,
      "loss": 0.6913,
      "step": 23920
    },
    {
      "epoch": 1.339511703438235,
      "grad_norm": 1.6212458610534668,
      "learning_rate": 9.897105263157896e-05,
      "loss": 0.5082,
      "step": 23921
    },
    {
      "epoch": 1.339567700750364,
      "grad_norm": 1.4157074689865112,
      "learning_rate": 9.897078947368422e-05,
      "loss": 0.4816,
      "step": 23922
    },
    {
      "epoch": 1.339623698062493,
      "grad_norm": 1.3685389757156372,
      "learning_rate": 9.897052631578948e-05,
      "loss": 0.4676,
      "step": 23923
    },
    {
      "epoch": 1.339679695374622,
      "grad_norm": 1.3349485397338867,
      "learning_rate": 9.897026315789474e-05,
      "loss": 0.7631,
      "step": 23924
    },
    {
      "epoch": 1.339735692686751,
      "grad_norm": 1.2802698612213135,
      "learning_rate": 9.897000000000001e-05,
      "loss": 0.5143,
      "step": 23925
    },
    {
      "epoch": 1.33979168999888,
      "grad_norm": 1.3084275722503662,
      "learning_rate": 9.896973684210527e-05,
      "loss": 0.4478,
      "step": 23926
    },
    {
      "epoch": 1.339847687311009,
      "grad_norm": 1.560441255569458,
      "learning_rate": 9.896947368421053e-05,
      "loss": 0.5547,
      "step": 23927
    },
    {
      "epoch": 1.339903684623138,
      "grad_norm": 1.327688217163086,
      "learning_rate": 9.896921052631579e-05,
      "loss": 0.6401,
      "step": 23928
    },
    {
      "epoch": 1.339959681935267,
      "grad_norm": 1.4617291688919067,
      "learning_rate": 9.896894736842106e-05,
      "loss": 0.4964,
      "step": 23929
    },
    {
      "epoch": 1.340015679247396,
      "grad_norm": 1.5066403150558472,
      "learning_rate": 9.896868421052632e-05,
      "loss": 0.4309,
      "step": 23930
    },
    {
      "epoch": 1.340071676559525,
      "grad_norm": 1.3324167728424072,
      "learning_rate": 9.896842105263158e-05,
      "loss": 0.4384,
      "step": 23931
    },
    {
      "epoch": 1.340127673871654,
      "grad_norm": 1.2845227718353271,
      "learning_rate": 9.896815789473684e-05,
      "loss": 0.4418,
      "step": 23932
    },
    {
      "epoch": 1.3401836711837831,
      "grad_norm": 1.507017731666565,
      "learning_rate": 9.896789473684211e-05,
      "loss": 0.5666,
      "step": 23933
    },
    {
      "epoch": 1.3402396684959121,
      "grad_norm": 1.5677680969238281,
      "learning_rate": 9.896763157894737e-05,
      "loss": 0.5223,
      "step": 23934
    },
    {
      "epoch": 1.3402956658080412,
      "grad_norm": 1.4154253005981445,
      "learning_rate": 9.896736842105263e-05,
      "loss": 0.5921,
      "step": 23935
    },
    {
      "epoch": 1.3403516631201702,
      "grad_norm": 1.358176589012146,
      "learning_rate": 9.89671052631579e-05,
      "loss": 0.4748,
      "step": 23936
    },
    {
      "epoch": 1.3404076604322992,
      "grad_norm": 1.3491337299346924,
      "learning_rate": 9.896684210526315e-05,
      "loss": 0.4248,
      "step": 23937
    },
    {
      "epoch": 1.3404636577444282,
      "grad_norm": 1.1043469905853271,
      "learning_rate": 9.896657894736843e-05,
      "loss": 0.4779,
      "step": 23938
    },
    {
      "epoch": 1.3405196550565572,
      "grad_norm": 1.4574761390686035,
      "learning_rate": 9.896631578947369e-05,
      "loss": 0.6477,
      "step": 23939
    },
    {
      "epoch": 1.3405756523686863,
      "grad_norm": 1.283100962638855,
      "learning_rate": 9.896605263157896e-05,
      "loss": 0.4059,
      "step": 23940
    },
    {
      "epoch": 1.3406316496808153,
      "grad_norm": 1.1675559282302856,
      "learning_rate": 9.89657894736842e-05,
      "loss": 0.3313,
      "step": 23941
    },
    {
      "epoch": 1.3406876469929443,
      "grad_norm": 1.175600290298462,
      "learning_rate": 9.896552631578948e-05,
      "loss": 0.3677,
      "step": 23942
    },
    {
      "epoch": 1.3407436443050733,
      "grad_norm": 1.2264043092727661,
      "learning_rate": 9.896526315789474e-05,
      "loss": 0.4248,
      "step": 23943
    },
    {
      "epoch": 1.3407996416172023,
      "grad_norm": 1.5873838663101196,
      "learning_rate": 9.896500000000001e-05,
      "loss": 0.487,
      "step": 23944
    },
    {
      "epoch": 1.3408556389293314,
      "grad_norm": 1.2819849252700806,
      "learning_rate": 9.896473684210527e-05,
      "loss": 0.4605,
      "step": 23945
    },
    {
      "epoch": 1.3409116362414604,
      "grad_norm": 1.4461268186569214,
      "learning_rate": 9.896447368421053e-05,
      "loss": 0.525,
      "step": 23946
    },
    {
      "epoch": 1.3409676335535894,
      "grad_norm": 1.2202770709991455,
      "learning_rate": 9.896421052631579e-05,
      "loss": 0.4893,
      "step": 23947
    },
    {
      "epoch": 1.3410236308657184,
      "grad_norm": 1.4147801399230957,
      "learning_rate": 9.896394736842106e-05,
      "loss": 0.4573,
      "step": 23948
    },
    {
      "epoch": 1.3410796281778474,
      "grad_norm": 1.9122450351715088,
      "learning_rate": 9.896368421052632e-05,
      "loss": 0.4832,
      "step": 23949
    },
    {
      "epoch": 1.3411356254899764,
      "grad_norm": 1.5094122886657715,
      "learning_rate": 9.896342105263158e-05,
      "loss": 0.4196,
      "step": 23950
    },
    {
      "epoch": 1.3411916228021055,
      "grad_norm": 1.4098691940307617,
      "learning_rate": 9.896315789473684e-05,
      "loss": 0.478,
      "step": 23951
    },
    {
      "epoch": 1.3412476201142345,
      "grad_norm": 1.3345513343811035,
      "learning_rate": 9.896289473684212e-05,
      "loss": 0.3645,
      "step": 23952
    },
    {
      "epoch": 1.3413036174263635,
      "grad_norm": 1.1226009130477905,
      "learning_rate": 9.896263157894738e-05,
      "loss": 0.3738,
      "step": 23953
    },
    {
      "epoch": 1.3413596147384925,
      "grad_norm": 1.3263577222824097,
      "learning_rate": 9.896236842105264e-05,
      "loss": 0.4815,
      "step": 23954
    },
    {
      "epoch": 1.3414156120506215,
      "grad_norm": 1.2852894067764282,
      "learning_rate": 9.89621052631579e-05,
      "loss": 0.4569,
      "step": 23955
    },
    {
      "epoch": 1.3414716093627506,
      "grad_norm": 1.4982022047042847,
      "learning_rate": 9.896184210526316e-05,
      "loss": 0.6568,
      "step": 23956
    },
    {
      "epoch": 1.3415276066748796,
      "grad_norm": 1.3840714693069458,
      "learning_rate": 9.896157894736843e-05,
      "loss": 0.4405,
      "step": 23957
    },
    {
      "epoch": 1.3415836039870086,
      "grad_norm": 1.18630051612854,
      "learning_rate": 9.896131578947369e-05,
      "loss": 0.5221,
      "step": 23958
    },
    {
      "epoch": 1.3416396012991376,
      "grad_norm": 1.3169230222702026,
      "learning_rate": 9.896105263157895e-05,
      "loss": 0.5608,
      "step": 23959
    },
    {
      "epoch": 1.3416955986112666,
      "grad_norm": 1.7112146615982056,
      "learning_rate": 9.896078947368421e-05,
      "loss": 0.5546,
      "step": 23960
    },
    {
      "epoch": 1.3417515959233957,
      "grad_norm": 1.5063049793243408,
      "learning_rate": 9.896052631578948e-05,
      "loss": 0.6101,
      "step": 23961
    },
    {
      "epoch": 1.3418075932355247,
      "grad_norm": 1.383370280265808,
      "learning_rate": 9.896026315789474e-05,
      "loss": 0.4475,
      "step": 23962
    },
    {
      "epoch": 1.3418635905476537,
      "grad_norm": 1.4847168922424316,
      "learning_rate": 9.896000000000001e-05,
      "loss": 0.5212,
      "step": 23963
    },
    {
      "epoch": 1.3419195878597827,
      "grad_norm": 1.172202229499817,
      "learning_rate": 9.895973684210526e-05,
      "loss": 0.4173,
      "step": 23964
    },
    {
      "epoch": 1.3419755851719117,
      "grad_norm": 1.6316022872924805,
      "learning_rate": 9.895947368421053e-05,
      "loss": 0.5424,
      "step": 23965
    },
    {
      "epoch": 1.3420315824840408,
      "grad_norm": 1.3216683864593506,
      "learning_rate": 9.89592105263158e-05,
      "loss": 0.4224,
      "step": 23966
    },
    {
      "epoch": 1.3420875797961698,
      "grad_norm": 1.0985703468322754,
      "learning_rate": 9.895894736842107e-05,
      "loss": 0.3667,
      "step": 23967
    },
    {
      "epoch": 1.3421435771082988,
      "grad_norm": 1.6286671161651611,
      "learning_rate": 9.895868421052631e-05,
      "loss": 0.5541,
      "step": 23968
    },
    {
      "epoch": 1.3421995744204278,
      "grad_norm": 1.5791877508163452,
      "learning_rate": 9.895842105263159e-05,
      "loss": 0.5442,
      "step": 23969
    },
    {
      "epoch": 1.3422555717325568,
      "grad_norm": 1.5616247653961182,
      "learning_rate": 9.895815789473685e-05,
      "loss": 0.6061,
      "step": 23970
    },
    {
      "epoch": 1.3423115690446858,
      "grad_norm": 1.5898517370224,
      "learning_rate": 9.89578947368421e-05,
      "loss": 0.4698,
      "step": 23971
    },
    {
      "epoch": 1.3423675663568149,
      "grad_norm": 1.1105499267578125,
      "learning_rate": 9.895763157894738e-05,
      "loss": 0.4383,
      "step": 23972
    },
    {
      "epoch": 1.3424235636689439,
      "grad_norm": 1.3896452188491821,
      "learning_rate": 9.895736842105263e-05,
      "loss": 0.4147,
      "step": 23973
    },
    {
      "epoch": 1.342479560981073,
      "grad_norm": 1.7090681791305542,
      "learning_rate": 9.89571052631579e-05,
      "loss": 0.4694,
      "step": 23974
    },
    {
      "epoch": 1.342535558293202,
      "grad_norm": 2.046135187149048,
      "learning_rate": 9.895684210526316e-05,
      "loss": 0.7754,
      "step": 23975
    },
    {
      "epoch": 1.342591555605331,
      "grad_norm": 1.567522644996643,
      "learning_rate": 9.895657894736843e-05,
      "loss": 0.5032,
      "step": 23976
    },
    {
      "epoch": 1.34264755291746,
      "grad_norm": 1.248820424079895,
      "learning_rate": 9.895631578947369e-05,
      "loss": 0.4866,
      "step": 23977
    },
    {
      "epoch": 1.342703550229589,
      "grad_norm": 1.32659113407135,
      "learning_rate": 9.895605263157895e-05,
      "loss": 0.5262,
      "step": 23978
    },
    {
      "epoch": 1.342759547541718,
      "grad_norm": 1.36756432056427,
      "learning_rate": 9.895578947368421e-05,
      "loss": 0.4342,
      "step": 23979
    },
    {
      "epoch": 1.342815544853847,
      "grad_norm": 1.3160245418548584,
      "learning_rate": 9.895552631578948e-05,
      "loss": 0.4198,
      "step": 23980
    },
    {
      "epoch": 1.342871542165976,
      "grad_norm": 1.4190582036972046,
      "learning_rate": 9.895526315789474e-05,
      "loss": 0.4506,
      "step": 23981
    },
    {
      "epoch": 1.342927539478105,
      "grad_norm": 1.7953743934631348,
      "learning_rate": 9.8955e-05,
      "loss": 0.4899,
      "step": 23982
    },
    {
      "epoch": 1.342983536790234,
      "grad_norm": 1.3152244091033936,
      "learning_rate": 9.895473684210526e-05,
      "loss": 0.4237,
      "step": 23983
    },
    {
      "epoch": 1.343039534102363,
      "grad_norm": 1.426226019859314,
      "learning_rate": 9.895447368421054e-05,
      "loss": 0.5225,
      "step": 23984
    },
    {
      "epoch": 1.3430955314144921,
      "grad_norm": 1.4641419649124146,
      "learning_rate": 9.89542105263158e-05,
      "loss": 0.476,
      "step": 23985
    },
    {
      "epoch": 1.3431515287266211,
      "grad_norm": 1.3623632192611694,
      "learning_rate": 9.895394736842106e-05,
      "loss": 0.5086,
      "step": 23986
    },
    {
      "epoch": 1.3432075260387502,
      "grad_norm": 1.0939357280731201,
      "learning_rate": 9.895368421052632e-05,
      "loss": 0.5236,
      "step": 23987
    },
    {
      "epoch": 1.3432635233508792,
      "grad_norm": 1.5585516691207886,
      "learning_rate": 9.895342105263158e-05,
      "loss": 0.594,
      "step": 23988
    },
    {
      "epoch": 1.3433195206630082,
      "grad_norm": 19.078744888305664,
      "learning_rate": 9.895315789473685e-05,
      "loss": 0.4286,
      "step": 23989
    },
    {
      "epoch": 1.3433755179751372,
      "grad_norm": 1.1451103687286377,
      "learning_rate": 9.895289473684211e-05,
      "loss": 0.4013,
      "step": 23990
    },
    {
      "epoch": 1.3434315152872662,
      "grad_norm": 1.6688843965530396,
      "learning_rate": 9.895263157894737e-05,
      "loss": 0.6271,
      "step": 23991
    },
    {
      "epoch": 1.3434875125993953,
      "grad_norm": 1.4637559652328491,
      "learning_rate": 9.895236842105263e-05,
      "loss": 0.6666,
      "step": 23992
    },
    {
      "epoch": 1.3435435099115243,
      "grad_norm": 1.3607884645462036,
      "learning_rate": 9.89521052631579e-05,
      "loss": 0.5131,
      "step": 23993
    },
    {
      "epoch": 1.3435995072236533,
      "grad_norm": 1.3291383981704712,
      "learning_rate": 9.895184210526316e-05,
      "loss": 0.4581,
      "step": 23994
    },
    {
      "epoch": 1.3436555045357823,
      "grad_norm": 1.298227071762085,
      "learning_rate": 9.895157894736843e-05,
      "loss": 0.5034,
      "step": 23995
    },
    {
      "epoch": 1.3437115018479113,
      "grad_norm": 1.3066940307617188,
      "learning_rate": 9.895131578947368e-05,
      "loss": 0.4796,
      "step": 23996
    },
    {
      "epoch": 1.3437674991600403,
      "grad_norm": 1.0873332023620605,
      "learning_rate": 9.895105263157895e-05,
      "loss": 0.3945,
      "step": 23997
    },
    {
      "epoch": 1.3438234964721694,
      "grad_norm": 1.6476949453353882,
      "learning_rate": 9.895078947368421e-05,
      "loss": 0.5423,
      "step": 23998
    },
    {
      "epoch": 1.3438794937842984,
      "grad_norm": 1.4215620756149292,
      "learning_rate": 9.895052631578949e-05,
      "loss": 0.4441,
      "step": 23999
    },
    {
      "epoch": 1.3439354910964274,
      "grad_norm": 1.2197273969650269,
      "learning_rate": 9.895026315789475e-05,
      "loss": 0.383,
      "step": 24000
    },
    {
      "epoch": 1.3439914884085564,
      "grad_norm": 1.3874455690383911,
      "learning_rate": 9.895e-05,
      "loss": 0.5279,
      "step": 24001
    },
    {
      "epoch": 1.3440474857206854,
      "grad_norm": 1.2576227188110352,
      "learning_rate": 9.894973684210527e-05,
      "loss": 0.4465,
      "step": 24002
    },
    {
      "epoch": 1.3441034830328145,
      "grad_norm": 1.453696846961975,
      "learning_rate": 9.894947368421054e-05,
      "loss": 0.4785,
      "step": 24003
    },
    {
      "epoch": 1.3441594803449435,
      "grad_norm": 1.446753978729248,
      "learning_rate": 9.89492105263158e-05,
      "loss": 0.5079,
      "step": 24004
    },
    {
      "epoch": 1.3442154776570725,
      "grad_norm": 1.7142091989517212,
      "learning_rate": 9.894894736842106e-05,
      "loss": 0.6639,
      "step": 24005
    },
    {
      "epoch": 1.3442714749692015,
      "grad_norm": 1.5219104290008545,
      "learning_rate": 9.894868421052632e-05,
      "loss": 0.4946,
      "step": 24006
    },
    {
      "epoch": 1.3443274722813305,
      "grad_norm": 1.135257601737976,
      "learning_rate": 9.894842105263158e-05,
      "loss": 0.3277,
      "step": 24007
    },
    {
      "epoch": 1.3443834695934596,
      "grad_norm": 1.2626960277557373,
      "learning_rate": 9.894815789473685e-05,
      "loss": 0.4215,
      "step": 24008
    },
    {
      "epoch": 1.3444394669055886,
      "grad_norm": 1.1269794702529907,
      "learning_rate": 9.894789473684211e-05,
      "loss": 0.4061,
      "step": 24009
    },
    {
      "epoch": 1.3444954642177176,
      "grad_norm": 1.2398040294647217,
      "learning_rate": 9.894763157894737e-05,
      "loss": 0.5051,
      "step": 24010
    },
    {
      "epoch": 1.3445514615298466,
      "grad_norm": 1.2162896394729614,
      "learning_rate": 9.894736842105263e-05,
      "loss": 0.3527,
      "step": 24011
    },
    {
      "epoch": 1.3446074588419756,
      "grad_norm": 1.1273400783538818,
      "learning_rate": 9.89471052631579e-05,
      "loss": 0.3632,
      "step": 24012
    },
    {
      "epoch": 1.3446634561541047,
      "grad_norm": 1.2960906028747559,
      "learning_rate": 9.894684210526316e-05,
      "loss": 0.4308,
      "step": 24013
    },
    {
      "epoch": 1.3447194534662337,
      "grad_norm": 1.1805799007415771,
      "learning_rate": 9.894657894736842e-05,
      "loss": 0.3707,
      "step": 24014
    },
    {
      "epoch": 1.3447754507783627,
      "grad_norm": 1.424858570098877,
      "learning_rate": 9.894631578947368e-05,
      "loss": 0.4553,
      "step": 24015
    },
    {
      "epoch": 1.3448314480904917,
      "grad_norm": 1.1969478130340576,
      "learning_rate": 9.894605263157896e-05,
      "loss": 0.5471,
      "step": 24016
    },
    {
      "epoch": 1.3448874454026207,
      "grad_norm": 1.5464682579040527,
      "learning_rate": 9.894578947368422e-05,
      "loss": 0.4241,
      "step": 24017
    },
    {
      "epoch": 1.3449434427147497,
      "grad_norm": 1.2001961469650269,
      "learning_rate": 9.894552631578949e-05,
      "loss": 0.4713,
      "step": 24018
    },
    {
      "epoch": 1.3449994400268788,
      "grad_norm": 1.5608412027359009,
      "learning_rate": 9.894526315789474e-05,
      "loss": 0.5646,
      "step": 24019
    },
    {
      "epoch": 1.3450554373390078,
      "grad_norm": 1.757699728012085,
      "learning_rate": 9.894500000000001e-05,
      "loss": 0.5386,
      "step": 24020
    },
    {
      "epoch": 1.3451114346511368,
      "grad_norm": 1.6653386354446411,
      "learning_rate": 9.894473684210527e-05,
      "loss": 0.593,
      "step": 24021
    },
    {
      "epoch": 1.3451674319632658,
      "grad_norm": 1.305739402770996,
      "learning_rate": 9.894447368421053e-05,
      "loss": 0.4264,
      "step": 24022
    },
    {
      "epoch": 1.3452234292753948,
      "grad_norm": 1.2630366086959839,
      "learning_rate": 9.894421052631579e-05,
      "loss": 0.5489,
      "step": 24023
    },
    {
      "epoch": 1.3452794265875239,
      "grad_norm": 1.4003286361694336,
      "learning_rate": 9.894394736842105e-05,
      "loss": 0.5305,
      "step": 24024
    },
    {
      "epoch": 1.3453354238996529,
      "grad_norm": 1.5072417259216309,
      "learning_rate": 9.894368421052632e-05,
      "loss": 0.5619,
      "step": 24025
    },
    {
      "epoch": 1.345391421211782,
      "grad_norm": 1.4055123329162598,
      "learning_rate": 9.894342105263158e-05,
      "loss": 0.4955,
      "step": 24026
    },
    {
      "epoch": 1.345447418523911,
      "grad_norm": 1.1914441585540771,
      "learning_rate": 9.894315789473685e-05,
      "loss": 0.3817,
      "step": 24027
    },
    {
      "epoch": 1.34550341583604,
      "grad_norm": 1.378165364265442,
      "learning_rate": 9.89428947368421e-05,
      "loss": 0.4743,
      "step": 24028
    },
    {
      "epoch": 1.345559413148169,
      "grad_norm": 1.399539828300476,
      "learning_rate": 9.894263157894737e-05,
      "loss": 0.4816,
      "step": 24029
    },
    {
      "epoch": 1.345615410460298,
      "grad_norm": 1.3169493675231934,
      "learning_rate": 9.894236842105263e-05,
      "loss": 0.4081,
      "step": 24030
    },
    {
      "epoch": 1.345671407772427,
      "grad_norm": 1.3407050371170044,
      "learning_rate": 9.89421052631579e-05,
      "loss": 0.3549,
      "step": 24031
    },
    {
      "epoch": 1.345727405084556,
      "grad_norm": 1.0616273880004883,
      "learning_rate": 9.894184210526317e-05,
      "loss": 0.3927,
      "step": 24032
    },
    {
      "epoch": 1.345783402396685,
      "grad_norm": 1.173591136932373,
      "learning_rate": 9.894157894736843e-05,
      "loss": 0.3702,
      "step": 24033
    },
    {
      "epoch": 1.345839399708814,
      "grad_norm": 1.3523924350738525,
      "learning_rate": 9.894131578947369e-05,
      "loss": 0.4446,
      "step": 24034
    },
    {
      "epoch": 1.345895397020943,
      "grad_norm": 1.2364277839660645,
      "learning_rate": 9.894105263157896e-05,
      "loss": 0.4579,
      "step": 24035
    },
    {
      "epoch": 1.345951394333072,
      "grad_norm": 1.3914684057235718,
      "learning_rate": 9.894078947368422e-05,
      "loss": 0.4652,
      "step": 24036
    },
    {
      "epoch": 1.3460073916452011,
      "grad_norm": 1.5705797672271729,
      "learning_rate": 9.894052631578948e-05,
      "loss": 0.4735,
      "step": 24037
    },
    {
      "epoch": 1.3460633889573301,
      "grad_norm": 1.4909099340438843,
      "learning_rate": 9.894026315789474e-05,
      "loss": 0.4696,
      "step": 24038
    },
    {
      "epoch": 1.346119386269459,
      "grad_norm": 1.4098501205444336,
      "learning_rate": 9.894e-05,
      "loss": 0.5234,
      "step": 24039
    },
    {
      "epoch": 1.346175383581588,
      "grad_norm": 1.753283143043518,
      "learning_rate": 9.893973684210527e-05,
      "loss": 0.3517,
      "step": 24040
    },
    {
      "epoch": 1.346231380893717,
      "grad_norm": 1.263336420059204,
      "learning_rate": 9.893947368421053e-05,
      "loss": 0.4078,
      "step": 24041
    },
    {
      "epoch": 1.346287378205846,
      "grad_norm": 1.461779236793518,
      "learning_rate": 9.893921052631579e-05,
      "loss": 0.4651,
      "step": 24042
    },
    {
      "epoch": 1.346343375517975,
      "grad_norm": 1.3641862869262695,
      "learning_rate": 9.893894736842105e-05,
      "loss": 0.5486,
      "step": 24043
    },
    {
      "epoch": 1.346399372830104,
      "grad_norm": 1.3653966188430786,
      "learning_rate": 9.893868421052632e-05,
      "loss": 0.4846,
      "step": 24044
    },
    {
      "epoch": 1.346455370142233,
      "grad_norm": 1.9145190715789795,
      "learning_rate": 9.893842105263158e-05,
      "loss": 0.3525,
      "step": 24045
    },
    {
      "epoch": 1.346511367454362,
      "grad_norm": 1.817022442817688,
      "learning_rate": 9.893815789473684e-05,
      "loss": 0.4692,
      "step": 24046
    },
    {
      "epoch": 1.346567364766491,
      "grad_norm": 1.4517470598220825,
      "learning_rate": 9.89378947368421e-05,
      "loss": 0.4655,
      "step": 24047
    },
    {
      "epoch": 1.34662336207862,
      "grad_norm": 1.4838329553604126,
      "learning_rate": 9.893763157894738e-05,
      "loss": 0.5176,
      "step": 24048
    },
    {
      "epoch": 1.3466793593907491,
      "grad_norm": 1.3544859886169434,
      "learning_rate": 9.893736842105264e-05,
      "loss": 0.435,
      "step": 24049
    },
    {
      "epoch": 1.3467353567028781,
      "grad_norm": 1.4888906478881836,
      "learning_rate": 9.893710526315791e-05,
      "loss": 0.4983,
      "step": 24050
    },
    {
      "epoch": 1.3467913540150072,
      "grad_norm": 1.4776861667633057,
      "learning_rate": 9.893684210526316e-05,
      "loss": 0.3908,
      "step": 24051
    },
    {
      "epoch": 1.3468473513271362,
      "grad_norm": 1.268038034439087,
      "learning_rate": 9.893657894736843e-05,
      "loss": 0.5198,
      "step": 24052
    },
    {
      "epoch": 1.3469033486392652,
      "grad_norm": 1.2676458358764648,
      "learning_rate": 9.893631578947369e-05,
      "loss": 0.4383,
      "step": 24053
    },
    {
      "epoch": 1.3469593459513942,
      "grad_norm": 1.3287932872772217,
      "learning_rate": 9.893605263157896e-05,
      "loss": 0.6178,
      "step": 24054
    },
    {
      "epoch": 1.3470153432635232,
      "grad_norm": 1.209586501121521,
      "learning_rate": 9.893578947368422e-05,
      "loss": 0.3452,
      "step": 24055
    },
    {
      "epoch": 1.3470713405756523,
      "grad_norm": 1.239275336265564,
      "learning_rate": 9.893552631578947e-05,
      "loss": 0.4227,
      "step": 24056
    },
    {
      "epoch": 1.3471273378877813,
      "grad_norm": 1.1617977619171143,
      "learning_rate": 9.893526315789474e-05,
      "loss": 0.5356,
      "step": 24057
    },
    {
      "epoch": 1.3471833351999103,
      "grad_norm": 1.5607088804244995,
      "learning_rate": 9.8935e-05,
      "loss": 0.4493,
      "step": 24058
    },
    {
      "epoch": 1.3472393325120393,
      "grad_norm": 1.4072494506835938,
      "learning_rate": 9.893473684210527e-05,
      "loss": 0.4303,
      "step": 24059
    },
    {
      "epoch": 1.3472953298241683,
      "grad_norm": 1.407301902770996,
      "learning_rate": 9.893447368421053e-05,
      "loss": 0.4385,
      "step": 24060
    },
    {
      "epoch": 1.3473513271362973,
      "grad_norm": 2.6843373775482178,
      "learning_rate": 9.89342105263158e-05,
      "loss": 0.5358,
      "step": 24061
    },
    {
      "epoch": 1.3474073244484264,
      "grad_norm": 1.351423978805542,
      "learning_rate": 9.893394736842105e-05,
      "loss": 0.4161,
      "step": 24062
    },
    {
      "epoch": 1.3474633217605554,
      "grad_norm": 1.351823329925537,
      "learning_rate": 9.893368421052633e-05,
      "loss": 0.4508,
      "step": 24063
    },
    {
      "epoch": 1.3475193190726844,
      "grad_norm": 1.767401933670044,
      "learning_rate": 9.893342105263159e-05,
      "loss": 0.7152,
      "step": 24064
    },
    {
      "epoch": 1.3475753163848134,
      "grad_norm": 6.961129188537598,
      "learning_rate": 9.893315789473685e-05,
      "loss": 0.5912,
      "step": 24065
    },
    {
      "epoch": 1.3476313136969424,
      "grad_norm": 1.4069809913635254,
      "learning_rate": 9.89328947368421e-05,
      "loss": 0.3986,
      "step": 24066
    },
    {
      "epoch": 1.3476873110090715,
      "grad_norm": 1.523929476737976,
      "learning_rate": 9.893263157894738e-05,
      "loss": 0.5491,
      "step": 24067
    },
    {
      "epoch": 1.3477433083212005,
      "grad_norm": 1.4899320602416992,
      "learning_rate": 9.893236842105264e-05,
      "loss": 0.474,
      "step": 24068
    },
    {
      "epoch": 1.3477993056333295,
      "grad_norm": 1.280630350112915,
      "learning_rate": 9.89321052631579e-05,
      "loss": 0.4654,
      "step": 24069
    },
    {
      "epoch": 1.3478553029454585,
      "grad_norm": 1.3929500579833984,
      "learning_rate": 9.893184210526316e-05,
      "loss": 0.3879,
      "step": 24070
    },
    {
      "epoch": 1.3479113002575875,
      "grad_norm": 1.3972034454345703,
      "learning_rate": 9.893157894736843e-05,
      "loss": 0.4501,
      "step": 24071
    },
    {
      "epoch": 1.3479672975697166,
      "grad_norm": 1.2639726400375366,
      "learning_rate": 9.893131578947369e-05,
      "loss": 0.4856,
      "step": 24072
    },
    {
      "epoch": 1.3480232948818456,
      "grad_norm": 1.5184447765350342,
      "learning_rate": 9.893105263157896e-05,
      "loss": 0.4357,
      "step": 24073
    },
    {
      "epoch": 1.3480792921939746,
      "grad_norm": 1.315146565437317,
      "learning_rate": 9.893078947368421e-05,
      "loss": 0.4788,
      "step": 24074
    },
    {
      "epoch": 1.3481352895061036,
      "grad_norm": 1.240248680114746,
      "learning_rate": 9.893052631578947e-05,
      "loss": 0.4533,
      "step": 24075
    },
    {
      "epoch": 1.3481912868182326,
      "grad_norm": 1.172282338142395,
      "learning_rate": 9.893026315789474e-05,
      "loss": 0.4746,
      "step": 24076
    },
    {
      "epoch": 1.3482472841303617,
      "grad_norm": 1.311255931854248,
      "learning_rate": 9.893e-05,
      "loss": 0.3963,
      "step": 24077
    },
    {
      "epoch": 1.3483032814424907,
      "grad_norm": 1.3184565305709839,
      "learning_rate": 9.892973684210526e-05,
      "loss": 0.539,
      "step": 24078
    },
    {
      "epoch": 1.3483592787546197,
      "grad_norm": 1.647714376449585,
      "learning_rate": 9.892947368421052e-05,
      "loss": 0.4814,
      "step": 24079
    },
    {
      "epoch": 1.3484152760667487,
      "grad_norm": 1.3690624237060547,
      "learning_rate": 9.89292105263158e-05,
      "loss": 0.5514,
      "step": 24080
    },
    {
      "epoch": 1.3484712733788777,
      "grad_norm": 1.2252135276794434,
      "learning_rate": 9.892894736842106e-05,
      "loss": 0.3129,
      "step": 24081
    },
    {
      "epoch": 1.3485272706910068,
      "grad_norm": 1.3502732515335083,
      "learning_rate": 9.892868421052633e-05,
      "loss": 0.5709,
      "step": 24082
    },
    {
      "epoch": 1.3485832680031358,
      "grad_norm": 1.294970989227295,
      "learning_rate": 9.892842105263158e-05,
      "loss": 0.4358,
      "step": 24083
    },
    {
      "epoch": 1.3486392653152648,
      "grad_norm": 1.469613790512085,
      "learning_rate": 9.892815789473685e-05,
      "loss": 0.4896,
      "step": 24084
    },
    {
      "epoch": 1.3486952626273938,
      "grad_norm": 1.8542873859405518,
      "learning_rate": 9.892789473684211e-05,
      "loss": 0.5623,
      "step": 24085
    },
    {
      "epoch": 1.3487512599395228,
      "grad_norm": 1.658245325088501,
      "learning_rate": 9.892763157894738e-05,
      "loss": 0.6528,
      "step": 24086
    },
    {
      "epoch": 1.3488072572516518,
      "grad_norm": 1.2902683019638062,
      "learning_rate": 9.892736842105264e-05,
      "loss": 0.5618,
      "step": 24087
    },
    {
      "epoch": 1.3488632545637809,
      "grad_norm": 1.1421093940734863,
      "learning_rate": 9.89271052631579e-05,
      "loss": 0.4831,
      "step": 24088
    },
    {
      "epoch": 1.3489192518759099,
      "grad_norm": 1.7525116205215454,
      "learning_rate": 9.892684210526316e-05,
      "loss": 0.5039,
      "step": 24089
    },
    {
      "epoch": 1.348975249188039,
      "grad_norm": 1.113273024559021,
      "learning_rate": 9.892657894736843e-05,
      "loss": 0.3923,
      "step": 24090
    },
    {
      "epoch": 1.349031246500168,
      "grad_norm": 1.418257713317871,
      "learning_rate": 9.89263157894737e-05,
      "loss": 0.5238,
      "step": 24091
    },
    {
      "epoch": 1.349087243812297,
      "grad_norm": 1.840466856956482,
      "learning_rate": 9.892605263157895e-05,
      "loss": 0.542,
      "step": 24092
    },
    {
      "epoch": 1.349143241124426,
      "grad_norm": 1.3966180086135864,
      "learning_rate": 9.892578947368421e-05,
      "loss": 0.4315,
      "step": 24093
    },
    {
      "epoch": 1.349199238436555,
      "grad_norm": 1.2987778186798096,
      "learning_rate": 9.892552631578947e-05,
      "loss": 0.5709,
      "step": 24094
    },
    {
      "epoch": 1.349255235748684,
      "grad_norm": 1.2208820581436157,
      "learning_rate": 9.892526315789475e-05,
      "loss": 0.45,
      "step": 24095
    },
    {
      "epoch": 1.349311233060813,
      "grad_norm": 2.0124058723449707,
      "learning_rate": 9.8925e-05,
      "loss": 0.544,
      "step": 24096
    },
    {
      "epoch": 1.349367230372942,
      "grad_norm": 1.3852250576019287,
      "learning_rate": 9.892473684210527e-05,
      "loss": 0.4652,
      "step": 24097
    },
    {
      "epoch": 1.349423227685071,
      "grad_norm": 1.2076325416564941,
      "learning_rate": 9.892447368421053e-05,
      "loss": 0.3717,
      "step": 24098
    },
    {
      "epoch": 1.3494792249972,
      "grad_norm": 1.1834232807159424,
      "learning_rate": 9.89242105263158e-05,
      "loss": 0.4067,
      "step": 24099
    },
    {
      "epoch": 1.349535222309329,
      "grad_norm": 1.4517765045166016,
      "learning_rate": 9.892394736842106e-05,
      "loss": 0.5416,
      "step": 24100
    },
    {
      "epoch": 1.3495912196214581,
      "grad_norm": 1.5725260972976685,
      "learning_rate": 9.892368421052632e-05,
      "loss": 0.607,
      "step": 24101
    },
    {
      "epoch": 1.3496472169335871,
      "grad_norm": 1.9018875360488892,
      "learning_rate": 9.892342105263158e-05,
      "loss": 0.6389,
      "step": 24102
    },
    {
      "epoch": 1.3497032142457162,
      "grad_norm": 1.2932662963867188,
      "learning_rate": 9.892315789473685e-05,
      "loss": 0.432,
      "step": 24103
    },
    {
      "epoch": 1.3497592115578452,
      "grad_norm": 1.3422033786773682,
      "learning_rate": 9.892289473684211e-05,
      "loss": 0.4565,
      "step": 24104
    },
    {
      "epoch": 1.3498152088699742,
      "grad_norm": 1.4826048612594604,
      "learning_rate": 9.892263157894738e-05,
      "loss": 0.5029,
      "step": 24105
    },
    {
      "epoch": 1.3498712061821032,
      "grad_norm": 1.4096723794937134,
      "learning_rate": 9.892236842105263e-05,
      "loss": 0.5243,
      "step": 24106
    },
    {
      "epoch": 1.3499272034942322,
      "grad_norm": 1.2899080514907837,
      "learning_rate": 9.89221052631579e-05,
      "loss": 0.4729,
      "step": 24107
    },
    {
      "epoch": 1.3499832008063612,
      "grad_norm": 3.5839273929595947,
      "learning_rate": 9.892184210526316e-05,
      "loss": 0.4041,
      "step": 24108
    },
    {
      "epoch": 1.3500391981184903,
      "grad_norm": 1.2935422658920288,
      "learning_rate": 9.892157894736842e-05,
      "loss": 0.5591,
      "step": 24109
    },
    {
      "epoch": 1.3500951954306193,
      "grad_norm": 1.571065068244934,
      "learning_rate": 9.89213157894737e-05,
      "loss": 0.4823,
      "step": 24110
    },
    {
      "epoch": 1.3501511927427483,
      "grad_norm": 1.3770251274108887,
      "learning_rate": 9.892105263157894e-05,
      "loss": 0.5661,
      "step": 24111
    },
    {
      "epoch": 1.3502071900548773,
      "grad_norm": 1.1110776662826538,
      "learning_rate": 9.892078947368422e-05,
      "loss": 0.2878,
      "step": 24112
    },
    {
      "epoch": 1.3502631873670063,
      "grad_norm": 1.5924330949783325,
      "learning_rate": 9.892052631578948e-05,
      "loss": 0.6316,
      "step": 24113
    },
    {
      "epoch": 1.3503191846791354,
      "grad_norm": 1.457471489906311,
      "learning_rate": 9.892026315789475e-05,
      "loss": 0.4762,
      "step": 24114
    },
    {
      "epoch": 1.3503751819912644,
      "grad_norm": 1.3572285175323486,
      "learning_rate": 9.892e-05,
      "loss": 0.6146,
      "step": 24115
    },
    {
      "epoch": 1.3504311793033934,
      "grad_norm": 1.2651491165161133,
      "learning_rate": 9.891973684210527e-05,
      "loss": 0.5161,
      "step": 24116
    },
    {
      "epoch": 1.3504871766155224,
      "grad_norm": 1.3892878293991089,
      "learning_rate": 9.891947368421053e-05,
      "loss": 0.4709,
      "step": 24117
    },
    {
      "epoch": 1.3505431739276514,
      "grad_norm": 1.5117753744125366,
      "learning_rate": 9.89192105263158e-05,
      "loss": 0.5472,
      "step": 24118
    },
    {
      "epoch": 1.3505991712397805,
      "grad_norm": 1.2897844314575195,
      "learning_rate": 9.891894736842106e-05,
      "loss": 0.4222,
      "step": 24119
    },
    {
      "epoch": 1.3506551685519095,
      "grad_norm": 1.2057937383651733,
      "learning_rate": 9.891868421052632e-05,
      "loss": 0.4167,
      "step": 24120
    },
    {
      "epoch": 1.3507111658640385,
      "grad_norm": 1.2438232898712158,
      "learning_rate": 9.891842105263158e-05,
      "loss": 0.3894,
      "step": 24121
    },
    {
      "epoch": 1.3507671631761675,
      "grad_norm": 1.4046286344528198,
      "learning_rate": 9.891815789473685e-05,
      "loss": 0.5186,
      "step": 24122
    },
    {
      "epoch": 1.3508231604882965,
      "grad_norm": 1.3658525943756104,
      "learning_rate": 9.891789473684211e-05,
      "loss": 0.4163,
      "step": 24123
    },
    {
      "epoch": 1.3508791578004256,
      "grad_norm": 1.5054686069488525,
      "learning_rate": 9.891763157894737e-05,
      "loss": 0.5208,
      "step": 24124
    },
    {
      "epoch": 1.3509351551125546,
      "grad_norm": 1.4575799703598022,
      "learning_rate": 9.891736842105263e-05,
      "loss": 0.439,
      "step": 24125
    },
    {
      "epoch": 1.3509911524246836,
      "grad_norm": 1.4247299432754517,
      "learning_rate": 9.891710526315789e-05,
      "loss": 0.4459,
      "step": 24126
    },
    {
      "epoch": 1.3510471497368126,
      "grad_norm": 1.237309455871582,
      "learning_rate": 9.891684210526317e-05,
      "loss": 0.5082,
      "step": 24127
    },
    {
      "epoch": 1.3511031470489416,
      "grad_norm": 1.267808198928833,
      "learning_rate": 9.891657894736843e-05,
      "loss": 0.3936,
      "step": 24128
    },
    {
      "epoch": 1.3511591443610707,
      "grad_norm": 1.9146208763122559,
      "learning_rate": 9.891631578947369e-05,
      "loss": 0.5779,
      "step": 24129
    },
    {
      "epoch": 1.3512151416731997,
      "grad_norm": 1.3047025203704834,
      "learning_rate": 9.891605263157894e-05,
      "loss": 0.5201,
      "step": 24130
    },
    {
      "epoch": 1.3512711389853287,
      "grad_norm": 1.3481351137161255,
      "learning_rate": 9.891578947368422e-05,
      "loss": 0.5738,
      "step": 24131
    },
    {
      "epoch": 1.3513271362974577,
      "grad_norm": 1.3294408321380615,
      "learning_rate": 9.891552631578948e-05,
      "loss": 0.4455,
      "step": 24132
    },
    {
      "epoch": 1.3513831336095867,
      "grad_norm": 1.4808603525161743,
      "learning_rate": 9.891526315789474e-05,
      "loss": 0.454,
      "step": 24133
    },
    {
      "epoch": 1.3514391309217157,
      "grad_norm": 1.704590082168579,
      "learning_rate": 9.8915e-05,
      "loss": 0.4827,
      "step": 24134
    },
    {
      "epoch": 1.3514951282338448,
      "grad_norm": 1.5826306343078613,
      "learning_rate": 9.891473684210527e-05,
      "loss": 0.3863,
      "step": 24135
    },
    {
      "epoch": 1.3515511255459738,
      "grad_norm": 1.5127642154693604,
      "learning_rate": 9.891447368421053e-05,
      "loss": 0.4407,
      "step": 24136
    },
    {
      "epoch": 1.3516071228581028,
      "grad_norm": 1.5637285709381104,
      "learning_rate": 9.89142105263158e-05,
      "loss": 0.5563,
      "step": 24137
    },
    {
      "epoch": 1.3516631201702318,
      "grad_norm": 1.279770851135254,
      "learning_rate": 9.891394736842105e-05,
      "loss": 0.4167,
      "step": 24138
    },
    {
      "epoch": 1.3517191174823608,
      "grad_norm": 1.5844379663467407,
      "learning_rate": 9.891368421052632e-05,
      "loss": 0.4667,
      "step": 24139
    },
    {
      "epoch": 1.3517751147944899,
      "grad_norm": 1.5427618026733398,
      "learning_rate": 9.891342105263158e-05,
      "loss": 0.546,
      "step": 24140
    },
    {
      "epoch": 1.3518311121066189,
      "grad_norm": 1.4364628791809082,
      "learning_rate": 9.891315789473686e-05,
      "loss": 0.5371,
      "step": 24141
    },
    {
      "epoch": 1.351887109418748,
      "grad_norm": 1.5178382396697998,
      "learning_rate": 9.891289473684212e-05,
      "loss": 0.5804,
      "step": 24142
    },
    {
      "epoch": 1.351943106730877,
      "grad_norm": 1.4384104013442993,
      "learning_rate": 9.891263157894736e-05,
      "loss": 0.3964,
      "step": 24143
    },
    {
      "epoch": 1.351999104043006,
      "grad_norm": 1.4983294010162354,
      "learning_rate": 9.891236842105264e-05,
      "loss": 0.4637,
      "step": 24144
    },
    {
      "epoch": 1.352055101355135,
      "grad_norm": 1.268157720565796,
      "learning_rate": 9.89121052631579e-05,
      "loss": 0.4143,
      "step": 24145
    },
    {
      "epoch": 1.352111098667264,
      "grad_norm": 2.0224101543426514,
      "learning_rate": 9.891184210526317e-05,
      "loss": 0.5155,
      "step": 24146
    },
    {
      "epoch": 1.352167095979393,
      "grad_norm": 1.149941086769104,
      "learning_rate": 9.891157894736843e-05,
      "loss": 0.4014,
      "step": 24147
    },
    {
      "epoch": 1.352223093291522,
      "grad_norm": 1.277801752090454,
      "learning_rate": 9.891131578947369e-05,
      "loss": 0.3095,
      "step": 24148
    },
    {
      "epoch": 1.352279090603651,
      "grad_norm": 1.5732860565185547,
      "learning_rate": 9.891105263157895e-05,
      "loss": 0.5345,
      "step": 24149
    },
    {
      "epoch": 1.35233508791578,
      "grad_norm": 1.5471495389938354,
      "learning_rate": 9.891078947368422e-05,
      "loss": 0.4654,
      "step": 24150
    },
    {
      "epoch": 1.352391085227909,
      "grad_norm": 1.4263899326324463,
      "learning_rate": 9.891052631578948e-05,
      "loss": 0.5575,
      "step": 24151
    },
    {
      "epoch": 1.352447082540038,
      "grad_norm": 1.419783115386963,
      "learning_rate": 9.891026315789474e-05,
      "loss": 0.5694,
      "step": 24152
    },
    {
      "epoch": 1.3525030798521671,
      "grad_norm": 1.1714369058609009,
      "learning_rate": 9.891e-05,
      "loss": 0.3831,
      "step": 24153
    },
    {
      "epoch": 1.3525590771642961,
      "grad_norm": 1.4590812921524048,
      "learning_rate": 9.890973684210527e-05,
      "loss": 0.4227,
      "step": 24154
    },
    {
      "epoch": 1.3526150744764251,
      "grad_norm": 1.3871046304702759,
      "learning_rate": 9.890947368421053e-05,
      "loss": 0.4155,
      "step": 24155
    },
    {
      "epoch": 1.3526710717885542,
      "grad_norm": 1.2001183032989502,
      "learning_rate": 9.890921052631579e-05,
      "loss": 0.506,
      "step": 24156
    },
    {
      "epoch": 1.3527270691006832,
      "grad_norm": 1.2502472400665283,
      "learning_rate": 9.890894736842105e-05,
      "loss": 0.4634,
      "step": 24157
    },
    {
      "epoch": 1.3527830664128122,
      "grad_norm": 1.3538122177124023,
      "learning_rate": 9.890868421052633e-05,
      "loss": 0.5772,
      "step": 24158
    },
    {
      "epoch": 1.3528390637249412,
      "grad_norm": 1.198918104171753,
      "learning_rate": 9.890842105263159e-05,
      "loss": 0.4198,
      "step": 24159
    },
    {
      "epoch": 1.3528950610370702,
      "grad_norm": 1.3008321523666382,
      "learning_rate": 9.890815789473685e-05,
      "loss": 0.4081,
      "step": 24160
    },
    {
      "epoch": 1.3529510583491993,
      "grad_norm": 1.3779500722885132,
      "learning_rate": 9.89078947368421e-05,
      "loss": 0.3471,
      "step": 24161
    },
    {
      "epoch": 1.3530070556613283,
      "grad_norm": 1.5330551862716675,
      "learning_rate": 9.890763157894736e-05,
      "loss": 0.5198,
      "step": 24162
    },
    {
      "epoch": 1.3530630529734573,
      "grad_norm": 1.5421743392944336,
      "learning_rate": 9.890736842105264e-05,
      "loss": 0.5889,
      "step": 24163
    },
    {
      "epoch": 1.3531190502855863,
      "grad_norm": 1.4606456756591797,
      "learning_rate": 9.89071052631579e-05,
      "loss": 0.5037,
      "step": 24164
    },
    {
      "epoch": 1.3531750475977153,
      "grad_norm": 1.3997637033462524,
      "learning_rate": 9.890684210526317e-05,
      "loss": 0.4339,
      "step": 24165
    },
    {
      "epoch": 1.3532310449098444,
      "grad_norm": 2.146996259689331,
      "learning_rate": 9.890657894736842e-05,
      "loss": 0.4112,
      "step": 24166
    },
    {
      "epoch": 1.3532870422219734,
      "grad_norm": 1.3905378580093384,
      "learning_rate": 9.890631578947369e-05,
      "loss": 0.4845,
      "step": 24167
    },
    {
      "epoch": 1.3533430395341024,
      "grad_norm": 1.1500251293182373,
      "learning_rate": 9.890605263157895e-05,
      "loss": 0.3876,
      "step": 24168
    },
    {
      "epoch": 1.3533990368462314,
      "grad_norm": 1.3702603578567505,
      "learning_rate": 9.890578947368422e-05,
      "loss": 0.5397,
      "step": 24169
    },
    {
      "epoch": 1.3534550341583604,
      "grad_norm": 1.3898779153823853,
      "learning_rate": 9.890552631578947e-05,
      "loss": 0.61,
      "step": 24170
    },
    {
      "epoch": 1.3535110314704895,
      "grad_norm": 1.6539660692214966,
      "learning_rate": 9.890526315789474e-05,
      "loss": 0.5015,
      "step": 24171
    },
    {
      "epoch": 1.3535670287826185,
      "grad_norm": 1.4206243753433228,
      "learning_rate": 9.8905e-05,
      "loss": 0.5337,
      "step": 24172
    },
    {
      "epoch": 1.3536230260947475,
      "grad_norm": 1.1216557025909424,
      "learning_rate": 9.890473684210528e-05,
      "loss": 0.4585,
      "step": 24173
    },
    {
      "epoch": 1.3536790234068765,
      "grad_norm": 1.3988527059555054,
      "learning_rate": 9.890447368421054e-05,
      "loss": 0.4261,
      "step": 24174
    },
    {
      "epoch": 1.3537350207190055,
      "grad_norm": 1.198021411895752,
      "learning_rate": 9.89042105263158e-05,
      "loss": 0.4139,
      "step": 24175
    },
    {
      "epoch": 1.3537910180311346,
      "grad_norm": 1.432989478111267,
      "learning_rate": 9.890394736842106e-05,
      "loss": 0.5061,
      "step": 24176
    },
    {
      "epoch": 1.3538470153432636,
      "grad_norm": 1.4509515762329102,
      "learning_rate": 9.890368421052631e-05,
      "loss": 0.4368,
      "step": 24177
    },
    {
      "epoch": 1.3539030126553926,
      "grad_norm": 1.2205407619476318,
      "learning_rate": 9.890342105263159e-05,
      "loss": 0.4162,
      "step": 24178
    },
    {
      "epoch": 1.3539590099675216,
      "grad_norm": 1.3447471857070923,
      "learning_rate": 9.890315789473685e-05,
      "loss": 0.4203,
      "step": 24179
    },
    {
      "epoch": 1.3540150072796506,
      "grad_norm": 1.5836881399154663,
      "learning_rate": 9.890289473684211e-05,
      "loss": 0.4985,
      "step": 24180
    },
    {
      "epoch": 1.3540710045917796,
      "grad_norm": 1.6705793142318726,
      "learning_rate": 9.890263157894737e-05,
      "loss": 0.4286,
      "step": 24181
    },
    {
      "epoch": 1.3541270019039087,
      "grad_norm": 1.2332497835159302,
      "learning_rate": 9.890236842105264e-05,
      "loss": 0.3292,
      "step": 24182
    },
    {
      "epoch": 1.3541829992160377,
      "grad_norm": 3.5343668460845947,
      "learning_rate": 9.89021052631579e-05,
      "loss": 0.5716,
      "step": 24183
    },
    {
      "epoch": 1.3542389965281667,
      "grad_norm": 1.3043307065963745,
      "learning_rate": 9.890184210526316e-05,
      "loss": 0.4799,
      "step": 24184
    },
    {
      "epoch": 1.3542949938402957,
      "grad_norm": 1.2603365182876587,
      "learning_rate": 9.890157894736842e-05,
      "loss": 0.4495,
      "step": 24185
    },
    {
      "epoch": 1.3543509911524247,
      "grad_norm": 1.324040412902832,
      "learning_rate": 9.890131578947369e-05,
      "loss": 0.3862,
      "step": 24186
    },
    {
      "epoch": 1.3544069884645538,
      "grad_norm": 1.3032180070877075,
      "learning_rate": 9.890105263157895e-05,
      "loss": 0.5301,
      "step": 24187
    },
    {
      "epoch": 1.3544629857766828,
      "grad_norm": 1.455611228942871,
      "learning_rate": 9.890078947368421e-05,
      "loss": 0.4353,
      "step": 24188
    },
    {
      "epoch": 1.3545189830888118,
      "grad_norm": 1.2120689153671265,
      "learning_rate": 9.890052631578947e-05,
      "loss": 0.429,
      "step": 24189
    },
    {
      "epoch": 1.3545749804009408,
      "grad_norm": 1.3751366138458252,
      "learning_rate": 9.890026315789475e-05,
      "loss": 0.4037,
      "step": 24190
    },
    {
      "epoch": 1.3546309777130698,
      "grad_norm": 1.38014817237854,
      "learning_rate": 9.89e-05,
      "loss": 0.5004,
      "step": 24191
    },
    {
      "epoch": 1.3546869750251989,
      "grad_norm": 1.1251559257507324,
      "learning_rate": 9.889973684210528e-05,
      "loss": 0.4815,
      "step": 24192
    },
    {
      "epoch": 1.3547429723373279,
      "grad_norm": 1.3140547275543213,
      "learning_rate": 9.889947368421052e-05,
      "loss": 0.5415,
      "step": 24193
    },
    {
      "epoch": 1.354798969649457,
      "grad_norm": 1.2560465335845947,
      "learning_rate": 9.88992105263158e-05,
      "loss": 0.4763,
      "step": 24194
    },
    {
      "epoch": 1.354854966961586,
      "grad_norm": 1.240114450454712,
      "learning_rate": 9.889894736842106e-05,
      "loss": 0.4366,
      "step": 24195
    },
    {
      "epoch": 1.354910964273715,
      "grad_norm": 1.2777663469314575,
      "learning_rate": 9.889868421052632e-05,
      "loss": 0.3919,
      "step": 24196
    },
    {
      "epoch": 1.354966961585844,
      "grad_norm": 1.5230880975723267,
      "learning_rate": 9.889842105263159e-05,
      "loss": 0.5446,
      "step": 24197
    },
    {
      "epoch": 1.355022958897973,
      "grad_norm": 1.1917310953140259,
      "learning_rate": 9.889815789473684e-05,
      "loss": 0.5723,
      "step": 24198
    },
    {
      "epoch": 1.355078956210102,
      "grad_norm": 1.3890283107757568,
      "learning_rate": 9.889789473684211e-05,
      "loss": 0.6764,
      "step": 24199
    },
    {
      "epoch": 1.355134953522231,
      "grad_norm": 1.5664749145507812,
      "learning_rate": 9.889763157894737e-05,
      "loss": 0.5056,
      "step": 24200
    },
    {
      "epoch": 1.35519095083436,
      "grad_norm": 1.706183671951294,
      "learning_rate": 9.889736842105264e-05,
      "loss": 0.4927,
      "step": 24201
    },
    {
      "epoch": 1.355246948146489,
      "grad_norm": 1.0862797498703003,
      "learning_rate": 9.88971052631579e-05,
      "loss": 0.3748,
      "step": 24202
    },
    {
      "epoch": 1.355302945458618,
      "grad_norm": 1.3727495670318604,
      "learning_rate": 9.889684210526316e-05,
      "loss": 0.4898,
      "step": 24203
    },
    {
      "epoch": 1.355358942770747,
      "grad_norm": 1.635277271270752,
      "learning_rate": 9.889657894736842e-05,
      "loss": 0.4993,
      "step": 24204
    },
    {
      "epoch": 1.355414940082876,
      "grad_norm": 1.2087209224700928,
      "learning_rate": 9.88963157894737e-05,
      "loss": 0.3636,
      "step": 24205
    },
    {
      "epoch": 1.3554709373950051,
      "grad_norm": 1.111750602722168,
      "learning_rate": 9.889605263157896e-05,
      "loss": 0.4543,
      "step": 24206
    },
    {
      "epoch": 1.3555269347071341,
      "grad_norm": 1.292447805404663,
      "learning_rate": 9.889578947368422e-05,
      "loss": 0.4484,
      "step": 24207
    },
    {
      "epoch": 1.3555829320192632,
      "grad_norm": 1.3316152095794678,
      "learning_rate": 9.889552631578947e-05,
      "loss": 0.3604,
      "step": 24208
    },
    {
      "epoch": 1.3556389293313922,
      "grad_norm": 1.314098834991455,
      "learning_rate": 9.889526315789475e-05,
      "loss": 0.3969,
      "step": 24209
    },
    {
      "epoch": 1.3556949266435212,
      "grad_norm": 1.5935890674591064,
      "learning_rate": 9.889500000000001e-05,
      "loss": 0.4482,
      "step": 24210
    },
    {
      "epoch": 1.3557509239556502,
      "grad_norm": 1.6971734762191772,
      "learning_rate": 9.889473684210527e-05,
      "loss": 0.4667,
      "step": 24211
    },
    {
      "epoch": 1.3558069212677792,
      "grad_norm": 1.528926134109497,
      "learning_rate": 9.889447368421053e-05,
      "loss": 0.4533,
      "step": 24212
    },
    {
      "epoch": 1.3558629185799083,
      "grad_norm": 1.1695584058761597,
      "learning_rate": 9.889421052631579e-05,
      "loss": 0.3628,
      "step": 24213
    },
    {
      "epoch": 1.3559189158920373,
      "grad_norm": 1.1592373847961426,
      "learning_rate": 9.889394736842106e-05,
      "loss": 0.4212,
      "step": 24214
    },
    {
      "epoch": 1.3559749132041663,
      "grad_norm": 1.202193021774292,
      "learning_rate": 9.889368421052632e-05,
      "loss": 0.4012,
      "step": 24215
    },
    {
      "epoch": 1.3560309105162953,
      "grad_norm": 1.3646345138549805,
      "learning_rate": 9.889342105263158e-05,
      "loss": 0.3992,
      "step": 24216
    },
    {
      "epoch": 1.3560869078284243,
      "grad_norm": 1.2469518184661865,
      "learning_rate": 9.889315789473684e-05,
      "loss": 0.5593,
      "step": 24217
    },
    {
      "epoch": 1.3561429051405534,
      "grad_norm": 1.3526692390441895,
      "learning_rate": 9.889289473684211e-05,
      "loss": 0.434,
      "step": 24218
    },
    {
      "epoch": 1.3561989024526824,
      "grad_norm": 1.6600333452224731,
      "learning_rate": 9.889263157894737e-05,
      "loss": 0.456,
      "step": 24219
    },
    {
      "epoch": 1.3562548997648114,
      "grad_norm": 1.4503874778747559,
      "learning_rate": 9.889236842105265e-05,
      "loss": 0.4817,
      "step": 24220
    },
    {
      "epoch": 1.3563108970769404,
      "grad_norm": 1.2051103115081787,
      "learning_rate": 9.889210526315789e-05,
      "loss": 0.48,
      "step": 24221
    },
    {
      "epoch": 1.3563668943890694,
      "grad_norm": 1.310248613357544,
      "learning_rate": 9.889184210526317e-05,
      "loss": 0.4754,
      "step": 24222
    },
    {
      "epoch": 1.3564228917011985,
      "grad_norm": 1.2252604961395264,
      "learning_rate": 9.889157894736842e-05,
      "loss": 0.4161,
      "step": 24223
    },
    {
      "epoch": 1.3564788890133275,
      "grad_norm": 1.4256223440170288,
      "learning_rate": 9.88913157894737e-05,
      "loss": 0.6532,
      "step": 24224
    },
    {
      "epoch": 1.3565348863254565,
      "grad_norm": 1.3690801858901978,
      "learning_rate": 9.889105263157894e-05,
      "loss": 0.5856,
      "step": 24225
    },
    {
      "epoch": 1.3565908836375855,
      "grad_norm": 1.3578503131866455,
      "learning_rate": 9.889078947368422e-05,
      "loss": 0.4091,
      "step": 24226
    },
    {
      "epoch": 1.3566468809497145,
      "grad_norm": 1.3648658990859985,
      "learning_rate": 9.889052631578948e-05,
      "loss": 0.5009,
      "step": 24227
    },
    {
      "epoch": 1.3567028782618435,
      "grad_norm": 1.122082233428955,
      "learning_rate": 9.889026315789475e-05,
      "loss": 0.4603,
      "step": 24228
    },
    {
      "epoch": 1.3567588755739726,
      "grad_norm": 1.250191330909729,
      "learning_rate": 9.889000000000001e-05,
      "loss": 0.4719,
      "step": 24229
    },
    {
      "epoch": 1.3568148728861016,
      "grad_norm": 1.492835283279419,
      "learning_rate": 9.888973684210526e-05,
      "loss": 0.3728,
      "step": 24230
    },
    {
      "epoch": 1.3568708701982306,
      "grad_norm": 1.8283497095108032,
      "learning_rate": 9.888947368421053e-05,
      "loss": 0.5613,
      "step": 24231
    },
    {
      "epoch": 1.3569268675103596,
      "grad_norm": 1.3256876468658447,
      "learning_rate": 9.888921052631579e-05,
      "loss": 0.458,
      "step": 24232
    },
    {
      "epoch": 1.3569828648224886,
      "grad_norm": 1.3450658321380615,
      "learning_rate": 9.888894736842106e-05,
      "loss": 0.5543,
      "step": 24233
    },
    {
      "epoch": 1.3570388621346177,
      "grad_norm": 1.1495815515518188,
      "learning_rate": 9.888868421052632e-05,
      "loss": 0.3296,
      "step": 24234
    },
    {
      "epoch": 1.3570948594467467,
      "grad_norm": 1.409485936164856,
      "learning_rate": 9.888842105263158e-05,
      "loss": 0.5735,
      "step": 24235
    },
    {
      "epoch": 1.3571508567588757,
      "grad_norm": 1.1729849576950073,
      "learning_rate": 9.888815789473684e-05,
      "loss": 0.4933,
      "step": 24236
    },
    {
      "epoch": 1.3572068540710047,
      "grad_norm": 1.3659942150115967,
      "learning_rate": 9.888789473684212e-05,
      "loss": 0.5493,
      "step": 24237
    },
    {
      "epoch": 1.3572628513831337,
      "grad_norm": 1.3520770072937012,
      "learning_rate": 9.888763157894737e-05,
      "loss": 0.5127,
      "step": 24238
    },
    {
      "epoch": 1.3573188486952628,
      "grad_norm": 1.4395102262496948,
      "learning_rate": 9.888736842105263e-05,
      "loss": 0.3862,
      "step": 24239
    },
    {
      "epoch": 1.3573748460073918,
      "grad_norm": 1.3926860094070435,
      "learning_rate": 9.88871052631579e-05,
      "loss": 0.476,
      "step": 24240
    },
    {
      "epoch": 1.3574308433195208,
      "grad_norm": 1.5400936603546143,
      "learning_rate": 9.888684210526317e-05,
      "loss": 0.415,
      "step": 24241
    },
    {
      "epoch": 1.3574868406316498,
      "grad_norm": 1.6989082098007202,
      "learning_rate": 9.888657894736843e-05,
      "loss": 0.4737,
      "step": 24242
    },
    {
      "epoch": 1.3575428379437788,
      "grad_norm": 1.441806674003601,
      "learning_rate": 9.888631578947369e-05,
      "loss": 0.3868,
      "step": 24243
    },
    {
      "epoch": 1.3575988352559079,
      "grad_norm": 1.4132057428359985,
      "learning_rate": 9.888605263157895e-05,
      "loss": 0.536,
      "step": 24244
    },
    {
      "epoch": 1.3576548325680367,
      "grad_norm": 1.4496694803237915,
      "learning_rate": 9.888578947368422e-05,
      "loss": 0.4312,
      "step": 24245
    },
    {
      "epoch": 1.3577108298801657,
      "grad_norm": 1.1595571041107178,
      "learning_rate": 9.888552631578948e-05,
      "loss": 0.3974,
      "step": 24246
    },
    {
      "epoch": 1.3577668271922947,
      "grad_norm": 1.4537824392318726,
      "learning_rate": 9.888526315789474e-05,
      "loss": 0.5266,
      "step": 24247
    },
    {
      "epoch": 1.3578228245044237,
      "grad_norm": 1.4500279426574707,
      "learning_rate": 9.8885e-05,
      "loss": 0.5608,
      "step": 24248
    },
    {
      "epoch": 1.3578788218165527,
      "grad_norm": 1.2889269590377808,
      "learning_rate": 9.888473684210526e-05,
      "loss": 0.4108,
      "step": 24249
    },
    {
      "epoch": 1.3579348191286817,
      "grad_norm": 1.3143959045410156,
      "learning_rate": 9.888447368421053e-05,
      "loss": 0.4612,
      "step": 24250
    },
    {
      "epoch": 1.3579908164408108,
      "grad_norm": 1.5159499645233154,
      "learning_rate": 9.888421052631579e-05,
      "loss": 0.5644,
      "step": 24251
    },
    {
      "epoch": 1.3580468137529398,
      "grad_norm": 1.3953338861465454,
      "learning_rate": 9.888394736842107e-05,
      "loss": 0.4103,
      "step": 24252
    },
    {
      "epoch": 1.3581028110650688,
      "grad_norm": 1.1751587390899658,
      "learning_rate": 9.888368421052631e-05,
      "loss": 0.4934,
      "step": 24253
    },
    {
      "epoch": 1.3581588083771978,
      "grad_norm": 1.4721421003341675,
      "learning_rate": 9.888342105263158e-05,
      "loss": 0.5593,
      "step": 24254
    },
    {
      "epoch": 1.3582148056893268,
      "grad_norm": 1.2335082292556763,
      "learning_rate": 9.888315789473684e-05,
      "loss": 0.5028,
      "step": 24255
    },
    {
      "epoch": 1.3582708030014559,
      "grad_norm": 1.506395697593689,
      "learning_rate": 9.888289473684212e-05,
      "loss": 0.4373,
      "step": 24256
    },
    {
      "epoch": 1.3583268003135849,
      "grad_norm": 1.208866000175476,
      "learning_rate": 9.888263157894738e-05,
      "loss": 0.4574,
      "step": 24257
    },
    {
      "epoch": 1.358382797625714,
      "grad_norm": 1.3434886932373047,
      "learning_rate": 9.888236842105264e-05,
      "loss": 0.5081,
      "step": 24258
    },
    {
      "epoch": 1.358438794937843,
      "grad_norm": 1.523879051208496,
      "learning_rate": 9.88821052631579e-05,
      "loss": 0.4654,
      "step": 24259
    },
    {
      "epoch": 1.358494792249972,
      "grad_norm": 1.5192832946777344,
      "learning_rate": 9.888184210526317e-05,
      "loss": 0.3732,
      "step": 24260
    },
    {
      "epoch": 1.358550789562101,
      "grad_norm": 1.497955560684204,
      "learning_rate": 9.888157894736843e-05,
      "loss": 0.4321,
      "step": 24261
    },
    {
      "epoch": 1.35860678687423,
      "grad_norm": 1.3706817626953125,
      "learning_rate": 9.888131578947369e-05,
      "loss": 0.5732,
      "step": 24262
    },
    {
      "epoch": 1.358662784186359,
      "grad_norm": 1.3490325212478638,
      "learning_rate": 9.888105263157895e-05,
      "loss": 0.4374,
      "step": 24263
    },
    {
      "epoch": 1.358718781498488,
      "grad_norm": 1.118749737739563,
      "learning_rate": 9.888078947368421e-05,
      "loss": 0.3573,
      "step": 24264
    },
    {
      "epoch": 1.358774778810617,
      "grad_norm": 1.126299262046814,
      "learning_rate": 9.888052631578948e-05,
      "loss": 0.2995,
      "step": 24265
    },
    {
      "epoch": 1.358830776122746,
      "grad_norm": 1.1920883655548096,
      "learning_rate": 9.888026315789474e-05,
      "loss": 0.4185,
      "step": 24266
    },
    {
      "epoch": 1.358886773434875,
      "grad_norm": 1.254758358001709,
      "learning_rate": 9.888e-05,
      "loss": 0.4563,
      "step": 24267
    },
    {
      "epoch": 1.358942770747004,
      "grad_norm": 1.0732206106185913,
      "learning_rate": 9.887973684210526e-05,
      "loss": 0.3767,
      "step": 24268
    },
    {
      "epoch": 1.358998768059133,
      "grad_norm": 1.278638243675232,
      "learning_rate": 9.887947368421053e-05,
      "loss": 0.4268,
      "step": 24269
    },
    {
      "epoch": 1.3590547653712621,
      "grad_norm": 1.693253517150879,
      "learning_rate": 9.88792105263158e-05,
      "loss": 0.4124,
      "step": 24270
    },
    {
      "epoch": 1.3591107626833911,
      "grad_norm": 1.4540988206863403,
      "learning_rate": 9.887894736842105e-05,
      "loss": 0.4139,
      "step": 24271
    },
    {
      "epoch": 1.3591667599955202,
      "grad_norm": 1.2501556873321533,
      "learning_rate": 9.887868421052631e-05,
      "loss": 0.467,
      "step": 24272
    },
    {
      "epoch": 1.3592227573076492,
      "grad_norm": 1.3515464067459106,
      "learning_rate": 9.887842105263159e-05,
      "loss": 0.5135,
      "step": 24273
    },
    {
      "epoch": 1.3592787546197782,
      "grad_norm": 1.2665053606033325,
      "learning_rate": 9.887815789473685e-05,
      "loss": 0.3851,
      "step": 24274
    },
    {
      "epoch": 1.3593347519319072,
      "grad_norm": 1.5510942935943604,
      "learning_rate": 9.887789473684212e-05,
      "loss": 0.5612,
      "step": 24275
    },
    {
      "epoch": 1.3593907492440362,
      "grad_norm": 1.9409633874893188,
      "learning_rate": 9.887763157894737e-05,
      "loss": 0.6694,
      "step": 24276
    },
    {
      "epoch": 1.3594467465561653,
      "grad_norm": 1.3130944967269897,
      "learning_rate": 9.887736842105264e-05,
      "loss": 0.4047,
      "step": 24277
    },
    {
      "epoch": 1.3595027438682943,
      "grad_norm": 1.4870585203170776,
      "learning_rate": 9.88771052631579e-05,
      "loss": 0.4797,
      "step": 24278
    },
    {
      "epoch": 1.3595587411804233,
      "grad_norm": 1.2149924039840698,
      "learning_rate": 9.887684210526317e-05,
      "loss": 0.3805,
      "step": 24279
    },
    {
      "epoch": 1.3596147384925523,
      "grad_norm": 1.4774595499038696,
      "learning_rate": 9.887657894736842e-05,
      "loss": 0.5419,
      "step": 24280
    },
    {
      "epoch": 1.3596707358046813,
      "grad_norm": 1.419573187828064,
      "learning_rate": 9.887631578947368e-05,
      "loss": 0.5009,
      "step": 24281
    },
    {
      "epoch": 1.3597267331168104,
      "grad_norm": 1.3416260480880737,
      "learning_rate": 9.887605263157895e-05,
      "loss": 0.4692,
      "step": 24282
    },
    {
      "epoch": 1.3597827304289394,
      "grad_norm": 1.3263635635375977,
      "learning_rate": 9.887578947368421e-05,
      "loss": 0.4656,
      "step": 24283
    },
    {
      "epoch": 1.3598387277410684,
      "grad_norm": 1.96281099319458,
      "learning_rate": 9.887552631578949e-05,
      "loss": 0.4804,
      "step": 24284
    },
    {
      "epoch": 1.3598947250531974,
      "grad_norm": 1.3477357625961304,
      "learning_rate": 9.887526315789473e-05,
      "loss": 0.4723,
      "step": 24285
    },
    {
      "epoch": 1.3599507223653264,
      "grad_norm": 1.3089745044708252,
      "learning_rate": 9.8875e-05,
      "loss": 0.5166,
      "step": 24286
    },
    {
      "epoch": 1.3600067196774555,
      "grad_norm": 1.3642199039459229,
      "learning_rate": 9.887473684210526e-05,
      "loss": 0.5577,
      "step": 24287
    },
    {
      "epoch": 1.3600627169895845,
      "grad_norm": 1.3563685417175293,
      "learning_rate": 9.887447368421054e-05,
      "loss": 0.6061,
      "step": 24288
    },
    {
      "epoch": 1.3601187143017135,
      "grad_norm": 1.2317285537719727,
      "learning_rate": 9.88742105263158e-05,
      "loss": 0.3844,
      "step": 24289
    },
    {
      "epoch": 1.3601747116138425,
      "grad_norm": 1.2608413696289062,
      "learning_rate": 9.887394736842106e-05,
      "loss": 0.5228,
      "step": 24290
    },
    {
      "epoch": 1.3602307089259715,
      "grad_norm": 1.5067768096923828,
      "learning_rate": 9.887368421052632e-05,
      "loss": 0.4704,
      "step": 24291
    },
    {
      "epoch": 1.3602867062381006,
      "grad_norm": 1.5356553792953491,
      "learning_rate": 9.887342105263159e-05,
      "loss": 0.7033,
      "step": 24292
    },
    {
      "epoch": 1.3603427035502296,
      "grad_norm": 1.1460360288619995,
      "learning_rate": 9.887315789473685e-05,
      "loss": 0.399,
      "step": 24293
    },
    {
      "epoch": 1.3603987008623586,
      "grad_norm": 1.5921638011932373,
      "learning_rate": 9.887289473684211e-05,
      "loss": 0.5124,
      "step": 24294
    },
    {
      "epoch": 1.3604546981744876,
      "grad_norm": 1.4019817113876343,
      "learning_rate": 9.887263157894737e-05,
      "loss": 0.4948,
      "step": 24295
    },
    {
      "epoch": 1.3605106954866166,
      "grad_norm": 1.5591027736663818,
      "learning_rate": 9.887236842105264e-05,
      "loss": 0.4862,
      "step": 24296
    },
    {
      "epoch": 1.3605666927987456,
      "grad_norm": 1.374293565750122,
      "learning_rate": 9.88721052631579e-05,
      "loss": 0.5368,
      "step": 24297
    },
    {
      "epoch": 1.3606226901108747,
      "grad_norm": 1.49769926071167,
      "learning_rate": 9.887184210526316e-05,
      "loss": 0.55,
      "step": 24298
    },
    {
      "epoch": 1.3606786874230037,
      "grad_norm": 1.1991252899169922,
      "learning_rate": 9.887157894736842e-05,
      "loss": 0.4687,
      "step": 24299
    },
    {
      "epoch": 1.3607346847351327,
      "grad_norm": 1.3198788166046143,
      "learning_rate": 9.887131578947368e-05,
      "loss": 0.4489,
      "step": 24300
    },
    {
      "epoch": 1.3607906820472617,
      "grad_norm": 1.2617061138153076,
      "learning_rate": 9.887105263157895e-05,
      "loss": 0.395,
      "step": 24301
    },
    {
      "epoch": 1.3608466793593907,
      "grad_norm": 1.3370567560195923,
      "learning_rate": 9.887078947368421e-05,
      "loss": 0.3999,
      "step": 24302
    },
    {
      "epoch": 1.3609026766715198,
      "grad_norm": 1.3850817680358887,
      "learning_rate": 9.887052631578947e-05,
      "loss": 0.5725,
      "step": 24303
    },
    {
      "epoch": 1.3609586739836488,
      "grad_norm": 1.5199081897735596,
      "learning_rate": 9.887026315789473e-05,
      "loss": 0.4016,
      "step": 24304
    },
    {
      "epoch": 1.3610146712957778,
      "grad_norm": 1.3331574201583862,
      "learning_rate": 9.887000000000001e-05,
      "loss": 0.4695,
      "step": 24305
    },
    {
      "epoch": 1.3610706686079068,
      "grad_norm": 1.2357096672058105,
      "learning_rate": 9.886973684210527e-05,
      "loss": 0.4227,
      "step": 24306
    },
    {
      "epoch": 1.3611266659200358,
      "grad_norm": 1.3251912593841553,
      "learning_rate": 9.886947368421054e-05,
      "loss": 0.4537,
      "step": 24307
    },
    {
      "epoch": 1.3611826632321649,
      "grad_norm": 1.1950063705444336,
      "learning_rate": 9.886921052631579e-05,
      "loss": 0.4806,
      "step": 24308
    },
    {
      "epoch": 1.3612386605442939,
      "grad_norm": 1.2474392652511597,
      "learning_rate": 9.886894736842106e-05,
      "loss": 0.4358,
      "step": 24309
    },
    {
      "epoch": 1.361294657856423,
      "grad_norm": 1.0946778059005737,
      "learning_rate": 9.886868421052632e-05,
      "loss": 0.3485,
      "step": 24310
    },
    {
      "epoch": 1.361350655168552,
      "grad_norm": 1.3305790424346924,
      "learning_rate": 9.886842105263159e-05,
      "loss": 0.3922,
      "step": 24311
    },
    {
      "epoch": 1.361406652480681,
      "grad_norm": 1.1710745096206665,
      "learning_rate": 9.886815789473685e-05,
      "loss": 0.3931,
      "step": 24312
    },
    {
      "epoch": 1.36146264979281,
      "grad_norm": 1.5039604902267456,
      "learning_rate": 9.886789473684211e-05,
      "loss": 0.4597,
      "step": 24313
    },
    {
      "epoch": 1.361518647104939,
      "grad_norm": 1.3262319564819336,
      "learning_rate": 9.886763157894737e-05,
      "loss": 0.5,
      "step": 24314
    },
    {
      "epoch": 1.361574644417068,
      "grad_norm": 1.3019124269485474,
      "learning_rate": 9.886736842105265e-05,
      "loss": 0.5419,
      "step": 24315
    },
    {
      "epoch": 1.361630641729197,
      "grad_norm": 1.5411773920059204,
      "learning_rate": 9.88671052631579e-05,
      "loss": 0.4171,
      "step": 24316
    },
    {
      "epoch": 1.361686639041326,
      "grad_norm": 1.4485447406768799,
      "learning_rate": 9.886684210526315e-05,
      "loss": 0.4519,
      "step": 24317
    },
    {
      "epoch": 1.361742636353455,
      "grad_norm": 1.3841382265090942,
      "learning_rate": 9.886657894736842e-05,
      "loss": 0.4242,
      "step": 24318
    },
    {
      "epoch": 1.361798633665584,
      "grad_norm": 1.1673097610473633,
      "learning_rate": 9.886631578947368e-05,
      "loss": 0.4415,
      "step": 24319
    },
    {
      "epoch": 1.361854630977713,
      "grad_norm": 1.6136304140090942,
      "learning_rate": 9.886605263157896e-05,
      "loss": 0.546,
      "step": 24320
    },
    {
      "epoch": 1.361910628289842,
      "grad_norm": 1.567229151725769,
      "learning_rate": 9.886578947368422e-05,
      "loss": 0.4276,
      "step": 24321
    },
    {
      "epoch": 1.3619666256019711,
      "grad_norm": 1.3506273031234741,
      "learning_rate": 9.886552631578948e-05,
      "loss": 0.4994,
      "step": 24322
    },
    {
      "epoch": 1.3620226229141001,
      "grad_norm": 1.3055624961853027,
      "learning_rate": 9.886526315789474e-05,
      "loss": 0.347,
      "step": 24323
    },
    {
      "epoch": 1.3620786202262292,
      "grad_norm": 1.38615882396698,
      "learning_rate": 9.886500000000001e-05,
      "loss": 0.4291,
      "step": 24324
    },
    {
      "epoch": 1.3621346175383582,
      "grad_norm": 1.3798707723617554,
      "learning_rate": 9.886473684210527e-05,
      "loss": 0.4276,
      "step": 24325
    },
    {
      "epoch": 1.3621906148504872,
      "grad_norm": 1.1843464374542236,
      "learning_rate": 9.886447368421053e-05,
      "loss": 0.4777,
      "step": 24326
    },
    {
      "epoch": 1.3622466121626162,
      "grad_norm": 1.2293568849563599,
      "learning_rate": 9.886421052631579e-05,
      "loss": 0.448,
      "step": 24327
    },
    {
      "epoch": 1.3623026094747452,
      "grad_norm": 1.2945222854614258,
      "learning_rate": 9.886394736842106e-05,
      "loss": 0.4568,
      "step": 24328
    },
    {
      "epoch": 1.3623586067868743,
      "grad_norm": 1.490443229675293,
      "learning_rate": 9.886368421052632e-05,
      "loss": 0.6576,
      "step": 24329
    },
    {
      "epoch": 1.3624146040990033,
      "grad_norm": 1.3081691265106201,
      "learning_rate": 9.88634210526316e-05,
      "loss": 0.516,
      "step": 24330
    },
    {
      "epoch": 1.3624706014111323,
      "grad_norm": 2.057906150817871,
      "learning_rate": 9.886315789473684e-05,
      "loss": 0.4726,
      "step": 24331
    },
    {
      "epoch": 1.3625265987232613,
      "grad_norm": 1.1586756706237793,
      "learning_rate": 9.886289473684211e-05,
      "loss": 0.364,
      "step": 24332
    },
    {
      "epoch": 1.3625825960353903,
      "grad_norm": 1.5228568315505981,
      "learning_rate": 9.886263157894737e-05,
      "loss": 0.5054,
      "step": 24333
    },
    {
      "epoch": 1.3626385933475194,
      "grad_norm": 1.2080132961273193,
      "learning_rate": 9.886236842105263e-05,
      "loss": 0.5405,
      "step": 24334
    },
    {
      "epoch": 1.3626945906596484,
      "grad_norm": 1.8775131702423096,
      "learning_rate": 9.88621052631579e-05,
      "loss": 0.5727,
      "step": 24335
    },
    {
      "epoch": 1.3627505879717774,
      "grad_norm": 1.4475866556167603,
      "learning_rate": 9.886184210526315e-05,
      "loss": 0.4302,
      "step": 24336
    },
    {
      "epoch": 1.3628065852839064,
      "grad_norm": 1.6076793670654297,
      "learning_rate": 9.886157894736843e-05,
      "loss": 0.5248,
      "step": 24337
    },
    {
      "epoch": 1.3628625825960354,
      "grad_norm": 1.269997000694275,
      "learning_rate": 9.886131578947369e-05,
      "loss": 0.3485,
      "step": 24338
    },
    {
      "epoch": 1.3629185799081645,
      "grad_norm": 1.2178103923797607,
      "learning_rate": 9.886105263157896e-05,
      "loss": 0.4373,
      "step": 24339
    },
    {
      "epoch": 1.3629745772202935,
      "grad_norm": 1.101955771446228,
      "learning_rate": 9.88607894736842e-05,
      "loss": 0.3823,
      "step": 24340
    },
    {
      "epoch": 1.3630305745324225,
      "grad_norm": 1.411412239074707,
      "learning_rate": 9.886052631578948e-05,
      "loss": 0.5211,
      "step": 24341
    },
    {
      "epoch": 1.3630865718445515,
      "grad_norm": 1.3933584690093994,
      "learning_rate": 9.886026315789474e-05,
      "loss": 0.5002,
      "step": 24342
    },
    {
      "epoch": 1.3631425691566805,
      "grad_norm": 1.3537825345993042,
      "learning_rate": 9.886000000000001e-05,
      "loss": 0.5249,
      "step": 24343
    },
    {
      "epoch": 1.3631985664688095,
      "grad_norm": 1.3923225402832031,
      "learning_rate": 9.885973684210527e-05,
      "loss": 0.4003,
      "step": 24344
    },
    {
      "epoch": 1.3632545637809386,
      "grad_norm": 1.1837642192840576,
      "learning_rate": 9.885947368421053e-05,
      "loss": 0.4657,
      "step": 24345
    },
    {
      "epoch": 1.3633105610930676,
      "grad_norm": 1.1448791027069092,
      "learning_rate": 9.885921052631579e-05,
      "loss": 0.4039,
      "step": 24346
    },
    {
      "epoch": 1.3633665584051966,
      "grad_norm": 1.3756505250930786,
      "learning_rate": 9.885894736842106e-05,
      "loss": 0.4797,
      "step": 24347
    },
    {
      "epoch": 1.3634225557173256,
      "grad_norm": 1.3280857801437378,
      "learning_rate": 9.885868421052632e-05,
      "loss": 0.4247,
      "step": 24348
    },
    {
      "epoch": 1.3634785530294546,
      "grad_norm": 1.4541010856628418,
      "learning_rate": 9.885842105263158e-05,
      "loss": 0.4844,
      "step": 24349
    },
    {
      "epoch": 1.3635345503415837,
      "grad_norm": 1.3154367208480835,
      "learning_rate": 9.885815789473684e-05,
      "loss": 0.5767,
      "step": 24350
    },
    {
      "epoch": 1.3635905476537127,
      "grad_norm": 1.5184102058410645,
      "learning_rate": 9.88578947368421e-05,
      "loss": 0.5135,
      "step": 24351
    },
    {
      "epoch": 1.3636465449658417,
      "grad_norm": 1.2757196426391602,
      "learning_rate": 9.885763157894738e-05,
      "loss": 0.4336,
      "step": 24352
    },
    {
      "epoch": 1.3637025422779707,
      "grad_norm": 1.3980156183242798,
      "learning_rate": 9.885736842105264e-05,
      "loss": 0.4931,
      "step": 24353
    },
    {
      "epoch": 1.3637585395900997,
      "grad_norm": 1.666050672531128,
      "learning_rate": 9.88571052631579e-05,
      "loss": 0.3957,
      "step": 24354
    },
    {
      "epoch": 1.3638145369022288,
      "grad_norm": 1.250495195388794,
      "learning_rate": 9.885684210526316e-05,
      "loss": 0.4734,
      "step": 24355
    },
    {
      "epoch": 1.3638705342143578,
      "grad_norm": 1.1907325983047485,
      "learning_rate": 9.885657894736843e-05,
      "loss": 0.4164,
      "step": 24356
    },
    {
      "epoch": 1.3639265315264868,
      "grad_norm": 1.5041922330856323,
      "learning_rate": 9.885631578947369e-05,
      "loss": 0.5274,
      "step": 24357
    },
    {
      "epoch": 1.3639825288386158,
      "grad_norm": 1.2572085857391357,
      "learning_rate": 9.885605263157895e-05,
      "loss": 0.4697,
      "step": 24358
    },
    {
      "epoch": 1.3640385261507448,
      "grad_norm": 1.5445501804351807,
      "learning_rate": 9.885578947368421e-05,
      "loss": 0.5151,
      "step": 24359
    },
    {
      "epoch": 1.3640945234628739,
      "grad_norm": 1.4259023666381836,
      "learning_rate": 9.885552631578948e-05,
      "loss": 0.5375,
      "step": 24360
    },
    {
      "epoch": 1.3641505207750029,
      "grad_norm": 1.1657230854034424,
      "learning_rate": 9.885526315789474e-05,
      "loss": 0.4456,
      "step": 24361
    },
    {
      "epoch": 1.364206518087132,
      "grad_norm": 1.5526230335235596,
      "learning_rate": 9.885500000000001e-05,
      "loss": 0.4723,
      "step": 24362
    },
    {
      "epoch": 1.364262515399261,
      "grad_norm": 1.306484580039978,
      "learning_rate": 9.885473684210526e-05,
      "loss": 0.4282,
      "step": 24363
    },
    {
      "epoch": 1.36431851271139,
      "grad_norm": 1.1856343746185303,
      "learning_rate": 9.885447368421053e-05,
      "loss": 0.4246,
      "step": 24364
    },
    {
      "epoch": 1.364374510023519,
      "grad_norm": 1.329923391342163,
      "learning_rate": 9.88542105263158e-05,
      "loss": 0.502,
      "step": 24365
    },
    {
      "epoch": 1.364430507335648,
      "grad_norm": 1.2879289388656616,
      "learning_rate": 9.885394736842107e-05,
      "loss": 0.4894,
      "step": 24366
    },
    {
      "epoch": 1.364486504647777,
      "grad_norm": 1.3006231784820557,
      "learning_rate": 9.885368421052633e-05,
      "loss": 0.462,
      "step": 24367
    },
    {
      "epoch": 1.364542501959906,
      "grad_norm": 1.650084376335144,
      "learning_rate": 9.885342105263157e-05,
      "loss": 0.387,
      "step": 24368
    },
    {
      "epoch": 1.3645984992720348,
      "grad_norm": 1.4683609008789062,
      "learning_rate": 9.885315789473685e-05,
      "loss": 0.3361,
      "step": 24369
    },
    {
      "epoch": 1.3646544965841638,
      "grad_norm": 1.5280910730361938,
      "learning_rate": 9.88528947368421e-05,
      "loss": 0.4098,
      "step": 24370
    },
    {
      "epoch": 1.3647104938962928,
      "grad_norm": 1.3542401790618896,
      "learning_rate": 9.885263157894738e-05,
      "loss": 0.4193,
      "step": 24371
    },
    {
      "epoch": 1.3647664912084219,
      "grad_norm": 1.2905116081237793,
      "learning_rate": 9.885236842105263e-05,
      "loss": 0.4167,
      "step": 24372
    },
    {
      "epoch": 1.3648224885205509,
      "grad_norm": 1.6238502264022827,
      "learning_rate": 9.88521052631579e-05,
      "loss": 0.495,
      "step": 24373
    },
    {
      "epoch": 1.36487848583268,
      "grad_norm": 1.3055897951126099,
      "learning_rate": 9.885184210526316e-05,
      "loss": 0.4439,
      "step": 24374
    },
    {
      "epoch": 1.364934483144809,
      "grad_norm": 1.3446067571640015,
      "learning_rate": 9.885157894736843e-05,
      "loss": 0.3404,
      "step": 24375
    },
    {
      "epoch": 1.364990480456938,
      "grad_norm": 1.355931043624878,
      "learning_rate": 9.885131578947369e-05,
      "loss": 0.4637,
      "step": 24376
    },
    {
      "epoch": 1.365046477769067,
      "grad_norm": 1.3010797500610352,
      "learning_rate": 9.885105263157895e-05,
      "loss": 0.457,
      "step": 24377
    },
    {
      "epoch": 1.365102475081196,
      "grad_norm": 1.3977206945419312,
      "learning_rate": 9.885078947368421e-05,
      "loss": 0.5038,
      "step": 24378
    },
    {
      "epoch": 1.365158472393325,
      "grad_norm": 1.1111963987350464,
      "learning_rate": 9.885052631578948e-05,
      "loss": 0.3715,
      "step": 24379
    },
    {
      "epoch": 1.365214469705454,
      "grad_norm": 1.1828562021255493,
      "learning_rate": 9.885026315789474e-05,
      "loss": 0.4013,
      "step": 24380
    },
    {
      "epoch": 1.365270467017583,
      "grad_norm": 1.5111533403396606,
      "learning_rate": 9.885e-05,
      "loss": 0.4277,
      "step": 24381
    },
    {
      "epoch": 1.365326464329712,
      "grad_norm": 24.50279426574707,
      "learning_rate": 9.884973684210526e-05,
      "loss": 0.5584,
      "step": 24382
    },
    {
      "epoch": 1.365382461641841,
      "grad_norm": 1.3131940364837646,
      "learning_rate": 9.884947368421054e-05,
      "loss": 0.3771,
      "step": 24383
    },
    {
      "epoch": 1.36543845895397,
      "grad_norm": 1.5126020908355713,
      "learning_rate": 9.88492105263158e-05,
      "loss": 0.5313,
      "step": 24384
    },
    {
      "epoch": 1.365494456266099,
      "grad_norm": 1.2867248058319092,
      "learning_rate": 9.884894736842106e-05,
      "loss": 0.4346,
      "step": 24385
    },
    {
      "epoch": 1.3655504535782281,
      "grad_norm": 1.7722824811935425,
      "learning_rate": 9.884868421052632e-05,
      "loss": 0.4468,
      "step": 24386
    },
    {
      "epoch": 1.3656064508903571,
      "grad_norm": 1.28387451171875,
      "learning_rate": 9.884842105263158e-05,
      "loss": 0.4015,
      "step": 24387
    },
    {
      "epoch": 1.3656624482024862,
      "grad_norm": 1.6635466814041138,
      "learning_rate": 9.884815789473685e-05,
      "loss": 0.4431,
      "step": 24388
    },
    {
      "epoch": 1.3657184455146152,
      "grad_norm": 1.243938684463501,
      "learning_rate": 9.884789473684211e-05,
      "loss": 0.47,
      "step": 24389
    },
    {
      "epoch": 1.3657744428267442,
      "grad_norm": 1.4119457006454468,
      "learning_rate": 9.884763157894737e-05,
      "loss": 0.5061,
      "step": 24390
    },
    {
      "epoch": 1.3658304401388732,
      "grad_norm": 1.2198487520217896,
      "learning_rate": 9.884736842105263e-05,
      "loss": 0.3484,
      "step": 24391
    },
    {
      "epoch": 1.3658864374510022,
      "grad_norm": 1.2817981243133545,
      "learning_rate": 9.88471052631579e-05,
      "loss": 0.5075,
      "step": 24392
    },
    {
      "epoch": 1.3659424347631313,
      "grad_norm": 1.5079666376113892,
      "learning_rate": 9.884684210526316e-05,
      "loss": 0.5416,
      "step": 24393
    },
    {
      "epoch": 1.3659984320752603,
      "grad_norm": 1.1374844312667847,
      "learning_rate": 9.884657894736843e-05,
      "loss": 0.3744,
      "step": 24394
    },
    {
      "epoch": 1.3660544293873893,
      "grad_norm": 1.438244342803955,
      "learning_rate": 9.884631578947368e-05,
      "loss": 0.4617,
      "step": 24395
    },
    {
      "epoch": 1.3661104266995183,
      "grad_norm": 1.3679901361465454,
      "learning_rate": 9.884605263157895e-05,
      "loss": 0.4839,
      "step": 24396
    },
    {
      "epoch": 1.3661664240116473,
      "grad_norm": 1.2486059665679932,
      "learning_rate": 9.884578947368421e-05,
      "loss": 0.4152,
      "step": 24397
    },
    {
      "epoch": 1.3662224213237764,
      "grad_norm": 1.423294186592102,
      "learning_rate": 9.884552631578949e-05,
      "loss": 0.4701,
      "step": 24398
    },
    {
      "epoch": 1.3662784186359054,
      "grad_norm": 1.3104181289672852,
      "learning_rate": 9.884526315789475e-05,
      "loss": 0.5171,
      "step": 24399
    },
    {
      "epoch": 1.3663344159480344,
      "grad_norm": 1.3045990467071533,
      "learning_rate": 9.8845e-05,
      "loss": 0.4298,
      "step": 24400
    },
    {
      "epoch": 1.3663904132601634,
      "grad_norm": 1.4200360774993896,
      "learning_rate": 9.884473684210527e-05,
      "loss": 0.5425,
      "step": 24401
    },
    {
      "epoch": 1.3664464105722924,
      "grad_norm": 1.6482840776443481,
      "learning_rate": 9.884447368421053e-05,
      "loss": 0.644,
      "step": 24402
    },
    {
      "epoch": 1.3665024078844215,
      "grad_norm": 1.2006323337554932,
      "learning_rate": 9.88442105263158e-05,
      "loss": 0.4565,
      "step": 24403
    },
    {
      "epoch": 1.3665584051965505,
      "grad_norm": 1.4098336696624756,
      "learning_rate": 9.884394736842106e-05,
      "loss": 0.4197,
      "step": 24404
    },
    {
      "epoch": 1.3666144025086795,
      "grad_norm": 1.4508843421936035,
      "learning_rate": 9.884368421052632e-05,
      "loss": 0.5249,
      "step": 24405
    },
    {
      "epoch": 1.3666703998208085,
      "grad_norm": 1.574019432067871,
      "learning_rate": 9.884342105263158e-05,
      "loss": 0.6243,
      "step": 24406
    },
    {
      "epoch": 1.3667263971329375,
      "grad_norm": 1.6158201694488525,
      "learning_rate": 9.884315789473685e-05,
      "loss": 0.5481,
      "step": 24407
    },
    {
      "epoch": 1.3667823944450666,
      "grad_norm": 1.360234260559082,
      "learning_rate": 9.884289473684211e-05,
      "loss": 0.5952,
      "step": 24408
    },
    {
      "epoch": 1.3668383917571956,
      "grad_norm": 1.18491530418396,
      "learning_rate": 9.884263157894737e-05,
      "loss": 0.2932,
      "step": 24409
    },
    {
      "epoch": 1.3668943890693246,
      "grad_norm": 1.4678871631622314,
      "learning_rate": 9.884236842105263e-05,
      "loss": 0.4682,
      "step": 24410
    },
    {
      "epoch": 1.3669503863814536,
      "grad_norm": 1.540473222732544,
      "learning_rate": 9.88421052631579e-05,
      "loss": 0.4994,
      "step": 24411
    },
    {
      "epoch": 1.3670063836935826,
      "grad_norm": 1.4704207181930542,
      "learning_rate": 9.884184210526316e-05,
      "loss": 0.4684,
      "step": 24412
    },
    {
      "epoch": 1.3670623810057116,
      "grad_norm": 1.3927764892578125,
      "learning_rate": 9.884157894736842e-05,
      "loss": 0.4538,
      "step": 24413
    },
    {
      "epoch": 1.3671183783178407,
      "grad_norm": 1.0648525953292847,
      "learning_rate": 9.884131578947368e-05,
      "loss": 0.3678,
      "step": 24414
    },
    {
      "epoch": 1.3671743756299697,
      "grad_norm": 1.3720592260360718,
      "learning_rate": 9.884105263157896e-05,
      "loss": 0.4999,
      "step": 24415
    },
    {
      "epoch": 1.3672303729420987,
      "grad_norm": 1.8324408531188965,
      "learning_rate": 9.884078947368422e-05,
      "loss": 0.5443,
      "step": 24416
    },
    {
      "epoch": 1.3672863702542277,
      "grad_norm": 1.3010083436965942,
      "learning_rate": 9.884052631578949e-05,
      "loss": 0.3828,
      "step": 24417
    },
    {
      "epoch": 1.3673423675663567,
      "grad_norm": 1.4595146179199219,
      "learning_rate": 9.884026315789474e-05,
      "loss": 0.6761,
      "step": 24418
    },
    {
      "epoch": 1.3673983648784858,
      "grad_norm": 1.4448872804641724,
      "learning_rate": 9.884e-05,
      "loss": 0.4775,
      "step": 24419
    },
    {
      "epoch": 1.3674543621906148,
      "grad_norm": 2.1411805152893066,
      "learning_rate": 9.883973684210527e-05,
      "loss": 0.7041,
      "step": 24420
    },
    {
      "epoch": 1.3675103595027438,
      "grad_norm": 1.5478935241699219,
      "learning_rate": 9.883947368421053e-05,
      "loss": 0.503,
      "step": 24421
    },
    {
      "epoch": 1.3675663568148728,
      "grad_norm": 1.2933239936828613,
      "learning_rate": 9.88392105263158e-05,
      "loss": 0.3763,
      "step": 24422
    },
    {
      "epoch": 1.3676223541270018,
      "grad_norm": 1.6128108501434326,
      "learning_rate": 9.883894736842105e-05,
      "loss": 0.5932,
      "step": 24423
    },
    {
      "epoch": 1.3676783514391309,
      "grad_norm": 1.8574882745742798,
      "learning_rate": 9.883868421052632e-05,
      "loss": 0.4911,
      "step": 24424
    },
    {
      "epoch": 1.3677343487512599,
      "grad_norm": 1.3461345434188843,
      "learning_rate": 9.883842105263158e-05,
      "loss": 0.4497,
      "step": 24425
    },
    {
      "epoch": 1.367790346063389,
      "grad_norm": 1.1728650331497192,
      "learning_rate": 9.883815789473685e-05,
      "loss": 0.3946,
      "step": 24426
    },
    {
      "epoch": 1.367846343375518,
      "grad_norm": 1.2892390489578247,
      "learning_rate": 9.88378947368421e-05,
      "loss": 0.4278,
      "step": 24427
    },
    {
      "epoch": 1.367902340687647,
      "grad_norm": 1.397377848625183,
      "learning_rate": 9.883763157894737e-05,
      "loss": 0.3746,
      "step": 24428
    },
    {
      "epoch": 1.367958337999776,
      "grad_norm": 1.3317534923553467,
      "learning_rate": 9.883736842105263e-05,
      "loss": 0.449,
      "step": 24429
    },
    {
      "epoch": 1.368014335311905,
      "grad_norm": 1.370968222618103,
      "learning_rate": 9.883710526315791e-05,
      "loss": 0.5822,
      "step": 24430
    },
    {
      "epoch": 1.368070332624034,
      "grad_norm": 1.4074839353561401,
      "learning_rate": 9.883684210526317e-05,
      "loss": 0.4826,
      "step": 24431
    },
    {
      "epoch": 1.368126329936163,
      "grad_norm": 1.3753305673599243,
      "learning_rate": 9.883657894736843e-05,
      "loss": 0.3796,
      "step": 24432
    },
    {
      "epoch": 1.368182327248292,
      "grad_norm": 1.620547890663147,
      "learning_rate": 9.883631578947369e-05,
      "loss": 0.5574,
      "step": 24433
    },
    {
      "epoch": 1.368238324560421,
      "grad_norm": 1.3297902345657349,
      "learning_rate": 9.883605263157896e-05,
      "loss": 0.4992,
      "step": 24434
    },
    {
      "epoch": 1.36829432187255,
      "grad_norm": 1.27519953250885,
      "learning_rate": 9.883578947368422e-05,
      "loss": 0.4256,
      "step": 24435
    },
    {
      "epoch": 1.368350319184679,
      "grad_norm": 1.8830041885375977,
      "learning_rate": 9.883552631578948e-05,
      "loss": 0.5545,
      "step": 24436
    },
    {
      "epoch": 1.368406316496808,
      "grad_norm": 1.4125512838363647,
      "learning_rate": 9.883526315789474e-05,
      "loss": 0.4259,
      "step": 24437
    },
    {
      "epoch": 1.3684623138089371,
      "grad_norm": 1.3391048908233643,
      "learning_rate": 9.8835e-05,
      "loss": 0.4256,
      "step": 24438
    },
    {
      "epoch": 1.3685183111210661,
      "grad_norm": 1.4954277276992798,
      "learning_rate": 9.883473684210527e-05,
      "loss": 0.5805,
      "step": 24439
    },
    {
      "epoch": 1.3685743084331952,
      "grad_norm": 1.4033195972442627,
      "learning_rate": 9.883447368421053e-05,
      "loss": 0.4628,
      "step": 24440
    },
    {
      "epoch": 1.3686303057453242,
      "grad_norm": 1.6419473886489868,
      "learning_rate": 9.883421052631579e-05,
      "loss": 0.4202,
      "step": 24441
    },
    {
      "epoch": 1.3686863030574532,
      "grad_norm": 1.9852774143218994,
      "learning_rate": 9.883394736842105e-05,
      "loss": 0.5977,
      "step": 24442
    },
    {
      "epoch": 1.3687423003695822,
      "grad_norm": 1.5373849868774414,
      "learning_rate": 9.883368421052632e-05,
      "loss": 0.7519,
      "step": 24443
    },
    {
      "epoch": 1.3687982976817112,
      "grad_norm": 1.4142190217971802,
      "learning_rate": 9.883342105263158e-05,
      "loss": 0.5456,
      "step": 24444
    },
    {
      "epoch": 1.3688542949938403,
      "grad_norm": 1.35452139377594,
      "learning_rate": 9.883315789473684e-05,
      "loss": 0.5445,
      "step": 24445
    },
    {
      "epoch": 1.3689102923059693,
      "grad_norm": 1.2676734924316406,
      "learning_rate": 9.88328947368421e-05,
      "loss": 0.3916,
      "step": 24446
    },
    {
      "epoch": 1.3689662896180983,
      "grad_norm": 1.472924828529358,
      "learning_rate": 9.883263157894738e-05,
      "loss": 0.5168,
      "step": 24447
    },
    {
      "epoch": 1.3690222869302273,
      "grad_norm": 1.5486114025115967,
      "learning_rate": 9.883236842105264e-05,
      "loss": 0.5228,
      "step": 24448
    },
    {
      "epoch": 1.3690782842423563,
      "grad_norm": 1.091369390487671,
      "learning_rate": 9.883210526315791e-05,
      "loss": 0.3869,
      "step": 24449
    },
    {
      "epoch": 1.3691342815544854,
      "grad_norm": 1.6277631521224976,
      "learning_rate": 9.883184210526316e-05,
      "loss": 0.4969,
      "step": 24450
    },
    {
      "epoch": 1.3691902788666144,
      "grad_norm": 1.6347336769104004,
      "learning_rate": 9.883157894736843e-05,
      "loss": 0.5751,
      "step": 24451
    },
    {
      "epoch": 1.3692462761787434,
      "grad_norm": 1.3268325328826904,
      "learning_rate": 9.883131578947369e-05,
      "loss": 0.3474,
      "step": 24452
    },
    {
      "epoch": 1.3693022734908724,
      "grad_norm": 1.3420542478561401,
      "learning_rate": 9.883105263157896e-05,
      "loss": 0.4211,
      "step": 24453
    },
    {
      "epoch": 1.3693582708030014,
      "grad_norm": 1.2873342037200928,
      "learning_rate": 9.883078947368422e-05,
      "loss": 0.5331,
      "step": 24454
    },
    {
      "epoch": 1.3694142681151305,
      "grad_norm": 1.4083830118179321,
      "learning_rate": 9.883052631578947e-05,
      "loss": 0.4748,
      "step": 24455
    },
    {
      "epoch": 1.3694702654272595,
      "grad_norm": 2.098632335662842,
      "learning_rate": 9.883026315789474e-05,
      "loss": 0.5454,
      "step": 24456
    },
    {
      "epoch": 1.3695262627393885,
      "grad_norm": 1.594950795173645,
      "learning_rate": 9.883e-05,
      "loss": 0.5077,
      "step": 24457
    },
    {
      "epoch": 1.3695822600515175,
      "grad_norm": 1.342111349105835,
      "learning_rate": 9.882973684210527e-05,
      "loss": 0.5786,
      "step": 24458
    },
    {
      "epoch": 1.3696382573636465,
      "grad_norm": 1.401429295539856,
      "learning_rate": 9.882947368421053e-05,
      "loss": 0.4681,
      "step": 24459
    },
    {
      "epoch": 1.3696942546757755,
      "grad_norm": 2.9357070922851562,
      "learning_rate": 9.88292105263158e-05,
      "loss": 0.6699,
      "step": 24460
    },
    {
      "epoch": 1.3697502519879046,
      "grad_norm": 1.313418984413147,
      "learning_rate": 9.882894736842105e-05,
      "loss": 0.4755,
      "step": 24461
    },
    {
      "epoch": 1.3698062493000336,
      "grad_norm": 1.44545316696167,
      "learning_rate": 9.882868421052633e-05,
      "loss": 0.4548,
      "step": 24462
    },
    {
      "epoch": 1.3698622466121626,
      "grad_norm": 1.3986786603927612,
      "learning_rate": 9.882842105263159e-05,
      "loss": 0.4006,
      "step": 24463
    },
    {
      "epoch": 1.3699182439242916,
      "grad_norm": 1.4873632192611694,
      "learning_rate": 9.882815789473685e-05,
      "loss": 0.5852,
      "step": 24464
    },
    {
      "epoch": 1.3699742412364206,
      "grad_norm": 1.25263512134552,
      "learning_rate": 9.88278947368421e-05,
      "loss": 0.3935,
      "step": 24465
    },
    {
      "epoch": 1.3700302385485497,
      "grad_norm": 1.2605167627334595,
      "learning_rate": 9.882763157894738e-05,
      "loss": 0.5369,
      "step": 24466
    },
    {
      "epoch": 1.3700862358606787,
      "grad_norm": 1.7462189197540283,
      "learning_rate": 9.882736842105264e-05,
      "loss": 0.5562,
      "step": 24467
    },
    {
      "epoch": 1.3701422331728077,
      "grad_norm": 1.29815673828125,
      "learning_rate": 9.88271052631579e-05,
      "loss": 0.4363,
      "step": 24468
    },
    {
      "epoch": 1.3701982304849367,
      "grad_norm": 1.4782339334487915,
      "learning_rate": 9.882684210526316e-05,
      "loss": 0.4121,
      "step": 24469
    },
    {
      "epoch": 1.3702542277970657,
      "grad_norm": 1.279172420501709,
      "learning_rate": 9.882657894736843e-05,
      "loss": 0.4359,
      "step": 24470
    },
    {
      "epoch": 1.3703102251091948,
      "grad_norm": 1.3147929906845093,
      "learning_rate": 9.882631578947369e-05,
      "loss": 0.4926,
      "step": 24471
    },
    {
      "epoch": 1.3703662224213238,
      "grad_norm": 1.593000888824463,
      "learning_rate": 9.882605263157895e-05,
      "loss": 0.43,
      "step": 24472
    },
    {
      "epoch": 1.3704222197334528,
      "grad_norm": 1.2197630405426025,
      "learning_rate": 9.882578947368421e-05,
      "loss": 0.424,
      "step": 24473
    },
    {
      "epoch": 1.3704782170455818,
      "grad_norm": 1.2020893096923828,
      "learning_rate": 9.882552631578947e-05,
      "loss": 0.4941,
      "step": 24474
    },
    {
      "epoch": 1.3705342143577108,
      "grad_norm": 1.323865532875061,
      "learning_rate": 9.882526315789474e-05,
      "loss": 0.6674,
      "step": 24475
    },
    {
      "epoch": 1.3705902116698399,
      "grad_norm": 1.4042918682098389,
      "learning_rate": 9.8825e-05,
      "loss": 0.3982,
      "step": 24476
    },
    {
      "epoch": 1.3706462089819689,
      "grad_norm": 1.2864902019500732,
      "learning_rate": 9.882473684210528e-05,
      "loss": 0.4238,
      "step": 24477
    },
    {
      "epoch": 1.370702206294098,
      "grad_norm": 1.6968425512313843,
      "learning_rate": 9.882447368421052e-05,
      "loss": 0.4565,
      "step": 24478
    },
    {
      "epoch": 1.370758203606227,
      "grad_norm": 1.2619678974151611,
      "learning_rate": 9.88242105263158e-05,
      "loss": 0.4263,
      "step": 24479
    },
    {
      "epoch": 1.370814200918356,
      "grad_norm": 1.5142074823379517,
      "learning_rate": 9.882394736842106e-05,
      "loss": 0.4625,
      "step": 24480
    },
    {
      "epoch": 1.370870198230485,
      "grad_norm": 1.8756927251815796,
      "learning_rate": 9.882368421052633e-05,
      "loss": 0.4626,
      "step": 24481
    },
    {
      "epoch": 1.370926195542614,
      "grad_norm": 1.309545874595642,
      "learning_rate": 9.882342105263158e-05,
      "loss": 0.4355,
      "step": 24482
    },
    {
      "epoch": 1.370982192854743,
      "grad_norm": 1.2559274435043335,
      "learning_rate": 9.882315789473685e-05,
      "loss": 0.4347,
      "step": 24483
    },
    {
      "epoch": 1.371038190166872,
      "grad_norm": 1.3520760536193848,
      "learning_rate": 9.882289473684211e-05,
      "loss": 0.4774,
      "step": 24484
    },
    {
      "epoch": 1.371094187479001,
      "grad_norm": 1.803807258605957,
      "learning_rate": 9.882263157894738e-05,
      "loss": 0.5191,
      "step": 24485
    },
    {
      "epoch": 1.37115018479113,
      "grad_norm": 1.264050006866455,
      "learning_rate": 9.882236842105264e-05,
      "loss": 0.42,
      "step": 24486
    },
    {
      "epoch": 1.371206182103259,
      "grad_norm": 1.4715521335601807,
      "learning_rate": 9.88221052631579e-05,
      "loss": 0.453,
      "step": 24487
    },
    {
      "epoch": 1.371262179415388,
      "grad_norm": 1.6300288438796997,
      "learning_rate": 9.882184210526316e-05,
      "loss": 0.4245,
      "step": 24488
    },
    {
      "epoch": 1.371318176727517,
      "grad_norm": 1.3610649108886719,
      "learning_rate": 9.882157894736842e-05,
      "loss": 0.5405,
      "step": 24489
    },
    {
      "epoch": 1.3713741740396461,
      "grad_norm": 1.3018198013305664,
      "learning_rate": 9.88213157894737e-05,
      "loss": 0.392,
      "step": 24490
    },
    {
      "epoch": 1.3714301713517751,
      "grad_norm": 1.224164366722107,
      "learning_rate": 9.882105263157895e-05,
      "loss": 0.478,
      "step": 24491
    },
    {
      "epoch": 1.3714861686639042,
      "grad_norm": 1.346011757850647,
      "learning_rate": 9.882078947368421e-05,
      "loss": 0.5077,
      "step": 24492
    },
    {
      "epoch": 1.3715421659760332,
      "grad_norm": 1.1635698080062866,
      "learning_rate": 9.882052631578947e-05,
      "loss": 0.4368,
      "step": 24493
    },
    {
      "epoch": 1.3715981632881622,
      "grad_norm": 0.9612644910812378,
      "learning_rate": 9.882026315789475e-05,
      "loss": 0.3163,
      "step": 24494
    },
    {
      "epoch": 1.3716541606002912,
      "grad_norm": 1.6849596500396729,
      "learning_rate": 9.882e-05,
      "loss": 0.4784,
      "step": 24495
    },
    {
      "epoch": 1.3717101579124202,
      "grad_norm": 1.464515209197998,
      "learning_rate": 9.881973684210527e-05,
      "loss": 0.5673,
      "step": 24496
    },
    {
      "epoch": 1.3717661552245493,
      "grad_norm": 1.300506830215454,
      "learning_rate": 9.881947368421053e-05,
      "loss": 0.4394,
      "step": 24497
    },
    {
      "epoch": 1.3718221525366783,
      "grad_norm": 1.3180797100067139,
      "learning_rate": 9.88192105263158e-05,
      "loss": 0.5123,
      "step": 24498
    },
    {
      "epoch": 1.3718781498488073,
      "grad_norm": 1.2333238124847412,
      "learning_rate": 9.881894736842106e-05,
      "loss": 0.4677,
      "step": 24499
    },
    {
      "epoch": 1.3719341471609363,
      "grad_norm": 1.2898578643798828,
      "learning_rate": 9.881868421052632e-05,
      "loss": 0.4185,
      "step": 24500
    },
    {
      "epoch": 1.3719901444730653,
      "grad_norm": 1.4654231071472168,
      "learning_rate": 9.881842105263158e-05,
      "loss": 0.5147,
      "step": 24501
    },
    {
      "epoch": 1.3720461417851944,
      "grad_norm": 1.352738618850708,
      "learning_rate": 9.881815789473685e-05,
      "loss": 0.5083,
      "step": 24502
    },
    {
      "epoch": 1.3721021390973234,
      "grad_norm": 1.1011550426483154,
      "learning_rate": 9.881789473684211e-05,
      "loss": 0.385,
      "step": 24503
    },
    {
      "epoch": 1.3721581364094524,
      "grad_norm": 3.091196060180664,
      "learning_rate": 9.881763157894738e-05,
      "loss": 0.4052,
      "step": 24504
    },
    {
      "epoch": 1.3722141337215814,
      "grad_norm": 1.392540454864502,
      "learning_rate": 9.881736842105263e-05,
      "loss": 0.4687,
      "step": 24505
    },
    {
      "epoch": 1.3722701310337104,
      "grad_norm": 1.606001853942871,
      "learning_rate": 9.881710526315789e-05,
      "loss": 0.4993,
      "step": 24506
    },
    {
      "epoch": 1.3723261283458394,
      "grad_norm": 1.2238637208938599,
      "learning_rate": 9.881684210526316e-05,
      "loss": 0.3676,
      "step": 24507
    },
    {
      "epoch": 1.3723821256579685,
      "grad_norm": 1.2948689460754395,
      "learning_rate": 9.881657894736842e-05,
      "loss": 0.5334,
      "step": 24508
    },
    {
      "epoch": 1.3724381229700975,
      "grad_norm": 1.380072832107544,
      "learning_rate": 9.88163157894737e-05,
      "loss": 0.4704,
      "step": 24509
    },
    {
      "epoch": 1.3724941202822265,
      "grad_norm": 1.105183482170105,
      "learning_rate": 9.881605263157894e-05,
      "loss": 0.4189,
      "step": 24510
    },
    {
      "epoch": 1.3725501175943555,
      "grad_norm": 1.1319661140441895,
      "learning_rate": 9.881578947368422e-05,
      "loss": 0.4072,
      "step": 24511
    },
    {
      "epoch": 1.3726061149064845,
      "grad_norm": 2.206179141998291,
      "learning_rate": 9.881552631578948e-05,
      "loss": 0.6171,
      "step": 24512
    },
    {
      "epoch": 1.3726621122186136,
      "grad_norm": 1.2761162519454956,
      "learning_rate": 9.881526315789475e-05,
      "loss": 0.3865,
      "step": 24513
    },
    {
      "epoch": 1.3727181095307426,
      "grad_norm": 1.4989609718322754,
      "learning_rate": 9.881500000000001e-05,
      "loss": 0.5733,
      "step": 24514
    },
    {
      "epoch": 1.3727741068428716,
      "grad_norm": 1.320183277130127,
      "learning_rate": 9.881473684210527e-05,
      "loss": 0.475,
      "step": 24515
    },
    {
      "epoch": 1.3728301041550006,
      "grad_norm": 1.1859266757965088,
      "learning_rate": 9.881447368421053e-05,
      "loss": 0.4993,
      "step": 24516
    },
    {
      "epoch": 1.3728861014671296,
      "grad_norm": 1.1573539972305298,
      "learning_rate": 9.88142105263158e-05,
      "loss": 0.3842,
      "step": 24517
    },
    {
      "epoch": 1.3729420987792587,
      "grad_norm": 1.4029028415679932,
      "learning_rate": 9.881394736842106e-05,
      "loss": 0.5271,
      "step": 24518
    },
    {
      "epoch": 1.3729980960913877,
      "grad_norm": 1.2326328754425049,
      "learning_rate": 9.881368421052632e-05,
      "loss": 0.4335,
      "step": 24519
    },
    {
      "epoch": 1.3730540934035167,
      "grad_norm": 3.316587209701538,
      "learning_rate": 9.881342105263158e-05,
      "loss": 0.5095,
      "step": 24520
    },
    {
      "epoch": 1.3731100907156457,
      "grad_norm": 1.5112648010253906,
      "learning_rate": 9.881315789473685e-05,
      "loss": 0.4672,
      "step": 24521
    },
    {
      "epoch": 1.3731660880277747,
      "grad_norm": 1.3780750036239624,
      "learning_rate": 9.881289473684211e-05,
      "loss": 0.446,
      "step": 24522
    },
    {
      "epoch": 1.3732220853399038,
      "grad_norm": 1.4295034408569336,
      "learning_rate": 9.881263157894737e-05,
      "loss": 0.3854,
      "step": 24523
    },
    {
      "epoch": 1.3732780826520328,
      "grad_norm": 1.6485295295715332,
      "learning_rate": 9.881236842105263e-05,
      "loss": 0.5406,
      "step": 24524
    },
    {
      "epoch": 1.3733340799641618,
      "grad_norm": 1.5437777042388916,
      "learning_rate": 9.881210526315789e-05,
      "loss": 0.6657,
      "step": 24525
    },
    {
      "epoch": 1.3733900772762908,
      "grad_norm": 1.3259806632995605,
      "learning_rate": 9.881184210526317e-05,
      "loss": 0.4393,
      "step": 24526
    },
    {
      "epoch": 1.3734460745884198,
      "grad_norm": 1.273875117301941,
      "learning_rate": 9.881157894736843e-05,
      "loss": 0.5589,
      "step": 24527
    },
    {
      "epoch": 1.3735020719005488,
      "grad_norm": 5.459080696105957,
      "learning_rate": 9.881131578947369e-05,
      "loss": 0.3946,
      "step": 24528
    },
    {
      "epoch": 1.3735580692126779,
      "grad_norm": 1.2246806621551514,
      "learning_rate": 9.881105263157895e-05,
      "loss": 0.4394,
      "step": 24529
    },
    {
      "epoch": 1.3736140665248069,
      "grad_norm": 1.2419131994247437,
      "learning_rate": 9.881078947368422e-05,
      "loss": 0.5603,
      "step": 24530
    },
    {
      "epoch": 1.373670063836936,
      "grad_norm": 1.3171749114990234,
      "learning_rate": 9.881052631578948e-05,
      "loss": 0.4098,
      "step": 24531
    },
    {
      "epoch": 1.373726061149065,
      "grad_norm": 1.3591086864471436,
      "learning_rate": 9.881026315789475e-05,
      "loss": 0.3953,
      "step": 24532
    },
    {
      "epoch": 1.373782058461194,
      "grad_norm": 1.4800583124160767,
      "learning_rate": 9.881e-05,
      "loss": 0.3976,
      "step": 24533
    },
    {
      "epoch": 1.373838055773323,
      "grad_norm": 1.4704041481018066,
      "learning_rate": 9.880973684210527e-05,
      "loss": 0.4332,
      "step": 24534
    },
    {
      "epoch": 1.373894053085452,
      "grad_norm": 1.4346450567245483,
      "learning_rate": 9.880947368421053e-05,
      "loss": 0.6352,
      "step": 24535
    },
    {
      "epoch": 1.373950050397581,
      "grad_norm": 1.7150028944015503,
      "learning_rate": 9.88092105263158e-05,
      "loss": 0.5394,
      "step": 24536
    },
    {
      "epoch": 1.37400604770971,
      "grad_norm": 1.4494582414627075,
      "learning_rate": 9.880894736842105e-05,
      "loss": 0.4575,
      "step": 24537
    },
    {
      "epoch": 1.374062045021839,
      "grad_norm": 1.6348971128463745,
      "learning_rate": 9.880868421052632e-05,
      "loss": 0.5308,
      "step": 24538
    },
    {
      "epoch": 1.374118042333968,
      "grad_norm": 1.4942302703857422,
      "learning_rate": 9.880842105263158e-05,
      "loss": 0.4096,
      "step": 24539
    },
    {
      "epoch": 1.374174039646097,
      "grad_norm": 1.2634156942367554,
      "learning_rate": 9.880815789473684e-05,
      "loss": 0.465,
      "step": 24540
    },
    {
      "epoch": 1.374230036958226,
      "grad_norm": 54.6109504699707,
      "learning_rate": 9.880789473684212e-05,
      "loss": 0.4818,
      "step": 24541
    },
    {
      "epoch": 1.3742860342703551,
      "grad_norm": 1.678976058959961,
      "learning_rate": 9.880763157894736e-05,
      "loss": 0.4527,
      "step": 24542
    },
    {
      "epoch": 1.3743420315824841,
      "grad_norm": 1.4438743591308594,
      "learning_rate": 9.880736842105264e-05,
      "loss": 0.4749,
      "step": 24543
    },
    {
      "epoch": 1.3743980288946132,
      "grad_norm": 1.649899959564209,
      "learning_rate": 9.88071052631579e-05,
      "loss": 0.3697,
      "step": 24544
    },
    {
      "epoch": 1.3744540262067422,
      "grad_norm": 1.311410903930664,
      "learning_rate": 9.880684210526317e-05,
      "loss": 0.3974,
      "step": 24545
    },
    {
      "epoch": 1.3745100235188712,
      "grad_norm": 1.3957568407058716,
      "learning_rate": 9.880657894736843e-05,
      "loss": 0.5027,
      "step": 24546
    },
    {
      "epoch": 1.3745660208310002,
      "grad_norm": 3.369692802429199,
      "learning_rate": 9.880631578947369e-05,
      "loss": 0.5027,
      "step": 24547
    },
    {
      "epoch": 1.3746220181431292,
      "grad_norm": 1.3832881450653076,
      "learning_rate": 9.880605263157895e-05,
      "loss": 0.4475,
      "step": 24548
    },
    {
      "epoch": 1.3746780154552583,
      "grad_norm": 1.0688807964324951,
      "learning_rate": 9.880578947368422e-05,
      "loss": 0.3709,
      "step": 24549
    },
    {
      "epoch": 1.3747340127673873,
      "grad_norm": 1.4251041412353516,
      "learning_rate": 9.880552631578948e-05,
      "loss": 0.5787,
      "step": 24550
    },
    {
      "epoch": 1.3747900100795163,
      "grad_norm": 3.9179627895355225,
      "learning_rate": 9.880526315789474e-05,
      "loss": 0.5285,
      "step": 24551
    },
    {
      "epoch": 1.3748460073916453,
      "grad_norm": 1.3410621881484985,
      "learning_rate": 9.8805e-05,
      "loss": 0.3814,
      "step": 24552
    },
    {
      "epoch": 1.3749020047037743,
      "grad_norm": 1.1759706735610962,
      "learning_rate": 9.880473684210527e-05,
      "loss": 0.4728,
      "step": 24553
    },
    {
      "epoch": 1.3749580020159033,
      "grad_norm": 1.222511649131775,
      "learning_rate": 9.880447368421053e-05,
      "loss": 0.3362,
      "step": 24554
    },
    {
      "epoch": 1.3750139993280324,
      "grad_norm": 1.8133656978607178,
      "learning_rate": 9.880421052631579e-05,
      "loss": 0.6572,
      "step": 24555
    },
    {
      "epoch": 1.3750699966401614,
      "grad_norm": 1.2357032299041748,
      "learning_rate": 9.880394736842105e-05,
      "loss": 0.4965,
      "step": 24556
    },
    {
      "epoch": 1.3751259939522904,
      "grad_norm": 1.3727672100067139,
      "learning_rate": 9.880368421052633e-05,
      "loss": 0.3667,
      "step": 24557
    },
    {
      "epoch": 1.3751819912644194,
      "grad_norm": 1.2672778367996216,
      "learning_rate": 9.880342105263159e-05,
      "loss": 0.3819,
      "step": 24558
    },
    {
      "epoch": 1.3752379885765484,
      "grad_norm": 1.537623643875122,
      "learning_rate": 9.880315789473685e-05,
      "loss": 0.4472,
      "step": 24559
    },
    {
      "epoch": 1.3752939858886775,
      "grad_norm": 1.3242238759994507,
      "learning_rate": 9.88028947368421e-05,
      "loss": 0.3838,
      "step": 24560
    },
    {
      "epoch": 1.3753499832008065,
      "grad_norm": 1.2333530187606812,
      "learning_rate": 9.880263157894736e-05,
      "loss": 0.4078,
      "step": 24561
    },
    {
      "epoch": 1.3754059805129355,
      "grad_norm": 1.4914171695709229,
      "learning_rate": 9.880236842105264e-05,
      "loss": 0.4885,
      "step": 24562
    },
    {
      "epoch": 1.3754619778250645,
      "grad_norm": 1.398218035697937,
      "learning_rate": 9.88021052631579e-05,
      "loss": 0.404,
      "step": 24563
    },
    {
      "epoch": 1.3755179751371935,
      "grad_norm": 1.4480798244476318,
      "learning_rate": 9.880184210526317e-05,
      "loss": 0.4943,
      "step": 24564
    },
    {
      "epoch": 1.3755739724493226,
      "grad_norm": 1.5293318033218384,
      "learning_rate": 9.880157894736842e-05,
      "loss": 0.5123,
      "step": 24565
    },
    {
      "epoch": 1.3756299697614516,
      "grad_norm": 2.8959763050079346,
      "learning_rate": 9.880131578947369e-05,
      "loss": 0.5938,
      "step": 24566
    },
    {
      "epoch": 1.3756859670735806,
      "grad_norm": 1.165149211883545,
      "learning_rate": 9.880105263157895e-05,
      "loss": 0.43,
      "step": 24567
    },
    {
      "epoch": 1.3757419643857096,
      "grad_norm": 1.2796895503997803,
      "learning_rate": 9.880078947368422e-05,
      "loss": 0.3938,
      "step": 24568
    },
    {
      "epoch": 1.3757979616978386,
      "grad_norm": 1.3704499006271362,
      "learning_rate": 9.880052631578948e-05,
      "loss": 0.4379,
      "step": 24569
    },
    {
      "epoch": 1.3758539590099677,
      "grad_norm": 1.426423192024231,
      "learning_rate": 9.880026315789474e-05,
      "loss": 0.4486,
      "step": 24570
    },
    {
      "epoch": 1.3759099563220967,
      "grad_norm": 1.323846459388733,
      "learning_rate": 9.88e-05,
      "loss": 0.3953,
      "step": 24571
    },
    {
      "epoch": 1.3759659536342257,
      "grad_norm": 1.355147361755371,
      "learning_rate": 9.879973684210528e-05,
      "loss": 0.4667,
      "step": 24572
    },
    {
      "epoch": 1.3760219509463547,
      "grad_norm": 1.044049620628357,
      "learning_rate": 9.879947368421054e-05,
      "loss": 0.374,
      "step": 24573
    },
    {
      "epoch": 1.3760779482584837,
      "grad_norm": 1.219199538230896,
      "learning_rate": 9.87992105263158e-05,
      "loss": 0.4622,
      "step": 24574
    },
    {
      "epoch": 1.3761339455706127,
      "grad_norm": 1.918542504310608,
      "learning_rate": 9.879894736842106e-05,
      "loss": 0.4706,
      "step": 24575
    },
    {
      "epoch": 1.3761899428827415,
      "grad_norm": 1.292521595954895,
      "learning_rate": 9.879868421052631e-05,
      "loss": 0.3774,
      "step": 24576
    },
    {
      "epoch": 1.3762459401948706,
      "grad_norm": 1.4552044868469238,
      "learning_rate": 9.879842105263159e-05,
      "loss": 0.5177,
      "step": 24577
    },
    {
      "epoch": 1.3763019375069996,
      "grad_norm": 1.1717822551727295,
      "learning_rate": 9.879815789473685e-05,
      "loss": 0.5617,
      "step": 24578
    },
    {
      "epoch": 1.3763579348191286,
      "grad_norm": 1.4983909130096436,
      "learning_rate": 9.879789473684211e-05,
      "loss": 0.4669,
      "step": 24579
    },
    {
      "epoch": 1.3764139321312576,
      "grad_norm": 1.1356526613235474,
      "learning_rate": 9.879763157894737e-05,
      "loss": 0.4193,
      "step": 24580
    },
    {
      "epoch": 1.3764699294433866,
      "grad_norm": 1.1955333948135376,
      "learning_rate": 9.879736842105264e-05,
      "loss": 0.3941,
      "step": 24581
    },
    {
      "epoch": 1.3765259267555157,
      "grad_norm": 1.644252896308899,
      "learning_rate": 9.87971052631579e-05,
      "loss": 0.5471,
      "step": 24582
    },
    {
      "epoch": 1.3765819240676447,
      "grad_norm": 1.354682445526123,
      "learning_rate": 9.879684210526316e-05,
      "loss": 0.4126,
      "step": 24583
    },
    {
      "epoch": 1.3766379213797737,
      "grad_norm": 1.320406436920166,
      "learning_rate": 9.879657894736842e-05,
      "loss": 0.4518,
      "step": 24584
    },
    {
      "epoch": 1.3766939186919027,
      "grad_norm": 1.5978227853775024,
      "learning_rate": 9.879631578947369e-05,
      "loss": 0.5316,
      "step": 24585
    },
    {
      "epoch": 1.3767499160040317,
      "grad_norm": 1.4413739442825317,
      "learning_rate": 9.879605263157895e-05,
      "loss": 0.5114,
      "step": 24586
    },
    {
      "epoch": 1.3768059133161608,
      "grad_norm": 1.6403529644012451,
      "learning_rate": 9.879578947368421e-05,
      "loss": 0.5069,
      "step": 24587
    },
    {
      "epoch": 1.3768619106282898,
      "grad_norm": 2.017141103744507,
      "learning_rate": 9.879552631578947e-05,
      "loss": 0.4421,
      "step": 24588
    },
    {
      "epoch": 1.3769179079404188,
      "grad_norm": 1.1616919040679932,
      "learning_rate": 9.879526315789475e-05,
      "loss": 0.3874,
      "step": 24589
    },
    {
      "epoch": 1.3769739052525478,
      "grad_norm": 1.409917950630188,
      "learning_rate": 9.8795e-05,
      "loss": 0.4899,
      "step": 24590
    },
    {
      "epoch": 1.3770299025646768,
      "grad_norm": 1.1836915016174316,
      "learning_rate": 9.879473684210528e-05,
      "loss": 0.4048,
      "step": 24591
    },
    {
      "epoch": 1.3770858998768059,
      "grad_norm": 1.5157136917114258,
      "learning_rate": 9.879447368421052e-05,
      "loss": 0.6242,
      "step": 24592
    },
    {
      "epoch": 1.3771418971889349,
      "grad_norm": 1.304404616355896,
      "learning_rate": 9.879421052631578e-05,
      "loss": 0.4373,
      "step": 24593
    },
    {
      "epoch": 1.377197894501064,
      "grad_norm": 1.431369423866272,
      "learning_rate": 9.879394736842106e-05,
      "loss": 0.4589,
      "step": 24594
    },
    {
      "epoch": 1.377253891813193,
      "grad_norm": 1.170019507408142,
      "learning_rate": 9.879368421052632e-05,
      "loss": 0.456,
      "step": 24595
    },
    {
      "epoch": 1.377309889125322,
      "grad_norm": 1.325610876083374,
      "learning_rate": 9.879342105263159e-05,
      "loss": 0.5171,
      "step": 24596
    },
    {
      "epoch": 1.377365886437451,
      "grad_norm": 1.6425721645355225,
      "learning_rate": 9.879315789473684e-05,
      "loss": 0.5185,
      "step": 24597
    },
    {
      "epoch": 1.37742188374958,
      "grad_norm": 1.7873594760894775,
      "learning_rate": 9.879289473684211e-05,
      "loss": 0.5937,
      "step": 24598
    },
    {
      "epoch": 1.377477881061709,
      "grad_norm": 1.5284775495529175,
      "learning_rate": 9.879263157894737e-05,
      "loss": 0.411,
      "step": 24599
    },
    {
      "epoch": 1.377533878373838,
      "grad_norm": 1.3774112462997437,
      "learning_rate": 9.879236842105264e-05,
      "loss": 0.4618,
      "step": 24600
    },
    {
      "epoch": 1.377589875685967,
      "grad_norm": 1.327060580253601,
      "learning_rate": 9.87921052631579e-05,
      "loss": 0.443,
      "step": 24601
    },
    {
      "epoch": 1.377645872998096,
      "grad_norm": 1.7170281410217285,
      "learning_rate": 9.879184210526316e-05,
      "loss": 0.5056,
      "step": 24602
    },
    {
      "epoch": 1.377701870310225,
      "grad_norm": 1.513201355934143,
      "learning_rate": 9.879157894736842e-05,
      "loss": 0.5068,
      "step": 24603
    },
    {
      "epoch": 1.377757867622354,
      "grad_norm": 1.3682736158370972,
      "learning_rate": 9.87913157894737e-05,
      "loss": 0.566,
      "step": 24604
    },
    {
      "epoch": 1.377813864934483,
      "grad_norm": 1.3634577989578247,
      "learning_rate": 9.879105263157896e-05,
      "loss": 0.4635,
      "step": 24605
    },
    {
      "epoch": 1.3778698622466121,
      "grad_norm": 1.2694344520568848,
      "learning_rate": 9.879078947368422e-05,
      "loss": 0.5066,
      "step": 24606
    },
    {
      "epoch": 1.3779258595587411,
      "grad_norm": 1.6849581003189087,
      "learning_rate": 9.879052631578947e-05,
      "loss": 0.5615,
      "step": 24607
    },
    {
      "epoch": 1.3779818568708702,
      "grad_norm": 1.1305550336837769,
      "learning_rate": 9.879026315789475e-05,
      "loss": 0.5246,
      "step": 24608
    },
    {
      "epoch": 1.3780378541829992,
      "grad_norm": 1.4769785404205322,
      "learning_rate": 9.879000000000001e-05,
      "loss": 0.4707,
      "step": 24609
    },
    {
      "epoch": 1.3780938514951282,
      "grad_norm": 1.4925611019134521,
      "learning_rate": 9.878973684210527e-05,
      "loss": 0.8428,
      "step": 24610
    },
    {
      "epoch": 1.3781498488072572,
      "grad_norm": 1.2915689945220947,
      "learning_rate": 9.878947368421053e-05,
      "loss": 0.4596,
      "step": 24611
    },
    {
      "epoch": 1.3782058461193862,
      "grad_norm": 1.2616046667099,
      "learning_rate": 9.878921052631579e-05,
      "loss": 0.4247,
      "step": 24612
    },
    {
      "epoch": 1.3782618434315153,
      "grad_norm": 1.432315707206726,
      "learning_rate": 9.878894736842106e-05,
      "loss": 0.4922,
      "step": 24613
    },
    {
      "epoch": 1.3783178407436443,
      "grad_norm": 1.3737187385559082,
      "learning_rate": 9.878868421052632e-05,
      "loss": 0.5139,
      "step": 24614
    },
    {
      "epoch": 1.3783738380557733,
      "grad_norm": 1.4290072917938232,
      "learning_rate": 9.878842105263158e-05,
      "loss": 0.3874,
      "step": 24615
    },
    {
      "epoch": 1.3784298353679023,
      "grad_norm": 1.3011133670806885,
      "learning_rate": 9.878815789473684e-05,
      "loss": 0.4629,
      "step": 24616
    },
    {
      "epoch": 1.3784858326800313,
      "grad_norm": 1.2998813390731812,
      "learning_rate": 9.878789473684211e-05,
      "loss": 0.4393,
      "step": 24617
    },
    {
      "epoch": 1.3785418299921604,
      "grad_norm": 1.4572488069534302,
      "learning_rate": 9.878763157894737e-05,
      "loss": 0.4486,
      "step": 24618
    },
    {
      "epoch": 1.3785978273042894,
      "grad_norm": 1.1944403648376465,
      "learning_rate": 9.878736842105265e-05,
      "loss": 0.3721,
      "step": 24619
    },
    {
      "epoch": 1.3786538246164184,
      "grad_norm": 1.4873815774917603,
      "learning_rate": 9.878710526315789e-05,
      "loss": 0.4823,
      "step": 24620
    },
    {
      "epoch": 1.3787098219285474,
      "grad_norm": 1.6921597719192505,
      "learning_rate": 9.878684210526317e-05,
      "loss": 0.5813,
      "step": 24621
    },
    {
      "epoch": 1.3787658192406764,
      "grad_norm": 1.540161371231079,
      "learning_rate": 9.878657894736843e-05,
      "loss": 0.5494,
      "step": 24622
    },
    {
      "epoch": 1.3788218165528054,
      "grad_norm": 1.5844582319259644,
      "learning_rate": 9.87863157894737e-05,
      "loss": 0.5868,
      "step": 24623
    },
    {
      "epoch": 1.3788778138649345,
      "grad_norm": 1.5227375030517578,
      "learning_rate": 9.878605263157896e-05,
      "loss": 0.4472,
      "step": 24624
    },
    {
      "epoch": 1.3789338111770635,
      "grad_norm": 1.295911431312561,
      "learning_rate": 9.878578947368422e-05,
      "loss": 0.434,
      "step": 24625
    },
    {
      "epoch": 1.3789898084891925,
      "grad_norm": 1.7489311695098877,
      "learning_rate": 9.878552631578948e-05,
      "loss": 0.4214,
      "step": 24626
    },
    {
      "epoch": 1.3790458058013215,
      "grad_norm": 1.0726250410079956,
      "learning_rate": 9.878526315789474e-05,
      "loss": 0.3382,
      "step": 24627
    },
    {
      "epoch": 1.3791018031134505,
      "grad_norm": 1.4631832838058472,
      "learning_rate": 9.878500000000001e-05,
      "loss": 0.4638,
      "step": 24628
    },
    {
      "epoch": 1.3791578004255796,
      "grad_norm": 1.142659068107605,
      "learning_rate": 9.878473684210526e-05,
      "loss": 0.351,
      "step": 24629
    },
    {
      "epoch": 1.3792137977377086,
      "grad_norm": 1.292729377746582,
      "learning_rate": 9.878447368421053e-05,
      "loss": 0.3391,
      "step": 24630
    },
    {
      "epoch": 1.3792697950498376,
      "grad_norm": 1.1582690477371216,
      "learning_rate": 9.878421052631579e-05,
      "loss": 0.4104,
      "step": 24631
    },
    {
      "epoch": 1.3793257923619666,
      "grad_norm": 1.4168444871902466,
      "learning_rate": 9.878394736842106e-05,
      "loss": 0.3806,
      "step": 24632
    },
    {
      "epoch": 1.3793817896740956,
      "grad_norm": 1.4440557956695557,
      "learning_rate": 9.878368421052632e-05,
      "loss": 0.5473,
      "step": 24633
    },
    {
      "epoch": 1.3794377869862247,
      "grad_norm": 1.5434112548828125,
      "learning_rate": 9.878342105263158e-05,
      "loss": 0.4289,
      "step": 24634
    },
    {
      "epoch": 1.3794937842983537,
      "grad_norm": 1.4226053953170776,
      "learning_rate": 9.878315789473684e-05,
      "loss": 0.4301,
      "step": 24635
    },
    {
      "epoch": 1.3795497816104827,
      "grad_norm": 1.3752599954605103,
      "learning_rate": 9.878289473684212e-05,
      "loss": 0.583,
      "step": 24636
    },
    {
      "epoch": 1.3796057789226117,
      "grad_norm": 1.5356531143188477,
      "learning_rate": 9.878263157894738e-05,
      "loss": 0.4519,
      "step": 24637
    },
    {
      "epoch": 1.3796617762347407,
      "grad_norm": 1.1383914947509766,
      "learning_rate": 9.878236842105263e-05,
      "loss": 0.4648,
      "step": 24638
    },
    {
      "epoch": 1.3797177735468698,
      "grad_norm": 1.3420947790145874,
      "learning_rate": 9.87821052631579e-05,
      "loss": 0.4335,
      "step": 24639
    },
    {
      "epoch": 1.3797737708589988,
      "grad_norm": 1.1258677244186401,
      "learning_rate": 9.878184210526317e-05,
      "loss": 0.4219,
      "step": 24640
    },
    {
      "epoch": 1.3798297681711278,
      "grad_norm": 1.1578919887542725,
      "learning_rate": 9.878157894736843e-05,
      "loss": 0.4617,
      "step": 24641
    },
    {
      "epoch": 1.3798857654832568,
      "grad_norm": 1.464247226715088,
      "learning_rate": 9.878131578947369e-05,
      "loss": 0.5192,
      "step": 24642
    },
    {
      "epoch": 1.3799417627953858,
      "grad_norm": 1.4965718984603882,
      "learning_rate": 9.878105263157895e-05,
      "loss": 0.4869,
      "step": 24643
    },
    {
      "epoch": 1.3799977601075148,
      "grad_norm": 2.279975652694702,
      "learning_rate": 9.878078947368421e-05,
      "loss": 0.4597,
      "step": 24644
    },
    {
      "epoch": 1.3800537574196439,
      "grad_norm": 1.3425018787384033,
      "learning_rate": 9.878052631578948e-05,
      "loss": 0.3994,
      "step": 24645
    },
    {
      "epoch": 1.3801097547317729,
      "grad_norm": 1.2083648443222046,
      "learning_rate": 9.878026315789474e-05,
      "loss": 0.3594,
      "step": 24646
    },
    {
      "epoch": 1.380165752043902,
      "grad_norm": 1.3597885370254517,
      "learning_rate": 9.878e-05,
      "loss": 0.4676,
      "step": 24647
    },
    {
      "epoch": 1.380221749356031,
      "grad_norm": 1.4435182809829712,
      "learning_rate": 9.877973684210526e-05,
      "loss": 0.4462,
      "step": 24648
    },
    {
      "epoch": 1.38027774666816,
      "grad_norm": 1.4416115283966064,
      "learning_rate": 9.877947368421053e-05,
      "loss": 0.5033,
      "step": 24649
    },
    {
      "epoch": 1.380333743980289,
      "grad_norm": 1.3227752447128296,
      "learning_rate": 9.877921052631579e-05,
      "loss": 0.4932,
      "step": 24650
    },
    {
      "epoch": 1.380389741292418,
      "grad_norm": 1.5899603366851807,
      "learning_rate": 9.877894736842107e-05,
      "loss": 0.6036,
      "step": 24651
    },
    {
      "epoch": 1.380445738604547,
      "grad_norm": 1.2183899879455566,
      "learning_rate": 9.877868421052631e-05,
      "loss": 0.4057,
      "step": 24652
    },
    {
      "epoch": 1.380501735916676,
      "grad_norm": 1.0418287515640259,
      "learning_rate": 9.877842105263159e-05,
      "loss": 0.3262,
      "step": 24653
    },
    {
      "epoch": 1.380557733228805,
      "grad_norm": 1.2916969060897827,
      "learning_rate": 9.877815789473684e-05,
      "loss": 0.4956,
      "step": 24654
    },
    {
      "epoch": 1.380613730540934,
      "grad_norm": 1.5164707899093628,
      "learning_rate": 9.877789473684212e-05,
      "loss": 0.46,
      "step": 24655
    },
    {
      "epoch": 1.380669727853063,
      "grad_norm": 1.257965326309204,
      "learning_rate": 9.877763157894738e-05,
      "loss": 0.5249,
      "step": 24656
    },
    {
      "epoch": 1.380725725165192,
      "grad_norm": 1.462067723274231,
      "learning_rate": 9.877736842105264e-05,
      "loss": 0.4827,
      "step": 24657
    },
    {
      "epoch": 1.3807817224773211,
      "grad_norm": 1.2253189086914062,
      "learning_rate": 9.87771052631579e-05,
      "loss": 0.379,
      "step": 24658
    },
    {
      "epoch": 1.3808377197894501,
      "grad_norm": 1.1978322267532349,
      "learning_rate": 9.877684210526317e-05,
      "loss": 0.3814,
      "step": 24659
    },
    {
      "epoch": 1.3808937171015792,
      "grad_norm": 1.9540668725967407,
      "learning_rate": 9.877657894736843e-05,
      "loss": 0.4833,
      "step": 24660
    },
    {
      "epoch": 1.3809497144137082,
      "grad_norm": 1.3607112169265747,
      "learning_rate": 9.877631578947369e-05,
      "loss": 0.5521,
      "step": 24661
    },
    {
      "epoch": 1.3810057117258372,
      "grad_norm": 1.5534237623214722,
      "learning_rate": 9.877605263157895e-05,
      "loss": 0.385,
      "step": 24662
    },
    {
      "epoch": 1.3810617090379662,
      "grad_norm": 1.330488681793213,
      "learning_rate": 9.877578947368421e-05,
      "loss": 0.5468,
      "step": 24663
    },
    {
      "epoch": 1.3811177063500952,
      "grad_norm": 1.4170063734054565,
      "learning_rate": 9.877552631578948e-05,
      "loss": 0.5088,
      "step": 24664
    },
    {
      "epoch": 1.3811737036622243,
      "grad_norm": 1.3981940746307373,
      "learning_rate": 9.877526315789474e-05,
      "loss": 0.5177,
      "step": 24665
    },
    {
      "epoch": 1.3812297009743533,
      "grad_norm": 1.315423846244812,
      "learning_rate": 9.8775e-05,
      "loss": 0.4008,
      "step": 24666
    },
    {
      "epoch": 1.3812856982864823,
      "grad_norm": 1.325331687927246,
      "learning_rate": 9.877473684210526e-05,
      "loss": 0.6356,
      "step": 24667
    },
    {
      "epoch": 1.3813416955986113,
      "grad_norm": 1.5587759017944336,
      "learning_rate": 9.877447368421054e-05,
      "loss": 0.5645,
      "step": 24668
    },
    {
      "epoch": 1.3813976929107403,
      "grad_norm": 1.4653170108795166,
      "learning_rate": 9.87742105263158e-05,
      "loss": 0.5262,
      "step": 24669
    },
    {
      "epoch": 1.3814536902228693,
      "grad_norm": 1.3540924787521362,
      "learning_rate": 9.877394736842105e-05,
      "loss": 0.3624,
      "step": 24670
    },
    {
      "epoch": 1.3815096875349984,
      "grad_norm": 1.3128352165222168,
      "learning_rate": 9.877368421052631e-05,
      "loss": 0.4842,
      "step": 24671
    },
    {
      "epoch": 1.3815656848471274,
      "grad_norm": 1.0599039793014526,
      "learning_rate": 9.877342105263159e-05,
      "loss": 0.3384,
      "step": 24672
    },
    {
      "epoch": 1.3816216821592564,
      "grad_norm": 1.2542729377746582,
      "learning_rate": 9.877315789473685e-05,
      "loss": 0.5512,
      "step": 24673
    },
    {
      "epoch": 1.3816776794713854,
      "grad_norm": 1.223557710647583,
      "learning_rate": 9.877289473684212e-05,
      "loss": 0.4321,
      "step": 24674
    },
    {
      "epoch": 1.3817336767835144,
      "grad_norm": 1.2568310499191284,
      "learning_rate": 9.877263157894737e-05,
      "loss": 0.5891,
      "step": 24675
    },
    {
      "epoch": 1.3817896740956435,
      "grad_norm": 1.4729034900665283,
      "learning_rate": 9.877236842105264e-05,
      "loss": 0.4209,
      "step": 24676
    },
    {
      "epoch": 1.3818456714077725,
      "grad_norm": 1.715546727180481,
      "learning_rate": 9.87721052631579e-05,
      "loss": 0.5552,
      "step": 24677
    },
    {
      "epoch": 1.3819016687199015,
      "grad_norm": 1.522024154663086,
      "learning_rate": 9.877184210526317e-05,
      "loss": 0.4391,
      "step": 24678
    },
    {
      "epoch": 1.3819576660320305,
      "grad_norm": 1.5103111267089844,
      "learning_rate": 9.877157894736843e-05,
      "loss": 0.6572,
      "step": 24679
    },
    {
      "epoch": 1.3820136633441595,
      "grad_norm": 1.2847040891647339,
      "learning_rate": 9.877131578947368e-05,
      "loss": 0.3829,
      "step": 24680
    },
    {
      "epoch": 1.3820696606562886,
      "grad_norm": 1.3065253496170044,
      "learning_rate": 9.877105263157895e-05,
      "loss": 0.364,
      "step": 24681
    },
    {
      "epoch": 1.3821256579684176,
      "grad_norm": 1.60619056224823,
      "learning_rate": 9.877078947368421e-05,
      "loss": 0.4733,
      "step": 24682
    },
    {
      "epoch": 1.3821816552805466,
      "grad_norm": 1.3917906284332275,
      "learning_rate": 9.877052631578949e-05,
      "loss": 0.5068,
      "step": 24683
    },
    {
      "epoch": 1.3822376525926756,
      "grad_norm": 1.093767523765564,
      "learning_rate": 9.877026315789473e-05,
      "loss": 0.3872,
      "step": 24684
    },
    {
      "epoch": 1.3822936499048046,
      "grad_norm": 1.4227591753005981,
      "learning_rate": 9.877e-05,
      "loss": 0.565,
      "step": 24685
    },
    {
      "epoch": 1.3823496472169337,
      "grad_norm": 1.251604437828064,
      "learning_rate": 9.876973684210526e-05,
      "loss": 0.5833,
      "step": 24686
    },
    {
      "epoch": 1.3824056445290627,
      "grad_norm": 1.1730982065200806,
      "learning_rate": 9.876947368421054e-05,
      "loss": 0.4711,
      "step": 24687
    },
    {
      "epoch": 1.3824616418411917,
      "grad_norm": 1.510393500328064,
      "learning_rate": 9.87692105263158e-05,
      "loss": 0.486,
      "step": 24688
    },
    {
      "epoch": 1.3825176391533207,
      "grad_norm": 1.3401352167129517,
      "learning_rate": 9.876894736842106e-05,
      "loss": 0.4406,
      "step": 24689
    },
    {
      "epoch": 1.3825736364654497,
      "grad_norm": 1.283668041229248,
      "learning_rate": 9.876868421052632e-05,
      "loss": 0.5077,
      "step": 24690
    },
    {
      "epoch": 1.3826296337775787,
      "grad_norm": 1.4496607780456543,
      "learning_rate": 9.876842105263159e-05,
      "loss": 0.5748,
      "step": 24691
    },
    {
      "epoch": 1.3826856310897078,
      "grad_norm": 1.473398208618164,
      "learning_rate": 9.876815789473685e-05,
      "loss": 0.4832,
      "step": 24692
    },
    {
      "epoch": 1.3827416284018368,
      "grad_norm": 1.3518096208572388,
      "learning_rate": 9.876789473684211e-05,
      "loss": 0.3901,
      "step": 24693
    },
    {
      "epoch": 1.3827976257139658,
      "grad_norm": 1.4092508554458618,
      "learning_rate": 9.876763157894737e-05,
      "loss": 0.4113,
      "step": 24694
    },
    {
      "epoch": 1.3828536230260948,
      "grad_norm": 1.2037180662155151,
      "learning_rate": 9.876736842105264e-05,
      "loss": 0.435,
      "step": 24695
    },
    {
      "epoch": 1.3829096203382238,
      "grad_norm": 1.2900842428207397,
      "learning_rate": 9.87671052631579e-05,
      "loss": 0.4582,
      "step": 24696
    },
    {
      "epoch": 1.3829656176503529,
      "grad_norm": 1.3820942640304565,
      "learning_rate": 9.876684210526316e-05,
      "loss": 0.5405,
      "step": 24697
    },
    {
      "epoch": 1.3830216149624819,
      "grad_norm": 1.4084657430648804,
      "learning_rate": 9.876657894736842e-05,
      "loss": 0.5001,
      "step": 24698
    },
    {
      "epoch": 1.383077612274611,
      "grad_norm": NaN,
      "learning_rate": 9.876657894736842e-05,
      "loss": 0.5658,
      "step": 24699
    },
    {
      "epoch": 1.3831336095867397,
      "grad_norm": 1.2814196348190308,
      "learning_rate": 9.876631578947368e-05,
      "loss": 0.4785,
      "step": 24700
    },
    {
      "epoch": 1.3831896068988687,
      "grad_norm": 1.270787239074707,
      "learning_rate": 9.876605263157895e-05,
      "loss": 0.4129,
      "step": 24701
    },
    {
      "epoch": 1.3832456042109977,
      "grad_norm": 1.2655234336853027,
      "learning_rate": 9.876578947368421e-05,
      "loss": 0.5402,
      "step": 24702
    },
    {
      "epoch": 1.3833016015231268,
      "grad_norm": 1.4280146360397339,
      "learning_rate": 9.876552631578947e-05,
      "loss": 0.5095,
      "step": 24703
    },
    {
      "epoch": 1.3833575988352558,
      "grad_norm": 1.4920822381973267,
      "learning_rate": 9.876526315789473e-05,
      "loss": 0.6155,
      "step": 24704
    },
    {
      "epoch": 1.3834135961473848,
      "grad_norm": 1.2366554737091064,
      "learning_rate": 9.876500000000001e-05,
      "loss": 0.5197,
      "step": 24705
    },
    {
      "epoch": 1.3834695934595138,
      "grad_norm": 1.5644608736038208,
      "learning_rate": 9.876473684210527e-05,
      "loss": 0.5479,
      "step": 24706
    },
    {
      "epoch": 1.3835255907716428,
      "grad_norm": 1.2335882186889648,
      "learning_rate": 9.876447368421054e-05,
      "loss": 0.3712,
      "step": 24707
    },
    {
      "epoch": 1.3835815880837719,
      "grad_norm": 1.6657646894454956,
      "learning_rate": 9.876421052631579e-05,
      "loss": 0.4121,
      "step": 24708
    },
    {
      "epoch": 1.3836375853959009,
      "grad_norm": 1.9175894260406494,
      "learning_rate": 9.876394736842106e-05,
      "loss": 0.4764,
      "step": 24709
    },
    {
      "epoch": 1.38369358270803,
      "grad_norm": 1.399888515472412,
      "learning_rate": 9.876368421052632e-05,
      "loss": 0.5589,
      "step": 24710
    },
    {
      "epoch": 1.383749580020159,
      "grad_norm": 1.2780975103378296,
      "learning_rate": 9.876342105263159e-05,
      "loss": 0.3981,
      "step": 24711
    },
    {
      "epoch": 1.383805577332288,
      "grad_norm": 1.5270181894302368,
      "learning_rate": 9.876315789473685e-05,
      "loss": 0.3951,
      "step": 24712
    },
    {
      "epoch": 1.383861574644417,
      "grad_norm": 1.413232684135437,
      "learning_rate": 9.876289473684211e-05,
      "loss": 0.5846,
      "step": 24713
    },
    {
      "epoch": 1.383917571956546,
      "grad_norm": 1.5790107250213623,
      "learning_rate": 9.876263157894737e-05,
      "loss": 0.5064,
      "step": 24714
    },
    {
      "epoch": 1.383973569268675,
      "grad_norm": 1.382285714149475,
      "learning_rate": 9.876236842105263e-05,
      "loss": 0.5088,
      "step": 24715
    },
    {
      "epoch": 1.384029566580804,
      "grad_norm": 1.5090510845184326,
      "learning_rate": 9.87621052631579e-05,
      "loss": 0.5117,
      "step": 24716
    },
    {
      "epoch": 1.384085563892933,
      "grad_norm": 1.3253145217895508,
      "learning_rate": 9.876184210526316e-05,
      "loss": 0.4963,
      "step": 24717
    },
    {
      "epoch": 1.384141561205062,
      "grad_norm": 1.4267219305038452,
      "learning_rate": 9.876157894736842e-05,
      "loss": 0.5101,
      "step": 24718
    },
    {
      "epoch": 1.384197558517191,
      "grad_norm": 1.1575123071670532,
      "learning_rate": 9.876131578947368e-05,
      "loss": 0.3897,
      "step": 24719
    },
    {
      "epoch": 1.38425355582932,
      "grad_norm": 1.4060261249542236,
      "learning_rate": 9.876105263157896e-05,
      "loss": 0.3516,
      "step": 24720
    },
    {
      "epoch": 1.384309553141449,
      "grad_norm": 1.3479305505752563,
      "learning_rate": 9.876078947368422e-05,
      "loss": 0.3674,
      "step": 24721
    },
    {
      "epoch": 1.3843655504535781,
      "grad_norm": 1.3310964107513428,
      "learning_rate": 9.876052631578948e-05,
      "loss": 0.435,
      "step": 24722
    },
    {
      "epoch": 1.3844215477657071,
      "grad_norm": 1.4235334396362305,
      "learning_rate": 9.876026315789474e-05,
      "loss": 0.4469,
      "step": 24723
    },
    {
      "epoch": 1.3844775450778362,
      "grad_norm": 1.3577880859375,
      "learning_rate": 9.876000000000001e-05,
      "loss": 0.4412,
      "step": 24724
    },
    {
      "epoch": 1.3845335423899652,
      "grad_norm": 1.4240859746932983,
      "learning_rate": 9.875973684210527e-05,
      "loss": 0.5463,
      "step": 24725
    },
    {
      "epoch": 1.3845895397020942,
      "grad_norm": 1.590160608291626,
      "learning_rate": 9.875947368421053e-05,
      "loss": 0.5334,
      "step": 24726
    },
    {
      "epoch": 1.3846455370142232,
      "grad_norm": 1.2518236637115479,
      "learning_rate": 9.875921052631579e-05,
      "loss": 0.4658,
      "step": 24727
    },
    {
      "epoch": 1.3847015343263522,
      "grad_norm": 1.2080436944961548,
      "learning_rate": 9.875894736842106e-05,
      "loss": 0.4117,
      "step": 24728
    },
    {
      "epoch": 1.3847575316384813,
      "grad_norm": 1.5010343790054321,
      "learning_rate": 9.875868421052632e-05,
      "loss": 0.4611,
      "step": 24729
    },
    {
      "epoch": 1.3848135289506103,
      "grad_norm": 1.06449294090271,
      "learning_rate": 9.87584210526316e-05,
      "loss": 0.3236,
      "step": 24730
    },
    {
      "epoch": 1.3848695262627393,
      "grad_norm": 1.2299138307571411,
      "learning_rate": 9.875815789473684e-05,
      "loss": 0.4756,
      "step": 24731
    },
    {
      "epoch": 1.3849255235748683,
      "grad_norm": 1.4848288297653198,
      "learning_rate": 9.87578947368421e-05,
      "loss": 0.4313,
      "step": 24732
    },
    {
      "epoch": 1.3849815208869973,
      "grad_norm": 1.2216533422470093,
      "learning_rate": 9.875763157894737e-05,
      "loss": 0.5013,
      "step": 24733
    },
    {
      "epoch": 1.3850375181991263,
      "grad_norm": 1.5379114151000977,
      "learning_rate": 9.875736842105263e-05,
      "loss": 0.5764,
      "step": 24734
    },
    {
      "epoch": 1.3850935155112554,
      "grad_norm": 1.2929130792617798,
      "learning_rate": 9.875710526315791e-05,
      "loss": 0.4394,
      "step": 24735
    },
    {
      "epoch": 1.3851495128233844,
      "grad_norm": 1.3166282176971436,
      "learning_rate": 9.875684210526315e-05,
      "loss": 0.3692,
      "step": 24736
    },
    {
      "epoch": 1.3852055101355134,
      "grad_norm": 1.3195898532867432,
      "learning_rate": 9.875657894736843e-05,
      "loss": 0.3983,
      "step": 24737
    },
    {
      "epoch": 1.3852615074476424,
      "grad_norm": 1.0964648723602295,
      "learning_rate": 9.875631578947369e-05,
      "loss": 0.3791,
      "step": 24738
    },
    {
      "epoch": 1.3853175047597714,
      "grad_norm": 1.157021164894104,
      "learning_rate": 9.875605263157896e-05,
      "loss": 0.4123,
      "step": 24739
    },
    {
      "epoch": 1.3853735020719005,
      "grad_norm": 1.494493842124939,
      "learning_rate": 9.87557894736842e-05,
      "loss": 0.5978,
      "step": 24740
    },
    {
      "epoch": 1.3854294993840295,
      "grad_norm": 1.2963590621948242,
      "learning_rate": 9.875552631578948e-05,
      "loss": 0.4444,
      "step": 24741
    },
    {
      "epoch": 1.3854854966961585,
      "grad_norm": 1.553191900253296,
      "learning_rate": 9.875526315789474e-05,
      "loss": 0.5829,
      "step": 24742
    },
    {
      "epoch": 1.3855414940082875,
      "grad_norm": 1.1810563802719116,
      "learning_rate": 9.875500000000001e-05,
      "loss": 0.3162,
      "step": 24743
    },
    {
      "epoch": 1.3855974913204165,
      "grad_norm": 1.3519984483718872,
      "learning_rate": 9.875473684210527e-05,
      "loss": 0.5755,
      "step": 24744
    },
    {
      "epoch": 1.3856534886325456,
      "grad_norm": 1.3369777202606201,
      "learning_rate": 9.875447368421053e-05,
      "loss": 0.5488,
      "step": 24745
    },
    {
      "epoch": 1.3857094859446746,
      "grad_norm": 1.6594231128692627,
      "learning_rate": 9.875421052631579e-05,
      "loss": 0.7104,
      "step": 24746
    },
    {
      "epoch": 1.3857654832568036,
      "grad_norm": 1.256836175918579,
      "learning_rate": 9.875394736842106e-05,
      "loss": 0.403,
      "step": 24747
    },
    {
      "epoch": 1.3858214805689326,
      "grad_norm": 1.3793790340423584,
      "learning_rate": 9.875368421052632e-05,
      "loss": 0.5349,
      "step": 24748
    },
    {
      "epoch": 1.3858774778810616,
      "grad_norm": 1.877210259437561,
      "learning_rate": 9.875342105263158e-05,
      "loss": 0.5041,
      "step": 24749
    },
    {
      "epoch": 1.3859334751931907,
      "grad_norm": 1.3058441877365112,
      "learning_rate": 9.875315789473684e-05,
      "loss": 0.5957,
      "step": 24750
    },
    {
      "epoch": 1.3859894725053197,
      "grad_norm": 1.1621700525283813,
      "learning_rate": 9.87528947368421e-05,
      "loss": 0.3936,
      "step": 24751
    },
    {
      "epoch": 1.3860454698174487,
      "grad_norm": 1.2263314723968506,
      "learning_rate": 9.875263157894738e-05,
      "loss": 0.4816,
      "step": 24752
    },
    {
      "epoch": 1.3861014671295777,
      "grad_norm": 1.1986078023910522,
      "learning_rate": 9.875236842105264e-05,
      "loss": 0.4512,
      "step": 24753
    },
    {
      "epoch": 1.3861574644417067,
      "grad_norm": 1.5553449392318726,
      "learning_rate": 9.87521052631579e-05,
      "loss": 0.4484,
      "step": 24754
    },
    {
      "epoch": 1.3862134617538358,
      "grad_norm": 1.3770147562026978,
      "learning_rate": 9.875184210526316e-05,
      "loss": 0.5147,
      "step": 24755
    },
    {
      "epoch": 1.3862694590659648,
      "grad_norm": 1.026626706123352,
      "learning_rate": 9.875157894736843e-05,
      "loss": 0.323,
      "step": 24756
    },
    {
      "epoch": 1.3863254563780938,
      "grad_norm": 1.424196481704712,
      "learning_rate": 9.875131578947369e-05,
      "loss": 0.5097,
      "step": 24757
    },
    {
      "epoch": 1.3863814536902228,
      "grad_norm": 1.755611777305603,
      "learning_rate": 9.875105263157895e-05,
      "loss": 0.52,
      "step": 24758
    },
    {
      "epoch": 1.3864374510023518,
      "grad_norm": 1.2150194644927979,
      "learning_rate": 9.875078947368421e-05,
      "loss": 0.4028,
      "step": 24759
    },
    {
      "epoch": 1.3864934483144808,
      "grad_norm": 1.298742413520813,
      "learning_rate": 9.875052631578948e-05,
      "loss": 0.4336,
      "step": 24760
    },
    {
      "epoch": 1.3865494456266099,
      "grad_norm": 1.5085780620574951,
      "learning_rate": 9.875026315789474e-05,
      "loss": 0.3935,
      "step": 24761
    },
    {
      "epoch": 1.3866054429387389,
      "grad_norm": 1.3242181539535522,
      "learning_rate": 9.875000000000002e-05,
      "loss": 0.51,
      "step": 24762
    },
    {
      "epoch": 1.386661440250868,
      "grad_norm": 1.444621205329895,
      "learning_rate": 9.874973684210526e-05,
      "loss": 0.4794,
      "step": 24763
    },
    {
      "epoch": 1.386717437562997,
      "grad_norm": 1.2220877408981323,
      "learning_rate": 9.874947368421053e-05,
      "loss": 0.4074,
      "step": 24764
    },
    {
      "epoch": 1.386773434875126,
      "grad_norm": 1.3091826438903809,
      "learning_rate": 9.87492105263158e-05,
      "loss": 0.4074,
      "step": 24765
    },
    {
      "epoch": 1.386829432187255,
      "grad_norm": 1.3461570739746094,
      "learning_rate": 9.874894736842105e-05,
      "loss": 0.4497,
      "step": 24766
    },
    {
      "epoch": 1.386885429499384,
      "grad_norm": 1.4616492986679077,
      "learning_rate": 9.874868421052633e-05,
      "loss": 0.4912,
      "step": 24767
    },
    {
      "epoch": 1.386941426811513,
      "grad_norm": 1.343125343322754,
      "learning_rate": 9.874842105263157e-05,
      "loss": 0.5187,
      "step": 24768
    },
    {
      "epoch": 1.386997424123642,
      "grad_norm": 1.45669424533844,
      "learning_rate": 9.874815789473685e-05,
      "loss": 0.5832,
      "step": 24769
    },
    {
      "epoch": 1.387053421435771,
      "grad_norm": 1.4133449792861938,
      "learning_rate": 9.87478947368421e-05,
      "loss": 0.4095,
      "step": 24770
    },
    {
      "epoch": 1.3871094187479,
      "grad_norm": 1.8313589096069336,
      "learning_rate": 9.874763157894738e-05,
      "loss": 0.6705,
      "step": 24771
    },
    {
      "epoch": 1.387165416060029,
      "grad_norm": 1.3852347135543823,
      "learning_rate": 9.874736842105264e-05,
      "loss": 0.5292,
      "step": 24772
    },
    {
      "epoch": 1.387221413372158,
      "grad_norm": 1.3958368301391602,
      "learning_rate": 9.87471052631579e-05,
      "loss": 0.3874,
      "step": 24773
    },
    {
      "epoch": 1.3872774106842871,
      "grad_norm": 1.496346354484558,
      "learning_rate": 9.874684210526316e-05,
      "loss": 0.5661,
      "step": 24774
    },
    {
      "epoch": 1.3873334079964161,
      "grad_norm": 1.2413619756698608,
      "learning_rate": 9.874657894736843e-05,
      "loss": 0.5856,
      "step": 24775
    },
    {
      "epoch": 1.3873894053085452,
      "grad_norm": 1.4150477647781372,
      "learning_rate": 9.874631578947369e-05,
      "loss": 0.4548,
      "step": 24776
    },
    {
      "epoch": 1.3874454026206742,
      "grad_norm": 1.2681676149368286,
      "learning_rate": 9.874605263157895e-05,
      "loss": 0.5001,
      "step": 24777
    },
    {
      "epoch": 1.3875013999328032,
      "grad_norm": 1.6500073671340942,
      "learning_rate": 9.874578947368421e-05,
      "loss": 0.5182,
      "step": 24778
    },
    {
      "epoch": 1.3875573972449322,
      "grad_norm": 1.3954724073410034,
      "learning_rate": 9.874552631578948e-05,
      "loss": 0.3805,
      "step": 24779
    },
    {
      "epoch": 1.3876133945570612,
      "grad_norm": 1.540405035018921,
      "learning_rate": 9.874526315789474e-05,
      "loss": 0.6096,
      "step": 24780
    },
    {
      "epoch": 1.3876693918691902,
      "grad_norm": 1.313897728919983,
      "learning_rate": 9.8745e-05,
      "loss": 0.4262,
      "step": 24781
    },
    {
      "epoch": 1.3877253891813193,
      "grad_norm": 1.4382654428482056,
      "learning_rate": 9.874473684210526e-05,
      "loss": 0.4548,
      "step": 24782
    },
    {
      "epoch": 1.3877813864934483,
      "grad_norm": 1.3728103637695312,
      "learning_rate": 9.874447368421052e-05,
      "loss": 0.4775,
      "step": 24783
    },
    {
      "epoch": 1.3878373838055773,
      "grad_norm": 1.2177050113677979,
      "learning_rate": 9.87442105263158e-05,
      "loss": 0.4215,
      "step": 24784
    },
    {
      "epoch": 1.3878933811177063,
      "grad_norm": 1.2666003704071045,
      "learning_rate": 9.874394736842106e-05,
      "loss": 0.5811,
      "step": 24785
    },
    {
      "epoch": 1.3879493784298353,
      "grad_norm": 1.3011101484298706,
      "learning_rate": 9.874368421052632e-05,
      "loss": 0.4707,
      "step": 24786
    },
    {
      "epoch": 1.3880053757419644,
      "grad_norm": 1.4403101205825806,
      "learning_rate": 9.874342105263158e-05,
      "loss": 0.5231,
      "step": 24787
    },
    {
      "epoch": 1.3880613730540934,
      "grad_norm": 1.5996503829956055,
      "learning_rate": 9.874315789473685e-05,
      "loss": 0.5047,
      "step": 24788
    },
    {
      "epoch": 1.3881173703662224,
      "grad_norm": 1.383208155632019,
      "learning_rate": 9.874289473684211e-05,
      "loss": 0.4274,
      "step": 24789
    },
    {
      "epoch": 1.3881733676783514,
      "grad_norm": 1.2859172821044922,
      "learning_rate": 9.874263157894737e-05,
      "loss": 0.4462,
      "step": 24790
    },
    {
      "epoch": 1.3882293649904804,
      "grad_norm": 1.318159818649292,
      "learning_rate": 9.874236842105263e-05,
      "loss": 0.4422,
      "step": 24791
    },
    {
      "epoch": 1.3882853623026095,
      "grad_norm": 1.6016712188720703,
      "learning_rate": 9.87421052631579e-05,
      "loss": 0.4696,
      "step": 24792
    },
    {
      "epoch": 1.3883413596147385,
      "grad_norm": 1.4052642583847046,
      "learning_rate": 9.874184210526316e-05,
      "loss": 0.534,
      "step": 24793
    },
    {
      "epoch": 1.3883973569268675,
      "grad_norm": 1.2752387523651123,
      "learning_rate": 9.874157894736843e-05,
      "loss": 0.4008,
      "step": 24794
    },
    {
      "epoch": 1.3884533542389965,
      "grad_norm": 1.1638542413711548,
      "learning_rate": 9.874131578947368e-05,
      "loss": 0.494,
      "step": 24795
    },
    {
      "epoch": 1.3885093515511255,
      "grad_norm": 1.2391552925109863,
      "learning_rate": 9.874105263157895e-05,
      "loss": 0.5531,
      "step": 24796
    },
    {
      "epoch": 1.3885653488632546,
      "grad_norm": 1.5980744361877441,
      "learning_rate": 9.874078947368421e-05,
      "loss": 0.3775,
      "step": 24797
    },
    {
      "epoch": 1.3886213461753836,
      "grad_norm": 1.2987377643585205,
      "learning_rate": 9.874052631578949e-05,
      "loss": 0.448,
      "step": 24798
    },
    {
      "epoch": 1.3886773434875126,
      "grad_norm": 1.1683251857757568,
      "learning_rate": 9.874026315789475e-05,
      "loss": 0.3695,
      "step": 24799
    },
    {
      "epoch": 1.3887333407996416,
      "grad_norm": 1.540535807609558,
      "learning_rate": 9.874e-05,
      "loss": 0.3995,
      "step": 24800
    },
    {
      "epoch": 1.3887893381117706,
      "grad_norm": 1.7004040479660034,
      "learning_rate": 9.873973684210527e-05,
      "loss": 0.6755,
      "step": 24801
    },
    {
      "epoch": 1.3888453354238997,
      "grad_norm": 1.4696658849716187,
      "learning_rate": 9.873947368421053e-05,
      "loss": 0.5824,
      "step": 24802
    },
    {
      "epoch": 1.3889013327360287,
      "grad_norm": 1.2471377849578857,
      "learning_rate": 9.87392105263158e-05,
      "loss": 0.4334,
      "step": 24803
    },
    {
      "epoch": 1.3889573300481577,
      "grad_norm": 1.641444444656372,
      "learning_rate": 9.873894736842106e-05,
      "loss": 0.4461,
      "step": 24804
    },
    {
      "epoch": 1.3890133273602867,
      "grad_norm": 1.4771136045455933,
      "learning_rate": 9.873868421052632e-05,
      "loss": 0.4074,
      "step": 24805
    },
    {
      "epoch": 1.3890693246724157,
      "grad_norm": 1.2724125385284424,
      "learning_rate": 9.873842105263158e-05,
      "loss": 0.4007,
      "step": 24806
    },
    {
      "epoch": 1.3891253219845447,
      "grad_norm": 1.336830973625183,
      "learning_rate": 9.873815789473685e-05,
      "loss": 0.4776,
      "step": 24807
    },
    {
      "epoch": 1.3891813192966738,
      "grad_norm": 1.6851872205734253,
      "learning_rate": 9.873789473684211e-05,
      "loss": 0.5717,
      "step": 24808
    },
    {
      "epoch": 1.3892373166088028,
      "grad_norm": 1.4466724395751953,
      "learning_rate": 9.873763157894737e-05,
      "loss": 0.5386,
      "step": 24809
    },
    {
      "epoch": 1.3892933139209318,
      "grad_norm": 1.3866513967514038,
      "learning_rate": 9.873736842105263e-05,
      "loss": 0.4549,
      "step": 24810
    },
    {
      "epoch": 1.3893493112330608,
      "grad_norm": 1.4376996755599976,
      "learning_rate": 9.87371052631579e-05,
      "loss": 0.5588,
      "step": 24811
    },
    {
      "epoch": 1.3894053085451898,
      "grad_norm": 1.3783830404281616,
      "learning_rate": 9.873684210526316e-05,
      "loss": 0.3829,
      "step": 24812
    },
    {
      "epoch": 1.3894613058573189,
      "grad_norm": 1.3257275819778442,
      "learning_rate": 9.873657894736842e-05,
      "loss": 0.4604,
      "step": 24813
    },
    {
      "epoch": 1.3895173031694479,
      "grad_norm": 1.4820529222488403,
      "learning_rate": 9.873631578947368e-05,
      "loss": 0.4265,
      "step": 24814
    },
    {
      "epoch": 1.389573300481577,
      "grad_norm": 1.401774287223816,
      "learning_rate": 9.873605263157896e-05,
      "loss": 0.4687,
      "step": 24815
    },
    {
      "epoch": 1.389629297793706,
      "grad_norm": 1.2722257375717163,
      "learning_rate": 9.873578947368422e-05,
      "loss": 0.4431,
      "step": 24816
    },
    {
      "epoch": 1.389685295105835,
      "grad_norm": 1.3507345914840698,
      "learning_rate": 9.873552631578949e-05,
      "loss": 0.499,
      "step": 24817
    },
    {
      "epoch": 1.389741292417964,
      "grad_norm": 1.3066329956054688,
      "learning_rate": 9.873526315789474e-05,
      "loss": 0.3926,
      "step": 24818
    },
    {
      "epoch": 1.389797289730093,
      "grad_norm": 1.2284746170043945,
      "learning_rate": 9.8735e-05,
      "loss": 0.4999,
      "step": 24819
    },
    {
      "epoch": 1.389853287042222,
      "grad_norm": 1.3337175846099854,
      "learning_rate": 9.873473684210527e-05,
      "loss": 0.4306,
      "step": 24820
    },
    {
      "epoch": 1.389909284354351,
      "grad_norm": 1.391411542892456,
      "learning_rate": 9.873447368421053e-05,
      "loss": 0.5146,
      "step": 24821
    },
    {
      "epoch": 1.38996528166648,
      "grad_norm": 1.3370726108551025,
      "learning_rate": 9.87342105263158e-05,
      "loss": 0.5088,
      "step": 24822
    },
    {
      "epoch": 1.390021278978609,
      "grad_norm": 1.313614010810852,
      "learning_rate": 9.873394736842105e-05,
      "loss": 0.545,
      "step": 24823
    },
    {
      "epoch": 1.390077276290738,
      "grad_norm": 1.2633293867111206,
      "learning_rate": 9.873368421052632e-05,
      "loss": 0.5053,
      "step": 24824
    },
    {
      "epoch": 1.390133273602867,
      "grad_norm": 1.53170907497406,
      "learning_rate": 9.873342105263158e-05,
      "loss": 0.541,
      "step": 24825
    },
    {
      "epoch": 1.3901892709149961,
      "grad_norm": 1.2162044048309326,
      "learning_rate": 9.873315789473685e-05,
      "loss": 0.4113,
      "step": 24826
    },
    {
      "epoch": 1.3902452682271251,
      "grad_norm": 1.4812928438186646,
      "learning_rate": 9.873289473684211e-05,
      "loss": 0.416,
      "step": 24827
    },
    {
      "epoch": 1.3903012655392541,
      "grad_norm": 1.2285056114196777,
      "learning_rate": 9.873263157894737e-05,
      "loss": 0.304,
      "step": 24828
    },
    {
      "epoch": 1.3903572628513832,
      "grad_norm": 1.3497986793518066,
      "learning_rate": 9.873236842105263e-05,
      "loss": 0.5413,
      "step": 24829
    },
    {
      "epoch": 1.3904132601635122,
      "grad_norm": 1.410062551498413,
      "learning_rate": 9.873210526315791e-05,
      "loss": 0.6767,
      "step": 24830
    },
    {
      "epoch": 1.3904692574756412,
      "grad_norm": 1.5519541501998901,
      "learning_rate": 9.873184210526317e-05,
      "loss": 0.5966,
      "step": 24831
    },
    {
      "epoch": 1.3905252547877702,
      "grad_norm": 1.116697907447815,
      "learning_rate": 9.873157894736843e-05,
      "loss": 0.474,
      "step": 24832
    },
    {
      "epoch": 1.3905812520998992,
      "grad_norm": 1.367313265800476,
      "learning_rate": 9.873131578947369e-05,
      "loss": 0.4658,
      "step": 24833
    },
    {
      "epoch": 1.3906372494120283,
      "grad_norm": 1.3255162239074707,
      "learning_rate": 9.873105263157896e-05,
      "loss": 0.5173,
      "step": 24834
    },
    {
      "epoch": 1.3906932467241573,
      "grad_norm": 1.3296301364898682,
      "learning_rate": 9.873078947368422e-05,
      "loss": 0.5345,
      "step": 24835
    },
    {
      "epoch": 1.3907492440362863,
      "grad_norm": 1.52505624294281,
      "learning_rate": 9.873052631578948e-05,
      "loss": 0.4608,
      "step": 24836
    },
    {
      "epoch": 1.3908052413484153,
      "grad_norm": 1.3161721229553223,
      "learning_rate": 9.873026315789474e-05,
      "loss": 0.5049,
      "step": 24837
    },
    {
      "epoch": 1.3908612386605443,
      "grad_norm": 1.3209775686264038,
      "learning_rate": 9.873e-05,
      "loss": 0.397,
      "step": 24838
    },
    {
      "epoch": 1.3909172359726734,
      "grad_norm": 1.4917718172073364,
      "learning_rate": 9.872973684210527e-05,
      "loss": 0.694,
      "step": 24839
    },
    {
      "epoch": 1.3909732332848024,
      "grad_norm": 1.4647622108459473,
      "learning_rate": 9.872947368421053e-05,
      "loss": 0.6038,
      "step": 24840
    },
    {
      "epoch": 1.3910292305969314,
      "grad_norm": 1.9479498863220215,
      "learning_rate": 9.872921052631579e-05,
      "loss": 0.5725,
      "step": 24841
    },
    {
      "epoch": 1.3910852279090604,
      "grad_norm": 1.154721736907959,
      "learning_rate": 9.872894736842105e-05,
      "loss": 0.3818,
      "step": 24842
    },
    {
      "epoch": 1.3911412252211894,
      "grad_norm": 1.3792271614074707,
      "learning_rate": 9.872868421052632e-05,
      "loss": 0.5345,
      "step": 24843
    },
    {
      "epoch": 1.3911972225333185,
      "grad_norm": 1.2258814573287964,
      "learning_rate": 9.872842105263158e-05,
      "loss": 0.4447,
      "step": 24844
    },
    {
      "epoch": 1.3912532198454475,
      "grad_norm": 1.2293710708618164,
      "learning_rate": 9.872815789473684e-05,
      "loss": 0.4907,
      "step": 24845
    },
    {
      "epoch": 1.3913092171575765,
      "grad_norm": 1.2616662979125977,
      "learning_rate": 9.87278947368421e-05,
      "loss": 0.5773,
      "step": 24846
    },
    {
      "epoch": 1.3913652144697055,
      "grad_norm": 1.1698933839797974,
      "learning_rate": 9.872763157894738e-05,
      "loss": 0.3963,
      "step": 24847
    },
    {
      "epoch": 1.3914212117818345,
      "grad_norm": 1.3462740182876587,
      "learning_rate": 9.872736842105264e-05,
      "loss": 0.4663,
      "step": 24848
    },
    {
      "epoch": 1.3914772090939636,
      "grad_norm": 1.2720924615859985,
      "learning_rate": 9.872710526315791e-05,
      "loss": 0.456,
      "step": 24849
    },
    {
      "epoch": 1.3915332064060926,
      "grad_norm": 1.2319821119308472,
      "learning_rate": 9.872684210526316e-05,
      "loss": 0.4582,
      "step": 24850
    },
    {
      "epoch": 1.3915892037182216,
      "grad_norm": 1.3817527294158936,
      "learning_rate": 9.872657894736843e-05,
      "loss": 0.558,
      "step": 24851
    },
    {
      "epoch": 1.3916452010303506,
      "grad_norm": 1.1048712730407715,
      "learning_rate": 9.872631578947369e-05,
      "loss": 0.405,
      "step": 24852
    },
    {
      "epoch": 1.3917011983424796,
      "grad_norm": 1.3013241291046143,
      "learning_rate": 9.872605263157895e-05,
      "loss": 0.4592,
      "step": 24853
    },
    {
      "epoch": 1.3917571956546086,
      "grad_norm": 1.7271647453308105,
      "learning_rate": 9.872578947368422e-05,
      "loss": 0.5207,
      "step": 24854
    },
    {
      "epoch": 1.3918131929667377,
      "grad_norm": 1.4949439764022827,
      "learning_rate": 9.872552631578947e-05,
      "loss": 0.4057,
      "step": 24855
    },
    {
      "epoch": 1.3918691902788667,
      "grad_norm": 1.1679201126098633,
      "learning_rate": 9.872526315789474e-05,
      "loss": 0.4945,
      "step": 24856
    },
    {
      "epoch": 1.3919251875909957,
      "grad_norm": 1.4353044033050537,
      "learning_rate": 9.8725e-05,
      "loss": 0.4496,
      "step": 24857
    },
    {
      "epoch": 1.3919811849031247,
      "grad_norm": 1.3244861364364624,
      "learning_rate": 9.872473684210527e-05,
      "loss": 0.5183,
      "step": 24858
    },
    {
      "epoch": 1.3920371822152537,
      "grad_norm": 1.666304588317871,
      "learning_rate": 9.872447368421053e-05,
      "loss": 0.4756,
      "step": 24859
    },
    {
      "epoch": 1.3920931795273828,
      "grad_norm": 1.4096368551254272,
      "learning_rate": 9.87242105263158e-05,
      "loss": 0.5145,
      "step": 24860
    },
    {
      "epoch": 1.3921491768395118,
      "grad_norm": 1.434039831161499,
      "learning_rate": 9.872394736842105e-05,
      "loss": 0.3844,
      "step": 24861
    },
    {
      "epoch": 1.3922051741516408,
      "grad_norm": 1.3839205503463745,
      "learning_rate": 9.872368421052633e-05,
      "loss": 0.4218,
      "step": 24862
    },
    {
      "epoch": 1.3922611714637698,
      "grad_norm": 1.1076642274856567,
      "learning_rate": 9.872342105263159e-05,
      "loss": 0.3004,
      "step": 24863
    },
    {
      "epoch": 1.3923171687758988,
      "grad_norm": 1.3232903480529785,
      "learning_rate": 9.872315789473685e-05,
      "loss": 0.4578,
      "step": 24864
    },
    {
      "epoch": 1.3923731660880279,
      "grad_norm": 1.181853175163269,
      "learning_rate": 9.87228947368421e-05,
      "loss": 0.4971,
      "step": 24865
    },
    {
      "epoch": 1.3924291634001569,
      "grad_norm": 1.3774648904800415,
      "learning_rate": 9.872263157894738e-05,
      "loss": 0.4604,
      "step": 24866
    },
    {
      "epoch": 1.392485160712286,
      "grad_norm": 1.4425255060195923,
      "learning_rate": 9.872236842105264e-05,
      "loss": 0.4997,
      "step": 24867
    },
    {
      "epoch": 1.392541158024415,
      "grad_norm": 1.4591766595840454,
      "learning_rate": 9.87221052631579e-05,
      "loss": 0.5154,
      "step": 24868
    },
    {
      "epoch": 1.392597155336544,
      "grad_norm": 1.1732569932937622,
      "learning_rate": 9.872184210526316e-05,
      "loss": 0.3848,
      "step": 24869
    },
    {
      "epoch": 1.392653152648673,
      "grad_norm": 1.3114980459213257,
      "learning_rate": 9.872157894736842e-05,
      "loss": 0.4703,
      "step": 24870
    },
    {
      "epoch": 1.392709149960802,
      "grad_norm": 1.6508270502090454,
      "learning_rate": 9.872131578947369e-05,
      "loss": 0.4463,
      "step": 24871
    },
    {
      "epoch": 1.392765147272931,
      "grad_norm": 1.3888405561447144,
      "learning_rate": 9.872105263157895e-05,
      "loss": 0.6197,
      "step": 24872
    },
    {
      "epoch": 1.39282114458506,
      "grad_norm": 1.3268065452575684,
      "learning_rate": 9.872078947368421e-05,
      "loss": 0.5795,
      "step": 24873
    },
    {
      "epoch": 1.392877141897189,
      "grad_norm": 1.2224912643432617,
      "learning_rate": 9.872052631578947e-05,
      "loss": 0.3589,
      "step": 24874
    },
    {
      "epoch": 1.392933139209318,
      "grad_norm": 1.2793575525283813,
      "learning_rate": 9.872026315789474e-05,
      "loss": 0.401,
      "step": 24875
    },
    {
      "epoch": 1.392989136521447,
      "grad_norm": 1.3169201612472534,
      "learning_rate": 9.872e-05,
      "loss": 0.5786,
      "step": 24876
    },
    {
      "epoch": 1.393045133833576,
      "grad_norm": 1.3333497047424316,
      "learning_rate": 9.871973684210528e-05,
      "loss": 0.5736,
      "step": 24877
    },
    {
      "epoch": 1.393101131145705,
      "grad_norm": 1.2146955728530884,
      "learning_rate": 9.871947368421052e-05,
      "loss": 0.4339,
      "step": 24878
    },
    {
      "epoch": 1.3931571284578341,
      "grad_norm": 1.5016287565231323,
      "learning_rate": 9.87192105263158e-05,
      "loss": 0.4464,
      "step": 24879
    },
    {
      "epoch": 1.3932131257699631,
      "grad_norm": 1.2557700872421265,
      "learning_rate": 9.871894736842106e-05,
      "loss": 0.437,
      "step": 24880
    },
    {
      "epoch": 1.3932691230820922,
      "grad_norm": 1.499138355255127,
      "learning_rate": 9.871868421052633e-05,
      "loss": 0.585,
      "step": 24881
    },
    {
      "epoch": 1.3933251203942212,
      "grad_norm": 1.5727580785751343,
      "learning_rate": 9.871842105263159e-05,
      "loss": 0.4809,
      "step": 24882
    },
    {
      "epoch": 1.3933811177063502,
      "grad_norm": 2.0146894454956055,
      "learning_rate": 9.871815789473685e-05,
      "loss": 0.6236,
      "step": 24883
    },
    {
      "epoch": 1.3934371150184792,
      "grad_norm": 1.3595463037490845,
      "learning_rate": 9.871789473684211e-05,
      "loss": 0.498,
      "step": 24884
    },
    {
      "epoch": 1.3934931123306082,
      "grad_norm": 1.3299007415771484,
      "learning_rate": 9.871763157894738e-05,
      "loss": 0.3855,
      "step": 24885
    },
    {
      "epoch": 1.3935491096427373,
      "grad_norm": 1.214621663093567,
      "learning_rate": 9.871736842105264e-05,
      "loss": 0.4616,
      "step": 24886
    },
    {
      "epoch": 1.3936051069548663,
      "grad_norm": 1.3083232641220093,
      "learning_rate": 9.871710526315789e-05,
      "loss": 0.4835,
      "step": 24887
    },
    {
      "epoch": 1.3936611042669953,
      "grad_norm": 1.3820178508758545,
      "learning_rate": 9.871684210526316e-05,
      "loss": 0.4902,
      "step": 24888
    },
    {
      "epoch": 1.3937171015791243,
      "grad_norm": 1.3684327602386475,
      "learning_rate": 9.871657894736842e-05,
      "loss": 0.5788,
      "step": 24889
    },
    {
      "epoch": 1.3937730988912533,
      "grad_norm": 1.7087435722351074,
      "learning_rate": 9.87163157894737e-05,
      "loss": 0.6771,
      "step": 24890
    },
    {
      "epoch": 1.3938290962033824,
      "grad_norm": 1.7902201414108276,
      "learning_rate": 9.871605263157895e-05,
      "loss": 0.7825,
      "step": 24891
    },
    {
      "epoch": 1.3938850935155114,
      "grad_norm": 1.5060359239578247,
      "learning_rate": 9.871578947368421e-05,
      "loss": 0.4653,
      "step": 24892
    },
    {
      "epoch": 1.3939410908276404,
      "grad_norm": 1.374214768409729,
      "learning_rate": 9.871552631578947e-05,
      "loss": 0.4788,
      "step": 24893
    },
    {
      "epoch": 1.3939970881397694,
      "grad_norm": 1.6973198652267456,
      "learning_rate": 9.871526315789475e-05,
      "loss": 0.5671,
      "step": 24894
    },
    {
      "epoch": 1.3940530854518984,
      "grad_norm": 1.3596285581588745,
      "learning_rate": 9.8715e-05,
      "loss": 0.4583,
      "step": 24895
    },
    {
      "epoch": 1.3941090827640275,
      "grad_norm": 1.5764806270599365,
      "learning_rate": 9.871473684210527e-05,
      "loss": 0.4715,
      "step": 24896
    },
    {
      "epoch": 1.3941650800761565,
      "grad_norm": 2.0739612579345703,
      "learning_rate": 9.871447368421053e-05,
      "loss": 0.5395,
      "step": 24897
    },
    {
      "epoch": 1.3942210773882855,
      "grad_norm": 1.426306128501892,
      "learning_rate": 9.87142105263158e-05,
      "loss": 0.4057,
      "step": 24898
    },
    {
      "epoch": 1.3942770747004145,
      "grad_norm": 1.3731775283813477,
      "learning_rate": 9.871394736842106e-05,
      "loss": 0.4563,
      "step": 24899
    },
    {
      "epoch": 1.3943330720125435,
      "grad_norm": 1.3683106899261475,
      "learning_rate": 9.871368421052632e-05,
      "loss": 0.4421,
      "step": 24900
    },
    {
      "epoch": 1.3943890693246725,
      "grad_norm": 1.569212555885315,
      "learning_rate": 9.871342105263158e-05,
      "loss": 0.5724,
      "step": 24901
    },
    {
      "epoch": 1.3944450666368016,
      "grad_norm": 1.333946704864502,
      "learning_rate": 9.871315789473685e-05,
      "loss": 0.4599,
      "step": 24902
    },
    {
      "epoch": 1.3945010639489306,
      "grad_norm": 1.504699945449829,
      "learning_rate": 9.871289473684211e-05,
      "loss": 0.4934,
      "step": 24903
    },
    {
      "epoch": 1.3945570612610596,
      "grad_norm": 1.2900937795639038,
      "learning_rate": 9.871263157894737e-05,
      "loss": 0.4915,
      "step": 24904
    },
    {
      "epoch": 1.3946130585731886,
      "grad_norm": 1.4753167629241943,
      "learning_rate": 9.871236842105263e-05,
      "loss": 0.4803,
      "step": 24905
    },
    {
      "epoch": 1.3946690558853176,
      "grad_norm": 2.1940629482269287,
      "learning_rate": 9.871210526315789e-05,
      "loss": 0.6225,
      "step": 24906
    },
    {
      "epoch": 1.3947250531974464,
      "grad_norm": 1.283613920211792,
      "learning_rate": 9.871184210526316e-05,
      "loss": 0.4688,
      "step": 24907
    },
    {
      "epoch": 1.3947810505095755,
      "grad_norm": 1.2940969467163086,
      "learning_rate": 9.871157894736842e-05,
      "loss": 0.5717,
      "step": 24908
    },
    {
      "epoch": 1.3948370478217045,
      "grad_norm": 1.5835634469985962,
      "learning_rate": 9.87113157894737e-05,
      "loss": 0.4983,
      "step": 24909
    },
    {
      "epoch": 1.3948930451338335,
      "grad_norm": 1.1694972515106201,
      "learning_rate": 9.871105263157894e-05,
      "loss": 0.3836,
      "step": 24910
    },
    {
      "epoch": 1.3949490424459625,
      "grad_norm": 1.377196192741394,
      "learning_rate": 9.871078947368422e-05,
      "loss": 0.4218,
      "step": 24911
    },
    {
      "epoch": 1.3950050397580915,
      "grad_norm": 1.129834771156311,
      "learning_rate": 9.871052631578948e-05,
      "loss": 0.4985,
      "step": 24912
    },
    {
      "epoch": 1.3950610370702206,
      "grad_norm": 1.351381540298462,
      "learning_rate": 9.871026315789475e-05,
      "loss": 0.5257,
      "step": 24913
    },
    {
      "epoch": 1.3951170343823496,
      "grad_norm": 1.200209379196167,
      "learning_rate": 9.871000000000001e-05,
      "loss": 0.4166,
      "step": 24914
    },
    {
      "epoch": 1.3951730316944786,
      "grad_norm": 1.3657712936401367,
      "learning_rate": 9.870973684210527e-05,
      "loss": 0.5301,
      "step": 24915
    },
    {
      "epoch": 1.3952290290066076,
      "grad_norm": 1.5016694068908691,
      "learning_rate": 9.870947368421053e-05,
      "loss": 0.4532,
      "step": 24916
    },
    {
      "epoch": 1.3952850263187366,
      "grad_norm": 1.2941175699234009,
      "learning_rate": 9.87092105263158e-05,
      "loss": 0.4956,
      "step": 24917
    },
    {
      "epoch": 1.3953410236308657,
      "grad_norm": 1.1918156147003174,
      "learning_rate": 9.870894736842106e-05,
      "loss": 0.427,
      "step": 24918
    },
    {
      "epoch": 1.3953970209429947,
      "grad_norm": 1.270900011062622,
      "learning_rate": 9.870868421052632e-05,
      "loss": 0.4563,
      "step": 24919
    },
    {
      "epoch": 1.3954530182551237,
      "grad_norm": 1.9775011539459229,
      "learning_rate": 9.870842105263158e-05,
      "loss": 0.5685,
      "step": 24920
    },
    {
      "epoch": 1.3955090155672527,
      "grad_norm": 1.4194166660308838,
      "learning_rate": 9.870815789473685e-05,
      "loss": 0.5493,
      "step": 24921
    },
    {
      "epoch": 1.3955650128793817,
      "grad_norm": 1.7846382856369019,
      "learning_rate": 9.870789473684211e-05,
      "loss": 0.4364,
      "step": 24922
    },
    {
      "epoch": 1.3956210101915107,
      "grad_norm": 1.2885242700576782,
      "learning_rate": 9.870763157894737e-05,
      "loss": 0.4624,
      "step": 24923
    },
    {
      "epoch": 1.3956770075036398,
      "grad_norm": 1.304019570350647,
      "learning_rate": 9.870736842105263e-05,
      "loss": 0.3475,
      "step": 24924
    },
    {
      "epoch": 1.3957330048157688,
      "grad_norm": 1.1312437057495117,
      "learning_rate": 9.870710526315789e-05,
      "loss": 0.4062,
      "step": 24925
    },
    {
      "epoch": 1.3957890021278978,
      "grad_norm": 1.23923921585083,
      "learning_rate": 9.870684210526317e-05,
      "loss": 0.3765,
      "step": 24926
    },
    {
      "epoch": 1.3958449994400268,
      "grad_norm": 1.3791269063949585,
      "learning_rate": 9.870657894736843e-05,
      "loss": 0.4871,
      "step": 24927
    },
    {
      "epoch": 1.3959009967521558,
      "grad_norm": 1.3312296867370605,
      "learning_rate": 9.870631578947369e-05,
      "loss": 0.3912,
      "step": 24928
    },
    {
      "epoch": 1.3959569940642849,
      "grad_norm": 1.7769653797149658,
      "learning_rate": 9.870605263157895e-05,
      "loss": 0.475,
      "step": 24929
    },
    {
      "epoch": 1.3960129913764139,
      "grad_norm": 1.4600893259048462,
      "learning_rate": 9.870578947368422e-05,
      "loss": 0.5491,
      "step": 24930
    },
    {
      "epoch": 1.396068988688543,
      "grad_norm": 1.404732346534729,
      "learning_rate": 9.870552631578948e-05,
      "loss": 0.4643,
      "step": 24931
    },
    {
      "epoch": 1.396124986000672,
      "grad_norm": 1.6797943115234375,
      "learning_rate": 9.870526315789475e-05,
      "loss": 0.7382,
      "step": 24932
    },
    {
      "epoch": 1.396180983312801,
      "grad_norm": 1.4181331396102905,
      "learning_rate": 9.8705e-05,
      "loss": 0.5598,
      "step": 24933
    },
    {
      "epoch": 1.39623698062493,
      "grad_norm": 1.5253980159759521,
      "learning_rate": 9.870473684210527e-05,
      "loss": 0.4718,
      "step": 24934
    },
    {
      "epoch": 1.396292977937059,
      "grad_norm": 1.158437728881836,
      "learning_rate": 9.870447368421053e-05,
      "loss": 0.3587,
      "step": 24935
    },
    {
      "epoch": 1.396348975249188,
      "grad_norm": 1.0369495153427124,
      "learning_rate": 9.87042105263158e-05,
      "loss": 0.2864,
      "step": 24936
    },
    {
      "epoch": 1.396404972561317,
      "grad_norm": 1.5150394439697266,
      "learning_rate": 9.870394736842105e-05,
      "loss": 0.4394,
      "step": 24937
    },
    {
      "epoch": 1.396460969873446,
      "grad_norm": 1.1184009313583374,
      "learning_rate": 9.870368421052632e-05,
      "loss": 0.3565,
      "step": 24938
    },
    {
      "epoch": 1.396516967185575,
      "grad_norm": 1.5306912660598755,
      "learning_rate": 9.870342105263158e-05,
      "loss": 0.5793,
      "step": 24939
    },
    {
      "epoch": 1.396572964497704,
      "grad_norm": 1.180704951286316,
      "learning_rate": 9.870315789473684e-05,
      "loss": 0.4901,
      "step": 24940
    },
    {
      "epoch": 1.396628961809833,
      "grad_norm": 1.214342474937439,
      "learning_rate": 9.870289473684212e-05,
      "loss": 0.527,
      "step": 24941
    },
    {
      "epoch": 1.396684959121962,
      "grad_norm": 1.1204359531402588,
      "learning_rate": 9.870263157894736e-05,
      "loss": 0.6086,
      "step": 24942
    },
    {
      "epoch": 1.3967409564340911,
      "grad_norm": 1.6315031051635742,
      "learning_rate": 9.870236842105264e-05,
      "loss": 0.6003,
      "step": 24943
    },
    {
      "epoch": 1.3967969537462201,
      "grad_norm": 1.6887123584747314,
      "learning_rate": 9.87021052631579e-05,
      "loss": 0.6169,
      "step": 24944
    },
    {
      "epoch": 1.3968529510583492,
      "grad_norm": 1.3773730993270874,
      "learning_rate": 9.870184210526317e-05,
      "loss": 0.4426,
      "step": 24945
    },
    {
      "epoch": 1.3969089483704782,
      "grad_norm": 1.2887552976608276,
      "learning_rate": 9.870157894736843e-05,
      "loss": 0.5258,
      "step": 24946
    },
    {
      "epoch": 1.3969649456826072,
      "grad_norm": 1.3308465480804443,
      "learning_rate": 9.870131578947369e-05,
      "loss": 0.5726,
      "step": 24947
    },
    {
      "epoch": 1.3970209429947362,
      "grad_norm": 1.4514648914337158,
      "learning_rate": 9.870105263157895e-05,
      "loss": 0.4428,
      "step": 24948
    },
    {
      "epoch": 1.3970769403068652,
      "grad_norm": 1.5108258724212646,
      "learning_rate": 9.870078947368422e-05,
      "loss": 0.4245,
      "step": 24949
    },
    {
      "epoch": 1.3971329376189943,
      "grad_norm": 1.440600872039795,
      "learning_rate": 9.870052631578948e-05,
      "loss": 0.4508,
      "step": 24950
    },
    {
      "epoch": 1.3971889349311233,
      "grad_norm": 1.2561041116714478,
      "learning_rate": 9.870026315789474e-05,
      "loss": 0.4105,
      "step": 24951
    },
    {
      "epoch": 1.3972449322432523,
      "grad_norm": 1.3195409774780273,
      "learning_rate": 9.87e-05,
      "loss": 0.5741,
      "step": 24952
    },
    {
      "epoch": 1.3973009295553813,
      "grad_norm": 1.2621251344680786,
      "learning_rate": 9.869973684210527e-05,
      "loss": 0.42,
      "step": 24953
    },
    {
      "epoch": 1.3973569268675103,
      "grad_norm": 1.4393939971923828,
      "learning_rate": 9.869947368421053e-05,
      "loss": 0.4588,
      "step": 24954
    },
    {
      "epoch": 1.3974129241796394,
      "grad_norm": 1.3184726238250732,
      "learning_rate": 9.869921052631579e-05,
      "loss": 0.3996,
      "step": 24955
    },
    {
      "epoch": 1.3974689214917684,
      "grad_norm": 1.3019930124282837,
      "learning_rate": 9.869894736842105e-05,
      "loss": 0.35,
      "step": 24956
    },
    {
      "epoch": 1.3975249188038974,
      "grad_norm": 1.455165982246399,
      "learning_rate": 9.869868421052631e-05,
      "loss": 0.471,
      "step": 24957
    },
    {
      "epoch": 1.3975809161160264,
      "grad_norm": 1.0577287673950195,
      "learning_rate": 9.869842105263159e-05,
      "loss": 0.3745,
      "step": 24958
    },
    {
      "epoch": 1.3976369134281554,
      "grad_norm": 1.6369513273239136,
      "learning_rate": 9.869815789473685e-05,
      "loss": 0.5191,
      "step": 24959
    },
    {
      "epoch": 1.3976929107402845,
      "grad_norm": 1.1356538534164429,
      "learning_rate": 9.86978947368421e-05,
      "loss": 0.4098,
      "step": 24960
    },
    {
      "epoch": 1.3977489080524135,
      "grad_norm": 1.5345463752746582,
      "learning_rate": 9.869763157894736e-05,
      "loss": 0.5791,
      "step": 24961
    },
    {
      "epoch": 1.3978049053645425,
      "grad_norm": 1.3994766473770142,
      "learning_rate": 9.869736842105264e-05,
      "loss": 0.4134,
      "step": 24962
    },
    {
      "epoch": 1.3978609026766715,
      "grad_norm": 1.4844574928283691,
      "learning_rate": 9.86971052631579e-05,
      "loss": 0.7353,
      "step": 24963
    },
    {
      "epoch": 1.3979168999888005,
      "grad_norm": 1.3428765535354614,
      "learning_rate": 9.869684210526317e-05,
      "loss": 0.4492,
      "step": 24964
    },
    {
      "epoch": 1.3979728973009296,
      "grad_norm": 1.1608526706695557,
      "learning_rate": 9.869657894736842e-05,
      "loss": 0.3797,
      "step": 24965
    },
    {
      "epoch": 1.3980288946130586,
      "grad_norm": 1.1632158756256104,
      "learning_rate": 9.869631578947369e-05,
      "loss": 0.3693,
      "step": 24966
    },
    {
      "epoch": 1.3980848919251876,
      "grad_norm": 1.0819597244262695,
      "learning_rate": 9.869605263157895e-05,
      "loss": 0.3702,
      "step": 24967
    },
    {
      "epoch": 1.3981408892373166,
      "grad_norm": 1.1602689027786255,
      "learning_rate": 9.869578947368422e-05,
      "loss": 0.4737,
      "step": 24968
    },
    {
      "epoch": 1.3981968865494456,
      "grad_norm": 1.334148645401001,
      "learning_rate": 9.869552631578948e-05,
      "loss": 0.4082,
      "step": 24969
    },
    {
      "epoch": 1.3982528838615746,
      "grad_norm": 1.1840577125549316,
      "learning_rate": 9.869526315789474e-05,
      "loss": 0.4302,
      "step": 24970
    },
    {
      "epoch": 1.3983088811737037,
      "grad_norm": 1.401780366897583,
      "learning_rate": 9.8695e-05,
      "loss": 0.4981,
      "step": 24971
    },
    {
      "epoch": 1.3983648784858327,
      "grad_norm": 1.2232701778411865,
      "learning_rate": 9.869473684210528e-05,
      "loss": 0.3785,
      "step": 24972
    },
    {
      "epoch": 1.3984208757979617,
      "grad_norm": 1.3001359701156616,
      "learning_rate": 9.869447368421054e-05,
      "loss": 0.6799,
      "step": 24973
    },
    {
      "epoch": 1.3984768731100907,
      "grad_norm": 1.3686200380325317,
      "learning_rate": 9.86942105263158e-05,
      "loss": 0.4898,
      "step": 24974
    },
    {
      "epoch": 1.3985328704222197,
      "grad_norm": 1.2565200328826904,
      "learning_rate": 9.869394736842106e-05,
      "loss": 0.5439,
      "step": 24975
    },
    {
      "epoch": 1.3985888677343488,
      "grad_norm": 1.2554477453231812,
      "learning_rate": 9.869368421052632e-05,
      "loss": 0.3778,
      "step": 24976
    },
    {
      "epoch": 1.3986448650464778,
      "grad_norm": 1.2617875337600708,
      "learning_rate": 9.869342105263159e-05,
      "loss": 0.5034,
      "step": 24977
    },
    {
      "epoch": 1.3987008623586068,
      "grad_norm": 1.0165064334869385,
      "learning_rate": 9.869315789473685e-05,
      "loss": 0.3976,
      "step": 24978
    },
    {
      "epoch": 1.3987568596707358,
      "grad_norm": 1.1564257144927979,
      "learning_rate": 9.869289473684211e-05,
      "loss": 0.4893,
      "step": 24979
    },
    {
      "epoch": 1.3988128569828648,
      "grad_norm": 1.5127946138381958,
      "learning_rate": 9.869263157894737e-05,
      "loss": 0.6844,
      "step": 24980
    },
    {
      "epoch": 1.3988688542949939,
      "grad_norm": 1.4281575679779053,
      "learning_rate": 9.869236842105264e-05,
      "loss": 0.481,
      "step": 24981
    },
    {
      "epoch": 1.3989248516071229,
      "grad_norm": 1.6895654201507568,
      "learning_rate": 9.86921052631579e-05,
      "loss": 0.5923,
      "step": 24982
    },
    {
      "epoch": 1.398980848919252,
      "grad_norm": 1.4220174551010132,
      "learning_rate": 9.869184210526316e-05,
      "loss": 0.4944,
      "step": 24983
    },
    {
      "epoch": 1.399036846231381,
      "grad_norm": 1.6604970693588257,
      "learning_rate": 9.869157894736842e-05,
      "loss": 0.5302,
      "step": 24984
    },
    {
      "epoch": 1.39909284354351,
      "grad_norm": 1.1861413717269897,
      "learning_rate": 9.86913157894737e-05,
      "loss": 0.4556,
      "step": 24985
    },
    {
      "epoch": 1.399148840855639,
      "grad_norm": 1.10104238986969,
      "learning_rate": 9.869105263157895e-05,
      "loss": 0.3691,
      "step": 24986
    },
    {
      "epoch": 1.399204838167768,
      "grad_norm": 1.4999996423721313,
      "learning_rate": 9.869078947368423e-05,
      "loss": 0.488,
      "step": 24987
    },
    {
      "epoch": 1.399260835479897,
      "grad_norm": 1.4314191341400146,
      "learning_rate": 9.869052631578947e-05,
      "loss": 0.5801,
      "step": 24988
    },
    {
      "epoch": 1.399316832792026,
      "grad_norm": 1.188765287399292,
      "learning_rate": 9.869026315789475e-05,
      "loss": 0.4722,
      "step": 24989
    },
    {
      "epoch": 1.399372830104155,
      "grad_norm": 1.447745442390442,
      "learning_rate": 9.869e-05,
      "loss": 0.4674,
      "step": 24990
    },
    {
      "epoch": 1.399428827416284,
      "grad_norm": 1.2270760536193848,
      "learning_rate": 9.868973684210527e-05,
      "loss": 0.4121,
      "step": 24991
    },
    {
      "epoch": 1.399484824728413,
      "grad_norm": 1.1734915971755981,
      "learning_rate": 9.868947368421052e-05,
      "loss": 0.385,
      "step": 24992
    },
    {
      "epoch": 1.399540822040542,
      "grad_norm": 1.1888699531555176,
      "learning_rate": 9.868921052631578e-05,
      "loss": 0.3492,
      "step": 24993
    },
    {
      "epoch": 1.399596819352671,
      "grad_norm": 1.3761533498764038,
      "learning_rate": 9.868894736842106e-05,
      "loss": 0.5162,
      "step": 24994
    },
    {
      "epoch": 1.3996528166648001,
      "grad_norm": 1.2712526321411133,
      "learning_rate": 9.868868421052632e-05,
      "loss": 0.5014,
      "step": 24995
    },
    {
      "epoch": 1.3997088139769291,
      "grad_norm": 1.1590772867202759,
      "learning_rate": 9.868842105263159e-05,
      "loss": 0.3405,
      "step": 24996
    },
    {
      "epoch": 1.3997648112890582,
      "grad_norm": 1.3925637006759644,
      "learning_rate": 9.868815789473684e-05,
      "loss": 0.4855,
      "step": 24997
    },
    {
      "epoch": 1.3998208086011872,
      "grad_norm": 1.4906693696975708,
      "learning_rate": 9.868789473684211e-05,
      "loss": 0.4965,
      "step": 24998
    },
    {
      "epoch": 1.3998768059133162,
      "grad_norm": 1.0911304950714111,
      "learning_rate": 9.868763157894737e-05,
      "loss": 0.5819,
      "step": 24999
    },
    {
      "epoch": 1.3999328032254452,
      "grad_norm": 1.2345770597457886,
      "learning_rate": 9.868736842105264e-05,
      "loss": 0.3439,
      "step": 25000
    },
    {
      "epoch": 1.3999888005375742,
      "grad_norm": 1.4910013675689697,
      "learning_rate": 9.86871052631579e-05,
      "loss": 0.4444,
      "step": 25001
    },
    {
      "epoch": 1.4000447978497033,
      "grad_norm": 1.181877613067627,
      "learning_rate": 9.868684210526316e-05,
      "loss": 0.4475,
      "step": 25002
    },
    {
      "epoch": 1.4001007951618323,
      "grad_norm": 1.181572675704956,
      "learning_rate": 9.868657894736842e-05,
      "loss": 0.4454,
      "step": 25003
    },
    {
      "epoch": 1.4001567924739613,
      "grad_norm": 1.3123735189437866,
      "learning_rate": 9.86863157894737e-05,
      "loss": 0.4238,
      "step": 25004
    },
    {
      "epoch": 1.4002127897860903,
      "grad_norm": 1.2522852420806885,
      "learning_rate": 9.868605263157896e-05,
      "loss": 0.3972,
      "step": 25005
    },
    {
      "epoch": 1.4002687870982193,
      "grad_norm": 1.5129053592681885,
      "learning_rate": 9.868578947368422e-05,
      "loss": 0.4779,
      "step": 25006
    },
    {
      "epoch": 1.4003247844103484,
      "grad_norm": 1.5927486419677734,
      "learning_rate": 9.868552631578948e-05,
      "loss": 0.3409,
      "step": 25007
    },
    {
      "epoch": 1.4003807817224774,
      "grad_norm": 1.570186734199524,
      "learning_rate": 9.868526315789473e-05,
      "loss": 0.4116,
      "step": 25008
    },
    {
      "epoch": 1.4004367790346064,
      "grad_norm": 1.2253022193908691,
      "learning_rate": 9.868500000000001e-05,
      "loss": 0.4927,
      "step": 25009
    },
    {
      "epoch": 1.4004927763467354,
      "grad_norm": 1.3859150409698486,
      "learning_rate": 9.868473684210527e-05,
      "loss": 0.4001,
      "step": 25010
    },
    {
      "epoch": 1.4005487736588644,
      "grad_norm": 1.146719217300415,
      "learning_rate": 9.868447368421053e-05,
      "loss": 0.4135,
      "step": 25011
    },
    {
      "epoch": 1.4006047709709935,
      "grad_norm": 1.5594269037246704,
      "learning_rate": 9.868421052631579e-05,
      "loss": 0.4072,
      "step": 25012
    },
    {
      "epoch": 1.4006607682831225,
      "grad_norm": 1.3745996952056885,
      "learning_rate": 9.868394736842106e-05,
      "loss": 0.4967,
      "step": 25013
    },
    {
      "epoch": 1.4007167655952515,
      "grad_norm": 1.658186912536621,
      "learning_rate": 9.868368421052632e-05,
      "loss": 0.5801,
      "step": 25014
    },
    {
      "epoch": 1.4007727629073805,
      "grad_norm": 1.9151132106781006,
      "learning_rate": 9.868342105263158e-05,
      "loss": 0.5393,
      "step": 25015
    },
    {
      "epoch": 1.4008287602195095,
      "grad_norm": 1.4492130279541016,
      "learning_rate": 9.868315789473684e-05,
      "loss": 0.4651,
      "step": 25016
    },
    {
      "epoch": 1.4008847575316385,
      "grad_norm": 1.218334436416626,
      "learning_rate": 9.868289473684211e-05,
      "loss": 0.4098,
      "step": 25017
    },
    {
      "epoch": 1.4009407548437676,
      "grad_norm": 1.4384442567825317,
      "learning_rate": 9.868263157894737e-05,
      "loss": 0.558,
      "step": 25018
    },
    {
      "epoch": 1.4009967521558966,
      "grad_norm": 1.6621073484420776,
      "learning_rate": 9.868236842105265e-05,
      "loss": 0.438,
      "step": 25019
    },
    {
      "epoch": 1.4010527494680256,
      "grad_norm": 1.2724928855895996,
      "learning_rate": 9.868210526315789e-05,
      "loss": 0.416,
      "step": 25020
    },
    {
      "epoch": 1.4011087467801546,
      "grad_norm": 1.4699560403823853,
      "learning_rate": 9.868184210526317e-05,
      "loss": 0.4839,
      "step": 25021
    },
    {
      "epoch": 1.4011647440922836,
      "grad_norm": 1.3664757013320923,
      "learning_rate": 9.868157894736843e-05,
      "loss": 0.5102,
      "step": 25022
    },
    {
      "epoch": 1.4012207414044127,
      "grad_norm": 1.3542282581329346,
      "learning_rate": 9.86813157894737e-05,
      "loss": 0.5138,
      "step": 25023
    },
    {
      "epoch": 1.4012767387165417,
      "grad_norm": 1.2294542789459229,
      "learning_rate": 9.868105263157896e-05,
      "loss": 0.4108,
      "step": 25024
    },
    {
      "epoch": 1.4013327360286707,
      "grad_norm": 1.1522819995880127,
      "learning_rate": 9.86807894736842e-05,
      "loss": 0.4243,
      "step": 25025
    },
    {
      "epoch": 1.4013887333407997,
      "grad_norm": 1.4976519346237183,
      "learning_rate": 9.868052631578948e-05,
      "loss": 0.6372,
      "step": 25026
    },
    {
      "epoch": 1.4014447306529287,
      "grad_norm": 1.2452642917633057,
      "learning_rate": 9.868026315789474e-05,
      "loss": 0.5069,
      "step": 25027
    },
    {
      "epoch": 1.4015007279650578,
      "grad_norm": 1.0968621969223022,
      "learning_rate": 9.868000000000001e-05,
      "loss": 0.4141,
      "step": 25028
    },
    {
      "epoch": 1.4015567252771868,
      "grad_norm": 1.0446137189865112,
      "learning_rate": 9.867973684210527e-05,
      "loss": 0.3676,
      "step": 25029
    },
    {
      "epoch": 1.4016127225893158,
      "grad_norm": 1.1803224086761475,
      "learning_rate": 9.867947368421053e-05,
      "loss": 0.5365,
      "step": 25030
    },
    {
      "epoch": 1.4016687199014446,
      "grad_norm": 1.3734524250030518,
      "learning_rate": 9.867921052631579e-05,
      "loss": 0.542,
      "step": 25031
    },
    {
      "epoch": 1.4017247172135736,
      "grad_norm": 1.3984856605529785,
      "learning_rate": 9.867894736842106e-05,
      "loss": 0.4396,
      "step": 25032
    },
    {
      "epoch": 1.4017807145257026,
      "grad_norm": 1.6063979864120483,
      "learning_rate": 9.867868421052632e-05,
      "loss": 0.5612,
      "step": 25033
    },
    {
      "epoch": 1.4018367118378317,
      "grad_norm": 1.452295184135437,
      "learning_rate": 9.867842105263158e-05,
      "loss": 0.5969,
      "step": 25034
    },
    {
      "epoch": 1.4018927091499607,
      "grad_norm": 1.2554432153701782,
      "learning_rate": 9.867815789473684e-05,
      "loss": 0.4216,
      "step": 25035
    },
    {
      "epoch": 1.4019487064620897,
      "grad_norm": 1.2694385051727295,
      "learning_rate": 9.867789473684212e-05,
      "loss": 0.4662,
      "step": 25036
    },
    {
      "epoch": 1.4020047037742187,
      "grad_norm": 1.1114654541015625,
      "learning_rate": 9.867763157894738e-05,
      "loss": 0.3756,
      "step": 25037
    },
    {
      "epoch": 1.4020607010863477,
      "grad_norm": 3.559643268585205,
      "learning_rate": 9.867736842105264e-05,
      "loss": 0.4125,
      "step": 25038
    },
    {
      "epoch": 1.4021166983984767,
      "grad_norm": 1.172831416130066,
      "learning_rate": 9.86771052631579e-05,
      "loss": 0.357,
      "step": 25039
    },
    {
      "epoch": 1.4021726957106058,
      "grad_norm": 1.3581202030181885,
      "learning_rate": 9.867684210526317e-05,
      "loss": 0.5453,
      "step": 25040
    },
    {
      "epoch": 1.4022286930227348,
      "grad_norm": 1.535194993019104,
      "learning_rate": 9.867657894736843e-05,
      "loss": 0.4958,
      "step": 25041
    },
    {
      "epoch": 1.4022846903348638,
      "grad_norm": 1.432216763496399,
      "learning_rate": 9.86763157894737e-05,
      "loss": 0.4126,
      "step": 25042
    },
    {
      "epoch": 1.4023406876469928,
      "grad_norm": 1.2078896760940552,
      "learning_rate": 9.867605263157895e-05,
      "loss": 0.4448,
      "step": 25043
    },
    {
      "epoch": 1.4023966849591218,
      "grad_norm": 1.6733994483947754,
      "learning_rate": 9.867578947368421e-05,
      "loss": 0.506,
      "step": 25044
    },
    {
      "epoch": 1.4024526822712509,
      "grad_norm": 1.4144052267074585,
      "learning_rate": 9.867552631578948e-05,
      "loss": 0.4617,
      "step": 25045
    },
    {
      "epoch": 1.4025086795833799,
      "grad_norm": 1.3199208974838257,
      "learning_rate": 9.867526315789474e-05,
      "loss": 0.3546,
      "step": 25046
    },
    {
      "epoch": 1.402564676895509,
      "grad_norm": 1.1983124017715454,
      "learning_rate": 9.8675e-05,
      "loss": 0.4099,
      "step": 25047
    },
    {
      "epoch": 1.402620674207638,
      "grad_norm": 1.9280728101730347,
      "learning_rate": 9.867473684210526e-05,
      "loss": 0.4807,
      "step": 25048
    },
    {
      "epoch": 1.402676671519767,
      "grad_norm": 1.4171066284179688,
      "learning_rate": 9.867447368421053e-05,
      "loss": 0.4783,
      "step": 25049
    },
    {
      "epoch": 1.402732668831896,
      "grad_norm": 1.6370997428894043,
      "learning_rate": 9.867421052631579e-05,
      "loss": 0.4487,
      "step": 25050
    },
    {
      "epoch": 1.402788666144025,
      "grad_norm": 1.3617645502090454,
      "learning_rate": 9.867394736842107e-05,
      "loss": 0.3976,
      "step": 25051
    },
    {
      "epoch": 1.402844663456154,
      "grad_norm": 1.1175732612609863,
      "learning_rate": 9.867368421052631e-05,
      "loss": 0.3508,
      "step": 25052
    },
    {
      "epoch": 1.402900660768283,
      "grad_norm": 5.507682800292969,
      "learning_rate": 9.867342105263159e-05,
      "loss": 0.3978,
      "step": 25053
    },
    {
      "epoch": 1.402956658080412,
      "grad_norm": 1.6868797540664673,
      "learning_rate": 9.867315789473684e-05,
      "loss": 0.5949,
      "step": 25054
    },
    {
      "epoch": 1.403012655392541,
      "grad_norm": 1.537165641784668,
      "learning_rate": 9.867289473684212e-05,
      "loss": 0.4219,
      "step": 25055
    },
    {
      "epoch": 1.40306865270467,
      "grad_norm": 1.173524022102356,
      "learning_rate": 9.867263157894738e-05,
      "loss": 0.4213,
      "step": 25056
    },
    {
      "epoch": 1.403124650016799,
      "grad_norm": 1.24052095413208,
      "learning_rate": 9.867236842105264e-05,
      "loss": 0.4199,
      "step": 25057
    },
    {
      "epoch": 1.403180647328928,
      "grad_norm": 1.5992597341537476,
      "learning_rate": 9.86721052631579e-05,
      "loss": 0.5079,
      "step": 25058
    },
    {
      "epoch": 1.4032366446410571,
      "grad_norm": 1.393551230430603,
      "learning_rate": 9.867184210526317e-05,
      "loss": 0.4988,
      "step": 25059
    },
    {
      "epoch": 1.4032926419531861,
      "grad_norm": 1.587053894996643,
      "learning_rate": 9.867157894736843e-05,
      "loss": 0.5091,
      "step": 25060
    },
    {
      "epoch": 1.4033486392653152,
      "grad_norm": 1.4986960887908936,
      "learning_rate": 9.867131578947369e-05,
      "loss": 0.4199,
      "step": 25061
    },
    {
      "epoch": 1.4034046365774442,
      "grad_norm": 1.3810064792633057,
      "learning_rate": 9.867105263157895e-05,
      "loss": 0.3867,
      "step": 25062
    },
    {
      "epoch": 1.4034606338895732,
      "grad_norm": 1.9216339588165283,
      "learning_rate": 9.867078947368421e-05,
      "loss": 0.4258,
      "step": 25063
    },
    {
      "epoch": 1.4035166312017022,
      "grad_norm": 1.329543113708496,
      "learning_rate": 9.867052631578948e-05,
      "loss": 0.437,
      "step": 25064
    },
    {
      "epoch": 1.4035726285138312,
      "grad_norm": 1.2471803426742554,
      "learning_rate": 9.867026315789474e-05,
      "loss": 0.4124,
      "step": 25065
    },
    {
      "epoch": 1.4036286258259603,
      "grad_norm": 1.6240177154541016,
      "learning_rate": 9.867e-05,
      "loss": 0.5176,
      "step": 25066
    },
    {
      "epoch": 1.4036846231380893,
      "grad_norm": 1.3207478523254395,
      "learning_rate": 9.866973684210526e-05,
      "loss": 0.4693,
      "step": 25067
    },
    {
      "epoch": 1.4037406204502183,
      "grad_norm": 1.6427310705184937,
      "learning_rate": 9.866947368421054e-05,
      "loss": 0.6914,
      "step": 25068
    },
    {
      "epoch": 1.4037966177623473,
      "grad_norm": 1.5911667346954346,
      "learning_rate": 9.86692105263158e-05,
      "loss": 0.5381,
      "step": 25069
    },
    {
      "epoch": 1.4038526150744763,
      "grad_norm": 8.670450210571289,
      "learning_rate": 9.866894736842105e-05,
      "loss": 0.5282,
      "step": 25070
    },
    {
      "epoch": 1.4039086123866054,
      "grad_norm": 1.9461487531661987,
      "learning_rate": 9.866868421052631e-05,
      "loss": 0.4911,
      "step": 25071
    },
    {
      "epoch": 1.4039646096987344,
      "grad_norm": 1.3057644367218018,
      "learning_rate": 9.866842105263159e-05,
      "loss": 0.5616,
      "step": 25072
    },
    {
      "epoch": 1.4040206070108634,
      "grad_norm": 1.534982442855835,
      "learning_rate": 9.866815789473685e-05,
      "loss": 0.4504,
      "step": 25073
    },
    {
      "epoch": 1.4040766043229924,
      "grad_norm": 1.2518433332443237,
      "learning_rate": 9.866789473684212e-05,
      "loss": 0.5359,
      "step": 25074
    },
    {
      "epoch": 1.4041326016351214,
      "grad_norm": 6.048405170440674,
      "learning_rate": 9.866763157894737e-05,
      "loss": 0.4873,
      "step": 25075
    },
    {
      "epoch": 1.4041885989472505,
      "grad_norm": 1.257227897644043,
      "learning_rate": 9.866736842105264e-05,
      "loss": 0.4599,
      "step": 25076
    },
    {
      "epoch": 1.4042445962593795,
      "grad_norm": 1.189857006072998,
      "learning_rate": 9.86671052631579e-05,
      "loss": 0.4897,
      "step": 25077
    },
    {
      "epoch": 1.4043005935715085,
      "grad_norm": 1.6713621616363525,
      "learning_rate": 9.866684210526316e-05,
      "loss": 0.5393,
      "step": 25078
    },
    {
      "epoch": 1.4043565908836375,
      "grad_norm": 1.223427176475525,
      "learning_rate": 9.866657894736843e-05,
      "loss": 0.4036,
      "step": 25079
    },
    {
      "epoch": 1.4044125881957665,
      "grad_norm": 1.4842644929885864,
      "learning_rate": 9.866631578947368e-05,
      "loss": 0.4918,
      "step": 25080
    },
    {
      "epoch": 1.4044685855078956,
      "grad_norm": 1.2758557796478271,
      "learning_rate": 9.866605263157895e-05,
      "loss": 0.417,
      "step": 25081
    },
    {
      "epoch": 1.4045245828200246,
      "grad_norm": 1.3707480430603027,
      "learning_rate": 9.866578947368421e-05,
      "loss": 0.4824,
      "step": 25082
    },
    {
      "epoch": 1.4045805801321536,
      "grad_norm": 1.6210976839065552,
      "learning_rate": 9.866552631578949e-05,
      "loss": 0.4343,
      "step": 25083
    },
    {
      "epoch": 1.4046365774442826,
      "grad_norm": 1.2979321479797363,
      "learning_rate": 9.866526315789475e-05,
      "loss": 0.3599,
      "step": 25084
    },
    {
      "epoch": 1.4046925747564116,
      "grad_norm": 1.4203746318817139,
      "learning_rate": 9.8665e-05,
      "loss": 0.4211,
      "step": 25085
    },
    {
      "epoch": 1.4047485720685406,
      "grad_norm": 1.2466440200805664,
      "learning_rate": 9.866473684210526e-05,
      "loss": 0.3942,
      "step": 25086
    },
    {
      "epoch": 1.4048045693806697,
      "grad_norm": 1.479690432548523,
      "learning_rate": 9.866447368421054e-05,
      "loss": 0.4823,
      "step": 25087
    },
    {
      "epoch": 1.4048605666927987,
      "grad_norm": 1.480630874633789,
      "learning_rate": 9.86642105263158e-05,
      "loss": 0.4906,
      "step": 25088
    },
    {
      "epoch": 1.4049165640049277,
      "grad_norm": 1.1486139297485352,
      "learning_rate": 9.866394736842106e-05,
      "loss": 0.4376,
      "step": 25089
    },
    {
      "epoch": 1.4049725613170567,
      "grad_norm": 1.2108925580978394,
      "learning_rate": 9.866368421052632e-05,
      "loss": 0.4693,
      "step": 25090
    },
    {
      "epoch": 1.4050285586291857,
      "grad_norm": 1.401533603668213,
      "learning_rate": 9.866342105263159e-05,
      "loss": 0.4328,
      "step": 25091
    },
    {
      "epoch": 1.4050845559413148,
      "grad_norm": 1.4225749969482422,
      "learning_rate": 9.866315789473685e-05,
      "loss": 0.5919,
      "step": 25092
    },
    {
      "epoch": 1.4051405532534438,
      "grad_norm": 1.5397193431854248,
      "learning_rate": 9.866289473684211e-05,
      "loss": 0.5583,
      "step": 25093
    },
    {
      "epoch": 1.4051965505655728,
      "grad_norm": 1.6258461475372314,
      "learning_rate": 9.866263157894737e-05,
      "loss": 0.5521,
      "step": 25094
    },
    {
      "epoch": 1.4052525478777018,
      "grad_norm": 1.1896156072616577,
      "learning_rate": 9.866236842105263e-05,
      "loss": 0.3568,
      "step": 25095
    },
    {
      "epoch": 1.4053085451898308,
      "grad_norm": 1.330722689628601,
      "learning_rate": 9.86621052631579e-05,
      "loss": 0.418,
      "step": 25096
    },
    {
      "epoch": 1.4053645425019599,
      "grad_norm": 1.491829514503479,
      "learning_rate": 9.866184210526316e-05,
      "loss": 0.4913,
      "step": 25097
    },
    {
      "epoch": 1.4054205398140889,
      "grad_norm": 1.9061870574951172,
      "learning_rate": 9.866157894736842e-05,
      "loss": 0.6653,
      "step": 25098
    },
    {
      "epoch": 1.405476537126218,
      "grad_norm": 1.0821285247802734,
      "learning_rate": 9.866131578947368e-05,
      "loss": 0.3845,
      "step": 25099
    },
    {
      "epoch": 1.405532534438347,
      "grad_norm": 1.4515200853347778,
      "learning_rate": 9.866105263157895e-05,
      "loss": 0.4841,
      "step": 25100
    },
    {
      "epoch": 1.405588531750476,
      "grad_norm": 1.3448429107666016,
      "learning_rate": 9.866078947368421e-05,
      "loss": 0.3733,
      "step": 25101
    },
    {
      "epoch": 1.405644529062605,
      "grad_norm": 1.5497682094573975,
      "learning_rate": 9.866052631578947e-05,
      "loss": 0.7214,
      "step": 25102
    },
    {
      "epoch": 1.405700526374734,
      "grad_norm": 1.6154425144195557,
      "learning_rate": 9.866026315789473e-05,
      "loss": 0.4119,
      "step": 25103
    },
    {
      "epoch": 1.405756523686863,
      "grad_norm": 1.3904379606246948,
      "learning_rate": 9.866000000000001e-05,
      "loss": 0.3991,
      "step": 25104
    },
    {
      "epoch": 1.405812520998992,
      "grad_norm": 1.2952815294265747,
      "learning_rate": 9.865973684210527e-05,
      "loss": 0.4578,
      "step": 25105
    },
    {
      "epoch": 1.405868518311121,
      "grad_norm": 1.5114688873291016,
      "learning_rate": 9.865947368421054e-05,
      "loss": 0.4603,
      "step": 25106
    },
    {
      "epoch": 1.40592451562325,
      "grad_norm": 1.2152762413024902,
      "learning_rate": 9.865921052631579e-05,
      "loss": 0.505,
      "step": 25107
    },
    {
      "epoch": 1.405980512935379,
      "grad_norm": 1.5251792669296265,
      "learning_rate": 9.865894736842106e-05,
      "loss": 0.4445,
      "step": 25108
    },
    {
      "epoch": 1.406036510247508,
      "grad_norm": 1.2245151996612549,
      "learning_rate": 9.865868421052632e-05,
      "loss": 0.4804,
      "step": 25109
    },
    {
      "epoch": 1.406092507559637,
      "grad_norm": 1.192751169204712,
      "learning_rate": 9.865842105263159e-05,
      "loss": 0.4295,
      "step": 25110
    },
    {
      "epoch": 1.4061485048717661,
      "grad_norm": 1.083000659942627,
      "learning_rate": 9.865815789473685e-05,
      "loss": 0.4174,
      "step": 25111
    },
    {
      "epoch": 1.4062045021838951,
      "grad_norm": 1.5058971643447876,
      "learning_rate": 9.86578947368421e-05,
      "loss": 0.4594,
      "step": 25112
    },
    {
      "epoch": 1.4062604994960242,
      "grad_norm": 1.3602256774902344,
      "learning_rate": 9.865763157894737e-05,
      "loss": 0.5681,
      "step": 25113
    },
    {
      "epoch": 1.4063164968081532,
      "grad_norm": 1.3668222427368164,
      "learning_rate": 9.865736842105263e-05,
      "loss": 0.4684,
      "step": 25114
    },
    {
      "epoch": 1.4063724941202822,
      "grad_norm": 1.3513565063476562,
      "learning_rate": 9.86571052631579e-05,
      "loss": 0.4067,
      "step": 25115
    },
    {
      "epoch": 1.4064284914324112,
      "grad_norm": 1.42878258228302,
      "learning_rate": 9.865684210526316e-05,
      "loss": 0.5119,
      "step": 25116
    },
    {
      "epoch": 1.4064844887445402,
      "grad_norm": 1.2328746318817139,
      "learning_rate": 9.865657894736842e-05,
      "loss": 0.49,
      "step": 25117
    },
    {
      "epoch": 1.4065404860566693,
      "grad_norm": 1.290832281112671,
      "learning_rate": 9.865631578947368e-05,
      "loss": 0.4261,
      "step": 25118
    },
    {
      "epoch": 1.4065964833687983,
      "grad_norm": 1.6286808252334595,
      "learning_rate": 9.865605263157896e-05,
      "loss": 0.548,
      "step": 25119
    },
    {
      "epoch": 1.4066524806809273,
      "grad_norm": 1.1603643894195557,
      "learning_rate": 9.865578947368422e-05,
      "loss": 0.4343,
      "step": 25120
    },
    {
      "epoch": 1.4067084779930563,
      "grad_norm": 1.461963415145874,
      "learning_rate": 9.865552631578948e-05,
      "loss": 0.5455,
      "step": 25121
    },
    {
      "epoch": 1.4067644753051853,
      "grad_norm": 1.9852609634399414,
      "learning_rate": 9.865526315789474e-05,
      "loss": 0.4646,
      "step": 25122
    },
    {
      "epoch": 1.4068204726173144,
      "grad_norm": 1.3366806507110596,
      "learning_rate": 9.865500000000001e-05,
      "loss": 0.4645,
      "step": 25123
    },
    {
      "epoch": 1.4068764699294434,
      "grad_norm": 1.3001078367233276,
      "learning_rate": 9.865473684210527e-05,
      "loss": 0.466,
      "step": 25124
    },
    {
      "epoch": 1.4069324672415724,
      "grad_norm": 1.2729926109313965,
      "learning_rate": 9.865447368421053e-05,
      "loss": 0.4699,
      "step": 25125
    },
    {
      "epoch": 1.4069884645537014,
      "grad_norm": 1.61940336227417,
      "learning_rate": 9.865421052631579e-05,
      "loss": 0.6571,
      "step": 25126
    },
    {
      "epoch": 1.4070444618658304,
      "grad_norm": 1.4772520065307617,
      "learning_rate": 9.865394736842106e-05,
      "loss": 0.5843,
      "step": 25127
    },
    {
      "epoch": 1.4071004591779595,
      "grad_norm": 1.5451966524124146,
      "learning_rate": 9.865368421052632e-05,
      "loss": 0.5362,
      "step": 25128
    },
    {
      "epoch": 1.4071564564900885,
      "grad_norm": 1.266040325164795,
      "learning_rate": 9.865342105263158e-05,
      "loss": 0.4418,
      "step": 25129
    },
    {
      "epoch": 1.4072124538022175,
      "grad_norm": 1.3463425636291504,
      "learning_rate": 9.865315789473684e-05,
      "loss": 0.4314,
      "step": 25130
    },
    {
      "epoch": 1.4072684511143465,
      "grad_norm": 1.5727450847625732,
      "learning_rate": 9.86528947368421e-05,
      "loss": 0.5435,
      "step": 25131
    },
    {
      "epoch": 1.4073244484264755,
      "grad_norm": 1.3411790132522583,
      "learning_rate": 9.865263157894737e-05,
      "loss": 0.4142,
      "step": 25132
    },
    {
      "epoch": 1.4073804457386045,
      "grad_norm": 1.4852484464645386,
      "learning_rate": 9.865236842105263e-05,
      "loss": 0.4494,
      "step": 25133
    },
    {
      "epoch": 1.4074364430507336,
      "grad_norm": 1.394225001335144,
      "learning_rate": 9.865210526315791e-05,
      "loss": 0.5541,
      "step": 25134
    },
    {
      "epoch": 1.4074924403628626,
      "grad_norm": 1.3585059642791748,
      "learning_rate": 9.865184210526315e-05,
      "loss": 0.3974,
      "step": 25135
    },
    {
      "epoch": 1.4075484376749916,
      "grad_norm": 1.5817770957946777,
      "learning_rate": 9.865157894736843e-05,
      "loss": 0.5899,
      "step": 25136
    },
    {
      "epoch": 1.4076044349871206,
      "grad_norm": 1.5890610218048096,
      "learning_rate": 9.865131578947369e-05,
      "loss": 0.5236,
      "step": 25137
    },
    {
      "epoch": 1.4076604322992496,
      "grad_norm": 1.1551625728607178,
      "learning_rate": 9.865105263157896e-05,
      "loss": 0.4185,
      "step": 25138
    },
    {
      "epoch": 1.4077164296113787,
      "grad_norm": 1.1804637908935547,
      "learning_rate": 9.86507894736842e-05,
      "loss": 0.4683,
      "step": 25139
    },
    {
      "epoch": 1.4077724269235077,
      "grad_norm": 1.4861499071121216,
      "learning_rate": 9.865052631578948e-05,
      "loss": 0.4231,
      "step": 25140
    },
    {
      "epoch": 1.4078284242356367,
      "grad_norm": 1.6176575422286987,
      "learning_rate": 9.865026315789474e-05,
      "loss": 0.4366,
      "step": 25141
    },
    {
      "epoch": 1.4078844215477657,
      "grad_norm": 1.3628233671188354,
      "learning_rate": 9.865000000000001e-05,
      "loss": 0.453,
      "step": 25142
    },
    {
      "epoch": 1.4079404188598947,
      "grad_norm": 1.26882004737854,
      "learning_rate": 9.864973684210527e-05,
      "loss": 0.4369,
      "step": 25143
    },
    {
      "epoch": 1.4079964161720238,
      "grad_norm": 1.4862089157104492,
      "learning_rate": 9.864947368421053e-05,
      "loss": 0.3508,
      "step": 25144
    },
    {
      "epoch": 1.4080524134841528,
      "grad_norm": 1.1362391710281372,
      "learning_rate": 9.864921052631579e-05,
      "loss": 0.3829,
      "step": 25145
    },
    {
      "epoch": 1.4081084107962818,
      "grad_norm": 1.2971348762512207,
      "learning_rate": 9.864894736842105e-05,
      "loss": 0.4346,
      "step": 25146
    },
    {
      "epoch": 1.4081644081084108,
      "grad_norm": 1.207412600517273,
      "learning_rate": 9.864868421052632e-05,
      "loss": 0.4799,
      "step": 25147
    },
    {
      "epoch": 1.4082204054205398,
      "grad_norm": 1.363423466682434,
      "learning_rate": 9.864842105263158e-05,
      "loss": 0.4393,
      "step": 25148
    },
    {
      "epoch": 1.4082764027326689,
      "grad_norm": 1.479589581489563,
      "learning_rate": 9.864815789473684e-05,
      "loss": 0.3937,
      "step": 25149
    },
    {
      "epoch": 1.4083324000447979,
      "grad_norm": 1.668657660484314,
      "learning_rate": 9.86478947368421e-05,
      "loss": 0.5922,
      "step": 25150
    },
    {
      "epoch": 1.408388397356927,
      "grad_norm": 1.1311315298080444,
      "learning_rate": 9.864763157894738e-05,
      "loss": 0.392,
      "step": 25151
    },
    {
      "epoch": 1.408444394669056,
      "grad_norm": 2.721068859100342,
      "learning_rate": 9.864736842105264e-05,
      "loss": 0.6064,
      "step": 25152
    },
    {
      "epoch": 1.408500391981185,
      "grad_norm": 1.382392168045044,
      "learning_rate": 9.86471052631579e-05,
      "loss": 0.4802,
      "step": 25153
    },
    {
      "epoch": 1.408556389293314,
      "grad_norm": 1.329283356666565,
      "learning_rate": 9.864684210526316e-05,
      "loss": 0.3837,
      "step": 25154
    },
    {
      "epoch": 1.408612386605443,
      "grad_norm": 1.3018931150436401,
      "learning_rate": 9.864657894736843e-05,
      "loss": 0.3788,
      "step": 25155
    },
    {
      "epoch": 1.408668383917572,
      "grad_norm": 1.6913831233978271,
      "learning_rate": 9.864631578947369e-05,
      "loss": 0.4156,
      "step": 25156
    },
    {
      "epoch": 1.408724381229701,
      "grad_norm": 1.326263427734375,
      "learning_rate": 9.864605263157895e-05,
      "loss": 0.4448,
      "step": 25157
    },
    {
      "epoch": 1.40878037854183,
      "grad_norm": 1.895554780960083,
      "learning_rate": 9.864578947368421e-05,
      "loss": 0.4293,
      "step": 25158
    },
    {
      "epoch": 1.408836375853959,
      "grad_norm": 1.3707953691482544,
      "learning_rate": 9.864552631578948e-05,
      "loss": 0.4818,
      "step": 25159
    },
    {
      "epoch": 1.408892373166088,
      "grad_norm": 1.4448717832565308,
      "learning_rate": 9.864526315789474e-05,
      "loss": 0.55,
      "step": 25160
    },
    {
      "epoch": 1.408948370478217,
      "grad_norm": 1.3688149452209473,
      "learning_rate": 9.864500000000002e-05,
      "loss": 0.6002,
      "step": 25161
    },
    {
      "epoch": 1.409004367790346,
      "grad_norm": 1.3279937505722046,
      "learning_rate": 9.864473684210526e-05,
      "loss": 0.5392,
      "step": 25162
    },
    {
      "epoch": 1.4090603651024751,
      "grad_norm": 1.3392213582992554,
      "learning_rate": 9.864447368421053e-05,
      "loss": 0.3843,
      "step": 25163
    },
    {
      "epoch": 1.4091163624146041,
      "grad_norm": 1.4783343076705933,
      "learning_rate": 9.86442105263158e-05,
      "loss": 0.4332,
      "step": 25164
    },
    {
      "epoch": 1.4091723597267332,
      "grad_norm": 1.4156527519226074,
      "learning_rate": 9.864394736842105e-05,
      "loss": 0.421,
      "step": 25165
    },
    {
      "epoch": 1.4092283570388622,
      "grad_norm": 1.6359426975250244,
      "learning_rate": 9.864368421052633e-05,
      "loss": 0.5679,
      "step": 25166
    },
    {
      "epoch": 1.4092843543509912,
      "grad_norm": 1.2974162101745605,
      "learning_rate": 9.864342105263157e-05,
      "loss": 0.3891,
      "step": 25167
    },
    {
      "epoch": 1.4093403516631202,
      "grad_norm": 1.4267315864562988,
      "learning_rate": 9.864315789473685e-05,
      "loss": 0.4329,
      "step": 25168
    },
    {
      "epoch": 1.4093963489752492,
      "grad_norm": 1.202324390411377,
      "learning_rate": 9.86428947368421e-05,
      "loss": 0.3673,
      "step": 25169
    },
    {
      "epoch": 1.4094523462873783,
      "grad_norm": 1.3575880527496338,
      "learning_rate": 9.864263157894738e-05,
      "loss": 0.4658,
      "step": 25170
    },
    {
      "epoch": 1.4095083435995073,
      "grad_norm": 1.1915929317474365,
      "learning_rate": 9.864236842105264e-05,
      "loss": 0.346,
      "step": 25171
    },
    {
      "epoch": 1.4095643409116363,
      "grad_norm": 1.1185379028320312,
      "learning_rate": 9.86421052631579e-05,
      "loss": 0.3682,
      "step": 25172
    },
    {
      "epoch": 1.4096203382237653,
      "grad_norm": 3.813021183013916,
      "learning_rate": 9.864184210526316e-05,
      "loss": 0.6769,
      "step": 25173
    },
    {
      "epoch": 1.4096763355358943,
      "grad_norm": 1.7576009035110474,
      "learning_rate": 9.864157894736843e-05,
      "loss": 0.5629,
      "step": 25174
    },
    {
      "epoch": 1.4097323328480234,
      "grad_norm": 1.261063575744629,
      "learning_rate": 9.864131578947369e-05,
      "loss": 0.4364,
      "step": 25175
    },
    {
      "epoch": 1.4097883301601524,
      "grad_norm": 1.2424596548080444,
      "learning_rate": 9.864105263157895e-05,
      "loss": 0.4822,
      "step": 25176
    },
    {
      "epoch": 1.4098443274722814,
      "grad_norm": 1.6005308628082275,
      "learning_rate": 9.864078947368421e-05,
      "loss": 0.4941,
      "step": 25177
    },
    {
      "epoch": 1.4099003247844104,
      "grad_norm": 1.1510900259017944,
      "learning_rate": 9.864052631578948e-05,
      "loss": 0.477,
      "step": 25178
    },
    {
      "epoch": 1.4099563220965394,
      "grad_norm": 1.8903448581695557,
      "learning_rate": 9.864026315789474e-05,
      "loss": 0.4593,
      "step": 25179
    },
    {
      "epoch": 1.4100123194086684,
      "grad_norm": 2.102847099304199,
      "learning_rate": 9.864e-05,
      "loss": 0.4552,
      "step": 25180
    },
    {
      "epoch": 1.4100683167207975,
      "grad_norm": 1.7312819957733154,
      "learning_rate": 9.863973684210526e-05,
      "loss": 0.4896,
      "step": 25181
    },
    {
      "epoch": 1.4101243140329265,
      "grad_norm": 1.3706915378570557,
      "learning_rate": 9.863947368421052e-05,
      "loss": 0.3614,
      "step": 25182
    },
    {
      "epoch": 1.4101803113450555,
      "grad_norm": 1.6786187887191772,
      "learning_rate": 9.86392105263158e-05,
      "loss": 0.5758,
      "step": 25183
    },
    {
      "epoch": 1.4102363086571845,
      "grad_norm": 1.2910107374191284,
      "learning_rate": 9.863894736842106e-05,
      "loss": 0.4112,
      "step": 25184
    },
    {
      "epoch": 1.4102923059693135,
      "grad_norm": 1.2806596755981445,
      "learning_rate": 9.863868421052632e-05,
      "loss": 0.4152,
      "step": 25185
    },
    {
      "epoch": 1.4103483032814426,
      "grad_norm": 1.6166380643844604,
      "learning_rate": 9.863842105263158e-05,
      "loss": 0.4378,
      "step": 25186
    },
    {
      "epoch": 1.4104043005935716,
      "grad_norm": 1.2381101846694946,
      "learning_rate": 9.863815789473685e-05,
      "loss": 0.3784,
      "step": 25187
    },
    {
      "epoch": 1.4104602979057006,
      "grad_norm": 1.633828043937683,
      "learning_rate": 9.863789473684211e-05,
      "loss": 0.4597,
      "step": 25188
    },
    {
      "epoch": 1.4105162952178296,
      "grad_norm": 1.0757710933685303,
      "learning_rate": 9.863763157894738e-05,
      "loss": 0.4191,
      "step": 25189
    },
    {
      "epoch": 1.4105722925299586,
      "grad_norm": 1.5068286657333374,
      "learning_rate": 9.863736842105263e-05,
      "loss": 0.452,
      "step": 25190
    },
    {
      "epoch": 1.4106282898420877,
      "grad_norm": 1.1757184267044067,
      "learning_rate": 9.86371052631579e-05,
      "loss": 0.3778,
      "step": 25191
    },
    {
      "epoch": 1.4106842871542167,
      "grad_norm": 1.402175784111023,
      "learning_rate": 9.863684210526316e-05,
      "loss": 0.7289,
      "step": 25192
    },
    {
      "epoch": 1.4107402844663457,
      "grad_norm": 1.231673240661621,
      "learning_rate": 9.863657894736843e-05,
      "loss": 0.4304,
      "step": 25193
    },
    {
      "epoch": 1.4107962817784747,
      "grad_norm": 1.4890073537826538,
      "learning_rate": 9.863631578947368e-05,
      "loss": 0.4979,
      "step": 25194
    },
    {
      "epoch": 1.4108522790906037,
      "grad_norm": 1.2157163619995117,
      "learning_rate": 9.863605263157895e-05,
      "loss": 0.4838,
      "step": 25195
    },
    {
      "epoch": 1.4109082764027328,
      "grad_norm": 1.2294938564300537,
      "learning_rate": 9.863578947368421e-05,
      "loss": 0.5075,
      "step": 25196
    },
    {
      "epoch": 1.4109642737148618,
      "grad_norm": 2.9420714378356934,
      "learning_rate": 9.863552631578949e-05,
      "loss": 0.4397,
      "step": 25197
    },
    {
      "epoch": 1.4110202710269908,
      "grad_norm": 1.3563071489334106,
      "learning_rate": 9.863526315789475e-05,
      "loss": 0.5007,
      "step": 25198
    },
    {
      "epoch": 1.4110762683391198,
      "grad_norm": 1.259868860244751,
      "learning_rate": 9.8635e-05,
      "loss": 0.6522,
      "step": 25199
    },
    {
      "epoch": 1.4111322656512488,
      "grad_norm": 1.3337322473526,
      "learning_rate": 9.863473684210527e-05,
      "loss": 0.4302,
      "step": 25200
    },
    {
      "epoch": 1.4111882629633778,
      "grad_norm": 1.2552626132965088,
      "learning_rate": 9.863447368421053e-05,
      "loss": 0.4105,
      "step": 25201
    },
    {
      "epoch": 1.4112442602755069,
      "grad_norm": 1.2617647647857666,
      "learning_rate": 9.86342105263158e-05,
      "loss": 0.3968,
      "step": 25202
    },
    {
      "epoch": 1.4113002575876359,
      "grad_norm": 1.4309154748916626,
      "learning_rate": 9.863394736842106e-05,
      "loss": 0.4875,
      "step": 25203
    },
    {
      "epoch": 1.411356254899765,
      "grad_norm": 1.6625126600265503,
      "learning_rate": 9.863368421052632e-05,
      "loss": 0.6625,
      "step": 25204
    },
    {
      "epoch": 1.411412252211894,
      "grad_norm": 1.1364010572433472,
      "learning_rate": 9.863342105263158e-05,
      "loss": 0.3535,
      "step": 25205
    },
    {
      "epoch": 1.411468249524023,
      "grad_norm": 1.4304853677749634,
      "learning_rate": 9.863315789473685e-05,
      "loss": 0.5349,
      "step": 25206
    },
    {
      "epoch": 1.411524246836152,
      "grad_norm": 1.3051401376724243,
      "learning_rate": 9.863289473684211e-05,
      "loss": 0.4308,
      "step": 25207
    },
    {
      "epoch": 1.411580244148281,
      "grad_norm": 1.8065649271011353,
      "learning_rate": 9.863263157894737e-05,
      "loss": 0.567,
      "step": 25208
    },
    {
      "epoch": 1.41163624146041,
      "grad_norm": 1.2617363929748535,
      "learning_rate": 9.863236842105263e-05,
      "loss": 0.5085,
      "step": 25209
    },
    {
      "epoch": 1.411692238772539,
      "grad_norm": 1.63270103931427,
      "learning_rate": 9.86321052631579e-05,
      "loss": 0.5207,
      "step": 25210
    },
    {
      "epoch": 1.411748236084668,
      "grad_norm": 1.464239478111267,
      "learning_rate": 9.863184210526316e-05,
      "loss": 0.4098,
      "step": 25211
    },
    {
      "epoch": 1.411804233396797,
      "grad_norm": 1.193290114402771,
      "learning_rate": 9.863157894736842e-05,
      "loss": 0.3687,
      "step": 25212
    },
    {
      "epoch": 1.411860230708926,
      "grad_norm": 1.1258714199066162,
      "learning_rate": 9.863131578947368e-05,
      "loss": 0.4189,
      "step": 25213
    },
    {
      "epoch": 1.411916228021055,
      "grad_norm": 1.2496358156204224,
      "learning_rate": 9.863105263157896e-05,
      "loss": 0.5213,
      "step": 25214
    },
    {
      "epoch": 1.4119722253331841,
      "grad_norm": 1.1736644506454468,
      "learning_rate": 9.863078947368422e-05,
      "loss": 0.5512,
      "step": 25215
    },
    {
      "epoch": 1.4120282226453131,
      "grad_norm": 1.31013023853302,
      "learning_rate": 9.863052631578948e-05,
      "loss": 0.4178,
      "step": 25216
    },
    {
      "epoch": 1.4120842199574422,
      "grad_norm": 1.6401429176330566,
      "learning_rate": 9.863026315789474e-05,
      "loss": 0.4042,
      "step": 25217
    },
    {
      "epoch": 1.4121402172695712,
      "grad_norm": 1.447288990020752,
      "learning_rate": 9.863e-05,
      "loss": 0.3885,
      "step": 25218
    },
    {
      "epoch": 1.4121962145817002,
      "grad_norm": 1.3267236948013306,
      "learning_rate": 9.862973684210527e-05,
      "loss": 0.6762,
      "step": 25219
    },
    {
      "epoch": 1.4122522118938292,
      "grad_norm": 1.245842456817627,
      "learning_rate": 9.862947368421053e-05,
      "loss": 0.6838,
      "step": 25220
    },
    {
      "epoch": 1.4123082092059582,
      "grad_norm": 1.2799817323684692,
      "learning_rate": 9.86292105263158e-05,
      "loss": 0.3342,
      "step": 25221
    },
    {
      "epoch": 1.4123642065180873,
      "grad_norm": 1.138277292251587,
      "learning_rate": 9.862894736842105e-05,
      "loss": 0.4473,
      "step": 25222
    },
    {
      "epoch": 1.4124202038302163,
      "grad_norm": 1.3019230365753174,
      "learning_rate": 9.862868421052632e-05,
      "loss": 0.4225,
      "step": 25223
    },
    {
      "epoch": 1.4124762011423453,
      "grad_norm": 1.2484922409057617,
      "learning_rate": 9.862842105263158e-05,
      "loss": 0.5595,
      "step": 25224
    },
    {
      "epoch": 1.4125321984544743,
      "grad_norm": 1.2911787033081055,
      "learning_rate": 9.862815789473685e-05,
      "loss": 0.5246,
      "step": 25225
    },
    {
      "epoch": 1.4125881957666033,
      "grad_norm": 1.1501758098602295,
      "learning_rate": 9.862789473684211e-05,
      "loss": 0.3478,
      "step": 25226
    },
    {
      "epoch": 1.4126441930787323,
      "grad_norm": 1.0365339517593384,
      "learning_rate": 9.862763157894737e-05,
      "loss": 0.3588,
      "step": 25227
    },
    {
      "epoch": 1.4127001903908614,
      "grad_norm": 1.8831053972244263,
      "learning_rate": 9.862736842105263e-05,
      "loss": 0.4659,
      "step": 25228
    },
    {
      "epoch": 1.4127561877029904,
      "grad_norm": 1.3875123262405396,
      "learning_rate": 9.862710526315791e-05,
      "loss": 0.4413,
      "step": 25229
    },
    {
      "epoch": 1.4128121850151194,
      "grad_norm": 1.4171642065048218,
      "learning_rate": 9.862684210526317e-05,
      "loss": 0.5081,
      "step": 25230
    },
    {
      "epoch": 1.4128681823272484,
      "grad_norm": 1.3800628185272217,
      "learning_rate": 9.862657894736843e-05,
      "loss": 0.3952,
      "step": 25231
    },
    {
      "epoch": 1.4129241796393774,
      "grad_norm": 1.1604880094528198,
      "learning_rate": 9.862631578947369e-05,
      "loss": 0.3431,
      "step": 25232
    },
    {
      "epoch": 1.4129801769515065,
      "grad_norm": 1.3972384929656982,
      "learning_rate": 9.862605263157895e-05,
      "loss": 0.5271,
      "step": 25233
    },
    {
      "epoch": 1.4130361742636355,
      "grad_norm": 1.6486868858337402,
      "learning_rate": 9.862578947368422e-05,
      "loss": 0.5212,
      "step": 25234
    },
    {
      "epoch": 1.4130921715757645,
      "grad_norm": 1.3751742839813232,
      "learning_rate": 9.862552631578948e-05,
      "loss": 0.3862,
      "step": 25235
    },
    {
      "epoch": 1.4131481688878935,
      "grad_norm": 1.791991949081421,
      "learning_rate": 9.862526315789474e-05,
      "loss": 0.5196,
      "step": 25236
    },
    {
      "epoch": 1.4132041662000223,
      "grad_norm": 1.3346039056777954,
      "learning_rate": 9.8625e-05,
      "loss": 0.5303,
      "step": 25237
    },
    {
      "epoch": 1.4132601635121513,
      "grad_norm": 1.5640456676483154,
      "learning_rate": 9.862473684210527e-05,
      "loss": 0.4898,
      "step": 25238
    },
    {
      "epoch": 1.4133161608242804,
      "grad_norm": 1.4504514932632446,
      "learning_rate": 9.862447368421053e-05,
      "loss": 0.4962,
      "step": 25239
    },
    {
      "epoch": 1.4133721581364094,
      "grad_norm": 1.5084350109100342,
      "learning_rate": 9.862421052631579e-05,
      "loss": 0.6098,
      "step": 25240
    },
    {
      "epoch": 1.4134281554485384,
      "grad_norm": 1.341499924659729,
      "learning_rate": 9.862394736842105e-05,
      "loss": 0.6482,
      "step": 25241
    },
    {
      "epoch": 1.4134841527606674,
      "grad_norm": 1.2496647834777832,
      "learning_rate": 9.862368421052632e-05,
      "loss": 0.3923,
      "step": 25242
    },
    {
      "epoch": 1.4135401500727964,
      "grad_norm": 1.2085115909576416,
      "learning_rate": 9.862342105263158e-05,
      "loss": 0.4895,
      "step": 25243
    },
    {
      "epoch": 1.4135961473849254,
      "grad_norm": 1.3878856897354126,
      "learning_rate": 9.862315789473686e-05,
      "loss": 0.4754,
      "step": 25244
    },
    {
      "epoch": 1.4136521446970545,
      "grad_norm": 1.2352678775787354,
      "learning_rate": 9.86228947368421e-05,
      "loss": 0.4851,
      "step": 25245
    },
    {
      "epoch": 1.4137081420091835,
      "grad_norm": 1.5618391036987305,
      "learning_rate": 9.862263157894738e-05,
      "loss": 0.5905,
      "step": 25246
    },
    {
      "epoch": 1.4137641393213125,
      "grad_norm": 1.4296767711639404,
      "learning_rate": 9.862236842105264e-05,
      "loss": 0.4608,
      "step": 25247
    },
    {
      "epoch": 1.4138201366334415,
      "grad_norm": 1.3693063259124756,
      "learning_rate": 9.862210526315791e-05,
      "loss": 0.3787,
      "step": 25248
    },
    {
      "epoch": 1.4138761339455705,
      "grad_norm": 1.2497797012329102,
      "learning_rate": 9.862184210526316e-05,
      "loss": 0.4279,
      "step": 25249
    },
    {
      "epoch": 1.4139321312576996,
      "grad_norm": 1.4723483324050903,
      "learning_rate": 9.862157894736842e-05,
      "loss": 0.5253,
      "step": 25250
    },
    {
      "epoch": 1.4139881285698286,
      "grad_norm": 1.3741096258163452,
      "learning_rate": 9.862131578947369e-05,
      "loss": 0.5041,
      "step": 25251
    },
    {
      "epoch": 1.4140441258819576,
      "grad_norm": 1.483561396598816,
      "learning_rate": 9.862105263157895e-05,
      "loss": 0.4966,
      "step": 25252
    },
    {
      "epoch": 1.4141001231940866,
      "grad_norm": 1.5171470642089844,
      "learning_rate": 9.862078947368422e-05,
      "loss": 0.4932,
      "step": 25253
    },
    {
      "epoch": 1.4141561205062156,
      "grad_norm": 1.2929694652557373,
      "learning_rate": 9.862052631578947e-05,
      "loss": 0.5337,
      "step": 25254
    },
    {
      "epoch": 1.4142121178183447,
      "grad_norm": 1.4575672149658203,
      "learning_rate": 9.862026315789474e-05,
      "loss": 0.4108,
      "step": 25255
    },
    {
      "epoch": 1.4142681151304737,
      "grad_norm": 1.2953664064407349,
      "learning_rate": 9.862e-05,
      "loss": 0.4451,
      "step": 25256
    },
    {
      "epoch": 1.4143241124426027,
      "grad_norm": 1.1995396614074707,
      "learning_rate": 9.861973684210527e-05,
      "loss": 0.5422,
      "step": 25257
    },
    {
      "epoch": 1.4143801097547317,
      "grad_norm": 1.1539461612701416,
      "learning_rate": 9.861947368421053e-05,
      "loss": 0.401,
      "step": 25258
    },
    {
      "epoch": 1.4144361070668607,
      "grad_norm": 1.0791876316070557,
      "learning_rate": 9.86192105263158e-05,
      "loss": 0.4076,
      "step": 25259
    },
    {
      "epoch": 1.4144921043789898,
      "grad_norm": 1.157410979270935,
      "learning_rate": 9.861894736842105e-05,
      "loss": 0.4216,
      "step": 25260
    },
    {
      "epoch": 1.4145481016911188,
      "grad_norm": 1.2240326404571533,
      "learning_rate": 9.861868421052633e-05,
      "loss": 0.3884,
      "step": 25261
    },
    {
      "epoch": 1.4146040990032478,
      "grad_norm": 1.393512487411499,
      "learning_rate": 9.861842105263159e-05,
      "loss": 0.4577,
      "step": 25262
    },
    {
      "epoch": 1.4146600963153768,
      "grad_norm": 1.1321724653244019,
      "learning_rate": 9.861815789473685e-05,
      "loss": 0.3056,
      "step": 25263
    },
    {
      "epoch": 1.4147160936275058,
      "grad_norm": 1.144692063331604,
      "learning_rate": 9.86178947368421e-05,
      "loss": 0.4127,
      "step": 25264
    },
    {
      "epoch": 1.4147720909396349,
      "grad_norm": 1.49162757396698,
      "learning_rate": 9.861763157894738e-05,
      "loss": 0.5832,
      "step": 25265
    },
    {
      "epoch": 1.4148280882517639,
      "grad_norm": 1.2007313966751099,
      "learning_rate": 9.861736842105264e-05,
      "loss": 0.4345,
      "step": 25266
    },
    {
      "epoch": 1.414884085563893,
      "grad_norm": 1.4854063987731934,
      "learning_rate": 9.86171052631579e-05,
      "loss": 0.5534,
      "step": 25267
    },
    {
      "epoch": 1.414940082876022,
      "grad_norm": 2.267275094985962,
      "learning_rate": 9.861684210526316e-05,
      "loss": 0.5159,
      "step": 25268
    },
    {
      "epoch": 1.414996080188151,
      "grad_norm": 1.5217339992523193,
      "learning_rate": 9.861657894736842e-05,
      "loss": 0.4637,
      "step": 25269
    },
    {
      "epoch": 1.41505207750028,
      "grad_norm": 1.3068915605545044,
      "learning_rate": 9.861631578947369e-05,
      "loss": 0.4396,
      "step": 25270
    },
    {
      "epoch": 1.415108074812409,
      "grad_norm": 1.3373641967773438,
      "learning_rate": 9.861605263157895e-05,
      "loss": 0.3759,
      "step": 25271
    },
    {
      "epoch": 1.415164072124538,
      "grad_norm": 1.6124898195266724,
      "learning_rate": 9.861578947368421e-05,
      "loss": 0.4239,
      "step": 25272
    },
    {
      "epoch": 1.415220069436667,
      "grad_norm": 1.4194447994232178,
      "learning_rate": 9.861552631578947e-05,
      "loss": 0.4973,
      "step": 25273
    },
    {
      "epoch": 1.415276066748796,
      "grad_norm": 1.24207603931427,
      "learning_rate": 9.861526315789474e-05,
      "loss": 0.4253,
      "step": 25274
    },
    {
      "epoch": 1.415332064060925,
      "grad_norm": 1.3166199922561646,
      "learning_rate": 9.8615e-05,
      "loss": 0.4236,
      "step": 25275
    },
    {
      "epoch": 1.415388061373054,
      "grad_norm": 1.2715953588485718,
      "learning_rate": 9.861473684210528e-05,
      "loss": 0.486,
      "step": 25276
    },
    {
      "epoch": 1.415444058685183,
      "grad_norm": 1.384478211402893,
      "learning_rate": 9.861447368421052e-05,
      "loss": 0.3994,
      "step": 25277
    },
    {
      "epoch": 1.415500055997312,
      "grad_norm": 1.9101049900054932,
      "learning_rate": 9.86142105263158e-05,
      "loss": 0.4845,
      "step": 25278
    },
    {
      "epoch": 1.4155560533094411,
      "grad_norm": 1.2240445613861084,
      "learning_rate": 9.861394736842106e-05,
      "loss": 0.4528,
      "step": 25279
    },
    {
      "epoch": 1.4156120506215701,
      "grad_norm": 1.328916072845459,
      "learning_rate": 9.861368421052633e-05,
      "loss": 0.3869,
      "step": 25280
    },
    {
      "epoch": 1.4156680479336992,
      "grad_norm": 1.400500774383545,
      "learning_rate": 9.861342105263159e-05,
      "loss": 0.5352,
      "step": 25281
    },
    {
      "epoch": 1.4157240452458282,
      "grad_norm": 1.5027445554733276,
      "learning_rate": 9.861315789473685e-05,
      "loss": 0.5391,
      "step": 25282
    },
    {
      "epoch": 1.4157800425579572,
      "grad_norm": 1.1852748394012451,
      "learning_rate": 9.861289473684211e-05,
      "loss": 0.4876,
      "step": 25283
    },
    {
      "epoch": 1.4158360398700862,
      "grad_norm": 1.3775509595870972,
      "learning_rate": 9.861263157894738e-05,
      "loss": 0.4517,
      "step": 25284
    },
    {
      "epoch": 1.4158920371822152,
      "grad_norm": 1.608871340751648,
      "learning_rate": 9.861236842105264e-05,
      "loss": 0.5324,
      "step": 25285
    },
    {
      "epoch": 1.4159480344943443,
      "grad_norm": 1.3725733757019043,
      "learning_rate": 9.86121052631579e-05,
      "loss": 0.4773,
      "step": 25286
    },
    {
      "epoch": 1.4160040318064733,
      "grad_norm": 1.2922005653381348,
      "learning_rate": 9.861184210526316e-05,
      "loss": 0.4044,
      "step": 25287
    },
    {
      "epoch": 1.4160600291186023,
      "grad_norm": 1.2618980407714844,
      "learning_rate": 9.861157894736842e-05,
      "loss": 0.4802,
      "step": 25288
    },
    {
      "epoch": 1.4161160264307313,
      "grad_norm": 1.3802038431167603,
      "learning_rate": 9.86113157894737e-05,
      "loss": 0.4104,
      "step": 25289
    },
    {
      "epoch": 1.4161720237428603,
      "grad_norm": 1.4444077014923096,
      "learning_rate": 9.861105263157895e-05,
      "loss": 0.4617,
      "step": 25290
    },
    {
      "epoch": 1.4162280210549893,
      "grad_norm": 3.483999013900757,
      "learning_rate": 9.861078947368421e-05,
      "loss": 0.5907,
      "step": 25291
    },
    {
      "epoch": 1.4162840183671184,
      "grad_norm": 1.1073352098464966,
      "learning_rate": 9.861052631578947e-05,
      "loss": 0.4187,
      "step": 25292
    },
    {
      "epoch": 1.4163400156792474,
      "grad_norm": 1.1422905921936035,
      "learning_rate": 9.861026315789475e-05,
      "loss": 0.3392,
      "step": 25293
    },
    {
      "epoch": 1.4163960129913764,
      "grad_norm": 1.6512439250946045,
      "learning_rate": 9.861e-05,
      "loss": 0.4894,
      "step": 25294
    },
    {
      "epoch": 1.4164520103035054,
      "grad_norm": 1.3519871234893799,
      "learning_rate": 9.860973684210527e-05,
      "loss": 0.5066,
      "step": 25295
    },
    {
      "epoch": 1.4165080076156344,
      "grad_norm": 1.090065360069275,
      "learning_rate": 9.860947368421053e-05,
      "loss": 0.3512,
      "step": 25296
    },
    {
      "epoch": 1.4165640049277635,
      "grad_norm": 1.3428910970687866,
      "learning_rate": 9.86092105263158e-05,
      "loss": 0.4256,
      "step": 25297
    },
    {
      "epoch": 1.4166200022398925,
      "grad_norm": 1.3911185264587402,
      "learning_rate": 9.860894736842106e-05,
      "loss": 0.4326,
      "step": 25298
    },
    {
      "epoch": 1.4166759995520215,
      "grad_norm": 1.390636920928955,
      "learning_rate": 9.860868421052633e-05,
      "loss": 0.4339,
      "step": 25299
    },
    {
      "epoch": 1.4167319968641505,
      "grad_norm": 1.4221104383468628,
      "learning_rate": 9.860842105263158e-05,
      "loss": 0.3952,
      "step": 25300
    },
    {
      "epoch": 1.4167879941762795,
      "grad_norm": 1.3638477325439453,
      "learning_rate": 9.860815789473685e-05,
      "loss": 0.4597,
      "step": 25301
    },
    {
      "epoch": 1.4168439914884086,
      "grad_norm": 1.2395983934402466,
      "learning_rate": 9.860789473684211e-05,
      "loss": 0.3628,
      "step": 25302
    },
    {
      "epoch": 1.4168999888005376,
      "grad_norm": 1.2597976922988892,
      "learning_rate": 9.860763157894737e-05,
      "loss": 0.422,
      "step": 25303
    },
    {
      "epoch": 1.4169559861126666,
      "grad_norm": 1.3745061159133911,
      "learning_rate": 9.860736842105263e-05,
      "loss": 0.5168,
      "step": 25304
    },
    {
      "epoch": 1.4170119834247956,
      "grad_norm": 1.2840275764465332,
      "learning_rate": 9.860710526315789e-05,
      "loss": 0.4069,
      "step": 25305
    },
    {
      "epoch": 1.4170679807369246,
      "grad_norm": 1.353759765625,
      "learning_rate": 9.860684210526316e-05,
      "loss": 0.4599,
      "step": 25306
    },
    {
      "epoch": 1.4171239780490537,
      "grad_norm": 1.3688899278640747,
      "learning_rate": 9.860657894736842e-05,
      "loss": 0.4981,
      "step": 25307
    },
    {
      "epoch": 1.4171799753611827,
      "grad_norm": 1.1914405822753906,
      "learning_rate": 9.86063157894737e-05,
      "loss": 0.3851,
      "step": 25308
    },
    {
      "epoch": 1.4172359726733117,
      "grad_norm": 1.4963513612747192,
      "learning_rate": 9.860605263157894e-05,
      "loss": 0.4801,
      "step": 25309
    },
    {
      "epoch": 1.4172919699854407,
      "grad_norm": 1.4012556076049805,
      "learning_rate": 9.860578947368422e-05,
      "loss": 0.4715,
      "step": 25310
    },
    {
      "epoch": 1.4173479672975697,
      "grad_norm": 1.372742772102356,
      "learning_rate": 9.860552631578948e-05,
      "loss": 0.4207,
      "step": 25311
    },
    {
      "epoch": 1.4174039646096988,
      "grad_norm": 1.3505504131317139,
      "learning_rate": 9.860526315789475e-05,
      "loss": 0.5519,
      "step": 25312
    },
    {
      "epoch": 1.4174599619218278,
      "grad_norm": 1.4344871044158936,
      "learning_rate": 9.860500000000001e-05,
      "loss": 0.4624,
      "step": 25313
    },
    {
      "epoch": 1.4175159592339568,
      "grad_norm": 1.3725773096084595,
      "learning_rate": 9.860473684210527e-05,
      "loss": 0.4659,
      "step": 25314
    },
    {
      "epoch": 1.4175719565460858,
      "grad_norm": 1.265283226966858,
      "learning_rate": 9.860447368421053e-05,
      "loss": 0.4751,
      "step": 25315
    },
    {
      "epoch": 1.4176279538582148,
      "grad_norm": 1.3512071371078491,
      "learning_rate": 9.86042105263158e-05,
      "loss": 0.5503,
      "step": 25316
    },
    {
      "epoch": 1.4176839511703438,
      "grad_norm": 1.2943168878555298,
      "learning_rate": 9.860394736842106e-05,
      "loss": 0.3472,
      "step": 25317
    },
    {
      "epoch": 1.4177399484824729,
      "grad_norm": 1.1657301187515259,
      "learning_rate": 9.860368421052632e-05,
      "loss": 0.4913,
      "step": 25318
    },
    {
      "epoch": 1.4177959457946019,
      "grad_norm": 1.530411958694458,
      "learning_rate": 9.860342105263158e-05,
      "loss": 0.4243,
      "step": 25319
    },
    {
      "epoch": 1.417851943106731,
      "grad_norm": 1.179490089416504,
      "learning_rate": 9.860315789473684e-05,
      "loss": 0.4927,
      "step": 25320
    },
    {
      "epoch": 1.41790794041886,
      "grad_norm": 1.3139373064041138,
      "learning_rate": 9.860289473684211e-05,
      "loss": 0.4186,
      "step": 25321
    },
    {
      "epoch": 1.417963937730989,
      "grad_norm": 1.1053187847137451,
      "learning_rate": 9.860263157894737e-05,
      "loss": 0.4252,
      "step": 25322
    },
    {
      "epoch": 1.418019935043118,
      "grad_norm": 1.361271858215332,
      "learning_rate": 9.860236842105263e-05,
      "loss": 0.5176,
      "step": 25323
    },
    {
      "epoch": 1.418075932355247,
      "grad_norm": 1.6484142541885376,
      "learning_rate": 9.860210526315789e-05,
      "loss": 0.5088,
      "step": 25324
    },
    {
      "epoch": 1.418131929667376,
      "grad_norm": 1.2425090074539185,
      "learning_rate": 9.860184210526317e-05,
      "loss": 0.4417,
      "step": 25325
    },
    {
      "epoch": 1.418187926979505,
      "grad_norm": 1.8077564239501953,
      "learning_rate": 9.860157894736843e-05,
      "loss": 0.5891,
      "step": 25326
    },
    {
      "epoch": 1.418243924291634,
      "grad_norm": 1.4657273292541504,
      "learning_rate": 9.860131578947369e-05,
      "loss": 0.5193,
      "step": 25327
    },
    {
      "epoch": 1.418299921603763,
      "grad_norm": 1.2672172784805298,
      "learning_rate": 9.860105263157895e-05,
      "loss": 0.398,
      "step": 25328
    },
    {
      "epoch": 1.418355918915892,
      "grad_norm": 1.22288978099823,
      "learning_rate": 9.860078947368422e-05,
      "loss": 0.5111,
      "step": 25329
    },
    {
      "epoch": 1.418411916228021,
      "grad_norm": 1.6638717651367188,
      "learning_rate": 9.860052631578948e-05,
      "loss": 0.5352,
      "step": 25330
    },
    {
      "epoch": 1.4184679135401501,
      "grad_norm": 1.5996172428131104,
      "learning_rate": 9.860026315789475e-05,
      "loss": 0.5242,
      "step": 25331
    },
    {
      "epoch": 1.4185239108522791,
      "grad_norm": 1.2739406824111938,
      "learning_rate": 9.86e-05,
      "loss": 0.4868,
      "step": 25332
    },
    {
      "epoch": 1.4185799081644082,
      "grad_norm": 1.4929602146148682,
      "learning_rate": 9.859973684210527e-05,
      "loss": 0.4906,
      "step": 25333
    },
    {
      "epoch": 1.4186359054765372,
      "grad_norm": 1.3288265466690063,
      "learning_rate": 9.859947368421053e-05,
      "loss": 0.4633,
      "step": 25334
    },
    {
      "epoch": 1.4186919027886662,
      "grad_norm": 1.4196847677230835,
      "learning_rate": 9.85992105263158e-05,
      "loss": 0.5721,
      "step": 25335
    },
    {
      "epoch": 1.4187479001007952,
      "grad_norm": 1.4367882013320923,
      "learning_rate": 9.859894736842106e-05,
      "loss": 0.5164,
      "step": 25336
    },
    {
      "epoch": 1.4188038974129242,
      "grad_norm": 1.402292013168335,
      "learning_rate": 9.859868421052631e-05,
      "loss": 0.5308,
      "step": 25337
    },
    {
      "epoch": 1.4188598947250533,
      "grad_norm": 1.1375635862350464,
      "learning_rate": 9.859842105263158e-05,
      "loss": 0.3699,
      "step": 25338
    },
    {
      "epoch": 1.4189158920371823,
      "grad_norm": 1.2534152269363403,
      "learning_rate": 9.859815789473684e-05,
      "loss": 0.5699,
      "step": 25339
    },
    {
      "epoch": 1.4189718893493113,
      "grad_norm": 1.2568928003311157,
      "learning_rate": 9.859789473684212e-05,
      "loss": 0.3924,
      "step": 25340
    },
    {
      "epoch": 1.4190278866614403,
      "grad_norm": 1.2098966836929321,
      "learning_rate": 9.859763157894736e-05,
      "loss": 0.5723,
      "step": 25341
    },
    {
      "epoch": 1.4190838839735693,
      "grad_norm": 1.161672592163086,
      "learning_rate": 9.859736842105264e-05,
      "loss": 0.4008,
      "step": 25342
    },
    {
      "epoch": 1.4191398812856983,
      "grad_norm": 1.3864883184432983,
      "learning_rate": 9.85971052631579e-05,
      "loss": 0.4221,
      "step": 25343
    },
    {
      "epoch": 1.4191958785978274,
      "grad_norm": 1.7194762229919434,
      "learning_rate": 9.859684210526317e-05,
      "loss": 0.571,
      "step": 25344
    },
    {
      "epoch": 1.4192518759099564,
      "grad_norm": 1.3612377643585205,
      "learning_rate": 9.859657894736843e-05,
      "loss": 0.3921,
      "step": 25345
    },
    {
      "epoch": 1.4193078732220854,
      "grad_norm": 1.2218958139419556,
      "learning_rate": 9.859631578947369e-05,
      "loss": 0.3954,
      "step": 25346
    },
    {
      "epoch": 1.4193638705342144,
      "grad_norm": 1.7550920248031616,
      "learning_rate": 9.859605263157895e-05,
      "loss": 0.529,
      "step": 25347
    },
    {
      "epoch": 1.4194198678463434,
      "grad_norm": 1.7339459657669067,
      "learning_rate": 9.859578947368422e-05,
      "loss": 0.5703,
      "step": 25348
    },
    {
      "epoch": 1.4194758651584725,
      "grad_norm": 1.1789673566818237,
      "learning_rate": 9.859552631578948e-05,
      "loss": 0.3967,
      "step": 25349
    },
    {
      "epoch": 1.4195318624706015,
      "grad_norm": 1.3035659790039062,
      "learning_rate": 9.859526315789474e-05,
      "loss": 0.416,
      "step": 25350
    },
    {
      "epoch": 1.4195878597827305,
      "grad_norm": 1.4353876113891602,
      "learning_rate": 9.8595e-05,
      "loss": 0.4244,
      "step": 25351
    },
    {
      "epoch": 1.4196438570948595,
      "grad_norm": 1.650483250617981,
      "learning_rate": 9.859473684210527e-05,
      "loss": 0.662,
      "step": 25352
    },
    {
      "epoch": 1.4196998544069885,
      "grad_norm": 1.2579351663589478,
      "learning_rate": 9.859447368421053e-05,
      "loss": 0.3497,
      "step": 25353
    },
    {
      "epoch": 1.4197558517191176,
      "grad_norm": 1.3151016235351562,
      "learning_rate": 9.859421052631579e-05,
      "loss": 0.4805,
      "step": 25354
    },
    {
      "epoch": 1.4198118490312466,
      "grad_norm": 2.101506471633911,
      "learning_rate": 9.859394736842105e-05,
      "loss": 0.5463,
      "step": 25355
    },
    {
      "epoch": 1.4198678463433756,
      "grad_norm": 1.4948760271072388,
      "learning_rate": 9.859368421052631e-05,
      "loss": 0.6012,
      "step": 25356
    },
    {
      "epoch": 1.4199238436555046,
      "grad_norm": 1.2718428373336792,
      "learning_rate": 9.859342105263159e-05,
      "loss": 0.5115,
      "step": 25357
    },
    {
      "epoch": 1.4199798409676336,
      "grad_norm": 1.3629120588302612,
      "learning_rate": 9.859315789473685e-05,
      "loss": 0.4204,
      "step": 25358
    },
    {
      "epoch": 1.4200358382797627,
      "grad_norm": 1.874666452407837,
      "learning_rate": 9.85928947368421e-05,
      "loss": 0.5246,
      "step": 25359
    },
    {
      "epoch": 1.4200918355918917,
      "grad_norm": 1.4612150192260742,
      "learning_rate": 9.859263157894737e-05,
      "loss": 0.4192,
      "step": 25360
    },
    {
      "epoch": 1.4201478329040205,
      "grad_norm": 1.6113919019699097,
      "learning_rate": 9.859236842105264e-05,
      "loss": 0.6654,
      "step": 25361
    },
    {
      "epoch": 1.4202038302161495,
      "grad_norm": 1.3224183320999146,
      "learning_rate": 9.85921052631579e-05,
      "loss": 0.5592,
      "step": 25362
    },
    {
      "epoch": 1.4202598275282785,
      "grad_norm": 1.343196153640747,
      "learning_rate": 9.859184210526317e-05,
      "loss": 0.4718,
      "step": 25363
    },
    {
      "epoch": 1.4203158248404075,
      "grad_norm": 1.2355575561523438,
      "learning_rate": 9.859157894736842e-05,
      "loss": 0.4856,
      "step": 25364
    },
    {
      "epoch": 1.4203718221525365,
      "grad_norm": 1.216900110244751,
      "learning_rate": 9.859131578947369e-05,
      "loss": 0.4172,
      "step": 25365
    },
    {
      "epoch": 1.4204278194646656,
      "grad_norm": 1.3724902868270874,
      "learning_rate": 9.859105263157895e-05,
      "loss": 0.4328,
      "step": 25366
    },
    {
      "epoch": 1.4204838167767946,
      "grad_norm": 1.2023346424102783,
      "learning_rate": 9.859078947368422e-05,
      "loss": 0.493,
      "step": 25367
    },
    {
      "epoch": 1.4205398140889236,
      "grad_norm": 1.1472097635269165,
      "learning_rate": 9.859052631578948e-05,
      "loss": 0.3719,
      "step": 25368
    },
    {
      "epoch": 1.4205958114010526,
      "grad_norm": 1.615952730178833,
      "learning_rate": 9.859026315789474e-05,
      "loss": 0.595,
      "step": 25369
    },
    {
      "epoch": 1.4206518087131816,
      "grad_norm": 1.4684040546417236,
      "learning_rate": 9.859e-05,
      "loss": 0.4789,
      "step": 25370
    },
    {
      "epoch": 1.4207078060253107,
      "grad_norm": 1.437103271484375,
      "learning_rate": 9.858973684210526e-05,
      "loss": 0.6406,
      "step": 25371
    },
    {
      "epoch": 1.4207638033374397,
      "grad_norm": 1.5414632558822632,
      "learning_rate": 9.858947368421054e-05,
      "loss": 0.3936,
      "step": 25372
    },
    {
      "epoch": 1.4208198006495687,
      "grad_norm": 1.5145574808120728,
      "learning_rate": 9.85892105263158e-05,
      "loss": 0.4852,
      "step": 25373
    },
    {
      "epoch": 1.4208757979616977,
      "grad_norm": 1.5883476734161377,
      "learning_rate": 9.858894736842106e-05,
      "loss": 0.5683,
      "step": 25374
    },
    {
      "epoch": 1.4209317952738267,
      "grad_norm": 1.665426254272461,
      "learning_rate": 9.858868421052632e-05,
      "loss": 0.4271,
      "step": 25375
    },
    {
      "epoch": 1.4209877925859558,
      "grad_norm": 1.6864166259765625,
      "learning_rate": 9.858842105263159e-05,
      "loss": 0.6829,
      "step": 25376
    },
    {
      "epoch": 1.4210437898980848,
      "grad_norm": 1.3218377828598022,
      "learning_rate": 9.858815789473685e-05,
      "loss": 0.4846,
      "step": 25377
    },
    {
      "epoch": 1.4210997872102138,
      "grad_norm": 1.1267671585083008,
      "learning_rate": 9.858789473684211e-05,
      "loss": 0.4406,
      "step": 25378
    },
    {
      "epoch": 1.4211557845223428,
      "grad_norm": 1.2661861181259155,
      "learning_rate": 9.858763157894737e-05,
      "loss": 0.4269,
      "step": 25379
    },
    {
      "epoch": 1.4212117818344718,
      "grad_norm": 1.487857460975647,
      "learning_rate": 9.858736842105264e-05,
      "loss": 0.5297,
      "step": 25380
    },
    {
      "epoch": 1.4212677791466009,
      "grad_norm": 1.3049241304397583,
      "learning_rate": 9.85871052631579e-05,
      "loss": 0.4583,
      "step": 25381
    },
    {
      "epoch": 1.4213237764587299,
      "grad_norm": 1.3917500972747803,
      "learning_rate": 9.858684210526316e-05,
      "loss": 0.444,
      "step": 25382
    },
    {
      "epoch": 1.421379773770859,
      "grad_norm": 1.7665787935256958,
      "learning_rate": 9.858657894736842e-05,
      "loss": 0.4787,
      "step": 25383
    },
    {
      "epoch": 1.421435771082988,
      "grad_norm": 2.91767954826355,
      "learning_rate": 9.85863157894737e-05,
      "loss": 0.471,
      "step": 25384
    },
    {
      "epoch": 1.421491768395117,
      "grad_norm": 1.3150043487548828,
      "learning_rate": 9.858605263157895e-05,
      "loss": 0.4157,
      "step": 25385
    },
    {
      "epoch": 1.421547765707246,
      "grad_norm": 1.415709137916565,
      "learning_rate": 9.858578947368423e-05,
      "loss": 0.4664,
      "step": 25386
    },
    {
      "epoch": 1.421603763019375,
      "grad_norm": 1.6887582540512085,
      "learning_rate": 9.858552631578947e-05,
      "loss": 0.5149,
      "step": 25387
    },
    {
      "epoch": 1.421659760331504,
      "grad_norm": 1.4312492609024048,
      "learning_rate": 9.858526315789473e-05,
      "loss": 0.5076,
      "step": 25388
    },
    {
      "epoch": 1.421715757643633,
      "grad_norm": 1.4411532878875732,
      "learning_rate": 9.8585e-05,
      "loss": 0.5419,
      "step": 25389
    },
    {
      "epoch": 1.421771754955762,
      "grad_norm": 1.3503779172897339,
      "learning_rate": 9.858473684210527e-05,
      "loss": 0.4765,
      "step": 25390
    },
    {
      "epoch": 1.421827752267891,
      "grad_norm": 1.2609636783599854,
      "learning_rate": 9.858447368421054e-05,
      "loss": 0.4416,
      "step": 25391
    },
    {
      "epoch": 1.42188374958002,
      "grad_norm": 1.4397571086883545,
      "learning_rate": 9.858421052631578e-05,
      "loss": 0.4568,
      "step": 25392
    },
    {
      "epoch": 1.421939746892149,
      "grad_norm": 1.258315086364746,
      "learning_rate": 9.858394736842106e-05,
      "loss": 0.4831,
      "step": 25393
    },
    {
      "epoch": 1.421995744204278,
      "grad_norm": 1.210988998413086,
      "learning_rate": 9.858368421052632e-05,
      "loss": 0.5015,
      "step": 25394
    },
    {
      "epoch": 1.4220517415164071,
      "grad_norm": 1.2439945936203003,
      "learning_rate": 9.858342105263159e-05,
      "loss": 0.3687,
      "step": 25395
    },
    {
      "epoch": 1.4221077388285361,
      "grad_norm": 1.1381407976150513,
      "learning_rate": 9.858315789473684e-05,
      "loss": 0.4425,
      "step": 25396
    },
    {
      "epoch": 1.4221637361406652,
      "grad_norm": 1.4158564805984497,
      "learning_rate": 9.858289473684211e-05,
      "loss": 0.4328,
      "step": 25397
    },
    {
      "epoch": 1.4222197334527942,
      "grad_norm": 1.231938123703003,
      "learning_rate": 9.858263157894737e-05,
      "loss": 0.3726,
      "step": 25398
    },
    {
      "epoch": 1.4222757307649232,
      "grad_norm": 1.071649193763733,
      "learning_rate": 9.858236842105264e-05,
      "loss": 0.3096,
      "step": 25399
    },
    {
      "epoch": 1.4223317280770522,
      "grad_norm": 3.990752696990967,
      "learning_rate": 9.85821052631579e-05,
      "loss": 0.3287,
      "step": 25400
    },
    {
      "epoch": 1.4223877253891812,
      "grad_norm": 1.2647135257720947,
      "learning_rate": 9.858184210526316e-05,
      "loss": 0.409,
      "step": 25401
    },
    {
      "epoch": 1.4224437227013103,
      "grad_norm": 1.5738040208816528,
      "learning_rate": 9.858157894736842e-05,
      "loss": 0.5038,
      "step": 25402
    },
    {
      "epoch": 1.4224997200134393,
      "grad_norm": 1.283447265625,
      "learning_rate": 9.85813157894737e-05,
      "loss": 0.4931,
      "step": 25403
    },
    {
      "epoch": 1.4225557173255683,
      "grad_norm": 1.3840880393981934,
      "learning_rate": 9.858105263157896e-05,
      "loss": 0.5407,
      "step": 25404
    },
    {
      "epoch": 1.4226117146376973,
      "grad_norm": 1.3599412441253662,
      "learning_rate": 9.858078947368422e-05,
      "loss": 0.5234,
      "step": 25405
    },
    {
      "epoch": 1.4226677119498263,
      "grad_norm": 1.2359181642532349,
      "learning_rate": 9.858052631578948e-05,
      "loss": 0.4504,
      "step": 25406
    },
    {
      "epoch": 1.4227237092619553,
      "grad_norm": 1.4266167879104614,
      "learning_rate": 9.858026315789473e-05,
      "loss": 0.5036,
      "step": 25407
    },
    {
      "epoch": 1.4227797065740844,
      "grad_norm": 1.2063062191009521,
      "learning_rate": 9.858000000000001e-05,
      "loss": 0.4117,
      "step": 25408
    },
    {
      "epoch": 1.4228357038862134,
      "grad_norm": 1.31067955493927,
      "learning_rate": 9.857973684210527e-05,
      "loss": 0.3727,
      "step": 25409
    },
    {
      "epoch": 1.4228917011983424,
      "grad_norm": 2.1875691413879395,
      "learning_rate": 9.857947368421053e-05,
      "loss": 0.4012,
      "step": 25410
    },
    {
      "epoch": 1.4229476985104714,
      "grad_norm": 1.1316533088684082,
      "learning_rate": 9.857921052631579e-05,
      "loss": 0.4388,
      "step": 25411
    },
    {
      "epoch": 1.4230036958226004,
      "grad_norm": 1.3008742332458496,
      "learning_rate": 9.857894736842106e-05,
      "loss": 0.3254,
      "step": 25412
    },
    {
      "epoch": 1.4230596931347295,
      "grad_norm": 1.3735431432724,
      "learning_rate": 9.857868421052632e-05,
      "loss": 0.4582,
      "step": 25413
    },
    {
      "epoch": 1.4231156904468585,
      "grad_norm": 1.4176539182662964,
      "learning_rate": 9.857842105263158e-05,
      "loss": 0.4259,
      "step": 25414
    },
    {
      "epoch": 1.4231716877589875,
      "grad_norm": 1.55917489528656,
      "learning_rate": 9.857815789473684e-05,
      "loss": 0.5155,
      "step": 25415
    },
    {
      "epoch": 1.4232276850711165,
      "grad_norm": 3.2967946529388428,
      "learning_rate": 9.857789473684211e-05,
      "loss": 0.4553,
      "step": 25416
    },
    {
      "epoch": 1.4232836823832455,
      "grad_norm": 2.0240895748138428,
      "learning_rate": 9.857763157894737e-05,
      "loss": 0.5449,
      "step": 25417
    },
    {
      "epoch": 1.4233396796953746,
      "grad_norm": 1.7926132678985596,
      "learning_rate": 9.857736842105265e-05,
      "loss": 0.4438,
      "step": 25418
    },
    {
      "epoch": 1.4233956770075036,
      "grad_norm": 1.3341935873031616,
      "learning_rate": 9.857710526315789e-05,
      "loss": 0.4905,
      "step": 25419
    },
    {
      "epoch": 1.4234516743196326,
      "grad_norm": 1.5490853786468506,
      "learning_rate": 9.857684210526317e-05,
      "loss": 0.5582,
      "step": 25420
    },
    {
      "epoch": 1.4235076716317616,
      "grad_norm": 1.5838794708251953,
      "learning_rate": 9.857657894736843e-05,
      "loss": 0.5194,
      "step": 25421
    },
    {
      "epoch": 1.4235636689438906,
      "grad_norm": 1.493714690208435,
      "learning_rate": 9.85763157894737e-05,
      "loss": 0.7161,
      "step": 25422
    },
    {
      "epoch": 1.4236196662560197,
      "grad_norm": 1.3017715215682983,
      "learning_rate": 9.857605263157896e-05,
      "loss": 0.4129,
      "step": 25423
    },
    {
      "epoch": 1.4236756635681487,
      "grad_norm": 2.0874288082122803,
      "learning_rate": 9.85757894736842e-05,
      "loss": 0.5038,
      "step": 25424
    },
    {
      "epoch": 1.4237316608802777,
      "grad_norm": 1.455629587173462,
      "learning_rate": 9.857552631578948e-05,
      "loss": 0.6834,
      "step": 25425
    },
    {
      "epoch": 1.4237876581924067,
      "grad_norm": 1.1289703845977783,
      "learning_rate": 9.857526315789474e-05,
      "loss": 0.3529,
      "step": 25426
    },
    {
      "epoch": 1.4238436555045357,
      "grad_norm": 1.2799420356750488,
      "learning_rate": 9.857500000000001e-05,
      "loss": 0.3936,
      "step": 25427
    },
    {
      "epoch": 1.4238996528166648,
      "grad_norm": 1.4523367881774902,
      "learning_rate": 9.857473684210527e-05,
      "loss": 0.4842,
      "step": 25428
    },
    {
      "epoch": 1.4239556501287938,
      "grad_norm": 1.5793325901031494,
      "learning_rate": 9.857447368421053e-05,
      "loss": 0.3518,
      "step": 25429
    },
    {
      "epoch": 1.4240116474409228,
      "grad_norm": 1.5430808067321777,
      "learning_rate": 9.857421052631579e-05,
      "loss": 0.5515,
      "step": 25430
    },
    {
      "epoch": 1.4240676447530518,
      "grad_norm": 1.567837119102478,
      "learning_rate": 9.857394736842106e-05,
      "loss": 0.4913,
      "step": 25431
    },
    {
      "epoch": 1.4241236420651808,
      "grad_norm": 1.3282862901687622,
      "learning_rate": 9.857368421052632e-05,
      "loss": 0.5311,
      "step": 25432
    },
    {
      "epoch": 1.4241796393773098,
      "grad_norm": 1.3393051624298096,
      "learning_rate": 9.857342105263158e-05,
      "loss": 0.4615,
      "step": 25433
    },
    {
      "epoch": 1.4242356366894389,
      "grad_norm": 1.48483145236969,
      "learning_rate": 9.857315789473684e-05,
      "loss": 0.6161,
      "step": 25434
    },
    {
      "epoch": 1.4242916340015679,
      "grad_norm": 1.5160679817199707,
      "learning_rate": 9.857289473684212e-05,
      "loss": 0.4744,
      "step": 25435
    },
    {
      "epoch": 1.424347631313697,
      "grad_norm": 1.438240647315979,
      "learning_rate": 9.857263157894738e-05,
      "loss": 0.5894,
      "step": 25436
    },
    {
      "epoch": 1.424403628625826,
      "grad_norm": 1.6397444009780884,
      "learning_rate": 9.857236842105264e-05,
      "loss": 0.6918,
      "step": 25437
    },
    {
      "epoch": 1.424459625937955,
      "grad_norm": 1.548452615737915,
      "learning_rate": 9.85721052631579e-05,
      "loss": 0.4115,
      "step": 25438
    },
    {
      "epoch": 1.424515623250084,
      "grad_norm": 1.4626857042312622,
      "learning_rate": 9.857184210526317e-05,
      "loss": 0.6428,
      "step": 25439
    },
    {
      "epoch": 1.424571620562213,
      "grad_norm": 1.2954314947128296,
      "learning_rate": 9.857157894736843e-05,
      "loss": 0.4795,
      "step": 25440
    },
    {
      "epoch": 1.424627617874342,
      "grad_norm": 1.250012755393982,
      "learning_rate": 9.857131578947369e-05,
      "loss": 0.3832,
      "step": 25441
    },
    {
      "epoch": 1.424683615186471,
      "grad_norm": 1.386883020401001,
      "learning_rate": 9.857105263157895e-05,
      "loss": 0.5218,
      "step": 25442
    },
    {
      "epoch": 1.4247396124986,
      "grad_norm": 1.4339525699615479,
      "learning_rate": 9.857078947368421e-05,
      "loss": 0.3912,
      "step": 25443
    },
    {
      "epoch": 1.424795609810729,
      "grad_norm": 1.4767746925354004,
      "learning_rate": 9.857052631578948e-05,
      "loss": 0.5203,
      "step": 25444
    },
    {
      "epoch": 1.424851607122858,
      "grad_norm": 1.441213607788086,
      "learning_rate": 9.857026315789474e-05,
      "loss": 0.4481,
      "step": 25445
    },
    {
      "epoch": 1.424907604434987,
      "grad_norm": 1.6515579223632812,
      "learning_rate": 9.857000000000001e-05,
      "loss": 0.4584,
      "step": 25446
    },
    {
      "epoch": 1.4249636017471161,
      "grad_norm": 1.3358708620071411,
      "learning_rate": 9.856973684210526e-05,
      "loss": 0.6889,
      "step": 25447
    },
    {
      "epoch": 1.4250195990592451,
      "grad_norm": 1.3359014987945557,
      "learning_rate": 9.856947368421053e-05,
      "loss": 0.4941,
      "step": 25448
    },
    {
      "epoch": 1.4250755963713742,
      "grad_norm": 1.5393649339675903,
      "learning_rate": 9.856921052631579e-05,
      "loss": 0.477,
      "step": 25449
    },
    {
      "epoch": 1.4251315936835032,
      "grad_norm": 1.1691579818725586,
      "learning_rate": 9.856894736842107e-05,
      "loss": 0.3377,
      "step": 25450
    },
    {
      "epoch": 1.4251875909956322,
      "grad_norm": 1.5221678018569946,
      "learning_rate": 9.856868421052631e-05,
      "loss": 0.6133,
      "step": 25451
    },
    {
      "epoch": 1.4252435883077612,
      "grad_norm": 1.448843002319336,
      "learning_rate": 9.856842105263159e-05,
      "loss": 0.4993,
      "step": 25452
    },
    {
      "epoch": 1.4252995856198902,
      "grad_norm": 1.2847239971160889,
      "learning_rate": 9.856815789473685e-05,
      "loss": 0.5198,
      "step": 25453
    },
    {
      "epoch": 1.4253555829320192,
      "grad_norm": 1.0127615928649902,
      "learning_rate": 9.856789473684212e-05,
      "loss": 0.3715,
      "step": 25454
    },
    {
      "epoch": 1.4254115802441483,
      "grad_norm": 1.3241877555847168,
      "learning_rate": 9.856763157894738e-05,
      "loss": 0.3097,
      "step": 25455
    },
    {
      "epoch": 1.4254675775562773,
      "grad_norm": 1.3852566480636597,
      "learning_rate": 9.856736842105264e-05,
      "loss": 0.4012,
      "step": 25456
    },
    {
      "epoch": 1.4255235748684063,
      "grad_norm": 1.2788549661636353,
      "learning_rate": 9.85671052631579e-05,
      "loss": 0.5921,
      "step": 25457
    },
    {
      "epoch": 1.4255795721805353,
      "grad_norm": 1.4329038858413696,
      "learning_rate": 9.856684210526316e-05,
      "loss": 0.5277,
      "step": 25458
    },
    {
      "epoch": 1.4256355694926643,
      "grad_norm": 1.3823659420013428,
      "learning_rate": 9.856657894736843e-05,
      "loss": 0.6759,
      "step": 25459
    },
    {
      "epoch": 1.4256915668047934,
      "grad_norm": 1.3341295719146729,
      "learning_rate": 9.856631578947369e-05,
      "loss": 0.4195,
      "step": 25460
    },
    {
      "epoch": 1.4257475641169224,
      "grad_norm": 1.3343552350997925,
      "learning_rate": 9.856605263157895e-05,
      "loss": 0.5346,
      "step": 25461
    },
    {
      "epoch": 1.4258035614290514,
      "grad_norm": 1.2055914402008057,
      "learning_rate": 9.856578947368421e-05,
      "loss": 0.4703,
      "step": 25462
    },
    {
      "epoch": 1.4258595587411804,
      "grad_norm": 2.759641170501709,
      "learning_rate": 9.856552631578948e-05,
      "loss": 0.4597,
      "step": 25463
    },
    {
      "epoch": 1.4259155560533094,
      "grad_norm": 1.228317141532898,
      "learning_rate": 9.856526315789474e-05,
      "loss": 0.427,
      "step": 25464
    },
    {
      "epoch": 1.4259715533654385,
      "grad_norm": 1.2668286561965942,
      "learning_rate": 9.8565e-05,
      "loss": 0.6717,
      "step": 25465
    },
    {
      "epoch": 1.4260275506775675,
      "grad_norm": 1.519857406616211,
      "learning_rate": 9.856473684210526e-05,
      "loss": 0.5231,
      "step": 25466
    },
    {
      "epoch": 1.4260835479896965,
      "grad_norm": 2.9786980152130127,
      "learning_rate": 9.856447368421054e-05,
      "loss": 0.5596,
      "step": 25467
    },
    {
      "epoch": 1.4261395453018255,
      "grad_norm": 1.5299054384231567,
      "learning_rate": 9.85642105263158e-05,
      "loss": 0.4107,
      "step": 25468
    },
    {
      "epoch": 1.4261955426139545,
      "grad_norm": 1.3316971063613892,
      "learning_rate": 9.856394736842105e-05,
      "loss": 0.4201,
      "step": 25469
    },
    {
      "epoch": 1.4262515399260836,
      "grad_norm": 1.4687387943267822,
      "learning_rate": 9.856368421052631e-05,
      "loss": 0.5388,
      "step": 25470
    },
    {
      "epoch": 1.4263075372382126,
      "grad_norm": 1.4971003532409668,
      "learning_rate": 9.856342105263159e-05,
      "loss": 0.4766,
      "step": 25471
    },
    {
      "epoch": 1.4263635345503416,
      "grad_norm": 1.244888186454773,
      "learning_rate": 9.856315789473685e-05,
      "loss": 0.4872,
      "step": 25472
    },
    {
      "epoch": 1.4264195318624706,
      "grad_norm": 1.1782761812210083,
      "learning_rate": 9.856289473684212e-05,
      "loss": 0.4772,
      "step": 25473
    },
    {
      "epoch": 1.4264755291745996,
      "grad_norm": 1.4044368267059326,
      "learning_rate": 9.856263157894737e-05,
      "loss": 0.4442,
      "step": 25474
    },
    {
      "epoch": 1.4265315264867287,
      "grad_norm": 1.432761311531067,
      "learning_rate": 9.856236842105263e-05,
      "loss": 0.4935,
      "step": 25475
    },
    {
      "epoch": 1.4265875237988577,
      "grad_norm": 1.283448338508606,
      "learning_rate": 9.85621052631579e-05,
      "loss": 0.38,
      "step": 25476
    },
    {
      "epoch": 1.4266435211109867,
      "grad_norm": 1.4212082624435425,
      "learning_rate": 9.856184210526316e-05,
      "loss": 0.539,
      "step": 25477
    },
    {
      "epoch": 1.4266995184231157,
      "grad_norm": 1.3751952648162842,
      "learning_rate": 9.856157894736843e-05,
      "loss": 0.399,
      "step": 25478
    },
    {
      "epoch": 1.4267555157352447,
      "grad_norm": 1.0430189371109009,
      "learning_rate": 9.856131578947368e-05,
      "loss": 0.3344,
      "step": 25479
    },
    {
      "epoch": 1.4268115130473737,
      "grad_norm": 1.3726401329040527,
      "learning_rate": 9.856105263157895e-05,
      "loss": 0.5293,
      "step": 25480
    },
    {
      "epoch": 1.4268675103595028,
      "grad_norm": 1.6846280097961426,
      "learning_rate": 9.856078947368421e-05,
      "loss": 0.5713,
      "step": 25481
    },
    {
      "epoch": 1.4269235076716318,
      "grad_norm": 1.4059313535690308,
      "learning_rate": 9.856052631578949e-05,
      "loss": 0.4264,
      "step": 25482
    },
    {
      "epoch": 1.4269795049837608,
      "grad_norm": 1.0575387477874756,
      "learning_rate": 9.856026315789475e-05,
      "loss": 0.3784,
      "step": 25483
    },
    {
      "epoch": 1.4270355022958898,
      "grad_norm": 1.3428961038589478,
      "learning_rate": 9.856e-05,
      "loss": 0.5346,
      "step": 25484
    },
    {
      "epoch": 1.4270914996080188,
      "grad_norm": 1.1390739679336548,
      "learning_rate": 9.855973684210526e-05,
      "loss": 0.3727,
      "step": 25485
    },
    {
      "epoch": 1.4271474969201479,
      "grad_norm": 1.6633925437927246,
      "learning_rate": 9.855947368421054e-05,
      "loss": 0.551,
      "step": 25486
    },
    {
      "epoch": 1.4272034942322769,
      "grad_norm": 1.5276228189468384,
      "learning_rate": 9.85592105263158e-05,
      "loss": 0.4956,
      "step": 25487
    },
    {
      "epoch": 1.427259491544406,
      "grad_norm": 1.3396916389465332,
      "learning_rate": 9.855894736842106e-05,
      "loss": 0.484,
      "step": 25488
    },
    {
      "epoch": 1.427315488856535,
      "grad_norm": 1.318367838859558,
      "learning_rate": 9.855868421052632e-05,
      "loss": 0.5128,
      "step": 25489
    },
    {
      "epoch": 1.427371486168664,
      "grad_norm": 1.217202067375183,
      "learning_rate": 9.855842105263159e-05,
      "loss": 0.4146,
      "step": 25490
    },
    {
      "epoch": 1.427427483480793,
      "grad_norm": 1.3000566959381104,
      "learning_rate": 9.855815789473685e-05,
      "loss": 0.5356,
      "step": 25491
    },
    {
      "epoch": 1.427483480792922,
      "grad_norm": 1.2982467412948608,
      "learning_rate": 9.855789473684211e-05,
      "loss": 0.4702,
      "step": 25492
    },
    {
      "epoch": 1.427539478105051,
      "grad_norm": 1.4222043752670288,
      "learning_rate": 9.855763157894737e-05,
      "loss": 0.4602,
      "step": 25493
    },
    {
      "epoch": 1.42759547541718,
      "grad_norm": 1.2356319427490234,
      "learning_rate": 9.855736842105263e-05,
      "loss": 0.4027,
      "step": 25494
    },
    {
      "epoch": 1.427651472729309,
      "grad_norm": 1.4177634716033936,
      "learning_rate": 9.85571052631579e-05,
      "loss": 0.3962,
      "step": 25495
    },
    {
      "epoch": 1.427707470041438,
      "grad_norm": 1.6410006284713745,
      "learning_rate": 9.855684210526316e-05,
      "loss": 0.4403,
      "step": 25496
    },
    {
      "epoch": 1.427763467353567,
      "grad_norm": 1.3921312093734741,
      "learning_rate": 9.855657894736842e-05,
      "loss": 0.4249,
      "step": 25497
    },
    {
      "epoch": 1.427819464665696,
      "grad_norm": 1.0755549669265747,
      "learning_rate": 9.855631578947368e-05,
      "loss": 0.3739,
      "step": 25498
    },
    {
      "epoch": 1.4278754619778251,
      "grad_norm": 1.423352599143982,
      "learning_rate": 9.855605263157896e-05,
      "loss": 0.514,
      "step": 25499
    },
    {
      "epoch": 1.4279314592899541,
      "grad_norm": 1.5200079679489136,
      "learning_rate": 9.855578947368421e-05,
      "loss": 0.5753,
      "step": 25500
    },
    {
      "epoch": 1.4279874566020831,
      "grad_norm": 1.4804867506027222,
      "learning_rate": 9.855552631578949e-05,
      "loss": 0.5534,
      "step": 25501
    },
    {
      "epoch": 1.4280434539142122,
      "grad_norm": 1.2663437128067017,
      "learning_rate": 9.855526315789473e-05,
      "loss": 0.4266,
      "step": 25502
    },
    {
      "epoch": 1.4280994512263412,
      "grad_norm": 1.291031002998352,
      "learning_rate": 9.855500000000001e-05,
      "loss": 0.4723,
      "step": 25503
    },
    {
      "epoch": 1.4281554485384702,
      "grad_norm": 1.6855638027191162,
      "learning_rate": 9.855473684210527e-05,
      "loss": 0.4717,
      "step": 25504
    },
    {
      "epoch": 1.4282114458505992,
      "grad_norm": 1.5187420845031738,
      "learning_rate": 9.855447368421054e-05,
      "loss": 0.4999,
      "step": 25505
    },
    {
      "epoch": 1.4282674431627282,
      "grad_norm": 1.2003096342086792,
      "learning_rate": 9.855421052631579e-05,
      "loss": 0.4537,
      "step": 25506
    },
    {
      "epoch": 1.4283234404748573,
      "grad_norm": 1.055364966392517,
      "learning_rate": 9.855394736842106e-05,
      "loss": 0.4704,
      "step": 25507
    },
    {
      "epoch": 1.4283794377869863,
      "grad_norm": 1.474247932434082,
      "learning_rate": 9.855368421052632e-05,
      "loss": 0.577,
      "step": 25508
    },
    {
      "epoch": 1.4284354350991153,
      "grad_norm": 1.2746540307998657,
      "learning_rate": 9.855342105263158e-05,
      "loss": 0.436,
      "step": 25509
    },
    {
      "epoch": 1.4284914324112443,
      "grad_norm": 1.1947314739227295,
      "learning_rate": 9.855315789473685e-05,
      "loss": 0.4319,
      "step": 25510
    },
    {
      "epoch": 1.4285474297233733,
      "grad_norm": 1.2716975212097168,
      "learning_rate": 9.85528947368421e-05,
      "loss": 0.4281,
      "step": 25511
    },
    {
      "epoch": 1.4286034270355024,
      "grad_norm": 1.1532578468322754,
      "learning_rate": 9.855263157894737e-05,
      "loss": 0.5804,
      "step": 25512
    },
    {
      "epoch": 1.4286594243476314,
      "grad_norm": 1.9912092685699463,
      "learning_rate": 9.855236842105263e-05,
      "loss": 0.5872,
      "step": 25513
    },
    {
      "epoch": 1.4287154216597604,
      "grad_norm": 1.3635388612747192,
      "learning_rate": 9.85521052631579e-05,
      "loss": 0.4668,
      "step": 25514
    },
    {
      "epoch": 1.4287714189718894,
      "grad_norm": 1.347943902015686,
      "learning_rate": 9.855184210526317e-05,
      "loss": 0.4532,
      "step": 25515
    },
    {
      "epoch": 1.4288274162840184,
      "grad_norm": 1.3383828401565552,
      "learning_rate": 9.855157894736842e-05,
      "loss": 0.4915,
      "step": 25516
    },
    {
      "epoch": 1.4288834135961475,
      "grad_norm": 1.4385755062103271,
      "learning_rate": 9.855131578947368e-05,
      "loss": 0.4985,
      "step": 25517
    },
    {
      "epoch": 1.4289394109082765,
      "grad_norm": 1.3665130138397217,
      "learning_rate": 9.855105263157896e-05,
      "loss": 0.6255,
      "step": 25518
    },
    {
      "epoch": 1.4289954082204055,
      "grad_norm": 1.445753574371338,
      "learning_rate": 9.855078947368422e-05,
      "loss": 0.3899,
      "step": 25519
    },
    {
      "epoch": 1.4290514055325345,
      "grad_norm": 1.3229690790176392,
      "learning_rate": 9.855052631578948e-05,
      "loss": 0.3873,
      "step": 25520
    },
    {
      "epoch": 1.4291074028446635,
      "grad_norm": 1.4007450342178345,
      "learning_rate": 9.855026315789474e-05,
      "loss": 0.4926,
      "step": 25521
    },
    {
      "epoch": 1.4291634001567926,
      "grad_norm": 1.5794925689697266,
      "learning_rate": 9.855000000000001e-05,
      "loss": 0.486,
      "step": 25522
    },
    {
      "epoch": 1.4292193974689216,
      "grad_norm": 2.4457929134368896,
      "learning_rate": 9.854973684210527e-05,
      "loss": 0.4961,
      "step": 25523
    },
    {
      "epoch": 1.4292753947810506,
      "grad_norm": 1.2906749248504639,
      "learning_rate": 9.854947368421053e-05,
      "loss": 0.4901,
      "step": 25524
    },
    {
      "epoch": 1.4293313920931796,
      "grad_norm": 1.2493500709533691,
      "learning_rate": 9.854921052631579e-05,
      "loss": 0.4229,
      "step": 25525
    },
    {
      "epoch": 1.4293873894053086,
      "grad_norm": 1.722590684890747,
      "learning_rate": 9.854894736842106e-05,
      "loss": 0.8219,
      "step": 25526
    },
    {
      "epoch": 1.4294433867174376,
      "grad_norm": 1.228349208831787,
      "learning_rate": 9.854868421052632e-05,
      "loss": 0.4558,
      "step": 25527
    },
    {
      "epoch": 1.4294993840295667,
      "grad_norm": 1.3449047803878784,
      "learning_rate": 9.854842105263158e-05,
      "loss": 0.5169,
      "step": 25528
    },
    {
      "epoch": 1.4295553813416957,
      "grad_norm": 1.2607173919677734,
      "learning_rate": 9.854815789473684e-05,
      "loss": 0.4035,
      "step": 25529
    },
    {
      "epoch": 1.4296113786538247,
      "grad_norm": 1.19171941280365,
      "learning_rate": 9.85478947368421e-05,
      "loss": 0.3917,
      "step": 25530
    },
    {
      "epoch": 1.4296673759659537,
      "grad_norm": 1.2546151876449585,
      "learning_rate": 9.854763157894737e-05,
      "loss": 0.4492,
      "step": 25531
    },
    {
      "epoch": 1.4297233732780827,
      "grad_norm": 1.5277622938156128,
      "learning_rate": 9.854736842105263e-05,
      "loss": 0.5667,
      "step": 25532
    },
    {
      "epoch": 1.4297793705902118,
      "grad_norm": 1.401011347770691,
      "learning_rate": 9.854710526315791e-05,
      "loss": 0.4184,
      "step": 25533
    },
    {
      "epoch": 1.4298353679023408,
      "grad_norm": 1.8958184719085693,
      "learning_rate": 9.854684210526315e-05,
      "loss": 0.497,
      "step": 25534
    },
    {
      "epoch": 1.4298913652144698,
      "grad_norm": 1.5363988876342773,
      "learning_rate": 9.854657894736843e-05,
      "loss": 0.4567,
      "step": 25535
    },
    {
      "epoch": 1.4299473625265988,
      "grad_norm": 1.2325515747070312,
      "learning_rate": 9.854631578947369e-05,
      "loss": 0.4315,
      "step": 25536
    },
    {
      "epoch": 1.4300033598387278,
      "grad_norm": 1.4724162817001343,
      "learning_rate": 9.854605263157896e-05,
      "loss": 0.3764,
      "step": 25537
    },
    {
      "epoch": 1.4300593571508569,
      "grad_norm": 1.2884479761123657,
      "learning_rate": 9.854578947368422e-05,
      "loss": 0.37,
      "step": 25538
    },
    {
      "epoch": 1.4301153544629859,
      "grad_norm": 1.3686155080795288,
      "learning_rate": 9.854552631578948e-05,
      "loss": 0.5295,
      "step": 25539
    },
    {
      "epoch": 1.430171351775115,
      "grad_norm": 1.3763350248336792,
      "learning_rate": 9.854526315789474e-05,
      "loss": 0.5457,
      "step": 25540
    },
    {
      "epoch": 1.430227349087244,
      "grad_norm": 2.0514578819274902,
      "learning_rate": 9.854500000000001e-05,
      "loss": 0.4575,
      "step": 25541
    },
    {
      "epoch": 1.430283346399373,
      "grad_norm": 1.3097364902496338,
      "learning_rate": 9.854473684210527e-05,
      "loss": 0.4334,
      "step": 25542
    },
    {
      "epoch": 1.430339343711502,
      "grad_norm": 1.1507169008255005,
      "learning_rate": 9.854447368421053e-05,
      "loss": 0.5702,
      "step": 25543
    },
    {
      "epoch": 1.430395341023631,
      "grad_norm": 1.8972383737564087,
      "learning_rate": 9.854421052631579e-05,
      "loss": 0.4563,
      "step": 25544
    },
    {
      "epoch": 1.43045133833576,
      "grad_norm": 1.3155434131622314,
      "learning_rate": 9.854394736842105e-05,
      "loss": 0.5349,
      "step": 25545
    },
    {
      "epoch": 1.430507335647889,
      "grad_norm": 5.150670051574707,
      "learning_rate": 9.854368421052632e-05,
      "loss": 0.5007,
      "step": 25546
    },
    {
      "epoch": 1.430563332960018,
      "grad_norm": 1.504660725593567,
      "learning_rate": 9.854342105263158e-05,
      "loss": 0.4913,
      "step": 25547
    },
    {
      "epoch": 1.430619330272147,
      "grad_norm": 1.3363568782806396,
      "learning_rate": 9.854315789473684e-05,
      "loss": 0.405,
      "step": 25548
    },
    {
      "epoch": 1.430675327584276,
      "grad_norm": 1.1625854969024658,
      "learning_rate": 9.85428947368421e-05,
      "loss": 0.4709,
      "step": 25549
    },
    {
      "epoch": 1.430731324896405,
      "grad_norm": 1.4491811990737915,
      "learning_rate": 9.854263157894738e-05,
      "loss": 0.3984,
      "step": 25550
    },
    {
      "epoch": 1.430787322208534,
      "grad_norm": 1.5567790269851685,
      "learning_rate": 9.854236842105264e-05,
      "loss": 0.4872,
      "step": 25551
    },
    {
      "epoch": 1.4308433195206631,
      "grad_norm": 1.5600478649139404,
      "learning_rate": 9.85421052631579e-05,
      "loss": 0.4427,
      "step": 25552
    },
    {
      "epoch": 1.4308993168327921,
      "grad_norm": 1.2961442470550537,
      "learning_rate": 9.854184210526316e-05,
      "loss": 0.3906,
      "step": 25553
    },
    {
      "epoch": 1.4309553141449212,
      "grad_norm": 1.2450376749038696,
      "learning_rate": 9.854157894736843e-05,
      "loss": 0.5177,
      "step": 25554
    },
    {
      "epoch": 1.4310113114570502,
      "grad_norm": 1.8768779039382935,
      "learning_rate": 9.854131578947369e-05,
      "loss": 0.4998,
      "step": 25555
    },
    {
      "epoch": 1.4310673087691792,
      "grad_norm": 1.3666603565216064,
      "learning_rate": 9.854105263157896e-05,
      "loss": 0.4617,
      "step": 25556
    },
    {
      "epoch": 1.4311233060813082,
      "grad_norm": 1.6488827466964722,
      "learning_rate": 9.854078947368421e-05,
      "loss": 0.6511,
      "step": 25557
    },
    {
      "epoch": 1.4311793033934372,
      "grad_norm": 1.5491153001785278,
      "learning_rate": 9.854052631578948e-05,
      "loss": 0.4367,
      "step": 25558
    },
    {
      "epoch": 1.4312353007055663,
      "grad_norm": 1.3581222295761108,
      "learning_rate": 9.854026315789474e-05,
      "loss": 0.5209,
      "step": 25559
    },
    {
      "epoch": 1.4312912980176953,
      "grad_norm": 1.5273687839508057,
      "learning_rate": 9.854000000000002e-05,
      "loss": 0.4487,
      "step": 25560
    },
    {
      "epoch": 1.4313472953298243,
      "grad_norm": 1.622373342514038,
      "learning_rate": 9.853973684210526e-05,
      "loss": 0.4901,
      "step": 25561
    },
    {
      "epoch": 1.4314032926419533,
      "grad_norm": 1.276328682899475,
      "learning_rate": 9.853947368421052e-05,
      "loss": 0.559,
      "step": 25562
    },
    {
      "epoch": 1.4314592899540823,
      "grad_norm": 1.4135698080062866,
      "learning_rate": 9.85392105263158e-05,
      "loss": 0.4358,
      "step": 25563
    },
    {
      "epoch": 1.4315152872662114,
      "grad_norm": 1.2505730390548706,
      "learning_rate": 9.853894736842105e-05,
      "loss": 0.494,
      "step": 25564
    },
    {
      "epoch": 1.4315712845783404,
      "grad_norm": 1.347671389579773,
      "learning_rate": 9.853868421052633e-05,
      "loss": 0.4177,
      "step": 25565
    },
    {
      "epoch": 1.4316272818904694,
      "grad_norm": 1.7066562175750732,
      "learning_rate": 9.853842105263157e-05,
      "loss": 0.9907,
      "step": 25566
    },
    {
      "epoch": 1.4316832792025984,
      "grad_norm": 1.3701480627059937,
      "learning_rate": 9.853815789473685e-05,
      "loss": 0.4696,
      "step": 25567
    },
    {
      "epoch": 1.4317392765147272,
      "grad_norm": 1.3462611436843872,
      "learning_rate": 9.85378947368421e-05,
      "loss": 0.4758,
      "step": 25568
    },
    {
      "epoch": 1.4317952738268562,
      "grad_norm": 1.2861237525939941,
      "learning_rate": 9.853763157894738e-05,
      "loss": 0.3977,
      "step": 25569
    },
    {
      "epoch": 1.4318512711389852,
      "grad_norm": 1.3685340881347656,
      "learning_rate": 9.853736842105264e-05,
      "loss": 0.4813,
      "step": 25570
    },
    {
      "epoch": 1.4319072684511143,
      "grad_norm": 1.4137378931045532,
      "learning_rate": 9.85371052631579e-05,
      "loss": 0.5633,
      "step": 25571
    },
    {
      "epoch": 1.4319632657632433,
      "grad_norm": 1.2090106010437012,
      "learning_rate": 9.853684210526316e-05,
      "loss": 0.4796,
      "step": 25572
    },
    {
      "epoch": 1.4320192630753723,
      "grad_norm": 1.4746901988983154,
      "learning_rate": 9.853657894736843e-05,
      "loss": 0.4088,
      "step": 25573
    },
    {
      "epoch": 1.4320752603875013,
      "grad_norm": 1.1521053314208984,
      "learning_rate": 9.853631578947369e-05,
      "loss": 0.3851,
      "step": 25574
    },
    {
      "epoch": 1.4321312576996303,
      "grad_norm": 1.3855780363082886,
      "learning_rate": 9.853605263157895e-05,
      "loss": 0.4598,
      "step": 25575
    },
    {
      "epoch": 1.4321872550117594,
      "grad_norm": 1.3446911573410034,
      "learning_rate": 9.853578947368421e-05,
      "loss": 0.4721,
      "step": 25576
    },
    {
      "epoch": 1.4322432523238884,
      "grad_norm": 1.3428047895431519,
      "learning_rate": 9.853552631578948e-05,
      "loss": 0.5576,
      "step": 25577
    },
    {
      "epoch": 1.4322992496360174,
      "grad_norm": 1.3166146278381348,
      "learning_rate": 9.853526315789474e-05,
      "loss": 0.4094,
      "step": 25578
    },
    {
      "epoch": 1.4323552469481464,
      "grad_norm": 1.4172027111053467,
      "learning_rate": 9.8535e-05,
      "loss": 0.4247,
      "step": 25579
    },
    {
      "epoch": 1.4324112442602754,
      "grad_norm": 1.9372472763061523,
      "learning_rate": 9.853473684210526e-05,
      "loss": 0.4461,
      "step": 25580
    },
    {
      "epoch": 1.4324672415724045,
      "grad_norm": 1.5230001211166382,
      "learning_rate": 9.853447368421052e-05,
      "loss": 0.4819,
      "step": 25581
    },
    {
      "epoch": 1.4325232388845335,
      "grad_norm": 1.5996944904327393,
      "learning_rate": 9.85342105263158e-05,
      "loss": 0.5124,
      "step": 25582
    },
    {
      "epoch": 1.4325792361966625,
      "grad_norm": 1.3001734018325806,
      "learning_rate": 9.853394736842106e-05,
      "loss": 0.4613,
      "step": 25583
    },
    {
      "epoch": 1.4326352335087915,
      "grad_norm": 1.233330249786377,
      "learning_rate": 9.853368421052632e-05,
      "loss": 0.3952,
      "step": 25584
    },
    {
      "epoch": 1.4326912308209205,
      "grad_norm": 1.220557451248169,
      "learning_rate": 9.853342105263158e-05,
      "loss": 0.4315,
      "step": 25585
    },
    {
      "epoch": 1.4327472281330496,
      "grad_norm": 1.838141679763794,
      "learning_rate": 9.853315789473685e-05,
      "loss": 0.5661,
      "step": 25586
    },
    {
      "epoch": 1.4328032254451786,
      "grad_norm": 1.4783620834350586,
      "learning_rate": 9.853289473684211e-05,
      "loss": 0.5511,
      "step": 25587
    },
    {
      "epoch": 1.4328592227573076,
      "grad_norm": 1.6853631734848022,
      "learning_rate": 9.853263157894738e-05,
      "loss": 0.666,
      "step": 25588
    },
    {
      "epoch": 1.4329152200694366,
      "grad_norm": 1.6526273488998413,
      "learning_rate": 9.853236842105263e-05,
      "loss": 0.4457,
      "step": 25589
    },
    {
      "epoch": 1.4329712173815656,
      "grad_norm": 1.7893047332763672,
      "learning_rate": 9.85321052631579e-05,
      "loss": 0.6248,
      "step": 25590
    },
    {
      "epoch": 1.4330272146936947,
      "grad_norm": 1.2511688470840454,
      "learning_rate": 9.853184210526316e-05,
      "loss": 0.4484,
      "step": 25591
    },
    {
      "epoch": 1.4330832120058237,
      "grad_norm": 2.7740895748138428,
      "learning_rate": 9.853157894736844e-05,
      "loss": 0.3603,
      "step": 25592
    },
    {
      "epoch": 1.4331392093179527,
      "grad_norm": 1.4321290254592896,
      "learning_rate": 9.85313157894737e-05,
      "loss": 0.4985,
      "step": 25593
    },
    {
      "epoch": 1.4331952066300817,
      "grad_norm": 1.5247721672058105,
      "learning_rate": 9.853105263157895e-05,
      "loss": 0.6196,
      "step": 25594
    },
    {
      "epoch": 1.4332512039422107,
      "grad_norm": 1.2495554685592651,
      "learning_rate": 9.853078947368421e-05,
      "loss": 0.4114,
      "step": 25595
    },
    {
      "epoch": 1.4333072012543397,
      "grad_norm": 1.3315740823745728,
      "learning_rate": 9.853052631578947e-05,
      "loss": 0.4687,
      "step": 25596
    },
    {
      "epoch": 1.4333631985664688,
      "grad_norm": 1.3829666376113892,
      "learning_rate": 9.853026315789475e-05,
      "loss": 0.4416,
      "step": 25597
    },
    {
      "epoch": 1.4334191958785978,
      "grad_norm": 1.4854471683502197,
      "learning_rate": 9.853e-05,
      "loss": 0.5246,
      "step": 25598
    },
    {
      "epoch": 1.4334751931907268,
      "grad_norm": 1.2463632822036743,
      "learning_rate": 9.852973684210527e-05,
      "loss": 0.438,
      "step": 25599
    },
    {
      "epoch": 1.4335311905028558,
      "grad_norm": 1.2956230640411377,
      "learning_rate": 9.852947368421053e-05,
      "loss": 0.524,
      "step": 25600
    },
    {
      "epoch": 1.4335871878149848,
      "grad_norm": 1.356801152229309,
      "learning_rate": 9.85292105263158e-05,
      "loss": 0.4658,
      "step": 25601
    },
    {
      "epoch": 1.4336431851271139,
      "grad_norm": 1.358520746231079,
      "learning_rate": 9.852894736842106e-05,
      "loss": 0.4039,
      "step": 25602
    },
    {
      "epoch": 1.4336991824392429,
      "grad_norm": 1.3137705326080322,
      "learning_rate": 9.852868421052632e-05,
      "loss": 0.4378,
      "step": 25603
    },
    {
      "epoch": 1.433755179751372,
      "grad_norm": 1.3565123081207275,
      "learning_rate": 9.852842105263158e-05,
      "loss": 0.4547,
      "step": 25604
    },
    {
      "epoch": 1.433811177063501,
      "grad_norm": 3.7248382568359375,
      "learning_rate": 9.852815789473685e-05,
      "loss": 0.6575,
      "step": 25605
    },
    {
      "epoch": 1.43386717437563,
      "grad_norm": 1.2829116582870483,
      "learning_rate": 9.852789473684211e-05,
      "loss": 0.405,
      "step": 25606
    },
    {
      "epoch": 1.433923171687759,
      "grad_norm": 2.3231306076049805,
      "learning_rate": 9.852763157894737e-05,
      "loss": 0.6263,
      "step": 25607
    },
    {
      "epoch": 1.433979168999888,
      "grad_norm": 1.3743995428085327,
      "learning_rate": 9.852736842105263e-05,
      "loss": 0.4791,
      "step": 25608
    },
    {
      "epoch": 1.434035166312017,
      "grad_norm": 1.517982840538025,
      "learning_rate": 9.85271052631579e-05,
      "loss": 0.5326,
      "step": 25609
    },
    {
      "epoch": 1.434091163624146,
      "grad_norm": 1.328944206237793,
      "learning_rate": 9.852684210526316e-05,
      "loss": 0.4551,
      "step": 25610
    },
    {
      "epoch": 1.434147160936275,
      "grad_norm": 1.4466986656188965,
      "learning_rate": 9.852657894736842e-05,
      "loss": 0.3798,
      "step": 25611
    },
    {
      "epoch": 1.434203158248404,
      "grad_norm": 1.442425012588501,
      "learning_rate": 9.852631578947368e-05,
      "loss": 0.478,
      "step": 25612
    },
    {
      "epoch": 1.434259155560533,
      "grad_norm": 1.1874006986618042,
      "learning_rate": 9.852605263157894e-05,
      "loss": 0.3621,
      "step": 25613
    },
    {
      "epoch": 1.434315152872662,
      "grad_norm": 1.4170418977737427,
      "learning_rate": 9.852578947368422e-05,
      "loss": 0.4213,
      "step": 25614
    },
    {
      "epoch": 1.434371150184791,
      "grad_norm": 1.4486767053604126,
      "learning_rate": 9.852552631578948e-05,
      "loss": 0.4419,
      "step": 25615
    },
    {
      "epoch": 1.4344271474969201,
      "grad_norm": 1.645767092704773,
      "learning_rate": 9.852526315789474e-05,
      "loss": 0.4913,
      "step": 25616
    },
    {
      "epoch": 1.4344831448090491,
      "grad_norm": 1.2231667041778564,
      "learning_rate": 9.8525e-05,
      "loss": 0.4702,
      "step": 25617
    },
    {
      "epoch": 1.4345391421211782,
      "grad_norm": 1.3708308935165405,
      "learning_rate": 9.852473684210527e-05,
      "loss": 0.4443,
      "step": 25618
    },
    {
      "epoch": 1.4345951394333072,
      "grad_norm": 1.3079771995544434,
      "learning_rate": 9.852447368421053e-05,
      "loss": 0.4473,
      "step": 25619
    },
    {
      "epoch": 1.4346511367454362,
      "grad_norm": 1.2420459985733032,
      "learning_rate": 9.85242105263158e-05,
      "loss": 0.4299,
      "step": 25620
    },
    {
      "epoch": 1.4347071340575652,
      "grad_norm": 1.3548718690872192,
      "learning_rate": 9.852394736842105e-05,
      "loss": 0.3791,
      "step": 25621
    },
    {
      "epoch": 1.4347631313696942,
      "grad_norm": 1.2706632614135742,
      "learning_rate": 9.852368421052632e-05,
      "loss": 0.4779,
      "step": 25622
    },
    {
      "epoch": 1.4348191286818233,
      "grad_norm": 1.0939626693725586,
      "learning_rate": 9.852342105263158e-05,
      "loss": 0.4237,
      "step": 25623
    },
    {
      "epoch": 1.4348751259939523,
      "grad_norm": 1.3191313743591309,
      "learning_rate": 9.852315789473685e-05,
      "loss": 0.4263,
      "step": 25624
    },
    {
      "epoch": 1.4349311233060813,
      "grad_norm": 1.2442973852157593,
      "learning_rate": 9.852289473684211e-05,
      "loss": 0.3207,
      "step": 25625
    },
    {
      "epoch": 1.4349871206182103,
      "grad_norm": 1.3126877546310425,
      "learning_rate": 9.852263157894737e-05,
      "loss": 0.4555,
      "step": 25626
    },
    {
      "epoch": 1.4350431179303393,
      "grad_norm": 1.4616296291351318,
      "learning_rate": 9.852236842105263e-05,
      "loss": 0.3951,
      "step": 25627
    },
    {
      "epoch": 1.4350991152424684,
      "grad_norm": 1.2831814289093018,
      "learning_rate": 9.852210526315791e-05,
      "loss": 0.423,
      "step": 25628
    },
    {
      "epoch": 1.4351551125545974,
      "grad_norm": 1.1951026916503906,
      "learning_rate": 9.852184210526317e-05,
      "loss": 0.3939,
      "step": 25629
    },
    {
      "epoch": 1.4352111098667264,
      "grad_norm": 1.3095088005065918,
      "learning_rate": 9.852157894736843e-05,
      "loss": 0.5734,
      "step": 25630
    },
    {
      "epoch": 1.4352671071788554,
      "grad_norm": 1.3465334177017212,
      "learning_rate": 9.852131578947369e-05,
      "loss": 0.4808,
      "step": 25631
    },
    {
      "epoch": 1.4353231044909844,
      "grad_norm": 1.1294889450073242,
      "learning_rate": 9.852105263157895e-05,
      "loss": 0.5764,
      "step": 25632
    },
    {
      "epoch": 1.4353791018031135,
      "grad_norm": 1.4509919881820679,
      "learning_rate": 9.852078947368422e-05,
      "loss": 0.543,
      "step": 25633
    },
    {
      "epoch": 1.4354350991152425,
      "grad_norm": 1.4717315435409546,
      "learning_rate": 9.852052631578948e-05,
      "loss": 0.5028,
      "step": 25634
    },
    {
      "epoch": 1.4354910964273715,
      "grad_norm": 1.5489089488983154,
      "learning_rate": 9.852026315789474e-05,
      "loss": 0.6137,
      "step": 25635
    },
    {
      "epoch": 1.4355470937395005,
      "grad_norm": 1.3831948041915894,
      "learning_rate": 9.852e-05,
      "loss": 0.3883,
      "step": 25636
    },
    {
      "epoch": 1.4356030910516295,
      "grad_norm": 1.2413886785507202,
      "learning_rate": 9.851973684210527e-05,
      "loss": 0.5447,
      "step": 25637
    },
    {
      "epoch": 1.4356590883637586,
      "grad_norm": 1.3069621324539185,
      "learning_rate": 9.851947368421053e-05,
      "loss": 0.469,
      "step": 25638
    },
    {
      "epoch": 1.4357150856758876,
      "grad_norm": 1.530605435371399,
      "learning_rate": 9.851921052631579e-05,
      "loss": 0.5449,
      "step": 25639
    },
    {
      "epoch": 1.4357710829880166,
      "grad_norm": 1.1206198930740356,
      "learning_rate": 9.851894736842105e-05,
      "loss": 0.41,
      "step": 25640
    },
    {
      "epoch": 1.4358270803001456,
      "grad_norm": 1.5072031021118164,
      "learning_rate": 9.851868421052632e-05,
      "loss": 0.4225,
      "step": 25641
    },
    {
      "epoch": 1.4358830776122746,
      "grad_norm": 1.2919793128967285,
      "learning_rate": 9.851842105263158e-05,
      "loss": 0.4009,
      "step": 25642
    },
    {
      "epoch": 1.4359390749244036,
      "grad_norm": 1.3727933168411255,
      "learning_rate": 9.851815789473686e-05,
      "loss": 0.4062,
      "step": 25643
    },
    {
      "epoch": 1.4359950722365327,
      "grad_norm": 1.3184592723846436,
      "learning_rate": 9.85178947368421e-05,
      "loss": 0.5137,
      "step": 25644
    },
    {
      "epoch": 1.4360510695486617,
      "grad_norm": 1.1559818983078003,
      "learning_rate": 9.851763157894738e-05,
      "loss": 0.4556,
      "step": 25645
    },
    {
      "epoch": 1.4361070668607907,
      "grad_norm": 1.5235875844955444,
      "learning_rate": 9.851736842105264e-05,
      "loss": 0.459,
      "step": 25646
    },
    {
      "epoch": 1.4361630641729197,
      "grad_norm": 1.1688374280929565,
      "learning_rate": 9.851710526315791e-05,
      "loss": 0.3956,
      "step": 25647
    },
    {
      "epoch": 1.4362190614850487,
      "grad_norm": 1.375508427619934,
      "learning_rate": 9.851684210526317e-05,
      "loss": 0.43,
      "step": 25648
    },
    {
      "epoch": 1.4362750587971778,
      "grad_norm": 1.9250904321670532,
      "learning_rate": 9.851657894736842e-05,
      "loss": 0.4892,
      "step": 25649
    },
    {
      "epoch": 1.4363310561093068,
      "grad_norm": 1.5735348463058472,
      "learning_rate": 9.851631578947369e-05,
      "loss": 0.6569,
      "step": 25650
    },
    {
      "epoch": 1.4363870534214358,
      "grad_norm": 1.421795129776001,
      "learning_rate": 9.851605263157895e-05,
      "loss": 0.4173,
      "step": 25651
    },
    {
      "epoch": 1.4364430507335648,
      "grad_norm": 1.58512282371521,
      "learning_rate": 9.851578947368422e-05,
      "loss": 0.393,
      "step": 25652
    },
    {
      "epoch": 1.4364990480456938,
      "grad_norm": 1.1600414514541626,
      "learning_rate": 9.851552631578947e-05,
      "loss": 0.4621,
      "step": 25653
    },
    {
      "epoch": 1.4365550453578229,
      "grad_norm": 1.1983907222747803,
      "learning_rate": 9.851526315789474e-05,
      "loss": 0.4011,
      "step": 25654
    },
    {
      "epoch": 1.4366110426699519,
      "grad_norm": 1.5610764026641846,
      "learning_rate": 9.8515e-05,
      "loss": 0.6253,
      "step": 25655
    },
    {
      "epoch": 1.436667039982081,
      "grad_norm": 1.3131215572357178,
      "learning_rate": 9.851473684210527e-05,
      "loss": 0.4615,
      "step": 25656
    },
    {
      "epoch": 1.43672303729421,
      "grad_norm": 1.3105984926223755,
      "learning_rate": 9.851447368421053e-05,
      "loss": 0.4597,
      "step": 25657
    },
    {
      "epoch": 1.436779034606339,
      "grad_norm": 1.5818188190460205,
      "learning_rate": 9.85142105263158e-05,
      "loss": 0.4973,
      "step": 25658
    },
    {
      "epoch": 1.436835031918468,
      "grad_norm": 1.244718313217163,
      "learning_rate": 9.851394736842105e-05,
      "loss": 0.4456,
      "step": 25659
    },
    {
      "epoch": 1.436891029230597,
      "grad_norm": 1.3325386047363281,
      "learning_rate": 9.851368421052633e-05,
      "loss": 0.4782,
      "step": 25660
    },
    {
      "epoch": 1.436947026542726,
      "grad_norm": 1.2712047100067139,
      "learning_rate": 9.851342105263159e-05,
      "loss": 0.4101,
      "step": 25661
    },
    {
      "epoch": 1.437003023854855,
      "grad_norm": 1.068215012550354,
      "learning_rate": 9.851315789473685e-05,
      "loss": 0.343,
      "step": 25662
    },
    {
      "epoch": 1.437059021166984,
      "grad_norm": 1.4893957376480103,
      "learning_rate": 9.85128947368421e-05,
      "loss": 0.5978,
      "step": 25663
    },
    {
      "epoch": 1.437115018479113,
      "grad_norm": 1.3513212203979492,
      "learning_rate": 9.851263157894738e-05,
      "loss": 0.4153,
      "step": 25664
    },
    {
      "epoch": 1.437171015791242,
      "grad_norm": 1.544857382774353,
      "learning_rate": 9.851236842105264e-05,
      "loss": 0.5701,
      "step": 25665
    },
    {
      "epoch": 1.437227013103371,
      "grad_norm": 1.2158294916152954,
      "learning_rate": 9.85121052631579e-05,
      "loss": 0.3734,
      "step": 25666
    },
    {
      "epoch": 1.4372830104155,
      "grad_norm": 2.1777915954589844,
      "learning_rate": 9.851184210526316e-05,
      "loss": 0.7021,
      "step": 25667
    },
    {
      "epoch": 1.4373390077276291,
      "grad_norm": 4.20610237121582,
      "learning_rate": 9.851157894736842e-05,
      "loss": 0.5588,
      "step": 25668
    },
    {
      "epoch": 1.4373950050397581,
      "grad_norm": 1.689322829246521,
      "learning_rate": 9.851131578947369e-05,
      "loss": 0.4976,
      "step": 25669
    },
    {
      "epoch": 1.4374510023518872,
      "grad_norm": 1.2596203088760376,
      "learning_rate": 9.851105263157895e-05,
      "loss": 0.4639,
      "step": 25670
    },
    {
      "epoch": 1.4375069996640162,
      "grad_norm": 1.1975135803222656,
      "learning_rate": 9.851078947368421e-05,
      "loss": 0.4729,
      "step": 25671
    },
    {
      "epoch": 1.4375629969761452,
      "grad_norm": 1.586650013923645,
      "learning_rate": 9.851052631578947e-05,
      "loss": 0.6526,
      "step": 25672
    },
    {
      "epoch": 1.4376189942882742,
      "grad_norm": 1.5360716581344604,
      "learning_rate": 9.851026315789474e-05,
      "loss": 0.4654,
      "step": 25673
    },
    {
      "epoch": 1.4376749916004032,
      "grad_norm": 1.3076252937316895,
      "learning_rate": 9.851e-05,
      "loss": 0.4586,
      "step": 25674
    },
    {
      "epoch": 1.4377309889125323,
      "grad_norm": 1.2574946880340576,
      "learning_rate": 9.850973684210528e-05,
      "loss": 0.394,
      "step": 25675
    },
    {
      "epoch": 1.4377869862246613,
      "grad_norm": 1.2691985368728638,
      "learning_rate": 9.850947368421052e-05,
      "loss": 0.4306,
      "step": 25676
    },
    {
      "epoch": 1.4378429835367903,
      "grad_norm": 1.185869812965393,
      "learning_rate": 9.85092105263158e-05,
      "loss": 0.3659,
      "step": 25677
    },
    {
      "epoch": 1.4378989808489193,
      "grad_norm": 1.2608500719070435,
      "learning_rate": 9.850894736842106e-05,
      "loss": 0.3736,
      "step": 25678
    },
    {
      "epoch": 1.4379549781610483,
      "grad_norm": 1.475943922996521,
      "learning_rate": 9.850868421052633e-05,
      "loss": 0.5287,
      "step": 25679
    },
    {
      "epoch": 1.4380109754731774,
      "grad_norm": 1.4908572435379028,
      "learning_rate": 9.850842105263159e-05,
      "loss": 0.5296,
      "step": 25680
    },
    {
      "epoch": 1.4380669727853064,
      "grad_norm": 1.2638475894927979,
      "learning_rate": 9.850815789473685e-05,
      "loss": 0.4952,
      "step": 25681
    },
    {
      "epoch": 1.4381229700974354,
      "grad_norm": 1.2012163400650024,
      "learning_rate": 9.850789473684211e-05,
      "loss": 0.5255,
      "step": 25682
    },
    {
      "epoch": 1.4381789674095644,
      "grad_norm": 2.8469507694244385,
      "learning_rate": 9.850763157894737e-05,
      "loss": 0.523,
      "step": 25683
    },
    {
      "epoch": 1.4382349647216934,
      "grad_norm": 1.1035187244415283,
      "learning_rate": 9.850736842105264e-05,
      "loss": 0.3706,
      "step": 25684
    },
    {
      "epoch": 1.4382909620338225,
      "grad_norm": 1.0298693180084229,
      "learning_rate": 9.85071052631579e-05,
      "loss": 0.3495,
      "step": 25685
    },
    {
      "epoch": 1.4383469593459515,
      "grad_norm": 1.5727527141571045,
      "learning_rate": 9.850684210526316e-05,
      "loss": 0.6739,
      "step": 25686
    },
    {
      "epoch": 1.4384029566580805,
      "grad_norm": 1.750372052192688,
      "learning_rate": 9.850657894736842e-05,
      "loss": 0.6113,
      "step": 25687
    },
    {
      "epoch": 1.4384589539702095,
      "grad_norm": 1.5725183486938477,
      "learning_rate": 9.85063157894737e-05,
      "loss": 0.5526,
      "step": 25688
    },
    {
      "epoch": 1.4385149512823385,
      "grad_norm": 1.3974330425262451,
      "learning_rate": 9.850605263157895e-05,
      "loss": 0.4032,
      "step": 25689
    },
    {
      "epoch": 1.4385709485944675,
      "grad_norm": 1.3261353969573975,
      "learning_rate": 9.850578947368421e-05,
      "loss": 0.4921,
      "step": 25690
    },
    {
      "epoch": 1.4386269459065966,
      "grad_norm": 1.3843958377838135,
      "learning_rate": 9.850552631578947e-05,
      "loss": 0.4347,
      "step": 25691
    },
    {
      "epoch": 1.4386829432187254,
      "grad_norm": 1.2524871826171875,
      "learning_rate": 9.850526315789475e-05,
      "loss": 0.4132,
      "step": 25692
    },
    {
      "epoch": 1.4387389405308544,
      "grad_norm": 1.102995753288269,
      "learning_rate": 9.8505e-05,
      "loss": 0.3919,
      "step": 25693
    },
    {
      "epoch": 1.4387949378429834,
      "grad_norm": 1.183918833732605,
      "learning_rate": 9.850473684210527e-05,
      "loss": 0.5171,
      "step": 25694
    },
    {
      "epoch": 1.4388509351551124,
      "grad_norm": 1.4520643949508667,
      "learning_rate": 9.850447368421053e-05,
      "loss": 0.4224,
      "step": 25695
    },
    {
      "epoch": 1.4389069324672414,
      "grad_norm": 1.2692177295684814,
      "learning_rate": 9.85042105263158e-05,
      "loss": 0.4775,
      "step": 25696
    },
    {
      "epoch": 1.4389629297793705,
      "grad_norm": 1.3434689044952393,
      "learning_rate": 9.850394736842106e-05,
      "loss": 0.4722,
      "step": 25697
    },
    {
      "epoch": 1.4390189270914995,
      "grad_norm": 1.5888018608093262,
      "learning_rate": 9.850368421052633e-05,
      "loss": 0.6064,
      "step": 25698
    },
    {
      "epoch": 1.4390749244036285,
      "grad_norm": 1.286224603652954,
      "learning_rate": 9.850342105263158e-05,
      "loss": 0.4078,
      "step": 25699
    },
    {
      "epoch": 1.4391309217157575,
      "grad_norm": 1.6147180795669556,
      "learning_rate": 9.850315789473684e-05,
      "loss": 0.6666,
      "step": 25700
    },
    {
      "epoch": 1.4391869190278865,
      "grad_norm": 1.2030236721038818,
      "learning_rate": 9.850289473684211e-05,
      "loss": 0.421,
      "step": 25701
    },
    {
      "epoch": 1.4392429163400156,
      "grad_norm": 1.427893042564392,
      "learning_rate": 9.850263157894737e-05,
      "loss": 0.4825,
      "step": 25702
    },
    {
      "epoch": 1.4392989136521446,
      "grad_norm": 1.3892394304275513,
      "learning_rate": 9.850236842105264e-05,
      "loss": 0.6434,
      "step": 25703
    },
    {
      "epoch": 1.4393549109642736,
      "grad_norm": 1.3688900470733643,
      "learning_rate": 9.850210526315789e-05,
      "loss": 0.5591,
      "step": 25704
    },
    {
      "epoch": 1.4394109082764026,
      "grad_norm": 1.286647081375122,
      "learning_rate": 9.850184210526316e-05,
      "loss": 0.5014,
      "step": 25705
    },
    {
      "epoch": 1.4394669055885316,
      "grad_norm": 1.2540347576141357,
      "learning_rate": 9.850157894736842e-05,
      "loss": 0.4826,
      "step": 25706
    },
    {
      "epoch": 1.4395229029006607,
      "grad_norm": 1.377650260925293,
      "learning_rate": 9.85013157894737e-05,
      "loss": 0.4497,
      "step": 25707
    },
    {
      "epoch": 1.4395789002127897,
      "grad_norm": 1.3599501848220825,
      "learning_rate": 9.850105263157894e-05,
      "loss": 0.4194,
      "step": 25708
    },
    {
      "epoch": 1.4396348975249187,
      "grad_norm": 1.356694221496582,
      "learning_rate": 9.850078947368422e-05,
      "loss": 0.3896,
      "step": 25709
    },
    {
      "epoch": 1.4396908948370477,
      "grad_norm": 1.2018773555755615,
      "learning_rate": 9.850052631578948e-05,
      "loss": 0.4454,
      "step": 25710
    },
    {
      "epoch": 1.4397468921491767,
      "grad_norm": 1.5941123962402344,
      "learning_rate": 9.850026315789475e-05,
      "loss": 0.5182,
      "step": 25711
    },
    {
      "epoch": 1.4398028894613057,
      "grad_norm": 1.4879815578460693,
      "learning_rate": 9.850000000000001e-05,
      "loss": 0.4818,
      "step": 25712
    },
    {
      "epoch": 1.4398588867734348,
      "grad_norm": 1.2622997760772705,
      "learning_rate": 9.849973684210527e-05,
      "loss": 0.4667,
      "step": 25713
    },
    {
      "epoch": 1.4399148840855638,
      "grad_norm": 1.4573948383331299,
      "learning_rate": 9.849947368421053e-05,
      "loss": 0.4054,
      "step": 25714
    },
    {
      "epoch": 1.4399708813976928,
      "grad_norm": 1.443192720413208,
      "learning_rate": 9.84992105263158e-05,
      "loss": 0.5356,
      "step": 25715
    },
    {
      "epoch": 1.4400268787098218,
      "grad_norm": 1.1048190593719482,
      "learning_rate": 9.849894736842106e-05,
      "loss": 0.3802,
      "step": 25716
    },
    {
      "epoch": 1.4400828760219508,
      "grad_norm": 1.1638503074645996,
      "learning_rate": 9.849868421052632e-05,
      "loss": 0.4394,
      "step": 25717
    },
    {
      "epoch": 1.4401388733340799,
      "grad_norm": 1.3394076824188232,
      "learning_rate": 9.849842105263158e-05,
      "loss": 0.5312,
      "step": 25718
    },
    {
      "epoch": 1.4401948706462089,
      "grad_norm": 1.2208266258239746,
      "learning_rate": 9.849815789473684e-05,
      "loss": 0.3839,
      "step": 25719
    },
    {
      "epoch": 1.440250867958338,
      "grad_norm": 1.4026274681091309,
      "learning_rate": 9.849789473684211e-05,
      "loss": 0.4958,
      "step": 25720
    },
    {
      "epoch": 1.440306865270467,
      "grad_norm": 1.4235761165618896,
      "learning_rate": 9.849763157894737e-05,
      "loss": 0.5339,
      "step": 25721
    },
    {
      "epoch": 1.440362862582596,
      "grad_norm": 1.5458778142929077,
      "learning_rate": 9.849736842105263e-05,
      "loss": 0.5653,
      "step": 25722
    },
    {
      "epoch": 1.440418859894725,
      "grad_norm": 7.980746746063232,
      "learning_rate": 9.849710526315789e-05,
      "loss": 0.4956,
      "step": 25723
    },
    {
      "epoch": 1.440474857206854,
      "grad_norm": 1.320092797279358,
      "learning_rate": 9.849684210526317e-05,
      "loss": 0.5172,
      "step": 25724
    },
    {
      "epoch": 1.440530854518983,
      "grad_norm": 1.689001441001892,
      "learning_rate": 9.849657894736843e-05,
      "loss": 0.5753,
      "step": 25725
    },
    {
      "epoch": 1.440586851831112,
      "grad_norm": 1.5204671621322632,
      "learning_rate": 9.849631578947369e-05,
      "loss": 0.5524,
      "step": 25726
    },
    {
      "epoch": 1.440642849143241,
      "grad_norm": 1.219130039215088,
      "learning_rate": 9.849605263157895e-05,
      "loss": 0.4574,
      "step": 25727
    },
    {
      "epoch": 1.44069884645537,
      "grad_norm": 1.1669539213180542,
      "learning_rate": 9.849578947368422e-05,
      "loss": 0.5025,
      "step": 25728
    },
    {
      "epoch": 1.440754843767499,
      "grad_norm": 1.4158577919006348,
      "learning_rate": 9.849552631578948e-05,
      "loss": 0.4848,
      "step": 25729
    },
    {
      "epoch": 1.440810841079628,
      "grad_norm": 1.3593709468841553,
      "learning_rate": 9.849526315789475e-05,
      "loss": 0.4785,
      "step": 25730
    },
    {
      "epoch": 1.440866838391757,
      "grad_norm": 1.3200407028198242,
      "learning_rate": 9.8495e-05,
      "loss": 0.3671,
      "step": 25731
    },
    {
      "epoch": 1.4409228357038861,
      "grad_norm": 2.0137245655059814,
      "learning_rate": 9.849473684210527e-05,
      "loss": 0.5586,
      "step": 25732
    },
    {
      "epoch": 1.4409788330160151,
      "grad_norm": 1.4319556951522827,
      "learning_rate": 9.849447368421053e-05,
      "loss": 0.4586,
      "step": 25733
    },
    {
      "epoch": 1.4410348303281442,
      "grad_norm": 1.4195220470428467,
      "learning_rate": 9.849421052631579e-05,
      "loss": 0.5305,
      "step": 25734
    },
    {
      "epoch": 1.4410908276402732,
      "grad_norm": 1.761801838874817,
      "learning_rate": 9.849394736842106e-05,
      "loss": 0.6981,
      "step": 25735
    },
    {
      "epoch": 1.4411468249524022,
      "grad_norm": 1.3044332265853882,
      "learning_rate": 9.849368421052631e-05,
      "loss": 0.4531,
      "step": 25736
    },
    {
      "epoch": 1.4412028222645312,
      "grad_norm": 1.449478030204773,
      "learning_rate": 9.849342105263158e-05,
      "loss": 0.5119,
      "step": 25737
    },
    {
      "epoch": 1.4412588195766602,
      "grad_norm": 1.167271375656128,
      "learning_rate": 9.849315789473684e-05,
      "loss": 0.4048,
      "step": 25738
    },
    {
      "epoch": 1.4413148168887893,
      "grad_norm": 1.1470009088516235,
      "learning_rate": 9.849289473684212e-05,
      "loss": 0.4402,
      "step": 25739
    },
    {
      "epoch": 1.4413708142009183,
      "grad_norm": 1.1541756391525269,
      "learning_rate": 9.849263157894738e-05,
      "loss": 0.4063,
      "step": 25740
    },
    {
      "epoch": 1.4414268115130473,
      "grad_norm": 1.43264901638031,
      "learning_rate": 9.849236842105264e-05,
      "loss": 0.4164,
      "step": 25741
    },
    {
      "epoch": 1.4414828088251763,
      "grad_norm": 1.517122745513916,
      "learning_rate": 9.84921052631579e-05,
      "loss": 0.6524,
      "step": 25742
    },
    {
      "epoch": 1.4415388061373053,
      "grad_norm": 1.531346082687378,
      "learning_rate": 9.849184210526317e-05,
      "loss": 0.4592,
      "step": 25743
    },
    {
      "epoch": 1.4415948034494344,
      "grad_norm": 1.2700045108795166,
      "learning_rate": 9.849157894736843e-05,
      "loss": 0.4308,
      "step": 25744
    },
    {
      "epoch": 1.4416508007615634,
      "grad_norm": 1.3897801637649536,
      "learning_rate": 9.849131578947369e-05,
      "loss": 0.5314,
      "step": 25745
    },
    {
      "epoch": 1.4417067980736924,
      "grad_norm": 1.5570229291915894,
      "learning_rate": 9.849105263157895e-05,
      "loss": 0.4738,
      "step": 25746
    },
    {
      "epoch": 1.4417627953858214,
      "grad_norm": 1.3779877424240112,
      "learning_rate": 9.849078947368422e-05,
      "loss": 0.6156,
      "step": 25747
    },
    {
      "epoch": 1.4418187926979504,
      "grad_norm": 1.4035085439682007,
      "learning_rate": 9.849052631578948e-05,
      "loss": 0.4309,
      "step": 25748
    },
    {
      "epoch": 1.4418747900100795,
      "grad_norm": 1.1910206079483032,
      "learning_rate": 9.849026315789474e-05,
      "loss": 0.4711,
      "step": 25749
    },
    {
      "epoch": 1.4419307873222085,
      "grad_norm": 1.4764854907989502,
      "learning_rate": 9.849e-05,
      "loss": 0.4317,
      "step": 25750
    },
    {
      "epoch": 1.4419867846343375,
      "grad_norm": 1.3121896982192993,
      "learning_rate": 9.848973684210526e-05,
      "loss": 0.4307,
      "step": 25751
    },
    {
      "epoch": 1.4420427819464665,
      "grad_norm": 1.616694450378418,
      "learning_rate": 9.848947368421053e-05,
      "loss": 0.7213,
      "step": 25752
    },
    {
      "epoch": 1.4420987792585955,
      "grad_norm": 1.1158812046051025,
      "learning_rate": 9.84892105263158e-05,
      "loss": 0.3822,
      "step": 25753
    },
    {
      "epoch": 1.4421547765707246,
      "grad_norm": 1.32943594455719,
      "learning_rate": 9.848894736842105e-05,
      "loss": 0.4025,
      "step": 25754
    },
    {
      "epoch": 1.4422107738828536,
      "grad_norm": 1.6812101602554321,
      "learning_rate": 9.848868421052631e-05,
      "loss": 0.557,
      "step": 25755
    },
    {
      "epoch": 1.4422667711949826,
      "grad_norm": 1.5972145795822144,
      "learning_rate": 9.848842105263159e-05,
      "loss": 0.4719,
      "step": 25756
    },
    {
      "epoch": 1.4423227685071116,
      "grad_norm": 1.3968019485473633,
      "learning_rate": 9.848815789473685e-05,
      "loss": 0.4173,
      "step": 25757
    },
    {
      "epoch": 1.4423787658192406,
      "grad_norm": 1.6362229585647583,
      "learning_rate": 9.848789473684212e-05,
      "loss": 0.5076,
      "step": 25758
    },
    {
      "epoch": 1.4424347631313696,
      "grad_norm": 1.2163110971450806,
      "learning_rate": 9.848763157894737e-05,
      "loss": 0.4992,
      "step": 25759
    },
    {
      "epoch": 1.4424907604434987,
      "grad_norm": 1.2565122842788696,
      "learning_rate": 9.848736842105264e-05,
      "loss": 0.4421,
      "step": 25760
    },
    {
      "epoch": 1.4425467577556277,
      "grad_norm": 1.3888131380081177,
      "learning_rate": 9.84871052631579e-05,
      "loss": 0.5243,
      "step": 25761
    },
    {
      "epoch": 1.4426027550677567,
      "grad_norm": 1.3142640590667725,
      "learning_rate": 9.848684210526317e-05,
      "loss": 0.3618,
      "step": 25762
    },
    {
      "epoch": 1.4426587523798857,
      "grad_norm": 1.2778666019439697,
      "learning_rate": 9.848657894736842e-05,
      "loss": 0.4543,
      "step": 25763
    },
    {
      "epoch": 1.4427147496920147,
      "grad_norm": 1.3430832624435425,
      "learning_rate": 9.848631578947369e-05,
      "loss": 0.5383,
      "step": 25764
    },
    {
      "epoch": 1.4427707470041438,
      "grad_norm": 1.625623106956482,
      "learning_rate": 9.848605263157895e-05,
      "loss": 0.5379,
      "step": 25765
    },
    {
      "epoch": 1.4428267443162728,
      "grad_norm": 1.27606999874115,
      "learning_rate": 9.848578947368422e-05,
      "loss": 0.3628,
      "step": 25766
    },
    {
      "epoch": 1.4428827416284018,
      "grad_norm": 1.6553044319152832,
      "learning_rate": 9.848552631578948e-05,
      "loss": 0.7632,
      "step": 25767
    },
    {
      "epoch": 1.4429387389405308,
      "grad_norm": 1.6357741355895996,
      "learning_rate": 9.848526315789474e-05,
      "loss": 0.5964,
      "step": 25768
    },
    {
      "epoch": 1.4429947362526598,
      "grad_norm": 1.7604966163635254,
      "learning_rate": 9.8485e-05,
      "loss": 0.7096,
      "step": 25769
    },
    {
      "epoch": 1.4430507335647889,
      "grad_norm": 1.5702502727508545,
      "learning_rate": 9.848473684210526e-05,
      "loss": 0.4116,
      "step": 25770
    },
    {
      "epoch": 1.4431067308769179,
      "grad_norm": 1.343562364578247,
      "learning_rate": 9.848447368421054e-05,
      "loss": 0.427,
      "step": 25771
    },
    {
      "epoch": 1.443162728189047,
      "grad_norm": 1.2983551025390625,
      "learning_rate": 9.84842105263158e-05,
      "loss": 0.38,
      "step": 25772
    },
    {
      "epoch": 1.443218725501176,
      "grad_norm": 1.2825380563735962,
      "learning_rate": 9.848394736842106e-05,
      "loss": 0.4135,
      "step": 25773
    },
    {
      "epoch": 1.443274722813305,
      "grad_norm": 1.3777891397476196,
      "learning_rate": 9.848368421052632e-05,
      "loss": 0.3667,
      "step": 25774
    },
    {
      "epoch": 1.443330720125434,
      "grad_norm": 1.2664874792099,
      "learning_rate": 9.848342105263159e-05,
      "loss": 0.4057,
      "step": 25775
    },
    {
      "epoch": 1.443386717437563,
      "grad_norm": 1.5738770961761475,
      "learning_rate": 9.848315789473685e-05,
      "loss": 0.511,
      "step": 25776
    },
    {
      "epoch": 1.443442714749692,
      "grad_norm": 1.3449558019638062,
      "learning_rate": 9.848289473684211e-05,
      "loss": 0.5309,
      "step": 25777
    },
    {
      "epoch": 1.443498712061821,
      "grad_norm": 1.1180827617645264,
      "learning_rate": 9.848263157894737e-05,
      "loss": 0.4378,
      "step": 25778
    },
    {
      "epoch": 1.44355470937395,
      "grad_norm": 1.8353078365325928,
      "learning_rate": 9.848236842105264e-05,
      "loss": 0.3912,
      "step": 25779
    },
    {
      "epoch": 1.443610706686079,
      "grad_norm": 2.383636951446533,
      "learning_rate": 9.84821052631579e-05,
      "loss": 0.6155,
      "step": 25780
    },
    {
      "epoch": 1.443666703998208,
      "grad_norm": 1.5396950244903564,
      "learning_rate": 9.848184210526316e-05,
      "loss": 0.4052,
      "step": 25781
    },
    {
      "epoch": 1.443722701310337,
      "grad_norm": 34.623844146728516,
      "learning_rate": 9.848157894736842e-05,
      "loss": 0.4543,
      "step": 25782
    },
    {
      "epoch": 1.443778698622466,
      "grad_norm": 1.4119480848312378,
      "learning_rate": 9.84813157894737e-05,
      "loss": 0.7073,
      "step": 25783
    },
    {
      "epoch": 1.4438346959345951,
      "grad_norm": 1.2975821495056152,
      "learning_rate": 9.848105263157895e-05,
      "loss": 0.4418,
      "step": 25784
    },
    {
      "epoch": 1.4438906932467241,
      "grad_norm": 1.4934360980987549,
      "learning_rate": 9.848078947368423e-05,
      "loss": 0.4745,
      "step": 25785
    },
    {
      "epoch": 1.4439466905588532,
      "grad_norm": 1.6379145383834839,
      "learning_rate": 9.848052631578947e-05,
      "loss": 0.4778,
      "step": 25786
    },
    {
      "epoch": 1.4440026878709822,
      "grad_norm": 1.3537036180496216,
      "learning_rate": 9.848026315789473e-05,
      "loss": 0.4766,
      "step": 25787
    },
    {
      "epoch": 1.4440586851831112,
      "grad_norm": 1.351130723953247,
      "learning_rate": 9.848e-05,
      "loss": 0.4779,
      "step": 25788
    },
    {
      "epoch": 1.4441146824952402,
      "grad_norm": 1.3500784635543823,
      "learning_rate": 9.847973684210527e-05,
      "loss": 0.4242,
      "step": 25789
    },
    {
      "epoch": 1.4441706798073692,
      "grad_norm": 1.1697924137115479,
      "learning_rate": 9.847947368421054e-05,
      "loss": 0.4517,
      "step": 25790
    },
    {
      "epoch": 1.4442266771194983,
      "grad_norm": 1.5843197107315063,
      "learning_rate": 9.847921052631578e-05,
      "loss": 0.3941,
      "step": 25791
    },
    {
      "epoch": 1.4442826744316273,
      "grad_norm": 1.3092375993728638,
      "learning_rate": 9.847894736842106e-05,
      "loss": 0.5552,
      "step": 25792
    },
    {
      "epoch": 1.4443386717437563,
      "grad_norm": 1.3153843879699707,
      "learning_rate": 9.847868421052632e-05,
      "loss": 0.3795,
      "step": 25793
    },
    {
      "epoch": 1.4443946690558853,
      "grad_norm": 1.3666855096817017,
      "learning_rate": 9.847842105263159e-05,
      "loss": 0.5007,
      "step": 25794
    },
    {
      "epoch": 1.4444506663680143,
      "grad_norm": 1.3877649307250977,
      "learning_rate": 9.847815789473685e-05,
      "loss": 0.5185,
      "step": 25795
    },
    {
      "epoch": 1.4445066636801434,
      "grad_norm": 3.332524299621582,
      "learning_rate": 9.847789473684211e-05,
      "loss": 0.4432,
      "step": 25796
    },
    {
      "epoch": 1.4445626609922724,
      "grad_norm": 1.4763113260269165,
      "learning_rate": 9.847763157894737e-05,
      "loss": 0.6981,
      "step": 25797
    },
    {
      "epoch": 1.4446186583044014,
      "grad_norm": 1.2583116292953491,
      "learning_rate": 9.847736842105264e-05,
      "loss": 0.4298,
      "step": 25798
    },
    {
      "epoch": 1.4446746556165304,
      "grad_norm": 1.458219289779663,
      "learning_rate": 9.84771052631579e-05,
      "loss": 0.5209,
      "step": 25799
    },
    {
      "epoch": 1.4447306529286594,
      "grad_norm": 1.372975468635559,
      "learning_rate": 9.847684210526316e-05,
      "loss": 0.5491,
      "step": 25800
    },
    {
      "epoch": 1.4447866502407885,
      "grad_norm": 1.1941468715667725,
      "learning_rate": 9.847657894736842e-05,
      "loss": 0.3895,
      "step": 25801
    },
    {
      "epoch": 1.4448426475529175,
      "grad_norm": 1.404276728630066,
      "learning_rate": 9.84763157894737e-05,
      "loss": 0.4141,
      "step": 25802
    },
    {
      "epoch": 1.4448986448650465,
      "grad_norm": 1.2524129152297974,
      "learning_rate": 9.847605263157896e-05,
      "loss": 0.5221,
      "step": 25803
    },
    {
      "epoch": 1.4449546421771755,
      "grad_norm": 1.4511653184890747,
      "learning_rate": 9.847578947368422e-05,
      "loss": 0.5579,
      "step": 25804
    },
    {
      "epoch": 1.4450106394893045,
      "grad_norm": 1.099122405052185,
      "learning_rate": 9.847552631578948e-05,
      "loss": 0.3619,
      "step": 25805
    },
    {
      "epoch": 1.4450666368014335,
      "grad_norm": 1.1699548959732056,
      "learning_rate": 9.847526315789474e-05,
      "loss": 0.3935,
      "step": 25806
    },
    {
      "epoch": 1.4451226341135626,
      "grad_norm": 1.9812945127487183,
      "learning_rate": 9.847500000000001e-05,
      "loss": 0.4948,
      "step": 25807
    },
    {
      "epoch": 1.4451786314256916,
      "grad_norm": 1.3612101078033447,
      "learning_rate": 9.847473684210527e-05,
      "loss": 0.5559,
      "step": 25808
    },
    {
      "epoch": 1.4452346287378206,
      "grad_norm": 1.381105899810791,
      "learning_rate": 9.847447368421053e-05,
      "loss": 0.323,
      "step": 25809
    },
    {
      "epoch": 1.4452906260499496,
      "grad_norm": 1.4454028606414795,
      "learning_rate": 9.847421052631579e-05,
      "loss": 0.5805,
      "step": 25810
    },
    {
      "epoch": 1.4453466233620786,
      "grad_norm": 2.1735076904296875,
      "learning_rate": 9.847394736842106e-05,
      "loss": 0.5439,
      "step": 25811
    },
    {
      "epoch": 1.4454026206742077,
      "grad_norm": 2.0774319171905518,
      "learning_rate": 9.847368421052632e-05,
      "loss": 0.559,
      "step": 25812
    },
    {
      "epoch": 1.4454586179863367,
      "grad_norm": 1.3191627264022827,
      "learning_rate": 9.847342105263158e-05,
      "loss": 0.4167,
      "step": 25813
    },
    {
      "epoch": 1.4455146152984657,
      "grad_norm": 1.2631181478500366,
      "learning_rate": 9.847315789473684e-05,
      "loss": 0.4546,
      "step": 25814
    },
    {
      "epoch": 1.4455706126105947,
      "grad_norm": 1.380804181098938,
      "learning_rate": 9.847289473684211e-05,
      "loss": 0.3802,
      "step": 25815
    },
    {
      "epoch": 1.4456266099227237,
      "grad_norm": 1.2480416297912598,
      "learning_rate": 9.847263157894737e-05,
      "loss": 0.5143,
      "step": 25816
    },
    {
      "epoch": 1.4456826072348528,
      "grad_norm": 1.4806478023529053,
      "learning_rate": 9.847236842105265e-05,
      "loss": 0.5376,
      "step": 25817
    },
    {
      "epoch": 1.4457386045469818,
      "grad_norm": 1.348549246788025,
      "learning_rate": 9.847210526315789e-05,
      "loss": 0.3942,
      "step": 25818
    },
    {
      "epoch": 1.4457946018591108,
      "grad_norm": 1.1411925554275513,
      "learning_rate": 9.847184210526317e-05,
      "loss": 0.3391,
      "step": 25819
    },
    {
      "epoch": 1.4458505991712398,
      "grad_norm": 1.7207562923431396,
      "learning_rate": 9.847157894736843e-05,
      "loss": 0.5623,
      "step": 25820
    },
    {
      "epoch": 1.4459065964833688,
      "grad_norm": 1.4202302694320679,
      "learning_rate": 9.847131578947369e-05,
      "loss": 0.5446,
      "step": 25821
    },
    {
      "epoch": 1.4459625937954979,
      "grad_norm": 1.5925122499465942,
      "learning_rate": 9.847105263157896e-05,
      "loss": 0.4418,
      "step": 25822
    },
    {
      "epoch": 1.4460185911076269,
      "grad_norm": 1.2614142894744873,
      "learning_rate": 9.84707894736842e-05,
      "loss": 0.4257,
      "step": 25823
    },
    {
      "epoch": 1.446074588419756,
      "grad_norm": 1.0207983255386353,
      "learning_rate": 9.847052631578948e-05,
      "loss": 0.3638,
      "step": 25824
    },
    {
      "epoch": 1.446130585731885,
      "grad_norm": 1.6025762557983398,
      "learning_rate": 9.847026315789474e-05,
      "loss": 0.5265,
      "step": 25825
    },
    {
      "epoch": 1.446186583044014,
      "grad_norm": 1.2102876901626587,
      "learning_rate": 9.847000000000001e-05,
      "loss": 0.4531,
      "step": 25826
    },
    {
      "epoch": 1.446242580356143,
      "grad_norm": 1.18339204788208,
      "learning_rate": 9.846973684210527e-05,
      "loss": 0.3625,
      "step": 25827
    },
    {
      "epoch": 1.446298577668272,
      "grad_norm": 1.980252981185913,
      "learning_rate": 9.846947368421053e-05,
      "loss": 0.5182,
      "step": 25828
    },
    {
      "epoch": 1.446354574980401,
      "grad_norm": 1.51128351688385,
      "learning_rate": 9.846921052631579e-05,
      "loss": 0.4307,
      "step": 25829
    },
    {
      "epoch": 1.44641057229253,
      "grad_norm": 1.2699722051620483,
      "learning_rate": 9.846894736842106e-05,
      "loss": 0.4264,
      "step": 25830
    },
    {
      "epoch": 1.446466569604659,
      "grad_norm": 1.3692348003387451,
      "learning_rate": 9.846868421052632e-05,
      "loss": 0.4324,
      "step": 25831
    },
    {
      "epoch": 1.446522566916788,
      "grad_norm": 1.2989691495895386,
      "learning_rate": 9.846842105263158e-05,
      "loss": 0.5484,
      "step": 25832
    },
    {
      "epoch": 1.446578564228917,
      "grad_norm": 1.4866071939468384,
      "learning_rate": 9.846815789473684e-05,
      "loss": 0.5034,
      "step": 25833
    },
    {
      "epoch": 1.446634561541046,
      "grad_norm": 1.1898126602172852,
      "learning_rate": 9.846789473684212e-05,
      "loss": 0.4293,
      "step": 25834
    },
    {
      "epoch": 1.446690558853175,
      "grad_norm": 1.5289809703826904,
      "learning_rate": 9.846763157894738e-05,
      "loss": 0.6081,
      "step": 25835
    },
    {
      "epoch": 1.4467465561653041,
      "grad_norm": 1.984678030014038,
      "learning_rate": 9.846736842105264e-05,
      "loss": 0.54,
      "step": 25836
    },
    {
      "epoch": 1.4468025534774331,
      "grad_norm": 1.222996711730957,
      "learning_rate": 9.84671052631579e-05,
      "loss": 0.3786,
      "step": 25837
    },
    {
      "epoch": 1.4468585507895622,
      "grad_norm": 1.2271722555160522,
      "learning_rate": 9.846684210526315e-05,
      "loss": 0.4339,
      "step": 25838
    },
    {
      "epoch": 1.4469145481016912,
      "grad_norm": 1.6566424369812012,
      "learning_rate": 9.846657894736843e-05,
      "loss": 0.3967,
      "step": 25839
    },
    {
      "epoch": 1.4469705454138202,
      "grad_norm": 1.3022581338882446,
      "learning_rate": 9.846631578947369e-05,
      "loss": 0.5282,
      "step": 25840
    },
    {
      "epoch": 1.4470265427259492,
      "grad_norm": 1.193652868270874,
      "learning_rate": 9.846605263157895e-05,
      "loss": 0.3789,
      "step": 25841
    },
    {
      "epoch": 1.4470825400380782,
      "grad_norm": 1.084812045097351,
      "learning_rate": 9.846578947368421e-05,
      "loss": 0.4475,
      "step": 25842
    },
    {
      "epoch": 1.4471385373502073,
      "grad_norm": 1.7399256229400635,
      "learning_rate": 9.846552631578948e-05,
      "loss": 0.4837,
      "step": 25843
    },
    {
      "epoch": 1.4471945346623363,
      "grad_norm": 4.695403099060059,
      "learning_rate": 9.846526315789474e-05,
      "loss": 0.4705,
      "step": 25844
    },
    {
      "epoch": 1.4472505319744653,
      "grad_norm": 7.097339630126953,
      "learning_rate": 9.846500000000001e-05,
      "loss": 0.4211,
      "step": 25845
    },
    {
      "epoch": 1.4473065292865943,
      "grad_norm": 1.3863807916641235,
      "learning_rate": 9.846473684210526e-05,
      "loss": 0.5693,
      "step": 25846
    },
    {
      "epoch": 1.4473625265987233,
      "grad_norm": 1.4710978269577026,
      "learning_rate": 9.846447368421053e-05,
      "loss": 0.4579,
      "step": 25847
    },
    {
      "epoch": 1.4474185239108524,
      "grad_norm": 1.4749094247817993,
      "learning_rate": 9.846421052631579e-05,
      "loss": 0.5118,
      "step": 25848
    },
    {
      "epoch": 1.4474745212229814,
      "grad_norm": 1.664351463317871,
      "learning_rate": 9.846394736842107e-05,
      "loss": 0.4544,
      "step": 25849
    },
    {
      "epoch": 1.4475305185351104,
      "grad_norm": 1.4030771255493164,
      "learning_rate": 9.846368421052633e-05,
      "loss": 0.4978,
      "step": 25850
    },
    {
      "epoch": 1.4475865158472394,
      "grad_norm": 1.3160074949264526,
      "learning_rate": 9.846342105263159e-05,
      "loss": 0.5448,
      "step": 25851
    },
    {
      "epoch": 1.4476425131593684,
      "grad_norm": 1.2401901483535767,
      "learning_rate": 9.846315789473685e-05,
      "loss": 0.4574,
      "step": 25852
    },
    {
      "epoch": 1.4476985104714974,
      "grad_norm": 1.3144906759262085,
      "learning_rate": 9.846289473684212e-05,
      "loss": 0.4755,
      "step": 25853
    },
    {
      "epoch": 1.4477545077836265,
      "grad_norm": 1.186714768409729,
      "learning_rate": 9.846263157894738e-05,
      "loss": 0.4043,
      "step": 25854
    },
    {
      "epoch": 1.4478105050957555,
      "grad_norm": 1.2592949867248535,
      "learning_rate": 9.846236842105262e-05,
      "loss": 0.377,
      "step": 25855
    },
    {
      "epoch": 1.4478665024078845,
      "grad_norm": 1.4551217555999756,
      "learning_rate": 9.84621052631579e-05,
      "loss": 0.5906,
      "step": 25856
    },
    {
      "epoch": 1.4479224997200135,
      "grad_norm": 1.4207539558410645,
      "learning_rate": 9.846184210526316e-05,
      "loss": 0.5401,
      "step": 25857
    },
    {
      "epoch": 1.4479784970321425,
      "grad_norm": 1.1180894374847412,
      "learning_rate": 9.846157894736843e-05,
      "loss": 0.3977,
      "step": 25858
    },
    {
      "epoch": 1.4480344943442716,
      "grad_norm": 1.841389775276184,
      "learning_rate": 9.846131578947369e-05,
      "loss": 0.5804,
      "step": 25859
    },
    {
      "epoch": 1.4480904916564006,
      "grad_norm": 1.4709042310714722,
      "learning_rate": 9.846105263157895e-05,
      "loss": 0.479,
      "step": 25860
    },
    {
      "epoch": 1.4481464889685296,
      "grad_norm": 1.5189865827560425,
      "learning_rate": 9.846078947368421e-05,
      "loss": 0.4432,
      "step": 25861
    },
    {
      "epoch": 1.4482024862806586,
      "grad_norm": 1.2005620002746582,
      "learning_rate": 9.846052631578948e-05,
      "loss": 0.3775,
      "step": 25862
    },
    {
      "epoch": 1.4482584835927876,
      "grad_norm": 1.397200584411621,
      "learning_rate": 9.846026315789474e-05,
      "loss": 0.6105,
      "step": 25863
    },
    {
      "epoch": 1.4483144809049167,
      "grad_norm": 1.3551275730133057,
      "learning_rate": 9.846e-05,
      "loss": 0.4215,
      "step": 25864
    },
    {
      "epoch": 1.4483704782170457,
      "grad_norm": 1.2143815755844116,
      "learning_rate": 9.845973684210526e-05,
      "loss": 0.4264,
      "step": 25865
    },
    {
      "epoch": 1.4484264755291747,
      "grad_norm": 1.278725266456604,
      "learning_rate": 9.845947368421054e-05,
      "loss": 0.6367,
      "step": 25866
    },
    {
      "epoch": 1.4484824728413037,
      "grad_norm": 1.7548737525939941,
      "learning_rate": 9.84592105263158e-05,
      "loss": 0.5817,
      "step": 25867
    },
    {
      "epoch": 1.4485384701534327,
      "grad_norm": 1.250089168548584,
      "learning_rate": 9.845894736842106e-05,
      "loss": 0.4424,
      "step": 25868
    },
    {
      "epoch": 1.4485944674655618,
      "grad_norm": 1.41702401638031,
      "learning_rate": 9.845868421052631e-05,
      "loss": 0.5236,
      "step": 25869
    },
    {
      "epoch": 1.4486504647776908,
      "grad_norm": 1.1638234853744507,
      "learning_rate": 9.845842105263159e-05,
      "loss": 0.4341,
      "step": 25870
    },
    {
      "epoch": 1.4487064620898198,
      "grad_norm": 1.0353578329086304,
      "learning_rate": 9.845815789473685e-05,
      "loss": 0.3386,
      "step": 25871
    },
    {
      "epoch": 1.4487624594019488,
      "grad_norm": 1.5625780820846558,
      "learning_rate": 9.845789473684211e-05,
      "loss": 0.395,
      "step": 25872
    },
    {
      "epoch": 1.4488184567140778,
      "grad_norm": 7.201679229736328,
      "learning_rate": 9.845763157894737e-05,
      "loss": 0.5571,
      "step": 25873
    },
    {
      "epoch": 1.4488744540262068,
      "grad_norm": 1.3730390071868896,
      "learning_rate": 9.845736842105263e-05,
      "loss": 0.4704,
      "step": 25874
    },
    {
      "epoch": 1.4489304513383359,
      "grad_norm": 1.3140690326690674,
      "learning_rate": 9.84571052631579e-05,
      "loss": 0.4497,
      "step": 25875
    },
    {
      "epoch": 1.4489864486504649,
      "grad_norm": 1.3097856044769287,
      "learning_rate": 9.845684210526316e-05,
      "loss": 0.452,
      "step": 25876
    },
    {
      "epoch": 1.449042445962594,
      "grad_norm": 1.4853609800338745,
      "learning_rate": 9.845657894736843e-05,
      "loss": 0.5977,
      "step": 25877
    },
    {
      "epoch": 1.449098443274723,
      "grad_norm": 1.4176387786865234,
      "learning_rate": 9.845631578947368e-05,
      "loss": 0.4417,
      "step": 25878
    },
    {
      "epoch": 1.449154440586852,
      "grad_norm": 1.1323189735412598,
      "learning_rate": 9.845605263157895e-05,
      "loss": 0.5364,
      "step": 25879
    },
    {
      "epoch": 1.449210437898981,
      "grad_norm": 1.172784686088562,
      "learning_rate": 9.845578947368421e-05,
      "loss": 0.4925,
      "step": 25880
    },
    {
      "epoch": 1.44926643521111,
      "grad_norm": 1.1581997871398926,
      "learning_rate": 9.845552631578949e-05,
      "loss": 0.3504,
      "step": 25881
    },
    {
      "epoch": 1.449322432523239,
      "grad_norm": 1.2002604007720947,
      "learning_rate": 9.845526315789475e-05,
      "loss": 0.4836,
      "step": 25882
    },
    {
      "epoch": 1.449378429835368,
      "grad_norm": 1.1245604753494263,
      "learning_rate": 9.8455e-05,
      "loss": 0.3703,
      "step": 25883
    },
    {
      "epoch": 1.449434427147497,
      "grad_norm": 1.520147442817688,
      "learning_rate": 9.845473684210526e-05,
      "loss": 0.5046,
      "step": 25884
    },
    {
      "epoch": 1.449490424459626,
      "grad_norm": 1.5617785453796387,
      "learning_rate": 9.845447368421054e-05,
      "loss": 0.4901,
      "step": 25885
    },
    {
      "epoch": 1.449546421771755,
      "grad_norm": 1.4862608909606934,
      "learning_rate": 9.84542105263158e-05,
      "loss": 0.4195,
      "step": 25886
    },
    {
      "epoch": 1.449602419083884,
      "grad_norm": 1.2302963733673096,
      "learning_rate": 9.845394736842106e-05,
      "loss": 0.4553,
      "step": 25887
    },
    {
      "epoch": 1.4496584163960131,
      "grad_norm": 1.2800395488739014,
      "learning_rate": 9.845368421052632e-05,
      "loss": 0.3787,
      "step": 25888
    },
    {
      "epoch": 1.4497144137081421,
      "grad_norm": 1.2763700485229492,
      "learning_rate": 9.845342105263159e-05,
      "loss": 0.3521,
      "step": 25889
    },
    {
      "epoch": 1.4497704110202712,
      "grad_norm": 1.4803738594055176,
      "learning_rate": 9.845315789473685e-05,
      "loss": 0.5863,
      "step": 25890
    },
    {
      "epoch": 1.4498264083324002,
      "grad_norm": 1.222286343574524,
      "learning_rate": 9.845289473684211e-05,
      "loss": 0.4811,
      "step": 25891
    },
    {
      "epoch": 1.4498824056445292,
      "grad_norm": 1.0914252996444702,
      "learning_rate": 9.845263157894737e-05,
      "loss": 0.3386,
      "step": 25892
    },
    {
      "epoch": 1.4499384029566582,
      "grad_norm": 1.4120428562164307,
      "learning_rate": 9.845236842105263e-05,
      "loss": 0.5278,
      "step": 25893
    },
    {
      "epoch": 1.4499944002687872,
      "grad_norm": 1.41368567943573,
      "learning_rate": 9.84521052631579e-05,
      "loss": 0.4883,
      "step": 25894
    },
    {
      "epoch": 1.4500503975809163,
      "grad_norm": 1.2134027481079102,
      "learning_rate": 9.845184210526316e-05,
      "loss": 0.4061,
      "step": 25895
    },
    {
      "epoch": 1.4501063948930453,
      "grad_norm": 1.4855021238327026,
      "learning_rate": 9.845157894736842e-05,
      "loss": 0.4442,
      "step": 25896
    },
    {
      "epoch": 1.4501623922051743,
      "grad_norm": 1.4125449657440186,
      "learning_rate": 9.845131578947368e-05,
      "loss": 0.4051,
      "step": 25897
    },
    {
      "epoch": 1.4502183895173033,
      "grad_norm": 1.275683879852295,
      "learning_rate": 9.845105263157896e-05,
      "loss": 0.4895,
      "step": 25898
    },
    {
      "epoch": 1.450274386829432,
      "grad_norm": 1.4906284809112549,
      "learning_rate": 9.845078947368422e-05,
      "loss": 0.4012,
      "step": 25899
    },
    {
      "epoch": 1.4503303841415611,
      "grad_norm": 1.4855666160583496,
      "learning_rate": 9.845052631578949e-05,
      "loss": 0.5737,
      "step": 25900
    },
    {
      "epoch": 1.4503863814536901,
      "grad_norm": 1.6058350801467896,
      "learning_rate": 9.845026315789473e-05,
      "loss": 0.5658,
      "step": 25901
    },
    {
      "epoch": 1.4504423787658192,
      "grad_norm": 1.2203283309936523,
      "learning_rate": 9.845000000000001e-05,
      "loss": 0.3587,
      "step": 25902
    },
    {
      "epoch": 1.4504983760779482,
      "grad_norm": 1.2490440607070923,
      "learning_rate": 9.844973684210527e-05,
      "loss": 0.3595,
      "step": 25903
    },
    {
      "epoch": 1.4505543733900772,
      "grad_norm": 1.5844202041625977,
      "learning_rate": 9.844947368421054e-05,
      "loss": 0.4712,
      "step": 25904
    },
    {
      "epoch": 1.4506103707022062,
      "grad_norm": 1.4154523611068726,
      "learning_rate": 9.84492105263158e-05,
      "loss": 0.4897,
      "step": 25905
    },
    {
      "epoch": 1.4506663680143352,
      "grad_norm": 1.3495726585388184,
      "learning_rate": 9.844894736842106e-05,
      "loss": 0.585,
      "step": 25906
    },
    {
      "epoch": 1.4507223653264643,
      "grad_norm": 1.4050761461257935,
      "learning_rate": 9.844868421052632e-05,
      "loss": 0.5269,
      "step": 25907
    },
    {
      "epoch": 1.4507783626385933,
      "grad_norm": 1.2094513177871704,
      "learning_rate": 9.844842105263158e-05,
      "loss": 0.4117,
      "step": 25908
    },
    {
      "epoch": 1.4508343599507223,
      "grad_norm": 1.6439192295074463,
      "learning_rate": 9.844815789473685e-05,
      "loss": 0.546,
      "step": 25909
    },
    {
      "epoch": 1.4508903572628513,
      "grad_norm": 1.701016902923584,
      "learning_rate": 9.84478947368421e-05,
      "loss": 0.4767,
      "step": 25910
    },
    {
      "epoch": 1.4509463545749803,
      "grad_norm": 1.4660980701446533,
      "learning_rate": 9.844763157894737e-05,
      "loss": 0.399,
      "step": 25911
    },
    {
      "epoch": 1.4510023518871094,
      "grad_norm": 1.4048539400100708,
      "learning_rate": 9.844736842105263e-05,
      "loss": 0.5477,
      "step": 25912
    },
    {
      "epoch": 1.4510583491992384,
      "grad_norm": 1.4888503551483154,
      "learning_rate": 9.84471052631579e-05,
      "loss": 0.6689,
      "step": 25913
    },
    {
      "epoch": 1.4511143465113674,
      "grad_norm": 1.1039063930511475,
      "learning_rate": 9.844684210526317e-05,
      "loss": 0.4155,
      "step": 25914
    },
    {
      "epoch": 1.4511703438234964,
      "grad_norm": 1.7937867641448975,
      "learning_rate": 9.844657894736842e-05,
      "loss": 0.4529,
      "step": 25915
    },
    {
      "epoch": 1.4512263411356254,
      "grad_norm": 1.480340600013733,
      "learning_rate": 9.844631578947368e-05,
      "loss": 0.4487,
      "step": 25916
    },
    {
      "epoch": 1.4512823384477544,
      "grad_norm": 1.8020457029342651,
      "learning_rate": 9.844605263157896e-05,
      "loss": 0.5597,
      "step": 25917
    },
    {
      "epoch": 1.4513383357598835,
      "grad_norm": 3.0687625408172607,
      "learning_rate": 9.844578947368422e-05,
      "loss": 0.5293,
      "step": 25918
    },
    {
      "epoch": 1.4513943330720125,
      "grad_norm": 1.4975465536117554,
      "learning_rate": 9.844552631578948e-05,
      "loss": 0.4141,
      "step": 25919
    },
    {
      "epoch": 1.4514503303841415,
      "grad_norm": 1.5277751684188843,
      "learning_rate": 9.844526315789474e-05,
      "loss": 0.6326,
      "step": 25920
    },
    {
      "epoch": 1.4515063276962705,
      "grad_norm": 1.370906949043274,
      "learning_rate": 9.844500000000001e-05,
      "loss": 0.468,
      "step": 25921
    },
    {
      "epoch": 1.4515623250083995,
      "grad_norm": 1.2885569334030151,
      "learning_rate": 9.844473684210527e-05,
      "loss": 0.5034,
      "step": 25922
    },
    {
      "epoch": 1.4516183223205286,
      "grad_norm": 1.2657023668289185,
      "learning_rate": 9.844447368421053e-05,
      "loss": 0.4651,
      "step": 25923
    },
    {
      "epoch": 1.4516743196326576,
      "grad_norm": 1.3978196382522583,
      "learning_rate": 9.844421052631579e-05,
      "loss": 0.4269,
      "step": 25924
    },
    {
      "epoch": 1.4517303169447866,
      "grad_norm": 1.3470312356948853,
      "learning_rate": 9.844394736842105e-05,
      "loss": 0.4954,
      "step": 25925
    },
    {
      "epoch": 1.4517863142569156,
      "grad_norm": 1.8148443698883057,
      "learning_rate": 9.844368421052632e-05,
      "loss": 0.6209,
      "step": 25926
    },
    {
      "epoch": 1.4518423115690446,
      "grad_norm": 5.657511234283447,
      "learning_rate": 9.844342105263158e-05,
      "loss": 0.4336,
      "step": 25927
    },
    {
      "epoch": 1.4518983088811737,
      "grad_norm": 1.5702624320983887,
      "learning_rate": 9.844315789473684e-05,
      "loss": 0.5375,
      "step": 25928
    },
    {
      "epoch": 1.4519543061933027,
      "grad_norm": 1.4084192514419556,
      "learning_rate": 9.84428947368421e-05,
      "loss": 0.5047,
      "step": 25929
    },
    {
      "epoch": 1.4520103035054317,
      "grad_norm": 1.1949462890625,
      "learning_rate": 9.844263157894738e-05,
      "loss": 0.3619,
      "step": 25930
    },
    {
      "epoch": 1.4520663008175607,
      "grad_norm": 1.2923543453216553,
      "learning_rate": 9.844236842105263e-05,
      "loss": 0.4469,
      "step": 25931
    },
    {
      "epoch": 1.4521222981296897,
      "grad_norm": 1.2760673761367798,
      "learning_rate": 9.844210526315791e-05,
      "loss": 0.467,
      "step": 25932
    },
    {
      "epoch": 1.4521782954418188,
      "grad_norm": 1.2177743911743164,
      "learning_rate": 9.844184210526315e-05,
      "loss": 0.406,
      "step": 25933
    },
    {
      "epoch": 1.4522342927539478,
      "grad_norm": 1.1399402618408203,
      "learning_rate": 9.844157894736843e-05,
      "loss": 0.3858,
      "step": 25934
    },
    {
      "epoch": 1.4522902900660768,
      "grad_norm": 1.6178439855575562,
      "learning_rate": 9.844131578947369e-05,
      "loss": 0.5365,
      "step": 25935
    },
    {
      "epoch": 1.4523462873782058,
      "grad_norm": 1.4049549102783203,
      "learning_rate": 9.844105263157896e-05,
      "loss": 0.5515,
      "step": 25936
    },
    {
      "epoch": 1.4524022846903348,
      "grad_norm": 1.2331537008285522,
      "learning_rate": 9.844078947368422e-05,
      "loss": 0.4917,
      "step": 25937
    },
    {
      "epoch": 1.4524582820024639,
      "grad_norm": 1.2904316186904907,
      "learning_rate": 9.844052631578948e-05,
      "loss": 0.4256,
      "step": 25938
    },
    {
      "epoch": 1.4525142793145929,
      "grad_norm": 1.261634349822998,
      "learning_rate": 9.844026315789474e-05,
      "loss": 0.458,
      "step": 25939
    },
    {
      "epoch": 1.452570276626722,
      "grad_norm": 1.430680513381958,
      "learning_rate": 9.844000000000001e-05,
      "loss": 0.4403,
      "step": 25940
    },
    {
      "epoch": 1.452626273938851,
      "grad_norm": 1.102910041809082,
      "learning_rate": 9.843973684210527e-05,
      "loss": 0.3499,
      "step": 25941
    },
    {
      "epoch": 1.45268227125098,
      "grad_norm": 1.4364904165267944,
      "learning_rate": 9.843947368421053e-05,
      "loss": 0.5855,
      "step": 25942
    },
    {
      "epoch": 1.452738268563109,
      "grad_norm": 1.155799388885498,
      "learning_rate": 9.843921052631579e-05,
      "loss": 0.3681,
      "step": 25943
    },
    {
      "epoch": 1.452794265875238,
      "grad_norm": 1.272122859954834,
      "learning_rate": 9.843894736842105e-05,
      "loss": 0.4879,
      "step": 25944
    },
    {
      "epoch": 1.452850263187367,
      "grad_norm": 1.3616063594818115,
      "learning_rate": 9.843868421052633e-05,
      "loss": 0.5572,
      "step": 25945
    },
    {
      "epoch": 1.452906260499496,
      "grad_norm": 1.530600905418396,
      "learning_rate": 9.843842105263158e-05,
      "loss": 0.4332,
      "step": 25946
    },
    {
      "epoch": 1.452962257811625,
      "grad_norm": 1.2354836463928223,
      "learning_rate": 9.843815789473684e-05,
      "loss": 0.4672,
      "step": 25947
    },
    {
      "epoch": 1.453018255123754,
      "grad_norm": 1.435318112373352,
      "learning_rate": 9.84378947368421e-05,
      "loss": 0.4326,
      "step": 25948
    },
    {
      "epoch": 1.453074252435883,
      "grad_norm": 1.4048984050750732,
      "learning_rate": 9.843763157894738e-05,
      "loss": 0.3772,
      "step": 25949
    },
    {
      "epoch": 1.453130249748012,
      "grad_norm": 2.3893043994903564,
      "learning_rate": 9.843736842105264e-05,
      "loss": 0.5347,
      "step": 25950
    },
    {
      "epoch": 1.453186247060141,
      "grad_norm": 1.403817057609558,
      "learning_rate": 9.84371052631579e-05,
      "loss": 0.5213,
      "step": 25951
    },
    {
      "epoch": 1.4532422443722701,
      "grad_norm": 1.2591506242752075,
      "learning_rate": 9.843684210526316e-05,
      "loss": 0.4081,
      "step": 25952
    },
    {
      "epoch": 1.4532982416843991,
      "grad_norm": 1.2045007944107056,
      "learning_rate": 9.843657894736843e-05,
      "loss": 0.5792,
      "step": 25953
    },
    {
      "epoch": 1.4533542389965282,
      "grad_norm": 1.2742711305618286,
      "learning_rate": 9.843631578947369e-05,
      "loss": 0.4952,
      "step": 25954
    },
    {
      "epoch": 1.4534102363086572,
      "grad_norm": 1.6678144931793213,
      "learning_rate": 9.843605263157896e-05,
      "loss": 0.5661,
      "step": 25955
    },
    {
      "epoch": 1.4534662336207862,
      "grad_norm": 1.4130090475082397,
      "learning_rate": 9.843578947368421e-05,
      "loss": 0.6704,
      "step": 25956
    },
    {
      "epoch": 1.4535222309329152,
      "grad_norm": 1.37088942527771,
      "learning_rate": 9.843552631578948e-05,
      "loss": 0.4189,
      "step": 25957
    },
    {
      "epoch": 1.4535782282450442,
      "grad_norm": 1.8538419008255005,
      "learning_rate": 9.843526315789474e-05,
      "loss": 0.4482,
      "step": 25958
    },
    {
      "epoch": 1.4536342255571733,
      "grad_norm": 1.3719291687011719,
      "learning_rate": 9.8435e-05,
      "loss": 0.3062,
      "step": 25959
    },
    {
      "epoch": 1.4536902228693023,
      "grad_norm": 1.3360399007797241,
      "learning_rate": 9.843473684210526e-05,
      "loss": 0.3929,
      "step": 25960
    },
    {
      "epoch": 1.4537462201814313,
      "grad_norm": 1.3749924898147583,
      "learning_rate": 9.843447368421052e-05,
      "loss": 0.4151,
      "step": 25961
    },
    {
      "epoch": 1.4538022174935603,
      "grad_norm": 1.381787657737732,
      "learning_rate": 9.84342105263158e-05,
      "loss": 0.4326,
      "step": 25962
    },
    {
      "epoch": 1.4538582148056893,
      "grad_norm": 1.9576159715652466,
      "learning_rate": 9.843394736842105e-05,
      "loss": 0.4619,
      "step": 25963
    },
    {
      "epoch": 1.4539142121178183,
      "grad_norm": 1.386579155921936,
      "learning_rate": 9.843368421052633e-05,
      "loss": 0.4705,
      "step": 25964
    },
    {
      "epoch": 1.4539702094299474,
      "grad_norm": 1.1891536712646484,
      "learning_rate": 9.843342105263157e-05,
      "loss": 0.4118,
      "step": 25965
    },
    {
      "epoch": 1.4540262067420764,
      "grad_norm": 1.2311022281646729,
      "learning_rate": 9.843315789473685e-05,
      "loss": 0.4656,
      "step": 25966
    },
    {
      "epoch": 1.4540822040542054,
      "grad_norm": 1.3781304359436035,
      "learning_rate": 9.843289473684211e-05,
      "loss": 0.5193,
      "step": 25967
    },
    {
      "epoch": 1.4541382013663344,
      "grad_norm": 1.1454792022705078,
      "learning_rate": 9.843263157894738e-05,
      "loss": 0.4095,
      "step": 25968
    },
    {
      "epoch": 1.4541941986784634,
      "grad_norm": 1.28517484664917,
      "learning_rate": 9.843236842105264e-05,
      "loss": 0.3822,
      "step": 25969
    },
    {
      "epoch": 1.4542501959905925,
      "grad_norm": 1.3925752639770508,
      "learning_rate": 9.84321052631579e-05,
      "loss": 0.4105,
      "step": 25970
    },
    {
      "epoch": 1.4543061933027215,
      "grad_norm": 3.689382553100586,
      "learning_rate": 9.843184210526316e-05,
      "loss": 0.603,
      "step": 25971
    },
    {
      "epoch": 1.4543621906148505,
      "grad_norm": 1.3937817811965942,
      "learning_rate": 9.843157894736843e-05,
      "loss": 0.4476,
      "step": 25972
    },
    {
      "epoch": 1.4544181879269795,
      "grad_norm": 77.36202239990234,
      "learning_rate": 9.843131578947369e-05,
      "loss": 0.3675,
      "step": 25973
    },
    {
      "epoch": 1.4544741852391085,
      "grad_norm": 1.22329580783844,
      "learning_rate": 9.843105263157895e-05,
      "loss": 0.3984,
      "step": 25974
    },
    {
      "epoch": 1.4545301825512376,
      "grad_norm": 1.4108954668045044,
      "learning_rate": 9.843078947368421e-05,
      "loss": 0.4543,
      "step": 25975
    },
    {
      "epoch": 1.4545861798633666,
      "grad_norm": 1.2376995086669922,
      "learning_rate": 9.843052631578947e-05,
      "loss": 0.4174,
      "step": 25976
    },
    {
      "epoch": 1.4546421771754956,
      "grad_norm": 1.5809969902038574,
      "learning_rate": 9.843026315789474e-05,
      "loss": 0.7597,
      "step": 25977
    },
    {
      "epoch": 1.4546981744876246,
      "grad_norm": 1.929602861404419,
      "learning_rate": 9.843e-05,
      "loss": 0.6487,
      "step": 25978
    },
    {
      "epoch": 1.4547541717997536,
      "grad_norm": 1.5038347244262695,
      "learning_rate": 9.842973684210526e-05,
      "loss": 0.6088,
      "step": 25979
    },
    {
      "epoch": 1.4548101691118827,
      "grad_norm": 1.659993290901184,
      "learning_rate": 9.842947368421052e-05,
      "loss": 0.4552,
      "step": 25980
    },
    {
      "epoch": 1.4548661664240117,
      "grad_norm": 1.1503784656524658,
      "learning_rate": 9.84292105263158e-05,
      "loss": 0.4481,
      "step": 25981
    },
    {
      "epoch": 1.4549221637361407,
      "grad_norm": 1.1233444213867188,
      "learning_rate": 9.842894736842106e-05,
      "loss": 0.3566,
      "step": 25982
    },
    {
      "epoch": 1.4549781610482697,
      "grad_norm": 1.5418497323989868,
      "learning_rate": 9.842868421052632e-05,
      "loss": 0.4217,
      "step": 25983
    },
    {
      "epoch": 1.4550341583603987,
      "grad_norm": 1.42060387134552,
      "learning_rate": 9.842842105263158e-05,
      "loss": 0.5069,
      "step": 25984
    },
    {
      "epoch": 1.4550901556725278,
      "grad_norm": 1.5929495096206665,
      "learning_rate": 9.842815789473685e-05,
      "loss": 0.4035,
      "step": 25985
    },
    {
      "epoch": 1.4551461529846568,
      "grad_norm": 1.476660966873169,
      "learning_rate": 9.842789473684211e-05,
      "loss": 0.5383,
      "step": 25986
    },
    {
      "epoch": 1.4552021502967858,
      "grad_norm": 1.4194258451461792,
      "learning_rate": 9.842763157894738e-05,
      "loss": 0.5295,
      "step": 25987
    },
    {
      "epoch": 1.4552581476089148,
      "grad_norm": 1.1278876066207886,
      "learning_rate": 9.842736842105263e-05,
      "loss": 0.4407,
      "step": 25988
    },
    {
      "epoch": 1.4553141449210438,
      "grad_norm": 1.4255927801132202,
      "learning_rate": 9.84271052631579e-05,
      "loss": 0.6305,
      "step": 25989
    },
    {
      "epoch": 1.4553701422331728,
      "grad_norm": 1.2105761766433716,
      "learning_rate": 9.842684210526316e-05,
      "loss": 0.3823,
      "step": 25990
    },
    {
      "epoch": 1.4554261395453019,
      "grad_norm": 1.2208470106124878,
      "learning_rate": 9.842657894736844e-05,
      "loss": 0.4616,
      "step": 25991
    },
    {
      "epoch": 1.4554821368574309,
      "grad_norm": 1.30439293384552,
      "learning_rate": 9.84263157894737e-05,
      "loss": 0.3807,
      "step": 25992
    },
    {
      "epoch": 1.45553813416956,
      "grad_norm": 1.7102527618408203,
      "learning_rate": 9.842605263157894e-05,
      "loss": 0.4421,
      "step": 25993
    },
    {
      "epoch": 1.455594131481689,
      "grad_norm": 1.3844481706619263,
      "learning_rate": 9.842578947368421e-05,
      "loss": 0.5318,
      "step": 25994
    },
    {
      "epoch": 1.455650128793818,
      "grad_norm": 1.1847615242004395,
      "learning_rate": 9.842552631578947e-05,
      "loss": 0.3416,
      "step": 25995
    },
    {
      "epoch": 1.455706126105947,
      "grad_norm": 1.346411943435669,
      "learning_rate": 9.842526315789475e-05,
      "loss": 0.5071,
      "step": 25996
    },
    {
      "epoch": 1.455762123418076,
      "grad_norm": 1.2523080110549927,
      "learning_rate": 9.842500000000001e-05,
      "loss": 0.5055,
      "step": 25997
    },
    {
      "epoch": 1.455818120730205,
      "grad_norm": 1.2922927141189575,
      "learning_rate": 9.842473684210527e-05,
      "loss": 0.4191,
      "step": 25998
    },
    {
      "epoch": 1.455874118042334,
      "grad_norm": 2.072600841522217,
      "learning_rate": 9.842447368421053e-05,
      "loss": 0.6591,
      "step": 25999
    },
    {
      "epoch": 1.455930115354463,
      "grad_norm": 1.4252487421035767,
      "learning_rate": 9.84242105263158e-05,
      "loss": 0.4047,
      "step": 26000
    },
    {
      "epoch": 1.455986112666592,
      "grad_norm": 1.4820553064346313,
      "learning_rate": 9.842394736842106e-05,
      "loss": 0.4762,
      "step": 26001
    },
    {
      "epoch": 1.456042109978721,
      "grad_norm": 1.4761065244674683,
      "learning_rate": 9.842368421052632e-05,
      "loss": 0.417,
      "step": 26002
    },
    {
      "epoch": 1.45609810729085,
      "grad_norm": 1.1952581405639648,
      "learning_rate": 9.842342105263158e-05,
      "loss": 0.5099,
      "step": 26003
    },
    {
      "epoch": 1.4561541046029791,
      "grad_norm": 1.2118871212005615,
      "learning_rate": 9.842315789473685e-05,
      "loss": 0.4029,
      "step": 26004
    },
    {
      "epoch": 1.4562101019151081,
      "grad_norm": 2.3767006397247314,
      "learning_rate": 9.842289473684211e-05,
      "loss": 0.4342,
      "step": 26005
    },
    {
      "epoch": 1.4562660992272372,
      "grad_norm": 1.329052448272705,
      "learning_rate": 9.842263157894737e-05,
      "loss": 0.4584,
      "step": 26006
    },
    {
      "epoch": 1.4563220965393662,
      "grad_norm": 1.3413516283035278,
      "learning_rate": 9.842236842105263e-05,
      "loss": 0.4386,
      "step": 26007
    },
    {
      "epoch": 1.4563780938514952,
      "grad_norm": 1.215601921081543,
      "learning_rate": 9.84221052631579e-05,
      "loss": 0.4626,
      "step": 26008
    },
    {
      "epoch": 1.4564340911636242,
      "grad_norm": 2.029501438140869,
      "learning_rate": 9.842184210526316e-05,
      "loss": 0.5052,
      "step": 26009
    },
    {
      "epoch": 1.4564900884757532,
      "grad_norm": 1.856543779373169,
      "learning_rate": 9.842157894736842e-05,
      "loss": 0.4854,
      "step": 26010
    },
    {
      "epoch": 1.4565460857878822,
      "grad_norm": 1.4724719524383545,
      "learning_rate": 9.842131578947368e-05,
      "loss": 0.5947,
      "step": 26011
    },
    {
      "epoch": 1.4566020831000113,
      "grad_norm": 1.2454043626785278,
      "learning_rate": 9.842105263157894e-05,
      "loss": 0.4048,
      "step": 26012
    },
    {
      "epoch": 1.4566580804121403,
      "grad_norm": 1.3093109130859375,
      "learning_rate": 9.842078947368422e-05,
      "loss": 0.4102,
      "step": 26013
    },
    {
      "epoch": 1.4567140777242693,
      "grad_norm": 1.3444128036499023,
      "learning_rate": 9.842052631578948e-05,
      "loss": 0.4793,
      "step": 26014
    },
    {
      "epoch": 1.4567700750363983,
      "grad_norm": 1.5629818439483643,
      "learning_rate": 9.842026315789474e-05,
      "loss": 0.4227,
      "step": 26015
    },
    {
      "epoch": 1.4568260723485273,
      "grad_norm": 1.532607078552246,
      "learning_rate": 9.842e-05,
      "loss": 0.6029,
      "step": 26016
    },
    {
      "epoch": 1.4568820696606564,
      "grad_norm": 1.3916041851043701,
      "learning_rate": 9.841973684210527e-05,
      "loss": 0.5419,
      "step": 26017
    },
    {
      "epoch": 1.4569380669727854,
      "grad_norm": 1.4412180185317993,
      "learning_rate": 9.841947368421053e-05,
      "loss": 0.4692,
      "step": 26018
    },
    {
      "epoch": 1.4569940642849144,
      "grad_norm": 1.4216548204421997,
      "learning_rate": 9.84192105263158e-05,
      "loss": 0.5973,
      "step": 26019
    },
    {
      "epoch": 1.4570500615970434,
      "grad_norm": 1.2250051498413086,
      "learning_rate": 9.841894736842105e-05,
      "loss": 0.5463,
      "step": 26020
    },
    {
      "epoch": 1.4571060589091724,
      "grad_norm": 1.220533847808838,
      "learning_rate": 9.841868421052632e-05,
      "loss": 0.4516,
      "step": 26021
    },
    {
      "epoch": 1.4571620562213015,
      "grad_norm": 1.1817001104354858,
      "learning_rate": 9.841842105263158e-05,
      "loss": 0.5348,
      "step": 26022
    },
    {
      "epoch": 1.4572180535334303,
      "grad_norm": 1.211058497428894,
      "learning_rate": 9.841815789473685e-05,
      "loss": 0.4168,
      "step": 26023
    },
    {
      "epoch": 1.4572740508455593,
      "grad_norm": 1.8724970817565918,
      "learning_rate": 9.841789473684211e-05,
      "loss": 0.4228,
      "step": 26024
    },
    {
      "epoch": 1.4573300481576883,
      "grad_norm": 1.1929121017456055,
      "learning_rate": 9.841763157894737e-05,
      "loss": 0.4154,
      "step": 26025
    },
    {
      "epoch": 1.4573860454698173,
      "grad_norm": 1.1333788633346558,
      "learning_rate": 9.841736842105263e-05,
      "loss": 0.3917,
      "step": 26026
    },
    {
      "epoch": 1.4574420427819463,
      "grad_norm": 1.3728362321853638,
      "learning_rate": 9.841710526315791e-05,
      "loss": 0.6962,
      "step": 26027
    },
    {
      "epoch": 1.4574980400940754,
      "grad_norm": 1.2028093338012695,
      "learning_rate": 9.841684210526317e-05,
      "loss": 0.3844,
      "step": 26028
    },
    {
      "epoch": 1.4575540374062044,
      "grad_norm": 1.2827948331832886,
      "learning_rate": 9.841657894736843e-05,
      "loss": 0.6314,
      "step": 26029
    },
    {
      "epoch": 1.4576100347183334,
      "grad_norm": 1.3257192373275757,
      "learning_rate": 9.841631578947369e-05,
      "loss": 0.5371,
      "step": 26030
    },
    {
      "epoch": 1.4576660320304624,
      "grad_norm": 1.401133418083191,
      "learning_rate": 9.841605263157895e-05,
      "loss": 0.4718,
      "step": 26031
    },
    {
      "epoch": 1.4577220293425914,
      "grad_norm": 1.27435302734375,
      "learning_rate": 9.841578947368422e-05,
      "loss": 0.4775,
      "step": 26032
    },
    {
      "epoch": 1.4577780266547204,
      "grad_norm": 1.7541958093643188,
      "learning_rate": 9.841552631578948e-05,
      "loss": 0.4958,
      "step": 26033
    },
    {
      "epoch": 1.4578340239668495,
      "grad_norm": 1.232792615890503,
      "learning_rate": 9.841526315789474e-05,
      "loss": 0.4311,
      "step": 26034
    },
    {
      "epoch": 1.4578900212789785,
      "grad_norm": 1.480499029159546,
      "learning_rate": 9.8415e-05,
      "loss": 0.5128,
      "step": 26035
    },
    {
      "epoch": 1.4579460185911075,
      "grad_norm": 1.412124514579773,
      "learning_rate": 9.841473684210527e-05,
      "loss": 0.4031,
      "step": 26036
    },
    {
      "epoch": 1.4580020159032365,
      "grad_norm": 1.420940637588501,
      "learning_rate": 9.841447368421053e-05,
      "loss": 0.4616,
      "step": 26037
    },
    {
      "epoch": 1.4580580132153655,
      "grad_norm": 1.3835965394973755,
      "learning_rate": 9.841421052631579e-05,
      "loss": 0.4683,
      "step": 26038
    },
    {
      "epoch": 1.4581140105274946,
      "grad_norm": 1.6787493228912354,
      "learning_rate": 9.841394736842105e-05,
      "loss": 0.5847,
      "step": 26039
    },
    {
      "epoch": 1.4581700078396236,
      "grad_norm": 1.1006306409835815,
      "learning_rate": 9.841368421052632e-05,
      "loss": 0.3483,
      "step": 26040
    },
    {
      "epoch": 1.4582260051517526,
      "grad_norm": 1.5119205713272095,
      "learning_rate": 9.841342105263158e-05,
      "loss": 0.5986,
      "step": 26041
    },
    {
      "epoch": 1.4582820024638816,
      "grad_norm": 1.6595722436904907,
      "learning_rate": 9.841315789473686e-05,
      "loss": 0.4327,
      "step": 26042
    },
    {
      "epoch": 1.4583379997760106,
      "grad_norm": 1.4033514261245728,
      "learning_rate": 9.84128947368421e-05,
      "loss": 0.499,
      "step": 26043
    },
    {
      "epoch": 1.4583939970881397,
      "grad_norm": 1.226958155632019,
      "learning_rate": 9.841263157894738e-05,
      "loss": 0.3799,
      "step": 26044
    },
    {
      "epoch": 1.4584499944002687,
      "grad_norm": 1.127150058746338,
      "learning_rate": 9.841236842105264e-05,
      "loss": 0.2782,
      "step": 26045
    },
    {
      "epoch": 1.4585059917123977,
      "grad_norm": 1.2230535745620728,
      "learning_rate": 9.84121052631579e-05,
      "loss": 0.4618,
      "step": 26046
    },
    {
      "epoch": 1.4585619890245267,
      "grad_norm": 1.1991167068481445,
      "learning_rate": 9.841184210526317e-05,
      "loss": 0.3825,
      "step": 26047
    },
    {
      "epoch": 1.4586179863366557,
      "grad_norm": 1.8252909183502197,
      "learning_rate": 9.841157894736842e-05,
      "loss": 0.5257,
      "step": 26048
    },
    {
      "epoch": 1.4586739836487848,
      "grad_norm": 1.4885025024414062,
      "learning_rate": 9.841131578947369e-05,
      "loss": 0.5272,
      "step": 26049
    },
    {
      "epoch": 1.4587299809609138,
      "grad_norm": 1.4274439811706543,
      "learning_rate": 9.841105263157895e-05,
      "loss": 0.488,
      "step": 26050
    },
    {
      "epoch": 1.4587859782730428,
      "grad_norm": 1.3002328872680664,
      "learning_rate": 9.841078947368422e-05,
      "loss": 0.4252,
      "step": 26051
    },
    {
      "epoch": 1.4588419755851718,
      "grad_norm": 1.3177926540374756,
      "learning_rate": 9.841052631578948e-05,
      "loss": 0.4527,
      "step": 26052
    },
    {
      "epoch": 1.4588979728973008,
      "grad_norm": 1.2957621812820435,
      "learning_rate": 9.841026315789474e-05,
      "loss": 0.4261,
      "step": 26053
    },
    {
      "epoch": 1.4589539702094299,
      "grad_norm": 1.5503376722335815,
      "learning_rate": 9.841e-05,
      "loss": 0.4861,
      "step": 26054
    },
    {
      "epoch": 1.4590099675215589,
      "grad_norm": 1.515969157218933,
      "learning_rate": 9.840973684210527e-05,
      "loss": 0.6048,
      "step": 26055
    },
    {
      "epoch": 1.459065964833688,
      "grad_norm": 1.3552058935165405,
      "learning_rate": 9.840947368421053e-05,
      "loss": 0.4295,
      "step": 26056
    },
    {
      "epoch": 1.459121962145817,
      "grad_norm": 1.1816718578338623,
      "learning_rate": 9.84092105263158e-05,
      "loss": 0.4525,
      "step": 26057
    },
    {
      "epoch": 1.459177959457946,
      "grad_norm": 1.6081931591033936,
      "learning_rate": 9.840894736842105e-05,
      "loss": 0.5457,
      "step": 26058
    },
    {
      "epoch": 1.459233956770075,
      "grad_norm": 1.6812665462493896,
      "learning_rate": 9.840868421052633e-05,
      "loss": 0.4257,
      "step": 26059
    },
    {
      "epoch": 1.459289954082204,
      "grad_norm": 1.229638695716858,
      "learning_rate": 9.840842105263159e-05,
      "loss": 0.4609,
      "step": 26060
    },
    {
      "epoch": 1.459345951394333,
      "grad_norm": 1.4026283025741577,
      "learning_rate": 9.840815789473685e-05,
      "loss": 0.5036,
      "step": 26061
    },
    {
      "epoch": 1.459401948706462,
      "grad_norm": 1.263132095336914,
      "learning_rate": 9.84078947368421e-05,
      "loss": 0.4865,
      "step": 26062
    },
    {
      "epoch": 1.459457946018591,
      "grad_norm": 1.5632518529891968,
      "learning_rate": 9.840763157894737e-05,
      "loss": 0.6191,
      "step": 26063
    },
    {
      "epoch": 1.45951394333072,
      "grad_norm": 1.5699266195297241,
      "learning_rate": 9.840736842105264e-05,
      "loss": 0.5325,
      "step": 26064
    },
    {
      "epoch": 1.459569940642849,
      "grad_norm": 1.2045263051986694,
      "learning_rate": 9.84071052631579e-05,
      "loss": 0.3987,
      "step": 26065
    },
    {
      "epoch": 1.459625937954978,
      "grad_norm": 5.441231727600098,
      "learning_rate": 9.840684210526316e-05,
      "loss": 0.5344,
      "step": 26066
    },
    {
      "epoch": 1.459681935267107,
      "grad_norm": 1.2703382968902588,
      "learning_rate": 9.840657894736842e-05,
      "loss": 0.5315,
      "step": 26067
    },
    {
      "epoch": 1.4597379325792361,
      "grad_norm": 1.1619566679000854,
      "learning_rate": 9.840631578947369e-05,
      "loss": 0.3904,
      "step": 26068
    },
    {
      "epoch": 1.4597939298913651,
      "grad_norm": 1.2699205875396729,
      "learning_rate": 9.840605263157895e-05,
      "loss": 0.4152,
      "step": 26069
    },
    {
      "epoch": 1.4598499272034942,
      "grad_norm": 1.5068904161453247,
      "learning_rate": 9.840578947368421e-05,
      "loss": 0.393,
      "step": 26070
    },
    {
      "epoch": 1.4599059245156232,
      "grad_norm": 1.146650791168213,
      "learning_rate": 9.840552631578947e-05,
      "loss": 0.3644,
      "step": 26071
    },
    {
      "epoch": 1.4599619218277522,
      "grad_norm": 1.2180830240249634,
      "learning_rate": 9.840526315789474e-05,
      "loss": 0.3651,
      "step": 26072
    },
    {
      "epoch": 1.4600179191398812,
      "grad_norm": 1.6138577461242676,
      "learning_rate": 9.8405e-05,
      "loss": 0.4821,
      "step": 26073
    },
    {
      "epoch": 1.4600739164520102,
      "grad_norm": 1.2507485151290894,
      "learning_rate": 9.840473684210528e-05,
      "loss": 0.6241,
      "step": 26074
    },
    {
      "epoch": 1.4601299137641393,
      "grad_norm": 1.2357007265090942,
      "learning_rate": 9.840447368421052e-05,
      "loss": 0.3541,
      "step": 26075
    },
    {
      "epoch": 1.4601859110762683,
      "grad_norm": 1.3801939487457275,
      "learning_rate": 9.84042105263158e-05,
      "loss": 0.7115,
      "step": 26076
    },
    {
      "epoch": 1.4602419083883973,
      "grad_norm": 1.485823631286621,
      "learning_rate": 9.840394736842106e-05,
      "loss": 0.4691,
      "step": 26077
    },
    {
      "epoch": 1.4602979057005263,
      "grad_norm": 8.340812683105469,
      "learning_rate": 9.840368421052633e-05,
      "loss": 0.3016,
      "step": 26078
    },
    {
      "epoch": 1.4603539030126553,
      "grad_norm": 1.2271965742111206,
      "learning_rate": 9.840342105263159e-05,
      "loss": 0.4139,
      "step": 26079
    },
    {
      "epoch": 1.4604099003247843,
      "grad_norm": 1.0837594270706177,
      "learning_rate": 9.840315789473684e-05,
      "loss": 0.4571,
      "step": 26080
    },
    {
      "epoch": 1.4604658976369134,
      "grad_norm": 1.594258427619934,
      "learning_rate": 9.840289473684211e-05,
      "loss": 0.4861,
      "step": 26081
    },
    {
      "epoch": 1.4605218949490424,
      "grad_norm": 1.315037727355957,
      "learning_rate": 9.840263157894737e-05,
      "loss": 0.4359,
      "step": 26082
    },
    {
      "epoch": 1.4605778922611714,
      "grad_norm": 1.315635085105896,
      "learning_rate": 9.840236842105264e-05,
      "loss": 0.5626,
      "step": 26083
    },
    {
      "epoch": 1.4606338895733004,
      "grad_norm": 1.4542587995529175,
      "learning_rate": 9.84021052631579e-05,
      "loss": 0.4359,
      "step": 26084
    },
    {
      "epoch": 1.4606898868854294,
      "grad_norm": 1.3752509355545044,
      "learning_rate": 9.840184210526316e-05,
      "loss": 0.4102,
      "step": 26085
    },
    {
      "epoch": 1.4607458841975585,
      "grad_norm": 1.7064623832702637,
      "learning_rate": 9.840157894736842e-05,
      "loss": 0.4918,
      "step": 26086
    },
    {
      "epoch": 1.4608018815096875,
      "grad_norm": 1.3330295085906982,
      "learning_rate": 9.84013157894737e-05,
      "loss": 0.4255,
      "step": 26087
    },
    {
      "epoch": 1.4608578788218165,
      "grad_norm": 1.7486516237258911,
      "learning_rate": 9.840105263157895e-05,
      "loss": 0.5374,
      "step": 26088
    },
    {
      "epoch": 1.4609138761339455,
      "grad_norm": 1.2772011756896973,
      "learning_rate": 9.840078947368421e-05,
      "loss": 0.4463,
      "step": 26089
    },
    {
      "epoch": 1.4609698734460745,
      "grad_norm": 1.404482126235962,
      "learning_rate": 9.840052631578947e-05,
      "loss": 0.3757,
      "step": 26090
    },
    {
      "epoch": 1.4610258707582036,
      "grad_norm": 1.397269368171692,
      "learning_rate": 9.840026315789475e-05,
      "loss": 0.4529,
      "step": 26091
    },
    {
      "epoch": 1.4610818680703326,
      "grad_norm": 1.510548710823059,
      "learning_rate": 9.84e-05,
      "loss": 0.5331,
      "step": 26092
    },
    {
      "epoch": 1.4611378653824616,
      "grad_norm": 1.2802950143814087,
      "learning_rate": 9.839973684210527e-05,
      "loss": 0.4482,
      "step": 26093
    },
    {
      "epoch": 1.4611938626945906,
      "grad_norm": 1.235029935836792,
      "learning_rate": 9.839947368421053e-05,
      "loss": 0.5423,
      "step": 26094
    },
    {
      "epoch": 1.4612498600067196,
      "grad_norm": 1.1043256521224976,
      "learning_rate": 9.83992105263158e-05,
      "loss": 0.4166,
      "step": 26095
    },
    {
      "epoch": 1.4613058573188487,
      "grad_norm": 1.2232075929641724,
      "learning_rate": 9.839894736842106e-05,
      "loss": 0.3702,
      "step": 26096
    },
    {
      "epoch": 1.4613618546309777,
      "grad_norm": 1.1902234554290771,
      "learning_rate": 9.839868421052632e-05,
      "loss": 0.4973,
      "step": 26097
    },
    {
      "epoch": 1.4614178519431067,
      "grad_norm": 1.5543293952941895,
      "learning_rate": 9.839842105263158e-05,
      "loss": 0.6131,
      "step": 26098
    },
    {
      "epoch": 1.4614738492552357,
      "grad_norm": 1.4217603206634521,
      "learning_rate": 9.839815789473684e-05,
      "loss": 0.4294,
      "step": 26099
    },
    {
      "epoch": 1.4615298465673647,
      "grad_norm": 1.4103598594665527,
      "learning_rate": 9.839789473684211e-05,
      "loss": 0.422,
      "step": 26100
    },
    {
      "epoch": 1.4615858438794938,
      "grad_norm": 1.322417974472046,
      "learning_rate": 9.839763157894737e-05,
      "loss": 0.6959,
      "step": 26101
    },
    {
      "epoch": 1.4616418411916228,
      "grad_norm": 1.3699936866760254,
      "learning_rate": 9.839736842105264e-05,
      "loss": 0.6449,
      "step": 26102
    },
    {
      "epoch": 1.4616978385037518,
      "grad_norm": 1.6707324981689453,
      "learning_rate": 9.839710526315789e-05,
      "loss": 0.4287,
      "step": 26103
    },
    {
      "epoch": 1.4617538358158808,
      "grad_norm": 1.2176315784454346,
      "learning_rate": 9.839684210526316e-05,
      "loss": 0.3065,
      "step": 26104
    },
    {
      "epoch": 1.4618098331280098,
      "grad_norm": 1.8233367204666138,
      "learning_rate": 9.839657894736842e-05,
      "loss": 0.531,
      "step": 26105
    },
    {
      "epoch": 1.4618658304401388,
      "grad_norm": 1.1831194162368774,
      "learning_rate": 9.83963157894737e-05,
      "loss": 0.3419,
      "step": 26106
    },
    {
      "epoch": 1.4619218277522679,
      "grad_norm": 1.393722414970398,
      "learning_rate": 9.839605263157896e-05,
      "loss": 0.522,
      "step": 26107
    },
    {
      "epoch": 1.4619778250643969,
      "grad_norm": 1.2852668762207031,
      "learning_rate": 9.839578947368422e-05,
      "loss": 0.5382,
      "step": 26108
    },
    {
      "epoch": 1.462033822376526,
      "grad_norm": 1.2282413244247437,
      "learning_rate": 9.839552631578948e-05,
      "loss": 0.4142,
      "step": 26109
    },
    {
      "epoch": 1.462089819688655,
      "grad_norm": 1.3288555145263672,
      "learning_rate": 9.839526315789475e-05,
      "loss": 0.3723,
      "step": 26110
    },
    {
      "epoch": 1.462145817000784,
      "grad_norm": 1.4714993238449097,
      "learning_rate": 9.839500000000001e-05,
      "loss": 0.4981,
      "step": 26111
    },
    {
      "epoch": 1.462201814312913,
      "grad_norm": 1.2625560760498047,
      "learning_rate": 9.839473684210527e-05,
      "loss": 0.5205,
      "step": 26112
    },
    {
      "epoch": 1.462257811625042,
      "grad_norm": 3.8642263412475586,
      "learning_rate": 9.839447368421053e-05,
      "loss": 0.5482,
      "step": 26113
    },
    {
      "epoch": 1.462313808937171,
      "grad_norm": 1.502594232559204,
      "learning_rate": 9.839421052631579e-05,
      "loss": 0.5093,
      "step": 26114
    },
    {
      "epoch": 1.4623698062493,
      "grad_norm": 1.234735131263733,
      "learning_rate": 9.839394736842106e-05,
      "loss": 0.434,
      "step": 26115
    },
    {
      "epoch": 1.462425803561429,
      "grad_norm": 1.3437854051589966,
      "learning_rate": 9.839368421052632e-05,
      "loss": 0.365,
      "step": 26116
    },
    {
      "epoch": 1.462481800873558,
      "grad_norm": 1.4200648069381714,
      "learning_rate": 9.839342105263158e-05,
      "loss": 0.4288,
      "step": 26117
    },
    {
      "epoch": 1.462537798185687,
      "grad_norm": 1.7690536975860596,
      "learning_rate": 9.839315789473684e-05,
      "loss": 0.5416,
      "step": 26118
    },
    {
      "epoch": 1.462593795497816,
      "grad_norm": 1.2500172853469849,
      "learning_rate": 9.839289473684211e-05,
      "loss": 0.374,
      "step": 26119
    },
    {
      "epoch": 1.4626497928099451,
      "grad_norm": 1.230628252029419,
      "learning_rate": 9.839263157894737e-05,
      "loss": 0.5791,
      "step": 26120
    },
    {
      "epoch": 1.4627057901220741,
      "grad_norm": 1.2279032468795776,
      "learning_rate": 9.839236842105263e-05,
      "loss": 0.4471,
      "step": 26121
    },
    {
      "epoch": 1.4627617874342032,
      "grad_norm": 1.8044170141220093,
      "learning_rate": 9.83921052631579e-05,
      "loss": 0.8309,
      "step": 26122
    },
    {
      "epoch": 1.4628177847463322,
      "grad_norm": 1.2737913131713867,
      "learning_rate": 9.839184210526317e-05,
      "loss": 0.5838,
      "step": 26123
    },
    {
      "epoch": 1.4628737820584612,
      "grad_norm": 1.3217850923538208,
      "learning_rate": 9.839157894736843e-05,
      "loss": 0.4428,
      "step": 26124
    },
    {
      "epoch": 1.4629297793705902,
      "grad_norm": 1.421263337135315,
      "learning_rate": 9.839131578947369e-05,
      "loss": 0.457,
      "step": 26125
    },
    {
      "epoch": 1.4629857766827192,
      "grad_norm": 1.8002113103866577,
      "learning_rate": 9.839105263157895e-05,
      "loss": 0.5559,
      "step": 26126
    },
    {
      "epoch": 1.4630417739948482,
      "grad_norm": 1.7728595733642578,
      "learning_rate": 9.839078947368422e-05,
      "loss": 0.6489,
      "step": 26127
    },
    {
      "epoch": 1.4630977713069773,
      "grad_norm": 1.3049659729003906,
      "learning_rate": 9.839052631578948e-05,
      "loss": 0.4105,
      "step": 26128
    },
    {
      "epoch": 1.4631537686191063,
      "grad_norm": 1.3368240594863892,
      "learning_rate": 9.839026315789475e-05,
      "loss": 0.4986,
      "step": 26129
    },
    {
      "epoch": 1.4632097659312353,
      "grad_norm": 1.1114985942840576,
      "learning_rate": 9.839e-05,
      "loss": 0.4899,
      "step": 26130
    },
    {
      "epoch": 1.4632657632433643,
      "grad_norm": 1.4967657327651978,
      "learning_rate": 9.838973684210526e-05,
      "loss": 0.419,
      "step": 26131
    },
    {
      "epoch": 1.4633217605554933,
      "grad_norm": 1.5984803438186646,
      "learning_rate": 9.838947368421053e-05,
      "loss": 0.4878,
      "step": 26132
    },
    {
      "epoch": 1.4633777578676224,
      "grad_norm": 1.439347505569458,
      "learning_rate": 9.838921052631579e-05,
      "loss": 0.4475,
      "step": 26133
    },
    {
      "epoch": 1.4634337551797514,
      "grad_norm": 1.2261884212493896,
      "learning_rate": 9.838894736842106e-05,
      "loss": 0.434,
      "step": 26134
    },
    {
      "epoch": 1.4634897524918804,
      "grad_norm": 1.3100343942642212,
      "learning_rate": 9.838868421052631e-05,
      "loss": 0.5021,
      "step": 26135
    },
    {
      "epoch": 1.4635457498040094,
      "grad_norm": 1.1510303020477295,
      "learning_rate": 9.838842105263158e-05,
      "loss": 0.4006,
      "step": 26136
    },
    {
      "epoch": 1.4636017471161384,
      "grad_norm": 1.1804864406585693,
      "learning_rate": 9.838815789473684e-05,
      "loss": 0.4323,
      "step": 26137
    },
    {
      "epoch": 1.4636577444282675,
      "grad_norm": 1.1206936836242676,
      "learning_rate": 9.838789473684212e-05,
      "loss": 0.4375,
      "step": 26138
    },
    {
      "epoch": 1.4637137417403965,
      "grad_norm": 1.62826669216156,
      "learning_rate": 9.838763157894738e-05,
      "loss": 0.5586,
      "step": 26139
    },
    {
      "epoch": 1.4637697390525255,
      "grad_norm": 1.252765417098999,
      "learning_rate": 9.838736842105264e-05,
      "loss": 0.4176,
      "step": 26140
    },
    {
      "epoch": 1.4638257363646545,
      "grad_norm": 1.4438307285308838,
      "learning_rate": 9.83871052631579e-05,
      "loss": 0.5453,
      "step": 26141
    },
    {
      "epoch": 1.4638817336767835,
      "grad_norm": 1.2927666902542114,
      "learning_rate": 9.838684210526317e-05,
      "loss": 0.4684,
      "step": 26142
    },
    {
      "epoch": 1.4639377309889126,
      "grad_norm": 1.4094184637069702,
      "learning_rate": 9.838657894736843e-05,
      "loss": 0.4477,
      "step": 26143
    },
    {
      "epoch": 1.4639937283010416,
      "grad_norm": 1.346919059753418,
      "learning_rate": 9.838631578947369e-05,
      "loss": 0.4416,
      "step": 26144
    },
    {
      "epoch": 1.4640497256131706,
      "grad_norm": 1.358778953552246,
      "learning_rate": 9.838605263157895e-05,
      "loss": 0.4921,
      "step": 26145
    },
    {
      "epoch": 1.4641057229252996,
      "grad_norm": 1.0951234102249146,
      "learning_rate": 9.838578947368422e-05,
      "loss": 0.4282,
      "step": 26146
    },
    {
      "epoch": 1.4641617202374286,
      "grad_norm": 1.1184499263763428,
      "learning_rate": 9.838552631578948e-05,
      "loss": 0.4633,
      "step": 26147
    },
    {
      "epoch": 1.4642177175495577,
      "grad_norm": 2.312195062637329,
      "learning_rate": 9.838526315789474e-05,
      "loss": 0.4097,
      "step": 26148
    },
    {
      "epoch": 1.4642737148616867,
      "grad_norm": 1.454043984413147,
      "learning_rate": 9.8385e-05,
      "loss": 0.4175,
      "step": 26149
    },
    {
      "epoch": 1.4643297121738157,
      "grad_norm": 1.3082998991012573,
      "learning_rate": 9.838473684210526e-05,
      "loss": 0.5793,
      "step": 26150
    },
    {
      "epoch": 1.4643857094859447,
      "grad_norm": 1.2758334875106812,
      "learning_rate": 9.838447368421053e-05,
      "loss": 0.4464,
      "step": 26151
    },
    {
      "epoch": 1.4644417067980737,
      "grad_norm": 1.262650728225708,
      "learning_rate": 9.83842105263158e-05,
      "loss": 0.4955,
      "step": 26152
    },
    {
      "epoch": 1.4644977041102027,
      "grad_norm": 1.4613919258117676,
      "learning_rate": 9.838394736842105e-05,
      "loss": 0.4926,
      "step": 26153
    },
    {
      "epoch": 1.4645537014223318,
      "grad_norm": 1.3332149982452393,
      "learning_rate": 9.838368421052631e-05,
      "loss": 0.3976,
      "step": 26154
    },
    {
      "epoch": 1.4646096987344608,
      "grad_norm": 2.993288278579712,
      "learning_rate": 9.838342105263159e-05,
      "loss": 0.5283,
      "step": 26155
    },
    {
      "epoch": 1.4646656960465898,
      "grad_norm": 1.2900582551956177,
      "learning_rate": 9.838315789473685e-05,
      "loss": 0.4609,
      "step": 26156
    },
    {
      "epoch": 1.4647216933587188,
      "grad_norm": 1.1441998481750488,
      "learning_rate": 9.838289473684212e-05,
      "loss": 0.4822,
      "step": 26157
    },
    {
      "epoch": 1.4647776906708478,
      "grad_norm": 2.2140674591064453,
      "learning_rate": 9.838263157894737e-05,
      "loss": 0.3641,
      "step": 26158
    },
    {
      "epoch": 1.4648336879829769,
      "grad_norm": 1.6417720317840576,
      "learning_rate": 9.838236842105264e-05,
      "loss": 0.474,
      "step": 26159
    },
    {
      "epoch": 1.4648896852951059,
      "grad_norm": 1.185815691947937,
      "learning_rate": 9.83821052631579e-05,
      "loss": 0.3657,
      "step": 26160
    },
    {
      "epoch": 1.464945682607235,
      "grad_norm": 1.2073918581008911,
      "learning_rate": 9.838184210526317e-05,
      "loss": 0.4163,
      "step": 26161
    },
    {
      "epoch": 1.465001679919364,
      "grad_norm": 1.1560301780700684,
      "learning_rate": 9.838157894736842e-05,
      "loss": 0.4289,
      "step": 26162
    },
    {
      "epoch": 1.465057677231493,
      "grad_norm": 1.4747960567474365,
      "learning_rate": 9.838131578947369e-05,
      "loss": 0.5165,
      "step": 26163
    },
    {
      "epoch": 1.465113674543622,
      "grad_norm": 1.760401725769043,
      "learning_rate": 9.838105263157895e-05,
      "loss": 0.5872,
      "step": 26164
    },
    {
      "epoch": 1.465169671855751,
      "grad_norm": 1.168134093284607,
      "learning_rate": 9.838078947368422e-05,
      "loss": 0.3327,
      "step": 26165
    },
    {
      "epoch": 1.46522566916788,
      "grad_norm": 1.3292665481567383,
      "learning_rate": 9.838052631578948e-05,
      "loss": 0.5577,
      "step": 26166
    },
    {
      "epoch": 1.465281666480009,
      "grad_norm": 1.2612038850784302,
      "learning_rate": 9.838026315789473e-05,
      "loss": 0.4686,
      "step": 26167
    },
    {
      "epoch": 1.465337663792138,
      "grad_norm": 1.4005955457687378,
      "learning_rate": 9.838e-05,
      "loss": 0.4274,
      "step": 26168
    },
    {
      "epoch": 1.465393661104267,
      "grad_norm": 1.398913025856018,
      "learning_rate": 9.837973684210526e-05,
      "loss": 0.4616,
      "step": 26169
    },
    {
      "epoch": 1.465449658416396,
      "grad_norm": 3.524045944213867,
      "learning_rate": 9.837947368421054e-05,
      "loss": 0.4242,
      "step": 26170
    },
    {
      "epoch": 1.465505655728525,
      "grad_norm": 1.2521146535873413,
      "learning_rate": 9.83792105263158e-05,
      "loss": 0.3885,
      "step": 26171
    },
    {
      "epoch": 1.465561653040654,
      "grad_norm": 1.2548400163650513,
      "learning_rate": 9.837894736842106e-05,
      "loss": 0.4356,
      "step": 26172
    },
    {
      "epoch": 1.4656176503527831,
      "grad_norm": 1.3966028690338135,
      "learning_rate": 9.837868421052632e-05,
      "loss": 0.5344,
      "step": 26173
    },
    {
      "epoch": 1.4656736476649121,
      "grad_norm": 1.1401585340499878,
      "learning_rate": 9.837842105263159e-05,
      "loss": 0.4329,
      "step": 26174
    },
    {
      "epoch": 1.4657296449770412,
      "grad_norm": 1.516364336013794,
      "learning_rate": 9.837815789473685e-05,
      "loss": 0.4669,
      "step": 26175
    },
    {
      "epoch": 1.4657856422891702,
      "grad_norm": 1.3908380270004272,
      "learning_rate": 9.837789473684211e-05,
      "loss": 0.5796,
      "step": 26176
    },
    {
      "epoch": 1.4658416396012992,
      "grad_norm": 1.1535483598709106,
      "learning_rate": 9.837763157894737e-05,
      "loss": 0.3513,
      "step": 26177
    },
    {
      "epoch": 1.4658976369134282,
      "grad_norm": 1.364871621131897,
      "learning_rate": 9.837736842105264e-05,
      "loss": 0.6552,
      "step": 26178
    },
    {
      "epoch": 1.4659536342255572,
      "grad_norm": 1.3948826789855957,
      "learning_rate": 9.83771052631579e-05,
      "loss": 0.5461,
      "step": 26179
    },
    {
      "epoch": 1.4660096315376863,
      "grad_norm": 1.451411485671997,
      "learning_rate": 9.837684210526316e-05,
      "loss": 0.5771,
      "step": 26180
    },
    {
      "epoch": 1.4660656288498153,
      "grad_norm": 1.616327166557312,
      "learning_rate": 9.837657894736842e-05,
      "loss": 0.4818,
      "step": 26181
    },
    {
      "epoch": 1.4661216261619443,
      "grad_norm": 1.4954359531402588,
      "learning_rate": 9.83763157894737e-05,
      "loss": 0.3666,
      "step": 26182
    },
    {
      "epoch": 1.4661776234740733,
      "grad_norm": 2.1145246028900146,
      "learning_rate": 9.837605263157895e-05,
      "loss": 0.5932,
      "step": 26183
    },
    {
      "epoch": 1.4662336207862023,
      "grad_norm": 1.4395397901535034,
      "learning_rate": 9.837578947368421e-05,
      "loss": 0.3956,
      "step": 26184
    },
    {
      "epoch": 1.4662896180983314,
      "grad_norm": 1.270071268081665,
      "learning_rate": 9.837552631578947e-05,
      "loss": 0.4866,
      "step": 26185
    },
    {
      "epoch": 1.4663456154104604,
      "grad_norm": 1.183920979499817,
      "learning_rate": 9.837526315789473e-05,
      "loss": 0.4086,
      "step": 26186
    },
    {
      "epoch": 1.4664016127225894,
      "grad_norm": 2.1491172313690186,
      "learning_rate": 9.8375e-05,
      "loss": 0.619,
      "step": 26187
    },
    {
      "epoch": 1.4664576100347184,
      "grad_norm": 1.3373596668243408,
      "learning_rate": 9.837473684210527e-05,
      "loss": 0.425,
      "step": 26188
    },
    {
      "epoch": 1.4665136073468474,
      "grad_norm": 1.3180291652679443,
      "learning_rate": 9.837447368421054e-05,
      "loss": 0.4633,
      "step": 26189
    },
    {
      "epoch": 1.4665696046589765,
      "grad_norm": 1.2733041048049927,
      "learning_rate": 9.837421052631579e-05,
      "loss": 0.5698,
      "step": 26190
    },
    {
      "epoch": 1.4666256019711055,
      "grad_norm": 1.4104869365692139,
      "learning_rate": 9.837394736842106e-05,
      "loss": 0.6728,
      "step": 26191
    },
    {
      "epoch": 1.4666815992832345,
      "grad_norm": 1.3825445175170898,
      "learning_rate": 9.837368421052632e-05,
      "loss": 0.4538,
      "step": 26192
    },
    {
      "epoch": 1.4667375965953635,
      "grad_norm": 1.294661283493042,
      "learning_rate": 9.837342105263159e-05,
      "loss": 0.4541,
      "step": 26193
    },
    {
      "epoch": 1.4667935939074925,
      "grad_norm": 1.3568086624145508,
      "learning_rate": 9.837315789473685e-05,
      "loss": 0.5706,
      "step": 26194
    },
    {
      "epoch": 1.4668495912196216,
      "grad_norm": 2.1935811042785645,
      "learning_rate": 9.837289473684211e-05,
      "loss": 0.5933,
      "step": 26195
    },
    {
      "epoch": 1.4669055885317506,
      "grad_norm": 1.3590644598007202,
      "learning_rate": 9.837263157894737e-05,
      "loss": 0.5384,
      "step": 26196
    },
    {
      "epoch": 1.4669615858438796,
      "grad_norm": 1.1283819675445557,
      "learning_rate": 9.837236842105264e-05,
      "loss": 0.4398,
      "step": 26197
    },
    {
      "epoch": 1.4670175831560086,
      "grad_norm": 1.3206721544265747,
      "learning_rate": 9.83721052631579e-05,
      "loss": 0.4794,
      "step": 26198
    },
    {
      "epoch": 1.4670735804681376,
      "grad_norm": 1.4902489185333252,
      "learning_rate": 9.837184210526316e-05,
      "loss": 0.5192,
      "step": 26199
    },
    {
      "epoch": 1.4671295777802666,
      "grad_norm": 1.4239227771759033,
      "learning_rate": 9.837157894736842e-05,
      "loss": 0.4689,
      "step": 26200
    },
    {
      "epoch": 1.4671855750923957,
      "grad_norm": 1.2795462608337402,
      "learning_rate": 9.837131578947368e-05,
      "loss": 0.428,
      "step": 26201
    },
    {
      "epoch": 1.4672415724045247,
      "grad_norm": 1.591731309890747,
      "learning_rate": 9.837105263157896e-05,
      "loss": 0.4965,
      "step": 26202
    },
    {
      "epoch": 1.4672975697166537,
      "grad_norm": 1.4575308561325073,
      "learning_rate": 9.837078947368422e-05,
      "loss": 0.4823,
      "step": 26203
    },
    {
      "epoch": 1.4673535670287827,
      "grad_norm": 1.3647435903549194,
      "learning_rate": 9.837052631578948e-05,
      "loss": 0.3855,
      "step": 26204
    },
    {
      "epoch": 1.4674095643409117,
      "grad_norm": 1.602904200553894,
      "learning_rate": 9.837026315789474e-05,
      "loss": 0.5449,
      "step": 26205
    },
    {
      "epoch": 1.4674655616530408,
      "grad_norm": 1.2654188871383667,
      "learning_rate": 9.837000000000001e-05,
      "loss": 0.4398,
      "step": 26206
    },
    {
      "epoch": 1.4675215589651698,
      "grad_norm": 1.1874247789382935,
      "learning_rate": 9.836973684210527e-05,
      "loss": 0.3172,
      "step": 26207
    },
    {
      "epoch": 1.4675775562772988,
      "grad_norm": 1.2212897539138794,
      "learning_rate": 9.836947368421053e-05,
      "loss": 0.4092,
      "step": 26208
    },
    {
      "epoch": 1.4676335535894278,
      "grad_norm": 1.4444659948349,
      "learning_rate": 9.836921052631579e-05,
      "loss": 0.4105,
      "step": 26209
    },
    {
      "epoch": 1.4676895509015568,
      "grad_norm": 1.3629571199417114,
      "learning_rate": 9.836894736842106e-05,
      "loss": 0.6394,
      "step": 26210
    },
    {
      "epoch": 1.4677455482136859,
      "grad_norm": 1.2981512546539307,
      "learning_rate": 9.836868421052632e-05,
      "loss": 0.5366,
      "step": 26211
    },
    {
      "epoch": 1.4678015455258149,
      "grad_norm": 1.6782872676849365,
      "learning_rate": 9.83684210526316e-05,
      "loss": 0.5427,
      "step": 26212
    },
    {
      "epoch": 1.467857542837944,
      "grad_norm": 1.6365340948104858,
      "learning_rate": 9.836815789473684e-05,
      "loss": 0.494,
      "step": 26213
    },
    {
      "epoch": 1.467913540150073,
      "grad_norm": 1.3263881206512451,
      "learning_rate": 9.836789473684211e-05,
      "loss": 0.4234,
      "step": 26214
    },
    {
      "epoch": 1.467969537462202,
      "grad_norm": 1.6355187892913818,
      "learning_rate": 9.836763157894737e-05,
      "loss": 0.6058,
      "step": 26215
    },
    {
      "epoch": 1.468025534774331,
      "grad_norm": 1.500312328338623,
      "learning_rate": 9.836736842105265e-05,
      "loss": 0.4829,
      "step": 26216
    },
    {
      "epoch": 1.46808153208646,
      "grad_norm": 1.261370301246643,
      "learning_rate": 9.836710526315789e-05,
      "loss": 0.5173,
      "step": 26217
    },
    {
      "epoch": 1.468137529398589,
      "grad_norm": 1.103760004043579,
      "learning_rate": 9.836684210526315e-05,
      "loss": 0.4576,
      "step": 26218
    },
    {
      "epoch": 1.468193526710718,
      "grad_norm": 1.3855925798416138,
      "learning_rate": 9.836657894736843e-05,
      "loss": 0.4705,
      "step": 26219
    },
    {
      "epoch": 1.468249524022847,
      "grad_norm": 1.7272679805755615,
      "learning_rate": 9.836631578947369e-05,
      "loss": 0.4376,
      "step": 26220
    },
    {
      "epoch": 1.468305521334976,
      "grad_norm": 1.2779830694198608,
      "learning_rate": 9.836605263157896e-05,
      "loss": 0.558,
      "step": 26221
    },
    {
      "epoch": 1.468361518647105,
      "grad_norm": 1.3490245342254639,
      "learning_rate": 9.83657894736842e-05,
      "loss": 0.4422,
      "step": 26222
    },
    {
      "epoch": 1.468417515959234,
      "grad_norm": 1.7253708839416504,
      "learning_rate": 9.836552631578948e-05,
      "loss": 0.5349,
      "step": 26223
    },
    {
      "epoch": 1.468473513271363,
      "grad_norm": 1.2533903121948242,
      "learning_rate": 9.836526315789474e-05,
      "loss": 0.489,
      "step": 26224
    },
    {
      "epoch": 1.4685295105834921,
      "grad_norm": 1.0124671459197998,
      "learning_rate": 9.836500000000001e-05,
      "loss": 0.3777,
      "step": 26225
    },
    {
      "epoch": 1.4685855078956211,
      "grad_norm": 1.3425922393798828,
      "learning_rate": 9.836473684210527e-05,
      "loss": 0.4844,
      "step": 26226
    },
    {
      "epoch": 1.4686415052077502,
      "grad_norm": 1.3471434116363525,
      "learning_rate": 9.836447368421053e-05,
      "loss": 0.4589,
      "step": 26227
    },
    {
      "epoch": 1.4686975025198792,
      "grad_norm": 1.4200447797775269,
      "learning_rate": 9.836421052631579e-05,
      "loss": 0.4118,
      "step": 26228
    },
    {
      "epoch": 1.468753499832008,
      "grad_norm": 1.3445628881454468,
      "learning_rate": 9.836394736842106e-05,
      "loss": 0.4383,
      "step": 26229
    },
    {
      "epoch": 1.468809497144137,
      "grad_norm": 1.3946771621704102,
      "learning_rate": 9.836368421052632e-05,
      "loss": 0.4094,
      "step": 26230
    },
    {
      "epoch": 1.468865494456266,
      "grad_norm": 1.3378466367721558,
      "learning_rate": 9.836342105263158e-05,
      "loss": 0.4681,
      "step": 26231
    },
    {
      "epoch": 1.468921491768395,
      "grad_norm": 1.1821906566619873,
      "learning_rate": 9.836315789473684e-05,
      "loss": 0.3714,
      "step": 26232
    },
    {
      "epoch": 1.468977489080524,
      "grad_norm": 1.356821894645691,
      "learning_rate": 9.836289473684212e-05,
      "loss": 0.5454,
      "step": 26233
    },
    {
      "epoch": 1.469033486392653,
      "grad_norm": 1.5115985870361328,
      "learning_rate": 9.836263157894738e-05,
      "loss": 0.5112,
      "step": 26234
    },
    {
      "epoch": 1.469089483704782,
      "grad_norm": 1.245773196220398,
      "learning_rate": 9.836236842105264e-05,
      "loss": 0.4485,
      "step": 26235
    },
    {
      "epoch": 1.4691454810169111,
      "grad_norm": 1.9896646738052368,
      "learning_rate": 9.83621052631579e-05,
      "loss": 0.6119,
      "step": 26236
    },
    {
      "epoch": 1.4692014783290401,
      "grad_norm": 1.217750072479248,
      "learning_rate": 9.836184210526315e-05,
      "loss": 0.4095,
      "step": 26237
    },
    {
      "epoch": 1.4692574756411692,
      "grad_norm": 1.9007140398025513,
      "learning_rate": 9.836157894736843e-05,
      "loss": 0.4894,
      "step": 26238
    },
    {
      "epoch": 1.4693134729532982,
      "grad_norm": 1.4336644411087036,
      "learning_rate": 9.836131578947369e-05,
      "loss": 0.4557,
      "step": 26239
    },
    {
      "epoch": 1.4693694702654272,
      "grad_norm": 1.2235615253448486,
      "learning_rate": 9.836105263157895e-05,
      "loss": 0.4204,
      "step": 26240
    },
    {
      "epoch": 1.4694254675775562,
      "grad_norm": 1.5203443765640259,
      "learning_rate": 9.836078947368421e-05,
      "loss": 0.6011,
      "step": 26241
    },
    {
      "epoch": 1.4694814648896852,
      "grad_norm": 1.2648966312408447,
      "learning_rate": 9.836052631578948e-05,
      "loss": 0.5083,
      "step": 26242
    },
    {
      "epoch": 1.4695374622018142,
      "grad_norm": 1.2400827407836914,
      "learning_rate": 9.836026315789474e-05,
      "loss": 0.5313,
      "step": 26243
    },
    {
      "epoch": 1.4695934595139433,
      "grad_norm": 1.233296513557434,
      "learning_rate": 9.836000000000001e-05,
      "loss": 0.4499,
      "step": 26244
    },
    {
      "epoch": 1.4696494568260723,
      "grad_norm": 1.275831937789917,
      "learning_rate": 9.835973684210526e-05,
      "loss": 0.539,
      "step": 26245
    },
    {
      "epoch": 1.4697054541382013,
      "grad_norm": 1.2991141080856323,
      "learning_rate": 9.835947368421053e-05,
      "loss": 0.502,
      "step": 26246
    },
    {
      "epoch": 1.4697614514503303,
      "grad_norm": 1.430525541305542,
      "learning_rate": 9.835921052631579e-05,
      "loss": 0.489,
      "step": 26247
    },
    {
      "epoch": 1.4698174487624593,
      "grad_norm": 1.268449068069458,
      "learning_rate": 9.835894736842107e-05,
      "loss": 0.3601,
      "step": 26248
    },
    {
      "epoch": 1.4698734460745884,
      "grad_norm": 1.5674430131912231,
      "learning_rate": 9.835868421052633e-05,
      "loss": 0.6267,
      "step": 26249
    },
    {
      "epoch": 1.4699294433867174,
      "grad_norm": 1.63973069190979,
      "learning_rate": 9.835842105263159e-05,
      "loss": 0.5781,
      "step": 26250
    },
    {
      "epoch": 1.4699854406988464,
      "grad_norm": 1.3429597616195679,
      "learning_rate": 9.835815789473685e-05,
      "loss": 0.3855,
      "step": 26251
    },
    {
      "epoch": 1.4700414380109754,
      "grad_norm": 1.366430640220642,
      "learning_rate": 9.83578947368421e-05,
      "loss": 0.416,
      "step": 26252
    },
    {
      "epoch": 1.4700974353231044,
      "grad_norm": 1.4533041715621948,
      "learning_rate": 9.835763157894738e-05,
      "loss": 0.45,
      "step": 26253
    },
    {
      "epoch": 1.4701534326352335,
      "grad_norm": 1.2154712677001953,
      "learning_rate": 9.835736842105264e-05,
      "loss": 0.4355,
      "step": 26254
    },
    {
      "epoch": 1.4702094299473625,
      "grad_norm": 1.096551537513733,
      "learning_rate": 9.83571052631579e-05,
      "loss": 0.4278,
      "step": 26255
    },
    {
      "epoch": 1.4702654272594915,
      "grad_norm": 1.5150096416473389,
      "learning_rate": 9.835684210526316e-05,
      "loss": 0.4331,
      "step": 26256
    },
    {
      "epoch": 1.4703214245716205,
      "grad_norm": 1.4409756660461426,
      "learning_rate": 9.835657894736843e-05,
      "loss": 0.4369,
      "step": 26257
    },
    {
      "epoch": 1.4703774218837495,
      "grad_norm": 1.529038667678833,
      "learning_rate": 9.835631578947369e-05,
      "loss": 0.5506,
      "step": 26258
    },
    {
      "epoch": 1.4704334191958786,
      "grad_norm": 1.58342444896698,
      "learning_rate": 9.835605263157895e-05,
      "loss": 0.4856,
      "step": 26259
    },
    {
      "epoch": 1.4704894165080076,
      "grad_norm": 1.3333591222763062,
      "learning_rate": 9.835578947368421e-05,
      "loss": 0.4603,
      "step": 26260
    },
    {
      "epoch": 1.4705454138201366,
      "grad_norm": 1.2099251747131348,
      "learning_rate": 9.835552631578948e-05,
      "loss": 0.402,
      "step": 26261
    },
    {
      "epoch": 1.4706014111322656,
      "grad_norm": 1.2804086208343506,
      "learning_rate": 9.835526315789474e-05,
      "loss": 0.3806,
      "step": 26262
    },
    {
      "epoch": 1.4706574084443946,
      "grad_norm": 1.2557204961776733,
      "learning_rate": 9.8355e-05,
      "loss": 0.383,
      "step": 26263
    },
    {
      "epoch": 1.4707134057565237,
      "grad_norm": 1.328728199005127,
      "learning_rate": 9.835473684210526e-05,
      "loss": 0.6845,
      "step": 26264
    },
    {
      "epoch": 1.4707694030686527,
      "grad_norm": 1.572045922279358,
      "learning_rate": 9.835447368421054e-05,
      "loss": 0.4997,
      "step": 26265
    },
    {
      "epoch": 1.4708254003807817,
      "grad_norm": 1.8102408647537231,
      "learning_rate": 9.83542105263158e-05,
      "loss": 0.5642,
      "step": 26266
    },
    {
      "epoch": 1.4708813976929107,
      "grad_norm": 1.502687692642212,
      "learning_rate": 9.835394736842107e-05,
      "loss": 0.5193,
      "step": 26267
    },
    {
      "epoch": 1.4709373950050397,
      "grad_norm": 1.3405128717422485,
      "learning_rate": 9.835368421052631e-05,
      "loss": 0.4186,
      "step": 26268
    },
    {
      "epoch": 1.4709933923171687,
      "grad_norm": 1.430985450744629,
      "learning_rate": 9.835342105263159e-05,
      "loss": 0.7133,
      "step": 26269
    },
    {
      "epoch": 1.4710493896292978,
      "grad_norm": 1.7925950288772583,
      "learning_rate": 9.835315789473685e-05,
      "loss": 0.4835,
      "step": 26270
    },
    {
      "epoch": 1.4711053869414268,
      "grad_norm": 1.115112543106079,
      "learning_rate": 9.835289473684211e-05,
      "loss": 0.4697,
      "step": 26271
    },
    {
      "epoch": 1.4711613842535558,
      "grad_norm": 1.8055635690689087,
      "learning_rate": 9.835263157894737e-05,
      "loss": 0.6008,
      "step": 26272
    },
    {
      "epoch": 1.4712173815656848,
      "grad_norm": 1.398441195487976,
      "learning_rate": 9.835236842105263e-05,
      "loss": 0.4506,
      "step": 26273
    },
    {
      "epoch": 1.4712733788778138,
      "grad_norm": 1.3960886001586914,
      "learning_rate": 9.83521052631579e-05,
      "loss": 0.5928,
      "step": 26274
    },
    {
      "epoch": 1.4713293761899429,
      "grad_norm": 1.3009015321731567,
      "learning_rate": 9.835184210526316e-05,
      "loss": 0.4612,
      "step": 26275
    },
    {
      "epoch": 1.4713853735020719,
      "grad_norm": 1.52046537399292,
      "learning_rate": 9.835157894736843e-05,
      "loss": 0.5243,
      "step": 26276
    },
    {
      "epoch": 1.471441370814201,
      "grad_norm": 1.382170557975769,
      "learning_rate": 9.835131578947368e-05,
      "loss": 0.5536,
      "step": 26277
    },
    {
      "epoch": 1.47149736812633,
      "grad_norm": 1.4639800786972046,
      "learning_rate": 9.835105263157895e-05,
      "loss": 0.4572,
      "step": 26278
    },
    {
      "epoch": 1.471553365438459,
      "grad_norm": 1.494409203529358,
      "learning_rate": 9.835078947368421e-05,
      "loss": 0.4796,
      "step": 26279
    },
    {
      "epoch": 1.471609362750588,
      "grad_norm": 1.181498646736145,
      "learning_rate": 9.835052631578949e-05,
      "loss": 0.4657,
      "step": 26280
    },
    {
      "epoch": 1.471665360062717,
      "grad_norm": 1.3346881866455078,
      "learning_rate": 9.835026315789475e-05,
      "loss": 0.5247,
      "step": 26281
    },
    {
      "epoch": 1.471721357374846,
      "grad_norm": 1.2633123397827148,
      "learning_rate": 9.835e-05,
      "loss": 0.3755,
      "step": 26282
    },
    {
      "epoch": 1.471777354686975,
      "grad_norm": 1.2065980434417725,
      "learning_rate": 9.834973684210527e-05,
      "loss": 0.5416,
      "step": 26283
    },
    {
      "epoch": 1.471833351999104,
      "grad_norm": 1.511853814125061,
      "learning_rate": 9.834947368421054e-05,
      "loss": 0.5346,
      "step": 26284
    },
    {
      "epoch": 1.471889349311233,
      "grad_norm": 1.6110639572143555,
      "learning_rate": 9.83492105263158e-05,
      "loss": 0.5677,
      "step": 26285
    },
    {
      "epoch": 1.471945346623362,
      "grad_norm": 1.2345631122589111,
      "learning_rate": 9.834894736842106e-05,
      "loss": 0.6015,
      "step": 26286
    },
    {
      "epoch": 1.472001343935491,
      "grad_norm": 1.659584879875183,
      "learning_rate": 9.834868421052632e-05,
      "loss": 0.4492,
      "step": 26287
    },
    {
      "epoch": 1.47205734124762,
      "grad_norm": 1.4842394590377808,
      "learning_rate": 9.834842105263158e-05,
      "loss": 0.5161,
      "step": 26288
    },
    {
      "epoch": 1.4721133385597491,
      "grad_norm": 1.5364598035812378,
      "learning_rate": 9.834815789473685e-05,
      "loss": 0.5089,
      "step": 26289
    },
    {
      "epoch": 1.4721693358718781,
      "grad_norm": 1.1319185495376587,
      "learning_rate": 9.834789473684211e-05,
      "loss": 0.3774,
      "step": 26290
    },
    {
      "epoch": 1.4722253331840072,
      "grad_norm": 1.2564584016799927,
      "learning_rate": 9.834763157894737e-05,
      "loss": 0.4365,
      "step": 26291
    },
    {
      "epoch": 1.4722813304961362,
      "grad_norm": 1.386336088180542,
      "learning_rate": 9.834736842105263e-05,
      "loss": 0.4626,
      "step": 26292
    },
    {
      "epoch": 1.4723373278082652,
      "grad_norm": 1.2889857292175293,
      "learning_rate": 9.83471052631579e-05,
      "loss": 0.469,
      "step": 26293
    },
    {
      "epoch": 1.4723933251203942,
      "grad_norm": 2.604438304901123,
      "learning_rate": 9.834684210526316e-05,
      "loss": 0.4053,
      "step": 26294
    },
    {
      "epoch": 1.4724493224325232,
      "grad_norm": 1.454038381576538,
      "learning_rate": 9.834657894736842e-05,
      "loss": 0.6295,
      "step": 26295
    },
    {
      "epoch": 1.4725053197446523,
      "grad_norm": 1.471243143081665,
      "learning_rate": 9.834631578947368e-05,
      "loss": 0.4956,
      "step": 26296
    },
    {
      "epoch": 1.4725613170567813,
      "grad_norm": 1.1576184034347534,
      "learning_rate": 9.834605263157896e-05,
      "loss": 0.4578,
      "step": 26297
    },
    {
      "epoch": 1.4726173143689103,
      "grad_norm": 1.247467279434204,
      "learning_rate": 9.834578947368422e-05,
      "loss": 0.4447,
      "step": 26298
    },
    {
      "epoch": 1.4726733116810393,
      "grad_norm": 1.2596161365509033,
      "learning_rate": 9.834552631578949e-05,
      "loss": 0.4248,
      "step": 26299
    },
    {
      "epoch": 1.4727293089931683,
      "grad_norm": 1.3921743631362915,
      "learning_rate": 9.834526315789473e-05,
      "loss": 0.4215,
      "step": 26300
    },
    {
      "epoch": 1.4727853063052974,
      "grad_norm": 1.458272099494934,
      "learning_rate": 9.834500000000001e-05,
      "loss": 0.4621,
      "step": 26301
    },
    {
      "epoch": 1.4728413036174264,
      "grad_norm": 1.153385043144226,
      "learning_rate": 9.834473684210527e-05,
      "loss": 0.4612,
      "step": 26302
    },
    {
      "epoch": 1.4728973009295554,
      "grad_norm": 1.3927161693572998,
      "learning_rate": 9.834447368421054e-05,
      "loss": 0.4447,
      "step": 26303
    },
    {
      "epoch": 1.4729532982416844,
      "grad_norm": 1.2911595106124878,
      "learning_rate": 9.83442105263158e-05,
      "loss": 0.4224,
      "step": 26304
    },
    {
      "epoch": 1.4730092955538134,
      "grad_norm": 1.290352702140808,
      "learning_rate": 9.834394736842105e-05,
      "loss": 0.3922,
      "step": 26305
    },
    {
      "epoch": 1.4730652928659425,
      "grad_norm": 1.5228631496429443,
      "learning_rate": 9.834368421052632e-05,
      "loss": 0.4679,
      "step": 26306
    },
    {
      "epoch": 1.4731212901780715,
      "grad_norm": 1.3852399587631226,
      "learning_rate": 9.834342105263158e-05,
      "loss": 0.3971,
      "step": 26307
    },
    {
      "epoch": 1.4731772874902005,
      "grad_norm": 1.2399355173110962,
      "learning_rate": 9.834315789473685e-05,
      "loss": 0.4984,
      "step": 26308
    },
    {
      "epoch": 1.4732332848023295,
      "grad_norm": 1.2135710716247559,
      "learning_rate": 9.83428947368421e-05,
      "loss": 0.4384,
      "step": 26309
    },
    {
      "epoch": 1.4732892821144585,
      "grad_norm": 1.644197940826416,
      "learning_rate": 9.834263157894737e-05,
      "loss": 0.5053,
      "step": 26310
    },
    {
      "epoch": 1.4733452794265876,
      "grad_norm": 1.3441396951675415,
      "learning_rate": 9.834236842105263e-05,
      "loss": 0.3923,
      "step": 26311
    },
    {
      "epoch": 1.4734012767387166,
      "grad_norm": 1.3883205652236938,
      "learning_rate": 9.83421052631579e-05,
      "loss": 0.6124,
      "step": 26312
    },
    {
      "epoch": 1.4734572740508456,
      "grad_norm": 1.3514333963394165,
      "learning_rate": 9.834184210526317e-05,
      "loss": 0.5058,
      "step": 26313
    },
    {
      "epoch": 1.4735132713629746,
      "grad_norm": 1.5079749822616577,
      "learning_rate": 9.834157894736843e-05,
      "loss": 0.5274,
      "step": 26314
    },
    {
      "epoch": 1.4735692686751036,
      "grad_norm": 1.2684351205825806,
      "learning_rate": 9.834131578947368e-05,
      "loss": 0.4921,
      "step": 26315
    },
    {
      "epoch": 1.4736252659872326,
      "grad_norm": 1.300086498260498,
      "learning_rate": 9.834105263157896e-05,
      "loss": 0.4784,
      "step": 26316
    },
    {
      "epoch": 1.4736812632993617,
      "grad_norm": 1.300095796585083,
      "learning_rate": 9.834078947368422e-05,
      "loss": 0.4917,
      "step": 26317
    },
    {
      "epoch": 1.4737372606114907,
      "grad_norm": 1.8674501180648804,
      "learning_rate": 9.834052631578948e-05,
      "loss": 0.5852,
      "step": 26318
    },
    {
      "epoch": 1.4737932579236197,
      "grad_norm": 1.654843807220459,
      "learning_rate": 9.834026315789474e-05,
      "loss": 0.5459,
      "step": 26319
    },
    {
      "epoch": 1.4738492552357487,
      "grad_norm": 1.3569914102554321,
      "learning_rate": 9.834000000000001e-05,
      "loss": 0.4422,
      "step": 26320
    },
    {
      "epoch": 1.4739052525478777,
      "grad_norm": 1.52809739112854,
      "learning_rate": 9.833973684210527e-05,
      "loss": 0.2912,
      "step": 26321
    },
    {
      "epoch": 1.4739612498600068,
      "grad_norm": 1.3407853841781616,
      "learning_rate": 9.833947368421053e-05,
      "loss": 0.4849,
      "step": 26322
    },
    {
      "epoch": 1.4740172471721358,
      "grad_norm": 1.2122225761413574,
      "learning_rate": 9.833921052631579e-05,
      "loss": 0.3636,
      "step": 26323
    },
    {
      "epoch": 1.4740732444842648,
      "grad_norm": 1.421341061592102,
      "learning_rate": 9.833894736842105e-05,
      "loss": 0.4648,
      "step": 26324
    },
    {
      "epoch": 1.4741292417963938,
      "grad_norm": 1.3635461330413818,
      "learning_rate": 9.833868421052632e-05,
      "loss": 0.3847,
      "step": 26325
    },
    {
      "epoch": 1.4741852391085228,
      "grad_norm": 1.5558712482452393,
      "learning_rate": 9.833842105263158e-05,
      "loss": 0.7002,
      "step": 26326
    },
    {
      "epoch": 1.4742412364206519,
      "grad_norm": 1.379497766494751,
      "learning_rate": 9.833815789473684e-05,
      "loss": 0.5231,
      "step": 26327
    },
    {
      "epoch": 1.4742972337327809,
      "grad_norm": 1.337708830833435,
      "learning_rate": 9.83378947368421e-05,
      "loss": 0.3953,
      "step": 26328
    },
    {
      "epoch": 1.47435323104491,
      "grad_norm": 1.4808558225631714,
      "learning_rate": 9.833763157894738e-05,
      "loss": 0.4949,
      "step": 26329
    },
    {
      "epoch": 1.474409228357039,
      "grad_norm": 1.286215901374817,
      "learning_rate": 9.833736842105263e-05,
      "loss": 0.352,
      "step": 26330
    },
    {
      "epoch": 1.474465225669168,
      "grad_norm": 1.3624666929244995,
      "learning_rate": 9.833710526315791e-05,
      "loss": 0.5038,
      "step": 26331
    },
    {
      "epoch": 1.474521222981297,
      "grad_norm": 1.7593283653259277,
      "learning_rate": 9.833684210526315e-05,
      "loss": 0.5098,
      "step": 26332
    },
    {
      "epoch": 1.474577220293426,
      "grad_norm": 1.313141107559204,
      "learning_rate": 9.833657894736843e-05,
      "loss": 0.4492,
      "step": 26333
    },
    {
      "epoch": 1.474633217605555,
      "grad_norm": 1.3633440732955933,
      "learning_rate": 9.833631578947369e-05,
      "loss": 0.3685,
      "step": 26334
    },
    {
      "epoch": 1.474689214917684,
      "grad_norm": 1.2808122634887695,
      "learning_rate": 9.833605263157896e-05,
      "loss": 0.4479,
      "step": 26335
    },
    {
      "epoch": 1.474745212229813,
      "grad_norm": 1.3971952199935913,
      "learning_rate": 9.833578947368422e-05,
      "loss": 0.4375,
      "step": 26336
    },
    {
      "epoch": 1.474801209541942,
      "grad_norm": 1.2194265127182007,
      "learning_rate": 9.833552631578948e-05,
      "loss": 0.4314,
      "step": 26337
    },
    {
      "epoch": 1.474857206854071,
      "grad_norm": 1.1891316175460815,
      "learning_rate": 9.833526315789474e-05,
      "loss": 0.3928,
      "step": 26338
    },
    {
      "epoch": 1.4749132041662,
      "grad_norm": 1.1536589860916138,
      "learning_rate": 9.8335e-05,
      "loss": 0.4065,
      "step": 26339
    },
    {
      "epoch": 1.474969201478329,
      "grad_norm": 1.2504948377609253,
      "learning_rate": 9.833473684210527e-05,
      "loss": 0.4255,
      "step": 26340
    },
    {
      "epoch": 1.4750251987904581,
      "grad_norm": 1.4145692586898804,
      "learning_rate": 9.833447368421053e-05,
      "loss": 0.4366,
      "step": 26341
    },
    {
      "epoch": 1.4750811961025871,
      "grad_norm": 1.2258678674697876,
      "learning_rate": 9.833421052631579e-05,
      "loss": 0.5979,
      "step": 26342
    },
    {
      "epoch": 1.4751371934147162,
      "grad_norm": 1.4275203943252563,
      "learning_rate": 9.833394736842105e-05,
      "loss": 0.4396,
      "step": 26343
    },
    {
      "epoch": 1.4751931907268452,
      "grad_norm": 1.2004852294921875,
      "learning_rate": 9.833368421052633e-05,
      "loss": 0.3964,
      "step": 26344
    },
    {
      "epoch": 1.4752491880389742,
      "grad_norm": 1.274658441543579,
      "learning_rate": 9.833342105263159e-05,
      "loss": 0.4803,
      "step": 26345
    },
    {
      "epoch": 1.4753051853511032,
      "grad_norm": 1.2332416772842407,
      "learning_rate": 9.833315789473684e-05,
      "loss": 0.4954,
      "step": 26346
    },
    {
      "epoch": 1.4753611826632322,
      "grad_norm": 1.2086886167526245,
      "learning_rate": 9.83328947368421e-05,
      "loss": 0.4858,
      "step": 26347
    },
    {
      "epoch": 1.4754171799753613,
      "grad_norm": 1.24374520778656,
      "learning_rate": 9.833263157894738e-05,
      "loss": 0.4812,
      "step": 26348
    },
    {
      "epoch": 1.4754731772874903,
      "grad_norm": 1.462717890739441,
      "learning_rate": 9.833236842105264e-05,
      "loss": 0.4936,
      "step": 26349
    },
    {
      "epoch": 1.4755291745996193,
      "grad_norm": 1.3813000917434692,
      "learning_rate": 9.83321052631579e-05,
      "loss": 0.4134,
      "step": 26350
    },
    {
      "epoch": 1.4755851719117483,
      "grad_norm": 1.4574353694915771,
      "learning_rate": 9.833184210526316e-05,
      "loss": 0.5203,
      "step": 26351
    },
    {
      "epoch": 1.4756411692238773,
      "grad_norm": 11.261170387268066,
      "learning_rate": 9.833157894736843e-05,
      "loss": 0.5446,
      "step": 26352
    },
    {
      "epoch": 1.4756971665360061,
      "grad_norm": 1.5375767946243286,
      "learning_rate": 9.833131578947369e-05,
      "loss": 0.4074,
      "step": 26353
    },
    {
      "epoch": 1.4757531638481352,
      "grad_norm": 1.4761403799057007,
      "learning_rate": 9.833105263157896e-05,
      "loss": 0.4767,
      "step": 26354
    },
    {
      "epoch": 1.4758091611602642,
      "grad_norm": 1.3679126501083374,
      "learning_rate": 9.833078947368421e-05,
      "loss": 0.4278,
      "step": 26355
    },
    {
      "epoch": 1.4758651584723932,
      "grad_norm": 1.2875906229019165,
      "learning_rate": 9.833052631578947e-05,
      "loss": 0.4687,
      "step": 26356
    },
    {
      "epoch": 1.4759211557845222,
      "grad_norm": 1.0745713710784912,
      "learning_rate": 9.833026315789474e-05,
      "loss": 0.3685,
      "step": 26357
    },
    {
      "epoch": 1.4759771530966512,
      "grad_norm": 1.2158112525939941,
      "learning_rate": 9.833e-05,
      "loss": 0.4848,
      "step": 26358
    },
    {
      "epoch": 1.4760331504087802,
      "grad_norm": 1.1623754501342773,
      "learning_rate": 9.832973684210528e-05,
      "loss": 0.5242,
      "step": 26359
    },
    {
      "epoch": 1.4760891477209093,
      "grad_norm": 1.2012572288513184,
      "learning_rate": 9.832947368421052e-05,
      "loss": 0.4513,
      "step": 26360
    },
    {
      "epoch": 1.4761451450330383,
      "grad_norm": 1.6301683187484741,
      "learning_rate": 9.83292105263158e-05,
      "loss": 0.4982,
      "step": 26361
    },
    {
      "epoch": 1.4762011423451673,
      "grad_norm": 1.2660621404647827,
      "learning_rate": 9.832894736842105e-05,
      "loss": 0.5017,
      "step": 26362
    },
    {
      "epoch": 1.4762571396572963,
      "grad_norm": 1.179836630821228,
      "learning_rate": 9.832868421052633e-05,
      "loss": 0.4138,
      "step": 26363
    },
    {
      "epoch": 1.4763131369694253,
      "grad_norm": 1.2577470541000366,
      "learning_rate": 9.832842105263157e-05,
      "loss": 0.4766,
      "step": 26364
    },
    {
      "epoch": 1.4763691342815544,
      "grad_norm": 1.3299623727798462,
      "learning_rate": 9.832815789473685e-05,
      "loss": 0.5242,
      "step": 26365
    },
    {
      "epoch": 1.4764251315936834,
      "grad_norm": 1.5384547710418701,
      "learning_rate": 9.832789473684211e-05,
      "loss": 0.453,
      "step": 26366
    },
    {
      "epoch": 1.4764811289058124,
      "grad_norm": 1.502936840057373,
      "learning_rate": 9.832763157894738e-05,
      "loss": 0.4914,
      "step": 26367
    },
    {
      "epoch": 1.4765371262179414,
      "grad_norm": 1.1734130382537842,
      "learning_rate": 9.832736842105264e-05,
      "loss": 0.3136,
      "step": 26368
    },
    {
      "epoch": 1.4765931235300704,
      "grad_norm": 1.788120150566101,
      "learning_rate": 9.83271052631579e-05,
      "loss": 0.4253,
      "step": 26369
    },
    {
      "epoch": 1.4766491208421995,
      "grad_norm": 1.5197820663452148,
      "learning_rate": 9.832684210526316e-05,
      "loss": 0.5413,
      "step": 26370
    },
    {
      "epoch": 1.4767051181543285,
      "grad_norm": 2.077280282974243,
      "learning_rate": 9.832657894736843e-05,
      "loss": 0.5113,
      "step": 26371
    },
    {
      "epoch": 1.4767611154664575,
      "grad_norm": 1.2023744583129883,
      "learning_rate": 9.832631578947369e-05,
      "loss": 0.5016,
      "step": 26372
    },
    {
      "epoch": 1.4768171127785865,
      "grad_norm": 1.2389616966247559,
      "learning_rate": 9.832605263157895e-05,
      "loss": 0.4866,
      "step": 26373
    },
    {
      "epoch": 1.4768731100907155,
      "grad_norm": 1.177794337272644,
      "learning_rate": 9.832578947368421e-05,
      "loss": 0.4466,
      "step": 26374
    },
    {
      "epoch": 1.4769291074028446,
      "grad_norm": 1.428767442703247,
      "learning_rate": 9.832552631578947e-05,
      "loss": 0.4747,
      "step": 26375
    },
    {
      "epoch": 1.4769851047149736,
      "grad_norm": 1.4767810106277466,
      "learning_rate": 9.832526315789475e-05,
      "loss": 0.4311,
      "step": 26376
    },
    {
      "epoch": 1.4770411020271026,
      "grad_norm": 1.1158396005630493,
      "learning_rate": 9.8325e-05,
      "loss": 0.373,
      "step": 26377
    },
    {
      "epoch": 1.4770970993392316,
      "grad_norm": 1.7182050943374634,
      "learning_rate": 9.832473684210526e-05,
      "loss": 0.4972,
      "step": 26378
    },
    {
      "epoch": 1.4771530966513606,
      "grad_norm": 1.5233622789382935,
      "learning_rate": 9.832447368421052e-05,
      "loss": 0.4996,
      "step": 26379
    },
    {
      "epoch": 1.4772090939634897,
      "grad_norm": 1.4264438152313232,
      "learning_rate": 9.83242105263158e-05,
      "loss": 0.5008,
      "step": 26380
    },
    {
      "epoch": 1.4772650912756187,
      "grad_norm": 1.1224011182785034,
      "learning_rate": 9.832394736842106e-05,
      "loss": 0.3306,
      "step": 26381
    },
    {
      "epoch": 1.4773210885877477,
      "grad_norm": 1.9449200630187988,
      "learning_rate": 9.832368421052632e-05,
      "loss": 0.53,
      "step": 26382
    },
    {
      "epoch": 1.4773770858998767,
      "grad_norm": 1.3462685346603394,
      "learning_rate": 9.832342105263158e-05,
      "loss": 0.464,
      "step": 26383
    },
    {
      "epoch": 1.4774330832120057,
      "grad_norm": 1.302512526512146,
      "learning_rate": 9.832315789473685e-05,
      "loss": 0.3426,
      "step": 26384
    },
    {
      "epoch": 1.4774890805241347,
      "grad_norm": 1.703502893447876,
      "learning_rate": 9.832289473684211e-05,
      "loss": 0.5233,
      "step": 26385
    },
    {
      "epoch": 1.4775450778362638,
      "grad_norm": 1.4388452768325806,
      "learning_rate": 9.832263157894738e-05,
      "loss": 0.569,
      "step": 26386
    },
    {
      "epoch": 1.4776010751483928,
      "grad_norm": 1.9214638471603394,
      "learning_rate": 9.832236842105263e-05,
      "loss": 0.495,
      "step": 26387
    },
    {
      "epoch": 1.4776570724605218,
      "grad_norm": 1.1944797039031982,
      "learning_rate": 9.83221052631579e-05,
      "loss": 0.4455,
      "step": 26388
    },
    {
      "epoch": 1.4777130697726508,
      "grad_norm": 1.778193712234497,
      "learning_rate": 9.832184210526316e-05,
      "loss": 0.5706,
      "step": 26389
    },
    {
      "epoch": 1.4777690670847798,
      "grad_norm": 1.3621065616607666,
      "learning_rate": 9.832157894736844e-05,
      "loss": 0.5761,
      "step": 26390
    },
    {
      "epoch": 1.4778250643969089,
      "grad_norm": 1.1708863973617554,
      "learning_rate": 9.83213157894737e-05,
      "loss": 0.3399,
      "step": 26391
    },
    {
      "epoch": 1.4778810617090379,
      "grad_norm": 1.17164146900177,
      "learning_rate": 9.832105263157894e-05,
      "loss": 0.326,
      "step": 26392
    },
    {
      "epoch": 1.477937059021167,
      "grad_norm": 1.2393851280212402,
      "learning_rate": 9.832078947368421e-05,
      "loss": 0.4197,
      "step": 26393
    },
    {
      "epoch": 1.477993056333296,
      "grad_norm": 1.4932615756988525,
      "learning_rate": 9.832052631578947e-05,
      "loss": 0.5323,
      "step": 26394
    },
    {
      "epoch": 1.478049053645425,
      "grad_norm": 1.2480852603912354,
      "learning_rate": 9.832026315789475e-05,
      "loss": 0.461,
      "step": 26395
    },
    {
      "epoch": 1.478105050957554,
      "grad_norm": 1.423901081085205,
      "learning_rate": 9.832000000000001e-05,
      "loss": 0.3795,
      "step": 26396
    },
    {
      "epoch": 1.478161048269683,
      "grad_norm": 1.1684621572494507,
      "learning_rate": 9.831973684210527e-05,
      "loss": 0.4044,
      "step": 26397
    },
    {
      "epoch": 1.478217045581812,
      "grad_norm": 1.1378426551818848,
      "learning_rate": 9.831947368421053e-05,
      "loss": 0.4358,
      "step": 26398
    },
    {
      "epoch": 1.478273042893941,
      "grad_norm": 1.8066250085830688,
      "learning_rate": 9.83192105263158e-05,
      "loss": 0.6716,
      "step": 26399
    },
    {
      "epoch": 1.47832904020607,
      "grad_norm": 1.414628505706787,
      "learning_rate": 9.831894736842106e-05,
      "loss": 0.4991,
      "step": 26400
    },
    {
      "epoch": 1.478385037518199,
      "grad_norm": 1.4492090940475464,
      "learning_rate": 9.831868421052632e-05,
      "loss": 0.4438,
      "step": 26401
    },
    {
      "epoch": 1.478441034830328,
      "grad_norm": 1.5771111249923706,
      "learning_rate": 9.831842105263158e-05,
      "loss": 0.4732,
      "step": 26402
    },
    {
      "epoch": 1.478497032142457,
      "grad_norm": 1.8152694702148438,
      "learning_rate": 9.831815789473685e-05,
      "loss": 0.4741,
      "step": 26403
    },
    {
      "epoch": 1.478553029454586,
      "grad_norm": 1.3209187984466553,
      "learning_rate": 9.831789473684211e-05,
      "loss": 0.3829,
      "step": 26404
    },
    {
      "epoch": 1.4786090267667151,
      "grad_norm": 1.21585214138031,
      "learning_rate": 9.831763157894737e-05,
      "loss": 0.4164,
      "step": 26405
    },
    {
      "epoch": 1.4786650240788441,
      "grad_norm": 1.0698697566986084,
      "learning_rate": 9.831736842105263e-05,
      "loss": 0.349,
      "step": 26406
    },
    {
      "epoch": 1.4787210213909732,
      "grad_norm": 1.439136266708374,
      "learning_rate": 9.83171052631579e-05,
      "loss": 0.3543,
      "step": 26407
    },
    {
      "epoch": 1.4787770187031022,
      "grad_norm": 1.2398730516433716,
      "learning_rate": 9.831684210526316e-05,
      "loss": 0.5895,
      "step": 26408
    },
    {
      "epoch": 1.4788330160152312,
      "grad_norm": 1.3039520978927612,
      "learning_rate": 9.831657894736842e-05,
      "loss": 0.5787,
      "step": 26409
    },
    {
      "epoch": 1.4788890133273602,
      "grad_norm": 1.1072537899017334,
      "learning_rate": 9.831631578947368e-05,
      "loss": 0.4565,
      "step": 26410
    },
    {
      "epoch": 1.4789450106394892,
      "grad_norm": 1.2011222839355469,
      "learning_rate": 9.831605263157894e-05,
      "loss": 0.4562,
      "step": 26411
    },
    {
      "epoch": 1.4790010079516183,
      "grad_norm": 1.3780157566070557,
      "learning_rate": 9.831578947368422e-05,
      "loss": 0.4488,
      "step": 26412
    },
    {
      "epoch": 1.4790570052637473,
      "grad_norm": 1.3226121664047241,
      "learning_rate": 9.831552631578948e-05,
      "loss": 0.5458,
      "step": 26413
    },
    {
      "epoch": 1.4791130025758763,
      "grad_norm": 1.4769147634506226,
      "learning_rate": 9.831526315789475e-05,
      "loss": 0.5004,
      "step": 26414
    },
    {
      "epoch": 1.4791689998880053,
      "grad_norm": 1.467332363128662,
      "learning_rate": 9.8315e-05,
      "loss": 0.6603,
      "step": 26415
    },
    {
      "epoch": 1.4792249972001343,
      "grad_norm": 1.6566319465637207,
      "learning_rate": 9.831473684210527e-05,
      "loss": 0.5555,
      "step": 26416
    },
    {
      "epoch": 1.4792809945122634,
      "grad_norm": 1.2249643802642822,
      "learning_rate": 9.831447368421053e-05,
      "loss": 0.4686,
      "step": 26417
    },
    {
      "epoch": 1.4793369918243924,
      "grad_norm": 1.3104848861694336,
      "learning_rate": 9.83142105263158e-05,
      "loss": 0.4912,
      "step": 26418
    },
    {
      "epoch": 1.4793929891365214,
      "grad_norm": 1.394104242324829,
      "learning_rate": 9.831394736842105e-05,
      "loss": 0.5911,
      "step": 26419
    },
    {
      "epoch": 1.4794489864486504,
      "grad_norm": 1.2208060026168823,
      "learning_rate": 9.831368421052632e-05,
      "loss": 0.3924,
      "step": 26420
    },
    {
      "epoch": 1.4795049837607794,
      "grad_norm": 1.481011986732483,
      "learning_rate": 9.831342105263158e-05,
      "loss": 0.4138,
      "step": 26421
    },
    {
      "epoch": 1.4795609810729085,
      "grad_norm": 1.4936903715133667,
      "learning_rate": 9.831315789473686e-05,
      "loss": 0.6404,
      "step": 26422
    },
    {
      "epoch": 1.4796169783850375,
      "grad_norm": 1.253900408744812,
      "learning_rate": 9.831289473684211e-05,
      "loss": 0.4427,
      "step": 26423
    },
    {
      "epoch": 1.4796729756971665,
      "grad_norm": 1.721214771270752,
      "learning_rate": 9.831263157894737e-05,
      "loss": 0.4268,
      "step": 26424
    },
    {
      "epoch": 1.4797289730092955,
      "grad_norm": 1.1525635719299316,
      "learning_rate": 9.831236842105263e-05,
      "loss": 0.4156,
      "step": 26425
    },
    {
      "epoch": 1.4797849703214245,
      "grad_norm": 1.419958233833313,
      "learning_rate": 9.83121052631579e-05,
      "loss": 0.5837,
      "step": 26426
    },
    {
      "epoch": 1.4798409676335536,
      "grad_norm": 1.310727596282959,
      "learning_rate": 9.831184210526317e-05,
      "loss": 0.4129,
      "step": 26427
    },
    {
      "epoch": 1.4798969649456826,
      "grad_norm": 1.3370429277420044,
      "learning_rate": 9.831157894736843e-05,
      "loss": 0.3879,
      "step": 26428
    },
    {
      "epoch": 1.4799529622578116,
      "grad_norm": 1.1163039207458496,
      "learning_rate": 9.831131578947369e-05,
      "loss": 0.377,
      "step": 26429
    },
    {
      "epoch": 1.4800089595699406,
      "grad_norm": 1.5141627788543701,
      "learning_rate": 9.831105263157895e-05,
      "loss": 0.7203,
      "step": 26430
    },
    {
      "epoch": 1.4800649568820696,
      "grad_norm": 1.1618309020996094,
      "learning_rate": 9.831078947368422e-05,
      "loss": 0.4064,
      "step": 26431
    },
    {
      "epoch": 1.4801209541941986,
      "grad_norm": 1.1684426069259644,
      "learning_rate": 9.831052631578948e-05,
      "loss": 0.4171,
      "step": 26432
    },
    {
      "epoch": 1.4801769515063277,
      "grad_norm": 1.3729358911514282,
      "learning_rate": 9.831026315789474e-05,
      "loss": 0.4878,
      "step": 26433
    },
    {
      "epoch": 1.4802329488184567,
      "grad_norm": 1.247262716293335,
      "learning_rate": 9.831e-05,
      "loss": 0.4128,
      "step": 26434
    },
    {
      "epoch": 1.4802889461305857,
      "grad_norm": 1.4308249950408936,
      "learning_rate": 9.830973684210527e-05,
      "loss": 0.5528,
      "step": 26435
    },
    {
      "epoch": 1.4803449434427147,
      "grad_norm": 1.364248275756836,
      "learning_rate": 9.830947368421053e-05,
      "loss": 0.4859,
      "step": 26436
    },
    {
      "epoch": 1.4804009407548437,
      "grad_norm": 1.3801158666610718,
      "learning_rate": 9.830921052631579e-05,
      "loss": 0.4584,
      "step": 26437
    },
    {
      "epoch": 1.4804569380669728,
      "grad_norm": 1.2886649370193481,
      "learning_rate": 9.830894736842105e-05,
      "loss": 0.5631,
      "step": 26438
    },
    {
      "epoch": 1.4805129353791018,
      "grad_norm": 1.2724788188934326,
      "learning_rate": 9.830868421052632e-05,
      "loss": 0.449,
      "step": 26439
    },
    {
      "epoch": 1.4805689326912308,
      "grad_norm": 1.3946900367736816,
      "learning_rate": 9.830842105263158e-05,
      "loss": 0.3947,
      "step": 26440
    },
    {
      "epoch": 1.4806249300033598,
      "grad_norm": 1.218000888824463,
      "learning_rate": 9.830815789473686e-05,
      "loss": 0.3982,
      "step": 26441
    },
    {
      "epoch": 1.4806809273154888,
      "grad_norm": 1.2052074670791626,
      "learning_rate": 9.83078947368421e-05,
      "loss": 0.4831,
      "step": 26442
    },
    {
      "epoch": 1.4807369246276179,
      "grad_norm": 1.4074491262435913,
      "learning_rate": 9.830763157894736e-05,
      "loss": 0.4657,
      "step": 26443
    },
    {
      "epoch": 1.4807929219397469,
      "grad_norm": 1.4324088096618652,
      "learning_rate": 9.830736842105264e-05,
      "loss": 0.4509,
      "step": 26444
    },
    {
      "epoch": 1.480848919251876,
      "grad_norm": 1.564112901687622,
      "learning_rate": 9.83071052631579e-05,
      "loss": 0.5721,
      "step": 26445
    },
    {
      "epoch": 1.480904916564005,
      "grad_norm": 1.3879790306091309,
      "learning_rate": 9.830684210526317e-05,
      "loss": 0.4773,
      "step": 26446
    },
    {
      "epoch": 1.480960913876134,
      "grad_norm": 1.097662329673767,
      "learning_rate": 9.830657894736842e-05,
      "loss": 0.5217,
      "step": 26447
    },
    {
      "epoch": 1.481016911188263,
      "grad_norm": 1.2054959535598755,
      "learning_rate": 9.830631578947369e-05,
      "loss": 0.4029,
      "step": 26448
    },
    {
      "epoch": 1.481072908500392,
      "grad_norm": 1.091911792755127,
      "learning_rate": 9.830605263157895e-05,
      "loss": 0.4091,
      "step": 26449
    },
    {
      "epoch": 1.481128905812521,
      "grad_norm": 1.366909146308899,
      "learning_rate": 9.830578947368422e-05,
      "loss": 0.5705,
      "step": 26450
    },
    {
      "epoch": 1.48118490312465,
      "grad_norm": 1.342989206314087,
      "learning_rate": 9.830552631578948e-05,
      "loss": 0.5374,
      "step": 26451
    },
    {
      "epoch": 1.481240900436779,
      "grad_norm": 1.3994578123092651,
      "learning_rate": 9.830526315789474e-05,
      "loss": 0.3981,
      "step": 26452
    },
    {
      "epoch": 1.481296897748908,
      "grad_norm": 1.430092215538025,
      "learning_rate": 9.8305e-05,
      "loss": 0.4674,
      "step": 26453
    },
    {
      "epoch": 1.481352895061037,
      "grad_norm": 1.160904884338379,
      "learning_rate": 9.830473684210527e-05,
      "loss": 0.3719,
      "step": 26454
    },
    {
      "epoch": 1.481408892373166,
      "grad_norm": 1.5587517023086548,
      "learning_rate": 9.830447368421053e-05,
      "loss": 0.5502,
      "step": 26455
    },
    {
      "epoch": 1.481464889685295,
      "grad_norm": 1.2606472969055176,
      "learning_rate": 9.83042105263158e-05,
      "loss": 0.4605,
      "step": 26456
    },
    {
      "epoch": 1.4815208869974241,
      "grad_norm": 1.5438977479934692,
      "learning_rate": 9.830394736842105e-05,
      "loss": 0.8613,
      "step": 26457
    },
    {
      "epoch": 1.4815768843095531,
      "grad_norm": 1.6708910465240479,
      "learning_rate": 9.830368421052633e-05,
      "loss": 0.5333,
      "step": 26458
    },
    {
      "epoch": 1.4816328816216822,
      "grad_norm": 1.3151296377182007,
      "learning_rate": 9.830342105263159e-05,
      "loss": 0.3697,
      "step": 26459
    },
    {
      "epoch": 1.4816888789338112,
      "grad_norm": 1.283218502998352,
      "learning_rate": 9.830315789473685e-05,
      "loss": 0.4876,
      "step": 26460
    },
    {
      "epoch": 1.4817448762459402,
      "grad_norm": 1.3207824230194092,
      "learning_rate": 9.83028947368421e-05,
      "loss": 0.4758,
      "step": 26461
    },
    {
      "epoch": 1.4818008735580692,
      "grad_norm": 1.426101803779602,
      "learning_rate": 9.830263157894737e-05,
      "loss": 0.5537,
      "step": 26462
    },
    {
      "epoch": 1.4818568708701982,
      "grad_norm": 1.3909064531326294,
      "learning_rate": 9.830236842105264e-05,
      "loss": 0.3975,
      "step": 26463
    },
    {
      "epoch": 1.4819128681823273,
      "grad_norm": 1.3793268203735352,
      "learning_rate": 9.83021052631579e-05,
      "loss": 0.5087,
      "step": 26464
    },
    {
      "epoch": 1.4819688654944563,
      "grad_norm": 1.632776141166687,
      "learning_rate": 9.830184210526316e-05,
      "loss": 0.4604,
      "step": 26465
    },
    {
      "epoch": 1.4820248628065853,
      "grad_norm": 1.5219885110855103,
      "learning_rate": 9.830157894736842e-05,
      "loss": 0.6344,
      "step": 26466
    },
    {
      "epoch": 1.4820808601187143,
      "grad_norm": 1.3433325290679932,
      "learning_rate": 9.830131578947369e-05,
      "loss": 0.4594,
      "step": 26467
    },
    {
      "epoch": 1.4821368574308433,
      "grad_norm": 1.540515661239624,
      "learning_rate": 9.830105263157895e-05,
      "loss": 0.3933,
      "step": 26468
    },
    {
      "epoch": 1.4821928547429724,
      "grad_norm": 1.087449312210083,
      "learning_rate": 9.830078947368422e-05,
      "loss": 0.4609,
      "step": 26469
    },
    {
      "epoch": 1.4822488520551014,
      "grad_norm": 1.4942513704299927,
      "learning_rate": 9.830052631578947e-05,
      "loss": 0.5734,
      "step": 26470
    },
    {
      "epoch": 1.4823048493672304,
      "grad_norm": 1.2333168983459473,
      "learning_rate": 9.830026315789474e-05,
      "loss": 0.3683,
      "step": 26471
    },
    {
      "epoch": 1.4823608466793594,
      "grad_norm": 1.3990916013717651,
      "learning_rate": 9.83e-05,
      "loss": 0.4677,
      "step": 26472
    },
    {
      "epoch": 1.4824168439914884,
      "grad_norm": 1.342435359954834,
      "learning_rate": 9.829973684210528e-05,
      "loss": 0.4344,
      "step": 26473
    },
    {
      "epoch": 1.4824728413036175,
      "grad_norm": 10.466675758361816,
      "learning_rate": 9.829947368421052e-05,
      "loss": 0.5972,
      "step": 26474
    },
    {
      "epoch": 1.4825288386157465,
      "grad_norm": 1.4302014112472534,
      "learning_rate": 9.82992105263158e-05,
      "loss": 0.5685,
      "step": 26475
    },
    {
      "epoch": 1.4825848359278755,
      "grad_norm": 1.4066733121871948,
      "learning_rate": 9.829894736842106e-05,
      "loss": 0.5278,
      "step": 26476
    },
    {
      "epoch": 1.4826408332400045,
      "grad_norm": 2.721774101257324,
      "learning_rate": 9.829868421052632e-05,
      "loss": 0.4604,
      "step": 26477
    },
    {
      "epoch": 1.4826968305521335,
      "grad_norm": 1.548677921295166,
      "learning_rate": 9.829842105263159e-05,
      "loss": 0.4053,
      "step": 26478
    },
    {
      "epoch": 1.4827528278642625,
      "grad_norm": 1.1114532947540283,
      "learning_rate": 9.829815789473684e-05,
      "loss": 0.5194,
      "step": 26479
    },
    {
      "epoch": 1.4828088251763916,
      "grad_norm": 1.2839246988296509,
      "learning_rate": 9.829789473684211e-05,
      "loss": 0.4681,
      "step": 26480
    },
    {
      "epoch": 1.4828648224885206,
      "grad_norm": 1.4902235269546509,
      "learning_rate": 9.829763157894737e-05,
      "loss": 0.4127,
      "step": 26481
    },
    {
      "epoch": 1.4829208198006496,
      "grad_norm": 1.416769027709961,
      "learning_rate": 9.829736842105264e-05,
      "loss": 0.5673,
      "step": 26482
    },
    {
      "epoch": 1.4829768171127786,
      "grad_norm": 1.4382866621017456,
      "learning_rate": 9.82971052631579e-05,
      "loss": 0.4693,
      "step": 26483
    },
    {
      "epoch": 1.4830328144249076,
      "grad_norm": 1.2511255741119385,
      "learning_rate": 9.829684210526316e-05,
      "loss": 0.439,
      "step": 26484
    },
    {
      "epoch": 1.4830888117370367,
      "grad_norm": 1.4404335021972656,
      "learning_rate": 9.829657894736842e-05,
      "loss": 0.5361,
      "step": 26485
    },
    {
      "epoch": 1.4831448090491657,
      "grad_norm": 1.4500209093093872,
      "learning_rate": 9.82963157894737e-05,
      "loss": 0.4741,
      "step": 26486
    },
    {
      "epoch": 1.4832008063612947,
      "grad_norm": 1.334502100944519,
      "learning_rate": 9.829605263157895e-05,
      "loss": 0.5394,
      "step": 26487
    },
    {
      "epoch": 1.4832568036734237,
      "grad_norm": 1.18729567527771,
      "learning_rate": 9.829578947368421e-05,
      "loss": 0.3905,
      "step": 26488
    },
    {
      "epoch": 1.4833128009855527,
      "grad_norm": 1.1093251705169678,
      "learning_rate": 9.829552631578947e-05,
      "loss": 0.3999,
      "step": 26489
    },
    {
      "epoch": 1.4833687982976818,
      "grad_norm": 1.262854814529419,
      "learning_rate": 9.829526315789475e-05,
      "loss": 0.4139,
      "step": 26490
    },
    {
      "epoch": 1.4834247956098108,
      "grad_norm": 1.491692066192627,
      "learning_rate": 9.8295e-05,
      "loss": 0.5313,
      "step": 26491
    },
    {
      "epoch": 1.4834807929219398,
      "grad_norm": 1.256239652633667,
      "learning_rate": 9.829473684210527e-05,
      "loss": 0.42,
      "step": 26492
    },
    {
      "epoch": 1.4835367902340688,
      "grad_norm": 1.0862276554107666,
      "learning_rate": 9.829447368421053e-05,
      "loss": 0.3264,
      "step": 26493
    },
    {
      "epoch": 1.4835927875461978,
      "grad_norm": 1.4838993549346924,
      "learning_rate": 9.829421052631579e-05,
      "loss": 0.395,
      "step": 26494
    },
    {
      "epoch": 1.4836487848583269,
      "grad_norm": 1.4637272357940674,
      "learning_rate": 9.829394736842106e-05,
      "loss": 0.4645,
      "step": 26495
    },
    {
      "epoch": 1.4837047821704559,
      "grad_norm": 1.218091368675232,
      "learning_rate": 9.829368421052632e-05,
      "loss": 0.5034,
      "step": 26496
    },
    {
      "epoch": 1.483760779482585,
      "grad_norm": 1.3952760696411133,
      "learning_rate": 9.829342105263158e-05,
      "loss": 0.3725,
      "step": 26497
    },
    {
      "epoch": 1.483816776794714,
      "grad_norm": 1.4923816919326782,
      "learning_rate": 9.829315789473684e-05,
      "loss": 0.4811,
      "step": 26498
    },
    {
      "epoch": 1.483872774106843,
      "grad_norm": 1.4141688346862793,
      "learning_rate": 9.829289473684211e-05,
      "loss": 0.6126,
      "step": 26499
    },
    {
      "epoch": 1.483928771418972,
      "grad_norm": 1.403441071510315,
      "learning_rate": 9.829263157894737e-05,
      "loss": 0.4566,
      "step": 26500
    },
    {
      "epoch": 1.483984768731101,
      "grad_norm": 1.1639097929000854,
      "learning_rate": 9.829236842105264e-05,
      "loss": 0.4863,
      "step": 26501
    },
    {
      "epoch": 1.48404076604323,
      "grad_norm": 1.2145073413848877,
      "learning_rate": 9.829210526315789e-05,
      "loss": 0.4488,
      "step": 26502
    },
    {
      "epoch": 1.484096763355359,
      "grad_norm": 1.349830150604248,
      "learning_rate": 9.829184210526316e-05,
      "loss": 0.4137,
      "step": 26503
    },
    {
      "epoch": 1.484152760667488,
      "grad_norm": 1.2369256019592285,
      "learning_rate": 9.829157894736842e-05,
      "loss": 0.3126,
      "step": 26504
    },
    {
      "epoch": 1.484208757979617,
      "grad_norm": 1.1879065036773682,
      "learning_rate": 9.82913157894737e-05,
      "loss": 0.4301,
      "step": 26505
    },
    {
      "epoch": 1.484264755291746,
      "grad_norm": 1.506127119064331,
      "learning_rate": 9.829105263157896e-05,
      "loss": 0.7175,
      "step": 26506
    },
    {
      "epoch": 1.484320752603875,
      "grad_norm": 1.4014273881912231,
      "learning_rate": 9.829078947368422e-05,
      "loss": 0.4257,
      "step": 26507
    },
    {
      "epoch": 1.484376749916004,
      "grad_norm": 1.3636915683746338,
      "learning_rate": 9.829052631578948e-05,
      "loss": 0.4488,
      "step": 26508
    },
    {
      "epoch": 1.4844327472281331,
      "grad_norm": 1.363595962524414,
      "learning_rate": 9.829026315789475e-05,
      "loss": 0.3705,
      "step": 26509
    },
    {
      "epoch": 1.4844887445402621,
      "grad_norm": 1.5012990236282349,
      "learning_rate": 9.829000000000001e-05,
      "loss": 0.412,
      "step": 26510
    },
    {
      "epoch": 1.4845447418523912,
      "grad_norm": 1.3049137592315674,
      "learning_rate": 9.828973684210527e-05,
      "loss": 0.4308,
      "step": 26511
    },
    {
      "epoch": 1.4846007391645202,
      "grad_norm": 1.3145856857299805,
      "learning_rate": 9.828947368421053e-05,
      "loss": 0.3557,
      "step": 26512
    },
    {
      "epoch": 1.4846567364766492,
      "grad_norm": 1.1711186170578003,
      "learning_rate": 9.828921052631579e-05,
      "loss": 0.4307,
      "step": 26513
    },
    {
      "epoch": 1.4847127337887782,
      "grad_norm": 1.9552361965179443,
      "learning_rate": 9.828894736842106e-05,
      "loss": 0.3842,
      "step": 26514
    },
    {
      "epoch": 1.4847687311009072,
      "grad_norm": 1.2192083597183228,
      "learning_rate": 9.828868421052632e-05,
      "loss": 0.4244,
      "step": 26515
    },
    {
      "epoch": 1.4848247284130363,
      "grad_norm": 1.537446141242981,
      "learning_rate": 9.828842105263158e-05,
      "loss": 0.4177,
      "step": 26516
    },
    {
      "epoch": 1.4848807257251653,
      "grad_norm": 1.4063435792922974,
      "learning_rate": 9.828815789473684e-05,
      "loss": 0.537,
      "step": 26517
    },
    {
      "epoch": 1.4849367230372943,
      "grad_norm": 1.3753093481063843,
      "learning_rate": 9.828789473684211e-05,
      "loss": 0.4318,
      "step": 26518
    },
    {
      "epoch": 1.4849927203494233,
      "grad_norm": 1.205532193183899,
      "learning_rate": 9.828763157894737e-05,
      "loss": 0.4066,
      "step": 26519
    },
    {
      "epoch": 1.4850487176615523,
      "grad_norm": 1.2799038887023926,
      "learning_rate": 9.828736842105263e-05,
      "loss": 0.4152,
      "step": 26520
    },
    {
      "epoch": 1.4851047149736814,
      "grad_norm": 1.3891645669937134,
      "learning_rate": 9.82871052631579e-05,
      "loss": 0.4474,
      "step": 26521
    },
    {
      "epoch": 1.4851607122858104,
      "grad_norm": 1.5356215238571167,
      "learning_rate": 9.828684210526317e-05,
      "loss": 0.5309,
      "step": 26522
    },
    {
      "epoch": 1.4852167095979394,
      "grad_norm": 1.4802745580673218,
      "learning_rate": 9.828657894736843e-05,
      "loss": 0.5008,
      "step": 26523
    },
    {
      "epoch": 1.4852727069100684,
      "grad_norm": 1.4387753009796143,
      "learning_rate": 9.82863157894737e-05,
      "loss": 0.5275,
      "step": 26524
    },
    {
      "epoch": 1.4853287042221974,
      "grad_norm": 1.2407060861587524,
      "learning_rate": 9.828605263157895e-05,
      "loss": 0.4245,
      "step": 26525
    },
    {
      "epoch": 1.4853847015343264,
      "grad_norm": 1.364608883857727,
      "learning_rate": 9.828578947368422e-05,
      "loss": 0.4472,
      "step": 26526
    },
    {
      "epoch": 1.4854406988464555,
      "grad_norm": 1.448456883430481,
      "learning_rate": 9.828552631578948e-05,
      "loss": 0.5022,
      "step": 26527
    },
    {
      "epoch": 1.4854966961585845,
      "grad_norm": 1.405077576637268,
      "learning_rate": 9.828526315789475e-05,
      "loss": 0.401,
      "step": 26528
    },
    {
      "epoch": 1.4855526934707135,
      "grad_norm": 1.231691598892212,
      "learning_rate": 9.8285e-05,
      "loss": 0.3612,
      "step": 26529
    },
    {
      "epoch": 1.4856086907828425,
      "grad_norm": 1.1891672611236572,
      "learning_rate": 9.828473684210526e-05,
      "loss": 0.5706,
      "step": 26530
    },
    {
      "epoch": 1.4856646880949715,
      "grad_norm": 1.8921364545822144,
      "learning_rate": 9.828447368421053e-05,
      "loss": 0.5376,
      "step": 26531
    },
    {
      "epoch": 1.4857206854071006,
      "grad_norm": 1.308678388595581,
      "learning_rate": 9.828421052631579e-05,
      "loss": 0.4693,
      "step": 26532
    },
    {
      "epoch": 1.4857766827192296,
      "grad_norm": 1.3780572414398193,
      "learning_rate": 9.828394736842106e-05,
      "loss": 0.4731,
      "step": 26533
    },
    {
      "epoch": 1.4858326800313586,
      "grad_norm": 1.3991550207138062,
      "learning_rate": 9.828368421052631e-05,
      "loss": 0.4826,
      "step": 26534
    },
    {
      "epoch": 1.4858886773434876,
      "grad_norm": 6.05579948425293,
      "learning_rate": 9.828342105263158e-05,
      "loss": 0.5507,
      "step": 26535
    },
    {
      "epoch": 1.4859446746556166,
      "grad_norm": 1.2682628631591797,
      "learning_rate": 9.828315789473684e-05,
      "loss": 0.4841,
      "step": 26536
    },
    {
      "epoch": 1.4860006719677457,
      "grad_norm": 1.2944921255111694,
      "learning_rate": 9.828289473684212e-05,
      "loss": 0.4768,
      "step": 26537
    },
    {
      "epoch": 1.4860566692798747,
      "grad_norm": 1.390641689300537,
      "learning_rate": 9.828263157894738e-05,
      "loss": 0.532,
      "step": 26538
    },
    {
      "epoch": 1.4861126665920037,
      "grad_norm": 1.2902518510818481,
      "learning_rate": 9.828236842105264e-05,
      "loss": 0.4296,
      "step": 26539
    },
    {
      "epoch": 1.4861686639041327,
      "grad_norm": 1.514411211013794,
      "learning_rate": 9.82821052631579e-05,
      "loss": 0.4976,
      "step": 26540
    },
    {
      "epoch": 1.4862246612162617,
      "grad_norm": 1.4082057476043701,
      "learning_rate": 9.828184210526317e-05,
      "loss": 0.4531,
      "step": 26541
    },
    {
      "epoch": 1.4862806585283908,
      "grad_norm": 1.2194361686706543,
      "learning_rate": 9.828157894736843e-05,
      "loss": 0.4623,
      "step": 26542
    },
    {
      "epoch": 1.4863366558405198,
      "grad_norm": 1.5673096179962158,
      "learning_rate": 9.828131578947369e-05,
      "loss": 0.4011,
      "step": 26543
    },
    {
      "epoch": 1.4863926531526488,
      "grad_norm": 1.132565975189209,
      "learning_rate": 9.828105263157895e-05,
      "loss": 0.3721,
      "step": 26544
    },
    {
      "epoch": 1.4864486504647778,
      "grad_norm": 1.6840846538543701,
      "learning_rate": 9.828078947368422e-05,
      "loss": 0.5935,
      "step": 26545
    },
    {
      "epoch": 1.4865046477769068,
      "grad_norm": 1.255689263343811,
      "learning_rate": 9.828052631578948e-05,
      "loss": 0.4315,
      "step": 26546
    },
    {
      "epoch": 1.4865606450890358,
      "grad_norm": 1.2471442222595215,
      "learning_rate": 9.828026315789474e-05,
      "loss": 0.4162,
      "step": 26547
    },
    {
      "epoch": 1.4866166424011649,
      "grad_norm": 1.4482321739196777,
      "learning_rate": 9.828e-05,
      "loss": 0.4751,
      "step": 26548
    },
    {
      "epoch": 1.4866726397132939,
      "grad_norm": 1.298810601234436,
      "learning_rate": 9.827973684210526e-05,
      "loss": 0.5159,
      "step": 26549
    },
    {
      "epoch": 1.486728637025423,
      "grad_norm": 1.200095295906067,
      "learning_rate": 9.827947368421053e-05,
      "loss": 0.4506,
      "step": 26550
    },
    {
      "epoch": 1.486784634337552,
      "grad_norm": 1.4362943172454834,
      "learning_rate": 9.82792105263158e-05,
      "loss": 0.475,
      "step": 26551
    },
    {
      "epoch": 1.486840631649681,
      "grad_norm": 1.1563717126846313,
      "learning_rate": 9.827894736842105e-05,
      "loss": 0.3387,
      "step": 26552
    },
    {
      "epoch": 1.48689662896181,
      "grad_norm": 1.1349667310714722,
      "learning_rate": 9.827868421052631e-05,
      "loss": 0.3534,
      "step": 26553
    },
    {
      "epoch": 1.486952626273939,
      "grad_norm": 1.1395677328109741,
      "learning_rate": 9.827842105263159e-05,
      "loss": 0.4657,
      "step": 26554
    },
    {
      "epoch": 1.487008623586068,
      "grad_norm": 1.8083597421646118,
      "learning_rate": 9.827815789473685e-05,
      "loss": 0.6034,
      "step": 26555
    },
    {
      "epoch": 1.487064620898197,
      "grad_norm": 1.188975214958191,
      "learning_rate": 9.827789473684212e-05,
      "loss": 0.4013,
      "step": 26556
    },
    {
      "epoch": 1.487120618210326,
      "grad_norm": 1.316445231437683,
      "learning_rate": 9.827763157894737e-05,
      "loss": 0.5189,
      "step": 26557
    },
    {
      "epoch": 1.487176615522455,
      "grad_norm": 1.5387048721313477,
      "learning_rate": 9.827736842105264e-05,
      "loss": 0.5348,
      "step": 26558
    },
    {
      "epoch": 1.487232612834584,
      "grad_norm": 1.3718448877334595,
      "learning_rate": 9.82771052631579e-05,
      "loss": 0.5667,
      "step": 26559
    },
    {
      "epoch": 1.4872886101467129,
      "grad_norm": 5.825066566467285,
      "learning_rate": 9.827684210526317e-05,
      "loss": 0.3749,
      "step": 26560
    },
    {
      "epoch": 1.487344607458842,
      "grad_norm": 1.566150426864624,
      "learning_rate": 9.827657894736843e-05,
      "loss": 0.403,
      "step": 26561
    },
    {
      "epoch": 1.487400604770971,
      "grad_norm": 1.1900097131729126,
      "learning_rate": 9.827631578947369e-05,
      "loss": 0.4285,
      "step": 26562
    },
    {
      "epoch": 1.4874566020831,
      "grad_norm": 1.266889214515686,
      "learning_rate": 9.827605263157895e-05,
      "loss": 0.4519,
      "step": 26563
    },
    {
      "epoch": 1.487512599395229,
      "grad_norm": 1.1044726371765137,
      "learning_rate": 9.827578947368421e-05,
      "loss": 0.4364,
      "step": 26564
    },
    {
      "epoch": 1.487568596707358,
      "grad_norm": 1.3785444498062134,
      "learning_rate": 9.827552631578948e-05,
      "loss": 0.4805,
      "step": 26565
    },
    {
      "epoch": 1.487624594019487,
      "grad_norm": 1.5386810302734375,
      "learning_rate": 9.827526315789473e-05,
      "loss": 0.5183,
      "step": 26566
    },
    {
      "epoch": 1.487680591331616,
      "grad_norm": 1.3156059980392456,
      "learning_rate": 9.8275e-05,
      "loss": 0.4641,
      "step": 26567
    },
    {
      "epoch": 1.487736588643745,
      "grad_norm": 1.3793963193893433,
      "learning_rate": 9.827473684210526e-05,
      "loss": 0.5021,
      "step": 26568
    },
    {
      "epoch": 1.487792585955874,
      "grad_norm": 1.1304057836532593,
      "learning_rate": 9.827447368421054e-05,
      "loss": 0.4386,
      "step": 26569
    },
    {
      "epoch": 1.487848583268003,
      "grad_norm": 1.4514892101287842,
      "learning_rate": 9.82742105263158e-05,
      "loss": 0.5732,
      "step": 26570
    },
    {
      "epoch": 1.487904580580132,
      "grad_norm": 1.2929202318191528,
      "learning_rate": 9.827394736842106e-05,
      "loss": 0.3627,
      "step": 26571
    },
    {
      "epoch": 1.487960577892261,
      "grad_norm": 1.4279794692993164,
      "learning_rate": 9.827368421052632e-05,
      "loss": 0.4568,
      "step": 26572
    },
    {
      "epoch": 1.4880165752043901,
      "grad_norm": 1.3467376232147217,
      "learning_rate": 9.827342105263159e-05,
      "loss": 0.4781,
      "step": 26573
    },
    {
      "epoch": 1.4880725725165191,
      "grad_norm": 1.162685751914978,
      "learning_rate": 9.827315789473685e-05,
      "loss": 0.373,
      "step": 26574
    },
    {
      "epoch": 1.4881285698286482,
      "grad_norm": 1.4191398620605469,
      "learning_rate": 9.827289473684211e-05,
      "loss": 0.6028,
      "step": 26575
    },
    {
      "epoch": 1.4881845671407772,
      "grad_norm": 1.132226586341858,
      "learning_rate": 9.827263157894737e-05,
      "loss": 0.3635,
      "step": 26576
    },
    {
      "epoch": 1.4882405644529062,
      "grad_norm": 1.084893822669983,
      "learning_rate": 9.827236842105264e-05,
      "loss": 0.463,
      "step": 26577
    },
    {
      "epoch": 1.4882965617650352,
      "grad_norm": 1.3243541717529297,
      "learning_rate": 9.82721052631579e-05,
      "loss": 0.4241,
      "step": 26578
    },
    {
      "epoch": 1.4883525590771642,
      "grad_norm": 1.4677218198776245,
      "learning_rate": 9.827184210526317e-05,
      "loss": 0.5732,
      "step": 26579
    },
    {
      "epoch": 1.4884085563892933,
      "grad_norm": 1.4174957275390625,
      "learning_rate": 9.827157894736842e-05,
      "loss": 0.4608,
      "step": 26580
    },
    {
      "epoch": 1.4884645537014223,
      "grad_norm": 1.155500888824463,
      "learning_rate": 9.827131578947368e-05,
      "loss": 0.358,
      "step": 26581
    },
    {
      "epoch": 1.4885205510135513,
      "grad_norm": 1.275414228439331,
      "learning_rate": 9.827105263157895e-05,
      "loss": 0.4646,
      "step": 26582
    },
    {
      "epoch": 1.4885765483256803,
      "grad_norm": 1.2806388139724731,
      "learning_rate": 9.827078947368421e-05,
      "loss": 0.4829,
      "step": 26583
    },
    {
      "epoch": 1.4886325456378093,
      "grad_norm": 1.2613698244094849,
      "learning_rate": 9.827052631578947e-05,
      "loss": 0.4348,
      "step": 26584
    },
    {
      "epoch": 1.4886885429499384,
      "grad_norm": 1.1614763736724854,
      "learning_rate": 9.827026315789473e-05,
      "loss": 0.4095,
      "step": 26585
    },
    {
      "epoch": 1.4887445402620674,
      "grad_norm": 1.1521501541137695,
      "learning_rate": 9.827e-05,
      "loss": 0.4157,
      "step": 26586
    },
    {
      "epoch": 1.4888005375741964,
      "grad_norm": 1.5595088005065918,
      "learning_rate": 9.826973684210527e-05,
      "loss": 0.4135,
      "step": 26587
    },
    {
      "epoch": 1.4888565348863254,
      "grad_norm": 1.1848524808883667,
      "learning_rate": 9.826947368421054e-05,
      "loss": 0.432,
      "step": 26588
    },
    {
      "epoch": 1.4889125321984544,
      "grad_norm": 2.0190978050231934,
      "learning_rate": 9.826921052631579e-05,
      "loss": 0.4802,
      "step": 26589
    },
    {
      "epoch": 1.4889685295105834,
      "grad_norm": 1.8962711095809937,
      "learning_rate": 9.826894736842106e-05,
      "loss": 0.6623,
      "step": 26590
    },
    {
      "epoch": 1.4890245268227125,
      "grad_norm": 1.5094536542892456,
      "learning_rate": 9.826868421052632e-05,
      "loss": 0.5381,
      "step": 26591
    },
    {
      "epoch": 1.4890805241348415,
      "grad_norm": 1.4123467206954956,
      "learning_rate": 9.826842105263159e-05,
      "loss": 0.4597,
      "step": 26592
    },
    {
      "epoch": 1.4891365214469705,
      "grad_norm": 1.146776795387268,
      "learning_rate": 9.826815789473685e-05,
      "loss": 0.4171,
      "step": 26593
    },
    {
      "epoch": 1.4891925187590995,
      "grad_norm": 1.3933452367782593,
      "learning_rate": 9.826789473684211e-05,
      "loss": 0.5097,
      "step": 26594
    },
    {
      "epoch": 1.4892485160712285,
      "grad_norm": 1.508174180984497,
      "learning_rate": 9.826763157894737e-05,
      "loss": 0.5648,
      "step": 26595
    },
    {
      "epoch": 1.4893045133833576,
      "grad_norm": 1.2280765771865845,
      "learning_rate": 9.826736842105264e-05,
      "loss": 0.4365,
      "step": 26596
    },
    {
      "epoch": 1.4893605106954866,
      "grad_norm": 1.496867299079895,
      "learning_rate": 9.82671052631579e-05,
      "loss": 0.4005,
      "step": 26597
    },
    {
      "epoch": 1.4894165080076156,
      "grad_norm": 1.4241769313812256,
      "learning_rate": 9.826684210526316e-05,
      "loss": 0.5276,
      "step": 26598
    },
    {
      "epoch": 1.4894725053197446,
      "grad_norm": 1.3316019773483276,
      "learning_rate": 9.826657894736842e-05,
      "loss": 0.4393,
      "step": 26599
    },
    {
      "epoch": 1.4895285026318736,
      "grad_norm": 1.1914223432540894,
      "learning_rate": 9.826631578947368e-05,
      "loss": 0.3833,
      "step": 26600
    },
    {
      "epoch": 1.4895844999440027,
      "grad_norm": 1.281269907951355,
      "learning_rate": 9.826605263157896e-05,
      "loss": 0.5015,
      "step": 26601
    },
    {
      "epoch": 1.4896404972561317,
      "grad_norm": 1.5901553630828857,
      "learning_rate": 9.826578947368422e-05,
      "loss": 0.4524,
      "step": 26602
    },
    {
      "epoch": 1.4896964945682607,
      "grad_norm": 1.4138697385787964,
      "learning_rate": 9.826552631578948e-05,
      "loss": 0.508,
      "step": 26603
    },
    {
      "epoch": 1.4897524918803897,
      "grad_norm": 1.2021832466125488,
      "learning_rate": 9.826526315789474e-05,
      "loss": 0.5529,
      "step": 26604
    },
    {
      "epoch": 1.4898084891925187,
      "grad_norm": 1.3018269538879395,
      "learning_rate": 9.826500000000001e-05,
      "loss": 0.4325,
      "step": 26605
    },
    {
      "epoch": 1.4898644865046478,
      "grad_norm": 1.1973603963851929,
      "learning_rate": 9.826473684210527e-05,
      "loss": 0.3586,
      "step": 26606
    },
    {
      "epoch": 1.4899204838167768,
      "grad_norm": 1.347508192062378,
      "learning_rate": 9.826447368421053e-05,
      "loss": 0.4202,
      "step": 26607
    },
    {
      "epoch": 1.4899764811289058,
      "grad_norm": 1.521946907043457,
      "learning_rate": 9.826421052631579e-05,
      "loss": 0.4604,
      "step": 26608
    },
    {
      "epoch": 1.4900324784410348,
      "grad_norm": 1.5154759883880615,
      "learning_rate": 9.826394736842106e-05,
      "loss": 0.6821,
      "step": 26609
    },
    {
      "epoch": 1.4900884757531638,
      "grad_norm": 1.2492790222167969,
      "learning_rate": 9.826368421052632e-05,
      "loss": 0.3793,
      "step": 26610
    },
    {
      "epoch": 1.4901444730652929,
      "grad_norm": 1.5363818407058716,
      "learning_rate": 9.82634210526316e-05,
      "loss": 0.4483,
      "step": 26611
    },
    {
      "epoch": 1.4902004703774219,
      "grad_norm": 1.3126407861709595,
      "learning_rate": 9.826315789473684e-05,
      "loss": 0.5083,
      "step": 26612
    },
    {
      "epoch": 1.490256467689551,
      "grad_norm": 1.3964883089065552,
      "learning_rate": 9.826289473684211e-05,
      "loss": 0.4706,
      "step": 26613
    },
    {
      "epoch": 1.49031246500168,
      "grad_norm": 1.463017463684082,
      "learning_rate": 9.826263157894737e-05,
      "loss": 0.4911,
      "step": 26614
    },
    {
      "epoch": 1.490368462313809,
      "grad_norm": 1.0788241624832153,
      "learning_rate": 9.826236842105263e-05,
      "loss": 0.3901,
      "step": 26615
    },
    {
      "epoch": 1.490424459625938,
      "grad_norm": 1.3994767665863037,
      "learning_rate": 9.82621052631579e-05,
      "loss": 0.5225,
      "step": 26616
    },
    {
      "epoch": 1.490480456938067,
      "grad_norm": 1.4026886224746704,
      "learning_rate": 9.826184210526315e-05,
      "loss": 0.5064,
      "step": 26617
    },
    {
      "epoch": 1.490536454250196,
      "grad_norm": 1.3137027025222778,
      "learning_rate": 9.826157894736843e-05,
      "loss": 0.5917,
      "step": 26618
    },
    {
      "epoch": 1.490592451562325,
      "grad_norm": 1.2471706867218018,
      "learning_rate": 9.826131578947369e-05,
      "loss": 0.5021,
      "step": 26619
    },
    {
      "epoch": 1.490648448874454,
      "grad_norm": 1.3848005533218384,
      "learning_rate": 9.826105263157896e-05,
      "loss": 0.6732,
      "step": 26620
    },
    {
      "epoch": 1.490704446186583,
      "grad_norm": 1.459981083869934,
      "learning_rate": 9.82607894736842e-05,
      "loss": 0.4569,
      "step": 26621
    },
    {
      "epoch": 1.490760443498712,
      "grad_norm": 1.3855366706848145,
      "learning_rate": 9.826052631578948e-05,
      "loss": 0.5632,
      "step": 26622
    },
    {
      "epoch": 1.490816440810841,
      "grad_norm": 1.3110512495040894,
      "learning_rate": 9.826026315789474e-05,
      "loss": 0.3802,
      "step": 26623
    },
    {
      "epoch": 1.49087243812297,
      "grad_norm": 1.2619868516921997,
      "learning_rate": 9.826000000000001e-05,
      "loss": 0.4903,
      "step": 26624
    },
    {
      "epoch": 1.4909284354350991,
      "grad_norm": 1.2366193532943726,
      "learning_rate": 9.825973684210527e-05,
      "loss": 0.5048,
      "step": 26625
    },
    {
      "epoch": 1.4909844327472281,
      "grad_norm": 1.1477752923965454,
      "learning_rate": 9.825947368421053e-05,
      "loss": 0.5035,
      "step": 26626
    },
    {
      "epoch": 1.4910404300593572,
      "grad_norm": 1.4129172563552856,
      "learning_rate": 9.825921052631579e-05,
      "loss": 0.5237,
      "step": 26627
    },
    {
      "epoch": 1.4910964273714862,
      "grad_norm": 1.224112868309021,
      "learning_rate": 9.825894736842106e-05,
      "loss": 0.3905,
      "step": 26628
    },
    {
      "epoch": 1.4911524246836152,
      "grad_norm": 1.2950512170791626,
      "learning_rate": 9.825868421052632e-05,
      "loss": 0.4303,
      "step": 26629
    },
    {
      "epoch": 1.4912084219957442,
      "grad_norm": 2.1667160987854004,
      "learning_rate": 9.825842105263158e-05,
      "loss": 0.4029,
      "step": 26630
    },
    {
      "epoch": 1.4912644193078732,
      "grad_norm": 1.4129606485366821,
      "learning_rate": 9.825815789473684e-05,
      "loss": 0.4251,
      "step": 26631
    },
    {
      "epoch": 1.4913204166200023,
      "grad_norm": 1.4479670524597168,
      "learning_rate": 9.825789473684212e-05,
      "loss": 0.4992,
      "step": 26632
    },
    {
      "epoch": 1.4913764139321313,
      "grad_norm": 1.3471579551696777,
      "learning_rate": 9.825763157894738e-05,
      "loss": 0.4921,
      "step": 26633
    },
    {
      "epoch": 1.4914324112442603,
      "grad_norm": 1.6021217107772827,
      "learning_rate": 9.825736842105264e-05,
      "loss": 0.4532,
      "step": 26634
    },
    {
      "epoch": 1.4914884085563893,
      "grad_norm": 1.3492587804794312,
      "learning_rate": 9.82571052631579e-05,
      "loss": 0.482,
      "step": 26635
    },
    {
      "epoch": 1.4915444058685183,
      "grad_norm": 1.9318296909332275,
      "learning_rate": 9.825684210526316e-05,
      "loss": 0.4424,
      "step": 26636
    },
    {
      "epoch": 1.4916004031806473,
      "grad_norm": 1.05913245677948,
      "learning_rate": 9.825657894736843e-05,
      "loss": 0.4489,
      "step": 26637
    },
    {
      "epoch": 1.4916564004927764,
      "grad_norm": 1.2017278671264648,
      "learning_rate": 9.825631578947369e-05,
      "loss": 0.4875,
      "step": 26638
    },
    {
      "epoch": 1.4917123978049054,
      "grad_norm": 1.4743341207504272,
      "learning_rate": 9.825605263157895e-05,
      "loss": 0.5475,
      "step": 26639
    },
    {
      "epoch": 1.4917683951170344,
      "grad_norm": 1.2858116626739502,
      "learning_rate": 9.825578947368421e-05,
      "loss": 0.4494,
      "step": 26640
    },
    {
      "epoch": 1.4918243924291634,
      "grad_norm": 1.4351807832717896,
      "learning_rate": 9.825552631578948e-05,
      "loss": 0.3805,
      "step": 26641
    },
    {
      "epoch": 1.4918803897412924,
      "grad_norm": 1.4434479475021362,
      "learning_rate": 9.825526315789474e-05,
      "loss": 0.5806,
      "step": 26642
    },
    {
      "epoch": 1.4919363870534215,
      "grad_norm": 1.21992027759552,
      "learning_rate": 9.825500000000001e-05,
      "loss": 0.413,
      "step": 26643
    },
    {
      "epoch": 1.4919923843655505,
      "grad_norm": 1.1825666427612305,
      "learning_rate": 9.825473684210526e-05,
      "loss": 0.4294,
      "step": 26644
    },
    {
      "epoch": 1.4920483816776795,
      "grad_norm": 1.4497736692428589,
      "learning_rate": 9.825447368421053e-05,
      "loss": 0.5618,
      "step": 26645
    },
    {
      "epoch": 1.4921043789898085,
      "grad_norm": 1.3041061162948608,
      "learning_rate": 9.825421052631579e-05,
      "loss": 0.5717,
      "step": 26646
    },
    {
      "epoch": 1.4921603763019375,
      "grad_norm": 1.6015552282333374,
      "learning_rate": 9.825394736842107e-05,
      "loss": 0.4817,
      "step": 26647
    },
    {
      "epoch": 1.4922163736140666,
      "grad_norm": 1.276550054550171,
      "learning_rate": 9.825368421052633e-05,
      "loss": 0.4927,
      "step": 26648
    },
    {
      "epoch": 1.4922723709261956,
      "grad_norm": 1.363005518913269,
      "learning_rate": 9.825342105263159e-05,
      "loss": 0.4021,
      "step": 26649
    },
    {
      "epoch": 1.4923283682383246,
      "grad_norm": 1.5803815126419067,
      "learning_rate": 9.825315789473685e-05,
      "loss": 0.6846,
      "step": 26650
    },
    {
      "epoch": 1.4923843655504536,
      "grad_norm": 1.558101773262024,
      "learning_rate": 9.82528947368421e-05,
      "loss": 0.4083,
      "step": 26651
    },
    {
      "epoch": 1.4924403628625826,
      "grad_norm": 1.2533341646194458,
      "learning_rate": 9.825263157894738e-05,
      "loss": 0.4739,
      "step": 26652
    },
    {
      "epoch": 1.4924963601747117,
      "grad_norm": 1.1889357566833496,
      "learning_rate": 9.825236842105264e-05,
      "loss": 0.3874,
      "step": 26653
    },
    {
      "epoch": 1.4925523574868407,
      "grad_norm": 1.213054895401001,
      "learning_rate": 9.82521052631579e-05,
      "loss": 0.424,
      "step": 26654
    },
    {
      "epoch": 1.4926083547989697,
      "grad_norm": 1.3365778923034668,
      "learning_rate": 9.825184210526316e-05,
      "loss": 0.3959,
      "step": 26655
    },
    {
      "epoch": 1.4926643521110987,
      "grad_norm": 1.2400928735733032,
      "learning_rate": 9.825157894736843e-05,
      "loss": 0.4036,
      "step": 26656
    },
    {
      "epoch": 1.4927203494232277,
      "grad_norm": 1.3384134769439697,
      "learning_rate": 9.825131578947369e-05,
      "loss": 0.4183,
      "step": 26657
    },
    {
      "epoch": 1.4927763467353568,
      "grad_norm": 1.3476084470748901,
      "learning_rate": 9.825105263157895e-05,
      "loss": 0.5117,
      "step": 26658
    },
    {
      "epoch": 1.4928323440474858,
      "grad_norm": 1.2372089624404907,
      "learning_rate": 9.825078947368421e-05,
      "loss": 0.4327,
      "step": 26659
    },
    {
      "epoch": 1.4928883413596148,
      "grad_norm": 1.314595341682434,
      "learning_rate": 9.825052631578948e-05,
      "loss": 0.522,
      "step": 26660
    },
    {
      "epoch": 1.4929443386717438,
      "grad_norm": 1.4243106842041016,
      "learning_rate": 9.825026315789474e-05,
      "loss": 0.5655,
      "step": 26661
    },
    {
      "epoch": 1.4930003359838728,
      "grad_norm": 1.6291049718856812,
      "learning_rate": 9.825e-05,
      "loss": 0.5341,
      "step": 26662
    },
    {
      "epoch": 1.4930563332960018,
      "grad_norm": 1.1582446098327637,
      "learning_rate": 9.824973684210526e-05,
      "loss": 0.3955,
      "step": 26663
    },
    {
      "epoch": 1.4931123306081309,
      "grad_norm": 1.3123834133148193,
      "learning_rate": 9.824947368421054e-05,
      "loss": 0.4636,
      "step": 26664
    },
    {
      "epoch": 1.4931683279202599,
      "grad_norm": 46.05010223388672,
      "learning_rate": 9.82492105263158e-05,
      "loss": 0.5557,
      "step": 26665
    },
    {
      "epoch": 1.493224325232389,
      "grad_norm": 1.4509391784667969,
      "learning_rate": 9.824894736842107e-05,
      "loss": 0.5353,
      "step": 26666
    },
    {
      "epoch": 1.493280322544518,
      "grad_norm": 1.1491566896438599,
      "learning_rate": 9.824868421052632e-05,
      "loss": 0.386,
      "step": 26667
    },
    {
      "epoch": 1.493336319856647,
      "grad_norm": 1.6083931922912598,
      "learning_rate": 9.824842105263157e-05,
      "loss": 0.581,
      "step": 26668
    },
    {
      "epoch": 1.493392317168776,
      "grad_norm": 1.2309685945510864,
      "learning_rate": 9.824815789473685e-05,
      "loss": 0.366,
      "step": 26669
    },
    {
      "epoch": 1.493448314480905,
      "grad_norm": 1.7661563158035278,
      "learning_rate": 9.824789473684211e-05,
      "loss": 0.6173,
      "step": 26670
    },
    {
      "epoch": 1.493504311793034,
      "grad_norm": 1.210553526878357,
      "learning_rate": 9.824763157894738e-05,
      "loss": 0.4531,
      "step": 26671
    },
    {
      "epoch": 1.493560309105163,
      "grad_norm": 1.2885973453521729,
      "learning_rate": 9.824736842105263e-05,
      "loss": 0.4018,
      "step": 26672
    },
    {
      "epoch": 1.493616306417292,
      "grad_norm": 1.4575316905975342,
      "learning_rate": 9.82471052631579e-05,
      "loss": 0.4969,
      "step": 26673
    },
    {
      "epoch": 1.493672303729421,
      "grad_norm": 1.6433037519454956,
      "learning_rate": 9.824684210526316e-05,
      "loss": 0.5275,
      "step": 26674
    },
    {
      "epoch": 1.49372830104155,
      "grad_norm": 1.4665077924728394,
      "learning_rate": 9.824657894736843e-05,
      "loss": 0.5303,
      "step": 26675
    },
    {
      "epoch": 1.493784298353679,
      "grad_norm": 1.18905770778656,
      "learning_rate": 9.824631578947368e-05,
      "loss": 0.4734,
      "step": 26676
    },
    {
      "epoch": 1.4938402956658081,
      "grad_norm": 1.1709564924240112,
      "learning_rate": 9.824605263157895e-05,
      "loss": 0.389,
      "step": 26677
    },
    {
      "epoch": 1.4938962929779371,
      "grad_norm": 1.2126034498214722,
      "learning_rate": 9.824578947368421e-05,
      "loss": 0.3979,
      "step": 26678
    },
    {
      "epoch": 1.4939522902900662,
      "grad_norm": 1.1228069067001343,
      "learning_rate": 9.824552631578949e-05,
      "loss": 0.3886,
      "step": 26679
    },
    {
      "epoch": 1.4940082876021952,
      "grad_norm": 1.4189866781234741,
      "learning_rate": 9.824526315789475e-05,
      "loss": 0.4907,
      "step": 26680
    },
    {
      "epoch": 1.4940642849143242,
      "grad_norm": 1.4428449869155884,
      "learning_rate": 9.8245e-05,
      "loss": 0.4597,
      "step": 26681
    },
    {
      "epoch": 1.4941202822264532,
      "grad_norm": 1.293521523475647,
      "learning_rate": 9.824473684210527e-05,
      "loss": 0.466,
      "step": 26682
    },
    {
      "epoch": 1.4941762795385822,
      "grad_norm": 1.3009419441223145,
      "learning_rate": 9.824447368421054e-05,
      "loss": 0.4997,
      "step": 26683
    },
    {
      "epoch": 1.494232276850711,
      "grad_norm": 1.138881802558899,
      "learning_rate": 9.82442105263158e-05,
      "loss": 0.3961,
      "step": 26684
    },
    {
      "epoch": 1.49428827416284,
      "grad_norm": 1.5244901180267334,
      "learning_rate": 9.824394736842106e-05,
      "loss": 0.4786,
      "step": 26685
    },
    {
      "epoch": 1.494344271474969,
      "grad_norm": 1.3683289289474487,
      "learning_rate": 9.824368421052632e-05,
      "loss": 0.5121,
      "step": 26686
    },
    {
      "epoch": 1.494400268787098,
      "grad_norm": 1.4406224489212036,
      "learning_rate": 9.824342105263158e-05,
      "loss": 0.4976,
      "step": 26687
    },
    {
      "epoch": 1.494456266099227,
      "grad_norm": 1.1860580444335938,
      "learning_rate": 9.824315789473685e-05,
      "loss": 0.3892,
      "step": 26688
    },
    {
      "epoch": 1.4945122634113561,
      "grad_norm": 1.1105409860610962,
      "learning_rate": 9.824289473684211e-05,
      "loss": 0.3432,
      "step": 26689
    },
    {
      "epoch": 1.4945682607234851,
      "grad_norm": 1.489783763885498,
      "learning_rate": 9.824263157894737e-05,
      "loss": 0.6195,
      "step": 26690
    },
    {
      "epoch": 1.4946242580356142,
      "grad_norm": 1.3480902910232544,
      "learning_rate": 9.824236842105263e-05,
      "loss": 0.4823,
      "step": 26691
    },
    {
      "epoch": 1.4946802553477432,
      "grad_norm": 1.6353440284729004,
      "learning_rate": 9.82421052631579e-05,
      "loss": 0.6926,
      "step": 26692
    },
    {
      "epoch": 1.4947362526598722,
      "grad_norm": 1.5540894269943237,
      "learning_rate": 9.824184210526316e-05,
      "loss": 0.4531,
      "step": 26693
    },
    {
      "epoch": 1.4947922499720012,
      "grad_norm": 1.8608779907226562,
      "learning_rate": 9.824157894736842e-05,
      "loss": 0.5218,
      "step": 26694
    },
    {
      "epoch": 1.4948482472841302,
      "grad_norm": 1.2090544700622559,
      "learning_rate": 9.824131578947368e-05,
      "loss": 0.4046,
      "step": 26695
    },
    {
      "epoch": 1.4949042445962593,
      "grad_norm": 1.2432013750076294,
      "learning_rate": 9.824105263157896e-05,
      "loss": 0.4473,
      "step": 26696
    },
    {
      "epoch": 1.4949602419083883,
      "grad_norm": 1.1945785284042358,
      "learning_rate": 9.824078947368422e-05,
      "loss": 0.3855,
      "step": 26697
    },
    {
      "epoch": 1.4950162392205173,
      "grad_norm": 1.077295184135437,
      "learning_rate": 9.824052631578949e-05,
      "loss": 0.3811,
      "step": 26698
    },
    {
      "epoch": 1.4950722365326463,
      "grad_norm": 1.4812551736831665,
      "learning_rate": 9.824026315789473e-05,
      "loss": 0.6106,
      "step": 26699
    },
    {
      "epoch": 1.4951282338447753,
      "grad_norm": 1.4711402654647827,
      "learning_rate": 9.824000000000001e-05,
      "loss": 0.3819,
      "step": 26700
    },
    {
      "epoch": 1.4951842311569044,
      "grad_norm": 1.2971477508544922,
      "learning_rate": 9.823973684210527e-05,
      "loss": 0.4607,
      "step": 26701
    },
    {
      "epoch": 1.4952402284690334,
      "grad_norm": 1.1944701671600342,
      "learning_rate": 9.823947368421053e-05,
      "loss": 0.4241,
      "step": 26702
    },
    {
      "epoch": 1.4952962257811624,
      "grad_norm": 1.2420014142990112,
      "learning_rate": 9.82392105263158e-05,
      "loss": 0.3327,
      "step": 26703
    },
    {
      "epoch": 1.4953522230932914,
      "grad_norm": 1.3738360404968262,
      "learning_rate": 9.823894736842105e-05,
      "loss": 0.4774,
      "step": 26704
    },
    {
      "epoch": 1.4954082204054204,
      "grad_norm": 1.2295901775360107,
      "learning_rate": 9.823868421052632e-05,
      "loss": 0.4331,
      "step": 26705
    },
    {
      "epoch": 1.4954642177175494,
      "grad_norm": 1.4630894660949707,
      "learning_rate": 9.823842105263158e-05,
      "loss": 0.6747,
      "step": 26706
    },
    {
      "epoch": 1.4955202150296785,
      "grad_norm": 1.5961787700653076,
      "learning_rate": 9.823815789473685e-05,
      "loss": 0.5482,
      "step": 26707
    },
    {
      "epoch": 1.4955762123418075,
      "grad_norm": 1.469846248626709,
      "learning_rate": 9.823789473684211e-05,
      "loss": 0.62,
      "step": 26708
    },
    {
      "epoch": 1.4956322096539365,
      "grad_norm": 1.65489661693573,
      "learning_rate": 9.823763157894737e-05,
      "loss": 0.5532,
      "step": 26709
    },
    {
      "epoch": 1.4956882069660655,
      "grad_norm": 1.1136822700500488,
      "learning_rate": 9.823736842105263e-05,
      "loss": 0.4125,
      "step": 26710
    },
    {
      "epoch": 1.4957442042781945,
      "grad_norm": 1.4769656658172607,
      "learning_rate": 9.82371052631579e-05,
      "loss": 0.4585,
      "step": 26711
    },
    {
      "epoch": 1.4958002015903236,
      "grad_norm": 1.5494664907455444,
      "learning_rate": 9.823684210526317e-05,
      "loss": 0.514,
      "step": 26712
    },
    {
      "epoch": 1.4958561989024526,
      "grad_norm": 1.2355320453643799,
      "learning_rate": 9.823657894736843e-05,
      "loss": 0.4382,
      "step": 26713
    },
    {
      "epoch": 1.4959121962145816,
      "grad_norm": 1.2471064329147339,
      "learning_rate": 9.823631578947368e-05,
      "loss": 0.4819,
      "step": 26714
    },
    {
      "epoch": 1.4959681935267106,
      "grad_norm": 1.4682189226150513,
      "learning_rate": 9.823605263157896e-05,
      "loss": 0.5551,
      "step": 26715
    },
    {
      "epoch": 1.4960241908388396,
      "grad_norm": 1.6817207336425781,
      "learning_rate": 9.823578947368422e-05,
      "loss": 0.6358,
      "step": 26716
    },
    {
      "epoch": 1.4960801881509687,
      "grad_norm": 1.5784755945205688,
      "learning_rate": 9.823552631578948e-05,
      "loss": 0.4629,
      "step": 26717
    },
    {
      "epoch": 1.4961361854630977,
      "grad_norm": 1.316011667251587,
      "learning_rate": 9.823526315789474e-05,
      "loss": 0.481,
      "step": 26718
    },
    {
      "epoch": 1.4961921827752267,
      "grad_norm": 1.3917642831802368,
      "learning_rate": 9.8235e-05,
      "loss": 0.4046,
      "step": 26719
    },
    {
      "epoch": 1.4962481800873557,
      "grad_norm": 1.3899905681610107,
      "learning_rate": 9.823473684210527e-05,
      "loss": 0.4627,
      "step": 26720
    },
    {
      "epoch": 1.4963041773994847,
      "grad_norm": 1.5084383487701416,
      "learning_rate": 9.823447368421053e-05,
      "loss": 0.3227,
      "step": 26721
    },
    {
      "epoch": 1.4963601747116138,
      "grad_norm": 1.2321195602416992,
      "learning_rate": 9.823421052631579e-05,
      "loss": 0.4581,
      "step": 26722
    },
    {
      "epoch": 1.4964161720237428,
      "grad_norm": 1.4483340978622437,
      "learning_rate": 9.823394736842105e-05,
      "loss": 0.6091,
      "step": 26723
    },
    {
      "epoch": 1.4964721693358718,
      "grad_norm": 1.0192410945892334,
      "learning_rate": 9.823368421052632e-05,
      "loss": 0.3762,
      "step": 26724
    },
    {
      "epoch": 1.4965281666480008,
      "grad_norm": 1.1380773782730103,
      "learning_rate": 9.823342105263158e-05,
      "loss": 0.3941,
      "step": 26725
    },
    {
      "epoch": 1.4965841639601298,
      "grad_norm": 1.5388333797454834,
      "learning_rate": 9.823315789473686e-05,
      "loss": 0.4075,
      "step": 26726
    },
    {
      "epoch": 1.4966401612722589,
      "grad_norm": 1.4268181324005127,
      "learning_rate": 9.82328947368421e-05,
      "loss": 0.4967,
      "step": 26727
    },
    {
      "epoch": 1.4966961585843879,
      "grad_norm": 1.0744291543960571,
      "learning_rate": 9.823263157894738e-05,
      "loss": 0.4012,
      "step": 26728
    },
    {
      "epoch": 1.496752155896517,
      "grad_norm": 1.3298625946044922,
      "learning_rate": 9.823236842105264e-05,
      "loss": 0.4907,
      "step": 26729
    },
    {
      "epoch": 1.496808153208646,
      "grad_norm": 1.501968264579773,
      "learning_rate": 9.823210526315791e-05,
      "loss": 0.6027,
      "step": 26730
    },
    {
      "epoch": 1.496864150520775,
      "grad_norm": 1.2917025089263916,
      "learning_rate": 9.823184210526315e-05,
      "loss": 0.4129,
      "step": 26731
    },
    {
      "epoch": 1.496920147832904,
      "grad_norm": 1.2158446311950684,
      "learning_rate": 9.823157894736843e-05,
      "loss": 0.3479,
      "step": 26732
    },
    {
      "epoch": 1.496976145145033,
      "grad_norm": 1.1381868124008179,
      "learning_rate": 9.823131578947369e-05,
      "loss": 0.4021,
      "step": 26733
    },
    {
      "epoch": 1.497032142457162,
      "grad_norm": 1.2020201683044434,
      "learning_rate": 9.823105263157896e-05,
      "loss": 0.3444,
      "step": 26734
    },
    {
      "epoch": 1.497088139769291,
      "grad_norm": 1.4698524475097656,
      "learning_rate": 9.823078947368422e-05,
      "loss": 0.4059,
      "step": 26735
    },
    {
      "epoch": 1.49714413708142,
      "grad_norm": 1.5952394008636475,
      "learning_rate": 9.823052631578947e-05,
      "loss": 0.4962,
      "step": 26736
    },
    {
      "epoch": 1.497200134393549,
      "grad_norm": 1.15773606300354,
      "learning_rate": 9.823026315789474e-05,
      "loss": 0.4159,
      "step": 26737
    },
    {
      "epoch": 1.497256131705678,
      "grad_norm": 1.5245182514190674,
      "learning_rate": 9.823e-05,
      "loss": 0.577,
      "step": 26738
    },
    {
      "epoch": 1.497312129017807,
      "grad_norm": 1.2623963356018066,
      "learning_rate": 9.822973684210527e-05,
      "loss": 0.3999,
      "step": 26739
    },
    {
      "epoch": 1.497368126329936,
      "grad_norm": 1.3014445304870605,
      "learning_rate": 9.822947368421053e-05,
      "loss": 0.395,
      "step": 26740
    },
    {
      "epoch": 1.4974241236420651,
      "grad_norm": 1.1897354125976562,
      "learning_rate": 9.822921052631579e-05,
      "loss": 0.4322,
      "step": 26741
    },
    {
      "epoch": 1.4974801209541941,
      "grad_norm": 1.2164031267166138,
      "learning_rate": 9.822894736842105e-05,
      "loss": 0.4825,
      "step": 26742
    },
    {
      "epoch": 1.4975361182663232,
      "grad_norm": 1.5846658945083618,
      "learning_rate": 9.822868421052633e-05,
      "loss": 0.6695,
      "step": 26743
    },
    {
      "epoch": 1.4975921155784522,
      "grad_norm": 1.182814598083496,
      "learning_rate": 9.822842105263159e-05,
      "loss": 0.421,
      "step": 26744
    },
    {
      "epoch": 1.4976481128905812,
      "grad_norm": 1.4049961566925049,
      "learning_rate": 9.822815789473684e-05,
      "loss": 0.3883,
      "step": 26745
    },
    {
      "epoch": 1.4977041102027102,
      "grad_norm": 1.538678765296936,
      "learning_rate": 9.82278947368421e-05,
      "loss": 0.5299,
      "step": 26746
    },
    {
      "epoch": 1.4977601075148392,
      "grad_norm": 1.246418833732605,
      "learning_rate": 9.822763157894738e-05,
      "loss": 0.3696,
      "step": 26747
    },
    {
      "epoch": 1.4978161048269683,
      "grad_norm": 1.547446846961975,
      "learning_rate": 9.822736842105264e-05,
      "loss": 0.5213,
      "step": 26748
    },
    {
      "epoch": 1.4978721021390973,
      "grad_norm": 1.4471827745437622,
      "learning_rate": 9.82271052631579e-05,
      "loss": 0.3878,
      "step": 26749
    },
    {
      "epoch": 1.4979280994512263,
      "grad_norm": 1.4013447761535645,
      "learning_rate": 9.822684210526316e-05,
      "loss": 0.5327,
      "step": 26750
    },
    {
      "epoch": 1.4979840967633553,
      "grad_norm": 1.5289849042892456,
      "learning_rate": 9.822657894736843e-05,
      "loss": 0.488,
      "step": 26751
    },
    {
      "epoch": 1.4980400940754843,
      "grad_norm": 1.0098973512649536,
      "learning_rate": 9.822631578947369e-05,
      "loss": 0.3345,
      "step": 26752
    },
    {
      "epoch": 1.4980960913876133,
      "grad_norm": 1.0826987028121948,
      "learning_rate": 9.822605263157896e-05,
      "loss": 0.5749,
      "step": 26753
    },
    {
      "epoch": 1.4981520886997424,
      "grad_norm": 1.5209635496139526,
      "learning_rate": 9.822578947368421e-05,
      "loss": 0.4544,
      "step": 26754
    },
    {
      "epoch": 1.4982080860118714,
      "grad_norm": 2.081101417541504,
      "learning_rate": 9.822552631578947e-05,
      "loss": 0.8189,
      "step": 26755
    },
    {
      "epoch": 1.4982640833240004,
      "grad_norm": 1.442839503288269,
      "learning_rate": 9.822526315789474e-05,
      "loss": 0.4608,
      "step": 26756
    },
    {
      "epoch": 1.4983200806361294,
      "grad_norm": 1.606909990310669,
      "learning_rate": 9.8225e-05,
      "loss": 0.4206,
      "step": 26757
    },
    {
      "epoch": 1.4983760779482584,
      "grad_norm": 1.4205514192581177,
      "learning_rate": 9.822473684210528e-05,
      "loss": 0.4728,
      "step": 26758
    },
    {
      "epoch": 1.4984320752603875,
      "grad_norm": 1.2935292720794678,
      "learning_rate": 9.822447368421052e-05,
      "loss": 0.4086,
      "step": 26759
    },
    {
      "epoch": 1.4984880725725165,
      "grad_norm": 1.299617052078247,
      "learning_rate": 9.82242105263158e-05,
      "loss": 0.483,
      "step": 26760
    },
    {
      "epoch": 1.4985440698846455,
      "grad_norm": 1.3414956331253052,
      "learning_rate": 9.822394736842105e-05,
      "loss": 0.3733,
      "step": 26761
    },
    {
      "epoch": 1.4986000671967745,
      "grad_norm": 1.2331242561340332,
      "learning_rate": 9.822368421052633e-05,
      "loss": 0.4835,
      "step": 26762
    },
    {
      "epoch": 1.4986560645089035,
      "grad_norm": 1.2244077920913696,
      "learning_rate": 9.822342105263159e-05,
      "loss": 0.3698,
      "step": 26763
    },
    {
      "epoch": 1.4987120618210326,
      "grad_norm": 1.3659173250198364,
      "learning_rate": 9.822315789473685e-05,
      "loss": 0.3211,
      "step": 26764
    },
    {
      "epoch": 1.4987680591331616,
      "grad_norm": 1.341306447982788,
      "learning_rate": 9.822289473684211e-05,
      "loss": 0.4194,
      "step": 26765
    },
    {
      "epoch": 1.4988240564452906,
      "grad_norm": 1.5292794704437256,
      "learning_rate": 9.822263157894738e-05,
      "loss": 0.464,
      "step": 26766
    },
    {
      "epoch": 1.4988800537574196,
      "grad_norm": 1.1992926597595215,
      "learning_rate": 9.822236842105264e-05,
      "loss": 0.4708,
      "step": 26767
    },
    {
      "epoch": 1.4989360510695486,
      "grad_norm": 1.3847891092300415,
      "learning_rate": 9.82221052631579e-05,
      "loss": 0.4973,
      "step": 26768
    },
    {
      "epoch": 1.4989920483816777,
      "grad_norm": 1.101475715637207,
      "learning_rate": 9.822184210526316e-05,
      "loss": 0.4568,
      "step": 26769
    },
    {
      "epoch": 1.4990480456938067,
      "grad_norm": 1.3191195726394653,
      "learning_rate": 9.822157894736843e-05,
      "loss": 0.3405,
      "step": 26770
    },
    {
      "epoch": 1.4991040430059357,
      "grad_norm": 1.291374921798706,
      "learning_rate": 9.822131578947369e-05,
      "loss": 0.4529,
      "step": 26771
    },
    {
      "epoch": 1.4991600403180647,
      "grad_norm": 1.3544799089431763,
      "learning_rate": 9.822105263157895e-05,
      "loss": 0.4364,
      "step": 26772
    },
    {
      "epoch": 1.4992160376301937,
      "grad_norm": 1.2974934577941895,
      "learning_rate": 9.822078947368421e-05,
      "loss": 0.407,
      "step": 26773
    },
    {
      "epoch": 1.4992720349423228,
      "grad_norm": 1.4829468727111816,
      "learning_rate": 9.822052631578947e-05,
      "loss": 0.4433,
      "step": 26774
    },
    {
      "epoch": 1.4993280322544518,
      "grad_norm": 2.106544017791748,
      "learning_rate": 9.822026315789475e-05,
      "loss": 0.4644,
      "step": 26775
    },
    {
      "epoch": 1.4993840295665808,
      "grad_norm": 1.395676612854004,
      "learning_rate": 9.822e-05,
      "loss": 0.4148,
      "step": 26776
    },
    {
      "epoch": 1.4994400268787098,
      "grad_norm": 1.3502802848815918,
      "learning_rate": 9.821973684210526e-05,
      "loss": 0.6398,
      "step": 26777
    },
    {
      "epoch": 1.4994960241908388,
      "grad_norm": 1.4034233093261719,
      "learning_rate": 9.821947368421052e-05,
      "loss": 0.6107,
      "step": 26778
    },
    {
      "epoch": 1.4995520215029678,
      "grad_norm": 1.351454257965088,
      "learning_rate": 9.82192105263158e-05,
      "loss": 0.4908,
      "step": 26779
    },
    {
      "epoch": 1.4996080188150969,
      "grad_norm": 1.7202394008636475,
      "learning_rate": 9.821894736842106e-05,
      "loss": 0.5555,
      "step": 26780
    },
    {
      "epoch": 1.4996640161272259,
      "grad_norm": 1.1028008460998535,
      "learning_rate": 9.821868421052632e-05,
      "loss": 0.3848,
      "step": 26781
    },
    {
      "epoch": 1.499720013439355,
      "grad_norm": 1.5374046564102173,
      "learning_rate": 9.821842105263158e-05,
      "loss": 0.4258,
      "step": 26782
    },
    {
      "epoch": 1.499776010751484,
      "grad_norm": 1.2351723909378052,
      "learning_rate": 9.821815789473685e-05,
      "loss": 0.4447,
      "step": 26783
    },
    {
      "epoch": 1.499832008063613,
      "grad_norm": 1.3446274995803833,
      "learning_rate": 9.821789473684211e-05,
      "loss": 0.4953,
      "step": 26784
    },
    {
      "epoch": 1.499888005375742,
      "grad_norm": 1.412572979927063,
      "learning_rate": 9.821763157894738e-05,
      "loss": 0.4591,
      "step": 26785
    },
    {
      "epoch": 1.499944002687871,
      "grad_norm": 1.1691511869430542,
      "learning_rate": 9.821736842105263e-05,
      "loss": 0.4391,
      "step": 26786
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2498557567596436,
      "learning_rate": 9.82171052631579e-05,
      "loss": 0.4797,
      "step": 26787
    },
    {
      "epoch": 1.500055997312129,
      "grad_norm": 1.5805020332336426,
      "learning_rate": 9.821684210526316e-05,
      "loss": 0.474,
      "step": 26788
    },
    {
      "epoch": 1.500111994624258,
      "grad_norm": 1.2805150747299194,
      "learning_rate": 9.821657894736842e-05,
      "loss": 0.3752,
      "step": 26789
    },
    {
      "epoch": 1.500167991936387,
      "grad_norm": 1.155444622039795,
      "learning_rate": 9.82163157894737e-05,
      "loss": 0.382,
      "step": 26790
    },
    {
      "epoch": 1.500223989248516,
      "grad_norm": 1.1691333055496216,
      "learning_rate": 9.821605263157894e-05,
      "loss": 0.3829,
      "step": 26791
    },
    {
      "epoch": 1.500279986560645,
      "grad_norm": 1.4243505001068115,
      "learning_rate": 9.821578947368421e-05,
      "loss": 0.5008,
      "step": 26792
    },
    {
      "epoch": 1.5003359838727741,
      "grad_norm": 1.4051177501678467,
      "learning_rate": 9.821552631578947e-05,
      "loss": 0.622,
      "step": 26793
    },
    {
      "epoch": 1.5003919811849031,
      "grad_norm": 1.3862048387527466,
      "learning_rate": 9.821526315789475e-05,
      "loss": 0.4147,
      "step": 26794
    },
    {
      "epoch": 1.5004479784970322,
      "grad_norm": 1.3812639713287354,
      "learning_rate": 9.821500000000001e-05,
      "loss": 0.5005,
      "step": 26795
    },
    {
      "epoch": 1.5005039758091612,
      "grad_norm": 1.407889485359192,
      "learning_rate": 9.821473684210527e-05,
      "loss": 0.5874,
      "step": 26796
    },
    {
      "epoch": 1.5005599731212902,
      "grad_norm": 1.387760043144226,
      "learning_rate": 9.821447368421053e-05,
      "loss": 0.4919,
      "step": 26797
    },
    {
      "epoch": 1.5006159704334192,
      "grad_norm": 1.3240712881088257,
      "learning_rate": 9.82142105263158e-05,
      "loss": 0.4559,
      "step": 26798
    },
    {
      "epoch": 1.5006719677455482,
      "grad_norm": 2.1726765632629395,
      "learning_rate": 9.821394736842106e-05,
      "loss": 0.5946,
      "step": 26799
    },
    {
      "epoch": 1.5007279650576772,
      "grad_norm": 1.0934633016586304,
      "learning_rate": 9.821368421052632e-05,
      "loss": 0.3353,
      "step": 26800
    },
    {
      "epoch": 1.5007839623698063,
      "grad_norm": 1.3092851638793945,
      "learning_rate": 9.821342105263158e-05,
      "loss": 0.5126,
      "step": 26801
    },
    {
      "epoch": 1.5008399596819353,
      "grad_norm": 1.3932744264602661,
      "learning_rate": 9.821315789473685e-05,
      "loss": 0.4586,
      "step": 26802
    },
    {
      "epoch": 1.5008959569940643,
      "grad_norm": 1.3432908058166504,
      "learning_rate": 9.821289473684211e-05,
      "loss": 0.4227,
      "step": 26803
    },
    {
      "epoch": 1.5009519543061933,
      "grad_norm": 1.2111737728118896,
      "learning_rate": 9.821263157894737e-05,
      "loss": 0.4225,
      "step": 26804
    },
    {
      "epoch": 1.5010079516183223,
      "grad_norm": 1.2058987617492676,
      "learning_rate": 9.821236842105263e-05,
      "loss": 0.3126,
      "step": 26805
    },
    {
      "epoch": 1.5010639489304514,
      "grad_norm": 1.4938188791275024,
      "learning_rate": 9.821210526315789e-05,
      "loss": 0.5548,
      "step": 26806
    },
    {
      "epoch": 1.5011199462425804,
      "grad_norm": 1.2865266799926758,
      "learning_rate": 9.821184210526316e-05,
      "loss": 0.4605,
      "step": 26807
    },
    {
      "epoch": 1.5011759435547094,
      "grad_norm": 1.329338550567627,
      "learning_rate": 9.821157894736842e-05,
      "loss": 0.4606,
      "step": 26808
    },
    {
      "epoch": 1.5012319408668384,
      "grad_norm": 1.2825850248336792,
      "learning_rate": 9.821131578947368e-05,
      "loss": 0.4284,
      "step": 26809
    },
    {
      "epoch": 1.5012879381789674,
      "grad_norm": 1.5459418296813965,
      "learning_rate": 9.821105263157894e-05,
      "loss": 0.6218,
      "step": 26810
    },
    {
      "epoch": 1.5013439354910965,
      "grad_norm": 1.3349864482879639,
      "learning_rate": 9.821078947368422e-05,
      "loss": 0.3773,
      "step": 26811
    },
    {
      "epoch": 1.5013999328032255,
      "grad_norm": 1.8708873987197876,
      "learning_rate": 9.821052631578948e-05,
      "loss": 0.4553,
      "step": 26812
    },
    {
      "epoch": 1.5014559301153545,
      "grad_norm": 1.0955673456192017,
      "learning_rate": 9.821026315789475e-05,
      "loss": 0.3634,
      "step": 26813
    },
    {
      "epoch": 1.5015119274274835,
      "grad_norm": 1.1382811069488525,
      "learning_rate": 9.821e-05,
      "loss": 0.4818,
      "step": 26814
    },
    {
      "epoch": 1.5015679247396125,
      "grad_norm": 1.322993516921997,
      "learning_rate": 9.820973684210527e-05,
      "loss": 0.4008,
      "step": 26815
    },
    {
      "epoch": 1.5016239220517416,
      "grad_norm": 1.3150700330734253,
      "learning_rate": 9.820947368421053e-05,
      "loss": 0.535,
      "step": 26816
    },
    {
      "epoch": 1.5016799193638706,
      "grad_norm": 1.4436053037643433,
      "learning_rate": 9.82092105263158e-05,
      "loss": 0.4855,
      "step": 26817
    },
    {
      "epoch": 1.5017359166759996,
      "grad_norm": 1.3064913749694824,
      "learning_rate": 9.820894736842106e-05,
      "loss": 0.4284,
      "step": 26818
    },
    {
      "epoch": 1.5017919139881286,
      "grad_norm": 1.1609876155853271,
      "learning_rate": 9.820868421052632e-05,
      "loss": 0.3699,
      "step": 26819
    },
    {
      "epoch": 1.5018479113002576,
      "grad_norm": 1.4147918224334717,
      "learning_rate": 9.820842105263158e-05,
      "loss": 0.5225,
      "step": 26820
    },
    {
      "epoch": 1.5019039086123867,
      "grad_norm": 1.2616310119628906,
      "learning_rate": 9.820815789473686e-05,
      "loss": 0.3796,
      "step": 26821
    },
    {
      "epoch": 1.5019599059245157,
      "grad_norm": 2.5908286571502686,
      "learning_rate": 9.820789473684211e-05,
      "loss": 0.5921,
      "step": 26822
    },
    {
      "epoch": 1.5020159032366447,
      "grad_norm": 1.4717801809310913,
      "learning_rate": 9.820763157894736e-05,
      "loss": 0.4755,
      "step": 26823
    },
    {
      "epoch": 1.5020719005487737,
      "grad_norm": 1.6030758619308472,
      "learning_rate": 9.820736842105263e-05,
      "loss": 0.5514,
      "step": 26824
    },
    {
      "epoch": 1.5021278978609027,
      "grad_norm": 1.283334493637085,
      "learning_rate": 9.82071052631579e-05,
      "loss": 0.5847,
      "step": 26825
    },
    {
      "epoch": 1.5021838951730317,
      "grad_norm": 1.3323065042495728,
      "learning_rate": 9.820684210526317e-05,
      "loss": 0.4857,
      "step": 26826
    },
    {
      "epoch": 1.5022398924851608,
      "grad_norm": 1.284218192100525,
      "learning_rate": 9.820657894736843e-05,
      "loss": 0.4279,
      "step": 26827
    },
    {
      "epoch": 1.5022958897972898,
      "grad_norm": 1.5113115310668945,
      "learning_rate": 9.820631578947369e-05,
      "loss": 0.4652,
      "step": 26828
    },
    {
      "epoch": 1.5023518871094188,
      "grad_norm": 1.7711883783340454,
      "learning_rate": 9.820605263157895e-05,
      "loss": 0.5026,
      "step": 26829
    },
    {
      "epoch": 1.5024078844215478,
      "grad_norm": 1.3830803632736206,
      "learning_rate": 9.820578947368422e-05,
      "loss": 0.4605,
      "step": 26830
    },
    {
      "epoch": 1.5024638817336768,
      "grad_norm": 1.2874218225479126,
      "learning_rate": 9.820552631578948e-05,
      "loss": 0.4895,
      "step": 26831
    },
    {
      "epoch": 1.5025198790458059,
      "grad_norm": 1.2400010824203491,
      "learning_rate": 9.820526315789474e-05,
      "loss": 0.3865,
      "step": 26832
    },
    {
      "epoch": 1.5025758763579349,
      "grad_norm": 1.394715666770935,
      "learning_rate": 9.8205e-05,
      "loss": 0.5973,
      "step": 26833
    },
    {
      "epoch": 1.502631873670064,
      "grad_norm": 1.2617863416671753,
      "learning_rate": 9.820473684210527e-05,
      "loss": 0.4719,
      "step": 26834
    },
    {
      "epoch": 1.502687870982193,
      "grad_norm": NaN,
      "learning_rate": 9.820473684210527e-05,
      "loss": 0.4233,
      "step": 26835
    },
    {
      "epoch": 1.502743868294322,
      "grad_norm": 1.1229867935180664,
      "learning_rate": 9.820447368421053e-05,
      "loss": 0.4105,
      "step": 26836
    },
    {
      "epoch": 1.502799865606451,
      "grad_norm": 1.460927963256836,
      "learning_rate": 9.820421052631579e-05,
      "loss": 0.5992,
      "step": 26837
    },
    {
      "epoch": 1.50285586291858,
      "grad_norm": 1.4900665283203125,
      "learning_rate": 9.820394736842105e-05,
      "loss": 0.5054,
      "step": 26838
    },
    {
      "epoch": 1.502911860230709,
      "grad_norm": 1.3573169708251953,
      "learning_rate": 9.820368421052632e-05,
      "loss": 0.5846,
      "step": 26839
    },
    {
      "epoch": 1.502967857542838,
      "grad_norm": 1.4125213623046875,
      "learning_rate": 9.820342105263158e-05,
      "loss": 0.4359,
      "step": 26840
    },
    {
      "epoch": 1.503023854854967,
      "grad_norm": 1.2597665786743164,
      "learning_rate": 9.820315789473684e-05,
      "loss": 0.4514,
      "step": 26841
    },
    {
      "epoch": 1.503079852167096,
      "grad_norm": 1.8424570560455322,
      "learning_rate": 9.82028947368421e-05,
      "loss": 0.6751,
      "step": 26842
    },
    {
      "epoch": 1.503135849479225,
      "grad_norm": 1.2865324020385742,
      "learning_rate": 9.820263157894736e-05,
      "loss": 0.4469,
      "step": 26843
    },
    {
      "epoch": 1.503191846791354,
      "grad_norm": 1.4428001642227173,
      "learning_rate": 9.820236842105264e-05,
      "loss": 0.5853,
      "step": 26844
    },
    {
      "epoch": 1.503247844103483,
      "grad_norm": 1.3592240810394287,
      "learning_rate": 9.82021052631579e-05,
      "loss": 0.4209,
      "step": 26845
    },
    {
      "epoch": 1.5033038414156121,
      "grad_norm": 1.2329879999160767,
      "learning_rate": 9.820184210526317e-05,
      "loss": 0.564,
      "step": 26846
    },
    {
      "epoch": 1.5033598387277411,
      "grad_norm": 1.3990342617034912,
      "learning_rate": 9.820157894736842e-05,
      "loss": 0.5239,
      "step": 26847
    },
    {
      "epoch": 1.5034158360398702,
      "grad_norm": 1.3297662734985352,
      "learning_rate": 9.820131578947369e-05,
      "loss": 0.4364,
      "step": 26848
    },
    {
      "epoch": 1.5034718333519992,
      "grad_norm": 1.827245831489563,
      "learning_rate": 9.820105263157895e-05,
      "loss": 0.5855,
      "step": 26849
    },
    {
      "epoch": 1.5035278306641282,
      "grad_norm": 1.5831985473632812,
      "learning_rate": 9.820078947368422e-05,
      "loss": 0.5982,
      "step": 26850
    },
    {
      "epoch": 1.5035838279762572,
      "grad_norm": 1.1328448057174683,
      "learning_rate": 9.820052631578948e-05,
      "loss": 0.4677,
      "step": 26851
    },
    {
      "epoch": 1.5036398252883862,
      "grad_norm": 1.227616310119629,
      "learning_rate": 9.820026315789474e-05,
      "loss": 0.5185,
      "step": 26852
    },
    {
      "epoch": 1.5036958226005153,
      "grad_norm": 1.471461296081543,
      "learning_rate": 9.82e-05,
      "loss": 0.4526,
      "step": 26853
    },
    {
      "epoch": 1.5037518199126443,
      "grad_norm": 1.071116328239441,
      "learning_rate": 9.819973684210527e-05,
      "loss": 0.3713,
      "step": 26854
    },
    {
      "epoch": 1.5038078172247733,
      "grad_norm": 1.463884949684143,
      "learning_rate": 9.819947368421053e-05,
      "loss": 0.5586,
      "step": 26855
    },
    {
      "epoch": 1.5038638145369023,
      "grad_norm": 1.183410406112671,
      "learning_rate": 9.81992105263158e-05,
      "loss": 0.4291,
      "step": 26856
    },
    {
      "epoch": 1.5039198118490313,
      "grad_norm": 1.2246249914169312,
      "learning_rate": 9.819894736842105e-05,
      "loss": 0.4141,
      "step": 26857
    },
    {
      "epoch": 1.5039758091611604,
      "grad_norm": 1.3548663854599,
      "learning_rate": 9.819868421052631e-05,
      "loss": 0.6363,
      "step": 26858
    },
    {
      "epoch": 1.5040318064732894,
      "grad_norm": 1.112684726715088,
      "learning_rate": 9.819842105263159e-05,
      "loss": 0.4017,
      "step": 26859
    },
    {
      "epoch": 1.5040878037854184,
      "grad_norm": 1.1287680864334106,
      "learning_rate": 9.819815789473685e-05,
      "loss": 0.3189,
      "step": 26860
    },
    {
      "epoch": 1.5041438010975474,
      "grad_norm": 1.3464311361312866,
      "learning_rate": 9.81978947368421e-05,
      "loss": 0.4012,
      "step": 26861
    },
    {
      "epoch": 1.5041997984096764,
      "grad_norm": 1.1868360042572021,
      "learning_rate": 9.819763157894737e-05,
      "loss": 0.3788,
      "step": 26862
    },
    {
      "epoch": 1.5042557957218055,
      "grad_norm": 1.2991905212402344,
      "learning_rate": 9.819736842105264e-05,
      "loss": 0.4808,
      "step": 26863
    },
    {
      "epoch": 1.5043117930339345,
      "grad_norm": 1.4344667196273804,
      "learning_rate": 9.81971052631579e-05,
      "loss": 0.5092,
      "step": 26864
    },
    {
      "epoch": 1.5043677903460635,
      "grad_norm": 1.2147564888000488,
      "learning_rate": 9.819684210526316e-05,
      "loss": 0.4904,
      "step": 26865
    },
    {
      "epoch": 1.5044237876581925,
      "grad_norm": 1.4568026065826416,
      "learning_rate": 9.819657894736842e-05,
      "loss": 0.5145,
      "step": 26866
    },
    {
      "epoch": 1.5044797849703215,
      "grad_norm": 1.2571946382522583,
      "learning_rate": 9.819631578947369e-05,
      "loss": 0.4085,
      "step": 26867
    },
    {
      "epoch": 1.5045357822824506,
      "grad_norm": 1.4784843921661377,
      "learning_rate": 9.819605263157895e-05,
      "loss": 0.5063,
      "step": 26868
    },
    {
      "epoch": 1.5045917795945796,
      "grad_norm": 1.479533314704895,
      "learning_rate": 9.819578947368423e-05,
      "loss": 0.4835,
      "step": 26869
    },
    {
      "epoch": 1.5046477769067086,
      "grad_norm": 1.1925956010818481,
      "learning_rate": 9.819552631578947e-05,
      "loss": 0.4318,
      "step": 26870
    },
    {
      "epoch": 1.5047037742188376,
      "grad_norm": 1.4827125072479248,
      "learning_rate": 9.819526315789474e-05,
      "loss": 0.64,
      "step": 26871
    },
    {
      "epoch": 1.5047597715309666,
      "grad_norm": 1.2384952306747437,
      "learning_rate": 9.8195e-05,
      "loss": 0.4034,
      "step": 26872
    },
    {
      "epoch": 1.5048157688430956,
      "grad_norm": 1.070129156112671,
      "learning_rate": 9.819473684210528e-05,
      "loss": 0.3833,
      "step": 26873
    },
    {
      "epoch": 1.5048717661552247,
      "grad_norm": 1.3075287342071533,
      "learning_rate": 9.819447368421054e-05,
      "loss": 0.4432,
      "step": 26874
    },
    {
      "epoch": 1.5049277634673537,
      "grad_norm": 1.3838526010513306,
      "learning_rate": 9.81942105263158e-05,
      "loss": 0.4524,
      "step": 26875
    },
    {
      "epoch": 1.5049837607794827,
      "grad_norm": 1.6466572284698486,
      "learning_rate": 9.819394736842106e-05,
      "loss": 0.5443,
      "step": 26876
    },
    {
      "epoch": 1.5050397580916117,
      "grad_norm": 1.976815104484558,
      "learning_rate": 9.819368421052632e-05,
      "loss": 0.5765,
      "step": 26877
    },
    {
      "epoch": 1.5050957554037407,
      "grad_norm": 1.3899425268173218,
      "learning_rate": 9.819342105263159e-05,
      "loss": 0.4691,
      "step": 26878
    },
    {
      "epoch": 1.5051517527158698,
      "grad_norm": 1.2887729406356812,
      "learning_rate": 9.819315789473684e-05,
      "loss": 0.5758,
      "step": 26879
    },
    {
      "epoch": 1.5052077500279988,
      "grad_norm": 1.201016902923584,
      "learning_rate": 9.819289473684211e-05,
      "loss": 0.3569,
      "step": 26880
    },
    {
      "epoch": 1.5052637473401278,
      "grad_norm": 1.3866976499557495,
      "learning_rate": 9.819263157894737e-05,
      "loss": 0.4409,
      "step": 26881
    },
    {
      "epoch": 1.5053197446522568,
      "grad_norm": 1.2868051528930664,
      "learning_rate": 9.819236842105264e-05,
      "loss": 0.4047,
      "step": 26882
    },
    {
      "epoch": 1.5053757419643858,
      "grad_norm": 2.1382479667663574,
      "learning_rate": 9.81921052631579e-05,
      "loss": 0.4798,
      "step": 26883
    },
    {
      "epoch": 1.5054317392765149,
      "grad_norm": 1.374638319015503,
      "learning_rate": 9.819184210526316e-05,
      "loss": 0.439,
      "step": 26884
    },
    {
      "epoch": 1.5054877365886439,
      "grad_norm": 1.6410070657730103,
      "learning_rate": 9.819157894736842e-05,
      "loss": 0.5119,
      "step": 26885
    },
    {
      "epoch": 1.505543733900773,
      "grad_norm": 1.246012568473816,
      "learning_rate": 9.81913157894737e-05,
      "loss": 0.4355,
      "step": 26886
    },
    {
      "epoch": 1.505599731212902,
      "grad_norm": 1.2205312252044678,
      "learning_rate": 9.819105263157895e-05,
      "loss": 0.4992,
      "step": 26887
    },
    {
      "epoch": 1.505655728525031,
      "grad_norm": 1.5243549346923828,
      "learning_rate": 9.819078947368421e-05,
      "loss": 0.5899,
      "step": 26888
    },
    {
      "epoch": 1.50571172583716,
      "grad_norm": 1.6237629652023315,
      "learning_rate": 9.819052631578947e-05,
      "loss": 0.4594,
      "step": 26889
    },
    {
      "epoch": 1.505767723149289,
      "grad_norm": 1.3966513872146606,
      "learning_rate": 9.819026315789475e-05,
      "loss": 0.4631,
      "step": 26890
    },
    {
      "epoch": 1.505823720461418,
      "grad_norm": 1.6498286724090576,
      "learning_rate": 9.819000000000001e-05,
      "loss": 0.4875,
      "step": 26891
    },
    {
      "epoch": 1.505879717773547,
      "grad_norm": 1.6096594333648682,
      "learning_rate": 9.818973684210527e-05,
      "loss": 0.3638,
      "step": 26892
    },
    {
      "epoch": 1.505935715085676,
      "grad_norm": 1.3800125122070312,
      "learning_rate": 9.818947368421053e-05,
      "loss": 0.4304,
      "step": 26893
    },
    {
      "epoch": 1.505991712397805,
      "grad_norm": 1.5419855117797852,
      "learning_rate": 9.818921052631579e-05,
      "loss": 0.4609,
      "step": 26894
    },
    {
      "epoch": 1.506047709709934,
      "grad_norm": 1.3288782835006714,
      "learning_rate": 9.818894736842106e-05,
      "loss": 0.4868,
      "step": 26895
    },
    {
      "epoch": 1.506103707022063,
      "grad_norm": 1.3769283294677734,
      "learning_rate": 9.818868421052632e-05,
      "loss": 0.4556,
      "step": 26896
    },
    {
      "epoch": 1.506159704334192,
      "grad_norm": 1.2915641069412231,
      "learning_rate": 9.818842105263158e-05,
      "loss": 0.513,
      "step": 26897
    },
    {
      "epoch": 1.5062157016463211,
      "grad_norm": 1.331145167350769,
      "learning_rate": 9.818815789473684e-05,
      "loss": 0.5951,
      "step": 26898
    },
    {
      "epoch": 1.5062716989584501,
      "grad_norm": 1.5310689210891724,
      "learning_rate": 9.818789473684211e-05,
      "loss": 0.5239,
      "step": 26899
    },
    {
      "epoch": 1.5063276962705792,
      "grad_norm": 1.2608551979064941,
      "learning_rate": 9.818763157894737e-05,
      "loss": 0.4522,
      "step": 26900
    },
    {
      "epoch": 1.5063836935827082,
      "grad_norm": 1.5724152326583862,
      "learning_rate": 9.818736842105264e-05,
      "loss": 0.5246,
      "step": 26901
    },
    {
      "epoch": 1.5064396908948372,
      "grad_norm": 1.3005868196487427,
      "learning_rate": 9.818710526315789e-05,
      "loss": 0.379,
      "step": 26902
    },
    {
      "epoch": 1.5064956882069662,
      "grad_norm": 1.192333459854126,
      "learning_rate": 9.818684210526316e-05,
      "loss": 0.3715,
      "step": 26903
    },
    {
      "epoch": 1.5065516855190952,
      "grad_norm": 1.3528265953063965,
      "learning_rate": 9.818657894736842e-05,
      "loss": 0.3949,
      "step": 26904
    },
    {
      "epoch": 1.5066076828312243,
      "grad_norm": 1.7287955284118652,
      "learning_rate": 9.81863157894737e-05,
      "loss": 0.3794,
      "step": 26905
    },
    {
      "epoch": 1.5066636801433533,
      "grad_norm": 2.0118093490600586,
      "learning_rate": 9.818605263157896e-05,
      "loss": 0.569,
      "step": 26906
    },
    {
      "epoch": 1.5067196774554823,
      "grad_norm": 1.28496515750885,
      "learning_rate": 9.818578947368422e-05,
      "loss": 0.4252,
      "step": 26907
    },
    {
      "epoch": 1.5067756747676113,
      "grad_norm": 1.237634301185608,
      "learning_rate": 9.818552631578948e-05,
      "loss": 0.4065,
      "step": 26908
    },
    {
      "epoch": 1.5068316720797403,
      "grad_norm": 1.317826509475708,
      "learning_rate": 9.818526315789475e-05,
      "loss": 0.4356,
      "step": 26909
    },
    {
      "epoch": 1.5068876693918694,
      "grad_norm": 1.404542326927185,
      "learning_rate": 9.818500000000001e-05,
      "loss": 0.5111,
      "step": 26910
    },
    {
      "epoch": 1.5069436667039984,
      "grad_norm": 1.5804247856140137,
      "learning_rate": 9.818473684210527e-05,
      "loss": 0.4187,
      "step": 26911
    },
    {
      "epoch": 1.5069996640161272,
      "grad_norm": 1.4498772621154785,
      "learning_rate": 9.818447368421053e-05,
      "loss": 0.5616,
      "step": 26912
    },
    {
      "epoch": 1.5070556613282562,
      "grad_norm": 1.2389779090881348,
      "learning_rate": 9.818421052631579e-05,
      "loss": 0.4663,
      "step": 26913
    },
    {
      "epoch": 1.5071116586403852,
      "grad_norm": 1.4195177555084229,
      "learning_rate": 9.818394736842106e-05,
      "loss": 0.5183,
      "step": 26914
    },
    {
      "epoch": 1.5071676559525142,
      "grad_norm": 1.356787919998169,
      "learning_rate": 9.818368421052632e-05,
      "loss": 0.3849,
      "step": 26915
    },
    {
      "epoch": 1.5072236532646432,
      "grad_norm": 1.1957062482833862,
      "learning_rate": 9.818342105263158e-05,
      "loss": 0.402,
      "step": 26916
    },
    {
      "epoch": 1.5072796505767723,
      "grad_norm": 1.26849365234375,
      "learning_rate": 9.818315789473684e-05,
      "loss": 0.538,
      "step": 26917
    },
    {
      "epoch": 1.5073356478889013,
      "grad_norm": 1.2174674272537231,
      "learning_rate": 9.818289473684211e-05,
      "loss": 0.4348,
      "step": 26918
    },
    {
      "epoch": 1.5073916452010303,
      "grad_norm": 1.4033805131912231,
      "learning_rate": 9.818263157894737e-05,
      "loss": 0.3808,
      "step": 26919
    },
    {
      "epoch": 1.5074476425131593,
      "grad_norm": 1.1879520416259766,
      "learning_rate": 9.818236842105263e-05,
      "loss": 0.5138,
      "step": 26920
    },
    {
      "epoch": 1.5075036398252883,
      "grad_norm": 1.3620471954345703,
      "learning_rate": 9.81821052631579e-05,
      "loss": 0.4985,
      "step": 26921
    },
    {
      "epoch": 1.5075596371374174,
      "grad_norm": 1.1453810930252075,
      "learning_rate": 9.818184210526317e-05,
      "loss": 0.4237,
      "step": 26922
    },
    {
      "epoch": 1.5076156344495464,
      "grad_norm": 18.394256591796875,
      "learning_rate": 9.818157894736843e-05,
      "loss": 0.5131,
      "step": 26923
    },
    {
      "epoch": 1.5076716317616754,
      "grad_norm": 1.3331096172332764,
      "learning_rate": 9.81813157894737e-05,
      "loss": 0.4732,
      "step": 26924
    },
    {
      "epoch": 1.5077276290738044,
      "grad_norm": 1.2102887630462646,
      "learning_rate": 9.818105263157895e-05,
      "loss": 0.4682,
      "step": 26925
    },
    {
      "epoch": 1.5077836263859334,
      "grad_norm": 1.5214927196502686,
      "learning_rate": 9.818078947368422e-05,
      "loss": 0.4753,
      "step": 26926
    },
    {
      "epoch": 1.5078396236980625,
      "grad_norm": 1.2080925703048706,
      "learning_rate": 9.818052631578948e-05,
      "loss": 0.4148,
      "step": 26927
    },
    {
      "epoch": 1.5078956210101915,
      "grad_norm": 7.116842269897461,
      "learning_rate": 9.818026315789474e-05,
      "loss": 0.4787,
      "step": 26928
    },
    {
      "epoch": 1.5079516183223205,
      "grad_norm": 1.38553786277771,
      "learning_rate": 9.818000000000001e-05,
      "loss": 0.4541,
      "step": 26929
    },
    {
      "epoch": 1.5080076156344495,
      "grad_norm": 3.073695659637451,
      "learning_rate": 9.817973684210526e-05,
      "loss": 0.4254,
      "step": 26930
    },
    {
      "epoch": 1.5080636129465785,
      "grad_norm": 1.1337478160858154,
      "learning_rate": 9.817947368421053e-05,
      "loss": 0.3835,
      "step": 26931
    },
    {
      "epoch": 1.5081196102587076,
      "grad_norm": 1.1407872438430786,
      "learning_rate": 9.817921052631579e-05,
      "loss": 0.3578,
      "step": 26932
    },
    {
      "epoch": 1.5081756075708366,
      "grad_norm": 1.1704356670379639,
      "learning_rate": 9.817894736842106e-05,
      "loss": 0.3821,
      "step": 26933
    },
    {
      "epoch": 1.5082316048829656,
      "grad_norm": 1.244095802307129,
      "learning_rate": 9.817868421052631e-05,
      "loss": 0.4444,
      "step": 26934
    },
    {
      "epoch": 1.5082876021950946,
      "grad_norm": 1.176902413368225,
      "learning_rate": 9.817842105263158e-05,
      "loss": 0.389,
      "step": 26935
    },
    {
      "epoch": 1.5083435995072236,
      "grad_norm": 1.5219768285751343,
      "learning_rate": 9.817815789473684e-05,
      "loss": 0.4255,
      "step": 26936
    },
    {
      "epoch": 1.5083995968193527,
      "grad_norm": 1.2690356969833374,
      "learning_rate": 9.817789473684212e-05,
      "loss": 0.4926,
      "step": 26937
    },
    {
      "epoch": 1.5084555941314817,
      "grad_norm": 1.209762454032898,
      "learning_rate": 9.817763157894738e-05,
      "loss": 0.4703,
      "step": 26938
    },
    {
      "epoch": 1.5085115914436107,
      "grad_norm": 1.5581369400024414,
      "learning_rate": 9.817736842105264e-05,
      "loss": 0.4605,
      "step": 26939
    },
    {
      "epoch": 1.5085675887557397,
      "grad_norm": 1.1573799848556519,
      "learning_rate": 9.81771052631579e-05,
      "loss": 0.4106,
      "step": 26940
    },
    {
      "epoch": 1.5086235860678687,
      "grad_norm": 1.1588919162750244,
      "learning_rate": 9.817684210526317e-05,
      "loss": 0.5242,
      "step": 26941
    },
    {
      "epoch": 1.5086795833799977,
      "grad_norm": 1.4145991802215576,
      "learning_rate": 9.817657894736843e-05,
      "loss": 0.4363,
      "step": 26942
    },
    {
      "epoch": 1.5087355806921268,
      "grad_norm": 1.0535063743591309,
      "learning_rate": 9.817631578947369e-05,
      "loss": 0.3595,
      "step": 26943
    },
    {
      "epoch": 1.5087915780042558,
      "grad_norm": 1.281191349029541,
      "learning_rate": 9.817605263157895e-05,
      "loss": 0.4164,
      "step": 26944
    },
    {
      "epoch": 1.5088475753163848,
      "grad_norm": 1.3760112524032593,
      "learning_rate": 9.817578947368421e-05,
      "loss": 0.459,
      "step": 26945
    },
    {
      "epoch": 1.5089035726285138,
      "grad_norm": 1.3487932682037354,
      "learning_rate": 9.817552631578948e-05,
      "loss": 0.4675,
      "step": 26946
    },
    {
      "epoch": 1.5089595699406428,
      "grad_norm": 1.274167776107788,
      "learning_rate": 9.817526315789474e-05,
      "loss": 0.4668,
      "step": 26947
    },
    {
      "epoch": 1.5090155672527719,
      "grad_norm": 1.2175053358078003,
      "learning_rate": 9.8175e-05,
      "loss": 0.6446,
      "step": 26948
    },
    {
      "epoch": 1.5090715645649009,
      "grad_norm": 1.1962106227874756,
      "learning_rate": 9.817473684210526e-05,
      "loss": 0.4663,
      "step": 26949
    },
    {
      "epoch": 1.50912756187703,
      "grad_norm": 7.010622024536133,
      "learning_rate": 9.817447368421053e-05,
      "loss": 0.2958,
      "step": 26950
    },
    {
      "epoch": 1.509183559189159,
      "grad_norm": 1.4108219146728516,
      "learning_rate": 9.81742105263158e-05,
      "loss": 0.4266,
      "step": 26951
    },
    {
      "epoch": 1.509239556501288,
      "grad_norm": 1.3778231143951416,
      "learning_rate": 9.817394736842105e-05,
      "loss": 0.5832,
      "step": 26952
    },
    {
      "epoch": 1.509295553813417,
      "grad_norm": 1.2143008708953857,
      "learning_rate": 9.817368421052631e-05,
      "loss": 0.49,
      "step": 26953
    },
    {
      "epoch": 1.509351551125546,
      "grad_norm": 1.4713767766952515,
      "learning_rate": 9.817342105263159e-05,
      "loss": 0.4437,
      "step": 26954
    },
    {
      "epoch": 1.509407548437675,
      "grad_norm": 1.3192007541656494,
      "learning_rate": 9.817315789473685e-05,
      "loss": 0.4068,
      "step": 26955
    },
    {
      "epoch": 1.509463545749804,
      "grad_norm": 1.194394826889038,
      "learning_rate": 9.817289473684212e-05,
      "loss": 0.4051,
      "step": 26956
    },
    {
      "epoch": 1.509519543061933,
      "grad_norm": 1.1465498208999634,
      "learning_rate": 9.817263157894737e-05,
      "loss": 0.4345,
      "step": 26957
    },
    {
      "epoch": 1.509575540374062,
      "grad_norm": 1.186548113822937,
      "learning_rate": 9.817236842105264e-05,
      "loss": 0.4653,
      "step": 26958
    },
    {
      "epoch": 1.509631537686191,
      "grad_norm": 1.673143744468689,
      "learning_rate": 9.81721052631579e-05,
      "loss": 0.8719,
      "step": 26959
    },
    {
      "epoch": 1.50968753499832,
      "grad_norm": 1.4337735176086426,
      "learning_rate": 9.817184210526317e-05,
      "loss": 0.4798,
      "step": 26960
    },
    {
      "epoch": 1.509743532310449,
      "grad_norm": 1.67737877368927,
      "learning_rate": 9.817157894736843e-05,
      "loss": 0.5259,
      "step": 26961
    },
    {
      "epoch": 1.5097995296225781,
      "grad_norm": 1.0275202989578247,
      "learning_rate": 9.817131578947368e-05,
      "loss": 0.4168,
      "step": 26962
    },
    {
      "epoch": 1.5098555269347071,
      "grad_norm": 1.2809327840805054,
      "learning_rate": 9.817105263157895e-05,
      "loss": 0.3889,
      "step": 26963
    },
    {
      "epoch": 1.5099115242468362,
      "grad_norm": 1.1116302013397217,
      "learning_rate": 9.817078947368421e-05,
      "loss": 0.4027,
      "step": 26964
    },
    {
      "epoch": 1.5099675215589652,
      "grad_norm": 1.626907229423523,
      "learning_rate": 9.817052631578948e-05,
      "loss": 0.5272,
      "step": 26965
    },
    {
      "epoch": 1.5100235188710942,
      "grad_norm": 1.2781729698181152,
      "learning_rate": 9.817026315789474e-05,
      "loss": 0.5919,
      "step": 26966
    },
    {
      "epoch": 1.5100795161832232,
      "grad_norm": 1.2536100149154663,
      "learning_rate": 9.817e-05,
      "loss": 0.3703,
      "step": 26967
    },
    {
      "epoch": 1.5101355134953522,
      "grad_norm": 1.2065856456756592,
      "learning_rate": 9.816973684210526e-05,
      "loss": 0.4474,
      "step": 26968
    },
    {
      "epoch": 1.5101915108074813,
      "grad_norm": 1.3996970653533936,
      "learning_rate": 9.816947368421054e-05,
      "loss": 0.4775,
      "step": 26969
    },
    {
      "epoch": 1.5102475081196103,
      "grad_norm": 1.4324966669082642,
      "learning_rate": 9.81692105263158e-05,
      "loss": 0.4612,
      "step": 26970
    },
    {
      "epoch": 1.5103035054317393,
      "grad_norm": 1.1891260147094727,
      "learning_rate": 9.816894736842106e-05,
      "loss": 0.5276,
      "step": 26971
    },
    {
      "epoch": 1.5103595027438683,
      "grad_norm": 1.230858564376831,
      "learning_rate": 9.816868421052632e-05,
      "loss": 0.592,
      "step": 26972
    },
    {
      "epoch": 1.5104155000559973,
      "grad_norm": 1.4232933521270752,
      "learning_rate": 9.816842105263159e-05,
      "loss": 0.499,
      "step": 26973
    },
    {
      "epoch": 1.5104714973681264,
      "grad_norm": 1.3536614179611206,
      "learning_rate": 9.816815789473685e-05,
      "loss": 0.3909,
      "step": 26974
    },
    {
      "epoch": 1.5105274946802554,
      "grad_norm": 2.4492759704589844,
      "learning_rate": 9.816789473684211e-05,
      "loss": 0.5383,
      "step": 26975
    },
    {
      "epoch": 1.5105834919923844,
      "grad_norm": 1.141347885131836,
      "learning_rate": 9.816763157894737e-05,
      "loss": 0.4357,
      "step": 26976
    },
    {
      "epoch": 1.5106394893045134,
      "grad_norm": 1.2195390462875366,
      "learning_rate": 9.816736842105264e-05,
      "loss": 0.5351,
      "step": 26977
    },
    {
      "epoch": 1.5106954866166424,
      "grad_norm": 1.432437539100647,
      "learning_rate": 9.81671052631579e-05,
      "loss": 0.6622,
      "step": 26978
    },
    {
      "epoch": 1.5107514839287715,
      "grad_norm": 1.1447179317474365,
      "learning_rate": 9.816684210526316e-05,
      "loss": 0.3834,
      "step": 26979
    },
    {
      "epoch": 1.5108074812409005,
      "grad_norm": 1.3678417205810547,
      "learning_rate": 9.816657894736842e-05,
      "loss": 0.5131,
      "step": 26980
    },
    {
      "epoch": 1.5108634785530295,
      "grad_norm": 1.1029322147369385,
      "learning_rate": 9.816631578947368e-05,
      "loss": 0.4077,
      "step": 26981
    },
    {
      "epoch": 1.5109194758651585,
      "grad_norm": 1.260974645614624,
      "learning_rate": 9.816605263157895e-05,
      "loss": 0.5159,
      "step": 26982
    },
    {
      "epoch": 1.5109754731772875,
      "grad_norm": 1.2160358428955078,
      "learning_rate": 9.816578947368421e-05,
      "loss": 0.4207,
      "step": 26983
    },
    {
      "epoch": 1.5110314704894166,
      "grad_norm": 1.2865164279937744,
      "learning_rate": 9.816552631578947e-05,
      "loss": 0.4125,
      "step": 26984
    },
    {
      "epoch": 1.5110874678015456,
      "grad_norm": 1.6537671089172363,
      "learning_rate": 9.816526315789473e-05,
      "loss": 0.6076,
      "step": 26985
    },
    {
      "epoch": 1.5111434651136746,
      "grad_norm": 1.1508857011795044,
      "learning_rate": 9.8165e-05,
      "loss": 0.4455,
      "step": 26986
    },
    {
      "epoch": 1.5111994624258036,
      "grad_norm": 1.6039947271347046,
      "learning_rate": 9.816473684210527e-05,
      "loss": 0.6007,
      "step": 26987
    },
    {
      "epoch": 1.5112554597379326,
      "grad_norm": 1.6592415571212769,
      "learning_rate": 9.816447368421054e-05,
      "loss": 0.5268,
      "step": 26988
    },
    {
      "epoch": 1.5113114570500616,
      "grad_norm": 1.210771083831787,
      "learning_rate": 9.816421052631579e-05,
      "loss": 0.4452,
      "step": 26989
    },
    {
      "epoch": 1.5113674543621907,
      "grad_norm": 1.40608549118042,
      "learning_rate": 9.816394736842106e-05,
      "loss": 0.4988,
      "step": 26990
    },
    {
      "epoch": 1.5114234516743197,
      "grad_norm": 2.0569067001342773,
      "learning_rate": 9.816368421052632e-05,
      "loss": 0.5539,
      "step": 26991
    },
    {
      "epoch": 1.5114794489864487,
      "grad_norm": 2.3576831817626953,
      "learning_rate": 9.816342105263159e-05,
      "loss": 0.6229,
      "step": 26992
    },
    {
      "epoch": 1.5115354462985777,
      "grad_norm": 1.2888567447662354,
      "learning_rate": 9.816315789473685e-05,
      "loss": 0.4938,
      "step": 26993
    },
    {
      "epoch": 1.5115914436107065,
      "grad_norm": 1.8300138711929321,
      "learning_rate": 9.816289473684211e-05,
      "loss": 0.6033,
      "step": 26994
    },
    {
      "epoch": 1.5116474409228355,
      "grad_norm": 1.290547251701355,
      "learning_rate": 9.816263157894737e-05,
      "loss": 0.4041,
      "step": 26995
    },
    {
      "epoch": 1.5117034382349646,
      "grad_norm": 1.5086348056793213,
      "learning_rate": 9.816236842105264e-05,
      "loss": 0.5601,
      "step": 26996
    },
    {
      "epoch": 1.5117594355470936,
      "grad_norm": 1.2090970277786255,
      "learning_rate": 9.81621052631579e-05,
      "loss": 0.4098,
      "step": 26997
    },
    {
      "epoch": 1.5118154328592226,
      "grad_norm": 1.125792384147644,
      "learning_rate": 9.816184210526316e-05,
      "loss": 0.3752,
      "step": 26998
    },
    {
      "epoch": 1.5118714301713516,
      "grad_norm": 1.3590054512023926,
      "learning_rate": 9.816157894736842e-05,
      "loss": 0.4998,
      "step": 26999
    },
    {
      "epoch": 1.5119274274834806,
      "grad_norm": 1.4590128660202026,
      "learning_rate": 9.816131578947368e-05,
      "loss": 0.4472,
      "step": 27000
    },
    {
      "epoch": 1.5119834247956097,
      "grad_norm": 1.3467227220535278,
      "learning_rate": 9.816105263157896e-05,
      "loss": 0.5055,
      "step": 27001
    },
    {
      "epoch": 1.5120394221077387,
      "grad_norm": 1.2745952606201172,
      "learning_rate": 9.816078947368422e-05,
      "loss": 0.5036,
      "step": 27002
    },
    {
      "epoch": 1.5120954194198677,
      "grad_norm": 1.2922078371047974,
      "learning_rate": 9.816052631578948e-05,
      "loss": 0.5639,
      "step": 27003
    },
    {
      "epoch": 1.5121514167319967,
      "grad_norm": 1.372616171836853,
      "learning_rate": 9.816026315789474e-05,
      "loss": 0.5246,
      "step": 27004
    },
    {
      "epoch": 1.5122074140441257,
      "grad_norm": 1.3429547548294067,
      "learning_rate": 9.816000000000001e-05,
      "loss": 0.419,
      "step": 27005
    },
    {
      "epoch": 1.5122634113562547,
      "grad_norm": 1.0865153074264526,
      "learning_rate": 9.815973684210527e-05,
      "loss": 0.3538,
      "step": 27006
    },
    {
      "epoch": 1.5123194086683838,
      "grad_norm": 1.3962829113006592,
      "learning_rate": 9.815947368421053e-05,
      "loss": 0.454,
      "step": 27007
    },
    {
      "epoch": 1.5123754059805128,
      "grad_norm": 1.2971525192260742,
      "learning_rate": 9.815921052631579e-05,
      "loss": 0.4568,
      "step": 27008
    },
    {
      "epoch": 1.5124314032926418,
      "grad_norm": 3.0007100105285645,
      "learning_rate": 9.815894736842106e-05,
      "loss": 0.3183,
      "step": 27009
    },
    {
      "epoch": 1.5124874006047708,
      "grad_norm": 1.372809648513794,
      "learning_rate": 9.815868421052632e-05,
      "loss": 0.5886,
      "step": 27010
    },
    {
      "epoch": 1.5125433979168998,
      "grad_norm": 1.3756669759750366,
      "learning_rate": 9.81584210526316e-05,
      "loss": 0.4924,
      "step": 27011
    },
    {
      "epoch": 1.5125993952290289,
      "grad_norm": 1.1371204853057861,
      "learning_rate": 9.815815789473684e-05,
      "loss": 0.3745,
      "step": 27012
    },
    {
      "epoch": 1.5126553925411579,
      "grad_norm": 1.4544603824615479,
      "learning_rate": 9.815789473684211e-05,
      "loss": 0.4664,
      "step": 27013
    },
    {
      "epoch": 1.512711389853287,
      "grad_norm": 1.2304860353469849,
      "learning_rate": 9.815763157894737e-05,
      "loss": 0.4754,
      "step": 27014
    },
    {
      "epoch": 1.512767387165416,
      "grad_norm": 1.2915074825286865,
      "learning_rate": 9.815736842105263e-05,
      "loss": 0.4477,
      "step": 27015
    },
    {
      "epoch": 1.512823384477545,
      "grad_norm": 2.008554697036743,
      "learning_rate": 9.81571052631579e-05,
      "loss": 0.7653,
      "step": 27016
    },
    {
      "epoch": 1.512879381789674,
      "grad_norm": 1.2905584573745728,
      "learning_rate": 9.815684210526315e-05,
      "loss": 0.4139,
      "step": 27017
    },
    {
      "epoch": 1.512935379101803,
      "grad_norm": 1.3069546222686768,
      "learning_rate": 9.815657894736843e-05,
      "loss": 0.5049,
      "step": 27018
    },
    {
      "epoch": 1.512991376413932,
      "grad_norm": 1.5047800540924072,
      "learning_rate": 9.815631578947369e-05,
      "loss": 0.4656,
      "step": 27019
    },
    {
      "epoch": 1.513047373726061,
      "grad_norm": 1.1808758974075317,
      "learning_rate": 9.815605263157896e-05,
      "loss": 0.3925,
      "step": 27020
    },
    {
      "epoch": 1.51310337103819,
      "grad_norm": 1.5783368349075317,
      "learning_rate": 9.815578947368422e-05,
      "loss": 0.4801,
      "step": 27021
    },
    {
      "epoch": 1.513159368350319,
      "grad_norm": 1.9102554321289062,
      "learning_rate": 9.815552631578948e-05,
      "loss": 0.5058,
      "step": 27022
    },
    {
      "epoch": 1.513215365662448,
      "grad_norm": 1.4686520099639893,
      "learning_rate": 9.815526315789474e-05,
      "loss": 0.5082,
      "step": 27023
    },
    {
      "epoch": 1.513271362974577,
      "grad_norm": 1.7134006023406982,
      "learning_rate": 9.815500000000001e-05,
      "loss": 0.5512,
      "step": 27024
    },
    {
      "epoch": 1.5133273602867061,
      "grad_norm": 1.4265936613082886,
      "learning_rate": 9.815473684210527e-05,
      "loss": 0.5363,
      "step": 27025
    },
    {
      "epoch": 1.5133833575988351,
      "grad_norm": 2.0115835666656494,
      "learning_rate": 9.815447368421053e-05,
      "loss": 0.5573,
      "step": 27026
    },
    {
      "epoch": 1.5134393549109642,
      "grad_norm": 1.341893196105957,
      "learning_rate": 9.815421052631579e-05,
      "loss": 0.5759,
      "step": 27027
    },
    {
      "epoch": 1.5134953522230932,
      "grad_norm": 4.912156581878662,
      "learning_rate": 9.815394736842106e-05,
      "loss": 0.5019,
      "step": 27028
    },
    {
      "epoch": 1.5135513495352222,
      "grad_norm": 1.7848292589187622,
      "learning_rate": 9.815368421052632e-05,
      "loss": 0.5088,
      "step": 27029
    },
    {
      "epoch": 1.5136073468473512,
      "grad_norm": 2.5112762451171875,
      "learning_rate": 9.815342105263158e-05,
      "loss": 0.5038,
      "step": 27030
    },
    {
      "epoch": 1.5136633441594802,
      "grad_norm": 1.1688530445098877,
      "learning_rate": 9.815315789473684e-05,
      "loss": 0.437,
      "step": 27031
    },
    {
      "epoch": 1.5137193414716092,
      "grad_norm": 1.1913703680038452,
      "learning_rate": 9.81528947368421e-05,
      "loss": 0.5217,
      "step": 27032
    },
    {
      "epoch": 1.5137753387837383,
      "grad_norm": 1.3893768787384033,
      "learning_rate": 9.815263157894738e-05,
      "loss": 0.4392,
      "step": 27033
    },
    {
      "epoch": 1.5138313360958673,
      "grad_norm": 1.5970072746276855,
      "learning_rate": 9.815236842105264e-05,
      "loss": 0.5538,
      "step": 27034
    },
    {
      "epoch": 1.5138873334079963,
      "grad_norm": 1.2803162336349487,
      "learning_rate": 9.81521052631579e-05,
      "loss": 0.4452,
      "step": 27035
    },
    {
      "epoch": 1.5139433307201253,
      "grad_norm": 1.2993170022964478,
      "learning_rate": 9.815184210526316e-05,
      "loss": 0.5912,
      "step": 27036
    },
    {
      "epoch": 1.5139993280322543,
      "grad_norm": 1.235471487045288,
      "learning_rate": 9.815157894736843e-05,
      "loss": 0.5795,
      "step": 27037
    },
    {
      "epoch": 1.5140553253443834,
      "grad_norm": 1.541257619857788,
      "learning_rate": 9.815131578947369e-05,
      "loss": 0.5014,
      "step": 27038
    },
    {
      "epoch": 1.5141113226565124,
      "grad_norm": 1.2752493619918823,
      "learning_rate": 9.815105263157895e-05,
      "loss": 0.4493,
      "step": 27039
    },
    {
      "epoch": 1.5141673199686414,
      "grad_norm": 1.2818028926849365,
      "learning_rate": 9.815078947368421e-05,
      "loss": 0.5289,
      "step": 27040
    },
    {
      "epoch": 1.5142233172807704,
      "grad_norm": 1.458085060119629,
      "learning_rate": 9.815052631578948e-05,
      "loss": 0.5055,
      "step": 27041
    },
    {
      "epoch": 1.5142793145928994,
      "grad_norm": 1.3541269302368164,
      "learning_rate": 9.815026315789474e-05,
      "loss": 0.447,
      "step": 27042
    },
    {
      "epoch": 1.5143353119050285,
      "grad_norm": 1.884742021560669,
      "learning_rate": 9.815000000000001e-05,
      "loss": 0.4487,
      "step": 27043
    },
    {
      "epoch": 1.5143913092171575,
      "grad_norm": 1.2877849340438843,
      "learning_rate": 9.814973684210526e-05,
      "loss": 0.5738,
      "step": 27044
    },
    {
      "epoch": 1.5144473065292865,
      "grad_norm": 1.541459560394287,
      "learning_rate": 9.814947368421053e-05,
      "loss": 0.5417,
      "step": 27045
    },
    {
      "epoch": 1.5145033038414155,
      "grad_norm": 1.1358022689819336,
      "learning_rate": 9.814921052631579e-05,
      "loss": 0.5055,
      "step": 27046
    },
    {
      "epoch": 1.5145593011535445,
      "grad_norm": 1.6166036128997803,
      "learning_rate": 9.814894736842107e-05,
      "loss": 0.3977,
      "step": 27047
    },
    {
      "epoch": 1.5146152984656736,
      "grad_norm": 1.3530287742614746,
      "learning_rate": 9.814868421052633e-05,
      "loss": 0.5103,
      "step": 27048
    },
    {
      "epoch": 1.5146712957778026,
      "grad_norm": 1.3396238088607788,
      "learning_rate": 9.814842105263157e-05,
      "loss": 0.3918,
      "step": 27049
    },
    {
      "epoch": 1.5147272930899316,
      "grad_norm": 1.286135196685791,
      "learning_rate": 9.814815789473685e-05,
      "loss": 0.3937,
      "step": 27050
    },
    {
      "epoch": 1.5147832904020606,
      "grad_norm": 1.4488623142242432,
      "learning_rate": 9.81478947368421e-05,
      "loss": 0.4194,
      "step": 27051
    },
    {
      "epoch": 1.5148392877141896,
      "grad_norm": 1.2895300388336182,
      "learning_rate": 9.814763157894738e-05,
      "loss": 0.4557,
      "step": 27052
    },
    {
      "epoch": 1.5148952850263186,
      "grad_norm": 1.3724356889724731,
      "learning_rate": 9.814736842105264e-05,
      "loss": 0.4399,
      "step": 27053
    },
    {
      "epoch": 1.5149512823384477,
      "grad_norm": 1.2225584983825684,
      "learning_rate": 9.81471052631579e-05,
      "loss": 0.4237,
      "step": 27054
    },
    {
      "epoch": 1.5150072796505767,
      "grad_norm": 1.3517637252807617,
      "learning_rate": 9.814684210526316e-05,
      "loss": 0.5057,
      "step": 27055
    },
    {
      "epoch": 1.5150632769627057,
      "grad_norm": 1.3071893453598022,
      "learning_rate": 9.814657894736843e-05,
      "loss": 0.4281,
      "step": 27056
    },
    {
      "epoch": 1.5151192742748347,
      "grad_norm": 1.7707455158233643,
      "learning_rate": 9.814631578947369e-05,
      "loss": 0.5988,
      "step": 27057
    },
    {
      "epoch": 1.5151752715869637,
      "grad_norm": 1.6087961196899414,
      "learning_rate": 9.814605263157895e-05,
      "loss": 0.6235,
      "step": 27058
    },
    {
      "epoch": 1.5152312688990928,
      "grad_norm": 1.6496331691741943,
      "learning_rate": 9.814578947368421e-05,
      "loss": 0.4991,
      "step": 27059
    },
    {
      "epoch": 1.5152872662112218,
      "grad_norm": 1.4258060455322266,
      "learning_rate": 9.814552631578948e-05,
      "loss": 0.4091,
      "step": 27060
    },
    {
      "epoch": 1.5153432635233508,
      "grad_norm": 1.502301812171936,
      "learning_rate": 9.814526315789474e-05,
      "loss": 0.4726,
      "step": 27061
    },
    {
      "epoch": 1.5153992608354798,
      "grad_norm": 1.0531821250915527,
      "learning_rate": 9.8145e-05,
      "loss": 0.4544,
      "step": 27062
    },
    {
      "epoch": 1.5154552581476088,
      "grad_norm": 1.3646159172058105,
      "learning_rate": 9.814473684210526e-05,
      "loss": 0.4808,
      "step": 27063
    },
    {
      "epoch": 1.5155112554597379,
      "grad_norm": 1.2903409004211426,
      "learning_rate": 9.814447368421054e-05,
      "loss": 0.455,
      "step": 27064
    },
    {
      "epoch": 1.5155672527718669,
      "grad_norm": 1.236262559890747,
      "learning_rate": 9.81442105263158e-05,
      "loss": 0.4925,
      "step": 27065
    },
    {
      "epoch": 1.515623250083996,
      "grad_norm": 1.5894114971160889,
      "learning_rate": 9.814394736842106e-05,
      "loss": 0.4423,
      "step": 27066
    },
    {
      "epoch": 1.515679247396125,
      "grad_norm": 1.3579305410385132,
      "learning_rate": 9.814368421052632e-05,
      "loss": 0.3986,
      "step": 27067
    },
    {
      "epoch": 1.515735244708254,
      "grad_norm": 1.482695460319519,
      "learning_rate": 9.814342105263157e-05,
      "loss": 0.4451,
      "step": 27068
    },
    {
      "epoch": 1.515791242020383,
      "grad_norm": 1.119757056236267,
      "learning_rate": 9.814315789473685e-05,
      "loss": 0.4088,
      "step": 27069
    },
    {
      "epoch": 1.515847239332512,
      "grad_norm": 1.4656729698181152,
      "learning_rate": 9.814289473684211e-05,
      "loss": 0.5124,
      "step": 27070
    },
    {
      "epoch": 1.515903236644641,
      "grad_norm": 1.4599612951278687,
      "learning_rate": 9.814263157894738e-05,
      "loss": 0.5764,
      "step": 27071
    },
    {
      "epoch": 1.51595923395677,
      "grad_norm": 1.2479785680770874,
      "learning_rate": 9.814236842105263e-05,
      "loss": 0.4453,
      "step": 27072
    },
    {
      "epoch": 1.516015231268899,
      "grad_norm": 1.2437714338302612,
      "learning_rate": 9.81421052631579e-05,
      "loss": 0.3318,
      "step": 27073
    },
    {
      "epoch": 1.516071228581028,
      "grad_norm": 1.187427043914795,
      "learning_rate": 9.814184210526316e-05,
      "loss": 0.408,
      "step": 27074
    },
    {
      "epoch": 1.516127225893157,
      "grad_norm": 2.3595211505889893,
      "learning_rate": 9.814157894736843e-05,
      "loss": 0.4369,
      "step": 27075
    },
    {
      "epoch": 1.516183223205286,
      "grad_norm": 1.1709308624267578,
      "learning_rate": 9.81413157894737e-05,
      "loss": 0.3735,
      "step": 27076
    },
    {
      "epoch": 1.516239220517415,
      "grad_norm": 1.4980545043945312,
      "learning_rate": 9.814105263157895e-05,
      "loss": 0.558,
      "step": 27077
    },
    {
      "epoch": 1.5162952178295441,
      "grad_norm": 1.3856656551361084,
      "learning_rate": 9.814078947368421e-05,
      "loss": 0.543,
      "step": 27078
    },
    {
      "epoch": 1.5163512151416731,
      "grad_norm": 1.2116948366165161,
      "learning_rate": 9.814052631578949e-05,
      "loss": 0.4328,
      "step": 27079
    },
    {
      "epoch": 1.5164072124538022,
      "grad_norm": 1.1827560663223267,
      "learning_rate": 9.814026315789475e-05,
      "loss": 0.4754,
      "step": 27080
    },
    {
      "epoch": 1.5164632097659312,
      "grad_norm": 1.2897484302520752,
      "learning_rate": 9.814e-05,
      "loss": 0.3885,
      "step": 27081
    },
    {
      "epoch": 1.5165192070780602,
      "grad_norm": 1.2172999382019043,
      "learning_rate": 9.813973684210527e-05,
      "loss": 0.4431,
      "step": 27082
    },
    {
      "epoch": 1.5165752043901892,
      "grad_norm": 1.2611318826675415,
      "learning_rate": 9.813947368421053e-05,
      "loss": 0.4159,
      "step": 27083
    },
    {
      "epoch": 1.5166312017023182,
      "grad_norm": 1.3078234195709229,
      "learning_rate": 9.81392105263158e-05,
      "loss": 0.4671,
      "step": 27084
    },
    {
      "epoch": 1.5166871990144473,
      "grad_norm": 1.3110862970352173,
      "learning_rate": 9.813894736842106e-05,
      "loss": 0.4017,
      "step": 27085
    },
    {
      "epoch": 1.5167431963265763,
      "grad_norm": 1.41542649269104,
      "learning_rate": 9.813868421052632e-05,
      "loss": 0.4007,
      "step": 27086
    },
    {
      "epoch": 1.5167991936387053,
      "grad_norm": 1.227420449256897,
      "learning_rate": 9.813842105263158e-05,
      "loss": 0.5873,
      "step": 27087
    },
    {
      "epoch": 1.5168551909508343,
      "grad_norm": 1.150922417640686,
      "learning_rate": 9.813815789473685e-05,
      "loss": 0.4562,
      "step": 27088
    },
    {
      "epoch": 1.5169111882629633,
      "grad_norm": 1.364452600479126,
      "learning_rate": 9.813789473684211e-05,
      "loss": 0.5006,
      "step": 27089
    },
    {
      "epoch": 1.5169671855750924,
      "grad_norm": 1.3885174989700317,
      "learning_rate": 9.813763157894737e-05,
      "loss": 0.4819,
      "step": 27090
    },
    {
      "epoch": 1.5170231828872214,
      "grad_norm": 1.5120718479156494,
      "learning_rate": 9.813736842105263e-05,
      "loss": 0.4843,
      "step": 27091
    },
    {
      "epoch": 1.5170791801993504,
      "grad_norm": 1.4497497081756592,
      "learning_rate": 9.81371052631579e-05,
      "loss": 0.413,
      "step": 27092
    },
    {
      "epoch": 1.5171351775114794,
      "grad_norm": 1.415647029876709,
      "learning_rate": 9.813684210526316e-05,
      "loss": 0.4852,
      "step": 27093
    },
    {
      "epoch": 1.5171911748236084,
      "grad_norm": 1.0697673559188843,
      "learning_rate": 9.813657894736842e-05,
      "loss": 0.3982,
      "step": 27094
    },
    {
      "epoch": 1.5172471721357375,
      "grad_norm": 1.2788331508636475,
      "learning_rate": 9.813631578947368e-05,
      "loss": 0.4982,
      "step": 27095
    },
    {
      "epoch": 1.5173031694478665,
      "grad_norm": 1.8143675327301025,
      "learning_rate": 9.813605263157896e-05,
      "loss": 0.4638,
      "step": 27096
    },
    {
      "epoch": 1.5173591667599955,
      "grad_norm": 1.411655306816101,
      "learning_rate": 9.813578947368422e-05,
      "loss": 0.52,
      "step": 27097
    },
    {
      "epoch": 1.5174151640721245,
      "grad_norm": 2.727475166320801,
      "learning_rate": 9.813552631578949e-05,
      "loss": 0.4718,
      "step": 27098
    },
    {
      "epoch": 1.5174711613842535,
      "grad_norm": 1.9955580234527588,
      "learning_rate": 9.813526315789473e-05,
      "loss": 0.6964,
      "step": 27099
    },
    {
      "epoch": 1.5175271586963825,
      "grad_norm": 1.6153862476348877,
      "learning_rate": 9.8135e-05,
      "loss": 0.4741,
      "step": 27100
    },
    {
      "epoch": 1.5175831560085116,
      "grad_norm": 1.4560762643814087,
      "learning_rate": 9.813473684210527e-05,
      "loss": 0.5971,
      "step": 27101
    },
    {
      "epoch": 1.5176391533206406,
      "grad_norm": 1.269814133644104,
      "learning_rate": 9.813447368421053e-05,
      "loss": 0.5312,
      "step": 27102
    },
    {
      "epoch": 1.5176951506327696,
      "grad_norm": 1.0865706205368042,
      "learning_rate": 9.81342105263158e-05,
      "loss": 0.3884,
      "step": 27103
    },
    {
      "epoch": 1.5177511479448986,
      "grad_norm": 1.2786933183670044,
      "learning_rate": 9.813394736842105e-05,
      "loss": 0.3779,
      "step": 27104
    },
    {
      "epoch": 1.5178071452570276,
      "grad_norm": 1.280717134475708,
      "learning_rate": 9.813368421052632e-05,
      "loss": 0.3973,
      "step": 27105
    },
    {
      "epoch": 1.5178631425691567,
      "grad_norm": 1.2063264846801758,
      "learning_rate": 9.813342105263158e-05,
      "loss": 0.4159,
      "step": 27106
    },
    {
      "epoch": 1.5179191398812857,
      "grad_norm": 1.1776610612869263,
      "learning_rate": 9.813315789473685e-05,
      "loss": 0.4355,
      "step": 27107
    },
    {
      "epoch": 1.5179751371934147,
      "grad_norm": 1.3531817197799683,
      "learning_rate": 9.813289473684211e-05,
      "loss": 0.4998,
      "step": 27108
    },
    {
      "epoch": 1.5180311345055437,
      "grad_norm": 1.4713177680969238,
      "learning_rate": 9.813263157894737e-05,
      "loss": 0.5086,
      "step": 27109
    },
    {
      "epoch": 1.5180871318176727,
      "grad_norm": 1.3408626317977905,
      "learning_rate": 9.813236842105263e-05,
      "loss": 0.3993,
      "step": 27110
    },
    {
      "epoch": 1.5181431291298018,
      "grad_norm": 1.6550896167755127,
      "learning_rate": 9.81321052631579e-05,
      "loss": 0.5474,
      "step": 27111
    },
    {
      "epoch": 1.5181991264419308,
      "grad_norm": 1.925497055053711,
      "learning_rate": 9.813184210526317e-05,
      "loss": 0.4303,
      "step": 27112
    },
    {
      "epoch": 1.5182551237540598,
      "grad_norm": 1.4134142398834229,
      "learning_rate": 9.813157894736843e-05,
      "loss": 0.502,
      "step": 27113
    },
    {
      "epoch": 1.5183111210661888,
      "grad_norm": 1.3809471130371094,
      "learning_rate": 9.813131578947369e-05,
      "loss": 0.5189,
      "step": 27114
    },
    {
      "epoch": 1.5183671183783178,
      "grad_norm": 1.0960626602172852,
      "learning_rate": 9.813105263157896e-05,
      "loss": 0.422,
      "step": 27115
    },
    {
      "epoch": 1.5184231156904469,
      "grad_norm": 1.3033056259155273,
      "learning_rate": 9.813078947368422e-05,
      "loss": 0.4158,
      "step": 27116
    },
    {
      "epoch": 1.5184791130025759,
      "grad_norm": 1.3114839792251587,
      "learning_rate": 9.813052631578948e-05,
      "loss": 0.4797,
      "step": 27117
    },
    {
      "epoch": 1.518535110314705,
      "grad_norm": 1.7351293563842773,
      "learning_rate": 9.813026315789474e-05,
      "loss": 0.4195,
      "step": 27118
    },
    {
      "epoch": 1.518591107626834,
      "grad_norm": 1.5991266965866089,
      "learning_rate": 9.813e-05,
      "loss": 0.3638,
      "step": 27119
    },
    {
      "epoch": 1.518647104938963,
      "grad_norm": 1.3073616027832031,
      "learning_rate": 9.812973684210527e-05,
      "loss": 0.3962,
      "step": 27120
    },
    {
      "epoch": 1.518703102251092,
      "grad_norm": 1.023710012435913,
      "learning_rate": 9.812947368421053e-05,
      "loss": 0.461,
      "step": 27121
    },
    {
      "epoch": 1.518759099563221,
      "grad_norm": 1.4045839309692383,
      "learning_rate": 9.812921052631579e-05,
      "loss": 0.4891,
      "step": 27122
    },
    {
      "epoch": 1.51881509687535,
      "grad_norm": 1.2008012533187866,
      "learning_rate": 9.812894736842105e-05,
      "loss": 0.3799,
      "step": 27123
    },
    {
      "epoch": 1.518871094187479,
      "grad_norm": 1.2304240465164185,
      "learning_rate": 9.812868421052632e-05,
      "loss": 0.4843,
      "step": 27124
    },
    {
      "epoch": 1.518927091499608,
      "grad_norm": 1.461506724357605,
      "learning_rate": 9.812842105263158e-05,
      "loss": 0.4795,
      "step": 27125
    },
    {
      "epoch": 1.518983088811737,
      "grad_norm": 1.3854840993881226,
      "learning_rate": 9.812815789473686e-05,
      "loss": 0.4384,
      "step": 27126
    },
    {
      "epoch": 1.519039086123866,
      "grad_norm": 1.8799121379852295,
      "learning_rate": 9.81278947368421e-05,
      "loss": 0.6213,
      "step": 27127
    },
    {
      "epoch": 1.519095083435995,
      "grad_norm": 1.2859798669815063,
      "learning_rate": 9.812763157894738e-05,
      "loss": 0.4172,
      "step": 27128
    },
    {
      "epoch": 1.519151080748124,
      "grad_norm": 1.3912931680679321,
      "learning_rate": 9.812736842105264e-05,
      "loss": 0.4523,
      "step": 27129
    },
    {
      "epoch": 1.5192070780602531,
      "grad_norm": 1.075352430343628,
      "learning_rate": 9.812710526315791e-05,
      "loss": 0.37,
      "step": 27130
    },
    {
      "epoch": 1.5192630753723821,
      "grad_norm": 1.4183050394058228,
      "learning_rate": 9.812684210526317e-05,
      "loss": 0.408,
      "step": 27131
    },
    {
      "epoch": 1.5193190726845112,
      "grad_norm": 1.3998409509658813,
      "learning_rate": 9.812657894736843e-05,
      "loss": 0.6337,
      "step": 27132
    },
    {
      "epoch": 1.5193750699966402,
      "grad_norm": 1.930840015411377,
      "learning_rate": 9.812631578947369e-05,
      "loss": 0.5001,
      "step": 27133
    },
    {
      "epoch": 1.5194310673087692,
      "grad_norm": 1.0706380605697632,
      "learning_rate": 9.812605263157896e-05,
      "loss": 0.3335,
      "step": 27134
    },
    {
      "epoch": 1.5194870646208982,
      "grad_norm": 2.313913106918335,
      "learning_rate": 9.812578947368422e-05,
      "loss": 0.6187,
      "step": 27135
    },
    {
      "epoch": 1.5195430619330272,
      "grad_norm": 1.260740041732788,
      "learning_rate": 9.812552631578947e-05,
      "loss": 0.4128,
      "step": 27136
    },
    {
      "epoch": 1.5195990592451563,
      "grad_norm": 3.1872777938842773,
      "learning_rate": 9.812526315789474e-05,
      "loss": 0.4337,
      "step": 27137
    },
    {
      "epoch": 1.5196550565572853,
      "grad_norm": 1.3023390769958496,
      "learning_rate": 9.8125e-05,
      "loss": 0.3654,
      "step": 27138
    },
    {
      "epoch": 1.5197110538694143,
      "grad_norm": 1.215165376663208,
      "learning_rate": 9.812473684210527e-05,
      "loss": 0.3756,
      "step": 27139
    },
    {
      "epoch": 1.5197670511815433,
      "grad_norm": 1.2889282703399658,
      "learning_rate": 9.812447368421053e-05,
      "loss": 0.4967,
      "step": 27140
    },
    {
      "epoch": 1.5198230484936723,
      "grad_norm": 1.33611261844635,
      "learning_rate": 9.812421052631579e-05,
      "loss": 0.3644,
      "step": 27141
    },
    {
      "epoch": 1.5198790458058014,
      "grad_norm": 1.5400924682617188,
      "learning_rate": 9.812394736842105e-05,
      "loss": 0.6058,
      "step": 27142
    },
    {
      "epoch": 1.5199350431179304,
      "grad_norm": 1.2969951629638672,
      "learning_rate": 9.812368421052633e-05,
      "loss": 0.707,
      "step": 27143
    },
    {
      "epoch": 1.5199910404300594,
      "grad_norm": 1.2953400611877441,
      "learning_rate": 9.812342105263159e-05,
      "loss": 0.5529,
      "step": 27144
    },
    {
      "epoch": 1.5200470377421884,
      "grad_norm": 1.1095150709152222,
      "learning_rate": 9.812315789473685e-05,
      "loss": 0.3975,
      "step": 27145
    },
    {
      "epoch": 1.5201030350543174,
      "grad_norm": 1.2372747659683228,
      "learning_rate": 9.81228947368421e-05,
      "loss": 0.5075,
      "step": 27146
    },
    {
      "epoch": 1.5201590323664464,
      "grad_norm": 1.6907185316085815,
      "learning_rate": 9.812263157894738e-05,
      "loss": 0.6013,
      "step": 27147
    },
    {
      "epoch": 1.5202150296785755,
      "grad_norm": 1.5320686101913452,
      "learning_rate": 9.812236842105264e-05,
      "loss": 0.5221,
      "step": 27148
    },
    {
      "epoch": 1.5202710269907045,
      "grad_norm": 1.3951154947280884,
      "learning_rate": 9.81221052631579e-05,
      "loss": 0.6272,
      "step": 27149
    },
    {
      "epoch": 1.5203270243028335,
      "grad_norm": 1.3134037256240845,
      "learning_rate": 9.812184210526316e-05,
      "loss": 0.4784,
      "step": 27150
    },
    {
      "epoch": 1.5203830216149625,
      "grad_norm": 1.2420580387115479,
      "learning_rate": 9.812157894736843e-05,
      "loss": 0.4082,
      "step": 27151
    },
    {
      "epoch": 1.5204390189270915,
      "grad_norm": 1.3023356199264526,
      "learning_rate": 9.812131578947369e-05,
      "loss": 0.5054,
      "step": 27152
    },
    {
      "epoch": 1.5204950162392206,
      "grad_norm": 4.242269515991211,
      "learning_rate": 9.812105263157895e-05,
      "loss": 0.3689,
      "step": 27153
    },
    {
      "epoch": 1.5205510135513496,
      "grad_norm": 1.17594575881958,
      "learning_rate": 9.812078947368421e-05,
      "loss": 0.391,
      "step": 27154
    },
    {
      "epoch": 1.5206070108634786,
      "grad_norm": 1.258170247077942,
      "learning_rate": 9.812052631578947e-05,
      "loss": 0.4083,
      "step": 27155
    },
    {
      "epoch": 1.5206630081756076,
      "grad_norm": 1.5187753438949585,
      "learning_rate": 9.812026315789474e-05,
      "loss": 0.6817,
      "step": 27156
    },
    {
      "epoch": 1.5207190054877366,
      "grad_norm": 1.2015737295150757,
      "learning_rate": 9.812e-05,
      "loss": 0.4681,
      "step": 27157
    },
    {
      "epoch": 1.5207750027998657,
      "grad_norm": 1.2637263536453247,
      "learning_rate": 9.811973684210528e-05,
      "loss": 0.3941,
      "step": 27158
    },
    {
      "epoch": 1.5208310001119947,
      "grad_norm": 1.280678153038025,
      "learning_rate": 9.811947368421052e-05,
      "loss": 0.4573,
      "step": 27159
    },
    {
      "epoch": 1.5208869974241237,
      "grad_norm": 1.2184869050979614,
      "learning_rate": 9.81192105263158e-05,
      "loss": 0.4618,
      "step": 27160
    },
    {
      "epoch": 1.5209429947362527,
      "grad_norm": 1.2928537130355835,
      "learning_rate": 9.811894736842105e-05,
      "loss": 0.3909,
      "step": 27161
    },
    {
      "epoch": 1.5209989920483817,
      "grad_norm": 1.3666313886642456,
      "learning_rate": 9.811868421052633e-05,
      "loss": 0.4929,
      "step": 27162
    },
    {
      "epoch": 1.5210549893605108,
      "grad_norm": 1.306142807006836,
      "learning_rate": 9.811842105263159e-05,
      "loss": 0.4229,
      "step": 27163
    },
    {
      "epoch": 1.5211109866726398,
      "grad_norm": 2.1070852279663086,
      "learning_rate": 9.811815789473685e-05,
      "loss": 0.4423,
      "step": 27164
    },
    {
      "epoch": 1.5211669839847688,
      "grad_norm": 1.2931374311447144,
      "learning_rate": 9.811789473684211e-05,
      "loss": 0.4635,
      "step": 27165
    },
    {
      "epoch": 1.5212229812968978,
      "grad_norm": 1.4965428113937378,
      "learning_rate": 9.811763157894738e-05,
      "loss": 0.5256,
      "step": 27166
    },
    {
      "epoch": 1.5212789786090268,
      "grad_norm": 1.2619374990463257,
      "learning_rate": 9.811736842105264e-05,
      "loss": 0.4944,
      "step": 27167
    },
    {
      "epoch": 1.5213349759211559,
      "grad_norm": 1.268546462059021,
      "learning_rate": 9.81171052631579e-05,
      "loss": 0.4044,
      "step": 27168
    },
    {
      "epoch": 1.5213909732332849,
      "grad_norm": 1.2187644243240356,
      "learning_rate": 9.811684210526316e-05,
      "loss": 0.3879,
      "step": 27169
    },
    {
      "epoch": 1.521446970545414,
      "grad_norm": 1.40371835231781,
      "learning_rate": 9.811657894736842e-05,
      "loss": 0.4912,
      "step": 27170
    },
    {
      "epoch": 1.521502967857543,
      "grad_norm": 2.225950002670288,
      "learning_rate": 9.811631578947369e-05,
      "loss": 0.512,
      "step": 27171
    },
    {
      "epoch": 1.521558965169672,
      "grad_norm": 2.1730637550354004,
      "learning_rate": 9.811605263157895e-05,
      "loss": 0.4606,
      "step": 27172
    },
    {
      "epoch": 1.521614962481801,
      "grad_norm": 1.4855023622512817,
      "learning_rate": 9.811578947368421e-05,
      "loss": 0.4031,
      "step": 27173
    },
    {
      "epoch": 1.52167095979393,
      "grad_norm": 1.6859471797943115,
      "learning_rate": 9.811552631578947e-05,
      "loss": 0.5472,
      "step": 27174
    },
    {
      "epoch": 1.521726957106059,
      "grad_norm": 1.3349955081939697,
      "learning_rate": 9.811526315789475e-05,
      "loss": 0.6914,
      "step": 27175
    },
    {
      "epoch": 1.521782954418188,
      "grad_norm": 1.1871305704116821,
      "learning_rate": 9.8115e-05,
      "loss": 0.3831,
      "step": 27176
    },
    {
      "epoch": 1.521838951730317,
      "grad_norm": 1.4136004447937012,
      "learning_rate": 9.811473684210526e-05,
      "loss": 0.4222,
      "step": 27177
    },
    {
      "epoch": 1.521894949042446,
      "grad_norm": 1.245827317237854,
      "learning_rate": 9.811447368421052e-05,
      "loss": 0.4735,
      "step": 27178
    },
    {
      "epoch": 1.521950946354575,
      "grad_norm": 1.1601252555847168,
      "learning_rate": 9.81142105263158e-05,
      "loss": 0.3037,
      "step": 27179
    },
    {
      "epoch": 1.522006943666704,
      "grad_norm": 1.2845308780670166,
      "learning_rate": 9.811394736842106e-05,
      "loss": 0.4207,
      "step": 27180
    },
    {
      "epoch": 1.522062940978833,
      "grad_norm": 1.2875018119812012,
      "learning_rate": 9.811368421052633e-05,
      "loss": 0.5284,
      "step": 27181
    },
    {
      "epoch": 1.5221189382909621,
      "grad_norm": 1.3341776132583618,
      "learning_rate": 9.811342105263158e-05,
      "loss": 0.4887,
      "step": 27182
    },
    {
      "epoch": 1.5221749356030911,
      "grad_norm": 1.4460558891296387,
      "learning_rate": 9.811315789473685e-05,
      "loss": 0.4795,
      "step": 27183
    },
    {
      "epoch": 1.5222309329152202,
      "grad_norm": 1.4187291860580444,
      "learning_rate": 9.811289473684211e-05,
      "loss": 0.6419,
      "step": 27184
    },
    {
      "epoch": 1.5222869302273492,
      "grad_norm": 1.1622968912124634,
      "learning_rate": 9.811263157894738e-05,
      "loss": 0.3809,
      "step": 27185
    },
    {
      "epoch": 1.5223429275394782,
      "grad_norm": 1.2642873525619507,
      "learning_rate": 9.811236842105263e-05,
      "loss": 0.4248,
      "step": 27186
    },
    {
      "epoch": 1.5223989248516072,
      "grad_norm": 1.4572975635528564,
      "learning_rate": 9.811210526315789e-05,
      "loss": 0.7081,
      "step": 27187
    },
    {
      "epoch": 1.5224549221637362,
      "grad_norm": 1.3987679481506348,
      "learning_rate": 9.811184210526316e-05,
      "loss": 0.606,
      "step": 27188
    },
    {
      "epoch": 1.5225109194758653,
      "grad_norm": 1.2974300384521484,
      "learning_rate": 9.811157894736842e-05,
      "loss": 0.4245,
      "step": 27189
    },
    {
      "epoch": 1.5225669167879943,
      "grad_norm": 1.381773829460144,
      "learning_rate": 9.81113157894737e-05,
      "loss": 0.465,
      "step": 27190
    },
    {
      "epoch": 1.5226229141001233,
      "grad_norm": 1.5028408765792847,
      "learning_rate": 9.811105263157894e-05,
      "loss": 0.4995,
      "step": 27191
    },
    {
      "epoch": 1.5226789114122523,
      "grad_norm": 1.1616239547729492,
      "learning_rate": 9.811078947368421e-05,
      "loss": 0.4463,
      "step": 27192
    },
    {
      "epoch": 1.5227349087243813,
      "grad_norm": 1.256756067276001,
      "learning_rate": 9.811052631578947e-05,
      "loss": 0.396,
      "step": 27193
    },
    {
      "epoch": 1.5227909060365103,
      "grad_norm": 1.2919772863388062,
      "learning_rate": 9.811026315789475e-05,
      "loss": 0.5616,
      "step": 27194
    },
    {
      "epoch": 1.5228469033486394,
      "grad_norm": 1.1837810277938843,
      "learning_rate": 9.811000000000001e-05,
      "loss": 0.4797,
      "step": 27195
    },
    {
      "epoch": 1.5229029006607684,
      "grad_norm": 2.169356346130371,
      "learning_rate": 9.810973684210527e-05,
      "loss": 0.4727,
      "step": 27196
    },
    {
      "epoch": 1.5229588979728974,
      "grad_norm": 1.384002923965454,
      "learning_rate": 9.810947368421053e-05,
      "loss": 0.4697,
      "step": 27197
    },
    {
      "epoch": 1.5230148952850264,
      "grad_norm": 1.2022258043289185,
      "learning_rate": 9.81092105263158e-05,
      "loss": 0.4269,
      "step": 27198
    },
    {
      "epoch": 1.5230708925971554,
      "grad_norm": 1.2444813251495361,
      "learning_rate": 9.810894736842106e-05,
      "loss": 0.5477,
      "step": 27199
    },
    {
      "epoch": 1.5231268899092845,
      "grad_norm": 1.3995620012283325,
      "learning_rate": 9.810868421052632e-05,
      "loss": 0.5408,
      "step": 27200
    },
    {
      "epoch": 1.5231828872214135,
      "grad_norm": 1.2903931140899658,
      "learning_rate": 9.810842105263158e-05,
      "loss": 0.3646,
      "step": 27201
    },
    {
      "epoch": 1.5232388845335425,
      "grad_norm": 1.0688526630401611,
      "learning_rate": 9.810815789473685e-05,
      "loss": 0.3268,
      "step": 27202
    },
    {
      "epoch": 1.5232948818456715,
      "grad_norm": 1.2227773666381836,
      "learning_rate": 9.810789473684211e-05,
      "loss": 0.4748,
      "step": 27203
    },
    {
      "epoch": 1.5233508791578005,
      "grad_norm": 1.2780729532241821,
      "learning_rate": 9.810763157894737e-05,
      "loss": 0.3487,
      "step": 27204
    },
    {
      "epoch": 1.5234068764699296,
      "grad_norm": 1.2940983772277832,
      "learning_rate": 9.810736842105263e-05,
      "loss": 0.6563,
      "step": 27205
    },
    {
      "epoch": 1.5234628737820586,
      "grad_norm": 1.4650967121124268,
      "learning_rate": 9.810710526315789e-05,
      "loss": 0.413,
      "step": 27206
    },
    {
      "epoch": 1.5235188710941876,
      "grad_norm": 1.5916510820388794,
      "learning_rate": 9.810684210526317e-05,
      "loss": 0.4701,
      "step": 27207
    },
    {
      "epoch": 1.5235748684063166,
      "grad_norm": 1.412753939628601,
      "learning_rate": 9.810657894736842e-05,
      "loss": 0.5446,
      "step": 27208
    },
    {
      "epoch": 1.5236308657184456,
      "grad_norm": 1.569606900215149,
      "learning_rate": 9.810631578947368e-05,
      "loss": 0.3985,
      "step": 27209
    },
    {
      "epoch": 1.5236868630305747,
      "grad_norm": 1.2182259559631348,
      "learning_rate": 9.810605263157894e-05,
      "loss": 0.3772,
      "step": 27210
    },
    {
      "epoch": 1.5237428603427037,
      "grad_norm": 1.3352302312850952,
      "learning_rate": 9.810578947368422e-05,
      "loss": 0.4234,
      "step": 27211
    },
    {
      "epoch": 1.5237988576548327,
      "grad_norm": 1.3450623750686646,
      "learning_rate": 9.810552631578948e-05,
      "loss": 0.5302,
      "step": 27212
    },
    {
      "epoch": 1.5238548549669617,
      "grad_norm": 1.124396800994873,
      "learning_rate": 9.810526315789475e-05,
      "loss": 0.4326,
      "step": 27213
    },
    {
      "epoch": 1.5239108522790907,
      "grad_norm": 1.2622078657150269,
      "learning_rate": 9.8105e-05,
      "loss": 0.4639,
      "step": 27214
    },
    {
      "epoch": 1.5239668495912198,
      "grad_norm": 1.2417024374008179,
      "learning_rate": 9.810473684210527e-05,
      "loss": 0.4054,
      "step": 27215
    },
    {
      "epoch": 1.5240228469033488,
      "grad_norm": 1.3757787942886353,
      "learning_rate": 9.810447368421053e-05,
      "loss": 0.656,
      "step": 27216
    },
    {
      "epoch": 1.5240788442154778,
      "grad_norm": 3.607560873031616,
      "learning_rate": 9.81042105263158e-05,
      "loss": 0.5444,
      "step": 27217
    },
    {
      "epoch": 1.5241348415276068,
      "grad_norm": 1.2746714353561401,
      "learning_rate": 9.810394736842106e-05,
      "loss": 0.4427,
      "step": 27218
    },
    {
      "epoch": 1.5241908388397358,
      "grad_norm": 1.492066740989685,
      "learning_rate": 9.810368421052632e-05,
      "loss": 0.5981,
      "step": 27219
    },
    {
      "epoch": 1.5242468361518648,
      "grad_norm": 1.1095901727676392,
      "learning_rate": 9.810342105263158e-05,
      "loss": 0.3708,
      "step": 27220
    },
    {
      "epoch": 1.5243028334639939,
      "grad_norm": 1.349563479423523,
      "learning_rate": 9.810315789473684e-05,
      "loss": 0.4297,
      "step": 27221
    },
    {
      "epoch": 1.5243588307761229,
      "grad_norm": 1.4883623123168945,
      "learning_rate": 9.810289473684212e-05,
      "loss": 0.3863,
      "step": 27222
    },
    {
      "epoch": 1.524414828088252,
      "grad_norm": 1.3130929470062256,
      "learning_rate": 9.810263157894737e-05,
      "loss": 0.406,
      "step": 27223
    },
    {
      "epoch": 1.524470825400381,
      "grad_norm": 1.2860057353973389,
      "learning_rate": 9.810236842105263e-05,
      "loss": 0.4408,
      "step": 27224
    },
    {
      "epoch": 1.52452682271251,
      "grad_norm": 1.1853996515274048,
      "learning_rate": 9.81021052631579e-05,
      "loss": 0.3751,
      "step": 27225
    },
    {
      "epoch": 1.524582820024639,
      "grad_norm": 1.4100369215011597,
      "learning_rate": 9.810184210526317e-05,
      "loss": 0.4522,
      "step": 27226
    },
    {
      "epoch": 1.524638817336768,
      "grad_norm": 1.4437059164047241,
      "learning_rate": 9.810157894736843e-05,
      "loss": 0.5082,
      "step": 27227
    },
    {
      "epoch": 1.524694814648897,
      "grad_norm": 0.9887460470199585,
      "learning_rate": 9.810131578947369e-05,
      "loss": 0.4007,
      "step": 27228
    },
    {
      "epoch": 1.524750811961026,
      "grad_norm": 1.3373279571533203,
      "learning_rate": 9.810105263157895e-05,
      "loss": 0.4514,
      "step": 27229
    },
    {
      "epoch": 1.524806809273155,
      "grad_norm": 1.519224762916565,
      "learning_rate": 9.810078947368422e-05,
      "loss": 0.4418,
      "step": 27230
    },
    {
      "epoch": 1.524862806585284,
      "grad_norm": 1.5326309204101562,
      "learning_rate": 9.810052631578948e-05,
      "loss": 0.6816,
      "step": 27231
    },
    {
      "epoch": 1.524918803897413,
      "grad_norm": 1.2395530939102173,
      "learning_rate": 9.810026315789474e-05,
      "loss": 0.4167,
      "step": 27232
    },
    {
      "epoch": 1.524974801209542,
      "grad_norm": 1.1414129734039307,
      "learning_rate": 9.81e-05,
      "loss": 0.4036,
      "step": 27233
    },
    {
      "epoch": 1.5250307985216711,
      "grad_norm": 1.3156803846359253,
      "learning_rate": 9.809973684210527e-05,
      "loss": 0.4922,
      "step": 27234
    },
    {
      "epoch": 1.5250867958338001,
      "grad_norm": 1.4350786209106445,
      "learning_rate": 9.809947368421053e-05,
      "loss": 0.5052,
      "step": 27235
    },
    {
      "epoch": 1.5251427931459292,
      "grad_norm": 1.3264518976211548,
      "learning_rate": 9.80992105263158e-05,
      "loss": 0.3631,
      "step": 27236
    },
    {
      "epoch": 1.5251987904580582,
      "grad_norm": 1.1088902950286865,
      "learning_rate": 9.809894736842105e-05,
      "loss": 0.3764,
      "step": 27237
    },
    {
      "epoch": 1.5252547877701872,
      "grad_norm": 1.1665536165237427,
      "learning_rate": 9.809868421052633e-05,
      "loss": 0.5014,
      "step": 27238
    },
    {
      "epoch": 1.5253107850823162,
      "grad_norm": 1.2686662673950195,
      "learning_rate": 9.809842105263158e-05,
      "loss": 0.4081,
      "step": 27239
    },
    {
      "epoch": 1.5253667823944452,
      "grad_norm": 2.089110851287842,
      "learning_rate": 9.809815789473684e-05,
      "loss": 0.4076,
      "step": 27240
    },
    {
      "epoch": 1.5254227797065743,
      "grad_norm": 1.2193576097488403,
      "learning_rate": 9.80978947368421e-05,
      "loss": 0.4736,
      "step": 27241
    },
    {
      "epoch": 1.525478777018703,
      "grad_norm": 1.1142656803131104,
      "learning_rate": 9.809763157894736e-05,
      "loss": 0.3628,
      "step": 27242
    },
    {
      "epoch": 1.525534774330832,
      "grad_norm": 2.5167746543884277,
      "learning_rate": 9.809736842105264e-05,
      "loss": 0.4503,
      "step": 27243
    },
    {
      "epoch": 1.525590771642961,
      "grad_norm": 1.2938402891159058,
      "learning_rate": 9.80971052631579e-05,
      "loss": 0.4976,
      "step": 27244
    },
    {
      "epoch": 1.52564676895509,
      "grad_norm": 1.2518161535263062,
      "learning_rate": 9.809684210526317e-05,
      "loss": 0.457,
      "step": 27245
    },
    {
      "epoch": 1.5257027662672191,
      "grad_norm": 1.3567074537277222,
      "learning_rate": 9.809657894736842e-05,
      "loss": 0.4738,
      "step": 27246
    },
    {
      "epoch": 1.5257587635793481,
      "grad_norm": 1.4210426807403564,
      "learning_rate": 9.809631578947369e-05,
      "loss": 0.4596,
      "step": 27247
    },
    {
      "epoch": 1.5258147608914772,
      "grad_norm": 1.4643398523330688,
      "learning_rate": 9.809605263157895e-05,
      "loss": 0.5329,
      "step": 27248
    },
    {
      "epoch": 1.5258707582036062,
      "grad_norm": 1.2194404602050781,
      "learning_rate": 9.809578947368422e-05,
      "loss": 0.4716,
      "step": 27249
    },
    {
      "epoch": 1.5259267555157352,
      "grad_norm": 1.2832481861114502,
      "learning_rate": 9.809552631578948e-05,
      "loss": 0.3971,
      "step": 27250
    },
    {
      "epoch": 1.5259827528278642,
      "grad_norm": 1.1515140533447266,
      "learning_rate": 9.809526315789474e-05,
      "loss": 0.4872,
      "step": 27251
    },
    {
      "epoch": 1.5260387501399932,
      "grad_norm": 1.1932960748672485,
      "learning_rate": 9.8095e-05,
      "loss": 0.3725,
      "step": 27252
    },
    {
      "epoch": 1.5260947474521223,
      "grad_norm": 1.0814008712768555,
      "learning_rate": 9.809473684210528e-05,
      "loss": 0.2995,
      "step": 27253
    },
    {
      "epoch": 1.5261507447642513,
      "grad_norm": 1.4640215635299683,
      "learning_rate": 9.809447368421053e-05,
      "loss": 0.4988,
      "step": 27254
    },
    {
      "epoch": 1.5262067420763803,
      "grad_norm": 1.5298998355865479,
      "learning_rate": 9.80942105263158e-05,
      "loss": 0.491,
      "step": 27255
    },
    {
      "epoch": 1.5262627393885093,
      "grad_norm": 1.3268486261367798,
      "learning_rate": 9.809394736842105e-05,
      "loss": 0.5103,
      "step": 27256
    },
    {
      "epoch": 1.5263187367006383,
      "grad_norm": 1.8547991514205933,
      "learning_rate": 9.809368421052631e-05,
      "loss": 0.4281,
      "step": 27257
    },
    {
      "epoch": 1.5263747340127674,
      "grad_norm": 1.2123308181762695,
      "learning_rate": 9.809342105263159e-05,
      "loss": 0.3843,
      "step": 27258
    },
    {
      "epoch": 1.5264307313248964,
      "grad_norm": 1.3726972341537476,
      "learning_rate": 9.809315789473685e-05,
      "loss": 0.5126,
      "step": 27259
    },
    {
      "epoch": 1.5264867286370254,
      "grad_norm": 1.2992104291915894,
      "learning_rate": 9.809289473684211e-05,
      "loss": 0.4645,
      "step": 27260
    },
    {
      "epoch": 1.5265427259491544,
      "grad_norm": 1.420084834098816,
      "learning_rate": 9.809263157894737e-05,
      "loss": 0.5628,
      "step": 27261
    },
    {
      "epoch": 1.5265987232612834,
      "grad_norm": 1.237008810043335,
      "learning_rate": 9.809236842105264e-05,
      "loss": 0.6578,
      "step": 27262
    },
    {
      "epoch": 1.5266547205734124,
      "grad_norm": 1.4518870115280151,
      "learning_rate": 9.80921052631579e-05,
      "loss": 0.4862,
      "step": 27263
    },
    {
      "epoch": 1.5267107178855415,
      "grad_norm": 1.1755800247192383,
      "learning_rate": 9.809184210526316e-05,
      "loss": 0.4391,
      "step": 27264
    },
    {
      "epoch": 1.5267667151976705,
      "grad_norm": 1.2907551527023315,
      "learning_rate": 9.809157894736842e-05,
      "loss": 0.5257,
      "step": 27265
    },
    {
      "epoch": 1.5268227125097995,
      "grad_norm": 1.1906359195709229,
      "learning_rate": 9.809131578947369e-05,
      "loss": 0.3457,
      "step": 27266
    },
    {
      "epoch": 1.5268787098219285,
      "grad_norm": 1.5713539123535156,
      "learning_rate": 9.809105263157895e-05,
      "loss": 0.467,
      "step": 27267
    },
    {
      "epoch": 1.5269347071340575,
      "grad_norm": 1.5531158447265625,
      "learning_rate": 9.809078947368423e-05,
      "loss": 0.5502,
      "step": 27268
    },
    {
      "epoch": 1.5269907044461866,
      "grad_norm": 1.4606424570083618,
      "learning_rate": 9.809052631578947e-05,
      "loss": 0.2967,
      "step": 27269
    },
    {
      "epoch": 1.5270467017583156,
      "grad_norm": 1.352168083190918,
      "learning_rate": 9.809026315789474e-05,
      "loss": 0.4274,
      "step": 27270
    },
    {
      "epoch": 1.5271026990704446,
      "grad_norm": 1.477839469909668,
      "learning_rate": 9.809e-05,
      "loss": 0.3866,
      "step": 27271
    },
    {
      "epoch": 1.5271586963825736,
      "grad_norm": 1.2686796188354492,
      "learning_rate": 9.808973684210528e-05,
      "loss": 0.4013,
      "step": 27272
    },
    {
      "epoch": 1.5272146936947026,
      "grad_norm": 1.5855923891067505,
      "learning_rate": 9.808947368421054e-05,
      "loss": 0.5514,
      "step": 27273
    },
    {
      "epoch": 1.5272706910068317,
      "grad_norm": 1.2596975564956665,
      "learning_rate": 9.808921052631578e-05,
      "loss": 0.4397,
      "step": 27274
    },
    {
      "epoch": 1.5273266883189607,
      "grad_norm": 1.4806995391845703,
      "learning_rate": 9.808894736842106e-05,
      "loss": 0.5705,
      "step": 27275
    },
    {
      "epoch": 1.5273826856310897,
      "grad_norm": 1.3230366706848145,
      "learning_rate": 9.808868421052632e-05,
      "loss": 0.6061,
      "step": 27276
    },
    {
      "epoch": 1.5274386829432187,
      "grad_norm": 1.2527214288711548,
      "learning_rate": 9.808842105263159e-05,
      "loss": 0.464,
      "step": 27277
    },
    {
      "epoch": 1.5274946802553477,
      "grad_norm": 1.3091689348220825,
      "learning_rate": 9.808815789473685e-05,
      "loss": 0.3889,
      "step": 27278
    },
    {
      "epoch": 1.5275506775674768,
      "grad_norm": 1.3826245069503784,
      "learning_rate": 9.808789473684211e-05,
      "loss": 0.4983,
      "step": 27279
    },
    {
      "epoch": 1.5276066748796058,
      "grad_norm": 1.2268179655075073,
      "learning_rate": 9.808763157894737e-05,
      "loss": 0.3406,
      "step": 27280
    },
    {
      "epoch": 1.5276626721917348,
      "grad_norm": 1.3147023916244507,
      "learning_rate": 9.808736842105264e-05,
      "loss": 0.6199,
      "step": 27281
    },
    {
      "epoch": 1.5277186695038638,
      "grad_norm": 1.1666747331619263,
      "learning_rate": 9.80871052631579e-05,
      "loss": 0.4304,
      "step": 27282
    },
    {
      "epoch": 1.5277746668159928,
      "grad_norm": 1.3330376148223877,
      "learning_rate": 9.808684210526316e-05,
      "loss": 0.4283,
      "step": 27283
    },
    {
      "epoch": 1.5278306641281219,
      "grad_norm": 1.4851858615875244,
      "learning_rate": 9.808657894736842e-05,
      "loss": 0.3865,
      "step": 27284
    },
    {
      "epoch": 1.5278866614402509,
      "grad_norm": 1.4621042013168335,
      "learning_rate": 9.80863157894737e-05,
      "loss": 0.5033,
      "step": 27285
    },
    {
      "epoch": 1.52794265875238,
      "grad_norm": 1.2530263662338257,
      "learning_rate": 9.808605263157895e-05,
      "loss": 0.4445,
      "step": 27286
    },
    {
      "epoch": 1.527998656064509,
      "grad_norm": 1.4377599954605103,
      "learning_rate": 9.808578947368421e-05,
      "loss": 0.4676,
      "step": 27287
    },
    {
      "epoch": 1.528054653376638,
      "grad_norm": 1.7416679859161377,
      "learning_rate": 9.808552631578947e-05,
      "loss": 0.5612,
      "step": 27288
    },
    {
      "epoch": 1.528110650688767,
      "grad_norm": 1.4890049695968628,
      "learning_rate": 9.808526315789475e-05,
      "loss": 0.4705,
      "step": 27289
    },
    {
      "epoch": 1.528166648000896,
      "grad_norm": 1.6244866847991943,
      "learning_rate": 9.808500000000001e-05,
      "loss": 0.5147,
      "step": 27290
    },
    {
      "epoch": 1.528222645313025,
      "grad_norm": 1.2908213138580322,
      "learning_rate": 9.808473684210527e-05,
      "loss": 0.3702,
      "step": 27291
    },
    {
      "epoch": 1.528278642625154,
      "grad_norm": 1.3829395771026611,
      "learning_rate": 9.808447368421053e-05,
      "loss": 0.4444,
      "step": 27292
    },
    {
      "epoch": 1.528334639937283,
      "grad_norm": 1.6316728591918945,
      "learning_rate": 9.808421052631579e-05,
      "loss": 0.4936,
      "step": 27293
    },
    {
      "epoch": 1.528390637249412,
      "grad_norm": 1.4428198337554932,
      "learning_rate": 9.808394736842106e-05,
      "loss": 0.5268,
      "step": 27294
    },
    {
      "epoch": 1.528446634561541,
      "grad_norm": 1.3690496683120728,
      "learning_rate": 9.808368421052632e-05,
      "loss": 0.4147,
      "step": 27295
    },
    {
      "epoch": 1.52850263187367,
      "grad_norm": 1.5757402181625366,
      "learning_rate": 9.808342105263158e-05,
      "loss": 0.6325,
      "step": 27296
    },
    {
      "epoch": 1.528558629185799,
      "grad_norm": 1.3949118852615356,
      "learning_rate": 9.808315789473684e-05,
      "loss": 0.4446,
      "step": 27297
    },
    {
      "epoch": 1.5286146264979281,
      "grad_norm": 1.326460838317871,
      "learning_rate": 9.808289473684211e-05,
      "loss": 0.5211,
      "step": 27298
    },
    {
      "epoch": 1.5286706238100571,
      "grad_norm": 1.2361222505569458,
      "learning_rate": 9.808263157894737e-05,
      "loss": 0.4878,
      "step": 27299
    },
    {
      "epoch": 1.5287266211221862,
      "grad_norm": 1.2712743282318115,
      "learning_rate": 9.808236842105264e-05,
      "loss": 0.3534,
      "step": 27300
    },
    {
      "epoch": 1.5287826184343152,
      "grad_norm": 1.9682635068893433,
      "learning_rate": 9.808210526315789e-05,
      "loss": 0.4473,
      "step": 27301
    },
    {
      "epoch": 1.5288386157464442,
      "grad_norm": 1.5164655447006226,
      "learning_rate": 9.808184210526316e-05,
      "loss": 0.5343,
      "step": 27302
    },
    {
      "epoch": 1.5288946130585732,
      "grad_norm": 1.496638298034668,
      "learning_rate": 9.808157894736842e-05,
      "loss": 0.5247,
      "step": 27303
    },
    {
      "epoch": 1.5289506103707022,
      "grad_norm": 1.265756607055664,
      "learning_rate": 9.80813157894737e-05,
      "loss": 0.3993,
      "step": 27304
    },
    {
      "epoch": 1.5290066076828313,
      "grad_norm": 1.3341575860977173,
      "learning_rate": 9.808105263157896e-05,
      "loss": 0.4568,
      "step": 27305
    },
    {
      "epoch": 1.5290626049949603,
      "grad_norm": 1.7430514097213745,
      "learning_rate": 9.808078947368422e-05,
      "loss": 0.4912,
      "step": 27306
    },
    {
      "epoch": 1.5291186023070893,
      "grad_norm": 1.522204041481018,
      "learning_rate": 9.808052631578948e-05,
      "loss": 0.4835,
      "step": 27307
    },
    {
      "epoch": 1.5291745996192183,
      "grad_norm": 8.616300582885742,
      "learning_rate": 9.808026315789474e-05,
      "loss": 0.6381,
      "step": 27308
    },
    {
      "epoch": 1.5292305969313473,
      "grad_norm": 1.8359129428863525,
      "learning_rate": 9.808000000000001e-05,
      "loss": 0.6387,
      "step": 27309
    },
    {
      "epoch": 1.5292865942434763,
      "grad_norm": 1.3288633823394775,
      "learning_rate": 9.807973684210527e-05,
      "loss": 0.4425,
      "step": 27310
    },
    {
      "epoch": 1.5293425915556054,
      "grad_norm": 1.3993644714355469,
      "learning_rate": 9.807947368421053e-05,
      "loss": 0.4144,
      "step": 27311
    },
    {
      "epoch": 1.5293985888677344,
      "grad_norm": 1.2721813917160034,
      "learning_rate": 9.807921052631579e-05,
      "loss": 0.3192,
      "step": 27312
    },
    {
      "epoch": 1.5294545861798634,
      "grad_norm": 1.2544965744018555,
      "learning_rate": 9.807894736842106e-05,
      "loss": 0.5001,
      "step": 27313
    },
    {
      "epoch": 1.5295105834919924,
      "grad_norm": 1.2644429206848145,
      "learning_rate": 9.807868421052632e-05,
      "loss": 0.4599,
      "step": 27314
    },
    {
      "epoch": 1.5295665808041214,
      "grad_norm": 1.3764448165893555,
      "learning_rate": 9.807842105263158e-05,
      "loss": 0.518,
      "step": 27315
    },
    {
      "epoch": 1.5296225781162505,
      "grad_norm": 1.488978624343872,
      "learning_rate": 9.807815789473684e-05,
      "loss": 0.5179,
      "step": 27316
    },
    {
      "epoch": 1.5296785754283795,
      "grad_norm": 1.5431971549987793,
      "learning_rate": 9.807789473684211e-05,
      "loss": 0.5134,
      "step": 27317
    },
    {
      "epoch": 1.5297345727405085,
      "grad_norm": 1.3441526889801025,
      "learning_rate": 9.807763157894737e-05,
      "loss": 0.3493,
      "step": 27318
    },
    {
      "epoch": 1.5297905700526375,
      "grad_norm": 1.3060107231140137,
      "learning_rate": 9.807736842105263e-05,
      "loss": 0.4902,
      "step": 27319
    },
    {
      "epoch": 1.5298465673647665,
      "grad_norm": 1.2472405433654785,
      "learning_rate": 9.80771052631579e-05,
      "loss": 0.3954,
      "step": 27320
    },
    {
      "epoch": 1.5299025646768956,
      "grad_norm": 1.2510164976119995,
      "learning_rate": 9.807684210526317e-05,
      "loss": 0.4524,
      "step": 27321
    },
    {
      "epoch": 1.5299585619890246,
      "grad_norm": 1.3180891275405884,
      "learning_rate": 9.807657894736843e-05,
      "loss": 0.3708,
      "step": 27322
    },
    {
      "epoch": 1.5300145593011536,
      "grad_norm": 1.217759370803833,
      "learning_rate": 9.80763157894737e-05,
      "loss": 0.4818,
      "step": 27323
    },
    {
      "epoch": 1.5300705566132826,
      "grad_norm": 1.3709975481033325,
      "learning_rate": 9.807605263157895e-05,
      "loss": 0.46,
      "step": 27324
    },
    {
      "epoch": 1.5301265539254114,
      "grad_norm": 1.1192548274993896,
      "learning_rate": 9.80757894736842e-05,
      "loss": 0.3857,
      "step": 27325
    },
    {
      "epoch": 1.5301825512375404,
      "grad_norm": 1.4618414640426636,
      "learning_rate": 9.807552631578948e-05,
      "loss": 0.3888,
      "step": 27326
    },
    {
      "epoch": 1.5302385485496695,
      "grad_norm": 1.6253644227981567,
      "learning_rate": 9.807526315789474e-05,
      "loss": 0.6086,
      "step": 27327
    },
    {
      "epoch": 1.5302945458617985,
      "grad_norm": 1.1323260068893433,
      "learning_rate": 9.807500000000001e-05,
      "loss": 0.4307,
      "step": 27328
    },
    {
      "epoch": 1.5303505431739275,
      "grad_norm": 1.3198050260543823,
      "learning_rate": 9.807473684210526e-05,
      "loss": 0.4646,
      "step": 27329
    },
    {
      "epoch": 1.5304065404860565,
      "grad_norm": 1.1652281284332275,
      "learning_rate": 9.807447368421053e-05,
      "loss": 0.4897,
      "step": 27330
    },
    {
      "epoch": 1.5304625377981855,
      "grad_norm": 1.1871758699417114,
      "learning_rate": 9.807421052631579e-05,
      "loss": 0.5071,
      "step": 27331
    },
    {
      "epoch": 1.5305185351103145,
      "grad_norm": 1.3732500076293945,
      "learning_rate": 9.807394736842106e-05,
      "loss": 0.3617,
      "step": 27332
    },
    {
      "epoch": 1.5305745324224436,
      "grad_norm": 1.1966280937194824,
      "learning_rate": 9.807368421052631e-05,
      "loss": 0.4051,
      "step": 27333
    },
    {
      "epoch": 1.5306305297345726,
      "grad_norm": 1.1944531202316284,
      "learning_rate": 9.807342105263158e-05,
      "loss": 0.387,
      "step": 27334
    },
    {
      "epoch": 1.5306865270467016,
      "grad_norm": 1.246779203414917,
      "learning_rate": 9.807315789473684e-05,
      "loss": 0.3806,
      "step": 27335
    },
    {
      "epoch": 1.5307425243588306,
      "grad_norm": 1.5883923768997192,
      "learning_rate": 9.807289473684212e-05,
      "loss": 0.5501,
      "step": 27336
    },
    {
      "epoch": 1.5307985216709596,
      "grad_norm": 1.4065109491348267,
      "learning_rate": 9.807263157894738e-05,
      "loss": 0.461,
      "step": 27337
    },
    {
      "epoch": 1.5308545189830887,
      "grad_norm": 1.3632166385650635,
      "learning_rate": 9.807236842105264e-05,
      "loss": 0.4515,
      "step": 27338
    },
    {
      "epoch": 1.5309105162952177,
      "grad_norm": 1.1589895486831665,
      "learning_rate": 9.80721052631579e-05,
      "loss": 0.506,
      "step": 27339
    },
    {
      "epoch": 1.5309665136073467,
      "grad_norm": 1.538962960243225,
      "learning_rate": 9.807184210526317e-05,
      "loss": 0.3962,
      "step": 27340
    },
    {
      "epoch": 1.5310225109194757,
      "grad_norm": 1.7433265447616577,
      "learning_rate": 9.807157894736843e-05,
      "loss": 0.5894,
      "step": 27341
    },
    {
      "epoch": 1.5310785082316047,
      "grad_norm": 1.180971384048462,
      "learning_rate": 9.807131578947369e-05,
      "loss": 0.2923,
      "step": 27342
    },
    {
      "epoch": 1.5311345055437338,
      "grad_norm": 1.596555471420288,
      "learning_rate": 9.807105263157895e-05,
      "loss": 0.4668,
      "step": 27343
    },
    {
      "epoch": 1.5311905028558628,
      "grad_norm": 1.2896409034729004,
      "learning_rate": 9.807078947368421e-05,
      "loss": 0.5266,
      "step": 27344
    },
    {
      "epoch": 1.5312465001679918,
      "grad_norm": 1.253602385520935,
      "learning_rate": 9.807052631578948e-05,
      "loss": 0.5117,
      "step": 27345
    },
    {
      "epoch": 1.5313024974801208,
      "grad_norm": 1.3735167980194092,
      "learning_rate": 9.807026315789474e-05,
      "loss": 0.4223,
      "step": 27346
    },
    {
      "epoch": 1.5313584947922498,
      "grad_norm": 1.3457038402557373,
      "learning_rate": 9.807e-05,
      "loss": 0.4288,
      "step": 27347
    },
    {
      "epoch": 1.5314144921043789,
      "grad_norm": 1.4801067113876343,
      "learning_rate": 9.806973684210526e-05,
      "loss": 0.4664,
      "step": 27348
    },
    {
      "epoch": 1.5314704894165079,
      "grad_norm": 1.544136643409729,
      "learning_rate": 9.806947368421053e-05,
      "loss": 0.449,
      "step": 27349
    },
    {
      "epoch": 1.531526486728637,
      "grad_norm": 1.1419053077697754,
      "learning_rate": 9.80692105263158e-05,
      "loss": 0.4573,
      "step": 27350
    },
    {
      "epoch": 1.531582484040766,
      "grad_norm": 1.8215211629867554,
      "learning_rate": 9.806894736842105e-05,
      "loss": 0.495,
      "step": 27351
    },
    {
      "epoch": 1.531638481352895,
      "grad_norm": 1.4061273336410522,
      "learning_rate": 9.806868421052631e-05,
      "loss": 0.5048,
      "step": 27352
    },
    {
      "epoch": 1.531694478665024,
      "grad_norm": 1.3769176006317139,
      "learning_rate": 9.806842105263159e-05,
      "loss": 0.5012,
      "step": 27353
    },
    {
      "epoch": 1.531750475977153,
      "grad_norm": 1.22798490524292,
      "learning_rate": 9.806815789473685e-05,
      "loss": 0.4844,
      "step": 27354
    },
    {
      "epoch": 1.531806473289282,
      "grad_norm": 1.7953495979309082,
      "learning_rate": 9.806789473684212e-05,
      "loss": 0.5709,
      "step": 27355
    },
    {
      "epoch": 1.531862470601411,
      "grad_norm": 1.2608802318572998,
      "learning_rate": 9.806763157894737e-05,
      "loss": 0.3773,
      "step": 27356
    },
    {
      "epoch": 1.53191846791354,
      "grad_norm": 1.4404892921447754,
      "learning_rate": 9.806736842105264e-05,
      "loss": 0.6066,
      "step": 27357
    },
    {
      "epoch": 1.531974465225669,
      "grad_norm": 1.2526389360427856,
      "learning_rate": 9.80671052631579e-05,
      "loss": 0.3322,
      "step": 27358
    },
    {
      "epoch": 1.532030462537798,
      "grad_norm": 1.2167657613754272,
      "learning_rate": 9.806684210526317e-05,
      "loss": 0.3577,
      "step": 27359
    },
    {
      "epoch": 1.532086459849927,
      "grad_norm": 1.2484211921691895,
      "learning_rate": 9.806657894736843e-05,
      "loss": 0.3917,
      "step": 27360
    },
    {
      "epoch": 1.532142457162056,
      "grad_norm": 1.107147455215454,
      "learning_rate": 9.806631578947368e-05,
      "loss": 0.4293,
      "step": 27361
    },
    {
      "epoch": 1.5321984544741851,
      "grad_norm": 1.4587103128433228,
      "learning_rate": 9.806605263157895e-05,
      "loss": 0.6043,
      "step": 27362
    },
    {
      "epoch": 1.5322544517863141,
      "grad_norm": 1.3617078065872192,
      "learning_rate": 9.806578947368421e-05,
      "loss": 0.5124,
      "step": 27363
    },
    {
      "epoch": 1.5323104490984432,
      "grad_norm": 1.3550008535385132,
      "learning_rate": 9.806552631578948e-05,
      "loss": 0.4097,
      "step": 27364
    },
    {
      "epoch": 1.5323664464105722,
      "grad_norm": 1.4102470874786377,
      "learning_rate": 9.806526315789474e-05,
      "loss": 0.4618,
      "step": 27365
    },
    {
      "epoch": 1.5324224437227012,
      "grad_norm": 1.5110503435134888,
      "learning_rate": 9.8065e-05,
      "loss": 0.5387,
      "step": 27366
    },
    {
      "epoch": 1.5324784410348302,
      "grad_norm": 1.5743473768234253,
      "learning_rate": 9.806473684210526e-05,
      "loss": 0.4776,
      "step": 27367
    },
    {
      "epoch": 1.5325344383469592,
      "grad_norm": 1.6093778610229492,
      "learning_rate": 9.806447368421054e-05,
      "loss": 0.3658,
      "step": 27368
    },
    {
      "epoch": 1.5325904356590883,
      "grad_norm": 1.5215086936950684,
      "learning_rate": 9.80642105263158e-05,
      "loss": 0.5366,
      "step": 27369
    },
    {
      "epoch": 1.5326464329712173,
      "grad_norm": 1.475719690322876,
      "learning_rate": 9.806394736842106e-05,
      "loss": 0.6122,
      "step": 27370
    },
    {
      "epoch": 1.5327024302833463,
      "grad_norm": 1.2523468732833862,
      "learning_rate": 9.806368421052632e-05,
      "loss": 0.398,
      "step": 27371
    },
    {
      "epoch": 1.5327584275954753,
      "grad_norm": 1.1293725967407227,
      "learning_rate": 9.806342105263159e-05,
      "loss": 0.3936,
      "step": 27372
    },
    {
      "epoch": 1.5328144249076043,
      "grad_norm": 1.5792362689971924,
      "learning_rate": 9.806315789473685e-05,
      "loss": 0.4651,
      "step": 27373
    },
    {
      "epoch": 1.5328704222197334,
      "grad_norm": 1.2675620317459106,
      "learning_rate": 9.806289473684211e-05,
      "loss": 0.6451,
      "step": 27374
    },
    {
      "epoch": 1.5329264195318624,
      "grad_norm": 1.1812338829040527,
      "learning_rate": 9.806263157894737e-05,
      "loss": 0.3571,
      "step": 27375
    },
    {
      "epoch": 1.5329824168439914,
      "grad_norm": 1.276490330696106,
      "learning_rate": 9.806236842105264e-05,
      "loss": 0.4349,
      "step": 27376
    },
    {
      "epoch": 1.5330384141561204,
      "grad_norm": 1.2429367303848267,
      "learning_rate": 9.80621052631579e-05,
      "loss": 0.4896,
      "step": 27377
    },
    {
      "epoch": 1.5330944114682494,
      "grad_norm": 1.464216947555542,
      "learning_rate": 9.806184210526316e-05,
      "loss": 0.4263,
      "step": 27378
    },
    {
      "epoch": 1.5331504087803784,
      "grad_norm": 1.555129051208496,
      "learning_rate": 9.806157894736842e-05,
      "loss": 0.5057,
      "step": 27379
    },
    {
      "epoch": 1.5332064060925075,
      "grad_norm": 1.5318166017532349,
      "learning_rate": 9.806131578947368e-05,
      "loss": 0.5248,
      "step": 27380
    },
    {
      "epoch": 1.5332624034046365,
      "grad_norm": 1.4204380512237549,
      "learning_rate": 9.806105263157895e-05,
      "loss": 0.5319,
      "step": 27381
    },
    {
      "epoch": 1.5333184007167655,
      "grad_norm": 1.737895131111145,
      "learning_rate": 9.806078947368421e-05,
      "loss": 0.543,
      "step": 27382
    },
    {
      "epoch": 1.5333743980288945,
      "grad_norm": 1.4844969511032104,
      "learning_rate": 9.806052631578949e-05,
      "loss": 0.492,
      "step": 27383
    },
    {
      "epoch": 1.5334303953410235,
      "grad_norm": 1.2135682106018066,
      "learning_rate": 9.806026315789473e-05,
      "loss": 0.4453,
      "step": 27384
    },
    {
      "epoch": 1.5334863926531526,
      "grad_norm": 1.1250629425048828,
      "learning_rate": 9.806e-05,
      "loss": 0.3666,
      "step": 27385
    },
    {
      "epoch": 1.5335423899652816,
      "grad_norm": 1.3407983779907227,
      "learning_rate": 9.805973684210527e-05,
      "loss": 0.4875,
      "step": 27386
    },
    {
      "epoch": 1.5335983872774106,
      "grad_norm": 1.3464614152908325,
      "learning_rate": 9.805947368421054e-05,
      "loss": 0.3705,
      "step": 27387
    },
    {
      "epoch": 1.5336543845895396,
      "grad_norm": 1.4388389587402344,
      "learning_rate": 9.805921052631579e-05,
      "loss": 0.3942,
      "step": 27388
    },
    {
      "epoch": 1.5337103819016686,
      "grad_norm": 1.7731399536132812,
      "learning_rate": 9.805894736842106e-05,
      "loss": 0.4749,
      "step": 27389
    },
    {
      "epoch": 1.5337663792137977,
      "grad_norm": 1.9700196981430054,
      "learning_rate": 9.805868421052632e-05,
      "loss": 0.3996,
      "step": 27390
    },
    {
      "epoch": 1.5338223765259267,
      "grad_norm": 1.5039012432098389,
      "learning_rate": 9.805842105263159e-05,
      "loss": 0.5317,
      "step": 27391
    },
    {
      "epoch": 1.5338783738380557,
      "grad_norm": 1.7295633554458618,
      "learning_rate": 9.805815789473685e-05,
      "loss": 0.6416,
      "step": 27392
    },
    {
      "epoch": 1.5339343711501847,
      "grad_norm": 1.3499912023544312,
      "learning_rate": 9.805789473684211e-05,
      "loss": 0.5603,
      "step": 27393
    },
    {
      "epoch": 1.5339903684623137,
      "grad_norm": 1.2199500799179077,
      "learning_rate": 9.805763157894737e-05,
      "loss": 0.4216,
      "step": 27394
    },
    {
      "epoch": 1.5340463657744428,
      "grad_norm": 1.0372475385665894,
      "learning_rate": 9.805736842105263e-05,
      "loss": 0.3845,
      "step": 27395
    },
    {
      "epoch": 1.5341023630865718,
      "grad_norm": 1.6075316667556763,
      "learning_rate": 9.80571052631579e-05,
      "loss": 0.7346,
      "step": 27396
    },
    {
      "epoch": 1.5341583603987008,
      "grad_norm": 1.2499035596847534,
      "learning_rate": 9.805684210526316e-05,
      "loss": 0.5194,
      "step": 27397
    },
    {
      "epoch": 1.5342143577108298,
      "grad_norm": 1.4742541313171387,
      "learning_rate": 9.805657894736842e-05,
      "loss": 0.424,
      "step": 27398
    },
    {
      "epoch": 1.5342703550229588,
      "grad_norm": 1.3422431945800781,
      "learning_rate": 9.805631578947368e-05,
      "loss": 0.4758,
      "step": 27399
    },
    {
      "epoch": 1.5343263523350879,
      "grad_norm": 1.3489582538604736,
      "learning_rate": 9.805605263157896e-05,
      "loss": 0.4106,
      "step": 27400
    },
    {
      "epoch": 1.5343823496472169,
      "grad_norm": 1.3509870767593384,
      "learning_rate": 9.805578947368422e-05,
      "loss": 0.3976,
      "step": 27401
    },
    {
      "epoch": 1.534438346959346,
      "grad_norm": 2.1054441928863525,
      "learning_rate": 9.805552631578948e-05,
      "loss": 0.5922,
      "step": 27402
    },
    {
      "epoch": 1.534494344271475,
      "grad_norm": 1.5807559490203857,
      "learning_rate": 9.805526315789474e-05,
      "loss": 0.4145,
      "step": 27403
    },
    {
      "epoch": 1.534550341583604,
      "grad_norm": 1.226464867591858,
      "learning_rate": 9.805500000000001e-05,
      "loss": 0.3998,
      "step": 27404
    },
    {
      "epoch": 1.534606338895733,
      "grad_norm": 1.1433545351028442,
      "learning_rate": 9.805473684210527e-05,
      "loss": 0.3917,
      "step": 27405
    },
    {
      "epoch": 1.534662336207862,
      "grad_norm": 1.6557573080062866,
      "learning_rate": 9.805447368421053e-05,
      "loss": 0.4927,
      "step": 27406
    },
    {
      "epoch": 1.534718333519991,
      "grad_norm": 1.523334264755249,
      "learning_rate": 9.805421052631579e-05,
      "loss": 0.3263,
      "step": 27407
    },
    {
      "epoch": 1.53477433083212,
      "grad_norm": 1.2454636096954346,
      "learning_rate": 9.805394736842106e-05,
      "loss": 0.4938,
      "step": 27408
    },
    {
      "epoch": 1.534830328144249,
      "grad_norm": 1.1688791513442993,
      "learning_rate": 9.805368421052632e-05,
      "loss": 0.3803,
      "step": 27409
    },
    {
      "epoch": 1.534886325456378,
      "grad_norm": 1.3370659351348877,
      "learning_rate": 9.80534210526316e-05,
      "loss": 0.3608,
      "step": 27410
    },
    {
      "epoch": 1.534942322768507,
      "grad_norm": 2.0097856521606445,
      "learning_rate": 9.805315789473684e-05,
      "loss": 0.5683,
      "step": 27411
    },
    {
      "epoch": 1.534998320080636,
      "grad_norm": 1.3716015815734863,
      "learning_rate": 9.80528947368421e-05,
      "loss": 0.3982,
      "step": 27412
    },
    {
      "epoch": 1.535054317392765,
      "grad_norm": 1.6543197631835938,
      "learning_rate": 9.805263157894737e-05,
      "loss": 0.4842,
      "step": 27413
    },
    {
      "epoch": 1.5351103147048941,
      "grad_norm": 1.7502961158752441,
      "learning_rate": 9.805236842105263e-05,
      "loss": 0.5106,
      "step": 27414
    },
    {
      "epoch": 1.5351663120170231,
      "grad_norm": 1.0914019346237183,
      "learning_rate": 9.80521052631579e-05,
      "loss": 0.3747,
      "step": 27415
    },
    {
      "epoch": 1.5352223093291522,
      "grad_norm": 1.7459182739257812,
      "learning_rate": 9.805184210526315e-05,
      "loss": 0.4082,
      "step": 27416
    },
    {
      "epoch": 1.5352783066412812,
      "grad_norm": 1.4182765483856201,
      "learning_rate": 9.805157894736843e-05,
      "loss": 0.4289,
      "step": 27417
    },
    {
      "epoch": 1.5353343039534102,
      "grad_norm": 1.6390981674194336,
      "learning_rate": 9.805131578947369e-05,
      "loss": 0.5726,
      "step": 27418
    },
    {
      "epoch": 1.5353903012655392,
      "grad_norm": 1.5273953676223755,
      "learning_rate": 9.805105263157896e-05,
      "loss": 0.4502,
      "step": 27419
    },
    {
      "epoch": 1.5354462985776682,
      "grad_norm": 1.197325587272644,
      "learning_rate": 9.805078947368422e-05,
      "loss": 0.4224,
      "step": 27420
    },
    {
      "epoch": 1.5355022958897973,
      "grad_norm": 1.758510708808899,
      "learning_rate": 9.805052631578948e-05,
      "loss": 0.5773,
      "step": 27421
    },
    {
      "epoch": 1.5355582932019263,
      "grad_norm": 1.3416218757629395,
      "learning_rate": 9.805026315789474e-05,
      "loss": 0.4875,
      "step": 27422
    },
    {
      "epoch": 1.5356142905140553,
      "grad_norm": 1.3372999429702759,
      "learning_rate": 9.805000000000001e-05,
      "loss": 0.3808,
      "step": 27423
    },
    {
      "epoch": 1.5356702878261843,
      "grad_norm": 1.239515781402588,
      "learning_rate": 9.804973684210527e-05,
      "loss": 0.6036,
      "step": 27424
    },
    {
      "epoch": 1.5357262851383133,
      "grad_norm": 1.2493059635162354,
      "learning_rate": 9.804947368421053e-05,
      "loss": 0.337,
      "step": 27425
    },
    {
      "epoch": 1.5357822824504423,
      "grad_norm": 1.2406386137008667,
      "learning_rate": 9.804921052631579e-05,
      "loss": 0.417,
      "step": 27426
    },
    {
      "epoch": 1.5358382797625714,
      "grad_norm": 1.2913198471069336,
      "learning_rate": 9.804894736842106e-05,
      "loss": 0.4527,
      "step": 27427
    },
    {
      "epoch": 1.5358942770747004,
      "grad_norm": 1.506883978843689,
      "learning_rate": 9.804868421052632e-05,
      "loss": 0.4886,
      "step": 27428
    },
    {
      "epoch": 1.5359502743868294,
      "grad_norm": 1.262926697731018,
      "learning_rate": 9.804842105263158e-05,
      "loss": 0.5099,
      "step": 27429
    },
    {
      "epoch": 1.5360062716989584,
      "grad_norm": 1.320450782775879,
      "learning_rate": 9.804815789473684e-05,
      "loss": 0.6103,
      "step": 27430
    },
    {
      "epoch": 1.5360622690110874,
      "grad_norm": 1.2099217176437378,
      "learning_rate": 9.80478947368421e-05,
      "loss": 0.3378,
      "step": 27431
    },
    {
      "epoch": 1.5361182663232165,
      "grad_norm": 1.5453951358795166,
      "learning_rate": 9.804763157894738e-05,
      "loss": 0.4624,
      "step": 27432
    },
    {
      "epoch": 1.5361742636353455,
      "grad_norm": 1.3994524478912354,
      "learning_rate": 9.804736842105264e-05,
      "loss": 0.6331,
      "step": 27433
    },
    {
      "epoch": 1.5362302609474745,
      "grad_norm": 1.4725806713104248,
      "learning_rate": 9.80471052631579e-05,
      "loss": 0.52,
      "step": 27434
    },
    {
      "epoch": 1.5362862582596035,
      "grad_norm": 1.5783519744873047,
      "learning_rate": 9.804684210526316e-05,
      "loss": 0.5181,
      "step": 27435
    },
    {
      "epoch": 1.5363422555717325,
      "grad_norm": 1.4294320344924927,
      "learning_rate": 9.804657894736843e-05,
      "loss": 0.6509,
      "step": 27436
    },
    {
      "epoch": 1.5363982528838616,
      "grad_norm": 1.1189452409744263,
      "learning_rate": 9.804631578947369e-05,
      "loss": 0.367,
      "step": 27437
    },
    {
      "epoch": 1.5364542501959906,
      "grad_norm": 1.5972261428833008,
      "learning_rate": 9.804605263157896e-05,
      "loss": 0.5024,
      "step": 27438
    },
    {
      "epoch": 1.5365102475081196,
      "grad_norm": 1.3196470737457275,
      "learning_rate": 9.804578947368421e-05,
      "loss": 0.4849,
      "step": 27439
    },
    {
      "epoch": 1.5365662448202486,
      "grad_norm": 1.0912482738494873,
      "learning_rate": 9.804552631578948e-05,
      "loss": 0.3987,
      "step": 27440
    },
    {
      "epoch": 1.5366222421323776,
      "grad_norm": 2.2294299602508545,
      "learning_rate": 9.804526315789474e-05,
      "loss": 0.4531,
      "step": 27441
    },
    {
      "epoch": 1.5366782394445067,
      "grad_norm": 1.0957279205322266,
      "learning_rate": 9.804500000000001e-05,
      "loss": 0.4376,
      "step": 27442
    },
    {
      "epoch": 1.5367342367566357,
      "grad_norm": 1.2432575225830078,
      "learning_rate": 9.804473684210526e-05,
      "loss": 0.4539,
      "step": 27443
    },
    {
      "epoch": 1.5367902340687647,
      "grad_norm": 2.164496898651123,
      "learning_rate": 9.804447368421053e-05,
      "loss": 0.31,
      "step": 27444
    },
    {
      "epoch": 1.5368462313808937,
      "grad_norm": 1.7686548233032227,
      "learning_rate": 9.80442105263158e-05,
      "loss": 0.5001,
      "step": 27445
    },
    {
      "epoch": 1.5369022286930227,
      "grad_norm": 1.1818535327911377,
      "learning_rate": 9.804394736842105e-05,
      "loss": 0.4528,
      "step": 27446
    },
    {
      "epoch": 1.5369582260051518,
      "grad_norm": 1.6264586448669434,
      "learning_rate": 9.804368421052633e-05,
      "loss": 0.4355,
      "step": 27447
    },
    {
      "epoch": 1.5370142233172808,
      "grad_norm": 1.404054045677185,
      "learning_rate": 9.804342105263157e-05,
      "loss": 0.6259,
      "step": 27448
    },
    {
      "epoch": 1.5370702206294098,
      "grad_norm": 1.3645541667938232,
      "learning_rate": 9.804315789473685e-05,
      "loss": 0.6429,
      "step": 27449
    },
    {
      "epoch": 1.5371262179415388,
      "grad_norm": 1.200858235359192,
      "learning_rate": 9.80428947368421e-05,
      "loss": 0.3692,
      "step": 27450
    },
    {
      "epoch": 1.5371822152536678,
      "grad_norm": 1.3735047578811646,
      "learning_rate": 9.804263157894738e-05,
      "loss": 0.5971,
      "step": 27451
    },
    {
      "epoch": 1.5372382125657968,
      "grad_norm": 1.3319628238677979,
      "learning_rate": 9.804236842105264e-05,
      "loss": 0.4273,
      "step": 27452
    },
    {
      "epoch": 1.5372942098779259,
      "grad_norm": 1.5300577878952026,
      "learning_rate": 9.80421052631579e-05,
      "loss": 0.4289,
      "step": 27453
    },
    {
      "epoch": 1.5373502071900549,
      "grad_norm": 1.256272315979004,
      "learning_rate": 9.804184210526316e-05,
      "loss": 0.4217,
      "step": 27454
    },
    {
      "epoch": 1.537406204502184,
      "grad_norm": 1.1546261310577393,
      "learning_rate": 9.804157894736843e-05,
      "loss": 0.5448,
      "step": 27455
    },
    {
      "epoch": 1.537462201814313,
      "grad_norm": 1.232764482498169,
      "learning_rate": 9.804131578947369e-05,
      "loss": 0.4131,
      "step": 27456
    },
    {
      "epoch": 1.537518199126442,
      "grad_norm": 1.5305607318878174,
      "learning_rate": 9.804105263157895e-05,
      "loss": 0.4478,
      "step": 27457
    },
    {
      "epoch": 1.537574196438571,
      "grad_norm": 1.6348007917404175,
      "learning_rate": 9.804078947368421e-05,
      "loss": 0.6888,
      "step": 27458
    },
    {
      "epoch": 1.5376301937507,
      "grad_norm": 1.250065565109253,
      "learning_rate": 9.804052631578948e-05,
      "loss": 0.4321,
      "step": 27459
    },
    {
      "epoch": 1.537686191062829,
      "grad_norm": 1.3260490894317627,
      "learning_rate": 9.804026315789474e-05,
      "loss": 0.463,
      "step": 27460
    },
    {
      "epoch": 1.537742188374958,
      "grad_norm": 1.4980632066726685,
      "learning_rate": 9.804e-05,
      "loss": 0.6361,
      "step": 27461
    },
    {
      "epoch": 1.537798185687087,
      "grad_norm": 1.5763019323349,
      "learning_rate": 9.803973684210526e-05,
      "loss": 0.8689,
      "step": 27462
    },
    {
      "epoch": 1.537854182999216,
      "grad_norm": 1.4502410888671875,
      "learning_rate": 9.803947368421052e-05,
      "loss": 0.3842,
      "step": 27463
    },
    {
      "epoch": 1.537910180311345,
      "grad_norm": 1.3471466302871704,
      "learning_rate": 9.80392105263158e-05,
      "loss": 0.4361,
      "step": 27464
    },
    {
      "epoch": 1.537966177623474,
      "grad_norm": 1.247658610343933,
      "learning_rate": 9.803894736842106e-05,
      "loss": 0.531,
      "step": 27465
    },
    {
      "epoch": 1.5380221749356031,
      "grad_norm": 1.1767892837524414,
      "learning_rate": 9.803868421052632e-05,
      "loss": 0.3491,
      "step": 27466
    },
    {
      "epoch": 1.5380781722477321,
      "grad_norm": 1.3079240322113037,
      "learning_rate": 9.803842105263158e-05,
      "loss": 0.5244,
      "step": 27467
    },
    {
      "epoch": 1.5381341695598612,
      "grad_norm": 1.3583015203475952,
      "learning_rate": 9.803815789473685e-05,
      "loss": 0.4296,
      "step": 27468
    },
    {
      "epoch": 1.5381901668719902,
      "grad_norm": 1.3113908767700195,
      "learning_rate": 9.803789473684211e-05,
      "loss": 0.4516,
      "step": 27469
    },
    {
      "epoch": 1.5382461641841192,
      "grad_norm": 1.5479438304901123,
      "learning_rate": 9.803763157894738e-05,
      "loss": 0.5803,
      "step": 27470
    },
    {
      "epoch": 1.5383021614962482,
      "grad_norm": 1.4623644351959229,
      "learning_rate": 9.803736842105263e-05,
      "loss": 0.4005,
      "step": 27471
    },
    {
      "epoch": 1.5383581588083772,
      "grad_norm": 1.529299259185791,
      "learning_rate": 9.80371052631579e-05,
      "loss": 0.4526,
      "step": 27472
    },
    {
      "epoch": 1.5384141561205062,
      "grad_norm": 1.45881986618042,
      "learning_rate": 9.803684210526316e-05,
      "loss": 0.6338,
      "step": 27473
    },
    {
      "epoch": 1.5384701534326353,
      "grad_norm": 1.1933166980743408,
      "learning_rate": 9.803657894736843e-05,
      "loss": 0.4632,
      "step": 27474
    },
    {
      "epoch": 1.5385261507447643,
      "grad_norm": 1.365078330039978,
      "learning_rate": 9.80363157894737e-05,
      "loss": 0.4295,
      "step": 27475
    },
    {
      "epoch": 1.5385821480568933,
      "grad_norm": 1.6030665636062622,
      "learning_rate": 9.803605263157895e-05,
      "loss": 0.5091,
      "step": 27476
    },
    {
      "epoch": 1.5386381453690223,
      "grad_norm": 1.1716169118881226,
      "learning_rate": 9.803578947368421e-05,
      "loss": 0.4437,
      "step": 27477
    },
    {
      "epoch": 1.5386941426811513,
      "grad_norm": 1.7651629447937012,
      "learning_rate": 9.803552631578949e-05,
      "loss": 0.5739,
      "step": 27478
    },
    {
      "epoch": 1.5387501399932804,
      "grad_norm": 1.3289557695388794,
      "learning_rate": 9.803526315789475e-05,
      "loss": 0.4221,
      "step": 27479
    },
    {
      "epoch": 1.5388061373054094,
      "grad_norm": 1.4761066436767578,
      "learning_rate": 9.8035e-05,
      "loss": 0.4091,
      "step": 27480
    },
    {
      "epoch": 1.5388621346175384,
      "grad_norm": 1.7415555715560913,
      "learning_rate": 9.803473684210527e-05,
      "loss": 0.465,
      "step": 27481
    },
    {
      "epoch": 1.5389181319296674,
      "grad_norm": 1.260891318321228,
      "learning_rate": 9.803447368421053e-05,
      "loss": 0.4759,
      "step": 27482
    },
    {
      "epoch": 1.5389741292417964,
      "grad_norm": 1.1866629123687744,
      "learning_rate": 9.80342105263158e-05,
      "loss": 0.4661,
      "step": 27483
    },
    {
      "epoch": 1.5390301265539255,
      "grad_norm": 1.501732587814331,
      "learning_rate": 9.803394736842106e-05,
      "loss": 0.5099,
      "step": 27484
    },
    {
      "epoch": 1.5390861238660545,
      "grad_norm": 1.0346893072128296,
      "learning_rate": 9.803368421052632e-05,
      "loss": 0.3779,
      "step": 27485
    },
    {
      "epoch": 1.5391421211781835,
      "grad_norm": 1.6514413356781006,
      "learning_rate": 9.803342105263158e-05,
      "loss": 0.4359,
      "step": 27486
    },
    {
      "epoch": 1.5391981184903125,
      "grad_norm": 1.1998839378356934,
      "learning_rate": 9.803315789473685e-05,
      "loss": 0.4138,
      "step": 27487
    },
    {
      "epoch": 1.5392541158024415,
      "grad_norm": 1.340802550315857,
      "learning_rate": 9.803289473684211e-05,
      "loss": 0.5049,
      "step": 27488
    },
    {
      "epoch": 1.5393101131145706,
      "grad_norm": 1.6594573259353638,
      "learning_rate": 9.803263157894737e-05,
      "loss": 0.5302,
      "step": 27489
    },
    {
      "epoch": 1.5393661104266996,
      "grad_norm": 1.2872958183288574,
      "learning_rate": 9.803236842105263e-05,
      "loss": 0.5179,
      "step": 27490
    },
    {
      "epoch": 1.5394221077388286,
      "grad_norm": 1.357077717781067,
      "learning_rate": 9.80321052631579e-05,
      "loss": 0.4203,
      "step": 27491
    },
    {
      "epoch": 1.5394781050509576,
      "grad_norm": 1.2538336515426636,
      "learning_rate": 9.803184210526316e-05,
      "loss": 0.4895,
      "step": 27492
    },
    {
      "epoch": 1.5395341023630866,
      "grad_norm": 1.4729819297790527,
      "learning_rate": 9.803157894736844e-05,
      "loss": 0.5603,
      "step": 27493
    },
    {
      "epoch": 1.5395900996752157,
      "grad_norm": 1.2647981643676758,
      "learning_rate": 9.803131578947368e-05,
      "loss": 0.4322,
      "step": 27494
    },
    {
      "epoch": 1.5396460969873447,
      "grad_norm": 1.2713780403137207,
      "learning_rate": 9.803105263157896e-05,
      "loss": 0.4704,
      "step": 27495
    },
    {
      "epoch": 1.5397020942994737,
      "grad_norm": 1.1072604656219482,
      "learning_rate": 9.803078947368422e-05,
      "loss": 0.4073,
      "step": 27496
    },
    {
      "epoch": 1.5397580916116027,
      "grad_norm": 1.9605051279067993,
      "learning_rate": 9.803052631578949e-05,
      "loss": 0.4553,
      "step": 27497
    },
    {
      "epoch": 1.5398140889237317,
      "grad_norm": 1.4391065835952759,
      "learning_rate": 9.803026315789474e-05,
      "loss": 0.5179,
      "step": 27498
    },
    {
      "epoch": 1.5398700862358607,
      "grad_norm": 1.2579585313796997,
      "learning_rate": 9.803e-05,
      "loss": 0.4785,
      "step": 27499
    },
    {
      "epoch": 1.5399260835479898,
      "grad_norm": 1.3812254667282104,
      "learning_rate": 9.802973684210527e-05,
      "loss": 0.4546,
      "step": 27500
    },
    {
      "epoch": 1.5399820808601188,
      "grad_norm": 1.4660743474960327,
      "learning_rate": 9.802947368421053e-05,
      "loss": 0.4336,
      "step": 27501
    },
    {
      "epoch": 1.5400380781722478,
      "grad_norm": 1.2951329946517944,
      "learning_rate": 9.80292105263158e-05,
      "loss": 0.3834,
      "step": 27502
    },
    {
      "epoch": 1.5400940754843768,
      "grad_norm": 1.3852564096450806,
      "learning_rate": 9.802894736842105e-05,
      "loss": 0.6174,
      "step": 27503
    },
    {
      "epoch": 1.5401500727965058,
      "grad_norm": 1.3390361070632935,
      "learning_rate": 9.802868421052632e-05,
      "loss": 0.573,
      "step": 27504
    },
    {
      "epoch": 1.5402060701086349,
      "grad_norm": 1.2705798149108887,
      "learning_rate": 9.802842105263158e-05,
      "loss": 0.3637,
      "step": 27505
    },
    {
      "epoch": 1.5402620674207639,
      "grad_norm": 1.1765118837356567,
      "learning_rate": 9.802815789473685e-05,
      "loss": 0.3652,
      "step": 27506
    },
    {
      "epoch": 1.540318064732893,
      "grad_norm": 1.1236757040023804,
      "learning_rate": 9.802789473684211e-05,
      "loss": 0.4043,
      "step": 27507
    },
    {
      "epoch": 1.540374062045022,
      "grad_norm": 1.32753586769104,
      "learning_rate": 9.802763157894737e-05,
      "loss": 0.4631,
      "step": 27508
    },
    {
      "epoch": 1.540430059357151,
      "grad_norm": 1.560129165649414,
      "learning_rate": 9.802736842105263e-05,
      "loss": 0.4467,
      "step": 27509
    },
    {
      "epoch": 1.54048605666928,
      "grad_norm": 1.3294612169265747,
      "learning_rate": 9.80271052631579e-05,
      "loss": 0.4659,
      "step": 27510
    },
    {
      "epoch": 1.540542053981409,
      "grad_norm": 1.3083581924438477,
      "learning_rate": 9.802684210526317e-05,
      "loss": 0.489,
      "step": 27511
    },
    {
      "epoch": 1.540598051293538,
      "grad_norm": 1.2318902015686035,
      "learning_rate": 9.802657894736843e-05,
      "loss": 0.4461,
      "step": 27512
    },
    {
      "epoch": 1.540654048605667,
      "grad_norm": 1.1843339204788208,
      "learning_rate": 9.802631578947369e-05,
      "loss": 0.4687,
      "step": 27513
    },
    {
      "epoch": 1.540710045917796,
      "grad_norm": 1.335110068321228,
      "learning_rate": 9.802605263157896e-05,
      "loss": 0.3646,
      "step": 27514
    },
    {
      "epoch": 1.540766043229925,
      "grad_norm": 1.146899700164795,
      "learning_rate": 9.802578947368422e-05,
      "loss": 0.4172,
      "step": 27515
    },
    {
      "epoch": 1.540822040542054,
      "grad_norm": 1.4995315074920654,
      "learning_rate": 9.802552631578948e-05,
      "loss": 0.4947,
      "step": 27516
    },
    {
      "epoch": 1.540878037854183,
      "grad_norm": 1.4839874505996704,
      "learning_rate": 9.802526315789474e-05,
      "loss": 0.4552,
      "step": 27517
    },
    {
      "epoch": 1.540934035166312,
      "grad_norm": 1.3880366086959839,
      "learning_rate": 9.8025e-05,
      "loss": 0.5506,
      "step": 27518
    },
    {
      "epoch": 1.5409900324784411,
      "grad_norm": 1.256199598312378,
      "learning_rate": 9.802473684210527e-05,
      "loss": 0.5615,
      "step": 27519
    },
    {
      "epoch": 1.5410460297905701,
      "grad_norm": 2.4414100646972656,
      "learning_rate": 9.802447368421053e-05,
      "loss": 0.4661,
      "step": 27520
    },
    {
      "epoch": 1.5411020271026992,
      "grad_norm": 1.535971999168396,
      "learning_rate": 9.802421052631579e-05,
      "loss": 0.5246,
      "step": 27521
    },
    {
      "epoch": 1.5411580244148282,
      "grad_norm": 1.4283818006515503,
      "learning_rate": 9.802394736842105e-05,
      "loss": 0.4692,
      "step": 27522
    },
    {
      "epoch": 1.5412140217269572,
      "grad_norm": 1.3877708911895752,
      "learning_rate": 9.802368421052632e-05,
      "loss": 0.5436,
      "step": 27523
    },
    {
      "epoch": 1.5412700190390862,
      "grad_norm": 1.5361599922180176,
      "learning_rate": 9.802342105263158e-05,
      "loss": 0.5783,
      "step": 27524
    },
    {
      "epoch": 1.5413260163512152,
      "grad_norm": 1.22494637966156,
      "learning_rate": 9.802315789473686e-05,
      "loss": 0.4429,
      "step": 27525
    },
    {
      "epoch": 1.5413820136633443,
      "grad_norm": 1.433502435684204,
      "learning_rate": 9.80228947368421e-05,
      "loss": 0.5215,
      "step": 27526
    },
    {
      "epoch": 1.5414380109754733,
      "grad_norm": 1.7901016473770142,
      "learning_rate": 9.802263157894738e-05,
      "loss": 0.4571,
      "step": 27527
    },
    {
      "epoch": 1.5414940082876023,
      "grad_norm": 1.2573442459106445,
      "learning_rate": 9.802236842105264e-05,
      "loss": 0.4885,
      "step": 27528
    },
    {
      "epoch": 1.5415500055997313,
      "grad_norm": 1.3599714040756226,
      "learning_rate": 9.802210526315791e-05,
      "loss": 0.32,
      "step": 27529
    },
    {
      "epoch": 1.5416060029118603,
      "grad_norm": 1.5652059316635132,
      "learning_rate": 9.802184210526317e-05,
      "loss": 0.4739,
      "step": 27530
    },
    {
      "epoch": 1.5416620002239894,
      "grad_norm": 1.3834737539291382,
      "learning_rate": 9.802157894736843e-05,
      "loss": 0.4994,
      "step": 27531
    },
    {
      "epoch": 1.5417179975361184,
      "grad_norm": 1.1089963912963867,
      "learning_rate": 9.802131578947369e-05,
      "loss": 0.4456,
      "step": 27532
    },
    {
      "epoch": 1.5417739948482474,
      "grad_norm": 1.2896775007247925,
      "learning_rate": 9.802105263157895e-05,
      "loss": 0.4007,
      "step": 27533
    },
    {
      "epoch": 1.5418299921603764,
      "grad_norm": 1.4454684257507324,
      "learning_rate": 9.802078947368422e-05,
      "loss": 0.4368,
      "step": 27534
    },
    {
      "epoch": 1.5418859894725054,
      "grad_norm": 1.4198540449142456,
      "learning_rate": 9.802052631578947e-05,
      "loss": 0.4667,
      "step": 27535
    },
    {
      "epoch": 1.5419419867846345,
      "grad_norm": 1.478713035583496,
      "learning_rate": 9.802026315789474e-05,
      "loss": 0.6273,
      "step": 27536
    },
    {
      "epoch": 1.5419979840967635,
      "grad_norm": 1.2896050214767456,
      "learning_rate": 9.802e-05,
      "loss": 0.4453,
      "step": 27537
    },
    {
      "epoch": 1.5420539814088925,
      "grad_norm": 1.3719547986984253,
      "learning_rate": 9.801973684210527e-05,
      "loss": 0.5107,
      "step": 27538
    },
    {
      "epoch": 1.5421099787210215,
      "grad_norm": 1.1835438013076782,
      "learning_rate": 9.801947368421053e-05,
      "loss": 0.4458,
      "step": 27539
    },
    {
      "epoch": 1.5421659760331505,
      "grad_norm": 1.4666827917099,
      "learning_rate": 9.801921052631579e-05,
      "loss": 0.5352,
      "step": 27540
    },
    {
      "epoch": 1.5422219733452796,
      "grad_norm": 1.3444392681121826,
      "learning_rate": 9.801894736842105e-05,
      "loss": 0.4657,
      "step": 27541
    },
    {
      "epoch": 1.5422779706574086,
      "grad_norm": 1.0673942565917969,
      "learning_rate": 9.801868421052633e-05,
      "loss": 0.4011,
      "step": 27542
    },
    {
      "epoch": 1.5423339679695376,
      "grad_norm": 1.4041234254837036,
      "learning_rate": 9.801842105263159e-05,
      "loss": 0.4219,
      "step": 27543
    },
    {
      "epoch": 1.5423899652816666,
      "grad_norm": 1.3166135549545288,
      "learning_rate": 9.801815789473685e-05,
      "loss": 0.3867,
      "step": 27544
    },
    {
      "epoch": 1.5424459625937956,
      "grad_norm": 1.516274094581604,
      "learning_rate": 9.80178947368421e-05,
      "loss": 0.4241,
      "step": 27545
    },
    {
      "epoch": 1.5425019599059246,
      "grad_norm": 1.1992155313491821,
      "learning_rate": 9.801763157894738e-05,
      "loss": 0.3902,
      "step": 27546
    },
    {
      "epoch": 1.5425579572180537,
      "grad_norm": 1.3640427589416504,
      "learning_rate": 9.801736842105264e-05,
      "loss": 0.5306,
      "step": 27547
    },
    {
      "epoch": 1.5426139545301827,
      "grad_norm": 1.326397180557251,
      "learning_rate": 9.801710526315791e-05,
      "loss": 0.4747,
      "step": 27548
    },
    {
      "epoch": 1.5426699518423117,
      "grad_norm": 1.6014254093170166,
      "learning_rate": 9.801684210526316e-05,
      "loss": 0.4637,
      "step": 27549
    },
    {
      "epoch": 1.5427259491544407,
      "grad_norm": 1.4529451131820679,
      "learning_rate": 9.801657894736842e-05,
      "loss": 0.4927,
      "step": 27550
    },
    {
      "epoch": 1.5427819464665697,
      "grad_norm": 1.4845919609069824,
      "learning_rate": 9.801631578947369e-05,
      "loss": 0.4053,
      "step": 27551
    },
    {
      "epoch": 1.5428379437786988,
      "grad_norm": 1.4558218717575073,
      "learning_rate": 9.801605263157895e-05,
      "loss": 0.572,
      "step": 27552
    },
    {
      "epoch": 1.5428939410908278,
      "grad_norm": 1.2350249290466309,
      "learning_rate": 9.801578947368421e-05,
      "loss": 0.324,
      "step": 27553
    },
    {
      "epoch": 1.5429499384029568,
      "grad_norm": 1.2948768138885498,
      "learning_rate": 9.801552631578947e-05,
      "loss": 0.6895,
      "step": 27554
    },
    {
      "epoch": 1.5430059357150858,
      "grad_norm": 1.2931283712387085,
      "learning_rate": 9.801526315789474e-05,
      "loss": 0.4341,
      "step": 27555
    },
    {
      "epoch": 1.5430619330272148,
      "grad_norm": 1.2458338737487793,
      "learning_rate": 9.8015e-05,
      "loss": 0.4292,
      "step": 27556
    },
    {
      "epoch": 1.5431179303393439,
      "grad_norm": 1.4769642353057861,
      "learning_rate": 9.801473684210528e-05,
      "loss": 0.7012,
      "step": 27557
    },
    {
      "epoch": 1.5431739276514729,
      "grad_norm": 1.3621057271957397,
      "learning_rate": 9.801447368421052e-05,
      "loss": 0.5053,
      "step": 27558
    },
    {
      "epoch": 1.543229924963602,
      "grad_norm": 1.2167919874191284,
      "learning_rate": 9.80142105263158e-05,
      "loss": 0.3819,
      "step": 27559
    },
    {
      "epoch": 1.543285922275731,
      "grad_norm": 1.462602972984314,
      "learning_rate": 9.801394736842106e-05,
      "loss": 0.474,
      "step": 27560
    },
    {
      "epoch": 1.54334191958786,
      "grad_norm": 1.011000633239746,
      "learning_rate": 9.801368421052633e-05,
      "loss": 0.3042,
      "step": 27561
    },
    {
      "epoch": 1.543397916899989,
      "grad_norm": 1.2537404298782349,
      "learning_rate": 9.801342105263159e-05,
      "loss": 0.4026,
      "step": 27562
    },
    {
      "epoch": 1.543453914212118,
      "grad_norm": 1.2451791763305664,
      "learning_rate": 9.801315789473685e-05,
      "loss": 0.4505,
      "step": 27563
    },
    {
      "epoch": 1.543509911524247,
      "grad_norm": 1.3682466745376587,
      "learning_rate": 9.801289473684211e-05,
      "loss": 0.4357,
      "step": 27564
    },
    {
      "epoch": 1.543565908836376,
      "grad_norm": 1.4733965396881104,
      "learning_rate": 9.801263157894738e-05,
      "loss": 0.4874,
      "step": 27565
    },
    {
      "epoch": 1.543621906148505,
      "grad_norm": 1.4038304090499878,
      "learning_rate": 9.801236842105264e-05,
      "loss": 0.4441,
      "step": 27566
    },
    {
      "epoch": 1.543677903460634,
      "grad_norm": 3.1235392093658447,
      "learning_rate": 9.80121052631579e-05,
      "loss": 0.4708,
      "step": 27567
    },
    {
      "epoch": 1.543733900772763,
      "grad_norm": 1.4425427913665771,
      "learning_rate": 9.801184210526316e-05,
      "loss": 0.518,
      "step": 27568
    },
    {
      "epoch": 1.543789898084892,
      "grad_norm": 1.268245816230774,
      "learning_rate": 9.801157894736842e-05,
      "loss": 0.6169,
      "step": 27569
    },
    {
      "epoch": 1.543845895397021,
      "grad_norm": 1.3493518829345703,
      "learning_rate": 9.801131578947369e-05,
      "loss": 0.4883,
      "step": 27570
    },
    {
      "epoch": 1.5439018927091501,
      "grad_norm": 1.2586694955825806,
      "learning_rate": 9.801105263157895e-05,
      "loss": 0.5377,
      "step": 27571
    },
    {
      "epoch": 1.5439578900212791,
      "grad_norm": 1.302017092704773,
      "learning_rate": 9.801078947368421e-05,
      "loss": 0.463,
      "step": 27572
    },
    {
      "epoch": 1.544013887333408,
      "grad_norm": 34.4649658203125,
      "learning_rate": 9.801052631578947e-05,
      "loss": 0.4456,
      "step": 27573
    },
    {
      "epoch": 1.544069884645537,
      "grad_norm": 1.345123291015625,
      "learning_rate": 9.801026315789475e-05,
      "loss": 0.4977,
      "step": 27574
    },
    {
      "epoch": 1.544125881957666,
      "grad_norm": 1.382767677307129,
      "learning_rate": 9.801e-05,
      "loss": 0.4965,
      "step": 27575
    },
    {
      "epoch": 1.544181879269795,
      "grad_norm": 1.341180443763733,
      "learning_rate": 9.800973684210526e-05,
      "loss": 0.4113,
      "step": 27576
    },
    {
      "epoch": 1.544237876581924,
      "grad_norm": 1.2358497381210327,
      "learning_rate": 9.800947368421052e-05,
      "loss": 0.5991,
      "step": 27577
    },
    {
      "epoch": 1.544293873894053,
      "grad_norm": 1.458097219467163,
      "learning_rate": 9.80092105263158e-05,
      "loss": 0.5126,
      "step": 27578
    },
    {
      "epoch": 1.544349871206182,
      "grad_norm": 1.3169153928756714,
      "learning_rate": 9.800894736842106e-05,
      "loss": 0.4325,
      "step": 27579
    },
    {
      "epoch": 1.544405868518311,
      "grad_norm": 1.228348970413208,
      "learning_rate": 9.800868421052633e-05,
      "loss": 0.3367,
      "step": 27580
    },
    {
      "epoch": 1.54446186583044,
      "grad_norm": 1.3079302310943604,
      "learning_rate": 9.800842105263158e-05,
      "loss": 0.4229,
      "step": 27581
    },
    {
      "epoch": 1.5445178631425691,
      "grad_norm": 1.2360742092132568,
      "learning_rate": 9.800815789473685e-05,
      "loss": 0.4752,
      "step": 27582
    },
    {
      "epoch": 1.5445738604546981,
      "grad_norm": 1.5535396337509155,
      "learning_rate": 9.800789473684211e-05,
      "loss": 0.4473,
      "step": 27583
    },
    {
      "epoch": 1.5446298577668272,
      "grad_norm": 1.1982449293136597,
      "learning_rate": 9.800763157894737e-05,
      "loss": 0.416,
      "step": 27584
    },
    {
      "epoch": 1.5446858550789562,
      "grad_norm": 1.2752082347869873,
      "learning_rate": 9.800736842105264e-05,
      "loss": 0.6395,
      "step": 27585
    },
    {
      "epoch": 1.5447418523910852,
      "grad_norm": 1.3502033948898315,
      "learning_rate": 9.800710526315789e-05,
      "loss": 0.4872,
      "step": 27586
    },
    {
      "epoch": 1.5447978497032142,
      "grad_norm": 1.4935429096221924,
      "learning_rate": 9.800684210526316e-05,
      "loss": 0.441,
      "step": 27587
    },
    {
      "epoch": 1.5448538470153432,
      "grad_norm": 1.4000657796859741,
      "learning_rate": 9.800657894736842e-05,
      "loss": 0.5606,
      "step": 27588
    },
    {
      "epoch": 1.5449098443274722,
      "grad_norm": 1.6481719017028809,
      "learning_rate": 9.80063157894737e-05,
      "loss": 0.4268,
      "step": 27589
    },
    {
      "epoch": 1.5449658416396013,
      "grad_norm": 1.4190216064453125,
      "learning_rate": 9.800605263157894e-05,
      "loss": 0.4317,
      "step": 27590
    },
    {
      "epoch": 1.5450218389517303,
      "grad_norm": 1.8343473672866821,
      "learning_rate": 9.800578947368422e-05,
      "loss": 0.4894,
      "step": 27591
    },
    {
      "epoch": 1.5450778362638593,
      "grad_norm": 1.4488444328308105,
      "learning_rate": 9.800552631578947e-05,
      "loss": 0.4326,
      "step": 27592
    },
    {
      "epoch": 1.5451338335759883,
      "grad_norm": 1.3605916500091553,
      "learning_rate": 9.800526315789475e-05,
      "loss": 0.4197,
      "step": 27593
    },
    {
      "epoch": 1.5451898308881173,
      "grad_norm": 1.5493638515472412,
      "learning_rate": 9.800500000000001e-05,
      "loss": 0.4657,
      "step": 27594
    },
    {
      "epoch": 1.5452458282002464,
      "grad_norm": 2.3098866939544678,
      "learning_rate": 9.800473684210527e-05,
      "loss": 0.437,
      "step": 27595
    },
    {
      "epoch": 1.5453018255123754,
      "grad_norm": 1.4588706493377686,
      "learning_rate": 9.800447368421053e-05,
      "loss": 0.4434,
      "step": 27596
    },
    {
      "epoch": 1.5453578228245044,
      "grad_norm": 1.5190616846084595,
      "learning_rate": 9.80042105263158e-05,
      "loss": 0.3889,
      "step": 27597
    },
    {
      "epoch": 1.5454138201366334,
      "grad_norm": 1.4056833982467651,
      "learning_rate": 9.800394736842106e-05,
      "loss": 0.4166,
      "step": 27598
    },
    {
      "epoch": 1.5454698174487624,
      "grad_norm": 1.3816081285476685,
      "learning_rate": 9.800368421052632e-05,
      "loss": 0.4344,
      "step": 27599
    },
    {
      "epoch": 1.5455258147608915,
      "grad_norm": 1.176428198814392,
      "learning_rate": 9.800342105263158e-05,
      "loss": 0.4996,
      "step": 27600
    },
    {
      "epoch": 1.5455818120730205,
      "grad_norm": 1.3347351551055908,
      "learning_rate": 9.800315789473685e-05,
      "loss": 0.641,
      "step": 27601
    },
    {
      "epoch": 1.5456378093851495,
      "grad_norm": 1.6254616975784302,
      "learning_rate": 9.800289473684211e-05,
      "loss": 0.567,
      "step": 27602
    },
    {
      "epoch": 1.5456938066972785,
      "grad_norm": 1.2517011165618896,
      "learning_rate": 9.800263157894737e-05,
      "loss": 0.3868,
      "step": 27603
    },
    {
      "epoch": 1.5457498040094075,
      "grad_norm": 1.9341506958007812,
      "learning_rate": 9.800236842105263e-05,
      "loss": 0.479,
      "step": 27604
    },
    {
      "epoch": 1.5458058013215366,
      "grad_norm": 1.3913823366165161,
      "learning_rate": 9.800210526315789e-05,
      "loss": 0.4766,
      "step": 27605
    },
    {
      "epoch": 1.5458617986336656,
      "grad_norm": 1.4235960245132446,
      "learning_rate": 9.800184210526317e-05,
      "loss": 0.5779,
      "step": 27606
    },
    {
      "epoch": 1.5459177959457946,
      "grad_norm": 1.5019274950027466,
      "learning_rate": 9.800157894736842e-05,
      "loss": 0.3857,
      "step": 27607
    },
    {
      "epoch": 1.5459737932579236,
      "grad_norm": 1.4286653995513916,
      "learning_rate": 9.800131578947368e-05,
      "loss": 0.4655,
      "step": 27608
    },
    {
      "epoch": 1.5460297905700526,
      "grad_norm": 1.2873685359954834,
      "learning_rate": 9.800105263157894e-05,
      "loss": 0.4759,
      "step": 27609
    },
    {
      "epoch": 1.5460857878821817,
      "grad_norm": 1.1733746528625488,
      "learning_rate": 9.800078947368422e-05,
      "loss": 0.4239,
      "step": 27610
    },
    {
      "epoch": 1.5461417851943107,
      "grad_norm": 1.183221697807312,
      "learning_rate": 9.800052631578948e-05,
      "loss": 0.3502,
      "step": 27611
    },
    {
      "epoch": 1.5461977825064397,
      "grad_norm": 1.1111266613006592,
      "learning_rate": 9.800026315789475e-05,
      "loss": 0.4377,
      "step": 27612
    },
    {
      "epoch": 1.5462537798185687,
      "grad_norm": 1.2624726295471191,
      "learning_rate": 9.8e-05,
      "loss": 0.5593,
      "step": 27613
    },
    {
      "epoch": 1.5463097771306977,
      "grad_norm": 1.170574426651001,
      "learning_rate": 9.799973684210527e-05,
      "loss": 0.5029,
      "step": 27614
    },
    {
      "epoch": 1.5463657744428267,
      "grad_norm": 1.469617247581482,
      "learning_rate": 9.799947368421053e-05,
      "loss": 0.4209,
      "step": 27615
    },
    {
      "epoch": 1.5464217717549558,
      "grad_norm": 1.3512747287750244,
      "learning_rate": 9.79992105263158e-05,
      "loss": 0.4437,
      "step": 27616
    },
    {
      "epoch": 1.5464777690670848,
      "grad_norm": 1.5204124450683594,
      "learning_rate": 9.799894736842106e-05,
      "loss": 0.5534,
      "step": 27617
    },
    {
      "epoch": 1.5465337663792138,
      "grad_norm": 1.490783929824829,
      "learning_rate": 9.799868421052632e-05,
      "loss": 0.5079,
      "step": 27618
    },
    {
      "epoch": 1.5465897636913428,
      "grad_norm": 1.3356685638427734,
      "learning_rate": 9.799842105263158e-05,
      "loss": 0.4309,
      "step": 27619
    },
    {
      "epoch": 1.5466457610034718,
      "grad_norm": 1.7717808485031128,
      "learning_rate": 9.799815789473684e-05,
      "loss": 0.4072,
      "step": 27620
    },
    {
      "epoch": 1.5467017583156009,
      "grad_norm": 1.167535662651062,
      "learning_rate": 9.799789473684212e-05,
      "loss": 0.3819,
      "step": 27621
    },
    {
      "epoch": 1.5467577556277299,
      "grad_norm": 1.2569087743759155,
      "learning_rate": 9.799763157894738e-05,
      "loss": 0.3743,
      "step": 27622
    },
    {
      "epoch": 1.546813752939859,
      "grad_norm": 1.3656669855117798,
      "learning_rate": 9.799736842105263e-05,
      "loss": 0.4982,
      "step": 27623
    },
    {
      "epoch": 1.546869750251988,
      "grad_norm": 1.3347703218460083,
      "learning_rate": 9.79971052631579e-05,
      "loss": 0.4499,
      "step": 27624
    },
    {
      "epoch": 1.546925747564117,
      "grad_norm": 1.5180779695510864,
      "learning_rate": 9.799684210526317e-05,
      "loss": 0.4667,
      "step": 27625
    },
    {
      "epoch": 1.546981744876246,
      "grad_norm": 1.380881428718567,
      "learning_rate": 9.799657894736843e-05,
      "loss": 0.3848,
      "step": 27626
    },
    {
      "epoch": 1.547037742188375,
      "grad_norm": 1.9279532432556152,
      "learning_rate": 9.799631578947369e-05,
      "loss": 0.7991,
      "step": 27627
    },
    {
      "epoch": 1.547093739500504,
      "grad_norm": 2.0329720973968506,
      "learning_rate": 9.799605263157895e-05,
      "loss": 0.4458,
      "step": 27628
    },
    {
      "epoch": 1.547149736812633,
      "grad_norm": 1.4294487237930298,
      "learning_rate": 9.799578947368422e-05,
      "loss": 0.4432,
      "step": 27629
    },
    {
      "epoch": 1.547205734124762,
      "grad_norm": 2.325901985168457,
      "learning_rate": 9.799552631578948e-05,
      "loss": 0.6328,
      "step": 27630
    },
    {
      "epoch": 1.547261731436891,
      "grad_norm": 1.354583978652954,
      "learning_rate": 9.799526315789474e-05,
      "loss": 0.3689,
      "step": 27631
    },
    {
      "epoch": 1.54731772874902,
      "grad_norm": 1.095689296722412,
      "learning_rate": 9.7995e-05,
      "loss": 0.3846,
      "step": 27632
    },
    {
      "epoch": 1.547373726061149,
      "grad_norm": 1.3646684885025024,
      "learning_rate": 9.799473684210527e-05,
      "loss": 0.5064,
      "step": 27633
    },
    {
      "epoch": 1.547429723373278,
      "grad_norm": 1.3029592037200928,
      "learning_rate": 9.799447368421053e-05,
      "loss": 0.5023,
      "step": 27634
    },
    {
      "epoch": 1.5474857206854071,
      "grad_norm": 1.477940320968628,
      "learning_rate": 9.79942105263158e-05,
      "loss": 0.454,
      "step": 27635
    },
    {
      "epoch": 1.5475417179975361,
      "grad_norm": 1.6404883861541748,
      "learning_rate": 9.799394736842105e-05,
      "loss": 0.4442,
      "step": 27636
    },
    {
      "epoch": 1.5475977153096652,
      "grad_norm": 1.6449779272079468,
      "learning_rate": 9.799368421052631e-05,
      "loss": 0.5236,
      "step": 27637
    },
    {
      "epoch": 1.5476537126217942,
      "grad_norm": 1.4596631526947021,
      "learning_rate": 9.799342105263158e-05,
      "loss": 0.3402,
      "step": 27638
    },
    {
      "epoch": 1.5477097099339232,
      "grad_norm": 1.3849345445632935,
      "learning_rate": 9.799315789473684e-05,
      "loss": 0.3877,
      "step": 27639
    },
    {
      "epoch": 1.5477657072460522,
      "grad_norm": 1.363826870918274,
      "learning_rate": 9.799289473684212e-05,
      "loss": 0.4869,
      "step": 27640
    },
    {
      "epoch": 1.5478217045581812,
      "grad_norm": 1.2477645874023438,
      "learning_rate": 9.799263157894736e-05,
      "loss": 0.4483,
      "step": 27641
    },
    {
      "epoch": 1.5478777018703103,
      "grad_norm": 1.1474741697311401,
      "learning_rate": 9.799236842105264e-05,
      "loss": 0.3734,
      "step": 27642
    },
    {
      "epoch": 1.5479336991824393,
      "grad_norm": 1.4541995525360107,
      "learning_rate": 9.79921052631579e-05,
      "loss": 0.4592,
      "step": 27643
    },
    {
      "epoch": 1.5479896964945683,
      "grad_norm": 1.234514832496643,
      "learning_rate": 9.799184210526317e-05,
      "loss": 0.5405,
      "step": 27644
    },
    {
      "epoch": 1.5480456938066973,
      "grad_norm": 1.9639506340026855,
      "learning_rate": 9.799157894736842e-05,
      "loss": 0.5765,
      "step": 27645
    },
    {
      "epoch": 1.5481016911188263,
      "grad_norm": 1.4586303234100342,
      "learning_rate": 9.799131578947369e-05,
      "loss": 0.5156,
      "step": 27646
    },
    {
      "epoch": 1.5481576884309554,
      "grad_norm": 1.235527515411377,
      "learning_rate": 9.799105263157895e-05,
      "loss": 0.4629,
      "step": 27647
    },
    {
      "epoch": 1.5482136857430844,
      "grad_norm": 1.2466105222702026,
      "learning_rate": 9.799078947368422e-05,
      "loss": 0.4021,
      "step": 27648
    },
    {
      "epoch": 1.5482696830552134,
      "grad_norm": 1.3709512948989868,
      "learning_rate": 9.799052631578948e-05,
      "loss": 0.4173,
      "step": 27649
    },
    {
      "epoch": 1.5483256803673424,
      "grad_norm": 1.4440704584121704,
      "learning_rate": 9.799026315789474e-05,
      "loss": 0.5999,
      "step": 27650
    },
    {
      "epoch": 1.5483816776794714,
      "grad_norm": 1.4406636953353882,
      "learning_rate": 9.799e-05,
      "loss": 0.4572,
      "step": 27651
    },
    {
      "epoch": 1.5484376749916005,
      "grad_norm": 1.3229143619537354,
      "learning_rate": 9.798973684210528e-05,
      "loss": 0.5996,
      "step": 27652
    },
    {
      "epoch": 1.5484936723037295,
      "grad_norm": 1.45196533203125,
      "learning_rate": 9.798947368421054e-05,
      "loss": 0.4739,
      "step": 27653
    },
    {
      "epoch": 1.5485496696158585,
      "grad_norm": 1.112792730331421,
      "learning_rate": 9.79892105263158e-05,
      "loss": 0.3605,
      "step": 27654
    },
    {
      "epoch": 1.5486056669279875,
      "grad_norm": 1.547577142715454,
      "learning_rate": 9.798894736842105e-05,
      "loss": 0.5193,
      "step": 27655
    },
    {
      "epoch": 1.5486616642401163,
      "grad_norm": 1.2999566793441772,
      "learning_rate": 9.798868421052631e-05,
      "loss": 0.404,
      "step": 27656
    },
    {
      "epoch": 1.5487176615522453,
      "grad_norm": 1.3730603456497192,
      "learning_rate": 9.798842105263159e-05,
      "loss": 0.442,
      "step": 27657
    },
    {
      "epoch": 1.5487736588643743,
      "grad_norm": 1.4913021326065063,
      "learning_rate": 9.798815789473685e-05,
      "loss": 0.4299,
      "step": 27658
    },
    {
      "epoch": 1.5488296561765034,
      "grad_norm": 1.3966526985168457,
      "learning_rate": 9.798789473684211e-05,
      "loss": 0.4235,
      "step": 27659
    },
    {
      "epoch": 1.5488856534886324,
      "grad_norm": 1.2151272296905518,
      "learning_rate": 9.798763157894737e-05,
      "loss": 0.4361,
      "step": 27660
    },
    {
      "epoch": 1.5489416508007614,
      "grad_norm": 1.390045404434204,
      "learning_rate": 9.798736842105264e-05,
      "loss": 0.4848,
      "step": 27661
    },
    {
      "epoch": 1.5489976481128904,
      "grad_norm": 1.304226279258728,
      "learning_rate": 9.79871052631579e-05,
      "loss": 0.3626,
      "step": 27662
    },
    {
      "epoch": 1.5490536454250194,
      "grad_norm": 1.1964772939682007,
      "learning_rate": 9.798684210526316e-05,
      "loss": 0.3906,
      "step": 27663
    },
    {
      "epoch": 1.5491096427371485,
      "grad_norm": 1.7195937633514404,
      "learning_rate": 9.798657894736842e-05,
      "loss": 0.5034,
      "step": 27664
    },
    {
      "epoch": 1.5491656400492775,
      "grad_norm": 1.2917834520339966,
      "learning_rate": 9.798631578947369e-05,
      "loss": 0.419,
      "step": 27665
    },
    {
      "epoch": 1.5492216373614065,
      "grad_norm": 1.176645278930664,
      "learning_rate": 9.798605263157895e-05,
      "loss": 0.4128,
      "step": 27666
    },
    {
      "epoch": 1.5492776346735355,
      "grad_norm": 1.359494924545288,
      "learning_rate": 9.798578947368423e-05,
      "loss": 0.4341,
      "step": 27667
    },
    {
      "epoch": 1.5493336319856645,
      "grad_norm": 1.2506375312805176,
      "learning_rate": 9.798552631578947e-05,
      "loss": 0.3823,
      "step": 27668
    },
    {
      "epoch": 1.5493896292977936,
      "grad_norm": 1.3744421005249023,
      "learning_rate": 9.798526315789474e-05,
      "loss": 0.5909,
      "step": 27669
    },
    {
      "epoch": 1.5494456266099226,
      "grad_norm": 1.2375952005386353,
      "learning_rate": 9.7985e-05,
      "loss": 0.63,
      "step": 27670
    },
    {
      "epoch": 1.5495016239220516,
      "grad_norm": 1.6736398935317993,
      "learning_rate": 9.798473684210526e-05,
      "loss": 0.6227,
      "step": 27671
    },
    {
      "epoch": 1.5495576212341806,
      "grad_norm": 1.6351205110549927,
      "learning_rate": 9.798447368421054e-05,
      "loss": 0.6249,
      "step": 27672
    },
    {
      "epoch": 1.5496136185463096,
      "grad_norm": 1.5866144895553589,
      "learning_rate": 9.798421052631578e-05,
      "loss": 0.6634,
      "step": 27673
    },
    {
      "epoch": 1.5496696158584387,
      "grad_norm": 1.447206974029541,
      "learning_rate": 9.798394736842106e-05,
      "loss": 0.4198,
      "step": 27674
    },
    {
      "epoch": 1.5497256131705677,
      "grad_norm": 1.0121902227401733,
      "learning_rate": 9.798368421052632e-05,
      "loss": 0.3474,
      "step": 27675
    },
    {
      "epoch": 1.5497816104826967,
      "grad_norm": 1.433874487876892,
      "learning_rate": 9.798342105263159e-05,
      "loss": 0.4346,
      "step": 27676
    },
    {
      "epoch": 1.5498376077948257,
      "grad_norm": 1.1715593338012695,
      "learning_rate": 9.798315789473685e-05,
      "loss": 0.4513,
      "step": 27677
    },
    {
      "epoch": 1.5498936051069547,
      "grad_norm": 1.19560968875885,
      "learning_rate": 9.798289473684211e-05,
      "loss": 0.4297,
      "step": 27678
    },
    {
      "epoch": 1.5499496024190837,
      "grad_norm": 1.3347197771072388,
      "learning_rate": 9.798263157894737e-05,
      "loss": 0.3472,
      "step": 27679
    },
    {
      "epoch": 1.5500055997312128,
      "grad_norm": 1.138706922531128,
      "learning_rate": 9.798236842105264e-05,
      "loss": 0.375,
      "step": 27680
    },
    {
      "epoch": 1.5500615970433418,
      "grad_norm": 1.112441062927246,
      "learning_rate": 9.79821052631579e-05,
      "loss": 0.4377,
      "step": 27681
    },
    {
      "epoch": 1.5501175943554708,
      "grad_norm": 1.169978380203247,
      "learning_rate": 9.798184210526316e-05,
      "loss": 0.4607,
      "step": 27682
    },
    {
      "epoch": 1.5501735916675998,
      "grad_norm": 1.4009206295013428,
      "learning_rate": 9.798157894736842e-05,
      "loss": 0.488,
      "step": 27683
    },
    {
      "epoch": 1.5502295889797288,
      "grad_norm": 1.5235466957092285,
      "learning_rate": 9.79813157894737e-05,
      "loss": 0.3927,
      "step": 27684
    },
    {
      "epoch": 1.5502855862918579,
      "grad_norm": 1.2733972072601318,
      "learning_rate": 9.798105263157895e-05,
      "loss": 0.4078,
      "step": 27685
    },
    {
      "epoch": 1.5503415836039869,
      "grad_norm": 1.9851861000061035,
      "learning_rate": 9.798078947368421e-05,
      "loss": 0.4431,
      "step": 27686
    },
    {
      "epoch": 1.550397580916116,
      "grad_norm": 1.2623870372772217,
      "learning_rate": 9.798052631578947e-05,
      "loss": 0.4077,
      "step": 27687
    },
    {
      "epoch": 1.550453578228245,
      "grad_norm": 1.3079915046691895,
      "learning_rate": 9.798026315789473e-05,
      "loss": 0.3512,
      "step": 27688
    },
    {
      "epoch": 1.550509575540374,
      "grad_norm": 1.4924261569976807,
      "learning_rate": 9.798000000000001e-05,
      "loss": 0.5004,
      "step": 27689
    },
    {
      "epoch": 1.550565572852503,
      "grad_norm": 1.1912822723388672,
      "learning_rate": 9.797973684210527e-05,
      "loss": 0.4008,
      "step": 27690
    },
    {
      "epoch": 1.550621570164632,
      "grad_norm": 1.5061736106872559,
      "learning_rate": 9.797947368421053e-05,
      "loss": 0.4683,
      "step": 27691
    },
    {
      "epoch": 1.550677567476761,
      "grad_norm": 1.206244707107544,
      "learning_rate": 9.797921052631579e-05,
      "loss": 0.4397,
      "step": 27692
    },
    {
      "epoch": 1.55073356478889,
      "grad_norm": 1.3236454725265503,
      "learning_rate": 9.797894736842106e-05,
      "loss": 0.416,
      "step": 27693
    },
    {
      "epoch": 1.550789562101019,
      "grad_norm": 1.4528212547302246,
      "learning_rate": 9.797868421052632e-05,
      "loss": 0.6181,
      "step": 27694
    },
    {
      "epoch": 1.550845559413148,
      "grad_norm": 1.3994133472442627,
      "learning_rate": 9.797842105263159e-05,
      "loss": 0.4998,
      "step": 27695
    },
    {
      "epoch": 1.550901556725277,
      "grad_norm": 1.1481215953826904,
      "learning_rate": 9.797815789473684e-05,
      "loss": 0.3476,
      "step": 27696
    },
    {
      "epoch": 1.550957554037406,
      "grad_norm": 1.178155779838562,
      "learning_rate": 9.797789473684211e-05,
      "loss": 0.3862,
      "step": 27697
    },
    {
      "epoch": 1.5510135513495351,
      "grad_norm": 1.1070185899734497,
      "learning_rate": 9.797763157894737e-05,
      "loss": 0.5307,
      "step": 27698
    },
    {
      "epoch": 1.5510695486616641,
      "grad_norm": 1.3176507949829102,
      "learning_rate": 9.797736842105265e-05,
      "loss": 0.3921,
      "step": 27699
    },
    {
      "epoch": 1.5511255459737932,
      "grad_norm": 1.0525776147842407,
      "learning_rate": 9.797710526315789e-05,
      "loss": 0.3283,
      "step": 27700
    },
    {
      "epoch": 1.5511815432859222,
      "grad_norm": 1.5223246812820435,
      "learning_rate": 9.797684210526316e-05,
      "loss": 0.5607,
      "step": 27701
    },
    {
      "epoch": 1.5512375405980512,
      "grad_norm": 1.5738961696624756,
      "learning_rate": 9.797657894736842e-05,
      "loss": 0.5562,
      "step": 27702
    },
    {
      "epoch": 1.5512935379101802,
      "grad_norm": 1.657384991645813,
      "learning_rate": 9.79763157894737e-05,
      "loss": 0.4221,
      "step": 27703
    },
    {
      "epoch": 1.5513495352223092,
      "grad_norm": 1.3452095985412598,
      "learning_rate": 9.797605263157896e-05,
      "loss": 0.483,
      "step": 27704
    },
    {
      "epoch": 1.5514055325344382,
      "grad_norm": 1.2362409830093384,
      "learning_rate": 9.79757894736842e-05,
      "loss": 0.3704,
      "step": 27705
    },
    {
      "epoch": 1.5514615298465673,
      "grad_norm": 1.1338156461715698,
      "learning_rate": 9.797552631578948e-05,
      "loss": 0.4091,
      "step": 27706
    },
    {
      "epoch": 1.5515175271586963,
      "grad_norm": 1.4207375049591064,
      "learning_rate": 9.797526315789474e-05,
      "loss": 0.4474,
      "step": 27707
    },
    {
      "epoch": 1.5515735244708253,
      "grad_norm": 1.397441029548645,
      "learning_rate": 9.797500000000001e-05,
      "loss": 0.4012,
      "step": 27708
    },
    {
      "epoch": 1.5516295217829543,
      "grad_norm": 1.2236764430999756,
      "learning_rate": 9.797473684210527e-05,
      "loss": 0.3869,
      "step": 27709
    },
    {
      "epoch": 1.5516855190950833,
      "grad_norm": 1.1548060178756714,
      "learning_rate": 9.797447368421053e-05,
      "loss": 0.3724,
      "step": 27710
    },
    {
      "epoch": 1.5517415164072124,
      "grad_norm": 1.5421934127807617,
      "learning_rate": 9.797421052631579e-05,
      "loss": 0.56,
      "step": 27711
    },
    {
      "epoch": 1.5517975137193414,
      "grad_norm": 1.4410067796707153,
      "learning_rate": 9.797394736842106e-05,
      "loss": 0.4808,
      "step": 27712
    },
    {
      "epoch": 1.5518535110314704,
      "grad_norm": 1.3346655368804932,
      "learning_rate": 9.797368421052632e-05,
      "loss": 0.4176,
      "step": 27713
    },
    {
      "epoch": 1.5519095083435994,
      "grad_norm": 1.3503823280334473,
      "learning_rate": 9.797342105263158e-05,
      "loss": 0.5093,
      "step": 27714
    },
    {
      "epoch": 1.5519655056557284,
      "grad_norm": 1.340509057044983,
      "learning_rate": 9.797315789473684e-05,
      "loss": 0.4658,
      "step": 27715
    },
    {
      "epoch": 1.5520215029678575,
      "grad_norm": 1.1793127059936523,
      "learning_rate": 9.797289473684211e-05,
      "loss": 0.4391,
      "step": 27716
    },
    {
      "epoch": 1.5520775002799865,
      "grad_norm": 1.1974103450775146,
      "learning_rate": 9.797263157894737e-05,
      "loss": 0.3996,
      "step": 27717
    },
    {
      "epoch": 1.5521334975921155,
      "grad_norm": 1.266251564025879,
      "learning_rate": 9.797236842105263e-05,
      "loss": 0.4152,
      "step": 27718
    },
    {
      "epoch": 1.5521894949042445,
      "grad_norm": 2.003234386444092,
      "learning_rate": 9.79721052631579e-05,
      "loss": 0.5629,
      "step": 27719
    },
    {
      "epoch": 1.5522454922163735,
      "grad_norm": 1.327091097831726,
      "learning_rate": 9.797184210526317e-05,
      "loss": 0.3728,
      "step": 27720
    },
    {
      "epoch": 1.5523014895285026,
      "grad_norm": 1.35039484500885,
      "learning_rate": 9.797157894736843e-05,
      "loss": 0.4403,
      "step": 27721
    },
    {
      "epoch": 1.5523574868406316,
      "grad_norm": 1.0910850763320923,
      "learning_rate": 9.79713157894737e-05,
      "loss": 0.366,
      "step": 27722
    },
    {
      "epoch": 1.5524134841527606,
      "grad_norm": 1.8816014528274536,
      "learning_rate": 9.797105263157895e-05,
      "loss": 0.6915,
      "step": 27723
    },
    {
      "epoch": 1.5524694814648896,
      "grad_norm": 1.3539272546768188,
      "learning_rate": 9.79707894736842e-05,
      "loss": 0.4375,
      "step": 27724
    },
    {
      "epoch": 1.5525254787770186,
      "grad_norm": 1.5180089473724365,
      "learning_rate": 9.797052631578948e-05,
      "loss": 0.4702,
      "step": 27725
    },
    {
      "epoch": 1.5525814760891476,
      "grad_norm": 1.1890053749084473,
      "learning_rate": 9.797026315789474e-05,
      "loss": 0.3834,
      "step": 27726
    },
    {
      "epoch": 1.5526374734012767,
      "grad_norm": 1.3108999729156494,
      "learning_rate": 9.797000000000001e-05,
      "loss": 0.4571,
      "step": 27727
    },
    {
      "epoch": 1.5526934707134057,
      "grad_norm": 1.4722704887390137,
      "learning_rate": 9.796973684210526e-05,
      "loss": 0.5284,
      "step": 27728
    },
    {
      "epoch": 1.5527494680255347,
      "grad_norm": 1.3678479194641113,
      "learning_rate": 9.796947368421053e-05,
      "loss": 0.5655,
      "step": 27729
    },
    {
      "epoch": 1.5528054653376637,
      "grad_norm": 1.1171141862869263,
      "learning_rate": 9.796921052631579e-05,
      "loss": 0.4052,
      "step": 27730
    },
    {
      "epoch": 1.5528614626497927,
      "grad_norm": 1.676128625869751,
      "learning_rate": 9.796894736842106e-05,
      "loss": 0.4494,
      "step": 27731
    },
    {
      "epoch": 1.5529174599619218,
      "grad_norm": 1.7069107294082642,
      "learning_rate": 9.796868421052632e-05,
      "loss": 0.6305,
      "step": 27732
    },
    {
      "epoch": 1.5529734572740508,
      "grad_norm": 1.9045878648757935,
      "learning_rate": 9.796842105263158e-05,
      "loss": 0.4019,
      "step": 27733
    },
    {
      "epoch": 1.5530294545861798,
      "grad_norm": 1.4582682847976685,
      "learning_rate": 9.796815789473684e-05,
      "loss": 0.335,
      "step": 27734
    },
    {
      "epoch": 1.5530854518983088,
      "grad_norm": 1.0639840364456177,
      "learning_rate": 9.796789473684212e-05,
      "loss": 0.4934,
      "step": 27735
    },
    {
      "epoch": 1.5531414492104378,
      "grad_norm": 1.4219121932983398,
      "learning_rate": 9.796763157894738e-05,
      "loss": 0.4423,
      "step": 27736
    },
    {
      "epoch": 1.5531974465225669,
      "grad_norm": 1.201599359512329,
      "learning_rate": 9.796736842105264e-05,
      "loss": 0.409,
      "step": 27737
    },
    {
      "epoch": 1.5532534438346959,
      "grad_norm": 1.385581612586975,
      "learning_rate": 9.79671052631579e-05,
      "loss": 0.4744,
      "step": 27738
    },
    {
      "epoch": 1.553309441146825,
      "grad_norm": 1.4093060493469238,
      "learning_rate": 9.796684210526317e-05,
      "loss": 0.3263,
      "step": 27739
    },
    {
      "epoch": 1.553365438458954,
      "grad_norm": 1.5411500930786133,
      "learning_rate": 9.796657894736843e-05,
      "loss": 0.4366,
      "step": 27740
    },
    {
      "epoch": 1.553421435771083,
      "grad_norm": 1.2060127258300781,
      "learning_rate": 9.796631578947369e-05,
      "loss": 0.4044,
      "step": 27741
    },
    {
      "epoch": 1.553477433083212,
      "grad_norm": 1.3991246223449707,
      "learning_rate": 9.796605263157895e-05,
      "loss": 0.4528,
      "step": 27742
    },
    {
      "epoch": 1.553533430395341,
      "grad_norm": 1.4705504179000854,
      "learning_rate": 9.796578947368421e-05,
      "loss": 0.549,
      "step": 27743
    },
    {
      "epoch": 1.55358942770747,
      "grad_norm": 1.3193557262420654,
      "learning_rate": 9.796552631578948e-05,
      "loss": 0.3969,
      "step": 27744
    },
    {
      "epoch": 1.553645425019599,
      "grad_norm": 1.5294543504714966,
      "learning_rate": 9.796526315789474e-05,
      "loss": 0.4579,
      "step": 27745
    },
    {
      "epoch": 1.553701422331728,
      "grad_norm": 1.4128329753875732,
      "learning_rate": 9.7965e-05,
      "loss": 0.5688,
      "step": 27746
    },
    {
      "epoch": 1.553757419643857,
      "grad_norm": 1.693296194076538,
      "learning_rate": 9.796473684210526e-05,
      "loss": 0.6006,
      "step": 27747
    },
    {
      "epoch": 1.553813416955986,
      "grad_norm": 1.4760183095932007,
      "learning_rate": 9.796447368421053e-05,
      "loss": 0.4203,
      "step": 27748
    },
    {
      "epoch": 1.553869414268115,
      "grad_norm": 1.2221639156341553,
      "learning_rate": 9.79642105263158e-05,
      "loss": 0.4608,
      "step": 27749
    },
    {
      "epoch": 1.553925411580244,
      "grad_norm": 1.3965953588485718,
      "learning_rate": 9.796394736842107e-05,
      "loss": 0.5166,
      "step": 27750
    },
    {
      "epoch": 1.5539814088923731,
      "grad_norm": 1.446052074432373,
      "learning_rate": 9.796368421052631e-05,
      "loss": 0.4521,
      "step": 27751
    },
    {
      "epoch": 1.5540374062045021,
      "grad_norm": 1.1243226528167725,
      "learning_rate": 9.796342105263159e-05,
      "loss": 0.2749,
      "step": 27752
    },
    {
      "epoch": 1.5540934035166312,
      "grad_norm": 1.5954065322875977,
      "learning_rate": 9.796315789473685e-05,
      "loss": 0.5175,
      "step": 27753
    },
    {
      "epoch": 1.5541494008287602,
      "grad_norm": 1.2738813161849976,
      "learning_rate": 9.796289473684212e-05,
      "loss": 0.4379,
      "step": 27754
    },
    {
      "epoch": 1.5542053981408892,
      "grad_norm": 1.3528434038162231,
      "learning_rate": 9.796263157894737e-05,
      "loss": 0.4364,
      "step": 27755
    },
    {
      "epoch": 1.5542613954530182,
      "grad_norm": 1.5497957468032837,
      "learning_rate": 9.796236842105264e-05,
      "loss": 0.5718,
      "step": 27756
    },
    {
      "epoch": 1.5543173927651472,
      "grad_norm": 1.4113351106643677,
      "learning_rate": 9.79621052631579e-05,
      "loss": 0.5735,
      "step": 27757
    },
    {
      "epoch": 1.5543733900772763,
      "grad_norm": 1.560281753540039,
      "learning_rate": 9.796184210526316e-05,
      "loss": 0.4684,
      "step": 27758
    },
    {
      "epoch": 1.5544293873894053,
      "grad_norm": 1.4054566621780396,
      "learning_rate": 9.796157894736843e-05,
      "loss": 0.3689,
      "step": 27759
    },
    {
      "epoch": 1.5544853847015343,
      "grad_norm": 1.5809017419815063,
      "learning_rate": 9.796131578947368e-05,
      "loss": 0.4974,
      "step": 27760
    },
    {
      "epoch": 1.5545413820136633,
      "grad_norm": 1.3012982606887817,
      "learning_rate": 9.796105263157895e-05,
      "loss": 0.5023,
      "step": 27761
    },
    {
      "epoch": 1.5545973793257923,
      "grad_norm": 1.0782115459442139,
      "learning_rate": 9.796078947368421e-05,
      "loss": 0.3798,
      "step": 27762
    },
    {
      "epoch": 1.5546533766379214,
      "grad_norm": 1.3033592700958252,
      "learning_rate": 9.796052631578948e-05,
      "loss": 0.4332,
      "step": 27763
    },
    {
      "epoch": 1.5547093739500504,
      "grad_norm": 1.4053047895431519,
      "learning_rate": 9.796026315789474e-05,
      "loss": 0.4463,
      "step": 27764
    },
    {
      "epoch": 1.5547653712621794,
      "grad_norm": 1.3142244815826416,
      "learning_rate": 9.796e-05,
      "loss": 0.4889,
      "step": 27765
    },
    {
      "epoch": 1.5548213685743084,
      "grad_norm": 1.018303632736206,
      "learning_rate": 9.795973684210526e-05,
      "loss": 0.3926,
      "step": 27766
    },
    {
      "epoch": 1.5548773658864374,
      "grad_norm": 1.1563303470611572,
      "learning_rate": 9.795947368421054e-05,
      "loss": 0.3991,
      "step": 27767
    },
    {
      "epoch": 1.5549333631985665,
      "grad_norm": 1.3148008584976196,
      "learning_rate": 9.79592105263158e-05,
      "loss": 0.4882,
      "step": 27768
    },
    {
      "epoch": 1.5549893605106955,
      "grad_norm": 1.119433879852295,
      "learning_rate": 9.795894736842106e-05,
      "loss": 0.3509,
      "step": 27769
    },
    {
      "epoch": 1.5550453578228245,
      "grad_norm": 1.9987049102783203,
      "learning_rate": 9.795868421052632e-05,
      "loss": 0.4775,
      "step": 27770
    },
    {
      "epoch": 1.5551013551349535,
      "grad_norm": 1.4459271430969238,
      "learning_rate": 9.795842105263159e-05,
      "loss": 0.5327,
      "step": 27771
    },
    {
      "epoch": 1.5551573524470825,
      "grad_norm": 22.486413955688477,
      "learning_rate": 9.795815789473685e-05,
      "loss": 0.6281,
      "step": 27772
    },
    {
      "epoch": 1.5552133497592115,
      "grad_norm": 1.9849474430084229,
      "learning_rate": 9.795789473684211e-05,
      "loss": 0.6185,
      "step": 27773
    },
    {
      "epoch": 1.5552693470713406,
      "grad_norm": 1.5695642232894897,
      "learning_rate": 9.795763157894737e-05,
      "loss": 0.3902,
      "step": 27774
    },
    {
      "epoch": 1.5553253443834696,
      "grad_norm": 1.440346121788025,
      "learning_rate": 9.795736842105263e-05,
      "loss": 0.5583,
      "step": 27775
    },
    {
      "epoch": 1.5553813416955986,
      "grad_norm": 1.3403582572937012,
      "learning_rate": 9.79571052631579e-05,
      "loss": 0.5714,
      "step": 27776
    },
    {
      "epoch": 1.5554373390077276,
      "grad_norm": 1.3569977283477783,
      "learning_rate": 9.795684210526316e-05,
      "loss": 0.3281,
      "step": 27777
    },
    {
      "epoch": 1.5554933363198566,
      "grad_norm": 1.1667449474334717,
      "learning_rate": 9.795657894736842e-05,
      "loss": 0.5078,
      "step": 27778
    },
    {
      "epoch": 1.5555493336319857,
      "grad_norm": 1.2146210670471191,
      "learning_rate": 9.795631578947368e-05,
      "loss": 0.5043,
      "step": 27779
    },
    {
      "epoch": 1.5556053309441147,
      "grad_norm": 2.4014410972595215,
      "learning_rate": 9.795605263157895e-05,
      "loss": 0.445,
      "step": 27780
    },
    {
      "epoch": 1.5556613282562437,
      "grad_norm": 1.3136094808578491,
      "learning_rate": 9.795578947368421e-05,
      "loss": 0.6192,
      "step": 27781
    },
    {
      "epoch": 1.5557173255683727,
      "grad_norm": 1.2894420623779297,
      "learning_rate": 9.795552631578949e-05,
      "loss": 0.4372,
      "step": 27782
    },
    {
      "epoch": 1.5557733228805017,
      "grad_norm": 1.4806616306304932,
      "learning_rate": 9.795526315789473e-05,
      "loss": 0.6182,
      "step": 27783
    },
    {
      "epoch": 1.5558293201926308,
      "grad_norm": 1.118759274482727,
      "learning_rate": 9.7955e-05,
      "loss": 0.4354,
      "step": 27784
    },
    {
      "epoch": 1.5558853175047598,
      "grad_norm": 1.2672228813171387,
      "learning_rate": 9.795473684210527e-05,
      "loss": 0.3302,
      "step": 27785
    },
    {
      "epoch": 1.5559413148168888,
      "grad_norm": 1.3909382820129395,
      "learning_rate": 9.795447368421054e-05,
      "loss": 0.619,
      "step": 27786
    },
    {
      "epoch": 1.5559973121290178,
      "grad_norm": 1.2416855096817017,
      "learning_rate": 9.79542105263158e-05,
      "loss": 0.4477,
      "step": 27787
    },
    {
      "epoch": 1.5560533094411468,
      "grad_norm": 1.565833330154419,
      "learning_rate": 9.795394736842106e-05,
      "loss": 0.5442,
      "step": 27788
    },
    {
      "epoch": 1.5561093067532759,
      "grad_norm": 1.4561254978179932,
      "learning_rate": 9.795368421052632e-05,
      "loss": 0.5995,
      "step": 27789
    },
    {
      "epoch": 1.5561653040654049,
      "grad_norm": 1.3448600769042969,
      "learning_rate": 9.795342105263159e-05,
      "loss": 0.4515,
      "step": 27790
    },
    {
      "epoch": 1.556221301377534,
      "grad_norm": 1.1499367952346802,
      "learning_rate": 9.795315789473685e-05,
      "loss": 0.445,
      "step": 27791
    },
    {
      "epoch": 1.556277298689663,
      "grad_norm": 1.1299775838851929,
      "learning_rate": 9.79528947368421e-05,
      "loss": 0.4119,
      "step": 27792
    },
    {
      "epoch": 1.556333296001792,
      "grad_norm": 1.1896737813949585,
      "learning_rate": 9.795263157894737e-05,
      "loss": 0.4403,
      "step": 27793
    },
    {
      "epoch": 1.556389293313921,
      "grad_norm": 1.3264724016189575,
      "learning_rate": 9.795236842105263e-05,
      "loss": 0.5096,
      "step": 27794
    },
    {
      "epoch": 1.55644529062605,
      "grad_norm": 1.5012086629867554,
      "learning_rate": 9.79521052631579e-05,
      "loss": 0.4116,
      "step": 27795
    },
    {
      "epoch": 1.556501287938179,
      "grad_norm": 2.2111494541168213,
      "learning_rate": 9.795184210526316e-05,
      "loss": 0.5572,
      "step": 27796
    },
    {
      "epoch": 1.556557285250308,
      "grad_norm": 1.1106387376785278,
      "learning_rate": 9.795157894736842e-05,
      "loss": 0.3883,
      "step": 27797
    },
    {
      "epoch": 1.556613282562437,
      "grad_norm": 1.485128402709961,
      "learning_rate": 9.795131578947368e-05,
      "loss": 0.5082,
      "step": 27798
    },
    {
      "epoch": 1.556669279874566,
      "grad_norm": 1.1329487562179565,
      "learning_rate": 9.795105263157896e-05,
      "loss": 0.4078,
      "step": 27799
    },
    {
      "epoch": 1.556725277186695,
      "grad_norm": 1.0622822046279907,
      "learning_rate": 9.795078947368422e-05,
      "loss": 0.473,
      "step": 27800
    },
    {
      "epoch": 1.556781274498824,
      "grad_norm": 1.2985491752624512,
      "learning_rate": 9.795052631578948e-05,
      "loss": 0.4488,
      "step": 27801
    },
    {
      "epoch": 1.556837271810953,
      "grad_norm": 1.0579886436462402,
      "learning_rate": 9.795026315789474e-05,
      "loss": 0.3947,
      "step": 27802
    },
    {
      "epoch": 1.5568932691230821,
      "grad_norm": 1.3231867551803589,
      "learning_rate": 9.795000000000001e-05,
      "loss": 0.4697,
      "step": 27803
    },
    {
      "epoch": 1.5569492664352111,
      "grad_norm": 1.3121984004974365,
      "learning_rate": 9.794973684210527e-05,
      "loss": 0.4894,
      "step": 27804
    },
    {
      "epoch": 1.5570052637473402,
      "grad_norm": 1.2421481609344482,
      "learning_rate": 9.794947368421053e-05,
      "loss": 0.4702,
      "step": 27805
    },
    {
      "epoch": 1.5570612610594692,
      "grad_norm": 1.358551025390625,
      "learning_rate": 9.794921052631579e-05,
      "loss": 0.4754,
      "step": 27806
    },
    {
      "epoch": 1.5571172583715982,
      "grad_norm": 1.1165051460266113,
      "learning_rate": 9.794894736842106e-05,
      "loss": 0.413,
      "step": 27807
    },
    {
      "epoch": 1.5571732556837272,
      "grad_norm": 1.1838672161102295,
      "learning_rate": 9.794868421052632e-05,
      "loss": 0.4276,
      "step": 27808
    },
    {
      "epoch": 1.5572292529958562,
      "grad_norm": 1.4677737951278687,
      "learning_rate": 9.794842105263158e-05,
      "loss": 0.5689,
      "step": 27809
    },
    {
      "epoch": 1.5572852503079853,
      "grad_norm": 2.69555926322937,
      "learning_rate": 9.794815789473684e-05,
      "loss": 0.6513,
      "step": 27810
    },
    {
      "epoch": 1.5573412476201143,
      "grad_norm": 1.8028641939163208,
      "learning_rate": 9.79478947368421e-05,
      "loss": 0.4972,
      "step": 27811
    },
    {
      "epoch": 1.5573972449322433,
      "grad_norm": 1.2707033157348633,
      "learning_rate": 9.794763157894737e-05,
      "loss": 0.4286,
      "step": 27812
    },
    {
      "epoch": 1.5574532422443723,
      "grad_norm": 1.3541085720062256,
      "learning_rate": 9.794736842105263e-05,
      "loss": 0.516,
      "step": 27813
    },
    {
      "epoch": 1.5575092395565013,
      "grad_norm": 45.87267303466797,
      "learning_rate": 9.794710526315791e-05,
      "loss": 0.4842,
      "step": 27814
    },
    {
      "epoch": 1.5575652368686304,
      "grad_norm": 1.760021686553955,
      "learning_rate": 9.794684210526315e-05,
      "loss": 0.7113,
      "step": 27815
    },
    {
      "epoch": 1.5576212341807594,
      "grad_norm": 1.8014436960220337,
      "learning_rate": 9.794657894736843e-05,
      "loss": 0.8422,
      "step": 27816
    },
    {
      "epoch": 1.5576772314928884,
      "grad_norm": 1.4810210466384888,
      "learning_rate": 9.794631578947369e-05,
      "loss": 0.441,
      "step": 27817
    },
    {
      "epoch": 1.5577332288050174,
      "grad_norm": 1.642518162727356,
      "learning_rate": 9.794605263157896e-05,
      "loss": 0.386,
      "step": 27818
    },
    {
      "epoch": 1.5577892261171464,
      "grad_norm": 1.0953466892242432,
      "learning_rate": 9.794578947368422e-05,
      "loss": 0.4178,
      "step": 27819
    },
    {
      "epoch": 1.5578452234292754,
      "grad_norm": 1.2690304517745972,
      "learning_rate": 9.794552631578948e-05,
      "loss": 0.4771,
      "step": 27820
    },
    {
      "epoch": 1.5579012207414045,
      "grad_norm": 1.3168573379516602,
      "learning_rate": 9.794526315789474e-05,
      "loss": 0.4387,
      "step": 27821
    },
    {
      "epoch": 1.5579572180535335,
      "grad_norm": 1.2579644918441772,
      "learning_rate": 9.794500000000001e-05,
      "loss": 0.5125,
      "step": 27822
    },
    {
      "epoch": 1.5580132153656625,
      "grad_norm": 1.160528540611267,
      "learning_rate": 9.794473684210527e-05,
      "loss": 0.4034,
      "step": 27823
    },
    {
      "epoch": 1.5580692126777915,
      "grad_norm": 1.2254694700241089,
      "learning_rate": 9.794447368421053e-05,
      "loss": 0.4466,
      "step": 27824
    },
    {
      "epoch": 1.5581252099899205,
      "grad_norm": 1.1999738216400146,
      "learning_rate": 9.794421052631579e-05,
      "loss": 0.441,
      "step": 27825
    },
    {
      "epoch": 1.5581812073020496,
      "grad_norm": 1.1831997632980347,
      "learning_rate": 9.794394736842105e-05,
      "loss": 0.4673,
      "step": 27826
    },
    {
      "epoch": 1.5582372046141786,
      "grad_norm": 1.330078125,
      "learning_rate": 9.794368421052632e-05,
      "loss": 0.5133,
      "step": 27827
    },
    {
      "epoch": 1.5582932019263076,
      "grad_norm": 1.326755404472351,
      "learning_rate": 9.794342105263158e-05,
      "loss": 0.4091,
      "step": 27828
    },
    {
      "epoch": 1.5583491992384366,
      "grad_norm": 1.13228440284729,
      "learning_rate": 9.794315789473684e-05,
      "loss": 0.3985,
      "step": 27829
    },
    {
      "epoch": 1.5584051965505656,
      "grad_norm": 1.2264310121536255,
      "learning_rate": 9.79428947368421e-05,
      "loss": 0.4694,
      "step": 27830
    },
    {
      "epoch": 1.5584611938626947,
      "grad_norm": 1.1563001871109009,
      "learning_rate": 9.794263157894738e-05,
      "loss": 0.3887,
      "step": 27831
    },
    {
      "epoch": 1.5585171911748237,
      "grad_norm": 1.2988418340682983,
      "learning_rate": 9.794236842105264e-05,
      "loss": 0.5275,
      "step": 27832
    },
    {
      "epoch": 1.5585731884869527,
      "grad_norm": 1.272973895072937,
      "learning_rate": 9.79421052631579e-05,
      "loss": 0.4014,
      "step": 27833
    },
    {
      "epoch": 1.5586291857990817,
      "grad_norm": 1.4969706535339355,
      "learning_rate": 9.794184210526316e-05,
      "loss": 0.5557,
      "step": 27834
    },
    {
      "epoch": 1.5586851831112107,
      "grad_norm": 1.4358437061309814,
      "learning_rate": 9.794157894736843e-05,
      "loss": 0.3992,
      "step": 27835
    },
    {
      "epoch": 1.5587411804233398,
      "grad_norm": 1.112850546836853,
      "learning_rate": 9.794131578947369e-05,
      "loss": 0.4536,
      "step": 27836
    },
    {
      "epoch": 1.5587971777354688,
      "grad_norm": 1.3522948026657104,
      "learning_rate": 9.794105263157896e-05,
      "loss": 0.4322,
      "step": 27837
    },
    {
      "epoch": 1.5588531750475978,
      "grad_norm": 1.5419671535491943,
      "learning_rate": 9.794078947368421e-05,
      "loss": 0.4437,
      "step": 27838
    },
    {
      "epoch": 1.5589091723597268,
      "grad_norm": 1.2560789585113525,
      "learning_rate": 9.794052631578948e-05,
      "loss": 0.414,
      "step": 27839
    },
    {
      "epoch": 1.5589651696718558,
      "grad_norm": 1.4753572940826416,
      "learning_rate": 9.794026315789474e-05,
      "loss": 0.4246,
      "step": 27840
    },
    {
      "epoch": 1.5590211669839849,
      "grad_norm": 1.5477051734924316,
      "learning_rate": 9.794000000000001e-05,
      "loss": 0.462,
      "step": 27841
    },
    {
      "epoch": 1.5590771642961139,
      "grad_norm": 1.2439004182815552,
      "learning_rate": 9.793973684210527e-05,
      "loss": 0.3825,
      "step": 27842
    },
    {
      "epoch": 1.559133161608243,
      "grad_norm": 1.1846262216567993,
      "learning_rate": 9.793947368421053e-05,
      "loss": 0.3773,
      "step": 27843
    },
    {
      "epoch": 1.559189158920372,
      "grad_norm": 1.2155591249465942,
      "learning_rate": 9.79392105263158e-05,
      "loss": 0.3896,
      "step": 27844
    },
    {
      "epoch": 1.559245156232501,
      "grad_norm": 1.6876803636550903,
      "learning_rate": 9.793894736842105e-05,
      "loss": 0.6784,
      "step": 27845
    },
    {
      "epoch": 1.55930115354463,
      "grad_norm": 1.1184388399124146,
      "learning_rate": 9.793868421052633e-05,
      "loss": 0.4407,
      "step": 27846
    },
    {
      "epoch": 1.559357150856759,
      "grad_norm": 1.3937392234802246,
      "learning_rate": 9.793842105263157e-05,
      "loss": 0.6646,
      "step": 27847
    },
    {
      "epoch": 1.559413148168888,
      "grad_norm": 1.4409432411193848,
      "learning_rate": 9.793815789473685e-05,
      "loss": 0.6223,
      "step": 27848
    },
    {
      "epoch": 1.559469145481017,
      "grad_norm": 2.5783779621124268,
      "learning_rate": 9.79378947368421e-05,
      "loss": 0.4499,
      "step": 27849
    },
    {
      "epoch": 1.559525142793146,
      "grad_norm": 1.3124620914459229,
      "learning_rate": 9.793763157894738e-05,
      "loss": 0.4684,
      "step": 27850
    },
    {
      "epoch": 1.559581140105275,
      "grad_norm": 1.2405046224594116,
      "learning_rate": 9.793736842105264e-05,
      "loss": 0.4707,
      "step": 27851
    },
    {
      "epoch": 1.559637137417404,
      "grad_norm": 1.333479642868042,
      "learning_rate": 9.79371052631579e-05,
      "loss": 0.4788,
      "step": 27852
    },
    {
      "epoch": 1.559693134729533,
      "grad_norm": 1.2092924118041992,
      "learning_rate": 9.793684210526316e-05,
      "loss": 0.4107,
      "step": 27853
    },
    {
      "epoch": 1.559749132041662,
      "grad_norm": 1.1981545686721802,
      "learning_rate": 9.793657894736843e-05,
      "loss": 0.377,
      "step": 27854
    },
    {
      "epoch": 1.5598051293537911,
      "grad_norm": 1.3340747356414795,
      "learning_rate": 9.793631578947369e-05,
      "loss": 0.4975,
      "step": 27855
    },
    {
      "epoch": 1.5598611266659201,
      "grad_norm": 1.3088123798370361,
      "learning_rate": 9.793605263157895e-05,
      "loss": 0.5356,
      "step": 27856
    },
    {
      "epoch": 1.5599171239780492,
      "grad_norm": 1.5157161951065063,
      "learning_rate": 9.793578947368421e-05,
      "loss": 0.5309,
      "step": 27857
    },
    {
      "epoch": 1.5599731212901782,
      "grad_norm": 1.4487133026123047,
      "learning_rate": 9.793552631578948e-05,
      "loss": 0.469,
      "step": 27858
    },
    {
      "epoch": 1.5600291186023072,
      "grad_norm": 1.2588891983032227,
      "learning_rate": 9.793526315789474e-05,
      "loss": 0.399,
      "step": 27859
    },
    {
      "epoch": 1.5600851159144362,
      "grad_norm": 1.49861478805542,
      "learning_rate": 9.7935e-05,
      "loss": 0.4292,
      "step": 27860
    },
    {
      "epoch": 1.5601411132265652,
      "grad_norm": 1.1538630723953247,
      "learning_rate": 9.793473684210526e-05,
      "loss": 0.4191,
      "step": 27861
    },
    {
      "epoch": 1.5601971105386943,
      "grad_norm": 1.2881295680999756,
      "learning_rate": 9.793447368421052e-05,
      "loss": 0.413,
      "step": 27862
    },
    {
      "epoch": 1.5602531078508233,
      "grad_norm": 1.2034615278244019,
      "learning_rate": 9.79342105263158e-05,
      "loss": 0.4059,
      "step": 27863
    },
    {
      "epoch": 1.5603091051629523,
      "grad_norm": 1.1906850337982178,
      "learning_rate": 9.793394736842106e-05,
      "loss": 0.4292,
      "step": 27864
    },
    {
      "epoch": 1.5603651024750813,
      "grad_norm": 1.0385959148406982,
      "learning_rate": 9.793368421052632e-05,
      "loss": 0.4155,
      "step": 27865
    },
    {
      "epoch": 1.5604210997872103,
      "grad_norm": 2.187077283859253,
      "learning_rate": 9.793342105263158e-05,
      "loss": 0.5637,
      "step": 27866
    },
    {
      "epoch": 1.5604770970993393,
      "grad_norm": 1.3084129095077515,
      "learning_rate": 9.793315789473685e-05,
      "loss": 0.4163,
      "step": 27867
    },
    {
      "epoch": 1.5605330944114684,
      "grad_norm": 1.4236375093460083,
      "learning_rate": 9.793289473684211e-05,
      "loss": 0.6847,
      "step": 27868
    },
    {
      "epoch": 1.5605890917235974,
      "grad_norm": 1.2820924520492554,
      "learning_rate": 9.793263157894738e-05,
      "loss": 0.4831,
      "step": 27869
    },
    {
      "epoch": 1.5606450890357264,
      "grad_norm": 1.1205179691314697,
      "learning_rate": 9.793236842105263e-05,
      "loss": 0.4814,
      "step": 27870
    },
    {
      "epoch": 1.5607010863478554,
      "grad_norm": 1.2113511562347412,
      "learning_rate": 9.79321052631579e-05,
      "loss": 0.342,
      "step": 27871
    },
    {
      "epoch": 1.5607570836599844,
      "grad_norm": 1.2579805850982666,
      "learning_rate": 9.793184210526316e-05,
      "loss": 0.3823,
      "step": 27872
    },
    {
      "epoch": 1.5608130809721135,
      "grad_norm": 1.8428640365600586,
      "learning_rate": 9.793157894736843e-05,
      "loss": 0.4281,
      "step": 27873
    },
    {
      "epoch": 1.5608690782842425,
      "grad_norm": 1.8847538232803345,
      "learning_rate": 9.79313157894737e-05,
      "loss": 0.3794,
      "step": 27874
    },
    {
      "epoch": 1.5609250755963715,
      "grad_norm": 1.2826124429702759,
      "learning_rate": 9.793105263157895e-05,
      "loss": 0.4482,
      "step": 27875
    },
    {
      "epoch": 1.5609810729085005,
      "grad_norm": 1.134809136390686,
      "learning_rate": 9.793078947368421e-05,
      "loss": 0.4436,
      "step": 27876
    },
    {
      "epoch": 1.5610370702206295,
      "grad_norm": 1.2939330339431763,
      "learning_rate": 9.793052631578949e-05,
      "loss": 0.4125,
      "step": 27877
    },
    {
      "epoch": 1.5610930675327586,
      "grad_norm": 1.9151451587677002,
      "learning_rate": 9.793026315789475e-05,
      "loss": 0.5838,
      "step": 27878
    },
    {
      "epoch": 1.5611490648448876,
      "grad_norm": 1.1742690801620483,
      "learning_rate": 9.793e-05,
      "loss": 0.3943,
      "step": 27879
    },
    {
      "epoch": 1.5612050621570166,
      "grad_norm": 1.6123944520950317,
      "learning_rate": 9.792973684210527e-05,
      "loss": 0.5769,
      "step": 27880
    },
    {
      "epoch": 1.5612610594691456,
      "grad_norm": 1.2266395092010498,
      "learning_rate": 9.792947368421053e-05,
      "loss": 0.3835,
      "step": 27881
    },
    {
      "epoch": 1.5613170567812746,
      "grad_norm": 1.2051206827163696,
      "learning_rate": 9.79292105263158e-05,
      "loss": 0.4799,
      "step": 27882
    },
    {
      "epoch": 1.5613730540934037,
      "grad_norm": 1.504221796989441,
      "learning_rate": 9.792894736842106e-05,
      "loss": 0.6255,
      "step": 27883
    },
    {
      "epoch": 1.5614290514055327,
      "grad_norm": 1.3478666543960571,
      "learning_rate": 9.792868421052632e-05,
      "loss": 0.4994,
      "step": 27884
    },
    {
      "epoch": 1.5614850487176617,
      "grad_norm": 1.4160724878311157,
      "learning_rate": 9.792842105263158e-05,
      "loss": 0.5038,
      "step": 27885
    },
    {
      "epoch": 1.5615410460297907,
      "grad_norm": 1.9105815887451172,
      "learning_rate": 9.792815789473685e-05,
      "loss": 0.5542,
      "step": 27886
    },
    {
      "epoch": 1.5615970433419197,
      "grad_norm": 1.240626335144043,
      "learning_rate": 9.792789473684211e-05,
      "loss": 0.3307,
      "step": 27887
    },
    {
      "epoch": 1.5616530406540488,
      "grad_norm": 1.2806501388549805,
      "learning_rate": 9.792763157894737e-05,
      "loss": 0.35,
      "step": 27888
    },
    {
      "epoch": 1.5617090379661778,
      "grad_norm": 1.2732311487197876,
      "learning_rate": 9.792736842105263e-05,
      "loss": 0.4266,
      "step": 27889
    },
    {
      "epoch": 1.5617650352783068,
      "grad_norm": 1.423905372619629,
      "learning_rate": 9.79271052631579e-05,
      "loss": 0.4768,
      "step": 27890
    },
    {
      "epoch": 1.5618210325904358,
      "grad_norm": 1.2721583843231201,
      "learning_rate": 9.792684210526316e-05,
      "loss": 0.4447,
      "step": 27891
    },
    {
      "epoch": 1.5618770299025648,
      "grad_norm": 1.4054096937179565,
      "learning_rate": 9.792657894736844e-05,
      "loss": 0.5952,
      "step": 27892
    },
    {
      "epoch": 1.5619330272146938,
      "grad_norm": 1.4433014392852783,
      "learning_rate": 9.792631578947368e-05,
      "loss": 0.4713,
      "step": 27893
    },
    {
      "epoch": 1.5619890245268229,
      "grad_norm": 1.7865170240402222,
      "learning_rate": 9.792605263157896e-05,
      "loss": 0.4889,
      "step": 27894
    },
    {
      "epoch": 1.5620450218389519,
      "grad_norm": 1.6442961692810059,
      "learning_rate": 9.792578947368422e-05,
      "loss": 0.3941,
      "step": 27895
    },
    {
      "epoch": 1.562101019151081,
      "grad_norm": 1.4772812128067017,
      "learning_rate": 9.792552631578948e-05,
      "loss": 0.4298,
      "step": 27896
    },
    {
      "epoch": 1.56215701646321,
      "grad_norm": 1.2540076971054077,
      "learning_rate": 9.792526315789475e-05,
      "loss": 0.381,
      "step": 27897
    },
    {
      "epoch": 1.562213013775339,
      "grad_norm": 1.2921191453933716,
      "learning_rate": 9.7925e-05,
      "loss": 0.4336,
      "step": 27898
    },
    {
      "epoch": 1.562269011087468,
      "grad_norm": 1.384541630744934,
      "learning_rate": 9.792473684210527e-05,
      "loss": 0.5008,
      "step": 27899
    },
    {
      "epoch": 1.562325008399597,
      "grad_norm": 1.2399932146072388,
      "learning_rate": 9.792447368421053e-05,
      "loss": 0.3815,
      "step": 27900
    },
    {
      "epoch": 1.562381005711726,
      "grad_norm": 1.7059587240219116,
      "learning_rate": 9.79242105263158e-05,
      "loss": 0.4672,
      "step": 27901
    },
    {
      "epoch": 1.562437003023855,
      "grad_norm": 1.260528564453125,
      "learning_rate": 9.792394736842105e-05,
      "loss": 0.4081,
      "step": 27902
    },
    {
      "epoch": 1.562493000335984,
      "grad_norm": 1.2083096504211426,
      "learning_rate": 9.792368421052632e-05,
      "loss": 0.4603,
      "step": 27903
    },
    {
      "epoch": 1.5625489976481128,
      "grad_norm": 1.1734106540679932,
      "learning_rate": 9.792342105263158e-05,
      "loss": 0.4077,
      "step": 27904
    },
    {
      "epoch": 1.5626049949602419,
      "grad_norm": 1.3308879137039185,
      "learning_rate": 9.792315789473685e-05,
      "loss": 0.3871,
      "step": 27905
    },
    {
      "epoch": 1.5626609922723709,
      "grad_norm": 1.4251302480697632,
      "learning_rate": 9.792289473684211e-05,
      "loss": 0.4728,
      "step": 27906
    },
    {
      "epoch": 1.5627169895845,
      "grad_norm": 1.292154312133789,
      "learning_rate": 9.792263157894737e-05,
      "loss": 0.4261,
      "step": 27907
    },
    {
      "epoch": 1.562772986896629,
      "grad_norm": 1.6118751764297485,
      "learning_rate": 9.792236842105263e-05,
      "loss": 0.462,
      "step": 27908
    },
    {
      "epoch": 1.562828984208758,
      "grad_norm": 1.6858899593353271,
      "learning_rate": 9.79221052631579e-05,
      "loss": 0.5896,
      "step": 27909
    },
    {
      "epoch": 1.562884981520887,
      "grad_norm": 1.938799262046814,
      "learning_rate": 9.792184210526317e-05,
      "loss": 0.5382,
      "step": 27910
    },
    {
      "epoch": 1.562940978833016,
      "grad_norm": 1.291366457939148,
      "learning_rate": 9.792157894736843e-05,
      "loss": 0.4255,
      "step": 27911
    },
    {
      "epoch": 1.562996976145145,
      "grad_norm": 1.2998098134994507,
      "learning_rate": 9.792131578947369e-05,
      "loss": 0.471,
      "step": 27912
    },
    {
      "epoch": 1.563052973457274,
      "grad_norm": 1.3475925922393799,
      "learning_rate": 9.792105263157895e-05,
      "loss": 0.4834,
      "step": 27913
    },
    {
      "epoch": 1.563108970769403,
      "grad_norm": 1.3017666339874268,
      "learning_rate": 9.792078947368422e-05,
      "loss": 0.5708,
      "step": 27914
    },
    {
      "epoch": 1.563164968081532,
      "grad_norm": 1.2234461307525635,
      "learning_rate": 9.792052631578948e-05,
      "loss": 0.5419,
      "step": 27915
    },
    {
      "epoch": 1.563220965393661,
      "grad_norm": 1.4304434061050415,
      "learning_rate": 9.792026315789474e-05,
      "loss": 0.4753,
      "step": 27916
    },
    {
      "epoch": 1.56327696270579,
      "grad_norm": 1.1569468975067139,
      "learning_rate": 9.792e-05,
      "loss": 0.3657,
      "step": 27917
    },
    {
      "epoch": 1.563332960017919,
      "grad_norm": 1.090216040611267,
      "learning_rate": 9.791973684210527e-05,
      "loss": 0.3583,
      "step": 27918
    },
    {
      "epoch": 1.5633889573300481,
      "grad_norm": 1.2005150318145752,
      "learning_rate": 9.791947368421053e-05,
      "loss": 0.3683,
      "step": 27919
    },
    {
      "epoch": 1.5634449546421771,
      "grad_norm": 1.5770893096923828,
      "learning_rate": 9.791921052631579e-05,
      "loss": 0.5088,
      "step": 27920
    },
    {
      "epoch": 1.5635009519543062,
      "grad_norm": 1.634976863861084,
      "learning_rate": 9.791894736842105e-05,
      "loss": 0.5555,
      "step": 27921
    },
    {
      "epoch": 1.5635569492664352,
      "grad_norm": 1.1582673788070679,
      "learning_rate": 9.791868421052632e-05,
      "loss": 0.4653,
      "step": 27922
    },
    {
      "epoch": 1.5636129465785642,
      "grad_norm": 1.4435118436813354,
      "learning_rate": 9.791842105263158e-05,
      "loss": 0.4989,
      "step": 27923
    },
    {
      "epoch": 1.5636689438906932,
      "grad_norm": 1.4571359157562256,
      "learning_rate": 9.791815789473686e-05,
      "loss": 0.4679,
      "step": 27924
    },
    {
      "epoch": 1.5637249412028222,
      "grad_norm": 1.1578705310821533,
      "learning_rate": 9.79178947368421e-05,
      "loss": 0.4218,
      "step": 27925
    },
    {
      "epoch": 1.5637809385149513,
      "grad_norm": 1.4565349817276,
      "learning_rate": 9.791763157894738e-05,
      "loss": 0.4999,
      "step": 27926
    },
    {
      "epoch": 1.5638369358270803,
      "grad_norm": 1.5129117965698242,
      "learning_rate": 9.791736842105264e-05,
      "loss": 0.4592,
      "step": 27927
    },
    {
      "epoch": 1.5638929331392093,
      "grad_norm": 1.1397032737731934,
      "learning_rate": 9.791710526315791e-05,
      "loss": 0.447,
      "step": 27928
    },
    {
      "epoch": 1.5639489304513383,
      "grad_norm": 1.3803645372390747,
      "learning_rate": 9.791684210526317e-05,
      "loss": 0.673,
      "step": 27929
    },
    {
      "epoch": 1.5640049277634673,
      "grad_norm": 1.7738953828811646,
      "learning_rate": 9.791657894736841e-05,
      "loss": 0.5228,
      "step": 27930
    },
    {
      "epoch": 1.5640609250755964,
      "grad_norm": 1.5509098768234253,
      "learning_rate": 9.791631578947369e-05,
      "loss": 0.4285,
      "step": 27931
    },
    {
      "epoch": 1.5641169223877254,
      "grad_norm": 1.4631469249725342,
      "learning_rate": 9.791605263157895e-05,
      "loss": 0.5835,
      "step": 27932
    },
    {
      "epoch": 1.5641729196998544,
      "grad_norm": 1.380316972732544,
      "learning_rate": 9.791578947368422e-05,
      "loss": 0.4579,
      "step": 27933
    },
    {
      "epoch": 1.5642289170119834,
      "grad_norm": 1.1702483892440796,
      "learning_rate": 9.791552631578948e-05,
      "loss": 0.5154,
      "step": 27934
    },
    {
      "epoch": 1.5642849143241124,
      "grad_norm": 1.2805495262145996,
      "learning_rate": 9.791526315789474e-05,
      "loss": 0.446,
      "step": 27935
    },
    {
      "epoch": 1.5643409116362414,
      "grad_norm": 1.351359248161316,
      "learning_rate": 9.7915e-05,
      "loss": 0.5345,
      "step": 27936
    },
    {
      "epoch": 1.5643969089483705,
      "grad_norm": 1.1455228328704834,
      "learning_rate": 9.791473684210527e-05,
      "loss": 0.3928,
      "step": 27937
    },
    {
      "epoch": 1.5644529062604995,
      "grad_norm": 1.5278799533843994,
      "learning_rate": 9.791447368421053e-05,
      "loss": 0.5924,
      "step": 27938
    },
    {
      "epoch": 1.5645089035726285,
      "grad_norm": 1.5728360414505005,
      "learning_rate": 9.791421052631579e-05,
      "loss": 0.4548,
      "step": 27939
    },
    {
      "epoch": 1.5645649008847575,
      "grad_norm": 1.5662719011306763,
      "learning_rate": 9.791394736842105e-05,
      "loss": 0.4648,
      "step": 27940
    },
    {
      "epoch": 1.5646208981968865,
      "grad_norm": 1.1243435144424438,
      "learning_rate": 9.791368421052633e-05,
      "loss": 0.339,
      "step": 27941
    },
    {
      "epoch": 1.5646768955090156,
      "grad_norm": 1.3633006811141968,
      "learning_rate": 9.791342105263159e-05,
      "loss": 0.4912,
      "step": 27942
    },
    {
      "epoch": 1.5647328928211446,
      "grad_norm": 1.3593796491622925,
      "learning_rate": 9.791315789473685e-05,
      "loss": 0.5445,
      "step": 27943
    },
    {
      "epoch": 1.5647888901332736,
      "grad_norm": 1.676698923110962,
      "learning_rate": 9.79128947368421e-05,
      "loss": 0.4787,
      "step": 27944
    },
    {
      "epoch": 1.5648448874454026,
      "grad_norm": 1.312387228012085,
      "learning_rate": 9.791263157894738e-05,
      "loss": 0.5799,
      "step": 27945
    },
    {
      "epoch": 1.5649008847575316,
      "grad_norm": 1.405980110168457,
      "learning_rate": 9.791236842105264e-05,
      "loss": 0.6163,
      "step": 27946
    },
    {
      "epoch": 1.5649568820696607,
      "grad_norm": 1.2081812620162964,
      "learning_rate": 9.79121052631579e-05,
      "loss": 0.4004,
      "step": 27947
    },
    {
      "epoch": 1.5650128793817897,
      "grad_norm": 1.5452519655227661,
      "learning_rate": 9.791184210526316e-05,
      "loss": 0.3953,
      "step": 27948
    },
    {
      "epoch": 1.5650688766939187,
      "grad_norm": 1.3677908182144165,
      "learning_rate": 9.791157894736842e-05,
      "loss": 0.474,
      "step": 27949
    },
    {
      "epoch": 1.5651248740060477,
      "grad_norm": 1.3281514644622803,
      "learning_rate": 9.791131578947369e-05,
      "loss": 0.3966,
      "step": 27950
    },
    {
      "epoch": 1.5651808713181767,
      "grad_norm": 1.2376335859298706,
      "learning_rate": 9.791105263157895e-05,
      "loss": 0.4162,
      "step": 27951
    },
    {
      "epoch": 1.5652368686303058,
      "grad_norm": 1.2949895858764648,
      "learning_rate": 9.791078947368422e-05,
      "loss": 0.4646,
      "step": 27952
    },
    {
      "epoch": 1.5652928659424348,
      "grad_norm": 1.4561299085617065,
      "learning_rate": 9.791052631578947e-05,
      "loss": 0.4796,
      "step": 27953
    },
    {
      "epoch": 1.5653488632545638,
      "grad_norm": 1.5170055627822876,
      "learning_rate": 9.791026315789474e-05,
      "loss": 0.5016,
      "step": 27954
    },
    {
      "epoch": 1.5654048605666928,
      "grad_norm": 2.710172176361084,
      "learning_rate": 9.791e-05,
      "loss": 0.3683,
      "step": 27955
    },
    {
      "epoch": 1.5654608578788218,
      "grad_norm": 1.7364965677261353,
      "learning_rate": 9.790973684210528e-05,
      "loss": 0.541,
      "step": 27956
    },
    {
      "epoch": 1.5655168551909509,
      "grad_norm": 1.2130730152130127,
      "learning_rate": 9.790947368421052e-05,
      "loss": 0.4467,
      "step": 27957
    },
    {
      "epoch": 1.5655728525030799,
      "grad_norm": 1.2386354207992554,
      "learning_rate": 9.79092105263158e-05,
      "loss": 0.4314,
      "step": 27958
    },
    {
      "epoch": 1.565628849815209,
      "grad_norm": 1.3954352140426636,
      "learning_rate": 9.790894736842106e-05,
      "loss": 0.4406,
      "step": 27959
    },
    {
      "epoch": 1.565684847127338,
      "grad_norm": 1.4999171495437622,
      "learning_rate": 9.790868421052633e-05,
      "loss": 0.505,
      "step": 27960
    },
    {
      "epoch": 1.565740844439467,
      "grad_norm": 1.12770676612854,
      "learning_rate": 9.790842105263159e-05,
      "loss": 0.4304,
      "step": 27961
    },
    {
      "epoch": 1.565796841751596,
      "grad_norm": 1.4203526973724365,
      "learning_rate": 9.790815789473685e-05,
      "loss": 0.522,
      "step": 27962
    },
    {
      "epoch": 1.565852839063725,
      "grad_norm": 1.2756706476211548,
      "learning_rate": 9.790789473684211e-05,
      "loss": 0.4641,
      "step": 27963
    },
    {
      "epoch": 1.565908836375854,
      "grad_norm": 1.3431830406188965,
      "learning_rate": 9.790763157894738e-05,
      "loss": 0.375,
      "step": 27964
    },
    {
      "epoch": 1.565964833687983,
      "grad_norm": 1.577451467514038,
      "learning_rate": 9.790736842105264e-05,
      "loss": 0.439,
      "step": 27965
    },
    {
      "epoch": 1.566020831000112,
      "grad_norm": 1.377826452255249,
      "learning_rate": 9.79071052631579e-05,
      "loss": 0.5604,
      "step": 27966
    },
    {
      "epoch": 1.566076828312241,
      "grad_norm": 1.374342679977417,
      "learning_rate": 9.790684210526316e-05,
      "loss": 0.4538,
      "step": 27967
    },
    {
      "epoch": 1.56613282562437,
      "grad_norm": 1.2720531225204468,
      "learning_rate": 9.790657894736842e-05,
      "loss": 0.3871,
      "step": 27968
    },
    {
      "epoch": 1.566188822936499,
      "grad_norm": 1.3962714672088623,
      "learning_rate": 9.790631578947369e-05,
      "loss": 0.5448,
      "step": 27969
    },
    {
      "epoch": 1.566244820248628,
      "grad_norm": 1.2228339910507202,
      "learning_rate": 9.790605263157895e-05,
      "loss": 0.3319,
      "step": 27970
    },
    {
      "epoch": 1.5663008175607571,
      "grad_norm": 1.1670565605163574,
      "learning_rate": 9.790578947368421e-05,
      "loss": 0.5092,
      "step": 27971
    },
    {
      "epoch": 1.5663568148728861,
      "grad_norm": 1.0824604034423828,
      "learning_rate": 9.790552631578947e-05,
      "loss": 0.3979,
      "step": 27972
    },
    {
      "epoch": 1.5664128121850152,
      "grad_norm": 1.3282324075698853,
      "learning_rate": 9.790526315789475e-05,
      "loss": 0.4293,
      "step": 27973
    },
    {
      "epoch": 1.5664688094971442,
      "grad_norm": 1.1624890565872192,
      "learning_rate": 9.7905e-05,
      "loss": 0.4407,
      "step": 27974
    },
    {
      "epoch": 1.5665248068092732,
      "grad_norm": 1.3137606382369995,
      "learning_rate": 9.790473684210527e-05,
      "loss": 0.4876,
      "step": 27975
    },
    {
      "epoch": 1.5665808041214022,
      "grad_norm": 1.2371691465377808,
      "learning_rate": 9.790447368421052e-05,
      "loss": 0.4415,
      "step": 27976
    },
    {
      "epoch": 1.5666368014335312,
      "grad_norm": 1.12775719165802,
      "learning_rate": 9.79042105263158e-05,
      "loss": 0.3587,
      "step": 27977
    },
    {
      "epoch": 1.5666927987456603,
      "grad_norm": 1.0950857400894165,
      "learning_rate": 9.790394736842106e-05,
      "loss": 0.3711,
      "step": 27978
    },
    {
      "epoch": 1.5667487960577893,
      "grad_norm": 1.4122695922851562,
      "learning_rate": 9.790368421052633e-05,
      "loss": 0.6856,
      "step": 27979
    },
    {
      "epoch": 1.5668047933699183,
      "grad_norm": 1.40877103805542,
      "learning_rate": 9.790342105263158e-05,
      "loss": 0.6257,
      "step": 27980
    },
    {
      "epoch": 1.5668607906820473,
      "grad_norm": 1.2427443265914917,
      "learning_rate": 9.790315789473685e-05,
      "loss": 0.3701,
      "step": 27981
    },
    {
      "epoch": 1.5669167879941763,
      "grad_norm": 1.1560802459716797,
      "learning_rate": 9.790289473684211e-05,
      "loss": 0.435,
      "step": 27982
    },
    {
      "epoch": 1.5669727853063053,
      "grad_norm": 1.0992628335952759,
      "learning_rate": 9.790263157894737e-05,
      "loss": 0.2943,
      "step": 27983
    },
    {
      "epoch": 1.5670287826184344,
      "grad_norm": 1.3091014623641968,
      "learning_rate": 9.790236842105264e-05,
      "loss": 0.4956,
      "step": 27984
    },
    {
      "epoch": 1.5670847799305634,
      "grad_norm": 1.0343049764633179,
      "learning_rate": 9.790210526315789e-05,
      "loss": 0.3894,
      "step": 27985
    },
    {
      "epoch": 1.5671407772426922,
      "grad_norm": 1.20504891872406,
      "learning_rate": 9.790184210526316e-05,
      "loss": 0.3722,
      "step": 27986
    },
    {
      "epoch": 1.5671967745548212,
      "grad_norm": 1.3490113019943237,
      "learning_rate": 9.790157894736842e-05,
      "loss": 0.4699,
      "step": 27987
    },
    {
      "epoch": 1.5672527718669502,
      "grad_norm": 1.3986599445343018,
      "learning_rate": 9.79013157894737e-05,
      "loss": 0.4752,
      "step": 27988
    },
    {
      "epoch": 1.5673087691790792,
      "grad_norm": 4.4811224937438965,
      "learning_rate": 9.790105263157896e-05,
      "loss": 0.4955,
      "step": 27989
    },
    {
      "epoch": 1.5673647664912083,
      "grad_norm": 1.4624607563018799,
      "learning_rate": 9.790078947368422e-05,
      "loss": 0.5015,
      "step": 27990
    },
    {
      "epoch": 1.5674207638033373,
      "grad_norm": 1.1344993114471436,
      "learning_rate": 9.790052631578947e-05,
      "loss": 0.4068,
      "step": 27991
    },
    {
      "epoch": 1.5674767611154663,
      "grad_norm": 1.3949861526489258,
      "learning_rate": 9.790026315789475e-05,
      "loss": 0.4201,
      "step": 27992
    },
    {
      "epoch": 1.5675327584275953,
      "grad_norm": 1.5104824304580688,
      "learning_rate": 9.790000000000001e-05,
      "loss": 0.3837,
      "step": 27993
    },
    {
      "epoch": 1.5675887557397243,
      "grad_norm": 1.2982854843139648,
      "learning_rate": 9.789973684210527e-05,
      "loss": 0.4254,
      "step": 27994
    },
    {
      "epoch": 1.5676447530518534,
      "grad_norm": 1.2567329406738281,
      "learning_rate": 9.789947368421053e-05,
      "loss": 0.4193,
      "step": 27995
    },
    {
      "epoch": 1.5677007503639824,
      "grad_norm": 1.7564815282821655,
      "learning_rate": 9.78992105263158e-05,
      "loss": 0.6307,
      "step": 27996
    },
    {
      "epoch": 1.5677567476761114,
      "grad_norm": 1.3201968669891357,
      "learning_rate": 9.789894736842106e-05,
      "loss": 0.412,
      "step": 27997
    },
    {
      "epoch": 1.5678127449882404,
      "grad_norm": 1.2468838691711426,
      "learning_rate": 9.789868421052632e-05,
      "loss": 0.3827,
      "step": 27998
    },
    {
      "epoch": 1.5678687423003694,
      "grad_norm": 1.3903758525848389,
      "learning_rate": 9.789842105263158e-05,
      "loss": 0.3819,
      "step": 27999
    },
    {
      "epoch": 1.5679247396124985,
      "grad_norm": 1.143258810043335,
      "learning_rate": 9.789815789473684e-05,
      "loss": 0.2986,
      "step": 28000
    },
    {
      "epoch": 1.5679807369246275,
      "grad_norm": 1.330857515335083,
      "learning_rate": 9.789789473684211e-05,
      "loss": 0.4789,
      "step": 28001
    },
    {
      "epoch": 1.5680367342367565,
      "grad_norm": 1.2843005657196045,
      "learning_rate": 9.789763157894737e-05,
      "loss": 0.4213,
      "step": 28002
    },
    {
      "epoch": 1.5680927315488855,
      "grad_norm": 1.3903347253799438,
      "learning_rate": 9.789736842105263e-05,
      "loss": 0.5385,
      "step": 28003
    },
    {
      "epoch": 1.5681487288610145,
      "grad_norm": 1.4101808071136475,
      "learning_rate": 9.789710526315789e-05,
      "loss": 0.4919,
      "step": 28004
    },
    {
      "epoch": 1.5682047261731435,
      "grad_norm": 1.4195345640182495,
      "learning_rate": 9.789684210526317e-05,
      "loss": 0.3698,
      "step": 28005
    },
    {
      "epoch": 1.5682607234852726,
      "grad_norm": 1.22687828540802,
      "learning_rate": 9.789657894736843e-05,
      "loss": 0.3312,
      "step": 28006
    },
    {
      "epoch": 1.5683167207974016,
      "grad_norm": 1.58999502658844,
      "learning_rate": 9.789631578947368e-05,
      "loss": 0.521,
      "step": 28007
    },
    {
      "epoch": 1.5683727181095306,
      "grad_norm": 1.1324678659439087,
      "learning_rate": 9.789605263157894e-05,
      "loss": 0.421,
      "step": 28008
    },
    {
      "epoch": 1.5684287154216596,
      "grad_norm": 1.395676612854004,
      "learning_rate": 9.789578947368422e-05,
      "loss": 0.5193,
      "step": 28009
    },
    {
      "epoch": 1.5684847127337886,
      "grad_norm": 1.327857255935669,
      "learning_rate": 9.789552631578948e-05,
      "loss": 0.5392,
      "step": 28010
    },
    {
      "epoch": 1.5685407100459177,
      "grad_norm": 1.1735122203826904,
      "learning_rate": 9.789526315789475e-05,
      "loss": 0.4637,
      "step": 28011
    },
    {
      "epoch": 1.5685967073580467,
      "grad_norm": 1.3529436588287354,
      "learning_rate": 9.7895e-05,
      "loss": 0.4668,
      "step": 28012
    },
    {
      "epoch": 1.5686527046701757,
      "grad_norm": 1.3784637451171875,
      "learning_rate": 9.789473684210527e-05,
      "loss": 0.4643,
      "step": 28013
    },
    {
      "epoch": 1.5687087019823047,
      "grad_norm": 1.3789427280426025,
      "learning_rate": 9.789447368421053e-05,
      "loss": 0.4272,
      "step": 28014
    },
    {
      "epoch": 1.5687646992944337,
      "grad_norm": 1.415741205215454,
      "learning_rate": 9.78942105263158e-05,
      "loss": 0.4292,
      "step": 28015
    },
    {
      "epoch": 1.5688206966065628,
      "grad_norm": 1.377379298210144,
      "learning_rate": 9.789394736842106e-05,
      "loss": 0.4687,
      "step": 28016
    },
    {
      "epoch": 1.5688766939186918,
      "grad_norm": 1.4200210571289062,
      "learning_rate": 9.789368421052631e-05,
      "loss": 0.4575,
      "step": 28017
    },
    {
      "epoch": 1.5689326912308208,
      "grad_norm": 1.5323363542556763,
      "learning_rate": 9.789342105263158e-05,
      "loss": 0.4414,
      "step": 28018
    },
    {
      "epoch": 1.5689886885429498,
      "grad_norm": 1.495772123336792,
      "learning_rate": 9.789315789473684e-05,
      "loss": 0.4087,
      "step": 28019
    },
    {
      "epoch": 1.5690446858550788,
      "grad_norm": 1.220541000366211,
      "learning_rate": 9.789289473684212e-05,
      "loss": 0.4097,
      "step": 28020
    },
    {
      "epoch": 1.5691006831672079,
      "grad_norm": 1.581607460975647,
      "learning_rate": 9.789263157894738e-05,
      "loss": 0.418,
      "step": 28021
    },
    {
      "epoch": 1.5691566804793369,
      "grad_norm": 1.676621437072754,
      "learning_rate": 9.789236842105263e-05,
      "loss": 0.4839,
      "step": 28022
    },
    {
      "epoch": 1.569212677791466,
      "grad_norm": 1.236293911933899,
      "learning_rate": 9.78921052631579e-05,
      "loss": 0.3658,
      "step": 28023
    },
    {
      "epoch": 1.569268675103595,
      "grad_norm": 1.4252465963363647,
      "learning_rate": 9.789184210526317e-05,
      "loss": 0.5402,
      "step": 28024
    },
    {
      "epoch": 1.569324672415724,
      "grad_norm": 1.4142038822174072,
      "learning_rate": 9.789157894736843e-05,
      "loss": 0.3963,
      "step": 28025
    },
    {
      "epoch": 1.569380669727853,
      "grad_norm": 1.4917858839035034,
      "learning_rate": 9.789131578947369e-05,
      "loss": 0.4539,
      "step": 28026
    },
    {
      "epoch": 1.569436667039982,
      "grad_norm": 1.422104835510254,
      "learning_rate": 9.789105263157895e-05,
      "loss": 0.5467,
      "step": 28027
    },
    {
      "epoch": 1.569492664352111,
      "grad_norm": 1.6085867881774902,
      "learning_rate": 9.789078947368422e-05,
      "loss": 0.6167,
      "step": 28028
    },
    {
      "epoch": 1.56954866166424,
      "grad_norm": 1.2367597818374634,
      "learning_rate": 9.789052631578948e-05,
      "loss": 0.4233,
      "step": 28029
    },
    {
      "epoch": 1.569604658976369,
      "grad_norm": 1.2459625005722046,
      "learning_rate": 9.789026315789474e-05,
      "loss": 0.4643,
      "step": 28030
    },
    {
      "epoch": 1.569660656288498,
      "grad_norm": 1.1102368831634521,
      "learning_rate": 9.789e-05,
      "loss": 0.4502,
      "step": 28031
    },
    {
      "epoch": 1.569716653600627,
      "grad_norm": 1.150658130645752,
      "learning_rate": 9.788973684210527e-05,
      "loss": 0.3848,
      "step": 28032
    },
    {
      "epoch": 1.569772650912756,
      "grad_norm": 1.266994595527649,
      "learning_rate": 9.788947368421053e-05,
      "loss": 0.3786,
      "step": 28033
    },
    {
      "epoch": 1.569828648224885,
      "grad_norm": 1.4353046417236328,
      "learning_rate": 9.788921052631579e-05,
      "loss": 0.4771,
      "step": 28034
    },
    {
      "epoch": 1.5698846455370141,
      "grad_norm": 1.238640308380127,
      "learning_rate": 9.788894736842105e-05,
      "loss": 0.519,
      "step": 28035
    },
    {
      "epoch": 1.5699406428491431,
      "grad_norm": 1.2137643098831177,
      "learning_rate": 9.788868421052631e-05,
      "loss": 0.4772,
      "step": 28036
    },
    {
      "epoch": 1.5699966401612722,
      "grad_norm": 1.268845558166504,
      "learning_rate": 9.788842105263159e-05,
      "loss": 0.424,
      "step": 28037
    },
    {
      "epoch": 1.5700526374734012,
      "grad_norm": 1.2281060218811035,
      "learning_rate": 9.788815789473684e-05,
      "loss": 0.4479,
      "step": 28038
    },
    {
      "epoch": 1.5701086347855302,
      "grad_norm": 4.048982620239258,
      "learning_rate": 9.788789473684212e-05,
      "loss": 0.4112,
      "step": 28039
    },
    {
      "epoch": 1.5701646320976592,
      "grad_norm": 1.7420209646224976,
      "learning_rate": 9.788763157894736e-05,
      "loss": 0.7577,
      "step": 28040
    },
    {
      "epoch": 1.5702206294097882,
      "grad_norm": 1.2172189950942993,
      "learning_rate": 9.788736842105264e-05,
      "loss": 0.3898,
      "step": 28041
    },
    {
      "epoch": 1.5702766267219173,
      "grad_norm": 1.5058587789535522,
      "learning_rate": 9.78871052631579e-05,
      "loss": 0.4789,
      "step": 28042
    },
    {
      "epoch": 1.5703326240340463,
      "grad_norm": 1.1051394939422607,
      "learning_rate": 9.788684210526317e-05,
      "loss": 0.3422,
      "step": 28043
    },
    {
      "epoch": 1.5703886213461753,
      "grad_norm": 1.3431580066680908,
      "learning_rate": 9.788657894736843e-05,
      "loss": 0.5644,
      "step": 28044
    },
    {
      "epoch": 1.5704446186583043,
      "grad_norm": 1.3015185594558716,
      "learning_rate": 9.788631578947369e-05,
      "loss": 0.4857,
      "step": 28045
    },
    {
      "epoch": 1.5705006159704333,
      "grad_norm": 1.2908891439437866,
      "learning_rate": 9.788605263157895e-05,
      "loss": 0.3975,
      "step": 28046
    },
    {
      "epoch": 1.5705566132825624,
      "grad_norm": 1.5780909061431885,
      "learning_rate": 9.788578947368422e-05,
      "loss": 0.4592,
      "step": 28047
    },
    {
      "epoch": 1.5706126105946914,
      "grad_norm": 1.4913671016693115,
      "learning_rate": 9.788552631578948e-05,
      "loss": 0.6319,
      "step": 28048
    },
    {
      "epoch": 1.5706686079068204,
      "grad_norm": 1.4054170846939087,
      "learning_rate": 9.788526315789474e-05,
      "loss": 0.5016,
      "step": 28049
    },
    {
      "epoch": 1.5707246052189494,
      "grad_norm": 1.1858580112457275,
      "learning_rate": 9.7885e-05,
      "loss": 0.4828,
      "step": 28050
    },
    {
      "epoch": 1.5707806025310784,
      "grad_norm": 1.3672168254852295,
      "learning_rate": 9.788473684210526e-05,
      "loss": 0.3784,
      "step": 28051
    },
    {
      "epoch": 1.5708365998432074,
      "grad_norm": 1.2569587230682373,
      "learning_rate": 9.788447368421054e-05,
      "loss": 0.4334,
      "step": 28052
    },
    {
      "epoch": 1.5708925971553365,
      "grad_norm": 1.2256430387496948,
      "learning_rate": 9.78842105263158e-05,
      "loss": 0.4234,
      "step": 28053
    },
    {
      "epoch": 1.5709485944674655,
      "grad_norm": 1.0204970836639404,
      "learning_rate": 9.788394736842105e-05,
      "loss": 0.3367,
      "step": 28054
    },
    {
      "epoch": 1.5710045917795945,
      "grad_norm": 1.4459164142608643,
      "learning_rate": 9.788368421052631e-05,
      "loss": 0.4671,
      "step": 28055
    },
    {
      "epoch": 1.5710605890917235,
      "grad_norm": 1.5175567865371704,
      "learning_rate": 9.788342105263159e-05,
      "loss": 0.5397,
      "step": 28056
    },
    {
      "epoch": 1.5711165864038525,
      "grad_norm": 1.364989161491394,
      "learning_rate": 9.788315789473685e-05,
      "loss": 0.483,
      "step": 28057
    },
    {
      "epoch": 1.5711725837159816,
      "grad_norm": 1.1792104244232178,
      "learning_rate": 9.788289473684211e-05,
      "loss": 0.352,
      "step": 28058
    },
    {
      "epoch": 1.5712285810281106,
      "grad_norm": 1.5198942422866821,
      "learning_rate": 9.788263157894737e-05,
      "loss": 0.4844,
      "step": 28059
    },
    {
      "epoch": 1.5712845783402396,
      "grad_norm": 1.224273681640625,
      "learning_rate": 9.788236842105264e-05,
      "loss": 0.4047,
      "step": 28060
    },
    {
      "epoch": 1.5713405756523686,
      "grad_norm": 1.302113652229309,
      "learning_rate": 9.78821052631579e-05,
      "loss": 0.4535,
      "step": 28061
    },
    {
      "epoch": 1.5713965729644976,
      "grad_norm": 1.559851050376892,
      "learning_rate": 9.788184210526316e-05,
      "loss": 0.4636,
      "step": 28062
    },
    {
      "epoch": 1.5714525702766267,
      "grad_norm": 1.35964834690094,
      "learning_rate": 9.788157894736842e-05,
      "loss": 0.5024,
      "step": 28063
    },
    {
      "epoch": 1.5715085675887557,
      "grad_norm": 1.6735707521438599,
      "learning_rate": 9.788131578947369e-05,
      "loss": 0.4968,
      "step": 28064
    },
    {
      "epoch": 1.5715645649008847,
      "grad_norm": 1.1186984777450562,
      "learning_rate": 9.788105263157895e-05,
      "loss": 0.3354,
      "step": 28065
    },
    {
      "epoch": 1.5716205622130137,
      "grad_norm": 1.3679900169372559,
      "learning_rate": 9.788078947368423e-05,
      "loss": 0.3556,
      "step": 28066
    },
    {
      "epoch": 1.5716765595251427,
      "grad_norm": 1.2580618858337402,
      "learning_rate": 9.788052631578947e-05,
      "loss": 0.466,
      "step": 28067
    },
    {
      "epoch": 1.5717325568372718,
      "grad_norm": 1.267106056213379,
      "learning_rate": 9.788026315789473e-05,
      "loss": 0.4127,
      "step": 28068
    },
    {
      "epoch": 1.5717885541494008,
      "grad_norm": 3.595611810684204,
      "learning_rate": 9.788e-05,
      "loss": 0.5909,
      "step": 28069
    },
    {
      "epoch": 1.5718445514615298,
      "grad_norm": NaN,
      "learning_rate": 9.788e-05,
      "loss": 0.599,
      "step": 28070
    },
    {
      "epoch": 1.5719005487736588,
      "grad_norm": 1.5328335762023926,
      "learning_rate": 9.787973684210526e-05,
      "loss": 0.4543,
      "step": 28071
    },
    {
      "epoch": 1.5719565460857878,
      "grad_norm": 1.3158411979675293,
      "learning_rate": 9.787947368421054e-05,
      "loss": 0.488,
      "step": 28072
    },
    {
      "epoch": 1.5720125433979169,
      "grad_norm": 2.418255090713501,
      "learning_rate": 9.787921052631578e-05,
      "loss": 0.5711,
      "step": 28073
    },
    {
      "epoch": 1.5720685407100459,
      "grad_norm": 1.469887614250183,
      "learning_rate": 9.787894736842106e-05,
      "loss": 0.5755,
      "step": 28074
    },
    {
      "epoch": 1.5721245380221749,
      "grad_norm": 1.3137468099594116,
      "learning_rate": 9.787868421052632e-05,
      "loss": 0.3472,
      "step": 28075
    },
    {
      "epoch": 1.572180535334304,
      "grad_norm": 1.1552168130874634,
      "learning_rate": 9.787842105263159e-05,
      "loss": 0.5308,
      "step": 28076
    },
    {
      "epoch": 1.572236532646433,
      "grad_norm": 1.41217839717865,
      "learning_rate": 9.787815789473685e-05,
      "loss": 0.3818,
      "step": 28077
    },
    {
      "epoch": 1.572292529958562,
      "grad_norm": 1.1353222131729126,
      "learning_rate": 9.787789473684211e-05,
      "loss": 0.3699,
      "step": 28078
    },
    {
      "epoch": 1.572348527270691,
      "grad_norm": 1.1848223209381104,
      "learning_rate": 9.787763157894737e-05,
      "loss": 0.5062,
      "step": 28079
    },
    {
      "epoch": 1.57240452458282,
      "grad_norm": 1.1055452823638916,
      "learning_rate": 9.787736842105264e-05,
      "loss": 0.3496,
      "step": 28080
    },
    {
      "epoch": 1.572460521894949,
      "grad_norm": 1.4840240478515625,
      "learning_rate": 9.78771052631579e-05,
      "loss": 0.4124,
      "step": 28081
    },
    {
      "epoch": 1.572516519207078,
      "grad_norm": 1.1247225999832153,
      "learning_rate": 9.787684210526316e-05,
      "loss": 0.5187,
      "step": 28082
    },
    {
      "epoch": 1.572572516519207,
      "grad_norm": 1.0302581787109375,
      "learning_rate": 9.787657894736842e-05,
      "loss": 0.412,
      "step": 28083
    },
    {
      "epoch": 1.572628513831336,
      "grad_norm": 1.3668094873428345,
      "learning_rate": 9.78763157894737e-05,
      "loss": 0.4253,
      "step": 28084
    },
    {
      "epoch": 1.572684511143465,
      "grad_norm": 1.3305730819702148,
      "learning_rate": 9.787605263157895e-05,
      "loss": 0.5992,
      "step": 28085
    },
    {
      "epoch": 1.572740508455594,
      "grad_norm": 1.515182375907898,
      "learning_rate": 9.787578947368421e-05,
      "loss": 0.5052,
      "step": 28086
    },
    {
      "epoch": 1.5727965057677231,
      "grad_norm": 1.3288472890853882,
      "learning_rate": 9.787552631578947e-05,
      "loss": 0.4967,
      "step": 28087
    },
    {
      "epoch": 1.5728525030798521,
      "grad_norm": 1.110041856765747,
      "learning_rate": 9.787526315789473e-05,
      "loss": 0.3674,
      "step": 28088
    },
    {
      "epoch": 1.5729085003919812,
      "grad_norm": 1.2525770664215088,
      "learning_rate": 9.787500000000001e-05,
      "loss": 0.414,
      "step": 28089
    },
    {
      "epoch": 1.5729644977041102,
      "grad_norm": 1.2663404941558838,
      "learning_rate": 9.787473684210527e-05,
      "loss": 0.3943,
      "step": 28090
    },
    {
      "epoch": 1.5730204950162392,
      "grad_norm": 1.105630874633789,
      "learning_rate": 9.787447368421053e-05,
      "loss": 0.3995,
      "step": 28091
    },
    {
      "epoch": 1.5730764923283682,
      "grad_norm": 1.1396989822387695,
      "learning_rate": 9.787421052631579e-05,
      "loss": 0.4185,
      "step": 28092
    },
    {
      "epoch": 1.5731324896404972,
      "grad_norm": 1.2848345041275024,
      "learning_rate": 9.787394736842106e-05,
      "loss": 0.4938,
      "step": 28093
    },
    {
      "epoch": 1.5731884869526263,
      "grad_norm": 1.0787601470947266,
      "learning_rate": 9.787368421052632e-05,
      "loss": 0.4144,
      "step": 28094
    },
    {
      "epoch": 1.5732444842647553,
      "grad_norm": 1.9113361835479736,
      "learning_rate": 9.787342105263159e-05,
      "loss": 0.6329,
      "step": 28095
    },
    {
      "epoch": 1.5733004815768843,
      "grad_norm": 1.2810189723968506,
      "learning_rate": 9.787315789473684e-05,
      "loss": 0.4292,
      "step": 28096
    },
    {
      "epoch": 1.5733564788890133,
      "grad_norm": 1.7841038703918457,
      "learning_rate": 9.787289473684211e-05,
      "loss": 0.448,
      "step": 28097
    },
    {
      "epoch": 1.5734124762011423,
      "grad_norm": 1.3965585231781006,
      "learning_rate": 9.787263157894737e-05,
      "loss": 0.4491,
      "step": 28098
    },
    {
      "epoch": 1.5734684735132713,
      "grad_norm": 1.3444855213165283,
      "learning_rate": 9.787236842105265e-05,
      "loss": 0.4838,
      "step": 28099
    },
    {
      "epoch": 1.5735244708254004,
      "grad_norm": 1.3435307741165161,
      "learning_rate": 9.78721052631579e-05,
      "loss": 0.5611,
      "step": 28100
    },
    {
      "epoch": 1.5735804681375294,
      "grad_norm": 1.0784518718719482,
      "learning_rate": 9.787184210526316e-05,
      "loss": 0.3621,
      "step": 28101
    },
    {
      "epoch": 1.5736364654496584,
      "grad_norm": 1.5495529174804688,
      "learning_rate": 9.787157894736842e-05,
      "loss": 0.5277,
      "step": 28102
    },
    {
      "epoch": 1.5736924627617874,
      "grad_norm": 2.4598546028137207,
      "learning_rate": 9.78713157894737e-05,
      "loss": 0.5536,
      "step": 28103
    },
    {
      "epoch": 1.5737484600739164,
      "grad_norm": 1.3591283559799194,
      "learning_rate": 9.787105263157896e-05,
      "loss": 0.5098,
      "step": 28104
    },
    {
      "epoch": 1.5738044573860455,
      "grad_norm": 1.3528375625610352,
      "learning_rate": 9.78707894736842e-05,
      "loss": 0.3944,
      "step": 28105
    },
    {
      "epoch": 1.5738604546981745,
      "grad_norm": 1.5302882194519043,
      "learning_rate": 9.787052631578948e-05,
      "loss": 0.4248,
      "step": 28106
    },
    {
      "epoch": 1.5739164520103035,
      "grad_norm": 2.221705675125122,
      "learning_rate": 9.787026315789474e-05,
      "loss": 0.5651,
      "step": 28107
    },
    {
      "epoch": 1.5739724493224325,
      "grad_norm": 1.2829450368881226,
      "learning_rate": 9.787000000000001e-05,
      "loss": 0.4317,
      "step": 28108
    },
    {
      "epoch": 1.5740284466345615,
      "grad_norm": 1.266735315322876,
      "learning_rate": 9.786973684210527e-05,
      "loss": 0.4263,
      "step": 28109
    },
    {
      "epoch": 1.5740844439466906,
      "grad_norm": 1.3932968378067017,
      "learning_rate": 9.786947368421053e-05,
      "loss": 0.4782,
      "step": 28110
    },
    {
      "epoch": 1.5741404412588196,
      "grad_norm": 1.2708779573440552,
      "learning_rate": 9.786921052631579e-05,
      "loss": 0.4157,
      "step": 28111
    },
    {
      "epoch": 1.5741964385709486,
      "grad_norm": 1.2368707656860352,
      "learning_rate": 9.786894736842106e-05,
      "loss": 0.4547,
      "step": 28112
    },
    {
      "epoch": 1.5742524358830776,
      "grad_norm": 1.4338250160217285,
      "learning_rate": 9.786868421052632e-05,
      "loss": 0.3706,
      "step": 28113
    },
    {
      "epoch": 1.5743084331952066,
      "grad_norm": 1.2648639678955078,
      "learning_rate": 9.786842105263158e-05,
      "loss": 0.3937,
      "step": 28114
    },
    {
      "epoch": 1.5743644305073357,
      "grad_norm": 1.6365704536437988,
      "learning_rate": 9.786815789473684e-05,
      "loss": 0.5993,
      "step": 28115
    },
    {
      "epoch": 1.5744204278194647,
      "grad_norm": 1.3124196529388428,
      "learning_rate": 9.786789473684211e-05,
      "loss": 0.4459,
      "step": 28116
    },
    {
      "epoch": 1.5744764251315937,
      "grad_norm": 1.3489229679107666,
      "learning_rate": 9.786763157894737e-05,
      "loss": 0.4817,
      "step": 28117
    },
    {
      "epoch": 1.5745324224437227,
      "grad_norm": 1.2093262672424316,
      "learning_rate": 9.786736842105263e-05,
      "loss": 0.413,
      "step": 28118
    },
    {
      "epoch": 1.5745884197558517,
      "grad_norm": 1.512413501739502,
      "learning_rate": 9.78671052631579e-05,
      "loss": 0.5071,
      "step": 28119
    },
    {
      "epoch": 1.5746444170679808,
      "grad_norm": 1.3144752979278564,
      "learning_rate": 9.786684210526317e-05,
      "loss": 0.5384,
      "step": 28120
    },
    {
      "epoch": 1.5747004143801098,
      "grad_norm": 1.8020782470703125,
      "learning_rate": 9.786657894736843e-05,
      "loss": 0.4502,
      "step": 28121
    },
    {
      "epoch": 1.5747564116922388,
      "grad_norm": 1.2363632917404175,
      "learning_rate": 9.786631578947369e-05,
      "loss": 0.3991,
      "step": 28122
    },
    {
      "epoch": 1.5748124090043678,
      "grad_norm": 1.388232946395874,
      "learning_rate": 9.786605263157895e-05,
      "loss": 0.5205,
      "step": 28123
    },
    {
      "epoch": 1.5748684063164968,
      "grad_norm": 11.61746883392334,
      "learning_rate": 9.78657894736842e-05,
      "loss": 0.6034,
      "step": 28124
    },
    {
      "epoch": 1.5749244036286258,
      "grad_norm": 1.5589476823806763,
      "learning_rate": 9.786552631578948e-05,
      "loss": 0.7174,
      "step": 28125
    },
    {
      "epoch": 1.5749804009407549,
      "grad_norm": 1.4372823238372803,
      "learning_rate": 9.786526315789474e-05,
      "loss": 0.5682,
      "step": 28126
    },
    {
      "epoch": 1.5750363982528839,
      "grad_norm": 1.3344392776489258,
      "learning_rate": 9.786500000000001e-05,
      "loss": 0.485,
      "step": 28127
    },
    {
      "epoch": 1.575092395565013,
      "grad_norm": 1.3974255323410034,
      "learning_rate": 9.786473684210526e-05,
      "loss": 0.4426,
      "step": 28128
    },
    {
      "epoch": 1.575148392877142,
      "grad_norm": 1.2909753322601318,
      "learning_rate": 9.786447368421053e-05,
      "loss": 0.4372,
      "step": 28129
    },
    {
      "epoch": 1.575204390189271,
      "grad_norm": 1.6748439073562622,
      "learning_rate": 9.786421052631579e-05,
      "loss": 0.4385,
      "step": 28130
    },
    {
      "epoch": 1.5752603875014,
      "grad_norm": 1.1811232566833496,
      "learning_rate": 9.786394736842106e-05,
      "loss": 0.4178,
      "step": 28131
    },
    {
      "epoch": 1.575316384813529,
      "grad_norm": 1.3954200744628906,
      "learning_rate": 9.786368421052632e-05,
      "loss": 0.4656,
      "step": 28132
    },
    {
      "epoch": 1.575372382125658,
      "grad_norm": 1.1845266819000244,
      "learning_rate": 9.786342105263158e-05,
      "loss": 0.4493,
      "step": 28133
    },
    {
      "epoch": 1.575428379437787,
      "grad_norm": 1.4437274932861328,
      "learning_rate": 9.786315789473684e-05,
      "loss": 0.5061,
      "step": 28134
    },
    {
      "epoch": 1.575484376749916,
      "grad_norm": 1.2547589540481567,
      "learning_rate": 9.786289473684212e-05,
      "loss": 0.3944,
      "step": 28135
    },
    {
      "epoch": 1.575540374062045,
      "grad_norm": 1.269227385520935,
      "learning_rate": 9.786263157894738e-05,
      "loss": 0.3511,
      "step": 28136
    },
    {
      "epoch": 1.575596371374174,
      "grad_norm": 1.3924602270126343,
      "learning_rate": 9.786236842105264e-05,
      "loss": 0.4636,
      "step": 28137
    },
    {
      "epoch": 1.575652368686303,
      "grad_norm": 1.0934702157974243,
      "learning_rate": 9.78621052631579e-05,
      "loss": 0.3141,
      "step": 28138
    },
    {
      "epoch": 1.5757083659984321,
      "grad_norm": 1.1891965866088867,
      "learning_rate": 9.786184210526316e-05,
      "loss": 0.5611,
      "step": 28139
    },
    {
      "epoch": 1.5757643633105611,
      "grad_norm": 1.4275307655334473,
      "learning_rate": 9.786157894736843e-05,
      "loss": 0.4026,
      "step": 28140
    },
    {
      "epoch": 1.5758203606226902,
      "grad_norm": 1.2542580366134644,
      "learning_rate": 9.786131578947369e-05,
      "loss": 0.4041,
      "step": 28141
    },
    {
      "epoch": 1.5758763579348192,
      "grad_norm": 1.2023520469665527,
      "learning_rate": 9.786105263157895e-05,
      "loss": 0.5344,
      "step": 28142
    },
    {
      "epoch": 1.5759323552469482,
      "grad_norm": 1.3916306495666504,
      "learning_rate": 9.786078947368421e-05,
      "loss": 0.5923,
      "step": 28143
    },
    {
      "epoch": 1.5759883525590772,
      "grad_norm": 1.2810475826263428,
      "learning_rate": 9.786052631578948e-05,
      "loss": 0.4102,
      "step": 28144
    },
    {
      "epoch": 1.5760443498712062,
      "grad_norm": 1.1698429584503174,
      "learning_rate": 9.786026315789474e-05,
      "loss": 0.3531,
      "step": 28145
    },
    {
      "epoch": 1.5761003471833352,
      "grad_norm": 1.3877931833267212,
      "learning_rate": 9.786e-05,
      "loss": 0.4117,
      "step": 28146
    },
    {
      "epoch": 1.5761563444954643,
      "grad_norm": 1.289060354232788,
      "learning_rate": 9.785973684210526e-05,
      "loss": 0.4003,
      "step": 28147
    },
    {
      "epoch": 1.5762123418075933,
      "grad_norm": 1.2534656524658203,
      "learning_rate": 9.785947368421053e-05,
      "loss": 0.4815,
      "step": 28148
    },
    {
      "epoch": 1.5762683391197223,
      "grad_norm": 1.4801337718963623,
      "learning_rate": 9.78592105263158e-05,
      "loss": 0.5747,
      "step": 28149
    },
    {
      "epoch": 1.5763243364318513,
      "grad_norm": 1.0988131761550903,
      "learning_rate": 9.785894736842107e-05,
      "loss": 0.4168,
      "step": 28150
    },
    {
      "epoch": 1.5763803337439803,
      "grad_norm": 1.3799777030944824,
      "learning_rate": 9.785868421052631e-05,
      "loss": 0.6514,
      "step": 28151
    },
    {
      "epoch": 1.5764363310561094,
      "grad_norm": 1.220173716545105,
      "learning_rate": 9.785842105263159e-05,
      "loss": 0.4265,
      "step": 28152
    },
    {
      "epoch": 1.5764923283682384,
      "grad_norm": 1.2092936038970947,
      "learning_rate": 9.785815789473685e-05,
      "loss": 0.4341,
      "step": 28153
    },
    {
      "epoch": 1.5765483256803674,
      "grad_norm": 1.393599271774292,
      "learning_rate": 9.785789473684212e-05,
      "loss": 0.3641,
      "step": 28154
    },
    {
      "epoch": 1.5766043229924964,
      "grad_norm": 1.2767813205718994,
      "learning_rate": 9.785763157894737e-05,
      "loss": 0.5336,
      "step": 28155
    },
    {
      "epoch": 1.5766603203046254,
      "grad_norm": 1.2935755252838135,
      "learning_rate": 9.785736842105263e-05,
      "loss": 0.5034,
      "step": 28156
    },
    {
      "epoch": 1.5767163176167545,
      "grad_norm": 1.108004093170166,
      "learning_rate": 9.78571052631579e-05,
      "loss": 0.4588,
      "step": 28157
    },
    {
      "epoch": 1.5767723149288835,
      "grad_norm": 1.0608549118041992,
      "learning_rate": 9.785684210526316e-05,
      "loss": 0.4156,
      "step": 28158
    },
    {
      "epoch": 1.5768283122410125,
      "grad_norm": 1.2897781133651733,
      "learning_rate": 9.785657894736843e-05,
      "loss": 0.4719,
      "step": 28159
    },
    {
      "epoch": 1.5768843095531415,
      "grad_norm": 1.090461015701294,
      "learning_rate": 9.785631578947368e-05,
      "loss": 0.3136,
      "step": 28160
    },
    {
      "epoch": 1.5769403068652705,
      "grad_norm": 1.243940830230713,
      "learning_rate": 9.785605263157895e-05,
      "loss": 0.3999,
      "step": 28161
    },
    {
      "epoch": 1.5769963041773996,
      "grad_norm": 1.2789543867111206,
      "learning_rate": 9.785578947368421e-05,
      "loss": 0.478,
      "step": 28162
    },
    {
      "epoch": 1.5770523014895286,
      "grad_norm": 1.2415505647659302,
      "learning_rate": 9.785552631578948e-05,
      "loss": 0.4176,
      "step": 28163
    },
    {
      "epoch": 1.5771082988016576,
      "grad_norm": 1.3886882066726685,
      "learning_rate": 9.785526315789474e-05,
      "loss": 0.4695,
      "step": 28164
    },
    {
      "epoch": 1.5771642961137866,
      "grad_norm": 1.1668152809143066,
      "learning_rate": 9.7855e-05,
      "loss": 0.3876,
      "step": 28165
    },
    {
      "epoch": 1.5772202934259156,
      "grad_norm": 1.349058747291565,
      "learning_rate": 9.785473684210526e-05,
      "loss": 0.5779,
      "step": 28166
    },
    {
      "epoch": 1.5772762907380447,
      "grad_norm": 1.3069305419921875,
      "learning_rate": 9.785447368421054e-05,
      "loss": 0.4907,
      "step": 28167
    },
    {
      "epoch": 1.5773322880501737,
      "grad_norm": 1.1267633438110352,
      "learning_rate": 9.78542105263158e-05,
      "loss": 0.3592,
      "step": 28168
    },
    {
      "epoch": 1.5773882853623027,
      "grad_norm": 1.205538272857666,
      "learning_rate": 9.785394736842106e-05,
      "loss": 0.4373,
      "step": 28169
    },
    {
      "epoch": 1.5774442826744317,
      "grad_norm": 1.0698776245117188,
      "learning_rate": 9.785368421052632e-05,
      "loss": 0.4397,
      "step": 28170
    },
    {
      "epoch": 1.5775002799865607,
      "grad_norm": 1.385589361190796,
      "learning_rate": 9.785342105263159e-05,
      "loss": 0.433,
      "step": 28171
    },
    {
      "epoch": 1.5775562772986897,
      "grad_norm": 1.3485287427902222,
      "learning_rate": 9.785315789473685e-05,
      "loss": 0.5212,
      "step": 28172
    },
    {
      "epoch": 1.5776122746108188,
      "grad_norm": 1.1175944805145264,
      "learning_rate": 9.785289473684211e-05,
      "loss": 0.3932,
      "step": 28173
    },
    {
      "epoch": 1.5776682719229478,
      "grad_norm": 1.5841163396835327,
      "learning_rate": 9.785263157894737e-05,
      "loss": 0.4892,
      "step": 28174
    },
    {
      "epoch": 1.5777242692350768,
      "grad_norm": 1.2950897216796875,
      "learning_rate": 9.785236842105263e-05,
      "loss": 0.4751,
      "step": 28175
    },
    {
      "epoch": 1.5777802665472058,
      "grad_norm": 1.267866611480713,
      "learning_rate": 9.78521052631579e-05,
      "loss": 0.3934,
      "step": 28176
    },
    {
      "epoch": 1.5778362638593348,
      "grad_norm": 1.0617599487304688,
      "learning_rate": 9.785184210526316e-05,
      "loss": 0.4001,
      "step": 28177
    },
    {
      "epoch": 1.5778922611714639,
      "grad_norm": 1.4428179264068604,
      "learning_rate": 9.785157894736842e-05,
      "loss": 0.5607,
      "step": 28178
    },
    {
      "epoch": 1.5779482584835929,
      "grad_norm": 1.1948655843734741,
      "learning_rate": 9.785131578947368e-05,
      "loss": 0.4326,
      "step": 28179
    },
    {
      "epoch": 1.578004255795722,
      "grad_norm": 1.6858242750167847,
      "learning_rate": 9.785105263157895e-05,
      "loss": 0.4088,
      "step": 28180
    },
    {
      "epoch": 1.578060253107851,
      "grad_norm": 1.232558250427246,
      "learning_rate": 9.785078947368421e-05,
      "loss": 0.4657,
      "step": 28181
    },
    {
      "epoch": 1.57811625041998,
      "grad_norm": 1.5232340097427368,
      "learning_rate": 9.785052631578949e-05,
      "loss": 0.5507,
      "step": 28182
    },
    {
      "epoch": 1.578172247732109,
      "grad_norm": 1.240832805633545,
      "learning_rate": 9.785026315789473e-05,
      "loss": 0.437,
      "step": 28183
    },
    {
      "epoch": 1.578228245044238,
      "grad_norm": 1.556320071220398,
      "learning_rate": 9.785e-05,
      "loss": 0.4443,
      "step": 28184
    },
    {
      "epoch": 1.578284242356367,
      "grad_norm": 1.4046788215637207,
      "learning_rate": 9.784973684210527e-05,
      "loss": 0.4333,
      "step": 28185
    },
    {
      "epoch": 1.578340239668496,
      "grad_norm": 1.3440078496932983,
      "learning_rate": 9.784947368421054e-05,
      "loss": 0.39,
      "step": 28186
    },
    {
      "epoch": 1.578396236980625,
      "grad_norm": 1.4519109725952148,
      "learning_rate": 9.78492105263158e-05,
      "loss": 0.5872,
      "step": 28187
    },
    {
      "epoch": 1.578452234292754,
      "grad_norm": 1.3294261693954468,
      "learning_rate": 9.784894736842106e-05,
      "loss": 0.5281,
      "step": 28188
    },
    {
      "epoch": 1.578508231604883,
      "grad_norm": 1.7994285821914673,
      "learning_rate": 9.784868421052632e-05,
      "loss": 0.4514,
      "step": 28189
    },
    {
      "epoch": 1.578564228917012,
      "grad_norm": 1.1187090873718262,
      "learning_rate": 9.784842105263158e-05,
      "loss": 0.4704,
      "step": 28190
    },
    {
      "epoch": 1.578620226229141,
      "grad_norm": 1.0622961521148682,
      "learning_rate": 9.784815789473685e-05,
      "loss": 0.3963,
      "step": 28191
    },
    {
      "epoch": 1.5786762235412701,
      "grad_norm": 1.4263650178909302,
      "learning_rate": 9.784789473684211e-05,
      "loss": 0.5271,
      "step": 28192
    },
    {
      "epoch": 1.5787322208533991,
      "grad_norm": 1.0818216800689697,
      "learning_rate": 9.784763157894737e-05,
      "loss": 0.4412,
      "step": 28193
    },
    {
      "epoch": 1.5787882181655282,
      "grad_norm": 1.3329949378967285,
      "learning_rate": 9.784736842105263e-05,
      "loss": 0.5599,
      "step": 28194
    },
    {
      "epoch": 1.5788442154776572,
      "grad_norm": 1.3518515825271606,
      "learning_rate": 9.78471052631579e-05,
      "loss": 0.3642,
      "step": 28195
    },
    {
      "epoch": 1.5789002127897862,
      "grad_norm": 1.2396793365478516,
      "learning_rate": 9.784684210526316e-05,
      "loss": 0.4187,
      "step": 28196
    },
    {
      "epoch": 1.5789562101019152,
      "grad_norm": 1.3476207256317139,
      "learning_rate": 9.784657894736842e-05,
      "loss": 0.3397,
      "step": 28197
    },
    {
      "epoch": 1.5790122074140442,
      "grad_norm": 1.346340537071228,
      "learning_rate": 9.784631578947368e-05,
      "loss": 0.4135,
      "step": 28198
    },
    {
      "epoch": 1.5790682047261733,
      "grad_norm": 1.3594069480895996,
      "learning_rate": 9.784605263157896e-05,
      "loss": 0.3856,
      "step": 28199
    },
    {
      "epoch": 1.5791242020383023,
      "grad_norm": 1.5098919868469238,
      "learning_rate": 9.784578947368422e-05,
      "loss": 0.3712,
      "step": 28200
    },
    {
      "epoch": 1.5791801993504313,
      "grad_norm": 1.3382796049118042,
      "learning_rate": 9.784552631578948e-05,
      "loss": 0.4768,
      "step": 28201
    },
    {
      "epoch": 1.5792361966625603,
      "grad_norm": 1.1051304340362549,
      "learning_rate": 9.784526315789474e-05,
      "loss": 0.3683,
      "step": 28202
    },
    {
      "epoch": 1.5792921939746893,
      "grad_norm": 1.24118971824646,
      "learning_rate": 9.784500000000001e-05,
      "loss": 0.4999,
      "step": 28203
    },
    {
      "epoch": 1.5793481912868184,
      "grad_norm": 1.661124348640442,
      "learning_rate": 9.784473684210527e-05,
      "loss": 0.7368,
      "step": 28204
    },
    {
      "epoch": 1.5794041885989474,
      "grad_norm": 1.280503511428833,
      "learning_rate": 9.784447368421054e-05,
      "loss": 0.5091,
      "step": 28205
    },
    {
      "epoch": 1.5794601859110764,
      "grad_norm": 1.2633830308914185,
      "learning_rate": 9.784421052631579e-05,
      "loss": 0.4254,
      "step": 28206
    },
    {
      "epoch": 1.5795161832232054,
      "grad_norm": 1.272430419921875,
      "learning_rate": 9.784394736842106e-05,
      "loss": 0.4289,
      "step": 28207
    },
    {
      "epoch": 1.5795721805353344,
      "grad_norm": 1.4057821035385132,
      "learning_rate": 9.784368421052632e-05,
      "loss": 0.3939,
      "step": 28208
    },
    {
      "epoch": 1.5796281778474635,
      "grad_norm": 1.5706238746643066,
      "learning_rate": 9.784342105263158e-05,
      "loss": 0.4638,
      "step": 28209
    },
    {
      "epoch": 1.5796841751595925,
      "grad_norm": 1.3825490474700928,
      "learning_rate": 9.784315789473684e-05,
      "loss": 0.397,
      "step": 28210
    },
    {
      "epoch": 1.5797401724717215,
      "grad_norm": 1.3392679691314697,
      "learning_rate": 9.78428947368421e-05,
      "loss": 0.4503,
      "step": 28211
    },
    {
      "epoch": 1.5797961697838505,
      "grad_norm": 1.4337555170059204,
      "learning_rate": 9.784263157894737e-05,
      "loss": 0.5699,
      "step": 28212
    },
    {
      "epoch": 1.5798521670959795,
      "grad_norm": 1.576265811920166,
      "learning_rate": 9.784236842105263e-05,
      "loss": 0.5449,
      "step": 28213
    },
    {
      "epoch": 1.5799081644081086,
      "grad_norm": 1.4735445976257324,
      "learning_rate": 9.784210526315791e-05,
      "loss": 0.4527,
      "step": 28214
    },
    {
      "epoch": 1.5799641617202376,
      "grad_norm": 1.4079476594924927,
      "learning_rate": 9.784184210526315e-05,
      "loss": 0.3421,
      "step": 28215
    },
    {
      "epoch": 1.5800201590323666,
      "grad_norm": 1.2776329517364502,
      "learning_rate": 9.784157894736843e-05,
      "loss": 0.461,
      "step": 28216
    },
    {
      "epoch": 1.5800761563444956,
      "grad_norm": 1.2142679691314697,
      "learning_rate": 9.784131578947369e-05,
      "loss": 0.3766,
      "step": 28217
    },
    {
      "epoch": 1.5801321536566246,
      "grad_norm": 1.3883777856826782,
      "learning_rate": 9.784105263157896e-05,
      "loss": 0.454,
      "step": 28218
    },
    {
      "epoch": 1.5801881509687536,
      "grad_norm": 1.1547354459762573,
      "learning_rate": 9.784078947368422e-05,
      "loss": 0.477,
      "step": 28219
    },
    {
      "epoch": 1.5802441482808827,
      "grad_norm": 1.2915985584259033,
      "learning_rate": 9.784052631578948e-05,
      "loss": 0.4815,
      "step": 28220
    },
    {
      "epoch": 1.5803001455930117,
      "grad_norm": 1.478183388710022,
      "learning_rate": 9.784026315789474e-05,
      "loss": 0.4279,
      "step": 28221
    },
    {
      "epoch": 1.5803561429051407,
      "grad_norm": 1.2574313879013062,
      "learning_rate": 9.784000000000001e-05,
      "loss": 0.508,
      "step": 28222
    },
    {
      "epoch": 1.5804121402172697,
      "grad_norm": 10.788265228271484,
      "learning_rate": 9.783973684210527e-05,
      "loss": 0.4428,
      "step": 28223
    },
    {
      "epoch": 1.5804681375293987,
      "grad_norm": 1.188437581062317,
      "learning_rate": 9.783947368421053e-05,
      "loss": 0.3691,
      "step": 28224
    },
    {
      "epoch": 1.5805241348415278,
      "grad_norm": 1.3362503051757812,
      "learning_rate": 9.783921052631579e-05,
      "loss": 0.4179,
      "step": 28225
    },
    {
      "epoch": 1.5805801321536568,
      "grad_norm": 1.9426575899124146,
      "learning_rate": 9.783894736842105e-05,
      "loss": 0.5131,
      "step": 28226
    },
    {
      "epoch": 1.5806361294657858,
      "grad_norm": 1.151312232017517,
      "learning_rate": 9.783868421052632e-05,
      "loss": 0.3808,
      "step": 28227
    },
    {
      "epoch": 1.5806921267779148,
      "grad_norm": 1.3491356372833252,
      "learning_rate": 9.783842105263158e-05,
      "loss": 0.4749,
      "step": 28228
    },
    {
      "epoch": 1.5807481240900438,
      "grad_norm": 1.3050971031188965,
      "learning_rate": 9.783815789473684e-05,
      "loss": 0.4173,
      "step": 28229
    },
    {
      "epoch": 1.5808041214021729,
      "grad_norm": 1.3836475610733032,
      "learning_rate": 9.78378947368421e-05,
      "loss": 0.4225,
      "step": 28230
    },
    {
      "epoch": 1.5808601187143019,
      "grad_norm": 1.2909274101257324,
      "learning_rate": 9.783763157894738e-05,
      "loss": 0.5205,
      "step": 28231
    },
    {
      "epoch": 1.580916116026431,
      "grad_norm": 1.6245766878128052,
      "learning_rate": 9.783736842105264e-05,
      "loss": 0.5249,
      "step": 28232
    },
    {
      "epoch": 1.58097211333856,
      "grad_norm": 1.3622593879699707,
      "learning_rate": 9.78371052631579e-05,
      "loss": 0.3795,
      "step": 28233
    },
    {
      "epoch": 1.5810281106506887,
      "grad_norm": 1.450543999671936,
      "learning_rate": 9.783684210526316e-05,
      "loss": 0.4643,
      "step": 28234
    },
    {
      "epoch": 1.5810841079628177,
      "grad_norm": 1.3439433574676514,
      "learning_rate": 9.783657894736843e-05,
      "loss": 0.6194,
      "step": 28235
    },
    {
      "epoch": 1.5811401052749467,
      "grad_norm": 1.3028042316436768,
      "learning_rate": 9.783631578947369e-05,
      "loss": 0.4211,
      "step": 28236
    },
    {
      "epoch": 1.5811961025870758,
      "grad_norm": 1.1072126626968384,
      "learning_rate": 9.783605263157896e-05,
      "loss": 0.3549,
      "step": 28237
    },
    {
      "epoch": 1.5812520998992048,
      "grad_norm": 1.2173089981079102,
      "learning_rate": 9.783578947368421e-05,
      "loss": 0.4091,
      "step": 28238
    },
    {
      "epoch": 1.5813080972113338,
      "grad_norm": 1.1502578258514404,
      "learning_rate": 9.783552631578948e-05,
      "loss": 0.4709,
      "step": 28239
    },
    {
      "epoch": 1.5813640945234628,
      "grad_norm": 1.3251762390136719,
      "learning_rate": 9.783526315789474e-05,
      "loss": 0.5133,
      "step": 28240
    },
    {
      "epoch": 1.5814200918355918,
      "grad_norm": 1.4343903064727783,
      "learning_rate": 9.783500000000001e-05,
      "loss": 0.4494,
      "step": 28241
    },
    {
      "epoch": 1.5814760891477209,
      "grad_norm": 1.1779705286026,
      "learning_rate": 9.783473684210527e-05,
      "loss": 0.55,
      "step": 28242
    },
    {
      "epoch": 1.5815320864598499,
      "grad_norm": 1.5313528776168823,
      "learning_rate": 9.783447368421052e-05,
      "loss": 0.4,
      "step": 28243
    },
    {
      "epoch": 1.581588083771979,
      "grad_norm": 1.4285385608673096,
      "learning_rate": 9.78342105263158e-05,
      "loss": 0.5774,
      "step": 28244
    },
    {
      "epoch": 1.581644081084108,
      "grad_norm": 1.9072041511535645,
      "learning_rate": 9.783394736842105e-05,
      "loss": 0.4054,
      "step": 28245
    },
    {
      "epoch": 1.581700078396237,
      "grad_norm": 1.3752843141555786,
      "learning_rate": 9.783368421052633e-05,
      "loss": 0.423,
      "step": 28246
    },
    {
      "epoch": 1.581756075708366,
      "grad_norm": 1.5470937490463257,
      "learning_rate": 9.783342105263159e-05,
      "loss": 0.4605,
      "step": 28247
    },
    {
      "epoch": 1.581812073020495,
      "grad_norm": 1.4123104810714722,
      "learning_rate": 9.783315789473685e-05,
      "loss": 0.5767,
      "step": 28248
    },
    {
      "epoch": 1.581868070332624,
      "grad_norm": 1.4098743200302124,
      "learning_rate": 9.78328947368421e-05,
      "loss": 0.4671,
      "step": 28249
    },
    {
      "epoch": 1.581924067644753,
      "grad_norm": 1.2538591623306274,
      "learning_rate": 9.783263157894738e-05,
      "loss": 0.3655,
      "step": 28250
    },
    {
      "epoch": 1.581980064956882,
      "grad_norm": 1.244978904724121,
      "learning_rate": 9.783236842105264e-05,
      "loss": 0.3907,
      "step": 28251
    },
    {
      "epoch": 1.582036062269011,
      "grad_norm": 1.2174460887908936,
      "learning_rate": 9.78321052631579e-05,
      "loss": 0.3742,
      "step": 28252
    },
    {
      "epoch": 1.58209205958114,
      "grad_norm": 16.026084899902344,
      "learning_rate": 9.783184210526316e-05,
      "loss": 0.5541,
      "step": 28253
    },
    {
      "epoch": 1.582148056893269,
      "grad_norm": 1.2309362888336182,
      "learning_rate": 9.783157894736843e-05,
      "loss": 0.4567,
      "step": 28254
    },
    {
      "epoch": 1.5822040542053981,
      "grad_norm": 1.18257474899292,
      "learning_rate": 9.783131578947369e-05,
      "loss": 0.457,
      "step": 28255
    },
    {
      "epoch": 1.5822600515175271,
      "grad_norm": 1.2617989778518677,
      "learning_rate": 9.783105263157895e-05,
      "loss": 0.4647,
      "step": 28256
    },
    {
      "epoch": 1.5823160488296562,
      "grad_norm": 1.1995033025741577,
      "learning_rate": 9.783078947368421e-05,
      "loss": 0.5024,
      "step": 28257
    },
    {
      "epoch": 1.5823720461417852,
      "grad_norm": 1.1797460317611694,
      "learning_rate": 9.783052631578948e-05,
      "loss": 0.5191,
      "step": 28258
    },
    {
      "epoch": 1.5824280434539142,
      "grad_norm": 1.3838032484054565,
      "learning_rate": 9.783026315789474e-05,
      "loss": 0.488,
      "step": 28259
    },
    {
      "epoch": 1.5824840407660432,
      "grad_norm": 1.5013700723648071,
      "learning_rate": 9.783e-05,
      "loss": 0.4724,
      "step": 28260
    },
    {
      "epoch": 1.5825400380781722,
      "grad_norm": 1.1968519687652588,
      "learning_rate": 9.782973684210526e-05,
      "loss": 0.4432,
      "step": 28261
    },
    {
      "epoch": 1.5825960353903012,
      "grad_norm": 1.2762856483459473,
      "learning_rate": 9.782947368421052e-05,
      "loss": 0.4813,
      "step": 28262
    },
    {
      "epoch": 1.5826520327024303,
      "grad_norm": 1.4965077638626099,
      "learning_rate": 9.78292105263158e-05,
      "loss": 0.4058,
      "step": 28263
    },
    {
      "epoch": 1.5827080300145593,
      "grad_norm": 1.2279053926467896,
      "learning_rate": 9.782894736842106e-05,
      "loss": 0.4609,
      "step": 28264
    },
    {
      "epoch": 1.5827640273266883,
      "grad_norm": 1.1062767505645752,
      "learning_rate": 9.782868421052632e-05,
      "loss": 0.3519,
      "step": 28265
    },
    {
      "epoch": 1.5828200246388173,
      "grad_norm": 1.3223053216934204,
      "learning_rate": 9.782842105263158e-05,
      "loss": 0.5096,
      "step": 28266
    },
    {
      "epoch": 1.5828760219509463,
      "grad_norm": 1.2049095630645752,
      "learning_rate": 9.782815789473685e-05,
      "loss": 0.4288,
      "step": 28267
    },
    {
      "epoch": 1.5829320192630754,
      "grad_norm": 1.3944287300109863,
      "learning_rate": 9.782789473684211e-05,
      "loss": 0.4298,
      "step": 28268
    },
    {
      "epoch": 1.5829880165752044,
      "grad_norm": 1.3010836839675903,
      "learning_rate": 9.782763157894738e-05,
      "loss": 0.3869,
      "step": 28269
    },
    {
      "epoch": 1.5830440138873334,
      "grad_norm": 1.4387359619140625,
      "learning_rate": 9.782736842105263e-05,
      "loss": 0.5562,
      "step": 28270
    },
    {
      "epoch": 1.5831000111994624,
      "grad_norm": 1.1513671875,
      "learning_rate": 9.78271052631579e-05,
      "loss": 0.4648,
      "step": 28271
    },
    {
      "epoch": 1.5831560085115914,
      "grad_norm": 1.276419997215271,
      "learning_rate": 9.782684210526316e-05,
      "loss": 0.5492,
      "step": 28272
    },
    {
      "epoch": 1.5832120058237205,
      "grad_norm": 1.277443766593933,
      "learning_rate": 9.782657894736843e-05,
      "loss": 0.4696,
      "step": 28273
    },
    {
      "epoch": 1.5832680031358495,
      "grad_norm": 1.727570652961731,
      "learning_rate": 9.78263157894737e-05,
      "loss": 0.5006,
      "step": 28274
    },
    {
      "epoch": 1.5833240004479785,
      "grad_norm": 1.3538728952407837,
      "learning_rate": 9.782605263157895e-05,
      "loss": 0.5115,
      "step": 28275
    },
    {
      "epoch": 1.5833799977601075,
      "grad_norm": 1.189645767211914,
      "learning_rate": 9.782578947368421e-05,
      "loss": 0.4491,
      "step": 28276
    },
    {
      "epoch": 1.5834359950722365,
      "grad_norm": 1.426240086555481,
      "learning_rate": 9.782552631578947e-05,
      "loss": 0.5695,
      "step": 28277
    },
    {
      "epoch": 1.5834919923843656,
      "grad_norm": 1.1678203344345093,
      "learning_rate": 9.782526315789475e-05,
      "loss": 0.4941,
      "step": 28278
    },
    {
      "epoch": 1.5835479896964946,
      "grad_norm": 1.3237804174423218,
      "learning_rate": 9.7825e-05,
      "loss": 0.4183,
      "step": 28279
    },
    {
      "epoch": 1.5836039870086236,
      "grad_norm": 1.8183512687683105,
      "learning_rate": 9.782473684210527e-05,
      "loss": 0.485,
      "step": 28280
    },
    {
      "epoch": 1.5836599843207526,
      "grad_norm": 1.3586292266845703,
      "learning_rate": 9.782447368421053e-05,
      "loss": 0.5133,
      "step": 28281
    },
    {
      "epoch": 1.5837159816328816,
      "grad_norm": 1.332970142364502,
      "learning_rate": 9.78242105263158e-05,
      "loss": 0.4322,
      "step": 28282
    },
    {
      "epoch": 1.5837719789450107,
      "grad_norm": 1.2447971105575562,
      "learning_rate": 9.782394736842106e-05,
      "loss": 0.4877,
      "step": 28283
    },
    {
      "epoch": 1.5838279762571397,
      "grad_norm": 1.1528841257095337,
      "learning_rate": 9.782368421052632e-05,
      "loss": 0.411,
      "step": 28284
    },
    {
      "epoch": 1.5838839735692687,
      "grad_norm": 1.331696629524231,
      "learning_rate": 9.782342105263158e-05,
      "loss": 0.5132,
      "step": 28285
    },
    {
      "epoch": 1.5839399708813977,
      "grad_norm": 1.526389718055725,
      "learning_rate": 9.782315789473685e-05,
      "loss": 0.5063,
      "step": 28286
    },
    {
      "epoch": 1.5839959681935267,
      "grad_norm": 1.220671534538269,
      "learning_rate": 9.782289473684211e-05,
      "loss": 0.4549,
      "step": 28287
    },
    {
      "epoch": 1.5840519655056557,
      "grad_norm": 1.4934096336364746,
      "learning_rate": 9.782263157894737e-05,
      "loss": 0.5103,
      "step": 28288
    },
    {
      "epoch": 1.5841079628177848,
      "grad_norm": 1.101176381111145,
      "learning_rate": 9.782236842105263e-05,
      "loss": 0.3865,
      "step": 28289
    },
    {
      "epoch": 1.5841639601299138,
      "grad_norm": 1.2829971313476562,
      "learning_rate": 9.78221052631579e-05,
      "loss": 0.3926,
      "step": 28290
    },
    {
      "epoch": 1.5842199574420428,
      "grad_norm": 1.3696972131729126,
      "learning_rate": 9.782184210526316e-05,
      "loss": 0.6527,
      "step": 28291
    },
    {
      "epoch": 1.5842759547541718,
      "grad_norm": 1.5396428108215332,
      "learning_rate": 9.782157894736844e-05,
      "loss": 0.5814,
      "step": 28292
    },
    {
      "epoch": 1.5843319520663008,
      "grad_norm": 1.616563081741333,
      "learning_rate": 9.782131578947368e-05,
      "loss": 0.5543,
      "step": 28293
    },
    {
      "epoch": 1.5843879493784299,
      "grad_norm": 1.4253442287445068,
      "learning_rate": 9.782105263157894e-05,
      "loss": 0.4605,
      "step": 28294
    },
    {
      "epoch": 1.5844439466905589,
      "grad_norm": 1.382038950920105,
      "learning_rate": 9.782078947368422e-05,
      "loss": 0.4453,
      "step": 28295
    },
    {
      "epoch": 1.584499944002688,
      "grad_norm": 1.2013399600982666,
      "learning_rate": 9.782052631578948e-05,
      "loss": 0.4356,
      "step": 28296
    },
    {
      "epoch": 1.584555941314817,
      "grad_norm": 1.2636768817901611,
      "learning_rate": 9.782026315789475e-05,
      "loss": 0.4449,
      "step": 28297
    },
    {
      "epoch": 1.584611938626946,
      "grad_norm": 1.3658510446548462,
      "learning_rate": 9.782e-05,
      "loss": 0.5393,
      "step": 28298
    },
    {
      "epoch": 1.584667935939075,
      "grad_norm": 1.285614252090454,
      "learning_rate": 9.781973684210527e-05,
      "loss": 0.4172,
      "step": 28299
    },
    {
      "epoch": 1.584723933251204,
      "grad_norm": 1.1509438753128052,
      "learning_rate": 9.781947368421053e-05,
      "loss": 0.4589,
      "step": 28300
    },
    {
      "epoch": 1.584779930563333,
      "grad_norm": 1.444385051727295,
      "learning_rate": 9.78192105263158e-05,
      "loss": 0.4409,
      "step": 28301
    },
    {
      "epoch": 1.584835927875462,
      "grad_norm": 1.872693657875061,
      "learning_rate": 9.781894736842106e-05,
      "loss": 0.5809,
      "step": 28302
    },
    {
      "epoch": 1.584891925187591,
      "grad_norm": 1.6845189332962036,
      "learning_rate": 9.781868421052632e-05,
      "loss": 0.4986,
      "step": 28303
    },
    {
      "epoch": 1.58494792249972,
      "grad_norm": 1.6680781841278076,
      "learning_rate": 9.781842105263158e-05,
      "loss": 0.5092,
      "step": 28304
    },
    {
      "epoch": 1.585003919811849,
      "grad_norm": 1.7148813009262085,
      "learning_rate": 9.781815789473685e-05,
      "loss": 0.564,
      "step": 28305
    },
    {
      "epoch": 1.585059917123978,
      "grad_norm": 1.5087623596191406,
      "learning_rate": 9.781789473684211e-05,
      "loss": 0.6449,
      "step": 28306
    },
    {
      "epoch": 1.585115914436107,
      "grad_norm": 1.4534099102020264,
      "learning_rate": 9.781763157894737e-05,
      "loss": 0.4322,
      "step": 28307
    },
    {
      "epoch": 1.5851719117482361,
      "grad_norm": 1.3164526224136353,
      "learning_rate": 9.781736842105263e-05,
      "loss": 0.354,
      "step": 28308
    },
    {
      "epoch": 1.5852279090603651,
      "grad_norm": 1.0275143384933472,
      "learning_rate": 9.78171052631579e-05,
      "loss": 0.2847,
      "step": 28309
    },
    {
      "epoch": 1.5852839063724942,
      "grad_norm": 1.2909725904464722,
      "learning_rate": 9.781684210526317e-05,
      "loss": 0.4754,
      "step": 28310
    },
    {
      "epoch": 1.5853399036846232,
      "grad_norm": 1.3653693199157715,
      "learning_rate": 9.781657894736843e-05,
      "loss": 0.3985,
      "step": 28311
    },
    {
      "epoch": 1.5853959009967522,
      "grad_norm": 1.3306163549423218,
      "learning_rate": 9.781631578947369e-05,
      "loss": 0.419,
      "step": 28312
    },
    {
      "epoch": 1.5854518983088812,
      "grad_norm": 1.3650118112564087,
      "learning_rate": 9.781605263157895e-05,
      "loss": 0.4289,
      "step": 28313
    },
    {
      "epoch": 1.5855078956210102,
      "grad_norm": 1.4015605449676514,
      "learning_rate": 9.781578947368422e-05,
      "loss": 0.4585,
      "step": 28314
    },
    {
      "epoch": 1.5855638929331393,
      "grad_norm": 1.2568800449371338,
      "learning_rate": 9.781552631578948e-05,
      "loss": 0.4393,
      "step": 28315
    },
    {
      "epoch": 1.5856198902452683,
      "grad_norm": 1.413943886756897,
      "learning_rate": 9.781526315789474e-05,
      "loss": 0.4012,
      "step": 28316
    },
    {
      "epoch": 1.585675887557397,
      "grad_norm": 1.41544771194458,
      "learning_rate": 9.7815e-05,
      "loss": 0.3876,
      "step": 28317
    },
    {
      "epoch": 1.585731884869526,
      "grad_norm": 1.333017349243164,
      "learning_rate": 9.781473684210527e-05,
      "loss": 0.5692,
      "step": 28318
    },
    {
      "epoch": 1.5857878821816551,
      "grad_norm": 1.1712980270385742,
      "learning_rate": 9.781447368421053e-05,
      "loss": 0.3676,
      "step": 28319
    },
    {
      "epoch": 1.5858438794937841,
      "grad_norm": 1.17991042137146,
      "learning_rate": 9.781421052631579e-05,
      "loss": 0.4407,
      "step": 28320
    },
    {
      "epoch": 1.5858998768059132,
      "grad_norm": 1.4239095449447632,
      "learning_rate": 9.781394736842105e-05,
      "loss": 0.5732,
      "step": 28321
    },
    {
      "epoch": 1.5859558741180422,
      "grad_norm": 2.375763177871704,
      "learning_rate": 9.781368421052632e-05,
      "loss": 0.54,
      "step": 28322
    },
    {
      "epoch": 1.5860118714301712,
      "grad_norm": 1.3151723146438599,
      "learning_rate": 9.781342105263158e-05,
      "loss": 0.3928,
      "step": 28323
    },
    {
      "epoch": 1.5860678687423002,
      "grad_norm": 1.134048342704773,
      "learning_rate": 9.781315789473686e-05,
      "loss": 0.4171,
      "step": 28324
    },
    {
      "epoch": 1.5861238660544292,
      "grad_norm": 1.9114692211151123,
      "learning_rate": 9.78128947368421e-05,
      "loss": 0.3269,
      "step": 28325
    },
    {
      "epoch": 1.5861798633665583,
      "grad_norm": 3.0973849296569824,
      "learning_rate": 9.781263157894738e-05,
      "loss": 0.4508,
      "step": 28326
    },
    {
      "epoch": 1.5862358606786873,
      "grad_norm": 1.5860693454742432,
      "learning_rate": 9.781236842105264e-05,
      "loss": 0.5947,
      "step": 28327
    },
    {
      "epoch": 1.5862918579908163,
      "grad_norm": 1.4285081624984741,
      "learning_rate": 9.781210526315791e-05,
      "loss": 0.5213,
      "step": 28328
    },
    {
      "epoch": 1.5863478553029453,
      "grad_norm": 1.274310827255249,
      "learning_rate": 9.781184210526317e-05,
      "loss": 0.5095,
      "step": 28329
    },
    {
      "epoch": 1.5864038526150743,
      "grad_norm": 1.354880928993225,
      "learning_rate": 9.781157894736841e-05,
      "loss": 0.4487,
      "step": 28330
    },
    {
      "epoch": 1.5864598499272033,
      "grad_norm": 1.4087774753570557,
      "learning_rate": 9.781131578947369e-05,
      "loss": 0.4681,
      "step": 28331
    },
    {
      "epoch": 1.5865158472393324,
      "grad_norm": 1.4856915473937988,
      "learning_rate": 9.781105263157895e-05,
      "loss": 0.4907,
      "step": 28332
    },
    {
      "epoch": 1.5865718445514614,
      "grad_norm": 1.165053129196167,
      "learning_rate": 9.781078947368422e-05,
      "loss": 0.4071,
      "step": 28333
    },
    {
      "epoch": 1.5866278418635904,
      "grad_norm": 1.8275299072265625,
      "learning_rate": 9.781052631578948e-05,
      "loss": 0.7704,
      "step": 28334
    },
    {
      "epoch": 1.5866838391757194,
      "grad_norm": 1.1157145500183105,
      "learning_rate": 9.781026315789474e-05,
      "loss": 0.4555,
      "step": 28335
    },
    {
      "epoch": 1.5867398364878484,
      "grad_norm": 1.1894867420196533,
      "learning_rate": 9.781e-05,
      "loss": 0.4554,
      "step": 28336
    },
    {
      "epoch": 1.5867958337999775,
      "grad_norm": 1.4696792364120483,
      "learning_rate": 9.780973684210527e-05,
      "loss": 0.4572,
      "step": 28337
    },
    {
      "epoch": 1.5868518311121065,
      "grad_norm": 1.2457644939422607,
      "learning_rate": 9.780947368421053e-05,
      "loss": 0.4419,
      "step": 28338
    },
    {
      "epoch": 1.5869078284242355,
      "grad_norm": 1.1074076890945435,
      "learning_rate": 9.780921052631579e-05,
      "loss": 0.5484,
      "step": 28339
    },
    {
      "epoch": 1.5869638257363645,
      "grad_norm": 1.3094606399536133,
      "learning_rate": 9.780894736842105e-05,
      "loss": 0.472,
      "step": 28340
    },
    {
      "epoch": 1.5870198230484935,
      "grad_norm": 1.7311792373657227,
      "learning_rate": 9.780868421052633e-05,
      "loss": 0.549,
      "step": 28341
    },
    {
      "epoch": 1.5870758203606226,
      "grad_norm": 1.266061782836914,
      "learning_rate": 9.780842105263159e-05,
      "loss": 0.4031,
      "step": 28342
    },
    {
      "epoch": 1.5871318176727516,
      "grad_norm": 1.1189930438995361,
      "learning_rate": 9.780815789473685e-05,
      "loss": 0.4839,
      "step": 28343
    },
    {
      "epoch": 1.5871878149848806,
      "grad_norm": 1.2545877695083618,
      "learning_rate": 9.78078947368421e-05,
      "loss": 0.6219,
      "step": 28344
    },
    {
      "epoch": 1.5872438122970096,
      "grad_norm": 1.3213335275650024,
      "learning_rate": 9.780763157894738e-05,
      "loss": 0.3883,
      "step": 28345
    },
    {
      "epoch": 1.5872998096091386,
      "grad_norm": 1.3805620670318604,
      "learning_rate": 9.780736842105264e-05,
      "loss": 0.4589,
      "step": 28346
    },
    {
      "epoch": 1.5873558069212677,
      "grad_norm": 1.2606295347213745,
      "learning_rate": 9.78071052631579e-05,
      "loss": 0.5068,
      "step": 28347
    },
    {
      "epoch": 1.5874118042333967,
      "grad_norm": 1.353511929512024,
      "learning_rate": 9.780684210526316e-05,
      "loss": 0.436,
      "step": 28348
    },
    {
      "epoch": 1.5874678015455257,
      "grad_norm": 1.7375603914260864,
      "learning_rate": 9.780657894736842e-05,
      "loss": 0.5207,
      "step": 28349
    },
    {
      "epoch": 1.5875237988576547,
      "grad_norm": 1.2530337572097778,
      "learning_rate": 9.780631578947369e-05,
      "loss": 0.3879,
      "step": 28350
    },
    {
      "epoch": 1.5875797961697837,
      "grad_norm": 1.5136656761169434,
      "learning_rate": 9.780605263157895e-05,
      "loss": 0.5764,
      "step": 28351
    },
    {
      "epoch": 1.5876357934819127,
      "grad_norm": 1.0596904754638672,
      "learning_rate": 9.780578947368422e-05,
      "loss": 0.3732,
      "step": 28352
    },
    {
      "epoch": 1.5876917907940418,
      "grad_norm": 1.2544870376586914,
      "learning_rate": 9.780552631578947e-05,
      "loss": 0.4187,
      "step": 28353
    },
    {
      "epoch": 1.5877477881061708,
      "grad_norm": 1.3218860626220703,
      "learning_rate": 9.780526315789474e-05,
      "loss": 0.5132,
      "step": 28354
    },
    {
      "epoch": 1.5878037854182998,
      "grad_norm": 1.3503254652023315,
      "learning_rate": 9.7805e-05,
      "loss": 0.4403,
      "step": 28355
    },
    {
      "epoch": 1.5878597827304288,
      "grad_norm": 1.5628440380096436,
      "learning_rate": 9.780473684210528e-05,
      "loss": 0.4707,
      "step": 28356
    },
    {
      "epoch": 1.5879157800425578,
      "grad_norm": 1.1598467826843262,
      "learning_rate": 9.780447368421052e-05,
      "loss": 0.4415,
      "step": 28357
    },
    {
      "epoch": 1.5879717773546869,
      "grad_norm": 1.7052439451217651,
      "learning_rate": 9.78042105263158e-05,
      "loss": 0.5187,
      "step": 28358
    },
    {
      "epoch": 1.5880277746668159,
      "grad_norm": 1.171898603439331,
      "learning_rate": 9.780394736842106e-05,
      "loss": 0.3662,
      "step": 28359
    },
    {
      "epoch": 1.588083771978945,
      "grad_norm": 1.285593032836914,
      "learning_rate": 9.780368421052633e-05,
      "loss": 0.4864,
      "step": 28360
    },
    {
      "epoch": 1.588139769291074,
      "grad_norm": 1.2117118835449219,
      "learning_rate": 9.780342105263159e-05,
      "loss": 0.3878,
      "step": 28361
    },
    {
      "epoch": 1.588195766603203,
      "grad_norm": 1.504900574684143,
      "learning_rate": 9.780315789473685e-05,
      "loss": 0.5126,
      "step": 28362
    },
    {
      "epoch": 1.588251763915332,
      "grad_norm": 1.3133801221847534,
      "learning_rate": 9.780289473684211e-05,
      "loss": 0.4801,
      "step": 28363
    },
    {
      "epoch": 1.588307761227461,
      "grad_norm": 1.1876076459884644,
      "learning_rate": 9.780263157894737e-05,
      "loss": 0.4214,
      "step": 28364
    },
    {
      "epoch": 1.58836375853959,
      "grad_norm": 1.3989534378051758,
      "learning_rate": 9.780236842105264e-05,
      "loss": 0.6245,
      "step": 28365
    },
    {
      "epoch": 1.588419755851719,
      "grad_norm": 1.4644654989242554,
      "learning_rate": 9.78021052631579e-05,
      "loss": 0.5663,
      "step": 28366
    },
    {
      "epoch": 1.588475753163848,
      "grad_norm": 1.4771467447280884,
      "learning_rate": 9.780184210526316e-05,
      "loss": 0.5066,
      "step": 28367
    },
    {
      "epoch": 1.588531750475977,
      "grad_norm": 1.3771470785140991,
      "learning_rate": 9.780157894736842e-05,
      "loss": 0.7344,
      "step": 28368
    },
    {
      "epoch": 1.588587747788106,
      "grad_norm": 1.4508684873580933,
      "learning_rate": 9.78013157894737e-05,
      "loss": 0.3952,
      "step": 28369
    },
    {
      "epoch": 1.588643745100235,
      "grad_norm": 1.3457491397857666,
      "learning_rate": 9.780105263157895e-05,
      "loss": 0.4476,
      "step": 28370
    },
    {
      "epoch": 1.5886997424123641,
      "grad_norm": 1.3605365753173828,
      "learning_rate": 9.780078947368421e-05,
      "loss": 0.556,
      "step": 28371
    },
    {
      "epoch": 1.5887557397244931,
      "grad_norm": 1.2433078289031982,
      "learning_rate": 9.780052631578947e-05,
      "loss": 0.4486,
      "step": 28372
    },
    {
      "epoch": 1.5888117370366222,
      "grad_norm": 1.3366619348526,
      "learning_rate": 9.780026315789475e-05,
      "loss": 0.4229,
      "step": 28373
    },
    {
      "epoch": 1.5888677343487512,
      "grad_norm": 1.091226577758789,
      "learning_rate": 9.78e-05,
      "loss": 0.4386,
      "step": 28374
    },
    {
      "epoch": 1.5889237316608802,
      "grad_norm": 1.1753994226455688,
      "learning_rate": 9.779973684210527e-05,
      "loss": 0.3889,
      "step": 28375
    },
    {
      "epoch": 1.5889797289730092,
      "grad_norm": 1.200480341911316,
      "learning_rate": 9.779947368421052e-05,
      "loss": 0.3553,
      "step": 28376
    },
    {
      "epoch": 1.5890357262851382,
      "grad_norm": 1.3582489490509033,
      "learning_rate": 9.77992105263158e-05,
      "loss": 0.5353,
      "step": 28377
    },
    {
      "epoch": 1.5890917235972672,
      "grad_norm": 1.1027886867523193,
      "learning_rate": 9.779894736842106e-05,
      "loss": 0.3692,
      "step": 28378
    },
    {
      "epoch": 1.5891477209093963,
      "grad_norm": 1.4722157716751099,
      "learning_rate": 9.779868421052633e-05,
      "loss": 0.3914,
      "step": 28379
    },
    {
      "epoch": 1.5892037182215253,
      "grad_norm": 1.501230001449585,
      "learning_rate": 9.779842105263158e-05,
      "loss": 0.594,
      "step": 28380
    },
    {
      "epoch": 1.5892597155336543,
      "grad_norm": 1.2601374387741089,
      "learning_rate": 9.779815789473684e-05,
      "loss": 0.4311,
      "step": 28381
    },
    {
      "epoch": 1.5893157128457833,
      "grad_norm": 1.4137557744979858,
      "learning_rate": 9.779789473684211e-05,
      "loss": 0.4823,
      "step": 28382
    },
    {
      "epoch": 1.5893717101579123,
      "grad_norm": 1.4185372591018677,
      "learning_rate": 9.779763157894737e-05,
      "loss": 0.3598,
      "step": 28383
    },
    {
      "epoch": 1.5894277074700414,
      "grad_norm": 1.58175790309906,
      "learning_rate": 9.779736842105264e-05,
      "loss": 0.5759,
      "step": 28384
    },
    {
      "epoch": 1.5894837047821704,
      "grad_norm": 1.2684824466705322,
      "learning_rate": 9.779710526315789e-05,
      "loss": 0.4728,
      "step": 28385
    },
    {
      "epoch": 1.5895397020942994,
      "grad_norm": 1.3118031024932861,
      "learning_rate": 9.779684210526316e-05,
      "loss": 0.4175,
      "step": 28386
    },
    {
      "epoch": 1.5895956994064284,
      "grad_norm": 1.0539162158966064,
      "learning_rate": 9.779657894736842e-05,
      "loss": 0.2494,
      "step": 28387
    },
    {
      "epoch": 1.5896516967185574,
      "grad_norm": 1.312360167503357,
      "learning_rate": 9.77963157894737e-05,
      "loss": 0.4386,
      "step": 28388
    },
    {
      "epoch": 1.5897076940306865,
      "grad_norm": 1.4711054563522339,
      "learning_rate": 9.779605263157896e-05,
      "loss": 0.3986,
      "step": 28389
    },
    {
      "epoch": 1.5897636913428155,
      "grad_norm": 1.13727605342865,
      "learning_rate": 9.779578947368422e-05,
      "loss": 0.4532,
      "step": 28390
    },
    {
      "epoch": 1.5898196886549445,
      "grad_norm": 1.7851922512054443,
      "learning_rate": 9.779552631578948e-05,
      "loss": 0.5963,
      "step": 28391
    },
    {
      "epoch": 1.5898756859670735,
      "grad_norm": 2.342245101928711,
      "learning_rate": 9.779526315789475e-05,
      "loss": 0.386,
      "step": 28392
    },
    {
      "epoch": 1.5899316832792025,
      "grad_norm": 1.5506318807601929,
      "learning_rate": 9.779500000000001e-05,
      "loss": 0.4646,
      "step": 28393
    },
    {
      "epoch": 1.5899876805913316,
      "grad_norm": 1.348465085029602,
      "learning_rate": 9.779473684210527e-05,
      "loss": 0.4819,
      "step": 28394
    },
    {
      "epoch": 1.5900436779034606,
      "grad_norm": 1.213765263557434,
      "learning_rate": 9.779447368421053e-05,
      "loss": 0.4695,
      "step": 28395
    },
    {
      "epoch": 1.5900996752155896,
      "grad_norm": 1.4216108322143555,
      "learning_rate": 9.77942105263158e-05,
      "loss": 0.6269,
      "step": 28396
    },
    {
      "epoch": 1.5901556725277186,
      "grad_norm": 1.424285650253296,
      "learning_rate": 9.779394736842106e-05,
      "loss": 0.4841,
      "step": 28397
    },
    {
      "epoch": 1.5902116698398476,
      "grad_norm": 1.2293741703033447,
      "learning_rate": 9.779368421052632e-05,
      "loss": 0.4536,
      "step": 28398
    },
    {
      "epoch": 1.5902676671519766,
      "grad_norm": 1.4259518384933472,
      "learning_rate": 9.779342105263158e-05,
      "loss": 0.4597,
      "step": 28399
    },
    {
      "epoch": 1.5903236644641057,
      "grad_norm": 1.1571725606918335,
      "learning_rate": 9.779315789473684e-05,
      "loss": 0.4649,
      "step": 28400
    },
    {
      "epoch": 1.5903796617762347,
      "grad_norm": 1.0336768627166748,
      "learning_rate": 9.779289473684211e-05,
      "loss": 0.4505,
      "step": 28401
    },
    {
      "epoch": 1.5904356590883637,
      "grad_norm": 1.3144545555114746,
      "learning_rate": 9.779263157894737e-05,
      "loss": 0.462,
      "step": 28402
    },
    {
      "epoch": 1.5904916564004927,
      "grad_norm": 1.402504801750183,
      "learning_rate": 9.779236842105263e-05,
      "loss": 0.5251,
      "step": 28403
    },
    {
      "epoch": 1.5905476537126217,
      "grad_norm": 1.2226722240447998,
      "learning_rate": 9.779210526315789e-05,
      "loss": 0.4761,
      "step": 28404
    },
    {
      "epoch": 1.5906036510247508,
      "grad_norm": 1.2393138408660889,
      "learning_rate": 9.779184210526317e-05,
      "loss": 0.3775,
      "step": 28405
    },
    {
      "epoch": 1.5906596483368798,
      "grad_norm": 1.0617609024047852,
      "learning_rate": 9.779157894736843e-05,
      "loss": 0.3888,
      "step": 28406
    },
    {
      "epoch": 1.5907156456490088,
      "grad_norm": 1.089975118637085,
      "learning_rate": 9.77913157894737e-05,
      "loss": 0.3558,
      "step": 28407
    },
    {
      "epoch": 1.5907716429611378,
      "grad_norm": 1.3500378131866455,
      "learning_rate": 9.779105263157894e-05,
      "loss": 0.4809,
      "step": 28408
    },
    {
      "epoch": 1.5908276402732668,
      "grad_norm": 1.1152628660202026,
      "learning_rate": 9.779078947368422e-05,
      "loss": 0.3475,
      "step": 28409
    },
    {
      "epoch": 1.5908836375853959,
      "grad_norm": 1.1033310890197754,
      "learning_rate": 9.779052631578948e-05,
      "loss": 0.3902,
      "step": 28410
    },
    {
      "epoch": 1.5909396348975249,
      "grad_norm": 1.1562005281448364,
      "learning_rate": 9.779026315789475e-05,
      "loss": 0.4295,
      "step": 28411
    },
    {
      "epoch": 1.590995632209654,
      "grad_norm": 1.3591700792312622,
      "learning_rate": 9.779e-05,
      "loss": 0.4184,
      "step": 28412
    },
    {
      "epoch": 1.591051629521783,
      "grad_norm": 1.2122689485549927,
      "learning_rate": 9.778973684210527e-05,
      "loss": 0.4766,
      "step": 28413
    },
    {
      "epoch": 1.591107626833912,
      "grad_norm": 1.1197035312652588,
      "learning_rate": 9.778947368421053e-05,
      "loss": 0.5783,
      "step": 28414
    },
    {
      "epoch": 1.591163624146041,
      "grad_norm": 1.5661566257476807,
      "learning_rate": 9.778921052631579e-05,
      "loss": 0.4492,
      "step": 28415
    },
    {
      "epoch": 1.59121962145817,
      "grad_norm": 1.2168840169906616,
      "learning_rate": 9.778894736842106e-05,
      "loss": 0.4883,
      "step": 28416
    },
    {
      "epoch": 1.591275618770299,
      "grad_norm": 1.9509795904159546,
      "learning_rate": 9.778868421052631e-05,
      "loss": 0.5233,
      "step": 28417
    },
    {
      "epoch": 1.591331616082428,
      "grad_norm": 1.128519058227539,
      "learning_rate": 9.778842105263158e-05,
      "loss": 0.373,
      "step": 28418
    },
    {
      "epoch": 1.591387613394557,
      "grad_norm": 1.1893380880355835,
      "learning_rate": 9.778815789473684e-05,
      "loss": 0.444,
      "step": 28419
    },
    {
      "epoch": 1.591443610706686,
      "grad_norm": 1.22463858127594,
      "learning_rate": 9.778789473684212e-05,
      "loss": 0.4,
      "step": 28420
    },
    {
      "epoch": 1.591499608018815,
      "grad_norm": 1.1782256364822388,
      "learning_rate": 9.778763157894738e-05,
      "loss": 0.3917,
      "step": 28421
    },
    {
      "epoch": 1.591555605330944,
      "grad_norm": 1.3224610090255737,
      "learning_rate": 9.778736842105264e-05,
      "loss": 0.4997,
      "step": 28422
    },
    {
      "epoch": 1.591611602643073,
      "grad_norm": 2.5644896030426025,
      "learning_rate": 9.77871052631579e-05,
      "loss": 0.6133,
      "step": 28423
    },
    {
      "epoch": 1.5916675999552021,
      "grad_norm": 1.3538347482681274,
      "learning_rate": 9.778684210526317e-05,
      "loss": 0.4019,
      "step": 28424
    },
    {
      "epoch": 1.5917235972673311,
      "grad_norm": 1.201648473739624,
      "learning_rate": 9.778657894736843e-05,
      "loss": 0.4126,
      "step": 28425
    },
    {
      "epoch": 1.5917795945794602,
      "grad_norm": 1.3906424045562744,
      "learning_rate": 9.778631578947369e-05,
      "loss": 0.4754,
      "step": 28426
    },
    {
      "epoch": 1.5918355918915892,
      "grad_norm": 1.3985551595687866,
      "learning_rate": 9.778605263157895e-05,
      "loss": 0.4963,
      "step": 28427
    },
    {
      "epoch": 1.5918915892037182,
      "grad_norm": 1.138534426689148,
      "learning_rate": 9.778578947368422e-05,
      "loss": 0.4352,
      "step": 28428
    },
    {
      "epoch": 1.5919475865158472,
      "grad_norm": 1.607399821281433,
      "learning_rate": 9.778552631578948e-05,
      "loss": 0.4495,
      "step": 28429
    },
    {
      "epoch": 1.5920035838279762,
      "grad_norm": 1.2148011922836304,
      "learning_rate": 9.778526315789474e-05,
      "loss": 0.4655,
      "step": 28430
    },
    {
      "epoch": 1.5920595811401053,
      "grad_norm": 1.1976314783096313,
      "learning_rate": 9.7785e-05,
      "loss": 0.4267,
      "step": 28431
    },
    {
      "epoch": 1.5921155784522343,
      "grad_norm": 1.3428587913513184,
      "learning_rate": 9.778473684210526e-05,
      "loss": 0.4297,
      "step": 28432
    },
    {
      "epoch": 1.5921715757643633,
      "grad_norm": 1.3928537368774414,
      "learning_rate": 9.778447368421053e-05,
      "loss": 0.4863,
      "step": 28433
    },
    {
      "epoch": 1.5922275730764923,
      "grad_norm": 1.0944451093673706,
      "learning_rate": 9.778421052631579e-05,
      "loss": 0.3974,
      "step": 28434
    },
    {
      "epoch": 1.5922835703886213,
      "grad_norm": 1.2076185941696167,
      "learning_rate": 9.778394736842105e-05,
      "loss": 0.5395,
      "step": 28435
    },
    {
      "epoch": 1.5923395677007504,
      "grad_norm": 1.1822516918182373,
      "learning_rate": 9.778368421052631e-05,
      "loss": 0.405,
      "step": 28436
    },
    {
      "epoch": 1.5923955650128794,
      "grad_norm": 1.5731171369552612,
      "learning_rate": 9.778342105263159e-05,
      "loss": 0.3564,
      "step": 28437
    },
    {
      "epoch": 1.5924515623250084,
      "grad_norm": 1.2677079439163208,
      "learning_rate": 9.778315789473684e-05,
      "loss": 0.4636,
      "step": 28438
    },
    {
      "epoch": 1.5925075596371374,
      "grad_norm": 1.1874058246612549,
      "learning_rate": 9.778289473684212e-05,
      "loss": 0.3787,
      "step": 28439
    },
    {
      "epoch": 1.5925635569492664,
      "grad_norm": 1.4953478574752808,
      "learning_rate": 9.778263157894736e-05,
      "loss": 0.4372,
      "step": 28440
    },
    {
      "epoch": 1.5926195542613955,
      "grad_norm": 1.2469935417175293,
      "learning_rate": 9.778236842105264e-05,
      "loss": 0.597,
      "step": 28441
    },
    {
      "epoch": 1.5926755515735245,
      "grad_norm": 1.3644754886627197,
      "learning_rate": 9.77821052631579e-05,
      "loss": 0.3941,
      "step": 28442
    },
    {
      "epoch": 1.5927315488856535,
      "grad_norm": 1.2577742338180542,
      "learning_rate": 9.778184210526317e-05,
      "loss": 0.4344,
      "step": 28443
    },
    {
      "epoch": 1.5927875461977825,
      "grad_norm": 1.7004826068878174,
      "learning_rate": 9.778157894736843e-05,
      "loss": 0.6271,
      "step": 28444
    },
    {
      "epoch": 1.5928435435099115,
      "grad_norm": 1.211025357246399,
      "learning_rate": 9.778131578947369e-05,
      "loss": 0.3728,
      "step": 28445
    },
    {
      "epoch": 1.5928995408220405,
      "grad_norm": 1.2380868196487427,
      "learning_rate": 9.778105263157895e-05,
      "loss": 0.5776,
      "step": 28446
    },
    {
      "epoch": 1.5929555381341696,
      "grad_norm": 2.2672958374023438,
      "learning_rate": 9.778078947368422e-05,
      "loss": 0.4709,
      "step": 28447
    },
    {
      "epoch": 1.5930115354462986,
      "grad_norm": 1.4772093296051025,
      "learning_rate": 9.778052631578948e-05,
      "loss": 0.4569,
      "step": 28448
    },
    {
      "epoch": 1.5930675327584276,
      "grad_norm": 1.0881645679473877,
      "learning_rate": 9.778026315789474e-05,
      "loss": 0.3936,
      "step": 28449
    },
    {
      "epoch": 1.5931235300705566,
      "grad_norm": 1.9581612348556519,
      "learning_rate": 9.778e-05,
      "loss": 0.4379,
      "step": 28450
    },
    {
      "epoch": 1.5931795273826856,
      "grad_norm": 1.2579345703125,
      "learning_rate": 9.777973684210526e-05,
      "loss": 0.4634,
      "step": 28451
    },
    {
      "epoch": 1.5932355246948147,
      "grad_norm": 1.2960530519485474,
      "learning_rate": 9.777947368421054e-05,
      "loss": 0.4623,
      "step": 28452
    },
    {
      "epoch": 1.5932915220069437,
      "grad_norm": 1.7639278173446655,
      "learning_rate": 9.77792105263158e-05,
      "loss": 0.3915,
      "step": 28453
    },
    {
      "epoch": 1.5933475193190727,
      "grad_norm": 1.5790356397628784,
      "learning_rate": 9.777894736842105e-05,
      "loss": 0.4792,
      "step": 28454
    },
    {
      "epoch": 1.5934035166312017,
      "grad_norm": 1.4003068208694458,
      "learning_rate": 9.777868421052631e-05,
      "loss": 0.4453,
      "step": 28455
    },
    {
      "epoch": 1.5934595139433307,
      "grad_norm": 1.4069652557373047,
      "learning_rate": 9.777842105263159e-05,
      "loss": 0.4734,
      "step": 28456
    },
    {
      "epoch": 1.5935155112554598,
      "grad_norm": 1.4430841207504272,
      "learning_rate": 9.777815789473685e-05,
      "loss": 0.4123,
      "step": 28457
    },
    {
      "epoch": 1.5935715085675888,
      "grad_norm": 1.223680019378662,
      "learning_rate": 9.777789473684211e-05,
      "loss": 0.3639,
      "step": 28458
    },
    {
      "epoch": 1.5936275058797178,
      "grad_norm": 1.3366825580596924,
      "learning_rate": 9.777763157894737e-05,
      "loss": 0.3767,
      "step": 28459
    },
    {
      "epoch": 1.5936835031918468,
      "grad_norm": 1.191957712173462,
      "learning_rate": 9.777736842105264e-05,
      "loss": 0.4411,
      "step": 28460
    },
    {
      "epoch": 1.5937395005039758,
      "grad_norm": 1.339077115058899,
      "learning_rate": 9.77771052631579e-05,
      "loss": 0.521,
      "step": 28461
    },
    {
      "epoch": 1.5937954978161049,
      "grad_norm": 1.3121421337127686,
      "learning_rate": 9.777684210526317e-05,
      "loss": 0.5048,
      "step": 28462
    },
    {
      "epoch": 1.5938514951282339,
      "grad_norm": 1.209811806678772,
      "learning_rate": 9.777657894736842e-05,
      "loss": 0.416,
      "step": 28463
    },
    {
      "epoch": 1.593907492440363,
      "grad_norm": 1.111407995223999,
      "learning_rate": 9.777631578947369e-05,
      "loss": 0.4355,
      "step": 28464
    },
    {
      "epoch": 1.593963489752492,
      "grad_norm": 1.4819329977035522,
      "learning_rate": 9.777605263157895e-05,
      "loss": 0.4893,
      "step": 28465
    },
    {
      "epoch": 1.594019487064621,
      "grad_norm": 1.4028327465057373,
      "learning_rate": 9.777578947368423e-05,
      "loss": 0.417,
      "step": 28466
    },
    {
      "epoch": 1.59407548437675,
      "grad_norm": 1.1550005674362183,
      "learning_rate": 9.777552631578947e-05,
      "loss": 0.4364,
      "step": 28467
    },
    {
      "epoch": 1.594131481688879,
      "grad_norm": 1.216228723526001,
      "learning_rate": 9.777526315789473e-05,
      "loss": 0.2954,
      "step": 28468
    },
    {
      "epoch": 1.594187479001008,
      "grad_norm": 1.1248055696487427,
      "learning_rate": 9.7775e-05,
      "loss": 0.3729,
      "step": 28469
    },
    {
      "epoch": 1.594243476313137,
      "grad_norm": 1.0551655292510986,
      "learning_rate": 9.777473684210526e-05,
      "loss": 0.3796,
      "step": 28470
    },
    {
      "epoch": 1.594299473625266,
      "grad_norm": 1.4725807905197144,
      "learning_rate": 9.777447368421054e-05,
      "loss": 0.5342,
      "step": 28471
    },
    {
      "epoch": 1.594355470937395,
      "grad_norm": 1.3018877506256104,
      "learning_rate": 9.777421052631578e-05,
      "loss": 0.5014,
      "step": 28472
    },
    {
      "epoch": 1.594411468249524,
      "grad_norm": 1.571959376335144,
      "learning_rate": 9.777394736842106e-05,
      "loss": 0.5121,
      "step": 28473
    },
    {
      "epoch": 1.594467465561653,
      "grad_norm": 1.4337856769561768,
      "learning_rate": 9.777368421052632e-05,
      "loss": 0.4807,
      "step": 28474
    },
    {
      "epoch": 1.594523462873782,
      "grad_norm": 1.2302309274673462,
      "learning_rate": 9.777342105263159e-05,
      "loss": 0.4135,
      "step": 28475
    },
    {
      "epoch": 1.5945794601859111,
      "grad_norm": 1.445213794708252,
      "learning_rate": 9.777315789473685e-05,
      "loss": 0.5878,
      "step": 28476
    },
    {
      "epoch": 1.5946354574980401,
      "grad_norm": 1.50583016872406,
      "learning_rate": 9.777289473684211e-05,
      "loss": 0.4031,
      "step": 28477
    },
    {
      "epoch": 1.5946914548101692,
      "grad_norm": 1.5212727785110474,
      "learning_rate": 9.777263157894737e-05,
      "loss": 0.4732,
      "step": 28478
    },
    {
      "epoch": 1.5947474521222982,
      "grad_norm": 1.2776482105255127,
      "learning_rate": 9.777236842105264e-05,
      "loss": 0.4414,
      "step": 28479
    },
    {
      "epoch": 1.5948034494344272,
      "grad_norm": 1.2676594257354736,
      "learning_rate": 9.77721052631579e-05,
      "loss": 0.5139,
      "step": 28480
    },
    {
      "epoch": 1.5948594467465562,
      "grad_norm": 1.4428002834320068,
      "learning_rate": 9.777184210526316e-05,
      "loss": 0.4788,
      "step": 28481
    },
    {
      "epoch": 1.5949154440586852,
      "grad_norm": 1.3846535682678223,
      "learning_rate": 9.777157894736842e-05,
      "loss": 0.5191,
      "step": 28482
    },
    {
      "epoch": 1.5949714413708143,
      "grad_norm": 1.5444915294647217,
      "learning_rate": 9.77713157894737e-05,
      "loss": 0.5192,
      "step": 28483
    },
    {
      "epoch": 1.5950274386829433,
      "grad_norm": 1.2789089679718018,
      "learning_rate": 9.777105263157896e-05,
      "loss": 0.4586,
      "step": 28484
    },
    {
      "epoch": 1.5950834359950723,
      "grad_norm": 1.6008487939834595,
      "learning_rate": 9.777078947368421e-05,
      "loss": 0.5281,
      "step": 28485
    },
    {
      "epoch": 1.5951394333072013,
      "grad_norm": 1.1789329051971436,
      "learning_rate": 9.777052631578947e-05,
      "loss": 0.4533,
      "step": 28486
    },
    {
      "epoch": 1.5951954306193303,
      "grad_norm": 1.3337887525558472,
      "learning_rate": 9.777026315789473e-05,
      "loss": 0.4568,
      "step": 28487
    },
    {
      "epoch": 1.5952514279314594,
      "grad_norm": 1.1670817136764526,
      "learning_rate": 9.777000000000001e-05,
      "loss": 0.4196,
      "step": 28488
    },
    {
      "epoch": 1.5953074252435884,
      "grad_norm": 1.2730642557144165,
      "learning_rate": 9.776973684210527e-05,
      "loss": 0.3497,
      "step": 28489
    },
    {
      "epoch": 1.5953634225557174,
      "grad_norm": 1.3642950057983398,
      "learning_rate": 9.776947368421053e-05,
      "loss": 0.4985,
      "step": 28490
    },
    {
      "epoch": 1.5954194198678464,
      "grad_norm": 1.2098121643066406,
      "learning_rate": 9.776921052631579e-05,
      "loss": 0.6366,
      "step": 28491
    },
    {
      "epoch": 1.5954754171799754,
      "grad_norm": 1.8024334907531738,
      "learning_rate": 9.776894736842106e-05,
      "loss": 0.4126,
      "step": 28492
    },
    {
      "epoch": 1.5955314144921044,
      "grad_norm": 1.2786428928375244,
      "learning_rate": 9.776868421052632e-05,
      "loss": 0.5756,
      "step": 28493
    },
    {
      "epoch": 1.5955874118042335,
      "grad_norm": 4.292102813720703,
      "learning_rate": 9.776842105263159e-05,
      "loss": 0.548,
      "step": 28494
    },
    {
      "epoch": 1.5956434091163625,
      "grad_norm": 1.196244239807129,
      "learning_rate": 9.776815789473684e-05,
      "loss": 0.3955,
      "step": 28495
    },
    {
      "epoch": 1.5956994064284915,
      "grad_norm": 1.4397553205490112,
      "learning_rate": 9.776789473684211e-05,
      "loss": 0.5208,
      "step": 28496
    },
    {
      "epoch": 1.5957554037406205,
      "grad_norm": 1.1940295696258545,
      "learning_rate": 9.776763157894737e-05,
      "loss": 0.3456,
      "step": 28497
    },
    {
      "epoch": 1.5958114010527495,
      "grad_norm": 1.2804096937179565,
      "learning_rate": 9.776736842105265e-05,
      "loss": 0.4411,
      "step": 28498
    },
    {
      "epoch": 1.5958673983648786,
      "grad_norm": 2.038516044616699,
      "learning_rate": 9.77671052631579e-05,
      "loss": 0.3775,
      "step": 28499
    },
    {
      "epoch": 1.5959233956770076,
      "grad_norm": 1.3518465757369995,
      "learning_rate": 9.776684210526316e-05,
      "loss": 0.4528,
      "step": 28500
    },
    {
      "epoch": 1.5959793929891366,
      "grad_norm": 1.3068658113479614,
      "learning_rate": 9.776657894736842e-05,
      "loss": 0.4894,
      "step": 28501
    },
    {
      "epoch": 1.5960353903012656,
      "grad_norm": 1.3878062963485718,
      "learning_rate": 9.776631578947368e-05,
      "loss": 0.4538,
      "step": 28502
    },
    {
      "epoch": 1.5960913876133946,
      "grad_norm": 1.7153552770614624,
      "learning_rate": 9.776605263157896e-05,
      "loss": 0.5061,
      "step": 28503
    },
    {
      "epoch": 1.5961473849255237,
      "grad_norm": 1.4197717905044556,
      "learning_rate": 9.776578947368422e-05,
      "loss": 0.5573,
      "step": 28504
    },
    {
      "epoch": 1.5962033822376527,
      "grad_norm": 1.4265944957733154,
      "learning_rate": 9.776552631578948e-05,
      "loss": 0.5851,
      "step": 28505
    },
    {
      "epoch": 1.5962593795497817,
      "grad_norm": 1.3374272584915161,
      "learning_rate": 9.776526315789474e-05,
      "loss": 0.334,
      "step": 28506
    },
    {
      "epoch": 1.5963153768619107,
      "grad_norm": 1.7477167844772339,
      "learning_rate": 9.776500000000001e-05,
      "loss": 0.6123,
      "step": 28507
    },
    {
      "epoch": 1.5963713741740397,
      "grad_norm": 1.1946483850479126,
      "learning_rate": 9.776473684210527e-05,
      "loss": 0.4356,
      "step": 28508
    },
    {
      "epoch": 1.5964273714861688,
      "grad_norm": 1.7405896186828613,
      "learning_rate": 9.776447368421053e-05,
      "loss": 0.4198,
      "step": 28509
    },
    {
      "epoch": 1.5964833687982978,
      "grad_norm": 1.3562170267105103,
      "learning_rate": 9.776421052631579e-05,
      "loss": 0.3167,
      "step": 28510
    },
    {
      "epoch": 1.5965393661104268,
      "grad_norm": 1.243408203125,
      "learning_rate": 9.776394736842106e-05,
      "loss": 0.5274,
      "step": 28511
    },
    {
      "epoch": 1.5965953634225558,
      "grad_norm": 2.603712320327759,
      "learning_rate": 9.776368421052632e-05,
      "loss": 0.4354,
      "step": 28512
    },
    {
      "epoch": 1.5966513607346848,
      "grad_norm": 1.4916447401046753,
      "learning_rate": 9.776342105263158e-05,
      "loss": 0.4733,
      "step": 28513
    },
    {
      "epoch": 1.5967073580468139,
      "grad_norm": 1.0743104219436646,
      "learning_rate": 9.776315789473684e-05,
      "loss": 0.3415,
      "step": 28514
    },
    {
      "epoch": 1.5967633553589429,
      "grad_norm": 1.333001971244812,
      "learning_rate": 9.776289473684212e-05,
      "loss": 0.6249,
      "step": 28515
    },
    {
      "epoch": 1.596819352671072,
      "grad_norm": 1.2960251569747925,
      "learning_rate": 9.776263157894737e-05,
      "loss": 0.4857,
      "step": 28516
    },
    {
      "epoch": 1.596875349983201,
      "grad_norm": 1.4187594652175903,
      "learning_rate": 9.776236842105265e-05,
      "loss": 0.5072,
      "step": 28517
    },
    {
      "epoch": 1.59693134729533,
      "grad_norm": 0.9220099449157715,
      "learning_rate": 9.77621052631579e-05,
      "loss": 0.3494,
      "step": 28518
    },
    {
      "epoch": 1.596987344607459,
      "grad_norm": 1.3434042930603027,
      "learning_rate": 9.776184210526315e-05,
      "loss": 0.4417,
      "step": 28519
    },
    {
      "epoch": 1.597043341919588,
      "grad_norm": 1.4903477430343628,
      "learning_rate": 9.776157894736843e-05,
      "loss": 0.5027,
      "step": 28520
    },
    {
      "epoch": 1.597099339231717,
      "grad_norm": 1.5842574834823608,
      "learning_rate": 9.776131578947369e-05,
      "loss": 0.688,
      "step": 28521
    },
    {
      "epoch": 1.597155336543846,
      "grad_norm": 1.6226140260696411,
      "learning_rate": 9.776105263157895e-05,
      "loss": 0.4676,
      "step": 28522
    },
    {
      "epoch": 1.597211333855975,
      "grad_norm": 1.1888446807861328,
      "learning_rate": 9.77607894736842e-05,
      "loss": 0.4191,
      "step": 28523
    },
    {
      "epoch": 1.597267331168104,
      "grad_norm": 1.2289389371871948,
      "learning_rate": 9.776052631578948e-05,
      "loss": 0.4819,
      "step": 28524
    },
    {
      "epoch": 1.597323328480233,
      "grad_norm": 1.3530793190002441,
      "learning_rate": 9.776026315789474e-05,
      "loss": 0.5033,
      "step": 28525
    },
    {
      "epoch": 1.597379325792362,
      "grad_norm": 1.10776948928833,
      "learning_rate": 9.776000000000001e-05,
      "loss": 0.3494,
      "step": 28526
    },
    {
      "epoch": 1.597435323104491,
      "grad_norm": 1.1916927099227905,
      "learning_rate": 9.775973684210526e-05,
      "loss": 0.4193,
      "step": 28527
    },
    {
      "epoch": 1.5974913204166201,
      "grad_norm": 1.2298662662506104,
      "learning_rate": 9.775947368421053e-05,
      "loss": 0.418,
      "step": 28528
    },
    {
      "epoch": 1.5975473177287491,
      "grad_norm": 1.2551629543304443,
      "learning_rate": 9.775921052631579e-05,
      "loss": 0.3814,
      "step": 28529
    },
    {
      "epoch": 1.5976033150408782,
      "grad_norm": 1.444276213645935,
      "learning_rate": 9.775894736842107e-05,
      "loss": 0.5472,
      "step": 28530
    },
    {
      "epoch": 1.5976593123530072,
      "grad_norm": 1.3685768842697144,
      "learning_rate": 9.775868421052632e-05,
      "loss": 0.6566,
      "step": 28531
    },
    {
      "epoch": 1.5977153096651362,
      "grad_norm": 1.209527850151062,
      "learning_rate": 9.775842105263158e-05,
      "loss": 0.6138,
      "step": 28532
    },
    {
      "epoch": 1.5977713069772652,
      "grad_norm": 1.4719892740249634,
      "learning_rate": 9.775815789473684e-05,
      "loss": 0.453,
      "step": 28533
    },
    {
      "epoch": 1.5978273042893942,
      "grad_norm": 1.3141635656356812,
      "learning_rate": 9.775789473684212e-05,
      "loss": 0.4432,
      "step": 28534
    },
    {
      "epoch": 1.5978833016015233,
      "grad_norm": 1.5745213031768799,
      "learning_rate": 9.775763157894738e-05,
      "loss": 0.5168,
      "step": 28535
    },
    {
      "epoch": 1.5979392989136523,
      "grad_norm": 1.1044379472732544,
      "learning_rate": 9.775736842105264e-05,
      "loss": 0.4065,
      "step": 28536
    },
    {
      "epoch": 1.5979952962257813,
      "grad_norm": 1.2080252170562744,
      "learning_rate": 9.77571052631579e-05,
      "loss": 0.4802,
      "step": 28537
    },
    {
      "epoch": 1.5980512935379103,
      "grad_norm": 1.4356034994125366,
      "learning_rate": 9.775684210526316e-05,
      "loss": 0.383,
      "step": 28538
    },
    {
      "epoch": 1.5981072908500393,
      "grad_norm": 1.205481767654419,
      "learning_rate": 9.775657894736843e-05,
      "loss": 0.3951,
      "step": 28539
    },
    {
      "epoch": 1.5981632881621683,
      "grad_norm": 2.1522939205169678,
      "learning_rate": 9.775631578947369e-05,
      "loss": 0.5754,
      "step": 28540
    },
    {
      "epoch": 1.5982192854742974,
      "grad_norm": 1.5407639741897583,
      "learning_rate": 9.775605263157895e-05,
      "loss": 0.5231,
      "step": 28541
    },
    {
      "epoch": 1.5982752827864264,
      "grad_norm": 1.5534600019454956,
      "learning_rate": 9.775578947368421e-05,
      "loss": 0.4517,
      "step": 28542
    },
    {
      "epoch": 1.5983312800985554,
      "grad_norm": 1.257481575012207,
      "learning_rate": 9.775552631578948e-05,
      "loss": 0.5159,
      "step": 28543
    },
    {
      "epoch": 1.5983872774106844,
      "grad_norm": 1.5003831386566162,
      "learning_rate": 9.775526315789474e-05,
      "loss": 0.4466,
      "step": 28544
    },
    {
      "epoch": 1.5984432747228134,
      "grad_norm": 1.46065354347229,
      "learning_rate": 9.7755e-05,
      "loss": 0.517,
      "step": 28545
    },
    {
      "epoch": 1.5984992720349425,
      "grad_norm": 1.3138266801834106,
      "learning_rate": 9.775473684210526e-05,
      "loss": 0.4196,
      "step": 28546
    },
    {
      "epoch": 1.5985552693470715,
      "grad_norm": 8.69722843170166,
      "learning_rate": 9.775447368421053e-05,
      "loss": 0.4998,
      "step": 28547
    },
    {
      "epoch": 1.5986112666592005,
      "grad_norm": 1.264536738395691,
      "learning_rate": 9.77542105263158e-05,
      "loss": 0.4926,
      "step": 28548
    },
    {
      "epoch": 1.5986672639713295,
      "grad_norm": 1.1531795263290405,
      "learning_rate": 9.775394736842107e-05,
      "loss": 0.4186,
      "step": 28549
    },
    {
      "epoch": 1.5987232612834585,
      "grad_norm": 1.7917656898498535,
      "learning_rate": 9.775368421052631e-05,
      "loss": 0.4706,
      "step": 28550
    },
    {
      "epoch": 1.5987792585955876,
      "grad_norm": 1.5780102014541626,
      "learning_rate": 9.775342105263159e-05,
      "loss": 0.4027,
      "step": 28551
    },
    {
      "epoch": 1.5988352559077166,
      "grad_norm": 1.246495246887207,
      "learning_rate": 9.775315789473685e-05,
      "loss": 0.5135,
      "step": 28552
    },
    {
      "epoch": 1.5988912532198456,
      "grad_norm": 1.2440063953399658,
      "learning_rate": 9.77528947368421e-05,
      "loss": 0.4332,
      "step": 28553
    },
    {
      "epoch": 1.5989472505319746,
      "grad_norm": 1.3712800741195679,
      "learning_rate": 9.775263157894738e-05,
      "loss": 0.4079,
      "step": 28554
    },
    {
      "epoch": 1.5990032478441036,
      "grad_norm": 1.4920446872711182,
      "learning_rate": 9.775236842105263e-05,
      "loss": 0.3994,
      "step": 28555
    },
    {
      "epoch": 1.5990592451562327,
      "grad_norm": 1.6190557479858398,
      "learning_rate": 9.77521052631579e-05,
      "loss": 0.4627,
      "step": 28556
    },
    {
      "epoch": 1.5991152424683617,
      "grad_norm": 1.5346342325210571,
      "learning_rate": 9.775184210526316e-05,
      "loss": 0.4299,
      "step": 28557
    },
    {
      "epoch": 1.5991712397804907,
      "grad_norm": 1.3129130601882935,
      "learning_rate": 9.775157894736843e-05,
      "loss": 0.5504,
      "step": 28558
    },
    {
      "epoch": 1.5992272370926197,
      "grad_norm": 1.2687749862670898,
      "learning_rate": 9.775131578947368e-05,
      "loss": 0.5004,
      "step": 28559
    },
    {
      "epoch": 1.5992832344047487,
      "grad_norm": 1.371648907661438,
      "learning_rate": 9.775105263157895e-05,
      "loss": 0.4729,
      "step": 28560
    },
    {
      "epoch": 1.5993392317168778,
      "grad_norm": 1.2943694591522217,
      "learning_rate": 9.775078947368421e-05,
      "loss": 0.6238,
      "step": 28561
    },
    {
      "epoch": 1.5993952290290068,
      "grad_norm": 1.118265151977539,
      "learning_rate": 9.775052631578948e-05,
      "loss": 0.3792,
      "step": 28562
    },
    {
      "epoch": 1.5994512263411358,
      "grad_norm": 1.4891771078109741,
      "learning_rate": 9.775026315789474e-05,
      "loss": 0.4695,
      "step": 28563
    },
    {
      "epoch": 1.5995072236532648,
      "grad_norm": 1.7308967113494873,
      "learning_rate": 9.775e-05,
      "loss": 0.4526,
      "step": 28564
    },
    {
      "epoch": 1.5995632209653936,
      "grad_norm": 1.083701252937317,
      "learning_rate": 9.774973684210526e-05,
      "loss": 0.3322,
      "step": 28565
    },
    {
      "epoch": 1.5996192182775226,
      "grad_norm": 1.39650559425354,
      "learning_rate": 9.774947368421054e-05,
      "loss": 0.3853,
      "step": 28566
    },
    {
      "epoch": 1.5996752155896516,
      "grad_norm": 1.2509013414382935,
      "learning_rate": 9.77492105263158e-05,
      "loss": 0.4567,
      "step": 28567
    },
    {
      "epoch": 1.5997312129017807,
      "grad_norm": 1.6539342403411865,
      "learning_rate": 9.774894736842106e-05,
      "loss": 0.5734,
      "step": 28568
    },
    {
      "epoch": 1.5997872102139097,
      "grad_norm": 1.1319507360458374,
      "learning_rate": 9.774868421052632e-05,
      "loss": 0.3607,
      "step": 28569
    },
    {
      "epoch": 1.5998432075260387,
      "grad_norm": 1.6434319019317627,
      "learning_rate": 9.774842105263159e-05,
      "loss": 0.4695,
      "step": 28570
    },
    {
      "epoch": 1.5998992048381677,
      "grad_norm": 1.5648937225341797,
      "learning_rate": 9.774815789473685e-05,
      "loss": 0.6171,
      "step": 28571
    },
    {
      "epoch": 1.5999552021502967,
      "grad_norm": 1.357996940612793,
      "learning_rate": 9.774789473684211e-05,
      "loss": 0.3983,
      "step": 28572
    },
    {
      "epoch": 1.6000111994624258,
      "grad_norm": 1.4389029741287231,
      "learning_rate": 9.774763157894737e-05,
      "loss": 0.4872,
      "step": 28573
    },
    {
      "epoch": 1.6000671967745548,
      "grad_norm": 1.4677770137786865,
      "learning_rate": 9.774736842105263e-05,
      "loss": 0.526,
      "step": 28574
    },
    {
      "epoch": 1.6001231940866838,
      "grad_norm": 1.207441806793213,
      "learning_rate": 9.77471052631579e-05,
      "loss": 0.5071,
      "step": 28575
    },
    {
      "epoch": 1.6001791913988128,
      "grad_norm": 1.2272001504898071,
      "learning_rate": 9.774684210526316e-05,
      "loss": 0.5347,
      "step": 28576
    },
    {
      "epoch": 1.6002351887109418,
      "grad_norm": 1.299938440322876,
      "learning_rate": 9.774657894736842e-05,
      "loss": 0.4146,
      "step": 28577
    },
    {
      "epoch": 1.6002911860230709,
      "grad_norm": 1.4660892486572266,
      "learning_rate": 9.774631578947368e-05,
      "loss": 0.5567,
      "step": 28578
    },
    {
      "epoch": 1.6003471833351999,
      "grad_norm": 1.6837488412857056,
      "learning_rate": 9.774605263157895e-05,
      "loss": 0.5576,
      "step": 28579
    },
    {
      "epoch": 1.600403180647329,
      "grad_norm": 1.3120943307876587,
      "learning_rate": 9.774578947368421e-05,
      "loss": 0.4611,
      "step": 28580
    },
    {
      "epoch": 1.600459177959458,
      "grad_norm": 1.2361335754394531,
      "learning_rate": 9.774552631578949e-05,
      "loss": 0.4284,
      "step": 28581
    },
    {
      "epoch": 1.600515175271587,
      "grad_norm": 1.3941798210144043,
      "learning_rate": 9.774526315789473e-05,
      "loss": 0.629,
      "step": 28582
    },
    {
      "epoch": 1.600571172583716,
      "grad_norm": 1.0770739316940308,
      "learning_rate": 9.774500000000001e-05,
      "loss": 0.479,
      "step": 28583
    },
    {
      "epoch": 1.600627169895845,
      "grad_norm": 1.4396823644638062,
      "learning_rate": 9.774473684210527e-05,
      "loss": 0.5685,
      "step": 28584
    },
    {
      "epoch": 1.600683167207974,
      "grad_norm": 1.361149549484253,
      "learning_rate": 9.774447368421054e-05,
      "loss": 0.4855,
      "step": 28585
    },
    {
      "epoch": 1.600739164520103,
      "grad_norm": 1.5855399370193481,
      "learning_rate": 9.77442105263158e-05,
      "loss": 0.5503,
      "step": 28586
    },
    {
      "epoch": 1.600795161832232,
      "grad_norm": 1.39113450050354,
      "learning_rate": 9.774394736842106e-05,
      "loss": 0.5595,
      "step": 28587
    },
    {
      "epoch": 1.600851159144361,
      "grad_norm": 1.7800246477127075,
      "learning_rate": 9.774368421052632e-05,
      "loss": 0.4223,
      "step": 28588
    },
    {
      "epoch": 1.60090715645649,
      "grad_norm": 1.1487984657287598,
      "learning_rate": 9.774342105263158e-05,
      "loss": 0.4232,
      "step": 28589
    },
    {
      "epoch": 1.600963153768619,
      "grad_norm": 2.425628900527954,
      "learning_rate": 9.774315789473685e-05,
      "loss": 0.5678,
      "step": 28590
    },
    {
      "epoch": 1.601019151080748,
      "grad_norm": 1.4537615776062012,
      "learning_rate": 9.774289473684211e-05,
      "loss": 0.5565,
      "step": 28591
    },
    {
      "epoch": 1.6010751483928771,
      "grad_norm": 1.319662094116211,
      "learning_rate": 9.774263157894737e-05,
      "loss": 0.4901,
      "step": 28592
    },
    {
      "epoch": 1.6011311457050061,
      "grad_norm": 1.9893040657043457,
      "learning_rate": 9.774236842105263e-05,
      "loss": 0.5555,
      "step": 28593
    },
    {
      "epoch": 1.6011871430171352,
      "grad_norm": 1.6523683071136475,
      "learning_rate": 9.77421052631579e-05,
      "loss": 0.4821,
      "step": 28594
    },
    {
      "epoch": 1.6012431403292642,
      "grad_norm": 1.3857654333114624,
      "learning_rate": 9.774184210526316e-05,
      "loss": 0.8286,
      "step": 28595
    },
    {
      "epoch": 1.6012991376413932,
      "grad_norm": 1.2868489027023315,
      "learning_rate": 9.774157894736842e-05,
      "loss": 0.4676,
      "step": 28596
    },
    {
      "epoch": 1.6013551349535222,
      "grad_norm": 1.4066622257232666,
      "learning_rate": 9.774131578947368e-05,
      "loss": 0.426,
      "step": 28597
    },
    {
      "epoch": 1.6014111322656512,
      "grad_norm": 1.2741644382476807,
      "learning_rate": 9.774105263157896e-05,
      "loss": 0.3972,
      "step": 28598
    },
    {
      "epoch": 1.6014671295777803,
      "grad_norm": 1.590599536895752,
      "learning_rate": 9.774078947368422e-05,
      "loss": 0.5083,
      "step": 28599
    },
    {
      "epoch": 1.6015231268899093,
      "grad_norm": 2.7208175659179688,
      "learning_rate": 9.774052631578948e-05,
      "loss": 0.6283,
      "step": 28600
    },
    {
      "epoch": 1.6015791242020383,
      "grad_norm": 1.4425791501998901,
      "learning_rate": 9.774026315789474e-05,
      "loss": 0.4716,
      "step": 28601
    },
    {
      "epoch": 1.6016351215141673,
      "grad_norm": 1.4351321458816528,
      "learning_rate": 9.774000000000001e-05,
      "loss": 0.445,
      "step": 28602
    },
    {
      "epoch": 1.6016911188262963,
      "grad_norm": 1.2157111167907715,
      "learning_rate": 9.773973684210527e-05,
      "loss": 0.3545,
      "step": 28603
    },
    {
      "epoch": 1.6017471161384254,
      "grad_norm": 1.4584052562713623,
      "learning_rate": 9.773947368421054e-05,
      "loss": 0.5592,
      "step": 28604
    },
    {
      "epoch": 1.6018031134505544,
      "grad_norm": 1.518430233001709,
      "learning_rate": 9.773921052631579e-05,
      "loss": 0.5112,
      "step": 28605
    },
    {
      "epoch": 1.6018591107626834,
      "grad_norm": 1.3541641235351562,
      "learning_rate": 9.773894736842105e-05,
      "loss": 0.4307,
      "step": 28606
    },
    {
      "epoch": 1.6019151080748124,
      "grad_norm": 1.5336047410964966,
      "learning_rate": 9.773868421052632e-05,
      "loss": 0.529,
      "step": 28607
    },
    {
      "epoch": 1.6019711053869414,
      "grad_norm": 1.3114652633666992,
      "learning_rate": 9.773842105263158e-05,
      "loss": 0.4296,
      "step": 28608
    },
    {
      "epoch": 1.6020271026990704,
      "grad_norm": 1.2435579299926758,
      "learning_rate": 9.773815789473685e-05,
      "loss": 0.435,
      "step": 28609
    },
    {
      "epoch": 1.6020831000111995,
      "grad_norm": 1.3253560066223145,
      "learning_rate": 9.77378947368421e-05,
      "loss": 0.4692,
      "step": 28610
    },
    {
      "epoch": 1.6021390973233285,
      "grad_norm": 1.359158992767334,
      "learning_rate": 9.773763157894737e-05,
      "loss": 0.4447,
      "step": 28611
    },
    {
      "epoch": 1.6021950946354575,
      "grad_norm": 1.135666012763977,
      "learning_rate": 9.773736842105263e-05,
      "loss": 0.4251,
      "step": 28612
    },
    {
      "epoch": 1.6022510919475865,
      "grad_norm": 1.4796688556671143,
      "learning_rate": 9.773710526315791e-05,
      "loss": 0.6012,
      "step": 28613
    },
    {
      "epoch": 1.6023070892597155,
      "grad_norm": 1.2790898084640503,
      "learning_rate": 9.773684210526315e-05,
      "loss": 0.5324,
      "step": 28614
    },
    {
      "epoch": 1.6023630865718446,
      "grad_norm": 1.2509945631027222,
      "learning_rate": 9.773657894736843e-05,
      "loss": 0.5256,
      "step": 28615
    },
    {
      "epoch": 1.6024190838839736,
      "grad_norm": 1.4188028573989868,
      "learning_rate": 9.773631578947369e-05,
      "loss": 0.5471,
      "step": 28616
    },
    {
      "epoch": 1.6024750811961026,
      "grad_norm": 1.5398504734039307,
      "learning_rate": 9.773605263157896e-05,
      "loss": 0.6039,
      "step": 28617
    },
    {
      "epoch": 1.6025310785082316,
      "grad_norm": 1.2628508806228638,
      "learning_rate": 9.773578947368422e-05,
      "loss": 0.3908,
      "step": 28618
    },
    {
      "epoch": 1.6025870758203606,
      "grad_norm": 1.200680136680603,
      "learning_rate": 9.773552631578948e-05,
      "loss": 0.3997,
      "step": 28619
    },
    {
      "epoch": 1.6026430731324897,
      "grad_norm": 1.3624876737594604,
      "learning_rate": 9.773526315789474e-05,
      "loss": 0.4948,
      "step": 28620
    },
    {
      "epoch": 1.6026990704446187,
      "grad_norm": 1.2947224378585815,
      "learning_rate": 9.773500000000001e-05,
      "loss": 0.4947,
      "step": 28621
    },
    {
      "epoch": 1.6027550677567477,
      "grad_norm": 1.4677554368972778,
      "learning_rate": 9.773473684210527e-05,
      "loss": 0.4754,
      "step": 28622
    },
    {
      "epoch": 1.6028110650688767,
      "grad_norm": 1.371080756187439,
      "learning_rate": 9.773447368421053e-05,
      "loss": 0.4582,
      "step": 28623
    },
    {
      "epoch": 1.6028670623810057,
      "grad_norm": 1.522147536277771,
      "learning_rate": 9.773421052631579e-05,
      "loss": 0.5,
      "step": 28624
    },
    {
      "epoch": 1.6029230596931348,
      "grad_norm": 1.4593405723571777,
      "learning_rate": 9.773394736842105e-05,
      "loss": 0.4705,
      "step": 28625
    },
    {
      "epoch": 1.6029790570052638,
      "grad_norm": 1.56641685962677,
      "learning_rate": 9.773368421052632e-05,
      "loss": 0.5664,
      "step": 28626
    },
    {
      "epoch": 1.6030350543173928,
      "grad_norm": 1.2927019596099854,
      "learning_rate": 9.773342105263158e-05,
      "loss": 0.4037,
      "step": 28627
    },
    {
      "epoch": 1.6030910516295218,
      "grad_norm": 1.29274320602417,
      "learning_rate": 9.773315789473684e-05,
      "loss": 0.4363,
      "step": 28628
    },
    {
      "epoch": 1.6031470489416508,
      "grad_norm": 1.2716351747512817,
      "learning_rate": 9.77328947368421e-05,
      "loss": 0.457,
      "step": 28629
    },
    {
      "epoch": 1.6032030462537799,
      "grad_norm": 1.4632830619812012,
      "learning_rate": 9.773263157894738e-05,
      "loss": 0.4265,
      "step": 28630
    },
    {
      "epoch": 1.6032590435659089,
      "grad_norm": 1.4021910429000854,
      "learning_rate": 9.773236842105264e-05,
      "loss": 0.4463,
      "step": 28631
    },
    {
      "epoch": 1.603315040878038,
      "grad_norm": 1.2475794553756714,
      "learning_rate": 9.77321052631579e-05,
      "loss": 0.5772,
      "step": 28632
    },
    {
      "epoch": 1.603371038190167,
      "grad_norm": 1.1905592679977417,
      "learning_rate": 9.773184210526316e-05,
      "loss": 0.3985,
      "step": 28633
    },
    {
      "epoch": 1.603427035502296,
      "grad_norm": 3.8043150901794434,
      "learning_rate": 9.773157894736843e-05,
      "loss": 0.4323,
      "step": 28634
    },
    {
      "epoch": 1.603483032814425,
      "grad_norm": 1.1995283365249634,
      "learning_rate": 9.773131578947369e-05,
      "loss": 0.3641,
      "step": 28635
    },
    {
      "epoch": 1.603539030126554,
      "grad_norm": 1.4894057512283325,
      "learning_rate": 9.773105263157896e-05,
      "loss": 0.4722,
      "step": 28636
    },
    {
      "epoch": 1.603595027438683,
      "grad_norm": 1.453940987586975,
      "learning_rate": 9.773078947368421e-05,
      "loss": 0.4959,
      "step": 28637
    },
    {
      "epoch": 1.603651024750812,
      "grad_norm": 1.3902662992477417,
      "learning_rate": 9.773052631578948e-05,
      "loss": 0.5107,
      "step": 28638
    },
    {
      "epoch": 1.603707022062941,
      "grad_norm": 1.1786566972732544,
      "learning_rate": 9.773026315789474e-05,
      "loss": 0.4885,
      "step": 28639
    },
    {
      "epoch": 1.60376301937507,
      "grad_norm": 1.3461360931396484,
      "learning_rate": 9.773e-05,
      "loss": 0.4654,
      "step": 28640
    },
    {
      "epoch": 1.603819016687199,
      "grad_norm": 1.120482325553894,
      "learning_rate": 9.772973684210527e-05,
      "loss": 0.4848,
      "step": 28641
    },
    {
      "epoch": 1.603875013999328,
      "grad_norm": 2.156015396118164,
      "learning_rate": 9.772947368421052e-05,
      "loss": 0.6806,
      "step": 28642
    },
    {
      "epoch": 1.603931011311457,
      "grad_norm": 1.0594418048858643,
      "learning_rate": 9.77292105263158e-05,
      "loss": 0.3164,
      "step": 28643
    },
    {
      "epoch": 1.6039870086235861,
      "grad_norm": 1.2195169925689697,
      "learning_rate": 9.772894736842105e-05,
      "loss": 0.4343,
      "step": 28644
    },
    {
      "epoch": 1.6040430059357151,
      "grad_norm": 1.37152898311615,
      "learning_rate": 9.772868421052633e-05,
      "loss": 0.4588,
      "step": 28645
    },
    {
      "epoch": 1.6040990032478442,
      "grad_norm": 1.2625083923339844,
      "learning_rate": 9.772842105263159e-05,
      "loss": 0.4527,
      "step": 28646
    },
    {
      "epoch": 1.6041550005599732,
      "grad_norm": 1.5454283952713013,
      "learning_rate": 9.772815789473685e-05,
      "loss": 0.6302,
      "step": 28647
    },
    {
      "epoch": 1.604210997872102,
      "grad_norm": 1.2913886308670044,
      "learning_rate": 9.77278947368421e-05,
      "loss": 0.5043,
      "step": 28648
    },
    {
      "epoch": 1.604266995184231,
      "grad_norm": 1.4128942489624023,
      "learning_rate": 9.772763157894738e-05,
      "loss": 0.4755,
      "step": 28649
    },
    {
      "epoch": 1.60432299249636,
      "grad_norm": 1.3026320934295654,
      "learning_rate": 9.772736842105264e-05,
      "loss": 0.4333,
      "step": 28650
    },
    {
      "epoch": 1.604378989808489,
      "grad_norm": 1.2849268913269043,
      "learning_rate": 9.77271052631579e-05,
      "loss": 0.5542,
      "step": 28651
    },
    {
      "epoch": 1.604434987120618,
      "grad_norm": 1.3272056579589844,
      "learning_rate": 9.772684210526316e-05,
      "loss": 0.3359,
      "step": 28652
    },
    {
      "epoch": 1.604490984432747,
      "grad_norm": 1.3442047834396362,
      "learning_rate": 9.772657894736843e-05,
      "loss": 0.4682,
      "step": 28653
    },
    {
      "epoch": 1.604546981744876,
      "grad_norm": 1.4577223062515259,
      "learning_rate": 9.772631578947369e-05,
      "loss": 0.4763,
      "step": 28654
    },
    {
      "epoch": 1.604602979057005,
      "grad_norm": 1.1649953126907349,
      "learning_rate": 9.772605263157895e-05,
      "loss": 0.4263,
      "step": 28655
    },
    {
      "epoch": 1.6046589763691341,
      "grad_norm": 1.4000461101531982,
      "learning_rate": 9.772578947368421e-05,
      "loss": 0.4431,
      "step": 28656
    },
    {
      "epoch": 1.6047149736812631,
      "grad_norm": 1.293175220489502,
      "learning_rate": 9.772552631578947e-05,
      "loss": 0.4242,
      "step": 28657
    },
    {
      "epoch": 1.6047709709933922,
      "grad_norm": 1.503591537475586,
      "learning_rate": 9.772526315789474e-05,
      "loss": 0.5596,
      "step": 28658
    },
    {
      "epoch": 1.6048269683055212,
      "grad_norm": 1.1230645179748535,
      "learning_rate": 9.7725e-05,
      "loss": 0.3241,
      "step": 28659
    },
    {
      "epoch": 1.6048829656176502,
      "grad_norm": 1.2284173965454102,
      "learning_rate": 9.772473684210526e-05,
      "loss": 0.334,
      "step": 28660
    },
    {
      "epoch": 1.6049389629297792,
      "grad_norm": 1.3745721578598022,
      "learning_rate": 9.772447368421052e-05,
      "loss": 0.3021,
      "step": 28661
    },
    {
      "epoch": 1.6049949602419082,
      "grad_norm": 1.3472864627838135,
      "learning_rate": 9.77242105263158e-05,
      "loss": 0.4249,
      "step": 28662
    },
    {
      "epoch": 1.6050509575540373,
      "grad_norm": 1.5999746322631836,
      "learning_rate": 9.772394736842106e-05,
      "loss": 0.5218,
      "step": 28663
    },
    {
      "epoch": 1.6051069548661663,
      "grad_norm": 2.2409257888793945,
      "learning_rate": 9.772368421052633e-05,
      "loss": 0.6774,
      "step": 28664
    },
    {
      "epoch": 1.6051629521782953,
      "grad_norm": 1.3826230764389038,
      "learning_rate": 9.772342105263158e-05,
      "loss": 0.4914,
      "step": 28665
    },
    {
      "epoch": 1.6052189494904243,
      "grad_norm": 1.2145477533340454,
      "learning_rate": 9.772315789473685e-05,
      "loss": 0.3978,
      "step": 28666
    },
    {
      "epoch": 1.6052749468025533,
      "grad_norm": 1.286562204360962,
      "learning_rate": 9.772289473684211e-05,
      "loss": 0.577,
      "step": 28667
    },
    {
      "epoch": 1.6053309441146824,
      "grad_norm": 1.5623323917388916,
      "learning_rate": 9.772263157894738e-05,
      "loss": 0.4105,
      "step": 28668
    },
    {
      "epoch": 1.6053869414268114,
      "grad_norm": 1.4411369562149048,
      "learning_rate": 9.772236842105263e-05,
      "loss": 0.4962,
      "step": 28669
    },
    {
      "epoch": 1.6054429387389404,
      "grad_norm": 1.319985270500183,
      "learning_rate": 9.77221052631579e-05,
      "loss": 0.6363,
      "step": 28670
    },
    {
      "epoch": 1.6054989360510694,
      "grad_norm": 1.4259133338928223,
      "learning_rate": 9.772184210526316e-05,
      "loss": 0.4932,
      "step": 28671
    },
    {
      "epoch": 1.6055549333631984,
      "grad_norm": 1.8027509450912476,
      "learning_rate": 9.772157894736843e-05,
      "loss": 0.4294,
      "step": 28672
    },
    {
      "epoch": 1.6056109306753275,
      "grad_norm": 1.3412691354751587,
      "learning_rate": 9.77213157894737e-05,
      "loss": 0.4014,
      "step": 28673
    },
    {
      "epoch": 1.6056669279874565,
      "grad_norm": 1.2715134620666504,
      "learning_rate": 9.772105263157894e-05,
      "loss": 0.3943,
      "step": 28674
    },
    {
      "epoch": 1.6057229252995855,
      "grad_norm": 1.5118589401245117,
      "learning_rate": 9.772078947368421e-05,
      "loss": 0.5511,
      "step": 28675
    },
    {
      "epoch": 1.6057789226117145,
      "grad_norm": 1.4125322103500366,
      "learning_rate": 9.772052631578947e-05,
      "loss": 0.4525,
      "step": 28676
    },
    {
      "epoch": 1.6058349199238435,
      "grad_norm": 1.3682007789611816,
      "learning_rate": 9.772026315789475e-05,
      "loss": 0.6081,
      "step": 28677
    },
    {
      "epoch": 1.6058909172359725,
      "grad_norm": 1.3445745706558228,
      "learning_rate": 9.772e-05,
      "loss": 0.4932,
      "step": 28678
    },
    {
      "epoch": 1.6059469145481016,
      "grad_norm": 3.1188995838165283,
      "learning_rate": 9.771973684210527e-05,
      "loss": 0.5,
      "step": 28679
    },
    {
      "epoch": 1.6060029118602306,
      "grad_norm": 1.4412713050842285,
      "learning_rate": 9.771947368421053e-05,
      "loss": 0.4545,
      "step": 28680
    },
    {
      "epoch": 1.6060589091723596,
      "grad_norm": 1.889561653137207,
      "learning_rate": 9.77192105263158e-05,
      "loss": 0.5177,
      "step": 28681
    },
    {
      "epoch": 1.6061149064844886,
      "grad_norm": 1.3179881572723389,
      "learning_rate": 9.771894736842106e-05,
      "loss": 0.5502,
      "step": 28682
    },
    {
      "epoch": 1.6061709037966176,
      "grad_norm": 1.4670768976211548,
      "learning_rate": 9.771868421052632e-05,
      "loss": 0.3974,
      "step": 28683
    },
    {
      "epoch": 1.6062269011087467,
      "grad_norm": 1.2923095226287842,
      "learning_rate": 9.771842105263158e-05,
      "loss": 0.6051,
      "step": 28684
    },
    {
      "epoch": 1.6062828984208757,
      "grad_norm": 1.2992335557937622,
      "learning_rate": 9.771815789473685e-05,
      "loss": 0.5566,
      "step": 28685
    },
    {
      "epoch": 1.6063388957330047,
      "grad_norm": 1.779698371887207,
      "learning_rate": 9.771789473684211e-05,
      "loss": 0.5703,
      "step": 28686
    },
    {
      "epoch": 1.6063948930451337,
      "grad_norm": 1.2247023582458496,
      "learning_rate": 9.771763157894737e-05,
      "loss": 0.4534,
      "step": 28687
    },
    {
      "epoch": 1.6064508903572627,
      "grad_norm": 1.4279910326004028,
      "learning_rate": 9.771736842105263e-05,
      "loss": 0.4967,
      "step": 28688
    },
    {
      "epoch": 1.6065068876693918,
      "grad_norm": 1.3476687669754028,
      "learning_rate": 9.77171052631579e-05,
      "loss": 0.6584,
      "step": 28689
    },
    {
      "epoch": 1.6065628849815208,
      "grad_norm": 1.3156377077102661,
      "learning_rate": 9.771684210526316e-05,
      "loss": 0.5392,
      "step": 28690
    },
    {
      "epoch": 1.6066188822936498,
      "grad_norm": 1.3912566900253296,
      "learning_rate": 9.771657894736844e-05,
      "loss": 0.4671,
      "step": 28691
    },
    {
      "epoch": 1.6066748796057788,
      "grad_norm": 1.198332667350769,
      "learning_rate": 9.771631578947368e-05,
      "loss": 0.3925,
      "step": 28692
    },
    {
      "epoch": 1.6067308769179078,
      "grad_norm": 1.2896145582199097,
      "learning_rate": 9.771605263157894e-05,
      "loss": 0.4474,
      "step": 28693
    },
    {
      "epoch": 1.6067868742300369,
      "grad_norm": 1.2734466791152954,
      "learning_rate": 9.771578947368422e-05,
      "loss": 0.4353,
      "step": 28694
    },
    {
      "epoch": 1.6068428715421659,
      "grad_norm": 1.3854820728302002,
      "learning_rate": 9.771552631578948e-05,
      "loss": 0.4362,
      "step": 28695
    },
    {
      "epoch": 1.606898868854295,
      "grad_norm": 1.2498865127563477,
      "learning_rate": 9.771526315789475e-05,
      "loss": 0.4349,
      "step": 28696
    },
    {
      "epoch": 1.606954866166424,
      "grad_norm": 1.5366005897521973,
      "learning_rate": 9.7715e-05,
      "loss": 0.533,
      "step": 28697
    },
    {
      "epoch": 1.607010863478553,
      "grad_norm": 1.1614415645599365,
      "learning_rate": 9.771473684210527e-05,
      "loss": 0.5351,
      "step": 28698
    },
    {
      "epoch": 1.607066860790682,
      "grad_norm": 1.7099281549453735,
      "learning_rate": 9.771447368421053e-05,
      "loss": 0.6309,
      "step": 28699
    },
    {
      "epoch": 1.607122858102811,
      "grad_norm": 1.2391961812973022,
      "learning_rate": 9.77142105263158e-05,
      "loss": 0.5167,
      "step": 28700
    },
    {
      "epoch": 1.60717885541494,
      "grad_norm": 1.4294838905334473,
      "learning_rate": 9.771394736842106e-05,
      "loss": 0.4542,
      "step": 28701
    },
    {
      "epoch": 1.607234852727069,
      "grad_norm": 0.9986810684204102,
      "learning_rate": 9.771368421052632e-05,
      "loss": 0.318,
      "step": 28702
    },
    {
      "epoch": 1.607290850039198,
      "grad_norm": 1.23465895652771,
      "learning_rate": 9.771342105263158e-05,
      "loss": 0.5482,
      "step": 28703
    },
    {
      "epoch": 1.607346847351327,
      "grad_norm": 1.420244574546814,
      "learning_rate": 9.771315789473685e-05,
      "loss": 0.4165,
      "step": 28704
    },
    {
      "epoch": 1.607402844663456,
      "grad_norm": 1.8302278518676758,
      "learning_rate": 9.771289473684211e-05,
      "loss": 0.5376,
      "step": 28705
    },
    {
      "epoch": 1.607458841975585,
      "grad_norm": 1.2272710800170898,
      "learning_rate": 9.771263157894737e-05,
      "loss": 0.4202,
      "step": 28706
    },
    {
      "epoch": 1.607514839287714,
      "grad_norm": 1.1919933557510376,
      "learning_rate": 9.771236842105263e-05,
      "loss": 0.4382,
      "step": 28707
    },
    {
      "epoch": 1.6075708365998431,
      "grad_norm": 1.2344907522201538,
      "learning_rate": 9.77121052631579e-05,
      "loss": 0.3364,
      "step": 28708
    },
    {
      "epoch": 1.6076268339119721,
      "grad_norm": 1.7100778818130493,
      "learning_rate": 9.771184210526317e-05,
      "loss": 0.5967,
      "step": 28709
    },
    {
      "epoch": 1.6076828312241012,
      "grad_norm": 1.5475964546203613,
      "learning_rate": 9.771157894736843e-05,
      "loss": 0.5107,
      "step": 28710
    },
    {
      "epoch": 1.6077388285362302,
      "grad_norm": 1.8974214792251587,
      "learning_rate": 9.771131578947369e-05,
      "loss": 0.7437,
      "step": 28711
    },
    {
      "epoch": 1.6077948258483592,
      "grad_norm": 1.195861577987671,
      "learning_rate": 9.771105263157895e-05,
      "loss": 0.3528,
      "step": 28712
    },
    {
      "epoch": 1.6078508231604882,
      "grad_norm": 1.2650530338287354,
      "learning_rate": 9.771078947368422e-05,
      "loss": 0.3817,
      "step": 28713
    },
    {
      "epoch": 1.6079068204726172,
      "grad_norm": 1.4194986820220947,
      "learning_rate": 9.771052631578948e-05,
      "loss": 0.509,
      "step": 28714
    },
    {
      "epoch": 1.6079628177847463,
      "grad_norm": 1.1955994367599487,
      "learning_rate": 9.771026315789474e-05,
      "loss": 0.4528,
      "step": 28715
    },
    {
      "epoch": 1.6080188150968753,
      "grad_norm": 1.1781166791915894,
      "learning_rate": 9.771e-05,
      "loss": 0.5074,
      "step": 28716
    },
    {
      "epoch": 1.6080748124090043,
      "grad_norm": 1.5795073509216309,
      "learning_rate": 9.770973684210527e-05,
      "loss": 0.3698,
      "step": 28717
    },
    {
      "epoch": 1.6081308097211333,
      "grad_norm": 1.37860107421875,
      "learning_rate": 9.770947368421053e-05,
      "loss": 0.3894,
      "step": 28718
    },
    {
      "epoch": 1.6081868070332623,
      "grad_norm": 1.3968303203582764,
      "learning_rate": 9.77092105263158e-05,
      "loss": 0.5085,
      "step": 28719
    },
    {
      "epoch": 1.6082428043453914,
      "grad_norm": 1.8028346300125122,
      "learning_rate": 9.770894736842105e-05,
      "loss": 0.4945,
      "step": 28720
    },
    {
      "epoch": 1.6082988016575204,
      "grad_norm": 1.4722049236297607,
      "learning_rate": 9.770868421052632e-05,
      "loss": 0.5267,
      "step": 28721
    },
    {
      "epoch": 1.6083547989696494,
      "grad_norm": 1.2737699747085571,
      "learning_rate": 9.770842105263158e-05,
      "loss": 0.5085,
      "step": 28722
    },
    {
      "epoch": 1.6084107962817784,
      "grad_norm": 1.464971661567688,
      "learning_rate": 9.770815789473686e-05,
      "loss": 0.5654,
      "step": 28723
    },
    {
      "epoch": 1.6084667935939074,
      "grad_norm": 1.0730468034744263,
      "learning_rate": 9.77078947368421e-05,
      "loss": 0.3335,
      "step": 28724
    },
    {
      "epoch": 1.6085227909060364,
      "grad_norm": 1.6783322095870972,
      "learning_rate": 9.770763157894738e-05,
      "loss": 0.5438,
      "step": 28725
    },
    {
      "epoch": 1.6085787882181655,
      "grad_norm": 1.5127896070480347,
      "learning_rate": 9.770736842105264e-05,
      "loss": 0.4101,
      "step": 28726
    },
    {
      "epoch": 1.6086347855302945,
      "grad_norm": 1.3831170797348022,
      "learning_rate": 9.77071052631579e-05,
      "loss": 0.529,
      "step": 28727
    },
    {
      "epoch": 1.6086907828424235,
      "grad_norm": 1.2516558170318604,
      "learning_rate": 9.770684210526317e-05,
      "loss": 0.5008,
      "step": 28728
    },
    {
      "epoch": 1.6087467801545525,
      "grad_norm": 1.1753110885620117,
      "learning_rate": 9.770657894736842e-05,
      "loss": 0.4279,
      "step": 28729
    },
    {
      "epoch": 1.6088027774666815,
      "grad_norm": 1.1439446210861206,
      "learning_rate": 9.770631578947369e-05,
      "loss": 0.4876,
      "step": 28730
    },
    {
      "epoch": 1.6088587747788106,
      "grad_norm": 1.350986361503601,
      "learning_rate": 9.770605263157895e-05,
      "loss": 0.4439,
      "step": 28731
    },
    {
      "epoch": 1.6089147720909396,
      "grad_norm": 1.4621875286102295,
      "learning_rate": 9.770578947368422e-05,
      "loss": 0.4695,
      "step": 28732
    },
    {
      "epoch": 1.6089707694030686,
      "grad_norm": 1.6829105615615845,
      "learning_rate": 9.770552631578948e-05,
      "loss": 0.4594,
      "step": 28733
    },
    {
      "epoch": 1.6090267667151976,
      "grad_norm": 1.4435217380523682,
      "learning_rate": 9.770526315789474e-05,
      "loss": 0.4287,
      "step": 28734
    },
    {
      "epoch": 1.6090827640273266,
      "grad_norm": 1.356022834777832,
      "learning_rate": 9.7705e-05,
      "loss": 0.3338,
      "step": 28735
    },
    {
      "epoch": 1.6091387613394557,
      "grad_norm": 1.1180651187896729,
      "learning_rate": 9.770473684210527e-05,
      "loss": 0.3526,
      "step": 28736
    },
    {
      "epoch": 1.6091947586515847,
      "grad_norm": 1.066878318786621,
      "learning_rate": 9.770447368421053e-05,
      "loss": 0.3508,
      "step": 28737
    },
    {
      "epoch": 1.6092507559637137,
      "grad_norm": 1.2516587972640991,
      "learning_rate": 9.77042105263158e-05,
      "loss": 0.3417,
      "step": 28738
    },
    {
      "epoch": 1.6093067532758427,
      "grad_norm": 1.305559754371643,
      "learning_rate": 9.770394736842105e-05,
      "loss": 0.4857,
      "step": 28739
    },
    {
      "epoch": 1.6093627505879717,
      "grad_norm": 3.1608378887176514,
      "learning_rate": 9.770368421052633e-05,
      "loss": 0.3657,
      "step": 28740
    },
    {
      "epoch": 1.6094187479001008,
      "grad_norm": 1.6207187175750732,
      "learning_rate": 9.770342105263159e-05,
      "loss": 0.5124,
      "step": 28741
    },
    {
      "epoch": 1.6094747452122298,
      "grad_norm": 1.2724627256393433,
      "learning_rate": 9.770315789473685e-05,
      "loss": 0.489,
      "step": 28742
    },
    {
      "epoch": 1.6095307425243588,
      "grad_norm": 1.23508620262146,
      "learning_rate": 9.77028947368421e-05,
      "loss": 0.4922,
      "step": 28743
    },
    {
      "epoch": 1.6095867398364878,
      "grad_norm": 1.3437213897705078,
      "learning_rate": 9.770263157894737e-05,
      "loss": 0.4651,
      "step": 28744
    },
    {
      "epoch": 1.6096427371486168,
      "grad_norm": 1.3021752834320068,
      "learning_rate": 9.770236842105264e-05,
      "loss": 0.4908,
      "step": 28745
    },
    {
      "epoch": 1.6096987344607459,
      "grad_norm": 1.061396837234497,
      "learning_rate": 9.77021052631579e-05,
      "loss": 0.3948,
      "step": 28746
    },
    {
      "epoch": 1.6097547317728749,
      "grad_norm": 1.5615110397338867,
      "learning_rate": 9.770184210526316e-05,
      "loss": 0.5421,
      "step": 28747
    },
    {
      "epoch": 1.6098107290850039,
      "grad_norm": 1.4910013675689697,
      "learning_rate": 9.770157894736842e-05,
      "loss": 0.6022,
      "step": 28748
    },
    {
      "epoch": 1.609866726397133,
      "grad_norm": 1.4547667503356934,
      "learning_rate": 9.770131578947369e-05,
      "loss": 0.5313,
      "step": 28749
    },
    {
      "epoch": 1.609922723709262,
      "grad_norm": 1.4698052406311035,
      "learning_rate": 9.770105263157895e-05,
      "loss": 0.539,
      "step": 28750
    },
    {
      "epoch": 1.609978721021391,
      "grad_norm": 1.3056294918060303,
      "learning_rate": 9.770078947368422e-05,
      "loss": 0.4185,
      "step": 28751
    },
    {
      "epoch": 1.61003471833352,
      "grad_norm": 1.2521520853042603,
      "learning_rate": 9.770052631578947e-05,
      "loss": 0.4903,
      "step": 28752
    },
    {
      "epoch": 1.610090715645649,
      "grad_norm": 1.549988031387329,
      "learning_rate": 9.770026315789474e-05,
      "loss": 0.502,
      "step": 28753
    },
    {
      "epoch": 1.610146712957778,
      "grad_norm": 1.5590424537658691,
      "learning_rate": 9.77e-05,
      "loss": 0.5469,
      "step": 28754
    },
    {
      "epoch": 1.610202710269907,
      "grad_norm": 1.0338451862335205,
      "learning_rate": 9.769973684210528e-05,
      "loss": 0.4221,
      "step": 28755
    },
    {
      "epoch": 1.610258707582036,
      "grad_norm": 1.9603238105773926,
      "learning_rate": 9.769947368421054e-05,
      "loss": 0.5875,
      "step": 28756
    },
    {
      "epoch": 1.610314704894165,
      "grad_norm": 1.5836817026138306,
      "learning_rate": 9.76992105263158e-05,
      "loss": 0.4774,
      "step": 28757
    },
    {
      "epoch": 1.610370702206294,
      "grad_norm": 1.1883763074874878,
      "learning_rate": 9.769894736842106e-05,
      "loss": 0.354,
      "step": 28758
    },
    {
      "epoch": 1.610426699518423,
      "grad_norm": 1.2072845697402954,
      "learning_rate": 9.769868421052633e-05,
      "loss": 0.3925,
      "step": 28759
    },
    {
      "epoch": 1.6104826968305521,
      "grad_norm": 1.1078659296035767,
      "learning_rate": 9.769842105263159e-05,
      "loss": 0.4503,
      "step": 28760
    },
    {
      "epoch": 1.6105386941426811,
      "grad_norm": 1.453686237335205,
      "learning_rate": 9.769815789473683e-05,
      "loss": 0.562,
      "step": 28761
    },
    {
      "epoch": 1.6105946914548102,
      "grad_norm": 1.4762941598892212,
      "learning_rate": 9.769789473684211e-05,
      "loss": 0.6003,
      "step": 28762
    },
    {
      "epoch": 1.6106506887669392,
      "grad_norm": 1.154151439666748,
      "learning_rate": 9.769763157894737e-05,
      "loss": 0.357,
      "step": 28763
    },
    {
      "epoch": 1.6107066860790682,
      "grad_norm": 1.1706411838531494,
      "learning_rate": 9.769736842105264e-05,
      "loss": 0.4488,
      "step": 28764
    },
    {
      "epoch": 1.6107626833911972,
      "grad_norm": 1.2597057819366455,
      "learning_rate": 9.76971052631579e-05,
      "loss": 0.4386,
      "step": 28765
    },
    {
      "epoch": 1.6108186807033262,
      "grad_norm": 1.1595110893249512,
      "learning_rate": 9.769684210526316e-05,
      "loss": 0.4152,
      "step": 28766
    },
    {
      "epoch": 1.6108746780154553,
      "grad_norm": 1.3331243991851807,
      "learning_rate": 9.769657894736842e-05,
      "loss": 0.5759,
      "step": 28767
    },
    {
      "epoch": 1.6109306753275843,
      "grad_norm": 1.391335129737854,
      "learning_rate": 9.76963157894737e-05,
      "loss": 0.6717,
      "step": 28768
    },
    {
      "epoch": 1.6109866726397133,
      "grad_norm": 1.2932710647583008,
      "learning_rate": 9.769605263157895e-05,
      "loss": 0.396,
      "step": 28769
    },
    {
      "epoch": 1.6110426699518423,
      "grad_norm": 1.1795724630355835,
      "learning_rate": 9.769578947368421e-05,
      "loss": 0.501,
      "step": 28770
    },
    {
      "epoch": 1.6110986672639713,
      "grad_norm": 1.2339893579483032,
      "learning_rate": 9.769552631578947e-05,
      "loss": 0.4543,
      "step": 28771
    },
    {
      "epoch": 1.6111546645761003,
      "grad_norm": 3.830420732498169,
      "learning_rate": 9.769526315789475e-05,
      "loss": 0.3997,
      "step": 28772
    },
    {
      "epoch": 1.6112106618882294,
      "grad_norm": 1.1677887439727783,
      "learning_rate": 9.7695e-05,
      "loss": 0.3785,
      "step": 28773
    },
    {
      "epoch": 1.6112666592003584,
      "grad_norm": 1.3042711019515991,
      "learning_rate": 9.769473684210528e-05,
      "loss": 0.6053,
      "step": 28774
    },
    {
      "epoch": 1.6113226565124874,
      "grad_norm": 1.1209880113601685,
      "learning_rate": 9.769447368421053e-05,
      "loss": 0.4068,
      "step": 28775
    },
    {
      "epoch": 1.6113786538246164,
      "grad_norm": 1.3490440845489502,
      "learning_rate": 9.76942105263158e-05,
      "loss": 0.5125,
      "step": 28776
    },
    {
      "epoch": 1.6114346511367454,
      "grad_norm": 1.358336329460144,
      "learning_rate": 9.769394736842106e-05,
      "loss": 0.5652,
      "step": 28777
    },
    {
      "epoch": 1.6114906484488745,
      "grad_norm": 1.1783603429794312,
      "learning_rate": 9.769368421052632e-05,
      "loss": 0.4084,
      "step": 28778
    },
    {
      "epoch": 1.6115466457610035,
      "grad_norm": 1.4164807796478271,
      "learning_rate": 9.769342105263158e-05,
      "loss": 0.5972,
      "step": 28779
    },
    {
      "epoch": 1.6116026430731325,
      "grad_norm": 1.284678339958191,
      "learning_rate": 9.769315789473684e-05,
      "loss": 0.4398,
      "step": 28780
    },
    {
      "epoch": 1.6116586403852615,
      "grad_norm": 1.2834359407424927,
      "learning_rate": 9.769289473684211e-05,
      "loss": 0.5424,
      "step": 28781
    },
    {
      "epoch": 1.6117146376973905,
      "grad_norm": 1.4140981435775757,
      "learning_rate": 9.769263157894737e-05,
      "loss": 0.4268,
      "step": 28782
    },
    {
      "epoch": 1.6117706350095196,
      "grad_norm": 1.4729633331298828,
      "learning_rate": 9.769236842105264e-05,
      "loss": 0.4791,
      "step": 28783
    },
    {
      "epoch": 1.6118266323216486,
      "grad_norm": 1.2354273796081543,
      "learning_rate": 9.769210526315789e-05,
      "loss": 0.489,
      "step": 28784
    },
    {
      "epoch": 1.6118826296337776,
      "grad_norm": 1.4806479215621948,
      "learning_rate": 9.769184210526316e-05,
      "loss": 0.4625,
      "step": 28785
    },
    {
      "epoch": 1.6119386269459066,
      "grad_norm": 1.2784717082977295,
      "learning_rate": 9.769157894736842e-05,
      "loss": 0.4417,
      "step": 28786
    },
    {
      "epoch": 1.6119946242580356,
      "grad_norm": 1.5859538316726685,
      "learning_rate": 9.76913157894737e-05,
      "loss": 0.5529,
      "step": 28787
    },
    {
      "epoch": 1.6120506215701647,
      "grad_norm": 1.3850970268249512,
      "learning_rate": 9.769105263157896e-05,
      "loss": 0.4521,
      "step": 28788
    },
    {
      "epoch": 1.6121066188822937,
      "grad_norm": 1.3167903423309326,
      "learning_rate": 9.769078947368422e-05,
      "loss": 0.4672,
      "step": 28789
    },
    {
      "epoch": 1.6121626161944227,
      "grad_norm": 1.0518642663955688,
      "learning_rate": 9.769052631578948e-05,
      "loss": 0.3206,
      "step": 28790
    },
    {
      "epoch": 1.6122186135065517,
      "grad_norm": 1.182880163192749,
      "learning_rate": 9.769026315789475e-05,
      "loss": 0.4477,
      "step": 28791
    },
    {
      "epoch": 1.6122746108186807,
      "grad_norm": 1.206493854522705,
      "learning_rate": 9.769000000000001e-05,
      "loss": 0.4195,
      "step": 28792
    },
    {
      "epoch": 1.6123306081308098,
      "grad_norm": 1.2070099115371704,
      "learning_rate": 9.768973684210527e-05,
      "loss": 0.5545,
      "step": 28793
    },
    {
      "epoch": 1.6123866054429388,
      "grad_norm": 1.3531379699707031,
      "learning_rate": 9.768947368421053e-05,
      "loss": 0.3874,
      "step": 28794
    },
    {
      "epoch": 1.6124426027550678,
      "grad_norm": 1.6257026195526123,
      "learning_rate": 9.768921052631579e-05,
      "loss": 0.4392,
      "step": 28795
    },
    {
      "epoch": 1.6124986000671968,
      "grad_norm": 1.2899463176727295,
      "learning_rate": 9.768894736842106e-05,
      "loss": 0.3819,
      "step": 28796
    },
    {
      "epoch": 1.6125545973793258,
      "grad_norm": 1.8360083103179932,
      "learning_rate": 9.768868421052632e-05,
      "loss": 0.4681,
      "step": 28797
    },
    {
      "epoch": 1.6126105946914548,
      "grad_norm": 1.408347249031067,
      "learning_rate": 9.768842105263158e-05,
      "loss": 0.4443,
      "step": 28798
    },
    {
      "epoch": 1.6126665920035839,
      "grad_norm": 16.916879653930664,
      "learning_rate": 9.768815789473684e-05,
      "loss": 0.5161,
      "step": 28799
    },
    {
      "epoch": 1.6127225893157129,
      "grad_norm": 1.405197262763977,
      "learning_rate": 9.768789473684211e-05,
      "loss": 0.4787,
      "step": 28800
    },
    {
      "epoch": 1.612778586627842,
      "grad_norm": 1.0854650735855103,
      "learning_rate": 9.768763157894737e-05,
      "loss": 0.4347,
      "step": 28801
    },
    {
      "epoch": 1.612834583939971,
      "grad_norm": 1.2529568672180176,
      "learning_rate": 9.768736842105263e-05,
      "loss": 0.419,
      "step": 28802
    },
    {
      "epoch": 1.6128905812521,
      "grad_norm": 1.1711766719818115,
      "learning_rate": 9.768710526315789e-05,
      "loss": 0.3822,
      "step": 28803
    },
    {
      "epoch": 1.612946578564229,
      "grad_norm": 1.5420488119125366,
      "learning_rate": 9.768684210526317e-05,
      "loss": 0.4751,
      "step": 28804
    },
    {
      "epoch": 1.613002575876358,
      "grad_norm": 1.1327838897705078,
      "learning_rate": 9.768657894736843e-05,
      "loss": 0.4577,
      "step": 28805
    },
    {
      "epoch": 1.613058573188487,
      "grad_norm": 1.602009892463684,
      "learning_rate": 9.76863157894737e-05,
      "loss": 0.6539,
      "step": 28806
    },
    {
      "epoch": 1.613114570500616,
      "grad_norm": 1.3847897052764893,
      "learning_rate": 9.768605263157894e-05,
      "loss": 0.5669,
      "step": 28807
    },
    {
      "epoch": 1.613170567812745,
      "grad_norm": 1.1550319194793701,
      "learning_rate": 9.768578947368422e-05,
      "loss": 0.4647,
      "step": 28808
    },
    {
      "epoch": 1.613226565124874,
      "grad_norm": 1.7570668458938599,
      "learning_rate": 9.768552631578948e-05,
      "loss": 0.4337,
      "step": 28809
    },
    {
      "epoch": 1.613282562437003,
      "grad_norm": 1.6952332258224487,
      "learning_rate": 9.768526315789475e-05,
      "loss": 0.7457,
      "step": 28810
    },
    {
      "epoch": 1.613338559749132,
      "grad_norm": 1.632666826248169,
      "learning_rate": 9.768500000000001e-05,
      "loss": 0.7305,
      "step": 28811
    },
    {
      "epoch": 1.6133945570612611,
      "grad_norm": 1.2979798316955566,
      "learning_rate": 9.768473684210527e-05,
      "loss": 0.5339,
      "step": 28812
    },
    {
      "epoch": 1.6134505543733901,
      "grad_norm": 1.3651719093322754,
      "learning_rate": 9.768447368421053e-05,
      "loss": 0.5517,
      "step": 28813
    },
    {
      "epoch": 1.6135065516855192,
      "grad_norm": 1.3211630582809448,
      "learning_rate": 9.768421052631579e-05,
      "loss": 0.48,
      "step": 28814
    },
    {
      "epoch": 1.6135625489976482,
      "grad_norm": 1.4898940324783325,
      "learning_rate": 9.768394736842106e-05,
      "loss": 0.5233,
      "step": 28815
    },
    {
      "epoch": 1.6136185463097772,
      "grad_norm": 1.2606719732284546,
      "learning_rate": 9.768368421052631e-05,
      "loss": 0.5858,
      "step": 28816
    },
    {
      "epoch": 1.6136745436219062,
      "grad_norm": 1.2960206270217896,
      "learning_rate": 9.768342105263158e-05,
      "loss": 0.4371,
      "step": 28817
    },
    {
      "epoch": 1.6137305409340352,
      "grad_norm": 1.3298046588897705,
      "learning_rate": 9.768315789473684e-05,
      "loss": 0.3939,
      "step": 28818
    },
    {
      "epoch": 1.6137865382461642,
      "grad_norm": 1.3246947526931763,
      "learning_rate": 9.768289473684212e-05,
      "loss": 0.4704,
      "step": 28819
    },
    {
      "epoch": 1.6138425355582933,
      "grad_norm": 1.300347924232483,
      "learning_rate": 9.768263157894738e-05,
      "loss": 0.4161,
      "step": 28820
    },
    {
      "epoch": 1.6138985328704223,
      "grad_norm": 1.2787528038024902,
      "learning_rate": 9.768236842105264e-05,
      "loss": 0.4934,
      "step": 28821
    },
    {
      "epoch": 1.6139545301825513,
      "grad_norm": 1.212634563446045,
      "learning_rate": 9.76821052631579e-05,
      "loss": 0.426,
      "step": 28822
    },
    {
      "epoch": 1.6140105274946803,
      "grad_norm": 1.5091456174850464,
      "learning_rate": 9.768184210526317e-05,
      "loss": 0.4461,
      "step": 28823
    },
    {
      "epoch": 1.6140665248068093,
      "grad_norm": 1.3671553134918213,
      "learning_rate": 9.768157894736843e-05,
      "loss": 0.4442,
      "step": 28824
    },
    {
      "epoch": 1.6141225221189384,
      "grad_norm": 1.0682740211486816,
      "learning_rate": 9.768131578947369e-05,
      "loss": 0.3582,
      "step": 28825
    },
    {
      "epoch": 1.6141785194310674,
      "grad_norm": 1.0848997831344604,
      "learning_rate": 9.768105263157895e-05,
      "loss": 0.3017,
      "step": 28826
    },
    {
      "epoch": 1.6142345167431964,
      "grad_norm": 1.3373178243637085,
      "learning_rate": 9.768078947368422e-05,
      "loss": 0.4526,
      "step": 28827
    },
    {
      "epoch": 1.6142905140553254,
      "grad_norm": 1.0836713314056396,
      "learning_rate": 9.768052631578948e-05,
      "loss": 0.3695,
      "step": 28828
    },
    {
      "epoch": 1.6143465113674544,
      "grad_norm": 1.3050196170806885,
      "learning_rate": 9.768026315789474e-05,
      "loss": 0.5845,
      "step": 28829
    },
    {
      "epoch": 1.6144025086795835,
      "grad_norm": 1.3665285110473633,
      "learning_rate": 9.768e-05,
      "loss": 0.4255,
      "step": 28830
    },
    {
      "epoch": 1.6144585059917125,
      "grad_norm": 1.3075628280639648,
      "learning_rate": 9.767973684210526e-05,
      "loss": 0.5029,
      "step": 28831
    },
    {
      "epoch": 1.6145145033038415,
      "grad_norm": 1.5388578176498413,
      "learning_rate": 9.767947368421053e-05,
      "loss": 0.5079,
      "step": 28832
    },
    {
      "epoch": 1.6145705006159705,
      "grad_norm": 1.5257887840270996,
      "learning_rate": 9.767921052631579e-05,
      "loss": 0.3415,
      "step": 28833
    },
    {
      "epoch": 1.6146264979280995,
      "grad_norm": 1.3473721742630005,
      "learning_rate": 9.767894736842105e-05,
      "loss": 0.5001,
      "step": 28834
    },
    {
      "epoch": 1.6146824952402286,
      "grad_norm": 1.303948163986206,
      "learning_rate": 9.767868421052631e-05,
      "loss": 0.4365,
      "step": 28835
    },
    {
      "epoch": 1.6147384925523576,
      "grad_norm": 1.313559651374817,
      "learning_rate": 9.767842105263159e-05,
      "loss": 0.4855,
      "step": 28836
    },
    {
      "epoch": 1.6147944898644866,
      "grad_norm": 1.1499208211898804,
      "learning_rate": 9.767815789473685e-05,
      "loss": 0.4064,
      "step": 28837
    },
    {
      "epoch": 1.6148504871766156,
      "grad_norm": 1.278723955154419,
      "learning_rate": 9.767789473684212e-05,
      "loss": 0.4662,
      "step": 28838
    },
    {
      "epoch": 1.6149064844887446,
      "grad_norm": 1.3083597421646118,
      "learning_rate": 9.767763157894736e-05,
      "loss": 0.534,
      "step": 28839
    },
    {
      "epoch": 1.6149624818008737,
      "grad_norm": 1.1727782487869263,
      "learning_rate": 9.767736842105264e-05,
      "loss": 0.4407,
      "step": 28840
    },
    {
      "epoch": 1.6150184791130027,
      "grad_norm": 1.8495877981185913,
      "learning_rate": 9.76771052631579e-05,
      "loss": 0.4931,
      "step": 28841
    },
    {
      "epoch": 1.6150744764251317,
      "grad_norm": 1.2436549663543701,
      "learning_rate": 9.767684210526317e-05,
      "loss": 0.3709,
      "step": 28842
    },
    {
      "epoch": 1.6151304737372607,
      "grad_norm": 1.3587554693222046,
      "learning_rate": 9.767657894736843e-05,
      "loss": 0.4508,
      "step": 28843
    },
    {
      "epoch": 1.6151864710493897,
      "grad_norm": 1.335359811782837,
      "learning_rate": 9.767631578947369e-05,
      "loss": 0.4305,
      "step": 28844
    },
    {
      "epoch": 1.6152424683615187,
      "grad_norm": 1.2961686849594116,
      "learning_rate": 9.767605263157895e-05,
      "loss": 0.4625,
      "step": 28845
    },
    {
      "epoch": 1.6152984656736478,
      "grad_norm": 1.3412424325942993,
      "learning_rate": 9.767578947368422e-05,
      "loss": 0.4367,
      "step": 28846
    },
    {
      "epoch": 1.6153544629857768,
      "grad_norm": 1.342451810836792,
      "learning_rate": 9.767552631578948e-05,
      "loss": 0.4499,
      "step": 28847
    },
    {
      "epoch": 1.6154104602979058,
      "grad_norm": 1.823162317276001,
      "learning_rate": 9.767526315789474e-05,
      "loss": 0.4182,
      "step": 28848
    },
    {
      "epoch": 1.6154664576100348,
      "grad_norm": 1.3667343854904175,
      "learning_rate": 9.7675e-05,
      "loss": 0.4889,
      "step": 28849
    },
    {
      "epoch": 1.6155224549221638,
      "grad_norm": 1.4782923460006714,
      "learning_rate": 9.767473684210526e-05,
      "loss": 0.5111,
      "step": 28850
    },
    {
      "epoch": 1.6155784522342929,
      "grad_norm": 1.3616951704025269,
      "learning_rate": 9.767447368421054e-05,
      "loss": 0.4375,
      "step": 28851
    },
    {
      "epoch": 1.6156344495464219,
      "grad_norm": 1.2682764530181885,
      "learning_rate": 9.76742105263158e-05,
      "loss": 0.4144,
      "step": 28852
    },
    {
      "epoch": 1.615690446858551,
      "grad_norm": 1.182246208190918,
      "learning_rate": 9.767394736842105e-05,
      "loss": 0.3703,
      "step": 28853
    },
    {
      "epoch": 1.61574644417068,
      "grad_norm": 1.3063621520996094,
      "learning_rate": 9.767368421052631e-05,
      "loss": 0.4596,
      "step": 28854
    },
    {
      "epoch": 1.615802441482809,
      "grad_norm": 1.2230753898620605,
      "learning_rate": 9.767342105263159e-05,
      "loss": 0.3699,
      "step": 28855
    },
    {
      "epoch": 1.615858438794938,
      "grad_norm": 2.1222188472747803,
      "learning_rate": 9.767315789473685e-05,
      "loss": 0.5617,
      "step": 28856
    },
    {
      "epoch": 1.615914436107067,
      "grad_norm": 1.2271397113800049,
      "learning_rate": 9.767289473684211e-05,
      "loss": 0.3465,
      "step": 28857
    },
    {
      "epoch": 1.615970433419196,
      "grad_norm": 2.0450069904327393,
      "learning_rate": 9.767263157894737e-05,
      "loss": 0.5504,
      "step": 28858
    },
    {
      "epoch": 1.616026430731325,
      "grad_norm": 1.3388069868087769,
      "learning_rate": 9.767236842105264e-05,
      "loss": 0.4284,
      "step": 28859
    },
    {
      "epoch": 1.616082428043454,
      "grad_norm": 1.4325214624404907,
      "learning_rate": 9.76721052631579e-05,
      "loss": 0.3864,
      "step": 28860
    },
    {
      "epoch": 1.616138425355583,
      "grad_norm": 1.2629402875900269,
      "learning_rate": 9.767184210526317e-05,
      "loss": 0.4763,
      "step": 28861
    },
    {
      "epoch": 1.616194422667712,
      "grad_norm": 1.146431803703308,
      "learning_rate": 9.767157894736842e-05,
      "loss": 0.3929,
      "step": 28862
    },
    {
      "epoch": 1.616250419979841,
      "grad_norm": 1.3569384813308716,
      "learning_rate": 9.767131578947369e-05,
      "loss": 0.4921,
      "step": 28863
    },
    {
      "epoch": 1.61630641729197,
      "grad_norm": 1.4281699657440186,
      "learning_rate": 9.767105263157895e-05,
      "loss": 0.454,
      "step": 28864
    },
    {
      "epoch": 1.6163624146040991,
      "grad_norm": 1.4064583778381348,
      "learning_rate": 9.767078947368421e-05,
      "loss": 0.4045,
      "step": 28865
    },
    {
      "epoch": 1.6164184119162281,
      "grad_norm": 1.4081312417984009,
      "learning_rate": 9.767052631578949e-05,
      "loss": 0.439,
      "step": 28866
    },
    {
      "epoch": 1.6164744092283572,
      "grad_norm": 1.4261716604232788,
      "learning_rate": 9.767026315789473e-05,
      "loss": 0.4116,
      "step": 28867
    },
    {
      "epoch": 1.6165304065404862,
      "grad_norm": 1.331688404083252,
      "learning_rate": 9.767e-05,
      "loss": 0.4301,
      "step": 28868
    },
    {
      "epoch": 1.6165864038526152,
      "grad_norm": 1.3330377340316772,
      "learning_rate": 9.766973684210526e-05,
      "loss": 0.4386,
      "step": 28869
    },
    {
      "epoch": 1.6166424011647442,
      "grad_norm": 1.7573132514953613,
      "learning_rate": 9.766947368421054e-05,
      "loss": 0.6382,
      "step": 28870
    },
    {
      "epoch": 1.6166983984768732,
      "grad_norm": 1.5500704050064087,
      "learning_rate": 9.766921052631578e-05,
      "loss": 0.4457,
      "step": 28871
    },
    {
      "epoch": 1.6167543957890023,
      "grad_norm": 1.4348666667938232,
      "learning_rate": 9.766894736842106e-05,
      "loss": 0.4841,
      "step": 28872
    },
    {
      "epoch": 1.6168103931011313,
      "grad_norm": 1.3392975330352783,
      "learning_rate": 9.766868421052632e-05,
      "loss": 0.5253,
      "step": 28873
    },
    {
      "epoch": 1.6168663904132603,
      "grad_norm": 1.9987597465515137,
      "learning_rate": 9.766842105263159e-05,
      "loss": 0.4609,
      "step": 28874
    },
    {
      "epoch": 1.6169223877253893,
      "grad_norm": 2.7324507236480713,
      "learning_rate": 9.766815789473685e-05,
      "loss": 0.6237,
      "step": 28875
    },
    {
      "epoch": 1.6169783850375183,
      "grad_norm": 1.508131742477417,
      "learning_rate": 9.766789473684211e-05,
      "loss": 0.5241,
      "step": 28876
    },
    {
      "epoch": 1.6170343823496474,
      "grad_norm": 1.9047346115112305,
      "learning_rate": 9.766763157894737e-05,
      "loss": 0.4972,
      "step": 28877
    },
    {
      "epoch": 1.6170903796617764,
      "grad_norm": 1.4013547897338867,
      "learning_rate": 9.766736842105264e-05,
      "loss": 0.51,
      "step": 28878
    },
    {
      "epoch": 1.6171463769739054,
      "grad_norm": 2.806600332260132,
      "learning_rate": 9.76671052631579e-05,
      "loss": 0.4834,
      "step": 28879
    },
    {
      "epoch": 1.6172023742860344,
      "grad_norm": 1.1944141387939453,
      "learning_rate": 9.766684210526316e-05,
      "loss": 0.3531,
      "step": 28880
    },
    {
      "epoch": 1.6172583715981634,
      "grad_norm": 1.2647945880889893,
      "learning_rate": 9.766657894736842e-05,
      "loss": 0.529,
      "step": 28881
    },
    {
      "epoch": 1.6173143689102925,
      "grad_norm": 1.6566176414489746,
      "learning_rate": 9.766631578947368e-05,
      "loss": 0.4407,
      "step": 28882
    },
    {
      "epoch": 1.6173703662224215,
      "grad_norm": 1.159731388092041,
      "learning_rate": 9.766605263157896e-05,
      "loss": 0.3763,
      "step": 28883
    },
    {
      "epoch": 1.6174263635345505,
      "grad_norm": 1.4550575017929077,
      "learning_rate": 9.766578947368421e-05,
      "loss": 0.5408,
      "step": 28884
    },
    {
      "epoch": 1.6174823608466795,
      "grad_norm": 1.4085232019424438,
      "learning_rate": 9.766552631578947e-05,
      "loss": 0.4368,
      "step": 28885
    },
    {
      "epoch": 1.6175383581588085,
      "grad_norm": 1.7718677520751953,
      "learning_rate": 9.766526315789473e-05,
      "loss": 0.4229,
      "step": 28886
    },
    {
      "epoch": 1.6175943554709376,
      "grad_norm": 1.944851040840149,
      "learning_rate": 9.766500000000001e-05,
      "loss": 0.4829,
      "step": 28887
    },
    {
      "epoch": 1.6176503527830666,
      "grad_norm": 1.1601150035858154,
      "learning_rate": 9.766473684210527e-05,
      "loss": 0.3789,
      "step": 28888
    },
    {
      "epoch": 1.6177063500951956,
      "grad_norm": 1.2976466417312622,
      "learning_rate": 9.766447368421053e-05,
      "loss": 0.4909,
      "step": 28889
    },
    {
      "epoch": 1.6177623474073246,
      "grad_norm": 1.249280333518982,
      "learning_rate": 9.766421052631579e-05,
      "loss": 0.4483,
      "step": 28890
    },
    {
      "epoch": 1.6178183447194536,
      "grad_norm": 1.6155037879943848,
      "learning_rate": 9.766394736842106e-05,
      "loss": 0.5474,
      "step": 28891
    },
    {
      "epoch": 1.6178743420315826,
      "grad_norm": 1.7790343761444092,
      "learning_rate": 9.766368421052632e-05,
      "loss": 0.4386,
      "step": 28892
    },
    {
      "epoch": 1.6179303393437117,
      "grad_norm": 1.2534726858139038,
      "learning_rate": 9.766342105263159e-05,
      "loss": 0.3634,
      "step": 28893
    },
    {
      "epoch": 1.6179863366558407,
      "grad_norm": 1.319286823272705,
      "learning_rate": 9.766315789473684e-05,
      "loss": 0.5808,
      "step": 28894
    },
    {
      "epoch": 1.6180423339679697,
      "grad_norm": 1.3928951025009155,
      "learning_rate": 9.766289473684211e-05,
      "loss": 0.3583,
      "step": 28895
    },
    {
      "epoch": 1.6180983312800985,
      "grad_norm": 1.4259158372879028,
      "learning_rate": 9.766263157894737e-05,
      "loss": 0.4083,
      "step": 28896
    },
    {
      "epoch": 1.6181543285922275,
      "grad_norm": 1.4068379402160645,
      "learning_rate": 9.766236842105265e-05,
      "loss": 0.6703,
      "step": 28897
    },
    {
      "epoch": 1.6182103259043565,
      "grad_norm": 1.14223051071167,
      "learning_rate": 9.76621052631579e-05,
      "loss": 0.4393,
      "step": 28898
    },
    {
      "epoch": 1.6182663232164856,
      "grad_norm": 1.1985951662063599,
      "learning_rate": 9.766184210526315e-05,
      "loss": 0.3575,
      "step": 28899
    },
    {
      "epoch": 1.6183223205286146,
      "grad_norm": 1.5247827768325806,
      "learning_rate": 9.766157894736842e-05,
      "loss": 0.5642,
      "step": 28900
    },
    {
      "epoch": 1.6183783178407436,
      "grad_norm": 1.274037480354309,
      "learning_rate": 9.766131578947368e-05,
      "loss": 0.4636,
      "step": 28901
    },
    {
      "epoch": 1.6184343151528726,
      "grad_norm": 1.0825755596160889,
      "learning_rate": 9.766105263157896e-05,
      "loss": 0.4752,
      "step": 28902
    },
    {
      "epoch": 1.6184903124650016,
      "grad_norm": 1.1186537742614746,
      "learning_rate": 9.766078947368422e-05,
      "loss": 0.4092,
      "step": 28903
    },
    {
      "epoch": 1.6185463097771307,
      "grad_norm": 1.2335047721862793,
      "learning_rate": 9.766052631578948e-05,
      "loss": 0.4991,
      "step": 28904
    },
    {
      "epoch": 1.6186023070892597,
      "grad_norm": 1.4402673244476318,
      "learning_rate": 9.766026315789474e-05,
      "loss": 0.4952,
      "step": 28905
    },
    {
      "epoch": 1.6186583044013887,
      "grad_norm": 1.5665110349655151,
      "learning_rate": 9.766000000000001e-05,
      "loss": 0.5344,
      "step": 28906
    },
    {
      "epoch": 1.6187143017135177,
      "grad_norm": 1.2146878242492676,
      "learning_rate": 9.765973684210527e-05,
      "loss": 0.4731,
      "step": 28907
    },
    {
      "epoch": 1.6187702990256467,
      "grad_norm": 1.4464036226272583,
      "learning_rate": 9.765947368421053e-05,
      "loss": 0.6648,
      "step": 28908
    },
    {
      "epoch": 1.6188262963377757,
      "grad_norm": 1.3497234582901,
      "learning_rate": 9.765921052631579e-05,
      "loss": 0.5241,
      "step": 28909
    },
    {
      "epoch": 1.6188822936499048,
      "grad_norm": 1.1629825830459595,
      "learning_rate": 9.765894736842106e-05,
      "loss": 0.3955,
      "step": 28910
    },
    {
      "epoch": 1.6189382909620338,
      "grad_norm": 1.1402976512908936,
      "learning_rate": 9.765868421052632e-05,
      "loss": 0.3949,
      "step": 28911
    },
    {
      "epoch": 1.6189942882741628,
      "grad_norm": 1.6595808267593384,
      "learning_rate": 9.765842105263158e-05,
      "loss": 0.5269,
      "step": 28912
    },
    {
      "epoch": 1.6190502855862918,
      "grad_norm": 1.1686064004898071,
      "learning_rate": 9.765815789473684e-05,
      "loss": 0.4283,
      "step": 28913
    },
    {
      "epoch": 1.6191062828984208,
      "grad_norm": 1.1514596939086914,
      "learning_rate": 9.765789473684212e-05,
      "loss": 0.4714,
      "step": 28914
    },
    {
      "epoch": 1.6191622802105499,
      "grad_norm": 1.1812119483947754,
      "learning_rate": 9.765763157894737e-05,
      "loss": 0.3457,
      "step": 28915
    },
    {
      "epoch": 1.6192182775226789,
      "grad_norm": 1.2720551490783691,
      "learning_rate": 9.765736842105263e-05,
      "loss": 0.4326,
      "step": 28916
    },
    {
      "epoch": 1.619274274834808,
      "grad_norm": 2.7686190605163574,
      "learning_rate": 9.76571052631579e-05,
      "loss": 0.701,
      "step": 28917
    },
    {
      "epoch": 1.619330272146937,
      "grad_norm": 1.5894356966018677,
      "learning_rate": 9.765684210526315e-05,
      "loss": 0.6887,
      "step": 28918
    },
    {
      "epoch": 1.619386269459066,
      "grad_norm": 1.4989386796951294,
      "learning_rate": 9.765657894736843e-05,
      "loss": 0.581,
      "step": 28919
    },
    {
      "epoch": 1.619442266771195,
      "grad_norm": 1.2035168409347534,
      "learning_rate": 9.765631578947369e-05,
      "loss": 0.4001,
      "step": 28920
    },
    {
      "epoch": 1.619498264083324,
      "grad_norm": 1.7772064208984375,
      "learning_rate": 9.765605263157896e-05,
      "loss": 0.4919,
      "step": 28921
    },
    {
      "epoch": 1.619554261395453,
      "grad_norm": 1.3816946744918823,
      "learning_rate": 9.76557894736842e-05,
      "loss": 0.4771,
      "step": 28922
    },
    {
      "epoch": 1.619610258707582,
      "grad_norm": 1.5412209033966064,
      "learning_rate": 9.765552631578948e-05,
      "loss": 0.4784,
      "step": 28923
    },
    {
      "epoch": 1.619666256019711,
      "grad_norm": 1.265815019607544,
      "learning_rate": 9.765526315789474e-05,
      "loss": 0.4187,
      "step": 28924
    },
    {
      "epoch": 1.61972225333184,
      "grad_norm": 1.6642132997512817,
      "learning_rate": 9.765500000000001e-05,
      "loss": 0.4875,
      "step": 28925
    },
    {
      "epoch": 1.619778250643969,
      "grad_norm": 1.099763035774231,
      "learning_rate": 9.765473684210526e-05,
      "loss": 0.3481,
      "step": 28926
    },
    {
      "epoch": 1.619834247956098,
      "grad_norm": 1.391595482826233,
      "learning_rate": 9.765447368421053e-05,
      "loss": 0.5049,
      "step": 28927
    },
    {
      "epoch": 1.6198902452682271,
      "grad_norm": 1.264090895652771,
      "learning_rate": 9.765421052631579e-05,
      "loss": 0.401,
      "step": 28928
    },
    {
      "epoch": 1.6199462425803561,
      "grad_norm": 1.3754740953445435,
      "learning_rate": 9.765394736842107e-05,
      "loss": 0.5046,
      "step": 28929
    },
    {
      "epoch": 1.6200022398924852,
      "grad_norm": 1.443232774734497,
      "learning_rate": 9.765368421052633e-05,
      "loss": 0.5517,
      "step": 28930
    },
    {
      "epoch": 1.6200582372046142,
      "grad_norm": 1.269391417503357,
      "learning_rate": 9.765342105263158e-05,
      "loss": 0.5542,
      "step": 28931
    },
    {
      "epoch": 1.6201142345167432,
      "grad_norm": 1.2511099576950073,
      "learning_rate": 9.765315789473684e-05,
      "loss": 0.4431,
      "step": 28932
    },
    {
      "epoch": 1.6201702318288722,
      "grad_norm": 1.5928725004196167,
      "learning_rate": 9.765289473684212e-05,
      "loss": 0.4087,
      "step": 28933
    },
    {
      "epoch": 1.6202262291410012,
      "grad_norm": 1.8670235872268677,
      "learning_rate": 9.765263157894738e-05,
      "loss": 0.478,
      "step": 28934
    },
    {
      "epoch": 1.6202822264531302,
      "grad_norm": 1.4178543090820312,
      "learning_rate": 9.765236842105264e-05,
      "loss": 0.4866,
      "step": 28935
    },
    {
      "epoch": 1.6203382237652593,
      "grad_norm": 1.7292839288711548,
      "learning_rate": 9.76521052631579e-05,
      "loss": 0.5253,
      "step": 28936
    },
    {
      "epoch": 1.6203942210773883,
      "grad_norm": 1.3527981042861938,
      "learning_rate": 9.765184210526316e-05,
      "loss": 0.5333,
      "step": 28937
    },
    {
      "epoch": 1.6204502183895173,
      "grad_norm": 1.0957695245742798,
      "learning_rate": 9.765157894736843e-05,
      "loss": 0.4332,
      "step": 28938
    },
    {
      "epoch": 1.6205062157016463,
      "grad_norm": 1.1085983514785767,
      "learning_rate": 9.765131578947369e-05,
      "loss": 0.3908,
      "step": 28939
    },
    {
      "epoch": 1.6205622130137753,
      "grad_norm": 1.4848521947860718,
      "learning_rate": 9.765105263157895e-05,
      "loss": 0.3817,
      "step": 28940
    },
    {
      "epoch": 1.6206182103259044,
      "grad_norm": 1.2339684963226318,
      "learning_rate": 9.765078947368421e-05,
      "loss": 0.4563,
      "step": 28941
    },
    {
      "epoch": 1.6206742076380334,
      "grad_norm": 1.0157705545425415,
      "learning_rate": 9.765052631578948e-05,
      "loss": 0.3975,
      "step": 28942
    },
    {
      "epoch": 1.6207302049501624,
      "grad_norm": 1.3542698621749878,
      "learning_rate": 9.765026315789474e-05,
      "loss": 0.5248,
      "step": 28943
    },
    {
      "epoch": 1.6207862022622914,
      "grad_norm": 1.5024199485778809,
      "learning_rate": 9.765e-05,
      "loss": 0.5143,
      "step": 28944
    },
    {
      "epoch": 1.6208421995744204,
      "grad_norm": 1.2035006284713745,
      "learning_rate": 9.764973684210526e-05,
      "loss": 0.4675,
      "step": 28945
    },
    {
      "epoch": 1.6208981968865495,
      "grad_norm": 4.292696475982666,
      "learning_rate": 9.764947368421053e-05,
      "loss": 0.7096,
      "step": 28946
    },
    {
      "epoch": 1.6209541941986785,
      "grad_norm": 1.3858609199523926,
      "learning_rate": 9.76492105263158e-05,
      "loss": 0.45,
      "step": 28947
    },
    {
      "epoch": 1.6210101915108075,
      "grad_norm": 1.3534660339355469,
      "learning_rate": 9.764894736842107e-05,
      "loss": 0.4051,
      "step": 28948
    },
    {
      "epoch": 1.6210661888229365,
      "grad_norm": 1.4373719692230225,
      "learning_rate": 9.764868421052631e-05,
      "loss": 0.4589,
      "step": 28949
    },
    {
      "epoch": 1.6211221861350655,
      "grad_norm": 1.2982500791549683,
      "learning_rate": 9.764842105263159e-05,
      "loss": 0.4363,
      "step": 28950
    },
    {
      "epoch": 1.6211781834471946,
      "grad_norm": 1.3825939893722534,
      "learning_rate": 9.764815789473685e-05,
      "loss": 0.5435,
      "step": 28951
    },
    {
      "epoch": 1.6212341807593236,
      "grad_norm": 1.072726845741272,
      "learning_rate": 9.764789473684211e-05,
      "loss": 0.4476,
      "step": 28952
    },
    {
      "epoch": 1.6212901780714526,
      "grad_norm": 1.3606057167053223,
      "learning_rate": 9.764763157894738e-05,
      "loss": 0.4894,
      "step": 28953
    },
    {
      "epoch": 1.6213461753835816,
      "grad_norm": 1.4802929162979126,
      "learning_rate": 9.764736842105263e-05,
      "loss": 0.4803,
      "step": 28954
    },
    {
      "epoch": 1.6214021726957106,
      "grad_norm": 1.1817505359649658,
      "learning_rate": 9.76471052631579e-05,
      "loss": 0.3624,
      "step": 28955
    },
    {
      "epoch": 1.6214581700078396,
      "grad_norm": 1.829140067100525,
      "learning_rate": 9.764684210526316e-05,
      "loss": 0.686,
      "step": 28956
    },
    {
      "epoch": 1.6215141673199687,
      "grad_norm": 1.3381738662719727,
      "learning_rate": 9.764657894736843e-05,
      "loss": 0.4526,
      "step": 28957
    },
    {
      "epoch": 1.6215701646320977,
      "grad_norm": 1.410481572151184,
      "learning_rate": 9.764631578947369e-05,
      "loss": 0.5079,
      "step": 28958
    },
    {
      "epoch": 1.6216261619442267,
      "grad_norm": 1.2489384412765503,
      "learning_rate": 9.764605263157895e-05,
      "loss": 0.4669,
      "step": 28959
    },
    {
      "epoch": 1.6216821592563557,
      "grad_norm": 1.709755301475525,
      "learning_rate": 9.764578947368421e-05,
      "loss": 0.6166,
      "step": 28960
    },
    {
      "epoch": 1.6217381565684847,
      "grad_norm": 1.248163104057312,
      "learning_rate": 9.764552631578949e-05,
      "loss": 0.491,
      "step": 28961
    },
    {
      "epoch": 1.6217941538806138,
      "grad_norm": 1.2984687089920044,
      "learning_rate": 9.764526315789474e-05,
      "loss": 0.4477,
      "step": 28962
    },
    {
      "epoch": 1.6218501511927428,
      "grad_norm": 1.3458508253097534,
      "learning_rate": 9.7645e-05,
      "loss": 0.4577,
      "step": 28963
    },
    {
      "epoch": 1.6219061485048718,
      "grad_norm": 1.5114479064941406,
      "learning_rate": 9.764473684210526e-05,
      "loss": 0.704,
      "step": 28964
    },
    {
      "epoch": 1.6219621458170008,
      "grad_norm": 1.4598270654678345,
      "learning_rate": 9.764447368421054e-05,
      "loss": 0.4764,
      "step": 28965
    },
    {
      "epoch": 1.6220181431291298,
      "grad_norm": 1.3080896139144897,
      "learning_rate": 9.76442105263158e-05,
      "loss": 0.4064,
      "step": 28966
    },
    {
      "epoch": 1.6220741404412589,
      "grad_norm": 1.2693126201629639,
      "learning_rate": 9.764394736842106e-05,
      "loss": 0.438,
      "step": 28967
    },
    {
      "epoch": 1.6221301377533879,
      "grad_norm": 1.0561363697052002,
      "learning_rate": 9.764368421052632e-05,
      "loss": 0.3789,
      "step": 28968
    },
    {
      "epoch": 1.622186135065517,
      "grad_norm": 1.2383800745010376,
      "learning_rate": 9.764342105263158e-05,
      "loss": 0.4333,
      "step": 28969
    },
    {
      "epoch": 1.622242132377646,
      "grad_norm": 1.2280343770980835,
      "learning_rate": 9.764315789473685e-05,
      "loss": 0.5021,
      "step": 28970
    },
    {
      "epoch": 1.622298129689775,
      "grad_norm": 1.3351688385009766,
      "learning_rate": 9.764289473684211e-05,
      "loss": 0.505,
      "step": 28971
    },
    {
      "epoch": 1.622354127001904,
      "grad_norm": 1.3184148073196411,
      "learning_rate": 9.764263157894737e-05,
      "loss": 0.5307,
      "step": 28972
    },
    {
      "epoch": 1.622410124314033,
      "grad_norm": 1.5563063621520996,
      "learning_rate": 9.764236842105263e-05,
      "loss": 0.4808,
      "step": 28973
    },
    {
      "epoch": 1.622466121626162,
      "grad_norm": 1.2767109870910645,
      "learning_rate": 9.76421052631579e-05,
      "loss": 0.5952,
      "step": 28974
    },
    {
      "epoch": 1.622522118938291,
      "grad_norm": 1.2463483810424805,
      "learning_rate": 9.764184210526316e-05,
      "loss": 0.377,
      "step": 28975
    },
    {
      "epoch": 1.62257811625042,
      "grad_norm": 1.4963417053222656,
      "learning_rate": 9.764157894736844e-05,
      "loss": 0.4538,
      "step": 28976
    },
    {
      "epoch": 1.622634113562549,
      "grad_norm": 1.3054777383804321,
      "learning_rate": 9.764131578947368e-05,
      "loss": 0.4315,
      "step": 28977
    },
    {
      "epoch": 1.6226901108746778,
      "grad_norm": 1.5979790687561035,
      "learning_rate": 9.764105263157895e-05,
      "loss": 0.4133,
      "step": 28978
    },
    {
      "epoch": 1.6227461081868069,
      "grad_norm": 1.4845672845840454,
      "learning_rate": 9.764078947368421e-05,
      "loss": 0.457,
      "step": 28979
    },
    {
      "epoch": 1.6228021054989359,
      "grad_norm": 1.5465277433395386,
      "learning_rate": 9.764052631578949e-05,
      "loss": 0.405,
      "step": 28980
    },
    {
      "epoch": 1.622858102811065,
      "grad_norm": 1.4176063537597656,
      "learning_rate": 9.764026315789473e-05,
      "loss": 0.384,
      "step": 28981
    },
    {
      "epoch": 1.622914100123194,
      "grad_norm": 1.4219379425048828,
      "learning_rate": 9.764000000000001e-05,
      "loss": 0.4869,
      "step": 28982
    },
    {
      "epoch": 1.622970097435323,
      "grad_norm": 1.5295010805130005,
      "learning_rate": 9.763973684210527e-05,
      "loss": 0.4806,
      "step": 28983
    },
    {
      "epoch": 1.623026094747452,
      "grad_norm": 1.2485733032226562,
      "learning_rate": 9.763947368421054e-05,
      "loss": 0.4526,
      "step": 28984
    },
    {
      "epoch": 1.623082092059581,
      "grad_norm": 1.1920751333236694,
      "learning_rate": 9.76392105263158e-05,
      "loss": 0.4755,
      "step": 28985
    },
    {
      "epoch": 1.62313808937171,
      "grad_norm": 1.9188928604125977,
      "learning_rate": 9.763894736842105e-05,
      "loss": 0.5597,
      "step": 28986
    },
    {
      "epoch": 1.623194086683839,
      "grad_norm": 1.6881300210952759,
      "learning_rate": 9.763868421052632e-05,
      "loss": 0.5532,
      "step": 28987
    },
    {
      "epoch": 1.623250083995968,
      "grad_norm": 1.4397168159484863,
      "learning_rate": 9.763842105263158e-05,
      "loss": 0.5602,
      "step": 28988
    },
    {
      "epoch": 1.623306081308097,
      "grad_norm": 1.4025870561599731,
      "learning_rate": 9.763815789473685e-05,
      "loss": 0.3857,
      "step": 28989
    },
    {
      "epoch": 1.623362078620226,
      "grad_norm": 14.822052955627441,
      "learning_rate": 9.763789473684211e-05,
      "loss": 0.4411,
      "step": 28990
    },
    {
      "epoch": 1.623418075932355,
      "grad_norm": 1.1370784044265747,
      "learning_rate": 9.763763157894737e-05,
      "loss": 0.3767,
      "step": 28991
    },
    {
      "epoch": 1.6234740732444841,
      "grad_norm": 1.361096978187561,
      "learning_rate": 9.763736842105263e-05,
      "loss": 0.6225,
      "step": 28992
    },
    {
      "epoch": 1.6235300705566131,
      "grad_norm": 1.27276611328125,
      "learning_rate": 9.76371052631579e-05,
      "loss": 0.465,
      "step": 28993
    },
    {
      "epoch": 1.6235860678687422,
      "grad_norm": 1.3719966411590576,
      "learning_rate": 9.763684210526316e-05,
      "loss": 0.6489,
      "step": 28994
    },
    {
      "epoch": 1.6236420651808712,
      "grad_norm": 1.4883934259414673,
      "learning_rate": 9.763657894736842e-05,
      "loss": 0.4197,
      "step": 28995
    },
    {
      "epoch": 1.6236980624930002,
      "grad_norm": 1.515711784362793,
      "learning_rate": 9.763631578947368e-05,
      "loss": 0.4197,
      "step": 28996
    },
    {
      "epoch": 1.6237540598051292,
      "grad_norm": 1.0941756963729858,
      "learning_rate": 9.763605263157896e-05,
      "loss": 0.4487,
      "step": 28997
    },
    {
      "epoch": 1.6238100571172582,
      "grad_norm": 1.2121586799621582,
      "learning_rate": 9.763578947368422e-05,
      "loss": 0.4212,
      "step": 28998
    },
    {
      "epoch": 1.6238660544293873,
      "grad_norm": 1.2333850860595703,
      "learning_rate": 9.763552631578948e-05,
      "loss": 0.4775,
      "step": 28999
    },
    {
      "epoch": 1.6239220517415163,
      "grad_norm": 1.2895197868347168,
      "learning_rate": 9.763526315789474e-05,
      "loss": 0.5964,
      "step": 29000
    },
    {
      "epoch": 1.6239780490536453,
      "grad_norm": 1.435691475868225,
      "learning_rate": 9.763500000000001e-05,
      "loss": 0.4094,
      "step": 29001
    },
    {
      "epoch": 1.6240340463657743,
      "grad_norm": 2.209624767303467,
      "learning_rate": 9.763473684210527e-05,
      "loss": 0.4834,
      "step": 29002
    },
    {
      "epoch": 1.6240900436779033,
      "grad_norm": 1.4166560173034668,
      "learning_rate": 9.763447368421053e-05,
      "loss": 0.6601,
      "step": 29003
    },
    {
      "epoch": 1.6241460409900323,
      "grad_norm": 1.151734471321106,
      "learning_rate": 9.763421052631579e-05,
      "loss": 0.4065,
      "step": 29004
    },
    {
      "epoch": 1.6242020383021614,
      "grad_norm": 1.2314025163650513,
      "learning_rate": 9.763394736842105e-05,
      "loss": 0.511,
      "step": 29005
    },
    {
      "epoch": 1.6242580356142904,
      "grad_norm": 1.1897450685501099,
      "learning_rate": 9.763368421052632e-05,
      "loss": 0.3494,
      "step": 29006
    },
    {
      "epoch": 1.6243140329264194,
      "grad_norm": 1.5910755395889282,
      "learning_rate": 9.763342105263158e-05,
      "loss": 0.5022,
      "step": 29007
    },
    {
      "epoch": 1.6243700302385484,
      "grad_norm": 1.2150752544403076,
      "learning_rate": 9.763315789473685e-05,
      "loss": 0.4724,
      "step": 29008
    },
    {
      "epoch": 1.6244260275506774,
      "grad_norm": 1.533077597618103,
      "learning_rate": 9.76328947368421e-05,
      "loss": 0.5095,
      "step": 29009
    },
    {
      "epoch": 1.6244820248628065,
      "grad_norm": 1.3829405307769775,
      "learning_rate": 9.763263157894737e-05,
      "loss": 0.4628,
      "step": 29010
    },
    {
      "epoch": 1.6245380221749355,
      "grad_norm": 1.334855318069458,
      "learning_rate": 9.763236842105263e-05,
      "loss": 0.3906,
      "step": 29011
    },
    {
      "epoch": 1.6245940194870645,
      "grad_norm": 1.2197556495666504,
      "learning_rate": 9.763210526315791e-05,
      "loss": 0.5492,
      "step": 29012
    },
    {
      "epoch": 1.6246500167991935,
      "grad_norm": 1.3573681116104126,
      "learning_rate": 9.763184210526317e-05,
      "loss": 0.5445,
      "step": 29013
    },
    {
      "epoch": 1.6247060141113225,
      "grad_norm": 1.6501281261444092,
      "learning_rate": 9.763157894736843e-05,
      "loss": 0.4317,
      "step": 29014
    },
    {
      "epoch": 1.6247620114234516,
      "grad_norm": 1.2500278949737549,
      "learning_rate": 9.763131578947369e-05,
      "loss": 0.385,
      "step": 29015
    },
    {
      "epoch": 1.6248180087355806,
      "grad_norm": 1.5370055437088013,
      "learning_rate": 9.763105263157896e-05,
      "loss": 0.522,
      "step": 29016
    },
    {
      "epoch": 1.6248740060477096,
      "grad_norm": 1.313562273979187,
      "learning_rate": 9.763078947368422e-05,
      "loss": 0.3867,
      "step": 29017
    },
    {
      "epoch": 1.6249300033598386,
      "grad_norm": 1.272716760635376,
      "learning_rate": 9.763052631578948e-05,
      "loss": 0.4413,
      "step": 29018
    },
    {
      "epoch": 1.6249860006719676,
      "grad_norm": 1.325479507446289,
      "learning_rate": 9.763026315789474e-05,
      "loss": 0.4495,
      "step": 29019
    },
    {
      "epoch": 1.6250419979840967,
      "grad_norm": 1.1775710582733154,
      "learning_rate": 9.763e-05,
      "loss": 0.6156,
      "step": 29020
    },
    {
      "epoch": 1.6250979952962257,
      "grad_norm": 1.4680395126342773,
      "learning_rate": 9.762973684210527e-05,
      "loss": 0.4178,
      "step": 29021
    },
    {
      "epoch": 1.6251539926083547,
      "grad_norm": 1.3455936908721924,
      "learning_rate": 9.762947368421053e-05,
      "loss": 0.3931,
      "step": 29022
    },
    {
      "epoch": 1.6252099899204837,
      "grad_norm": 1.1602425575256348,
      "learning_rate": 9.762921052631579e-05,
      "loss": 0.4051,
      "step": 29023
    },
    {
      "epoch": 1.6252659872326127,
      "grad_norm": 1.522652506828308,
      "learning_rate": 9.762894736842105e-05,
      "loss": 0.4474,
      "step": 29024
    },
    {
      "epoch": 1.6253219845447417,
      "grad_norm": 1.1786167621612549,
      "learning_rate": 9.762868421052632e-05,
      "loss": 0.6032,
      "step": 29025
    },
    {
      "epoch": 1.6253779818568708,
      "grad_norm": 1.346570372581482,
      "learning_rate": 9.762842105263158e-05,
      "loss": 0.4033,
      "step": 29026
    },
    {
      "epoch": 1.6254339791689998,
      "grad_norm": 1.21056067943573,
      "learning_rate": 9.762815789473684e-05,
      "loss": 0.5341,
      "step": 29027
    },
    {
      "epoch": 1.6254899764811288,
      "grad_norm": 1.5269578695297241,
      "learning_rate": 9.76278947368421e-05,
      "loss": 0.5068,
      "step": 29028
    },
    {
      "epoch": 1.6255459737932578,
      "grad_norm": 1.4292340278625488,
      "learning_rate": 9.762763157894738e-05,
      "loss": 0.4894,
      "step": 29029
    },
    {
      "epoch": 1.6256019711053868,
      "grad_norm": 1.1982123851776123,
      "learning_rate": 9.762736842105264e-05,
      "loss": 0.4273,
      "step": 29030
    },
    {
      "epoch": 1.6256579684175159,
      "grad_norm": 1.709243893623352,
      "learning_rate": 9.76271052631579e-05,
      "loss": 0.5359,
      "step": 29031
    },
    {
      "epoch": 1.6257139657296449,
      "grad_norm": 1.1783006191253662,
      "learning_rate": 9.762684210526316e-05,
      "loss": 0.3656,
      "step": 29032
    },
    {
      "epoch": 1.625769963041774,
      "grad_norm": 1.061718225479126,
      "learning_rate": 9.762657894736843e-05,
      "loss": 0.4325,
      "step": 29033
    },
    {
      "epoch": 1.625825960353903,
      "grad_norm": 2.1916258335113525,
      "learning_rate": 9.762631578947369e-05,
      "loss": 0.4931,
      "step": 29034
    },
    {
      "epoch": 1.625881957666032,
      "grad_norm": 1.1895729303359985,
      "learning_rate": 9.762605263157896e-05,
      "loss": 0.3862,
      "step": 29035
    },
    {
      "epoch": 1.625937954978161,
      "grad_norm": 1.0891497135162354,
      "learning_rate": 9.762578947368421e-05,
      "loss": 0.4223,
      "step": 29036
    },
    {
      "epoch": 1.62599395229029,
      "grad_norm": 1.2871347665786743,
      "learning_rate": 9.762552631578947e-05,
      "loss": 0.4835,
      "step": 29037
    },
    {
      "epoch": 1.626049949602419,
      "grad_norm": 1.1227421760559082,
      "learning_rate": 9.762526315789474e-05,
      "loss": 0.3586,
      "step": 29038
    },
    {
      "epoch": 1.626105946914548,
      "grad_norm": 1.5562020540237427,
      "learning_rate": 9.7625e-05,
      "loss": 0.3576,
      "step": 29039
    },
    {
      "epoch": 1.626161944226677,
      "grad_norm": 1.2298904657363892,
      "learning_rate": 9.762473684210527e-05,
      "loss": 0.5008,
      "step": 29040
    },
    {
      "epoch": 1.626217941538806,
      "grad_norm": 1.3883846998214722,
      "learning_rate": 9.762447368421052e-05,
      "loss": 0.4788,
      "step": 29041
    },
    {
      "epoch": 1.626273938850935,
      "grad_norm": 1.3229891061782837,
      "learning_rate": 9.76242105263158e-05,
      "loss": 0.503,
      "step": 29042
    },
    {
      "epoch": 1.626329936163064,
      "grad_norm": 1.4288462400436401,
      "learning_rate": 9.762394736842105e-05,
      "loss": 0.5985,
      "step": 29043
    },
    {
      "epoch": 1.6263859334751931,
      "grad_norm": 1.5787339210510254,
      "learning_rate": 9.762368421052633e-05,
      "loss": 0.4384,
      "step": 29044
    },
    {
      "epoch": 1.6264419307873221,
      "grad_norm": 1.0647637844085693,
      "learning_rate": 9.762342105263159e-05,
      "loss": 0.3867,
      "step": 29045
    },
    {
      "epoch": 1.6264979280994512,
      "grad_norm": 1.180756688117981,
      "learning_rate": 9.762315789473685e-05,
      "loss": 0.4342,
      "step": 29046
    },
    {
      "epoch": 1.6265539254115802,
      "grad_norm": 1.1512324810028076,
      "learning_rate": 9.76228947368421e-05,
      "loss": 0.4566,
      "step": 29047
    },
    {
      "epoch": 1.6266099227237092,
      "grad_norm": 1.1414687633514404,
      "learning_rate": 9.762263157894738e-05,
      "loss": 0.3501,
      "step": 29048
    },
    {
      "epoch": 1.6266659200358382,
      "grad_norm": 1.374139428138733,
      "learning_rate": 9.762236842105264e-05,
      "loss": 0.4822,
      "step": 29049
    },
    {
      "epoch": 1.6267219173479672,
      "grad_norm": 1.401052713394165,
      "learning_rate": 9.76221052631579e-05,
      "loss": 0.4372,
      "step": 29050
    },
    {
      "epoch": 1.6267779146600962,
      "grad_norm": 1.2061361074447632,
      "learning_rate": 9.762184210526316e-05,
      "loss": 0.3528,
      "step": 29051
    },
    {
      "epoch": 1.6268339119722253,
      "grad_norm": 1.4091142416000366,
      "learning_rate": 9.762157894736843e-05,
      "loss": 0.5102,
      "step": 29052
    },
    {
      "epoch": 1.6268899092843543,
      "grad_norm": 1.4907644987106323,
      "learning_rate": 9.762131578947369e-05,
      "loss": 0.6561,
      "step": 29053
    },
    {
      "epoch": 1.6269459065964833,
      "grad_norm": 1.1247915029525757,
      "learning_rate": 9.762105263157895e-05,
      "loss": 0.4596,
      "step": 29054
    },
    {
      "epoch": 1.6270019039086123,
      "grad_norm": 1.152758002281189,
      "learning_rate": 9.762078947368421e-05,
      "loss": 0.4631,
      "step": 29055
    },
    {
      "epoch": 1.6270579012207413,
      "grad_norm": 1.2452892065048218,
      "learning_rate": 9.762052631578947e-05,
      "loss": 0.3635,
      "step": 29056
    },
    {
      "epoch": 1.6271138985328704,
      "grad_norm": 1.5018380880355835,
      "learning_rate": 9.762026315789474e-05,
      "loss": 0.5134,
      "step": 29057
    },
    {
      "epoch": 1.6271698958449994,
      "grad_norm": 1.26292085647583,
      "learning_rate": 9.762e-05,
      "loss": 0.4954,
      "step": 29058
    },
    {
      "epoch": 1.6272258931571284,
      "grad_norm": 1.3581428527832031,
      "learning_rate": 9.761973684210526e-05,
      "loss": 0.451,
      "step": 29059
    },
    {
      "epoch": 1.6272818904692574,
      "grad_norm": 1.4662487506866455,
      "learning_rate": 9.761947368421052e-05,
      "loss": 0.4825,
      "step": 29060
    },
    {
      "epoch": 1.6273378877813864,
      "grad_norm": 1.0710781812667847,
      "learning_rate": 9.76192105263158e-05,
      "loss": 0.3624,
      "step": 29061
    },
    {
      "epoch": 1.6273938850935155,
      "grad_norm": 1.219622254371643,
      "learning_rate": 9.761894736842106e-05,
      "loss": 0.43,
      "step": 29062
    },
    {
      "epoch": 1.6274498824056445,
      "grad_norm": 1.3126916885375977,
      "learning_rate": 9.761868421052633e-05,
      "loss": 0.4779,
      "step": 29063
    },
    {
      "epoch": 1.6275058797177735,
      "grad_norm": 1.0866769552230835,
      "learning_rate": 9.761842105263158e-05,
      "loss": 0.4377,
      "step": 29064
    },
    {
      "epoch": 1.6275618770299025,
      "grad_norm": 1.2525542974472046,
      "learning_rate": 9.761815789473685e-05,
      "loss": 0.4171,
      "step": 29065
    },
    {
      "epoch": 1.6276178743420315,
      "grad_norm": 1.3107547760009766,
      "learning_rate": 9.761789473684211e-05,
      "loss": 0.4772,
      "step": 29066
    },
    {
      "epoch": 1.6276738716541606,
      "grad_norm": 1.738831639289856,
      "learning_rate": 9.761763157894738e-05,
      "loss": 0.5713,
      "step": 29067
    },
    {
      "epoch": 1.6277298689662896,
      "grad_norm": 1.6793580055236816,
      "learning_rate": 9.761736842105264e-05,
      "loss": 0.4374,
      "step": 29068
    },
    {
      "epoch": 1.6277858662784186,
      "grad_norm": 1.2854893207550049,
      "learning_rate": 9.76171052631579e-05,
      "loss": 0.3926,
      "step": 29069
    },
    {
      "epoch": 1.6278418635905476,
      "grad_norm": 1.1788318157196045,
      "learning_rate": 9.761684210526316e-05,
      "loss": 0.4633,
      "step": 29070
    },
    {
      "epoch": 1.6278978609026766,
      "grad_norm": 1.4562842845916748,
      "learning_rate": 9.761657894736843e-05,
      "loss": 0.3869,
      "step": 29071
    },
    {
      "epoch": 1.6279538582148056,
      "grad_norm": 1.302661657333374,
      "learning_rate": 9.76163157894737e-05,
      "loss": 0.4805,
      "step": 29072
    },
    {
      "epoch": 1.6280098555269347,
      "grad_norm": 1.3763971328735352,
      "learning_rate": 9.761605263157894e-05,
      "loss": 0.3982,
      "step": 29073
    },
    {
      "epoch": 1.6280658528390637,
      "grad_norm": 1.9479624032974243,
      "learning_rate": 9.761578947368421e-05,
      "loss": 0.4567,
      "step": 29074
    },
    {
      "epoch": 1.6281218501511927,
      "grad_norm": 1.659568190574646,
      "learning_rate": 9.761552631578947e-05,
      "loss": 0.71,
      "step": 29075
    },
    {
      "epoch": 1.6281778474633217,
      "grad_norm": 1.4795762300491333,
      "learning_rate": 9.761526315789475e-05,
      "loss": 0.4887,
      "step": 29076
    },
    {
      "epoch": 1.6282338447754507,
      "grad_norm": 1.327354073524475,
      "learning_rate": 9.7615e-05,
      "loss": 0.5031,
      "step": 29077
    },
    {
      "epoch": 1.6282898420875798,
      "grad_norm": 1.9933340549468994,
      "learning_rate": 9.761473684210527e-05,
      "loss": 0.3958,
      "step": 29078
    },
    {
      "epoch": 1.6283458393997088,
      "grad_norm": 1.9714298248291016,
      "learning_rate": 9.761447368421053e-05,
      "loss": 0.5657,
      "step": 29079
    },
    {
      "epoch": 1.6284018367118378,
      "grad_norm": 1.4459640979766846,
      "learning_rate": 9.76142105263158e-05,
      "loss": 0.4235,
      "step": 29080
    },
    {
      "epoch": 1.6284578340239668,
      "grad_norm": 1.193831443786621,
      "learning_rate": 9.761394736842106e-05,
      "loss": 0.441,
      "step": 29081
    },
    {
      "epoch": 1.6285138313360958,
      "grad_norm": 1.1125028133392334,
      "learning_rate": 9.761368421052632e-05,
      "loss": 0.4223,
      "step": 29082
    },
    {
      "epoch": 1.6285698286482249,
      "grad_norm": 1.3093454837799072,
      "learning_rate": 9.761342105263158e-05,
      "loss": 0.4256,
      "step": 29083
    },
    {
      "epoch": 1.6286258259603539,
      "grad_norm": 2.2786388397216797,
      "learning_rate": 9.761315789473685e-05,
      "loss": 0.54,
      "step": 29084
    },
    {
      "epoch": 1.628681823272483,
      "grad_norm": 1.323691964149475,
      "learning_rate": 9.761289473684211e-05,
      "loss": 0.5686,
      "step": 29085
    },
    {
      "epoch": 1.628737820584612,
      "grad_norm": 1.310255765914917,
      "learning_rate": 9.761263157894737e-05,
      "loss": 0.5394,
      "step": 29086
    },
    {
      "epoch": 1.628793817896741,
      "grad_norm": 1.7696998119354248,
      "learning_rate": 9.761236842105263e-05,
      "loss": 0.5722,
      "step": 29087
    },
    {
      "epoch": 1.62884981520887,
      "grad_norm": 1.2077407836914062,
      "learning_rate": 9.76121052631579e-05,
      "loss": 0.3398,
      "step": 29088
    },
    {
      "epoch": 1.628905812520999,
      "grad_norm": 1.19025719165802,
      "learning_rate": 9.761184210526316e-05,
      "loss": 0.4873,
      "step": 29089
    },
    {
      "epoch": 1.628961809833128,
      "grad_norm": 1.1577588319778442,
      "learning_rate": 9.761157894736842e-05,
      "loss": 0.3893,
      "step": 29090
    },
    {
      "epoch": 1.629017807145257,
      "grad_norm": 1.648565649986267,
      "learning_rate": 9.761131578947368e-05,
      "loss": 0.5877,
      "step": 29091
    },
    {
      "epoch": 1.629073804457386,
      "grad_norm": 1.1934082508087158,
      "learning_rate": 9.761105263157894e-05,
      "loss": 0.4315,
      "step": 29092
    },
    {
      "epoch": 1.629129801769515,
      "grad_norm": 1.4596374034881592,
      "learning_rate": 9.761078947368422e-05,
      "loss": 0.5484,
      "step": 29093
    },
    {
      "epoch": 1.629185799081644,
      "grad_norm": 1.3889191150665283,
      "learning_rate": 9.761052631578948e-05,
      "loss": 0.4731,
      "step": 29094
    },
    {
      "epoch": 1.629241796393773,
      "grad_norm": 1.4306739568710327,
      "learning_rate": 9.761026315789475e-05,
      "loss": 0.6296,
      "step": 29095
    },
    {
      "epoch": 1.629297793705902,
      "grad_norm": 1.3662248849868774,
      "learning_rate": 9.761e-05,
      "loss": 0.4153,
      "step": 29096
    },
    {
      "epoch": 1.6293537910180311,
      "grad_norm": 1.737163782119751,
      "learning_rate": 9.760973684210527e-05,
      "loss": 0.5097,
      "step": 29097
    },
    {
      "epoch": 1.6294097883301601,
      "grad_norm": 1.5683296918869019,
      "learning_rate": 9.760947368421053e-05,
      "loss": 0.4128,
      "step": 29098
    },
    {
      "epoch": 1.6294657856422892,
      "grad_norm": 1.4416677951812744,
      "learning_rate": 9.76092105263158e-05,
      "loss": 0.4875,
      "step": 29099
    },
    {
      "epoch": 1.6295217829544182,
      "grad_norm": 1.2504794597625732,
      "learning_rate": 9.760894736842106e-05,
      "loss": 0.3317,
      "step": 29100
    },
    {
      "epoch": 1.6295777802665472,
      "grad_norm": 1.3671859502792358,
      "learning_rate": 9.760868421052632e-05,
      "loss": 0.513,
      "step": 29101
    },
    {
      "epoch": 1.6296337775786762,
      "grad_norm": 1.8179361820220947,
      "learning_rate": 9.760842105263158e-05,
      "loss": 0.5065,
      "step": 29102
    },
    {
      "epoch": 1.6296897748908052,
      "grad_norm": 1.1108883619308472,
      "learning_rate": 9.760815789473685e-05,
      "loss": 0.3864,
      "step": 29103
    },
    {
      "epoch": 1.6297457722029343,
      "grad_norm": 15.233407020568848,
      "learning_rate": 9.760789473684211e-05,
      "loss": 0.3712,
      "step": 29104
    },
    {
      "epoch": 1.6298017695150633,
      "grad_norm": 1.2472684383392334,
      "learning_rate": 9.760763157894737e-05,
      "loss": 0.3965,
      "step": 29105
    },
    {
      "epoch": 1.6298577668271923,
      "grad_norm": 2.189598560333252,
      "learning_rate": 9.760736842105263e-05,
      "loss": 0.559,
      "step": 29106
    },
    {
      "epoch": 1.6299137641393213,
      "grad_norm": 1.4176472425460815,
      "learning_rate": 9.760710526315789e-05,
      "loss": 0.443,
      "step": 29107
    },
    {
      "epoch": 1.6299697614514503,
      "grad_norm": 1.387403964996338,
      "learning_rate": 9.760684210526317e-05,
      "loss": 0.5454,
      "step": 29108
    },
    {
      "epoch": 1.6300257587635794,
      "grad_norm": 1.6350126266479492,
      "learning_rate": 9.760657894736843e-05,
      "loss": 0.4206,
      "step": 29109
    },
    {
      "epoch": 1.6300817560757084,
      "grad_norm": 1.2087743282318115,
      "learning_rate": 9.760631578947369e-05,
      "loss": 0.4521,
      "step": 29110
    },
    {
      "epoch": 1.6301377533878374,
      "grad_norm": 1.3117601871490479,
      "learning_rate": 9.760605263157895e-05,
      "loss": 0.5464,
      "step": 29111
    },
    {
      "epoch": 1.6301937506999664,
      "grad_norm": 1.255249261856079,
      "learning_rate": 9.760578947368422e-05,
      "loss": 0.4812,
      "step": 29112
    },
    {
      "epoch": 1.6302497480120954,
      "grad_norm": 1.740799903869629,
      "learning_rate": 9.760552631578948e-05,
      "loss": 0.4692,
      "step": 29113
    },
    {
      "epoch": 1.6303057453242245,
      "grad_norm": 1.5625931024551392,
      "learning_rate": 9.760526315789474e-05,
      "loss": 0.4163,
      "step": 29114
    },
    {
      "epoch": 1.6303617426363535,
      "grad_norm": 1.2539629936218262,
      "learning_rate": 9.7605e-05,
      "loss": 0.4085,
      "step": 29115
    },
    {
      "epoch": 1.6304177399484825,
      "grad_norm": 1.3470768928527832,
      "learning_rate": 9.760473684210527e-05,
      "loss": 0.4537,
      "step": 29116
    },
    {
      "epoch": 1.6304737372606115,
      "grad_norm": 1.1457774639129639,
      "learning_rate": 9.760447368421053e-05,
      "loss": 0.451,
      "step": 29117
    },
    {
      "epoch": 1.6305297345727405,
      "grad_norm": 1.6563763618469238,
      "learning_rate": 9.76042105263158e-05,
      "loss": 0.6096,
      "step": 29118
    },
    {
      "epoch": 1.6305857318848695,
      "grad_norm": 1.1781911849975586,
      "learning_rate": 9.760394736842105e-05,
      "loss": 0.4572,
      "step": 29119
    },
    {
      "epoch": 1.6306417291969986,
      "grad_norm": 1.3582149744033813,
      "learning_rate": 9.760368421052632e-05,
      "loss": 0.5164,
      "step": 29120
    },
    {
      "epoch": 1.6306977265091276,
      "grad_norm": 1.073562502861023,
      "learning_rate": 9.760342105263158e-05,
      "loss": 0.3497,
      "step": 29121
    },
    {
      "epoch": 1.6307537238212566,
      "grad_norm": 1.3410911560058594,
      "learning_rate": 9.760315789473686e-05,
      "loss": 0.488,
      "step": 29122
    },
    {
      "epoch": 1.6308097211333856,
      "grad_norm": 1.3580961227416992,
      "learning_rate": 9.760289473684212e-05,
      "loss": 0.5321,
      "step": 29123
    },
    {
      "epoch": 1.6308657184455146,
      "grad_norm": 1.290549635887146,
      "learning_rate": 9.760263157894736e-05,
      "loss": 0.5001,
      "step": 29124
    },
    {
      "epoch": 1.6309217157576437,
      "grad_norm": 1.5435512065887451,
      "learning_rate": 9.760236842105264e-05,
      "loss": 0.6849,
      "step": 29125
    },
    {
      "epoch": 1.6309777130697727,
      "grad_norm": 1.1113272905349731,
      "learning_rate": 9.76021052631579e-05,
      "loss": 0.3134,
      "step": 29126
    },
    {
      "epoch": 1.6310337103819017,
      "grad_norm": 1.1967477798461914,
      "learning_rate": 9.760184210526317e-05,
      "loss": 0.3766,
      "step": 29127
    },
    {
      "epoch": 1.6310897076940307,
      "grad_norm": 1.4602841138839722,
      "learning_rate": 9.760157894736842e-05,
      "loss": 0.5771,
      "step": 29128
    },
    {
      "epoch": 1.6311457050061597,
      "grad_norm": 1.3339039087295532,
      "learning_rate": 9.760131578947369e-05,
      "loss": 0.4594,
      "step": 29129
    },
    {
      "epoch": 1.6312017023182888,
      "grad_norm": 1.2210233211517334,
      "learning_rate": 9.760105263157895e-05,
      "loss": 0.4366,
      "step": 29130
    },
    {
      "epoch": 1.6312576996304178,
      "grad_norm": 1.192659616470337,
      "learning_rate": 9.760078947368422e-05,
      "loss": 0.5058,
      "step": 29131
    },
    {
      "epoch": 1.6313136969425468,
      "grad_norm": 1.5911970138549805,
      "learning_rate": 9.760052631578948e-05,
      "loss": 0.4466,
      "step": 29132
    },
    {
      "epoch": 1.6313696942546758,
      "grad_norm": 1.1969664096832275,
      "learning_rate": 9.760026315789474e-05,
      "loss": 0.4689,
      "step": 29133
    },
    {
      "epoch": 1.6314256915668048,
      "grad_norm": 1.3877958059310913,
      "learning_rate": 9.76e-05,
      "loss": 0.3713,
      "step": 29134
    },
    {
      "epoch": 1.6314816888789339,
      "grad_norm": 1.6576367616653442,
      "learning_rate": 9.759973684210527e-05,
      "loss": 0.5546,
      "step": 29135
    },
    {
      "epoch": 1.6315376861910629,
      "grad_norm": 1.4091688394546509,
      "learning_rate": 9.759947368421053e-05,
      "loss": 0.4724,
      "step": 29136
    },
    {
      "epoch": 1.631593683503192,
      "grad_norm": 1.2410529851913452,
      "learning_rate": 9.75992105263158e-05,
      "loss": 0.4976,
      "step": 29137
    },
    {
      "epoch": 1.631649680815321,
      "grad_norm": 1.290068507194519,
      "learning_rate": 9.759894736842105e-05,
      "loss": 0.5175,
      "step": 29138
    },
    {
      "epoch": 1.63170567812745,
      "grad_norm": 1.3230829238891602,
      "learning_rate": 9.759868421052633e-05,
      "loss": 0.357,
      "step": 29139
    },
    {
      "epoch": 1.631761675439579,
      "grad_norm": 1.3626868724822998,
      "learning_rate": 9.759842105263159e-05,
      "loss": 0.6765,
      "step": 29140
    },
    {
      "epoch": 1.631817672751708,
      "grad_norm": 1.5146164894104004,
      "learning_rate": 9.759815789473685e-05,
      "loss": 0.5233,
      "step": 29141
    },
    {
      "epoch": 1.631873670063837,
      "grad_norm": 1.4057866334915161,
      "learning_rate": 9.75978947368421e-05,
      "loss": 0.3804,
      "step": 29142
    },
    {
      "epoch": 1.631929667375966,
      "grad_norm": 1.1456196308135986,
      "learning_rate": 9.759763157894737e-05,
      "loss": 0.4014,
      "step": 29143
    },
    {
      "epoch": 1.631985664688095,
      "grad_norm": 1.3079019784927368,
      "learning_rate": 9.759736842105264e-05,
      "loss": 0.5138,
      "step": 29144
    },
    {
      "epoch": 1.632041662000224,
      "grad_norm": 1.0659583806991577,
      "learning_rate": 9.75971052631579e-05,
      "loss": 0.4265,
      "step": 29145
    },
    {
      "epoch": 1.632097659312353,
      "grad_norm": 1.377334713935852,
      "learning_rate": 9.759684210526316e-05,
      "loss": 0.4598,
      "step": 29146
    },
    {
      "epoch": 1.632153656624482,
      "grad_norm": 1.3103435039520264,
      "learning_rate": 9.759657894736842e-05,
      "loss": 0.3911,
      "step": 29147
    },
    {
      "epoch": 1.632209653936611,
      "grad_norm": 1.2591474056243896,
      "learning_rate": 9.759631578947369e-05,
      "loss": 0.4053,
      "step": 29148
    },
    {
      "epoch": 1.6322656512487401,
      "grad_norm": 1.6633414030075073,
      "learning_rate": 9.759605263157895e-05,
      "loss": 0.4674,
      "step": 29149
    },
    {
      "epoch": 1.6323216485608691,
      "grad_norm": 1.1004396677017212,
      "learning_rate": 9.759578947368422e-05,
      "loss": 0.3968,
      "step": 29150
    },
    {
      "epoch": 1.6323776458729982,
      "grad_norm": 1.4744101762771606,
      "learning_rate": 9.759552631578947e-05,
      "loss": 0.4244,
      "step": 29151
    },
    {
      "epoch": 1.6324336431851272,
      "grad_norm": 1.2575689554214478,
      "learning_rate": 9.759526315789474e-05,
      "loss": 0.4432,
      "step": 29152
    },
    {
      "epoch": 1.6324896404972562,
      "grad_norm": 1.2009389400482178,
      "learning_rate": 9.7595e-05,
      "loss": 0.4098,
      "step": 29153
    },
    {
      "epoch": 1.6325456378093852,
      "grad_norm": 1.4981553554534912,
      "learning_rate": 9.759473684210528e-05,
      "loss": 0.4769,
      "step": 29154
    },
    {
      "epoch": 1.6326016351215142,
      "grad_norm": 1.6977664232254028,
      "learning_rate": 9.759447368421054e-05,
      "loss": 0.4882,
      "step": 29155
    },
    {
      "epoch": 1.6326576324336433,
      "grad_norm": 0.9824967384338379,
      "learning_rate": 9.75942105263158e-05,
      "loss": 0.3928,
      "step": 29156
    },
    {
      "epoch": 1.6327136297457723,
      "grad_norm": 1.3059180974960327,
      "learning_rate": 9.759394736842106e-05,
      "loss": 0.5373,
      "step": 29157
    },
    {
      "epoch": 1.6327696270579013,
      "grad_norm": 1.1779322624206543,
      "learning_rate": 9.759368421052632e-05,
      "loss": 0.4314,
      "step": 29158
    },
    {
      "epoch": 1.6328256243700303,
      "grad_norm": 1.2194240093231201,
      "learning_rate": 9.759342105263159e-05,
      "loss": 0.4762,
      "step": 29159
    },
    {
      "epoch": 1.6328816216821593,
      "grad_norm": 1.175771713256836,
      "learning_rate": 9.759315789473685e-05,
      "loss": 0.4765,
      "step": 29160
    },
    {
      "epoch": 1.6329376189942884,
      "grad_norm": 1.2643389701843262,
      "learning_rate": 9.759289473684211e-05,
      "loss": 0.4175,
      "step": 29161
    },
    {
      "epoch": 1.6329936163064174,
      "grad_norm": 1.2959729433059692,
      "learning_rate": 9.759263157894737e-05,
      "loss": 0.452,
      "step": 29162
    },
    {
      "epoch": 1.6330496136185464,
      "grad_norm": 1.3463983535766602,
      "learning_rate": 9.759236842105264e-05,
      "loss": 0.3897,
      "step": 29163
    },
    {
      "epoch": 1.6331056109306754,
      "grad_norm": 1.2260736227035522,
      "learning_rate": 9.75921052631579e-05,
      "loss": 0.46,
      "step": 29164
    },
    {
      "epoch": 1.6331616082428044,
      "grad_norm": 1.4508298635482788,
      "learning_rate": 9.759184210526316e-05,
      "loss": 0.5266,
      "step": 29165
    },
    {
      "epoch": 1.6332176055549334,
      "grad_norm": 1.2655651569366455,
      "learning_rate": 9.759157894736842e-05,
      "loss": 0.4594,
      "step": 29166
    },
    {
      "epoch": 1.6332736028670625,
      "grad_norm": 1.4045745134353638,
      "learning_rate": 9.75913157894737e-05,
      "loss": 0.7315,
      "step": 29167
    },
    {
      "epoch": 1.6333296001791915,
      "grad_norm": 1.1083928346633911,
      "learning_rate": 9.759105263157895e-05,
      "loss": 0.4991,
      "step": 29168
    },
    {
      "epoch": 1.6333855974913205,
      "grad_norm": 1.4445360898971558,
      "learning_rate": 9.759078947368421e-05,
      "loss": 0.5848,
      "step": 29169
    },
    {
      "epoch": 1.6334415948034495,
      "grad_norm": 1.4294407367706299,
      "learning_rate": 9.759052631578947e-05,
      "loss": 0.5486,
      "step": 29170
    },
    {
      "epoch": 1.6334975921155785,
      "grad_norm": 1.4833860397338867,
      "learning_rate": 9.759026315789475e-05,
      "loss": 0.5012,
      "step": 29171
    },
    {
      "epoch": 1.6335535894277076,
      "grad_norm": 1.4099397659301758,
      "learning_rate": 9.759e-05,
      "loss": 0.3433,
      "step": 29172
    },
    {
      "epoch": 1.6336095867398366,
      "grad_norm": 5.987528324127197,
      "learning_rate": 9.758973684210528e-05,
      "loss": 0.5535,
      "step": 29173
    },
    {
      "epoch": 1.6336655840519656,
      "grad_norm": 1.8035095930099487,
      "learning_rate": 9.758947368421053e-05,
      "loss": 0.4982,
      "step": 29174
    },
    {
      "epoch": 1.6337215813640946,
      "grad_norm": 1.0976383686065674,
      "learning_rate": 9.75892105263158e-05,
      "loss": 0.3637,
      "step": 29175
    },
    {
      "epoch": 1.6337775786762236,
      "grad_norm": 1.1581015586853027,
      "learning_rate": 9.758894736842106e-05,
      "loss": 0.4259,
      "step": 29176
    },
    {
      "epoch": 1.6338335759883527,
      "grad_norm": 1.3488553762435913,
      "learning_rate": 9.758868421052632e-05,
      "loss": 0.409,
      "step": 29177
    },
    {
      "epoch": 1.6338895733004817,
      "grad_norm": 1.1652346849441528,
      "learning_rate": 9.758842105263158e-05,
      "loss": 0.478,
      "step": 29178
    },
    {
      "epoch": 1.6339455706126107,
      "grad_norm": 1.2191673517227173,
      "learning_rate": 9.758815789473684e-05,
      "loss": 0.3877,
      "step": 29179
    },
    {
      "epoch": 1.6340015679247397,
      "grad_norm": 1.2604730129241943,
      "learning_rate": 9.758789473684211e-05,
      "loss": 0.4978,
      "step": 29180
    },
    {
      "epoch": 1.6340575652368687,
      "grad_norm": 1.6360100507736206,
      "learning_rate": 9.758763157894737e-05,
      "loss": 0.3549,
      "step": 29181
    },
    {
      "epoch": 1.6341135625489978,
      "grad_norm": 1.3707400560379028,
      "learning_rate": 9.758736842105264e-05,
      "loss": 0.4374,
      "step": 29182
    },
    {
      "epoch": 1.6341695598611268,
      "grad_norm": 1.3481628894805908,
      "learning_rate": 9.758710526315789e-05,
      "loss": 0.546,
      "step": 29183
    },
    {
      "epoch": 1.6342255571732558,
      "grad_norm": 1.3352290391921997,
      "learning_rate": 9.758684210526316e-05,
      "loss": 0.5488,
      "step": 29184
    },
    {
      "epoch": 1.6342815544853848,
      "grad_norm": 1.357681393623352,
      "learning_rate": 9.758657894736842e-05,
      "loss": 0.4132,
      "step": 29185
    },
    {
      "epoch": 1.6343375517975138,
      "grad_norm": 1.2435659170150757,
      "learning_rate": 9.75863157894737e-05,
      "loss": 0.3325,
      "step": 29186
    },
    {
      "epoch": 1.6343935491096429,
      "grad_norm": 1.1963164806365967,
      "learning_rate": 9.758605263157896e-05,
      "loss": 0.448,
      "step": 29187
    },
    {
      "epoch": 1.6344495464217719,
      "grad_norm": 1.3451859951019287,
      "learning_rate": 9.758578947368422e-05,
      "loss": 0.5199,
      "step": 29188
    },
    {
      "epoch": 1.634505543733901,
      "grad_norm": 1.3353017568588257,
      "learning_rate": 9.758552631578948e-05,
      "loss": 0.6076,
      "step": 29189
    },
    {
      "epoch": 1.63456154104603,
      "grad_norm": 1.2593022584915161,
      "learning_rate": 9.758526315789475e-05,
      "loss": 0.53,
      "step": 29190
    },
    {
      "epoch": 1.634617538358159,
      "grad_norm": 1.5316128730773926,
      "learning_rate": 9.758500000000001e-05,
      "loss": 0.4194,
      "step": 29191
    },
    {
      "epoch": 1.634673535670288,
      "grad_norm": 1.4630426168441772,
      "learning_rate": 9.758473684210527e-05,
      "loss": 0.472,
      "step": 29192
    },
    {
      "epoch": 1.634729532982417,
      "grad_norm": 1.186896800994873,
      "learning_rate": 9.758447368421053e-05,
      "loss": 0.4558,
      "step": 29193
    },
    {
      "epoch": 1.634785530294546,
      "grad_norm": 1.35330331325531,
      "learning_rate": 9.758421052631579e-05,
      "loss": 0.4215,
      "step": 29194
    },
    {
      "epoch": 1.634841527606675,
      "grad_norm": 1.1766036748886108,
      "learning_rate": 9.758394736842106e-05,
      "loss": 0.3758,
      "step": 29195
    },
    {
      "epoch": 1.634897524918804,
      "grad_norm": 2.030928373336792,
      "learning_rate": 9.758368421052632e-05,
      "loss": 0.485,
      "step": 29196
    },
    {
      "epoch": 1.634953522230933,
      "grad_norm": 1.5365872383117676,
      "learning_rate": 9.758342105263158e-05,
      "loss": 0.7332,
      "step": 29197
    },
    {
      "epoch": 1.635009519543062,
      "grad_norm": 1.1595895290374756,
      "learning_rate": 9.758315789473684e-05,
      "loss": 0.3959,
      "step": 29198
    },
    {
      "epoch": 1.635065516855191,
      "grad_norm": 1.3030779361724854,
      "learning_rate": 9.758289473684211e-05,
      "loss": 0.4442,
      "step": 29199
    },
    {
      "epoch": 1.63512151416732,
      "grad_norm": 1.7303502559661865,
      "learning_rate": 9.758263157894737e-05,
      "loss": 0.411,
      "step": 29200
    },
    {
      "epoch": 1.6351775114794491,
      "grad_norm": 1.6569221019744873,
      "learning_rate": 9.758236842105263e-05,
      "loss": 0.6445,
      "step": 29201
    },
    {
      "epoch": 1.6352335087915781,
      "grad_norm": 1.3060539960861206,
      "learning_rate": 9.758210526315789e-05,
      "loss": 0.4981,
      "step": 29202
    },
    {
      "epoch": 1.6352895061037072,
      "grad_norm": 1.2461727857589722,
      "learning_rate": 9.758184210526317e-05,
      "loss": 0.5195,
      "step": 29203
    },
    {
      "epoch": 1.6353455034158362,
      "grad_norm": 1.407566785812378,
      "learning_rate": 9.758157894736843e-05,
      "loss": 0.4992,
      "step": 29204
    },
    {
      "epoch": 1.6354015007279652,
      "grad_norm": 1.8369650840759277,
      "learning_rate": 9.75813157894737e-05,
      "loss": 0.4087,
      "step": 29205
    },
    {
      "epoch": 1.6354574980400942,
      "grad_norm": 1.3014521598815918,
      "learning_rate": 9.758105263157895e-05,
      "loss": 0.4605,
      "step": 29206
    },
    {
      "epoch": 1.6355134953522232,
      "grad_norm": 1.6740087270736694,
      "learning_rate": 9.758078947368422e-05,
      "loss": 0.4178,
      "step": 29207
    },
    {
      "epoch": 1.6355694926643523,
      "grad_norm": 1.2325019836425781,
      "learning_rate": 9.758052631578948e-05,
      "loss": 0.405,
      "step": 29208
    },
    {
      "epoch": 1.6356254899764813,
      "grad_norm": 1.3273875713348389,
      "learning_rate": 9.758026315789475e-05,
      "loss": 0.3583,
      "step": 29209
    },
    {
      "epoch": 1.6356814872886103,
      "grad_norm": 1.1015675067901611,
      "learning_rate": 9.758000000000001e-05,
      "loss": 0.3004,
      "step": 29210
    },
    {
      "epoch": 1.6357374846007393,
      "grad_norm": 1.7572027444839478,
      "learning_rate": 9.757973684210526e-05,
      "loss": 0.5247,
      "step": 29211
    },
    {
      "epoch": 1.6357934819128683,
      "grad_norm": 1.3972452878952026,
      "learning_rate": 9.757947368421053e-05,
      "loss": 0.4335,
      "step": 29212
    },
    {
      "epoch": 1.6358494792249973,
      "grad_norm": 1.4027369022369385,
      "learning_rate": 9.757921052631579e-05,
      "loss": 0.4834,
      "step": 29213
    },
    {
      "epoch": 1.6359054765371264,
      "grad_norm": 1.4487584829330444,
      "learning_rate": 9.757894736842106e-05,
      "loss": 0.4702,
      "step": 29214
    },
    {
      "epoch": 1.6359614738492554,
      "grad_norm": 1.4246277809143066,
      "learning_rate": 9.757868421052632e-05,
      "loss": 0.4236,
      "step": 29215
    },
    {
      "epoch": 1.6360174711613844,
      "grad_norm": 1.6195552349090576,
      "learning_rate": 9.757842105263158e-05,
      "loss": 0.407,
      "step": 29216
    },
    {
      "epoch": 1.6360734684735134,
      "grad_norm": 1.6442829370498657,
      "learning_rate": 9.757815789473684e-05,
      "loss": 0.4091,
      "step": 29217
    },
    {
      "epoch": 1.6361294657856424,
      "grad_norm": 1.4494110345840454,
      "learning_rate": 9.757789473684212e-05,
      "loss": 0.4088,
      "step": 29218
    },
    {
      "epoch": 1.6361854630977715,
      "grad_norm": 1.286902666091919,
      "learning_rate": 9.757763157894738e-05,
      "loss": 0.4466,
      "step": 29219
    },
    {
      "epoch": 1.6362414604099005,
      "grad_norm": 1.5643664598464966,
      "learning_rate": 9.757736842105264e-05,
      "loss": 0.6074,
      "step": 29220
    },
    {
      "epoch": 1.6362974577220295,
      "grad_norm": 1.476778507232666,
      "learning_rate": 9.75771052631579e-05,
      "loss": 0.4695,
      "step": 29221
    },
    {
      "epoch": 1.6363534550341585,
      "grad_norm": 1.1702582836151123,
      "learning_rate": 9.757684210526317e-05,
      "loss": 0.4855,
      "step": 29222
    },
    {
      "epoch": 1.6364094523462875,
      "grad_norm": 1.292647123336792,
      "learning_rate": 9.757657894736843e-05,
      "loss": 0.4206,
      "step": 29223
    },
    {
      "epoch": 1.6364654496584166,
      "grad_norm": 1.072305679321289,
      "learning_rate": 9.757631578947369e-05,
      "loss": 0.4266,
      "step": 29224
    },
    {
      "epoch": 1.6365214469705456,
      "grad_norm": 2.024962902069092,
      "learning_rate": 9.757605263157895e-05,
      "loss": 0.5598,
      "step": 29225
    },
    {
      "epoch": 1.6365774442826744,
      "grad_norm": 1.2907689809799194,
      "learning_rate": 9.757578947368422e-05,
      "loss": 0.4269,
      "step": 29226
    },
    {
      "epoch": 1.6366334415948034,
      "grad_norm": 1.6650328636169434,
      "learning_rate": 9.757552631578948e-05,
      "loss": 0.635,
      "step": 29227
    },
    {
      "epoch": 1.6366894389069324,
      "grad_norm": 1.3826417922973633,
      "learning_rate": 9.757526315789474e-05,
      "loss": 0.4345,
      "step": 29228
    },
    {
      "epoch": 1.6367454362190614,
      "grad_norm": 2.3610739707946777,
      "learning_rate": 9.7575e-05,
      "loss": 0.5797,
      "step": 29229
    },
    {
      "epoch": 1.6368014335311905,
      "grad_norm": 1.8455467224121094,
      "learning_rate": 9.757473684210526e-05,
      "loss": 0.7451,
      "step": 29230
    },
    {
      "epoch": 1.6368574308433195,
      "grad_norm": 1.270052433013916,
      "learning_rate": 9.757447368421053e-05,
      "loss": 0.4652,
      "step": 29231
    },
    {
      "epoch": 1.6369134281554485,
      "grad_norm": 1.677120566368103,
      "learning_rate": 9.757421052631579e-05,
      "loss": 0.4804,
      "step": 29232
    },
    {
      "epoch": 1.6369694254675775,
      "grad_norm": 1.7527693510055542,
      "learning_rate": 9.757394736842105e-05,
      "loss": 0.3979,
      "step": 29233
    },
    {
      "epoch": 1.6370254227797065,
      "grad_norm": 1.2408288717269897,
      "learning_rate": 9.757368421052631e-05,
      "loss": 0.4502,
      "step": 29234
    },
    {
      "epoch": 1.6370814200918355,
      "grad_norm": 1.2545652389526367,
      "learning_rate": 9.757342105263159e-05,
      "loss": 0.4549,
      "step": 29235
    },
    {
      "epoch": 1.6371374174039646,
      "grad_norm": 1.446791648864746,
      "learning_rate": 9.757315789473685e-05,
      "loss": 0.5472,
      "step": 29236
    },
    {
      "epoch": 1.6371934147160936,
      "grad_norm": 1.5149366855621338,
      "learning_rate": 9.757289473684212e-05,
      "loss": 0.4647,
      "step": 29237
    },
    {
      "epoch": 1.6372494120282226,
      "grad_norm": 1.4112406969070435,
      "learning_rate": 9.757263157894736e-05,
      "loss": 0.5926,
      "step": 29238
    },
    {
      "epoch": 1.6373054093403516,
      "grad_norm": 1.5062649250030518,
      "learning_rate": 9.757236842105264e-05,
      "loss": 0.3953,
      "step": 29239
    },
    {
      "epoch": 1.6373614066524806,
      "grad_norm": 1.1949267387390137,
      "learning_rate": 9.75721052631579e-05,
      "loss": 0.3226,
      "step": 29240
    },
    {
      "epoch": 1.6374174039646097,
      "grad_norm": 1.3616310358047485,
      "learning_rate": 9.757184210526317e-05,
      "loss": 0.5263,
      "step": 29241
    },
    {
      "epoch": 1.6374734012767387,
      "grad_norm": 1.2105578184127808,
      "learning_rate": 9.757157894736843e-05,
      "loss": 0.4264,
      "step": 29242
    },
    {
      "epoch": 1.6375293985888677,
      "grad_norm": 1.3974268436431885,
      "learning_rate": 9.757131578947369e-05,
      "loss": 0.5389,
      "step": 29243
    },
    {
      "epoch": 1.6375853959009967,
      "grad_norm": 1.3129873275756836,
      "learning_rate": 9.757105263157895e-05,
      "loss": 0.4387,
      "step": 29244
    },
    {
      "epoch": 1.6376413932131257,
      "grad_norm": 1.1442760229110718,
      "learning_rate": 9.757078947368421e-05,
      "loss": 0.4952,
      "step": 29245
    },
    {
      "epoch": 1.6376973905252548,
      "grad_norm": 1.4596924781799316,
      "learning_rate": 9.757052631578948e-05,
      "loss": 0.4531,
      "step": 29246
    },
    {
      "epoch": 1.6377533878373838,
      "grad_norm": 1.195422649383545,
      "learning_rate": 9.757026315789474e-05,
      "loss": 0.4395,
      "step": 29247
    },
    {
      "epoch": 1.6378093851495128,
      "grad_norm": 1.3700623512268066,
      "learning_rate": 9.757e-05,
      "loss": 0.3675,
      "step": 29248
    },
    {
      "epoch": 1.6378653824616418,
      "grad_norm": 1.3969826698303223,
      "learning_rate": 9.756973684210526e-05,
      "loss": 0.4904,
      "step": 29249
    },
    {
      "epoch": 1.6379213797737708,
      "grad_norm": 1.2775415182113647,
      "learning_rate": 9.756947368421054e-05,
      "loss": 0.528,
      "step": 29250
    },
    {
      "epoch": 1.6379773770858999,
      "grad_norm": 1.1494224071502686,
      "learning_rate": 9.75692105263158e-05,
      "loss": 0.3905,
      "step": 29251
    },
    {
      "epoch": 1.6380333743980289,
      "grad_norm": 1.369499683380127,
      "learning_rate": 9.756894736842106e-05,
      "loss": 0.5188,
      "step": 29252
    },
    {
      "epoch": 1.638089371710158,
      "grad_norm": 1.2361911535263062,
      "learning_rate": 9.756868421052631e-05,
      "loss": 0.6881,
      "step": 29253
    },
    {
      "epoch": 1.638145369022287,
      "grad_norm": 1.3087455034255981,
      "learning_rate": 9.756842105263159e-05,
      "loss": 0.5997,
      "step": 29254
    },
    {
      "epoch": 1.638201366334416,
      "grad_norm": 1.2577117681503296,
      "learning_rate": 9.756815789473685e-05,
      "loss": 0.429,
      "step": 29255
    },
    {
      "epoch": 1.638257363646545,
      "grad_norm": 1.2478843927383423,
      "learning_rate": 9.756789473684211e-05,
      "loss": 0.5154,
      "step": 29256
    },
    {
      "epoch": 1.638313360958674,
      "grad_norm": 1.2429277896881104,
      "learning_rate": 9.756763157894737e-05,
      "loss": 0.4897,
      "step": 29257
    },
    {
      "epoch": 1.638369358270803,
      "grad_norm": 1.3642586469650269,
      "learning_rate": 9.756736842105264e-05,
      "loss": 0.4276,
      "step": 29258
    },
    {
      "epoch": 1.638425355582932,
      "grad_norm": 1.3517820835113525,
      "learning_rate": 9.75671052631579e-05,
      "loss": 0.4568,
      "step": 29259
    },
    {
      "epoch": 1.638481352895061,
      "grad_norm": 1.2205671072006226,
      "learning_rate": 9.756684210526317e-05,
      "loss": 0.4527,
      "step": 29260
    },
    {
      "epoch": 1.63853735020719,
      "grad_norm": 1.2526224851608276,
      "learning_rate": 9.756657894736842e-05,
      "loss": 0.5031,
      "step": 29261
    },
    {
      "epoch": 1.638593347519319,
      "grad_norm": 1.354042649269104,
      "learning_rate": 9.756631578947368e-05,
      "loss": 0.3161,
      "step": 29262
    },
    {
      "epoch": 1.638649344831448,
      "grad_norm": 1.4771968126296997,
      "learning_rate": 9.756605263157895e-05,
      "loss": 0.5773,
      "step": 29263
    },
    {
      "epoch": 1.638705342143577,
      "grad_norm": 1.1587164402008057,
      "learning_rate": 9.756578947368421e-05,
      "loss": 0.3288,
      "step": 29264
    },
    {
      "epoch": 1.6387613394557061,
      "grad_norm": 1.5697356462478638,
      "learning_rate": 9.756552631578949e-05,
      "loss": 0.5518,
      "step": 29265
    },
    {
      "epoch": 1.6388173367678351,
      "grad_norm": 1.3110226392745972,
      "learning_rate": 9.756526315789473e-05,
      "loss": 0.4752,
      "step": 29266
    },
    {
      "epoch": 1.6388733340799642,
      "grad_norm": 1.389346718788147,
      "learning_rate": 9.7565e-05,
      "loss": 0.5271,
      "step": 29267
    },
    {
      "epoch": 1.6389293313920932,
      "grad_norm": 1.3497350215911865,
      "learning_rate": 9.756473684210526e-05,
      "loss": 0.341,
      "step": 29268
    },
    {
      "epoch": 1.6389853287042222,
      "grad_norm": 1.2732964754104614,
      "learning_rate": 9.756447368421054e-05,
      "loss": 0.4535,
      "step": 29269
    },
    {
      "epoch": 1.6390413260163512,
      "grad_norm": 1.4268783330917358,
      "learning_rate": 9.75642105263158e-05,
      "loss": 0.5771,
      "step": 29270
    },
    {
      "epoch": 1.6390973233284802,
      "grad_norm": 1.372564435005188,
      "learning_rate": 9.756394736842106e-05,
      "loss": 0.4146,
      "step": 29271
    },
    {
      "epoch": 1.6391533206406093,
      "grad_norm": 1.3734673261642456,
      "learning_rate": 9.756368421052632e-05,
      "loss": 0.4694,
      "step": 29272
    },
    {
      "epoch": 1.6392093179527383,
      "grad_norm": 1.1530158519744873,
      "learning_rate": 9.756342105263159e-05,
      "loss": 0.4468,
      "step": 29273
    },
    {
      "epoch": 1.6392653152648673,
      "grad_norm": 9.442240715026855,
      "learning_rate": 9.756315789473685e-05,
      "loss": 0.4221,
      "step": 29274
    },
    {
      "epoch": 1.6393213125769963,
      "grad_norm": 1.2248588800430298,
      "learning_rate": 9.756289473684211e-05,
      "loss": 0.4177,
      "step": 29275
    },
    {
      "epoch": 1.6393773098891253,
      "grad_norm": 1.0724364519119263,
      "learning_rate": 9.756263157894737e-05,
      "loss": 0.2904,
      "step": 29276
    },
    {
      "epoch": 1.6394333072012544,
      "grad_norm": 1.2486066818237305,
      "learning_rate": 9.756236842105264e-05,
      "loss": 0.4967,
      "step": 29277
    },
    {
      "epoch": 1.6394893045133834,
      "grad_norm": 1.1921827793121338,
      "learning_rate": 9.75621052631579e-05,
      "loss": 0.4251,
      "step": 29278
    },
    {
      "epoch": 1.6395453018255124,
      "grad_norm": 1.30080246925354,
      "learning_rate": 9.756184210526316e-05,
      "loss": 0.4643,
      "step": 29279
    },
    {
      "epoch": 1.6396012991376414,
      "grad_norm": 1.4581222534179688,
      "learning_rate": 9.756157894736842e-05,
      "loss": 0.5637,
      "step": 29280
    },
    {
      "epoch": 1.6396572964497704,
      "grad_norm": 1.3218766450881958,
      "learning_rate": 9.756131578947368e-05,
      "loss": 0.416,
      "step": 29281
    },
    {
      "epoch": 1.6397132937618994,
      "grad_norm": 1.4028072357177734,
      "learning_rate": 9.756105263157896e-05,
      "loss": 0.4125,
      "step": 29282
    },
    {
      "epoch": 1.6397692910740285,
      "grad_norm": 1.1699087619781494,
      "learning_rate": 9.756078947368422e-05,
      "loss": 0.3245,
      "step": 29283
    },
    {
      "epoch": 1.6398252883861575,
      "grad_norm": 1.3882259130477905,
      "learning_rate": 9.756052631578947e-05,
      "loss": 0.456,
      "step": 29284
    },
    {
      "epoch": 1.6398812856982865,
      "grad_norm": 1.126474142074585,
      "learning_rate": 9.756026315789473e-05,
      "loss": 0.3478,
      "step": 29285
    },
    {
      "epoch": 1.6399372830104155,
      "grad_norm": 1.2113491296768188,
      "learning_rate": 9.756000000000001e-05,
      "loss": 0.4296,
      "step": 29286
    },
    {
      "epoch": 1.6399932803225445,
      "grad_norm": 1.5506101846694946,
      "learning_rate": 9.755973684210527e-05,
      "loss": 0.5387,
      "step": 29287
    },
    {
      "epoch": 1.6400492776346736,
      "grad_norm": 1.283271074295044,
      "learning_rate": 9.755947368421053e-05,
      "loss": 0.5906,
      "step": 29288
    },
    {
      "epoch": 1.6401052749468026,
      "grad_norm": 1.549760103225708,
      "learning_rate": 9.755921052631579e-05,
      "loss": 0.7714,
      "step": 29289
    },
    {
      "epoch": 1.6401612722589316,
      "grad_norm": 1.3226968050003052,
      "learning_rate": 9.755894736842106e-05,
      "loss": 0.3748,
      "step": 29290
    },
    {
      "epoch": 1.6402172695710606,
      "grad_norm": 1.3081238269805908,
      "learning_rate": 9.755868421052632e-05,
      "loss": 0.409,
      "step": 29291
    },
    {
      "epoch": 1.6402732668831896,
      "grad_norm": 1.3023624420166016,
      "learning_rate": 9.75584210526316e-05,
      "loss": 0.5124,
      "step": 29292
    },
    {
      "epoch": 1.6403292641953187,
      "grad_norm": 1.170420527458191,
      "learning_rate": 9.755815789473684e-05,
      "loss": 0.4215,
      "step": 29293
    },
    {
      "epoch": 1.6403852615074477,
      "grad_norm": 1.242762804031372,
      "learning_rate": 9.755789473684211e-05,
      "loss": 0.4159,
      "step": 29294
    },
    {
      "epoch": 1.6404412588195767,
      "grad_norm": 1.4003139734268188,
      "learning_rate": 9.755763157894737e-05,
      "loss": 0.4027,
      "step": 29295
    },
    {
      "epoch": 1.6404972561317057,
      "grad_norm": 1.252991795539856,
      "learning_rate": 9.755736842105265e-05,
      "loss": 0.4914,
      "step": 29296
    },
    {
      "epoch": 1.6405532534438347,
      "grad_norm": 1.2703192234039307,
      "learning_rate": 9.75571052631579e-05,
      "loss": 0.4134,
      "step": 29297
    },
    {
      "epoch": 1.6406092507559638,
      "grad_norm": 1.4573822021484375,
      "learning_rate": 9.755684210526315e-05,
      "loss": 0.4338,
      "step": 29298
    },
    {
      "epoch": 1.6406652480680928,
      "grad_norm": 1.3850125074386597,
      "learning_rate": 9.755657894736842e-05,
      "loss": 0.4316,
      "step": 29299
    },
    {
      "epoch": 1.6407212453802218,
      "grad_norm": 1.1426774263381958,
      "learning_rate": 9.755631578947368e-05,
      "loss": 0.4613,
      "step": 29300
    },
    {
      "epoch": 1.6407772426923508,
      "grad_norm": 1.572794795036316,
      "learning_rate": 9.755605263157896e-05,
      "loss": 0.5873,
      "step": 29301
    },
    {
      "epoch": 1.6408332400044798,
      "grad_norm": 1.3179900646209717,
      "learning_rate": 9.755578947368422e-05,
      "loss": 0.4128,
      "step": 29302
    },
    {
      "epoch": 1.6408892373166089,
      "grad_norm": 1.2013174295425415,
      "learning_rate": 9.755552631578948e-05,
      "loss": 0.4532,
      "step": 29303
    },
    {
      "epoch": 1.6409452346287379,
      "grad_norm": 1.488812804222107,
      "learning_rate": 9.755526315789474e-05,
      "loss": 0.502,
      "step": 29304
    },
    {
      "epoch": 1.641001231940867,
      "grad_norm": 1.9621952772140503,
      "learning_rate": 9.755500000000001e-05,
      "loss": 0.434,
      "step": 29305
    },
    {
      "epoch": 1.641057229252996,
      "grad_norm": 1.46525239944458,
      "learning_rate": 9.755473684210527e-05,
      "loss": 0.5215,
      "step": 29306
    },
    {
      "epoch": 1.641113226565125,
      "grad_norm": 1.2031933069229126,
      "learning_rate": 9.755447368421053e-05,
      "loss": 0.373,
      "step": 29307
    },
    {
      "epoch": 1.641169223877254,
      "grad_norm": 1.5745117664337158,
      "learning_rate": 9.755421052631579e-05,
      "loss": 0.6136,
      "step": 29308
    },
    {
      "epoch": 1.6412252211893827,
      "grad_norm": 1.5272430181503296,
      "learning_rate": 9.755394736842106e-05,
      "loss": 0.5901,
      "step": 29309
    },
    {
      "epoch": 1.6412812185015118,
      "grad_norm": 1.1108843088150024,
      "learning_rate": 9.755368421052632e-05,
      "loss": 0.3861,
      "step": 29310
    },
    {
      "epoch": 1.6413372158136408,
      "grad_norm": 1.524139404296875,
      "learning_rate": 9.755342105263158e-05,
      "loss": 0.5559,
      "step": 29311
    },
    {
      "epoch": 1.6413932131257698,
      "grad_norm": 1.4420135021209717,
      "learning_rate": 9.755315789473684e-05,
      "loss": 0.4656,
      "step": 29312
    },
    {
      "epoch": 1.6414492104378988,
      "grad_norm": 1.3306641578674316,
      "learning_rate": 9.755289473684212e-05,
      "loss": 0.4025,
      "step": 29313
    },
    {
      "epoch": 1.6415052077500278,
      "grad_norm": 1.4485255479812622,
      "learning_rate": 9.755263157894738e-05,
      "loss": 0.4637,
      "step": 29314
    },
    {
      "epoch": 1.6415612050621569,
      "grad_norm": 1.257196307182312,
      "learning_rate": 9.755236842105263e-05,
      "loss": 0.4802,
      "step": 29315
    },
    {
      "epoch": 1.6416172023742859,
      "grad_norm": 1.5029897689819336,
      "learning_rate": 9.75521052631579e-05,
      "loss": 0.4612,
      "step": 29316
    },
    {
      "epoch": 1.641673199686415,
      "grad_norm": 1.9011369943618774,
      "learning_rate": 9.755184210526315e-05,
      "loss": 0.6143,
      "step": 29317
    },
    {
      "epoch": 1.641729196998544,
      "grad_norm": 1.32305908203125,
      "learning_rate": 9.755157894736843e-05,
      "loss": 0.4435,
      "step": 29318
    },
    {
      "epoch": 1.641785194310673,
      "grad_norm": 0.9825717806816101,
      "learning_rate": 9.755131578947369e-05,
      "loss": 0.3063,
      "step": 29319
    },
    {
      "epoch": 1.641841191622802,
      "grad_norm": 1.2988083362579346,
      "learning_rate": 9.755105263157896e-05,
      "loss": 0.3966,
      "step": 29320
    },
    {
      "epoch": 1.641897188934931,
      "grad_norm": 1.243069052696228,
      "learning_rate": 9.75507894736842e-05,
      "loss": 0.4338,
      "step": 29321
    },
    {
      "epoch": 1.64195318624706,
      "grad_norm": 1.6195769309997559,
      "learning_rate": 9.755052631578948e-05,
      "loss": 0.4953,
      "step": 29322
    },
    {
      "epoch": 1.642009183559189,
      "grad_norm": 1.1538461446762085,
      "learning_rate": 9.755026315789474e-05,
      "loss": 0.4261,
      "step": 29323
    },
    {
      "epoch": 1.642065180871318,
      "grad_norm": 1.189353346824646,
      "learning_rate": 9.755000000000001e-05,
      "loss": 0.3824,
      "step": 29324
    },
    {
      "epoch": 1.642121178183447,
      "grad_norm": 1.3259837627410889,
      "learning_rate": 9.754973684210527e-05,
      "loss": 0.4563,
      "step": 29325
    },
    {
      "epoch": 1.642177175495576,
      "grad_norm": 1.227063775062561,
      "learning_rate": 9.754947368421053e-05,
      "loss": 0.3955,
      "step": 29326
    },
    {
      "epoch": 1.642233172807705,
      "grad_norm": 1.155016303062439,
      "learning_rate": 9.754921052631579e-05,
      "loss": 0.3072,
      "step": 29327
    },
    {
      "epoch": 1.642289170119834,
      "grad_norm": 1.4582222700119019,
      "learning_rate": 9.754894736842107e-05,
      "loss": 0.5228,
      "step": 29328
    },
    {
      "epoch": 1.6423451674319631,
      "grad_norm": 1.249298095703125,
      "learning_rate": 9.754868421052633e-05,
      "loss": 0.3776,
      "step": 29329
    },
    {
      "epoch": 1.6424011647440921,
      "grad_norm": 1.5514057874679565,
      "learning_rate": 9.754842105263158e-05,
      "loss": 0.4551,
      "step": 29330
    },
    {
      "epoch": 1.6424571620562212,
      "grad_norm": 1.3318390846252441,
      "learning_rate": 9.754815789473684e-05,
      "loss": 0.5298,
      "step": 29331
    },
    {
      "epoch": 1.6425131593683502,
      "grad_norm": 1.2294528484344482,
      "learning_rate": 9.75478947368421e-05,
      "loss": 0.4337,
      "step": 29332
    },
    {
      "epoch": 1.6425691566804792,
      "grad_norm": 1.2888906002044678,
      "learning_rate": 9.754763157894738e-05,
      "loss": 0.4862,
      "step": 29333
    },
    {
      "epoch": 1.6426251539926082,
      "grad_norm": 1.1327520608901978,
      "learning_rate": 9.754736842105264e-05,
      "loss": 0.4729,
      "step": 29334
    },
    {
      "epoch": 1.6426811513047372,
      "grad_norm": 1.1546499729156494,
      "learning_rate": 9.75471052631579e-05,
      "loss": 0.4146,
      "step": 29335
    },
    {
      "epoch": 1.6427371486168663,
      "grad_norm": 1.6673691272735596,
      "learning_rate": 9.754684210526316e-05,
      "loss": 0.6231,
      "step": 29336
    },
    {
      "epoch": 1.6427931459289953,
      "grad_norm": 1.2196277379989624,
      "learning_rate": 9.754657894736843e-05,
      "loss": 0.4973,
      "step": 29337
    },
    {
      "epoch": 1.6428491432411243,
      "grad_norm": 1.2602683305740356,
      "learning_rate": 9.754631578947369e-05,
      "loss": 0.5665,
      "step": 29338
    },
    {
      "epoch": 1.6429051405532533,
      "grad_norm": 1.1988778114318848,
      "learning_rate": 9.754605263157895e-05,
      "loss": 0.4264,
      "step": 29339
    },
    {
      "epoch": 1.6429611378653823,
      "grad_norm": 1.1710362434387207,
      "learning_rate": 9.754578947368421e-05,
      "loss": 0.3061,
      "step": 29340
    },
    {
      "epoch": 1.6430171351775114,
      "grad_norm": 1.3546621799468994,
      "learning_rate": 9.754552631578948e-05,
      "loss": 0.6396,
      "step": 29341
    },
    {
      "epoch": 1.6430731324896404,
      "grad_norm": 1.3007144927978516,
      "learning_rate": 9.754526315789474e-05,
      "loss": 0.435,
      "step": 29342
    },
    {
      "epoch": 1.6431291298017694,
      "grad_norm": 1.340864896774292,
      "learning_rate": 9.7545e-05,
      "loss": 0.4531,
      "step": 29343
    },
    {
      "epoch": 1.6431851271138984,
      "grad_norm": 1.3586357831954956,
      "learning_rate": 9.754473684210526e-05,
      "loss": 0.4163,
      "step": 29344
    },
    {
      "epoch": 1.6432411244260274,
      "grad_norm": 1.363587498664856,
      "learning_rate": 9.754447368421054e-05,
      "loss": 0.4542,
      "step": 29345
    },
    {
      "epoch": 1.6432971217381565,
      "grad_norm": 1.5507924556732178,
      "learning_rate": 9.75442105263158e-05,
      "loss": 0.4726,
      "step": 29346
    },
    {
      "epoch": 1.6433531190502855,
      "grad_norm": 1.2276865243911743,
      "learning_rate": 9.754394736842107e-05,
      "loss": 0.5614,
      "step": 29347
    },
    {
      "epoch": 1.6434091163624145,
      "grad_norm": 1.294532060623169,
      "learning_rate": 9.754368421052631e-05,
      "loss": 0.3719,
      "step": 29348
    },
    {
      "epoch": 1.6434651136745435,
      "grad_norm": 1.558737874031067,
      "learning_rate": 9.754342105263157e-05,
      "loss": 0.5353,
      "step": 29349
    },
    {
      "epoch": 1.6435211109866725,
      "grad_norm": 1.4003148078918457,
      "learning_rate": 9.754315789473685e-05,
      "loss": 0.5072,
      "step": 29350
    },
    {
      "epoch": 1.6435771082988015,
      "grad_norm": 1.0688210725784302,
      "learning_rate": 9.754289473684211e-05,
      "loss": 0.3453,
      "step": 29351
    },
    {
      "epoch": 1.6436331056109306,
      "grad_norm": 1.2977280616760254,
      "learning_rate": 9.754263157894738e-05,
      "loss": 0.4658,
      "step": 29352
    },
    {
      "epoch": 1.6436891029230596,
      "grad_norm": 1.1841076612472534,
      "learning_rate": 9.754236842105263e-05,
      "loss": 0.4808,
      "step": 29353
    },
    {
      "epoch": 1.6437451002351886,
      "grad_norm": 1.3413748741149902,
      "learning_rate": 9.75421052631579e-05,
      "loss": 0.5303,
      "step": 29354
    },
    {
      "epoch": 1.6438010975473176,
      "grad_norm": 1.2622005939483643,
      "learning_rate": 9.754184210526316e-05,
      "loss": 0.4256,
      "step": 29355
    },
    {
      "epoch": 1.6438570948594466,
      "grad_norm": 1.2882167100906372,
      "learning_rate": 9.754157894736843e-05,
      "loss": 0.4495,
      "step": 29356
    },
    {
      "epoch": 1.6439130921715757,
      "grad_norm": 1.2747324705123901,
      "learning_rate": 9.754131578947369e-05,
      "loss": 0.4351,
      "step": 29357
    },
    {
      "epoch": 1.6439690894837047,
      "grad_norm": 1.3000568151474,
      "learning_rate": 9.754105263157895e-05,
      "loss": 0.6083,
      "step": 29358
    },
    {
      "epoch": 1.6440250867958337,
      "grad_norm": 1.176451325416565,
      "learning_rate": 9.754078947368421e-05,
      "loss": 0.463,
      "step": 29359
    },
    {
      "epoch": 1.6440810841079627,
      "grad_norm": 1.1218034029006958,
      "learning_rate": 9.754052631578949e-05,
      "loss": 0.3896,
      "step": 29360
    },
    {
      "epoch": 1.6441370814200917,
      "grad_norm": 1.5688334703445435,
      "learning_rate": 9.754026315789474e-05,
      "loss": 0.4835,
      "step": 29361
    },
    {
      "epoch": 1.6441930787322208,
      "grad_norm": 1.240452527999878,
      "learning_rate": 9.754e-05,
      "loss": 0.359,
      "step": 29362
    },
    {
      "epoch": 1.6442490760443498,
      "grad_norm": 1.2389189004898071,
      "learning_rate": 9.753973684210526e-05,
      "loss": 0.6237,
      "step": 29363
    },
    {
      "epoch": 1.6443050733564788,
      "grad_norm": 1.3271735906600952,
      "learning_rate": 9.753947368421054e-05,
      "loss": 0.4437,
      "step": 29364
    },
    {
      "epoch": 1.6443610706686078,
      "grad_norm": 1.3391735553741455,
      "learning_rate": 9.75392105263158e-05,
      "loss": 0.433,
      "step": 29365
    },
    {
      "epoch": 1.6444170679807368,
      "grad_norm": 1.1607493162155151,
      "learning_rate": 9.753894736842106e-05,
      "loss": 0.3328,
      "step": 29366
    },
    {
      "epoch": 1.6444730652928659,
      "grad_norm": 1.331188678741455,
      "learning_rate": 9.753868421052632e-05,
      "loss": 0.5072,
      "step": 29367
    },
    {
      "epoch": 1.6445290626049949,
      "grad_norm": 1.3958109617233276,
      "learning_rate": 9.753842105263158e-05,
      "loss": 0.4293,
      "step": 29368
    },
    {
      "epoch": 1.644585059917124,
      "grad_norm": 1.2311660051345825,
      "learning_rate": 9.753815789473685e-05,
      "loss": 0.4149,
      "step": 29369
    },
    {
      "epoch": 1.644641057229253,
      "grad_norm": 1.254151463508606,
      "learning_rate": 9.753789473684211e-05,
      "loss": 0.3842,
      "step": 29370
    },
    {
      "epoch": 1.644697054541382,
      "grad_norm": 1.3861286640167236,
      "learning_rate": 9.753763157894737e-05,
      "loss": 0.435,
      "step": 29371
    },
    {
      "epoch": 1.644753051853511,
      "grad_norm": 1.5580637454986572,
      "learning_rate": 9.753736842105263e-05,
      "loss": 0.5089,
      "step": 29372
    },
    {
      "epoch": 1.64480904916564,
      "grad_norm": 1.239936113357544,
      "learning_rate": 9.75371052631579e-05,
      "loss": 0.3747,
      "step": 29373
    },
    {
      "epoch": 1.644865046477769,
      "grad_norm": 1.2191531658172607,
      "learning_rate": 9.753684210526316e-05,
      "loss": 0.408,
      "step": 29374
    },
    {
      "epoch": 1.644921043789898,
      "grad_norm": 1.3743925094604492,
      "learning_rate": 9.753657894736844e-05,
      "loss": 0.4911,
      "step": 29375
    },
    {
      "epoch": 1.644977041102027,
      "grad_norm": 1.1639271974563599,
      "learning_rate": 9.753631578947368e-05,
      "loss": 0.3753,
      "step": 29376
    },
    {
      "epoch": 1.645033038414156,
      "grad_norm": 1.3580440282821655,
      "learning_rate": 9.753605263157895e-05,
      "loss": 0.4591,
      "step": 29377
    },
    {
      "epoch": 1.645089035726285,
      "grad_norm": 1.3186320066452026,
      "learning_rate": 9.753578947368421e-05,
      "loss": 0.5989,
      "step": 29378
    },
    {
      "epoch": 1.645145033038414,
      "grad_norm": 1.604811191558838,
      "learning_rate": 9.753552631578949e-05,
      "loss": 0.6839,
      "step": 29379
    },
    {
      "epoch": 1.645201030350543,
      "grad_norm": 9.71956729888916,
      "learning_rate": 9.753526315789473e-05,
      "loss": 0.5981,
      "step": 29380
    },
    {
      "epoch": 1.6452570276626721,
      "grad_norm": 1.2815930843353271,
      "learning_rate": 9.753500000000001e-05,
      "loss": 0.3662,
      "step": 29381
    },
    {
      "epoch": 1.6453130249748011,
      "grad_norm": 1.2173564434051514,
      "learning_rate": 9.753473684210527e-05,
      "loss": 0.3752,
      "step": 29382
    },
    {
      "epoch": 1.6453690222869302,
      "grad_norm": 1.4138680696487427,
      "learning_rate": 9.753447368421053e-05,
      "loss": 0.3849,
      "step": 29383
    },
    {
      "epoch": 1.6454250195990592,
      "grad_norm": 1.145095944404602,
      "learning_rate": 9.75342105263158e-05,
      "loss": 0.3243,
      "step": 29384
    },
    {
      "epoch": 1.6454810169111882,
      "grad_norm": 1.850839614868164,
      "learning_rate": 9.753394736842105e-05,
      "loss": 0.5586,
      "step": 29385
    },
    {
      "epoch": 1.6455370142233172,
      "grad_norm": 1.1812289953231812,
      "learning_rate": 9.753368421052632e-05,
      "loss": 0.3993,
      "step": 29386
    },
    {
      "epoch": 1.6455930115354462,
      "grad_norm": 1.272046446800232,
      "learning_rate": 9.753342105263158e-05,
      "loss": 0.4131,
      "step": 29387
    },
    {
      "epoch": 1.6456490088475753,
      "grad_norm": 1.8550727367401123,
      "learning_rate": 9.753315789473685e-05,
      "loss": 0.5772,
      "step": 29388
    },
    {
      "epoch": 1.6457050061597043,
      "grad_norm": 1.3328020572662354,
      "learning_rate": 9.753289473684211e-05,
      "loss": 0.5455,
      "step": 29389
    },
    {
      "epoch": 1.6457610034718333,
      "grad_norm": 1.267332673072815,
      "learning_rate": 9.753263157894737e-05,
      "loss": 0.4384,
      "step": 29390
    },
    {
      "epoch": 1.6458170007839623,
      "grad_norm": 1.1775612831115723,
      "learning_rate": 9.753236842105263e-05,
      "loss": 0.4278,
      "step": 29391
    },
    {
      "epoch": 1.6458729980960913,
      "grad_norm": 1.369011402130127,
      "learning_rate": 9.75321052631579e-05,
      "loss": 0.4961,
      "step": 29392
    },
    {
      "epoch": 1.6459289954082204,
      "grad_norm": 1.5223098993301392,
      "learning_rate": 9.753184210526316e-05,
      "loss": 0.5209,
      "step": 29393
    },
    {
      "epoch": 1.6459849927203494,
      "grad_norm": 1.4535424709320068,
      "learning_rate": 9.753157894736842e-05,
      "loss": 0.5633,
      "step": 29394
    },
    {
      "epoch": 1.6460409900324784,
      "grad_norm": 1.1435127258300781,
      "learning_rate": 9.753131578947368e-05,
      "loss": 0.4205,
      "step": 29395
    },
    {
      "epoch": 1.6460969873446074,
      "grad_norm": 1.2056618928909302,
      "learning_rate": 9.753105263157896e-05,
      "loss": 0.3655,
      "step": 29396
    },
    {
      "epoch": 1.6461529846567364,
      "grad_norm": 1.3217023611068726,
      "learning_rate": 9.753078947368422e-05,
      "loss": 0.3516,
      "step": 29397
    },
    {
      "epoch": 1.6462089819688654,
      "grad_norm": 1.221899151802063,
      "learning_rate": 9.753052631578948e-05,
      "loss": 0.4635,
      "step": 29398
    },
    {
      "epoch": 1.6462649792809945,
      "grad_norm": 0.9793908596038818,
      "learning_rate": 9.753026315789474e-05,
      "loss": 0.3526,
      "step": 29399
    },
    {
      "epoch": 1.6463209765931235,
      "grad_norm": 1.0542079210281372,
      "learning_rate": 9.753e-05,
      "loss": 0.3677,
      "step": 29400
    },
    {
      "epoch": 1.6463769739052525,
      "grad_norm": 1.3768837451934814,
      "learning_rate": 9.752973684210527e-05,
      "loss": 0.4844,
      "step": 29401
    },
    {
      "epoch": 1.6464329712173815,
      "grad_norm": 1.3428592681884766,
      "learning_rate": 9.752947368421053e-05,
      "loss": 0.661,
      "step": 29402
    },
    {
      "epoch": 1.6464889685295105,
      "grad_norm": 1.7047210931777954,
      "learning_rate": 9.752921052631579e-05,
      "loss": 0.5118,
      "step": 29403
    },
    {
      "epoch": 1.6465449658416396,
      "grad_norm": 1.5451815128326416,
      "learning_rate": 9.752894736842105e-05,
      "loss": 0.4501,
      "step": 29404
    },
    {
      "epoch": 1.6466009631537686,
      "grad_norm": 1.2636375427246094,
      "learning_rate": 9.752868421052632e-05,
      "loss": 0.4475,
      "step": 29405
    },
    {
      "epoch": 1.6466569604658976,
      "grad_norm": 1.2503951787948608,
      "learning_rate": 9.752842105263158e-05,
      "loss": 0.3915,
      "step": 29406
    },
    {
      "epoch": 1.6467129577780266,
      "grad_norm": 1.2569408416748047,
      "learning_rate": 9.752815789473686e-05,
      "loss": 0.4041,
      "step": 29407
    },
    {
      "epoch": 1.6467689550901556,
      "grad_norm": 1.645736575126648,
      "learning_rate": 9.75278947368421e-05,
      "loss": 0.4398,
      "step": 29408
    },
    {
      "epoch": 1.6468249524022847,
      "grad_norm": 1.7948026657104492,
      "learning_rate": 9.752763157894737e-05,
      "loss": 0.5674,
      "step": 29409
    },
    {
      "epoch": 1.6468809497144137,
      "grad_norm": 1.4449021816253662,
      "learning_rate": 9.752736842105263e-05,
      "loss": 0.5032,
      "step": 29410
    },
    {
      "epoch": 1.6469369470265427,
      "grad_norm": 1.430108904838562,
      "learning_rate": 9.752710526315791e-05,
      "loss": 0.4738,
      "step": 29411
    },
    {
      "epoch": 1.6469929443386717,
      "grad_norm": 1.1391552686691284,
      "learning_rate": 9.752684210526317e-05,
      "loss": 0.4393,
      "step": 29412
    },
    {
      "epoch": 1.6470489416508007,
      "grad_norm": 1.2336457967758179,
      "learning_rate": 9.752657894736843e-05,
      "loss": 0.4446,
      "step": 29413
    },
    {
      "epoch": 1.6471049389629298,
      "grad_norm": 1.372226357460022,
      "learning_rate": 9.752631578947369e-05,
      "loss": 0.406,
      "step": 29414
    },
    {
      "epoch": 1.6471609362750588,
      "grad_norm": 1.3093339204788208,
      "learning_rate": 9.752605263157896e-05,
      "loss": 0.5155,
      "step": 29415
    },
    {
      "epoch": 1.6472169335871878,
      "grad_norm": 1.4605350494384766,
      "learning_rate": 9.752578947368422e-05,
      "loss": 0.4429,
      "step": 29416
    },
    {
      "epoch": 1.6472729308993168,
      "grad_norm": 1.4476892948150635,
      "learning_rate": 9.752552631578948e-05,
      "loss": 0.415,
      "step": 29417
    },
    {
      "epoch": 1.6473289282114458,
      "grad_norm": 1.3584709167480469,
      "learning_rate": 9.752526315789474e-05,
      "loss": 0.5558,
      "step": 29418
    },
    {
      "epoch": 1.6473849255235749,
      "grad_norm": 1.3899284601211548,
      "learning_rate": 9.7525e-05,
      "loss": 0.5136,
      "step": 29419
    },
    {
      "epoch": 1.6474409228357039,
      "grad_norm": 1.0593681335449219,
      "learning_rate": 9.752473684210527e-05,
      "loss": 0.4007,
      "step": 29420
    },
    {
      "epoch": 1.6474969201478329,
      "grad_norm": 1.3471382856369019,
      "learning_rate": 9.752447368421053e-05,
      "loss": 0.4306,
      "step": 29421
    },
    {
      "epoch": 1.647552917459962,
      "grad_norm": 1.8453447818756104,
      "learning_rate": 9.752421052631579e-05,
      "loss": 0.569,
      "step": 29422
    },
    {
      "epoch": 1.647608914772091,
      "grad_norm": 1.2868098020553589,
      "learning_rate": 9.752394736842105e-05,
      "loss": 0.4398,
      "step": 29423
    },
    {
      "epoch": 1.64766491208422,
      "grad_norm": 1.1600334644317627,
      "learning_rate": 9.752368421052632e-05,
      "loss": 0.45,
      "step": 29424
    },
    {
      "epoch": 1.647720909396349,
      "grad_norm": 1.535599946975708,
      "learning_rate": 9.752342105263158e-05,
      "loss": 0.5377,
      "step": 29425
    },
    {
      "epoch": 1.647776906708478,
      "grad_norm": 1.270965814590454,
      "learning_rate": 9.752315789473684e-05,
      "loss": 0.4662,
      "step": 29426
    },
    {
      "epoch": 1.647832904020607,
      "grad_norm": 1.2857426404953003,
      "learning_rate": 9.75228947368421e-05,
      "loss": 0.5464,
      "step": 29427
    },
    {
      "epoch": 1.647888901332736,
      "grad_norm": 1.3699164390563965,
      "learning_rate": 9.752263157894738e-05,
      "loss": 0.3875,
      "step": 29428
    },
    {
      "epoch": 1.647944898644865,
      "grad_norm": 1.0707435607910156,
      "learning_rate": 9.752236842105264e-05,
      "loss": 0.4614,
      "step": 29429
    },
    {
      "epoch": 1.648000895956994,
      "grad_norm": 1.468240737915039,
      "learning_rate": 9.752210526315791e-05,
      "loss": 0.4208,
      "step": 29430
    },
    {
      "epoch": 1.648056893269123,
      "grad_norm": 1.38775634765625,
      "learning_rate": 9.752184210526316e-05,
      "loss": 0.5273,
      "step": 29431
    },
    {
      "epoch": 1.648112890581252,
      "grad_norm": 1.7929494380950928,
      "learning_rate": 9.752157894736843e-05,
      "loss": 0.4566,
      "step": 29432
    },
    {
      "epoch": 1.6481688878933811,
      "grad_norm": 1.2799029350280762,
      "learning_rate": 9.752131578947369e-05,
      "loss": 0.5098,
      "step": 29433
    },
    {
      "epoch": 1.6482248852055101,
      "grad_norm": 1.3467371463775635,
      "learning_rate": 9.752105263157896e-05,
      "loss": 0.4895,
      "step": 29434
    },
    {
      "epoch": 1.6482808825176392,
      "grad_norm": 1.270985722541809,
      "learning_rate": 9.752078947368421e-05,
      "loss": 0.6181,
      "step": 29435
    },
    {
      "epoch": 1.6483368798297682,
      "grad_norm": 1.092190146446228,
      "learning_rate": 9.752052631578947e-05,
      "loss": 0.387,
      "step": 29436
    },
    {
      "epoch": 1.6483928771418972,
      "grad_norm": 1.0694005489349365,
      "learning_rate": 9.752026315789474e-05,
      "loss": 0.4356,
      "step": 29437
    },
    {
      "epoch": 1.6484488744540262,
      "grad_norm": 1.276921272277832,
      "learning_rate": 9.752e-05,
      "loss": 0.4261,
      "step": 29438
    },
    {
      "epoch": 1.6485048717661552,
      "grad_norm": 1.4633861780166626,
      "learning_rate": 9.751973684210527e-05,
      "loss": 0.5498,
      "step": 29439
    },
    {
      "epoch": 1.6485608690782843,
      "grad_norm": 1.5083926916122437,
      "learning_rate": 9.751947368421052e-05,
      "loss": 0.4237,
      "step": 29440
    },
    {
      "epoch": 1.6486168663904133,
      "grad_norm": 1.3734607696533203,
      "learning_rate": 9.75192105263158e-05,
      "loss": 0.3969,
      "step": 29441
    },
    {
      "epoch": 1.6486728637025423,
      "grad_norm": 1.3441766500473022,
      "learning_rate": 9.751894736842105e-05,
      "loss": 0.3779,
      "step": 29442
    },
    {
      "epoch": 1.6487288610146713,
      "grad_norm": 1.093099594116211,
      "learning_rate": 9.751868421052633e-05,
      "loss": 0.3103,
      "step": 29443
    },
    {
      "epoch": 1.6487848583268003,
      "grad_norm": 1.3892213106155396,
      "learning_rate": 9.751842105263159e-05,
      "loss": 0.4943,
      "step": 29444
    },
    {
      "epoch": 1.6488408556389293,
      "grad_norm": 1.4378942251205444,
      "learning_rate": 9.751815789473685e-05,
      "loss": 0.4575,
      "step": 29445
    },
    {
      "epoch": 1.6488968529510584,
      "grad_norm": 1.298954725265503,
      "learning_rate": 9.75178947368421e-05,
      "loss": 0.4933,
      "step": 29446
    },
    {
      "epoch": 1.6489528502631874,
      "grad_norm": 1.314711332321167,
      "learning_rate": 9.751763157894738e-05,
      "loss": 0.4495,
      "step": 29447
    },
    {
      "epoch": 1.6490088475753164,
      "grad_norm": 1.2707839012145996,
      "learning_rate": 9.751736842105264e-05,
      "loss": 0.4522,
      "step": 29448
    },
    {
      "epoch": 1.6490648448874454,
      "grad_norm": 1.2846564054489136,
      "learning_rate": 9.75171052631579e-05,
      "loss": 0.3352,
      "step": 29449
    },
    {
      "epoch": 1.6491208421995744,
      "grad_norm": 1.248152494430542,
      "learning_rate": 9.751684210526316e-05,
      "loss": 0.4574,
      "step": 29450
    },
    {
      "epoch": 1.6491768395117035,
      "grad_norm": 1.2577346563339233,
      "learning_rate": 9.751657894736843e-05,
      "loss": 0.4628,
      "step": 29451
    },
    {
      "epoch": 1.6492328368238325,
      "grad_norm": 1.3272029161453247,
      "learning_rate": 9.751631578947369e-05,
      "loss": 0.4397,
      "step": 29452
    },
    {
      "epoch": 1.6492888341359615,
      "grad_norm": 1.3512473106384277,
      "learning_rate": 9.751605263157895e-05,
      "loss": 0.3505,
      "step": 29453
    },
    {
      "epoch": 1.6493448314480905,
      "grad_norm": 1.1984714269638062,
      "learning_rate": 9.751578947368421e-05,
      "loss": 0.6082,
      "step": 29454
    },
    {
      "epoch": 1.6494008287602195,
      "grad_norm": 1.3861230611801147,
      "learning_rate": 9.751552631578947e-05,
      "loss": 0.4913,
      "step": 29455
    },
    {
      "epoch": 1.6494568260723486,
      "grad_norm": 1.2818530797958374,
      "learning_rate": 9.751526315789474e-05,
      "loss": 0.443,
      "step": 29456
    },
    {
      "epoch": 1.6495128233844776,
      "grad_norm": 1.317275881767273,
      "learning_rate": 9.7515e-05,
      "loss": 0.3658,
      "step": 29457
    },
    {
      "epoch": 1.6495688206966066,
      "grad_norm": 1.3239867687225342,
      "learning_rate": 9.751473684210526e-05,
      "loss": 0.4324,
      "step": 29458
    },
    {
      "epoch": 1.6496248180087356,
      "grad_norm": 1.4137225151062012,
      "learning_rate": 9.751447368421052e-05,
      "loss": 0.4373,
      "step": 29459
    },
    {
      "epoch": 1.6496808153208646,
      "grad_norm": 1.5117820501327515,
      "learning_rate": 9.75142105263158e-05,
      "loss": 0.5513,
      "step": 29460
    },
    {
      "epoch": 1.6497368126329937,
      "grad_norm": 1.5019891262054443,
      "learning_rate": 9.751394736842106e-05,
      "loss": 0.5148,
      "step": 29461
    },
    {
      "epoch": 1.6497928099451227,
      "grad_norm": 1.271453857421875,
      "learning_rate": 9.751368421052633e-05,
      "loss": 0.4867,
      "step": 29462
    },
    {
      "epoch": 1.6498488072572517,
      "grad_norm": 1.3284165859222412,
      "learning_rate": 9.751342105263158e-05,
      "loss": 0.4884,
      "step": 29463
    },
    {
      "epoch": 1.6499048045693807,
      "grad_norm": 1.3907990455627441,
      "learning_rate": 9.751315789473685e-05,
      "loss": 0.471,
      "step": 29464
    },
    {
      "epoch": 1.6499608018815097,
      "grad_norm": 1.2383745908737183,
      "learning_rate": 9.751289473684211e-05,
      "loss": 0.318,
      "step": 29465
    },
    {
      "epoch": 1.6500167991936388,
      "grad_norm": 1.1266353130340576,
      "learning_rate": 9.751263157894738e-05,
      "loss": 0.3896,
      "step": 29466
    },
    {
      "epoch": 1.6500727965057678,
      "grad_norm": 1.3946961164474487,
      "learning_rate": 9.751236842105264e-05,
      "loss": 0.4474,
      "step": 29467
    },
    {
      "epoch": 1.6501287938178968,
      "grad_norm": 1.186700463294983,
      "learning_rate": 9.75121052631579e-05,
      "loss": 0.319,
      "step": 29468
    },
    {
      "epoch": 1.6501847911300258,
      "grad_norm": 1.1210553646087646,
      "learning_rate": 9.751184210526316e-05,
      "loss": 0.3125,
      "step": 29469
    },
    {
      "epoch": 1.6502407884421548,
      "grad_norm": 1.0811258554458618,
      "learning_rate": 9.751157894736842e-05,
      "loss": 0.3655,
      "step": 29470
    },
    {
      "epoch": 1.6502967857542838,
      "grad_norm": 1.7336432933807373,
      "learning_rate": 9.75113157894737e-05,
      "loss": 0.4085,
      "step": 29471
    },
    {
      "epoch": 1.6503527830664129,
      "grad_norm": 1.5337207317352295,
      "learning_rate": 9.751105263157895e-05,
      "loss": 0.4996,
      "step": 29472
    },
    {
      "epoch": 1.6504087803785419,
      "grad_norm": 1.4256982803344727,
      "learning_rate": 9.751078947368421e-05,
      "loss": 0.3553,
      "step": 29473
    },
    {
      "epoch": 1.650464777690671,
      "grad_norm": 1.38419508934021,
      "learning_rate": 9.751052631578947e-05,
      "loss": 0.4517,
      "step": 29474
    },
    {
      "epoch": 1.6505207750028,
      "grad_norm": 1.5306450128555298,
      "learning_rate": 9.751026315789475e-05,
      "loss": 0.4817,
      "step": 29475
    },
    {
      "epoch": 1.650576772314929,
      "grad_norm": 1.411915898323059,
      "learning_rate": 9.751e-05,
      "loss": 0.4983,
      "step": 29476
    },
    {
      "epoch": 1.650632769627058,
      "grad_norm": 1.1324421167373657,
      "learning_rate": 9.750973684210527e-05,
      "loss": 0.3834,
      "step": 29477
    },
    {
      "epoch": 1.650688766939187,
      "grad_norm": 1.4064148664474487,
      "learning_rate": 9.750947368421053e-05,
      "loss": 0.5747,
      "step": 29478
    },
    {
      "epoch": 1.650744764251316,
      "grad_norm": 1.3565051555633545,
      "learning_rate": 9.75092105263158e-05,
      "loss": 0.4658,
      "step": 29479
    },
    {
      "epoch": 1.650800761563445,
      "grad_norm": 1.220345377922058,
      "learning_rate": 9.750894736842106e-05,
      "loss": 0.4655,
      "step": 29480
    },
    {
      "epoch": 1.650856758875574,
      "grad_norm": 1.3398367166519165,
      "learning_rate": 9.750868421052632e-05,
      "loss": 0.5441,
      "step": 29481
    },
    {
      "epoch": 1.650912756187703,
      "grad_norm": 3.0499606132507324,
      "learning_rate": 9.750842105263158e-05,
      "loss": 0.4552,
      "step": 29482
    },
    {
      "epoch": 1.650968753499832,
      "grad_norm": 1.3581091165542603,
      "learning_rate": 9.750815789473685e-05,
      "loss": 0.5208,
      "step": 29483
    },
    {
      "epoch": 1.651024750811961,
      "grad_norm": 1.7729673385620117,
      "learning_rate": 9.750789473684211e-05,
      "loss": 0.6271,
      "step": 29484
    },
    {
      "epoch": 1.6510807481240901,
      "grad_norm": 1.3497742414474487,
      "learning_rate": 9.750763157894738e-05,
      "loss": 0.4637,
      "step": 29485
    },
    {
      "epoch": 1.6511367454362191,
      "grad_norm": 1.1094225645065308,
      "learning_rate": 9.750736842105263e-05,
      "loss": 0.5067,
      "step": 29486
    },
    {
      "epoch": 1.6511927427483482,
      "grad_norm": 1.2953336238861084,
      "learning_rate": 9.750710526315789e-05,
      "loss": 0.4458,
      "step": 29487
    },
    {
      "epoch": 1.6512487400604772,
      "grad_norm": 1.1549729108810425,
      "learning_rate": 9.750684210526316e-05,
      "loss": 0.437,
      "step": 29488
    },
    {
      "epoch": 1.6513047373726062,
      "grad_norm": 1.3755946159362793,
      "learning_rate": 9.750657894736842e-05,
      "loss": 0.4402,
      "step": 29489
    },
    {
      "epoch": 1.6513607346847352,
      "grad_norm": 1.4747800827026367,
      "learning_rate": 9.750631578947368e-05,
      "loss": 0.4794,
      "step": 29490
    },
    {
      "epoch": 1.6514167319968642,
      "grad_norm": 1.548931360244751,
      "learning_rate": 9.750605263157894e-05,
      "loss": 0.5268,
      "step": 29491
    },
    {
      "epoch": 1.6514727293089932,
      "grad_norm": 1.3267422914505005,
      "learning_rate": 9.750578947368422e-05,
      "loss": 0.4324,
      "step": 29492
    },
    {
      "epoch": 1.6515287266211223,
      "grad_norm": 1.1592317819595337,
      "learning_rate": 9.750552631578948e-05,
      "loss": 0.4064,
      "step": 29493
    },
    {
      "epoch": 1.6515847239332513,
      "grad_norm": 1.4331164360046387,
      "learning_rate": 9.750526315789475e-05,
      "loss": 0.6338,
      "step": 29494
    },
    {
      "epoch": 1.6516407212453803,
      "grad_norm": 1.419296383857727,
      "learning_rate": 9.7505e-05,
      "loss": 0.5461,
      "step": 29495
    },
    {
      "epoch": 1.6516967185575093,
      "grad_norm": 1.5112463235855103,
      "learning_rate": 9.750473684210527e-05,
      "loss": 0.401,
      "step": 29496
    },
    {
      "epoch": 1.6517527158696383,
      "grad_norm": 1.4628111124038696,
      "learning_rate": 9.750447368421053e-05,
      "loss": 0.5966,
      "step": 29497
    },
    {
      "epoch": 1.6518087131817674,
      "grad_norm": 1.2177109718322754,
      "learning_rate": 9.75042105263158e-05,
      "loss": 0.3109,
      "step": 29498
    },
    {
      "epoch": 1.6518647104938964,
      "grad_norm": 1.3104015588760376,
      "learning_rate": 9.750394736842106e-05,
      "loss": 0.4858,
      "step": 29499
    },
    {
      "epoch": 1.6519207078060254,
      "grad_norm": 1.5395573377609253,
      "learning_rate": 9.750368421052632e-05,
      "loss": 0.437,
      "step": 29500
    },
    {
      "epoch": 1.6519767051181544,
      "grad_norm": 2.065742254257202,
      "learning_rate": 9.750342105263158e-05,
      "loss": 0.5739,
      "step": 29501
    },
    {
      "epoch": 1.6520327024302834,
      "grad_norm": 1.1836086511611938,
      "learning_rate": 9.750315789473685e-05,
      "loss": 0.4297,
      "step": 29502
    },
    {
      "epoch": 1.6520886997424125,
      "grad_norm": 1.1465251445770264,
      "learning_rate": 9.750289473684211e-05,
      "loss": 0.4416,
      "step": 29503
    },
    {
      "epoch": 1.6521446970545415,
      "grad_norm": 1.2871326208114624,
      "learning_rate": 9.750263157894737e-05,
      "loss": 0.5569,
      "step": 29504
    },
    {
      "epoch": 1.6522006943666705,
      "grad_norm": 1.2531226873397827,
      "learning_rate": 9.750236842105263e-05,
      "loss": 0.3886,
      "step": 29505
    },
    {
      "epoch": 1.6522566916787995,
      "grad_norm": 1.8033958673477173,
      "learning_rate": 9.75021052631579e-05,
      "loss": 0.4523,
      "step": 29506
    },
    {
      "epoch": 1.6523126889909285,
      "grad_norm": 1.0945812463760376,
      "learning_rate": 9.750184210526317e-05,
      "loss": 0.3309,
      "step": 29507
    },
    {
      "epoch": 1.6523686863030576,
      "grad_norm": 1.440437912940979,
      "learning_rate": 9.750157894736843e-05,
      "loss": 0.5318,
      "step": 29508
    },
    {
      "epoch": 1.6524246836151866,
      "grad_norm": 1.3866740465164185,
      "learning_rate": 9.750131578947369e-05,
      "loss": 0.6054,
      "step": 29509
    },
    {
      "epoch": 1.6524806809273156,
      "grad_norm": 1.3404566049575806,
      "learning_rate": 9.750105263157895e-05,
      "loss": 0.7541,
      "step": 29510
    },
    {
      "epoch": 1.6525366782394446,
      "grad_norm": 1.138606071472168,
      "learning_rate": 9.750078947368422e-05,
      "loss": 0.3472,
      "step": 29511
    },
    {
      "epoch": 1.6525926755515736,
      "grad_norm": 1.5618456602096558,
      "learning_rate": 9.750052631578948e-05,
      "loss": 0.5744,
      "step": 29512
    },
    {
      "epoch": 1.6526486728637027,
      "grad_norm": 1.288644790649414,
      "learning_rate": 9.750026315789474e-05,
      "loss": 0.5016,
      "step": 29513
    },
    {
      "epoch": 1.6527046701758317,
      "grad_norm": 1.5030759572982788,
      "learning_rate": 9.75e-05,
      "loss": 0.6377,
      "step": 29514
    },
    {
      "epoch": 1.6527606674879607,
      "grad_norm": 3.824960231781006,
      "learning_rate": 9.749973684210527e-05,
      "loss": 0.4982,
      "step": 29515
    },
    {
      "epoch": 1.6528166648000897,
      "grad_norm": 1.2412022352218628,
      "learning_rate": 9.749947368421053e-05,
      "loss": 0.4165,
      "step": 29516
    },
    {
      "epoch": 1.6528726621122187,
      "grad_norm": 1.4901292324066162,
      "learning_rate": 9.74992105263158e-05,
      "loss": 0.4076,
      "step": 29517
    },
    {
      "epoch": 1.6529286594243477,
      "grad_norm": 1.3059476613998413,
      "learning_rate": 9.749894736842105e-05,
      "loss": 0.4026,
      "step": 29518
    },
    {
      "epoch": 1.6529846567364768,
      "grad_norm": 1.6507205963134766,
      "learning_rate": 9.749868421052632e-05,
      "loss": 0.5171,
      "step": 29519
    },
    {
      "epoch": 1.6530406540486058,
      "grad_norm": 1.561182975769043,
      "learning_rate": 9.749842105263158e-05,
      "loss": 0.5634,
      "step": 29520
    },
    {
      "epoch": 1.6530966513607348,
      "grad_norm": 1.259454607963562,
      "learning_rate": 9.749815789473684e-05,
      "loss": 0.4905,
      "step": 29521
    },
    {
      "epoch": 1.6531526486728638,
      "grad_norm": 2.1463735103607178,
      "learning_rate": 9.749789473684212e-05,
      "loss": 0.6679,
      "step": 29522
    },
    {
      "epoch": 1.6532086459849928,
      "grad_norm": 5.757894515991211,
      "learning_rate": 9.749763157894736e-05,
      "loss": 0.6395,
      "step": 29523
    },
    {
      "epoch": 1.6532646432971219,
      "grad_norm": 1.4047274589538574,
      "learning_rate": 9.749736842105264e-05,
      "loss": 0.4141,
      "step": 29524
    },
    {
      "epoch": 1.6533206406092509,
      "grad_norm": 1.2539740800857544,
      "learning_rate": 9.74971052631579e-05,
      "loss": 0.4272,
      "step": 29525
    },
    {
      "epoch": 1.65337663792138,
      "grad_norm": 1.2732815742492676,
      "learning_rate": 9.749684210526317e-05,
      "loss": 0.5182,
      "step": 29526
    },
    {
      "epoch": 1.653432635233509,
      "grad_norm": 1.3196169137954712,
      "learning_rate": 9.749657894736842e-05,
      "loss": 0.3876,
      "step": 29527
    },
    {
      "epoch": 1.653488632545638,
      "grad_norm": 1.3361912965774536,
      "learning_rate": 9.749631578947369e-05,
      "loss": 0.4089,
      "step": 29528
    },
    {
      "epoch": 1.653544629857767,
      "grad_norm": 1.3709325790405273,
      "learning_rate": 9.749605263157895e-05,
      "loss": 0.6227,
      "step": 29529
    },
    {
      "epoch": 1.653600627169896,
      "grad_norm": 1.0868281126022339,
      "learning_rate": 9.749578947368422e-05,
      "loss": 0.3535,
      "step": 29530
    },
    {
      "epoch": 1.653656624482025,
      "grad_norm": 1.2153334617614746,
      "learning_rate": 9.749552631578948e-05,
      "loss": 0.3779,
      "step": 29531
    },
    {
      "epoch": 1.653712621794154,
      "grad_norm": 1.5079317092895508,
      "learning_rate": 9.749526315789474e-05,
      "loss": 0.4953,
      "step": 29532
    },
    {
      "epoch": 1.653768619106283,
      "grad_norm": 1.44725501537323,
      "learning_rate": 9.7495e-05,
      "loss": 0.5007,
      "step": 29533
    },
    {
      "epoch": 1.653824616418412,
      "grad_norm": 1.10235595703125,
      "learning_rate": 9.749473684210527e-05,
      "loss": 0.4212,
      "step": 29534
    },
    {
      "epoch": 1.653880613730541,
      "grad_norm": 1.637251615524292,
      "learning_rate": 9.749447368421053e-05,
      "loss": 0.5369,
      "step": 29535
    },
    {
      "epoch": 1.65393661104267,
      "grad_norm": 1.3870211839675903,
      "learning_rate": 9.74942105263158e-05,
      "loss": 0.4615,
      "step": 29536
    },
    {
      "epoch": 1.653992608354799,
      "grad_norm": 1.8148595094680786,
      "learning_rate": 9.749394736842105e-05,
      "loss": 0.608,
      "step": 29537
    },
    {
      "epoch": 1.6540486056669281,
      "grad_norm": 1.1375174522399902,
      "learning_rate": 9.749368421052633e-05,
      "loss": 0.4065,
      "step": 29538
    },
    {
      "epoch": 1.6541046029790571,
      "grad_norm": 1.3817225694656372,
      "learning_rate": 9.749342105263159e-05,
      "loss": 0.5522,
      "step": 29539
    },
    {
      "epoch": 1.6541606002911862,
      "grad_norm": 1.1288411617279053,
      "learning_rate": 9.749315789473685e-05,
      "loss": 0.429,
      "step": 29540
    },
    {
      "epoch": 1.6542165976033152,
      "grad_norm": 1.1628162860870361,
      "learning_rate": 9.74928947368421e-05,
      "loss": 0.3498,
      "step": 29541
    },
    {
      "epoch": 1.6542725949154442,
      "grad_norm": 1.3848342895507812,
      "learning_rate": 9.749263157894737e-05,
      "loss": 0.4717,
      "step": 29542
    },
    {
      "epoch": 1.6543285922275732,
      "grad_norm": 1.5429258346557617,
      "learning_rate": 9.749236842105264e-05,
      "loss": 0.458,
      "step": 29543
    },
    {
      "epoch": 1.6543845895397022,
      "grad_norm": 1.1181485652923584,
      "learning_rate": 9.74921052631579e-05,
      "loss": 0.4494,
      "step": 29544
    },
    {
      "epoch": 1.6544405868518313,
      "grad_norm": 1.3679903745651245,
      "learning_rate": 9.749184210526316e-05,
      "loss": 0.5011,
      "step": 29545
    },
    {
      "epoch": 1.6544965841639603,
      "grad_norm": 1.237748622894287,
      "learning_rate": 9.749157894736842e-05,
      "loss": 0.3805,
      "step": 29546
    },
    {
      "epoch": 1.6545525814760893,
      "grad_norm": 1.2433940172195435,
      "learning_rate": 9.749131578947369e-05,
      "loss": 0.3774,
      "step": 29547
    },
    {
      "epoch": 1.6546085787882183,
      "grad_norm": 1.468341588973999,
      "learning_rate": 9.749105263157895e-05,
      "loss": 0.4329,
      "step": 29548
    },
    {
      "epoch": 1.6546645761003473,
      "grad_norm": 1.4150248765945435,
      "learning_rate": 9.749078947368422e-05,
      "loss": 0.4416,
      "step": 29549
    },
    {
      "epoch": 1.6547205734124764,
      "grad_norm": 1.1630582809448242,
      "learning_rate": 9.749052631578947e-05,
      "loss": 0.4249,
      "step": 29550
    },
    {
      "epoch": 1.6547765707246054,
      "grad_norm": 1.161422610282898,
      "learning_rate": 9.749026315789474e-05,
      "loss": 0.3848,
      "step": 29551
    },
    {
      "epoch": 1.6548325680367344,
      "grad_norm": 1.278017520904541,
      "learning_rate": 9.749e-05,
      "loss": 0.402,
      "step": 29552
    },
    {
      "epoch": 1.6548885653488634,
      "grad_norm": 1.96346914768219,
      "learning_rate": 9.748973684210528e-05,
      "loss": 0.6422,
      "step": 29553
    },
    {
      "epoch": 1.6549445626609924,
      "grad_norm": 1.2192671298980713,
      "learning_rate": 9.748947368421054e-05,
      "loss": 0.5708,
      "step": 29554
    },
    {
      "epoch": 1.6550005599731215,
      "grad_norm": 1.2060295343399048,
      "learning_rate": 9.74892105263158e-05,
      "loss": 0.5261,
      "step": 29555
    },
    {
      "epoch": 1.6550565572852505,
      "grad_norm": 1.8292222023010254,
      "learning_rate": 9.748894736842106e-05,
      "loss": 0.5154,
      "step": 29556
    },
    {
      "epoch": 1.6551125545973793,
      "grad_norm": 1.2270359992980957,
      "learning_rate": 9.748868421052632e-05,
      "loss": 0.5626,
      "step": 29557
    },
    {
      "epoch": 1.6551685519095083,
      "grad_norm": 1.1451836824417114,
      "learning_rate": 9.748842105263159e-05,
      "loss": 0.5148,
      "step": 29558
    },
    {
      "epoch": 1.6552245492216373,
      "grad_norm": 1.3507838249206543,
      "learning_rate": 9.748815789473685e-05,
      "loss": 0.6799,
      "step": 29559
    },
    {
      "epoch": 1.6552805465337663,
      "grad_norm": 1.8475145101547241,
      "learning_rate": 9.748789473684211e-05,
      "loss": 0.6658,
      "step": 29560
    },
    {
      "epoch": 1.6553365438458953,
      "grad_norm": 1.3059606552124023,
      "learning_rate": 9.748763157894737e-05,
      "loss": 0.4534,
      "step": 29561
    },
    {
      "epoch": 1.6553925411580244,
      "grad_norm": 1.292354941368103,
      "learning_rate": 9.748736842105264e-05,
      "loss": 0.432,
      "step": 29562
    },
    {
      "epoch": 1.6554485384701534,
      "grad_norm": 1.2703644037246704,
      "learning_rate": 9.74871052631579e-05,
      "loss": 0.5647,
      "step": 29563
    },
    {
      "epoch": 1.6555045357822824,
      "grad_norm": 1.7154474258422852,
      "learning_rate": 9.748684210526316e-05,
      "loss": 0.5775,
      "step": 29564
    },
    {
      "epoch": 1.6555605330944114,
      "grad_norm": 1.4255478382110596,
      "learning_rate": 9.748657894736842e-05,
      "loss": 0.407,
      "step": 29565
    },
    {
      "epoch": 1.6556165304065404,
      "grad_norm": 1.5124664306640625,
      "learning_rate": 9.74863157894737e-05,
      "loss": 0.4397,
      "step": 29566
    },
    {
      "epoch": 1.6556725277186695,
      "grad_norm": 1.2534633874893188,
      "learning_rate": 9.748605263157895e-05,
      "loss": 0.4175,
      "step": 29567
    },
    {
      "epoch": 1.6557285250307985,
      "grad_norm": 1.333579182624817,
      "learning_rate": 9.748578947368421e-05,
      "loss": 0.515,
      "step": 29568
    },
    {
      "epoch": 1.6557845223429275,
      "grad_norm": 1.228956937789917,
      "learning_rate": 9.748552631578947e-05,
      "loss": 0.518,
      "step": 29569
    },
    {
      "epoch": 1.6558405196550565,
      "grad_norm": 1.1497122049331665,
      "learning_rate": 9.748526315789475e-05,
      "loss": 0.4025,
      "step": 29570
    },
    {
      "epoch": 1.6558965169671855,
      "grad_norm": 1.0443910360336304,
      "learning_rate": 9.7485e-05,
      "loss": 0.427,
      "step": 29571
    },
    {
      "epoch": 1.6559525142793146,
      "grad_norm": 1.344438076019287,
      "learning_rate": 9.748473684210528e-05,
      "loss": 0.45,
      "step": 29572
    },
    {
      "epoch": 1.6560085115914436,
      "grad_norm": 1.1079256534576416,
      "learning_rate": 9.748447368421053e-05,
      "loss": 0.4042,
      "step": 29573
    },
    {
      "epoch": 1.6560645089035726,
      "grad_norm": 1.3417983055114746,
      "learning_rate": 9.748421052631579e-05,
      "loss": 0.435,
      "step": 29574
    },
    {
      "epoch": 1.6561205062157016,
      "grad_norm": 1.4069679975509644,
      "learning_rate": 9.748394736842106e-05,
      "loss": 0.46,
      "step": 29575
    },
    {
      "epoch": 1.6561765035278306,
      "grad_norm": 1.1575227975845337,
      "learning_rate": 9.748368421052632e-05,
      "loss": 0.4175,
      "step": 29576
    },
    {
      "epoch": 1.6562325008399597,
      "grad_norm": 1.3027757406234741,
      "learning_rate": 9.748342105263159e-05,
      "loss": 0.4284,
      "step": 29577
    },
    {
      "epoch": 1.6562884981520887,
      "grad_norm": 1.4100687503814697,
      "learning_rate": 9.748315789473684e-05,
      "loss": 0.8479,
      "step": 29578
    },
    {
      "epoch": 1.6563444954642177,
      "grad_norm": 1.2550665140151978,
      "learning_rate": 9.748289473684211e-05,
      "loss": 0.5583,
      "step": 29579
    },
    {
      "epoch": 1.6564004927763467,
      "grad_norm": 1.2159197330474854,
      "learning_rate": 9.748263157894737e-05,
      "loss": 0.5085,
      "step": 29580
    },
    {
      "epoch": 1.6564564900884757,
      "grad_norm": 1.5163697004318237,
      "learning_rate": 9.748236842105264e-05,
      "loss": 0.4916,
      "step": 29581
    },
    {
      "epoch": 1.6565124874006047,
      "grad_norm": 1.3901549577713013,
      "learning_rate": 9.748210526315789e-05,
      "loss": 0.4707,
      "step": 29582
    },
    {
      "epoch": 1.6565684847127338,
      "grad_norm": 1.4969396591186523,
      "learning_rate": 9.748184210526316e-05,
      "loss": 0.5083,
      "step": 29583
    },
    {
      "epoch": 1.6566244820248628,
      "grad_norm": 1.331526279449463,
      "learning_rate": 9.748157894736842e-05,
      "loss": 0.3855,
      "step": 29584
    },
    {
      "epoch": 1.6566804793369918,
      "grad_norm": 1.305368423461914,
      "learning_rate": 9.74813157894737e-05,
      "loss": 0.4413,
      "step": 29585
    },
    {
      "epoch": 1.6567364766491208,
      "grad_norm": 1.43300461769104,
      "learning_rate": 9.748105263157896e-05,
      "loss": 0.5209,
      "step": 29586
    },
    {
      "epoch": 1.6567924739612498,
      "grad_norm": 1.1488232612609863,
      "learning_rate": 9.748078947368422e-05,
      "loss": 0.4211,
      "step": 29587
    },
    {
      "epoch": 1.6568484712733789,
      "grad_norm": 1.2985970973968506,
      "learning_rate": 9.748052631578948e-05,
      "loss": 0.531,
      "step": 29588
    },
    {
      "epoch": 1.6569044685855079,
      "grad_norm": 1.1614956855773926,
      "learning_rate": 9.748026315789475e-05,
      "loss": 0.4647,
      "step": 29589
    },
    {
      "epoch": 1.656960465897637,
      "grad_norm": 1.4463664293289185,
      "learning_rate": 9.748000000000001e-05,
      "loss": 0.5643,
      "step": 29590
    },
    {
      "epoch": 1.657016463209766,
      "grad_norm": 1.2314857244491577,
      "learning_rate": 9.747973684210527e-05,
      "loss": 0.478,
      "step": 29591
    },
    {
      "epoch": 1.657072460521895,
      "grad_norm": 1.4421050548553467,
      "learning_rate": 9.747947368421053e-05,
      "loss": 0.4519,
      "step": 29592
    },
    {
      "epoch": 1.657128457834024,
      "grad_norm": 1.3177601099014282,
      "learning_rate": 9.747921052631579e-05,
      "loss": 0.4995,
      "step": 29593
    },
    {
      "epoch": 1.657184455146153,
      "grad_norm": 1.2215352058410645,
      "learning_rate": 9.747894736842106e-05,
      "loss": 0.3068,
      "step": 29594
    },
    {
      "epoch": 1.657240452458282,
      "grad_norm": 1.2740132808685303,
      "learning_rate": 9.747868421052632e-05,
      "loss": 0.5852,
      "step": 29595
    },
    {
      "epoch": 1.657296449770411,
      "grad_norm": 1.2289226055145264,
      "learning_rate": 9.747842105263158e-05,
      "loss": 0.5125,
      "step": 29596
    },
    {
      "epoch": 1.65735244708254,
      "grad_norm": 1.533870816230774,
      "learning_rate": 9.747815789473684e-05,
      "loss": 0.7458,
      "step": 29597
    },
    {
      "epoch": 1.657408444394669,
      "grad_norm": 1.202687382698059,
      "learning_rate": 9.747789473684211e-05,
      "loss": 0.4193,
      "step": 29598
    },
    {
      "epoch": 1.657464441706798,
      "grad_norm": 1.5871564149856567,
      "learning_rate": 9.747763157894737e-05,
      "loss": 0.6005,
      "step": 29599
    },
    {
      "epoch": 1.657520439018927,
      "grad_norm": 1.3361384868621826,
      "learning_rate": 9.747736842105263e-05,
      "loss": 0.4297,
      "step": 29600
    },
    {
      "epoch": 1.6575764363310561,
      "grad_norm": 1.468178629875183,
      "learning_rate": 9.747710526315789e-05,
      "loss": 0.5172,
      "step": 29601
    },
    {
      "epoch": 1.6576324336431851,
      "grad_norm": 1.1791592836380005,
      "learning_rate": 9.747684210526317e-05,
      "loss": 0.3931,
      "step": 29602
    },
    {
      "epoch": 1.6576884309553142,
      "grad_norm": 1.1586310863494873,
      "learning_rate": 9.747657894736843e-05,
      "loss": 0.4085,
      "step": 29603
    },
    {
      "epoch": 1.6577444282674432,
      "grad_norm": 1.1029143333435059,
      "learning_rate": 9.74763157894737e-05,
      "loss": 0.4226,
      "step": 29604
    },
    {
      "epoch": 1.6578004255795722,
      "grad_norm": 1.5106743574142456,
      "learning_rate": 9.747605263157895e-05,
      "loss": 0.4974,
      "step": 29605
    },
    {
      "epoch": 1.6578564228917012,
      "grad_norm": 1.2638213634490967,
      "learning_rate": 9.747578947368422e-05,
      "loss": 0.3985,
      "step": 29606
    },
    {
      "epoch": 1.6579124202038302,
      "grad_norm": 1.1854311227798462,
      "learning_rate": 9.747552631578948e-05,
      "loss": 0.3951,
      "step": 29607
    },
    {
      "epoch": 1.6579684175159592,
      "grad_norm": 1.3831970691680908,
      "learning_rate": 9.747526315789474e-05,
      "loss": 0.4718,
      "step": 29608
    },
    {
      "epoch": 1.6580244148280883,
      "grad_norm": 1.2144850492477417,
      "learning_rate": 9.747500000000001e-05,
      "loss": 0.5548,
      "step": 29609
    },
    {
      "epoch": 1.6580804121402173,
      "grad_norm": 1.3014781475067139,
      "learning_rate": 9.747473684210526e-05,
      "loss": 0.3956,
      "step": 29610
    },
    {
      "epoch": 1.6581364094523463,
      "grad_norm": 1.3508234024047852,
      "learning_rate": 9.747447368421053e-05,
      "loss": 0.4022,
      "step": 29611
    },
    {
      "epoch": 1.6581924067644753,
      "grad_norm": 1.2428338527679443,
      "learning_rate": 9.747421052631579e-05,
      "loss": 0.4778,
      "step": 29612
    },
    {
      "epoch": 1.6582484040766043,
      "grad_norm": 1.5273547172546387,
      "learning_rate": 9.747394736842106e-05,
      "loss": 0.542,
      "step": 29613
    },
    {
      "epoch": 1.6583044013887334,
      "grad_norm": 1.193739891052246,
      "learning_rate": 9.747368421052632e-05,
      "loss": 0.4343,
      "step": 29614
    },
    {
      "epoch": 1.6583603987008624,
      "grad_norm": 2.0253703594207764,
      "learning_rate": 9.747342105263158e-05,
      "loss": 0.6708,
      "step": 29615
    },
    {
      "epoch": 1.6584163960129914,
      "grad_norm": 1.3252277374267578,
      "learning_rate": 9.747315789473684e-05,
      "loss": 0.3905,
      "step": 29616
    },
    {
      "epoch": 1.6584723933251204,
      "grad_norm": 1.2533987760543823,
      "learning_rate": 9.747289473684212e-05,
      "loss": 0.384,
      "step": 29617
    },
    {
      "epoch": 1.6585283906372494,
      "grad_norm": 1.2431951761245728,
      "learning_rate": 9.747263157894738e-05,
      "loss": 0.579,
      "step": 29618
    },
    {
      "epoch": 1.6585843879493785,
      "grad_norm": 1.0912152528762817,
      "learning_rate": 9.747236842105264e-05,
      "loss": 0.4041,
      "step": 29619
    },
    {
      "epoch": 1.6586403852615075,
      "grad_norm": 1.4213159084320068,
      "learning_rate": 9.74721052631579e-05,
      "loss": 0.6027,
      "step": 29620
    },
    {
      "epoch": 1.6586963825736365,
      "grad_norm": 1.2781519889831543,
      "learning_rate": 9.747184210526317e-05,
      "loss": 0.4213,
      "step": 29621
    },
    {
      "epoch": 1.6587523798857655,
      "grad_norm": 1.4699957370758057,
      "learning_rate": 9.747157894736843e-05,
      "loss": 0.5215,
      "step": 29622
    },
    {
      "epoch": 1.6588083771978945,
      "grad_norm": 2.326874017715454,
      "learning_rate": 9.747131578947369e-05,
      "loss": 0.3919,
      "step": 29623
    },
    {
      "epoch": 1.6588643745100236,
      "grad_norm": 1.2619982957839966,
      "learning_rate": 9.747105263157895e-05,
      "loss": 0.4346,
      "step": 29624
    },
    {
      "epoch": 1.6589203718221526,
      "grad_norm": 1.6012333631515503,
      "learning_rate": 9.747078947368421e-05,
      "loss": 0.4691,
      "step": 29625
    },
    {
      "epoch": 1.6589763691342816,
      "grad_norm": 1.347254753112793,
      "learning_rate": 9.747052631578948e-05,
      "loss": 0.3613,
      "step": 29626
    },
    {
      "epoch": 1.6590323664464106,
      "grad_norm": 1.4736576080322266,
      "learning_rate": 9.747026315789474e-05,
      "loss": 0.4603,
      "step": 29627
    },
    {
      "epoch": 1.6590883637585396,
      "grad_norm": 1.2075624465942383,
      "learning_rate": 9.747e-05,
      "loss": 0.4067,
      "step": 29628
    },
    {
      "epoch": 1.6591443610706686,
      "grad_norm": 1.2575129270553589,
      "learning_rate": 9.746973684210526e-05,
      "loss": 0.4704,
      "step": 29629
    },
    {
      "epoch": 1.6592003583827977,
      "grad_norm": 1.3325769901275635,
      "learning_rate": 9.746947368421053e-05,
      "loss": 0.5049,
      "step": 29630
    },
    {
      "epoch": 1.6592563556949267,
      "grad_norm": 1.4154554605484009,
      "learning_rate": 9.746921052631579e-05,
      "loss": 0.5334,
      "step": 29631
    },
    {
      "epoch": 1.6593123530070557,
      "grad_norm": 1.592248558998108,
      "learning_rate": 9.746894736842107e-05,
      "loss": 0.5429,
      "step": 29632
    },
    {
      "epoch": 1.6593683503191847,
      "grad_norm": 1.3198601007461548,
      "learning_rate": 9.746868421052631e-05,
      "loss": 0.4146,
      "step": 29633
    },
    {
      "epoch": 1.6594243476313137,
      "grad_norm": 1.1203079223632812,
      "learning_rate": 9.746842105263159e-05,
      "loss": 0.4375,
      "step": 29634
    },
    {
      "epoch": 1.6594803449434428,
      "grad_norm": 1.3273859024047852,
      "learning_rate": 9.746815789473685e-05,
      "loss": 0.4493,
      "step": 29635
    },
    {
      "epoch": 1.6595363422555718,
      "grad_norm": 1.3148910999298096,
      "learning_rate": 9.746789473684212e-05,
      "loss": 0.5294,
      "step": 29636
    },
    {
      "epoch": 1.6595923395677008,
      "grad_norm": 1.3773329257965088,
      "learning_rate": 9.746763157894736e-05,
      "loss": 0.4416,
      "step": 29637
    },
    {
      "epoch": 1.6596483368798298,
      "grad_norm": 1.2627286911010742,
      "learning_rate": 9.746736842105264e-05,
      "loss": 0.4194,
      "step": 29638
    },
    {
      "epoch": 1.6597043341919588,
      "grad_norm": 1.3446952104568481,
      "learning_rate": 9.74671052631579e-05,
      "loss": 0.4122,
      "step": 29639
    },
    {
      "epoch": 1.6597603315040876,
      "grad_norm": 1.124110460281372,
      "learning_rate": 9.746684210526317e-05,
      "loss": 0.5054,
      "step": 29640
    },
    {
      "epoch": 1.6598163288162167,
      "grad_norm": 1.5453792810440063,
      "learning_rate": 9.746657894736843e-05,
      "loss": 0.5256,
      "step": 29641
    },
    {
      "epoch": 1.6598723261283457,
      "grad_norm": 1.4115320444107056,
      "learning_rate": 9.746631578947368e-05,
      "loss": 0.4691,
      "step": 29642
    },
    {
      "epoch": 1.6599283234404747,
      "grad_norm": 1.2420213222503662,
      "learning_rate": 9.746605263157895e-05,
      "loss": 0.452,
      "step": 29643
    },
    {
      "epoch": 1.6599843207526037,
      "grad_norm": 1.2624112367630005,
      "learning_rate": 9.746578947368421e-05,
      "loss": 0.4987,
      "step": 29644
    },
    {
      "epoch": 1.6600403180647327,
      "grad_norm": 1.2748420238494873,
      "learning_rate": 9.746552631578948e-05,
      "loss": 0.5516,
      "step": 29645
    },
    {
      "epoch": 1.6600963153768618,
      "grad_norm": 1.506902813911438,
      "learning_rate": 9.746526315789474e-05,
      "loss": 0.4746,
      "step": 29646
    },
    {
      "epoch": 1.6601523126889908,
      "grad_norm": 1.2483278512954712,
      "learning_rate": 9.7465e-05,
      "loss": 0.4323,
      "step": 29647
    },
    {
      "epoch": 1.6602083100011198,
      "grad_norm": 1.122362732887268,
      "learning_rate": 9.746473684210526e-05,
      "loss": 0.3422,
      "step": 29648
    },
    {
      "epoch": 1.6602643073132488,
      "grad_norm": 1.3935191631317139,
      "learning_rate": 9.746447368421054e-05,
      "loss": 0.3623,
      "step": 29649
    },
    {
      "epoch": 1.6603203046253778,
      "grad_norm": 1.3148080110549927,
      "learning_rate": 9.74642105263158e-05,
      "loss": 0.4853,
      "step": 29650
    },
    {
      "epoch": 1.6603763019375068,
      "grad_norm": 1.0957515239715576,
      "learning_rate": 9.746394736842106e-05,
      "loss": 0.4409,
      "step": 29651
    },
    {
      "epoch": 1.6604322992496359,
      "grad_norm": 1.3667657375335693,
      "learning_rate": 9.746368421052631e-05,
      "loss": 0.5785,
      "step": 29652
    },
    {
      "epoch": 1.6604882965617649,
      "grad_norm": 1.3684552907943726,
      "learning_rate": 9.746342105263159e-05,
      "loss": 0.4376,
      "step": 29653
    },
    {
      "epoch": 1.660544293873894,
      "grad_norm": 1.152071475982666,
      "learning_rate": 9.746315789473685e-05,
      "loss": 0.4768,
      "step": 29654
    },
    {
      "epoch": 1.660600291186023,
      "grad_norm": 3.4706151485443115,
      "learning_rate": 9.746289473684211e-05,
      "loss": 0.5972,
      "step": 29655
    },
    {
      "epoch": 1.660656288498152,
      "grad_norm": 1.246666431427002,
      "learning_rate": 9.746263157894737e-05,
      "loss": 0.4079,
      "step": 29656
    },
    {
      "epoch": 1.660712285810281,
      "grad_norm": 1.2707175016403198,
      "learning_rate": 9.746236842105264e-05,
      "loss": 0.3829,
      "step": 29657
    },
    {
      "epoch": 1.66076828312241,
      "grad_norm": 1.4588186740875244,
      "learning_rate": 9.74621052631579e-05,
      "loss": 0.4542,
      "step": 29658
    },
    {
      "epoch": 1.660824280434539,
      "grad_norm": 1.4935686588287354,
      "learning_rate": 9.746184210526317e-05,
      "loss": 0.393,
      "step": 29659
    },
    {
      "epoch": 1.660880277746668,
      "grad_norm": 1.3570958375930786,
      "learning_rate": 9.746157894736842e-05,
      "loss": 0.3965,
      "step": 29660
    },
    {
      "epoch": 1.660936275058797,
      "grad_norm": 1.4252172708511353,
      "learning_rate": 9.746131578947368e-05,
      "loss": 0.5035,
      "step": 29661
    },
    {
      "epoch": 1.660992272370926,
      "grad_norm": 1.4909636974334717,
      "learning_rate": 9.746105263157895e-05,
      "loss": 0.4325,
      "step": 29662
    },
    {
      "epoch": 1.661048269683055,
      "grad_norm": 1.0536448955535889,
      "learning_rate": 9.746078947368421e-05,
      "loss": 0.2782,
      "step": 29663
    },
    {
      "epoch": 1.661104266995184,
      "grad_norm": 1.175025463104248,
      "learning_rate": 9.746052631578949e-05,
      "loss": 0.3818,
      "step": 29664
    },
    {
      "epoch": 1.6611602643073131,
      "grad_norm": 1.2664695978164673,
      "learning_rate": 9.746026315789473e-05,
      "loss": 0.399,
      "step": 29665
    },
    {
      "epoch": 1.6612162616194421,
      "grad_norm": 1.3744924068450928,
      "learning_rate": 9.746e-05,
      "loss": 0.5242,
      "step": 29666
    },
    {
      "epoch": 1.6612722589315712,
      "grad_norm": 1.2288700342178345,
      "learning_rate": 9.745973684210527e-05,
      "loss": 0.4353,
      "step": 29667
    },
    {
      "epoch": 1.6613282562437002,
      "grad_norm": 1.1383012533187866,
      "learning_rate": 9.745947368421054e-05,
      "loss": 0.3404,
      "step": 29668
    },
    {
      "epoch": 1.6613842535558292,
      "grad_norm": 1.3235902786254883,
      "learning_rate": 9.74592105263158e-05,
      "loss": 0.4674,
      "step": 29669
    },
    {
      "epoch": 1.6614402508679582,
      "grad_norm": 1.4621721506118774,
      "learning_rate": 9.745894736842106e-05,
      "loss": 0.5222,
      "step": 29670
    },
    {
      "epoch": 1.6614962481800872,
      "grad_norm": 1.462956428527832,
      "learning_rate": 9.745868421052632e-05,
      "loss": 0.4607,
      "step": 29671
    },
    {
      "epoch": 1.6615522454922163,
      "grad_norm": 1.5089340209960938,
      "learning_rate": 9.745842105263159e-05,
      "loss": 0.4257,
      "step": 29672
    },
    {
      "epoch": 1.6616082428043453,
      "grad_norm": 1.2238178253173828,
      "learning_rate": 9.745815789473685e-05,
      "loss": 0.5535,
      "step": 29673
    },
    {
      "epoch": 1.6616642401164743,
      "grad_norm": 1.066439151763916,
      "learning_rate": 9.745789473684211e-05,
      "loss": 0.4257,
      "step": 29674
    },
    {
      "epoch": 1.6617202374286033,
      "grad_norm": 1.2708643674850464,
      "learning_rate": 9.745763157894737e-05,
      "loss": 0.3578,
      "step": 29675
    },
    {
      "epoch": 1.6617762347407323,
      "grad_norm": 1.596035361289978,
      "learning_rate": 9.745736842105264e-05,
      "loss": 0.7007,
      "step": 29676
    },
    {
      "epoch": 1.6618322320528613,
      "grad_norm": 1.338188648223877,
      "learning_rate": 9.74571052631579e-05,
      "loss": 0.4382,
      "step": 29677
    },
    {
      "epoch": 1.6618882293649904,
      "grad_norm": 1.3567324876785278,
      "learning_rate": 9.745684210526316e-05,
      "loss": 0.5165,
      "step": 29678
    },
    {
      "epoch": 1.6619442266771194,
      "grad_norm": 1.4828284978866577,
      "learning_rate": 9.745657894736842e-05,
      "loss": 0.693,
      "step": 29679
    },
    {
      "epoch": 1.6620002239892484,
      "grad_norm": 1.2866010665893555,
      "learning_rate": 9.745631578947368e-05,
      "loss": 0.4845,
      "step": 29680
    },
    {
      "epoch": 1.6620562213013774,
      "grad_norm": 1.3637531995773315,
      "learning_rate": 9.745605263157896e-05,
      "loss": 0.3814,
      "step": 29681
    },
    {
      "epoch": 1.6621122186135064,
      "grad_norm": 1.154879093170166,
      "learning_rate": 9.745578947368422e-05,
      "loss": 0.4708,
      "step": 29682
    },
    {
      "epoch": 1.6621682159256355,
      "grad_norm": 1.2102686166763306,
      "learning_rate": 9.745552631578947e-05,
      "loss": 0.543,
      "step": 29683
    },
    {
      "epoch": 1.6622242132377645,
      "grad_norm": 1.2043113708496094,
      "learning_rate": 9.745526315789473e-05,
      "loss": 0.4621,
      "step": 29684
    },
    {
      "epoch": 1.6622802105498935,
      "grad_norm": 1.490852952003479,
      "learning_rate": 9.745500000000001e-05,
      "loss": 0.399,
      "step": 29685
    },
    {
      "epoch": 1.6623362078620225,
      "grad_norm": 1.358458161354065,
      "learning_rate": 9.745473684210527e-05,
      "loss": 0.5194,
      "step": 29686
    },
    {
      "epoch": 1.6623922051741515,
      "grad_norm": 1.3399975299835205,
      "learning_rate": 9.745447368421054e-05,
      "loss": 0.5748,
      "step": 29687
    },
    {
      "epoch": 1.6624482024862806,
      "grad_norm": 1.2152906656265259,
      "learning_rate": 9.745421052631579e-05,
      "loss": 0.4133,
      "step": 29688
    },
    {
      "epoch": 1.6625041997984096,
      "grad_norm": 1.3359047174453735,
      "learning_rate": 9.745394736842106e-05,
      "loss": 0.5731,
      "step": 29689
    },
    {
      "epoch": 1.6625601971105386,
      "grad_norm": 1.3178585767745972,
      "learning_rate": 9.745368421052632e-05,
      "loss": 0.4571,
      "step": 29690
    },
    {
      "epoch": 1.6626161944226676,
      "grad_norm": 1.138414978981018,
      "learning_rate": 9.74534210526316e-05,
      "loss": 0.4158,
      "step": 29691
    },
    {
      "epoch": 1.6626721917347966,
      "grad_norm": 1.1555383205413818,
      "learning_rate": 9.745315789473684e-05,
      "loss": 0.3688,
      "step": 29692
    },
    {
      "epoch": 1.6627281890469257,
      "grad_norm": 1.456110954284668,
      "learning_rate": 9.745289473684211e-05,
      "loss": 0.4896,
      "step": 29693
    },
    {
      "epoch": 1.6627841863590547,
      "grad_norm": 1.2308326959609985,
      "learning_rate": 9.745263157894737e-05,
      "loss": 0.3622,
      "step": 29694
    },
    {
      "epoch": 1.6628401836711837,
      "grad_norm": 1.0821744203567505,
      "learning_rate": 9.745236842105263e-05,
      "loss": 0.3717,
      "step": 29695
    },
    {
      "epoch": 1.6628961809833127,
      "grad_norm": 1.3416963815689087,
      "learning_rate": 9.74521052631579e-05,
      "loss": 0.455,
      "step": 29696
    },
    {
      "epoch": 1.6629521782954417,
      "grad_norm": 1.3744860887527466,
      "learning_rate": 9.745184210526315e-05,
      "loss": 0.3445,
      "step": 29697
    },
    {
      "epoch": 1.6630081756075707,
      "grad_norm": 1.2955223321914673,
      "learning_rate": 9.745157894736843e-05,
      "loss": 0.4848,
      "step": 29698
    },
    {
      "epoch": 1.6630641729196998,
      "grad_norm": 1.7026212215423584,
      "learning_rate": 9.745131578947368e-05,
      "loss": 0.4915,
      "step": 29699
    },
    {
      "epoch": 1.6631201702318288,
      "grad_norm": 1.5713258981704712,
      "learning_rate": 9.745105263157896e-05,
      "loss": 0.4763,
      "step": 29700
    },
    {
      "epoch": 1.6631761675439578,
      "grad_norm": 1.4540106058120728,
      "learning_rate": 9.745078947368422e-05,
      "loss": 0.3707,
      "step": 29701
    },
    {
      "epoch": 1.6632321648560868,
      "grad_norm": 1.380915880203247,
      "learning_rate": 9.745052631578948e-05,
      "loss": 0.477,
      "step": 29702
    },
    {
      "epoch": 1.6632881621682158,
      "grad_norm": 1.3537237644195557,
      "learning_rate": 9.745026315789474e-05,
      "loss": 0.5506,
      "step": 29703
    },
    {
      "epoch": 1.6633441594803449,
      "grad_norm": 1.5406357049942017,
      "learning_rate": 9.745000000000001e-05,
      "loss": 0.767,
      "step": 29704
    },
    {
      "epoch": 1.6634001567924739,
      "grad_norm": 1.2187687158584595,
      "learning_rate": 9.744973684210527e-05,
      "loss": 0.4966,
      "step": 29705
    },
    {
      "epoch": 1.663456154104603,
      "grad_norm": 1.3390312194824219,
      "learning_rate": 9.744947368421053e-05,
      "loss": 0.4701,
      "step": 29706
    },
    {
      "epoch": 1.663512151416732,
      "grad_norm": 1.160866618156433,
      "learning_rate": 9.744921052631579e-05,
      "loss": 0.4631,
      "step": 29707
    },
    {
      "epoch": 1.663568148728861,
      "grad_norm": 1.4030683040618896,
      "learning_rate": 9.744894736842106e-05,
      "loss": 0.5906,
      "step": 29708
    },
    {
      "epoch": 1.66362414604099,
      "grad_norm": 1.3373100757598877,
      "learning_rate": 9.744868421052632e-05,
      "loss": 0.4799,
      "step": 29709
    },
    {
      "epoch": 1.663680143353119,
      "grad_norm": 1.0900112390518188,
      "learning_rate": 9.744842105263158e-05,
      "loss": 0.3698,
      "step": 29710
    },
    {
      "epoch": 1.663736140665248,
      "grad_norm": 1.1846494674682617,
      "learning_rate": 9.744815789473684e-05,
      "loss": 0.5321,
      "step": 29711
    },
    {
      "epoch": 1.663792137977377,
      "grad_norm": 1.3239916563034058,
      "learning_rate": 9.74478947368421e-05,
      "loss": 0.4737,
      "step": 29712
    },
    {
      "epoch": 1.663848135289506,
      "grad_norm": 1.345616102218628,
      "learning_rate": 9.744763157894738e-05,
      "loss": 0.4982,
      "step": 29713
    },
    {
      "epoch": 1.663904132601635,
      "grad_norm": 1.169311285018921,
      "learning_rate": 9.744736842105263e-05,
      "loss": 0.3796,
      "step": 29714
    },
    {
      "epoch": 1.663960129913764,
      "grad_norm": 1.1881449222564697,
      "learning_rate": 9.74471052631579e-05,
      "loss": 0.3696,
      "step": 29715
    },
    {
      "epoch": 1.664016127225893,
      "grad_norm": 1.3634545803070068,
      "learning_rate": 9.744684210526315e-05,
      "loss": 0.4455,
      "step": 29716
    },
    {
      "epoch": 1.6640721245380221,
      "grad_norm": 1.3136138916015625,
      "learning_rate": 9.744657894736843e-05,
      "loss": 0.538,
      "step": 29717
    },
    {
      "epoch": 1.6641281218501511,
      "grad_norm": 1.3176722526550293,
      "learning_rate": 9.744631578947369e-05,
      "loss": 0.5301,
      "step": 29718
    },
    {
      "epoch": 1.6641841191622802,
      "grad_norm": 1.1613926887512207,
      "learning_rate": 9.744605263157896e-05,
      "loss": 0.4518,
      "step": 29719
    },
    {
      "epoch": 1.6642401164744092,
      "grad_norm": 1.5034033060073853,
      "learning_rate": 9.744578947368421e-05,
      "loss": 0.5178,
      "step": 29720
    },
    {
      "epoch": 1.6642961137865382,
      "grad_norm": 1.380826711654663,
      "learning_rate": 9.744552631578948e-05,
      "loss": 0.3322,
      "step": 29721
    },
    {
      "epoch": 1.6643521110986672,
      "grad_norm": 1.1774296760559082,
      "learning_rate": 9.744526315789474e-05,
      "loss": 0.4113,
      "step": 29722
    },
    {
      "epoch": 1.6644081084107962,
      "grad_norm": 1.336766004562378,
      "learning_rate": 9.744500000000001e-05,
      "loss": 0.5364,
      "step": 29723
    },
    {
      "epoch": 1.6644641057229252,
      "grad_norm": 1.1515580415725708,
      "learning_rate": 9.744473684210527e-05,
      "loss": 0.4323,
      "step": 29724
    },
    {
      "epoch": 1.6645201030350543,
      "grad_norm": 1.148767352104187,
      "learning_rate": 9.744447368421053e-05,
      "loss": 0.3941,
      "step": 29725
    },
    {
      "epoch": 1.6645761003471833,
      "grad_norm": 1.420762538909912,
      "learning_rate": 9.744421052631579e-05,
      "loss": 0.4618,
      "step": 29726
    },
    {
      "epoch": 1.6646320976593123,
      "grad_norm": 1.3800610303878784,
      "learning_rate": 9.744394736842107e-05,
      "loss": 0.5246,
      "step": 29727
    },
    {
      "epoch": 1.6646880949714413,
      "grad_norm": 1.1994104385375977,
      "learning_rate": 9.744368421052633e-05,
      "loss": 0.4552,
      "step": 29728
    },
    {
      "epoch": 1.6647440922835703,
      "grad_norm": 1.433768391609192,
      "learning_rate": 9.744342105263157e-05,
      "loss": 0.4113,
      "step": 29729
    },
    {
      "epoch": 1.6648000895956994,
      "grad_norm": 1.3951174020767212,
      "learning_rate": 9.744315789473684e-05,
      "loss": 0.6001,
      "step": 29730
    },
    {
      "epoch": 1.6648560869078284,
      "grad_norm": 1.1915291547775269,
      "learning_rate": 9.74428947368421e-05,
      "loss": 0.484,
      "step": 29731
    },
    {
      "epoch": 1.6649120842199574,
      "grad_norm": 4.183198928833008,
      "learning_rate": 9.744263157894738e-05,
      "loss": 0.4532,
      "step": 29732
    },
    {
      "epoch": 1.6649680815320864,
      "grad_norm": 1.5106087923049927,
      "learning_rate": 9.744236842105264e-05,
      "loss": 0.4033,
      "step": 29733
    },
    {
      "epoch": 1.6650240788442154,
      "grad_norm": 1.2340867519378662,
      "learning_rate": 9.74421052631579e-05,
      "loss": 0.3915,
      "step": 29734
    },
    {
      "epoch": 1.6650800761563445,
      "grad_norm": 1.3168102502822876,
      "learning_rate": 9.744184210526316e-05,
      "loss": 0.4267,
      "step": 29735
    },
    {
      "epoch": 1.6651360734684735,
      "grad_norm": 1.1634732484817505,
      "learning_rate": 9.744157894736843e-05,
      "loss": 0.36,
      "step": 29736
    },
    {
      "epoch": 1.6651920707806025,
      "grad_norm": 1.2173901796340942,
      "learning_rate": 9.744131578947369e-05,
      "loss": 0.5981,
      "step": 29737
    },
    {
      "epoch": 1.6652480680927315,
      "grad_norm": 1.174443244934082,
      "learning_rate": 9.744105263157895e-05,
      "loss": 0.5204,
      "step": 29738
    },
    {
      "epoch": 1.6653040654048605,
      "grad_norm": 1.5484994649887085,
      "learning_rate": 9.744078947368421e-05,
      "loss": 0.3388,
      "step": 29739
    },
    {
      "epoch": 1.6653600627169896,
      "grad_norm": 1.8700287342071533,
      "learning_rate": 9.744052631578948e-05,
      "loss": 0.6348,
      "step": 29740
    },
    {
      "epoch": 1.6654160600291186,
      "grad_norm": 1.558946132659912,
      "learning_rate": 9.744026315789474e-05,
      "loss": 0.4442,
      "step": 29741
    },
    {
      "epoch": 1.6654720573412476,
      "grad_norm": 1.4027788639068604,
      "learning_rate": 9.744000000000002e-05,
      "loss": 0.4474,
      "step": 29742
    },
    {
      "epoch": 1.6655280546533766,
      "grad_norm": 1.3163743019104004,
      "learning_rate": 9.743973684210526e-05,
      "loss": 0.3866,
      "step": 29743
    },
    {
      "epoch": 1.6655840519655056,
      "grad_norm": 1.152384638786316,
      "learning_rate": 9.743947368421054e-05,
      "loss": 0.4037,
      "step": 29744
    },
    {
      "epoch": 1.6656400492776346,
      "grad_norm": 1.4696251153945923,
      "learning_rate": 9.74392105263158e-05,
      "loss": 0.5558,
      "step": 29745
    },
    {
      "epoch": 1.6656960465897637,
      "grad_norm": 1.2139456272125244,
      "learning_rate": 9.743894736842105e-05,
      "loss": 0.3967,
      "step": 29746
    },
    {
      "epoch": 1.6657520439018927,
      "grad_norm": 1.3266783952713013,
      "learning_rate": 9.743868421052631e-05,
      "loss": 0.4887,
      "step": 29747
    },
    {
      "epoch": 1.6658080412140217,
      "grad_norm": 1.4412658214569092,
      "learning_rate": 9.743842105263157e-05,
      "loss": 0.5102,
      "step": 29748
    },
    {
      "epoch": 1.6658640385261507,
      "grad_norm": 1.185046672821045,
      "learning_rate": 9.743815789473685e-05,
      "loss": 0.341,
      "step": 29749
    },
    {
      "epoch": 1.6659200358382797,
      "grad_norm": 1.3606398105621338,
      "learning_rate": 9.743789473684211e-05,
      "loss": 0.5179,
      "step": 29750
    },
    {
      "epoch": 1.6659760331504088,
      "grad_norm": 1.2505133152008057,
      "learning_rate": 9.743763157894738e-05,
      "loss": 0.4396,
      "step": 29751
    },
    {
      "epoch": 1.6660320304625378,
      "grad_norm": 1.2264609336853027,
      "learning_rate": 9.743736842105263e-05,
      "loss": 0.3688,
      "step": 29752
    },
    {
      "epoch": 1.6660880277746668,
      "grad_norm": 1.205929160118103,
      "learning_rate": 9.74371052631579e-05,
      "loss": 0.4127,
      "step": 29753
    },
    {
      "epoch": 1.6661440250867958,
      "grad_norm": 1.192292332649231,
      "learning_rate": 9.743684210526316e-05,
      "loss": 0.4382,
      "step": 29754
    },
    {
      "epoch": 1.6662000223989248,
      "grad_norm": 15.911165237426758,
      "learning_rate": 9.743657894736843e-05,
      "loss": 0.3512,
      "step": 29755
    },
    {
      "epoch": 1.6662560197110539,
      "grad_norm": 1.7799100875854492,
      "learning_rate": 9.743631578947369e-05,
      "loss": 0.5259,
      "step": 29756
    },
    {
      "epoch": 1.6663120170231829,
      "grad_norm": 3.8827314376831055,
      "learning_rate": 9.743605263157895e-05,
      "loss": 0.5644,
      "step": 29757
    },
    {
      "epoch": 1.666368014335312,
      "grad_norm": 1.453129529953003,
      "learning_rate": 9.743578947368421e-05,
      "loss": 0.6702,
      "step": 29758
    },
    {
      "epoch": 1.666424011647441,
      "grad_norm": 1.8428001403808594,
      "learning_rate": 9.743552631578949e-05,
      "loss": 0.4429,
      "step": 29759
    },
    {
      "epoch": 1.66648000895957,
      "grad_norm": 1.1079034805297852,
      "learning_rate": 9.743526315789475e-05,
      "loss": 0.4014,
      "step": 29760
    },
    {
      "epoch": 1.666536006271699,
      "grad_norm": 1.9630281925201416,
      "learning_rate": 9.7435e-05,
      "loss": 0.5883,
      "step": 29761
    },
    {
      "epoch": 1.666592003583828,
      "grad_norm": 2.018019914627075,
      "learning_rate": 9.743473684210526e-05,
      "loss": 0.6178,
      "step": 29762
    },
    {
      "epoch": 1.666648000895957,
      "grad_norm": 1.1073989868164062,
      "learning_rate": 9.743447368421052e-05,
      "loss": 0.4937,
      "step": 29763
    },
    {
      "epoch": 1.666703998208086,
      "grad_norm": 1.2717527151107788,
      "learning_rate": 9.74342105263158e-05,
      "loss": 0.5471,
      "step": 29764
    },
    {
      "epoch": 1.666759995520215,
      "grad_norm": 1.579235315322876,
      "learning_rate": 9.743394736842106e-05,
      "loss": 0.4843,
      "step": 29765
    },
    {
      "epoch": 1.666815992832344,
      "grad_norm": 1.0931422710418701,
      "learning_rate": 9.743368421052632e-05,
      "loss": 0.3776,
      "step": 29766
    },
    {
      "epoch": 1.666871990144473,
      "grad_norm": 1.1754850149154663,
      "learning_rate": 9.743342105263158e-05,
      "loss": 0.3256,
      "step": 29767
    },
    {
      "epoch": 1.666927987456602,
      "grad_norm": 1.2110705375671387,
      "learning_rate": 9.743315789473685e-05,
      "loss": 0.4423,
      "step": 29768
    },
    {
      "epoch": 1.666983984768731,
      "grad_norm": 1.223608374595642,
      "learning_rate": 9.743289473684211e-05,
      "loss": 0.4686,
      "step": 29769
    },
    {
      "epoch": 1.6670399820808601,
      "grad_norm": 1.2251781225204468,
      "learning_rate": 9.743263157894737e-05,
      "loss": 0.432,
      "step": 29770
    },
    {
      "epoch": 1.6670959793929891,
      "grad_norm": 1.308613657951355,
      "learning_rate": 9.743236842105263e-05,
      "loss": 0.531,
      "step": 29771
    },
    {
      "epoch": 1.6671519767051182,
      "grad_norm": 1.401737928390503,
      "learning_rate": 9.74321052631579e-05,
      "loss": 0.5821,
      "step": 29772
    },
    {
      "epoch": 1.6672079740172472,
      "grad_norm": 1.3729923963546753,
      "learning_rate": 9.743184210526316e-05,
      "loss": 0.4739,
      "step": 29773
    },
    {
      "epoch": 1.6672639713293762,
      "grad_norm": 1.2187713384628296,
      "learning_rate": 9.743157894736844e-05,
      "loss": 0.5502,
      "step": 29774
    },
    {
      "epoch": 1.6673199686415052,
      "grad_norm": 1.1433850526809692,
      "learning_rate": 9.743131578947368e-05,
      "loss": 0.443,
      "step": 29775
    },
    {
      "epoch": 1.6673759659536342,
      "grad_norm": 1.6556144952774048,
      "learning_rate": 9.743105263157895e-05,
      "loss": 0.5669,
      "step": 29776
    },
    {
      "epoch": 1.6674319632657633,
      "grad_norm": 1.1630947589874268,
      "learning_rate": 9.743078947368421e-05,
      "loss": 0.3854,
      "step": 29777
    },
    {
      "epoch": 1.6674879605778923,
      "grad_norm": 1.1792995929718018,
      "learning_rate": 9.743052631578949e-05,
      "loss": 0.4185,
      "step": 29778
    },
    {
      "epoch": 1.6675439578900213,
      "grad_norm": 1.5685478448867798,
      "learning_rate": 9.743026315789475e-05,
      "loss": 0.5321,
      "step": 29779
    },
    {
      "epoch": 1.6675999552021503,
      "grad_norm": 1.180819034576416,
      "learning_rate": 9.743000000000001e-05,
      "loss": 0.3929,
      "step": 29780
    },
    {
      "epoch": 1.6676559525142793,
      "grad_norm": 1.367296814918518,
      "learning_rate": 9.742973684210527e-05,
      "loss": 0.4154,
      "step": 29781
    },
    {
      "epoch": 1.6677119498264084,
      "grad_norm": 1.4089455604553223,
      "learning_rate": 9.742947368421053e-05,
      "loss": 0.4036,
      "step": 29782
    },
    {
      "epoch": 1.6677679471385374,
      "grad_norm": 1.3270117044448853,
      "learning_rate": 9.74292105263158e-05,
      "loss": 0.5185,
      "step": 29783
    },
    {
      "epoch": 1.6678239444506664,
      "grad_norm": 1.379282832145691,
      "learning_rate": 9.742894736842105e-05,
      "loss": 0.3957,
      "step": 29784
    },
    {
      "epoch": 1.6678799417627954,
      "grad_norm": 1.2152642011642456,
      "learning_rate": 9.742868421052632e-05,
      "loss": 0.4632,
      "step": 29785
    },
    {
      "epoch": 1.6679359390749244,
      "grad_norm": 1.2232376337051392,
      "learning_rate": 9.742842105263158e-05,
      "loss": 0.4157,
      "step": 29786
    },
    {
      "epoch": 1.6679919363870535,
      "grad_norm": 2.0165438652038574,
      "learning_rate": 9.742815789473685e-05,
      "loss": 0.414,
      "step": 29787
    },
    {
      "epoch": 1.6680479336991825,
      "grad_norm": 44.752079010009766,
      "learning_rate": 9.742789473684211e-05,
      "loss": 0.388,
      "step": 29788
    },
    {
      "epoch": 1.6681039310113115,
      "grad_norm": 1.298831820487976,
      "learning_rate": 9.742763157894737e-05,
      "loss": 0.6055,
      "step": 29789
    },
    {
      "epoch": 1.6681599283234405,
      "grad_norm": 1.1803098917007446,
      "learning_rate": 9.742736842105263e-05,
      "loss": 0.3863,
      "step": 29790
    },
    {
      "epoch": 1.6682159256355695,
      "grad_norm": 1.267655849456787,
      "learning_rate": 9.74271052631579e-05,
      "loss": 0.4228,
      "step": 29791
    },
    {
      "epoch": 1.6682719229476985,
      "grad_norm": 1.3102296590805054,
      "learning_rate": 9.742684210526316e-05,
      "loss": 0.4232,
      "step": 29792
    },
    {
      "epoch": 1.6683279202598276,
      "grad_norm": 1.2054402828216553,
      "learning_rate": 9.742657894736842e-05,
      "loss": 0.4745,
      "step": 29793
    },
    {
      "epoch": 1.6683839175719566,
      "grad_norm": 1.2526195049285889,
      "learning_rate": 9.742631578947368e-05,
      "loss": 0.3803,
      "step": 29794
    },
    {
      "epoch": 1.6684399148840856,
      "grad_norm": 1.429464340209961,
      "learning_rate": 9.742605263157896e-05,
      "loss": 0.5537,
      "step": 29795
    },
    {
      "epoch": 1.6684959121962146,
      "grad_norm": 1.1986188888549805,
      "learning_rate": 9.742578947368422e-05,
      "loss": 0.3575,
      "step": 29796
    },
    {
      "epoch": 1.6685519095083436,
      "grad_norm": 1.2700387239456177,
      "learning_rate": 9.742552631578949e-05,
      "loss": 0.4072,
      "step": 29797
    },
    {
      "epoch": 1.6686079068204727,
      "grad_norm": 1.1243160963058472,
      "learning_rate": 9.742526315789474e-05,
      "loss": 0.453,
      "step": 29798
    },
    {
      "epoch": 1.6686639041326017,
      "grad_norm": 1.2256789207458496,
      "learning_rate": 9.7425e-05,
      "loss": 0.3341,
      "step": 29799
    },
    {
      "epoch": 1.6687199014447307,
      "grad_norm": 1.7087154388427734,
      "learning_rate": 9.742473684210527e-05,
      "loss": 0.4214,
      "step": 29800
    },
    {
      "epoch": 1.6687758987568597,
      "grad_norm": 1.2494076490402222,
      "learning_rate": 9.742447368421053e-05,
      "loss": 0.609,
      "step": 29801
    },
    {
      "epoch": 1.6688318960689887,
      "grad_norm": 1.8516823053359985,
      "learning_rate": 9.742421052631579e-05,
      "loss": 0.411,
      "step": 29802
    },
    {
      "epoch": 1.6688878933811178,
      "grad_norm": 1.2978487014770508,
      "learning_rate": 9.742394736842105e-05,
      "loss": 0.5197,
      "step": 29803
    },
    {
      "epoch": 1.6689438906932468,
      "grad_norm": 1.579337239265442,
      "learning_rate": 9.742368421052632e-05,
      "loss": 0.5147,
      "step": 29804
    },
    {
      "epoch": 1.6689998880053758,
      "grad_norm": 1.4517598152160645,
      "learning_rate": 9.742342105263158e-05,
      "loss": 0.5645,
      "step": 29805
    },
    {
      "epoch": 1.6690558853175048,
      "grad_norm": 1.1208311319351196,
      "learning_rate": 9.742315789473686e-05,
      "loss": 0.4752,
      "step": 29806
    },
    {
      "epoch": 1.6691118826296338,
      "grad_norm": 1.1679915189743042,
      "learning_rate": 9.74228947368421e-05,
      "loss": 0.4083,
      "step": 29807
    },
    {
      "epoch": 1.6691678799417629,
      "grad_norm": 1.585705041885376,
      "learning_rate": 9.742263157894737e-05,
      "loss": 0.4569,
      "step": 29808
    },
    {
      "epoch": 1.6692238772538919,
      "grad_norm": 1.0873582363128662,
      "learning_rate": 9.742236842105263e-05,
      "loss": 0.39,
      "step": 29809
    },
    {
      "epoch": 1.669279874566021,
      "grad_norm": 1.1331446170806885,
      "learning_rate": 9.742210526315791e-05,
      "loss": 0.4536,
      "step": 29810
    },
    {
      "epoch": 1.66933587187815,
      "grad_norm": 1.2082889080047607,
      "learning_rate": 9.742184210526317e-05,
      "loss": 0.4531,
      "step": 29811
    },
    {
      "epoch": 1.669391869190279,
      "grad_norm": 1.4118036031723022,
      "learning_rate": 9.742157894736843e-05,
      "loss": 0.3652,
      "step": 29812
    },
    {
      "epoch": 1.669447866502408,
      "grad_norm": 1.1882736682891846,
      "learning_rate": 9.742131578947369e-05,
      "loss": 0.3823,
      "step": 29813
    },
    {
      "epoch": 1.669503863814537,
      "grad_norm": 1.3445663452148438,
      "learning_rate": 9.742105263157896e-05,
      "loss": 0.4056,
      "step": 29814
    },
    {
      "epoch": 1.669559861126666,
      "grad_norm": 1.6700886487960815,
      "learning_rate": 9.742078947368422e-05,
      "loss": 0.7545,
      "step": 29815
    },
    {
      "epoch": 1.669615858438795,
      "grad_norm": 1.2657943964004517,
      "learning_rate": 9.742052631578948e-05,
      "loss": 0.4648,
      "step": 29816
    },
    {
      "epoch": 1.669671855750924,
      "grad_norm": 1.2922332286834717,
      "learning_rate": 9.742026315789474e-05,
      "loss": 0.4243,
      "step": 29817
    },
    {
      "epoch": 1.669727853063053,
      "grad_norm": 1.136527419090271,
      "learning_rate": 9.742e-05,
      "loss": 0.4334,
      "step": 29818
    },
    {
      "epoch": 1.669783850375182,
      "grad_norm": 1.1991740465164185,
      "learning_rate": 9.741973684210527e-05,
      "loss": 0.4208,
      "step": 29819
    },
    {
      "epoch": 1.669839847687311,
      "grad_norm": 1.3565469980239868,
      "learning_rate": 9.741947368421053e-05,
      "loss": 0.5002,
      "step": 29820
    },
    {
      "epoch": 1.66989584499944,
      "grad_norm": 1.618970274925232,
      "learning_rate": 9.741921052631579e-05,
      "loss": 0.446,
      "step": 29821
    },
    {
      "epoch": 1.6699518423115691,
      "grad_norm": 2.0439717769622803,
      "learning_rate": 9.741894736842105e-05,
      "loss": 0.5478,
      "step": 29822
    },
    {
      "epoch": 1.6700078396236981,
      "grad_norm": 1.4054450988769531,
      "learning_rate": 9.741868421052632e-05,
      "loss": 0.4755,
      "step": 29823
    },
    {
      "epoch": 1.6700638369358272,
      "grad_norm": 1.5218238830566406,
      "learning_rate": 9.741842105263158e-05,
      "loss": 0.4832,
      "step": 29824
    },
    {
      "epoch": 1.6701198342479562,
      "grad_norm": 1.2585930824279785,
      "learning_rate": 9.741815789473684e-05,
      "loss": 0.6058,
      "step": 29825
    },
    {
      "epoch": 1.6701758315600852,
      "grad_norm": 1.4107081890106201,
      "learning_rate": 9.74178947368421e-05,
      "loss": 0.4306,
      "step": 29826
    },
    {
      "epoch": 1.6702318288722142,
      "grad_norm": 1.1761633157730103,
      "learning_rate": 9.741763157894738e-05,
      "loss": 0.4319,
      "step": 29827
    },
    {
      "epoch": 1.6702878261843432,
      "grad_norm": 1.3887361288070679,
      "learning_rate": 9.741736842105264e-05,
      "loss": 0.6162,
      "step": 29828
    },
    {
      "epoch": 1.6703438234964723,
      "grad_norm": 1.3429697751998901,
      "learning_rate": 9.741710526315791e-05,
      "loss": 0.4642,
      "step": 29829
    },
    {
      "epoch": 1.6703998208086013,
      "grad_norm": 1.3777918815612793,
      "learning_rate": 9.741684210526316e-05,
      "loss": 0.5079,
      "step": 29830
    },
    {
      "epoch": 1.6704558181207303,
      "grad_norm": 1.3541017770767212,
      "learning_rate": 9.741657894736843e-05,
      "loss": 0.513,
      "step": 29831
    },
    {
      "epoch": 1.6705118154328593,
      "grad_norm": 1.4516394138336182,
      "learning_rate": 9.741631578947369e-05,
      "loss": 0.4262,
      "step": 29832
    },
    {
      "epoch": 1.6705678127449883,
      "grad_norm": 1.2368522882461548,
      "learning_rate": 9.741605263157895e-05,
      "loss": 0.4667,
      "step": 29833
    },
    {
      "epoch": 1.6706238100571174,
      "grad_norm": 1.223478078842163,
      "learning_rate": 9.741578947368422e-05,
      "loss": 0.4711,
      "step": 29834
    },
    {
      "epoch": 1.6706798073692464,
      "grad_norm": 1.3523377180099487,
      "learning_rate": 9.741552631578947e-05,
      "loss": 0.4679,
      "step": 29835
    },
    {
      "epoch": 1.6707358046813754,
      "grad_norm": 1.3053621053695679,
      "learning_rate": 9.741526315789474e-05,
      "loss": 0.3716,
      "step": 29836
    },
    {
      "epoch": 1.6707918019935044,
      "grad_norm": 2.675323247909546,
      "learning_rate": 9.7415e-05,
      "loss": 0.3351,
      "step": 29837
    },
    {
      "epoch": 1.6708477993056334,
      "grad_norm": 1.5996848344802856,
      "learning_rate": 9.741473684210527e-05,
      "loss": 0.3857,
      "step": 29838
    },
    {
      "epoch": 1.6709037966177624,
      "grad_norm": 1.1878904104232788,
      "learning_rate": 9.741447368421052e-05,
      "loss": 0.4158,
      "step": 29839
    },
    {
      "epoch": 1.6709597939298915,
      "grad_norm": 1.2107948064804077,
      "learning_rate": 9.74142105263158e-05,
      "loss": 0.4259,
      "step": 29840
    },
    {
      "epoch": 1.6710157912420205,
      "grad_norm": 2.0347845554351807,
      "learning_rate": 9.741394736842105e-05,
      "loss": 0.4988,
      "step": 29841
    },
    {
      "epoch": 1.6710717885541495,
      "grad_norm": 1.2873926162719727,
      "learning_rate": 9.741368421052633e-05,
      "loss": 0.4446,
      "step": 29842
    },
    {
      "epoch": 1.6711277858662785,
      "grad_norm": 1.201278805732727,
      "learning_rate": 9.741342105263159e-05,
      "loss": 0.4185,
      "step": 29843
    },
    {
      "epoch": 1.6711837831784075,
      "grad_norm": 1.1728787422180176,
      "learning_rate": 9.741315789473685e-05,
      "loss": 0.3828,
      "step": 29844
    },
    {
      "epoch": 1.6712397804905366,
      "grad_norm": 1.3575496673583984,
      "learning_rate": 9.74128947368421e-05,
      "loss": 0.4228,
      "step": 29845
    },
    {
      "epoch": 1.6712957778026656,
      "grad_norm": 1.3685311079025269,
      "learning_rate": 9.741263157894738e-05,
      "loss": 0.4682,
      "step": 29846
    },
    {
      "epoch": 1.6713517751147946,
      "grad_norm": 1.4752414226531982,
      "learning_rate": 9.741236842105264e-05,
      "loss": 0.3884,
      "step": 29847
    },
    {
      "epoch": 1.6714077724269236,
      "grad_norm": 1.190846562385559,
      "learning_rate": 9.74121052631579e-05,
      "loss": 0.4626,
      "step": 29848
    },
    {
      "epoch": 1.6714637697390526,
      "grad_norm": 1.1097655296325684,
      "learning_rate": 9.741184210526316e-05,
      "loss": 0.414,
      "step": 29849
    },
    {
      "epoch": 1.6715197670511817,
      "grad_norm": 1.0734790563583374,
      "learning_rate": 9.741157894736842e-05,
      "loss": 0.4411,
      "step": 29850
    },
    {
      "epoch": 1.6715757643633107,
      "grad_norm": 1.2061095237731934,
      "learning_rate": 9.741131578947369e-05,
      "loss": 0.4699,
      "step": 29851
    },
    {
      "epoch": 1.6716317616754397,
      "grad_norm": 2.0681514739990234,
      "learning_rate": 9.741105263157895e-05,
      "loss": 0.4441,
      "step": 29852
    },
    {
      "epoch": 1.6716877589875687,
      "grad_norm": 1.153032660484314,
      "learning_rate": 9.741078947368421e-05,
      "loss": 0.4403,
      "step": 29853
    },
    {
      "epoch": 1.6717437562996977,
      "grad_norm": 1.0701205730438232,
      "learning_rate": 9.741052631578947e-05,
      "loss": 0.3503,
      "step": 29854
    },
    {
      "epoch": 1.6717997536118268,
      "grad_norm": 1.1991156339645386,
      "learning_rate": 9.741026315789474e-05,
      "loss": 0.4324,
      "step": 29855
    },
    {
      "epoch": 1.6718557509239558,
      "grad_norm": 1.3591337203979492,
      "learning_rate": 9.741e-05,
      "loss": 0.5427,
      "step": 29856
    },
    {
      "epoch": 1.6719117482360848,
      "grad_norm": 1.1301558017730713,
      "learning_rate": 9.740973684210526e-05,
      "loss": 0.4113,
      "step": 29857
    },
    {
      "epoch": 1.6719677455482138,
      "grad_norm": 1.305549144744873,
      "learning_rate": 9.740947368421052e-05,
      "loss": 0.5043,
      "step": 29858
    },
    {
      "epoch": 1.6720237428603428,
      "grad_norm": 3.242582321166992,
      "learning_rate": 9.74092105263158e-05,
      "loss": 0.5073,
      "step": 29859
    },
    {
      "epoch": 1.6720797401724719,
      "grad_norm": 1.270106315612793,
      "learning_rate": 9.740894736842106e-05,
      "loss": 0.3616,
      "step": 29860
    },
    {
      "epoch": 1.6721357374846009,
      "grad_norm": 1.0806063413619995,
      "learning_rate": 9.740868421052633e-05,
      "loss": 0.4177,
      "step": 29861
    },
    {
      "epoch": 1.67219173479673,
      "grad_norm": 1.2932080030441284,
      "learning_rate": 9.740842105263158e-05,
      "loss": 0.4994,
      "step": 29862
    },
    {
      "epoch": 1.672247732108859,
      "grad_norm": 1.2522165775299072,
      "learning_rate": 9.740815789473685e-05,
      "loss": 0.3813,
      "step": 29863
    },
    {
      "epoch": 1.672303729420988,
      "grad_norm": 1.3221081495285034,
      "learning_rate": 9.740789473684211e-05,
      "loss": 0.5046,
      "step": 29864
    },
    {
      "epoch": 1.672359726733117,
      "grad_norm": 3.6724979877471924,
      "learning_rate": 9.740763157894738e-05,
      "loss": 0.4389,
      "step": 29865
    },
    {
      "epoch": 1.672415724045246,
      "grad_norm": 1.2772105932235718,
      "learning_rate": 9.740736842105264e-05,
      "loss": 0.4686,
      "step": 29866
    },
    {
      "epoch": 1.672471721357375,
      "grad_norm": 1.2992759943008423,
      "learning_rate": 9.740710526315789e-05,
      "loss": 0.5575,
      "step": 29867
    },
    {
      "epoch": 1.672527718669504,
      "grad_norm": 1.3870601654052734,
      "learning_rate": 9.740684210526316e-05,
      "loss": 0.4174,
      "step": 29868
    },
    {
      "epoch": 1.672583715981633,
      "grad_norm": 1.2118297815322876,
      "learning_rate": 9.740657894736842e-05,
      "loss": 0.4362,
      "step": 29869
    },
    {
      "epoch": 1.672639713293762,
      "grad_norm": 1.3227561712265015,
      "learning_rate": 9.74063157894737e-05,
      "loss": 0.3491,
      "step": 29870
    },
    {
      "epoch": 1.672695710605891,
      "grad_norm": 1.1811689138412476,
      "learning_rate": 9.740605263157895e-05,
      "loss": 0.5631,
      "step": 29871
    },
    {
      "epoch": 1.67275170791802,
      "grad_norm": 1.6219584941864014,
      "learning_rate": 9.740578947368421e-05,
      "loss": 0.6717,
      "step": 29872
    },
    {
      "epoch": 1.672807705230149,
      "grad_norm": 1.3326748609542847,
      "learning_rate": 9.740552631578947e-05,
      "loss": 0.4784,
      "step": 29873
    },
    {
      "epoch": 1.6728637025422781,
      "grad_norm": 1.8884485960006714,
      "learning_rate": 9.740526315789475e-05,
      "loss": 0.3688,
      "step": 29874
    },
    {
      "epoch": 1.6729196998544071,
      "grad_norm": 1.4678939580917358,
      "learning_rate": 9.7405e-05,
      "loss": 0.4771,
      "step": 29875
    },
    {
      "epoch": 1.6729756971665362,
      "grad_norm": 1.3649877309799194,
      "learning_rate": 9.740473684210527e-05,
      "loss": 0.4531,
      "step": 29876
    },
    {
      "epoch": 1.6730316944786652,
      "grad_norm": 1.416639804840088,
      "learning_rate": 9.740447368421053e-05,
      "loss": 0.7262,
      "step": 29877
    },
    {
      "epoch": 1.6730876917907942,
      "grad_norm": 1.2157752513885498,
      "learning_rate": 9.74042105263158e-05,
      "loss": 0.4003,
      "step": 29878
    },
    {
      "epoch": 1.6731436891029232,
      "grad_norm": 1.248354196548462,
      "learning_rate": 9.740394736842106e-05,
      "loss": 0.5036,
      "step": 29879
    },
    {
      "epoch": 1.6731996864150522,
      "grad_norm": 1.21266770362854,
      "learning_rate": 9.740368421052632e-05,
      "loss": 0.432,
      "step": 29880
    },
    {
      "epoch": 1.6732556837271813,
      "grad_norm": 1.730705976486206,
      "learning_rate": 9.740342105263158e-05,
      "loss": 0.4821,
      "step": 29881
    },
    {
      "epoch": 1.6733116810393103,
      "grad_norm": 1.1938798427581787,
      "learning_rate": 9.740315789473685e-05,
      "loss": 0.4414,
      "step": 29882
    },
    {
      "epoch": 1.6733676783514393,
      "grad_norm": 7.105051040649414,
      "learning_rate": 9.740289473684211e-05,
      "loss": 0.4072,
      "step": 29883
    },
    {
      "epoch": 1.6734236756635683,
      "grad_norm": 1.474406123161316,
      "learning_rate": 9.740263157894737e-05,
      "loss": 0.5356,
      "step": 29884
    },
    {
      "epoch": 1.6734796729756973,
      "grad_norm": 1.5218783617019653,
      "learning_rate": 9.740236842105263e-05,
      "loss": 0.5101,
      "step": 29885
    },
    {
      "epoch": 1.6735356702878263,
      "grad_norm": 1.516426682472229,
      "learning_rate": 9.740210526315789e-05,
      "loss": 0.5078,
      "step": 29886
    },
    {
      "epoch": 1.6735916675999554,
      "grad_norm": 1.0792291164398193,
      "learning_rate": 9.740184210526316e-05,
      "loss": 0.4085,
      "step": 29887
    },
    {
      "epoch": 1.6736476649120842,
      "grad_norm": 1.207808256149292,
      "learning_rate": 9.740157894736842e-05,
      "loss": 0.3622,
      "step": 29888
    },
    {
      "epoch": 1.6737036622242132,
      "grad_norm": 1.3278062343597412,
      "learning_rate": 9.74013157894737e-05,
      "loss": 0.4181,
      "step": 29889
    },
    {
      "epoch": 1.6737596595363422,
      "grad_norm": 1.1830133199691772,
      "learning_rate": 9.740105263157894e-05,
      "loss": 0.3895,
      "step": 29890
    },
    {
      "epoch": 1.6738156568484712,
      "grad_norm": 1.3510818481445312,
      "learning_rate": 9.740078947368422e-05,
      "loss": 0.3787,
      "step": 29891
    },
    {
      "epoch": 1.6738716541606002,
      "grad_norm": 1.8418278694152832,
      "learning_rate": 9.740052631578948e-05,
      "loss": 0.4635,
      "step": 29892
    },
    {
      "epoch": 1.6739276514727293,
      "grad_norm": 1.1231051683425903,
      "learning_rate": 9.740026315789475e-05,
      "loss": 0.417,
      "step": 29893
    },
    {
      "epoch": 1.6739836487848583,
      "grad_norm": 1.301915168762207,
      "learning_rate": 9.74e-05,
      "loss": 0.3922,
      "step": 29894
    },
    {
      "epoch": 1.6740396460969873,
      "grad_norm": 1.3169394731521606,
      "learning_rate": 9.739973684210527e-05,
      "loss": 0.372,
      "step": 29895
    },
    {
      "epoch": 1.6740956434091163,
      "grad_norm": 1.2419829368591309,
      "learning_rate": 9.739947368421053e-05,
      "loss": 0.4359,
      "step": 29896
    },
    {
      "epoch": 1.6741516407212453,
      "grad_norm": 1.370413899421692,
      "learning_rate": 9.73992105263158e-05,
      "loss": 0.4605,
      "step": 29897
    },
    {
      "epoch": 1.6742076380333744,
      "grad_norm": 1.1843878030776978,
      "learning_rate": 9.739894736842106e-05,
      "loss": 0.4267,
      "step": 29898
    },
    {
      "epoch": 1.6742636353455034,
      "grad_norm": 1.4774214029312134,
      "learning_rate": 9.739868421052632e-05,
      "loss": 0.4291,
      "step": 29899
    },
    {
      "epoch": 1.6743196326576324,
      "grad_norm": 1.1830426454544067,
      "learning_rate": 9.739842105263158e-05,
      "loss": 0.3761,
      "step": 29900
    },
    {
      "epoch": 1.6743756299697614,
      "grad_norm": 1.2707479000091553,
      "learning_rate": 9.739815789473685e-05,
      "loss": 0.3663,
      "step": 29901
    },
    {
      "epoch": 1.6744316272818904,
      "grad_norm": 1.1479617357254028,
      "learning_rate": 9.739789473684211e-05,
      "loss": 0.376,
      "step": 29902
    },
    {
      "epoch": 1.6744876245940195,
      "grad_norm": 1.423410415649414,
      "learning_rate": 9.739763157894737e-05,
      "loss": 0.5771,
      "step": 29903
    },
    {
      "epoch": 1.6745436219061485,
      "grad_norm": 1.1610666513442993,
      "learning_rate": 9.739736842105263e-05,
      "loss": 0.4341,
      "step": 29904
    },
    {
      "epoch": 1.6745996192182775,
      "grad_norm": 1.1887913942337036,
      "learning_rate": 9.73971052631579e-05,
      "loss": 0.5333,
      "step": 29905
    },
    {
      "epoch": 1.6746556165304065,
      "grad_norm": 1.1377102136611938,
      "learning_rate": 9.739684210526317e-05,
      "loss": 0.4063,
      "step": 29906
    },
    {
      "epoch": 1.6747116138425355,
      "grad_norm": 1.2968591451644897,
      "learning_rate": 9.739657894736843e-05,
      "loss": 0.5483,
      "step": 29907
    },
    {
      "epoch": 1.6747676111546645,
      "grad_norm": 1.3600558042526245,
      "learning_rate": 9.739631578947369e-05,
      "loss": 0.4918,
      "step": 29908
    },
    {
      "epoch": 1.6748236084667936,
      "grad_norm": 1.417197585105896,
      "learning_rate": 9.739605263157895e-05,
      "loss": 0.4186,
      "step": 29909
    },
    {
      "epoch": 1.6748796057789226,
      "grad_norm": 1.2363240718841553,
      "learning_rate": 9.739578947368422e-05,
      "loss": 0.452,
      "step": 29910
    },
    {
      "epoch": 1.6749356030910516,
      "grad_norm": 1.2443267107009888,
      "learning_rate": 9.739552631578948e-05,
      "loss": 0.5973,
      "step": 29911
    },
    {
      "epoch": 1.6749916004031806,
      "grad_norm": 1.374777913093567,
      "learning_rate": 9.739526315789474e-05,
      "loss": 0.5798,
      "step": 29912
    },
    {
      "epoch": 1.6750475977153096,
      "grad_norm": 1.2771215438842773,
      "learning_rate": 9.7395e-05,
      "loss": 0.4863,
      "step": 29913
    },
    {
      "epoch": 1.6751035950274387,
      "grad_norm": 1.2088505029678345,
      "learning_rate": 9.739473684210527e-05,
      "loss": 0.5217,
      "step": 29914
    },
    {
      "epoch": 1.6751595923395677,
      "grad_norm": 1.2444523572921753,
      "learning_rate": 9.739447368421053e-05,
      "loss": 0.3946,
      "step": 29915
    },
    {
      "epoch": 1.6752155896516967,
      "grad_norm": 1.221674919128418,
      "learning_rate": 9.73942105263158e-05,
      "loss": 0.4118,
      "step": 29916
    },
    {
      "epoch": 1.6752715869638257,
      "grad_norm": 1.1710373163223267,
      "learning_rate": 9.739394736842105e-05,
      "loss": 0.4843,
      "step": 29917
    },
    {
      "epoch": 1.6753275842759547,
      "grad_norm": 1.4426542520523071,
      "learning_rate": 9.739368421052632e-05,
      "loss": 0.3877,
      "step": 29918
    },
    {
      "epoch": 1.6753835815880838,
      "grad_norm": 1.4047588109970093,
      "learning_rate": 9.739342105263158e-05,
      "loss": 0.5278,
      "step": 29919
    },
    {
      "epoch": 1.6754395789002128,
      "grad_norm": 1.557431697845459,
      "learning_rate": 9.739315789473684e-05,
      "loss": 0.3851,
      "step": 29920
    },
    {
      "epoch": 1.6754955762123418,
      "grad_norm": 1.5626405477523804,
      "learning_rate": 9.739289473684212e-05,
      "loss": 0.5033,
      "step": 29921
    },
    {
      "epoch": 1.6755515735244708,
      "grad_norm": 1.4701859951019287,
      "learning_rate": 9.739263157894736e-05,
      "loss": 0.4843,
      "step": 29922
    },
    {
      "epoch": 1.6756075708365998,
      "grad_norm": 1.4334521293640137,
      "learning_rate": 9.739236842105264e-05,
      "loss": 0.5705,
      "step": 29923
    },
    {
      "epoch": 1.6756635681487289,
      "grad_norm": 1.1340359449386597,
      "learning_rate": 9.73921052631579e-05,
      "loss": 0.3748,
      "step": 29924
    },
    {
      "epoch": 1.6757195654608579,
      "grad_norm": 1.2505853176116943,
      "learning_rate": 9.739184210526317e-05,
      "loss": 0.4332,
      "step": 29925
    },
    {
      "epoch": 1.675775562772987,
      "grad_norm": 1.6236600875854492,
      "learning_rate": 9.739157894736843e-05,
      "loss": 0.5216,
      "step": 29926
    },
    {
      "epoch": 1.675831560085116,
      "grad_norm": 1.5575350522994995,
      "learning_rate": 9.739131578947369e-05,
      "loss": 0.6268,
      "step": 29927
    },
    {
      "epoch": 1.675887557397245,
      "grad_norm": 1.2122241258621216,
      "learning_rate": 9.739105263157895e-05,
      "loss": 0.3967,
      "step": 29928
    },
    {
      "epoch": 1.675943554709374,
      "grad_norm": 1.1989843845367432,
      "learning_rate": 9.739078947368422e-05,
      "loss": 0.4996,
      "step": 29929
    },
    {
      "epoch": 1.675999552021503,
      "grad_norm": 1.2929365634918213,
      "learning_rate": 9.739052631578948e-05,
      "loss": 0.4767,
      "step": 29930
    },
    {
      "epoch": 1.676055549333632,
      "grad_norm": 1.5503559112548828,
      "learning_rate": 9.739026315789474e-05,
      "loss": 0.4401,
      "step": 29931
    },
    {
      "epoch": 1.676111546645761,
      "grad_norm": 1.1933990716934204,
      "learning_rate": 9.739e-05,
      "loss": 0.4673,
      "step": 29932
    },
    {
      "epoch": 1.67616754395789,
      "grad_norm": 1.1944936513900757,
      "learning_rate": 9.738973684210527e-05,
      "loss": 0.3465,
      "step": 29933
    },
    {
      "epoch": 1.676223541270019,
      "grad_norm": 1.291595697402954,
      "learning_rate": 9.738947368421053e-05,
      "loss": 0.4419,
      "step": 29934
    },
    {
      "epoch": 1.676279538582148,
      "grad_norm": 1.2692008018493652,
      "learning_rate": 9.73892105263158e-05,
      "loss": 0.4615,
      "step": 29935
    },
    {
      "epoch": 1.676335535894277,
      "grad_norm": 1.3068135976791382,
      "learning_rate": 9.738894736842105e-05,
      "loss": 0.5755,
      "step": 29936
    },
    {
      "epoch": 1.676391533206406,
      "grad_norm": 1.5347583293914795,
      "learning_rate": 9.738868421052631e-05,
      "loss": 0.4634,
      "step": 29937
    },
    {
      "epoch": 1.6764475305185351,
      "grad_norm": 1.3170474767684937,
      "learning_rate": 9.738842105263159e-05,
      "loss": 0.4509,
      "step": 29938
    },
    {
      "epoch": 1.6765035278306641,
      "grad_norm": 1.2580183744430542,
      "learning_rate": 9.738815789473685e-05,
      "loss": 0.3957,
      "step": 29939
    },
    {
      "epoch": 1.6765595251427932,
      "grad_norm": 1.0235823392868042,
      "learning_rate": 9.73878947368421e-05,
      "loss": 0.2971,
      "step": 29940
    },
    {
      "epoch": 1.6766155224549222,
      "grad_norm": 1.274720311164856,
      "learning_rate": 9.738763157894737e-05,
      "loss": 0.401,
      "step": 29941
    },
    {
      "epoch": 1.6766715197670512,
      "grad_norm": 1.6044840812683105,
      "learning_rate": 9.738736842105264e-05,
      "loss": 0.7239,
      "step": 29942
    },
    {
      "epoch": 1.6767275170791802,
      "grad_norm": 1.2107619047164917,
      "learning_rate": 9.73871052631579e-05,
      "loss": 0.4301,
      "step": 29943
    },
    {
      "epoch": 1.6767835143913092,
      "grad_norm": 1.5246989727020264,
      "learning_rate": 9.738684210526317e-05,
      "loss": 0.5253,
      "step": 29944
    },
    {
      "epoch": 1.6768395117034383,
      "grad_norm": 1.3754037618637085,
      "learning_rate": 9.738657894736842e-05,
      "loss": 0.4578,
      "step": 29945
    },
    {
      "epoch": 1.6768955090155673,
      "grad_norm": 1.1871083974838257,
      "learning_rate": 9.738631578947369e-05,
      "loss": 0.3561,
      "step": 29946
    },
    {
      "epoch": 1.6769515063276963,
      "grad_norm": 1.3134440183639526,
      "learning_rate": 9.738605263157895e-05,
      "loss": 0.5167,
      "step": 29947
    },
    {
      "epoch": 1.6770075036398253,
      "grad_norm": 1.5911232233047485,
      "learning_rate": 9.738578947368422e-05,
      "loss": 0.4224,
      "step": 29948
    },
    {
      "epoch": 1.6770635009519543,
      "grad_norm": 1.1805120706558228,
      "learning_rate": 9.738552631578947e-05,
      "loss": 0.4332,
      "step": 29949
    },
    {
      "epoch": 1.6771194982640834,
      "grad_norm": 1.2180368900299072,
      "learning_rate": 9.738526315789474e-05,
      "loss": 0.4843,
      "step": 29950
    },
    {
      "epoch": 1.6771754955762124,
      "grad_norm": 0.9737374782562256,
      "learning_rate": 9.7385e-05,
      "loss": 0.3797,
      "step": 29951
    },
    {
      "epoch": 1.6772314928883414,
      "grad_norm": 1.3471542596817017,
      "learning_rate": 9.738473684210528e-05,
      "loss": 0.5777,
      "step": 29952
    },
    {
      "epoch": 1.6772874902004704,
      "grad_norm": 1.407422661781311,
      "learning_rate": 9.738447368421054e-05,
      "loss": 0.4742,
      "step": 29953
    },
    {
      "epoch": 1.6773434875125994,
      "grad_norm": 1.4307019710540771,
      "learning_rate": 9.738421052631578e-05,
      "loss": 0.4776,
      "step": 29954
    },
    {
      "epoch": 1.6773994848247284,
      "grad_norm": 1.4423656463623047,
      "learning_rate": 9.738394736842106e-05,
      "loss": 0.6163,
      "step": 29955
    },
    {
      "epoch": 1.6774554821368575,
      "grad_norm": 1.8022441864013672,
      "learning_rate": 9.738368421052632e-05,
      "loss": 0.4509,
      "step": 29956
    },
    {
      "epoch": 1.6775114794489865,
      "grad_norm": 1.2700351476669312,
      "learning_rate": 9.738342105263159e-05,
      "loss": 0.4093,
      "step": 29957
    },
    {
      "epoch": 1.6775674767611155,
      "grad_norm": 1.5666680335998535,
      "learning_rate": 9.738315789473685e-05,
      "loss": 0.5372,
      "step": 29958
    },
    {
      "epoch": 1.6776234740732445,
      "grad_norm": 1.3759901523590088,
      "learning_rate": 9.738289473684211e-05,
      "loss": 0.4121,
      "step": 29959
    },
    {
      "epoch": 1.6776794713853735,
      "grad_norm": 1.5998016595840454,
      "learning_rate": 9.738263157894737e-05,
      "loss": 0.5589,
      "step": 29960
    },
    {
      "epoch": 1.6777354686975026,
      "grad_norm": 1.2518067359924316,
      "learning_rate": 9.738236842105264e-05,
      "loss": 0.3976,
      "step": 29961
    },
    {
      "epoch": 1.6777914660096316,
      "grad_norm": 1.2326905727386475,
      "learning_rate": 9.73821052631579e-05,
      "loss": 0.4092,
      "step": 29962
    },
    {
      "epoch": 1.6778474633217606,
      "grad_norm": 1.0324153900146484,
      "learning_rate": 9.738184210526316e-05,
      "loss": 0.433,
      "step": 29963
    },
    {
      "epoch": 1.6779034606338896,
      "grad_norm": 1.4028562307357788,
      "learning_rate": 9.738157894736842e-05,
      "loss": 0.5518,
      "step": 29964
    },
    {
      "epoch": 1.6779594579460186,
      "grad_norm": 1.1557691097259521,
      "learning_rate": 9.73813157894737e-05,
      "loss": 0.4234,
      "step": 29965
    },
    {
      "epoch": 1.6780154552581477,
      "grad_norm": 1.2471976280212402,
      "learning_rate": 9.738105263157895e-05,
      "loss": 0.4519,
      "step": 29966
    },
    {
      "epoch": 1.6780714525702767,
      "grad_norm": 1.1806623935699463,
      "learning_rate": 9.738078947368421e-05,
      "loss": 0.4599,
      "step": 29967
    },
    {
      "epoch": 1.6781274498824057,
      "grad_norm": 1.3362605571746826,
      "learning_rate": 9.738052631578947e-05,
      "loss": 0.6427,
      "step": 29968
    },
    {
      "epoch": 1.6781834471945347,
      "grad_norm": 1.3070813417434692,
      "learning_rate": 9.738026315789475e-05,
      "loss": 0.5511,
      "step": 29969
    },
    {
      "epoch": 1.6782394445066637,
      "grad_norm": 1.2259621620178223,
      "learning_rate": 9.738e-05,
      "loss": 0.3998,
      "step": 29970
    },
    {
      "epoch": 1.6782954418187925,
      "grad_norm": 1.2757374048233032,
      "learning_rate": 9.737973684210527e-05,
      "loss": 0.4778,
      "step": 29971
    },
    {
      "epoch": 1.6783514391309216,
      "grad_norm": 1.2138557434082031,
      "learning_rate": 9.737947368421053e-05,
      "loss": 0.3277,
      "step": 29972
    },
    {
      "epoch": 1.6784074364430506,
      "grad_norm": 1.3146134614944458,
      "learning_rate": 9.737921052631579e-05,
      "loss": 0.3642,
      "step": 29973
    },
    {
      "epoch": 1.6784634337551796,
      "grad_norm": 1.1987791061401367,
      "learning_rate": 9.737894736842106e-05,
      "loss": 0.4592,
      "step": 29974
    },
    {
      "epoch": 1.6785194310673086,
      "grad_norm": 1.5726855993270874,
      "learning_rate": 9.737868421052632e-05,
      "loss": 0.6088,
      "step": 29975
    },
    {
      "epoch": 1.6785754283794376,
      "grad_norm": 1.094915509223938,
      "learning_rate": 9.737842105263159e-05,
      "loss": 0.4098,
      "step": 29976
    },
    {
      "epoch": 1.6786314256915666,
      "grad_norm": 1.1610496044158936,
      "learning_rate": 9.737815789473684e-05,
      "loss": 0.4652,
      "step": 29977
    },
    {
      "epoch": 1.6786874230036957,
      "grad_norm": 1.342151165008545,
      "learning_rate": 9.737789473684211e-05,
      "loss": 0.392,
      "step": 29978
    },
    {
      "epoch": 1.6787434203158247,
      "grad_norm": 1.1701411008834839,
      "learning_rate": 9.737763157894737e-05,
      "loss": 0.4119,
      "step": 29979
    },
    {
      "epoch": 1.6787994176279537,
      "grad_norm": 1.7732369899749756,
      "learning_rate": 9.737736842105264e-05,
      "loss": 0.4794,
      "step": 29980
    },
    {
      "epoch": 1.6788554149400827,
      "grad_norm": 1.4591057300567627,
      "learning_rate": 9.73771052631579e-05,
      "loss": 0.4294,
      "step": 29981
    },
    {
      "epoch": 1.6789114122522117,
      "grad_norm": 1.2426187992095947,
      "learning_rate": 9.737684210526316e-05,
      "loss": 0.5808,
      "step": 29982
    },
    {
      "epoch": 1.6789674095643408,
      "grad_norm": 1.490771770477295,
      "learning_rate": 9.737657894736842e-05,
      "loss": 0.5741,
      "step": 29983
    },
    {
      "epoch": 1.6790234068764698,
      "grad_norm": 1.3071575164794922,
      "learning_rate": 9.73763157894737e-05,
      "loss": 0.4386,
      "step": 29984
    },
    {
      "epoch": 1.6790794041885988,
      "grad_norm": 1.275038480758667,
      "learning_rate": 9.737605263157896e-05,
      "loss": 0.3716,
      "step": 29985
    },
    {
      "epoch": 1.6791354015007278,
      "grad_norm": 1.1046198606491089,
      "learning_rate": 9.737578947368422e-05,
      "loss": 0.4326,
      "step": 29986
    },
    {
      "epoch": 1.6791913988128568,
      "grad_norm": 1.302770972251892,
      "learning_rate": 9.737552631578948e-05,
      "loss": 0.4648,
      "step": 29987
    },
    {
      "epoch": 1.6792473961249859,
      "grad_norm": 1.6733739376068115,
      "learning_rate": 9.737526315789474e-05,
      "loss": 0.4772,
      "step": 29988
    },
    {
      "epoch": 1.6793033934371149,
      "grad_norm": 1.2860265970230103,
      "learning_rate": 9.737500000000001e-05,
      "loss": 0.312,
      "step": 29989
    },
    {
      "epoch": 1.679359390749244,
      "grad_norm": 1.3980674743652344,
      "learning_rate": 9.737473684210527e-05,
      "loss": 0.517,
      "step": 29990
    },
    {
      "epoch": 1.679415388061373,
      "grad_norm": 1.5490573644638062,
      "learning_rate": 9.737447368421053e-05,
      "loss": 0.6049,
      "step": 29991
    },
    {
      "epoch": 1.679471385373502,
      "grad_norm": 1.5674378871917725,
      "learning_rate": 9.737421052631579e-05,
      "loss": 0.5021,
      "step": 29992
    },
    {
      "epoch": 1.679527382685631,
      "grad_norm": 1.4125771522521973,
      "learning_rate": 9.737394736842106e-05,
      "loss": 0.4054,
      "step": 29993
    },
    {
      "epoch": 1.67958337999776,
      "grad_norm": 4.377709865570068,
      "learning_rate": 9.737368421052632e-05,
      "loss": 0.2836,
      "step": 29994
    },
    {
      "epoch": 1.679639377309889,
      "grad_norm": 1.6075434684753418,
      "learning_rate": 9.737342105263158e-05,
      "loss": 0.3507,
      "step": 29995
    },
    {
      "epoch": 1.679695374622018,
      "grad_norm": 1.4482122659683228,
      "learning_rate": 9.737315789473684e-05,
      "loss": 0.4288,
      "step": 29996
    },
    {
      "epoch": 1.679751371934147,
      "grad_norm": 1.3585349321365356,
      "learning_rate": 9.737289473684211e-05,
      "loss": 0.5288,
      "step": 29997
    },
    {
      "epoch": 1.679807369246276,
      "grad_norm": 1.1839277744293213,
      "learning_rate": 9.737263157894737e-05,
      "loss": 0.4808,
      "step": 29998
    },
    {
      "epoch": 1.679863366558405,
      "grad_norm": 1.5054590702056885,
      "learning_rate": 9.737236842105263e-05,
      "loss": 0.4966,
      "step": 29999
    },
    {
      "epoch": 1.679919363870534,
      "grad_norm": 1.5555325746536255,
      "learning_rate": 9.737210526315789e-05,
      "loss": 0.4613,
      "step": 30000
    },
    {
      "epoch": 1.679975361182663,
      "grad_norm": 1.2994585037231445,
      "learning_rate": 9.737184210526317e-05,
      "loss": 0.4076,
      "step": 30001
    },
    {
      "epoch": 1.6800313584947921,
      "grad_norm": 1.3233389854431152,
      "learning_rate": 9.737157894736843e-05,
      "loss": 0.4217,
      "step": 30002
    },
    {
      "epoch": 1.6800873558069211,
      "grad_norm": 1.224224328994751,
      "learning_rate": 9.73713157894737e-05,
      "loss": 0.3775,
      "step": 30003
    },
    {
      "epoch": 1.6801433531190502,
      "grad_norm": 1.1796660423278809,
      "learning_rate": 9.737105263157895e-05,
      "loss": 0.4234,
      "step": 30004
    },
    {
      "epoch": 1.6801993504311792,
      "grad_norm": 1.211627721786499,
      "learning_rate": 9.73707894736842e-05,
      "loss": 0.4587,
      "step": 30005
    },
    {
      "epoch": 1.6802553477433082,
      "grad_norm": 1.2654528617858887,
      "learning_rate": 9.737052631578948e-05,
      "loss": 0.4753,
      "step": 30006
    },
    {
      "epoch": 1.6803113450554372,
      "grad_norm": 1.498504400253296,
      "learning_rate": 9.737026315789474e-05,
      "loss": 0.3996,
      "step": 30007
    },
    {
      "epoch": 1.6803673423675662,
      "grad_norm": 1.0283262729644775,
      "learning_rate": 9.737000000000001e-05,
      "loss": 0.3684,
      "step": 30008
    },
    {
      "epoch": 1.6804233396796953,
      "grad_norm": 1.4832878112792969,
      "learning_rate": 9.736973684210526e-05,
      "loss": 0.5157,
      "step": 30009
    },
    {
      "epoch": 1.6804793369918243,
      "grad_norm": 1.3012858629226685,
      "learning_rate": 9.736947368421053e-05,
      "loss": 0.5147,
      "step": 30010
    },
    {
      "epoch": 1.6805353343039533,
      "grad_norm": 1.2735188007354736,
      "learning_rate": 9.736921052631579e-05,
      "loss": 0.5182,
      "step": 30011
    },
    {
      "epoch": 1.6805913316160823,
      "grad_norm": 1.273616909980774,
      "learning_rate": 9.736894736842106e-05,
      "loss": 0.4831,
      "step": 30012
    },
    {
      "epoch": 1.6806473289282113,
      "grad_norm": 1.2145116329193115,
      "learning_rate": 9.736868421052632e-05,
      "loss": 0.4781,
      "step": 30013
    },
    {
      "epoch": 1.6807033262403404,
      "grad_norm": 1.4690943956375122,
      "learning_rate": 9.736842105263158e-05,
      "loss": 0.677,
      "step": 30014
    },
    {
      "epoch": 1.6807593235524694,
      "grad_norm": 1.4068059921264648,
      "learning_rate": 9.736815789473684e-05,
      "loss": 0.4209,
      "step": 30015
    },
    {
      "epoch": 1.6808153208645984,
      "grad_norm": 1.0696946382522583,
      "learning_rate": 9.736789473684212e-05,
      "loss": 0.4339,
      "step": 30016
    },
    {
      "epoch": 1.6808713181767274,
      "grad_norm": 1.1998757123947144,
      "learning_rate": 9.736763157894738e-05,
      "loss": 0.4082,
      "step": 30017
    },
    {
      "epoch": 1.6809273154888564,
      "grad_norm": 1.4615795612335205,
      "learning_rate": 9.736736842105264e-05,
      "loss": 0.4241,
      "step": 30018
    },
    {
      "epoch": 1.6809833128009855,
      "grad_norm": 1.3013535737991333,
      "learning_rate": 9.73671052631579e-05,
      "loss": 0.4069,
      "step": 30019
    },
    {
      "epoch": 1.6810393101131145,
      "grad_norm": 1.12428617477417,
      "learning_rate": 9.736684210526317e-05,
      "loss": 0.4252,
      "step": 30020
    },
    {
      "epoch": 1.6810953074252435,
      "grad_norm": 1.2561031579971313,
      "learning_rate": 9.736657894736843e-05,
      "loss": 0.4285,
      "step": 30021
    },
    {
      "epoch": 1.6811513047373725,
      "grad_norm": 1.296723484992981,
      "learning_rate": 9.736631578947369e-05,
      "loss": 0.5986,
      "step": 30022
    },
    {
      "epoch": 1.6812073020495015,
      "grad_norm": 1.1818569898605347,
      "learning_rate": 9.736605263157895e-05,
      "loss": 0.4809,
      "step": 30023
    },
    {
      "epoch": 1.6812632993616305,
      "grad_norm": 1.256476879119873,
      "learning_rate": 9.736578947368421e-05,
      "loss": 0.4643,
      "step": 30024
    },
    {
      "epoch": 1.6813192966737596,
      "grad_norm": 1.585464596748352,
      "learning_rate": 9.736552631578948e-05,
      "loss": 0.4811,
      "step": 30025
    },
    {
      "epoch": 1.6813752939858886,
      "grad_norm": 1.2282109260559082,
      "learning_rate": 9.736526315789474e-05,
      "loss": 0.4674,
      "step": 30026
    },
    {
      "epoch": 1.6814312912980176,
      "grad_norm": 2.262535810470581,
      "learning_rate": 9.7365e-05,
      "loss": 0.5656,
      "step": 30027
    },
    {
      "epoch": 1.6814872886101466,
      "grad_norm": 1.455905795097351,
      "learning_rate": 9.736473684210526e-05,
      "loss": 0.6624,
      "step": 30028
    },
    {
      "epoch": 1.6815432859222756,
      "grad_norm": 1.2129590511322021,
      "learning_rate": 9.736447368421053e-05,
      "loss": 0.3836,
      "step": 30029
    },
    {
      "epoch": 1.6815992832344047,
      "grad_norm": 1.4859960079193115,
      "learning_rate": 9.736421052631579e-05,
      "loss": 0.401,
      "step": 30030
    },
    {
      "epoch": 1.6816552805465337,
      "grad_norm": 1.437681794166565,
      "learning_rate": 9.736394736842107e-05,
      "loss": 0.4562,
      "step": 30031
    },
    {
      "epoch": 1.6817112778586627,
      "grad_norm": 1.2419911623001099,
      "learning_rate": 9.736368421052631e-05,
      "loss": 0.5159,
      "step": 30032
    },
    {
      "epoch": 1.6817672751707917,
      "grad_norm": 1.2178417444229126,
      "learning_rate": 9.736342105263159e-05,
      "loss": 0.4272,
      "step": 30033
    },
    {
      "epoch": 1.6818232724829207,
      "grad_norm": 1.9885791540145874,
      "learning_rate": 9.736315789473685e-05,
      "loss": 0.7109,
      "step": 30034
    },
    {
      "epoch": 1.6818792697950498,
      "grad_norm": 1.2620553970336914,
      "learning_rate": 9.736289473684212e-05,
      "loss": 0.4715,
      "step": 30035
    },
    {
      "epoch": 1.6819352671071788,
      "grad_norm": 1.548359990119934,
      "learning_rate": 9.736263157894738e-05,
      "loss": 0.4702,
      "step": 30036
    },
    {
      "epoch": 1.6819912644193078,
      "grad_norm": 2.5032172203063965,
      "learning_rate": 9.736236842105264e-05,
      "loss": 0.5865,
      "step": 30037
    },
    {
      "epoch": 1.6820472617314368,
      "grad_norm": 1.3945519924163818,
      "learning_rate": 9.73621052631579e-05,
      "loss": 0.4448,
      "step": 30038
    },
    {
      "epoch": 1.6821032590435658,
      "grad_norm": 1.1375930309295654,
      "learning_rate": 9.736184210526317e-05,
      "loss": 0.3363,
      "step": 30039
    },
    {
      "epoch": 1.6821592563556949,
      "grad_norm": 1.2704322338104248,
      "learning_rate": 9.736157894736843e-05,
      "loss": 0.4223,
      "step": 30040
    },
    {
      "epoch": 1.6822152536678239,
      "grad_norm": 1.4568498134613037,
      "learning_rate": 9.736131578947368e-05,
      "loss": 0.5237,
      "step": 30041
    },
    {
      "epoch": 1.682271250979953,
      "grad_norm": 1.1758419275283813,
      "learning_rate": 9.736105263157895e-05,
      "loss": 0.4233,
      "step": 30042
    },
    {
      "epoch": 1.682327248292082,
      "grad_norm": 1.3888391256332397,
      "learning_rate": 9.736078947368421e-05,
      "loss": 0.4914,
      "step": 30043
    },
    {
      "epoch": 1.682383245604211,
      "grad_norm": 1.6250340938568115,
      "learning_rate": 9.736052631578948e-05,
      "loss": 0.5952,
      "step": 30044
    },
    {
      "epoch": 1.68243924291634,
      "grad_norm": 2.2349987030029297,
      "learning_rate": 9.736026315789474e-05,
      "loss": 0.5461,
      "step": 30045
    },
    {
      "epoch": 1.682495240228469,
      "grad_norm": 1.2714875936508179,
      "learning_rate": 9.736e-05,
      "loss": 0.301,
      "step": 30046
    },
    {
      "epoch": 1.682551237540598,
      "grad_norm": 1.311805009841919,
      "learning_rate": 9.735973684210526e-05,
      "loss": 0.4269,
      "step": 30047
    },
    {
      "epoch": 1.682607234852727,
      "grad_norm": 1.1518783569335938,
      "learning_rate": 9.735947368421054e-05,
      "loss": 0.6915,
      "step": 30048
    },
    {
      "epoch": 1.682663232164856,
      "grad_norm": 1.5244580507278442,
      "learning_rate": 9.73592105263158e-05,
      "loss": 0.599,
      "step": 30049
    },
    {
      "epoch": 1.682719229476985,
      "grad_norm": 1.5039039850234985,
      "learning_rate": 9.735894736842106e-05,
      "loss": 0.4116,
      "step": 30050
    },
    {
      "epoch": 1.682775226789114,
      "grad_norm": 1.5512877702713013,
      "learning_rate": 9.735868421052632e-05,
      "loss": 0.4742,
      "step": 30051
    },
    {
      "epoch": 1.682831224101243,
      "grad_norm": 1.1976828575134277,
      "learning_rate": 9.735842105263159e-05,
      "loss": 0.4459,
      "step": 30052
    },
    {
      "epoch": 1.682887221413372,
      "grad_norm": 1.203157901763916,
      "learning_rate": 9.735815789473685e-05,
      "loss": 0.393,
      "step": 30053
    },
    {
      "epoch": 1.6829432187255011,
      "grad_norm": 1.2579879760742188,
      "learning_rate": 9.735789473684211e-05,
      "loss": 0.4515,
      "step": 30054
    },
    {
      "epoch": 1.6829992160376301,
      "grad_norm": 1.4593232870101929,
      "learning_rate": 9.735763157894737e-05,
      "loss": 0.5174,
      "step": 30055
    },
    {
      "epoch": 1.6830552133497592,
      "grad_norm": 1.5238014459609985,
      "learning_rate": 9.735736842105264e-05,
      "loss": 0.5091,
      "step": 30056
    },
    {
      "epoch": 1.6831112106618882,
      "grad_norm": 1.6702297925949097,
      "learning_rate": 9.73571052631579e-05,
      "loss": 0.5301,
      "step": 30057
    },
    {
      "epoch": 1.6831672079740172,
      "grad_norm": 1.409104824066162,
      "learning_rate": 9.735684210526316e-05,
      "loss": 0.5162,
      "step": 30058
    },
    {
      "epoch": 1.6832232052861462,
      "grad_norm": 1.4003217220306396,
      "learning_rate": 9.735657894736842e-05,
      "loss": 0.4034,
      "step": 30059
    },
    {
      "epoch": 1.6832792025982752,
      "grad_norm": 1.0628284215927124,
      "learning_rate": 9.735631578947368e-05,
      "loss": 0.3403,
      "step": 30060
    },
    {
      "epoch": 1.6833351999104043,
      "grad_norm": 1.4191914796829224,
      "learning_rate": 9.735605263157895e-05,
      "loss": 0.4711,
      "step": 30061
    },
    {
      "epoch": 1.6833911972225333,
      "grad_norm": 1.4225420951843262,
      "learning_rate": 9.735578947368421e-05,
      "loss": 0.4781,
      "step": 30062
    },
    {
      "epoch": 1.6834471945346623,
      "grad_norm": 1.2523552179336548,
      "learning_rate": 9.735552631578949e-05,
      "loss": 0.3511,
      "step": 30063
    },
    {
      "epoch": 1.6835031918467913,
      "grad_norm": 1.1969324350357056,
      "learning_rate": 9.735526315789473e-05,
      "loss": 0.4782,
      "step": 30064
    },
    {
      "epoch": 1.6835591891589203,
      "grad_norm": 1.314456820487976,
      "learning_rate": 9.7355e-05,
      "loss": 0.3676,
      "step": 30065
    },
    {
      "epoch": 1.6836151864710494,
      "grad_norm": 3.1269609928131104,
      "learning_rate": 9.735473684210527e-05,
      "loss": 0.4836,
      "step": 30066
    },
    {
      "epoch": 1.6836711837831784,
      "grad_norm": 1.3982963562011719,
      "learning_rate": 9.735447368421054e-05,
      "loss": 0.5443,
      "step": 30067
    },
    {
      "epoch": 1.6837271810953074,
      "grad_norm": 1.2770054340362549,
      "learning_rate": 9.73542105263158e-05,
      "loss": 0.3972,
      "step": 30068
    },
    {
      "epoch": 1.6837831784074364,
      "grad_norm": 1.6281217336654663,
      "learning_rate": 9.735394736842106e-05,
      "loss": 0.4614,
      "step": 30069
    },
    {
      "epoch": 1.6838391757195654,
      "grad_norm": 1.2229751348495483,
      "learning_rate": 9.735368421052632e-05,
      "loss": 0.617,
      "step": 30070
    },
    {
      "epoch": 1.6838951730316944,
      "grad_norm": 1.2579398155212402,
      "learning_rate": 9.735342105263159e-05,
      "loss": 0.5079,
      "step": 30071
    },
    {
      "epoch": 1.6839511703438235,
      "grad_norm": 1.5189852714538574,
      "learning_rate": 9.735315789473685e-05,
      "loss": 0.4688,
      "step": 30072
    },
    {
      "epoch": 1.6840071676559525,
      "grad_norm": 1.2876662015914917,
      "learning_rate": 9.735289473684211e-05,
      "loss": 0.4895,
      "step": 30073
    },
    {
      "epoch": 1.6840631649680815,
      "grad_norm": 1.3178309202194214,
      "learning_rate": 9.735263157894737e-05,
      "loss": 0.5764,
      "step": 30074
    },
    {
      "epoch": 1.6841191622802105,
      "grad_norm": 1.5533462762832642,
      "learning_rate": 9.735236842105263e-05,
      "loss": 0.5602,
      "step": 30075
    },
    {
      "epoch": 1.6841751595923395,
      "grad_norm": 1.2621676921844482,
      "learning_rate": 9.73521052631579e-05,
      "loss": 0.4298,
      "step": 30076
    },
    {
      "epoch": 1.6842311569044686,
      "grad_norm": 1.3253891468048096,
      "learning_rate": 9.735184210526316e-05,
      "loss": 0.6581,
      "step": 30077
    },
    {
      "epoch": 1.6842871542165976,
      "grad_norm": 1.3398269414901733,
      "learning_rate": 9.735157894736842e-05,
      "loss": 0.5109,
      "step": 30078
    },
    {
      "epoch": 1.6843431515287266,
      "grad_norm": 1.0978977680206299,
      "learning_rate": 9.735131578947368e-05,
      "loss": 0.3186,
      "step": 30079
    },
    {
      "epoch": 1.6843991488408556,
      "grad_norm": 1.2364259958267212,
      "learning_rate": 9.735105263157896e-05,
      "loss": 0.4888,
      "step": 30080
    },
    {
      "epoch": 1.6844551461529846,
      "grad_norm": 1.2050082683563232,
      "learning_rate": 9.735078947368422e-05,
      "loss": 0.3848,
      "step": 30081
    },
    {
      "epoch": 1.6845111434651137,
      "grad_norm": 1.363664150238037,
      "learning_rate": 9.735052631578948e-05,
      "loss": 0.3568,
      "step": 30082
    },
    {
      "epoch": 1.6845671407772427,
      "grad_norm": 1.1327582597732544,
      "learning_rate": 9.735026315789473e-05,
      "loss": 0.3377,
      "step": 30083
    },
    {
      "epoch": 1.6846231380893717,
      "grad_norm": 1.2636672258377075,
      "learning_rate": 9.735000000000001e-05,
      "loss": 0.3855,
      "step": 30084
    },
    {
      "epoch": 1.6846791354015007,
      "grad_norm": 1.3640165328979492,
      "learning_rate": 9.734973684210527e-05,
      "loss": 0.3801,
      "step": 30085
    },
    {
      "epoch": 1.6847351327136297,
      "grad_norm": 1.1210427284240723,
      "learning_rate": 9.734947368421054e-05,
      "loss": 0.4005,
      "step": 30086
    },
    {
      "epoch": 1.6847911300257588,
      "grad_norm": 1.3407551050186157,
      "learning_rate": 9.734921052631579e-05,
      "loss": 0.5224,
      "step": 30087
    },
    {
      "epoch": 1.6848471273378878,
      "grad_norm": 1.1256792545318604,
      "learning_rate": 9.734894736842106e-05,
      "loss": 0.3795,
      "step": 30088
    },
    {
      "epoch": 1.6849031246500168,
      "grad_norm": 1.897878885269165,
      "learning_rate": 9.734868421052632e-05,
      "loss": 0.7088,
      "step": 30089
    },
    {
      "epoch": 1.6849591219621458,
      "grad_norm": 1.8243905305862427,
      "learning_rate": 9.73484210526316e-05,
      "loss": 0.6018,
      "step": 30090
    },
    {
      "epoch": 1.6850151192742748,
      "grad_norm": 1.3680564165115356,
      "learning_rate": 9.734815789473685e-05,
      "loss": 0.4472,
      "step": 30091
    },
    {
      "epoch": 1.6850711165864038,
      "grad_norm": 1.147234320640564,
      "learning_rate": 9.73478947368421e-05,
      "loss": 0.3787,
      "step": 30092
    },
    {
      "epoch": 1.6851271138985329,
      "grad_norm": 1.3384244441986084,
      "learning_rate": 9.734763157894737e-05,
      "loss": 0.5763,
      "step": 30093
    },
    {
      "epoch": 1.6851831112106619,
      "grad_norm": 1.4484529495239258,
      "learning_rate": 9.734736842105263e-05,
      "loss": 0.5563,
      "step": 30094
    },
    {
      "epoch": 1.685239108522791,
      "grad_norm": 1.1401987075805664,
      "learning_rate": 9.73471052631579e-05,
      "loss": 0.3952,
      "step": 30095
    },
    {
      "epoch": 1.68529510583492,
      "grad_norm": 1.1800525188446045,
      "learning_rate": 9.734684210526315e-05,
      "loss": 0.4631,
      "step": 30096
    },
    {
      "epoch": 1.685351103147049,
      "grad_norm": 1.7282980680465698,
      "learning_rate": 9.734657894736843e-05,
      "loss": 0.6157,
      "step": 30097
    },
    {
      "epoch": 1.685407100459178,
      "grad_norm": 1.2098634243011475,
      "learning_rate": 9.734631578947368e-05,
      "loss": 0.49,
      "step": 30098
    },
    {
      "epoch": 1.685463097771307,
      "grad_norm": 1.2357712984085083,
      "learning_rate": 9.734605263157896e-05,
      "loss": 0.4857,
      "step": 30099
    },
    {
      "epoch": 1.685519095083436,
      "grad_norm": 1.075027585029602,
      "learning_rate": 9.734578947368422e-05,
      "loss": 0.4665,
      "step": 30100
    },
    {
      "epoch": 1.685575092395565,
      "grad_norm": 1.2004560232162476,
      "learning_rate": 9.734552631578948e-05,
      "loss": 0.5046,
      "step": 30101
    },
    {
      "epoch": 1.685631089707694,
      "grad_norm": 1.4363986253738403,
      "learning_rate": 9.734526315789474e-05,
      "loss": 0.4104,
      "step": 30102
    },
    {
      "epoch": 1.685687087019823,
      "grad_norm": 1.3236922025680542,
      "learning_rate": 9.734500000000001e-05,
      "loss": 0.4833,
      "step": 30103
    },
    {
      "epoch": 1.685743084331952,
      "grad_norm": 1.2907474040985107,
      "learning_rate": 9.734473684210527e-05,
      "loss": 0.5518,
      "step": 30104
    },
    {
      "epoch": 1.685799081644081,
      "grad_norm": 1.906058430671692,
      "learning_rate": 9.734447368421053e-05,
      "loss": 0.6542,
      "step": 30105
    },
    {
      "epoch": 1.6858550789562101,
      "grad_norm": 1.234045386314392,
      "learning_rate": 9.734421052631579e-05,
      "loss": 0.3836,
      "step": 30106
    },
    {
      "epoch": 1.6859110762683391,
      "grad_norm": 1.3779826164245605,
      "learning_rate": 9.734394736842106e-05,
      "loss": 0.5039,
      "step": 30107
    },
    {
      "epoch": 1.6859670735804682,
      "grad_norm": 1.2399606704711914,
      "learning_rate": 9.734368421052632e-05,
      "loss": 0.4581,
      "step": 30108
    },
    {
      "epoch": 1.6860230708925972,
      "grad_norm": 1.3214467763900757,
      "learning_rate": 9.734342105263158e-05,
      "loss": 0.4162,
      "step": 30109
    },
    {
      "epoch": 1.6860790682047262,
      "grad_norm": 1.1875327825546265,
      "learning_rate": 9.734315789473684e-05,
      "loss": 0.4735,
      "step": 30110
    },
    {
      "epoch": 1.6861350655168552,
      "grad_norm": 1.963280200958252,
      "learning_rate": 9.73428947368421e-05,
      "loss": 0.6456,
      "step": 30111
    },
    {
      "epoch": 1.6861910628289842,
      "grad_norm": 1.2360131740570068,
      "learning_rate": 9.734263157894738e-05,
      "loss": 0.5305,
      "step": 30112
    },
    {
      "epoch": 1.6862470601411133,
      "grad_norm": 1.077210545539856,
      "learning_rate": 9.734236842105264e-05,
      "loss": 0.3636,
      "step": 30113
    },
    {
      "epoch": 1.6863030574532423,
      "grad_norm": 1.512688398361206,
      "learning_rate": 9.73421052631579e-05,
      "loss": 0.5859,
      "step": 30114
    },
    {
      "epoch": 1.6863590547653713,
      "grad_norm": 2.2748279571533203,
      "learning_rate": 9.734184210526315e-05,
      "loss": 0.5212,
      "step": 30115
    },
    {
      "epoch": 1.6864150520775003,
      "grad_norm": 1.3436267375946045,
      "learning_rate": 9.734157894736843e-05,
      "loss": 0.411,
      "step": 30116
    },
    {
      "epoch": 1.6864710493896293,
      "grad_norm": 1.226438045501709,
      "learning_rate": 9.734131578947369e-05,
      "loss": 0.385,
      "step": 30117
    },
    {
      "epoch": 1.6865270467017583,
      "grad_norm": 1.1846208572387695,
      "learning_rate": 9.734105263157896e-05,
      "loss": 0.3664,
      "step": 30118
    },
    {
      "epoch": 1.6865830440138874,
      "grad_norm": 1.4303958415985107,
      "learning_rate": 9.734078947368421e-05,
      "loss": 0.6189,
      "step": 30119
    },
    {
      "epoch": 1.6866390413260164,
      "grad_norm": 1.2793289422988892,
      "learning_rate": 9.734052631578948e-05,
      "loss": 0.4675,
      "step": 30120
    },
    {
      "epoch": 1.6866950386381454,
      "grad_norm": 1.8236483335494995,
      "learning_rate": 9.734026315789474e-05,
      "loss": 0.5422,
      "step": 30121
    },
    {
      "epoch": 1.6867510359502744,
      "grad_norm": 1.384316086769104,
      "learning_rate": 9.734000000000001e-05,
      "loss": 0.4779,
      "step": 30122
    },
    {
      "epoch": 1.6868070332624034,
      "grad_norm": 1.2739804983139038,
      "learning_rate": 9.733973684210527e-05,
      "loss": 0.5885,
      "step": 30123
    },
    {
      "epoch": 1.6868630305745325,
      "grad_norm": 1.3124525547027588,
      "learning_rate": 9.733947368421053e-05,
      "loss": 0.3851,
      "step": 30124
    },
    {
      "epoch": 1.6869190278866615,
      "grad_norm": 5.500936031341553,
      "learning_rate": 9.733921052631579e-05,
      "loss": 0.5684,
      "step": 30125
    },
    {
      "epoch": 1.6869750251987905,
      "grad_norm": 1.1165839433670044,
      "learning_rate": 9.733894736842105e-05,
      "loss": 0.4695,
      "step": 30126
    },
    {
      "epoch": 1.6870310225109195,
      "grad_norm": 17.776029586791992,
      "learning_rate": 9.733868421052633e-05,
      "loss": 0.4529,
      "step": 30127
    },
    {
      "epoch": 1.6870870198230485,
      "grad_norm": 14.615177154541016,
      "learning_rate": 9.733842105263159e-05,
      "loss": 0.6338,
      "step": 30128
    },
    {
      "epoch": 1.6871430171351776,
      "grad_norm": 1.2860603332519531,
      "learning_rate": 9.733815789473684e-05,
      "loss": 0.5732,
      "step": 30129
    },
    {
      "epoch": 1.6871990144473066,
      "grad_norm": 1.4584449529647827,
      "learning_rate": 9.73378947368421e-05,
      "loss": 0.5255,
      "step": 30130
    },
    {
      "epoch": 1.6872550117594356,
      "grad_norm": 1.2533122301101685,
      "learning_rate": 9.733763157894738e-05,
      "loss": 0.4487,
      "step": 30131
    },
    {
      "epoch": 1.6873110090715646,
      "grad_norm": 1.1864601373672485,
      "learning_rate": 9.733736842105264e-05,
      "loss": 0.3937,
      "step": 30132
    },
    {
      "epoch": 1.6873670063836936,
      "grad_norm": 1.6280063390731812,
      "learning_rate": 9.73371052631579e-05,
      "loss": 0.4634,
      "step": 30133
    },
    {
      "epoch": 1.6874230036958227,
      "grad_norm": 1.5736427307128906,
      "learning_rate": 9.733684210526316e-05,
      "loss": 0.4416,
      "step": 30134
    },
    {
      "epoch": 1.6874790010079517,
      "grad_norm": 3.5148539543151855,
      "learning_rate": 9.733657894736843e-05,
      "loss": 0.578,
      "step": 30135
    },
    {
      "epoch": 1.6875349983200807,
      "grad_norm": 1.5513710975646973,
      "learning_rate": 9.733631578947369e-05,
      "loss": 0.4843,
      "step": 30136
    },
    {
      "epoch": 1.6875909956322097,
      "grad_norm": 1.3073911666870117,
      "learning_rate": 9.733605263157895e-05,
      "loss": 0.3689,
      "step": 30137
    },
    {
      "epoch": 1.6876469929443387,
      "grad_norm": 1.873498558998108,
      "learning_rate": 9.733578947368421e-05,
      "loss": 0.5152,
      "step": 30138
    },
    {
      "epoch": 1.6877029902564677,
      "grad_norm": 1.4693621397018433,
      "learning_rate": 9.733552631578948e-05,
      "loss": 0.5006,
      "step": 30139
    },
    {
      "epoch": 1.6877589875685968,
      "grad_norm": 1.352457880973816,
      "learning_rate": 9.733526315789474e-05,
      "loss": 0.5083,
      "step": 30140
    },
    {
      "epoch": 1.6878149848807258,
      "grad_norm": 1.1852613687515259,
      "learning_rate": 9.733500000000002e-05,
      "loss": 0.3672,
      "step": 30141
    },
    {
      "epoch": 1.6878709821928548,
      "grad_norm": 1.3586770296096802,
      "learning_rate": 9.733473684210526e-05,
      "loss": 0.5802,
      "step": 30142
    },
    {
      "epoch": 1.6879269795049838,
      "grad_norm": 1.1743924617767334,
      "learning_rate": 9.733447368421054e-05,
      "loss": 0.3909,
      "step": 30143
    },
    {
      "epoch": 1.6879829768171128,
      "grad_norm": 1.2252435684204102,
      "learning_rate": 9.73342105263158e-05,
      "loss": 0.3919,
      "step": 30144
    },
    {
      "epoch": 1.6880389741292419,
      "grad_norm": 1.4737225770950317,
      "learning_rate": 9.733394736842105e-05,
      "loss": 0.4831,
      "step": 30145
    },
    {
      "epoch": 1.6880949714413709,
      "grad_norm": 1.5884093046188354,
      "learning_rate": 9.733368421052633e-05,
      "loss": 0.3723,
      "step": 30146
    },
    {
      "epoch": 1.6881509687535,
      "grad_norm": 1.301109790802002,
      "learning_rate": 9.733342105263157e-05,
      "loss": 0.5235,
      "step": 30147
    },
    {
      "epoch": 1.688206966065629,
      "grad_norm": 1.361692190170288,
      "learning_rate": 9.733315789473685e-05,
      "loss": 0.5165,
      "step": 30148
    },
    {
      "epoch": 1.688262963377758,
      "grad_norm": 1.2293809652328491,
      "learning_rate": 9.733289473684211e-05,
      "loss": 0.4645,
      "step": 30149
    },
    {
      "epoch": 1.688318960689887,
      "grad_norm": 1.2032302618026733,
      "learning_rate": 9.733263157894738e-05,
      "loss": 0.5969,
      "step": 30150
    },
    {
      "epoch": 1.688374958002016,
      "grad_norm": 1.4731817245483398,
      "learning_rate": 9.733236842105263e-05,
      "loss": 0.4357,
      "step": 30151
    },
    {
      "epoch": 1.688430955314145,
      "grad_norm": 1.2632147073745728,
      "learning_rate": 9.73321052631579e-05,
      "loss": 0.4602,
      "step": 30152
    },
    {
      "epoch": 1.688486952626274,
      "grad_norm": 1.2067543268203735,
      "learning_rate": 9.733184210526316e-05,
      "loss": 0.3663,
      "step": 30153
    },
    {
      "epoch": 1.688542949938403,
      "grad_norm": 1.1950796842575073,
      "learning_rate": 9.733157894736843e-05,
      "loss": 0.359,
      "step": 30154
    },
    {
      "epoch": 1.688598947250532,
      "grad_norm": 1.176719069480896,
      "learning_rate": 9.733131578947369e-05,
      "loss": 0.4632,
      "step": 30155
    },
    {
      "epoch": 1.688654944562661,
      "grad_norm": 1.5808255672454834,
      "learning_rate": 9.733105263157895e-05,
      "loss": 0.4925,
      "step": 30156
    },
    {
      "epoch": 1.68871094187479,
      "grad_norm": 1.0797353982925415,
      "learning_rate": 9.733078947368421e-05,
      "loss": 0.3291,
      "step": 30157
    },
    {
      "epoch": 1.6887669391869191,
      "grad_norm": 1.3204952478408813,
      "learning_rate": 9.733052631578949e-05,
      "loss": 0.3987,
      "step": 30158
    },
    {
      "epoch": 1.6888229364990481,
      "grad_norm": 1.4973983764648438,
      "learning_rate": 9.733026315789475e-05,
      "loss": 0.4919,
      "step": 30159
    },
    {
      "epoch": 1.6888789338111772,
      "grad_norm": 1.1468061208724976,
      "learning_rate": 9.733e-05,
      "loss": 0.394,
      "step": 30160
    },
    {
      "epoch": 1.6889349311233062,
      "grad_norm": 1.3484851121902466,
      "learning_rate": 9.732973684210526e-05,
      "loss": 0.383,
      "step": 30161
    },
    {
      "epoch": 1.6889909284354352,
      "grad_norm": 1.2262699604034424,
      "learning_rate": 9.732947368421052e-05,
      "loss": 0.4681,
      "step": 30162
    },
    {
      "epoch": 1.6890469257475642,
      "grad_norm": 1.2382701635360718,
      "learning_rate": 9.73292105263158e-05,
      "loss": 0.3634,
      "step": 30163
    },
    {
      "epoch": 1.6891029230596932,
      "grad_norm": 1.236656904220581,
      "learning_rate": 9.732894736842106e-05,
      "loss": 0.5847,
      "step": 30164
    },
    {
      "epoch": 1.6891589203718222,
      "grad_norm": 1.5243562459945679,
      "learning_rate": 9.732868421052632e-05,
      "loss": 0.4452,
      "step": 30165
    },
    {
      "epoch": 1.6892149176839513,
      "grad_norm": 1.3775279521942139,
      "learning_rate": 9.732842105263158e-05,
      "loss": 0.466,
      "step": 30166
    },
    {
      "epoch": 1.6892709149960803,
      "grad_norm": 1.2446755170822144,
      "learning_rate": 9.732815789473685e-05,
      "loss": 0.4259,
      "step": 30167
    },
    {
      "epoch": 1.6893269123082093,
      "grad_norm": 1.2202911376953125,
      "learning_rate": 9.732789473684211e-05,
      "loss": 0.4717,
      "step": 30168
    },
    {
      "epoch": 1.6893829096203383,
      "grad_norm": 1.2155464887619019,
      "learning_rate": 9.732763157894737e-05,
      "loss": 0.4708,
      "step": 30169
    },
    {
      "epoch": 1.6894389069324673,
      "grad_norm": 1.3304054737091064,
      "learning_rate": 9.732736842105263e-05,
      "loss": 0.4863,
      "step": 30170
    },
    {
      "epoch": 1.6894949042445964,
      "grad_norm": 1.3756487369537354,
      "learning_rate": 9.73271052631579e-05,
      "loss": 0.5803,
      "step": 30171
    },
    {
      "epoch": 1.6895509015567254,
      "grad_norm": 1.2009427547454834,
      "learning_rate": 9.732684210526316e-05,
      "loss": 0.415,
      "step": 30172
    },
    {
      "epoch": 1.6896068988688544,
      "grad_norm": 1.3241255283355713,
      "learning_rate": 9.732657894736844e-05,
      "loss": 0.5297,
      "step": 30173
    },
    {
      "epoch": 1.6896628961809834,
      "grad_norm": 1.2313463687896729,
      "learning_rate": 9.732631578947368e-05,
      "loss": 0.4625,
      "step": 30174
    },
    {
      "epoch": 1.6897188934931124,
      "grad_norm": 1.45566725730896,
      "learning_rate": 9.732605263157896e-05,
      "loss": 0.461,
      "step": 30175
    },
    {
      "epoch": 1.6897748908052415,
      "grad_norm": 1.1152372360229492,
      "learning_rate": 9.732578947368421e-05,
      "loss": 0.4818,
      "step": 30176
    },
    {
      "epoch": 1.6898308881173705,
      "grad_norm": 1.3938652276992798,
      "learning_rate": 9.732552631578949e-05,
      "loss": 0.3975,
      "step": 30177
    },
    {
      "epoch": 1.6898868854294995,
      "grad_norm": 1.3016711473464966,
      "learning_rate": 9.732526315789475e-05,
      "loss": 0.4593,
      "step": 30178
    },
    {
      "epoch": 1.6899428827416285,
      "grad_norm": 1.3438842296600342,
      "learning_rate": 9.7325e-05,
      "loss": 0.4991,
      "step": 30179
    },
    {
      "epoch": 1.6899988800537575,
      "grad_norm": 1.6723095178604126,
      "learning_rate": 9.732473684210527e-05,
      "loss": 0.4055,
      "step": 30180
    },
    {
      "epoch": 1.6900548773658866,
      "grad_norm": 1.1403666734695435,
      "learning_rate": 9.732447368421053e-05,
      "loss": 0.4457,
      "step": 30181
    },
    {
      "epoch": 1.6901108746780156,
      "grad_norm": 1.7278417348861694,
      "learning_rate": 9.73242105263158e-05,
      "loss": 0.6594,
      "step": 30182
    },
    {
      "epoch": 1.6901668719901446,
      "grad_norm": 1.2462953329086304,
      "learning_rate": 9.732394736842106e-05,
      "loss": 0.3927,
      "step": 30183
    },
    {
      "epoch": 1.6902228693022736,
      "grad_norm": 1.1908814907073975,
      "learning_rate": 9.732368421052632e-05,
      "loss": 0.4063,
      "step": 30184
    },
    {
      "epoch": 1.6902788666144026,
      "grad_norm": 1.24465811252594,
      "learning_rate": 9.732342105263158e-05,
      "loss": 0.4597,
      "step": 30185
    },
    {
      "epoch": 1.6903348639265316,
      "grad_norm": 1.2822506427764893,
      "learning_rate": 9.732315789473685e-05,
      "loss": 0.3647,
      "step": 30186
    },
    {
      "epoch": 1.6903908612386607,
      "grad_norm": 1.6926192045211792,
      "learning_rate": 9.732289473684211e-05,
      "loss": 0.5835,
      "step": 30187
    },
    {
      "epoch": 1.6904468585507897,
      "grad_norm": 1.2964590787887573,
      "learning_rate": 9.732263157894737e-05,
      "loss": 0.4284,
      "step": 30188
    },
    {
      "epoch": 1.6905028558629187,
      "grad_norm": 1.1388252973556519,
      "learning_rate": 9.732236842105263e-05,
      "loss": 0.5062,
      "step": 30189
    },
    {
      "epoch": 1.6905588531750477,
      "grad_norm": 1.3133678436279297,
      "learning_rate": 9.73221052631579e-05,
      "loss": 0.4705,
      "step": 30190
    },
    {
      "epoch": 1.6906148504871767,
      "grad_norm": 1.310320496559143,
      "learning_rate": 9.732184210526316e-05,
      "loss": 0.5138,
      "step": 30191
    },
    {
      "epoch": 1.6906708477993058,
      "grad_norm": 1.4054476022720337,
      "learning_rate": 9.732157894736842e-05,
      "loss": 0.5325,
      "step": 30192
    },
    {
      "epoch": 1.6907268451114348,
      "grad_norm": 1.578825831413269,
      "learning_rate": 9.732131578947368e-05,
      "loss": 0.4885,
      "step": 30193
    },
    {
      "epoch": 1.6907828424235638,
      "grad_norm": 1.1037230491638184,
      "learning_rate": 9.732105263157896e-05,
      "loss": 0.3731,
      "step": 30194
    },
    {
      "epoch": 1.6908388397356928,
      "grad_norm": 1.074224829673767,
      "learning_rate": 9.732078947368422e-05,
      "loss": 0.4612,
      "step": 30195
    },
    {
      "epoch": 1.6908948370478218,
      "grad_norm": 1.2621155977249146,
      "learning_rate": 9.732052631578948e-05,
      "loss": 0.41,
      "step": 30196
    },
    {
      "epoch": 1.6909508343599509,
      "grad_norm": 1.3471922874450684,
      "learning_rate": 9.732026315789474e-05,
      "loss": 0.4656,
      "step": 30197
    },
    {
      "epoch": 1.6910068316720799,
      "grad_norm": 1.1260102987289429,
      "learning_rate": 9.732e-05,
      "loss": 0.5334,
      "step": 30198
    },
    {
      "epoch": 1.691062828984209,
      "grad_norm": 1.2001433372497559,
      "learning_rate": 9.731973684210527e-05,
      "loss": 0.4569,
      "step": 30199
    },
    {
      "epoch": 1.691118826296338,
      "grad_norm": 1.347712516784668,
      "learning_rate": 9.731947368421053e-05,
      "loss": 0.553,
      "step": 30200
    },
    {
      "epoch": 1.691174823608467,
      "grad_norm": 1.417059302330017,
      "learning_rate": 9.731921052631579e-05,
      "loss": 0.6363,
      "step": 30201
    },
    {
      "epoch": 1.691230820920596,
      "grad_norm": 1.2156375646591187,
      "learning_rate": 9.731894736842105e-05,
      "loss": 0.4198,
      "step": 30202
    },
    {
      "epoch": 1.691286818232725,
      "grad_norm": 1.3916971683502197,
      "learning_rate": 9.731868421052632e-05,
      "loss": 0.4246,
      "step": 30203
    },
    {
      "epoch": 1.691342815544854,
      "grad_norm": 1.130895972251892,
      "learning_rate": 9.731842105263158e-05,
      "loss": 0.3977,
      "step": 30204
    },
    {
      "epoch": 1.691398812856983,
      "grad_norm": 1.2875492572784424,
      "learning_rate": 9.731815789473686e-05,
      "loss": 0.5469,
      "step": 30205
    },
    {
      "epoch": 1.691454810169112,
      "grad_norm": 1.2050881385803223,
      "learning_rate": 9.73178947368421e-05,
      "loss": 0.4667,
      "step": 30206
    },
    {
      "epoch": 1.691510807481241,
      "grad_norm": 1.1337792873382568,
      "learning_rate": 9.731763157894737e-05,
      "loss": 0.4197,
      "step": 30207
    },
    {
      "epoch": 1.69156680479337,
      "grad_norm": 1.0673487186431885,
      "learning_rate": 9.731736842105263e-05,
      "loss": 0.3491,
      "step": 30208
    },
    {
      "epoch": 1.691622802105499,
      "grad_norm": 1.392971158027649,
      "learning_rate": 9.731710526315791e-05,
      "loss": 0.6227,
      "step": 30209
    },
    {
      "epoch": 1.691678799417628,
      "grad_norm": 1.138738989830017,
      "learning_rate": 9.731684210526317e-05,
      "loss": 0.4073,
      "step": 30210
    },
    {
      "epoch": 1.6917347967297571,
      "grad_norm": 1.3383424282073975,
      "learning_rate": 9.731657894736843e-05,
      "loss": 0.4484,
      "step": 30211
    },
    {
      "epoch": 1.6917907940418861,
      "grad_norm": 1.3517189025878906,
      "learning_rate": 9.731631578947369e-05,
      "loss": 0.3765,
      "step": 30212
    },
    {
      "epoch": 1.6918467913540152,
      "grad_norm": 1.1367801427841187,
      "learning_rate": 9.731605263157895e-05,
      "loss": 0.3876,
      "step": 30213
    },
    {
      "epoch": 1.6919027886661442,
      "grad_norm": 1.1613292694091797,
      "learning_rate": 9.731578947368422e-05,
      "loss": 0.4214,
      "step": 30214
    },
    {
      "epoch": 1.6919587859782732,
      "grad_norm": 1.3283835649490356,
      "learning_rate": 9.731552631578948e-05,
      "loss": 0.382,
      "step": 30215
    },
    {
      "epoch": 1.6920147832904022,
      "grad_norm": 2.5240118503570557,
      "learning_rate": 9.731526315789474e-05,
      "loss": 0.4319,
      "step": 30216
    },
    {
      "epoch": 1.6920707806025312,
      "grad_norm": 1.573872685432434,
      "learning_rate": 9.7315e-05,
      "loss": 0.4483,
      "step": 30217
    },
    {
      "epoch": 1.6921267779146603,
      "grad_norm": 1.1742709875106812,
      "learning_rate": 9.731473684210527e-05,
      "loss": 0.5286,
      "step": 30218
    },
    {
      "epoch": 1.692182775226789,
      "grad_norm": 1.2130649089813232,
      "learning_rate": 9.731447368421053e-05,
      "loss": 0.4015,
      "step": 30219
    },
    {
      "epoch": 1.692238772538918,
      "grad_norm": 1.6368238925933838,
      "learning_rate": 9.731421052631579e-05,
      "loss": 0.6805,
      "step": 30220
    },
    {
      "epoch": 1.692294769851047,
      "grad_norm": 1.2728393077850342,
      "learning_rate": 9.731394736842105e-05,
      "loss": 0.4743,
      "step": 30221
    },
    {
      "epoch": 1.6923507671631761,
      "grad_norm": 1.1846604347229004,
      "learning_rate": 9.731368421052632e-05,
      "loss": 0.4199,
      "step": 30222
    },
    {
      "epoch": 1.6924067644753051,
      "grad_norm": 1.1958200931549072,
      "learning_rate": 9.731342105263158e-05,
      "loss": 0.5363,
      "step": 30223
    },
    {
      "epoch": 1.6924627617874342,
      "grad_norm": 1.304478645324707,
      "learning_rate": 9.731315789473684e-05,
      "loss": 0.5076,
      "step": 30224
    },
    {
      "epoch": 1.6925187590995632,
      "grad_norm": 1.5974457263946533,
      "learning_rate": 9.73128947368421e-05,
      "loss": 0.5178,
      "step": 30225
    },
    {
      "epoch": 1.6925747564116922,
      "grad_norm": 0.9744231700897217,
      "learning_rate": 9.731263157894738e-05,
      "loss": 0.4195,
      "step": 30226
    },
    {
      "epoch": 1.6926307537238212,
      "grad_norm": 1.188340663909912,
      "learning_rate": 9.731236842105264e-05,
      "loss": 0.4446,
      "step": 30227
    },
    {
      "epoch": 1.6926867510359502,
      "grad_norm": 1.4804326295852661,
      "learning_rate": 9.731210526315791e-05,
      "loss": 0.587,
      "step": 30228
    },
    {
      "epoch": 1.6927427483480793,
      "grad_norm": 1.5981879234313965,
      "learning_rate": 9.731184210526316e-05,
      "loss": 0.4969,
      "step": 30229
    },
    {
      "epoch": 1.6927987456602083,
      "grad_norm": 1.0407193899154663,
      "learning_rate": 9.731157894736842e-05,
      "loss": 0.365,
      "step": 30230
    },
    {
      "epoch": 1.6928547429723373,
      "grad_norm": 1.2792677879333496,
      "learning_rate": 9.731131578947369e-05,
      "loss": 0.584,
      "step": 30231
    },
    {
      "epoch": 1.6929107402844663,
      "grad_norm": 1.2894998788833618,
      "learning_rate": 9.731105263157895e-05,
      "loss": 0.4905,
      "step": 30232
    },
    {
      "epoch": 1.6929667375965953,
      "grad_norm": 1.29068124294281,
      "learning_rate": 9.731078947368422e-05,
      "loss": 0.4458,
      "step": 30233
    },
    {
      "epoch": 1.6930227349087243,
      "grad_norm": 1.3974711894989014,
      "learning_rate": 9.731052631578947e-05,
      "loss": 0.5001,
      "step": 30234
    },
    {
      "epoch": 1.6930787322208534,
      "grad_norm": 1.2442253828048706,
      "learning_rate": 9.731026315789474e-05,
      "loss": 0.2887,
      "step": 30235
    },
    {
      "epoch": 1.6931347295329824,
      "grad_norm": 1.3117127418518066,
      "learning_rate": 9.731e-05,
      "loss": 0.4423,
      "step": 30236
    },
    {
      "epoch": 1.6931907268451114,
      "grad_norm": 1.3299121856689453,
      "learning_rate": 9.730973684210528e-05,
      "loss": 0.5066,
      "step": 30237
    },
    {
      "epoch": 1.6932467241572404,
      "grad_norm": 1.3412389755249023,
      "learning_rate": 9.730947368421053e-05,
      "loss": 0.3981,
      "step": 30238
    },
    {
      "epoch": 1.6933027214693694,
      "grad_norm": 1.9414891004562378,
      "learning_rate": 9.73092105263158e-05,
      "loss": 0.5383,
      "step": 30239
    },
    {
      "epoch": 1.6933587187814985,
      "grad_norm": 1.3416924476623535,
      "learning_rate": 9.730894736842105e-05,
      "loss": 0.3939,
      "step": 30240
    },
    {
      "epoch": 1.6934147160936275,
      "grad_norm": 1.4639008045196533,
      "learning_rate": 9.730868421052633e-05,
      "loss": 0.4688,
      "step": 30241
    },
    {
      "epoch": 1.6934707134057565,
      "grad_norm": 1.4314970970153809,
      "learning_rate": 9.730842105263159e-05,
      "loss": 0.4478,
      "step": 30242
    },
    {
      "epoch": 1.6935267107178855,
      "grad_norm": 1.1324588060379028,
      "learning_rate": 9.730815789473685e-05,
      "loss": 0.4582,
      "step": 30243
    },
    {
      "epoch": 1.6935827080300145,
      "grad_norm": 1.1943910121917725,
      "learning_rate": 9.73078947368421e-05,
      "loss": 0.3673,
      "step": 30244
    },
    {
      "epoch": 1.6936387053421436,
      "grad_norm": 1.440931797027588,
      "learning_rate": 9.730763157894738e-05,
      "loss": 0.5169,
      "step": 30245
    },
    {
      "epoch": 1.6936947026542726,
      "grad_norm": 1.5068817138671875,
      "learning_rate": 9.730736842105264e-05,
      "loss": 0.4397,
      "step": 30246
    },
    {
      "epoch": 1.6937506999664016,
      "grad_norm": 1.0686894655227661,
      "learning_rate": 9.73071052631579e-05,
      "loss": 0.4375,
      "step": 30247
    },
    {
      "epoch": 1.6938066972785306,
      "grad_norm": 1.2193092107772827,
      "learning_rate": 9.730684210526316e-05,
      "loss": 0.4532,
      "step": 30248
    },
    {
      "epoch": 1.6938626945906596,
      "grad_norm": 1.2892290353775024,
      "learning_rate": 9.730657894736842e-05,
      "loss": 0.5527,
      "step": 30249
    },
    {
      "epoch": 1.6939186919027887,
      "grad_norm": 1.4779232740402222,
      "learning_rate": 9.730631578947369e-05,
      "loss": 0.4601,
      "step": 30250
    },
    {
      "epoch": 1.6939746892149177,
      "grad_norm": 1.5048413276672363,
      "learning_rate": 9.730605263157895e-05,
      "loss": 0.4469,
      "step": 30251
    },
    {
      "epoch": 1.6940306865270467,
      "grad_norm": 1.171181082725525,
      "learning_rate": 9.730578947368421e-05,
      "loss": 0.422,
      "step": 30252
    },
    {
      "epoch": 1.6940866838391757,
      "grad_norm": 1.3899632692337036,
      "learning_rate": 9.730552631578947e-05,
      "loss": 0.378,
      "step": 30253
    },
    {
      "epoch": 1.6941426811513047,
      "grad_norm": 1.1190667152404785,
      "learning_rate": 9.730526315789474e-05,
      "loss": 0.3367,
      "step": 30254
    },
    {
      "epoch": 1.6941986784634337,
      "grad_norm": 1.3638272285461426,
      "learning_rate": 9.7305e-05,
      "loss": 0.5276,
      "step": 30255
    },
    {
      "epoch": 1.6942546757755628,
      "grad_norm": 1.307783603668213,
      "learning_rate": 9.730473684210526e-05,
      "loss": 0.5087,
      "step": 30256
    },
    {
      "epoch": 1.6943106730876918,
      "grad_norm": 1.137140154838562,
      "learning_rate": 9.730447368421052e-05,
      "loss": 0.4028,
      "step": 30257
    },
    {
      "epoch": 1.6943666703998208,
      "grad_norm": 1.8036515712738037,
      "learning_rate": 9.73042105263158e-05,
      "loss": 0.5313,
      "step": 30258
    },
    {
      "epoch": 1.6944226677119498,
      "grad_norm": 1.3460479974746704,
      "learning_rate": 9.730394736842106e-05,
      "loss": 0.4697,
      "step": 30259
    },
    {
      "epoch": 1.6944786650240788,
      "grad_norm": 1.4281642436981201,
      "learning_rate": 9.730368421052633e-05,
      "loss": 0.5596,
      "step": 30260
    },
    {
      "epoch": 1.6945346623362079,
      "grad_norm": 1.271577000617981,
      "learning_rate": 9.730342105263158e-05,
      "loss": 0.3794,
      "step": 30261
    },
    {
      "epoch": 1.6945906596483369,
      "grad_norm": 1.4523308277130127,
      "learning_rate": 9.730315789473685e-05,
      "loss": 0.5573,
      "step": 30262
    },
    {
      "epoch": 1.694646656960466,
      "grad_norm": 1.1428872346878052,
      "learning_rate": 9.730289473684211e-05,
      "loss": 0.4664,
      "step": 30263
    },
    {
      "epoch": 1.694702654272595,
      "grad_norm": 1.3204830884933472,
      "learning_rate": 9.730263157894738e-05,
      "loss": 0.611,
      "step": 30264
    },
    {
      "epoch": 1.694758651584724,
      "grad_norm": 1.1934540271759033,
      "learning_rate": 9.730236842105264e-05,
      "loss": 0.3084,
      "step": 30265
    },
    {
      "epoch": 1.694814648896853,
      "grad_norm": 1.0935078859329224,
      "learning_rate": 9.730210526315789e-05,
      "loss": 0.41,
      "step": 30266
    },
    {
      "epoch": 1.694870646208982,
      "grad_norm": 1.3033454418182373,
      "learning_rate": 9.730184210526316e-05,
      "loss": 0.5077,
      "step": 30267
    },
    {
      "epoch": 1.694926643521111,
      "grad_norm": 1.4620641469955444,
      "learning_rate": 9.730157894736842e-05,
      "loss": 0.4244,
      "step": 30268
    },
    {
      "epoch": 1.69498264083324,
      "grad_norm": 1.3887606859207153,
      "learning_rate": 9.73013157894737e-05,
      "loss": 0.5175,
      "step": 30269
    },
    {
      "epoch": 1.695038638145369,
      "grad_norm": 1.2040472030639648,
      "learning_rate": 9.730105263157895e-05,
      "loss": 0.4533,
      "step": 30270
    },
    {
      "epoch": 1.695094635457498,
      "grad_norm": 1.8510829210281372,
      "learning_rate": 9.730078947368421e-05,
      "loss": 0.5017,
      "step": 30271
    },
    {
      "epoch": 1.695150632769627,
      "grad_norm": 1.2853960990905762,
      "learning_rate": 9.730052631578947e-05,
      "loss": 0.3734,
      "step": 30272
    },
    {
      "epoch": 1.695206630081756,
      "grad_norm": 1.4359183311462402,
      "learning_rate": 9.730026315789475e-05,
      "loss": 0.4473,
      "step": 30273
    },
    {
      "epoch": 1.6952626273938851,
      "grad_norm": 1.0984580516815186,
      "learning_rate": 9.730000000000001e-05,
      "loss": 0.3766,
      "step": 30274
    },
    {
      "epoch": 1.6953186247060141,
      "grad_norm": 1.4857829809188843,
      "learning_rate": 9.729973684210527e-05,
      "loss": 0.5342,
      "step": 30275
    },
    {
      "epoch": 1.6953746220181432,
      "grad_norm": 1.3552742004394531,
      "learning_rate": 9.729947368421053e-05,
      "loss": 0.3975,
      "step": 30276
    },
    {
      "epoch": 1.6954306193302722,
      "grad_norm": 1.2708369493484497,
      "learning_rate": 9.72992105263158e-05,
      "loss": 0.4116,
      "step": 30277
    },
    {
      "epoch": 1.6954866166424012,
      "grad_norm": 1.426080584526062,
      "learning_rate": 9.729894736842106e-05,
      "loss": 0.5349,
      "step": 30278
    },
    {
      "epoch": 1.6955426139545302,
      "grad_norm": 1.2011563777923584,
      "learning_rate": 9.729868421052632e-05,
      "loss": 0.388,
      "step": 30279
    },
    {
      "epoch": 1.6955986112666592,
      "grad_norm": 1.3876416683197021,
      "learning_rate": 9.729842105263158e-05,
      "loss": 0.3935,
      "step": 30280
    },
    {
      "epoch": 1.6956546085787882,
      "grad_norm": 1.0139832496643066,
      "learning_rate": 9.729815789473685e-05,
      "loss": 0.4902,
      "step": 30281
    },
    {
      "epoch": 1.6957106058909173,
      "grad_norm": 1.6276042461395264,
      "learning_rate": 9.729789473684211e-05,
      "loss": 0.388,
      "step": 30282
    },
    {
      "epoch": 1.6957666032030463,
      "grad_norm": 1.4506871700286865,
      "learning_rate": 9.729763157894737e-05,
      "loss": 0.4478,
      "step": 30283
    },
    {
      "epoch": 1.6958226005151753,
      "grad_norm": 1.4444221258163452,
      "learning_rate": 9.729736842105263e-05,
      "loss": 0.4424,
      "step": 30284
    },
    {
      "epoch": 1.6958785978273043,
      "grad_norm": 1.1168495416641235,
      "learning_rate": 9.729710526315789e-05,
      "loss": 0.3262,
      "step": 30285
    },
    {
      "epoch": 1.6959345951394333,
      "grad_norm": 1.3173397779464722,
      "learning_rate": 9.729684210526316e-05,
      "loss": 0.4328,
      "step": 30286
    },
    {
      "epoch": 1.6959905924515624,
      "grad_norm": 1.8876169919967651,
      "learning_rate": 9.729657894736842e-05,
      "loss": 0.4391,
      "step": 30287
    },
    {
      "epoch": 1.6960465897636914,
      "grad_norm": 1.3588616847991943,
      "learning_rate": 9.72963157894737e-05,
      "loss": 0.3272,
      "step": 30288
    },
    {
      "epoch": 1.6961025870758204,
      "grad_norm": 1.0950651168823242,
      "learning_rate": 9.729605263157894e-05,
      "loss": 0.4377,
      "step": 30289
    },
    {
      "epoch": 1.6961585843879494,
      "grad_norm": 1.5283358097076416,
      "learning_rate": 9.729578947368422e-05,
      "loss": 0.4147,
      "step": 30290
    },
    {
      "epoch": 1.6962145817000784,
      "grad_norm": 1.4009937047958374,
      "learning_rate": 9.729552631578948e-05,
      "loss": 0.5298,
      "step": 30291
    },
    {
      "epoch": 1.6962705790122075,
      "grad_norm": 1.4278610944747925,
      "learning_rate": 9.729526315789475e-05,
      "loss": 0.4376,
      "step": 30292
    },
    {
      "epoch": 1.6963265763243365,
      "grad_norm": 2.084202289581299,
      "learning_rate": 9.729500000000001e-05,
      "loss": 0.6569,
      "step": 30293
    },
    {
      "epoch": 1.6963825736364655,
      "grad_norm": 1.378486156463623,
      "learning_rate": 9.729473684210527e-05,
      "loss": 0.4371,
      "step": 30294
    },
    {
      "epoch": 1.6964385709485945,
      "grad_norm": 1.3103843927383423,
      "learning_rate": 9.729447368421053e-05,
      "loss": 0.4739,
      "step": 30295
    },
    {
      "epoch": 1.6964945682607235,
      "grad_norm": 1.1588494777679443,
      "learning_rate": 9.72942105263158e-05,
      "loss": 0.3367,
      "step": 30296
    },
    {
      "epoch": 1.6965505655728526,
      "grad_norm": 1.5184040069580078,
      "learning_rate": 9.729394736842106e-05,
      "loss": 0.7053,
      "step": 30297
    },
    {
      "epoch": 1.6966065628849816,
      "grad_norm": 1.2309514284133911,
      "learning_rate": 9.729368421052632e-05,
      "loss": 0.3411,
      "step": 30298
    },
    {
      "epoch": 1.6966625601971106,
      "grad_norm": 1.477604866027832,
      "learning_rate": 9.729342105263158e-05,
      "loss": 0.6026,
      "step": 30299
    },
    {
      "epoch": 1.6967185575092396,
      "grad_norm": 1.1118342876434326,
      "learning_rate": 9.729315789473684e-05,
      "loss": 0.4385,
      "step": 30300
    },
    {
      "epoch": 1.6967745548213684,
      "grad_norm": 1.553444266319275,
      "learning_rate": 9.729289473684211e-05,
      "loss": 0.6059,
      "step": 30301
    },
    {
      "epoch": 1.6968305521334974,
      "grad_norm": 1.1063417196273804,
      "learning_rate": 9.729263157894737e-05,
      "loss": 0.3791,
      "step": 30302
    },
    {
      "epoch": 1.6968865494456264,
      "grad_norm": 1.128703236579895,
      "learning_rate": 9.729236842105263e-05,
      "loss": 0.3458,
      "step": 30303
    },
    {
      "epoch": 1.6969425467577555,
      "grad_norm": 1.374710202217102,
      "learning_rate": 9.72921052631579e-05,
      "loss": 0.5214,
      "step": 30304
    },
    {
      "epoch": 1.6969985440698845,
      "grad_norm": 1.659647822380066,
      "learning_rate": 9.729184210526317e-05,
      "loss": 0.5203,
      "step": 30305
    },
    {
      "epoch": 1.6970545413820135,
      "grad_norm": 1.23285710811615,
      "learning_rate": 9.729157894736843e-05,
      "loss": 0.588,
      "step": 30306
    },
    {
      "epoch": 1.6971105386941425,
      "grad_norm": 1.2525181770324707,
      "learning_rate": 9.729131578947369e-05,
      "loss": 0.3612,
      "step": 30307
    },
    {
      "epoch": 1.6971665360062715,
      "grad_norm": 1.2933460474014282,
      "learning_rate": 9.729105263157895e-05,
      "loss": 0.4537,
      "step": 30308
    },
    {
      "epoch": 1.6972225333184006,
      "grad_norm": 1.0990279912948608,
      "learning_rate": 9.729078947368422e-05,
      "loss": 0.3874,
      "step": 30309
    },
    {
      "epoch": 1.6972785306305296,
      "grad_norm": 1.3627982139587402,
      "learning_rate": 9.729052631578948e-05,
      "loss": 0.4063,
      "step": 30310
    },
    {
      "epoch": 1.6973345279426586,
      "grad_norm": 1.2585375308990479,
      "learning_rate": 9.729026315789474e-05,
      "loss": 0.3325,
      "step": 30311
    },
    {
      "epoch": 1.6973905252547876,
      "grad_norm": 1.0967957973480225,
      "learning_rate": 9.729e-05,
      "loss": 0.3409,
      "step": 30312
    },
    {
      "epoch": 1.6974465225669166,
      "grad_norm": 1.286448359489441,
      "learning_rate": 9.728973684210527e-05,
      "loss": 0.3923,
      "step": 30313
    },
    {
      "epoch": 1.6975025198790457,
      "grad_norm": 1.4949727058410645,
      "learning_rate": 9.728947368421053e-05,
      "loss": 0.4739,
      "step": 30314
    },
    {
      "epoch": 1.6975585171911747,
      "grad_norm": 1.621005654335022,
      "learning_rate": 9.72892105263158e-05,
      "loss": 0.6308,
      "step": 30315
    },
    {
      "epoch": 1.6976145145033037,
      "grad_norm": 1.20448637008667,
      "learning_rate": 9.728894736842105e-05,
      "loss": 0.4985,
      "step": 30316
    },
    {
      "epoch": 1.6976705118154327,
      "grad_norm": 1.2671122550964355,
      "learning_rate": 9.728868421052631e-05,
      "loss": 0.387,
      "step": 30317
    },
    {
      "epoch": 1.6977265091275617,
      "grad_norm": 1.430680751800537,
      "learning_rate": 9.728842105263158e-05,
      "loss": 0.574,
      "step": 30318
    },
    {
      "epoch": 1.6977825064396908,
      "grad_norm": 1.3269802331924438,
      "learning_rate": 9.728815789473684e-05,
      "loss": 0.5057,
      "step": 30319
    },
    {
      "epoch": 1.6978385037518198,
      "grad_norm": 2.183607339859009,
      "learning_rate": 9.728789473684212e-05,
      "loss": 0.5895,
      "step": 30320
    },
    {
      "epoch": 1.6978945010639488,
      "grad_norm": 1.1722875833511353,
      "learning_rate": 9.728763157894736e-05,
      "loss": 0.3825,
      "step": 30321
    },
    {
      "epoch": 1.6979504983760778,
      "grad_norm": 1.1483943462371826,
      "learning_rate": 9.728736842105264e-05,
      "loss": 0.3443,
      "step": 30322
    },
    {
      "epoch": 1.6980064956882068,
      "grad_norm": 1.5114575624465942,
      "learning_rate": 9.72871052631579e-05,
      "loss": 0.4387,
      "step": 30323
    },
    {
      "epoch": 1.6980624930003358,
      "grad_norm": 1.5740960836410522,
      "learning_rate": 9.728684210526317e-05,
      "loss": 0.4974,
      "step": 30324
    },
    {
      "epoch": 1.6981184903124649,
      "grad_norm": 1.171128273010254,
      "learning_rate": 9.728657894736843e-05,
      "loss": 0.4277,
      "step": 30325
    },
    {
      "epoch": 1.6981744876245939,
      "grad_norm": 1.138498067855835,
      "learning_rate": 9.728631578947369e-05,
      "loss": 0.4725,
      "step": 30326
    },
    {
      "epoch": 1.698230484936723,
      "grad_norm": 1.3775825500488281,
      "learning_rate": 9.728605263157895e-05,
      "loss": 0.5069,
      "step": 30327
    },
    {
      "epoch": 1.698286482248852,
      "grad_norm": 1.1714247465133667,
      "learning_rate": 9.728578947368422e-05,
      "loss": 0.3434,
      "step": 30328
    },
    {
      "epoch": 1.698342479560981,
      "grad_norm": 1.5268391370773315,
      "learning_rate": 9.728552631578948e-05,
      "loss": 0.3696,
      "step": 30329
    },
    {
      "epoch": 1.69839847687311,
      "grad_norm": 1.343772292137146,
      "learning_rate": 9.728526315789474e-05,
      "loss": 0.5687,
      "step": 30330
    },
    {
      "epoch": 1.698454474185239,
      "grad_norm": 64.54393005371094,
      "learning_rate": 9.7285e-05,
      "loss": 0.4503,
      "step": 30331
    },
    {
      "epoch": 1.698510471497368,
      "grad_norm": 1.3343149423599243,
      "learning_rate": 9.728473684210527e-05,
      "loss": 0.4719,
      "step": 30332
    },
    {
      "epoch": 1.698566468809497,
      "grad_norm": 1.236086130142212,
      "learning_rate": 9.728447368421053e-05,
      "loss": 0.4541,
      "step": 30333
    },
    {
      "epoch": 1.698622466121626,
      "grad_norm": 1.2689766883850098,
      "learning_rate": 9.72842105263158e-05,
      "loss": 0.4627,
      "step": 30334
    },
    {
      "epoch": 1.698678463433755,
      "grad_norm": 1.5581690073013306,
      "learning_rate": 9.728394736842105e-05,
      "loss": 0.576,
      "step": 30335
    },
    {
      "epoch": 1.698734460745884,
      "grad_norm": 1.436468482017517,
      "learning_rate": 9.728368421052631e-05,
      "loss": 0.4485,
      "step": 30336
    },
    {
      "epoch": 1.698790458058013,
      "grad_norm": 1.2061339616775513,
      "learning_rate": 9.728342105263159e-05,
      "loss": 0.4498,
      "step": 30337
    },
    {
      "epoch": 1.6988464553701421,
      "grad_norm": 1.6089482307434082,
      "learning_rate": 9.728315789473685e-05,
      "loss": 0.5003,
      "step": 30338
    },
    {
      "epoch": 1.6989024526822711,
      "grad_norm": 1.0775550603866577,
      "learning_rate": 9.72828947368421e-05,
      "loss": 0.3458,
      "step": 30339
    },
    {
      "epoch": 1.6989584499944002,
      "grad_norm": 1.5468053817749023,
      "learning_rate": 9.728263157894737e-05,
      "loss": 0.7323,
      "step": 30340
    },
    {
      "epoch": 1.6990144473065292,
      "grad_norm": 1.2649682760238647,
      "learning_rate": 9.728236842105264e-05,
      "loss": 0.4295,
      "step": 30341
    },
    {
      "epoch": 1.6990704446186582,
      "grad_norm": 1.3385552167892456,
      "learning_rate": 9.72821052631579e-05,
      "loss": 0.5028,
      "step": 30342
    },
    {
      "epoch": 1.6991264419307872,
      "grad_norm": 1.2830957174301147,
      "learning_rate": 9.728184210526317e-05,
      "loss": 0.4697,
      "step": 30343
    },
    {
      "epoch": 1.6991824392429162,
      "grad_norm": 2.1357758045196533,
      "learning_rate": 9.728157894736842e-05,
      "loss": 0.4494,
      "step": 30344
    },
    {
      "epoch": 1.6992384365550453,
      "grad_norm": 1.2747297286987305,
      "learning_rate": 9.728131578947369e-05,
      "loss": 0.5061,
      "step": 30345
    },
    {
      "epoch": 1.6992944338671743,
      "grad_norm": 1.3982797861099243,
      "learning_rate": 9.728105263157895e-05,
      "loss": 0.4909,
      "step": 30346
    },
    {
      "epoch": 1.6993504311793033,
      "grad_norm": 1.4929579496383667,
      "learning_rate": 9.728078947368422e-05,
      "loss": 0.5826,
      "step": 30347
    },
    {
      "epoch": 1.6994064284914323,
      "grad_norm": 1.6980502605438232,
      "learning_rate": 9.728052631578948e-05,
      "loss": 0.5291,
      "step": 30348
    },
    {
      "epoch": 1.6994624258035613,
      "grad_norm": 1.1305783987045288,
      "learning_rate": 9.728026315789474e-05,
      "loss": 0.4975,
      "step": 30349
    },
    {
      "epoch": 1.6995184231156903,
      "grad_norm": 1.2212214469909668,
      "learning_rate": 9.728e-05,
      "loss": 0.4564,
      "step": 30350
    },
    {
      "epoch": 1.6995744204278194,
      "grad_norm": 1.3931679725646973,
      "learning_rate": 9.727973684210526e-05,
      "loss": 0.6455,
      "step": 30351
    },
    {
      "epoch": 1.6996304177399484,
      "grad_norm": 1.271008014678955,
      "learning_rate": 9.727947368421054e-05,
      "loss": 0.4099,
      "step": 30352
    },
    {
      "epoch": 1.6996864150520774,
      "grad_norm": 1.3081786632537842,
      "learning_rate": 9.727921052631578e-05,
      "loss": 0.4813,
      "step": 30353
    },
    {
      "epoch": 1.6997424123642064,
      "grad_norm": 1.37801194190979,
      "learning_rate": 9.727894736842106e-05,
      "loss": 0.451,
      "step": 30354
    },
    {
      "epoch": 1.6997984096763354,
      "grad_norm": 1.640711784362793,
      "learning_rate": 9.727868421052632e-05,
      "loss": 0.5858,
      "step": 30355
    },
    {
      "epoch": 1.6998544069884645,
      "grad_norm": 1.121232032775879,
      "learning_rate": 9.727842105263159e-05,
      "loss": 0.4415,
      "step": 30356
    },
    {
      "epoch": 1.6999104043005935,
      "grad_norm": 1.3477303981781006,
      "learning_rate": 9.727815789473685e-05,
      "loss": 0.4505,
      "step": 30357
    },
    {
      "epoch": 1.6999664016127225,
      "grad_norm": 2.0943679809570312,
      "learning_rate": 9.727789473684211e-05,
      "loss": 0.4611,
      "step": 30358
    },
    {
      "epoch": 1.7000223989248515,
      "grad_norm": 1.7255926132202148,
      "learning_rate": 9.727763157894737e-05,
      "loss": 0.4871,
      "step": 30359
    },
    {
      "epoch": 1.7000783962369805,
      "grad_norm": 1.3967081308364868,
      "learning_rate": 9.727736842105264e-05,
      "loss": 0.4439,
      "step": 30360
    },
    {
      "epoch": 1.7001343935491096,
      "grad_norm": 1.5152515172958374,
      "learning_rate": 9.72771052631579e-05,
      "loss": 0.4556,
      "step": 30361
    },
    {
      "epoch": 1.7001903908612386,
      "grad_norm": 1.40230131149292,
      "learning_rate": 9.727684210526316e-05,
      "loss": 0.5182,
      "step": 30362
    },
    {
      "epoch": 1.7002463881733676,
      "grad_norm": 1.2221062183380127,
      "learning_rate": 9.727657894736842e-05,
      "loss": 0.4226,
      "step": 30363
    },
    {
      "epoch": 1.7003023854854966,
      "grad_norm": 6.97371768951416,
      "learning_rate": 9.72763157894737e-05,
      "loss": 0.4474,
      "step": 30364
    },
    {
      "epoch": 1.7003583827976256,
      "grad_norm": 1.3779109716415405,
      "learning_rate": 9.727605263157895e-05,
      "loss": 0.4449,
      "step": 30365
    },
    {
      "epoch": 1.7004143801097547,
      "grad_norm": 1.1985836029052734,
      "learning_rate": 9.727578947368421e-05,
      "loss": 0.3976,
      "step": 30366
    },
    {
      "epoch": 1.7004703774218837,
      "grad_norm": 1.293669581413269,
      "learning_rate": 9.727552631578947e-05,
      "loss": 0.4866,
      "step": 30367
    },
    {
      "epoch": 1.7005263747340127,
      "grad_norm": 1.30210280418396,
      "learning_rate": 9.727526315789473e-05,
      "loss": 0.4353,
      "step": 30368
    },
    {
      "epoch": 1.7005823720461417,
      "grad_norm": 1.4130703210830688,
      "learning_rate": 9.7275e-05,
      "loss": 0.4157,
      "step": 30369
    },
    {
      "epoch": 1.7006383693582707,
      "grad_norm": 1.3582077026367188,
      "learning_rate": 9.727473684210527e-05,
      "loss": 0.5089,
      "step": 30370
    },
    {
      "epoch": 1.7006943666703997,
      "grad_norm": 1.5402058362960815,
      "learning_rate": 9.727447368421053e-05,
      "loss": 0.5706,
      "step": 30371
    },
    {
      "epoch": 1.7007503639825288,
      "grad_norm": 1.088727593421936,
      "learning_rate": 9.727421052631579e-05,
      "loss": 0.3592,
      "step": 30372
    },
    {
      "epoch": 1.7008063612946578,
      "grad_norm": 1.4988527297973633,
      "learning_rate": 9.727394736842106e-05,
      "loss": 0.3488,
      "step": 30373
    },
    {
      "epoch": 1.7008623586067868,
      "grad_norm": 1.0835514068603516,
      "learning_rate": 9.727368421052632e-05,
      "loss": 0.3906,
      "step": 30374
    },
    {
      "epoch": 1.7009183559189158,
      "grad_norm": 1.2537224292755127,
      "learning_rate": 9.727342105263159e-05,
      "loss": 0.4103,
      "step": 30375
    },
    {
      "epoch": 1.7009743532310448,
      "grad_norm": 1.2662135362625122,
      "learning_rate": 9.727315789473684e-05,
      "loss": 0.4934,
      "step": 30376
    },
    {
      "epoch": 1.7010303505431739,
      "grad_norm": 1.1604135036468506,
      "learning_rate": 9.727289473684211e-05,
      "loss": 0.3871,
      "step": 30377
    },
    {
      "epoch": 1.7010863478553029,
      "grad_norm": 1.2488824129104614,
      "learning_rate": 9.727263157894737e-05,
      "loss": 0.3867,
      "step": 30378
    },
    {
      "epoch": 1.701142345167432,
      "grad_norm": 1.395504355430603,
      "learning_rate": 9.727236842105264e-05,
      "loss": 0.4575,
      "step": 30379
    },
    {
      "epoch": 1.701198342479561,
      "grad_norm": 1.2531800270080566,
      "learning_rate": 9.72721052631579e-05,
      "loss": 0.4608,
      "step": 30380
    },
    {
      "epoch": 1.70125433979169,
      "grad_norm": 1.281958818435669,
      "learning_rate": 9.727184210526316e-05,
      "loss": 0.4547,
      "step": 30381
    },
    {
      "epoch": 1.701310337103819,
      "grad_norm": 1.3307981491088867,
      "learning_rate": 9.727157894736842e-05,
      "loss": 0.3593,
      "step": 30382
    },
    {
      "epoch": 1.701366334415948,
      "grad_norm": 1.6306116580963135,
      "learning_rate": 9.72713157894737e-05,
      "loss": 0.58,
      "step": 30383
    },
    {
      "epoch": 1.701422331728077,
      "grad_norm": 1.287005066871643,
      "learning_rate": 9.727105263157896e-05,
      "loss": 0.4737,
      "step": 30384
    },
    {
      "epoch": 1.701478329040206,
      "grad_norm": 1.359609603881836,
      "learning_rate": 9.727078947368422e-05,
      "loss": 0.5492,
      "step": 30385
    },
    {
      "epoch": 1.701534326352335,
      "grad_norm": 1.3711774349212646,
      "learning_rate": 9.727052631578948e-05,
      "loss": 0.4229,
      "step": 30386
    },
    {
      "epoch": 1.701590323664464,
      "grad_norm": 1.237109899520874,
      "learning_rate": 9.727026315789474e-05,
      "loss": 0.4392,
      "step": 30387
    },
    {
      "epoch": 1.701646320976593,
      "grad_norm": 1.3945149183273315,
      "learning_rate": 9.727000000000001e-05,
      "loss": 0.4245,
      "step": 30388
    },
    {
      "epoch": 1.701702318288722,
      "grad_norm": 1.4769854545593262,
      "learning_rate": 9.726973684210527e-05,
      "loss": 0.4906,
      "step": 30389
    },
    {
      "epoch": 1.7017583156008511,
      "grad_norm": 1.082001805305481,
      "learning_rate": 9.726947368421053e-05,
      "loss": 0.3421,
      "step": 30390
    },
    {
      "epoch": 1.7018143129129801,
      "grad_norm": 1.4194084405899048,
      "learning_rate": 9.726921052631579e-05,
      "loss": 0.4602,
      "step": 30391
    },
    {
      "epoch": 1.7018703102251092,
      "grad_norm": 1.1614925861358643,
      "learning_rate": 9.726894736842106e-05,
      "loss": 0.5332,
      "step": 30392
    },
    {
      "epoch": 1.7019263075372382,
      "grad_norm": 1.0861917734146118,
      "learning_rate": 9.726868421052632e-05,
      "loss": 0.377,
      "step": 30393
    },
    {
      "epoch": 1.7019823048493672,
      "grad_norm": 1.1531882286071777,
      "learning_rate": 9.726842105263158e-05,
      "loss": 0.4217,
      "step": 30394
    },
    {
      "epoch": 1.7020383021614962,
      "grad_norm": 1.6344202756881714,
      "learning_rate": 9.726815789473684e-05,
      "loss": 0.4903,
      "step": 30395
    },
    {
      "epoch": 1.7020942994736252,
      "grad_norm": 1.258263349533081,
      "learning_rate": 9.726789473684211e-05,
      "loss": 0.486,
      "step": 30396
    },
    {
      "epoch": 1.7021502967857542,
      "grad_norm": 1.3069782257080078,
      "learning_rate": 9.726763157894737e-05,
      "loss": 0.4587,
      "step": 30397
    },
    {
      "epoch": 1.7022062940978833,
      "grad_norm": 1.4569857120513916,
      "learning_rate": 9.726736842105265e-05,
      "loss": 0.4588,
      "step": 30398
    },
    {
      "epoch": 1.7022622914100123,
      "grad_norm": 1.7946926355361938,
      "learning_rate": 9.726710526315789e-05,
      "loss": 0.6473,
      "step": 30399
    },
    {
      "epoch": 1.7023182887221413,
      "grad_norm": 1.431936502456665,
      "learning_rate": 9.726684210526317e-05,
      "loss": 0.457,
      "step": 30400
    },
    {
      "epoch": 1.7023742860342703,
      "grad_norm": 1.1445733308792114,
      "learning_rate": 9.726657894736843e-05,
      "loss": 0.4208,
      "step": 30401
    },
    {
      "epoch": 1.7024302833463993,
      "grad_norm": 1.2694015502929688,
      "learning_rate": 9.72663157894737e-05,
      "loss": 0.3792,
      "step": 30402
    },
    {
      "epoch": 1.7024862806585284,
      "grad_norm": 1.0210872888565063,
      "learning_rate": 9.726605263157895e-05,
      "loss": 0.4022,
      "step": 30403
    },
    {
      "epoch": 1.7025422779706574,
      "grad_norm": 4.105430603027344,
      "learning_rate": 9.72657894736842e-05,
      "loss": 0.584,
      "step": 30404
    },
    {
      "epoch": 1.7025982752827864,
      "grad_norm": 1.0736076831817627,
      "learning_rate": 9.726552631578948e-05,
      "loss": 0.4297,
      "step": 30405
    },
    {
      "epoch": 1.7026542725949154,
      "grad_norm": 1.2866970300674438,
      "learning_rate": 9.726526315789474e-05,
      "loss": 0.4977,
      "step": 30406
    },
    {
      "epoch": 1.7027102699070444,
      "grad_norm": 1.7026922702789307,
      "learning_rate": 9.726500000000001e-05,
      "loss": 0.5121,
      "step": 30407
    },
    {
      "epoch": 1.7027662672191735,
      "grad_norm": 1.3127936124801636,
      "learning_rate": 9.726473684210526e-05,
      "loss": 0.4201,
      "step": 30408
    },
    {
      "epoch": 1.7028222645313025,
      "grad_norm": 1.456664800643921,
      "learning_rate": 9.726447368421053e-05,
      "loss": 0.6082,
      "step": 30409
    },
    {
      "epoch": 1.7028782618434315,
      "grad_norm": 1.3231877088546753,
      "learning_rate": 9.726421052631579e-05,
      "loss": 0.496,
      "step": 30410
    },
    {
      "epoch": 1.7029342591555605,
      "grad_norm": 1.4809296131134033,
      "learning_rate": 9.726394736842106e-05,
      "loss": 0.4325,
      "step": 30411
    },
    {
      "epoch": 1.7029902564676895,
      "grad_norm": 1.2596771717071533,
      "learning_rate": 9.726368421052632e-05,
      "loss": 0.4379,
      "step": 30412
    },
    {
      "epoch": 1.7030462537798186,
      "grad_norm": 1.4474269151687622,
      "learning_rate": 9.726342105263158e-05,
      "loss": 0.5362,
      "step": 30413
    },
    {
      "epoch": 1.7031022510919476,
      "grad_norm": 1.1639429330825806,
      "learning_rate": 9.726315789473684e-05,
      "loss": 0.353,
      "step": 30414
    },
    {
      "epoch": 1.7031582484040766,
      "grad_norm": 1.225325107574463,
      "learning_rate": 9.726289473684212e-05,
      "loss": 0.3527,
      "step": 30415
    },
    {
      "epoch": 1.7032142457162056,
      "grad_norm": 1.3113058805465698,
      "learning_rate": 9.726263157894738e-05,
      "loss": 0.5815,
      "step": 30416
    },
    {
      "epoch": 1.7032702430283346,
      "grad_norm": 1.1792465448379517,
      "learning_rate": 9.726236842105264e-05,
      "loss": 0.5102,
      "step": 30417
    },
    {
      "epoch": 1.7033262403404636,
      "grad_norm": 1.6666264533996582,
      "learning_rate": 9.72621052631579e-05,
      "loss": 0.5822,
      "step": 30418
    },
    {
      "epoch": 1.7033822376525927,
      "grad_norm": 1.060243010520935,
      "learning_rate": 9.726184210526317e-05,
      "loss": 0.408,
      "step": 30419
    },
    {
      "epoch": 1.7034382349647217,
      "grad_norm": 1.1443876028060913,
      "learning_rate": 9.726157894736843e-05,
      "loss": 0.3858,
      "step": 30420
    },
    {
      "epoch": 1.7034942322768507,
      "grad_norm": 1.060678482055664,
      "learning_rate": 9.726131578947369e-05,
      "loss": 0.399,
      "step": 30421
    },
    {
      "epoch": 1.7035502295889797,
      "grad_norm": 1.4833637475967407,
      "learning_rate": 9.726105263157895e-05,
      "loss": 0.466,
      "step": 30422
    },
    {
      "epoch": 1.7036062269011087,
      "grad_norm": 1.1668732166290283,
      "learning_rate": 9.726078947368421e-05,
      "loss": 0.3098,
      "step": 30423
    },
    {
      "epoch": 1.7036622242132378,
      "grad_norm": 1.498699426651001,
      "learning_rate": 9.726052631578948e-05,
      "loss": 0.543,
      "step": 30424
    },
    {
      "epoch": 1.7037182215253668,
      "grad_norm": 1.2083587646484375,
      "learning_rate": 9.726026315789474e-05,
      "loss": 0.3934,
      "step": 30425
    },
    {
      "epoch": 1.7037742188374958,
      "grad_norm": 1.2850645780563354,
      "learning_rate": 9.726e-05,
      "loss": 0.5364,
      "step": 30426
    },
    {
      "epoch": 1.7038302161496248,
      "grad_norm": 1.451698899269104,
      "learning_rate": 9.725973684210526e-05,
      "loss": 0.4178,
      "step": 30427
    },
    {
      "epoch": 1.7038862134617538,
      "grad_norm": 1.3936408758163452,
      "learning_rate": 9.725947368421053e-05,
      "loss": 0.4768,
      "step": 30428
    },
    {
      "epoch": 1.7039422107738829,
      "grad_norm": 1.2527719736099243,
      "learning_rate": 9.72592105263158e-05,
      "loss": 0.3909,
      "step": 30429
    },
    {
      "epoch": 1.7039982080860119,
      "grad_norm": 1.081490159034729,
      "learning_rate": 9.725894736842107e-05,
      "loss": 0.4346,
      "step": 30430
    },
    {
      "epoch": 1.704054205398141,
      "grad_norm": 1.1686931848526,
      "learning_rate": 9.725868421052631e-05,
      "loss": 0.4694,
      "step": 30431
    },
    {
      "epoch": 1.70411020271027,
      "grad_norm": 1.3607361316680908,
      "learning_rate": 9.725842105263159e-05,
      "loss": 0.4416,
      "step": 30432
    },
    {
      "epoch": 1.704166200022399,
      "grad_norm": 1.2696168422698975,
      "learning_rate": 9.725815789473685e-05,
      "loss": 0.4423,
      "step": 30433
    },
    {
      "epoch": 1.704222197334528,
      "grad_norm": 1.1625663042068481,
      "learning_rate": 9.725789473684212e-05,
      "loss": 0.3651,
      "step": 30434
    },
    {
      "epoch": 1.704278194646657,
      "grad_norm": 1.2001128196716309,
      "learning_rate": 9.725763157894738e-05,
      "loss": 0.3825,
      "step": 30435
    },
    {
      "epoch": 1.704334191958786,
      "grad_norm": 1.3356788158416748,
      "learning_rate": 9.725736842105264e-05,
      "loss": 0.4432,
      "step": 30436
    },
    {
      "epoch": 1.704390189270915,
      "grad_norm": 1.427314043045044,
      "learning_rate": 9.72571052631579e-05,
      "loss": 0.4516,
      "step": 30437
    },
    {
      "epoch": 1.704446186583044,
      "grad_norm": 1.3588881492614746,
      "learning_rate": 9.725684210526316e-05,
      "loss": 0.5084,
      "step": 30438
    },
    {
      "epoch": 1.704502183895173,
      "grad_norm": 1.1967135667800903,
      "learning_rate": 9.725657894736843e-05,
      "loss": 0.3756,
      "step": 30439
    },
    {
      "epoch": 1.704558181207302,
      "grad_norm": 1.4664340019226074,
      "learning_rate": 9.725631578947369e-05,
      "loss": 0.5995,
      "step": 30440
    },
    {
      "epoch": 1.704614178519431,
      "grad_norm": 1.5603938102722168,
      "learning_rate": 9.725605263157895e-05,
      "loss": 0.6902,
      "step": 30441
    },
    {
      "epoch": 1.70467017583156,
      "grad_norm": 1.2841347455978394,
      "learning_rate": 9.725578947368421e-05,
      "loss": 0.3792,
      "step": 30442
    },
    {
      "epoch": 1.7047261731436891,
      "grad_norm": 1.312217354774475,
      "learning_rate": 9.725552631578948e-05,
      "loss": 0.4777,
      "step": 30443
    },
    {
      "epoch": 1.7047821704558181,
      "grad_norm": 1.2578442096710205,
      "learning_rate": 9.725526315789474e-05,
      "loss": 0.4364,
      "step": 30444
    },
    {
      "epoch": 1.7048381677679472,
      "grad_norm": 1.2755889892578125,
      "learning_rate": 9.7255e-05,
      "loss": 0.4185,
      "step": 30445
    },
    {
      "epoch": 1.7048941650800762,
      "grad_norm": 1.2359391450881958,
      "learning_rate": 9.725473684210526e-05,
      "loss": 0.4579,
      "step": 30446
    },
    {
      "epoch": 1.7049501623922052,
      "grad_norm": 1.344716191291809,
      "learning_rate": 9.725447368421054e-05,
      "loss": 0.5177,
      "step": 30447
    },
    {
      "epoch": 1.7050061597043342,
      "grad_norm": 1.423856258392334,
      "learning_rate": 9.72542105263158e-05,
      "loss": 0.5039,
      "step": 30448
    },
    {
      "epoch": 1.7050621570164632,
      "grad_norm": 1.1566288471221924,
      "learning_rate": 9.725394736842106e-05,
      "loss": 0.4561,
      "step": 30449
    },
    {
      "epoch": 1.7051181543285923,
      "grad_norm": 1.3783998489379883,
      "learning_rate": 9.725368421052632e-05,
      "loss": 0.4719,
      "step": 30450
    },
    {
      "epoch": 1.7051741516407213,
      "grad_norm": 1.2949869632720947,
      "learning_rate": 9.725342105263159e-05,
      "loss": 0.613,
      "step": 30451
    },
    {
      "epoch": 1.7052301489528503,
      "grad_norm": 1.324256181716919,
      "learning_rate": 9.725315789473685e-05,
      "loss": 0.3451,
      "step": 30452
    },
    {
      "epoch": 1.7052861462649793,
      "grad_norm": 1.0943481922149658,
      "learning_rate": 9.725289473684212e-05,
      "loss": 0.3402,
      "step": 30453
    },
    {
      "epoch": 1.7053421435771083,
      "grad_norm": 1.1040737628936768,
      "learning_rate": 9.725263157894737e-05,
      "loss": 0.4539,
      "step": 30454
    },
    {
      "epoch": 1.7053981408892374,
      "grad_norm": 1.1457338333129883,
      "learning_rate": 9.725236842105263e-05,
      "loss": 0.4564,
      "step": 30455
    },
    {
      "epoch": 1.7054541382013664,
      "grad_norm": 1.266891360282898,
      "learning_rate": 9.72521052631579e-05,
      "loss": 0.3382,
      "step": 30456
    },
    {
      "epoch": 1.7055101355134954,
      "grad_norm": 1.2966458797454834,
      "learning_rate": 9.725184210526316e-05,
      "loss": 0.4611,
      "step": 30457
    },
    {
      "epoch": 1.7055661328256244,
      "grad_norm": 1.471893548965454,
      "learning_rate": 9.725157894736842e-05,
      "loss": 0.4955,
      "step": 30458
    },
    {
      "epoch": 1.7056221301377534,
      "grad_norm": 1.342434048652649,
      "learning_rate": 9.725131578947368e-05,
      "loss": 0.4862,
      "step": 30459
    },
    {
      "epoch": 1.7056781274498825,
      "grad_norm": 1.1808223724365234,
      "learning_rate": 9.725105263157895e-05,
      "loss": 0.3575,
      "step": 30460
    },
    {
      "epoch": 1.7057341247620115,
      "grad_norm": 1.2905265092849731,
      "learning_rate": 9.725078947368421e-05,
      "loss": 0.5352,
      "step": 30461
    },
    {
      "epoch": 1.7057901220741405,
      "grad_norm": 1.2809385061264038,
      "learning_rate": 9.725052631578949e-05,
      "loss": 0.5049,
      "step": 30462
    },
    {
      "epoch": 1.7058461193862695,
      "grad_norm": 1.2980560064315796,
      "learning_rate": 9.725026315789473e-05,
      "loss": 0.3592,
      "step": 30463
    },
    {
      "epoch": 1.7059021166983985,
      "grad_norm": 1.2398675680160522,
      "learning_rate": 9.725e-05,
      "loss": 0.4474,
      "step": 30464
    },
    {
      "epoch": 1.7059581140105275,
      "grad_norm": 1.3723958730697632,
      "learning_rate": 9.724973684210527e-05,
      "loss": 0.6417,
      "step": 30465
    },
    {
      "epoch": 1.7060141113226566,
      "grad_norm": 1.1518769264221191,
      "learning_rate": 9.724947368421054e-05,
      "loss": 0.3915,
      "step": 30466
    },
    {
      "epoch": 1.7060701086347856,
      "grad_norm": 1.6506885290145874,
      "learning_rate": 9.72492105263158e-05,
      "loss": 0.4685,
      "step": 30467
    },
    {
      "epoch": 1.7061261059469146,
      "grad_norm": 1.1885184049606323,
      "learning_rate": 9.724894736842106e-05,
      "loss": 0.4079,
      "step": 30468
    },
    {
      "epoch": 1.7061821032590436,
      "grad_norm": 1.6102256774902344,
      "learning_rate": 9.724868421052632e-05,
      "loss": 0.4723,
      "step": 30469
    },
    {
      "epoch": 1.7062381005711726,
      "grad_norm": 1.0847893953323364,
      "learning_rate": 9.724842105263159e-05,
      "loss": 0.4572,
      "step": 30470
    },
    {
      "epoch": 1.7062940978833017,
      "grad_norm": 1.1210451126098633,
      "learning_rate": 9.724815789473685e-05,
      "loss": 0.3528,
      "step": 30471
    },
    {
      "epoch": 1.7063500951954307,
      "grad_norm": 1.3248612880706787,
      "learning_rate": 9.724789473684211e-05,
      "loss": 0.5108,
      "step": 30472
    },
    {
      "epoch": 1.7064060925075597,
      "grad_norm": 1.2521291971206665,
      "learning_rate": 9.724763157894737e-05,
      "loss": 0.4766,
      "step": 30473
    },
    {
      "epoch": 1.7064620898196887,
      "grad_norm": 1.1710253953933716,
      "learning_rate": 9.724736842105263e-05,
      "loss": 0.5097,
      "step": 30474
    },
    {
      "epoch": 1.7065180871318177,
      "grad_norm": 1.3462587594985962,
      "learning_rate": 9.72471052631579e-05,
      "loss": 0.5612,
      "step": 30475
    },
    {
      "epoch": 1.7065740844439468,
      "grad_norm": 1.2117505073547363,
      "learning_rate": 9.724684210526316e-05,
      "loss": 0.4326,
      "step": 30476
    },
    {
      "epoch": 1.7066300817560758,
      "grad_norm": 1.5765436887741089,
      "learning_rate": 9.724657894736842e-05,
      "loss": 0.7118,
      "step": 30477
    },
    {
      "epoch": 1.7066860790682048,
      "grad_norm": 1.361477017402649,
      "learning_rate": 9.724631578947368e-05,
      "loss": 0.3543,
      "step": 30478
    },
    {
      "epoch": 1.7067420763803338,
      "grad_norm": 1.6338622570037842,
      "learning_rate": 9.724605263157896e-05,
      "loss": 0.528,
      "step": 30479
    },
    {
      "epoch": 1.7067980736924628,
      "grad_norm": 1.4930986166000366,
      "learning_rate": 9.724578947368422e-05,
      "loss": 0.5079,
      "step": 30480
    },
    {
      "epoch": 1.7068540710045919,
      "grad_norm": 1.363598108291626,
      "learning_rate": 9.724552631578948e-05,
      "loss": 0.4515,
      "step": 30481
    },
    {
      "epoch": 1.7069100683167209,
      "grad_norm": 1.3108935356140137,
      "learning_rate": 9.724526315789474e-05,
      "loss": 0.5058,
      "step": 30482
    },
    {
      "epoch": 1.70696606562885,
      "grad_norm": 1.2128300666809082,
      "learning_rate": 9.724500000000001e-05,
      "loss": 0.5126,
      "step": 30483
    },
    {
      "epoch": 1.707022062940979,
      "grad_norm": 1.2338123321533203,
      "learning_rate": 9.724473684210527e-05,
      "loss": 0.3588,
      "step": 30484
    },
    {
      "epoch": 1.707078060253108,
      "grad_norm": 1.3296399116516113,
      "learning_rate": 9.724447368421054e-05,
      "loss": 0.4617,
      "step": 30485
    },
    {
      "epoch": 1.707134057565237,
      "grad_norm": 1.0531251430511475,
      "learning_rate": 9.724421052631579e-05,
      "loss": 0.3619,
      "step": 30486
    },
    {
      "epoch": 1.707190054877366,
      "grad_norm": 1.1226277351379395,
      "learning_rate": 9.724394736842106e-05,
      "loss": 0.3713,
      "step": 30487
    },
    {
      "epoch": 1.707246052189495,
      "grad_norm": 1.1927179098129272,
      "learning_rate": 9.724368421052632e-05,
      "loss": 0.3629,
      "step": 30488
    },
    {
      "epoch": 1.707302049501624,
      "grad_norm": 1.546406865119934,
      "learning_rate": 9.724342105263158e-05,
      "loss": 0.6288,
      "step": 30489
    },
    {
      "epoch": 1.707358046813753,
      "grad_norm": 1.3373161554336548,
      "learning_rate": 9.724315789473685e-05,
      "loss": 0.3687,
      "step": 30490
    },
    {
      "epoch": 1.707414044125882,
      "grad_norm": 1.1055089235305786,
      "learning_rate": 9.72428947368421e-05,
      "loss": 0.4005,
      "step": 30491
    },
    {
      "epoch": 1.707470041438011,
      "grad_norm": 1.1816829442977905,
      "learning_rate": 9.724263157894737e-05,
      "loss": 0.4471,
      "step": 30492
    },
    {
      "epoch": 1.70752603875014,
      "grad_norm": 1.4974868297576904,
      "learning_rate": 9.724236842105263e-05,
      "loss": 0.4795,
      "step": 30493
    },
    {
      "epoch": 1.707582036062269,
      "grad_norm": 1.2631924152374268,
      "learning_rate": 9.72421052631579e-05,
      "loss": 0.4412,
      "step": 30494
    },
    {
      "epoch": 1.7076380333743981,
      "grad_norm": 1.2478843927383423,
      "learning_rate": 9.724184210526317e-05,
      "loss": 0.5724,
      "step": 30495
    },
    {
      "epoch": 1.7076940306865271,
      "grad_norm": 1.5070134401321411,
      "learning_rate": 9.724157894736843e-05,
      "loss": 0.5999,
      "step": 30496
    },
    {
      "epoch": 1.7077500279986562,
      "grad_norm": 1.4186972379684448,
      "learning_rate": 9.724131578947369e-05,
      "loss": 0.4678,
      "step": 30497
    },
    {
      "epoch": 1.7078060253107852,
      "grad_norm": 1.4076356887817383,
      "learning_rate": 9.724105263157896e-05,
      "loss": 0.471,
      "step": 30498
    },
    {
      "epoch": 1.7078620226229142,
      "grad_norm": 1.4378857612609863,
      "learning_rate": 9.724078947368422e-05,
      "loss": 0.4323,
      "step": 30499
    },
    {
      "epoch": 1.7079180199350432,
      "grad_norm": 1.23332941532135,
      "learning_rate": 9.724052631578948e-05,
      "loss": 0.5648,
      "step": 30500
    },
    {
      "epoch": 1.7079740172471722,
      "grad_norm": 1.2035006284713745,
      "learning_rate": 9.724026315789474e-05,
      "loss": 0.4644,
      "step": 30501
    },
    {
      "epoch": 1.7080300145593013,
      "grad_norm": 1.1639639139175415,
      "learning_rate": 9.724000000000001e-05,
      "loss": 0.4612,
      "step": 30502
    },
    {
      "epoch": 1.7080860118714303,
      "grad_norm": 1.1637418270111084,
      "learning_rate": 9.723973684210527e-05,
      "loss": 0.3284,
      "step": 30503
    },
    {
      "epoch": 1.7081420091835593,
      "grad_norm": 1.4097344875335693,
      "learning_rate": 9.723947368421053e-05,
      "loss": 0.5028,
      "step": 30504
    },
    {
      "epoch": 1.7081980064956883,
      "grad_norm": 1.4470453262329102,
      "learning_rate": 9.723921052631579e-05,
      "loss": 0.4567,
      "step": 30505
    },
    {
      "epoch": 1.7082540038078173,
      "grad_norm": 1.3159265518188477,
      "learning_rate": 9.723894736842106e-05,
      "loss": 0.4034,
      "step": 30506
    },
    {
      "epoch": 1.7083100011199464,
      "grad_norm": 2.2175371646881104,
      "learning_rate": 9.723868421052632e-05,
      "loss": 0.4138,
      "step": 30507
    },
    {
      "epoch": 1.7083659984320754,
      "grad_norm": 1.3051400184631348,
      "learning_rate": 9.723842105263158e-05,
      "loss": 0.4557,
      "step": 30508
    },
    {
      "epoch": 1.7084219957442044,
      "grad_norm": 1.4340587854385376,
      "learning_rate": 9.723815789473684e-05,
      "loss": 0.3551,
      "step": 30509
    },
    {
      "epoch": 1.7084779930563334,
      "grad_norm": 1.2266170978546143,
      "learning_rate": 9.72378947368421e-05,
      "loss": 0.4332,
      "step": 30510
    },
    {
      "epoch": 1.7085339903684624,
      "grad_norm": 1.3109291791915894,
      "learning_rate": 9.723763157894738e-05,
      "loss": 0.41,
      "step": 30511
    },
    {
      "epoch": 1.7085899876805914,
      "grad_norm": 1.437774658203125,
      "learning_rate": 9.723736842105264e-05,
      "loss": 0.4451,
      "step": 30512
    },
    {
      "epoch": 1.7086459849927205,
      "grad_norm": 1.1644784212112427,
      "learning_rate": 9.72371052631579e-05,
      "loss": 0.2779,
      "step": 30513
    },
    {
      "epoch": 1.7087019823048495,
      "grad_norm": 1.1141005754470825,
      "learning_rate": 9.723684210526315e-05,
      "loss": 0.4475,
      "step": 30514
    },
    {
      "epoch": 1.7087579796169785,
      "grad_norm": 1.309321403503418,
      "learning_rate": 9.723657894736843e-05,
      "loss": 0.3756,
      "step": 30515
    },
    {
      "epoch": 1.7088139769291075,
      "grad_norm": 1.527395248413086,
      "learning_rate": 9.723631578947369e-05,
      "loss": 0.5072,
      "step": 30516
    },
    {
      "epoch": 1.7088699742412365,
      "grad_norm": 1.4535458087921143,
      "learning_rate": 9.723605263157896e-05,
      "loss": 0.408,
      "step": 30517
    },
    {
      "epoch": 1.7089259715533656,
      "grad_norm": 1.2554166316986084,
      "learning_rate": 9.723578947368421e-05,
      "loss": 0.5063,
      "step": 30518
    },
    {
      "epoch": 1.7089819688654946,
      "grad_norm": 1.2566369771957397,
      "learning_rate": 9.723552631578948e-05,
      "loss": 0.4132,
      "step": 30519
    },
    {
      "epoch": 1.7090379661776236,
      "grad_norm": 1.10805344581604,
      "learning_rate": 9.723526315789474e-05,
      "loss": 0.4375,
      "step": 30520
    },
    {
      "epoch": 1.7090939634897526,
      "grad_norm": 1.84591543674469,
      "learning_rate": 9.723500000000001e-05,
      "loss": 0.5896,
      "step": 30521
    },
    {
      "epoch": 1.7091499608018816,
      "grad_norm": 1.3341203927993774,
      "learning_rate": 9.723473684210527e-05,
      "loss": 0.3631,
      "step": 30522
    },
    {
      "epoch": 1.7092059581140107,
      "grad_norm": 1.2211353778839111,
      "learning_rate": 9.723447368421053e-05,
      "loss": 0.4661,
      "step": 30523
    },
    {
      "epoch": 1.7092619554261397,
      "grad_norm": 1.682071566581726,
      "learning_rate": 9.723421052631579e-05,
      "loss": 0.5394,
      "step": 30524
    },
    {
      "epoch": 1.7093179527382687,
      "grad_norm": 1.1894397735595703,
      "learning_rate": 9.723394736842105e-05,
      "loss": 0.3991,
      "step": 30525
    },
    {
      "epoch": 1.7093739500503977,
      "grad_norm": 1.1776165962219238,
      "learning_rate": 9.723368421052633e-05,
      "loss": 0.3585,
      "step": 30526
    },
    {
      "epoch": 1.7094299473625267,
      "grad_norm": 1.2605681419372559,
      "learning_rate": 9.723342105263159e-05,
      "loss": 0.485,
      "step": 30527
    },
    {
      "epoch": 1.7094859446746558,
      "grad_norm": 1.5534437894821167,
      "learning_rate": 9.723315789473685e-05,
      "loss": 0.5367,
      "step": 30528
    },
    {
      "epoch": 1.7095419419867848,
      "grad_norm": 1.427203893661499,
      "learning_rate": 9.72328947368421e-05,
      "loss": 0.4518,
      "step": 30529
    },
    {
      "epoch": 1.7095979392989138,
      "grad_norm": 1.098892092704773,
      "learning_rate": 9.723263157894738e-05,
      "loss": 0.3718,
      "step": 30530
    },
    {
      "epoch": 1.7096539366110428,
      "grad_norm": 1.3872413635253906,
      "learning_rate": 9.723236842105264e-05,
      "loss": 0.4475,
      "step": 30531
    },
    {
      "epoch": 1.7097099339231718,
      "grad_norm": 1.3392170667648315,
      "learning_rate": 9.72321052631579e-05,
      "loss": 0.5314,
      "step": 30532
    },
    {
      "epoch": 1.7097659312353009,
      "grad_norm": 1.3937572240829468,
      "learning_rate": 9.723184210526316e-05,
      "loss": 0.4211,
      "step": 30533
    },
    {
      "epoch": 1.7098219285474299,
      "grad_norm": 1.1660832166671753,
      "learning_rate": 9.723157894736843e-05,
      "loss": 0.4424,
      "step": 30534
    },
    {
      "epoch": 1.709877925859559,
      "grad_norm": 1.8650375604629517,
      "learning_rate": 9.723131578947369e-05,
      "loss": 0.4601,
      "step": 30535
    },
    {
      "epoch": 1.709933923171688,
      "grad_norm": 1.4749939441680908,
      "learning_rate": 9.723105263157895e-05,
      "loss": 0.4636,
      "step": 30536
    },
    {
      "epoch": 1.709989920483817,
      "grad_norm": 1.7622720003128052,
      "learning_rate": 9.723078947368421e-05,
      "loss": 0.5261,
      "step": 30537
    },
    {
      "epoch": 1.710045917795946,
      "grad_norm": 1.921120047569275,
      "learning_rate": 9.723052631578948e-05,
      "loss": 0.4489,
      "step": 30538
    },
    {
      "epoch": 1.710101915108075,
      "grad_norm": 1.7017356157302856,
      "learning_rate": 9.723026315789474e-05,
      "loss": 0.4543,
      "step": 30539
    },
    {
      "epoch": 1.710157912420204,
      "grad_norm": 1.1812225580215454,
      "learning_rate": 9.723000000000002e-05,
      "loss": 0.3206,
      "step": 30540
    },
    {
      "epoch": 1.710213909732333,
      "grad_norm": 1.267765760421753,
      "learning_rate": 9.722973684210526e-05,
      "loss": 0.4081,
      "step": 30541
    },
    {
      "epoch": 1.710269907044462,
      "grad_norm": 5.9477667808532715,
      "learning_rate": 9.722947368421052e-05,
      "loss": 0.4174,
      "step": 30542
    },
    {
      "epoch": 1.710325904356591,
      "grad_norm": 1.1593431234359741,
      "learning_rate": 9.72292105263158e-05,
      "loss": 0.4871,
      "step": 30543
    },
    {
      "epoch": 1.71038190166872,
      "grad_norm": 1.4129647016525269,
      "learning_rate": 9.722894736842105e-05,
      "loss": 0.5077,
      "step": 30544
    },
    {
      "epoch": 1.710437898980849,
      "grad_norm": 1.1454287767410278,
      "learning_rate": 9.722868421052633e-05,
      "loss": 0.596,
      "step": 30545
    },
    {
      "epoch": 1.710493896292978,
      "grad_norm": 1.4092313051223755,
      "learning_rate": 9.722842105263157e-05,
      "loss": 0.5816,
      "step": 30546
    },
    {
      "epoch": 1.7105498936051071,
      "grad_norm": 1.3189082145690918,
      "learning_rate": 9.722815789473685e-05,
      "loss": 0.4882,
      "step": 30547
    },
    {
      "epoch": 1.7106058909172361,
      "grad_norm": 1.5342934131622314,
      "learning_rate": 9.722789473684211e-05,
      "loss": 0.3852,
      "step": 30548
    },
    {
      "epoch": 1.710661888229365,
      "grad_norm": 1.220430850982666,
      "learning_rate": 9.722763157894738e-05,
      "loss": 0.5128,
      "step": 30549
    },
    {
      "epoch": 1.710717885541494,
      "grad_norm": 1.8775285482406616,
      "learning_rate": 9.722736842105263e-05,
      "loss": 0.5091,
      "step": 30550
    },
    {
      "epoch": 1.710773882853623,
      "grad_norm": 1.3726714849472046,
      "learning_rate": 9.72271052631579e-05,
      "loss": 0.5295,
      "step": 30551
    },
    {
      "epoch": 1.710829880165752,
      "grad_norm": 1.4516642093658447,
      "learning_rate": 9.722684210526316e-05,
      "loss": 0.4134,
      "step": 30552
    },
    {
      "epoch": 1.710885877477881,
      "grad_norm": 1.3525302410125732,
      "learning_rate": 9.722657894736843e-05,
      "loss": 0.4878,
      "step": 30553
    },
    {
      "epoch": 1.71094187479001,
      "grad_norm": 1.3447599411010742,
      "learning_rate": 9.722631578947369e-05,
      "loss": 0.565,
      "step": 30554
    },
    {
      "epoch": 1.710997872102139,
      "grad_norm": 1.1941272020339966,
      "learning_rate": 9.722605263157895e-05,
      "loss": 0.3563,
      "step": 30555
    },
    {
      "epoch": 1.711053869414268,
      "grad_norm": 1.3645598888397217,
      "learning_rate": 9.722578947368421e-05,
      "loss": 0.5775,
      "step": 30556
    },
    {
      "epoch": 1.711109866726397,
      "grad_norm": 1.2135733366012573,
      "learning_rate": 9.722552631578949e-05,
      "loss": 0.4698,
      "step": 30557
    },
    {
      "epoch": 1.711165864038526,
      "grad_norm": 1.188438057899475,
      "learning_rate": 9.722526315789475e-05,
      "loss": 0.3848,
      "step": 30558
    },
    {
      "epoch": 1.7112218613506551,
      "grad_norm": 1.270774006843567,
      "learning_rate": 9.7225e-05,
      "loss": 0.4707,
      "step": 30559
    },
    {
      "epoch": 1.7112778586627841,
      "grad_norm": 1.282051920890808,
      "learning_rate": 9.722473684210526e-05,
      "loss": 0.3673,
      "step": 30560
    },
    {
      "epoch": 1.7113338559749132,
      "grad_norm": 1.2208714485168457,
      "learning_rate": 9.722447368421052e-05,
      "loss": 0.406,
      "step": 30561
    },
    {
      "epoch": 1.7113898532870422,
      "grad_norm": 1.078134536743164,
      "learning_rate": 9.72242105263158e-05,
      "loss": 0.4927,
      "step": 30562
    },
    {
      "epoch": 1.7114458505991712,
      "grad_norm": 7.428740978240967,
      "learning_rate": 9.722394736842106e-05,
      "loss": 0.4742,
      "step": 30563
    },
    {
      "epoch": 1.7115018479113002,
      "grad_norm": 1.3124948740005493,
      "learning_rate": 9.722368421052632e-05,
      "loss": 0.4801,
      "step": 30564
    },
    {
      "epoch": 1.7115578452234292,
      "grad_norm": 1.4366769790649414,
      "learning_rate": 9.722342105263158e-05,
      "loss": 0.4579,
      "step": 30565
    },
    {
      "epoch": 1.7116138425355583,
      "grad_norm": 1.1832019090652466,
      "learning_rate": 9.722315789473685e-05,
      "loss": 0.4563,
      "step": 30566
    },
    {
      "epoch": 1.7116698398476873,
      "grad_norm": 1.219072699546814,
      "learning_rate": 9.722289473684211e-05,
      "loss": 0.3572,
      "step": 30567
    },
    {
      "epoch": 1.7117258371598163,
      "grad_norm": 1.352976679801941,
      "learning_rate": 9.722263157894737e-05,
      "loss": 0.4948,
      "step": 30568
    },
    {
      "epoch": 1.7117818344719453,
      "grad_norm": 2.1166634559631348,
      "learning_rate": 9.722236842105263e-05,
      "loss": 0.3773,
      "step": 30569
    },
    {
      "epoch": 1.7118378317840743,
      "grad_norm": 1.0951555967330933,
      "learning_rate": 9.72221052631579e-05,
      "loss": 0.4427,
      "step": 30570
    },
    {
      "epoch": 1.7118938290962034,
      "grad_norm": 1.319106936454773,
      "learning_rate": 9.722184210526316e-05,
      "loss": 0.4357,
      "step": 30571
    },
    {
      "epoch": 1.7119498264083324,
      "grad_norm": 1.6315269470214844,
      "learning_rate": 9.722157894736844e-05,
      "loss": 0.4057,
      "step": 30572
    },
    {
      "epoch": 1.7120058237204614,
      "grad_norm": 1.246069073677063,
      "learning_rate": 9.722131578947368e-05,
      "loss": 0.3893,
      "step": 30573
    },
    {
      "epoch": 1.7120618210325904,
      "grad_norm": 1.2473217248916626,
      "learning_rate": 9.722105263157896e-05,
      "loss": 0.4203,
      "step": 30574
    },
    {
      "epoch": 1.7121178183447194,
      "grad_norm": 1.3081737756729126,
      "learning_rate": 9.722078947368421e-05,
      "loss": 0.4838,
      "step": 30575
    },
    {
      "epoch": 1.7121738156568485,
      "grad_norm": 1.5456491708755493,
      "learning_rate": 9.722052631578947e-05,
      "loss": 0.4271,
      "step": 30576
    },
    {
      "epoch": 1.7122298129689775,
      "grad_norm": 1.4192774295806885,
      "learning_rate": 9.722026315789475e-05,
      "loss": 0.4473,
      "step": 30577
    },
    {
      "epoch": 1.7122858102811065,
      "grad_norm": 1.259068489074707,
      "learning_rate": 9.722e-05,
      "loss": 0.4821,
      "step": 30578
    },
    {
      "epoch": 1.7123418075932355,
      "grad_norm": 1.2192338705062866,
      "learning_rate": 9.721973684210527e-05,
      "loss": 0.4147,
      "step": 30579
    },
    {
      "epoch": 1.7123978049053645,
      "grad_norm": 1.3094992637634277,
      "learning_rate": 9.721947368421053e-05,
      "loss": 0.4099,
      "step": 30580
    },
    {
      "epoch": 1.7124538022174935,
      "grad_norm": 1.2361643314361572,
      "learning_rate": 9.72192105263158e-05,
      "loss": 0.4245,
      "step": 30581
    },
    {
      "epoch": 1.7125097995296226,
      "grad_norm": 1.2182124853134155,
      "learning_rate": 9.721894736842106e-05,
      "loss": 0.5122,
      "step": 30582
    },
    {
      "epoch": 1.7125657968417516,
      "grad_norm": 1.3774763345718384,
      "learning_rate": 9.721868421052632e-05,
      "loss": 0.4316,
      "step": 30583
    },
    {
      "epoch": 1.7126217941538806,
      "grad_norm": 1.2150533199310303,
      "learning_rate": 9.721842105263158e-05,
      "loss": 0.4453,
      "step": 30584
    },
    {
      "epoch": 1.7126777914660096,
      "grad_norm": 1.3658932447433472,
      "learning_rate": 9.721815789473685e-05,
      "loss": 0.5215,
      "step": 30585
    },
    {
      "epoch": 1.7127337887781386,
      "grad_norm": 1.7840657234191895,
      "learning_rate": 9.721789473684211e-05,
      "loss": 0.4515,
      "step": 30586
    },
    {
      "epoch": 1.7127897860902677,
      "grad_norm": 12.474908828735352,
      "learning_rate": 9.721763157894737e-05,
      "loss": 0.397,
      "step": 30587
    },
    {
      "epoch": 1.7128457834023967,
      "grad_norm": 1.279429316520691,
      "learning_rate": 9.721736842105263e-05,
      "loss": 0.4489,
      "step": 30588
    },
    {
      "epoch": 1.7129017807145257,
      "grad_norm": 1.2631796598434448,
      "learning_rate": 9.72171052631579e-05,
      "loss": 0.4396,
      "step": 30589
    },
    {
      "epoch": 1.7129577780266547,
      "grad_norm": 1.2091093063354492,
      "learning_rate": 9.721684210526317e-05,
      "loss": 0.4535,
      "step": 30590
    },
    {
      "epoch": 1.7130137753387837,
      "grad_norm": 1.2190179824829102,
      "learning_rate": 9.721657894736842e-05,
      "loss": 0.3891,
      "step": 30591
    },
    {
      "epoch": 1.7130697726509128,
      "grad_norm": 4.941834926605225,
      "learning_rate": 9.721631578947368e-05,
      "loss": 0.5027,
      "step": 30592
    },
    {
      "epoch": 1.7131257699630418,
      "grad_norm": 1.272477388381958,
      "learning_rate": 9.721605263157894e-05,
      "loss": 0.4147,
      "step": 30593
    },
    {
      "epoch": 1.7131817672751708,
      "grad_norm": 1.1352440118789673,
      "learning_rate": 9.721578947368422e-05,
      "loss": 0.4997,
      "step": 30594
    },
    {
      "epoch": 1.7132377645872998,
      "grad_norm": 1.238242506980896,
      "learning_rate": 9.721552631578948e-05,
      "loss": 0.5159,
      "step": 30595
    },
    {
      "epoch": 1.7132937618994288,
      "grad_norm": 1.2218987941741943,
      "learning_rate": 9.721526315789474e-05,
      "loss": 0.5147,
      "step": 30596
    },
    {
      "epoch": 1.7133497592115579,
      "grad_norm": 1.0865470170974731,
      "learning_rate": 9.7215e-05,
      "loss": 0.3208,
      "step": 30597
    },
    {
      "epoch": 1.7134057565236869,
      "grad_norm": 1.231279969215393,
      "learning_rate": 9.721473684210527e-05,
      "loss": 0.5322,
      "step": 30598
    },
    {
      "epoch": 1.713461753835816,
      "grad_norm": 1.792553186416626,
      "learning_rate": 9.721447368421053e-05,
      "loss": 0.6053,
      "step": 30599
    },
    {
      "epoch": 1.713517751147945,
      "grad_norm": 16.398744583129883,
      "learning_rate": 9.72142105263158e-05,
      "loss": 0.6179,
      "step": 30600
    },
    {
      "epoch": 1.713573748460074,
      "grad_norm": 1.4515624046325684,
      "learning_rate": 9.721394736842105e-05,
      "loss": 0.4621,
      "step": 30601
    },
    {
      "epoch": 1.713629745772203,
      "grad_norm": 1.3135499954223633,
      "learning_rate": 9.721368421052632e-05,
      "loss": 0.4713,
      "step": 30602
    },
    {
      "epoch": 1.713685743084332,
      "grad_norm": 1.1159043312072754,
      "learning_rate": 9.721342105263158e-05,
      "loss": 0.3756,
      "step": 30603
    },
    {
      "epoch": 1.713741740396461,
      "grad_norm": 1.162896752357483,
      "learning_rate": 9.721315789473686e-05,
      "loss": 0.3915,
      "step": 30604
    },
    {
      "epoch": 1.71379773770859,
      "grad_norm": 1.2234514951705933,
      "learning_rate": 9.72128947368421e-05,
      "loss": 0.426,
      "step": 30605
    },
    {
      "epoch": 1.713853735020719,
      "grad_norm": 1.1834098100662231,
      "learning_rate": 9.721263157894737e-05,
      "loss": 0.4264,
      "step": 30606
    },
    {
      "epoch": 1.713909732332848,
      "grad_norm": 1.1783487796783447,
      "learning_rate": 9.721236842105263e-05,
      "loss": 0.3993,
      "step": 30607
    },
    {
      "epoch": 1.713965729644977,
      "grad_norm": 1.6386569738388062,
      "learning_rate": 9.721210526315791e-05,
      "loss": 0.5254,
      "step": 30608
    },
    {
      "epoch": 1.714021726957106,
      "grad_norm": 1.2830673456192017,
      "learning_rate": 9.721184210526317e-05,
      "loss": 0.3077,
      "step": 30609
    },
    {
      "epoch": 1.714077724269235,
      "grad_norm": 1.170202612876892,
      "learning_rate": 9.721157894736841e-05,
      "loss": 0.3864,
      "step": 30610
    },
    {
      "epoch": 1.7141337215813641,
      "grad_norm": 2.121738910675049,
      "learning_rate": 9.721131578947369e-05,
      "loss": 0.5497,
      "step": 30611
    },
    {
      "epoch": 1.7141897188934931,
      "grad_norm": 1.3131927251815796,
      "learning_rate": 9.721105263157895e-05,
      "loss": 0.461,
      "step": 30612
    },
    {
      "epoch": 1.7142457162056222,
      "grad_norm": 1.1239851713180542,
      "learning_rate": 9.721078947368422e-05,
      "loss": 0.4226,
      "step": 30613
    },
    {
      "epoch": 1.7143017135177512,
      "grad_norm": 1.1671003103256226,
      "learning_rate": 9.721052631578948e-05,
      "loss": 0.5343,
      "step": 30614
    },
    {
      "epoch": 1.7143577108298802,
      "grad_norm": 1.367677927017212,
      "learning_rate": 9.721026315789474e-05,
      "loss": 0.6234,
      "step": 30615
    },
    {
      "epoch": 1.7144137081420092,
      "grad_norm": 1.6681021451950073,
      "learning_rate": 9.721e-05,
      "loss": 0.3351,
      "step": 30616
    },
    {
      "epoch": 1.7144697054541382,
      "grad_norm": 1.8264193534851074,
      "learning_rate": 9.720973684210527e-05,
      "loss": 0.5302,
      "step": 30617
    },
    {
      "epoch": 1.7145257027662673,
      "grad_norm": 1.5470703840255737,
      "learning_rate": 9.720947368421053e-05,
      "loss": 0.594,
      "step": 30618
    },
    {
      "epoch": 1.7145817000783963,
      "grad_norm": 1.2262579202651978,
      "learning_rate": 9.720921052631579e-05,
      "loss": 0.3708,
      "step": 30619
    },
    {
      "epoch": 1.7146376973905253,
      "grad_norm": 1.1791127920150757,
      "learning_rate": 9.720894736842105e-05,
      "loss": 0.4498,
      "step": 30620
    },
    {
      "epoch": 1.7146936947026543,
      "grad_norm": 1.3748281002044678,
      "learning_rate": 9.720868421052633e-05,
      "loss": 0.4342,
      "step": 30621
    },
    {
      "epoch": 1.7147496920147833,
      "grad_norm": 1.1307182312011719,
      "learning_rate": 9.720842105263158e-05,
      "loss": 0.4498,
      "step": 30622
    },
    {
      "epoch": 1.7148056893269124,
      "grad_norm": 1.339543104171753,
      "learning_rate": 9.720815789473684e-05,
      "loss": 0.5279,
      "step": 30623
    },
    {
      "epoch": 1.7148616866390414,
      "grad_norm": 1.3946506977081299,
      "learning_rate": 9.72078947368421e-05,
      "loss": 0.5304,
      "step": 30624
    },
    {
      "epoch": 1.7149176839511704,
      "grad_norm": 2.007319688796997,
      "learning_rate": 9.720763157894738e-05,
      "loss": 0.4803,
      "step": 30625
    },
    {
      "epoch": 1.7149736812632994,
      "grad_norm": 1.4103554487228394,
      "learning_rate": 9.720736842105264e-05,
      "loss": 0.4206,
      "step": 30626
    },
    {
      "epoch": 1.7150296785754284,
      "grad_norm": 1.0439146757125854,
      "learning_rate": 9.720710526315791e-05,
      "loss": 0.5129,
      "step": 30627
    },
    {
      "epoch": 1.7150856758875574,
      "grad_norm": 1.2654073238372803,
      "learning_rate": 9.720684210526316e-05,
      "loss": 0.4236,
      "step": 30628
    },
    {
      "epoch": 1.7151416731996865,
      "grad_norm": 1.275828242301941,
      "learning_rate": 9.720657894736842e-05,
      "loss": 0.4752,
      "step": 30629
    },
    {
      "epoch": 1.7151976705118155,
      "grad_norm": 1.4743921756744385,
      "learning_rate": 9.720631578947369e-05,
      "loss": 0.4481,
      "step": 30630
    },
    {
      "epoch": 1.7152536678239445,
      "grad_norm": 1.5907094478607178,
      "learning_rate": 9.720605263157895e-05,
      "loss": 0.4339,
      "step": 30631
    },
    {
      "epoch": 1.7153096651360733,
      "grad_norm": 2.2020251750946045,
      "learning_rate": 9.720578947368422e-05,
      "loss": 0.4501,
      "step": 30632
    },
    {
      "epoch": 1.7153656624482023,
      "grad_norm": 1.5741947889328003,
      "learning_rate": 9.720552631578947e-05,
      "loss": 0.5433,
      "step": 30633
    },
    {
      "epoch": 1.7154216597603313,
      "grad_norm": 1.4771791696548462,
      "learning_rate": 9.720526315789474e-05,
      "loss": 0.4377,
      "step": 30634
    },
    {
      "epoch": 1.7154776570724604,
      "grad_norm": 1.3025338649749756,
      "learning_rate": 9.7205e-05,
      "loss": 0.4054,
      "step": 30635
    },
    {
      "epoch": 1.7155336543845894,
      "grad_norm": 1.1977874040603638,
      "learning_rate": 9.720473684210528e-05,
      "loss": 0.4227,
      "step": 30636
    },
    {
      "epoch": 1.7155896516967184,
      "grad_norm": 1.209682583808899,
      "learning_rate": 9.720447368421053e-05,
      "loss": 0.3793,
      "step": 30637
    },
    {
      "epoch": 1.7156456490088474,
      "grad_norm": 1.4564845561981201,
      "learning_rate": 9.72042105263158e-05,
      "loss": 0.4768,
      "step": 30638
    },
    {
      "epoch": 1.7157016463209764,
      "grad_norm": 1.0919175148010254,
      "learning_rate": 9.720394736842105e-05,
      "loss": 0.4748,
      "step": 30639
    },
    {
      "epoch": 1.7157576436331055,
      "grad_norm": 1.1943644285202026,
      "learning_rate": 9.720368421052633e-05,
      "loss": 0.3998,
      "step": 30640
    },
    {
      "epoch": 1.7158136409452345,
      "grad_norm": 1.596779704093933,
      "learning_rate": 9.720342105263159e-05,
      "loss": 0.4699,
      "step": 30641
    },
    {
      "epoch": 1.7158696382573635,
      "grad_norm": 1.8138458728790283,
      "learning_rate": 9.720315789473685e-05,
      "loss": 0.6418,
      "step": 30642
    },
    {
      "epoch": 1.7159256355694925,
      "grad_norm": 1.606316089630127,
      "learning_rate": 9.720289473684211e-05,
      "loss": 0.602,
      "step": 30643
    },
    {
      "epoch": 1.7159816328816215,
      "grad_norm": 1.4772661924362183,
      "learning_rate": 9.720263157894738e-05,
      "loss": 0.4297,
      "step": 30644
    },
    {
      "epoch": 1.7160376301937506,
      "grad_norm": 2.572352409362793,
      "learning_rate": 9.720236842105264e-05,
      "loss": 0.3914,
      "step": 30645
    },
    {
      "epoch": 1.7160936275058796,
      "grad_norm": 1.6091371774673462,
      "learning_rate": 9.72021052631579e-05,
      "loss": 0.3716,
      "step": 30646
    },
    {
      "epoch": 1.7161496248180086,
      "grad_norm": 1.4084526300430298,
      "learning_rate": 9.720184210526316e-05,
      "loss": 0.4196,
      "step": 30647
    },
    {
      "epoch": 1.7162056221301376,
      "grad_norm": 1.1943351030349731,
      "learning_rate": 9.720157894736842e-05,
      "loss": 0.3734,
      "step": 30648
    },
    {
      "epoch": 1.7162616194422666,
      "grad_norm": 1.3019176721572876,
      "learning_rate": 9.720131578947369e-05,
      "loss": 0.5291,
      "step": 30649
    },
    {
      "epoch": 1.7163176167543956,
      "grad_norm": 1.2358137369155884,
      "learning_rate": 9.720105263157895e-05,
      "loss": 0.4952,
      "step": 30650
    },
    {
      "epoch": 1.7163736140665247,
      "grad_norm": 1.4986878633499146,
      "learning_rate": 9.720078947368421e-05,
      "loss": 0.6412,
      "step": 30651
    },
    {
      "epoch": 1.7164296113786537,
      "grad_norm": 1.4451597929000854,
      "learning_rate": 9.720052631578947e-05,
      "loss": 0.479,
      "step": 30652
    },
    {
      "epoch": 1.7164856086907827,
      "grad_norm": 1.2005128860473633,
      "learning_rate": 9.720026315789474e-05,
      "loss": 0.5046,
      "step": 30653
    },
    {
      "epoch": 1.7165416060029117,
      "grad_norm": 1.1480522155761719,
      "learning_rate": 9.72e-05,
      "loss": 0.489,
      "step": 30654
    },
    {
      "epoch": 1.7165976033150407,
      "grad_norm": 1.278332233428955,
      "learning_rate": 9.719973684210528e-05,
      "loss": 0.5082,
      "step": 30655
    },
    {
      "epoch": 1.7166536006271698,
      "grad_norm": 1.2766473293304443,
      "learning_rate": 9.719947368421052e-05,
      "loss": 0.4704,
      "step": 30656
    },
    {
      "epoch": 1.7167095979392988,
      "grad_norm": 1.597607135772705,
      "learning_rate": 9.71992105263158e-05,
      "loss": 0.4532,
      "step": 30657
    },
    {
      "epoch": 1.7167655952514278,
      "grad_norm": 1.3771840333938599,
      "learning_rate": 9.719894736842106e-05,
      "loss": 0.4287,
      "step": 30658
    },
    {
      "epoch": 1.7168215925635568,
      "grad_norm": 1.1253502368927002,
      "learning_rate": 9.719868421052633e-05,
      "loss": 0.3985,
      "step": 30659
    },
    {
      "epoch": 1.7168775898756858,
      "grad_norm": 1.0468671321868896,
      "learning_rate": 9.719842105263158e-05,
      "loss": 0.3813,
      "step": 30660
    },
    {
      "epoch": 1.7169335871878149,
      "grad_norm": 1.1901935338974,
      "learning_rate": 9.719815789473685e-05,
      "loss": 0.4057,
      "step": 30661
    },
    {
      "epoch": 1.7169895844999439,
      "grad_norm": 1.9568490982055664,
      "learning_rate": 9.719789473684211e-05,
      "loss": 0.3452,
      "step": 30662
    },
    {
      "epoch": 1.717045581812073,
      "grad_norm": 1.33143150806427,
      "learning_rate": 9.719763157894737e-05,
      "loss": 0.5552,
      "step": 30663
    },
    {
      "epoch": 1.717101579124202,
      "grad_norm": 1.3214884996414185,
      "learning_rate": 9.719736842105264e-05,
      "loss": 0.4655,
      "step": 30664
    },
    {
      "epoch": 1.717157576436331,
      "grad_norm": 1.1933867931365967,
      "learning_rate": 9.719710526315789e-05,
      "loss": 0.4986,
      "step": 30665
    },
    {
      "epoch": 1.71721357374846,
      "grad_norm": 1.3275268077850342,
      "learning_rate": 9.719684210526316e-05,
      "loss": 0.3995,
      "step": 30666
    },
    {
      "epoch": 1.717269571060589,
      "grad_norm": 1.329003930091858,
      "learning_rate": 9.719657894736842e-05,
      "loss": 0.4352,
      "step": 30667
    },
    {
      "epoch": 1.717325568372718,
      "grad_norm": 1.2409541606903076,
      "learning_rate": 9.71963157894737e-05,
      "loss": 0.4716,
      "step": 30668
    },
    {
      "epoch": 1.717381565684847,
      "grad_norm": 1.4531842470169067,
      "learning_rate": 9.719605263157895e-05,
      "loss": 0.4473,
      "step": 30669
    },
    {
      "epoch": 1.717437562996976,
      "grad_norm": 1.2469524145126343,
      "learning_rate": 9.719578947368421e-05,
      "loss": 0.47,
      "step": 30670
    },
    {
      "epoch": 1.717493560309105,
      "grad_norm": 1.8494130373001099,
      "learning_rate": 9.719552631578947e-05,
      "loss": 0.42,
      "step": 30671
    },
    {
      "epoch": 1.717549557621234,
      "grad_norm": 1.1138805150985718,
      "learning_rate": 9.719526315789475e-05,
      "loss": 0.4293,
      "step": 30672
    },
    {
      "epoch": 1.717605554933363,
      "grad_norm": 1.6776938438415527,
      "learning_rate": 9.719500000000001e-05,
      "loss": 0.4414,
      "step": 30673
    },
    {
      "epoch": 1.717661552245492,
      "grad_norm": 1.2268189191818237,
      "learning_rate": 9.719473684210527e-05,
      "loss": 0.4444,
      "step": 30674
    },
    {
      "epoch": 1.7177175495576211,
      "grad_norm": 1.4933257102966309,
      "learning_rate": 9.719447368421053e-05,
      "loss": 0.5096,
      "step": 30675
    },
    {
      "epoch": 1.7177735468697501,
      "grad_norm": 1.1103745698928833,
      "learning_rate": 9.71942105263158e-05,
      "loss": 0.3784,
      "step": 30676
    },
    {
      "epoch": 1.7178295441818792,
      "grad_norm": 1.309709072113037,
      "learning_rate": 9.719394736842106e-05,
      "loss": 0.5753,
      "step": 30677
    },
    {
      "epoch": 1.7178855414940082,
      "grad_norm": 1.2489408254623413,
      "learning_rate": 9.719368421052632e-05,
      "loss": 0.3883,
      "step": 30678
    },
    {
      "epoch": 1.7179415388061372,
      "grad_norm": 1.180154800415039,
      "learning_rate": 9.719342105263158e-05,
      "loss": 0.3999,
      "step": 30679
    },
    {
      "epoch": 1.7179975361182662,
      "grad_norm": 1.634352445602417,
      "learning_rate": 9.719315789473684e-05,
      "loss": 0.4748,
      "step": 30680
    },
    {
      "epoch": 1.7180535334303952,
      "grad_norm": 1.1118837594985962,
      "learning_rate": 9.719289473684211e-05,
      "loss": 0.4365,
      "step": 30681
    },
    {
      "epoch": 1.7181095307425243,
      "grad_norm": 1.3902416229248047,
      "learning_rate": 9.719263157894737e-05,
      "loss": 0.5039,
      "step": 30682
    },
    {
      "epoch": 1.7181655280546533,
      "grad_norm": 1.4275425672531128,
      "learning_rate": 9.719236842105263e-05,
      "loss": 0.4861,
      "step": 30683
    },
    {
      "epoch": 1.7182215253667823,
      "grad_norm": 1.4224351644515991,
      "learning_rate": 9.719210526315789e-05,
      "loss": 0.583,
      "step": 30684
    },
    {
      "epoch": 1.7182775226789113,
      "grad_norm": 1.3549447059631348,
      "learning_rate": 9.719184210526316e-05,
      "loss": 0.4253,
      "step": 30685
    },
    {
      "epoch": 1.7183335199910403,
      "grad_norm": 1.197293758392334,
      "learning_rate": 9.719157894736842e-05,
      "loss": 0.3436,
      "step": 30686
    },
    {
      "epoch": 1.7183895173031694,
      "grad_norm": 1.121922254562378,
      "learning_rate": 9.71913157894737e-05,
      "loss": 0.426,
      "step": 30687
    },
    {
      "epoch": 1.7184455146152984,
      "grad_norm": 1.2227044105529785,
      "learning_rate": 9.719105263157894e-05,
      "loss": 0.433,
      "step": 30688
    },
    {
      "epoch": 1.7185015119274274,
      "grad_norm": 1.234965443611145,
      "learning_rate": 9.719078947368422e-05,
      "loss": 0.4819,
      "step": 30689
    },
    {
      "epoch": 1.7185575092395564,
      "grad_norm": 1.185038685798645,
      "learning_rate": 9.719052631578948e-05,
      "loss": 0.4845,
      "step": 30690
    },
    {
      "epoch": 1.7186135065516854,
      "grad_norm": 1.0895192623138428,
      "learning_rate": 9.719026315789475e-05,
      "loss": 0.416,
      "step": 30691
    },
    {
      "epoch": 1.7186695038638145,
      "grad_norm": 1.145519733428955,
      "learning_rate": 9.719000000000001e-05,
      "loss": 0.5857,
      "step": 30692
    },
    {
      "epoch": 1.7187255011759435,
      "grad_norm": 1.2973244190216064,
      "learning_rate": 9.718973684210527e-05,
      "loss": 0.5575,
      "step": 30693
    },
    {
      "epoch": 1.7187814984880725,
      "grad_norm": 1.195210337638855,
      "learning_rate": 9.718947368421053e-05,
      "loss": 0.3612,
      "step": 30694
    },
    {
      "epoch": 1.7188374958002015,
      "grad_norm": 1.2516976594924927,
      "learning_rate": 9.71892105263158e-05,
      "loss": 0.5291,
      "step": 30695
    },
    {
      "epoch": 1.7188934931123305,
      "grad_norm": 1.301780104637146,
      "learning_rate": 9.718894736842106e-05,
      "loss": 0.4843,
      "step": 30696
    },
    {
      "epoch": 1.7189494904244595,
      "grad_norm": 1.1459444761276245,
      "learning_rate": 9.718868421052632e-05,
      "loss": 0.479,
      "step": 30697
    },
    {
      "epoch": 1.7190054877365886,
      "grad_norm": 1.6155439615249634,
      "learning_rate": 9.718842105263158e-05,
      "loss": 0.5189,
      "step": 30698
    },
    {
      "epoch": 1.7190614850487176,
      "grad_norm": 1.2038792371749878,
      "learning_rate": 9.718815789473684e-05,
      "loss": 0.3827,
      "step": 30699
    },
    {
      "epoch": 1.7191174823608466,
      "grad_norm": 1.0903184413909912,
      "learning_rate": 9.718789473684211e-05,
      "loss": 0.3966,
      "step": 30700
    },
    {
      "epoch": 1.7191734796729756,
      "grad_norm": 1.5096611976623535,
      "learning_rate": 9.718763157894737e-05,
      "loss": 0.4004,
      "step": 30701
    },
    {
      "epoch": 1.7192294769851046,
      "grad_norm": 1.1890404224395752,
      "learning_rate": 9.718736842105263e-05,
      "loss": 0.5064,
      "step": 30702
    },
    {
      "epoch": 1.7192854742972337,
      "grad_norm": 1.2429345846176147,
      "learning_rate": 9.71871052631579e-05,
      "loss": 0.5223,
      "step": 30703
    },
    {
      "epoch": 1.7193414716093627,
      "grad_norm": 1.2582592964172363,
      "learning_rate": 9.718684210526317e-05,
      "loss": 0.5147,
      "step": 30704
    },
    {
      "epoch": 1.7193974689214917,
      "grad_norm": 1.1780757904052734,
      "learning_rate": 9.718657894736843e-05,
      "loss": 0.4832,
      "step": 30705
    },
    {
      "epoch": 1.7194534662336207,
      "grad_norm": 1.1705541610717773,
      "learning_rate": 9.718631578947369e-05,
      "loss": 0.4167,
      "step": 30706
    },
    {
      "epoch": 1.7195094635457497,
      "grad_norm": 1.7118525505065918,
      "learning_rate": 9.718605263157895e-05,
      "loss": 0.5343,
      "step": 30707
    },
    {
      "epoch": 1.7195654608578788,
      "grad_norm": 1.2762616872787476,
      "learning_rate": 9.718578947368422e-05,
      "loss": 0.441,
      "step": 30708
    },
    {
      "epoch": 1.7196214581700078,
      "grad_norm": 1.195202350616455,
      "learning_rate": 9.718552631578948e-05,
      "loss": 0.5421,
      "step": 30709
    },
    {
      "epoch": 1.7196774554821368,
      "grad_norm": 1.2236769199371338,
      "learning_rate": 9.718526315789475e-05,
      "loss": 0.4745,
      "step": 30710
    },
    {
      "epoch": 1.7197334527942658,
      "grad_norm": 1.5898139476776123,
      "learning_rate": 9.7185e-05,
      "loss": 0.5193,
      "step": 30711
    },
    {
      "epoch": 1.7197894501063948,
      "grad_norm": 1.3720322847366333,
      "learning_rate": 9.718473684210527e-05,
      "loss": 0.4336,
      "step": 30712
    },
    {
      "epoch": 1.7198454474185239,
      "grad_norm": 1.2265517711639404,
      "learning_rate": 9.718447368421053e-05,
      "loss": 0.443,
      "step": 30713
    },
    {
      "epoch": 1.7199014447306529,
      "grad_norm": 1.514292597770691,
      "learning_rate": 9.718421052631579e-05,
      "loss": 0.4443,
      "step": 30714
    },
    {
      "epoch": 1.719957442042782,
      "grad_norm": 1.1352652311325073,
      "learning_rate": 9.718394736842105e-05,
      "loss": 0.3963,
      "step": 30715
    },
    {
      "epoch": 1.720013439354911,
      "grad_norm": 1.0488125085830688,
      "learning_rate": 9.718368421052631e-05,
      "loss": 0.4215,
      "step": 30716
    },
    {
      "epoch": 1.72006943666704,
      "grad_norm": 1.1707981824874878,
      "learning_rate": 9.718342105263158e-05,
      "loss": 0.4291,
      "step": 30717
    },
    {
      "epoch": 1.720125433979169,
      "grad_norm": 1.696366786956787,
      "learning_rate": 9.718315789473684e-05,
      "loss": 0.5962,
      "step": 30718
    },
    {
      "epoch": 1.720181431291298,
      "grad_norm": 1.5346957445144653,
      "learning_rate": 9.718289473684212e-05,
      "loss": 0.5974,
      "step": 30719
    },
    {
      "epoch": 1.720237428603427,
      "grad_norm": 1.9283950328826904,
      "learning_rate": 9.718263157894736e-05,
      "loss": 0.4745,
      "step": 30720
    },
    {
      "epoch": 1.720293425915556,
      "grad_norm": 1.3071030378341675,
      "learning_rate": 9.718236842105264e-05,
      "loss": 0.5455,
      "step": 30721
    },
    {
      "epoch": 1.720349423227685,
      "grad_norm": 1.199967384338379,
      "learning_rate": 9.71821052631579e-05,
      "loss": 0.4828,
      "step": 30722
    },
    {
      "epoch": 1.720405420539814,
      "grad_norm": 1.2231509685516357,
      "learning_rate": 9.718184210526317e-05,
      "loss": 0.4276,
      "step": 30723
    },
    {
      "epoch": 1.720461417851943,
      "grad_norm": 1.6901391744613647,
      "learning_rate": 9.718157894736843e-05,
      "loss": 0.5318,
      "step": 30724
    },
    {
      "epoch": 1.720517415164072,
      "grad_norm": 1.1251397132873535,
      "learning_rate": 9.718131578947369e-05,
      "loss": 0.4106,
      "step": 30725
    },
    {
      "epoch": 1.720573412476201,
      "grad_norm": 1.2451826333999634,
      "learning_rate": 9.718105263157895e-05,
      "loss": 0.4808,
      "step": 30726
    },
    {
      "epoch": 1.7206294097883301,
      "grad_norm": 1.2075666189193726,
      "learning_rate": 9.718078947368422e-05,
      "loss": 0.4009,
      "step": 30727
    },
    {
      "epoch": 1.7206854071004591,
      "grad_norm": 1.3249578475952148,
      "learning_rate": 9.718052631578948e-05,
      "loss": 0.4715,
      "step": 30728
    },
    {
      "epoch": 1.7207414044125882,
      "grad_norm": 1.2314748764038086,
      "learning_rate": 9.718026315789474e-05,
      "loss": 0.6087,
      "step": 30729
    },
    {
      "epoch": 1.7207974017247172,
      "grad_norm": 1.3579349517822266,
      "learning_rate": 9.718e-05,
      "loss": 0.5412,
      "step": 30730
    },
    {
      "epoch": 1.7208533990368462,
      "grad_norm": 1.256733775138855,
      "learning_rate": 9.717973684210526e-05,
      "loss": 0.4667,
      "step": 30731
    },
    {
      "epoch": 1.7209093963489752,
      "grad_norm": 1.2888498306274414,
      "learning_rate": 9.717947368421053e-05,
      "loss": 0.4215,
      "step": 30732
    },
    {
      "epoch": 1.7209653936611042,
      "grad_norm": 1.2255321741104126,
      "learning_rate": 9.71792105263158e-05,
      "loss": 0.4204,
      "step": 30733
    },
    {
      "epoch": 1.7210213909732333,
      "grad_norm": 1.2704070806503296,
      "learning_rate": 9.717894736842105e-05,
      "loss": 0.4192,
      "step": 30734
    },
    {
      "epoch": 1.7210773882853623,
      "grad_norm": 1.1680527925491333,
      "learning_rate": 9.717868421052631e-05,
      "loss": 0.5187,
      "step": 30735
    },
    {
      "epoch": 1.7211333855974913,
      "grad_norm": 1.2011053562164307,
      "learning_rate": 9.717842105263159e-05,
      "loss": 0.5848,
      "step": 30736
    },
    {
      "epoch": 1.7211893829096203,
      "grad_norm": 1.4032227993011475,
      "learning_rate": 9.717815789473685e-05,
      "loss": 0.4313,
      "step": 30737
    },
    {
      "epoch": 1.7212453802217493,
      "grad_norm": 1.2838280200958252,
      "learning_rate": 9.71778947368421e-05,
      "loss": 0.5346,
      "step": 30738
    },
    {
      "epoch": 1.7213013775338784,
      "grad_norm": 1.3266584873199463,
      "learning_rate": 9.717763157894737e-05,
      "loss": 0.4865,
      "step": 30739
    },
    {
      "epoch": 1.7213573748460074,
      "grad_norm": 1.2213317155838013,
      "learning_rate": 9.717736842105264e-05,
      "loss": 0.5428,
      "step": 30740
    },
    {
      "epoch": 1.7214133721581364,
      "grad_norm": 1.1680879592895508,
      "learning_rate": 9.71771052631579e-05,
      "loss": 0.4599,
      "step": 30741
    },
    {
      "epoch": 1.7214693694702654,
      "grad_norm": 1.1384347677230835,
      "learning_rate": 9.717684210526317e-05,
      "loss": 0.3859,
      "step": 30742
    },
    {
      "epoch": 1.7215253667823944,
      "grad_norm": 1.2647172212600708,
      "learning_rate": 9.717657894736842e-05,
      "loss": 0.4213,
      "step": 30743
    },
    {
      "epoch": 1.7215813640945234,
      "grad_norm": 1.1774641275405884,
      "learning_rate": 9.717631578947369e-05,
      "loss": 0.4956,
      "step": 30744
    },
    {
      "epoch": 1.7216373614066525,
      "grad_norm": 1.4863076210021973,
      "learning_rate": 9.717605263157895e-05,
      "loss": 0.4511,
      "step": 30745
    },
    {
      "epoch": 1.7216933587187815,
      "grad_norm": 1.4243805408477783,
      "learning_rate": 9.717578947368422e-05,
      "loss": 0.4707,
      "step": 30746
    },
    {
      "epoch": 1.7217493560309105,
      "grad_norm": 1.3286114931106567,
      "learning_rate": 9.717552631578948e-05,
      "loss": 0.4154,
      "step": 30747
    },
    {
      "epoch": 1.7218053533430395,
      "grad_norm": 1.3599361181259155,
      "learning_rate": 9.717526315789474e-05,
      "loss": 0.5631,
      "step": 30748
    },
    {
      "epoch": 1.7218613506551685,
      "grad_norm": 1.6556535959243774,
      "learning_rate": 9.7175e-05,
      "loss": 0.4746,
      "step": 30749
    },
    {
      "epoch": 1.7219173479672976,
      "grad_norm": 1.2787960767745972,
      "learning_rate": 9.717473684210526e-05,
      "loss": 0.4807,
      "step": 30750
    },
    {
      "epoch": 1.7219733452794266,
      "grad_norm": 1.2840867042541504,
      "learning_rate": 9.717447368421054e-05,
      "loss": 0.3179,
      "step": 30751
    },
    {
      "epoch": 1.7220293425915556,
      "grad_norm": 1.2753773927688599,
      "learning_rate": 9.717421052631578e-05,
      "loss": 0.389,
      "step": 30752
    },
    {
      "epoch": 1.7220853399036846,
      "grad_norm": 1.3589175939559937,
      "learning_rate": 9.717394736842106e-05,
      "loss": 0.6011,
      "step": 30753
    },
    {
      "epoch": 1.7221413372158136,
      "grad_norm": 1.447645902633667,
      "learning_rate": 9.717368421052632e-05,
      "loss": 0.3273,
      "step": 30754
    },
    {
      "epoch": 1.7221973345279427,
      "grad_norm": 1.3061238527297974,
      "learning_rate": 9.717342105263159e-05,
      "loss": 0.4034,
      "step": 30755
    },
    {
      "epoch": 1.7222533318400717,
      "grad_norm": 1.231913685798645,
      "learning_rate": 9.717315789473685e-05,
      "loss": 0.4542,
      "step": 30756
    },
    {
      "epoch": 1.7223093291522007,
      "grad_norm": 1.2924394607543945,
      "learning_rate": 9.717289473684211e-05,
      "loss": 0.5165,
      "step": 30757
    },
    {
      "epoch": 1.7223653264643297,
      "grad_norm": 1.3542739152908325,
      "learning_rate": 9.717263157894737e-05,
      "loss": 0.3954,
      "step": 30758
    },
    {
      "epoch": 1.7224213237764587,
      "grad_norm": 1.2190207242965698,
      "learning_rate": 9.717236842105264e-05,
      "loss": 0.3373,
      "step": 30759
    },
    {
      "epoch": 1.7224773210885878,
      "grad_norm": 1.327471375465393,
      "learning_rate": 9.71721052631579e-05,
      "loss": 0.5234,
      "step": 30760
    },
    {
      "epoch": 1.7225333184007168,
      "grad_norm": 1.2556498050689697,
      "learning_rate": 9.717184210526316e-05,
      "loss": 0.3365,
      "step": 30761
    },
    {
      "epoch": 1.7225893157128458,
      "grad_norm": 1.273229956626892,
      "learning_rate": 9.717157894736842e-05,
      "loss": 0.4334,
      "step": 30762
    },
    {
      "epoch": 1.7226453130249748,
      "grad_norm": 1.2415874004364014,
      "learning_rate": 9.71713157894737e-05,
      "loss": 0.3701,
      "step": 30763
    },
    {
      "epoch": 1.7227013103371038,
      "grad_norm": 1.1257222890853882,
      "learning_rate": 9.717105263157895e-05,
      "loss": 0.4524,
      "step": 30764
    },
    {
      "epoch": 1.7227573076492328,
      "grad_norm": 1.111696720123291,
      "learning_rate": 9.717078947368423e-05,
      "loss": 0.4103,
      "step": 30765
    },
    {
      "epoch": 1.7228133049613619,
      "grad_norm": 1.157478928565979,
      "learning_rate": 9.717052631578947e-05,
      "loss": 0.4071,
      "step": 30766
    },
    {
      "epoch": 1.7228693022734909,
      "grad_norm": 1.3755488395690918,
      "learning_rate": 9.717026315789473e-05,
      "loss": 0.441,
      "step": 30767
    },
    {
      "epoch": 1.72292529958562,
      "grad_norm": 1.4772863388061523,
      "learning_rate": 9.717e-05,
      "loss": 0.5367,
      "step": 30768
    },
    {
      "epoch": 1.722981296897749,
      "grad_norm": 1.3974822759628296,
      "learning_rate": 9.716973684210527e-05,
      "loss": 0.4845,
      "step": 30769
    },
    {
      "epoch": 1.723037294209878,
      "grad_norm": 1.4198808670043945,
      "learning_rate": 9.716947368421053e-05,
      "loss": 0.4812,
      "step": 30770
    },
    {
      "epoch": 1.723093291522007,
      "grad_norm": 1.354041337966919,
      "learning_rate": 9.716921052631579e-05,
      "loss": 0.4878,
      "step": 30771
    },
    {
      "epoch": 1.723149288834136,
      "grad_norm": 1.297965407371521,
      "learning_rate": 9.716894736842106e-05,
      "loss": 0.411,
      "step": 30772
    },
    {
      "epoch": 1.723205286146265,
      "grad_norm": 1.2439676523208618,
      "learning_rate": 9.716868421052632e-05,
      "loss": 0.4477,
      "step": 30773
    },
    {
      "epoch": 1.723261283458394,
      "grad_norm": 1.2469125986099243,
      "learning_rate": 9.716842105263159e-05,
      "loss": 0.48,
      "step": 30774
    },
    {
      "epoch": 1.723317280770523,
      "grad_norm": 1.226881980895996,
      "learning_rate": 9.716815789473684e-05,
      "loss": 0.4352,
      "step": 30775
    },
    {
      "epoch": 1.723373278082652,
      "grad_norm": 1.4902311563491821,
      "learning_rate": 9.716789473684211e-05,
      "loss": 0.3364,
      "step": 30776
    },
    {
      "epoch": 1.723429275394781,
      "grad_norm": 1.266294002532959,
      "learning_rate": 9.716763157894737e-05,
      "loss": 0.3488,
      "step": 30777
    },
    {
      "epoch": 1.72348527270691,
      "grad_norm": 1.145354986190796,
      "learning_rate": 9.716736842105264e-05,
      "loss": 0.3891,
      "step": 30778
    },
    {
      "epoch": 1.7235412700190391,
      "grad_norm": 1.250402569770813,
      "learning_rate": 9.71671052631579e-05,
      "loss": 0.5013,
      "step": 30779
    },
    {
      "epoch": 1.7235972673311681,
      "grad_norm": 1.360511064529419,
      "learning_rate": 9.716684210526316e-05,
      "loss": 0.4564,
      "step": 30780
    },
    {
      "epoch": 1.7236532646432972,
      "grad_norm": 1.220169186592102,
      "learning_rate": 9.716657894736842e-05,
      "loss": 0.4816,
      "step": 30781
    },
    {
      "epoch": 1.7237092619554262,
      "grad_norm": 1.2172590494155884,
      "learning_rate": 9.71663157894737e-05,
      "loss": 0.4198,
      "step": 30782
    },
    {
      "epoch": 1.7237652592675552,
      "grad_norm": 1.2992833852767944,
      "learning_rate": 9.716605263157896e-05,
      "loss": 0.5251,
      "step": 30783
    },
    {
      "epoch": 1.7238212565796842,
      "grad_norm": 1.2512496709823608,
      "learning_rate": 9.716578947368422e-05,
      "loss": 0.4944,
      "step": 30784
    },
    {
      "epoch": 1.7238772538918132,
      "grad_norm": 1.255361557006836,
      "learning_rate": 9.716552631578948e-05,
      "loss": 0.4415,
      "step": 30785
    },
    {
      "epoch": 1.7239332512039423,
      "grad_norm": 1.219032883644104,
      "learning_rate": 9.716526315789474e-05,
      "loss": 0.4437,
      "step": 30786
    },
    {
      "epoch": 1.7239892485160713,
      "grad_norm": 1.262127161026001,
      "learning_rate": 9.716500000000001e-05,
      "loss": 0.4603,
      "step": 30787
    },
    {
      "epoch": 1.7240452458282003,
      "grad_norm": 1.7772603034973145,
      "learning_rate": 9.716473684210527e-05,
      "loss": 0.5027,
      "step": 30788
    },
    {
      "epoch": 1.7241012431403293,
      "grad_norm": 1.373513102531433,
      "learning_rate": 9.716447368421053e-05,
      "loss": 0.4535,
      "step": 30789
    },
    {
      "epoch": 1.7241572404524583,
      "grad_norm": 1.1996124982833862,
      "learning_rate": 9.716421052631579e-05,
      "loss": 0.429,
      "step": 30790
    },
    {
      "epoch": 1.7242132377645873,
      "grad_norm": 1.2211042642593384,
      "learning_rate": 9.716394736842106e-05,
      "loss": 0.4429,
      "step": 30791
    },
    {
      "epoch": 1.7242692350767164,
      "grad_norm": 1.1607037782669067,
      "learning_rate": 9.716368421052632e-05,
      "loss": 0.3886,
      "step": 30792
    },
    {
      "epoch": 1.7243252323888454,
      "grad_norm": 1.408096194267273,
      "learning_rate": 9.716342105263158e-05,
      "loss": 0.5755,
      "step": 30793
    },
    {
      "epoch": 1.7243812297009744,
      "grad_norm": 1.4828472137451172,
      "learning_rate": 9.716315789473684e-05,
      "loss": 0.4418,
      "step": 30794
    },
    {
      "epoch": 1.7244372270131034,
      "grad_norm": 1.319539189338684,
      "learning_rate": 9.716289473684211e-05,
      "loss": 0.4372,
      "step": 30795
    },
    {
      "epoch": 1.7244932243252324,
      "grad_norm": 1.5043408870697021,
      "learning_rate": 9.716263157894737e-05,
      "loss": 0.4326,
      "step": 30796
    },
    {
      "epoch": 1.7245492216373615,
      "grad_norm": 1.0972951650619507,
      "learning_rate": 9.716236842105265e-05,
      "loss": 0.5003,
      "step": 30797
    },
    {
      "epoch": 1.7246052189494905,
      "grad_norm": 1.3848352432250977,
      "learning_rate": 9.716210526315789e-05,
      "loss": 0.4515,
      "step": 30798
    },
    {
      "epoch": 1.7246612162616195,
      "grad_norm": 1.3718483448028564,
      "learning_rate": 9.716184210526317e-05,
      "loss": 0.7014,
      "step": 30799
    },
    {
      "epoch": 1.7247172135737485,
      "grad_norm": 1.258340835571289,
      "learning_rate": 9.716157894736843e-05,
      "loss": 0.4507,
      "step": 30800
    },
    {
      "epoch": 1.7247732108858775,
      "grad_norm": 1.5159603357315063,
      "learning_rate": 9.716131578947369e-05,
      "loss": 0.5084,
      "step": 30801
    },
    {
      "epoch": 1.7248292081980066,
      "grad_norm": 1.6316993236541748,
      "learning_rate": 9.716105263157896e-05,
      "loss": 0.481,
      "step": 30802
    },
    {
      "epoch": 1.7248852055101356,
      "grad_norm": 1.4929754734039307,
      "learning_rate": 9.71607894736842e-05,
      "loss": 0.4773,
      "step": 30803
    },
    {
      "epoch": 1.7249412028222646,
      "grad_norm": 1.174686312675476,
      "learning_rate": 9.716052631578948e-05,
      "loss": 0.4631,
      "step": 30804
    },
    {
      "epoch": 1.7249972001343936,
      "grad_norm": 1.2033119201660156,
      "learning_rate": 9.716026315789474e-05,
      "loss": 0.5065,
      "step": 30805
    },
    {
      "epoch": 1.7250531974465226,
      "grad_norm": 1.6159331798553467,
      "learning_rate": 9.716000000000001e-05,
      "loss": 0.3934,
      "step": 30806
    },
    {
      "epoch": 1.7251091947586517,
      "grad_norm": 1.3112629652023315,
      "learning_rate": 9.715973684210526e-05,
      "loss": 0.3701,
      "step": 30807
    },
    {
      "epoch": 1.7251651920707807,
      "grad_norm": 1.2754032611846924,
      "learning_rate": 9.715947368421053e-05,
      "loss": 0.4287,
      "step": 30808
    },
    {
      "epoch": 1.7252211893829097,
      "grad_norm": 2.0877411365509033,
      "learning_rate": 9.715921052631579e-05,
      "loss": 0.6157,
      "step": 30809
    },
    {
      "epoch": 1.7252771866950387,
      "grad_norm": 1.2816237211227417,
      "learning_rate": 9.715894736842106e-05,
      "loss": 0.4074,
      "step": 30810
    },
    {
      "epoch": 1.7253331840071677,
      "grad_norm": 1.1529109477996826,
      "learning_rate": 9.715868421052632e-05,
      "loss": 0.4474,
      "step": 30811
    },
    {
      "epoch": 1.7253891813192967,
      "grad_norm": 1.130661129951477,
      "learning_rate": 9.715842105263158e-05,
      "loss": 0.5181,
      "step": 30812
    },
    {
      "epoch": 1.7254451786314258,
      "grad_norm": 1.5895882844924927,
      "learning_rate": 9.715815789473684e-05,
      "loss": 0.4612,
      "step": 30813
    },
    {
      "epoch": 1.7255011759435548,
      "grad_norm": 1.087876796722412,
      "learning_rate": 9.715789473684212e-05,
      "loss": 0.3793,
      "step": 30814
    },
    {
      "epoch": 1.7255571732556838,
      "grad_norm": 2.8522260189056396,
      "learning_rate": 9.715763157894738e-05,
      "loss": 0.3504,
      "step": 30815
    },
    {
      "epoch": 1.7256131705678128,
      "grad_norm": 2.2068569660186768,
      "learning_rate": 9.715736842105264e-05,
      "loss": 0.5478,
      "step": 30816
    },
    {
      "epoch": 1.7256691678799418,
      "grad_norm": 1.260810136795044,
      "learning_rate": 9.71571052631579e-05,
      "loss": 0.4399,
      "step": 30817
    },
    {
      "epoch": 1.7257251651920709,
      "grad_norm": 1.6022473573684692,
      "learning_rate": 9.715684210526316e-05,
      "loss": 0.3176,
      "step": 30818
    },
    {
      "epoch": 1.7257811625041999,
      "grad_norm": 1.4457985162734985,
      "learning_rate": 9.715657894736843e-05,
      "loss": 0.4669,
      "step": 30819
    },
    {
      "epoch": 1.725837159816329,
      "grad_norm": 1.2545794248580933,
      "learning_rate": 9.715631578947369e-05,
      "loss": 0.4905,
      "step": 30820
    },
    {
      "epoch": 1.725893157128458,
      "grad_norm": 1.3766297101974487,
      "learning_rate": 9.715605263157895e-05,
      "loss": 0.432,
      "step": 30821
    },
    {
      "epoch": 1.725949154440587,
      "grad_norm": 1.2831355333328247,
      "learning_rate": 9.715578947368421e-05,
      "loss": 0.419,
      "step": 30822
    },
    {
      "epoch": 1.726005151752716,
      "grad_norm": 1.179022192955017,
      "learning_rate": 9.715552631578948e-05,
      "loss": 0.4321,
      "step": 30823
    },
    {
      "epoch": 1.726061149064845,
      "grad_norm": 1.3134585618972778,
      "learning_rate": 9.715526315789474e-05,
      "loss": 0.5058,
      "step": 30824
    },
    {
      "epoch": 1.726117146376974,
      "grad_norm": 1.1801894903182983,
      "learning_rate": 9.7155e-05,
      "loss": 0.5043,
      "step": 30825
    },
    {
      "epoch": 1.726173143689103,
      "grad_norm": 1.549644947052002,
      "learning_rate": 9.715473684210526e-05,
      "loss": 0.4859,
      "step": 30826
    },
    {
      "epoch": 1.726229141001232,
      "grad_norm": 1.0734951496124268,
      "learning_rate": 9.715447368421053e-05,
      "loss": 0.414,
      "step": 30827
    },
    {
      "epoch": 1.726285138313361,
      "grad_norm": 1.5300109386444092,
      "learning_rate": 9.71542105263158e-05,
      "loss": 0.488,
      "step": 30828
    },
    {
      "epoch": 1.72634113562549,
      "grad_norm": 3.6961076259613037,
      "learning_rate": 9.715394736842107e-05,
      "loss": 0.5406,
      "step": 30829
    },
    {
      "epoch": 1.726397132937619,
      "grad_norm": 1.129594326019287,
      "learning_rate": 9.715368421052631e-05,
      "loss": 0.3016,
      "step": 30830
    },
    {
      "epoch": 1.7264531302497481,
      "grad_norm": 1.2709170579910278,
      "learning_rate": 9.715342105263159e-05,
      "loss": 0.452,
      "step": 30831
    },
    {
      "epoch": 1.7265091275618771,
      "grad_norm": 1.4917163848876953,
      "learning_rate": 9.715315789473685e-05,
      "loss": 0.4523,
      "step": 30832
    },
    {
      "epoch": 1.7265651248740062,
      "grad_norm": 1.2385468482971191,
      "learning_rate": 9.715289473684212e-05,
      "loss": 0.3755,
      "step": 30833
    },
    {
      "epoch": 1.7266211221861352,
      "grad_norm": 1.2770428657531738,
      "learning_rate": 9.715263157894738e-05,
      "loss": 0.4445,
      "step": 30834
    },
    {
      "epoch": 1.7266771194982642,
      "grad_norm": 1.444815754890442,
      "learning_rate": 9.715236842105263e-05,
      "loss": 0.4496,
      "step": 30835
    },
    {
      "epoch": 1.7267331168103932,
      "grad_norm": 1.4520589113235474,
      "learning_rate": 9.71521052631579e-05,
      "loss": 0.5309,
      "step": 30836
    },
    {
      "epoch": 1.7267891141225222,
      "grad_norm": 1.0482993125915527,
      "learning_rate": 9.715184210526316e-05,
      "loss": 0.4375,
      "step": 30837
    },
    {
      "epoch": 1.7268451114346512,
      "grad_norm": 1.4906712770462036,
      "learning_rate": 9.715157894736843e-05,
      "loss": 0.5349,
      "step": 30838
    },
    {
      "epoch": 1.7269011087467803,
      "grad_norm": 2.543030261993408,
      "learning_rate": 9.715131578947369e-05,
      "loss": 0.6079,
      "step": 30839
    },
    {
      "epoch": 1.7269571060589093,
      "grad_norm": 1.0798964500427246,
      "learning_rate": 9.715105263157895e-05,
      "loss": 0.4067,
      "step": 30840
    },
    {
      "epoch": 1.7270131033710383,
      "grad_norm": 1.2734647989273071,
      "learning_rate": 9.715078947368421e-05,
      "loss": 0.4286,
      "step": 30841
    },
    {
      "epoch": 1.7270691006831673,
      "grad_norm": 1.318042516708374,
      "learning_rate": 9.715052631578948e-05,
      "loss": 0.5077,
      "step": 30842
    },
    {
      "epoch": 1.7271250979952963,
      "grad_norm": 1.216072678565979,
      "learning_rate": 9.715026315789474e-05,
      "loss": 0.5351,
      "step": 30843
    },
    {
      "epoch": 1.7271810953074254,
      "grad_norm": 1.4882640838623047,
      "learning_rate": 9.715e-05,
      "loss": 0.475,
      "step": 30844
    },
    {
      "epoch": 1.7272370926195544,
      "grad_norm": 1.2969797849655151,
      "learning_rate": 9.714973684210526e-05,
      "loss": 0.422,
      "step": 30845
    },
    {
      "epoch": 1.7272930899316834,
      "grad_norm": 1.624219536781311,
      "learning_rate": 9.714947368421054e-05,
      "loss": 0.464,
      "step": 30846
    },
    {
      "epoch": 1.7273490872438124,
      "grad_norm": 1.4371284246444702,
      "learning_rate": 9.71492105263158e-05,
      "loss": 0.4519,
      "step": 30847
    },
    {
      "epoch": 1.7274050845559414,
      "grad_norm": 1.3296457529067993,
      "learning_rate": 9.714894736842106e-05,
      "loss": 0.4431,
      "step": 30848
    },
    {
      "epoch": 1.7274610818680705,
      "grad_norm": 1.1049184799194336,
      "learning_rate": 9.714868421052632e-05,
      "loss": 0.376,
      "step": 30849
    },
    {
      "epoch": 1.7275170791801995,
      "grad_norm": 1.1859382390975952,
      "learning_rate": 9.714842105263159e-05,
      "loss": 0.4267,
      "step": 30850
    },
    {
      "epoch": 1.7275730764923285,
      "grad_norm": 1.1705151796340942,
      "learning_rate": 9.714815789473685e-05,
      "loss": 0.3619,
      "step": 30851
    },
    {
      "epoch": 1.7276290738044575,
      "grad_norm": 1.3142000436782837,
      "learning_rate": 9.714789473684211e-05,
      "loss": 0.4021,
      "step": 30852
    },
    {
      "epoch": 1.7276850711165865,
      "grad_norm": 1.1487438678741455,
      "learning_rate": 9.714763157894737e-05,
      "loss": 0.4081,
      "step": 30853
    },
    {
      "epoch": 1.7277410684287156,
      "grad_norm": 1.276232361793518,
      "learning_rate": 9.714736842105263e-05,
      "loss": 0.4626,
      "step": 30854
    },
    {
      "epoch": 1.7277970657408446,
      "grad_norm": 1.2493582963943481,
      "learning_rate": 9.71471052631579e-05,
      "loss": 0.4618,
      "step": 30855
    },
    {
      "epoch": 1.7278530630529736,
      "grad_norm": 1.2609328031539917,
      "learning_rate": 9.714684210526316e-05,
      "loss": 0.3786,
      "step": 30856
    },
    {
      "epoch": 1.7279090603651026,
      "grad_norm": 1.2448943853378296,
      "learning_rate": 9.714657894736843e-05,
      "loss": 0.5076,
      "step": 30857
    },
    {
      "epoch": 1.7279650576772316,
      "grad_norm": 1.2144376039505005,
      "learning_rate": 9.714631578947368e-05,
      "loss": 0.5064,
      "step": 30858
    },
    {
      "epoch": 1.7280210549893606,
      "grad_norm": 1.4060462713241577,
      "learning_rate": 9.714605263157895e-05,
      "loss": 0.5556,
      "step": 30859
    },
    {
      "epoch": 1.7280770523014897,
      "grad_norm": 1.2787256240844727,
      "learning_rate": 9.714578947368421e-05,
      "loss": 0.3218,
      "step": 30860
    },
    {
      "epoch": 1.7281330496136187,
      "grad_norm": 1.5079458951950073,
      "learning_rate": 9.714552631578949e-05,
      "loss": 0.6093,
      "step": 30861
    },
    {
      "epoch": 1.7281890469257477,
      "grad_norm": 1.6231735944747925,
      "learning_rate": 9.714526315789473e-05,
      "loss": 0.5613,
      "step": 30862
    },
    {
      "epoch": 1.7282450442378767,
      "grad_norm": 1.4514434337615967,
      "learning_rate": 9.7145e-05,
      "loss": 0.4394,
      "step": 30863
    },
    {
      "epoch": 1.7283010415500057,
      "grad_norm": 1.2759157419204712,
      "learning_rate": 9.714473684210527e-05,
      "loss": 0.4208,
      "step": 30864
    },
    {
      "epoch": 1.7283570388621348,
      "grad_norm": 1.2782208919525146,
      "learning_rate": 9.714447368421054e-05,
      "loss": 0.5318,
      "step": 30865
    },
    {
      "epoch": 1.7284130361742638,
      "grad_norm": 2.5487778186798096,
      "learning_rate": 9.71442105263158e-05,
      "loss": 0.4648,
      "step": 30866
    },
    {
      "epoch": 1.7284690334863928,
      "grad_norm": 1.0467078685760498,
      "learning_rate": 9.714394736842106e-05,
      "loss": 0.3735,
      "step": 30867
    },
    {
      "epoch": 1.7285250307985218,
      "grad_norm": 1.2292864322662354,
      "learning_rate": 9.714368421052632e-05,
      "loss": 0.3864,
      "step": 30868
    },
    {
      "epoch": 1.7285810281106508,
      "grad_norm": 1.3292547464370728,
      "learning_rate": 9.714342105263159e-05,
      "loss": 0.6821,
      "step": 30869
    },
    {
      "epoch": 1.7286370254227799,
      "grad_norm": 1.1492819786071777,
      "learning_rate": 9.714315789473685e-05,
      "loss": 0.4084,
      "step": 30870
    },
    {
      "epoch": 1.7286930227349089,
      "grad_norm": 2.3479971885681152,
      "learning_rate": 9.714289473684211e-05,
      "loss": 0.4776,
      "step": 30871
    },
    {
      "epoch": 1.728749020047038,
      "grad_norm": 1.6894288063049316,
      "learning_rate": 9.714263157894737e-05,
      "loss": 0.5442,
      "step": 30872
    },
    {
      "epoch": 1.728805017359167,
      "grad_norm": 1.538828730583191,
      "learning_rate": 9.714236842105263e-05,
      "loss": 0.5265,
      "step": 30873
    },
    {
      "epoch": 1.728861014671296,
      "grad_norm": 1.340962290763855,
      "learning_rate": 9.71421052631579e-05,
      "loss": 0.5284,
      "step": 30874
    },
    {
      "epoch": 1.728917011983425,
      "grad_norm": 1.1624521017074585,
      "learning_rate": 9.714184210526316e-05,
      "loss": 0.2955,
      "step": 30875
    },
    {
      "epoch": 1.728973009295554,
      "grad_norm": 1.3382991552352905,
      "learning_rate": 9.714157894736842e-05,
      "loss": 0.4681,
      "step": 30876
    },
    {
      "epoch": 1.729029006607683,
      "grad_norm": 1.409085988998413,
      "learning_rate": 9.714131578947368e-05,
      "loss": 0.4515,
      "step": 30877
    },
    {
      "epoch": 1.729085003919812,
      "grad_norm": 1.0673515796661377,
      "learning_rate": 9.714105263157896e-05,
      "loss": 0.5178,
      "step": 30878
    },
    {
      "epoch": 1.729141001231941,
      "grad_norm": 1.5937429666519165,
      "learning_rate": 9.714078947368422e-05,
      "loss": 0.4162,
      "step": 30879
    },
    {
      "epoch": 1.7291969985440698,
      "grad_norm": 1.2507046461105347,
      "learning_rate": 9.714052631578948e-05,
      "loss": 0.392,
      "step": 30880
    },
    {
      "epoch": 1.7292529958561988,
      "grad_norm": 1.7141634225845337,
      "learning_rate": 9.714026315789474e-05,
      "loss": 0.3829,
      "step": 30881
    },
    {
      "epoch": 1.7293089931683279,
      "grad_norm": 1.924500584602356,
      "learning_rate": 9.714000000000001e-05,
      "loss": 0.5439,
      "step": 30882
    },
    {
      "epoch": 1.7293649904804569,
      "grad_norm": 1.5594667196273804,
      "learning_rate": 9.713973684210527e-05,
      "loss": 0.5527,
      "step": 30883
    },
    {
      "epoch": 1.729420987792586,
      "grad_norm": 1.3248144388198853,
      "learning_rate": 9.713947368421054e-05,
      "loss": 0.4547,
      "step": 30884
    },
    {
      "epoch": 1.729476985104715,
      "grad_norm": 1.5802890062332153,
      "learning_rate": 9.713921052631579e-05,
      "loss": 0.6615,
      "step": 30885
    },
    {
      "epoch": 1.729532982416844,
      "grad_norm": 1.5077229738235474,
      "learning_rate": 9.713894736842106e-05,
      "loss": 0.4718,
      "step": 30886
    },
    {
      "epoch": 1.729588979728973,
      "grad_norm": 1.4542040824890137,
      "learning_rate": 9.713868421052632e-05,
      "loss": 0.6412,
      "step": 30887
    },
    {
      "epoch": 1.729644977041102,
      "grad_norm": 1.3625273704528809,
      "learning_rate": 9.713842105263158e-05,
      "loss": 0.369,
      "step": 30888
    },
    {
      "epoch": 1.729700974353231,
      "grad_norm": 1.2317105531692505,
      "learning_rate": 9.713815789473685e-05,
      "loss": 0.326,
      "step": 30889
    },
    {
      "epoch": 1.72975697166536,
      "grad_norm": 1.2325578927993774,
      "learning_rate": 9.71378947368421e-05,
      "loss": 0.4449,
      "step": 30890
    },
    {
      "epoch": 1.729812968977489,
      "grad_norm": 1.545949935913086,
      "learning_rate": 9.713763157894737e-05,
      "loss": 0.4058,
      "step": 30891
    },
    {
      "epoch": 1.729868966289618,
      "grad_norm": 1.4828636646270752,
      "learning_rate": 9.713736842105263e-05,
      "loss": 0.5084,
      "step": 30892
    },
    {
      "epoch": 1.729924963601747,
      "grad_norm": 1.7670361995697021,
      "learning_rate": 9.71371052631579e-05,
      "loss": 0.5689,
      "step": 30893
    },
    {
      "epoch": 1.729980960913876,
      "grad_norm": 1.1305797100067139,
      "learning_rate": 9.713684210526317e-05,
      "loss": 0.4592,
      "step": 30894
    },
    {
      "epoch": 1.7300369582260051,
      "grad_norm": 1.1148333549499512,
      "learning_rate": 9.713657894736843e-05,
      "loss": 0.5318,
      "step": 30895
    },
    {
      "epoch": 1.7300929555381341,
      "grad_norm": 1.5832405090332031,
      "learning_rate": 9.713631578947369e-05,
      "loss": 0.4372,
      "step": 30896
    },
    {
      "epoch": 1.7301489528502632,
      "grad_norm": 1.43873929977417,
      "learning_rate": 9.713605263157896e-05,
      "loss": 0.47,
      "step": 30897
    },
    {
      "epoch": 1.7302049501623922,
      "grad_norm": 1.550362229347229,
      "learning_rate": 9.713578947368422e-05,
      "loss": 0.5809,
      "step": 30898
    },
    {
      "epoch": 1.7302609474745212,
      "grad_norm": 1.765970230102539,
      "learning_rate": 9.713552631578948e-05,
      "loss": 0.4819,
      "step": 30899
    },
    {
      "epoch": 1.7303169447866502,
      "grad_norm": 1.1776527166366577,
      "learning_rate": 9.713526315789474e-05,
      "loss": 0.4204,
      "step": 30900
    },
    {
      "epoch": 1.7303729420987792,
      "grad_norm": 1.1927144527435303,
      "learning_rate": 9.713500000000001e-05,
      "loss": 0.4343,
      "step": 30901
    },
    {
      "epoch": 1.7304289394109083,
      "grad_norm": 1.1371490955352783,
      "learning_rate": 9.713473684210527e-05,
      "loss": 0.5609,
      "step": 30902
    },
    {
      "epoch": 1.7304849367230373,
      "grad_norm": 1.316361665725708,
      "learning_rate": 9.713447368421053e-05,
      "loss": 0.5007,
      "step": 30903
    },
    {
      "epoch": 1.7305409340351663,
      "grad_norm": 1.2518609762191772,
      "learning_rate": 9.713421052631579e-05,
      "loss": 0.3775,
      "step": 30904
    },
    {
      "epoch": 1.7305969313472953,
      "grad_norm": 1.2294044494628906,
      "learning_rate": 9.713394736842105e-05,
      "loss": 0.4788,
      "step": 30905
    },
    {
      "epoch": 1.7306529286594243,
      "grad_norm": 1.3479539155960083,
      "learning_rate": 9.713368421052632e-05,
      "loss": 0.4313,
      "step": 30906
    },
    {
      "epoch": 1.7307089259715533,
      "grad_norm": 1.2183607816696167,
      "learning_rate": 9.713342105263158e-05,
      "loss": 0.3896,
      "step": 30907
    },
    {
      "epoch": 1.7307649232836824,
      "grad_norm": 17.11246109008789,
      "learning_rate": 9.713315789473684e-05,
      "loss": 0.636,
      "step": 30908
    },
    {
      "epoch": 1.7308209205958114,
      "grad_norm": 1.1182912588119507,
      "learning_rate": 9.71328947368421e-05,
      "loss": 0.3731,
      "step": 30909
    },
    {
      "epoch": 1.7308769179079404,
      "grad_norm": 1.2550162076950073,
      "learning_rate": 9.713263157894738e-05,
      "loss": 0.4245,
      "step": 30910
    },
    {
      "epoch": 1.7309329152200694,
      "grad_norm": 1.3438735008239746,
      "learning_rate": 9.713236842105264e-05,
      "loss": 0.4981,
      "step": 30911
    },
    {
      "epoch": 1.7309889125321984,
      "grad_norm": 1.236170768737793,
      "learning_rate": 9.713210526315791e-05,
      "loss": 0.4917,
      "step": 30912
    },
    {
      "epoch": 1.7310449098443275,
      "grad_norm": 1.380021333694458,
      "learning_rate": 9.713184210526315e-05,
      "loss": 0.5406,
      "step": 30913
    },
    {
      "epoch": 1.7311009071564565,
      "grad_norm": 1.2908653020858765,
      "learning_rate": 9.713157894736843e-05,
      "loss": 0.4543,
      "step": 30914
    },
    {
      "epoch": 1.7311569044685855,
      "grad_norm": 2.054720640182495,
      "learning_rate": 9.713131578947369e-05,
      "loss": 0.4492,
      "step": 30915
    },
    {
      "epoch": 1.7312129017807145,
      "grad_norm": 1.1338237524032593,
      "learning_rate": 9.713105263157896e-05,
      "loss": 0.3648,
      "step": 30916
    },
    {
      "epoch": 1.7312688990928435,
      "grad_norm": 1.230956792831421,
      "learning_rate": 9.713078947368421e-05,
      "loss": 0.3829,
      "step": 30917
    },
    {
      "epoch": 1.7313248964049726,
      "grad_norm": 1.3563530445098877,
      "learning_rate": 9.713052631578948e-05,
      "loss": 0.4956,
      "step": 30918
    },
    {
      "epoch": 1.7313808937171016,
      "grad_norm": 1.4030046463012695,
      "learning_rate": 9.713026315789474e-05,
      "loss": 0.4191,
      "step": 30919
    },
    {
      "epoch": 1.7314368910292306,
      "grad_norm": 1.2252717018127441,
      "learning_rate": 9.713000000000001e-05,
      "loss": 0.4479,
      "step": 30920
    },
    {
      "epoch": 1.7314928883413596,
      "grad_norm": 1.3108750581741333,
      "learning_rate": 9.712973684210527e-05,
      "loss": 0.5127,
      "step": 30921
    },
    {
      "epoch": 1.7315488856534886,
      "grad_norm": 1.378910779953003,
      "learning_rate": 9.712947368421052e-05,
      "loss": 0.4461,
      "step": 30922
    },
    {
      "epoch": 1.7316048829656177,
      "grad_norm": 1.148742437362671,
      "learning_rate": 9.712921052631579e-05,
      "loss": 0.3799,
      "step": 30923
    },
    {
      "epoch": 1.7316608802777467,
      "grad_norm": 1.2744581699371338,
      "learning_rate": 9.712894736842105e-05,
      "loss": 0.5466,
      "step": 30924
    },
    {
      "epoch": 1.7317168775898757,
      "grad_norm": 1.3952937126159668,
      "learning_rate": 9.712868421052633e-05,
      "loss": 0.4239,
      "step": 30925
    },
    {
      "epoch": 1.7317728749020047,
      "grad_norm": 1.8824533224105835,
      "learning_rate": 9.712842105263159e-05,
      "loss": 0.4215,
      "step": 30926
    },
    {
      "epoch": 1.7318288722141337,
      "grad_norm": 1.224982500076294,
      "learning_rate": 9.712815789473685e-05,
      "loss": 0.5098,
      "step": 30927
    },
    {
      "epoch": 1.7318848695262627,
      "grad_norm": 1.2281935214996338,
      "learning_rate": 9.71278947368421e-05,
      "loss": 0.5545,
      "step": 30928
    },
    {
      "epoch": 1.7319408668383918,
      "grad_norm": 1.7226656675338745,
      "learning_rate": 9.712763157894738e-05,
      "loss": 0.4299,
      "step": 30929
    },
    {
      "epoch": 1.7319968641505208,
      "grad_norm": 1.298384189605713,
      "learning_rate": 9.712736842105264e-05,
      "loss": 0.4545,
      "step": 30930
    },
    {
      "epoch": 1.7320528614626498,
      "grad_norm": 1.3409208059310913,
      "learning_rate": 9.71271052631579e-05,
      "loss": 0.4533,
      "step": 30931
    },
    {
      "epoch": 1.7321088587747788,
      "grad_norm": 1.1323697566986084,
      "learning_rate": 9.712684210526316e-05,
      "loss": 0.4413,
      "step": 30932
    },
    {
      "epoch": 1.7321648560869078,
      "grad_norm": 1.2897166013717651,
      "learning_rate": 9.712657894736843e-05,
      "loss": 0.3455,
      "step": 30933
    },
    {
      "epoch": 1.7322208533990369,
      "grad_norm": 1.7649589776992798,
      "learning_rate": 9.712631578947369e-05,
      "loss": 0.4775,
      "step": 30934
    },
    {
      "epoch": 1.7322768507111659,
      "grad_norm": 1.309889316558838,
      "learning_rate": 9.712605263157895e-05,
      "loss": 0.5041,
      "step": 30935
    },
    {
      "epoch": 1.732332848023295,
      "grad_norm": 1.0974708795547485,
      "learning_rate": 9.712578947368421e-05,
      "loss": 0.4333,
      "step": 30936
    },
    {
      "epoch": 1.732388845335424,
      "grad_norm": 1.4677408933639526,
      "learning_rate": 9.712552631578948e-05,
      "loss": 0.5098,
      "step": 30937
    },
    {
      "epoch": 1.732444842647553,
      "grad_norm": 1.055536150932312,
      "learning_rate": 9.712526315789474e-05,
      "loss": 0.3425,
      "step": 30938
    },
    {
      "epoch": 1.732500839959682,
      "grad_norm": 1.0858639478683472,
      "learning_rate": 9.7125e-05,
      "loss": 0.4275,
      "step": 30939
    },
    {
      "epoch": 1.732556837271811,
      "grad_norm": 1.2390285730361938,
      "learning_rate": 9.712473684210526e-05,
      "loss": 0.5106,
      "step": 30940
    },
    {
      "epoch": 1.73261283458394,
      "grad_norm": 1.2363107204437256,
      "learning_rate": 9.712447368421052e-05,
      "loss": 0.3923,
      "step": 30941
    },
    {
      "epoch": 1.732668831896069,
      "grad_norm": 1.291670322418213,
      "learning_rate": 9.71242105263158e-05,
      "loss": 0.4968,
      "step": 30942
    },
    {
      "epoch": 1.732724829208198,
      "grad_norm": 1.3040119409561157,
      "learning_rate": 9.712394736842106e-05,
      "loss": 0.53,
      "step": 30943
    },
    {
      "epoch": 1.732780826520327,
      "grad_norm": 1.4760606288909912,
      "learning_rate": 9.712368421052633e-05,
      "loss": 0.6047,
      "step": 30944
    },
    {
      "epoch": 1.732836823832456,
      "grad_norm": 1.2323769330978394,
      "learning_rate": 9.712342105263157e-05,
      "loss": 0.401,
      "step": 30945
    },
    {
      "epoch": 1.732892821144585,
      "grad_norm": 1.1582738161087036,
      "learning_rate": 9.712315789473685e-05,
      "loss": 0.4747,
      "step": 30946
    },
    {
      "epoch": 1.7329488184567141,
      "grad_norm": 1.2419651746749878,
      "learning_rate": 9.712289473684211e-05,
      "loss": 0.5276,
      "step": 30947
    },
    {
      "epoch": 1.7330048157688431,
      "grad_norm": 1.2945401668548584,
      "learning_rate": 9.712263157894738e-05,
      "loss": 0.4313,
      "step": 30948
    },
    {
      "epoch": 1.7330608130809722,
      "grad_norm": 1.580261468887329,
      "learning_rate": 9.712236842105264e-05,
      "loss": 0.4265,
      "step": 30949
    },
    {
      "epoch": 1.7331168103931012,
      "grad_norm": 1.2586312294006348,
      "learning_rate": 9.71221052631579e-05,
      "loss": 0.4663,
      "step": 30950
    },
    {
      "epoch": 1.7331728077052302,
      "grad_norm": 2.0549309253692627,
      "learning_rate": 9.712184210526316e-05,
      "loss": 0.4111,
      "step": 30951
    },
    {
      "epoch": 1.7332288050173592,
      "grad_norm": 1.2690858840942383,
      "learning_rate": 9.712157894736843e-05,
      "loss": 0.4778,
      "step": 30952
    },
    {
      "epoch": 1.7332848023294882,
      "grad_norm": 1.483779788017273,
      "learning_rate": 9.712131578947369e-05,
      "loss": 0.5175,
      "step": 30953
    },
    {
      "epoch": 1.7333407996416172,
      "grad_norm": 1.263841152191162,
      "learning_rate": 9.712105263157895e-05,
      "loss": 0.4895,
      "step": 30954
    },
    {
      "epoch": 1.7333967969537463,
      "grad_norm": 1.1458534002304077,
      "learning_rate": 9.712078947368421e-05,
      "loss": 0.4272,
      "step": 30955
    },
    {
      "epoch": 1.7334527942658753,
      "grad_norm": 1.2037091255187988,
      "learning_rate": 9.712052631578947e-05,
      "loss": 0.3797,
      "step": 30956
    },
    {
      "epoch": 1.7335087915780043,
      "grad_norm": 1.631466269493103,
      "learning_rate": 9.712026315789475e-05,
      "loss": 0.6362,
      "step": 30957
    },
    {
      "epoch": 1.7335647888901333,
      "grad_norm": 1.3013267517089844,
      "learning_rate": 9.712e-05,
      "loss": 0.5013,
      "step": 30958
    },
    {
      "epoch": 1.7336207862022623,
      "grad_norm": 1.4832836389541626,
      "learning_rate": 9.711973684210526e-05,
      "loss": 0.5935,
      "step": 30959
    },
    {
      "epoch": 1.7336767835143914,
      "grad_norm": 1.1693936586380005,
      "learning_rate": 9.711947368421052e-05,
      "loss": 0.3655,
      "step": 30960
    },
    {
      "epoch": 1.7337327808265204,
      "grad_norm": 1.3943352699279785,
      "learning_rate": 9.71192105263158e-05,
      "loss": 0.5045,
      "step": 30961
    },
    {
      "epoch": 1.7337887781386494,
      "grad_norm": 1.4526394605636597,
      "learning_rate": 9.711894736842106e-05,
      "loss": 0.4856,
      "step": 30962
    },
    {
      "epoch": 1.7338447754507782,
      "grad_norm": 1.68649423122406,
      "learning_rate": 9.711868421052632e-05,
      "loss": 0.4439,
      "step": 30963
    },
    {
      "epoch": 1.7339007727629072,
      "grad_norm": 1.2721911668777466,
      "learning_rate": 9.711842105263158e-05,
      "loss": 0.5415,
      "step": 30964
    },
    {
      "epoch": 1.7339567700750362,
      "grad_norm": 1.460144281387329,
      "learning_rate": 9.711815789473685e-05,
      "loss": 0.552,
      "step": 30965
    },
    {
      "epoch": 1.7340127673871653,
      "grad_norm": 1.6503968238830566,
      "learning_rate": 9.711789473684211e-05,
      "loss": 0.6655,
      "step": 30966
    },
    {
      "epoch": 1.7340687646992943,
      "grad_norm": 1.4066368341445923,
      "learning_rate": 9.711763157894738e-05,
      "loss": 0.5155,
      "step": 30967
    },
    {
      "epoch": 1.7341247620114233,
      "grad_norm": 1.1384308338165283,
      "learning_rate": 9.711736842105263e-05,
      "loss": 0.4905,
      "step": 30968
    },
    {
      "epoch": 1.7341807593235523,
      "grad_norm": 1.749942660331726,
      "learning_rate": 9.71171052631579e-05,
      "loss": 0.5055,
      "step": 30969
    },
    {
      "epoch": 1.7342367566356813,
      "grad_norm": 1.4464553594589233,
      "learning_rate": 9.711684210526316e-05,
      "loss": 0.452,
      "step": 30970
    },
    {
      "epoch": 1.7342927539478104,
      "grad_norm": 1.3154737949371338,
      "learning_rate": 9.711657894736844e-05,
      "loss": 0.4312,
      "step": 30971
    },
    {
      "epoch": 1.7343487512599394,
      "grad_norm": 1.425906777381897,
      "learning_rate": 9.711631578947368e-05,
      "loss": 0.615,
      "step": 30972
    },
    {
      "epoch": 1.7344047485720684,
      "grad_norm": 1.1637827157974243,
      "learning_rate": 9.711605263157894e-05,
      "loss": 0.4332,
      "step": 30973
    },
    {
      "epoch": 1.7344607458841974,
      "grad_norm": 1.2237335443496704,
      "learning_rate": 9.711578947368422e-05,
      "loss": 0.3171,
      "step": 30974
    },
    {
      "epoch": 1.7345167431963264,
      "grad_norm": 1.2054741382598877,
      "learning_rate": 9.711552631578947e-05,
      "loss": 0.4881,
      "step": 30975
    },
    {
      "epoch": 1.7345727405084554,
      "grad_norm": 1.2754313945770264,
      "learning_rate": 9.711526315789475e-05,
      "loss": 0.4256,
      "step": 30976
    },
    {
      "epoch": 1.7346287378205845,
      "grad_norm": 1.0154242515563965,
      "learning_rate": 9.7115e-05,
      "loss": 0.3056,
      "step": 30977
    },
    {
      "epoch": 1.7346847351327135,
      "grad_norm": 1.235609531402588,
      "learning_rate": 9.711473684210527e-05,
      "loss": 0.459,
      "step": 30978
    },
    {
      "epoch": 1.7347407324448425,
      "grad_norm": 1.4229402542114258,
      "learning_rate": 9.711447368421053e-05,
      "loss": 0.4944,
      "step": 30979
    },
    {
      "epoch": 1.7347967297569715,
      "grad_norm": 1.2148220539093018,
      "learning_rate": 9.71142105263158e-05,
      "loss": 0.5224,
      "step": 30980
    },
    {
      "epoch": 1.7348527270691005,
      "grad_norm": 1.2697978019714355,
      "learning_rate": 9.711394736842106e-05,
      "loss": 0.3654,
      "step": 30981
    },
    {
      "epoch": 1.7349087243812296,
      "grad_norm": 1.1354392766952515,
      "learning_rate": 9.711368421052632e-05,
      "loss": 0.4228,
      "step": 30982
    },
    {
      "epoch": 1.7349647216933586,
      "grad_norm": 1.4376211166381836,
      "learning_rate": 9.711342105263158e-05,
      "loss": 0.4614,
      "step": 30983
    },
    {
      "epoch": 1.7350207190054876,
      "grad_norm": 1.1127361059188843,
      "learning_rate": 9.711315789473685e-05,
      "loss": 0.3225,
      "step": 30984
    },
    {
      "epoch": 1.7350767163176166,
      "grad_norm": 1.2308012247085571,
      "learning_rate": 9.711289473684211e-05,
      "loss": 0.4746,
      "step": 30985
    },
    {
      "epoch": 1.7351327136297456,
      "grad_norm": 1.4758014678955078,
      "learning_rate": 9.711263157894737e-05,
      "loss": 0.4164,
      "step": 30986
    },
    {
      "epoch": 1.7351887109418747,
      "grad_norm": 6.427015781402588,
      "learning_rate": 9.711236842105263e-05,
      "loss": 0.5271,
      "step": 30987
    },
    {
      "epoch": 1.7352447082540037,
      "grad_norm": 1.6755577325820923,
      "learning_rate": 9.71121052631579e-05,
      "loss": 0.5436,
      "step": 30988
    },
    {
      "epoch": 1.7353007055661327,
      "grad_norm": 1.2384088039398193,
      "learning_rate": 9.711184210526317e-05,
      "loss": 0.3598,
      "step": 30989
    },
    {
      "epoch": 1.7353567028782617,
      "grad_norm": 1.1982392072677612,
      "learning_rate": 9.711157894736842e-05,
      "loss": 0.3884,
      "step": 30990
    },
    {
      "epoch": 1.7354127001903907,
      "grad_norm": 1.1121225357055664,
      "learning_rate": 9.711131578947368e-05,
      "loss": 0.4059,
      "step": 30991
    },
    {
      "epoch": 1.7354686975025198,
      "grad_norm": 1.5421463251113892,
      "learning_rate": 9.711105263157894e-05,
      "loss": 0.395,
      "step": 30992
    },
    {
      "epoch": 1.7355246948146488,
      "grad_norm": 1.4363654851913452,
      "learning_rate": 9.711078947368422e-05,
      "loss": 0.5026,
      "step": 30993
    },
    {
      "epoch": 1.7355806921267778,
      "grad_norm": 1.8139078617095947,
      "learning_rate": 9.711052631578948e-05,
      "loss": 0.6181,
      "step": 30994
    },
    {
      "epoch": 1.7356366894389068,
      "grad_norm": 1.1919149160385132,
      "learning_rate": 9.711026315789474e-05,
      "loss": 0.4262,
      "step": 30995
    },
    {
      "epoch": 1.7356926867510358,
      "grad_norm": 1.5225441455841064,
      "learning_rate": 9.711e-05,
      "loss": 0.5237,
      "step": 30996
    },
    {
      "epoch": 1.7357486840631648,
      "grad_norm": 1.6076759099960327,
      "learning_rate": 9.710973684210527e-05,
      "loss": 0.4948,
      "step": 30997
    },
    {
      "epoch": 1.7358046813752939,
      "grad_norm": 1.1972556114196777,
      "learning_rate": 9.710947368421053e-05,
      "loss": 0.3819,
      "step": 30998
    },
    {
      "epoch": 1.7358606786874229,
      "grad_norm": 1.541182041168213,
      "learning_rate": 9.71092105263158e-05,
      "loss": 0.5491,
      "step": 30999
    },
    {
      "epoch": 1.735916675999552,
      "grad_norm": 1.5775507688522339,
      "learning_rate": 9.710894736842105e-05,
      "loss": 0.5367,
      "step": 31000
    },
    {
      "epoch": 1.735972673311681,
      "grad_norm": 1.3196219205856323,
      "learning_rate": 9.710868421052632e-05,
      "loss": 0.508,
      "step": 31001
    },
    {
      "epoch": 1.73602867062381,
      "grad_norm": 1.4850828647613525,
      "learning_rate": 9.710842105263158e-05,
      "loss": 0.5157,
      "step": 31002
    },
    {
      "epoch": 1.736084667935939,
      "grad_norm": 1.3666611909866333,
      "learning_rate": 9.710815789473686e-05,
      "loss": 0.5597,
      "step": 31003
    },
    {
      "epoch": 1.736140665248068,
      "grad_norm": 1.231253981590271,
      "learning_rate": 9.710789473684212e-05,
      "loss": 0.4525,
      "step": 31004
    },
    {
      "epoch": 1.736196662560197,
      "grad_norm": 1.0799951553344727,
      "learning_rate": 9.710763157894738e-05,
      "loss": 0.387,
      "step": 31005
    },
    {
      "epoch": 1.736252659872326,
      "grad_norm": 1.2046945095062256,
      "learning_rate": 9.710736842105263e-05,
      "loss": 0.4121,
      "step": 31006
    },
    {
      "epoch": 1.736308657184455,
      "grad_norm": 1.1753010749816895,
      "learning_rate": 9.710710526315791e-05,
      "loss": 0.3987,
      "step": 31007
    },
    {
      "epoch": 1.736364654496584,
      "grad_norm": 1.2389849424362183,
      "learning_rate": 9.710684210526317e-05,
      "loss": 0.5673,
      "step": 31008
    },
    {
      "epoch": 1.736420651808713,
      "grad_norm": 1.3578776121139526,
      "learning_rate": 9.710657894736841e-05,
      "loss": 0.4629,
      "step": 31009
    },
    {
      "epoch": 1.736476649120842,
      "grad_norm": 1.1498032808303833,
      "learning_rate": 9.710631578947369e-05,
      "loss": 0.5048,
      "step": 31010
    },
    {
      "epoch": 1.7365326464329711,
      "grad_norm": 1.311370611190796,
      "learning_rate": 9.710605263157895e-05,
      "loss": 0.4367,
      "step": 31011
    },
    {
      "epoch": 1.7365886437451001,
      "grad_norm": 1.094753623008728,
      "learning_rate": 9.710578947368422e-05,
      "loss": 0.4911,
      "step": 31012
    },
    {
      "epoch": 1.7366446410572292,
      "grad_norm": 1.3457807302474976,
      "learning_rate": 9.710552631578948e-05,
      "loss": 0.3923,
      "step": 31013
    },
    {
      "epoch": 1.7367006383693582,
      "grad_norm": 1.321081280708313,
      "learning_rate": 9.710526315789474e-05,
      "loss": 0.402,
      "step": 31014
    },
    {
      "epoch": 1.7367566356814872,
      "grad_norm": 1.2904243469238281,
      "learning_rate": 9.7105e-05,
      "loss": 0.4171,
      "step": 31015
    },
    {
      "epoch": 1.7368126329936162,
      "grad_norm": 1.4380868673324585,
      "learning_rate": 9.710473684210527e-05,
      "loss": 0.3509,
      "step": 31016
    },
    {
      "epoch": 1.7368686303057452,
      "grad_norm": 1.2378318309783936,
      "learning_rate": 9.710447368421053e-05,
      "loss": 0.4268,
      "step": 31017
    },
    {
      "epoch": 1.7369246276178743,
      "grad_norm": 1.30984628200531,
      "learning_rate": 9.710421052631579e-05,
      "loss": 0.4519,
      "step": 31018
    },
    {
      "epoch": 1.7369806249300033,
      "grad_norm": 1.7147924900054932,
      "learning_rate": 9.710394736842105e-05,
      "loss": 0.536,
      "step": 31019
    },
    {
      "epoch": 1.7370366222421323,
      "grad_norm": 1.2327755689620972,
      "learning_rate": 9.710368421052633e-05,
      "loss": 0.4454,
      "step": 31020
    },
    {
      "epoch": 1.7370926195542613,
      "grad_norm": 1.0665234327316284,
      "learning_rate": 9.710342105263158e-05,
      "loss": 0.3624,
      "step": 31021
    },
    {
      "epoch": 1.7371486168663903,
      "grad_norm": 1.1231725215911865,
      "learning_rate": 9.710315789473684e-05,
      "loss": 0.398,
      "step": 31022
    },
    {
      "epoch": 1.7372046141785193,
      "grad_norm": 1.2381103038787842,
      "learning_rate": 9.71028947368421e-05,
      "loss": 0.445,
      "step": 31023
    },
    {
      "epoch": 1.7372606114906484,
      "grad_norm": 1.2365427017211914,
      "learning_rate": 9.710263157894738e-05,
      "loss": 0.4542,
      "step": 31024
    },
    {
      "epoch": 1.7373166088027774,
      "grad_norm": 1.264270305633545,
      "learning_rate": 9.710236842105264e-05,
      "loss": 0.5108,
      "step": 31025
    },
    {
      "epoch": 1.7373726061149064,
      "grad_norm": 1.1615774631500244,
      "learning_rate": 9.71021052631579e-05,
      "loss": 0.4658,
      "step": 31026
    },
    {
      "epoch": 1.7374286034270354,
      "grad_norm": 1.5205755233764648,
      "learning_rate": 9.710184210526316e-05,
      "loss": 0.7027,
      "step": 31027
    },
    {
      "epoch": 1.7374846007391644,
      "grad_norm": 1.4298266172409058,
      "learning_rate": 9.710157894736842e-05,
      "loss": 0.4847,
      "step": 31028
    },
    {
      "epoch": 1.7375405980512935,
      "grad_norm": 1.3484586477279663,
      "learning_rate": 9.710131578947369e-05,
      "loss": 0.5279,
      "step": 31029
    },
    {
      "epoch": 1.7375965953634225,
      "grad_norm": 1.2147866487503052,
      "learning_rate": 9.710105263157895e-05,
      "loss": 0.4323,
      "step": 31030
    },
    {
      "epoch": 1.7376525926755515,
      "grad_norm": 1.224869966506958,
      "learning_rate": 9.710078947368422e-05,
      "loss": 0.5024,
      "step": 31031
    },
    {
      "epoch": 1.7377085899876805,
      "grad_norm": 1.0645188093185425,
      "learning_rate": 9.710052631578947e-05,
      "loss": 0.5958,
      "step": 31032
    },
    {
      "epoch": 1.7377645872998095,
      "grad_norm": 1.4227310419082642,
      "learning_rate": 9.710026315789474e-05,
      "loss": 0.4599,
      "step": 31033
    },
    {
      "epoch": 1.7378205846119386,
      "grad_norm": 1.4048056602478027,
      "learning_rate": 9.71e-05,
      "loss": 0.4401,
      "step": 31034
    },
    {
      "epoch": 1.7378765819240676,
      "grad_norm": 1.2840206623077393,
      "learning_rate": 9.709973684210528e-05,
      "loss": 0.4556,
      "step": 31035
    },
    {
      "epoch": 1.7379325792361966,
      "grad_norm": 1.3036516904830933,
      "learning_rate": 9.709947368421054e-05,
      "loss": 0.4384,
      "step": 31036
    },
    {
      "epoch": 1.7379885765483256,
      "grad_norm": 1.2935075759887695,
      "learning_rate": 9.70992105263158e-05,
      "loss": 0.4812,
      "step": 31037
    },
    {
      "epoch": 1.7380445738604546,
      "grad_norm": 1.361349105834961,
      "learning_rate": 9.709894736842105e-05,
      "loss": 0.42,
      "step": 31038
    },
    {
      "epoch": 1.7381005711725837,
      "grad_norm": 1.3498167991638184,
      "learning_rate": 9.709868421052633e-05,
      "loss": 0.458,
      "step": 31039
    },
    {
      "epoch": 1.7381565684847127,
      "grad_norm": 1.1751586198806763,
      "learning_rate": 9.709842105263159e-05,
      "loss": 0.4148,
      "step": 31040
    },
    {
      "epoch": 1.7382125657968417,
      "grad_norm": 1.1283135414123535,
      "learning_rate": 9.709815789473685e-05,
      "loss": 0.2819,
      "step": 31041
    },
    {
      "epoch": 1.7382685631089707,
      "grad_norm": 1.1883455514907837,
      "learning_rate": 9.709789473684211e-05,
      "loss": 0.4182,
      "step": 31042
    },
    {
      "epoch": 1.7383245604210997,
      "grad_norm": 1.396022081375122,
      "learning_rate": 9.709763157894737e-05,
      "loss": 0.4821,
      "step": 31043
    },
    {
      "epoch": 1.7383805577332287,
      "grad_norm": 1.1936545372009277,
      "learning_rate": 9.709736842105264e-05,
      "loss": 0.4972,
      "step": 31044
    },
    {
      "epoch": 1.7384365550453578,
      "grad_norm": 1.2382044792175293,
      "learning_rate": 9.70971052631579e-05,
      "loss": 0.5357,
      "step": 31045
    },
    {
      "epoch": 1.7384925523574868,
      "grad_norm": 1.3015687465667725,
      "learning_rate": 9.709684210526316e-05,
      "loss": 0.4523,
      "step": 31046
    },
    {
      "epoch": 1.7385485496696158,
      "grad_norm": 1.3992559909820557,
      "learning_rate": 9.709657894736842e-05,
      "loss": 0.3672,
      "step": 31047
    },
    {
      "epoch": 1.7386045469817448,
      "grad_norm": 1.282843828201294,
      "learning_rate": 9.709631578947369e-05,
      "loss": 0.4421,
      "step": 31048
    },
    {
      "epoch": 1.7386605442938738,
      "grad_norm": 1.4199289083480835,
      "learning_rate": 9.709605263157895e-05,
      "loss": 0.5129,
      "step": 31049
    },
    {
      "epoch": 1.7387165416060029,
      "grad_norm": 0.9760697484016418,
      "learning_rate": 9.709578947368421e-05,
      "loss": 0.3753,
      "step": 31050
    },
    {
      "epoch": 1.7387725389181319,
      "grad_norm": 1.2729848623275757,
      "learning_rate": 9.709552631578947e-05,
      "loss": 0.4728,
      "step": 31051
    },
    {
      "epoch": 1.738828536230261,
      "grad_norm": 1.1899892091751099,
      "learning_rate": 9.709526315789474e-05,
      "loss": 0.4492,
      "step": 31052
    },
    {
      "epoch": 1.73888453354239,
      "grad_norm": 1.205531358718872,
      "learning_rate": 9.7095e-05,
      "loss": 0.5223,
      "step": 31053
    },
    {
      "epoch": 1.738940530854519,
      "grad_norm": 1.3818005323410034,
      "learning_rate": 9.709473684210528e-05,
      "loss": 0.3802,
      "step": 31054
    },
    {
      "epoch": 1.738996528166648,
      "grad_norm": 1.2059881687164307,
      "learning_rate": 9.709447368421052e-05,
      "loss": 0.4753,
      "step": 31055
    },
    {
      "epoch": 1.739052525478777,
      "grad_norm": 1.1895244121551514,
      "learning_rate": 9.70942105263158e-05,
      "loss": 0.4158,
      "step": 31056
    },
    {
      "epoch": 1.739108522790906,
      "grad_norm": 1.4219253063201904,
      "learning_rate": 9.709394736842106e-05,
      "loss": 0.398,
      "step": 31057
    },
    {
      "epoch": 1.739164520103035,
      "grad_norm": 1.3253235816955566,
      "learning_rate": 9.709368421052633e-05,
      "loss": 0.5903,
      "step": 31058
    },
    {
      "epoch": 1.739220517415164,
      "grad_norm": 1.251280665397644,
      "learning_rate": 9.709342105263159e-05,
      "loss": 0.484,
      "step": 31059
    },
    {
      "epoch": 1.739276514727293,
      "grad_norm": 1.4518582820892334,
      "learning_rate": 9.709315789473684e-05,
      "loss": 0.5893,
      "step": 31060
    },
    {
      "epoch": 1.739332512039422,
      "grad_norm": 1.418822169303894,
      "learning_rate": 9.709289473684211e-05,
      "loss": 0.5994,
      "step": 31061
    },
    {
      "epoch": 1.739388509351551,
      "grad_norm": 1.2382903099060059,
      "learning_rate": 9.709263157894737e-05,
      "loss": 0.4333,
      "step": 31062
    },
    {
      "epoch": 1.7394445066636801,
      "grad_norm": 1.23009192943573,
      "learning_rate": 9.709236842105264e-05,
      "loss": 0.5559,
      "step": 31063
    },
    {
      "epoch": 1.7395005039758091,
      "grad_norm": 1.3467671871185303,
      "learning_rate": 9.709210526315789e-05,
      "loss": 0.441,
      "step": 31064
    },
    {
      "epoch": 1.7395565012879382,
      "grad_norm": 1.2569304704666138,
      "learning_rate": 9.709184210526316e-05,
      "loss": 0.4839,
      "step": 31065
    },
    {
      "epoch": 1.7396124986000672,
      "grad_norm": 1.2604433298110962,
      "learning_rate": 9.709157894736842e-05,
      "loss": 0.4684,
      "step": 31066
    },
    {
      "epoch": 1.7396684959121962,
      "grad_norm": 1.1469084024429321,
      "learning_rate": 9.70913157894737e-05,
      "loss": 0.4653,
      "step": 31067
    },
    {
      "epoch": 1.7397244932243252,
      "grad_norm": 1.407639503479004,
      "learning_rate": 9.709105263157895e-05,
      "loss": 0.4579,
      "step": 31068
    },
    {
      "epoch": 1.7397804905364542,
      "grad_norm": 1.30097496509552,
      "learning_rate": 9.709078947368421e-05,
      "loss": 0.6514,
      "step": 31069
    },
    {
      "epoch": 1.7398364878485832,
      "grad_norm": 1.5303374528884888,
      "learning_rate": 9.709052631578947e-05,
      "loss": 0.477,
      "step": 31070
    },
    {
      "epoch": 1.7398924851607123,
      "grad_norm": 1.188430905342102,
      "learning_rate": 9.709026315789475e-05,
      "loss": 0.4511,
      "step": 31071
    },
    {
      "epoch": 1.7399484824728413,
      "grad_norm": 1.0792704820632935,
      "learning_rate": 9.709000000000001e-05,
      "loss": 0.3328,
      "step": 31072
    },
    {
      "epoch": 1.7400044797849703,
      "grad_norm": 1.2438716888427734,
      "learning_rate": 9.708973684210527e-05,
      "loss": 0.3821,
      "step": 31073
    },
    {
      "epoch": 1.7400604770970993,
      "grad_norm": 1.2731200456619263,
      "learning_rate": 9.708947368421053e-05,
      "loss": 0.4448,
      "step": 31074
    },
    {
      "epoch": 1.7401164744092283,
      "grad_norm": 1.2387651205062866,
      "learning_rate": 9.70892105263158e-05,
      "loss": 0.4412,
      "step": 31075
    },
    {
      "epoch": 1.7401724717213574,
      "grad_norm": 1.2949508428573608,
      "learning_rate": 9.708894736842106e-05,
      "loss": 0.47,
      "step": 31076
    },
    {
      "epoch": 1.7402284690334864,
      "grad_norm": 1.228320598602295,
      "learning_rate": 9.708868421052632e-05,
      "loss": 0.408,
      "step": 31077
    },
    {
      "epoch": 1.7402844663456154,
      "grad_norm": 1.0673197507858276,
      "learning_rate": 9.708842105263158e-05,
      "loss": 0.3355,
      "step": 31078
    },
    {
      "epoch": 1.7403404636577444,
      "grad_norm": 1.6339677572250366,
      "learning_rate": 9.708815789473684e-05,
      "loss": 0.5258,
      "step": 31079
    },
    {
      "epoch": 1.7403964609698734,
      "grad_norm": 1.0141102075576782,
      "learning_rate": 9.708789473684211e-05,
      "loss": 0.3434,
      "step": 31080
    },
    {
      "epoch": 1.7404524582820025,
      "grad_norm": 1.1632977724075317,
      "learning_rate": 9.708763157894737e-05,
      "loss": 0.4494,
      "step": 31081
    },
    {
      "epoch": 1.7405084555941315,
      "grad_norm": 1.386582374572754,
      "learning_rate": 9.708736842105263e-05,
      "loss": 0.5816,
      "step": 31082
    },
    {
      "epoch": 1.7405644529062605,
      "grad_norm": 1.2117961645126343,
      "learning_rate": 9.708710526315789e-05,
      "loss": 0.3829,
      "step": 31083
    },
    {
      "epoch": 1.7406204502183895,
      "grad_norm": 1.2344613075256348,
      "learning_rate": 9.708684210526316e-05,
      "loss": 0.6839,
      "step": 31084
    },
    {
      "epoch": 1.7406764475305185,
      "grad_norm": 1.557342529296875,
      "learning_rate": 9.708657894736842e-05,
      "loss": 0.6371,
      "step": 31085
    },
    {
      "epoch": 1.7407324448426476,
      "grad_norm": 1.478472352027893,
      "learning_rate": 9.70863157894737e-05,
      "loss": 0.4407,
      "step": 31086
    },
    {
      "epoch": 1.7407884421547766,
      "grad_norm": 1.3969019651412964,
      "learning_rate": 9.708605263157894e-05,
      "loss": 0.4745,
      "step": 31087
    },
    {
      "epoch": 1.7408444394669056,
      "grad_norm": 1.1291042566299438,
      "learning_rate": 9.708578947368422e-05,
      "loss": 0.3991,
      "step": 31088
    },
    {
      "epoch": 1.7409004367790346,
      "grad_norm": 1.3396730422973633,
      "learning_rate": 9.708552631578948e-05,
      "loss": 0.4702,
      "step": 31089
    },
    {
      "epoch": 1.7409564340911636,
      "grad_norm": 1.233237385749817,
      "learning_rate": 9.708526315789475e-05,
      "loss": 0.4785,
      "step": 31090
    },
    {
      "epoch": 1.7410124314032926,
      "grad_norm": 1.2310079336166382,
      "learning_rate": 9.708500000000001e-05,
      "loss": 0.3986,
      "step": 31091
    },
    {
      "epoch": 1.7410684287154217,
      "grad_norm": 1.2281066179275513,
      "learning_rate": 9.708473684210527e-05,
      "loss": 0.3578,
      "step": 31092
    },
    {
      "epoch": 1.7411244260275507,
      "grad_norm": 1.6963021755218506,
      "learning_rate": 9.708447368421053e-05,
      "loss": 0.7621,
      "step": 31093
    },
    {
      "epoch": 1.7411804233396797,
      "grad_norm": 1.2462592124938965,
      "learning_rate": 9.708421052631579e-05,
      "loss": 0.5078,
      "step": 31094
    },
    {
      "epoch": 1.7412364206518087,
      "grad_norm": 1.1407694816589355,
      "learning_rate": 9.708394736842106e-05,
      "loss": 0.464,
      "step": 31095
    },
    {
      "epoch": 1.7412924179639377,
      "grad_norm": 1.5024384260177612,
      "learning_rate": 9.708368421052632e-05,
      "loss": 0.354,
      "step": 31096
    },
    {
      "epoch": 1.7413484152760668,
      "grad_norm": 1.2973623275756836,
      "learning_rate": 9.708342105263158e-05,
      "loss": 0.3644,
      "step": 31097
    },
    {
      "epoch": 1.7414044125881958,
      "grad_norm": 1.157495141029358,
      "learning_rate": 9.708315789473684e-05,
      "loss": 0.4756,
      "step": 31098
    },
    {
      "epoch": 1.7414604099003248,
      "grad_norm": 1.5629620552062988,
      "learning_rate": 9.708289473684211e-05,
      "loss": 0.5234,
      "step": 31099
    },
    {
      "epoch": 1.7415164072124538,
      "grad_norm": 1.3443243503570557,
      "learning_rate": 9.708263157894737e-05,
      "loss": 0.456,
      "step": 31100
    },
    {
      "epoch": 1.7415724045245828,
      "grad_norm": 1.2040119171142578,
      "learning_rate": 9.708236842105263e-05,
      "loss": 0.3734,
      "step": 31101
    },
    {
      "epoch": 1.7416284018367119,
      "grad_norm": 1.5180034637451172,
      "learning_rate": 9.70821052631579e-05,
      "loss": 0.4813,
      "step": 31102
    },
    {
      "epoch": 1.7416843991488409,
      "grad_norm": 1.3192877769470215,
      "learning_rate": 9.708184210526317e-05,
      "loss": 0.4648,
      "step": 31103
    },
    {
      "epoch": 1.74174039646097,
      "grad_norm": 1.2469027042388916,
      "learning_rate": 9.708157894736843e-05,
      "loss": 0.4622,
      "step": 31104
    },
    {
      "epoch": 1.741796393773099,
      "grad_norm": 1.5461679697036743,
      "learning_rate": 9.708131578947369e-05,
      "loss": 0.3631,
      "step": 31105
    },
    {
      "epoch": 1.741852391085228,
      "grad_norm": 1.1623367071151733,
      "learning_rate": 9.708105263157895e-05,
      "loss": 0.5347,
      "step": 31106
    },
    {
      "epoch": 1.741908388397357,
      "grad_norm": 1.2339093685150146,
      "learning_rate": 9.708078947368422e-05,
      "loss": 0.4605,
      "step": 31107
    },
    {
      "epoch": 1.741964385709486,
      "grad_norm": 1.197686791419983,
      "learning_rate": 9.708052631578948e-05,
      "loss": 0.5984,
      "step": 31108
    },
    {
      "epoch": 1.742020383021615,
      "grad_norm": 1.1987754106521606,
      "learning_rate": 9.708026315789475e-05,
      "loss": 0.3691,
      "step": 31109
    },
    {
      "epoch": 1.742076380333744,
      "grad_norm": 1.3697737455368042,
      "learning_rate": 9.708e-05,
      "loss": 0.446,
      "step": 31110
    },
    {
      "epoch": 1.742132377645873,
      "grad_norm": 1.2985669374465942,
      "learning_rate": 9.707973684210527e-05,
      "loss": 0.4275,
      "step": 31111
    },
    {
      "epoch": 1.742188374958002,
      "grad_norm": 1.3137775659561157,
      "learning_rate": 9.707947368421053e-05,
      "loss": 0.4253,
      "step": 31112
    },
    {
      "epoch": 1.742244372270131,
      "grad_norm": 1.1247609853744507,
      "learning_rate": 9.707921052631579e-05,
      "loss": 0.4312,
      "step": 31113
    },
    {
      "epoch": 1.74230036958226,
      "grad_norm": 1.3682801723480225,
      "learning_rate": 9.707894736842106e-05,
      "loss": 0.5176,
      "step": 31114
    },
    {
      "epoch": 1.742356366894389,
      "grad_norm": 1.2950327396392822,
      "learning_rate": 9.707868421052631e-05,
      "loss": 0.4822,
      "step": 31115
    },
    {
      "epoch": 1.7424123642065181,
      "grad_norm": 1.2705243825912476,
      "learning_rate": 9.707842105263158e-05,
      "loss": 0.4913,
      "step": 31116
    },
    {
      "epoch": 1.7424683615186471,
      "grad_norm": 1.3573381900787354,
      "learning_rate": 9.707815789473684e-05,
      "loss": 0.4824,
      "step": 31117
    },
    {
      "epoch": 1.7425243588307762,
      "grad_norm": 1.2273638248443604,
      "learning_rate": 9.707789473684212e-05,
      "loss": 0.4039,
      "step": 31118
    },
    {
      "epoch": 1.7425803561429052,
      "grad_norm": 1.455743670463562,
      "learning_rate": 9.707763157894736e-05,
      "loss": 0.4769,
      "step": 31119
    },
    {
      "epoch": 1.7426363534550342,
      "grad_norm": 1.5062799453735352,
      "learning_rate": 9.707736842105264e-05,
      "loss": 0.3431,
      "step": 31120
    },
    {
      "epoch": 1.7426923507671632,
      "grad_norm": 1.1785157918930054,
      "learning_rate": 9.70771052631579e-05,
      "loss": 0.6031,
      "step": 31121
    },
    {
      "epoch": 1.7427483480792922,
      "grad_norm": 1.2476261854171753,
      "learning_rate": 9.707684210526317e-05,
      "loss": 0.3797,
      "step": 31122
    },
    {
      "epoch": 1.7428043453914213,
      "grad_norm": 1.257611870765686,
      "learning_rate": 9.707657894736843e-05,
      "loss": 0.4104,
      "step": 31123
    },
    {
      "epoch": 1.7428603427035503,
      "grad_norm": 1.2240301370620728,
      "learning_rate": 9.707631578947369e-05,
      "loss": 0.4853,
      "step": 31124
    },
    {
      "epoch": 1.7429163400156793,
      "grad_norm": 1.4875454902648926,
      "learning_rate": 9.707605263157895e-05,
      "loss": 0.4447,
      "step": 31125
    },
    {
      "epoch": 1.7429723373278083,
      "grad_norm": 1.1892359256744385,
      "learning_rate": 9.707578947368422e-05,
      "loss": 0.4327,
      "step": 31126
    },
    {
      "epoch": 1.7430283346399373,
      "grad_norm": 1.1689327955245972,
      "learning_rate": 9.707552631578948e-05,
      "loss": 0.3766,
      "step": 31127
    },
    {
      "epoch": 1.7430843319520664,
      "grad_norm": 1.5304299592971802,
      "learning_rate": 9.707526315789474e-05,
      "loss": 0.6333,
      "step": 31128
    },
    {
      "epoch": 1.7431403292641954,
      "grad_norm": 1.2294846773147583,
      "learning_rate": 9.7075e-05,
      "loss": 0.4744,
      "step": 31129
    },
    {
      "epoch": 1.7431963265763244,
      "grad_norm": 1.062276005744934,
      "learning_rate": 9.707473684210526e-05,
      "loss": 0.3565,
      "step": 31130
    },
    {
      "epoch": 1.7432523238884534,
      "grad_norm": 1.1798160076141357,
      "learning_rate": 9.707447368421053e-05,
      "loss": 0.3753,
      "step": 31131
    },
    {
      "epoch": 1.7433083212005824,
      "grad_norm": 1.4706741571426392,
      "learning_rate": 9.70742105263158e-05,
      "loss": 0.6998,
      "step": 31132
    },
    {
      "epoch": 1.7433643185127115,
      "grad_norm": 1.335087776184082,
      "learning_rate": 9.707394736842105e-05,
      "loss": 0.4379,
      "step": 31133
    },
    {
      "epoch": 1.7434203158248405,
      "grad_norm": 1.7656105756759644,
      "learning_rate": 9.707368421052631e-05,
      "loss": 0.4847,
      "step": 31134
    },
    {
      "epoch": 1.7434763131369695,
      "grad_norm": 1.0609756708145142,
      "learning_rate": 9.707342105263159e-05,
      "loss": 0.3443,
      "step": 31135
    },
    {
      "epoch": 1.7435323104490985,
      "grad_norm": 1.1438069343566895,
      "learning_rate": 9.707315789473685e-05,
      "loss": 0.4442,
      "step": 31136
    },
    {
      "epoch": 1.7435883077612275,
      "grad_norm": 1.1084822416305542,
      "learning_rate": 9.70728947368421e-05,
      "loss": 0.4142,
      "step": 31137
    },
    {
      "epoch": 1.7436443050733565,
      "grad_norm": 1.147504210472107,
      "learning_rate": 9.707263157894737e-05,
      "loss": 0.3977,
      "step": 31138
    },
    {
      "epoch": 1.7437003023854856,
      "grad_norm": 1.682477355003357,
      "learning_rate": 9.707236842105264e-05,
      "loss": 0.4326,
      "step": 31139
    },
    {
      "epoch": 1.7437562996976146,
      "grad_norm": 1.200075387954712,
      "learning_rate": 9.70721052631579e-05,
      "loss": 0.3948,
      "step": 31140
    },
    {
      "epoch": 1.7438122970097436,
      "grad_norm": 1.2038331031799316,
      "learning_rate": 9.707184210526317e-05,
      "loss": 0.3892,
      "step": 31141
    },
    {
      "epoch": 1.7438682943218726,
      "grad_norm": 1.3430256843566895,
      "learning_rate": 9.707157894736842e-05,
      "loss": 0.499,
      "step": 31142
    },
    {
      "epoch": 1.7439242916340016,
      "grad_norm": 1.3392972946166992,
      "learning_rate": 9.707131578947369e-05,
      "loss": 0.5093,
      "step": 31143
    },
    {
      "epoch": 1.7439802889461307,
      "grad_norm": 1.2368052005767822,
      "learning_rate": 9.707105263157895e-05,
      "loss": 0.3795,
      "step": 31144
    },
    {
      "epoch": 1.7440362862582597,
      "grad_norm": 1.3254647254943848,
      "learning_rate": 9.707078947368422e-05,
      "loss": 0.41,
      "step": 31145
    },
    {
      "epoch": 1.7440922835703887,
      "grad_norm": 1.5145714282989502,
      "learning_rate": 9.707052631578948e-05,
      "loss": 0.4794,
      "step": 31146
    },
    {
      "epoch": 1.7441482808825177,
      "grad_norm": 1.277005910873413,
      "learning_rate": 9.707026315789473e-05,
      "loss": 0.4421,
      "step": 31147
    },
    {
      "epoch": 1.7442042781946467,
      "grad_norm": 1.0948920249938965,
      "learning_rate": 9.707e-05,
      "loss": 0.3123,
      "step": 31148
    },
    {
      "epoch": 1.7442602755067758,
      "grad_norm": 1.3008092641830444,
      "learning_rate": 9.706973684210526e-05,
      "loss": 0.5022,
      "step": 31149
    },
    {
      "epoch": 1.7443162728189048,
      "grad_norm": 1.3784140348434448,
      "learning_rate": 9.706947368421054e-05,
      "loss": 0.4771,
      "step": 31150
    },
    {
      "epoch": 1.7443722701310338,
      "grad_norm": 1.32255220413208,
      "learning_rate": 9.70692105263158e-05,
      "loss": 0.4237,
      "step": 31151
    },
    {
      "epoch": 1.7444282674431628,
      "grad_norm": 1.2068743705749512,
      "learning_rate": 9.706894736842106e-05,
      "loss": 0.4902,
      "step": 31152
    },
    {
      "epoch": 1.7444842647552918,
      "grad_norm": 1.5408138036727905,
      "learning_rate": 9.706868421052632e-05,
      "loss": 0.5773,
      "step": 31153
    },
    {
      "epoch": 1.7445402620674209,
      "grad_norm": 2.509183883666992,
      "learning_rate": 9.706842105263159e-05,
      "loss": 0.5349,
      "step": 31154
    },
    {
      "epoch": 1.7445962593795499,
      "grad_norm": 1.1979178190231323,
      "learning_rate": 9.706815789473685e-05,
      "loss": 0.3273,
      "step": 31155
    },
    {
      "epoch": 1.744652256691679,
      "grad_norm": 1.3609654903411865,
      "learning_rate": 9.706789473684211e-05,
      "loss": 0.4675,
      "step": 31156
    },
    {
      "epoch": 1.744708254003808,
      "grad_norm": 1.5484447479248047,
      "learning_rate": 9.706763157894737e-05,
      "loss": 0.5337,
      "step": 31157
    },
    {
      "epoch": 1.744764251315937,
      "grad_norm": 1.31448495388031,
      "learning_rate": 9.706736842105264e-05,
      "loss": 0.4891,
      "step": 31158
    },
    {
      "epoch": 1.744820248628066,
      "grad_norm": 1.3487775325775146,
      "learning_rate": 9.70671052631579e-05,
      "loss": 0.553,
      "step": 31159
    },
    {
      "epoch": 1.744876245940195,
      "grad_norm": 1.3651710748672485,
      "learning_rate": 9.706684210526316e-05,
      "loss": 0.6034,
      "step": 31160
    },
    {
      "epoch": 1.744932243252324,
      "grad_norm": 1.2114300727844238,
      "learning_rate": 9.706657894736842e-05,
      "loss": 0.3901,
      "step": 31161
    },
    {
      "epoch": 1.744988240564453,
      "grad_norm": 1.1772379875183105,
      "learning_rate": 9.70663157894737e-05,
      "loss": 0.4121,
      "step": 31162
    },
    {
      "epoch": 1.745044237876582,
      "grad_norm": 1.1749367713928223,
      "learning_rate": 9.706605263157895e-05,
      "loss": 0.4912,
      "step": 31163
    },
    {
      "epoch": 1.745100235188711,
      "grad_norm": 1.1842727661132812,
      "learning_rate": 9.706578947368421e-05,
      "loss": 0.4274,
      "step": 31164
    },
    {
      "epoch": 1.74515623250084,
      "grad_norm": 1.3142340183258057,
      "learning_rate": 9.706552631578947e-05,
      "loss": 0.5152,
      "step": 31165
    },
    {
      "epoch": 1.745212229812969,
      "grad_norm": 1.4865119457244873,
      "learning_rate": 9.706526315789473e-05,
      "loss": 0.486,
      "step": 31166
    },
    {
      "epoch": 1.745268227125098,
      "grad_norm": 1.3705570697784424,
      "learning_rate": 9.7065e-05,
      "loss": 0.5238,
      "step": 31167
    },
    {
      "epoch": 1.7453242244372271,
      "grad_norm": 1.315129041671753,
      "learning_rate": 9.706473684210527e-05,
      "loss": 0.5359,
      "step": 31168
    },
    {
      "epoch": 1.7453802217493561,
      "grad_norm": 1.2519727945327759,
      "learning_rate": 9.706447368421054e-05,
      "loss": 0.3968,
      "step": 31169
    },
    {
      "epoch": 1.7454362190614852,
      "grad_norm": 1.2539435625076294,
      "learning_rate": 9.706421052631579e-05,
      "loss": 0.3755,
      "step": 31170
    },
    {
      "epoch": 1.7454922163736142,
      "grad_norm": 1.0977662801742554,
      "learning_rate": 9.706394736842106e-05,
      "loss": 0.4074,
      "step": 31171
    },
    {
      "epoch": 1.7455482136857432,
      "grad_norm": 1.293164849281311,
      "learning_rate": 9.706368421052632e-05,
      "loss": 0.3691,
      "step": 31172
    },
    {
      "epoch": 1.7456042109978722,
      "grad_norm": 1.315238356590271,
      "learning_rate": 9.706342105263159e-05,
      "loss": 0.4713,
      "step": 31173
    },
    {
      "epoch": 1.7456602083100012,
      "grad_norm": 1.282739520072937,
      "learning_rate": 9.706315789473684e-05,
      "loss": 0.3478,
      "step": 31174
    },
    {
      "epoch": 1.7457162056221303,
      "grad_norm": 1.2086079120635986,
      "learning_rate": 9.706289473684211e-05,
      "loss": 0.5047,
      "step": 31175
    },
    {
      "epoch": 1.7457722029342593,
      "grad_norm": 2.1796000003814697,
      "learning_rate": 9.706263157894737e-05,
      "loss": 0.5282,
      "step": 31176
    },
    {
      "epoch": 1.7458282002463883,
      "grad_norm": 1.298432469367981,
      "learning_rate": 9.706236842105264e-05,
      "loss": 0.4878,
      "step": 31177
    },
    {
      "epoch": 1.7458841975585173,
      "grad_norm": 1.4929630756378174,
      "learning_rate": 9.70621052631579e-05,
      "loss": 0.505,
      "step": 31178
    },
    {
      "epoch": 1.7459401948706463,
      "grad_norm": 1.1581120491027832,
      "learning_rate": 9.706184210526316e-05,
      "loss": 0.3978,
      "step": 31179
    },
    {
      "epoch": 1.7459961921827754,
      "grad_norm": 1.0985528230667114,
      "learning_rate": 9.706157894736842e-05,
      "loss": 0.4282,
      "step": 31180
    },
    {
      "epoch": 1.7460521894949044,
      "grad_norm": 1.657619595527649,
      "learning_rate": 9.706131578947368e-05,
      "loss": 0.7524,
      "step": 31181
    },
    {
      "epoch": 1.7461081868070334,
      "grad_norm": 1.113402009010315,
      "learning_rate": 9.706105263157896e-05,
      "loss": 0.3417,
      "step": 31182
    },
    {
      "epoch": 1.7461641841191624,
      "grad_norm": 1.1263939142227173,
      "learning_rate": 9.706078947368422e-05,
      "loss": 0.4483,
      "step": 31183
    },
    {
      "epoch": 1.7462201814312914,
      "grad_norm": 1.3888018131256104,
      "learning_rate": 9.706052631578948e-05,
      "loss": 0.6088,
      "step": 31184
    },
    {
      "epoch": 1.7462761787434204,
      "grad_norm": 1.1755828857421875,
      "learning_rate": 9.706026315789474e-05,
      "loss": 0.423,
      "step": 31185
    },
    {
      "epoch": 1.7463321760555495,
      "grad_norm": 1.1729487180709839,
      "learning_rate": 9.706000000000001e-05,
      "loss": 0.4509,
      "step": 31186
    },
    {
      "epoch": 1.7463881733676785,
      "grad_norm": 1.492467999458313,
      "learning_rate": 9.705973684210527e-05,
      "loss": 0.5417,
      "step": 31187
    },
    {
      "epoch": 1.7464441706798075,
      "grad_norm": 1.4126147031784058,
      "learning_rate": 9.705947368421053e-05,
      "loss": 0.5848,
      "step": 31188
    },
    {
      "epoch": 1.7465001679919365,
      "grad_norm": 1.270298719406128,
      "learning_rate": 9.705921052631579e-05,
      "loss": 0.473,
      "step": 31189
    },
    {
      "epoch": 1.7465561653040655,
      "grad_norm": 1.3339388370513916,
      "learning_rate": 9.705894736842106e-05,
      "loss": 0.5668,
      "step": 31190
    },
    {
      "epoch": 1.7466121626161946,
      "grad_norm": 1.4679417610168457,
      "learning_rate": 9.705868421052632e-05,
      "loss": 0.6043,
      "step": 31191
    },
    {
      "epoch": 1.7466681599283236,
      "grad_norm": 1.3213615417480469,
      "learning_rate": 9.705842105263158e-05,
      "loss": 0.4504,
      "step": 31192
    },
    {
      "epoch": 1.7467241572404526,
      "grad_norm": 1.5918020009994507,
      "learning_rate": 9.705815789473684e-05,
      "loss": 0.5611,
      "step": 31193
    },
    {
      "epoch": 1.7467801545525816,
      "grad_norm": 1.4029831886291504,
      "learning_rate": 9.705789473684211e-05,
      "loss": 0.3647,
      "step": 31194
    },
    {
      "epoch": 1.7468361518647106,
      "grad_norm": 1.5055946111679077,
      "learning_rate": 9.705763157894737e-05,
      "loss": 0.5016,
      "step": 31195
    },
    {
      "epoch": 1.7468921491768397,
      "grad_norm": 1.0396732091903687,
      "learning_rate": 9.705736842105265e-05,
      "loss": 0.3701,
      "step": 31196
    },
    {
      "epoch": 1.7469481464889687,
      "grad_norm": 1.2922356128692627,
      "learning_rate": 9.70571052631579e-05,
      "loss": 0.5646,
      "step": 31197
    },
    {
      "epoch": 1.7470041438010977,
      "grad_norm": 1.3413194417953491,
      "learning_rate": 9.705684210526315e-05,
      "loss": 0.4088,
      "step": 31198
    },
    {
      "epoch": 1.7470601411132267,
      "grad_norm": 1.216722846031189,
      "learning_rate": 9.705657894736843e-05,
      "loss": 0.5008,
      "step": 31199
    },
    {
      "epoch": 1.7471161384253557,
      "grad_norm": 1.3155068159103394,
      "learning_rate": 9.705631578947369e-05,
      "loss": 0.4174,
      "step": 31200
    },
    {
      "epoch": 1.7471721357374848,
      "grad_norm": 1.207533836364746,
      "learning_rate": 9.705605263157896e-05,
      "loss": 0.5191,
      "step": 31201
    },
    {
      "epoch": 1.7472281330496138,
      "grad_norm": 1.6117228269577026,
      "learning_rate": 9.70557894736842e-05,
      "loss": 0.5692,
      "step": 31202
    },
    {
      "epoch": 1.7472841303617428,
      "grad_norm": 1.204028844833374,
      "learning_rate": 9.705552631578948e-05,
      "loss": 0.4536,
      "step": 31203
    },
    {
      "epoch": 1.7473401276738718,
      "grad_norm": 1.4153550863265991,
      "learning_rate": 9.705526315789474e-05,
      "loss": 0.438,
      "step": 31204
    },
    {
      "epoch": 1.7473961249860008,
      "grad_norm": 1.072204828262329,
      "learning_rate": 9.705500000000001e-05,
      "loss": 0.3561,
      "step": 31205
    },
    {
      "epoch": 1.7474521222981299,
      "grad_norm": 1.1738938093185425,
      "learning_rate": 9.705473684210527e-05,
      "loss": 0.411,
      "step": 31206
    },
    {
      "epoch": 1.7475081196102589,
      "grad_norm": 1.049439549446106,
      "learning_rate": 9.705447368421053e-05,
      "loss": 0.3586,
      "step": 31207
    },
    {
      "epoch": 1.747564116922388,
      "grad_norm": 1.2313178777694702,
      "learning_rate": 9.705421052631579e-05,
      "loss": 0.4807,
      "step": 31208
    },
    {
      "epoch": 1.747620114234517,
      "grad_norm": 1.2798324823379517,
      "learning_rate": 9.705394736842106e-05,
      "loss": 0.3847,
      "step": 31209
    },
    {
      "epoch": 1.747676111546646,
      "grad_norm": 1.2703906297683716,
      "learning_rate": 9.705368421052632e-05,
      "loss": 0.4997,
      "step": 31210
    },
    {
      "epoch": 1.7477321088587747,
      "grad_norm": 1.3671642541885376,
      "learning_rate": 9.705342105263158e-05,
      "loss": 0.4343,
      "step": 31211
    },
    {
      "epoch": 1.7477881061709037,
      "grad_norm": 1.385200023651123,
      "learning_rate": 9.705315789473684e-05,
      "loss": 0.478,
      "step": 31212
    },
    {
      "epoch": 1.7478441034830328,
      "grad_norm": 1.0814567804336548,
      "learning_rate": 9.705289473684212e-05,
      "loss": 0.3617,
      "step": 31213
    },
    {
      "epoch": 1.7479001007951618,
      "grad_norm": 1.0792593955993652,
      "learning_rate": 9.705263157894738e-05,
      "loss": 0.3795,
      "step": 31214
    },
    {
      "epoch": 1.7479560981072908,
      "grad_norm": 1.7391676902770996,
      "learning_rate": 9.705236842105264e-05,
      "loss": 0.4514,
      "step": 31215
    },
    {
      "epoch": 1.7480120954194198,
      "grad_norm": 1.5550775527954102,
      "learning_rate": 9.70521052631579e-05,
      "loss": 0.6263,
      "step": 31216
    },
    {
      "epoch": 1.7480680927315488,
      "grad_norm": 1.538326382637024,
      "learning_rate": 9.705184210526316e-05,
      "loss": 0.5031,
      "step": 31217
    },
    {
      "epoch": 1.7481240900436779,
      "grad_norm": 1.129977822303772,
      "learning_rate": 9.705157894736843e-05,
      "loss": 0.3993,
      "step": 31218
    },
    {
      "epoch": 1.7481800873558069,
      "grad_norm": 1.1480122804641724,
      "learning_rate": 9.705131578947369e-05,
      "loss": 0.4013,
      "step": 31219
    },
    {
      "epoch": 1.748236084667936,
      "grad_norm": 1.175819754600525,
      "learning_rate": 9.705105263157895e-05,
      "loss": 0.3819,
      "step": 31220
    },
    {
      "epoch": 1.748292081980065,
      "grad_norm": 1.3679624795913696,
      "learning_rate": 9.705078947368421e-05,
      "loss": 0.442,
      "step": 31221
    },
    {
      "epoch": 1.748348079292194,
      "grad_norm": 1.1933364868164062,
      "learning_rate": 9.705052631578948e-05,
      "loss": 0.4248,
      "step": 31222
    },
    {
      "epoch": 1.748404076604323,
      "grad_norm": 1.593381404876709,
      "learning_rate": 9.705026315789474e-05,
      "loss": 0.3883,
      "step": 31223
    },
    {
      "epoch": 1.748460073916452,
      "grad_norm": 1.7808988094329834,
      "learning_rate": 9.705e-05,
      "loss": 0.517,
      "step": 31224
    },
    {
      "epoch": 1.748516071228581,
      "grad_norm": 1.6892181634902954,
      "learning_rate": 9.704973684210526e-05,
      "loss": 0.5065,
      "step": 31225
    },
    {
      "epoch": 1.74857206854071,
      "grad_norm": 1.612975001335144,
      "learning_rate": 9.704947368421053e-05,
      "loss": 0.5153,
      "step": 31226
    },
    {
      "epoch": 1.748628065852839,
      "grad_norm": 1.4767022132873535,
      "learning_rate": 9.70492105263158e-05,
      "loss": 0.4598,
      "step": 31227
    },
    {
      "epoch": 1.748684063164968,
      "grad_norm": 1.2587836980819702,
      "learning_rate": 9.704894736842107e-05,
      "loss": 0.4505,
      "step": 31228
    },
    {
      "epoch": 1.748740060477097,
      "grad_norm": 1.2305335998535156,
      "learning_rate": 9.704868421052631e-05,
      "loss": 0.4602,
      "step": 31229
    },
    {
      "epoch": 1.748796057789226,
      "grad_norm": 1.1922690868377686,
      "learning_rate": 9.704842105263159e-05,
      "loss": 0.5075,
      "step": 31230
    },
    {
      "epoch": 1.748852055101355,
      "grad_norm": 1.8206361532211304,
      "learning_rate": 9.704815789473685e-05,
      "loss": 0.482,
      "step": 31231
    },
    {
      "epoch": 1.7489080524134841,
      "grad_norm": 1.375199556350708,
      "learning_rate": 9.704789473684212e-05,
      "loss": 0.4182,
      "step": 31232
    },
    {
      "epoch": 1.7489640497256131,
      "grad_norm": 1.3915208578109741,
      "learning_rate": 9.704763157894738e-05,
      "loss": 0.4135,
      "step": 31233
    },
    {
      "epoch": 1.7490200470377422,
      "grad_norm": 1.5195192098617554,
      "learning_rate": 9.704736842105263e-05,
      "loss": 0.4485,
      "step": 31234
    },
    {
      "epoch": 1.7490760443498712,
      "grad_norm": 1.5032994747161865,
      "learning_rate": 9.70471052631579e-05,
      "loss": 0.4478,
      "step": 31235
    },
    {
      "epoch": 1.7491320416620002,
      "grad_norm": 1.0254724025726318,
      "learning_rate": 9.704684210526316e-05,
      "loss": 0.3381,
      "step": 31236
    },
    {
      "epoch": 1.7491880389741292,
      "grad_norm": 1.2206236124038696,
      "learning_rate": 9.704657894736843e-05,
      "loss": 0.4974,
      "step": 31237
    },
    {
      "epoch": 1.7492440362862582,
      "grad_norm": 1.2313346862792969,
      "learning_rate": 9.704631578947369e-05,
      "loss": 0.3825,
      "step": 31238
    },
    {
      "epoch": 1.7493000335983873,
      "grad_norm": 1.2586702108383179,
      "learning_rate": 9.704605263157895e-05,
      "loss": 0.3527,
      "step": 31239
    },
    {
      "epoch": 1.7493560309105163,
      "grad_norm": 1.4547477960586548,
      "learning_rate": 9.704578947368421e-05,
      "loss": 0.5049,
      "step": 31240
    },
    {
      "epoch": 1.7494120282226453,
      "grad_norm": 1.1846121549606323,
      "learning_rate": 9.704552631578948e-05,
      "loss": 0.3511,
      "step": 31241
    },
    {
      "epoch": 1.7494680255347743,
      "grad_norm": 1.0702062845230103,
      "learning_rate": 9.704526315789474e-05,
      "loss": 0.3378,
      "step": 31242
    },
    {
      "epoch": 1.7495240228469033,
      "grad_norm": 1.2872754335403442,
      "learning_rate": 9.7045e-05,
      "loss": 0.3932,
      "step": 31243
    },
    {
      "epoch": 1.7495800201590324,
      "grad_norm": 1.1747994422912598,
      "learning_rate": 9.704473684210526e-05,
      "loss": 0.4573,
      "step": 31244
    },
    {
      "epoch": 1.7496360174711614,
      "grad_norm": 1.3304767608642578,
      "learning_rate": 9.704447368421054e-05,
      "loss": 0.4008,
      "step": 31245
    },
    {
      "epoch": 1.7496920147832904,
      "grad_norm": 1.310982584953308,
      "learning_rate": 9.70442105263158e-05,
      "loss": 0.4587,
      "step": 31246
    },
    {
      "epoch": 1.7497480120954194,
      "grad_norm": 1.4259971380233765,
      "learning_rate": 9.704394736842106e-05,
      "loss": 0.4796,
      "step": 31247
    },
    {
      "epoch": 1.7498040094075484,
      "grad_norm": 1.1854883432388306,
      "learning_rate": 9.704368421052632e-05,
      "loss": 0.5095,
      "step": 31248
    },
    {
      "epoch": 1.7498600067196775,
      "grad_norm": 1.2379854917526245,
      "learning_rate": 9.704342105263159e-05,
      "loss": 0.4267,
      "step": 31249
    },
    {
      "epoch": 1.7499160040318065,
      "grad_norm": 1.5443087816238403,
      "learning_rate": 9.704315789473685e-05,
      "loss": 0.6044,
      "step": 31250
    },
    {
      "epoch": 1.7499720013439355,
      "grad_norm": 1.7336145639419556,
      "learning_rate": 9.704289473684211e-05,
      "loss": 0.5171,
      "step": 31251
    },
    {
      "epoch": 1.7500279986560645,
      "grad_norm": 1.352027177810669,
      "learning_rate": 9.704263157894737e-05,
      "loss": 0.5536,
      "step": 31252
    },
    {
      "epoch": 1.7500839959681935,
      "grad_norm": 1.3359984159469604,
      "learning_rate": 9.704236842105263e-05,
      "loss": 0.4543,
      "step": 31253
    },
    {
      "epoch": 1.7501399932803225,
      "grad_norm": 1.0503405332565308,
      "learning_rate": 9.70421052631579e-05,
      "loss": 0.3971,
      "step": 31254
    },
    {
      "epoch": 1.7501959905924516,
      "grad_norm": 1.1917916536331177,
      "learning_rate": 9.704184210526316e-05,
      "loss": 0.3951,
      "step": 31255
    },
    {
      "epoch": 1.7502519879045806,
      "grad_norm": 1.2114452123641968,
      "learning_rate": 9.704157894736843e-05,
      "loss": 0.4025,
      "step": 31256
    },
    {
      "epoch": 1.7503079852167096,
      "grad_norm": 1.156714677810669,
      "learning_rate": 9.704131578947368e-05,
      "loss": 0.4568,
      "step": 31257
    },
    {
      "epoch": 1.7503639825288386,
      "grad_norm": 1.41642165184021,
      "learning_rate": 9.704105263157895e-05,
      "loss": 0.5325,
      "step": 31258
    },
    {
      "epoch": 1.7504199798409676,
      "grad_norm": 1.2985272407531738,
      "learning_rate": 9.704078947368421e-05,
      "loss": 0.4317,
      "step": 31259
    },
    {
      "epoch": 1.7504759771530967,
      "grad_norm": 1.1710045337677002,
      "learning_rate": 9.704052631578949e-05,
      "loss": 0.4385,
      "step": 31260
    },
    {
      "epoch": 1.7505319744652257,
      "grad_norm": 1.4962447881698608,
      "learning_rate": 9.704026315789475e-05,
      "loss": 0.4705,
      "step": 31261
    },
    {
      "epoch": 1.7505879717773547,
      "grad_norm": 1.2892462015151978,
      "learning_rate": 9.704e-05,
      "loss": 0.4027,
      "step": 31262
    },
    {
      "epoch": 1.7506439690894837,
      "grad_norm": 1.217913031578064,
      "learning_rate": 9.703973684210527e-05,
      "loss": 0.5367,
      "step": 31263
    },
    {
      "epoch": 1.7506999664016127,
      "grad_norm": 1.1747641563415527,
      "learning_rate": 9.703947368421054e-05,
      "loss": 0.4703,
      "step": 31264
    },
    {
      "epoch": 1.7507559637137418,
      "grad_norm": 1.229805588722229,
      "learning_rate": 9.70392105263158e-05,
      "loss": 0.3878,
      "step": 31265
    },
    {
      "epoch": 1.7508119610258708,
      "grad_norm": 1.64512038230896,
      "learning_rate": 9.703894736842106e-05,
      "loss": 0.5059,
      "step": 31266
    },
    {
      "epoch": 1.7508679583379998,
      "grad_norm": 1.3795711994171143,
      "learning_rate": 9.703868421052632e-05,
      "loss": 0.59,
      "step": 31267
    },
    {
      "epoch": 1.7509239556501288,
      "grad_norm": 1.3502846956253052,
      "learning_rate": 9.703842105263158e-05,
      "loss": 0.5191,
      "step": 31268
    },
    {
      "epoch": 1.7509799529622578,
      "grad_norm": 1.1269354820251465,
      "learning_rate": 9.703815789473685e-05,
      "loss": 0.4288,
      "step": 31269
    },
    {
      "epoch": 1.7510359502743869,
      "grad_norm": 1.2391060590744019,
      "learning_rate": 9.703789473684211e-05,
      "loss": 0.3725,
      "step": 31270
    },
    {
      "epoch": 1.7510919475865159,
      "grad_norm": 1.2322804927825928,
      "learning_rate": 9.703763157894737e-05,
      "loss": 0.5085,
      "step": 31271
    },
    {
      "epoch": 1.751147944898645,
      "grad_norm": 1.2614835500717163,
      "learning_rate": 9.703736842105263e-05,
      "loss": 0.4143,
      "step": 31272
    },
    {
      "epoch": 1.751203942210774,
      "grad_norm": 1.2989394664764404,
      "learning_rate": 9.70371052631579e-05,
      "loss": 0.5407,
      "step": 31273
    },
    {
      "epoch": 1.751259939522903,
      "grad_norm": 1.300110101699829,
      "learning_rate": 9.703684210526316e-05,
      "loss": 0.5379,
      "step": 31274
    },
    {
      "epoch": 1.751315936835032,
      "grad_norm": 1.4079710245132446,
      "learning_rate": 9.703657894736842e-05,
      "loss": 0.4106,
      "step": 31275
    },
    {
      "epoch": 1.751371934147161,
      "grad_norm": 1.0841021537780762,
      "learning_rate": 9.703631578947368e-05,
      "loss": 0.3562,
      "step": 31276
    },
    {
      "epoch": 1.75142793145929,
      "grad_norm": 41.36603927612305,
      "learning_rate": 9.703605263157896e-05,
      "loss": 0.4181,
      "step": 31277
    },
    {
      "epoch": 1.751483928771419,
      "grad_norm": 1.1688511371612549,
      "learning_rate": 9.703578947368422e-05,
      "loss": 0.3864,
      "step": 31278
    },
    {
      "epoch": 1.751539926083548,
      "grad_norm": 1.1154115200042725,
      "learning_rate": 9.703552631578948e-05,
      "loss": 0.3512,
      "step": 31279
    },
    {
      "epoch": 1.751595923395677,
      "grad_norm": 1.5570732355117798,
      "learning_rate": 9.703526315789474e-05,
      "loss": 0.5818,
      "step": 31280
    },
    {
      "epoch": 1.751651920707806,
      "grad_norm": 1.2714945077896118,
      "learning_rate": 9.703500000000001e-05,
      "loss": 0.3954,
      "step": 31281
    },
    {
      "epoch": 1.751707918019935,
      "grad_norm": 1.1105530261993408,
      "learning_rate": 9.703473684210527e-05,
      "loss": 0.4644,
      "step": 31282
    },
    {
      "epoch": 1.751763915332064,
      "grad_norm": 1.3542370796203613,
      "learning_rate": 9.703447368421054e-05,
      "loss": 0.4844,
      "step": 31283
    },
    {
      "epoch": 1.7518199126441931,
      "grad_norm": 1.2920011281967163,
      "learning_rate": 9.703421052631579e-05,
      "loss": 0.3696,
      "step": 31284
    },
    {
      "epoch": 1.7518759099563221,
      "grad_norm": 1.2199699878692627,
      "learning_rate": 9.703394736842105e-05,
      "loss": 0.3776,
      "step": 31285
    },
    {
      "epoch": 1.7519319072684512,
      "grad_norm": 1.2068777084350586,
      "learning_rate": 9.703368421052632e-05,
      "loss": 0.4286,
      "step": 31286
    },
    {
      "epoch": 1.7519879045805802,
      "grad_norm": 1.2400809526443481,
      "learning_rate": 9.703342105263158e-05,
      "loss": 0.4208,
      "step": 31287
    },
    {
      "epoch": 1.7520439018927092,
      "grad_norm": 1.454540491104126,
      "learning_rate": 9.703315789473685e-05,
      "loss": 0.5424,
      "step": 31288
    },
    {
      "epoch": 1.7520998992048382,
      "grad_norm": 1.419374942779541,
      "learning_rate": 9.70328947368421e-05,
      "loss": 0.4436,
      "step": 31289
    },
    {
      "epoch": 1.7521558965169672,
      "grad_norm": 1.0190922021865845,
      "learning_rate": 9.703263157894737e-05,
      "loss": 0.4034,
      "step": 31290
    },
    {
      "epoch": 1.7522118938290963,
      "grad_norm": 1.2758339643478394,
      "learning_rate": 9.703236842105263e-05,
      "loss": 0.5161,
      "step": 31291
    },
    {
      "epoch": 1.7522678911412253,
      "grad_norm": 1.1510674953460693,
      "learning_rate": 9.70321052631579e-05,
      "loss": 0.4194,
      "step": 31292
    },
    {
      "epoch": 1.752323888453354,
      "grad_norm": 1.3816848993301392,
      "learning_rate": 9.703184210526317e-05,
      "loss": 0.5782,
      "step": 31293
    },
    {
      "epoch": 1.752379885765483,
      "grad_norm": 1.1880531311035156,
      "learning_rate": 9.703157894736843e-05,
      "loss": 0.4683,
      "step": 31294
    },
    {
      "epoch": 1.752435883077612,
      "grad_norm": 1.1447632312774658,
      "learning_rate": 9.703131578947369e-05,
      "loss": 0.5645,
      "step": 31295
    },
    {
      "epoch": 1.7524918803897411,
      "grad_norm": 1.6359468698501587,
      "learning_rate": 9.703105263157896e-05,
      "loss": 0.6439,
      "step": 31296
    },
    {
      "epoch": 1.7525478777018701,
      "grad_norm": 1.5488758087158203,
      "learning_rate": 9.703078947368422e-05,
      "loss": 0.4205,
      "step": 31297
    },
    {
      "epoch": 1.7526038750139992,
      "grad_norm": 1.2353792190551758,
      "learning_rate": 9.703052631578948e-05,
      "loss": 0.3828,
      "step": 31298
    },
    {
      "epoch": 1.7526598723261282,
      "grad_norm": 1.3690663576126099,
      "learning_rate": 9.703026315789474e-05,
      "loss": 0.4517,
      "step": 31299
    },
    {
      "epoch": 1.7527158696382572,
      "grad_norm": 5.144920825958252,
      "learning_rate": 9.703000000000001e-05,
      "loss": 0.5015,
      "step": 31300
    },
    {
      "epoch": 1.7527718669503862,
      "grad_norm": 1.2079733610153198,
      "learning_rate": 9.702973684210527e-05,
      "loss": 0.5603,
      "step": 31301
    },
    {
      "epoch": 1.7528278642625152,
      "grad_norm": 1.635956048965454,
      "learning_rate": 9.702947368421053e-05,
      "loss": 0.438,
      "step": 31302
    },
    {
      "epoch": 1.7528838615746443,
      "grad_norm": 1.3473219871520996,
      "learning_rate": 9.702921052631579e-05,
      "loss": 0.4354,
      "step": 31303
    },
    {
      "epoch": 1.7529398588867733,
      "grad_norm": 1.3026502132415771,
      "learning_rate": 9.702894736842105e-05,
      "loss": 0.3847,
      "step": 31304
    },
    {
      "epoch": 1.7529958561989023,
      "grad_norm": 1.169182300567627,
      "learning_rate": 9.702868421052632e-05,
      "loss": 0.4121,
      "step": 31305
    },
    {
      "epoch": 1.7530518535110313,
      "grad_norm": 0.9932965040206909,
      "learning_rate": 9.702842105263158e-05,
      "loss": 0.3512,
      "step": 31306
    },
    {
      "epoch": 1.7531078508231603,
      "grad_norm": 1.2884613275527954,
      "learning_rate": 9.702815789473684e-05,
      "loss": 0.373,
      "step": 31307
    },
    {
      "epoch": 1.7531638481352894,
      "grad_norm": 1.0677106380462646,
      "learning_rate": 9.70278947368421e-05,
      "loss": 0.3906,
      "step": 31308
    },
    {
      "epoch": 1.7532198454474184,
      "grad_norm": 1.265514612197876,
      "learning_rate": 9.702763157894738e-05,
      "loss": 0.6163,
      "step": 31309
    },
    {
      "epoch": 1.7532758427595474,
      "grad_norm": 1.2068476676940918,
      "learning_rate": 9.702736842105264e-05,
      "loss": 0.3683,
      "step": 31310
    },
    {
      "epoch": 1.7533318400716764,
      "grad_norm": 1.3675965070724487,
      "learning_rate": 9.702710526315791e-05,
      "loss": 0.4875,
      "step": 31311
    },
    {
      "epoch": 1.7533878373838054,
      "grad_norm": 1.547469139099121,
      "learning_rate": 9.702684210526316e-05,
      "loss": 0.4268,
      "step": 31312
    },
    {
      "epoch": 1.7534438346959345,
      "grad_norm": 1.2892094850540161,
      "learning_rate": 9.702657894736843e-05,
      "loss": 0.6642,
      "step": 31313
    },
    {
      "epoch": 1.7534998320080635,
      "grad_norm": 1.1886186599731445,
      "learning_rate": 9.702631578947369e-05,
      "loss": 0.4739,
      "step": 31314
    },
    {
      "epoch": 1.7535558293201925,
      "grad_norm": 1.0835310220718384,
      "learning_rate": 9.702605263157896e-05,
      "loss": 0.4139,
      "step": 31315
    },
    {
      "epoch": 1.7536118266323215,
      "grad_norm": 1.205816388130188,
      "learning_rate": 9.702578947368422e-05,
      "loss": 0.4162,
      "step": 31316
    },
    {
      "epoch": 1.7536678239444505,
      "grad_norm": 1.6776663064956665,
      "learning_rate": 9.702552631578948e-05,
      "loss": 0.4073,
      "step": 31317
    },
    {
      "epoch": 1.7537238212565796,
      "grad_norm": 4.810730934143066,
      "learning_rate": 9.702526315789474e-05,
      "loss": 0.4364,
      "step": 31318
    },
    {
      "epoch": 1.7537798185687086,
      "grad_norm": 1.4386926889419556,
      "learning_rate": 9.7025e-05,
      "loss": 0.4643,
      "step": 31319
    },
    {
      "epoch": 1.7538358158808376,
      "grad_norm": 1.1380045413970947,
      "learning_rate": 9.702473684210527e-05,
      "loss": 0.4737,
      "step": 31320
    },
    {
      "epoch": 1.7538918131929666,
      "grad_norm": 1.463484764099121,
      "learning_rate": 9.702447368421052e-05,
      "loss": 0.3943,
      "step": 31321
    },
    {
      "epoch": 1.7539478105050956,
      "grad_norm": 1.4343619346618652,
      "learning_rate": 9.702421052631579e-05,
      "loss": 0.3837,
      "step": 31322
    },
    {
      "epoch": 1.7540038078172246,
      "grad_norm": 1.1751424074172974,
      "learning_rate": 9.702394736842105e-05,
      "loss": 0.397,
      "step": 31323
    },
    {
      "epoch": 1.7540598051293537,
      "grad_norm": 1.3465496301651,
      "learning_rate": 9.702368421052633e-05,
      "loss": 0.5806,
      "step": 31324
    },
    {
      "epoch": 1.7541158024414827,
      "grad_norm": 1.3034881353378296,
      "learning_rate": 9.702342105263159e-05,
      "loss": 0.4874,
      "step": 31325
    },
    {
      "epoch": 1.7541717997536117,
      "grad_norm": 1.4874991178512573,
      "learning_rate": 9.702315789473685e-05,
      "loss": 0.492,
      "step": 31326
    },
    {
      "epoch": 1.7542277970657407,
      "grad_norm": 1.558430552482605,
      "learning_rate": 9.70228947368421e-05,
      "loss": 0.5006,
      "step": 31327
    },
    {
      "epoch": 1.7542837943778697,
      "grad_norm": 1.315670132637024,
      "learning_rate": 9.702263157894738e-05,
      "loss": 0.5163,
      "step": 31328
    },
    {
      "epoch": 1.7543397916899988,
      "grad_norm": 1.3746200799942017,
      "learning_rate": 9.702236842105264e-05,
      "loss": 0.4011,
      "step": 31329
    },
    {
      "epoch": 1.7543957890021278,
      "grad_norm": 1.3804941177368164,
      "learning_rate": 9.70221052631579e-05,
      "loss": 0.5946,
      "step": 31330
    },
    {
      "epoch": 1.7544517863142568,
      "grad_norm": 1.374734878540039,
      "learning_rate": 9.702184210526316e-05,
      "loss": 0.3677,
      "step": 31331
    },
    {
      "epoch": 1.7545077836263858,
      "grad_norm": 1.3483905792236328,
      "learning_rate": 9.702157894736843e-05,
      "loss": 0.5115,
      "step": 31332
    },
    {
      "epoch": 1.7545637809385148,
      "grad_norm": 1.217339277267456,
      "learning_rate": 9.702131578947369e-05,
      "loss": 0.4084,
      "step": 31333
    },
    {
      "epoch": 1.7546197782506439,
      "grad_norm": 1.4455842971801758,
      "learning_rate": 9.702105263157895e-05,
      "loss": 0.5797,
      "step": 31334
    },
    {
      "epoch": 1.7546757755627729,
      "grad_norm": 1.247218370437622,
      "learning_rate": 9.702078947368421e-05,
      "loss": 0.3732,
      "step": 31335
    },
    {
      "epoch": 1.754731772874902,
      "grad_norm": 1.3854154348373413,
      "learning_rate": 9.702052631578947e-05,
      "loss": 0.5129,
      "step": 31336
    },
    {
      "epoch": 1.754787770187031,
      "grad_norm": 1.4090889692306519,
      "learning_rate": 9.702026315789474e-05,
      "loss": 0.5575,
      "step": 31337
    },
    {
      "epoch": 1.75484376749916,
      "grad_norm": 1.278238296508789,
      "learning_rate": 9.702e-05,
      "loss": 0.5516,
      "step": 31338
    },
    {
      "epoch": 1.754899764811289,
      "grad_norm": 1.2656220197677612,
      "learning_rate": 9.701973684210526e-05,
      "loss": 0.4156,
      "step": 31339
    },
    {
      "epoch": 1.754955762123418,
      "grad_norm": 1.3355220556259155,
      "learning_rate": 9.701947368421052e-05,
      "loss": 0.4433,
      "step": 31340
    },
    {
      "epoch": 1.755011759435547,
      "grad_norm": 1.4254103899002075,
      "learning_rate": 9.70192105263158e-05,
      "loss": 0.4262,
      "step": 31341
    },
    {
      "epoch": 1.755067756747676,
      "grad_norm": 1.6784453392028809,
      "learning_rate": 9.701894736842106e-05,
      "loss": 0.5253,
      "step": 31342
    },
    {
      "epoch": 1.755123754059805,
      "grad_norm": 1.319821834564209,
      "learning_rate": 9.701868421052633e-05,
      "loss": 0.592,
      "step": 31343
    },
    {
      "epoch": 1.755179751371934,
      "grad_norm": 1.1004695892333984,
      "learning_rate": 9.701842105263157e-05,
      "loss": 0.3536,
      "step": 31344
    },
    {
      "epoch": 1.755235748684063,
      "grad_norm": 1.2576149702072144,
      "learning_rate": 9.701815789473685e-05,
      "loss": 0.4572,
      "step": 31345
    },
    {
      "epoch": 1.755291745996192,
      "grad_norm": 1.5715019702911377,
      "learning_rate": 9.701789473684211e-05,
      "loss": 0.4802,
      "step": 31346
    },
    {
      "epoch": 1.755347743308321,
      "grad_norm": 1.1008951663970947,
      "learning_rate": 9.701763157894738e-05,
      "loss": 0.4025,
      "step": 31347
    },
    {
      "epoch": 1.7554037406204501,
      "grad_norm": 1.357969045639038,
      "learning_rate": 9.701736842105264e-05,
      "loss": 0.386,
      "step": 31348
    },
    {
      "epoch": 1.7554597379325791,
      "grad_norm": 1.2820796966552734,
      "learning_rate": 9.70171052631579e-05,
      "loss": 0.4113,
      "step": 31349
    },
    {
      "epoch": 1.7555157352447082,
      "grad_norm": 1.2205705642700195,
      "learning_rate": 9.701684210526316e-05,
      "loss": 0.4281,
      "step": 31350
    },
    {
      "epoch": 1.7555717325568372,
      "grad_norm": 1.1640247106552124,
      "learning_rate": 9.701657894736843e-05,
      "loss": 0.4124,
      "step": 31351
    },
    {
      "epoch": 1.7556277298689662,
      "grad_norm": 1.018678903579712,
      "learning_rate": 9.70163157894737e-05,
      "loss": 0.3908,
      "step": 31352
    },
    {
      "epoch": 1.7556837271810952,
      "grad_norm": 1.4704970121383667,
      "learning_rate": 9.701605263157895e-05,
      "loss": 0.4808,
      "step": 31353
    },
    {
      "epoch": 1.7557397244932242,
      "grad_norm": 1.1616512537002563,
      "learning_rate": 9.701578947368421e-05,
      "loss": 0.3945,
      "step": 31354
    },
    {
      "epoch": 1.7557957218053533,
      "grad_norm": 1.2916069030761719,
      "learning_rate": 9.701552631578947e-05,
      "loss": 0.4441,
      "step": 31355
    },
    {
      "epoch": 1.7558517191174823,
      "grad_norm": 1.2617357969284058,
      "learning_rate": 9.701526315789475e-05,
      "loss": 0.4459,
      "step": 31356
    },
    {
      "epoch": 1.7559077164296113,
      "grad_norm": 1.7488200664520264,
      "learning_rate": 9.7015e-05,
      "loss": 0.5532,
      "step": 31357
    },
    {
      "epoch": 1.7559637137417403,
      "grad_norm": 1.1443363428115845,
      "learning_rate": 9.701473684210527e-05,
      "loss": 0.3739,
      "step": 31358
    },
    {
      "epoch": 1.7560197110538693,
      "grad_norm": 1.3983464241027832,
      "learning_rate": 9.701447368421052e-05,
      "loss": 0.4532,
      "step": 31359
    },
    {
      "epoch": 1.7560757083659984,
      "grad_norm": 1.4170891046524048,
      "learning_rate": 9.70142105263158e-05,
      "loss": 0.4413,
      "step": 31360
    },
    {
      "epoch": 1.7561317056781274,
      "grad_norm": 1.5757806301116943,
      "learning_rate": 9.701394736842106e-05,
      "loss": 0.5593,
      "step": 31361
    },
    {
      "epoch": 1.7561877029902564,
      "grad_norm": 1.6156803369522095,
      "learning_rate": 9.701368421052632e-05,
      "loss": 0.456,
      "step": 31362
    },
    {
      "epoch": 1.7562437003023854,
      "grad_norm": 1.2452237606048584,
      "learning_rate": 9.701342105263158e-05,
      "loss": 0.3817,
      "step": 31363
    },
    {
      "epoch": 1.7562996976145144,
      "grad_norm": 1.4902656078338623,
      "learning_rate": 9.701315789473685e-05,
      "loss": 0.5332,
      "step": 31364
    },
    {
      "epoch": 1.7563556949266435,
      "grad_norm": 1.3083916902542114,
      "learning_rate": 9.701289473684211e-05,
      "loss": 0.5033,
      "step": 31365
    },
    {
      "epoch": 1.7564116922387725,
      "grad_norm": 1.261059284210205,
      "learning_rate": 9.701263157894738e-05,
      "loss": 0.5927,
      "step": 31366
    },
    {
      "epoch": 1.7564676895509015,
      "grad_norm": 1.3278404474258423,
      "learning_rate": 9.701236842105263e-05,
      "loss": 0.3786,
      "step": 31367
    },
    {
      "epoch": 1.7565236868630305,
      "grad_norm": 1.5977632999420166,
      "learning_rate": 9.70121052631579e-05,
      "loss": 0.4335,
      "step": 31368
    },
    {
      "epoch": 1.7565796841751595,
      "grad_norm": 1.1443051099777222,
      "learning_rate": 9.701184210526316e-05,
      "loss": 0.4152,
      "step": 31369
    },
    {
      "epoch": 1.7566356814872885,
      "grad_norm": 1.3411881923675537,
      "learning_rate": 9.701157894736844e-05,
      "loss": 0.443,
      "step": 31370
    },
    {
      "epoch": 1.7566916787994176,
      "grad_norm": 1.2317705154418945,
      "learning_rate": 9.701131578947368e-05,
      "loss": 0.4129,
      "step": 31371
    },
    {
      "epoch": 1.7567476761115466,
      "grad_norm": 1.2970744371414185,
      "learning_rate": 9.701105263157894e-05,
      "loss": 0.5434,
      "step": 31372
    },
    {
      "epoch": 1.7568036734236756,
      "grad_norm": 1.1788445711135864,
      "learning_rate": 9.701078947368422e-05,
      "loss": 0.3618,
      "step": 31373
    },
    {
      "epoch": 1.7568596707358046,
      "grad_norm": 1.4097920656204224,
      "learning_rate": 9.701052631578948e-05,
      "loss": 0.513,
      "step": 31374
    },
    {
      "epoch": 1.7569156680479336,
      "grad_norm": 1.1962306499481201,
      "learning_rate": 9.701026315789475e-05,
      "loss": 0.5054,
      "step": 31375
    },
    {
      "epoch": 1.7569716653600627,
      "grad_norm": 1.5471967458724976,
      "learning_rate": 9.701e-05,
      "loss": 0.4983,
      "step": 31376
    },
    {
      "epoch": 1.7570276626721917,
      "grad_norm": 1.1592347621917725,
      "learning_rate": 9.700973684210527e-05,
      "loss": 0.6716,
      "step": 31377
    },
    {
      "epoch": 1.7570836599843207,
      "grad_norm": 1.21383535861969,
      "learning_rate": 9.700947368421053e-05,
      "loss": 0.4102,
      "step": 31378
    },
    {
      "epoch": 1.7571396572964497,
      "grad_norm": 1.246001958847046,
      "learning_rate": 9.70092105263158e-05,
      "loss": 0.4585,
      "step": 31379
    },
    {
      "epoch": 1.7571956546085787,
      "grad_norm": 1.214787244796753,
      "learning_rate": 9.700894736842106e-05,
      "loss": 0.3326,
      "step": 31380
    },
    {
      "epoch": 1.7572516519207078,
      "grad_norm": 1.2083290815353394,
      "learning_rate": 9.700868421052632e-05,
      "loss": 0.3861,
      "step": 31381
    },
    {
      "epoch": 1.7573076492328368,
      "grad_norm": 1.724625825881958,
      "learning_rate": 9.700842105263158e-05,
      "loss": 0.7243,
      "step": 31382
    },
    {
      "epoch": 1.7573636465449658,
      "grad_norm": 1.2969224452972412,
      "learning_rate": 9.700815789473685e-05,
      "loss": 0.561,
      "step": 31383
    },
    {
      "epoch": 1.7574196438570948,
      "grad_norm": 1.1480454206466675,
      "learning_rate": 9.700789473684211e-05,
      "loss": 0.5084,
      "step": 31384
    },
    {
      "epoch": 1.7574756411692238,
      "grad_norm": 1.3230129480361938,
      "learning_rate": 9.700763157894737e-05,
      "loss": 0.5435,
      "step": 31385
    },
    {
      "epoch": 1.7575316384813529,
      "grad_norm": 1.4467577934265137,
      "learning_rate": 9.700736842105263e-05,
      "loss": 0.5748,
      "step": 31386
    },
    {
      "epoch": 1.7575876357934819,
      "grad_norm": 1.3462480306625366,
      "learning_rate": 9.70071052631579e-05,
      "loss": 0.429,
      "step": 31387
    },
    {
      "epoch": 1.757643633105611,
      "grad_norm": 1.230965495109558,
      "learning_rate": 9.700684210526317e-05,
      "loss": 0.5786,
      "step": 31388
    },
    {
      "epoch": 1.75769963041774,
      "grad_norm": 1.4086639881134033,
      "learning_rate": 9.700657894736843e-05,
      "loss": 0.4782,
      "step": 31389
    },
    {
      "epoch": 1.757755627729869,
      "grad_norm": 1.478774905204773,
      "learning_rate": 9.700631578947368e-05,
      "loss": 0.3529,
      "step": 31390
    },
    {
      "epoch": 1.757811625041998,
      "grad_norm": 1.061188817024231,
      "learning_rate": 9.700605263157894e-05,
      "loss": 0.3758,
      "step": 31391
    },
    {
      "epoch": 1.757867622354127,
      "grad_norm": 1.043335199356079,
      "learning_rate": 9.700578947368422e-05,
      "loss": 0.435,
      "step": 31392
    },
    {
      "epoch": 1.757923619666256,
      "grad_norm": 1.258521318435669,
      "learning_rate": 9.700552631578948e-05,
      "loss": 0.3658,
      "step": 31393
    },
    {
      "epoch": 1.757979616978385,
      "grad_norm": 4.0094757080078125,
      "learning_rate": 9.700526315789474e-05,
      "loss": 0.3804,
      "step": 31394
    },
    {
      "epoch": 1.758035614290514,
      "grad_norm": 1.100940227508545,
      "learning_rate": 9.7005e-05,
      "loss": 0.3954,
      "step": 31395
    },
    {
      "epoch": 1.758091611602643,
      "grad_norm": 1.1431269645690918,
      "learning_rate": 9.700473684210527e-05,
      "loss": 0.3382,
      "step": 31396
    },
    {
      "epoch": 1.758147608914772,
      "grad_norm": 1.1409684419631958,
      "learning_rate": 9.700447368421053e-05,
      "loss": 0.5586,
      "step": 31397
    },
    {
      "epoch": 1.758203606226901,
      "grad_norm": 1.3768917322158813,
      "learning_rate": 9.70042105263158e-05,
      "loss": 0.4479,
      "step": 31398
    },
    {
      "epoch": 1.75825960353903,
      "grad_norm": 1.5339504480361938,
      "learning_rate": 9.700394736842105e-05,
      "loss": 0.4439,
      "step": 31399
    },
    {
      "epoch": 1.7583156008511591,
      "grad_norm": 1.4344311952590942,
      "learning_rate": 9.700368421052632e-05,
      "loss": 0.4942,
      "step": 31400
    },
    {
      "epoch": 1.7583715981632881,
      "grad_norm": 1.1855796575546265,
      "learning_rate": 9.700342105263158e-05,
      "loss": 0.4133,
      "step": 31401
    },
    {
      "epoch": 1.7584275954754172,
      "grad_norm": 1.260161280632019,
      "learning_rate": 9.700315789473686e-05,
      "loss": 0.3815,
      "step": 31402
    },
    {
      "epoch": 1.7584835927875462,
      "grad_norm": 1.1264280080795288,
      "learning_rate": 9.700289473684212e-05,
      "loss": 0.5768,
      "step": 31403
    },
    {
      "epoch": 1.7585395900996752,
      "grad_norm": 1.344635248184204,
      "learning_rate": 9.700263157894738e-05,
      "loss": 0.5187,
      "step": 31404
    },
    {
      "epoch": 1.7585955874118042,
      "grad_norm": 1.3495181798934937,
      "learning_rate": 9.700236842105263e-05,
      "loss": 0.5547,
      "step": 31405
    },
    {
      "epoch": 1.7586515847239332,
      "grad_norm": 1.0829442739486694,
      "learning_rate": 9.70021052631579e-05,
      "loss": 0.3704,
      "step": 31406
    },
    {
      "epoch": 1.7587075820360623,
      "grad_norm": 1.1291688680648804,
      "learning_rate": 9.700184210526317e-05,
      "loss": 0.3036,
      "step": 31407
    },
    {
      "epoch": 1.7587635793481913,
      "grad_norm": 1.5994068384170532,
      "learning_rate": 9.700157894736843e-05,
      "loss": 0.4838,
      "step": 31408
    },
    {
      "epoch": 1.7588195766603203,
      "grad_norm": 1.3823555707931519,
      "learning_rate": 9.700131578947369e-05,
      "loss": 0.3572,
      "step": 31409
    },
    {
      "epoch": 1.7588755739724493,
      "grad_norm": 1.4652913808822632,
      "learning_rate": 9.700105263157895e-05,
      "loss": 0.7145,
      "step": 31410
    },
    {
      "epoch": 1.7589315712845783,
      "grad_norm": 1.3560162782669067,
      "learning_rate": 9.700078947368422e-05,
      "loss": 0.4733,
      "step": 31411
    },
    {
      "epoch": 1.7589875685967074,
      "grad_norm": 1.3950577974319458,
      "learning_rate": 9.700052631578948e-05,
      "loss": 0.5525,
      "step": 31412
    },
    {
      "epoch": 1.7590435659088364,
      "grad_norm": 1.1825885772705078,
      "learning_rate": 9.700026315789474e-05,
      "loss": 0.3966,
      "step": 31413
    },
    {
      "epoch": 1.7590995632209654,
      "grad_norm": 1.2298526763916016,
      "learning_rate": 9.7e-05,
      "loss": 0.4035,
      "step": 31414
    },
    {
      "epoch": 1.7591555605330944,
      "grad_norm": 1.301370620727539,
      "learning_rate": 9.699973684210527e-05,
      "loss": 0.5156,
      "step": 31415
    },
    {
      "epoch": 1.7592115578452234,
      "grad_norm": 1.4733564853668213,
      "learning_rate": 9.699947368421053e-05,
      "loss": 0.4461,
      "step": 31416
    },
    {
      "epoch": 1.7592675551573524,
      "grad_norm": 1.7607483863830566,
      "learning_rate": 9.699921052631579e-05,
      "loss": 0.5734,
      "step": 31417
    },
    {
      "epoch": 1.7593235524694815,
      "grad_norm": 1.5247420072555542,
      "learning_rate": 9.699894736842105e-05,
      "loss": 0.4377,
      "step": 31418
    },
    {
      "epoch": 1.7593795497816105,
      "grad_norm": 1.4009897708892822,
      "learning_rate": 9.699868421052633e-05,
      "loss": 0.4704,
      "step": 31419
    },
    {
      "epoch": 1.7594355470937395,
      "grad_norm": 1.2151381969451904,
      "learning_rate": 9.699842105263159e-05,
      "loss": 0.4475,
      "step": 31420
    },
    {
      "epoch": 1.7594915444058685,
      "grad_norm": 1.4483447074890137,
      "learning_rate": 9.699815789473686e-05,
      "loss": 0.463,
      "step": 31421
    },
    {
      "epoch": 1.7595475417179975,
      "grad_norm": 1.5088638067245483,
      "learning_rate": 9.69978947368421e-05,
      "loss": 0.4674,
      "step": 31422
    },
    {
      "epoch": 1.7596035390301266,
      "grad_norm": 1.3373827934265137,
      "learning_rate": 9.699763157894736e-05,
      "loss": 0.4543,
      "step": 31423
    },
    {
      "epoch": 1.7596595363422556,
      "grad_norm": 1.1299997568130493,
      "learning_rate": 9.699736842105264e-05,
      "loss": 0.4847,
      "step": 31424
    },
    {
      "epoch": 1.7597155336543846,
      "grad_norm": 1.358533501625061,
      "learning_rate": 9.69971052631579e-05,
      "loss": 0.3672,
      "step": 31425
    },
    {
      "epoch": 1.7597715309665136,
      "grad_norm": 1.4248813390731812,
      "learning_rate": 9.699684210526316e-05,
      "loss": 0.3376,
      "step": 31426
    },
    {
      "epoch": 1.7598275282786426,
      "grad_norm": 1.6210975646972656,
      "learning_rate": 9.699657894736842e-05,
      "loss": 0.602,
      "step": 31427
    },
    {
      "epoch": 1.7598835255907717,
      "grad_norm": 1.2547893524169922,
      "learning_rate": 9.699631578947369e-05,
      "loss": 0.4421,
      "step": 31428
    },
    {
      "epoch": 1.7599395229029007,
      "grad_norm": 1.2690328359603882,
      "learning_rate": 9.699605263157895e-05,
      "loss": 0.432,
      "step": 31429
    },
    {
      "epoch": 1.7599955202150297,
      "grad_norm": 1.2691785097122192,
      "learning_rate": 9.699578947368422e-05,
      "loss": 0.4426,
      "step": 31430
    },
    {
      "epoch": 1.7600515175271587,
      "grad_norm": 1.4508098363876343,
      "learning_rate": 9.699552631578947e-05,
      "loss": 0.5126,
      "step": 31431
    },
    {
      "epoch": 1.7601075148392877,
      "grad_norm": 1.277583360671997,
      "learning_rate": 9.699526315789474e-05,
      "loss": 0.4527,
      "step": 31432
    },
    {
      "epoch": 1.7601635121514168,
      "grad_norm": 1.5339820384979248,
      "learning_rate": 9.6995e-05,
      "loss": 0.459,
      "step": 31433
    },
    {
      "epoch": 1.7602195094635458,
      "grad_norm": 1.3452469110488892,
      "learning_rate": 9.699473684210528e-05,
      "loss": 0.5097,
      "step": 31434
    },
    {
      "epoch": 1.7602755067756748,
      "grad_norm": 1.2519586086273193,
      "learning_rate": 9.699447368421054e-05,
      "loss": 0.4339,
      "step": 31435
    },
    {
      "epoch": 1.7603315040878038,
      "grad_norm": 1.7836685180664062,
      "learning_rate": 9.69942105263158e-05,
      "loss": 0.4599,
      "step": 31436
    },
    {
      "epoch": 1.7603875013999328,
      "grad_norm": 1.384204387664795,
      "learning_rate": 9.699394736842105e-05,
      "loss": 0.5123,
      "step": 31437
    },
    {
      "epoch": 1.7604434987120618,
      "grad_norm": 1.1255943775177002,
      "learning_rate": 9.699368421052633e-05,
      "loss": 0.4688,
      "step": 31438
    },
    {
      "epoch": 1.7604994960241909,
      "grad_norm": 1.056462287902832,
      "learning_rate": 9.699342105263159e-05,
      "loss": 0.333,
      "step": 31439
    },
    {
      "epoch": 1.7605554933363199,
      "grad_norm": 1.203423261642456,
      "learning_rate": 9.699315789473685e-05,
      "loss": 0.4784,
      "step": 31440
    },
    {
      "epoch": 1.760611490648449,
      "grad_norm": 5.010548114776611,
      "learning_rate": 9.699289473684211e-05,
      "loss": 0.5922,
      "step": 31441
    },
    {
      "epoch": 1.760667487960578,
      "grad_norm": 1.019890546798706,
      "learning_rate": 9.699263157894737e-05,
      "loss": 0.4258,
      "step": 31442
    },
    {
      "epoch": 1.760723485272707,
      "grad_norm": 1.202873945236206,
      "learning_rate": 9.699236842105264e-05,
      "loss": 0.539,
      "step": 31443
    },
    {
      "epoch": 1.760779482584836,
      "grad_norm": 1.1808617115020752,
      "learning_rate": 9.69921052631579e-05,
      "loss": 0.5889,
      "step": 31444
    },
    {
      "epoch": 1.760835479896965,
      "grad_norm": 1.1226969957351685,
      "learning_rate": 9.699184210526316e-05,
      "loss": 0.374,
      "step": 31445
    },
    {
      "epoch": 1.760891477209094,
      "grad_norm": 1.376338243484497,
      "learning_rate": 9.699157894736842e-05,
      "loss": 0.3882,
      "step": 31446
    },
    {
      "epoch": 1.760947474521223,
      "grad_norm": 1.1124764680862427,
      "learning_rate": 9.699131578947369e-05,
      "loss": 0.4151,
      "step": 31447
    },
    {
      "epoch": 1.761003471833352,
      "grad_norm": 1.4831949472427368,
      "learning_rate": 9.699105263157895e-05,
      "loss": 0.4305,
      "step": 31448
    },
    {
      "epoch": 1.761059469145481,
      "grad_norm": 1.3852840662002563,
      "learning_rate": 9.699078947368421e-05,
      "loss": 0.3966,
      "step": 31449
    },
    {
      "epoch": 1.76111546645761,
      "grad_norm": 1.6167633533477783,
      "learning_rate": 9.699052631578947e-05,
      "loss": 0.411,
      "step": 31450
    },
    {
      "epoch": 1.761171463769739,
      "grad_norm": 1.3998875617980957,
      "learning_rate": 9.699026315789475e-05,
      "loss": 0.5188,
      "step": 31451
    },
    {
      "epoch": 1.7612274610818681,
      "grad_norm": 1.45525062084198,
      "learning_rate": 9.699e-05,
      "loss": 0.5463,
      "step": 31452
    },
    {
      "epoch": 1.7612834583939971,
      "grad_norm": 0.9640327095985413,
      "learning_rate": 9.698973684210528e-05,
      "loss": 0.3042,
      "step": 31453
    },
    {
      "epoch": 1.7613394557061262,
      "grad_norm": 1.1937028169631958,
      "learning_rate": 9.698947368421052e-05,
      "loss": 0.4544,
      "step": 31454
    },
    {
      "epoch": 1.7613954530182552,
      "grad_norm": 1.2516857385635376,
      "learning_rate": 9.69892105263158e-05,
      "loss": 0.4241,
      "step": 31455
    },
    {
      "epoch": 1.7614514503303842,
      "grad_norm": 1.1950222253799438,
      "learning_rate": 9.698894736842106e-05,
      "loss": 0.4083,
      "step": 31456
    },
    {
      "epoch": 1.7615074476425132,
      "grad_norm": 1.199872374534607,
      "learning_rate": 9.698868421052632e-05,
      "loss": 0.5317,
      "step": 31457
    },
    {
      "epoch": 1.7615634449546422,
      "grad_norm": 1.4942198991775513,
      "learning_rate": 9.698842105263159e-05,
      "loss": 0.3739,
      "step": 31458
    },
    {
      "epoch": 1.7616194422667713,
      "grad_norm": 2.0371768474578857,
      "learning_rate": 9.698815789473684e-05,
      "loss": 0.5374,
      "step": 31459
    },
    {
      "epoch": 1.7616754395789003,
      "grad_norm": 1.3608322143554688,
      "learning_rate": 9.698789473684211e-05,
      "loss": 0.3797,
      "step": 31460
    },
    {
      "epoch": 1.7617314368910293,
      "grad_norm": 1.410219430923462,
      "learning_rate": 9.698763157894737e-05,
      "loss": 0.3973,
      "step": 31461
    },
    {
      "epoch": 1.7617874342031583,
      "grad_norm": 1.2892036437988281,
      "learning_rate": 9.698736842105264e-05,
      "loss": 0.5025,
      "step": 31462
    },
    {
      "epoch": 1.7618434315152873,
      "grad_norm": 1.354514241218567,
      "learning_rate": 9.69871052631579e-05,
      "loss": 0.584,
      "step": 31463
    },
    {
      "epoch": 1.7618994288274163,
      "grad_norm": 1.2871614694595337,
      "learning_rate": 9.698684210526316e-05,
      "loss": 0.4542,
      "step": 31464
    },
    {
      "epoch": 1.7619554261395454,
      "grad_norm": 0.9832156300544739,
      "learning_rate": 9.698657894736842e-05,
      "loss": 0.3554,
      "step": 31465
    },
    {
      "epoch": 1.7620114234516744,
      "grad_norm": 1.3597266674041748,
      "learning_rate": 9.69863157894737e-05,
      "loss": 0.5007,
      "step": 31466
    },
    {
      "epoch": 1.7620674207638034,
      "grad_norm": 1.1829832792282104,
      "learning_rate": 9.698605263157895e-05,
      "loss": 0.4606,
      "step": 31467
    },
    {
      "epoch": 1.7621234180759324,
      "grad_norm": 1.450077772140503,
      "learning_rate": 9.698578947368421e-05,
      "loss": 0.4254,
      "step": 31468
    },
    {
      "epoch": 1.7621794153880614,
      "grad_norm": 1.1420611143112183,
      "learning_rate": 9.698552631578947e-05,
      "loss": 0.4703,
      "step": 31469
    },
    {
      "epoch": 1.7622354127001905,
      "grad_norm": 1.348101019859314,
      "learning_rate": 9.698526315789475e-05,
      "loss": 0.514,
      "step": 31470
    },
    {
      "epoch": 1.7622914100123195,
      "grad_norm": 1.2099101543426514,
      "learning_rate": 9.698500000000001e-05,
      "loss": 0.4555,
      "step": 31471
    },
    {
      "epoch": 1.7623474073244485,
      "grad_norm": 1.2844172716140747,
      "learning_rate": 9.698473684210527e-05,
      "loss": 0.5856,
      "step": 31472
    },
    {
      "epoch": 1.7624034046365775,
      "grad_norm": 1.1239879131317139,
      "learning_rate": 9.698447368421053e-05,
      "loss": 0.3856,
      "step": 31473
    },
    {
      "epoch": 1.7624594019487065,
      "grad_norm": 1.1519743204116821,
      "learning_rate": 9.69842105263158e-05,
      "loss": 0.3621,
      "step": 31474
    },
    {
      "epoch": 1.7625153992608356,
      "grad_norm": 1.4256031513214111,
      "learning_rate": 9.698394736842106e-05,
      "loss": 0.4518,
      "step": 31475
    },
    {
      "epoch": 1.7625713965729646,
      "grad_norm": 1.134522557258606,
      "learning_rate": 9.698368421052632e-05,
      "loss": 0.3928,
      "step": 31476
    },
    {
      "epoch": 1.7626273938850936,
      "grad_norm": 1.2252823114395142,
      "learning_rate": 9.698342105263158e-05,
      "loss": 0.4037,
      "step": 31477
    },
    {
      "epoch": 1.7626833911972226,
      "grad_norm": 1.4766465425491333,
      "learning_rate": 9.698315789473684e-05,
      "loss": 0.6407,
      "step": 31478
    },
    {
      "epoch": 1.7627393885093516,
      "grad_norm": 1.9454439878463745,
      "learning_rate": 9.698289473684211e-05,
      "loss": 0.5116,
      "step": 31479
    },
    {
      "epoch": 1.7627953858214807,
      "grad_norm": 1.4040120840072632,
      "learning_rate": 9.698263157894737e-05,
      "loss": 0.5018,
      "step": 31480
    },
    {
      "epoch": 1.7628513831336097,
      "grad_norm": 1.3570436239242554,
      "learning_rate": 9.698236842105263e-05,
      "loss": 0.4183,
      "step": 31481
    },
    {
      "epoch": 1.7629073804457387,
      "grad_norm": 1.0709501504898071,
      "learning_rate": 9.698210526315789e-05,
      "loss": 0.4102,
      "step": 31482
    },
    {
      "epoch": 1.7629633777578677,
      "grad_norm": 1.4571372270584106,
      "learning_rate": 9.698184210526316e-05,
      "loss": 0.6113,
      "step": 31483
    },
    {
      "epoch": 1.7630193750699967,
      "grad_norm": 1.3944772481918335,
      "learning_rate": 9.698157894736842e-05,
      "loss": 0.4673,
      "step": 31484
    },
    {
      "epoch": 1.7630753723821257,
      "grad_norm": 1.5022156238555908,
      "learning_rate": 9.69813157894737e-05,
      "loss": 0.5727,
      "step": 31485
    },
    {
      "epoch": 1.7631313696942548,
      "grad_norm": 1.2592860460281372,
      "learning_rate": 9.698105263157894e-05,
      "loss": 0.4493,
      "step": 31486
    },
    {
      "epoch": 1.7631873670063838,
      "grad_norm": 1.3246036767959595,
      "learning_rate": 9.698078947368422e-05,
      "loss": 0.4089,
      "step": 31487
    },
    {
      "epoch": 1.7632433643185128,
      "grad_norm": 1.2681739330291748,
      "learning_rate": 9.698052631578948e-05,
      "loss": 0.4553,
      "step": 31488
    },
    {
      "epoch": 1.7632993616306418,
      "grad_norm": 1.3387362957000732,
      "learning_rate": 9.698026315789475e-05,
      "loss": 0.409,
      "step": 31489
    },
    {
      "epoch": 1.7633553589427708,
      "grad_norm": 1.2997803688049316,
      "learning_rate": 9.698000000000001e-05,
      "loss": 0.4797,
      "step": 31490
    },
    {
      "epoch": 1.7634113562548999,
      "grad_norm": 1.271246075630188,
      "learning_rate": 9.697973684210527e-05,
      "loss": 0.3911,
      "step": 31491
    },
    {
      "epoch": 1.7634673535670289,
      "grad_norm": 1.261385440826416,
      "learning_rate": 9.697947368421053e-05,
      "loss": 0.4139,
      "step": 31492
    },
    {
      "epoch": 1.763523350879158,
      "grad_norm": 1.2858843803405762,
      "learning_rate": 9.697921052631579e-05,
      "loss": 0.5789,
      "step": 31493
    },
    {
      "epoch": 1.763579348191287,
      "grad_norm": 1.1422197818756104,
      "learning_rate": 9.697894736842106e-05,
      "loss": 0.3957,
      "step": 31494
    },
    {
      "epoch": 1.763635345503416,
      "grad_norm": 1.1950825452804565,
      "learning_rate": 9.697868421052632e-05,
      "loss": 0.3616,
      "step": 31495
    },
    {
      "epoch": 1.763691342815545,
      "grad_norm": 1.162674903869629,
      "learning_rate": 9.697842105263158e-05,
      "loss": 0.4708,
      "step": 31496
    },
    {
      "epoch": 1.763747340127674,
      "grad_norm": 1.1652240753173828,
      "learning_rate": 9.697815789473684e-05,
      "loss": 0.4366,
      "step": 31497
    },
    {
      "epoch": 1.763803337439803,
      "grad_norm": 1.3446221351623535,
      "learning_rate": 9.697789473684211e-05,
      "loss": 0.5055,
      "step": 31498
    },
    {
      "epoch": 1.763859334751932,
      "grad_norm": 1.241191029548645,
      "learning_rate": 9.697763157894737e-05,
      "loss": 0.4349,
      "step": 31499
    },
    {
      "epoch": 1.763915332064061,
      "grad_norm": 1.277579426765442,
      "learning_rate": 9.697736842105263e-05,
      "loss": 0.5438,
      "step": 31500
    },
    {
      "epoch": 1.76397132937619,
      "grad_norm": 1.2783100605010986,
      "learning_rate": 9.69771052631579e-05,
      "loss": 0.4395,
      "step": 31501
    },
    {
      "epoch": 1.764027326688319,
      "grad_norm": 1.133917212486267,
      "learning_rate": 9.697684210526317e-05,
      "loss": 0.4213,
      "step": 31502
    },
    {
      "epoch": 1.764083324000448,
      "grad_norm": 1.1544266939163208,
      "learning_rate": 9.697657894736843e-05,
      "loss": 0.4736,
      "step": 31503
    },
    {
      "epoch": 1.7641393213125771,
      "grad_norm": 1.3574148416519165,
      "learning_rate": 9.697631578947369e-05,
      "loss": 0.3828,
      "step": 31504
    },
    {
      "epoch": 1.7641953186247061,
      "grad_norm": 1.3751143217086792,
      "learning_rate": 9.697605263157895e-05,
      "loss": 0.4168,
      "step": 31505
    },
    {
      "epoch": 1.7642513159368352,
      "grad_norm": 1.498292326927185,
      "learning_rate": 9.697578947368422e-05,
      "loss": 0.5091,
      "step": 31506
    },
    {
      "epoch": 1.7643073132489642,
      "grad_norm": 1.330779790878296,
      "learning_rate": 9.697552631578948e-05,
      "loss": 0.5447,
      "step": 31507
    },
    {
      "epoch": 1.7643633105610932,
      "grad_norm": 1.4283132553100586,
      "learning_rate": 9.697526315789475e-05,
      "loss": 0.389,
      "step": 31508
    },
    {
      "epoch": 1.7644193078732222,
      "grad_norm": 1.2465740442276,
      "learning_rate": 9.6975e-05,
      "loss": 0.379,
      "step": 31509
    },
    {
      "epoch": 1.7644753051853512,
      "grad_norm": 1.3268877267837524,
      "learning_rate": 9.697473684210526e-05,
      "loss": 0.5558,
      "step": 31510
    },
    {
      "epoch": 1.7645313024974802,
      "grad_norm": 1.5753016471862793,
      "learning_rate": 9.697447368421053e-05,
      "loss": 0.4791,
      "step": 31511
    },
    {
      "epoch": 1.7645872998096093,
      "grad_norm": 1.1190284490585327,
      "learning_rate": 9.697421052631579e-05,
      "loss": 0.4404,
      "step": 31512
    },
    {
      "epoch": 1.7646432971217383,
      "grad_norm": 1.1840065717697144,
      "learning_rate": 9.697394736842107e-05,
      "loss": 0.4407,
      "step": 31513
    },
    {
      "epoch": 1.7646992944338673,
      "grad_norm": 1.1299233436584473,
      "learning_rate": 9.697368421052631e-05,
      "loss": 0.4273,
      "step": 31514
    },
    {
      "epoch": 1.7647552917459963,
      "grad_norm": 1.1534067392349243,
      "learning_rate": 9.697342105263158e-05,
      "loss": 0.4088,
      "step": 31515
    },
    {
      "epoch": 1.7648112890581253,
      "grad_norm": 1.4165939092636108,
      "learning_rate": 9.697315789473684e-05,
      "loss": 0.3707,
      "step": 31516
    },
    {
      "epoch": 1.7648672863702544,
      "grad_norm": 1.1244193315505981,
      "learning_rate": 9.697289473684212e-05,
      "loss": 0.5268,
      "step": 31517
    },
    {
      "epoch": 1.7649232836823834,
      "grad_norm": 1.745214819908142,
      "learning_rate": 9.697263157894738e-05,
      "loss": 0.5335,
      "step": 31518
    },
    {
      "epoch": 1.7649792809945124,
      "grad_norm": 1.2726930379867554,
      "learning_rate": 9.697236842105264e-05,
      "loss": 0.4248,
      "step": 31519
    },
    {
      "epoch": 1.7650352783066414,
      "grad_norm": 1.469014286994934,
      "learning_rate": 9.69721052631579e-05,
      "loss": 0.5316,
      "step": 31520
    },
    {
      "epoch": 1.7650912756187704,
      "grad_norm": 1.2842581272125244,
      "learning_rate": 9.697184210526317e-05,
      "loss": 0.5588,
      "step": 31521
    },
    {
      "epoch": 1.7651472729308995,
      "grad_norm": 1.3293139934539795,
      "learning_rate": 9.697157894736843e-05,
      "loss": 0.5061,
      "step": 31522
    },
    {
      "epoch": 1.7652032702430285,
      "grad_norm": 1.244208574295044,
      "learning_rate": 9.697131578947369e-05,
      "loss": 0.4003,
      "step": 31523
    },
    {
      "epoch": 1.7652592675551575,
      "grad_norm": 1.1883312463760376,
      "learning_rate": 9.697105263157895e-05,
      "loss": 0.4022,
      "step": 31524
    },
    {
      "epoch": 1.7653152648672865,
      "grad_norm": 0.9889773726463318,
      "learning_rate": 9.697078947368422e-05,
      "loss": 0.2647,
      "step": 31525
    },
    {
      "epoch": 1.7653712621794155,
      "grad_norm": 1.2561534643173218,
      "learning_rate": 9.697052631578948e-05,
      "loss": 0.4171,
      "step": 31526
    },
    {
      "epoch": 1.7654272594915446,
      "grad_norm": 1.361660361289978,
      "learning_rate": 9.697026315789474e-05,
      "loss": 0.4491,
      "step": 31527
    },
    {
      "epoch": 1.7654832568036736,
      "grad_norm": 1.2190622091293335,
      "learning_rate": 9.697e-05,
      "loss": 0.4789,
      "step": 31528
    },
    {
      "epoch": 1.7655392541158026,
      "grad_norm": 1.2869620323181152,
      "learning_rate": 9.696973684210526e-05,
      "loss": 0.4522,
      "step": 31529
    },
    {
      "epoch": 1.7655952514279316,
      "grad_norm": 1.0969651937484741,
      "learning_rate": 9.696947368421053e-05,
      "loss": 0.4566,
      "step": 31530
    },
    {
      "epoch": 1.7656512487400606,
      "grad_norm": 1.3023502826690674,
      "learning_rate": 9.69692105263158e-05,
      "loss": 0.4476,
      "step": 31531
    },
    {
      "epoch": 1.7657072460521896,
      "grad_norm": 1.3227877616882324,
      "learning_rate": 9.696894736842105e-05,
      "loss": 0.5335,
      "step": 31532
    },
    {
      "epoch": 1.7657632433643187,
      "grad_norm": 1.197417140007019,
      "learning_rate": 9.696868421052631e-05,
      "loss": 0.444,
      "step": 31533
    },
    {
      "epoch": 1.7658192406764477,
      "grad_norm": 1.3425732851028442,
      "learning_rate": 9.696842105263159e-05,
      "loss": 0.5227,
      "step": 31534
    },
    {
      "epoch": 1.7658752379885767,
      "grad_norm": 1.4290789365768433,
      "learning_rate": 9.696815789473685e-05,
      "loss": 0.5945,
      "step": 31535
    },
    {
      "epoch": 1.7659312353007057,
      "grad_norm": 1.2781115770339966,
      "learning_rate": 9.69678947368421e-05,
      "loss": 0.3768,
      "step": 31536
    },
    {
      "epoch": 1.7659872326128347,
      "grad_norm": 1.3153653144836426,
      "learning_rate": 9.696763157894737e-05,
      "loss": 0.4119,
      "step": 31537
    },
    {
      "epoch": 1.7660432299249638,
      "grad_norm": 1.438663363456726,
      "learning_rate": 9.696736842105264e-05,
      "loss": 0.3812,
      "step": 31538
    },
    {
      "epoch": 1.7660992272370928,
      "grad_norm": 1.6345983743667603,
      "learning_rate": 9.69671052631579e-05,
      "loss": 0.808,
      "step": 31539
    },
    {
      "epoch": 1.7661552245492218,
      "grad_norm": 1.1660674810409546,
      "learning_rate": 9.696684210526317e-05,
      "loss": 0.4568,
      "step": 31540
    },
    {
      "epoch": 1.7662112218613506,
      "grad_norm": 1.3210314512252808,
      "learning_rate": 9.696657894736842e-05,
      "loss": 0.5017,
      "step": 31541
    },
    {
      "epoch": 1.7662672191734796,
      "grad_norm": 1.3592729568481445,
      "learning_rate": 9.696631578947369e-05,
      "loss": 0.5468,
      "step": 31542
    },
    {
      "epoch": 1.7663232164856086,
      "grad_norm": 1.4642986059188843,
      "learning_rate": 9.696605263157895e-05,
      "loss": 0.4787,
      "step": 31543
    },
    {
      "epoch": 1.7663792137977377,
      "grad_norm": 1.237945318222046,
      "learning_rate": 9.696578947368421e-05,
      "loss": 0.3869,
      "step": 31544
    },
    {
      "epoch": 1.7664352111098667,
      "grad_norm": 1.135084867477417,
      "learning_rate": 9.696552631578948e-05,
      "loss": 0.3375,
      "step": 31545
    },
    {
      "epoch": 1.7664912084219957,
      "grad_norm": 1.292121171951294,
      "learning_rate": 9.696526315789473e-05,
      "loss": 0.4826,
      "step": 31546
    },
    {
      "epoch": 1.7665472057341247,
      "grad_norm": 1.1377503871917725,
      "learning_rate": 9.6965e-05,
      "loss": 0.442,
      "step": 31547
    },
    {
      "epoch": 1.7666032030462537,
      "grad_norm": 1.2419637441635132,
      "learning_rate": 9.696473684210526e-05,
      "loss": 0.5338,
      "step": 31548
    },
    {
      "epoch": 1.7666592003583828,
      "grad_norm": 1.1220835447311401,
      "learning_rate": 9.696447368421054e-05,
      "loss": 0.3862,
      "step": 31549
    },
    {
      "epoch": 1.7667151976705118,
      "grad_norm": 1.1769870519638062,
      "learning_rate": 9.69642105263158e-05,
      "loss": 0.3768,
      "step": 31550
    },
    {
      "epoch": 1.7667711949826408,
      "grad_norm": 1.450187087059021,
      "learning_rate": 9.696394736842106e-05,
      "loss": 0.4583,
      "step": 31551
    },
    {
      "epoch": 1.7668271922947698,
      "grad_norm": 1.1313263177871704,
      "learning_rate": 9.696368421052632e-05,
      "loss": 0.4579,
      "step": 31552
    },
    {
      "epoch": 1.7668831896068988,
      "grad_norm": 1.4563243389129639,
      "learning_rate": 9.696342105263159e-05,
      "loss": 0.5501,
      "step": 31553
    },
    {
      "epoch": 1.7669391869190278,
      "grad_norm": 1.1168112754821777,
      "learning_rate": 9.696315789473685e-05,
      "loss": 0.359,
      "step": 31554
    },
    {
      "epoch": 1.7669951842311569,
      "grad_norm": 1.2646751403808594,
      "learning_rate": 9.696289473684211e-05,
      "loss": 0.4124,
      "step": 31555
    },
    {
      "epoch": 1.7670511815432859,
      "grad_norm": 1.4349724054336548,
      "learning_rate": 9.696263157894737e-05,
      "loss": 0.5039,
      "step": 31556
    },
    {
      "epoch": 1.767107178855415,
      "grad_norm": 1.2223607301712036,
      "learning_rate": 9.696236842105264e-05,
      "loss": 0.3452,
      "step": 31557
    },
    {
      "epoch": 1.767163176167544,
      "grad_norm": 1.445101022720337,
      "learning_rate": 9.69621052631579e-05,
      "loss": 0.5003,
      "step": 31558
    },
    {
      "epoch": 1.767219173479673,
      "grad_norm": 1.5038913488388062,
      "learning_rate": 9.696184210526316e-05,
      "loss": 0.5452,
      "step": 31559
    },
    {
      "epoch": 1.767275170791802,
      "grad_norm": 1.1562330722808838,
      "learning_rate": 9.696157894736842e-05,
      "loss": 0.4638,
      "step": 31560
    },
    {
      "epoch": 1.767331168103931,
      "grad_norm": 1.3902146816253662,
      "learning_rate": 9.696131578947368e-05,
      "loss": 0.4532,
      "step": 31561
    },
    {
      "epoch": 1.76738716541606,
      "grad_norm": 1.4367204904556274,
      "learning_rate": 9.696105263157895e-05,
      "loss": 0.6274,
      "step": 31562
    },
    {
      "epoch": 1.767443162728189,
      "grad_norm": 1.3518548011779785,
      "learning_rate": 9.696078947368421e-05,
      "loss": 0.4238,
      "step": 31563
    },
    {
      "epoch": 1.767499160040318,
      "grad_norm": 1.3402560949325562,
      "learning_rate": 9.696052631578947e-05,
      "loss": 0.4032,
      "step": 31564
    },
    {
      "epoch": 1.767555157352447,
      "grad_norm": 1.5832061767578125,
      "learning_rate": 9.696026315789473e-05,
      "loss": 0.4415,
      "step": 31565
    },
    {
      "epoch": 1.767611154664576,
      "grad_norm": 1.2894824743270874,
      "learning_rate": 9.696000000000001e-05,
      "loss": 0.3704,
      "step": 31566
    },
    {
      "epoch": 1.767667151976705,
      "grad_norm": 1.0190800428390503,
      "learning_rate": 9.695973684210527e-05,
      "loss": 0.3231,
      "step": 31567
    },
    {
      "epoch": 1.7677231492888341,
      "grad_norm": 1.3909094333648682,
      "learning_rate": 9.695947368421054e-05,
      "loss": 0.5809,
      "step": 31568
    },
    {
      "epoch": 1.7677791466009631,
      "grad_norm": 1.6655607223510742,
      "learning_rate": 9.695921052631579e-05,
      "loss": 0.7583,
      "step": 31569
    },
    {
      "epoch": 1.7678351439130922,
      "grad_norm": 4.349919319152832,
      "learning_rate": 9.695894736842106e-05,
      "loss": 0.5639,
      "step": 31570
    },
    {
      "epoch": 1.7678911412252212,
      "grad_norm": 9.490988731384277,
      "learning_rate": 9.695868421052632e-05,
      "loss": 0.3731,
      "step": 31571
    },
    {
      "epoch": 1.7679471385373502,
      "grad_norm": 1.2213760614395142,
      "learning_rate": 9.695842105263159e-05,
      "loss": 0.4255,
      "step": 31572
    },
    {
      "epoch": 1.7680031358494792,
      "grad_norm": 1.6073302030563354,
      "learning_rate": 9.695815789473684e-05,
      "loss": 0.4538,
      "step": 31573
    },
    {
      "epoch": 1.7680591331616082,
      "grad_norm": 1.3122426271438599,
      "learning_rate": 9.695789473684211e-05,
      "loss": 0.3353,
      "step": 31574
    },
    {
      "epoch": 1.7681151304737373,
      "grad_norm": 1.3224055767059326,
      "learning_rate": 9.695763157894737e-05,
      "loss": 0.3719,
      "step": 31575
    },
    {
      "epoch": 1.7681711277858663,
      "grad_norm": 1.2959257364273071,
      "learning_rate": 9.695736842105264e-05,
      "loss": 0.5298,
      "step": 31576
    },
    {
      "epoch": 1.7682271250979953,
      "grad_norm": 1.9519927501678467,
      "learning_rate": 9.69571052631579e-05,
      "loss": 0.5065,
      "step": 31577
    },
    {
      "epoch": 1.7682831224101243,
      "grad_norm": 1.4113036394119263,
      "learning_rate": 9.695684210526315e-05,
      "loss": 0.5254,
      "step": 31578
    },
    {
      "epoch": 1.7683391197222533,
      "grad_norm": 1.2891052961349487,
      "learning_rate": 9.695657894736842e-05,
      "loss": 0.6011,
      "step": 31579
    },
    {
      "epoch": 1.7683951170343823,
      "grad_norm": 1.668357491493225,
      "learning_rate": 9.695631578947368e-05,
      "loss": 0.4809,
      "step": 31580
    },
    {
      "epoch": 1.7684511143465114,
      "grad_norm": 1.218103051185608,
      "learning_rate": 9.695605263157896e-05,
      "loss": 0.5375,
      "step": 31581
    },
    {
      "epoch": 1.7685071116586404,
      "grad_norm": 1.2777475118637085,
      "learning_rate": 9.695578947368422e-05,
      "loss": 0.3912,
      "step": 31582
    },
    {
      "epoch": 1.7685631089707694,
      "grad_norm": 1.4177066087722778,
      "learning_rate": 9.695552631578948e-05,
      "loss": 0.4315,
      "step": 31583
    },
    {
      "epoch": 1.7686191062828984,
      "grad_norm": 1.2517859935760498,
      "learning_rate": 9.695526315789474e-05,
      "loss": 0.4109,
      "step": 31584
    },
    {
      "epoch": 1.7686751035950274,
      "grad_norm": 1.3994783163070679,
      "learning_rate": 9.695500000000001e-05,
      "loss": 0.3294,
      "step": 31585
    },
    {
      "epoch": 1.7687311009071565,
      "grad_norm": 1.345992088317871,
      "learning_rate": 9.695473684210527e-05,
      "loss": 0.4645,
      "step": 31586
    },
    {
      "epoch": 1.7687870982192855,
      "grad_norm": 1.1893301010131836,
      "learning_rate": 9.695447368421053e-05,
      "loss": 0.4279,
      "step": 31587
    },
    {
      "epoch": 1.7688430955314145,
      "grad_norm": 1.310773253440857,
      "learning_rate": 9.695421052631579e-05,
      "loss": 0.4225,
      "step": 31588
    },
    {
      "epoch": 1.7688990928435435,
      "grad_norm": 1.2024987936019897,
      "learning_rate": 9.695394736842106e-05,
      "loss": 0.4784,
      "step": 31589
    },
    {
      "epoch": 1.7689550901556725,
      "grad_norm": 1.2498165369033813,
      "learning_rate": 9.695368421052632e-05,
      "loss": 0.5249,
      "step": 31590
    },
    {
      "epoch": 1.7690110874678016,
      "grad_norm": 1.200305700302124,
      "learning_rate": 9.695342105263158e-05,
      "loss": 0.3937,
      "step": 31591
    },
    {
      "epoch": 1.7690670847799306,
      "grad_norm": 1.3510410785675049,
      "learning_rate": 9.695315789473684e-05,
      "loss": 0.606,
      "step": 31592
    },
    {
      "epoch": 1.7691230820920596,
      "grad_norm": 1.2939341068267822,
      "learning_rate": 9.695289473684211e-05,
      "loss": 0.4131,
      "step": 31593
    },
    {
      "epoch": 1.7691790794041886,
      "grad_norm": 1.2193002700805664,
      "learning_rate": 9.695263157894737e-05,
      "loss": 0.3932,
      "step": 31594
    },
    {
      "epoch": 1.7692350767163176,
      "grad_norm": 2.1687841415405273,
      "learning_rate": 9.695236842105265e-05,
      "loss": 0.5081,
      "step": 31595
    },
    {
      "epoch": 1.7692910740284467,
      "grad_norm": 1.2129395008087158,
      "learning_rate": 9.69521052631579e-05,
      "loss": 0.3913,
      "step": 31596
    },
    {
      "epoch": 1.7693470713405757,
      "grad_norm": 1.280544638633728,
      "learning_rate": 9.695184210526315e-05,
      "loss": 0.4863,
      "step": 31597
    },
    {
      "epoch": 1.7694030686527047,
      "grad_norm": 1.1567386388778687,
      "learning_rate": 9.695157894736843e-05,
      "loss": 0.4065,
      "step": 31598
    },
    {
      "epoch": 1.7694590659648337,
      "grad_norm": 1.4139891862869263,
      "learning_rate": 9.695131578947369e-05,
      "loss": 0.3961,
      "step": 31599
    },
    {
      "epoch": 1.7695150632769627,
      "grad_norm": 1.68169367313385,
      "learning_rate": 9.695105263157896e-05,
      "loss": 0.454,
      "step": 31600
    },
    {
      "epoch": 1.7695710605890917,
      "grad_norm": 1.2753052711486816,
      "learning_rate": 9.69507894736842e-05,
      "loss": 0.4978,
      "step": 31601
    },
    {
      "epoch": 1.7696270579012208,
      "grad_norm": 1.3094966411590576,
      "learning_rate": 9.695052631578948e-05,
      "loss": 0.4663,
      "step": 31602
    },
    {
      "epoch": 1.7696830552133498,
      "grad_norm": 1.4657502174377441,
      "learning_rate": 9.695026315789474e-05,
      "loss": 0.428,
      "step": 31603
    },
    {
      "epoch": 1.7697390525254788,
      "grad_norm": 1.3512753248214722,
      "learning_rate": 9.695000000000001e-05,
      "loss": 0.4366,
      "step": 31604
    },
    {
      "epoch": 1.7697950498376078,
      "grad_norm": 1.6124919652938843,
      "learning_rate": 9.694973684210527e-05,
      "loss": 0.4766,
      "step": 31605
    },
    {
      "epoch": 1.7698510471497368,
      "grad_norm": 1.2926450967788696,
      "learning_rate": 9.694947368421053e-05,
      "loss": 0.5164,
      "step": 31606
    },
    {
      "epoch": 1.7699070444618659,
      "grad_norm": 1.2213493585586548,
      "learning_rate": 9.694921052631579e-05,
      "loss": 0.3471,
      "step": 31607
    },
    {
      "epoch": 1.7699630417739949,
      "grad_norm": 1.5887097120285034,
      "learning_rate": 9.694894736842106e-05,
      "loss": 0.4045,
      "step": 31608
    },
    {
      "epoch": 1.770019039086124,
      "grad_norm": 1.4204438924789429,
      "learning_rate": 9.694868421052632e-05,
      "loss": 0.5412,
      "step": 31609
    },
    {
      "epoch": 1.770075036398253,
      "grad_norm": 1.4192122220993042,
      "learning_rate": 9.694842105263158e-05,
      "loss": 0.3931,
      "step": 31610
    },
    {
      "epoch": 1.770131033710382,
      "grad_norm": 1.5752617120742798,
      "learning_rate": 9.694815789473684e-05,
      "loss": 0.6127,
      "step": 31611
    },
    {
      "epoch": 1.770187031022511,
      "grad_norm": 1.273874044418335,
      "learning_rate": 9.694789473684212e-05,
      "loss": 0.3776,
      "step": 31612
    },
    {
      "epoch": 1.77024302833464,
      "grad_norm": 1.2905644178390503,
      "learning_rate": 9.694763157894738e-05,
      "loss": 0.4609,
      "step": 31613
    },
    {
      "epoch": 1.770299025646769,
      "grad_norm": 1.2650513648986816,
      "learning_rate": 9.694736842105264e-05,
      "loss": 0.5233,
      "step": 31614
    },
    {
      "epoch": 1.770355022958898,
      "grad_norm": 1.0819251537322998,
      "learning_rate": 9.69471052631579e-05,
      "loss": 0.4006,
      "step": 31615
    },
    {
      "epoch": 1.770411020271027,
      "grad_norm": 1.757305383682251,
      "learning_rate": 9.694684210526316e-05,
      "loss": 0.5248,
      "step": 31616
    },
    {
      "epoch": 1.770467017583156,
      "grad_norm": 1.2096786499023438,
      "learning_rate": 9.694657894736843e-05,
      "loss": 0.4087,
      "step": 31617
    },
    {
      "epoch": 1.770523014895285,
      "grad_norm": 1.274585247039795,
      "learning_rate": 9.694631578947369e-05,
      "loss": 0.5563,
      "step": 31618
    },
    {
      "epoch": 1.770579012207414,
      "grad_norm": 1.2701423168182373,
      "learning_rate": 9.694605263157895e-05,
      "loss": 0.4777,
      "step": 31619
    },
    {
      "epoch": 1.7706350095195431,
      "grad_norm": 1.6738379001617432,
      "learning_rate": 9.694578947368421e-05,
      "loss": 0.5118,
      "step": 31620
    },
    {
      "epoch": 1.7706910068316721,
      "grad_norm": 1.7709014415740967,
      "learning_rate": 9.694552631578948e-05,
      "loss": 0.4175,
      "step": 31621
    },
    {
      "epoch": 1.7707470041438012,
      "grad_norm": 1.1367491483688354,
      "learning_rate": 9.694526315789474e-05,
      "loss": 0.4296,
      "step": 31622
    },
    {
      "epoch": 1.7708030014559302,
      "grad_norm": 1.3250112533569336,
      "learning_rate": 9.694500000000001e-05,
      "loss": 0.4725,
      "step": 31623
    },
    {
      "epoch": 1.770858998768059,
      "grad_norm": 1.6657941341400146,
      "learning_rate": 9.694473684210526e-05,
      "loss": 0.4841,
      "step": 31624
    },
    {
      "epoch": 1.770914996080188,
      "grad_norm": 1.638959527015686,
      "learning_rate": 9.694447368421053e-05,
      "loss": 0.4254,
      "step": 31625
    },
    {
      "epoch": 1.770970993392317,
      "grad_norm": 1.305229663848877,
      "learning_rate": 9.69442105263158e-05,
      "loss": 0.7032,
      "step": 31626
    },
    {
      "epoch": 1.771026990704446,
      "grad_norm": 1.2624949216842651,
      "learning_rate": 9.694394736842107e-05,
      "loss": 0.4501,
      "step": 31627
    },
    {
      "epoch": 1.771082988016575,
      "grad_norm": 1.4390772581100464,
      "learning_rate": 9.694368421052631e-05,
      "loss": 0.4499,
      "step": 31628
    },
    {
      "epoch": 1.771138985328704,
      "grad_norm": 1.0659310817718506,
      "learning_rate": 9.694342105263159e-05,
      "loss": 0.4712,
      "step": 31629
    },
    {
      "epoch": 1.771194982640833,
      "grad_norm": 1.2920441627502441,
      "learning_rate": 9.694315789473685e-05,
      "loss": 0.6832,
      "step": 31630
    },
    {
      "epoch": 1.771250979952962,
      "grad_norm": 1.3546836376190186,
      "learning_rate": 9.69428947368421e-05,
      "loss": 0.4744,
      "step": 31631
    },
    {
      "epoch": 1.7713069772650911,
      "grad_norm": 1.0631438493728638,
      "learning_rate": 9.694263157894738e-05,
      "loss": 0.3273,
      "step": 31632
    },
    {
      "epoch": 1.7713629745772201,
      "grad_norm": 1.3528141975402832,
      "learning_rate": 9.694236842105263e-05,
      "loss": 0.5658,
      "step": 31633
    },
    {
      "epoch": 1.7714189718893492,
      "grad_norm": 1.164036750793457,
      "learning_rate": 9.69421052631579e-05,
      "loss": 0.5891,
      "step": 31634
    },
    {
      "epoch": 1.7714749692014782,
      "grad_norm": 1.2094208002090454,
      "learning_rate": 9.694184210526316e-05,
      "loss": 0.4431,
      "step": 31635
    },
    {
      "epoch": 1.7715309665136072,
      "grad_norm": 1.2523390054702759,
      "learning_rate": 9.694157894736843e-05,
      "loss": 0.4231,
      "step": 31636
    },
    {
      "epoch": 1.7715869638257362,
      "grad_norm": 1.4777216911315918,
      "learning_rate": 9.694131578947369e-05,
      "loss": 0.5889,
      "step": 31637
    },
    {
      "epoch": 1.7716429611378652,
      "grad_norm": 1.0440623760223389,
      "learning_rate": 9.694105263157895e-05,
      "loss": 0.4355,
      "step": 31638
    },
    {
      "epoch": 1.7716989584499943,
      "grad_norm": 2.2610154151916504,
      "learning_rate": 9.694078947368421e-05,
      "loss": 0.5215,
      "step": 31639
    },
    {
      "epoch": 1.7717549557621233,
      "grad_norm": 1.2607110738754272,
      "learning_rate": 9.694052631578948e-05,
      "loss": 0.4782,
      "step": 31640
    },
    {
      "epoch": 1.7718109530742523,
      "grad_norm": 1.078346848487854,
      "learning_rate": 9.694026315789474e-05,
      "loss": 0.3725,
      "step": 31641
    },
    {
      "epoch": 1.7718669503863813,
      "grad_norm": 2.2641713619232178,
      "learning_rate": 9.694e-05,
      "loss": 0.3142,
      "step": 31642
    },
    {
      "epoch": 1.7719229476985103,
      "grad_norm": 1.4688613414764404,
      "learning_rate": 9.693973684210526e-05,
      "loss": 0.3142,
      "step": 31643
    },
    {
      "epoch": 1.7719789450106394,
      "grad_norm": 1.177298665046692,
      "learning_rate": 9.693947368421054e-05,
      "loss": 0.3972,
      "step": 31644
    },
    {
      "epoch": 1.7720349423227684,
      "grad_norm": 1.3002843856811523,
      "learning_rate": 9.69392105263158e-05,
      "loss": 0.439,
      "step": 31645
    },
    {
      "epoch": 1.7720909396348974,
      "grad_norm": 1.273055911064148,
      "learning_rate": 9.693894736842106e-05,
      "loss": 0.4135,
      "step": 31646
    },
    {
      "epoch": 1.7721469369470264,
      "grad_norm": 1.3855077028274536,
      "learning_rate": 9.693868421052632e-05,
      "loss": 0.4321,
      "step": 31647
    },
    {
      "epoch": 1.7722029342591554,
      "grad_norm": 1.1521739959716797,
      "learning_rate": 9.693842105263158e-05,
      "loss": 0.3381,
      "step": 31648
    },
    {
      "epoch": 1.7722589315712844,
      "grad_norm": 1.1715768575668335,
      "learning_rate": 9.693815789473685e-05,
      "loss": 0.4246,
      "step": 31649
    },
    {
      "epoch": 1.7723149288834135,
      "grad_norm": 1.096937894821167,
      "learning_rate": 9.693789473684211e-05,
      "loss": 0.4936,
      "step": 31650
    },
    {
      "epoch": 1.7723709261955425,
      "grad_norm": 1.292404055595398,
      "learning_rate": 9.693763157894737e-05,
      "loss": 0.4372,
      "step": 31651
    },
    {
      "epoch": 1.7724269235076715,
      "grad_norm": 1.322475790977478,
      "learning_rate": 9.693736842105263e-05,
      "loss": 0.3626,
      "step": 31652
    },
    {
      "epoch": 1.7724829208198005,
      "grad_norm": 1.5499192476272583,
      "learning_rate": 9.69371052631579e-05,
      "loss": 0.5435,
      "step": 31653
    },
    {
      "epoch": 1.7725389181319295,
      "grad_norm": 1.8029277324676514,
      "learning_rate": 9.693684210526316e-05,
      "loss": 0.5153,
      "step": 31654
    },
    {
      "epoch": 1.7725949154440586,
      "grad_norm": 1.4051363468170166,
      "learning_rate": 9.693657894736843e-05,
      "loss": 0.4287,
      "step": 31655
    },
    {
      "epoch": 1.7726509127561876,
      "grad_norm": 1.1059962511062622,
      "learning_rate": 9.693631578947368e-05,
      "loss": 0.4496,
      "step": 31656
    },
    {
      "epoch": 1.7727069100683166,
      "grad_norm": 1.195513129234314,
      "learning_rate": 9.693605263157895e-05,
      "loss": 0.4725,
      "step": 31657
    },
    {
      "epoch": 1.7727629073804456,
      "grad_norm": 1.163993239402771,
      "learning_rate": 9.693578947368421e-05,
      "loss": 0.4145,
      "step": 31658
    },
    {
      "epoch": 1.7728189046925746,
      "grad_norm": 1.1289546489715576,
      "learning_rate": 9.693552631578949e-05,
      "loss": 0.3838,
      "step": 31659
    },
    {
      "epoch": 1.7728749020047037,
      "grad_norm": 1.1771734952926636,
      "learning_rate": 9.693526315789475e-05,
      "loss": 0.5962,
      "step": 31660
    },
    {
      "epoch": 1.7729308993168327,
      "grad_norm": 1.2909501791000366,
      "learning_rate": 9.6935e-05,
      "loss": 0.4819,
      "step": 31661
    },
    {
      "epoch": 1.7729868966289617,
      "grad_norm": 1.3093260526657104,
      "learning_rate": 9.693473684210527e-05,
      "loss": 0.4384,
      "step": 31662
    },
    {
      "epoch": 1.7730428939410907,
      "grad_norm": 1.278106927871704,
      "learning_rate": 9.693447368421054e-05,
      "loss": 0.5425,
      "step": 31663
    },
    {
      "epoch": 1.7730988912532197,
      "grad_norm": 1.4459409713745117,
      "learning_rate": 9.69342105263158e-05,
      "loss": 0.689,
      "step": 31664
    },
    {
      "epoch": 1.7731548885653488,
      "grad_norm": 1.3192808628082275,
      "learning_rate": 9.693394736842106e-05,
      "loss": 0.4512,
      "step": 31665
    },
    {
      "epoch": 1.7732108858774778,
      "grad_norm": 1.1529093980789185,
      "learning_rate": 9.693368421052632e-05,
      "loss": 0.4549,
      "step": 31666
    },
    {
      "epoch": 1.7732668831896068,
      "grad_norm": 1.417559266090393,
      "learning_rate": 9.693342105263158e-05,
      "loss": 0.4787,
      "step": 31667
    },
    {
      "epoch": 1.7733228805017358,
      "grad_norm": 1.1452620029449463,
      "learning_rate": 9.693315789473685e-05,
      "loss": 0.4548,
      "step": 31668
    },
    {
      "epoch": 1.7733788778138648,
      "grad_norm": 1.2665936946868896,
      "learning_rate": 9.693289473684211e-05,
      "loss": 0.4287,
      "step": 31669
    },
    {
      "epoch": 1.7734348751259938,
      "grad_norm": 1.3512331247329712,
      "learning_rate": 9.693263157894737e-05,
      "loss": 0.4704,
      "step": 31670
    },
    {
      "epoch": 1.7734908724381229,
      "grad_norm": 1.2202987670898438,
      "learning_rate": 9.693236842105263e-05,
      "loss": 0.4467,
      "step": 31671
    },
    {
      "epoch": 1.7735468697502519,
      "grad_norm": 1.2009615898132324,
      "learning_rate": 9.69321052631579e-05,
      "loss": 0.4032,
      "step": 31672
    },
    {
      "epoch": 1.773602867062381,
      "grad_norm": 1.3964484930038452,
      "learning_rate": 9.693184210526316e-05,
      "loss": 0.4818,
      "step": 31673
    },
    {
      "epoch": 1.77365886437451,
      "grad_norm": 1.4268285036087036,
      "learning_rate": 9.693157894736842e-05,
      "loss": 0.4191,
      "step": 31674
    },
    {
      "epoch": 1.773714861686639,
      "grad_norm": 1.3885416984558105,
      "learning_rate": 9.693131578947368e-05,
      "loss": 0.4025,
      "step": 31675
    },
    {
      "epoch": 1.773770858998768,
      "grad_norm": 1.4444143772125244,
      "learning_rate": 9.693105263157896e-05,
      "loss": 0.4656,
      "step": 31676
    },
    {
      "epoch": 1.773826856310897,
      "grad_norm": 1.0805639028549194,
      "learning_rate": 9.693078947368422e-05,
      "loss": 0.2582,
      "step": 31677
    },
    {
      "epoch": 1.773882853623026,
      "grad_norm": 1.2887513637542725,
      "learning_rate": 9.693052631578949e-05,
      "loss": 0.4735,
      "step": 31678
    },
    {
      "epoch": 1.773938850935155,
      "grad_norm": 1.3559648990631104,
      "learning_rate": 9.693026315789474e-05,
      "loss": 0.4658,
      "step": 31679
    },
    {
      "epoch": 1.773994848247284,
      "grad_norm": 1.340680718421936,
      "learning_rate": 9.693000000000001e-05,
      "loss": 0.4844,
      "step": 31680
    },
    {
      "epoch": 1.774050845559413,
      "grad_norm": 1.3505420684814453,
      "learning_rate": 9.692973684210527e-05,
      "loss": 0.4911,
      "step": 31681
    },
    {
      "epoch": 1.774106842871542,
      "grad_norm": 1.1470973491668701,
      "learning_rate": 9.692947368421053e-05,
      "loss": 0.4863,
      "step": 31682
    },
    {
      "epoch": 1.774162840183671,
      "grad_norm": 1.459671139717102,
      "learning_rate": 9.692921052631579e-05,
      "loss": 0.6615,
      "step": 31683
    },
    {
      "epoch": 1.7742188374958001,
      "grad_norm": 1.1560617685317993,
      "learning_rate": 9.692894736842105e-05,
      "loss": 0.3681,
      "step": 31684
    },
    {
      "epoch": 1.7742748348079291,
      "grad_norm": 1.1295814514160156,
      "learning_rate": 9.692868421052632e-05,
      "loss": 0.4043,
      "step": 31685
    },
    {
      "epoch": 1.7743308321200582,
      "grad_norm": 1.4021835327148438,
      "learning_rate": 9.692842105263158e-05,
      "loss": 0.5041,
      "step": 31686
    },
    {
      "epoch": 1.7743868294321872,
      "grad_norm": 1.4055055379867554,
      "learning_rate": 9.692815789473685e-05,
      "loss": 0.41,
      "step": 31687
    },
    {
      "epoch": 1.7744428267443162,
      "grad_norm": 3.6340811252593994,
      "learning_rate": 9.69278947368421e-05,
      "loss": 0.3583,
      "step": 31688
    },
    {
      "epoch": 1.7744988240564452,
      "grad_norm": 1.1289743185043335,
      "learning_rate": 9.692763157894737e-05,
      "loss": 0.4102,
      "step": 31689
    },
    {
      "epoch": 1.7745548213685742,
      "grad_norm": 1.2529717683792114,
      "learning_rate": 9.692736842105263e-05,
      "loss": 0.3426,
      "step": 31690
    },
    {
      "epoch": 1.7746108186807033,
      "grad_norm": 1.2105739116668701,
      "learning_rate": 9.69271052631579e-05,
      "loss": 0.3889,
      "step": 31691
    },
    {
      "epoch": 1.7746668159928323,
      "grad_norm": 1.1601111888885498,
      "learning_rate": 9.692684210526317e-05,
      "loss": 0.3919,
      "step": 31692
    },
    {
      "epoch": 1.7747228133049613,
      "grad_norm": 1.2139086723327637,
      "learning_rate": 9.692657894736843e-05,
      "loss": 0.3789,
      "step": 31693
    },
    {
      "epoch": 1.7747788106170903,
      "grad_norm": 1.1330914497375488,
      "learning_rate": 9.692631578947369e-05,
      "loss": 0.6112,
      "step": 31694
    },
    {
      "epoch": 1.7748348079292193,
      "grad_norm": 1.6613965034484863,
      "learning_rate": 9.692605263157896e-05,
      "loss": 0.5128,
      "step": 31695
    },
    {
      "epoch": 1.7748908052413483,
      "grad_norm": 1.1898295879364014,
      "learning_rate": 9.692578947368422e-05,
      "loss": 0.4033,
      "step": 31696
    },
    {
      "epoch": 1.7749468025534774,
      "grad_norm": 1.0605305433273315,
      "learning_rate": 9.692552631578948e-05,
      "loss": 0.4209,
      "step": 31697
    },
    {
      "epoch": 1.7750027998656064,
      "grad_norm": 1.1873576641082764,
      "learning_rate": 9.692526315789474e-05,
      "loss": 0.4714,
      "step": 31698
    },
    {
      "epoch": 1.7750587971777354,
      "grad_norm": 1.2370866537094116,
      "learning_rate": 9.6925e-05,
      "loss": 0.3921,
      "step": 31699
    },
    {
      "epoch": 1.7751147944898644,
      "grad_norm": 1.138181447982788,
      "learning_rate": 9.692473684210527e-05,
      "loss": 0.4276,
      "step": 31700
    },
    {
      "epoch": 1.7751707918019934,
      "grad_norm": 1.3021535873413086,
      "learning_rate": 9.692447368421053e-05,
      "loss": 0.3714,
      "step": 31701
    },
    {
      "epoch": 1.7752267891141225,
      "grad_norm": 1.5191811323165894,
      "learning_rate": 9.692421052631579e-05,
      "loss": 0.4694,
      "step": 31702
    },
    {
      "epoch": 1.7752827864262515,
      "grad_norm": 1.4879953861236572,
      "learning_rate": 9.692394736842105e-05,
      "loss": 0.5514,
      "step": 31703
    },
    {
      "epoch": 1.7753387837383805,
      "grad_norm": 1.4922763109207153,
      "learning_rate": 9.692368421052632e-05,
      "loss": 0.637,
      "step": 31704
    },
    {
      "epoch": 1.7753947810505095,
      "grad_norm": 1.2755488157272339,
      "learning_rate": 9.692342105263158e-05,
      "loss": 0.473,
      "step": 31705
    },
    {
      "epoch": 1.7754507783626385,
      "grad_norm": 1.2252918481826782,
      "learning_rate": 9.692315789473684e-05,
      "loss": 0.3126,
      "step": 31706
    },
    {
      "epoch": 1.7755067756747676,
      "grad_norm": 1.1464699506759644,
      "learning_rate": 9.69228947368421e-05,
      "loss": 0.4335,
      "step": 31707
    },
    {
      "epoch": 1.7755627729868966,
      "grad_norm": 1.3130724430084229,
      "learning_rate": 9.692263157894738e-05,
      "loss": 0.3764,
      "step": 31708
    },
    {
      "epoch": 1.7756187702990256,
      "grad_norm": 2.37186861038208,
      "learning_rate": 9.692236842105264e-05,
      "loss": 0.4217,
      "step": 31709
    },
    {
      "epoch": 1.7756747676111546,
      "grad_norm": 1.2758076190948486,
      "learning_rate": 9.692210526315791e-05,
      "loss": 0.4001,
      "step": 31710
    },
    {
      "epoch": 1.7757307649232836,
      "grad_norm": 1.1145694255828857,
      "learning_rate": 9.692184210526316e-05,
      "loss": 0.5205,
      "step": 31711
    },
    {
      "epoch": 1.7757867622354127,
      "grad_norm": 1.9133871793746948,
      "learning_rate": 9.692157894736843e-05,
      "loss": 0.4757,
      "step": 31712
    },
    {
      "epoch": 1.7758427595475417,
      "grad_norm": 1.3190933465957642,
      "learning_rate": 9.692131578947369e-05,
      "loss": 0.5608,
      "step": 31713
    },
    {
      "epoch": 1.7758987568596707,
      "grad_norm": 2.1781973838806152,
      "learning_rate": 9.692105263157896e-05,
      "loss": 0.4955,
      "step": 31714
    },
    {
      "epoch": 1.7759547541717997,
      "grad_norm": 1.279032826423645,
      "learning_rate": 9.692078947368422e-05,
      "loss": 0.4946,
      "step": 31715
    },
    {
      "epoch": 1.7760107514839287,
      "grad_norm": 1.3642264604568481,
      "learning_rate": 9.692052631578948e-05,
      "loss": 0.6054,
      "step": 31716
    },
    {
      "epoch": 1.7760667487960577,
      "grad_norm": 1.3150238990783691,
      "learning_rate": 9.692026315789474e-05,
      "loss": 0.3583,
      "step": 31717
    },
    {
      "epoch": 1.7761227461081868,
      "grad_norm": 1.5483708381652832,
      "learning_rate": 9.692e-05,
      "loss": 0.5425,
      "step": 31718
    },
    {
      "epoch": 1.7761787434203158,
      "grad_norm": 1.29545259475708,
      "learning_rate": 9.691973684210527e-05,
      "loss": 0.569,
      "step": 31719
    },
    {
      "epoch": 1.7762347407324448,
      "grad_norm": 1.1991771459579468,
      "learning_rate": 9.691947368421053e-05,
      "loss": 0.5426,
      "step": 31720
    },
    {
      "epoch": 1.7762907380445738,
      "grad_norm": 1.1908551454544067,
      "learning_rate": 9.691921052631579e-05,
      "loss": 0.4921,
      "step": 31721
    },
    {
      "epoch": 1.7763467353567028,
      "grad_norm": 1.421525001525879,
      "learning_rate": 9.691894736842105e-05,
      "loss": 0.6218,
      "step": 31722
    },
    {
      "epoch": 1.7764027326688319,
      "grad_norm": 1.1339898109436035,
      "learning_rate": 9.691868421052633e-05,
      "loss": 0.4113,
      "step": 31723
    },
    {
      "epoch": 1.7764587299809609,
      "grad_norm": 1.093232274055481,
      "learning_rate": 9.691842105263159e-05,
      "loss": 0.362,
      "step": 31724
    },
    {
      "epoch": 1.77651472729309,
      "grad_norm": 1.2974497079849243,
      "learning_rate": 9.691815789473685e-05,
      "loss": 0.378,
      "step": 31725
    },
    {
      "epoch": 1.776570724605219,
      "grad_norm": 1.2342931032180786,
      "learning_rate": 9.69178947368421e-05,
      "loss": 0.4887,
      "step": 31726
    },
    {
      "epoch": 1.776626721917348,
      "grad_norm": 1.3725272417068481,
      "learning_rate": 9.691763157894738e-05,
      "loss": 0.4088,
      "step": 31727
    },
    {
      "epoch": 1.776682719229477,
      "grad_norm": 1.853804349899292,
      "learning_rate": 9.691736842105264e-05,
      "loss": 0.4583,
      "step": 31728
    },
    {
      "epoch": 1.776738716541606,
      "grad_norm": 1.1662169694900513,
      "learning_rate": 9.69171052631579e-05,
      "loss": 0.3787,
      "step": 31729
    },
    {
      "epoch": 1.776794713853735,
      "grad_norm": 1.2693150043487549,
      "learning_rate": 9.691684210526316e-05,
      "loss": 0.4017,
      "step": 31730
    },
    {
      "epoch": 1.776850711165864,
      "grad_norm": 1.52790367603302,
      "learning_rate": 9.691657894736843e-05,
      "loss": 0.5531,
      "step": 31731
    },
    {
      "epoch": 1.776906708477993,
      "grad_norm": 1.3320693969726562,
      "learning_rate": 9.691631578947369e-05,
      "loss": 0.4455,
      "step": 31732
    },
    {
      "epoch": 1.776962705790122,
      "grad_norm": 1.2238317728042603,
      "learning_rate": 9.691605263157896e-05,
      "loss": 0.5328,
      "step": 31733
    },
    {
      "epoch": 1.777018703102251,
      "grad_norm": 1.2253999710083008,
      "learning_rate": 9.691578947368421e-05,
      "loss": 0.4084,
      "step": 31734
    },
    {
      "epoch": 1.77707470041438,
      "grad_norm": 1.1530673503875732,
      "learning_rate": 9.691552631578947e-05,
      "loss": 0.3968,
      "step": 31735
    },
    {
      "epoch": 1.7771306977265091,
      "grad_norm": 1.1781492233276367,
      "learning_rate": 9.691526315789474e-05,
      "loss": 0.4335,
      "step": 31736
    },
    {
      "epoch": 1.7771866950386381,
      "grad_norm": 1.2266974449157715,
      "learning_rate": 9.6915e-05,
      "loss": 0.4273,
      "step": 31737
    },
    {
      "epoch": 1.7772426923507672,
      "grad_norm": 1.2361249923706055,
      "learning_rate": 9.691473684210526e-05,
      "loss": 0.4069,
      "step": 31738
    },
    {
      "epoch": 1.7772986896628962,
      "grad_norm": 1.9679622650146484,
      "learning_rate": 9.691447368421052e-05,
      "loss": 0.5161,
      "step": 31739
    },
    {
      "epoch": 1.7773546869750252,
      "grad_norm": 1.1759806871414185,
      "learning_rate": 9.69142105263158e-05,
      "loss": 0.4516,
      "step": 31740
    },
    {
      "epoch": 1.7774106842871542,
      "grad_norm": 1.8541347980499268,
      "learning_rate": 9.691394736842106e-05,
      "loss": 0.4834,
      "step": 31741
    },
    {
      "epoch": 1.7774666815992832,
      "grad_norm": 1.1747126579284668,
      "learning_rate": 9.691368421052633e-05,
      "loss": 0.3733,
      "step": 31742
    },
    {
      "epoch": 1.7775226789114122,
      "grad_norm": 1.3951380252838135,
      "learning_rate": 9.691342105263157e-05,
      "loss": 0.4915,
      "step": 31743
    },
    {
      "epoch": 1.7775786762235413,
      "grad_norm": 1.2836487293243408,
      "learning_rate": 9.691315789473685e-05,
      "loss": 0.4666,
      "step": 31744
    },
    {
      "epoch": 1.7776346735356703,
      "grad_norm": 1.1611849069595337,
      "learning_rate": 9.691289473684211e-05,
      "loss": 0.4434,
      "step": 31745
    },
    {
      "epoch": 1.7776906708477993,
      "grad_norm": 1.3393611907958984,
      "learning_rate": 9.691263157894738e-05,
      "loss": 0.4784,
      "step": 31746
    },
    {
      "epoch": 1.7777466681599283,
      "grad_norm": 1.1358977556228638,
      "learning_rate": 9.691236842105264e-05,
      "loss": 0.4414,
      "step": 31747
    },
    {
      "epoch": 1.7778026654720573,
      "grad_norm": 1.4379245042800903,
      "learning_rate": 9.69121052631579e-05,
      "loss": 0.6311,
      "step": 31748
    },
    {
      "epoch": 1.7778586627841864,
      "grad_norm": 1.019580364227295,
      "learning_rate": 9.691184210526316e-05,
      "loss": 0.3647,
      "step": 31749
    },
    {
      "epoch": 1.7779146600963154,
      "grad_norm": 1.4219930171966553,
      "learning_rate": 9.691157894736843e-05,
      "loss": 0.5379,
      "step": 31750
    },
    {
      "epoch": 1.7779706574084444,
      "grad_norm": 1.107913851737976,
      "learning_rate": 9.69113157894737e-05,
      "loss": 0.3647,
      "step": 31751
    },
    {
      "epoch": 1.7780266547205734,
      "grad_norm": 1.3188385963439941,
      "learning_rate": 9.691105263157895e-05,
      "loss": 0.3607,
      "step": 31752
    },
    {
      "epoch": 1.7780826520327024,
      "grad_norm": 1.370132565498352,
      "learning_rate": 9.691078947368421e-05,
      "loss": 0.4885,
      "step": 31753
    },
    {
      "epoch": 1.7781386493448315,
      "grad_norm": 1.3688284158706665,
      "learning_rate": 9.691052631578947e-05,
      "loss": 0.4173,
      "step": 31754
    },
    {
      "epoch": 1.7781946466569605,
      "grad_norm": 1.4062292575836182,
      "learning_rate": 9.691026315789475e-05,
      "loss": 0.3259,
      "step": 31755
    },
    {
      "epoch": 1.7782506439690895,
      "grad_norm": 1.1700189113616943,
      "learning_rate": 9.691e-05,
      "loss": 0.4681,
      "step": 31756
    },
    {
      "epoch": 1.7783066412812185,
      "grad_norm": 1.588115930557251,
      "learning_rate": 9.690973684210527e-05,
      "loss": 0.4796,
      "step": 31757
    },
    {
      "epoch": 1.7783626385933475,
      "grad_norm": 1.4379734992980957,
      "learning_rate": 9.690947368421053e-05,
      "loss": 0.5249,
      "step": 31758
    },
    {
      "epoch": 1.7784186359054766,
      "grad_norm": 1.4565213918685913,
      "learning_rate": 9.69092105263158e-05,
      "loss": 0.4669,
      "step": 31759
    },
    {
      "epoch": 1.7784746332176056,
      "grad_norm": 1.3071175813674927,
      "learning_rate": 9.690894736842106e-05,
      "loss": 0.4272,
      "step": 31760
    },
    {
      "epoch": 1.7785306305297346,
      "grad_norm": 1.2956658601760864,
      "learning_rate": 9.690868421052632e-05,
      "loss": 0.4045,
      "step": 31761
    },
    {
      "epoch": 1.7785866278418636,
      "grad_norm": 1.0321497917175293,
      "learning_rate": 9.690842105263158e-05,
      "loss": 0.3443,
      "step": 31762
    },
    {
      "epoch": 1.7786426251539926,
      "grad_norm": 1.2439674139022827,
      "learning_rate": 9.690815789473685e-05,
      "loss": 0.422,
      "step": 31763
    },
    {
      "epoch": 1.7786986224661216,
      "grad_norm": 1.3908679485321045,
      "learning_rate": 9.690789473684211e-05,
      "loss": 0.5081,
      "step": 31764
    },
    {
      "epoch": 1.7787546197782507,
      "grad_norm": 1.2918809652328491,
      "learning_rate": 9.690763157894738e-05,
      "loss": 0.5505,
      "step": 31765
    },
    {
      "epoch": 1.7788106170903797,
      "grad_norm": 1.4847381114959717,
      "learning_rate": 9.690736842105263e-05,
      "loss": 0.6401,
      "step": 31766
    },
    {
      "epoch": 1.7788666144025087,
      "grad_norm": 1.180594801902771,
      "learning_rate": 9.69071052631579e-05,
      "loss": 0.3993,
      "step": 31767
    },
    {
      "epoch": 1.7789226117146377,
      "grad_norm": 1.1299070119857788,
      "learning_rate": 9.690684210526316e-05,
      "loss": 0.3554,
      "step": 31768
    },
    {
      "epoch": 1.7789786090267667,
      "grad_norm": 1.3463633060455322,
      "learning_rate": 9.690657894736842e-05,
      "loss": 0.3062,
      "step": 31769
    },
    {
      "epoch": 1.7790346063388958,
      "grad_norm": 0.982157826423645,
      "learning_rate": 9.69063157894737e-05,
      "loss": 0.3706,
      "step": 31770
    },
    {
      "epoch": 1.7790906036510248,
      "grad_norm": 1.5274629592895508,
      "learning_rate": 9.690605263157894e-05,
      "loss": 0.4789,
      "step": 31771
    },
    {
      "epoch": 1.7791466009631538,
      "grad_norm": 1.5049922466278076,
      "learning_rate": 9.690578947368422e-05,
      "loss": 0.5137,
      "step": 31772
    },
    {
      "epoch": 1.7792025982752828,
      "grad_norm": 1.152225375175476,
      "learning_rate": 9.690552631578948e-05,
      "loss": 0.4106,
      "step": 31773
    },
    {
      "epoch": 1.7792585955874118,
      "grad_norm": 1.4260683059692383,
      "learning_rate": 9.690526315789475e-05,
      "loss": 0.5347,
      "step": 31774
    },
    {
      "epoch": 1.7793145928995409,
      "grad_norm": 2.1656501293182373,
      "learning_rate": 9.6905e-05,
      "loss": 0.6394,
      "step": 31775
    },
    {
      "epoch": 1.7793705902116699,
      "grad_norm": 1.3388731479644775,
      "learning_rate": 9.690473684210527e-05,
      "loss": 0.4603,
      "step": 31776
    },
    {
      "epoch": 1.779426587523799,
      "grad_norm": 1.748445749282837,
      "learning_rate": 9.690447368421053e-05,
      "loss": 0.5135,
      "step": 31777
    },
    {
      "epoch": 1.779482584835928,
      "grad_norm": 1.1211421489715576,
      "learning_rate": 9.69042105263158e-05,
      "loss": 0.3823,
      "step": 31778
    },
    {
      "epoch": 1.779538582148057,
      "grad_norm": 1.2308905124664307,
      "learning_rate": 9.690394736842106e-05,
      "loss": 0.6448,
      "step": 31779
    },
    {
      "epoch": 1.779594579460186,
      "grad_norm": 1.1907392740249634,
      "learning_rate": 9.690368421052632e-05,
      "loss": 0.4445,
      "step": 31780
    },
    {
      "epoch": 1.779650576772315,
      "grad_norm": 1.2433491945266724,
      "learning_rate": 9.690342105263158e-05,
      "loss": 0.4855,
      "step": 31781
    },
    {
      "epoch": 1.779706574084444,
      "grad_norm": 1.5997979640960693,
      "learning_rate": 9.690315789473685e-05,
      "loss": 0.5141,
      "step": 31782
    },
    {
      "epoch": 1.779762571396573,
      "grad_norm": 1.5376241207122803,
      "learning_rate": 9.690289473684211e-05,
      "loss": 0.4068,
      "step": 31783
    },
    {
      "epoch": 1.779818568708702,
      "grad_norm": 1.3327856063842773,
      "learning_rate": 9.690263157894737e-05,
      "loss": 0.4554,
      "step": 31784
    },
    {
      "epoch": 1.779874566020831,
      "grad_norm": 1.2942867279052734,
      "learning_rate": 9.690236842105263e-05,
      "loss": 0.4102,
      "step": 31785
    },
    {
      "epoch": 1.77993056333296,
      "grad_norm": 1.1606141328811646,
      "learning_rate": 9.690210526315789e-05,
      "loss": 0.4507,
      "step": 31786
    },
    {
      "epoch": 1.779986560645089,
      "grad_norm": 1.7103277444839478,
      "learning_rate": 9.690184210526317e-05,
      "loss": 0.6373,
      "step": 31787
    },
    {
      "epoch": 1.780042557957218,
      "grad_norm": 1.3943488597869873,
      "learning_rate": 9.690157894736843e-05,
      "loss": 0.4965,
      "step": 31788
    },
    {
      "epoch": 1.7800985552693471,
      "grad_norm": 1.1858793497085571,
      "learning_rate": 9.690131578947369e-05,
      "loss": 0.4665,
      "step": 31789
    },
    {
      "epoch": 1.7801545525814761,
      "grad_norm": 1.5315145254135132,
      "learning_rate": 9.690105263157894e-05,
      "loss": 0.4587,
      "step": 31790
    },
    {
      "epoch": 1.7802105498936052,
      "grad_norm": 1.3065170049667358,
      "learning_rate": 9.690078947368422e-05,
      "loss": 0.4388,
      "step": 31791
    },
    {
      "epoch": 1.7802665472057342,
      "grad_norm": 1.2521973848342896,
      "learning_rate": 9.690052631578948e-05,
      "loss": 0.3834,
      "step": 31792
    },
    {
      "epoch": 1.7803225445178632,
      "grad_norm": 1.2697452306747437,
      "learning_rate": 9.690026315789474e-05,
      "loss": 0.5289,
      "step": 31793
    },
    {
      "epoch": 1.7803785418299922,
      "grad_norm": 1.2355313301086426,
      "learning_rate": 9.69e-05,
      "loss": 0.4728,
      "step": 31794
    },
    {
      "epoch": 1.7804345391421212,
      "grad_norm": 1.1875896453857422,
      "learning_rate": 9.689973684210527e-05,
      "loss": 0.4261,
      "step": 31795
    },
    {
      "epoch": 1.7804905364542503,
      "grad_norm": 1.2282297611236572,
      "learning_rate": 9.689947368421053e-05,
      "loss": 0.5512,
      "step": 31796
    },
    {
      "epoch": 1.7805465337663793,
      "grad_norm": 1.7952834367752075,
      "learning_rate": 9.68992105263158e-05,
      "loss": 0.4245,
      "step": 31797
    },
    {
      "epoch": 1.7806025310785083,
      "grad_norm": 2.2149720191955566,
      "learning_rate": 9.689894736842105e-05,
      "loss": 0.4552,
      "step": 31798
    },
    {
      "epoch": 1.7806585283906373,
      "grad_norm": 1.2255440950393677,
      "learning_rate": 9.689868421052632e-05,
      "loss": 0.4927,
      "step": 31799
    },
    {
      "epoch": 1.7807145257027663,
      "grad_norm": 1.1667736768722534,
      "learning_rate": 9.689842105263158e-05,
      "loss": 0.3893,
      "step": 31800
    },
    {
      "epoch": 1.7807705230148954,
      "grad_norm": 1.3435007333755493,
      "learning_rate": 9.689815789473686e-05,
      "loss": 0.4568,
      "step": 31801
    },
    {
      "epoch": 1.7808265203270244,
      "grad_norm": 1.2837270498275757,
      "learning_rate": 9.689789473684212e-05,
      "loss": 0.4608,
      "step": 31802
    },
    {
      "epoch": 1.7808825176391534,
      "grad_norm": 1.213241457939148,
      "learning_rate": 9.689763157894736e-05,
      "loss": 0.4059,
      "step": 31803
    },
    {
      "epoch": 1.7809385149512824,
      "grad_norm": 1.6467320919036865,
      "learning_rate": 9.689736842105264e-05,
      "loss": 0.4102,
      "step": 31804
    },
    {
      "epoch": 1.7809945122634114,
      "grad_norm": 1.0240404605865479,
      "learning_rate": 9.68971052631579e-05,
      "loss": 0.3627,
      "step": 31805
    },
    {
      "epoch": 1.7810505095755405,
      "grad_norm": 1.316407561302185,
      "learning_rate": 9.689684210526317e-05,
      "loss": 0.4339,
      "step": 31806
    },
    {
      "epoch": 1.7811065068876695,
      "grad_norm": 1.584269404411316,
      "learning_rate": 9.689657894736843e-05,
      "loss": 0.5503,
      "step": 31807
    },
    {
      "epoch": 1.7811625041997985,
      "grad_norm": 1.0722568035125732,
      "learning_rate": 9.689631578947369e-05,
      "loss": 0.4073,
      "step": 31808
    },
    {
      "epoch": 1.7812185015119275,
      "grad_norm": 1.6375370025634766,
      "learning_rate": 9.689605263157895e-05,
      "loss": 0.6101,
      "step": 31809
    },
    {
      "epoch": 1.7812744988240565,
      "grad_norm": 1.1459671258926392,
      "learning_rate": 9.689578947368422e-05,
      "loss": 0.4613,
      "step": 31810
    },
    {
      "epoch": 1.7813304961361855,
      "grad_norm": 1.2400115728378296,
      "learning_rate": 9.689552631578948e-05,
      "loss": 0.4566,
      "step": 31811
    },
    {
      "epoch": 1.7813864934483146,
      "grad_norm": 1.147714614868164,
      "learning_rate": 9.689526315789474e-05,
      "loss": 0.4565,
      "step": 31812
    },
    {
      "epoch": 1.7814424907604436,
      "grad_norm": 1.325195550918579,
      "learning_rate": 9.6895e-05,
      "loss": 0.4775,
      "step": 31813
    },
    {
      "epoch": 1.7814984880725726,
      "grad_norm": 1.9836220741271973,
      "learning_rate": 9.689473684210527e-05,
      "loss": 0.5257,
      "step": 31814
    },
    {
      "epoch": 1.7815544853847016,
      "grad_norm": 1.124781847000122,
      "learning_rate": 9.689447368421053e-05,
      "loss": 0.391,
      "step": 31815
    },
    {
      "epoch": 1.7816104826968306,
      "grad_norm": 1.200110673904419,
      "learning_rate": 9.689421052631579e-05,
      "loss": 0.4523,
      "step": 31816
    },
    {
      "epoch": 1.7816664800089597,
      "grad_norm": 1.1694585084915161,
      "learning_rate": 9.689394736842105e-05,
      "loss": 0.3514,
      "step": 31817
    },
    {
      "epoch": 1.7817224773210887,
      "grad_norm": 1.551188588142395,
      "learning_rate": 9.689368421052633e-05,
      "loss": 0.5292,
      "step": 31818
    },
    {
      "epoch": 1.7817784746332177,
      "grad_norm": 1.1470431089401245,
      "learning_rate": 9.689342105263159e-05,
      "loss": 0.5614,
      "step": 31819
    },
    {
      "epoch": 1.7818344719453467,
      "grad_norm": 1.3186626434326172,
      "learning_rate": 9.689315789473684e-05,
      "loss": 0.4121,
      "step": 31820
    },
    {
      "epoch": 1.7818904692574757,
      "grad_norm": 0.9956709742546082,
      "learning_rate": 9.68928947368421e-05,
      "loss": 0.3453,
      "step": 31821
    },
    {
      "epoch": 1.7819464665696048,
      "grad_norm": 1.278331995010376,
      "learning_rate": 9.689263157894736e-05,
      "loss": 0.3733,
      "step": 31822
    },
    {
      "epoch": 1.7820024638817338,
      "grad_norm": 1.5626790523529053,
      "learning_rate": 9.689236842105264e-05,
      "loss": 0.477,
      "step": 31823
    },
    {
      "epoch": 1.7820584611938628,
      "grad_norm": 1.3329910039901733,
      "learning_rate": 9.68921052631579e-05,
      "loss": 0.5075,
      "step": 31824
    },
    {
      "epoch": 1.7821144585059918,
      "grad_norm": 1.240760087966919,
      "learning_rate": 9.689184210526317e-05,
      "loss": 0.4999,
      "step": 31825
    },
    {
      "epoch": 1.7821704558181208,
      "grad_norm": 1.1359931230545044,
      "learning_rate": 9.689157894736842e-05,
      "loss": 0.3785,
      "step": 31826
    },
    {
      "epoch": 1.7822264531302499,
      "grad_norm": 12.997991561889648,
      "learning_rate": 9.689131578947369e-05,
      "loss": 0.5553,
      "step": 31827
    },
    {
      "epoch": 1.7822824504423789,
      "grad_norm": 1.2757093906402588,
      "learning_rate": 9.689105263157895e-05,
      "loss": 0.4396,
      "step": 31828
    },
    {
      "epoch": 1.782338447754508,
      "grad_norm": 1.2319806814193726,
      "learning_rate": 9.689078947368422e-05,
      "loss": 0.4742,
      "step": 31829
    },
    {
      "epoch": 1.782394445066637,
      "grad_norm": 1.4098560810089111,
      "learning_rate": 9.689052631578947e-05,
      "loss": 0.4182,
      "step": 31830
    },
    {
      "epoch": 1.782450442378766,
      "grad_norm": 1.340309500694275,
      "learning_rate": 9.689026315789474e-05,
      "loss": 0.4386,
      "step": 31831
    },
    {
      "epoch": 1.782506439690895,
      "grad_norm": 1.3434604406356812,
      "learning_rate": 9.689e-05,
      "loss": 0.4141,
      "step": 31832
    },
    {
      "epoch": 1.782562437003024,
      "grad_norm": 1.1592700481414795,
      "learning_rate": 9.688973684210528e-05,
      "loss": 0.3565,
      "step": 31833
    },
    {
      "epoch": 1.782618434315153,
      "grad_norm": 1.4475698471069336,
      "learning_rate": 9.688947368421054e-05,
      "loss": 0.526,
      "step": 31834
    },
    {
      "epoch": 1.782674431627282,
      "grad_norm": 1.2637836933135986,
      "learning_rate": 9.68892105263158e-05,
      "loss": 0.5198,
      "step": 31835
    },
    {
      "epoch": 1.782730428939411,
      "grad_norm": 1.7472718954086304,
      "learning_rate": 9.688894736842105e-05,
      "loss": 0.5239,
      "step": 31836
    },
    {
      "epoch": 1.78278642625154,
      "grad_norm": 1.2594273090362549,
      "learning_rate": 9.688868421052633e-05,
      "loss": 0.6498,
      "step": 31837
    },
    {
      "epoch": 1.782842423563669,
      "grad_norm": 1.727853775024414,
      "learning_rate": 9.688842105263159e-05,
      "loss": 0.6166,
      "step": 31838
    },
    {
      "epoch": 1.782898420875798,
      "grad_norm": 1.330910086631775,
      "learning_rate": 9.688815789473685e-05,
      "loss": 0.5106,
      "step": 31839
    },
    {
      "epoch": 1.782954418187927,
      "grad_norm": 1.4381299018859863,
      "learning_rate": 9.688789473684211e-05,
      "loss": 0.4625,
      "step": 31840
    },
    {
      "epoch": 1.7830104155000561,
      "grad_norm": 1.1630643606185913,
      "learning_rate": 9.688763157894737e-05,
      "loss": 0.5307,
      "step": 31841
    },
    {
      "epoch": 1.7830664128121851,
      "grad_norm": 1.0656017065048218,
      "learning_rate": 9.688736842105264e-05,
      "loss": 0.431,
      "step": 31842
    },
    {
      "epoch": 1.7831224101243142,
      "grad_norm": 1.2093061208724976,
      "learning_rate": 9.68871052631579e-05,
      "loss": 0.3468,
      "step": 31843
    },
    {
      "epoch": 1.7831784074364432,
      "grad_norm": 1.1307623386383057,
      "learning_rate": 9.688684210526316e-05,
      "loss": 0.3215,
      "step": 31844
    },
    {
      "epoch": 1.7832344047485722,
      "grad_norm": 1.2576932907104492,
      "learning_rate": 9.688657894736842e-05,
      "loss": 0.4666,
      "step": 31845
    },
    {
      "epoch": 1.7832904020607012,
      "grad_norm": 1.3892356157302856,
      "learning_rate": 9.688631578947369e-05,
      "loss": 0.3826,
      "step": 31846
    },
    {
      "epoch": 1.7833463993728302,
      "grad_norm": 1.03448486328125,
      "learning_rate": 9.688605263157895e-05,
      "loss": 0.3517,
      "step": 31847
    },
    {
      "epoch": 1.7834023966849593,
      "grad_norm": 1.18561851978302,
      "learning_rate": 9.688578947368421e-05,
      "loss": 0.4066,
      "step": 31848
    },
    {
      "epoch": 1.7834583939970883,
      "grad_norm": 1.6200153827667236,
      "learning_rate": 9.688552631578947e-05,
      "loss": 0.4927,
      "step": 31849
    },
    {
      "epoch": 1.7835143913092173,
      "grad_norm": 1.7550350427627563,
      "learning_rate": 9.688526315789475e-05,
      "loss": 0.4092,
      "step": 31850
    },
    {
      "epoch": 1.7835703886213463,
      "grad_norm": 1.1973174810409546,
      "learning_rate": 9.6885e-05,
      "loss": 0.4021,
      "step": 31851
    },
    {
      "epoch": 1.7836263859334753,
      "grad_norm": 1.3861836194992065,
      "learning_rate": 9.688473684210528e-05,
      "loss": 0.5933,
      "step": 31852
    },
    {
      "epoch": 1.7836823832456044,
      "grad_norm": 1.1747161149978638,
      "learning_rate": 9.688447368421052e-05,
      "loss": 0.5082,
      "step": 31853
    },
    {
      "epoch": 1.7837383805577334,
      "grad_norm": 1.2632472515106201,
      "learning_rate": 9.68842105263158e-05,
      "loss": 0.4313,
      "step": 31854
    },
    {
      "epoch": 1.7837943778698624,
      "grad_norm": 1.205872893333435,
      "learning_rate": 9.688394736842106e-05,
      "loss": 0.4356,
      "step": 31855
    },
    {
      "epoch": 1.7838503751819914,
      "grad_norm": 1.2383676767349243,
      "learning_rate": 9.688368421052632e-05,
      "loss": 0.3702,
      "step": 31856
    },
    {
      "epoch": 1.7839063724941204,
      "grad_norm": 1.3884590864181519,
      "learning_rate": 9.688342105263159e-05,
      "loss": 0.6019,
      "step": 31857
    },
    {
      "epoch": 1.7839623698062494,
      "grad_norm": 1.1572792530059814,
      "learning_rate": 9.688315789473684e-05,
      "loss": 0.3724,
      "step": 31858
    },
    {
      "epoch": 1.7840183671183785,
      "grad_norm": 1.1126223802566528,
      "learning_rate": 9.688289473684211e-05,
      "loss": 0.4468,
      "step": 31859
    },
    {
      "epoch": 1.7840743644305075,
      "grad_norm": 1.447487235069275,
      "learning_rate": 9.688263157894737e-05,
      "loss": 0.4622,
      "step": 31860
    },
    {
      "epoch": 1.7841303617426365,
      "grad_norm": 1.2140978574752808,
      "learning_rate": 9.688236842105264e-05,
      "loss": 0.4691,
      "step": 31861
    },
    {
      "epoch": 1.7841863590547655,
      "grad_norm": 1.0958054065704346,
      "learning_rate": 9.68821052631579e-05,
      "loss": 0.5264,
      "step": 31862
    },
    {
      "epoch": 1.7842423563668945,
      "grad_norm": 1.5366441011428833,
      "learning_rate": 9.688184210526316e-05,
      "loss": 0.4639,
      "step": 31863
    },
    {
      "epoch": 1.7842983536790236,
      "grad_norm": 1.129201054573059,
      "learning_rate": 9.688157894736842e-05,
      "loss": 0.4285,
      "step": 31864
    },
    {
      "epoch": 1.7843543509911526,
      "grad_norm": 1.1247044801712036,
      "learning_rate": 9.68813157894737e-05,
      "loss": 0.3693,
      "step": 31865
    },
    {
      "epoch": 1.7844103483032816,
      "grad_norm": 1.5763261318206787,
      "learning_rate": 9.688105263157896e-05,
      "loss": 0.5131,
      "step": 31866
    },
    {
      "epoch": 1.7844663456154106,
      "grad_norm": 1.141681432723999,
      "learning_rate": 9.688078947368421e-05,
      "loss": 0.5572,
      "step": 31867
    },
    {
      "epoch": 1.7845223429275396,
      "grad_norm": 1.4665062427520752,
      "learning_rate": 9.688052631578947e-05,
      "loss": 0.5142,
      "step": 31868
    },
    {
      "epoch": 1.7845783402396687,
      "grad_norm": 1.1503149271011353,
      "learning_rate": 9.688026315789475e-05,
      "loss": 0.3358,
      "step": 31869
    },
    {
      "epoch": 1.7846343375517977,
      "grad_norm": 1.065591812133789,
      "learning_rate": 9.688000000000001e-05,
      "loss": 0.3363,
      "step": 31870
    },
    {
      "epoch": 1.7846903348639267,
      "grad_norm": 1.360927700996399,
      "learning_rate": 9.687973684210527e-05,
      "loss": 0.3881,
      "step": 31871
    },
    {
      "epoch": 1.7847463321760555,
      "grad_norm": 1.1576305627822876,
      "learning_rate": 9.687947368421053e-05,
      "loss": 0.4286,
      "step": 31872
    },
    {
      "epoch": 1.7848023294881845,
      "grad_norm": 1.765459656715393,
      "learning_rate": 9.687921052631579e-05,
      "loss": 0.6039,
      "step": 31873
    },
    {
      "epoch": 1.7848583268003135,
      "grad_norm": 1.1548362970352173,
      "learning_rate": 9.687894736842106e-05,
      "loss": 0.5625,
      "step": 31874
    },
    {
      "epoch": 1.7849143241124426,
      "grad_norm": 1.1519672870635986,
      "learning_rate": 9.687868421052632e-05,
      "loss": 0.4924,
      "step": 31875
    },
    {
      "epoch": 1.7849703214245716,
      "grad_norm": 1.2811933755874634,
      "learning_rate": 9.687842105263158e-05,
      "loss": 0.492,
      "step": 31876
    },
    {
      "epoch": 1.7850263187367006,
      "grad_norm": 1.201781153678894,
      "learning_rate": 9.687815789473684e-05,
      "loss": 0.3948,
      "step": 31877
    },
    {
      "epoch": 1.7850823160488296,
      "grad_norm": 1.2015191316604614,
      "learning_rate": 9.687789473684211e-05,
      "loss": 0.3957,
      "step": 31878
    },
    {
      "epoch": 1.7851383133609586,
      "grad_norm": 1.3081945180892944,
      "learning_rate": 9.687763157894737e-05,
      "loss": 0.4933,
      "step": 31879
    },
    {
      "epoch": 1.7851943106730876,
      "grad_norm": 1.3227640390396118,
      "learning_rate": 9.687736842105265e-05,
      "loss": 0.4365,
      "step": 31880
    },
    {
      "epoch": 1.7852503079852167,
      "grad_norm": 1.1125644445419312,
      "learning_rate": 9.687710526315789e-05,
      "loss": 0.4648,
      "step": 31881
    },
    {
      "epoch": 1.7853063052973457,
      "grad_norm": 1.2398297786712646,
      "learning_rate": 9.687684210526316e-05,
      "loss": 0.387,
      "step": 31882
    },
    {
      "epoch": 1.7853623026094747,
      "grad_norm": 1.2380250692367554,
      "learning_rate": 9.687657894736842e-05,
      "loss": 0.4519,
      "step": 31883
    },
    {
      "epoch": 1.7854182999216037,
      "grad_norm": 1.1946462392807007,
      "learning_rate": 9.68763157894737e-05,
      "loss": 0.476,
      "step": 31884
    },
    {
      "epoch": 1.7854742972337327,
      "grad_norm": 1.0793193578720093,
      "learning_rate": 9.687605263157894e-05,
      "loss": 0.4237,
      "step": 31885
    },
    {
      "epoch": 1.7855302945458618,
      "grad_norm": 1.292043685913086,
      "learning_rate": 9.687578947368422e-05,
      "loss": 0.4286,
      "step": 31886
    },
    {
      "epoch": 1.7855862918579908,
      "grad_norm": 1.3193550109863281,
      "learning_rate": 9.687552631578948e-05,
      "loss": 0.4822,
      "step": 31887
    },
    {
      "epoch": 1.7856422891701198,
      "grad_norm": 1.0490700006484985,
      "learning_rate": 9.687526315789475e-05,
      "loss": 0.4018,
      "step": 31888
    },
    {
      "epoch": 1.7856982864822488,
      "grad_norm": 2.052964210510254,
      "learning_rate": 9.687500000000001e-05,
      "loss": 0.5584,
      "step": 31889
    },
    {
      "epoch": 1.7857542837943778,
      "grad_norm": 1.2843341827392578,
      "learning_rate": 9.687473684210526e-05,
      "loss": 0.4075,
      "step": 31890
    },
    {
      "epoch": 1.7858102811065069,
      "grad_norm": 1.354588270187378,
      "learning_rate": 9.687447368421053e-05,
      "loss": 0.3725,
      "step": 31891
    },
    {
      "epoch": 1.7858662784186359,
      "grad_norm": 1.2183945178985596,
      "learning_rate": 9.687421052631579e-05,
      "loss": 0.4345,
      "step": 31892
    },
    {
      "epoch": 1.785922275730765,
      "grad_norm": 1.8452256917953491,
      "learning_rate": 9.687394736842106e-05,
      "loss": 0.4525,
      "step": 31893
    },
    {
      "epoch": 1.785978273042894,
      "grad_norm": 1.266893744468689,
      "learning_rate": 9.687368421052632e-05,
      "loss": 0.4361,
      "step": 31894
    },
    {
      "epoch": 1.786034270355023,
      "grad_norm": 1.3150335550308228,
      "learning_rate": 9.687342105263158e-05,
      "loss": 0.4323,
      "step": 31895
    },
    {
      "epoch": 1.786090267667152,
      "grad_norm": 1.535183072090149,
      "learning_rate": 9.687315789473684e-05,
      "loss": 0.4749,
      "step": 31896
    },
    {
      "epoch": 1.786146264979281,
      "grad_norm": 1.301011085510254,
      "learning_rate": 9.687289473684212e-05,
      "loss": 0.4234,
      "step": 31897
    },
    {
      "epoch": 1.78620226229141,
      "grad_norm": 1.994390606880188,
      "learning_rate": 9.687263157894737e-05,
      "loss": 0.4824,
      "step": 31898
    },
    {
      "epoch": 1.786258259603539,
      "grad_norm": 1.3605796098709106,
      "learning_rate": 9.687236842105263e-05,
      "loss": 0.457,
      "step": 31899
    },
    {
      "epoch": 1.786314256915668,
      "grad_norm": 1.096146583557129,
      "learning_rate": 9.68721052631579e-05,
      "loss": 0.3761,
      "step": 31900
    },
    {
      "epoch": 1.786370254227797,
      "grad_norm": 1.2055675983428955,
      "learning_rate": 9.687184210526317e-05,
      "loss": 0.4838,
      "step": 31901
    },
    {
      "epoch": 1.786426251539926,
      "grad_norm": 1.2578544616699219,
      "learning_rate": 9.687157894736843e-05,
      "loss": 0.4031,
      "step": 31902
    },
    {
      "epoch": 1.786482248852055,
      "grad_norm": 1.4588252305984497,
      "learning_rate": 9.687131578947369e-05,
      "loss": 0.4556,
      "step": 31903
    },
    {
      "epoch": 1.786538246164184,
      "grad_norm": 2.3814427852630615,
      "learning_rate": 9.687105263157895e-05,
      "loss": 0.4625,
      "step": 31904
    },
    {
      "epoch": 1.7865942434763131,
      "grad_norm": 1.379195213317871,
      "learning_rate": 9.687078947368422e-05,
      "loss": 0.495,
      "step": 31905
    },
    {
      "epoch": 1.7866502407884421,
      "grad_norm": 1.1372191905975342,
      "learning_rate": 9.687052631578948e-05,
      "loss": 0.3828,
      "step": 31906
    },
    {
      "epoch": 1.7867062381005712,
      "grad_norm": 1.2545710802078247,
      "learning_rate": 9.687026315789474e-05,
      "loss": 0.6164,
      "step": 31907
    },
    {
      "epoch": 1.7867622354127002,
      "grad_norm": 1.309998631477356,
      "learning_rate": 9.687e-05,
      "loss": 0.4626,
      "step": 31908
    },
    {
      "epoch": 1.7868182327248292,
      "grad_norm": 1.2679873704910278,
      "learning_rate": 9.686973684210526e-05,
      "loss": 0.4296,
      "step": 31909
    },
    {
      "epoch": 1.7868742300369582,
      "grad_norm": 1.700130581855774,
      "learning_rate": 9.686947368421053e-05,
      "loss": 0.5619,
      "step": 31910
    },
    {
      "epoch": 1.7869302273490872,
      "grad_norm": 1.367583155632019,
      "learning_rate": 9.686921052631579e-05,
      "loss": 0.6026,
      "step": 31911
    },
    {
      "epoch": 1.7869862246612163,
      "grad_norm": 1.5523277521133423,
      "learning_rate": 9.686894736842107e-05,
      "loss": 0.3854,
      "step": 31912
    },
    {
      "epoch": 1.7870422219733453,
      "grad_norm": 1.2146656513214111,
      "learning_rate": 9.686868421052631e-05,
      "loss": 0.4527,
      "step": 31913
    },
    {
      "epoch": 1.7870982192854743,
      "grad_norm": 1.5013121366500854,
      "learning_rate": 9.686842105263158e-05,
      "loss": 0.4224,
      "step": 31914
    },
    {
      "epoch": 1.7871542165976033,
      "grad_norm": 1.2803595066070557,
      "learning_rate": 9.686815789473684e-05,
      "loss": 0.4494,
      "step": 31915
    },
    {
      "epoch": 1.7872102139097323,
      "grad_norm": 1.6458550691604614,
      "learning_rate": 9.686789473684212e-05,
      "loss": 0.4969,
      "step": 31916
    },
    {
      "epoch": 1.7872662112218614,
      "grad_norm": 1.4115514755249023,
      "learning_rate": 9.686763157894738e-05,
      "loss": 0.4788,
      "step": 31917
    },
    {
      "epoch": 1.7873222085339904,
      "grad_norm": 1.4544378519058228,
      "learning_rate": 9.686736842105264e-05,
      "loss": 0.5141,
      "step": 31918
    },
    {
      "epoch": 1.7873782058461194,
      "grad_norm": 1.1197162866592407,
      "learning_rate": 9.68671052631579e-05,
      "loss": 0.4807,
      "step": 31919
    },
    {
      "epoch": 1.7874342031582484,
      "grad_norm": 1.3073194026947021,
      "learning_rate": 9.686684210526317e-05,
      "loss": 0.5076,
      "step": 31920
    },
    {
      "epoch": 1.7874902004703774,
      "grad_norm": 1.1209121942520142,
      "learning_rate": 9.686657894736843e-05,
      "loss": 0.4475,
      "step": 31921
    },
    {
      "epoch": 1.7875461977825065,
      "grad_norm": 1.3270071744918823,
      "learning_rate": 9.686631578947369e-05,
      "loss": 0.5461,
      "step": 31922
    },
    {
      "epoch": 1.7876021950946355,
      "grad_norm": 1.3773369789123535,
      "learning_rate": 9.686605263157895e-05,
      "loss": 0.3995,
      "step": 31923
    },
    {
      "epoch": 1.7876581924067645,
      "grad_norm": 1.25208580493927,
      "learning_rate": 9.686578947368421e-05,
      "loss": 0.422,
      "step": 31924
    },
    {
      "epoch": 1.7877141897188935,
      "grad_norm": 1.4375327825546265,
      "learning_rate": 9.686552631578948e-05,
      "loss": 0.4481,
      "step": 31925
    },
    {
      "epoch": 1.7877701870310225,
      "grad_norm": 1.1717000007629395,
      "learning_rate": 9.686526315789474e-05,
      "loss": 0.3859,
      "step": 31926
    },
    {
      "epoch": 1.7878261843431515,
      "grad_norm": 1.3534276485443115,
      "learning_rate": 9.6865e-05,
      "loss": 0.5196,
      "step": 31927
    },
    {
      "epoch": 1.7878821816552806,
      "grad_norm": 1.1850370168685913,
      "learning_rate": 9.686473684210526e-05,
      "loss": 0.4482,
      "step": 31928
    },
    {
      "epoch": 1.7879381789674096,
      "grad_norm": 1.0730195045471191,
      "learning_rate": 9.686447368421053e-05,
      "loss": 0.3513,
      "step": 31929
    },
    {
      "epoch": 1.7879941762795386,
      "grad_norm": 1.3491233587265015,
      "learning_rate": 9.68642105263158e-05,
      "loss": 0.466,
      "step": 31930
    },
    {
      "epoch": 1.7880501735916676,
      "grad_norm": 1.3401660919189453,
      "learning_rate": 9.686394736842105e-05,
      "loss": 0.5137,
      "step": 31931
    },
    {
      "epoch": 1.7881061709037966,
      "grad_norm": 1.0419365167617798,
      "learning_rate": 9.686368421052631e-05,
      "loss": 0.3629,
      "step": 31932
    },
    {
      "epoch": 1.7881621682159257,
      "grad_norm": 1.366532564163208,
      "learning_rate": 9.686342105263159e-05,
      "loss": 0.4584,
      "step": 31933
    },
    {
      "epoch": 1.7882181655280547,
      "grad_norm": 1.2646071910858154,
      "learning_rate": 9.686315789473685e-05,
      "loss": 0.5024,
      "step": 31934
    },
    {
      "epoch": 1.7882741628401837,
      "grad_norm": 1.360987663269043,
      "learning_rate": 9.686289473684212e-05,
      "loss": 0.3943,
      "step": 31935
    },
    {
      "epoch": 1.7883301601523127,
      "grad_norm": 1.5033934116363525,
      "learning_rate": 9.686263157894737e-05,
      "loss": 0.4614,
      "step": 31936
    },
    {
      "epoch": 1.7883861574644417,
      "grad_norm": 1.2284682989120483,
      "learning_rate": 9.686236842105264e-05,
      "loss": 0.4583,
      "step": 31937
    },
    {
      "epoch": 1.7884421547765708,
      "grad_norm": 1.2431849241256714,
      "learning_rate": 9.68621052631579e-05,
      "loss": 0.4344,
      "step": 31938
    },
    {
      "epoch": 1.7884981520886998,
      "grad_norm": 1.0632922649383545,
      "learning_rate": 9.686184210526317e-05,
      "loss": 0.3701,
      "step": 31939
    },
    {
      "epoch": 1.7885541494008288,
      "grad_norm": 1.4324793815612793,
      "learning_rate": 9.686157894736842e-05,
      "loss": 0.5288,
      "step": 31940
    },
    {
      "epoch": 1.7886101467129578,
      "grad_norm": 1.2565683126449585,
      "learning_rate": 9.686131578947368e-05,
      "loss": 0.4015,
      "step": 31941
    },
    {
      "epoch": 1.7886661440250868,
      "grad_norm": 1.568869709968567,
      "learning_rate": 9.686105263157895e-05,
      "loss": 0.4316,
      "step": 31942
    },
    {
      "epoch": 1.7887221413372159,
      "grad_norm": 1.173990249633789,
      "learning_rate": 9.686078947368421e-05,
      "loss": 0.4195,
      "step": 31943
    },
    {
      "epoch": 1.7887781386493449,
      "grad_norm": 1.4450441598892212,
      "learning_rate": 9.686052631578948e-05,
      "loss": 0.4647,
      "step": 31944
    },
    {
      "epoch": 1.788834135961474,
      "grad_norm": 1.2189844846725464,
      "learning_rate": 9.686026315789473e-05,
      "loss": 0.5044,
      "step": 31945
    },
    {
      "epoch": 1.788890133273603,
      "grad_norm": 1.457152009010315,
      "learning_rate": 9.686e-05,
      "loss": 0.5644,
      "step": 31946
    },
    {
      "epoch": 1.788946130585732,
      "grad_norm": 1.3943636417388916,
      "learning_rate": 9.685973684210526e-05,
      "loss": 0.363,
      "step": 31947
    },
    {
      "epoch": 1.789002127897861,
      "grad_norm": 1.5694568157196045,
      "learning_rate": 9.685947368421054e-05,
      "loss": 0.416,
      "step": 31948
    },
    {
      "epoch": 1.78905812520999,
      "grad_norm": 1.3233616352081299,
      "learning_rate": 9.68592105263158e-05,
      "loss": 0.5417,
      "step": 31949
    },
    {
      "epoch": 1.789114122522119,
      "grad_norm": 1.9907273054122925,
      "learning_rate": 9.685894736842106e-05,
      "loss": 0.4553,
      "step": 31950
    },
    {
      "epoch": 1.789170119834248,
      "grad_norm": 1.4094562530517578,
      "learning_rate": 9.685868421052632e-05,
      "loss": 0.5634,
      "step": 31951
    },
    {
      "epoch": 1.789226117146377,
      "grad_norm": 1.2532178163528442,
      "learning_rate": 9.685842105263159e-05,
      "loss": 0.4212,
      "step": 31952
    },
    {
      "epoch": 1.789282114458506,
      "grad_norm": 1.1450448036193848,
      "learning_rate": 9.685815789473685e-05,
      "loss": 0.4174,
      "step": 31953
    },
    {
      "epoch": 1.789338111770635,
      "grad_norm": 5.6791582107543945,
      "learning_rate": 9.685789473684211e-05,
      "loss": 0.4486,
      "step": 31954
    },
    {
      "epoch": 1.7893941090827639,
      "grad_norm": 1.223522424697876,
      "learning_rate": 9.685763157894737e-05,
      "loss": 0.5301,
      "step": 31955
    },
    {
      "epoch": 1.7894501063948929,
      "grad_norm": 1.2860651016235352,
      "learning_rate": 9.685736842105264e-05,
      "loss": 0.6105,
      "step": 31956
    },
    {
      "epoch": 1.789506103707022,
      "grad_norm": 1.1884591579437256,
      "learning_rate": 9.68571052631579e-05,
      "loss": 0.4867,
      "step": 31957
    },
    {
      "epoch": 1.789562101019151,
      "grad_norm": 1.3899544477462769,
      "learning_rate": 9.685684210526316e-05,
      "loss": 0.5301,
      "step": 31958
    },
    {
      "epoch": 1.78961809833128,
      "grad_norm": 1.0374916791915894,
      "learning_rate": 9.685657894736842e-05,
      "loss": 0.3354,
      "step": 31959
    },
    {
      "epoch": 1.789674095643409,
      "grad_norm": 1.2505470514297485,
      "learning_rate": 9.685631578947368e-05,
      "loss": 0.4811,
      "step": 31960
    },
    {
      "epoch": 1.789730092955538,
      "grad_norm": 3.660677671432495,
      "learning_rate": 9.685605263157895e-05,
      "loss": 0.4431,
      "step": 31961
    },
    {
      "epoch": 1.789786090267667,
      "grad_norm": 1.2718796730041504,
      "learning_rate": 9.685578947368421e-05,
      "loss": 0.439,
      "step": 31962
    },
    {
      "epoch": 1.789842087579796,
      "grad_norm": 1.0270181894302368,
      "learning_rate": 9.685552631578947e-05,
      "loss": 0.3657,
      "step": 31963
    },
    {
      "epoch": 1.789898084891925,
      "grad_norm": 1.5077179670333862,
      "learning_rate": 9.685526315789473e-05,
      "loss": 0.7349,
      "step": 31964
    },
    {
      "epoch": 1.789954082204054,
      "grad_norm": 1.1391503810882568,
      "learning_rate": 9.685500000000001e-05,
      "loss": 0.3734,
      "step": 31965
    },
    {
      "epoch": 1.790010079516183,
      "grad_norm": 1.2708402872085571,
      "learning_rate": 9.685473684210527e-05,
      "loss": 0.4743,
      "step": 31966
    },
    {
      "epoch": 1.790066076828312,
      "grad_norm": 1.5185188055038452,
      "learning_rate": 9.685447368421054e-05,
      "loss": 0.3838,
      "step": 31967
    },
    {
      "epoch": 1.790122074140441,
      "grad_norm": 1.0767401456832886,
      "learning_rate": 9.685421052631579e-05,
      "loss": 0.332,
      "step": 31968
    },
    {
      "epoch": 1.7901780714525701,
      "grad_norm": 2.1739299297332764,
      "learning_rate": 9.685394736842106e-05,
      "loss": 0.5035,
      "step": 31969
    },
    {
      "epoch": 1.7902340687646991,
      "grad_norm": 1.1915934085845947,
      "learning_rate": 9.685368421052632e-05,
      "loss": 0.3914,
      "step": 31970
    },
    {
      "epoch": 1.7902900660768282,
      "grad_norm": 1.3421036005020142,
      "learning_rate": 9.685342105263159e-05,
      "loss": 0.4314,
      "step": 31971
    },
    {
      "epoch": 1.7903460633889572,
      "grad_norm": 1.072020411491394,
      "learning_rate": 9.685315789473685e-05,
      "loss": 0.3705,
      "step": 31972
    },
    {
      "epoch": 1.7904020607010862,
      "grad_norm": 1.4651178121566772,
      "learning_rate": 9.685289473684211e-05,
      "loss": 0.4829,
      "step": 31973
    },
    {
      "epoch": 1.7904580580132152,
      "grad_norm": 1.6425532102584839,
      "learning_rate": 9.685263157894737e-05,
      "loss": 0.5433,
      "step": 31974
    },
    {
      "epoch": 1.7905140553253442,
      "grad_norm": 1.3565775156021118,
      "learning_rate": 9.685236842105264e-05,
      "loss": 0.4022,
      "step": 31975
    },
    {
      "epoch": 1.7905700526374733,
      "grad_norm": 1.4639005661010742,
      "learning_rate": 9.68521052631579e-05,
      "loss": 0.3853,
      "step": 31976
    },
    {
      "epoch": 1.7906260499496023,
      "grad_norm": 1.363292932510376,
      "learning_rate": 9.685184210526315e-05,
      "loss": 0.5404,
      "step": 31977
    },
    {
      "epoch": 1.7906820472617313,
      "grad_norm": 1.3989355564117432,
      "learning_rate": 9.685157894736842e-05,
      "loss": 0.439,
      "step": 31978
    },
    {
      "epoch": 1.7907380445738603,
      "grad_norm": 1.1867221593856812,
      "learning_rate": 9.685131578947368e-05,
      "loss": 0.351,
      "step": 31979
    },
    {
      "epoch": 1.7907940418859893,
      "grad_norm": 1.2818231582641602,
      "learning_rate": 9.685105263157896e-05,
      "loss": 0.3876,
      "step": 31980
    },
    {
      "epoch": 1.7908500391981184,
      "grad_norm": 1.0945988893508911,
      "learning_rate": 9.685078947368422e-05,
      "loss": 0.4235,
      "step": 31981
    },
    {
      "epoch": 1.7909060365102474,
      "grad_norm": 1.424734354019165,
      "learning_rate": 9.685052631578948e-05,
      "loss": 0.4813,
      "step": 31982
    },
    {
      "epoch": 1.7909620338223764,
      "grad_norm": 1.2352250814437866,
      "learning_rate": 9.685026315789474e-05,
      "loss": 0.4418,
      "step": 31983
    },
    {
      "epoch": 1.7910180311345054,
      "grad_norm": 1.2557188272476196,
      "learning_rate": 9.685000000000001e-05,
      "loss": 0.4996,
      "step": 31984
    },
    {
      "epoch": 1.7910740284466344,
      "grad_norm": 1.4204963445663452,
      "learning_rate": 9.684973684210527e-05,
      "loss": 0.5375,
      "step": 31985
    },
    {
      "epoch": 1.7911300257587635,
      "grad_norm": 1.0954538583755493,
      "learning_rate": 9.684947368421053e-05,
      "loss": 0.4255,
      "step": 31986
    },
    {
      "epoch": 1.7911860230708925,
      "grad_norm": 1.6155771017074585,
      "learning_rate": 9.684921052631579e-05,
      "loss": 0.5792,
      "step": 31987
    },
    {
      "epoch": 1.7912420203830215,
      "grad_norm": 1.3473423719406128,
      "learning_rate": 9.684894736842106e-05,
      "loss": 0.6393,
      "step": 31988
    },
    {
      "epoch": 1.7912980176951505,
      "grad_norm": 1.2579264640808105,
      "learning_rate": 9.684868421052632e-05,
      "loss": 0.4093,
      "step": 31989
    },
    {
      "epoch": 1.7913540150072795,
      "grad_norm": 1.2967183589935303,
      "learning_rate": 9.68484210526316e-05,
      "loss": 0.501,
      "step": 31990
    },
    {
      "epoch": 1.7914100123194086,
      "grad_norm": 1.2125723361968994,
      "learning_rate": 9.684815789473684e-05,
      "loss": 0.4765,
      "step": 31991
    },
    {
      "epoch": 1.7914660096315376,
      "grad_norm": 1.2356135845184326,
      "learning_rate": 9.684789473684211e-05,
      "loss": 0.4129,
      "step": 31992
    },
    {
      "epoch": 1.7915220069436666,
      "grad_norm": 1.2769567966461182,
      "learning_rate": 9.684763157894737e-05,
      "loss": 0.4385,
      "step": 31993
    },
    {
      "epoch": 1.7915780042557956,
      "grad_norm": 1.0065739154815674,
      "learning_rate": 9.684736842105263e-05,
      "loss": 0.381,
      "step": 31994
    },
    {
      "epoch": 1.7916340015679246,
      "grad_norm": 1.2326743602752686,
      "learning_rate": 9.68471052631579e-05,
      "loss": 0.5205,
      "step": 31995
    },
    {
      "epoch": 1.7916899988800536,
      "grad_norm": 1.2099952697753906,
      "learning_rate": 9.684684210526315e-05,
      "loss": 0.4812,
      "step": 31996
    },
    {
      "epoch": 1.7917459961921827,
      "grad_norm": 1.4518769979476929,
      "learning_rate": 9.684657894736843e-05,
      "loss": 0.5582,
      "step": 31997
    },
    {
      "epoch": 1.7918019935043117,
      "grad_norm": 1.2784314155578613,
      "learning_rate": 9.684631578947369e-05,
      "loss": 0.4311,
      "step": 31998
    },
    {
      "epoch": 1.7918579908164407,
      "grad_norm": 1.170072317123413,
      "learning_rate": 9.684605263157896e-05,
      "loss": 0.454,
      "step": 31999
    },
    {
      "epoch": 1.7919139881285697,
      "grad_norm": 1.7834513187408447,
      "learning_rate": 9.68457894736842e-05,
      "loss": 0.5746,
      "step": 32000
    },
    {
      "epoch": 1.7919699854406987,
      "grad_norm": 1.2325236797332764,
      "learning_rate": 9.684552631578948e-05,
      "loss": 0.4664,
      "step": 32001
    },
    {
      "epoch": 1.7920259827528278,
      "grad_norm": 1.687134861946106,
      "learning_rate": 9.684526315789474e-05,
      "loss": 0.635,
      "step": 32002
    },
    {
      "epoch": 1.7920819800649568,
      "grad_norm": 1.6744015216827393,
      "learning_rate": 9.684500000000001e-05,
      "loss": 0.6643,
      "step": 32003
    },
    {
      "epoch": 1.7921379773770858,
      "grad_norm": 1.270850658416748,
      "learning_rate": 9.684473684210527e-05,
      "loss": 0.4488,
      "step": 32004
    },
    {
      "epoch": 1.7921939746892148,
      "grad_norm": 1.3037875890731812,
      "learning_rate": 9.684447368421053e-05,
      "loss": 0.4997,
      "step": 32005
    },
    {
      "epoch": 1.7922499720013438,
      "grad_norm": 1.092988133430481,
      "learning_rate": 9.684421052631579e-05,
      "loss": 0.3485,
      "step": 32006
    },
    {
      "epoch": 1.7923059693134729,
      "grad_norm": 1.4699571132659912,
      "learning_rate": 9.684394736842106e-05,
      "loss": 0.5074,
      "step": 32007
    },
    {
      "epoch": 1.7923619666256019,
      "grad_norm": 1.195314645767212,
      "learning_rate": 9.684368421052632e-05,
      "loss": 0.3733,
      "step": 32008
    },
    {
      "epoch": 1.792417963937731,
      "grad_norm": 1.9048974514007568,
      "learning_rate": 9.684342105263158e-05,
      "loss": 0.4565,
      "step": 32009
    },
    {
      "epoch": 1.79247396124986,
      "grad_norm": 1.0765550136566162,
      "learning_rate": 9.684315789473684e-05,
      "loss": 0.4448,
      "step": 32010
    },
    {
      "epoch": 1.792529958561989,
      "grad_norm": 1.8611201047897339,
      "learning_rate": 9.68428947368421e-05,
      "loss": 0.6514,
      "step": 32011
    },
    {
      "epoch": 1.792585955874118,
      "grad_norm": 1.2042350769042969,
      "learning_rate": 9.684263157894738e-05,
      "loss": 0.3938,
      "step": 32012
    },
    {
      "epoch": 1.792641953186247,
      "grad_norm": 1.2873826026916504,
      "learning_rate": 9.684236842105264e-05,
      "loss": 0.463,
      "step": 32013
    },
    {
      "epoch": 1.792697950498376,
      "grad_norm": 1.1961383819580078,
      "learning_rate": 9.68421052631579e-05,
      "loss": 0.3753,
      "step": 32014
    },
    {
      "epoch": 1.792753947810505,
      "grad_norm": 1.4698563814163208,
      "learning_rate": 9.684184210526316e-05,
      "loss": 0.3901,
      "step": 32015
    },
    {
      "epoch": 1.792809945122634,
      "grad_norm": 1.2973783016204834,
      "learning_rate": 9.684157894736843e-05,
      "loss": 0.4112,
      "step": 32016
    },
    {
      "epoch": 1.792865942434763,
      "grad_norm": 1.3621046543121338,
      "learning_rate": 9.684131578947369e-05,
      "loss": 0.4495,
      "step": 32017
    },
    {
      "epoch": 1.792921939746892,
      "grad_norm": 1.4783670902252197,
      "learning_rate": 9.684105263157895e-05,
      "loss": 0.5238,
      "step": 32018
    },
    {
      "epoch": 1.792977937059021,
      "grad_norm": 1.2862879037857056,
      "learning_rate": 9.684078947368421e-05,
      "loss": 0.4602,
      "step": 32019
    },
    {
      "epoch": 1.79303393437115,
      "grad_norm": 1.5063560009002686,
      "learning_rate": 9.684052631578948e-05,
      "loss": 0.4594,
      "step": 32020
    },
    {
      "epoch": 1.7930899316832791,
      "grad_norm": 1.4773988723754883,
      "learning_rate": 9.684026315789474e-05,
      "loss": 0.6903,
      "step": 32021
    },
    {
      "epoch": 1.7931459289954081,
      "grad_norm": 1.222888469696045,
      "learning_rate": 9.684000000000001e-05,
      "loss": 0.4066,
      "step": 32022
    },
    {
      "epoch": 1.7932019263075372,
      "grad_norm": 0.9128860831260681,
      "learning_rate": 9.683973684210526e-05,
      "loss": 0.3873,
      "step": 32023
    },
    {
      "epoch": 1.7932579236196662,
      "grad_norm": 1.1375364065170288,
      "learning_rate": 9.683947368421053e-05,
      "loss": 0.3439,
      "step": 32024
    },
    {
      "epoch": 1.7933139209317952,
      "grad_norm": 1.0395842790603638,
      "learning_rate": 9.68392105263158e-05,
      "loss": 0.3702,
      "step": 32025
    },
    {
      "epoch": 1.7933699182439242,
      "grad_norm": 1.6284608840942383,
      "learning_rate": 9.683894736842107e-05,
      "loss": 0.5268,
      "step": 32026
    },
    {
      "epoch": 1.7934259155560532,
      "grad_norm": 1.306712031364441,
      "learning_rate": 9.683868421052633e-05,
      "loss": 0.6079,
      "step": 32027
    },
    {
      "epoch": 1.7934819128681823,
      "grad_norm": 1.5549505949020386,
      "learning_rate": 9.683842105263157e-05,
      "loss": 0.302,
      "step": 32028
    },
    {
      "epoch": 1.7935379101803113,
      "grad_norm": 1.2894011735916138,
      "learning_rate": 9.683815789473685e-05,
      "loss": 0.5774,
      "step": 32029
    },
    {
      "epoch": 1.7935939074924403,
      "grad_norm": 1.305033564567566,
      "learning_rate": 9.68378947368421e-05,
      "loss": 0.5882,
      "step": 32030
    },
    {
      "epoch": 1.7936499048045693,
      "grad_norm": 1.203795313835144,
      "learning_rate": 9.683763157894738e-05,
      "loss": 0.4354,
      "step": 32031
    },
    {
      "epoch": 1.7937059021166983,
      "grad_norm": 1.35097074508667,
      "learning_rate": 9.683736842105263e-05,
      "loss": 0.4676,
      "step": 32032
    },
    {
      "epoch": 1.7937618994288274,
      "grad_norm": 1.9423344135284424,
      "learning_rate": 9.68371052631579e-05,
      "loss": 0.4425,
      "step": 32033
    },
    {
      "epoch": 1.7938178967409564,
      "grad_norm": 1.3653912544250488,
      "learning_rate": 9.683684210526316e-05,
      "loss": 0.4419,
      "step": 32034
    },
    {
      "epoch": 1.7938738940530854,
      "grad_norm": 1.712348461151123,
      "learning_rate": 9.683657894736843e-05,
      "loss": 0.5226,
      "step": 32035
    },
    {
      "epoch": 1.7939298913652144,
      "grad_norm": 1.261184573173523,
      "learning_rate": 9.683631578947369e-05,
      "loss": 0.6008,
      "step": 32036
    },
    {
      "epoch": 1.7939858886773434,
      "grad_norm": 1.3308205604553223,
      "learning_rate": 9.683605263157895e-05,
      "loss": 0.4221,
      "step": 32037
    },
    {
      "epoch": 1.7940418859894725,
      "grad_norm": 1.3777445554733276,
      "learning_rate": 9.683578947368421e-05,
      "loss": 0.5306,
      "step": 32038
    },
    {
      "epoch": 1.7940978833016015,
      "grad_norm": 1.1889625787734985,
      "learning_rate": 9.683552631578948e-05,
      "loss": 0.4811,
      "step": 32039
    },
    {
      "epoch": 1.7941538806137305,
      "grad_norm": 1.2197210788726807,
      "learning_rate": 9.683526315789474e-05,
      "loss": 0.4505,
      "step": 32040
    },
    {
      "epoch": 1.7942098779258595,
      "grad_norm": 2.1066439151763916,
      "learning_rate": 9.6835e-05,
      "loss": 0.4286,
      "step": 32041
    },
    {
      "epoch": 1.7942658752379885,
      "grad_norm": 1.5215115547180176,
      "learning_rate": 9.683473684210526e-05,
      "loss": 0.4157,
      "step": 32042
    },
    {
      "epoch": 1.7943218725501175,
      "grad_norm": 1.2734651565551758,
      "learning_rate": 9.683447368421054e-05,
      "loss": 0.4494,
      "step": 32043
    },
    {
      "epoch": 1.7943778698622466,
      "grad_norm": 1.4336392879486084,
      "learning_rate": 9.68342105263158e-05,
      "loss": 0.5462,
      "step": 32044
    },
    {
      "epoch": 1.7944338671743756,
      "grad_norm": 1.415024757385254,
      "learning_rate": 9.683394736842106e-05,
      "loss": 0.5185,
      "step": 32045
    },
    {
      "epoch": 1.7944898644865046,
      "grad_norm": 1.0889935493469238,
      "learning_rate": 9.683368421052632e-05,
      "loss": 0.3652,
      "step": 32046
    },
    {
      "epoch": 1.7945458617986336,
      "grad_norm": 1.3765455484390259,
      "learning_rate": 9.683342105263158e-05,
      "loss": 0.5184,
      "step": 32047
    },
    {
      "epoch": 1.7946018591107626,
      "grad_norm": 1.792376160621643,
      "learning_rate": 9.683315789473685e-05,
      "loss": 0.4826,
      "step": 32048
    },
    {
      "epoch": 1.7946578564228917,
      "grad_norm": 1.0766433477401733,
      "learning_rate": 9.683289473684211e-05,
      "loss": 0.386,
      "step": 32049
    },
    {
      "epoch": 1.7947138537350207,
      "grad_norm": 1.2214709520339966,
      "learning_rate": 9.683263157894737e-05,
      "loss": 0.4808,
      "step": 32050
    },
    {
      "epoch": 1.7947698510471497,
      "grad_norm": 1.2832372188568115,
      "learning_rate": 9.683236842105263e-05,
      "loss": 0.4893,
      "step": 32051
    },
    {
      "epoch": 1.7948258483592787,
      "grad_norm": 2.639239549636841,
      "learning_rate": 9.68321052631579e-05,
      "loss": 0.7086,
      "step": 32052
    },
    {
      "epoch": 1.7948818456714077,
      "grad_norm": 1.6005090475082397,
      "learning_rate": 9.683184210526316e-05,
      "loss": 0.4862,
      "step": 32053
    },
    {
      "epoch": 1.7949378429835368,
      "grad_norm": 1.281174898147583,
      "learning_rate": 9.683157894736843e-05,
      "loss": 0.5162,
      "step": 32054
    },
    {
      "epoch": 1.7949938402956658,
      "grad_norm": 1.1073137521743774,
      "learning_rate": 9.683131578947368e-05,
      "loss": 0.4177,
      "step": 32055
    },
    {
      "epoch": 1.7950498376077948,
      "grad_norm": 1.5641192197799683,
      "learning_rate": 9.683105263157895e-05,
      "loss": 0.4397,
      "step": 32056
    },
    {
      "epoch": 1.7951058349199238,
      "grad_norm": 1.0586190223693848,
      "learning_rate": 9.683078947368421e-05,
      "loss": 0.3676,
      "step": 32057
    },
    {
      "epoch": 1.7951618322320528,
      "grad_norm": 1.2605412006378174,
      "learning_rate": 9.683052631578949e-05,
      "loss": 0.4423,
      "step": 32058
    },
    {
      "epoch": 1.7952178295441819,
      "grad_norm": 1.313506841659546,
      "learning_rate": 9.683026315789475e-05,
      "loss": 0.3356,
      "step": 32059
    },
    {
      "epoch": 1.7952738268563109,
      "grad_norm": 1.2559566497802734,
      "learning_rate": 9.683e-05,
      "loss": 0.5682,
      "step": 32060
    },
    {
      "epoch": 1.79532982416844,
      "grad_norm": 1.969361424446106,
      "learning_rate": 9.682973684210527e-05,
      "loss": 0.5176,
      "step": 32061
    },
    {
      "epoch": 1.795385821480569,
      "grad_norm": 1.121721863746643,
      "learning_rate": 9.682947368421053e-05,
      "loss": 0.3406,
      "step": 32062
    },
    {
      "epoch": 1.795441818792698,
      "grad_norm": 1.276720643043518,
      "learning_rate": 9.68292105263158e-05,
      "loss": 0.6006,
      "step": 32063
    },
    {
      "epoch": 1.795497816104827,
      "grad_norm": 1.2242330312728882,
      "learning_rate": 9.682894736842106e-05,
      "loss": 0.3921,
      "step": 32064
    },
    {
      "epoch": 1.795553813416956,
      "grad_norm": 1.31707763671875,
      "learning_rate": 9.682868421052632e-05,
      "loss": 0.4271,
      "step": 32065
    },
    {
      "epoch": 1.795609810729085,
      "grad_norm": 1.3440417051315308,
      "learning_rate": 9.682842105263158e-05,
      "loss": 0.4761,
      "step": 32066
    },
    {
      "epoch": 1.795665808041214,
      "grad_norm": 1.528328776359558,
      "learning_rate": 9.682815789473685e-05,
      "loss": 0.5074,
      "step": 32067
    },
    {
      "epoch": 1.795721805353343,
      "grad_norm": 1.130800724029541,
      "learning_rate": 9.682789473684211e-05,
      "loss": 0.3965,
      "step": 32068
    },
    {
      "epoch": 1.795777802665472,
      "grad_norm": 1.5954233407974243,
      "learning_rate": 9.682763157894737e-05,
      "loss": 0.4566,
      "step": 32069
    },
    {
      "epoch": 1.795833799977601,
      "grad_norm": 1.1379814147949219,
      "learning_rate": 9.682736842105263e-05,
      "loss": 0.4734,
      "step": 32070
    },
    {
      "epoch": 1.79588979728973,
      "grad_norm": 1.2935895919799805,
      "learning_rate": 9.68271052631579e-05,
      "loss": 0.3702,
      "step": 32071
    },
    {
      "epoch": 1.795945794601859,
      "grad_norm": 1.4047520160675049,
      "learning_rate": 9.682684210526316e-05,
      "loss": 0.5346,
      "step": 32072
    },
    {
      "epoch": 1.7960017919139881,
      "grad_norm": 1.4247809648513794,
      "learning_rate": 9.682657894736842e-05,
      "loss": 0.4072,
      "step": 32073
    },
    {
      "epoch": 1.7960577892261171,
      "grad_norm": 1.657099723815918,
      "learning_rate": 9.682631578947368e-05,
      "loss": 0.5088,
      "step": 32074
    },
    {
      "epoch": 1.7961137865382462,
      "grad_norm": 1.5160071849822998,
      "learning_rate": 9.682605263157896e-05,
      "loss": 0.5435,
      "step": 32075
    },
    {
      "epoch": 1.7961697838503752,
      "grad_norm": 1.0926828384399414,
      "learning_rate": 9.682578947368422e-05,
      "loss": 0.3499,
      "step": 32076
    },
    {
      "epoch": 1.7962257811625042,
      "grad_norm": 1.184757947921753,
      "learning_rate": 9.682552631578949e-05,
      "loss": 0.3446,
      "step": 32077
    },
    {
      "epoch": 1.7962817784746332,
      "grad_norm": 1.2774674892425537,
      "learning_rate": 9.682526315789474e-05,
      "loss": 0.4119,
      "step": 32078
    },
    {
      "epoch": 1.7963377757867622,
      "grad_norm": 1.1742981672286987,
      "learning_rate": 9.682500000000001e-05,
      "loss": 0.4501,
      "step": 32079
    },
    {
      "epoch": 1.7963937730988913,
      "grad_norm": 1.0841269493103027,
      "learning_rate": 9.682473684210527e-05,
      "loss": 0.3399,
      "step": 32080
    },
    {
      "epoch": 1.7964497704110203,
      "grad_norm": 1.441339373588562,
      "learning_rate": 9.682447368421053e-05,
      "loss": 0.6248,
      "step": 32081
    },
    {
      "epoch": 1.7965057677231493,
      "grad_norm": 1.1356699466705322,
      "learning_rate": 9.68242105263158e-05,
      "loss": 0.3581,
      "step": 32082
    },
    {
      "epoch": 1.7965617650352783,
      "grad_norm": 1.0941351652145386,
      "learning_rate": 9.682394736842105e-05,
      "loss": 0.4131,
      "step": 32083
    },
    {
      "epoch": 1.7966177623474073,
      "grad_norm": 1.2052688598632812,
      "learning_rate": 9.682368421052632e-05,
      "loss": 0.4974,
      "step": 32084
    },
    {
      "epoch": 1.7966737596595364,
      "grad_norm": 1.3144172430038452,
      "learning_rate": 9.682342105263158e-05,
      "loss": 0.5425,
      "step": 32085
    },
    {
      "epoch": 1.7967297569716654,
      "grad_norm": 1.3686069250106812,
      "learning_rate": 9.682315789473685e-05,
      "loss": 0.439,
      "step": 32086
    },
    {
      "epoch": 1.7967857542837944,
      "grad_norm": 1.1614123582839966,
      "learning_rate": 9.68228947368421e-05,
      "loss": 0.355,
      "step": 32087
    },
    {
      "epoch": 1.7968417515959234,
      "grad_norm": 1.3086341619491577,
      "learning_rate": 9.682263157894737e-05,
      "loss": 0.4057,
      "step": 32088
    },
    {
      "epoch": 1.7968977489080524,
      "grad_norm": 1.2517374753952026,
      "learning_rate": 9.682236842105263e-05,
      "loss": 0.4739,
      "step": 32089
    },
    {
      "epoch": 1.7969537462201814,
      "grad_norm": 1.3251738548278809,
      "learning_rate": 9.68221052631579e-05,
      "loss": 0.4241,
      "step": 32090
    },
    {
      "epoch": 1.7970097435323105,
      "grad_norm": 1.3720954656600952,
      "learning_rate": 9.682184210526317e-05,
      "loss": 0.5435,
      "step": 32091
    },
    {
      "epoch": 1.7970657408444395,
      "grad_norm": 1.362906813621521,
      "learning_rate": 9.682157894736843e-05,
      "loss": 0.4403,
      "step": 32092
    },
    {
      "epoch": 1.7971217381565685,
      "grad_norm": 1.5304017066955566,
      "learning_rate": 9.682131578947369e-05,
      "loss": 0.3982,
      "step": 32093
    },
    {
      "epoch": 1.7971777354686975,
      "grad_norm": 1.428594946861267,
      "learning_rate": 9.682105263157896e-05,
      "loss": 0.4535,
      "step": 32094
    },
    {
      "epoch": 1.7972337327808265,
      "grad_norm": 1.2646770477294922,
      "learning_rate": 9.682078947368422e-05,
      "loss": 0.4245,
      "step": 32095
    },
    {
      "epoch": 1.7972897300929556,
      "grad_norm": 1.387730598449707,
      "learning_rate": 9.682052631578948e-05,
      "loss": 0.3672,
      "step": 32096
    },
    {
      "epoch": 1.7973457274050846,
      "grad_norm": 1.1369807720184326,
      "learning_rate": 9.682026315789474e-05,
      "loss": 0.4255,
      "step": 32097
    },
    {
      "epoch": 1.7974017247172136,
      "grad_norm": 1.5108939409255981,
      "learning_rate": 9.682e-05,
      "loss": 0.5703,
      "step": 32098
    },
    {
      "epoch": 1.7974577220293426,
      "grad_norm": 1.3130950927734375,
      "learning_rate": 9.681973684210527e-05,
      "loss": 0.503,
      "step": 32099
    },
    {
      "epoch": 1.7975137193414716,
      "grad_norm": 1.1229069232940674,
      "learning_rate": 9.681947368421053e-05,
      "loss": 0.3066,
      "step": 32100
    },
    {
      "epoch": 1.7975697166536007,
      "grad_norm": 1.3163306713104248,
      "learning_rate": 9.681921052631579e-05,
      "loss": 0.577,
      "step": 32101
    },
    {
      "epoch": 1.7976257139657297,
      "grad_norm": 1.2117362022399902,
      "learning_rate": 9.681894736842105e-05,
      "loss": 0.4055,
      "step": 32102
    },
    {
      "epoch": 1.7976817112778587,
      "grad_norm": 1.4718432426452637,
      "learning_rate": 9.681868421052632e-05,
      "loss": 0.453,
      "step": 32103
    },
    {
      "epoch": 1.7977377085899877,
      "grad_norm": 2.0896828174591064,
      "learning_rate": 9.681842105263158e-05,
      "loss": 0.3959,
      "step": 32104
    },
    {
      "epoch": 1.7977937059021167,
      "grad_norm": 1.7026898860931396,
      "learning_rate": 9.681815789473684e-05,
      "loss": 0.6358,
      "step": 32105
    },
    {
      "epoch": 1.7978497032142458,
      "grad_norm": 1.3427798748016357,
      "learning_rate": 9.68178947368421e-05,
      "loss": 0.5346,
      "step": 32106
    },
    {
      "epoch": 1.7979057005263748,
      "grad_norm": 1.5636309385299683,
      "learning_rate": 9.681763157894738e-05,
      "loss": 0.4554,
      "step": 32107
    },
    {
      "epoch": 1.7979616978385038,
      "grad_norm": 1.7977672815322876,
      "learning_rate": 9.681736842105264e-05,
      "loss": 0.4401,
      "step": 32108
    },
    {
      "epoch": 1.7980176951506328,
      "grad_norm": 1.3516571521759033,
      "learning_rate": 9.681710526315791e-05,
      "loss": 0.4898,
      "step": 32109
    },
    {
      "epoch": 1.7980736924627618,
      "grad_norm": 1.267030119895935,
      "learning_rate": 9.681684210526316e-05,
      "loss": 0.4257,
      "step": 32110
    },
    {
      "epoch": 1.7981296897748908,
      "grad_norm": 0.9707457423210144,
      "learning_rate": 9.681657894736843e-05,
      "loss": 0.402,
      "step": 32111
    },
    {
      "epoch": 1.7981856870870199,
      "grad_norm": 1.283591389656067,
      "learning_rate": 9.681631578947369e-05,
      "loss": 0.3039,
      "step": 32112
    },
    {
      "epoch": 1.7982416843991489,
      "grad_norm": 1.1905903816223145,
      "learning_rate": 9.681605263157896e-05,
      "loss": 0.4965,
      "step": 32113
    },
    {
      "epoch": 1.798297681711278,
      "grad_norm": 1.1125209331512451,
      "learning_rate": 9.681578947368422e-05,
      "loss": 0.3219,
      "step": 32114
    },
    {
      "epoch": 1.798353679023407,
      "grad_norm": 1.3652342557907104,
      "learning_rate": 9.681552631578947e-05,
      "loss": 0.4651,
      "step": 32115
    },
    {
      "epoch": 1.798409676335536,
      "grad_norm": 1.111520528793335,
      "learning_rate": 9.681526315789474e-05,
      "loss": 0.3665,
      "step": 32116
    },
    {
      "epoch": 1.798465673647665,
      "grad_norm": 1.493578314781189,
      "learning_rate": 9.6815e-05,
      "loss": 0.5879,
      "step": 32117
    },
    {
      "epoch": 1.798521670959794,
      "grad_norm": 1.2755286693572998,
      "learning_rate": 9.681473684210527e-05,
      "loss": 0.3884,
      "step": 32118
    },
    {
      "epoch": 1.798577668271923,
      "grad_norm": 1.6973812580108643,
      "learning_rate": 9.681447368421053e-05,
      "loss": 0.4738,
      "step": 32119
    },
    {
      "epoch": 1.798633665584052,
      "grad_norm": 1.2047200202941895,
      "learning_rate": 9.68142105263158e-05,
      "loss": 0.4954,
      "step": 32120
    },
    {
      "epoch": 1.798689662896181,
      "grad_norm": 1.3098191022872925,
      "learning_rate": 9.681394736842105e-05,
      "loss": 0.4377,
      "step": 32121
    },
    {
      "epoch": 1.79874566020831,
      "grad_norm": 1.3078348636627197,
      "learning_rate": 9.681368421052633e-05,
      "loss": 0.3742,
      "step": 32122
    },
    {
      "epoch": 1.798801657520439,
      "grad_norm": 4.851028919219971,
      "learning_rate": 9.681342105263159e-05,
      "loss": 0.5733,
      "step": 32123
    },
    {
      "epoch": 1.798857654832568,
      "grad_norm": 1.3678643703460693,
      "learning_rate": 9.681315789473685e-05,
      "loss": 0.352,
      "step": 32124
    },
    {
      "epoch": 1.7989136521446971,
      "grad_norm": 1.1013264656066895,
      "learning_rate": 9.68128947368421e-05,
      "loss": 0.4205,
      "step": 32125
    },
    {
      "epoch": 1.7989696494568261,
      "grad_norm": 1.7529017925262451,
      "learning_rate": 9.681263157894738e-05,
      "loss": 0.6738,
      "step": 32126
    },
    {
      "epoch": 1.7990256467689552,
      "grad_norm": 1.278640627861023,
      "learning_rate": 9.681236842105264e-05,
      "loss": 0.3978,
      "step": 32127
    },
    {
      "epoch": 1.7990816440810842,
      "grad_norm": 1.36661958694458,
      "learning_rate": 9.68121052631579e-05,
      "loss": 0.4554,
      "step": 32128
    },
    {
      "epoch": 1.7991376413932132,
      "grad_norm": 1.1438835859298706,
      "learning_rate": 9.681184210526316e-05,
      "loss": 0.3538,
      "step": 32129
    },
    {
      "epoch": 1.7991936387053422,
      "grad_norm": 1.1635104417800903,
      "learning_rate": 9.681157894736843e-05,
      "loss": 0.3943,
      "step": 32130
    },
    {
      "epoch": 1.7992496360174712,
      "grad_norm": 1.2324329614639282,
      "learning_rate": 9.681131578947369e-05,
      "loss": 0.3721,
      "step": 32131
    },
    {
      "epoch": 1.7993056333296003,
      "grad_norm": 1.1140331029891968,
      "learning_rate": 9.681105263157895e-05,
      "loss": 0.3159,
      "step": 32132
    },
    {
      "epoch": 1.7993616306417293,
      "grad_norm": 1.3687115907669067,
      "learning_rate": 9.681078947368421e-05,
      "loss": 0.5195,
      "step": 32133
    },
    {
      "epoch": 1.7994176279538583,
      "grad_norm": 1.2846589088439941,
      "learning_rate": 9.681052631578947e-05,
      "loss": 0.4109,
      "step": 32134
    },
    {
      "epoch": 1.7994736252659873,
      "grad_norm": 1.5153899192810059,
      "learning_rate": 9.681026315789474e-05,
      "loss": 0.3978,
      "step": 32135
    },
    {
      "epoch": 1.7995296225781163,
      "grad_norm": 1.623278021812439,
      "learning_rate": 9.681e-05,
      "loss": 0.5264,
      "step": 32136
    },
    {
      "epoch": 1.7995856198902453,
      "grad_norm": 1.4748739004135132,
      "learning_rate": 9.680973684210528e-05,
      "loss": 0.4594,
      "step": 32137
    },
    {
      "epoch": 1.7996416172023744,
      "grad_norm": 1.1571379899978638,
      "learning_rate": 9.680947368421052e-05,
      "loss": 0.3855,
      "step": 32138
    },
    {
      "epoch": 1.7996976145145034,
      "grad_norm": 1.15643310546875,
      "learning_rate": 9.68092105263158e-05,
      "loss": 0.5464,
      "step": 32139
    },
    {
      "epoch": 1.7997536118266324,
      "grad_norm": 1.2689732313156128,
      "learning_rate": 9.680894736842106e-05,
      "loss": 0.5113,
      "step": 32140
    },
    {
      "epoch": 1.7998096091387614,
      "grad_norm": 1.3981406688690186,
      "learning_rate": 9.680868421052633e-05,
      "loss": 0.4687,
      "step": 32141
    },
    {
      "epoch": 1.7998656064508904,
      "grad_norm": 1.1974506378173828,
      "learning_rate": 9.680842105263158e-05,
      "loss": 0.4869,
      "step": 32142
    },
    {
      "epoch": 1.7999216037630195,
      "grad_norm": 1.3745614290237427,
      "learning_rate": 9.680815789473685e-05,
      "loss": 0.5289,
      "step": 32143
    },
    {
      "epoch": 1.7999776010751485,
      "grad_norm": 1.3292382955551147,
      "learning_rate": 9.680789473684211e-05,
      "loss": 0.3645,
      "step": 32144
    },
    {
      "epoch": 1.8000335983872775,
      "grad_norm": 1.2569174766540527,
      "learning_rate": 9.680763157894738e-05,
      "loss": 0.3685,
      "step": 32145
    },
    {
      "epoch": 1.8000895956994065,
      "grad_norm": 1.0837546586990356,
      "learning_rate": 9.680736842105264e-05,
      "loss": 0.3407,
      "step": 32146
    },
    {
      "epoch": 1.8001455930115355,
      "grad_norm": 1.473798155784607,
      "learning_rate": 9.68071052631579e-05,
      "loss": 0.4752,
      "step": 32147
    },
    {
      "epoch": 1.8002015903236646,
      "grad_norm": 1.2655588388442993,
      "learning_rate": 9.680684210526316e-05,
      "loss": 0.4597,
      "step": 32148
    },
    {
      "epoch": 1.8002575876357936,
      "grad_norm": 1.2842222452163696,
      "learning_rate": 9.680657894736842e-05,
      "loss": 0.3583,
      "step": 32149
    },
    {
      "epoch": 1.8003135849479226,
      "grad_norm": 1.1543163061141968,
      "learning_rate": 9.68063157894737e-05,
      "loss": 0.3795,
      "step": 32150
    },
    {
      "epoch": 1.8003695822600516,
      "grad_norm": 1.3438326120376587,
      "learning_rate": 9.680605263157895e-05,
      "loss": 0.4056,
      "step": 32151
    },
    {
      "epoch": 1.8004255795721806,
      "grad_norm": 1.7318296432495117,
      "learning_rate": 9.680578947368421e-05,
      "loss": 0.5532,
      "step": 32152
    },
    {
      "epoch": 1.8004815768843097,
      "grad_norm": 1.3657329082489014,
      "learning_rate": 9.680552631578947e-05,
      "loss": 0.5227,
      "step": 32153
    },
    {
      "epoch": 1.8005375741964387,
      "grad_norm": 1.2669062614440918,
      "learning_rate": 9.680526315789475e-05,
      "loss": 0.4564,
      "step": 32154
    },
    {
      "epoch": 1.8005935715085677,
      "grad_norm": 1.3922449350357056,
      "learning_rate": 9.6805e-05,
      "loss": 0.4705,
      "step": 32155
    },
    {
      "epoch": 1.8006495688206967,
      "grad_norm": 1.5206490755081177,
      "learning_rate": 9.680473684210527e-05,
      "loss": 0.5274,
      "step": 32156
    },
    {
      "epoch": 1.8007055661328257,
      "grad_norm": 1.0941153764724731,
      "learning_rate": 9.680447368421053e-05,
      "loss": 0.3552,
      "step": 32157
    },
    {
      "epoch": 1.8007615634449547,
      "grad_norm": 1.8165775537490845,
      "learning_rate": 9.68042105263158e-05,
      "loss": 0.606,
      "step": 32158
    },
    {
      "epoch": 1.8008175607570838,
      "grad_norm": 1.0419973134994507,
      "learning_rate": 9.680394736842106e-05,
      "loss": 0.4076,
      "step": 32159
    },
    {
      "epoch": 1.8008735580692128,
      "grad_norm": 1.3721517324447632,
      "learning_rate": 9.680368421052632e-05,
      "loss": 0.5063,
      "step": 32160
    },
    {
      "epoch": 1.8009295553813418,
      "grad_norm": 1.3869708776474,
      "learning_rate": 9.680342105263158e-05,
      "loss": 0.4236,
      "step": 32161
    },
    {
      "epoch": 1.8009855526934708,
      "grad_norm": 1.7303255796432495,
      "learning_rate": 9.680315789473685e-05,
      "loss": 0.4363,
      "step": 32162
    },
    {
      "epoch": 1.8010415500055998,
      "grad_norm": 1.1910163164138794,
      "learning_rate": 9.680289473684211e-05,
      "loss": 0.4146,
      "step": 32163
    },
    {
      "epoch": 1.8010975473177289,
      "grad_norm": 1.8176252841949463,
      "learning_rate": 9.680263157894738e-05,
      "loss": 0.6962,
      "step": 32164
    },
    {
      "epoch": 1.8011535446298579,
      "grad_norm": 1.0348756313323975,
      "learning_rate": 9.680236842105263e-05,
      "loss": 0.3322,
      "step": 32165
    },
    {
      "epoch": 1.801209541941987,
      "grad_norm": 1.2159212827682495,
      "learning_rate": 9.680210526315789e-05,
      "loss": 0.4055,
      "step": 32166
    },
    {
      "epoch": 1.801265539254116,
      "grad_norm": 1.4628657102584839,
      "learning_rate": 9.680184210526316e-05,
      "loss": 0.4393,
      "step": 32167
    },
    {
      "epoch": 1.801321536566245,
      "grad_norm": 1.2295887470245361,
      "learning_rate": 9.680157894736842e-05,
      "loss": 0.5086,
      "step": 32168
    },
    {
      "epoch": 1.801377533878374,
      "grad_norm": 1.1949251890182495,
      "learning_rate": 9.68013157894737e-05,
      "loss": 0.4459,
      "step": 32169
    },
    {
      "epoch": 1.801433531190503,
      "grad_norm": 1.3208088874816895,
      "learning_rate": 9.680105263157894e-05,
      "loss": 0.6299,
      "step": 32170
    },
    {
      "epoch": 1.801489528502632,
      "grad_norm": 1.6019909381866455,
      "learning_rate": 9.680078947368422e-05,
      "loss": 0.5921,
      "step": 32171
    },
    {
      "epoch": 1.801545525814761,
      "grad_norm": 1.39640212059021,
      "learning_rate": 9.680052631578948e-05,
      "loss": 0.4855,
      "step": 32172
    },
    {
      "epoch": 1.80160152312689,
      "grad_norm": 1.4144102334976196,
      "learning_rate": 9.680026315789475e-05,
      "loss": 0.4963,
      "step": 32173
    },
    {
      "epoch": 1.801657520439019,
      "grad_norm": 1.6579128503799438,
      "learning_rate": 9.680000000000001e-05,
      "loss": 0.2844,
      "step": 32174
    },
    {
      "epoch": 1.801713517751148,
      "grad_norm": 1.77386474609375,
      "learning_rate": 9.679973684210527e-05,
      "loss": 0.5623,
      "step": 32175
    },
    {
      "epoch": 1.801769515063277,
      "grad_norm": 1.1909027099609375,
      "learning_rate": 9.679947368421053e-05,
      "loss": 0.3624,
      "step": 32176
    },
    {
      "epoch": 1.8018255123754061,
      "grad_norm": 1.2934516668319702,
      "learning_rate": 9.67992105263158e-05,
      "loss": 0.3771,
      "step": 32177
    },
    {
      "epoch": 1.8018815096875351,
      "grad_norm": 1.3932722806930542,
      "learning_rate": 9.679894736842106e-05,
      "loss": 0.4694,
      "step": 32178
    },
    {
      "epoch": 1.8019375069996642,
      "grad_norm": 1.1344374418258667,
      "learning_rate": 9.679868421052632e-05,
      "loss": 0.4734,
      "step": 32179
    },
    {
      "epoch": 1.8019935043117932,
      "grad_norm": 1.27701735496521,
      "learning_rate": 9.679842105263158e-05,
      "loss": 0.4842,
      "step": 32180
    },
    {
      "epoch": 1.8020495016239222,
      "grad_norm": 1.308442234992981,
      "learning_rate": 9.679815789473685e-05,
      "loss": 0.3844,
      "step": 32181
    },
    {
      "epoch": 1.8021054989360512,
      "grad_norm": 1.227443814277649,
      "learning_rate": 9.679789473684211e-05,
      "loss": 0.4875,
      "step": 32182
    },
    {
      "epoch": 1.8021614962481802,
      "grad_norm": 1.7548967599868774,
      "learning_rate": 9.679763157894737e-05,
      "loss": 0.424,
      "step": 32183
    },
    {
      "epoch": 1.8022174935603092,
      "grad_norm": 1.1256011724472046,
      "learning_rate": 9.679736842105263e-05,
      "loss": 0.4127,
      "step": 32184
    },
    {
      "epoch": 1.8022734908724383,
      "grad_norm": 1.2418335676193237,
      "learning_rate": 9.679710526315789e-05,
      "loss": 0.3754,
      "step": 32185
    },
    {
      "epoch": 1.8023294881845673,
      "grad_norm": 1.2596659660339355,
      "learning_rate": 9.679684210526317e-05,
      "loss": 0.4198,
      "step": 32186
    },
    {
      "epoch": 1.8023854854966963,
      "grad_norm": 1.1821552515029907,
      "learning_rate": 9.679657894736843e-05,
      "loss": 0.398,
      "step": 32187
    },
    {
      "epoch": 1.8024414828088253,
      "grad_norm": 1.5141074657440186,
      "learning_rate": 9.679631578947369e-05,
      "loss": 0.5065,
      "step": 32188
    },
    {
      "epoch": 1.8024974801209543,
      "grad_norm": 1.3615210056304932,
      "learning_rate": 9.679605263157894e-05,
      "loss": 0.3789,
      "step": 32189
    },
    {
      "epoch": 1.8025534774330834,
      "grad_norm": 1.0209513902664185,
      "learning_rate": 9.679578947368422e-05,
      "loss": 0.3548,
      "step": 32190
    },
    {
      "epoch": 1.8026094747452124,
      "grad_norm": 1.6605288982391357,
      "learning_rate": 9.679552631578948e-05,
      "loss": 0.3921,
      "step": 32191
    },
    {
      "epoch": 1.8026654720573414,
      "grad_norm": 1.5985584259033203,
      "learning_rate": 9.679526315789475e-05,
      "loss": 0.2913,
      "step": 32192
    },
    {
      "epoch": 1.8027214693694704,
      "grad_norm": 1.179266333580017,
      "learning_rate": 9.6795e-05,
      "loss": 0.3706,
      "step": 32193
    },
    {
      "epoch": 1.8027774666815994,
      "grad_norm": 1.19562828540802,
      "learning_rate": 9.679473684210527e-05,
      "loss": 0.4548,
      "step": 32194
    },
    {
      "epoch": 1.8028334639937285,
      "grad_norm": 1.4101299047470093,
      "learning_rate": 9.679447368421053e-05,
      "loss": 0.5092,
      "step": 32195
    },
    {
      "epoch": 1.8028894613058575,
      "grad_norm": 1.1909759044647217,
      "learning_rate": 9.67942105263158e-05,
      "loss": 0.4377,
      "step": 32196
    },
    {
      "epoch": 1.8029454586179865,
      "grad_norm": 1.3512917757034302,
      "learning_rate": 9.679394736842105e-05,
      "loss": 0.472,
      "step": 32197
    },
    {
      "epoch": 1.8030014559301155,
      "grad_norm": 0.9307965636253357,
      "learning_rate": 9.679368421052632e-05,
      "loss": 0.4077,
      "step": 32198
    },
    {
      "epoch": 1.8030574532422445,
      "grad_norm": 1.3607311248779297,
      "learning_rate": 9.679342105263158e-05,
      "loss": 0.4763,
      "step": 32199
    },
    {
      "epoch": 1.8031134505543736,
      "grad_norm": 1.2147271633148193,
      "learning_rate": 9.679315789473686e-05,
      "loss": 0.5062,
      "step": 32200
    },
    {
      "epoch": 1.8031694478665026,
      "grad_norm": 1.5019962787628174,
      "learning_rate": 9.679289473684212e-05,
      "loss": 0.4833,
      "step": 32201
    },
    {
      "epoch": 1.8032254451786316,
      "grad_norm": 1.2628660202026367,
      "learning_rate": 9.679263157894736e-05,
      "loss": 0.4927,
      "step": 32202
    },
    {
      "epoch": 1.8032814424907604,
      "grad_norm": 1.1674035787582397,
      "learning_rate": 9.679236842105264e-05,
      "loss": 0.4991,
      "step": 32203
    },
    {
      "epoch": 1.8033374398028894,
      "grad_norm": 1.3083025217056274,
      "learning_rate": 9.67921052631579e-05,
      "loss": 0.4124,
      "step": 32204
    },
    {
      "epoch": 1.8033934371150184,
      "grad_norm": 1.4513611793518066,
      "learning_rate": 9.679184210526317e-05,
      "loss": 0.4372,
      "step": 32205
    },
    {
      "epoch": 1.8034494344271474,
      "grad_norm": 1.7680569887161255,
      "learning_rate": 9.679157894736843e-05,
      "loss": 0.4754,
      "step": 32206
    },
    {
      "epoch": 1.8035054317392765,
      "grad_norm": 1.0999014377593994,
      "learning_rate": 9.679131578947369e-05,
      "loss": 0.4982,
      "step": 32207
    },
    {
      "epoch": 1.8035614290514055,
      "grad_norm": 1.3361412286758423,
      "learning_rate": 9.679105263157895e-05,
      "loss": 0.4791,
      "step": 32208
    },
    {
      "epoch": 1.8036174263635345,
      "grad_norm": 1.3928239345550537,
      "learning_rate": 9.679078947368422e-05,
      "loss": 0.5503,
      "step": 32209
    },
    {
      "epoch": 1.8036734236756635,
      "grad_norm": 1.30655038356781,
      "learning_rate": 9.679052631578948e-05,
      "loss": 0.45,
      "step": 32210
    },
    {
      "epoch": 1.8037294209877925,
      "grad_norm": 1.2056602239608765,
      "learning_rate": 9.679026315789474e-05,
      "loss": 0.4359,
      "step": 32211
    },
    {
      "epoch": 1.8037854182999216,
      "grad_norm": 1.2480945587158203,
      "learning_rate": 9.679e-05,
      "loss": 0.503,
      "step": 32212
    },
    {
      "epoch": 1.8038414156120506,
      "grad_norm": 1.1976988315582275,
      "learning_rate": 9.678973684210527e-05,
      "loss": 0.3591,
      "step": 32213
    },
    {
      "epoch": 1.8038974129241796,
      "grad_norm": 1.318771243095398,
      "learning_rate": 9.678947368421053e-05,
      "loss": 0.5894,
      "step": 32214
    },
    {
      "epoch": 1.8039534102363086,
      "grad_norm": 1.0990079641342163,
      "learning_rate": 9.678921052631579e-05,
      "loss": 0.4771,
      "step": 32215
    },
    {
      "epoch": 1.8040094075484376,
      "grad_norm": 1.2563648223876953,
      "learning_rate": 9.678894736842105e-05,
      "loss": 0.5148,
      "step": 32216
    },
    {
      "epoch": 1.8040654048605667,
      "grad_norm": 1.1544779539108276,
      "learning_rate": 9.678868421052633e-05,
      "loss": 0.4065,
      "step": 32217
    },
    {
      "epoch": 1.8041214021726957,
      "grad_norm": 1.2515053749084473,
      "learning_rate": 9.678842105263159e-05,
      "loss": 0.3505,
      "step": 32218
    },
    {
      "epoch": 1.8041773994848247,
      "grad_norm": 1.3231868743896484,
      "learning_rate": 9.678815789473685e-05,
      "loss": 0.4272,
      "step": 32219
    },
    {
      "epoch": 1.8042333967969537,
      "grad_norm": 1.3882676362991333,
      "learning_rate": 9.67878947368421e-05,
      "loss": 0.4257,
      "step": 32220
    },
    {
      "epoch": 1.8042893941090827,
      "grad_norm": 1.2684223651885986,
      "learning_rate": 9.678763157894736e-05,
      "loss": 0.5404,
      "step": 32221
    },
    {
      "epoch": 1.8043453914212118,
      "grad_norm": 1.5299326181411743,
      "learning_rate": 9.678736842105264e-05,
      "loss": 0.4626,
      "step": 32222
    },
    {
      "epoch": 1.8044013887333408,
      "grad_norm": 1.2657458782196045,
      "learning_rate": 9.67871052631579e-05,
      "loss": 0.4272,
      "step": 32223
    },
    {
      "epoch": 1.8044573860454698,
      "grad_norm": 1.9942419528961182,
      "learning_rate": 9.678684210526317e-05,
      "loss": 0.4734,
      "step": 32224
    },
    {
      "epoch": 1.8045133833575988,
      "grad_norm": 1.2680203914642334,
      "learning_rate": 9.678657894736842e-05,
      "loss": 0.4573,
      "step": 32225
    },
    {
      "epoch": 1.8045693806697278,
      "grad_norm": 1.1932395696640015,
      "learning_rate": 9.678631578947369e-05,
      "loss": 0.4803,
      "step": 32226
    },
    {
      "epoch": 1.8046253779818568,
      "grad_norm": 1.2549781799316406,
      "learning_rate": 9.678605263157895e-05,
      "loss": 0.5361,
      "step": 32227
    },
    {
      "epoch": 1.8046813752939859,
      "grad_norm": 1.649200677871704,
      "learning_rate": 9.678578947368422e-05,
      "loss": 0.4761,
      "step": 32228
    },
    {
      "epoch": 1.8047373726061149,
      "grad_norm": 1.0116018056869507,
      "learning_rate": 9.678552631578948e-05,
      "loss": 0.2765,
      "step": 32229
    },
    {
      "epoch": 1.804793369918244,
      "grad_norm": 1.1805944442749023,
      "learning_rate": 9.678526315789474e-05,
      "loss": 0.4441,
      "step": 32230
    },
    {
      "epoch": 1.804849367230373,
      "grad_norm": 1.7720093727111816,
      "learning_rate": 9.6785e-05,
      "loss": 0.4193,
      "step": 32231
    },
    {
      "epoch": 1.804905364542502,
      "grad_norm": 1.3531529903411865,
      "learning_rate": 9.678473684210528e-05,
      "loss": 0.6227,
      "step": 32232
    },
    {
      "epoch": 1.804961361854631,
      "grad_norm": 1.4623477458953857,
      "learning_rate": 9.678447368421054e-05,
      "loss": 0.5265,
      "step": 32233
    },
    {
      "epoch": 1.80501735916676,
      "grad_norm": 1.2007571458816528,
      "learning_rate": 9.67842105263158e-05,
      "loss": 0.3738,
      "step": 32234
    },
    {
      "epoch": 1.805073356478889,
      "grad_norm": 1.392640471458435,
      "learning_rate": 9.678394736842106e-05,
      "loss": 0.6393,
      "step": 32235
    },
    {
      "epoch": 1.805129353791018,
      "grad_norm": 1.4702069759368896,
      "learning_rate": 9.678368421052631e-05,
      "loss": 0.4482,
      "step": 32236
    },
    {
      "epoch": 1.805185351103147,
      "grad_norm": 1.3520073890686035,
      "learning_rate": 9.678342105263159e-05,
      "loss": 0.6072,
      "step": 32237
    },
    {
      "epoch": 1.805241348415276,
      "grad_norm": 1.0187300443649292,
      "learning_rate": 9.678315789473685e-05,
      "loss": 0.3648,
      "step": 32238
    },
    {
      "epoch": 1.805297345727405,
      "grad_norm": 1.127990961074829,
      "learning_rate": 9.678289473684211e-05,
      "loss": 0.4802,
      "step": 32239
    },
    {
      "epoch": 1.805353343039534,
      "grad_norm": 1.6357054710388184,
      "learning_rate": 9.678263157894737e-05,
      "loss": 0.6482,
      "step": 32240
    },
    {
      "epoch": 1.8054093403516631,
      "grad_norm": 1.160599708557129,
      "learning_rate": 9.678236842105264e-05,
      "loss": 0.4286,
      "step": 32241
    },
    {
      "epoch": 1.8054653376637921,
      "grad_norm": 2.0232622623443604,
      "learning_rate": 9.67821052631579e-05,
      "loss": 0.7219,
      "step": 32242
    },
    {
      "epoch": 1.8055213349759212,
      "grad_norm": 1.1855010986328125,
      "learning_rate": 9.678184210526316e-05,
      "loss": 0.3872,
      "step": 32243
    },
    {
      "epoch": 1.8055773322880502,
      "grad_norm": 1.3260722160339355,
      "learning_rate": 9.678157894736842e-05,
      "loss": 0.4659,
      "step": 32244
    },
    {
      "epoch": 1.8056333296001792,
      "grad_norm": 1.2078649997711182,
      "learning_rate": 9.678131578947369e-05,
      "loss": 0.359,
      "step": 32245
    },
    {
      "epoch": 1.8056893269123082,
      "grad_norm": 1.1123757362365723,
      "learning_rate": 9.678105263157895e-05,
      "loss": 0.3495,
      "step": 32246
    },
    {
      "epoch": 1.8057453242244372,
      "grad_norm": 1.3541125059127808,
      "learning_rate": 9.678078947368421e-05,
      "loss": 0.4439,
      "step": 32247
    },
    {
      "epoch": 1.8058013215365663,
      "grad_norm": 1.3888378143310547,
      "learning_rate": 9.678052631578947e-05,
      "loss": 0.4234,
      "step": 32248
    },
    {
      "epoch": 1.8058573188486953,
      "grad_norm": 1.2677805423736572,
      "learning_rate": 9.678026315789475e-05,
      "loss": 0.4105,
      "step": 32249
    },
    {
      "epoch": 1.8059133161608243,
      "grad_norm": 1.654213309288025,
      "learning_rate": 9.678e-05,
      "loss": 0.5667,
      "step": 32250
    },
    {
      "epoch": 1.8059693134729533,
      "grad_norm": 1.298555850982666,
      "learning_rate": 9.677973684210528e-05,
      "loss": 0.4278,
      "step": 32251
    },
    {
      "epoch": 1.8060253107850823,
      "grad_norm": 1.2293976545333862,
      "learning_rate": 9.677947368421052e-05,
      "loss": 0.3329,
      "step": 32252
    },
    {
      "epoch": 1.8060813080972113,
      "grad_norm": 1.1662614345550537,
      "learning_rate": 9.677921052631578e-05,
      "loss": 0.4153,
      "step": 32253
    },
    {
      "epoch": 1.8061373054093404,
      "grad_norm": 1.3072854280471802,
      "learning_rate": 9.677894736842106e-05,
      "loss": 0.4323,
      "step": 32254
    },
    {
      "epoch": 1.8061933027214694,
      "grad_norm": 1.1795058250427246,
      "learning_rate": 9.677868421052632e-05,
      "loss": 0.4744,
      "step": 32255
    },
    {
      "epoch": 1.8062493000335984,
      "grad_norm": 1.3275505304336548,
      "learning_rate": 9.677842105263159e-05,
      "loss": 0.5024,
      "step": 32256
    },
    {
      "epoch": 1.8063052973457274,
      "grad_norm": 1.6467268466949463,
      "learning_rate": 9.677815789473684e-05,
      "loss": 0.5493,
      "step": 32257
    },
    {
      "epoch": 1.8063612946578564,
      "grad_norm": 1.3063864707946777,
      "learning_rate": 9.677789473684211e-05,
      "loss": 0.3845,
      "step": 32258
    },
    {
      "epoch": 1.8064172919699855,
      "grad_norm": 1.3723528385162354,
      "learning_rate": 9.677763157894737e-05,
      "loss": 0.3789,
      "step": 32259
    },
    {
      "epoch": 1.8064732892821145,
      "grad_norm": 1.4124640226364136,
      "learning_rate": 9.677736842105264e-05,
      "loss": 0.4726,
      "step": 32260
    },
    {
      "epoch": 1.8065292865942435,
      "grad_norm": 1.2545688152313232,
      "learning_rate": 9.67771052631579e-05,
      "loss": 0.3276,
      "step": 32261
    },
    {
      "epoch": 1.8065852839063725,
      "grad_norm": 1.3047953844070435,
      "learning_rate": 9.677684210526316e-05,
      "loss": 0.5004,
      "step": 32262
    },
    {
      "epoch": 1.8066412812185015,
      "grad_norm": 1.2162398099899292,
      "learning_rate": 9.677657894736842e-05,
      "loss": 0.6282,
      "step": 32263
    },
    {
      "epoch": 1.8066972785306306,
      "grad_norm": 1.263633131980896,
      "learning_rate": 9.67763157894737e-05,
      "loss": 0.4557,
      "step": 32264
    },
    {
      "epoch": 1.8067532758427596,
      "grad_norm": 1.1215318441390991,
      "learning_rate": 9.677605263157896e-05,
      "loss": 0.3735,
      "step": 32265
    },
    {
      "epoch": 1.8068092731548886,
      "grad_norm": 1.6310820579528809,
      "learning_rate": 9.677578947368421e-05,
      "loss": 0.379,
      "step": 32266
    },
    {
      "epoch": 1.8068652704670176,
      "grad_norm": 1.388453483581543,
      "learning_rate": 9.677552631578947e-05,
      "loss": 0.6514,
      "step": 32267
    },
    {
      "epoch": 1.8069212677791466,
      "grad_norm": 1.3009912967681885,
      "learning_rate": 9.677526315789475e-05,
      "loss": 0.3912,
      "step": 32268
    },
    {
      "epoch": 1.8069772650912757,
      "grad_norm": 1.580469012260437,
      "learning_rate": 9.677500000000001e-05,
      "loss": 0.3718,
      "step": 32269
    },
    {
      "epoch": 1.8070332624034047,
      "grad_norm": 1.0518136024475098,
      "learning_rate": 9.677473684210527e-05,
      "loss": 0.5209,
      "step": 32270
    },
    {
      "epoch": 1.8070892597155337,
      "grad_norm": 1.1770780086517334,
      "learning_rate": 9.677447368421053e-05,
      "loss": 0.4068,
      "step": 32271
    },
    {
      "epoch": 1.8071452570276627,
      "grad_norm": 1.210191011428833,
      "learning_rate": 9.677421052631579e-05,
      "loss": 0.417,
      "step": 32272
    },
    {
      "epoch": 1.8072012543397917,
      "grad_norm": 1.0887278318405151,
      "learning_rate": 9.677394736842106e-05,
      "loss": 0.397,
      "step": 32273
    },
    {
      "epoch": 1.8072572516519207,
      "grad_norm": 1.2882227897644043,
      "learning_rate": 9.677368421052632e-05,
      "loss": 0.4956,
      "step": 32274
    },
    {
      "epoch": 1.8073132489640498,
      "grad_norm": 1.2671078443527222,
      "learning_rate": 9.677342105263158e-05,
      "loss": 0.4725,
      "step": 32275
    },
    {
      "epoch": 1.8073692462761788,
      "grad_norm": 1.2704538106918335,
      "learning_rate": 9.677315789473684e-05,
      "loss": 0.3748,
      "step": 32276
    },
    {
      "epoch": 1.8074252435883078,
      "grad_norm": 1.7363781929016113,
      "learning_rate": 9.677289473684211e-05,
      "loss": 0.5365,
      "step": 32277
    },
    {
      "epoch": 1.8074812409004368,
      "grad_norm": 1.450387954711914,
      "learning_rate": 9.677263157894737e-05,
      "loss": 0.5902,
      "step": 32278
    },
    {
      "epoch": 1.8075372382125658,
      "grad_norm": 1.348443627357483,
      "learning_rate": 9.677236842105265e-05,
      "loss": 0.4706,
      "step": 32279
    },
    {
      "epoch": 1.8075932355246949,
      "grad_norm": 1.037840485572815,
      "learning_rate": 9.677210526315789e-05,
      "loss": 0.4654,
      "step": 32280
    },
    {
      "epoch": 1.8076492328368239,
      "grad_norm": 1.7464728355407715,
      "learning_rate": 9.677184210526317e-05,
      "loss": 0.4769,
      "step": 32281
    },
    {
      "epoch": 1.807705230148953,
      "grad_norm": 1.4880002737045288,
      "learning_rate": 9.677157894736842e-05,
      "loss": 0.5658,
      "step": 32282
    },
    {
      "epoch": 1.807761227461082,
      "grad_norm": 1.1665414571762085,
      "learning_rate": 9.67713157894737e-05,
      "loss": 0.4121,
      "step": 32283
    },
    {
      "epoch": 1.807817224773211,
      "grad_norm": 1.2785245180130005,
      "learning_rate": 9.677105263157896e-05,
      "loss": 0.7103,
      "step": 32284
    },
    {
      "epoch": 1.8078732220853397,
      "grad_norm": 1.2456238269805908,
      "learning_rate": 9.677078947368422e-05,
      "loss": 0.5982,
      "step": 32285
    },
    {
      "epoch": 1.8079292193974688,
      "grad_norm": 1.2603510618209839,
      "learning_rate": 9.677052631578948e-05,
      "loss": 0.4032,
      "step": 32286
    },
    {
      "epoch": 1.8079852167095978,
      "grad_norm": 1.1457760334014893,
      "learning_rate": 9.677026315789474e-05,
      "loss": 0.3193,
      "step": 32287
    },
    {
      "epoch": 1.8080412140217268,
      "grad_norm": 1.6529200077056885,
      "learning_rate": 9.677000000000001e-05,
      "loss": 0.3488,
      "step": 32288
    },
    {
      "epoch": 1.8080972113338558,
      "grad_norm": 1.6476494073867798,
      "learning_rate": 9.676973684210526e-05,
      "loss": 0.5689,
      "step": 32289
    },
    {
      "epoch": 1.8081532086459848,
      "grad_norm": 1.4062116146087646,
      "learning_rate": 9.676947368421053e-05,
      "loss": 0.4178,
      "step": 32290
    },
    {
      "epoch": 1.8082092059581139,
      "grad_norm": 1.2759218215942383,
      "learning_rate": 9.676921052631579e-05,
      "loss": 0.4394,
      "step": 32291
    },
    {
      "epoch": 1.8082652032702429,
      "grad_norm": 1.1573610305786133,
      "learning_rate": 9.676894736842106e-05,
      "loss": 0.4067,
      "step": 32292
    },
    {
      "epoch": 1.808321200582372,
      "grad_norm": 1.9033753871917725,
      "learning_rate": 9.676868421052632e-05,
      "loss": 0.4422,
      "step": 32293
    },
    {
      "epoch": 1.808377197894501,
      "grad_norm": 1.420331358909607,
      "learning_rate": 9.676842105263158e-05,
      "loss": 0.4602,
      "step": 32294
    },
    {
      "epoch": 1.80843319520663,
      "grad_norm": 1.4490327835083008,
      "learning_rate": 9.676815789473684e-05,
      "loss": 0.4131,
      "step": 32295
    },
    {
      "epoch": 1.808489192518759,
      "grad_norm": 1.15997314453125,
      "learning_rate": 9.676789473684212e-05,
      "loss": 0.4344,
      "step": 32296
    },
    {
      "epoch": 1.808545189830888,
      "grad_norm": 1.6639409065246582,
      "learning_rate": 9.676763157894737e-05,
      "loss": 0.3995,
      "step": 32297
    },
    {
      "epoch": 1.808601187143017,
      "grad_norm": 1.2038909196853638,
      "learning_rate": 9.676736842105263e-05,
      "loss": 0.5114,
      "step": 32298
    },
    {
      "epoch": 1.808657184455146,
      "grad_norm": 4.691858291625977,
      "learning_rate": 9.67671052631579e-05,
      "loss": 0.4514,
      "step": 32299
    },
    {
      "epoch": 1.808713181767275,
      "grad_norm": 1.3198199272155762,
      "learning_rate": 9.676684210526317e-05,
      "loss": 0.4075,
      "step": 32300
    },
    {
      "epoch": 1.808769179079404,
      "grad_norm": 1.3034695386886597,
      "learning_rate": 9.676657894736843e-05,
      "loss": 0.3631,
      "step": 32301
    },
    {
      "epoch": 1.808825176391533,
      "grad_norm": 1.3898499011993408,
      "learning_rate": 9.676631578947369e-05,
      "loss": 0.4603,
      "step": 32302
    },
    {
      "epoch": 1.808881173703662,
      "grad_norm": 1.1379055976867676,
      "learning_rate": 9.676605263157895e-05,
      "loss": 0.4313,
      "step": 32303
    },
    {
      "epoch": 1.808937171015791,
      "grad_norm": 1.540268898010254,
      "learning_rate": 9.67657894736842e-05,
      "loss": 0.3844,
      "step": 32304
    },
    {
      "epoch": 1.8089931683279201,
      "grad_norm": 1.2744961977005005,
      "learning_rate": 9.676552631578948e-05,
      "loss": 0.4449,
      "step": 32305
    },
    {
      "epoch": 1.8090491656400491,
      "grad_norm": 1.347895622253418,
      "learning_rate": 9.676526315789474e-05,
      "loss": 0.5563,
      "step": 32306
    },
    {
      "epoch": 1.8091051629521782,
      "grad_norm": 1.3699471950531006,
      "learning_rate": 9.6765e-05,
      "loss": 0.4339,
      "step": 32307
    },
    {
      "epoch": 1.8091611602643072,
      "grad_norm": 1.4137951135635376,
      "learning_rate": 9.676473684210526e-05,
      "loss": 0.4656,
      "step": 32308
    },
    {
      "epoch": 1.8092171575764362,
      "grad_norm": 1.2649034261703491,
      "learning_rate": 9.676447368421053e-05,
      "loss": 0.6339,
      "step": 32309
    },
    {
      "epoch": 1.8092731548885652,
      "grad_norm": 1.2294228076934814,
      "learning_rate": 9.676421052631579e-05,
      "loss": 0.4258,
      "step": 32310
    },
    {
      "epoch": 1.8093291522006942,
      "grad_norm": 1.2274796962738037,
      "learning_rate": 9.676394736842107e-05,
      "loss": 0.4107,
      "step": 32311
    },
    {
      "epoch": 1.8093851495128233,
      "grad_norm": 1.0296804904937744,
      "learning_rate": 9.676368421052631e-05,
      "loss": 0.34,
      "step": 32312
    },
    {
      "epoch": 1.8094411468249523,
      "grad_norm": 1.4007890224456787,
      "learning_rate": 9.676342105263158e-05,
      "loss": 0.5687,
      "step": 32313
    },
    {
      "epoch": 1.8094971441370813,
      "grad_norm": 1.0892633199691772,
      "learning_rate": 9.676315789473684e-05,
      "loss": 0.379,
      "step": 32314
    },
    {
      "epoch": 1.8095531414492103,
      "grad_norm": 1.4093111753463745,
      "learning_rate": 9.676289473684212e-05,
      "loss": 0.4331,
      "step": 32315
    },
    {
      "epoch": 1.8096091387613393,
      "grad_norm": 1.26890230178833,
      "learning_rate": 9.676263157894738e-05,
      "loss": 0.4225,
      "step": 32316
    },
    {
      "epoch": 1.8096651360734684,
      "grad_norm": 1.212093710899353,
      "learning_rate": 9.676236842105264e-05,
      "loss": 0.448,
      "step": 32317
    },
    {
      "epoch": 1.8097211333855974,
      "grad_norm": 1.0250269174575806,
      "learning_rate": 9.67621052631579e-05,
      "loss": 0.3897,
      "step": 32318
    },
    {
      "epoch": 1.8097771306977264,
      "grad_norm": 1.362004280090332,
      "learning_rate": 9.676184210526317e-05,
      "loss": 0.4284,
      "step": 32319
    },
    {
      "epoch": 1.8098331280098554,
      "grad_norm": 1.3526875972747803,
      "learning_rate": 9.676157894736843e-05,
      "loss": 0.4642,
      "step": 32320
    },
    {
      "epoch": 1.8098891253219844,
      "grad_norm": 1.1725151538848877,
      "learning_rate": 9.676131578947369e-05,
      "loss": 0.2806,
      "step": 32321
    },
    {
      "epoch": 1.8099451226341134,
      "grad_norm": 1.3579639196395874,
      "learning_rate": 9.676105263157895e-05,
      "loss": 0.3285,
      "step": 32322
    },
    {
      "epoch": 1.8100011199462425,
      "grad_norm": 1.3228201866149902,
      "learning_rate": 9.676078947368421e-05,
      "loss": 0.3841,
      "step": 32323
    },
    {
      "epoch": 1.8100571172583715,
      "grad_norm": 1.3598320484161377,
      "learning_rate": 9.676052631578948e-05,
      "loss": 0.4721,
      "step": 32324
    },
    {
      "epoch": 1.8101131145705005,
      "grad_norm": 1.3178752660751343,
      "learning_rate": 9.676026315789474e-05,
      "loss": 0.5529,
      "step": 32325
    },
    {
      "epoch": 1.8101691118826295,
      "grad_norm": 1.1229150295257568,
      "learning_rate": 9.676e-05,
      "loss": 0.3938,
      "step": 32326
    },
    {
      "epoch": 1.8102251091947585,
      "grad_norm": 1.2453727722167969,
      "learning_rate": 9.675973684210526e-05,
      "loss": 0.4427,
      "step": 32327
    },
    {
      "epoch": 1.8102811065068876,
      "grad_norm": 1.4503259658813477,
      "learning_rate": 9.675947368421053e-05,
      "loss": 0.6826,
      "step": 32328
    },
    {
      "epoch": 1.8103371038190166,
      "grad_norm": 1.2733267545700073,
      "learning_rate": 9.67592105263158e-05,
      "loss": 0.4243,
      "step": 32329
    },
    {
      "epoch": 1.8103931011311456,
      "grad_norm": 1.0814517736434937,
      "learning_rate": 9.675894736842105e-05,
      "loss": 0.3606,
      "step": 32330
    },
    {
      "epoch": 1.8104490984432746,
      "grad_norm": 1.253149151802063,
      "learning_rate": 9.675868421052631e-05,
      "loss": 0.4945,
      "step": 32331
    },
    {
      "epoch": 1.8105050957554036,
      "grad_norm": 1.5092262029647827,
      "learning_rate": 9.675842105263159e-05,
      "loss": 0.504,
      "step": 32332
    },
    {
      "epoch": 1.8105610930675327,
      "grad_norm": 1.3534983396530151,
      "learning_rate": 9.675815789473685e-05,
      "loss": 0.4628,
      "step": 32333
    },
    {
      "epoch": 1.8106170903796617,
      "grad_norm": 1.501723289489746,
      "learning_rate": 9.675789473684212e-05,
      "loss": 0.4423,
      "step": 32334
    },
    {
      "epoch": 1.8106730876917907,
      "grad_norm": 1.143212080001831,
      "learning_rate": 9.675763157894737e-05,
      "loss": 0.4591,
      "step": 32335
    },
    {
      "epoch": 1.8107290850039197,
      "grad_norm": 1.273430585861206,
      "learning_rate": 9.675736842105264e-05,
      "loss": 0.4814,
      "step": 32336
    },
    {
      "epoch": 1.8107850823160487,
      "grad_norm": 1.1064444780349731,
      "learning_rate": 9.67571052631579e-05,
      "loss": 0.3152,
      "step": 32337
    },
    {
      "epoch": 1.8108410796281778,
      "grad_norm": 1.2755224704742432,
      "learning_rate": 9.675684210526317e-05,
      "loss": 0.4356,
      "step": 32338
    },
    {
      "epoch": 1.8108970769403068,
      "grad_norm": 1.177262544631958,
      "learning_rate": 9.675657894736843e-05,
      "loss": 0.4795,
      "step": 32339
    },
    {
      "epoch": 1.8109530742524358,
      "grad_norm": 2.422985315322876,
      "learning_rate": 9.675631578947368e-05,
      "loss": 0.4553,
      "step": 32340
    },
    {
      "epoch": 1.8110090715645648,
      "grad_norm": 1.331923246383667,
      "learning_rate": 9.675605263157895e-05,
      "loss": 0.5829,
      "step": 32341
    },
    {
      "epoch": 1.8110650688766938,
      "grad_norm": 1.192797064781189,
      "learning_rate": 9.675578947368421e-05,
      "loss": 0.4697,
      "step": 32342
    },
    {
      "epoch": 1.8111210661888228,
      "grad_norm": 1.3705257177352905,
      "learning_rate": 9.675552631578949e-05,
      "loss": 0.4342,
      "step": 32343
    },
    {
      "epoch": 1.8111770635009519,
      "grad_norm": 1.352030873298645,
      "learning_rate": 9.675526315789473e-05,
      "loss": 0.4028,
      "step": 32344
    },
    {
      "epoch": 1.8112330608130809,
      "grad_norm": 1.074386477470398,
      "learning_rate": 9.6755e-05,
      "loss": 0.3142,
      "step": 32345
    },
    {
      "epoch": 1.81128905812521,
      "grad_norm": 1.3192023038864136,
      "learning_rate": 9.675473684210526e-05,
      "loss": 0.3898,
      "step": 32346
    },
    {
      "epoch": 1.811345055437339,
      "grad_norm": 1.1558268070220947,
      "learning_rate": 9.675447368421054e-05,
      "loss": 0.4478,
      "step": 32347
    },
    {
      "epoch": 1.811401052749468,
      "grad_norm": 1.1593337059020996,
      "learning_rate": 9.67542105263158e-05,
      "loss": 0.45,
      "step": 32348
    },
    {
      "epoch": 1.811457050061597,
      "grad_norm": 1.3755301237106323,
      "learning_rate": 9.675394736842106e-05,
      "loss": 0.4612,
      "step": 32349
    },
    {
      "epoch": 1.811513047373726,
      "grad_norm": 1.855124592781067,
      "learning_rate": 9.675368421052632e-05,
      "loss": 0.519,
      "step": 32350
    },
    {
      "epoch": 1.811569044685855,
      "grad_norm": 1.2857928276062012,
      "learning_rate": 9.675342105263159e-05,
      "loss": 0.3673,
      "step": 32351
    },
    {
      "epoch": 1.811625041997984,
      "grad_norm": 1.2704052925109863,
      "learning_rate": 9.675315789473685e-05,
      "loss": 0.4789,
      "step": 32352
    },
    {
      "epoch": 1.811681039310113,
      "grad_norm": 1.3460575342178345,
      "learning_rate": 9.675289473684211e-05,
      "loss": 0.5163,
      "step": 32353
    },
    {
      "epoch": 1.811737036622242,
      "grad_norm": 1.560444712638855,
      "learning_rate": 9.675263157894737e-05,
      "loss": 0.7829,
      "step": 32354
    },
    {
      "epoch": 1.811793033934371,
      "grad_norm": 2.1030690670013428,
      "learning_rate": 9.675236842105264e-05,
      "loss": 0.4952,
      "step": 32355
    },
    {
      "epoch": 1.8118490312465,
      "grad_norm": 1.3004571199417114,
      "learning_rate": 9.67521052631579e-05,
      "loss": 0.4162,
      "step": 32356
    },
    {
      "epoch": 1.8119050285586291,
      "grad_norm": 1.7338188886642456,
      "learning_rate": 9.675184210526316e-05,
      "loss": 0.6515,
      "step": 32357
    },
    {
      "epoch": 1.8119610258707581,
      "grad_norm": 1.035826563835144,
      "learning_rate": 9.675157894736842e-05,
      "loss": 0.4084,
      "step": 32358
    },
    {
      "epoch": 1.8120170231828872,
      "grad_norm": 1.3207197189331055,
      "learning_rate": 9.675131578947368e-05,
      "loss": 0.5052,
      "step": 32359
    },
    {
      "epoch": 1.8120730204950162,
      "grad_norm": 0.9274834990501404,
      "learning_rate": 9.675105263157895e-05,
      "loss": 0.3125,
      "step": 32360
    },
    {
      "epoch": 1.8121290178071452,
      "grad_norm": 1.6074193716049194,
      "learning_rate": 9.675078947368421e-05,
      "loss": 0.4257,
      "step": 32361
    },
    {
      "epoch": 1.8121850151192742,
      "grad_norm": 1.6241343021392822,
      "learning_rate": 9.675052631578947e-05,
      "loss": 0.5435,
      "step": 32362
    },
    {
      "epoch": 1.8122410124314032,
      "grad_norm": 1.4978107213974,
      "learning_rate": 9.675026315789473e-05,
      "loss": 0.374,
      "step": 32363
    },
    {
      "epoch": 1.8122970097435323,
      "grad_norm": 1.2861928939819336,
      "learning_rate": 9.675000000000001e-05,
      "loss": 0.4076,
      "step": 32364
    },
    {
      "epoch": 1.8123530070556613,
      "grad_norm": 1.4866325855255127,
      "learning_rate": 9.674973684210527e-05,
      "loss": 0.5525,
      "step": 32365
    },
    {
      "epoch": 1.8124090043677903,
      "grad_norm": 1.2640074491500854,
      "learning_rate": 9.674947368421054e-05,
      "loss": 0.4721,
      "step": 32366
    },
    {
      "epoch": 1.8124650016799193,
      "grad_norm": 1.274003505706787,
      "learning_rate": 9.674921052631579e-05,
      "loss": 0.5322,
      "step": 32367
    },
    {
      "epoch": 1.8125209989920483,
      "grad_norm": 1.17131507396698,
      "learning_rate": 9.674894736842106e-05,
      "loss": 0.4904,
      "step": 32368
    },
    {
      "epoch": 1.8125769963041773,
      "grad_norm": 1.154022455215454,
      "learning_rate": 9.674868421052632e-05,
      "loss": 0.3773,
      "step": 32369
    },
    {
      "epoch": 1.8126329936163064,
      "grad_norm": 1.114092469215393,
      "learning_rate": 9.674842105263159e-05,
      "loss": 0.608,
      "step": 32370
    },
    {
      "epoch": 1.8126889909284354,
      "grad_norm": 1.2449208498001099,
      "learning_rate": 9.674815789473685e-05,
      "loss": 0.4599,
      "step": 32371
    },
    {
      "epoch": 1.8127449882405644,
      "grad_norm": 1.368186354637146,
      "learning_rate": 9.674789473684211e-05,
      "loss": 0.4416,
      "step": 32372
    },
    {
      "epoch": 1.8128009855526934,
      "grad_norm": 1.2514532804489136,
      "learning_rate": 9.674763157894737e-05,
      "loss": 0.4161,
      "step": 32373
    },
    {
      "epoch": 1.8128569828648224,
      "grad_norm": 1.328495979309082,
      "learning_rate": 9.674736842105263e-05,
      "loss": 0.4454,
      "step": 32374
    },
    {
      "epoch": 1.8129129801769515,
      "grad_norm": 1.437473177909851,
      "learning_rate": 9.67471052631579e-05,
      "loss": 0.499,
      "step": 32375
    },
    {
      "epoch": 1.8129689774890805,
      "grad_norm": 1.3084028959274292,
      "learning_rate": 9.674684210526316e-05,
      "loss": 0.4292,
      "step": 32376
    },
    {
      "epoch": 1.8130249748012095,
      "grad_norm": 1.2749313116073608,
      "learning_rate": 9.674657894736842e-05,
      "loss": 0.4161,
      "step": 32377
    },
    {
      "epoch": 1.8130809721133385,
      "grad_norm": 1.0798845291137695,
      "learning_rate": 9.674631578947368e-05,
      "loss": 0.3706,
      "step": 32378
    },
    {
      "epoch": 1.8131369694254675,
      "grad_norm": 1.4673467874526978,
      "learning_rate": 9.674605263157896e-05,
      "loss": 0.4363,
      "step": 32379
    },
    {
      "epoch": 1.8131929667375966,
      "grad_norm": 1.444225549697876,
      "learning_rate": 9.674578947368422e-05,
      "loss": 0.488,
      "step": 32380
    },
    {
      "epoch": 1.8132489640497256,
      "grad_norm": 1.188883900642395,
      "learning_rate": 9.674552631578948e-05,
      "loss": 0.3659,
      "step": 32381
    },
    {
      "epoch": 1.8133049613618546,
      "grad_norm": 1.378848671913147,
      "learning_rate": 9.674526315789474e-05,
      "loss": 0.4077,
      "step": 32382
    },
    {
      "epoch": 1.8133609586739836,
      "grad_norm": NaN,
      "learning_rate": 9.674526315789474e-05,
      "loss": 0.6469,
      "step": 32383
    },
    {
      "epoch": 1.8134169559861126,
      "grad_norm": 1.1309772729873657,
      "learning_rate": 9.674500000000001e-05,
      "loss": 0.3757,
      "step": 32384
    },
    {
      "epoch": 1.8134729532982417,
      "grad_norm": 1.1356582641601562,
      "learning_rate": 9.674473684210527e-05,
      "loss": 0.5102,
      "step": 32385
    },
    {
      "epoch": 1.8135289506103707,
      "grad_norm": 1.3671808242797852,
      "learning_rate": 9.674447368421053e-05,
      "loss": 0.3884,
      "step": 32386
    },
    {
      "epoch": 1.8135849479224997,
      "grad_norm": 1.0610895156860352,
      "learning_rate": 9.674421052631579e-05,
      "loss": 0.4281,
      "step": 32387
    },
    {
      "epoch": 1.8136409452346287,
      "grad_norm": 1.7117756605148315,
      "learning_rate": 9.674394736842106e-05,
      "loss": 0.4621,
      "step": 32388
    },
    {
      "epoch": 1.8136969425467577,
      "grad_norm": 1.286011815071106,
      "learning_rate": 9.674368421052632e-05,
      "loss": 0.3887,
      "step": 32389
    },
    {
      "epoch": 1.8137529398588867,
      "grad_norm": 1.394737720489502,
      "learning_rate": 9.67434210526316e-05,
      "loss": 0.4349,
      "step": 32390
    },
    {
      "epoch": 1.8138089371710158,
      "grad_norm": 2.456839084625244,
      "learning_rate": 9.674315789473684e-05,
      "loss": 0.4555,
      "step": 32391
    },
    {
      "epoch": 1.8138649344831448,
      "grad_norm": 1.1063789129257202,
      "learning_rate": 9.67428947368421e-05,
      "loss": 0.3573,
      "step": 32392
    },
    {
      "epoch": 1.8139209317952738,
      "grad_norm": 1.1253880262374878,
      "learning_rate": 9.674263157894737e-05,
      "loss": 0.3738,
      "step": 32393
    },
    {
      "epoch": 1.8139769291074028,
      "grad_norm": 1.2422535419464111,
      "learning_rate": 9.674236842105263e-05,
      "loss": 0.5004,
      "step": 32394
    },
    {
      "epoch": 1.8140329264195318,
      "grad_norm": 2.7551474571228027,
      "learning_rate": 9.67421052631579e-05,
      "loss": 0.6218,
      "step": 32395
    },
    {
      "epoch": 1.8140889237316609,
      "grad_norm": 1.719006896018982,
      "learning_rate": 9.674184210526315e-05,
      "loss": 0.6325,
      "step": 32396
    },
    {
      "epoch": 1.8141449210437899,
      "grad_norm": 1.2585749626159668,
      "learning_rate": 9.674157894736843e-05,
      "loss": 0.4662,
      "step": 32397
    },
    {
      "epoch": 1.814200918355919,
      "grad_norm": 1.2235743999481201,
      "learning_rate": 9.674131578947369e-05,
      "loss": 0.5555,
      "step": 32398
    },
    {
      "epoch": 1.814256915668048,
      "grad_norm": 1.2404729127883911,
      "learning_rate": 9.674105263157896e-05,
      "loss": 0.4315,
      "step": 32399
    },
    {
      "epoch": 1.814312912980177,
      "grad_norm": 1.3855149745941162,
      "learning_rate": 9.67407894736842e-05,
      "loss": 0.6888,
      "step": 32400
    },
    {
      "epoch": 1.814368910292306,
      "grad_norm": 1.2371548414230347,
      "learning_rate": 9.674052631578948e-05,
      "loss": 0.4986,
      "step": 32401
    },
    {
      "epoch": 1.814424907604435,
      "grad_norm": 1.169758677482605,
      "learning_rate": 9.674026315789474e-05,
      "loss": 0.3782,
      "step": 32402
    },
    {
      "epoch": 1.814480904916564,
      "grad_norm": 1.3540791273117065,
      "learning_rate": 9.674000000000001e-05,
      "loss": 0.5137,
      "step": 32403
    },
    {
      "epoch": 1.814536902228693,
      "grad_norm": 1.1753042936325073,
      "learning_rate": 9.673973684210527e-05,
      "loss": 0.5769,
      "step": 32404
    },
    {
      "epoch": 1.814592899540822,
      "grad_norm": 1.234282374382019,
      "learning_rate": 9.673947368421053e-05,
      "loss": 0.4231,
      "step": 32405
    },
    {
      "epoch": 1.814648896852951,
      "grad_norm": 1.2665352821350098,
      "learning_rate": 9.673921052631579e-05,
      "loss": 0.4462,
      "step": 32406
    },
    {
      "epoch": 1.81470489416508,
      "grad_norm": 1.6300742626190186,
      "learning_rate": 9.673894736842106e-05,
      "loss": 0.3469,
      "step": 32407
    },
    {
      "epoch": 1.814760891477209,
      "grad_norm": 1.1760287284851074,
      "learning_rate": 9.673868421052632e-05,
      "loss": 0.3751,
      "step": 32408
    },
    {
      "epoch": 1.8148168887893381,
      "grad_norm": 1.4402811527252197,
      "learning_rate": 9.673842105263158e-05,
      "loss": 0.4324,
      "step": 32409
    },
    {
      "epoch": 1.8148728861014671,
      "grad_norm": 1.525953769683838,
      "learning_rate": 9.673815789473684e-05,
      "loss": 0.4798,
      "step": 32410
    },
    {
      "epoch": 1.8149288834135962,
      "grad_norm": 1.136301040649414,
      "learning_rate": 9.67378947368421e-05,
      "loss": 0.351,
      "step": 32411
    },
    {
      "epoch": 1.8149848807257252,
      "grad_norm": 1.1505835056304932,
      "learning_rate": 9.673763157894738e-05,
      "loss": 0.4527,
      "step": 32412
    },
    {
      "epoch": 1.8150408780378542,
      "grad_norm": 1.2191072702407837,
      "learning_rate": 9.673736842105264e-05,
      "loss": 0.4146,
      "step": 32413
    },
    {
      "epoch": 1.8150968753499832,
      "grad_norm": 1.3683855533599854,
      "learning_rate": 9.67371052631579e-05,
      "loss": 0.5296,
      "step": 32414
    },
    {
      "epoch": 1.8151528726621122,
      "grad_norm": 1.335317611694336,
      "learning_rate": 9.673684210526316e-05,
      "loss": 0.3824,
      "step": 32415
    },
    {
      "epoch": 1.8152088699742412,
      "grad_norm": 1.258230209350586,
      "learning_rate": 9.673657894736843e-05,
      "loss": 0.408,
      "step": 32416
    },
    {
      "epoch": 1.8152648672863703,
      "grad_norm": 1.4180176258087158,
      "learning_rate": 9.673631578947369e-05,
      "loss": 0.3647,
      "step": 32417
    },
    {
      "epoch": 1.8153208645984993,
      "grad_norm": 1.151563286781311,
      "learning_rate": 9.673605263157895e-05,
      "loss": 0.3705,
      "step": 32418
    },
    {
      "epoch": 1.8153768619106283,
      "grad_norm": 1.4188717603683472,
      "learning_rate": 9.673578947368421e-05,
      "loss": 0.5693,
      "step": 32419
    },
    {
      "epoch": 1.8154328592227573,
      "grad_norm": 1.3015892505645752,
      "learning_rate": 9.673552631578948e-05,
      "loss": 0.396,
      "step": 32420
    },
    {
      "epoch": 1.8154888565348863,
      "grad_norm": 1.1772204637527466,
      "learning_rate": 9.673526315789474e-05,
      "loss": 0.4458,
      "step": 32421
    },
    {
      "epoch": 1.8155448538470154,
      "grad_norm": 1.3385714292526245,
      "learning_rate": 9.673500000000001e-05,
      "loss": 0.4674,
      "step": 32422
    },
    {
      "epoch": 1.8156008511591444,
      "grad_norm": 1.2404509782791138,
      "learning_rate": 9.673473684210526e-05,
      "loss": 0.4117,
      "step": 32423
    },
    {
      "epoch": 1.8156568484712734,
      "grad_norm": 1.6911565065383911,
      "learning_rate": 9.673447368421053e-05,
      "loss": 0.4973,
      "step": 32424
    },
    {
      "epoch": 1.8157128457834024,
      "grad_norm": 1.200289011001587,
      "learning_rate": 9.67342105263158e-05,
      "loss": 0.3987,
      "step": 32425
    },
    {
      "epoch": 1.8157688430955314,
      "grad_norm": 1.3031913042068481,
      "learning_rate": 9.673394736842105e-05,
      "loss": 0.4437,
      "step": 32426
    },
    {
      "epoch": 1.8158248404076605,
      "grad_norm": 1.4410406351089478,
      "learning_rate": 9.673368421052633e-05,
      "loss": 0.4704,
      "step": 32427
    },
    {
      "epoch": 1.8158808377197895,
      "grad_norm": 3.1850626468658447,
      "learning_rate": 9.673342105263157e-05,
      "loss": 0.4591,
      "step": 32428
    },
    {
      "epoch": 1.8159368350319185,
      "grad_norm": 1.22998046875,
      "learning_rate": 9.673315789473685e-05,
      "loss": 0.5411,
      "step": 32429
    },
    {
      "epoch": 1.8159928323440475,
      "grad_norm": 1.366072416305542,
      "learning_rate": 9.67328947368421e-05,
      "loss": 0.5403,
      "step": 32430
    },
    {
      "epoch": 1.8160488296561765,
      "grad_norm": 1.1950949430465698,
      "learning_rate": 9.673263157894738e-05,
      "loss": 0.5019,
      "step": 32431
    },
    {
      "epoch": 1.8161048269683056,
      "grad_norm": 1.0988450050354004,
      "learning_rate": 9.673236842105264e-05,
      "loss": 0.377,
      "step": 32432
    },
    {
      "epoch": 1.8161608242804346,
      "grad_norm": 1.191611886024475,
      "learning_rate": 9.67321052631579e-05,
      "loss": 0.3688,
      "step": 32433
    },
    {
      "epoch": 1.8162168215925636,
      "grad_norm": 1.55448579788208,
      "learning_rate": 9.673184210526316e-05,
      "loss": 0.4016,
      "step": 32434
    },
    {
      "epoch": 1.8162728189046926,
      "grad_norm": 1.2630044221878052,
      "learning_rate": 9.673157894736843e-05,
      "loss": 0.4086,
      "step": 32435
    },
    {
      "epoch": 1.8163288162168216,
      "grad_norm": 1.627791166305542,
      "learning_rate": 9.673131578947369e-05,
      "loss": 0.4262,
      "step": 32436
    },
    {
      "epoch": 1.8163848135289506,
      "grad_norm": 1.4962151050567627,
      "learning_rate": 9.673105263157895e-05,
      "loss": 0.4298,
      "step": 32437
    },
    {
      "epoch": 1.8164408108410797,
      "grad_norm": 1.174643635749817,
      "learning_rate": 9.673078947368421e-05,
      "loss": 0.4499,
      "step": 32438
    },
    {
      "epoch": 1.8164968081532087,
      "grad_norm": 1.398247480392456,
      "learning_rate": 9.673052631578948e-05,
      "loss": 0.4514,
      "step": 32439
    },
    {
      "epoch": 1.8165528054653377,
      "grad_norm": 1.5056617259979248,
      "learning_rate": 9.673026315789474e-05,
      "loss": 0.4985,
      "step": 32440
    },
    {
      "epoch": 1.8166088027774667,
      "grad_norm": 1.16782808303833,
      "learning_rate": 9.673e-05,
      "loss": 0.4234,
      "step": 32441
    },
    {
      "epoch": 1.8166648000895957,
      "grad_norm": 1.407913327217102,
      "learning_rate": 9.672973684210526e-05,
      "loss": 0.455,
      "step": 32442
    },
    {
      "epoch": 1.8167207974017248,
      "grad_norm": 1.251463532447815,
      "learning_rate": 9.672947368421054e-05,
      "loss": 0.4097,
      "step": 32443
    },
    {
      "epoch": 1.8167767947138538,
      "grad_norm": 1.2283661365509033,
      "learning_rate": 9.67292105263158e-05,
      "loss": 0.3914,
      "step": 32444
    },
    {
      "epoch": 1.8168327920259828,
      "grad_norm": 1.2987653017044067,
      "learning_rate": 9.672894736842106e-05,
      "loss": 0.447,
      "step": 32445
    },
    {
      "epoch": 1.8168887893381118,
      "grad_norm": 1.072947382926941,
      "learning_rate": 9.672868421052632e-05,
      "loss": 0.395,
      "step": 32446
    },
    {
      "epoch": 1.8169447866502408,
      "grad_norm": 1.221902847290039,
      "learning_rate": 9.672842105263158e-05,
      "loss": 0.4584,
      "step": 32447
    },
    {
      "epoch": 1.8170007839623699,
      "grad_norm": 1.433893084526062,
      "learning_rate": 9.672815789473685e-05,
      "loss": 0.4965,
      "step": 32448
    },
    {
      "epoch": 1.8170567812744989,
      "grad_norm": 1.2009406089782715,
      "learning_rate": 9.672789473684211e-05,
      "loss": 0.3997,
      "step": 32449
    },
    {
      "epoch": 1.817112778586628,
      "grad_norm": 1.0657004117965698,
      "learning_rate": 9.672763157894737e-05,
      "loss": 0.3921,
      "step": 32450
    },
    {
      "epoch": 1.817168775898757,
      "grad_norm": 1.0397655963897705,
      "learning_rate": 9.672736842105263e-05,
      "loss": 0.319,
      "step": 32451
    },
    {
      "epoch": 1.817224773210886,
      "grad_norm": 1.0295485258102417,
      "learning_rate": 9.67271052631579e-05,
      "loss": 0.3316,
      "step": 32452
    },
    {
      "epoch": 1.817280770523015,
      "grad_norm": 1.3678115606307983,
      "learning_rate": 9.672684210526316e-05,
      "loss": 0.4833,
      "step": 32453
    },
    {
      "epoch": 1.817336767835144,
      "grad_norm": 1.404530644416809,
      "learning_rate": 9.672657894736843e-05,
      "loss": 0.4843,
      "step": 32454
    },
    {
      "epoch": 1.817392765147273,
      "grad_norm": 1.1422580480575562,
      "learning_rate": 9.672631578947368e-05,
      "loss": 0.4187,
      "step": 32455
    },
    {
      "epoch": 1.817448762459402,
      "grad_norm": 1.2389475107192993,
      "learning_rate": 9.672605263157895e-05,
      "loss": 0.4367,
      "step": 32456
    },
    {
      "epoch": 1.817504759771531,
      "grad_norm": 1.4369456768035889,
      "learning_rate": 9.672578947368421e-05,
      "loss": 0.5171,
      "step": 32457
    },
    {
      "epoch": 1.81756075708366,
      "grad_norm": 1.3177380561828613,
      "learning_rate": 9.672552631578949e-05,
      "loss": 0.3973,
      "step": 32458
    },
    {
      "epoch": 1.817616754395789,
      "grad_norm": 1.4693036079406738,
      "learning_rate": 9.672526315789475e-05,
      "loss": 0.484,
      "step": 32459
    },
    {
      "epoch": 1.817672751707918,
      "grad_norm": 1.2984145879745483,
      "learning_rate": 9.6725e-05,
      "loss": 0.4323,
      "step": 32460
    },
    {
      "epoch": 1.817728749020047,
      "grad_norm": 1.3219741582870483,
      "learning_rate": 9.672473684210527e-05,
      "loss": 0.462,
      "step": 32461
    },
    {
      "epoch": 1.8177847463321761,
      "grad_norm": 1.6226540803909302,
      "learning_rate": 9.672447368421053e-05,
      "loss": 0.5406,
      "step": 32462
    },
    {
      "epoch": 1.8178407436443051,
      "grad_norm": 1.251319408416748,
      "learning_rate": 9.67242105263158e-05,
      "loss": 0.436,
      "step": 32463
    },
    {
      "epoch": 1.8178967409564342,
      "grad_norm": 1.5320277214050293,
      "learning_rate": 9.672394736842106e-05,
      "loss": 0.4934,
      "step": 32464
    },
    {
      "epoch": 1.8179527382685632,
      "grad_norm": 1.1777857542037964,
      "learning_rate": 9.672368421052632e-05,
      "loss": 0.3865,
      "step": 32465
    },
    {
      "epoch": 1.8180087355806922,
      "grad_norm": 1.2484936714172363,
      "learning_rate": 9.672342105263158e-05,
      "loss": 0.3981,
      "step": 32466
    },
    {
      "epoch": 1.8180647328928212,
      "grad_norm": 1.3128679990768433,
      "learning_rate": 9.672315789473685e-05,
      "loss": 0.4972,
      "step": 32467
    },
    {
      "epoch": 1.8181207302049502,
      "grad_norm": 1.1874022483825684,
      "learning_rate": 9.672289473684211e-05,
      "loss": 0.4869,
      "step": 32468
    },
    {
      "epoch": 1.8181767275170793,
      "grad_norm": 1.3044192790985107,
      "learning_rate": 9.672263157894737e-05,
      "loss": 0.5496,
      "step": 32469
    },
    {
      "epoch": 1.8182327248292083,
      "grad_norm": 25.047182083129883,
      "learning_rate": 9.672236842105263e-05,
      "loss": 0.5854,
      "step": 32470
    },
    {
      "epoch": 1.8182887221413373,
      "grad_norm": 1.2622507810592651,
      "learning_rate": 9.67221052631579e-05,
      "loss": 0.3639,
      "step": 32471
    },
    {
      "epoch": 1.8183447194534663,
      "grad_norm": 1.3041987419128418,
      "learning_rate": 9.672184210526316e-05,
      "loss": 0.4311,
      "step": 32472
    },
    {
      "epoch": 1.8184007167655953,
      "grad_norm": 1.299324631690979,
      "learning_rate": 9.672157894736842e-05,
      "loss": 0.4381,
      "step": 32473
    },
    {
      "epoch": 1.8184567140777244,
      "grad_norm": 1.4443858861923218,
      "learning_rate": 9.672131578947368e-05,
      "loss": 0.4631,
      "step": 32474
    },
    {
      "epoch": 1.8185127113898534,
      "grad_norm": 1.4813605546951294,
      "learning_rate": 9.672105263157896e-05,
      "loss": 0.4762,
      "step": 32475
    },
    {
      "epoch": 1.8185687087019824,
      "grad_norm": 1.339863657951355,
      "learning_rate": 9.672078947368422e-05,
      "loss": 0.4089,
      "step": 32476
    },
    {
      "epoch": 1.8186247060141114,
      "grad_norm": 1.1854147911071777,
      "learning_rate": 9.672052631578949e-05,
      "loss": 0.5234,
      "step": 32477
    },
    {
      "epoch": 1.8186807033262404,
      "grad_norm": 1.4232052564620972,
      "learning_rate": 9.672026315789474e-05,
      "loss": 0.5493,
      "step": 32478
    },
    {
      "epoch": 1.8187367006383695,
      "grad_norm": 1.1105598211288452,
      "learning_rate": 9.672e-05,
      "loss": 0.3632,
      "step": 32479
    },
    {
      "epoch": 1.8187926979504985,
      "grad_norm": 1.3750697374343872,
      "learning_rate": 9.671973684210527e-05,
      "loss": 0.5164,
      "step": 32480
    },
    {
      "epoch": 1.8188486952626275,
      "grad_norm": 1.3002640008926392,
      "learning_rate": 9.671947368421053e-05,
      "loss": 0.4005,
      "step": 32481
    },
    {
      "epoch": 1.8189046925747565,
      "grad_norm": 1.077797532081604,
      "learning_rate": 9.67192105263158e-05,
      "loss": 0.3966,
      "step": 32482
    },
    {
      "epoch": 1.8189606898868855,
      "grad_norm": 1.2418121099472046,
      "learning_rate": 9.671894736842105e-05,
      "loss": 0.4458,
      "step": 32483
    },
    {
      "epoch": 1.8190166871990145,
      "grad_norm": 1.2358874082565308,
      "learning_rate": 9.671868421052632e-05,
      "loss": 0.4766,
      "step": 32484
    },
    {
      "epoch": 1.8190726845111436,
      "grad_norm": 1.1712127923965454,
      "learning_rate": 9.671842105263158e-05,
      "loss": 0.4278,
      "step": 32485
    },
    {
      "epoch": 1.8191286818232726,
      "grad_norm": 1.1019481420516968,
      "learning_rate": 9.671815789473685e-05,
      "loss": 0.4316,
      "step": 32486
    },
    {
      "epoch": 1.8191846791354016,
      "grad_norm": 1.5419946908950806,
      "learning_rate": 9.671789473684211e-05,
      "loss": 0.5452,
      "step": 32487
    },
    {
      "epoch": 1.8192406764475306,
      "grad_norm": 0.9683425426483154,
      "learning_rate": 9.671763157894737e-05,
      "loss": 0.3032,
      "step": 32488
    },
    {
      "epoch": 1.8192966737596596,
      "grad_norm": 1.3100649118423462,
      "learning_rate": 9.671736842105263e-05,
      "loss": 0.4757,
      "step": 32489
    },
    {
      "epoch": 1.8193526710717887,
      "grad_norm": 1.3169389963150024,
      "learning_rate": 9.67171052631579e-05,
      "loss": 0.4721,
      "step": 32490
    },
    {
      "epoch": 1.8194086683839177,
      "grad_norm": 1.4245240688323975,
      "learning_rate": 9.671684210526317e-05,
      "loss": 0.4404,
      "step": 32491
    },
    {
      "epoch": 1.8194646656960467,
      "grad_norm": 1.2592575550079346,
      "learning_rate": 9.671657894736843e-05,
      "loss": 0.5125,
      "step": 32492
    },
    {
      "epoch": 1.8195206630081757,
      "grad_norm": 1.2397279739379883,
      "learning_rate": 9.671631578947369e-05,
      "loss": 0.5608,
      "step": 32493
    },
    {
      "epoch": 1.8195766603203047,
      "grad_norm": 1.2415943145751953,
      "learning_rate": 9.671605263157896e-05,
      "loss": 0.4949,
      "step": 32494
    },
    {
      "epoch": 1.8196326576324338,
      "grad_norm": 1.1079696416854858,
      "learning_rate": 9.671578947368422e-05,
      "loss": 0.3893,
      "step": 32495
    },
    {
      "epoch": 1.8196886549445628,
      "grad_norm": 1.2162058353424072,
      "learning_rate": 9.671552631578948e-05,
      "loss": 0.3831,
      "step": 32496
    },
    {
      "epoch": 1.8197446522566918,
      "grad_norm": 1.2957499027252197,
      "learning_rate": 9.671526315789474e-05,
      "loss": 0.474,
      "step": 32497
    },
    {
      "epoch": 1.8198006495688208,
      "grad_norm": 1.0710455179214478,
      "learning_rate": 9.6715e-05,
      "loss": 0.3684,
      "step": 32498
    },
    {
      "epoch": 1.8198566468809498,
      "grad_norm": 1.2035348415374756,
      "learning_rate": 9.671473684210527e-05,
      "loss": 0.3746,
      "step": 32499
    },
    {
      "epoch": 1.8199126441930789,
      "grad_norm": 1.1825233697891235,
      "learning_rate": 9.671447368421053e-05,
      "loss": 0.4409,
      "step": 32500
    },
    {
      "epoch": 1.8199686415052079,
      "grad_norm": 1.3556849956512451,
      "learning_rate": 9.671421052631579e-05,
      "loss": 0.5354,
      "step": 32501
    },
    {
      "epoch": 1.820024638817337,
      "grad_norm": 1.3185393810272217,
      "learning_rate": 9.671394736842105e-05,
      "loss": 0.4968,
      "step": 32502
    },
    {
      "epoch": 1.820080636129466,
      "grad_norm": 1.3182179927825928,
      "learning_rate": 9.671368421052632e-05,
      "loss": 0.5185,
      "step": 32503
    },
    {
      "epoch": 1.820136633441595,
      "grad_norm": 2.057490587234497,
      "learning_rate": 9.671342105263158e-05,
      "loss": 0.4132,
      "step": 32504
    },
    {
      "epoch": 1.820192630753724,
      "grad_norm": 1.2419458627700806,
      "learning_rate": 9.671315789473684e-05,
      "loss": 0.6465,
      "step": 32505
    },
    {
      "epoch": 1.820248628065853,
      "grad_norm": 1.45686936378479,
      "learning_rate": 9.67128947368421e-05,
      "loss": 0.4034,
      "step": 32506
    },
    {
      "epoch": 1.820304625377982,
      "grad_norm": 1.3267873525619507,
      "learning_rate": 9.671263157894738e-05,
      "loss": 0.5541,
      "step": 32507
    },
    {
      "epoch": 1.820360622690111,
      "grad_norm": 1.2031878232955933,
      "learning_rate": 9.671236842105264e-05,
      "loss": 0.5194,
      "step": 32508
    },
    {
      "epoch": 1.82041662000224,
      "grad_norm": 1.5494134426116943,
      "learning_rate": 9.671210526315791e-05,
      "loss": 0.5716,
      "step": 32509
    },
    {
      "epoch": 1.820472617314369,
      "grad_norm": 1.2340202331542969,
      "learning_rate": 9.671184210526316e-05,
      "loss": 0.437,
      "step": 32510
    },
    {
      "epoch": 1.820528614626498,
      "grad_norm": 1.1425443887710571,
      "learning_rate": 9.671157894736843e-05,
      "loss": 0.3416,
      "step": 32511
    },
    {
      "epoch": 1.820584611938627,
      "grad_norm": 1.1883059740066528,
      "learning_rate": 9.671131578947369e-05,
      "loss": 0.528,
      "step": 32512
    },
    {
      "epoch": 1.820640609250756,
      "grad_norm": 1.4798719882965088,
      "learning_rate": 9.671105263157895e-05,
      "loss": 0.5063,
      "step": 32513
    },
    {
      "epoch": 1.8206966065628851,
      "grad_norm": 1.866953730583191,
      "learning_rate": 9.671078947368422e-05,
      "loss": 0.4992,
      "step": 32514
    },
    {
      "epoch": 1.8207526038750141,
      "grad_norm": 1.6121031045913696,
      "learning_rate": 9.671052631578947e-05,
      "loss": 0.4008,
      "step": 32515
    },
    {
      "epoch": 1.8208086011871432,
      "grad_norm": 1.46975576877594,
      "learning_rate": 9.671026315789474e-05,
      "loss": 0.6517,
      "step": 32516
    },
    {
      "epoch": 1.8208645984992722,
      "grad_norm": 1.711916208267212,
      "learning_rate": 9.671e-05,
      "loss": 0.6506,
      "step": 32517
    },
    {
      "epoch": 1.8209205958114012,
      "grad_norm": 1.4283182621002197,
      "learning_rate": 9.670973684210527e-05,
      "loss": 0.4752,
      "step": 32518
    },
    {
      "epoch": 1.8209765931235302,
      "grad_norm": 1.1654165983200073,
      "learning_rate": 9.670947368421053e-05,
      "loss": 0.4088,
      "step": 32519
    },
    {
      "epoch": 1.8210325904356592,
      "grad_norm": 1.0469211339950562,
      "learning_rate": 9.67092105263158e-05,
      "loss": 0.4589,
      "step": 32520
    },
    {
      "epoch": 1.8210885877477883,
      "grad_norm": 1.2741965055465698,
      "learning_rate": 9.670894736842105e-05,
      "loss": 0.46,
      "step": 32521
    },
    {
      "epoch": 1.8211445850599173,
      "grad_norm": 1.0342941284179688,
      "learning_rate": 9.670868421052633e-05,
      "loss": 0.4866,
      "step": 32522
    },
    {
      "epoch": 1.8212005823720463,
      "grad_norm": 1.4391822814941406,
      "learning_rate": 9.670842105263159e-05,
      "loss": 0.4368,
      "step": 32523
    },
    {
      "epoch": 1.8212565796841753,
      "grad_norm": 1.0916335582733154,
      "learning_rate": 9.670815789473685e-05,
      "loss": 0.4136,
      "step": 32524
    },
    {
      "epoch": 1.8213125769963043,
      "grad_norm": 1.1497094631195068,
      "learning_rate": 9.67078947368421e-05,
      "loss": 0.468,
      "step": 32525
    },
    {
      "epoch": 1.8213685743084334,
      "grad_norm": 1.96045982837677,
      "learning_rate": 9.670763157894738e-05,
      "loss": 0.3914,
      "step": 32526
    },
    {
      "epoch": 1.8214245716205624,
      "grad_norm": 1.3185077905654907,
      "learning_rate": 9.670736842105264e-05,
      "loss": 0.5111,
      "step": 32527
    },
    {
      "epoch": 1.8214805689326914,
      "grad_norm": 1.5654627084732056,
      "learning_rate": 9.67071052631579e-05,
      "loss": 0.5916,
      "step": 32528
    },
    {
      "epoch": 1.8215365662448204,
      "grad_norm": 1.1321732997894287,
      "learning_rate": 9.670684210526316e-05,
      "loss": 0.3828,
      "step": 32529
    },
    {
      "epoch": 1.8215925635569494,
      "grad_norm": 1.2071489095687866,
      "learning_rate": 9.670657894736842e-05,
      "loss": 0.3852,
      "step": 32530
    },
    {
      "epoch": 1.8216485608690784,
      "grad_norm": 1.2667101621627808,
      "learning_rate": 9.670631578947369e-05,
      "loss": 0.5004,
      "step": 32531
    },
    {
      "epoch": 1.8217045581812075,
      "grad_norm": 1.2658908367156982,
      "learning_rate": 9.670605263157895e-05,
      "loss": 0.424,
      "step": 32532
    },
    {
      "epoch": 1.8217605554933363,
      "grad_norm": 1.3126044273376465,
      "learning_rate": 9.670578947368421e-05,
      "loss": 0.3907,
      "step": 32533
    },
    {
      "epoch": 1.8218165528054653,
      "grad_norm": 1.6182126998901367,
      "learning_rate": 9.670552631578947e-05,
      "loss": 0.3787,
      "step": 32534
    },
    {
      "epoch": 1.8218725501175943,
      "grad_norm": 1.3860315084457397,
      "learning_rate": 9.670526315789474e-05,
      "loss": 0.4279,
      "step": 32535
    },
    {
      "epoch": 1.8219285474297233,
      "grad_norm": 1.235561728477478,
      "learning_rate": 9.6705e-05,
      "loss": 0.5008,
      "step": 32536
    },
    {
      "epoch": 1.8219845447418523,
      "grad_norm": 1.148931860923767,
      "learning_rate": 9.670473684210528e-05,
      "loss": 0.4007,
      "step": 32537
    },
    {
      "epoch": 1.8220405420539814,
      "grad_norm": 1.3664186000823975,
      "learning_rate": 9.670447368421052e-05,
      "loss": 0.4464,
      "step": 32538
    },
    {
      "epoch": 1.8220965393661104,
      "grad_norm": 1.6627272367477417,
      "learning_rate": 9.67042105263158e-05,
      "loss": 0.5376,
      "step": 32539
    },
    {
      "epoch": 1.8221525366782394,
      "grad_norm": 1.3919153213500977,
      "learning_rate": 9.670394736842106e-05,
      "loss": 0.4488,
      "step": 32540
    },
    {
      "epoch": 1.8222085339903684,
      "grad_norm": 1.199384093284607,
      "learning_rate": 9.670368421052633e-05,
      "loss": 0.5649,
      "step": 32541
    },
    {
      "epoch": 1.8222645313024974,
      "grad_norm": 1.414018988609314,
      "learning_rate": 9.670342105263159e-05,
      "loss": 0.5207,
      "step": 32542
    },
    {
      "epoch": 1.8223205286146265,
      "grad_norm": 1.116434097290039,
      "learning_rate": 9.670315789473685e-05,
      "loss": 0.3832,
      "step": 32543
    },
    {
      "epoch": 1.8223765259267555,
      "grad_norm": 1.2003717422485352,
      "learning_rate": 9.670289473684211e-05,
      "loss": 0.441,
      "step": 32544
    },
    {
      "epoch": 1.8224325232388845,
      "grad_norm": 1.4894652366638184,
      "learning_rate": 9.670263157894738e-05,
      "loss": 0.4912,
      "step": 32545
    },
    {
      "epoch": 1.8224885205510135,
      "grad_norm": 1.1877217292785645,
      "learning_rate": 9.670236842105264e-05,
      "loss": 0.4977,
      "step": 32546
    },
    {
      "epoch": 1.8225445178631425,
      "grad_norm": 1.110384464263916,
      "learning_rate": 9.670210526315789e-05,
      "loss": 0.3216,
      "step": 32547
    },
    {
      "epoch": 1.8226005151752716,
      "grad_norm": 1.4273176193237305,
      "learning_rate": 9.670184210526316e-05,
      "loss": 0.4597,
      "step": 32548
    },
    {
      "epoch": 1.8226565124874006,
      "grad_norm": 1.3708255290985107,
      "learning_rate": 9.670157894736842e-05,
      "loss": 0.4172,
      "step": 32549
    },
    {
      "epoch": 1.8227125097995296,
      "grad_norm": 1.3241392374038696,
      "learning_rate": 9.67013157894737e-05,
      "loss": 0.4817,
      "step": 32550
    },
    {
      "epoch": 1.8227685071116586,
      "grad_norm": 1.4388078451156616,
      "learning_rate": 9.670105263157895e-05,
      "loss": 0.3829,
      "step": 32551
    },
    {
      "epoch": 1.8228245044237876,
      "grad_norm": 1.3396120071411133,
      "learning_rate": 9.670078947368421e-05,
      "loss": 0.3631,
      "step": 32552
    },
    {
      "epoch": 1.8228805017359166,
      "grad_norm": 1.3354370594024658,
      "learning_rate": 9.670052631578947e-05,
      "loss": 0.5774,
      "step": 32553
    },
    {
      "epoch": 1.8229364990480457,
      "grad_norm": 1.1911250352859497,
      "learning_rate": 9.670026315789475e-05,
      "loss": 0.3018,
      "step": 32554
    },
    {
      "epoch": 1.8229924963601747,
      "grad_norm": 1.2693029642105103,
      "learning_rate": 9.67e-05,
      "loss": 0.518,
      "step": 32555
    },
    {
      "epoch": 1.8230484936723037,
      "grad_norm": 1.420599102973938,
      "learning_rate": 9.669973684210527e-05,
      "loss": 0.4161,
      "step": 32556
    },
    {
      "epoch": 1.8231044909844327,
      "grad_norm": 1.2358683347702026,
      "learning_rate": 9.669947368421053e-05,
      "loss": 0.4386,
      "step": 32557
    },
    {
      "epoch": 1.8231604882965617,
      "grad_norm": 1.0605735778808594,
      "learning_rate": 9.66992105263158e-05,
      "loss": 0.4498,
      "step": 32558
    },
    {
      "epoch": 1.8232164856086908,
      "grad_norm": 3.1415019035339355,
      "learning_rate": 9.669894736842106e-05,
      "loss": 0.5537,
      "step": 32559
    },
    {
      "epoch": 1.8232724829208198,
      "grad_norm": 1.2564799785614014,
      "learning_rate": 9.669868421052632e-05,
      "loss": 0.4851,
      "step": 32560
    },
    {
      "epoch": 1.8233284802329488,
      "grad_norm": 1.4091824293136597,
      "learning_rate": 9.669842105263158e-05,
      "loss": 0.4864,
      "step": 32561
    },
    {
      "epoch": 1.8233844775450778,
      "grad_norm": 1.2649959325790405,
      "learning_rate": 9.669815789473685e-05,
      "loss": 0.4714,
      "step": 32562
    },
    {
      "epoch": 1.8234404748572068,
      "grad_norm": 1.2252275943756104,
      "learning_rate": 9.669789473684211e-05,
      "loss": 0.56,
      "step": 32563
    },
    {
      "epoch": 1.8234964721693359,
      "grad_norm": 1.1678951978683472,
      "learning_rate": 9.669763157894738e-05,
      "loss": 0.4308,
      "step": 32564
    },
    {
      "epoch": 1.8235524694814649,
      "grad_norm": 1.6501609086990356,
      "learning_rate": 9.669736842105263e-05,
      "loss": 0.4936,
      "step": 32565
    },
    {
      "epoch": 1.823608466793594,
      "grad_norm": 1.3688122034072876,
      "learning_rate": 9.669710526315789e-05,
      "loss": 0.5517,
      "step": 32566
    },
    {
      "epoch": 1.823664464105723,
      "grad_norm": 1.4094111919403076,
      "learning_rate": 9.669684210526316e-05,
      "loss": 0.5128,
      "step": 32567
    },
    {
      "epoch": 1.823720461417852,
      "grad_norm": 1.262662649154663,
      "learning_rate": 9.669657894736842e-05,
      "loss": 0.4281,
      "step": 32568
    },
    {
      "epoch": 1.823776458729981,
      "grad_norm": 1.237623929977417,
      "learning_rate": 9.66963157894737e-05,
      "loss": 0.4589,
      "step": 32569
    },
    {
      "epoch": 1.82383245604211,
      "grad_norm": 1.4218124151229858,
      "learning_rate": 9.669605263157894e-05,
      "loss": 0.4109,
      "step": 32570
    },
    {
      "epoch": 1.823888453354239,
      "grad_norm": 1.2873589992523193,
      "learning_rate": 9.669578947368422e-05,
      "loss": 0.4809,
      "step": 32571
    },
    {
      "epoch": 1.823944450666368,
      "grad_norm": 1.2886723279953003,
      "learning_rate": 9.669552631578948e-05,
      "loss": 0.5073,
      "step": 32572
    },
    {
      "epoch": 1.824000447978497,
      "grad_norm": 1.300972580909729,
      "learning_rate": 9.669526315789475e-05,
      "loss": 0.5037,
      "step": 32573
    },
    {
      "epoch": 1.824056445290626,
      "grad_norm": 1.2771070003509521,
      "learning_rate": 9.669500000000001e-05,
      "loss": 0.5444,
      "step": 32574
    },
    {
      "epoch": 1.824112442602755,
      "grad_norm": 1.7403132915496826,
      "learning_rate": 9.669473684210527e-05,
      "loss": 0.5305,
      "step": 32575
    },
    {
      "epoch": 1.824168439914884,
      "grad_norm": 1.533210277557373,
      "learning_rate": 9.669447368421053e-05,
      "loss": 0.5951,
      "step": 32576
    },
    {
      "epoch": 1.824224437227013,
      "grad_norm": 1.3646222352981567,
      "learning_rate": 9.66942105263158e-05,
      "loss": 0.5527,
      "step": 32577
    },
    {
      "epoch": 1.8242804345391421,
      "grad_norm": 1.0207258462905884,
      "learning_rate": 9.669394736842106e-05,
      "loss": 0.4674,
      "step": 32578
    },
    {
      "epoch": 1.8243364318512711,
      "grad_norm": 1.1555756330490112,
      "learning_rate": 9.669368421052632e-05,
      "loss": 0.4146,
      "step": 32579
    },
    {
      "epoch": 1.8243924291634002,
      "grad_norm": 1.2460813522338867,
      "learning_rate": 9.669342105263158e-05,
      "loss": 0.4015,
      "step": 32580
    },
    {
      "epoch": 1.8244484264755292,
      "grad_norm": 1.219386100769043,
      "learning_rate": 9.669315789473685e-05,
      "loss": 0.4457,
      "step": 32581
    },
    {
      "epoch": 1.8245044237876582,
      "grad_norm": 1.1244335174560547,
      "learning_rate": 9.669289473684211e-05,
      "loss": 0.4759,
      "step": 32582
    },
    {
      "epoch": 1.8245604210997872,
      "grad_norm": 1.125978946685791,
      "learning_rate": 9.669263157894737e-05,
      "loss": 0.503,
      "step": 32583
    },
    {
      "epoch": 1.8246164184119162,
      "grad_norm": 1.4441940784454346,
      "learning_rate": 9.669236842105263e-05,
      "loss": 0.5164,
      "step": 32584
    },
    {
      "epoch": 1.8246724157240453,
      "grad_norm": 1.0562191009521484,
      "learning_rate": 9.669210526315789e-05,
      "loss": 0.4711,
      "step": 32585
    },
    {
      "epoch": 1.8247284130361743,
      "grad_norm": 1.3122776746749878,
      "learning_rate": 9.669184210526317e-05,
      "loss": 0.4623,
      "step": 32586
    },
    {
      "epoch": 1.8247844103483033,
      "grad_norm": 1.3073694705963135,
      "learning_rate": 9.669157894736843e-05,
      "loss": 0.4393,
      "step": 32587
    },
    {
      "epoch": 1.8248404076604323,
      "grad_norm": 2.049767017364502,
      "learning_rate": 9.669131578947369e-05,
      "loss": 0.5651,
      "step": 32588
    },
    {
      "epoch": 1.8248964049725613,
      "grad_norm": 1.6820993423461914,
      "learning_rate": 9.669105263157895e-05,
      "loss": 0.3945,
      "step": 32589
    },
    {
      "epoch": 1.8249524022846904,
      "grad_norm": 1.2036643028259277,
      "learning_rate": 9.669078947368422e-05,
      "loss": 0.3957,
      "step": 32590
    },
    {
      "epoch": 1.8250083995968194,
      "grad_norm": 1.2300925254821777,
      "learning_rate": 9.669052631578948e-05,
      "loss": 0.4814,
      "step": 32591
    },
    {
      "epoch": 1.8250643969089484,
      "grad_norm": 1.1175167560577393,
      "learning_rate": 9.669026315789475e-05,
      "loss": 0.4314,
      "step": 32592
    },
    {
      "epoch": 1.8251203942210774,
      "grad_norm": 1.4281058311462402,
      "learning_rate": 9.669e-05,
      "loss": 0.4205,
      "step": 32593
    },
    {
      "epoch": 1.8251763915332064,
      "grad_norm": 1.2252016067504883,
      "learning_rate": 9.668973684210527e-05,
      "loss": 0.4933,
      "step": 32594
    },
    {
      "epoch": 1.8252323888453355,
      "grad_norm": 1.2222179174423218,
      "learning_rate": 9.668947368421053e-05,
      "loss": 0.3959,
      "step": 32595
    },
    {
      "epoch": 1.8252883861574645,
      "grad_norm": 1.1796728372573853,
      "learning_rate": 9.66892105263158e-05,
      "loss": 0.4052,
      "step": 32596
    },
    {
      "epoch": 1.8253443834695935,
      "grad_norm": 1.116337537765503,
      "learning_rate": 9.668894736842105e-05,
      "loss": 0.4055,
      "step": 32597
    },
    {
      "epoch": 1.8254003807817225,
      "grad_norm": 1.5841273069381714,
      "learning_rate": 9.668868421052632e-05,
      "loss": 0.3562,
      "step": 32598
    },
    {
      "epoch": 1.8254563780938515,
      "grad_norm": 1.1964962482452393,
      "learning_rate": 9.668842105263158e-05,
      "loss": 0.4176,
      "step": 32599
    },
    {
      "epoch": 1.8255123754059805,
      "grad_norm": 1.1066843271255493,
      "learning_rate": 9.668815789473684e-05,
      "loss": 0.3333,
      "step": 32600
    },
    {
      "epoch": 1.8255683727181096,
      "grad_norm": 1.0237269401550293,
      "learning_rate": 9.668789473684212e-05,
      "loss": 0.4636,
      "step": 32601
    },
    {
      "epoch": 1.8256243700302386,
      "grad_norm": 1.3789377212524414,
      "learning_rate": 9.668763157894736e-05,
      "loss": 0.4183,
      "step": 32602
    },
    {
      "epoch": 1.8256803673423676,
      "grad_norm": 1.4113359451293945,
      "learning_rate": 9.668736842105264e-05,
      "loss": 0.5721,
      "step": 32603
    },
    {
      "epoch": 1.8257363646544966,
      "grad_norm": 1.2128055095672607,
      "learning_rate": 9.66871052631579e-05,
      "loss": 0.4906,
      "step": 32604
    },
    {
      "epoch": 1.8257923619666256,
      "grad_norm": 1.167681336402893,
      "learning_rate": 9.668684210526317e-05,
      "loss": 0.3832,
      "step": 32605
    },
    {
      "epoch": 1.8258483592787547,
      "grad_norm": 1.196630597114563,
      "learning_rate": 9.668657894736843e-05,
      "loss": 0.4218,
      "step": 32606
    },
    {
      "epoch": 1.8259043565908837,
      "grad_norm": 1.136807918548584,
      "learning_rate": 9.668631578947369e-05,
      "loss": 0.5089,
      "step": 32607
    },
    {
      "epoch": 1.8259603539030127,
      "grad_norm": 1.463698387145996,
      "learning_rate": 9.668605263157895e-05,
      "loss": 0.548,
      "step": 32608
    },
    {
      "epoch": 1.8260163512151417,
      "grad_norm": 1.3486815690994263,
      "learning_rate": 9.668578947368422e-05,
      "loss": 0.4634,
      "step": 32609
    },
    {
      "epoch": 1.8260723485272707,
      "grad_norm": 5.1178998947143555,
      "learning_rate": 9.668552631578948e-05,
      "loss": 0.4325,
      "step": 32610
    },
    {
      "epoch": 1.8261283458393998,
      "grad_norm": 1.2187914848327637,
      "learning_rate": 9.668526315789474e-05,
      "loss": 0.4171,
      "step": 32611
    },
    {
      "epoch": 1.8261843431515288,
      "grad_norm": 1.1664485931396484,
      "learning_rate": 9.6685e-05,
      "loss": 0.408,
      "step": 32612
    },
    {
      "epoch": 1.8262403404636578,
      "grad_norm": 1.2440263032913208,
      "learning_rate": 9.668473684210527e-05,
      "loss": 0.4748,
      "step": 32613
    },
    {
      "epoch": 1.8262963377757868,
      "grad_norm": 1.2527198791503906,
      "learning_rate": 9.668447368421053e-05,
      "loss": 0.514,
      "step": 32614
    },
    {
      "epoch": 1.8263523350879158,
      "grad_norm": 1.047395944595337,
      "learning_rate": 9.668421052631579e-05,
      "loss": 0.4122,
      "step": 32615
    },
    {
      "epoch": 1.8264083324000446,
      "grad_norm": 1.3422802686691284,
      "learning_rate": 9.668394736842105e-05,
      "loss": 0.7273,
      "step": 32616
    },
    {
      "epoch": 1.8264643297121737,
      "grad_norm": 1.388197660446167,
      "learning_rate": 9.668368421052631e-05,
      "loss": 0.4851,
      "step": 32617
    },
    {
      "epoch": 1.8265203270243027,
      "grad_norm": 1.5944206714630127,
      "learning_rate": 9.668342105263159e-05,
      "loss": 0.5703,
      "step": 32618
    },
    {
      "epoch": 1.8265763243364317,
      "grad_norm": 1.9509798288345337,
      "learning_rate": 9.668315789473685e-05,
      "loss": 0.4805,
      "step": 32619
    },
    {
      "epoch": 1.8266323216485607,
      "grad_norm": 1.3672857284545898,
      "learning_rate": 9.66828947368421e-05,
      "loss": 0.5626,
      "step": 32620
    },
    {
      "epoch": 1.8266883189606897,
      "grad_norm": 1.4626967906951904,
      "learning_rate": 9.668263157894736e-05,
      "loss": 0.3086,
      "step": 32621
    },
    {
      "epoch": 1.8267443162728187,
      "grad_norm": 1.1000856161117554,
      "learning_rate": 9.668236842105264e-05,
      "loss": 0.3826,
      "step": 32622
    },
    {
      "epoch": 1.8268003135849478,
      "grad_norm": 1.2412537336349487,
      "learning_rate": 9.66821052631579e-05,
      "loss": 0.3716,
      "step": 32623
    },
    {
      "epoch": 1.8268563108970768,
      "grad_norm": 1.2429136037826538,
      "learning_rate": 9.668184210526317e-05,
      "loss": 0.4128,
      "step": 32624
    },
    {
      "epoch": 1.8269123082092058,
      "grad_norm": 1.3121851682662964,
      "learning_rate": 9.668157894736842e-05,
      "loss": 0.4995,
      "step": 32625
    },
    {
      "epoch": 1.8269683055213348,
      "grad_norm": 1.2313607931137085,
      "learning_rate": 9.668131578947369e-05,
      "loss": 0.489,
      "step": 32626
    },
    {
      "epoch": 1.8270243028334638,
      "grad_norm": 1.1633257865905762,
      "learning_rate": 9.668105263157895e-05,
      "loss": 0.4584,
      "step": 32627
    },
    {
      "epoch": 1.8270803001455929,
      "grad_norm": 1.3403372764587402,
      "learning_rate": 9.668078947368422e-05,
      "loss": 0.4156,
      "step": 32628
    },
    {
      "epoch": 1.8271362974577219,
      "grad_norm": 1.3637710809707642,
      "learning_rate": 9.668052631578948e-05,
      "loss": 0.5571,
      "step": 32629
    },
    {
      "epoch": 1.827192294769851,
      "grad_norm": 1.3844184875488281,
      "learning_rate": 9.668026315789474e-05,
      "loss": 0.483,
      "step": 32630
    },
    {
      "epoch": 1.82724829208198,
      "grad_norm": 1.2805343866348267,
      "learning_rate": 9.668e-05,
      "loss": 0.4295,
      "step": 32631
    },
    {
      "epoch": 1.827304289394109,
      "grad_norm": 1.489109992980957,
      "learning_rate": 9.667973684210528e-05,
      "loss": 0.5253,
      "step": 32632
    },
    {
      "epoch": 1.827360286706238,
      "grad_norm": 1.432491421699524,
      "learning_rate": 9.667947368421054e-05,
      "loss": 0.5053,
      "step": 32633
    },
    {
      "epoch": 1.827416284018367,
      "grad_norm": 1.2370774745941162,
      "learning_rate": 9.66792105263158e-05,
      "loss": 0.386,
      "step": 32634
    },
    {
      "epoch": 1.827472281330496,
      "grad_norm": 1.4065133333206177,
      "learning_rate": 9.667894736842106e-05,
      "loss": 0.4296,
      "step": 32635
    },
    {
      "epoch": 1.827528278642625,
      "grad_norm": 1.4006291627883911,
      "learning_rate": 9.667868421052631e-05,
      "loss": 0.4126,
      "step": 32636
    },
    {
      "epoch": 1.827584275954754,
      "grad_norm": 1.0816266536712646,
      "learning_rate": 9.667842105263159e-05,
      "loss": 0.3676,
      "step": 32637
    },
    {
      "epoch": 1.827640273266883,
      "grad_norm": 1.4172054529190063,
      "learning_rate": 9.667815789473685e-05,
      "loss": 0.44,
      "step": 32638
    },
    {
      "epoch": 1.827696270579012,
      "grad_norm": 1.3221161365509033,
      "learning_rate": 9.667789473684211e-05,
      "loss": 0.467,
      "step": 32639
    },
    {
      "epoch": 1.827752267891141,
      "grad_norm": 1.667936086654663,
      "learning_rate": 9.667763157894737e-05,
      "loss": 0.6794,
      "step": 32640
    },
    {
      "epoch": 1.82780826520327,
      "grad_norm": 1.3499414920806885,
      "learning_rate": 9.667736842105264e-05,
      "loss": 0.4399,
      "step": 32641
    },
    {
      "epoch": 1.8278642625153991,
      "grad_norm": 1.3167911767959595,
      "learning_rate": 9.66771052631579e-05,
      "loss": 0.511,
      "step": 32642
    },
    {
      "epoch": 1.8279202598275281,
      "grad_norm": 1.287710189819336,
      "learning_rate": 9.667684210526316e-05,
      "loss": 0.4156,
      "step": 32643
    },
    {
      "epoch": 1.8279762571396572,
      "grad_norm": 1.2779874801635742,
      "learning_rate": 9.667657894736842e-05,
      "loss": 0.4393,
      "step": 32644
    },
    {
      "epoch": 1.8280322544517862,
      "grad_norm": 1.7091296911239624,
      "learning_rate": 9.667631578947369e-05,
      "loss": 0.4468,
      "step": 32645
    },
    {
      "epoch": 1.8280882517639152,
      "grad_norm": 1.4625667333602905,
      "learning_rate": 9.667605263157895e-05,
      "loss": 0.4958,
      "step": 32646
    },
    {
      "epoch": 1.8281442490760442,
      "grad_norm": 1.1907398700714111,
      "learning_rate": 9.667578947368423e-05,
      "loss": 0.3929,
      "step": 32647
    },
    {
      "epoch": 1.8282002463881732,
      "grad_norm": 1.2688395977020264,
      "learning_rate": 9.667552631578947e-05,
      "loss": 0.4314,
      "step": 32648
    },
    {
      "epoch": 1.8282562437003023,
      "grad_norm": 1.3119951486587524,
      "learning_rate": 9.667526315789475e-05,
      "loss": 0.4976,
      "step": 32649
    },
    {
      "epoch": 1.8283122410124313,
      "grad_norm": 1.057431697845459,
      "learning_rate": 9.6675e-05,
      "loss": 0.3274,
      "step": 32650
    },
    {
      "epoch": 1.8283682383245603,
      "grad_norm": 1.3666818141937256,
      "learning_rate": 9.667473684210527e-05,
      "loss": 0.435,
      "step": 32651
    },
    {
      "epoch": 1.8284242356366893,
      "grad_norm": 1.2738220691680908,
      "learning_rate": 9.667447368421052e-05,
      "loss": 0.4332,
      "step": 32652
    },
    {
      "epoch": 1.8284802329488183,
      "grad_norm": 1.235003113746643,
      "learning_rate": 9.667421052631578e-05,
      "loss": 0.379,
      "step": 32653
    },
    {
      "epoch": 1.8285362302609474,
      "grad_norm": 1.3275306224822998,
      "learning_rate": 9.667394736842106e-05,
      "loss": 0.4661,
      "step": 32654
    },
    {
      "epoch": 1.8285922275730764,
      "grad_norm": 1.3707810640335083,
      "learning_rate": 9.667368421052632e-05,
      "loss": 0.3822,
      "step": 32655
    },
    {
      "epoch": 1.8286482248852054,
      "grad_norm": 1.2254188060760498,
      "learning_rate": 9.667342105263159e-05,
      "loss": 0.5657,
      "step": 32656
    },
    {
      "epoch": 1.8287042221973344,
      "grad_norm": 1.3151730298995972,
      "learning_rate": 9.667315789473684e-05,
      "loss": 0.5042,
      "step": 32657
    },
    {
      "epoch": 1.8287602195094634,
      "grad_norm": 1.1538809537887573,
      "learning_rate": 9.667289473684211e-05,
      "loss": 0.4685,
      "step": 32658
    },
    {
      "epoch": 1.8288162168215925,
      "grad_norm": 1.4197208881378174,
      "learning_rate": 9.667263157894737e-05,
      "loss": 0.5985,
      "step": 32659
    },
    {
      "epoch": 1.8288722141337215,
      "grad_norm": 1.1300543546676636,
      "learning_rate": 9.667236842105264e-05,
      "loss": 0.4008,
      "step": 32660
    },
    {
      "epoch": 1.8289282114458505,
      "grad_norm": 1.0233668088912964,
      "learning_rate": 9.66721052631579e-05,
      "loss": 0.3974,
      "step": 32661
    },
    {
      "epoch": 1.8289842087579795,
      "grad_norm": 1.1356687545776367,
      "learning_rate": 9.667184210526316e-05,
      "loss": 0.4677,
      "step": 32662
    },
    {
      "epoch": 1.8290402060701085,
      "grad_norm": 1.3866469860076904,
      "learning_rate": 9.667157894736842e-05,
      "loss": 0.4227,
      "step": 32663
    },
    {
      "epoch": 1.8290962033822376,
      "grad_norm": 1.1879624128341675,
      "learning_rate": 9.66713157894737e-05,
      "loss": 0.3934,
      "step": 32664
    },
    {
      "epoch": 1.8291522006943666,
      "grad_norm": 1.0241892337799072,
      "learning_rate": 9.667105263157896e-05,
      "loss": 0.3762,
      "step": 32665
    },
    {
      "epoch": 1.8292081980064956,
      "grad_norm": 1.2172296047210693,
      "learning_rate": 9.667078947368422e-05,
      "loss": 0.401,
      "step": 32666
    },
    {
      "epoch": 1.8292641953186246,
      "grad_norm": 1.5045950412750244,
      "learning_rate": 9.667052631578947e-05,
      "loss": 0.4328,
      "step": 32667
    },
    {
      "epoch": 1.8293201926307536,
      "grad_norm": 1.503819227218628,
      "learning_rate": 9.667026315789473e-05,
      "loss": 0.5246,
      "step": 32668
    },
    {
      "epoch": 1.8293761899428826,
      "grad_norm": 1.2372878789901733,
      "learning_rate": 9.667000000000001e-05,
      "loss": 0.521,
      "step": 32669
    },
    {
      "epoch": 1.8294321872550117,
      "grad_norm": 1.2774956226348877,
      "learning_rate": 9.666973684210527e-05,
      "loss": 0.4547,
      "step": 32670
    },
    {
      "epoch": 1.8294881845671407,
      "grad_norm": 1.1652895212173462,
      "learning_rate": 9.666947368421053e-05,
      "loss": 0.2893,
      "step": 32671
    },
    {
      "epoch": 1.8295441818792697,
      "grad_norm": 1.0408943891525269,
      "learning_rate": 9.666921052631579e-05,
      "loss": 0.4493,
      "step": 32672
    },
    {
      "epoch": 1.8296001791913987,
      "grad_norm": 1.4637556076049805,
      "learning_rate": 9.666894736842106e-05,
      "loss": 0.5793,
      "step": 32673
    },
    {
      "epoch": 1.8296561765035277,
      "grad_norm": 1.8593851327896118,
      "learning_rate": 9.666868421052632e-05,
      "loss": 0.5905,
      "step": 32674
    },
    {
      "epoch": 1.8297121738156568,
      "grad_norm": 1.6842243671417236,
      "learning_rate": 9.666842105263158e-05,
      "loss": 0.528,
      "step": 32675
    },
    {
      "epoch": 1.8297681711277858,
      "grad_norm": 1.177577257156372,
      "learning_rate": 9.666815789473684e-05,
      "loss": 0.4878,
      "step": 32676
    },
    {
      "epoch": 1.8298241684399148,
      "grad_norm": 1.218920111656189,
      "learning_rate": 9.666789473684211e-05,
      "loss": 0.4243,
      "step": 32677
    },
    {
      "epoch": 1.8298801657520438,
      "grad_norm": 1.0772826671600342,
      "learning_rate": 9.666763157894737e-05,
      "loss": 0.4324,
      "step": 32678
    },
    {
      "epoch": 1.8299361630641728,
      "grad_norm": 1.2124522924423218,
      "learning_rate": 9.666736842105265e-05,
      "loss": 0.4393,
      "step": 32679
    },
    {
      "epoch": 1.8299921603763019,
      "grad_norm": 1.0846668481826782,
      "learning_rate": 9.666710526315789e-05,
      "loss": 0.3997,
      "step": 32680
    },
    {
      "epoch": 1.8300481576884309,
      "grad_norm": 1.091623306274414,
      "learning_rate": 9.666684210526317e-05,
      "loss": 0.3132,
      "step": 32681
    },
    {
      "epoch": 1.83010415500056,
      "grad_norm": 1.125016689300537,
      "learning_rate": 9.666657894736842e-05,
      "loss": 0.4582,
      "step": 32682
    },
    {
      "epoch": 1.830160152312689,
      "grad_norm": 1.3759264945983887,
      "learning_rate": 9.66663157894737e-05,
      "loss": 0.4266,
      "step": 32683
    },
    {
      "epoch": 1.830216149624818,
      "grad_norm": 1.172193169593811,
      "learning_rate": 9.666605263157896e-05,
      "loss": 0.3767,
      "step": 32684
    },
    {
      "epoch": 1.830272146936947,
      "grad_norm": 1.1010347604751587,
      "learning_rate": 9.666578947368422e-05,
      "loss": 0.3532,
      "step": 32685
    },
    {
      "epoch": 1.830328144249076,
      "grad_norm": 1.3016178607940674,
      "learning_rate": 9.666552631578948e-05,
      "loss": 0.3954,
      "step": 32686
    },
    {
      "epoch": 1.830384141561205,
      "grad_norm": 1.693276286125183,
      "learning_rate": 9.666526315789474e-05,
      "loss": 0.5813,
      "step": 32687
    },
    {
      "epoch": 1.830440138873334,
      "grad_norm": 1.37324857711792,
      "learning_rate": 9.666500000000001e-05,
      "loss": 0.5363,
      "step": 32688
    },
    {
      "epoch": 1.830496136185463,
      "grad_norm": 1.3337602615356445,
      "learning_rate": 9.666473684210527e-05,
      "loss": 0.3974,
      "step": 32689
    },
    {
      "epoch": 1.830552133497592,
      "grad_norm": 1.5725585222244263,
      "learning_rate": 9.666447368421053e-05,
      "loss": 0.4479,
      "step": 32690
    },
    {
      "epoch": 1.830608130809721,
      "grad_norm": 1.1100361347198486,
      "learning_rate": 9.666421052631579e-05,
      "loss": 0.3584,
      "step": 32691
    },
    {
      "epoch": 1.83066412812185,
      "grad_norm": 1.5025217533111572,
      "learning_rate": 9.666394736842106e-05,
      "loss": 0.5089,
      "step": 32692
    },
    {
      "epoch": 1.830720125433979,
      "grad_norm": 1.9290908575057983,
      "learning_rate": 9.666368421052632e-05,
      "loss": 0.3646,
      "step": 32693
    },
    {
      "epoch": 1.8307761227461081,
      "grad_norm": 1.5080006122589111,
      "learning_rate": 9.666342105263158e-05,
      "loss": 0.561,
      "step": 32694
    },
    {
      "epoch": 1.8308321200582371,
      "grad_norm": 1.1862131357192993,
      "learning_rate": 9.666315789473684e-05,
      "loss": 0.3779,
      "step": 32695
    },
    {
      "epoch": 1.8308881173703662,
      "grad_norm": 1.2814775705337524,
      "learning_rate": 9.666289473684212e-05,
      "loss": 0.4245,
      "step": 32696
    },
    {
      "epoch": 1.8309441146824952,
      "grad_norm": 1.4017432928085327,
      "learning_rate": 9.666263157894738e-05,
      "loss": 0.4706,
      "step": 32697
    },
    {
      "epoch": 1.8310001119946242,
      "grad_norm": 1.4507789611816406,
      "learning_rate": 9.666236842105263e-05,
      "loss": 0.4321,
      "step": 32698
    },
    {
      "epoch": 1.8310561093067532,
      "grad_norm": 1.382832646369934,
      "learning_rate": 9.66621052631579e-05,
      "loss": 0.4984,
      "step": 32699
    },
    {
      "epoch": 1.8311121066188822,
      "grad_norm": 1.1069839000701904,
      "learning_rate": 9.666184210526317e-05,
      "loss": 0.4409,
      "step": 32700
    },
    {
      "epoch": 1.8311681039310113,
      "grad_norm": 1.4302692413330078,
      "learning_rate": 9.666157894736843e-05,
      "loss": 0.4594,
      "step": 32701
    },
    {
      "epoch": 1.8312241012431403,
      "grad_norm": 1.1549898386001587,
      "learning_rate": 9.66613157894737e-05,
      "loss": 0.3996,
      "step": 32702
    },
    {
      "epoch": 1.8312800985552693,
      "grad_norm": 1.374919056892395,
      "learning_rate": 9.666105263157895e-05,
      "loss": 0.4857,
      "step": 32703
    },
    {
      "epoch": 1.8313360958673983,
      "grad_norm": 1.1142358779907227,
      "learning_rate": 9.66607894736842e-05,
      "loss": 0.3632,
      "step": 32704
    },
    {
      "epoch": 1.8313920931795273,
      "grad_norm": 1.3335648775100708,
      "learning_rate": 9.666052631578948e-05,
      "loss": 0.5292,
      "step": 32705
    },
    {
      "epoch": 1.8314480904916564,
      "grad_norm": 1.409423589706421,
      "learning_rate": 9.666026315789474e-05,
      "loss": 0.3871,
      "step": 32706
    },
    {
      "epoch": 1.8315040878037854,
      "grad_norm": 1.2985585927963257,
      "learning_rate": 9.666e-05,
      "loss": 0.5195,
      "step": 32707
    },
    {
      "epoch": 1.8315600851159144,
      "grad_norm": 1.509708046913147,
      "learning_rate": 9.665973684210526e-05,
      "loss": 0.4044,
      "step": 32708
    },
    {
      "epoch": 1.8316160824280434,
      "grad_norm": 1.2600221633911133,
      "learning_rate": 9.665947368421053e-05,
      "loss": 0.3483,
      "step": 32709
    },
    {
      "epoch": 1.8316720797401724,
      "grad_norm": 1.5613957643508911,
      "learning_rate": 9.665921052631579e-05,
      "loss": 0.4571,
      "step": 32710
    },
    {
      "epoch": 1.8317280770523015,
      "grad_norm": 1.3759835958480835,
      "learning_rate": 9.665894736842107e-05,
      "loss": 0.4701,
      "step": 32711
    },
    {
      "epoch": 1.8317840743644305,
      "grad_norm": 1.1702947616577148,
      "learning_rate": 9.665868421052631e-05,
      "loss": 0.3795,
      "step": 32712
    },
    {
      "epoch": 1.8318400716765595,
      "grad_norm": 3.083773374557495,
      "learning_rate": 9.665842105263158e-05,
      "loss": 0.3771,
      "step": 32713
    },
    {
      "epoch": 1.8318960689886885,
      "grad_norm": 1.134027361869812,
      "learning_rate": 9.665815789473684e-05,
      "loss": 0.5371,
      "step": 32714
    },
    {
      "epoch": 1.8319520663008175,
      "grad_norm": 9.868935585021973,
      "learning_rate": 9.665789473684212e-05,
      "loss": 0.7914,
      "step": 32715
    },
    {
      "epoch": 1.8320080636129465,
      "grad_norm": 1.3018718957901,
      "learning_rate": 9.665763157894738e-05,
      "loss": 0.4281,
      "step": 32716
    },
    {
      "epoch": 1.8320640609250756,
      "grad_norm": 1.113433837890625,
      "learning_rate": 9.665736842105264e-05,
      "loss": 0.4189,
      "step": 32717
    },
    {
      "epoch": 1.8321200582372046,
      "grad_norm": 1.243665337562561,
      "learning_rate": 9.66571052631579e-05,
      "loss": 0.49,
      "step": 32718
    },
    {
      "epoch": 1.8321760555493336,
      "grad_norm": 1.505263090133667,
      "learning_rate": 9.665684210526317e-05,
      "loss": 0.4262,
      "step": 32719
    },
    {
      "epoch": 1.8322320528614626,
      "grad_norm": 1.5680044889450073,
      "learning_rate": 9.665657894736843e-05,
      "loss": 0.4489,
      "step": 32720
    },
    {
      "epoch": 1.8322880501735916,
      "grad_norm": 1.7051466703414917,
      "learning_rate": 9.665631578947369e-05,
      "loss": 0.559,
      "step": 32721
    },
    {
      "epoch": 1.8323440474857207,
      "grad_norm": 1.1974444389343262,
      "learning_rate": 9.665605263157895e-05,
      "loss": 0.4361,
      "step": 32722
    },
    {
      "epoch": 1.8324000447978497,
      "grad_norm": 1.1413769721984863,
      "learning_rate": 9.665578947368421e-05,
      "loss": 0.4257,
      "step": 32723
    },
    {
      "epoch": 1.8324560421099787,
      "grad_norm": 0.9643688201904297,
      "learning_rate": 9.665552631578948e-05,
      "loss": 0.3265,
      "step": 32724
    },
    {
      "epoch": 1.8325120394221077,
      "grad_norm": 1.388958215713501,
      "learning_rate": 9.665526315789474e-05,
      "loss": 0.5034,
      "step": 32725
    },
    {
      "epoch": 1.8325680367342367,
      "grad_norm": 1.190285563468933,
      "learning_rate": 9.6655e-05,
      "loss": 0.4158,
      "step": 32726
    },
    {
      "epoch": 1.8326240340463658,
      "grad_norm": 1.7620315551757812,
      "learning_rate": 9.665473684210526e-05,
      "loss": 0.3913,
      "step": 32727
    },
    {
      "epoch": 1.8326800313584948,
      "grad_norm": 1.3899251222610474,
      "learning_rate": 9.665447368421054e-05,
      "loss": 0.3877,
      "step": 32728
    },
    {
      "epoch": 1.8327360286706238,
      "grad_norm": 1.5617096424102783,
      "learning_rate": 9.66542105263158e-05,
      "loss": 0.4683,
      "step": 32729
    },
    {
      "epoch": 1.8327920259827528,
      "grad_norm": 1.2852299213409424,
      "learning_rate": 9.665394736842105e-05,
      "loss": 0.4646,
      "step": 32730
    },
    {
      "epoch": 1.8328480232948818,
      "grad_norm": 1.3972654342651367,
      "learning_rate": 9.665368421052631e-05,
      "loss": 0.5794,
      "step": 32731
    },
    {
      "epoch": 1.8329040206070109,
      "grad_norm": 1.4230235815048218,
      "learning_rate": 9.665342105263159e-05,
      "loss": 0.3749,
      "step": 32732
    },
    {
      "epoch": 1.8329600179191399,
      "grad_norm": 1.1495620012283325,
      "learning_rate": 9.665315789473685e-05,
      "loss": 0.3909,
      "step": 32733
    },
    {
      "epoch": 1.833016015231269,
      "grad_norm": 1.1372021436691284,
      "learning_rate": 9.665289473684212e-05,
      "loss": 0.4301,
      "step": 32734
    },
    {
      "epoch": 1.833072012543398,
      "grad_norm": 1.7445937395095825,
      "learning_rate": 9.665263157894737e-05,
      "loss": 0.4723,
      "step": 32735
    },
    {
      "epoch": 1.833128009855527,
      "grad_norm": 1.34993314743042,
      "learning_rate": 9.665236842105264e-05,
      "loss": 0.3778,
      "step": 32736
    },
    {
      "epoch": 1.833184007167656,
      "grad_norm": 1.1281663179397583,
      "learning_rate": 9.66521052631579e-05,
      "loss": 0.4953,
      "step": 32737
    },
    {
      "epoch": 1.833240004479785,
      "grad_norm": 1.7411876916885376,
      "learning_rate": 9.665184210526316e-05,
      "loss": 0.5088,
      "step": 32738
    },
    {
      "epoch": 1.833296001791914,
      "grad_norm": 8.135942459106445,
      "learning_rate": 9.665157894736843e-05,
      "loss": 0.3537,
      "step": 32739
    },
    {
      "epoch": 1.833351999104043,
      "grad_norm": 1.7054061889648438,
      "learning_rate": 9.665131578947368e-05,
      "loss": 0.4314,
      "step": 32740
    },
    {
      "epoch": 1.833407996416172,
      "grad_norm": 1.3007142543792725,
      "learning_rate": 9.665105263157895e-05,
      "loss": 0.5351,
      "step": 32741
    },
    {
      "epoch": 1.833463993728301,
      "grad_norm": 1.2788482904434204,
      "learning_rate": 9.665078947368421e-05,
      "loss": 0.4453,
      "step": 32742
    },
    {
      "epoch": 1.83351999104043,
      "grad_norm": 1.6615954637527466,
      "learning_rate": 9.665052631578949e-05,
      "loss": 0.6915,
      "step": 32743
    },
    {
      "epoch": 1.833575988352559,
      "grad_norm": 0.9844016432762146,
      "learning_rate": 9.665026315789474e-05,
      "loss": 0.3056,
      "step": 32744
    },
    {
      "epoch": 1.833631985664688,
      "grad_norm": 1.385599136352539,
      "learning_rate": 9.665e-05,
      "loss": 0.3915,
      "step": 32745
    },
    {
      "epoch": 1.8336879829768171,
      "grad_norm": 1.181168794631958,
      "learning_rate": 9.664973684210526e-05,
      "loss": 0.4216,
      "step": 32746
    },
    {
      "epoch": 1.8337439802889461,
      "grad_norm": 1.248872995376587,
      "learning_rate": 9.664947368421054e-05,
      "loss": 0.4273,
      "step": 32747
    },
    {
      "epoch": 1.8337999776010752,
      "grad_norm": 1.19883131980896,
      "learning_rate": 9.66492105263158e-05,
      "loss": 0.4783,
      "step": 32748
    },
    {
      "epoch": 1.8338559749132042,
      "grad_norm": 1.430114507675171,
      "learning_rate": 9.664894736842106e-05,
      "loss": 0.4207,
      "step": 32749
    },
    {
      "epoch": 1.8339119722253332,
      "grad_norm": 1.2618263959884644,
      "learning_rate": 9.664868421052632e-05,
      "loss": 0.4088,
      "step": 32750
    },
    {
      "epoch": 1.8339679695374622,
      "grad_norm": 1.3619344234466553,
      "learning_rate": 9.664842105263159e-05,
      "loss": 0.4562,
      "step": 32751
    },
    {
      "epoch": 1.8340239668495912,
      "grad_norm": 1.2100218534469604,
      "learning_rate": 9.664815789473685e-05,
      "loss": 0.4778,
      "step": 32752
    },
    {
      "epoch": 1.8340799641617203,
      "grad_norm": 1.2954472303390503,
      "learning_rate": 9.664789473684211e-05,
      "loss": 0.3898,
      "step": 32753
    },
    {
      "epoch": 1.8341359614738493,
      "grad_norm": 1.3339674472808838,
      "learning_rate": 9.664763157894737e-05,
      "loss": 0.4605,
      "step": 32754
    },
    {
      "epoch": 1.8341919587859783,
      "grad_norm": 1.3754647970199585,
      "learning_rate": 9.664736842105263e-05,
      "loss": 0.5608,
      "step": 32755
    },
    {
      "epoch": 1.8342479560981073,
      "grad_norm": 1.32127046585083,
      "learning_rate": 9.66471052631579e-05,
      "loss": 0.4764,
      "step": 32756
    },
    {
      "epoch": 1.8343039534102363,
      "grad_norm": 1.9327669143676758,
      "learning_rate": 9.664684210526316e-05,
      "loss": 0.6438,
      "step": 32757
    },
    {
      "epoch": 1.8343599507223654,
      "grad_norm": 1.1660802364349365,
      "learning_rate": 9.664657894736842e-05,
      "loss": 0.3647,
      "step": 32758
    },
    {
      "epoch": 1.8344159480344944,
      "grad_norm": 1.2363452911376953,
      "learning_rate": 9.664631578947368e-05,
      "loss": 0.3862,
      "step": 32759
    },
    {
      "epoch": 1.8344719453466234,
      "grad_norm": 1.5387428998947144,
      "learning_rate": 9.664605263157895e-05,
      "loss": 0.4448,
      "step": 32760
    },
    {
      "epoch": 1.8345279426587524,
      "grad_norm": 1.165092945098877,
      "learning_rate": 9.664578947368421e-05,
      "loss": 0.4194,
      "step": 32761
    },
    {
      "epoch": 1.8345839399708814,
      "grad_norm": 1.6571848392486572,
      "learning_rate": 9.664552631578947e-05,
      "loss": 0.5064,
      "step": 32762
    },
    {
      "epoch": 1.8346399372830104,
      "grad_norm": 1.0997291803359985,
      "learning_rate": 9.664526315789473e-05,
      "loss": 0.463,
      "step": 32763
    },
    {
      "epoch": 1.8346959345951395,
      "grad_norm": 1.1814830303192139,
      "learning_rate": 9.664500000000001e-05,
      "loss": 0.562,
      "step": 32764
    },
    {
      "epoch": 1.8347519319072685,
      "grad_norm": 1.591491937637329,
      "learning_rate": 9.664473684210527e-05,
      "loss": 0.4498,
      "step": 32765
    },
    {
      "epoch": 1.8348079292193975,
      "grad_norm": 1.3598792552947998,
      "learning_rate": 9.664447368421054e-05,
      "loss": 0.4368,
      "step": 32766
    },
    {
      "epoch": 1.8348639265315265,
      "grad_norm": 1.3111408948898315,
      "learning_rate": 9.664421052631579e-05,
      "loss": 0.4398,
      "step": 32767
    },
    {
      "epoch": 1.8349199238436555,
      "grad_norm": 1.3196347951889038,
      "learning_rate": 9.664394736842106e-05,
      "loss": 0.3836,
      "step": 32768
    },
    {
      "epoch": 1.8349759211557846,
      "grad_norm": 1.611499547958374,
      "learning_rate": 9.664368421052632e-05,
      "loss": 0.513,
      "step": 32769
    },
    {
      "epoch": 1.8350319184679136,
      "grad_norm": 1.3398953676223755,
      "learning_rate": 9.664342105263159e-05,
      "loss": 0.3723,
      "step": 32770
    },
    {
      "epoch": 1.8350879157800426,
      "grad_norm": 1.3389477729797363,
      "learning_rate": 9.664315789473685e-05,
      "loss": 0.4886,
      "step": 32771
    },
    {
      "epoch": 1.8351439130921716,
      "grad_norm": 8.145891189575195,
      "learning_rate": 9.66428947368421e-05,
      "loss": 0.4429,
      "step": 32772
    },
    {
      "epoch": 1.8351999104043006,
      "grad_norm": 1.5232269763946533,
      "learning_rate": 9.664263157894737e-05,
      "loss": 0.4692,
      "step": 32773
    },
    {
      "epoch": 1.8352559077164297,
      "grad_norm": 1.6759427785873413,
      "learning_rate": 9.664236842105263e-05,
      "loss": 0.4715,
      "step": 32774
    },
    {
      "epoch": 1.8353119050285587,
      "grad_norm": 1.2519711256027222,
      "learning_rate": 9.66421052631579e-05,
      "loss": 0.6044,
      "step": 32775
    },
    {
      "epoch": 1.8353679023406877,
      "grad_norm": 1.3026199340820312,
      "learning_rate": 9.664184210526316e-05,
      "loss": 0.5726,
      "step": 32776
    },
    {
      "epoch": 1.8354238996528167,
      "grad_norm": 1.4376665353775024,
      "learning_rate": 9.664157894736842e-05,
      "loss": 0.4096,
      "step": 32777
    },
    {
      "epoch": 1.8354798969649457,
      "grad_norm": 1.316648244857788,
      "learning_rate": 9.664131578947368e-05,
      "loss": 0.4464,
      "step": 32778
    },
    {
      "epoch": 1.8355358942770748,
      "grad_norm": 1.3668608665466309,
      "learning_rate": 9.664105263157896e-05,
      "loss": 0.6071,
      "step": 32779
    },
    {
      "epoch": 1.8355918915892038,
      "grad_norm": 1.3676471710205078,
      "learning_rate": 9.664078947368422e-05,
      "loss": 0.4454,
      "step": 32780
    },
    {
      "epoch": 1.8356478889013328,
      "grad_norm": 1.1532204151153564,
      "learning_rate": 9.664052631578948e-05,
      "loss": 0.5273,
      "step": 32781
    },
    {
      "epoch": 1.8357038862134618,
      "grad_norm": 1.2275989055633545,
      "learning_rate": 9.664026315789474e-05,
      "loss": 0.4498,
      "step": 32782
    },
    {
      "epoch": 1.8357598835255908,
      "grad_norm": 1.100254774093628,
      "learning_rate": 9.664000000000001e-05,
      "loss": 0.5322,
      "step": 32783
    },
    {
      "epoch": 1.8358158808377198,
      "grad_norm": 1.539417028427124,
      "learning_rate": 9.663973684210527e-05,
      "loss": 0.5364,
      "step": 32784
    },
    {
      "epoch": 1.8358718781498489,
      "grad_norm": 1.8416458368301392,
      "learning_rate": 9.663947368421053e-05,
      "loss": 0.528,
      "step": 32785
    },
    {
      "epoch": 1.8359278754619779,
      "grad_norm": 1.192286729812622,
      "learning_rate": 9.663921052631579e-05,
      "loss": 0.4131,
      "step": 32786
    },
    {
      "epoch": 1.835983872774107,
      "grad_norm": 1.1289178133010864,
      "learning_rate": 9.663894736842106e-05,
      "loss": 0.4501,
      "step": 32787
    },
    {
      "epoch": 1.836039870086236,
      "grad_norm": 1.2684918642044067,
      "learning_rate": 9.663868421052632e-05,
      "loss": 0.4582,
      "step": 32788
    },
    {
      "epoch": 1.836095867398365,
      "grad_norm": 1.1682430505752563,
      "learning_rate": 9.663842105263158e-05,
      "loss": 0.458,
      "step": 32789
    },
    {
      "epoch": 1.836151864710494,
      "grad_norm": 1.2826062440872192,
      "learning_rate": 9.663815789473684e-05,
      "loss": 0.4753,
      "step": 32790
    },
    {
      "epoch": 1.836207862022623,
      "grad_norm": 1.398698329925537,
      "learning_rate": 9.66378947368421e-05,
      "loss": 0.4009,
      "step": 32791
    },
    {
      "epoch": 1.836263859334752,
      "grad_norm": 1.1002782583236694,
      "learning_rate": 9.663763157894737e-05,
      "loss": 0.3559,
      "step": 32792
    },
    {
      "epoch": 1.836319856646881,
      "grad_norm": 1.801236629486084,
      "learning_rate": 9.663736842105263e-05,
      "loss": 0.4569,
      "step": 32793
    },
    {
      "epoch": 1.83637585395901,
      "grad_norm": 1.076566219329834,
      "learning_rate": 9.663710526315791e-05,
      "loss": 0.4302,
      "step": 32794
    },
    {
      "epoch": 1.836431851271139,
      "grad_norm": 1.3344539403915405,
      "learning_rate": 9.663684210526315e-05,
      "loss": 0.4285,
      "step": 32795
    },
    {
      "epoch": 1.836487848583268,
      "grad_norm": 1.2668132781982422,
      "learning_rate": 9.663657894736843e-05,
      "loss": 0.4649,
      "step": 32796
    },
    {
      "epoch": 1.836543845895397,
      "grad_norm": 43.08137512207031,
      "learning_rate": 9.663631578947369e-05,
      "loss": 0.4311,
      "step": 32797
    },
    {
      "epoch": 1.8365998432075261,
      "grad_norm": 1.227246642112732,
      "learning_rate": 9.663605263157896e-05,
      "loss": 0.3829,
      "step": 32798
    },
    {
      "epoch": 1.8366558405196551,
      "grad_norm": 1.2849862575531006,
      "learning_rate": 9.66357894736842e-05,
      "loss": 0.4545,
      "step": 32799
    },
    {
      "epoch": 1.8367118378317842,
      "grad_norm": 1.5289303064346313,
      "learning_rate": 9.663552631578948e-05,
      "loss": 0.5596,
      "step": 32800
    },
    {
      "epoch": 1.8367678351439132,
      "grad_norm": 1.6089023351669312,
      "learning_rate": 9.663526315789474e-05,
      "loss": 0.6024,
      "step": 32801
    },
    {
      "epoch": 1.8368238324560422,
      "grad_norm": 1.6643784046173096,
      "learning_rate": 9.663500000000001e-05,
      "loss": 0.5891,
      "step": 32802
    },
    {
      "epoch": 1.8368798297681712,
      "grad_norm": 1.2046257257461548,
      "learning_rate": 9.663473684210527e-05,
      "loss": 0.5249,
      "step": 32803
    },
    {
      "epoch": 1.8369358270803002,
      "grad_norm": 1.3768064975738525,
      "learning_rate": 9.663447368421053e-05,
      "loss": 0.6448,
      "step": 32804
    },
    {
      "epoch": 1.8369918243924293,
      "grad_norm": 1.4686816930770874,
      "learning_rate": 9.663421052631579e-05,
      "loss": 0.4023,
      "step": 32805
    },
    {
      "epoch": 1.8370478217045583,
      "grad_norm": 1.0558440685272217,
      "learning_rate": 9.663394736842106e-05,
      "loss": 0.3994,
      "step": 32806
    },
    {
      "epoch": 1.8371038190166873,
      "grad_norm": 1.4201335906982422,
      "learning_rate": 9.663368421052632e-05,
      "loss": 0.6545,
      "step": 32807
    },
    {
      "epoch": 1.8371598163288163,
      "grad_norm": 1.1541748046875,
      "learning_rate": 9.663342105263158e-05,
      "loss": 0.4763,
      "step": 32808
    },
    {
      "epoch": 1.8372158136409453,
      "grad_norm": 1.4335594177246094,
      "learning_rate": 9.663315789473684e-05,
      "loss": 0.4626,
      "step": 32809
    },
    {
      "epoch": 1.8372718109530743,
      "grad_norm": 1.263968825340271,
      "learning_rate": 9.66328947368421e-05,
      "loss": 0.4137,
      "step": 32810
    },
    {
      "epoch": 1.8373278082652034,
      "grad_norm": 1.3395044803619385,
      "learning_rate": 9.663263157894738e-05,
      "loss": 0.5898,
      "step": 32811
    },
    {
      "epoch": 1.8373838055773324,
      "grad_norm": 1.0702292919158936,
      "learning_rate": 9.663236842105264e-05,
      "loss": 0.3403,
      "step": 32812
    },
    {
      "epoch": 1.8374398028894614,
      "grad_norm": 1.3203871250152588,
      "learning_rate": 9.66321052631579e-05,
      "loss": 0.4055,
      "step": 32813
    },
    {
      "epoch": 1.8374958002015904,
      "grad_norm": 1.0626285076141357,
      "learning_rate": 9.663184210526316e-05,
      "loss": 0.4178,
      "step": 32814
    },
    {
      "epoch": 1.8375517975137194,
      "grad_norm": 1.1373602151870728,
      "learning_rate": 9.663157894736843e-05,
      "loss": 0.4189,
      "step": 32815
    },
    {
      "epoch": 1.8376077948258485,
      "grad_norm": 1.1357685327529907,
      "learning_rate": 9.663131578947369e-05,
      "loss": 0.3559,
      "step": 32816
    },
    {
      "epoch": 1.8376637921379775,
      "grad_norm": 1.2538461685180664,
      "learning_rate": 9.663105263157895e-05,
      "loss": 0.4359,
      "step": 32817
    },
    {
      "epoch": 1.8377197894501065,
      "grad_norm": 1.2521655559539795,
      "learning_rate": 9.663078947368421e-05,
      "loss": 0.4569,
      "step": 32818
    },
    {
      "epoch": 1.8377757867622355,
      "grad_norm": 1.338199496269226,
      "learning_rate": 9.663052631578948e-05,
      "loss": 0.454,
      "step": 32819
    },
    {
      "epoch": 1.8378317840743645,
      "grad_norm": 1.312105417251587,
      "learning_rate": 9.663026315789474e-05,
      "loss": 0.3525,
      "step": 32820
    },
    {
      "epoch": 1.8378877813864936,
      "grad_norm": 1.2985937595367432,
      "learning_rate": 9.663000000000002e-05,
      "loss": 0.4823,
      "step": 32821
    },
    {
      "epoch": 1.8379437786986226,
      "grad_norm": 1.1555944681167603,
      "learning_rate": 9.662973684210526e-05,
      "loss": 0.5203,
      "step": 32822
    },
    {
      "epoch": 1.8379997760107516,
      "grad_norm": 1.1490230560302734,
      "learning_rate": 9.662947368421053e-05,
      "loss": 0.4,
      "step": 32823
    },
    {
      "epoch": 1.8380557733228806,
      "grad_norm": 1.2443671226501465,
      "learning_rate": 9.66292105263158e-05,
      "loss": 0.4949,
      "step": 32824
    },
    {
      "epoch": 1.8381117706350096,
      "grad_norm": 1.3705300092697144,
      "learning_rate": 9.662894736842105e-05,
      "loss": 0.4172,
      "step": 32825
    },
    {
      "epoch": 1.8381677679471387,
      "grad_norm": 1.2172324657440186,
      "learning_rate": 9.662868421052633e-05,
      "loss": 0.485,
      "step": 32826
    },
    {
      "epoch": 1.8382237652592677,
      "grad_norm": 1.6878315210342407,
      "learning_rate": 9.662842105263157e-05,
      "loss": 0.4268,
      "step": 32827
    },
    {
      "epoch": 1.8382797625713967,
      "grad_norm": 1.7038004398345947,
      "learning_rate": 9.662815789473685e-05,
      "loss": 0.5283,
      "step": 32828
    },
    {
      "epoch": 1.8383357598835257,
      "grad_norm": 1.224631667137146,
      "learning_rate": 9.66278947368421e-05,
      "loss": 0.5011,
      "step": 32829
    },
    {
      "epoch": 1.8383917571956547,
      "grad_norm": 1.2574737071990967,
      "learning_rate": 9.662763157894738e-05,
      "loss": 0.4075,
      "step": 32830
    },
    {
      "epoch": 1.8384477545077837,
      "grad_norm": 1.220641016960144,
      "learning_rate": 9.662736842105264e-05,
      "loss": 0.483,
      "step": 32831
    },
    {
      "epoch": 1.8385037518199128,
      "grad_norm": 1.5085011720657349,
      "learning_rate": 9.66271052631579e-05,
      "loss": 0.491,
      "step": 32832
    },
    {
      "epoch": 1.8385597491320418,
      "grad_norm": 1.2420427799224854,
      "learning_rate": 9.662684210526316e-05,
      "loss": 0.4541,
      "step": 32833
    },
    {
      "epoch": 1.8386157464441708,
      "grad_norm": 1.1556251049041748,
      "learning_rate": 9.662657894736843e-05,
      "loss": 0.5116,
      "step": 32834
    },
    {
      "epoch": 1.8386717437562998,
      "grad_norm": 1.3470628261566162,
      "learning_rate": 9.662631578947369e-05,
      "loss": 0.4289,
      "step": 32835
    },
    {
      "epoch": 1.8387277410684288,
      "grad_norm": 1.3866633176803589,
      "learning_rate": 9.662605263157895e-05,
      "loss": 0.6079,
      "step": 32836
    },
    {
      "epoch": 1.8387837383805579,
      "grad_norm": 1.2391531467437744,
      "learning_rate": 9.662578947368421e-05,
      "loss": 0.3687,
      "step": 32837
    },
    {
      "epoch": 1.8388397356926869,
      "grad_norm": 1.0916754007339478,
      "learning_rate": 9.662552631578948e-05,
      "loss": 0.4057,
      "step": 32838
    },
    {
      "epoch": 1.838895733004816,
      "grad_norm": 1.2940764427185059,
      "learning_rate": 9.662526315789474e-05,
      "loss": 0.4639,
      "step": 32839
    },
    {
      "epoch": 1.838951730316945,
      "grad_norm": 1.2522588968276978,
      "learning_rate": 9.6625e-05,
      "loss": 0.6026,
      "step": 32840
    },
    {
      "epoch": 1.839007727629074,
      "grad_norm": 1.3938361406326294,
      "learning_rate": 9.662473684210526e-05,
      "loss": 0.5013,
      "step": 32841
    },
    {
      "epoch": 1.839063724941203,
      "grad_norm": 1.405850887298584,
      "learning_rate": 9.662447368421052e-05,
      "loss": 0.5068,
      "step": 32842
    },
    {
      "epoch": 1.839119722253332,
      "grad_norm": 1.6099187135696411,
      "learning_rate": 9.66242105263158e-05,
      "loss": 0.3717,
      "step": 32843
    },
    {
      "epoch": 1.839175719565461,
      "grad_norm": 1.1712437868118286,
      "learning_rate": 9.662394736842106e-05,
      "loss": 0.4962,
      "step": 32844
    },
    {
      "epoch": 1.83923171687759,
      "grad_norm": 1.2748817205429077,
      "learning_rate": 9.662368421052632e-05,
      "loss": 0.489,
      "step": 32845
    },
    {
      "epoch": 1.839287714189719,
      "grad_norm": 1.3727037906646729,
      "learning_rate": 9.662342105263158e-05,
      "loss": 0.4432,
      "step": 32846
    },
    {
      "epoch": 1.839343711501848,
      "grad_norm": 1.0529649257659912,
      "learning_rate": 9.662315789473685e-05,
      "loss": 0.4623,
      "step": 32847
    },
    {
      "epoch": 1.839399708813977,
      "grad_norm": 1.2337660789489746,
      "learning_rate": 9.662289473684211e-05,
      "loss": 0.3982,
      "step": 32848
    },
    {
      "epoch": 1.839455706126106,
      "grad_norm": 1.2151973247528076,
      "learning_rate": 9.662263157894738e-05,
      "loss": 0.3726,
      "step": 32849
    },
    {
      "epoch": 1.8395117034382351,
      "grad_norm": 1.1094932556152344,
      "learning_rate": 9.662236842105263e-05,
      "loss": 0.5162,
      "step": 32850
    },
    {
      "epoch": 1.8395677007503641,
      "grad_norm": 1.5417227745056152,
      "learning_rate": 9.66221052631579e-05,
      "loss": 0.5746,
      "step": 32851
    },
    {
      "epoch": 1.8396236980624932,
      "grad_norm": 1.2124770879745483,
      "learning_rate": 9.662184210526316e-05,
      "loss": 0.4133,
      "step": 32852
    },
    {
      "epoch": 1.8396796953746222,
      "grad_norm": 1.2000820636749268,
      "learning_rate": 9.662157894736843e-05,
      "loss": 0.428,
      "step": 32853
    },
    {
      "epoch": 1.8397356926867512,
      "grad_norm": 1.2374606132507324,
      "learning_rate": 9.662131578947368e-05,
      "loss": 0.5061,
      "step": 32854
    },
    {
      "epoch": 1.8397916899988802,
      "grad_norm": 1.285317063331604,
      "learning_rate": 9.662105263157895e-05,
      "loss": 0.534,
      "step": 32855
    },
    {
      "epoch": 1.8398476873110092,
      "grad_norm": 0.9246919751167297,
      "learning_rate": 9.662078947368421e-05,
      "loss": 0.3633,
      "step": 32856
    },
    {
      "epoch": 1.8399036846231382,
      "grad_norm": 1.323366641998291,
      "learning_rate": 9.662052631578949e-05,
      "loss": 0.5477,
      "step": 32857
    },
    {
      "epoch": 1.8399596819352673,
      "grad_norm": 1.421041488647461,
      "learning_rate": 9.662026315789475e-05,
      "loss": 0.4274,
      "step": 32858
    },
    {
      "epoch": 1.8400156792473963,
      "grad_norm": 1.307328462600708,
      "learning_rate": 9.661999999999999e-05,
      "loss": 0.3705,
      "step": 32859
    },
    {
      "epoch": 1.8400716765595253,
      "grad_norm": 1.5341895818710327,
      "learning_rate": 9.661973684210527e-05,
      "loss": 0.6223,
      "step": 32860
    },
    {
      "epoch": 1.8401276738716543,
      "grad_norm": 1.3287591934204102,
      "learning_rate": 9.661947368421053e-05,
      "loss": 0.4585,
      "step": 32861
    },
    {
      "epoch": 1.8401836711837833,
      "grad_norm": 1.434287190437317,
      "learning_rate": 9.66192105263158e-05,
      "loss": 0.4387,
      "step": 32862
    },
    {
      "epoch": 1.8402396684959124,
      "grad_norm": 1.4952448606491089,
      "learning_rate": 9.661894736842106e-05,
      "loss": 0.5614,
      "step": 32863
    },
    {
      "epoch": 1.8402956658080412,
      "grad_norm": 1.2806533575057983,
      "learning_rate": 9.661868421052632e-05,
      "loss": 0.4939,
      "step": 32864
    },
    {
      "epoch": 1.8403516631201702,
      "grad_norm": 2.0783755779266357,
      "learning_rate": 9.661842105263158e-05,
      "loss": 0.4356,
      "step": 32865
    },
    {
      "epoch": 1.8404076604322992,
      "grad_norm": 1.4120310544967651,
      "learning_rate": 9.661815789473685e-05,
      "loss": 0.5665,
      "step": 32866
    },
    {
      "epoch": 1.8404636577444282,
      "grad_norm": 1.0840588808059692,
      "learning_rate": 9.661789473684211e-05,
      "loss": 0.3623,
      "step": 32867
    },
    {
      "epoch": 1.8405196550565572,
      "grad_norm": 1.24738347530365,
      "learning_rate": 9.661763157894737e-05,
      "loss": 0.5108,
      "step": 32868
    },
    {
      "epoch": 1.8405756523686863,
      "grad_norm": 1.4922033548355103,
      "learning_rate": 9.661736842105263e-05,
      "loss": 0.4668,
      "step": 32869
    },
    {
      "epoch": 1.8406316496808153,
      "grad_norm": 1.4111509323120117,
      "learning_rate": 9.66171052631579e-05,
      "loss": 0.4468,
      "step": 32870
    },
    {
      "epoch": 1.8406876469929443,
      "grad_norm": 2.189815044403076,
      "learning_rate": 9.661684210526316e-05,
      "loss": 0.4086,
      "step": 32871
    },
    {
      "epoch": 1.8407436443050733,
      "grad_norm": 1.1416786909103394,
      "learning_rate": 9.661657894736842e-05,
      "loss": 0.4294,
      "step": 32872
    },
    {
      "epoch": 1.8407996416172023,
      "grad_norm": 1.2118918895721436,
      "learning_rate": 9.661631578947368e-05,
      "loss": 0.4384,
      "step": 32873
    },
    {
      "epoch": 1.8408556389293314,
      "grad_norm": 1.250840425491333,
      "learning_rate": 9.661605263157896e-05,
      "loss": 0.4667,
      "step": 32874
    },
    {
      "epoch": 1.8409116362414604,
      "grad_norm": 1.3781054019927979,
      "learning_rate": 9.661578947368422e-05,
      "loss": 0.5255,
      "step": 32875
    },
    {
      "epoch": 1.8409676335535894,
      "grad_norm": 1.709642767906189,
      "learning_rate": 9.661552631578948e-05,
      "loss": 0.7597,
      "step": 32876
    },
    {
      "epoch": 1.8410236308657184,
      "grad_norm": 1.3381803035736084,
      "learning_rate": 9.661526315789474e-05,
      "loss": 0.5986,
      "step": 32877
    },
    {
      "epoch": 1.8410796281778474,
      "grad_norm": 1.1220204830169678,
      "learning_rate": 9.6615e-05,
      "loss": 0.4215,
      "step": 32878
    },
    {
      "epoch": 1.8411356254899764,
      "grad_norm": 1.6010150909423828,
      "learning_rate": 9.661473684210527e-05,
      "loss": 0.4468,
      "step": 32879
    },
    {
      "epoch": 1.8411916228021055,
      "grad_norm": 1.2880184650421143,
      "learning_rate": 9.661447368421053e-05,
      "loss": 0.4617,
      "step": 32880
    },
    {
      "epoch": 1.8412476201142345,
      "grad_norm": 1.3595713376998901,
      "learning_rate": 9.66142105263158e-05,
      "loss": 0.5685,
      "step": 32881
    },
    {
      "epoch": 1.8413036174263635,
      "grad_norm": 1.3904837369918823,
      "learning_rate": 9.661394736842105e-05,
      "loss": 0.5028,
      "step": 32882
    },
    {
      "epoch": 1.8413596147384925,
      "grad_norm": 1.2443723678588867,
      "learning_rate": 9.661368421052632e-05,
      "loss": 0.3857,
      "step": 32883
    },
    {
      "epoch": 1.8414156120506215,
      "grad_norm": 1.3925625085830688,
      "learning_rate": 9.661342105263158e-05,
      "loss": 0.5692,
      "step": 32884
    },
    {
      "epoch": 1.8414716093627506,
      "grad_norm": 1.2612394094467163,
      "learning_rate": 9.661315789473685e-05,
      "loss": 0.4527,
      "step": 32885
    },
    {
      "epoch": 1.8415276066748796,
      "grad_norm": 1.4269003868103027,
      "learning_rate": 9.661289473684211e-05,
      "loss": 0.5568,
      "step": 32886
    },
    {
      "epoch": 1.8415836039870086,
      "grad_norm": 1.358361005783081,
      "learning_rate": 9.661263157894737e-05,
      "loss": 0.472,
      "step": 32887
    },
    {
      "epoch": 1.8416396012991376,
      "grad_norm": 1.2917922735214233,
      "learning_rate": 9.661236842105263e-05,
      "loss": 0.3534,
      "step": 32888
    },
    {
      "epoch": 1.8416955986112666,
      "grad_norm": 1.4576170444488525,
      "learning_rate": 9.661210526315791e-05,
      "loss": 0.5876,
      "step": 32889
    },
    {
      "epoch": 1.8417515959233957,
      "grad_norm": 1.164176344871521,
      "learning_rate": 9.661184210526317e-05,
      "loss": 0.4028,
      "step": 32890
    },
    {
      "epoch": 1.8418075932355247,
      "grad_norm": 1.0955661535263062,
      "learning_rate": 9.661157894736843e-05,
      "loss": 0.3932,
      "step": 32891
    },
    {
      "epoch": 1.8418635905476537,
      "grad_norm": 1.2213622331619263,
      "learning_rate": 9.661131578947369e-05,
      "loss": 0.3994,
      "step": 32892
    },
    {
      "epoch": 1.8419195878597827,
      "grad_norm": 1.3236523866653442,
      "learning_rate": 9.661105263157895e-05,
      "loss": 0.3853,
      "step": 32893
    },
    {
      "epoch": 1.8419755851719117,
      "grad_norm": 1.3061505556106567,
      "learning_rate": 9.661078947368422e-05,
      "loss": 0.5589,
      "step": 32894
    },
    {
      "epoch": 1.8420315824840408,
      "grad_norm": 1.585598111152649,
      "learning_rate": 9.661052631578948e-05,
      "loss": 0.4717,
      "step": 32895
    },
    {
      "epoch": 1.8420875797961698,
      "grad_norm": 1.441094994544983,
      "learning_rate": 9.661026315789474e-05,
      "loss": 0.5385,
      "step": 32896
    },
    {
      "epoch": 1.8421435771082988,
      "grad_norm": 1.3344049453735352,
      "learning_rate": 9.661e-05,
      "loss": 0.4717,
      "step": 32897
    },
    {
      "epoch": 1.8421995744204278,
      "grad_norm": 1.2032164335250854,
      "learning_rate": 9.660973684210527e-05,
      "loss": 0.336,
      "step": 32898
    },
    {
      "epoch": 1.8422555717325568,
      "grad_norm": 1.133961796760559,
      "learning_rate": 9.660947368421053e-05,
      "loss": 0.4464,
      "step": 32899
    },
    {
      "epoch": 1.8423115690446858,
      "grad_norm": 33.66267395019531,
      "learning_rate": 9.660921052631579e-05,
      "loss": 0.4082,
      "step": 32900
    },
    {
      "epoch": 1.8423675663568149,
      "grad_norm": 1.3558391332626343,
      "learning_rate": 9.660894736842105e-05,
      "loss": 0.4596,
      "step": 32901
    },
    {
      "epoch": 1.8424235636689439,
      "grad_norm": 1.2404605150222778,
      "learning_rate": 9.660868421052632e-05,
      "loss": 0.3912,
      "step": 32902
    },
    {
      "epoch": 1.842479560981073,
      "grad_norm": 1.6435612440109253,
      "learning_rate": 9.660842105263158e-05,
      "loss": 0.5293,
      "step": 32903
    },
    {
      "epoch": 1.842535558293202,
      "grad_norm": 1.2336878776550293,
      "learning_rate": 9.660815789473686e-05,
      "loss": 0.4404,
      "step": 32904
    },
    {
      "epoch": 1.842591555605331,
      "grad_norm": 1.19773268699646,
      "learning_rate": 9.66078947368421e-05,
      "loss": 0.4471,
      "step": 32905
    },
    {
      "epoch": 1.84264755291746,
      "grad_norm": 1.265150547027588,
      "learning_rate": 9.660763157894738e-05,
      "loss": 0.4126,
      "step": 32906
    },
    {
      "epoch": 1.842703550229589,
      "grad_norm": 1.2897471189498901,
      "learning_rate": 9.660736842105264e-05,
      "loss": 0.3866,
      "step": 32907
    },
    {
      "epoch": 1.842759547541718,
      "grad_norm": 1.2433923482894897,
      "learning_rate": 9.660710526315791e-05,
      "loss": 0.4181,
      "step": 32908
    },
    {
      "epoch": 1.842815544853847,
      "grad_norm": 1.3655455112457275,
      "learning_rate": 9.660684210526316e-05,
      "loss": 0.405,
      "step": 32909
    },
    {
      "epoch": 1.842871542165976,
      "grad_norm": 1.5096124410629272,
      "learning_rate": 9.660657894736842e-05,
      "loss": 0.3937,
      "step": 32910
    },
    {
      "epoch": 1.842927539478105,
      "grad_norm": 1.2894468307495117,
      "learning_rate": 9.660631578947369e-05,
      "loss": 0.3344,
      "step": 32911
    },
    {
      "epoch": 1.842983536790234,
      "grad_norm": 1.274783730506897,
      "learning_rate": 9.660605263157895e-05,
      "loss": 0.483,
      "step": 32912
    },
    {
      "epoch": 1.843039534102363,
      "grad_norm": 1.2660008668899536,
      "learning_rate": 9.660578947368422e-05,
      "loss": 0.429,
      "step": 32913
    },
    {
      "epoch": 1.8430955314144921,
      "grad_norm": 1.3275043964385986,
      "learning_rate": 9.660552631578947e-05,
      "loss": 0.3887,
      "step": 32914
    },
    {
      "epoch": 1.8431515287266211,
      "grad_norm": 1.3385419845581055,
      "learning_rate": 9.660526315789474e-05,
      "loss": 0.4337,
      "step": 32915
    },
    {
      "epoch": 1.8432075260387502,
      "grad_norm": 1.1293500661849976,
      "learning_rate": 9.6605e-05,
      "loss": 0.3122,
      "step": 32916
    },
    {
      "epoch": 1.8432635233508792,
      "grad_norm": 1.2302464246749878,
      "learning_rate": 9.660473684210527e-05,
      "loss": 0.4735,
      "step": 32917
    },
    {
      "epoch": 1.8433195206630082,
      "grad_norm": 1.0381650924682617,
      "learning_rate": 9.660447368421053e-05,
      "loss": 0.3316,
      "step": 32918
    },
    {
      "epoch": 1.8433755179751372,
      "grad_norm": 1.338551640510559,
      "learning_rate": 9.66042105263158e-05,
      "loss": 0.4724,
      "step": 32919
    },
    {
      "epoch": 1.8434315152872662,
      "grad_norm": 1.4086332321166992,
      "learning_rate": 9.660394736842105e-05,
      "loss": 0.3751,
      "step": 32920
    },
    {
      "epoch": 1.8434875125993953,
      "grad_norm": 2.265753746032715,
      "learning_rate": 9.660368421052633e-05,
      "loss": 0.5968,
      "step": 32921
    },
    {
      "epoch": 1.8435435099115243,
      "grad_norm": 1.1701689958572388,
      "learning_rate": 9.660342105263159e-05,
      "loss": 0.4198,
      "step": 32922
    },
    {
      "epoch": 1.8435995072236533,
      "grad_norm": 1.2380021810531616,
      "learning_rate": 9.660315789473685e-05,
      "loss": 0.3859,
      "step": 32923
    },
    {
      "epoch": 1.8436555045357823,
      "grad_norm": 1.7832541465759277,
      "learning_rate": 9.66028947368421e-05,
      "loss": 0.4776,
      "step": 32924
    },
    {
      "epoch": 1.8437115018479113,
      "grad_norm": 1.1675121784210205,
      "learning_rate": 9.660263157894738e-05,
      "loss": 0.3027,
      "step": 32925
    },
    {
      "epoch": 1.8437674991600403,
      "grad_norm": 1.0559121370315552,
      "learning_rate": 9.660236842105264e-05,
      "loss": 0.3486,
      "step": 32926
    },
    {
      "epoch": 1.8438234964721694,
      "grad_norm": 1.1947718858718872,
      "learning_rate": 9.66021052631579e-05,
      "loss": 0.384,
      "step": 32927
    },
    {
      "epoch": 1.8438794937842984,
      "grad_norm": 1.0962809324264526,
      "learning_rate": 9.660184210526316e-05,
      "loss": 0.4411,
      "step": 32928
    },
    {
      "epoch": 1.8439354910964274,
      "grad_norm": 1.8183112144470215,
      "learning_rate": 9.660157894736842e-05,
      "loss": 0.5891,
      "step": 32929
    },
    {
      "epoch": 1.8439914884085564,
      "grad_norm": 1.551383137702942,
      "learning_rate": 9.660131578947369e-05,
      "loss": 0.5877,
      "step": 32930
    },
    {
      "epoch": 1.8440474857206854,
      "grad_norm": 1.1563749313354492,
      "learning_rate": 9.660105263157895e-05,
      "loss": 0.4114,
      "step": 32931
    },
    {
      "epoch": 1.8441034830328145,
      "grad_norm": 1.4932472705841064,
      "learning_rate": 9.660078947368421e-05,
      "loss": 0.6754,
      "step": 32932
    },
    {
      "epoch": 1.8441594803449435,
      "grad_norm": 1.3673840761184692,
      "learning_rate": 9.660052631578947e-05,
      "loss": 0.539,
      "step": 32933
    },
    {
      "epoch": 1.8442154776570725,
      "grad_norm": 0.9999203085899353,
      "learning_rate": 9.660026315789474e-05,
      "loss": 0.4372,
      "step": 32934
    },
    {
      "epoch": 1.8442714749692015,
      "grad_norm": 1.6258403062820435,
      "learning_rate": 9.66e-05,
      "loss": 0.4996,
      "step": 32935
    },
    {
      "epoch": 1.8443274722813305,
      "grad_norm": 1.0690977573394775,
      "learning_rate": 9.659973684210528e-05,
      "loss": 0.3502,
      "step": 32936
    },
    {
      "epoch": 1.8443834695934596,
      "grad_norm": 1.2587487697601318,
      "learning_rate": 9.659947368421052e-05,
      "loss": 0.504,
      "step": 32937
    },
    {
      "epoch": 1.8444394669055886,
      "grad_norm": 1.3194084167480469,
      "learning_rate": 9.65992105263158e-05,
      "loss": 0.5542,
      "step": 32938
    },
    {
      "epoch": 1.8444954642177176,
      "grad_norm": 1.546574354171753,
      "learning_rate": 9.659894736842106e-05,
      "loss": 0.506,
      "step": 32939
    },
    {
      "epoch": 1.8445514615298466,
      "grad_norm": 1.2619967460632324,
      "learning_rate": 9.659868421052633e-05,
      "loss": 0.4009,
      "step": 32940
    },
    {
      "epoch": 1.8446074588419756,
      "grad_norm": 1.1226036548614502,
      "learning_rate": 9.659842105263159e-05,
      "loss": 0.3289,
      "step": 32941
    },
    {
      "epoch": 1.8446634561541047,
      "grad_norm": 1.6209882497787476,
      "learning_rate": 9.659815789473685e-05,
      "loss": 0.4991,
      "step": 32942
    },
    {
      "epoch": 1.8447194534662337,
      "grad_norm": 1.3435505628585815,
      "learning_rate": 9.659789473684211e-05,
      "loss": 0.4163,
      "step": 32943
    },
    {
      "epoch": 1.8447754507783627,
      "grad_norm": 1.4991600513458252,
      "learning_rate": 9.659763157894738e-05,
      "loss": 0.4847,
      "step": 32944
    },
    {
      "epoch": 1.8448314480904917,
      "grad_norm": 1.529256820678711,
      "learning_rate": 9.659736842105264e-05,
      "loss": 0.4851,
      "step": 32945
    },
    {
      "epoch": 1.8448874454026207,
      "grad_norm": 1.960693597793579,
      "learning_rate": 9.659710526315789e-05,
      "loss": 0.5274,
      "step": 32946
    },
    {
      "epoch": 1.8449434427147495,
      "grad_norm": 1.2479568719863892,
      "learning_rate": 9.659684210526316e-05,
      "loss": 0.3408,
      "step": 32947
    },
    {
      "epoch": 1.8449994400268785,
      "grad_norm": 1.3500275611877441,
      "learning_rate": 9.659657894736842e-05,
      "loss": 0.385,
      "step": 32948
    },
    {
      "epoch": 1.8450554373390076,
      "grad_norm": 1.42543625831604,
      "learning_rate": 9.65963157894737e-05,
      "loss": 0.4586,
      "step": 32949
    },
    {
      "epoch": 1.8451114346511366,
      "grad_norm": 1.253745198249817,
      "learning_rate": 9.659605263157895e-05,
      "loss": 0.4494,
      "step": 32950
    },
    {
      "epoch": 1.8451674319632656,
      "grad_norm": 1.1142747402191162,
      "learning_rate": 9.659578947368421e-05,
      "loss": 0.3687,
      "step": 32951
    },
    {
      "epoch": 1.8452234292753946,
      "grad_norm": 1.284039855003357,
      "learning_rate": 9.659552631578947e-05,
      "loss": 0.4253,
      "step": 32952
    },
    {
      "epoch": 1.8452794265875236,
      "grad_norm": 1.646706223487854,
      "learning_rate": 9.659526315789475e-05,
      "loss": 0.4558,
      "step": 32953
    },
    {
      "epoch": 1.8453354238996527,
      "grad_norm": 1.3232616186141968,
      "learning_rate": 9.6595e-05,
      "loss": 0.4442,
      "step": 32954
    },
    {
      "epoch": 1.8453914212117817,
      "grad_norm": 1.055118441581726,
      "learning_rate": 9.659473684210527e-05,
      "loss": 0.3636,
      "step": 32955
    },
    {
      "epoch": 1.8454474185239107,
      "grad_norm": 1.055234670639038,
      "learning_rate": 9.659447368421053e-05,
      "loss": 0.383,
      "step": 32956
    },
    {
      "epoch": 1.8455034158360397,
      "grad_norm": 1.223559856414795,
      "learning_rate": 9.65942105263158e-05,
      "loss": 0.4492,
      "step": 32957
    },
    {
      "epoch": 1.8455594131481687,
      "grad_norm": 1.0541549921035767,
      "learning_rate": 9.659394736842106e-05,
      "loss": 0.4253,
      "step": 32958
    },
    {
      "epoch": 1.8456154104602978,
      "grad_norm": 1.2929306030273438,
      "learning_rate": 9.659368421052633e-05,
      "loss": 0.4008,
      "step": 32959
    },
    {
      "epoch": 1.8456714077724268,
      "grad_norm": 1.321595311164856,
      "learning_rate": 9.659342105263158e-05,
      "loss": 0.5088,
      "step": 32960
    },
    {
      "epoch": 1.8457274050845558,
      "grad_norm": 1.2284960746765137,
      "learning_rate": 9.659315789473685e-05,
      "loss": 0.3958,
      "step": 32961
    },
    {
      "epoch": 1.8457834023966848,
      "grad_norm": 1.2104394435882568,
      "learning_rate": 9.659289473684211e-05,
      "loss": 0.4704,
      "step": 32962
    },
    {
      "epoch": 1.8458393997088138,
      "grad_norm": 1.3578954935073853,
      "learning_rate": 9.659263157894737e-05,
      "loss": 0.411,
      "step": 32963
    },
    {
      "epoch": 1.8458953970209429,
      "grad_norm": 1.5557917356491089,
      "learning_rate": 9.659236842105263e-05,
      "loss": 0.3553,
      "step": 32964
    },
    {
      "epoch": 1.8459513943330719,
      "grad_norm": 1.177769660949707,
      "learning_rate": 9.659210526315789e-05,
      "loss": 0.3891,
      "step": 32965
    },
    {
      "epoch": 1.846007391645201,
      "grad_norm": 1.1529127359390259,
      "learning_rate": 9.659184210526316e-05,
      "loss": 0.5096,
      "step": 32966
    },
    {
      "epoch": 1.84606338895733,
      "grad_norm": 1.1034342050552368,
      "learning_rate": 9.659157894736842e-05,
      "loss": 0.3543,
      "step": 32967
    },
    {
      "epoch": 1.846119386269459,
      "grad_norm": 1.303011178970337,
      "learning_rate": 9.65913157894737e-05,
      "loss": 0.5574,
      "step": 32968
    },
    {
      "epoch": 1.846175383581588,
      "grad_norm": 1.259000539779663,
      "learning_rate": 9.659105263157894e-05,
      "loss": 0.5287,
      "step": 32969
    },
    {
      "epoch": 1.846231380893717,
      "grad_norm": 1.2804316282272339,
      "learning_rate": 9.659078947368422e-05,
      "loss": 0.4154,
      "step": 32970
    },
    {
      "epoch": 1.846287378205846,
      "grad_norm": 1.7210476398468018,
      "learning_rate": 9.659052631578948e-05,
      "loss": 0.4942,
      "step": 32971
    },
    {
      "epoch": 1.846343375517975,
      "grad_norm": 1.3382210731506348,
      "learning_rate": 9.659026315789475e-05,
      "loss": 0.603,
      "step": 32972
    },
    {
      "epoch": 1.846399372830104,
      "grad_norm": 1.4095916748046875,
      "learning_rate": 9.659000000000001e-05,
      "loss": 0.5391,
      "step": 32973
    },
    {
      "epoch": 1.846455370142233,
      "grad_norm": 1.493009090423584,
      "learning_rate": 9.658973684210527e-05,
      "loss": 0.5056,
      "step": 32974
    },
    {
      "epoch": 1.846511367454362,
      "grad_norm": 1.3601634502410889,
      "learning_rate": 9.658947368421053e-05,
      "loss": 0.5551,
      "step": 32975
    },
    {
      "epoch": 1.846567364766491,
      "grad_norm": 1.2282062768936157,
      "learning_rate": 9.65892105263158e-05,
      "loss": 0.4257,
      "step": 32976
    },
    {
      "epoch": 1.84662336207862,
      "grad_norm": 1.2335641384124756,
      "learning_rate": 9.658894736842106e-05,
      "loss": 0.4919,
      "step": 32977
    },
    {
      "epoch": 1.8466793593907491,
      "grad_norm": 1.1923980712890625,
      "learning_rate": 9.658868421052632e-05,
      "loss": 0.4678,
      "step": 32978
    },
    {
      "epoch": 1.8467353567028781,
      "grad_norm": 1.159430742263794,
      "learning_rate": 9.658842105263158e-05,
      "loss": 0.4693,
      "step": 32979
    },
    {
      "epoch": 1.8467913540150072,
      "grad_norm": 2.119084119796753,
      "learning_rate": 9.658815789473684e-05,
      "loss": 0.5571,
      "step": 32980
    },
    {
      "epoch": 1.8468473513271362,
      "grad_norm": 1.2369295358657837,
      "learning_rate": 9.658789473684211e-05,
      "loss": 0.5934,
      "step": 32981
    },
    {
      "epoch": 1.8469033486392652,
      "grad_norm": 1.211553692817688,
      "learning_rate": 9.658763157894737e-05,
      "loss": 0.4896,
      "step": 32982
    },
    {
      "epoch": 1.8469593459513942,
      "grad_norm": 1.5806540250778198,
      "learning_rate": 9.658736842105263e-05,
      "loss": 0.5435,
      "step": 32983
    },
    {
      "epoch": 1.8470153432635232,
      "grad_norm": 1.3230743408203125,
      "learning_rate": 9.658710526315789e-05,
      "loss": 0.507,
      "step": 32984
    },
    {
      "epoch": 1.8470713405756523,
      "grad_norm": 1.4513412714004517,
      "learning_rate": 9.658684210526317e-05,
      "loss": 0.4663,
      "step": 32985
    },
    {
      "epoch": 1.8471273378877813,
      "grad_norm": 1.4557348489761353,
      "learning_rate": 9.658657894736843e-05,
      "loss": 0.3542,
      "step": 32986
    },
    {
      "epoch": 1.8471833351999103,
      "grad_norm": 1.183685302734375,
      "learning_rate": 9.658631578947369e-05,
      "loss": 0.4862,
      "step": 32987
    },
    {
      "epoch": 1.8472393325120393,
      "grad_norm": 1.3003326654434204,
      "learning_rate": 9.658605263157895e-05,
      "loss": 0.5666,
      "step": 32988
    },
    {
      "epoch": 1.8472953298241683,
      "grad_norm": 1.2486693859100342,
      "learning_rate": 9.658578947368422e-05,
      "loss": 0.451,
      "step": 32989
    },
    {
      "epoch": 1.8473513271362973,
      "grad_norm": 1.376805305480957,
      "learning_rate": 9.658552631578948e-05,
      "loss": 0.5394,
      "step": 32990
    },
    {
      "epoch": 1.8474073244484264,
      "grad_norm": 1.3795126676559448,
      "learning_rate": 9.658526315789475e-05,
      "loss": 0.4197,
      "step": 32991
    },
    {
      "epoch": 1.8474633217605554,
      "grad_norm": 1.3613721132278442,
      "learning_rate": 9.6585e-05,
      "loss": 0.4942,
      "step": 32992
    },
    {
      "epoch": 1.8475193190726844,
      "grad_norm": 1.049103021621704,
      "learning_rate": 9.658473684210527e-05,
      "loss": 0.3842,
      "step": 32993
    },
    {
      "epoch": 1.8475753163848134,
      "grad_norm": 1.1341272592544556,
      "learning_rate": 9.658447368421053e-05,
      "loss": 0.4122,
      "step": 32994
    },
    {
      "epoch": 1.8476313136969424,
      "grad_norm": 4.548651218414307,
      "learning_rate": 9.65842105263158e-05,
      "loss": 0.3641,
      "step": 32995
    },
    {
      "epoch": 1.8476873110090715,
      "grad_norm": 1.4399057626724243,
      "learning_rate": 9.658394736842106e-05,
      "loss": 0.5778,
      "step": 32996
    },
    {
      "epoch": 1.8477433083212005,
      "grad_norm": 1.3400887250900269,
      "learning_rate": 9.658368421052631e-05,
      "loss": 0.3781,
      "step": 32997
    },
    {
      "epoch": 1.8477993056333295,
      "grad_norm": 1.4371615648269653,
      "learning_rate": 9.658342105263158e-05,
      "loss": 0.397,
      "step": 32998
    },
    {
      "epoch": 1.8478553029454585,
      "grad_norm": 1.1246460676193237,
      "learning_rate": 9.658315789473684e-05,
      "loss": 0.3377,
      "step": 32999
    },
    {
      "epoch": 1.8479113002575875,
      "grad_norm": 1.2970534563064575,
      "learning_rate": 9.658289473684212e-05,
      "loss": 0.4319,
      "step": 33000
    },
    {
      "epoch": 1.8479672975697166,
      "grad_norm": 1.1445130109786987,
      "learning_rate": 9.658263157894736e-05,
      "loss": 0.4152,
      "step": 33001
    },
    {
      "epoch": 1.8480232948818456,
      "grad_norm": 1.2730820178985596,
      "learning_rate": 9.658236842105264e-05,
      "loss": 0.498,
      "step": 33002
    },
    {
      "epoch": 1.8480792921939746,
      "grad_norm": 1.1291395425796509,
      "learning_rate": 9.65821052631579e-05,
      "loss": 0.3832,
      "step": 33003
    },
    {
      "epoch": 1.8481352895061036,
      "grad_norm": 1.567795753479004,
      "learning_rate": 9.658184210526317e-05,
      "loss": 0.4268,
      "step": 33004
    },
    {
      "epoch": 1.8481912868182326,
      "grad_norm": 1.1752840280532837,
      "learning_rate": 9.658157894736843e-05,
      "loss": 0.501,
      "step": 33005
    },
    {
      "epoch": 1.8482472841303617,
      "grad_norm": 1.0759519338607788,
      "learning_rate": 9.658131578947369e-05,
      "loss": 0.3402,
      "step": 33006
    },
    {
      "epoch": 1.8483032814424907,
      "grad_norm": 1.3437083959579468,
      "learning_rate": 9.658105263157895e-05,
      "loss": 0.4244,
      "step": 33007
    },
    {
      "epoch": 1.8483592787546197,
      "grad_norm": 1.3966819047927856,
      "learning_rate": 9.658078947368422e-05,
      "loss": 0.5113,
      "step": 33008
    },
    {
      "epoch": 1.8484152760667487,
      "grad_norm": 7.844576835632324,
      "learning_rate": 9.658052631578948e-05,
      "loss": 0.5597,
      "step": 33009
    },
    {
      "epoch": 1.8484712733788777,
      "grad_norm": 1.2463321685791016,
      "learning_rate": 9.658026315789474e-05,
      "loss": 0.3794,
      "step": 33010
    },
    {
      "epoch": 1.8485272706910068,
      "grad_norm": 1.2907909154891968,
      "learning_rate": 9.658e-05,
      "loss": 0.4675,
      "step": 33011
    },
    {
      "epoch": 1.8485832680031358,
      "grad_norm": 1.0820895433425903,
      "learning_rate": 9.657973684210527e-05,
      "loss": 0.4019,
      "step": 33012
    },
    {
      "epoch": 1.8486392653152648,
      "grad_norm": 1.0687875747680664,
      "learning_rate": 9.657947368421053e-05,
      "loss": 0.3793,
      "step": 33013
    },
    {
      "epoch": 1.8486952626273938,
      "grad_norm": 1.2128746509552002,
      "learning_rate": 9.657921052631579e-05,
      "loss": 0.4496,
      "step": 33014
    },
    {
      "epoch": 1.8487512599395228,
      "grad_norm": 1.3045095205307007,
      "learning_rate": 9.657894736842105e-05,
      "loss": 0.5148,
      "step": 33015
    },
    {
      "epoch": 1.8488072572516518,
      "grad_norm": 1.2198787927627563,
      "learning_rate": 9.657868421052631e-05,
      "loss": 0.444,
      "step": 33016
    },
    {
      "epoch": 1.8488632545637809,
      "grad_norm": 1.0958393812179565,
      "learning_rate": 9.657842105263159e-05,
      "loss": 0.4635,
      "step": 33017
    },
    {
      "epoch": 1.8489192518759099,
      "grad_norm": 1.2258250713348389,
      "learning_rate": 9.657815789473685e-05,
      "loss": 0.4792,
      "step": 33018
    },
    {
      "epoch": 1.848975249188039,
      "grad_norm": 1.2343659400939941,
      "learning_rate": 9.65778947368421e-05,
      "loss": 0.5383,
      "step": 33019
    },
    {
      "epoch": 1.849031246500168,
      "grad_norm": 1.6550458669662476,
      "learning_rate": 9.657763157894736e-05,
      "loss": 0.4201,
      "step": 33020
    },
    {
      "epoch": 1.849087243812297,
      "grad_norm": 1.3469651937484741,
      "learning_rate": 9.657736842105264e-05,
      "loss": 0.3987,
      "step": 33021
    },
    {
      "epoch": 1.849143241124426,
      "grad_norm": 1.2976679801940918,
      "learning_rate": 9.65771052631579e-05,
      "loss": 0.4981,
      "step": 33022
    },
    {
      "epoch": 1.849199238436555,
      "grad_norm": 1.2716412544250488,
      "learning_rate": 9.657684210526317e-05,
      "loss": 0.4696,
      "step": 33023
    },
    {
      "epoch": 1.849255235748684,
      "grad_norm": 1.000976324081421,
      "learning_rate": 9.657657894736842e-05,
      "loss": 0.3443,
      "step": 33024
    },
    {
      "epoch": 1.849311233060813,
      "grad_norm": 1.4338499307632446,
      "learning_rate": 9.657631578947369e-05,
      "loss": 0.4904,
      "step": 33025
    },
    {
      "epoch": 1.849367230372942,
      "grad_norm": 1.2978743314743042,
      "learning_rate": 9.657605263157895e-05,
      "loss": 0.6887,
      "step": 33026
    },
    {
      "epoch": 1.849423227685071,
      "grad_norm": 1.2637507915496826,
      "learning_rate": 9.657578947368422e-05,
      "loss": 0.4435,
      "step": 33027
    },
    {
      "epoch": 1.8494792249972,
      "grad_norm": 1.7632912397384644,
      "learning_rate": 9.657552631578948e-05,
      "loss": 0.4266,
      "step": 33028
    },
    {
      "epoch": 1.849535222309329,
      "grad_norm": 1.2291784286499023,
      "learning_rate": 9.657526315789474e-05,
      "loss": 0.4696,
      "step": 33029
    },
    {
      "epoch": 1.8495912196214581,
      "grad_norm": 1.2759946584701538,
      "learning_rate": 9.6575e-05,
      "loss": 0.445,
      "step": 33030
    },
    {
      "epoch": 1.8496472169335871,
      "grad_norm": 1.1573654413223267,
      "learning_rate": 9.657473684210526e-05,
      "loss": 0.4667,
      "step": 33031
    },
    {
      "epoch": 1.8497032142457162,
      "grad_norm": 1.4651817083358765,
      "learning_rate": 9.657447368421054e-05,
      "loss": 0.4681,
      "step": 33032
    },
    {
      "epoch": 1.8497592115578452,
      "grad_norm": 1.3305976390838623,
      "learning_rate": 9.65742105263158e-05,
      "loss": 0.6046,
      "step": 33033
    },
    {
      "epoch": 1.8498152088699742,
      "grad_norm": 1.2923963069915771,
      "learning_rate": 9.657394736842106e-05,
      "loss": 0.7058,
      "step": 33034
    },
    {
      "epoch": 1.8498712061821032,
      "grad_norm": 1.3880723714828491,
      "learning_rate": 9.657368421052632e-05,
      "loss": 0.5578,
      "step": 33035
    },
    {
      "epoch": 1.8499272034942322,
      "grad_norm": 1.334620714187622,
      "learning_rate": 9.657342105263159e-05,
      "loss": 0.4341,
      "step": 33036
    },
    {
      "epoch": 1.8499832008063612,
      "grad_norm": 1.2041887044906616,
      "learning_rate": 9.657315789473685e-05,
      "loss": 0.3855,
      "step": 33037
    },
    {
      "epoch": 1.8500391981184903,
      "grad_norm": 1.3424389362335205,
      "learning_rate": 9.657289473684211e-05,
      "loss": 0.4479,
      "step": 33038
    },
    {
      "epoch": 1.8500951954306193,
      "grad_norm": 1.2787641286849976,
      "learning_rate": 9.657263157894737e-05,
      "loss": 0.4199,
      "step": 33039
    },
    {
      "epoch": 1.8501511927427483,
      "grad_norm": 1.672153353691101,
      "learning_rate": 9.657236842105264e-05,
      "loss": 0.6242,
      "step": 33040
    },
    {
      "epoch": 1.8502071900548773,
      "grad_norm": 1.437294840812683,
      "learning_rate": 9.65721052631579e-05,
      "loss": 0.5694,
      "step": 33041
    },
    {
      "epoch": 1.8502631873670063,
      "grad_norm": 1.0827196836471558,
      "learning_rate": 9.657184210526316e-05,
      "loss": 0.3911,
      "step": 33042
    },
    {
      "epoch": 1.8503191846791354,
      "grad_norm": 1.386171817779541,
      "learning_rate": 9.657157894736842e-05,
      "loss": 0.6313,
      "step": 33043
    },
    {
      "epoch": 1.8503751819912644,
      "grad_norm": 1.3338085412979126,
      "learning_rate": 9.65713157894737e-05,
      "loss": 0.4517,
      "step": 33044
    },
    {
      "epoch": 1.8504311793033934,
      "grad_norm": 1.337959885597229,
      "learning_rate": 9.657105263157895e-05,
      "loss": 0.3323,
      "step": 33045
    },
    {
      "epoch": 1.8504871766155224,
      "grad_norm": 1.3833059072494507,
      "learning_rate": 9.657078947368423e-05,
      "loss": 0.4781,
      "step": 33046
    },
    {
      "epoch": 1.8505431739276514,
      "grad_norm": 1.4422019720077515,
      "learning_rate": 9.657052631578947e-05,
      "loss": 0.5218,
      "step": 33047
    },
    {
      "epoch": 1.8505991712397805,
      "grad_norm": 1.213630199432373,
      "learning_rate": 9.657026315789475e-05,
      "loss": 0.3653,
      "step": 33048
    },
    {
      "epoch": 1.8506551685519095,
      "grad_norm": 1.1104912757873535,
      "learning_rate": 9.657e-05,
      "loss": 0.4329,
      "step": 33049
    },
    {
      "epoch": 1.8507111658640385,
      "grad_norm": 1.2492437362670898,
      "learning_rate": 9.656973684210527e-05,
      "loss": 0.3078,
      "step": 33050
    },
    {
      "epoch": 1.8507671631761675,
      "grad_norm": 1.1975221633911133,
      "learning_rate": 9.656947368421054e-05,
      "loss": 0.4118,
      "step": 33051
    },
    {
      "epoch": 1.8508231604882965,
      "grad_norm": 1.2960602045059204,
      "learning_rate": 9.656921052631578e-05,
      "loss": 0.4938,
      "step": 33052
    },
    {
      "epoch": 1.8508791578004256,
      "grad_norm": 3.541465997695923,
      "learning_rate": 9.656894736842106e-05,
      "loss": 0.5232,
      "step": 33053
    },
    {
      "epoch": 1.8509351551125546,
      "grad_norm": 1.584956169128418,
      "learning_rate": 9.656868421052632e-05,
      "loss": 0.6109,
      "step": 33054
    },
    {
      "epoch": 1.8509911524246836,
      "grad_norm": 1.1987066268920898,
      "learning_rate": 9.656842105263159e-05,
      "loss": 0.4364,
      "step": 33055
    },
    {
      "epoch": 1.8510471497368126,
      "grad_norm": 1.2029398679733276,
      "learning_rate": 9.656815789473684e-05,
      "loss": 0.4021,
      "step": 33056
    },
    {
      "epoch": 1.8511031470489416,
      "grad_norm": 1.3097554445266724,
      "learning_rate": 9.656789473684211e-05,
      "loss": 0.4785,
      "step": 33057
    },
    {
      "epoch": 1.8511591443610707,
      "grad_norm": 4.961433410644531,
      "learning_rate": 9.656763157894737e-05,
      "loss": 0.4511,
      "step": 33058
    },
    {
      "epoch": 1.8512151416731997,
      "grad_norm": 1.3062509298324585,
      "learning_rate": 9.656736842105264e-05,
      "loss": 0.5214,
      "step": 33059
    },
    {
      "epoch": 1.8512711389853287,
      "grad_norm": 0.9713279604911804,
      "learning_rate": 9.65671052631579e-05,
      "loss": 0.2653,
      "step": 33060
    },
    {
      "epoch": 1.8513271362974577,
      "grad_norm": 1.042035460472107,
      "learning_rate": 9.656684210526316e-05,
      "loss": 0.4177,
      "step": 33061
    },
    {
      "epoch": 1.8513831336095867,
      "grad_norm": 1.2576196193695068,
      "learning_rate": 9.656657894736842e-05,
      "loss": 0.4957,
      "step": 33062
    },
    {
      "epoch": 1.8514391309217157,
      "grad_norm": 1.2618991136550903,
      "learning_rate": 9.65663157894737e-05,
      "loss": 0.4992,
      "step": 33063
    },
    {
      "epoch": 1.8514951282338448,
      "grad_norm": 1.0386995077133179,
      "learning_rate": 9.656605263157896e-05,
      "loss": 0.3648,
      "step": 33064
    },
    {
      "epoch": 1.8515511255459738,
      "grad_norm": 1.4311909675598145,
      "learning_rate": 9.656578947368422e-05,
      "loss": 0.4981,
      "step": 33065
    },
    {
      "epoch": 1.8516071228581028,
      "grad_norm": 1.0725550651550293,
      "learning_rate": 9.656552631578948e-05,
      "loss": 0.425,
      "step": 33066
    },
    {
      "epoch": 1.8516631201702318,
      "grad_norm": 1.2727642059326172,
      "learning_rate": 9.656526315789473e-05,
      "loss": 0.464,
      "step": 33067
    },
    {
      "epoch": 1.8517191174823608,
      "grad_norm": 1.134121060371399,
      "learning_rate": 9.656500000000001e-05,
      "loss": 0.3897,
      "step": 33068
    },
    {
      "epoch": 1.8517751147944899,
      "grad_norm": 1.1708431243896484,
      "learning_rate": 9.656473684210527e-05,
      "loss": 0.4454,
      "step": 33069
    },
    {
      "epoch": 1.8518311121066189,
      "grad_norm": 1.1136401891708374,
      "learning_rate": 9.656447368421053e-05,
      "loss": 0.3273,
      "step": 33070
    },
    {
      "epoch": 1.851887109418748,
      "grad_norm": 1.343088150024414,
      "learning_rate": 9.656421052631579e-05,
      "loss": 0.4572,
      "step": 33071
    },
    {
      "epoch": 1.851943106730877,
      "grad_norm": 1.1251026391983032,
      "learning_rate": 9.656394736842106e-05,
      "loss": 0.3703,
      "step": 33072
    },
    {
      "epoch": 1.851999104043006,
      "grad_norm": 1.3409059047698975,
      "learning_rate": 9.656368421052632e-05,
      "loss": 0.5208,
      "step": 33073
    },
    {
      "epoch": 1.852055101355135,
      "grad_norm": 1.2337473630905151,
      "learning_rate": 9.656342105263158e-05,
      "loss": 0.4543,
      "step": 33074
    },
    {
      "epoch": 1.852111098667264,
      "grad_norm": 1.4777143001556396,
      "learning_rate": 9.656315789473684e-05,
      "loss": 0.6273,
      "step": 33075
    },
    {
      "epoch": 1.852167095979393,
      "grad_norm": 3.0945606231689453,
      "learning_rate": 9.656289473684211e-05,
      "loss": 0.4552,
      "step": 33076
    },
    {
      "epoch": 1.852223093291522,
      "grad_norm": 1.309761643409729,
      "learning_rate": 9.656263157894737e-05,
      "loss": 0.5515,
      "step": 33077
    },
    {
      "epoch": 1.852279090603651,
      "grad_norm": 1.0365557670593262,
      "learning_rate": 9.656236842105265e-05,
      "loss": 0.4608,
      "step": 33078
    },
    {
      "epoch": 1.85233508791578,
      "grad_norm": 1.2894647121429443,
      "learning_rate": 9.656210526315789e-05,
      "loss": 0.5498,
      "step": 33079
    },
    {
      "epoch": 1.852391085227909,
      "grad_norm": 1.4224568605422974,
      "learning_rate": 9.656184210526317e-05,
      "loss": 0.4155,
      "step": 33080
    },
    {
      "epoch": 1.852447082540038,
      "grad_norm": 1.1105976104736328,
      "learning_rate": 9.656157894736843e-05,
      "loss": 0.38,
      "step": 33081
    },
    {
      "epoch": 1.8525030798521671,
      "grad_norm": 1.318962574005127,
      "learning_rate": 9.65613157894737e-05,
      "loss": 0.5226,
      "step": 33082
    },
    {
      "epoch": 1.8525590771642961,
      "grad_norm": 1.0393366813659668,
      "learning_rate": 9.656105263157896e-05,
      "loss": 0.3868,
      "step": 33083
    },
    {
      "epoch": 1.8526150744764251,
      "grad_norm": 1.1849257946014404,
      "learning_rate": 9.65607894736842e-05,
      "loss": 0.5016,
      "step": 33084
    },
    {
      "epoch": 1.8526710717885542,
      "grad_norm": 1.0333153009414673,
      "learning_rate": 9.656052631578948e-05,
      "loss": 0.4353,
      "step": 33085
    },
    {
      "epoch": 1.8527270691006832,
      "grad_norm": 1.3058979511260986,
      "learning_rate": 9.656026315789474e-05,
      "loss": 0.4755,
      "step": 33086
    },
    {
      "epoch": 1.8527830664128122,
      "grad_norm": 1.299809217453003,
      "learning_rate": 9.656000000000001e-05,
      "loss": 0.4371,
      "step": 33087
    },
    {
      "epoch": 1.8528390637249412,
      "grad_norm": 1.0798301696777344,
      "learning_rate": 9.655973684210527e-05,
      "loss": 0.3373,
      "step": 33088
    },
    {
      "epoch": 1.8528950610370702,
      "grad_norm": 1.1431992053985596,
      "learning_rate": 9.655947368421053e-05,
      "loss": 0.3428,
      "step": 33089
    },
    {
      "epoch": 1.8529510583491993,
      "grad_norm": 1.3724161386489868,
      "learning_rate": 9.655921052631579e-05,
      "loss": 0.4823,
      "step": 33090
    },
    {
      "epoch": 1.8530070556613283,
      "grad_norm": 1.2289235591888428,
      "learning_rate": 9.655894736842106e-05,
      "loss": 0.4292,
      "step": 33091
    },
    {
      "epoch": 1.8530630529734573,
      "grad_norm": 1.367658257484436,
      "learning_rate": 9.655868421052632e-05,
      "loss": 0.5845,
      "step": 33092
    },
    {
      "epoch": 1.8531190502855863,
      "grad_norm": 1.4230901002883911,
      "learning_rate": 9.655842105263158e-05,
      "loss": 0.5229,
      "step": 33093
    },
    {
      "epoch": 1.8531750475977153,
      "grad_norm": 1.417458415031433,
      "learning_rate": 9.655815789473684e-05,
      "loss": 0.556,
      "step": 33094
    },
    {
      "epoch": 1.8532310449098444,
      "grad_norm": 1.4302271604537964,
      "learning_rate": 9.655789473684212e-05,
      "loss": 0.6765,
      "step": 33095
    },
    {
      "epoch": 1.8532870422219734,
      "grad_norm": 1.3495972156524658,
      "learning_rate": 9.655763157894738e-05,
      "loss": 0.5772,
      "step": 33096
    },
    {
      "epoch": 1.8533430395341024,
      "grad_norm": 1.2109516859054565,
      "learning_rate": 9.655736842105264e-05,
      "loss": 0.5467,
      "step": 33097
    },
    {
      "epoch": 1.8533990368462314,
      "grad_norm": 1.260568618774414,
      "learning_rate": 9.65571052631579e-05,
      "loss": 0.309,
      "step": 33098
    },
    {
      "epoch": 1.8534550341583604,
      "grad_norm": 1.3763041496276855,
      "learning_rate": 9.655684210526317e-05,
      "loss": 0.4497,
      "step": 33099
    },
    {
      "epoch": 1.8535110314704895,
      "grad_norm": 1.0809141397476196,
      "learning_rate": 9.655657894736843e-05,
      "loss": 0.3706,
      "step": 33100
    },
    {
      "epoch": 1.8535670287826185,
      "grad_norm": 1.1887532472610474,
      "learning_rate": 9.655631578947369e-05,
      "loss": 0.3976,
      "step": 33101
    },
    {
      "epoch": 1.8536230260947475,
      "grad_norm": 1.2558282613754272,
      "learning_rate": 9.655605263157895e-05,
      "loss": 0.4437,
      "step": 33102
    },
    {
      "epoch": 1.8536790234068765,
      "grad_norm": 1.4799175262451172,
      "learning_rate": 9.655578947368421e-05,
      "loss": 0.4076,
      "step": 33103
    },
    {
      "epoch": 1.8537350207190055,
      "grad_norm": 1.3490275144577026,
      "learning_rate": 9.655552631578948e-05,
      "loss": 0.3874,
      "step": 33104
    },
    {
      "epoch": 1.8537910180311346,
      "grad_norm": 1.267166018486023,
      "learning_rate": 9.655526315789474e-05,
      "loss": 0.4538,
      "step": 33105
    },
    {
      "epoch": 1.8538470153432636,
      "grad_norm": 2.162872314453125,
      "learning_rate": 9.655500000000001e-05,
      "loss": 0.4441,
      "step": 33106
    },
    {
      "epoch": 1.8539030126553926,
      "grad_norm": 1.268487572669983,
      "learning_rate": 9.655473684210526e-05,
      "loss": 0.607,
      "step": 33107
    },
    {
      "epoch": 1.8539590099675216,
      "grad_norm": 1.4336848258972168,
      "learning_rate": 9.655447368421053e-05,
      "loss": 0.493,
      "step": 33108
    },
    {
      "epoch": 1.8540150072796506,
      "grad_norm": 1.1673438549041748,
      "learning_rate": 9.655421052631579e-05,
      "loss": 0.4071,
      "step": 33109
    },
    {
      "epoch": 1.8540710045917796,
      "grad_norm": 1.1715099811553955,
      "learning_rate": 9.655394736842107e-05,
      "loss": 0.6009,
      "step": 33110
    },
    {
      "epoch": 1.8541270019039087,
      "grad_norm": 1.4830306768417358,
      "learning_rate": 9.655368421052631e-05,
      "loss": 0.4813,
      "step": 33111
    },
    {
      "epoch": 1.8541829992160377,
      "grad_norm": 1.2398089170455933,
      "learning_rate": 9.655342105263159e-05,
      "loss": 0.4535,
      "step": 33112
    },
    {
      "epoch": 1.8542389965281667,
      "grad_norm": 1.3025530576705933,
      "learning_rate": 9.655315789473684e-05,
      "loss": 0.5483,
      "step": 33113
    },
    {
      "epoch": 1.8542949938402957,
      "grad_norm": 1.1998889446258545,
      "learning_rate": 9.655289473684212e-05,
      "loss": 0.3307,
      "step": 33114
    },
    {
      "epoch": 1.8543509911524247,
      "grad_norm": 1.0280098915100098,
      "learning_rate": 9.655263157894738e-05,
      "loss": 0.4335,
      "step": 33115
    },
    {
      "epoch": 1.8544069884645538,
      "grad_norm": 1.661843180656433,
      "learning_rate": 9.655236842105264e-05,
      "loss": 0.5408,
      "step": 33116
    },
    {
      "epoch": 1.8544629857766828,
      "grad_norm": 1.2318053245544434,
      "learning_rate": 9.65521052631579e-05,
      "loss": 0.4169,
      "step": 33117
    },
    {
      "epoch": 1.8545189830888118,
      "grad_norm": 1.513507604598999,
      "learning_rate": 9.655184210526316e-05,
      "loss": 0.7214,
      "step": 33118
    },
    {
      "epoch": 1.8545749804009408,
      "grad_norm": 1.455310344696045,
      "learning_rate": 9.655157894736843e-05,
      "loss": 0.5072,
      "step": 33119
    },
    {
      "epoch": 1.8546309777130698,
      "grad_norm": 1.4041128158569336,
      "learning_rate": 9.655131578947369e-05,
      "loss": 0.5939,
      "step": 33120
    },
    {
      "epoch": 1.8546869750251989,
      "grad_norm": 1.0689117908477783,
      "learning_rate": 9.655105263157895e-05,
      "loss": 0.3959,
      "step": 33121
    },
    {
      "epoch": 1.8547429723373279,
      "grad_norm": 4.283168315887451,
      "learning_rate": 9.655078947368421e-05,
      "loss": 0.4627,
      "step": 33122
    },
    {
      "epoch": 1.854798969649457,
      "grad_norm": 1.3845758438110352,
      "learning_rate": 9.655052631578948e-05,
      "loss": 0.533,
      "step": 33123
    },
    {
      "epoch": 1.854854966961586,
      "grad_norm": 1.3015806674957275,
      "learning_rate": 9.655026315789474e-05,
      "loss": 0.4367,
      "step": 33124
    },
    {
      "epoch": 1.854910964273715,
      "grad_norm": 1.154928207397461,
      "learning_rate": 9.655e-05,
      "loss": 0.4463,
      "step": 33125
    },
    {
      "epoch": 1.854966961585844,
      "grad_norm": 1.3839025497436523,
      "learning_rate": 9.654973684210526e-05,
      "loss": 0.5426,
      "step": 33126
    },
    {
      "epoch": 1.855022958897973,
      "grad_norm": 1.185211181640625,
      "learning_rate": 9.654947368421054e-05,
      "loss": 0.3559,
      "step": 33127
    },
    {
      "epoch": 1.855078956210102,
      "grad_norm": 1.0509347915649414,
      "learning_rate": 9.65492105263158e-05,
      "loss": 0.426,
      "step": 33128
    },
    {
      "epoch": 1.855134953522231,
      "grad_norm": 1.454164743423462,
      "learning_rate": 9.654894736842105e-05,
      "loss": 0.3985,
      "step": 33129
    },
    {
      "epoch": 1.85519095083436,
      "grad_norm": 1.1675007343292236,
      "learning_rate": 9.654868421052631e-05,
      "loss": 0.3877,
      "step": 33130
    },
    {
      "epoch": 1.855246948146489,
      "grad_norm": 1.4339954853057861,
      "learning_rate": 9.654842105263159e-05,
      "loss": 0.5345,
      "step": 33131
    },
    {
      "epoch": 1.855302945458618,
      "grad_norm": 1.2303017377853394,
      "learning_rate": 9.654815789473685e-05,
      "loss": 0.4575,
      "step": 33132
    },
    {
      "epoch": 1.855358942770747,
      "grad_norm": 1.3257334232330322,
      "learning_rate": 9.654789473684212e-05,
      "loss": 0.4836,
      "step": 33133
    },
    {
      "epoch": 1.855414940082876,
      "grad_norm": 1.131144642829895,
      "learning_rate": 9.654763157894737e-05,
      "loss": 0.3676,
      "step": 33134
    },
    {
      "epoch": 1.8554709373950051,
      "grad_norm": 2.5463688373565674,
      "learning_rate": 9.654736842105263e-05,
      "loss": 0.4261,
      "step": 33135
    },
    {
      "epoch": 1.8555269347071341,
      "grad_norm": 1.3660928010940552,
      "learning_rate": 9.65471052631579e-05,
      "loss": 0.4662,
      "step": 33136
    },
    {
      "epoch": 1.8555829320192632,
      "grad_norm": 1.2900288105010986,
      "learning_rate": 9.654684210526316e-05,
      "loss": 0.4106,
      "step": 33137
    },
    {
      "epoch": 1.8556389293313922,
      "grad_norm": 1.391201138496399,
      "learning_rate": 9.654657894736843e-05,
      "loss": 0.382,
      "step": 33138
    },
    {
      "epoch": 1.8556949266435212,
      "grad_norm": 1.3388479948043823,
      "learning_rate": 9.654631578947368e-05,
      "loss": 0.5254,
      "step": 33139
    },
    {
      "epoch": 1.8557509239556502,
      "grad_norm": 1.160375952720642,
      "learning_rate": 9.654605263157895e-05,
      "loss": 0.3951,
      "step": 33140
    },
    {
      "epoch": 1.8558069212677792,
      "grad_norm": 1.3148711919784546,
      "learning_rate": 9.654578947368421e-05,
      "loss": 0.4271,
      "step": 33141
    },
    {
      "epoch": 1.8558629185799083,
      "grad_norm": 1.4214383363723755,
      "learning_rate": 9.654552631578949e-05,
      "loss": 0.5326,
      "step": 33142
    },
    {
      "epoch": 1.8559189158920373,
      "grad_norm": 1.1944375038146973,
      "learning_rate": 9.654526315789475e-05,
      "loss": 0.5269,
      "step": 33143
    },
    {
      "epoch": 1.8559749132041663,
      "grad_norm": 1.4981383085250854,
      "learning_rate": 9.6545e-05,
      "loss": 0.4033,
      "step": 33144
    },
    {
      "epoch": 1.8560309105162953,
      "grad_norm": 1.5162910223007202,
      "learning_rate": 9.654473684210526e-05,
      "loss": 0.4073,
      "step": 33145
    },
    {
      "epoch": 1.8560869078284243,
      "grad_norm": 1.4173718690872192,
      "learning_rate": 9.654447368421054e-05,
      "loss": 0.5271,
      "step": 33146
    },
    {
      "epoch": 1.8561429051405534,
      "grad_norm": 1.2792532444000244,
      "learning_rate": 9.65442105263158e-05,
      "loss": 0.4023,
      "step": 33147
    },
    {
      "epoch": 1.8561989024526824,
      "grad_norm": 1.521446943283081,
      "learning_rate": 9.654394736842106e-05,
      "loss": 0.5062,
      "step": 33148
    },
    {
      "epoch": 1.8562548997648114,
      "grad_norm": 1.3226897716522217,
      "learning_rate": 9.654368421052632e-05,
      "loss": 0.3883,
      "step": 33149
    },
    {
      "epoch": 1.8563108970769404,
      "grad_norm": 1.6951515674591064,
      "learning_rate": 9.654342105263159e-05,
      "loss": 0.5469,
      "step": 33150
    },
    {
      "epoch": 1.8563668943890694,
      "grad_norm": 1.3238255977630615,
      "learning_rate": 9.654315789473685e-05,
      "loss": 0.4881,
      "step": 33151
    },
    {
      "epoch": 1.8564228917011985,
      "grad_norm": 1.434852957725525,
      "learning_rate": 9.654289473684211e-05,
      "loss": 0.5531,
      "step": 33152
    },
    {
      "epoch": 1.8564788890133275,
      "grad_norm": 1.1392138004302979,
      "learning_rate": 9.654263157894737e-05,
      "loss": 0.4223,
      "step": 33153
    },
    {
      "epoch": 1.8565348863254565,
      "grad_norm": 1.3729078769683838,
      "learning_rate": 9.654236842105263e-05,
      "loss": 0.4292,
      "step": 33154
    },
    {
      "epoch": 1.8565908836375855,
      "grad_norm": 1.61236572265625,
      "learning_rate": 9.65421052631579e-05,
      "loss": 0.5432,
      "step": 33155
    },
    {
      "epoch": 1.8566468809497145,
      "grad_norm": 1.3132075071334839,
      "learning_rate": 9.654184210526316e-05,
      "loss": 0.4583,
      "step": 33156
    },
    {
      "epoch": 1.8567028782618435,
      "grad_norm": 1.4339098930358887,
      "learning_rate": 9.654157894736842e-05,
      "loss": 0.4041,
      "step": 33157
    },
    {
      "epoch": 1.8567588755739726,
      "grad_norm": 1.225520372390747,
      "learning_rate": 9.654131578947368e-05,
      "loss": 0.3977,
      "step": 33158
    },
    {
      "epoch": 1.8568148728861016,
      "grad_norm": 1.1398911476135254,
      "learning_rate": 9.654105263157895e-05,
      "loss": 0.3641,
      "step": 33159
    },
    {
      "epoch": 1.8568708701982306,
      "grad_norm": 1.2526735067367554,
      "learning_rate": 9.654078947368421e-05,
      "loss": 0.4355,
      "step": 33160
    },
    {
      "epoch": 1.8569268675103596,
      "grad_norm": 1.4194422960281372,
      "learning_rate": 9.654052631578949e-05,
      "loss": 0.4494,
      "step": 33161
    },
    {
      "epoch": 1.8569828648224886,
      "grad_norm": 1.4049956798553467,
      "learning_rate": 9.654026315789473e-05,
      "loss": 0.4442,
      "step": 33162
    },
    {
      "epoch": 1.8570388621346177,
      "grad_norm": 1.2219620943069458,
      "learning_rate": 9.654000000000001e-05,
      "loss": 0.3741,
      "step": 33163
    },
    {
      "epoch": 1.8570948594467467,
      "grad_norm": 1.4150627851486206,
      "learning_rate": 9.653973684210527e-05,
      "loss": 0.3954,
      "step": 33164
    },
    {
      "epoch": 1.8571508567588757,
      "grad_norm": 1.2717299461364746,
      "learning_rate": 9.653947368421054e-05,
      "loss": 0.4213,
      "step": 33165
    },
    {
      "epoch": 1.8572068540710047,
      "grad_norm": 1.4136308431625366,
      "learning_rate": 9.653921052631579e-05,
      "loss": 0.4058,
      "step": 33166
    },
    {
      "epoch": 1.8572628513831337,
      "grad_norm": 1.196578860282898,
      "learning_rate": 9.653894736842106e-05,
      "loss": 0.4078,
      "step": 33167
    },
    {
      "epoch": 1.8573188486952628,
      "grad_norm": 1.098273515701294,
      "learning_rate": 9.653868421052632e-05,
      "loss": 0.5009,
      "step": 33168
    },
    {
      "epoch": 1.8573748460073918,
      "grad_norm": 1.2358180284500122,
      "learning_rate": 9.653842105263159e-05,
      "loss": 0.5651,
      "step": 33169
    },
    {
      "epoch": 1.8574308433195208,
      "grad_norm": 1.186604380607605,
      "learning_rate": 9.653815789473685e-05,
      "loss": 0.4409,
      "step": 33170
    },
    {
      "epoch": 1.8574868406316498,
      "grad_norm": 1.2703826427459717,
      "learning_rate": 9.65378947368421e-05,
      "loss": 0.5416,
      "step": 33171
    },
    {
      "epoch": 1.8575428379437788,
      "grad_norm": 1.4217480421066284,
      "learning_rate": 9.653763157894737e-05,
      "loss": 0.491,
      "step": 33172
    },
    {
      "epoch": 1.8575988352559079,
      "grad_norm": 1.5214414596557617,
      "learning_rate": 9.653736842105263e-05,
      "loss": 0.4847,
      "step": 33173
    },
    {
      "epoch": 1.8576548325680369,
      "grad_norm": 1.191049575805664,
      "learning_rate": 9.65371052631579e-05,
      "loss": 0.3773,
      "step": 33174
    },
    {
      "epoch": 1.857710829880166,
      "grad_norm": 1.1918625831604004,
      "learning_rate": 9.653684210526316e-05,
      "loss": 0.5194,
      "step": 33175
    },
    {
      "epoch": 1.857766827192295,
      "grad_norm": 1.0788966417312622,
      "learning_rate": 9.653657894736842e-05,
      "loss": 0.3965,
      "step": 33176
    },
    {
      "epoch": 1.857822824504424,
      "grad_norm": 1.2535103559494019,
      "learning_rate": 9.653631578947368e-05,
      "loss": 0.4623,
      "step": 33177
    },
    {
      "epoch": 1.857878821816553,
      "grad_norm": 1.4373008012771606,
      "learning_rate": 9.653605263157896e-05,
      "loss": 0.5303,
      "step": 33178
    },
    {
      "epoch": 1.857934819128682,
      "grad_norm": 1.4039499759674072,
      "learning_rate": 9.653578947368422e-05,
      "loss": 0.3667,
      "step": 33179
    },
    {
      "epoch": 1.857990816440811,
      "grad_norm": 1.1685330867767334,
      "learning_rate": 9.653552631578948e-05,
      "loss": 0.5415,
      "step": 33180
    },
    {
      "epoch": 1.85804681375294,
      "grad_norm": 1.0712639093399048,
      "learning_rate": 9.653526315789474e-05,
      "loss": 0.3915,
      "step": 33181
    },
    {
      "epoch": 1.858102811065069,
      "grad_norm": 1.3238636255264282,
      "learning_rate": 9.653500000000001e-05,
      "loss": 0.4885,
      "step": 33182
    },
    {
      "epoch": 1.858158808377198,
      "grad_norm": 1.2522668838500977,
      "learning_rate": 9.653473684210527e-05,
      "loss": 0.5375,
      "step": 33183
    },
    {
      "epoch": 1.858214805689327,
      "grad_norm": 1.290726661682129,
      "learning_rate": 9.653447368421053e-05,
      "loss": 0.5262,
      "step": 33184
    },
    {
      "epoch": 1.858270803001456,
      "grad_norm": 1.303221583366394,
      "learning_rate": 9.653421052631579e-05,
      "loss": 0.4288,
      "step": 33185
    },
    {
      "epoch": 1.858326800313585,
      "grad_norm": 1.161790132522583,
      "learning_rate": 9.653394736842106e-05,
      "loss": 0.3933,
      "step": 33186
    },
    {
      "epoch": 1.8583827976257141,
      "grad_norm": 1.4410653114318848,
      "learning_rate": 9.653368421052632e-05,
      "loss": 0.4534,
      "step": 33187
    },
    {
      "epoch": 1.8584387949378431,
      "grad_norm": 1.4781135320663452,
      "learning_rate": 9.653342105263158e-05,
      "loss": 0.7443,
      "step": 33188
    },
    {
      "epoch": 1.8584947922499722,
      "grad_norm": 1.5559567213058472,
      "learning_rate": 9.653315789473684e-05,
      "loss": 0.6003,
      "step": 33189
    },
    {
      "epoch": 1.8585507895621012,
      "grad_norm": 1.182310938835144,
      "learning_rate": 9.65328947368421e-05,
      "loss": 0.5594,
      "step": 33190
    },
    {
      "epoch": 1.8586067868742302,
      "grad_norm": 1.2010245323181152,
      "learning_rate": 9.653263157894737e-05,
      "loss": 0.5012,
      "step": 33191
    },
    {
      "epoch": 1.8586627841863592,
      "grad_norm": 1.1277631521224976,
      "learning_rate": 9.653236842105263e-05,
      "loss": 0.3654,
      "step": 33192
    },
    {
      "epoch": 1.8587187814984882,
      "grad_norm": 1.3679944276809692,
      "learning_rate": 9.653210526315791e-05,
      "loss": 0.4288,
      "step": 33193
    },
    {
      "epoch": 1.8587747788106173,
      "grad_norm": 1.3539291620254517,
      "learning_rate": 9.653184210526315e-05,
      "loss": 0.4613,
      "step": 33194
    },
    {
      "epoch": 1.858830776122746,
      "grad_norm": 1.3745754957199097,
      "learning_rate": 9.653157894736843e-05,
      "loss": 0.6353,
      "step": 33195
    },
    {
      "epoch": 1.858886773434875,
      "grad_norm": 1.0956120491027832,
      "learning_rate": 9.653131578947369e-05,
      "loss": 0.344,
      "step": 33196
    },
    {
      "epoch": 1.858942770747004,
      "grad_norm": 1.1513943672180176,
      "learning_rate": 9.653105263157896e-05,
      "loss": 0.4839,
      "step": 33197
    },
    {
      "epoch": 1.858998768059133,
      "grad_norm": 1.3860986232757568,
      "learning_rate": 9.653078947368422e-05,
      "loss": 0.4771,
      "step": 33198
    },
    {
      "epoch": 1.8590547653712621,
      "grad_norm": 1.429257869720459,
      "learning_rate": 9.653052631578948e-05,
      "loss": 0.5066,
      "step": 33199
    },
    {
      "epoch": 1.8591107626833911,
      "grad_norm": 1.1928224563598633,
      "learning_rate": 9.653026315789474e-05,
      "loss": 0.3391,
      "step": 33200
    },
    {
      "epoch": 1.8591667599955202,
      "grad_norm": 1.1016384363174438,
      "learning_rate": 9.653000000000001e-05,
      "loss": 0.3275,
      "step": 33201
    },
    {
      "epoch": 1.8592227573076492,
      "grad_norm": 1.5377134084701538,
      "learning_rate": 9.652973684210527e-05,
      "loss": 0.3946,
      "step": 33202
    },
    {
      "epoch": 1.8592787546197782,
      "grad_norm": 1.344175100326538,
      "learning_rate": 9.652947368421053e-05,
      "loss": 0.3915,
      "step": 33203
    },
    {
      "epoch": 1.8593347519319072,
      "grad_norm": 1.3472548723220825,
      "learning_rate": 9.652921052631579e-05,
      "loss": 0.4253,
      "step": 33204
    },
    {
      "epoch": 1.8593907492440362,
      "grad_norm": 1.1115021705627441,
      "learning_rate": 9.652894736842105e-05,
      "loss": 0.3779,
      "step": 33205
    },
    {
      "epoch": 1.8594467465561653,
      "grad_norm": 1.5202127695083618,
      "learning_rate": 9.652868421052632e-05,
      "loss": 0.5438,
      "step": 33206
    },
    {
      "epoch": 1.8595027438682943,
      "grad_norm": 1.1402655839920044,
      "learning_rate": 9.652842105263158e-05,
      "loss": 0.4357,
      "step": 33207
    },
    {
      "epoch": 1.8595587411804233,
      "grad_norm": 1.2354209423065186,
      "learning_rate": 9.652815789473684e-05,
      "loss": 0.5218,
      "step": 33208
    },
    {
      "epoch": 1.8596147384925523,
      "grad_norm": 0.9355956315994263,
      "learning_rate": 9.65278947368421e-05,
      "loss": 0.3268,
      "step": 33209
    },
    {
      "epoch": 1.8596707358046813,
      "grad_norm": 1.586888313293457,
      "learning_rate": 9.652763157894738e-05,
      "loss": 0.552,
      "step": 33210
    },
    {
      "epoch": 1.8597267331168104,
      "grad_norm": 1.3255527019500732,
      "learning_rate": 9.652736842105264e-05,
      "loss": 0.4361,
      "step": 33211
    },
    {
      "epoch": 1.8597827304289394,
      "grad_norm": 1.2221790552139282,
      "learning_rate": 9.65271052631579e-05,
      "loss": 0.3618,
      "step": 33212
    },
    {
      "epoch": 1.8598387277410684,
      "grad_norm": 1.3526649475097656,
      "learning_rate": 9.652684210526316e-05,
      "loss": 0.5011,
      "step": 33213
    },
    {
      "epoch": 1.8598947250531974,
      "grad_norm": 1.154591679573059,
      "learning_rate": 9.652657894736843e-05,
      "loss": 0.3266,
      "step": 33214
    },
    {
      "epoch": 1.8599507223653264,
      "grad_norm": 1.2744574546813965,
      "learning_rate": 9.652631578947369e-05,
      "loss": 0.401,
      "step": 33215
    },
    {
      "epoch": 1.8600067196774555,
      "grad_norm": 1.0954231023788452,
      "learning_rate": 9.652605263157895e-05,
      "loss": 0.3587,
      "step": 33216
    },
    {
      "epoch": 1.8600627169895845,
      "grad_norm": 1.2176766395568848,
      "learning_rate": 9.652578947368421e-05,
      "loss": 0.3546,
      "step": 33217
    },
    {
      "epoch": 1.8601187143017135,
      "grad_norm": 1.1566481590270996,
      "learning_rate": 9.652552631578948e-05,
      "loss": 0.5207,
      "step": 33218
    },
    {
      "epoch": 1.8601747116138425,
      "grad_norm": 1.3136396408081055,
      "learning_rate": 9.652526315789474e-05,
      "loss": 0.5068,
      "step": 33219
    },
    {
      "epoch": 1.8602307089259715,
      "grad_norm": 1.2912509441375732,
      "learning_rate": 9.652500000000002e-05,
      "loss": 0.3516,
      "step": 33220
    },
    {
      "epoch": 1.8602867062381006,
      "grad_norm": 1.2112452983856201,
      "learning_rate": 9.652473684210526e-05,
      "loss": 0.3653,
      "step": 33221
    },
    {
      "epoch": 1.8603427035502296,
      "grad_norm": 1.2557883262634277,
      "learning_rate": 9.652447368421052e-05,
      "loss": 0.4438,
      "step": 33222
    },
    {
      "epoch": 1.8603987008623586,
      "grad_norm": 1.3698614835739136,
      "learning_rate": 9.65242105263158e-05,
      "loss": 0.481,
      "step": 33223
    },
    {
      "epoch": 1.8604546981744876,
      "grad_norm": 2.157306671142578,
      "learning_rate": 9.652394736842105e-05,
      "loss": 0.4831,
      "step": 33224
    },
    {
      "epoch": 1.8605106954866166,
      "grad_norm": 1.425485610961914,
      "learning_rate": 9.652368421052633e-05,
      "loss": 0.3729,
      "step": 33225
    },
    {
      "epoch": 1.8605666927987456,
      "grad_norm": 1.3635287284851074,
      "learning_rate": 9.652342105263157e-05,
      "loss": 0.5146,
      "step": 33226
    },
    {
      "epoch": 1.8606226901108747,
      "grad_norm": 1.1912692785263062,
      "learning_rate": 9.652315789473685e-05,
      "loss": 0.3423,
      "step": 33227
    },
    {
      "epoch": 1.8606786874230037,
      "grad_norm": 1.2394254207611084,
      "learning_rate": 9.65228947368421e-05,
      "loss": 0.474,
      "step": 33228
    },
    {
      "epoch": 1.8607346847351327,
      "grad_norm": 1.2628141641616821,
      "learning_rate": 9.652263157894738e-05,
      "loss": 0.4587,
      "step": 33229
    },
    {
      "epoch": 1.8607906820472617,
      "grad_norm": 1.4132494926452637,
      "learning_rate": 9.652236842105264e-05,
      "loss": 0.5303,
      "step": 33230
    },
    {
      "epoch": 1.8608466793593907,
      "grad_norm": 1.8759733438491821,
      "learning_rate": 9.65221052631579e-05,
      "loss": 0.4537,
      "step": 33231
    },
    {
      "epoch": 1.8609026766715198,
      "grad_norm": 1.3873984813690186,
      "learning_rate": 9.652184210526316e-05,
      "loss": 0.5849,
      "step": 33232
    },
    {
      "epoch": 1.8609586739836488,
      "grad_norm": 1.183600902557373,
      "learning_rate": 9.652157894736843e-05,
      "loss": 0.4262,
      "step": 33233
    },
    {
      "epoch": 1.8610146712957778,
      "grad_norm": 1.477653980255127,
      "learning_rate": 9.652131578947369e-05,
      "loss": 0.4572,
      "step": 33234
    },
    {
      "epoch": 1.8610706686079068,
      "grad_norm": 1.4078872203826904,
      "learning_rate": 9.652105263157895e-05,
      "loss": 0.5162,
      "step": 33235
    },
    {
      "epoch": 1.8611266659200358,
      "grad_norm": 1.0280321836471558,
      "learning_rate": 9.652078947368421e-05,
      "loss": 0.3673,
      "step": 33236
    },
    {
      "epoch": 1.8611826632321649,
      "grad_norm": 1.45988929271698,
      "learning_rate": 9.652052631578948e-05,
      "loss": 0.495,
      "step": 33237
    },
    {
      "epoch": 1.8612386605442939,
      "grad_norm": 1.3883593082427979,
      "learning_rate": 9.652026315789474e-05,
      "loss": 0.4717,
      "step": 33238
    },
    {
      "epoch": 1.861294657856423,
      "grad_norm": 1.2350587844848633,
      "learning_rate": 9.652e-05,
      "loss": 0.4214,
      "step": 33239
    },
    {
      "epoch": 1.861350655168552,
      "grad_norm": 1.4167972803115845,
      "learning_rate": 9.651973684210526e-05,
      "loss": 0.5187,
      "step": 33240
    },
    {
      "epoch": 1.861406652480681,
      "grad_norm": 1.1706527471542358,
      "learning_rate": 9.651947368421052e-05,
      "loss": 0.3587,
      "step": 33241
    },
    {
      "epoch": 1.86146264979281,
      "grad_norm": 1.1903153657913208,
      "learning_rate": 9.65192105263158e-05,
      "loss": 0.4871,
      "step": 33242
    },
    {
      "epoch": 1.861518647104939,
      "grad_norm": 1.4145921468734741,
      "learning_rate": 9.651894736842106e-05,
      "loss": 0.5622,
      "step": 33243
    },
    {
      "epoch": 1.861574644417068,
      "grad_norm": 1.3552058935165405,
      "learning_rate": 9.651868421052632e-05,
      "loss": 0.4162,
      "step": 33244
    },
    {
      "epoch": 1.861630641729197,
      "grad_norm": 1.501378059387207,
      "learning_rate": 9.651842105263158e-05,
      "loss": 0.5254,
      "step": 33245
    },
    {
      "epoch": 1.861686639041326,
      "grad_norm": 33.333866119384766,
      "learning_rate": 9.651815789473685e-05,
      "loss": 0.4166,
      "step": 33246
    },
    {
      "epoch": 1.861742636353455,
      "grad_norm": 1.1257805824279785,
      "learning_rate": 9.651789473684211e-05,
      "loss": 0.4534,
      "step": 33247
    },
    {
      "epoch": 1.861798633665584,
      "grad_norm": 1.1850035190582275,
      "learning_rate": 9.651763157894738e-05,
      "loss": 0.3522,
      "step": 33248
    },
    {
      "epoch": 1.861854630977713,
      "grad_norm": 1.1929658651351929,
      "learning_rate": 9.651736842105263e-05,
      "loss": 0.3706,
      "step": 33249
    },
    {
      "epoch": 1.861910628289842,
      "grad_norm": 1.1902594566345215,
      "learning_rate": 9.65171052631579e-05,
      "loss": 0.35,
      "step": 33250
    },
    {
      "epoch": 1.8619666256019711,
      "grad_norm": 1.2255616188049316,
      "learning_rate": 9.651684210526316e-05,
      "loss": 0.4522,
      "step": 33251
    },
    {
      "epoch": 1.8620226229141001,
      "grad_norm": 3.014446973800659,
      "learning_rate": 9.651657894736843e-05,
      "loss": 0.5083,
      "step": 33252
    },
    {
      "epoch": 1.8620786202262292,
      "grad_norm": 1.3079462051391602,
      "learning_rate": 9.65163157894737e-05,
      "loss": 0.4072,
      "step": 33253
    },
    {
      "epoch": 1.8621346175383582,
      "grad_norm": 1.3268287181854248,
      "learning_rate": 9.651605263157895e-05,
      "loss": 0.3646,
      "step": 33254
    },
    {
      "epoch": 1.8621906148504872,
      "grad_norm": 1.6336162090301514,
      "learning_rate": 9.651578947368421e-05,
      "loss": 0.4509,
      "step": 33255
    },
    {
      "epoch": 1.8622466121626162,
      "grad_norm": 1.2794172763824463,
      "learning_rate": 9.651552631578947e-05,
      "loss": 0.5047,
      "step": 33256
    },
    {
      "epoch": 1.8623026094747452,
      "grad_norm": 1.1605744361877441,
      "learning_rate": 9.651526315789475e-05,
      "loss": 0.4259,
      "step": 33257
    },
    {
      "epoch": 1.8623586067868743,
      "grad_norm": 1.38705313205719,
      "learning_rate": 9.6515e-05,
      "loss": 0.4562,
      "step": 33258
    },
    {
      "epoch": 1.8624146040990033,
      "grad_norm": 1.2819559574127197,
      "learning_rate": 9.651473684210527e-05,
      "loss": 0.422,
      "step": 33259
    },
    {
      "epoch": 1.8624706014111323,
      "grad_norm": 1.3177659511566162,
      "learning_rate": 9.651447368421053e-05,
      "loss": 0.4142,
      "step": 33260
    },
    {
      "epoch": 1.8625265987232613,
      "grad_norm": 1.313639760017395,
      "learning_rate": 9.65142105263158e-05,
      "loss": 0.4397,
      "step": 33261
    },
    {
      "epoch": 1.8625825960353903,
      "grad_norm": 1.1930993795394897,
      "learning_rate": 9.651394736842106e-05,
      "loss": 0.5392,
      "step": 33262
    },
    {
      "epoch": 1.8626385933475194,
      "grad_norm": 1.1592968702316284,
      "learning_rate": 9.651368421052632e-05,
      "loss": 0.3514,
      "step": 33263
    },
    {
      "epoch": 1.8626945906596484,
      "grad_norm": 1.1909130811691284,
      "learning_rate": 9.651342105263158e-05,
      "loss": 0.5344,
      "step": 33264
    },
    {
      "epoch": 1.8627505879717774,
      "grad_norm": 1.3716959953308105,
      "learning_rate": 9.651315789473685e-05,
      "loss": 0.4311,
      "step": 33265
    },
    {
      "epoch": 1.8628065852839064,
      "grad_norm": 1.1980502605438232,
      "learning_rate": 9.651289473684211e-05,
      "loss": 0.3682,
      "step": 33266
    },
    {
      "epoch": 1.8628625825960354,
      "grad_norm": 1.3581562042236328,
      "learning_rate": 9.651263157894737e-05,
      "loss": 0.3964,
      "step": 33267
    },
    {
      "epoch": 1.8629185799081645,
      "grad_norm": 1.4218684434890747,
      "learning_rate": 9.651236842105263e-05,
      "loss": 0.4884,
      "step": 33268
    },
    {
      "epoch": 1.8629745772202935,
      "grad_norm": 1.450047492980957,
      "learning_rate": 9.65121052631579e-05,
      "loss": 0.7257,
      "step": 33269
    },
    {
      "epoch": 1.8630305745324225,
      "grad_norm": 1.3057308197021484,
      "learning_rate": 9.651184210526316e-05,
      "loss": 0.408,
      "step": 33270
    },
    {
      "epoch": 1.8630865718445515,
      "grad_norm": 1.2041081190109253,
      "learning_rate": 9.651157894736842e-05,
      "loss": 0.4326,
      "step": 33271
    },
    {
      "epoch": 1.8631425691566805,
      "grad_norm": 1.2635146379470825,
      "learning_rate": 9.651131578947368e-05,
      "loss": 0.4321,
      "step": 33272
    },
    {
      "epoch": 1.8631985664688095,
      "grad_norm": 1.2276989221572876,
      "learning_rate": 9.651105263157894e-05,
      "loss": 0.3515,
      "step": 33273
    },
    {
      "epoch": 1.8632545637809386,
      "grad_norm": 1.303195595741272,
      "learning_rate": 9.651078947368422e-05,
      "loss": 0.4348,
      "step": 33274
    },
    {
      "epoch": 1.8633105610930676,
      "grad_norm": 1.6392093896865845,
      "learning_rate": 9.651052631578948e-05,
      "loss": 0.6247,
      "step": 33275
    },
    {
      "epoch": 1.8633665584051966,
      "grad_norm": 1.3800324201583862,
      "learning_rate": 9.651026315789474e-05,
      "loss": 0.5736,
      "step": 33276
    },
    {
      "epoch": 1.8634225557173256,
      "grad_norm": 1.1216520071029663,
      "learning_rate": 9.651e-05,
      "loss": 0.3868,
      "step": 33277
    },
    {
      "epoch": 1.8634785530294544,
      "grad_norm": 1.1675200462341309,
      "learning_rate": 9.650973684210527e-05,
      "loss": 0.3946,
      "step": 33278
    },
    {
      "epoch": 1.8635345503415834,
      "grad_norm": 1.2029776573181152,
      "learning_rate": 9.650947368421053e-05,
      "loss": 0.4378,
      "step": 33279
    },
    {
      "epoch": 1.8635905476537125,
      "grad_norm": 1.215852975845337,
      "learning_rate": 9.65092105263158e-05,
      "loss": 0.4484,
      "step": 33280
    },
    {
      "epoch": 1.8636465449658415,
      "grad_norm": 1.428413987159729,
      "learning_rate": 9.650894736842105e-05,
      "loss": 0.4041,
      "step": 33281
    },
    {
      "epoch": 1.8637025422779705,
      "grad_norm": 1.229519009590149,
      "learning_rate": 9.650868421052632e-05,
      "loss": 0.3805,
      "step": 33282
    },
    {
      "epoch": 1.8637585395900995,
      "grad_norm": 1.3234349489212036,
      "learning_rate": 9.650842105263158e-05,
      "loss": 0.5095,
      "step": 33283
    },
    {
      "epoch": 1.8638145369022285,
      "grad_norm": 1.1571629047393799,
      "learning_rate": 9.650815789473685e-05,
      "loss": 0.4227,
      "step": 33284
    },
    {
      "epoch": 1.8638705342143576,
      "grad_norm": 1.1840343475341797,
      "learning_rate": 9.650789473684211e-05,
      "loss": 0.464,
      "step": 33285
    },
    {
      "epoch": 1.8639265315264866,
      "grad_norm": 1.4128124713897705,
      "learning_rate": 9.650763157894737e-05,
      "loss": 0.4179,
      "step": 33286
    },
    {
      "epoch": 1.8639825288386156,
      "grad_norm": 1.331292986869812,
      "learning_rate": 9.650736842105263e-05,
      "loss": 0.528,
      "step": 33287
    },
    {
      "epoch": 1.8640385261507446,
      "grad_norm": 1.3400959968566895,
      "learning_rate": 9.650710526315791e-05,
      "loss": 0.4923,
      "step": 33288
    },
    {
      "epoch": 1.8640945234628736,
      "grad_norm": 1.6587700843811035,
      "learning_rate": 9.650684210526317e-05,
      "loss": 0.3992,
      "step": 33289
    },
    {
      "epoch": 1.8641505207750027,
      "grad_norm": 1.353323221206665,
      "learning_rate": 9.650657894736843e-05,
      "loss": 0.5483,
      "step": 33290
    },
    {
      "epoch": 1.8642065180871317,
      "grad_norm": 1.2740854024887085,
      "learning_rate": 9.650631578947369e-05,
      "loss": 0.3513,
      "step": 33291
    },
    {
      "epoch": 1.8642625153992607,
      "grad_norm": 1.1472843885421753,
      "learning_rate": 9.650605263157895e-05,
      "loss": 0.4573,
      "step": 33292
    },
    {
      "epoch": 1.8643185127113897,
      "grad_norm": 1.3171521425247192,
      "learning_rate": 9.650578947368422e-05,
      "loss": 0.6538,
      "step": 33293
    },
    {
      "epoch": 1.8643745100235187,
      "grad_norm": 1.2200098037719727,
      "learning_rate": 9.650552631578948e-05,
      "loss": 0.402,
      "step": 33294
    },
    {
      "epoch": 1.8644305073356477,
      "grad_norm": 1.4275422096252441,
      "learning_rate": 9.650526315789474e-05,
      "loss": 0.4681,
      "step": 33295
    },
    {
      "epoch": 1.8644865046477768,
      "grad_norm": 1.1108968257904053,
      "learning_rate": 9.6505e-05,
      "loss": 0.3743,
      "step": 33296
    },
    {
      "epoch": 1.8645425019599058,
      "grad_norm": 1.392356514930725,
      "learning_rate": 9.650473684210527e-05,
      "loss": 0.4675,
      "step": 33297
    },
    {
      "epoch": 1.8645984992720348,
      "grad_norm": 1.3910579681396484,
      "learning_rate": 9.650447368421053e-05,
      "loss": 0.4508,
      "step": 33298
    },
    {
      "epoch": 1.8646544965841638,
      "grad_norm": 1.153457522392273,
      "learning_rate": 9.650421052631579e-05,
      "loss": 0.331,
      "step": 33299
    },
    {
      "epoch": 1.8647104938962928,
      "grad_norm": 1.8094803094863892,
      "learning_rate": 9.650394736842105e-05,
      "loss": 0.5507,
      "step": 33300
    },
    {
      "epoch": 1.8647664912084219,
      "grad_norm": 1.2175483703613281,
      "learning_rate": 9.650368421052632e-05,
      "loss": 0.4674,
      "step": 33301
    },
    {
      "epoch": 1.8648224885205509,
      "grad_norm": 1.3713819980621338,
      "learning_rate": 9.650342105263158e-05,
      "loss": 0.5527,
      "step": 33302
    },
    {
      "epoch": 1.86487848583268,
      "grad_norm": 1.3288484811782837,
      "learning_rate": 9.650315789473686e-05,
      "loss": 0.4117,
      "step": 33303
    },
    {
      "epoch": 1.864934483144809,
      "grad_norm": 1.3322553634643555,
      "learning_rate": 9.65028947368421e-05,
      "loss": 0.4769,
      "step": 33304
    },
    {
      "epoch": 1.864990480456938,
      "grad_norm": 1.8681646585464478,
      "learning_rate": 9.650263157894738e-05,
      "loss": 0.6378,
      "step": 33305
    },
    {
      "epoch": 1.865046477769067,
      "grad_norm": 1.0922811031341553,
      "learning_rate": 9.650236842105264e-05,
      "loss": 0.4184,
      "step": 33306
    },
    {
      "epoch": 1.865102475081196,
      "grad_norm": 1.4009759426116943,
      "learning_rate": 9.650210526315791e-05,
      "loss": 0.6117,
      "step": 33307
    },
    {
      "epoch": 1.865158472393325,
      "grad_norm": 1.2001932859420776,
      "learning_rate": 9.650184210526317e-05,
      "loss": 0.3913,
      "step": 33308
    },
    {
      "epoch": 1.865214469705454,
      "grad_norm": 1.260614037513733,
      "learning_rate": 9.650157894736842e-05,
      "loss": 0.5306,
      "step": 33309
    },
    {
      "epoch": 1.865270467017583,
      "grad_norm": 1.0835615396499634,
      "learning_rate": 9.650131578947369e-05,
      "loss": 0.405,
      "step": 33310
    },
    {
      "epoch": 1.865326464329712,
      "grad_norm": 1.2805088758468628,
      "learning_rate": 9.650105263157895e-05,
      "loss": 0.4737,
      "step": 33311
    },
    {
      "epoch": 1.865382461641841,
      "grad_norm": 1.0372954607009888,
      "learning_rate": 9.650078947368422e-05,
      "loss": 0.4574,
      "step": 33312
    },
    {
      "epoch": 1.86543845895397,
      "grad_norm": 2.122236728668213,
      "learning_rate": 9.650052631578947e-05,
      "loss": 0.6584,
      "step": 33313
    },
    {
      "epoch": 1.865494456266099,
      "grad_norm": 1.1040592193603516,
      "learning_rate": 9.650026315789474e-05,
      "loss": 0.4956,
      "step": 33314
    },
    {
      "epoch": 1.8655504535782281,
      "grad_norm": 1.2051090002059937,
      "learning_rate": 9.65e-05,
      "loss": 0.4011,
      "step": 33315
    },
    {
      "epoch": 1.8656064508903571,
      "grad_norm": 1.1285511255264282,
      "learning_rate": 9.649973684210527e-05,
      "loss": 0.4076,
      "step": 33316
    },
    {
      "epoch": 1.8656624482024862,
      "grad_norm": 1.331356406211853,
      "learning_rate": 9.649947368421053e-05,
      "loss": 0.5501,
      "step": 33317
    },
    {
      "epoch": 1.8657184455146152,
      "grad_norm": 1.309225082397461,
      "learning_rate": 9.64992105263158e-05,
      "loss": 0.5664,
      "step": 33318
    },
    {
      "epoch": 1.8657744428267442,
      "grad_norm": 1.356074333190918,
      "learning_rate": 9.649894736842105e-05,
      "loss": 0.4073,
      "step": 33319
    },
    {
      "epoch": 1.8658304401388732,
      "grad_norm": 1.1765908002853394,
      "learning_rate": 9.649868421052633e-05,
      "loss": 0.3849,
      "step": 33320
    },
    {
      "epoch": 1.8658864374510022,
      "grad_norm": 1.1801725625991821,
      "learning_rate": 9.649842105263159e-05,
      "loss": 0.3199,
      "step": 33321
    },
    {
      "epoch": 1.8659424347631313,
      "grad_norm": 1.316142201423645,
      "learning_rate": 9.649815789473685e-05,
      "loss": 0.4739,
      "step": 33322
    },
    {
      "epoch": 1.8659984320752603,
      "grad_norm": 1.6283364295959473,
      "learning_rate": 9.64978947368421e-05,
      "loss": 0.4789,
      "step": 33323
    },
    {
      "epoch": 1.8660544293873893,
      "grad_norm": 1.1305875778198242,
      "learning_rate": 9.649763157894738e-05,
      "loss": 0.392,
      "step": 33324
    },
    {
      "epoch": 1.8661104266995183,
      "grad_norm": 1.1613478660583496,
      "learning_rate": 9.649736842105264e-05,
      "loss": 0.4457,
      "step": 33325
    },
    {
      "epoch": 1.8661664240116473,
      "grad_norm": 1.3853996992111206,
      "learning_rate": 9.64971052631579e-05,
      "loss": 0.4711,
      "step": 33326
    },
    {
      "epoch": 1.8662224213237764,
      "grad_norm": 1.3434357643127441,
      "learning_rate": 9.649684210526316e-05,
      "loss": 0.4273,
      "step": 33327
    },
    {
      "epoch": 1.8662784186359054,
      "grad_norm": 1.1406131982803345,
      "learning_rate": 9.649657894736842e-05,
      "loss": 0.4457,
      "step": 33328
    },
    {
      "epoch": 1.8663344159480344,
      "grad_norm": 1.2307461500167847,
      "learning_rate": 9.649631578947369e-05,
      "loss": 0.4473,
      "step": 33329
    },
    {
      "epoch": 1.8663904132601634,
      "grad_norm": 1.5934114456176758,
      "learning_rate": 9.649605263157895e-05,
      "loss": 0.5564,
      "step": 33330
    },
    {
      "epoch": 1.8664464105722924,
      "grad_norm": 1.3289058208465576,
      "learning_rate": 9.649578947368421e-05,
      "loss": 0.45,
      "step": 33331
    },
    {
      "epoch": 1.8665024078844215,
      "grad_norm": 2.0612432956695557,
      "learning_rate": 9.649552631578947e-05,
      "loss": 0.4376,
      "step": 33332
    },
    {
      "epoch": 1.8665584051965505,
      "grad_norm": 1.287265419960022,
      "learning_rate": 9.649526315789474e-05,
      "loss": 0.4721,
      "step": 33333
    },
    {
      "epoch": 1.8666144025086795,
      "grad_norm": 1.308773159980774,
      "learning_rate": 9.6495e-05,
      "loss": 0.3741,
      "step": 33334
    },
    {
      "epoch": 1.8666703998208085,
      "grad_norm": 1.1810001134872437,
      "learning_rate": 9.649473684210528e-05,
      "loss": 0.373,
      "step": 33335
    },
    {
      "epoch": 1.8667263971329375,
      "grad_norm": 1.1118175983428955,
      "learning_rate": 9.649447368421052e-05,
      "loss": 0.5305,
      "step": 33336
    },
    {
      "epoch": 1.8667823944450666,
      "grad_norm": 1.206966519355774,
      "learning_rate": 9.64942105263158e-05,
      "loss": 0.437,
      "step": 33337
    },
    {
      "epoch": 1.8668383917571956,
      "grad_norm": 1.597222924232483,
      "learning_rate": 9.649394736842106e-05,
      "loss": 0.449,
      "step": 33338
    },
    {
      "epoch": 1.8668943890693246,
      "grad_norm": 1.008517861366272,
      "learning_rate": 9.649368421052633e-05,
      "loss": 0.3517,
      "step": 33339
    },
    {
      "epoch": 1.8669503863814536,
      "grad_norm": 1.286507248878479,
      "learning_rate": 9.649342105263159e-05,
      "loss": 0.4505,
      "step": 33340
    },
    {
      "epoch": 1.8670063836935826,
      "grad_norm": 1.3307602405548096,
      "learning_rate": 9.649315789473685e-05,
      "loss": 0.57,
      "step": 33341
    },
    {
      "epoch": 1.8670623810057116,
      "grad_norm": 1.8110291957855225,
      "learning_rate": 9.649289473684211e-05,
      "loss": 0.5791,
      "step": 33342
    },
    {
      "epoch": 1.8671183783178407,
      "grad_norm": 66.24345397949219,
      "learning_rate": 9.649263157894737e-05,
      "loss": 0.4023,
      "step": 33343
    },
    {
      "epoch": 1.8671743756299697,
      "grad_norm": 1.3057340383529663,
      "learning_rate": 9.649236842105264e-05,
      "loss": 0.4617,
      "step": 33344
    },
    {
      "epoch": 1.8672303729420987,
      "grad_norm": 1.3963127136230469,
      "learning_rate": 9.64921052631579e-05,
      "loss": 0.4739,
      "step": 33345
    },
    {
      "epoch": 1.8672863702542277,
      "grad_norm": 1.2040624618530273,
      "learning_rate": 9.649184210526316e-05,
      "loss": 0.4258,
      "step": 33346
    },
    {
      "epoch": 1.8673423675663567,
      "grad_norm": 1.2081611156463623,
      "learning_rate": 9.649157894736842e-05,
      "loss": 0.4192,
      "step": 33347
    },
    {
      "epoch": 1.8673983648784858,
      "grad_norm": 1.3619356155395508,
      "learning_rate": 9.64913157894737e-05,
      "loss": 0.5874,
      "step": 33348
    },
    {
      "epoch": 1.8674543621906148,
      "grad_norm": 1.1492903232574463,
      "learning_rate": 9.649105263157895e-05,
      "loss": 0.4294,
      "step": 33349
    },
    {
      "epoch": 1.8675103595027438,
      "grad_norm": 1.2217748165130615,
      "learning_rate": 9.649078947368421e-05,
      "loss": 0.475,
      "step": 33350
    },
    {
      "epoch": 1.8675663568148728,
      "grad_norm": 1.3813090324401855,
      "learning_rate": 9.649052631578947e-05,
      "loss": 0.3238,
      "step": 33351
    },
    {
      "epoch": 1.8676223541270018,
      "grad_norm": 1.1766788959503174,
      "learning_rate": 9.649026315789475e-05,
      "loss": 0.3393,
      "step": 33352
    },
    {
      "epoch": 1.8676783514391309,
      "grad_norm": 1.260389804840088,
      "learning_rate": 9.649e-05,
      "loss": 0.3905,
      "step": 33353
    },
    {
      "epoch": 1.8677343487512599,
      "grad_norm": 1.2006301879882812,
      "learning_rate": 9.648973684210527e-05,
      "loss": 0.5319,
      "step": 33354
    },
    {
      "epoch": 1.867790346063389,
      "grad_norm": 1.167595624923706,
      "learning_rate": 9.648947368421053e-05,
      "loss": 0.5305,
      "step": 33355
    },
    {
      "epoch": 1.867846343375518,
      "grad_norm": 2.3871161937713623,
      "learning_rate": 9.64892105263158e-05,
      "loss": 0.4194,
      "step": 33356
    },
    {
      "epoch": 1.867902340687647,
      "grad_norm": 1.462841272354126,
      "learning_rate": 9.648894736842106e-05,
      "loss": 0.5383,
      "step": 33357
    },
    {
      "epoch": 1.867958337999776,
      "grad_norm": 1.3849934339523315,
      "learning_rate": 9.648868421052633e-05,
      "loss": 0.5103,
      "step": 33358
    },
    {
      "epoch": 1.868014335311905,
      "grad_norm": 1.3383766412734985,
      "learning_rate": 9.648842105263158e-05,
      "loss": 0.4018,
      "step": 33359
    },
    {
      "epoch": 1.868070332624034,
      "grad_norm": 1.3043535947799683,
      "learning_rate": 9.648815789473684e-05,
      "loss": 0.3786,
      "step": 33360
    },
    {
      "epoch": 1.868126329936163,
      "grad_norm": 1.200724720954895,
      "learning_rate": 9.648789473684211e-05,
      "loss": 0.4577,
      "step": 33361
    },
    {
      "epoch": 1.868182327248292,
      "grad_norm": 1.3673505783081055,
      "learning_rate": 9.648763157894737e-05,
      "loss": 0.442,
      "step": 33362
    },
    {
      "epoch": 1.868238324560421,
      "grad_norm": 1.3371307849884033,
      "learning_rate": 9.648736842105264e-05,
      "loss": 0.687,
      "step": 33363
    },
    {
      "epoch": 1.86829432187255,
      "grad_norm": 1.0764373540878296,
      "learning_rate": 9.648710526315789e-05,
      "loss": 0.376,
      "step": 33364
    },
    {
      "epoch": 1.868350319184679,
      "grad_norm": 1.60707426071167,
      "learning_rate": 9.648684210526316e-05,
      "loss": 0.5271,
      "step": 33365
    },
    {
      "epoch": 1.868406316496808,
      "grad_norm": 1.3147951364517212,
      "learning_rate": 9.648657894736842e-05,
      "loss": 0.412,
      "step": 33366
    },
    {
      "epoch": 1.8684623138089371,
      "grad_norm": 1.316485047340393,
      "learning_rate": 9.64863157894737e-05,
      "loss": 0.3975,
      "step": 33367
    },
    {
      "epoch": 1.8685183111210661,
      "grad_norm": 0.9729315042495728,
      "learning_rate": 9.648605263157894e-05,
      "loss": 0.3303,
      "step": 33368
    },
    {
      "epoch": 1.8685743084331952,
      "grad_norm": 1.1686781644821167,
      "learning_rate": 9.648578947368422e-05,
      "loss": 0.4727,
      "step": 33369
    },
    {
      "epoch": 1.8686303057453242,
      "grad_norm": 1.7420340776443481,
      "learning_rate": 9.648552631578948e-05,
      "loss": 0.4967,
      "step": 33370
    },
    {
      "epoch": 1.8686863030574532,
      "grad_norm": 2.7304697036743164,
      "learning_rate": 9.648526315789475e-05,
      "loss": 0.5604,
      "step": 33371
    },
    {
      "epoch": 1.8687423003695822,
      "grad_norm": 1.4076279401779175,
      "learning_rate": 9.648500000000001e-05,
      "loss": 0.5262,
      "step": 33372
    },
    {
      "epoch": 1.8687982976817112,
      "grad_norm": 2.1522576808929443,
      "learning_rate": 9.648473684210527e-05,
      "loss": 0.4765,
      "step": 33373
    },
    {
      "epoch": 1.8688542949938403,
      "grad_norm": 1.5706709623336792,
      "learning_rate": 9.648447368421053e-05,
      "loss": 0.3541,
      "step": 33374
    },
    {
      "epoch": 1.8689102923059693,
      "grad_norm": 1.5873135328292847,
      "learning_rate": 9.64842105263158e-05,
      "loss": 0.5126,
      "step": 33375
    },
    {
      "epoch": 1.8689662896180983,
      "grad_norm": 1.2355859279632568,
      "learning_rate": 9.648394736842106e-05,
      "loss": 0.4603,
      "step": 33376
    },
    {
      "epoch": 1.8690222869302273,
      "grad_norm": 1.0957920551300049,
      "learning_rate": 9.648368421052632e-05,
      "loss": 0.4055,
      "step": 33377
    },
    {
      "epoch": 1.8690782842423563,
      "grad_norm": 1.0139875411987305,
      "learning_rate": 9.648342105263158e-05,
      "loss": 0.3791,
      "step": 33378
    },
    {
      "epoch": 1.8691342815544854,
      "grad_norm": 1.4458435773849487,
      "learning_rate": 9.648315789473684e-05,
      "loss": 0.3773,
      "step": 33379
    },
    {
      "epoch": 1.8691902788666144,
      "grad_norm": 1.202217936515808,
      "learning_rate": 9.648289473684211e-05,
      "loss": 0.3829,
      "step": 33380
    },
    {
      "epoch": 1.8692462761787434,
      "grad_norm": 1.1900771856307983,
      "learning_rate": 9.648263157894737e-05,
      "loss": 0.5546,
      "step": 33381
    },
    {
      "epoch": 1.8693022734908724,
      "grad_norm": 1.2505542039871216,
      "learning_rate": 9.648236842105263e-05,
      "loss": 0.3581,
      "step": 33382
    },
    {
      "epoch": 1.8693582708030014,
      "grad_norm": 1.3131285905838013,
      "learning_rate": 9.648210526315789e-05,
      "loss": 0.5366,
      "step": 33383
    },
    {
      "epoch": 1.8694142681151305,
      "grad_norm": 1.3099415302276611,
      "learning_rate": 9.648184210526317e-05,
      "loss": 0.434,
      "step": 33384
    },
    {
      "epoch": 1.8694702654272595,
      "grad_norm": 1.3272502422332764,
      "learning_rate": 9.648157894736843e-05,
      "loss": 0.4881,
      "step": 33385
    },
    {
      "epoch": 1.8695262627393885,
      "grad_norm": 1.1791683435440063,
      "learning_rate": 9.648131578947369e-05,
      "loss": 0.3329,
      "step": 33386
    },
    {
      "epoch": 1.8695822600515175,
      "grad_norm": 1.0309921503067017,
      "learning_rate": 9.648105263157895e-05,
      "loss": 0.4096,
      "step": 33387
    },
    {
      "epoch": 1.8696382573636465,
      "grad_norm": 1.122999906539917,
      "learning_rate": 9.648078947368422e-05,
      "loss": 0.4852,
      "step": 33388
    },
    {
      "epoch": 1.8696942546757755,
      "grad_norm": 1.4413858652114868,
      "learning_rate": 9.648052631578948e-05,
      "loss": 0.541,
      "step": 33389
    },
    {
      "epoch": 1.8697502519879046,
      "grad_norm": 1.4153752326965332,
      "learning_rate": 9.648026315789475e-05,
      "loss": 0.501,
      "step": 33390
    },
    {
      "epoch": 1.8698062493000336,
      "grad_norm": 1.1813348531723022,
      "learning_rate": 9.648e-05,
      "loss": 0.4236,
      "step": 33391
    },
    {
      "epoch": 1.8698622466121626,
      "grad_norm": 1.3457003831863403,
      "learning_rate": 9.647973684210527e-05,
      "loss": 0.5359,
      "step": 33392
    },
    {
      "epoch": 1.8699182439242916,
      "grad_norm": 1.1282265186309814,
      "learning_rate": 9.647947368421053e-05,
      "loss": 0.367,
      "step": 33393
    },
    {
      "epoch": 1.8699742412364206,
      "grad_norm": 1.2186192274093628,
      "learning_rate": 9.647921052631579e-05,
      "loss": 0.4307,
      "step": 33394
    },
    {
      "epoch": 1.8700302385485497,
      "grad_norm": 1.0650784969329834,
      "learning_rate": 9.647894736842106e-05,
      "loss": 0.443,
      "step": 33395
    },
    {
      "epoch": 1.8700862358606787,
      "grad_norm": 1.222720980644226,
      "learning_rate": 9.647868421052631e-05,
      "loss": 0.4682,
      "step": 33396
    },
    {
      "epoch": 1.8701422331728077,
      "grad_norm": 1.4409343004226685,
      "learning_rate": 9.647842105263158e-05,
      "loss": 0.4807,
      "step": 33397
    },
    {
      "epoch": 1.8701982304849367,
      "grad_norm": 1.3927421569824219,
      "learning_rate": 9.647815789473684e-05,
      "loss": 0.4209,
      "step": 33398
    },
    {
      "epoch": 1.8702542277970657,
      "grad_norm": 1.0722668170928955,
      "learning_rate": 9.647789473684212e-05,
      "loss": 0.4513,
      "step": 33399
    },
    {
      "epoch": 1.8703102251091948,
      "grad_norm": 1.2874648571014404,
      "learning_rate": 9.647763157894738e-05,
      "loss": 0.549,
      "step": 33400
    },
    {
      "epoch": 1.8703662224213238,
      "grad_norm": 1.316003441810608,
      "learning_rate": 9.647736842105264e-05,
      "loss": 0.5344,
      "step": 33401
    },
    {
      "epoch": 1.8704222197334528,
      "grad_norm": 1.159010887145996,
      "learning_rate": 9.64771052631579e-05,
      "loss": 0.3841,
      "step": 33402
    },
    {
      "epoch": 1.8704782170455818,
      "grad_norm": 1.185390830039978,
      "learning_rate": 9.647684210526317e-05,
      "loss": 0.5582,
      "step": 33403
    },
    {
      "epoch": 1.8705342143577108,
      "grad_norm": 1.2026081085205078,
      "learning_rate": 9.647657894736843e-05,
      "loss": 0.4247,
      "step": 33404
    },
    {
      "epoch": 1.8705902116698399,
      "grad_norm": 1.1749247312545776,
      "learning_rate": 9.647631578947369e-05,
      "loss": 0.4169,
      "step": 33405
    },
    {
      "epoch": 1.8706462089819689,
      "grad_norm": 1.0058047771453857,
      "learning_rate": 9.647605263157895e-05,
      "loss": 0.4166,
      "step": 33406
    },
    {
      "epoch": 1.870702206294098,
      "grad_norm": 1.3511548042297363,
      "learning_rate": 9.647578947368422e-05,
      "loss": 0.4144,
      "step": 33407
    },
    {
      "epoch": 1.870758203606227,
      "grad_norm": 1.2476104497909546,
      "learning_rate": 9.647552631578948e-05,
      "loss": 0.5436,
      "step": 33408
    },
    {
      "epoch": 1.870814200918356,
      "grad_norm": 1.3797138929367065,
      "learning_rate": 9.647526315789474e-05,
      "loss": 0.4511,
      "step": 33409
    },
    {
      "epoch": 1.870870198230485,
      "grad_norm": 1.3588685989379883,
      "learning_rate": 9.6475e-05,
      "loss": 0.3568,
      "step": 33410
    },
    {
      "epoch": 1.870926195542614,
      "grad_norm": 1.2693657875061035,
      "learning_rate": 9.647473684210527e-05,
      "loss": 0.4852,
      "step": 33411
    },
    {
      "epoch": 1.870982192854743,
      "grad_norm": 1.2818776369094849,
      "learning_rate": 9.647447368421053e-05,
      "loss": 0.519,
      "step": 33412
    },
    {
      "epoch": 1.871038190166872,
      "grad_norm": 1.0625146627426147,
      "learning_rate": 9.647421052631579e-05,
      "loss": 0.4105,
      "step": 33413
    },
    {
      "epoch": 1.871094187479001,
      "grad_norm": 1.251999020576477,
      "learning_rate": 9.647394736842105e-05,
      "loss": 0.386,
      "step": 33414
    },
    {
      "epoch": 1.87115018479113,
      "grad_norm": 1.3598161935806274,
      "learning_rate": 9.647368421052631e-05,
      "loss": 0.4513,
      "step": 33415
    },
    {
      "epoch": 1.871206182103259,
      "grad_norm": 1.6231831312179565,
      "learning_rate": 9.647342105263159e-05,
      "loss": 0.3562,
      "step": 33416
    },
    {
      "epoch": 1.871262179415388,
      "grad_norm": 1.338484764099121,
      "learning_rate": 9.647315789473685e-05,
      "loss": 0.5784,
      "step": 33417
    },
    {
      "epoch": 1.871318176727517,
      "grad_norm": 1.1802111864089966,
      "learning_rate": 9.64728947368421e-05,
      "loss": 0.3312,
      "step": 33418
    },
    {
      "epoch": 1.8713741740396461,
      "grad_norm": 1.439077615737915,
      "learning_rate": 9.647263157894737e-05,
      "loss": 0.5292,
      "step": 33419
    },
    {
      "epoch": 1.8714301713517751,
      "grad_norm": 1.4299787282943726,
      "learning_rate": 9.647236842105264e-05,
      "loss": 0.5413,
      "step": 33420
    },
    {
      "epoch": 1.8714861686639042,
      "grad_norm": 1.0642756223678589,
      "learning_rate": 9.64721052631579e-05,
      "loss": 0.3577,
      "step": 33421
    },
    {
      "epoch": 1.8715421659760332,
      "grad_norm": 1.4010738134384155,
      "learning_rate": 9.647184210526317e-05,
      "loss": 0.5869,
      "step": 33422
    },
    {
      "epoch": 1.8715981632881622,
      "grad_norm": 1.336345911026001,
      "learning_rate": 9.647157894736842e-05,
      "loss": 0.4308,
      "step": 33423
    },
    {
      "epoch": 1.8716541606002912,
      "grad_norm": 1.9138880968093872,
      "learning_rate": 9.647131578947369e-05,
      "loss": 0.442,
      "step": 33424
    },
    {
      "epoch": 1.8717101579124202,
      "grad_norm": 1.2148171663284302,
      "learning_rate": 9.647105263157895e-05,
      "loss": 0.4135,
      "step": 33425
    },
    {
      "epoch": 1.8717661552245493,
      "grad_norm": 1.213161826133728,
      "learning_rate": 9.647078947368422e-05,
      "loss": 0.3819,
      "step": 33426
    },
    {
      "epoch": 1.8718221525366783,
      "grad_norm": 1.2208739519119263,
      "learning_rate": 9.647052631578948e-05,
      "loss": 0.4355,
      "step": 33427
    },
    {
      "epoch": 1.8718781498488073,
      "grad_norm": 1.170845627784729,
      "learning_rate": 9.647026315789474e-05,
      "loss": 0.432,
      "step": 33428
    },
    {
      "epoch": 1.8719341471609363,
      "grad_norm": 1.3409026861190796,
      "learning_rate": 9.647e-05,
      "loss": 0.4973,
      "step": 33429
    },
    {
      "epoch": 1.8719901444730653,
      "grad_norm": 0.9035952091217041,
      "learning_rate": 9.646973684210526e-05,
      "loss": 0.3667,
      "step": 33430
    },
    {
      "epoch": 1.8720461417851944,
      "grad_norm": 1.4720350503921509,
      "learning_rate": 9.646947368421054e-05,
      "loss": 0.4517,
      "step": 33431
    },
    {
      "epoch": 1.8721021390973234,
      "grad_norm": 1.2141084671020508,
      "learning_rate": 9.64692105263158e-05,
      "loss": 0.6211,
      "step": 33432
    },
    {
      "epoch": 1.8721581364094524,
      "grad_norm": 1.2069872617721558,
      "learning_rate": 9.646894736842106e-05,
      "loss": 0.3285,
      "step": 33433
    },
    {
      "epoch": 1.8722141337215814,
      "grad_norm": 1.1581941843032837,
      "learning_rate": 9.646868421052632e-05,
      "loss": 0.3489,
      "step": 33434
    },
    {
      "epoch": 1.8722701310337104,
      "grad_norm": 1.3856080770492554,
      "learning_rate": 9.646842105263159e-05,
      "loss": 0.4284,
      "step": 33435
    },
    {
      "epoch": 1.8723261283458394,
      "grad_norm": 1.283517599105835,
      "learning_rate": 9.646815789473685e-05,
      "loss": 0.4719,
      "step": 33436
    },
    {
      "epoch": 1.8723821256579685,
      "grad_norm": 1.2817213535308838,
      "learning_rate": 9.646789473684211e-05,
      "loss": 0.3189,
      "step": 33437
    },
    {
      "epoch": 1.8724381229700975,
      "grad_norm": 1.4360096454620361,
      "learning_rate": 9.646763157894737e-05,
      "loss": 0.4909,
      "step": 33438
    },
    {
      "epoch": 1.8724941202822265,
      "grad_norm": 1.327418565750122,
      "learning_rate": 9.646736842105264e-05,
      "loss": 0.4441,
      "step": 33439
    },
    {
      "epoch": 1.8725501175943555,
      "grad_norm": 1.2192895412445068,
      "learning_rate": 9.64671052631579e-05,
      "loss": 0.4229,
      "step": 33440
    },
    {
      "epoch": 1.8726061149064845,
      "grad_norm": 1.1235969066619873,
      "learning_rate": 9.646684210526316e-05,
      "loss": 0.3288,
      "step": 33441
    },
    {
      "epoch": 1.8726621122186136,
      "grad_norm": 1.4548892974853516,
      "learning_rate": 9.646657894736842e-05,
      "loss": 0.4478,
      "step": 33442
    },
    {
      "epoch": 1.8727181095307426,
      "grad_norm": 1.042901873588562,
      "learning_rate": 9.64663157894737e-05,
      "loss": 0.4034,
      "step": 33443
    },
    {
      "epoch": 1.8727741068428716,
      "grad_norm": 1.234602689743042,
      "learning_rate": 9.646605263157895e-05,
      "loss": 0.4703,
      "step": 33444
    },
    {
      "epoch": 1.8728301041550006,
      "grad_norm": 1.1242870092391968,
      "learning_rate": 9.646578947368423e-05,
      "loss": 0.3587,
      "step": 33445
    },
    {
      "epoch": 1.8728861014671296,
      "grad_norm": 19.38405418395996,
      "learning_rate": 9.646552631578947e-05,
      "loss": 0.6063,
      "step": 33446
    },
    {
      "epoch": 1.8729420987792587,
      "grad_norm": 1.3827544450759888,
      "learning_rate": 9.646526315789473e-05,
      "loss": 0.5936,
      "step": 33447
    },
    {
      "epoch": 1.8729980960913877,
      "grad_norm": 1.390112280845642,
      "learning_rate": 9.6465e-05,
      "loss": 0.4984,
      "step": 33448
    },
    {
      "epoch": 1.8730540934035167,
      "grad_norm": 1.5721423625946045,
      "learning_rate": 9.646473684210527e-05,
      "loss": 0.4716,
      "step": 33449
    },
    {
      "epoch": 1.8731100907156457,
      "grad_norm": 1.161504864692688,
      "learning_rate": 9.646447368421054e-05,
      "loss": 0.5542,
      "step": 33450
    },
    {
      "epoch": 1.8731660880277747,
      "grad_norm": 1.4575273990631104,
      "learning_rate": 9.646421052631578e-05,
      "loss": 0.4453,
      "step": 33451
    },
    {
      "epoch": 1.8732220853399038,
      "grad_norm": 1.0841702222824097,
      "learning_rate": 9.646394736842106e-05,
      "loss": 0.4081,
      "step": 33452
    },
    {
      "epoch": 1.8732780826520328,
      "grad_norm": 2.4907963275909424,
      "learning_rate": 9.646368421052632e-05,
      "loss": 0.5758,
      "step": 33453
    },
    {
      "epoch": 1.8733340799641618,
      "grad_norm": 1.3458235263824463,
      "learning_rate": 9.646342105263159e-05,
      "loss": 0.3684,
      "step": 33454
    },
    {
      "epoch": 1.8733900772762908,
      "grad_norm": 2.1697590351104736,
      "learning_rate": 9.646315789473685e-05,
      "loss": 0.4024,
      "step": 33455
    },
    {
      "epoch": 1.8734460745884198,
      "grad_norm": 1.2358863353729248,
      "learning_rate": 9.646289473684211e-05,
      "loss": 0.4061,
      "step": 33456
    },
    {
      "epoch": 1.8735020719005488,
      "grad_norm": 1.3101366758346558,
      "learning_rate": 9.646263157894737e-05,
      "loss": 0.3517,
      "step": 33457
    },
    {
      "epoch": 1.8735580692126779,
      "grad_norm": 1.1429742574691772,
      "learning_rate": 9.646236842105264e-05,
      "loss": 0.4721,
      "step": 33458
    },
    {
      "epoch": 1.8736140665248069,
      "grad_norm": 1.4095683097839355,
      "learning_rate": 9.64621052631579e-05,
      "loss": 0.4284,
      "step": 33459
    },
    {
      "epoch": 1.873670063836936,
      "grad_norm": 1.1719272136688232,
      "learning_rate": 9.646184210526316e-05,
      "loss": 0.3813,
      "step": 33460
    },
    {
      "epoch": 1.873726061149065,
      "grad_norm": 1.4283121824264526,
      "learning_rate": 9.646157894736842e-05,
      "loss": 0.378,
      "step": 33461
    },
    {
      "epoch": 1.873782058461194,
      "grad_norm": 1.238408088684082,
      "learning_rate": 9.64613157894737e-05,
      "loss": 0.3886,
      "step": 33462
    },
    {
      "epoch": 1.873838055773323,
      "grad_norm": 1.2752630710601807,
      "learning_rate": 9.646105263157896e-05,
      "loss": 0.5643,
      "step": 33463
    },
    {
      "epoch": 1.873894053085452,
      "grad_norm": 1.4761980772018433,
      "learning_rate": 9.646078947368422e-05,
      "loss": 0.4346,
      "step": 33464
    },
    {
      "epoch": 1.873950050397581,
      "grad_norm": 1.5497807264328003,
      "learning_rate": 9.646052631578948e-05,
      "loss": 0.4997,
      "step": 33465
    },
    {
      "epoch": 1.87400604770971,
      "grad_norm": 1.2519959211349487,
      "learning_rate": 9.646026315789473e-05,
      "loss": 0.3908,
      "step": 33466
    },
    {
      "epoch": 1.874062045021839,
      "grad_norm": 1.2081587314605713,
      "learning_rate": 9.646000000000001e-05,
      "loss": 0.3579,
      "step": 33467
    },
    {
      "epoch": 1.874118042333968,
      "grad_norm": 1.389553189277649,
      "learning_rate": 9.645973684210527e-05,
      "loss": 0.5409,
      "step": 33468
    },
    {
      "epoch": 1.874174039646097,
      "grad_norm": 1.1664785146713257,
      "learning_rate": 9.645947368421053e-05,
      "loss": 0.4342,
      "step": 33469
    },
    {
      "epoch": 1.874230036958226,
      "grad_norm": 1.5058810710906982,
      "learning_rate": 9.645921052631579e-05,
      "loss": 0.4564,
      "step": 33470
    },
    {
      "epoch": 1.8742860342703551,
      "grad_norm": 1.4256490468978882,
      "learning_rate": 9.645894736842106e-05,
      "loss": 0.7544,
      "step": 33471
    },
    {
      "epoch": 1.8743420315824841,
      "grad_norm": 1.1871120929718018,
      "learning_rate": 9.645868421052632e-05,
      "loss": 0.3614,
      "step": 33472
    },
    {
      "epoch": 1.8743980288946132,
      "grad_norm": 1.2680423259735107,
      "learning_rate": 9.645842105263158e-05,
      "loss": 0.6132,
      "step": 33473
    },
    {
      "epoch": 1.8744540262067422,
      "grad_norm": 1.3779940605163574,
      "learning_rate": 9.645815789473684e-05,
      "loss": 0.5614,
      "step": 33474
    },
    {
      "epoch": 1.8745100235188712,
      "grad_norm": 1.3336397409439087,
      "learning_rate": 9.645789473684211e-05,
      "loss": 0.7411,
      "step": 33475
    },
    {
      "epoch": 1.8745660208310002,
      "grad_norm": 1.2560302019119263,
      "learning_rate": 9.645763157894737e-05,
      "loss": 0.5966,
      "step": 33476
    },
    {
      "epoch": 1.8746220181431292,
      "grad_norm": 1.2894598245620728,
      "learning_rate": 9.645736842105265e-05,
      "loss": 0.5726,
      "step": 33477
    },
    {
      "epoch": 1.8746780154552583,
      "grad_norm": 1.4899004697799683,
      "learning_rate": 9.645710526315789e-05,
      "loss": 0.5685,
      "step": 33478
    },
    {
      "epoch": 1.8747340127673873,
      "grad_norm": 1.2647489309310913,
      "learning_rate": 9.645684210526317e-05,
      "loss": 0.5478,
      "step": 33479
    },
    {
      "epoch": 1.8747900100795163,
      "grad_norm": 1.6174085140228271,
      "learning_rate": 9.645657894736843e-05,
      "loss": 0.5126,
      "step": 33480
    },
    {
      "epoch": 1.8748460073916453,
      "grad_norm": 1.1691871881484985,
      "learning_rate": 9.645631578947369e-05,
      "loss": 0.3347,
      "step": 33481
    },
    {
      "epoch": 1.8749020047037743,
      "grad_norm": 1.0986146926879883,
      "learning_rate": 9.645605263157896e-05,
      "loss": 0.4416,
      "step": 33482
    },
    {
      "epoch": 1.8749580020159033,
      "grad_norm": 1.0803313255310059,
      "learning_rate": 9.64557894736842e-05,
      "loss": 0.3994,
      "step": 33483
    },
    {
      "epoch": 1.8750139993280324,
      "grad_norm": 1.4178773164749146,
      "learning_rate": 9.645552631578948e-05,
      "loss": 0.4352,
      "step": 33484
    },
    {
      "epoch": 1.8750699966401614,
      "grad_norm": 1.1567668914794922,
      "learning_rate": 9.645526315789474e-05,
      "loss": 0.3542,
      "step": 33485
    },
    {
      "epoch": 1.8751259939522904,
      "grad_norm": 1.1704045534133911,
      "learning_rate": 9.645500000000001e-05,
      "loss": 0.3638,
      "step": 33486
    },
    {
      "epoch": 1.8751819912644194,
      "grad_norm": 1.0589802265167236,
      "learning_rate": 9.645473684210527e-05,
      "loss": 0.3783,
      "step": 33487
    },
    {
      "epoch": 1.8752379885765484,
      "grad_norm": 1.2962638139724731,
      "learning_rate": 9.645447368421053e-05,
      "loss": 0.3587,
      "step": 33488
    },
    {
      "epoch": 1.8752939858886775,
      "grad_norm": 1.381291151046753,
      "learning_rate": 9.645421052631579e-05,
      "loss": 0.4869,
      "step": 33489
    },
    {
      "epoch": 1.8753499832008065,
      "grad_norm": 1.4825259447097778,
      "learning_rate": 9.645394736842106e-05,
      "loss": 0.3757,
      "step": 33490
    },
    {
      "epoch": 1.8754059805129355,
      "grad_norm": 1.277748942375183,
      "learning_rate": 9.645368421052632e-05,
      "loss": 0.448,
      "step": 33491
    },
    {
      "epoch": 1.8754619778250645,
      "grad_norm": 1.4495632648468018,
      "learning_rate": 9.645342105263158e-05,
      "loss": 0.4545,
      "step": 33492
    },
    {
      "epoch": 1.8755179751371935,
      "grad_norm": 2.9469432830810547,
      "learning_rate": 9.645315789473684e-05,
      "loss": 0.5421,
      "step": 33493
    },
    {
      "epoch": 1.8755739724493226,
      "grad_norm": 1.1439541578292847,
      "learning_rate": 9.645289473684212e-05,
      "loss": 0.4455,
      "step": 33494
    },
    {
      "epoch": 1.8756299697614516,
      "grad_norm": 1.2018928527832031,
      "learning_rate": 9.645263157894738e-05,
      "loss": 0.4064,
      "step": 33495
    },
    {
      "epoch": 1.8756859670735806,
      "grad_norm": 1.1954567432403564,
      "learning_rate": 9.645236842105264e-05,
      "loss": 0.4607,
      "step": 33496
    },
    {
      "epoch": 1.8757419643857096,
      "grad_norm": 1.1736812591552734,
      "learning_rate": 9.64521052631579e-05,
      "loss": 0.4126,
      "step": 33497
    },
    {
      "epoch": 1.8757979616978386,
      "grad_norm": 1.3685011863708496,
      "learning_rate": 9.645184210526315e-05,
      "loss": 0.537,
      "step": 33498
    },
    {
      "epoch": 1.8758539590099677,
      "grad_norm": 1.4805583953857422,
      "learning_rate": 9.645157894736843e-05,
      "loss": 0.4641,
      "step": 33499
    },
    {
      "epoch": 1.8759099563220967,
      "grad_norm": 1.1129140853881836,
      "learning_rate": 9.645131578947369e-05,
      "loss": 0.5136,
      "step": 33500
    },
    {
      "epoch": 1.8759659536342257,
      "grad_norm": 1.281879186630249,
      "learning_rate": 9.645105263157895e-05,
      "loss": 0.5016,
      "step": 33501
    },
    {
      "epoch": 1.8760219509463547,
      "grad_norm": 1.2343182563781738,
      "learning_rate": 9.645078947368421e-05,
      "loss": 0.4978,
      "step": 33502
    },
    {
      "epoch": 1.8760779482584837,
      "grad_norm": 2.5606625080108643,
      "learning_rate": 9.645052631578948e-05,
      "loss": 0.6544,
      "step": 33503
    },
    {
      "epoch": 1.8761339455706127,
      "grad_norm": 1.1873128414154053,
      "learning_rate": 9.645026315789474e-05,
      "loss": 0.4298,
      "step": 33504
    },
    {
      "epoch": 1.8761899428827418,
      "grad_norm": 1.2903032302856445,
      "learning_rate": 9.645000000000001e-05,
      "loss": 0.3659,
      "step": 33505
    },
    {
      "epoch": 1.8762459401948708,
      "grad_norm": 1.234182596206665,
      "learning_rate": 9.644973684210526e-05,
      "loss": 0.5044,
      "step": 33506
    },
    {
      "epoch": 1.8763019375069998,
      "grad_norm": 1.7295570373535156,
      "learning_rate": 9.644947368421053e-05,
      "loss": 0.5064,
      "step": 33507
    },
    {
      "epoch": 1.8763579348191288,
      "grad_norm": 1.4970859289169312,
      "learning_rate": 9.644921052631579e-05,
      "loss": 0.4238,
      "step": 33508
    },
    {
      "epoch": 1.8764139321312578,
      "grad_norm": 1.4024280309677124,
      "learning_rate": 9.644894736842107e-05,
      "loss": 0.3762,
      "step": 33509
    },
    {
      "epoch": 1.8764699294433869,
      "grad_norm": 1.2532130479812622,
      "learning_rate": 9.644868421052633e-05,
      "loss": 0.4599,
      "step": 33510
    },
    {
      "epoch": 1.8765259267555159,
      "grad_norm": 1.2962455749511719,
      "learning_rate": 9.644842105263159e-05,
      "loss": 0.466,
      "step": 33511
    },
    {
      "epoch": 1.876581924067645,
      "grad_norm": 1.477338194847107,
      "learning_rate": 9.644815789473685e-05,
      "loss": 0.4867,
      "step": 33512
    },
    {
      "epoch": 1.876637921379774,
      "grad_norm": 1.3582735061645508,
      "learning_rate": 9.644789473684212e-05,
      "loss": 0.5876,
      "step": 33513
    },
    {
      "epoch": 1.876693918691903,
      "grad_norm": 1.4817018508911133,
      "learning_rate": 9.644763157894738e-05,
      "loss": 0.5833,
      "step": 33514
    },
    {
      "epoch": 1.876749916004032,
      "grad_norm": 1.1551045179367065,
      "learning_rate": 9.644736842105262e-05,
      "loss": 0.366,
      "step": 33515
    },
    {
      "epoch": 1.876805913316161,
      "grad_norm": 1.5476411581039429,
      "learning_rate": 9.64471052631579e-05,
      "loss": 0.4366,
      "step": 33516
    },
    {
      "epoch": 1.87686191062829,
      "grad_norm": 1.493558406829834,
      "learning_rate": 9.644684210526316e-05,
      "loss": 0.3961,
      "step": 33517
    },
    {
      "epoch": 1.876917907940419,
      "grad_norm": 1.3267251253128052,
      "learning_rate": 9.644657894736843e-05,
      "loss": 0.5182,
      "step": 33518
    },
    {
      "epoch": 1.876973905252548,
      "grad_norm": 1.1937638521194458,
      "learning_rate": 9.644631578947369e-05,
      "loss": 0.536,
      "step": 33519
    },
    {
      "epoch": 1.877029902564677,
      "grad_norm": 1.0580928325653076,
      "learning_rate": 9.644605263157895e-05,
      "loss": 0.375,
      "step": 33520
    },
    {
      "epoch": 1.877085899876806,
      "grad_norm": 1.1075624227523804,
      "learning_rate": 9.644578947368421e-05,
      "loss": 0.4144,
      "step": 33521
    },
    {
      "epoch": 1.877141897188935,
      "grad_norm": 1.1278550624847412,
      "learning_rate": 9.644552631578948e-05,
      "loss": 0.3824,
      "step": 33522
    },
    {
      "epoch": 1.8771978945010641,
      "grad_norm": 1.2467658519744873,
      "learning_rate": 9.644526315789474e-05,
      "loss": 0.4265,
      "step": 33523
    },
    {
      "epoch": 1.8772538918131931,
      "grad_norm": 1.258202314376831,
      "learning_rate": 9.6445e-05,
      "loss": 0.4668,
      "step": 33524
    },
    {
      "epoch": 1.8773098891253222,
      "grad_norm": 1.1626869440078735,
      "learning_rate": 9.644473684210526e-05,
      "loss": 0.4467,
      "step": 33525
    },
    {
      "epoch": 1.877365886437451,
      "grad_norm": 1.2683888673782349,
      "learning_rate": 9.644447368421054e-05,
      "loss": 0.6625,
      "step": 33526
    },
    {
      "epoch": 1.87742188374958,
      "grad_norm": 1.3738536834716797,
      "learning_rate": 9.64442105263158e-05,
      "loss": 0.5151,
      "step": 33527
    },
    {
      "epoch": 1.877477881061709,
      "grad_norm": 1.4201312065124512,
      "learning_rate": 9.644394736842105e-05,
      "loss": 0.6824,
      "step": 33528
    },
    {
      "epoch": 1.877533878373838,
      "grad_norm": 1.239016056060791,
      "learning_rate": 9.644368421052631e-05,
      "loss": 0.4848,
      "step": 33529
    },
    {
      "epoch": 1.877589875685967,
      "grad_norm": 1.2624435424804688,
      "learning_rate": 9.644342105263159e-05,
      "loss": 0.3683,
      "step": 33530
    },
    {
      "epoch": 1.877645872998096,
      "grad_norm": 1.3580355644226074,
      "learning_rate": 9.644315789473685e-05,
      "loss": 0.493,
      "step": 33531
    },
    {
      "epoch": 1.877701870310225,
      "grad_norm": 6.130359172821045,
      "learning_rate": 9.644289473684212e-05,
      "loss": 0.4353,
      "step": 33532
    },
    {
      "epoch": 1.877757867622354,
      "grad_norm": 1.0849494934082031,
      "learning_rate": 9.644263157894737e-05,
      "loss": 0.5841,
      "step": 33533
    },
    {
      "epoch": 1.877813864934483,
      "grad_norm": 1.3526138067245483,
      "learning_rate": 9.644236842105263e-05,
      "loss": 0.5025,
      "step": 33534
    },
    {
      "epoch": 1.8778698622466121,
      "grad_norm": 1.5321365594863892,
      "learning_rate": 9.64421052631579e-05,
      "loss": 0.4215,
      "step": 33535
    },
    {
      "epoch": 1.8779258595587411,
      "grad_norm": 1.3473364114761353,
      "learning_rate": 9.644184210526316e-05,
      "loss": 0.4768,
      "step": 33536
    },
    {
      "epoch": 1.8779818568708702,
      "grad_norm": 2.2610108852386475,
      "learning_rate": 9.644157894736843e-05,
      "loss": 0.4832,
      "step": 33537
    },
    {
      "epoch": 1.8780378541829992,
      "grad_norm": 1.3744839429855347,
      "learning_rate": 9.644131578947368e-05,
      "loss": 0.3333,
      "step": 33538
    },
    {
      "epoch": 1.8780938514951282,
      "grad_norm": 1.569636583328247,
      "learning_rate": 9.644105263157895e-05,
      "loss": 0.4473,
      "step": 33539
    },
    {
      "epoch": 1.8781498488072572,
      "grad_norm": 1.3599305152893066,
      "learning_rate": 9.644078947368421e-05,
      "loss": 0.4995,
      "step": 33540
    },
    {
      "epoch": 1.8782058461193862,
      "grad_norm": 1.6744434833526611,
      "learning_rate": 9.644052631578949e-05,
      "loss": 0.4744,
      "step": 33541
    },
    {
      "epoch": 1.8782618434315153,
      "grad_norm": 1.747742772102356,
      "learning_rate": 9.644026315789475e-05,
      "loss": 0.6652,
      "step": 33542
    },
    {
      "epoch": 1.8783178407436443,
      "grad_norm": 1.4932689666748047,
      "learning_rate": 9.644e-05,
      "loss": 0.461,
      "step": 33543
    },
    {
      "epoch": 1.8783738380557733,
      "grad_norm": 1.44297194480896,
      "learning_rate": 9.643973684210526e-05,
      "loss": 0.4669,
      "step": 33544
    },
    {
      "epoch": 1.8784298353679023,
      "grad_norm": 1.219820499420166,
      "learning_rate": 9.643947368421054e-05,
      "loss": 0.4247,
      "step": 33545
    },
    {
      "epoch": 1.8784858326800313,
      "grad_norm": 1.0487533807754517,
      "learning_rate": 9.64392105263158e-05,
      "loss": 0.3219,
      "step": 33546
    },
    {
      "epoch": 1.8785418299921604,
      "grad_norm": 1.5364068746566772,
      "learning_rate": 9.643894736842106e-05,
      "loss": 0.4218,
      "step": 33547
    },
    {
      "epoch": 1.8785978273042894,
      "grad_norm": 1.849379062652588,
      "learning_rate": 9.643868421052632e-05,
      "loss": 0.5092,
      "step": 33548
    },
    {
      "epoch": 1.8786538246164184,
      "grad_norm": 1.3615421056747437,
      "learning_rate": 9.643842105263159e-05,
      "loss": 0.4887,
      "step": 33549
    },
    {
      "epoch": 1.8787098219285474,
      "grad_norm": 1.3537588119506836,
      "learning_rate": 9.643815789473685e-05,
      "loss": 0.3424,
      "step": 33550
    },
    {
      "epoch": 1.8787658192406764,
      "grad_norm": 1.3269686698913574,
      "learning_rate": 9.643789473684211e-05,
      "loss": 0.4016,
      "step": 33551
    },
    {
      "epoch": 1.8788218165528054,
      "grad_norm": 1.248303771018982,
      "learning_rate": 9.643763157894737e-05,
      "loss": 0.4266,
      "step": 33552
    },
    {
      "epoch": 1.8788778138649345,
      "grad_norm": 1.1459150314331055,
      "learning_rate": 9.643736842105263e-05,
      "loss": 0.5064,
      "step": 33553
    },
    {
      "epoch": 1.8789338111770635,
      "grad_norm": 1.3242383003234863,
      "learning_rate": 9.64371052631579e-05,
      "loss": 0.4664,
      "step": 33554
    },
    {
      "epoch": 1.8789898084891925,
      "grad_norm": 1.175794005393982,
      "learning_rate": 9.643684210526316e-05,
      "loss": 0.6039,
      "step": 33555
    },
    {
      "epoch": 1.8790458058013215,
      "grad_norm": 1.266629695892334,
      "learning_rate": 9.643657894736842e-05,
      "loss": 0.4808,
      "step": 33556
    },
    {
      "epoch": 1.8791018031134505,
      "grad_norm": 1.1700021028518677,
      "learning_rate": 9.643631578947368e-05,
      "loss": 0.4547,
      "step": 33557
    },
    {
      "epoch": 1.8791578004255796,
      "grad_norm": 1.6373611688613892,
      "learning_rate": 9.643605263157896e-05,
      "loss": 0.8479,
      "step": 33558
    },
    {
      "epoch": 1.8792137977377086,
      "grad_norm": 1.2791857719421387,
      "learning_rate": 9.643578947368421e-05,
      "loss": 0.6051,
      "step": 33559
    },
    {
      "epoch": 1.8792697950498376,
      "grad_norm": 2.331620454788208,
      "learning_rate": 9.643552631578949e-05,
      "loss": 0.4451,
      "step": 33560
    },
    {
      "epoch": 1.8793257923619666,
      "grad_norm": 1.241905927658081,
      "learning_rate": 9.643526315789473e-05,
      "loss": 0.4582,
      "step": 33561
    },
    {
      "epoch": 1.8793817896740956,
      "grad_norm": 1.7495547533035278,
      "learning_rate": 9.643500000000001e-05,
      "loss": 0.5253,
      "step": 33562
    },
    {
      "epoch": 1.8794377869862247,
      "grad_norm": 1.262455701828003,
      "learning_rate": 9.643473684210527e-05,
      "loss": 0.4419,
      "step": 33563
    },
    {
      "epoch": 1.8794937842983537,
      "grad_norm": 1.1354089975357056,
      "learning_rate": 9.643447368421054e-05,
      "loss": 0.3548,
      "step": 33564
    },
    {
      "epoch": 1.8795497816104827,
      "grad_norm": 1.259466528892517,
      "learning_rate": 9.64342105263158e-05,
      "loss": 0.432,
      "step": 33565
    },
    {
      "epoch": 1.8796057789226117,
      "grad_norm": 1.0078903436660767,
      "learning_rate": 9.643394736842106e-05,
      "loss": 0.5474,
      "step": 33566
    },
    {
      "epoch": 1.8796617762347407,
      "grad_norm": 1.3523410558700562,
      "learning_rate": 9.643368421052632e-05,
      "loss": 0.5023,
      "step": 33567
    },
    {
      "epoch": 1.8797177735468698,
      "grad_norm": 1.131030797958374,
      "learning_rate": 9.643342105263158e-05,
      "loss": 0.4898,
      "step": 33568
    },
    {
      "epoch": 1.8797737708589988,
      "grad_norm": 1.4040687084197998,
      "learning_rate": 9.643315789473685e-05,
      "loss": 0.4081,
      "step": 33569
    },
    {
      "epoch": 1.8798297681711278,
      "grad_norm": 1.508417010307312,
      "learning_rate": 9.64328947368421e-05,
      "loss": 0.5026,
      "step": 33570
    },
    {
      "epoch": 1.8798857654832568,
      "grad_norm": 1.6678227186203003,
      "learning_rate": 9.643263157894737e-05,
      "loss": 0.4174,
      "step": 33571
    },
    {
      "epoch": 1.8799417627953858,
      "grad_norm": 1.1624722480773926,
      "learning_rate": 9.643236842105263e-05,
      "loss": 0.3999,
      "step": 33572
    },
    {
      "epoch": 1.8799977601075148,
      "grad_norm": 1.4845346212387085,
      "learning_rate": 9.64321052631579e-05,
      "loss": 0.5066,
      "step": 33573
    },
    {
      "epoch": 1.8800537574196439,
      "grad_norm": 1.162156105041504,
      "learning_rate": 9.643184210526316e-05,
      "loss": 0.4714,
      "step": 33574
    },
    {
      "epoch": 1.8801097547317729,
      "grad_norm": 1.4354486465454102,
      "learning_rate": 9.643157894736842e-05,
      "loss": 0.5478,
      "step": 33575
    },
    {
      "epoch": 1.880165752043902,
      "grad_norm": 1.3707231283187866,
      "learning_rate": 9.643131578947368e-05,
      "loss": 0.4741,
      "step": 33576
    },
    {
      "epoch": 1.880221749356031,
      "grad_norm": 1.2601408958435059,
      "learning_rate": 9.643105263157896e-05,
      "loss": 0.5409,
      "step": 33577
    },
    {
      "epoch": 1.88027774666816,
      "grad_norm": 1.170703411102295,
      "learning_rate": 9.643078947368422e-05,
      "loss": 0.5179,
      "step": 33578
    },
    {
      "epoch": 1.880333743980289,
      "grad_norm": 1.0930501222610474,
      "learning_rate": 9.643052631578948e-05,
      "loss": 0.3924,
      "step": 33579
    },
    {
      "epoch": 1.880389741292418,
      "grad_norm": 1.5169048309326172,
      "learning_rate": 9.643026315789474e-05,
      "loss": 0.478,
      "step": 33580
    },
    {
      "epoch": 1.880445738604547,
      "grad_norm": 1.275112509727478,
      "learning_rate": 9.643000000000001e-05,
      "loss": 0.4494,
      "step": 33581
    },
    {
      "epoch": 1.880501735916676,
      "grad_norm": 1.0645058155059814,
      "learning_rate": 9.642973684210527e-05,
      "loss": 0.3672,
      "step": 33582
    },
    {
      "epoch": 1.880557733228805,
      "grad_norm": 1.5216699838638306,
      "learning_rate": 9.642947368421053e-05,
      "loss": 0.406,
      "step": 33583
    },
    {
      "epoch": 1.880613730540934,
      "grad_norm": 1.1006428003311157,
      "learning_rate": 9.642921052631579e-05,
      "loss": 0.3145,
      "step": 33584
    },
    {
      "epoch": 1.880669727853063,
      "grad_norm": 1.1141917705535889,
      "learning_rate": 9.642894736842105e-05,
      "loss": 0.4025,
      "step": 33585
    },
    {
      "epoch": 1.880725725165192,
      "grad_norm": 1.3039439916610718,
      "learning_rate": 9.642868421052632e-05,
      "loss": 0.3697,
      "step": 33586
    },
    {
      "epoch": 1.8807817224773211,
      "grad_norm": 1.3562519550323486,
      "learning_rate": 9.642842105263158e-05,
      "loss": 0.4074,
      "step": 33587
    },
    {
      "epoch": 1.8808377197894501,
      "grad_norm": 1.293365240097046,
      "learning_rate": 9.642815789473684e-05,
      "loss": 0.4704,
      "step": 33588
    },
    {
      "epoch": 1.8808937171015792,
      "grad_norm": 1.014214277267456,
      "learning_rate": 9.64278947368421e-05,
      "loss": 0.3254,
      "step": 33589
    },
    {
      "epoch": 1.8809497144137082,
      "grad_norm": 1.2974371910095215,
      "learning_rate": 9.642763157894737e-05,
      "loss": 0.4649,
      "step": 33590
    },
    {
      "epoch": 1.8810057117258372,
      "grad_norm": 1.162210464477539,
      "learning_rate": 9.642736842105263e-05,
      "loss": 0.3937,
      "step": 33591
    },
    {
      "epoch": 1.8810617090379662,
      "grad_norm": 1.4351285696029663,
      "learning_rate": 9.642710526315791e-05,
      "loss": 0.3878,
      "step": 33592
    },
    {
      "epoch": 1.8811177063500952,
      "grad_norm": 1.082622766494751,
      "learning_rate": 9.642684210526315e-05,
      "loss": 0.3281,
      "step": 33593
    },
    {
      "epoch": 1.8811737036622243,
      "grad_norm": 1.0994406938552856,
      "learning_rate": 9.642657894736843e-05,
      "loss": 0.4361,
      "step": 33594
    },
    {
      "epoch": 1.8812297009743533,
      "grad_norm": 1.3685424327850342,
      "learning_rate": 9.642631578947369e-05,
      "loss": 0.3647,
      "step": 33595
    },
    {
      "epoch": 1.8812856982864823,
      "grad_norm": 1.3866658210754395,
      "learning_rate": 9.642605263157896e-05,
      "loss": 0.4928,
      "step": 33596
    },
    {
      "epoch": 1.8813416955986113,
      "grad_norm": 1.6045397520065308,
      "learning_rate": 9.642578947368422e-05,
      "loss": 0.439,
      "step": 33597
    },
    {
      "epoch": 1.8813976929107403,
      "grad_norm": 1.2860578298568726,
      "learning_rate": 9.642552631578948e-05,
      "loss": 0.4012,
      "step": 33598
    },
    {
      "epoch": 1.8814536902228693,
      "grad_norm": 1.388733983039856,
      "learning_rate": 9.642526315789474e-05,
      "loss": 0.6161,
      "step": 33599
    },
    {
      "epoch": 1.8815096875349984,
      "grad_norm": 1.3261902332305908,
      "learning_rate": 9.642500000000001e-05,
      "loss": 0.5005,
      "step": 33600
    },
    {
      "epoch": 1.8815656848471274,
      "grad_norm": 1.3303438425064087,
      "learning_rate": 9.642473684210527e-05,
      "loss": 0.3799,
      "step": 33601
    },
    {
      "epoch": 1.8816216821592564,
      "grad_norm": 1.154757022857666,
      "learning_rate": 9.642447368421053e-05,
      "loss": 0.4278,
      "step": 33602
    },
    {
      "epoch": 1.8816776794713854,
      "grad_norm": 1.2198487520217896,
      "learning_rate": 9.642421052631579e-05,
      "loss": 0.4519,
      "step": 33603
    },
    {
      "epoch": 1.8817336767835144,
      "grad_norm": 1.1428169012069702,
      "learning_rate": 9.642394736842105e-05,
      "loss": 0.4733,
      "step": 33604
    },
    {
      "epoch": 1.8817896740956435,
      "grad_norm": 1.6005162000656128,
      "learning_rate": 9.642368421052632e-05,
      "loss": 0.4505,
      "step": 33605
    },
    {
      "epoch": 1.8818456714077725,
      "grad_norm": 1.1878433227539062,
      "learning_rate": 9.642342105263158e-05,
      "loss": 0.3358,
      "step": 33606
    },
    {
      "epoch": 1.8819016687199015,
      "grad_norm": 1.1796928644180298,
      "learning_rate": 9.642315789473684e-05,
      "loss": 0.3803,
      "step": 33607
    },
    {
      "epoch": 1.8819576660320303,
      "grad_norm": 1.4113816022872925,
      "learning_rate": 9.64228947368421e-05,
      "loss": 0.3453,
      "step": 33608
    },
    {
      "epoch": 1.8820136633441593,
      "grad_norm": 1.5846142768859863,
      "learning_rate": 9.642263157894738e-05,
      "loss": 0.4481,
      "step": 33609
    },
    {
      "epoch": 1.8820696606562883,
      "grad_norm": 1.2397921085357666,
      "learning_rate": 9.642236842105264e-05,
      "loss": 0.3869,
      "step": 33610
    },
    {
      "epoch": 1.8821256579684174,
      "grad_norm": 1.4078173637390137,
      "learning_rate": 9.64221052631579e-05,
      "loss": 0.4009,
      "step": 33611
    },
    {
      "epoch": 1.8821816552805464,
      "grad_norm": 1.3057299852371216,
      "learning_rate": 9.642184210526316e-05,
      "loss": 0.3502,
      "step": 33612
    },
    {
      "epoch": 1.8822376525926754,
      "grad_norm": 1.337715983390808,
      "learning_rate": 9.642157894736843e-05,
      "loss": 0.374,
      "step": 33613
    },
    {
      "epoch": 1.8822936499048044,
      "grad_norm": 1.0337460041046143,
      "learning_rate": 9.642131578947369e-05,
      "loss": 0.332,
      "step": 33614
    },
    {
      "epoch": 1.8823496472169334,
      "grad_norm": 1.1945194005966187,
      "learning_rate": 9.642105263157896e-05,
      "loss": 0.4678,
      "step": 33615
    },
    {
      "epoch": 1.8824056445290624,
      "grad_norm": 1.33255934715271,
      "learning_rate": 9.642078947368421e-05,
      "loss": 0.6063,
      "step": 33616
    },
    {
      "epoch": 1.8824616418411915,
      "grad_norm": 1.1708885431289673,
      "learning_rate": 9.642052631578948e-05,
      "loss": 0.4045,
      "step": 33617
    },
    {
      "epoch": 1.8825176391533205,
      "grad_norm": 1.1798646450042725,
      "learning_rate": 9.642026315789474e-05,
      "loss": 0.4694,
      "step": 33618
    },
    {
      "epoch": 1.8825736364654495,
      "grad_norm": 1.2565056085586548,
      "learning_rate": 9.642e-05,
      "loss": 0.4368,
      "step": 33619
    },
    {
      "epoch": 1.8826296337775785,
      "grad_norm": 1.0882625579833984,
      "learning_rate": 9.641973684210526e-05,
      "loss": 0.3722,
      "step": 33620
    },
    {
      "epoch": 1.8826856310897075,
      "grad_norm": 1.31728994846344,
      "learning_rate": 9.641947368421052e-05,
      "loss": 0.4617,
      "step": 33621
    },
    {
      "epoch": 1.8827416284018366,
      "grad_norm": 1.18839693069458,
      "learning_rate": 9.64192105263158e-05,
      "loss": 0.5075,
      "step": 33622
    },
    {
      "epoch": 1.8827976257139656,
      "grad_norm": 1.2895548343658447,
      "learning_rate": 9.641894736842105e-05,
      "loss": 0.4276,
      "step": 33623
    },
    {
      "epoch": 1.8828536230260946,
      "grad_norm": 1.4911162853240967,
      "learning_rate": 9.641868421052633e-05,
      "loss": 0.445,
      "step": 33624
    },
    {
      "epoch": 1.8829096203382236,
      "grad_norm": 1.3225804567337036,
      "learning_rate": 9.641842105263157e-05,
      "loss": 0.5065,
      "step": 33625
    },
    {
      "epoch": 1.8829656176503526,
      "grad_norm": 1.3103877305984497,
      "learning_rate": 9.641815789473685e-05,
      "loss": 0.4557,
      "step": 33626
    },
    {
      "epoch": 1.8830216149624817,
      "grad_norm": 1.2465382814407349,
      "learning_rate": 9.64178947368421e-05,
      "loss": 0.3737,
      "step": 33627
    },
    {
      "epoch": 1.8830776122746107,
      "grad_norm": 1.0563464164733887,
      "learning_rate": 9.641763157894738e-05,
      "loss": 0.3558,
      "step": 33628
    },
    {
      "epoch": 1.8831336095867397,
      "grad_norm": 1.0182586908340454,
      "learning_rate": 9.641736842105264e-05,
      "loss": 0.3644,
      "step": 33629
    },
    {
      "epoch": 1.8831896068988687,
      "grad_norm": 1.4864789247512817,
      "learning_rate": 9.64171052631579e-05,
      "loss": 0.5125,
      "step": 33630
    },
    {
      "epoch": 1.8832456042109977,
      "grad_norm": 1.320773720741272,
      "learning_rate": 9.641684210526316e-05,
      "loss": 0.4455,
      "step": 33631
    },
    {
      "epoch": 1.8833016015231268,
      "grad_norm": 1.162069320678711,
      "learning_rate": 9.641657894736843e-05,
      "loss": 0.4008,
      "step": 33632
    },
    {
      "epoch": 1.8833575988352558,
      "grad_norm": 1.2817904949188232,
      "learning_rate": 9.641631578947369e-05,
      "loss": 0.5957,
      "step": 33633
    },
    {
      "epoch": 1.8834135961473848,
      "grad_norm": 1.4701032638549805,
      "learning_rate": 9.641605263157895e-05,
      "loss": 0.4714,
      "step": 33634
    },
    {
      "epoch": 1.8834695934595138,
      "grad_norm": 1.1117165088653564,
      "learning_rate": 9.641578947368421e-05,
      "loss": 0.3716,
      "step": 33635
    },
    {
      "epoch": 1.8835255907716428,
      "grad_norm": 1.3376896381378174,
      "learning_rate": 9.641552631578947e-05,
      "loss": 0.4571,
      "step": 33636
    },
    {
      "epoch": 1.8835815880837719,
      "grad_norm": 1.3433526754379272,
      "learning_rate": 9.641526315789474e-05,
      "loss": 0.492,
      "step": 33637
    },
    {
      "epoch": 1.8836375853959009,
      "grad_norm": 1.2727469205856323,
      "learning_rate": 9.6415e-05,
      "loss": 0.5312,
      "step": 33638
    },
    {
      "epoch": 1.88369358270803,
      "grad_norm": 1.3819561004638672,
      "learning_rate": 9.641473684210526e-05,
      "loss": 0.3681,
      "step": 33639
    },
    {
      "epoch": 1.883749580020159,
      "grad_norm": 3.575835943222046,
      "learning_rate": 9.641447368421052e-05,
      "loss": 0.7263,
      "step": 33640
    },
    {
      "epoch": 1.883805577332288,
      "grad_norm": 1.2515321969985962,
      "learning_rate": 9.64142105263158e-05,
      "loss": 0.3699,
      "step": 33641
    },
    {
      "epoch": 1.883861574644417,
      "grad_norm": 1.6168538331985474,
      "learning_rate": 9.641394736842106e-05,
      "loss": 0.5083,
      "step": 33642
    },
    {
      "epoch": 1.883917571956546,
      "grad_norm": 1.1859713792800903,
      "learning_rate": 9.641368421052632e-05,
      "loss": 0.4167,
      "step": 33643
    },
    {
      "epoch": 1.883973569268675,
      "grad_norm": 1.1871665716171265,
      "learning_rate": 9.641342105263158e-05,
      "loss": 0.4458,
      "step": 33644
    },
    {
      "epoch": 1.884029566580804,
      "grad_norm": 1.9021202325820923,
      "learning_rate": 9.641315789473685e-05,
      "loss": 0.6099,
      "step": 33645
    },
    {
      "epoch": 1.884085563892933,
      "grad_norm": 1.1225911378860474,
      "learning_rate": 9.641289473684211e-05,
      "loss": 0.4321,
      "step": 33646
    },
    {
      "epoch": 1.884141561205062,
      "grad_norm": 1.3472858667373657,
      "learning_rate": 9.641263157894738e-05,
      "loss": 0.4517,
      "step": 33647
    },
    {
      "epoch": 1.884197558517191,
      "grad_norm": 1.1004210710525513,
      "learning_rate": 9.641236842105263e-05,
      "loss": 0.392,
      "step": 33648
    },
    {
      "epoch": 1.88425355582932,
      "grad_norm": 1.0171610116958618,
      "learning_rate": 9.64121052631579e-05,
      "loss": 0.3757,
      "step": 33649
    },
    {
      "epoch": 1.884309553141449,
      "grad_norm": 1.1212135553359985,
      "learning_rate": 9.641184210526316e-05,
      "loss": 0.3261,
      "step": 33650
    },
    {
      "epoch": 1.8843655504535781,
      "grad_norm": 1.3765286207199097,
      "learning_rate": 9.641157894736844e-05,
      "loss": 0.4012,
      "step": 33651
    },
    {
      "epoch": 1.8844215477657071,
      "grad_norm": 1.331695556640625,
      "learning_rate": 9.64113157894737e-05,
      "loss": 0.4816,
      "step": 33652
    },
    {
      "epoch": 1.8844775450778362,
      "grad_norm": 1.4452427625656128,
      "learning_rate": 9.641105263157895e-05,
      "loss": 0.4796,
      "step": 33653
    },
    {
      "epoch": 1.8845335423899652,
      "grad_norm": 1.0920870304107666,
      "learning_rate": 9.641078947368421e-05,
      "loss": 0.3887,
      "step": 33654
    },
    {
      "epoch": 1.8845895397020942,
      "grad_norm": 1.507569432258606,
      "learning_rate": 9.641052631578947e-05,
      "loss": 0.5712,
      "step": 33655
    },
    {
      "epoch": 1.8846455370142232,
      "grad_norm": 1.39818274974823,
      "learning_rate": 9.641026315789475e-05,
      "loss": 0.5199,
      "step": 33656
    },
    {
      "epoch": 1.8847015343263522,
      "grad_norm": 1.2738715410232544,
      "learning_rate": 9.641000000000001e-05,
      "loss": 0.3823,
      "step": 33657
    },
    {
      "epoch": 1.8847575316384813,
      "grad_norm": 1.2792837619781494,
      "learning_rate": 9.640973684210527e-05,
      "loss": 0.5718,
      "step": 33658
    },
    {
      "epoch": 1.8848135289506103,
      "grad_norm": 1.134690284729004,
      "learning_rate": 9.640947368421053e-05,
      "loss": 0.4987,
      "step": 33659
    },
    {
      "epoch": 1.8848695262627393,
      "grad_norm": 1.654996395111084,
      "learning_rate": 9.64092105263158e-05,
      "loss": 0.506,
      "step": 33660
    },
    {
      "epoch": 1.8849255235748683,
      "grad_norm": 1.5433547496795654,
      "learning_rate": 9.640894736842106e-05,
      "loss": 0.4502,
      "step": 33661
    },
    {
      "epoch": 1.8849815208869973,
      "grad_norm": 1.2343541383743286,
      "learning_rate": 9.640868421052632e-05,
      "loss": 0.5119,
      "step": 33662
    },
    {
      "epoch": 1.8850375181991263,
      "grad_norm": 1.1132186651229858,
      "learning_rate": 9.640842105263158e-05,
      "loss": 0.4188,
      "step": 33663
    },
    {
      "epoch": 1.8850935155112554,
      "grad_norm": 1.0647146701812744,
      "learning_rate": 9.640815789473685e-05,
      "loss": 0.3496,
      "step": 33664
    },
    {
      "epoch": 1.8851495128233844,
      "grad_norm": 1.1932886838912964,
      "learning_rate": 9.640789473684211e-05,
      "loss": 0.3946,
      "step": 33665
    },
    {
      "epoch": 1.8852055101355134,
      "grad_norm": 1.2747304439544678,
      "learning_rate": 9.640763157894737e-05,
      "loss": 0.5523,
      "step": 33666
    },
    {
      "epoch": 1.8852615074476424,
      "grad_norm": 1.1313902139663696,
      "learning_rate": 9.640736842105263e-05,
      "loss": 0.3612,
      "step": 33667
    },
    {
      "epoch": 1.8853175047597714,
      "grad_norm": 1.4558825492858887,
      "learning_rate": 9.64071052631579e-05,
      "loss": 0.4322,
      "step": 33668
    },
    {
      "epoch": 1.8853735020719005,
      "grad_norm": 1.230788230895996,
      "learning_rate": 9.640684210526316e-05,
      "loss": 0.4886,
      "step": 33669
    },
    {
      "epoch": 1.8854294993840295,
      "grad_norm": 1.6095679998397827,
      "learning_rate": 9.640657894736844e-05,
      "loss": 0.5022,
      "step": 33670
    },
    {
      "epoch": 1.8854854966961585,
      "grad_norm": 1.2208847999572754,
      "learning_rate": 9.640631578947368e-05,
      "loss": 0.3561,
      "step": 33671
    },
    {
      "epoch": 1.8855414940082875,
      "grad_norm": 1.3562772274017334,
      "learning_rate": 9.640605263157894e-05,
      "loss": 0.5832,
      "step": 33672
    },
    {
      "epoch": 1.8855974913204165,
      "grad_norm": 1.2518420219421387,
      "learning_rate": 9.640578947368422e-05,
      "loss": 0.4103,
      "step": 33673
    },
    {
      "epoch": 1.8856534886325456,
      "grad_norm": 1.2258661985397339,
      "learning_rate": 9.640552631578948e-05,
      "loss": 0.3607,
      "step": 33674
    },
    {
      "epoch": 1.8857094859446746,
      "grad_norm": 1.1410584449768066,
      "learning_rate": 9.640526315789474e-05,
      "loss": 0.3617,
      "step": 33675
    },
    {
      "epoch": 1.8857654832568036,
      "grad_norm": 1.5106875896453857,
      "learning_rate": 9.6405e-05,
      "loss": 0.4099,
      "step": 33676
    },
    {
      "epoch": 1.8858214805689326,
      "grad_norm": 1.303658366203308,
      "learning_rate": 9.640473684210527e-05,
      "loss": 0.4013,
      "step": 33677
    },
    {
      "epoch": 1.8858774778810616,
      "grad_norm": 1.1303516626358032,
      "learning_rate": 9.640447368421053e-05,
      "loss": 0.4688,
      "step": 33678
    },
    {
      "epoch": 1.8859334751931907,
      "grad_norm": 1.2534388303756714,
      "learning_rate": 9.64042105263158e-05,
      "loss": 0.5761,
      "step": 33679
    },
    {
      "epoch": 1.8859894725053197,
      "grad_norm": 1.3551287651062012,
      "learning_rate": 9.640394736842105e-05,
      "loss": 0.5568,
      "step": 33680
    },
    {
      "epoch": 1.8860454698174487,
      "grad_norm": 1.0844429731369019,
      "learning_rate": 9.640368421052632e-05,
      "loss": 0.4228,
      "step": 33681
    },
    {
      "epoch": 1.8861014671295777,
      "grad_norm": 1.2325901985168457,
      "learning_rate": 9.640342105263158e-05,
      "loss": 0.3286,
      "step": 33682
    },
    {
      "epoch": 1.8861574644417067,
      "grad_norm": 1.2700951099395752,
      "learning_rate": 9.640315789473685e-05,
      "loss": 0.5235,
      "step": 33683
    },
    {
      "epoch": 1.8862134617538358,
      "grad_norm": 1.2671318054199219,
      "learning_rate": 9.640289473684211e-05,
      "loss": 0.3488,
      "step": 33684
    },
    {
      "epoch": 1.8862694590659648,
      "grad_norm": 1.3767714500427246,
      "learning_rate": 9.640263157894737e-05,
      "loss": 0.535,
      "step": 33685
    },
    {
      "epoch": 1.8863254563780938,
      "grad_norm": 1.3846298456192017,
      "learning_rate": 9.640236842105263e-05,
      "loss": 0.4825,
      "step": 33686
    },
    {
      "epoch": 1.8863814536902228,
      "grad_norm": 1.3426178693771362,
      "learning_rate": 9.640210526315791e-05,
      "loss": 0.4473,
      "step": 33687
    },
    {
      "epoch": 1.8864374510023518,
      "grad_norm": 1.4409888982772827,
      "learning_rate": 9.640184210526317e-05,
      "loss": 0.456,
      "step": 33688
    },
    {
      "epoch": 1.8864934483144808,
      "grad_norm": 1.2739231586456299,
      "learning_rate": 9.640157894736843e-05,
      "loss": 0.5298,
      "step": 33689
    },
    {
      "epoch": 1.8865494456266099,
      "grad_norm": 1.292602777481079,
      "learning_rate": 9.640131578947369e-05,
      "loss": 0.4456,
      "step": 33690
    },
    {
      "epoch": 1.8866054429387389,
      "grad_norm": 1.380937099456787,
      "learning_rate": 9.640105263157895e-05,
      "loss": 0.4619,
      "step": 33691
    },
    {
      "epoch": 1.886661440250868,
      "grad_norm": 1.3528653383255005,
      "learning_rate": 9.640078947368422e-05,
      "loss": 0.7742,
      "step": 33692
    },
    {
      "epoch": 1.886717437562997,
      "grad_norm": 1.2947343587875366,
      "learning_rate": 9.640052631578948e-05,
      "loss": 0.4531,
      "step": 33693
    },
    {
      "epoch": 1.886773434875126,
      "grad_norm": 1.711308240890503,
      "learning_rate": 9.640026315789474e-05,
      "loss": 0.5245,
      "step": 33694
    },
    {
      "epoch": 1.886829432187255,
      "grad_norm": 1.9468669891357422,
      "learning_rate": 9.64e-05,
      "loss": 0.4723,
      "step": 33695
    },
    {
      "epoch": 1.886885429499384,
      "grad_norm": 1.1978245973587036,
      "learning_rate": 9.639973684210527e-05,
      "loss": 0.4676,
      "step": 33696
    },
    {
      "epoch": 1.886941426811513,
      "grad_norm": 1.2457702159881592,
      "learning_rate": 9.639947368421053e-05,
      "loss": 0.4324,
      "step": 33697
    },
    {
      "epoch": 1.886997424123642,
      "grad_norm": 1.3889532089233398,
      "learning_rate": 9.639921052631579e-05,
      "loss": 0.4338,
      "step": 33698
    },
    {
      "epoch": 1.887053421435771,
      "grad_norm": 3.0156965255737305,
      "learning_rate": 9.639894736842105e-05,
      "loss": 0.407,
      "step": 33699
    },
    {
      "epoch": 1.8871094187479,
      "grad_norm": 2.5987277030944824,
      "learning_rate": 9.639868421052632e-05,
      "loss": 0.4327,
      "step": 33700
    },
    {
      "epoch": 1.887165416060029,
      "grad_norm": 1.2317811250686646,
      "learning_rate": 9.639842105263158e-05,
      "loss": 0.4153,
      "step": 33701
    },
    {
      "epoch": 1.887221413372158,
      "grad_norm": 1.1623668670654297,
      "learning_rate": 9.639815789473686e-05,
      "loss": 0.427,
      "step": 33702
    },
    {
      "epoch": 1.8872774106842871,
      "grad_norm": 1.4816343784332275,
      "learning_rate": 9.63978947368421e-05,
      "loss": 0.3947,
      "step": 33703
    },
    {
      "epoch": 1.8873334079964161,
      "grad_norm": 1.0645761489868164,
      "learning_rate": 9.639763157894738e-05,
      "loss": 0.4634,
      "step": 33704
    },
    {
      "epoch": 1.8873894053085452,
      "grad_norm": 1.100671648979187,
      "learning_rate": 9.639736842105264e-05,
      "loss": 0.3889,
      "step": 33705
    },
    {
      "epoch": 1.8874454026206742,
      "grad_norm": 1.3065826892852783,
      "learning_rate": 9.63971052631579e-05,
      "loss": 0.3958,
      "step": 33706
    },
    {
      "epoch": 1.8875013999328032,
      "grad_norm": 1.1387032270431519,
      "learning_rate": 9.639684210526317e-05,
      "loss": 0.3602,
      "step": 33707
    },
    {
      "epoch": 1.8875573972449322,
      "grad_norm": 1.325945496559143,
      "learning_rate": 9.639657894736842e-05,
      "loss": 0.5236,
      "step": 33708
    },
    {
      "epoch": 1.8876133945570612,
      "grad_norm": 2.0520529747009277,
      "learning_rate": 9.639631578947369e-05,
      "loss": 0.5497,
      "step": 33709
    },
    {
      "epoch": 1.8876693918691902,
      "grad_norm": 2.4582436084747314,
      "learning_rate": 9.639605263157895e-05,
      "loss": 0.526,
      "step": 33710
    },
    {
      "epoch": 1.8877253891813193,
      "grad_norm": 1.141645908355713,
      "learning_rate": 9.639578947368422e-05,
      "loss": 0.4367,
      "step": 33711
    },
    {
      "epoch": 1.8877813864934483,
      "grad_norm": 1.6599266529083252,
      "learning_rate": 9.639552631578948e-05,
      "loss": 0.5174,
      "step": 33712
    },
    {
      "epoch": 1.8878373838055773,
      "grad_norm": 1.2973520755767822,
      "learning_rate": 9.639526315789474e-05,
      "loss": 0.4233,
      "step": 33713
    },
    {
      "epoch": 1.8878933811177063,
      "grad_norm": 1.4130158424377441,
      "learning_rate": 9.6395e-05,
      "loss": 0.4095,
      "step": 33714
    },
    {
      "epoch": 1.8879493784298353,
      "grad_norm": 1.0638209581375122,
      "learning_rate": 9.639473684210527e-05,
      "loss": 0.3541,
      "step": 33715
    },
    {
      "epoch": 1.8880053757419644,
      "grad_norm": 1.392927885055542,
      "learning_rate": 9.639447368421053e-05,
      "loss": 0.4833,
      "step": 33716
    },
    {
      "epoch": 1.8880613730540934,
      "grad_norm": 1.2576876878738403,
      "learning_rate": 9.63942105263158e-05,
      "loss": 0.4412,
      "step": 33717
    },
    {
      "epoch": 1.8881173703662224,
      "grad_norm": 1.2127974033355713,
      "learning_rate": 9.639394736842105e-05,
      "loss": 0.4262,
      "step": 33718
    },
    {
      "epoch": 1.8881733676783514,
      "grad_norm": 1.4530575275421143,
      "learning_rate": 9.639368421052633e-05,
      "loss": 0.4753,
      "step": 33719
    },
    {
      "epoch": 1.8882293649904804,
      "grad_norm": 1.5028923749923706,
      "learning_rate": 9.639342105263159e-05,
      "loss": 0.4946,
      "step": 33720
    },
    {
      "epoch": 1.8882853623026095,
      "grad_norm": 1.3612123727798462,
      "learning_rate": 9.639315789473685e-05,
      "loss": 0.5098,
      "step": 33721
    },
    {
      "epoch": 1.8883413596147385,
      "grad_norm": 1.2808541059494019,
      "learning_rate": 9.63928947368421e-05,
      "loss": 0.3995,
      "step": 33722
    },
    {
      "epoch": 1.8883973569268675,
      "grad_norm": 1.3019863367080688,
      "learning_rate": 9.639263157894737e-05,
      "loss": 0.4485,
      "step": 33723
    },
    {
      "epoch": 1.8884533542389965,
      "grad_norm": 1.5253281593322754,
      "learning_rate": 9.639236842105264e-05,
      "loss": 0.3523,
      "step": 33724
    },
    {
      "epoch": 1.8885093515511255,
      "grad_norm": 1.2433851957321167,
      "learning_rate": 9.63921052631579e-05,
      "loss": 0.3545,
      "step": 33725
    },
    {
      "epoch": 1.8885653488632546,
      "grad_norm": 1.3459885120391846,
      "learning_rate": 9.639184210526316e-05,
      "loss": 0.3981,
      "step": 33726
    },
    {
      "epoch": 1.8886213461753836,
      "grad_norm": 1.536413550376892,
      "learning_rate": 9.639157894736842e-05,
      "loss": 0.5784,
      "step": 33727
    },
    {
      "epoch": 1.8886773434875126,
      "grad_norm": 1.5997157096862793,
      "learning_rate": 9.639131578947369e-05,
      "loss": 0.5754,
      "step": 33728
    },
    {
      "epoch": 1.8887333407996416,
      "grad_norm": 1.4149631261825562,
      "learning_rate": 9.639105263157895e-05,
      "loss": 0.4858,
      "step": 33729
    },
    {
      "epoch": 1.8887893381117706,
      "grad_norm": 1.2886008024215698,
      "learning_rate": 9.639078947368421e-05,
      "loss": 0.3614,
      "step": 33730
    },
    {
      "epoch": 1.8888453354238997,
      "grad_norm": 1.094068169593811,
      "learning_rate": 9.639052631578947e-05,
      "loss": 0.3469,
      "step": 33731
    },
    {
      "epoch": 1.8889013327360287,
      "grad_norm": 1.321854591369629,
      "learning_rate": 9.639026315789474e-05,
      "loss": 0.537,
      "step": 33732
    },
    {
      "epoch": 1.8889573300481577,
      "grad_norm": 1.2715879678726196,
      "learning_rate": 9.639e-05,
      "loss": 0.3926,
      "step": 33733
    },
    {
      "epoch": 1.8890133273602867,
      "grad_norm": 1.2562849521636963,
      "learning_rate": 9.638973684210528e-05,
      "loss": 0.5515,
      "step": 33734
    },
    {
      "epoch": 1.8890693246724157,
      "grad_norm": 1.1653449535369873,
      "learning_rate": 9.638947368421052e-05,
      "loss": 0.3282,
      "step": 33735
    },
    {
      "epoch": 1.8891253219845447,
      "grad_norm": 1.2798750400543213,
      "learning_rate": 9.63892105263158e-05,
      "loss": 0.375,
      "step": 33736
    },
    {
      "epoch": 1.8891813192966738,
      "grad_norm": 1.352647066116333,
      "learning_rate": 9.638894736842106e-05,
      "loss": 0.5331,
      "step": 33737
    },
    {
      "epoch": 1.8892373166088028,
      "grad_norm": 1.0662155151367188,
      "learning_rate": 9.638868421052633e-05,
      "loss": 0.3511,
      "step": 33738
    },
    {
      "epoch": 1.8892933139209318,
      "grad_norm": 1.7140498161315918,
      "learning_rate": 9.638842105263159e-05,
      "loss": 0.628,
      "step": 33739
    },
    {
      "epoch": 1.8893493112330608,
      "grad_norm": 1.380328893661499,
      "learning_rate": 9.638815789473684e-05,
      "loss": 0.3845,
      "step": 33740
    },
    {
      "epoch": 1.8894053085451898,
      "grad_norm": 1.2696951627731323,
      "learning_rate": 9.638789473684211e-05,
      "loss": 0.4262,
      "step": 33741
    },
    {
      "epoch": 1.8894613058573189,
      "grad_norm": 1.3804117441177368,
      "learning_rate": 9.638763157894737e-05,
      "loss": 0.4314,
      "step": 33742
    },
    {
      "epoch": 1.8895173031694479,
      "grad_norm": 1.4543256759643555,
      "learning_rate": 9.638736842105264e-05,
      "loss": 0.3924,
      "step": 33743
    },
    {
      "epoch": 1.889573300481577,
      "grad_norm": 1.3162803649902344,
      "learning_rate": 9.63871052631579e-05,
      "loss": 0.4269,
      "step": 33744
    },
    {
      "epoch": 1.889629297793706,
      "grad_norm": 1.6088993549346924,
      "learning_rate": 9.638684210526316e-05,
      "loss": 0.4083,
      "step": 33745
    },
    {
      "epoch": 1.889685295105835,
      "grad_norm": 1.1304014921188354,
      "learning_rate": 9.638657894736842e-05,
      "loss": 0.3764,
      "step": 33746
    },
    {
      "epoch": 1.889741292417964,
      "grad_norm": 1.3694664239883423,
      "learning_rate": 9.63863157894737e-05,
      "loss": 0.5474,
      "step": 33747
    },
    {
      "epoch": 1.889797289730093,
      "grad_norm": 1.7699271440505981,
      "learning_rate": 9.638605263157895e-05,
      "loss": 0.6235,
      "step": 33748
    },
    {
      "epoch": 1.889853287042222,
      "grad_norm": 1.3306398391723633,
      "learning_rate": 9.638578947368421e-05,
      "loss": 0.5235,
      "step": 33749
    },
    {
      "epoch": 1.889909284354351,
      "grad_norm": 1.003732442855835,
      "learning_rate": 9.638552631578947e-05,
      "loss": 0.413,
      "step": 33750
    },
    {
      "epoch": 1.88996528166648,
      "grad_norm": 1.5121312141418457,
      "learning_rate": 9.638526315789475e-05,
      "loss": 0.5089,
      "step": 33751
    },
    {
      "epoch": 1.890021278978609,
      "grad_norm": 1.380193829536438,
      "learning_rate": 9.6385e-05,
      "loss": 0.4384,
      "step": 33752
    },
    {
      "epoch": 1.890077276290738,
      "grad_norm": 1.448320746421814,
      "learning_rate": 9.638473684210527e-05,
      "loss": 0.4362,
      "step": 33753
    },
    {
      "epoch": 1.890133273602867,
      "grad_norm": 1.176749348640442,
      "learning_rate": 9.638447368421053e-05,
      "loss": 0.4464,
      "step": 33754
    },
    {
      "epoch": 1.8901892709149961,
      "grad_norm": 1.3431552648544312,
      "learning_rate": 9.63842105263158e-05,
      "loss": 0.4593,
      "step": 33755
    },
    {
      "epoch": 1.8902452682271251,
      "grad_norm": 1.3157732486724854,
      "learning_rate": 9.638394736842106e-05,
      "loss": 0.5824,
      "step": 33756
    },
    {
      "epoch": 1.8903012655392541,
      "grad_norm": 1.3365122079849243,
      "learning_rate": 9.638368421052632e-05,
      "loss": 0.4323,
      "step": 33757
    },
    {
      "epoch": 1.8903572628513832,
      "grad_norm": 1.0848917961120605,
      "learning_rate": 9.638342105263158e-05,
      "loss": 0.4097,
      "step": 33758
    },
    {
      "epoch": 1.8904132601635122,
      "grad_norm": 1.2415964603424072,
      "learning_rate": 9.638315789473684e-05,
      "loss": 0.5244,
      "step": 33759
    },
    {
      "epoch": 1.8904692574756412,
      "grad_norm": 1.2789194583892822,
      "learning_rate": 9.638289473684211e-05,
      "loss": 0.4942,
      "step": 33760
    },
    {
      "epoch": 1.8905252547877702,
      "grad_norm": 1.1298048496246338,
      "learning_rate": 9.638263157894737e-05,
      "loss": 0.3647,
      "step": 33761
    },
    {
      "epoch": 1.8905812520998992,
      "grad_norm": 1.3419326543807983,
      "learning_rate": 9.638236842105264e-05,
      "loss": 0.4633,
      "step": 33762
    },
    {
      "epoch": 1.8906372494120283,
      "grad_norm": 1.2517443895339966,
      "learning_rate": 9.638210526315789e-05,
      "loss": 0.3826,
      "step": 33763
    },
    {
      "epoch": 1.8906932467241573,
      "grad_norm": 1.4213787317276,
      "learning_rate": 9.638184210526316e-05,
      "loss": 0.4486,
      "step": 33764
    },
    {
      "epoch": 1.8907492440362863,
      "grad_norm": 1.257437825202942,
      "learning_rate": 9.638157894736842e-05,
      "loss": 0.4895,
      "step": 33765
    },
    {
      "epoch": 1.8908052413484153,
      "grad_norm": 1.1475830078125,
      "learning_rate": 9.63813157894737e-05,
      "loss": 0.4839,
      "step": 33766
    },
    {
      "epoch": 1.8908612386605443,
      "grad_norm": 1.2176473140716553,
      "learning_rate": 9.638105263157894e-05,
      "loss": 0.4739,
      "step": 33767
    },
    {
      "epoch": 1.8909172359726734,
      "grad_norm": 1.2745640277862549,
      "learning_rate": 9.638078947368422e-05,
      "loss": 0.4743,
      "step": 33768
    },
    {
      "epoch": 1.8909732332848024,
      "grad_norm": 1.2797621488571167,
      "learning_rate": 9.638052631578948e-05,
      "loss": 0.5023,
      "step": 33769
    },
    {
      "epoch": 1.8910292305969314,
      "grad_norm": 1.512215495109558,
      "learning_rate": 9.638026315789475e-05,
      "loss": 0.5836,
      "step": 33770
    },
    {
      "epoch": 1.8910852279090604,
      "grad_norm": 1.1744507551193237,
      "learning_rate": 9.638000000000001e-05,
      "loss": 0.4569,
      "step": 33771
    },
    {
      "epoch": 1.8911412252211894,
      "grad_norm": 1.1797175407409668,
      "learning_rate": 9.637973684210527e-05,
      "loss": 0.4027,
      "step": 33772
    },
    {
      "epoch": 1.8911972225333185,
      "grad_norm": 1.2333705425262451,
      "learning_rate": 9.637947368421053e-05,
      "loss": 0.3951,
      "step": 33773
    },
    {
      "epoch": 1.8912532198454475,
      "grad_norm": 1.27533757686615,
      "learning_rate": 9.63792105263158e-05,
      "loss": 0.5331,
      "step": 33774
    },
    {
      "epoch": 1.8913092171575765,
      "grad_norm": 1.1606650352478027,
      "learning_rate": 9.637894736842106e-05,
      "loss": 0.3501,
      "step": 33775
    },
    {
      "epoch": 1.8913652144697055,
      "grad_norm": 1.5292431116104126,
      "learning_rate": 9.637868421052632e-05,
      "loss": 0.494,
      "step": 33776
    },
    {
      "epoch": 1.8914212117818345,
      "grad_norm": 1.5637319087982178,
      "learning_rate": 9.637842105263158e-05,
      "loss": 0.4728,
      "step": 33777
    },
    {
      "epoch": 1.8914772090939636,
      "grad_norm": 1.486598253250122,
      "learning_rate": 9.637815789473684e-05,
      "loss": 0.48,
      "step": 33778
    },
    {
      "epoch": 1.8915332064060926,
      "grad_norm": 1.3211822509765625,
      "learning_rate": 9.637789473684211e-05,
      "loss": 0.4402,
      "step": 33779
    },
    {
      "epoch": 1.8915892037182216,
      "grad_norm": 1.488054871559143,
      "learning_rate": 9.637763157894737e-05,
      "loss": 0.4652,
      "step": 33780
    },
    {
      "epoch": 1.8916452010303506,
      "grad_norm": 2.684025764465332,
      "learning_rate": 9.637736842105263e-05,
      "loss": 0.4414,
      "step": 33781
    },
    {
      "epoch": 1.8917011983424796,
      "grad_norm": 1.5047484636306763,
      "learning_rate": 9.637710526315789e-05,
      "loss": 0.629,
      "step": 33782
    },
    {
      "epoch": 1.8917571956546086,
      "grad_norm": 1.1654138565063477,
      "learning_rate": 9.637684210526317e-05,
      "loss": 0.4215,
      "step": 33783
    },
    {
      "epoch": 1.8918131929667377,
      "grad_norm": 1.5539867877960205,
      "learning_rate": 9.637657894736843e-05,
      "loss": 0.4363,
      "step": 33784
    },
    {
      "epoch": 1.8918691902788667,
      "grad_norm": 1.1967475414276123,
      "learning_rate": 9.637631578947369e-05,
      "loss": 0.4965,
      "step": 33785
    },
    {
      "epoch": 1.8919251875909957,
      "grad_norm": 1.3217979669570923,
      "learning_rate": 9.637605263157895e-05,
      "loss": 0.4689,
      "step": 33786
    },
    {
      "epoch": 1.8919811849031247,
      "grad_norm": 1.4112365245819092,
      "learning_rate": 9.637578947368422e-05,
      "loss": 0.411,
      "step": 33787
    },
    {
      "epoch": 1.8920371822152537,
      "grad_norm": 1.2487647533416748,
      "learning_rate": 9.637552631578948e-05,
      "loss": 0.4502,
      "step": 33788
    },
    {
      "epoch": 1.8920931795273828,
      "grad_norm": 1.1966150999069214,
      "learning_rate": 9.637526315789475e-05,
      "loss": 0.4188,
      "step": 33789
    },
    {
      "epoch": 1.8921491768395118,
      "grad_norm": 1.0316742658615112,
      "learning_rate": 9.6375e-05,
      "loss": 0.3524,
      "step": 33790
    },
    {
      "epoch": 1.8922051741516408,
      "grad_norm": 1.2423183917999268,
      "learning_rate": 9.637473684210527e-05,
      "loss": 0.3844,
      "step": 33791
    },
    {
      "epoch": 1.8922611714637698,
      "grad_norm": 1.5044188499450684,
      "learning_rate": 9.637447368421053e-05,
      "loss": 0.4557,
      "step": 33792
    },
    {
      "epoch": 1.8923171687758988,
      "grad_norm": 1.5450776815414429,
      "learning_rate": 9.637421052631579e-05,
      "loss": 0.5864,
      "step": 33793
    },
    {
      "epoch": 1.8923731660880279,
      "grad_norm": 1.2004612684249878,
      "learning_rate": 9.637394736842106e-05,
      "loss": 0.4908,
      "step": 33794
    },
    {
      "epoch": 1.8924291634001569,
      "grad_norm": 1.4424163103103638,
      "learning_rate": 9.637368421052631e-05,
      "loss": 0.4344,
      "step": 33795
    },
    {
      "epoch": 1.892485160712286,
      "grad_norm": 1.1905769109725952,
      "learning_rate": 9.637342105263158e-05,
      "loss": 0.3851,
      "step": 33796
    },
    {
      "epoch": 1.892541158024415,
      "grad_norm": 1.1438097953796387,
      "learning_rate": 9.637315789473684e-05,
      "loss": 0.5024,
      "step": 33797
    },
    {
      "epoch": 1.892597155336544,
      "grad_norm": 1.3093501329421997,
      "learning_rate": 9.637289473684212e-05,
      "loss": 0.5167,
      "step": 33798
    },
    {
      "epoch": 1.892653152648673,
      "grad_norm": 1.4427118301391602,
      "learning_rate": 9.637263157894738e-05,
      "loss": 0.4535,
      "step": 33799
    },
    {
      "epoch": 1.892709149960802,
      "grad_norm": 1.4023902416229248,
      "learning_rate": 9.637236842105264e-05,
      "loss": 0.4296,
      "step": 33800
    },
    {
      "epoch": 1.892765147272931,
      "grad_norm": 1.0676928758621216,
      "learning_rate": 9.63721052631579e-05,
      "loss": 0.4416,
      "step": 33801
    },
    {
      "epoch": 1.89282114458506,
      "grad_norm": 1.659006118774414,
      "learning_rate": 9.637184210526317e-05,
      "loss": 0.3841,
      "step": 33802
    },
    {
      "epoch": 1.892877141897189,
      "grad_norm": 1.1680701971054077,
      "learning_rate": 9.637157894736843e-05,
      "loss": 0.3974,
      "step": 33803
    },
    {
      "epoch": 1.892933139209318,
      "grad_norm": 1.1408295631408691,
      "learning_rate": 9.637131578947369e-05,
      "loss": 0.5063,
      "step": 33804
    },
    {
      "epoch": 1.892989136521447,
      "grad_norm": 1.110579490661621,
      "learning_rate": 9.637105263157895e-05,
      "loss": 0.4753,
      "step": 33805
    },
    {
      "epoch": 1.893045133833576,
      "grad_norm": 1.7396602630615234,
      "learning_rate": 9.637078947368422e-05,
      "loss": 0.5928,
      "step": 33806
    },
    {
      "epoch": 1.893101131145705,
      "grad_norm": 3.463256597518921,
      "learning_rate": 9.637052631578948e-05,
      "loss": 0.6267,
      "step": 33807
    },
    {
      "epoch": 1.8931571284578341,
      "grad_norm": 1.2343287467956543,
      "learning_rate": 9.637026315789474e-05,
      "loss": 0.5322,
      "step": 33808
    },
    {
      "epoch": 1.8932131257699631,
      "grad_norm": 1.302688717842102,
      "learning_rate": 9.637e-05,
      "loss": 0.5278,
      "step": 33809
    },
    {
      "epoch": 1.8932691230820922,
      "grad_norm": 1.4164820909500122,
      "learning_rate": 9.636973684210526e-05,
      "loss": 0.5605,
      "step": 33810
    },
    {
      "epoch": 1.8933251203942212,
      "grad_norm": 1.3477282524108887,
      "learning_rate": 9.636947368421053e-05,
      "loss": 0.5601,
      "step": 33811
    },
    {
      "epoch": 1.8933811177063502,
      "grad_norm": 1.483593463897705,
      "learning_rate": 9.63692105263158e-05,
      "loss": 0.5226,
      "step": 33812
    },
    {
      "epoch": 1.8934371150184792,
      "grad_norm": 1.3437167406082153,
      "learning_rate": 9.636894736842105e-05,
      "loss": 0.4311,
      "step": 33813
    },
    {
      "epoch": 1.8934931123306082,
      "grad_norm": 1.099973440170288,
      "learning_rate": 9.636868421052631e-05,
      "loss": 0.3685,
      "step": 33814
    },
    {
      "epoch": 1.8935491096427373,
      "grad_norm": 1.405934453010559,
      "learning_rate": 9.636842105263159e-05,
      "loss": 0.4477,
      "step": 33815
    },
    {
      "epoch": 1.8936051069548663,
      "grad_norm": 1.0704610347747803,
      "learning_rate": 9.636815789473685e-05,
      "loss": 0.3752,
      "step": 33816
    },
    {
      "epoch": 1.8936611042669953,
      "grad_norm": 1.0581151247024536,
      "learning_rate": 9.636789473684212e-05,
      "loss": 0.2809,
      "step": 33817
    },
    {
      "epoch": 1.8937171015791243,
      "grad_norm": 1.4210178852081299,
      "learning_rate": 9.636763157894737e-05,
      "loss": 0.4863,
      "step": 33818
    },
    {
      "epoch": 1.8937730988912533,
      "grad_norm": 1.2307201623916626,
      "learning_rate": 9.636736842105264e-05,
      "loss": 0.4447,
      "step": 33819
    },
    {
      "epoch": 1.8938290962033824,
      "grad_norm": 1.2953604459762573,
      "learning_rate": 9.63671052631579e-05,
      "loss": 0.3506,
      "step": 33820
    },
    {
      "epoch": 1.8938850935155114,
      "grad_norm": 1.4039552211761475,
      "learning_rate": 9.636684210526317e-05,
      "loss": 0.4966,
      "step": 33821
    },
    {
      "epoch": 1.8939410908276404,
      "grad_norm": 1.3947612047195435,
      "learning_rate": 9.636657894736842e-05,
      "loss": 0.6898,
      "step": 33822
    },
    {
      "epoch": 1.8939970881397694,
      "grad_norm": 1.1440707445144653,
      "learning_rate": 9.636631578947369e-05,
      "loss": 0.3723,
      "step": 33823
    },
    {
      "epoch": 1.8940530854518984,
      "grad_norm": 1.3068008422851562,
      "learning_rate": 9.636605263157895e-05,
      "loss": 0.4593,
      "step": 33824
    },
    {
      "epoch": 1.8941090827640275,
      "grad_norm": 1.2232640981674194,
      "learning_rate": 9.636578947368422e-05,
      "loss": 0.4409,
      "step": 33825
    },
    {
      "epoch": 1.8941650800761565,
      "grad_norm": 5.0967278480529785,
      "learning_rate": 9.636552631578948e-05,
      "loss": 0.6335,
      "step": 33826
    },
    {
      "epoch": 1.8942210773882855,
      "grad_norm": 1.1405287981033325,
      "learning_rate": 9.636526315789473e-05,
      "loss": 0.5729,
      "step": 33827
    },
    {
      "epoch": 1.8942770747004145,
      "grad_norm": 1.2232614755630493,
      "learning_rate": 9.6365e-05,
      "loss": 0.4041,
      "step": 33828
    },
    {
      "epoch": 1.8943330720125435,
      "grad_norm": 1.1981899738311768,
      "learning_rate": 9.636473684210526e-05,
      "loss": 0.4102,
      "step": 33829
    },
    {
      "epoch": 1.8943890693246725,
      "grad_norm": 1.337570309638977,
      "learning_rate": 9.636447368421054e-05,
      "loss": 0.5812,
      "step": 33830
    },
    {
      "epoch": 1.8944450666368016,
      "grad_norm": 1.2408257722854614,
      "learning_rate": 9.63642105263158e-05,
      "loss": 0.4288,
      "step": 33831
    },
    {
      "epoch": 1.8945010639489306,
      "grad_norm": 1.336145281791687,
      "learning_rate": 9.636394736842106e-05,
      "loss": 0.3697,
      "step": 33832
    },
    {
      "epoch": 1.8945570612610596,
      "grad_norm": 1.1942615509033203,
      "learning_rate": 9.636368421052632e-05,
      "loss": 0.4404,
      "step": 33833
    },
    {
      "epoch": 1.8946130585731886,
      "grad_norm": 1.264601230621338,
      "learning_rate": 9.636342105263159e-05,
      "loss": 0.4135,
      "step": 33834
    },
    {
      "epoch": 1.8946690558853176,
      "grad_norm": 1.4116460084915161,
      "learning_rate": 9.636315789473685e-05,
      "loss": 0.5765,
      "step": 33835
    },
    {
      "epoch": 1.8947250531974467,
      "grad_norm": 1.257911205291748,
      "learning_rate": 9.636289473684211e-05,
      "loss": 0.3804,
      "step": 33836
    },
    {
      "epoch": 1.8947810505095757,
      "grad_norm": 1.2506638765335083,
      "learning_rate": 9.636263157894737e-05,
      "loss": 0.5428,
      "step": 33837
    },
    {
      "epoch": 1.8948370478217047,
      "grad_norm": 1.0231753587722778,
      "learning_rate": 9.636236842105264e-05,
      "loss": 0.3928,
      "step": 33838
    },
    {
      "epoch": 1.8948930451338337,
      "grad_norm": 1.066957950592041,
      "learning_rate": 9.63621052631579e-05,
      "loss": 0.3921,
      "step": 33839
    },
    {
      "epoch": 1.8949490424459627,
      "grad_norm": 1.5812498331069946,
      "learning_rate": 9.636184210526316e-05,
      "loss": 0.5301,
      "step": 33840
    },
    {
      "epoch": 1.8950050397580918,
      "grad_norm": 1.1676290035247803,
      "learning_rate": 9.636157894736842e-05,
      "loss": 0.3537,
      "step": 33841
    },
    {
      "epoch": 1.8950610370702208,
      "grad_norm": 1.1587707996368408,
      "learning_rate": 9.63613157894737e-05,
      "loss": 0.4013,
      "step": 33842
    },
    {
      "epoch": 1.8951170343823498,
      "grad_norm": 1.092280626296997,
      "learning_rate": 9.636105263157895e-05,
      "loss": 0.3923,
      "step": 33843
    },
    {
      "epoch": 1.8951730316944788,
      "grad_norm": 1.4090932607650757,
      "learning_rate": 9.636078947368421e-05,
      "loss": 0.5199,
      "step": 33844
    },
    {
      "epoch": 1.8952290290066078,
      "grad_norm": 1.1344330310821533,
      "learning_rate": 9.636052631578947e-05,
      "loss": 0.367,
      "step": 33845
    },
    {
      "epoch": 1.8952850263187369,
      "grad_norm": 1.2157957553863525,
      "learning_rate": 9.636026315789473e-05,
      "loss": 0.4706,
      "step": 33846
    },
    {
      "epoch": 1.8953410236308659,
      "grad_norm": 1.0233741998672485,
      "learning_rate": 9.636e-05,
      "loss": 0.3999,
      "step": 33847
    },
    {
      "epoch": 1.895397020942995,
      "grad_norm": 1.2454618215560913,
      "learning_rate": 9.635973684210527e-05,
      "loss": 0.4491,
      "step": 33848
    },
    {
      "epoch": 1.895453018255124,
      "grad_norm": 1.4090995788574219,
      "learning_rate": 9.635947368421054e-05,
      "loss": 0.5261,
      "step": 33849
    },
    {
      "epoch": 1.895509015567253,
      "grad_norm": 1.2223213911056519,
      "learning_rate": 9.635921052631578e-05,
      "loss": 0.3924,
      "step": 33850
    },
    {
      "epoch": 1.895565012879382,
      "grad_norm": 1.1465595960617065,
      "learning_rate": 9.635894736842106e-05,
      "loss": 0.4231,
      "step": 33851
    },
    {
      "epoch": 1.895621010191511,
      "grad_norm": 1.3092809915542603,
      "learning_rate": 9.635868421052632e-05,
      "loss": 0.4809,
      "step": 33852
    },
    {
      "epoch": 1.89567700750364,
      "grad_norm": 1.2736303806304932,
      "learning_rate": 9.635842105263159e-05,
      "loss": 0.3895,
      "step": 33853
    },
    {
      "epoch": 1.895733004815769,
      "grad_norm": 1.46797513961792,
      "learning_rate": 9.635815789473685e-05,
      "loss": 0.6356,
      "step": 33854
    },
    {
      "epoch": 1.895789002127898,
      "grad_norm": 13.827521324157715,
      "learning_rate": 9.635789473684211e-05,
      "loss": 0.5151,
      "step": 33855
    },
    {
      "epoch": 1.8958449994400268,
      "grad_norm": 1.1565792560577393,
      "learning_rate": 9.635763157894737e-05,
      "loss": 0.3499,
      "step": 33856
    },
    {
      "epoch": 1.8959009967521558,
      "grad_norm": 1.4003320932388306,
      "learning_rate": 9.635736842105264e-05,
      "loss": 0.5926,
      "step": 33857
    },
    {
      "epoch": 1.8959569940642849,
      "grad_norm": 1.3092973232269287,
      "learning_rate": 9.63571052631579e-05,
      "loss": 0.4114,
      "step": 33858
    },
    {
      "epoch": 1.8960129913764139,
      "grad_norm": 1.2511399984359741,
      "learning_rate": 9.635684210526316e-05,
      "loss": 0.404,
      "step": 33859
    },
    {
      "epoch": 1.896068988688543,
      "grad_norm": 1.4181357622146606,
      "learning_rate": 9.635657894736842e-05,
      "loss": 0.4621,
      "step": 33860
    },
    {
      "epoch": 1.896124986000672,
      "grad_norm": 1.2863298654556274,
      "learning_rate": 9.635631578947368e-05,
      "loss": 0.4247,
      "step": 33861
    },
    {
      "epoch": 1.896180983312801,
      "grad_norm": 1.435805320739746,
      "learning_rate": 9.635605263157896e-05,
      "loss": 0.6558,
      "step": 33862
    },
    {
      "epoch": 1.89623698062493,
      "grad_norm": 1.1426249742507935,
      "learning_rate": 9.635578947368422e-05,
      "loss": 0.4401,
      "step": 33863
    },
    {
      "epoch": 1.896292977937059,
      "grad_norm": 1.1203047037124634,
      "learning_rate": 9.635552631578948e-05,
      "loss": 0.4032,
      "step": 33864
    },
    {
      "epoch": 1.896348975249188,
      "grad_norm": 1.984034776687622,
      "learning_rate": 9.635526315789474e-05,
      "loss": 0.5787,
      "step": 33865
    },
    {
      "epoch": 1.896404972561317,
      "grad_norm": 1.1878093481063843,
      "learning_rate": 9.635500000000001e-05,
      "loss": 0.5409,
      "step": 33866
    },
    {
      "epoch": 1.896460969873446,
      "grad_norm": 1.182365894317627,
      "learning_rate": 9.635473684210527e-05,
      "loss": 0.4109,
      "step": 33867
    },
    {
      "epoch": 1.896516967185575,
      "grad_norm": 1.4006723165512085,
      "learning_rate": 9.635447368421053e-05,
      "loss": 0.5323,
      "step": 33868
    },
    {
      "epoch": 1.896572964497704,
      "grad_norm": 1.240382432937622,
      "learning_rate": 9.635421052631579e-05,
      "loss": 0.3969,
      "step": 33869
    },
    {
      "epoch": 1.896628961809833,
      "grad_norm": 1.2382484674453735,
      "learning_rate": 9.635394736842106e-05,
      "loss": 0.4247,
      "step": 33870
    },
    {
      "epoch": 1.896684959121962,
      "grad_norm": 1.303580641746521,
      "learning_rate": 9.635368421052632e-05,
      "loss": 0.5061,
      "step": 33871
    },
    {
      "epoch": 1.8967409564340911,
      "grad_norm": 1.0175176858901978,
      "learning_rate": 9.63534210526316e-05,
      "loss": 0.4051,
      "step": 33872
    },
    {
      "epoch": 1.8967969537462201,
      "grad_norm": 1.1349375247955322,
      "learning_rate": 9.635315789473684e-05,
      "loss": 0.3357,
      "step": 33873
    },
    {
      "epoch": 1.8968529510583492,
      "grad_norm": 1.1460795402526855,
      "learning_rate": 9.635289473684211e-05,
      "loss": 0.5061,
      "step": 33874
    },
    {
      "epoch": 1.8969089483704782,
      "grad_norm": 1.1077089309692383,
      "learning_rate": 9.635263157894737e-05,
      "loss": 0.4761,
      "step": 33875
    },
    {
      "epoch": 1.8969649456826072,
      "grad_norm": 1.2439892292022705,
      "learning_rate": 9.635236842105265e-05,
      "loss": 0.6103,
      "step": 33876
    },
    {
      "epoch": 1.8970209429947362,
      "grad_norm": 1.1198794841766357,
      "learning_rate": 9.635210526315789e-05,
      "loss": 0.377,
      "step": 33877
    },
    {
      "epoch": 1.8970769403068652,
      "grad_norm": 1.1713459491729736,
      "learning_rate": 9.635184210526315e-05,
      "loss": 0.356,
      "step": 33878
    },
    {
      "epoch": 1.8971329376189943,
      "grad_norm": 1.2828829288482666,
      "learning_rate": 9.635157894736843e-05,
      "loss": 0.4217,
      "step": 33879
    },
    {
      "epoch": 1.8971889349311233,
      "grad_norm": 1.3622703552246094,
      "learning_rate": 9.635131578947369e-05,
      "loss": 0.4352,
      "step": 33880
    },
    {
      "epoch": 1.8972449322432523,
      "grad_norm": 1.1960508823394775,
      "learning_rate": 9.635105263157896e-05,
      "loss": 0.3662,
      "step": 33881
    },
    {
      "epoch": 1.8973009295553813,
      "grad_norm": 1.1093484163284302,
      "learning_rate": 9.63507894736842e-05,
      "loss": 0.4225,
      "step": 33882
    },
    {
      "epoch": 1.8973569268675103,
      "grad_norm": 1.1489733457565308,
      "learning_rate": 9.635052631578948e-05,
      "loss": 0.3717,
      "step": 33883
    },
    {
      "epoch": 1.8974129241796394,
      "grad_norm": 1.3215217590332031,
      "learning_rate": 9.635026315789474e-05,
      "loss": 0.5286,
      "step": 33884
    },
    {
      "epoch": 1.8974689214917684,
      "grad_norm": 1.3089959621429443,
      "learning_rate": 9.635000000000001e-05,
      "loss": 0.5382,
      "step": 33885
    },
    {
      "epoch": 1.8975249188038974,
      "grad_norm": 1.2140651941299438,
      "learning_rate": 9.634973684210527e-05,
      "loss": 0.4458,
      "step": 33886
    },
    {
      "epoch": 1.8975809161160264,
      "grad_norm": 1.1865999698638916,
      "learning_rate": 9.634947368421053e-05,
      "loss": 0.4096,
      "step": 33887
    },
    {
      "epoch": 1.8976369134281554,
      "grad_norm": 1.132714867591858,
      "learning_rate": 9.634921052631579e-05,
      "loss": 0.4003,
      "step": 33888
    },
    {
      "epoch": 1.8976929107402845,
      "grad_norm": 1.1732120513916016,
      "learning_rate": 9.634894736842106e-05,
      "loss": 0.3859,
      "step": 33889
    },
    {
      "epoch": 1.8977489080524135,
      "grad_norm": 1.196721076965332,
      "learning_rate": 9.634868421052632e-05,
      "loss": 0.4944,
      "step": 33890
    },
    {
      "epoch": 1.8978049053645425,
      "grad_norm": 2.0257256031036377,
      "learning_rate": 9.634842105263158e-05,
      "loss": 0.3944,
      "step": 33891
    },
    {
      "epoch": 1.8978609026766715,
      "grad_norm": 1.5345579385757446,
      "learning_rate": 9.634815789473684e-05,
      "loss": 0.4686,
      "step": 33892
    },
    {
      "epoch": 1.8979168999888005,
      "grad_norm": 1.37324857711792,
      "learning_rate": 9.634789473684212e-05,
      "loss": 0.3133,
      "step": 33893
    },
    {
      "epoch": 1.8979728973009296,
      "grad_norm": 1.2435295581817627,
      "learning_rate": 9.634763157894738e-05,
      "loss": 0.421,
      "step": 33894
    },
    {
      "epoch": 1.8980288946130586,
      "grad_norm": 1.1709814071655273,
      "learning_rate": 9.634736842105264e-05,
      "loss": 0.3355,
      "step": 33895
    },
    {
      "epoch": 1.8980848919251876,
      "grad_norm": 1.3459051847457886,
      "learning_rate": 9.63471052631579e-05,
      "loss": 0.432,
      "step": 33896
    },
    {
      "epoch": 1.8981408892373166,
      "grad_norm": 1.2831532955169678,
      "learning_rate": 9.634684210526315e-05,
      "loss": 0.5152,
      "step": 33897
    },
    {
      "epoch": 1.8981968865494456,
      "grad_norm": 2.883971691131592,
      "learning_rate": 9.634657894736843e-05,
      "loss": 0.3491,
      "step": 33898
    },
    {
      "epoch": 1.8982528838615746,
      "grad_norm": 1.1753042936325073,
      "learning_rate": 9.634631578947369e-05,
      "loss": 0.508,
      "step": 33899
    },
    {
      "epoch": 1.8983088811737037,
      "grad_norm": 1.3692599534988403,
      "learning_rate": 9.634605263157895e-05,
      "loss": 0.4621,
      "step": 33900
    },
    {
      "epoch": 1.8983648784858327,
      "grad_norm": 1.031097650527954,
      "learning_rate": 9.634578947368421e-05,
      "loss": 0.4122,
      "step": 33901
    },
    {
      "epoch": 1.8984208757979617,
      "grad_norm": 1.2313507795333862,
      "learning_rate": 9.634552631578948e-05,
      "loss": 0.3716,
      "step": 33902
    },
    {
      "epoch": 1.8984768731100907,
      "grad_norm": 1.2132445573806763,
      "learning_rate": 9.634526315789474e-05,
      "loss": 0.5237,
      "step": 33903
    },
    {
      "epoch": 1.8985328704222197,
      "grad_norm": 1.228338599205017,
      "learning_rate": 9.634500000000001e-05,
      "loss": 0.44,
      "step": 33904
    },
    {
      "epoch": 1.8985888677343488,
      "grad_norm": 1.3325189352035522,
      "learning_rate": 9.634473684210526e-05,
      "loss": 0.4337,
      "step": 33905
    },
    {
      "epoch": 1.8986448650464778,
      "grad_norm": 1.539389967918396,
      "learning_rate": 9.634447368421053e-05,
      "loss": 0.5252,
      "step": 33906
    },
    {
      "epoch": 1.8987008623586068,
      "grad_norm": 1.1752235889434814,
      "learning_rate": 9.634421052631579e-05,
      "loss": 0.4842,
      "step": 33907
    },
    {
      "epoch": 1.8987568596707358,
      "grad_norm": 1.2581361532211304,
      "learning_rate": 9.634394736842107e-05,
      "loss": 0.4375,
      "step": 33908
    },
    {
      "epoch": 1.8988128569828648,
      "grad_norm": 1.5986545085906982,
      "learning_rate": 9.634368421052633e-05,
      "loss": 0.6611,
      "step": 33909
    },
    {
      "epoch": 1.8988688542949939,
      "grad_norm": 1.125227689743042,
      "learning_rate": 9.634342105263159e-05,
      "loss": 0.4173,
      "step": 33910
    },
    {
      "epoch": 1.8989248516071229,
      "grad_norm": 1.1550233364105225,
      "learning_rate": 9.634315789473685e-05,
      "loss": 0.4341,
      "step": 33911
    },
    {
      "epoch": 1.898980848919252,
      "grad_norm": 0.9847545027732849,
      "learning_rate": 9.634289473684212e-05,
      "loss": 0.3214,
      "step": 33912
    },
    {
      "epoch": 1.899036846231381,
      "grad_norm": 1.723526954650879,
      "learning_rate": 9.634263157894738e-05,
      "loss": 0.4387,
      "step": 33913
    },
    {
      "epoch": 1.89909284354351,
      "grad_norm": 1.1564288139343262,
      "learning_rate": 9.634236842105264e-05,
      "loss": 0.3953,
      "step": 33914
    },
    {
      "epoch": 1.899148840855639,
      "grad_norm": 1.2269929647445679,
      "learning_rate": 9.63421052631579e-05,
      "loss": 0.4532,
      "step": 33915
    },
    {
      "epoch": 1.899204838167768,
      "grad_norm": 1.3798385858535767,
      "learning_rate": 9.634184210526316e-05,
      "loss": 0.4787,
      "step": 33916
    },
    {
      "epoch": 1.899260835479897,
      "grad_norm": 1.1531494855880737,
      "learning_rate": 9.634157894736843e-05,
      "loss": 0.5046,
      "step": 33917
    },
    {
      "epoch": 1.899316832792026,
      "grad_norm": 1.664993405342102,
      "learning_rate": 9.634131578947369e-05,
      "loss": 0.684,
      "step": 33918
    },
    {
      "epoch": 1.899372830104155,
      "grad_norm": 1.21270751953125,
      "learning_rate": 9.634105263157895e-05,
      "loss": 0.3913,
      "step": 33919
    },
    {
      "epoch": 1.899428827416284,
      "grad_norm": 1.0564695596694946,
      "learning_rate": 9.634078947368421e-05,
      "loss": 0.3431,
      "step": 33920
    },
    {
      "epoch": 1.899484824728413,
      "grad_norm": 1.189333200454712,
      "learning_rate": 9.634052631578948e-05,
      "loss": 0.4347,
      "step": 33921
    },
    {
      "epoch": 1.899540822040542,
      "grad_norm": 1.3843694925308228,
      "learning_rate": 9.634026315789474e-05,
      "loss": 0.4178,
      "step": 33922
    },
    {
      "epoch": 1.899596819352671,
      "grad_norm": 1.4496698379516602,
      "learning_rate": 9.634e-05,
      "loss": 0.5207,
      "step": 33923
    },
    {
      "epoch": 1.8996528166648001,
      "grad_norm": 1.14940345287323,
      "learning_rate": 9.633973684210526e-05,
      "loss": 0.3536,
      "step": 33924
    },
    {
      "epoch": 1.8997088139769291,
      "grad_norm": 1.3769302368164062,
      "learning_rate": 9.633947368421054e-05,
      "loss": 0.641,
      "step": 33925
    },
    {
      "epoch": 1.8997648112890582,
      "grad_norm": 1.2607393264770508,
      "learning_rate": 9.63392105263158e-05,
      "loss": 0.4933,
      "step": 33926
    },
    {
      "epoch": 1.8998208086011872,
      "grad_norm": 1.3049705028533936,
      "learning_rate": 9.633894736842107e-05,
      "loss": 0.4616,
      "step": 33927
    },
    {
      "epoch": 1.8998768059133162,
      "grad_norm": 1.2081362009048462,
      "learning_rate": 9.633868421052631e-05,
      "loss": 0.4235,
      "step": 33928
    },
    {
      "epoch": 1.8999328032254452,
      "grad_norm": 1.797351598739624,
      "learning_rate": 9.633842105263159e-05,
      "loss": 0.5012,
      "step": 33929
    },
    {
      "epoch": 1.8999888005375742,
      "grad_norm": 1.1874183416366577,
      "learning_rate": 9.633815789473685e-05,
      "loss": 0.4893,
      "step": 33930
    },
    {
      "epoch": 1.9000447978497033,
      "grad_norm": 1.5537874698638916,
      "learning_rate": 9.633789473684211e-05,
      "loss": 0.553,
      "step": 33931
    },
    {
      "epoch": 1.9001007951618323,
      "grad_norm": 4.517096996307373,
      "learning_rate": 9.633763157894737e-05,
      "loss": 0.4328,
      "step": 33932
    },
    {
      "epoch": 1.9001567924739613,
      "grad_norm": 1.2272001504898071,
      "learning_rate": 9.633736842105263e-05,
      "loss": 0.4294,
      "step": 33933
    },
    {
      "epoch": 1.9002127897860903,
      "grad_norm": 1.1718171834945679,
      "learning_rate": 9.63371052631579e-05,
      "loss": 0.3851,
      "step": 33934
    },
    {
      "epoch": 1.9002687870982193,
      "grad_norm": 1.3043668270111084,
      "learning_rate": 9.633684210526316e-05,
      "loss": 0.6295,
      "step": 33935
    },
    {
      "epoch": 1.9003247844103484,
      "grad_norm": 1.1041791439056396,
      "learning_rate": 9.633657894736843e-05,
      "loss": 0.3721,
      "step": 33936
    },
    {
      "epoch": 1.9003807817224774,
      "grad_norm": 1.1693410873413086,
      "learning_rate": 9.633631578947368e-05,
      "loss": 0.392,
      "step": 33937
    },
    {
      "epoch": 1.9004367790346064,
      "grad_norm": 1.0420026779174805,
      "learning_rate": 9.633605263157895e-05,
      "loss": 0.4449,
      "step": 33938
    },
    {
      "epoch": 1.9004927763467352,
      "grad_norm": 1.079854130744934,
      "learning_rate": 9.633578947368421e-05,
      "loss": 0.3689,
      "step": 33939
    },
    {
      "epoch": 1.9005487736588642,
      "grad_norm": 1.185213327407837,
      "learning_rate": 9.633552631578949e-05,
      "loss": 0.4248,
      "step": 33940
    },
    {
      "epoch": 1.9006047709709932,
      "grad_norm": 1.264661192893982,
      "learning_rate": 9.633526315789475e-05,
      "loss": 0.4596,
      "step": 33941
    },
    {
      "epoch": 1.9006607682831222,
      "grad_norm": 1.2667551040649414,
      "learning_rate": 9.6335e-05,
      "loss": 0.42,
      "step": 33942
    },
    {
      "epoch": 1.9007167655952513,
      "grad_norm": 1.214684247970581,
      "learning_rate": 9.633473684210526e-05,
      "loss": 0.3905,
      "step": 33943
    },
    {
      "epoch": 1.9007727629073803,
      "grad_norm": 1.3427232503890991,
      "learning_rate": 9.633447368421054e-05,
      "loss": 0.4673,
      "step": 33944
    },
    {
      "epoch": 1.9008287602195093,
      "grad_norm": 1.7394793033599854,
      "learning_rate": 9.63342105263158e-05,
      "loss": 0.4895,
      "step": 33945
    },
    {
      "epoch": 1.9008847575316383,
      "grad_norm": 1.4160462617874146,
      "learning_rate": 9.633394736842106e-05,
      "loss": 0.5204,
      "step": 33946
    },
    {
      "epoch": 1.9009407548437673,
      "grad_norm": 1.0840052366256714,
      "learning_rate": 9.633368421052632e-05,
      "loss": 0.3965,
      "step": 33947
    },
    {
      "epoch": 1.9009967521558964,
      "grad_norm": 1.3159178495407104,
      "learning_rate": 9.633342105263158e-05,
      "loss": 0.5057,
      "step": 33948
    },
    {
      "epoch": 1.9010527494680254,
      "grad_norm": 1.166968584060669,
      "learning_rate": 9.633315789473685e-05,
      "loss": 0.3823,
      "step": 33949
    },
    {
      "epoch": 1.9011087467801544,
      "grad_norm": 1.4747583866119385,
      "learning_rate": 9.633289473684211e-05,
      "loss": 0.5492,
      "step": 33950
    },
    {
      "epoch": 1.9011647440922834,
      "grad_norm": 1.4184335470199585,
      "learning_rate": 9.633263157894737e-05,
      "loss": 0.4097,
      "step": 33951
    },
    {
      "epoch": 1.9012207414044124,
      "grad_norm": 1.1197212934494019,
      "learning_rate": 9.633236842105263e-05,
      "loss": 0.362,
      "step": 33952
    },
    {
      "epoch": 1.9012767387165415,
      "grad_norm": 1.1916241645812988,
      "learning_rate": 9.63321052631579e-05,
      "loss": 0.3744,
      "step": 33953
    },
    {
      "epoch": 1.9013327360286705,
      "grad_norm": 1.0502461194992065,
      "learning_rate": 9.633184210526316e-05,
      "loss": 0.3703,
      "step": 33954
    },
    {
      "epoch": 1.9013887333407995,
      "grad_norm": 1.2252848148345947,
      "learning_rate": 9.633157894736842e-05,
      "loss": 0.4205,
      "step": 33955
    },
    {
      "epoch": 1.9014447306529285,
      "grad_norm": 1.6981635093688965,
      "learning_rate": 9.633131578947368e-05,
      "loss": 0.6063,
      "step": 33956
    },
    {
      "epoch": 1.9015007279650575,
      "grad_norm": 1.4446275234222412,
      "learning_rate": 9.633105263157896e-05,
      "loss": 0.4902,
      "step": 33957
    },
    {
      "epoch": 1.9015567252771866,
      "grad_norm": 1.41056489944458,
      "learning_rate": 9.633078947368422e-05,
      "loss": 0.4143,
      "step": 33958
    },
    {
      "epoch": 1.9016127225893156,
      "grad_norm": 1.5445547103881836,
      "learning_rate": 9.633052631578949e-05,
      "loss": 0.5362,
      "step": 33959
    },
    {
      "epoch": 1.9016687199014446,
      "grad_norm": 1.2787317037582397,
      "learning_rate": 9.633026315789473e-05,
      "loss": 0.5314,
      "step": 33960
    },
    {
      "epoch": 1.9017247172135736,
      "grad_norm": 1.3120055198669434,
      "learning_rate": 9.633000000000001e-05,
      "loss": 0.6227,
      "step": 33961
    },
    {
      "epoch": 1.9017807145257026,
      "grad_norm": 1.3405863046646118,
      "learning_rate": 9.632973684210527e-05,
      "loss": 0.5112,
      "step": 33962
    },
    {
      "epoch": 1.9018367118378317,
      "grad_norm": 1.1099889278411865,
      "learning_rate": 9.632947368421054e-05,
      "loss": 0.3395,
      "step": 33963
    },
    {
      "epoch": 1.9018927091499607,
      "grad_norm": 1.5446983575820923,
      "learning_rate": 9.63292105263158e-05,
      "loss": 0.4659,
      "step": 33964
    },
    {
      "epoch": 1.9019487064620897,
      "grad_norm": 1.3510504961013794,
      "learning_rate": 9.632894736842105e-05,
      "loss": 0.39,
      "step": 33965
    },
    {
      "epoch": 1.9020047037742187,
      "grad_norm": 1.6096017360687256,
      "learning_rate": 9.632868421052632e-05,
      "loss": 0.4869,
      "step": 33966
    },
    {
      "epoch": 1.9020607010863477,
      "grad_norm": 1.2717180252075195,
      "learning_rate": 9.632842105263158e-05,
      "loss": 0.3454,
      "step": 33967
    },
    {
      "epoch": 1.9021166983984767,
      "grad_norm": 1.4995439052581787,
      "learning_rate": 9.632815789473685e-05,
      "loss": 0.4375,
      "step": 33968
    },
    {
      "epoch": 1.9021726957106058,
      "grad_norm": 1.2514420747756958,
      "learning_rate": 9.63278947368421e-05,
      "loss": 0.398,
      "step": 33969
    },
    {
      "epoch": 1.9022286930227348,
      "grad_norm": 1.4572794437408447,
      "learning_rate": 9.632763157894737e-05,
      "loss": 0.4694,
      "step": 33970
    },
    {
      "epoch": 1.9022846903348638,
      "grad_norm": 1.3768348693847656,
      "learning_rate": 9.632736842105263e-05,
      "loss": 0.6099,
      "step": 33971
    },
    {
      "epoch": 1.9023406876469928,
      "grad_norm": 4.976465225219727,
      "learning_rate": 9.63271052631579e-05,
      "loss": 0.513,
      "step": 33972
    },
    {
      "epoch": 1.9023966849591218,
      "grad_norm": 1.2748594284057617,
      "learning_rate": 9.632684210526317e-05,
      "loss": 0.4999,
      "step": 33973
    },
    {
      "epoch": 1.9024526822712509,
      "grad_norm": 1.338942527770996,
      "learning_rate": 9.632657894736842e-05,
      "loss": 0.4815,
      "step": 33974
    },
    {
      "epoch": 1.9025086795833799,
      "grad_norm": 1.2265136241912842,
      "learning_rate": 9.632631578947368e-05,
      "loss": 0.3827,
      "step": 33975
    },
    {
      "epoch": 1.902564676895509,
      "grad_norm": 1.1530029773712158,
      "learning_rate": 9.632605263157896e-05,
      "loss": 0.3824,
      "step": 33976
    },
    {
      "epoch": 1.902620674207638,
      "grad_norm": 1.1153985261917114,
      "learning_rate": 9.632578947368422e-05,
      "loss": 0.4111,
      "step": 33977
    },
    {
      "epoch": 1.902676671519767,
      "grad_norm": 1.3617947101593018,
      "learning_rate": 9.632552631578948e-05,
      "loss": 0.5695,
      "step": 33978
    },
    {
      "epoch": 1.902732668831896,
      "grad_norm": 1.4597316980361938,
      "learning_rate": 9.632526315789474e-05,
      "loss": 0.6447,
      "step": 33979
    },
    {
      "epoch": 1.902788666144025,
      "grad_norm": 1.3488487005233765,
      "learning_rate": 9.632500000000001e-05,
      "loss": 0.4099,
      "step": 33980
    },
    {
      "epoch": 1.902844663456154,
      "grad_norm": 1.186000108718872,
      "learning_rate": 9.632473684210527e-05,
      "loss": 0.4242,
      "step": 33981
    },
    {
      "epoch": 1.902900660768283,
      "grad_norm": 1.384997010231018,
      "learning_rate": 9.632447368421053e-05,
      "loss": 0.5924,
      "step": 33982
    },
    {
      "epoch": 1.902956658080412,
      "grad_norm": 1.370645523071289,
      "learning_rate": 9.632421052631579e-05,
      "loss": 0.3987,
      "step": 33983
    },
    {
      "epoch": 1.903012655392541,
      "grad_norm": 1.3196059465408325,
      "learning_rate": 9.632394736842105e-05,
      "loss": 0.5177,
      "step": 33984
    },
    {
      "epoch": 1.90306865270467,
      "grad_norm": 1.0273501873016357,
      "learning_rate": 9.632368421052632e-05,
      "loss": 0.4669,
      "step": 33985
    },
    {
      "epoch": 1.903124650016799,
      "grad_norm": 1.3591892719268799,
      "learning_rate": 9.632342105263158e-05,
      "loss": 0.4202,
      "step": 33986
    },
    {
      "epoch": 1.903180647328928,
      "grad_norm": 1.6412771940231323,
      "learning_rate": 9.632315789473684e-05,
      "loss": 0.4595,
      "step": 33987
    },
    {
      "epoch": 1.9032366446410571,
      "grad_norm": 1.127336859703064,
      "learning_rate": 9.63228947368421e-05,
      "loss": 0.3776,
      "step": 33988
    },
    {
      "epoch": 1.9032926419531861,
      "grad_norm": 1.1677864789962769,
      "learning_rate": 9.632263157894737e-05,
      "loss": 0.4411,
      "step": 33989
    },
    {
      "epoch": 1.9033486392653152,
      "grad_norm": 1.3534612655639648,
      "learning_rate": 9.632236842105263e-05,
      "loss": 0.4277,
      "step": 33990
    },
    {
      "epoch": 1.9034046365774442,
      "grad_norm": 1.0454528331756592,
      "learning_rate": 9.632210526315791e-05,
      "loss": 0.4218,
      "step": 33991
    },
    {
      "epoch": 1.9034606338895732,
      "grad_norm": 1.1414886713027954,
      "learning_rate": 9.632184210526315e-05,
      "loss": 0.3341,
      "step": 33992
    },
    {
      "epoch": 1.9035166312017022,
      "grad_norm": 1.2816598415374756,
      "learning_rate": 9.632157894736843e-05,
      "loss": 0.4588,
      "step": 33993
    },
    {
      "epoch": 1.9035726285138312,
      "grad_norm": 1.324050784111023,
      "learning_rate": 9.632131578947369e-05,
      "loss": 0.413,
      "step": 33994
    },
    {
      "epoch": 1.9036286258259603,
      "grad_norm": 1.1996866464614868,
      "learning_rate": 9.632105263157896e-05,
      "loss": 0.3494,
      "step": 33995
    },
    {
      "epoch": 1.9036846231380893,
      "grad_norm": 1.3523858785629272,
      "learning_rate": 9.632078947368422e-05,
      "loss": 0.3639,
      "step": 33996
    },
    {
      "epoch": 1.9037406204502183,
      "grad_norm": 1.1181321144104004,
      "learning_rate": 9.632052631578948e-05,
      "loss": 0.3104,
      "step": 33997
    },
    {
      "epoch": 1.9037966177623473,
      "grad_norm": 1.1143064498901367,
      "learning_rate": 9.632026315789474e-05,
      "loss": 0.3711,
      "step": 33998
    },
    {
      "epoch": 1.9038526150744763,
      "grad_norm": 1.3833227157592773,
      "learning_rate": 9.632e-05,
      "loss": 0.6783,
      "step": 33999
    },
    {
      "epoch": 1.9039086123866054,
      "grad_norm": 1.2196251153945923,
      "learning_rate": 9.631973684210527e-05,
      "loss": 0.3824,
      "step": 34000
    },
    {
      "epoch": 1.9039646096987344,
      "grad_norm": 1.4445823431015015,
      "learning_rate": 9.631947368421053e-05,
      "loss": 0.3967,
      "step": 34001
    },
    {
      "epoch": 1.9040206070108634,
      "grad_norm": 1.404177188873291,
      "learning_rate": 9.631921052631579e-05,
      "loss": 0.4164,
      "step": 34002
    },
    {
      "epoch": 1.9040766043229924,
      "grad_norm": 1.4708819389343262,
      "learning_rate": 9.631894736842105e-05,
      "loss": 0.4517,
      "step": 34003
    },
    {
      "epoch": 1.9041326016351214,
      "grad_norm": 1.2899205684661865,
      "learning_rate": 9.631868421052633e-05,
      "loss": 0.5623,
      "step": 34004
    },
    {
      "epoch": 1.9041885989472505,
      "grad_norm": 1.2500195503234863,
      "learning_rate": 9.631842105263158e-05,
      "loss": 0.4628,
      "step": 34005
    },
    {
      "epoch": 1.9042445962593795,
      "grad_norm": 1.3206158876419067,
      "learning_rate": 9.631815789473684e-05,
      "loss": 0.4163,
      "step": 34006
    },
    {
      "epoch": 1.9043005935715085,
      "grad_norm": 1.0282695293426514,
      "learning_rate": 9.63178947368421e-05,
      "loss": 0.3138,
      "step": 34007
    },
    {
      "epoch": 1.9043565908836375,
      "grad_norm": 1.237765908241272,
      "learning_rate": 9.631763157894738e-05,
      "loss": 0.6517,
      "step": 34008
    },
    {
      "epoch": 1.9044125881957665,
      "grad_norm": 1.2659456729888916,
      "learning_rate": 9.631736842105264e-05,
      "loss": 0.4954,
      "step": 34009
    },
    {
      "epoch": 1.9044685855078956,
      "grad_norm": 1.0131499767303467,
      "learning_rate": 9.63171052631579e-05,
      "loss": 0.2964,
      "step": 34010
    },
    {
      "epoch": 1.9045245828200246,
      "grad_norm": 1.1911507844924927,
      "learning_rate": 9.631684210526316e-05,
      "loss": 0.5041,
      "step": 34011
    },
    {
      "epoch": 1.9045805801321536,
      "grad_norm": 1.6755865812301636,
      "learning_rate": 9.631657894736843e-05,
      "loss": 0.5075,
      "step": 34012
    },
    {
      "epoch": 1.9046365774442826,
      "grad_norm": 1.1088289022445679,
      "learning_rate": 9.631631578947369e-05,
      "loss": 0.3312,
      "step": 34013
    },
    {
      "epoch": 1.9046925747564116,
      "grad_norm": 0.9510724544525146,
      "learning_rate": 9.631605263157896e-05,
      "loss": 0.4505,
      "step": 34014
    },
    {
      "epoch": 1.9047485720685406,
      "grad_norm": 1.2416331768035889,
      "learning_rate": 9.631578947368421e-05,
      "loss": 0.3983,
      "step": 34015
    },
    {
      "epoch": 1.9048045693806697,
      "grad_norm": 1.8037056922912598,
      "learning_rate": 9.631552631578948e-05,
      "loss": 0.4658,
      "step": 34016
    },
    {
      "epoch": 1.9048605666927987,
      "grad_norm": 1.2995556592941284,
      "learning_rate": 9.631526315789474e-05,
      "loss": 0.5357,
      "step": 34017
    },
    {
      "epoch": 1.9049165640049277,
      "grad_norm": 1.3226078748703003,
      "learning_rate": 9.6315e-05,
      "loss": 0.4247,
      "step": 34018
    },
    {
      "epoch": 1.9049725613170567,
      "grad_norm": 1.5135236978530884,
      "learning_rate": 9.631473684210528e-05,
      "loss": 0.5026,
      "step": 34019
    },
    {
      "epoch": 1.9050285586291857,
      "grad_norm": 1.210835576057434,
      "learning_rate": 9.631447368421052e-05,
      "loss": 0.4925,
      "step": 34020
    },
    {
      "epoch": 1.9050845559413148,
      "grad_norm": 1.1402785778045654,
      "learning_rate": 9.63142105263158e-05,
      "loss": 0.3449,
      "step": 34021
    },
    {
      "epoch": 1.9051405532534438,
      "grad_norm": 1.345220923423767,
      "learning_rate": 9.631394736842105e-05,
      "loss": 0.4847,
      "step": 34022
    },
    {
      "epoch": 1.9051965505655728,
      "grad_norm": 1.2168327569961548,
      "learning_rate": 9.631368421052633e-05,
      "loss": 0.5165,
      "step": 34023
    },
    {
      "epoch": 1.9052525478777018,
      "grad_norm": 1.2520406246185303,
      "learning_rate": 9.631342105263157e-05,
      "loss": 0.3482,
      "step": 34024
    },
    {
      "epoch": 1.9053085451898308,
      "grad_norm": 1.5670303106307983,
      "learning_rate": 9.631315789473685e-05,
      "loss": 0.4262,
      "step": 34025
    },
    {
      "epoch": 1.9053645425019599,
      "grad_norm": 1.2887228727340698,
      "learning_rate": 9.631289473684211e-05,
      "loss": 0.4468,
      "step": 34026
    },
    {
      "epoch": 1.9054205398140889,
      "grad_norm": 1.4221678972244263,
      "learning_rate": 9.631263157894738e-05,
      "loss": 0.3648,
      "step": 34027
    },
    {
      "epoch": 1.905476537126218,
      "grad_norm": 1.2144310474395752,
      "learning_rate": 9.631236842105264e-05,
      "loss": 0.5628,
      "step": 34028
    },
    {
      "epoch": 1.905532534438347,
      "grad_norm": 1.2756747007369995,
      "learning_rate": 9.63121052631579e-05,
      "loss": 0.4993,
      "step": 34029
    },
    {
      "epoch": 1.905588531750476,
      "grad_norm": 1.1108474731445312,
      "learning_rate": 9.631184210526316e-05,
      "loss": 0.4143,
      "step": 34030
    },
    {
      "epoch": 1.905644529062605,
      "grad_norm": 2.419938802719116,
      "learning_rate": 9.631157894736843e-05,
      "loss": 0.5744,
      "step": 34031
    },
    {
      "epoch": 1.905700526374734,
      "grad_norm": 1.4356378316879272,
      "learning_rate": 9.631131578947369e-05,
      "loss": 0.5235,
      "step": 34032
    },
    {
      "epoch": 1.905756523686863,
      "grad_norm": 1.190591812133789,
      "learning_rate": 9.631105263157895e-05,
      "loss": 0.4297,
      "step": 34033
    },
    {
      "epoch": 1.905812520998992,
      "grad_norm": 1.5549372434616089,
      "learning_rate": 9.631078947368421e-05,
      "loss": 0.5141,
      "step": 34034
    },
    {
      "epoch": 1.905868518311121,
      "grad_norm": 1.0839934349060059,
      "learning_rate": 9.631052631578947e-05,
      "loss": 0.3773,
      "step": 34035
    },
    {
      "epoch": 1.90592451562325,
      "grad_norm": 1.4727147817611694,
      "learning_rate": 9.631026315789474e-05,
      "loss": 0.497,
      "step": 34036
    },
    {
      "epoch": 1.905980512935379,
      "grad_norm": 1.1691676378250122,
      "learning_rate": 9.631e-05,
      "loss": 0.4149,
      "step": 34037
    },
    {
      "epoch": 1.906036510247508,
      "grad_norm": 1.3910284042358398,
      "learning_rate": 9.630973684210526e-05,
      "loss": 0.5442,
      "step": 34038
    },
    {
      "epoch": 1.906092507559637,
      "grad_norm": 1.3977830410003662,
      "learning_rate": 9.630947368421052e-05,
      "loss": 0.5295,
      "step": 34039
    },
    {
      "epoch": 1.9061485048717661,
      "grad_norm": 1.445651888847351,
      "learning_rate": 9.63092105263158e-05,
      "loss": 0.4602,
      "step": 34040
    },
    {
      "epoch": 1.9062045021838951,
      "grad_norm": 1.5516797304153442,
      "learning_rate": 9.630894736842106e-05,
      "loss": 0.636,
      "step": 34041
    },
    {
      "epoch": 1.9062604994960242,
      "grad_norm": 1.1154232025146484,
      "learning_rate": 9.630868421052632e-05,
      "loss": 0.3463,
      "step": 34042
    },
    {
      "epoch": 1.9063164968081532,
      "grad_norm": 1.4853843450546265,
      "learning_rate": 9.630842105263158e-05,
      "loss": 0.4056,
      "step": 34043
    },
    {
      "epoch": 1.9063724941202822,
      "grad_norm": 1.3673241138458252,
      "learning_rate": 9.630815789473685e-05,
      "loss": 0.398,
      "step": 34044
    },
    {
      "epoch": 1.9064284914324112,
      "grad_norm": 1.294242262840271,
      "learning_rate": 9.630789473684211e-05,
      "loss": 0.5059,
      "step": 34045
    },
    {
      "epoch": 1.9064844887445402,
      "grad_norm": 1.346583604812622,
      "learning_rate": 9.630763157894738e-05,
      "loss": 0.4852,
      "step": 34046
    },
    {
      "epoch": 1.9065404860566693,
      "grad_norm": 1.1595357656478882,
      "learning_rate": 9.630736842105263e-05,
      "loss": 0.499,
      "step": 34047
    },
    {
      "epoch": 1.9065964833687983,
      "grad_norm": 1.107212781906128,
      "learning_rate": 9.63071052631579e-05,
      "loss": 0.4304,
      "step": 34048
    },
    {
      "epoch": 1.9066524806809273,
      "grad_norm": 1.377042293548584,
      "learning_rate": 9.630684210526316e-05,
      "loss": 0.6586,
      "step": 34049
    },
    {
      "epoch": 1.9067084779930563,
      "grad_norm": 1.1463828086853027,
      "learning_rate": 9.630657894736844e-05,
      "loss": 0.4227,
      "step": 34050
    },
    {
      "epoch": 1.9067644753051853,
      "grad_norm": 1.401996374130249,
      "learning_rate": 9.63063157894737e-05,
      "loss": 0.5736,
      "step": 34051
    },
    {
      "epoch": 1.9068204726173144,
      "grad_norm": 1.1936086416244507,
      "learning_rate": 9.630605263157894e-05,
      "loss": 0.3585,
      "step": 34052
    },
    {
      "epoch": 1.9068764699294434,
      "grad_norm": 1.1592646837234497,
      "learning_rate": 9.630578947368421e-05,
      "loss": 0.4199,
      "step": 34053
    },
    {
      "epoch": 1.9069324672415724,
      "grad_norm": 1.3792665004730225,
      "learning_rate": 9.630552631578947e-05,
      "loss": 0.5398,
      "step": 34054
    },
    {
      "epoch": 1.9069884645537014,
      "grad_norm": 1.6987804174423218,
      "learning_rate": 9.630526315789475e-05,
      "loss": 0.4082,
      "step": 34055
    },
    {
      "epoch": 1.9070444618658304,
      "grad_norm": 1.2151567935943604,
      "learning_rate": 9.630500000000001e-05,
      "loss": 0.4407,
      "step": 34056
    },
    {
      "epoch": 1.9071004591779595,
      "grad_norm": 1.377261996269226,
      "learning_rate": 9.630473684210527e-05,
      "loss": 0.4117,
      "step": 34057
    },
    {
      "epoch": 1.9071564564900885,
      "grad_norm": 1.2642101049423218,
      "learning_rate": 9.630447368421053e-05,
      "loss": 0.5015,
      "step": 34058
    },
    {
      "epoch": 1.9072124538022175,
      "grad_norm": 1.256766438484192,
      "learning_rate": 9.63042105263158e-05,
      "loss": 0.4659,
      "step": 34059
    },
    {
      "epoch": 1.9072684511143465,
      "grad_norm": 1.3358551263809204,
      "learning_rate": 9.630394736842106e-05,
      "loss": 0.3662,
      "step": 34060
    },
    {
      "epoch": 1.9073244484264755,
      "grad_norm": 1.1349291801452637,
      "learning_rate": 9.630368421052632e-05,
      "loss": 0.3684,
      "step": 34061
    },
    {
      "epoch": 1.9073804457386045,
      "grad_norm": 2.011260509490967,
      "learning_rate": 9.630342105263158e-05,
      "loss": 0.3968,
      "step": 34062
    },
    {
      "epoch": 1.9074364430507336,
      "grad_norm": 2.6817777156829834,
      "learning_rate": 9.630315789473685e-05,
      "loss": 0.4376,
      "step": 34063
    },
    {
      "epoch": 1.9074924403628626,
      "grad_norm": 1.0032994747161865,
      "learning_rate": 9.630289473684211e-05,
      "loss": 0.3115,
      "step": 34064
    },
    {
      "epoch": 1.9075484376749916,
      "grad_norm": 1.1072883605957031,
      "learning_rate": 9.630263157894737e-05,
      "loss": 0.3222,
      "step": 34065
    },
    {
      "epoch": 1.9076044349871206,
      "grad_norm": 1.3170570135116577,
      "learning_rate": 9.630236842105263e-05,
      "loss": 0.5636,
      "step": 34066
    },
    {
      "epoch": 1.9076604322992496,
      "grad_norm": 1.0354145765304565,
      "learning_rate": 9.63021052631579e-05,
      "loss": 0.3679,
      "step": 34067
    },
    {
      "epoch": 1.9077164296113787,
      "grad_norm": 1.7400500774383545,
      "learning_rate": 9.630184210526316e-05,
      "loss": 0.5328,
      "step": 34068
    },
    {
      "epoch": 1.9077724269235077,
      "grad_norm": 1.0497963428497314,
      "learning_rate": 9.630157894736842e-05,
      "loss": 0.4762,
      "step": 34069
    },
    {
      "epoch": 1.9078284242356367,
      "grad_norm": 1.284080147743225,
      "learning_rate": 9.630131578947368e-05,
      "loss": 0.4976,
      "step": 34070
    },
    {
      "epoch": 1.9078844215477657,
      "grad_norm": 1.3971341848373413,
      "learning_rate": 9.630105263157894e-05,
      "loss": 0.5977,
      "step": 34071
    },
    {
      "epoch": 1.9079404188598947,
      "grad_norm": 1.1857272386550903,
      "learning_rate": 9.630078947368422e-05,
      "loss": 0.3346,
      "step": 34072
    },
    {
      "epoch": 1.9079964161720238,
      "grad_norm": 1.2894997596740723,
      "learning_rate": 9.630052631578948e-05,
      "loss": 0.4208,
      "step": 34073
    },
    {
      "epoch": 1.9080524134841528,
      "grad_norm": 1.0345476865768433,
      "learning_rate": 9.630026315789475e-05,
      "loss": 0.3871,
      "step": 34074
    },
    {
      "epoch": 1.9081084107962818,
      "grad_norm": 1.2449591159820557,
      "learning_rate": 9.63e-05,
      "loss": 0.509,
      "step": 34075
    },
    {
      "epoch": 1.9081644081084108,
      "grad_norm": 1.2070410251617432,
      "learning_rate": 9.629973684210527e-05,
      "loss": 0.463,
      "step": 34076
    },
    {
      "epoch": 1.9082204054205398,
      "grad_norm": 1.2701133489608765,
      "learning_rate": 9.629947368421053e-05,
      "loss": 0.4276,
      "step": 34077
    },
    {
      "epoch": 1.9082764027326689,
      "grad_norm": 1.3997082710266113,
      "learning_rate": 9.62992105263158e-05,
      "loss": 0.5083,
      "step": 34078
    },
    {
      "epoch": 1.9083324000447979,
      "grad_norm": 1.109686255455017,
      "learning_rate": 9.629894736842105e-05,
      "loss": 0.6153,
      "step": 34079
    },
    {
      "epoch": 1.908388397356927,
      "grad_norm": 1.3531676530838013,
      "learning_rate": 9.629868421052632e-05,
      "loss": 0.3774,
      "step": 34080
    },
    {
      "epoch": 1.908444394669056,
      "grad_norm": 1.209232211112976,
      "learning_rate": 9.629842105263158e-05,
      "loss": 0.4172,
      "step": 34081
    },
    {
      "epoch": 1.908500391981185,
      "grad_norm": 1.4462954998016357,
      "learning_rate": 9.629815789473685e-05,
      "loss": 0.473,
      "step": 34082
    },
    {
      "epoch": 1.908556389293314,
      "grad_norm": 1.3191204071044922,
      "learning_rate": 9.629789473684211e-05,
      "loss": 0.4976,
      "step": 34083
    },
    {
      "epoch": 1.908612386605443,
      "grad_norm": 1.070173978805542,
      "learning_rate": 9.629763157894737e-05,
      "loss": 0.3638,
      "step": 34084
    },
    {
      "epoch": 1.908668383917572,
      "grad_norm": 1.3106045722961426,
      "learning_rate": 9.629736842105263e-05,
      "loss": 0.5458,
      "step": 34085
    },
    {
      "epoch": 1.908724381229701,
      "grad_norm": 1.342474341392517,
      "learning_rate": 9.62971052631579e-05,
      "loss": 0.4291,
      "step": 34086
    },
    {
      "epoch": 1.90878037854183,
      "grad_norm": 1.2556720972061157,
      "learning_rate": 9.629684210526317e-05,
      "loss": 0.5439,
      "step": 34087
    },
    {
      "epoch": 1.908836375853959,
      "grad_norm": 1.4221272468566895,
      "learning_rate": 9.629657894736843e-05,
      "loss": 0.3934,
      "step": 34088
    },
    {
      "epoch": 1.908892373166088,
      "grad_norm": 1.4182671308517456,
      "learning_rate": 9.629631578947369e-05,
      "loss": 0.4742,
      "step": 34089
    },
    {
      "epoch": 1.908948370478217,
      "grad_norm": 1.1342058181762695,
      "learning_rate": 9.629605263157895e-05,
      "loss": 0.4728,
      "step": 34090
    },
    {
      "epoch": 1.909004367790346,
      "grad_norm": 1.3984627723693848,
      "learning_rate": 9.629578947368422e-05,
      "loss": 0.5415,
      "step": 34091
    },
    {
      "epoch": 1.9090603651024751,
      "grad_norm": 1.110479474067688,
      "learning_rate": 9.629552631578948e-05,
      "loss": 0.4191,
      "step": 34092
    },
    {
      "epoch": 1.9091163624146041,
      "grad_norm": 1.3373271226882935,
      "learning_rate": 9.629526315789474e-05,
      "loss": 0.4739,
      "step": 34093
    },
    {
      "epoch": 1.9091723597267332,
      "grad_norm": 1.0889606475830078,
      "learning_rate": 9.6295e-05,
      "loss": 0.4729,
      "step": 34094
    },
    {
      "epoch": 1.9092283570388622,
      "grad_norm": 1.1331942081451416,
      "learning_rate": 9.629473684210527e-05,
      "loss": 0.4056,
      "step": 34095
    },
    {
      "epoch": 1.9092843543509912,
      "grad_norm": 1.4616351127624512,
      "learning_rate": 9.629447368421053e-05,
      "loss": 0.3804,
      "step": 34096
    },
    {
      "epoch": 1.9093403516631202,
      "grad_norm": 1.1372205018997192,
      "learning_rate": 9.629421052631579e-05,
      "loss": 0.3689,
      "step": 34097
    },
    {
      "epoch": 1.9093963489752492,
      "grad_norm": 1.503621220588684,
      "learning_rate": 9.629394736842105e-05,
      "loss": 0.5472,
      "step": 34098
    },
    {
      "epoch": 1.9094523462873783,
      "grad_norm": 1.1998803615570068,
      "learning_rate": 9.629368421052632e-05,
      "loss": 0.4041,
      "step": 34099
    },
    {
      "epoch": 1.9095083435995073,
      "grad_norm": 1.2902586460113525,
      "learning_rate": 9.629342105263158e-05,
      "loss": 0.4447,
      "step": 34100
    },
    {
      "epoch": 1.9095643409116363,
      "grad_norm": 1.6612448692321777,
      "learning_rate": 9.629315789473686e-05,
      "loss": 0.4459,
      "step": 34101
    },
    {
      "epoch": 1.9096203382237653,
      "grad_norm": 1.2382559776306152,
      "learning_rate": 9.62928947368421e-05,
      "loss": 0.435,
      "step": 34102
    },
    {
      "epoch": 1.9096763355358943,
      "grad_norm": 1.1899651288986206,
      "learning_rate": 9.629263157894736e-05,
      "loss": 0.396,
      "step": 34103
    },
    {
      "epoch": 1.9097323328480234,
      "grad_norm": 1.619368314743042,
      "learning_rate": 9.629236842105264e-05,
      "loss": 0.4779,
      "step": 34104
    },
    {
      "epoch": 1.9097883301601524,
      "grad_norm": 1.3264882564544678,
      "learning_rate": 9.62921052631579e-05,
      "loss": 0.5627,
      "step": 34105
    },
    {
      "epoch": 1.9098443274722814,
      "grad_norm": 1.3368408679962158,
      "learning_rate": 9.629184210526317e-05,
      "loss": 0.3888,
      "step": 34106
    },
    {
      "epoch": 1.9099003247844104,
      "grad_norm": 1.3359593152999878,
      "learning_rate": 9.629157894736842e-05,
      "loss": 0.4386,
      "step": 34107
    },
    {
      "epoch": 1.9099563220965394,
      "grad_norm": 2.0662670135498047,
      "learning_rate": 9.629131578947369e-05,
      "loss": 0.4334,
      "step": 34108
    },
    {
      "epoch": 1.9100123194086684,
      "grad_norm": 1.4234237670898438,
      "learning_rate": 9.629105263157895e-05,
      "loss": 0.5127,
      "step": 34109
    },
    {
      "epoch": 1.9100683167207975,
      "grad_norm": 1.157505750656128,
      "learning_rate": 9.629078947368422e-05,
      "loss": 0.5082,
      "step": 34110
    },
    {
      "epoch": 1.9101243140329265,
      "grad_norm": 1.3347121477127075,
      "learning_rate": 9.629052631578948e-05,
      "loss": 0.4526,
      "step": 34111
    },
    {
      "epoch": 1.9101803113450555,
      "grad_norm": 1.220191478729248,
      "learning_rate": 9.629026315789474e-05,
      "loss": 0.4028,
      "step": 34112
    },
    {
      "epoch": 1.9102363086571845,
      "grad_norm": 1.5360329151153564,
      "learning_rate": 9.629e-05,
      "loss": 0.52,
      "step": 34113
    },
    {
      "epoch": 1.9102923059693135,
      "grad_norm": 1.1935615539550781,
      "learning_rate": 9.628973684210527e-05,
      "loss": 0.4619,
      "step": 34114
    },
    {
      "epoch": 1.9103483032814426,
      "grad_norm": 1.2846730947494507,
      "learning_rate": 9.628947368421053e-05,
      "loss": 0.4897,
      "step": 34115
    },
    {
      "epoch": 1.9104043005935716,
      "grad_norm": 1.267817497253418,
      "learning_rate": 9.62892105263158e-05,
      "loss": 0.4019,
      "step": 34116
    },
    {
      "epoch": 1.9104602979057006,
      "grad_norm": 1.3004090785980225,
      "learning_rate": 9.628894736842105e-05,
      "loss": 0.4717,
      "step": 34117
    },
    {
      "epoch": 1.9105162952178296,
      "grad_norm": 1.3779370784759521,
      "learning_rate": 9.628868421052633e-05,
      "loss": 0.4668,
      "step": 34118
    },
    {
      "epoch": 1.9105722925299586,
      "grad_norm": 1.5220592021942139,
      "learning_rate": 9.628842105263159e-05,
      "loss": 0.649,
      "step": 34119
    },
    {
      "epoch": 1.9106282898420877,
      "grad_norm": 1.109362006187439,
      "learning_rate": 9.628815789473685e-05,
      "loss": 0.4501,
      "step": 34120
    },
    {
      "epoch": 1.9106842871542167,
      "grad_norm": 1.2758779525756836,
      "learning_rate": 9.62878947368421e-05,
      "loss": 0.5198,
      "step": 34121
    },
    {
      "epoch": 1.9107402844663457,
      "grad_norm": 1.2492681741714478,
      "learning_rate": 9.628763157894737e-05,
      "loss": 0.4845,
      "step": 34122
    },
    {
      "epoch": 1.9107962817784747,
      "grad_norm": 1.2996468544006348,
      "learning_rate": 9.628736842105264e-05,
      "loss": 0.4167,
      "step": 34123
    },
    {
      "epoch": 1.9108522790906037,
      "grad_norm": 1.2472577095031738,
      "learning_rate": 9.62871052631579e-05,
      "loss": 0.5334,
      "step": 34124
    },
    {
      "epoch": 1.9109082764027328,
      "grad_norm": 1.2333887815475464,
      "learning_rate": 9.628684210526316e-05,
      "loss": 0.4408,
      "step": 34125
    },
    {
      "epoch": 1.9109642737148618,
      "grad_norm": 1.2807999849319458,
      "learning_rate": 9.628657894736842e-05,
      "loss": 0.4937,
      "step": 34126
    },
    {
      "epoch": 1.9110202710269908,
      "grad_norm": 1.3753105401992798,
      "learning_rate": 9.628631578947369e-05,
      "loss": 0.4081,
      "step": 34127
    },
    {
      "epoch": 1.9110762683391198,
      "grad_norm": 1.077946424484253,
      "learning_rate": 9.628605263157895e-05,
      "loss": 0.3922,
      "step": 34128
    },
    {
      "epoch": 1.9111322656512488,
      "grad_norm": 1.2039012908935547,
      "learning_rate": 9.628578947368422e-05,
      "loss": 0.4253,
      "step": 34129
    },
    {
      "epoch": 1.9111882629633778,
      "grad_norm": 1.1264578104019165,
      "learning_rate": 9.628552631578947e-05,
      "loss": 0.3595,
      "step": 34130
    },
    {
      "epoch": 1.9112442602755069,
      "grad_norm": 1.2084472179412842,
      "learning_rate": 9.628526315789474e-05,
      "loss": 0.4605,
      "step": 34131
    },
    {
      "epoch": 1.9113002575876359,
      "grad_norm": 1.6101555824279785,
      "learning_rate": 9.6285e-05,
      "loss": 0.5062,
      "step": 34132
    },
    {
      "epoch": 1.911356254899765,
      "grad_norm": 1.0796869993209839,
      "learning_rate": 9.628473684210528e-05,
      "loss": 0.423,
      "step": 34133
    },
    {
      "epoch": 1.911412252211894,
      "grad_norm": 1.3164533376693726,
      "learning_rate": 9.628447368421052e-05,
      "loss": 0.455,
      "step": 34134
    },
    {
      "epoch": 1.911468249524023,
      "grad_norm": 1.3260061740875244,
      "learning_rate": 9.62842105263158e-05,
      "loss": 0.4756,
      "step": 34135
    },
    {
      "epoch": 1.911524246836152,
      "grad_norm": 1.1585819721221924,
      "learning_rate": 9.628394736842106e-05,
      "loss": 0.3987,
      "step": 34136
    },
    {
      "epoch": 1.911580244148281,
      "grad_norm": 1.0840131044387817,
      "learning_rate": 9.628368421052633e-05,
      "loss": 0.3771,
      "step": 34137
    },
    {
      "epoch": 1.91163624146041,
      "grad_norm": 1.3092949390411377,
      "learning_rate": 9.628342105263159e-05,
      "loss": 0.2978,
      "step": 34138
    },
    {
      "epoch": 1.911692238772539,
      "grad_norm": 1.2766376733779907,
      "learning_rate": 9.628315789473684e-05,
      "loss": 0.5586,
      "step": 34139
    },
    {
      "epoch": 1.911748236084668,
      "grad_norm": 1.1610255241394043,
      "learning_rate": 9.628289473684211e-05,
      "loss": 0.3512,
      "step": 34140
    },
    {
      "epoch": 1.911804233396797,
      "grad_norm": 1.5795600414276123,
      "learning_rate": 9.628263157894737e-05,
      "loss": 0.7731,
      "step": 34141
    },
    {
      "epoch": 1.911860230708926,
      "grad_norm": 1.1372798681259155,
      "learning_rate": 9.628236842105264e-05,
      "loss": 0.3679,
      "step": 34142
    },
    {
      "epoch": 1.911916228021055,
      "grad_norm": 1.3664319515228271,
      "learning_rate": 9.62821052631579e-05,
      "loss": 0.4945,
      "step": 34143
    },
    {
      "epoch": 1.9119722253331841,
      "grad_norm": 1.1209534406661987,
      "learning_rate": 9.628184210526316e-05,
      "loss": 0.3983,
      "step": 34144
    },
    {
      "epoch": 1.9120282226453131,
      "grad_norm": 1.691123366355896,
      "learning_rate": 9.628157894736842e-05,
      "loss": 0.3647,
      "step": 34145
    },
    {
      "epoch": 1.9120842199574422,
      "grad_norm": 1.096861481666565,
      "learning_rate": 9.62813157894737e-05,
      "loss": 0.3945,
      "step": 34146
    },
    {
      "epoch": 1.9121402172695712,
      "grad_norm": 1.6279759407043457,
      "learning_rate": 9.628105263157895e-05,
      "loss": 0.6269,
      "step": 34147
    },
    {
      "epoch": 1.9121962145817002,
      "grad_norm": 1.212383508682251,
      "learning_rate": 9.628078947368421e-05,
      "loss": 0.6039,
      "step": 34148
    },
    {
      "epoch": 1.9122522118938292,
      "grad_norm": 1.4479109048843384,
      "learning_rate": 9.628052631578947e-05,
      "loss": 0.4781,
      "step": 34149
    },
    {
      "epoch": 1.9123082092059582,
      "grad_norm": 1.1678768396377563,
      "learning_rate": 9.628026315789475e-05,
      "loss": 0.369,
      "step": 34150
    },
    {
      "epoch": 1.9123642065180873,
      "grad_norm": 1.184769630432129,
      "learning_rate": 9.628e-05,
      "loss": 0.3619,
      "step": 34151
    },
    {
      "epoch": 1.9124202038302163,
      "grad_norm": 1.2355364561080933,
      "learning_rate": 9.627973684210527e-05,
      "loss": 0.5006,
      "step": 34152
    },
    {
      "epoch": 1.9124762011423453,
      "grad_norm": 1.4290616512298584,
      "learning_rate": 9.627947368421053e-05,
      "loss": 0.4971,
      "step": 34153
    },
    {
      "epoch": 1.9125321984544743,
      "grad_norm": 1.1668577194213867,
      "learning_rate": 9.62792105263158e-05,
      "loss": 0.5216,
      "step": 34154
    },
    {
      "epoch": 1.9125881957666033,
      "grad_norm": 1.20247220993042,
      "learning_rate": 9.627894736842106e-05,
      "loss": 0.5353,
      "step": 34155
    },
    {
      "epoch": 1.9126441930787323,
      "grad_norm": 1.2770265340805054,
      "learning_rate": 9.627868421052632e-05,
      "loss": 0.4636,
      "step": 34156
    },
    {
      "epoch": 1.9127001903908614,
      "grad_norm": 1.2879416942596436,
      "learning_rate": 9.627842105263158e-05,
      "loss": 0.3738,
      "step": 34157
    },
    {
      "epoch": 1.9127561877029904,
      "grad_norm": 1.2735261917114258,
      "learning_rate": 9.627815789473684e-05,
      "loss": 0.4967,
      "step": 34158
    },
    {
      "epoch": 1.9128121850151194,
      "grad_norm": 1.1289223432540894,
      "learning_rate": 9.627789473684211e-05,
      "loss": 0.4111,
      "step": 34159
    },
    {
      "epoch": 1.9128681823272484,
      "grad_norm": 1.3695653676986694,
      "learning_rate": 9.627763157894737e-05,
      "loss": 0.4389,
      "step": 34160
    },
    {
      "epoch": 1.9129241796393774,
      "grad_norm": 1.3141605854034424,
      "learning_rate": 9.627736842105264e-05,
      "loss": 0.5207,
      "step": 34161
    },
    {
      "epoch": 1.9129801769515065,
      "grad_norm": 1.2922433614730835,
      "learning_rate": 9.627710526315789e-05,
      "loss": 0.3467,
      "step": 34162
    },
    {
      "epoch": 1.9130361742636355,
      "grad_norm": 1.3904390335083008,
      "learning_rate": 9.627684210526316e-05,
      "loss": 0.6212,
      "step": 34163
    },
    {
      "epoch": 1.9130921715757645,
      "grad_norm": 1.306141972541809,
      "learning_rate": 9.627657894736842e-05,
      "loss": 0.4378,
      "step": 34164
    },
    {
      "epoch": 1.9131481688878935,
      "grad_norm": 3.934844732284546,
      "learning_rate": 9.62763157894737e-05,
      "loss": 0.3961,
      "step": 34165
    },
    {
      "epoch": 1.9132041662000225,
      "grad_norm": 1.4914439916610718,
      "learning_rate": 9.627605263157896e-05,
      "loss": 0.4424,
      "step": 34166
    },
    {
      "epoch": 1.9132601635121516,
      "grad_norm": 1.3697915077209473,
      "learning_rate": 9.627578947368422e-05,
      "loss": 0.3805,
      "step": 34167
    },
    {
      "epoch": 1.9133161608242806,
      "grad_norm": 1.213944673538208,
      "learning_rate": 9.627552631578948e-05,
      "loss": 0.5198,
      "step": 34168
    },
    {
      "epoch": 1.9133721581364096,
      "grad_norm": 1.2957360744476318,
      "learning_rate": 9.627526315789475e-05,
      "loss": 0.5333,
      "step": 34169
    },
    {
      "epoch": 1.9134281554485386,
      "grad_norm": 1.2282534837722778,
      "learning_rate": 9.627500000000001e-05,
      "loss": 0.4358,
      "step": 34170
    },
    {
      "epoch": 1.9134841527606676,
      "grad_norm": 1.2574893236160278,
      "learning_rate": 9.627473684210527e-05,
      "loss": 0.447,
      "step": 34171
    },
    {
      "epoch": 1.9135401500727967,
      "grad_norm": 1.2336598634719849,
      "learning_rate": 9.627447368421053e-05,
      "loss": 0.3728,
      "step": 34172
    },
    {
      "epoch": 1.9135961473849257,
      "grad_norm": 1.1450051069259644,
      "learning_rate": 9.627421052631579e-05,
      "loss": 0.4009,
      "step": 34173
    },
    {
      "epoch": 1.9136521446970547,
      "grad_norm": 1.0521119832992554,
      "learning_rate": 9.627394736842106e-05,
      "loss": 0.3865,
      "step": 34174
    },
    {
      "epoch": 1.9137081420091837,
      "grad_norm": 1.1690138578414917,
      "learning_rate": 9.627368421052632e-05,
      "loss": 0.3348,
      "step": 34175
    },
    {
      "epoch": 1.9137641393213127,
      "grad_norm": 1.4462471008300781,
      "learning_rate": 9.627342105263158e-05,
      "loss": 0.533,
      "step": 34176
    },
    {
      "epoch": 1.9138201366334417,
      "grad_norm": 1.2312036752700806,
      "learning_rate": 9.627315789473684e-05,
      "loss": 0.4244,
      "step": 34177
    },
    {
      "epoch": 1.9138761339455708,
      "grad_norm": 1.2292710542678833,
      "learning_rate": 9.627289473684211e-05,
      "loss": 0.4981,
      "step": 34178
    },
    {
      "epoch": 1.9139321312576998,
      "grad_norm": 1.1272227764129639,
      "learning_rate": 9.627263157894737e-05,
      "loss": 0.46,
      "step": 34179
    },
    {
      "epoch": 1.9139881285698288,
      "grad_norm": 1.3013900518417358,
      "learning_rate": 9.627236842105263e-05,
      "loss": 0.4791,
      "step": 34180
    },
    {
      "epoch": 1.9140441258819578,
      "grad_norm": 1.3461647033691406,
      "learning_rate": 9.62721052631579e-05,
      "loss": 0.388,
      "step": 34181
    },
    {
      "epoch": 1.9141001231940868,
      "grad_norm": 1.478873372077942,
      "learning_rate": 9.627184210526317e-05,
      "loss": 0.5939,
      "step": 34182
    },
    {
      "epoch": 1.9141561205062159,
      "grad_norm": 1.1562503576278687,
      "learning_rate": 9.627157894736843e-05,
      "loss": 0.4228,
      "step": 34183
    },
    {
      "epoch": 1.9142121178183449,
      "grad_norm": 1.2001547813415527,
      "learning_rate": 9.62713157894737e-05,
      "loss": 0.4393,
      "step": 34184
    },
    {
      "epoch": 1.914268115130474,
      "grad_norm": 1.3186362981796265,
      "learning_rate": 9.627105263157895e-05,
      "loss": 0.4394,
      "step": 34185
    },
    {
      "epoch": 1.914324112442603,
      "grad_norm": 1.3413364887237549,
      "learning_rate": 9.627078947368422e-05,
      "loss": 0.4327,
      "step": 34186
    },
    {
      "epoch": 1.9143801097547317,
      "grad_norm": 1.077527642250061,
      "learning_rate": 9.627052631578948e-05,
      "loss": 0.434,
      "step": 34187
    },
    {
      "epoch": 1.9144361070668607,
      "grad_norm": 1.1904281377792358,
      "learning_rate": 9.627026315789475e-05,
      "loss": 0.4208,
      "step": 34188
    },
    {
      "epoch": 1.9144921043789898,
      "grad_norm": 1.4086194038391113,
      "learning_rate": 9.627e-05,
      "loss": 0.4608,
      "step": 34189
    },
    {
      "epoch": 1.9145481016911188,
      "grad_norm": 1.7058290243148804,
      "learning_rate": 9.626973684210526e-05,
      "loss": 0.5709,
      "step": 34190
    },
    {
      "epoch": 1.9146040990032478,
      "grad_norm": 2.118589162826538,
      "learning_rate": 9.626947368421053e-05,
      "loss": 0.7264,
      "step": 34191
    },
    {
      "epoch": 1.9146600963153768,
      "grad_norm": 1.8971775770187378,
      "learning_rate": 9.626921052631579e-05,
      "loss": 0.4616,
      "step": 34192
    },
    {
      "epoch": 1.9147160936275058,
      "grad_norm": 1.6888585090637207,
      "learning_rate": 9.626894736842106e-05,
      "loss": 0.4564,
      "step": 34193
    },
    {
      "epoch": 1.9147720909396349,
      "grad_norm": 1.201965570449829,
      "learning_rate": 9.626868421052631e-05,
      "loss": 0.3806,
      "step": 34194
    },
    {
      "epoch": 1.9148280882517639,
      "grad_norm": 1.3241913318634033,
      "learning_rate": 9.626842105263158e-05,
      "loss": 0.4463,
      "step": 34195
    },
    {
      "epoch": 1.914884085563893,
      "grad_norm": 1.21066415309906,
      "learning_rate": 9.626815789473684e-05,
      "loss": 0.564,
      "step": 34196
    },
    {
      "epoch": 1.914940082876022,
      "grad_norm": 1.0874874591827393,
      "learning_rate": 9.626789473684212e-05,
      "loss": 0.3959,
      "step": 34197
    },
    {
      "epoch": 1.914996080188151,
      "grad_norm": 1.0003881454467773,
      "learning_rate": 9.626763157894738e-05,
      "loss": 0.3957,
      "step": 34198
    },
    {
      "epoch": 1.91505207750028,
      "grad_norm": 1.2091902494430542,
      "learning_rate": 9.626736842105264e-05,
      "loss": 0.4708,
      "step": 34199
    },
    {
      "epoch": 1.915108074812409,
      "grad_norm": 1.31708562374115,
      "learning_rate": 9.62671052631579e-05,
      "loss": 0.4529,
      "step": 34200
    },
    {
      "epoch": 1.915164072124538,
      "grad_norm": 1.1357673406600952,
      "learning_rate": 9.626684210526317e-05,
      "loss": 0.4211,
      "step": 34201
    },
    {
      "epoch": 1.915220069436667,
      "grad_norm": 1.5314455032348633,
      "learning_rate": 9.626657894736843e-05,
      "loss": 0.4606,
      "step": 34202
    },
    {
      "epoch": 1.915276066748796,
      "grad_norm": 1.7351003885269165,
      "learning_rate": 9.626631578947369e-05,
      "loss": 0.5584,
      "step": 34203
    },
    {
      "epoch": 1.915332064060925,
      "grad_norm": 1.1706600189208984,
      "learning_rate": 9.626605263157895e-05,
      "loss": 0.4167,
      "step": 34204
    },
    {
      "epoch": 1.915388061373054,
      "grad_norm": 1.2618739604949951,
      "learning_rate": 9.626578947368422e-05,
      "loss": 0.3504,
      "step": 34205
    },
    {
      "epoch": 1.915444058685183,
      "grad_norm": 2.1442837715148926,
      "learning_rate": 9.626552631578948e-05,
      "loss": 0.4904,
      "step": 34206
    },
    {
      "epoch": 1.915500055997312,
      "grad_norm": 1.4252557754516602,
      "learning_rate": 9.626526315789474e-05,
      "loss": 0.6217,
      "step": 34207
    },
    {
      "epoch": 1.9155560533094411,
      "grad_norm": 1.251801609992981,
      "learning_rate": 9.6265e-05,
      "loss": 0.4026,
      "step": 34208
    },
    {
      "epoch": 1.9156120506215701,
      "grad_norm": 1.2736146450042725,
      "learning_rate": 9.626473684210526e-05,
      "loss": 0.4117,
      "step": 34209
    },
    {
      "epoch": 1.9156680479336992,
      "grad_norm": 1.229191780090332,
      "learning_rate": 9.626447368421053e-05,
      "loss": 0.3847,
      "step": 34210
    },
    {
      "epoch": 1.9157240452458282,
      "grad_norm": 1.1892379522323608,
      "learning_rate": 9.62642105263158e-05,
      "loss": 0.3912,
      "step": 34211
    },
    {
      "epoch": 1.9157800425579572,
      "grad_norm": 0.9947324395179749,
      "learning_rate": 9.626394736842105e-05,
      "loss": 0.3347,
      "step": 34212
    },
    {
      "epoch": 1.9158360398700862,
      "grad_norm": 1.2207244634628296,
      "learning_rate": 9.626368421052631e-05,
      "loss": 0.4815,
      "step": 34213
    },
    {
      "epoch": 1.9158920371822152,
      "grad_norm": 1.3298957347869873,
      "learning_rate": 9.626342105263159e-05,
      "loss": 0.4308,
      "step": 34214
    },
    {
      "epoch": 1.9159480344943443,
      "grad_norm": 1.2501661777496338,
      "learning_rate": 9.626315789473685e-05,
      "loss": 0.4146,
      "step": 34215
    },
    {
      "epoch": 1.9160040318064733,
      "grad_norm": 1.244620442390442,
      "learning_rate": 9.626289473684212e-05,
      "loss": 0.4207,
      "step": 34216
    },
    {
      "epoch": 1.9160600291186023,
      "grad_norm": 1.4204108715057373,
      "learning_rate": 9.626263157894737e-05,
      "loss": 0.4191,
      "step": 34217
    },
    {
      "epoch": 1.9161160264307313,
      "grad_norm": 1.298211693763733,
      "learning_rate": 9.626236842105264e-05,
      "loss": 0.4649,
      "step": 34218
    },
    {
      "epoch": 1.9161720237428603,
      "grad_norm": 1.203240156173706,
      "learning_rate": 9.62621052631579e-05,
      "loss": 0.4979,
      "step": 34219
    },
    {
      "epoch": 1.9162280210549893,
      "grad_norm": 1.6693533658981323,
      "learning_rate": 9.626184210526317e-05,
      "loss": 0.4998,
      "step": 34220
    },
    {
      "epoch": 1.9162840183671184,
      "grad_norm": 1.4997769594192505,
      "learning_rate": 9.626157894736843e-05,
      "loss": 0.5273,
      "step": 34221
    },
    {
      "epoch": 1.9163400156792474,
      "grad_norm": 1.489009976387024,
      "learning_rate": 9.626131578947369e-05,
      "loss": 0.4585,
      "step": 34222
    },
    {
      "epoch": 1.9163960129913764,
      "grad_norm": 3.5480387210845947,
      "learning_rate": 9.626105263157895e-05,
      "loss": 0.4267,
      "step": 34223
    },
    {
      "epoch": 1.9164520103035054,
      "grad_norm": 1.8703699111938477,
      "learning_rate": 9.626078947368421e-05,
      "loss": 0.494,
      "step": 34224
    },
    {
      "epoch": 1.9165080076156344,
      "grad_norm": 1.5600225925445557,
      "learning_rate": 9.626052631578948e-05,
      "loss": 0.4672,
      "step": 34225
    },
    {
      "epoch": 1.9165640049277635,
      "grad_norm": 1.241592288017273,
      "learning_rate": 9.626026315789473e-05,
      "loss": 0.4302,
      "step": 34226
    },
    {
      "epoch": 1.9166200022398925,
      "grad_norm": 1.4656599760055542,
      "learning_rate": 9.626e-05,
      "loss": 0.6568,
      "step": 34227
    },
    {
      "epoch": 1.9166759995520215,
      "grad_norm": 1.0733925104141235,
      "learning_rate": 9.625973684210526e-05,
      "loss": 0.3584,
      "step": 34228
    },
    {
      "epoch": 1.9167319968641505,
      "grad_norm": 1.3101646900177002,
      "learning_rate": 9.625947368421054e-05,
      "loss": 0.5529,
      "step": 34229
    },
    {
      "epoch": 1.9167879941762795,
      "grad_norm": 1.2405332326889038,
      "learning_rate": 9.62592105263158e-05,
      "loss": 0.4275,
      "step": 34230
    },
    {
      "epoch": 1.9168439914884086,
      "grad_norm": 1.251200556755066,
      "learning_rate": 9.625894736842106e-05,
      "loss": 0.3557,
      "step": 34231
    },
    {
      "epoch": 1.9168999888005376,
      "grad_norm": 1.316773772239685,
      "learning_rate": 9.625868421052632e-05,
      "loss": 0.5058,
      "step": 34232
    },
    {
      "epoch": 1.9169559861126666,
      "grad_norm": 1.567847490310669,
      "learning_rate": 9.625842105263159e-05,
      "loss": 0.4152,
      "step": 34233
    },
    {
      "epoch": 1.9170119834247956,
      "grad_norm": 1.6433348655700684,
      "learning_rate": 9.625815789473685e-05,
      "loss": 0.5167,
      "step": 34234
    },
    {
      "epoch": 1.9170679807369246,
      "grad_norm": 1.2534894943237305,
      "learning_rate": 9.625789473684211e-05,
      "loss": 0.3704,
      "step": 34235
    },
    {
      "epoch": 1.9171239780490537,
      "grad_norm": 1.6590468883514404,
      "learning_rate": 9.625763157894737e-05,
      "loss": 0.4596,
      "step": 34236
    },
    {
      "epoch": 1.9171799753611827,
      "grad_norm": 1.4893572330474854,
      "learning_rate": 9.625736842105264e-05,
      "loss": 0.6242,
      "step": 34237
    },
    {
      "epoch": 1.9172359726733117,
      "grad_norm": 1.1379451751708984,
      "learning_rate": 9.62571052631579e-05,
      "loss": 0.4604,
      "step": 34238
    },
    {
      "epoch": 1.9172919699854407,
      "grad_norm": 1.4879343509674072,
      "learning_rate": 9.625684210526316e-05,
      "loss": 0.6344,
      "step": 34239
    },
    {
      "epoch": 1.9173479672975697,
      "grad_norm": 1.1161693334579468,
      "learning_rate": 9.625657894736842e-05,
      "loss": 0.4123,
      "step": 34240
    },
    {
      "epoch": 1.9174039646096988,
      "grad_norm": 1.2039239406585693,
      "learning_rate": 9.625631578947368e-05,
      "loss": 0.4103,
      "step": 34241
    },
    {
      "epoch": 1.9174599619218278,
      "grad_norm": 1.24535071849823,
      "learning_rate": 9.625605263157895e-05,
      "loss": 0.4896,
      "step": 34242
    },
    {
      "epoch": 1.9175159592339568,
      "grad_norm": 1.0158687829971313,
      "learning_rate": 9.625578947368421e-05,
      "loss": 0.3767,
      "step": 34243
    },
    {
      "epoch": 1.9175719565460858,
      "grad_norm": 1.027435541152954,
      "learning_rate": 9.625552631578947e-05,
      "loss": 0.4488,
      "step": 34244
    },
    {
      "epoch": 1.9176279538582148,
      "grad_norm": 1.675846815109253,
      "learning_rate": 9.625526315789473e-05,
      "loss": 0.6597,
      "step": 34245
    },
    {
      "epoch": 1.9176839511703438,
      "grad_norm": 1.5310463905334473,
      "learning_rate": 9.6255e-05,
      "loss": 0.549,
      "step": 34246
    },
    {
      "epoch": 1.9177399484824729,
      "grad_norm": 1.384231448173523,
      "learning_rate": 9.625473684210527e-05,
      "loss": 0.4122,
      "step": 34247
    },
    {
      "epoch": 1.9177959457946019,
      "grad_norm": 1.0827211141586304,
      "learning_rate": 9.625447368421054e-05,
      "loss": 0.3269,
      "step": 34248
    },
    {
      "epoch": 1.917851943106731,
      "grad_norm": 1.2690229415893555,
      "learning_rate": 9.625421052631579e-05,
      "loss": 0.4455,
      "step": 34249
    },
    {
      "epoch": 1.91790794041886,
      "grad_norm": 1.355848789215088,
      "learning_rate": 9.625394736842106e-05,
      "loss": 0.468,
      "step": 34250
    },
    {
      "epoch": 1.917963937730989,
      "grad_norm": 1.2106949090957642,
      "learning_rate": 9.625368421052632e-05,
      "loss": 0.4324,
      "step": 34251
    },
    {
      "epoch": 1.918019935043118,
      "grad_norm": 1.2169551849365234,
      "learning_rate": 9.625342105263159e-05,
      "loss": 0.5631,
      "step": 34252
    },
    {
      "epoch": 1.918075932355247,
      "grad_norm": 1.2722358703613281,
      "learning_rate": 9.625315789473685e-05,
      "loss": 0.3317,
      "step": 34253
    },
    {
      "epoch": 1.918131929667376,
      "grad_norm": 1.2074919939041138,
      "learning_rate": 9.625289473684211e-05,
      "loss": 0.5792,
      "step": 34254
    },
    {
      "epoch": 1.918187926979505,
      "grad_norm": 1.2313235998153687,
      "learning_rate": 9.625263157894737e-05,
      "loss": 0.4815,
      "step": 34255
    },
    {
      "epoch": 1.918243924291634,
      "grad_norm": 1.1061846017837524,
      "learning_rate": 9.625236842105264e-05,
      "loss": 0.4265,
      "step": 34256
    },
    {
      "epoch": 1.918299921603763,
      "grad_norm": 1.473401427268982,
      "learning_rate": 9.62521052631579e-05,
      "loss": 0.4969,
      "step": 34257
    },
    {
      "epoch": 1.918355918915892,
      "grad_norm": 1.4823957681655884,
      "learning_rate": 9.625184210526316e-05,
      "loss": 0.3321,
      "step": 34258
    },
    {
      "epoch": 1.918411916228021,
      "grad_norm": 1.396756887435913,
      "learning_rate": 9.625157894736842e-05,
      "loss": 0.5285,
      "step": 34259
    },
    {
      "epoch": 1.9184679135401501,
      "grad_norm": 1.2933837175369263,
      "learning_rate": 9.625131578947368e-05,
      "loss": 0.4284,
      "step": 34260
    },
    {
      "epoch": 1.9185239108522791,
      "grad_norm": 1.6208646297454834,
      "learning_rate": 9.625105263157896e-05,
      "loss": 0.496,
      "step": 34261
    },
    {
      "epoch": 1.9185799081644082,
      "grad_norm": 1.1522470712661743,
      "learning_rate": 9.625078947368422e-05,
      "loss": 0.5184,
      "step": 34262
    },
    {
      "epoch": 1.9186359054765372,
      "grad_norm": 1.5574489831924438,
      "learning_rate": 9.625052631578948e-05,
      "loss": 0.3969,
      "step": 34263
    },
    {
      "epoch": 1.9186919027886662,
      "grad_norm": 1.2130779027938843,
      "learning_rate": 9.625026315789474e-05,
      "loss": 0.3794,
      "step": 34264
    },
    {
      "epoch": 1.9187479001007952,
      "grad_norm": 1.3396785259246826,
      "learning_rate": 9.625000000000001e-05,
      "loss": 0.338,
      "step": 34265
    },
    {
      "epoch": 1.9188038974129242,
      "grad_norm": 1.2985972166061401,
      "learning_rate": 9.624973684210527e-05,
      "loss": 0.488,
      "step": 34266
    },
    {
      "epoch": 1.9188598947250533,
      "grad_norm": 1.207932472229004,
      "learning_rate": 9.624947368421053e-05,
      "loss": 0.4687,
      "step": 34267
    },
    {
      "epoch": 1.9189158920371823,
      "grad_norm": 1.3236061334609985,
      "learning_rate": 9.624921052631579e-05,
      "loss": 0.3557,
      "step": 34268
    },
    {
      "epoch": 1.9189718893493113,
      "grad_norm": 1.395732045173645,
      "learning_rate": 9.624894736842106e-05,
      "loss": 0.7787,
      "step": 34269
    },
    {
      "epoch": 1.91902788666144,
      "grad_norm": 1.53282630443573,
      "learning_rate": 9.624868421052632e-05,
      "loss": 0.4336,
      "step": 34270
    },
    {
      "epoch": 1.919083883973569,
      "grad_norm": 1.1582788228988647,
      "learning_rate": 9.62484210526316e-05,
      "loss": 0.3405,
      "step": 34271
    },
    {
      "epoch": 1.9191398812856981,
      "grad_norm": 1.5056872367858887,
      "learning_rate": 9.624815789473684e-05,
      "loss": 0.5928,
      "step": 34272
    },
    {
      "epoch": 1.9191958785978271,
      "grad_norm": 1.3590638637542725,
      "learning_rate": 9.624789473684211e-05,
      "loss": 0.5295,
      "step": 34273
    },
    {
      "epoch": 1.9192518759099562,
      "grad_norm": 1.653045892715454,
      "learning_rate": 9.624763157894737e-05,
      "loss": 0.4595,
      "step": 34274
    },
    {
      "epoch": 1.9193078732220852,
      "grad_norm": 1.2127115726470947,
      "learning_rate": 9.624736842105265e-05,
      "loss": 0.485,
      "step": 34275
    },
    {
      "epoch": 1.9193638705342142,
      "grad_norm": 1.2260850667953491,
      "learning_rate": 9.62471052631579e-05,
      "loss": 0.4518,
      "step": 34276
    },
    {
      "epoch": 1.9194198678463432,
      "grad_norm": 1.5495177507400513,
      "learning_rate": 9.624684210526315e-05,
      "loss": 0.5465,
      "step": 34277
    },
    {
      "epoch": 1.9194758651584722,
      "grad_norm": 1.3879324197769165,
      "learning_rate": 9.624657894736843e-05,
      "loss": 0.6113,
      "step": 34278
    },
    {
      "epoch": 1.9195318624706013,
      "grad_norm": 1.4819798469543457,
      "learning_rate": 9.624631578947369e-05,
      "loss": 0.5579,
      "step": 34279
    },
    {
      "epoch": 1.9195878597827303,
      "grad_norm": 1.6818416118621826,
      "learning_rate": 9.624605263157896e-05,
      "loss": 0.7239,
      "step": 34280
    },
    {
      "epoch": 1.9196438570948593,
      "grad_norm": 1.1855067014694214,
      "learning_rate": 9.62457894736842e-05,
      "loss": 0.425,
      "step": 34281
    },
    {
      "epoch": 1.9196998544069883,
      "grad_norm": 1.3974952697753906,
      "learning_rate": 9.624552631578948e-05,
      "loss": 0.6626,
      "step": 34282
    },
    {
      "epoch": 1.9197558517191173,
      "grad_norm": 1.5829144716262817,
      "learning_rate": 9.624526315789474e-05,
      "loss": 0.5308,
      "step": 34283
    },
    {
      "epoch": 1.9198118490312464,
      "grad_norm": 1.2658365964889526,
      "learning_rate": 9.624500000000001e-05,
      "loss": 0.4676,
      "step": 34284
    },
    {
      "epoch": 1.9198678463433754,
      "grad_norm": 1.507955551147461,
      "learning_rate": 9.624473684210527e-05,
      "loss": 0.5393,
      "step": 34285
    },
    {
      "epoch": 1.9199238436555044,
      "grad_norm": 1.2962963581085205,
      "learning_rate": 9.624447368421053e-05,
      "loss": 0.442,
      "step": 34286
    },
    {
      "epoch": 1.9199798409676334,
      "grad_norm": 1.0144649744033813,
      "learning_rate": 9.624421052631579e-05,
      "loss": 0.3287,
      "step": 34287
    },
    {
      "epoch": 1.9200358382797624,
      "grad_norm": 1.274845004081726,
      "learning_rate": 9.624394736842106e-05,
      "loss": 0.3465,
      "step": 34288
    },
    {
      "epoch": 1.9200918355918914,
      "grad_norm": 1.1636922359466553,
      "learning_rate": 9.624368421052632e-05,
      "loss": 0.3874,
      "step": 34289
    },
    {
      "epoch": 1.9201478329040205,
      "grad_norm": 1.9486541748046875,
      "learning_rate": 9.624342105263158e-05,
      "loss": 0.3908,
      "step": 34290
    },
    {
      "epoch": 1.9202038302161495,
      "grad_norm": 1.5582003593444824,
      "learning_rate": 9.624315789473684e-05,
      "loss": 0.6293,
      "step": 34291
    },
    {
      "epoch": 1.9202598275282785,
      "grad_norm": 1.5689884424209595,
      "learning_rate": 9.624289473684212e-05,
      "loss": 0.519,
      "step": 34292
    },
    {
      "epoch": 1.9203158248404075,
      "grad_norm": 1.3671067953109741,
      "learning_rate": 9.624263157894738e-05,
      "loss": 0.4712,
      "step": 34293
    },
    {
      "epoch": 1.9203718221525365,
      "grad_norm": 1.3742132186889648,
      "learning_rate": 9.624236842105264e-05,
      "loss": 0.3905,
      "step": 34294
    },
    {
      "epoch": 1.9204278194646656,
      "grad_norm": 2.128695011138916,
      "learning_rate": 9.62421052631579e-05,
      "loss": 0.4296,
      "step": 34295
    },
    {
      "epoch": 1.9204838167767946,
      "grad_norm": 1.2733598947525024,
      "learning_rate": 9.624184210526315e-05,
      "loss": 0.4557,
      "step": 34296
    },
    {
      "epoch": 1.9205398140889236,
      "grad_norm": 1.1369787454605103,
      "learning_rate": 9.624157894736843e-05,
      "loss": 0.3241,
      "step": 34297
    },
    {
      "epoch": 1.9205958114010526,
      "grad_norm": 1.1625676155090332,
      "learning_rate": 9.624131578947369e-05,
      "loss": 0.4057,
      "step": 34298
    },
    {
      "epoch": 1.9206518087131816,
      "grad_norm": 1.3650845289230347,
      "learning_rate": 9.624105263157895e-05,
      "loss": 0.5414,
      "step": 34299
    },
    {
      "epoch": 1.9207078060253107,
      "grad_norm": 1.284866213798523,
      "learning_rate": 9.624078947368421e-05,
      "loss": 0.615,
      "step": 34300
    },
    {
      "epoch": 1.9207638033374397,
      "grad_norm": 1.1933172941207886,
      "learning_rate": 9.624052631578948e-05,
      "loss": 0.4239,
      "step": 34301
    },
    {
      "epoch": 1.9208198006495687,
      "grad_norm": 1.2017443180084229,
      "learning_rate": 9.624026315789474e-05,
      "loss": 0.4367,
      "step": 34302
    },
    {
      "epoch": 1.9208757979616977,
      "grad_norm": 1.1451778411865234,
      "learning_rate": 9.624000000000001e-05,
      "loss": 0.4988,
      "step": 34303
    },
    {
      "epoch": 1.9209317952738267,
      "grad_norm": 1.1466450691223145,
      "learning_rate": 9.623973684210526e-05,
      "loss": 0.3327,
      "step": 34304
    },
    {
      "epoch": 1.9209877925859558,
      "grad_norm": 1.1683193445205688,
      "learning_rate": 9.623947368421053e-05,
      "loss": 0.4286,
      "step": 34305
    },
    {
      "epoch": 1.9210437898980848,
      "grad_norm": 2.1987383365631104,
      "learning_rate": 9.623921052631579e-05,
      "loss": 0.5721,
      "step": 34306
    },
    {
      "epoch": 1.9210997872102138,
      "grad_norm": 1.4699146747589111,
      "learning_rate": 9.623894736842107e-05,
      "loss": 0.3228,
      "step": 34307
    },
    {
      "epoch": 1.9211557845223428,
      "grad_norm": 1.1275240182876587,
      "learning_rate": 9.623868421052633e-05,
      "loss": 0.4266,
      "step": 34308
    },
    {
      "epoch": 1.9212117818344718,
      "grad_norm": 1.1163877248764038,
      "learning_rate": 9.623842105263159e-05,
      "loss": 0.3134,
      "step": 34309
    },
    {
      "epoch": 1.9212677791466009,
      "grad_norm": 1.2665023803710938,
      "learning_rate": 9.623815789473685e-05,
      "loss": 0.4593,
      "step": 34310
    },
    {
      "epoch": 1.9213237764587299,
      "grad_norm": 1.1689059734344482,
      "learning_rate": 9.62378947368421e-05,
      "loss": 0.3862,
      "step": 34311
    },
    {
      "epoch": 1.921379773770859,
      "grad_norm": 2.7677831649780273,
      "learning_rate": 9.623763157894738e-05,
      "loss": 0.6233,
      "step": 34312
    },
    {
      "epoch": 1.921435771082988,
      "grad_norm": 1.601834774017334,
      "learning_rate": 9.623736842105264e-05,
      "loss": 0.4761,
      "step": 34313
    },
    {
      "epoch": 1.921491768395117,
      "grad_norm": 1.2268656492233276,
      "learning_rate": 9.62371052631579e-05,
      "loss": 0.4992,
      "step": 34314
    },
    {
      "epoch": 1.921547765707246,
      "grad_norm": 1.2537119388580322,
      "learning_rate": 9.623684210526316e-05,
      "loss": 0.3951,
      "step": 34315
    },
    {
      "epoch": 1.921603763019375,
      "grad_norm": 1.186688780784607,
      "learning_rate": 9.623657894736843e-05,
      "loss": 0.4199,
      "step": 34316
    },
    {
      "epoch": 1.921659760331504,
      "grad_norm": 1.0517016649246216,
      "learning_rate": 9.623631578947369e-05,
      "loss": 0.3358,
      "step": 34317
    },
    {
      "epoch": 1.921715757643633,
      "grad_norm": 1.111101508140564,
      "learning_rate": 9.623605263157895e-05,
      "loss": 0.5862,
      "step": 34318
    },
    {
      "epoch": 1.921771754955762,
      "grad_norm": 1.5175127983093262,
      "learning_rate": 9.623578947368421e-05,
      "loss": 0.5376,
      "step": 34319
    },
    {
      "epoch": 1.921827752267891,
      "grad_norm": 1.1358342170715332,
      "learning_rate": 9.623552631578948e-05,
      "loss": 0.4517,
      "step": 34320
    },
    {
      "epoch": 1.92188374958002,
      "grad_norm": 1.2149403095245361,
      "learning_rate": 9.623526315789474e-05,
      "loss": 0.3261,
      "step": 34321
    },
    {
      "epoch": 1.921939746892149,
      "grad_norm": 1.4926055669784546,
      "learning_rate": 9.6235e-05,
      "loss": 0.3937,
      "step": 34322
    },
    {
      "epoch": 1.921995744204278,
      "grad_norm": 1.2512887716293335,
      "learning_rate": 9.623473684210526e-05,
      "loss": 0.4034,
      "step": 34323
    },
    {
      "epoch": 1.9220517415164071,
      "grad_norm": 1.4327195882797241,
      "learning_rate": 9.623447368421054e-05,
      "loss": 0.5098,
      "step": 34324
    },
    {
      "epoch": 1.9221077388285361,
      "grad_norm": 3.018702745437622,
      "learning_rate": 9.62342105263158e-05,
      "loss": 0.4752,
      "step": 34325
    },
    {
      "epoch": 1.9221637361406652,
      "grad_norm": 1.2417402267456055,
      "learning_rate": 9.623394736842107e-05,
      "loss": 0.4311,
      "step": 34326
    },
    {
      "epoch": 1.9222197334527942,
      "grad_norm": 1.3895093202590942,
      "learning_rate": 9.623368421052631e-05,
      "loss": 0.7125,
      "step": 34327
    },
    {
      "epoch": 1.9222757307649232,
      "grad_norm": 1.3928189277648926,
      "learning_rate": 9.623342105263157e-05,
      "loss": 0.4179,
      "step": 34328
    },
    {
      "epoch": 1.9223317280770522,
      "grad_norm": 1.251164197921753,
      "learning_rate": 9.623315789473685e-05,
      "loss": 0.4703,
      "step": 34329
    },
    {
      "epoch": 1.9223877253891812,
      "grad_norm": 1.836476445198059,
      "learning_rate": 9.623289473684211e-05,
      "loss": 0.3875,
      "step": 34330
    },
    {
      "epoch": 1.9224437227013103,
      "grad_norm": 1.1445401906967163,
      "learning_rate": 9.623263157894738e-05,
      "loss": 0.4549,
      "step": 34331
    },
    {
      "epoch": 1.9224997200134393,
      "grad_norm": 1.2746981382369995,
      "learning_rate": 9.623236842105263e-05,
      "loss": 0.5712,
      "step": 34332
    },
    {
      "epoch": 1.9225557173255683,
      "grad_norm": 1.4008656740188599,
      "learning_rate": 9.62321052631579e-05,
      "loss": 0.5445,
      "step": 34333
    },
    {
      "epoch": 1.9226117146376973,
      "grad_norm": 1.5213289260864258,
      "learning_rate": 9.623184210526316e-05,
      "loss": 0.4742,
      "step": 34334
    },
    {
      "epoch": 1.9226677119498263,
      "grad_norm": 1.2487987279891968,
      "learning_rate": 9.623157894736843e-05,
      "loss": 0.4239,
      "step": 34335
    },
    {
      "epoch": 1.9227237092619553,
      "grad_norm": 1.5003831386566162,
      "learning_rate": 9.623131578947368e-05,
      "loss": 0.6254,
      "step": 34336
    },
    {
      "epoch": 1.9227797065740844,
      "grad_norm": 1.2354856729507446,
      "learning_rate": 9.623105263157895e-05,
      "loss": 0.4269,
      "step": 34337
    },
    {
      "epoch": 1.9228357038862134,
      "grad_norm": 1.4254043102264404,
      "learning_rate": 9.623078947368421e-05,
      "loss": 0.6442,
      "step": 34338
    },
    {
      "epoch": 1.9228917011983424,
      "grad_norm": 1.3578306436538696,
      "learning_rate": 9.623052631578949e-05,
      "loss": 0.4991,
      "step": 34339
    },
    {
      "epoch": 1.9229476985104714,
      "grad_norm": 1.0492668151855469,
      "learning_rate": 9.623026315789475e-05,
      "loss": 0.3883,
      "step": 34340
    },
    {
      "epoch": 1.9230036958226004,
      "grad_norm": 1.4757280349731445,
      "learning_rate": 9.623e-05,
      "loss": 0.4509,
      "step": 34341
    },
    {
      "epoch": 1.9230596931347295,
      "grad_norm": 1.4086908102035522,
      "learning_rate": 9.622973684210527e-05,
      "loss": 0.4788,
      "step": 34342
    },
    {
      "epoch": 1.9231156904468585,
      "grad_norm": 1.3750075101852417,
      "learning_rate": 9.622947368421054e-05,
      "loss": 0.5515,
      "step": 34343
    },
    {
      "epoch": 1.9231716877589875,
      "grad_norm": 1.1854740381240845,
      "learning_rate": 9.62292105263158e-05,
      "loss": 0.4634,
      "step": 34344
    },
    {
      "epoch": 1.9232276850711165,
      "grad_norm": 1.2228724956512451,
      "learning_rate": 9.622894736842106e-05,
      "loss": 0.4002,
      "step": 34345
    },
    {
      "epoch": 1.9232836823832455,
      "grad_norm": 1.1499624252319336,
      "learning_rate": 9.622868421052632e-05,
      "loss": 0.5372,
      "step": 34346
    },
    {
      "epoch": 1.9233396796953746,
      "grad_norm": 1.2971465587615967,
      "learning_rate": 9.622842105263158e-05,
      "loss": 0.4611,
      "step": 34347
    },
    {
      "epoch": 1.9233956770075036,
      "grad_norm": 1.2563623189926147,
      "learning_rate": 9.622815789473685e-05,
      "loss": 0.4014,
      "step": 34348
    },
    {
      "epoch": 1.9234516743196326,
      "grad_norm": 1.2086490392684937,
      "learning_rate": 9.622789473684211e-05,
      "loss": 0.5549,
      "step": 34349
    },
    {
      "epoch": 1.9235076716317616,
      "grad_norm": 1.3588988780975342,
      "learning_rate": 9.622763157894737e-05,
      "loss": 0.5435,
      "step": 34350
    },
    {
      "epoch": 1.9235636689438906,
      "grad_norm": 1.1672650575637817,
      "learning_rate": 9.622736842105263e-05,
      "loss": 0.6131,
      "step": 34351
    },
    {
      "epoch": 1.9236196662560197,
      "grad_norm": 1.523383617401123,
      "learning_rate": 9.62271052631579e-05,
      "loss": 0.4516,
      "step": 34352
    },
    {
      "epoch": 1.9236756635681487,
      "grad_norm": 1.1716790199279785,
      "learning_rate": 9.622684210526316e-05,
      "loss": 0.4739,
      "step": 34353
    },
    {
      "epoch": 1.9237316608802777,
      "grad_norm": 1.0879019498825073,
      "learning_rate": 9.622657894736842e-05,
      "loss": 0.442,
      "step": 34354
    },
    {
      "epoch": 1.9237876581924067,
      "grad_norm": 1.4493409395217896,
      "learning_rate": 9.622631578947368e-05,
      "loss": 0.4492,
      "step": 34355
    },
    {
      "epoch": 1.9238436555045357,
      "grad_norm": 1.2001742124557495,
      "learning_rate": 9.622605263157896e-05,
      "loss": 0.4314,
      "step": 34356
    },
    {
      "epoch": 1.9238996528166648,
      "grad_norm": 1.4988598823547363,
      "learning_rate": 9.622578947368422e-05,
      "loss": 0.3817,
      "step": 34357
    },
    {
      "epoch": 1.9239556501287938,
      "grad_norm": 1.5891660451889038,
      "learning_rate": 9.622552631578949e-05,
      "loss": 0.5123,
      "step": 34358
    },
    {
      "epoch": 1.9240116474409228,
      "grad_norm": 3.146915912628174,
      "learning_rate": 9.622526315789473e-05,
      "loss": 0.4199,
      "step": 34359
    },
    {
      "epoch": 1.9240676447530518,
      "grad_norm": 1.290448784828186,
      "learning_rate": 9.622500000000001e-05,
      "loss": 0.4193,
      "step": 34360
    },
    {
      "epoch": 1.9241236420651808,
      "grad_norm": 1.2967422008514404,
      "learning_rate": 9.622473684210527e-05,
      "loss": 0.6199,
      "step": 34361
    },
    {
      "epoch": 1.9241796393773098,
      "grad_norm": 1.1139644384384155,
      "learning_rate": 9.622447368421053e-05,
      "loss": 0.4343,
      "step": 34362
    },
    {
      "epoch": 1.9242356366894389,
      "grad_norm": 2.0440478324890137,
      "learning_rate": 9.62242105263158e-05,
      "loss": 0.4569,
      "step": 34363
    },
    {
      "epoch": 1.9242916340015679,
      "grad_norm": 1.8078516721725464,
      "learning_rate": 9.622394736842105e-05,
      "loss": 0.5124,
      "step": 34364
    },
    {
      "epoch": 1.924347631313697,
      "grad_norm": 1.470478892326355,
      "learning_rate": 9.622368421052632e-05,
      "loss": 0.4409,
      "step": 34365
    },
    {
      "epoch": 1.924403628625826,
      "grad_norm": 1.3908190727233887,
      "learning_rate": 9.622342105263158e-05,
      "loss": 0.4864,
      "step": 34366
    },
    {
      "epoch": 1.924459625937955,
      "grad_norm": 1.82606041431427,
      "learning_rate": 9.622315789473685e-05,
      "loss": 0.6269,
      "step": 34367
    },
    {
      "epoch": 1.924515623250084,
      "grad_norm": 1.1784051656723022,
      "learning_rate": 9.622289473684211e-05,
      "loss": 0.4439,
      "step": 34368
    },
    {
      "epoch": 1.924571620562213,
      "grad_norm": 2.00205397605896,
      "learning_rate": 9.622263157894737e-05,
      "loss": 0.6574,
      "step": 34369
    },
    {
      "epoch": 1.924627617874342,
      "grad_norm": 1.235109806060791,
      "learning_rate": 9.622236842105263e-05,
      "loss": 0.5686,
      "step": 34370
    },
    {
      "epoch": 1.924683615186471,
      "grad_norm": 1.3443928956985474,
      "learning_rate": 9.62221052631579e-05,
      "loss": 0.4126,
      "step": 34371
    },
    {
      "epoch": 1.9247396124986,
      "grad_norm": 1.8188425302505493,
      "learning_rate": 9.622184210526317e-05,
      "loss": 0.63,
      "step": 34372
    },
    {
      "epoch": 1.924795609810729,
      "grad_norm": 0.9170876741409302,
      "learning_rate": 9.622157894736843e-05,
      "loss": 0.3381,
      "step": 34373
    },
    {
      "epoch": 1.924851607122858,
      "grad_norm": 1.3750289678573608,
      "learning_rate": 9.622131578947368e-05,
      "loss": 0.4304,
      "step": 34374
    },
    {
      "epoch": 1.924907604434987,
      "grad_norm": 1.1242152452468872,
      "learning_rate": 9.622105263157896e-05,
      "loss": 0.3476,
      "step": 34375
    },
    {
      "epoch": 1.9249636017471161,
      "grad_norm": 1.2126795053482056,
      "learning_rate": 9.622078947368422e-05,
      "loss": 0.3974,
      "step": 34376
    },
    {
      "epoch": 1.9250195990592451,
      "grad_norm": 1.937725305557251,
      "learning_rate": 9.622052631578948e-05,
      "loss": 0.4846,
      "step": 34377
    },
    {
      "epoch": 1.9250755963713742,
      "grad_norm": 1.781497597694397,
      "learning_rate": 9.622026315789474e-05,
      "loss": 0.512,
      "step": 34378
    },
    {
      "epoch": 1.9251315936835032,
      "grad_norm": 1.374320149421692,
      "learning_rate": 9.622000000000001e-05,
      "loss": 0.4105,
      "step": 34379
    },
    {
      "epoch": 1.9251875909956322,
      "grad_norm": 1.2288329601287842,
      "learning_rate": 9.621973684210527e-05,
      "loss": 0.4994,
      "step": 34380
    },
    {
      "epoch": 1.9252435883077612,
      "grad_norm": 1.423977017402649,
      "learning_rate": 9.621947368421053e-05,
      "loss": 0.5332,
      "step": 34381
    },
    {
      "epoch": 1.9252995856198902,
      "grad_norm": 1.233642339706421,
      "learning_rate": 9.621921052631579e-05,
      "loss": 0.4645,
      "step": 34382
    },
    {
      "epoch": 1.9253555829320192,
      "grad_norm": 1.2181953191757202,
      "learning_rate": 9.621894736842105e-05,
      "loss": 0.3515,
      "step": 34383
    },
    {
      "epoch": 1.9254115802441483,
      "grad_norm": 1.1944061517715454,
      "learning_rate": 9.621868421052632e-05,
      "loss": 0.3579,
      "step": 34384
    },
    {
      "epoch": 1.9254675775562773,
      "grad_norm": 1.306414246559143,
      "learning_rate": 9.621842105263158e-05,
      "loss": 0.4828,
      "step": 34385
    },
    {
      "epoch": 1.9255235748684063,
      "grad_norm": 1.5317976474761963,
      "learning_rate": 9.621815789473686e-05,
      "loss": 0.5148,
      "step": 34386
    },
    {
      "epoch": 1.9255795721805353,
      "grad_norm": 1.1759620904922485,
      "learning_rate": 9.62178947368421e-05,
      "loss": 0.3584,
      "step": 34387
    },
    {
      "epoch": 1.9256355694926643,
      "grad_norm": 1.2523274421691895,
      "learning_rate": 9.621763157894738e-05,
      "loss": 0.4177,
      "step": 34388
    },
    {
      "epoch": 1.9256915668047934,
      "grad_norm": 1.1623507738113403,
      "learning_rate": 9.621736842105263e-05,
      "loss": 0.4529,
      "step": 34389
    },
    {
      "epoch": 1.9257475641169224,
      "grad_norm": 1.6678543090820312,
      "learning_rate": 9.621710526315791e-05,
      "loss": 0.517,
      "step": 34390
    },
    {
      "epoch": 1.9258035614290514,
      "grad_norm": 1.3336390256881714,
      "learning_rate": 9.621684210526315e-05,
      "loss": 0.6577,
      "step": 34391
    },
    {
      "epoch": 1.9258595587411804,
      "grad_norm": 1.9441115856170654,
      "learning_rate": 9.621657894736843e-05,
      "loss": 0.5595,
      "step": 34392
    },
    {
      "epoch": 1.9259155560533094,
      "grad_norm": 1.3937158584594727,
      "learning_rate": 9.621631578947369e-05,
      "loss": 0.515,
      "step": 34393
    },
    {
      "epoch": 1.9259715533654385,
      "grad_norm": 1.1185667514801025,
      "learning_rate": 9.621605263157896e-05,
      "loss": 0.372,
      "step": 34394
    },
    {
      "epoch": 1.9260275506775675,
      "grad_norm": 1.1725096702575684,
      "learning_rate": 9.621578947368422e-05,
      "loss": 0.3736,
      "step": 34395
    },
    {
      "epoch": 1.9260835479896965,
      "grad_norm": 1.2225680351257324,
      "learning_rate": 9.621552631578948e-05,
      "loss": 0.434,
      "step": 34396
    },
    {
      "epoch": 1.9261395453018255,
      "grad_norm": 1.1347873210906982,
      "learning_rate": 9.621526315789474e-05,
      "loss": 0.3333,
      "step": 34397
    },
    {
      "epoch": 1.9261955426139545,
      "grad_norm": 1.5155844688415527,
      "learning_rate": 9.6215e-05,
      "loss": 0.4664,
      "step": 34398
    },
    {
      "epoch": 1.9262515399260836,
      "grad_norm": 1.376240611076355,
      "learning_rate": 9.621473684210527e-05,
      "loss": 0.4451,
      "step": 34399
    },
    {
      "epoch": 1.9263075372382126,
      "grad_norm": 1.0406979322433472,
      "learning_rate": 9.621447368421053e-05,
      "loss": 0.3307,
      "step": 34400
    },
    {
      "epoch": 1.9263635345503416,
      "grad_norm": 1.2049260139465332,
      "learning_rate": 9.621421052631579e-05,
      "loss": 0.5193,
      "step": 34401
    },
    {
      "epoch": 1.9264195318624706,
      "grad_norm": 1.3064241409301758,
      "learning_rate": 9.621394736842105e-05,
      "loss": 0.4781,
      "step": 34402
    },
    {
      "epoch": 1.9264755291745996,
      "grad_norm": 1.2402961254119873,
      "learning_rate": 9.621368421052633e-05,
      "loss": 0.4365,
      "step": 34403
    },
    {
      "epoch": 1.9265315264867287,
      "grad_norm": 1.203587532043457,
      "learning_rate": 9.621342105263158e-05,
      "loss": 0.3792,
      "step": 34404
    },
    {
      "epoch": 1.9265875237988577,
      "grad_norm": 1.3120694160461426,
      "learning_rate": 9.621315789473684e-05,
      "loss": 0.412,
      "step": 34405
    },
    {
      "epoch": 1.9266435211109867,
      "grad_norm": 1.1383551359176636,
      "learning_rate": 9.62128947368421e-05,
      "loss": 0.3611,
      "step": 34406
    },
    {
      "epoch": 1.9266995184231157,
      "grad_norm": 1.8300246000289917,
      "learning_rate": 9.621263157894738e-05,
      "loss": 0.577,
      "step": 34407
    },
    {
      "epoch": 1.9267555157352447,
      "grad_norm": 1.2871698141098022,
      "learning_rate": 9.621236842105264e-05,
      "loss": 0.405,
      "step": 34408
    },
    {
      "epoch": 1.9268115130473737,
      "grad_norm": 1.2269333600997925,
      "learning_rate": 9.62121052631579e-05,
      "loss": 0.4462,
      "step": 34409
    },
    {
      "epoch": 1.9268675103595028,
      "grad_norm": 1.1481660604476929,
      "learning_rate": 9.621184210526316e-05,
      "loss": 0.4579,
      "step": 34410
    },
    {
      "epoch": 1.9269235076716318,
      "grad_norm": 1.5203206539154053,
      "learning_rate": 9.621157894736843e-05,
      "loss": 0.4338,
      "step": 34411
    },
    {
      "epoch": 1.9269795049837608,
      "grad_norm": 1.0951775312423706,
      "learning_rate": 9.621131578947369e-05,
      "loss": 0.4529,
      "step": 34412
    },
    {
      "epoch": 1.9270355022958898,
      "grad_norm": 1.3437350988388062,
      "learning_rate": 9.621105263157896e-05,
      "loss": 0.4369,
      "step": 34413
    },
    {
      "epoch": 1.9270914996080188,
      "grad_norm": 1.1797075271606445,
      "learning_rate": 9.621078947368421e-05,
      "loss": 0.4991,
      "step": 34414
    },
    {
      "epoch": 1.9271474969201479,
      "grad_norm": 1.4596105813980103,
      "learning_rate": 9.621052631578947e-05,
      "loss": 0.4578,
      "step": 34415
    },
    {
      "epoch": 1.9272034942322769,
      "grad_norm": 1.2185550928115845,
      "learning_rate": 9.621026315789474e-05,
      "loss": 0.3894,
      "step": 34416
    },
    {
      "epoch": 1.927259491544406,
      "grad_norm": 1.472474217414856,
      "learning_rate": 9.621e-05,
      "loss": 0.3932,
      "step": 34417
    },
    {
      "epoch": 1.927315488856535,
      "grad_norm": 1.316421627998352,
      "learning_rate": 9.620973684210528e-05,
      "loss": 0.4032,
      "step": 34418
    },
    {
      "epoch": 1.927371486168664,
      "grad_norm": 1.1761480569839478,
      "learning_rate": 9.620947368421052e-05,
      "loss": 0.4049,
      "step": 34419
    },
    {
      "epoch": 1.927427483480793,
      "grad_norm": 1.1489301919937134,
      "learning_rate": 9.62092105263158e-05,
      "loss": 0.4376,
      "step": 34420
    },
    {
      "epoch": 1.927483480792922,
      "grad_norm": 1.3674211502075195,
      "learning_rate": 9.620894736842105e-05,
      "loss": 0.324,
      "step": 34421
    },
    {
      "epoch": 1.927539478105051,
      "grad_norm": 1.2294304370880127,
      "learning_rate": 9.620868421052633e-05,
      "loss": 0.5572,
      "step": 34422
    },
    {
      "epoch": 1.92759547541718,
      "grad_norm": 1.4763152599334717,
      "learning_rate": 9.620842105263159e-05,
      "loss": 0.4053,
      "step": 34423
    },
    {
      "epoch": 1.927651472729309,
      "grad_norm": 1.1901642084121704,
      "learning_rate": 9.620815789473685e-05,
      "loss": 0.4247,
      "step": 34424
    },
    {
      "epoch": 1.927707470041438,
      "grad_norm": 1.2052173614501953,
      "learning_rate": 9.620789473684211e-05,
      "loss": 0.4945,
      "step": 34425
    },
    {
      "epoch": 1.927763467353567,
      "grad_norm": 2.0695602893829346,
      "learning_rate": 9.620763157894738e-05,
      "loss": 0.464,
      "step": 34426
    },
    {
      "epoch": 1.927819464665696,
      "grad_norm": 1.0273263454437256,
      "learning_rate": 9.620736842105264e-05,
      "loss": 0.307,
      "step": 34427
    },
    {
      "epoch": 1.9278754619778251,
      "grad_norm": 1.4338146448135376,
      "learning_rate": 9.62071052631579e-05,
      "loss": 0.5388,
      "step": 34428
    },
    {
      "epoch": 1.9279314592899541,
      "grad_norm": 1.180341362953186,
      "learning_rate": 9.620684210526316e-05,
      "loss": 0.3366,
      "step": 34429
    },
    {
      "epoch": 1.9279874566020831,
      "grad_norm": 1.208812952041626,
      "learning_rate": 9.620657894736843e-05,
      "loss": 0.4029,
      "step": 34430
    },
    {
      "epoch": 1.9280434539142122,
      "grad_norm": 1.3337002992630005,
      "learning_rate": 9.620631578947369e-05,
      "loss": 0.5443,
      "step": 34431
    },
    {
      "epoch": 1.9280994512263412,
      "grad_norm": 1.5183266401290894,
      "learning_rate": 9.620605263157895e-05,
      "loss": 0.6829,
      "step": 34432
    },
    {
      "epoch": 1.9281554485384702,
      "grad_norm": 0.9501885175704956,
      "learning_rate": 9.620578947368421e-05,
      "loss": 0.2856,
      "step": 34433
    },
    {
      "epoch": 1.9282114458505992,
      "grad_norm": 1.473787546157837,
      "learning_rate": 9.620552631578947e-05,
      "loss": 0.6641,
      "step": 34434
    },
    {
      "epoch": 1.9282674431627282,
      "grad_norm": 1.1880011558532715,
      "learning_rate": 9.620526315789474e-05,
      "loss": 0.4315,
      "step": 34435
    },
    {
      "epoch": 1.9283234404748573,
      "grad_norm": 3.8558762073516846,
      "learning_rate": 9.6205e-05,
      "loss": 0.432,
      "step": 34436
    },
    {
      "epoch": 1.9283794377869863,
      "grad_norm": 1.1921058893203735,
      "learning_rate": 9.620473684210526e-05,
      "loss": 0.4802,
      "step": 34437
    },
    {
      "epoch": 1.9284354350991153,
      "grad_norm": 1.0664736032485962,
      "learning_rate": 9.620447368421052e-05,
      "loss": 0.4363,
      "step": 34438
    },
    {
      "epoch": 1.9284914324112443,
      "grad_norm": 1.2945587635040283,
      "learning_rate": 9.62042105263158e-05,
      "loss": 0.3728,
      "step": 34439
    },
    {
      "epoch": 1.9285474297233733,
      "grad_norm": 1.3429768085479736,
      "learning_rate": 9.620394736842106e-05,
      "loss": 0.4986,
      "step": 34440
    },
    {
      "epoch": 1.9286034270355024,
      "grad_norm": 1.4482563734054565,
      "learning_rate": 9.620368421052632e-05,
      "loss": 0.3196,
      "step": 34441
    },
    {
      "epoch": 1.9286594243476314,
      "grad_norm": 1.043186902999878,
      "learning_rate": 9.620342105263158e-05,
      "loss": 0.3654,
      "step": 34442
    },
    {
      "epoch": 1.9287154216597604,
      "grad_norm": 1.4350152015686035,
      "learning_rate": 9.620315789473685e-05,
      "loss": 0.6509,
      "step": 34443
    },
    {
      "epoch": 1.9287714189718894,
      "grad_norm": 1.1923483610153198,
      "learning_rate": 9.620289473684211e-05,
      "loss": 0.3241,
      "step": 34444
    },
    {
      "epoch": 1.9288274162840184,
      "grad_norm": 1.2620973587036133,
      "learning_rate": 9.620263157894738e-05,
      "loss": 0.4405,
      "step": 34445
    },
    {
      "epoch": 1.9288834135961475,
      "grad_norm": 1.403353214263916,
      "learning_rate": 9.620236842105263e-05,
      "loss": 0.4476,
      "step": 34446
    },
    {
      "epoch": 1.9289394109082765,
      "grad_norm": 1.1871943473815918,
      "learning_rate": 9.62021052631579e-05,
      "loss": 0.5248,
      "step": 34447
    },
    {
      "epoch": 1.9289954082204055,
      "grad_norm": 1.3543674945831299,
      "learning_rate": 9.620184210526316e-05,
      "loss": 0.4392,
      "step": 34448
    },
    {
      "epoch": 1.9290514055325345,
      "grad_norm": 1.3094923496246338,
      "learning_rate": 9.620157894736842e-05,
      "loss": 0.4446,
      "step": 34449
    },
    {
      "epoch": 1.9291074028446635,
      "grad_norm": 1.3161166906356812,
      "learning_rate": 9.62013157894737e-05,
      "loss": 0.4432,
      "step": 34450
    },
    {
      "epoch": 1.9291634001567926,
      "grad_norm": 1.849827766418457,
      "learning_rate": 9.620105263157894e-05,
      "loss": 0.4328,
      "step": 34451
    },
    {
      "epoch": 1.9292193974689216,
      "grad_norm": 1.1753054857254028,
      "learning_rate": 9.620078947368421e-05,
      "loss": 0.3896,
      "step": 34452
    },
    {
      "epoch": 1.9292753947810506,
      "grad_norm": 1.3388580083847046,
      "learning_rate": 9.620052631578947e-05,
      "loss": 0.5128,
      "step": 34453
    },
    {
      "epoch": 1.9293313920931796,
      "grad_norm": 1.2258212566375732,
      "learning_rate": 9.620026315789475e-05,
      "loss": 0.3657,
      "step": 34454
    },
    {
      "epoch": 1.9293873894053086,
      "grad_norm": 1.276568055152893,
      "learning_rate": 9.620000000000001e-05,
      "loss": 0.3678,
      "step": 34455
    },
    {
      "epoch": 1.9294433867174376,
      "grad_norm": 1.1893410682678223,
      "learning_rate": 9.619973684210527e-05,
      "loss": 0.4745,
      "step": 34456
    },
    {
      "epoch": 1.9294993840295667,
      "grad_norm": 1.2533750534057617,
      "learning_rate": 9.619947368421053e-05,
      "loss": 0.4409,
      "step": 34457
    },
    {
      "epoch": 1.9295553813416957,
      "grad_norm": 7.574336528778076,
      "learning_rate": 9.61992105263158e-05,
      "loss": 0.4121,
      "step": 34458
    },
    {
      "epoch": 1.9296113786538247,
      "grad_norm": 1.3594753742218018,
      "learning_rate": 9.619894736842106e-05,
      "loss": 0.411,
      "step": 34459
    },
    {
      "epoch": 1.9296673759659537,
      "grad_norm": 1.322891116142273,
      "learning_rate": 9.619868421052632e-05,
      "loss": 0.3755,
      "step": 34460
    },
    {
      "epoch": 1.9297233732780827,
      "grad_norm": 1.327690601348877,
      "learning_rate": 9.619842105263158e-05,
      "loss": 0.4629,
      "step": 34461
    },
    {
      "epoch": 1.9297793705902118,
      "grad_norm": 1.4175219535827637,
      "learning_rate": 9.619815789473685e-05,
      "loss": 0.5439,
      "step": 34462
    },
    {
      "epoch": 1.9298353679023408,
      "grad_norm": 1.6120556592941284,
      "learning_rate": 9.619789473684211e-05,
      "loss": 0.4773,
      "step": 34463
    },
    {
      "epoch": 1.9298913652144698,
      "grad_norm": 1.348920464515686,
      "learning_rate": 9.619763157894737e-05,
      "loss": 0.3932,
      "step": 34464
    },
    {
      "epoch": 1.9299473625265988,
      "grad_norm": 1.186859369277954,
      "learning_rate": 9.619736842105263e-05,
      "loss": 0.5529,
      "step": 34465
    },
    {
      "epoch": 1.9300033598387278,
      "grad_norm": 0.9572727680206299,
      "learning_rate": 9.619710526315789e-05,
      "loss": 0.3139,
      "step": 34466
    },
    {
      "epoch": 1.9300593571508569,
      "grad_norm": 1.2366197109222412,
      "learning_rate": 9.619684210526316e-05,
      "loss": 0.3454,
      "step": 34467
    },
    {
      "epoch": 1.9301153544629859,
      "grad_norm": 1.113550066947937,
      "learning_rate": 9.619657894736842e-05,
      "loss": 0.4842,
      "step": 34468
    },
    {
      "epoch": 1.930171351775115,
      "grad_norm": 0.9857762455940247,
      "learning_rate": 9.619631578947368e-05,
      "loss": 0.3599,
      "step": 34469
    },
    {
      "epoch": 1.930227349087244,
      "grad_norm": 1.1586251258850098,
      "learning_rate": 9.619605263157894e-05,
      "loss": 0.3874,
      "step": 34470
    },
    {
      "epoch": 1.930283346399373,
      "grad_norm": 1.244394063949585,
      "learning_rate": 9.619578947368422e-05,
      "loss": 0.4,
      "step": 34471
    },
    {
      "epoch": 1.930339343711502,
      "grad_norm": 1.2778180837631226,
      "learning_rate": 9.619552631578948e-05,
      "loss": 0.4722,
      "step": 34472
    },
    {
      "epoch": 1.930395341023631,
      "grad_norm": 5.057855606079102,
      "learning_rate": 9.619526315789475e-05,
      "loss": 0.447,
      "step": 34473
    },
    {
      "epoch": 1.93045133833576,
      "grad_norm": 1.5484793186187744,
      "learning_rate": 9.6195e-05,
      "loss": 0.5107,
      "step": 34474
    },
    {
      "epoch": 1.930507335647889,
      "grad_norm": 1.6093711853027344,
      "learning_rate": 9.619473684210527e-05,
      "loss": 0.6132,
      "step": 34475
    },
    {
      "epoch": 1.930563332960018,
      "grad_norm": 1.5866259336471558,
      "learning_rate": 9.619447368421053e-05,
      "loss": 0.4621,
      "step": 34476
    },
    {
      "epoch": 1.930619330272147,
      "grad_norm": 1.2830190658569336,
      "learning_rate": 9.61942105263158e-05,
      "loss": 0.4974,
      "step": 34477
    },
    {
      "epoch": 1.930675327584276,
      "grad_norm": 1.5656461715698242,
      "learning_rate": 9.619394736842106e-05,
      "loss": 0.4832,
      "step": 34478
    },
    {
      "epoch": 1.930731324896405,
      "grad_norm": 1.625942349433899,
      "learning_rate": 9.619368421052632e-05,
      "loss": 0.4119,
      "step": 34479
    },
    {
      "epoch": 1.930787322208534,
      "grad_norm": 1.3033757209777832,
      "learning_rate": 9.619342105263158e-05,
      "loss": 0.689,
      "step": 34480
    },
    {
      "epoch": 1.9308433195206631,
      "grad_norm": 1.2183730602264404,
      "learning_rate": 9.619315789473686e-05,
      "loss": 0.4074,
      "step": 34481
    },
    {
      "epoch": 1.9308993168327921,
      "grad_norm": 1.1757538318634033,
      "learning_rate": 9.619289473684211e-05,
      "loss": 0.3777,
      "step": 34482
    },
    {
      "epoch": 1.9309553141449212,
      "grad_norm": 2.666792392730713,
      "learning_rate": 9.619263157894736e-05,
      "loss": 0.467,
      "step": 34483
    },
    {
      "epoch": 1.9310113114570502,
      "grad_norm": 1.1633663177490234,
      "learning_rate": 9.619236842105263e-05,
      "loss": 0.4984,
      "step": 34484
    },
    {
      "epoch": 1.9310673087691792,
      "grad_norm": 1.1002119779586792,
      "learning_rate": 9.61921052631579e-05,
      "loss": 0.4097,
      "step": 34485
    },
    {
      "epoch": 1.9311233060813082,
      "grad_norm": 1.0700511932373047,
      "learning_rate": 9.619184210526317e-05,
      "loss": 0.403,
      "step": 34486
    },
    {
      "epoch": 1.9311793033934372,
      "grad_norm": 1.2530592679977417,
      "learning_rate": 9.619157894736843e-05,
      "loss": 0.5182,
      "step": 34487
    },
    {
      "epoch": 1.9312353007055663,
      "grad_norm": 1.1896612644195557,
      "learning_rate": 9.619131578947369e-05,
      "loss": 0.4402,
      "step": 34488
    },
    {
      "epoch": 1.9312912980176953,
      "grad_norm": 1.3995798826217651,
      "learning_rate": 9.619105263157895e-05,
      "loss": 0.4661,
      "step": 34489
    },
    {
      "epoch": 1.9313472953298243,
      "grad_norm": 1.2056599855422974,
      "learning_rate": 9.619078947368422e-05,
      "loss": 0.3773,
      "step": 34490
    },
    {
      "epoch": 1.9314032926419533,
      "grad_norm": 1.3707834482192993,
      "learning_rate": 9.619052631578948e-05,
      "loss": 0.4755,
      "step": 34491
    },
    {
      "epoch": 1.9314592899540823,
      "grad_norm": 1.335739254951477,
      "learning_rate": 9.619026315789474e-05,
      "loss": 0.4215,
      "step": 34492
    },
    {
      "epoch": 1.9315152872662114,
      "grad_norm": 1.5348047018051147,
      "learning_rate": 9.619e-05,
      "loss": 0.4623,
      "step": 34493
    },
    {
      "epoch": 1.9315712845783404,
      "grad_norm": 1.5599212646484375,
      "learning_rate": 9.618973684210527e-05,
      "loss": 0.4511,
      "step": 34494
    },
    {
      "epoch": 1.9316272818904694,
      "grad_norm": 1.3879213333129883,
      "learning_rate": 9.618947368421053e-05,
      "loss": 0.3944,
      "step": 34495
    },
    {
      "epoch": 1.9316832792025984,
      "grad_norm": 1.3457173109054565,
      "learning_rate": 9.618921052631579e-05,
      "loss": 0.4529,
      "step": 34496
    },
    {
      "epoch": 1.9317392765147274,
      "grad_norm": 1.2318543195724487,
      "learning_rate": 9.618894736842105e-05,
      "loss": 0.5488,
      "step": 34497
    },
    {
      "epoch": 1.9317952738268565,
      "grad_norm": 1.1020747423171997,
      "learning_rate": 9.618868421052632e-05,
      "loss": 0.3901,
      "step": 34498
    },
    {
      "epoch": 1.9318512711389855,
      "grad_norm": 5.155155658721924,
      "learning_rate": 9.618842105263158e-05,
      "loss": 0.5578,
      "step": 34499
    },
    {
      "epoch": 1.9319072684511145,
      "grad_norm": 1.4636818170547485,
      "learning_rate": 9.618815789473686e-05,
      "loss": 0.3723,
      "step": 34500
    },
    {
      "epoch": 1.9319632657632435,
      "grad_norm": 1.1043155193328857,
      "learning_rate": 9.61878947368421e-05,
      "loss": 0.5302,
      "step": 34501
    },
    {
      "epoch": 1.9320192630753725,
      "grad_norm": 1.0371291637420654,
      "learning_rate": 9.618763157894736e-05,
      "loss": 0.3842,
      "step": 34502
    },
    {
      "epoch": 1.9320752603875015,
      "grad_norm": 1.4405776262283325,
      "learning_rate": 9.618736842105264e-05,
      "loss": 0.4688,
      "step": 34503
    },
    {
      "epoch": 1.9321312576996306,
      "grad_norm": 1.1060543060302734,
      "learning_rate": 9.61871052631579e-05,
      "loss": 0.3822,
      "step": 34504
    },
    {
      "epoch": 1.9321872550117596,
      "grad_norm": 1.216041922569275,
      "learning_rate": 9.618684210526317e-05,
      "loss": 0.415,
      "step": 34505
    },
    {
      "epoch": 1.9322432523238886,
      "grad_norm": 1.065531611442566,
      "learning_rate": 9.618657894736842e-05,
      "loss": 0.4025,
      "step": 34506
    },
    {
      "epoch": 1.9322992496360176,
      "grad_norm": 1.2006736993789673,
      "learning_rate": 9.618631578947369e-05,
      "loss": 0.4577,
      "step": 34507
    },
    {
      "epoch": 1.9323552469481466,
      "grad_norm": 1.2080492973327637,
      "learning_rate": 9.618605263157895e-05,
      "loss": 0.4196,
      "step": 34508
    },
    {
      "epoch": 1.9324112442602757,
      "grad_norm": 1.1670516729354858,
      "learning_rate": 9.618578947368422e-05,
      "loss": 0.4129,
      "step": 34509
    },
    {
      "epoch": 1.9324672415724047,
      "grad_norm": 1.0263789892196655,
      "learning_rate": 9.618552631578948e-05,
      "loss": 0.3277,
      "step": 34510
    },
    {
      "epoch": 1.9325232388845337,
      "grad_norm": 1.4198226928710938,
      "learning_rate": 9.618526315789474e-05,
      "loss": 0.3526,
      "step": 34511
    },
    {
      "epoch": 1.9325792361966627,
      "grad_norm": 1.7080702781677246,
      "learning_rate": 9.6185e-05,
      "loss": 0.457,
      "step": 34512
    },
    {
      "epoch": 1.9326352335087917,
      "grad_norm": 1.5729179382324219,
      "learning_rate": 9.618473684210527e-05,
      "loss": 0.4614,
      "step": 34513
    },
    {
      "epoch": 1.9326912308209208,
      "grad_norm": 1.2717292308807373,
      "learning_rate": 9.618447368421053e-05,
      "loss": 0.3934,
      "step": 34514
    },
    {
      "epoch": 1.9327472281330498,
      "grad_norm": 2.0297207832336426,
      "learning_rate": 9.61842105263158e-05,
      "loss": 0.4119,
      "step": 34515
    },
    {
      "epoch": 1.9328032254451788,
      "grad_norm": 1.2161298990249634,
      "learning_rate": 9.618394736842105e-05,
      "loss": 0.4797,
      "step": 34516
    },
    {
      "epoch": 1.9328592227573078,
      "grad_norm": 1.2220958471298218,
      "learning_rate": 9.618368421052633e-05,
      "loss": 0.4447,
      "step": 34517
    },
    {
      "epoch": 1.9329152200694366,
      "grad_norm": 1.3285832405090332,
      "learning_rate": 9.618342105263159e-05,
      "loss": 0.4445,
      "step": 34518
    },
    {
      "epoch": 1.9329712173815656,
      "grad_norm": 1.2737245559692383,
      "learning_rate": 9.618315789473685e-05,
      "loss": 0.4862,
      "step": 34519
    },
    {
      "epoch": 1.9330272146936947,
      "grad_norm": 1.5205929279327393,
      "learning_rate": 9.61828947368421e-05,
      "loss": 0.6421,
      "step": 34520
    },
    {
      "epoch": 1.9330832120058237,
      "grad_norm": 1.1003897190093994,
      "learning_rate": 9.618263157894737e-05,
      "loss": 0.4398,
      "step": 34521
    },
    {
      "epoch": 1.9331392093179527,
      "grad_norm": 1.3158518075942993,
      "learning_rate": 9.618236842105264e-05,
      "loss": 0.3468,
      "step": 34522
    },
    {
      "epoch": 1.9331952066300817,
      "grad_norm": 1.0968856811523438,
      "learning_rate": 9.61821052631579e-05,
      "loss": 0.5677,
      "step": 34523
    },
    {
      "epoch": 1.9332512039422107,
      "grad_norm": 1.324910044670105,
      "learning_rate": 9.618184210526316e-05,
      "loss": 0.4995,
      "step": 34524
    },
    {
      "epoch": 1.9333072012543397,
      "grad_norm": 1.264108419418335,
      "learning_rate": 9.618157894736842e-05,
      "loss": 0.4034,
      "step": 34525
    },
    {
      "epoch": 1.9333631985664688,
      "grad_norm": 1.2521198987960815,
      "learning_rate": 9.618131578947369e-05,
      "loss": 0.4265,
      "step": 34526
    },
    {
      "epoch": 1.9334191958785978,
      "grad_norm": 1.1728318929672241,
      "learning_rate": 9.618105263157895e-05,
      "loss": 0.4831,
      "step": 34527
    },
    {
      "epoch": 1.9334751931907268,
      "grad_norm": 1.2251523733139038,
      "learning_rate": 9.618078947368422e-05,
      "loss": 0.3995,
      "step": 34528
    },
    {
      "epoch": 1.9335311905028558,
      "grad_norm": 1.1888328790664673,
      "learning_rate": 9.618052631578947e-05,
      "loss": 0.6765,
      "step": 34529
    },
    {
      "epoch": 1.9335871878149848,
      "grad_norm": 1.165860891342163,
      "learning_rate": 9.618026315789474e-05,
      "loss": 0.3834,
      "step": 34530
    },
    {
      "epoch": 1.9336431851271139,
      "grad_norm": 1.2785743474960327,
      "learning_rate": 9.618e-05,
      "loss": 0.4218,
      "step": 34531
    },
    {
      "epoch": 1.9336991824392429,
      "grad_norm": 1.2545628547668457,
      "learning_rate": 9.617973684210528e-05,
      "loss": 0.5435,
      "step": 34532
    },
    {
      "epoch": 1.933755179751372,
      "grad_norm": 1.1461656093597412,
      "learning_rate": 9.617947368421054e-05,
      "loss": 0.3866,
      "step": 34533
    },
    {
      "epoch": 1.933811177063501,
      "grad_norm": 1.384101390838623,
      "learning_rate": 9.61792105263158e-05,
      "loss": 0.4265,
      "step": 34534
    },
    {
      "epoch": 1.93386717437563,
      "grad_norm": 1.1068389415740967,
      "learning_rate": 9.617894736842106e-05,
      "loss": 0.4518,
      "step": 34535
    },
    {
      "epoch": 1.933923171687759,
      "grad_norm": 1.266049861907959,
      "learning_rate": 9.617868421052632e-05,
      "loss": 0.5008,
      "step": 34536
    },
    {
      "epoch": 1.933979168999888,
      "grad_norm": 1.5910966396331787,
      "learning_rate": 9.617842105263159e-05,
      "loss": 0.5164,
      "step": 34537
    },
    {
      "epoch": 1.934035166312017,
      "grad_norm": 1.6353645324707031,
      "learning_rate": 9.617815789473684e-05,
      "loss": 0.4413,
      "step": 34538
    },
    {
      "epoch": 1.934091163624146,
      "grad_norm": 1.3025604486465454,
      "learning_rate": 9.617789473684211e-05,
      "loss": 0.4211,
      "step": 34539
    },
    {
      "epoch": 1.934147160936275,
      "grad_norm": 1.4288135766983032,
      "learning_rate": 9.617763157894737e-05,
      "loss": 0.6824,
      "step": 34540
    },
    {
      "epoch": 1.934203158248404,
      "grad_norm": 1.650879144668579,
      "learning_rate": 9.617736842105264e-05,
      "loss": 0.4972,
      "step": 34541
    },
    {
      "epoch": 1.934259155560533,
      "grad_norm": 1.1612884998321533,
      "learning_rate": 9.61771052631579e-05,
      "loss": 0.4288,
      "step": 34542
    },
    {
      "epoch": 1.934315152872662,
      "grad_norm": 1.3062304258346558,
      "learning_rate": 9.617684210526316e-05,
      "loss": 0.4666,
      "step": 34543
    },
    {
      "epoch": 1.934371150184791,
      "grad_norm": 1.217829704284668,
      "learning_rate": 9.617657894736842e-05,
      "loss": 0.4608,
      "step": 34544
    },
    {
      "epoch": 1.9344271474969201,
      "grad_norm": 1.2446705102920532,
      "learning_rate": 9.61763157894737e-05,
      "loss": 0.4185,
      "step": 34545
    },
    {
      "epoch": 1.9344831448090491,
      "grad_norm": 1.3282055854797363,
      "learning_rate": 9.617605263157895e-05,
      "loss": 0.4442,
      "step": 34546
    },
    {
      "epoch": 1.9345391421211782,
      "grad_norm": 1.196389079093933,
      "learning_rate": 9.617578947368421e-05,
      "loss": 0.3877,
      "step": 34547
    },
    {
      "epoch": 1.9345951394333072,
      "grad_norm": 1.754717469215393,
      "learning_rate": 9.617552631578947e-05,
      "loss": 0.4662,
      "step": 34548
    },
    {
      "epoch": 1.9346511367454362,
      "grad_norm": 1.138705849647522,
      "learning_rate": 9.617526315789475e-05,
      "loss": 0.4915,
      "step": 34549
    },
    {
      "epoch": 1.9347071340575652,
      "grad_norm": 1.3363920450210571,
      "learning_rate": 9.6175e-05,
      "loss": 0.4322,
      "step": 34550
    },
    {
      "epoch": 1.9347631313696942,
      "grad_norm": 1.40372896194458,
      "learning_rate": 9.617473684210527e-05,
      "loss": 0.3927,
      "step": 34551
    },
    {
      "epoch": 1.9348191286818233,
      "grad_norm": 1.1117277145385742,
      "learning_rate": 9.617447368421053e-05,
      "loss": 0.4147,
      "step": 34552
    },
    {
      "epoch": 1.9348751259939523,
      "grad_norm": 1.5525130033493042,
      "learning_rate": 9.617421052631579e-05,
      "loss": 0.4016,
      "step": 34553
    },
    {
      "epoch": 1.9349311233060813,
      "grad_norm": 1.1578190326690674,
      "learning_rate": 9.617394736842106e-05,
      "loss": 0.4673,
      "step": 34554
    },
    {
      "epoch": 1.9349871206182103,
      "grad_norm": 1.208340048789978,
      "learning_rate": 9.617368421052632e-05,
      "loss": 0.4391,
      "step": 34555
    },
    {
      "epoch": 1.9350431179303393,
      "grad_norm": 1.1800040006637573,
      "learning_rate": 9.617342105263158e-05,
      "loss": 0.4267,
      "step": 34556
    },
    {
      "epoch": 1.9350991152424684,
      "grad_norm": 1.0394799709320068,
      "learning_rate": 9.617315789473684e-05,
      "loss": 0.4812,
      "step": 34557
    },
    {
      "epoch": 1.9351551125545974,
      "grad_norm": 1.281531572341919,
      "learning_rate": 9.617289473684211e-05,
      "loss": 0.4384,
      "step": 34558
    },
    {
      "epoch": 1.9352111098667264,
      "grad_norm": 1.0990893840789795,
      "learning_rate": 9.617263157894737e-05,
      "loss": 0.4131,
      "step": 34559
    },
    {
      "epoch": 1.9352671071788554,
      "grad_norm": 1.271069884300232,
      "learning_rate": 9.617236842105264e-05,
      "loss": 0.4101,
      "step": 34560
    },
    {
      "epoch": 1.9353231044909844,
      "grad_norm": 1.4581999778747559,
      "learning_rate": 9.617210526315789e-05,
      "loss": 0.5431,
      "step": 34561
    },
    {
      "epoch": 1.9353791018031135,
      "grad_norm": 1.7357470989227295,
      "learning_rate": 9.617184210526316e-05,
      "loss": 0.6159,
      "step": 34562
    },
    {
      "epoch": 1.9354350991152425,
      "grad_norm": 1.191070318222046,
      "learning_rate": 9.617157894736842e-05,
      "loss": 0.3845,
      "step": 34563
    },
    {
      "epoch": 1.9354910964273715,
      "grad_norm": 1.4071005582809448,
      "learning_rate": 9.61713157894737e-05,
      "loss": 0.6476,
      "step": 34564
    },
    {
      "epoch": 1.9355470937395005,
      "grad_norm": 1.4647846221923828,
      "learning_rate": 9.617105263157896e-05,
      "loss": 0.4518,
      "step": 34565
    },
    {
      "epoch": 1.9356030910516295,
      "grad_norm": 1.1088801622390747,
      "learning_rate": 9.617078947368422e-05,
      "loss": 0.4062,
      "step": 34566
    },
    {
      "epoch": 1.9356590883637586,
      "grad_norm": 1.1932700872421265,
      "learning_rate": 9.617052631578948e-05,
      "loss": 0.4368,
      "step": 34567
    },
    {
      "epoch": 1.9357150856758876,
      "grad_norm": 1.3187135457992554,
      "learning_rate": 9.617026315789475e-05,
      "loss": 0.5061,
      "step": 34568
    },
    {
      "epoch": 1.9357710829880166,
      "grad_norm": 1.1500115394592285,
      "learning_rate": 9.617000000000001e-05,
      "loss": 0.4545,
      "step": 34569
    },
    {
      "epoch": 1.9358270803001456,
      "grad_norm": 1.350380778312683,
      "learning_rate": 9.616973684210527e-05,
      "loss": 0.4808,
      "step": 34570
    },
    {
      "epoch": 1.9358830776122746,
      "grad_norm": 1.2474403381347656,
      "learning_rate": 9.616947368421053e-05,
      "loss": 0.4078,
      "step": 34571
    },
    {
      "epoch": 1.9359390749244036,
      "grad_norm": 1.3160289525985718,
      "learning_rate": 9.616921052631579e-05,
      "loss": 0.552,
      "step": 34572
    },
    {
      "epoch": 1.9359950722365327,
      "grad_norm": 1.6678022146224976,
      "learning_rate": 9.616894736842106e-05,
      "loss": 0.4072,
      "step": 34573
    },
    {
      "epoch": 1.9360510695486617,
      "grad_norm": 1.8979817628860474,
      "learning_rate": 9.616868421052632e-05,
      "loss": 0.4502,
      "step": 34574
    },
    {
      "epoch": 1.9361070668607907,
      "grad_norm": 1.8159265518188477,
      "learning_rate": 9.616842105263158e-05,
      "loss": 0.4216,
      "step": 34575
    },
    {
      "epoch": 1.9361630641729197,
      "grad_norm": 1.3641353845596313,
      "learning_rate": 9.616815789473684e-05,
      "loss": 0.2937,
      "step": 34576
    },
    {
      "epoch": 1.9362190614850487,
      "grad_norm": 1.1954460144042969,
      "learning_rate": 9.616789473684211e-05,
      "loss": 0.4449,
      "step": 34577
    },
    {
      "epoch": 1.9362750587971778,
      "grad_norm": 1.7169078588485718,
      "learning_rate": 9.616763157894737e-05,
      "loss": 0.6248,
      "step": 34578
    },
    {
      "epoch": 1.9363310561093068,
      "grad_norm": 1.1840386390686035,
      "learning_rate": 9.616736842105263e-05,
      "loss": 0.5037,
      "step": 34579
    },
    {
      "epoch": 1.9363870534214358,
      "grad_norm": 1.3257672786712646,
      "learning_rate": 9.61671052631579e-05,
      "loss": 0.4713,
      "step": 34580
    },
    {
      "epoch": 1.9364430507335648,
      "grad_norm": 1.4435157775878906,
      "learning_rate": 9.616684210526317e-05,
      "loss": 0.4189,
      "step": 34581
    },
    {
      "epoch": 1.9364990480456938,
      "grad_norm": 1.2479151487350464,
      "learning_rate": 9.616657894736843e-05,
      "loss": 0.4395,
      "step": 34582
    },
    {
      "epoch": 1.9365550453578229,
      "grad_norm": 1.1391650438308716,
      "learning_rate": 9.61663157894737e-05,
      "loss": 0.359,
      "step": 34583
    },
    {
      "epoch": 1.9366110426699519,
      "grad_norm": 1.1199320554733276,
      "learning_rate": 9.616605263157895e-05,
      "loss": 0.4245,
      "step": 34584
    },
    {
      "epoch": 1.936667039982081,
      "grad_norm": 1.3649721145629883,
      "learning_rate": 9.616578947368422e-05,
      "loss": 0.4402,
      "step": 34585
    },
    {
      "epoch": 1.93672303729421,
      "grad_norm": 1.2751247882843018,
      "learning_rate": 9.616552631578948e-05,
      "loss": 0.4888,
      "step": 34586
    },
    {
      "epoch": 1.936779034606339,
      "grad_norm": 1.2136073112487793,
      "learning_rate": 9.616526315789474e-05,
      "loss": 0.4776,
      "step": 34587
    },
    {
      "epoch": 1.936835031918468,
      "grad_norm": 1.2153640985488892,
      "learning_rate": 9.616500000000001e-05,
      "loss": 0.5074,
      "step": 34588
    },
    {
      "epoch": 1.936891029230597,
      "grad_norm": 1.2673554420471191,
      "learning_rate": 9.616473684210526e-05,
      "loss": 0.4076,
      "step": 34589
    },
    {
      "epoch": 1.936947026542726,
      "grad_norm": 1.2590012550354004,
      "learning_rate": 9.616447368421053e-05,
      "loss": 0.4068,
      "step": 34590
    },
    {
      "epoch": 1.937003023854855,
      "grad_norm": 1.1679997444152832,
      "learning_rate": 9.616421052631579e-05,
      "loss": 0.5273,
      "step": 34591
    },
    {
      "epoch": 1.937059021166984,
      "grad_norm": 1.422092318534851,
      "learning_rate": 9.616394736842106e-05,
      "loss": 0.413,
      "step": 34592
    },
    {
      "epoch": 1.937115018479113,
      "grad_norm": 1.30519700050354,
      "learning_rate": 9.616368421052631e-05,
      "loss": 0.458,
      "step": 34593
    },
    {
      "epoch": 1.937171015791242,
      "grad_norm": 1.289374828338623,
      "learning_rate": 9.616342105263158e-05,
      "loss": 0.3867,
      "step": 34594
    },
    {
      "epoch": 1.937227013103371,
      "grad_norm": 1.5045796632766724,
      "learning_rate": 9.616315789473684e-05,
      "loss": 0.5411,
      "step": 34595
    },
    {
      "epoch": 1.9372830104155,
      "grad_norm": 1.4894495010375977,
      "learning_rate": 9.616289473684212e-05,
      "loss": 0.4823,
      "step": 34596
    },
    {
      "epoch": 1.9373390077276291,
      "grad_norm": 1.1748589277267456,
      "learning_rate": 9.616263157894738e-05,
      "loss": 0.4262,
      "step": 34597
    },
    {
      "epoch": 1.9373950050397581,
      "grad_norm": 1.3518860340118408,
      "learning_rate": 9.616236842105264e-05,
      "loss": 0.4314,
      "step": 34598
    },
    {
      "epoch": 1.9374510023518872,
      "grad_norm": 1.6337566375732422,
      "learning_rate": 9.61621052631579e-05,
      "loss": 0.4184,
      "step": 34599
    },
    {
      "epoch": 1.937506999664016,
      "grad_norm": 1.634097933769226,
      "learning_rate": 9.616184210526317e-05,
      "loss": 0.3884,
      "step": 34600
    },
    {
      "epoch": 1.937562996976145,
      "grad_norm": 1.6530539989471436,
      "learning_rate": 9.616157894736843e-05,
      "loss": 0.4732,
      "step": 34601
    },
    {
      "epoch": 1.937618994288274,
      "grad_norm": 1.3201847076416016,
      "learning_rate": 9.616131578947369e-05,
      "loss": 0.4004,
      "step": 34602
    },
    {
      "epoch": 1.937674991600403,
      "grad_norm": 1.0525574684143066,
      "learning_rate": 9.616105263157895e-05,
      "loss": 0.3802,
      "step": 34603
    },
    {
      "epoch": 1.937730988912532,
      "grad_norm": 1.343254804611206,
      "learning_rate": 9.616078947368421e-05,
      "loss": 0.4366,
      "step": 34604
    },
    {
      "epoch": 1.937786986224661,
      "grad_norm": 1.4546691179275513,
      "learning_rate": 9.616052631578948e-05,
      "loss": 0.5366,
      "step": 34605
    },
    {
      "epoch": 1.93784298353679,
      "grad_norm": 1.1656324863433838,
      "learning_rate": 9.616026315789474e-05,
      "loss": 0.488,
      "step": 34606
    },
    {
      "epoch": 1.937898980848919,
      "grad_norm": 1.2189165353775024,
      "learning_rate": 9.616e-05,
      "loss": 0.4051,
      "step": 34607
    },
    {
      "epoch": 1.9379549781610481,
      "grad_norm": 1.1972731351852417,
      "learning_rate": 9.615973684210526e-05,
      "loss": 0.5443,
      "step": 34608
    },
    {
      "epoch": 1.9380109754731771,
      "grad_norm": 1.43869149684906,
      "learning_rate": 9.615947368421053e-05,
      "loss": 0.4481,
      "step": 34609
    },
    {
      "epoch": 1.9380669727853062,
      "grad_norm": 1.5553572177886963,
      "learning_rate": 9.61592105263158e-05,
      "loss": 0.5815,
      "step": 34610
    },
    {
      "epoch": 1.9381229700974352,
      "grad_norm": 1.141389012336731,
      "learning_rate": 9.615894736842105e-05,
      "loss": 0.4766,
      "step": 34611
    },
    {
      "epoch": 1.9381789674095642,
      "grad_norm": 1.2720290422439575,
      "learning_rate": 9.615868421052631e-05,
      "loss": 0.48,
      "step": 34612
    },
    {
      "epoch": 1.9382349647216932,
      "grad_norm": 1.0373291969299316,
      "learning_rate": 9.615842105263159e-05,
      "loss": 0.4641,
      "step": 34613
    },
    {
      "epoch": 1.9382909620338222,
      "grad_norm": 1.37346351146698,
      "learning_rate": 9.615815789473685e-05,
      "loss": 0.498,
      "step": 34614
    },
    {
      "epoch": 1.9383469593459512,
      "grad_norm": 1.2834287881851196,
      "learning_rate": 9.615789473684212e-05,
      "loss": 0.4999,
      "step": 34615
    },
    {
      "epoch": 1.9384029566580803,
      "grad_norm": 1.1452680826187134,
      "learning_rate": 9.615763157894737e-05,
      "loss": 0.4217,
      "step": 34616
    },
    {
      "epoch": 1.9384589539702093,
      "grad_norm": 1.2596207857131958,
      "learning_rate": 9.615736842105264e-05,
      "loss": 0.5259,
      "step": 34617
    },
    {
      "epoch": 1.9385149512823383,
      "grad_norm": 1.3268545866012573,
      "learning_rate": 9.61571052631579e-05,
      "loss": 0.3686,
      "step": 34618
    },
    {
      "epoch": 1.9385709485944673,
      "grad_norm": 1.4154986143112183,
      "learning_rate": 9.615684210526317e-05,
      "loss": 0.4974,
      "step": 34619
    },
    {
      "epoch": 1.9386269459065963,
      "grad_norm": 0.9902254939079285,
      "learning_rate": 9.615657894736843e-05,
      "loss": 0.3613,
      "step": 34620
    },
    {
      "epoch": 1.9386829432187254,
      "grad_norm": 1.471644401550293,
      "learning_rate": 9.615631578947369e-05,
      "loss": 0.4858,
      "step": 34621
    },
    {
      "epoch": 1.9387389405308544,
      "grad_norm": 1.3542932271957397,
      "learning_rate": 9.615605263157895e-05,
      "loss": 0.4886,
      "step": 34622
    },
    {
      "epoch": 1.9387949378429834,
      "grad_norm": 1.045552134513855,
      "learning_rate": 9.615578947368421e-05,
      "loss": 0.3742,
      "step": 34623
    },
    {
      "epoch": 1.9388509351551124,
      "grad_norm": 1.6668481826782227,
      "learning_rate": 9.615552631578948e-05,
      "loss": 0.5614,
      "step": 34624
    },
    {
      "epoch": 1.9389069324672414,
      "grad_norm": 1.5938810110092163,
      "learning_rate": 9.615526315789474e-05,
      "loss": 0.3317,
      "step": 34625
    },
    {
      "epoch": 1.9389629297793705,
      "grad_norm": 1.4155513048171997,
      "learning_rate": 9.6155e-05,
      "loss": 0.4982,
      "step": 34626
    },
    {
      "epoch": 1.9390189270914995,
      "grad_norm": 1.1545783281326294,
      "learning_rate": 9.615473684210526e-05,
      "loss": 0.4106,
      "step": 34627
    },
    {
      "epoch": 1.9390749244036285,
      "grad_norm": 1.1907387971878052,
      "learning_rate": 9.615447368421054e-05,
      "loss": 0.3877,
      "step": 34628
    },
    {
      "epoch": 1.9391309217157575,
      "grad_norm": 1.9938383102416992,
      "learning_rate": 9.61542105263158e-05,
      "loss": 0.4124,
      "step": 34629
    },
    {
      "epoch": 1.9391869190278865,
      "grad_norm": 1.1957322359085083,
      "learning_rate": 9.615394736842106e-05,
      "loss": 0.5507,
      "step": 34630
    },
    {
      "epoch": 1.9392429163400156,
      "grad_norm": 1.529099941253662,
      "learning_rate": 9.615368421052632e-05,
      "loss": 0.3914,
      "step": 34631
    },
    {
      "epoch": 1.9392989136521446,
      "grad_norm": 1.3665045499801636,
      "learning_rate": 9.615342105263159e-05,
      "loss": 0.5274,
      "step": 34632
    },
    {
      "epoch": 1.9393549109642736,
      "grad_norm": 1.3016235828399658,
      "learning_rate": 9.615315789473685e-05,
      "loss": 0.5063,
      "step": 34633
    },
    {
      "epoch": 1.9394109082764026,
      "grad_norm": 1.4059265851974487,
      "learning_rate": 9.615289473684211e-05,
      "loss": 0.4747,
      "step": 34634
    },
    {
      "epoch": 1.9394669055885316,
      "grad_norm": 1.2913436889648438,
      "learning_rate": 9.615263157894737e-05,
      "loss": 0.4082,
      "step": 34635
    },
    {
      "epoch": 1.9395229029006607,
      "grad_norm": 1.2261064052581787,
      "learning_rate": 9.615236842105264e-05,
      "loss": 0.4238,
      "step": 34636
    },
    {
      "epoch": 1.9395789002127897,
      "grad_norm": 1.096225619316101,
      "learning_rate": 9.61521052631579e-05,
      "loss": 0.4179,
      "step": 34637
    },
    {
      "epoch": 1.9396348975249187,
      "grad_norm": 1.0602062940597534,
      "learning_rate": 9.615184210526317e-05,
      "loss": 0.4675,
      "step": 34638
    },
    {
      "epoch": 1.9396908948370477,
      "grad_norm": 1.3400038480758667,
      "learning_rate": 9.615157894736842e-05,
      "loss": 0.3999,
      "step": 34639
    },
    {
      "epoch": 1.9397468921491767,
      "grad_norm": 1.1630491018295288,
      "learning_rate": 9.615131578947368e-05,
      "loss": 0.3586,
      "step": 34640
    },
    {
      "epoch": 1.9398028894613057,
      "grad_norm": 1.1611605882644653,
      "learning_rate": 9.615105263157895e-05,
      "loss": 0.54,
      "step": 34641
    },
    {
      "epoch": 1.9398588867734348,
      "grad_norm": 1.250673770904541,
      "learning_rate": 9.615078947368421e-05,
      "loss": 0.4018,
      "step": 34642
    },
    {
      "epoch": 1.9399148840855638,
      "grad_norm": 1.2914491891860962,
      "learning_rate": 9.615052631578947e-05,
      "loss": 0.566,
      "step": 34643
    },
    {
      "epoch": 1.9399708813976928,
      "grad_norm": 1.1089885234832764,
      "learning_rate": 9.615026315789473e-05,
      "loss": 0.5059,
      "step": 34644
    },
    {
      "epoch": 1.9400268787098218,
      "grad_norm": 1.4148918390274048,
      "learning_rate": 9.615e-05,
      "loss": 0.3945,
      "step": 34645
    },
    {
      "epoch": 1.9400828760219508,
      "grad_norm": 1.3594365119934082,
      "learning_rate": 9.614973684210527e-05,
      "loss": 0.4928,
      "step": 34646
    },
    {
      "epoch": 1.9401388733340799,
      "grad_norm": 1.026719331741333,
      "learning_rate": 9.614947368421054e-05,
      "loss": 0.2757,
      "step": 34647
    },
    {
      "epoch": 1.9401948706462089,
      "grad_norm": 1.564383625984192,
      "learning_rate": 9.614921052631579e-05,
      "loss": 0.6749,
      "step": 34648
    },
    {
      "epoch": 1.940250867958338,
      "grad_norm": 1.378625750541687,
      "learning_rate": 9.614894736842106e-05,
      "loss": 0.333,
      "step": 34649
    },
    {
      "epoch": 1.940306865270467,
      "grad_norm": 1.2058113813400269,
      "learning_rate": 9.614868421052632e-05,
      "loss": 0.4204,
      "step": 34650
    },
    {
      "epoch": 1.940362862582596,
      "grad_norm": 1.3051544427871704,
      "learning_rate": 9.614842105263159e-05,
      "loss": 0.3762,
      "step": 34651
    },
    {
      "epoch": 1.940418859894725,
      "grad_norm": 1.471980094909668,
      "learning_rate": 9.614815789473685e-05,
      "loss": 0.3494,
      "step": 34652
    },
    {
      "epoch": 1.940474857206854,
      "grad_norm": 1.1293929815292358,
      "learning_rate": 9.614789473684211e-05,
      "loss": 0.4105,
      "step": 34653
    },
    {
      "epoch": 1.940530854518983,
      "grad_norm": 1.0405559539794922,
      "learning_rate": 9.614763157894737e-05,
      "loss": 0.4182,
      "step": 34654
    },
    {
      "epoch": 1.940586851831112,
      "grad_norm": 1.2232109308242798,
      "learning_rate": 9.614736842105264e-05,
      "loss": 0.3722,
      "step": 34655
    },
    {
      "epoch": 1.940642849143241,
      "grad_norm": 1.5417680740356445,
      "learning_rate": 9.61471052631579e-05,
      "loss": 0.4721,
      "step": 34656
    },
    {
      "epoch": 1.94069884645537,
      "grad_norm": 1.6847895383834839,
      "learning_rate": 9.614684210526316e-05,
      "loss": 0.4847,
      "step": 34657
    },
    {
      "epoch": 1.940754843767499,
      "grad_norm": 1.2873189449310303,
      "learning_rate": 9.614657894736842e-05,
      "loss": 0.4079,
      "step": 34658
    },
    {
      "epoch": 1.940810841079628,
      "grad_norm": 1.2061678171157837,
      "learning_rate": 9.614631578947368e-05,
      "loss": 0.5552,
      "step": 34659
    },
    {
      "epoch": 1.940866838391757,
      "grad_norm": 31.84278106689453,
      "learning_rate": 9.614605263157896e-05,
      "loss": 0.354,
      "step": 34660
    },
    {
      "epoch": 1.9409228357038861,
      "grad_norm": 1.3271760940551758,
      "learning_rate": 9.614578947368422e-05,
      "loss": 0.4848,
      "step": 34661
    },
    {
      "epoch": 1.9409788330160151,
      "grad_norm": 1.1252739429473877,
      "learning_rate": 9.614552631578948e-05,
      "loss": 0.4151,
      "step": 34662
    },
    {
      "epoch": 1.9410348303281442,
      "grad_norm": 1.287384271621704,
      "learning_rate": 9.614526315789474e-05,
      "loss": 0.5251,
      "step": 34663
    },
    {
      "epoch": 1.9410908276402732,
      "grad_norm": 1.4087750911712646,
      "learning_rate": 9.614500000000001e-05,
      "loss": 0.5181,
      "step": 34664
    },
    {
      "epoch": 1.9411468249524022,
      "grad_norm": 1.1600946187973022,
      "learning_rate": 9.614473684210527e-05,
      "loss": 0.3616,
      "step": 34665
    },
    {
      "epoch": 1.9412028222645312,
      "grad_norm": 1.1475861072540283,
      "learning_rate": 9.614447368421053e-05,
      "loss": 0.5085,
      "step": 34666
    },
    {
      "epoch": 1.9412588195766602,
      "grad_norm": 1.1850228309631348,
      "learning_rate": 9.614421052631579e-05,
      "loss": 0.4361,
      "step": 34667
    },
    {
      "epoch": 1.9413148168887893,
      "grad_norm": 1.1274324655532837,
      "learning_rate": 9.614394736842106e-05,
      "loss": 0.3552,
      "step": 34668
    },
    {
      "epoch": 1.9413708142009183,
      "grad_norm": 1.6500643491744995,
      "learning_rate": 9.614368421052632e-05,
      "loss": 0.4626,
      "step": 34669
    },
    {
      "epoch": 1.9414268115130473,
      "grad_norm": 1.5146517753601074,
      "learning_rate": 9.61434210526316e-05,
      "loss": 0.6171,
      "step": 34670
    },
    {
      "epoch": 1.9414828088251763,
      "grad_norm": 1.4760044813156128,
      "learning_rate": 9.614315789473684e-05,
      "loss": 0.5422,
      "step": 34671
    },
    {
      "epoch": 1.9415388061373053,
      "grad_norm": 1.1378986835479736,
      "learning_rate": 9.614289473684211e-05,
      "loss": 0.4081,
      "step": 34672
    },
    {
      "epoch": 1.9415948034494344,
      "grad_norm": 1.2853001356124878,
      "learning_rate": 9.614263157894737e-05,
      "loss": 0.5046,
      "step": 34673
    },
    {
      "epoch": 1.9416508007615634,
      "grad_norm": 1.8204479217529297,
      "learning_rate": 9.614236842105263e-05,
      "loss": 0.5259,
      "step": 34674
    },
    {
      "epoch": 1.9417067980736924,
      "grad_norm": 1.2837580442428589,
      "learning_rate": 9.61421052631579e-05,
      "loss": 0.4325,
      "step": 34675
    },
    {
      "epoch": 1.9417627953858214,
      "grad_norm": 1.2756407260894775,
      "learning_rate": 9.614184210526315e-05,
      "loss": 0.4198,
      "step": 34676
    },
    {
      "epoch": 1.9418187926979504,
      "grad_norm": 1.205938696861267,
      "learning_rate": 9.614157894736843e-05,
      "loss": 0.4451,
      "step": 34677
    },
    {
      "epoch": 1.9418747900100795,
      "grad_norm": 1.0049076080322266,
      "learning_rate": 9.614131578947369e-05,
      "loss": 0.401,
      "step": 34678
    },
    {
      "epoch": 1.9419307873222085,
      "grad_norm": 1.1169884204864502,
      "learning_rate": 9.614105263157896e-05,
      "loss": 0.3919,
      "step": 34679
    },
    {
      "epoch": 1.9419867846343375,
      "grad_norm": 1.808873176574707,
      "learning_rate": 9.614078947368422e-05,
      "loss": 0.3812,
      "step": 34680
    },
    {
      "epoch": 1.9420427819464665,
      "grad_norm": 1.1550613641738892,
      "learning_rate": 9.614052631578948e-05,
      "loss": 0.3846,
      "step": 34681
    },
    {
      "epoch": 1.9420987792585955,
      "grad_norm": 1.4847160577774048,
      "learning_rate": 9.614026315789474e-05,
      "loss": 0.4704,
      "step": 34682
    },
    {
      "epoch": 1.9421547765707246,
      "grad_norm": 1.2507047653198242,
      "learning_rate": 9.614000000000001e-05,
      "loss": 0.4187,
      "step": 34683
    },
    {
      "epoch": 1.9422107738828536,
      "grad_norm": 1.1527881622314453,
      "learning_rate": 9.613973684210527e-05,
      "loss": 0.4345,
      "step": 34684
    },
    {
      "epoch": 1.9422667711949826,
      "grad_norm": 1.1177562475204468,
      "learning_rate": 9.613947368421053e-05,
      "loss": 0.4265,
      "step": 34685
    },
    {
      "epoch": 1.9423227685071116,
      "grad_norm": 1.0668704509735107,
      "learning_rate": 9.613921052631579e-05,
      "loss": 0.4395,
      "step": 34686
    },
    {
      "epoch": 1.9423787658192406,
      "grad_norm": 1.3068642616271973,
      "learning_rate": 9.613894736842106e-05,
      "loss": 0.3916,
      "step": 34687
    },
    {
      "epoch": 1.9424347631313696,
      "grad_norm": 1.35270094871521,
      "learning_rate": 9.613868421052632e-05,
      "loss": 0.3953,
      "step": 34688
    },
    {
      "epoch": 1.9424907604434987,
      "grad_norm": 1.1668734550476074,
      "learning_rate": 9.613842105263158e-05,
      "loss": 0.4772,
      "step": 34689
    },
    {
      "epoch": 1.9425467577556277,
      "grad_norm": 1.3212227821350098,
      "learning_rate": 9.613815789473684e-05,
      "loss": 0.4353,
      "step": 34690
    },
    {
      "epoch": 1.9426027550677567,
      "grad_norm": 1.2311409711837769,
      "learning_rate": 9.61378947368421e-05,
      "loss": 0.4164,
      "step": 34691
    },
    {
      "epoch": 1.9426587523798857,
      "grad_norm": 0.9628854393959045,
      "learning_rate": 9.613763157894738e-05,
      "loss": 0.3499,
      "step": 34692
    },
    {
      "epoch": 1.9427147496920147,
      "grad_norm": 1.4028321504592896,
      "learning_rate": 9.613736842105264e-05,
      "loss": 0.3137,
      "step": 34693
    },
    {
      "epoch": 1.9427707470041438,
      "grad_norm": 1.1342920064926147,
      "learning_rate": 9.61371052631579e-05,
      "loss": 0.363,
      "step": 34694
    },
    {
      "epoch": 1.9428267443162728,
      "grad_norm": 1.2898186445236206,
      "learning_rate": 9.613684210526316e-05,
      "loss": 0.4357,
      "step": 34695
    },
    {
      "epoch": 1.9428827416284018,
      "grad_norm": 1.3033515214920044,
      "learning_rate": 9.613657894736843e-05,
      "loss": 0.3959,
      "step": 34696
    },
    {
      "epoch": 1.9429387389405308,
      "grad_norm": 1.0831857919692993,
      "learning_rate": 9.613631578947369e-05,
      "loss": 0.5126,
      "step": 34697
    },
    {
      "epoch": 1.9429947362526598,
      "grad_norm": 1.411524772644043,
      "learning_rate": 9.613605263157895e-05,
      "loss": 0.4417,
      "step": 34698
    },
    {
      "epoch": 1.9430507335647889,
      "grad_norm": 1.2662649154663086,
      "learning_rate": 9.613578947368421e-05,
      "loss": 0.46,
      "step": 34699
    },
    {
      "epoch": 1.9431067308769179,
      "grad_norm": 0.9521045088768005,
      "learning_rate": 9.613552631578948e-05,
      "loss": 0.2692,
      "step": 34700
    },
    {
      "epoch": 1.943162728189047,
      "grad_norm": 1.4040699005126953,
      "learning_rate": 9.613526315789474e-05,
      "loss": 0.4689,
      "step": 34701
    },
    {
      "epoch": 1.943218725501176,
      "grad_norm": 1.1717408895492554,
      "learning_rate": 9.613500000000001e-05,
      "loss": 0.3754,
      "step": 34702
    },
    {
      "epoch": 1.943274722813305,
      "grad_norm": 25.933670043945312,
      "learning_rate": 9.613473684210526e-05,
      "loss": 0.5469,
      "step": 34703
    },
    {
      "epoch": 1.943330720125434,
      "grad_norm": 1.6813251972198486,
      "learning_rate": 9.613447368421053e-05,
      "loss": 0.5512,
      "step": 34704
    },
    {
      "epoch": 1.943386717437563,
      "grad_norm": 1.6740896701812744,
      "learning_rate": 9.613421052631579e-05,
      "loss": 0.5387,
      "step": 34705
    },
    {
      "epoch": 1.943442714749692,
      "grad_norm": 1.402829885482788,
      "learning_rate": 9.613394736842107e-05,
      "loss": 0.4147,
      "step": 34706
    },
    {
      "epoch": 1.943498712061821,
      "grad_norm": 1.3453236818313599,
      "learning_rate": 9.613368421052633e-05,
      "loss": 0.7019,
      "step": 34707
    },
    {
      "epoch": 1.94355470937395,
      "grad_norm": 1.191650390625,
      "learning_rate": 9.613342105263157e-05,
      "loss": 0.4072,
      "step": 34708
    },
    {
      "epoch": 1.943610706686079,
      "grad_norm": 0.9482975006103516,
      "learning_rate": 9.613315789473685e-05,
      "loss": 0.2927,
      "step": 34709
    },
    {
      "epoch": 1.943666703998208,
      "grad_norm": 1.2613580226898193,
      "learning_rate": 9.61328947368421e-05,
      "loss": 0.4612,
      "step": 34710
    },
    {
      "epoch": 1.943722701310337,
      "grad_norm": 1.2545610666275024,
      "learning_rate": 9.613263157894738e-05,
      "loss": 0.5167,
      "step": 34711
    },
    {
      "epoch": 1.943778698622466,
      "grad_norm": 1.2578083276748657,
      "learning_rate": 9.613236842105264e-05,
      "loss": 0.4072,
      "step": 34712
    },
    {
      "epoch": 1.9438346959345951,
      "grad_norm": 1.382576584815979,
      "learning_rate": 9.61321052631579e-05,
      "loss": 0.516,
      "step": 34713
    },
    {
      "epoch": 1.9438906932467241,
      "grad_norm": 1.1403379440307617,
      "learning_rate": 9.613184210526316e-05,
      "loss": 0.3981,
      "step": 34714
    },
    {
      "epoch": 1.9439466905588532,
      "grad_norm": 1.2607324123382568,
      "learning_rate": 9.613157894736843e-05,
      "loss": 0.5121,
      "step": 34715
    },
    {
      "epoch": 1.9440026878709822,
      "grad_norm": 1.231096625328064,
      "learning_rate": 9.613131578947369e-05,
      "loss": 0.3824,
      "step": 34716
    },
    {
      "epoch": 1.9440586851831112,
      "grad_norm": 1.4585281610488892,
      "learning_rate": 9.613105263157895e-05,
      "loss": 0.575,
      "step": 34717
    },
    {
      "epoch": 1.9441146824952402,
      "grad_norm": 1.4718551635742188,
      "learning_rate": 9.613078947368421e-05,
      "loss": 0.5371,
      "step": 34718
    },
    {
      "epoch": 1.9441706798073692,
      "grad_norm": 1.2319722175598145,
      "learning_rate": 9.613052631578948e-05,
      "loss": 0.4254,
      "step": 34719
    },
    {
      "epoch": 1.9442266771194983,
      "grad_norm": 1.2203540802001953,
      "learning_rate": 9.613026315789474e-05,
      "loss": 0.4326,
      "step": 34720
    },
    {
      "epoch": 1.9442826744316273,
      "grad_norm": 1.7114763259887695,
      "learning_rate": 9.613e-05,
      "loss": 0.3861,
      "step": 34721
    },
    {
      "epoch": 1.9443386717437563,
      "grad_norm": 1.3191580772399902,
      "learning_rate": 9.612973684210526e-05,
      "loss": 0.4873,
      "step": 34722
    },
    {
      "epoch": 1.9443946690558853,
      "grad_norm": 1.222227931022644,
      "learning_rate": 9.612947368421054e-05,
      "loss": 0.5539,
      "step": 34723
    },
    {
      "epoch": 1.9444506663680143,
      "grad_norm": 1.337488055229187,
      "learning_rate": 9.61292105263158e-05,
      "loss": 0.4003,
      "step": 34724
    },
    {
      "epoch": 1.9445066636801434,
      "grad_norm": 1.2201447486877441,
      "learning_rate": 9.612894736842106e-05,
      "loss": 0.3754,
      "step": 34725
    },
    {
      "epoch": 1.9445626609922724,
      "grad_norm": 1.3758704662322998,
      "learning_rate": 9.612868421052632e-05,
      "loss": 0.4832,
      "step": 34726
    },
    {
      "epoch": 1.9446186583044014,
      "grad_norm": 1.0471324920654297,
      "learning_rate": 9.612842105263157e-05,
      "loss": 0.4696,
      "step": 34727
    },
    {
      "epoch": 1.9446746556165304,
      "grad_norm": 2.534958839416504,
      "learning_rate": 9.612815789473685e-05,
      "loss": 0.6174,
      "step": 34728
    },
    {
      "epoch": 1.9447306529286594,
      "grad_norm": 1.1390732526779175,
      "learning_rate": 9.612789473684211e-05,
      "loss": 0.4587,
      "step": 34729
    },
    {
      "epoch": 1.9447866502407885,
      "grad_norm": 1.5098438262939453,
      "learning_rate": 9.612763157894738e-05,
      "loss": 0.4986,
      "step": 34730
    },
    {
      "epoch": 1.9448426475529175,
      "grad_norm": 1.1739249229431152,
      "learning_rate": 9.612736842105263e-05,
      "loss": 0.5055,
      "step": 34731
    },
    {
      "epoch": 1.9448986448650465,
      "grad_norm": 1.2637979984283447,
      "learning_rate": 9.61271052631579e-05,
      "loss": 0.4017,
      "step": 34732
    },
    {
      "epoch": 1.9449546421771755,
      "grad_norm": 1.1154446601867676,
      "learning_rate": 9.612684210526316e-05,
      "loss": 0.4069,
      "step": 34733
    },
    {
      "epoch": 1.9450106394893045,
      "grad_norm": 1.1956051588058472,
      "learning_rate": 9.612657894736843e-05,
      "loss": 0.406,
      "step": 34734
    },
    {
      "epoch": 1.9450666368014335,
      "grad_norm": 1.6822385787963867,
      "learning_rate": 9.61263157894737e-05,
      "loss": 0.4457,
      "step": 34735
    },
    {
      "epoch": 1.9451226341135626,
      "grad_norm": 1.1216477155685425,
      "learning_rate": 9.612605263157895e-05,
      "loss": 0.4441,
      "step": 34736
    },
    {
      "epoch": 1.9451786314256916,
      "grad_norm": 1.3763424158096313,
      "learning_rate": 9.612578947368421e-05,
      "loss": 0.4056,
      "step": 34737
    },
    {
      "epoch": 1.9452346287378206,
      "grad_norm": 1.4477684497833252,
      "learning_rate": 9.612552631578949e-05,
      "loss": 0.5025,
      "step": 34738
    },
    {
      "epoch": 1.9452906260499496,
      "grad_norm": 1.247282862663269,
      "learning_rate": 9.612526315789475e-05,
      "loss": 0.4293,
      "step": 34739
    },
    {
      "epoch": 1.9453466233620786,
      "grad_norm": 1.4127693176269531,
      "learning_rate": 9.6125e-05,
      "loss": 0.4844,
      "step": 34740
    },
    {
      "epoch": 1.9454026206742077,
      "grad_norm": 1.2757805585861206,
      "learning_rate": 9.612473684210527e-05,
      "loss": 0.5124,
      "step": 34741
    },
    {
      "epoch": 1.9454586179863367,
      "grad_norm": 1.2004910707473755,
      "learning_rate": 9.612447368421054e-05,
      "loss": 0.4907,
      "step": 34742
    },
    {
      "epoch": 1.9455146152984657,
      "grad_norm": 3.766026735305786,
      "learning_rate": 9.61242105263158e-05,
      "loss": 0.4712,
      "step": 34743
    },
    {
      "epoch": 1.9455706126105947,
      "grad_norm": 1.2574459314346313,
      "learning_rate": 9.612394736842106e-05,
      "loss": 0.5574,
      "step": 34744
    },
    {
      "epoch": 1.9456266099227237,
      "grad_norm": 1.0154584646224976,
      "learning_rate": 9.612368421052632e-05,
      "loss": 0.3641,
      "step": 34745
    },
    {
      "epoch": 1.9456826072348528,
      "grad_norm": 1.2783890962600708,
      "learning_rate": 9.612342105263158e-05,
      "loss": 0.4259,
      "step": 34746
    },
    {
      "epoch": 1.9457386045469818,
      "grad_norm": 1.3876056671142578,
      "learning_rate": 9.612315789473685e-05,
      "loss": 0.4498,
      "step": 34747
    },
    {
      "epoch": 1.9457946018591108,
      "grad_norm": 1.3685499429702759,
      "learning_rate": 9.612289473684211e-05,
      "loss": 0.6399,
      "step": 34748
    },
    {
      "epoch": 1.9458505991712398,
      "grad_norm": 1.4972935914993286,
      "learning_rate": 9.612263157894737e-05,
      "loss": 0.4284,
      "step": 34749
    },
    {
      "epoch": 1.9459065964833688,
      "grad_norm": 1.1798095703125,
      "learning_rate": 9.612236842105263e-05,
      "loss": 0.3664,
      "step": 34750
    },
    {
      "epoch": 1.9459625937954979,
      "grad_norm": 1.0813255310058594,
      "learning_rate": 9.61221052631579e-05,
      "loss": 0.3635,
      "step": 34751
    },
    {
      "epoch": 1.9460185911076269,
      "grad_norm": 1.1028759479522705,
      "learning_rate": 9.612184210526316e-05,
      "loss": 0.3612,
      "step": 34752
    },
    {
      "epoch": 1.946074588419756,
      "grad_norm": 1.4469988346099854,
      "learning_rate": 9.612157894736842e-05,
      "loss": 0.5838,
      "step": 34753
    },
    {
      "epoch": 1.946130585731885,
      "grad_norm": 1.3017935752868652,
      "learning_rate": 9.612131578947368e-05,
      "loss": 0.3959,
      "step": 34754
    },
    {
      "epoch": 1.946186583044014,
      "grad_norm": 1.4254018068313599,
      "learning_rate": 9.612105263157896e-05,
      "loss": 0.5403,
      "step": 34755
    },
    {
      "epoch": 1.946242580356143,
      "grad_norm": 1.5793259143829346,
      "learning_rate": 9.612078947368422e-05,
      "loss": 0.3522,
      "step": 34756
    },
    {
      "epoch": 1.946298577668272,
      "grad_norm": 1.7903027534484863,
      "learning_rate": 9.612052631578949e-05,
      "loss": 0.4963,
      "step": 34757
    },
    {
      "epoch": 1.946354574980401,
      "grad_norm": 1.037590503692627,
      "learning_rate": 9.612026315789473e-05,
      "loss": 0.335,
      "step": 34758
    },
    {
      "epoch": 1.94641057229253,
      "grad_norm": 1.1669597625732422,
      "learning_rate": 9.612000000000001e-05,
      "loss": 0.557,
      "step": 34759
    },
    {
      "epoch": 1.946466569604659,
      "grad_norm": 1.1176064014434814,
      "learning_rate": 9.611973684210527e-05,
      "loss": 0.34,
      "step": 34760
    },
    {
      "epoch": 1.946522566916788,
      "grad_norm": 1.3815439939498901,
      "learning_rate": 9.611947368421053e-05,
      "loss": 0.5729,
      "step": 34761
    },
    {
      "epoch": 1.946578564228917,
      "grad_norm": 1.4713155031204224,
      "learning_rate": 9.61192105263158e-05,
      "loss": 0.6223,
      "step": 34762
    },
    {
      "epoch": 1.946634561541046,
      "grad_norm": 1.3211703300476074,
      "learning_rate": 9.611894736842105e-05,
      "loss": 0.4572,
      "step": 34763
    },
    {
      "epoch": 1.946690558853175,
      "grad_norm": 1.4970693588256836,
      "learning_rate": 9.611868421052632e-05,
      "loss": 0.4681,
      "step": 34764
    },
    {
      "epoch": 1.9467465561653041,
      "grad_norm": 1.1335227489471436,
      "learning_rate": 9.611842105263158e-05,
      "loss": 0.3592,
      "step": 34765
    },
    {
      "epoch": 1.9468025534774331,
      "grad_norm": 1.2893025875091553,
      "learning_rate": 9.611815789473685e-05,
      "loss": 0.428,
      "step": 34766
    },
    {
      "epoch": 1.9468585507895622,
      "grad_norm": 1.251278281211853,
      "learning_rate": 9.611789473684211e-05,
      "loss": 0.4115,
      "step": 34767
    },
    {
      "epoch": 1.9469145481016912,
      "grad_norm": 1.3840177059173584,
      "learning_rate": 9.611763157894737e-05,
      "loss": 0.5638,
      "step": 34768
    },
    {
      "epoch": 1.9469705454138202,
      "grad_norm": 1.361457109451294,
      "learning_rate": 9.611736842105263e-05,
      "loss": 0.5049,
      "step": 34769
    },
    {
      "epoch": 1.9470265427259492,
      "grad_norm": 1.1503407955169678,
      "learning_rate": 9.61171052631579e-05,
      "loss": 0.4856,
      "step": 34770
    },
    {
      "epoch": 1.9470825400380782,
      "grad_norm": 1.0933759212493896,
      "learning_rate": 9.611684210526317e-05,
      "loss": 0.4899,
      "step": 34771
    },
    {
      "epoch": 1.9471385373502073,
      "grad_norm": 1.3940249681472778,
      "learning_rate": 9.611657894736843e-05,
      "loss": 0.5338,
      "step": 34772
    },
    {
      "epoch": 1.9471945346623363,
      "grad_norm": 1.3528846502304077,
      "learning_rate": 9.611631578947368e-05,
      "loss": 0.4647,
      "step": 34773
    },
    {
      "epoch": 1.9472505319744653,
      "grad_norm": 1.2399296760559082,
      "learning_rate": 9.611605263157896e-05,
      "loss": 0.4262,
      "step": 34774
    },
    {
      "epoch": 1.9473065292865943,
      "grad_norm": 1.2943211793899536,
      "learning_rate": 9.611578947368422e-05,
      "loss": 0.4055,
      "step": 34775
    },
    {
      "epoch": 1.9473625265987233,
      "grad_norm": 1.335343837738037,
      "learning_rate": 9.611552631578948e-05,
      "loss": 0.4456,
      "step": 34776
    },
    {
      "epoch": 1.9474185239108524,
      "grad_norm": 1.2676340341567993,
      "learning_rate": 9.611526315789474e-05,
      "loss": 0.422,
      "step": 34777
    },
    {
      "epoch": 1.9474745212229814,
      "grad_norm": 1.2184985876083374,
      "learning_rate": 9.6115e-05,
      "loss": 0.3713,
      "step": 34778
    },
    {
      "epoch": 1.9475305185351104,
      "grad_norm": 1.112838625907898,
      "learning_rate": 9.611473684210527e-05,
      "loss": 0.4254,
      "step": 34779
    },
    {
      "epoch": 1.9475865158472394,
      "grad_norm": 1.2566959857940674,
      "learning_rate": 9.611447368421053e-05,
      "loss": 0.3828,
      "step": 34780
    },
    {
      "epoch": 1.9476425131593684,
      "grad_norm": 1.3487221002578735,
      "learning_rate": 9.611421052631579e-05,
      "loss": 0.4829,
      "step": 34781
    },
    {
      "epoch": 1.9476985104714974,
      "grad_norm": 1.2248846292495728,
      "learning_rate": 9.611394736842105e-05,
      "loss": 0.5197,
      "step": 34782
    },
    {
      "epoch": 1.9477545077836265,
      "grad_norm": 1.3060572147369385,
      "learning_rate": 9.611368421052632e-05,
      "loss": 0.3828,
      "step": 34783
    },
    {
      "epoch": 1.9478105050957555,
      "grad_norm": 1.038963794708252,
      "learning_rate": 9.611342105263158e-05,
      "loss": 0.3626,
      "step": 34784
    },
    {
      "epoch": 1.9478665024078845,
      "grad_norm": 1.215692400932312,
      "learning_rate": 9.611315789473686e-05,
      "loss": 0.4817,
      "step": 34785
    },
    {
      "epoch": 1.9479224997200135,
      "grad_norm": 1.2252408266067505,
      "learning_rate": 9.61128947368421e-05,
      "loss": 0.3279,
      "step": 34786
    },
    {
      "epoch": 1.9479784970321425,
      "grad_norm": 1.3991550207138062,
      "learning_rate": 9.611263157894738e-05,
      "loss": 0.4552,
      "step": 34787
    },
    {
      "epoch": 1.9480344943442716,
      "grad_norm": 1.2616218328475952,
      "learning_rate": 9.611236842105264e-05,
      "loss": 0.4275,
      "step": 34788
    },
    {
      "epoch": 1.9480904916564006,
      "grad_norm": 1.5421687364578247,
      "learning_rate": 9.611210526315791e-05,
      "loss": 0.5387,
      "step": 34789
    },
    {
      "epoch": 1.9481464889685296,
      "grad_norm": 1.352452278137207,
      "learning_rate": 9.611184210526315e-05,
      "loss": 0.4082,
      "step": 34790
    },
    {
      "epoch": 1.9482024862806586,
      "grad_norm": 1.048331379890442,
      "learning_rate": 9.611157894736843e-05,
      "loss": 0.3665,
      "step": 34791
    },
    {
      "epoch": 1.9482584835927876,
      "grad_norm": 1.5675545930862427,
      "learning_rate": 9.611131578947369e-05,
      "loss": 0.4322,
      "step": 34792
    },
    {
      "epoch": 1.9483144809049167,
      "grad_norm": 1.3555084466934204,
      "learning_rate": 9.611105263157896e-05,
      "loss": 0.4585,
      "step": 34793
    },
    {
      "epoch": 1.9483704782170457,
      "grad_norm": 1.2935302257537842,
      "learning_rate": 9.611078947368422e-05,
      "loss": 0.4026,
      "step": 34794
    },
    {
      "epoch": 1.9484264755291747,
      "grad_norm": 1.2858777046203613,
      "learning_rate": 9.611052631578947e-05,
      "loss": 0.4629,
      "step": 34795
    },
    {
      "epoch": 1.9484824728413037,
      "grad_norm": 1.4356635808944702,
      "learning_rate": 9.611026315789474e-05,
      "loss": 0.457,
      "step": 34796
    },
    {
      "epoch": 1.9485384701534327,
      "grad_norm": 1.1528637409210205,
      "learning_rate": 9.611e-05,
      "loss": 0.4287,
      "step": 34797
    },
    {
      "epoch": 1.9485944674655618,
      "grad_norm": 1.2664382457733154,
      "learning_rate": 9.610973684210527e-05,
      "loss": 0.4413,
      "step": 34798
    },
    {
      "epoch": 1.9486504647776908,
      "grad_norm": 1.2803250551223755,
      "learning_rate": 9.610947368421053e-05,
      "loss": 0.3677,
      "step": 34799
    },
    {
      "epoch": 1.9487064620898198,
      "grad_norm": 1.3596066236495972,
      "learning_rate": 9.610921052631579e-05,
      "loss": 0.4128,
      "step": 34800
    },
    {
      "epoch": 1.9487624594019488,
      "grad_norm": 1.3163148164749146,
      "learning_rate": 9.610894736842105e-05,
      "loss": 0.4701,
      "step": 34801
    },
    {
      "epoch": 1.9488184567140778,
      "grad_norm": 1.1898585557937622,
      "learning_rate": 9.610868421052633e-05,
      "loss": 0.4409,
      "step": 34802
    },
    {
      "epoch": 1.9488744540262068,
      "grad_norm": 1.1987719535827637,
      "learning_rate": 9.610842105263159e-05,
      "loss": 0.4209,
      "step": 34803
    },
    {
      "epoch": 1.9489304513383359,
      "grad_norm": 1.1841790676116943,
      "learning_rate": 9.610815789473684e-05,
      "loss": 0.4202,
      "step": 34804
    },
    {
      "epoch": 1.9489864486504649,
      "grad_norm": 5.456706523895264,
      "learning_rate": 9.61078947368421e-05,
      "loss": 0.404,
      "step": 34805
    },
    {
      "epoch": 1.949042445962594,
      "grad_norm": 1.2145456075668335,
      "learning_rate": 9.610763157894738e-05,
      "loss": 0.4286,
      "step": 34806
    },
    {
      "epoch": 1.949098443274723,
      "grad_norm": 1.3066177368164062,
      "learning_rate": 9.610736842105264e-05,
      "loss": 0.4138,
      "step": 34807
    },
    {
      "epoch": 1.949154440586852,
      "grad_norm": 1.01899254322052,
      "learning_rate": 9.61071052631579e-05,
      "loss": 0.4498,
      "step": 34808
    },
    {
      "epoch": 1.949210437898981,
      "grad_norm": 1.2671171426773071,
      "learning_rate": 9.610684210526316e-05,
      "loss": 0.4301,
      "step": 34809
    },
    {
      "epoch": 1.94926643521111,
      "grad_norm": 1.398147702217102,
      "learning_rate": 9.610657894736843e-05,
      "loss": 0.5394,
      "step": 34810
    },
    {
      "epoch": 1.949322432523239,
      "grad_norm": 1.791809320449829,
      "learning_rate": 9.610631578947369e-05,
      "loss": 0.5838,
      "step": 34811
    },
    {
      "epoch": 1.949378429835368,
      "grad_norm": 2.0739619731903076,
      "learning_rate": 9.610605263157895e-05,
      "loss": 0.3496,
      "step": 34812
    },
    {
      "epoch": 1.949434427147497,
      "grad_norm": 1.506271243095398,
      "learning_rate": 9.610578947368421e-05,
      "loss": 0.5248,
      "step": 34813
    },
    {
      "epoch": 1.949490424459626,
      "grad_norm": 1.480880618095398,
      "learning_rate": 9.610552631578947e-05,
      "loss": 0.4734,
      "step": 34814
    },
    {
      "epoch": 1.949546421771755,
      "grad_norm": 1.0420907735824585,
      "learning_rate": 9.610526315789474e-05,
      "loss": 0.3872,
      "step": 34815
    },
    {
      "epoch": 1.949602419083884,
      "grad_norm": 1.2384741306304932,
      "learning_rate": 9.6105e-05,
      "loss": 0.5088,
      "step": 34816
    },
    {
      "epoch": 1.9496584163960131,
      "grad_norm": 1.5229109525680542,
      "learning_rate": 9.610473684210528e-05,
      "loss": 0.5538,
      "step": 34817
    },
    {
      "epoch": 1.9497144137081421,
      "grad_norm": 1.1750011444091797,
      "learning_rate": 9.610447368421052e-05,
      "loss": 0.49,
      "step": 34818
    },
    {
      "epoch": 1.9497704110202712,
      "grad_norm": 1.2158242464065552,
      "learning_rate": 9.61042105263158e-05,
      "loss": 0.5002,
      "step": 34819
    },
    {
      "epoch": 1.9498264083324002,
      "grad_norm": 1.3357243537902832,
      "learning_rate": 9.610394736842105e-05,
      "loss": 0.4838,
      "step": 34820
    },
    {
      "epoch": 1.9498824056445292,
      "grad_norm": 1.097816824913025,
      "learning_rate": 9.610368421052633e-05,
      "loss": 0.4567,
      "step": 34821
    },
    {
      "epoch": 1.9499384029566582,
      "grad_norm": 1.206210732460022,
      "learning_rate": 9.610342105263159e-05,
      "loss": 0.4362,
      "step": 34822
    },
    {
      "epoch": 1.9499944002687872,
      "grad_norm": 1.408158540725708,
      "learning_rate": 9.610315789473685e-05,
      "loss": 0.5342,
      "step": 34823
    },
    {
      "epoch": 1.9500503975809163,
      "grad_norm": 1.2987158298492432,
      "learning_rate": 9.610289473684211e-05,
      "loss": 0.4069,
      "step": 34824
    },
    {
      "epoch": 1.9501063948930453,
      "grad_norm": 1.2456448078155518,
      "learning_rate": 9.610263157894738e-05,
      "loss": 0.4683,
      "step": 34825
    },
    {
      "epoch": 1.9501623922051743,
      "grad_norm": 1.2461268901824951,
      "learning_rate": 9.610236842105264e-05,
      "loss": 0.4657,
      "step": 34826
    },
    {
      "epoch": 1.9502183895173033,
      "grad_norm": 1.597239375114441,
      "learning_rate": 9.61021052631579e-05,
      "loss": 0.5867,
      "step": 34827
    },
    {
      "epoch": 1.9502743868294323,
      "grad_norm": 1.554483413696289,
      "learning_rate": 9.610184210526316e-05,
      "loss": 0.4291,
      "step": 34828
    },
    {
      "epoch": 1.9503303841415613,
      "grad_norm": 1.6899200677871704,
      "learning_rate": 9.610157894736842e-05,
      "loss": 0.4518,
      "step": 34829
    },
    {
      "epoch": 1.9503863814536904,
      "grad_norm": 2.320903778076172,
      "learning_rate": 9.610131578947369e-05,
      "loss": 0.6214,
      "step": 34830
    },
    {
      "epoch": 1.9504423787658194,
      "grad_norm": 1.515985369682312,
      "learning_rate": 9.610105263157895e-05,
      "loss": 0.6437,
      "step": 34831
    },
    {
      "epoch": 1.9504983760779484,
      "grad_norm": 1.6087281703948975,
      "learning_rate": 9.610078947368421e-05,
      "loss": 0.4387,
      "step": 34832
    },
    {
      "epoch": 1.9505543733900774,
      "grad_norm": 1.1448251008987427,
      "learning_rate": 9.610052631578947e-05,
      "loss": 0.5262,
      "step": 34833
    },
    {
      "epoch": 1.9506103707022064,
      "grad_norm": 1.3280251026153564,
      "learning_rate": 9.610026315789475e-05,
      "loss": 0.4794,
      "step": 34834
    },
    {
      "epoch": 1.9506663680143355,
      "grad_norm": 1.1608219146728516,
      "learning_rate": 9.61e-05,
      "loss": 0.4233,
      "step": 34835
    },
    {
      "epoch": 1.9507223653264645,
      "grad_norm": 1.6304550170898438,
      "learning_rate": 9.609973684210526e-05,
      "loss": 0.5261,
      "step": 34836
    },
    {
      "epoch": 1.9507783626385935,
      "grad_norm": 1.2674919366836548,
      "learning_rate": 9.609947368421052e-05,
      "loss": 0.4454,
      "step": 34837
    },
    {
      "epoch": 1.9508343599507225,
      "grad_norm": 1.1969736814498901,
      "learning_rate": 9.60992105263158e-05,
      "loss": 0.3657,
      "step": 34838
    },
    {
      "epoch": 1.9508903572628515,
      "grad_norm": 1.1888172626495361,
      "learning_rate": 9.609894736842106e-05,
      "loss": 0.43,
      "step": 34839
    },
    {
      "epoch": 1.9509463545749806,
      "grad_norm": 1.346352458000183,
      "learning_rate": 9.609868421052633e-05,
      "loss": 0.388,
      "step": 34840
    },
    {
      "epoch": 1.9510023518871096,
      "grad_norm": 1.221828579902649,
      "learning_rate": 9.609842105263158e-05,
      "loss": 0.3512,
      "step": 34841
    },
    {
      "epoch": 1.9510583491992386,
      "grad_norm": 1.2034492492675781,
      "learning_rate": 9.609815789473685e-05,
      "loss": 0.4418,
      "step": 34842
    },
    {
      "epoch": 1.9511143465113676,
      "grad_norm": 1.2048699855804443,
      "learning_rate": 9.609789473684211e-05,
      "loss": 0.4099,
      "step": 34843
    },
    {
      "epoch": 1.9511703438234966,
      "grad_norm": 1.17389976978302,
      "learning_rate": 9.609763157894738e-05,
      "loss": 0.3761,
      "step": 34844
    },
    {
      "epoch": 1.9512263411356257,
      "grad_norm": 1.4838505983352661,
      "learning_rate": 9.609736842105263e-05,
      "loss": 0.4248,
      "step": 34845
    },
    {
      "epoch": 1.9512823384477547,
      "grad_norm": 1.2922229766845703,
      "learning_rate": 9.609710526315789e-05,
      "loss": 0.3797,
      "step": 34846
    },
    {
      "epoch": 1.9513383357598837,
      "grad_norm": 1.8989818096160889,
      "learning_rate": 9.609684210526316e-05,
      "loss": 0.373,
      "step": 34847
    },
    {
      "epoch": 1.9513943330720125,
      "grad_norm": 1.9176208972930908,
      "learning_rate": 9.609657894736842e-05,
      "loss": 0.5793,
      "step": 34848
    },
    {
      "epoch": 1.9514503303841415,
      "grad_norm": 1.0938307046890259,
      "learning_rate": 9.60963157894737e-05,
      "loss": 0.3922,
      "step": 34849
    },
    {
      "epoch": 1.9515063276962705,
      "grad_norm": 1.2032626867294312,
      "learning_rate": 9.609605263157894e-05,
      "loss": 0.3815,
      "step": 34850
    },
    {
      "epoch": 1.9515623250083995,
      "grad_norm": 1.1053210496902466,
      "learning_rate": 9.609578947368421e-05,
      "loss": 0.3817,
      "step": 34851
    },
    {
      "epoch": 1.9516183223205286,
      "grad_norm": 1.1857454776763916,
      "learning_rate": 9.609552631578947e-05,
      "loss": 0.5533,
      "step": 34852
    },
    {
      "epoch": 1.9516743196326576,
      "grad_norm": 1.1610991954803467,
      "learning_rate": 9.609526315789475e-05,
      "loss": 0.3912,
      "step": 34853
    },
    {
      "epoch": 1.9517303169447866,
      "grad_norm": 1.2906380891799927,
      "learning_rate": 9.609500000000001e-05,
      "loss": 0.5849,
      "step": 34854
    },
    {
      "epoch": 1.9517863142569156,
      "grad_norm": 1.6633220911026,
      "learning_rate": 9.609473684210527e-05,
      "loss": 0.4415,
      "step": 34855
    },
    {
      "epoch": 1.9518423115690446,
      "grad_norm": 1.0566508769989014,
      "learning_rate": 9.609447368421053e-05,
      "loss": 0.4489,
      "step": 34856
    },
    {
      "epoch": 1.9518983088811737,
      "grad_norm": 1.4683705568313599,
      "learning_rate": 9.60942105263158e-05,
      "loss": 0.4707,
      "step": 34857
    },
    {
      "epoch": 1.9519543061933027,
      "grad_norm": 1.2812076807022095,
      "learning_rate": 9.609394736842106e-05,
      "loss": 0.3667,
      "step": 34858
    },
    {
      "epoch": 1.9520103035054317,
      "grad_norm": 1.259263038635254,
      "learning_rate": 9.609368421052632e-05,
      "loss": 0.3893,
      "step": 34859
    },
    {
      "epoch": 1.9520663008175607,
      "grad_norm": 1.582943081855774,
      "learning_rate": 9.609342105263158e-05,
      "loss": 0.4785,
      "step": 34860
    },
    {
      "epoch": 1.9521222981296897,
      "grad_norm": 1.4122960567474365,
      "learning_rate": 9.609315789473685e-05,
      "loss": 0.489,
      "step": 34861
    },
    {
      "epoch": 1.9521782954418188,
      "grad_norm": 1.5275075435638428,
      "learning_rate": 9.609289473684211e-05,
      "loss": 0.5193,
      "step": 34862
    },
    {
      "epoch": 1.9522342927539478,
      "grad_norm": 1.2714399099349976,
      "learning_rate": 9.609263157894737e-05,
      "loss": 0.3157,
      "step": 34863
    },
    {
      "epoch": 1.9522902900660768,
      "grad_norm": 1.472785234451294,
      "learning_rate": 9.609236842105263e-05,
      "loss": 0.6184,
      "step": 34864
    },
    {
      "epoch": 1.9523462873782058,
      "grad_norm": 0.9683521389961243,
      "learning_rate": 9.609210526315789e-05,
      "loss": 0.3978,
      "step": 34865
    },
    {
      "epoch": 1.9524022846903348,
      "grad_norm": 1.248137354850769,
      "learning_rate": 9.609184210526316e-05,
      "loss": 0.4805,
      "step": 34866
    },
    {
      "epoch": 1.9524582820024639,
      "grad_norm": 1.1852248907089233,
      "learning_rate": 9.609157894736842e-05,
      "loss": 0.4315,
      "step": 34867
    },
    {
      "epoch": 1.9525142793145929,
      "grad_norm": 1.2343366146087646,
      "learning_rate": 9.609131578947368e-05,
      "loss": 0.3617,
      "step": 34868
    },
    {
      "epoch": 1.952570276626722,
      "grad_norm": 1.2805016040802002,
      "learning_rate": 9.609105263157894e-05,
      "loss": 0.4281,
      "step": 34869
    },
    {
      "epoch": 1.952626273938851,
      "grad_norm": 1.263858675956726,
      "learning_rate": 9.609078947368422e-05,
      "loss": 0.4489,
      "step": 34870
    },
    {
      "epoch": 1.95268227125098,
      "grad_norm": 1.2028876543045044,
      "learning_rate": 9.609052631578948e-05,
      "loss": 0.4111,
      "step": 34871
    },
    {
      "epoch": 1.952738268563109,
      "grad_norm": 1.2021613121032715,
      "learning_rate": 9.609026315789475e-05,
      "loss": 0.3796,
      "step": 34872
    },
    {
      "epoch": 1.952794265875238,
      "grad_norm": 1.22563636302948,
      "learning_rate": 9.609e-05,
      "loss": 0.3896,
      "step": 34873
    },
    {
      "epoch": 1.952850263187367,
      "grad_norm": 1.5378974676132202,
      "learning_rate": 9.608973684210527e-05,
      "loss": 0.4585,
      "step": 34874
    },
    {
      "epoch": 1.952906260499496,
      "grad_norm": 1.2581027746200562,
      "learning_rate": 9.608947368421053e-05,
      "loss": 0.458,
      "step": 34875
    },
    {
      "epoch": 1.952962257811625,
      "grad_norm": 1.5139538049697876,
      "learning_rate": 9.60892105263158e-05,
      "loss": 0.4948,
      "step": 34876
    },
    {
      "epoch": 1.953018255123754,
      "grad_norm": 1.8972543478012085,
      "learning_rate": 9.608894736842106e-05,
      "loss": 0.4417,
      "step": 34877
    },
    {
      "epoch": 1.953074252435883,
      "grad_norm": 1.1931805610656738,
      "learning_rate": 9.608868421052632e-05,
      "loss": 0.5338,
      "step": 34878
    },
    {
      "epoch": 1.953130249748012,
      "grad_norm": 1.2430371046066284,
      "learning_rate": 9.608842105263158e-05,
      "loss": 0.4099,
      "step": 34879
    },
    {
      "epoch": 1.953186247060141,
      "grad_norm": 1.3579869270324707,
      "learning_rate": 9.608815789473686e-05,
      "loss": 0.6221,
      "step": 34880
    },
    {
      "epoch": 1.9532422443722701,
      "grad_norm": 1.0779386758804321,
      "learning_rate": 9.608789473684211e-05,
      "loss": 0.361,
      "step": 34881
    },
    {
      "epoch": 1.9532982416843991,
      "grad_norm": 1.4587692022323608,
      "learning_rate": 9.608763157894737e-05,
      "loss": 0.4174,
      "step": 34882
    },
    {
      "epoch": 1.9533542389965282,
      "grad_norm": 1.1698614358901978,
      "learning_rate": 9.608736842105263e-05,
      "loss": 0.4295,
      "step": 34883
    },
    {
      "epoch": 1.9534102363086572,
      "grad_norm": 1.2938059568405151,
      "learning_rate": 9.60871052631579e-05,
      "loss": 0.4211,
      "step": 34884
    },
    {
      "epoch": 1.9534662336207862,
      "grad_norm": 1.0886733531951904,
      "learning_rate": 9.608684210526317e-05,
      "loss": 0.4006,
      "step": 34885
    },
    {
      "epoch": 1.9535222309329152,
      "grad_norm": 1.319171667098999,
      "learning_rate": 9.608657894736843e-05,
      "loss": 0.3378,
      "step": 34886
    },
    {
      "epoch": 1.9535782282450442,
      "grad_norm": 1.29500412940979,
      "learning_rate": 9.608631578947369e-05,
      "loss": 0.449,
      "step": 34887
    },
    {
      "epoch": 1.9536342255571733,
      "grad_norm": 1.24713134765625,
      "learning_rate": 9.608605263157895e-05,
      "loss": 0.4236,
      "step": 34888
    },
    {
      "epoch": 1.9536902228693023,
      "grad_norm": 6.881576061248779,
      "learning_rate": 9.608578947368422e-05,
      "loss": 0.3765,
      "step": 34889
    },
    {
      "epoch": 1.9537462201814313,
      "grad_norm": 1.4172170162200928,
      "learning_rate": 9.608552631578948e-05,
      "loss": 0.4722,
      "step": 34890
    },
    {
      "epoch": 1.9538022174935603,
      "grad_norm": 1.6852965354919434,
      "learning_rate": 9.608526315789474e-05,
      "loss": 0.4372,
      "step": 34891
    },
    {
      "epoch": 1.9538582148056893,
      "grad_norm": 1.4383089542388916,
      "learning_rate": 9.6085e-05,
      "loss": 0.5763,
      "step": 34892
    },
    {
      "epoch": 1.9539142121178183,
      "grad_norm": 1.2315882444381714,
      "learning_rate": 9.608473684210527e-05,
      "loss": 0.3473,
      "step": 34893
    },
    {
      "epoch": 1.9539702094299474,
      "grad_norm": 1.3040111064910889,
      "learning_rate": 9.608447368421053e-05,
      "loss": 0.432,
      "step": 34894
    },
    {
      "epoch": 1.9540262067420764,
      "grad_norm": 1.5814647674560547,
      "learning_rate": 9.60842105263158e-05,
      "loss": 0.5433,
      "step": 34895
    },
    {
      "epoch": 1.9540822040542054,
      "grad_norm": 1.3125485181808472,
      "learning_rate": 9.608394736842105e-05,
      "loss": 0.4704,
      "step": 34896
    },
    {
      "epoch": 1.9541382013663344,
      "grad_norm": 1.228325605392456,
      "learning_rate": 9.608368421052632e-05,
      "loss": 0.3786,
      "step": 34897
    },
    {
      "epoch": 1.9541941986784634,
      "grad_norm": 1.4973866939544678,
      "learning_rate": 9.608342105263158e-05,
      "loss": 0.39,
      "step": 34898
    },
    {
      "epoch": 1.9542501959905925,
      "grad_norm": 1.1977494955062866,
      "learning_rate": 9.608315789473684e-05,
      "loss": 0.4249,
      "step": 34899
    },
    {
      "epoch": 1.9543061933027215,
      "grad_norm": 1.2408409118652344,
      "learning_rate": 9.60828947368421e-05,
      "loss": 0.4704,
      "step": 34900
    },
    {
      "epoch": 1.9543621906148505,
      "grad_norm": 1.1670175790786743,
      "learning_rate": 9.608263157894736e-05,
      "loss": 0.5145,
      "step": 34901
    },
    {
      "epoch": 1.9544181879269795,
      "grad_norm": 1.4870555400848389,
      "learning_rate": 9.608236842105264e-05,
      "loss": 0.4691,
      "step": 34902
    },
    {
      "epoch": 1.9544741852391085,
      "grad_norm": 1.1747153997421265,
      "learning_rate": 9.60821052631579e-05,
      "loss": 0.3679,
      "step": 34903
    },
    {
      "epoch": 1.9545301825512376,
      "grad_norm": 1.1523088216781616,
      "learning_rate": 9.608184210526317e-05,
      "loss": 0.4698,
      "step": 34904
    },
    {
      "epoch": 1.9545861798633666,
      "grad_norm": 1.3853576183319092,
      "learning_rate": 9.608157894736842e-05,
      "loss": 0.4367,
      "step": 34905
    },
    {
      "epoch": 1.9546421771754956,
      "grad_norm": 1.1110098361968994,
      "learning_rate": 9.608131578947369e-05,
      "loss": 0.456,
      "step": 34906
    },
    {
      "epoch": 1.9546981744876246,
      "grad_norm": 0.9134936332702637,
      "learning_rate": 9.608105263157895e-05,
      "loss": 0.2954,
      "step": 34907
    },
    {
      "epoch": 1.9547541717997536,
      "grad_norm": 1.0795981884002686,
      "learning_rate": 9.608078947368422e-05,
      "loss": 0.3645,
      "step": 34908
    },
    {
      "epoch": 1.9548101691118827,
      "grad_norm": 1.0911567211151123,
      "learning_rate": 9.608052631578948e-05,
      "loss": 0.376,
      "step": 34909
    },
    {
      "epoch": 1.9548661664240117,
      "grad_norm": 1.2310270071029663,
      "learning_rate": 9.608026315789474e-05,
      "loss": 0.5349,
      "step": 34910
    },
    {
      "epoch": 1.9549221637361407,
      "grad_norm": 1.1885923147201538,
      "learning_rate": 9.608e-05,
      "loss": 0.3642,
      "step": 34911
    },
    {
      "epoch": 1.9549781610482697,
      "grad_norm": 1.2332559823989868,
      "learning_rate": 9.607973684210527e-05,
      "loss": 0.4253,
      "step": 34912
    },
    {
      "epoch": 1.9550341583603987,
      "grad_norm": 1.3652539253234863,
      "learning_rate": 9.607947368421053e-05,
      "loss": 0.3514,
      "step": 34913
    },
    {
      "epoch": 1.9550901556725278,
      "grad_norm": 1.0595307350158691,
      "learning_rate": 9.60792105263158e-05,
      "loss": 0.3562,
      "step": 34914
    },
    {
      "epoch": 1.9551461529846568,
      "grad_norm": 1.3542340993881226,
      "learning_rate": 9.607894736842105e-05,
      "loss": 0.342,
      "step": 34915
    },
    {
      "epoch": 1.9552021502967858,
      "grad_norm": 1.07725191116333,
      "learning_rate": 9.607868421052631e-05,
      "loss": 0.3868,
      "step": 34916
    },
    {
      "epoch": 1.9552581476089148,
      "grad_norm": 1.141705870628357,
      "learning_rate": 9.607842105263159e-05,
      "loss": 0.3354,
      "step": 34917
    },
    {
      "epoch": 1.9553141449210438,
      "grad_norm": 1.359151005744934,
      "learning_rate": 9.607815789473685e-05,
      "loss": 0.5093,
      "step": 34918
    },
    {
      "epoch": 1.9553701422331728,
      "grad_norm": 1.2323857545852661,
      "learning_rate": 9.60778947368421e-05,
      "loss": 0.4207,
      "step": 34919
    },
    {
      "epoch": 1.9554261395453019,
      "grad_norm": 1.1758780479431152,
      "learning_rate": 9.607763157894737e-05,
      "loss": 0.4891,
      "step": 34920
    },
    {
      "epoch": 1.9554821368574309,
      "grad_norm": 1.0693532228469849,
      "learning_rate": 9.607736842105264e-05,
      "loss": 0.3488,
      "step": 34921
    },
    {
      "epoch": 1.95553813416956,
      "grad_norm": 1.350893497467041,
      "learning_rate": 9.60771052631579e-05,
      "loss": 0.4233,
      "step": 34922
    },
    {
      "epoch": 1.955594131481689,
      "grad_norm": 1.453223705291748,
      "learning_rate": 9.607684210526316e-05,
      "loss": 0.4068,
      "step": 34923
    },
    {
      "epoch": 1.955650128793818,
      "grad_norm": 1.3900030851364136,
      "learning_rate": 9.607657894736842e-05,
      "loss": 0.4189,
      "step": 34924
    },
    {
      "epoch": 1.955706126105947,
      "grad_norm": 1.5687403678894043,
      "learning_rate": 9.607631578947369e-05,
      "loss": 0.5702,
      "step": 34925
    },
    {
      "epoch": 1.955762123418076,
      "grad_norm": 1.1448888778686523,
      "learning_rate": 9.607605263157895e-05,
      "loss": 0.5064,
      "step": 34926
    },
    {
      "epoch": 1.955818120730205,
      "grad_norm": 1.2428531646728516,
      "learning_rate": 9.607578947368423e-05,
      "loss": 0.4836,
      "step": 34927
    },
    {
      "epoch": 1.955874118042334,
      "grad_norm": 1.1179463863372803,
      "learning_rate": 9.607552631578947e-05,
      "loss": 0.406,
      "step": 34928
    },
    {
      "epoch": 1.955930115354463,
      "grad_norm": 1.4065165519714355,
      "learning_rate": 9.607526315789474e-05,
      "loss": 0.4229,
      "step": 34929
    },
    {
      "epoch": 1.955986112666592,
      "grad_norm": 1.1828871965408325,
      "learning_rate": 9.6075e-05,
      "loss": 0.569,
      "step": 34930
    },
    {
      "epoch": 1.9560421099787209,
      "grad_norm": 1.2322351932525635,
      "learning_rate": 9.607473684210528e-05,
      "loss": 0.4409,
      "step": 34931
    },
    {
      "epoch": 1.9560981072908499,
      "grad_norm": 1.1226717233657837,
      "learning_rate": 9.607447368421054e-05,
      "loss": 0.4559,
      "step": 34932
    },
    {
      "epoch": 1.956154104602979,
      "grad_norm": 1.2956340312957764,
      "learning_rate": 9.607421052631578e-05,
      "loss": 0.3332,
      "step": 34933
    },
    {
      "epoch": 1.956210101915108,
      "grad_norm": 1.3241641521453857,
      "learning_rate": 9.607394736842106e-05,
      "loss": 0.4315,
      "step": 34934
    },
    {
      "epoch": 1.956266099227237,
      "grad_norm": 1.2096668481826782,
      "learning_rate": 9.607368421052632e-05,
      "loss": 0.5941,
      "step": 34935
    },
    {
      "epoch": 1.956322096539366,
      "grad_norm": 1.2623282670974731,
      "learning_rate": 9.607342105263159e-05,
      "loss": 0.4363,
      "step": 34936
    },
    {
      "epoch": 1.956378093851495,
      "grad_norm": 1.1424297094345093,
      "learning_rate": 9.607315789473685e-05,
      "loss": 0.4514,
      "step": 34937
    },
    {
      "epoch": 1.956434091163624,
      "grad_norm": 1.1829099655151367,
      "learning_rate": 9.607289473684211e-05,
      "loss": 0.595,
      "step": 34938
    },
    {
      "epoch": 1.956490088475753,
      "grad_norm": 1.2078707218170166,
      "learning_rate": 9.607263157894737e-05,
      "loss": 0.5205,
      "step": 34939
    },
    {
      "epoch": 1.956546085787882,
      "grad_norm": 1.5366955995559692,
      "learning_rate": 9.607236842105264e-05,
      "loss": 0.4761,
      "step": 34940
    },
    {
      "epoch": 1.956602083100011,
      "grad_norm": 1.944976806640625,
      "learning_rate": 9.60721052631579e-05,
      "loss": 0.5052,
      "step": 34941
    },
    {
      "epoch": 1.95665808041214,
      "grad_norm": 1.1572849750518799,
      "learning_rate": 9.607184210526316e-05,
      "loss": 0.4216,
      "step": 34942
    },
    {
      "epoch": 1.956714077724269,
      "grad_norm": 1.136043906211853,
      "learning_rate": 9.607157894736842e-05,
      "loss": 0.5111,
      "step": 34943
    },
    {
      "epoch": 1.956770075036398,
      "grad_norm": 1.2086769342422485,
      "learning_rate": 9.60713157894737e-05,
      "loss": 0.4202,
      "step": 34944
    },
    {
      "epoch": 1.9568260723485271,
      "grad_norm": 1.2215800285339355,
      "learning_rate": 9.607105263157895e-05,
      "loss": 0.3991,
      "step": 34945
    },
    {
      "epoch": 1.9568820696606561,
      "grad_norm": 1.1990710496902466,
      "learning_rate": 9.607078947368421e-05,
      "loss": 0.6209,
      "step": 34946
    },
    {
      "epoch": 1.9569380669727852,
      "grad_norm": 1.2651270627975464,
      "learning_rate": 9.607052631578947e-05,
      "loss": 0.5738,
      "step": 34947
    },
    {
      "epoch": 1.9569940642849142,
      "grad_norm": 1.4180033206939697,
      "learning_rate": 9.607026315789475e-05,
      "loss": 0.5229,
      "step": 34948
    },
    {
      "epoch": 1.9570500615970432,
      "grad_norm": 1.2521084547042847,
      "learning_rate": 9.607000000000001e-05,
      "loss": 0.449,
      "step": 34949
    },
    {
      "epoch": 1.9571060589091722,
      "grad_norm": 1.58376944065094,
      "learning_rate": 9.606973684210527e-05,
      "loss": 0.4505,
      "step": 34950
    },
    {
      "epoch": 1.9571620562213012,
      "grad_norm": 1.3174633979797363,
      "learning_rate": 9.606947368421053e-05,
      "loss": 0.4375,
      "step": 34951
    },
    {
      "epoch": 1.9572180535334303,
      "grad_norm": 5.596231937408447,
      "learning_rate": 9.606921052631579e-05,
      "loss": 0.5268,
      "step": 34952
    },
    {
      "epoch": 1.9572740508455593,
      "grad_norm": 1.0874046087265015,
      "learning_rate": 9.606894736842106e-05,
      "loss": 0.3645,
      "step": 34953
    },
    {
      "epoch": 1.9573300481576883,
      "grad_norm": 1.0293734073638916,
      "learning_rate": 9.606868421052632e-05,
      "loss": 0.4514,
      "step": 34954
    },
    {
      "epoch": 1.9573860454698173,
      "grad_norm": 1.1275116205215454,
      "learning_rate": 9.606842105263158e-05,
      "loss": 0.366,
      "step": 34955
    },
    {
      "epoch": 1.9574420427819463,
      "grad_norm": 1.2125002145767212,
      "learning_rate": 9.606815789473684e-05,
      "loss": 0.3208,
      "step": 34956
    },
    {
      "epoch": 1.9574980400940754,
      "grad_norm": 1.4821226596832275,
      "learning_rate": 9.606789473684211e-05,
      "loss": 0.4257,
      "step": 34957
    },
    {
      "epoch": 1.9575540374062044,
      "grad_norm": 1.2202311754226685,
      "learning_rate": 9.606763157894737e-05,
      "loss": 0.4591,
      "step": 34958
    },
    {
      "epoch": 1.9576100347183334,
      "grad_norm": 1.2566643953323364,
      "learning_rate": 9.606736842105264e-05,
      "loss": 0.4185,
      "step": 34959
    },
    {
      "epoch": 1.9576660320304624,
      "grad_norm": 1.2568758726119995,
      "learning_rate": 9.606710526315789e-05,
      "loss": 0.4621,
      "step": 34960
    },
    {
      "epoch": 1.9577220293425914,
      "grad_norm": 1.2447750568389893,
      "learning_rate": 9.606684210526316e-05,
      "loss": 0.3596,
      "step": 34961
    },
    {
      "epoch": 1.9577780266547204,
      "grad_norm": 1.2768527269363403,
      "learning_rate": 9.606657894736842e-05,
      "loss": 0.3643,
      "step": 34962
    },
    {
      "epoch": 1.9578340239668495,
      "grad_norm": 1.2024080753326416,
      "learning_rate": 9.60663157894737e-05,
      "loss": 0.3704,
      "step": 34963
    },
    {
      "epoch": 1.9578900212789785,
      "grad_norm": 1.0775671005249023,
      "learning_rate": 9.606605263157896e-05,
      "loss": 0.4106,
      "step": 34964
    },
    {
      "epoch": 1.9579460185911075,
      "grad_norm": 1.339693546295166,
      "learning_rate": 9.606578947368422e-05,
      "loss": 0.4383,
      "step": 34965
    },
    {
      "epoch": 1.9580020159032365,
      "grad_norm": 1.3563727140426636,
      "learning_rate": 9.606552631578948e-05,
      "loss": 0.6099,
      "step": 34966
    },
    {
      "epoch": 1.9580580132153655,
      "grad_norm": 1.1151005029678345,
      "learning_rate": 9.606526315789474e-05,
      "loss": 0.4951,
      "step": 34967
    },
    {
      "epoch": 1.9581140105274946,
      "grad_norm": 1.5692121982574463,
      "learning_rate": 9.606500000000001e-05,
      "loss": 0.5409,
      "step": 34968
    },
    {
      "epoch": 1.9581700078396236,
      "grad_norm": 1.467974305152893,
      "learning_rate": 9.606473684210527e-05,
      "loss": 0.5498,
      "step": 34969
    },
    {
      "epoch": 1.9582260051517526,
      "grad_norm": 1.652017593383789,
      "learning_rate": 9.606447368421053e-05,
      "loss": 0.6711,
      "step": 34970
    },
    {
      "epoch": 1.9582820024638816,
      "grad_norm": 1.2267835140228271,
      "learning_rate": 9.606421052631579e-05,
      "loss": 0.4458,
      "step": 34971
    },
    {
      "epoch": 1.9583379997760106,
      "grad_norm": 1.3156280517578125,
      "learning_rate": 9.606394736842106e-05,
      "loss": 0.5704,
      "step": 34972
    },
    {
      "epoch": 1.9583939970881397,
      "grad_norm": 1.1479923725128174,
      "learning_rate": 9.606368421052632e-05,
      "loss": 0.4117,
      "step": 34973
    },
    {
      "epoch": 1.9584499944002687,
      "grad_norm": 1.268945574760437,
      "learning_rate": 9.606342105263158e-05,
      "loss": 0.4164,
      "step": 34974
    },
    {
      "epoch": 1.9585059917123977,
      "grad_norm": 1.2137863636016846,
      "learning_rate": 9.606315789473684e-05,
      "loss": 0.5185,
      "step": 34975
    },
    {
      "epoch": 1.9585619890245267,
      "grad_norm": 1.1835424900054932,
      "learning_rate": 9.606289473684211e-05,
      "loss": 0.4456,
      "step": 34976
    },
    {
      "epoch": 1.9586179863366557,
      "grad_norm": 1.148664951324463,
      "learning_rate": 9.606263157894737e-05,
      "loss": 0.574,
      "step": 34977
    },
    {
      "epoch": 1.9586739836487848,
      "grad_norm": 1.1673240661621094,
      "learning_rate": 9.606236842105263e-05,
      "loss": 0.3927,
      "step": 34978
    },
    {
      "epoch": 1.9587299809609138,
      "grad_norm": 1.0870554447174072,
      "learning_rate": 9.60621052631579e-05,
      "loss": 0.488,
      "step": 34979
    },
    {
      "epoch": 1.9587859782730428,
      "grad_norm": 1.343524694442749,
      "learning_rate": 9.606184210526317e-05,
      "loss": 0.6659,
      "step": 34980
    },
    {
      "epoch": 1.9588419755851718,
      "grad_norm": 1.2094429731369019,
      "learning_rate": 9.606157894736843e-05,
      "loss": 0.3796,
      "step": 34981
    },
    {
      "epoch": 1.9588979728973008,
      "grad_norm": 1.2557742595672607,
      "learning_rate": 9.60613157894737e-05,
      "loss": 0.4631,
      "step": 34982
    },
    {
      "epoch": 1.9589539702094299,
      "grad_norm": 1.2624608278274536,
      "learning_rate": 9.606105263157895e-05,
      "loss": 0.3857,
      "step": 34983
    },
    {
      "epoch": 1.9590099675215589,
      "grad_norm": 1.361048698425293,
      "learning_rate": 9.606078947368422e-05,
      "loss": 0.4774,
      "step": 34984
    },
    {
      "epoch": 1.959065964833688,
      "grad_norm": 1.2811528444290161,
      "learning_rate": 9.606052631578948e-05,
      "loss": 0.5225,
      "step": 34985
    },
    {
      "epoch": 1.959121962145817,
      "grad_norm": 1.0644234418869019,
      "learning_rate": 9.606026315789474e-05,
      "loss": 0.4803,
      "step": 34986
    },
    {
      "epoch": 1.959177959457946,
      "grad_norm": 1.3712126016616821,
      "learning_rate": 9.606000000000001e-05,
      "loss": 0.5319,
      "step": 34987
    },
    {
      "epoch": 1.959233956770075,
      "grad_norm": 1.1343204975128174,
      "learning_rate": 9.605973684210526e-05,
      "loss": 0.3498,
      "step": 34988
    },
    {
      "epoch": 1.959289954082204,
      "grad_norm": 1.7177571058273315,
      "learning_rate": 9.605947368421053e-05,
      "loss": 0.603,
      "step": 34989
    },
    {
      "epoch": 1.959345951394333,
      "grad_norm": 1.4027554988861084,
      "learning_rate": 9.605921052631579e-05,
      "loss": 0.506,
      "step": 34990
    },
    {
      "epoch": 1.959401948706462,
      "grad_norm": 1.1770683526992798,
      "learning_rate": 9.605894736842106e-05,
      "loss": 0.491,
      "step": 34991
    },
    {
      "epoch": 1.959457946018591,
      "grad_norm": 1.2766592502593994,
      "learning_rate": 9.605868421052631e-05,
      "loss": 0.545,
      "step": 34992
    },
    {
      "epoch": 1.95951394333072,
      "grad_norm": 1.336032748222351,
      "learning_rate": 9.605842105263158e-05,
      "loss": 0.4822,
      "step": 34993
    },
    {
      "epoch": 1.959569940642849,
      "grad_norm": 1.2164918184280396,
      "learning_rate": 9.605815789473684e-05,
      "loss": 0.402,
      "step": 34994
    },
    {
      "epoch": 1.959625937954978,
      "grad_norm": 1.2803086042404175,
      "learning_rate": 9.605789473684212e-05,
      "loss": 0.4547,
      "step": 34995
    },
    {
      "epoch": 1.959681935267107,
      "grad_norm": 1.3348603248596191,
      "learning_rate": 9.605763157894738e-05,
      "loss": 0.467,
      "step": 34996
    },
    {
      "epoch": 1.9597379325792361,
      "grad_norm": 1.2480894327163696,
      "learning_rate": 9.605736842105264e-05,
      "loss": 0.4815,
      "step": 34997
    },
    {
      "epoch": 1.9597939298913651,
      "grad_norm": 1.3180657625198364,
      "learning_rate": 9.60571052631579e-05,
      "loss": 0.3948,
      "step": 34998
    },
    {
      "epoch": 1.9598499272034942,
      "grad_norm": 1.760079264640808,
      "learning_rate": 9.605684210526317e-05,
      "loss": 0.4587,
      "step": 34999
    },
    {
      "epoch": 1.9599059245156232,
      "grad_norm": 1.2017862796783447,
      "learning_rate": 9.605657894736843e-05,
      "loss": 0.384,
      "step": 35000
    },
    {
      "epoch": 1.9599619218277522,
      "grad_norm": 1.046779751777649,
      "learning_rate": 9.605631578947369e-05,
      "loss": 0.4045,
      "step": 35001
    },
    {
      "epoch": 1.9600179191398812,
      "grad_norm": 1.3169909715652466,
      "learning_rate": 9.605605263157895e-05,
      "loss": 0.5204,
      "step": 35002
    },
    {
      "epoch": 1.9600739164520102,
      "grad_norm": 1.1759002208709717,
      "learning_rate": 9.605578947368421e-05,
      "loss": 0.5466,
      "step": 35003
    },
    {
      "epoch": 1.9601299137641393,
      "grad_norm": 1.4461716413497925,
      "learning_rate": 9.605552631578948e-05,
      "loss": 0.511,
      "step": 35004
    },
    {
      "epoch": 1.9601859110762683,
      "grad_norm": 1.152579426765442,
      "learning_rate": 9.605526315789474e-05,
      "loss": 0.387,
      "step": 35005
    },
    {
      "epoch": 1.9602419083883973,
      "grad_norm": 1.0224753618240356,
      "learning_rate": 9.6055e-05,
      "loss": 0.5103,
      "step": 35006
    },
    {
      "epoch": 1.9602979057005263,
      "grad_norm": 1.0869301557540894,
      "learning_rate": 9.605473684210526e-05,
      "loss": 0.3547,
      "step": 35007
    },
    {
      "epoch": 1.9603539030126553,
      "grad_norm": 1.0646460056304932,
      "learning_rate": 9.605447368421053e-05,
      "loss": 0.3916,
      "step": 35008
    },
    {
      "epoch": 1.9604099003247843,
      "grad_norm": 1.681850552558899,
      "learning_rate": 9.60542105263158e-05,
      "loss": 0.3659,
      "step": 35009
    },
    {
      "epoch": 1.9604658976369134,
      "grad_norm": 1.224524974822998,
      "learning_rate": 9.605394736842105e-05,
      "loss": 0.4143,
      "step": 35010
    },
    {
      "epoch": 1.9605218949490424,
      "grad_norm": 2.4472427368164062,
      "learning_rate": 9.605368421052631e-05,
      "loss": 0.3781,
      "step": 35011
    },
    {
      "epoch": 1.9605778922611714,
      "grad_norm": 1.1017571687698364,
      "learning_rate": 9.605342105263159e-05,
      "loss": 0.4201,
      "step": 35012
    },
    {
      "epoch": 1.9606338895733004,
      "grad_norm": 1.1993101835250854,
      "learning_rate": 9.605315789473685e-05,
      "loss": 0.4085,
      "step": 35013
    },
    {
      "epoch": 1.9606898868854294,
      "grad_norm": 1.0888237953186035,
      "learning_rate": 9.605289473684212e-05,
      "loss": 0.4382,
      "step": 35014
    },
    {
      "epoch": 1.9607458841975585,
      "grad_norm": 1.1229703426361084,
      "learning_rate": 9.605263157894737e-05,
      "loss": 0.379,
      "step": 35015
    },
    {
      "epoch": 1.9608018815096875,
      "grad_norm": 1.3178211450576782,
      "learning_rate": 9.605236842105264e-05,
      "loss": 0.3898,
      "step": 35016
    },
    {
      "epoch": 1.9608578788218165,
      "grad_norm": 1.2040702104568481,
      "learning_rate": 9.60521052631579e-05,
      "loss": 0.5029,
      "step": 35017
    },
    {
      "epoch": 1.9609138761339455,
      "grad_norm": 1.3323034048080444,
      "learning_rate": 9.605184210526317e-05,
      "loss": 0.4765,
      "step": 35018
    },
    {
      "epoch": 1.9609698734460745,
      "grad_norm": 1.4703279733657837,
      "learning_rate": 9.605157894736843e-05,
      "loss": 0.5044,
      "step": 35019
    },
    {
      "epoch": 1.9610258707582036,
      "grad_norm": 1.125908374786377,
      "learning_rate": 9.605131578947368e-05,
      "loss": 0.4539,
      "step": 35020
    },
    {
      "epoch": 1.9610818680703326,
      "grad_norm": 1.385450839996338,
      "learning_rate": 9.605105263157895e-05,
      "loss": 0.3706,
      "step": 35021
    },
    {
      "epoch": 1.9611378653824616,
      "grad_norm": 1.18792724609375,
      "learning_rate": 9.605078947368421e-05,
      "loss": 0.3762,
      "step": 35022
    },
    {
      "epoch": 1.9611938626945906,
      "grad_norm": 1.3209445476531982,
      "learning_rate": 9.605052631578948e-05,
      "loss": 0.4242,
      "step": 35023
    },
    {
      "epoch": 1.9612498600067196,
      "grad_norm": 1.3063178062438965,
      "learning_rate": 9.605026315789474e-05,
      "loss": 0.429,
      "step": 35024
    },
    {
      "epoch": 1.9613058573188487,
      "grad_norm": 1.4127497673034668,
      "learning_rate": 9.605e-05,
      "loss": 0.4488,
      "step": 35025
    },
    {
      "epoch": 1.9613618546309777,
      "grad_norm": 1.2043349742889404,
      "learning_rate": 9.604973684210526e-05,
      "loss": 0.3887,
      "step": 35026
    },
    {
      "epoch": 1.9614178519431067,
      "grad_norm": 1.132593035697937,
      "learning_rate": 9.604947368421054e-05,
      "loss": 0.4677,
      "step": 35027
    },
    {
      "epoch": 1.9614738492552357,
      "grad_norm": 1.2407653331756592,
      "learning_rate": 9.60492105263158e-05,
      "loss": 0.3957,
      "step": 35028
    },
    {
      "epoch": 1.9615298465673647,
      "grad_norm": 1.4393200874328613,
      "learning_rate": 9.604894736842106e-05,
      "loss": 0.4343,
      "step": 35029
    },
    {
      "epoch": 1.9615858438794938,
      "grad_norm": 1.5096509456634521,
      "learning_rate": 9.604868421052632e-05,
      "loss": 0.3698,
      "step": 35030
    },
    {
      "epoch": 1.9616418411916228,
      "grad_norm": 1.0236238241195679,
      "learning_rate": 9.604842105263159e-05,
      "loss": 0.3599,
      "step": 35031
    },
    {
      "epoch": 1.9616978385037518,
      "grad_norm": 1.436355471611023,
      "learning_rate": 9.604815789473685e-05,
      "loss": 0.5627,
      "step": 35032
    },
    {
      "epoch": 1.9617538358158808,
      "grad_norm": 1.5737653970718384,
      "learning_rate": 9.604789473684211e-05,
      "loss": 0.5378,
      "step": 35033
    },
    {
      "epoch": 1.9618098331280098,
      "grad_norm": 1.166304588317871,
      "learning_rate": 9.604763157894737e-05,
      "loss": 0.489,
      "step": 35034
    },
    {
      "epoch": 1.9618658304401388,
      "grad_norm": 1.2401537895202637,
      "learning_rate": 9.604736842105264e-05,
      "loss": 0.4317,
      "step": 35035
    },
    {
      "epoch": 1.9619218277522679,
      "grad_norm": 1.3277195692062378,
      "learning_rate": 9.60471052631579e-05,
      "loss": 0.357,
      "step": 35036
    },
    {
      "epoch": 1.9619778250643969,
      "grad_norm": 3.106485605239868,
      "learning_rate": 9.604684210526316e-05,
      "loss": 0.5722,
      "step": 35037
    },
    {
      "epoch": 1.962033822376526,
      "grad_norm": 1.1998252868652344,
      "learning_rate": 9.604657894736842e-05,
      "loss": 0.4039,
      "step": 35038
    },
    {
      "epoch": 1.962089819688655,
      "grad_norm": 1.1706624031066895,
      "learning_rate": 9.604631578947368e-05,
      "loss": 0.4741,
      "step": 35039
    },
    {
      "epoch": 1.962145817000784,
      "grad_norm": 1.3105096817016602,
      "learning_rate": 9.604605263157895e-05,
      "loss": 0.6032,
      "step": 35040
    },
    {
      "epoch": 1.962201814312913,
      "grad_norm": 1.3602538108825684,
      "learning_rate": 9.604578947368421e-05,
      "loss": 0.4094,
      "step": 35041
    },
    {
      "epoch": 1.962257811625042,
      "grad_norm": 1.3215928077697754,
      "learning_rate": 9.604552631578949e-05,
      "loss": 0.4288,
      "step": 35042
    },
    {
      "epoch": 1.962313808937171,
      "grad_norm": 1.5521959066390991,
      "learning_rate": 9.604526315789473e-05,
      "loss": 0.5863,
      "step": 35043
    },
    {
      "epoch": 1.9623698062493,
      "grad_norm": 1.3338600397109985,
      "learning_rate": 9.6045e-05,
      "loss": 0.4718,
      "step": 35044
    },
    {
      "epoch": 1.962425803561429,
      "grad_norm": 1.1608291864395142,
      "learning_rate": 9.604473684210527e-05,
      "loss": 0.4493,
      "step": 35045
    },
    {
      "epoch": 1.962481800873558,
      "grad_norm": 1.3158161640167236,
      "learning_rate": 9.604447368421054e-05,
      "loss": 0.4648,
      "step": 35046
    },
    {
      "epoch": 1.962537798185687,
      "grad_norm": 1.10756254196167,
      "learning_rate": 9.604421052631579e-05,
      "loss": 0.4545,
      "step": 35047
    },
    {
      "epoch": 1.962593795497816,
      "grad_norm": 1.2917457818984985,
      "learning_rate": 9.604394736842106e-05,
      "loss": 0.454,
      "step": 35048
    },
    {
      "epoch": 1.9626497928099451,
      "grad_norm": 1.4271832704544067,
      "learning_rate": 9.604368421052632e-05,
      "loss": 0.4267,
      "step": 35049
    },
    {
      "epoch": 1.9627057901220741,
      "grad_norm": 1.290459156036377,
      "learning_rate": 9.604342105263159e-05,
      "loss": 0.3639,
      "step": 35050
    },
    {
      "epoch": 1.9627617874342032,
      "grad_norm": 1.2243764400482178,
      "learning_rate": 9.604315789473685e-05,
      "loss": 0.3729,
      "step": 35051
    },
    {
      "epoch": 1.9628177847463322,
      "grad_norm": 1.481799840927124,
      "learning_rate": 9.604289473684211e-05,
      "loss": 0.507,
      "step": 35052
    },
    {
      "epoch": 1.9628737820584612,
      "grad_norm": 1.734131932258606,
      "learning_rate": 9.604263157894737e-05,
      "loss": 0.5044,
      "step": 35053
    },
    {
      "epoch": 1.9629297793705902,
      "grad_norm": 1.3061918020248413,
      "learning_rate": 9.604236842105263e-05,
      "loss": 0.5672,
      "step": 35054
    },
    {
      "epoch": 1.9629857766827192,
      "grad_norm": 1.0983169078826904,
      "learning_rate": 9.60421052631579e-05,
      "loss": 0.3827,
      "step": 35055
    },
    {
      "epoch": 1.9630417739948482,
      "grad_norm": 1.2631009817123413,
      "learning_rate": 9.604184210526316e-05,
      "loss": 0.422,
      "step": 35056
    },
    {
      "epoch": 1.9630977713069773,
      "grad_norm": 1.4536410570144653,
      "learning_rate": 9.604157894736842e-05,
      "loss": 0.5509,
      "step": 35057
    },
    {
      "epoch": 1.9631537686191063,
      "grad_norm": 1.2138433456420898,
      "learning_rate": 9.604131578947368e-05,
      "loss": 0.5279,
      "step": 35058
    },
    {
      "epoch": 1.9632097659312353,
      "grad_norm": 1.2809592485427856,
      "learning_rate": 9.604105263157896e-05,
      "loss": 0.4929,
      "step": 35059
    },
    {
      "epoch": 1.9632657632433643,
      "grad_norm": 1.2691136598587036,
      "learning_rate": 9.604078947368422e-05,
      "loss": 0.4245,
      "step": 35060
    },
    {
      "epoch": 1.9633217605554933,
      "grad_norm": 1.5423649549484253,
      "learning_rate": 9.604052631578948e-05,
      "loss": 0.5136,
      "step": 35061
    },
    {
      "epoch": 1.9633777578676224,
      "grad_norm": 1.168105959892273,
      "learning_rate": 9.604026315789474e-05,
      "loss": 0.4363,
      "step": 35062
    },
    {
      "epoch": 1.9634337551797514,
      "grad_norm": 1.3851975202560425,
      "learning_rate": 9.604000000000001e-05,
      "loss": 0.4109,
      "step": 35063
    },
    {
      "epoch": 1.9634897524918804,
      "grad_norm": 1.7474687099456787,
      "learning_rate": 9.603973684210527e-05,
      "loss": 0.4735,
      "step": 35064
    },
    {
      "epoch": 1.9635457498040094,
      "grad_norm": 1.3522891998291016,
      "learning_rate": 9.603947368421053e-05,
      "loss": 0.4313,
      "step": 35065
    },
    {
      "epoch": 1.9636017471161384,
      "grad_norm": 1.3846291303634644,
      "learning_rate": 9.603921052631579e-05,
      "loss": 0.4365,
      "step": 35066
    },
    {
      "epoch": 1.9636577444282675,
      "grad_norm": 1.3774664402008057,
      "learning_rate": 9.603894736842106e-05,
      "loss": 0.5081,
      "step": 35067
    },
    {
      "epoch": 1.9637137417403965,
      "grad_norm": 1.0783714056015015,
      "learning_rate": 9.603868421052632e-05,
      "loss": 0.3773,
      "step": 35068
    },
    {
      "epoch": 1.9637697390525255,
      "grad_norm": 1.2375034093856812,
      "learning_rate": 9.60384210526316e-05,
      "loss": 0.431,
      "step": 35069
    },
    {
      "epoch": 1.9638257363646545,
      "grad_norm": 1.3854904174804688,
      "learning_rate": 9.603815789473684e-05,
      "loss": 0.5805,
      "step": 35070
    },
    {
      "epoch": 1.9638817336767835,
      "grad_norm": 1.1335335969924927,
      "learning_rate": 9.60378947368421e-05,
      "loss": 0.4852,
      "step": 35071
    },
    {
      "epoch": 1.9639377309889126,
      "grad_norm": 1.6117701530456543,
      "learning_rate": 9.603763157894737e-05,
      "loss": 0.4522,
      "step": 35072
    },
    {
      "epoch": 1.9639937283010416,
      "grad_norm": 1.0952625274658203,
      "learning_rate": 9.603736842105263e-05,
      "loss": 0.4284,
      "step": 35073
    },
    {
      "epoch": 1.9640497256131706,
      "grad_norm": 1.6762878894805908,
      "learning_rate": 9.60371052631579e-05,
      "loss": 0.4316,
      "step": 35074
    },
    {
      "epoch": 1.9641057229252996,
      "grad_norm": 1.3426074981689453,
      "learning_rate": 9.603684210526315e-05,
      "loss": 0.4859,
      "step": 35075
    },
    {
      "epoch": 1.9641617202374286,
      "grad_norm": 1.27347731590271,
      "learning_rate": 9.603657894736843e-05,
      "loss": 0.5075,
      "step": 35076
    },
    {
      "epoch": 1.9642177175495577,
      "grad_norm": 1.244230031967163,
      "learning_rate": 9.603631578947369e-05,
      "loss": 0.4861,
      "step": 35077
    },
    {
      "epoch": 1.9642737148616867,
      "grad_norm": 1.130865216255188,
      "learning_rate": 9.603605263157896e-05,
      "loss": 0.5262,
      "step": 35078
    },
    {
      "epoch": 1.9643297121738157,
      "grad_norm": 1.3578885793685913,
      "learning_rate": 9.603578947368422e-05,
      "loss": 0.437,
      "step": 35079
    },
    {
      "epoch": 1.9643857094859447,
      "grad_norm": 1.1074336767196655,
      "learning_rate": 9.603552631578948e-05,
      "loss": 0.4588,
      "step": 35080
    },
    {
      "epoch": 1.9644417067980737,
      "grad_norm": 1.6396186351776123,
      "learning_rate": 9.603526315789474e-05,
      "loss": 0.3646,
      "step": 35081
    },
    {
      "epoch": 1.9644977041102027,
      "grad_norm": 0.9948999881744385,
      "learning_rate": 9.603500000000001e-05,
      "loss": 0.3672,
      "step": 35082
    },
    {
      "epoch": 1.9645537014223318,
      "grad_norm": 1.91913902759552,
      "learning_rate": 9.603473684210527e-05,
      "loss": 0.5184,
      "step": 35083
    },
    {
      "epoch": 1.9646096987344608,
      "grad_norm": 2.5451409816741943,
      "learning_rate": 9.603447368421053e-05,
      "loss": 0.4887,
      "step": 35084
    },
    {
      "epoch": 1.9646656960465898,
      "grad_norm": 1.2871602773666382,
      "learning_rate": 9.603421052631579e-05,
      "loss": 0.4339,
      "step": 35085
    },
    {
      "epoch": 1.9647216933587188,
      "grad_norm": 1.2026474475860596,
      "learning_rate": 9.603394736842106e-05,
      "loss": 0.3858,
      "step": 35086
    },
    {
      "epoch": 1.9647776906708478,
      "grad_norm": 1.516147494316101,
      "learning_rate": 9.603368421052632e-05,
      "loss": 0.4047,
      "step": 35087
    },
    {
      "epoch": 1.9648336879829769,
      "grad_norm": 1.432829737663269,
      "learning_rate": 9.603342105263158e-05,
      "loss": 0.4611,
      "step": 35088
    },
    {
      "epoch": 1.9648896852951059,
      "grad_norm": 1.2240043878555298,
      "learning_rate": 9.603315789473684e-05,
      "loss": 0.402,
      "step": 35089
    },
    {
      "epoch": 1.964945682607235,
      "grad_norm": 1.1273319721221924,
      "learning_rate": 9.60328947368421e-05,
      "loss": 0.4584,
      "step": 35090
    },
    {
      "epoch": 1.965001679919364,
      "grad_norm": 1.4825795888900757,
      "learning_rate": 9.603263157894738e-05,
      "loss": 0.6356,
      "step": 35091
    },
    {
      "epoch": 1.965057677231493,
      "grad_norm": 1.0835208892822266,
      "learning_rate": 9.603236842105264e-05,
      "loss": 0.4487,
      "step": 35092
    },
    {
      "epoch": 1.965113674543622,
      "grad_norm": 1.8915773630142212,
      "learning_rate": 9.60321052631579e-05,
      "loss": 0.5894,
      "step": 35093
    },
    {
      "epoch": 1.965169671855751,
      "grad_norm": 1.3294094800949097,
      "learning_rate": 9.603184210526316e-05,
      "loss": 0.452,
      "step": 35094
    },
    {
      "epoch": 1.96522566916788,
      "grad_norm": 1.1674472093582153,
      "learning_rate": 9.603157894736843e-05,
      "loss": 0.3874,
      "step": 35095
    },
    {
      "epoch": 1.965281666480009,
      "grad_norm": 1.1369107961654663,
      "learning_rate": 9.603131578947369e-05,
      "loss": 0.5593,
      "step": 35096
    },
    {
      "epoch": 1.965337663792138,
      "grad_norm": 1.2223368883132935,
      "learning_rate": 9.603105263157896e-05,
      "loss": 0.4747,
      "step": 35097
    },
    {
      "epoch": 1.965393661104267,
      "grad_norm": 1.2368556261062622,
      "learning_rate": 9.603078947368421e-05,
      "loss": 0.5067,
      "step": 35098
    },
    {
      "epoch": 1.965449658416396,
      "grad_norm": 1.1085302829742432,
      "learning_rate": 9.603052631578948e-05,
      "loss": 0.4162,
      "step": 35099
    },
    {
      "epoch": 1.965505655728525,
      "grad_norm": 1.0740025043487549,
      "learning_rate": 9.603026315789474e-05,
      "loss": 0.3344,
      "step": 35100
    },
    {
      "epoch": 1.965561653040654,
      "grad_norm": 1.219307541847229,
      "learning_rate": 9.603000000000001e-05,
      "loss": 0.536,
      "step": 35101
    },
    {
      "epoch": 1.9656176503527831,
      "grad_norm": 1.2384260892868042,
      "learning_rate": 9.602973684210526e-05,
      "loss": 0.361,
      "step": 35102
    },
    {
      "epoch": 1.9656736476649121,
      "grad_norm": 1.2756088972091675,
      "learning_rate": 9.602947368421053e-05,
      "loss": 0.4436,
      "step": 35103
    },
    {
      "epoch": 1.9657296449770412,
      "grad_norm": 1.2073445320129395,
      "learning_rate": 9.602921052631579e-05,
      "loss": 0.3952,
      "step": 35104
    },
    {
      "epoch": 1.9657856422891702,
      "grad_norm": 1.1042813062667847,
      "learning_rate": 9.602894736842107e-05,
      "loss": 0.3114,
      "step": 35105
    },
    {
      "epoch": 1.9658416396012992,
      "grad_norm": 1.1160542964935303,
      "learning_rate": 9.602868421052633e-05,
      "loss": 0.4083,
      "step": 35106
    },
    {
      "epoch": 1.9658976369134282,
      "grad_norm": 1.1463243961334229,
      "learning_rate": 9.602842105263157e-05,
      "loss": 0.4085,
      "step": 35107
    },
    {
      "epoch": 1.9659536342255572,
      "grad_norm": 1.1548633575439453,
      "learning_rate": 9.602815789473685e-05,
      "loss": 0.3938,
      "step": 35108
    },
    {
      "epoch": 1.9660096315376863,
      "grad_norm": 1.4418126344680786,
      "learning_rate": 9.60278947368421e-05,
      "loss": 0.4592,
      "step": 35109
    },
    {
      "epoch": 1.9660656288498153,
      "grad_norm": 1.1367462873458862,
      "learning_rate": 9.602763157894738e-05,
      "loss": 0.3902,
      "step": 35110
    },
    {
      "epoch": 1.9661216261619443,
      "grad_norm": 1.1419309377670288,
      "learning_rate": 9.602736842105264e-05,
      "loss": 0.4165,
      "step": 35111
    },
    {
      "epoch": 1.9661776234740733,
      "grad_norm": 1.337330937385559,
      "learning_rate": 9.60271052631579e-05,
      "loss": 0.4509,
      "step": 35112
    },
    {
      "epoch": 1.9662336207862023,
      "grad_norm": 1.4922298192977905,
      "learning_rate": 9.602684210526316e-05,
      "loss": 0.3353,
      "step": 35113
    },
    {
      "epoch": 1.9662896180983314,
      "grad_norm": 1.513649344444275,
      "learning_rate": 9.602657894736843e-05,
      "loss": 0.5482,
      "step": 35114
    },
    {
      "epoch": 1.9663456154104604,
      "grad_norm": 1.0241405963897705,
      "learning_rate": 9.602631578947369e-05,
      "loss": 0.4572,
      "step": 35115
    },
    {
      "epoch": 1.9664016127225894,
      "grad_norm": 1.1362054347991943,
      "learning_rate": 9.602605263157895e-05,
      "loss": 0.4123,
      "step": 35116
    },
    {
      "epoch": 1.9664576100347184,
      "grad_norm": 1.225488543510437,
      "learning_rate": 9.602578947368421e-05,
      "loss": 0.5372,
      "step": 35117
    },
    {
      "epoch": 1.9665136073468474,
      "grad_norm": 1.6217612028121948,
      "learning_rate": 9.602552631578948e-05,
      "loss": 0.4319,
      "step": 35118
    },
    {
      "epoch": 1.9665696046589765,
      "grad_norm": 1.2428851127624512,
      "learning_rate": 9.602526315789474e-05,
      "loss": 0.4407,
      "step": 35119
    },
    {
      "epoch": 1.9666256019711055,
      "grad_norm": 1.1516507863998413,
      "learning_rate": 9.6025e-05,
      "loss": 0.5326,
      "step": 35120
    },
    {
      "epoch": 1.9666815992832345,
      "grad_norm": 1.1538132429122925,
      "learning_rate": 9.602473684210526e-05,
      "loss": 0.3408,
      "step": 35121
    },
    {
      "epoch": 1.9667375965953635,
      "grad_norm": 1.265317678451538,
      "learning_rate": 9.602447368421054e-05,
      "loss": 0.4446,
      "step": 35122
    },
    {
      "epoch": 1.9667935939074925,
      "grad_norm": 1.145520806312561,
      "learning_rate": 9.60242105263158e-05,
      "loss": 0.3781,
      "step": 35123
    },
    {
      "epoch": 1.9668495912196216,
      "grad_norm": 1.13011634349823,
      "learning_rate": 9.602394736842106e-05,
      "loss": 0.4152,
      "step": 35124
    },
    {
      "epoch": 1.9669055885317506,
      "grad_norm": 1.4513499736785889,
      "learning_rate": 9.602368421052632e-05,
      "loss": 0.5631,
      "step": 35125
    },
    {
      "epoch": 1.9669615858438796,
      "grad_norm": 1.3004796504974365,
      "learning_rate": 9.602342105263157e-05,
      "loss": 0.4229,
      "step": 35126
    },
    {
      "epoch": 1.9670175831560086,
      "grad_norm": 1.1182887554168701,
      "learning_rate": 9.602315789473685e-05,
      "loss": 0.3526,
      "step": 35127
    },
    {
      "epoch": 1.9670735804681376,
      "grad_norm": 1.6951583623886108,
      "learning_rate": 9.602289473684211e-05,
      "loss": 0.3822,
      "step": 35128
    },
    {
      "epoch": 1.9671295777802666,
      "grad_norm": 1.7151211500167847,
      "learning_rate": 9.602263157894738e-05,
      "loss": 0.5269,
      "step": 35129
    },
    {
      "epoch": 1.9671855750923957,
      "grad_norm": 1.3853042125701904,
      "learning_rate": 9.602236842105263e-05,
      "loss": 0.4884,
      "step": 35130
    },
    {
      "epoch": 1.9672415724045247,
      "grad_norm": 1.8636727333068848,
      "learning_rate": 9.60221052631579e-05,
      "loss": 0.4545,
      "step": 35131
    },
    {
      "epoch": 1.9672975697166537,
      "grad_norm": 1.3396034240722656,
      "learning_rate": 9.602184210526316e-05,
      "loss": 0.5325,
      "step": 35132
    },
    {
      "epoch": 1.9673535670287827,
      "grad_norm": 5.555102348327637,
      "learning_rate": 9.602157894736843e-05,
      "loss": 0.4265,
      "step": 35133
    },
    {
      "epoch": 1.9674095643409117,
      "grad_norm": 1.3711774349212646,
      "learning_rate": 9.60213157894737e-05,
      "loss": 0.3228,
      "step": 35134
    },
    {
      "epoch": 1.9674655616530408,
      "grad_norm": 1.1278173923492432,
      "learning_rate": 9.602105263157895e-05,
      "loss": 0.4273,
      "step": 35135
    },
    {
      "epoch": 1.9675215589651698,
      "grad_norm": 1.406720519065857,
      "learning_rate": 9.602078947368421e-05,
      "loss": 0.507,
      "step": 35136
    },
    {
      "epoch": 1.9675775562772988,
      "grad_norm": 1.2888717651367188,
      "learning_rate": 9.602052631578949e-05,
      "loss": 0.5419,
      "step": 35137
    },
    {
      "epoch": 1.9676335535894278,
      "grad_norm": 2.0798962116241455,
      "learning_rate": 9.602026315789475e-05,
      "loss": 0.5126,
      "step": 35138
    },
    {
      "epoch": 1.9676895509015568,
      "grad_norm": 1.1820476055145264,
      "learning_rate": 9.602e-05,
      "loss": 0.5528,
      "step": 35139
    },
    {
      "epoch": 1.9677455482136859,
      "grad_norm": 1.4389581680297852,
      "learning_rate": 9.601973684210527e-05,
      "loss": 0.4277,
      "step": 35140
    },
    {
      "epoch": 1.9678015455258149,
      "grad_norm": 1.1537623405456543,
      "learning_rate": 9.601947368421053e-05,
      "loss": 0.4237,
      "step": 35141
    },
    {
      "epoch": 1.967857542837944,
      "grad_norm": 1.3462848663330078,
      "learning_rate": 9.60192105263158e-05,
      "loss": 0.4196,
      "step": 35142
    },
    {
      "epoch": 1.967913540150073,
      "grad_norm": 1.516627550125122,
      "learning_rate": 9.601894736842106e-05,
      "loss": 0.5463,
      "step": 35143
    },
    {
      "epoch": 1.967969537462202,
      "grad_norm": 1.5292978286743164,
      "learning_rate": 9.601868421052632e-05,
      "loss": 0.4922,
      "step": 35144
    },
    {
      "epoch": 1.968025534774331,
      "grad_norm": 1.6563143730163574,
      "learning_rate": 9.601842105263158e-05,
      "loss": 0.6334,
      "step": 35145
    },
    {
      "epoch": 1.96808153208646,
      "grad_norm": 1.3477717638015747,
      "learning_rate": 9.601815789473685e-05,
      "loss": 0.3483,
      "step": 35146
    },
    {
      "epoch": 1.968137529398589,
      "grad_norm": 1.1159671545028687,
      "learning_rate": 9.601789473684211e-05,
      "loss": 0.4548,
      "step": 35147
    },
    {
      "epoch": 1.968193526710718,
      "grad_norm": 1.2092177867889404,
      "learning_rate": 9.601763157894737e-05,
      "loss": 0.461,
      "step": 35148
    },
    {
      "epoch": 1.968249524022847,
      "grad_norm": 1.286135196685791,
      "learning_rate": 9.601736842105263e-05,
      "loss": 0.448,
      "step": 35149
    },
    {
      "epoch": 1.968305521334976,
      "grad_norm": 1.833780288696289,
      "learning_rate": 9.60171052631579e-05,
      "loss": 0.5254,
      "step": 35150
    },
    {
      "epoch": 1.968361518647105,
      "grad_norm": 1.0451806783676147,
      "learning_rate": 9.601684210526316e-05,
      "loss": 0.4319,
      "step": 35151
    },
    {
      "epoch": 1.968417515959234,
      "grad_norm": 1.2672401666641235,
      "learning_rate": 9.601657894736844e-05,
      "loss": 0.4771,
      "step": 35152
    },
    {
      "epoch": 1.968473513271363,
      "grad_norm": 1.365888237953186,
      "learning_rate": 9.601631578947368e-05,
      "loss": 0.3488,
      "step": 35153
    },
    {
      "epoch": 1.9685295105834921,
      "grad_norm": 1.1900831460952759,
      "learning_rate": 9.601605263157896e-05,
      "loss": 0.3894,
      "step": 35154
    },
    {
      "epoch": 1.9685855078956211,
      "grad_norm": 1.4087401628494263,
      "learning_rate": 9.601578947368422e-05,
      "loss": 0.3823,
      "step": 35155
    },
    {
      "epoch": 1.9686415052077502,
      "grad_norm": 1.3083221912384033,
      "learning_rate": 9.601552631578949e-05,
      "loss": 0.5053,
      "step": 35156
    },
    {
      "epoch": 1.9686975025198792,
      "grad_norm": 1.2480884790420532,
      "learning_rate": 9.601526315789473e-05,
      "loss": 0.4257,
      "step": 35157
    },
    {
      "epoch": 1.9687534998320082,
      "grad_norm": 1.2438790798187256,
      "learning_rate": 9.6015e-05,
      "loss": 0.3915,
      "step": 35158
    },
    {
      "epoch": 1.9688094971441372,
      "grad_norm": 1.1964640617370605,
      "learning_rate": 9.601473684210527e-05,
      "loss": 0.3754,
      "step": 35159
    },
    {
      "epoch": 1.9688654944562662,
      "grad_norm": 1.8770874738693237,
      "learning_rate": 9.601447368421053e-05,
      "loss": 0.3471,
      "step": 35160
    },
    {
      "epoch": 1.9689214917683953,
      "grad_norm": 1.4485785961151123,
      "learning_rate": 9.60142105263158e-05,
      "loss": 0.6088,
      "step": 35161
    },
    {
      "epoch": 1.9689774890805243,
      "grad_norm": 1.4133427143096924,
      "learning_rate": 9.601394736842105e-05,
      "loss": 0.5865,
      "step": 35162
    },
    {
      "epoch": 1.9690334863926533,
      "grad_norm": 1.1782273054122925,
      "learning_rate": 9.601368421052632e-05,
      "loss": 0.4794,
      "step": 35163
    },
    {
      "epoch": 1.9690894837047823,
      "grad_norm": 1.2187813520431519,
      "learning_rate": 9.601342105263158e-05,
      "loss": 0.3984,
      "step": 35164
    },
    {
      "epoch": 1.9691454810169113,
      "grad_norm": 1.7584083080291748,
      "learning_rate": 9.601315789473685e-05,
      "loss": 0.496,
      "step": 35165
    },
    {
      "epoch": 1.9692014783290404,
      "grad_norm": 1.381347894668579,
      "learning_rate": 9.601289473684211e-05,
      "loss": 0.4182,
      "step": 35166
    },
    {
      "epoch": 1.9692574756411694,
      "grad_norm": 1.3327884674072266,
      "learning_rate": 9.601263157894737e-05,
      "loss": 0.5346,
      "step": 35167
    },
    {
      "epoch": 1.9693134729532984,
      "grad_norm": 1.390777349472046,
      "learning_rate": 9.601236842105263e-05,
      "loss": 0.4846,
      "step": 35168
    },
    {
      "epoch": 1.9693694702654274,
      "grad_norm": 1.268707275390625,
      "learning_rate": 9.60121052631579e-05,
      "loss": 0.3648,
      "step": 35169
    },
    {
      "epoch": 1.9694254675775564,
      "grad_norm": 1.09173583984375,
      "learning_rate": 9.601184210526317e-05,
      "loss": 0.368,
      "step": 35170
    },
    {
      "epoch": 1.9694814648896855,
      "grad_norm": 1.4485660791397095,
      "learning_rate": 9.601157894736843e-05,
      "loss": 0.462,
      "step": 35171
    },
    {
      "epoch": 1.9695374622018145,
      "grad_norm": 1.302014708518982,
      "learning_rate": 9.601131578947369e-05,
      "loss": 0.4345,
      "step": 35172
    },
    {
      "epoch": 1.9695934595139435,
      "grad_norm": 1.0750302076339722,
      "learning_rate": 9.601105263157896e-05,
      "loss": 0.3762,
      "step": 35173
    },
    {
      "epoch": 1.9696494568260725,
      "grad_norm": 1.3170346021652222,
      "learning_rate": 9.601078947368422e-05,
      "loss": 0.4624,
      "step": 35174
    },
    {
      "epoch": 1.9697054541382015,
      "grad_norm": 1.2083524465560913,
      "learning_rate": 9.601052631578948e-05,
      "loss": 0.4129,
      "step": 35175
    },
    {
      "epoch": 1.9697614514503305,
      "grad_norm": 1.0780861377716064,
      "learning_rate": 9.601026315789474e-05,
      "loss": 0.3793,
      "step": 35176
    },
    {
      "epoch": 1.9698174487624596,
      "grad_norm": 1.239497184753418,
      "learning_rate": 9.601e-05,
      "loss": 0.5152,
      "step": 35177
    },
    {
      "epoch": 1.9698734460745886,
      "grad_norm": 1.0269932746887207,
      "learning_rate": 9.600973684210527e-05,
      "loss": 0.3444,
      "step": 35178
    },
    {
      "epoch": 1.9699294433867174,
      "grad_norm": 1.3105684518814087,
      "learning_rate": 9.600947368421053e-05,
      "loss": 0.4618,
      "step": 35179
    },
    {
      "epoch": 1.9699854406988464,
      "grad_norm": 1.0756056308746338,
      "learning_rate": 9.600921052631579e-05,
      "loss": 0.3688,
      "step": 35180
    },
    {
      "epoch": 1.9700414380109754,
      "grad_norm": 1.748188853263855,
      "learning_rate": 9.600894736842105e-05,
      "loss": 0.6373,
      "step": 35181
    },
    {
      "epoch": 1.9700974353231044,
      "grad_norm": 1.3499687910079956,
      "learning_rate": 9.600868421052632e-05,
      "loss": 0.5859,
      "step": 35182
    },
    {
      "epoch": 1.9701534326352335,
      "grad_norm": 1.5669392347335815,
      "learning_rate": 9.600842105263158e-05,
      "loss": 0.4076,
      "step": 35183
    },
    {
      "epoch": 1.9702094299473625,
      "grad_norm": 1.1693569421768188,
      "learning_rate": 9.600815789473686e-05,
      "loss": 0.5019,
      "step": 35184
    },
    {
      "epoch": 1.9702654272594915,
      "grad_norm": 1.2108407020568848,
      "learning_rate": 9.60078947368421e-05,
      "loss": 0.4216,
      "step": 35185
    },
    {
      "epoch": 1.9703214245716205,
      "grad_norm": 1.2133065462112427,
      "learning_rate": 9.600763157894738e-05,
      "loss": 0.3984,
      "step": 35186
    },
    {
      "epoch": 1.9703774218837495,
      "grad_norm": 3.837827205657959,
      "learning_rate": 9.600736842105264e-05,
      "loss": 0.3495,
      "step": 35187
    },
    {
      "epoch": 1.9704334191958786,
      "grad_norm": 1.2708792686462402,
      "learning_rate": 9.600710526315791e-05,
      "loss": 0.4606,
      "step": 35188
    },
    {
      "epoch": 1.9704894165080076,
      "grad_norm": 1.0991908311843872,
      "learning_rate": 9.600684210526317e-05,
      "loss": 0.3249,
      "step": 35189
    },
    {
      "epoch": 1.9705454138201366,
      "grad_norm": 1.2925313711166382,
      "learning_rate": 9.600657894736843e-05,
      "loss": 0.4992,
      "step": 35190
    },
    {
      "epoch": 1.9706014111322656,
      "grad_norm": 1.2970471382141113,
      "learning_rate": 9.600631578947369e-05,
      "loss": 0.5117,
      "step": 35191
    },
    {
      "epoch": 1.9706574084443946,
      "grad_norm": 1.265629529953003,
      "learning_rate": 9.600605263157895e-05,
      "loss": 0.4824,
      "step": 35192
    },
    {
      "epoch": 1.9707134057565237,
      "grad_norm": 1.2016805410385132,
      "learning_rate": 9.600578947368422e-05,
      "loss": 0.4069,
      "step": 35193
    },
    {
      "epoch": 1.9707694030686527,
      "grad_norm": 1.1600769758224487,
      "learning_rate": 9.600552631578947e-05,
      "loss": 0.3889,
      "step": 35194
    },
    {
      "epoch": 1.9708254003807817,
      "grad_norm": 2.4386913776397705,
      "learning_rate": 9.600526315789474e-05,
      "loss": 0.3638,
      "step": 35195
    },
    {
      "epoch": 1.9708813976929107,
      "grad_norm": 1.1993727684020996,
      "learning_rate": 9.6005e-05,
      "loss": 0.5336,
      "step": 35196
    },
    {
      "epoch": 1.9709373950050397,
      "grad_norm": 1.071684718132019,
      "learning_rate": 9.600473684210527e-05,
      "loss": 0.4054,
      "step": 35197
    },
    {
      "epoch": 1.9709933923171687,
      "grad_norm": 1.1378376483917236,
      "learning_rate": 9.600447368421053e-05,
      "loss": 0.4152,
      "step": 35198
    },
    {
      "epoch": 1.9710493896292978,
      "grad_norm": 1.4601490497589111,
      "learning_rate": 9.600421052631579e-05,
      "loss": 0.5032,
      "step": 35199
    },
    {
      "epoch": 1.9711053869414268,
      "grad_norm": 1.1496762037277222,
      "learning_rate": 9.600394736842105e-05,
      "loss": 0.3943,
      "step": 35200
    },
    {
      "epoch": 1.9711613842535558,
      "grad_norm": 1.4457305669784546,
      "learning_rate": 9.600368421052633e-05,
      "loss": 0.4355,
      "step": 35201
    },
    {
      "epoch": 1.9712173815656848,
      "grad_norm": 1.0772647857666016,
      "learning_rate": 9.600342105263159e-05,
      "loss": 0.4467,
      "step": 35202
    },
    {
      "epoch": 1.9712733788778138,
      "grad_norm": 1.120788812637329,
      "learning_rate": 9.600315789473685e-05,
      "loss": 0.3564,
      "step": 35203
    },
    {
      "epoch": 1.9713293761899429,
      "grad_norm": 1.5987862348556519,
      "learning_rate": 9.60028947368421e-05,
      "loss": 0.4844,
      "step": 35204
    },
    {
      "epoch": 1.9713853735020719,
      "grad_norm": 1.2389825582504272,
      "learning_rate": 9.600263157894738e-05,
      "loss": 0.5305,
      "step": 35205
    },
    {
      "epoch": 1.971441370814201,
      "grad_norm": 1.3866772651672363,
      "learning_rate": 9.600236842105264e-05,
      "loss": 0.4768,
      "step": 35206
    },
    {
      "epoch": 1.97149736812633,
      "grad_norm": 1.2331037521362305,
      "learning_rate": 9.600210526315791e-05,
      "loss": 0.5245,
      "step": 35207
    },
    {
      "epoch": 1.971553365438459,
      "grad_norm": 1.3958488702774048,
      "learning_rate": 9.600184210526316e-05,
      "loss": 0.5676,
      "step": 35208
    },
    {
      "epoch": 1.971609362750588,
      "grad_norm": 1.4546359777450562,
      "learning_rate": 9.600157894736842e-05,
      "loss": 0.4586,
      "step": 35209
    },
    {
      "epoch": 1.971665360062717,
      "grad_norm": 1.0753148794174194,
      "learning_rate": 9.600131578947369e-05,
      "loss": 0.4909,
      "step": 35210
    },
    {
      "epoch": 1.971721357374846,
      "grad_norm": 1.093936800956726,
      "learning_rate": 9.600105263157895e-05,
      "loss": 0.4042,
      "step": 35211
    },
    {
      "epoch": 1.971777354686975,
      "grad_norm": 1.210041880607605,
      "learning_rate": 9.600078947368421e-05,
      "loss": 0.4656,
      "step": 35212
    },
    {
      "epoch": 1.971833351999104,
      "grad_norm": 1.2268257141113281,
      "learning_rate": 9.600052631578947e-05,
      "loss": 0.5345,
      "step": 35213
    },
    {
      "epoch": 1.971889349311233,
      "grad_norm": 1.0761581659317017,
      "learning_rate": 9.600026315789474e-05,
      "loss": 0.3414,
      "step": 35214
    },
    {
      "epoch": 1.971945346623362,
      "grad_norm": 1.282917857170105,
      "learning_rate": 9.6e-05,
      "loss": 0.4907,
      "step": 35215
    },
    {
      "epoch": 1.972001343935491,
      "grad_norm": 1.3124475479125977,
      "learning_rate": 9.599973684210528e-05,
      "loss": 0.4386,
      "step": 35216
    },
    {
      "epoch": 1.97205734124762,
      "grad_norm": 1.1063408851623535,
      "learning_rate": 9.599947368421052e-05,
      "loss": 0.3806,
      "step": 35217
    },
    {
      "epoch": 1.9721133385597491,
      "grad_norm": 1.3533333539962769,
      "learning_rate": 9.59992105263158e-05,
      "loss": 0.4063,
      "step": 35218
    },
    {
      "epoch": 1.9721693358718781,
      "grad_norm": 1.1129554510116577,
      "learning_rate": 9.599894736842105e-05,
      "loss": 0.4131,
      "step": 35219
    },
    {
      "epoch": 1.9722253331840072,
      "grad_norm": 1.182013750076294,
      "learning_rate": 9.599868421052633e-05,
      "loss": 0.4233,
      "step": 35220
    },
    {
      "epoch": 1.9722813304961362,
      "grad_norm": 1.1397662162780762,
      "learning_rate": 9.599842105263159e-05,
      "loss": 0.4027,
      "step": 35221
    },
    {
      "epoch": 1.9723373278082652,
      "grad_norm": 1.0963093042373657,
      "learning_rate": 9.599815789473685e-05,
      "loss": 0.3613,
      "step": 35222
    },
    {
      "epoch": 1.9723933251203942,
      "grad_norm": 1.1787842512130737,
      "learning_rate": 9.599789473684211e-05,
      "loss": 0.4715,
      "step": 35223
    },
    {
      "epoch": 1.9724493224325232,
      "grad_norm": 1.245827078819275,
      "learning_rate": 9.599763157894738e-05,
      "loss": 0.448,
      "step": 35224
    },
    {
      "epoch": 1.9725053197446523,
      "grad_norm": 1.415168046951294,
      "learning_rate": 9.599736842105264e-05,
      "loss": 0.4246,
      "step": 35225
    },
    {
      "epoch": 1.9725613170567813,
      "grad_norm": 1.21942937374115,
      "learning_rate": 9.59971052631579e-05,
      "loss": 0.5603,
      "step": 35226
    },
    {
      "epoch": 1.9726173143689103,
      "grad_norm": 1.5196648836135864,
      "learning_rate": 9.599684210526316e-05,
      "loss": 0.4951,
      "step": 35227
    },
    {
      "epoch": 1.9726733116810393,
      "grad_norm": 1.3526822328567505,
      "learning_rate": 9.599657894736842e-05,
      "loss": 0.5213,
      "step": 35228
    },
    {
      "epoch": 1.9727293089931683,
      "grad_norm": 1.5978784561157227,
      "learning_rate": 9.599631578947369e-05,
      "loss": 0.4612,
      "step": 35229
    },
    {
      "epoch": 1.9727853063052974,
      "grad_norm": 1.035378336906433,
      "learning_rate": 9.599605263157895e-05,
      "loss": 0.3565,
      "step": 35230
    },
    {
      "epoch": 1.9728413036174264,
      "grad_norm": 1.2160054445266724,
      "learning_rate": 9.599578947368421e-05,
      "loss": 0.4434,
      "step": 35231
    },
    {
      "epoch": 1.9728973009295554,
      "grad_norm": 1.7078431844711304,
      "learning_rate": 9.599552631578947e-05,
      "loss": 0.5087,
      "step": 35232
    },
    {
      "epoch": 1.9729532982416844,
      "grad_norm": 1.4370431900024414,
      "learning_rate": 9.599526315789475e-05,
      "loss": 0.6214,
      "step": 35233
    },
    {
      "epoch": 1.9730092955538134,
      "grad_norm": 1.3612529039382935,
      "learning_rate": 9.5995e-05,
      "loss": 0.3608,
      "step": 35234
    },
    {
      "epoch": 1.9730652928659425,
      "grad_norm": 1.5312025547027588,
      "learning_rate": 9.599473684210526e-05,
      "loss": 0.6006,
      "step": 35235
    },
    {
      "epoch": 1.9731212901780715,
      "grad_norm": 1.21440851688385,
      "learning_rate": 9.599447368421052e-05,
      "loss": 0.4015,
      "step": 35236
    },
    {
      "epoch": 1.9731772874902005,
      "grad_norm": 1.1395652294158936,
      "learning_rate": 9.59942105263158e-05,
      "loss": 0.3825,
      "step": 35237
    },
    {
      "epoch": 1.9732332848023295,
      "grad_norm": 1.1796191930770874,
      "learning_rate": 9.599394736842106e-05,
      "loss": 0.3271,
      "step": 35238
    },
    {
      "epoch": 1.9732892821144585,
      "grad_norm": 1.3698086738586426,
      "learning_rate": 9.599368421052633e-05,
      "loss": 0.4568,
      "step": 35239
    },
    {
      "epoch": 1.9733452794265876,
      "grad_norm": 1.324042797088623,
      "learning_rate": 9.599342105263158e-05,
      "loss": 0.6767,
      "step": 35240
    },
    {
      "epoch": 1.9734012767387166,
      "grad_norm": 1.1679190397262573,
      "learning_rate": 9.599315789473685e-05,
      "loss": 0.4078,
      "step": 35241
    },
    {
      "epoch": 1.9734572740508456,
      "grad_norm": 1.365000605583191,
      "learning_rate": 9.599289473684211e-05,
      "loss": 0.4339,
      "step": 35242
    },
    {
      "epoch": 1.9735132713629746,
      "grad_norm": 1.2375390529632568,
      "learning_rate": 9.599263157894738e-05,
      "loss": 0.4164,
      "step": 35243
    },
    {
      "epoch": 1.9735692686751036,
      "grad_norm": 1.109804630279541,
      "learning_rate": 9.599236842105264e-05,
      "loss": 0.4037,
      "step": 35244
    },
    {
      "epoch": 1.9736252659872326,
      "grad_norm": 1.1338765621185303,
      "learning_rate": 9.599210526315789e-05,
      "loss": 0.406,
      "step": 35245
    },
    {
      "epoch": 1.9736812632993617,
      "grad_norm": 1.3704458475112915,
      "learning_rate": 9.599184210526316e-05,
      "loss": 0.5781,
      "step": 35246
    },
    {
      "epoch": 1.9737372606114907,
      "grad_norm": 1.5784547328948975,
      "learning_rate": 9.599157894736842e-05,
      "loss": 0.5274,
      "step": 35247
    },
    {
      "epoch": 1.9737932579236197,
      "grad_norm": 1.4383635520935059,
      "learning_rate": 9.59913157894737e-05,
      "loss": 0.5551,
      "step": 35248
    },
    {
      "epoch": 1.9738492552357487,
      "grad_norm": 1.0930918455123901,
      "learning_rate": 9.599105263157894e-05,
      "loss": 0.4649,
      "step": 35249
    },
    {
      "epoch": 1.9739052525478777,
      "grad_norm": 1.5893990993499756,
      "learning_rate": 9.599078947368421e-05,
      "loss": 0.3964,
      "step": 35250
    },
    {
      "epoch": 1.9739612498600068,
      "grad_norm": 1.1147282123565674,
      "learning_rate": 9.599052631578947e-05,
      "loss": 0.5089,
      "step": 35251
    },
    {
      "epoch": 1.9740172471721358,
      "grad_norm": 1.2224620580673218,
      "learning_rate": 9.599026315789475e-05,
      "loss": 0.384,
      "step": 35252
    },
    {
      "epoch": 1.9740732444842648,
      "grad_norm": 1.1296567916870117,
      "learning_rate": 9.599000000000001e-05,
      "loss": 0.3237,
      "step": 35253
    },
    {
      "epoch": 1.9741292417963938,
      "grad_norm": 1.2474920749664307,
      "learning_rate": 9.598973684210527e-05,
      "loss": 0.3414,
      "step": 35254
    },
    {
      "epoch": 1.9741852391085228,
      "grad_norm": 1.3443057537078857,
      "learning_rate": 9.598947368421053e-05,
      "loss": 0.3491,
      "step": 35255
    },
    {
      "epoch": 1.9742412364206519,
      "grad_norm": 1.239501714706421,
      "learning_rate": 9.59892105263158e-05,
      "loss": 0.4116,
      "step": 35256
    },
    {
      "epoch": 1.9742972337327809,
      "grad_norm": 1.2772403955459595,
      "learning_rate": 9.598894736842106e-05,
      "loss": 0.4355,
      "step": 35257
    },
    {
      "epoch": 1.97435323104491,
      "grad_norm": 1.468505620956421,
      "learning_rate": 9.598868421052632e-05,
      "loss": 0.5805,
      "step": 35258
    },
    {
      "epoch": 1.974409228357039,
      "grad_norm": 1.5236645936965942,
      "learning_rate": 9.598842105263158e-05,
      "loss": 0.5462,
      "step": 35259
    },
    {
      "epoch": 1.974465225669168,
      "grad_norm": 1.3489094972610474,
      "learning_rate": 9.598815789473685e-05,
      "loss": 0.4622,
      "step": 35260
    },
    {
      "epoch": 1.974521222981297,
      "grad_norm": 1.4805444478988647,
      "learning_rate": 9.598789473684211e-05,
      "loss": 0.4942,
      "step": 35261
    },
    {
      "epoch": 1.9745772202934257,
      "grad_norm": 1.370435118675232,
      "learning_rate": 9.598763157894737e-05,
      "loss": 0.3483,
      "step": 35262
    },
    {
      "epoch": 1.9746332176055548,
      "grad_norm": 1.4066795110702515,
      "learning_rate": 9.598736842105263e-05,
      "loss": 0.5351,
      "step": 35263
    },
    {
      "epoch": 1.9746892149176838,
      "grad_norm": 1.5087624788284302,
      "learning_rate": 9.598710526315789e-05,
      "loss": 0.5398,
      "step": 35264
    },
    {
      "epoch": 1.9747452122298128,
      "grad_norm": 1.315524935722351,
      "learning_rate": 9.598684210526317e-05,
      "loss": 0.4398,
      "step": 35265
    },
    {
      "epoch": 1.9748012095419418,
      "grad_norm": 1.300810694694519,
      "learning_rate": 9.598657894736842e-05,
      "loss": 0.4372,
      "step": 35266
    },
    {
      "epoch": 1.9748572068540708,
      "grad_norm": 1.5306553840637207,
      "learning_rate": 9.598631578947368e-05,
      "loss": 0.4294,
      "step": 35267
    },
    {
      "epoch": 1.9749132041661999,
      "grad_norm": 1.1755743026733398,
      "learning_rate": 9.598605263157894e-05,
      "loss": 0.4432,
      "step": 35268
    },
    {
      "epoch": 1.9749692014783289,
      "grad_norm": 1.3118199110031128,
      "learning_rate": 9.598578947368422e-05,
      "loss": 0.4627,
      "step": 35269
    },
    {
      "epoch": 1.975025198790458,
      "grad_norm": 1.2914881706237793,
      "learning_rate": 9.598552631578948e-05,
      "loss": 0.4053,
      "step": 35270
    },
    {
      "epoch": 1.975081196102587,
      "grad_norm": 1.8204461336135864,
      "learning_rate": 9.598526315789475e-05,
      "loss": 0.3838,
      "step": 35271
    },
    {
      "epoch": 1.975137193414716,
      "grad_norm": 1.2272690534591675,
      "learning_rate": 9.5985e-05,
      "loss": 0.3862,
      "step": 35272
    },
    {
      "epoch": 1.975193190726845,
      "grad_norm": 1.3409082889556885,
      "learning_rate": 9.598473684210527e-05,
      "loss": 0.4829,
      "step": 35273
    },
    {
      "epoch": 1.975249188038974,
      "grad_norm": 1.3153965473175049,
      "learning_rate": 9.598447368421053e-05,
      "loss": 0.421,
      "step": 35274
    },
    {
      "epoch": 1.975305185351103,
      "grad_norm": 1.3150523900985718,
      "learning_rate": 9.59842105263158e-05,
      "loss": 0.3628,
      "step": 35275
    },
    {
      "epoch": 1.975361182663232,
      "grad_norm": 1.2833306789398193,
      "learning_rate": 9.598394736842106e-05,
      "loss": 0.4401,
      "step": 35276
    },
    {
      "epoch": 1.975417179975361,
      "grad_norm": 1.3704347610473633,
      "learning_rate": 9.598368421052632e-05,
      "loss": 0.3553,
      "step": 35277
    },
    {
      "epoch": 1.97547317728749,
      "grad_norm": 1.6373692750930786,
      "learning_rate": 9.598342105263158e-05,
      "loss": 0.5005,
      "step": 35278
    },
    {
      "epoch": 1.975529174599619,
      "grad_norm": 1.0903794765472412,
      "learning_rate": 9.598315789473684e-05,
      "loss": 0.3671,
      "step": 35279
    },
    {
      "epoch": 1.975585171911748,
      "grad_norm": 1.225213646888733,
      "learning_rate": 9.598289473684212e-05,
      "loss": 0.341,
      "step": 35280
    },
    {
      "epoch": 1.9756411692238771,
      "grad_norm": 1.1427005529403687,
      "learning_rate": 9.598263157894737e-05,
      "loss": 0.551,
      "step": 35281
    },
    {
      "epoch": 1.9756971665360061,
      "grad_norm": 1.1421732902526855,
      "learning_rate": 9.598236842105263e-05,
      "loss": 0.3506,
      "step": 35282
    },
    {
      "epoch": 1.9757531638481352,
      "grad_norm": 1.3825337886810303,
      "learning_rate": 9.59821052631579e-05,
      "loss": 0.3827,
      "step": 35283
    },
    {
      "epoch": 1.9758091611602642,
      "grad_norm": 1.250551700592041,
      "learning_rate": 9.598184210526317e-05,
      "loss": 0.3922,
      "step": 35284
    },
    {
      "epoch": 1.9758651584723932,
      "grad_norm": 1.4467190504074097,
      "learning_rate": 9.598157894736843e-05,
      "loss": 0.608,
      "step": 35285
    },
    {
      "epoch": 1.9759211557845222,
      "grad_norm": 1.1037590503692627,
      "learning_rate": 9.598131578947369e-05,
      "loss": 0.4356,
      "step": 35286
    },
    {
      "epoch": 1.9759771530966512,
      "grad_norm": 1.0927555561065674,
      "learning_rate": 9.598105263157895e-05,
      "loss": 0.4336,
      "step": 35287
    },
    {
      "epoch": 1.9760331504087802,
      "grad_norm": 1.1293150186538696,
      "learning_rate": 9.598078947368422e-05,
      "loss": 0.3492,
      "step": 35288
    },
    {
      "epoch": 1.9760891477209093,
      "grad_norm": 1.0118589401245117,
      "learning_rate": 9.598052631578948e-05,
      "loss": 0.3937,
      "step": 35289
    },
    {
      "epoch": 1.9761451450330383,
      "grad_norm": 1.2155112028121948,
      "learning_rate": 9.598026315789474e-05,
      "loss": 0.4235,
      "step": 35290
    },
    {
      "epoch": 1.9762011423451673,
      "grad_norm": 1.6978845596313477,
      "learning_rate": 9.598e-05,
      "loss": 0.447,
      "step": 35291
    },
    {
      "epoch": 1.9762571396572963,
      "grad_norm": 1.2969763278961182,
      "learning_rate": 9.597973684210527e-05,
      "loss": 0.5313,
      "step": 35292
    },
    {
      "epoch": 1.9763131369694253,
      "grad_norm": 1.598415732383728,
      "learning_rate": 9.597947368421053e-05,
      "loss": 0.5412,
      "step": 35293
    },
    {
      "epoch": 1.9763691342815544,
      "grad_norm": 1.4855791330337524,
      "learning_rate": 9.59792105263158e-05,
      "loss": 0.5057,
      "step": 35294
    },
    {
      "epoch": 1.9764251315936834,
      "grad_norm": 1.07868492603302,
      "learning_rate": 9.597894736842105e-05,
      "loss": 0.6179,
      "step": 35295
    },
    {
      "epoch": 1.9764811289058124,
      "grad_norm": 1.5606224536895752,
      "learning_rate": 9.597868421052631e-05,
      "loss": 0.4144,
      "step": 35296
    },
    {
      "epoch": 1.9765371262179414,
      "grad_norm": 1.1112538576126099,
      "learning_rate": 9.597842105263158e-05,
      "loss": 0.401,
      "step": 35297
    },
    {
      "epoch": 1.9765931235300704,
      "grad_norm": 1.1307101249694824,
      "learning_rate": 9.597815789473684e-05,
      "loss": 0.4075,
      "step": 35298
    },
    {
      "epoch": 1.9766491208421995,
      "grad_norm": 1.7817635536193848,
      "learning_rate": 9.597789473684212e-05,
      "loss": 0.4521,
      "step": 35299
    },
    {
      "epoch": 1.9767051181543285,
      "grad_norm": 1.455461859703064,
      "learning_rate": 9.597763157894736e-05,
      "loss": 0.4451,
      "step": 35300
    },
    {
      "epoch": 1.9767611154664575,
      "grad_norm": 1.3352333307266235,
      "learning_rate": 9.597736842105264e-05,
      "loss": 0.408,
      "step": 35301
    },
    {
      "epoch": 1.9768171127785865,
      "grad_norm": 1.4200162887573242,
      "learning_rate": 9.59771052631579e-05,
      "loss": 0.6055,
      "step": 35302
    },
    {
      "epoch": 1.9768731100907155,
      "grad_norm": 1.4622924327850342,
      "learning_rate": 9.597684210526317e-05,
      "loss": 0.5753,
      "step": 35303
    },
    {
      "epoch": 1.9769291074028446,
      "grad_norm": 1.3756738901138306,
      "learning_rate": 9.597657894736842e-05,
      "loss": 0.3705,
      "step": 35304
    },
    {
      "epoch": 1.9769851047149736,
      "grad_norm": 1.5039740800857544,
      "learning_rate": 9.597631578947369e-05,
      "loss": 0.5865,
      "step": 35305
    },
    {
      "epoch": 1.9770411020271026,
      "grad_norm": 1.6791962385177612,
      "learning_rate": 9.597605263157895e-05,
      "loss": 0.4755,
      "step": 35306
    },
    {
      "epoch": 1.9770970993392316,
      "grad_norm": 1.1733907461166382,
      "learning_rate": 9.597578947368422e-05,
      "loss": 0.5278,
      "step": 35307
    },
    {
      "epoch": 1.9771530966513606,
      "grad_norm": 1.375545859336853,
      "learning_rate": 9.597552631578948e-05,
      "loss": 0.4223,
      "step": 35308
    },
    {
      "epoch": 1.9772090939634897,
      "grad_norm": 1.0620664358139038,
      "learning_rate": 9.597526315789474e-05,
      "loss": 0.3884,
      "step": 35309
    },
    {
      "epoch": 1.9772650912756187,
      "grad_norm": 1.1718782186508179,
      "learning_rate": 9.5975e-05,
      "loss": 0.4194,
      "step": 35310
    },
    {
      "epoch": 1.9773210885877477,
      "grad_norm": 1.4529876708984375,
      "learning_rate": 9.597473684210528e-05,
      "loss": 0.4884,
      "step": 35311
    },
    {
      "epoch": 1.9773770858998767,
      "grad_norm": 1.1633386611938477,
      "learning_rate": 9.597447368421053e-05,
      "loss": 0.4522,
      "step": 35312
    },
    {
      "epoch": 1.9774330832120057,
      "grad_norm": 1.3067140579223633,
      "learning_rate": 9.59742105263158e-05,
      "loss": 0.5606,
      "step": 35313
    },
    {
      "epoch": 1.9774890805241347,
      "grad_norm": 1.4889006614685059,
      "learning_rate": 9.597394736842105e-05,
      "loss": 0.3932,
      "step": 35314
    },
    {
      "epoch": 1.9775450778362638,
      "grad_norm": 1.1453803777694702,
      "learning_rate": 9.597368421052631e-05,
      "loss": 0.4174,
      "step": 35315
    },
    {
      "epoch": 1.9776010751483928,
      "grad_norm": 1.5066471099853516,
      "learning_rate": 9.597342105263159e-05,
      "loss": 0.4884,
      "step": 35316
    },
    {
      "epoch": 1.9776570724605218,
      "grad_norm": 1.270275592803955,
      "learning_rate": 9.597315789473685e-05,
      "loss": 0.4955,
      "step": 35317
    },
    {
      "epoch": 1.9777130697726508,
      "grad_norm": 1.5205512046813965,
      "learning_rate": 9.59728947368421e-05,
      "loss": 0.4563,
      "step": 35318
    },
    {
      "epoch": 1.9777690670847798,
      "grad_norm": 1.362046718597412,
      "learning_rate": 9.597263157894737e-05,
      "loss": 0.5449,
      "step": 35319
    },
    {
      "epoch": 1.9778250643969089,
      "grad_norm": 1.1937780380249023,
      "learning_rate": 9.597236842105264e-05,
      "loss": 0.4954,
      "step": 35320
    },
    {
      "epoch": 1.9778810617090379,
      "grad_norm": 1.38205885887146,
      "learning_rate": 9.59721052631579e-05,
      "loss": 0.5093,
      "step": 35321
    },
    {
      "epoch": 1.977937059021167,
      "grad_norm": 1.2679609060287476,
      "learning_rate": 9.597184210526316e-05,
      "loss": 0.3756,
      "step": 35322
    },
    {
      "epoch": 1.977993056333296,
      "grad_norm": 1.0841118097305298,
      "learning_rate": 9.597157894736842e-05,
      "loss": 0.3772,
      "step": 35323
    },
    {
      "epoch": 1.978049053645425,
      "grad_norm": 1.3491220474243164,
      "learning_rate": 9.597131578947369e-05,
      "loss": 0.4203,
      "step": 35324
    },
    {
      "epoch": 1.978105050957554,
      "grad_norm": 1.5062943696975708,
      "learning_rate": 9.597105263157895e-05,
      "loss": 0.4433,
      "step": 35325
    },
    {
      "epoch": 1.978161048269683,
      "grad_norm": 1.08420991897583,
      "learning_rate": 9.597078947368423e-05,
      "loss": 0.4247,
      "step": 35326
    },
    {
      "epoch": 1.978217045581812,
      "grad_norm": 1.4956640005111694,
      "learning_rate": 9.597052631578947e-05,
      "loss": 0.5669,
      "step": 35327
    },
    {
      "epoch": 1.978273042893941,
      "grad_norm": 1.1101624965667725,
      "learning_rate": 9.597026315789474e-05,
      "loss": 0.4106,
      "step": 35328
    },
    {
      "epoch": 1.97832904020607,
      "grad_norm": 1.8630064725875854,
      "learning_rate": 9.597e-05,
      "loss": 0.7349,
      "step": 35329
    },
    {
      "epoch": 1.978385037518199,
      "grad_norm": 1.4439631700515747,
      "learning_rate": 9.596973684210526e-05,
      "loss": 0.4387,
      "step": 35330
    },
    {
      "epoch": 1.978441034830328,
      "grad_norm": 1.2561832666397095,
      "learning_rate": 9.596947368421054e-05,
      "loss": 0.4245,
      "step": 35331
    },
    {
      "epoch": 1.978497032142457,
      "grad_norm": 1.7012962102890015,
      "learning_rate": 9.596921052631578e-05,
      "loss": 0.5488,
      "step": 35332
    },
    {
      "epoch": 1.978553029454586,
      "grad_norm": 1.9780102968215942,
      "learning_rate": 9.596894736842106e-05,
      "loss": 0.3328,
      "step": 35333
    },
    {
      "epoch": 1.9786090267667151,
      "grad_norm": 1.3113173246383667,
      "learning_rate": 9.596868421052632e-05,
      "loss": 0.4253,
      "step": 35334
    },
    {
      "epoch": 1.9786650240788441,
      "grad_norm": 1.534112811088562,
      "learning_rate": 9.596842105263159e-05,
      "loss": 0.5875,
      "step": 35335
    },
    {
      "epoch": 1.9787210213909732,
      "grad_norm": 1.8889906406402588,
      "learning_rate": 9.596815789473685e-05,
      "loss": 0.4843,
      "step": 35336
    },
    {
      "epoch": 1.9787770187031022,
      "grad_norm": 1.081467866897583,
      "learning_rate": 9.596789473684211e-05,
      "loss": 0.3644,
      "step": 35337
    },
    {
      "epoch": 1.9788330160152312,
      "grad_norm": 1.1088752746582031,
      "learning_rate": 9.596763157894737e-05,
      "loss": 0.3999,
      "step": 35338
    },
    {
      "epoch": 1.9788890133273602,
      "grad_norm": 1.027759313583374,
      "learning_rate": 9.596736842105264e-05,
      "loss": 0.36,
      "step": 35339
    },
    {
      "epoch": 1.9789450106394892,
      "grad_norm": 1.1924670934677124,
      "learning_rate": 9.59671052631579e-05,
      "loss": 0.3514,
      "step": 35340
    },
    {
      "epoch": 1.9790010079516183,
      "grad_norm": 1.1977674961090088,
      "learning_rate": 9.596684210526316e-05,
      "loss": 0.3228,
      "step": 35341
    },
    {
      "epoch": 1.9790570052637473,
      "grad_norm": 3.1098453998565674,
      "learning_rate": 9.596657894736842e-05,
      "loss": 0.5448,
      "step": 35342
    },
    {
      "epoch": 1.9791130025758763,
      "grad_norm": 1.3640179634094238,
      "learning_rate": 9.59663157894737e-05,
      "loss": 0.4023,
      "step": 35343
    },
    {
      "epoch": 1.9791689998880053,
      "grad_norm": 1.2926201820373535,
      "learning_rate": 9.596605263157895e-05,
      "loss": 0.4943,
      "step": 35344
    },
    {
      "epoch": 1.9792249972001343,
      "grad_norm": 1.4971150159835815,
      "learning_rate": 9.596578947368421e-05,
      "loss": 0.6926,
      "step": 35345
    },
    {
      "epoch": 1.9792809945122634,
      "grad_norm": 1.3296997547149658,
      "learning_rate": 9.596552631578947e-05,
      "loss": 0.5384,
      "step": 35346
    },
    {
      "epoch": 1.9793369918243924,
      "grad_norm": 1.3212518692016602,
      "learning_rate": 9.596526315789475e-05,
      "loss": 0.4779,
      "step": 35347
    },
    {
      "epoch": 1.9793929891365214,
      "grad_norm": 1.2288786172866821,
      "learning_rate": 9.596500000000001e-05,
      "loss": 0.3691,
      "step": 35348
    },
    {
      "epoch": 1.9794489864486504,
      "grad_norm": 1.3563345670700073,
      "learning_rate": 9.596473684210527e-05,
      "loss": 0.5572,
      "step": 35349
    },
    {
      "epoch": 1.9795049837607794,
      "grad_norm": 1.5157183408737183,
      "learning_rate": 9.596447368421053e-05,
      "loss": 0.4621,
      "step": 35350
    },
    {
      "epoch": 1.9795609810729085,
      "grad_norm": 1.6226670742034912,
      "learning_rate": 9.596421052631579e-05,
      "loss": 0.5195,
      "step": 35351
    },
    {
      "epoch": 1.9796169783850375,
      "grad_norm": 1.340502381324768,
      "learning_rate": 9.596394736842106e-05,
      "loss": 0.4098,
      "step": 35352
    },
    {
      "epoch": 1.9796729756971665,
      "grad_norm": 1.1395243406295776,
      "learning_rate": 9.596368421052632e-05,
      "loss": 0.5622,
      "step": 35353
    },
    {
      "epoch": 1.9797289730092955,
      "grad_norm": 1.2077263593673706,
      "learning_rate": 9.596342105263159e-05,
      "loss": 0.4037,
      "step": 35354
    },
    {
      "epoch": 1.9797849703214245,
      "grad_norm": 1.1225641965866089,
      "learning_rate": 9.596315789473684e-05,
      "loss": 0.3685,
      "step": 35355
    },
    {
      "epoch": 1.9798409676335536,
      "grad_norm": 1.0156373977661133,
      "learning_rate": 9.596289473684211e-05,
      "loss": 0.4296,
      "step": 35356
    },
    {
      "epoch": 1.9798969649456826,
      "grad_norm": 1.2100605964660645,
      "learning_rate": 9.596263157894737e-05,
      "loss": 0.3318,
      "step": 35357
    },
    {
      "epoch": 1.9799529622578116,
      "grad_norm": 1.1853482723236084,
      "learning_rate": 9.596236842105264e-05,
      "loss": 0.3517,
      "step": 35358
    },
    {
      "epoch": 1.9800089595699406,
      "grad_norm": 1.1259390115737915,
      "learning_rate": 9.596210526315789e-05,
      "loss": 0.4718,
      "step": 35359
    },
    {
      "epoch": 1.9800649568820696,
      "grad_norm": 0.9772047996520996,
      "learning_rate": 9.596184210526316e-05,
      "loss": 0.3278,
      "step": 35360
    },
    {
      "epoch": 1.9801209541941986,
      "grad_norm": 1.421338677406311,
      "learning_rate": 9.596157894736842e-05,
      "loss": 0.5354,
      "step": 35361
    },
    {
      "epoch": 1.9801769515063277,
      "grad_norm": 1.678537368774414,
      "learning_rate": 9.59613157894737e-05,
      "loss": 0.4781,
      "step": 35362
    },
    {
      "epoch": 1.9802329488184567,
      "grad_norm": 1.3583216667175293,
      "learning_rate": 9.596105263157896e-05,
      "loss": 0.5838,
      "step": 35363
    },
    {
      "epoch": 1.9802889461305857,
      "grad_norm": 1.3110512495040894,
      "learning_rate": 9.596078947368422e-05,
      "loss": 0.4631,
      "step": 35364
    },
    {
      "epoch": 1.9803449434427147,
      "grad_norm": 1.4307591915130615,
      "learning_rate": 9.596052631578948e-05,
      "loss": 0.4516,
      "step": 35365
    },
    {
      "epoch": 1.9804009407548437,
      "grad_norm": 1.3773220777511597,
      "learning_rate": 9.596026315789474e-05,
      "loss": 0.5839,
      "step": 35366
    },
    {
      "epoch": 1.9804569380669728,
      "grad_norm": 1.1750822067260742,
      "learning_rate": 9.596000000000001e-05,
      "loss": 0.4467,
      "step": 35367
    },
    {
      "epoch": 1.9805129353791018,
      "grad_norm": 1.1129287481307983,
      "learning_rate": 9.595973684210527e-05,
      "loss": 0.5654,
      "step": 35368
    },
    {
      "epoch": 1.9805689326912308,
      "grad_norm": 1.5263557434082031,
      "learning_rate": 9.595947368421053e-05,
      "loss": 0.4007,
      "step": 35369
    },
    {
      "epoch": 1.9806249300033598,
      "grad_norm": 1.5154064893722534,
      "learning_rate": 9.595921052631579e-05,
      "loss": 0.5398,
      "step": 35370
    },
    {
      "epoch": 1.9806809273154888,
      "grad_norm": 1.120880126953125,
      "learning_rate": 9.595894736842106e-05,
      "loss": 0.368,
      "step": 35371
    },
    {
      "epoch": 1.9807369246276179,
      "grad_norm": 1.1270954608917236,
      "learning_rate": 9.595868421052632e-05,
      "loss": 0.4904,
      "step": 35372
    },
    {
      "epoch": 1.9807929219397469,
      "grad_norm": 1.2989213466644287,
      "learning_rate": 9.595842105263158e-05,
      "loss": 0.4836,
      "step": 35373
    },
    {
      "epoch": 1.980848919251876,
      "grad_norm": 0.9655188918113708,
      "learning_rate": 9.595815789473684e-05,
      "loss": 0.3337,
      "step": 35374
    },
    {
      "epoch": 1.980904916564005,
      "grad_norm": 1.3269989490509033,
      "learning_rate": 9.595789473684211e-05,
      "loss": 0.4409,
      "step": 35375
    },
    {
      "epoch": 1.980960913876134,
      "grad_norm": 1.2914983034133911,
      "learning_rate": 9.595763157894737e-05,
      "loss": 0.466,
      "step": 35376
    },
    {
      "epoch": 1.981016911188263,
      "grad_norm": 1.32680082321167,
      "learning_rate": 9.595736842105263e-05,
      "loss": 0.466,
      "step": 35377
    },
    {
      "epoch": 1.981072908500392,
      "grad_norm": 1.374399185180664,
      "learning_rate": 9.59571052631579e-05,
      "loss": 0.5656,
      "step": 35378
    },
    {
      "epoch": 1.981128905812521,
      "grad_norm": 1.424728512763977,
      "learning_rate": 9.595684210526317e-05,
      "loss": 0.4942,
      "step": 35379
    },
    {
      "epoch": 1.98118490312465,
      "grad_norm": 1.3107686042785645,
      "learning_rate": 9.595657894736843e-05,
      "loss": 0.3764,
      "step": 35380
    },
    {
      "epoch": 1.981240900436779,
      "grad_norm": 1.2910714149475098,
      "learning_rate": 9.59563157894737e-05,
      "loss": 0.3745,
      "step": 35381
    },
    {
      "epoch": 1.981296897748908,
      "grad_norm": 1.1032922267913818,
      "learning_rate": 9.595605263157895e-05,
      "loss": 0.3212,
      "step": 35382
    },
    {
      "epoch": 1.981352895061037,
      "grad_norm": 0.9643816947937012,
      "learning_rate": 9.59557894736842e-05,
      "loss": 0.3388,
      "step": 35383
    },
    {
      "epoch": 1.981408892373166,
      "grad_norm": 1.2272729873657227,
      "learning_rate": 9.595552631578948e-05,
      "loss": 0.3363,
      "step": 35384
    },
    {
      "epoch": 1.981464889685295,
      "grad_norm": 1.1242787837982178,
      "learning_rate": 9.595526315789474e-05,
      "loss": 0.4673,
      "step": 35385
    },
    {
      "epoch": 1.9815208869974241,
      "grad_norm": 1.1283066272735596,
      "learning_rate": 9.595500000000001e-05,
      "loss": 0.4302,
      "step": 35386
    },
    {
      "epoch": 1.9815768843095531,
      "grad_norm": 1.655296802520752,
      "learning_rate": 9.595473684210526e-05,
      "loss": 0.4352,
      "step": 35387
    },
    {
      "epoch": 1.9816328816216822,
      "grad_norm": 1.2834354639053345,
      "learning_rate": 9.595447368421053e-05,
      "loss": 0.4599,
      "step": 35388
    },
    {
      "epoch": 1.9816888789338112,
      "grad_norm": 1.2974046468734741,
      "learning_rate": 9.595421052631579e-05,
      "loss": 0.5437,
      "step": 35389
    },
    {
      "epoch": 1.9817448762459402,
      "grad_norm": 1.1857694387435913,
      "learning_rate": 9.595394736842106e-05,
      "loss": 0.477,
      "step": 35390
    },
    {
      "epoch": 1.9818008735580692,
      "grad_norm": 1.22774076461792,
      "learning_rate": 9.595368421052632e-05,
      "loss": 0.4467,
      "step": 35391
    },
    {
      "epoch": 1.9818568708701982,
      "grad_norm": 1.5466434955596924,
      "learning_rate": 9.595342105263158e-05,
      "loss": 0.5265,
      "step": 35392
    },
    {
      "epoch": 1.9819128681823273,
      "grad_norm": 1.5745201110839844,
      "learning_rate": 9.595315789473684e-05,
      "loss": 0.4461,
      "step": 35393
    },
    {
      "epoch": 1.9819688654944563,
      "grad_norm": 1.8737821578979492,
      "learning_rate": 9.595289473684212e-05,
      "loss": 0.6758,
      "step": 35394
    },
    {
      "epoch": 1.9820248628065853,
      "grad_norm": 1.8726141452789307,
      "learning_rate": 9.595263157894738e-05,
      "loss": 0.6814,
      "step": 35395
    },
    {
      "epoch": 1.9820808601187143,
      "grad_norm": 1.350630760192871,
      "learning_rate": 9.595236842105264e-05,
      "loss": 0.3601,
      "step": 35396
    },
    {
      "epoch": 1.9821368574308433,
      "grad_norm": 1.2303060293197632,
      "learning_rate": 9.59521052631579e-05,
      "loss": 0.407,
      "step": 35397
    },
    {
      "epoch": 1.9821928547429724,
      "grad_norm": 1.2057762145996094,
      "learning_rate": 9.595184210526317e-05,
      "loss": 0.4318,
      "step": 35398
    },
    {
      "epoch": 1.9822488520551014,
      "grad_norm": 1.0356110334396362,
      "learning_rate": 9.595157894736843e-05,
      "loss": 0.4149,
      "step": 35399
    },
    {
      "epoch": 1.9823048493672304,
      "grad_norm": 1.197208285331726,
      "learning_rate": 9.595131578947369e-05,
      "loss": 0.4863,
      "step": 35400
    },
    {
      "epoch": 1.9823608466793594,
      "grad_norm": 1.2359369993209839,
      "learning_rate": 9.595105263157895e-05,
      "loss": 0.4589,
      "step": 35401
    },
    {
      "epoch": 1.9824168439914884,
      "grad_norm": 1.2901225090026855,
      "learning_rate": 9.595078947368421e-05,
      "loss": 0.471,
      "step": 35402
    },
    {
      "epoch": 1.9824728413036175,
      "grad_norm": 1.406449317932129,
      "learning_rate": 9.595052631578948e-05,
      "loss": 0.4205,
      "step": 35403
    },
    {
      "epoch": 1.9825288386157465,
      "grad_norm": 1.2992604970932007,
      "learning_rate": 9.595026315789474e-05,
      "loss": 0.4876,
      "step": 35404
    },
    {
      "epoch": 1.9825848359278755,
      "grad_norm": 1.2177059650421143,
      "learning_rate": 9.595e-05,
      "loss": 0.3949,
      "step": 35405
    },
    {
      "epoch": 1.9826408332400045,
      "grad_norm": 1.4046787023544312,
      "learning_rate": 9.594973684210526e-05,
      "loss": 0.467,
      "step": 35406
    },
    {
      "epoch": 1.9826968305521335,
      "grad_norm": 1.054486632347107,
      "learning_rate": 9.594947368421053e-05,
      "loss": 0.3462,
      "step": 35407
    },
    {
      "epoch": 1.9827528278642625,
      "grad_norm": 1.1303553581237793,
      "learning_rate": 9.59492105263158e-05,
      "loss": 0.3987,
      "step": 35408
    },
    {
      "epoch": 1.9828088251763916,
      "grad_norm": 1.245105266571045,
      "learning_rate": 9.594894736842107e-05,
      "loss": 0.5129,
      "step": 35409
    },
    {
      "epoch": 1.9828648224885206,
      "grad_norm": 1.9922682046890259,
      "learning_rate": 9.594868421052631e-05,
      "loss": 0.3798,
      "step": 35410
    },
    {
      "epoch": 1.9829208198006496,
      "grad_norm": 1.4982882738113403,
      "learning_rate": 9.594842105263159e-05,
      "loss": 0.4651,
      "step": 35411
    },
    {
      "epoch": 1.9829768171127786,
      "grad_norm": 1.1572262048721313,
      "learning_rate": 9.594815789473685e-05,
      "loss": 0.4673,
      "step": 35412
    },
    {
      "epoch": 1.9830328144249076,
      "grad_norm": 1.0734959840774536,
      "learning_rate": 9.594789473684212e-05,
      "loss": 0.5939,
      "step": 35413
    },
    {
      "epoch": 1.9830888117370367,
      "grad_norm": 1.112282633781433,
      "learning_rate": 9.594763157894737e-05,
      "loss": 0.344,
      "step": 35414
    },
    {
      "epoch": 1.9831448090491657,
      "grad_norm": 1.1996608972549438,
      "learning_rate": 9.594736842105264e-05,
      "loss": 0.658,
      "step": 35415
    },
    {
      "epoch": 1.9832008063612947,
      "grad_norm": 1.3755563497543335,
      "learning_rate": 9.59471052631579e-05,
      "loss": 0.407,
      "step": 35416
    },
    {
      "epoch": 1.9832568036734237,
      "grad_norm": 1.1868852376937866,
      "learning_rate": 9.594684210526316e-05,
      "loss": 0.4186,
      "step": 35417
    },
    {
      "epoch": 1.9833128009855527,
      "grad_norm": 1.6150925159454346,
      "learning_rate": 9.594657894736843e-05,
      "loss": 0.5411,
      "step": 35418
    },
    {
      "epoch": 1.9833687982976818,
      "grad_norm": 1.4090509414672852,
      "learning_rate": 9.594631578947368e-05,
      "loss": 0.5802,
      "step": 35419
    },
    {
      "epoch": 1.9834247956098108,
      "grad_norm": 1.4140605926513672,
      "learning_rate": 9.594605263157895e-05,
      "loss": 0.4161,
      "step": 35420
    },
    {
      "epoch": 1.9834807929219398,
      "grad_norm": 1.141640305519104,
      "learning_rate": 9.594578947368421e-05,
      "loss": 0.5105,
      "step": 35421
    },
    {
      "epoch": 1.9835367902340688,
      "grad_norm": 1.4441699981689453,
      "learning_rate": 9.594552631578948e-05,
      "loss": 0.401,
      "step": 35422
    },
    {
      "epoch": 1.9835927875461978,
      "grad_norm": 1.2866114377975464,
      "learning_rate": 9.594526315789474e-05,
      "loss": 0.5019,
      "step": 35423
    },
    {
      "epoch": 1.9836487848583269,
      "grad_norm": 1.2182586193084717,
      "learning_rate": 9.5945e-05,
      "loss": 0.4528,
      "step": 35424
    },
    {
      "epoch": 1.9837047821704559,
      "grad_norm": 1.0991075038909912,
      "learning_rate": 9.594473684210526e-05,
      "loss": 0.413,
      "step": 35425
    },
    {
      "epoch": 1.983760779482585,
      "grad_norm": 1.2246930599212646,
      "learning_rate": 9.594447368421054e-05,
      "loss": 0.5544,
      "step": 35426
    },
    {
      "epoch": 1.983816776794714,
      "grad_norm": 1.139660120010376,
      "learning_rate": 9.59442105263158e-05,
      "loss": 0.3881,
      "step": 35427
    },
    {
      "epoch": 1.983872774106843,
      "grad_norm": 1.272331714630127,
      "learning_rate": 9.594394736842106e-05,
      "loss": 0.4805,
      "step": 35428
    },
    {
      "epoch": 1.983928771418972,
      "grad_norm": 1.4612683057785034,
      "learning_rate": 9.594368421052632e-05,
      "loss": 0.4432,
      "step": 35429
    },
    {
      "epoch": 1.983984768731101,
      "grad_norm": 1.177104115486145,
      "learning_rate": 9.594342105263159e-05,
      "loss": 0.4355,
      "step": 35430
    },
    {
      "epoch": 1.98404076604323,
      "grad_norm": 1.148566484451294,
      "learning_rate": 9.594315789473685e-05,
      "loss": 0.3758,
      "step": 35431
    },
    {
      "epoch": 1.984096763355359,
      "grad_norm": 1.3939623832702637,
      "learning_rate": 9.594289473684211e-05,
      "loss": 0.4636,
      "step": 35432
    },
    {
      "epoch": 1.984152760667488,
      "grad_norm": 1.4306120872497559,
      "learning_rate": 9.594263157894737e-05,
      "loss": 0.3576,
      "step": 35433
    },
    {
      "epoch": 1.984208757979617,
      "grad_norm": 1.126480221748352,
      "learning_rate": 9.594236842105263e-05,
      "loss": 0.4908,
      "step": 35434
    },
    {
      "epoch": 1.984264755291746,
      "grad_norm": 1.2852604389190674,
      "learning_rate": 9.59421052631579e-05,
      "loss": 0.3717,
      "step": 35435
    },
    {
      "epoch": 1.984320752603875,
      "grad_norm": 1.189316987991333,
      "learning_rate": 9.594184210526316e-05,
      "loss": 0.4749,
      "step": 35436
    },
    {
      "epoch": 1.984376749916004,
      "grad_norm": 1.1891878843307495,
      "learning_rate": 9.594157894736842e-05,
      "loss": 0.4139,
      "step": 35437
    },
    {
      "epoch": 1.9844327472281331,
      "grad_norm": 1.2388026714324951,
      "learning_rate": 9.594131578947368e-05,
      "loss": 0.5164,
      "step": 35438
    },
    {
      "epoch": 1.9844887445402621,
      "grad_norm": 1.4272000789642334,
      "learning_rate": 9.594105263157895e-05,
      "loss": 0.4422,
      "step": 35439
    },
    {
      "epoch": 1.9845447418523912,
      "grad_norm": 1.5697245597839355,
      "learning_rate": 9.594078947368421e-05,
      "loss": 0.3517,
      "step": 35440
    },
    {
      "epoch": 1.9846007391645202,
      "grad_norm": 1.6420040130615234,
      "learning_rate": 9.594052631578949e-05,
      "loss": 0.4172,
      "step": 35441
    },
    {
      "epoch": 1.9846567364766492,
      "grad_norm": 1.081558346748352,
      "learning_rate": 9.594026315789473e-05,
      "loss": 0.3497,
      "step": 35442
    },
    {
      "epoch": 1.9847127337887782,
      "grad_norm": 1.1123874187469482,
      "learning_rate": 9.594e-05,
      "loss": 0.3874,
      "step": 35443
    },
    {
      "epoch": 1.9847687311009072,
      "grad_norm": 1.313855528831482,
      "learning_rate": 9.593973684210527e-05,
      "loss": 0.5523,
      "step": 35444
    },
    {
      "epoch": 1.9848247284130363,
      "grad_norm": 1.5975446701049805,
      "learning_rate": 9.593947368421054e-05,
      "loss": 0.4747,
      "step": 35445
    },
    {
      "epoch": 1.9848807257251653,
      "grad_norm": 1.0752410888671875,
      "learning_rate": 9.59392105263158e-05,
      "loss": 0.439,
      "step": 35446
    },
    {
      "epoch": 1.9849367230372943,
      "grad_norm": 1.4096988439559937,
      "learning_rate": 9.593894736842106e-05,
      "loss": 0.605,
      "step": 35447
    },
    {
      "epoch": 1.9849927203494233,
      "grad_norm": 1.220782995223999,
      "learning_rate": 9.593868421052632e-05,
      "loss": 0.4369,
      "step": 35448
    },
    {
      "epoch": 1.9850487176615523,
      "grad_norm": 1.346310019493103,
      "learning_rate": 9.593842105263159e-05,
      "loss": 0.5336,
      "step": 35449
    },
    {
      "epoch": 1.9851047149736814,
      "grad_norm": 1.2538485527038574,
      "learning_rate": 9.593815789473685e-05,
      "loss": 0.4565,
      "step": 35450
    },
    {
      "epoch": 1.9851607122858104,
      "grad_norm": 1.2544595003128052,
      "learning_rate": 9.59378947368421e-05,
      "loss": 0.4645,
      "step": 35451
    },
    {
      "epoch": 1.9852167095979394,
      "grad_norm": 1.2146624326705933,
      "learning_rate": 9.593763157894737e-05,
      "loss": 0.3921,
      "step": 35452
    },
    {
      "epoch": 1.9852727069100684,
      "grad_norm": 1.038583517074585,
      "learning_rate": 9.593736842105263e-05,
      "loss": 0.3153,
      "step": 35453
    },
    {
      "epoch": 1.9853287042221974,
      "grad_norm": 1.4133057594299316,
      "learning_rate": 9.59371052631579e-05,
      "loss": 0.4592,
      "step": 35454
    },
    {
      "epoch": 1.9853847015343264,
      "grad_norm": 1.3205456733703613,
      "learning_rate": 9.593684210526316e-05,
      "loss": 0.3537,
      "step": 35455
    },
    {
      "epoch": 1.9854406988464555,
      "grad_norm": 1.0945783853530884,
      "learning_rate": 9.593657894736842e-05,
      "loss": 0.4419,
      "step": 35456
    },
    {
      "epoch": 1.9854966961585845,
      "grad_norm": 1.373948097229004,
      "learning_rate": 9.593631578947368e-05,
      "loss": 0.5903,
      "step": 35457
    },
    {
      "epoch": 1.9855526934707135,
      "grad_norm": 1.7060059309005737,
      "learning_rate": 9.593605263157896e-05,
      "loss": 0.5437,
      "step": 35458
    },
    {
      "epoch": 1.9856086907828425,
      "grad_norm": 3.1456387042999268,
      "learning_rate": 9.593578947368422e-05,
      "loss": 0.5352,
      "step": 35459
    },
    {
      "epoch": 1.9856646880949715,
      "grad_norm": 1.105337142944336,
      "learning_rate": 9.593552631578948e-05,
      "loss": 0.3794,
      "step": 35460
    },
    {
      "epoch": 1.9857206854071006,
      "grad_norm": 1.2466400861740112,
      "learning_rate": 9.593526315789474e-05,
      "loss": 0.4434,
      "step": 35461
    },
    {
      "epoch": 1.9857766827192296,
      "grad_norm": 1.237947702407837,
      "learning_rate": 9.593500000000001e-05,
      "loss": 0.3995,
      "step": 35462
    },
    {
      "epoch": 1.9858326800313586,
      "grad_norm": 1.208706259727478,
      "learning_rate": 9.593473684210527e-05,
      "loss": 0.4239,
      "step": 35463
    },
    {
      "epoch": 1.9858886773434876,
      "grad_norm": 1.50117027759552,
      "learning_rate": 9.593447368421053e-05,
      "loss": 0.4797,
      "step": 35464
    },
    {
      "epoch": 1.9859446746556166,
      "grad_norm": 1.306710958480835,
      "learning_rate": 9.593421052631579e-05,
      "loss": 0.503,
      "step": 35465
    },
    {
      "epoch": 1.9860006719677457,
      "grad_norm": 4.107534408569336,
      "learning_rate": 9.593394736842106e-05,
      "loss": 0.7216,
      "step": 35466
    },
    {
      "epoch": 1.9860566692798747,
      "grad_norm": 1.4712620973587036,
      "learning_rate": 9.593368421052632e-05,
      "loss": 0.5996,
      "step": 35467
    },
    {
      "epoch": 1.9861126665920037,
      "grad_norm": 1.2280992269515991,
      "learning_rate": 9.59334210526316e-05,
      "loss": 0.4118,
      "step": 35468
    },
    {
      "epoch": 1.9861686639041327,
      "grad_norm": 1.087149977684021,
      "learning_rate": 9.593315789473684e-05,
      "loss": 0.3858,
      "step": 35469
    },
    {
      "epoch": 1.9862246612162617,
      "grad_norm": 1.139447569847107,
      "learning_rate": 9.59328947368421e-05,
      "loss": 0.5448,
      "step": 35470
    },
    {
      "epoch": 1.9862806585283908,
      "grad_norm": 1.401938557624817,
      "learning_rate": 9.593263157894737e-05,
      "loss": 0.4129,
      "step": 35471
    },
    {
      "epoch": 1.9863366558405198,
      "grad_norm": 1.299832820892334,
      "learning_rate": 9.593236842105263e-05,
      "loss": 0.4094,
      "step": 35472
    },
    {
      "epoch": 1.9863926531526488,
      "grad_norm": 1.196219801902771,
      "learning_rate": 9.59321052631579e-05,
      "loss": 0.4212,
      "step": 35473
    },
    {
      "epoch": 1.9864486504647778,
      "grad_norm": 1.2520298957824707,
      "learning_rate": 9.593184210526315e-05,
      "loss": 0.5987,
      "step": 35474
    },
    {
      "epoch": 1.9865046477769068,
      "grad_norm": 1.2894660234451294,
      "learning_rate": 9.593157894736843e-05,
      "loss": 0.4341,
      "step": 35475
    },
    {
      "epoch": 1.9865606450890358,
      "grad_norm": 1.1469789743423462,
      "learning_rate": 9.593131578947369e-05,
      "loss": 0.4556,
      "step": 35476
    },
    {
      "epoch": 1.9866166424011649,
      "grad_norm": 1.070273995399475,
      "learning_rate": 9.593105263157896e-05,
      "loss": 0.404,
      "step": 35477
    },
    {
      "epoch": 1.9866726397132939,
      "grad_norm": 1.1394150257110596,
      "learning_rate": 9.593078947368422e-05,
      "loss": 0.5053,
      "step": 35478
    },
    {
      "epoch": 1.986728637025423,
      "grad_norm": 1.0825096368789673,
      "learning_rate": 9.593052631578948e-05,
      "loss": 0.3979,
      "step": 35479
    },
    {
      "epoch": 1.986784634337552,
      "grad_norm": 1.304796814918518,
      "learning_rate": 9.593026315789474e-05,
      "loss": 0.4139,
      "step": 35480
    },
    {
      "epoch": 1.986840631649681,
      "grad_norm": 1.3459256887435913,
      "learning_rate": 9.593000000000001e-05,
      "loss": 0.5829,
      "step": 35481
    },
    {
      "epoch": 1.98689662896181,
      "grad_norm": 1.2086055278778076,
      "learning_rate": 9.592973684210527e-05,
      "loss": 0.4489,
      "step": 35482
    },
    {
      "epoch": 1.986952626273939,
      "grad_norm": 1.0581413507461548,
      "learning_rate": 9.592947368421053e-05,
      "loss": 0.5134,
      "step": 35483
    },
    {
      "epoch": 1.987008623586068,
      "grad_norm": 1.3694082498550415,
      "learning_rate": 9.592921052631579e-05,
      "loss": 0.5376,
      "step": 35484
    },
    {
      "epoch": 1.987064620898197,
      "grad_norm": 1.1337991952896118,
      "learning_rate": 9.592894736842106e-05,
      "loss": 0.4801,
      "step": 35485
    },
    {
      "epoch": 1.987120618210326,
      "grad_norm": 1.1556999683380127,
      "learning_rate": 9.592868421052632e-05,
      "loss": 0.4499,
      "step": 35486
    },
    {
      "epoch": 1.987176615522455,
      "grad_norm": 1.1705234050750732,
      "learning_rate": 9.592842105263158e-05,
      "loss": 0.3402,
      "step": 35487
    },
    {
      "epoch": 1.987232612834584,
      "grad_norm": 1.2650971412658691,
      "learning_rate": 9.592815789473684e-05,
      "loss": 0.4602,
      "step": 35488
    },
    {
      "epoch": 1.987288610146713,
      "grad_norm": 1.4428532123565674,
      "learning_rate": 9.59278947368421e-05,
      "loss": 0.4304,
      "step": 35489
    },
    {
      "epoch": 1.9873446074588421,
      "grad_norm": 1.4514931440353394,
      "learning_rate": 9.592763157894738e-05,
      "loss": 0.4974,
      "step": 35490
    },
    {
      "epoch": 1.9874006047709711,
      "grad_norm": 1.5715280771255493,
      "learning_rate": 9.592736842105264e-05,
      "loss": 0.5744,
      "step": 35491
    },
    {
      "epoch": 1.9874566020831002,
      "grad_norm": 1.3635389804840088,
      "learning_rate": 9.59271052631579e-05,
      "loss": 0.3976,
      "step": 35492
    },
    {
      "epoch": 1.9875125993952292,
      "grad_norm": 1.2272080183029175,
      "learning_rate": 9.592684210526316e-05,
      "loss": 0.4523,
      "step": 35493
    },
    {
      "epoch": 1.9875685967073582,
      "grad_norm": 1.288355827331543,
      "learning_rate": 9.592657894736843e-05,
      "loss": 0.4742,
      "step": 35494
    },
    {
      "epoch": 1.9876245940194872,
      "grad_norm": 1.020338773727417,
      "learning_rate": 9.592631578947369e-05,
      "loss": 0.4291,
      "step": 35495
    },
    {
      "epoch": 1.9876805913316162,
      "grad_norm": 1.267419457435608,
      "learning_rate": 9.592605263157896e-05,
      "loss": 0.4002,
      "step": 35496
    },
    {
      "epoch": 1.9877365886437453,
      "grad_norm": 1.3968697786331177,
      "learning_rate": 9.592578947368421e-05,
      "loss": 0.5103,
      "step": 35497
    },
    {
      "epoch": 1.9877925859558743,
      "grad_norm": 1.4301570653915405,
      "learning_rate": 9.592552631578948e-05,
      "loss": 0.4989,
      "step": 35498
    },
    {
      "epoch": 1.9878485832680033,
      "grad_norm": 1.1554651260375977,
      "learning_rate": 9.592526315789474e-05,
      "loss": 0.5512,
      "step": 35499
    },
    {
      "epoch": 1.9879045805801323,
      "grad_norm": 1.2478835582733154,
      "learning_rate": 9.592500000000001e-05,
      "loss": 0.4239,
      "step": 35500
    },
    {
      "epoch": 1.9879605778922613,
      "grad_norm": 1.1728347539901733,
      "learning_rate": 9.592473684210527e-05,
      "loss": 0.4987,
      "step": 35501
    },
    {
      "epoch": 1.9880165752043903,
      "grad_norm": 1.4364714622497559,
      "learning_rate": 9.592447368421053e-05,
      "loss": 0.3506,
      "step": 35502
    },
    {
      "epoch": 1.9880725725165194,
      "grad_norm": 1.3623409271240234,
      "learning_rate": 9.59242105263158e-05,
      "loss": 0.6151,
      "step": 35503
    },
    {
      "epoch": 1.9881285698286484,
      "grad_norm": 1.1457279920578003,
      "learning_rate": 9.592394736842105e-05,
      "loss": 0.4293,
      "step": 35504
    },
    {
      "epoch": 1.9881845671407774,
      "grad_norm": 1.4999864101409912,
      "learning_rate": 9.592368421052633e-05,
      "loss": 0.5702,
      "step": 35505
    },
    {
      "epoch": 1.9882405644529064,
      "grad_norm": 1.3519184589385986,
      "learning_rate": 9.592342105263157e-05,
      "loss": 0.443,
      "step": 35506
    },
    {
      "epoch": 1.9882965617650354,
      "grad_norm": 2.0611536502838135,
      "learning_rate": 9.592315789473685e-05,
      "loss": 0.5067,
      "step": 35507
    },
    {
      "epoch": 1.9883525590771645,
      "grad_norm": 1.4516867399215698,
      "learning_rate": 9.59228947368421e-05,
      "loss": 0.5665,
      "step": 35508
    },
    {
      "epoch": 1.9884085563892935,
      "grad_norm": 1.0792677402496338,
      "learning_rate": 9.592263157894738e-05,
      "loss": 0.3083,
      "step": 35509
    },
    {
      "epoch": 1.9884645537014223,
      "grad_norm": 1.0905338525772095,
      "learning_rate": 9.592236842105264e-05,
      "loss": 0.4128,
      "step": 35510
    },
    {
      "epoch": 1.9885205510135513,
      "grad_norm": 1.0899282693862915,
      "learning_rate": 9.59221052631579e-05,
      "loss": 0.4503,
      "step": 35511
    },
    {
      "epoch": 1.9885765483256803,
      "grad_norm": 1.273254156112671,
      "learning_rate": 9.592184210526316e-05,
      "loss": 0.4026,
      "step": 35512
    },
    {
      "epoch": 1.9886325456378093,
      "grad_norm": 1.2546851634979248,
      "learning_rate": 9.592157894736843e-05,
      "loss": 0.4344,
      "step": 35513
    },
    {
      "epoch": 1.9886885429499384,
      "grad_norm": 1.2890563011169434,
      "learning_rate": 9.592131578947369e-05,
      "loss": 0.5002,
      "step": 35514
    },
    {
      "epoch": 1.9887445402620674,
      "grad_norm": 1.2466799020767212,
      "learning_rate": 9.592105263157895e-05,
      "loss": 0.4146,
      "step": 35515
    },
    {
      "epoch": 1.9888005375741964,
      "grad_norm": 1.6226491928100586,
      "learning_rate": 9.592078947368421e-05,
      "loss": 0.4102,
      "step": 35516
    },
    {
      "epoch": 1.9888565348863254,
      "grad_norm": 1.557784914970398,
      "learning_rate": 9.592052631578948e-05,
      "loss": 0.4721,
      "step": 35517
    },
    {
      "epoch": 1.9889125321984544,
      "grad_norm": 1.1472299098968506,
      "learning_rate": 9.592026315789474e-05,
      "loss": 0.4538,
      "step": 35518
    },
    {
      "epoch": 1.9889685295105834,
      "grad_norm": 1.212196707725525,
      "learning_rate": 9.592e-05,
      "loss": 0.4624,
      "step": 35519
    },
    {
      "epoch": 1.9890245268227125,
      "grad_norm": 1.2951369285583496,
      "learning_rate": 9.591973684210526e-05,
      "loss": 0.361,
      "step": 35520
    },
    {
      "epoch": 1.9890805241348415,
      "grad_norm": 1.0945374965667725,
      "learning_rate": 9.591947368421052e-05,
      "loss": 0.4849,
      "step": 35521
    },
    {
      "epoch": 1.9891365214469705,
      "grad_norm": 1.3309515714645386,
      "learning_rate": 9.59192105263158e-05,
      "loss": 0.6063,
      "step": 35522
    },
    {
      "epoch": 1.9891925187590995,
      "grad_norm": 1.2102982997894287,
      "learning_rate": 9.591894736842106e-05,
      "loss": 0.4288,
      "step": 35523
    },
    {
      "epoch": 1.9892485160712285,
      "grad_norm": 1.8596972227096558,
      "learning_rate": 9.591868421052632e-05,
      "loss": 0.5483,
      "step": 35524
    },
    {
      "epoch": 1.9893045133833576,
      "grad_norm": 1.093278408050537,
      "learning_rate": 9.591842105263158e-05,
      "loss": 0.3525,
      "step": 35525
    },
    {
      "epoch": 1.9893605106954866,
      "grad_norm": 1.171967625617981,
      "learning_rate": 9.591815789473685e-05,
      "loss": 0.4379,
      "step": 35526
    },
    {
      "epoch": 1.9894165080076156,
      "grad_norm": 1.3306617736816406,
      "learning_rate": 9.591789473684211e-05,
      "loss": 0.5521,
      "step": 35527
    },
    {
      "epoch": 1.9894725053197446,
      "grad_norm": 1.178497314453125,
      "learning_rate": 9.591763157894738e-05,
      "loss": 0.3866,
      "step": 35528
    },
    {
      "epoch": 1.9895285026318736,
      "grad_norm": 1.358841061592102,
      "learning_rate": 9.591736842105263e-05,
      "loss": 0.3975,
      "step": 35529
    },
    {
      "epoch": 1.9895844999440027,
      "grad_norm": 1.3383830785751343,
      "learning_rate": 9.59171052631579e-05,
      "loss": 0.4068,
      "step": 35530
    },
    {
      "epoch": 1.9896404972561317,
      "grad_norm": 1.126112461090088,
      "learning_rate": 9.591684210526316e-05,
      "loss": 0.4247,
      "step": 35531
    },
    {
      "epoch": 1.9896964945682607,
      "grad_norm": 1.2307801246643066,
      "learning_rate": 9.591657894736843e-05,
      "loss": 0.4971,
      "step": 35532
    },
    {
      "epoch": 1.9897524918803897,
      "grad_norm": 1.073792815208435,
      "learning_rate": 9.59163157894737e-05,
      "loss": 0.4611,
      "step": 35533
    },
    {
      "epoch": 1.9898084891925187,
      "grad_norm": 1.284218192100525,
      "learning_rate": 9.591605263157895e-05,
      "loss": 0.3711,
      "step": 35534
    },
    {
      "epoch": 1.9898644865046478,
      "grad_norm": 2.1291611194610596,
      "learning_rate": 9.591578947368421e-05,
      "loss": 0.5127,
      "step": 35535
    },
    {
      "epoch": 1.9899204838167768,
      "grad_norm": 0.9030032753944397,
      "learning_rate": 9.591552631578949e-05,
      "loss": 0.208,
      "step": 35536
    },
    {
      "epoch": 1.9899764811289058,
      "grad_norm": 1.2215075492858887,
      "learning_rate": 9.591526315789475e-05,
      "loss": 0.35,
      "step": 35537
    },
    {
      "epoch": 1.9900324784410348,
      "grad_norm": 1.2426862716674805,
      "learning_rate": 9.5915e-05,
      "loss": 0.3986,
      "step": 35538
    },
    {
      "epoch": 1.9900884757531638,
      "grad_norm": 1.435858130455017,
      "learning_rate": 9.591473684210527e-05,
      "loss": 0.4092,
      "step": 35539
    },
    {
      "epoch": 1.9901444730652929,
      "grad_norm": 1.5430365800857544,
      "learning_rate": 9.591447368421053e-05,
      "loss": 0.5414,
      "step": 35540
    },
    {
      "epoch": 1.9902004703774219,
      "grad_norm": 1.3427220582962036,
      "learning_rate": 9.59142105263158e-05,
      "loss": 0.444,
      "step": 35541
    },
    {
      "epoch": 1.990256467689551,
      "grad_norm": 1.5588865280151367,
      "learning_rate": 9.591394736842106e-05,
      "loss": 0.6043,
      "step": 35542
    },
    {
      "epoch": 1.99031246500168,
      "grad_norm": 1.4000496864318848,
      "learning_rate": 9.591368421052632e-05,
      "loss": 0.4929,
      "step": 35543
    },
    {
      "epoch": 1.990368462313809,
      "grad_norm": 1.0791112184524536,
      "learning_rate": 9.591342105263158e-05,
      "loss": 0.3904,
      "step": 35544
    },
    {
      "epoch": 1.990424459625938,
      "grad_norm": 1.1854305267333984,
      "learning_rate": 9.591315789473685e-05,
      "loss": 0.3382,
      "step": 35545
    },
    {
      "epoch": 1.990480456938067,
      "grad_norm": 1.3216811418533325,
      "learning_rate": 9.591289473684211e-05,
      "loss": 0.5435,
      "step": 35546
    },
    {
      "epoch": 1.990536454250196,
      "grad_norm": 1.1134005784988403,
      "learning_rate": 9.591263157894737e-05,
      "loss": 0.3386,
      "step": 35547
    },
    {
      "epoch": 1.990592451562325,
      "grad_norm": 1.3041775226593018,
      "learning_rate": 9.591236842105263e-05,
      "loss": 0.5231,
      "step": 35548
    },
    {
      "epoch": 1.990648448874454,
      "grad_norm": 1.1259886026382446,
      "learning_rate": 9.59121052631579e-05,
      "loss": 0.4198,
      "step": 35549
    },
    {
      "epoch": 1.990704446186583,
      "grad_norm": 1.188585638999939,
      "learning_rate": 9.591184210526316e-05,
      "loss": 0.4411,
      "step": 35550
    },
    {
      "epoch": 1.990760443498712,
      "grad_norm": 1.2205523252487183,
      "learning_rate": 9.591157894736844e-05,
      "loss": 0.3702,
      "step": 35551
    },
    {
      "epoch": 1.990816440810841,
      "grad_norm": 1.3695698976516724,
      "learning_rate": 9.591131578947368e-05,
      "loss": 0.4383,
      "step": 35552
    },
    {
      "epoch": 1.99087243812297,
      "grad_norm": 1.3751276731491089,
      "learning_rate": 9.591105263157896e-05,
      "loss": 0.4639,
      "step": 35553
    },
    {
      "epoch": 1.9909284354350991,
      "grad_norm": 1.538450837135315,
      "learning_rate": 9.591078947368422e-05,
      "loss": 0.4652,
      "step": 35554
    },
    {
      "epoch": 1.9909844327472281,
      "grad_norm": 1.5518817901611328,
      "learning_rate": 9.591052631578948e-05,
      "loss": 0.5109,
      "step": 35555
    },
    {
      "epoch": 1.9910404300593572,
      "grad_norm": 1.3193330764770508,
      "learning_rate": 9.591026315789475e-05,
      "loss": 0.4631,
      "step": 35556
    },
    {
      "epoch": 1.9910964273714862,
      "grad_norm": 1.244462490081787,
      "learning_rate": 9.591e-05,
      "loss": 0.422,
      "step": 35557
    },
    {
      "epoch": 1.9911524246836152,
      "grad_norm": 1.370192050933838,
      "learning_rate": 9.590973684210527e-05,
      "loss": 0.4698,
      "step": 35558
    },
    {
      "epoch": 1.9912084219957442,
      "grad_norm": 1.4153363704681396,
      "learning_rate": 9.590947368421053e-05,
      "loss": 0.5244,
      "step": 35559
    },
    {
      "epoch": 1.9912644193078732,
      "grad_norm": 1.6399774551391602,
      "learning_rate": 9.59092105263158e-05,
      "loss": 0.4686,
      "step": 35560
    },
    {
      "epoch": 1.9913204166200023,
      "grad_norm": 1.1923521757125854,
      "learning_rate": 9.590894736842105e-05,
      "loss": 0.3494,
      "step": 35561
    },
    {
      "epoch": 1.9913764139321313,
      "grad_norm": 1.447489857673645,
      "learning_rate": 9.590868421052632e-05,
      "loss": 0.4389,
      "step": 35562
    },
    {
      "epoch": 1.9914324112442603,
      "grad_norm": 1.191446304321289,
      "learning_rate": 9.590842105263158e-05,
      "loss": 0.4822,
      "step": 35563
    },
    {
      "epoch": 1.9914884085563893,
      "grad_norm": 2.585723876953125,
      "learning_rate": 9.590815789473685e-05,
      "loss": 0.5502,
      "step": 35564
    },
    {
      "epoch": 1.9915444058685183,
      "grad_norm": 1.7500213384628296,
      "learning_rate": 9.590789473684211e-05,
      "loss": 0.3284,
      "step": 35565
    },
    {
      "epoch": 1.9916004031806473,
      "grad_norm": 1.3805952072143555,
      "learning_rate": 9.590763157894737e-05,
      "loss": 0.5546,
      "step": 35566
    },
    {
      "epoch": 1.9916564004927764,
      "grad_norm": 1.2405415773391724,
      "learning_rate": 9.590736842105263e-05,
      "loss": 0.4891,
      "step": 35567
    },
    {
      "epoch": 1.9917123978049054,
      "grad_norm": 1.6159594058990479,
      "learning_rate": 9.59071052631579e-05,
      "loss": 0.5214,
      "step": 35568
    },
    {
      "epoch": 1.9917683951170344,
      "grad_norm": 1.3045800924301147,
      "learning_rate": 9.590684210526317e-05,
      "loss": 0.5356,
      "step": 35569
    },
    {
      "epoch": 1.9918243924291634,
      "grad_norm": 1.3108443021774292,
      "learning_rate": 9.590657894736843e-05,
      "loss": 0.4313,
      "step": 35570
    },
    {
      "epoch": 1.9918803897412924,
      "grad_norm": 1.1155539751052856,
      "learning_rate": 9.590631578947369e-05,
      "loss": 0.3851,
      "step": 35571
    },
    {
      "epoch": 1.9919363870534215,
      "grad_norm": 1.0812451839447021,
      "learning_rate": 9.590605263157894e-05,
      "loss": 0.4684,
      "step": 35572
    },
    {
      "epoch": 1.9919923843655505,
      "grad_norm": 1.2073543071746826,
      "learning_rate": 9.590578947368422e-05,
      "loss": 0.541,
      "step": 35573
    },
    {
      "epoch": 1.9920483816776795,
      "grad_norm": 1.2165262699127197,
      "learning_rate": 9.590552631578948e-05,
      "loss": 0.3822,
      "step": 35574
    },
    {
      "epoch": 1.9921043789898085,
      "grad_norm": 1.5016649961471558,
      "learning_rate": 9.590526315789474e-05,
      "loss": 0.5133,
      "step": 35575
    },
    {
      "epoch": 1.9921603763019375,
      "grad_norm": 1.2360635995864868,
      "learning_rate": 9.5905e-05,
      "loss": 0.4435,
      "step": 35576
    },
    {
      "epoch": 1.9922163736140666,
      "grad_norm": 1.1395368576049805,
      "learning_rate": 9.590473684210527e-05,
      "loss": 0.4172,
      "step": 35577
    },
    {
      "epoch": 1.9922723709261956,
      "grad_norm": 1.3615334033966064,
      "learning_rate": 9.590447368421053e-05,
      "loss": 0.4527,
      "step": 35578
    },
    {
      "epoch": 1.9923283682383246,
      "grad_norm": 1.1413980722427368,
      "learning_rate": 9.590421052631579e-05,
      "loss": 0.4149,
      "step": 35579
    },
    {
      "epoch": 1.9923843655504536,
      "grad_norm": 1.5299428701400757,
      "learning_rate": 9.590394736842105e-05,
      "loss": 0.4049,
      "step": 35580
    },
    {
      "epoch": 1.9924403628625826,
      "grad_norm": 1.1139276027679443,
      "learning_rate": 9.590368421052632e-05,
      "loss": 0.5362,
      "step": 35581
    },
    {
      "epoch": 1.9924963601747117,
      "grad_norm": 1.3127061128616333,
      "learning_rate": 9.590342105263158e-05,
      "loss": 0.4276,
      "step": 35582
    },
    {
      "epoch": 1.9925523574868407,
      "grad_norm": 3.077648878097534,
      "learning_rate": 9.590315789473686e-05,
      "loss": 0.3951,
      "step": 35583
    },
    {
      "epoch": 1.9926083547989697,
      "grad_norm": 1.3226358890533447,
      "learning_rate": 9.59028947368421e-05,
      "loss": 0.3977,
      "step": 35584
    },
    {
      "epoch": 1.9926643521110987,
      "grad_norm": 1.651672601699829,
      "learning_rate": 9.590263157894738e-05,
      "loss": 0.4528,
      "step": 35585
    },
    {
      "epoch": 1.9927203494232277,
      "grad_norm": 1.2801048755645752,
      "learning_rate": 9.590236842105264e-05,
      "loss": 0.401,
      "step": 35586
    },
    {
      "epoch": 1.9927763467353568,
      "grad_norm": 1.3332183361053467,
      "learning_rate": 9.590210526315791e-05,
      "loss": 0.3641,
      "step": 35587
    },
    {
      "epoch": 1.9928323440474858,
      "grad_norm": 1.0959124565124512,
      "learning_rate": 9.590184210526317e-05,
      "loss": 0.3598,
      "step": 35588
    },
    {
      "epoch": 1.9928883413596148,
      "grad_norm": 1.2634996175765991,
      "learning_rate": 9.590157894736843e-05,
      "loss": 0.4075,
      "step": 35589
    },
    {
      "epoch": 1.9929443386717438,
      "grad_norm": 1.2024871110916138,
      "learning_rate": 9.590131578947369e-05,
      "loss": 0.4666,
      "step": 35590
    },
    {
      "epoch": 1.9930003359838728,
      "grad_norm": 1.1232930421829224,
      "learning_rate": 9.590105263157895e-05,
      "loss": 0.3872,
      "step": 35591
    },
    {
      "epoch": 1.9930563332960016,
      "grad_norm": 3.0282576084136963,
      "learning_rate": 9.590078947368422e-05,
      "loss": 0.3748,
      "step": 35592
    },
    {
      "epoch": 1.9931123306081306,
      "grad_norm": 1.2452056407928467,
      "learning_rate": 9.590052631578948e-05,
      "loss": 0.3512,
      "step": 35593
    },
    {
      "epoch": 1.9931683279202597,
      "grad_norm": 1.3730556964874268,
      "learning_rate": 9.590026315789474e-05,
      "loss": 0.603,
      "step": 35594
    },
    {
      "epoch": 1.9932243252323887,
      "grad_norm": 1.37942373752594,
      "learning_rate": 9.59e-05,
      "loss": 0.4558,
      "step": 35595
    },
    {
      "epoch": 1.9932803225445177,
      "grad_norm": 1.5861713886260986,
      "learning_rate": 9.589973684210527e-05,
      "loss": 0.5294,
      "step": 35596
    },
    {
      "epoch": 1.9933363198566467,
      "grad_norm": 1.404109001159668,
      "learning_rate": 9.589947368421053e-05,
      "loss": 0.447,
      "step": 35597
    },
    {
      "epoch": 1.9933923171687757,
      "grad_norm": 1.2659451961517334,
      "learning_rate": 9.589921052631579e-05,
      "loss": 0.5307,
      "step": 35598
    },
    {
      "epoch": 1.9934483144809048,
      "grad_norm": 1.0489815473556519,
      "learning_rate": 9.589894736842105e-05,
      "loss": 0.355,
      "step": 35599
    },
    {
      "epoch": 1.9935043117930338,
      "grad_norm": 1.2136744260787964,
      "learning_rate": 9.589868421052633e-05,
      "loss": 0.441,
      "step": 35600
    },
    {
      "epoch": 1.9935603091051628,
      "grad_norm": 1.2241686582565308,
      "learning_rate": 9.589842105263159e-05,
      "loss": 0.3985,
      "step": 35601
    },
    {
      "epoch": 1.9936163064172918,
      "grad_norm": 7.419262886047363,
      "learning_rate": 9.589815789473685e-05,
      "loss": 0.5303,
      "step": 35602
    },
    {
      "epoch": 1.9936723037294208,
      "grad_norm": 1.1942400932312012,
      "learning_rate": 9.58978947368421e-05,
      "loss": 0.2767,
      "step": 35603
    },
    {
      "epoch": 1.9937283010415499,
      "grad_norm": 1.4366904497146606,
      "learning_rate": 9.589763157894738e-05,
      "loss": 0.4674,
      "step": 35604
    },
    {
      "epoch": 1.9937842983536789,
      "grad_norm": 1.48849618434906,
      "learning_rate": 9.589736842105264e-05,
      "loss": 0.5505,
      "step": 35605
    },
    {
      "epoch": 1.993840295665808,
      "grad_norm": 1.5724461078643799,
      "learning_rate": 9.589710526315791e-05,
      "loss": 0.5464,
      "step": 35606
    },
    {
      "epoch": 1.993896292977937,
      "grad_norm": 1.2811603546142578,
      "learning_rate": 9.589684210526316e-05,
      "loss": 0.4419,
      "step": 35607
    },
    {
      "epoch": 1.993952290290066,
      "grad_norm": 1.5148351192474365,
      "learning_rate": 9.589657894736842e-05,
      "loss": 0.3906,
      "step": 35608
    },
    {
      "epoch": 1.994008287602195,
      "grad_norm": 1.3868930339813232,
      "learning_rate": 9.589631578947369e-05,
      "loss": 0.3752,
      "step": 35609
    },
    {
      "epoch": 1.994064284914324,
      "grad_norm": 1.1747761964797974,
      "learning_rate": 9.589605263157895e-05,
      "loss": 0.4691,
      "step": 35610
    },
    {
      "epoch": 1.994120282226453,
      "grad_norm": 1.198513150215149,
      "learning_rate": 9.589578947368421e-05,
      "loss": 0.4682,
      "step": 35611
    },
    {
      "epoch": 1.994176279538582,
      "grad_norm": 1.2227568626403809,
      "learning_rate": 9.589552631578947e-05,
      "loss": 0.4135,
      "step": 35612
    },
    {
      "epoch": 1.994232276850711,
      "grad_norm": 1.4125950336456299,
      "learning_rate": 9.589526315789474e-05,
      "loss": 0.5474,
      "step": 35613
    },
    {
      "epoch": 1.99428827416284,
      "grad_norm": 1.240922212600708,
      "learning_rate": 9.5895e-05,
      "loss": 0.5191,
      "step": 35614
    },
    {
      "epoch": 1.994344271474969,
      "grad_norm": 1.2878451347351074,
      "learning_rate": 9.589473684210528e-05,
      "loss": 0.5735,
      "step": 35615
    },
    {
      "epoch": 1.994400268787098,
      "grad_norm": 1.459401249885559,
      "learning_rate": 9.589447368421052e-05,
      "loss": 0.5433,
      "step": 35616
    },
    {
      "epoch": 1.994456266099227,
      "grad_norm": 1.2540624141693115,
      "learning_rate": 9.58942105263158e-05,
      "loss": 0.4477,
      "step": 35617
    },
    {
      "epoch": 1.9945122634113561,
      "grad_norm": 1.094946026802063,
      "learning_rate": 9.589394736842106e-05,
      "loss": 0.354,
      "step": 35618
    },
    {
      "epoch": 1.9945682607234851,
      "grad_norm": 1.1434732675552368,
      "learning_rate": 9.589368421052633e-05,
      "loss": 0.4327,
      "step": 35619
    },
    {
      "epoch": 1.9946242580356142,
      "grad_norm": 1.153523564338684,
      "learning_rate": 9.589342105263159e-05,
      "loss": 0.3438,
      "step": 35620
    },
    {
      "epoch": 1.9946802553477432,
      "grad_norm": 1.1065376996994019,
      "learning_rate": 9.589315789473685e-05,
      "loss": 0.4217,
      "step": 35621
    },
    {
      "epoch": 1.9947362526598722,
      "grad_norm": 1.3761993646621704,
      "learning_rate": 9.589289473684211e-05,
      "loss": 0.4477,
      "step": 35622
    },
    {
      "epoch": 1.9947922499720012,
      "grad_norm": 1.2070022821426392,
      "learning_rate": 9.589263157894738e-05,
      "loss": 0.4723,
      "step": 35623
    },
    {
      "epoch": 1.9948482472841302,
      "grad_norm": 1.2414604425430298,
      "learning_rate": 9.589236842105264e-05,
      "loss": 0.422,
      "step": 35624
    },
    {
      "epoch": 1.9949042445962593,
      "grad_norm": 1.2085981369018555,
      "learning_rate": 9.58921052631579e-05,
      "loss": 0.531,
      "step": 35625
    },
    {
      "epoch": 1.9949602419083883,
      "grad_norm": 1.2342300415039062,
      "learning_rate": 9.589184210526316e-05,
      "loss": 0.4963,
      "step": 35626
    },
    {
      "epoch": 1.9950162392205173,
      "grad_norm": 1.2143809795379639,
      "learning_rate": 9.589157894736842e-05,
      "loss": 0.512,
      "step": 35627
    },
    {
      "epoch": 1.9950722365326463,
      "grad_norm": 1.24850332736969,
      "learning_rate": 9.589131578947369e-05,
      "loss": 0.4538,
      "step": 35628
    },
    {
      "epoch": 1.9951282338447753,
      "grad_norm": 1.295248031616211,
      "learning_rate": 9.589105263157895e-05,
      "loss": 0.5132,
      "step": 35629
    },
    {
      "epoch": 1.9951842311569044,
      "grad_norm": 1.2572659254074097,
      "learning_rate": 9.589078947368421e-05,
      "loss": 0.393,
      "step": 35630
    },
    {
      "epoch": 1.9952402284690334,
      "grad_norm": 1.2328392267227173,
      "learning_rate": 9.589052631578947e-05,
      "loss": 0.4205,
      "step": 35631
    },
    {
      "epoch": 1.9952962257811624,
      "grad_norm": 15.718681335449219,
      "learning_rate": 9.589026315789475e-05,
      "loss": 0.4222,
      "step": 35632
    },
    {
      "epoch": 1.9953522230932914,
      "grad_norm": 0.9863313436508179,
      "learning_rate": 9.589e-05,
      "loss": 0.3564,
      "step": 35633
    },
    {
      "epoch": 1.9954082204054204,
      "grad_norm": 1.335634469985962,
      "learning_rate": 9.588973684210526e-05,
      "loss": 0.4485,
      "step": 35634
    },
    {
      "epoch": 1.9954642177175494,
      "grad_norm": 1.133293628692627,
      "learning_rate": 9.588947368421052e-05,
      "loss": 0.3722,
      "step": 35635
    },
    {
      "epoch": 1.9955202150296785,
      "grad_norm": 1.2661166191101074,
      "learning_rate": 9.58892105263158e-05,
      "loss": 0.367,
      "step": 35636
    },
    {
      "epoch": 1.9955762123418075,
      "grad_norm": 1.321421504020691,
      "learning_rate": 9.588894736842106e-05,
      "loss": 0.4103,
      "step": 35637
    },
    {
      "epoch": 1.9956322096539365,
      "grad_norm": 2.1491875648498535,
      "learning_rate": 9.588868421052633e-05,
      "loss": 0.4192,
      "step": 35638
    },
    {
      "epoch": 1.9956882069660655,
      "grad_norm": 1.3580129146575928,
      "learning_rate": 9.588842105263158e-05,
      "loss": 0.4743,
      "step": 35639
    },
    {
      "epoch": 1.9957442042781945,
      "grad_norm": 1.3840094804763794,
      "learning_rate": 9.588815789473685e-05,
      "loss": 0.6083,
      "step": 35640
    },
    {
      "epoch": 1.9958002015903236,
      "grad_norm": 1.2663124799728394,
      "learning_rate": 9.588789473684211e-05,
      "loss": 0.3718,
      "step": 35641
    },
    {
      "epoch": 1.9958561989024526,
      "grad_norm": 1.0742474794387817,
      "learning_rate": 9.588763157894737e-05,
      "loss": 0.411,
      "step": 35642
    },
    {
      "epoch": 1.9959121962145816,
      "grad_norm": 1.621517300605774,
      "learning_rate": 9.588736842105264e-05,
      "loss": 0.402,
      "step": 35643
    },
    {
      "epoch": 1.9959681935267106,
      "grad_norm": 1.3091566562652588,
      "learning_rate": 9.588710526315789e-05,
      "loss": 0.4042,
      "step": 35644
    },
    {
      "epoch": 1.9960241908388396,
      "grad_norm": 1.3030307292938232,
      "learning_rate": 9.588684210526316e-05,
      "loss": 0.4972,
      "step": 35645
    },
    {
      "epoch": 1.9960801881509687,
      "grad_norm": 1.2336479425430298,
      "learning_rate": 9.588657894736842e-05,
      "loss": 0.3389,
      "step": 35646
    },
    {
      "epoch": 1.9961361854630977,
      "grad_norm": 1.6403483152389526,
      "learning_rate": 9.58863157894737e-05,
      "loss": 0.5168,
      "step": 35647
    },
    {
      "epoch": 1.9961921827752267,
      "grad_norm": 1.3678525686264038,
      "learning_rate": 9.588605263157896e-05,
      "loss": 0.3554,
      "step": 35648
    },
    {
      "epoch": 1.9962481800873557,
      "grad_norm": 1.223106026649475,
      "learning_rate": 9.588578947368422e-05,
      "loss": 0.5403,
      "step": 35649
    },
    {
      "epoch": 1.9963041773994847,
      "grad_norm": 1.5367188453674316,
      "learning_rate": 9.588552631578947e-05,
      "loss": 0.5844,
      "step": 35650
    },
    {
      "epoch": 1.9963601747116138,
      "grad_norm": 1.2097409963607788,
      "learning_rate": 9.588526315789475e-05,
      "loss": 0.5129,
      "step": 35651
    },
    {
      "epoch": 1.9964161720237428,
      "grad_norm": 1.184790849685669,
      "learning_rate": 9.588500000000001e-05,
      "loss": 0.55,
      "step": 35652
    },
    {
      "epoch": 1.9964721693358718,
      "grad_norm": 1.5261874198913574,
      "learning_rate": 9.588473684210527e-05,
      "loss": 0.4259,
      "step": 35653
    },
    {
      "epoch": 1.9965281666480008,
      "grad_norm": 1.1492356061935425,
      "learning_rate": 9.588447368421053e-05,
      "loss": 0.5255,
      "step": 35654
    },
    {
      "epoch": 1.9965841639601298,
      "grad_norm": 1.4155291318893433,
      "learning_rate": 9.58842105263158e-05,
      "loss": 0.6317,
      "step": 35655
    },
    {
      "epoch": 1.9966401612722589,
      "grad_norm": 1.1190556287765503,
      "learning_rate": 9.588394736842106e-05,
      "loss": 0.4293,
      "step": 35656
    },
    {
      "epoch": 1.9966961585843879,
      "grad_norm": 1.1112204790115356,
      "learning_rate": 9.588368421052632e-05,
      "loss": 0.4493,
      "step": 35657
    },
    {
      "epoch": 1.996752155896517,
      "grad_norm": 1.1368733644485474,
      "learning_rate": 9.588342105263158e-05,
      "loss": 0.378,
      "step": 35658
    },
    {
      "epoch": 1.996808153208646,
      "grad_norm": 1.2593837976455688,
      "learning_rate": 9.588315789473684e-05,
      "loss": 0.3516,
      "step": 35659
    },
    {
      "epoch": 1.996864150520775,
      "grad_norm": 1.0613576173782349,
      "learning_rate": 9.588289473684211e-05,
      "loss": 0.3938,
      "step": 35660
    },
    {
      "epoch": 1.996920147832904,
      "grad_norm": 1.4086543321609497,
      "learning_rate": 9.588263157894737e-05,
      "loss": 0.596,
      "step": 35661
    },
    {
      "epoch": 1.996976145145033,
      "grad_norm": 1.358246088027954,
      "learning_rate": 9.588236842105263e-05,
      "loss": 0.4761,
      "step": 35662
    },
    {
      "epoch": 1.997032142457162,
      "grad_norm": 1.2722880840301514,
      "learning_rate": 9.588210526315789e-05,
      "loss": 0.3117,
      "step": 35663
    },
    {
      "epoch": 1.997088139769291,
      "grad_norm": 1.9498416185379028,
      "learning_rate": 9.588184210526317e-05,
      "loss": 0.8083,
      "step": 35664
    },
    {
      "epoch": 1.99714413708142,
      "grad_norm": 1.12445068359375,
      "learning_rate": 9.588157894736842e-05,
      "loss": 0.3728,
      "step": 35665
    },
    {
      "epoch": 1.997200134393549,
      "grad_norm": 1.3108285665512085,
      "learning_rate": 9.588131578947368e-05,
      "loss": 0.4414,
      "step": 35666
    },
    {
      "epoch": 1.997256131705678,
      "grad_norm": 1.588301658630371,
      "learning_rate": 9.588105263157894e-05,
      "loss": 0.5313,
      "step": 35667
    },
    {
      "epoch": 1.997312129017807,
      "grad_norm": 1.1827144622802734,
      "learning_rate": 9.588078947368422e-05,
      "loss": 0.3846,
      "step": 35668
    },
    {
      "epoch": 1.997368126329936,
      "grad_norm": 1.1902782917022705,
      "learning_rate": 9.588052631578948e-05,
      "loss": 0.3731,
      "step": 35669
    },
    {
      "epoch": 1.9974241236420651,
      "grad_norm": 1.0918718576431274,
      "learning_rate": 9.588026315789475e-05,
      "loss": 0.4079,
      "step": 35670
    },
    {
      "epoch": 1.9974801209541941,
      "grad_norm": 1.2798343896865845,
      "learning_rate": 9.588e-05,
      "loss": 0.4568,
      "step": 35671
    },
    {
      "epoch": 1.9975361182663232,
      "grad_norm": 1.6583809852600098,
      "learning_rate": 9.587973684210527e-05,
      "loss": 0.5875,
      "step": 35672
    },
    {
      "epoch": 1.9975921155784522,
      "grad_norm": 1.1150529384613037,
      "learning_rate": 9.587947368421053e-05,
      "loss": 0.4319,
      "step": 35673
    },
    {
      "epoch": 1.9976481128905812,
      "grad_norm": 1.0876377820968628,
      "learning_rate": 9.58792105263158e-05,
      "loss": 0.4841,
      "step": 35674
    },
    {
      "epoch": 1.9977041102027102,
      "grad_norm": 1.3036870956420898,
      "learning_rate": 9.587894736842106e-05,
      "loss": 0.4038,
      "step": 35675
    },
    {
      "epoch": 1.9977601075148392,
      "grad_norm": 1.0480951070785522,
      "learning_rate": 9.587868421052631e-05,
      "loss": 0.3024,
      "step": 35676
    },
    {
      "epoch": 1.9978161048269683,
      "grad_norm": 1.4002890586853027,
      "learning_rate": 9.587842105263158e-05,
      "loss": 0.4513,
      "step": 35677
    },
    {
      "epoch": 1.9978721021390973,
      "grad_norm": 1.1184619665145874,
      "learning_rate": 9.587815789473684e-05,
      "loss": 0.5511,
      "step": 35678
    },
    {
      "epoch": 1.9979280994512263,
      "grad_norm": 1.3167643547058105,
      "learning_rate": 9.587789473684212e-05,
      "loss": 0.3597,
      "step": 35679
    },
    {
      "epoch": 1.9979840967633553,
      "grad_norm": 1.3305026292800903,
      "learning_rate": 9.587763157894738e-05,
      "loss": 0.4304,
      "step": 35680
    },
    {
      "epoch": 1.9980400940754843,
      "grad_norm": 1.1540409326553345,
      "learning_rate": 9.587736842105263e-05,
      "loss": 0.4654,
      "step": 35681
    },
    {
      "epoch": 1.9980960913876133,
      "grad_norm": 1.2186146974563599,
      "learning_rate": 9.58771052631579e-05,
      "loss": 0.4228,
      "step": 35682
    },
    {
      "epoch": 1.9981520886997424,
      "grad_norm": 1.2066798210144043,
      "learning_rate": 9.587684210526317e-05,
      "loss": 0.3922,
      "step": 35683
    },
    {
      "epoch": 1.9982080860118714,
      "grad_norm": 1.2881914377212524,
      "learning_rate": 9.587657894736843e-05,
      "loss": 0.6669,
      "step": 35684
    },
    {
      "epoch": 1.9982640833240004,
      "grad_norm": 1.508577585220337,
      "learning_rate": 9.587631578947369e-05,
      "loss": 0.6911,
      "step": 35685
    },
    {
      "epoch": 1.9983200806361294,
      "grad_norm": 1.2731300592422485,
      "learning_rate": 9.587605263157895e-05,
      "loss": 0.4137,
      "step": 35686
    },
    {
      "epoch": 1.9983760779482584,
      "grad_norm": 1.101730465888977,
      "learning_rate": 9.587578947368422e-05,
      "loss": 0.3451,
      "step": 35687
    },
    {
      "epoch": 1.9984320752603875,
      "grad_norm": 1.026218295097351,
      "learning_rate": 9.587552631578948e-05,
      "loss": 0.3838,
      "step": 35688
    },
    {
      "epoch": 1.9984880725725165,
      "grad_norm": 1.415786623954773,
      "learning_rate": 9.587526315789474e-05,
      "loss": 0.4131,
      "step": 35689
    },
    {
      "epoch": 1.9985440698846455,
      "grad_norm": 1.2542513608932495,
      "learning_rate": 9.5875e-05,
      "loss": 0.4029,
      "step": 35690
    },
    {
      "epoch": 1.9986000671967745,
      "grad_norm": 1.2519434690475464,
      "learning_rate": 9.587473684210527e-05,
      "loss": 0.4694,
      "step": 35691
    },
    {
      "epoch": 1.9986560645089035,
      "grad_norm": 1.3884254693984985,
      "learning_rate": 9.587447368421053e-05,
      "loss": 0.5692,
      "step": 35692
    },
    {
      "epoch": 1.9987120618210326,
      "grad_norm": 1.1429213285446167,
      "learning_rate": 9.587421052631579e-05,
      "loss": 0.3987,
      "step": 35693
    },
    {
      "epoch": 1.9987680591331616,
      "grad_norm": 1.4305481910705566,
      "learning_rate": 9.587394736842105e-05,
      "loss": 0.4446,
      "step": 35694
    },
    {
      "epoch": 1.9988240564452906,
      "grad_norm": 1.106413722038269,
      "learning_rate": 9.587368421052631e-05,
      "loss": 0.3701,
      "step": 35695
    },
    {
      "epoch": 1.9988800537574196,
      "grad_norm": 1.6494601964950562,
      "learning_rate": 9.587342105263158e-05,
      "loss": 0.6221,
      "step": 35696
    },
    {
      "epoch": 1.9989360510695486,
      "grad_norm": 1.0754121541976929,
      "learning_rate": 9.587315789473684e-05,
      "loss": 0.4384,
      "step": 35697
    },
    {
      "epoch": 1.9989920483816777,
      "grad_norm": 1.6394118070602417,
      "learning_rate": 9.587289473684212e-05,
      "loss": 0.4564,
      "step": 35698
    },
    {
      "epoch": 1.9990480456938067,
      "grad_norm": 1.4092940092086792,
      "learning_rate": 9.587263157894736e-05,
      "loss": 0.4772,
      "step": 35699
    },
    {
      "epoch": 1.9991040430059357,
      "grad_norm": 1.2801393270492554,
      "learning_rate": 9.587236842105264e-05,
      "loss": 0.4322,
      "step": 35700
    },
    {
      "epoch": 1.9991600403180647,
      "grad_norm": 1.380327582359314,
      "learning_rate": 9.58721052631579e-05,
      "loss": 0.4788,
      "step": 35701
    },
    {
      "epoch": 1.9992160376301937,
      "grad_norm": 1.2255809307098389,
      "learning_rate": 9.587184210526317e-05,
      "loss": 0.5594,
      "step": 35702
    },
    {
      "epoch": 1.9992720349423228,
      "grad_norm": 1.1452220678329468,
      "learning_rate": 9.587157894736843e-05,
      "loss": 0.395,
      "step": 35703
    },
    {
      "epoch": 1.9993280322544518,
      "grad_norm": 1.340274691581726,
      "learning_rate": 9.587131578947369e-05,
      "loss": 0.4126,
      "step": 35704
    },
    {
      "epoch": 1.9993840295665808,
      "grad_norm": 1.1576058864593506,
      "learning_rate": 9.587105263157895e-05,
      "loss": 0.4399,
      "step": 35705
    },
    {
      "epoch": 1.9994400268787098,
      "grad_norm": 1.8450422286987305,
      "learning_rate": 9.587078947368422e-05,
      "loss": 0.6133,
      "step": 35706
    },
    {
      "epoch": 1.9994960241908388,
      "grad_norm": 1.3574519157409668,
      "learning_rate": 9.587052631578948e-05,
      "loss": 0.491,
      "step": 35707
    },
    {
      "epoch": 1.9995520215029678,
      "grad_norm": 1.2326457500457764,
      "learning_rate": 9.587026315789474e-05,
      "loss": 0.4273,
      "step": 35708
    },
    {
      "epoch": 1.9996080188150969,
      "grad_norm": 1.2883415222167969,
      "learning_rate": 9.587e-05,
      "loss": 0.5397,
      "step": 35709
    },
    {
      "epoch": 1.9996640161272259,
      "grad_norm": 1.2907953262329102,
      "learning_rate": 9.586973684210528e-05,
      "loss": 0.4966,
      "step": 35710
    },
    {
      "epoch": 1.999720013439355,
      "grad_norm": 1.2141108512878418,
      "learning_rate": 9.586947368421053e-05,
      "loss": 0.4072,
      "step": 35711
    },
    {
      "epoch": 1.999776010751484,
      "grad_norm": 1.2814353704452515,
      "learning_rate": 9.58692105263158e-05,
      "loss": 0.2963,
      "step": 35712
    },
    {
      "epoch": 1.999832008063613,
      "grad_norm": 1.3092714548110962,
      "learning_rate": 9.586894736842105e-05,
      "loss": 0.5815,
      "step": 35713
    },
    {
      "epoch": 1.999888005375742,
      "grad_norm": 1.191420316696167,
      "learning_rate": 9.586868421052631e-05,
      "loss": 0.5152,
      "step": 35714
    },
    {
      "epoch": 1.999944002687871,
      "grad_norm": 3.37546968460083,
      "learning_rate": 9.586842105263159e-05,
      "loss": 0.4066,
      "step": 35715
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.749148368835449,
      "learning_rate": 9.586815789473685e-05,
      "loss": 0.3993,
      "step": 35716
    },
    {
      "epoch": 2.000055997312129,
      "grad_norm": 1.182232141494751,
      "learning_rate": 9.586789473684211e-05,
      "loss": 0.3358,
      "step": 35717
    },
    {
      "epoch": 2.000111994624258,
      "grad_norm": 1.2029112577438354,
      "learning_rate": 9.586763157894737e-05,
      "loss": 0.4234,
      "step": 35718
    },
    {
      "epoch": 2.000167991936387,
      "grad_norm": 1.4242031574249268,
      "learning_rate": 9.586736842105264e-05,
      "loss": 0.4894,
      "step": 35719
    },
    {
      "epoch": 2.000223989248516,
      "grad_norm": 4.750683307647705,
      "learning_rate": 9.58671052631579e-05,
      "loss": 0.5317,
      "step": 35720
    },
    {
      "epoch": 2.000279986560645,
      "grad_norm": 1.2182492017745972,
      "learning_rate": 9.586684210526316e-05,
      "loss": 0.3997,
      "step": 35721
    },
    {
      "epoch": 2.000335983872774,
      "grad_norm": 1.1296474933624268,
      "learning_rate": 9.586657894736842e-05,
      "loss": 0.3267,
      "step": 35722
    },
    {
      "epoch": 2.000391981184903,
      "grad_norm": 1.637326717376709,
      "learning_rate": 9.586631578947369e-05,
      "loss": 0.363,
      "step": 35723
    },
    {
      "epoch": 2.000447978497032,
      "grad_norm": 1.2048490047454834,
      "learning_rate": 9.586605263157895e-05,
      "loss": 0.3873,
      "step": 35724
    },
    {
      "epoch": 2.000503975809161,
      "grad_norm": 1.240618109703064,
      "learning_rate": 9.586578947368423e-05,
      "loss": 0.3913,
      "step": 35725
    },
    {
      "epoch": 2.00055997312129,
      "grad_norm": 1.374437928199768,
      "learning_rate": 9.586552631578947e-05,
      "loss": 0.4084,
      "step": 35726
    },
    {
      "epoch": 2.000615970433419,
      "grad_norm": 1.311750054359436,
      "learning_rate": 9.586526315789474e-05,
      "loss": 0.3994,
      "step": 35727
    },
    {
      "epoch": 2.0006719677455482,
      "grad_norm": 1.074459195137024,
      "learning_rate": 9.5865e-05,
      "loss": 0.3234,
      "step": 35728
    },
    {
      "epoch": 2.0007279650576772,
      "grad_norm": 1.2244187593460083,
      "learning_rate": 9.586473684210526e-05,
      "loss": 0.4227,
      "step": 35729
    },
    {
      "epoch": 2.0007839623698063,
      "grad_norm": 1.313676357269287,
      "learning_rate": 9.586447368421054e-05,
      "loss": 0.3667,
      "step": 35730
    },
    {
      "epoch": 2.0008399596819353,
      "grad_norm": 1.1889480352401733,
      "learning_rate": 9.586421052631578e-05,
      "loss": 0.3467,
      "step": 35731
    },
    {
      "epoch": 2.0008959569940643,
      "grad_norm": 1.0301470756530762,
      "learning_rate": 9.586394736842106e-05,
      "loss": 0.3365,
      "step": 35732
    },
    {
      "epoch": 2.0009519543061933,
      "grad_norm": 2.0698623657226562,
      "learning_rate": 9.586368421052632e-05,
      "loss": 0.5078,
      "step": 35733
    },
    {
      "epoch": 2.0010079516183223,
      "grad_norm": 1.75558340549469,
      "learning_rate": 9.586342105263159e-05,
      "loss": 0.489,
      "step": 35734
    },
    {
      "epoch": 2.0010639489304514,
      "grad_norm": 1.4140416383743286,
      "learning_rate": 9.586315789473685e-05,
      "loss": 0.6721,
      "step": 35735
    },
    {
      "epoch": 2.0011199462425804,
      "grad_norm": 1.1331034898757935,
      "learning_rate": 9.586289473684211e-05,
      "loss": 0.471,
      "step": 35736
    },
    {
      "epoch": 2.0011759435547094,
      "grad_norm": 1.2293968200683594,
      "learning_rate": 9.586263157894737e-05,
      "loss": 0.4646,
      "step": 35737
    },
    {
      "epoch": 2.0012319408668384,
      "grad_norm": 1.138527750968933,
      "learning_rate": 9.586236842105264e-05,
      "loss": 0.3917,
      "step": 35738
    },
    {
      "epoch": 2.0012879381789674,
      "grad_norm": 1.1666724681854248,
      "learning_rate": 9.58621052631579e-05,
      "loss": 0.3377,
      "step": 35739
    },
    {
      "epoch": 2.0013439354910965,
      "grad_norm": 1.2568024396896362,
      "learning_rate": 9.586184210526316e-05,
      "loss": 0.4457,
      "step": 35740
    },
    {
      "epoch": 2.0013999328032255,
      "grad_norm": 1.2787812948226929,
      "learning_rate": 9.586157894736842e-05,
      "loss": 0.5511,
      "step": 35741
    },
    {
      "epoch": 2.0014559301153545,
      "grad_norm": 1.2110960483551025,
      "learning_rate": 9.58613157894737e-05,
      "loss": 0.423,
      "step": 35742
    },
    {
      "epoch": 2.0015119274274835,
      "grad_norm": 1.179447054862976,
      "learning_rate": 9.586105263157895e-05,
      "loss": 0.4133,
      "step": 35743
    },
    {
      "epoch": 2.0015679247396125,
      "grad_norm": 0.9979178309440613,
      "learning_rate": 9.586078947368421e-05,
      "loss": 0.397,
      "step": 35744
    },
    {
      "epoch": 2.0016239220517416,
      "grad_norm": 1.2861733436584473,
      "learning_rate": 9.586052631578947e-05,
      "loss": 0.3606,
      "step": 35745
    },
    {
      "epoch": 2.0016799193638706,
      "grad_norm": 1.187597393989563,
      "learning_rate": 9.586026315789473e-05,
      "loss": 0.4863,
      "step": 35746
    },
    {
      "epoch": 2.0017359166759996,
      "grad_norm": 1.4008166790008545,
      "learning_rate": 9.586000000000001e-05,
      "loss": 0.4478,
      "step": 35747
    },
    {
      "epoch": 2.0017919139881286,
      "grad_norm": 1.7283551692962646,
      "learning_rate": 9.585973684210527e-05,
      "loss": 0.475,
      "step": 35748
    },
    {
      "epoch": 2.0018479113002576,
      "grad_norm": 0.9884861707687378,
      "learning_rate": 9.585947368421053e-05,
      "loss": 0.3559,
      "step": 35749
    },
    {
      "epoch": 2.0019039086123867,
      "grad_norm": 1.0272537469863892,
      "learning_rate": 9.585921052631579e-05,
      "loss": 0.3473,
      "step": 35750
    },
    {
      "epoch": 2.0019599059245157,
      "grad_norm": 1.2969883680343628,
      "learning_rate": 9.585894736842106e-05,
      "loss": 0.4908,
      "step": 35751
    },
    {
      "epoch": 2.0020159032366447,
      "grad_norm": 1.3099888563156128,
      "learning_rate": 9.585868421052632e-05,
      "loss": 0.5012,
      "step": 35752
    },
    {
      "epoch": 2.0020719005487737,
      "grad_norm": 1.1673057079315186,
      "learning_rate": 9.585842105263159e-05,
      "loss": 0.3225,
      "step": 35753
    },
    {
      "epoch": 2.0021278978609027,
      "grad_norm": 1.207336187362671,
      "learning_rate": 9.585815789473684e-05,
      "loss": 0.388,
      "step": 35754
    },
    {
      "epoch": 2.0021838951730317,
      "grad_norm": 1.1550965309143066,
      "learning_rate": 9.585789473684211e-05,
      "loss": 0.3822,
      "step": 35755
    },
    {
      "epoch": 2.0022398924851608,
      "grad_norm": 1.2960766553878784,
      "learning_rate": 9.585763157894737e-05,
      "loss": 0.4392,
      "step": 35756
    },
    {
      "epoch": 2.00229588979729,
      "grad_norm": 1.2397187948226929,
      "learning_rate": 9.585736842105265e-05,
      "loss": 0.4618,
      "step": 35757
    },
    {
      "epoch": 2.002351887109419,
      "grad_norm": 1.3082901239395142,
      "learning_rate": 9.58571052631579e-05,
      "loss": 0.3647,
      "step": 35758
    },
    {
      "epoch": 2.002407884421548,
      "grad_norm": 1.0328633785247803,
      "learning_rate": 9.585684210526316e-05,
      "loss": 0.3228,
      "step": 35759
    },
    {
      "epoch": 2.002463881733677,
      "grad_norm": 1.1539751291275024,
      "learning_rate": 9.585657894736842e-05,
      "loss": 0.4277,
      "step": 35760
    },
    {
      "epoch": 2.002519879045806,
      "grad_norm": 1.3577837944030762,
      "learning_rate": 9.58563157894737e-05,
      "loss": 0.3331,
      "step": 35761
    },
    {
      "epoch": 2.002575876357935,
      "grad_norm": 1.2352749109268188,
      "learning_rate": 9.585605263157896e-05,
      "loss": 0.3721,
      "step": 35762
    },
    {
      "epoch": 2.002631873670064,
      "grad_norm": 1.2029948234558105,
      "learning_rate": 9.58557894736842e-05,
      "loss": 0.4044,
      "step": 35763
    },
    {
      "epoch": 2.002687870982193,
      "grad_norm": 1.164831280708313,
      "learning_rate": 9.585552631578948e-05,
      "loss": 0.4242,
      "step": 35764
    },
    {
      "epoch": 2.002743868294322,
      "grad_norm": 1.022634744644165,
      "learning_rate": 9.585526315789474e-05,
      "loss": 0.3323,
      "step": 35765
    },
    {
      "epoch": 2.002799865606451,
      "grad_norm": 1.1480488777160645,
      "learning_rate": 9.585500000000001e-05,
      "loss": 0.4344,
      "step": 35766
    },
    {
      "epoch": 2.00285586291858,
      "grad_norm": 1.2631078958511353,
      "learning_rate": 9.585473684210527e-05,
      "loss": 0.3399,
      "step": 35767
    },
    {
      "epoch": 2.002911860230709,
      "grad_norm": 1.3309322595596313,
      "learning_rate": 9.585447368421053e-05,
      "loss": 0.3777,
      "step": 35768
    },
    {
      "epoch": 2.002967857542838,
      "grad_norm": 1.3151357173919678,
      "learning_rate": 9.585421052631579e-05,
      "loss": 0.4689,
      "step": 35769
    },
    {
      "epoch": 2.003023854854967,
      "grad_norm": 1.2066999673843384,
      "learning_rate": 9.585394736842106e-05,
      "loss": 0.4101,
      "step": 35770
    },
    {
      "epoch": 2.003079852167096,
      "grad_norm": 1.628219485282898,
      "learning_rate": 9.585368421052632e-05,
      "loss": 0.4464,
      "step": 35771
    },
    {
      "epoch": 2.003135849479225,
      "grad_norm": 1.8368597030639648,
      "learning_rate": 9.585342105263158e-05,
      "loss": 0.376,
      "step": 35772
    },
    {
      "epoch": 2.003191846791354,
      "grad_norm": 1.2289546728134155,
      "learning_rate": 9.585315789473684e-05,
      "loss": 0.3091,
      "step": 35773
    },
    {
      "epoch": 2.003247844103483,
      "grad_norm": 1.1499255895614624,
      "learning_rate": 9.585289473684211e-05,
      "loss": 0.3235,
      "step": 35774
    },
    {
      "epoch": 2.003303841415612,
      "grad_norm": 1.0664421319961548,
      "learning_rate": 9.585263157894737e-05,
      "loss": 0.3417,
      "step": 35775
    },
    {
      "epoch": 2.003359838727741,
      "grad_norm": 1.402164340019226,
      "learning_rate": 9.585236842105263e-05,
      "loss": 0.3967,
      "step": 35776
    },
    {
      "epoch": 2.00341583603987,
      "grad_norm": 1.247530460357666,
      "learning_rate": 9.58521052631579e-05,
      "loss": 0.4341,
      "step": 35777
    },
    {
      "epoch": 2.003471833351999,
      "grad_norm": 1.1795834302902222,
      "learning_rate": 9.585184210526317e-05,
      "loss": 0.3836,
      "step": 35778
    },
    {
      "epoch": 2.003527830664128,
      "grad_norm": 1.086114764213562,
      "learning_rate": 9.585157894736843e-05,
      "loss": 0.41,
      "step": 35779
    },
    {
      "epoch": 2.0035838279762572,
      "grad_norm": 1.1867990493774414,
      "learning_rate": 9.585131578947369e-05,
      "loss": 0.3922,
      "step": 35780
    },
    {
      "epoch": 2.0036398252883862,
      "grad_norm": 1.5759308338165283,
      "learning_rate": 9.585105263157895e-05,
      "loss": 0.4954,
      "step": 35781
    },
    {
      "epoch": 2.0036958226005153,
      "grad_norm": 1.1702313423156738,
      "learning_rate": 9.58507894736842e-05,
      "loss": 0.4298,
      "step": 35782
    },
    {
      "epoch": 2.0037518199126443,
      "grad_norm": 1.2212941646575928,
      "learning_rate": 9.585052631578948e-05,
      "loss": 0.4765,
      "step": 35783
    },
    {
      "epoch": 2.0038078172247733,
      "grad_norm": 1.2862499952316284,
      "learning_rate": 9.585026315789474e-05,
      "loss": 0.3819,
      "step": 35784
    },
    {
      "epoch": 2.0038638145369023,
      "grad_norm": 1.1117898225784302,
      "learning_rate": 9.585000000000001e-05,
      "loss": 0.4198,
      "step": 35785
    },
    {
      "epoch": 2.0039198118490313,
      "grad_norm": 1.0650211572647095,
      "learning_rate": 9.584973684210526e-05,
      "loss": 0.2853,
      "step": 35786
    },
    {
      "epoch": 2.0039758091611604,
      "grad_norm": 1.1335151195526123,
      "learning_rate": 9.584947368421053e-05,
      "loss": 0.3703,
      "step": 35787
    },
    {
      "epoch": 2.0040318064732894,
      "grad_norm": 1.1390008926391602,
      "learning_rate": 9.584921052631579e-05,
      "loss": 0.4315,
      "step": 35788
    },
    {
      "epoch": 2.0040878037854184,
      "grad_norm": 1.2367141246795654,
      "learning_rate": 9.584894736842106e-05,
      "loss": 0.3785,
      "step": 35789
    },
    {
      "epoch": 2.0041438010975474,
      "grad_norm": 1.0569877624511719,
      "learning_rate": 9.584868421052632e-05,
      "loss": 0.3036,
      "step": 35790
    },
    {
      "epoch": 2.0041997984096764,
      "grad_norm": 1.3080319166183472,
      "learning_rate": 9.584842105263158e-05,
      "loss": 0.3466,
      "step": 35791
    },
    {
      "epoch": 2.0042557957218055,
      "grad_norm": 1.15290367603302,
      "learning_rate": 9.584815789473684e-05,
      "loss": 0.3132,
      "step": 35792
    },
    {
      "epoch": 2.0043117930339345,
      "grad_norm": 1.5112502574920654,
      "learning_rate": 9.584789473684212e-05,
      "loss": 0.2645,
      "step": 35793
    },
    {
      "epoch": 2.0043677903460635,
      "grad_norm": 1.4089183807373047,
      "learning_rate": 9.584763157894738e-05,
      "loss": 0.4406,
      "step": 35794
    },
    {
      "epoch": 2.0044237876581925,
      "grad_norm": 1.319968581199646,
      "learning_rate": 9.584736842105264e-05,
      "loss": 0.4964,
      "step": 35795
    },
    {
      "epoch": 2.0044797849703215,
      "grad_norm": 1.488022804260254,
      "learning_rate": 9.58471052631579e-05,
      "loss": 0.4173,
      "step": 35796
    },
    {
      "epoch": 2.0045357822824506,
      "grad_norm": 0.944221019744873,
      "learning_rate": 9.584684210526316e-05,
      "loss": 0.3124,
      "step": 35797
    },
    {
      "epoch": 2.0045917795945796,
      "grad_norm": 1.1327488422393799,
      "learning_rate": 9.584657894736843e-05,
      "loss": 0.4095,
      "step": 35798
    },
    {
      "epoch": 2.0046477769067086,
      "grad_norm": 1.058314323425293,
      "learning_rate": 9.584631578947369e-05,
      "loss": 0.3432,
      "step": 35799
    },
    {
      "epoch": 2.0047037742188376,
      "grad_norm": 1.3376262187957764,
      "learning_rate": 9.584605263157895e-05,
      "loss": 0.4135,
      "step": 35800
    },
    {
      "epoch": 2.0047597715309666,
      "grad_norm": 1.2323520183563232,
      "learning_rate": 9.584578947368421e-05,
      "loss": 0.3806,
      "step": 35801
    },
    {
      "epoch": 2.0048157688430956,
      "grad_norm": 1.4351063966751099,
      "learning_rate": 9.584552631578948e-05,
      "loss": 0.4225,
      "step": 35802
    },
    {
      "epoch": 2.0048717661552247,
      "grad_norm": 1.1267834901809692,
      "learning_rate": 9.584526315789474e-05,
      "loss": 0.4466,
      "step": 35803
    },
    {
      "epoch": 2.0049277634673537,
      "grad_norm": 1.6285756826400757,
      "learning_rate": 9.5845e-05,
      "loss": 0.3928,
      "step": 35804
    },
    {
      "epoch": 2.0049837607794827,
      "grad_norm": 1.2738834619522095,
      "learning_rate": 9.584473684210526e-05,
      "loss": 0.4749,
      "step": 35805
    },
    {
      "epoch": 2.0050397580916117,
      "grad_norm": 1.0531673431396484,
      "learning_rate": 9.584447368421053e-05,
      "loss": 0.2921,
      "step": 35806
    },
    {
      "epoch": 2.0050957554037407,
      "grad_norm": 1.0331834554672241,
      "learning_rate": 9.58442105263158e-05,
      "loss": 0.3885,
      "step": 35807
    },
    {
      "epoch": 2.0051517527158698,
      "grad_norm": 1.1604411602020264,
      "learning_rate": 9.584394736842107e-05,
      "loss": 0.4515,
      "step": 35808
    },
    {
      "epoch": 2.005207750027999,
      "grad_norm": 1.1311315298080444,
      "learning_rate": 9.584368421052631e-05,
      "loss": 0.4762,
      "step": 35809
    },
    {
      "epoch": 2.005263747340128,
      "grad_norm": 1.404622197151184,
      "learning_rate": 9.584342105263159e-05,
      "loss": 0.4433,
      "step": 35810
    },
    {
      "epoch": 2.005319744652257,
      "grad_norm": 1.1629540920257568,
      "learning_rate": 9.584315789473685e-05,
      "loss": 0.4171,
      "step": 35811
    },
    {
      "epoch": 2.005375741964386,
      "grad_norm": 1.1784030199050903,
      "learning_rate": 9.584289473684212e-05,
      "loss": 0.3564,
      "step": 35812
    },
    {
      "epoch": 2.005431739276515,
      "grad_norm": 1.2254221439361572,
      "learning_rate": 9.584263157894737e-05,
      "loss": 0.5103,
      "step": 35813
    },
    {
      "epoch": 2.005487736588644,
      "grad_norm": 1.1892682313919067,
      "learning_rate": 9.584236842105263e-05,
      "loss": 0.3648,
      "step": 35814
    },
    {
      "epoch": 2.005543733900773,
      "grad_norm": 1.1096426248550415,
      "learning_rate": 9.58421052631579e-05,
      "loss": 0.3227,
      "step": 35815
    },
    {
      "epoch": 2.005599731212902,
      "grad_norm": 1.273844599723816,
      "learning_rate": 9.584184210526316e-05,
      "loss": 0.4548,
      "step": 35816
    },
    {
      "epoch": 2.005655728525031,
      "grad_norm": 1.563796877861023,
      "learning_rate": 9.584157894736843e-05,
      "loss": 0.5533,
      "step": 35817
    },
    {
      "epoch": 2.00571172583716,
      "grad_norm": 1.0604662895202637,
      "learning_rate": 9.584131578947368e-05,
      "loss": 0.3694,
      "step": 35818
    },
    {
      "epoch": 2.005767723149289,
      "grad_norm": 1.0590765476226807,
      "learning_rate": 9.584105263157895e-05,
      "loss": 0.3599,
      "step": 35819
    },
    {
      "epoch": 2.005823720461418,
      "grad_norm": 1.3057653903961182,
      "learning_rate": 9.584078947368421e-05,
      "loss": 0.3673,
      "step": 35820
    },
    {
      "epoch": 2.005879717773547,
      "grad_norm": 1.3801923990249634,
      "learning_rate": 9.584052631578948e-05,
      "loss": 0.4259,
      "step": 35821
    },
    {
      "epoch": 2.005935715085676,
      "grad_norm": 1.0347604751586914,
      "learning_rate": 9.584026315789474e-05,
      "loss": 0.4183,
      "step": 35822
    },
    {
      "epoch": 2.005991712397805,
      "grad_norm": 1.06027352809906,
      "learning_rate": 9.584e-05,
      "loss": 0.4672,
      "step": 35823
    },
    {
      "epoch": 2.006047709709934,
      "grad_norm": 1.2466121912002563,
      "learning_rate": 9.583973684210526e-05,
      "loss": 0.3761,
      "step": 35824
    },
    {
      "epoch": 2.006103707022063,
      "grad_norm": 1.213773250579834,
      "learning_rate": 9.583947368421054e-05,
      "loss": 0.3395,
      "step": 35825
    },
    {
      "epoch": 2.006159704334192,
      "grad_norm": 1.1831530332565308,
      "learning_rate": 9.58392105263158e-05,
      "loss": 0.4747,
      "step": 35826
    },
    {
      "epoch": 2.006215701646321,
      "grad_norm": 1.307273507118225,
      "learning_rate": 9.583894736842106e-05,
      "loss": 0.4007,
      "step": 35827
    },
    {
      "epoch": 2.00627169895845,
      "grad_norm": 1.3315942287445068,
      "learning_rate": 9.583868421052632e-05,
      "loss": 0.4129,
      "step": 35828
    },
    {
      "epoch": 2.006327696270579,
      "grad_norm": 1.1211820840835571,
      "learning_rate": 9.583842105263159e-05,
      "loss": 0.4427,
      "step": 35829
    },
    {
      "epoch": 2.006383693582708,
      "grad_norm": 1.229960560798645,
      "learning_rate": 9.583815789473685e-05,
      "loss": 0.3709,
      "step": 35830
    },
    {
      "epoch": 2.006439690894837,
      "grad_norm": 1.319773554801941,
      "learning_rate": 9.583789473684211e-05,
      "loss": 0.3947,
      "step": 35831
    },
    {
      "epoch": 2.006495688206966,
      "grad_norm": 1.1925987005233765,
      "learning_rate": 9.583763157894737e-05,
      "loss": 0.4663,
      "step": 35832
    },
    {
      "epoch": 2.0065516855190952,
      "grad_norm": 1.1169219017028809,
      "learning_rate": 9.583736842105263e-05,
      "loss": 0.5099,
      "step": 35833
    },
    {
      "epoch": 2.0066076828312243,
      "grad_norm": 1.2888444662094116,
      "learning_rate": 9.58371052631579e-05,
      "loss": 0.4546,
      "step": 35834
    },
    {
      "epoch": 2.0066636801433533,
      "grad_norm": 0.917158305644989,
      "learning_rate": 9.583684210526316e-05,
      "loss": 0.4137,
      "step": 35835
    },
    {
      "epoch": 2.0067196774554823,
      "grad_norm": 1.3366646766662598,
      "learning_rate": 9.583657894736842e-05,
      "loss": 0.5159,
      "step": 35836
    },
    {
      "epoch": 2.0067756747676113,
      "grad_norm": 1.1978360414505005,
      "learning_rate": 9.583631578947368e-05,
      "loss": 0.3692,
      "step": 35837
    },
    {
      "epoch": 2.0068316720797403,
      "grad_norm": 1.040445327758789,
      "learning_rate": 9.583605263157895e-05,
      "loss": 0.3494,
      "step": 35838
    },
    {
      "epoch": 2.0068876693918694,
      "grad_norm": 1.335199236869812,
      "learning_rate": 9.583578947368421e-05,
      "loss": 0.4833,
      "step": 35839
    },
    {
      "epoch": 2.0069436667039984,
      "grad_norm": 1.7000359296798706,
      "learning_rate": 9.583552631578949e-05,
      "loss": 0.5331,
      "step": 35840
    },
    {
      "epoch": 2.0069996640161274,
      "grad_norm": 1.0351409912109375,
      "learning_rate": 9.583526315789473e-05,
      "loss": 0.3562,
      "step": 35841
    },
    {
      "epoch": 2.0070556613282564,
      "grad_norm": 1.117228627204895,
      "learning_rate": 9.5835e-05,
      "loss": 0.3918,
      "step": 35842
    },
    {
      "epoch": 2.0071116586403854,
      "grad_norm": 1.4779503345489502,
      "learning_rate": 9.583473684210527e-05,
      "loss": 0.3577,
      "step": 35843
    },
    {
      "epoch": 2.0071676559525145,
      "grad_norm": 1.2578717470169067,
      "learning_rate": 9.583447368421054e-05,
      "loss": 0.4472,
      "step": 35844
    },
    {
      "epoch": 2.0072236532646435,
      "grad_norm": 1.2031910419464111,
      "learning_rate": 9.58342105263158e-05,
      "loss": 0.5332,
      "step": 35845
    },
    {
      "epoch": 2.0072796505767725,
      "grad_norm": 2.2340481281280518,
      "learning_rate": 9.583394736842106e-05,
      "loss": 0.4782,
      "step": 35846
    },
    {
      "epoch": 2.0073356478889015,
      "grad_norm": 1.074960708618164,
      "learning_rate": 9.583368421052632e-05,
      "loss": 0.4155,
      "step": 35847
    },
    {
      "epoch": 2.0073916452010305,
      "grad_norm": 1.1378623247146606,
      "learning_rate": 9.583342105263159e-05,
      "loss": 0.4665,
      "step": 35848
    },
    {
      "epoch": 2.0074476425131595,
      "grad_norm": 1.3077210187911987,
      "learning_rate": 9.583315789473685e-05,
      "loss": 0.386,
      "step": 35849
    },
    {
      "epoch": 2.0075036398252886,
      "grad_norm": 1.2918747663497925,
      "learning_rate": 9.583289473684211e-05,
      "loss": 0.3627,
      "step": 35850
    },
    {
      "epoch": 2.0075596371374176,
      "grad_norm": 1.1277105808258057,
      "learning_rate": 9.583263157894737e-05,
      "loss": 0.3799,
      "step": 35851
    },
    {
      "epoch": 2.0076156344495466,
      "grad_norm": 1.3728315830230713,
      "learning_rate": 9.583236842105263e-05,
      "loss": 0.5168,
      "step": 35852
    },
    {
      "epoch": 2.0076716317616756,
      "grad_norm": 1.2692402601242065,
      "learning_rate": 9.58321052631579e-05,
      "loss": 0.3778,
      "step": 35853
    },
    {
      "epoch": 2.0077276290738046,
      "grad_norm": 1.251347303390503,
      "learning_rate": 9.583184210526316e-05,
      "loss": 0.4026,
      "step": 35854
    },
    {
      "epoch": 2.0077836263859337,
      "grad_norm": 1.8793201446533203,
      "learning_rate": 9.583157894736842e-05,
      "loss": 0.3749,
      "step": 35855
    },
    {
      "epoch": 2.0078396236980627,
      "grad_norm": 1.4593137502670288,
      "learning_rate": 9.583131578947368e-05,
      "loss": 0.4253,
      "step": 35856
    },
    {
      "epoch": 2.0078956210101917,
      "grad_norm": 1.2291958332061768,
      "learning_rate": 9.583105263157896e-05,
      "loss": 0.4263,
      "step": 35857
    },
    {
      "epoch": 2.0079516183223207,
      "grad_norm": 1.1300537586212158,
      "learning_rate": 9.583078947368422e-05,
      "loss": 0.442,
      "step": 35858
    },
    {
      "epoch": 2.0080076156344497,
      "grad_norm": 2.294628381729126,
      "learning_rate": 9.583052631578948e-05,
      "loss": 0.5886,
      "step": 35859
    },
    {
      "epoch": 2.0080636129465788,
      "grad_norm": 1.2356338500976562,
      "learning_rate": 9.583026315789474e-05,
      "loss": 0.3758,
      "step": 35860
    },
    {
      "epoch": 2.0081196102587078,
      "grad_norm": 1.3109859228134155,
      "learning_rate": 9.583000000000001e-05,
      "loss": 0.437,
      "step": 35861
    },
    {
      "epoch": 2.008175607570837,
      "grad_norm": 1.2050403356552124,
      "learning_rate": 9.582973684210527e-05,
      "loss": 0.4448,
      "step": 35862
    },
    {
      "epoch": 2.008231604882966,
      "grad_norm": 1.7827585935592651,
      "learning_rate": 9.582947368421054e-05,
      "loss": 0.4063,
      "step": 35863
    },
    {
      "epoch": 2.008287602195095,
      "grad_norm": 1.5390233993530273,
      "learning_rate": 9.582921052631579e-05,
      "loss": 0.5479,
      "step": 35864
    },
    {
      "epoch": 2.008343599507224,
      "grad_norm": 1.1264935731887817,
      "learning_rate": 9.582894736842106e-05,
      "loss": 0.3283,
      "step": 35865
    },
    {
      "epoch": 2.008399596819353,
      "grad_norm": 1.103168249130249,
      "learning_rate": 9.582868421052632e-05,
      "loss": 0.4165,
      "step": 35866
    },
    {
      "epoch": 2.008455594131482,
      "grad_norm": 1.0895768404006958,
      "learning_rate": 9.582842105263158e-05,
      "loss": 0.3414,
      "step": 35867
    },
    {
      "epoch": 2.008511591443611,
      "grad_norm": 1.1353875398635864,
      "learning_rate": 9.582815789473684e-05,
      "loss": 0.4349,
      "step": 35868
    },
    {
      "epoch": 2.00856758875574,
      "grad_norm": 1.701114535331726,
      "learning_rate": 9.58278947368421e-05,
      "loss": 0.4884,
      "step": 35869
    },
    {
      "epoch": 2.008623586067869,
      "grad_norm": 1.1399396657943726,
      "learning_rate": 9.582763157894737e-05,
      "loss": 0.3099,
      "step": 35870
    },
    {
      "epoch": 2.008679583379998,
      "grad_norm": 1.2674838304519653,
      "learning_rate": 9.582736842105263e-05,
      "loss": 0.3562,
      "step": 35871
    },
    {
      "epoch": 2.008735580692127,
      "grad_norm": 1.2159181833267212,
      "learning_rate": 9.582710526315791e-05,
      "loss": 0.3508,
      "step": 35872
    },
    {
      "epoch": 2.008791578004256,
      "grad_norm": 1.099257469177246,
      "learning_rate": 9.582684210526315e-05,
      "loss": 0.3758,
      "step": 35873
    },
    {
      "epoch": 2.008847575316385,
      "grad_norm": 1.1834862232208252,
      "learning_rate": 9.582657894736843e-05,
      "loss": 0.3256,
      "step": 35874
    },
    {
      "epoch": 2.008903572628514,
      "grad_norm": 1.1165449619293213,
      "learning_rate": 9.582631578947369e-05,
      "loss": 0.3158,
      "step": 35875
    },
    {
      "epoch": 2.008959569940643,
      "grad_norm": 1.0332053899765015,
      "learning_rate": 9.582605263157896e-05,
      "loss": 0.3302,
      "step": 35876
    },
    {
      "epoch": 2.009015567252772,
      "grad_norm": 1.2526644468307495,
      "learning_rate": 9.582578947368422e-05,
      "loss": 0.3243,
      "step": 35877
    },
    {
      "epoch": 2.009071564564901,
      "grad_norm": 1.3846008777618408,
      "learning_rate": 9.582552631578948e-05,
      "loss": 0.5322,
      "step": 35878
    },
    {
      "epoch": 2.00912756187703,
      "grad_norm": 1.2860394716262817,
      "learning_rate": 9.582526315789474e-05,
      "loss": 0.4885,
      "step": 35879
    },
    {
      "epoch": 2.009183559189159,
      "grad_norm": 1.0912970304489136,
      "learning_rate": 9.582500000000001e-05,
      "loss": 0.3656,
      "step": 35880
    },
    {
      "epoch": 2.0092395565012877,
      "grad_norm": 1.3248655796051025,
      "learning_rate": 9.582473684210527e-05,
      "loss": 0.5045,
      "step": 35881
    },
    {
      "epoch": 2.0092955538134167,
      "grad_norm": 8.629794120788574,
      "learning_rate": 9.582447368421053e-05,
      "loss": 0.5859,
      "step": 35882
    },
    {
      "epoch": 2.0093515511255458,
      "grad_norm": 1.2444722652435303,
      "learning_rate": 9.582421052631579e-05,
      "loss": 0.3668,
      "step": 35883
    },
    {
      "epoch": 2.0094075484376748,
      "grad_norm": 1.117751121520996,
      "learning_rate": 9.582394736842105e-05,
      "loss": 0.3771,
      "step": 35884
    },
    {
      "epoch": 2.009463545749804,
      "grad_norm": 1.234378457069397,
      "learning_rate": 9.582368421052632e-05,
      "loss": 0.316,
      "step": 35885
    },
    {
      "epoch": 2.009519543061933,
      "grad_norm": 1.1099599599838257,
      "learning_rate": 9.582342105263158e-05,
      "loss": 0.3597,
      "step": 35886
    },
    {
      "epoch": 2.009575540374062,
      "grad_norm": 1.1704127788543701,
      "learning_rate": 9.582315789473684e-05,
      "loss": 0.6218,
      "step": 35887
    },
    {
      "epoch": 2.009631537686191,
      "grad_norm": 1.6880762577056885,
      "learning_rate": 9.58228947368421e-05,
      "loss": 0.5584,
      "step": 35888
    },
    {
      "epoch": 2.00968753499832,
      "grad_norm": 1.1258209943771362,
      "learning_rate": 9.582263157894738e-05,
      "loss": 0.3726,
      "step": 35889
    },
    {
      "epoch": 2.009743532310449,
      "grad_norm": 1.1790224313735962,
      "learning_rate": 9.582236842105264e-05,
      "loss": 0.3754,
      "step": 35890
    },
    {
      "epoch": 2.009799529622578,
      "grad_norm": 1.0744763612747192,
      "learning_rate": 9.58221052631579e-05,
      "loss": 0.4505,
      "step": 35891
    },
    {
      "epoch": 2.009855526934707,
      "grad_norm": 1.2784233093261719,
      "learning_rate": 9.582184210526316e-05,
      "loss": 0.3827,
      "step": 35892
    },
    {
      "epoch": 2.009911524246836,
      "grad_norm": 1.1017351150512695,
      "learning_rate": 9.582157894736843e-05,
      "loss": 0.3326,
      "step": 35893
    },
    {
      "epoch": 2.009967521558965,
      "grad_norm": 1.0899889469146729,
      "learning_rate": 9.582131578947369e-05,
      "loss": 0.3688,
      "step": 35894
    },
    {
      "epoch": 2.010023518871094,
      "grad_norm": 1.0779321193695068,
      "learning_rate": 9.582105263157896e-05,
      "loss": 0.456,
      "step": 35895
    },
    {
      "epoch": 2.010079516183223,
      "grad_norm": 1.6367486715316772,
      "learning_rate": 9.582078947368421e-05,
      "loss": 0.41,
      "step": 35896
    },
    {
      "epoch": 2.010135513495352,
      "grad_norm": 1.3015451431274414,
      "learning_rate": 9.582052631578948e-05,
      "loss": 0.5063,
      "step": 35897
    },
    {
      "epoch": 2.010191510807481,
      "grad_norm": 1.2744941711425781,
      "learning_rate": 9.582026315789474e-05,
      "loss": 0.3712,
      "step": 35898
    },
    {
      "epoch": 2.01024750811961,
      "grad_norm": 1.3015434741973877,
      "learning_rate": 9.582000000000001e-05,
      "loss": 0.4476,
      "step": 35899
    },
    {
      "epoch": 2.010303505431739,
      "grad_norm": 1.184593915939331,
      "learning_rate": 9.581973684210527e-05,
      "loss": 0.3752,
      "step": 35900
    },
    {
      "epoch": 2.010359502743868,
      "grad_norm": 1.2621227502822876,
      "learning_rate": 9.581947368421052e-05,
      "loss": 0.5791,
      "step": 35901
    },
    {
      "epoch": 2.010415500055997,
      "grad_norm": 1.1744240522384644,
      "learning_rate": 9.58192105263158e-05,
      "loss": 0.5119,
      "step": 35902
    },
    {
      "epoch": 2.010471497368126,
      "grad_norm": 1.249359130859375,
      "learning_rate": 9.581894736842105e-05,
      "loss": 0.4097,
      "step": 35903
    },
    {
      "epoch": 2.010527494680255,
      "grad_norm": 1.1291635036468506,
      "learning_rate": 9.581868421052633e-05,
      "loss": 0.5405,
      "step": 35904
    },
    {
      "epoch": 2.010583491992384,
      "grad_norm": 1.0455520153045654,
      "learning_rate": 9.581842105263159e-05,
      "loss": 0.2966,
      "step": 35905
    },
    {
      "epoch": 2.010639489304513,
      "grad_norm": 1.5298330783843994,
      "learning_rate": 9.581815789473685e-05,
      "loss": 0.3256,
      "step": 35906
    },
    {
      "epoch": 2.010695486616642,
      "grad_norm": 1.1614567041397095,
      "learning_rate": 9.58178947368421e-05,
      "loss": 0.3451,
      "step": 35907
    },
    {
      "epoch": 2.0107514839287712,
      "grad_norm": 1.3921276330947876,
      "learning_rate": 9.581763157894738e-05,
      "loss": 0.5074,
      "step": 35908
    },
    {
      "epoch": 2.0108074812409003,
      "grad_norm": 1.4734488725662231,
      "learning_rate": 9.581736842105264e-05,
      "loss": 0.4119,
      "step": 35909
    },
    {
      "epoch": 2.0108634785530293,
      "grad_norm": 1.422045350074768,
      "learning_rate": 9.58171052631579e-05,
      "loss": 0.4492,
      "step": 35910
    },
    {
      "epoch": 2.0109194758651583,
      "grad_norm": 0.9959626197814941,
      "learning_rate": 9.581684210526316e-05,
      "loss": 0.3575,
      "step": 35911
    },
    {
      "epoch": 2.0109754731772873,
      "grad_norm": 1.044052004814148,
      "learning_rate": 9.581657894736843e-05,
      "loss": 0.3031,
      "step": 35912
    },
    {
      "epoch": 2.0110314704894163,
      "grad_norm": 1.3160734176635742,
      "learning_rate": 9.581631578947369e-05,
      "loss": 0.3027,
      "step": 35913
    },
    {
      "epoch": 2.0110874678015453,
      "grad_norm": 1.310926079750061,
      "learning_rate": 9.581605263157895e-05,
      "loss": 0.5398,
      "step": 35914
    },
    {
      "epoch": 2.0111434651136744,
      "grad_norm": 1.1180884838104248,
      "learning_rate": 9.581578947368421e-05,
      "loss": 0.3767,
      "step": 35915
    },
    {
      "epoch": 2.0111994624258034,
      "grad_norm": 1.181401252746582,
      "learning_rate": 9.581552631578948e-05,
      "loss": 0.4468,
      "step": 35916
    },
    {
      "epoch": 2.0112554597379324,
      "grad_norm": 0.9832403063774109,
      "learning_rate": 9.581526315789474e-05,
      "loss": 0.3448,
      "step": 35917
    },
    {
      "epoch": 2.0113114570500614,
      "grad_norm": 1.194062352180481,
      "learning_rate": 9.5815e-05,
      "loss": 0.3539,
      "step": 35918
    },
    {
      "epoch": 2.0113674543621904,
      "grad_norm": 1.1809194087982178,
      "learning_rate": 9.581473684210526e-05,
      "loss": 0.5476,
      "step": 35919
    },
    {
      "epoch": 2.0114234516743195,
      "grad_norm": 1.3839787244796753,
      "learning_rate": 9.581447368421052e-05,
      "loss": 0.3048,
      "step": 35920
    },
    {
      "epoch": 2.0114794489864485,
      "grad_norm": 1.1443463563919067,
      "learning_rate": 9.58142105263158e-05,
      "loss": 0.4274,
      "step": 35921
    },
    {
      "epoch": 2.0115354462985775,
      "grad_norm": 1.2282607555389404,
      "learning_rate": 9.581394736842106e-05,
      "loss": 0.4215,
      "step": 35922
    },
    {
      "epoch": 2.0115914436107065,
      "grad_norm": 1.3101545572280884,
      "learning_rate": 9.581368421052632e-05,
      "loss": 0.4682,
      "step": 35923
    },
    {
      "epoch": 2.0116474409228355,
      "grad_norm": 1.539891242980957,
      "learning_rate": 9.581342105263158e-05,
      "loss": 0.5432,
      "step": 35924
    },
    {
      "epoch": 2.0117034382349646,
      "grad_norm": 1.0843690633773804,
      "learning_rate": 9.581315789473685e-05,
      "loss": 0.3139,
      "step": 35925
    },
    {
      "epoch": 2.0117594355470936,
      "grad_norm": 1.157828688621521,
      "learning_rate": 9.581289473684211e-05,
      "loss": 0.4349,
      "step": 35926
    },
    {
      "epoch": 2.0118154328592226,
      "grad_norm": 1.4255735874176025,
      "learning_rate": 9.581263157894738e-05,
      "loss": 0.4029,
      "step": 35927
    },
    {
      "epoch": 2.0118714301713516,
      "grad_norm": 1.1901848316192627,
      "learning_rate": 9.581236842105263e-05,
      "loss": 0.3723,
      "step": 35928
    },
    {
      "epoch": 2.0119274274834806,
      "grad_norm": 4.131755828857422,
      "learning_rate": 9.58121052631579e-05,
      "loss": 0.5211,
      "step": 35929
    },
    {
      "epoch": 2.0119834247956097,
      "grad_norm": 1.1934020519256592,
      "learning_rate": 9.581184210526316e-05,
      "loss": 0.3866,
      "step": 35930
    },
    {
      "epoch": 2.0120394221077387,
      "grad_norm": 1.4586005210876465,
      "learning_rate": 9.581157894736843e-05,
      "loss": 0.4046,
      "step": 35931
    },
    {
      "epoch": 2.0120954194198677,
      "grad_norm": 0.9839408993721008,
      "learning_rate": 9.58113157894737e-05,
      "loss": 0.3388,
      "step": 35932
    },
    {
      "epoch": 2.0121514167319967,
      "grad_norm": 1.388522982597351,
      "learning_rate": 9.581105263157895e-05,
      "loss": 0.4598,
      "step": 35933
    },
    {
      "epoch": 2.0122074140441257,
      "grad_norm": 1.2487350702285767,
      "learning_rate": 9.581078947368421e-05,
      "loss": 0.4325,
      "step": 35934
    },
    {
      "epoch": 2.0122634113562547,
      "grad_norm": 1.2178984880447388,
      "learning_rate": 9.581052631578947e-05,
      "loss": 0.4311,
      "step": 35935
    },
    {
      "epoch": 2.0123194086683838,
      "grad_norm": 1.1398550271987915,
      "learning_rate": 9.581026315789475e-05,
      "loss": 0.4934,
      "step": 35936
    },
    {
      "epoch": 2.012375405980513,
      "grad_norm": 1.0782445669174194,
      "learning_rate": 9.581e-05,
      "loss": 0.4676,
      "step": 35937
    },
    {
      "epoch": 2.012431403292642,
      "grad_norm": 1.114208698272705,
      "learning_rate": 9.580973684210527e-05,
      "loss": 0.3577,
      "step": 35938
    },
    {
      "epoch": 2.012487400604771,
      "grad_norm": 1.2817232608795166,
      "learning_rate": 9.580947368421053e-05,
      "loss": 0.4958,
      "step": 35939
    },
    {
      "epoch": 2.0125433979169,
      "grad_norm": 1.174094796180725,
      "learning_rate": 9.58092105263158e-05,
      "loss": 0.5193,
      "step": 35940
    },
    {
      "epoch": 2.012599395229029,
      "grad_norm": 1.1445204019546509,
      "learning_rate": 9.580894736842106e-05,
      "loss": 0.3182,
      "step": 35941
    },
    {
      "epoch": 2.012655392541158,
      "grad_norm": 1.3254717588424683,
      "learning_rate": 9.580868421052632e-05,
      "loss": 0.4159,
      "step": 35942
    },
    {
      "epoch": 2.012711389853287,
      "grad_norm": 3.3294131755828857,
      "learning_rate": 9.580842105263158e-05,
      "loss": 0.3783,
      "step": 35943
    },
    {
      "epoch": 2.012767387165416,
      "grad_norm": 1.3575527667999268,
      "learning_rate": 9.580815789473685e-05,
      "loss": 0.4383,
      "step": 35944
    },
    {
      "epoch": 2.012823384477545,
      "grad_norm": 1.1147736310958862,
      "learning_rate": 9.580789473684211e-05,
      "loss": 0.4005,
      "step": 35945
    },
    {
      "epoch": 2.012879381789674,
      "grad_norm": 0.9356734752655029,
      "learning_rate": 9.580763157894737e-05,
      "loss": 0.3653,
      "step": 35946
    },
    {
      "epoch": 2.012935379101803,
      "grad_norm": 1.1593421697616577,
      "learning_rate": 9.580736842105263e-05,
      "loss": 0.3229,
      "step": 35947
    },
    {
      "epoch": 2.012991376413932,
      "grad_norm": 1.0800411701202393,
      "learning_rate": 9.58071052631579e-05,
      "loss": 0.5072,
      "step": 35948
    },
    {
      "epoch": 2.013047373726061,
      "grad_norm": 1.129819631576538,
      "learning_rate": 9.580684210526316e-05,
      "loss": 0.317,
      "step": 35949
    },
    {
      "epoch": 2.01310337103819,
      "grad_norm": 1.2670780420303345,
      "learning_rate": 9.580657894736844e-05,
      "loss": 0.4502,
      "step": 35950
    },
    {
      "epoch": 2.013159368350319,
      "grad_norm": 1.0401873588562012,
      "learning_rate": 9.580631578947368e-05,
      "loss": 0.2882,
      "step": 35951
    },
    {
      "epoch": 2.013215365662448,
      "grad_norm": 1.3120824098587036,
      "learning_rate": 9.580605263157896e-05,
      "loss": 0.4081,
      "step": 35952
    },
    {
      "epoch": 2.013271362974577,
      "grad_norm": 1.2080658674240112,
      "learning_rate": 9.580578947368422e-05,
      "loss": 0.2895,
      "step": 35953
    },
    {
      "epoch": 2.013327360286706,
      "grad_norm": 1.1258790493011475,
      "learning_rate": 9.580552631578948e-05,
      "loss": 0.3108,
      "step": 35954
    },
    {
      "epoch": 2.013383357598835,
      "grad_norm": 1.244515061378479,
      "learning_rate": 9.580526315789475e-05,
      "loss": 0.4727,
      "step": 35955
    },
    {
      "epoch": 2.013439354910964,
      "grad_norm": 1.1742671728134155,
      "learning_rate": 9.5805e-05,
      "loss": 0.3761,
      "step": 35956
    },
    {
      "epoch": 2.013495352223093,
      "grad_norm": 1.2236824035644531,
      "learning_rate": 9.580473684210527e-05,
      "loss": 0.3268,
      "step": 35957
    },
    {
      "epoch": 2.013551349535222,
      "grad_norm": 0.9712657928466797,
      "learning_rate": 9.580447368421053e-05,
      "loss": 0.3369,
      "step": 35958
    },
    {
      "epoch": 2.013607346847351,
      "grad_norm": 1.0330934524536133,
      "learning_rate": 9.58042105263158e-05,
      "loss": 0.3252,
      "step": 35959
    },
    {
      "epoch": 2.0136633441594802,
      "grad_norm": 1.1708682775497437,
      "learning_rate": 9.580394736842106e-05,
      "loss": 0.4575,
      "step": 35960
    },
    {
      "epoch": 2.0137193414716092,
      "grad_norm": 1.1404727697372437,
      "learning_rate": 9.580368421052632e-05,
      "loss": 0.3624,
      "step": 35961
    },
    {
      "epoch": 2.0137753387837383,
      "grad_norm": 1.5483572483062744,
      "learning_rate": 9.580342105263158e-05,
      "loss": 0.3849,
      "step": 35962
    },
    {
      "epoch": 2.0138313360958673,
      "grad_norm": 1.1538753509521484,
      "learning_rate": 9.580315789473685e-05,
      "loss": 0.4729,
      "step": 35963
    },
    {
      "epoch": 2.0138873334079963,
      "grad_norm": 1.1414966583251953,
      "learning_rate": 9.580289473684211e-05,
      "loss": 0.4258,
      "step": 35964
    },
    {
      "epoch": 2.0139433307201253,
      "grad_norm": 1.4914534091949463,
      "learning_rate": 9.580263157894737e-05,
      "loss": 0.3606,
      "step": 35965
    },
    {
      "epoch": 2.0139993280322543,
      "grad_norm": 1.128311038017273,
      "learning_rate": 9.580236842105263e-05,
      "loss": 0.4615,
      "step": 35966
    },
    {
      "epoch": 2.0140553253443834,
      "grad_norm": 1.1732231378555298,
      "learning_rate": 9.58021052631579e-05,
      "loss": 0.4487,
      "step": 35967
    },
    {
      "epoch": 2.0141113226565124,
      "grad_norm": 1.0852938890457153,
      "learning_rate": 9.580184210526317e-05,
      "loss": 0.3207,
      "step": 35968
    },
    {
      "epoch": 2.0141673199686414,
      "grad_norm": 1.2399272918701172,
      "learning_rate": 9.580157894736843e-05,
      "loss": 0.4097,
      "step": 35969
    },
    {
      "epoch": 2.0142233172807704,
      "grad_norm": 1.2835215330123901,
      "learning_rate": 9.580131578947369e-05,
      "loss": 0.4627,
      "step": 35970
    },
    {
      "epoch": 2.0142793145928994,
      "grad_norm": 1.1350926160812378,
      "learning_rate": 9.580105263157895e-05,
      "loss": 0.3804,
      "step": 35971
    },
    {
      "epoch": 2.0143353119050285,
      "grad_norm": 1.1614155769348145,
      "learning_rate": 9.580078947368422e-05,
      "loss": 0.4744,
      "step": 35972
    },
    {
      "epoch": 2.0143913092171575,
      "grad_norm": 1.3877421617507935,
      "learning_rate": 9.580052631578948e-05,
      "loss": 0.5546,
      "step": 35973
    },
    {
      "epoch": 2.0144473065292865,
      "grad_norm": 1.0699504613876343,
      "learning_rate": 9.580026315789474e-05,
      "loss": 0.4023,
      "step": 35974
    },
    {
      "epoch": 2.0145033038414155,
      "grad_norm": 1.6108322143554688,
      "learning_rate": 9.58e-05,
      "loss": 0.4144,
      "step": 35975
    },
    {
      "epoch": 2.0145593011535445,
      "grad_norm": 1.1560313701629639,
      "learning_rate": 9.579973684210527e-05,
      "loss": 0.4657,
      "step": 35976
    },
    {
      "epoch": 2.0146152984656736,
      "grad_norm": 1.2327479124069214,
      "learning_rate": 9.579947368421053e-05,
      "loss": 0.3902,
      "step": 35977
    },
    {
      "epoch": 2.0146712957778026,
      "grad_norm": 1.1427966356277466,
      "learning_rate": 9.579921052631579e-05,
      "loss": 0.3219,
      "step": 35978
    },
    {
      "epoch": 2.0147272930899316,
      "grad_norm": 1.204358458518982,
      "learning_rate": 9.579894736842105e-05,
      "loss": 0.3552,
      "step": 35979
    },
    {
      "epoch": 2.0147832904020606,
      "grad_norm": 1.119761347770691,
      "learning_rate": 9.579868421052632e-05,
      "loss": 0.3405,
      "step": 35980
    },
    {
      "epoch": 2.0148392877141896,
      "grad_norm": 1.1006317138671875,
      "learning_rate": 9.579842105263158e-05,
      "loss": 0.3676,
      "step": 35981
    },
    {
      "epoch": 2.0148952850263186,
      "grad_norm": 1.3845229148864746,
      "learning_rate": 9.579815789473686e-05,
      "loss": 0.391,
      "step": 35982
    },
    {
      "epoch": 2.0149512823384477,
      "grad_norm": 1.2657369375228882,
      "learning_rate": 9.57978947368421e-05,
      "loss": 0.3135,
      "step": 35983
    },
    {
      "epoch": 2.0150072796505767,
      "grad_norm": 1.659509301185608,
      "learning_rate": 9.579763157894738e-05,
      "loss": 0.4413,
      "step": 35984
    },
    {
      "epoch": 2.0150632769627057,
      "grad_norm": 1.4364237785339355,
      "learning_rate": 9.579736842105264e-05,
      "loss": 0.3885,
      "step": 35985
    },
    {
      "epoch": 2.0151192742748347,
      "grad_norm": 1.399691104888916,
      "learning_rate": 9.579710526315791e-05,
      "loss": 0.4291,
      "step": 35986
    },
    {
      "epoch": 2.0151752715869637,
      "grad_norm": 1.1163642406463623,
      "learning_rate": 9.579684210526317e-05,
      "loss": 0.3634,
      "step": 35987
    },
    {
      "epoch": 2.0152312688990928,
      "grad_norm": 1.1183875799179077,
      "learning_rate": 9.579657894736841e-05,
      "loss": 0.476,
      "step": 35988
    },
    {
      "epoch": 2.015287266211222,
      "grad_norm": 1.3187004327774048,
      "learning_rate": 9.579631578947369e-05,
      "loss": 0.4876,
      "step": 35989
    },
    {
      "epoch": 2.015343263523351,
      "grad_norm": 1.2039744853973389,
      "learning_rate": 9.579605263157895e-05,
      "loss": 0.3554,
      "step": 35990
    },
    {
      "epoch": 2.01539926083548,
      "grad_norm": 1.1927218437194824,
      "learning_rate": 9.579578947368422e-05,
      "loss": 0.4803,
      "step": 35991
    },
    {
      "epoch": 2.015455258147609,
      "grad_norm": 0.9970502257347107,
      "learning_rate": 9.579552631578948e-05,
      "loss": 0.3802,
      "step": 35992
    },
    {
      "epoch": 2.015511255459738,
      "grad_norm": 1.1429189443588257,
      "learning_rate": 9.579526315789474e-05,
      "loss": 0.3696,
      "step": 35993
    },
    {
      "epoch": 2.015567252771867,
      "grad_norm": 1.4234963655471802,
      "learning_rate": 9.5795e-05,
      "loss": 0.435,
      "step": 35994
    },
    {
      "epoch": 2.015623250083996,
      "grad_norm": 1.411230444908142,
      "learning_rate": 9.579473684210527e-05,
      "loss": 0.5022,
      "step": 35995
    },
    {
      "epoch": 2.015679247396125,
      "grad_norm": 1.1422884464263916,
      "learning_rate": 9.579447368421053e-05,
      "loss": 0.3783,
      "step": 35996
    },
    {
      "epoch": 2.015735244708254,
      "grad_norm": 1.1443277597427368,
      "learning_rate": 9.579421052631579e-05,
      "loss": 0.3681,
      "step": 35997
    },
    {
      "epoch": 2.015791242020383,
      "grad_norm": 1.1956120729446411,
      "learning_rate": 9.579394736842105e-05,
      "loss": 0.5532,
      "step": 35998
    },
    {
      "epoch": 2.015847239332512,
      "grad_norm": 1.4882780313491821,
      "learning_rate": 9.579368421052633e-05,
      "loss": 0.4715,
      "step": 35999
    },
    {
      "epoch": 2.015903236644641,
      "grad_norm": 1.3563767671585083,
      "learning_rate": 9.579342105263159e-05,
      "loss": 0.3056,
      "step": 36000
    },
    {
      "epoch": 2.01595923395677,
      "grad_norm": 1.0231021642684937,
      "learning_rate": 9.579315789473685e-05,
      "loss": 0.3587,
      "step": 36001
    },
    {
      "epoch": 2.016015231268899,
      "grad_norm": 1.1333285570144653,
      "learning_rate": 9.57928947368421e-05,
      "loss": 0.3389,
      "step": 36002
    },
    {
      "epoch": 2.016071228581028,
      "grad_norm": 1.1697754859924316,
      "learning_rate": 9.579263157894738e-05,
      "loss": 0.4175,
      "step": 36003
    },
    {
      "epoch": 2.016127225893157,
      "grad_norm": 1.0883378982543945,
      "learning_rate": 9.579236842105264e-05,
      "loss": 0.3309,
      "step": 36004
    },
    {
      "epoch": 2.016183223205286,
      "grad_norm": 1.1815637350082397,
      "learning_rate": 9.57921052631579e-05,
      "loss": 0.3602,
      "step": 36005
    },
    {
      "epoch": 2.016239220517415,
      "grad_norm": 1.048217535018921,
      "learning_rate": 9.579184210526316e-05,
      "loss": 0.2999,
      "step": 36006
    },
    {
      "epoch": 2.016295217829544,
      "grad_norm": 1.4316154718399048,
      "learning_rate": 9.579157894736842e-05,
      "loss": 0.608,
      "step": 36007
    },
    {
      "epoch": 2.016351215141673,
      "grad_norm": 1.4802874326705933,
      "learning_rate": 9.579131578947369e-05,
      "loss": 0.4365,
      "step": 36008
    },
    {
      "epoch": 2.016407212453802,
      "grad_norm": 1.0527253150939941,
      "learning_rate": 9.579105263157895e-05,
      "loss": 0.3922,
      "step": 36009
    },
    {
      "epoch": 2.016463209765931,
      "grad_norm": 1.1508678197860718,
      "learning_rate": 9.579078947368422e-05,
      "loss": 0.4145,
      "step": 36010
    },
    {
      "epoch": 2.01651920707806,
      "grad_norm": 1.1785411834716797,
      "learning_rate": 9.579052631578947e-05,
      "loss": 0.481,
      "step": 36011
    },
    {
      "epoch": 2.0165752043901892,
      "grad_norm": 1.0644607543945312,
      "learning_rate": 9.579026315789474e-05,
      "loss": 0.423,
      "step": 36012
    },
    {
      "epoch": 2.0166312017023182,
      "grad_norm": 1.374876856803894,
      "learning_rate": 9.579e-05,
      "loss": 0.3902,
      "step": 36013
    },
    {
      "epoch": 2.0166871990144473,
      "grad_norm": 1.7422996759414673,
      "learning_rate": 9.578973684210528e-05,
      "loss": 0.3413,
      "step": 36014
    },
    {
      "epoch": 2.0167431963265763,
      "grad_norm": 1.0457087755203247,
      "learning_rate": 9.578947368421052e-05,
      "loss": 0.337,
      "step": 36015
    },
    {
      "epoch": 2.0167991936387053,
      "grad_norm": 1.2817370891571045,
      "learning_rate": 9.57892105263158e-05,
      "loss": 0.3393,
      "step": 36016
    },
    {
      "epoch": 2.0168551909508343,
      "grad_norm": 1.215639591217041,
      "learning_rate": 9.578894736842106e-05,
      "loss": 0.3777,
      "step": 36017
    },
    {
      "epoch": 2.0169111882629633,
      "grad_norm": 1.1111069917678833,
      "learning_rate": 9.578868421052633e-05,
      "loss": 0.365,
      "step": 36018
    },
    {
      "epoch": 2.0169671855750924,
      "grad_norm": 1.325737714767456,
      "learning_rate": 9.578842105263159e-05,
      "loss": 0.405,
      "step": 36019
    },
    {
      "epoch": 2.0170231828872214,
      "grad_norm": 1.307184100151062,
      "learning_rate": 9.578815789473685e-05,
      "loss": 0.4497,
      "step": 36020
    },
    {
      "epoch": 2.0170791801993504,
      "grad_norm": 1.1664663553237915,
      "learning_rate": 9.578789473684211e-05,
      "loss": 0.3073,
      "step": 36021
    },
    {
      "epoch": 2.0171351775114794,
      "grad_norm": 1.1457533836364746,
      "learning_rate": 9.578763157894737e-05,
      "loss": 0.4955,
      "step": 36022
    },
    {
      "epoch": 2.0171911748236084,
      "grad_norm": 1.065338373184204,
      "learning_rate": 9.578736842105264e-05,
      "loss": 0.384,
      "step": 36023
    },
    {
      "epoch": 2.0172471721357375,
      "grad_norm": 1.027599573135376,
      "learning_rate": 9.57871052631579e-05,
      "loss": 0.3274,
      "step": 36024
    },
    {
      "epoch": 2.0173031694478665,
      "grad_norm": 1.402746558189392,
      "learning_rate": 9.578684210526316e-05,
      "loss": 0.3838,
      "step": 36025
    },
    {
      "epoch": 2.0173591667599955,
      "grad_norm": 1.1714805364608765,
      "learning_rate": 9.578657894736842e-05,
      "loss": 0.3866,
      "step": 36026
    },
    {
      "epoch": 2.0174151640721245,
      "grad_norm": 1.1373165845870972,
      "learning_rate": 9.578631578947369e-05,
      "loss": 0.3817,
      "step": 36027
    },
    {
      "epoch": 2.0174711613842535,
      "grad_norm": 1.0959217548370361,
      "learning_rate": 9.578605263157895e-05,
      "loss": 0.3602,
      "step": 36028
    },
    {
      "epoch": 2.0175271586963825,
      "grad_norm": 1.457277536392212,
      "learning_rate": 9.578578947368421e-05,
      "loss": 0.5121,
      "step": 36029
    },
    {
      "epoch": 2.0175831560085116,
      "grad_norm": 1.3429263830184937,
      "learning_rate": 9.578552631578947e-05,
      "loss": 0.4155,
      "step": 36030
    },
    {
      "epoch": 2.0176391533206406,
      "grad_norm": 1.2870543003082275,
      "learning_rate": 9.578526315789475e-05,
      "loss": 0.4816,
      "step": 36031
    },
    {
      "epoch": 2.0176951506327696,
      "grad_norm": 1.4878095388412476,
      "learning_rate": 9.5785e-05,
      "loss": 0.5135,
      "step": 36032
    },
    {
      "epoch": 2.0177511479448986,
      "grad_norm": 1.2382071018218994,
      "learning_rate": 9.578473684210527e-05,
      "loss": 0.3768,
      "step": 36033
    },
    {
      "epoch": 2.0178071452570276,
      "grad_norm": 1.2249504327774048,
      "learning_rate": 9.578447368421052e-05,
      "loss": 0.3401,
      "step": 36034
    },
    {
      "epoch": 2.0178631425691567,
      "grad_norm": 1.1332892179489136,
      "learning_rate": 9.57842105263158e-05,
      "loss": 0.3329,
      "step": 36035
    },
    {
      "epoch": 2.0179191398812857,
      "grad_norm": 1.102394700050354,
      "learning_rate": 9.578394736842106e-05,
      "loss": 0.3353,
      "step": 36036
    },
    {
      "epoch": 2.0179751371934147,
      "grad_norm": 1.173794150352478,
      "learning_rate": 9.578368421052633e-05,
      "loss": 0.3378,
      "step": 36037
    },
    {
      "epoch": 2.0180311345055437,
      "grad_norm": 1.0940375328063965,
      "learning_rate": 9.578342105263158e-05,
      "loss": 0.3304,
      "step": 36038
    },
    {
      "epoch": 2.0180871318176727,
      "grad_norm": 1.1070879697799683,
      "learning_rate": 9.578315789473684e-05,
      "loss": 0.2866,
      "step": 36039
    },
    {
      "epoch": 2.0181431291298018,
      "grad_norm": 1.2510130405426025,
      "learning_rate": 9.578289473684211e-05,
      "loss": 0.3679,
      "step": 36040
    },
    {
      "epoch": 2.018199126441931,
      "grad_norm": 1.1776942014694214,
      "learning_rate": 9.578263157894737e-05,
      "loss": 0.3003,
      "step": 36041
    },
    {
      "epoch": 2.01825512375406,
      "grad_norm": 1.3806794881820679,
      "learning_rate": 9.578236842105264e-05,
      "loss": 0.4139,
      "step": 36042
    },
    {
      "epoch": 2.018311121066189,
      "grad_norm": 1.2915743589401245,
      "learning_rate": 9.578210526315789e-05,
      "loss": 0.3807,
      "step": 36043
    },
    {
      "epoch": 2.018367118378318,
      "grad_norm": 1.0765838623046875,
      "learning_rate": 9.578184210526316e-05,
      "loss": 0.3254,
      "step": 36044
    },
    {
      "epoch": 2.018423115690447,
      "grad_norm": 1.1900556087493896,
      "learning_rate": 9.578157894736842e-05,
      "loss": 0.3332,
      "step": 36045
    },
    {
      "epoch": 2.018479113002576,
      "grad_norm": 1.3256704807281494,
      "learning_rate": 9.57813157894737e-05,
      "loss": 0.3772,
      "step": 36046
    },
    {
      "epoch": 2.018535110314705,
      "grad_norm": 1.034179925918579,
      "learning_rate": 9.578105263157896e-05,
      "loss": 0.335,
      "step": 36047
    },
    {
      "epoch": 2.018591107626834,
      "grad_norm": 1.486627221107483,
      "learning_rate": 9.578078947368422e-05,
      "loss": 0.4926,
      "step": 36048
    },
    {
      "epoch": 2.018647104938963,
      "grad_norm": 1.0806629657745361,
      "learning_rate": 9.578052631578947e-05,
      "loss": 0.4818,
      "step": 36049
    },
    {
      "epoch": 2.018703102251092,
      "grad_norm": 1.1111974716186523,
      "learning_rate": 9.578026315789475e-05,
      "loss": 0.3146,
      "step": 36050
    },
    {
      "epoch": 2.018759099563221,
      "grad_norm": 1.6517853736877441,
      "learning_rate": 9.578000000000001e-05,
      "loss": 0.3759,
      "step": 36051
    },
    {
      "epoch": 2.01881509687535,
      "grad_norm": 1.2819726467132568,
      "learning_rate": 9.577973684210527e-05,
      "loss": 0.3637,
      "step": 36052
    },
    {
      "epoch": 2.018871094187479,
      "grad_norm": 1.3480576276779175,
      "learning_rate": 9.577947368421053e-05,
      "loss": 0.4443,
      "step": 36053
    },
    {
      "epoch": 2.018927091499608,
      "grad_norm": 1.1057201623916626,
      "learning_rate": 9.57792105263158e-05,
      "loss": 0.4054,
      "step": 36054
    },
    {
      "epoch": 2.018983088811737,
      "grad_norm": 1.2916324138641357,
      "learning_rate": 9.577894736842106e-05,
      "loss": 0.4222,
      "step": 36055
    },
    {
      "epoch": 2.019039086123866,
      "grad_norm": 1.1482441425323486,
      "learning_rate": 9.577868421052632e-05,
      "loss": 0.3983,
      "step": 36056
    },
    {
      "epoch": 2.019095083435995,
      "grad_norm": 1.3948460817337036,
      "learning_rate": 9.577842105263158e-05,
      "loss": 0.4187,
      "step": 36057
    },
    {
      "epoch": 2.019151080748124,
      "grad_norm": 1.170255422592163,
      "learning_rate": 9.577815789473684e-05,
      "loss": 0.3615,
      "step": 36058
    },
    {
      "epoch": 2.019207078060253,
      "grad_norm": 1.0794289112091064,
      "learning_rate": 9.577789473684211e-05,
      "loss": 0.3487,
      "step": 36059
    },
    {
      "epoch": 2.019263075372382,
      "grad_norm": 1.2683566808700562,
      "learning_rate": 9.577763157894737e-05,
      "loss": 0.3541,
      "step": 36060
    },
    {
      "epoch": 2.019319072684511,
      "grad_norm": 1.210978627204895,
      "learning_rate": 9.577736842105263e-05,
      "loss": 0.359,
      "step": 36061
    },
    {
      "epoch": 2.01937506999664,
      "grad_norm": 1.1657421588897705,
      "learning_rate": 9.577710526315789e-05,
      "loss": 0.3847,
      "step": 36062
    },
    {
      "epoch": 2.019431067308769,
      "grad_norm": 1.3235756158828735,
      "learning_rate": 9.577684210526317e-05,
      "loss": 0.369,
      "step": 36063
    },
    {
      "epoch": 2.019487064620898,
      "grad_norm": 1.171451449394226,
      "learning_rate": 9.577657894736843e-05,
      "loss": 0.4012,
      "step": 36064
    },
    {
      "epoch": 2.0195430619330272,
      "grad_norm": 1.0404460430145264,
      "learning_rate": 9.57763157894737e-05,
      "loss": 0.3322,
      "step": 36065
    },
    {
      "epoch": 2.0195990592451563,
      "grad_norm": 1.0139400959014893,
      "learning_rate": 9.577605263157894e-05,
      "loss": 0.3745,
      "step": 36066
    },
    {
      "epoch": 2.0196550565572853,
      "grad_norm": 1.1442186832427979,
      "learning_rate": 9.577578947368422e-05,
      "loss": 0.464,
      "step": 36067
    },
    {
      "epoch": 2.0197110538694143,
      "grad_norm": 1.219778060913086,
      "learning_rate": 9.577552631578948e-05,
      "loss": 0.4836,
      "step": 36068
    },
    {
      "epoch": 2.0197670511815433,
      "grad_norm": 1.1053717136383057,
      "learning_rate": 9.577526315789475e-05,
      "loss": 0.3311,
      "step": 36069
    },
    {
      "epoch": 2.0198230484936723,
      "grad_norm": 1.0756784677505493,
      "learning_rate": 9.5775e-05,
      "loss": 0.4954,
      "step": 36070
    },
    {
      "epoch": 2.0198790458058014,
      "grad_norm": 1.1235644817352295,
      "learning_rate": 9.577473684210527e-05,
      "loss": 0.325,
      "step": 36071
    },
    {
      "epoch": 2.0199350431179304,
      "grad_norm": 1.427750587463379,
      "learning_rate": 9.577447368421053e-05,
      "loss": 0.3647,
      "step": 36072
    },
    {
      "epoch": 2.0199910404300594,
      "grad_norm": 1.2528737783432007,
      "learning_rate": 9.57742105263158e-05,
      "loss": 0.3711,
      "step": 36073
    },
    {
      "epoch": 2.0200470377421884,
      "grad_norm": 1.7312415838241577,
      "learning_rate": 9.577394736842106e-05,
      "loss": 0.4634,
      "step": 36074
    },
    {
      "epoch": 2.0201030350543174,
      "grad_norm": 1.1633621454238892,
      "learning_rate": 9.577368421052631e-05,
      "loss": 0.2956,
      "step": 36075
    },
    {
      "epoch": 2.0201590323664464,
      "grad_norm": 1.3151518106460571,
      "learning_rate": 9.577342105263158e-05,
      "loss": 0.4186,
      "step": 36076
    },
    {
      "epoch": 2.0202150296785755,
      "grad_norm": 1.3769540786743164,
      "learning_rate": 9.577315789473684e-05,
      "loss": 0.516,
      "step": 36077
    },
    {
      "epoch": 2.0202710269907045,
      "grad_norm": 1.6785788536071777,
      "learning_rate": 9.577289473684212e-05,
      "loss": 0.4451,
      "step": 36078
    },
    {
      "epoch": 2.0203270243028335,
      "grad_norm": 1.1740608215332031,
      "learning_rate": 9.577263157894738e-05,
      "loss": 0.3494,
      "step": 36079
    },
    {
      "epoch": 2.0203830216149625,
      "grad_norm": 1.2428419589996338,
      "learning_rate": 9.577236842105263e-05,
      "loss": 0.3836,
      "step": 36080
    },
    {
      "epoch": 2.0204390189270915,
      "grad_norm": 1.0339562892913818,
      "learning_rate": 9.57721052631579e-05,
      "loss": 0.3294,
      "step": 36081
    },
    {
      "epoch": 2.0204950162392206,
      "grad_norm": 0.9753482937812805,
      "learning_rate": 9.577184210526317e-05,
      "loss": 0.2998,
      "step": 36082
    },
    {
      "epoch": 2.0205510135513496,
      "grad_norm": 1.2974295616149902,
      "learning_rate": 9.577157894736843e-05,
      "loss": 0.4364,
      "step": 36083
    },
    {
      "epoch": 2.0206070108634786,
      "grad_norm": 1.1775233745574951,
      "learning_rate": 9.577131578947369e-05,
      "loss": 0.3805,
      "step": 36084
    },
    {
      "epoch": 2.0206630081756076,
      "grad_norm": 1.050278902053833,
      "learning_rate": 9.577105263157895e-05,
      "loss": 0.3641,
      "step": 36085
    },
    {
      "epoch": 2.0207190054877366,
      "grad_norm": 1.3361601829528809,
      "learning_rate": 9.577078947368422e-05,
      "loss": 0.4103,
      "step": 36086
    },
    {
      "epoch": 2.0207750027998657,
      "grad_norm": 1.1317722797393799,
      "learning_rate": 9.577052631578948e-05,
      "loss": 0.4682,
      "step": 36087
    },
    {
      "epoch": 2.0208310001119947,
      "grad_norm": 1.1627200841903687,
      "learning_rate": 9.577026315789474e-05,
      "loss": 0.5429,
      "step": 36088
    },
    {
      "epoch": 2.0208869974241237,
      "grad_norm": 1.1527284383773804,
      "learning_rate": 9.577e-05,
      "loss": 0.3745,
      "step": 36089
    },
    {
      "epoch": 2.0209429947362527,
      "grad_norm": 1.3439584970474243,
      "learning_rate": 9.576973684210527e-05,
      "loss": 0.4224,
      "step": 36090
    },
    {
      "epoch": 2.0209989920483817,
      "grad_norm": 1.7126879692077637,
      "learning_rate": 9.576947368421053e-05,
      "loss": 0.4834,
      "step": 36091
    },
    {
      "epoch": 2.0210549893605108,
      "grad_norm": 1.350543737411499,
      "learning_rate": 9.576921052631579e-05,
      "loss": 0.4606,
      "step": 36092
    },
    {
      "epoch": 2.0211109866726398,
      "grad_norm": 1.043258786201477,
      "learning_rate": 9.576894736842105e-05,
      "loss": 0.3486,
      "step": 36093
    },
    {
      "epoch": 2.021166983984769,
      "grad_norm": 1.2732752561569214,
      "learning_rate": 9.576868421052631e-05,
      "loss": 0.3951,
      "step": 36094
    },
    {
      "epoch": 2.021222981296898,
      "grad_norm": 1.2452353239059448,
      "learning_rate": 9.576842105263159e-05,
      "loss": 0.4,
      "step": 36095
    },
    {
      "epoch": 2.021278978609027,
      "grad_norm": 1.2646691799163818,
      "learning_rate": 9.576815789473684e-05,
      "loss": 0.3664,
      "step": 36096
    },
    {
      "epoch": 2.021334975921156,
      "grad_norm": 2.23759388923645,
      "learning_rate": 9.576789473684212e-05,
      "loss": 0.4669,
      "step": 36097
    },
    {
      "epoch": 2.021390973233285,
      "grad_norm": 1.27659010887146,
      "learning_rate": 9.576763157894736e-05,
      "loss": 0.3555,
      "step": 36098
    },
    {
      "epoch": 2.021446970545414,
      "grad_norm": 1.1369566917419434,
      "learning_rate": 9.576736842105264e-05,
      "loss": 0.4216,
      "step": 36099
    },
    {
      "epoch": 2.021502967857543,
      "grad_norm": 1.1875971555709839,
      "learning_rate": 9.57671052631579e-05,
      "loss": 0.4602,
      "step": 36100
    },
    {
      "epoch": 2.021558965169672,
      "grad_norm": 1.323009729385376,
      "learning_rate": 9.576684210526317e-05,
      "loss": 0.4474,
      "step": 36101
    },
    {
      "epoch": 2.021614962481801,
      "grad_norm": 1.0964728593826294,
      "learning_rate": 9.576657894736843e-05,
      "loss": 0.3779,
      "step": 36102
    },
    {
      "epoch": 2.02167095979393,
      "grad_norm": 1.629712700843811,
      "learning_rate": 9.576631578947369e-05,
      "loss": 0.4766,
      "step": 36103
    },
    {
      "epoch": 2.021726957106059,
      "grad_norm": 0.9526477456092834,
      "learning_rate": 9.576605263157895e-05,
      "loss": 0.347,
      "step": 36104
    },
    {
      "epoch": 2.021782954418188,
      "grad_norm": 1.4993746280670166,
      "learning_rate": 9.576578947368422e-05,
      "loss": 0.4092,
      "step": 36105
    },
    {
      "epoch": 2.021838951730317,
      "grad_norm": 1.4700255393981934,
      "learning_rate": 9.576552631578948e-05,
      "loss": 0.4275,
      "step": 36106
    },
    {
      "epoch": 2.021894949042446,
      "grad_norm": 1.217809796333313,
      "learning_rate": 9.576526315789474e-05,
      "loss": 0.4027,
      "step": 36107
    },
    {
      "epoch": 2.021950946354575,
      "grad_norm": 1.048000693321228,
      "learning_rate": 9.5765e-05,
      "loss": 0.3954,
      "step": 36108
    },
    {
      "epoch": 2.022006943666704,
      "grad_norm": 1.0968892574310303,
      "learning_rate": 9.576473684210526e-05,
      "loss": 0.377,
      "step": 36109
    },
    {
      "epoch": 2.022062940978833,
      "grad_norm": 1.4301968812942505,
      "learning_rate": 9.576447368421054e-05,
      "loss": 0.381,
      "step": 36110
    },
    {
      "epoch": 2.022118938290962,
      "grad_norm": 1.4264206886291504,
      "learning_rate": 9.57642105263158e-05,
      "loss": 0.4448,
      "step": 36111
    },
    {
      "epoch": 2.022174935603091,
      "grad_norm": 1.0612566471099854,
      "learning_rate": 9.576394736842105e-05,
      "loss": 0.371,
      "step": 36112
    },
    {
      "epoch": 2.02223093291522,
      "grad_norm": 1.2199323177337646,
      "learning_rate": 9.576368421052631e-05,
      "loss": 0.3894,
      "step": 36113
    },
    {
      "epoch": 2.022286930227349,
      "grad_norm": 1.1175472736358643,
      "learning_rate": 9.576342105263159e-05,
      "loss": 0.3095,
      "step": 36114
    },
    {
      "epoch": 2.022342927539478,
      "grad_norm": 1.0955042839050293,
      "learning_rate": 9.576315789473685e-05,
      "loss": 0.4419,
      "step": 36115
    },
    {
      "epoch": 2.022398924851607,
      "grad_norm": 1.4427772760391235,
      "learning_rate": 9.576289473684211e-05,
      "loss": 0.6101,
      "step": 36116
    },
    {
      "epoch": 2.0224549221637362,
      "grad_norm": 1.2829735279083252,
      "learning_rate": 9.576263157894737e-05,
      "loss": 0.4075,
      "step": 36117
    },
    {
      "epoch": 2.0225109194758653,
      "grad_norm": 1.0631839036941528,
      "learning_rate": 9.576236842105264e-05,
      "loss": 0.3797,
      "step": 36118
    },
    {
      "epoch": 2.0225669167879943,
      "grad_norm": 0.9853391051292419,
      "learning_rate": 9.57621052631579e-05,
      "loss": 0.2773,
      "step": 36119
    },
    {
      "epoch": 2.0226229141001233,
      "grad_norm": 1.6444292068481445,
      "learning_rate": 9.576184210526317e-05,
      "loss": 0.4226,
      "step": 36120
    },
    {
      "epoch": 2.0226789114122523,
      "grad_norm": 1.163071632385254,
      "learning_rate": 9.576157894736842e-05,
      "loss": 0.4074,
      "step": 36121
    },
    {
      "epoch": 2.0227349087243813,
      "grad_norm": 1.309867262840271,
      "learning_rate": 9.576131578947369e-05,
      "loss": 0.3962,
      "step": 36122
    },
    {
      "epoch": 2.0227909060365103,
      "grad_norm": 1.4894245862960815,
      "learning_rate": 9.576105263157895e-05,
      "loss": 0.5097,
      "step": 36123
    },
    {
      "epoch": 2.0228469033486394,
      "grad_norm": 1.0927071571350098,
      "learning_rate": 9.576078947368423e-05,
      "loss": 0.2955,
      "step": 36124
    },
    {
      "epoch": 2.0229029006607684,
      "grad_norm": 1.075710415840149,
      "learning_rate": 9.576052631578947e-05,
      "loss": 0.3185,
      "step": 36125
    },
    {
      "epoch": 2.0229588979728974,
      "grad_norm": 1.3109391927719116,
      "learning_rate": 9.576026315789473e-05,
      "loss": 0.4119,
      "step": 36126
    },
    {
      "epoch": 2.0230148952850264,
      "grad_norm": 1.2229167222976685,
      "learning_rate": 9.576e-05,
      "loss": 0.4157,
      "step": 36127
    },
    {
      "epoch": 2.0230708925971554,
      "grad_norm": 1.074326515197754,
      "learning_rate": 9.575973684210526e-05,
      "loss": 0.2969,
      "step": 36128
    },
    {
      "epoch": 2.0231268899092845,
      "grad_norm": 1.3686977624893188,
      "learning_rate": 9.575947368421054e-05,
      "loss": 0.42,
      "step": 36129
    },
    {
      "epoch": 2.0231828872214135,
      "grad_norm": 1.3230191469192505,
      "learning_rate": 9.575921052631578e-05,
      "loss": 0.414,
      "step": 36130
    },
    {
      "epoch": 2.0232388845335425,
      "grad_norm": 1.3710585832595825,
      "learning_rate": 9.575894736842106e-05,
      "loss": 0.4667,
      "step": 36131
    },
    {
      "epoch": 2.0232948818456715,
      "grad_norm": 1.0677279233932495,
      "learning_rate": 9.575868421052632e-05,
      "loss": 0.2917,
      "step": 36132
    },
    {
      "epoch": 2.0233508791578005,
      "grad_norm": 1.244718313217163,
      "learning_rate": 9.575842105263159e-05,
      "loss": 0.4061,
      "step": 36133
    },
    {
      "epoch": 2.0234068764699296,
      "grad_norm": 1.2666012048721313,
      "learning_rate": 9.575815789473685e-05,
      "loss": 0.5106,
      "step": 36134
    },
    {
      "epoch": 2.0234628737820586,
      "grad_norm": 1.3248134851455688,
      "learning_rate": 9.575789473684211e-05,
      "loss": 0.4333,
      "step": 36135
    },
    {
      "epoch": 2.0235188710941876,
      "grad_norm": 1.3805012702941895,
      "learning_rate": 9.575763157894737e-05,
      "loss": 0.4819,
      "step": 36136
    },
    {
      "epoch": 2.0235748684063166,
      "grad_norm": 1.1341010332107544,
      "learning_rate": 9.575736842105264e-05,
      "loss": 0.4437,
      "step": 36137
    },
    {
      "epoch": 2.0236308657184456,
      "grad_norm": 1.5677810907363892,
      "learning_rate": 9.57571052631579e-05,
      "loss": 0.3928,
      "step": 36138
    },
    {
      "epoch": 2.0236868630305747,
      "grad_norm": 1.3011412620544434,
      "learning_rate": 9.575684210526316e-05,
      "loss": 0.4063,
      "step": 36139
    },
    {
      "epoch": 2.0237428603427037,
      "grad_norm": 1.318723201751709,
      "learning_rate": 9.575657894736842e-05,
      "loss": 0.4089,
      "step": 36140
    },
    {
      "epoch": 2.0237988576548327,
      "grad_norm": 1.2499979734420776,
      "learning_rate": 9.57563157894737e-05,
      "loss": 0.3874,
      "step": 36141
    },
    {
      "epoch": 2.0238548549669617,
      "grad_norm": 1.162208080291748,
      "learning_rate": 9.575605263157895e-05,
      "loss": 0.3703,
      "step": 36142
    },
    {
      "epoch": 2.0239108522790907,
      "grad_norm": 1.1510460376739502,
      "learning_rate": 9.575578947368421e-05,
      "loss": 0.4627,
      "step": 36143
    },
    {
      "epoch": 2.0239668495912198,
      "grad_norm": 1.3512641191482544,
      "learning_rate": 9.575552631578947e-05,
      "loss": 0.544,
      "step": 36144
    },
    {
      "epoch": 2.0240228469033488,
      "grad_norm": 1.150254249572754,
      "learning_rate": 9.575526315789473e-05,
      "loss": 0.5287,
      "step": 36145
    },
    {
      "epoch": 2.024078844215478,
      "grad_norm": 1.109472632408142,
      "learning_rate": 9.575500000000001e-05,
      "loss": 0.3934,
      "step": 36146
    },
    {
      "epoch": 2.024134841527607,
      "grad_norm": 1.3962764739990234,
      "learning_rate": 9.575473684210527e-05,
      "loss": 0.3851,
      "step": 36147
    },
    {
      "epoch": 2.024190838839736,
      "grad_norm": 1.1645554304122925,
      "learning_rate": 9.575447368421053e-05,
      "loss": 0.3797,
      "step": 36148
    },
    {
      "epoch": 2.024246836151865,
      "grad_norm": 1.3559650182724,
      "learning_rate": 9.575421052631579e-05,
      "loss": 0.4167,
      "step": 36149
    },
    {
      "epoch": 2.024302833463994,
      "grad_norm": 1.0167299509048462,
      "learning_rate": 9.575394736842106e-05,
      "loss": 0.395,
      "step": 36150
    },
    {
      "epoch": 2.024358830776123,
      "grad_norm": 0.9587291479110718,
      "learning_rate": 9.575368421052632e-05,
      "loss": 0.3134,
      "step": 36151
    },
    {
      "epoch": 2.024414828088252,
      "grad_norm": 1.1310670375823975,
      "learning_rate": 9.575342105263159e-05,
      "loss": 0.3498,
      "step": 36152
    },
    {
      "epoch": 2.024470825400381,
      "grad_norm": 1.0973423719406128,
      "learning_rate": 9.575315789473684e-05,
      "loss": 0.3597,
      "step": 36153
    },
    {
      "epoch": 2.02452682271251,
      "grad_norm": 0.9995084404945374,
      "learning_rate": 9.575289473684211e-05,
      "loss": 0.352,
      "step": 36154
    },
    {
      "epoch": 2.024582820024639,
      "grad_norm": 1.0361990928649902,
      "learning_rate": 9.575263157894737e-05,
      "loss": 0.3934,
      "step": 36155
    },
    {
      "epoch": 2.024638817336768,
      "grad_norm": 1.2091307640075684,
      "learning_rate": 9.575236842105265e-05,
      "loss": 0.5091,
      "step": 36156
    },
    {
      "epoch": 2.024694814648897,
      "grad_norm": 1.361951231956482,
      "learning_rate": 9.57521052631579e-05,
      "loss": 0.6277,
      "step": 36157
    },
    {
      "epoch": 2.024750811961026,
      "grad_norm": 1.2040526866912842,
      "learning_rate": 9.575184210526316e-05,
      "loss": 0.3802,
      "step": 36158
    },
    {
      "epoch": 2.024806809273155,
      "grad_norm": 1.1696475744247437,
      "learning_rate": 9.575157894736842e-05,
      "loss": 0.3907,
      "step": 36159
    },
    {
      "epoch": 2.024862806585284,
      "grad_norm": 1.2047109603881836,
      "learning_rate": 9.575131578947368e-05,
      "loss": 0.3811,
      "step": 36160
    },
    {
      "epoch": 2.024918803897413,
      "grad_norm": 1.2403868436813354,
      "learning_rate": 9.575105263157896e-05,
      "loss": 0.4177,
      "step": 36161
    },
    {
      "epoch": 2.024974801209542,
      "grad_norm": 1.7188231945037842,
      "learning_rate": 9.57507894736842e-05,
      "loss": 0.4631,
      "step": 36162
    },
    {
      "epoch": 2.025030798521671,
      "grad_norm": 1.296467661857605,
      "learning_rate": 9.575052631578948e-05,
      "loss": 0.4341,
      "step": 36163
    },
    {
      "epoch": 2.0250867958338,
      "grad_norm": 1.1680549383163452,
      "learning_rate": 9.575026315789474e-05,
      "loss": 0.4633,
      "step": 36164
    },
    {
      "epoch": 2.025142793145929,
      "grad_norm": 1.1548320055007935,
      "learning_rate": 9.575000000000001e-05,
      "loss": 0.3898,
      "step": 36165
    },
    {
      "epoch": 2.025198790458058,
      "grad_norm": 1.3048478364944458,
      "learning_rate": 9.574973684210527e-05,
      "loss": 0.3727,
      "step": 36166
    },
    {
      "epoch": 2.025254787770187,
      "grad_norm": 0.9240673780441284,
      "learning_rate": 9.574947368421053e-05,
      "loss": 0.2686,
      "step": 36167
    },
    {
      "epoch": 2.025310785082316,
      "grad_norm": 1.121976613998413,
      "learning_rate": 9.574921052631579e-05,
      "loss": 0.3149,
      "step": 36168
    },
    {
      "epoch": 2.0253667823944452,
      "grad_norm": 1.2317181825637817,
      "learning_rate": 9.574894736842106e-05,
      "loss": 0.3576,
      "step": 36169
    },
    {
      "epoch": 2.0254227797065743,
      "grad_norm": 1.183701515197754,
      "learning_rate": 9.574868421052632e-05,
      "loss": 0.4173,
      "step": 36170
    },
    {
      "epoch": 2.0254787770187033,
      "grad_norm": 1.2518954277038574,
      "learning_rate": 9.574842105263158e-05,
      "loss": 0.3794,
      "step": 36171
    },
    {
      "epoch": 2.0255347743308323,
      "grad_norm": 1.0745656490325928,
      "learning_rate": 9.574815789473684e-05,
      "loss": 0.3689,
      "step": 36172
    },
    {
      "epoch": 2.0255907716429613,
      "grad_norm": 1.2014179229736328,
      "learning_rate": 9.574789473684211e-05,
      "loss": 0.4485,
      "step": 36173
    },
    {
      "epoch": 2.0256467689550903,
      "grad_norm": 1.394370675086975,
      "learning_rate": 9.574763157894737e-05,
      "loss": 0.5683,
      "step": 36174
    },
    {
      "epoch": 2.0257027662672193,
      "grad_norm": 1.0620108842849731,
      "learning_rate": 9.574736842105265e-05,
      "loss": 0.306,
      "step": 36175
    },
    {
      "epoch": 2.0257587635793484,
      "grad_norm": 0.99882572889328,
      "learning_rate": 9.57471052631579e-05,
      "loss": 0.3377,
      "step": 36176
    },
    {
      "epoch": 2.0258147608914774,
      "grad_norm": 1.4434428215026855,
      "learning_rate": 9.574684210526315e-05,
      "loss": 0.3261,
      "step": 36177
    },
    {
      "epoch": 2.0258707582036064,
      "grad_norm": 1.2017641067504883,
      "learning_rate": 9.574657894736843e-05,
      "loss": 0.3859,
      "step": 36178
    },
    {
      "epoch": 2.0259267555157354,
      "grad_norm": 1.434136986732483,
      "learning_rate": 9.574631578947369e-05,
      "loss": 0.3284,
      "step": 36179
    },
    {
      "epoch": 2.0259827528278644,
      "grad_norm": 1.265540361404419,
      "learning_rate": 9.574605263157895e-05,
      "loss": 0.4311,
      "step": 36180
    },
    {
      "epoch": 2.0260387501399935,
      "grad_norm": 1.2686388492584229,
      "learning_rate": 9.57457894736842e-05,
      "loss": 0.3066,
      "step": 36181
    },
    {
      "epoch": 2.0260947474521225,
      "grad_norm": 1.158517837524414,
      "learning_rate": 9.574552631578948e-05,
      "loss": 0.3474,
      "step": 36182
    },
    {
      "epoch": 2.0261507447642515,
      "grad_norm": 1.011753797531128,
      "learning_rate": 9.574526315789474e-05,
      "loss": 0.328,
      "step": 36183
    },
    {
      "epoch": 2.0262067420763805,
      "grad_norm": 0.9872798323631287,
      "learning_rate": 9.574500000000001e-05,
      "loss": 0.4186,
      "step": 36184
    },
    {
      "epoch": 2.0262627393885095,
      "grad_norm": 1.162466287612915,
      "learning_rate": 9.574473684210526e-05,
      "loss": 0.4061,
      "step": 36185
    },
    {
      "epoch": 2.0263187367006386,
      "grad_norm": 1.0803300142288208,
      "learning_rate": 9.574447368421053e-05,
      "loss": 0.3805,
      "step": 36186
    },
    {
      "epoch": 2.0263747340127676,
      "grad_norm": 1.0123308897018433,
      "learning_rate": 9.574421052631579e-05,
      "loss": 0.3352,
      "step": 36187
    },
    {
      "epoch": 2.0264307313248966,
      "grad_norm": 1.1157196760177612,
      "learning_rate": 9.574394736842106e-05,
      "loss": 0.4009,
      "step": 36188
    },
    {
      "epoch": 2.0264867286370256,
      "grad_norm": 1.3922655582427979,
      "learning_rate": 9.574368421052632e-05,
      "loss": 0.588,
      "step": 36189
    },
    {
      "epoch": 2.0265427259491546,
      "grad_norm": 2.8528687953948975,
      "learning_rate": 9.574342105263158e-05,
      "loss": 0.4321,
      "step": 36190
    },
    {
      "epoch": 2.0265987232612837,
      "grad_norm": 1.2426711320877075,
      "learning_rate": 9.574315789473684e-05,
      "loss": 0.3454,
      "step": 36191
    },
    {
      "epoch": 2.0266547205734127,
      "grad_norm": 1.2248778343200684,
      "learning_rate": 9.574289473684212e-05,
      "loss": 0.4249,
      "step": 36192
    },
    {
      "epoch": 2.0267107178855417,
      "grad_norm": 1.218500018119812,
      "learning_rate": 9.574263157894738e-05,
      "loss": 0.4198,
      "step": 36193
    },
    {
      "epoch": 2.0267667151976707,
      "grad_norm": 1.039161205291748,
      "learning_rate": 9.574236842105264e-05,
      "loss": 0.327,
      "step": 36194
    },
    {
      "epoch": 2.0268227125097997,
      "grad_norm": 1.3586393594741821,
      "learning_rate": 9.57421052631579e-05,
      "loss": 0.4643,
      "step": 36195
    },
    {
      "epoch": 2.0268787098219287,
      "grad_norm": 1.3972834348678589,
      "learning_rate": 9.574184210526316e-05,
      "loss": 0.5018,
      "step": 36196
    },
    {
      "epoch": 2.0269347071340578,
      "grad_norm": 1.2016923427581787,
      "learning_rate": 9.574157894736843e-05,
      "loss": 0.4195,
      "step": 36197
    },
    {
      "epoch": 2.026990704446187,
      "grad_norm": 1.1543623208999634,
      "learning_rate": 9.574131578947369e-05,
      "loss": 0.5065,
      "step": 36198
    },
    {
      "epoch": 2.027046701758316,
      "grad_norm": 1.2000309228897095,
      "learning_rate": 9.574105263157895e-05,
      "loss": 0.4289,
      "step": 36199
    },
    {
      "epoch": 2.027102699070445,
      "grad_norm": 1.1484936475753784,
      "learning_rate": 9.574078947368421e-05,
      "loss": 0.3308,
      "step": 36200
    },
    {
      "epoch": 2.027158696382574,
      "grad_norm": 1.168464183807373,
      "learning_rate": 9.574052631578948e-05,
      "loss": 0.4177,
      "step": 36201
    },
    {
      "epoch": 2.027214693694703,
      "grad_norm": 1.098473072052002,
      "learning_rate": 9.574026315789474e-05,
      "loss": 0.3013,
      "step": 36202
    },
    {
      "epoch": 2.027270691006832,
      "grad_norm": 1.334673285484314,
      "learning_rate": 9.574e-05,
      "loss": 0.4392,
      "step": 36203
    },
    {
      "epoch": 2.027326688318961,
      "grad_norm": 1.0094263553619385,
      "learning_rate": 9.573973684210526e-05,
      "loss": 0.2937,
      "step": 36204
    },
    {
      "epoch": 2.02738268563109,
      "grad_norm": 1.306365728378296,
      "learning_rate": 9.573947368421053e-05,
      "loss": 0.3798,
      "step": 36205
    },
    {
      "epoch": 2.027438682943219,
      "grad_norm": 1.7061251401901245,
      "learning_rate": 9.57392105263158e-05,
      "loss": 0.4361,
      "step": 36206
    },
    {
      "epoch": 2.027494680255348,
      "grad_norm": 1.1141397953033447,
      "learning_rate": 9.573894736842107e-05,
      "loss": 0.4646,
      "step": 36207
    },
    {
      "epoch": 2.027550677567477,
      "grad_norm": 1.0769239664077759,
      "learning_rate": 9.573868421052631e-05,
      "loss": 0.4355,
      "step": 36208
    },
    {
      "epoch": 2.027606674879606,
      "grad_norm": 1.003715991973877,
      "learning_rate": 9.573842105263159e-05,
      "loss": 0.4171,
      "step": 36209
    },
    {
      "epoch": 2.027662672191735,
      "grad_norm": 1.2618032693862915,
      "learning_rate": 9.573815789473685e-05,
      "loss": 0.4016,
      "step": 36210
    },
    {
      "epoch": 2.027718669503864,
      "grad_norm": 1.1518477201461792,
      "learning_rate": 9.573789473684212e-05,
      "loss": 0.2621,
      "step": 36211
    },
    {
      "epoch": 2.0277746668159926,
      "grad_norm": 1.785696029663086,
      "learning_rate": 9.573763157894738e-05,
      "loss": 0.4088,
      "step": 36212
    },
    {
      "epoch": 2.027830664128122,
      "grad_norm": 1.5694187879562378,
      "learning_rate": 9.573736842105263e-05,
      "loss": 0.3665,
      "step": 36213
    },
    {
      "epoch": 2.0278866614402506,
      "grad_norm": 1.0264112949371338,
      "learning_rate": 9.57371052631579e-05,
      "loss": 0.3302,
      "step": 36214
    },
    {
      "epoch": 2.0279426587523797,
      "grad_norm": 1.345012903213501,
      "learning_rate": 9.573684210526316e-05,
      "loss": 0.4916,
      "step": 36215
    },
    {
      "epoch": 2.0279986560645087,
      "grad_norm": 1.4593931436538696,
      "learning_rate": 9.573657894736843e-05,
      "loss": 0.6602,
      "step": 36216
    },
    {
      "epoch": 2.0280546533766377,
      "grad_norm": 1.0988857746124268,
      "learning_rate": 9.573631578947368e-05,
      "loss": 0.4485,
      "step": 36217
    },
    {
      "epoch": 2.0281106506887667,
      "grad_norm": 1.1506752967834473,
      "learning_rate": 9.573605263157895e-05,
      "loss": 0.3753,
      "step": 36218
    },
    {
      "epoch": 2.0281666480008957,
      "grad_norm": 1.1675409078598022,
      "learning_rate": 9.573578947368421e-05,
      "loss": 0.3075,
      "step": 36219
    },
    {
      "epoch": 2.0282226453130248,
      "grad_norm": 1.2139896154403687,
      "learning_rate": 9.573552631578948e-05,
      "loss": 0.376,
      "step": 36220
    },
    {
      "epoch": 2.028278642625154,
      "grad_norm": 1.1897326707839966,
      "learning_rate": 9.573526315789474e-05,
      "loss": 0.5029,
      "step": 36221
    },
    {
      "epoch": 2.028334639937283,
      "grad_norm": 1.5655843019485474,
      "learning_rate": 9.5735e-05,
      "loss": 0.3787,
      "step": 36222
    },
    {
      "epoch": 2.028390637249412,
      "grad_norm": 1.1911202669143677,
      "learning_rate": 9.573473684210526e-05,
      "loss": 0.3281,
      "step": 36223
    },
    {
      "epoch": 2.028446634561541,
      "grad_norm": 1.3864094018936157,
      "learning_rate": 9.573447368421054e-05,
      "loss": 0.4893,
      "step": 36224
    },
    {
      "epoch": 2.02850263187367,
      "grad_norm": 1.345996379852295,
      "learning_rate": 9.57342105263158e-05,
      "loss": 0.4782,
      "step": 36225
    },
    {
      "epoch": 2.028558629185799,
      "grad_norm": 1.1793789863586426,
      "learning_rate": 9.573394736842106e-05,
      "loss": 0.5286,
      "step": 36226
    },
    {
      "epoch": 2.028614626497928,
      "grad_norm": 1.0208584070205688,
      "learning_rate": 9.573368421052632e-05,
      "loss": 0.4353,
      "step": 36227
    },
    {
      "epoch": 2.028670623810057,
      "grad_norm": 1.3628425598144531,
      "learning_rate": 9.573342105263159e-05,
      "loss": 0.4194,
      "step": 36228
    },
    {
      "epoch": 2.028726621122186,
      "grad_norm": 1.5805126428604126,
      "learning_rate": 9.573315789473685e-05,
      "loss": 0.469,
      "step": 36229
    },
    {
      "epoch": 2.028782618434315,
      "grad_norm": 0.9696323275566101,
      "learning_rate": 9.573289473684211e-05,
      "loss": 0.2908,
      "step": 36230
    },
    {
      "epoch": 2.028838615746444,
      "grad_norm": 1.1465381383895874,
      "learning_rate": 9.573263157894737e-05,
      "loss": 0.3607,
      "step": 36231
    },
    {
      "epoch": 2.028894613058573,
      "grad_norm": 1.2499195337295532,
      "learning_rate": 9.573236842105263e-05,
      "loss": 0.6442,
      "step": 36232
    },
    {
      "epoch": 2.028950610370702,
      "grad_norm": 1.4390406608581543,
      "learning_rate": 9.57321052631579e-05,
      "loss": 0.4236,
      "step": 36233
    },
    {
      "epoch": 2.029006607682831,
      "grad_norm": 1.2855228185653687,
      "learning_rate": 9.573184210526316e-05,
      "loss": 0.4247,
      "step": 36234
    },
    {
      "epoch": 2.02906260499496,
      "grad_norm": 1.1419548988342285,
      "learning_rate": 9.573157894736842e-05,
      "loss": 0.4055,
      "step": 36235
    },
    {
      "epoch": 2.029118602307089,
      "grad_norm": 0.9791093468666077,
      "learning_rate": 9.573131578947368e-05,
      "loss": 0.3513,
      "step": 36236
    },
    {
      "epoch": 2.029174599619218,
      "grad_norm": 1.3781180381774902,
      "learning_rate": 9.573105263157895e-05,
      "loss": 0.4697,
      "step": 36237
    },
    {
      "epoch": 2.029230596931347,
      "grad_norm": 1.1535961627960205,
      "learning_rate": 9.573078947368421e-05,
      "loss": 0.3634,
      "step": 36238
    },
    {
      "epoch": 2.029286594243476,
      "grad_norm": 3.7169346809387207,
      "learning_rate": 9.573052631578949e-05,
      "loss": 0.3134,
      "step": 36239
    },
    {
      "epoch": 2.029342591555605,
      "grad_norm": 1.3666712045669556,
      "learning_rate": 9.573026315789473e-05,
      "loss": 0.3581,
      "step": 36240
    },
    {
      "epoch": 2.029398588867734,
      "grad_norm": 1.0228607654571533,
      "learning_rate": 9.573e-05,
      "loss": 0.3921,
      "step": 36241
    },
    {
      "epoch": 2.029454586179863,
      "grad_norm": 1.2561143636703491,
      "learning_rate": 9.572973684210527e-05,
      "loss": 0.3713,
      "step": 36242
    },
    {
      "epoch": 2.029510583491992,
      "grad_norm": 1.2982616424560547,
      "learning_rate": 9.572947368421054e-05,
      "loss": 0.4248,
      "step": 36243
    },
    {
      "epoch": 2.0295665808041212,
      "grad_norm": 1.1033931970596313,
      "learning_rate": 9.57292105263158e-05,
      "loss": 0.4484,
      "step": 36244
    },
    {
      "epoch": 2.0296225781162502,
      "grad_norm": 1.2534352540969849,
      "learning_rate": 9.572894736842106e-05,
      "loss": 0.3661,
      "step": 36245
    },
    {
      "epoch": 2.0296785754283793,
      "grad_norm": 1.1837644577026367,
      "learning_rate": 9.572868421052632e-05,
      "loss": 0.4713,
      "step": 36246
    },
    {
      "epoch": 2.0297345727405083,
      "grad_norm": 1.0372940301895142,
      "learning_rate": 9.572842105263158e-05,
      "loss": 0.4233,
      "step": 36247
    },
    {
      "epoch": 2.0297905700526373,
      "grad_norm": 1.0914605855941772,
      "learning_rate": 9.572815789473685e-05,
      "loss": 0.4265,
      "step": 36248
    },
    {
      "epoch": 2.0298465673647663,
      "grad_norm": 0.9688382744789124,
      "learning_rate": 9.572789473684211e-05,
      "loss": 0.3353,
      "step": 36249
    },
    {
      "epoch": 2.0299025646768953,
      "grad_norm": 1.4337506294250488,
      "learning_rate": 9.572763157894737e-05,
      "loss": 0.3795,
      "step": 36250
    },
    {
      "epoch": 2.0299585619890244,
      "grad_norm": 1.0688773393630981,
      "learning_rate": 9.572736842105263e-05,
      "loss": 0.3368,
      "step": 36251
    },
    {
      "epoch": 2.0300145593011534,
      "grad_norm": 1.1917976140975952,
      "learning_rate": 9.57271052631579e-05,
      "loss": 0.3923,
      "step": 36252
    },
    {
      "epoch": 2.0300705566132824,
      "grad_norm": 1.224181056022644,
      "learning_rate": 9.572684210526316e-05,
      "loss": 0.3594,
      "step": 36253
    },
    {
      "epoch": 2.0301265539254114,
      "grad_norm": 1.2333035469055176,
      "learning_rate": 9.572657894736842e-05,
      "loss": 0.4228,
      "step": 36254
    },
    {
      "epoch": 2.0301825512375404,
      "grad_norm": 1.189501404762268,
      "learning_rate": 9.572631578947368e-05,
      "loss": 0.496,
      "step": 36255
    },
    {
      "epoch": 2.0302385485496695,
      "grad_norm": 1.214761734008789,
      "learning_rate": 9.572605263157896e-05,
      "loss": 0.4171,
      "step": 36256
    },
    {
      "epoch": 2.0302945458617985,
      "grad_norm": 1.0537689924240112,
      "learning_rate": 9.572578947368422e-05,
      "loss": 0.3346,
      "step": 36257
    },
    {
      "epoch": 2.0303505431739275,
      "grad_norm": 1.2068097591400146,
      "learning_rate": 9.572552631578948e-05,
      "loss": 0.3823,
      "step": 36258
    },
    {
      "epoch": 2.0304065404860565,
      "grad_norm": 1.2687188386917114,
      "learning_rate": 9.572526315789474e-05,
      "loss": 0.446,
      "step": 36259
    },
    {
      "epoch": 2.0304625377981855,
      "grad_norm": 1.2734394073486328,
      "learning_rate": 9.572500000000001e-05,
      "loss": 0.3045,
      "step": 36260
    },
    {
      "epoch": 2.0305185351103145,
      "grad_norm": 1.3959215879440308,
      "learning_rate": 9.572473684210527e-05,
      "loss": 0.3452,
      "step": 36261
    },
    {
      "epoch": 2.0305745324224436,
      "grad_norm": 1.2616201639175415,
      "learning_rate": 9.572447368421054e-05,
      "loss": 0.3113,
      "step": 36262
    },
    {
      "epoch": 2.0306305297345726,
      "grad_norm": 1.3425836563110352,
      "learning_rate": 9.572421052631579e-05,
      "loss": 0.3892,
      "step": 36263
    },
    {
      "epoch": 2.0306865270467016,
      "grad_norm": 0.9696201086044312,
      "learning_rate": 9.572394736842105e-05,
      "loss": 0.3914,
      "step": 36264
    },
    {
      "epoch": 2.0307425243588306,
      "grad_norm": 1.2023886442184448,
      "learning_rate": 9.572368421052632e-05,
      "loss": 0.3857,
      "step": 36265
    },
    {
      "epoch": 2.0307985216709596,
      "grad_norm": 1.0438746213912964,
      "learning_rate": 9.572342105263158e-05,
      "loss": 0.4836,
      "step": 36266
    },
    {
      "epoch": 2.0308545189830887,
      "grad_norm": 1.060734510421753,
      "learning_rate": 9.572315789473685e-05,
      "loss": 0.3443,
      "step": 36267
    },
    {
      "epoch": 2.0309105162952177,
      "grad_norm": 1.1171808242797852,
      "learning_rate": 9.57228947368421e-05,
      "loss": 0.5194,
      "step": 36268
    },
    {
      "epoch": 2.0309665136073467,
      "grad_norm": 1.2432234287261963,
      "learning_rate": 9.572263157894737e-05,
      "loss": 0.3651,
      "step": 36269
    },
    {
      "epoch": 2.0310225109194757,
      "grad_norm": 1.3481745719909668,
      "learning_rate": 9.572236842105263e-05,
      "loss": 0.4087,
      "step": 36270
    },
    {
      "epoch": 2.0310785082316047,
      "grad_norm": 1.2913042306900024,
      "learning_rate": 9.572210526315791e-05,
      "loss": 0.3931,
      "step": 36271
    },
    {
      "epoch": 2.0311345055437338,
      "grad_norm": 1.2349194288253784,
      "learning_rate": 9.572184210526315e-05,
      "loss": 0.4094,
      "step": 36272
    },
    {
      "epoch": 2.031190502855863,
      "grad_norm": 1.427011489868164,
      "learning_rate": 9.572157894736843e-05,
      "loss": 0.5283,
      "step": 36273
    },
    {
      "epoch": 2.031246500167992,
      "grad_norm": 1.7662633657455444,
      "learning_rate": 9.572131578947369e-05,
      "loss": 0.5626,
      "step": 36274
    },
    {
      "epoch": 2.031302497480121,
      "grad_norm": 1.3289101123809814,
      "learning_rate": 9.572105263157896e-05,
      "loss": 0.459,
      "step": 36275
    },
    {
      "epoch": 2.03135849479225,
      "grad_norm": 1.0823477506637573,
      "learning_rate": 9.572078947368422e-05,
      "loss": 0.3981,
      "step": 36276
    },
    {
      "epoch": 2.031414492104379,
      "grad_norm": 1.23638916015625,
      "learning_rate": 9.572052631578948e-05,
      "loss": 0.4047,
      "step": 36277
    },
    {
      "epoch": 2.031470489416508,
      "grad_norm": 1.3539667129516602,
      "learning_rate": 9.572026315789474e-05,
      "loss": 0.4088,
      "step": 36278
    },
    {
      "epoch": 2.031526486728637,
      "grad_norm": 1.067913293838501,
      "learning_rate": 9.572000000000001e-05,
      "loss": 0.4061,
      "step": 36279
    },
    {
      "epoch": 2.031582484040766,
      "grad_norm": 1.0562572479248047,
      "learning_rate": 9.571973684210527e-05,
      "loss": 0.4266,
      "step": 36280
    },
    {
      "epoch": 2.031638481352895,
      "grad_norm": 1.4439635276794434,
      "learning_rate": 9.571947368421053e-05,
      "loss": 0.3692,
      "step": 36281
    },
    {
      "epoch": 2.031694478665024,
      "grad_norm": 1.1870864629745483,
      "learning_rate": 9.571921052631579e-05,
      "loss": 0.4067,
      "step": 36282
    },
    {
      "epoch": 2.031750475977153,
      "grad_norm": 1.0863558053970337,
      "learning_rate": 9.571894736842105e-05,
      "loss": 0.3113,
      "step": 36283
    },
    {
      "epoch": 2.031806473289282,
      "grad_norm": 1.1814287900924683,
      "learning_rate": 9.571868421052632e-05,
      "loss": 0.4776,
      "step": 36284
    },
    {
      "epoch": 2.031862470601411,
      "grad_norm": 1.1033085584640503,
      "learning_rate": 9.571842105263158e-05,
      "loss": 0.36,
      "step": 36285
    },
    {
      "epoch": 2.03191846791354,
      "grad_norm": 1.1536273956298828,
      "learning_rate": 9.571815789473684e-05,
      "loss": 0.4738,
      "step": 36286
    },
    {
      "epoch": 2.031974465225669,
      "grad_norm": 1.1841963529586792,
      "learning_rate": 9.57178947368421e-05,
      "loss": 0.4056,
      "step": 36287
    },
    {
      "epoch": 2.032030462537798,
      "grad_norm": 0.9759486317634583,
      "learning_rate": 9.571763157894738e-05,
      "loss": 0.3167,
      "step": 36288
    },
    {
      "epoch": 2.032086459849927,
      "grad_norm": 1.9350852966308594,
      "learning_rate": 9.571736842105264e-05,
      "loss": 0.6437,
      "step": 36289
    },
    {
      "epoch": 2.032142457162056,
      "grad_norm": 1.4433387517929077,
      "learning_rate": 9.57171052631579e-05,
      "loss": 0.4003,
      "step": 36290
    },
    {
      "epoch": 2.032198454474185,
      "grad_norm": 1.247506856918335,
      "learning_rate": 9.571684210526316e-05,
      "loss": 0.3645,
      "step": 36291
    },
    {
      "epoch": 2.032254451786314,
      "grad_norm": 1.0808452367782593,
      "learning_rate": 9.571657894736843e-05,
      "loss": 0.3193,
      "step": 36292
    },
    {
      "epoch": 2.032310449098443,
      "grad_norm": 1.2305854558944702,
      "learning_rate": 9.571631578947369e-05,
      "loss": 0.348,
      "step": 36293
    },
    {
      "epoch": 2.032366446410572,
      "grad_norm": 1.2643438577651978,
      "learning_rate": 9.571605263157896e-05,
      "loss": 0.5225,
      "step": 36294
    },
    {
      "epoch": 2.032422443722701,
      "grad_norm": 1.1478748321533203,
      "learning_rate": 9.571578947368421e-05,
      "loss": 0.3745,
      "step": 36295
    },
    {
      "epoch": 2.03247844103483,
      "grad_norm": 1.0920768976211548,
      "learning_rate": 9.571552631578948e-05,
      "loss": 0.3905,
      "step": 36296
    },
    {
      "epoch": 2.0325344383469592,
      "grad_norm": 1.1317269802093506,
      "learning_rate": 9.571526315789474e-05,
      "loss": 0.458,
      "step": 36297
    },
    {
      "epoch": 2.0325904356590883,
      "grad_norm": 0.9771817922592163,
      "learning_rate": 9.5715e-05,
      "loss": 0.3913,
      "step": 36298
    },
    {
      "epoch": 2.0326464329712173,
      "grad_norm": 1.0951465368270874,
      "learning_rate": 9.571473684210527e-05,
      "loss": 0.3489,
      "step": 36299
    },
    {
      "epoch": 2.0327024302833463,
      "grad_norm": 1.08945631980896,
      "learning_rate": 9.571447368421052e-05,
      "loss": 0.4503,
      "step": 36300
    },
    {
      "epoch": 2.0327584275954753,
      "grad_norm": 1.3406882286071777,
      "learning_rate": 9.57142105263158e-05,
      "loss": 0.3598,
      "step": 36301
    },
    {
      "epoch": 2.0328144249076043,
      "grad_norm": 1.3288211822509766,
      "learning_rate": 9.571394736842105e-05,
      "loss": 0.3867,
      "step": 36302
    },
    {
      "epoch": 2.0328704222197334,
      "grad_norm": 1.317568302154541,
      "learning_rate": 9.571368421052633e-05,
      "loss": 0.4894,
      "step": 36303
    },
    {
      "epoch": 2.0329264195318624,
      "grad_norm": 1.071881651878357,
      "learning_rate": 9.571342105263159e-05,
      "loss": 0.4839,
      "step": 36304
    },
    {
      "epoch": 2.0329824168439914,
      "grad_norm": 1.4124839305877686,
      "learning_rate": 9.571315789473685e-05,
      "loss": 0.4885,
      "step": 36305
    },
    {
      "epoch": 2.0330384141561204,
      "grad_norm": 1.169803261756897,
      "learning_rate": 9.57128947368421e-05,
      "loss": 0.4676,
      "step": 36306
    },
    {
      "epoch": 2.0330944114682494,
      "grad_norm": 1.0720146894454956,
      "learning_rate": 9.571263157894738e-05,
      "loss": 0.2855,
      "step": 36307
    },
    {
      "epoch": 2.0331504087803784,
      "grad_norm": 1.2608129978179932,
      "learning_rate": 9.571236842105264e-05,
      "loss": 0.3826,
      "step": 36308
    },
    {
      "epoch": 2.0332064060925075,
      "grad_norm": 1.0856701135635376,
      "learning_rate": 9.57121052631579e-05,
      "loss": 0.322,
      "step": 36309
    },
    {
      "epoch": 2.0332624034046365,
      "grad_norm": 1.2446951866149902,
      "learning_rate": 9.571184210526316e-05,
      "loss": 0.317,
      "step": 36310
    },
    {
      "epoch": 2.0333184007167655,
      "grad_norm": 1.3527703285217285,
      "learning_rate": 9.571157894736843e-05,
      "loss": 0.3809,
      "step": 36311
    },
    {
      "epoch": 2.0333743980288945,
      "grad_norm": 1.168303370475769,
      "learning_rate": 9.571131578947369e-05,
      "loss": 0.3102,
      "step": 36312
    },
    {
      "epoch": 2.0334303953410235,
      "grad_norm": 1.1110092401504517,
      "learning_rate": 9.571105263157895e-05,
      "loss": 0.3578,
      "step": 36313
    },
    {
      "epoch": 2.0334863926531526,
      "grad_norm": 1.3893972635269165,
      "learning_rate": 9.571078947368421e-05,
      "loss": 0.3737,
      "step": 36314
    },
    {
      "epoch": 2.0335423899652816,
      "grad_norm": 1.1075295209884644,
      "learning_rate": 9.571052631578948e-05,
      "loss": 0.3095,
      "step": 36315
    },
    {
      "epoch": 2.0335983872774106,
      "grad_norm": 1.4017988443374634,
      "learning_rate": 9.571026315789474e-05,
      "loss": 0.4303,
      "step": 36316
    },
    {
      "epoch": 2.0336543845895396,
      "grad_norm": 1.1054296493530273,
      "learning_rate": 9.571e-05,
      "loss": 0.4027,
      "step": 36317
    },
    {
      "epoch": 2.0337103819016686,
      "grad_norm": 1.3854578733444214,
      "learning_rate": 9.570973684210526e-05,
      "loss": 0.4309,
      "step": 36318
    },
    {
      "epoch": 2.0337663792137977,
      "grad_norm": 1.179198980331421,
      "learning_rate": 9.570947368421052e-05,
      "loss": 0.3861,
      "step": 36319
    },
    {
      "epoch": 2.0338223765259267,
      "grad_norm": 1.0702497959136963,
      "learning_rate": 9.57092105263158e-05,
      "loss": 0.4061,
      "step": 36320
    },
    {
      "epoch": 2.0338783738380557,
      "grad_norm": 1.2873876094818115,
      "learning_rate": 9.570894736842106e-05,
      "loss": 0.3976,
      "step": 36321
    },
    {
      "epoch": 2.0339343711501847,
      "grad_norm": 1.2305032014846802,
      "learning_rate": 9.570868421052633e-05,
      "loss": 0.427,
      "step": 36322
    },
    {
      "epoch": 2.0339903684623137,
      "grad_norm": 1.0279797315597534,
      "learning_rate": 9.570842105263158e-05,
      "loss": 0.2978,
      "step": 36323
    },
    {
      "epoch": 2.0340463657744428,
      "grad_norm": 1.109320878982544,
      "learning_rate": 9.570815789473685e-05,
      "loss": 0.417,
      "step": 36324
    },
    {
      "epoch": 2.0341023630865718,
      "grad_norm": 1.2711796760559082,
      "learning_rate": 9.570789473684211e-05,
      "loss": 0.5205,
      "step": 36325
    },
    {
      "epoch": 2.034158360398701,
      "grad_norm": 1.1478742361068726,
      "learning_rate": 9.570763157894738e-05,
      "loss": 0.3033,
      "step": 36326
    },
    {
      "epoch": 2.03421435771083,
      "grad_norm": 1.1631360054016113,
      "learning_rate": 9.570736842105263e-05,
      "loss": 0.3176,
      "step": 36327
    },
    {
      "epoch": 2.034270355022959,
      "grad_norm": 1.0805273056030273,
      "learning_rate": 9.57071052631579e-05,
      "loss": 0.4554,
      "step": 36328
    },
    {
      "epoch": 2.034326352335088,
      "grad_norm": 1.1540974378585815,
      "learning_rate": 9.570684210526316e-05,
      "loss": 0.3262,
      "step": 36329
    },
    {
      "epoch": 2.034382349647217,
      "grad_norm": 1.108595609664917,
      "learning_rate": 9.570657894736843e-05,
      "loss": 0.3023,
      "step": 36330
    },
    {
      "epoch": 2.034438346959346,
      "grad_norm": 1.475795865058899,
      "learning_rate": 9.57063157894737e-05,
      "loss": 0.383,
      "step": 36331
    },
    {
      "epoch": 2.034494344271475,
      "grad_norm": 1.4235014915466309,
      "learning_rate": 9.570605263157895e-05,
      "loss": 0.4574,
      "step": 36332
    },
    {
      "epoch": 2.034550341583604,
      "grad_norm": 1.1427555084228516,
      "learning_rate": 9.570578947368421e-05,
      "loss": 0.3371,
      "step": 36333
    },
    {
      "epoch": 2.034606338895733,
      "grad_norm": 1.497515082359314,
      "learning_rate": 9.570552631578947e-05,
      "loss": 0.5595,
      "step": 36334
    },
    {
      "epoch": 2.034662336207862,
      "grad_norm": 1.7947185039520264,
      "learning_rate": 9.570526315789475e-05,
      "loss": 0.4902,
      "step": 36335
    },
    {
      "epoch": 2.034718333519991,
      "grad_norm": 1.197310447692871,
      "learning_rate": 9.5705e-05,
      "loss": 0.3562,
      "step": 36336
    },
    {
      "epoch": 2.03477433083212,
      "grad_norm": 1.3201531171798706,
      "learning_rate": 9.570473684210527e-05,
      "loss": 0.358,
      "step": 36337
    },
    {
      "epoch": 2.034830328144249,
      "grad_norm": 1.0975604057312012,
      "learning_rate": 9.570447368421053e-05,
      "loss": 0.4022,
      "step": 36338
    },
    {
      "epoch": 2.034886325456378,
      "grad_norm": 1.3169515132904053,
      "learning_rate": 9.57042105263158e-05,
      "loss": 0.4562,
      "step": 36339
    },
    {
      "epoch": 2.034942322768507,
      "grad_norm": 1.1284713745117188,
      "learning_rate": 9.570394736842106e-05,
      "loss": 0.3666,
      "step": 36340
    },
    {
      "epoch": 2.034998320080636,
      "grad_norm": 1.9148318767547607,
      "learning_rate": 9.570368421052632e-05,
      "loss": 0.5468,
      "step": 36341
    },
    {
      "epoch": 2.035054317392765,
      "grad_norm": 1.2590042352676392,
      "learning_rate": 9.570342105263158e-05,
      "loss": 0.3115,
      "step": 36342
    },
    {
      "epoch": 2.035110314704894,
      "grad_norm": 1.3185571432113647,
      "learning_rate": 9.570315789473685e-05,
      "loss": 0.45,
      "step": 36343
    },
    {
      "epoch": 2.035166312017023,
      "grad_norm": 1.2616349458694458,
      "learning_rate": 9.570289473684211e-05,
      "loss": 0.4616,
      "step": 36344
    },
    {
      "epoch": 2.035222309329152,
      "grad_norm": 1.0101739168167114,
      "learning_rate": 9.570263157894737e-05,
      "loss": 0.3045,
      "step": 36345
    },
    {
      "epoch": 2.035278306641281,
      "grad_norm": 1.2091948986053467,
      "learning_rate": 9.570236842105263e-05,
      "loss": 0.4284,
      "step": 36346
    },
    {
      "epoch": 2.03533430395341,
      "grad_norm": 1.174495816230774,
      "learning_rate": 9.57021052631579e-05,
      "loss": 0.4017,
      "step": 36347
    },
    {
      "epoch": 2.035390301265539,
      "grad_norm": 1.139573335647583,
      "learning_rate": 9.570184210526316e-05,
      "loss": 0.5467,
      "step": 36348
    },
    {
      "epoch": 2.0354462985776682,
      "grad_norm": 1.0842411518096924,
      "learning_rate": 9.570157894736844e-05,
      "loss": 0.4468,
      "step": 36349
    },
    {
      "epoch": 2.0355022958897973,
      "grad_norm": 0.9883029460906982,
      "learning_rate": 9.570131578947368e-05,
      "loss": 0.2937,
      "step": 36350
    },
    {
      "epoch": 2.0355582932019263,
      "grad_norm": 1.8533591032028198,
      "learning_rate": 9.570105263157894e-05,
      "loss": 0.4119,
      "step": 36351
    },
    {
      "epoch": 2.0356142905140553,
      "grad_norm": 1.1652733087539673,
      "learning_rate": 9.570078947368422e-05,
      "loss": 0.3923,
      "step": 36352
    },
    {
      "epoch": 2.0356702878261843,
      "grad_norm": 1.3475103378295898,
      "learning_rate": 9.570052631578948e-05,
      "loss": 0.4403,
      "step": 36353
    },
    {
      "epoch": 2.0357262851383133,
      "grad_norm": 1.1191695928573608,
      "learning_rate": 9.570026315789475e-05,
      "loss": 0.4404,
      "step": 36354
    },
    {
      "epoch": 2.0357822824504423,
      "grad_norm": 1.1079308986663818,
      "learning_rate": 9.57e-05,
      "loss": 0.2964,
      "step": 36355
    },
    {
      "epoch": 2.0358382797625714,
      "grad_norm": 1.2974822521209717,
      "learning_rate": 9.569973684210527e-05,
      "loss": 0.4669,
      "step": 36356
    },
    {
      "epoch": 2.0358942770747004,
      "grad_norm": 1.1397373676300049,
      "learning_rate": 9.569947368421053e-05,
      "loss": 0.4338,
      "step": 36357
    },
    {
      "epoch": 2.0359502743868294,
      "grad_norm": 1.115784764289856,
      "learning_rate": 9.56992105263158e-05,
      "loss": 0.37,
      "step": 36358
    },
    {
      "epoch": 2.0360062716989584,
      "grad_norm": 1.3924771547317505,
      "learning_rate": 9.569894736842106e-05,
      "loss": 0.5537,
      "step": 36359
    },
    {
      "epoch": 2.0360622690110874,
      "grad_norm": 1.061295986175537,
      "learning_rate": 9.569868421052632e-05,
      "loss": 0.3646,
      "step": 36360
    },
    {
      "epoch": 2.0361182663232165,
      "grad_norm": 1.1816364526748657,
      "learning_rate": 9.569842105263158e-05,
      "loss": 0.4101,
      "step": 36361
    },
    {
      "epoch": 2.0361742636353455,
      "grad_norm": 1.3380966186523438,
      "learning_rate": 9.569815789473685e-05,
      "loss": 0.4725,
      "step": 36362
    },
    {
      "epoch": 2.0362302609474745,
      "grad_norm": 1.329868197441101,
      "learning_rate": 9.569789473684211e-05,
      "loss": 0.4685,
      "step": 36363
    },
    {
      "epoch": 2.0362862582596035,
      "grad_norm": 2.372368097305298,
      "learning_rate": 9.569763157894737e-05,
      "loss": 0.355,
      "step": 36364
    },
    {
      "epoch": 2.0363422555717325,
      "grad_norm": 1.1149588823318481,
      "learning_rate": 9.569736842105263e-05,
      "loss": 0.397,
      "step": 36365
    },
    {
      "epoch": 2.0363982528838616,
      "grad_norm": 1.4281861782073975,
      "learning_rate": 9.56971052631579e-05,
      "loss": 0.4194,
      "step": 36366
    },
    {
      "epoch": 2.0364542501959906,
      "grad_norm": 1.0964142084121704,
      "learning_rate": 9.569684210526317e-05,
      "loss": 0.464,
      "step": 36367
    },
    {
      "epoch": 2.0365102475081196,
      "grad_norm": 1.3055789470672607,
      "learning_rate": 9.569657894736843e-05,
      "loss": 0.4803,
      "step": 36368
    },
    {
      "epoch": 2.0365662448202486,
      "grad_norm": 1.1180106401443481,
      "learning_rate": 9.569631578947369e-05,
      "loss": 0.3562,
      "step": 36369
    },
    {
      "epoch": 2.0366222421323776,
      "grad_norm": 1.1111477613449097,
      "learning_rate": 9.569605263157895e-05,
      "loss": 0.3256,
      "step": 36370
    },
    {
      "epoch": 2.0366782394445067,
      "grad_norm": 1.2732056379318237,
      "learning_rate": 9.569578947368422e-05,
      "loss": 0.5059,
      "step": 36371
    },
    {
      "epoch": 2.0367342367566357,
      "grad_norm": 0.9448058009147644,
      "learning_rate": 9.569552631578948e-05,
      "loss": 0.3496,
      "step": 36372
    },
    {
      "epoch": 2.0367902340687647,
      "grad_norm": 1.2061527967453003,
      "learning_rate": 9.569526315789474e-05,
      "loss": 0.4017,
      "step": 36373
    },
    {
      "epoch": 2.0368462313808937,
      "grad_norm": 1.293936014175415,
      "learning_rate": 9.5695e-05,
      "loss": 0.3569,
      "step": 36374
    },
    {
      "epoch": 2.0369022286930227,
      "grad_norm": 1.329754114151001,
      "learning_rate": 9.569473684210527e-05,
      "loss": 0.4799,
      "step": 36375
    },
    {
      "epoch": 2.0369582260051518,
      "grad_norm": 1.2398200035095215,
      "learning_rate": 9.569447368421053e-05,
      "loss": 0.4471,
      "step": 36376
    },
    {
      "epoch": 2.0370142233172808,
      "grad_norm": 1.3133492469787598,
      "learning_rate": 9.56942105263158e-05,
      "loss": 0.3627,
      "step": 36377
    },
    {
      "epoch": 2.03707022062941,
      "grad_norm": 1.4421110153198242,
      "learning_rate": 9.569394736842105e-05,
      "loss": 0.5083,
      "step": 36378
    },
    {
      "epoch": 2.037126217941539,
      "grad_norm": 15.928050994873047,
      "learning_rate": 9.569368421052632e-05,
      "loss": 0.3977,
      "step": 36379
    },
    {
      "epoch": 2.037182215253668,
      "grad_norm": 1.1906312704086304,
      "learning_rate": 9.569342105263158e-05,
      "loss": 0.3724,
      "step": 36380
    },
    {
      "epoch": 2.037238212565797,
      "grad_norm": 1.338821291923523,
      "learning_rate": 9.569315789473686e-05,
      "loss": 0.3934,
      "step": 36381
    },
    {
      "epoch": 2.037294209877926,
      "grad_norm": 1.6535534858703613,
      "learning_rate": 9.56928947368421e-05,
      "loss": 0.4688,
      "step": 36382
    },
    {
      "epoch": 2.037350207190055,
      "grad_norm": 1.2338567972183228,
      "learning_rate": 9.569263157894738e-05,
      "loss": 0.363,
      "step": 36383
    },
    {
      "epoch": 2.037406204502184,
      "grad_norm": 1.1587836742401123,
      "learning_rate": 9.569236842105264e-05,
      "loss": 0.3548,
      "step": 36384
    },
    {
      "epoch": 2.037462201814313,
      "grad_norm": 1.517890214920044,
      "learning_rate": 9.56921052631579e-05,
      "loss": 0.5015,
      "step": 36385
    },
    {
      "epoch": 2.037518199126442,
      "grad_norm": 1.1395021677017212,
      "learning_rate": 9.569184210526317e-05,
      "loss": 0.4391,
      "step": 36386
    },
    {
      "epoch": 2.037574196438571,
      "grad_norm": 2.258161783218384,
      "learning_rate": 9.569157894736841e-05,
      "loss": 0.4614,
      "step": 36387
    },
    {
      "epoch": 2.0376301937507,
      "grad_norm": 1.120851755142212,
      "learning_rate": 9.569131578947369e-05,
      "loss": 0.3919,
      "step": 36388
    },
    {
      "epoch": 2.037686191062829,
      "grad_norm": 1.4719966650009155,
      "learning_rate": 9.569105263157895e-05,
      "loss": 0.4211,
      "step": 36389
    },
    {
      "epoch": 2.037742188374958,
      "grad_norm": 1.056656837463379,
      "learning_rate": 9.569078947368422e-05,
      "loss": 0.3292,
      "step": 36390
    },
    {
      "epoch": 2.037798185687087,
      "grad_norm": 1.1554722785949707,
      "learning_rate": 9.569052631578948e-05,
      "loss": 0.3393,
      "step": 36391
    },
    {
      "epoch": 2.037854182999216,
      "grad_norm": 1.3846722841262817,
      "learning_rate": 9.569026315789474e-05,
      "loss": 0.4035,
      "step": 36392
    },
    {
      "epoch": 2.037910180311345,
      "grad_norm": 1.2527151107788086,
      "learning_rate": 9.569e-05,
      "loss": 0.3506,
      "step": 36393
    },
    {
      "epoch": 2.037966177623474,
      "grad_norm": 1.1692259311676025,
      "learning_rate": 9.568973684210527e-05,
      "loss": 0.3963,
      "step": 36394
    },
    {
      "epoch": 2.038022174935603,
      "grad_norm": 1.0423853397369385,
      "learning_rate": 9.568947368421053e-05,
      "loss": 0.3444,
      "step": 36395
    },
    {
      "epoch": 2.038078172247732,
      "grad_norm": 1.27071213722229,
      "learning_rate": 9.568921052631579e-05,
      "loss": 0.3218,
      "step": 36396
    },
    {
      "epoch": 2.038134169559861,
      "grad_norm": 1.1114684343338013,
      "learning_rate": 9.568894736842105e-05,
      "loss": 0.3699,
      "step": 36397
    },
    {
      "epoch": 2.03819016687199,
      "grad_norm": 1.2327271699905396,
      "learning_rate": 9.568868421052633e-05,
      "loss": 0.3797,
      "step": 36398
    },
    {
      "epoch": 2.038246164184119,
      "grad_norm": 1.254450798034668,
      "learning_rate": 9.568842105263159e-05,
      "loss": 0.5084,
      "step": 36399
    },
    {
      "epoch": 2.038302161496248,
      "grad_norm": 1.3284070491790771,
      "learning_rate": 9.568815789473685e-05,
      "loss": 0.3726,
      "step": 36400
    },
    {
      "epoch": 2.0383581588083772,
      "grad_norm": 1.1129344701766968,
      "learning_rate": 9.56878947368421e-05,
      "loss": 0.4085,
      "step": 36401
    },
    {
      "epoch": 2.0384141561205062,
      "grad_norm": 1.1655981540679932,
      "learning_rate": 9.568763157894736e-05,
      "loss": 0.3665,
      "step": 36402
    },
    {
      "epoch": 2.0384701534326353,
      "grad_norm": 1.2138686180114746,
      "learning_rate": 9.568736842105264e-05,
      "loss": 0.421,
      "step": 36403
    },
    {
      "epoch": 2.0385261507447643,
      "grad_norm": 1.2618060111999512,
      "learning_rate": 9.56871052631579e-05,
      "loss": 0.5025,
      "step": 36404
    },
    {
      "epoch": 2.0385821480568933,
      "grad_norm": 1.064089298248291,
      "learning_rate": 9.568684210526316e-05,
      "loss": 0.4081,
      "step": 36405
    },
    {
      "epoch": 2.0386381453690223,
      "grad_norm": 1.2715030908584595,
      "learning_rate": 9.568657894736842e-05,
      "loss": 0.4756,
      "step": 36406
    },
    {
      "epoch": 2.0386941426811513,
      "grad_norm": 1.3624147176742554,
      "learning_rate": 9.568631578947369e-05,
      "loss": 0.4342,
      "step": 36407
    },
    {
      "epoch": 2.0387501399932804,
      "grad_norm": 1.3513011932373047,
      "learning_rate": 9.568605263157895e-05,
      "loss": 0.3672,
      "step": 36408
    },
    {
      "epoch": 2.0388061373054094,
      "grad_norm": 1.0731122493743896,
      "learning_rate": 9.568578947368422e-05,
      "loss": 0.3693,
      "step": 36409
    },
    {
      "epoch": 2.0388621346175384,
      "grad_norm": 1.1919761896133423,
      "learning_rate": 9.568552631578947e-05,
      "loss": 0.3531,
      "step": 36410
    },
    {
      "epoch": 2.0389181319296674,
      "grad_norm": 1.0832617282867432,
      "learning_rate": 9.568526315789474e-05,
      "loss": 0.3769,
      "step": 36411
    },
    {
      "epoch": 2.0389741292417964,
      "grad_norm": 1.413251280784607,
      "learning_rate": 9.5685e-05,
      "loss": 0.5257,
      "step": 36412
    },
    {
      "epoch": 2.0390301265539255,
      "grad_norm": 1.0677169561386108,
      "learning_rate": 9.568473684210528e-05,
      "loss": 0.4086,
      "step": 36413
    },
    {
      "epoch": 2.0390861238660545,
      "grad_norm": 1.272536039352417,
      "learning_rate": 9.568447368421054e-05,
      "loss": 0.3315,
      "step": 36414
    },
    {
      "epoch": 2.0391421211781835,
      "grad_norm": 1.1808593273162842,
      "learning_rate": 9.56842105263158e-05,
      "loss": 0.3294,
      "step": 36415
    },
    {
      "epoch": 2.0391981184903125,
      "grad_norm": 1.25516939163208,
      "learning_rate": 9.568394736842106e-05,
      "loss": 0.3966,
      "step": 36416
    },
    {
      "epoch": 2.0392541158024415,
      "grad_norm": 1.088417410850525,
      "learning_rate": 9.568368421052633e-05,
      "loss": 0.2943,
      "step": 36417
    },
    {
      "epoch": 2.0393101131145706,
      "grad_norm": 1.3179112672805786,
      "learning_rate": 9.568342105263159e-05,
      "loss": 0.4835,
      "step": 36418
    },
    {
      "epoch": 2.0393661104266996,
      "grad_norm": 1.8994653224945068,
      "learning_rate": 9.568315789473683e-05,
      "loss": 0.546,
      "step": 36419
    },
    {
      "epoch": 2.0394221077388286,
      "grad_norm": 1.1465600728988647,
      "learning_rate": 9.568289473684211e-05,
      "loss": 0.3404,
      "step": 36420
    },
    {
      "epoch": 2.0394781050509576,
      "grad_norm": 1.4661141633987427,
      "learning_rate": 9.568263157894737e-05,
      "loss": 0.4156,
      "step": 36421
    },
    {
      "epoch": 2.0395341023630866,
      "grad_norm": 1.3896393775939941,
      "learning_rate": 9.568236842105264e-05,
      "loss": 0.4228,
      "step": 36422
    },
    {
      "epoch": 2.0395900996752157,
      "grad_norm": 1.2230881452560425,
      "learning_rate": 9.56821052631579e-05,
      "loss": 0.3906,
      "step": 36423
    },
    {
      "epoch": 2.0396460969873447,
      "grad_norm": 1.1439485549926758,
      "learning_rate": 9.568184210526316e-05,
      "loss": 0.3726,
      "step": 36424
    },
    {
      "epoch": 2.0397020942994737,
      "grad_norm": 1.0547151565551758,
      "learning_rate": 9.568157894736842e-05,
      "loss": 0.3595,
      "step": 36425
    },
    {
      "epoch": 2.0397580916116027,
      "grad_norm": 1.3836079835891724,
      "learning_rate": 9.56813157894737e-05,
      "loss": 0.5307,
      "step": 36426
    },
    {
      "epoch": 2.0398140889237317,
      "grad_norm": 1.029758095741272,
      "learning_rate": 9.568105263157895e-05,
      "loss": 0.377,
      "step": 36427
    },
    {
      "epoch": 2.0398700862358607,
      "grad_norm": 0.9970435500144958,
      "learning_rate": 9.568078947368421e-05,
      "loss": 0.3412,
      "step": 36428
    },
    {
      "epoch": 2.0399260835479898,
      "grad_norm": 1.1629329919815063,
      "learning_rate": 9.568052631578947e-05,
      "loss": 0.3815,
      "step": 36429
    },
    {
      "epoch": 2.039982080860119,
      "grad_norm": 1.4850358963012695,
      "learning_rate": 9.568026315789475e-05,
      "loss": 0.4581,
      "step": 36430
    },
    {
      "epoch": 2.040038078172248,
      "grad_norm": 1.164542317390442,
      "learning_rate": 9.568e-05,
      "loss": 0.3491,
      "step": 36431
    },
    {
      "epoch": 2.040094075484377,
      "grad_norm": 1.3895856142044067,
      "learning_rate": 9.567973684210527e-05,
      "loss": 0.3756,
      "step": 36432
    },
    {
      "epoch": 2.040150072796506,
      "grad_norm": 1.2311501502990723,
      "learning_rate": 9.567947368421052e-05,
      "loss": 0.5004,
      "step": 36433
    },
    {
      "epoch": 2.040206070108635,
      "grad_norm": 1.1164169311523438,
      "learning_rate": 9.56792105263158e-05,
      "loss": 0.311,
      "step": 36434
    },
    {
      "epoch": 2.040262067420764,
      "grad_norm": 1.1234296560287476,
      "learning_rate": 9.567894736842106e-05,
      "loss": 0.3489,
      "step": 36435
    },
    {
      "epoch": 2.040318064732893,
      "grad_norm": 1.1429530382156372,
      "learning_rate": 9.567868421052633e-05,
      "loss": 0.4242,
      "step": 36436
    },
    {
      "epoch": 2.040374062045022,
      "grad_norm": 1.1247172355651855,
      "learning_rate": 9.567842105263158e-05,
      "loss": 0.4346,
      "step": 36437
    },
    {
      "epoch": 2.040430059357151,
      "grad_norm": 1.36590576171875,
      "learning_rate": 9.567815789473684e-05,
      "loss": 0.4237,
      "step": 36438
    },
    {
      "epoch": 2.04048605666928,
      "grad_norm": 1.1380928754806519,
      "learning_rate": 9.567789473684211e-05,
      "loss": 0.4228,
      "step": 36439
    },
    {
      "epoch": 2.040542053981409,
      "grad_norm": 1.4909753799438477,
      "learning_rate": 9.567763157894737e-05,
      "loss": 0.4079,
      "step": 36440
    },
    {
      "epoch": 2.040598051293538,
      "grad_norm": 1.1781961917877197,
      "learning_rate": 9.567736842105264e-05,
      "loss": 0.3907,
      "step": 36441
    },
    {
      "epoch": 2.040654048605667,
      "grad_norm": 1.4378325939178467,
      "learning_rate": 9.567710526315789e-05,
      "loss": 0.4051,
      "step": 36442
    },
    {
      "epoch": 2.040710045917796,
      "grad_norm": 1.249224305152893,
      "learning_rate": 9.567684210526316e-05,
      "loss": 0.4587,
      "step": 36443
    },
    {
      "epoch": 2.040766043229925,
      "grad_norm": 1.0237922668457031,
      "learning_rate": 9.567657894736842e-05,
      "loss": 0.3823,
      "step": 36444
    },
    {
      "epoch": 2.040822040542054,
      "grad_norm": 0.989255428314209,
      "learning_rate": 9.56763157894737e-05,
      "loss": 0.3499,
      "step": 36445
    },
    {
      "epoch": 2.040878037854183,
      "grad_norm": 1.6551977396011353,
      "learning_rate": 9.567605263157896e-05,
      "loss": 0.3919,
      "step": 36446
    },
    {
      "epoch": 2.040934035166312,
      "grad_norm": 1.178074836730957,
      "learning_rate": 9.567578947368422e-05,
      "loss": 0.3257,
      "step": 36447
    },
    {
      "epoch": 2.040990032478441,
      "grad_norm": 1.409286618232727,
      "learning_rate": 9.567552631578948e-05,
      "loss": 0.5432,
      "step": 36448
    },
    {
      "epoch": 2.04104602979057,
      "grad_norm": 1.3017104864120483,
      "learning_rate": 9.567526315789475e-05,
      "loss": 0.3588,
      "step": 36449
    },
    {
      "epoch": 2.041102027102699,
      "grad_norm": 1.182559847831726,
      "learning_rate": 9.567500000000001e-05,
      "loss": 0.3466,
      "step": 36450
    },
    {
      "epoch": 2.041158024414828,
      "grad_norm": 1.0540881156921387,
      "learning_rate": 9.567473684210527e-05,
      "loss": 0.377,
      "step": 36451
    },
    {
      "epoch": 2.041214021726957,
      "grad_norm": 1.5845394134521484,
      "learning_rate": 9.567447368421053e-05,
      "loss": 0.3711,
      "step": 36452
    },
    {
      "epoch": 2.0412700190390862,
      "grad_norm": 1.032497763633728,
      "learning_rate": 9.56742105263158e-05,
      "loss": 0.3749,
      "step": 36453
    },
    {
      "epoch": 2.0413260163512152,
      "grad_norm": 1.2808810472488403,
      "learning_rate": 9.567394736842106e-05,
      "loss": 0.388,
      "step": 36454
    },
    {
      "epoch": 2.0413820136633443,
      "grad_norm": 1.5016989707946777,
      "learning_rate": 9.567368421052632e-05,
      "loss": 0.3864,
      "step": 36455
    },
    {
      "epoch": 2.0414380109754733,
      "grad_norm": 1.0901206731796265,
      "learning_rate": 9.567342105263158e-05,
      "loss": 0.3969,
      "step": 36456
    },
    {
      "epoch": 2.0414940082876023,
      "grad_norm": 1.341513752937317,
      "learning_rate": 9.567315789473684e-05,
      "loss": 0.4181,
      "step": 36457
    },
    {
      "epoch": 2.0415500055997313,
      "grad_norm": 1.3783624172210693,
      "learning_rate": 9.567289473684211e-05,
      "loss": 0.5157,
      "step": 36458
    },
    {
      "epoch": 2.0416060029118603,
      "grad_norm": 1.3884556293487549,
      "learning_rate": 9.567263157894737e-05,
      "loss": 0.4305,
      "step": 36459
    },
    {
      "epoch": 2.0416620002239894,
      "grad_norm": 1.177894949913025,
      "learning_rate": 9.567236842105263e-05,
      "loss": 0.5656,
      "step": 36460
    },
    {
      "epoch": 2.0417179975361184,
      "grad_norm": 1.1024378538131714,
      "learning_rate": 9.567210526315789e-05,
      "loss": 0.4329,
      "step": 36461
    },
    {
      "epoch": 2.0417739948482474,
      "grad_norm": 1.2017886638641357,
      "learning_rate": 9.567184210526317e-05,
      "loss": 0.3496,
      "step": 36462
    },
    {
      "epoch": 2.0418299921603764,
      "grad_norm": 1.2188318967819214,
      "learning_rate": 9.567157894736843e-05,
      "loss": 0.376,
      "step": 36463
    },
    {
      "epoch": 2.0418859894725054,
      "grad_norm": 1.315497875213623,
      "learning_rate": 9.56713157894737e-05,
      "loss": 0.4797,
      "step": 36464
    },
    {
      "epoch": 2.0419419867846345,
      "grad_norm": 1.1133599281311035,
      "learning_rate": 9.567105263157894e-05,
      "loss": 0.374,
      "step": 36465
    },
    {
      "epoch": 2.0419979840967635,
      "grad_norm": 1.440004587173462,
      "learning_rate": 9.567078947368422e-05,
      "loss": 0.4988,
      "step": 36466
    },
    {
      "epoch": 2.0420539814088925,
      "grad_norm": 1.364035725593567,
      "learning_rate": 9.567052631578948e-05,
      "loss": 0.5888,
      "step": 36467
    },
    {
      "epoch": 2.0421099787210215,
      "grad_norm": 1.1423653364181519,
      "learning_rate": 9.567026315789475e-05,
      "loss": 0.3866,
      "step": 36468
    },
    {
      "epoch": 2.0421659760331505,
      "grad_norm": 1.2399284839630127,
      "learning_rate": 9.567000000000001e-05,
      "loss": 0.4427,
      "step": 36469
    },
    {
      "epoch": 2.0422219733452796,
      "grad_norm": 2.5457022190093994,
      "learning_rate": 9.566973684210527e-05,
      "loss": 0.3276,
      "step": 36470
    },
    {
      "epoch": 2.0422779706574086,
      "grad_norm": 1.1251628398895264,
      "learning_rate": 9.566947368421053e-05,
      "loss": 0.3402,
      "step": 36471
    },
    {
      "epoch": 2.0423339679695376,
      "grad_norm": 1.359660029411316,
      "learning_rate": 9.566921052631579e-05,
      "loss": 0.4478,
      "step": 36472
    },
    {
      "epoch": 2.0423899652816666,
      "grad_norm": 1.1996190547943115,
      "learning_rate": 9.566894736842106e-05,
      "loss": 0.3956,
      "step": 36473
    },
    {
      "epoch": 2.0424459625937956,
      "grad_norm": 1.0779095888137817,
      "learning_rate": 9.566868421052631e-05,
      "loss": 0.4106,
      "step": 36474
    },
    {
      "epoch": 2.0425019599059246,
      "grad_norm": 1.1836365461349487,
      "learning_rate": 9.566842105263158e-05,
      "loss": 0.4242,
      "step": 36475
    },
    {
      "epoch": 2.0425579572180537,
      "grad_norm": 1.069923758506775,
      "learning_rate": 9.566815789473684e-05,
      "loss": 0.3652,
      "step": 36476
    },
    {
      "epoch": 2.0426139545301827,
      "grad_norm": 1.194624423980713,
      "learning_rate": 9.566789473684212e-05,
      "loss": 0.4332,
      "step": 36477
    },
    {
      "epoch": 2.0426699518423117,
      "grad_norm": 1.1308766603469849,
      "learning_rate": 9.566763157894738e-05,
      "loss": 0.3135,
      "step": 36478
    },
    {
      "epoch": 2.0427259491544407,
      "grad_norm": 1.3823795318603516,
      "learning_rate": 9.566736842105264e-05,
      "loss": 0.2934,
      "step": 36479
    },
    {
      "epoch": 2.0427819464665697,
      "grad_norm": 1.156907081604004,
      "learning_rate": 9.56671052631579e-05,
      "loss": 0.3417,
      "step": 36480
    },
    {
      "epoch": 2.0428379437786988,
      "grad_norm": 2.377918243408203,
      "learning_rate": 9.566684210526317e-05,
      "loss": 0.4117,
      "step": 36481
    },
    {
      "epoch": 2.042893941090828,
      "grad_norm": 1.466359257698059,
      "learning_rate": 9.566657894736843e-05,
      "loss": 0.402,
      "step": 36482
    },
    {
      "epoch": 2.042949938402957,
      "grad_norm": 1.3489794731140137,
      "learning_rate": 9.566631578947369e-05,
      "loss": 0.4012,
      "step": 36483
    },
    {
      "epoch": 2.043005935715086,
      "grad_norm": 1.6134717464447021,
      "learning_rate": 9.566605263157895e-05,
      "loss": 0.4087,
      "step": 36484
    },
    {
      "epoch": 2.043061933027215,
      "grad_norm": 1.2730505466461182,
      "learning_rate": 9.566578947368422e-05,
      "loss": 0.3468,
      "step": 36485
    },
    {
      "epoch": 2.043117930339344,
      "grad_norm": 1.3391703367233276,
      "learning_rate": 9.566552631578948e-05,
      "loss": 0.5541,
      "step": 36486
    },
    {
      "epoch": 2.043173927651473,
      "grad_norm": 1.3690104484558105,
      "learning_rate": 9.566526315789474e-05,
      "loss": 0.455,
      "step": 36487
    },
    {
      "epoch": 2.043229924963602,
      "grad_norm": 1.2672092914581299,
      "learning_rate": 9.5665e-05,
      "loss": 0.3785,
      "step": 36488
    },
    {
      "epoch": 2.043285922275731,
      "grad_norm": 1.551269292831421,
      "learning_rate": 9.566473684210526e-05,
      "loss": 0.4424,
      "step": 36489
    },
    {
      "epoch": 2.04334191958786,
      "grad_norm": 1.2057708501815796,
      "learning_rate": 9.566447368421053e-05,
      "loss": 0.3383,
      "step": 36490
    },
    {
      "epoch": 2.043397916899989,
      "grad_norm": 1.1033844947814941,
      "learning_rate": 9.566421052631579e-05,
      "loss": 0.3056,
      "step": 36491
    },
    {
      "epoch": 2.043453914212118,
      "grad_norm": 1.648995280265808,
      "learning_rate": 9.566394736842105e-05,
      "loss": 0.3907,
      "step": 36492
    },
    {
      "epoch": 2.043509911524247,
      "grad_norm": 1.2089917659759521,
      "learning_rate": 9.566368421052631e-05,
      "loss": 0.346,
      "step": 36493
    },
    {
      "epoch": 2.043565908836376,
      "grad_norm": 1.0456515550613403,
      "learning_rate": 9.566342105263159e-05,
      "loss": 0.426,
      "step": 36494
    },
    {
      "epoch": 2.043621906148505,
      "grad_norm": 1.1843595504760742,
      "learning_rate": 9.566315789473684e-05,
      "loss": 0.3329,
      "step": 36495
    },
    {
      "epoch": 2.043677903460634,
      "grad_norm": 1.4157453775405884,
      "learning_rate": 9.566289473684212e-05,
      "loss": 0.3362,
      "step": 36496
    },
    {
      "epoch": 2.043733900772763,
      "grad_norm": 1.4117177724838257,
      "learning_rate": 9.566263157894736e-05,
      "loss": 0.3274,
      "step": 36497
    },
    {
      "epoch": 2.043789898084892,
      "grad_norm": 1.9223222732543945,
      "learning_rate": 9.566236842105264e-05,
      "loss": 0.4338,
      "step": 36498
    },
    {
      "epoch": 2.043845895397021,
      "grad_norm": 1.2246278524398804,
      "learning_rate": 9.56621052631579e-05,
      "loss": 0.5161,
      "step": 36499
    },
    {
      "epoch": 2.04390189270915,
      "grad_norm": 1.1343804597854614,
      "learning_rate": 9.566184210526317e-05,
      "loss": 0.3901,
      "step": 36500
    },
    {
      "epoch": 2.043957890021279,
      "grad_norm": 1.1578418016433716,
      "learning_rate": 9.566157894736843e-05,
      "loss": 0.3797,
      "step": 36501
    },
    {
      "epoch": 2.044013887333408,
      "grad_norm": 1.0334910154342651,
      "learning_rate": 9.566131578947369e-05,
      "loss": 0.3632,
      "step": 36502
    },
    {
      "epoch": 2.044069884645537,
      "grad_norm": 1.2720609903335571,
      "learning_rate": 9.566105263157895e-05,
      "loss": 0.4558,
      "step": 36503
    },
    {
      "epoch": 2.044125881957666,
      "grad_norm": 1.2591471672058105,
      "learning_rate": 9.566078947368422e-05,
      "loss": 0.5534,
      "step": 36504
    },
    {
      "epoch": 2.044181879269795,
      "grad_norm": 1.1023025512695312,
      "learning_rate": 9.566052631578948e-05,
      "loss": 0.4034,
      "step": 36505
    },
    {
      "epoch": 2.0442378765819242,
      "grad_norm": 1.1416971683502197,
      "learning_rate": 9.566026315789474e-05,
      "loss": 0.4341,
      "step": 36506
    },
    {
      "epoch": 2.0442938738940533,
      "grad_norm": 1.21024489402771,
      "learning_rate": 9.566e-05,
      "loss": 0.4344,
      "step": 36507
    },
    {
      "epoch": 2.0443498712061823,
      "grad_norm": 1.142454743385315,
      "learning_rate": 9.565973684210526e-05,
      "loss": 0.4636,
      "step": 36508
    },
    {
      "epoch": 2.0444058685183113,
      "grad_norm": 1.2204978466033936,
      "learning_rate": 9.565947368421054e-05,
      "loss": 0.6929,
      "step": 36509
    },
    {
      "epoch": 2.0444618658304403,
      "grad_norm": 1.3453527688980103,
      "learning_rate": 9.56592105263158e-05,
      "loss": 0.4727,
      "step": 36510
    },
    {
      "epoch": 2.0445178631425693,
      "grad_norm": 1.1923243999481201,
      "learning_rate": 9.565894736842105e-05,
      "loss": 0.4869,
      "step": 36511
    },
    {
      "epoch": 2.0445738604546984,
      "grad_norm": 1.0655434131622314,
      "learning_rate": 9.565868421052631e-05,
      "loss": 0.3825,
      "step": 36512
    },
    {
      "epoch": 2.0446298577668274,
      "grad_norm": 0.9945422410964966,
      "learning_rate": 9.565842105263159e-05,
      "loss": 0.325,
      "step": 36513
    },
    {
      "epoch": 2.0446858550789564,
      "grad_norm": 1.1645029783248901,
      "learning_rate": 9.565815789473685e-05,
      "loss": 0.3556,
      "step": 36514
    },
    {
      "epoch": 2.0447418523910854,
      "grad_norm": 1.2809237241744995,
      "learning_rate": 9.565789473684211e-05,
      "loss": 0.4214,
      "step": 36515
    },
    {
      "epoch": 2.0447978497032144,
      "grad_norm": 1.6623132228851318,
      "learning_rate": 9.565763157894737e-05,
      "loss": 0.3907,
      "step": 36516
    },
    {
      "epoch": 2.0448538470153435,
      "grad_norm": 1.1635187864303589,
      "learning_rate": 9.565736842105264e-05,
      "loss": 0.3445,
      "step": 36517
    },
    {
      "epoch": 2.0449098443274725,
      "grad_norm": 1.2451426982879639,
      "learning_rate": 9.56571052631579e-05,
      "loss": 0.3665,
      "step": 36518
    },
    {
      "epoch": 2.0449658416396015,
      "grad_norm": 1.2910524606704712,
      "learning_rate": 9.565684210526317e-05,
      "loss": 0.4116,
      "step": 36519
    },
    {
      "epoch": 2.0450218389517305,
      "grad_norm": 1.2206307649612427,
      "learning_rate": 9.565657894736842e-05,
      "loss": 0.4924,
      "step": 36520
    },
    {
      "epoch": 2.0450778362638595,
      "grad_norm": 1.2187786102294922,
      "learning_rate": 9.565631578947369e-05,
      "loss": 0.3896,
      "step": 36521
    },
    {
      "epoch": 2.0451338335759885,
      "grad_norm": 1.055055856704712,
      "learning_rate": 9.565605263157895e-05,
      "loss": 0.3306,
      "step": 36522
    },
    {
      "epoch": 2.0451898308881176,
      "grad_norm": 1.1805987358093262,
      "learning_rate": 9.565578947368421e-05,
      "loss": 0.3498,
      "step": 36523
    },
    {
      "epoch": 2.0452458282002466,
      "grad_norm": 1.7351652383804321,
      "learning_rate": 9.565552631578949e-05,
      "loss": 0.3427,
      "step": 36524
    },
    {
      "epoch": 2.0453018255123756,
      "grad_norm": 1.3048440217971802,
      "learning_rate": 9.565526315789473e-05,
      "loss": 0.3714,
      "step": 36525
    },
    {
      "epoch": 2.0453578228245046,
      "grad_norm": 1.1941004991531372,
      "learning_rate": 9.5655e-05,
      "loss": 0.3627,
      "step": 36526
    },
    {
      "epoch": 2.0454138201366336,
      "grad_norm": 1.2364619970321655,
      "learning_rate": 9.565473684210526e-05,
      "loss": 0.4008,
      "step": 36527
    },
    {
      "epoch": 2.0454698174487627,
      "grad_norm": 1.2263951301574707,
      "learning_rate": 9.565447368421054e-05,
      "loss": 0.3941,
      "step": 36528
    },
    {
      "epoch": 2.0455258147608917,
      "grad_norm": 1.3391637802124023,
      "learning_rate": 9.565421052631578e-05,
      "loss": 0.3986,
      "step": 36529
    },
    {
      "epoch": 2.0455818120730207,
      "grad_norm": 1.071968674659729,
      "learning_rate": 9.565394736842106e-05,
      "loss": 0.3317,
      "step": 36530
    },
    {
      "epoch": 2.0456378093851497,
      "grad_norm": 1.5218069553375244,
      "learning_rate": 9.565368421052632e-05,
      "loss": 0.485,
      "step": 36531
    },
    {
      "epoch": 2.0456938066972787,
      "grad_norm": 1.1150939464569092,
      "learning_rate": 9.565342105263159e-05,
      "loss": 0.3682,
      "step": 36532
    },
    {
      "epoch": 2.0457498040094078,
      "grad_norm": 1.6196460723876953,
      "learning_rate": 9.565315789473685e-05,
      "loss": 0.4308,
      "step": 36533
    },
    {
      "epoch": 2.0458058013215368,
      "grad_norm": 1.3216304779052734,
      "learning_rate": 9.565289473684211e-05,
      "loss": 0.3898,
      "step": 36534
    },
    {
      "epoch": 2.045861798633666,
      "grad_norm": 1.3647665977478027,
      "learning_rate": 9.565263157894737e-05,
      "loss": 0.368,
      "step": 36535
    },
    {
      "epoch": 2.045917795945795,
      "grad_norm": 1.0381462574005127,
      "learning_rate": 9.565236842105264e-05,
      "loss": 0.258,
      "step": 36536
    },
    {
      "epoch": 2.045973793257924,
      "grad_norm": 1.904181957244873,
      "learning_rate": 9.56521052631579e-05,
      "loss": 0.4845,
      "step": 36537
    },
    {
      "epoch": 2.046029790570053,
      "grad_norm": 1.441455364227295,
      "learning_rate": 9.565184210526316e-05,
      "loss": 0.6119,
      "step": 36538
    },
    {
      "epoch": 2.046085787882182,
      "grad_norm": 1.2802162170410156,
      "learning_rate": 9.565157894736842e-05,
      "loss": 0.411,
      "step": 36539
    },
    {
      "epoch": 2.046141785194311,
      "grad_norm": 1.4507910013198853,
      "learning_rate": 9.565131578947368e-05,
      "loss": 0.4691,
      "step": 36540
    },
    {
      "epoch": 2.0461977825064395,
      "grad_norm": 1.3461358547210693,
      "learning_rate": 9.565105263157896e-05,
      "loss": 0.304,
      "step": 36541
    },
    {
      "epoch": 2.046253779818569,
      "grad_norm": 1.265123724937439,
      "learning_rate": 9.565078947368421e-05,
      "loss": 0.4243,
      "step": 36542
    },
    {
      "epoch": 2.0463097771306975,
      "grad_norm": 1.1044375896453857,
      "learning_rate": 9.565052631578947e-05,
      "loss": 0.4004,
      "step": 36543
    },
    {
      "epoch": 2.046365774442827,
      "grad_norm": 1.3630045652389526,
      "learning_rate": 9.565026315789473e-05,
      "loss": 0.4935,
      "step": 36544
    },
    {
      "epoch": 2.0464217717549555,
      "grad_norm": 1.0712352991104126,
      "learning_rate": 9.565000000000001e-05,
      "loss": 0.3826,
      "step": 36545
    },
    {
      "epoch": 2.0464777690670846,
      "grad_norm": 1.3000266551971436,
      "learning_rate": 9.564973684210527e-05,
      "loss": 0.469,
      "step": 36546
    },
    {
      "epoch": 2.0465337663792136,
      "grad_norm": 1.2789708375930786,
      "learning_rate": 9.564947368421053e-05,
      "loss": 0.4323,
      "step": 36547
    },
    {
      "epoch": 2.0465897636913426,
      "grad_norm": 1.2918531894683838,
      "learning_rate": 9.564921052631579e-05,
      "loss": 0.4387,
      "step": 36548
    },
    {
      "epoch": 2.0466457610034716,
      "grad_norm": 0.9122598767280579,
      "learning_rate": 9.564894736842106e-05,
      "loss": 0.297,
      "step": 36549
    },
    {
      "epoch": 2.0467017583156006,
      "grad_norm": 1.3538532257080078,
      "learning_rate": 9.564868421052632e-05,
      "loss": 0.4952,
      "step": 36550
    },
    {
      "epoch": 2.0467577556277297,
      "grad_norm": 1.0562924146652222,
      "learning_rate": 9.564842105263159e-05,
      "loss": 0.4113,
      "step": 36551
    },
    {
      "epoch": 2.0468137529398587,
      "grad_norm": 1.6232126951217651,
      "learning_rate": 9.564815789473684e-05,
      "loss": 0.5564,
      "step": 36552
    },
    {
      "epoch": 2.0468697502519877,
      "grad_norm": 1.0307356119155884,
      "learning_rate": 9.564789473684211e-05,
      "loss": 0.3868,
      "step": 36553
    },
    {
      "epoch": 2.0469257475641167,
      "grad_norm": 1.0779378414154053,
      "learning_rate": 9.564763157894737e-05,
      "loss": 0.357,
      "step": 36554
    },
    {
      "epoch": 2.0469817448762457,
      "grad_norm": 1.2754429578781128,
      "learning_rate": 9.564736842105265e-05,
      "loss": 0.3737,
      "step": 36555
    },
    {
      "epoch": 2.0470377421883748,
      "grad_norm": 1.0556129217147827,
      "learning_rate": 9.56471052631579e-05,
      "loss": 0.4356,
      "step": 36556
    },
    {
      "epoch": 2.0470937395005038,
      "grad_norm": 1.3854846954345703,
      "learning_rate": 9.564684210526316e-05,
      "loss": 0.3546,
      "step": 36557
    },
    {
      "epoch": 2.047149736812633,
      "grad_norm": 1.2450940608978271,
      "learning_rate": 9.564657894736842e-05,
      "loss": 0.4489,
      "step": 36558
    },
    {
      "epoch": 2.047205734124762,
      "grad_norm": 1.2619494199752808,
      "learning_rate": 9.564631578947368e-05,
      "loss": 0.396,
      "step": 36559
    },
    {
      "epoch": 2.047261731436891,
      "grad_norm": 1.088112711906433,
      "learning_rate": 9.564605263157896e-05,
      "loss": 0.338,
      "step": 36560
    },
    {
      "epoch": 2.04731772874902,
      "grad_norm": 1.1738293170928955,
      "learning_rate": 9.564578947368422e-05,
      "loss": 0.3759,
      "step": 36561
    },
    {
      "epoch": 2.047373726061149,
      "grad_norm": 1.100028157234192,
      "learning_rate": 9.564552631578948e-05,
      "loss": 0.4259,
      "step": 36562
    },
    {
      "epoch": 2.047429723373278,
      "grad_norm": 1.1526410579681396,
      "learning_rate": 9.564526315789474e-05,
      "loss": 0.3414,
      "step": 36563
    },
    {
      "epoch": 2.047485720685407,
      "grad_norm": 1.3450732231140137,
      "learning_rate": 9.564500000000001e-05,
      "loss": 0.482,
      "step": 36564
    },
    {
      "epoch": 2.047541717997536,
      "grad_norm": 1.2071354389190674,
      "learning_rate": 9.564473684210527e-05,
      "loss": 0.3974,
      "step": 36565
    },
    {
      "epoch": 2.047597715309665,
      "grad_norm": 1.3167060613632202,
      "learning_rate": 9.564447368421053e-05,
      "loss": 0.4767,
      "step": 36566
    },
    {
      "epoch": 2.047653712621794,
      "grad_norm": 1.1307607889175415,
      "learning_rate": 9.564421052631579e-05,
      "loss": 0.3596,
      "step": 36567
    },
    {
      "epoch": 2.047709709933923,
      "grad_norm": 1.1424685716629028,
      "learning_rate": 9.564394736842106e-05,
      "loss": 0.4913,
      "step": 36568
    },
    {
      "epoch": 2.047765707246052,
      "grad_norm": 1.3867213726043701,
      "learning_rate": 9.564368421052632e-05,
      "loss": 0.468,
      "step": 36569
    },
    {
      "epoch": 2.047821704558181,
      "grad_norm": 0.9291280508041382,
      "learning_rate": 9.564342105263158e-05,
      "loss": 0.311,
      "step": 36570
    },
    {
      "epoch": 2.04787770187031,
      "grad_norm": 1.4469846487045288,
      "learning_rate": 9.564315789473684e-05,
      "loss": 0.4706,
      "step": 36571
    },
    {
      "epoch": 2.047933699182439,
      "grad_norm": 1.6718295812606812,
      "learning_rate": 9.564289473684211e-05,
      "loss": 0.3678,
      "step": 36572
    },
    {
      "epoch": 2.047989696494568,
      "grad_norm": 1.3264329433441162,
      "learning_rate": 9.564263157894737e-05,
      "loss": 0.4753,
      "step": 36573
    },
    {
      "epoch": 2.048045693806697,
      "grad_norm": 1.3022792339324951,
      "learning_rate": 9.564236842105265e-05,
      "loss": 0.3413,
      "step": 36574
    },
    {
      "epoch": 2.048101691118826,
      "grad_norm": 1.0563435554504395,
      "learning_rate": 9.56421052631579e-05,
      "loss": 0.4405,
      "step": 36575
    },
    {
      "epoch": 2.048157688430955,
      "grad_norm": 1.2141474485397339,
      "learning_rate": 9.564184210526315e-05,
      "loss": 0.3618,
      "step": 36576
    },
    {
      "epoch": 2.048213685743084,
      "grad_norm": 1.396917700767517,
      "learning_rate": 9.564157894736843e-05,
      "loss": 0.4116,
      "step": 36577
    },
    {
      "epoch": 2.048269683055213,
      "grad_norm": 1.1717616319656372,
      "learning_rate": 9.564131578947369e-05,
      "loss": 0.3369,
      "step": 36578
    },
    {
      "epoch": 2.048325680367342,
      "grad_norm": 1.2501972913742065,
      "learning_rate": 9.564105263157896e-05,
      "loss": 0.3603,
      "step": 36579
    },
    {
      "epoch": 2.048381677679471,
      "grad_norm": 1.0641067028045654,
      "learning_rate": 9.56407894736842e-05,
      "loss": 0.317,
      "step": 36580
    },
    {
      "epoch": 2.0484376749916002,
      "grad_norm": 1.1840697526931763,
      "learning_rate": 9.564052631578948e-05,
      "loss": 0.3738,
      "step": 36581
    },
    {
      "epoch": 2.0484936723037293,
      "grad_norm": 1.3396750688552856,
      "learning_rate": 9.564026315789474e-05,
      "loss": 0.4202,
      "step": 36582
    },
    {
      "epoch": 2.0485496696158583,
      "grad_norm": 1.1548391580581665,
      "learning_rate": 9.564000000000001e-05,
      "loss": 0.4973,
      "step": 36583
    },
    {
      "epoch": 2.0486056669279873,
      "grad_norm": 0.9568310379981995,
      "learning_rate": 9.563973684210526e-05,
      "loss": 0.2957,
      "step": 36584
    },
    {
      "epoch": 2.0486616642401163,
      "grad_norm": 1.3702946901321411,
      "learning_rate": 9.563947368421053e-05,
      "loss": 0.4046,
      "step": 36585
    },
    {
      "epoch": 2.0487176615522453,
      "grad_norm": 1.2944976091384888,
      "learning_rate": 9.563921052631579e-05,
      "loss": 0.3728,
      "step": 36586
    },
    {
      "epoch": 2.0487736588643743,
      "grad_norm": 6.153049468994141,
      "learning_rate": 9.563894736842107e-05,
      "loss": 0.4144,
      "step": 36587
    },
    {
      "epoch": 2.0488296561765034,
      "grad_norm": 1.2084838151931763,
      "learning_rate": 9.563868421052632e-05,
      "loss": 0.4585,
      "step": 36588
    },
    {
      "epoch": 2.0488856534886324,
      "grad_norm": 1.2019833326339722,
      "learning_rate": 9.563842105263158e-05,
      "loss": 0.4591,
      "step": 36589
    },
    {
      "epoch": 2.0489416508007614,
      "grad_norm": 1.2909711599349976,
      "learning_rate": 9.563815789473684e-05,
      "loss": 0.5021,
      "step": 36590
    },
    {
      "epoch": 2.0489976481128904,
      "grad_norm": 1.2167983055114746,
      "learning_rate": 9.563789473684212e-05,
      "loss": 0.301,
      "step": 36591
    },
    {
      "epoch": 2.0490536454250194,
      "grad_norm": 1.307668685913086,
      "learning_rate": 9.563763157894738e-05,
      "loss": 0.4496,
      "step": 36592
    },
    {
      "epoch": 2.0491096427371485,
      "grad_norm": 1.1638180017471313,
      "learning_rate": 9.563736842105264e-05,
      "loss": 0.3485,
      "step": 36593
    },
    {
      "epoch": 2.0491656400492775,
      "grad_norm": 1.4204975366592407,
      "learning_rate": 9.56371052631579e-05,
      "loss": 0.5176,
      "step": 36594
    },
    {
      "epoch": 2.0492216373614065,
      "grad_norm": 2.032364845275879,
      "learning_rate": 9.563684210526316e-05,
      "loss": 0.3821,
      "step": 36595
    },
    {
      "epoch": 2.0492776346735355,
      "grad_norm": 1.2145661115646362,
      "learning_rate": 9.563657894736843e-05,
      "loss": 0.3932,
      "step": 36596
    },
    {
      "epoch": 2.0493336319856645,
      "grad_norm": 1.2356343269348145,
      "learning_rate": 9.563631578947369e-05,
      "loss": 0.3657,
      "step": 36597
    },
    {
      "epoch": 2.0493896292977936,
      "grad_norm": 1.2230480909347534,
      "learning_rate": 9.563605263157895e-05,
      "loss": 0.3893,
      "step": 36598
    },
    {
      "epoch": 2.0494456266099226,
      "grad_norm": 1.3757137060165405,
      "learning_rate": 9.563578947368421e-05,
      "loss": 0.4167,
      "step": 36599
    },
    {
      "epoch": 2.0495016239220516,
      "grad_norm": 1.1271175146102905,
      "learning_rate": 9.563552631578948e-05,
      "loss": 0.3841,
      "step": 36600
    },
    {
      "epoch": 2.0495576212341806,
      "grad_norm": 1.030759334564209,
      "learning_rate": 9.563526315789474e-05,
      "loss": 0.3127,
      "step": 36601
    },
    {
      "epoch": 2.0496136185463096,
      "grad_norm": 1.291495442390442,
      "learning_rate": 9.5635e-05,
      "loss": 0.51,
      "step": 36602
    },
    {
      "epoch": 2.0496696158584387,
      "grad_norm": 0.9779292941093445,
      "learning_rate": 9.563473684210526e-05,
      "loss": 0.3252,
      "step": 36603
    },
    {
      "epoch": 2.0497256131705677,
      "grad_norm": 1.4492605924606323,
      "learning_rate": 9.563447368421053e-05,
      "loss": 0.5175,
      "step": 36604
    },
    {
      "epoch": 2.0497816104826967,
      "grad_norm": 1.5373119115829468,
      "learning_rate": 9.56342105263158e-05,
      "loss": 0.3834,
      "step": 36605
    },
    {
      "epoch": 2.0498376077948257,
      "grad_norm": 1.162558913230896,
      "learning_rate": 9.563394736842107e-05,
      "loss": 0.3834,
      "step": 36606
    },
    {
      "epoch": 2.0498936051069547,
      "grad_norm": 1.3218050003051758,
      "learning_rate": 9.563368421052631e-05,
      "loss": 0.3625,
      "step": 36607
    },
    {
      "epoch": 2.0499496024190837,
      "grad_norm": 1.0814683437347412,
      "learning_rate": 9.563342105263159e-05,
      "loss": 0.3973,
      "step": 36608
    },
    {
      "epoch": 2.0500055997312128,
      "grad_norm": 1.1715924739837646,
      "learning_rate": 9.563315789473685e-05,
      "loss": 0.3398,
      "step": 36609
    },
    {
      "epoch": 2.050061597043342,
      "grad_norm": 1.4802566766738892,
      "learning_rate": 9.56328947368421e-05,
      "loss": 0.4495,
      "step": 36610
    },
    {
      "epoch": 2.050117594355471,
      "grad_norm": 1.1524264812469482,
      "learning_rate": 9.563263157894738e-05,
      "loss": 0.3766,
      "step": 36611
    },
    {
      "epoch": 2.0501735916676,
      "grad_norm": 1.2625375986099243,
      "learning_rate": 9.563236842105263e-05,
      "loss": 0.3898,
      "step": 36612
    },
    {
      "epoch": 2.050229588979729,
      "grad_norm": 1.2026548385620117,
      "learning_rate": 9.56321052631579e-05,
      "loss": 0.4932,
      "step": 36613
    },
    {
      "epoch": 2.050285586291858,
      "grad_norm": 1.5375016927719116,
      "learning_rate": 9.563184210526316e-05,
      "loss": 0.5922,
      "step": 36614
    },
    {
      "epoch": 2.050341583603987,
      "grad_norm": 1.1345345973968506,
      "learning_rate": 9.563157894736843e-05,
      "loss": 0.2752,
      "step": 36615
    },
    {
      "epoch": 2.050397580916116,
      "grad_norm": 1.2007031440734863,
      "learning_rate": 9.563131578947369e-05,
      "loss": 0.4468,
      "step": 36616
    },
    {
      "epoch": 2.050453578228245,
      "grad_norm": 1.315387487411499,
      "learning_rate": 9.563105263157895e-05,
      "loss": 0.5099,
      "step": 36617
    },
    {
      "epoch": 2.050509575540374,
      "grad_norm": 1.5523185729980469,
      "learning_rate": 9.563078947368421e-05,
      "loss": 0.4931,
      "step": 36618
    },
    {
      "epoch": 2.050565572852503,
      "grad_norm": 1.0699571371078491,
      "learning_rate": 9.563052631578948e-05,
      "loss": 0.3211,
      "step": 36619
    },
    {
      "epoch": 2.050621570164632,
      "grad_norm": 1.1341267824172974,
      "learning_rate": 9.563026315789474e-05,
      "loss": 0.287,
      "step": 36620
    },
    {
      "epoch": 2.050677567476761,
      "grad_norm": 1.1612727642059326,
      "learning_rate": 9.563e-05,
      "loss": 0.3456,
      "step": 36621
    },
    {
      "epoch": 2.05073356478889,
      "grad_norm": 1.3249485492706299,
      "learning_rate": 9.562973684210526e-05,
      "loss": 0.4748,
      "step": 36622
    },
    {
      "epoch": 2.050789562101019,
      "grad_norm": 1.222671389579773,
      "learning_rate": 9.562947368421054e-05,
      "loss": 0.4131,
      "step": 36623
    },
    {
      "epoch": 2.050845559413148,
      "grad_norm": 1.3996583223342896,
      "learning_rate": 9.56292105263158e-05,
      "loss": 0.3747,
      "step": 36624
    },
    {
      "epoch": 2.050901556725277,
      "grad_norm": 1.175726056098938,
      "learning_rate": 9.562894736842106e-05,
      "loss": 0.356,
      "step": 36625
    },
    {
      "epoch": 2.050957554037406,
      "grad_norm": 1.2267580032348633,
      "learning_rate": 9.562868421052632e-05,
      "loss": 0.5676,
      "step": 36626
    },
    {
      "epoch": 2.051013551349535,
      "grad_norm": 1.312058925628662,
      "learning_rate": 9.562842105263158e-05,
      "loss": 0.4798,
      "step": 36627
    },
    {
      "epoch": 2.051069548661664,
      "grad_norm": 1.1690709590911865,
      "learning_rate": 9.562815789473685e-05,
      "loss": 0.4803,
      "step": 36628
    },
    {
      "epoch": 2.051125545973793,
      "grad_norm": 1.144781231880188,
      "learning_rate": 9.562789473684211e-05,
      "loss": 0.3738,
      "step": 36629
    },
    {
      "epoch": 2.051181543285922,
      "grad_norm": 1.0730394124984741,
      "learning_rate": 9.562763157894737e-05,
      "loss": 0.3401,
      "step": 36630
    },
    {
      "epoch": 2.051237540598051,
      "grad_norm": 1.4585849046707153,
      "learning_rate": 9.562736842105263e-05,
      "loss": 0.5125,
      "step": 36631
    },
    {
      "epoch": 2.05129353791018,
      "grad_norm": 1.3115953207015991,
      "learning_rate": 9.56271052631579e-05,
      "loss": 0.4958,
      "step": 36632
    },
    {
      "epoch": 2.0513495352223092,
      "grad_norm": 1.5912495851516724,
      "learning_rate": 9.562684210526316e-05,
      "loss": 0.3862,
      "step": 36633
    },
    {
      "epoch": 2.0514055325344382,
      "grad_norm": 1.532239317893982,
      "learning_rate": 9.562657894736842e-05,
      "loss": 0.3606,
      "step": 36634
    },
    {
      "epoch": 2.0514615298465673,
      "grad_norm": 1.2204035520553589,
      "learning_rate": 9.562631578947368e-05,
      "loss": 0.4069,
      "step": 36635
    },
    {
      "epoch": 2.0515175271586963,
      "grad_norm": 1.1472162008285522,
      "learning_rate": 9.562605263157895e-05,
      "loss": 0.295,
      "step": 36636
    },
    {
      "epoch": 2.0515735244708253,
      "grad_norm": 1.5692294836044312,
      "learning_rate": 9.562578947368421e-05,
      "loss": 0.3696,
      "step": 36637
    },
    {
      "epoch": 2.0516295217829543,
      "grad_norm": 1.2900413274765015,
      "learning_rate": 9.562552631578949e-05,
      "loss": 0.3143,
      "step": 36638
    },
    {
      "epoch": 2.0516855190950833,
      "grad_norm": 1.001652479171753,
      "learning_rate": 9.562526315789473e-05,
      "loss": 0.2978,
      "step": 36639
    },
    {
      "epoch": 2.0517415164072124,
      "grad_norm": 1.2048441171646118,
      "learning_rate": 9.562500000000001e-05,
      "loss": 0.4021,
      "step": 36640
    },
    {
      "epoch": 2.0517975137193414,
      "grad_norm": 1.0572541952133179,
      "learning_rate": 9.562473684210527e-05,
      "loss": 0.4368,
      "step": 36641
    },
    {
      "epoch": 2.0518535110314704,
      "grad_norm": 1.110068678855896,
      "learning_rate": 9.562447368421054e-05,
      "loss": 0.3574,
      "step": 36642
    },
    {
      "epoch": 2.0519095083435994,
      "grad_norm": 1.2250903844833374,
      "learning_rate": 9.56242105263158e-05,
      "loss": 0.3853,
      "step": 36643
    },
    {
      "epoch": 2.0519655056557284,
      "grad_norm": 1.0479674339294434,
      "learning_rate": 9.562394736842105e-05,
      "loss": 0.2793,
      "step": 36644
    },
    {
      "epoch": 2.0520215029678575,
      "grad_norm": 1.0499855279922485,
      "learning_rate": 9.562368421052632e-05,
      "loss": 0.362,
      "step": 36645
    },
    {
      "epoch": 2.0520775002799865,
      "grad_norm": 1.1197271347045898,
      "learning_rate": 9.562342105263158e-05,
      "loss": 0.389,
      "step": 36646
    },
    {
      "epoch": 2.0521334975921155,
      "grad_norm": 1.198747158050537,
      "learning_rate": 9.562315789473685e-05,
      "loss": 0.365,
      "step": 36647
    },
    {
      "epoch": 2.0521894949042445,
      "grad_norm": 2.027738094329834,
      "learning_rate": 9.562289473684211e-05,
      "loss": 0.4487,
      "step": 36648
    },
    {
      "epoch": 2.0522454922163735,
      "grad_norm": 1.1502078771591187,
      "learning_rate": 9.562263157894737e-05,
      "loss": 0.4557,
      "step": 36649
    },
    {
      "epoch": 2.0523014895285026,
      "grad_norm": 1.2851338386535645,
      "learning_rate": 9.562236842105263e-05,
      "loss": 0.5174,
      "step": 36650
    },
    {
      "epoch": 2.0523574868406316,
      "grad_norm": 1.2322546243667603,
      "learning_rate": 9.56221052631579e-05,
      "loss": 0.4128,
      "step": 36651
    },
    {
      "epoch": 2.0524134841527606,
      "grad_norm": 1.146069884300232,
      "learning_rate": 9.562184210526316e-05,
      "loss": 0.376,
      "step": 36652
    },
    {
      "epoch": 2.0524694814648896,
      "grad_norm": 1.13925039768219,
      "learning_rate": 9.562157894736842e-05,
      "loss": 0.3688,
      "step": 36653
    },
    {
      "epoch": 2.0525254787770186,
      "grad_norm": 1.842946171760559,
      "learning_rate": 9.562131578947368e-05,
      "loss": 0.4901,
      "step": 36654
    },
    {
      "epoch": 2.0525814760891476,
      "grad_norm": 1.4088022708892822,
      "learning_rate": 9.562105263157896e-05,
      "loss": 0.5086,
      "step": 36655
    },
    {
      "epoch": 2.0526374734012767,
      "grad_norm": 1.1978005170822144,
      "learning_rate": 9.562078947368422e-05,
      "loss": 0.4408,
      "step": 36656
    },
    {
      "epoch": 2.0526934707134057,
      "grad_norm": 1.1670106649398804,
      "learning_rate": 9.562052631578948e-05,
      "loss": 0.3121,
      "step": 36657
    },
    {
      "epoch": 2.0527494680255347,
      "grad_norm": 1.2237215042114258,
      "learning_rate": 9.562026315789474e-05,
      "loss": 0.3633,
      "step": 36658
    },
    {
      "epoch": 2.0528054653376637,
      "grad_norm": 1.2898637056350708,
      "learning_rate": 9.562000000000001e-05,
      "loss": 0.2586,
      "step": 36659
    },
    {
      "epoch": 2.0528614626497927,
      "grad_norm": 1.11089289188385,
      "learning_rate": 9.561973684210527e-05,
      "loss": 0.3754,
      "step": 36660
    },
    {
      "epoch": 2.0529174599619218,
      "grad_norm": 1.5122790336608887,
      "learning_rate": 9.561947368421053e-05,
      "loss": 0.3777,
      "step": 36661
    },
    {
      "epoch": 2.052973457274051,
      "grad_norm": 1.288523554801941,
      "learning_rate": 9.561921052631579e-05,
      "loss": 0.3686,
      "step": 36662
    },
    {
      "epoch": 2.05302945458618,
      "grad_norm": 1.0326403379440308,
      "learning_rate": 9.561894736842105e-05,
      "loss": 0.3671,
      "step": 36663
    },
    {
      "epoch": 2.053085451898309,
      "grad_norm": 1.1743468046188354,
      "learning_rate": 9.561868421052632e-05,
      "loss": 0.3457,
      "step": 36664
    },
    {
      "epoch": 2.053141449210438,
      "grad_norm": 1.2513277530670166,
      "learning_rate": 9.561842105263158e-05,
      "loss": 0.3704,
      "step": 36665
    },
    {
      "epoch": 2.053197446522567,
      "grad_norm": 1.179674506187439,
      "learning_rate": 9.561815789473685e-05,
      "loss": 0.3202,
      "step": 36666
    },
    {
      "epoch": 2.053253443834696,
      "grad_norm": 1.2520854473114014,
      "learning_rate": 9.56178947368421e-05,
      "loss": 0.3949,
      "step": 36667
    },
    {
      "epoch": 2.053309441146825,
      "grad_norm": 1.4296391010284424,
      "learning_rate": 9.561763157894737e-05,
      "loss": 0.3317,
      "step": 36668
    },
    {
      "epoch": 2.053365438458954,
      "grad_norm": 1.0658658742904663,
      "learning_rate": 9.561736842105263e-05,
      "loss": 0.3226,
      "step": 36669
    },
    {
      "epoch": 2.053421435771083,
      "grad_norm": 1.2129640579223633,
      "learning_rate": 9.561710526315791e-05,
      "loss": 0.3908,
      "step": 36670
    },
    {
      "epoch": 2.053477433083212,
      "grad_norm": 1.2850233316421509,
      "learning_rate": 9.561684210526317e-05,
      "loss": 0.4048,
      "step": 36671
    },
    {
      "epoch": 2.053533430395341,
      "grad_norm": 1.4278819561004639,
      "learning_rate": 9.561657894736843e-05,
      "loss": 0.4404,
      "step": 36672
    },
    {
      "epoch": 2.05358942770747,
      "grad_norm": 1.3692359924316406,
      "learning_rate": 9.561631578947369e-05,
      "loss": 0.3954,
      "step": 36673
    },
    {
      "epoch": 2.053645425019599,
      "grad_norm": 1.4084349870681763,
      "learning_rate": 9.561605263157896e-05,
      "loss": 0.6925,
      "step": 36674
    },
    {
      "epoch": 2.053701422331728,
      "grad_norm": 1.0658395290374756,
      "learning_rate": 9.561578947368422e-05,
      "loss": 0.3643,
      "step": 36675
    },
    {
      "epoch": 2.053757419643857,
      "grad_norm": 1.1298015117645264,
      "learning_rate": 9.561552631578948e-05,
      "loss": 0.3538,
      "step": 36676
    },
    {
      "epoch": 2.053813416955986,
      "grad_norm": 1.3170356750488281,
      "learning_rate": 9.561526315789474e-05,
      "loss": 0.5105,
      "step": 36677
    },
    {
      "epoch": 2.053869414268115,
      "grad_norm": 1.5675780773162842,
      "learning_rate": 9.561500000000001e-05,
      "loss": 0.3579,
      "step": 36678
    },
    {
      "epoch": 2.053925411580244,
      "grad_norm": 1.1146208047866821,
      "learning_rate": 9.561473684210527e-05,
      "loss": 0.3776,
      "step": 36679
    },
    {
      "epoch": 2.053981408892373,
      "grad_norm": 1.0981762409210205,
      "learning_rate": 9.561447368421053e-05,
      "loss": 0.3284,
      "step": 36680
    },
    {
      "epoch": 2.054037406204502,
      "grad_norm": 1.0860785245895386,
      "learning_rate": 9.561421052631579e-05,
      "loss": 0.4252,
      "step": 36681
    },
    {
      "epoch": 2.054093403516631,
      "grad_norm": 1.0068347454071045,
      "learning_rate": 9.561394736842105e-05,
      "loss": 0.3325,
      "step": 36682
    },
    {
      "epoch": 2.05414940082876,
      "grad_norm": 1.085824728012085,
      "learning_rate": 9.561368421052632e-05,
      "loss": 0.4278,
      "step": 36683
    },
    {
      "epoch": 2.054205398140889,
      "grad_norm": 1.3052728176116943,
      "learning_rate": 9.561342105263158e-05,
      "loss": 0.3933,
      "step": 36684
    },
    {
      "epoch": 2.0542613954530182,
      "grad_norm": 1.1264413595199585,
      "learning_rate": 9.561315789473684e-05,
      "loss": 0.4286,
      "step": 36685
    },
    {
      "epoch": 2.0543173927651472,
      "grad_norm": 1.3334736824035645,
      "learning_rate": 9.56128947368421e-05,
      "loss": 0.3909,
      "step": 36686
    },
    {
      "epoch": 2.0543733900772763,
      "grad_norm": 1.1422628164291382,
      "learning_rate": 9.561263157894738e-05,
      "loss": 0.3922,
      "step": 36687
    },
    {
      "epoch": 2.0544293873894053,
      "grad_norm": 1.257146954536438,
      "learning_rate": 9.561236842105264e-05,
      "loss": 0.3271,
      "step": 36688
    },
    {
      "epoch": 2.0544853847015343,
      "grad_norm": 1.1288529634475708,
      "learning_rate": 9.56121052631579e-05,
      "loss": 0.3966,
      "step": 36689
    },
    {
      "epoch": 2.0545413820136633,
      "grad_norm": 1.121405005455017,
      "learning_rate": 9.561184210526316e-05,
      "loss": 0.3233,
      "step": 36690
    },
    {
      "epoch": 2.0545973793257923,
      "grad_norm": 1.1121511459350586,
      "learning_rate": 9.561157894736843e-05,
      "loss": 0.3844,
      "step": 36691
    },
    {
      "epoch": 2.0546533766379214,
      "grad_norm": 1.195992112159729,
      "learning_rate": 9.561131578947369e-05,
      "loss": 0.3577,
      "step": 36692
    },
    {
      "epoch": 2.0547093739500504,
      "grad_norm": 1.3259330987930298,
      "learning_rate": 9.561105263157896e-05,
      "loss": 0.372,
      "step": 36693
    },
    {
      "epoch": 2.0547653712621794,
      "grad_norm": 1.043801188468933,
      "learning_rate": 9.561078947368421e-05,
      "loss": 0.3794,
      "step": 36694
    },
    {
      "epoch": 2.0548213685743084,
      "grad_norm": 1.0895954370498657,
      "learning_rate": 9.561052631578948e-05,
      "loss": 0.4274,
      "step": 36695
    },
    {
      "epoch": 2.0548773658864374,
      "grad_norm": 1.207619309425354,
      "learning_rate": 9.561026315789474e-05,
      "loss": 0.4047,
      "step": 36696
    },
    {
      "epoch": 2.0549333631985665,
      "grad_norm": 1.4847220182418823,
      "learning_rate": 9.561e-05,
      "loss": 0.5094,
      "step": 36697
    },
    {
      "epoch": 2.0549893605106955,
      "grad_norm": 1.1867951154708862,
      "learning_rate": 9.560973684210527e-05,
      "loss": 0.2936,
      "step": 36698
    },
    {
      "epoch": 2.0550453578228245,
      "grad_norm": 1.5140495300292969,
      "learning_rate": 9.560947368421052e-05,
      "loss": 0.507,
      "step": 36699
    },
    {
      "epoch": 2.0551013551349535,
      "grad_norm": 1.1378545761108398,
      "learning_rate": 9.56092105263158e-05,
      "loss": 0.371,
      "step": 36700
    },
    {
      "epoch": 2.0551573524470825,
      "grad_norm": 1.1760331392288208,
      "learning_rate": 9.560894736842105e-05,
      "loss": 0.395,
      "step": 36701
    },
    {
      "epoch": 2.0552133497592115,
      "grad_norm": 1.1640903949737549,
      "learning_rate": 9.560868421052633e-05,
      "loss": 0.4455,
      "step": 36702
    },
    {
      "epoch": 2.0552693470713406,
      "grad_norm": 1.2893195152282715,
      "learning_rate": 9.560842105263159e-05,
      "loss": 0.5465,
      "step": 36703
    },
    {
      "epoch": 2.0553253443834696,
      "grad_norm": 1.3953769207000732,
      "learning_rate": 9.560815789473685e-05,
      "loss": 0.4445,
      "step": 36704
    },
    {
      "epoch": 2.0553813416955986,
      "grad_norm": 1.0773518085479736,
      "learning_rate": 9.56078947368421e-05,
      "loss": 0.3723,
      "step": 36705
    },
    {
      "epoch": 2.0554373390077276,
      "grad_norm": 1.1542632579803467,
      "learning_rate": 9.560763157894738e-05,
      "loss": 0.4239,
      "step": 36706
    },
    {
      "epoch": 2.0554933363198566,
      "grad_norm": 1.3278210163116455,
      "learning_rate": 9.560736842105264e-05,
      "loss": 0.4774,
      "step": 36707
    },
    {
      "epoch": 2.0555493336319857,
      "grad_norm": 1.103522777557373,
      "learning_rate": 9.56071052631579e-05,
      "loss": 0.392,
      "step": 36708
    },
    {
      "epoch": 2.0556053309441147,
      "grad_norm": 1.2399300336837769,
      "learning_rate": 9.560684210526316e-05,
      "loss": 0.5026,
      "step": 36709
    },
    {
      "epoch": 2.0556613282562437,
      "grad_norm": 1.3378663063049316,
      "learning_rate": 9.560657894736843e-05,
      "loss": 0.4297,
      "step": 36710
    },
    {
      "epoch": 2.0557173255683727,
      "grad_norm": 1.0820890665054321,
      "learning_rate": 9.560631578947369e-05,
      "loss": 0.37,
      "step": 36711
    },
    {
      "epoch": 2.0557733228805017,
      "grad_norm": 1.1408289670944214,
      "learning_rate": 9.560605263157895e-05,
      "loss": 0.4069,
      "step": 36712
    },
    {
      "epoch": 2.0558293201926308,
      "grad_norm": 1.0384246110916138,
      "learning_rate": 9.560578947368421e-05,
      "loss": 0.349,
      "step": 36713
    },
    {
      "epoch": 2.05588531750476,
      "grad_norm": 1.3504265546798706,
      "learning_rate": 9.560552631578947e-05,
      "loss": 0.5331,
      "step": 36714
    },
    {
      "epoch": 2.055941314816889,
      "grad_norm": 1.1632843017578125,
      "learning_rate": 9.560526315789474e-05,
      "loss": 0.3979,
      "step": 36715
    },
    {
      "epoch": 2.055997312129018,
      "grad_norm": 1.5368894338607788,
      "learning_rate": 9.5605e-05,
      "loss": 0.5003,
      "step": 36716
    },
    {
      "epoch": 2.056053309441147,
      "grad_norm": 1.1824567317962646,
      "learning_rate": 9.560473684210526e-05,
      "loss": 0.4659,
      "step": 36717
    },
    {
      "epoch": 2.056109306753276,
      "grad_norm": 1.1657074689865112,
      "learning_rate": 9.560447368421052e-05,
      "loss": 0.4018,
      "step": 36718
    },
    {
      "epoch": 2.056165304065405,
      "grad_norm": 1.632127285003662,
      "learning_rate": 9.56042105263158e-05,
      "loss": 0.3002,
      "step": 36719
    },
    {
      "epoch": 2.056221301377534,
      "grad_norm": 1.1505695581436157,
      "learning_rate": 9.560394736842106e-05,
      "loss": 0.3992,
      "step": 36720
    },
    {
      "epoch": 2.056277298689663,
      "grad_norm": 1.2248297929763794,
      "learning_rate": 9.560368421052633e-05,
      "loss": 0.42,
      "step": 36721
    },
    {
      "epoch": 2.056333296001792,
      "grad_norm": 1.1423290967941284,
      "learning_rate": 9.560342105263158e-05,
      "loss": 0.3792,
      "step": 36722
    },
    {
      "epoch": 2.056389293313921,
      "grad_norm": 1.2099378108978271,
      "learning_rate": 9.560315789473685e-05,
      "loss": 0.3225,
      "step": 36723
    },
    {
      "epoch": 2.05644529062605,
      "grad_norm": 1.0739911794662476,
      "learning_rate": 9.560289473684211e-05,
      "loss": 0.4086,
      "step": 36724
    },
    {
      "epoch": 2.056501287938179,
      "grad_norm": 1.005111813545227,
      "learning_rate": 9.560263157894738e-05,
      "loss": 0.3213,
      "step": 36725
    },
    {
      "epoch": 2.056557285250308,
      "grad_norm": 1.2234336137771606,
      "learning_rate": 9.560236842105264e-05,
      "loss": 0.487,
      "step": 36726
    },
    {
      "epoch": 2.056613282562437,
      "grad_norm": 1.187186360359192,
      "learning_rate": 9.56021052631579e-05,
      "loss": 0.3481,
      "step": 36727
    },
    {
      "epoch": 2.056669279874566,
      "grad_norm": 1.2229480743408203,
      "learning_rate": 9.560184210526316e-05,
      "loss": 0.3494,
      "step": 36728
    },
    {
      "epoch": 2.056725277186695,
      "grad_norm": 1.3513422012329102,
      "learning_rate": 9.560157894736843e-05,
      "loss": 0.4719,
      "step": 36729
    },
    {
      "epoch": 2.056781274498824,
      "grad_norm": 1.0294100046157837,
      "learning_rate": 9.56013157894737e-05,
      "loss": 0.3285,
      "step": 36730
    },
    {
      "epoch": 2.056837271810953,
      "grad_norm": 1.206562876701355,
      "learning_rate": 9.560105263157894e-05,
      "loss": 0.3717,
      "step": 36731
    },
    {
      "epoch": 2.056893269123082,
      "grad_norm": 1.3168164491653442,
      "learning_rate": 9.560078947368421e-05,
      "loss": 0.3847,
      "step": 36732
    },
    {
      "epoch": 2.056949266435211,
      "grad_norm": 1.4654356241226196,
      "learning_rate": 9.560052631578947e-05,
      "loss": 0.443,
      "step": 36733
    },
    {
      "epoch": 2.05700526374734,
      "grad_norm": 1.2013304233551025,
      "learning_rate": 9.560026315789475e-05,
      "loss": 0.4463,
      "step": 36734
    },
    {
      "epoch": 2.057061261059469,
      "grad_norm": 1.1533199548721313,
      "learning_rate": 9.56e-05,
      "loss": 0.3376,
      "step": 36735
    },
    {
      "epoch": 2.057117258371598,
      "grad_norm": 1.1251952648162842,
      "learning_rate": 9.559973684210527e-05,
      "loss": 0.3469,
      "step": 36736
    },
    {
      "epoch": 2.057173255683727,
      "grad_norm": 1.171920657157898,
      "learning_rate": 9.559947368421053e-05,
      "loss": 0.3549,
      "step": 36737
    },
    {
      "epoch": 2.0572292529958562,
      "grad_norm": 1.260909914970398,
      "learning_rate": 9.55992105263158e-05,
      "loss": 0.435,
      "step": 36738
    },
    {
      "epoch": 2.0572852503079853,
      "grad_norm": 1.2944713830947876,
      "learning_rate": 9.559894736842106e-05,
      "loss": 0.5194,
      "step": 36739
    },
    {
      "epoch": 2.0573412476201143,
      "grad_norm": 1.2429436445236206,
      "learning_rate": 9.559868421052632e-05,
      "loss": 0.3908,
      "step": 36740
    },
    {
      "epoch": 2.0573972449322433,
      "grad_norm": 1.3853048086166382,
      "learning_rate": 9.559842105263158e-05,
      "loss": 0.5268,
      "step": 36741
    },
    {
      "epoch": 2.0574532422443723,
      "grad_norm": 1.4531614780426025,
      "learning_rate": 9.559815789473685e-05,
      "loss": 0.4978,
      "step": 36742
    },
    {
      "epoch": 2.0575092395565013,
      "grad_norm": 1.424088478088379,
      "learning_rate": 9.559789473684211e-05,
      "loss": 0.5105,
      "step": 36743
    },
    {
      "epoch": 2.0575652368686304,
      "grad_norm": 1.2815285921096802,
      "learning_rate": 9.559763157894737e-05,
      "loss": 0.4422,
      "step": 36744
    },
    {
      "epoch": 2.0576212341807594,
      "grad_norm": 1.1937334537506104,
      "learning_rate": 9.559736842105263e-05,
      "loss": 0.338,
      "step": 36745
    },
    {
      "epoch": 2.0576772314928884,
      "grad_norm": 1.1087217330932617,
      "learning_rate": 9.55971052631579e-05,
      "loss": 0.3856,
      "step": 36746
    },
    {
      "epoch": 2.0577332288050174,
      "grad_norm": 1.173322319984436,
      "learning_rate": 9.559684210526316e-05,
      "loss": 0.3706,
      "step": 36747
    },
    {
      "epoch": 2.0577892261171464,
      "grad_norm": 1.1126890182495117,
      "learning_rate": 9.559657894736842e-05,
      "loss": 0.3902,
      "step": 36748
    },
    {
      "epoch": 2.0578452234292754,
      "grad_norm": 1.0003615617752075,
      "learning_rate": 9.559631578947368e-05,
      "loss": 0.3936,
      "step": 36749
    },
    {
      "epoch": 2.0579012207414045,
      "grad_norm": 1.1534682512283325,
      "learning_rate": 9.559605263157894e-05,
      "loss": 0.404,
      "step": 36750
    },
    {
      "epoch": 2.0579572180535335,
      "grad_norm": 1.2277191877365112,
      "learning_rate": 9.559578947368422e-05,
      "loss": 0.3616,
      "step": 36751
    },
    {
      "epoch": 2.0580132153656625,
      "grad_norm": 1.795446753501892,
      "learning_rate": 9.559552631578948e-05,
      "loss": 0.4194,
      "step": 36752
    },
    {
      "epoch": 2.0580692126777915,
      "grad_norm": 1.590470790863037,
      "learning_rate": 9.559526315789475e-05,
      "loss": 0.583,
      "step": 36753
    },
    {
      "epoch": 2.0581252099899205,
      "grad_norm": 1.2584571838378906,
      "learning_rate": 9.5595e-05,
      "loss": 0.413,
      "step": 36754
    },
    {
      "epoch": 2.0581812073020496,
      "grad_norm": 1.623884677886963,
      "learning_rate": 9.559473684210527e-05,
      "loss": 0.5619,
      "step": 36755
    },
    {
      "epoch": 2.0582372046141786,
      "grad_norm": 1.3221566677093506,
      "learning_rate": 9.559447368421053e-05,
      "loss": 0.3693,
      "step": 36756
    },
    {
      "epoch": 2.0582932019263076,
      "grad_norm": 1.0448098182678223,
      "learning_rate": 9.55942105263158e-05,
      "loss": 0.5021,
      "step": 36757
    },
    {
      "epoch": 2.0583491992384366,
      "grad_norm": 1.2021024227142334,
      "learning_rate": 9.559394736842106e-05,
      "loss": 0.4689,
      "step": 36758
    },
    {
      "epoch": 2.0584051965505656,
      "grad_norm": 1.207700490951538,
      "learning_rate": 9.559368421052632e-05,
      "loss": 0.4446,
      "step": 36759
    },
    {
      "epoch": 2.0584611938626947,
      "grad_norm": 1.25325345993042,
      "learning_rate": 9.559342105263158e-05,
      "loss": 0.4801,
      "step": 36760
    },
    {
      "epoch": 2.0585171911748237,
      "grad_norm": 1.231303334236145,
      "learning_rate": 9.559315789473685e-05,
      "loss": 0.3779,
      "step": 36761
    },
    {
      "epoch": 2.0585731884869527,
      "grad_norm": 1.3286206722259521,
      "learning_rate": 9.559289473684211e-05,
      "loss": 0.7109,
      "step": 36762
    },
    {
      "epoch": 2.0586291857990817,
      "grad_norm": 1.3455702066421509,
      "learning_rate": 9.559263157894737e-05,
      "loss": 0.4081,
      "step": 36763
    },
    {
      "epoch": 2.0586851831112107,
      "grad_norm": 1.2441279888153076,
      "learning_rate": 9.559236842105263e-05,
      "loss": 0.4919,
      "step": 36764
    },
    {
      "epoch": 2.0587411804233398,
      "grad_norm": 1.1973718404769897,
      "learning_rate": 9.559210526315789e-05,
      "loss": 0.5086,
      "step": 36765
    },
    {
      "epoch": 2.0587971777354688,
      "grad_norm": 1.0812766551971436,
      "learning_rate": 9.559184210526317e-05,
      "loss": 0.2827,
      "step": 36766
    },
    {
      "epoch": 2.058853175047598,
      "grad_norm": 1.2745712995529175,
      "learning_rate": 9.559157894736843e-05,
      "loss": 0.5086,
      "step": 36767
    },
    {
      "epoch": 2.058909172359727,
      "grad_norm": 1.1130268573760986,
      "learning_rate": 9.559131578947369e-05,
      "loss": 0.4475,
      "step": 36768
    },
    {
      "epoch": 2.058965169671856,
      "grad_norm": 1.0718028545379639,
      "learning_rate": 9.559105263157895e-05,
      "loss": 0.3888,
      "step": 36769
    },
    {
      "epoch": 2.059021166983985,
      "grad_norm": 1.0981756448745728,
      "learning_rate": 9.559078947368422e-05,
      "loss": 0.4471,
      "step": 36770
    },
    {
      "epoch": 2.059077164296114,
      "grad_norm": 1.0213072299957275,
      "learning_rate": 9.559052631578948e-05,
      "loss": 0.3512,
      "step": 36771
    },
    {
      "epoch": 2.059133161608243,
      "grad_norm": 1.099674105644226,
      "learning_rate": 9.559026315789474e-05,
      "loss": 0.3833,
      "step": 36772
    },
    {
      "epoch": 2.059189158920372,
      "grad_norm": 1.295213222503662,
      "learning_rate": 9.559e-05,
      "loss": 0.3492,
      "step": 36773
    },
    {
      "epoch": 2.059245156232501,
      "grad_norm": 1.330477237701416,
      "learning_rate": 9.558973684210527e-05,
      "loss": 0.5201,
      "step": 36774
    },
    {
      "epoch": 2.05930115354463,
      "grad_norm": 1.1716090440750122,
      "learning_rate": 9.558947368421053e-05,
      "loss": 0.367,
      "step": 36775
    },
    {
      "epoch": 2.059357150856759,
      "grad_norm": 1.090935468673706,
      "learning_rate": 9.55892105263158e-05,
      "loss": 0.312,
      "step": 36776
    },
    {
      "epoch": 2.059413148168888,
      "grad_norm": 1.1700232028961182,
      "learning_rate": 9.558894736842105e-05,
      "loss": 0.3055,
      "step": 36777
    },
    {
      "epoch": 2.059469145481017,
      "grad_norm": 1.268692135810852,
      "learning_rate": 9.558868421052632e-05,
      "loss": 0.3626,
      "step": 36778
    },
    {
      "epoch": 2.059525142793146,
      "grad_norm": 0.9433198571205139,
      "learning_rate": 9.558842105263158e-05,
      "loss": 0.3012,
      "step": 36779
    },
    {
      "epoch": 2.059581140105275,
      "grad_norm": 1.2524492740631104,
      "learning_rate": 9.558815789473686e-05,
      "loss": 0.429,
      "step": 36780
    },
    {
      "epoch": 2.059637137417404,
      "grad_norm": 1.4334501028060913,
      "learning_rate": 9.558789473684212e-05,
      "loss": 0.4208,
      "step": 36781
    },
    {
      "epoch": 2.059693134729533,
      "grad_norm": 1.0466660261154175,
      "learning_rate": 9.558763157894736e-05,
      "loss": 0.3498,
      "step": 36782
    },
    {
      "epoch": 2.059749132041662,
      "grad_norm": 1.1548635959625244,
      "learning_rate": 9.558736842105264e-05,
      "loss": 0.3169,
      "step": 36783
    },
    {
      "epoch": 2.059805129353791,
      "grad_norm": 1.2209688425064087,
      "learning_rate": 9.55871052631579e-05,
      "loss": 0.4727,
      "step": 36784
    },
    {
      "epoch": 2.05986112666592,
      "grad_norm": 1.0632095336914062,
      "learning_rate": 9.558684210526317e-05,
      "loss": 0.3624,
      "step": 36785
    },
    {
      "epoch": 2.059917123978049,
      "grad_norm": 1.1065341234207153,
      "learning_rate": 9.558657894736842e-05,
      "loss": 0.3032,
      "step": 36786
    },
    {
      "epoch": 2.059973121290178,
      "grad_norm": 1.1085914373397827,
      "learning_rate": 9.558631578947369e-05,
      "loss": 0.3335,
      "step": 36787
    },
    {
      "epoch": 2.060029118602307,
      "grad_norm": 1.2254257202148438,
      "learning_rate": 9.558605263157895e-05,
      "loss": 0.4998,
      "step": 36788
    },
    {
      "epoch": 2.060085115914436,
      "grad_norm": 1.5903806686401367,
      "learning_rate": 9.558578947368422e-05,
      "loss": 0.3586,
      "step": 36789
    },
    {
      "epoch": 2.0601411132265652,
      "grad_norm": 1.1134637594223022,
      "learning_rate": 9.558552631578948e-05,
      "loss": 0.3995,
      "step": 36790
    },
    {
      "epoch": 2.0601971105386943,
      "grad_norm": 1.1553540229797363,
      "learning_rate": 9.558526315789474e-05,
      "loss": 0.44,
      "step": 36791
    },
    {
      "epoch": 2.0602531078508233,
      "grad_norm": 1.0799517631530762,
      "learning_rate": 9.5585e-05,
      "loss": 0.3405,
      "step": 36792
    },
    {
      "epoch": 2.0603091051629523,
      "grad_norm": 1.3857663869857788,
      "learning_rate": 9.558473684210527e-05,
      "loss": 0.475,
      "step": 36793
    },
    {
      "epoch": 2.0603651024750813,
      "grad_norm": 1.163667917251587,
      "learning_rate": 9.558447368421053e-05,
      "loss": 0.4205,
      "step": 36794
    },
    {
      "epoch": 2.0604210997872103,
      "grad_norm": 1.2941070795059204,
      "learning_rate": 9.558421052631579e-05,
      "loss": 0.4197,
      "step": 36795
    },
    {
      "epoch": 2.0604770970993393,
      "grad_norm": 1.1041325330734253,
      "learning_rate": 9.558394736842105e-05,
      "loss": 0.5431,
      "step": 36796
    },
    {
      "epoch": 2.0605330944114684,
      "grad_norm": 1.222534418106079,
      "learning_rate": 9.558368421052633e-05,
      "loss": 0.424,
      "step": 36797
    },
    {
      "epoch": 2.0605890917235974,
      "grad_norm": 1.4718118906021118,
      "learning_rate": 9.558342105263159e-05,
      "loss": 0.4878,
      "step": 36798
    },
    {
      "epoch": 2.0606450890357264,
      "grad_norm": 1.2191308736801147,
      "learning_rate": 9.558315789473685e-05,
      "loss": 0.3631,
      "step": 36799
    },
    {
      "epoch": 2.0607010863478554,
      "grad_norm": 1.1393415927886963,
      "learning_rate": 9.55828947368421e-05,
      "loss": 0.3554,
      "step": 36800
    },
    {
      "epoch": 2.0607570836599844,
      "grad_norm": 1.2141400575637817,
      "learning_rate": 9.558263157894737e-05,
      "loss": 0.4108,
      "step": 36801
    },
    {
      "epoch": 2.0608130809721135,
      "grad_norm": 1.2095710039138794,
      "learning_rate": 9.558236842105264e-05,
      "loss": 0.4936,
      "step": 36802
    },
    {
      "epoch": 2.0608690782842425,
      "grad_norm": 1.177917718887329,
      "learning_rate": 9.55821052631579e-05,
      "loss": 0.4903,
      "step": 36803
    },
    {
      "epoch": 2.0609250755963715,
      "grad_norm": 1.1616653203964233,
      "learning_rate": 9.558184210526316e-05,
      "loss": 0.3547,
      "step": 36804
    },
    {
      "epoch": 2.0609810729085005,
      "grad_norm": 1.171865463256836,
      "learning_rate": 9.558157894736842e-05,
      "loss": 0.3489,
      "step": 36805
    },
    {
      "epoch": 2.0610370702206295,
      "grad_norm": 1.3349839448928833,
      "learning_rate": 9.558131578947369e-05,
      "loss": 0.3907,
      "step": 36806
    },
    {
      "epoch": 2.0610930675327586,
      "grad_norm": 1.1776931285858154,
      "learning_rate": 9.558105263157895e-05,
      "loss": 0.3885,
      "step": 36807
    },
    {
      "epoch": 2.0611490648448876,
      "grad_norm": 1.3368803262710571,
      "learning_rate": 9.558078947368422e-05,
      "loss": 0.565,
      "step": 36808
    },
    {
      "epoch": 2.0612050621570166,
      "grad_norm": 1.3812310695648193,
      "learning_rate": 9.558052631578947e-05,
      "loss": 0.4339,
      "step": 36809
    },
    {
      "epoch": 2.0612610594691456,
      "grad_norm": 1.1804356575012207,
      "learning_rate": 9.558026315789474e-05,
      "loss": 0.3922,
      "step": 36810
    },
    {
      "epoch": 2.0613170567812746,
      "grad_norm": 1.4295260906219482,
      "learning_rate": 9.558e-05,
      "loss": 0.3222,
      "step": 36811
    },
    {
      "epoch": 2.0613730540934037,
      "grad_norm": 1.125641107559204,
      "learning_rate": 9.557973684210528e-05,
      "loss": 0.3687,
      "step": 36812
    },
    {
      "epoch": 2.0614290514055327,
      "grad_norm": 1.2327799797058105,
      "learning_rate": 9.557947368421054e-05,
      "loss": 0.3693,
      "step": 36813
    },
    {
      "epoch": 2.0614850487176617,
      "grad_norm": 1.180492639541626,
      "learning_rate": 9.55792105263158e-05,
      "loss": 0.4699,
      "step": 36814
    },
    {
      "epoch": 2.0615410460297907,
      "grad_norm": 1.3446556329727173,
      "learning_rate": 9.557894736842106e-05,
      "loss": 0.4611,
      "step": 36815
    },
    {
      "epoch": 2.0615970433419197,
      "grad_norm": 1.1068192720413208,
      "learning_rate": 9.557868421052633e-05,
      "loss": 0.335,
      "step": 36816
    },
    {
      "epoch": 2.0616530406540488,
      "grad_norm": 1.1895300149917603,
      "learning_rate": 9.557842105263159e-05,
      "loss": 0.3293,
      "step": 36817
    },
    {
      "epoch": 2.0617090379661778,
      "grad_norm": 1.337397813796997,
      "learning_rate": 9.557815789473685e-05,
      "loss": 0.4127,
      "step": 36818
    },
    {
      "epoch": 2.061765035278307,
      "grad_norm": 1.1548151969909668,
      "learning_rate": 9.557789473684211e-05,
      "loss": 0.3771,
      "step": 36819
    },
    {
      "epoch": 2.061821032590436,
      "grad_norm": 1.3712074756622314,
      "learning_rate": 9.557763157894737e-05,
      "loss": 0.47,
      "step": 36820
    },
    {
      "epoch": 2.061877029902565,
      "grad_norm": 1.128764033317566,
      "learning_rate": 9.557736842105264e-05,
      "loss": 0.4657,
      "step": 36821
    },
    {
      "epoch": 2.061933027214694,
      "grad_norm": 1.2151983976364136,
      "learning_rate": 9.55771052631579e-05,
      "loss": 0.5197,
      "step": 36822
    },
    {
      "epoch": 2.061989024526823,
      "grad_norm": 1.4019395112991333,
      "learning_rate": 9.557684210526316e-05,
      "loss": 0.4865,
      "step": 36823
    },
    {
      "epoch": 2.062045021838952,
      "grad_norm": 1.0516685247421265,
      "learning_rate": 9.557657894736842e-05,
      "loss": 0.3587,
      "step": 36824
    },
    {
      "epoch": 2.062101019151081,
      "grad_norm": 1.0749855041503906,
      "learning_rate": 9.55763157894737e-05,
      "loss": 0.4393,
      "step": 36825
    },
    {
      "epoch": 2.06215701646321,
      "grad_norm": 1.1283698081970215,
      "learning_rate": 9.557605263157895e-05,
      "loss": 0.3505,
      "step": 36826
    },
    {
      "epoch": 2.062213013775339,
      "grad_norm": 1.0680946111679077,
      "learning_rate": 9.557578947368421e-05,
      "loss": 0.3366,
      "step": 36827
    },
    {
      "epoch": 2.062269011087468,
      "grad_norm": 1.1984539031982422,
      "learning_rate": 9.557552631578947e-05,
      "loss": 0.3697,
      "step": 36828
    },
    {
      "epoch": 2.062325008399597,
      "grad_norm": 0.9972808957099915,
      "learning_rate": 9.557526315789475e-05,
      "loss": 0.3212,
      "step": 36829
    },
    {
      "epoch": 2.062381005711726,
      "grad_norm": 1.148760437965393,
      "learning_rate": 9.5575e-05,
      "loss": 0.4308,
      "step": 36830
    },
    {
      "epoch": 2.062437003023855,
      "grad_norm": 1.0589383840560913,
      "learning_rate": 9.557473684210528e-05,
      "loss": 0.3277,
      "step": 36831
    },
    {
      "epoch": 2.062493000335984,
      "grad_norm": 1.4211186170578003,
      "learning_rate": 9.557447368421053e-05,
      "loss": 0.4505,
      "step": 36832
    },
    {
      "epoch": 2.062548997648113,
      "grad_norm": 1.2813228368759155,
      "learning_rate": 9.55742105263158e-05,
      "loss": 0.528,
      "step": 36833
    },
    {
      "epoch": 2.062604994960242,
      "grad_norm": 1.42642343044281,
      "learning_rate": 9.557394736842106e-05,
      "loss": 0.3632,
      "step": 36834
    },
    {
      "epoch": 2.062660992272371,
      "grad_norm": 1.537554144859314,
      "learning_rate": 9.557368421052632e-05,
      "loss": 0.4565,
      "step": 36835
    },
    {
      "epoch": 2.0627169895845,
      "grad_norm": 1.445787787437439,
      "learning_rate": 9.557342105263158e-05,
      "loss": 0.4884,
      "step": 36836
    },
    {
      "epoch": 2.062772986896629,
      "grad_norm": 1.2845577001571655,
      "learning_rate": 9.557315789473684e-05,
      "loss": 0.3517,
      "step": 36837
    },
    {
      "epoch": 2.062828984208758,
      "grad_norm": 1.3901147842407227,
      "learning_rate": 9.557289473684211e-05,
      "loss": 0.4411,
      "step": 36838
    },
    {
      "epoch": 2.062884981520887,
      "grad_norm": 1.1573408842086792,
      "learning_rate": 9.557263157894737e-05,
      "loss": 0.3688,
      "step": 36839
    },
    {
      "epoch": 2.062940978833016,
      "grad_norm": 1.2008999586105347,
      "learning_rate": 9.557236842105264e-05,
      "loss": 0.3334,
      "step": 36840
    },
    {
      "epoch": 2.062996976145145,
      "grad_norm": 1.3475545644760132,
      "learning_rate": 9.557210526315789e-05,
      "loss": 0.3896,
      "step": 36841
    },
    {
      "epoch": 2.0630529734572742,
      "grad_norm": 0.9694061279296875,
      "learning_rate": 9.557184210526316e-05,
      "loss": 0.3302,
      "step": 36842
    },
    {
      "epoch": 2.0631089707694032,
      "grad_norm": 1.5760122537612915,
      "learning_rate": 9.557157894736842e-05,
      "loss": 0.3356,
      "step": 36843
    },
    {
      "epoch": 2.0631649680815323,
      "grad_norm": 1.191697359085083,
      "learning_rate": 9.55713157894737e-05,
      "loss": 0.3838,
      "step": 36844
    },
    {
      "epoch": 2.0632209653936613,
      "grad_norm": 1.1822952032089233,
      "learning_rate": 9.557105263157896e-05,
      "loss": 0.3592,
      "step": 36845
    },
    {
      "epoch": 2.0632769627057903,
      "grad_norm": 1.1636078357696533,
      "learning_rate": 9.557078947368422e-05,
      "loss": 0.396,
      "step": 36846
    },
    {
      "epoch": 2.0633329600179193,
      "grad_norm": 1.4038398265838623,
      "learning_rate": 9.557052631578948e-05,
      "loss": 0.459,
      "step": 36847
    },
    {
      "epoch": 2.0633889573300483,
      "grad_norm": 1.333531141281128,
      "learning_rate": 9.557026315789475e-05,
      "loss": 0.4123,
      "step": 36848
    },
    {
      "epoch": 2.0634449546421774,
      "grad_norm": 1.1316839456558228,
      "learning_rate": 9.557000000000001e-05,
      "loss": 0.3526,
      "step": 36849
    },
    {
      "epoch": 2.0635009519543064,
      "grad_norm": 1.2709190845489502,
      "learning_rate": 9.556973684210527e-05,
      "loss": 0.4048,
      "step": 36850
    },
    {
      "epoch": 2.0635569492664354,
      "grad_norm": 1.1169583797454834,
      "learning_rate": 9.556947368421053e-05,
      "loss": 0.4244,
      "step": 36851
    },
    {
      "epoch": 2.0636129465785644,
      "grad_norm": 1.2875707149505615,
      "learning_rate": 9.556921052631579e-05,
      "loss": 0.4093,
      "step": 36852
    },
    {
      "epoch": 2.0636689438906934,
      "grad_norm": 1.0595837831497192,
      "learning_rate": 9.556894736842106e-05,
      "loss": 0.3113,
      "step": 36853
    },
    {
      "epoch": 2.0637249412028225,
      "grad_norm": 1.3138606548309326,
      "learning_rate": 9.556868421052632e-05,
      "loss": 0.5959,
      "step": 36854
    },
    {
      "epoch": 2.0637809385149515,
      "grad_norm": 1.1631929874420166,
      "learning_rate": 9.556842105263158e-05,
      "loss": 0.4212,
      "step": 36855
    },
    {
      "epoch": 2.0638369358270805,
      "grad_norm": 1.0259475708007812,
      "learning_rate": 9.556815789473684e-05,
      "loss": 0.3314,
      "step": 36856
    },
    {
      "epoch": 2.0638929331392095,
      "grad_norm": 1.2289806604385376,
      "learning_rate": 9.556789473684211e-05,
      "loss": 0.4292,
      "step": 36857
    },
    {
      "epoch": 2.0639489304513385,
      "grad_norm": 1.1908369064331055,
      "learning_rate": 9.556763157894737e-05,
      "loss": 0.3178,
      "step": 36858
    },
    {
      "epoch": 2.0640049277634676,
      "grad_norm": 1.1122509241104126,
      "learning_rate": 9.556736842105263e-05,
      "loss": 0.3663,
      "step": 36859
    },
    {
      "epoch": 2.0640609250755966,
      "grad_norm": 1.2240458726882935,
      "learning_rate": 9.556710526315789e-05,
      "loss": 0.4061,
      "step": 36860
    },
    {
      "epoch": 2.0641169223877256,
      "grad_norm": 1.0180732011795044,
      "learning_rate": 9.556684210526317e-05,
      "loss": 0.4708,
      "step": 36861
    },
    {
      "epoch": 2.0641729196998546,
      "grad_norm": 1.2493585348129272,
      "learning_rate": 9.556657894736843e-05,
      "loss": 0.4976,
      "step": 36862
    },
    {
      "epoch": 2.0642289170119836,
      "grad_norm": 1.1532797813415527,
      "learning_rate": 9.55663157894737e-05,
      "loss": 0.4071,
      "step": 36863
    },
    {
      "epoch": 2.0642849143241127,
      "grad_norm": 1.1726077795028687,
      "learning_rate": 9.556605263157894e-05,
      "loss": 0.4969,
      "step": 36864
    },
    {
      "epoch": 2.0643409116362417,
      "grad_norm": 1.117709755897522,
      "learning_rate": 9.556578947368422e-05,
      "loss": 0.4236,
      "step": 36865
    },
    {
      "epoch": 2.0643969089483707,
      "grad_norm": 1.331722378730774,
      "learning_rate": 9.556552631578948e-05,
      "loss": 0.4288,
      "step": 36866
    },
    {
      "epoch": 2.0644529062604997,
      "grad_norm": 1.3825278282165527,
      "learning_rate": 9.556526315789475e-05,
      "loss": 0.4228,
      "step": 36867
    },
    {
      "epoch": 2.0645089035726287,
      "grad_norm": 1.1908832788467407,
      "learning_rate": 9.556500000000001e-05,
      "loss": 0.3601,
      "step": 36868
    },
    {
      "epoch": 2.0645649008847577,
      "grad_norm": 1.097708821296692,
      "learning_rate": 9.556473684210526e-05,
      "loss": 0.4395,
      "step": 36869
    },
    {
      "epoch": 2.0646208981968863,
      "grad_norm": 1.173099398612976,
      "learning_rate": 9.556447368421053e-05,
      "loss": 0.4075,
      "step": 36870
    },
    {
      "epoch": 2.064676895509016,
      "grad_norm": 1.0582363605499268,
      "learning_rate": 9.556421052631579e-05,
      "loss": 0.4063,
      "step": 36871
    },
    {
      "epoch": 2.0647328928211444,
      "grad_norm": 1.4180046319961548,
      "learning_rate": 9.556394736842106e-05,
      "loss": 0.3628,
      "step": 36872
    },
    {
      "epoch": 2.064788890133274,
      "grad_norm": 1.1703236103057861,
      "learning_rate": 9.556368421052632e-05,
      "loss": 0.4244,
      "step": 36873
    },
    {
      "epoch": 2.0648448874454024,
      "grad_norm": 1.1510074138641357,
      "learning_rate": 9.556342105263158e-05,
      "loss": 0.469,
      "step": 36874
    },
    {
      "epoch": 2.064900884757532,
      "grad_norm": 1.5087875127792358,
      "learning_rate": 9.556315789473684e-05,
      "loss": 0.4926,
      "step": 36875
    },
    {
      "epoch": 2.0649568820696604,
      "grad_norm": 1.0679322481155396,
      "learning_rate": 9.556289473684212e-05,
      "loss": 0.4444,
      "step": 36876
    },
    {
      "epoch": 2.06501287938179,
      "grad_norm": 1.0772632360458374,
      "learning_rate": 9.556263157894738e-05,
      "loss": 0.2651,
      "step": 36877
    },
    {
      "epoch": 2.0650688766939185,
      "grad_norm": 1.3106554746627808,
      "learning_rate": 9.556236842105264e-05,
      "loss": 0.4676,
      "step": 36878
    },
    {
      "epoch": 2.0651248740060475,
      "grad_norm": 1.3430856466293335,
      "learning_rate": 9.55621052631579e-05,
      "loss": 0.5481,
      "step": 36879
    },
    {
      "epoch": 2.0651808713181765,
      "grad_norm": 1.429925560951233,
      "learning_rate": 9.556184210526317e-05,
      "loss": 0.5065,
      "step": 36880
    },
    {
      "epoch": 2.0652368686303055,
      "grad_norm": 1.942503809928894,
      "learning_rate": 9.556157894736843e-05,
      "loss": 0.5501,
      "step": 36881
    },
    {
      "epoch": 2.0652928659424346,
      "grad_norm": 1.3585002422332764,
      "learning_rate": 9.556131578947369e-05,
      "loss": 0.3248,
      "step": 36882
    },
    {
      "epoch": 2.0653488632545636,
      "grad_norm": 1.0788979530334473,
      "learning_rate": 9.556105263157895e-05,
      "loss": 0.3938,
      "step": 36883
    },
    {
      "epoch": 2.0654048605666926,
      "grad_norm": 1.2493810653686523,
      "learning_rate": 9.556078947368422e-05,
      "loss": 0.4479,
      "step": 36884
    },
    {
      "epoch": 2.0654608578788216,
      "grad_norm": 1.247441053390503,
      "learning_rate": 9.556052631578948e-05,
      "loss": 0.3806,
      "step": 36885
    },
    {
      "epoch": 2.0655168551909506,
      "grad_norm": 1.31350576877594,
      "learning_rate": 9.556026315789474e-05,
      "loss": 0.4712,
      "step": 36886
    },
    {
      "epoch": 2.0655728525030796,
      "grad_norm": 1.229966640472412,
      "learning_rate": 9.556e-05,
      "loss": 0.3978,
      "step": 36887
    },
    {
      "epoch": 2.0656288498152087,
      "grad_norm": 1.129414439201355,
      "learning_rate": 9.555973684210526e-05,
      "loss": 0.4401,
      "step": 36888
    },
    {
      "epoch": 2.0656848471273377,
      "grad_norm": 1.2380365133285522,
      "learning_rate": 9.555947368421053e-05,
      "loss": 0.5639,
      "step": 36889
    },
    {
      "epoch": 2.0657408444394667,
      "grad_norm": 1.3425729274749756,
      "learning_rate": 9.555921052631579e-05,
      "loss": 0.467,
      "step": 36890
    },
    {
      "epoch": 2.0657968417515957,
      "grad_norm": 1.3067071437835693,
      "learning_rate": 9.555894736842105e-05,
      "loss": 0.5442,
      "step": 36891
    },
    {
      "epoch": 2.0658528390637247,
      "grad_norm": 0.9647210836410522,
      "learning_rate": 9.555868421052631e-05,
      "loss": 0.3612,
      "step": 36892
    },
    {
      "epoch": 2.0659088363758538,
      "grad_norm": 1.132678508758545,
      "learning_rate": 9.555842105263159e-05,
      "loss": 0.459,
      "step": 36893
    },
    {
      "epoch": 2.065964833687983,
      "grad_norm": 1.348985195159912,
      "learning_rate": 9.555815789473685e-05,
      "loss": 0.3985,
      "step": 36894
    },
    {
      "epoch": 2.066020831000112,
      "grad_norm": 1.496087670326233,
      "learning_rate": 9.555789473684212e-05,
      "loss": 0.4345,
      "step": 36895
    },
    {
      "epoch": 2.066076828312241,
      "grad_norm": 1.0980764627456665,
      "learning_rate": 9.555763157894736e-05,
      "loss": 0.3042,
      "step": 36896
    },
    {
      "epoch": 2.06613282562437,
      "grad_norm": 1.1323881149291992,
      "learning_rate": 9.555736842105264e-05,
      "loss": 0.3571,
      "step": 36897
    },
    {
      "epoch": 2.066188822936499,
      "grad_norm": 1.0366615056991577,
      "learning_rate": 9.55571052631579e-05,
      "loss": 0.4156,
      "step": 36898
    },
    {
      "epoch": 2.066244820248628,
      "grad_norm": 1.3057920932769775,
      "learning_rate": 9.555684210526317e-05,
      "loss": 0.3661,
      "step": 36899
    },
    {
      "epoch": 2.066300817560757,
      "grad_norm": 1.364570140838623,
      "learning_rate": 9.555657894736843e-05,
      "loss": 0.4026,
      "step": 36900
    },
    {
      "epoch": 2.066356814872886,
      "grad_norm": 1.1887586116790771,
      "learning_rate": 9.555631578947369e-05,
      "loss": 0.4218,
      "step": 36901
    },
    {
      "epoch": 2.066412812185015,
      "grad_norm": 1.0178182125091553,
      "learning_rate": 9.555605263157895e-05,
      "loss": 0.4016,
      "step": 36902
    },
    {
      "epoch": 2.066468809497144,
      "grad_norm": 1.231207013130188,
      "learning_rate": 9.555578947368421e-05,
      "loss": 0.4729,
      "step": 36903
    },
    {
      "epoch": 2.066524806809273,
      "grad_norm": 1.0676156282424927,
      "learning_rate": 9.555552631578948e-05,
      "loss": 0.3729,
      "step": 36904
    },
    {
      "epoch": 2.066580804121402,
      "grad_norm": 1.0702403783798218,
      "learning_rate": 9.555526315789474e-05,
      "loss": 0.3773,
      "step": 36905
    },
    {
      "epoch": 2.066636801433531,
      "grad_norm": 1.214680790901184,
      "learning_rate": 9.5555e-05,
      "loss": 0.3497,
      "step": 36906
    },
    {
      "epoch": 2.06669279874566,
      "grad_norm": 1.245222568511963,
      "learning_rate": 9.555473684210526e-05,
      "loss": 0.4053,
      "step": 36907
    },
    {
      "epoch": 2.066748796057789,
      "grad_norm": 0.9513360261917114,
      "learning_rate": 9.555447368421054e-05,
      "loss": 0.2561,
      "step": 36908
    },
    {
      "epoch": 2.066804793369918,
      "grad_norm": 1.1846604347229004,
      "learning_rate": 9.55542105263158e-05,
      "loss": 0.3632,
      "step": 36909
    },
    {
      "epoch": 2.066860790682047,
      "grad_norm": 1.0984668731689453,
      "learning_rate": 9.555394736842105e-05,
      "loss": 0.5049,
      "step": 36910
    },
    {
      "epoch": 2.066916787994176,
      "grad_norm": 1.2024413347244263,
      "learning_rate": 9.555368421052631e-05,
      "loss": 0.4477,
      "step": 36911
    },
    {
      "epoch": 2.066972785306305,
      "grad_norm": 1.6307432651519775,
      "learning_rate": 9.555342105263159e-05,
      "loss": 0.5437,
      "step": 36912
    },
    {
      "epoch": 2.067028782618434,
      "grad_norm": 1.186331868171692,
      "learning_rate": 9.555315789473685e-05,
      "loss": 0.3038,
      "step": 36913
    },
    {
      "epoch": 2.067084779930563,
      "grad_norm": 1.291262149810791,
      "learning_rate": 9.555289473684211e-05,
      "loss": 0.3901,
      "step": 36914
    },
    {
      "epoch": 2.067140777242692,
      "grad_norm": 1.1227818727493286,
      "learning_rate": 9.555263157894737e-05,
      "loss": 0.3818,
      "step": 36915
    },
    {
      "epoch": 2.067196774554821,
      "grad_norm": 1.1458094120025635,
      "learning_rate": 9.555236842105264e-05,
      "loss": 0.5112,
      "step": 36916
    },
    {
      "epoch": 2.0672527718669502,
      "grad_norm": 1.233750343322754,
      "learning_rate": 9.55521052631579e-05,
      "loss": 0.347,
      "step": 36917
    },
    {
      "epoch": 2.0673087691790792,
      "grad_norm": 1.522252082824707,
      "learning_rate": 9.555184210526317e-05,
      "loss": 0.459,
      "step": 36918
    },
    {
      "epoch": 2.0673647664912083,
      "grad_norm": 1.088221549987793,
      "learning_rate": 9.555157894736842e-05,
      "loss": 0.3287,
      "step": 36919
    },
    {
      "epoch": 2.0674207638033373,
      "grad_norm": 1.0218133926391602,
      "learning_rate": 9.555131578947369e-05,
      "loss": 0.3374,
      "step": 36920
    },
    {
      "epoch": 2.0674767611154663,
      "grad_norm": 1.201037049293518,
      "learning_rate": 9.555105263157895e-05,
      "loss": 0.4101,
      "step": 36921
    },
    {
      "epoch": 2.0675327584275953,
      "grad_norm": 1.1600815057754517,
      "learning_rate": 9.555078947368421e-05,
      "loss": 0.365,
      "step": 36922
    },
    {
      "epoch": 2.0675887557397243,
      "grad_norm": 1.1735756397247314,
      "learning_rate": 9.555052631578949e-05,
      "loss": 0.4629,
      "step": 36923
    },
    {
      "epoch": 2.0676447530518534,
      "grad_norm": 1.163258671760559,
      "learning_rate": 9.555026315789473e-05,
      "loss": 0.3683,
      "step": 36924
    },
    {
      "epoch": 2.0677007503639824,
      "grad_norm": 1.4746806621551514,
      "learning_rate": 9.555e-05,
      "loss": 0.4052,
      "step": 36925
    },
    {
      "epoch": 2.0677567476761114,
      "grad_norm": 1.3272709846496582,
      "learning_rate": 9.554973684210526e-05,
      "loss": 0.4694,
      "step": 36926
    },
    {
      "epoch": 2.0678127449882404,
      "grad_norm": 1.1801903247833252,
      "learning_rate": 9.554947368421054e-05,
      "loss": 0.3871,
      "step": 36927
    },
    {
      "epoch": 2.0678687423003694,
      "grad_norm": 1.1899926662445068,
      "learning_rate": 9.55492105263158e-05,
      "loss": 0.4908,
      "step": 36928
    },
    {
      "epoch": 2.0679247396124985,
      "grad_norm": 1.074407935142517,
      "learning_rate": 9.554894736842106e-05,
      "loss": 0.3928,
      "step": 36929
    },
    {
      "epoch": 2.0679807369246275,
      "grad_norm": 1.132913589477539,
      "learning_rate": 9.554868421052632e-05,
      "loss": 0.5435,
      "step": 36930
    },
    {
      "epoch": 2.0680367342367565,
      "grad_norm": 1.0559227466583252,
      "learning_rate": 9.554842105263159e-05,
      "loss": 0.3458,
      "step": 36931
    },
    {
      "epoch": 2.0680927315488855,
      "grad_norm": 1.2586069107055664,
      "learning_rate": 9.554815789473685e-05,
      "loss": 0.3804,
      "step": 36932
    },
    {
      "epoch": 2.0681487288610145,
      "grad_norm": 1.3245611190795898,
      "learning_rate": 9.554789473684211e-05,
      "loss": 0.5137,
      "step": 36933
    },
    {
      "epoch": 2.0682047261731435,
      "grad_norm": 1.4646902084350586,
      "learning_rate": 9.554763157894737e-05,
      "loss": 0.4418,
      "step": 36934
    },
    {
      "epoch": 2.0682607234852726,
      "grad_norm": 1.1135940551757812,
      "learning_rate": 9.554736842105264e-05,
      "loss": 0.3913,
      "step": 36935
    },
    {
      "epoch": 2.0683167207974016,
      "grad_norm": 1.0342915058135986,
      "learning_rate": 9.55471052631579e-05,
      "loss": 0.3362,
      "step": 36936
    },
    {
      "epoch": 2.0683727181095306,
      "grad_norm": 1.3088523149490356,
      "learning_rate": 9.554684210526316e-05,
      "loss": 0.4989,
      "step": 36937
    },
    {
      "epoch": 2.0684287154216596,
      "grad_norm": 1.4924994707107544,
      "learning_rate": 9.554657894736842e-05,
      "loss": 0.4135,
      "step": 36938
    },
    {
      "epoch": 2.0684847127337886,
      "grad_norm": 1.2817860841751099,
      "learning_rate": 9.554631578947368e-05,
      "loss": 0.467,
      "step": 36939
    },
    {
      "epoch": 2.0685407100459177,
      "grad_norm": 1.0914523601531982,
      "learning_rate": 9.554605263157896e-05,
      "loss": 0.3482,
      "step": 36940
    },
    {
      "epoch": 2.0685967073580467,
      "grad_norm": 1.3932172060012817,
      "learning_rate": 9.554578947368421e-05,
      "loss": 0.3755,
      "step": 36941
    },
    {
      "epoch": 2.0686527046701757,
      "grad_norm": 1.1626859903335571,
      "learning_rate": 9.554552631578947e-05,
      "loss": 0.356,
      "step": 36942
    },
    {
      "epoch": 2.0687087019823047,
      "grad_norm": 1.333052158355713,
      "learning_rate": 9.554526315789473e-05,
      "loss": 0.4533,
      "step": 36943
    },
    {
      "epoch": 2.0687646992944337,
      "grad_norm": 1.1898797750473022,
      "learning_rate": 9.554500000000001e-05,
      "loss": 0.3425,
      "step": 36944
    },
    {
      "epoch": 2.0688206966065628,
      "grad_norm": 1.055832862854004,
      "learning_rate": 9.554473684210527e-05,
      "loss": 0.3349,
      "step": 36945
    },
    {
      "epoch": 2.068876693918692,
      "grad_norm": 1.3232004642486572,
      "learning_rate": 9.554447368421053e-05,
      "loss": 0.3759,
      "step": 36946
    },
    {
      "epoch": 2.068932691230821,
      "grad_norm": 1.226897120475769,
      "learning_rate": 9.554421052631579e-05,
      "loss": 0.3898,
      "step": 36947
    },
    {
      "epoch": 2.06898868854295,
      "grad_norm": 1.1153355836868286,
      "learning_rate": 9.554394736842106e-05,
      "loss": 0.3346,
      "step": 36948
    },
    {
      "epoch": 2.069044685855079,
      "grad_norm": 1.2123972177505493,
      "learning_rate": 9.554368421052632e-05,
      "loss": 0.4014,
      "step": 36949
    },
    {
      "epoch": 2.069100683167208,
      "grad_norm": 1.141378402709961,
      "learning_rate": 9.554342105263159e-05,
      "loss": 0.4785,
      "step": 36950
    },
    {
      "epoch": 2.069156680479337,
      "grad_norm": 1.3106846809387207,
      "learning_rate": 9.554315789473684e-05,
      "loss": 0.3313,
      "step": 36951
    },
    {
      "epoch": 2.069212677791466,
      "grad_norm": 1.13906991481781,
      "learning_rate": 9.554289473684211e-05,
      "loss": 0.3155,
      "step": 36952
    },
    {
      "epoch": 2.069268675103595,
      "grad_norm": 1.2419027090072632,
      "learning_rate": 9.554263157894737e-05,
      "loss": 0.3622,
      "step": 36953
    },
    {
      "epoch": 2.069324672415724,
      "grad_norm": 1.2661066055297852,
      "learning_rate": 9.554236842105265e-05,
      "loss": 0.4009,
      "step": 36954
    },
    {
      "epoch": 2.069380669727853,
      "grad_norm": 1.1473249197006226,
      "learning_rate": 9.55421052631579e-05,
      "loss": 0.3986,
      "step": 36955
    },
    {
      "epoch": 2.069436667039982,
      "grad_norm": 1.9443079233169556,
      "learning_rate": 9.554184210526315e-05,
      "loss": 0.3873,
      "step": 36956
    },
    {
      "epoch": 2.069492664352111,
      "grad_norm": 1.073199987411499,
      "learning_rate": 9.554157894736842e-05,
      "loss": 0.4132,
      "step": 36957
    },
    {
      "epoch": 2.06954866166424,
      "grad_norm": 1.2984119653701782,
      "learning_rate": 9.554131578947368e-05,
      "loss": 0.362,
      "step": 36958
    },
    {
      "epoch": 2.069604658976369,
      "grad_norm": 1.2506521940231323,
      "learning_rate": 9.554105263157896e-05,
      "loss": 0.6003,
      "step": 36959
    },
    {
      "epoch": 2.069660656288498,
      "grad_norm": 1.4506969451904297,
      "learning_rate": 9.554078947368422e-05,
      "loss": 0.3888,
      "step": 36960
    },
    {
      "epoch": 2.069716653600627,
      "grad_norm": 1.0450642108917236,
      "learning_rate": 9.554052631578948e-05,
      "loss": 0.3111,
      "step": 36961
    },
    {
      "epoch": 2.069772650912756,
      "grad_norm": 1.128321886062622,
      "learning_rate": 9.554026315789474e-05,
      "loss": 0.4417,
      "step": 36962
    },
    {
      "epoch": 2.069828648224885,
      "grad_norm": 1.2313975095748901,
      "learning_rate": 9.554000000000001e-05,
      "loss": 0.4691,
      "step": 36963
    },
    {
      "epoch": 2.069884645537014,
      "grad_norm": 0.8868594169616699,
      "learning_rate": 9.553973684210527e-05,
      "loss": 0.2942,
      "step": 36964
    },
    {
      "epoch": 2.069940642849143,
      "grad_norm": 1.2188931703567505,
      "learning_rate": 9.553947368421053e-05,
      "loss": 0.4984,
      "step": 36965
    },
    {
      "epoch": 2.069996640161272,
      "grad_norm": 1.3257720470428467,
      "learning_rate": 9.553921052631579e-05,
      "loss": 0.442,
      "step": 36966
    },
    {
      "epoch": 2.070052637473401,
      "grad_norm": 0.9913079738616943,
      "learning_rate": 9.553894736842106e-05,
      "loss": 0.3307,
      "step": 36967
    },
    {
      "epoch": 2.07010863478553,
      "grad_norm": 1.6526544094085693,
      "learning_rate": 9.553868421052632e-05,
      "loss": 0.3479,
      "step": 36968
    },
    {
      "epoch": 2.070164632097659,
      "grad_norm": 1.077796220779419,
      "learning_rate": 9.553842105263158e-05,
      "loss": 0.4324,
      "step": 36969
    },
    {
      "epoch": 2.0702206294097882,
      "grad_norm": 1.4535760879516602,
      "learning_rate": 9.553815789473684e-05,
      "loss": 0.4425,
      "step": 36970
    },
    {
      "epoch": 2.0702766267219173,
      "grad_norm": 1.2667109966278076,
      "learning_rate": 9.553789473684212e-05,
      "loss": 0.4282,
      "step": 36971
    },
    {
      "epoch": 2.0703326240340463,
      "grad_norm": 1.4694347381591797,
      "learning_rate": 9.553763157894737e-05,
      "loss": 0.3872,
      "step": 36972
    },
    {
      "epoch": 2.0703886213461753,
      "grad_norm": 1.3678876161575317,
      "learning_rate": 9.553736842105263e-05,
      "loss": 0.395,
      "step": 36973
    },
    {
      "epoch": 2.0704446186583043,
      "grad_norm": 1.1957201957702637,
      "learning_rate": 9.55371052631579e-05,
      "loss": 0.4489,
      "step": 36974
    },
    {
      "epoch": 2.0705006159704333,
      "grad_norm": 1.1299457550048828,
      "learning_rate": 9.553684210526315e-05,
      "loss": 0.3781,
      "step": 36975
    },
    {
      "epoch": 2.0705566132825624,
      "grad_norm": 1.2196074724197388,
      "learning_rate": 9.553657894736843e-05,
      "loss": 0.4459,
      "step": 36976
    },
    {
      "epoch": 2.0706126105946914,
      "grad_norm": 1.035150408744812,
      "learning_rate": 9.553631578947369e-05,
      "loss": 0.4042,
      "step": 36977
    },
    {
      "epoch": 2.0706686079068204,
      "grad_norm": 1.177608847618103,
      "learning_rate": 9.553605263157896e-05,
      "loss": 0.3662,
      "step": 36978
    },
    {
      "epoch": 2.0707246052189494,
      "grad_norm": 1.466491460800171,
      "learning_rate": 9.55357894736842e-05,
      "loss": 0.4265,
      "step": 36979
    },
    {
      "epoch": 2.0707806025310784,
      "grad_norm": 1.8743799924850464,
      "learning_rate": 9.553552631578948e-05,
      "loss": 0.4758,
      "step": 36980
    },
    {
      "epoch": 2.0708365998432074,
      "grad_norm": 1.1449819803237915,
      "learning_rate": 9.553526315789474e-05,
      "loss": 0.3368,
      "step": 36981
    },
    {
      "epoch": 2.0708925971553365,
      "grad_norm": 1.0853910446166992,
      "learning_rate": 9.553500000000001e-05,
      "loss": 0.3319,
      "step": 36982
    },
    {
      "epoch": 2.0709485944674655,
      "grad_norm": 0.9453714489936829,
      "learning_rate": 9.553473684210526e-05,
      "loss": 0.308,
      "step": 36983
    },
    {
      "epoch": 2.0710045917795945,
      "grad_norm": 1.0927577018737793,
      "learning_rate": 9.553447368421053e-05,
      "loss": 0.4955,
      "step": 36984
    },
    {
      "epoch": 2.0710605890917235,
      "grad_norm": 1.1222269535064697,
      "learning_rate": 9.553421052631579e-05,
      "loss": 0.3911,
      "step": 36985
    },
    {
      "epoch": 2.0711165864038525,
      "grad_norm": 1.1565266847610474,
      "learning_rate": 9.553394736842107e-05,
      "loss": 0.4005,
      "step": 36986
    },
    {
      "epoch": 2.0711725837159816,
      "grad_norm": 1.2788268327713013,
      "learning_rate": 9.553368421052633e-05,
      "loss": 0.3073,
      "step": 36987
    },
    {
      "epoch": 2.0712285810281106,
      "grad_norm": 1.2404124736785889,
      "learning_rate": 9.553342105263158e-05,
      "loss": 0.3663,
      "step": 36988
    },
    {
      "epoch": 2.0712845783402396,
      "grad_norm": 1.246645212173462,
      "learning_rate": 9.553315789473684e-05,
      "loss": 0.4515,
      "step": 36989
    },
    {
      "epoch": 2.0713405756523686,
      "grad_norm": 1.140640139579773,
      "learning_rate": 9.55328947368421e-05,
      "loss": 0.3704,
      "step": 36990
    },
    {
      "epoch": 2.0713965729644976,
      "grad_norm": 1.2302244901657104,
      "learning_rate": 9.553263157894738e-05,
      "loss": 0.451,
      "step": 36991
    },
    {
      "epoch": 2.0714525702766267,
      "grad_norm": 1.0759280920028687,
      "learning_rate": 9.553236842105264e-05,
      "loss": 0.4187,
      "step": 36992
    },
    {
      "epoch": 2.0715085675887557,
      "grad_norm": 0.9919374585151672,
      "learning_rate": 9.55321052631579e-05,
      "loss": 0.3183,
      "step": 36993
    },
    {
      "epoch": 2.0715645649008847,
      "grad_norm": 1.2650642395019531,
      "learning_rate": 9.553184210526316e-05,
      "loss": 0.4595,
      "step": 36994
    },
    {
      "epoch": 2.0716205622130137,
      "grad_norm": 1.0574581623077393,
      "learning_rate": 9.553157894736843e-05,
      "loss": 0.3302,
      "step": 36995
    },
    {
      "epoch": 2.0716765595251427,
      "grad_norm": 1.3501040935516357,
      "learning_rate": 9.553131578947369e-05,
      "loss": 0.522,
      "step": 36996
    },
    {
      "epoch": 2.0717325568372718,
      "grad_norm": 1.5992681980133057,
      "learning_rate": 9.553105263157895e-05,
      "loss": 0.3718,
      "step": 36997
    },
    {
      "epoch": 2.0717885541494008,
      "grad_norm": 1.3846632242202759,
      "learning_rate": 9.553078947368421e-05,
      "loss": 0.4699,
      "step": 36998
    },
    {
      "epoch": 2.07184455146153,
      "grad_norm": 1.1945549249649048,
      "learning_rate": 9.553052631578948e-05,
      "loss": 0.3624,
      "step": 36999
    },
    {
      "epoch": 2.071900548773659,
      "grad_norm": 1.1957957744598389,
      "learning_rate": 9.553026315789474e-05,
      "loss": 0.3635,
      "step": 37000
    },
    {
      "epoch": 2.071956546085788,
      "grad_norm": 1.0470725297927856,
      "learning_rate": 9.553e-05,
      "loss": 0.3627,
      "step": 37001
    },
    {
      "epoch": 2.072012543397917,
      "grad_norm": 1.5645742416381836,
      "learning_rate": 9.552973684210526e-05,
      "loss": 0.5567,
      "step": 37002
    },
    {
      "epoch": 2.072068540710046,
      "grad_norm": 1.2169232368469238,
      "learning_rate": 9.552947368421053e-05,
      "loss": 0.5565,
      "step": 37003
    },
    {
      "epoch": 2.072124538022175,
      "grad_norm": 1.127480387687683,
      "learning_rate": 9.55292105263158e-05,
      "loss": 0.3291,
      "step": 37004
    },
    {
      "epoch": 2.072180535334304,
      "grad_norm": 1.265182375907898,
      "learning_rate": 9.552894736842107e-05,
      "loss": 0.4051,
      "step": 37005
    },
    {
      "epoch": 2.072236532646433,
      "grad_norm": 1.5591259002685547,
      "learning_rate": 9.552868421052631e-05,
      "loss": 0.3732,
      "step": 37006
    },
    {
      "epoch": 2.072292529958562,
      "grad_norm": 1.1799579858779907,
      "learning_rate": 9.552842105263157e-05,
      "loss": 0.4394,
      "step": 37007
    },
    {
      "epoch": 2.072348527270691,
      "grad_norm": 1.1281867027282715,
      "learning_rate": 9.552815789473685e-05,
      "loss": 0.3402,
      "step": 37008
    },
    {
      "epoch": 2.07240452458282,
      "grad_norm": 1.11223304271698,
      "learning_rate": 9.552789473684211e-05,
      "loss": 0.3626,
      "step": 37009
    },
    {
      "epoch": 2.072460521894949,
      "grad_norm": 1.16005277633667,
      "learning_rate": 9.552763157894738e-05,
      "loss": 0.3698,
      "step": 37010
    },
    {
      "epoch": 2.072516519207078,
      "grad_norm": 1.0963557958602905,
      "learning_rate": 9.552736842105263e-05,
      "loss": 0.333,
      "step": 37011
    },
    {
      "epoch": 2.072572516519207,
      "grad_norm": 1.276484489440918,
      "learning_rate": 9.55271052631579e-05,
      "loss": 0.4115,
      "step": 37012
    },
    {
      "epoch": 2.072628513831336,
      "grad_norm": 1.0017668008804321,
      "learning_rate": 9.552684210526316e-05,
      "loss": 0.2998,
      "step": 37013
    },
    {
      "epoch": 2.072684511143465,
      "grad_norm": 1.1222561597824097,
      "learning_rate": 9.552657894736843e-05,
      "loss": 0.4406,
      "step": 37014
    },
    {
      "epoch": 2.072740508455594,
      "grad_norm": 1.30161452293396,
      "learning_rate": 9.552631578947369e-05,
      "loss": 0.3866,
      "step": 37015
    },
    {
      "epoch": 2.072796505767723,
      "grad_norm": 1.3920040130615234,
      "learning_rate": 9.552605263157895e-05,
      "loss": 0.4152,
      "step": 37016
    },
    {
      "epoch": 2.072852503079852,
      "grad_norm": 1.2607569694519043,
      "learning_rate": 9.552578947368421e-05,
      "loss": 0.3916,
      "step": 37017
    },
    {
      "epoch": 2.072908500391981,
      "grad_norm": 1.0879734754562378,
      "learning_rate": 9.552552631578948e-05,
      "loss": 0.4235,
      "step": 37018
    },
    {
      "epoch": 2.07296449770411,
      "grad_norm": 1.5044060945510864,
      "learning_rate": 9.552526315789474e-05,
      "loss": 0.3682,
      "step": 37019
    },
    {
      "epoch": 2.073020495016239,
      "grad_norm": 1.2094651460647583,
      "learning_rate": 9.5525e-05,
      "loss": 0.3546,
      "step": 37020
    },
    {
      "epoch": 2.073076492328368,
      "grad_norm": 1.2888457775115967,
      "learning_rate": 9.552473684210526e-05,
      "loss": 0.3719,
      "step": 37021
    },
    {
      "epoch": 2.0731324896404972,
      "grad_norm": 1.185799479484558,
      "learning_rate": 9.552447368421054e-05,
      "loss": 0.6407,
      "step": 37022
    },
    {
      "epoch": 2.0731884869526263,
      "grad_norm": 1.4544696807861328,
      "learning_rate": 9.55242105263158e-05,
      "loss": 0.5527,
      "step": 37023
    },
    {
      "epoch": 2.0732444842647553,
      "grad_norm": 1.117397665977478,
      "learning_rate": 9.552394736842106e-05,
      "loss": 0.3854,
      "step": 37024
    },
    {
      "epoch": 2.0733004815768843,
      "grad_norm": 1.1623637676239014,
      "learning_rate": 9.552368421052632e-05,
      "loss": 0.368,
      "step": 37025
    },
    {
      "epoch": 2.0733564788890133,
      "grad_norm": 1.2776652574539185,
      "learning_rate": 9.552342105263158e-05,
      "loss": 0.3656,
      "step": 37026
    },
    {
      "epoch": 2.0734124762011423,
      "grad_norm": 1.075451135635376,
      "learning_rate": 9.552315789473685e-05,
      "loss": 0.4179,
      "step": 37027
    },
    {
      "epoch": 2.0734684735132713,
      "grad_norm": 1.8438770771026611,
      "learning_rate": 9.552289473684211e-05,
      "loss": 0.442,
      "step": 37028
    },
    {
      "epoch": 2.0735244708254004,
      "grad_norm": 1.3850990533828735,
      "learning_rate": 9.552263157894737e-05,
      "loss": 0.3629,
      "step": 37029
    },
    {
      "epoch": 2.0735804681375294,
      "grad_norm": 2.116248369216919,
      "learning_rate": 9.552236842105263e-05,
      "loss": 0.3262,
      "step": 37030
    },
    {
      "epoch": 2.0736364654496584,
      "grad_norm": 1.1774441003799438,
      "learning_rate": 9.55221052631579e-05,
      "loss": 0.3812,
      "step": 37031
    },
    {
      "epoch": 2.0736924627617874,
      "grad_norm": 1.1282743215560913,
      "learning_rate": 9.552184210526316e-05,
      "loss": 0.449,
      "step": 37032
    },
    {
      "epoch": 2.0737484600739164,
      "grad_norm": 1.1806365251541138,
      "learning_rate": 9.552157894736844e-05,
      "loss": 0.4776,
      "step": 37033
    },
    {
      "epoch": 2.0738044573860455,
      "grad_norm": 1.3845775127410889,
      "learning_rate": 9.552131578947368e-05,
      "loss": 0.39,
      "step": 37034
    },
    {
      "epoch": 2.0738604546981745,
      "grad_norm": 1.1486878395080566,
      "learning_rate": 9.552105263157895e-05,
      "loss": 0.372,
      "step": 37035
    },
    {
      "epoch": 2.0739164520103035,
      "grad_norm": 1.1615314483642578,
      "learning_rate": 9.552078947368421e-05,
      "loss": 0.4529,
      "step": 37036
    },
    {
      "epoch": 2.0739724493224325,
      "grad_norm": 1.048534631729126,
      "learning_rate": 9.552052631578949e-05,
      "loss": 0.3942,
      "step": 37037
    },
    {
      "epoch": 2.0740284466345615,
      "grad_norm": 1.3325378894805908,
      "learning_rate": 9.552026315789473e-05,
      "loss": 0.4338,
      "step": 37038
    },
    {
      "epoch": 2.0740844439466906,
      "grad_norm": 1.5767453908920288,
      "learning_rate": 9.552000000000001e-05,
      "loss": 0.4125,
      "step": 37039
    },
    {
      "epoch": 2.0741404412588196,
      "grad_norm": 1.3049664497375488,
      "learning_rate": 9.551973684210527e-05,
      "loss": 0.4525,
      "step": 37040
    },
    {
      "epoch": 2.0741964385709486,
      "grad_norm": 1.1605595350265503,
      "learning_rate": 9.551947368421054e-05,
      "loss": 0.3554,
      "step": 37041
    },
    {
      "epoch": 2.0742524358830776,
      "grad_norm": 1.7675814628601074,
      "learning_rate": 9.55192105263158e-05,
      "loss": 0.4183,
      "step": 37042
    },
    {
      "epoch": 2.0743084331952066,
      "grad_norm": 2.703983783721924,
      "learning_rate": 9.551894736842105e-05,
      "loss": 0.4311,
      "step": 37043
    },
    {
      "epoch": 2.0743644305073357,
      "grad_norm": 1.2000855207443237,
      "learning_rate": 9.551868421052632e-05,
      "loss": 0.4255,
      "step": 37044
    },
    {
      "epoch": 2.0744204278194647,
      "grad_norm": 1.2760854959487915,
      "learning_rate": 9.551842105263158e-05,
      "loss": 0.5025,
      "step": 37045
    },
    {
      "epoch": 2.0744764251315937,
      "grad_norm": 1.0725868940353394,
      "learning_rate": 9.551815789473685e-05,
      "loss": 0.2992,
      "step": 37046
    },
    {
      "epoch": 2.0745324224437227,
      "grad_norm": 1.1692692041397095,
      "learning_rate": 9.551789473684211e-05,
      "loss": 0.5226,
      "step": 37047
    },
    {
      "epoch": 2.0745884197558517,
      "grad_norm": 1.4657577276229858,
      "learning_rate": 9.551763157894737e-05,
      "loss": 0.3711,
      "step": 37048
    },
    {
      "epoch": 2.0746444170679808,
      "grad_norm": 2.478576421737671,
      "learning_rate": 9.551736842105263e-05,
      "loss": 0.3464,
      "step": 37049
    },
    {
      "epoch": 2.0747004143801098,
      "grad_norm": 1.212652325630188,
      "learning_rate": 9.55171052631579e-05,
      "loss": 0.3771,
      "step": 37050
    },
    {
      "epoch": 2.074756411692239,
      "grad_norm": 2.1155765056610107,
      "learning_rate": 9.551684210526316e-05,
      "loss": 0.4901,
      "step": 37051
    },
    {
      "epoch": 2.074812409004368,
      "grad_norm": 1.8258616924285889,
      "learning_rate": 9.551657894736842e-05,
      "loss": 0.7175,
      "step": 37052
    },
    {
      "epoch": 2.074868406316497,
      "grad_norm": 1.197744607925415,
      "learning_rate": 9.551631578947368e-05,
      "loss": 0.3696,
      "step": 37053
    },
    {
      "epoch": 2.074924403628626,
      "grad_norm": 1.0415812730789185,
      "learning_rate": 9.551605263157896e-05,
      "loss": 0.3591,
      "step": 37054
    },
    {
      "epoch": 2.074980400940755,
      "grad_norm": 1.1505001783370972,
      "learning_rate": 9.551578947368422e-05,
      "loss": 0.3208,
      "step": 37055
    },
    {
      "epoch": 2.075036398252884,
      "grad_norm": 1.2407912015914917,
      "learning_rate": 9.551552631578948e-05,
      "loss": 0.4863,
      "step": 37056
    },
    {
      "epoch": 2.075092395565013,
      "grad_norm": 1.2999475002288818,
      "learning_rate": 9.551526315789474e-05,
      "loss": 0.3376,
      "step": 37057
    },
    {
      "epoch": 2.075148392877142,
      "grad_norm": 1.7863974571228027,
      "learning_rate": 9.551500000000001e-05,
      "loss": 0.3528,
      "step": 37058
    },
    {
      "epoch": 2.075204390189271,
      "grad_norm": 1.103898525238037,
      "learning_rate": 9.551473684210527e-05,
      "loss": 0.4201,
      "step": 37059
    },
    {
      "epoch": 2.0752603875014,
      "grad_norm": 1.8107049465179443,
      "learning_rate": 9.551447368421053e-05,
      "loss": 0.3832,
      "step": 37060
    },
    {
      "epoch": 2.075316384813529,
      "grad_norm": 1.3128491640090942,
      "learning_rate": 9.551421052631579e-05,
      "loss": 0.4369,
      "step": 37061
    },
    {
      "epoch": 2.075372382125658,
      "grad_norm": 1.1766630411148071,
      "learning_rate": 9.551394736842105e-05,
      "loss": 0.3435,
      "step": 37062
    },
    {
      "epoch": 2.075428379437787,
      "grad_norm": 1.3892816305160522,
      "learning_rate": 9.551368421052632e-05,
      "loss": 0.4505,
      "step": 37063
    },
    {
      "epoch": 2.075484376749916,
      "grad_norm": 1.3339816331863403,
      "learning_rate": 9.551342105263158e-05,
      "loss": 0.3003,
      "step": 37064
    },
    {
      "epoch": 2.075540374062045,
      "grad_norm": 1.2571492195129395,
      "learning_rate": 9.551315789473685e-05,
      "loss": 0.3149,
      "step": 37065
    },
    {
      "epoch": 2.075596371374174,
      "grad_norm": 1.3235915899276733,
      "learning_rate": 9.55128947368421e-05,
      "loss": 0.437,
      "step": 37066
    },
    {
      "epoch": 2.075652368686303,
      "grad_norm": 1.3600482940673828,
      "learning_rate": 9.551263157894737e-05,
      "loss": 0.3727,
      "step": 37067
    },
    {
      "epoch": 2.075708365998432,
      "grad_norm": 1.1924700736999512,
      "learning_rate": 9.551236842105263e-05,
      "loss": 0.4372,
      "step": 37068
    },
    {
      "epoch": 2.075764363310561,
      "grad_norm": 1.0132958889007568,
      "learning_rate": 9.551210526315791e-05,
      "loss": 0.3187,
      "step": 37069
    },
    {
      "epoch": 2.07582036062269,
      "grad_norm": 1.1740894317626953,
      "learning_rate": 9.551184210526317e-05,
      "loss": 0.3479,
      "step": 37070
    },
    {
      "epoch": 2.075876357934819,
      "grad_norm": 1.0848509073257446,
      "learning_rate": 9.551157894736843e-05,
      "loss": 0.3972,
      "step": 37071
    },
    {
      "epoch": 2.075932355246948,
      "grad_norm": 1.0278829336166382,
      "learning_rate": 9.551131578947369e-05,
      "loss": 0.3416,
      "step": 37072
    },
    {
      "epoch": 2.075988352559077,
      "grad_norm": 1.0650432109832764,
      "learning_rate": 9.551105263157896e-05,
      "loss": 0.321,
      "step": 37073
    },
    {
      "epoch": 2.0760443498712062,
      "grad_norm": 1.2951693534851074,
      "learning_rate": 9.551078947368422e-05,
      "loss": 0.4047,
      "step": 37074
    },
    {
      "epoch": 2.0761003471833352,
      "grad_norm": 1.3263237476348877,
      "learning_rate": 9.551052631578948e-05,
      "loss": 0.5355,
      "step": 37075
    },
    {
      "epoch": 2.0761563444954643,
      "grad_norm": 1.0165356397628784,
      "learning_rate": 9.551026315789474e-05,
      "loss": 0.35,
      "step": 37076
    },
    {
      "epoch": 2.0762123418075933,
      "grad_norm": 1.2454731464385986,
      "learning_rate": 9.551e-05,
      "loss": 0.3496,
      "step": 37077
    },
    {
      "epoch": 2.0762683391197223,
      "grad_norm": 1.1699351072311401,
      "learning_rate": 9.550973684210527e-05,
      "loss": 0.4812,
      "step": 37078
    },
    {
      "epoch": 2.0763243364318513,
      "grad_norm": 1.3235853910446167,
      "learning_rate": 9.550947368421053e-05,
      "loss": 0.3165,
      "step": 37079
    },
    {
      "epoch": 2.0763803337439803,
      "grad_norm": 1.433682918548584,
      "learning_rate": 9.550921052631579e-05,
      "loss": 0.4355,
      "step": 37080
    },
    {
      "epoch": 2.0764363310561094,
      "grad_norm": 1.1682240962982178,
      "learning_rate": 9.550894736842105e-05,
      "loss": 0.4433,
      "step": 37081
    },
    {
      "epoch": 2.0764923283682384,
      "grad_norm": 1.4480180740356445,
      "learning_rate": 9.550868421052632e-05,
      "loss": 0.268,
      "step": 37082
    },
    {
      "epoch": 2.0765483256803674,
      "grad_norm": 1.1202014684677124,
      "learning_rate": 9.550842105263158e-05,
      "loss": 0.3283,
      "step": 37083
    },
    {
      "epoch": 2.0766043229924964,
      "grad_norm": 1.251604676246643,
      "learning_rate": 9.550815789473684e-05,
      "loss": 0.4132,
      "step": 37084
    },
    {
      "epoch": 2.0766603203046254,
      "grad_norm": 1.2931939363479614,
      "learning_rate": 9.55078947368421e-05,
      "loss": 0.3654,
      "step": 37085
    },
    {
      "epoch": 2.0767163176167545,
      "grad_norm": 1.807619571685791,
      "learning_rate": 9.550763157894738e-05,
      "loss": 0.6579,
      "step": 37086
    },
    {
      "epoch": 2.0767723149288835,
      "grad_norm": 0.9698479771614075,
      "learning_rate": 9.550736842105264e-05,
      "loss": 0.3679,
      "step": 37087
    },
    {
      "epoch": 2.0768283122410125,
      "grad_norm": 1.1213072538375854,
      "learning_rate": 9.550710526315791e-05,
      "loss": 0.4671,
      "step": 37088
    },
    {
      "epoch": 2.0768843095531415,
      "grad_norm": 1.1348891258239746,
      "learning_rate": 9.550684210526316e-05,
      "loss": 0.3589,
      "step": 37089
    },
    {
      "epoch": 2.0769403068652705,
      "grad_norm": 1.299492359161377,
      "learning_rate": 9.550657894736843e-05,
      "loss": 0.4255,
      "step": 37090
    },
    {
      "epoch": 2.0769963041773996,
      "grad_norm": 1.1695098876953125,
      "learning_rate": 9.550631578947369e-05,
      "loss": 0.4946,
      "step": 37091
    },
    {
      "epoch": 2.0770523014895286,
      "grad_norm": 1.1541613340377808,
      "learning_rate": 9.550605263157896e-05,
      "loss": 0.3691,
      "step": 37092
    },
    {
      "epoch": 2.0771082988016576,
      "grad_norm": 1.1366599798202515,
      "learning_rate": 9.550578947368421e-05,
      "loss": 0.3645,
      "step": 37093
    },
    {
      "epoch": 2.0771642961137866,
      "grad_norm": 1.00677490234375,
      "learning_rate": 9.550552631578947e-05,
      "loss": 0.3854,
      "step": 37094
    },
    {
      "epoch": 2.0772202934259156,
      "grad_norm": 1.0176050662994385,
      "learning_rate": 9.550526315789474e-05,
      "loss": 0.3488,
      "step": 37095
    },
    {
      "epoch": 2.0772762907380447,
      "grad_norm": 1.3522952795028687,
      "learning_rate": 9.5505e-05,
      "loss": 0.4183,
      "step": 37096
    },
    {
      "epoch": 2.0773322880501737,
      "grad_norm": 1.144392490386963,
      "learning_rate": 9.550473684210527e-05,
      "loss": 0.4108,
      "step": 37097
    },
    {
      "epoch": 2.0773882853623027,
      "grad_norm": 1.8258731365203857,
      "learning_rate": 9.550447368421052e-05,
      "loss": 0.4787,
      "step": 37098
    },
    {
      "epoch": 2.0774442826744317,
      "grad_norm": 1.1547350883483887,
      "learning_rate": 9.55042105263158e-05,
      "loss": 0.3333,
      "step": 37099
    },
    {
      "epoch": 2.0775002799865607,
      "grad_norm": 1.4513922929763794,
      "learning_rate": 9.550394736842105e-05,
      "loss": 0.5042,
      "step": 37100
    },
    {
      "epoch": 2.0775562772986897,
      "grad_norm": 1.0561192035675049,
      "learning_rate": 9.550368421052633e-05,
      "loss": 0.3526,
      "step": 37101
    },
    {
      "epoch": 2.0776122746108188,
      "grad_norm": 1.2201570272445679,
      "learning_rate": 9.550342105263159e-05,
      "loss": 0.5021,
      "step": 37102
    },
    {
      "epoch": 2.077668271922948,
      "grad_norm": 1.092822790145874,
      "learning_rate": 9.550315789473685e-05,
      "loss": 0.3379,
      "step": 37103
    },
    {
      "epoch": 2.077724269235077,
      "grad_norm": 1.3406414985656738,
      "learning_rate": 9.55028947368421e-05,
      "loss": 0.4214,
      "step": 37104
    },
    {
      "epoch": 2.077780266547206,
      "grad_norm": 1.097962498664856,
      "learning_rate": 9.550263157894738e-05,
      "loss": 0.4013,
      "step": 37105
    },
    {
      "epoch": 2.077836263859335,
      "grad_norm": 1.019471526145935,
      "learning_rate": 9.550236842105264e-05,
      "loss": 0.2978,
      "step": 37106
    },
    {
      "epoch": 2.077892261171464,
      "grad_norm": 1.233687400817871,
      "learning_rate": 9.55021052631579e-05,
      "loss": 0.4046,
      "step": 37107
    },
    {
      "epoch": 2.077948258483593,
      "grad_norm": 1.3127537965774536,
      "learning_rate": 9.550184210526316e-05,
      "loss": 0.3596,
      "step": 37108
    },
    {
      "epoch": 2.078004255795722,
      "grad_norm": 1.0794765949249268,
      "learning_rate": 9.550157894736843e-05,
      "loss": 0.3634,
      "step": 37109
    },
    {
      "epoch": 2.078060253107851,
      "grad_norm": 1.2233608961105347,
      "learning_rate": 9.550131578947369e-05,
      "loss": 0.4409,
      "step": 37110
    },
    {
      "epoch": 2.07811625041998,
      "grad_norm": 1.390821099281311,
      "learning_rate": 9.550105263157895e-05,
      "loss": 0.4429,
      "step": 37111
    },
    {
      "epoch": 2.078172247732109,
      "grad_norm": 1.3868762254714966,
      "learning_rate": 9.550078947368421e-05,
      "loss": 0.4688,
      "step": 37112
    },
    {
      "epoch": 2.078228245044238,
      "grad_norm": 1.585021734237671,
      "learning_rate": 9.550052631578947e-05,
      "loss": 0.4357,
      "step": 37113
    },
    {
      "epoch": 2.078284242356367,
      "grad_norm": 1.3313313722610474,
      "learning_rate": 9.550026315789474e-05,
      "loss": 0.4094,
      "step": 37114
    },
    {
      "epoch": 2.078340239668496,
      "grad_norm": 1.19365394115448,
      "learning_rate": 9.55e-05,
      "loss": 0.4417,
      "step": 37115
    },
    {
      "epoch": 2.078396236980625,
      "grad_norm": 1.2019215822219849,
      "learning_rate": 9.549973684210526e-05,
      "loss": 0.3783,
      "step": 37116
    },
    {
      "epoch": 2.078452234292754,
      "grad_norm": 1.3380273580551147,
      "learning_rate": 9.549947368421052e-05,
      "loss": 0.4253,
      "step": 37117
    },
    {
      "epoch": 2.078508231604883,
      "grad_norm": 1.0407902002334595,
      "learning_rate": 9.54992105263158e-05,
      "loss": 0.3846,
      "step": 37118
    },
    {
      "epoch": 2.078564228917012,
      "grad_norm": 1.1597486734390259,
      "learning_rate": 9.549894736842106e-05,
      "loss": 0.3388,
      "step": 37119
    },
    {
      "epoch": 2.078620226229141,
      "grad_norm": 1.4331289529800415,
      "learning_rate": 9.549868421052633e-05,
      "loss": 0.3774,
      "step": 37120
    },
    {
      "epoch": 2.07867622354127,
      "grad_norm": 1.128772497177124,
      "learning_rate": 9.549842105263158e-05,
      "loss": 0.4266,
      "step": 37121
    },
    {
      "epoch": 2.078732220853399,
      "grad_norm": 2.545440435409546,
      "learning_rate": 9.549815789473685e-05,
      "loss": 0.4797,
      "step": 37122
    },
    {
      "epoch": 2.078788218165528,
      "grad_norm": 1.110619306564331,
      "learning_rate": 9.549789473684211e-05,
      "loss": 0.3519,
      "step": 37123
    },
    {
      "epoch": 2.078844215477657,
      "grad_norm": 1.1616355180740356,
      "learning_rate": 9.549763157894738e-05,
      "loss": 0.3368,
      "step": 37124
    },
    {
      "epoch": 2.078900212789786,
      "grad_norm": 1.4003827571868896,
      "learning_rate": 9.549736842105264e-05,
      "loss": 0.5483,
      "step": 37125
    },
    {
      "epoch": 2.0789562101019152,
      "grad_norm": 1.0419765710830688,
      "learning_rate": 9.54971052631579e-05,
      "loss": 0.3862,
      "step": 37126
    },
    {
      "epoch": 2.0790122074140442,
      "grad_norm": 1.1480308771133423,
      "learning_rate": 9.549684210526316e-05,
      "loss": 0.393,
      "step": 37127
    },
    {
      "epoch": 2.0790682047261733,
      "grad_norm": 1.9641711711883545,
      "learning_rate": 9.549657894736842e-05,
      "loss": 0.5263,
      "step": 37128
    },
    {
      "epoch": 2.0791242020383023,
      "grad_norm": 1.1566872596740723,
      "learning_rate": 9.54963157894737e-05,
      "loss": 0.3837,
      "step": 37129
    },
    {
      "epoch": 2.0791801993504313,
      "grad_norm": 1.0126267671585083,
      "learning_rate": 9.549605263157895e-05,
      "loss": 0.3531,
      "step": 37130
    },
    {
      "epoch": 2.0792361966625603,
      "grad_norm": 1.4017198085784912,
      "learning_rate": 9.549578947368421e-05,
      "loss": 0.4071,
      "step": 37131
    },
    {
      "epoch": 2.0792921939746893,
      "grad_norm": 1.0598618984222412,
      "learning_rate": 9.549552631578947e-05,
      "loss": 0.3409,
      "step": 37132
    },
    {
      "epoch": 2.0793481912868184,
      "grad_norm": 1.2616815567016602,
      "learning_rate": 9.549526315789475e-05,
      "loss": 0.3222,
      "step": 37133
    },
    {
      "epoch": 2.0794041885989474,
      "grad_norm": 1.5376297235488892,
      "learning_rate": 9.5495e-05,
      "loss": 0.3813,
      "step": 37134
    },
    {
      "epoch": 2.0794601859110764,
      "grad_norm": 1.044512152671814,
      "learning_rate": 9.549473684210527e-05,
      "loss": 0.4686,
      "step": 37135
    },
    {
      "epoch": 2.0795161832232054,
      "grad_norm": 1.1966419219970703,
      "learning_rate": 9.549447368421053e-05,
      "loss": 0.3618,
      "step": 37136
    },
    {
      "epoch": 2.0795721805353344,
      "grad_norm": 1.2195829153060913,
      "learning_rate": 9.54942105263158e-05,
      "loss": 0.361,
      "step": 37137
    },
    {
      "epoch": 2.0796281778474635,
      "grad_norm": 1.0981051921844482,
      "learning_rate": 9.549394736842106e-05,
      "loss": 0.3469,
      "step": 37138
    },
    {
      "epoch": 2.0796841751595925,
      "grad_norm": 1.322159767150879,
      "learning_rate": 9.549368421052632e-05,
      "loss": 0.4314,
      "step": 37139
    },
    {
      "epoch": 2.0797401724717215,
      "grad_norm": 1.1160557270050049,
      "learning_rate": 9.549342105263158e-05,
      "loss": 0.3706,
      "step": 37140
    },
    {
      "epoch": 2.0797961697838505,
      "grad_norm": 0.995877742767334,
      "learning_rate": 9.549315789473685e-05,
      "loss": 0.355,
      "step": 37141
    },
    {
      "epoch": 2.0798521670959795,
      "grad_norm": 1.0276455879211426,
      "learning_rate": 9.549289473684211e-05,
      "loss": 0.408,
      "step": 37142
    },
    {
      "epoch": 2.0799081644081086,
      "grad_norm": 1.3109102249145508,
      "learning_rate": 9.549263157894738e-05,
      "loss": 0.5016,
      "step": 37143
    },
    {
      "epoch": 2.0799641617202376,
      "grad_norm": 1.0983550548553467,
      "learning_rate": 9.549236842105263e-05,
      "loss": 0.4917,
      "step": 37144
    },
    {
      "epoch": 2.0800201590323666,
      "grad_norm": 1.2872225046157837,
      "learning_rate": 9.549210526315789e-05,
      "loss": 0.3916,
      "step": 37145
    },
    {
      "epoch": 2.0800761563444956,
      "grad_norm": 1.0223556756973267,
      "learning_rate": 9.549184210526316e-05,
      "loss": 0.3483,
      "step": 37146
    },
    {
      "epoch": 2.0801321536566246,
      "grad_norm": 1.2347514629364014,
      "learning_rate": 9.549157894736842e-05,
      "loss": 0.3918,
      "step": 37147
    },
    {
      "epoch": 2.0801881509687536,
      "grad_norm": 1.2691410779953003,
      "learning_rate": 9.549131578947368e-05,
      "loss": 0.4163,
      "step": 37148
    },
    {
      "epoch": 2.0802441482808827,
      "grad_norm": 1.1786844730377197,
      "learning_rate": 9.549105263157894e-05,
      "loss": 0.3879,
      "step": 37149
    },
    {
      "epoch": 2.0803001455930117,
      "grad_norm": 1.1877529621124268,
      "learning_rate": 9.549078947368422e-05,
      "loss": 0.4794,
      "step": 37150
    },
    {
      "epoch": 2.0803561429051407,
      "grad_norm": 1.334516167640686,
      "learning_rate": 9.549052631578948e-05,
      "loss": 0.4243,
      "step": 37151
    },
    {
      "epoch": 2.0804121402172697,
      "grad_norm": 1.169430136680603,
      "learning_rate": 9.549026315789475e-05,
      "loss": 0.3405,
      "step": 37152
    },
    {
      "epoch": 2.0804681375293987,
      "grad_norm": 1.3155646324157715,
      "learning_rate": 9.549e-05,
      "loss": 0.3505,
      "step": 37153
    },
    {
      "epoch": 2.0805241348415278,
      "grad_norm": 1.1126888990402222,
      "learning_rate": 9.548973684210527e-05,
      "loss": 0.3629,
      "step": 37154
    },
    {
      "epoch": 2.080580132153657,
      "grad_norm": 1.2429786920547485,
      "learning_rate": 9.548947368421053e-05,
      "loss": 0.3864,
      "step": 37155
    },
    {
      "epoch": 2.080636129465786,
      "grad_norm": 1.2071207761764526,
      "learning_rate": 9.54892105263158e-05,
      "loss": 0.4656,
      "step": 37156
    },
    {
      "epoch": 2.080692126777915,
      "grad_norm": 1.2748379707336426,
      "learning_rate": 9.548894736842106e-05,
      "loss": 0.4055,
      "step": 37157
    },
    {
      "epoch": 2.080748124090044,
      "grad_norm": 1.146135687828064,
      "learning_rate": 9.548868421052632e-05,
      "loss": 0.3264,
      "step": 37158
    },
    {
      "epoch": 2.080804121402173,
      "grad_norm": 1.2361005544662476,
      "learning_rate": 9.548842105263158e-05,
      "loss": 0.3694,
      "step": 37159
    },
    {
      "epoch": 2.080860118714302,
      "grad_norm": 1.1005116701126099,
      "learning_rate": 9.548815789473685e-05,
      "loss": 0.3446,
      "step": 37160
    },
    {
      "epoch": 2.080916116026431,
      "grad_norm": 1.1418181657791138,
      "learning_rate": 9.548789473684211e-05,
      "loss": 0.3424,
      "step": 37161
    },
    {
      "epoch": 2.08097211333856,
      "grad_norm": 1.3484691381454468,
      "learning_rate": 9.548763157894737e-05,
      "loss": 0.4909,
      "step": 37162
    },
    {
      "epoch": 2.081028110650689,
      "grad_norm": 1.1424063444137573,
      "learning_rate": 9.548736842105263e-05,
      "loss": 0.4426,
      "step": 37163
    },
    {
      "epoch": 2.081084107962818,
      "grad_norm": 1.1802825927734375,
      "learning_rate": 9.548710526315789e-05,
      "loss": 0.4285,
      "step": 37164
    },
    {
      "epoch": 2.081140105274947,
      "grad_norm": 1.1364519596099854,
      "learning_rate": 9.548684210526317e-05,
      "loss": 0.4595,
      "step": 37165
    },
    {
      "epoch": 2.081196102587076,
      "grad_norm": 1.2519359588623047,
      "learning_rate": 9.548657894736843e-05,
      "loss": 0.3414,
      "step": 37166
    },
    {
      "epoch": 2.081252099899205,
      "grad_norm": 1.1235076189041138,
      "learning_rate": 9.548631578947369e-05,
      "loss": 0.4133,
      "step": 37167
    },
    {
      "epoch": 2.081308097211334,
      "grad_norm": 1.2762763500213623,
      "learning_rate": 9.548605263157895e-05,
      "loss": 0.4115,
      "step": 37168
    },
    {
      "epoch": 2.081364094523463,
      "grad_norm": 1.2176369428634644,
      "learning_rate": 9.548578947368422e-05,
      "loss": 0.4538,
      "step": 37169
    },
    {
      "epoch": 2.081420091835592,
      "grad_norm": 1.1449079513549805,
      "learning_rate": 9.548552631578948e-05,
      "loss": 0.4054,
      "step": 37170
    },
    {
      "epoch": 2.081476089147721,
      "grad_norm": 1.3267099857330322,
      "learning_rate": 9.548526315789474e-05,
      "loss": 0.4364,
      "step": 37171
    },
    {
      "epoch": 2.08153208645985,
      "grad_norm": 1.049601674079895,
      "learning_rate": 9.5485e-05,
      "loss": 0.3338,
      "step": 37172
    },
    {
      "epoch": 2.081588083771979,
      "grad_norm": 1.4149084091186523,
      "learning_rate": 9.548473684210527e-05,
      "loss": 0.4344,
      "step": 37173
    },
    {
      "epoch": 2.081644081084108,
      "grad_norm": 1.3202974796295166,
      "learning_rate": 9.548447368421053e-05,
      "loss": 0.3549,
      "step": 37174
    },
    {
      "epoch": 2.081700078396237,
      "grad_norm": 1.240950345993042,
      "learning_rate": 9.54842105263158e-05,
      "loss": 0.3334,
      "step": 37175
    },
    {
      "epoch": 2.081756075708366,
      "grad_norm": 1.1495739221572876,
      "learning_rate": 9.548394736842105e-05,
      "loss": 0.3302,
      "step": 37176
    },
    {
      "epoch": 2.081812073020495,
      "grad_norm": 1.6975127458572388,
      "learning_rate": 9.548368421052632e-05,
      "loss": 0.4443,
      "step": 37177
    },
    {
      "epoch": 2.081868070332624,
      "grad_norm": 0.9897533059120178,
      "learning_rate": 9.548342105263158e-05,
      "loss": 0.2651,
      "step": 37178
    },
    {
      "epoch": 2.0819240676447532,
      "grad_norm": 1.1337271928787231,
      "learning_rate": 9.548315789473686e-05,
      "loss": 0.3466,
      "step": 37179
    },
    {
      "epoch": 2.0819800649568823,
      "grad_norm": 1.257933497428894,
      "learning_rate": 9.548289473684212e-05,
      "loss": 0.3768,
      "step": 37180
    },
    {
      "epoch": 2.0820360622690113,
      "grad_norm": 1.4609490633010864,
      "learning_rate": 9.548263157894736e-05,
      "loss": 0.4275,
      "step": 37181
    },
    {
      "epoch": 2.0820920595811403,
      "grad_norm": 1.231135606765747,
      "learning_rate": 9.548236842105264e-05,
      "loss": 0.5232,
      "step": 37182
    },
    {
      "epoch": 2.0821480568932693,
      "grad_norm": 1.0626839399337769,
      "learning_rate": 9.54821052631579e-05,
      "loss": 0.4598,
      "step": 37183
    },
    {
      "epoch": 2.0822040542053983,
      "grad_norm": 1.3281725645065308,
      "learning_rate": 9.548184210526317e-05,
      "loss": 0.4482,
      "step": 37184
    },
    {
      "epoch": 2.0822600515175274,
      "grad_norm": 1.34035062789917,
      "learning_rate": 9.548157894736842e-05,
      "loss": 0.4053,
      "step": 37185
    },
    {
      "epoch": 2.0823160488296564,
      "grad_norm": 1.2800393104553223,
      "learning_rate": 9.548131578947369e-05,
      "loss": 0.4128,
      "step": 37186
    },
    {
      "epoch": 2.0823720461417854,
      "grad_norm": 1.0975431203842163,
      "learning_rate": 9.548105263157895e-05,
      "loss": 0.3698,
      "step": 37187
    },
    {
      "epoch": 2.0824280434539144,
      "grad_norm": 1.3764777183532715,
      "learning_rate": 9.548078947368422e-05,
      "loss": 0.3715,
      "step": 37188
    },
    {
      "epoch": 2.0824840407660434,
      "grad_norm": 1.9758543968200684,
      "learning_rate": 9.548052631578948e-05,
      "loss": 0.3885,
      "step": 37189
    },
    {
      "epoch": 2.0825400380781725,
      "grad_norm": 1.273565411567688,
      "learning_rate": 9.548026315789474e-05,
      "loss": 0.4156,
      "step": 37190
    },
    {
      "epoch": 2.0825960353903015,
      "grad_norm": 1.8587754964828491,
      "learning_rate": 9.548e-05,
      "loss": 0.3303,
      "step": 37191
    },
    {
      "epoch": 2.0826520327024305,
      "grad_norm": 1.2714428901672363,
      "learning_rate": 9.547973684210527e-05,
      "loss": 0.3596,
      "step": 37192
    },
    {
      "epoch": 2.0827080300145595,
      "grad_norm": 1.3159571886062622,
      "learning_rate": 9.547947368421053e-05,
      "loss": 0.393,
      "step": 37193
    },
    {
      "epoch": 2.0827640273266885,
      "grad_norm": 1.1187297105789185,
      "learning_rate": 9.54792105263158e-05,
      "loss": 0.3307,
      "step": 37194
    },
    {
      "epoch": 2.0828200246388175,
      "grad_norm": 1.2388211488723755,
      "learning_rate": 9.547894736842105e-05,
      "loss": 0.4144,
      "step": 37195
    },
    {
      "epoch": 2.0828760219509466,
      "grad_norm": 1.1398829221725464,
      "learning_rate": 9.547868421052633e-05,
      "loss": 0.44,
      "step": 37196
    },
    {
      "epoch": 2.0829320192630756,
      "grad_norm": 1.4079337120056152,
      "learning_rate": 9.547842105263159e-05,
      "loss": 0.4342,
      "step": 37197
    },
    {
      "epoch": 2.0829880165752046,
      "grad_norm": 1.1257526874542236,
      "learning_rate": 9.547815789473685e-05,
      "loss": 0.3424,
      "step": 37198
    },
    {
      "epoch": 2.083044013887333,
      "grad_norm": 1.1795446872711182,
      "learning_rate": 9.54778947368421e-05,
      "loss": 0.4446,
      "step": 37199
    },
    {
      "epoch": 2.0831000111994626,
      "grad_norm": 1.1247566938400269,
      "learning_rate": 9.547763157894737e-05,
      "loss": 0.3227,
      "step": 37200
    },
    {
      "epoch": 2.083156008511591,
      "grad_norm": 1.0809983015060425,
      "learning_rate": 9.547736842105264e-05,
      "loss": 0.3475,
      "step": 37201
    },
    {
      "epoch": 2.0832120058237207,
      "grad_norm": 1.2458832263946533,
      "learning_rate": 9.54771052631579e-05,
      "loss": 0.3239,
      "step": 37202
    },
    {
      "epoch": 2.0832680031358493,
      "grad_norm": 1.3078027963638306,
      "learning_rate": 9.547684210526316e-05,
      "loss": 0.4875,
      "step": 37203
    },
    {
      "epoch": 2.0833240004479787,
      "grad_norm": 1.0511339902877808,
      "learning_rate": 9.547657894736842e-05,
      "loss": 0.3621,
      "step": 37204
    },
    {
      "epoch": 2.0833799977601073,
      "grad_norm": 1.3642507791519165,
      "learning_rate": 9.547631578947369e-05,
      "loss": 0.3654,
      "step": 37205
    },
    {
      "epoch": 2.0834359950722368,
      "grad_norm": 1.306526780128479,
      "learning_rate": 9.547605263157895e-05,
      "loss": 0.3201,
      "step": 37206
    },
    {
      "epoch": 2.0834919923843653,
      "grad_norm": 1.1872780323028564,
      "learning_rate": 9.547578947368422e-05,
      "loss": 0.3796,
      "step": 37207
    },
    {
      "epoch": 2.083547989696495,
      "grad_norm": 1.3992371559143066,
      "learning_rate": 9.547552631578947e-05,
      "loss": 0.6462,
      "step": 37208
    },
    {
      "epoch": 2.0836039870086234,
      "grad_norm": 1.2817180156707764,
      "learning_rate": 9.547526315789474e-05,
      "loss": 0.4204,
      "step": 37209
    },
    {
      "epoch": 2.0836599843207524,
      "grad_norm": 1.1339430809020996,
      "learning_rate": 9.5475e-05,
      "loss": 0.3755,
      "step": 37210
    },
    {
      "epoch": 2.0837159816328814,
      "grad_norm": 1.2353686094284058,
      "learning_rate": 9.547473684210528e-05,
      "loss": 0.3471,
      "step": 37211
    },
    {
      "epoch": 2.0837719789450104,
      "grad_norm": 1.0560917854309082,
      "learning_rate": 9.547447368421054e-05,
      "loss": 0.3901,
      "step": 37212
    },
    {
      "epoch": 2.0838279762571394,
      "grad_norm": 1.170471429824829,
      "learning_rate": 9.54742105263158e-05,
      "loss": 0.3709,
      "step": 37213
    },
    {
      "epoch": 2.0838839735692685,
      "grad_norm": 1.3219298124313354,
      "learning_rate": 9.547394736842106e-05,
      "loss": 0.4167,
      "step": 37214
    },
    {
      "epoch": 2.0839399708813975,
      "grad_norm": 3.4093730449676514,
      "learning_rate": 9.547368421052632e-05,
      "loss": 0.4378,
      "step": 37215
    },
    {
      "epoch": 2.0839959681935265,
      "grad_norm": 1.3299362659454346,
      "learning_rate": 9.547342105263159e-05,
      "loss": 0.3874,
      "step": 37216
    },
    {
      "epoch": 2.0840519655056555,
      "grad_norm": 1.4559849500656128,
      "learning_rate": 9.547315789473685e-05,
      "loss": 0.5026,
      "step": 37217
    },
    {
      "epoch": 2.0841079628177845,
      "grad_norm": 1.4420535564422607,
      "learning_rate": 9.547289473684211e-05,
      "loss": 0.4487,
      "step": 37218
    },
    {
      "epoch": 2.0841639601299136,
      "grad_norm": 1.241303563117981,
      "learning_rate": 9.547263157894737e-05,
      "loss": 0.4259,
      "step": 37219
    },
    {
      "epoch": 2.0842199574420426,
      "grad_norm": 1.065914511680603,
      "learning_rate": 9.547236842105264e-05,
      "loss": 0.2619,
      "step": 37220
    },
    {
      "epoch": 2.0842759547541716,
      "grad_norm": 1.2404170036315918,
      "learning_rate": 9.54721052631579e-05,
      "loss": 0.3328,
      "step": 37221
    },
    {
      "epoch": 2.0843319520663006,
      "grad_norm": 1.225048303604126,
      "learning_rate": 9.547184210526316e-05,
      "loss": 0.3751,
      "step": 37222
    },
    {
      "epoch": 2.0843879493784296,
      "grad_norm": 1.1892751455307007,
      "learning_rate": 9.547157894736842e-05,
      "loss": 0.3604,
      "step": 37223
    },
    {
      "epoch": 2.0844439466905587,
      "grad_norm": 1.1867549419403076,
      "learning_rate": 9.54713157894737e-05,
      "loss": 0.3135,
      "step": 37224
    },
    {
      "epoch": 2.0844999440026877,
      "grad_norm": 1.1025311946868896,
      "learning_rate": 9.547105263157895e-05,
      "loss": 0.441,
      "step": 37225
    },
    {
      "epoch": 2.0845559413148167,
      "grad_norm": 1.0714370012283325,
      "learning_rate": 9.547078947368421e-05,
      "loss": 0.3109,
      "step": 37226
    },
    {
      "epoch": 2.0846119386269457,
      "grad_norm": 1.4138555526733398,
      "learning_rate": 9.547052631578947e-05,
      "loss": 0.3932,
      "step": 37227
    },
    {
      "epoch": 2.0846679359390747,
      "grad_norm": 1.2056083679199219,
      "learning_rate": 9.547026315789475e-05,
      "loss": 0.35,
      "step": 37228
    },
    {
      "epoch": 2.0847239332512038,
      "grad_norm": 2.1281239986419678,
      "learning_rate": 9.547e-05,
      "loss": 0.5038,
      "step": 37229
    },
    {
      "epoch": 2.0847799305633328,
      "grad_norm": 0.9300613403320312,
      "learning_rate": 9.546973684210528e-05,
      "loss": 0.2665,
      "step": 37230
    },
    {
      "epoch": 2.084835927875462,
      "grad_norm": 1.1529935598373413,
      "learning_rate": 9.546947368421053e-05,
      "loss": 0.3517,
      "step": 37231
    },
    {
      "epoch": 2.084891925187591,
      "grad_norm": 1.3257426023483276,
      "learning_rate": 9.546921052631578e-05,
      "loss": 0.4568,
      "step": 37232
    },
    {
      "epoch": 2.08494792249972,
      "grad_norm": Infinity,
      "learning_rate": 9.546921052631578e-05,
      "loss": 0.3722,
      "step": 37233
    },
    {
      "epoch": 2.085003919811849,
      "grad_norm": 1.205477237701416,
      "learning_rate": 9.546894736842106e-05,
      "loss": 0.4136,
      "step": 37234
    },
    {
      "epoch": 2.085059917123978,
      "grad_norm": 1.3416367769241333,
      "learning_rate": 9.546868421052632e-05,
      "loss": 0.4135,
      "step": 37235
    },
    {
      "epoch": 2.085115914436107,
      "grad_norm": 1.068905234336853,
      "learning_rate": 9.546842105263159e-05,
      "loss": 0.3123,
      "step": 37236
    },
    {
      "epoch": 2.085171911748236,
      "grad_norm": 1.1558846235275269,
      "learning_rate": 9.546815789473684e-05,
      "loss": 0.3197,
      "step": 37237
    },
    {
      "epoch": 2.085227909060365,
      "grad_norm": 1.882606863975525,
      "learning_rate": 9.546789473684211e-05,
      "loss": 0.4793,
      "step": 37238
    },
    {
      "epoch": 2.085283906372494,
      "grad_norm": 1.1726343631744385,
      "learning_rate": 9.546763157894737e-05,
      "loss": 0.4171,
      "step": 37239
    },
    {
      "epoch": 2.085339903684623,
      "grad_norm": 3.1281239986419678,
      "learning_rate": 9.546736842105264e-05,
      "loss": 0.472,
      "step": 37240
    },
    {
      "epoch": 2.085395900996752,
      "grad_norm": 1.4461760520935059,
      "learning_rate": 9.546710526315789e-05,
      "loss": 0.5411,
      "step": 37241
    },
    {
      "epoch": 2.085451898308881,
      "grad_norm": 1.129280686378479,
      "learning_rate": 9.546684210526316e-05,
      "loss": 0.3339,
      "step": 37242
    },
    {
      "epoch": 2.08550789562101,
      "grad_norm": 1.1016062498092651,
      "learning_rate": 9.546657894736842e-05,
      "loss": 0.4184,
      "step": 37243
    },
    {
      "epoch": 2.085563892933139,
      "grad_norm": 1.110470175743103,
      "learning_rate": 9.54663157894737e-05,
      "loss": 0.512,
      "step": 37244
    },
    {
      "epoch": 2.085619890245268,
      "grad_norm": 1.012777328491211,
      "learning_rate": 9.546605263157896e-05,
      "loss": 0.2953,
      "step": 37245
    },
    {
      "epoch": 2.085675887557397,
      "grad_norm": 1.058820366859436,
      "learning_rate": 9.546578947368422e-05,
      "loss": 0.3438,
      "step": 37246
    },
    {
      "epoch": 2.085731884869526,
      "grad_norm": 1.2653052806854248,
      "learning_rate": 9.546552631578948e-05,
      "loss": 0.3219,
      "step": 37247
    },
    {
      "epoch": 2.085787882181655,
      "grad_norm": 1.0162802934646606,
      "learning_rate": 9.546526315789475e-05,
      "loss": 0.3695,
      "step": 37248
    },
    {
      "epoch": 2.085843879493784,
      "grad_norm": 1.2537152767181396,
      "learning_rate": 9.546500000000001e-05,
      "loss": 0.4265,
      "step": 37249
    },
    {
      "epoch": 2.085899876805913,
      "grad_norm": 1.3454452753067017,
      "learning_rate": 9.546473684210527e-05,
      "loss": 0.383,
      "step": 37250
    },
    {
      "epoch": 2.085955874118042,
      "grad_norm": 1.2169373035430908,
      "learning_rate": 9.546447368421053e-05,
      "loss": 0.3409,
      "step": 37251
    },
    {
      "epoch": 2.086011871430171,
      "grad_norm": 1.344805359840393,
      "learning_rate": 9.546421052631579e-05,
      "loss": 0.4391,
      "step": 37252
    },
    {
      "epoch": 2.0860678687423,
      "grad_norm": 1.2403202056884766,
      "learning_rate": 9.546394736842106e-05,
      "loss": 0.3225,
      "step": 37253
    },
    {
      "epoch": 2.0861238660544292,
      "grad_norm": 1.2060853242874146,
      "learning_rate": 9.546368421052632e-05,
      "loss": 0.4632,
      "step": 37254
    },
    {
      "epoch": 2.0861798633665583,
      "grad_norm": 1.178431749343872,
      "learning_rate": 9.546342105263158e-05,
      "loss": 0.3967,
      "step": 37255
    },
    {
      "epoch": 2.0862358606786873,
      "grad_norm": 1.0614399909973145,
      "learning_rate": 9.546315789473684e-05,
      "loss": 0.3317,
      "step": 37256
    },
    {
      "epoch": 2.0862918579908163,
      "grad_norm": 1.1768767833709717,
      "learning_rate": 9.546289473684211e-05,
      "loss": 0.4176,
      "step": 37257
    },
    {
      "epoch": 2.0863478553029453,
      "grad_norm": 1.1512436866760254,
      "learning_rate": 9.546263157894737e-05,
      "loss": 0.3549,
      "step": 37258
    },
    {
      "epoch": 2.0864038526150743,
      "grad_norm": 2.1010122299194336,
      "learning_rate": 9.546236842105263e-05,
      "loss": 0.496,
      "step": 37259
    },
    {
      "epoch": 2.0864598499272033,
      "grad_norm": 1.272742748260498,
      "learning_rate": 9.546210526315789e-05,
      "loss": 0.3605,
      "step": 37260
    },
    {
      "epoch": 2.0865158472393324,
      "grad_norm": 1.2747138738632202,
      "learning_rate": 9.546184210526317e-05,
      "loss": 0.4403,
      "step": 37261
    },
    {
      "epoch": 2.0865718445514614,
      "grad_norm": 1.161557912826538,
      "learning_rate": 9.546157894736843e-05,
      "loss": 0.378,
      "step": 37262
    },
    {
      "epoch": 2.0866278418635904,
      "grad_norm": 1.02803635597229,
      "learning_rate": 9.54613157894737e-05,
      "loss": 0.3952,
      "step": 37263
    },
    {
      "epoch": 2.0866838391757194,
      "grad_norm": 1.4038516283035278,
      "learning_rate": 9.546105263157894e-05,
      "loss": 0.3973,
      "step": 37264
    },
    {
      "epoch": 2.0867398364878484,
      "grad_norm": 0.9738485217094421,
      "learning_rate": 9.546078947368422e-05,
      "loss": 0.3993,
      "step": 37265
    },
    {
      "epoch": 2.0867958337999775,
      "grad_norm": 1.2438774108886719,
      "learning_rate": 9.546052631578948e-05,
      "loss": 0.3888,
      "step": 37266
    },
    {
      "epoch": 2.0868518311121065,
      "grad_norm": 1.4068541526794434,
      "learning_rate": 9.546026315789474e-05,
      "loss": 0.4481,
      "step": 37267
    },
    {
      "epoch": 2.0869078284242355,
      "grad_norm": 1.2054994106292725,
      "learning_rate": 9.546000000000001e-05,
      "loss": 0.4418,
      "step": 37268
    },
    {
      "epoch": 2.0869638257363645,
      "grad_norm": 1.1550027132034302,
      "learning_rate": 9.545973684210526e-05,
      "loss": 0.3843,
      "step": 37269
    },
    {
      "epoch": 2.0870198230484935,
      "grad_norm": 1.2914761304855347,
      "learning_rate": 9.545947368421053e-05,
      "loss": 0.4585,
      "step": 37270
    },
    {
      "epoch": 2.0870758203606226,
      "grad_norm": 1.358154535293579,
      "learning_rate": 9.545921052631579e-05,
      "loss": 0.4164,
      "step": 37271
    },
    {
      "epoch": 2.0871318176727516,
      "grad_norm": 1.304789662361145,
      "learning_rate": 9.545894736842106e-05,
      "loss": 0.5312,
      "step": 37272
    },
    {
      "epoch": 2.0871878149848806,
      "grad_norm": 1.2800456285476685,
      "learning_rate": 9.545868421052632e-05,
      "loss": 0.4901,
      "step": 37273
    },
    {
      "epoch": 2.0872438122970096,
      "grad_norm": 1.2947059869766235,
      "learning_rate": 9.545842105263158e-05,
      "loss": 0.3777,
      "step": 37274
    },
    {
      "epoch": 2.0872998096091386,
      "grad_norm": 1.2609716653823853,
      "learning_rate": 9.545815789473684e-05,
      "loss": 0.3797,
      "step": 37275
    },
    {
      "epoch": 2.0873558069212677,
      "grad_norm": 1.1620147228240967,
      "learning_rate": 9.545789473684212e-05,
      "loss": 0.4261,
      "step": 37276
    },
    {
      "epoch": 2.0874118042333967,
      "grad_norm": 1.3531605005264282,
      "learning_rate": 9.545763157894738e-05,
      "loss": 0.3702,
      "step": 37277
    },
    {
      "epoch": 2.0874678015455257,
      "grad_norm": 1.0374218225479126,
      "learning_rate": 9.545736842105264e-05,
      "loss": 0.3551,
      "step": 37278
    },
    {
      "epoch": 2.0875237988576547,
      "grad_norm": 1.289520263671875,
      "learning_rate": 9.54571052631579e-05,
      "loss": 0.3349,
      "step": 37279
    },
    {
      "epoch": 2.0875797961697837,
      "grad_norm": 1.2119457721710205,
      "learning_rate": 9.545684210526317e-05,
      "loss": 0.4018,
      "step": 37280
    },
    {
      "epoch": 2.0876357934819127,
      "grad_norm": 1.100595474243164,
      "learning_rate": 9.545657894736843e-05,
      "loss": 0.4861,
      "step": 37281
    },
    {
      "epoch": 2.0876917907940418,
      "grad_norm": 1.183851957321167,
      "learning_rate": 9.545631578947369e-05,
      "loss": 0.5076,
      "step": 37282
    },
    {
      "epoch": 2.087747788106171,
      "grad_norm": 1.2981088161468506,
      "learning_rate": 9.545605263157895e-05,
      "loss": 0.506,
      "step": 37283
    },
    {
      "epoch": 2.0878037854183,
      "grad_norm": 1.307874321937561,
      "learning_rate": 9.545578947368422e-05,
      "loss": 0.4809,
      "step": 37284
    },
    {
      "epoch": 2.087859782730429,
      "grad_norm": 1.1047242879867554,
      "learning_rate": 9.545552631578948e-05,
      "loss": 0.368,
      "step": 37285
    },
    {
      "epoch": 2.087915780042558,
      "grad_norm": 1.3026601076126099,
      "learning_rate": 9.545526315789474e-05,
      "loss": 0.4654,
      "step": 37286
    },
    {
      "epoch": 2.087971777354687,
      "grad_norm": 0.9914889931678772,
      "learning_rate": 9.5455e-05,
      "loss": 0.3641,
      "step": 37287
    },
    {
      "epoch": 2.088027774666816,
      "grad_norm": 1.3385239839553833,
      "learning_rate": 9.545473684210526e-05,
      "loss": 0.3314,
      "step": 37288
    },
    {
      "epoch": 2.088083771978945,
      "grad_norm": 1.364428997039795,
      "learning_rate": 9.545447368421053e-05,
      "loss": 0.6244,
      "step": 37289
    },
    {
      "epoch": 2.088139769291074,
      "grad_norm": 1.5114381313323975,
      "learning_rate": 9.545421052631579e-05,
      "loss": 0.3504,
      "step": 37290
    },
    {
      "epoch": 2.088195766603203,
      "grad_norm": 0.9901379346847534,
      "learning_rate": 9.545394736842107e-05,
      "loss": 0.302,
      "step": 37291
    },
    {
      "epoch": 2.088251763915332,
      "grad_norm": 1.1798889636993408,
      "learning_rate": 9.545368421052631e-05,
      "loss": 0.4512,
      "step": 37292
    },
    {
      "epoch": 2.088307761227461,
      "grad_norm": 1.1171339750289917,
      "learning_rate": 9.545342105263159e-05,
      "loss": 0.4173,
      "step": 37293
    },
    {
      "epoch": 2.08836375853959,
      "grad_norm": 1.1986720561981201,
      "learning_rate": 9.545315789473685e-05,
      "loss": 0.4038,
      "step": 37294
    },
    {
      "epoch": 2.088419755851719,
      "grad_norm": 1.1060482263565063,
      "learning_rate": 9.545289473684212e-05,
      "loss": 0.3582,
      "step": 37295
    },
    {
      "epoch": 2.088475753163848,
      "grad_norm": 1.1861315965652466,
      "learning_rate": 9.545263157894736e-05,
      "loss": 0.4357,
      "step": 37296
    },
    {
      "epoch": 2.088531750475977,
      "grad_norm": 1.3403488397598267,
      "learning_rate": 9.545236842105264e-05,
      "loss": 0.3187,
      "step": 37297
    },
    {
      "epoch": 2.088587747788106,
      "grad_norm": 1.2128984928131104,
      "learning_rate": 9.54521052631579e-05,
      "loss": 0.4774,
      "step": 37298
    },
    {
      "epoch": 2.088643745100235,
      "grad_norm": 1.0585711002349854,
      "learning_rate": 9.545184210526317e-05,
      "loss": 0.3591,
      "step": 37299
    },
    {
      "epoch": 2.088699742412364,
      "grad_norm": 1.1522780656814575,
      "learning_rate": 9.545157894736843e-05,
      "loss": 0.4074,
      "step": 37300
    },
    {
      "epoch": 2.088755739724493,
      "grad_norm": 1.0165148973464966,
      "learning_rate": 9.545131578947369e-05,
      "loss": 0.2194,
      "step": 37301
    },
    {
      "epoch": 2.088811737036622,
      "grad_norm": 1.1626001596450806,
      "learning_rate": 9.545105263157895e-05,
      "loss": 0.3202,
      "step": 37302
    },
    {
      "epoch": 2.088867734348751,
      "grad_norm": 1.2696199417114258,
      "learning_rate": 9.545078947368421e-05,
      "loss": 0.353,
      "step": 37303
    },
    {
      "epoch": 2.08892373166088,
      "grad_norm": 1.1939014196395874,
      "learning_rate": 9.545052631578948e-05,
      "loss": 0.4128,
      "step": 37304
    },
    {
      "epoch": 2.088979728973009,
      "grad_norm": 1.1139200925827026,
      "learning_rate": 9.545026315789474e-05,
      "loss": 0.3994,
      "step": 37305
    },
    {
      "epoch": 2.0890357262851382,
      "grad_norm": 1.2029638290405273,
      "learning_rate": 9.545e-05,
      "loss": 0.6494,
      "step": 37306
    },
    {
      "epoch": 2.0890917235972672,
      "grad_norm": 1.1315141916275024,
      "learning_rate": 9.544973684210526e-05,
      "loss": 0.3814,
      "step": 37307
    },
    {
      "epoch": 2.0891477209093963,
      "grad_norm": 1.1952275037765503,
      "learning_rate": 9.544947368421054e-05,
      "loss": 0.5266,
      "step": 37308
    },
    {
      "epoch": 2.0892037182215253,
      "grad_norm": 1.12501859664917,
      "learning_rate": 9.54492105263158e-05,
      "loss": 0.4222,
      "step": 37309
    },
    {
      "epoch": 2.0892597155336543,
      "grad_norm": 1.1186892986297607,
      "learning_rate": 9.544894736842106e-05,
      "loss": 0.406,
      "step": 37310
    },
    {
      "epoch": 2.0893157128457833,
      "grad_norm": 1.1692185401916504,
      "learning_rate": 9.544868421052631e-05,
      "loss": 0.3359,
      "step": 37311
    },
    {
      "epoch": 2.0893717101579123,
      "grad_norm": 1.2075005769729614,
      "learning_rate": 9.544842105263159e-05,
      "loss": 0.4156,
      "step": 37312
    },
    {
      "epoch": 2.0894277074700414,
      "grad_norm": 1.2053308486938477,
      "learning_rate": 9.544815789473685e-05,
      "loss": 0.3962,
      "step": 37313
    },
    {
      "epoch": 2.0894837047821704,
      "grad_norm": 1.2531945705413818,
      "learning_rate": 9.544789473684211e-05,
      "loss": 0.3937,
      "step": 37314
    },
    {
      "epoch": 2.0895397020942994,
      "grad_norm": 1.2690480947494507,
      "learning_rate": 9.544763157894737e-05,
      "loss": 0.4204,
      "step": 37315
    },
    {
      "epoch": 2.0895956994064284,
      "grad_norm": 1.3338758945465088,
      "learning_rate": 9.544736842105264e-05,
      "loss": 0.4211,
      "step": 37316
    },
    {
      "epoch": 2.0896516967185574,
      "grad_norm": 1.301088809967041,
      "learning_rate": 9.54471052631579e-05,
      "loss": 0.3492,
      "step": 37317
    },
    {
      "epoch": 2.0897076940306865,
      "grad_norm": 1.335763931274414,
      "learning_rate": 9.544684210526317e-05,
      "loss": 0.345,
      "step": 37318
    },
    {
      "epoch": 2.0897636913428155,
      "grad_norm": 1.5093432664871216,
      "learning_rate": 9.544657894736842e-05,
      "loss": 0.3776,
      "step": 37319
    },
    {
      "epoch": 2.0898196886549445,
      "grad_norm": 1.188510537147522,
      "learning_rate": 9.544631578947368e-05,
      "loss": 0.4294,
      "step": 37320
    },
    {
      "epoch": 2.0898756859670735,
      "grad_norm": 0.995898425579071,
      "learning_rate": 9.544605263157895e-05,
      "loss": 0.2933,
      "step": 37321
    },
    {
      "epoch": 2.0899316832792025,
      "grad_norm": 1.7647291421890259,
      "learning_rate": 9.544578947368421e-05,
      "loss": 0.5691,
      "step": 37322
    },
    {
      "epoch": 2.0899876805913316,
      "grad_norm": 1.2489018440246582,
      "learning_rate": 9.544552631578949e-05,
      "loss": 0.4336,
      "step": 37323
    },
    {
      "epoch": 2.0900436779034606,
      "grad_norm": 1.1707091331481934,
      "learning_rate": 9.544526315789473e-05,
      "loss": 0.4447,
      "step": 37324
    },
    {
      "epoch": 2.0900996752155896,
      "grad_norm": 1.239990234375,
      "learning_rate": 9.5445e-05,
      "loss": 0.4153,
      "step": 37325
    },
    {
      "epoch": 2.0901556725277186,
      "grad_norm": 1.0833947658538818,
      "learning_rate": 9.544473684210526e-05,
      "loss": 0.3389,
      "step": 37326
    },
    {
      "epoch": 2.0902116698398476,
      "grad_norm": 1.1745514869689941,
      "learning_rate": 9.544447368421054e-05,
      "loss": 0.4703,
      "step": 37327
    },
    {
      "epoch": 2.0902676671519766,
      "grad_norm": 1.8172285556793213,
      "learning_rate": 9.54442105263158e-05,
      "loss": 0.4035,
      "step": 37328
    },
    {
      "epoch": 2.0903236644641057,
      "grad_norm": 1.1282519102096558,
      "learning_rate": 9.544394736842106e-05,
      "loss": 0.332,
      "step": 37329
    },
    {
      "epoch": 2.0903796617762347,
      "grad_norm": 1.2324405908584595,
      "learning_rate": 9.544368421052632e-05,
      "loss": 0.4185,
      "step": 37330
    },
    {
      "epoch": 2.0904356590883637,
      "grad_norm": 1.5262449979782104,
      "learning_rate": 9.544342105263159e-05,
      "loss": 0.3879,
      "step": 37331
    },
    {
      "epoch": 2.0904916564004927,
      "grad_norm": 1.236594557762146,
      "learning_rate": 9.544315789473685e-05,
      "loss": 0.4264,
      "step": 37332
    },
    {
      "epoch": 2.0905476537126217,
      "grad_norm": 1.2415767908096313,
      "learning_rate": 9.544289473684211e-05,
      "loss": 0.4047,
      "step": 37333
    },
    {
      "epoch": 2.0906036510247508,
      "grad_norm": 1.2237541675567627,
      "learning_rate": 9.544263157894737e-05,
      "loss": 0.4503,
      "step": 37334
    },
    {
      "epoch": 2.09065964833688,
      "grad_norm": 1.3254679441452026,
      "learning_rate": 9.544236842105264e-05,
      "loss": 0.332,
      "step": 37335
    },
    {
      "epoch": 2.090715645649009,
      "grad_norm": 1.1052825450897217,
      "learning_rate": 9.54421052631579e-05,
      "loss": 0.3597,
      "step": 37336
    },
    {
      "epoch": 2.090771642961138,
      "grad_norm": 1.513141393661499,
      "learning_rate": 9.544184210526316e-05,
      "loss": 0.5085,
      "step": 37337
    },
    {
      "epoch": 2.090827640273267,
      "grad_norm": 1.5670636892318726,
      "learning_rate": 9.544157894736842e-05,
      "loss": 0.3812,
      "step": 37338
    },
    {
      "epoch": 2.090883637585396,
      "grad_norm": 1.4884428977966309,
      "learning_rate": 9.544131578947368e-05,
      "loss": 0.3865,
      "step": 37339
    },
    {
      "epoch": 2.090939634897525,
      "grad_norm": 1.1959185600280762,
      "learning_rate": 9.544105263157896e-05,
      "loss": 0.4192,
      "step": 37340
    },
    {
      "epoch": 2.090995632209654,
      "grad_norm": 1.0133477449417114,
      "learning_rate": 9.544078947368422e-05,
      "loss": 0.4587,
      "step": 37341
    },
    {
      "epoch": 2.091051629521783,
      "grad_norm": 1.1428687572479248,
      "learning_rate": 9.544052631578947e-05,
      "loss": 0.4325,
      "step": 37342
    },
    {
      "epoch": 2.091107626833912,
      "grad_norm": 1.1978991031646729,
      "learning_rate": 9.544026315789473e-05,
      "loss": 0.4032,
      "step": 37343
    },
    {
      "epoch": 2.091163624146041,
      "grad_norm": 1.2987525463104248,
      "learning_rate": 9.544000000000001e-05,
      "loss": 0.3846,
      "step": 37344
    },
    {
      "epoch": 2.09121962145817,
      "grad_norm": 1.6236987113952637,
      "learning_rate": 9.543973684210527e-05,
      "loss": 0.4488,
      "step": 37345
    },
    {
      "epoch": 2.091275618770299,
      "grad_norm": 1.1632187366485596,
      "learning_rate": 9.543947368421054e-05,
      "loss": 0.3504,
      "step": 37346
    },
    {
      "epoch": 2.091331616082428,
      "grad_norm": 1.1244224309921265,
      "learning_rate": 9.543921052631579e-05,
      "loss": 0.4028,
      "step": 37347
    },
    {
      "epoch": 2.091387613394557,
      "grad_norm": 1.3994784355163574,
      "learning_rate": 9.543894736842106e-05,
      "loss": 0.4271,
      "step": 37348
    },
    {
      "epoch": 2.091443610706686,
      "grad_norm": 1.2536182403564453,
      "learning_rate": 9.543868421052632e-05,
      "loss": 0.4585,
      "step": 37349
    },
    {
      "epoch": 2.091499608018815,
      "grad_norm": 4.12711238861084,
      "learning_rate": 9.54384210526316e-05,
      "loss": 0.4133,
      "step": 37350
    },
    {
      "epoch": 2.091555605330944,
      "grad_norm": 1.0105020999908447,
      "learning_rate": 9.543815789473684e-05,
      "loss": 0.3117,
      "step": 37351
    },
    {
      "epoch": 2.091611602643073,
      "grad_norm": 1.1619453430175781,
      "learning_rate": 9.543789473684211e-05,
      "loss": 0.3117,
      "step": 37352
    },
    {
      "epoch": 2.091667599955202,
      "grad_norm": 1.6516146659851074,
      "learning_rate": 9.543763157894737e-05,
      "loss": 0.409,
      "step": 37353
    },
    {
      "epoch": 2.091723597267331,
      "grad_norm": 1.3395391702651978,
      "learning_rate": 9.543736842105263e-05,
      "loss": 0.4325,
      "step": 37354
    },
    {
      "epoch": 2.09177959457946,
      "grad_norm": 1.1015377044677734,
      "learning_rate": 9.54371052631579e-05,
      "loss": 0.3454,
      "step": 37355
    },
    {
      "epoch": 2.091835591891589,
      "grad_norm": 1.5405291318893433,
      "learning_rate": 9.543684210526315e-05,
      "loss": 0.5277,
      "step": 37356
    },
    {
      "epoch": 2.091891589203718,
      "grad_norm": 1.5420892238616943,
      "learning_rate": 9.543657894736842e-05,
      "loss": 0.4734,
      "step": 37357
    },
    {
      "epoch": 2.0919475865158472,
      "grad_norm": 1.1422734260559082,
      "learning_rate": 9.543631578947368e-05,
      "loss": 0.4361,
      "step": 37358
    },
    {
      "epoch": 2.0920035838279762,
      "grad_norm": 1.1816012859344482,
      "learning_rate": 9.543605263157896e-05,
      "loss": 0.39,
      "step": 37359
    },
    {
      "epoch": 2.0920595811401053,
      "grad_norm": 1.4430456161499023,
      "learning_rate": 9.543578947368422e-05,
      "loss": 0.4375,
      "step": 37360
    },
    {
      "epoch": 2.0921155784522343,
      "grad_norm": 1.3723641633987427,
      "learning_rate": 9.543552631578948e-05,
      "loss": 0.618,
      "step": 37361
    },
    {
      "epoch": 2.0921715757643633,
      "grad_norm": 1.2190183401107788,
      "learning_rate": 9.543526315789474e-05,
      "loss": 0.4008,
      "step": 37362
    },
    {
      "epoch": 2.0922275730764923,
      "grad_norm": 1.0896131992340088,
      "learning_rate": 9.543500000000001e-05,
      "loss": 0.3852,
      "step": 37363
    },
    {
      "epoch": 2.0922835703886213,
      "grad_norm": 3.0846312046051025,
      "learning_rate": 9.543473684210527e-05,
      "loss": 0.3818,
      "step": 37364
    },
    {
      "epoch": 2.0923395677007504,
      "grad_norm": 1.1485261917114258,
      "learning_rate": 9.543447368421053e-05,
      "loss": 0.4209,
      "step": 37365
    },
    {
      "epoch": 2.0923955650128794,
      "grad_norm": 1.19721257686615,
      "learning_rate": 9.543421052631579e-05,
      "loss": 0.3192,
      "step": 37366
    },
    {
      "epoch": 2.0924515623250084,
      "grad_norm": 1.2039201259613037,
      "learning_rate": 9.543394736842106e-05,
      "loss": 0.369,
      "step": 37367
    },
    {
      "epoch": 2.0925075596371374,
      "grad_norm": 1.1867198944091797,
      "learning_rate": 9.543368421052632e-05,
      "loss": 0.4234,
      "step": 37368
    },
    {
      "epoch": 2.0925635569492664,
      "grad_norm": 1.2044740915298462,
      "learning_rate": 9.543342105263158e-05,
      "loss": 0.3206,
      "step": 37369
    },
    {
      "epoch": 2.0926195542613955,
      "grad_norm": 2.129072427749634,
      "learning_rate": 9.543315789473684e-05,
      "loss": 0.3831,
      "step": 37370
    },
    {
      "epoch": 2.0926755515735245,
      "grad_norm": 1.0805439949035645,
      "learning_rate": 9.54328947368421e-05,
      "loss": 0.3292,
      "step": 37371
    },
    {
      "epoch": 2.0927315488856535,
      "grad_norm": 1.4952677488327026,
      "learning_rate": 9.543263157894738e-05,
      "loss": 0.4247,
      "step": 37372
    },
    {
      "epoch": 2.0927875461977825,
      "grad_norm": 1.153788447380066,
      "learning_rate": 9.543236842105263e-05,
      "loss": 0.3438,
      "step": 37373
    },
    {
      "epoch": 2.0928435435099115,
      "grad_norm": 1.24036705493927,
      "learning_rate": 9.54321052631579e-05,
      "loss": 0.4539,
      "step": 37374
    },
    {
      "epoch": 2.0928995408220405,
      "grad_norm": 1.5538272857666016,
      "learning_rate": 9.543184210526315e-05,
      "loss": 0.6366,
      "step": 37375
    },
    {
      "epoch": 2.0929555381341696,
      "grad_norm": 1.5553739070892334,
      "learning_rate": 9.543157894736843e-05,
      "loss": 0.4747,
      "step": 37376
    },
    {
      "epoch": 2.0930115354462986,
      "grad_norm": 1.358181357383728,
      "learning_rate": 9.543131578947369e-05,
      "loss": 0.5924,
      "step": 37377
    },
    {
      "epoch": 2.0930675327584276,
      "grad_norm": 1.3236382007598877,
      "learning_rate": 9.543105263157896e-05,
      "loss": 0.3966,
      "step": 37378
    },
    {
      "epoch": 2.0931235300705566,
      "grad_norm": 1.1184443235397339,
      "learning_rate": 9.54307894736842e-05,
      "loss": 0.3685,
      "step": 37379
    },
    {
      "epoch": 2.0931795273826856,
      "grad_norm": 1.0814545154571533,
      "learning_rate": 9.543052631578948e-05,
      "loss": 0.3307,
      "step": 37380
    },
    {
      "epoch": 2.0932355246948147,
      "grad_norm": 1.1240146160125732,
      "learning_rate": 9.543026315789474e-05,
      "loss": 0.3621,
      "step": 37381
    },
    {
      "epoch": 2.0932915220069437,
      "grad_norm": 1.158893346786499,
      "learning_rate": 9.543000000000001e-05,
      "loss": 0.3994,
      "step": 37382
    },
    {
      "epoch": 2.0933475193190727,
      "grad_norm": 1.1639890670776367,
      "learning_rate": 9.542973684210527e-05,
      "loss": 0.4934,
      "step": 37383
    },
    {
      "epoch": 2.0934035166312017,
      "grad_norm": 1.4643397331237793,
      "learning_rate": 9.542947368421053e-05,
      "loss": 0.4854,
      "step": 37384
    },
    {
      "epoch": 2.0934595139433307,
      "grad_norm": 1.1275420188903809,
      "learning_rate": 9.542921052631579e-05,
      "loss": 0.3062,
      "step": 37385
    },
    {
      "epoch": 2.0935155112554598,
      "grad_norm": 1.1237006187438965,
      "learning_rate": 9.542894736842107e-05,
      "loss": 0.5781,
      "step": 37386
    },
    {
      "epoch": 2.093571508567589,
      "grad_norm": 1.3043904304504395,
      "learning_rate": 9.542868421052633e-05,
      "loss": 0.493,
      "step": 37387
    },
    {
      "epoch": 2.093627505879718,
      "grad_norm": 1.6650962829589844,
      "learning_rate": 9.542842105263157e-05,
      "loss": 0.5845,
      "step": 37388
    },
    {
      "epoch": 2.093683503191847,
      "grad_norm": 1.3205933570861816,
      "learning_rate": 9.542815789473684e-05,
      "loss": 0.3947,
      "step": 37389
    },
    {
      "epoch": 2.093739500503976,
      "grad_norm": 1.439923644065857,
      "learning_rate": 9.54278947368421e-05,
      "loss": 0.5636,
      "step": 37390
    },
    {
      "epoch": 2.093795497816105,
      "grad_norm": 1.2145929336547852,
      "learning_rate": 9.542763157894738e-05,
      "loss": 0.38,
      "step": 37391
    },
    {
      "epoch": 2.093851495128234,
      "grad_norm": 1.3981770277023315,
      "learning_rate": 9.542736842105264e-05,
      "loss": 0.3531,
      "step": 37392
    },
    {
      "epoch": 2.093907492440363,
      "grad_norm": 1.2881981134414673,
      "learning_rate": 9.54271052631579e-05,
      "loss": 0.4195,
      "step": 37393
    },
    {
      "epoch": 2.093963489752492,
      "grad_norm": 1.1832435131072998,
      "learning_rate": 9.542684210526316e-05,
      "loss": 0.439,
      "step": 37394
    },
    {
      "epoch": 2.094019487064621,
      "grad_norm": 1.1978384256362915,
      "learning_rate": 9.542657894736843e-05,
      "loss": 0.4055,
      "step": 37395
    },
    {
      "epoch": 2.09407548437675,
      "grad_norm": 1.3463934659957886,
      "learning_rate": 9.542631578947369e-05,
      "loss": 0.3306,
      "step": 37396
    },
    {
      "epoch": 2.094131481688879,
      "grad_norm": 1.0849336385726929,
      "learning_rate": 9.542605263157895e-05,
      "loss": 0.3562,
      "step": 37397
    },
    {
      "epoch": 2.094187479001008,
      "grad_norm": 1.3308988809585571,
      "learning_rate": 9.542578947368421e-05,
      "loss": 0.3903,
      "step": 37398
    },
    {
      "epoch": 2.094243476313137,
      "grad_norm": 1.2387291193008423,
      "learning_rate": 9.542552631578948e-05,
      "loss": 0.4327,
      "step": 37399
    },
    {
      "epoch": 2.094299473625266,
      "grad_norm": 1.2718130350112915,
      "learning_rate": 9.542526315789474e-05,
      "loss": 0.5289,
      "step": 37400
    },
    {
      "epoch": 2.094355470937395,
      "grad_norm": 1.0964674949645996,
      "learning_rate": 9.542500000000002e-05,
      "loss": 0.2977,
      "step": 37401
    },
    {
      "epoch": 2.094411468249524,
      "grad_norm": 1.032365322113037,
      "learning_rate": 9.542473684210526e-05,
      "loss": 0.3181,
      "step": 37402
    },
    {
      "epoch": 2.094467465561653,
      "grad_norm": 1.5739465951919556,
      "learning_rate": 9.542447368421054e-05,
      "loss": 0.4535,
      "step": 37403
    },
    {
      "epoch": 2.094523462873782,
      "grad_norm": 1.0292078256607056,
      "learning_rate": 9.54242105263158e-05,
      "loss": 0.4825,
      "step": 37404
    },
    {
      "epoch": 2.094579460185911,
      "grad_norm": 1.3064666986465454,
      "learning_rate": 9.542394736842107e-05,
      "loss": 0.4716,
      "step": 37405
    },
    {
      "epoch": 2.09463545749804,
      "grad_norm": 1.0690101385116577,
      "learning_rate": 9.542368421052631e-05,
      "loss": 0.5079,
      "step": 37406
    },
    {
      "epoch": 2.094691454810169,
      "grad_norm": 1.1750473976135254,
      "learning_rate": 9.542342105263157e-05,
      "loss": 0.4538,
      "step": 37407
    },
    {
      "epoch": 2.094747452122298,
      "grad_norm": 1.2519437074661255,
      "learning_rate": 9.542315789473685e-05,
      "loss": 0.4664,
      "step": 37408
    },
    {
      "epoch": 2.094803449434427,
      "grad_norm": 1.2530477046966553,
      "learning_rate": 9.542289473684211e-05,
      "loss": 0.4613,
      "step": 37409
    },
    {
      "epoch": 2.094859446746556,
      "grad_norm": 1.133747935295105,
      "learning_rate": 9.542263157894738e-05,
      "loss": 0.4216,
      "step": 37410
    },
    {
      "epoch": 2.0949154440586852,
      "grad_norm": 1.041358232498169,
      "learning_rate": 9.542236842105263e-05,
      "loss": 0.3232,
      "step": 37411
    },
    {
      "epoch": 2.0949714413708143,
      "grad_norm": 1.2870550155639648,
      "learning_rate": 9.54221052631579e-05,
      "loss": 0.4394,
      "step": 37412
    },
    {
      "epoch": 2.0950274386829433,
      "grad_norm": 1.2573827505111694,
      "learning_rate": 9.542184210526316e-05,
      "loss": 0.3643,
      "step": 37413
    },
    {
      "epoch": 2.0950834359950723,
      "grad_norm": 1.0182030200958252,
      "learning_rate": 9.542157894736843e-05,
      "loss": 0.413,
      "step": 37414
    },
    {
      "epoch": 2.0951394333072013,
      "grad_norm": 1.5206785202026367,
      "learning_rate": 9.542131578947369e-05,
      "loss": 0.3778,
      "step": 37415
    },
    {
      "epoch": 2.0951954306193303,
      "grad_norm": 1.2366689443588257,
      "learning_rate": 9.542105263157895e-05,
      "loss": 0.3302,
      "step": 37416
    },
    {
      "epoch": 2.0952514279314594,
      "grad_norm": 1.2743369340896606,
      "learning_rate": 9.542078947368421e-05,
      "loss": 0.3246,
      "step": 37417
    },
    {
      "epoch": 2.0953074252435884,
      "grad_norm": 1.2836215496063232,
      "learning_rate": 9.542052631578949e-05,
      "loss": 0.5246,
      "step": 37418
    },
    {
      "epoch": 2.0953634225557174,
      "grad_norm": 1.1655752658843994,
      "learning_rate": 9.542026315789474e-05,
      "loss": 0.407,
      "step": 37419
    },
    {
      "epoch": 2.0954194198678464,
      "grad_norm": 1.2362353801727295,
      "learning_rate": 9.542e-05,
      "loss": 0.3755,
      "step": 37420
    },
    {
      "epoch": 2.0954754171799754,
      "grad_norm": 1.2940351963043213,
      "learning_rate": 9.541973684210526e-05,
      "loss": 0.4432,
      "step": 37421
    },
    {
      "epoch": 2.0955314144921044,
      "grad_norm": 1.0072232484817505,
      "learning_rate": 9.541947368421054e-05,
      "loss": 0.3858,
      "step": 37422
    },
    {
      "epoch": 2.0955874118042335,
      "grad_norm": 1.354242205619812,
      "learning_rate": 9.54192105263158e-05,
      "loss": 0.4157,
      "step": 37423
    },
    {
      "epoch": 2.0956434091163625,
      "grad_norm": 1.4414793252944946,
      "learning_rate": 9.541894736842106e-05,
      "loss": 0.4629,
      "step": 37424
    },
    {
      "epoch": 2.0956994064284915,
      "grad_norm": 1.1099823713302612,
      "learning_rate": 9.541868421052632e-05,
      "loss": 0.3359,
      "step": 37425
    },
    {
      "epoch": 2.0957554037406205,
      "grad_norm": 1.3464715480804443,
      "learning_rate": 9.541842105263158e-05,
      "loss": 0.3824,
      "step": 37426
    },
    {
      "epoch": 2.0958114010527495,
      "grad_norm": 1.2769676446914673,
      "learning_rate": 9.541815789473685e-05,
      "loss": 0.4031,
      "step": 37427
    },
    {
      "epoch": 2.0958673983648786,
      "grad_norm": 1.2056983709335327,
      "learning_rate": 9.541789473684211e-05,
      "loss": 0.3576,
      "step": 37428
    },
    {
      "epoch": 2.0959233956770076,
      "grad_norm": 1.1970973014831543,
      "learning_rate": 9.541763157894737e-05,
      "loss": 0.3541,
      "step": 37429
    },
    {
      "epoch": 2.0959793929891366,
      "grad_norm": 1.2029637098312378,
      "learning_rate": 9.541736842105263e-05,
      "loss": 0.4307,
      "step": 37430
    },
    {
      "epoch": 2.0960353903012656,
      "grad_norm": 1.3212459087371826,
      "learning_rate": 9.54171052631579e-05,
      "loss": 0.3502,
      "step": 37431
    },
    {
      "epoch": 2.0960913876133946,
      "grad_norm": 1.5098742246627808,
      "learning_rate": 9.541684210526316e-05,
      "loss": 0.3903,
      "step": 37432
    },
    {
      "epoch": 2.0961473849255237,
      "grad_norm": 1.0971882343292236,
      "learning_rate": 9.541657894736844e-05,
      "loss": 0.3606,
      "step": 37433
    },
    {
      "epoch": 2.0962033822376527,
      "grad_norm": 1.0959628820419312,
      "learning_rate": 9.541631578947368e-05,
      "loss": 0.3228,
      "step": 37434
    },
    {
      "epoch": 2.0962593795497817,
      "grad_norm": 1.2527858018875122,
      "learning_rate": 9.541605263157895e-05,
      "loss": 0.3596,
      "step": 37435
    },
    {
      "epoch": 2.0963153768619107,
      "grad_norm": 1.2567248344421387,
      "learning_rate": 9.541578947368421e-05,
      "loss": 0.4379,
      "step": 37436
    },
    {
      "epoch": 2.0963713741740397,
      "grad_norm": 1.4920306205749512,
      "learning_rate": 9.541552631578949e-05,
      "loss": 0.5387,
      "step": 37437
    },
    {
      "epoch": 2.0964273714861688,
      "grad_norm": 1.2100061178207397,
      "learning_rate": 9.541526315789475e-05,
      "loss": 0.3701,
      "step": 37438
    },
    {
      "epoch": 2.0964833687982978,
      "grad_norm": 1.143375039100647,
      "learning_rate": 9.541500000000001e-05,
      "loss": 0.4088,
      "step": 37439
    },
    {
      "epoch": 2.096539366110427,
      "grad_norm": 1.1372005939483643,
      "learning_rate": 9.541473684210527e-05,
      "loss": 0.3769,
      "step": 37440
    },
    {
      "epoch": 2.096595363422556,
      "grad_norm": 1.1026577949523926,
      "learning_rate": 9.541447368421053e-05,
      "loss": 0.4592,
      "step": 37441
    },
    {
      "epoch": 2.096651360734685,
      "grad_norm": 1.075384259223938,
      "learning_rate": 9.54142105263158e-05,
      "loss": 0.4273,
      "step": 37442
    },
    {
      "epoch": 2.096707358046814,
      "grad_norm": 1.0644259452819824,
      "learning_rate": 9.541394736842105e-05,
      "loss": 0.3473,
      "step": 37443
    },
    {
      "epoch": 2.096763355358943,
      "grad_norm": 1.162556529045105,
      "learning_rate": 9.541368421052632e-05,
      "loss": 0.3998,
      "step": 37444
    },
    {
      "epoch": 2.096819352671072,
      "grad_norm": 1.1349822282791138,
      "learning_rate": 9.541342105263158e-05,
      "loss": 0.315,
      "step": 37445
    },
    {
      "epoch": 2.096875349983201,
      "grad_norm": 0.964417040348053,
      "learning_rate": 9.541315789473685e-05,
      "loss": 0.3121,
      "step": 37446
    },
    {
      "epoch": 2.09693134729533,
      "grad_norm": 1.3284751176834106,
      "learning_rate": 9.541289473684211e-05,
      "loss": 0.4745,
      "step": 37447
    },
    {
      "epoch": 2.096987344607459,
      "grad_norm": 1.365309238433838,
      "learning_rate": 9.541263157894737e-05,
      "loss": 0.4909,
      "step": 37448
    },
    {
      "epoch": 2.097043341919588,
      "grad_norm": 1.2811161279678345,
      "learning_rate": 9.541236842105263e-05,
      "loss": 0.4636,
      "step": 37449
    },
    {
      "epoch": 2.097099339231717,
      "grad_norm": 1.3605387210845947,
      "learning_rate": 9.54121052631579e-05,
      "loss": 0.4437,
      "step": 37450
    },
    {
      "epoch": 2.097155336543846,
      "grad_norm": 1.6411168575286865,
      "learning_rate": 9.541184210526316e-05,
      "loss": 0.3933,
      "step": 37451
    },
    {
      "epoch": 2.097211333855975,
      "grad_norm": 1.4493459463119507,
      "learning_rate": 9.541157894736842e-05,
      "loss": 0.4003,
      "step": 37452
    },
    {
      "epoch": 2.097267331168104,
      "grad_norm": 1.049507737159729,
      "learning_rate": 9.541131578947368e-05,
      "loss": 0.4041,
      "step": 37453
    },
    {
      "epoch": 2.097323328480233,
      "grad_norm": 1.447320818901062,
      "learning_rate": 9.541105263157896e-05,
      "loss": 0.5007,
      "step": 37454
    },
    {
      "epoch": 2.097379325792362,
      "grad_norm": 1.0521488189697266,
      "learning_rate": 9.541078947368422e-05,
      "loss": 0.4576,
      "step": 37455
    },
    {
      "epoch": 2.097435323104491,
      "grad_norm": 1.0134435892105103,
      "learning_rate": 9.541052631578948e-05,
      "loss": 0.365,
      "step": 37456
    },
    {
      "epoch": 2.09749132041662,
      "grad_norm": 1.0665197372436523,
      "learning_rate": 9.541026315789474e-05,
      "loss": 0.3322,
      "step": 37457
    },
    {
      "epoch": 2.097547317728749,
      "grad_norm": 1.5423494577407837,
      "learning_rate": 9.541e-05,
      "loss": 0.5235,
      "step": 37458
    },
    {
      "epoch": 2.097603315040878,
      "grad_norm": 1.1237189769744873,
      "learning_rate": 9.540973684210527e-05,
      "loss": 0.3981,
      "step": 37459
    },
    {
      "epoch": 2.097659312353007,
      "grad_norm": 1.0664355754852295,
      "learning_rate": 9.540947368421053e-05,
      "loss": 0.3673,
      "step": 37460
    },
    {
      "epoch": 2.097715309665136,
      "grad_norm": 1.0892367362976074,
      "learning_rate": 9.540921052631579e-05,
      "loss": 0.433,
      "step": 37461
    },
    {
      "epoch": 2.097771306977265,
      "grad_norm": 1.2048622369766235,
      "learning_rate": 9.540894736842105e-05,
      "loss": 0.4602,
      "step": 37462
    },
    {
      "epoch": 2.0978273042893942,
      "grad_norm": 1.3197017908096313,
      "learning_rate": 9.540868421052632e-05,
      "loss": 0.5158,
      "step": 37463
    },
    {
      "epoch": 2.0978833016015233,
      "grad_norm": 1.4500712156295776,
      "learning_rate": 9.540842105263158e-05,
      "loss": 0.4806,
      "step": 37464
    },
    {
      "epoch": 2.0979392989136523,
      "grad_norm": 1.0754746198654175,
      "learning_rate": 9.540815789473685e-05,
      "loss": 0.3819,
      "step": 37465
    },
    {
      "epoch": 2.0979952962257813,
      "grad_norm": 1.3551568984985352,
      "learning_rate": 9.54078947368421e-05,
      "loss": 0.4861,
      "step": 37466
    },
    {
      "epoch": 2.0980512935379103,
      "grad_norm": 1.2907880544662476,
      "learning_rate": 9.540763157894737e-05,
      "loss": 0.4613,
      "step": 37467
    },
    {
      "epoch": 2.0981072908500393,
      "grad_norm": 1.043502926826477,
      "learning_rate": 9.540736842105263e-05,
      "loss": 0.4111,
      "step": 37468
    },
    {
      "epoch": 2.0981632881621683,
      "grad_norm": 1.3776546716690063,
      "learning_rate": 9.540710526315791e-05,
      "loss": 0.4498,
      "step": 37469
    },
    {
      "epoch": 2.0982192854742974,
      "grad_norm": 1.6057136058807373,
      "learning_rate": 9.540684210526317e-05,
      "loss": 0.4522,
      "step": 37470
    },
    {
      "epoch": 2.0982752827864264,
      "grad_norm": 1.2646669149398804,
      "learning_rate": 9.540657894736843e-05,
      "loss": 0.4493,
      "step": 37471
    },
    {
      "epoch": 2.0983312800985554,
      "grad_norm": 1.35685133934021,
      "learning_rate": 9.540631578947369e-05,
      "loss": 0.4374,
      "step": 37472
    },
    {
      "epoch": 2.0983872774106844,
      "grad_norm": 1.2141730785369873,
      "learning_rate": 9.540605263157896e-05,
      "loss": 0.336,
      "step": 37473
    },
    {
      "epoch": 2.0984432747228134,
      "grad_norm": 0.959006130695343,
      "learning_rate": 9.540578947368422e-05,
      "loss": 0.3191,
      "step": 37474
    },
    {
      "epoch": 2.0984992720349425,
      "grad_norm": 1.154911756515503,
      "learning_rate": 9.540552631578948e-05,
      "loss": 0.4295,
      "step": 37475
    },
    {
      "epoch": 2.0985552693470715,
      "grad_norm": 1.1139824390411377,
      "learning_rate": 9.540526315789474e-05,
      "loss": 0.3457,
      "step": 37476
    },
    {
      "epoch": 2.0986112666592005,
      "grad_norm": 1.1385635137557983,
      "learning_rate": 9.5405e-05,
      "loss": 0.4435,
      "step": 37477
    },
    {
      "epoch": 2.0986672639713295,
      "grad_norm": 1.2122125625610352,
      "learning_rate": 9.540473684210527e-05,
      "loss": 0.4658,
      "step": 37478
    },
    {
      "epoch": 2.0987232612834585,
      "grad_norm": 1.038793921470642,
      "learning_rate": 9.540447368421053e-05,
      "loss": 0.3161,
      "step": 37479
    },
    {
      "epoch": 2.0987792585955876,
      "grad_norm": 1.1674559116363525,
      "learning_rate": 9.540421052631579e-05,
      "loss": 0.3696,
      "step": 37480
    },
    {
      "epoch": 2.0988352559077166,
      "grad_norm": 1.4023323059082031,
      "learning_rate": 9.540394736842105e-05,
      "loss": 0.5009,
      "step": 37481
    },
    {
      "epoch": 2.0988912532198456,
      "grad_norm": 1.4067128896713257,
      "learning_rate": 9.540368421052632e-05,
      "loss": 0.3737,
      "step": 37482
    },
    {
      "epoch": 2.0989472505319746,
      "grad_norm": 1.057881474494934,
      "learning_rate": 9.540342105263158e-05,
      "loss": 0.4268,
      "step": 37483
    },
    {
      "epoch": 2.0990032478441036,
      "grad_norm": Infinity,
      "learning_rate": 9.540342105263158e-05,
      "loss": 0.5112,
      "step": 37484
    },
    {
      "epoch": 2.0990592451562327,
      "grad_norm": 1.0927484035491943,
      "learning_rate": 9.540315789473684e-05,
      "loss": 0.3716,
      "step": 37485
    },
    {
      "epoch": 2.0991152424683617,
      "grad_norm": 1.2988111972808838,
      "learning_rate": 9.54028947368421e-05,
      "loss": 0.4031,
      "step": 37486
    },
    {
      "epoch": 2.0991712397804907,
      "grad_norm": 1.22554349899292,
      "learning_rate": 9.540263157894738e-05,
      "loss": 0.3964,
      "step": 37487
    },
    {
      "epoch": 2.0992272370926197,
      "grad_norm": 1.038845181465149,
      "learning_rate": 9.540236842105264e-05,
      "loss": 0.4681,
      "step": 37488
    },
    {
      "epoch": 2.0992832344047487,
      "grad_norm": 1.2106190919876099,
      "learning_rate": 9.540210526315791e-05,
      "loss": 0.3466,
      "step": 37489
    },
    {
      "epoch": 2.0993392317168778,
      "grad_norm": 1.1918106079101562,
      "learning_rate": 9.540184210526316e-05,
      "loss": 0.3871,
      "step": 37490
    },
    {
      "epoch": 2.0993952290290068,
      "grad_norm": 1.5124584436416626,
      "learning_rate": 9.540157894736843e-05,
      "loss": 0.5388,
      "step": 37491
    },
    {
      "epoch": 2.099451226341136,
      "grad_norm": 1.2367541790008545,
      "learning_rate": 9.540131578947369e-05,
      "loss": 0.4239,
      "step": 37492
    },
    {
      "epoch": 2.099507223653265,
      "grad_norm": 1.241233468055725,
      "learning_rate": 9.540105263157895e-05,
      "loss": 0.4121,
      "step": 37493
    },
    {
      "epoch": 2.099563220965394,
      "grad_norm": 1.0106979608535767,
      "learning_rate": 9.540078947368422e-05,
      "loss": 0.3556,
      "step": 37494
    },
    {
      "epoch": 2.099619218277523,
      "grad_norm": 1.0835684537887573,
      "learning_rate": 9.540052631578947e-05,
      "loss": 0.371,
      "step": 37495
    },
    {
      "epoch": 2.099675215589652,
      "grad_norm": 1.2277915477752686,
      "learning_rate": 9.540026315789474e-05,
      "loss": 0.3772,
      "step": 37496
    },
    {
      "epoch": 2.099731212901781,
      "grad_norm": 1.3299797773361206,
      "learning_rate": 9.54e-05,
      "loss": 0.3928,
      "step": 37497
    },
    {
      "epoch": 2.09978721021391,
      "grad_norm": 1.2439411878585815,
      "learning_rate": 9.539973684210527e-05,
      "loss": 0.3833,
      "step": 37498
    },
    {
      "epoch": 2.099843207526039,
      "grad_norm": 1.1844677925109863,
      "learning_rate": 9.539947368421052e-05,
      "loss": 0.4501,
      "step": 37499
    },
    {
      "epoch": 2.099899204838168,
      "grad_norm": 1.5768829584121704,
      "learning_rate": 9.53992105263158e-05,
      "loss": 0.3532,
      "step": 37500
    },
    {
      "epoch": 2.099955202150297,
      "grad_norm": 1.0400227308273315,
      "learning_rate": 9.539894736842105e-05,
      "loss": 0.3932,
      "step": 37501
    },
    {
      "epoch": 2.100011199462426,
      "grad_norm": 1.2879549264907837,
      "learning_rate": 9.539868421052633e-05,
      "loss": 0.3829,
      "step": 37502
    },
    {
      "epoch": 2.100067196774555,
      "grad_norm": 1.179744005203247,
      "learning_rate": 9.539842105263159e-05,
      "loss": 0.3992,
      "step": 37503
    },
    {
      "epoch": 2.100123194086684,
      "grad_norm": 1.1425442695617676,
      "learning_rate": 9.539815789473685e-05,
      "loss": 0.4249,
      "step": 37504
    },
    {
      "epoch": 2.100179191398813,
      "grad_norm": 1.0603622198104858,
      "learning_rate": 9.53978947368421e-05,
      "loss": 0.311,
      "step": 37505
    },
    {
      "epoch": 2.100235188710942,
      "grad_norm": 1.3630765676498413,
      "learning_rate": 9.539763157894738e-05,
      "loss": 0.3658,
      "step": 37506
    },
    {
      "epoch": 2.100291186023071,
      "grad_norm": 1.2215807437896729,
      "learning_rate": 9.539736842105264e-05,
      "loss": 0.5665,
      "step": 37507
    },
    {
      "epoch": 2.1003471833352,
      "grad_norm": 1.2204254865646362,
      "learning_rate": 9.53971052631579e-05,
      "loss": 0.365,
      "step": 37508
    },
    {
      "epoch": 2.100403180647329,
      "grad_norm": 1.1640019416809082,
      "learning_rate": 9.539684210526316e-05,
      "loss": 0.5645,
      "step": 37509
    },
    {
      "epoch": 2.100459177959458,
      "grad_norm": 1.2654796838760376,
      "learning_rate": 9.539657894736842e-05,
      "loss": 0.4639,
      "step": 37510
    },
    {
      "epoch": 2.100515175271587,
      "grad_norm": 1.2631561756134033,
      "learning_rate": 9.539631578947369e-05,
      "loss": 0.4163,
      "step": 37511
    },
    {
      "epoch": 2.100571172583716,
      "grad_norm": 1.1944363117218018,
      "learning_rate": 9.539605263157895e-05,
      "loss": 0.4792,
      "step": 37512
    },
    {
      "epoch": 2.100627169895845,
      "grad_norm": 1.024147629737854,
      "learning_rate": 9.539578947368421e-05,
      "loss": 0.3747,
      "step": 37513
    },
    {
      "epoch": 2.100683167207974,
      "grad_norm": 1.2537106275558472,
      "learning_rate": 9.539552631578947e-05,
      "loss": 0.427,
      "step": 37514
    },
    {
      "epoch": 2.1007391645201032,
      "grad_norm": 1.1115913391113281,
      "learning_rate": 9.539526315789474e-05,
      "loss": 0.416,
      "step": 37515
    },
    {
      "epoch": 2.1007951618322322,
      "grad_norm": 1.4281648397445679,
      "learning_rate": 9.5395e-05,
      "loss": 0.4033,
      "step": 37516
    },
    {
      "epoch": 2.1008511591443613,
      "grad_norm": 1.257993459701538,
      "learning_rate": 9.539473684210526e-05,
      "loss": 0.3923,
      "step": 37517
    },
    {
      "epoch": 2.1009071564564903,
      "grad_norm": 1.2031153440475464,
      "learning_rate": 9.539447368421052e-05,
      "loss": 0.3821,
      "step": 37518
    },
    {
      "epoch": 2.1009631537686193,
      "grad_norm": 1.0563081502914429,
      "learning_rate": 9.53942105263158e-05,
      "loss": 0.4296,
      "step": 37519
    },
    {
      "epoch": 2.1010191510807483,
      "grad_norm": 1.3949379920959473,
      "learning_rate": 9.539394736842106e-05,
      "loss": 0.4826,
      "step": 37520
    },
    {
      "epoch": 2.1010751483928773,
      "grad_norm": 1.1932613849639893,
      "learning_rate": 9.539368421052633e-05,
      "loss": 0.3783,
      "step": 37521
    },
    {
      "epoch": 2.1011311457050064,
      "grad_norm": 1.34486985206604,
      "learning_rate": 9.539342105263158e-05,
      "loss": 0.4214,
      "step": 37522
    },
    {
      "epoch": 2.1011871430171354,
      "grad_norm": 1.2657222747802734,
      "learning_rate": 9.539315789473685e-05,
      "loss": 0.4101,
      "step": 37523
    },
    {
      "epoch": 2.1012431403292644,
      "grad_norm": 1.351082444190979,
      "learning_rate": 9.539289473684211e-05,
      "loss": 0.3538,
      "step": 37524
    },
    {
      "epoch": 2.1012991376413934,
      "grad_norm": 1.369026780128479,
      "learning_rate": 9.539263157894738e-05,
      "loss": 0.4197,
      "step": 37525
    },
    {
      "epoch": 2.1013551349535224,
      "grad_norm": 1.1988039016723633,
      "learning_rate": 9.539236842105264e-05,
      "loss": 0.4993,
      "step": 37526
    },
    {
      "epoch": 2.1014111322656515,
      "grad_norm": 1.323513388633728,
      "learning_rate": 9.53921052631579e-05,
      "loss": 0.4617,
      "step": 37527
    },
    {
      "epoch": 2.1014671295777805,
      "grad_norm": 1.4121928215026855,
      "learning_rate": 9.539184210526316e-05,
      "loss": 0.3061,
      "step": 37528
    },
    {
      "epoch": 2.1015231268899095,
      "grad_norm": 1.47354257106781,
      "learning_rate": 9.539157894736842e-05,
      "loss": 0.5402,
      "step": 37529
    },
    {
      "epoch": 2.101579124202038,
      "grad_norm": 1.6160423755645752,
      "learning_rate": 9.53913157894737e-05,
      "loss": 0.3643,
      "step": 37530
    },
    {
      "epoch": 2.1016351215141675,
      "grad_norm": 1.4825937747955322,
      "learning_rate": 9.539105263157895e-05,
      "loss": 0.5207,
      "step": 37531
    },
    {
      "epoch": 2.101691118826296,
      "grad_norm": 1.3338812589645386,
      "learning_rate": 9.539078947368421e-05,
      "loss": 0.3811,
      "step": 37532
    },
    {
      "epoch": 2.1017471161384256,
      "grad_norm": 1.099802851676941,
      "learning_rate": 9.539052631578947e-05,
      "loss": 0.4325,
      "step": 37533
    },
    {
      "epoch": 2.101803113450554,
      "grad_norm": 1.2754861116409302,
      "learning_rate": 9.539026315789475e-05,
      "loss": 0.4251,
      "step": 37534
    },
    {
      "epoch": 2.1018591107626836,
      "grad_norm": 1.2157913446426392,
      "learning_rate": 9.539e-05,
      "loss": 0.3712,
      "step": 37535
    },
    {
      "epoch": 2.101915108074812,
      "grad_norm": 2.3159897327423096,
      "learning_rate": 9.538973684210527e-05,
      "loss": 0.3095,
      "step": 37536
    },
    {
      "epoch": 2.1019711053869417,
      "grad_norm": 1.369210124015808,
      "learning_rate": 9.538947368421053e-05,
      "loss": 0.3814,
      "step": 37537
    },
    {
      "epoch": 2.1020271026990702,
      "grad_norm": 1.0498032569885254,
      "learning_rate": 9.53892105263158e-05,
      "loss": 0.2957,
      "step": 37538
    },
    {
      "epoch": 2.1020831000111997,
      "grad_norm": 1.3048136234283447,
      "learning_rate": 9.538894736842106e-05,
      "loss": 0.4662,
      "step": 37539
    },
    {
      "epoch": 2.1021390973233283,
      "grad_norm": 1.214530348777771,
      "learning_rate": 9.538868421052632e-05,
      "loss": 0.4265,
      "step": 37540
    },
    {
      "epoch": 2.1021950946354573,
      "grad_norm": 1.177441120147705,
      "learning_rate": 9.538842105263158e-05,
      "loss": 0.3527,
      "step": 37541
    },
    {
      "epoch": 2.1022510919475863,
      "grad_norm": 1.0436906814575195,
      "learning_rate": 9.538815789473685e-05,
      "loss": 0.3506,
      "step": 37542
    },
    {
      "epoch": 2.1023070892597153,
      "grad_norm": 1.4231051206588745,
      "learning_rate": 9.538789473684211e-05,
      "loss": 0.6043,
      "step": 37543
    },
    {
      "epoch": 2.1023630865718443,
      "grad_norm": 1.1546419858932495,
      "learning_rate": 9.538763157894738e-05,
      "loss": 0.3679,
      "step": 37544
    },
    {
      "epoch": 2.1024190838839734,
      "grad_norm": 1.2269538640975952,
      "learning_rate": 9.538736842105263e-05,
      "loss": 0.4475,
      "step": 37545
    },
    {
      "epoch": 2.1024750811961024,
      "grad_norm": 1.3043169975280762,
      "learning_rate": 9.538710526315789e-05,
      "loss": 0.4066,
      "step": 37546
    },
    {
      "epoch": 2.1025310785082314,
      "grad_norm": 1.1699858903884888,
      "learning_rate": 9.538684210526316e-05,
      "loss": 0.4026,
      "step": 37547
    },
    {
      "epoch": 2.1025870758203604,
      "grad_norm": 1.221286654472351,
      "learning_rate": 9.538657894736842e-05,
      "loss": 0.4903,
      "step": 37548
    },
    {
      "epoch": 2.1026430731324894,
      "grad_norm": 1.325402021408081,
      "learning_rate": 9.53863157894737e-05,
      "loss": 0.4644,
      "step": 37549
    },
    {
      "epoch": 2.1026990704446185,
      "grad_norm": 1.1518927812576294,
      "learning_rate": 9.538605263157894e-05,
      "loss": 0.3624,
      "step": 37550
    },
    {
      "epoch": 2.1027550677567475,
      "grad_norm": 1.0352146625518799,
      "learning_rate": 9.538578947368422e-05,
      "loss": 0.3069,
      "step": 37551
    },
    {
      "epoch": 2.1028110650688765,
      "grad_norm": 1.0786614418029785,
      "learning_rate": 9.538552631578948e-05,
      "loss": 0.368,
      "step": 37552
    },
    {
      "epoch": 2.1028670623810055,
      "grad_norm": 1.1645514965057373,
      "learning_rate": 9.538526315789475e-05,
      "loss": 0.4692,
      "step": 37553
    },
    {
      "epoch": 2.1029230596931345,
      "grad_norm": 1.149208903312683,
      "learning_rate": 9.5385e-05,
      "loss": 0.474,
      "step": 37554
    },
    {
      "epoch": 2.1029790570052636,
      "grad_norm": 1.4023059606552124,
      "learning_rate": 9.538473684210527e-05,
      "loss": 0.4279,
      "step": 37555
    },
    {
      "epoch": 2.1030350543173926,
      "grad_norm": 1.1420704126358032,
      "learning_rate": 9.538447368421053e-05,
      "loss": 0.4171,
      "step": 37556
    },
    {
      "epoch": 2.1030910516295216,
      "grad_norm": 1.4176591634750366,
      "learning_rate": 9.53842105263158e-05,
      "loss": 0.4781,
      "step": 37557
    },
    {
      "epoch": 2.1031470489416506,
      "grad_norm": 1.1842955350875854,
      "learning_rate": 9.538394736842106e-05,
      "loss": 0.3584,
      "step": 37558
    },
    {
      "epoch": 2.1032030462537796,
      "grad_norm": 1.2433788776397705,
      "learning_rate": 9.538368421052632e-05,
      "loss": 0.383,
      "step": 37559
    },
    {
      "epoch": 2.1032590435659086,
      "grad_norm": 1.3103694915771484,
      "learning_rate": 9.538342105263158e-05,
      "loss": 0.4348,
      "step": 37560
    },
    {
      "epoch": 2.1033150408780377,
      "grad_norm": 1.1505239009857178,
      "learning_rate": 9.538315789473685e-05,
      "loss": 0.3615,
      "step": 37561
    },
    {
      "epoch": 2.1033710381901667,
      "grad_norm": 1.1136572360992432,
      "learning_rate": 9.538289473684211e-05,
      "loss": 0.3285,
      "step": 37562
    },
    {
      "epoch": 2.1034270355022957,
      "grad_norm": 1.0530961751937866,
      "learning_rate": 9.538263157894737e-05,
      "loss": 0.3297,
      "step": 37563
    },
    {
      "epoch": 2.1034830328144247,
      "grad_norm": 1.1275643110275269,
      "learning_rate": 9.538236842105263e-05,
      "loss": 0.3673,
      "step": 37564
    },
    {
      "epoch": 2.1035390301265537,
      "grad_norm": 1.1657040119171143,
      "learning_rate": 9.53821052631579e-05,
      "loss": 0.4207,
      "step": 37565
    },
    {
      "epoch": 2.1035950274386828,
      "grad_norm": 1.2637416124343872,
      "learning_rate": 9.538184210526317e-05,
      "loss": 0.3207,
      "step": 37566
    },
    {
      "epoch": 2.103651024750812,
      "grad_norm": 1.205443263053894,
      "learning_rate": 9.538157894736843e-05,
      "loss": 0.3905,
      "step": 37567
    },
    {
      "epoch": 2.103707022062941,
      "grad_norm": 1.214951992034912,
      "learning_rate": 9.538131578947369e-05,
      "loss": 0.4477,
      "step": 37568
    },
    {
      "epoch": 2.10376301937507,
      "grad_norm": 1.2780406475067139,
      "learning_rate": 9.538105263157895e-05,
      "loss": 0.3984,
      "step": 37569
    },
    {
      "epoch": 2.103819016687199,
      "grad_norm": 1.2695943117141724,
      "learning_rate": 9.538078947368422e-05,
      "loss": 0.3483,
      "step": 37570
    },
    {
      "epoch": 2.103875013999328,
      "grad_norm": 1.2124972343444824,
      "learning_rate": 9.538052631578948e-05,
      "loss": 0.3902,
      "step": 37571
    },
    {
      "epoch": 2.103931011311457,
      "grad_norm": 1.0899230241775513,
      "learning_rate": 9.538026315789474e-05,
      "loss": 0.3797,
      "step": 37572
    },
    {
      "epoch": 2.103987008623586,
      "grad_norm": 1.3242824077606201,
      "learning_rate": 9.538e-05,
      "loss": 0.3386,
      "step": 37573
    },
    {
      "epoch": 2.104043005935715,
      "grad_norm": 1.321236491203308,
      "learning_rate": 9.537973684210527e-05,
      "loss": 0.3262,
      "step": 37574
    },
    {
      "epoch": 2.104099003247844,
      "grad_norm": 1.4211359024047852,
      "learning_rate": 9.537947368421053e-05,
      "loss": 0.3566,
      "step": 37575
    },
    {
      "epoch": 2.104155000559973,
      "grad_norm": 1.4502272605895996,
      "learning_rate": 9.53792105263158e-05,
      "loss": 0.4676,
      "step": 37576
    },
    {
      "epoch": 2.104210997872102,
      "grad_norm": 1.1795663833618164,
      "learning_rate": 9.537894736842105e-05,
      "loss": 0.4295,
      "step": 37577
    },
    {
      "epoch": 2.104266995184231,
      "grad_norm": 1.2600855827331543,
      "learning_rate": 9.537868421052632e-05,
      "loss": 0.4081,
      "step": 37578
    },
    {
      "epoch": 2.10432299249636,
      "grad_norm": 1.0279121398925781,
      "learning_rate": 9.537842105263158e-05,
      "loss": 0.4388,
      "step": 37579
    },
    {
      "epoch": 2.104378989808489,
      "grad_norm": 1.7014554738998413,
      "learning_rate": 9.537815789473684e-05,
      "loss": 0.4917,
      "step": 37580
    },
    {
      "epoch": 2.104434987120618,
      "grad_norm": 1.2564527988433838,
      "learning_rate": 9.537789473684212e-05,
      "loss": 0.4139,
      "step": 37581
    },
    {
      "epoch": 2.104490984432747,
      "grad_norm": 1.1442245244979858,
      "learning_rate": 9.537763157894736e-05,
      "loss": 0.4013,
      "step": 37582
    },
    {
      "epoch": 2.104546981744876,
      "grad_norm": 1.162278652191162,
      "learning_rate": 9.537736842105264e-05,
      "loss": 0.5206,
      "step": 37583
    },
    {
      "epoch": 2.104602979057005,
      "grad_norm": 2.273815155029297,
      "learning_rate": 9.53771052631579e-05,
      "loss": 0.439,
      "step": 37584
    },
    {
      "epoch": 2.104658976369134,
      "grad_norm": 1.1859922409057617,
      "learning_rate": 9.537684210526317e-05,
      "loss": 0.4164,
      "step": 37585
    },
    {
      "epoch": 2.104714973681263,
      "grad_norm": 1.170497179031372,
      "learning_rate": 9.537657894736843e-05,
      "loss": 0.3093,
      "step": 37586
    },
    {
      "epoch": 2.104770970993392,
      "grad_norm": 1.0040929317474365,
      "learning_rate": 9.537631578947369e-05,
      "loss": 0.4037,
      "step": 37587
    },
    {
      "epoch": 2.104826968305521,
      "grad_norm": 1.472621202468872,
      "learning_rate": 9.537605263157895e-05,
      "loss": 0.3692,
      "step": 37588
    },
    {
      "epoch": 2.10488296561765,
      "grad_norm": 1.2446300983428955,
      "learning_rate": 9.537578947368422e-05,
      "loss": 0.5026,
      "step": 37589
    },
    {
      "epoch": 2.1049389629297792,
      "grad_norm": 1.1514365673065186,
      "learning_rate": 9.537552631578948e-05,
      "loss": 0.4952,
      "step": 37590
    },
    {
      "epoch": 2.1049949602419082,
      "grad_norm": 1.3575093746185303,
      "learning_rate": 9.537526315789474e-05,
      "loss": 0.4979,
      "step": 37591
    },
    {
      "epoch": 2.1050509575540373,
      "grad_norm": 1.142269253730774,
      "learning_rate": 9.5375e-05,
      "loss": 0.3675,
      "step": 37592
    },
    {
      "epoch": 2.1051069548661663,
      "grad_norm": 1.1615089178085327,
      "learning_rate": 9.537473684210527e-05,
      "loss": 0.4012,
      "step": 37593
    },
    {
      "epoch": 2.1051629521782953,
      "grad_norm": 1.3611550331115723,
      "learning_rate": 9.537447368421053e-05,
      "loss": 0.3028,
      "step": 37594
    },
    {
      "epoch": 2.1052189494904243,
      "grad_norm": 1.0073506832122803,
      "learning_rate": 9.53742105263158e-05,
      "loss": 0.3443,
      "step": 37595
    },
    {
      "epoch": 2.1052749468025533,
      "grad_norm": 1.3089855909347534,
      "learning_rate": 9.537394736842105e-05,
      "loss": 0.4662,
      "step": 37596
    },
    {
      "epoch": 2.1053309441146824,
      "grad_norm": 1.0892695188522339,
      "learning_rate": 9.537368421052631e-05,
      "loss": 0.3493,
      "step": 37597
    },
    {
      "epoch": 2.1053869414268114,
      "grad_norm": 1.1969753503799438,
      "learning_rate": 9.537342105263159e-05,
      "loss": 0.4227,
      "step": 37598
    },
    {
      "epoch": 2.1054429387389404,
      "grad_norm": 1.1530171632766724,
      "learning_rate": 9.537315789473685e-05,
      "loss": 0.3752,
      "step": 37599
    },
    {
      "epoch": 2.1054989360510694,
      "grad_norm": 1.081833004951477,
      "learning_rate": 9.53728947368421e-05,
      "loss": 0.4024,
      "step": 37600
    },
    {
      "epoch": 2.1055549333631984,
      "grad_norm": 1.1977484226226807,
      "learning_rate": 9.537263157894737e-05,
      "loss": 0.5045,
      "step": 37601
    },
    {
      "epoch": 2.1056109306753275,
      "grad_norm": 1.2235177755355835,
      "learning_rate": 9.537236842105264e-05,
      "loss": 0.4183,
      "step": 37602
    },
    {
      "epoch": 2.1056669279874565,
      "grad_norm": 1.6584830284118652,
      "learning_rate": 9.53721052631579e-05,
      "loss": 0.492,
      "step": 37603
    },
    {
      "epoch": 2.1057229252995855,
      "grad_norm": 1.2708359956741333,
      "learning_rate": 9.537184210526317e-05,
      "loss": 0.5033,
      "step": 37604
    },
    {
      "epoch": 2.1057789226117145,
      "grad_norm": 1.1725993156433105,
      "learning_rate": 9.537157894736842e-05,
      "loss": 0.423,
      "step": 37605
    },
    {
      "epoch": 2.1058349199238435,
      "grad_norm": 1.2848418951034546,
      "learning_rate": 9.537131578947369e-05,
      "loss": 0.4571,
      "step": 37606
    },
    {
      "epoch": 2.1058909172359725,
      "grad_norm": 1.1580767631530762,
      "learning_rate": 9.537105263157895e-05,
      "loss": 0.3126,
      "step": 37607
    },
    {
      "epoch": 2.1059469145481016,
      "grad_norm": 1.1102182865142822,
      "learning_rate": 9.537078947368422e-05,
      "loss": 0.3116,
      "step": 37608
    },
    {
      "epoch": 2.1060029118602306,
      "grad_norm": 1.4429696798324585,
      "learning_rate": 9.537052631578947e-05,
      "loss": 0.4199,
      "step": 37609
    },
    {
      "epoch": 2.1060589091723596,
      "grad_norm": 1.2626240253448486,
      "learning_rate": 9.537026315789474e-05,
      "loss": 0.4992,
      "step": 37610
    },
    {
      "epoch": 2.1061149064844886,
      "grad_norm": 1.210782527923584,
      "learning_rate": 9.537e-05,
      "loss": 0.3782,
      "step": 37611
    },
    {
      "epoch": 2.1061709037966176,
      "grad_norm": 1.3715052604675293,
      "learning_rate": 9.536973684210528e-05,
      "loss": 0.4497,
      "step": 37612
    },
    {
      "epoch": 2.1062269011087467,
      "grad_norm": 1.0635639429092407,
      "learning_rate": 9.536947368421054e-05,
      "loss": 0.4196,
      "step": 37613
    },
    {
      "epoch": 2.1062828984208757,
      "grad_norm": 1.2242220640182495,
      "learning_rate": 9.536921052631578e-05,
      "loss": 0.4658,
      "step": 37614
    },
    {
      "epoch": 2.1063388957330047,
      "grad_norm": 1.2983136177062988,
      "learning_rate": 9.536894736842106e-05,
      "loss": 0.3303,
      "step": 37615
    },
    {
      "epoch": 2.1063948930451337,
      "grad_norm": 1.2525765895843506,
      "learning_rate": 9.536868421052632e-05,
      "loss": 0.4423,
      "step": 37616
    },
    {
      "epoch": 2.1064508903572627,
      "grad_norm": 1.5390067100524902,
      "learning_rate": 9.536842105263159e-05,
      "loss": 0.4236,
      "step": 37617
    },
    {
      "epoch": 2.1065068876693918,
      "grad_norm": 1.3755855560302734,
      "learning_rate": 9.536815789473685e-05,
      "loss": 0.4819,
      "step": 37618
    },
    {
      "epoch": 2.1065628849815208,
      "grad_norm": 1.3707313537597656,
      "learning_rate": 9.536789473684211e-05,
      "loss": 0.4312,
      "step": 37619
    },
    {
      "epoch": 2.10661888229365,
      "grad_norm": 1.4791666269302368,
      "learning_rate": 9.536763157894737e-05,
      "loss": 0.4804,
      "step": 37620
    },
    {
      "epoch": 2.106674879605779,
      "grad_norm": 1.3952860832214355,
      "learning_rate": 9.536736842105264e-05,
      "loss": 0.3225,
      "step": 37621
    },
    {
      "epoch": 2.106730876917908,
      "grad_norm": 1.162414312362671,
      "learning_rate": 9.53671052631579e-05,
      "loss": 0.3696,
      "step": 37622
    },
    {
      "epoch": 2.106786874230037,
      "grad_norm": 1.1685415506362915,
      "learning_rate": 9.536684210526316e-05,
      "loss": 0.4196,
      "step": 37623
    },
    {
      "epoch": 2.106842871542166,
      "grad_norm": 1.2862279415130615,
      "learning_rate": 9.536657894736842e-05,
      "loss": 0.355,
      "step": 37624
    },
    {
      "epoch": 2.106898868854295,
      "grad_norm": 1.0106096267700195,
      "learning_rate": 9.53663157894737e-05,
      "loss": 0.2846,
      "step": 37625
    },
    {
      "epoch": 2.106954866166424,
      "grad_norm": 1.1744362115859985,
      "learning_rate": 9.536605263157895e-05,
      "loss": 0.3198,
      "step": 37626
    },
    {
      "epoch": 2.107010863478553,
      "grad_norm": 1.1389715671539307,
      "learning_rate": 9.536578947368421e-05,
      "loss": 0.4217,
      "step": 37627
    },
    {
      "epoch": 2.107066860790682,
      "grad_norm": 1.3454861640930176,
      "learning_rate": 9.536552631578947e-05,
      "loss": 0.4608,
      "step": 37628
    },
    {
      "epoch": 2.107122858102811,
      "grad_norm": 1.1700462102890015,
      "learning_rate": 9.536526315789475e-05,
      "loss": 0.3859,
      "step": 37629
    },
    {
      "epoch": 2.10717885541494,
      "grad_norm": 1.2105263471603394,
      "learning_rate": 9.5365e-05,
      "loss": 0.4089,
      "step": 37630
    },
    {
      "epoch": 2.107234852727069,
      "grad_norm": 1.2925617694854736,
      "learning_rate": 9.536473684210527e-05,
      "loss": 0.47,
      "step": 37631
    },
    {
      "epoch": 2.107290850039198,
      "grad_norm": 1.2375229597091675,
      "learning_rate": 9.536447368421053e-05,
      "loss": 0.4665,
      "step": 37632
    },
    {
      "epoch": 2.107346847351327,
      "grad_norm": 1.166219711303711,
      "learning_rate": 9.536421052631579e-05,
      "loss": 0.3837,
      "step": 37633
    },
    {
      "epoch": 2.107402844663456,
      "grad_norm": 1.2813639640808105,
      "learning_rate": 9.536394736842106e-05,
      "loss": 0.4506,
      "step": 37634
    },
    {
      "epoch": 2.107458841975585,
      "grad_norm": 1.2452691793441772,
      "learning_rate": 9.536368421052632e-05,
      "loss": 0.4071,
      "step": 37635
    },
    {
      "epoch": 2.107514839287714,
      "grad_norm": 1.2973533868789673,
      "learning_rate": 9.536342105263159e-05,
      "loss": 0.3378,
      "step": 37636
    },
    {
      "epoch": 2.107570836599843,
      "grad_norm": 1.2390633821487427,
      "learning_rate": 9.536315789473684e-05,
      "loss": 0.3924,
      "step": 37637
    },
    {
      "epoch": 2.107626833911972,
      "grad_norm": 1.1294937133789062,
      "learning_rate": 9.536289473684211e-05,
      "loss": 0.3611,
      "step": 37638
    },
    {
      "epoch": 2.107682831224101,
      "grad_norm": 1.0345213413238525,
      "learning_rate": 9.536263157894737e-05,
      "loss": 0.3974,
      "step": 37639
    },
    {
      "epoch": 2.10773882853623,
      "grad_norm": 1.1751327514648438,
      "learning_rate": 9.536236842105264e-05,
      "loss": 0.4803,
      "step": 37640
    },
    {
      "epoch": 2.107794825848359,
      "grad_norm": 1.4237899780273438,
      "learning_rate": 9.53621052631579e-05,
      "loss": 0.4054,
      "step": 37641
    },
    {
      "epoch": 2.107850823160488,
      "grad_norm": 1.1355594396591187,
      "learning_rate": 9.536184210526316e-05,
      "loss": 0.3758,
      "step": 37642
    },
    {
      "epoch": 2.1079068204726172,
      "grad_norm": 1.2845560312271118,
      "learning_rate": 9.536157894736842e-05,
      "loss": 0.4636,
      "step": 37643
    },
    {
      "epoch": 2.1079628177847463,
      "grad_norm": 1.09945809841156,
      "learning_rate": 9.53613157894737e-05,
      "loss": 0.2978,
      "step": 37644
    },
    {
      "epoch": 2.1080188150968753,
      "grad_norm": 1.336234211921692,
      "learning_rate": 9.536105263157896e-05,
      "loss": 0.3787,
      "step": 37645
    },
    {
      "epoch": 2.1080748124090043,
      "grad_norm": 1.6262624263763428,
      "learning_rate": 9.536078947368422e-05,
      "loss": 0.4172,
      "step": 37646
    },
    {
      "epoch": 2.1081308097211333,
      "grad_norm": 1.1292541027069092,
      "learning_rate": 9.536052631578948e-05,
      "loss": 0.3687,
      "step": 37647
    },
    {
      "epoch": 2.1081868070332623,
      "grad_norm": 1.4260233640670776,
      "learning_rate": 9.536026315789475e-05,
      "loss": 0.3782,
      "step": 37648
    },
    {
      "epoch": 2.1082428043453914,
      "grad_norm": 1.0964329242706299,
      "learning_rate": 9.536000000000001e-05,
      "loss": 0.3028,
      "step": 37649
    },
    {
      "epoch": 2.1082988016575204,
      "grad_norm": 1.1549230813980103,
      "learning_rate": 9.535973684210527e-05,
      "loss": 0.3356,
      "step": 37650
    },
    {
      "epoch": 2.1083547989696494,
      "grad_norm": 1.5794603824615479,
      "learning_rate": 9.535947368421053e-05,
      "loss": 0.6498,
      "step": 37651
    },
    {
      "epoch": 2.1084107962817784,
      "grad_norm": 1.0998942852020264,
      "learning_rate": 9.535921052631579e-05,
      "loss": 0.2676,
      "step": 37652
    },
    {
      "epoch": 2.1084667935939074,
      "grad_norm": 1.2723844051361084,
      "learning_rate": 9.535894736842106e-05,
      "loss": 0.3411,
      "step": 37653
    },
    {
      "epoch": 2.1085227909060364,
      "grad_norm": 1.1733119487762451,
      "learning_rate": 9.535868421052632e-05,
      "loss": 0.3587,
      "step": 37654
    },
    {
      "epoch": 2.1085787882181655,
      "grad_norm": 1.2792655229568481,
      "learning_rate": 9.535842105263158e-05,
      "loss": 0.4295,
      "step": 37655
    },
    {
      "epoch": 2.1086347855302945,
      "grad_norm": 1.0961569547653198,
      "learning_rate": 9.535815789473684e-05,
      "loss": 0.3743,
      "step": 37656
    },
    {
      "epoch": 2.1086907828424235,
      "grad_norm": 1.2600672245025635,
      "learning_rate": 9.535789473684211e-05,
      "loss": 0.4055,
      "step": 37657
    },
    {
      "epoch": 2.1087467801545525,
      "grad_norm": 1.284856915473938,
      "learning_rate": 9.535763157894737e-05,
      "loss": 0.4472,
      "step": 37658
    },
    {
      "epoch": 2.1088027774666815,
      "grad_norm": 1.4620368480682373,
      "learning_rate": 9.535736842105263e-05,
      "loss": 0.4192,
      "step": 37659
    },
    {
      "epoch": 2.1088587747788106,
      "grad_norm": 1.1823731660842896,
      "learning_rate": 9.535710526315789e-05,
      "loss": 0.3756,
      "step": 37660
    },
    {
      "epoch": 2.1089147720909396,
      "grad_norm": 1.25752592086792,
      "learning_rate": 9.535684210526317e-05,
      "loss": 0.4007,
      "step": 37661
    },
    {
      "epoch": 2.1089707694030686,
      "grad_norm": 1.2487444877624512,
      "learning_rate": 9.535657894736843e-05,
      "loss": 0.4453,
      "step": 37662
    },
    {
      "epoch": 2.1090267667151976,
      "grad_norm": 1.2192504405975342,
      "learning_rate": 9.53563157894737e-05,
      "loss": 0.3499,
      "step": 37663
    },
    {
      "epoch": 2.1090827640273266,
      "grad_norm": 1.2093487977981567,
      "learning_rate": 9.535605263157895e-05,
      "loss": 0.3824,
      "step": 37664
    },
    {
      "epoch": 2.1091387613394557,
      "grad_norm": 1.2120347023010254,
      "learning_rate": 9.535578947368422e-05,
      "loss": 0.4127,
      "step": 37665
    },
    {
      "epoch": 2.1091947586515847,
      "grad_norm": 1.254859209060669,
      "learning_rate": 9.535552631578948e-05,
      "loss": 0.4518,
      "step": 37666
    },
    {
      "epoch": 2.1092507559637137,
      "grad_norm": 1.2412467002868652,
      "learning_rate": 9.535526315789474e-05,
      "loss": 0.4724,
      "step": 37667
    },
    {
      "epoch": 2.1093067532758427,
      "grad_norm": 1.1160590648651123,
      "learning_rate": 9.535500000000001e-05,
      "loss": 0.3399,
      "step": 37668
    },
    {
      "epoch": 2.1093627505879717,
      "grad_norm": 1.1172974109649658,
      "learning_rate": 9.535473684210526e-05,
      "loss": 0.3388,
      "step": 37669
    },
    {
      "epoch": 2.1094187479001008,
      "grad_norm": 1.2220187187194824,
      "learning_rate": 9.535447368421053e-05,
      "loss": 0.4007,
      "step": 37670
    },
    {
      "epoch": 2.1094747452122298,
      "grad_norm": 1.247684359550476,
      "learning_rate": 9.535421052631579e-05,
      "loss": 0.3795,
      "step": 37671
    },
    {
      "epoch": 2.109530742524359,
      "grad_norm": 1.02691650390625,
      "learning_rate": 9.535394736842106e-05,
      "loss": 0.296,
      "step": 37672
    },
    {
      "epoch": 2.109586739836488,
      "grad_norm": 1.1002277135849,
      "learning_rate": 9.535368421052632e-05,
      "loss": 0.3334,
      "step": 37673
    },
    {
      "epoch": 2.109642737148617,
      "grad_norm": 1.1732851266860962,
      "learning_rate": 9.535342105263158e-05,
      "loss": 0.3932,
      "step": 37674
    },
    {
      "epoch": 2.109698734460746,
      "grad_norm": 1.5924240350723267,
      "learning_rate": 9.535315789473684e-05,
      "loss": 0.4279,
      "step": 37675
    },
    {
      "epoch": 2.109754731772875,
      "grad_norm": 1.443625569343567,
      "learning_rate": 9.535289473684212e-05,
      "loss": 0.5602,
      "step": 37676
    },
    {
      "epoch": 2.109810729085004,
      "grad_norm": 1.2515921592712402,
      "learning_rate": 9.535263157894738e-05,
      "loss": 0.419,
      "step": 37677
    },
    {
      "epoch": 2.109866726397133,
      "grad_norm": 1.3528836965560913,
      "learning_rate": 9.535236842105264e-05,
      "loss": 0.663,
      "step": 37678
    },
    {
      "epoch": 2.109922723709262,
      "grad_norm": 1.3246128559112549,
      "learning_rate": 9.53521052631579e-05,
      "loss": 0.3789,
      "step": 37679
    },
    {
      "epoch": 2.109978721021391,
      "grad_norm": 1.1721516847610474,
      "learning_rate": 9.535184210526317e-05,
      "loss": 0.3756,
      "step": 37680
    },
    {
      "epoch": 2.11003471833352,
      "grad_norm": 1.4207085371017456,
      "learning_rate": 9.535157894736843e-05,
      "loss": 0.4669,
      "step": 37681
    },
    {
      "epoch": 2.110090715645649,
      "grad_norm": 1.2498027086257935,
      "learning_rate": 9.535131578947369e-05,
      "loss": 0.3731,
      "step": 37682
    },
    {
      "epoch": 2.110146712957778,
      "grad_norm": 1.4895986318588257,
      "learning_rate": 9.535105263157895e-05,
      "loss": 0.4104,
      "step": 37683
    },
    {
      "epoch": 2.110202710269907,
      "grad_norm": 1.1882609128952026,
      "learning_rate": 9.535078947368421e-05,
      "loss": 0.4324,
      "step": 37684
    },
    {
      "epoch": 2.110258707582036,
      "grad_norm": 1.3308801651000977,
      "learning_rate": 9.535052631578948e-05,
      "loss": 0.3848,
      "step": 37685
    },
    {
      "epoch": 2.110314704894165,
      "grad_norm": 1.0326311588287354,
      "learning_rate": 9.535026315789474e-05,
      "loss": 0.2912,
      "step": 37686
    },
    {
      "epoch": 2.110370702206294,
      "grad_norm": 1.1226708889007568,
      "learning_rate": 9.535e-05,
      "loss": 0.3299,
      "step": 37687
    },
    {
      "epoch": 2.110426699518423,
      "grad_norm": 1.4871490001678467,
      "learning_rate": 9.534973684210526e-05,
      "loss": 0.4005,
      "step": 37688
    },
    {
      "epoch": 2.110482696830552,
      "grad_norm": 1.2377749681472778,
      "learning_rate": 9.534947368421053e-05,
      "loss": 0.422,
      "step": 37689
    },
    {
      "epoch": 2.110538694142681,
      "grad_norm": 1.1320468187332153,
      "learning_rate": 9.534921052631579e-05,
      "loss": 0.3589,
      "step": 37690
    },
    {
      "epoch": 2.11059469145481,
      "grad_norm": 1.4200750589370728,
      "learning_rate": 9.534894736842107e-05,
      "loss": 0.4861,
      "step": 37691
    },
    {
      "epoch": 2.110650688766939,
      "grad_norm": 1.0389201641082764,
      "learning_rate": 9.534868421052631e-05,
      "loss": 0.5698,
      "step": 37692
    },
    {
      "epoch": 2.110706686079068,
      "grad_norm": 0.9634940028190613,
      "learning_rate": 9.534842105263159e-05,
      "loss": 0.3202,
      "step": 37693
    },
    {
      "epoch": 2.110762683391197,
      "grad_norm": 1.0997002124786377,
      "learning_rate": 9.534815789473685e-05,
      "loss": 0.333,
      "step": 37694
    },
    {
      "epoch": 2.1108186807033262,
      "grad_norm": 1.1376430988311768,
      "learning_rate": 9.534789473684212e-05,
      "loss": 0.3694,
      "step": 37695
    },
    {
      "epoch": 2.1108746780154553,
      "grad_norm": 1.2030510902404785,
      "learning_rate": 9.534763157894738e-05,
      "loss": 0.4139,
      "step": 37696
    },
    {
      "epoch": 2.1109306753275843,
      "grad_norm": 1.1908349990844727,
      "learning_rate": 9.534736842105264e-05,
      "loss": 0.3961,
      "step": 37697
    },
    {
      "epoch": 2.1109866726397133,
      "grad_norm": 1.1226392984390259,
      "learning_rate": 9.53471052631579e-05,
      "loss": 0.4358,
      "step": 37698
    },
    {
      "epoch": 2.1110426699518423,
      "grad_norm": 1.2626878023147583,
      "learning_rate": 9.534684210526317e-05,
      "loss": 0.4907,
      "step": 37699
    },
    {
      "epoch": 2.1110986672639713,
      "grad_norm": 1.3042765855789185,
      "learning_rate": 9.534657894736843e-05,
      "loss": 0.4645,
      "step": 37700
    },
    {
      "epoch": 2.1111546645761003,
      "grad_norm": 1.2931530475616455,
      "learning_rate": 9.534631578947368e-05,
      "loss": 0.4219,
      "step": 37701
    },
    {
      "epoch": 2.1112106618882294,
      "grad_norm": 1.3562021255493164,
      "learning_rate": 9.534605263157895e-05,
      "loss": 0.4596,
      "step": 37702
    },
    {
      "epoch": 2.1112666592003584,
      "grad_norm": 1.2898374795913696,
      "learning_rate": 9.534578947368421e-05,
      "loss": 0.3581,
      "step": 37703
    },
    {
      "epoch": 2.1113226565124874,
      "grad_norm": 1.2061854600906372,
      "learning_rate": 9.534552631578948e-05,
      "loss": 0.3977,
      "step": 37704
    },
    {
      "epoch": 2.1113786538246164,
      "grad_norm": 1.3235297203063965,
      "learning_rate": 9.534526315789474e-05,
      "loss": 0.3571,
      "step": 37705
    },
    {
      "epoch": 2.1114346511367454,
      "grad_norm": 1.2746213674545288,
      "learning_rate": 9.5345e-05,
      "loss": 0.5094,
      "step": 37706
    },
    {
      "epoch": 2.1114906484488745,
      "grad_norm": 1.3411320447921753,
      "learning_rate": 9.534473684210526e-05,
      "loss": 0.4216,
      "step": 37707
    },
    {
      "epoch": 2.1115466457610035,
      "grad_norm": 1.3013726472854614,
      "learning_rate": 9.534447368421054e-05,
      "loss": 0.3305,
      "step": 37708
    },
    {
      "epoch": 2.1116026430731325,
      "grad_norm": 1.1931748390197754,
      "learning_rate": 9.53442105263158e-05,
      "loss": 0.4516,
      "step": 37709
    },
    {
      "epoch": 2.1116586403852615,
      "grad_norm": 0.997417688369751,
      "learning_rate": 9.534394736842106e-05,
      "loss": 0.3268,
      "step": 37710
    },
    {
      "epoch": 2.1117146376973905,
      "grad_norm": 1.3128045797348022,
      "learning_rate": 9.534368421052631e-05,
      "loss": 0.5035,
      "step": 37711
    },
    {
      "epoch": 2.1117706350095196,
      "grad_norm": 1.5369590520858765,
      "learning_rate": 9.534342105263159e-05,
      "loss": 0.3783,
      "step": 37712
    },
    {
      "epoch": 2.1118266323216486,
      "grad_norm": 0.994907557964325,
      "learning_rate": 9.534315789473685e-05,
      "loss": 0.3859,
      "step": 37713
    },
    {
      "epoch": 2.1118826296337776,
      "grad_norm": 1.630784273147583,
      "learning_rate": 9.534289473684211e-05,
      "loss": 0.5884,
      "step": 37714
    },
    {
      "epoch": 2.1119386269459066,
      "grad_norm": 1.5292789936065674,
      "learning_rate": 9.534263157894737e-05,
      "loss": 0.4662,
      "step": 37715
    },
    {
      "epoch": 2.1119946242580356,
      "grad_norm": 1.374040961265564,
      "learning_rate": 9.534236842105264e-05,
      "loss": 0.4858,
      "step": 37716
    },
    {
      "epoch": 2.1120506215701647,
      "grad_norm": 1.1735118627548218,
      "learning_rate": 9.53421052631579e-05,
      "loss": 0.366,
      "step": 37717
    },
    {
      "epoch": 2.1121066188822937,
      "grad_norm": 1.1391663551330566,
      "learning_rate": 9.534184210526316e-05,
      "loss": 0.4214,
      "step": 37718
    },
    {
      "epoch": 2.1121626161944227,
      "grad_norm": 1.172099232673645,
      "learning_rate": 9.534157894736842e-05,
      "loss": 0.3972,
      "step": 37719
    },
    {
      "epoch": 2.1122186135065517,
      "grad_norm": 1.3714842796325684,
      "learning_rate": 9.534131578947368e-05,
      "loss": 0.5416,
      "step": 37720
    },
    {
      "epoch": 2.1122746108186807,
      "grad_norm": 1.390973687171936,
      "learning_rate": 9.534105263157895e-05,
      "loss": 0.4445,
      "step": 37721
    },
    {
      "epoch": 2.1123306081308098,
      "grad_norm": 1.1501284837722778,
      "learning_rate": 9.534078947368421e-05,
      "loss": 0.4735,
      "step": 37722
    },
    {
      "epoch": 2.1123866054429388,
      "grad_norm": 1.2320268154144287,
      "learning_rate": 9.534052631578949e-05,
      "loss": 0.4413,
      "step": 37723
    },
    {
      "epoch": 2.112442602755068,
      "grad_norm": 1.2600257396697998,
      "learning_rate": 9.534026315789473e-05,
      "loss": 0.3476,
      "step": 37724
    },
    {
      "epoch": 2.112498600067197,
      "grad_norm": 1.1833323240280151,
      "learning_rate": 9.534e-05,
      "loss": 0.3807,
      "step": 37725
    },
    {
      "epoch": 2.112554597379326,
      "grad_norm": 1.3143657445907593,
      "learning_rate": 9.533973684210527e-05,
      "loss": 0.4744,
      "step": 37726
    },
    {
      "epoch": 2.112610594691455,
      "grad_norm": 1.4492616653442383,
      "learning_rate": 9.533947368421054e-05,
      "loss": 0.4555,
      "step": 37727
    },
    {
      "epoch": 2.112666592003584,
      "grad_norm": 1.362603783607483,
      "learning_rate": 9.53392105263158e-05,
      "loss": 0.4049,
      "step": 37728
    },
    {
      "epoch": 2.112722589315713,
      "grad_norm": 1.4482049942016602,
      "learning_rate": 9.533894736842106e-05,
      "loss": 0.4499,
      "step": 37729
    },
    {
      "epoch": 2.112778586627842,
      "grad_norm": 1.1095775365829468,
      "learning_rate": 9.533868421052632e-05,
      "loss": 0.3733,
      "step": 37730
    },
    {
      "epoch": 2.112834583939971,
      "grad_norm": 1.288974404335022,
      "learning_rate": 9.533842105263159e-05,
      "loss": 0.3501,
      "step": 37731
    },
    {
      "epoch": 2.1128905812521,
      "grad_norm": 1.22319757938385,
      "learning_rate": 9.533815789473685e-05,
      "loss": 0.3843,
      "step": 37732
    },
    {
      "epoch": 2.112946578564229,
      "grad_norm": 1.2581007480621338,
      "learning_rate": 9.533789473684211e-05,
      "loss": 0.4432,
      "step": 37733
    },
    {
      "epoch": 2.113002575876358,
      "grad_norm": 1.4334641695022583,
      "learning_rate": 9.533763157894737e-05,
      "loss": 0.3485,
      "step": 37734
    },
    {
      "epoch": 2.113058573188487,
      "grad_norm": 1.1329319477081299,
      "learning_rate": 9.533736842105263e-05,
      "loss": 0.4525,
      "step": 37735
    },
    {
      "epoch": 2.113114570500616,
      "grad_norm": 1.234907627105713,
      "learning_rate": 9.53371052631579e-05,
      "loss": 0.386,
      "step": 37736
    },
    {
      "epoch": 2.113170567812745,
      "grad_norm": 10.424367904663086,
      "learning_rate": 9.533684210526316e-05,
      "loss": 0.4617,
      "step": 37737
    },
    {
      "epoch": 2.113226565124874,
      "grad_norm": 0.9995739459991455,
      "learning_rate": 9.533657894736842e-05,
      "loss": 0.2939,
      "step": 37738
    },
    {
      "epoch": 2.113282562437003,
      "grad_norm": 1.154493808746338,
      "learning_rate": 9.533631578947368e-05,
      "loss": 0.3866,
      "step": 37739
    },
    {
      "epoch": 2.113338559749132,
      "grad_norm": 1.1501070261001587,
      "learning_rate": 9.533605263157896e-05,
      "loss": 0.4242,
      "step": 37740
    },
    {
      "epoch": 2.113394557061261,
      "grad_norm": 1.2766205072402954,
      "learning_rate": 9.533578947368422e-05,
      "loss": 0.3907,
      "step": 37741
    },
    {
      "epoch": 2.11345055437339,
      "grad_norm": 1.2891227006912231,
      "learning_rate": 9.533552631578947e-05,
      "loss": 0.4394,
      "step": 37742
    },
    {
      "epoch": 2.113506551685519,
      "grad_norm": 1.2441725730895996,
      "learning_rate": 9.533526315789473e-05,
      "loss": 0.3491,
      "step": 37743
    },
    {
      "epoch": 2.113562548997648,
      "grad_norm": 1.2892502546310425,
      "learning_rate": 9.533500000000001e-05,
      "loss": 0.4216,
      "step": 37744
    },
    {
      "epoch": 2.113618546309777,
      "grad_norm": 1.5214523077011108,
      "learning_rate": 9.533473684210527e-05,
      "loss": 0.3599,
      "step": 37745
    },
    {
      "epoch": 2.113674543621906,
      "grad_norm": 1.1385174989700317,
      "learning_rate": 9.533447368421054e-05,
      "loss": 0.4494,
      "step": 37746
    },
    {
      "epoch": 2.1137305409340352,
      "grad_norm": 1.390559196472168,
      "learning_rate": 9.533421052631579e-05,
      "loss": 0.3402,
      "step": 37747
    },
    {
      "epoch": 2.1137865382461642,
      "grad_norm": 1.2365553379058838,
      "learning_rate": 9.533394736842106e-05,
      "loss": 0.4832,
      "step": 37748
    },
    {
      "epoch": 2.1138425355582933,
      "grad_norm": 1.30582857131958,
      "learning_rate": 9.533368421052632e-05,
      "loss": 0.4236,
      "step": 37749
    },
    {
      "epoch": 2.1138985328704223,
      "grad_norm": 1.213720440864563,
      "learning_rate": 9.53334210526316e-05,
      "loss": 0.4795,
      "step": 37750
    },
    {
      "epoch": 2.1139545301825513,
      "grad_norm": 1.1774375438690186,
      "learning_rate": 9.533315789473685e-05,
      "loss": 0.3157,
      "step": 37751
    },
    {
      "epoch": 2.1140105274946803,
      "grad_norm": 1.2122899293899536,
      "learning_rate": 9.53328947368421e-05,
      "loss": 0.3471,
      "step": 37752
    },
    {
      "epoch": 2.1140665248068093,
      "grad_norm": 1.4899814128875732,
      "learning_rate": 9.533263157894737e-05,
      "loss": 0.3924,
      "step": 37753
    },
    {
      "epoch": 2.1141225221189384,
      "grad_norm": 1.2417616844177246,
      "learning_rate": 9.533236842105263e-05,
      "loss": 0.355,
      "step": 37754
    },
    {
      "epoch": 2.1141785194310674,
      "grad_norm": 1.3443083763122559,
      "learning_rate": 9.53321052631579e-05,
      "loss": 0.4508,
      "step": 37755
    },
    {
      "epoch": 2.1142345167431964,
      "grad_norm": 1.0989341735839844,
      "learning_rate": 9.533184210526315e-05,
      "loss": 0.6193,
      "step": 37756
    },
    {
      "epoch": 2.1142905140553254,
      "grad_norm": 1.1604912281036377,
      "learning_rate": 9.533157894736843e-05,
      "loss": 0.5548,
      "step": 37757
    },
    {
      "epoch": 2.1143465113674544,
      "grad_norm": 1.5853008031845093,
      "learning_rate": 9.533131578947368e-05,
      "loss": 0.4636,
      "step": 37758
    },
    {
      "epoch": 2.1144025086795835,
      "grad_norm": 1.2316597700119019,
      "learning_rate": 9.533105263157896e-05,
      "loss": 0.3978,
      "step": 37759
    },
    {
      "epoch": 2.1144585059917125,
      "grad_norm": 1.0920389890670776,
      "learning_rate": 9.533078947368422e-05,
      "loss": 0.3358,
      "step": 37760
    },
    {
      "epoch": 2.1145145033038415,
      "grad_norm": 1.0193519592285156,
      "learning_rate": 9.533052631578948e-05,
      "loss": 0.3429,
      "step": 37761
    },
    {
      "epoch": 2.1145705006159705,
      "grad_norm": 1.4410526752471924,
      "learning_rate": 9.533026315789474e-05,
      "loss": 0.4416,
      "step": 37762
    },
    {
      "epoch": 2.1146264979280995,
      "grad_norm": 1.339468002319336,
      "learning_rate": 9.533000000000001e-05,
      "loss": 0.5131,
      "step": 37763
    },
    {
      "epoch": 2.1146824952402286,
      "grad_norm": 1.0867359638214111,
      "learning_rate": 9.532973684210527e-05,
      "loss": 0.3936,
      "step": 37764
    },
    {
      "epoch": 2.1147384925523576,
      "grad_norm": 1.1157872676849365,
      "learning_rate": 9.532947368421053e-05,
      "loss": 0.2791,
      "step": 37765
    },
    {
      "epoch": 2.1147944898644866,
      "grad_norm": 1.4042932987213135,
      "learning_rate": 9.532921052631579e-05,
      "loss": 0.4413,
      "step": 37766
    },
    {
      "epoch": 2.1148504871766156,
      "grad_norm": 1.2471221685409546,
      "learning_rate": 9.532894736842106e-05,
      "loss": 0.5011,
      "step": 37767
    },
    {
      "epoch": 2.1149064844887446,
      "grad_norm": 1.0515024662017822,
      "learning_rate": 9.532868421052632e-05,
      "loss": 0.3093,
      "step": 37768
    },
    {
      "epoch": 2.1149624818008737,
      "grad_norm": 1.1320897340774536,
      "learning_rate": 9.532842105263158e-05,
      "loss": 0.3541,
      "step": 37769
    },
    {
      "epoch": 2.1150184791130027,
      "grad_norm": 1.1313374042510986,
      "learning_rate": 9.532815789473684e-05,
      "loss": 0.4539,
      "step": 37770
    },
    {
      "epoch": 2.1150744764251317,
      "grad_norm": 1.2079766988754272,
      "learning_rate": 9.53278947368421e-05,
      "loss": 0.4676,
      "step": 37771
    },
    {
      "epoch": 2.1151304737372607,
      "grad_norm": 1.1099145412445068,
      "learning_rate": 9.532763157894738e-05,
      "loss": 0.3352,
      "step": 37772
    },
    {
      "epoch": 2.1151864710493897,
      "grad_norm": 1.165467619895935,
      "learning_rate": 9.532736842105263e-05,
      "loss": 0.374,
      "step": 37773
    },
    {
      "epoch": 2.1152424683615187,
      "grad_norm": 1.6281195878982544,
      "learning_rate": 9.53271052631579e-05,
      "loss": 0.489,
      "step": 37774
    },
    {
      "epoch": 2.1152984656736478,
      "grad_norm": 1.1673880815505981,
      "learning_rate": 9.532684210526315e-05,
      "loss": 0.5044,
      "step": 37775
    },
    {
      "epoch": 2.115354462985777,
      "grad_norm": 1.1003457307815552,
      "learning_rate": 9.532657894736843e-05,
      "loss": 0.3416,
      "step": 37776
    },
    {
      "epoch": 2.115410460297906,
      "grad_norm": 1.307185411453247,
      "learning_rate": 9.532631578947369e-05,
      "loss": 0.4734,
      "step": 37777
    },
    {
      "epoch": 2.115466457610035,
      "grad_norm": 1.1056904792785645,
      "learning_rate": 9.532605263157896e-05,
      "loss": 0.3534,
      "step": 37778
    },
    {
      "epoch": 2.115522454922164,
      "grad_norm": 1.336822509765625,
      "learning_rate": 9.532578947368421e-05,
      "loss": 0.432,
      "step": 37779
    },
    {
      "epoch": 2.115578452234293,
      "grad_norm": 1.2675280570983887,
      "learning_rate": 9.532552631578948e-05,
      "loss": 0.5916,
      "step": 37780
    },
    {
      "epoch": 2.115634449546422,
      "grad_norm": 1.0618001222610474,
      "learning_rate": 9.532526315789474e-05,
      "loss": 0.3309,
      "step": 37781
    },
    {
      "epoch": 2.115690446858551,
      "grad_norm": 1.125723958015442,
      "learning_rate": 9.532500000000001e-05,
      "loss": 0.2804,
      "step": 37782
    },
    {
      "epoch": 2.11574644417068,
      "grad_norm": 1.0606441497802734,
      "learning_rate": 9.532473684210527e-05,
      "loss": 0.3503,
      "step": 37783
    },
    {
      "epoch": 2.115802441482809,
      "grad_norm": 1.1353930234909058,
      "learning_rate": 9.532447368421053e-05,
      "loss": 0.4429,
      "step": 37784
    },
    {
      "epoch": 2.115858438794938,
      "grad_norm": 1.134667158126831,
      "learning_rate": 9.532421052631579e-05,
      "loss": 0.3666,
      "step": 37785
    },
    {
      "epoch": 2.115914436107067,
      "grad_norm": 1.098807454109192,
      "learning_rate": 9.532394736842107e-05,
      "loss": 0.4427,
      "step": 37786
    },
    {
      "epoch": 2.115970433419196,
      "grad_norm": 1.4602035284042358,
      "learning_rate": 9.532368421052633e-05,
      "loss": 0.3726,
      "step": 37787
    },
    {
      "epoch": 2.116026430731325,
      "grad_norm": 1.1076409816741943,
      "learning_rate": 9.532342105263159e-05,
      "loss": 0.289,
      "step": 37788
    },
    {
      "epoch": 2.116082428043454,
      "grad_norm": 1.1125380992889404,
      "learning_rate": 9.532315789473684e-05,
      "loss": 0.3272,
      "step": 37789
    },
    {
      "epoch": 2.116138425355583,
      "grad_norm": 1.3662583827972412,
      "learning_rate": 9.53228947368421e-05,
      "loss": 0.388,
      "step": 37790
    },
    {
      "epoch": 2.116194422667712,
      "grad_norm": 1.3298346996307373,
      "learning_rate": 9.532263157894738e-05,
      "loss": 0.4258,
      "step": 37791
    },
    {
      "epoch": 2.116250419979841,
      "grad_norm": 1.2065690755844116,
      "learning_rate": 9.532236842105264e-05,
      "loss": 0.432,
      "step": 37792
    },
    {
      "epoch": 2.11630641729197,
      "grad_norm": 1.2760372161865234,
      "learning_rate": 9.53221052631579e-05,
      "loss": 0.3999,
      "step": 37793
    },
    {
      "epoch": 2.116362414604099,
      "grad_norm": 1.139719009399414,
      "learning_rate": 9.532184210526316e-05,
      "loss": 0.3156,
      "step": 37794
    },
    {
      "epoch": 2.116418411916228,
      "grad_norm": 1.0459928512573242,
      "learning_rate": 9.532157894736843e-05,
      "loss": 0.3737,
      "step": 37795
    },
    {
      "epoch": 2.116474409228357,
      "grad_norm": 1.007080316543579,
      "learning_rate": 9.532131578947369e-05,
      "loss": 0.3101,
      "step": 37796
    },
    {
      "epoch": 2.116530406540486,
      "grad_norm": 1.2657349109649658,
      "learning_rate": 9.532105263157895e-05,
      "loss": 0.3466,
      "step": 37797
    },
    {
      "epoch": 2.116586403852615,
      "grad_norm": 1.2727843523025513,
      "learning_rate": 9.532078947368421e-05,
      "loss": 0.3888,
      "step": 37798
    },
    {
      "epoch": 2.1166424011647442,
      "grad_norm": 1.3665987253189087,
      "learning_rate": 9.532052631578948e-05,
      "loss": 0.5027,
      "step": 37799
    },
    {
      "epoch": 2.1166983984768732,
      "grad_norm": 1.14633047580719,
      "learning_rate": 9.532026315789474e-05,
      "loss": 0.4561,
      "step": 37800
    },
    {
      "epoch": 2.1167543957890023,
      "grad_norm": 1.150771141052246,
      "learning_rate": 9.532000000000002e-05,
      "loss": 0.429,
      "step": 37801
    },
    {
      "epoch": 2.1168103931011313,
      "grad_norm": 1.1088114976882935,
      "learning_rate": 9.531973684210526e-05,
      "loss": 0.3064,
      "step": 37802
    },
    {
      "epoch": 2.1168663904132603,
      "grad_norm": 1.6099441051483154,
      "learning_rate": 9.531947368421054e-05,
      "loss": 0.5734,
      "step": 37803
    },
    {
      "epoch": 2.1169223877253893,
      "grad_norm": 1.4852136373519897,
      "learning_rate": 9.53192105263158e-05,
      "loss": 0.6458,
      "step": 37804
    },
    {
      "epoch": 2.1169783850375183,
      "grad_norm": 1.2405089139938354,
      "learning_rate": 9.531894736842105e-05,
      "loss": 0.3487,
      "step": 37805
    },
    {
      "epoch": 2.1170343823496474,
      "grad_norm": 0.9049539566040039,
      "learning_rate": 9.531868421052633e-05,
      "loss": 0.2875,
      "step": 37806
    },
    {
      "epoch": 2.1170903796617764,
      "grad_norm": 1.2981488704681396,
      "learning_rate": 9.531842105263157e-05,
      "loss": 0.4661,
      "step": 37807
    },
    {
      "epoch": 2.1171463769739054,
      "grad_norm": 3.335503339767456,
      "learning_rate": 9.531815789473685e-05,
      "loss": 0.5825,
      "step": 37808
    },
    {
      "epoch": 2.1172023742860344,
      "grad_norm": 1.1003698110580444,
      "learning_rate": 9.531789473684211e-05,
      "loss": 0.4532,
      "step": 37809
    },
    {
      "epoch": 2.1172583715981634,
      "grad_norm": 1.1118097305297852,
      "learning_rate": 9.531763157894738e-05,
      "loss": 0.3212,
      "step": 37810
    },
    {
      "epoch": 2.1173143689102925,
      "grad_norm": 1.362941026687622,
      "learning_rate": 9.531736842105263e-05,
      "loss": 0.3852,
      "step": 37811
    },
    {
      "epoch": 2.1173703662224215,
      "grad_norm": 1.21826171875,
      "learning_rate": 9.53171052631579e-05,
      "loss": 0.4628,
      "step": 37812
    },
    {
      "epoch": 2.1174263635345505,
      "grad_norm": 1.2498557567596436,
      "learning_rate": 9.531684210526316e-05,
      "loss": 0.6538,
      "step": 37813
    },
    {
      "epoch": 2.1174823608466795,
      "grad_norm": 1.3448197841644287,
      "learning_rate": 9.531657894736843e-05,
      "loss": 0.4049,
      "step": 37814
    },
    {
      "epoch": 2.1175383581588085,
      "grad_norm": 1.0381081104278564,
      "learning_rate": 9.531631578947369e-05,
      "loss": 0.3324,
      "step": 37815
    },
    {
      "epoch": 2.1175943554709376,
      "grad_norm": 1.150907278060913,
      "learning_rate": 9.531605263157895e-05,
      "loss": 0.3745,
      "step": 37816
    },
    {
      "epoch": 2.1176503527830666,
      "grad_norm": 1.1191520690917969,
      "learning_rate": 9.531578947368421e-05,
      "loss": 0.317,
      "step": 37817
    },
    {
      "epoch": 2.1177063500951956,
      "grad_norm": 1.0300110578536987,
      "learning_rate": 9.531552631578949e-05,
      "loss": 0.4101,
      "step": 37818
    },
    {
      "epoch": 2.1177623474073246,
      "grad_norm": 1.178125262260437,
      "learning_rate": 9.531526315789475e-05,
      "loss": 0.3649,
      "step": 37819
    },
    {
      "epoch": 2.1178183447194536,
      "grad_norm": 1.2245209217071533,
      "learning_rate": 9.5315e-05,
      "loss": 0.387,
      "step": 37820
    },
    {
      "epoch": 2.1178743420315826,
      "grad_norm": 1.3217214345932007,
      "learning_rate": 9.531473684210526e-05,
      "loss": 0.4777,
      "step": 37821
    },
    {
      "epoch": 2.1179303393437117,
      "grad_norm": 1.4592339992523193,
      "learning_rate": 9.531447368421052e-05,
      "loss": 0.3873,
      "step": 37822
    },
    {
      "epoch": 2.1179863366558407,
      "grad_norm": 1.208885908126831,
      "learning_rate": 9.53142105263158e-05,
      "loss": 0.3682,
      "step": 37823
    },
    {
      "epoch": 2.1180423339679697,
      "grad_norm": 1.3035715818405151,
      "learning_rate": 9.531394736842106e-05,
      "loss": 0.4042,
      "step": 37824
    },
    {
      "epoch": 2.1180983312800987,
      "grad_norm": 1.3901745080947876,
      "learning_rate": 9.531368421052632e-05,
      "loss": 0.4977,
      "step": 37825
    },
    {
      "epoch": 2.1181543285922277,
      "grad_norm": 1.1876839399337769,
      "learning_rate": 9.531342105263158e-05,
      "loss": 0.4223,
      "step": 37826
    },
    {
      "epoch": 2.1182103259043568,
      "grad_norm": 1.1811094284057617,
      "learning_rate": 9.531315789473685e-05,
      "loss": 0.4048,
      "step": 37827
    },
    {
      "epoch": 2.118266323216486,
      "grad_norm": 1.0523566007614136,
      "learning_rate": 9.531289473684211e-05,
      "loss": 0.3743,
      "step": 37828
    },
    {
      "epoch": 2.118322320528615,
      "grad_norm": 1.228696346282959,
      "learning_rate": 9.531263157894737e-05,
      "loss": 0.3872,
      "step": 37829
    },
    {
      "epoch": 2.118378317840744,
      "grad_norm": 1.5651582479476929,
      "learning_rate": 9.531236842105263e-05,
      "loss": 0.4446,
      "step": 37830
    },
    {
      "epoch": 2.118434315152873,
      "grad_norm": 1.2424721717834473,
      "learning_rate": 9.53121052631579e-05,
      "loss": 0.3981,
      "step": 37831
    },
    {
      "epoch": 2.118490312465002,
      "grad_norm": 1.9074064493179321,
      "learning_rate": 9.531184210526316e-05,
      "loss": 0.4753,
      "step": 37832
    },
    {
      "epoch": 2.118546309777131,
      "grad_norm": 1.414995789527893,
      "learning_rate": 9.531157894736844e-05,
      "loss": 0.4431,
      "step": 37833
    },
    {
      "epoch": 2.11860230708926,
      "grad_norm": 1.2317641973495483,
      "learning_rate": 9.531131578947368e-05,
      "loss": 0.5211,
      "step": 37834
    },
    {
      "epoch": 2.118658304401389,
      "grad_norm": 1.1901524066925049,
      "learning_rate": 9.531105263157895e-05,
      "loss": 0.3705,
      "step": 37835
    },
    {
      "epoch": 2.118714301713518,
      "grad_norm": 1.1597578525543213,
      "learning_rate": 9.531078947368421e-05,
      "loss": 0.4003,
      "step": 37836
    },
    {
      "epoch": 2.118770299025647,
      "grad_norm": 1.245458960533142,
      "learning_rate": 9.531052631578949e-05,
      "loss": 0.3511,
      "step": 37837
    },
    {
      "epoch": 2.118826296337776,
      "grad_norm": 1.221401572227478,
      "learning_rate": 9.531026315789475e-05,
      "loss": 0.3562,
      "step": 37838
    },
    {
      "epoch": 2.118882293649905,
      "grad_norm": 1.2704401016235352,
      "learning_rate": 9.531e-05,
      "loss": 0.408,
      "step": 37839
    },
    {
      "epoch": 2.118938290962034,
      "grad_norm": 1.1648112535476685,
      "learning_rate": 9.530973684210527e-05,
      "loss": 0.3679,
      "step": 37840
    },
    {
      "epoch": 2.118994288274163,
      "grad_norm": 1.776847004890442,
      "learning_rate": 9.530947368421053e-05,
      "loss": 0.474,
      "step": 37841
    },
    {
      "epoch": 2.119050285586292,
      "grad_norm": 1.1837369203567505,
      "learning_rate": 9.53092105263158e-05,
      "loss": 0.4325,
      "step": 37842
    },
    {
      "epoch": 2.119106282898421,
      "grad_norm": 1.2718595266342163,
      "learning_rate": 9.530894736842106e-05,
      "loss": 0.3621,
      "step": 37843
    },
    {
      "epoch": 2.11916228021055,
      "grad_norm": 1.0920614004135132,
      "learning_rate": 9.530868421052632e-05,
      "loss": 0.4358,
      "step": 37844
    },
    {
      "epoch": 2.119218277522679,
      "grad_norm": 1.2388941049575806,
      "learning_rate": 9.530842105263158e-05,
      "loss": 0.4012,
      "step": 37845
    },
    {
      "epoch": 2.119274274834808,
      "grad_norm": 1.3358289003372192,
      "learning_rate": 9.530815789473685e-05,
      "loss": 0.4192,
      "step": 37846
    },
    {
      "epoch": 2.119330272146937,
      "grad_norm": 1.178553581237793,
      "learning_rate": 9.530789473684211e-05,
      "loss": 0.3461,
      "step": 37847
    },
    {
      "epoch": 2.119386269459066,
      "grad_norm": 1.004017949104309,
      "learning_rate": 9.530763157894737e-05,
      "loss": 0.3324,
      "step": 37848
    },
    {
      "epoch": 2.119442266771195,
      "grad_norm": 1.5627021789550781,
      "learning_rate": 9.530736842105263e-05,
      "loss": 0.5497,
      "step": 37849
    },
    {
      "epoch": 2.119498264083324,
      "grad_norm": 1.0687642097473145,
      "learning_rate": 9.53071052631579e-05,
      "loss": 0.3485,
      "step": 37850
    },
    {
      "epoch": 2.119554261395453,
      "grad_norm": 1.380363941192627,
      "learning_rate": 9.530684210526316e-05,
      "loss": 0.3637,
      "step": 37851
    },
    {
      "epoch": 2.1196102587075822,
      "grad_norm": 1.372911810874939,
      "learning_rate": 9.530657894736842e-05,
      "loss": 0.3865,
      "step": 37852
    },
    {
      "epoch": 2.1196662560197113,
      "grad_norm": 1.0048118829727173,
      "learning_rate": 9.530631578947368e-05,
      "loss": 0.3525,
      "step": 37853
    },
    {
      "epoch": 2.1197222533318403,
      "grad_norm": 1.1297627687454224,
      "learning_rate": 9.530605263157896e-05,
      "loss": 0.3891,
      "step": 37854
    },
    {
      "epoch": 2.1197782506439693,
      "grad_norm": 1.726850986480713,
      "learning_rate": 9.530578947368422e-05,
      "loss": 0.5057,
      "step": 37855
    },
    {
      "epoch": 2.1198342479560983,
      "grad_norm": 1.1801815032958984,
      "learning_rate": 9.530552631578948e-05,
      "loss": 0.3669,
      "step": 37856
    },
    {
      "epoch": 2.1198902452682273,
      "grad_norm": 1.9252707958221436,
      "learning_rate": 9.530526315789474e-05,
      "loss": 0.4337,
      "step": 37857
    },
    {
      "epoch": 2.1199462425803564,
      "grad_norm": 1.2330302000045776,
      "learning_rate": 9.5305e-05,
      "loss": 0.3446,
      "step": 37858
    },
    {
      "epoch": 2.1200022398924854,
      "grad_norm": 1.2027246952056885,
      "learning_rate": 9.530473684210527e-05,
      "loss": 0.4215,
      "step": 37859
    },
    {
      "epoch": 2.1200582372046144,
      "grad_norm": 11.59250259399414,
      "learning_rate": 9.530447368421053e-05,
      "loss": 0.3774,
      "step": 37860
    },
    {
      "epoch": 2.120114234516743,
      "grad_norm": 1.3247222900390625,
      "learning_rate": 9.530421052631579e-05,
      "loss": 0.3576,
      "step": 37861
    },
    {
      "epoch": 2.1201702318288724,
      "grad_norm": 1.2677847146987915,
      "learning_rate": 9.530394736842105e-05,
      "loss": 0.3704,
      "step": 37862
    },
    {
      "epoch": 2.120226229141001,
      "grad_norm": 1.1607929468154907,
      "learning_rate": 9.530368421052632e-05,
      "loss": 0.3964,
      "step": 37863
    },
    {
      "epoch": 2.1202822264531305,
      "grad_norm": 1.180167555809021,
      "learning_rate": 9.530342105263158e-05,
      "loss": 0.3506,
      "step": 37864
    },
    {
      "epoch": 2.120338223765259,
      "grad_norm": 1.253099799156189,
      "learning_rate": 9.530315789473686e-05,
      "loss": 0.368,
      "step": 37865
    },
    {
      "epoch": 2.1203942210773885,
      "grad_norm": 1.6238375902175903,
      "learning_rate": 9.53028947368421e-05,
      "loss": 0.5655,
      "step": 37866
    },
    {
      "epoch": 2.120450218389517,
      "grad_norm": 0.9868596792221069,
      "learning_rate": 9.530263157894737e-05,
      "loss": 0.3393,
      "step": 37867
    },
    {
      "epoch": 2.1205062157016465,
      "grad_norm": 0.935724139213562,
      "learning_rate": 9.530236842105263e-05,
      "loss": 0.2918,
      "step": 37868
    },
    {
      "epoch": 2.120562213013775,
      "grad_norm": 1.0049982070922852,
      "learning_rate": 9.530210526315791e-05,
      "loss": 0.34,
      "step": 37869
    },
    {
      "epoch": 2.1206182103259046,
      "grad_norm": 2.395608425140381,
      "learning_rate": 9.530184210526317e-05,
      "loss": 0.5362,
      "step": 37870
    },
    {
      "epoch": 2.120674207638033,
      "grad_norm": 1.216915249824524,
      "learning_rate": 9.530157894736843e-05,
      "loss": 0.3044,
      "step": 37871
    },
    {
      "epoch": 2.120730204950162,
      "grad_norm": 1.6804417371749878,
      "learning_rate": 9.530131578947369e-05,
      "loss": 0.5489,
      "step": 37872
    },
    {
      "epoch": 2.120786202262291,
      "grad_norm": 1.2425605058670044,
      "learning_rate": 9.530105263157895e-05,
      "loss": 0.5188,
      "step": 37873
    },
    {
      "epoch": 2.12084219957442,
      "grad_norm": 1.4513862133026123,
      "learning_rate": 9.530078947368422e-05,
      "loss": 0.3784,
      "step": 37874
    },
    {
      "epoch": 2.1208981968865492,
      "grad_norm": 1.2104915380477905,
      "learning_rate": 9.530052631578948e-05,
      "loss": 0.4571,
      "step": 37875
    },
    {
      "epoch": 2.1209541941986783,
      "grad_norm": 1.4345799684524536,
      "learning_rate": 9.530026315789474e-05,
      "loss": 0.409,
      "step": 37876
    },
    {
      "epoch": 2.1210101915108073,
      "grad_norm": 1.3612395524978638,
      "learning_rate": 9.53e-05,
      "loss": 0.3945,
      "step": 37877
    },
    {
      "epoch": 2.1210661888229363,
      "grad_norm": 1.274493932723999,
      "learning_rate": 9.529973684210527e-05,
      "loss": 0.4298,
      "step": 37878
    },
    {
      "epoch": 2.1211221861350653,
      "grad_norm": 1.0498849153518677,
      "learning_rate": 9.529947368421053e-05,
      "loss": 0.3003,
      "step": 37879
    },
    {
      "epoch": 2.1211781834471943,
      "grad_norm": 1.6037408113479614,
      "learning_rate": 9.529921052631579e-05,
      "loss": 0.4317,
      "step": 37880
    },
    {
      "epoch": 2.1212341807593234,
      "grad_norm": 1.1873399019241333,
      "learning_rate": 9.529894736842105e-05,
      "loss": 0.3375,
      "step": 37881
    },
    {
      "epoch": 2.1212901780714524,
      "grad_norm": 2.2799975872039795,
      "learning_rate": 9.529868421052632e-05,
      "loss": 0.4741,
      "step": 37882
    },
    {
      "epoch": 2.1213461753835814,
      "grad_norm": 1.4707082509994507,
      "learning_rate": 9.529842105263158e-05,
      "loss": 0.4066,
      "step": 37883
    },
    {
      "epoch": 2.1214021726957104,
      "grad_norm": 1.120848536491394,
      "learning_rate": 9.529815789473684e-05,
      "loss": 0.3951,
      "step": 37884
    },
    {
      "epoch": 2.1214581700078394,
      "grad_norm": 2.239795207977295,
      "learning_rate": 9.52978947368421e-05,
      "loss": 0.4015,
      "step": 37885
    },
    {
      "epoch": 2.1215141673199684,
      "grad_norm": 1.1888052225112915,
      "learning_rate": 9.529763157894738e-05,
      "loss": 0.3552,
      "step": 37886
    },
    {
      "epoch": 2.1215701646320975,
      "grad_norm": 1.222002625465393,
      "learning_rate": 9.529736842105264e-05,
      "loss": 0.3734,
      "step": 37887
    },
    {
      "epoch": 2.1216261619442265,
      "grad_norm": 1.1781657934188843,
      "learning_rate": 9.529710526315791e-05,
      "loss": 0.4107,
      "step": 37888
    },
    {
      "epoch": 2.1216821592563555,
      "grad_norm": 1.4246891736984253,
      "learning_rate": 9.529684210526316e-05,
      "loss": 0.4274,
      "step": 37889
    },
    {
      "epoch": 2.1217381565684845,
      "grad_norm": 1.1937236785888672,
      "learning_rate": 9.529657894736842e-05,
      "loss": 0.4264,
      "step": 37890
    },
    {
      "epoch": 2.1217941538806135,
      "grad_norm": 1.1374826431274414,
      "learning_rate": 9.529631578947369e-05,
      "loss": 0.4189,
      "step": 37891
    },
    {
      "epoch": 2.1218501511927426,
      "grad_norm": 1.240546464920044,
      "learning_rate": 9.529605263157895e-05,
      "loss": 0.384,
      "step": 37892
    },
    {
      "epoch": 2.1219061485048716,
      "grad_norm": 1.2603422403335571,
      "learning_rate": 9.529578947368422e-05,
      "loss": 0.4497,
      "step": 37893
    },
    {
      "epoch": 2.1219621458170006,
      "grad_norm": 1.3641451597213745,
      "learning_rate": 9.529552631578947e-05,
      "loss": 0.3536,
      "step": 37894
    },
    {
      "epoch": 2.1220181431291296,
      "grad_norm": 1.1944833993911743,
      "learning_rate": 9.529526315789474e-05,
      "loss": 0.4059,
      "step": 37895
    },
    {
      "epoch": 2.1220741404412586,
      "grad_norm": 1.2889608144760132,
      "learning_rate": 9.5295e-05,
      "loss": 0.3696,
      "step": 37896
    },
    {
      "epoch": 2.1221301377533877,
      "grad_norm": 1.3130391836166382,
      "learning_rate": 9.529473684210527e-05,
      "loss": 0.4156,
      "step": 37897
    },
    {
      "epoch": 2.1221861350655167,
      "grad_norm": 1.4661166667938232,
      "learning_rate": 9.529447368421053e-05,
      "loss": 0.3641,
      "step": 37898
    },
    {
      "epoch": 2.1222421323776457,
      "grad_norm": 1.3656140565872192,
      "learning_rate": 9.52942105263158e-05,
      "loss": 0.5186,
      "step": 37899
    },
    {
      "epoch": 2.1222981296897747,
      "grad_norm": 1.2916266918182373,
      "learning_rate": 9.529394736842105e-05,
      "loss": 0.4432,
      "step": 37900
    },
    {
      "epoch": 2.1223541270019037,
      "grad_norm": 1.222621202468872,
      "learning_rate": 9.529368421052633e-05,
      "loss": 0.576,
      "step": 37901
    },
    {
      "epoch": 2.1224101243140328,
      "grad_norm": 1.6173008680343628,
      "learning_rate": 9.529342105263159e-05,
      "loss": 0.4105,
      "step": 37902
    },
    {
      "epoch": 2.1224661216261618,
      "grad_norm": 1.3266522884368896,
      "learning_rate": 9.529315789473685e-05,
      "loss": 0.5658,
      "step": 37903
    },
    {
      "epoch": 2.122522118938291,
      "grad_norm": 1.2293651103973389,
      "learning_rate": 9.52928947368421e-05,
      "loss": 0.4712,
      "step": 37904
    },
    {
      "epoch": 2.12257811625042,
      "grad_norm": 1.1439555883407593,
      "learning_rate": 9.529263157894738e-05,
      "loss": 0.3701,
      "step": 37905
    },
    {
      "epoch": 2.122634113562549,
      "grad_norm": 1.230742335319519,
      "learning_rate": 9.529236842105264e-05,
      "loss": 0.449,
      "step": 37906
    },
    {
      "epoch": 2.122690110874678,
      "grad_norm": 1.2715702056884766,
      "learning_rate": 9.52921052631579e-05,
      "loss": 0.4344,
      "step": 37907
    },
    {
      "epoch": 2.122746108186807,
      "grad_norm": 1.171204924583435,
      "learning_rate": 9.529184210526316e-05,
      "loss": 0.4105,
      "step": 37908
    },
    {
      "epoch": 2.122802105498936,
      "grad_norm": 0.9493831396102905,
      "learning_rate": 9.529157894736842e-05,
      "loss": 0.3483,
      "step": 37909
    },
    {
      "epoch": 2.122858102811065,
      "grad_norm": 1.241207480430603,
      "learning_rate": 9.529131578947369e-05,
      "loss": 0.398,
      "step": 37910
    },
    {
      "epoch": 2.122914100123194,
      "grad_norm": 1.1960015296936035,
      "learning_rate": 9.529105263157895e-05,
      "loss": 0.3897,
      "step": 37911
    },
    {
      "epoch": 2.122970097435323,
      "grad_norm": 1.1621278524398804,
      "learning_rate": 9.529078947368421e-05,
      "loss": 0.3482,
      "step": 37912
    },
    {
      "epoch": 2.123026094747452,
      "grad_norm": 0.9802743792533875,
      "learning_rate": 9.529052631578947e-05,
      "loss": 0.3122,
      "step": 37913
    },
    {
      "epoch": 2.123082092059581,
      "grad_norm": 1.3529890775680542,
      "learning_rate": 9.529026315789474e-05,
      "loss": 0.4806,
      "step": 37914
    },
    {
      "epoch": 2.12313808937171,
      "grad_norm": 1.2725598812103271,
      "learning_rate": 9.529e-05,
      "loss": 0.3694,
      "step": 37915
    },
    {
      "epoch": 2.123194086683839,
      "grad_norm": 1.2651251554489136,
      "learning_rate": 9.528973684210526e-05,
      "loss": 0.5421,
      "step": 37916
    },
    {
      "epoch": 2.123250083995968,
      "grad_norm": 1.0582131147384644,
      "learning_rate": 9.528947368421052e-05,
      "loss": 0.3326,
      "step": 37917
    },
    {
      "epoch": 2.123306081308097,
      "grad_norm": 2.0830466747283936,
      "learning_rate": 9.52892105263158e-05,
      "loss": 0.542,
      "step": 37918
    },
    {
      "epoch": 2.123362078620226,
      "grad_norm": 1.1611149311065674,
      "learning_rate": 9.528894736842106e-05,
      "loss": 0.4467,
      "step": 37919
    },
    {
      "epoch": 2.123418075932355,
      "grad_norm": 1.0782891511917114,
      "learning_rate": 9.528868421052633e-05,
      "loss": 0.289,
      "step": 37920
    },
    {
      "epoch": 2.123474073244484,
      "grad_norm": 1.3379461765289307,
      "learning_rate": 9.528842105263158e-05,
      "loss": 0.5336,
      "step": 37921
    },
    {
      "epoch": 2.123530070556613,
      "grad_norm": 1.0055097341537476,
      "learning_rate": 9.528815789473685e-05,
      "loss": 0.3341,
      "step": 37922
    },
    {
      "epoch": 2.123586067868742,
      "grad_norm": 1.2222527265548706,
      "learning_rate": 9.528789473684211e-05,
      "loss": 0.4326,
      "step": 37923
    },
    {
      "epoch": 2.123642065180871,
      "grad_norm": 1.134134292602539,
      "learning_rate": 9.528763157894738e-05,
      "loss": 0.35,
      "step": 37924
    },
    {
      "epoch": 2.123698062493,
      "grad_norm": 1.1304410696029663,
      "learning_rate": 9.528736842105264e-05,
      "loss": 0.381,
      "step": 37925
    },
    {
      "epoch": 2.123754059805129,
      "grad_norm": 1.1868478059768677,
      "learning_rate": 9.528710526315789e-05,
      "loss": 0.4347,
      "step": 37926
    },
    {
      "epoch": 2.1238100571172582,
      "grad_norm": 1.1191680431365967,
      "learning_rate": 9.528684210526316e-05,
      "loss": 0.4864,
      "step": 37927
    },
    {
      "epoch": 2.1238660544293873,
      "grad_norm": 1.53757643699646,
      "learning_rate": 9.528657894736842e-05,
      "loss": 0.4159,
      "step": 37928
    },
    {
      "epoch": 2.1239220517415163,
      "grad_norm": 1.367113471031189,
      "learning_rate": 9.52863157894737e-05,
      "loss": 0.3963,
      "step": 37929
    },
    {
      "epoch": 2.1239780490536453,
      "grad_norm": 1.32086980342865,
      "learning_rate": 9.528605263157895e-05,
      "loss": 0.3874,
      "step": 37930
    },
    {
      "epoch": 2.1240340463657743,
      "grad_norm": 1.1955931186676025,
      "learning_rate": 9.528578947368421e-05,
      "loss": 0.3927,
      "step": 37931
    },
    {
      "epoch": 2.1240900436779033,
      "grad_norm": 1.0497366189956665,
      "learning_rate": 9.528552631578947e-05,
      "loss": 0.4363,
      "step": 37932
    },
    {
      "epoch": 2.1241460409900323,
      "grad_norm": 1.3721774816513062,
      "learning_rate": 9.528526315789475e-05,
      "loss": 0.3975,
      "step": 37933
    },
    {
      "epoch": 2.1242020383021614,
      "grad_norm": 2.4244110584259033,
      "learning_rate": 9.5285e-05,
      "loss": 0.4301,
      "step": 37934
    },
    {
      "epoch": 2.1242580356142904,
      "grad_norm": 1.2817292213439941,
      "learning_rate": 9.528473684210527e-05,
      "loss": 0.38,
      "step": 37935
    },
    {
      "epoch": 2.1243140329264194,
      "grad_norm": 1.9610052108764648,
      "learning_rate": 9.528447368421053e-05,
      "loss": 0.4149,
      "step": 37936
    },
    {
      "epoch": 2.1243700302385484,
      "grad_norm": 1.0222722291946411,
      "learning_rate": 9.52842105263158e-05,
      "loss": 0.2925,
      "step": 37937
    },
    {
      "epoch": 2.1244260275506774,
      "grad_norm": 1.5136748552322388,
      "learning_rate": 9.528394736842106e-05,
      "loss": 0.5081,
      "step": 37938
    },
    {
      "epoch": 2.1244820248628065,
      "grad_norm": 1.4547631740570068,
      "learning_rate": 9.528368421052632e-05,
      "loss": 0.3788,
      "step": 37939
    },
    {
      "epoch": 2.1245380221749355,
      "grad_norm": 1.1482230424880981,
      "learning_rate": 9.528342105263158e-05,
      "loss": 0.3097,
      "step": 37940
    },
    {
      "epoch": 2.1245940194870645,
      "grad_norm": 1.0308619737625122,
      "learning_rate": 9.528315789473685e-05,
      "loss": 0.3182,
      "step": 37941
    },
    {
      "epoch": 2.1246500167991935,
      "grad_norm": 1.0513734817504883,
      "learning_rate": 9.528289473684211e-05,
      "loss": 0.3506,
      "step": 37942
    },
    {
      "epoch": 2.1247060141113225,
      "grad_norm": 0.9860881567001343,
      "learning_rate": 9.528263157894737e-05,
      "loss": 0.2575,
      "step": 37943
    },
    {
      "epoch": 2.1247620114234516,
      "grad_norm": 1.2484816312789917,
      "learning_rate": 9.528236842105263e-05,
      "loss": 0.3139,
      "step": 37944
    },
    {
      "epoch": 2.1248180087355806,
      "grad_norm": 1.031117558479309,
      "learning_rate": 9.528210526315789e-05,
      "loss": 0.2977,
      "step": 37945
    },
    {
      "epoch": 2.1248740060477096,
      "grad_norm": 1.2670975923538208,
      "learning_rate": 9.528184210526316e-05,
      "loss": 0.4885,
      "step": 37946
    },
    {
      "epoch": 2.1249300033598386,
      "grad_norm": 1.2101235389709473,
      "learning_rate": 9.528157894736842e-05,
      "loss": 0.4491,
      "step": 37947
    },
    {
      "epoch": 2.1249860006719676,
      "grad_norm": 1.1838452816009521,
      "learning_rate": 9.52813157894737e-05,
      "loss": 0.4123,
      "step": 37948
    },
    {
      "epoch": 2.1250419979840967,
      "grad_norm": 1.2386916875839233,
      "learning_rate": 9.528105263157894e-05,
      "loss": 0.461,
      "step": 37949
    },
    {
      "epoch": 2.1250979952962257,
      "grad_norm": 1.1315258741378784,
      "learning_rate": 9.528078947368422e-05,
      "loss": 0.3907,
      "step": 37950
    },
    {
      "epoch": 2.1251539926083547,
      "grad_norm": 1.2928507328033447,
      "learning_rate": 9.528052631578948e-05,
      "loss": 0.4226,
      "step": 37951
    },
    {
      "epoch": 2.1252099899204837,
      "grad_norm": 1.2939456701278687,
      "learning_rate": 9.528026315789475e-05,
      "loss": 0.5222,
      "step": 37952
    },
    {
      "epoch": 2.1252659872326127,
      "grad_norm": 1.379740834236145,
      "learning_rate": 9.528000000000001e-05,
      "loss": 0.3743,
      "step": 37953
    },
    {
      "epoch": 2.1253219845447417,
      "grad_norm": 0.9714986681938171,
      "learning_rate": 9.527973684210527e-05,
      "loss": 0.3069,
      "step": 37954
    },
    {
      "epoch": 2.1253779818568708,
      "grad_norm": 1.1989487409591675,
      "learning_rate": 9.527947368421053e-05,
      "loss": 0.3927,
      "step": 37955
    },
    {
      "epoch": 2.125433979169,
      "grad_norm": 1.2335395812988281,
      "learning_rate": 9.52792105263158e-05,
      "loss": 0.4218,
      "step": 37956
    },
    {
      "epoch": 2.125489976481129,
      "grad_norm": 1.1995121240615845,
      "learning_rate": 9.527894736842106e-05,
      "loss": 0.3731,
      "step": 37957
    },
    {
      "epoch": 2.125545973793258,
      "grad_norm": 1.0925146341323853,
      "learning_rate": 9.527868421052632e-05,
      "loss": 0.4277,
      "step": 37958
    },
    {
      "epoch": 2.125601971105387,
      "grad_norm": 1.2077703475952148,
      "learning_rate": 9.527842105263158e-05,
      "loss": 0.4163,
      "step": 37959
    },
    {
      "epoch": 2.125657968417516,
      "grad_norm": 1.1684659719467163,
      "learning_rate": 9.527815789473684e-05,
      "loss": 0.4115,
      "step": 37960
    },
    {
      "epoch": 2.125713965729645,
      "grad_norm": 2.821614980697632,
      "learning_rate": 9.527789473684211e-05,
      "loss": 0.4261,
      "step": 37961
    },
    {
      "epoch": 2.125769963041774,
      "grad_norm": 1.2369601726531982,
      "learning_rate": 9.527763157894737e-05,
      "loss": 0.4649,
      "step": 37962
    },
    {
      "epoch": 2.125825960353903,
      "grad_norm": 1.170612096786499,
      "learning_rate": 9.527736842105263e-05,
      "loss": 0.4416,
      "step": 37963
    },
    {
      "epoch": 2.125881957666032,
      "grad_norm": 1.1098328828811646,
      "learning_rate": 9.52771052631579e-05,
      "loss": 0.405,
      "step": 37964
    },
    {
      "epoch": 2.125937954978161,
      "grad_norm": 1.4585012197494507,
      "learning_rate": 9.527684210526317e-05,
      "loss": 0.423,
      "step": 37965
    },
    {
      "epoch": 2.12599395229029,
      "grad_norm": 1.250841498374939,
      "learning_rate": 9.527657894736843e-05,
      "loss": 0.3501,
      "step": 37966
    },
    {
      "epoch": 2.126049949602419,
      "grad_norm": 1.2169735431671143,
      "learning_rate": 9.527631578947369e-05,
      "loss": 0.3859,
      "step": 37967
    },
    {
      "epoch": 2.126105946914548,
      "grad_norm": 1.583406925201416,
      "learning_rate": 9.527605263157895e-05,
      "loss": 0.3602,
      "step": 37968
    },
    {
      "epoch": 2.126161944226677,
      "grad_norm": 1.2548919916152954,
      "learning_rate": 9.527578947368422e-05,
      "loss": 0.378,
      "step": 37969
    },
    {
      "epoch": 2.126217941538806,
      "grad_norm": 1.0360541343688965,
      "learning_rate": 9.527552631578948e-05,
      "loss": 0.3592,
      "step": 37970
    },
    {
      "epoch": 2.126273938850935,
      "grad_norm": 1.138854742050171,
      "learning_rate": 9.527526315789474e-05,
      "loss": 0.3268,
      "step": 37971
    },
    {
      "epoch": 2.126329936163064,
      "grad_norm": 1.2766937017440796,
      "learning_rate": 9.5275e-05,
      "loss": 0.4271,
      "step": 37972
    },
    {
      "epoch": 2.126385933475193,
      "grad_norm": 1.1715785264968872,
      "learning_rate": 9.527473684210527e-05,
      "loss": 0.3522,
      "step": 37973
    },
    {
      "epoch": 2.126441930787322,
      "grad_norm": 1.1503130197525024,
      "learning_rate": 9.527447368421053e-05,
      "loss": 0.3519,
      "step": 37974
    },
    {
      "epoch": 2.126497928099451,
      "grad_norm": 1.5244576930999756,
      "learning_rate": 9.52742105263158e-05,
      "loss": 0.4394,
      "step": 37975
    },
    {
      "epoch": 2.12655392541158,
      "grad_norm": 1.2267115116119385,
      "learning_rate": 9.527394736842105e-05,
      "loss": 0.5937,
      "step": 37976
    },
    {
      "epoch": 2.126609922723709,
      "grad_norm": 1.0399476289749146,
      "learning_rate": 9.527368421052631e-05,
      "loss": 0.4108,
      "step": 37977
    },
    {
      "epoch": 2.126665920035838,
      "grad_norm": 3.3570339679718018,
      "learning_rate": 9.527342105263158e-05,
      "loss": 0.351,
      "step": 37978
    },
    {
      "epoch": 2.1267219173479672,
      "grad_norm": 1.0552146434783936,
      "learning_rate": 9.527315789473684e-05,
      "loss": 0.3609,
      "step": 37979
    },
    {
      "epoch": 2.1267779146600962,
      "grad_norm": 1.0554651021957397,
      "learning_rate": 9.527289473684212e-05,
      "loss": 0.2997,
      "step": 37980
    },
    {
      "epoch": 2.1268339119722253,
      "grad_norm": 1.498835563659668,
      "learning_rate": 9.527263157894736e-05,
      "loss": 0.4488,
      "step": 37981
    },
    {
      "epoch": 2.1268899092843543,
      "grad_norm": 0.9943491220474243,
      "learning_rate": 9.527236842105264e-05,
      "loss": 0.2958,
      "step": 37982
    },
    {
      "epoch": 2.1269459065964833,
      "grad_norm": 1.1603772640228271,
      "learning_rate": 9.52721052631579e-05,
      "loss": 0.3737,
      "step": 37983
    },
    {
      "epoch": 2.1270019039086123,
      "grad_norm": 1.308395504951477,
      "learning_rate": 9.527184210526317e-05,
      "loss": 0.5178,
      "step": 37984
    },
    {
      "epoch": 2.1270579012207413,
      "grad_norm": 1.1913459300994873,
      "learning_rate": 9.527157894736843e-05,
      "loss": 0.4921,
      "step": 37985
    },
    {
      "epoch": 2.1271138985328704,
      "grad_norm": 1.0885347127914429,
      "learning_rate": 9.527131578947369e-05,
      "loss": 0.4256,
      "step": 37986
    },
    {
      "epoch": 2.1271698958449994,
      "grad_norm": 1.210465908050537,
      "learning_rate": 9.527105263157895e-05,
      "loss": 0.4189,
      "step": 37987
    },
    {
      "epoch": 2.1272258931571284,
      "grad_norm": 1.1525578498840332,
      "learning_rate": 9.527078947368422e-05,
      "loss": 0.392,
      "step": 37988
    },
    {
      "epoch": 2.1272818904692574,
      "grad_norm": 1.4139964580535889,
      "learning_rate": 9.527052631578948e-05,
      "loss": 0.4049,
      "step": 37989
    },
    {
      "epoch": 2.1273378877813864,
      "grad_norm": 1.1509031057357788,
      "learning_rate": 9.527026315789474e-05,
      "loss": 0.396,
      "step": 37990
    },
    {
      "epoch": 2.1273938850935155,
      "grad_norm": 1.540934443473816,
      "learning_rate": 9.527e-05,
      "loss": 0.5453,
      "step": 37991
    },
    {
      "epoch": 2.1274498824056445,
      "grad_norm": 1.2494606971740723,
      "learning_rate": 9.526973684210527e-05,
      "loss": 0.5135,
      "step": 37992
    },
    {
      "epoch": 2.1275058797177735,
      "grad_norm": 1.2312878370285034,
      "learning_rate": 9.526947368421053e-05,
      "loss": 0.4202,
      "step": 37993
    },
    {
      "epoch": 2.1275618770299025,
      "grad_norm": 1.0873737335205078,
      "learning_rate": 9.52692105263158e-05,
      "loss": 0.4458,
      "step": 37994
    },
    {
      "epoch": 2.1276178743420315,
      "grad_norm": 1.1822694540023804,
      "learning_rate": 9.526894736842105e-05,
      "loss": 0.4514,
      "step": 37995
    },
    {
      "epoch": 2.1276738716541606,
      "grad_norm": 1.933834433555603,
      "learning_rate": 9.526868421052631e-05,
      "loss": 0.4042,
      "step": 37996
    },
    {
      "epoch": 2.1277298689662896,
      "grad_norm": 1.212084412574768,
      "learning_rate": 9.526842105263159e-05,
      "loss": 0.4588,
      "step": 37997
    },
    {
      "epoch": 2.1277858662784186,
      "grad_norm": 1.2162895202636719,
      "learning_rate": 9.526815789473685e-05,
      "loss": 0.3609,
      "step": 37998
    },
    {
      "epoch": 2.1278418635905476,
      "grad_norm": 1.2796589136123657,
      "learning_rate": 9.52678947368421e-05,
      "loss": 0.4637,
      "step": 37999
    },
    {
      "epoch": 2.1278978609026766,
      "grad_norm": 1.2076795101165771,
      "learning_rate": 9.526763157894737e-05,
      "loss": 0.3873,
      "step": 38000
    },
    {
      "epoch": 2.1279538582148056,
      "grad_norm": 1.3360549211502075,
      "learning_rate": 9.526736842105264e-05,
      "loss": 0.4939,
      "step": 38001
    },
    {
      "epoch": 2.1280098555269347,
      "grad_norm": 1.0555404424667358,
      "learning_rate": 9.52671052631579e-05,
      "loss": 0.326,
      "step": 38002
    },
    {
      "epoch": 2.1280658528390637,
      "grad_norm": 1.303219199180603,
      "learning_rate": 9.526684210526317e-05,
      "loss": 0.4291,
      "step": 38003
    },
    {
      "epoch": 2.1281218501511927,
      "grad_norm": 1.0729445219039917,
      "learning_rate": 9.526657894736842e-05,
      "loss": 0.3746,
      "step": 38004
    },
    {
      "epoch": 2.1281778474633217,
      "grad_norm": 1.295015811920166,
      "learning_rate": 9.526631578947369e-05,
      "loss": 0.3456,
      "step": 38005
    },
    {
      "epoch": 2.1282338447754507,
      "grad_norm": 1.159035563468933,
      "learning_rate": 9.526605263157895e-05,
      "loss": 0.4296,
      "step": 38006
    },
    {
      "epoch": 2.1282898420875798,
      "grad_norm": 1.1793303489685059,
      "learning_rate": 9.526578947368422e-05,
      "loss": 0.3101,
      "step": 38007
    },
    {
      "epoch": 2.128345839399709,
      "grad_norm": 1.3309861421585083,
      "learning_rate": 9.526552631578947e-05,
      "loss": 0.472,
      "step": 38008
    },
    {
      "epoch": 2.128401836711838,
      "grad_norm": 1.6163116693496704,
      "learning_rate": 9.526526315789474e-05,
      "loss": 0.3426,
      "step": 38009
    },
    {
      "epoch": 2.128457834023967,
      "grad_norm": 1.2657898664474487,
      "learning_rate": 9.5265e-05,
      "loss": 0.4352,
      "step": 38010
    },
    {
      "epoch": 2.128513831336096,
      "grad_norm": 1.3555500507354736,
      "learning_rate": 9.526473684210526e-05,
      "loss": 0.5425,
      "step": 38011
    },
    {
      "epoch": 2.128569828648225,
      "grad_norm": 1.3209218978881836,
      "learning_rate": 9.526447368421054e-05,
      "loss": 0.5877,
      "step": 38012
    },
    {
      "epoch": 2.128625825960354,
      "grad_norm": 1.2382484674453735,
      "learning_rate": 9.526421052631578e-05,
      "loss": 0.4607,
      "step": 38013
    },
    {
      "epoch": 2.128681823272483,
      "grad_norm": 0.9914510846138,
      "learning_rate": 9.526394736842106e-05,
      "loss": 0.2801,
      "step": 38014
    },
    {
      "epoch": 2.128737820584612,
      "grad_norm": 1.3365525007247925,
      "learning_rate": 9.526368421052632e-05,
      "loss": 0.5038,
      "step": 38015
    },
    {
      "epoch": 2.128793817896741,
      "grad_norm": 1.103435754776001,
      "learning_rate": 9.526342105263159e-05,
      "loss": 0.2661,
      "step": 38016
    },
    {
      "epoch": 2.12884981520887,
      "grad_norm": 1.072560429573059,
      "learning_rate": 9.526315789473685e-05,
      "loss": 0.3636,
      "step": 38017
    },
    {
      "epoch": 2.128905812520999,
      "grad_norm": 1.1811704635620117,
      "learning_rate": 9.526289473684211e-05,
      "loss": 0.4714,
      "step": 38018
    },
    {
      "epoch": 2.128961809833128,
      "grad_norm": 1.3292332887649536,
      "learning_rate": 9.526263157894737e-05,
      "loss": 0.3906,
      "step": 38019
    },
    {
      "epoch": 2.129017807145257,
      "grad_norm": 1.183525800704956,
      "learning_rate": 9.526236842105264e-05,
      "loss": 0.3852,
      "step": 38020
    },
    {
      "epoch": 2.129073804457386,
      "grad_norm": 1.075858235359192,
      "learning_rate": 9.52621052631579e-05,
      "loss": 0.3572,
      "step": 38021
    },
    {
      "epoch": 2.129129801769515,
      "grad_norm": 1.056778073310852,
      "learning_rate": 9.526184210526316e-05,
      "loss": 0.3609,
      "step": 38022
    },
    {
      "epoch": 2.129185799081644,
      "grad_norm": 1.314979076385498,
      "learning_rate": 9.526157894736842e-05,
      "loss": 0.4256,
      "step": 38023
    },
    {
      "epoch": 2.129241796393773,
      "grad_norm": 1.4851086139678955,
      "learning_rate": 9.52613157894737e-05,
      "loss": 0.569,
      "step": 38024
    },
    {
      "epoch": 2.129297793705902,
      "grad_norm": 1.003501534461975,
      "learning_rate": 9.526105263157895e-05,
      "loss": 0.3412,
      "step": 38025
    },
    {
      "epoch": 2.129353791018031,
      "grad_norm": 1.280484676361084,
      "learning_rate": 9.526078947368421e-05,
      "loss": 0.3972,
      "step": 38026
    },
    {
      "epoch": 2.12940978833016,
      "grad_norm": 1.2305500507354736,
      "learning_rate": 9.526052631578947e-05,
      "loss": 0.4467,
      "step": 38027
    },
    {
      "epoch": 2.129465785642289,
      "grad_norm": 1.06526780128479,
      "learning_rate": 9.526026315789475e-05,
      "loss": 0.4148,
      "step": 38028
    },
    {
      "epoch": 2.129521782954418,
      "grad_norm": 1.4010645151138306,
      "learning_rate": 9.526e-05,
      "loss": 0.6029,
      "step": 38029
    },
    {
      "epoch": 2.129577780266547,
      "grad_norm": 1.3162074089050293,
      "learning_rate": 9.525973684210527e-05,
      "loss": 0.4352,
      "step": 38030
    },
    {
      "epoch": 2.1296337775786762,
      "grad_norm": 1.3628954887390137,
      "learning_rate": 9.525947368421053e-05,
      "loss": 0.4788,
      "step": 38031
    },
    {
      "epoch": 2.1296897748908052,
      "grad_norm": 1.086837887763977,
      "learning_rate": 9.525921052631579e-05,
      "loss": 0.4691,
      "step": 38032
    },
    {
      "epoch": 2.1297457722029343,
      "grad_norm": 1.3622840642929077,
      "learning_rate": 9.525894736842106e-05,
      "loss": 0.379,
      "step": 38033
    },
    {
      "epoch": 2.1298017695150633,
      "grad_norm": 1.2818411588668823,
      "learning_rate": 9.525868421052632e-05,
      "loss": 0.3972,
      "step": 38034
    },
    {
      "epoch": 2.1298577668271923,
      "grad_norm": 1.116562843322754,
      "learning_rate": 9.525842105263159e-05,
      "loss": 0.3853,
      "step": 38035
    },
    {
      "epoch": 2.1299137641393213,
      "grad_norm": 1.5651719570159912,
      "learning_rate": 9.525815789473684e-05,
      "loss": 0.5475,
      "step": 38036
    },
    {
      "epoch": 2.1299697614514503,
      "grad_norm": 1.1923152208328247,
      "learning_rate": 9.525789473684211e-05,
      "loss": 0.3647,
      "step": 38037
    },
    {
      "epoch": 2.1300257587635794,
      "grad_norm": 1.0621285438537598,
      "learning_rate": 9.525763157894737e-05,
      "loss": 0.3732,
      "step": 38038
    },
    {
      "epoch": 2.1300817560757084,
      "grad_norm": 1.2186707258224487,
      "learning_rate": 9.525736842105264e-05,
      "loss": 0.3878,
      "step": 38039
    },
    {
      "epoch": 2.1301377533878374,
      "grad_norm": 1.339374303817749,
      "learning_rate": 9.52571052631579e-05,
      "loss": 0.4424,
      "step": 38040
    },
    {
      "epoch": 2.1301937506999664,
      "grad_norm": 1.1642976999282837,
      "learning_rate": 9.525684210526316e-05,
      "loss": 0.374,
      "step": 38041
    },
    {
      "epoch": 2.1302497480120954,
      "grad_norm": 1.3978774547576904,
      "learning_rate": 9.525657894736842e-05,
      "loss": 0.4784,
      "step": 38042
    },
    {
      "epoch": 2.1303057453242245,
      "grad_norm": 1.1733441352844238,
      "learning_rate": 9.52563157894737e-05,
      "loss": 0.3921,
      "step": 38043
    },
    {
      "epoch": 2.1303617426363535,
      "grad_norm": 1.6524600982666016,
      "learning_rate": 9.525605263157896e-05,
      "loss": 0.6389,
      "step": 38044
    },
    {
      "epoch": 2.1304177399484825,
      "grad_norm": 1.2763721942901611,
      "learning_rate": 9.525578947368422e-05,
      "loss": 0.4434,
      "step": 38045
    },
    {
      "epoch": 2.1304737372606115,
      "grad_norm": 1.1687251329421997,
      "learning_rate": 9.525552631578948e-05,
      "loss": 0.3839,
      "step": 38046
    },
    {
      "epoch": 2.1305297345727405,
      "grad_norm": 1.3363008499145508,
      "learning_rate": 9.525526315789474e-05,
      "loss": 0.5429,
      "step": 38047
    },
    {
      "epoch": 2.1305857318848695,
      "grad_norm": 1.1057367324829102,
      "learning_rate": 9.525500000000001e-05,
      "loss": 0.4435,
      "step": 38048
    },
    {
      "epoch": 2.1306417291969986,
      "grad_norm": 1.234102725982666,
      "learning_rate": 9.525473684210527e-05,
      "loss": 0.3411,
      "step": 38049
    },
    {
      "epoch": 2.1306977265091276,
      "grad_norm": 1.1591323614120483,
      "learning_rate": 9.525447368421053e-05,
      "loss": 0.3871,
      "step": 38050
    },
    {
      "epoch": 2.1307537238212566,
      "grad_norm": 1.2242653369903564,
      "learning_rate": 9.525421052631579e-05,
      "loss": 0.4891,
      "step": 38051
    },
    {
      "epoch": 2.1308097211333856,
      "grad_norm": 1.1745156049728394,
      "learning_rate": 9.525394736842106e-05,
      "loss": 0.3484,
      "step": 38052
    },
    {
      "epoch": 2.1308657184455146,
      "grad_norm": 1.4954283237457275,
      "learning_rate": 9.525368421052632e-05,
      "loss": 0.5606,
      "step": 38053
    },
    {
      "epoch": 2.1309217157576437,
      "grad_norm": 1.1327989101409912,
      "learning_rate": 9.525342105263158e-05,
      "loss": 0.3961,
      "step": 38054
    },
    {
      "epoch": 2.1309777130697727,
      "grad_norm": 1.109371542930603,
      "learning_rate": 9.525315789473684e-05,
      "loss": 0.3639,
      "step": 38055
    },
    {
      "epoch": 2.1310337103819017,
      "grad_norm": 1.0768954753875732,
      "learning_rate": 9.525289473684211e-05,
      "loss": 0.3222,
      "step": 38056
    },
    {
      "epoch": 2.1310897076940307,
      "grad_norm": 1.4685007333755493,
      "learning_rate": 9.525263157894737e-05,
      "loss": 0.3647,
      "step": 38057
    },
    {
      "epoch": 2.1311457050061597,
      "grad_norm": 1.146191120147705,
      "learning_rate": 9.525236842105265e-05,
      "loss": 0.3827,
      "step": 38058
    },
    {
      "epoch": 2.1312017023182888,
      "grad_norm": 1.1066203117370605,
      "learning_rate": 9.525210526315789e-05,
      "loss": 0.3453,
      "step": 38059
    },
    {
      "epoch": 2.131257699630418,
      "grad_norm": 1.2790555953979492,
      "learning_rate": 9.525184210526317e-05,
      "loss": 0.3675,
      "step": 38060
    },
    {
      "epoch": 2.131313696942547,
      "grad_norm": 1.3151731491088867,
      "learning_rate": 9.525157894736843e-05,
      "loss": 0.3577,
      "step": 38061
    },
    {
      "epoch": 2.131369694254676,
      "grad_norm": 1.1565687656402588,
      "learning_rate": 9.52513157894737e-05,
      "loss": 0.5799,
      "step": 38062
    },
    {
      "epoch": 2.131425691566805,
      "grad_norm": 1.2114107608795166,
      "learning_rate": 9.525105263157895e-05,
      "loss": 0.3779,
      "step": 38063
    },
    {
      "epoch": 2.131481688878934,
      "grad_norm": 1.0612064599990845,
      "learning_rate": 9.52507894736842e-05,
      "loss": 0.399,
      "step": 38064
    },
    {
      "epoch": 2.131537686191063,
      "grad_norm": 1.068852424621582,
      "learning_rate": 9.525052631578948e-05,
      "loss": 0.3575,
      "step": 38065
    },
    {
      "epoch": 2.131593683503192,
      "grad_norm": 0.9930023550987244,
      "learning_rate": 9.525026315789474e-05,
      "loss": 0.3041,
      "step": 38066
    },
    {
      "epoch": 2.131649680815321,
      "grad_norm": 1.4951878786087036,
      "learning_rate": 9.525000000000001e-05,
      "loss": 0.2975,
      "step": 38067
    },
    {
      "epoch": 2.13170567812745,
      "grad_norm": 1.0131851434707642,
      "learning_rate": 9.524973684210526e-05,
      "loss": 0.3439,
      "step": 38068
    },
    {
      "epoch": 2.131761675439579,
      "grad_norm": 1.3838732242584229,
      "learning_rate": 9.524947368421053e-05,
      "loss": 0.4531,
      "step": 38069
    },
    {
      "epoch": 2.131817672751708,
      "grad_norm": 1.0750999450683594,
      "learning_rate": 9.524921052631579e-05,
      "loss": 0.4624,
      "step": 38070
    },
    {
      "epoch": 2.131873670063837,
      "grad_norm": 1.3043855428695679,
      "learning_rate": 9.524894736842106e-05,
      "loss": 0.3655,
      "step": 38071
    },
    {
      "epoch": 2.131929667375966,
      "grad_norm": 1.5056402683258057,
      "learning_rate": 9.524868421052632e-05,
      "loss": 0.593,
      "step": 38072
    },
    {
      "epoch": 2.131985664688095,
      "grad_norm": 1.312848687171936,
      "learning_rate": 9.524842105263158e-05,
      "loss": 0.3436,
      "step": 38073
    },
    {
      "epoch": 2.132041662000224,
      "grad_norm": 3.372086763381958,
      "learning_rate": 9.524815789473684e-05,
      "loss": 0.4041,
      "step": 38074
    },
    {
      "epoch": 2.132097659312353,
      "grad_norm": 1.2546002864837646,
      "learning_rate": 9.524789473684212e-05,
      "loss": 0.4795,
      "step": 38075
    },
    {
      "epoch": 2.132153656624482,
      "grad_norm": 1.248281717300415,
      "learning_rate": 9.524763157894738e-05,
      "loss": 0.3266,
      "step": 38076
    },
    {
      "epoch": 2.132209653936611,
      "grad_norm": 1.1348991394042969,
      "learning_rate": 9.524736842105264e-05,
      "loss": 0.4241,
      "step": 38077
    },
    {
      "epoch": 2.13226565124874,
      "grad_norm": 1.1503900289535522,
      "learning_rate": 9.52471052631579e-05,
      "loss": 0.4737,
      "step": 38078
    },
    {
      "epoch": 2.132321648560869,
      "grad_norm": 1.0026459693908691,
      "learning_rate": 9.524684210526317e-05,
      "loss": 0.2794,
      "step": 38079
    },
    {
      "epoch": 2.132377645872998,
      "grad_norm": 1.325727939605713,
      "learning_rate": 9.524657894736843e-05,
      "loss": 0.4621,
      "step": 38080
    },
    {
      "epoch": 2.132433643185127,
      "grad_norm": 1.221501350402832,
      "learning_rate": 9.524631578947369e-05,
      "loss": 0.385,
      "step": 38081
    },
    {
      "epoch": 2.132489640497256,
      "grad_norm": 1.2808952331542969,
      "learning_rate": 9.524605263157895e-05,
      "loss": 0.5001,
      "step": 38082
    },
    {
      "epoch": 2.132545637809385,
      "grad_norm": 1.339376449584961,
      "learning_rate": 9.524578947368421e-05,
      "loss": 0.3182,
      "step": 38083
    },
    {
      "epoch": 2.1326016351215142,
      "grad_norm": 1.396331548690796,
      "learning_rate": 9.524552631578948e-05,
      "loss": 0.6019,
      "step": 38084
    },
    {
      "epoch": 2.1326576324336433,
      "grad_norm": 1.290798306465149,
      "learning_rate": 9.524526315789474e-05,
      "loss": 0.4234,
      "step": 38085
    },
    {
      "epoch": 2.1327136297457723,
      "grad_norm": 1.3265392780303955,
      "learning_rate": 9.5245e-05,
      "loss": 0.471,
      "step": 38086
    },
    {
      "epoch": 2.1327696270579013,
      "grad_norm": 1.3487203121185303,
      "learning_rate": 9.524473684210526e-05,
      "loss": 0.3755,
      "step": 38087
    },
    {
      "epoch": 2.1328256243700303,
      "grad_norm": 1.2713758945465088,
      "learning_rate": 9.524447368421053e-05,
      "loss": 0.3466,
      "step": 38088
    },
    {
      "epoch": 2.1328816216821593,
      "grad_norm": 1.2795480489730835,
      "learning_rate": 9.524421052631579e-05,
      "loss": 0.4042,
      "step": 38089
    },
    {
      "epoch": 2.1329376189942884,
      "grad_norm": 1.3855855464935303,
      "learning_rate": 9.524394736842107e-05,
      "loss": 0.3244,
      "step": 38090
    },
    {
      "epoch": 2.1329936163064174,
      "grad_norm": 1.3090237379074097,
      "learning_rate": 9.524368421052631e-05,
      "loss": 0.4619,
      "step": 38091
    },
    {
      "epoch": 2.1330496136185464,
      "grad_norm": 1.1355621814727783,
      "learning_rate": 9.524342105263159e-05,
      "loss": 0.3734,
      "step": 38092
    },
    {
      "epoch": 2.1331056109306754,
      "grad_norm": 1.2517218589782715,
      "learning_rate": 9.524315789473685e-05,
      "loss": 0.4236,
      "step": 38093
    },
    {
      "epoch": 2.1331616082428044,
      "grad_norm": 1.0171056985855103,
      "learning_rate": 9.524289473684212e-05,
      "loss": 0.3167,
      "step": 38094
    },
    {
      "epoch": 2.1332176055549334,
      "grad_norm": 1.0247516632080078,
      "learning_rate": 9.524263157894738e-05,
      "loss": 0.3004,
      "step": 38095
    },
    {
      "epoch": 2.1332736028670625,
      "grad_norm": 1.3456711769104004,
      "learning_rate": 9.524236842105264e-05,
      "loss": 0.4945,
      "step": 38096
    },
    {
      "epoch": 2.1333296001791915,
      "grad_norm": 1.6759661436080933,
      "learning_rate": 9.52421052631579e-05,
      "loss": 0.5181,
      "step": 38097
    },
    {
      "epoch": 2.1333855974913205,
      "grad_norm": 1.3364152908325195,
      "learning_rate": 9.524184210526316e-05,
      "loss": 0.5071,
      "step": 38098
    },
    {
      "epoch": 2.1334415948034495,
      "grad_norm": 1.2659976482391357,
      "learning_rate": 9.524157894736843e-05,
      "loss": 0.3458,
      "step": 38099
    },
    {
      "epoch": 2.1334975921155785,
      "grad_norm": 1.069238543510437,
      "learning_rate": 9.524131578947369e-05,
      "loss": 0.4014,
      "step": 38100
    },
    {
      "epoch": 2.1335535894277076,
      "grad_norm": 1.1336287260055542,
      "learning_rate": 9.524105263157895e-05,
      "loss": 0.3361,
      "step": 38101
    },
    {
      "epoch": 2.1336095867398366,
      "grad_norm": 1.1548731327056885,
      "learning_rate": 9.524078947368421e-05,
      "loss": 0.3298,
      "step": 38102
    },
    {
      "epoch": 2.1336655840519656,
      "grad_norm": 1.1080031394958496,
      "learning_rate": 9.524052631578948e-05,
      "loss": 0.3494,
      "step": 38103
    },
    {
      "epoch": 2.1337215813640946,
      "grad_norm": 1.2724939584732056,
      "learning_rate": 9.524026315789474e-05,
      "loss": 0.4123,
      "step": 38104
    },
    {
      "epoch": 2.1337775786762236,
      "grad_norm": 1.1961098909378052,
      "learning_rate": 9.524e-05,
      "loss": 0.4559,
      "step": 38105
    },
    {
      "epoch": 2.1338335759883527,
      "grad_norm": 1.2123501300811768,
      "learning_rate": 9.523973684210526e-05,
      "loss": 0.4127,
      "step": 38106
    },
    {
      "epoch": 2.1338895733004817,
      "grad_norm": 1.1664334535598755,
      "learning_rate": 9.523947368421054e-05,
      "loss": 0.3587,
      "step": 38107
    },
    {
      "epoch": 2.1339455706126107,
      "grad_norm": 1.2061718702316284,
      "learning_rate": 9.52392105263158e-05,
      "loss": 0.4268,
      "step": 38108
    },
    {
      "epoch": 2.1340015679247397,
      "grad_norm": 1.4267489910125732,
      "learning_rate": 9.523894736842106e-05,
      "loss": 0.4335,
      "step": 38109
    },
    {
      "epoch": 2.1340575652368687,
      "grad_norm": 1.1007779836654663,
      "learning_rate": 9.523868421052632e-05,
      "loss": 0.3474,
      "step": 38110
    },
    {
      "epoch": 2.1341135625489978,
      "grad_norm": 1.1919065713882446,
      "learning_rate": 9.523842105263159e-05,
      "loss": 0.4011,
      "step": 38111
    },
    {
      "epoch": 2.1341695598611268,
      "grad_norm": 1.30142343044281,
      "learning_rate": 9.523815789473685e-05,
      "loss": 0.321,
      "step": 38112
    },
    {
      "epoch": 2.134225557173256,
      "grad_norm": 1.7247234582901,
      "learning_rate": 9.523789473684212e-05,
      "loss": 0.5547,
      "step": 38113
    },
    {
      "epoch": 2.134281554485385,
      "grad_norm": 17.529237747192383,
      "learning_rate": 9.523763157894737e-05,
      "loss": 0.5024,
      "step": 38114
    },
    {
      "epoch": 2.134337551797514,
      "grad_norm": 1.0507961511611938,
      "learning_rate": 9.523736842105263e-05,
      "loss": 0.4108,
      "step": 38115
    },
    {
      "epoch": 2.134393549109643,
      "grad_norm": 1.1761435270309448,
      "learning_rate": 9.52371052631579e-05,
      "loss": 0.3861,
      "step": 38116
    },
    {
      "epoch": 2.134449546421772,
      "grad_norm": 1.165257215499878,
      "learning_rate": 9.523684210526316e-05,
      "loss": 0.3681,
      "step": 38117
    },
    {
      "epoch": 2.134505543733901,
      "grad_norm": 1.3846526145935059,
      "learning_rate": 9.523657894736842e-05,
      "loss": 0.4921,
      "step": 38118
    },
    {
      "epoch": 2.13456154104603,
      "grad_norm": 1.1170088052749634,
      "learning_rate": 9.523631578947368e-05,
      "loss": 0.3884,
      "step": 38119
    },
    {
      "epoch": 2.134617538358159,
      "grad_norm": 1.1309449672698975,
      "learning_rate": 9.523605263157895e-05,
      "loss": 0.4525,
      "step": 38120
    },
    {
      "epoch": 2.134673535670288,
      "grad_norm": 7.3944091796875,
      "learning_rate": 9.523578947368421e-05,
      "loss": 0.3676,
      "step": 38121
    },
    {
      "epoch": 2.134729532982417,
      "grad_norm": 1.280097484588623,
      "learning_rate": 9.523552631578949e-05,
      "loss": 0.4587,
      "step": 38122
    },
    {
      "epoch": 2.134785530294546,
      "grad_norm": 1.298851490020752,
      "learning_rate": 9.523526315789473e-05,
      "loss": 0.5293,
      "step": 38123
    },
    {
      "epoch": 2.134841527606675,
      "grad_norm": 1.1458475589752197,
      "learning_rate": 9.5235e-05,
      "loss": 0.4269,
      "step": 38124
    },
    {
      "epoch": 2.134897524918804,
      "grad_norm": 1.4874486923217773,
      "learning_rate": 9.523473684210527e-05,
      "loss": 0.4289,
      "step": 38125
    },
    {
      "epoch": 2.134953522230933,
      "grad_norm": 1.2012462615966797,
      "learning_rate": 9.523447368421054e-05,
      "loss": 0.4538,
      "step": 38126
    },
    {
      "epoch": 2.135009519543062,
      "grad_norm": 0.940038800239563,
      "learning_rate": 9.52342105263158e-05,
      "loss": 0.2816,
      "step": 38127
    },
    {
      "epoch": 2.135065516855191,
      "grad_norm": 1.2139818668365479,
      "learning_rate": 9.523394736842106e-05,
      "loss": 0.4267,
      "step": 38128
    },
    {
      "epoch": 2.13512151416732,
      "grad_norm": 1.2821401357650757,
      "learning_rate": 9.523368421052632e-05,
      "loss": 0.4666,
      "step": 38129
    },
    {
      "epoch": 2.135177511479449,
      "grad_norm": 1.4604532718658447,
      "learning_rate": 9.523342105263159e-05,
      "loss": 0.4266,
      "step": 38130
    },
    {
      "epoch": 2.135233508791578,
      "grad_norm": 1.1600923538208008,
      "learning_rate": 9.523315789473685e-05,
      "loss": 0.3544,
      "step": 38131
    },
    {
      "epoch": 2.135289506103707,
      "grad_norm": 1.4497381448745728,
      "learning_rate": 9.523289473684211e-05,
      "loss": 0.6255,
      "step": 38132
    },
    {
      "epoch": 2.135345503415836,
      "grad_norm": 1.1776341199874878,
      "learning_rate": 9.523263157894737e-05,
      "loss": 0.4341,
      "step": 38133
    },
    {
      "epoch": 2.135401500727965,
      "grad_norm": 1.1329070329666138,
      "learning_rate": 9.523236842105263e-05,
      "loss": 0.3535,
      "step": 38134
    },
    {
      "epoch": 2.135457498040094,
      "grad_norm": 1.3169541358947754,
      "learning_rate": 9.52321052631579e-05,
      "loss": 0.4009,
      "step": 38135
    },
    {
      "epoch": 2.1355134953522232,
      "grad_norm": 1.129478096961975,
      "learning_rate": 9.523184210526316e-05,
      "loss": 0.3754,
      "step": 38136
    },
    {
      "epoch": 2.1355694926643523,
      "grad_norm": 1.1229019165039062,
      "learning_rate": 9.523157894736842e-05,
      "loss": 0.3577,
      "step": 38137
    },
    {
      "epoch": 2.1356254899764813,
      "grad_norm": 1.3719444274902344,
      "learning_rate": 9.523131578947368e-05,
      "loss": 0.4032,
      "step": 38138
    },
    {
      "epoch": 2.1356814872886103,
      "grad_norm": 1.0451343059539795,
      "learning_rate": 9.523105263157896e-05,
      "loss": 0.3173,
      "step": 38139
    },
    {
      "epoch": 2.1357374846007393,
      "grad_norm": 1.2154014110565186,
      "learning_rate": 9.523078947368422e-05,
      "loss": 0.3402,
      "step": 38140
    },
    {
      "epoch": 2.1357934819128683,
      "grad_norm": 1.2046551704406738,
      "learning_rate": 9.523052631578948e-05,
      "loss": 0.2937,
      "step": 38141
    },
    {
      "epoch": 2.1358494792249973,
      "grad_norm": 1.037703275680542,
      "learning_rate": 9.523026315789473e-05,
      "loss": 0.4117,
      "step": 38142
    },
    {
      "epoch": 2.1359054765371264,
      "grad_norm": 1.2204055786132812,
      "learning_rate": 9.523000000000001e-05,
      "loss": 0.3595,
      "step": 38143
    },
    {
      "epoch": 2.1359614738492554,
      "grad_norm": 1.3470975160598755,
      "learning_rate": 9.522973684210527e-05,
      "loss": 0.3916,
      "step": 38144
    },
    {
      "epoch": 2.1360174711613844,
      "grad_norm": 1.1628057956695557,
      "learning_rate": 9.522947368421054e-05,
      "loss": 0.4497,
      "step": 38145
    },
    {
      "epoch": 2.1360734684735134,
      "grad_norm": 1.1477094888687134,
      "learning_rate": 9.522921052631579e-05,
      "loss": 0.3363,
      "step": 38146
    },
    {
      "epoch": 2.1361294657856424,
      "grad_norm": 0.9284495115280151,
      "learning_rate": 9.522894736842106e-05,
      "loss": 0.3003,
      "step": 38147
    },
    {
      "epoch": 2.1361854630977715,
      "grad_norm": 1.283875823020935,
      "learning_rate": 9.522868421052632e-05,
      "loss": 0.3802,
      "step": 38148
    },
    {
      "epoch": 2.1362414604099005,
      "grad_norm": 1.2610158920288086,
      "learning_rate": 9.52284210526316e-05,
      "loss": 0.4136,
      "step": 38149
    },
    {
      "epoch": 2.1362974577220295,
      "grad_norm": 1.7163478136062622,
      "learning_rate": 9.522815789473685e-05,
      "loss": 0.391,
      "step": 38150
    },
    {
      "epoch": 2.1363534550341585,
      "grad_norm": 1.3285573720932007,
      "learning_rate": 9.52278947368421e-05,
      "loss": 0.583,
      "step": 38151
    },
    {
      "epoch": 2.1364094523462875,
      "grad_norm": 1.1929993629455566,
      "learning_rate": 9.522763157894737e-05,
      "loss": 0.4423,
      "step": 38152
    },
    {
      "epoch": 2.1364654496584166,
      "grad_norm": 1.0414402484893799,
      "learning_rate": 9.522736842105263e-05,
      "loss": 0.3851,
      "step": 38153
    },
    {
      "epoch": 2.1365214469705456,
      "grad_norm": 1.5661691427230835,
      "learning_rate": 9.52271052631579e-05,
      "loss": 0.4844,
      "step": 38154
    },
    {
      "epoch": 2.1365774442826746,
      "grad_norm": 1.4953093528747559,
      "learning_rate": 9.522684210526317e-05,
      "loss": 0.4092,
      "step": 38155
    },
    {
      "epoch": 2.1366334415948036,
      "grad_norm": 1.1428807973861694,
      "learning_rate": 9.522657894736843e-05,
      "loss": 0.4427,
      "step": 38156
    },
    {
      "epoch": 2.1366894389069326,
      "grad_norm": 1.4431251287460327,
      "learning_rate": 9.522631578947368e-05,
      "loss": 0.3928,
      "step": 38157
    },
    {
      "epoch": 2.1367454362190617,
      "grad_norm": 1.0334142446517944,
      "learning_rate": 9.522605263157896e-05,
      "loss": 0.3657,
      "step": 38158
    },
    {
      "epoch": 2.1368014335311907,
      "grad_norm": 1.386918544769287,
      "learning_rate": 9.522578947368422e-05,
      "loss": 0.5782,
      "step": 38159
    },
    {
      "epoch": 2.1368574308433197,
      "grad_norm": 1.8670121431350708,
      "learning_rate": 9.522552631578948e-05,
      "loss": 0.5832,
      "step": 38160
    },
    {
      "epoch": 2.1369134281554487,
      "grad_norm": 1.3642046451568604,
      "learning_rate": 9.522526315789474e-05,
      "loss": 0.4891,
      "step": 38161
    },
    {
      "epoch": 2.1369694254675777,
      "grad_norm": 1.1905229091644287,
      "learning_rate": 9.522500000000001e-05,
      "loss": 0.394,
      "step": 38162
    },
    {
      "epoch": 2.1370254227797068,
      "grad_norm": 1.1040364503860474,
      "learning_rate": 9.522473684210527e-05,
      "loss": 0.4327,
      "step": 38163
    },
    {
      "epoch": 2.1370814200918358,
      "grad_norm": 1.4536675214767456,
      "learning_rate": 9.522447368421053e-05,
      "loss": 0.3779,
      "step": 38164
    },
    {
      "epoch": 2.137137417403965,
      "grad_norm": 1.1703200340270996,
      "learning_rate": 9.522421052631579e-05,
      "loss": 0.5,
      "step": 38165
    },
    {
      "epoch": 2.137193414716094,
      "grad_norm": 1.2313188314437866,
      "learning_rate": 9.522394736842106e-05,
      "loss": 0.4218,
      "step": 38166
    },
    {
      "epoch": 2.137249412028223,
      "grad_norm": 1.3737343549728394,
      "learning_rate": 9.522368421052632e-05,
      "loss": 0.4801,
      "step": 38167
    },
    {
      "epoch": 2.137305409340352,
      "grad_norm": 1.1587690114974976,
      "learning_rate": 9.522342105263158e-05,
      "loss": 0.449,
      "step": 38168
    },
    {
      "epoch": 2.137361406652481,
      "grad_norm": 1.213584065437317,
      "learning_rate": 9.522315789473684e-05,
      "loss": 0.6608,
      "step": 38169
    },
    {
      "epoch": 2.13741740396461,
      "grad_norm": 1.144053339958191,
      "learning_rate": 9.52228947368421e-05,
      "loss": 0.3027,
      "step": 38170
    },
    {
      "epoch": 2.137473401276739,
      "grad_norm": 1.5866066217422485,
      "learning_rate": 9.522263157894738e-05,
      "loss": 0.3931,
      "step": 38171
    },
    {
      "epoch": 2.137529398588868,
      "grad_norm": 1.2629061937332153,
      "learning_rate": 9.522236842105264e-05,
      "loss": 0.3728,
      "step": 38172
    },
    {
      "epoch": 2.137585395900997,
      "grad_norm": 1.3687466382980347,
      "learning_rate": 9.52221052631579e-05,
      "loss": 0.5041,
      "step": 38173
    },
    {
      "epoch": 2.137641393213126,
      "grad_norm": 1.656934142112732,
      "learning_rate": 9.522184210526315e-05,
      "loss": 0.405,
      "step": 38174
    },
    {
      "epoch": 2.137697390525255,
      "grad_norm": 1.3004800081253052,
      "learning_rate": 9.522157894736843e-05,
      "loss": 0.4654,
      "step": 38175
    },
    {
      "epoch": 2.137753387837384,
      "grad_norm": 1.2191650867462158,
      "learning_rate": 9.522131578947369e-05,
      "loss": 0.4331,
      "step": 38176
    },
    {
      "epoch": 2.137809385149513,
      "grad_norm": 1.1870242357254028,
      "learning_rate": 9.522105263157896e-05,
      "loss": 0.4856,
      "step": 38177
    },
    {
      "epoch": 2.137865382461642,
      "grad_norm": 1.1643965244293213,
      "learning_rate": 9.522078947368421e-05,
      "loss": 0.5596,
      "step": 38178
    },
    {
      "epoch": 2.137921379773771,
      "grad_norm": 1.1035288572311401,
      "learning_rate": 9.522052631578948e-05,
      "loss": 0.3734,
      "step": 38179
    },
    {
      "epoch": 2.1379773770859,
      "grad_norm": 1.2940701246261597,
      "learning_rate": 9.522026315789474e-05,
      "loss": 0.4103,
      "step": 38180
    },
    {
      "epoch": 2.138033374398029,
      "grad_norm": 1.2635301351547241,
      "learning_rate": 9.522000000000001e-05,
      "loss": 0.3733,
      "step": 38181
    },
    {
      "epoch": 2.138089371710158,
      "grad_norm": 1.1572139263153076,
      "learning_rate": 9.521973684210527e-05,
      "loss": 0.3266,
      "step": 38182
    },
    {
      "epoch": 2.138145369022287,
      "grad_norm": 1.041929841041565,
      "learning_rate": 9.521947368421053e-05,
      "loss": 0.3002,
      "step": 38183
    },
    {
      "epoch": 2.138201366334416,
      "grad_norm": 0.955590546131134,
      "learning_rate": 9.521921052631579e-05,
      "loss": 0.2736,
      "step": 38184
    },
    {
      "epoch": 2.138257363646545,
      "grad_norm": 1.9749939441680908,
      "learning_rate": 9.521894736842105e-05,
      "loss": 0.3939,
      "step": 38185
    },
    {
      "epoch": 2.1383133609586737,
      "grad_norm": 1.3457142114639282,
      "learning_rate": 9.521868421052633e-05,
      "loss": 0.422,
      "step": 38186
    },
    {
      "epoch": 2.138369358270803,
      "grad_norm": 1.1366219520568848,
      "learning_rate": 9.521842105263159e-05,
      "loss": 0.3412,
      "step": 38187
    },
    {
      "epoch": 2.138425355582932,
      "grad_norm": 1.5393035411834717,
      "learning_rate": 9.521815789473684e-05,
      "loss": 0.4015,
      "step": 38188
    },
    {
      "epoch": 2.1384813528950612,
      "grad_norm": 1.14662766456604,
      "learning_rate": 9.52178947368421e-05,
      "loss": 0.329,
      "step": 38189
    },
    {
      "epoch": 2.13853735020719,
      "grad_norm": 1.3876137733459473,
      "learning_rate": 9.521763157894738e-05,
      "loss": 0.4296,
      "step": 38190
    },
    {
      "epoch": 2.1385933475193193,
      "grad_norm": 1.0128370523452759,
      "learning_rate": 9.521736842105264e-05,
      "loss": 0.3615,
      "step": 38191
    },
    {
      "epoch": 2.138649344831448,
      "grad_norm": 1.2008684873580933,
      "learning_rate": 9.52171052631579e-05,
      "loss": 0.3374,
      "step": 38192
    },
    {
      "epoch": 2.1387053421435773,
      "grad_norm": 1.228963017463684,
      "learning_rate": 9.521684210526316e-05,
      "loss": 0.4207,
      "step": 38193
    },
    {
      "epoch": 2.138761339455706,
      "grad_norm": 1.266869068145752,
      "learning_rate": 9.521657894736843e-05,
      "loss": 0.4245,
      "step": 38194
    },
    {
      "epoch": 2.1388173367678354,
      "grad_norm": 1.4772392511367798,
      "learning_rate": 9.521631578947369e-05,
      "loss": 0.4317,
      "step": 38195
    },
    {
      "epoch": 2.138873334079964,
      "grad_norm": 1.2737996578216553,
      "learning_rate": 9.521605263157895e-05,
      "loss": 0.5218,
      "step": 38196
    },
    {
      "epoch": 2.1389293313920934,
      "grad_norm": 1.0991106033325195,
      "learning_rate": 9.521578947368421e-05,
      "loss": 0.3129,
      "step": 38197
    },
    {
      "epoch": 2.138985328704222,
      "grad_norm": 1.2733256816864014,
      "learning_rate": 9.521552631578948e-05,
      "loss": 0.3836,
      "step": 38198
    },
    {
      "epoch": 2.1390413260163514,
      "grad_norm": 1.081042766571045,
      "learning_rate": 9.521526315789474e-05,
      "loss": 0.2377,
      "step": 38199
    },
    {
      "epoch": 2.13909732332848,
      "grad_norm": 1.481172800064087,
      "learning_rate": 9.521500000000002e-05,
      "loss": 0.4217,
      "step": 38200
    },
    {
      "epoch": 2.1391533206406095,
      "grad_norm": 1.7473317384719849,
      "learning_rate": 9.521473684210526e-05,
      "loss": 0.3945,
      "step": 38201
    },
    {
      "epoch": 2.139209317952738,
      "grad_norm": 1.254936933517456,
      "learning_rate": 9.521447368421052e-05,
      "loss": 0.3956,
      "step": 38202
    },
    {
      "epoch": 2.1392653152648675,
      "grad_norm": 1.0650924444198608,
      "learning_rate": 9.52142105263158e-05,
      "loss": 0.3619,
      "step": 38203
    },
    {
      "epoch": 2.139321312576996,
      "grad_norm": 1.7617086172103882,
      "learning_rate": 9.521394736842105e-05,
      "loss": 0.4086,
      "step": 38204
    },
    {
      "epoch": 2.1393773098891256,
      "grad_norm": 1.22715163230896,
      "learning_rate": 9.521368421052633e-05,
      "loss": 0.3465,
      "step": 38205
    },
    {
      "epoch": 2.139433307201254,
      "grad_norm": 1.155155062675476,
      "learning_rate": 9.521342105263157e-05,
      "loss": 0.3816,
      "step": 38206
    },
    {
      "epoch": 2.139489304513383,
      "grad_norm": 1.2315603494644165,
      "learning_rate": 9.521315789473685e-05,
      "loss": 0.3732,
      "step": 38207
    },
    {
      "epoch": 2.139545301825512,
      "grad_norm": 1.3409041166305542,
      "learning_rate": 9.521289473684211e-05,
      "loss": 0.3569,
      "step": 38208
    },
    {
      "epoch": 2.139601299137641,
      "grad_norm": 1.539539098739624,
      "learning_rate": 9.521263157894738e-05,
      "loss": 0.4029,
      "step": 38209
    },
    {
      "epoch": 2.13965729644977,
      "grad_norm": 1.4609636068344116,
      "learning_rate": 9.521236842105263e-05,
      "loss": 0.473,
      "step": 38210
    },
    {
      "epoch": 2.1397132937618992,
      "grad_norm": 1.2497127056121826,
      "learning_rate": 9.52121052631579e-05,
      "loss": 0.4548,
      "step": 38211
    },
    {
      "epoch": 2.1397692910740282,
      "grad_norm": 1.1959890127182007,
      "learning_rate": 9.521184210526316e-05,
      "loss": 0.4805,
      "step": 38212
    },
    {
      "epoch": 2.1398252883861573,
      "grad_norm": 1.2456157207489014,
      "learning_rate": 9.521157894736843e-05,
      "loss": 0.6036,
      "step": 38213
    },
    {
      "epoch": 2.1398812856982863,
      "grad_norm": 1.2958616018295288,
      "learning_rate": 9.521131578947369e-05,
      "loss": 0.4659,
      "step": 38214
    },
    {
      "epoch": 2.1399372830104153,
      "grad_norm": 1.251543402671814,
      "learning_rate": 9.521105263157895e-05,
      "loss": 0.3911,
      "step": 38215
    },
    {
      "epoch": 2.1399932803225443,
      "grad_norm": 1.1401814222335815,
      "learning_rate": 9.521078947368421e-05,
      "loss": 0.4113,
      "step": 38216
    },
    {
      "epoch": 2.1400492776346733,
      "grad_norm": 1.1797345876693726,
      "learning_rate": 9.521052631578949e-05,
      "loss": 0.3466,
      "step": 38217
    },
    {
      "epoch": 2.1401052749468024,
      "grad_norm": 1.1843957901000977,
      "learning_rate": 9.521026315789475e-05,
      "loss": 0.3757,
      "step": 38218
    },
    {
      "epoch": 2.1401612722589314,
      "grad_norm": 1.1159995794296265,
      "learning_rate": 9.521e-05,
      "loss": 0.3955,
      "step": 38219
    },
    {
      "epoch": 2.1402172695710604,
      "grad_norm": 2.6713151931762695,
      "learning_rate": 9.520973684210526e-05,
      "loss": 0.5151,
      "step": 38220
    },
    {
      "epoch": 2.1402732668831894,
      "grad_norm": 1.3884177207946777,
      "learning_rate": 9.520947368421052e-05,
      "loss": 0.4047,
      "step": 38221
    },
    {
      "epoch": 2.1403292641953184,
      "grad_norm": 1.573036551475525,
      "learning_rate": 9.52092105263158e-05,
      "loss": 0.4764,
      "step": 38222
    },
    {
      "epoch": 2.1403852615074475,
      "grad_norm": 2.4388699531555176,
      "learning_rate": 9.520894736842106e-05,
      "loss": 0.3302,
      "step": 38223
    },
    {
      "epoch": 2.1404412588195765,
      "grad_norm": 1.01044499874115,
      "learning_rate": 9.520868421052632e-05,
      "loss": 0.4048,
      "step": 38224
    },
    {
      "epoch": 2.1404972561317055,
      "grad_norm": 1.3563491106033325,
      "learning_rate": 9.520842105263158e-05,
      "loss": 0.3775,
      "step": 38225
    },
    {
      "epoch": 2.1405532534438345,
      "grad_norm": 1.245753288269043,
      "learning_rate": 9.520815789473685e-05,
      "loss": 0.3903,
      "step": 38226
    },
    {
      "epoch": 2.1406092507559635,
      "grad_norm": 1.3333702087402344,
      "learning_rate": 9.520789473684211e-05,
      "loss": 0.3899,
      "step": 38227
    },
    {
      "epoch": 2.1406652480680926,
      "grad_norm": 1.3277714252471924,
      "learning_rate": 9.520763157894737e-05,
      "loss": 0.3534,
      "step": 38228
    },
    {
      "epoch": 2.1407212453802216,
      "grad_norm": 1.1180626153945923,
      "learning_rate": 9.520736842105263e-05,
      "loss": 0.3965,
      "step": 38229
    },
    {
      "epoch": 2.1407772426923506,
      "grad_norm": 1.2243865728378296,
      "learning_rate": 9.52071052631579e-05,
      "loss": 0.416,
      "step": 38230
    },
    {
      "epoch": 2.1408332400044796,
      "grad_norm": 1.4420181512832642,
      "learning_rate": 9.520684210526316e-05,
      "loss": 0.4152,
      "step": 38231
    },
    {
      "epoch": 2.1408892373166086,
      "grad_norm": 1.1621487140655518,
      "learning_rate": 9.520657894736844e-05,
      "loss": 0.3386,
      "step": 38232
    },
    {
      "epoch": 2.1409452346287376,
      "grad_norm": 1.1099042892456055,
      "learning_rate": 9.520631578947368e-05,
      "loss": 0.356,
      "step": 38233
    },
    {
      "epoch": 2.1410012319408667,
      "grad_norm": 1.3326021432876587,
      "learning_rate": 9.520605263157896e-05,
      "loss": 0.549,
      "step": 38234
    },
    {
      "epoch": 2.1410572292529957,
      "grad_norm": 1.1555290222167969,
      "learning_rate": 9.520578947368421e-05,
      "loss": 0.3943,
      "step": 38235
    },
    {
      "epoch": 2.1411132265651247,
      "grad_norm": 1.0788766145706177,
      "learning_rate": 9.520552631578947e-05,
      "loss": 0.4022,
      "step": 38236
    },
    {
      "epoch": 2.1411692238772537,
      "grad_norm": 1.3333680629730225,
      "learning_rate": 9.520526315789475e-05,
      "loss": 0.4609,
      "step": 38237
    },
    {
      "epoch": 2.1412252211893827,
      "grad_norm": 1.1839755773544312,
      "learning_rate": 9.5205e-05,
      "loss": 0.4888,
      "step": 38238
    },
    {
      "epoch": 2.1412812185015118,
      "grad_norm": 1.1685596704483032,
      "learning_rate": 9.520473684210527e-05,
      "loss": 0.3532,
      "step": 38239
    },
    {
      "epoch": 2.141337215813641,
      "grad_norm": 1.1523849964141846,
      "learning_rate": 9.520447368421053e-05,
      "loss": 0.4416,
      "step": 38240
    },
    {
      "epoch": 2.14139321312577,
      "grad_norm": 1.5175058841705322,
      "learning_rate": 9.52042105263158e-05,
      "loss": 0.3381,
      "step": 38241
    },
    {
      "epoch": 2.141449210437899,
      "grad_norm": 1.2360601425170898,
      "learning_rate": 9.520394736842106e-05,
      "loss": 0.3368,
      "step": 38242
    },
    {
      "epoch": 2.141505207750028,
      "grad_norm": 1.612764596939087,
      "learning_rate": 9.520368421052632e-05,
      "loss": 0.4384,
      "step": 38243
    },
    {
      "epoch": 2.141561205062157,
      "grad_norm": 1.1987849473953247,
      "learning_rate": 9.520342105263158e-05,
      "loss": 0.4229,
      "step": 38244
    },
    {
      "epoch": 2.141617202374286,
      "grad_norm": 1.4699829816818237,
      "learning_rate": 9.520315789473685e-05,
      "loss": 0.3507,
      "step": 38245
    },
    {
      "epoch": 2.141673199686415,
      "grad_norm": 1.0830490589141846,
      "learning_rate": 9.520289473684211e-05,
      "loss": 0.3972,
      "step": 38246
    },
    {
      "epoch": 2.141729196998544,
      "grad_norm": 1.5250779390335083,
      "learning_rate": 9.520263157894737e-05,
      "loss": 0.3755,
      "step": 38247
    },
    {
      "epoch": 2.141785194310673,
      "grad_norm": 1.159103512763977,
      "learning_rate": 9.520236842105263e-05,
      "loss": 0.4339,
      "step": 38248
    },
    {
      "epoch": 2.141841191622802,
      "grad_norm": 1.3004153966903687,
      "learning_rate": 9.52021052631579e-05,
      "loss": 0.5246,
      "step": 38249
    },
    {
      "epoch": 2.141897188934931,
      "grad_norm": 1.6057721376419067,
      "learning_rate": 9.520184210526316e-05,
      "loss": 0.6392,
      "step": 38250
    },
    {
      "epoch": 2.14195318624706,
      "grad_norm": 1.3801565170288086,
      "learning_rate": 9.520157894736842e-05,
      "loss": 0.4938,
      "step": 38251
    },
    {
      "epoch": 2.142009183559189,
      "grad_norm": 1.3770190477371216,
      "learning_rate": 9.520131578947368e-05,
      "loss": 0.4209,
      "step": 38252
    },
    {
      "epoch": 2.142065180871318,
      "grad_norm": 1.5340890884399414,
      "learning_rate": 9.520105263157894e-05,
      "loss": 0.3501,
      "step": 38253
    },
    {
      "epoch": 2.142121178183447,
      "grad_norm": 1.3469096422195435,
      "learning_rate": 9.520078947368422e-05,
      "loss": 0.3959,
      "step": 38254
    },
    {
      "epoch": 2.142177175495576,
      "grad_norm": 2.013986349105835,
      "learning_rate": 9.520052631578948e-05,
      "loss": 0.3397,
      "step": 38255
    },
    {
      "epoch": 2.142233172807705,
      "grad_norm": 1.247554063796997,
      "learning_rate": 9.520026315789474e-05,
      "loss": 0.4052,
      "step": 38256
    },
    {
      "epoch": 2.142289170119834,
      "grad_norm": 1.3199414014816284,
      "learning_rate": 9.52e-05,
      "loss": 0.4163,
      "step": 38257
    },
    {
      "epoch": 2.142345167431963,
      "grad_norm": 1.2954035997390747,
      "learning_rate": 9.519973684210527e-05,
      "loss": 0.3952,
      "step": 38258
    },
    {
      "epoch": 2.142401164744092,
      "grad_norm": 3.7618207931518555,
      "learning_rate": 9.519947368421053e-05,
      "loss": 0.3623,
      "step": 38259
    },
    {
      "epoch": 2.142457162056221,
      "grad_norm": 1.317212462425232,
      "learning_rate": 9.51992105263158e-05,
      "loss": 0.3576,
      "step": 38260
    },
    {
      "epoch": 2.14251315936835,
      "grad_norm": 1.2333412170410156,
      "learning_rate": 9.519894736842105e-05,
      "loss": 0.3835,
      "step": 38261
    },
    {
      "epoch": 2.142569156680479,
      "grad_norm": 1.303763747215271,
      "learning_rate": 9.519868421052632e-05,
      "loss": 0.4328,
      "step": 38262
    },
    {
      "epoch": 2.1426251539926082,
      "grad_norm": 1.2166461944580078,
      "learning_rate": 9.519842105263158e-05,
      "loss": 0.4558,
      "step": 38263
    },
    {
      "epoch": 2.1426811513047372,
      "grad_norm": 1.468721628189087,
      "learning_rate": 9.519815789473686e-05,
      "loss": 0.3667,
      "step": 38264
    },
    {
      "epoch": 2.1427371486168663,
      "grad_norm": 1.43351411819458,
      "learning_rate": 9.51978947368421e-05,
      "loss": 0.4482,
      "step": 38265
    },
    {
      "epoch": 2.1427931459289953,
      "grad_norm": 1.1382423639297485,
      "learning_rate": 9.519763157894737e-05,
      "loss": 0.3623,
      "step": 38266
    },
    {
      "epoch": 2.1428491432411243,
      "grad_norm": 1.515245795249939,
      "learning_rate": 9.519736842105263e-05,
      "loss": 0.4972,
      "step": 38267
    },
    {
      "epoch": 2.1429051405532533,
      "grad_norm": 1.1439354419708252,
      "learning_rate": 9.519710526315791e-05,
      "loss": 0.2823,
      "step": 38268
    },
    {
      "epoch": 2.1429611378653823,
      "grad_norm": 1.1343514919281006,
      "learning_rate": 9.519684210526317e-05,
      "loss": 0.5263,
      "step": 38269
    },
    {
      "epoch": 2.1430171351775114,
      "grad_norm": 1.220656394958496,
      "learning_rate": 9.519657894736843e-05,
      "loss": 0.4349,
      "step": 38270
    },
    {
      "epoch": 2.1430731324896404,
      "grad_norm": 1.3684910535812378,
      "learning_rate": 9.519631578947369e-05,
      "loss": 0.4427,
      "step": 38271
    },
    {
      "epoch": 2.1431291298017694,
      "grad_norm": 1.428341269493103,
      "learning_rate": 9.519605263157895e-05,
      "loss": 0.317,
      "step": 38272
    },
    {
      "epoch": 2.1431851271138984,
      "grad_norm": 1.359214186668396,
      "learning_rate": 9.519578947368422e-05,
      "loss": 0.5043,
      "step": 38273
    },
    {
      "epoch": 2.1432411244260274,
      "grad_norm": 1.393699288368225,
      "learning_rate": 9.519552631578948e-05,
      "loss": 0.5634,
      "step": 38274
    },
    {
      "epoch": 2.1432971217381565,
      "grad_norm": 1.0113399028778076,
      "learning_rate": 9.519526315789474e-05,
      "loss": 0.3596,
      "step": 38275
    },
    {
      "epoch": 2.1433531190502855,
      "grad_norm": 1.1781200170516968,
      "learning_rate": 9.5195e-05,
      "loss": 0.515,
      "step": 38276
    },
    {
      "epoch": 2.1434091163624145,
      "grad_norm": 1.2815715074539185,
      "learning_rate": 9.519473684210527e-05,
      "loss": 0.7271,
      "step": 38277
    },
    {
      "epoch": 2.1434651136745435,
      "grad_norm": 0.9573467373847961,
      "learning_rate": 9.519447368421053e-05,
      "loss": 0.3249,
      "step": 38278
    },
    {
      "epoch": 2.1435211109866725,
      "grad_norm": 1.0952003002166748,
      "learning_rate": 9.519421052631579e-05,
      "loss": 0.3179,
      "step": 38279
    },
    {
      "epoch": 2.1435771082988015,
      "grad_norm": 1.7654109001159668,
      "learning_rate": 9.519394736842105e-05,
      "loss": 0.5171,
      "step": 38280
    },
    {
      "epoch": 2.1436331056109306,
      "grad_norm": 1.0280646085739136,
      "learning_rate": 9.519368421052632e-05,
      "loss": 0.347,
      "step": 38281
    },
    {
      "epoch": 2.1436891029230596,
      "grad_norm": 1.328781008720398,
      "learning_rate": 9.519342105263158e-05,
      "loss": 0.4942,
      "step": 38282
    },
    {
      "epoch": 2.1437451002351886,
      "grad_norm": 1.1049416065216064,
      "learning_rate": 9.519315789473684e-05,
      "loss": 0.4046,
      "step": 38283
    },
    {
      "epoch": 2.1438010975473176,
      "grad_norm": 1.1152710914611816,
      "learning_rate": 9.51928947368421e-05,
      "loss": 0.3752,
      "step": 38284
    },
    {
      "epoch": 2.1438570948594466,
      "grad_norm": 1.3638591766357422,
      "learning_rate": 9.519263157894738e-05,
      "loss": 0.4311,
      "step": 38285
    },
    {
      "epoch": 2.1439130921715757,
      "grad_norm": 1.081524133682251,
      "learning_rate": 9.519236842105264e-05,
      "loss": 0.3546,
      "step": 38286
    },
    {
      "epoch": 2.1439690894837047,
      "grad_norm": 1.1396889686584473,
      "learning_rate": 9.519210526315791e-05,
      "loss": 0.3938,
      "step": 38287
    },
    {
      "epoch": 2.1440250867958337,
      "grad_norm": 1.2552344799041748,
      "learning_rate": 9.519184210526316e-05,
      "loss": 0.3956,
      "step": 38288
    },
    {
      "epoch": 2.1440810841079627,
      "grad_norm": 1.2583619356155396,
      "learning_rate": 9.519157894736842e-05,
      "loss": 0.3633,
      "step": 38289
    },
    {
      "epoch": 2.1441370814200917,
      "grad_norm": 1.4059014320373535,
      "learning_rate": 9.519131578947369e-05,
      "loss": 0.3834,
      "step": 38290
    },
    {
      "epoch": 2.1441930787322208,
      "grad_norm": 22.061979293823242,
      "learning_rate": 9.519105263157895e-05,
      "loss": 0.482,
      "step": 38291
    },
    {
      "epoch": 2.1442490760443498,
      "grad_norm": 1.2164620161056519,
      "learning_rate": 9.519078947368422e-05,
      "loss": 0.4478,
      "step": 38292
    },
    {
      "epoch": 2.144305073356479,
      "grad_norm": 1.2906829118728638,
      "learning_rate": 9.519052631578947e-05,
      "loss": 0.4366,
      "step": 38293
    },
    {
      "epoch": 2.144361070668608,
      "grad_norm": 1.21834397315979,
      "learning_rate": 9.519026315789474e-05,
      "loss": 0.3554,
      "step": 38294
    },
    {
      "epoch": 2.144417067980737,
      "grad_norm": 1.115110158920288,
      "learning_rate": 9.519e-05,
      "loss": 0.3624,
      "step": 38295
    },
    {
      "epoch": 2.144473065292866,
      "grad_norm": 2.6654458045959473,
      "learning_rate": 9.518973684210527e-05,
      "loss": 0.4215,
      "step": 38296
    },
    {
      "epoch": 2.144529062604995,
      "grad_norm": 1.1112185716629028,
      "learning_rate": 9.518947368421053e-05,
      "loss": 0.3235,
      "step": 38297
    },
    {
      "epoch": 2.144585059917124,
      "grad_norm": 1.0276960134506226,
      "learning_rate": 9.51892105263158e-05,
      "loss": 0.3116,
      "step": 38298
    },
    {
      "epoch": 2.144641057229253,
      "grad_norm": 1.3983666896820068,
      "learning_rate": 9.518894736842105e-05,
      "loss": 0.4774,
      "step": 38299
    },
    {
      "epoch": 2.144697054541382,
      "grad_norm": 1.3449548482894897,
      "learning_rate": 9.518868421052633e-05,
      "loss": 0.4011,
      "step": 38300
    },
    {
      "epoch": 2.144753051853511,
      "grad_norm": 1.0634759664535522,
      "learning_rate": 9.518842105263159e-05,
      "loss": 0.3559,
      "step": 38301
    },
    {
      "epoch": 2.14480904916564,
      "grad_norm": 1.1814496517181396,
      "learning_rate": 9.518815789473685e-05,
      "loss": 0.4522,
      "step": 38302
    },
    {
      "epoch": 2.144865046477769,
      "grad_norm": 1.2169724702835083,
      "learning_rate": 9.51878947368421e-05,
      "loss": 0.3815,
      "step": 38303
    },
    {
      "epoch": 2.144921043789898,
      "grad_norm": 1.457496166229248,
      "learning_rate": 9.518763157894738e-05,
      "loss": 0.4407,
      "step": 38304
    },
    {
      "epoch": 2.144977041102027,
      "grad_norm": 1.0918076038360596,
      "learning_rate": 9.518736842105264e-05,
      "loss": 0.5094,
      "step": 38305
    },
    {
      "epoch": 2.145033038414156,
      "grad_norm": 1.5331761837005615,
      "learning_rate": 9.51871052631579e-05,
      "loss": 0.4693,
      "step": 38306
    },
    {
      "epoch": 2.145089035726285,
      "grad_norm": 1.241992712020874,
      "learning_rate": 9.518684210526316e-05,
      "loss": 0.3811,
      "step": 38307
    },
    {
      "epoch": 2.145145033038414,
      "grad_norm": 1.1199181079864502,
      "learning_rate": 9.518657894736842e-05,
      "loss": 0.355,
      "step": 38308
    },
    {
      "epoch": 2.145201030350543,
      "grad_norm": 1.2314311265945435,
      "learning_rate": 9.518631578947369e-05,
      "loss": 0.3225,
      "step": 38309
    },
    {
      "epoch": 2.145257027662672,
      "grad_norm": 1.224465250968933,
      "learning_rate": 9.518605263157895e-05,
      "loss": 0.4511,
      "step": 38310
    },
    {
      "epoch": 2.145313024974801,
      "grad_norm": 1.66157865524292,
      "learning_rate": 9.518578947368421e-05,
      "loss": 0.3905,
      "step": 38311
    },
    {
      "epoch": 2.14536902228693,
      "grad_norm": 1.0555822849273682,
      "learning_rate": 9.518552631578947e-05,
      "loss": 0.4319,
      "step": 38312
    },
    {
      "epoch": 2.145425019599059,
      "grad_norm": 1.0612894296646118,
      "learning_rate": 9.518526315789474e-05,
      "loss": 0.3882,
      "step": 38313
    },
    {
      "epoch": 2.145481016911188,
      "grad_norm": 1.2183176279067993,
      "learning_rate": 9.5185e-05,
      "loss": 0.3918,
      "step": 38314
    },
    {
      "epoch": 2.145537014223317,
      "grad_norm": 1.1744695901870728,
      "learning_rate": 9.518473684210528e-05,
      "loss": 0.4009,
      "step": 38315
    },
    {
      "epoch": 2.1455930115354462,
      "grad_norm": 1.217829704284668,
      "learning_rate": 9.518447368421052e-05,
      "loss": 0.4671,
      "step": 38316
    },
    {
      "epoch": 2.1456490088475753,
      "grad_norm": 1.1557682752609253,
      "learning_rate": 9.51842105263158e-05,
      "loss": 0.3153,
      "step": 38317
    },
    {
      "epoch": 2.1457050061597043,
      "grad_norm": 1.12213134765625,
      "learning_rate": 9.518394736842106e-05,
      "loss": 0.3185,
      "step": 38318
    },
    {
      "epoch": 2.1457610034718333,
      "grad_norm": 1.2991222143173218,
      "learning_rate": 9.518368421052633e-05,
      "loss": 0.3617,
      "step": 38319
    },
    {
      "epoch": 2.1458170007839623,
      "grad_norm": 1.6040012836456299,
      "learning_rate": 9.518342105263158e-05,
      "loss": 0.4461,
      "step": 38320
    },
    {
      "epoch": 2.1458729980960913,
      "grad_norm": 1.1934152841567993,
      "learning_rate": 9.518315789473685e-05,
      "loss": 0.3457,
      "step": 38321
    },
    {
      "epoch": 2.1459289954082204,
      "grad_norm": 1.2900394201278687,
      "learning_rate": 9.518289473684211e-05,
      "loss": 0.4278,
      "step": 38322
    },
    {
      "epoch": 2.1459849927203494,
      "grad_norm": 1.6887977123260498,
      "learning_rate": 9.518263157894737e-05,
      "loss": 0.4584,
      "step": 38323
    },
    {
      "epoch": 2.1460409900324784,
      "grad_norm": 1.2036949396133423,
      "learning_rate": 9.518236842105264e-05,
      "loss": 0.4191,
      "step": 38324
    },
    {
      "epoch": 2.1460969873446074,
      "grad_norm": 1.3448704481124878,
      "learning_rate": 9.518210526315789e-05,
      "loss": 0.3949,
      "step": 38325
    },
    {
      "epoch": 2.1461529846567364,
      "grad_norm": 1.206503987312317,
      "learning_rate": 9.518184210526316e-05,
      "loss": 0.4296,
      "step": 38326
    },
    {
      "epoch": 2.1462089819688654,
      "grad_norm": 1.0800708532333374,
      "learning_rate": 9.518157894736842e-05,
      "loss": 0.3174,
      "step": 38327
    },
    {
      "epoch": 2.1462649792809945,
      "grad_norm": 1.0248748064041138,
      "learning_rate": 9.51813157894737e-05,
      "loss": 0.2715,
      "step": 38328
    },
    {
      "epoch": 2.1463209765931235,
      "grad_norm": 1.1427592039108276,
      "learning_rate": 9.518105263157895e-05,
      "loss": 0.3067,
      "step": 38329
    },
    {
      "epoch": 2.1463769739052525,
      "grad_norm": 1.1758934259414673,
      "learning_rate": 9.518078947368421e-05,
      "loss": 0.3684,
      "step": 38330
    },
    {
      "epoch": 2.1464329712173815,
      "grad_norm": 1.4410475492477417,
      "learning_rate": 9.518052631578947e-05,
      "loss": 0.4653,
      "step": 38331
    },
    {
      "epoch": 2.1464889685295105,
      "grad_norm": 1.2631841897964478,
      "learning_rate": 9.518026315789475e-05,
      "loss": 0.4607,
      "step": 38332
    },
    {
      "epoch": 2.1465449658416396,
      "grad_norm": 1.0833930969238281,
      "learning_rate": 9.518000000000001e-05,
      "loss": 0.3683,
      "step": 38333
    },
    {
      "epoch": 2.1466009631537686,
      "grad_norm": 1.2091177701950073,
      "learning_rate": 9.517973684210527e-05,
      "loss": 0.3553,
      "step": 38334
    },
    {
      "epoch": 2.1466569604658976,
      "grad_norm": 1.1710312366485596,
      "learning_rate": 9.517947368421053e-05,
      "loss": 0.3622,
      "step": 38335
    },
    {
      "epoch": 2.1467129577780266,
      "grad_norm": 1.8413041830062866,
      "learning_rate": 9.51792105263158e-05,
      "loss": 0.4118,
      "step": 38336
    },
    {
      "epoch": 2.1467689550901556,
      "grad_norm": 1.2330278158187866,
      "learning_rate": 9.517894736842106e-05,
      "loss": 0.3719,
      "step": 38337
    },
    {
      "epoch": 2.1468249524022847,
      "grad_norm": 1.2583216428756714,
      "learning_rate": 9.517868421052632e-05,
      "loss": 0.371,
      "step": 38338
    },
    {
      "epoch": 2.1468809497144137,
      "grad_norm": 1.4155641794204712,
      "learning_rate": 9.517842105263158e-05,
      "loss": 0.3507,
      "step": 38339
    },
    {
      "epoch": 2.1469369470265427,
      "grad_norm": 1.2163387537002563,
      "learning_rate": 9.517815789473684e-05,
      "loss": 0.4214,
      "step": 38340
    },
    {
      "epoch": 2.1469929443386717,
      "grad_norm": 1.347472071647644,
      "learning_rate": 9.517789473684211e-05,
      "loss": 0.511,
      "step": 38341
    },
    {
      "epoch": 2.1470489416508007,
      "grad_norm": 1.351393222808838,
      "learning_rate": 9.517763157894737e-05,
      "loss": 0.5125,
      "step": 38342
    },
    {
      "epoch": 2.1471049389629298,
      "grad_norm": 1.3511576652526855,
      "learning_rate": 9.517736842105263e-05,
      "loss": 0.4792,
      "step": 38343
    },
    {
      "epoch": 2.1471609362750588,
      "grad_norm": 1.2691364288330078,
      "learning_rate": 9.517710526315789e-05,
      "loss": 0.3979,
      "step": 38344
    },
    {
      "epoch": 2.147216933587188,
      "grad_norm": 1.173933506011963,
      "learning_rate": 9.517684210526316e-05,
      "loss": 0.4042,
      "step": 38345
    },
    {
      "epoch": 2.147272930899317,
      "grad_norm": 1.3882774114608765,
      "learning_rate": 9.517657894736842e-05,
      "loss": 0.3967,
      "step": 38346
    },
    {
      "epoch": 2.147328928211446,
      "grad_norm": 1.3339409828186035,
      "learning_rate": 9.51763157894737e-05,
      "loss": 0.3959,
      "step": 38347
    },
    {
      "epoch": 2.147384925523575,
      "grad_norm": 1.1768335103988647,
      "learning_rate": 9.517605263157894e-05,
      "loss": 0.3841,
      "step": 38348
    },
    {
      "epoch": 2.147440922835704,
      "grad_norm": 1.2612472772598267,
      "learning_rate": 9.517578947368422e-05,
      "loss": 0.4125,
      "step": 38349
    },
    {
      "epoch": 2.147496920147833,
      "grad_norm": 1.4095532894134521,
      "learning_rate": 9.517552631578948e-05,
      "loss": 0.4211,
      "step": 38350
    },
    {
      "epoch": 2.147552917459962,
      "grad_norm": 1.1696006059646606,
      "learning_rate": 9.517526315789475e-05,
      "loss": 0.4731,
      "step": 38351
    },
    {
      "epoch": 2.147608914772091,
      "grad_norm": 1.128891110420227,
      "learning_rate": 9.517500000000001e-05,
      "loss": 0.3521,
      "step": 38352
    },
    {
      "epoch": 2.14766491208422,
      "grad_norm": 1.2698031663894653,
      "learning_rate": 9.517473684210527e-05,
      "loss": 0.4201,
      "step": 38353
    },
    {
      "epoch": 2.147720909396349,
      "grad_norm": 1.259680986404419,
      "learning_rate": 9.517447368421053e-05,
      "loss": 0.4834,
      "step": 38354
    },
    {
      "epoch": 2.147776906708478,
      "grad_norm": 1.3092985153198242,
      "learning_rate": 9.51742105263158e-05,
      "loss": 0.4251,
      "step": 38355
    },
    {
      "epoch": 2.147832904020607,
      "grad_norm": 1.0928853750228882,
      "learning_rate": 9.517394736842106e-05,
      "loss": 0.3274,
      "step": 38356
    },
    {
      "epoch": 2.147888901332736,
      "grad_norm": 1.5518240928649902,
      "learning_rate": 9.517368421052631e-05,
      "loss": 0.4858,
      "step": 38357
    },
    {
      "epoch": 2.147944898644865,
      "grad_norm": 1.2851063013076782,
      "learning_rate": 9.517342105263158e-05,
      "loss": 0.6377,
      "step": 38358
    },
    {
      "epoch": 2.148000895956994,
      "grad_norm": 1.3254694938659668,
      "learning_rate": 9.517315789473684e-05,
      "loss": 0.3525,
      "step": 38359
    },
    {
      "epoch": 2.148056893269123,
      "grad_norm": 1.1813671588897705,
      "learning_rate": 9.517289473684211e-05,
      "loss": 0.3759,
      "step": 38360
    },
    {
      "epoch": 2.148112890581252,
      "grad_norm": 1.207115888595581,
      "learning_rate": 9.517263157894737e-05,
      "loss": 0.308,
      "step": 38361
    },
    {
      "epoch": 2.148168887893381,
      "grad_norm": 1.1749515533447266,
      "learning_rate": 9.517236842105263e-05,
      "loss": 0.3349,
      "step": 38362
    },
    {
      "epoch": 2.14822488520551,
      "grad_norm": 1.2420547008514404,
      "learning_rate": 9.51721052631579e-05,
      "loss": 0.3061,
      "step": 38363
    },
    {
      "epoch": 2.148280882517639,
      "grad_norm": 1.067697286605835,
      "learning_rate": 9.517184210526317e-05,
      "loss": 0.335,
      "step": 38364
    },
    {
      "epoch": 2.148336879829768,
      "grad_norm": 1.0185019969940186,
      "learning_rate": 9.517157894736843e-05,
      "loss": 0.3493,
      "step": 38365
    },
    {
      "epoch": 2.148392877141897,
      "grad_norm": 1.1054452657699585,
      "learning_rate": 9.517131578947369e-05,
      "loss": 0.3525,
      "step": 38366
    },
    {
      "epoch": 2.148448874454026,
      "grad_norm": 1.2548161745071411,
      "learning_rate": 9.517105263157895e-05,
      "loss": 0.4089,
      "step": 38367
    },
    {
      "epoch": 2.1485048717661552,
      "grad_norm": 0.9476507306098938,
      "learning_rate": 9.517078947368422e-05,
      "loss": 0.2851,
      "step": 38368
    },
    {
      "epoch": 2.1485608690782843,
      "grad_norm": 1.0133048295974731,
      "learning_rate": 9.517052631578948e-05,
      "loss": 0.3709,
      "step": 38369
    },
    {
      "epoch": 2.1486168663904133,
      "grad_norm": 1.2617453336715698,
      "learning_rate": 9.517026315789475e-05,
      "loss": 0.4156,
      "step": 38370
    },
    {
      "epoch": 2.1486728637025423,
      "grad_norm": 1.101865291595459,
      "learning_rate": 9.517e-05,
      "loss": 0.3994,
      "step": 38371
    },
    {
      "epoch": 2.1487288610146713,
      "grad_norm": 1.0678901672363281,
      "learning_rate": 9.516973684210527e-05,
      "loss": 0.3432,
      "step": 38372
    },
    {
      "epoch": 2.1487848583268003,
      "grad_norm": 1.4221538305282593,
      "learning_rate": 9.516947368421053e-05,
      "loss": 0.4337,
      "step": 38373
    },
    {
      "epoch": 2.1488408556389293,
      "grad_norm": 1.2667040824890137,
      "learning_rate": 9.516921052631579e-05,
      "loss": 0.4032,
      "step": 38374
    },
    {
      "epoch": 2.1488968529510584,
      "grad_norm": 1.187917709350586,
      "learning_rate": 9.516894736842105e-05,
      "loss": 0.4213,
      "step": 38375
    },
    {
      "epoch": 2.1489528502631874,
      "grad_norm": 1.1725702285766602,
      "learning_rate": 9.516868421052631e-05,
      "loss": 0.5062,
      "step": 38376
    },
    {
      "epoch": 2.1490088475753164,
      "grad_norm": 1.0603768825531006,
      "learning_rate": 9.516842105263158e-05,
      "loss": 0.3374,
      "step": 38377
    },
    {
      "epoch": 2.1490648448874454,
      "grad_norm": 1.3903053998947144,
      "learning_rate": 9.516815789473684e-05,
      "loss": 0.5,
      "step": 38378
    },
    {
      "epoch": 2.1491208421995744,
      "grad_norm": 1.1216073036193848,
      "learning_rate": 9.516789473684212e-05,
      "loss": 0.3625,
      "step": 38379
    },
    {
      "epoch": 2.1491768395117035,
      "grad_norm": 1.4432493448257446,
      "learning_rate": 9.516763157894736e-05,
      "loss": 0.5352,
      "step": 38380
    },
    {
      "epoch": 2.1492328368238325,
      "grad_norm": 1.377672553062439,
      "learning_rate": 9.516736842105264e-05,
      "loss": 0.3987,
      "step": 38381
    },
    {
      "epoch": 2.1492888341359615,
      "grad_norm": 1.30591881275177,
      "learning_rate": 9.51671052631579e-05,
      "loss": 0.5589,
      "step": 38382
    },
    {
      "epoch": 2.1493448314480905,
      "grad_norm": 1.307314395904541,
      "learning_rate": 9.516684210526317e-05,
      "loss": 0.4652,
      "step": 38383
    },
    {
      "epoch": 2.1494008287602195,
      "grad_norm": 1.6501095294952393,
      "learning_rate": 9.516657894736843e-05,
      "loss": 0.4011,
      "step": 38384
    },
    {
      "epoch": 2.1494568260723486,
      "grad_norm": 1.1767069101333618,
      "learning_rate": 9.516631578947369e-05,
      "loss": 0.3429,
      "step": 38385
    },
    {
      "epoch": 2.1495128233844776,
      "grad_norm": 1.1041499376296997,
      "learning_rate": 9.516605263157895e-05,
      "loss": 0.5331,
      "step": 38386
    },
    {
      "epoch": 2.1495688206966066,
      "grad_norm": 1.075477957725525,
      "learning_rate": 9.516578947368422e-05,
      "loss": 0.3662,
      "step": 38387
    },
    {
      "epoch": 2.1496248180087356,
      "grad_norm": 1.2789161205291748,
      "learning_rate": 9.516552631578948e-05,
      "loss": 0.5727,
      "step": 38388
    },
    {
      "epoch": 2.1496808153208646,
      "grad_norm": 1.06875479221344,
      "learning_rate": 9.516526315789474e-05,
      "loss": 0.2913,
      "step": 38389
    },
    {
      "epoch": 2.1497368126329937,
      "grad_norm": 1.1192445755004883,
      "learning_rate": 9.5165e-05,
      "loss": 0.3445,
      "step": 38390
    },
    {
      "epoch": 2.1497928099451227,
      "grad_norm": 1.203905701637268,
      "learning_rate": 9.516473684210527e-05,
      "loss": 0.3686,
      "step": 38391
    },
    {
      "epoch": 2.1498488072572517,
      "grad_norm": 1.228430151939392,
      "learning_rate": 9.516447368421053e-05,
      "loss": 0.4566,
      "step": 38392
    },
    {
      "epoch": 2.1499048045693807,
      "grad_norm": 0.8909831643104553,
      "learning_rate": 9.51642105263158e-05,
      "loss": 0.2665,
      "step": 38393
    },
    {
      "epoch": 2.1499608018815097,
      "grad_norm": 1.204802393913269,
      "learning_rate": 9.516394736842105e-05,
      "loss": 0.416,
      "step": 38394
    },
    {
      "epoch": 2.1500167991936388,
      "grad_norm": 1.3512656688690186,
      "learning_rate": 9.516368421052631e-05,
      "loss": 0.41,
      "step": 38395
    },
    {
      "epoch": 2.1500727965057678,
      "grad_norm": 1.1512117385864258,
      "learning_rate": 9.516342105263159e-05,
      "loss": 0.3702,
      "step": 38396
    },
    {
      "epoch": 2.150128793817897,
      "grad_norm": 1.8756872415542603,
      "learning_rate": 9.516315789473685e-05,
      "loss": 0.4072,
      "step": 38397
    },
    {
      "epoch": 2.150184791130026,
      "grad_norm": 1.3012136220932007,
      "learning_rate": 9.51628947368421e-05,
      "loss": 0.3271,
      "step": 38398
    },
    {
      "epoch": 2.150240788442155,
      "grad_norm": 1.1595178842544556,
      "learning_rate": 9.516263157894737e-05,
      "loss": 0.3225,
      "step": 38399
    },
    {
      "epoch": 2.150296785754284,
      "grad_norm": 2.2717444896698,
      "learning_rate": 9.516236842105264e-05,
      "loss": 0.3537,
      "step": 38400
    },
    {
      "epoch": 2.150352783066413,
      "grad_norm": 1.4404935836791992,
      "learning_rate": 9.51621052631579e-05,
      "loss": 0.4387,
      "step": 38401
    },
    {
      "epoch": 2.150408780378542,
      "grad_norm": 1.1425600051879883,
      "learning_rate": 9.516184210526317e-05,
      "loss": 0.3943,
      "step": 38402
    },
    {
      "epoch": 2.150464777690671,
      "grad_norm": 1.2277195453643799,
      "learning_rate": 9.516157894736842e-05,
      "loss": 0.4301,
      "step": 38403
    },
    {
      "epoch": 2.1505207750028,
      "grad_norm": 1.088707685470581,
      "learning_rate": 9.516131578947369e-05,
      "loss": 0.377,
      "step": 38404
    },
    {
      "epoch": 2.150576772314929,
      "grad_norm": 1.5279866456985474,
      "learning_rate": 9.516105263157895e-05,
      "loss": 0.4505,
      "step": 38405
    },
    {
      "epoch": 2.150632769627058,
      "grad_norm": 1.4106463193893433,
      "learning_rate": 9.516078947368422e-05,
      "loss": 0.3173,
      "step": 38406
    },
    {
      "epoch": 2.150688766939187,
      "grad_norm": 1.3112070560455322,
      "learning_rate": 9.516052631578948e-05,
      "loss": 0.4362,
      "step": 38407
    },
    {
      "epoch": 2.150744764251316,
      "grad_norm": 1.2518410682678223,
      "learning_rate": 9.516026315789474e-05,
      "loss": 0.3829,
      "step": 38408
    },
    {
      "epoch": 2.150800761563445,
      "grad_norm": 1.3488497734069824,
      "learning_rate": 9.516e-05,
      "loss": 0.3634,
      "step": 38409
    },
    {
      "epoch": 2.150856758875574,
      "grad_norm": 1.2787147760391235,
      "learning_rate": 9.515973684210526e-05,
      "loss": 0.4027,
      "step": 38410
    },
    {
      "epoch": 2.150912756187703,
      "grad_norm": 1.1850714683532715,
      "learning_rate": 9.515947368421054e-05,
      "loss": 0.4072,
      "step": 38411
    },
    {
      "epoch": 2.150968753499832,
      "grad_norm": 1.5273966789245605,
      "learning_rate": 9.515921052631578e-05,
      "loss": 0.3271,
      "step": 38412
    },
    {
      "epoch": 2.151024750811961,
      "grad_norm": 1.3151838779449463,
      "learning_rate": 9.515894736842106e-05,
      "loss": 0.5206,
      "step": 38413
    },
    {
      "epoch": 2.15108074812409,
      "grad_norm": 1.1229515075683594,
      "learning_rate": 9.515868421052632e-05,
      "loss": 0.3385,
      "step": 38414
    },
    {
      "epoch": 2.151136745436219,
      "grad_norm": 2.3135693073272705,
      "learning_rate": 9.515842105263159e-05,
      "loss": 0.4188,
      "step": 38415
    },
    {
      "epoch": 2.151192742748348,
      "grad_norm": 1.2588485479354858,
      "learning_rate": 9.515815789473685e-05,
      "loss": 0.3691,
      "step": 38416
    },
    {
      "epoch": 2.151248740060477,
      "grad_norm": 0.9635387659072876,
      "learning_rate": 9.515789473684211e-05,
      "loss": 0.2888,
      "step": 38417
    },
    {
      "epoch": 2.151304737372606,
      "grad_norm": 1.3263167142868042,
      "learning_rate": 9.515763157894737e-05,
      "loss": 0.5061,
      "step": 38418
    },
    {
      "epoch": 2.151360734684735,
      "grad_norm": 1.085422158241272,
      "learning_rate": 9.515736842105264e-05,
      "loss": 0.4358,
      "step": 38419
    },
    {
      "epoch": 2.1514167319968642,
      "grad_norm": 1.3631871938705444,
      "learning_rate": 9.51571052631579e-05,
      "loss": 0.3652,
      "step": 38420
    },
    {
      "epoch": 2.1514727293089932,
      "grad_norm": 1.1401987075805664,
      "learning_rate": 9.515684210526316e-05,
      "loss": 0.4013,
      "step": 38421
    },
    {
      "epoch": 2.1515287266211223,
      "grad_norm": 1.1191740036010742,
      "learning_rate": 9.515657894736842e-05,
      "loss": 0.3712,
      "step": 38422
    },
    {
      "epoch": 2.1515847239332513,
      "grad_norm": 1.2889227867126465,
      "learning_rate": 9.51563157894737e-05,
      "loss": 0.4646,
      "step": 38423
    },
    {
      "epoch": 2.1516407212453803,
      "grad_norm": 1.3274955749511719,
      "learning_rate": 9.515605263157895e-05,
      "loss": 0.4719,
      "step": 38424
    },
    {
      "epoch": 2.1516967185575093,
      "grad_norm": 5.06099271774292,
      "learning_rate": 9.515578947368423e-05,
      "loss": 0.3651,
      "step": 38425
    },
    {
      "epoch": 2.1517527158696383,
      "grad_norm": 0.9237107634544373,
      "learning_rate": 9.515552631578947e-05,
      "loss": 0.3798,
      "step": 38426
    },
    {
      "epoch": 2.1518087131817674,
      "grad_norm": 1.1915366649627686,
      "learning_rate": 9.515526315789473e-05,
      "loss": 0.4381,
      "step": 38427
    },
    {
      "epoch": 2.1518647104938964,
      "grad_norm": 1.8770909309387207,
      "learning_rate": 9.5155e-05,
      "loss": 0.4434,
      "step": 38428
    },
    {
      "epoch": 2.1519207078060254,
      "grad_norm": 1.025797724723816,
      "learning_rate": 9.515473684210527e-05,
      "loss": 0.3957,
      "step": 38429
    },
    {
      "epoch": 2.1519767051181544,
      "grad_norm": 1.0619704723358154,
      "learning_rate": 9.515447368421053e-05,
      "loss": 0.3415,
      "step": 38430
    },
    {
      "epoch": 2.1520327024302834,
      "grad_norm": 1.0002723932266235,
      "learning_rate": 9.515421052631579e-05,
      "loss": 0.2805,
      "step": 38431
    },
    {
      "epoch": 2.1520886997424125,
      "grad_norm": 1.3483352661132812,
      "learning_rate": 9.515394736842106e-05,
      "loss": 0.3768,
      "step": 38432
    },
    {
      "epoch": 2.1521446970545415,
      "grad_norm": 1.174026370048523,
      "learning_rate": 9.515368421052632e-05,
      "loss": 0.3947,
      "step": 38433
    },
    {
      "epoch": 2.1522006943666705,
      "grad_norm": 1.195227026939392,
      "learning_rate": 9.515342105263159e-05,
      "loss": 0.4498,
      "step": 38434
    },
    {
      "epoch": 2.1522566916787995,
      "grad_norm": 1.5190393924713135,
      "learning_rate": 9.515315789473684e-05,
      "loss": 0.569,
      "step": 38435
    },
    {
      "epoch": 2.1523126889909285,
      "grad_norm": 1.0727328062057495,
      "learning_rate": 9.515289473684211e-05,
      "loss": 0.3135,
      "step": 38436
    },
    {
      "epoch": 2.1523686863030576,
      "grad_norm": 1.1796677112579346,
      "learning_rate": 9.515263157894737e-05,
      "loss": 0.3885,
      "step": 38437
    },
    {
      "epoch": 2.1524246836151866,
      "grad_norm": 1.220680832862854,
      "learning_rate": 9.515236842105264e-05,
      "loss": 0.4243,
      "step": 38438
    },
    {
      "epoch": 2.1524806809273156,
      "grad_norm": 1.3221995830535889,
      "learning_rate": 9.51521052631579e-05,
      "loss": 0.6364,
      "step": 38439
    },
    {
      "epoch": 2.1525366782394446,
      "grad_norm": 1.0980709791183472,
      "learning_rate": 9.515184210526316e-05,
      "loss": 0.413,
      "step": 38440
    },
    {
      "epoch": 2.1525926755515736,
      "grad_norm": 1.1699408292770386,
      "learning_rate": 9.515157894736842e-05,
      "loss": 0.3476,
      "step": 38441
    },
    {
      "epoch": 2.1526486728637027,
      "grad_norm": 1.2593698501586914,
      "learning_rate": 9.51513157894737e-05,
      "loss": 0.3814,
      "step": 38442
    },
    {
      "epoch": 2.1527046701758317,
      "grad_norm": 1.216633915901184,
      "learning_rate": 9.515105263157896e-05,
      "loss": 0.4571,
      "step": 38443
    },
    {
      "epoch": 2.1527606674879607,
      "grad_norm": 1.1833136081695557,
      "learning_rate": 9.515078947368422e-05,
      "loss": 0.3987,
      "step": 38444
    },
    {
      "epoch": 2.1528166648000897,
      "grad_norm": 1.7841136455535889,
      "learning_rate": 9.515052631578948e-05,
      "loss": 0.3522,
      "step": 38445
    },
    {
      "epoch": 2.1528726621122187,
      "grad_norm": 1.0907915830612183,
      "learning_rate": 9.515026315789474e-05,
      "loss": 0.3427,
      "step": 38446
    },
    {
      "epoch": 2.1529286594243477,
      "grad_norm": 1.0720961093902588,
      "learning_rate": 9.515000000000001e-05,
      "loss": 0.3173,
      "step": 38447
    },
    {
      "epoch": 2.1529846567364768,
      "grad_norm": 1.3579915761947632,
      "learning_rate": 9.514973684210527e-05,
      "loss": 0.3721,
      "step": 38448
    },
    {
      "epoch": 2.153040654048606,
      "grad_norm": 1.539716362953186,
      "learning_rate": 9.514947368421053e-05,
      "loss": 0.4071,
      "step": 38449
    },
    {
      "epoch": 2.153096651360735,
      "grad_norm": 0.9702361822128296,
      "learning_rate": 9.514921052631579e-05,
      "loss": 0.365,
      "step": 38450
    },
    {
      "epoch": 2.153152648672864,
      "grad_norm": 1.2700371742248535,
      "learning_rate": 9.514894736842106e-05,
      "loss": 0.5815,
      "step": 38451
    },
    {
      "epoch": 2.153208645984993,
      "grad_norm": 1.013090968132019,
      "learning_rate": 9.514868421052632e-05,
      "loss": 0.2962,
      "step": 38452
    },
    {
      "epoch": 2.153264643297122,
      "grad_norm": 1.250405192375183,
      "learning_rate": 9.514842105263158e-05,
      "loss": 0.4018,
      "step": 38453
    },
    {
      "epoch": 2.153320640609251,
      "grad_norm": 1.1727354526519775,
      "learning_rate": 9.514815789473684e-05,
      "loss": 0.4397,
      "step": 38454
    },
    {
      "epoch": 2.15337663792138,
      "grad_norm": 1.2441362142562866,
      "learning_rate": 9.514789473684211e-05,
      "loss": 0.3624,
      "step": 38455
    },
    {
      "epoch": 2.153432635233509,
      "grad_norm": 2.902956247329712,
      "learning_rate": 9.514763157894737e-05,
      "loss": 0.383,
      "step": 38456
    },
    {
      "epoch": 2.153488632545638,
      "grad_norm": 1.148582935333252,
      "learning_rate": 9.514736842105265e-05,
      "loss": 0.4701,
      "step": 38457
    },
    {
      "epoch": 2.153544629857767,
      "grad_norm": 1.849380373954773,
      "learning_rate": 9.514710526315789e-05,
      "loss": 0.256,
      "step": 38458
    },
    {
      "epoch": 2.153600627169896,
      "grad_norm": 1.0875487327575684,
      "learning_rate": 9.514684210526317e-05,
      "loss": 0.3613,
      "step": 38459
    },
    {
      "epoch": 2.153656624482025,
      "grad_norm": 1.0818275213241577,
      "learning_rate": 9.514657894736843e-05,
      "loss": 0.4268,
      "step": 38460
    },
    {
      "epoch": 2.153712621794154,
      "grad_norm": 1.1120792627334595,
      "learning_rate": 9.514631578947369e-05,
      "loss": 0.3878,
      "step": 38461
    },
    {
      "epoch": 2.153768619106283,
      "grad_norm": 1.3657582998275757,
      "learning_rate": 9.514605263157896e-05,
      "loss": 0.4561,
      "step": 38462
    },
    {
      "epoch": 2.153824616418412,
      "grad_norm": 1.220436453819275,
      "learning_rate": 9.51457894736842e-05,
      "loss": 0.364,
      "step": 38463
    },
    {
      "epoch": 2.153880613730541,
      "grad_norm": 1.2224228382110596,
      "learning_rate": 9.514552631578948e-05,
      "loss": 0.4847,
      "step": 38464
    },
    {
      "epoch": 2.15393661104267,
      "grad_norm": 1.2699791193008423,
      "learning_rate": 9.514526315789474e-05,
      "loss": 0.3903,
      "step": 38465
    },
    {
      "epoch": 2.153992608354799,
      "grad_norm": 1.104504942893982,
      "learning_rate": 9.514500000000001e-05,
      "loss": 0.3714,
      "step": 38466
    },
    {
      "epoch": 2.154048605666928,
      "grad_norm": 1.4629234075546265,
      "learning_rate": 9.514473684210526e-05,
      "loss": 0.4596,
      "step": 38467
    },
    {
      "epoch": 2.154104602979057,
      "grad_norm": 1.3545204401016235,
      "learning_rate": 9.514447368421053e-05,
      "loss": 0.3256,
      "step": 38468
    },
    {
      "epoch": 2.154160600291186,
      "grad_norm": 1.1996327638626099,
      "learning_rate": 9.514421052631579e-05,
      "loss": 0.504,
      "step": 38469
    },
    {
      "epoch": 2.154216597603315,
      "grad_norm": 1.231961727142334,
      "learning_rate": 9.514394736842106e-05,
      "loss": 0.3281,
      "step": 38470
    },
    {
      "epoch": 2.154272594915444,
      "grad_norm": 1.2423095703125,
      "learning_rate": 9.514368421052632e-05,
      "loss": 0.315,
      "step": 38471
    },
    {
      "epoch": 2.1543285922275732,
      "grad_norm": 1.1325401067733765,
      "learning_rate": 9.514342105263158e-05,
      "loss": 0.3833,
      "step": 38472
    },
    {
      "epoch": 2.1543845895397022,
      "grad_norm": 1.2642079591751099,
      "learning_rate": 9.514315789473684e-05,
      "loss": 0.3332,
      "step": 38473
    },
    {
      "epoch": 2.1544405868518313,
      "grad_norm": 1.121393084526062,
      "learning_rate": 9.514289473684212e-05,
      "loss": 0.3456,
      "step": 38474
    },
    {
      "epoch": 2.1544965841639603,
      "grad_norm": 1.3007762432098389,
      "learning_rate": 9.514263157894738e-05,
      "loss": 0.5197,
      "step": 38475
    },
    {
      "epoch": 2.1545525814760893,
      "grad_norm": 1.2785131931304932,
      "learning_rate": 9.514236842105264e-05,
      "loss": 0.7217,
      "step": 38476
    },
    {
      "epoch": 2.1546085787882183,
      "grad_norm": 1.2964714765548706,
      "learning_rate": 9.51421052631579e-05,
      "loss": 0.3526,
      "step": 38477
    },
    {
      "epoch": 2.1546645761003473,
      "grad_norm": 1.0315666198730469,
      "learning_rate": 9.514184210526316e-05,
      "loss": 0.4063,
      "step": 38478
    },
    {
      "epoch": 2.1547205734124764,
      "grad_norm": 1.1242539882659912,
      "learning_rate": 9.514157894736843e-05,
      "loss": 0.3544,
      "step": 38479
    },
    {
      "epoch": 2.1547765707246054,
      "grad_norm": 1.4843276739120483,
      "learning_rate": 9.514131578947369e-05,
      "loss": 0.3062,
      "step": 38480
    },
    {
      "epoch": 2.1548325680367344,
      "grad_norm": 1.1322059631347656,
      "learning_rate": 9.514105263157895e-05,
      "loss": 0.3404,
      "step": 38481
    },
    {
      "epoch": 2.1548885653488634,
      "grad_norm": 1.428338885307312,
      "learning_rate": 9.514078947368421e-05,
      "loss": 0.3516,
      "step": 38482
    },
    {
      "epoch": 2.1549445626609924,
      "grad_norm": 1.294930100440979,
      "learning_rate": 9.514052631578948e-05,
      "loss": 0.4035,
      "step": 38483
    },
    {
      "epoch": 2.1550005599731215,
      "grad_norm": 1.3391445875167847,
      "learning_rate": 9.514026315789474e-05,
      "loss": 0.3593,
      "step": 38484
    },
    {
      "epoch": 2.1550565572852505,
      "grad_norm": 1.3055952787399292,
      "learning_rate": 9.514e-05,
      "loss": 0.4217,
      "step": 38485
    },
    {
      "epoch": 2.1551125545973795,
      "grad_norm": 1.3343820571899414,
      "learning_rate": 9.513973684210526e-05,
      "loss": 0.4283,
      "step": 38486
    },
    {
      "epoch": 2.1551685519095085,
      "grad_norm": 1.2045443058013916,
      "learning_rate": 9.513947368421053e-05,
      "loss": 0.3519,
      "step": 38487
    },
    {
      "epoch": 2.1552245492216375,
      "grad_norm": 1.249302864074707,
      "learning_rate": 9.51392105263158e-05,
      "loss": 0.3885,
      "step": 38488
    },
    {
      "epoch": 2.1552805465337666,
      "grad_norm": 1.2215322256088257,
      "learning_rate": 9.513894736842107e-05,
      "loss": 0.3797,
      "step": 38489
    },
    {
      "epoch": 2.1553365438458956,
      "grad_norm": 1.0655783414840698,
      "learning_rate": 9.513868421052631e-05,
      "loss": 0.298,
      "step": 38490
    },
    {
      "epoch": 2.1553925411580246,
      "grad_norm": 1.4894440174102783,
      "learning_rate": 9.513842105263159e-05,
      "loss": 0.3956,
      "step": 38491
    },
    {
      "epoch": 2.1554485384701536,
      "grad_norm": 1.0665535926818848,
      "learning_rate": 9.513815789473685e-05,
      "loss": 0.4009,
      "step": 38492
    },
    {
      "epoch": 2.1555045357822826,
      "grad_norm": 1.4605998992919922,
      "learning_rate": 9.513789473684212e-05,
      "loss": 0.4195,
      "step": 38493
    },
    {
      "epoch": 2.1555605330944116,
      "grad_norm": 1.0981621742248535,
      "learning_rate": 9.513763157894738e-05,
      "loss": 0.3732,
      "step": 38494
    },
    {
      "epoch": 2.1556165304065407,
      "grad_norm": 1.1166752576828003,
      "learning_rate": 9.513736842105262e-05,
      "loss": 0.4115,
      "step": 38495
    },
    {
      "epoch": 2.1556725277186697,
      "grad_norm": 1.3041073083877563,
      "learning_rate": 9.51371052631579e-05,
      "loss": 0.4104,
      "step": 38496
    },
    {
      "epoch": 2.1557285250307987,
      "grad_norm": 1.2176666259765625,
      "learning_rate": 9.513684210526316e-05,
      "loss": 0.3595,
      "step": 38497
    },
    {
      "epoch": 2.1557845223429277,
      "grad_norm": 1.1245518922805786,
      "learning_rate": 9.513657894736843e-05,
      "loss": 0.4238,
      "step": 38498
    },
    {
      "epoch": 2.1558405196550567,
      "grad_norm": 1.2398622035980225,
      "learning_rate": 9.513631578947369e-05,
      "loss": 0.4483,
      "step": 38499
    },
    {
      "epoch": 2.1558965169671858,
      "grad_norm": 1.4633852243423462,
      "learning_rate": 9.513605263157895e-05,
      "loss": 0.3133,
      "step": 38500
    },
    {
      "epoch": 2.155952514279315,
      "grad_norm": 1.0684674978256226,
      "learning_rate": 9.513578947368421e-05,
      "loss": 0.3676,
      "step": 38501
    },
    {
      "epoch": 2.156008511591444,
      "grad_norm": 1.0375701189041138,
      "learning_rate": 9.513552631578948e-05,
      "loss": 0.3438,
      "step": 38502
    },
    {
      "epoch": 2.156064508903573,
      "grad_norm": 1.0976521968841553,
      "learning_rate": 9.513526315789474e-05,
      "loss": 0.341,
      "step": 38503
    },
    {
      "epoch": 2.156120506215702,
      "grad_norm": 1.1272724866867065,
      "learning_rate": 9.5135e-05,
      "loss": 0.297,
      "step": 38504
    },
    {
      "epoch": 2.156176503527831,
      "grad_norm": 1.0749926567077637,
      "learning_rate": 9.513473684210526e-05,
      "loss": 0.3416,
      "step": 38505
    },
    {
      "epoch": 2.15623250083996,
      "grad_norm": 1.091835618019104,
      "learning_rate": 9.513447368421054e-05,
      "loss": 0.3801,
      "step": 38506
    },
    {
      "epoch": 2.156288498152089,
      "grad_norm": 1.025221347808838,
      "learning_rate": 9.51342105263158e-05,
      "loss": 0.3643,
      "step": 38507
    },
    {
      "epoch": 2.156344495464218,
      "grad_norm": 1.303847074508667,
      "learning_rate": 9.513394736842106e-05,
      "loss": 0.4585,
      "step": 38508
    },
    {
      "epoch": 2.156400492776347,
      "grad_norm": 1.105273962020874,
      "learning_rate": 9.513368421052632e-05,
      "loss": 0.3048,
      "step": 38509
    },
    {
      "epoch": 2.156456490088476,
      "grad_norm": 1.1365665197372437,
      "learning_rate": 9.513342105263159e-05,
      "loss": 0.421,
      "step": 38510
    },
    {
      "epoch": 2.156512487400605,
      "grad_norm": 1.1094828844070435,
      "learning_rate": 9.513315789473685e-05,
      "loss": 0.3764,
      "step": 38511
    },
    {
      "epoch": 2.156568484712734,
      "grad_norm": 1.1896651983261108,
      "learning_rate": 9.513289473684212e-05,
      "loss": 0.4012,
      "step": 38512
    },
    {
      "epoch": 2.156624482024863,
      "grad_norm": 1.4077340364456177,
      "learning_rate": 9.513263157894737e-05,
      "loss": 0.3851,
      "step": 38513
    },
    {
      "epoch": 2.156680479336992,
      "grad_norm": 1.1899935007095337,
      "learning_rate": 9.513236842105263e-05,
      "loss": 0.3729,
      "step": 38514
    },
    {
      "epoch": 2.156736476649121,
      "grad_norm": 1.1590992212295532,
      "learning_rate": 9.51321052631579e-05,
      "loss": 0.4018,
      "step": 38515
    },
    {
      "epoch": 2.15679247396125,
      "grad_norm": 1.2674232721328735,
      "learning_rate": 9.513184210526316e-05,
      "loss": 0.5133,
      "step": 38516
    },
    {
      "epoch": 2.1568484712733786,
      "grad_norm": 1.2413398027420044,
      "learning_rate": 9.513157894736843e-05,
      "loss": 0.508,
      "step": 38517
    },
    {
      "epoch": 2.156904468585508,
      "grad_norm": 1.1949427127838135,
      "learning_rate": 9.513131578947368e-05,
      "loss": 0.4393,
      "step": 38518
    },
    {
      "epoch": 2.1569604658976367,
      "grad_norm": 1.2321666479110718,
      "learning_rate": 9.513105263157895e-05,
      "loss": 0.3572,
      "step": 38519
    },
    {
      "epoch": 2.157016463209766,
      "grad_norm": 1.2356003522872925,
      "learning_rate": 9.513078947368421e-05,
      "loss": 0.4017,
      "step": 38520
    },
    {
      "epoch": 2.1570724605218947,
      "grad_norm": 1.1150670051574707,
      "learning_rate": 9.513052631578949e-05,
      "loss": 0.3228,
      "step": 38521
    },
    {
      "epoch": 2.157128457834024,
      "grad_norm": 1.595812439918518,
      "learning_rate": 9.513026315789473e-05,
      "loss": 0.4999,
      "step": 38522
    },
    {
      "epoch": 2.1571844551461528,
      "grad_norm": 1.9882651567459106,
      "learning_rate": 9.513e-05,
      "loss": 0.3013,
      "step": 38523
    },
    {
      "epoch": 2.157240452458282,
      "grad_norm": 1.2250685691833496,
      "learning_rate": 9.512973684210527e-05,
      "loss": 0.362,
      "step": 38524
    },
    {
      "epoch": 2.157296449770411,
      "grad_norm": 1.7353954315185547,
      "learning_rate": 9.512947368421054e-05,
      "loss": 0.5075,
      "step": 38525
    },
    {
      "epoch": 2.1573524470825403,
      "grad_norm": 1.0507529973983765,
      "learning_rate": 9.51292105263158e-05,
      "loss": 0.3354,
      "step": 38526
    },
    {
      "epoch": 2.157408444394669,
      "grad_norm": 1.3162833452224731,
      "learning_rate": 9.512894736842106e-05,
      "loss": 0.4163,
      "step": 38527
    },
    {
      "epoch": 2.1574644417067983,
      "grad_norm": 1.0403687953948975,
      "learning_rate": 9.512868421052632e-05,
      "loss": 0.4029,
      "step": 38528
    },
    {
      "epoch": 2.157520439018927,
      "grad_norm": 1.1419215202331543,
      "learning_rate": 9.512842105263159e-05,
      "loss": 0.3681,
      "step": 38529
    },
    {
      "epoch": 2.1575764363310563,
      "grad_norm": 1.597672462463379,
      "learning_rate": 9.512815789473685e-05,
      "loss": 0.3261,
      "step": 38530
    },
    {
      "epoch": 2.157632433643185,
      "grad_norm": 2.1670682430267334,
      "learning_rate": 9.512789473684211e-05,
      "loss": 0.3101,
      "step": 38531
    },
    {
      "epoch": 2.1576884309553144,
      "grad_norm": 1.0850896835327148,
      "learning_rate": 9.512763157894737e-05,
      "loss": 0.4121,
      "step": 38532
    },
    {
      "epoch": 2.157744428267443,
      "grad_norm": 1.089418649673462,
      "learning_rate": 9.512736842105263e-05,
      "loss": 0.3356,
      "step": 38533
    },
    {
      "epoch": 2.1578004255795724,
      "grad_norm": 1.0577048063278198,
      "learning_rate": 9.51271052631579e-05,
      "loss": 0.3781,
      "step": 38534
    },
    {
      "epoch": 2.157856422891701,
      "grad_norm": 1.6464054584503174,
      "learning_rate": 9.512684210526316e-05,
      "loss": 0.4049,
      "step": 38535
    },
    {
      "epoch": 2.1579124202038305,
      "grad_norm": 1.7340925931930542,
      "learning_rate": 9.512657894736842e-05,
      "loss": 0.5827,
      "step": 38536
    },
    {
      "epoch": 2.157968417515959,
      "grad_norm": 1.3729428052902222,
      "learning_rate": 9.512631578947368e-05,
      "loss": 0.4285,
      "step": 38537
    },
    {
      "epoch": 2.158024414828088,
      "grad_norm": 1.1906861066818237,
      "learning_rate": 9.512605263157896e-05,
      "loss": 0.2826,
      "step": 38538
    },
    {
      "epoch": 2.158080412140217,
      "grad_norm": 0.9225759506225586,
      "learning_rate": 9.512578947368422e-05,
      "loss": 0.2453,
      "step": 38539
    },
    {
      "epoch": 2.158136409452346,
      "grad_norm": 1.1134223937988281,
      "learning_rate": 9.512552631578948e-05,
      "loss": 0.3685,
      "step": 38540
    },
    {
      "epoch": 2.158192406764475,
      "grad_norm": 1.5738328695297241,
      "learning_rate": 9.512526315789473e-05,
      "loss": 0.5515,
      "step": 38541
    },
    {
      "epoch": 2.158248404076604,
      "grad_norm": 1.5049524307250977,
      "learning_rate": 9.512500000000001e-05,
      "loss": 0.3536,
      "step": 38542
    },
    {
      "epoch": 2.158304401388733,
      "grad_norm": 1.2111343145370483,
      "learning_rate": 9.512473684210527e-05,
      "loss": 0.5285,
      "step": 38543
    },
    {
      "epoch": 2.158360398700862,
      "grad_norm": 1.3672007322311401,
      "learning_rate": 9.512447368421054e-05,
      "loss": 0.5118,
      "step": 38544
    },
    {
      "epoch": 2.158416396012991,
      "grad_norm": 1.1374894380569458,
      "learning_rate": 9.512421052631579e-05,
      "loss": 0.3471,
      "step": 38545
    },
    {
      "epoch": 2.15847239332512,
      "grad_norm": 1.3705662488937378,
      "learning_rate": 9.512394736842106e-05,
      "loss": 0.4479,
      "step": 38546
    },
    {
      "epoch": 2.158528390637249,
      "grad_norm": 1.0276386737823486,
      "learning_rate": 9.512368421052632e-05,
      "loss": 0.4133,
      "step": 38547
    },
    {
      "epoch": 2.1585843879493782,
      "grad_norm": 1.2824528217315674,
      "learning_rate": 9.512342105263158e-05,
      "loss": 0.4397,
      "step": 38548
    },
    {
      "epoch": 2.1586403852615073,
      "grad_norm": 1.2494326829910278,
      "learning_rate": 9.512315789473685e-05,
      "loss": 0.3303,
      "step": 38549
    },
    {
      "epoch": 2.1586963825736363,
      "grad_norm": 1.5131142139434814,
      "learning_rate": 9.51228947368421e-05,
      "loss": 0.702,
      "step": 38550
    },
    {
      "epoch": 2.1587523798857653,
      "grad_norm": 1.3123862743377686,
      "learning_rate": 9.512263157894737e-05,
      "loss": 0.3819,
      "step": 38551
    },
    {
      "epoch": 2.1588083771978943,
      "grad_norm": 1.3677548170089722,
      "learning_rate": 9.512236842105263e-05,
      "loss": 0.449,
      "step": 38552
    },
    {
      "epoch": 2.1588643745100233,
      "grad_norm": 1.2566975355148315,
      "learning_rate": 9.51221052631579e-05,
      "loss": 0.434,
      "step": 38553
    },
    {
      "epoch": 2.1589203718221524,
      "grad_norm": 1.159308910369873,
      "learning_rate": 9.512184210526317e-05,
      "loss": 0.5404,
      "step": 38554
    },
    {
      "epoch": 2.1589763691342814,
      "grad_norm": 1.2370284795761108,
      "learning_rate": 9.512157894736843e-05,
      "loss": 0.5489,
      "step": 38555
    },
    {
      "epoch": 2.1590323664464104,
      "grad_norm": 1.1077632904052734,
      "learning_rate": 9.512131578947369e-05,
      "loss": 0.3503,
      "step": 38556
    },
    {
      "epoch": 2.1590883637585394,
      "grad_norm": 1.2936064004898071,
      "learning_rate": 9.512105263157896e-05,
      "loss": 0.4271,
      "step": 38557
    },
    {
      "epoch": 2.1591443610706684,
      "grad_norm": 0.9969408512115479,
      "learning_rate": 9.512078947368422e-05,
      "loss": 0.4391,
      "step": 38558
    },
    {
      "epoch": 2.1592003583827974,
      "grad_norm": 1.1851403713226318,
      "learning_rate": 9.512052631578948e-05,
      "loss": 0.3519,
      "step": 38559
    },
    {
      "epoch": 2.1592563556949265,
      "grad_norm": 1.053171157836914,
      "learning_rate": 9.512026315789474e-05,
      "loss": 0.342,
      "step": 38560
    },
    {
      "epoch": 2.1593123530070555,
      "grad_norm": 1.3716280460357666,
      "learning_rate": 9.512000000000001e-05,
      "loss": 0.2969,
      "step": 38561
    },
    {
      "epoch": 2.1593683503191845,
      "grad_norm": 1.2639951705932617,
      "learning_rate": 9.511973684210527e-05,
      "loss": 0.3877,
      "step": 38562
    },
    {
      "epoch": 2.1594243476313135,
      "grad_norm": 1.4132440090179443,
      "learning_rate": 9.511947368421053e-05,
      "loss": 0.3946,
      "step": 38563
    },
    {
      "epoch": 2.1594803449434425,
      "grad_norm": 1.3084564208984375,
      "learning_rate": 9.511921052631579e-05,
      "loss": 0.4713,
      "step": 38564
    },
    {
      "epoch": 2.1595363422555716,
      "grad_norm": 1.294042706489563,
      "learning_rate": 9.511894736842105e-05,
      "loss": 0.3917,
      "step": 38565
    },
    {
      "epoch": 2.1595923395677006,
      "grad_norm": 1.09099280834198,
      "learning_rate": 9.511868421052632e-05,
      "loss": 0.2852,
      "step": 38566
    },
    {
      "epoch": 2.1596483368798296,
      "grad_norm": 1.2776057720184326,
      "learning_rate": 9.511842105263158e-05,
      "loss": 0.3396,
      "step": 38567
    },
    {
      "epoch": 2.1597043341919586,
      "grad_norm": 1.1086061000823975,
      "learning_rate": 9.511815789473684e-05,
      "loss": 0.3762,
      "step": 38568
    },
    {
      "epoch": 2.1597603315040876,
      "grad_norm": 1.0837737321853638,
      "learning_rate": 9.51178947368421e-05,
      "loss": 0.2789,
      "step": 38569
    },
    {
      "epoch": 2.1598163288162167,
      "grad_norm": 1.5426263809204102,
      "learning_rate": 9.511763157894738e-05,
      "loss": 0.5148,
      "step": 38570
    },
    {
      "epoch": 2.1598723261283457,
      "grad_norm": 0.9703961610794067,
      "learning_rate": 9.511736842105264e-05,
      "loss": 0.4398,
      "step": 38571
    },
    {
      "epoch": 2.1599283234404747,
      "grad_norm": 1.5045948028564453,
      "learning_rate": 9.511710526315791e-05,
      "loss": 0.7126,
      "step": 38572
    },
    {
      "epoch": 2.1599843207526037,
      "grad_norm": 1.4359759092330933,
      "learning_rate": 9.511684210526315e-05,
      "loss": 0.3646,
      "step": 38573
    },
    {
      "epoch": 2.1600403180647327,
      "grad_norm": 1.2386369705200195,
      "learning_rate": 9.511657894736843e-05,
      "loss": 0.4214,
      "step": 38574
    },
    {
      "epoch": 2.1600963153768618,
      "grad_norm": 1.1075921058654785,
      "learning_rate": 9.511631578947369e-05,
      "loss": 0.3562,
      "step": 38575
    },
    {
      "epoch": 2.1601523126889908,
      "grad_norm": 1.3298624753952026,
      "learning_rate": 9.511605263157896e-05,
      "loss": 0.447,
      "step": 38576
    },
    {
      "epoch": 2.16020831000112,
      "grad_norm": 1.3276902437210083,
      "learning_rate": 9.511578947368421e-05,
      "loss": 0.3599,
      "step": 38577
    },
    {
      "epoch": 2.160264307313249,
      "grad_norm": 1.250604271888733,
      "learning_rate": 9.511552631578948e-05,
      "loss": 0.4126,
      "step": 38578
    },
    {
      "epoch": 2.160320304625378,
      "grad_norm": 1.123768925666809,
      "learning_rate": 9.511526315789474e-05,
      "loss": 0.3555,
      "step": 38579
    },
    {
      "epoch": 2.160376301937507,
      "grad_norm": 1.360512137413025,
      "learning_rate": 9.511500000000001e-05,
      "loss": 0.4484,
      "step": 38580
    },
    {
      "epoch": 2.160432299249636,
      "grad_norm": 1.6691123247146606,
      "learning_rate": 9.511473684210527e-05,
      "loss": 0.3395,
      "step": 38581
    },
    {
      "epoch": 2.160488296561765,
      "grad_norm": 1.8766708374023438,
      "learning_rate": 9.511447368421052e-05,
      "loss": 0.5455,
      "step": 38582
    },
    {
      "epoch": 2.160544293873894,
      "grad_norm": 1.3959908485412598,
      "learning_rate": 9.511421052631579e-05,
      "loss": 0.5452,
      "step": 38583
    },
    {
      "epoch": 2.160600291186023,
      "grad_norm": 1.4481756687164307,
      "learning_rate": 9.511394736842105e-05,
      "loss": 0.4623,
      "step": 38584
    },
    {
      "epoch": 2.160656288498152,
      "grad_norm": 1.4850200414657593,
      "learning_rate": 9.511368421052633e-05,
      "loss": 0.4645,
      "step": 38585
    },
    {
      "epoch": 2.160712285810281,
      "grad_norm": 1.1922173500061035,
      "learning_rate": 9.511342105263159e-05,
      "loss": 0.3469,
      "step": 38586
    },
    {
      "epoch": 2.16076828312241,
      "grad_norm": 1.4706612825393677,
      "learning_rate": 9.511315789473685e-05,
      "loss": 0.4546,
      "step": 38587
    },
    {
      "epoch": 2.160824280434539,
      "grad_norm": 1.2319934368133545,
      "learning_rate": 9.51128947368421e-05,
      "loss": 0.4516,
      "step": 38588
    },
    {
      "epoch": 2.160880277746668,
      "grad_norm": 1.1282672882080078,
      "learning_rate": 9.511263157894738e-05,
      "loss": 0.345,
      "step": 38589
    },
    {
      "epoch": 2.160936275058797,
      "grad_norm": 3.1275134086608887,
      "learning_rate": 9.511236842105264e-05,
      "loss": 0.5066,
      "step": 38590
    },
    {
      "epoch": 2.160992272370926,
      "grad_norm": 1.3236494064331055,
      "learning_rate": 9.51121052631579e-05,
      "loss": 0.4097,
      "step": 38591
    },
    {
      "epoch": 2.161048269683055,
      "grad_norm": 1.289269208908081,
      "learning_rate": 9.511184210526316e-05,
      "loss": 0.4153,
      "step": 38592
    },
    {
      "epoch": 2.161104266995184,
      "grad_norm": 1.4659175872802734,
      "learning_rate": 9.511157894736843e-05,
      "loss": 0.5203,
      "step": 38593
    },
    {
      "epoch": 2.161160264307313,
      "grad_norm": 1.179755449295044,
      "learning_rate": 9.511131578947369e-05,
      "loss": 0.4724,
      "step": 38594
    },
    {
      "epoch": 2.161216261619442,
      "grad_norm": 1.3480043411254883,
      "learning_rate": 9.511105263157895e-05,
      "loss": 0.393,
      "step": 38595
    },
    {
      "epoch": 2.161272258931571,
      "grad_norm": 1.3042832612991333,
      "learning_rate": 9.511078947368421e-05,
      "loss": 0.5268,
      "step": 38596
    },
    {
      "epoch": 2.1613282562437,
      "grad_norm": 1.2216871976852417,
      "learning_rate": 9.511052631578948e-05,
      "loss": 0.4966,
      "step": 38597
    },
    {
      "epoch": 2.161384253555829,
      "grad_norm": 1.2597949504852295,
      "learning_rate": 9.511026315789474e-05,
      "loss": 0.39,
      "step": 38598
    },
    {
      "epoch": 2.161440250867958,
      "grad_norm": 1.4922704696655273,
      "learning_rate": 9.511e-05,
      "loss": 0.4078,
      "step": 38599
    },
    {
      "epoch": 2.1614962481800872,
      "grad_norm": 1.1868919134140015,
      "learning_rate": 9.510973684210526e-05,
      "loss": 0.4372,
      "step": 38600
    },
    {
      "epoch": 2.1615522454922163,
      "grad_norm": 1.1306382417678833,
      "learning_rate": 9.510947368421052e-05,
      "loss": 0.3356,
      "step": 38601
    },
    {
      "epoch": 2.1616082428043453,
      "grad_norm": 1.0911821126937866,
      "learning_rate": 9.51092105263158e-05,
      "loss": 0.3762,
      "step": 38602
    },
    {
      "epoch": 2.1616642401164743,
      "grad_norm": 1.5354933738708496,
      "learning_rate": 9.510894736842105e-05,
      "loss": 0.3249,
      "step": 38603
    },
    {
      "epoch": 2.1617202374286033,
      "grad_norm": 1.2692798376083374,
      "learning_rate": 9.510868421052633e-05,
      "loss": 0.3939,
      "step": 38604
    },
    {
      "epoch": 2.1617762347407323,
      "grad_norm": 1.6741389036178589,
      "learning_rate": 9.510842105263157e-05,
      "loss": 0.3377,
      "step": 38605
    },
    {
      "epoch": 2.1618322320528613,
      "grad_norm": 1.1014587879180908,
      "learning_rate": 9.510815789473685e-05,
      "loss": 0.4165,
      "step": 38606
    },
    {
      "epoch": 2.1618882293649904,
      "grad_norm": 1.2921602725982666,
      "learning_rate": 9.510789473684211e-05,
      "loss": 0.3577,
      "step": 38607
    },
    {
      "epoch": 2.1619442266771194,
      "grad_norm": 2.498472213745117,
      "learning_rate": 9.510763157894738e-05,
      "loss": 0.3721,
      "step": 38608
    },
    {
      "epoch": 2.1620002239892484,
      "grad_norm": 1.0836174488067627,
      "learning_rate": 9.510736842105264e-05,
      "loss": 0.3331,
      "step": 38609
    },
    {
      "epoch": 2.1620562213013774,
      "grad_norm": 1.6303850412368774,
      "learning_rate": 9.51071052631579e-05,
      "loss": 0.4466,
      "step": 38610
    },
    {
      "epoch": 2.1621122186135064,
      "grad_norm": 1.271372675895691,
      "learning_rate": 9.510684210526316e-05,
      "loss": 0.4084,
      "step": 38611
    },
    {
      "epoch": 2.1621682159256355,
      "grad_norm": 1.1850817203521729,
      "learning_rate": 9.510657894736843e-05,
      "loss": 0.2733,
      "step": 38612
    },
    {
      "epoch": 2.1622242132377645,
      "grad_norm": 1.1912884712219238,
      "learning_rate": 9.510631578947369e-05,
      "loss": 0.4199,
      "step": 38613
    },
    {
      "epoch": 2.1622802105498935,
      "grad_norm": 1.3928399085998535,
      "learning_rate": 9.510605263157895e-05,
      "loss": 0.352,
      "step": 38614
    },
    {
      "epoch": 2.1623362078620225,
      "grad_norm": 4.831754684448242,
      "learning_rate": 9.510578947368421e-05,
      "loss": 0.3855,
      "step": 38615
    },
    {
      "epoch": 2.1623922051741515,
      "grad_norm": 1.2795354127883911,
      "learning_rate": 9.510552631578947e-05,
      "loss": 0.3873,
      "step": 38616
    },
    {
      "epoch": 2.1624482024862806,
      "grad_norm": 1.6733497381210327,
      "learning_rate": 9.510526315789475e-05,
      "loss": 0.4981,
      "step": 38617
    },
    {
      "epoch": 2.1625041997984096,
      "grad_norm": 1.0682317018508911,
      "learning_rate": 9.5105e-05,
      "loss": 0.3443,
      "step": 38618
    },
    {
      "epoch": 2.1625601971105386,
      "grad_norm": 6.7518310546875,
      "learning_rate": 9.510473684210526e-05,
      "loss": 0.358,
      "step": 38619
    },
    {
      "epoch": 2.1626161944226676,
      "grad_norm": 1.6293147802352905,
      "learning_rate": 9.510447368421052e-05,
      "loss": 0.4795,
      "step": 38620
    },
    {
      "epoch": 2.1626721917347966,
      "grad_norm": 1.2021028995513916,
      "learning_rate": 9.51042105263158e-05,
      "loss": 0.3671,
      "step": 38621
    },
    {
      "epoch": 2.1627281890469257,
      "grad_norm": 1.1868031024932861,
      "learning_rate": 9.510394736842106e-05,
      "loss": 0.4678,
      "step": 38622
    },
    {
      "epoch": 2.1627841863590547,
      "grad_norm": 1.2829930782318115,
      "learning_rate": 9.510368421052632e-05,
      "loss": 0.5021,
      "step": 38623
    },
    {
      "epoch": 2.1628401836711837,
      "grad_norm": 2.4438936710357666,
      "learning_rate": 9.510342105263158e-05,
      "loss": 0.426,
      "step": 38624
    },
    {
      "epoch": 2.1628961809833127,
      "grad_norm": 1.3038809299468994,
      "learning_rate": 9.510315789473685e-05,
      "loss": 0.3525,
      "step": 38625
    },
    {
      "epoch": 2.1629521782954417,
      "grad_norm": 1.0626068115234375,
      "learning_rate": 9.510289473684211e-05,
      "loss": 0.267,
      "step": 38626
    },
    {
      "epoch": 2.1630081756075707,
      "grad_norm": 0.9845471978187561,
      "learning_rate": 9.510263157894738e-05,
      "loss": 0.313,
      "step": 38627
    },
    {
      "epoch": 2.1630641729196998,
      "grad_norm": 1.1851896047592163,
      "learning_rate": 9.510236842105263e-05,
      "loss": 0.4291,
      "step": 38628
    },
    {
      "epoch": 2.163120170231829,
      "grad_norm": 1.3013083934783936,
      "learning_rate": 9.51021052631579e-05,
      "loss": 0.378,
      "step": 38629
    },
    {
      "epoch": 2.163176167543958,
      "grad_norm": 1.3607717752456665,
      "learning_rate": 9.510184210526316e-05,
      "loss": 0.3257,
      "step": 38630
    },
    {
      "epoch": 2.163232164856087,
      "grad_norm": 1.3301739692687988,
      "learning_rate": 9.510157894736844e-05,
      "loss": 0.4685,
      "step": 38631
    },
    {
      "epoch": 2.163288162168216,
      "grad_norm": 1.148032307624817,
      "learning_rate": 9.510131578947368e-05,
      "loss": 0.3056,
      "step": 38632
    },
    {
      "epoch": 2.163344159480345,
      "grad_norm": 1.2843828201293945,
      "learning_rate": 9.510105263157896e-05,
      "loss": 0.3811,
      "step": 38633
    },
    {
      "epoch": 2.163400156792474,
      "grad_norm": 1.2604568004608154,
      "learning_rate": 9.510078947368421e-05,
      "loss": 0.3152,
      "step": 38634
    },
    {
      "epoch": 2.163456154104603,
      "grad_norm": 1.4230263233184814,
      "learning_rate": 9.510052631578947e-05,
      "loss": 0.4336,
      "step": 38635
    },
    {
      "epoch": 2.163512151416732,
      "grad_norm": 1.424559473991394,
      "learning_rate": 9.510026315789475e-05,
      "loss": 0.401,
      "step": 38636
    },
    {
      "epoch": 2.163568148728861,
      "grad_norm": 1.4415538311004639,
      "learning_rate": 9.51e-05,
      "loss": 0.5129,
      "step": 38637
    },
    {
      "epoch": 2.16362414604099,
      "grad_norm": 1.2011644840240479,
      "learning_rate": 9.509973684210527e-05,
      "loss": 0.493,
      "step": 38638
    },
    {
      "epoch": 2.163680143353119,
      "grad_norm": 1.1000211238861084,
      "learning_rate": 9.509947368421053e-05,
      "loss": 0.4089,
      "step": 38639
    },
    {
      "epoch": 2.163736140665248,
      "grad_norm": 1.1017917394638062,
      "learning_rate": 9.50992105263158e-05,
      "loss": 0.3115,
      "step": 38640
    },
    {
      "epoch": 2.163792137977377,
      "grad_norm": 2.239315986633301,
      "learning_rate": 9.509894736842106e-05,
      "loss": 0.3906,
      "step": 38641
    },
    {
      "epoch": 2.163848135289506,
      "grad_norm": 1.5324336290359497,
      "learning_rate": 9.509868421052632e-05,
      "loss": 0.3539,
      "step": 38642
    },
    {
      "epoch": 2.163904132601635,
      "grad_norm": 1.2346194982528687,
      "learning_rate": 9.509842105263158e-05,
      "loss": 0.3852,
      "step": 38643
    },
    {
      "epoch": 2.163960129913764,
      "grad_norm": 1.1673130989074707,
      "learning_rate": 9.509815789473685e-05,
      "loss": 0.3684,
      "step": 38644
    },
    {
      "epoch": 2.164016127225893,
      "grad_norm": 1.2488288879394531,
      "learning_rate": 9.509789473684211e-05,
      "loss": 0.4818,
      "step": 38645
    },
    {
      "epoch": 2.164072124538022,
      "grad_norm": 1.3335537910461426,
      "learning_rate": 9.509763157894737e-05,
      "loss": 0.3676,
      "step": 38646
    },
    {
      "epoch": 2.164128121850151,
      "grad_norm": 1.3128639459609985,
      "learning_rate": 9.509736842105263e-05,
      "loss": 0.3896,
      "step": 38647
    },
    {
      "epoch": 2.16418411916228,
      "grad_norm": 1.1093014478683472,
      "learning_rate": 9.50971052631579e-05,
      "loss": 0.4251,
      "step": 38648
    },
    {
      "epoch": 2.164240116474409,
      "grad_norm": 1.3227241039276123,
      "learning_rate": 9.509684210526317e-05,
      "loss": 0.5072,
      "step": 38649
    },
    {
      "epoch": 2.164296113786538,
      "grad_norm": 1.337886929512024,
      "learning_rate": 9.509657894736842e-05,
      "loss": 0.4624,
      "step": 38650
    },
    {
      "epoch": 2.164352111098667,
      "grad_norm": 1.1644375324249268,
      "learning_rate": 9.509631578947368e-05,
      "loss": 0.3698,
      "step": 38651
    },
    {
      "epoch": 2.1644081084107962,
      "grad_norm": 1.3222434520721436,
      "learning_rate": 9.509605263157894e-05,
      "loss": 0.441,
      "step": 38652
    },
    {
      "epoch": 2.1644641057229252,
      "grad_norm": 1.2937666177749634,
      "learning_rate": 9.509578947368422e-05,
      "loss": 0.356,
      "step": 38653
    },
    {
      "epoch": 2.1645201030350543,
      "grad_norm": 1.115323543548584,
      "learning_rate": 9.509552631578948e-05,
      "loss": 0.3832,
      "step": 38654
    },
    {
      "epoch": 2.1645761003471833,
      "grad_norm": 1.278096318244934,
      "learning_rate": 9.509526315789474e-05,
      "loss": 0.3858,
      "step": 38655
    },
    {
      "epoch": 2.1646320976593123,
      "grad_norm": 1.301234245300293,
      "learning_rate": 9.5095e-05,
      "loss": 0.5348,
      "step": 38656
    },
    {
      "epoch": 2.1646880949714413,
      "grad_norm": 1.2598164081573486,
      "learning_rate": 9.509473684210527e-05,
      "loss": 0.4293,
      "step": 38657
    },
    {
      "epoch": 2.1647440922835703,
      "grad_norm": 1.2415803670883179,
      "learning_rate": 9.509447368421053e-05,
      "loss": 0.3764,
      "step": 38658
    },
    {
      "epoch": 2.1648000895956994,
      "grad_norm": 1.280566930770874,
      "learning_rate": 9.50942105263158e-05,
      "loss": 0.4864,
      "step": 38659
    },
    {
      "epoch": 2.1648560869078284,
      "grad_norm": 1.072081446647644,
      "learning_rate": 9.509394736842105e-05,
      "loss": 0.3543,
      "step": 38660
    },
    {
      "epoch": 2.1649120842199574,
      "grad_norm": 1.2830604314804077,
      "learning_rate": 9.509368421052632e-05,
      "loss": 0.5242,
      "step": 38661
    },
    {
      "epoch": 2.1649680815320864,
      "grad_norm": 1.5358259677886963,
      "learning_rate": 9.509342105263158e-05,
      "loss": 0.3182,
      "step": 38662
    },
    {
      "epoch": 2.1650240788442154,
      "grad_norm": 1.0890122652053833,
      "learning_rate": 9.509315789473686e-05,
      "loss": 0.3241,
      "step": 38663
    },
    {
      "epoch": 2.1650800761563445,
      "grad_norm": 1.2559322118759155,
      "learning_rate": 9.509289473684212e-05,
      "loss": 0.3738,
      "step": 38664
    },
    {
      "epoch": 2.1651360734684735,
      "grad_norm": 1.1010996103286743,
      "learning_rate": 9.509263157894737e-05,
      "loss": 0.3821,
      "step": 38665
    },
    {
      "epoch": 2.1651920707806025,
      "grad_norm": 1.3612381219863892,
      "learning_rate": 9.509236842105263e-05,
      "loss": 0.4368,
      "step": 38666
    },
    {
      "epoch": 2.1652480680927315,
      "grad_norm": 1.3079760074615479,
      "learning_rate": 9.509210526315791e-05,
      "loss": 0.321,
      "step": 38667
    },
    {
      "epoch": 2.1653040654048605,
      "grad_norm": 1.1332366466522217,
      "learning_rate": 9.509184210526317e-05,
      "loss": 0.4093,
      "step": 38668
    },
    {
      "epoch": 2.1653600627169896,
      "grad_norm": 1.2772388458251953,
      "learning_rate": 9.509157894736841e-05,
      "loss": 0.5936,
      "step": 38669
    },
    {
      "epoch": 2.1654160600291186,
      "grad_norm": 1.3397071361541748,
      "learning_rate": 9.509131578947369e-05,
      "loss": 0.5131,
      "step": 38670
    },
    {
      "epoch": 2.1654720573412476,
      "grad_norm": 1.2153857946395874,
      "learning_rate": 9.509105263157895e-05,
      "loss": 0.3978,
      "step": 38671
    },
    {
      "epoch": 2.1655280546533766,
      "grad_norm": 1.2067170143127441,
      "learning_rate": 9.509078947368422e-05,
      "loss": 0.4489,
      "step": 38672
    },
    {
      "epoch": 2.1655840519655056,
      "grad_norm": 1.1188828945159912,
      "learning_rate": 9.509052631578948e-05,
      "loss": 0.3822,
      "step": 38673
    },
    {
      "epoch": 2.1656400492776346,
      "grad_norm": 1.2397791147232056,
      "learning_rate": 9.509026315789474e-05,
      "loss": 0.4053,
      "step": 38674
    },
    {
      "epoch": 2.1656960465897637,
      "grad_norm": 1.6326227188110352,
      "learning_rate": 9.509e-05,
      "loss": 0.5112,
      "step": 38675
    },
    {
      "epoch": 2.1657520439018927,
      "grad_norm": 2.9639601707458496,
      "learning_rate": 9.508973684210527e-05,
      "loss": 0.6638,
      "step": 38676
    },
    {
      "epoch": 2.1658080412140217,
      "grad_norm": 1.2753801345825195,
      "learning_rate": 9.508947368421053e-05,
      "loss": 0.411,
      "step": 38677
    },
    {
      "epoch": 2.1658640385261507,
      "grad_norm": 1.3385000228881836,
      "learning_rate": 9.508921052631579e-05,
      "loss": 0.3314,
      "step": 38678
    },
    {
      "epoch": 2.1659200358382797,
      "grad_norm": 1.5822936296463013,
      "learning_rate": 9.508894736842105e-05,
      "loss": 0.4313,
      "step": 38679
    },
    {
      "epoch": 2.1659760331504088,
      "grad_norm": 1.1356408596038818,
      "learning_rate": 9.508868421052633e-05,
      "loss": 0.2771,
      "step": 38680
    },
    {
      "epoch": 2.166032030462538,
      "grad_norm": 1.179540991783142,
      "learning_rate": 9.508842105263158e-05,
      "loss": 0.2924,
      "step": 38681
    },
    {
      "epoch": 2.166088027774667,
      "grad_norm": 1.1818360090255737,
      "learning_rate": 9.508815789473684e-05,
      "loss": 0.386,
      "step": 38682
    },
    {
      "epoch": 2.166144025086796,
      "grad_norm": 1.357463002204895,
      "learning_rate": 9.50878947368421e-05,
      "loss": 0.4024,
      "step": 38683
    },
    {
      "epoch": 2.166200022398925,
      "grad_norm": 1.4847887754440308,
      "learning_rate": 9.508763157894738e-05,
      "loss": 0.3718,
      "step": 38684
    },
    {
      "epoch": 2.166256019711054,
      "grad_norm": 1.1857059001922607,
      "learning_rate": 9.508736842105264e-05,
      "loss": 0.4537,
      "step": 38685
    },
    {
      "epoch": 2.166312017023183,
      "grad_norm": 1.2659543752670288,
      "learning_rate": 9.50871052631579e-05,
      "loss": 0.459,
      "step": 38686
    },
    {
      "epoch": 2.166368014335312,
      "grad_norm": 1.1082799434661865,
      "learning_rate": 9.508684210526316e-05,
      "loss": 0.3,
      "step": 38687
    },
    {
      "epoch": 2.166424011647441,
      "grad_norm": 1.6420667171478271,
      "learning_rate": 9.508657894736842e-05,
      "loss": 0.4268,
      "step": 38688
    },
    {
      "epoch": 2.16648000895957,
      "grad_norm": 1.2505706548690796,
      "learning_rate": 9.508631578947369e-05,
      "loss": 0.3536,
      "step": 38689
    },
    {
      "epoch": 2.166536006271699,
      "grad_norm": 1.760909914970398,
      "learning_rate": 9.508605263157895e-05,
      "loss": 0.4157,
      "step": 38690
    },
    {
      "epoch": 2.166592003583828,
      "grad_norm": 1.2517143487930298,
      "learning_rate": 9.508578947368422e-05,
      "loss": 0.4143,
      "step": 38691
    },
    {
      "epoch": 2.166648000895957,
      "grad_norm": 1.3516215085983276,
      "learning_rate": 9.508552631578947e-05,
      "loss": 0.439,
      "step": 38692
    },
    {
      "epoch": 2.166703998208086,
      "grad_norm": 1.533014178276062,
      "learning_rate": 9.508526315789474e-05,
      "loss": 0.4483,
      "step": 38693
    },
    {
      "epoch": 2.166759995520215,
      "grad_norm": 1.5135496854782104,
      "learning_rate": 9.5085e-05,
      "loss": 0.4472,
      "step": 38694
    },
    {
      "epoch": 2.166815992832344,
      "grad_norm": 1.148119330406189,
      "learning_rate": 9.508473684210528e-05,
      "loss": 0.4208,
      "step": 38695
    },
    {
      "epoch": 2.166871990144473,
      "grad_norm": 2.8705575466156006,
      "learning_rate": 9.508447368421053e-05,
      "loss": 0.3584,
      "step": 38696
    },
    {
      "epoch": 2.166927987456602,
      "grad_norm": 1.128533959388733,
      "learning_rate": 9.50842105263158e-05,
      "loss": 0.2863,
      "step": 38697
    },
    {
      "epoch": 2.166983984768731,
      "grad_norm": 1.2889177799224854,
      "learning_rate": 9.508394736842105e-05,
      "loss": 0.5247,
      "step": 38698
    },
    {
      "epoch": 2.16703998208086,
      "grad_norm": 1.1939932107925415,
      "learning_rate": 9.508368421052633e-05,
      "loss": 0.3645,
      "step": 38699
    },
    {
      "epoch": 2.167095979392989,
      "grad_norm": 1.2405543327331543,
      "learning_rate": 9.508342105263159e-05,
      "loss": 0.3339,
      "step": 38700
    },
    {
      "epoch": 2.167151976705118,
      "grad_norm": 1.2746635675430298,
      "learning_rate": 9.508315789473685e-05,
      "loss": 0.4621,
      "step": 38701
    },
    {
      "epoch": 2.167207974017247,
      "grad_norm": 1.770918607711792,
      "learning_rate": 9.508289473684211e-05,
      "loss": 0.4877,
      "step": 38702
    },
    {
      "epoch": 2.167263971329376,
      "grad_norm": 1.111429214477539,
      "learning_rate": 9.508263157894737e-05,
      "loss": 0.318,
      "step": 38703
    },
    {
      "epoch": 2.1673199686415052,
      "grad_norm": 0.9885323643684387,
      "learning_rate": 9.508236842105264e-05,
      "loss": 0.3386,
      "step": 38704
    },
    {
      "epoch": 2.1673759659536342,
      "grad_norm": 1.0358213186264038,
      "learning_rate": 9.50821052631579e-05,
      "loss": 0.4759,
      "step": 38705
    },
    {
      "epoch": 2.1674319632657633,
      "grad_norm": 1.1039458513259888,
      "learning_rate": 9.508184210526316e-05,
      "loss": 0.3714,
      "step": 38706
    },
    {
      "epoch": 2.1674879605778923,
      "grad_norm": 1.2194938659667969,
      "learning_rate": 9.508157894736842e-05,
      "loss": 0.5097,
      "step": 38707
    },
    {
      "epoch": 2.1675439578900213,
      "grad_norm": 1.103987693786621,
      "learning_rate": 9.508131578947369e-05,
      "loss": 0.3322,
      "step": 38708
    },
    {
      "epoch": 2.1675999552021503,
      "grad_norm": 1.4788658618927002,
      "learning_rate": 9.508105263157895e-05,
      "loss": 0.517,
      "step": 38709
    },
    {
      "epoch": 2.1676559525142793,
      "grad_norm": 1.2633532285690308,
      "learning_rate": 9.508078947368421e-05,
      "loss": 0.366,
      "step": 38710
    },
    {
      "epoch": 2.1677119498264084,
      "grad_norm": 1.207823395729065,
      "learning_rate": 9.508052631578947e-05,
      "loss": 0.3953,
      "step": 38711
    },
    {
      "epoch": 2.1677679471385374,
      "grad_norm": 1.4169831275939941,
      "learning_rate": 9.508026315789474e-05,
      "loss": 0.3455,
      "step": 38712
    },
    {
      "epoch": 2.1678239444506664,
      "grad_norm": 1.0815608501434326,
      "learning_rate": 9.508e-05,
      "loss": 0.3824,
      "step": 38713
    },
    {
      "epoch": 2.1678799417627954,
      "grad_norm": 11.572774887084961,
      "learning_rate": 9.507973684210528e-05,
      "loss": 0.3728,
      "step": 38714
    },
    {
      "epoch": 2.1679359390749244,
      "grad_norm": 1.1930299997329712,
      "learning_rate": 9.507947368421052e-05,
      "loss": 0.3353,
      "step": 38715
    },
    {
      "epoch": 2.1679919363870535,
      "grad_norm": 1.3133292198181152,
      "learning_rate": 9.50792105263158e-05,
      "loss": 0.5097,
      "step": 38716
    },
    {
      "epoch": 2.1680479336991825,
      "grad_norm": 1.9562718868255615,
      "learning_rate": 9.507894736842106e-05,
      "loss": 0.3721,
      "step": 38717
    },
    {
      "epoch": 2.1681039310113115,
      "grad_norm": 1.2233657836914062,
      "learning_rate": 9.507868421052633e-05,
      "loss": 0.4567,
      "step": 38718
    },
    {
      "epoch": 2.1681599283234405,
      "grad_norm": 1.4319254159927368,
      "learning_rate": 9.507842105263159e-05,
      "loss": 0.5813,
      "step": 38719
    },
    {
      "epoch": 2.1682159256355695,
      "grad_norm": 1.3189743757247925,
      "learning_rate": 9.507815789473684e-05,
      "loss": 0.4681,
      "step": 38720
    },
    {
      "epoch": 2.1682719229476985,
      "grad_norm": 1.2775558233261108,
      "learning_rate": 9.507789473684211e-05,
      "loss": 0.4193,
      "step": 38721
    },
    {
      "epoch": 2.1683279202598276,
      "grad_norm": 1.1394903659820557,
      "learning_rate": 9.507763157894737e-05,
      "loss": 0.3815,
      "step": 38722
    },
    {
      "epoch": 2.1683839175719566,
      "grad_norm": 0.9606124758720398,
      "learning_rate": 9.507736842105264e-05,
      "loss": 0.3349,
      "step": 38723
    },
    {
      "epoch": 2.1684399148840856,
      "grad_norm": 1.2792342901229858,
      "learning_rate": 9.507710526315789e-05,
      "loss": 0.4129,
      "step": 38724
    },
    {
      "epoch": 2.1684959121962146,
      "grad_norm": 1.188530683517456,
      "learning_rate": 9.507684210526316e-05,
      "loss": 0.4346,
      "step": 38725
    },
    {
      "epoch": 2.1685519095083436,
      "grad_norm": 1.5447030067443848,
      "learning_rate": 9.507657894736842e-05,
      "loss": 0.4115,
      "step": 38726
    },
    {
      "epoch": 2.1686079068204727,
      "grad_norm": 1.632058024406433,
      "learning_rate": 9.50763157894737e-05,
      "loss": 0.3921,
      "step": 38727
    },
    {
      "epoch": 2.1686639041326017,
      "grad_norm": 1.1843558549880981,
      "learning_rate": 9.507605263157895e-05,
      "loss": 0.5779,
      "step": 38728
    },
    {
      "epoch": 2.1687199014447307,
      "grad_norm": 1.3355997800827026,
      "learning_rate": 9.507578947368421e-05,
      "loss": 0.3578,
      "step": 38729
    },
    {
      "epoch": 2.1687758987568597,
      "grad_norm": 1.1126041412353516,
      "learning_rate": 9.507552631578947e-05,
      "loss": 0.377,
      "step": 38730
    },
    {
      "epoch": 2.1688318960689887,
      "grad_norm": 1.345935344696045,
      "learning_rate": 9.507526315789475e-05,
      "loss": 0.5817,
      "step": 38731
    },
    {
      "epoch": 2.1688878933811178,
      "grad_norm": 1.5539087057113647,
      "learning_rate": 9.507500000000001e-05,
      "loss": 0.4198,
      "step": 38732
    },
    {
      "epoch": 2.168943890693247,
      "grad_norm": 1.0643341541290283,
      "learning_rate": 9.507473684210527e-05,
      "loss": 0.3843,
      "step": 38733
    },
    {
      "epoch": 2.168999888005376,
      "grad_norm": 1.3023912906646729,
      "learning_rate": 9.507447368421053e-05,
      "loss": 0.4587,
      "step": 38734
    },
    {
      "epoch": 2.169055885317505,
      "grad_norm": 1.0790605545043945,
      "learning_rate": 9.50742105263158e-05,
      "loss": 0.5219,
      "step": 38735
    },
    {
      "epoch": 2.169111882629634,
      "grad_norm": 1.103473424911499,
      "learning_rate": 9.507394736842106e-05,
      "loss": 0.3692,
      "step": 38736
    },
    {
      "epoch": 2.169167879941763,
      "grad_norm": 1.1826369762420654,
      "learning_rate": 9.507368421052632e-05,
      "loss": 0.3282,
      "step": 38737
    },
    {
      "epoch": 2.169223877253892,
      "grad_norm": 1.1878516674041748,
      "learning_rate": 9.507342105263158e-05,
      "loss": 0.3549,
      "step": 38738
    },
    {
      "epoch": 2.169279874566021,
      "grad_norm": 1.3450536727905273,
      "learning_rate": 9.507315789473684e-05,
      "loss": 0.3837,
      "step": 38739
    },
    {
      "epoch": 2.16933587187815,
      "grad_norm": 18.298437118530273,
      "learning_rate": 9.507289473684211e-05,
      "loss": 0.448,
      "step": 38740
    },
    {
      "epoch": 2.169391869190279,
      "grad_norm": 1.3776415586471558,
      "learning_rate": 9.507263157894737e-05,
      "loss": 0.3615,
      "step": 38741
    },
    {
      "epoch": 2.169447866502408,
      "grad_norm": 1.2406986951828003,
      "learning_rate": 9.507236842105263e-05,
      "loss": 0.5144,
      "step": 38742
    },
    {
      "epoch": 2.169503863814537,
      "grad_norm": 1.4912259578704834,
      "learning_rate": 9.507210526315789e-05,
      "loss": 0.4983,
      "step": 38743
    },
    {
      "epoch": 2.169559861126666,
      "grad_norm": 1.1874314546585083,
      "learning_rate": 9.507184210526316e-05,
      "loss": 0.4494,
      "step": 38744
    },
    {
      "epoch": 2.169615858438795,
      "grad_norm": 1.2723397016525269,
      "learning_rate": 9.507157894736842e-05,
      "loss": 0.307,
      "step": 38745
    },
    {
      "epoch": 2.169671855750924,
      "grad_norm": 1.0741443634033203,
      "learning_rate": 9.50713157894737e-05,
      "loss": 0.3702,
      "step": 38746
    },
    {
      "epoch": 2.169727853063053,
      "grad_norm": 1.28151273727417,
      "learning_rate": 9.507105263157894e-05,
      "loss": 0.4373,
      "step": 38747
    },
    {
      "epoch": 2.169783850375182,
      "grad_norm": 1.0656479597091675,
      "learning_rate": 9.507078947368422e-05,
      "loss": 0.3111,
      "step": 38748
    },
    {
      "epoch": 2.169839847687311,
      "grad_norm": 1.0278630256652832,
      "learning_rate": 9.507052631578948e-05,
      "loss": 0.3361,
      "step": 38749
    },
    {
      "epoch": 2.16989584499944,
      "grad_norm": 1.1096421480178833,
      "learning_rate": 9.507026315789475e-05,
      "loss": 0.2895,
      "step": 38750
    },
    {
      "epoch": 2.169951842311569,
      "grad_norm": 1.2937290668487549,
      "learning_rate": 9.507000000000001e-05,
      "loss": 0.3763,
      "step": 38751
    },
    {
      "epoch": 2.170007839623698,
      "grad_norm": 1.3802119493484497,
      "learning_rate": 9.506973684210527e-05,
      "loss": 0.3737,
      "step": 38752
    },
    {
      "epoch": 2.170063836935827,
      "grad_norm": 1.3635809421539307,
      "learning_rate": 9.506947368421053e-05,
      "loss": 0.3415,
      "step": 38753
    },
    {
      "epoch": 2.170119834247956,
      "grad_norm": 1.760245680809021,
      "learning_rate": 9.50692105263158e-05,
      "loss": 0.3595,
      "step": 38754
    },
    {
      "epoch": 2.170175831560085,
      "grad_norm": 7.00363826751709,
      "learning_rate": 9.506894736842106e-05,
      "loss": 0.4028,
      "step": 38755
    },
    {
      "epoch": 2.170231828872214,
      "grad_norm": 1.265249252319336,
      "learning_rate": 9.506868421052632e-05,
      "loss": 0.4439,
      "step": 38756
    },
    {
      "epoch": 2.1702878261843432,
      "grad_norm": 1.1187540292739868,
      "learning_rate": 9.506842105263158e-05,
      "loss": 0.3813,
      "step": 38757
    },
    {
      "epoch": 2.1703438234964723,
      "grad_norm": 1.2081091403961182,
      "learning_rate": 9.506815789473684e-05,
      "loss": 0.3272,
      "step": 38758
    },
    {
      "epoch": 2.1703998208086013,
      "grad_norm": 1.191619634628296,
      "learning_rate": 9.506789473684211e-05,
      "loss": 0.3484,
      "step": 38759
    },
    {
      "epoch": 2.1704558181207303,
      "grad_norm": 1.1724276542663574,
      "learning_rate": 9.506763157894737e-05,
      "loss": 0.4558,
      "step": 38760
    },
    {
      "epoch": 2.1705118154328593,
      "grad_norm": 1.1554993391036987,
      "learning_rate": 9.506736842105263e-05,
      "loss": 0.385,
      "step": 38761
    },
    {
      "epoch": 2.1705678127449883,
      "grad_norm": 1.2074999809265137,
      "learning_rate": 9.50671052631579e-05,
      "loss": 0.337,
      "step": 38762
    },
    {
      "epoch": 2.1706238100571174,
      "grad_norm": 1.16044282913208,
      "learning_rate": 9.506684210526317e-05,
      "loss": 0.445,
      "step": 38763
    },
    {
      "epoch": 2.1706798073692464,
      "grad_norm": 1.5974719524383545,
      "learning_rate": 9.506657894736843e-05,
      "loss": 0.6338,
      "step": 38764
    },
    {
      "epoch": 2.1707358046813754,
      "grad_norm": 1.3173381090164185,
      "learning_rate": 9.506631578947369e-05,
      "loss": 0.4665,
      "step": 38765
    },
    {
      "epoch": 2.1707918019935044,
      "grad_norm": 1.090227484703064,
      "learning_rate": 9.506605263157895e-05,
      "loss": 0.5187,
      "step": 38766
    },
    {
      "epoch": 2.1708477993056334,
      "grad_norm": 1.4474667310714722,
      "learning_rate": 9.506578947368422e-05,
      "loss": 0.4419,
      "step": 38767
    },
    {
      "epoch": 2.1709037966177624,
      "grad_norm": 1.0742278099060059,
      "learning_rate": 9.506552631578948e-05,
      "loss": 0.357,
      "step": 38768
    },
    {
      "epoch": 2.1709597939298915,
      "grad_norm": 1.0377870798110962,
      "learning_rate": 9.506526315789475e-05,
      "loss": 0.3934,
      "step": 38769
    },
    {
      "epoch": 2.1710157912420205,
      "grad_norm": 1.4398410320281982,
      "learning_rate": 9.5065e-05,
      "loss": 0.3485,
      "step": 38770
    },
    {
      "epoch": 2.1710717885541495,
      "grad_norm": 1.1837376356124878,
      "learning_rate": 9.506473684210527e-05,
      "loss": 0.3261,
      "step": 38771
    },
    {
      "epoch": 2.1711277858662785,
      "grad_norm": 1.4360090494155884,
      "learning_rate": 9.506447368421053e-05,
      "loss": 0.4374,
      "step": 38772
    },
    {
      "epoch": 2.1711837831784075,
      "grad_norm": 1.3159102201461792,
      "learning_rate": 9.506421052631579e-05,
      "loss": 0.4219,
      "step": 38773
    },
    {
      "epoch": 2.1712397804905366,
      "grad_norm": 1.34866201877594,
      "learning_rate": 9.506394736842106e-05,
      "loss": 0.5321,
      "step": 38774
    },
    {
      "epoch": 2.1712957778026656,
      "grad_norm": 1.5206176042556763,
      "learning_rate": 9.506368421052631e-05,
      "loss": 0.4511,
      "step": 38775
    },
    {
      "epoch": 2.1713517751147946,
      "grad_norm": 1.2116981744766235,
      "learning_rate": 9.506342105263158e-05,
      "loss": 0.3733,
      "step": 38776
    },
    {
      "epoch": 2.1714077724269236,
      "grad_norm": 1.347051978111267,
      "learning_rate": 9.506315789473684e-05,
      "loss": 0.3345,
      "step": 38777
    },
    {
      "epoch": 2.1714637697390526,
      "grad_norm": 1.2784743309020996,
      "learning_rate": 9.506289473684212e-05,
      "loss": 0.4808,
      "step": 38778
    },
    {
      "epoch": 2.1715197670511817,
      "grad_norm": 2.8104066848754883,
      "learning_rate": 9.506263157894736e-05,
      "loss": 0.4559,
      "step": 38779
    },
    {
      "epoch": 2.1715757643633107,
      "grad_norm": 1.1485052108764648,
      "learning_rate": 9.506236842105264e-05,
      "loss": 0.3383,
      "step": 38780
    },
    {
      "epoch": 2.1716317616754397,
      "grad_norm": 1.295506477355957,
      "learning_rate": 9.50621052631579e-05,
      "loss": 0.3663,
      "step": 38781
    },
    {
      "epoch": 2.1716877589875687,
      "grad_norm": 1.2281622886657715,
      "learning_rate": 9.506184210526317e-05,
      "loss": 0.4737,
      "step": 38782
    },
    {
      "epoch": 2.1717437562996977,
      "grad_norm": 1.2863173484802246,
      "learning_rate": 9.506157894736843e-05,
      "loss": 0.4569,
      "step": 38783
    },
    {
      "epoch": 2.1717997536118268,
      "grad_norm": 0.9083560109138489,
      "learning_rate": 9.506131578947369e-05,
      "loss": 0.3385,
      "step": 38784
    },
    {
      "epoch": 2.1718557509239558,
      "grad_norm": 1.363257646560669,
      "learning_rate": 9.506105263157895e-05,
      "loss": 0.4225,
      "step": 38785
    },
    {
      "epoch": 2.171911748236085,
      "grad_norm": 1.5215632915496826,
      "learning_rate": 9.506078947368422e-05,
      "loss": 0.4476,
      "step": 38786
    },
    {
      "epoch": 2.171967745548214,
      "grad_norm": 1.3151311874389648,
      "learning_rate": 9.506052631578948e-05,
      "loss": 0.4814,
      "step": 38787
    },
    {
      "epoch": 2.172023742860343,
      "grad_norm": 1.5748416185379028,
      "learning_rate": 9.506026315789474e-05,
      "loss": 0.5159,
      "step": 38788
    },
    {
      "epoch": 2.172079740172472,
      "grad_norm": 1.0687793493270874,
      "learning_rate": 9.506e-05,
      "loss": 0.3712,
      "step": 38789
    },
    {
      "epoch": 2.172135737484601,
      "grad_norm": 1.0927104949951172,
      "learning_rate": 9.505973684210526e-05,
      "loss": 0.3703,
      "step": 38790
    },
    {
      "epoch": 2.17219173479673,
      "grad_norm": 1.2010008096694946,
      "learning_rate": 9.505947368421053e-05,
      "loss": 0.4677,
      "step": 38791
    },
    {
      "epoch": 2.172247732108859,
      "grad_norm": 1.1475380659103394,
      "learning_rate": 9.50592105263158e-05,
      "loss": 0.4434,
      "step": 38792
    },
    {
      "epoch": 2.172303729420988,
      "grad_norm": 1.2391051054000854,
      "learning_rate": 9.505894736842105e-05,
      "loss": 0.4522,
      "step": 38793
    },
    {
      "epoch": 2.172359726733117,
      "grad_norm": 1.1092345714569092,
      "learning_rate": 9.505868421052631e-05,
      "loss": 0.3942,
      "step": 38794
    },
    {
      "epoch": 2.172415724045246,
      "grad_norm": 1.4943710565567017,
      "learning_rate": 9.505842105263159e-05,
      "loss": 0.3755,
      "step": 38795
    },
    {
      "epoch": 2.172471721357375,
      "grad_norm": 1.1359182596206665,
      "learning_rate": 9.505815789473685e-05,
      "loss": 0.3567,
      "step": 38796
    },
    {
      "epoch": 2.172527718669504,
      "grad_norm": 0.9820088744163513,
      "learning_rate": 9.50578947368421e-05,
      "loss": 0.3297,
      "step": 38797
    },
    {
      "epoch": 2.172583715981633,
      "grad_norm": 1.015222191810608,
      "learning_rate": 9.505763157894737e-05,
      "loss": 0.379,
      "step": 38798
    },
    {
      "epoch": 2.172639713293762,
      "grad_norm": 1.2182691097259521,
      "learning_rate": 9.505736842105264e-05,
      "loss": 0.3359,
      "step": 38799
    },
    {
      "epoch": 2.172695710605891,
      "grad_norm": 1.1907962560653687,
      "learning_rate": 9.50571052631579e-05,
      "loss": 0.3002,
      "step": 38800
    },
    {
      "epoch": 2.17275170791802,
      "grad_norm": 1.1289174556732178,
      "learning_rate": 9.505684210526317e-05,
      "loss": 0.3434,
      "step": 38801
    },
    {
      "epoch": 2.172807705230149,
      "grad_norm": 1.3047112226486206,
      "learning_rate": 9.505657894736842e-05,
      "loss": 0.4215,
      "step": 38802
    },
    {
      "epoch": 2.172863702542278,
      "grad_norm": 1.1560428142547607,
      "learning_rate": 9.505631578947369e-05,
      "loss": 0.3619,
      "step": 38803
    },
    {
      "epoch": 2.172919699854407,
      "grad_norm": 1.3775564432144165,
      "learning_rate": 9.505605263157895e-05,
      "loss": 0.5222,
      "step": 38804
    },
    {
      "epoch": 2.172975697166536,
      "grad_norm": 1.239606261253357,
      "learning_rate": 9.505578947368422e-05,
      "loss": 0.3483,
      "step": 38805
    },
    {
      "epoch": 2.173031694478665,
      "grad_norm": 1.0560238361358643,
      "learning_rate": 9.505552631578948e-05,
      "loss": 0.4161,
      "step": 38806
    },
    {
      "epoch": 2.173087691790794,
      "grad_norm": 10.787339210510254,
      "learning_rate": 9.505526315789473e-05,
      "loss": 0.4249,
      "step": 38807
    },
    {
      "epoch": 2.173143689102923,
      "grad_norm": 1.4692264795303345,
      "learning_rate": 9.5055e-05,
      "loss": 0.4207,
      "step": 38808
    },
    {
      "epoch": 2.1731996864150522,
      "grad_norm": 1.3276922702789307,
      "learning_rate": 9.505473684210526e-05,
      "loss": 0.5401,
      "step": 38809
    },
    {
      "epoch": 2.1732556837271813,
      "grad_norm": 1.1649796962738037,
      "learning_rate": 9.505447368421054e-05,
      "loss": 0.3488,
      "step": 38810
    },
    {
      "epoch": 2.1733116810393103,
      "grad_norm": 1.348006010055542,
      "learning_rate": 9.50542105263158e-05,
      "loss": 0.3728,
      "step": 38811
    },
    {
      "epoch": 2.1733676783514393,
      "grad_norm": 1.331128716468811,
      "learning_rate": 9.505394736842106e-05,
      "loss": 0.5421,
      "step": 38812
    },
    {
      "epoch": 2.1734236756635683,
      "grad_norm": 1.311055302619934,
      "learning_rate": 9.505368421052632e-05,
      "loss": 0.4138,
      "step": 38813
    },
    {
      "epoch": 2.1734796729756973,
      "grad_norm": 1.1107230186462402,
      "learning_rate": 9.505342105263159e-05,
      "loss": 0.3522,
      "step": 38814
    },
    {
      "epoch": 2.1735356702878263,
      "grad_norm": 1.0440689325332642,
      "learning_rate": 9.505315789473685e-05,
      "loss": 0.3763,
      "step": 38815
    },
    {
      "epoch": 2.1735916675999554,
      "grad_norm": 1.1436115503311157,
      "learning_rate": 9.505289473684211e-05,
      "loss": 0.4915,
      "step": 38816
    },
    {
      "epoch": 2.1736476649120844,
      "grad_norm": 1.4141136407852173,
      "learning_rate": 9.505263157894737e-05,
      "loss": 0.5187,
      "step": 38817
    },
    {
      "epoch": 2.1737036622242134,
      "grad_norm": 1.1843149662017822,
      "learning_rate": 9.505236842105264e-05,
      "loss": 0.4248,
      "step": 38818
    },
    {
      "epoch": 2.1737596595363424,
      "grad_norm": 1.2413588762283325,
      "learning_rate": 9.50521052631579e-05,
      "loss": 0.3571,
      "step": 38819
    },
    {
      "epoch": 2.1738156568484714,
      "grad_norm": 1.5059640407562256,
      "learning_rate": 9.505184210526316e-05,
      "loss": 0.4477,
      "step": 38820
    },
    {
      "epoch": 2.1738716541606005,
      "grad_norm": 1.0821102857589722,
      "learning_rate": 9.505157894736842e-05,
      "loss": 0.3638,
      "step": 38821
    },
    {
      "epoch": 2.1739276514727295,
      "grad_norm": 1.2391413450241089,
      "learning_rate": 9.50513157894737e-05,
      "loss": 0.3356,
      "step": 38822
    },
    {
      "epoch": 2.1739836487848585,
      "grad_norm": 2.526907444000244,
      "learning_rate": 9.505105263157895e-05,
      "loss": 0.3537,
      "step": 38823
    },
    {
      "epoch": 2.1740396460969875,
      "grad_norm": 1.299836277961731,
      "learning_rate": 9.505078947368421e-05,
      "loss": 0.5154,
      "step": 38824
    },
    {
      "epoch": 2.1740956434091165,
      "grad_norm": 1.2905185222625732,
      "learning_rate": 9.505052631578947e-05,
      "loss": 0.4221,
      "step": 38825
    },
    {
      "epoch": 2.1741516407212456,
      "grad_norm": 1.4165515899658203,
      "learning_rate": 9.505026315789473e-05,
      "loss": 0.4411,
      "step": 38826
    },
    {
      "epoch": 2.1742076380333746,
      "grad_norm": 1.1649396419525146,
      "learning_rate": 9.505e-05,
      "loss": 0.3733,
      "step": 38827
    },
    {
      "epoch": 2.1742636353455036,
      "grad_norm": 1.7880675792694092,
      "learning_rate": 9.504973684210527e-05,
      "loss": 0.5652,
      "step": 38828
    },
    {
      "epoch": 2.1743196326576326,
      "grad_norm": 1.4706186056137085,
      "learning_rate": 9.504947368421053e-05,
      "loss": 0.5711,
      "step": 38829
    },
    {
      "epoch": 2.1743756299697616,
      "grad_norm": 1.3184335231781006,
      "learning_rate": 9.504921052631579e-05,
      "loss": 0.6148,
      "step": 38830
    },
    {
      "epoch": 2.1744316272818907,
      "grad_norm": 1.4265294075012207,
      "learning_rate": 9.504894736842106e-05,
      "loss": 0.5209,
      "step": 38831
    },
    {
      "epoch": 2.1744876245940197,
      "grad_norm": 1.400057315826416,
      "learning_rate": 9.504868421052632e-05,
      "loss": 0.4364,
      "step": 38832
    },
    {
      "epoch": 2.1745436219061487,
      "grad_norm": 1.2918931245803833,
      "learning_rate": 9.504842105263159e-05,
      "loss": 0.4272,
      "step": 38833
    },
    {
      "epoch": 2.1745996192182777,
      "grad_norm": 1.2086308002471924,
      "learning_rate": 9.504815789473684e-05,
      "loss": 0.4111,
      "step": 38834
    },
    {
      "epoch": 2.1746556165304067,
      "grad_norm": 1.6088145971298218,
      "learning_rate": 9.504789473684211e-05,
      "loss": 0.4501,
      "step": 38835
    },
    {
      "epoch": 2.1747116138425358,
      "grad_norm": 1.1741492748260498,
      "learning_rate": 9.504763157894737e-05,
      "loss": 0.4131,
      "step": 38836
    },
    {
      "epoch": 2.1747676111546648,
      "grad_norm": 1.002426028251648,
      "learning_rate": 9.504736842105264e-05,
      "loss": 0.3974,
      "step": 38837
    },
    {
      "epoch": 2.174823608466794,
      "grad_norm": 1.4331179857254028,
      "learning_rate": 9.50471052631579e-05,
      "loss": 0.4252,
      "step": 38838
    },
    {
      "epoch": 2.174879605778923,
      "grad_norm": 1.3485172986984253,
      "learning_rate": 9.504684210526316e-05,
      "loss": 0.3749,
      "step": 38839
    },
    {
      "epoch": 2.174935603091052,
      "grad_norm": 1.1548750400543213,
      "learning_rate": 9.504657894736842e-05,
      "loss": 0.4604,
      "step": 38840
    },
    {
      "epoch": 2.174991600403181,
      "grad_norm": 1.2099618911743164,
      "learning_rate": 9.504631578947368e-05,
      "loss": 0.3838,
      "step": 38841
    },
    {
      "epoch": 2.17504759771531,
      "grad_norm": 1.1643341779708862,
      "learning_rate": 9.504605263157896e-05,
      "loss": 0.3674,
      "step": 38842
    },
    {
      "epoch": 2.175103595027439,
      "grad_norm": 1.2454111576080322,
      "learning_rate": 9.504578947368422e-05,
      "loss": 0.4027,
      "step": 38843
    },
    {
      "epoch": 2.175159592339568,
      "grad_norm": 1.327427625656128,
      "learning_rate": 9.504552631578948e-05,
      "loss": 0.4797,
      "step": 38844
    },
    {
      "epoch": 2.175215589651697,
      "grad_norm": 1.336377739906311,
      "learning_rate": 9.504526315789474e-05,
      "loss": 0.333,
      "step": 38845
    },
    {
      "epoch": 2.175271586963826,
      "grad_norm": 1.1384215354919434,
      "learning_rate": 9.504500000000001e-05,
      "loss": 0.364,
      "step": 38846
    },
    {
      "epoch": 2.175327584275955,
      "grad_norm": 1.369988203048706,
      "learning_rate": 9.504473684210527e-05,
      "loss": 0.5326,
      "step": 38847
    },
    {
      "epoch": 2.1753835815880835,
      "grad_norm": 1.1162896156311035,
      "learning_rate": 9.504447368421053e-05,
      "loss": 0.356,
      "step": 38848
    },
    {
      "epoch": 2.175439578900213,
      "grad_norm": 1.1583393812179565,
      "learning_rate": 9.504421052631579e-05,
      "loss": 0.4827,
      "step": 38849
    },
    {
      "epoch": 2.1754955762123416,
      "grad_norm": 0.9568096399307251,
      "learning_rate": 9.504394736842106e-05,
      "loss": 0.2738,
      "step": 38850
    },
    {
      "epoch": 2.175551573524471,
      "grad_norm": 1.0295175313949585,
      "learning_rate": 9.504368421052632e-05,
      "loss": 0.3141,
      "step": 38851
    },
    {
      "epoch": 2.1756075708365996,
      "grad_norm": 1.140512228012085,
      "learning_rate": 9.504342105263158e-05,
      "loss": 0.4043,
      "step": 38852
    },
    {
      "epoch": 2.175663568148729,
      "grad_norm": 1.1284157037734985,
      "learning_rate": 9.504315789473684e-05,
      "loss": 0.4028,
      "step": 38853
    },
    {
      "epoch": 2.1757195654608577,
      "grad_norm": 1.0963139533996582,
      "learning_rate": 9.504289473684211e-05,
      "loss": 0.397,
      "step": 38854
    },
    {
      "epoch": 2.175775562772987,
      "grad_norm": 1.193366289138794,
      "learning_rate": 9.504263157894737e-05,
      "loss": 0.4722,
      "step": 38855
    },
    {
      "epoch": 2.1758315600851157,
      "grad_norm": 1.4941381216049194,
      "learning_rate": 9.504236842105265e-05,
      "loss": 0.4345,
      "step": 38856
    },
    {
      "epoch": 2.175887557397245,
      "grad_norm": 1.2314658164978027,
      "learning_rate": 9.504210526315789e-05,
      "loss": 0.4036,
      "step": 38857
    },
    {
      "epoch": 2.1759435547093737,
      "grad_norm": 1.2378042936325073,
      "learning_rate": 9.504184210526315e-05,
      "loss": 0.4267,
      "step": 38858
    },
    {
      "epoch": 2.175999552021503,
      "grad_norm": 1.2242639064788818,
      "learning_rate": 9.504157894736843e-05,
      "loss": 0.3896,
      "step": 38859
    },
    {
      "epoch": 2.1760555493336318,
      "grad_norm": 1.1625510454177856,
      "learning_rate": 9.504131578947369e-05,
      "loss": 0.4591,
      "step": 38860
    },
    {
      "epoch": 2.1761115466457612,
      "grad_norm": 1.1546874046325684,
      "learning_rate": 9.504105263157896e-05,
      "loss": 0.324,
      "step": 38861
    },
    {
      "epoch": 2.17616754395789,
      "grad_norm": 1.0996016263961792,
      "learning_rate": 9.50407894736842e-05,
      "loss": 0.3849,
      "step": 38862
    },
    {
      "epoch": 2.1762235412700193,
      "grad_norm": 1.4061026573181152,
      "learning_rate": 9.504052631578948e-05,
      "loss": 0.63,
      "step": 38863
    },
    {
      "epoch": 2.176279538582148,
      "grad_norm": 1.5557388067245483,
      "learning_rate": 9.504026315789474e-05,
      "loss": 0.511,
      "step": 38864
    },
    {
      "epoch": 2.1763355358942773,
      "grad_norm": 1.2355982065200806,
      "learning_rate": 9.504000000000001e-05,
      "loss": 0.4664,
      "step": 38865
    },
    {
      "epoch": 2.176391533206406,
      "grad_norm": 1.4220056533813477,
      "learning_rate": 9.503973684210527e-05,
      "loss": 0.3973,
      "step": 38866
    },
    {
      "epoch": 2.1764475305185353,
      "grad_norm": 1.2815178632736206,
      "learning_rate": 9.503947368421053e-05,
      "loss": 0.3512,
      "step": 38867
    },
    {
      "epoch": 2.176503527830664,
      "grad_norm": 1.3415242433547974,
      "learning_rate": 9.503921052631579e-05,
      "loss": 0.4323,
      "step": 38868
    },
    {
      "epoch": 2.176559525142793,
      "grad_norm": 0.9945501685142517,
      "learning_rate": 9.503894736842106e-05,
      "loss": 0.3058,
      "step": 38869
    },
    {
      "epoch": 2.176615522454922,
      "grad_norm": 1.1742113828659058,
      "learning_rate": 9.503868421052632e-05,
      "loss": 0.42,
      "step": 38870
    },
    {
      "epoch": 2.176671519767051,
      "grad_norm": 1.1134363412857056,
      "learning_rate": 9.503842105263158e-05,
      "loss": 0.3647,
      "step": 38871
    },
    {
      "epoch": 2.17672751707918,
      "grad_norm": 1.309715986251831,
      "learning_rate": 9.503815789473684e-05,
      "loss": 0.393,
      "step": 38872
    },
    {
      "epoch": 2.176783514391309,
      "grad_norm": 1.3498668670654297,
      "learning_rate": 9.503789473684212e-05,
      "loss": 0.3947,
      "step": 38873
    },
    {
      "epoch": 2.176839511703438,
      "grad_norm": 1.7171504497528076,
      "learning_rate": 9.503763157894738e-05,
      "loss": 0.401,
      "step": 38874
    },
    {
      "epoch": 2.176895509015567,
      "grad_norm": 1.3124399185180664,
      "learning_rate": 9.503736842105264e-05,
      "loss": 0.4468,
      "step": 38875
    },
    {
      "epoch": 2.176951506327696,
      "grad_norm": 1.4594861268997192,
      "learning_rate": 9.50371052631579e-05,
      "loss": 0.6204,
      "step": 38876
    },
    {
      "epoch": 2.177007503639825,
      "grad_norm": 1.3782179355621338,
      "learning_rate": 9.503684210526316e-05,
      "loss": 0.413,
      "step": 38877
    },
    {
      "epoch": 2.177063500951954,
      "grad_norm": 1.1675093173980713,
      "learning_rate": 9.503657894736843e-05,
      "loss": 0.3121,
      "step": 38878
    },
    {
      "epoch": 2.177119498264083,
      "grad_norm": 1.2666668891906738,
      "learning_rate": 9.503631578947369e-05,
      "loss": 0.5612,
      "step": 38879
    },
    {
      "epoch": 2.177175495576212,
      "grad_norm": 1.3125929832458496,
      "learning_rate": 9.503605263157895e-05,
      "loss": 0.3641,
      "step": 38880
    },
    {
      "epoch": 2.177231492888341,
      "grad_norm": 1.1350667476654053,
      "learning_rate": 9.503578947368421e-05,
      "loss": 0.4192,
      "step": 38881
    },
    {
      "epoch": 2.17728749020047,
      "grad_norm": 1.1488652229309082,
      "learning_rate": 9.503552631578948e-05,
      "loss": 0.4399,
      "step": 38882
    },
    {
      "epoch": 2.177343487512599,
      "grad_norm": 1.3907078504562378,
      "learning_rate": 9.503526315789474e-05,
      "loss": 0.4261,
      "step": 38883
    },
    {
      "epoch": 2.1773994848247282,
      "grad_norm": 1.2135865688323975,
      "learning_rate": 9.5035e-05,
      "loss": 0.4621,
      "step": 38884
    },
    {
      "epoch": 2.1774554821368572,
      "grad_norm": 1.1741116046905518,
      "learning_rate": 9.503473684210526e-05,
      "loss": 0.3871,
      "step": 38885
    },
    {
      "epoch": 2.1775114794489863,
      "grad_norm": 1.2427276372909546,
      "learning_rate": 9.503447368421053e-05,
      "loss": 0.4548,
      "step": 38886
    },
    {
      "epoch": 2.1775674767611153,
      "grad_norm": 1.0792986154556274,
      "learning_rate": 9.50342105263158e-05,
      "loss": 0.4223,
      "step": 38887
    },
    {
      "epoch": 2.1776234740732443,
      "grad_norm": 1.2899301052093506,
      "learning_rate": 9.503394736842107e-05,
      "loss": 0.4178,
      "step": 38888
    },
    {
      "epoch": 2.1776794713853733,
      "grad_norm": 1.2221159934997559,
      "learning_rate": 9.503368421052631e-05,
      "loss": 0.3752,
      "step": 38889
    },
    {
      "epoch": 2.1777354686975023,
      "grad_norm": 1.1396558284759521,
      "learning_rate": 9.503342105263159e-05,
      "loss": 0.3783,
      "step": 38890
    },
    {
      "epoch": 2.1777914660096314,
      "grad_norm": 1.1767154932022095,
      "learning_rate": 9.503315789473685e-05,
      "loss": 0.4357,
      "step": 38891
    },
    {
      "epoch": 2.1778474633217604,
      "grad_norm": 1.1137216091156006,
      "learning_rate": 9.503289473684212e-05,
      "loss": 0.3027,
      "step": 38892
    },
    {
      "epoch": 2.1779034606338894,
      "grad_norm": 1.1826591491699219,
      "learning_rate": 9.503263157894738e-05,
      "loss": 0.3199,
      "step": 38893
    },
    {
      "epoch": 2.1779594579460184,
      "grad_norm": 1.4285383224487305,
      "learning_rate": 9.503236842105263e-05,
      "loss": 0.3135,
      "step": 38894
    },
    {
      "epoch": 2.1780154552581474,
      "grad_norm": 1.256919264793396,
      "learning_rate": 9.50321052631579e-05,
      "loss": 0.3256,
      "step": 38895
    },
    {
      "epoch": 2.1780714525702765,
      "grad_norm": 1.0414938926696777,
      "learning_rate": 9.503184210526316e-05,
      "loss": 0.348,
      "step": 38896
    },
    {
      "epoch": 2.1781274498824055,
      "grad_norm": 1.1992238759994507,
      "learning_rate": 9.503157894736843e-05,
      "loss": 0.3693,
      "step": 38897
    },
    {
      "epoch": 2.1781834471945345,
      "grad_norm": 1.8195973634719849,
      "learning_rate": 9.503131578947369e-05,
      "loss": 0.4111,
      "step": 38898
    },
    {
      "epoch": 2.1782394445066635,
      "grad_norm": 1.1915786266326904,
      "learning_rate": 9.503105263157895e-05,
      "loss": 0.4478,
      "step": 38899
    },
    {
      "epoch": 2.1782954418187925,
      "grad_norm": 1.3864026069641113,
      "learning_rate": 9.503078947368421e-05,
      "loss": 0.4103,
      "step": 38900
    },
    {
      "epoch": 2.1783514391309216,
      "grad_norm": 1.1396385431289673,
      "learning_rate": 9.503052631578948e-05,
      "loss": 0.3759,
      "step": 38901
    },
    {
      "epoch": 2.1784074364430506,
      "grad_norm": 1.4047677516937256,
      "learning_rate": 9.503026315789474e-05,
      "loss": 0.305,
      "step": 38902
    },
    {
      "epoch": 2.1784634337551796,
      "grad_norm": 1.1481488943099976,
      "learning_rate": 9.503e-05,
      "loss": 0.3261,
      "step": 38903
    },
    {
      "epoch": 2.1785194310673086,
      "grad_norm": 0.9933944344520569,
      "learning_rate": 9.502973684210526e-05,
      "loss": 0.4122,
      "step": 38904
    },
    {
      "epoch": 2.1785754283794376,
      "grad_norm": 1.276698350906372,
      "learning_rate": 9.502947368421054e-05,
      "loss": 0.4318,
      "step": 38905
    },
    {
      "epoch": 2.1786314256915666,
      "grad_norm": 1.16350257396698,
      "learning_rate": 9.50292105263158e-05,
      "loss": 0.3173,
      "step": 38906
    },
    {
      "epoch": 2.1786874230036957,
      "grad_norm": 1.2647807598114014,
      "learning_rate": 9.502894736842106e-05,
      "loss": 0.5815,
      "step": 38907
    },
    {
      "epoch": 2.1787434203158247,
      "grad_norm": 1.2193032503128052,
      "learning_rate": 9.502868421052632e-05,
      "loss": 0.3209,
      "step": 38908
    },
    {
      "epoch": 2.1787994176279537,
      "grad_norm": 1.2528175115585327,
      "learning_rate": 9.502842105263159e-05,
      "loss": 0.33,
      "step": 38909
    },
    {
      "epoch": 2.1788554149400827,
      "grad_norm": 1.1707937717437744,
      "learning_rate": 9.502815789473685e-05,
      "loss": 0.3684,
      "step": 38910
    },
    {
      "epoch": 2.1789114122522117,
      "grad_norm": 1.7491612434387207,
      "learning_rate": 9.502789473684211e-05,
      "loss": 0.4934,
      "step": 38911
    },
    {
      "epoch": 2.1789674095643408,
      "grad_norm": 1.155828833580017,
      "learning_rate": 9.502763157894737e-05,
      "loss": 0.3752,
      "step": 38912
    },
    {
      "epoch": 2.17902340687647,
      "grad_norm": 1.2812103033065796,
      "learning_rate": 9.502736842105263e-05,
      "loss": 0.4206,
      "step": 38913
    },
    {
      "epoch": 2.179079404188599,
      "grad_norm": 1.065891146659851,
      "learning_rate": 9.50271052631579e-05,
      "loss": 0.3271,
      "step": 38914
    },
    {
      "epoch": 2.179135401500728,
      "grad_norm": 1.0294924974441528,
      "learning_rate": 9.502684210526316e-05,
      "loss": 0.3177,
      "step": 38915
    },
    {
      "epoch": 2.179191398812857,
      "grad_norm": 1.785454511642456,
      "learning_rate": 9.502657894736843e-05,
      "loss": 0.394,
      "step": 38916
    },
    {
      "epoch": 2.179247396124986,
      "grad_norm": 1.1948015689849854,
      "learning_rate": 9.502631578947368e-05,
      "loss": 0.4498,
      "step": 38917
    },
    {
      "epoch": 2.179303393437115,
      "grad_norm": 1.1637802124023438,
      "learning_rate": 9.502605263157895e-05,
      "loss": 0.4534,
      "step": 38918
    },
    {
      "epoch": 2.179359390749244,
      "grad_norm": 1.2530790567398071,
      "learning_rate": 9.502578947368421e-05,
      "loss": 0.3578,
      "step": 38919
    },
    {
      "epoch": 2.179415388061373,
      "grad_norm": 1.146479845046997,
      "learning_rate": 9.502552631578949e-05,
      "loss": 0.4398,
      "step": 38920
    },
    {
      "epoch": 2.179471385373502,
      "grad_norm": 1.2815388441085815,
      "learning_rate": 9.502526315789475e-05,
      "loss": 0.5147,
      "step": 38921
    },
    {
      "epoch": 2.179527382685631,
      "grad_norm": 1.3130314350128174,
      "learning_rate": 9.5025e-05,
      "loss": 0.3835,
      "step": 38922
    },
    {
      "epoch": 2.17958337999776,
      "grad_norm": 1.3097599744796753,
      "learning_rate": 9.502473684210527e-05,
      "loss": 0.6164,
      "step": 38923
    },
    {
      "epoch": 2.179639377309889,
      "grad_norm": 1.446223258972168,
      "learning_rate": 9.502447368421054e-05,
      "loss": 0.5908,
      "step": 38924
    },
    {
      "epoch": 2.179695374622018,
      "grad_norm": 3.3467094898223877,
      "learning_rate": 9.50242105263158e-05,
      "loss": 0.4512,
      "step": 38925
    },
    {
      "epoch": 2.179751371934147,
      "grad_norm": 1.2375922203063965,
      "learning_rate": 9.502394736842106e-05,
      "loss": 0.3563,
      "step": 38926
    },
    {
      "epoch": 2.179807369246276,
      "grad_norm": 1.1299866437911987,
      "learning_rate": 9.502368421052632e-05,
      "loss": 0.3449,
      "step": 38927
    },
    {
      "epoch": 2.179863366558405,
      "grad_norm": 1.254575252532959,
      "learning_rate": 9.502342105263158e-05,
      "loss": 0.3399,
      "step": 38928
    },
    {
      "epoch": 2.179919363870534,
      "grad_norm": 1.0637726783752441,
      "learning_rate": 9.502315789473685e-05,
      "loss": 0.3475,
      "step": 38929
    },
    {
      "epoch": 2.179975361182663,
      "grad_norm": 1.1697044372558594,
      "learning_rate": 9.502289473684211e-05,
      "loss": 0.3728,
      "step": 38930
    },
    {
      "epoch": 2.180031358494792,
      "grad_norm": 1.3098459243774414,
      "learning_rate": 9.502263157894737e-05,
      "loss": 0.4322,
      "step": 38931
    },
    {
      "epoch": 2.180087355806921,
      "grad_norm": 2.233530282974243,
      "learning_rate": 9.502236842105263e-05,
      "loss": 0.3817,
      "step": 38932
    },
    {
      "epoch": 2.18014335311905,
      "grad_norm": 1.1003053188323975,
      "learning_rate": 9.50221052631579e-05,
      "loss": 0.4172,
      "step": 38933
    },
    {
      "epoch": 2.180199350431179,
      "grad_norm": 1.119368314743042,
      "learning_rate": 9.502184210526316e-05,
      "loss": 0.5839,
      "step": 38934
    },
    {
      "epoch": 2.180255347743308,
      "grad_norm": 1.164452075958252,
      "learning_rate": 9.502157894736842e-05,
      "loss": 0.4346,
      "step": 38935
    },
    {
      "epoch": 2.1803113450554372,
      "grad_norm": 1.368444561958313,
      "learning_rate": 9.502131578947368e-05,
      "loss": 0.4748,
      "step": 38936
    },
    {
      "epoch": 2.1803673423675662,
      "grad_norm": 1.2676095962524414,
      "learning_rate": 9.502105263157896e-05,
      "loss": 0.3016,
      "step": 38937
    },
    {
      "epoch": 2.1804233396796953,
      "grad_norm": 1.023878812789917,
      "learning_rate": 9.502078947368422e-05,
      "loss": 0.3663,
      "step": 38938
    },
    {
      "epoch": 2.1804793369918243,
      "grad_norm": 1.0366082191467285,
      "learning_rate": 9.502052631578948e-05,
      "loss": 0.4657,
      "step": 38939
    },
    {
      "epoch": 2.1805353343039533,
      "grad_norm": 1.2221672534942627,
      "learning_rate": 9.502026315789474e-05,
      "loss": 0.3695,
      "step": 38940
    },
    {
      "epoch": 2.1805913316160823,
      "grad_norm": 1.192507028579712,
      "learning_rate": 9.502000000000001e-05,
      "loss": 0.4121,
      "step": 38941
    },
    {
      "epoch": 2.1806473289282113,
      "grad_norm": 1.4165290594100952,
      "learning_rate": 9.501973684210527e-05,
      "loss": 0.4447,
      "step": 38942
    },
    {
      "epoch": 2.1807033262403404,
      "grad_norm": 1.3653814792633057,
      "learning_rate": 9.501947368421054e-05,
      "loss": 0.4148,
      "step": 38943
    },
    {
      "epoch": 2.1807593235524694,
      "grad_norm": 1.22435462474823,
      "learning_rate": 9.501921052631579e-05,
      "loss": 0.3519,
      "step": 38944
    },
    {
      "epoch": 2.1808153208645984,
      "grad_norm": 1.050353765487671,
      "learning_rate": 9.501894736842105e-05,
      "loss": 0.352,
      "step": 38945
    },
    {
      "epoch": 2.1808713181767274,
      "grad_norm": 1.0864853858947754,
      "learning_rate": 9.501868421052632e-05,
      "loss": 0.4096,
      "step": 38946
    },
    {
      "epoch": 2.1809273154888564,
      "grad_norm": 1.4147504568099976,
      "learning_rate": 9.501842105263158e-05,
      "loss": 0.4296,
      "step": 38947
    },
    {
      "epoch": 2.1809833128009855,
      "grad_norm": 1.0325373411178589,
      "learning_rate": 9.501815789473685e-05,
      "loss": 0.3209,
      "step": 38948
    },
    {
      "epoch": 2.1810393101131145,
      "grad_norm": 1.2250773906707764,
      "learning_rate": 9.50178947368421e-05,
      "loss": 0.4107,
      "step": 38949
    },
    {
      "epoch": 2.1810953074252435,
      "grad_norm": 1.2440638542175293,
      "learning_rate": 9.501763157894737e-05,
      "loss": 0.4684,
      "step": 38950
    },
    {
      "epoch": 2.1811513047373725,
      "grad_norm": 1.2981528043746948,
      "learning_rate": 9.501736842105263e-05,
      "loss": 0.6772,
      "step": 38951
    },
    {
      "epoch": 2.1812073020495015,
      "grad_norm": 1.3794368505477905,
      "learning_rate": 9.50171052631579e-05,
      "loss": 0.5427,
      "step": 38952
    },
    {
      "epoch": 2.1812632993616305,
      "grad_norm": 1.3692600727081299,
      "learning_rate": 9.501684210526317e-05,
      "loss": 0.4986,
      "step": 38953
    },
    {
      "epoch": 2.1813192966737596,
      "grad_norm": 0.9568555951118469,
      "learning_rate": 9.501657894736843e-05,
      "loss": 0.2829,
      "step": 38954
    },
    {
      "epoch": 2.1813752939858886,
      "grad_norm": 1.1541242599487305,
      "learning_rate": 9.501631578947369e-05,
      "loss": 0.3728,
      "step": 38955
    },
    {
      "epoch": 2.1814312912980176,
      "grad_norm": 1.4491171836853027,
      "learning_rate": 9.501605263157896e-05,
      "loss": 0.5582,
      "step": 38956
    },
    {
      "epoch": 2.1814872886101466,
      "grad_norm": 1.1712186336517334,
      "learning_rate": 9.501578947368422e-05,
      "loss": 0.4367,
      "step": 38957
    },
    {
      "epoch": 2.1815432859222756,
      "grad_norm": 1.528090476989746,
      "learning_rate": 9.501552631578948e-05,
      "loss": 0.4122,
      "step": 38958
    },
    {
      "epoch": 2.1815992832344047,
      "grad_norm": 1.26664400100708,
      "learning_rate": 9.501526315789474e-05,
      "loss": 0.4434,
      "step": 38959
    },
    {
      "epoch": 2.1816552805465337,
      "grad_norm": 1.0228183269500732,
      "learning_rate": 9.501500000000001e-05,
      "loss": 0.3562,
      "step": 38960
    },
    {
      "epoch": 2.1817112778586627,
      "grad_norm": 1.4269508123397827,
      "learning_rate": 9.501473684210527e-05,
      "loss": 0.5148,
      "step": 38961
    },
    {
      "epoch": 2.1817672751707917,
      "grad_norm": 1.1177386045455933,
      "learning_rate": 9.501447368421053e-05,
      "loss": 0.4435,
      "step": 38962
    },
    {
      "epoch": 2.1818232724829207,
      "grad_norm": 1.1529814004898071,
      "learning_rate": 9.501421052631579e-05,
      "loss": 0.4385,
      "step": 38963
    },
    {
      "epoch": 2.1818792697950498,
      "grad_norm": 1.5116426944732666,
      "learning_rate": 9.501394736842105e-05,
      "loss": 0.3738,
      "step": 38964
    },
    {
      "epoch": 2.1819352671071788,
      "grad_norm": 1.2246263027191162,
      "learning_rate": 9.501368421052632e-05,
      "loss": 0.4642,
      "step": 38965
    },
    {
      "epoch": 2.181991264419308,
      "grad_norm": 1.2727391719818115,
      "learning_rate": 9.501342105263158e-05,
      "loss": 0.3506,
      "step": 38966
    },
    {
      "epoch": 2.182047261731437,
      "grad_norm": 1.125381588935852,
      "learning_rate": 9.501315789473684e-05,
      "loss": 0.4078,
      "step": 38967
    },
    {
      "epoch": 2.182103259043566,
      "grad_norm": 1.204925775527954,
      "learning_rate": 9.50128947368421e-05,
      "loss": 0.4381,
      "step": 38968
    },
    {
      "epoch": 2.182159256355695,
      "grad_norm": 1.3298226594924927,
      "learning_rate": 9.501263157894738e-05,
      "loss": 0.5204,
      "step": 38969
    },
    {
      "epoch": 2.182215253667824,
      "grad_norm": 1.2462958097457886,
      "learning_rate": 9.501236842105264e-05,
      "loss": 0.3601,
      "step": 38970
    },
    {
      "epoch": 2.182271250979953,
      "grad_norm": 1.0171583890914917,
      "learning_rate": 9.501210526315791e-05,
      "loss": 0.2963,
      "step": 38971
    },
    {
      "epoch": 2.182327248292082,
      "grad_norm": 1.1733615398406982,
      "learning_rate": 9.501184210526315e-05,
      "loss": 0.3303,
      "step": 38972
    },
    {
      "epoch": 2.182383245604211,
      "grad_norm": 1.3030155897140503,
      "learning_rate": 9.501157894736843e-05,
      "loss": 0.5021,
      "step": 38973
    },
    {
      "epoch": 2.18243924291634,
      "grad_norm": 1.178606390953064,
      "learning_rate": 9.501131578947369e-05,
      "loss": 0.444,
      "step": 38974
    },
    {
      "epoch": 2.182495240228469,
      "grad_norm": 1.173682689666748,
      "learning_rate": 9.501105263157896e-05,
      "loss": 0.3879,
      "step": 38975
    },
    {
      "epoch": 2.182551237540598,
      "grad_norm": 1.5737130641937256,
      "learning_rate": 9.501078947368422e-05,
      "loss": 0.5326,
      "step": 38976
    },
    {
      "epoch": 2.182607234852727,
      "grad_norm": 1.249324083328247,
      "learning_rate": 9.501052631578948e-05,
      "loss": 0.547,
      "step": 38977
    },
    {
      "epoch": 2.182663232164856,
      "grad_norm": 1.126025676727295,
      "learning_rate": 9.501026315789474e-05,
      "loss": 0.3829,
      "step": 38978
    },
    {
      "epoch": 2.182719229476985,
      "grad_norm": 1.2087044715881348,
      "learning_rate": 9.501e-05,
      "loss": 0.3842,
      "step": 38979
    },
    {
      "epoch": 2.182775226789114,
      "grad_norm": 1.2227627038955688,
      "learning_rate": 9.500973684210527e-05,
      "loss": 0.3454,
      "step": 38980
    },
    {
      "epoch": 2.182831224101243,
      "grad_norm": 1.2327896356582642,
      "learning_rate": 9.500947368421052e-05,
      "loss": 0.5793,
      "step": 38981
    },
    {
      "epoch": 2.182887221413372,
      "grad_norm": 1.4385664463043213,
      "learning_rate": 9.500921052631579e-05,
      "loss": 0.4619,
      "step": 38982
    },
    {
      "epoch": 2.182943218725501,
      "grad_norm": 1.091143012046814,
      "learning_rate": 9.500894736842105e-05,
      "loss": 0.3438,
      "step": 38983
    },
    {
      "epoch": 2.18299921603763,
      "grad_norm": 1.3226444721221924,
      "learning_rate": 9.500868421052633e-05,
      "loss": 0.3826,
      "step": 38984
    },
    {
      "epoch": 2.183055213349759,
      "grad_norm": 2.773717164993286,
      "learning_rate": 9.500842105263159e-05,
      "loss": 0.5174,
      "step": 38985
    },
    {
      "epoch": 2.183111210661888,
      "grad_norm": 1.127301812171936,
      "learning_rate": 9.500815789473685e-05,
      "loss": 0.3555,
      "step": 38986
    },
    {
      "epoch": 2.183167207974017,
      "grad_norm": 1.1886684894561768,
      "learning_rate": 9.50078947368421e-05,
      "loss": 0.4,
      "step": 38987
    },
    {
      "epoch": 2.183223205286146,
      "grad_norm": 1.497908592224121,
      "learning_rate": 9.500763157894738e-05,
      "loss": 0.4662,
      "step": 38988
    },
    {
      "epoch": 2.1832792025982752,
      "grad_norm": 0.9962058067321777,
      "learning_rate": 9.500736842105264e-05,
      "loss": 0.3578,
      "step": 38989
    },
    {
      "epoch": 2.1833351999104043,
      "grad_norm": 1.0867133140563965,
      "learning_rate": 9.50071052631579e-05,
      "loss": 0.3901,
      "step": 38990
    },
    {
      "epoch": 2.1833911972225333,
      "grad_norm": 1.0843846797943115,
      "learning_rate": 9.500684210526316e-05,
      "loss": 0.3555,
      "step": 38991
    },
    {
      "epoch": 2.1834471945346623,
      "grad_norm": 1.1588462591171265,
      "learning_rate": 9.500657894736843e-05,
      "loss": 0.3438,
      "step": 38992
    },
    {
      "epoch": 2.1835031918467913,
      "grad_norm": 1.1109051704406738,
      "learning_rate": 9.500631578947369e-05,
      "loss": 0.3453,
      "step": 38993
    },
    {
      "epoch": 2.1835591891589203,
      "grad_norm": 1.0532950162887573,
      "learning_rate": 9.500605263157895e-05,
      "loss": 0.3334,
      "step": 38994
    },
    {
      "epoch": 2.1836151864710494,
      "grad_norm": 1.0967183113098145,
      "learning_rate": 9.500578947368421e-05,
      "loss": 0.4175,
      "step": 38995
    },
    {
      "epoch": 2.1836711837831784,
      "grad_norm": 1.1317375898361206,
      "learning_rate": 9.500552631578948e-05,
      "loss": 0.404,
      "step": 38996
    },
    {
      "epoch": 2.1837271810953074,
      "grad_norm": 1.2214254140853882,
      "learning_rate": 9.500526315789474e-05,
      "loss": 0.3777,
      "step": 38997
    },
    {
      "epoch": 2.1837831784074364,
      "grad_norm": 1.44541335105896,
      "learning_rate": 9.5005e-05,
      "loss": 0.5532,
      "step": 38998
    },
    {
      "epoch": 2.1838391757195654,
      "grad_norm": 1.2402639389038086,
      "learning_rate": 9.500473684210526e-05,
      "loss": 0.3779,
      "step": 38999
    },
    {
      "epoch": 2.1838951730316944,
      "grad_norm": 1.216313123703003,
      "learning_rate": 9.500447368421052e-05,
      "loss": 0.3283,
      "step": 39000
    },
    {
      "epoch": 2.1839511703438235,
      "grad_norm": 1.1168580055236816,
      "learning_rate": 9.50042105263158e-05,
      "loss": 0.3606,
      "step": 39001
    },
    {
      "epoch": 2.1840071676559525,
      "grad_norm": 1.0124750137329102,
      "learning_rate": 9.500394736842106e-05,
      "loss": 0.2679,
      "step": 39002
    },
    {
      "epoch": 2.1840631649680815,
      "grad_norm": 1.1533573865890503,
      "learning_rate": 9.500368421052633e-05,
      "loss": 0.4544,
      "step": 39003
    },
    {
      "epoch": 2.1841191622802105,
      "grad_norm": 1.3082711696624756,
      "learning_rate": 9.500342105263157e-05,
      "loss": 0.4118,
      "step": 39004
    },
    {
      "epoch": 2.1841751595923395,
      "grad_norm": 1.6087048053741455,
      "learning_rate": 9.500315789473685e-05,
      "loss": 0.3556,
      "step": 39005
    },
    {
      "epoch": 2.1842311569044686,
      "grad_norm": 1.2887152433395386,
      "learning_rate": 9.500289473684211e-05,
      "loss": 0.3823,
      "step": 39006
    },
    {
      "epoch": 2.1842871542165976,
      "grad_norm": 1.2029297351837158,
      "learning_rate": 9.500263157894738e-05,
      "loss": 0.4293,
      "step": 39007
    },
    {
      "epoch": 2.1843431515287266,
      "grad_norm": 0.9717657566070557,
      "learning_rate": 9.500236842105264e-05,
      "loss": 0.3112,
      "step": 39008
    },
    {
      "epoch": 2.1843991488408556,
      "grad_norm": 1.5726242065429688,
      "learning_rate": 9.50021052631579e-05,
      "loss": 0.5459,
      "step": 39009
    },
    {
      "epoch": 2.1844551461529846,
      "grad_norm": 1.1057627201080322,
      "learning_rate": 9.500184210526316e-05,
      "loss": 0.4304,
      "step": 39010
    },
    {
      "epoch": 2.1845111434651137,
      "grad_norm": 1.2038341760635376,
      "learning_rate": 9.500157894736843e-05,
      "loss": 0.4084,
      "step": 39011
    },
    {
      "epoch": 2.1845671407772427,
      "grad_norm": 1.1715941429138184,
      "learning_rate": 9.500131578947369e-05,
      "loss": 0.4198,
      "step": 39012
    },
    {
      "epoch": 2.1846231380893717,
      "grad_norm": 1.3885571956634521,
      "learning_rate": 9.500105263157895e-05,
      "loss": 0.4682,
      "step": 39013
    },
    {
      "epoch": 2.1846791354015007,
      "grad_norm": 1.2786542177200317,
      "learning_rate": 9.500078947368421e-05,
      "loss": 0.4038,
      "step": 39014
    },
    {
      "epoch": 2.1847351327136297,
      "grad_norm": 1.22564697265625,
      "learning_rate": 9.500052631578947e-05,
      "loss": 0.3938,
      "step": 39015
    },
    {
      "epoch": 2.1847911300257588,
      "grad_norm": 1.1310079097747803,
      "learning_rate": 9.500026315789475e-05,
      "loss": 0.4345,
      "step": 39016
    },
    {
      "epoch": 2.1848471273378878,
      "grad_norm": 1.3271491527557373,
      "learning_rate": 9.5e-05,
      "loss": 0.4416,
      "step": 39017
    },
    {
      "epoch": 2.184903124650017,
      "grad_norm": 1.1095949411392212,
      "learning_rate": 9.499973684210526e-05,
      "loss": 0.3353,
      "step": 39018
    },
    {
      "epoch": 2.184959121962146,
      "grad_norm": 2.6231560707092285,
      "learning_rate": 9.499947368421052e-05,
      "loss": 0.4896,
      "step": 39019
    },
    {
      "epoch": 2.185015119274275,
      "grad_norm": 1.1061480045318604,
      "learning_rate": 9.49992105263158e-05,
      "loss": 0.5028,
      "step": 39020
    },
    {
      "epoch": 2.185071116586404,
      "grad_norm": 1.3535009622573853,
      "learning_rate": 9.499894736842106e-05,
      "loss": 0.4763,
      "step": 39021
    },
    {
      "epoch": 2.185127113898533,
      "grad_norm": 3.0489447116851807,
      "learning_rate": 9.499868421052632e-05,
      "loss": 0.38,
      "step": 39022
    },
    {
      "epoch": 2.185183111210662,
      "grad_norm": 1.338590383529663,
      "learning_rate": 9.499842105263158e-05,
      "loss": 0.3659,
      "step": 39023
    },
    {
      "epoch": 2.185239108522791,
      "grad_norm": 1.6380319595336914,
      "learning_rate": 9.499815789473685e-05,
      "loss": 0.3535,
      "step": 39024
    },
    {
      "epoch": 2.18529510583492,
      "grad_norm": 1.4939478635787964,
      "learning_rate": 9.499789473684211e-05,
      "loss": 0.3089,
      "step": 39025
    },
    {
      "epoch": 2.185351103147049,
      "grad_norm": 1.0415600538253784,
      "learning_rate": 9.499763157894738e-05,
      "loss": 0.3007,
      "step": 39026
    },
    {
      "epoch": 2.185407100459178,
      "grad_norm": 1.3063600063323975,
      "learning_rate": 9.499736842105263e-05,
      "loss": 0.3285,
      "step": 39027
    },
    {
      "epoch": 2.185463097771307,
      "grad_norm": 1.5945566892623901,
      "learning_rate": 9.49971052631579e-05,
      "loss": 0.5552,
      "step": 39028
    },
    {
      "epoch": 2.185519095083436,
      "grad_norm": 1.6054874658584595,
      "learning_rate": 9.499684210526316e-05,
      "loss": 0.4378,
      "step": 39029
    },
    {
      "epoch": 2.185575092395565,
      "grad_norm": 1.6225864887237549,
      "learning_rate": 9.499657894736844e-05,
      "loss": 0.4031,
      "step": 39030
    },
    {
      "epoch": 2.185631089707694,
      "grad_norm": 1.2703971862792969,
      "learning_rate": 9.499631578947368e-05,
      "loss": 0.413,
      "step": 39031
    },
    {
      "epoch": 2.185687087019823,
      "grad_norm": 0.9448586702346802,
      "learning_rate": 9.499605263157894e-05,
      "loss": 0.3488,
      "step": 39032
    },
    {
      "epoch": 2.185743084331952,
      "grad_norm": 1.1336658000946045,
      "learning_rate": 9.499578947368422e-05,
      "loss": 0.3389,
      "step": 39033
    },
    {
      "epoch": 2.185799081644081,
      "grad_norm": 1.075467586517334,
      "learning_rate": 9.499552631578947e-05,
      "loss": 0.3926,
      "step": 39034
    },
    {
      "epoch": 2.18585507895621,
      "grad_norm": 1.138867974281311,
      "learning_rate": 9.499526315789475e-05,
      "loss": 0.3904,
      "step": 39035
    },
    {
      "epoch": 2.185911076268339,
      "grad_norm": 1.709425687789917,
      "learning_rate": 9.4995e-05,
      "loss": 0.5505,
      "step": 39036
    },
    {
      "epoch": 2.185967073580468,
      "grad_norm": 1.1988377571105957,
      "learning_rate": 9.499473684210527e-05,
      "loss": 0.3754,
      "step": 39037
    },
    {
      "epoch": 2.186023070892597,
      "grad_norm": 1.3486446142196655,
      "learning_rate": 9.499447368421053e-05,
      "loss": 0.4484,
      "step": 39038
    },
    {
      "epoch": 2.186079068204726,
      "grad_norm": 1.1207929849624634,
      "learning_rate": 9.49942105263158e-05,
      "loss": 0.35,
      "step": 39039
    },
    {
      "epoch": 2.186135065516855,
      "grad_norm": 1.3796926736831665,
      "learning_rate": 9.499394736842106e-05,
      "loss": 0.4969,
      "step": 39040
    },
    {
      "epoch": 2.1861910628289842,
      "grad_norm": 1.2409188747406006,
      "learning_rate": 9.499368421052632e-05,
      "loss": 0.5177,
      "step": 39041
    },
    {
      "epoch": 2.1862470601411133,
      "grad_norm": 1.2011913061141968,
      "learning_rate": 9.499342105263158e-05,
      "loss": 0.3566,
      "step": 39042
    },
    {
      "epoch": 2.1863030574532423,
      "grad_norm": 1.0961623191833496,
      "learning_rate": 9.499315789473685e-05,
      "loss": 0.4385,
      "step": 39043
    },
    {
      "epoch": 2.1863590547653713,
      "grad_norm": 1.2823792695999146,
      "learning_rate": 9.499289473684211e-05,
      "loss": 0.3847,
      "step": 39044
    },
    {
      "epoch": 2.1864150520775003,
      "grad_norm": 1.1282930374145508,
      "learning_rate": 9.499263157894737e-05,
      "loss": 0.3827,
      "step": 39045
    },
    {
      "epoch": 2.1864710493896293,
      "grad_norm": 1.1703003644943237,
      "learning_rate": 9.499236842105263e-05,
      "loss": 0.4588,
      "step": 39046
    },
    {
      "epoch": 2.1865270467017583,
      "grad_norm": 1.125441551208496,
      "learning_rate": 9.49921052631579e-05,
      "loss": 0.3284,
      "step": 39047
    },
    {
      "epoch": 2.1865830440138874,
      "grad_norm": 1.1358660459518433,
      "learning_rate": 9.499184210526317e-05,
      "loss": 0.4166,
      "step": 39048
    },
    {
      "epoch": 2.1866390413260164,
      "grad_norm": 1.1321810483932495,
      "learning_rate": 9.499157894736842e-05,
      "loss": 0.3167,
      "step": 39049
    },
    {
      "epoch": 2.1866950386381454,
      "grad_norm": 1.1149832010269165,
      "learning_rate": 9.499131578947368e-05,
      "loss": 0.3864,
      "step": 39050
    },
    {
      "epoch": 2.1867510359502744,
      "grad_norm": 1.2042311429977417,
      "learning_rate": 9.499105263157894e-05,
      "loss": 0.4259,
      "step": 39051
    },
    {
      "epoch": 2.1868070332624034,
      "grad_norm": 1.3041541576385498,
      "learning_rate": 9.499078947368422e-05,
      "loss": 0.3633,
      "step": 39052
    },
    {
      "epoch": 2.1868630305745325,
      "grad_norm": 1.0885785818099976,
      "learning_rate": 9.499052631578948e-05,
      "loss": 0.2723,
      "step": 39053
    },
    {
      "epoch": 2.1869190278866615,
      "grad_norm": 1.215593695640564,
      "learning_rate": 9.499026315789474e-05,
      "loss": 0.3927,
      "step": 39054
    },
    {
      "epoch": 2.1869750251987905,
      "grad_norm": 1.653286099433899,
      "learning_rate": 9.499e-05,
      "loss": 0.5376,
      "step": 39055
    },
    {
      "epoch": 2.1870310225109195,
      "grad_norm": 1.3529754877090454,
      "learning_rate": 9.498973684210527e-05,
      "loss": 0.4994,
      "step": 39056
    },
    {
      "epoch": 2.1870870198230485,
      "grad_norm": 1.3690624237060547,
      "learning_rate": 9.498947368421053e-05,
      "loss": 0.4035,
      "step": 39057
    },
    {
      "epoch": 2.1871430171351776,
      "grad_norm": 1.157879114151001,
      "learning_rate": 9.49892105263158e-05,
      "loss": 0.3911,
      "step": 39058
    },
    {
      "epoch": 2.1871990144473066,
      "grad_norm": 1.4271876811981201,
      "learning_rate": 9.498894736842105e-05,
      "loss": 0.4178,
      "step": 39059
    },
    {
      "epoch": 2.1872550117594356,
      "grad_norm": 1.2080905437469482,
      "learning_rate": 9.498868421052632e-05,
      "loss": 0.356,
      "step": 39060
    },
    {
      "epoch": 2.1873110090715646,
      "grad_norm": 1.517770767211914,
      "learning_rate": 9.498842105263158e-05,
      "loss": 0.5933,
      "step": 39061
    },
    {
      "epoch": 2.1873670063836936,
      "grad_norm": 1.3313344717025757,
      "learning_rate": 9.498815789473686e-05,
      "loss": 0.3613,
      "step": 39062
    },
    {
      "epoch": 2.1874230036958227,
      "grad_norm": 1.301638126373291,
      "learning_rate": 9.498789473684212e-05,
      "loss": 0.5225,
      "step": 39063
    },
    {
      "epoch": 2.1874790010079517,
      "grad_norm": 1.6095978021621704,
      "learning_rate": 9.498763157894738e-05,
      "loss": 0.5183,
      "step": 39064
    },
    {
      "epoch": 2.1875349983200807,
      "grad_norm": 1.1433076858520508,
      "learning_rate": 9.498736842105263e-05,
      "loss": 0.348,
      "step": 39065
    },
    {
      "epoch": 2.1875909956322097,
      "grad_norm": 1.3681985139846802,
      "learning_rate": 9.49871052631579e-05,
      "loss": 0.4488,
      "step": 39066
    },
    {
      "epoch": 2.1876469929443387,
      "grad_norm": 1.0436224937438965,
      "learning_rate": 9.498684210526317e-05,
      "loss": 0.3897,
      "step": 39067
    },
    {
      "epoch": 2.1877029902564677,
      "grad_norm": 1.19947350025177,
      "learning_rate": 9.498657894736843e-05,
      "loss": 0.5267,
      "step": 39068
    },
    {
      "epoch": 2.1877589875685968,
      "grad_norm": 1.2829921245574951,
      "learning_rate": 9.498631578947369e-05,
      "loss": 0.4798,
      "step": 39069
    },
    {
      "epoch": 2.187814984880726,
      "grad_norm": 1.0748436450958252,
      "learning_rate": 9.498605263157895e-05,
      "loss": 0.3202,
      "step": 39070
    },
    {
      "epoch": 2.187870982192855,
      "grad_norm": 1.782092809677124,
      "learning_rate": 9.498578947368422e-05,
      "loss": 0.4477,
      "step": 39071
    },
    {
      "epoch": 2.187926979504984,
      "grad_norm": 1.437738299369812,
      "learning_rate": 9.498552631578948e-05,
      "loss": 0.5512,
      "step": 39072
    },
    {
      "epoch": 2.187982976817113,
      "grad_norm": 1.4653900861740112,
      "learning_rate": 9.498526315789474e-05,
      "loss": 0.5765,
      "step": 39073
    },
    {
      "epoch": 2.188038974129242,
      "grad_norm": 1.160523772239685,
      "learning_rate": 9.4985e-05,
      "loss": 0.4265,
      "step": 39074
    },
    {
      "epoch": 2.188094971441371,
      "grad_norm": 1.1257492303848267,
      "learning_rate": 9.498473684210527e-05,
      "loss": 0.3653,
      "step": 39075
    },
    {
      "epoch": 2.1881509687535,
      "grad_norm": 1.2313061952590942,
      "learning_rate": 9.498447368421053e-05,
      "loss": 0.4876,
      "step": 39076
    },
    {
      "epoch": 2.188206966065629,
      "grad_norm": 1.2742465734481812,
      "learning_rate": 9.498421052631579e-05,
      "loss": 0.4261,
      "step": 39077
    },
    {
      "epoch": 2.188262963377758,
      "grad_norm": 1.4065688848495483,
      "learning_rate": 9.498394736842105e-05,
      "loss": 0.5457,
      "step": 39078
    },
    {
      "epoch": 2.188318960689887,
      "grad_norm": 1.0316917896270752,
      "learning_rate": 9.498368421052633e-05,
      "loss": 0.387,
      "step": 39079
    },
    {
      "epoch": 2.188374958002016,
      "grad_norm": 1.0288453102111816,
      "learning_rate": 9.498342105263158e-05,
      "loss": 0.3717,
      "step": 39080
    },
    {
      "epoch": 2.188430955314145,
      "grad_norm": 1.3102045059204102,
      "learning_rate": 9.498315789473686e-05,
      "loss": 0.4779,
      "step": 39081
    },
    {
      "epoch": 2.188486952626274,
      "grad_norm": 1.369468092918396,
      "learning_rate": 9.49828947368421e-05,
      "loss": 0.45,
      "step": 39082
    },
    {
      "epoch": 2.188542949938403,
      "grad_norm": 1.1277834177017212,
      "learning_rate": 9.498263157894736e-05,
      "loss": 0.2852,
      "step": 39083
    },
    {
      "epoch": 2.188598947250532,
      "grad_norm": 1.4799185991287231,
      "learning_rate": 9.498236842105264e-05,
      "loss": 0.4513,
      "step": 39084
    },
    {
      "epoch": 2.188654944562661,
      "grad_norm": 1.1144607067108154,
      "learning_rate": 9.49821052631579e-05,
      "loss": 0.3621,
      "step": 39085
    },
    {
      "epoch": 2.18871094187479,
      "grad_norm": 1.2686020135879517,
      "learning_rate": 9.498184210526316e-05,
      "loss": 0.4895,
      "step": 39086
    },
    {
      "epoch": 2.188766939186919,
      "grad_norm": 0.9396287202835083,
      "learning_rate": 9.498157894736842e-05,
      "loss": 0.3632,
      "step": 39087
    },
    {
      "epoch": 2.188822936499048,
      "grad_norm": 1.519822120666504,
      "learning_rate": 9.498131578947369e-05,
      "loss": 0.4151,
      "step": 39088
    },
    {
      "epoch": 2.188878933811177,
      "grad_norm": 1.1211975812911987,
      "learning_rate": 9.498105263157895e-05,
      "loss": 0.3343,
      "step": 39089
    },
    {
      "epoch": 2.188934931123306,
      "grad_norm": 1.1856366395950317,
      "learning_rate": 9.498078947368422e-05,
      "loss": 0.3753,
      "step": 39090
    },
    {
      "epoch": 2.188990928435435,
      "grad_norm": 1.2592588663101196,
      "learning_rate": 9.498052631578947e-05,
      "loss": 0.575,
      "step": 39091
    },
    {
      "epoch": 2.189046925747564,
      "grad_norm": 1.1531543731689453,
      "learning_rate": 9.498026315789474e-05,
      "loss": 0.3017,
      "step": 39092
    },
    {
      "epoch": 2.1891029230596932,
      "grad_norm": 1.2448511123657227,
      "learning_rate": 9.498e-05,
      "loss": 0.4248,
      "step": 39093
    },
    {
      "epoch": 2.1891589203718222,
      "grad_norm": 1.5831552743911743,
      "learning_rate": 9.497973684210528e-05,
      "loss": 0.5818,
      "step": 39094
    },
    {
      "epoch": 2.1892149176839513,
      "grad_norm": 1.2447991371154785,
      "learning_rate": 9.497947368421054e-05,
      "loss": 0.3859,
      "step": 39095
    },
    {
      "epoch": 2.1892709149960803,
      "grad_norm": 1.3265899419784546,
      "learning_rate": 9.49792105263158e-05,
      "loss": 0.4058,
      "step": 39096
    },
    {
      "epoch": 2.1893269123082093,
      "grad_norm": 1.16269850730896,
      "learning_rate": 9.497894736842105e-05,
      "loss": 0.4267,
      "step": 39097
    },
    {
      "epoch": 2.1893829096203383,
      "grad_norm": 1.2243558168411255,
      "learning_rate": 9.497868421052633e-05,
      "loss": 0.3839,
      "step": 39098
    },
    {
      "epoch": 2.1894389069324673,
      "grad_norm": 1.1689585447311401,
      "learning_rate": 9.497842105263159e-05,
      "loss": 0.4997,
      "step": 39099
    },
    {
      "epoch": 2.1894949042445964,
      "grad_norm": 1.2064203023910522,
      "learning_rate": 9.497815789473685e-05,
      "loss": 0.3026,
      "step": 39100
    },
    {
      "epoch": 2.1895509015567254,
      "grad_norm": 1.153178334236145,
      "learning_rate": 9.497789473684211e-05,
      "loss": 0.4539,
      "step": 39101
    },
    {
      "epoch": 2.1896068988688544,
      "grad_norm": 1.1924258470535278,
      "learning_rate": 9.497763157894737e-05,
      "loss": 0.4493,
      "step": 39102
    },
    {
      "epoch": 2.1896628961809834,
      "grad_norm": 1.712356448173523,
      "learning_rate": 9.497736842105264e-05,
      "loss": 0.389,
      "step": 39103
    },
    {
      "epoch": 2.1897188934931124,
      "grad_norm": 1.312074065208435,
      "learning_rate": 9.49771052631579e-05,
      "loss": 0.3904,
      "step": 39104
    },
    {
      "epoch": 2.1897748908052415,
      "grad_norm": 1.186565637588501,
      "learning_rate": 9.497684210526316e-05,
      "loss": 0.466,
      "step": 39105
    },
    {
      "epoch": 2.1898308881173705,
      "grad_norm": 1.7509232759475708,
      "learning_rate": 9.497657894736842e-05,
      "loss": 0.5128,
      "step": 39106
    },
    {
      "epoch": 2.1898868854294995,
      "grad_norm": 1.4431231021881104,
      "learning_rate": 9.497631578947369e-05,
      "loss": 0.5513,
      "step": 39107
    },
    {
      "epoch": 2.1899428827416285,
      "grad_norm": 1.2687747478485107,
      "learning_rate": 9.497605263157895e-05,
      "loss": 0.3674,
      "step": 39108
    },
    {
      "epoch": 2.1899988800537575,
      "grad_norm": 1.2940678596496582,
      "learning_rate": 9.497578947368421e-05,
      "loss": 0.4532,
      "step": 39109
    },
    {
      "epoch": 2.1900548773658866,
      "grad_norm": 1.352541208267212,
      "learning_rate": 9.497552631578947e-05,
      "loss": 0.4748,
      "step": 39110
    },
    {
      "epoch": 2.1901108746780156,
      "grad_norm": 1.1367441415786743,
      "learning_rate": 9.497526315789474e-05,
      "loss": 0.3074,
      "step": 39111
    },
    {
      "epoch": 2.1901668719901446,
      "grad_norm": 1.5073899030685425,
      "learning_rate": 9.4975e-05,
      "loss": 0.5365,
      "step": 39112
    },
    {
      "epoch": 2.1902228693022736,
      "grad_norm": 1.1930011510849,
      "learning_rate": 9.497473684210528e-05,
      "loss": 0.3782,
      "step": 39113
    },
    {
      "epoch": 2.1902788666144026,
      "grad_norm": 1.1974886655807495,
      "learning_rate": 9.497447368421052e-05,
      "loss": 0.519,
      "step": 39114
    },
    {
      "epoch": 2.1903348639265316,
      "grad_norm": 1.2613402605056763,
      "learning_rate": 9.49742105263158e-05,
      "loss": 0.3358,
      "step": 39115
    },
    {
      "epoch": 2.1903908612386607,
      "grad_norm": 1.0909866094589233,
      "learning_rate": 9.497394736842106e-05,
      "loss": 0.3243,
      "step": 39116
    },
    {
      "epoch": 2.1904468585507897,
      "grad_norm": 1.2143921852111816,
      "learning_rate": 9.497368421052633e-05,
      "loss": 0.5245,
      "step": 39117
    },
    {
      "epoch": 2.1905028558629187,
      "grad_norm": 1.3832098245620728,
      "learning_rate": 9.497342105263159e-05,
      "loss": 0.4174,
      "step": 39118
    },
    {
      "epoch": 2.1905588531750477,
      "grad_norm": 1.1252905130386353,
      "learning_rate": 9.497315789473684e-05,
      "loss": 0.3326,
      "step": 39119
    },
    {
      "epoch": 2.1906148504871767,
      "grad_norm": 1.20613431930542,
      "learning_rate": 9.497289473684211e-05,
      "loss": 0.4339,
      "step": 39120
    },
    {
      "epoch": 2.1906708477993058,
      "grad_norm": 1.369518756866455,
      "learning_rate": 9.497263157894737e-05,
      "loss": 0.4737,
      "step": 39121
    },
    {
      "epoch": 2.190726845111435,
      "grad_norm": 1.3958098888397217,
      "learning_rate": 9.497236842105264e-05,
      "loss": 0.3378,
      "step": 39122
    },
    {
      "epoch": 2.190782842423564,
      "grad_norm": 1.2123152017593384,
      "learning_rate": 9.49721052631579e-05,
      "loss": 0.3848,
      "step": 39123
    },
    {
      "epoch": 2.190838839735693,
      "grad_norm": 1.1625896692276,
      "learning_rate": 9.497184210526316e-05,
      "loss": 0.405,
      "step": 39124
    },
    {
      "epoch": 2.190894837047822,
      "grad_norm": 1.1821825504302979,
      "learning_rate": 9.497157894736842e-05,
      "loss": 0.4239,
      "step": 39125
    },
    {
      "epoch": 2.190950834359951,
      "grad_norm": 1.431579828262329,
      "learning_rate": 9.49713157894737e-05,
      "loss": 0.3814,
      "step": 39126
    },
    {
      "epoch": 2.19100683167208,
      "grad_norm": 1.3663723468780518,
      "learning_rate": 9.497105263157895e-05,
      "loss": 0.5754,
      "step": 39127
    },
    {
      "epoch": 2.191062828984209,
      "grad_norm": 1.3225445747375488,
      "learning_rate": 9.497078947368421e-05,
      "loss": 0.4511,
      "step": 39128
    },
    {
      "epoch": 2.191118826296338,
      "grad_norm": 1.0950003862380981,
      "learning_rate": 9.497052631578947e-05,
      "loss": 0.3255,
      "step": 39129
    },
    {
      "epoch": 2.191174823608467,
      "grad_norm": 1.2019085884094238,
      "learning_rate": 9.497026315789475e-05,
      "loss": 0.4056,
      "step": 39130
    },
    {
      "epoch": 2.191230820920596,
      "grad_norm": 1.0218220949172974,
      "learning_rate": 9.497000000000001e-05,
      "loss": 0.2836,
      "step": 39131
    },
    {
      "epoch": 2.191286818232725,
      "grad_norm": 1.5577071905136108,
      "learning_rate": 9.496973684210527e-05,
      "loss": 0.3909,
      "step": 39132
    },
    {
      "epoch": 2.191342815544854,
      "grad_norm": 1.2853047847747803,
      "learning_rate": 9.496947368421053e-05,
      "loss": 0.3458,
      "step": 39133
    },
    {
      "epoch": 2.191398812856983,
      "grad_norm": 1.121080756187439,
      "learning_rate": 9.49692105263158e-05,
      "loss": 0.394,
      "step": 39134
    },
    {
      "epoch": 2.191454810169112,
      "grad_norm": 1.3960609436035156,
      "learning_rate": 9.496894736842106e-05,
      "loss": 0.433,
      "step": 39135
    },
    {
      "epoch": 2.191510807481241,
      "grad_norm": 1.69021737575531,
      "learning_rate": 9.496868421052632e-05,
      "loss": 0.3527,
      "step": 39136
    },
    {
      "epoch": 2.19156680479337,
      "grad_norm": 1.6987029314041138,
      "learning_rate": 9.496842105263158e-05,
      "loss": 0.4539,
      "step": 39137
    },
    {
      "epoch": 2.191622802105499,
      "grad_norm": 1.0741510391235352,
      "learning_rate": 9.496815789473684e-05,
      "loss": 0.3644,
      "step": 39138
    },
    {
      "epoch": 2.191678799417628,
      "grad_norm": 1.169575810432434,
      "learning_rate": 9.496789473684211e-05,
      "loss": 0.3835,
      "step": 39139
    },
    {
      "epoch": 2.191734796729757,
      "grad_norm": 1.0368984937667847,
      "learning_rate": 9.496763157894737e-05,
      "loss": 0.3543,
      "step": 39140
    },
    {
      "epoch": 2.191790794041886,
      "grad_norm": 1.200516939163208,
      "learning_rate": 9.496736842105263e-05,
      "loss": 0.3347,
      "step": 39141
    },
    {
      "epoch": 2.191846791354015,
      "grad_norm": 1.1778619289398193,
      "learning_rate": 9.496710526315789e-05,
      "loss": 0.5514,
      "step": 39142
    },
    {
      "epoch": 2.191902788666144,
      "grad_norm": 1.0610679388046265,
      "learning_rate": 9.496684210526316e-05,
      "loss": 0.4898,
      "step": 39143
    },
    {
      "epoch": 2.191958785978273,
      "grad_norm": 1.0308139324188232,
      "learning_rate": 9.496657894736842e-05,
      "loss": 0.3861,
      "step": 39144
    },
    {
      "epoch": 2.1920147832904022,
      "grad_norm": 1.3206017017364502,
      "learning_rate": 9.49663157894737e-05,
      "loss": 0.4364,
      "step": 39145
    },
    {
      "epoch": 2.1920707806025312,
      "grad_norm": 1.1329728364944458,
      "learning_rate": 9.496605263157894e-05,
      "loss": 0.4449,
      "step": 39146
    },
    {
      "epoch": 2.1921267779146603,
      "grad_norm": 1.07108736038208,
      "learning_rate": 9.496578947368422e-05,
      "loss": 0.4644,
      "step": 39147
    },
    {
      "epoch": 2.1921827752267893,
      "grad_norm": 5.464520454406738,
      "learning_rate": 9.496552631578948e-05,
      "loss": 0.3794,
      "step": 39148
    },
    {
      "epoch": 2.1922387725389183,
      "grad_norm": 1.1187776327133179,
      "learning_rate": 9.496526315789475e-05,
      "loss": 0.3259,
      "step": 39149
    },
    {
      "epoch": 2.1922947698510473,
      "grad_norm": 1.1216533184051514,
      "learning_rate": 9.496500000000001e-05,
      "loss": 0.3556,
      "step": 39150
    },
    {
      "epoch": 2.1923507671631763,
      "grad_norm": 1.1532158851623535,
      "learning_rate": 9.496473684210527e-05,
      "loss": 0.4255,
      "step": 39151
    },
    {
      "epoch": 2.1924067644753054,
      "grad_norm": 0.9703332185745239,
      "learning_rate": 9.496447368421053e-05,
      "loss": 0.3977,
      "step": 39152
    },
    {
      "epoch": 2.1924627617874344,
      "grad_norm": 1.062602162361145,
      "learning_rate": 9.496421052631579e-05,
      "loss": 0.3796,
      "step": 39153
    },
    {
      "epoch": 2.1925187590995634,
      "grad_norm": 1.2606024742126465,
      "learning_rate": 9.496394736842106e-05,
      "loss": 0.4109,
      "step": 39154
    },
    {
      "epoch": 2.1925747564116924,
      "grad_norm": 1.25972580909729,
      "learning_rate": 9.496368421052632e-05,
      "loss": 0.4907,
      "step": 39155
    },
    {
      "epoch": 2.1926307537238214,
      "grad_norm": 1.2315218448638916,
      "learning_rate": 9.496342105263158e-05,
      "loss": 0.3272,
      "step": 39156
    },
    {
      "epoch": 2.1926867510359505,
      "grad_norm": 1.318179965019226,
      "learning_rate": 9.496315789473684e-05,
      "loss": 0.3284,
      "step": 39157
    },
    {
      "epoch": 2.1927427483480795,
      "grad_norm": 1.233273983001709,
      "learning_rate": 9.496289473684211e-05,
      "loss": 0.3327,
      "step": 39158
    },
    {
      "epoch": 2.1927987456602085,
      "grad_norm": 1.1989058256149292,
      "learning_rate": 9.496263157894737e-05,
      "loss": 0.3619,
      "step": 39159
    },
    {
      "epoch": 2.1928547429723375,
      "grad_norm": 1.437318205833435,
      "learning_rate": 9.496236842105263e-05,
      "loss": 0.3528,
      "step": 39160
    },
    {
      "epoch": 2.1929107402844665,
      "grad_norm": 1.3835985660552979,
      "learning_rate": 9.49621052631579e-05,
      "loss": 0.5303,
      "step": 39161
    },
    {
      "epoch": 2.1929667375965956,
      "grad_norm": 1.2616150379180908,
      "learning_rate": 9.496184210526317e-05,
      "loss": 0.3978,
      "step": 39162
    },
    {
      "epoch": 2.1930227349087246,
      "grad_norm": 1.1376595497131348,
      "learning_rate": 9.496157894736843e-05,
      "loss": 0.3847,
      "step": 39163
    },
    {
      "epoch": 2.1930787322208536,
      "grad_norm": 1.135696530342102,
      "learning_rate": 9.496131578947369e-05,
      "loss": 0.4026,
      "step": 39164
    },
    {
      "epoch": 2.1931347295329826,
      "grad_norm": 1.1316195726394653,
      "learning_rate": 9.496105263157895e-05,
      "loss": 0.3641,
      "step": 39165
    },
    {
      "epoch": 2.1931907268451116,
      "grad_norm": 1.1594762802124023,
      "learning_rate": 9.496078947368422e-05,
      "loss": 0.5395,
      "step": 39166
    },
    {
      "epoch": 2.1932467241572406,
      "grad_norm": 1.3209244012832642,
      "learning_rate": 9.496052631578948e-05,
      "loss": 0.4982,
      "step": 39167
    },
    {
      "epoch": 2.1933027214693697,
      "grad_norm": 1.1222604513168335,
      "learning_rate": 9.496026315789475e-05,
      "loss": 0.4387,
      "step": 39168
    },
    {
      "epoch": 2.1933587187814987,
      "grad_norm": 1.0970358848571777,
      "learning_rate": 9.496e-05,
      "loss": 0.4285,
      "step": 39169
    },
    {
      "epoch": 2.1934147160936277,
      "grad_norm": 0.9503394961357117,
      "learning_rate": 9.495973684210526e-05,
      "loss": 0.3276,
      "step": 39170
    },
    {
      "epoch": 2.1934707134057567,
      "grad_norm": 1.6275936365127563,
      "learning_rate": 9.495947368421053e-05,
      "loss": 0.401,
      "step": 39171
    },
    {
      "epoch": 2.1935267107178857,
      "grad_norm": 1.111403226852417,
      "learning_rate": 9.495921052631579e-05,
      "loss": 0.3822,
      "step": 39172
    },
    {
      "epoch": 2.1935827080300148,
      "grad_norm": 1.0329989194869995,
      "learning_rate": 9.495894736842106e-05,
      "loss": 0.4374,
      "step": 39173
    },
    {
      "epoch": 2.193638705342144,
      "grad_norm": 1.0841001272201538,
      "learning_rate": 9.495868421052631e-05,
      "loss": 0.4015,
      "step": 39174
    },
    {
      "epoch": 2.193694702654273,
      "grad_norm": 1.3559916019439697,
      "learning_rate": 9.495842105263158e-05,
      "loss": 0.5634,
      "step": 39175
    },
    {
      "epoch": 2.193750699966402,
      "grad_norm": 1.3796883821487427,
      "learning_rate": 9.495815789473684e-05,
      "loss": 0.4402,
      "step": 39176
    },
    {
      "epoch": 2.193806697278531,
      "grad_norm": 1.0877370834350586,
      "learning_rate": 9.495789473684212e-05,
      "loss": 0.3181,
      "step": 39177
    },
    {
      "epoch": 2.19386269459066,
      "grad_norm": 1.078378677368164,
      "learning_rate": 9.495763157894738e-05,
      "loss": 0.3726,
      "step": 39178
    },
    {
      "epoch": 2.1939186919027884,
      "grad_norm": 1.0643137693405151,
      "learning_rate": 9.495736842105264e-05,
      "loss": 0.3799,
      "step": 39179
    },
    {
      "epoch": 2.193974689214918,
      "grad_norm": 1.2806888818740845,
      "learning_rate": 9.49571052631579e-05,
      "loss": 0.39,
      "step": 39180
    },
    {
      "epoch": 2.1940306865270465,
      "grad_norm": 1.2856566905975342,
      "learning_rate": 9.495684210526317e-05,
      "loss": 0.3795,
      "step": 39181
    },
    {
      "epoch": 2.194086683839176,
      "grad_norm": 1.1614986658096313,
      "learning_rate": 9.495657894736843e-05,
      "loss": 0.3582,
      "step": 39182
    },
    {
      "epoch": 2.1941426811513045,
      "grad_norm": 1.077746868133545,
      "learning_rate": 9.495631578947369e-05,
      "loss": 0.3703,
      "step": 39183
    },
    {
      "epoch": 2.194198678463434,
      "grad_norm": 1.4261763095855713,
      "learning_rate": 9.495605263157895e-05,
      "loss": 0.4189,
      "step": 39184
    },
    {
      "epoch": 2.1942546757755625,
      "grad_norm": 1.3532723188400269,
      "learning_rate": 9.495578947368422e-05,
      "loss": 0.4273,
      "step": 39185
    },
    {
      "epoch": 2.194310673087692,
      "grad_norm": 1.1887974739074707,
      "learning_rate": 9.495552631578948e-05,
      "loss": 0.3335,
      "step": 39186
    },
    {
      "epoch": 2.1943666703998206,
      "grad_norm": 1.2685467004776,
      "learning_rate": 9.495526315789474e-05,
      "loss": 0.4662,
      "step": 39187
    },
    {
      "epoch": 2.19442266771195,
      "grad_norm": 0.9770505428314209,
      "learning_rate": 9.4955e-05,
      "loss": 0.3361,
      "step": 39188
    },
    {
      "epoch": 2.1944786650240786,
      "grad_norm": 1.0768907070159912,
      "learning_rate": 9.495473684210526e-05,
      "loss": 0.4428,
      "step": 39189
    },
    {
      "epoch": 2.194534662336208,
      "grad_norm": 1.177499771118164,
      "learning_rate": 9.495447368421053e-05,
      "loss": 0.5098,
      "step": 39190
    },
    {
      "epoch": 2.1945906596483367,
      "grad_norm": 1.2610516548156738,
      "learning_rate": 9.49542105263158e-05,
      "loss": 0.4874,
      "step": 39191
    },
    {
      "epoch": 2.194646656960466,
      "grad_norm": 1.0482573509216309,
      "learning_rate": 9.495394736842105e-05,
      "loss": 0.4516,
      "step": 39192
    },
    {
      "epoch": 2.1947026542725947,
      "grad_norm": 1.1462467908859253,
      "learning_rate": 9.495368421052631e-05,
      "loss": 0.3628,
      "step": 39193
    },
    {
      "epoch": 2.194758651584724,
      "grad_norm": 1.2358667850494385,
      "learning_rate": 9.495342105263159e-05,
      "loss": 0.4096,
      "step": 39194
    },
    {
      "epoch": 2.1948146488968527,
      "grad_norm": 1.7542204856872559,
      "learning_rate": 9.495315789473685e-05,
      "loss": 0.5005,
      "step": 39195
    },
    {
      "epoch": 2.194870646208982,
      "grad_norm": 1.1441692113876343,
      "learning_rate": 9.49528947368421e-05,
      "loss": 0.3844,
      "step": 39196
    },
    {
      "epoch": 2.1949266435211108,
      "grad_norm": 1.5919365882873535,
      "learning_rate": 9.495263157894737e-05,
      "loss": 0.4196,
      "step": 39197
    },
    {
      "epoch": 2.1949826408332402,
      "grad_norm": 1.2509034872055054,
      "learning_rate": 9.495236842105264e-05,
      "loss": 0.4728,
      "step": 39198
    },
    {
      "epoch": 2.195038638145369,
      "grad_norm": 1.468296766281128,
      "learning_rate": 9.49521052631579e-05,
      "loss": 0.5532,
      "step": 39199
    },
    {
      "epoch": 2.195094635457498,
      "grad_norm": 1.3390840291976929,
      "learning_rate": 9.495184210526317e-05,
      "loss": 0.403,
      "step": 39200
    },
    {
      "epoch": 2.195150632769627,
      "grad_norm": 1.035071611404419,
      "learning_rate": 9.495157894736842e-05,
      "loss": 0.4697,
      "step": 39201
    },
    {
      "epoch": 2.195206630081756,
      "grad_norm": 1.354354739189148,
      "learning_rate": 9.495131578947369e-05,
      "loss": 0.4857,
      "step": 39202
    },
    {
      "epoch": 2.195262627393885,
      "grad_norm": 1.4106533527374268,
      "learning_rate": 9.495105263157895e-05,
      "loss": 0.4241,
      "step": 39203
    },
    {
      "epoch": 2.195318624706014,
      "grad_norm": 1.2615022659301758,
      "learning_rate": 9.495078947368421e-05,
      "loss": 0.5512,
      "step": 39204
    },
    {
      "epoch": 2.195374622018143,
      "grad_norm": 1.1615214347839355,
      "learning_rate": 9.495052631578948e-05,
      "loss": 0.2905,
      "step": 39205
    },
    {
      "epoch": 2.195430619330272,
      "grad_norm": 1.1964410543441772,
      "learning_rate": 9.495026315789473e-05,
      "loss": 0.2503,
      "step": 39206
    },
    {
      "epoch": 2.195486616642401,
      "grad_norm": 0.9163525700569153,
      "learning_rate": 9.495e-05,
      "loss": 0.305,
      "step": 39207
    },
    {
      "epoch": 2.19554261395453,
      "grad_norm": 1.2203083038330078,
      "learning_rate": 9.494973684210526e-05,
      "loss": 0.3645,
      "step": 39208
    },
    {
      "epoch": 2.195598611266659,
      "grad_norm": 0.9435203671455383,
      "learning_rate": 9.494947368421054e-05,
      "loss": 0.3135,
      "step": 39209
    },
    {
      "epoch": 2.195654608578788,
      "grad_norm": 1.3386194705963135,
      "learning_rate": 9.49492105263158e-05,
      "loss": 0.5953,
      "step": 39210
    },
    {
      "epoch": 2.195710605890917,
      "grad_norm": 1.1914485692977905,
      "learning_rate": 9.494894736842106e-05,
      "loss": 0.5013,
      "step": 39211
    },
    {
      "epoch": 2.195766603203046,
      "grad_norm": 1.083156943321228,
      "learning_rate": 9.494868421052632e-05,
      "loss": 0.2811,
      "step": 39212
    },
    {
      "epoch": 2.195822600515175,
      "grad_norm": 1.305131435394287,
      "learning_rate": 9.494842105263159e-05,
      "loss": 0.4749,
      "step": 39213
    },
    {
      "epoch": 2.195878597827304,
      "grad_norm": 1.2930506467819214,
      "learning_rate": 9.494815789473685e-05,
      "loss": 0.3796,
      "step": 39214
    },
    {
      "epoch": 2.195934595139433,
      "grad_norm": 1.2020272016525269,
      "learning_rate": 9.494789473684211e-05,
      "loss": 0.4032,
      "step": 39215
    },
    {
      "epoch": 2.195990592451562,
      "grad_norm": 1.2056829929351807,
      "learning_rate": 9.494763157894737e-05,
      "loss": 0.4452,
      "step": 39216
    },
    {
      "epoch": 2.196046589763691,
      "grad_norm": 1.2149534225463867,
      "learning_rate": 9.494736842105264e-05,
      "loss": 0.3319,
      "step": 39217
    },
    {
      "epoch": 2.19610258707582,
      "grad_norm": 1.3055657148361206,
      "learning_rate": 9.49471052631579e-05,
      "loss": 0.5154,
      "step": 39218
    },
    {
      "epoch": 2.196158584387949,
      "grad_norm": 1.0483262538909912,
      "learning_rate": 9.494684210526316e-05,
      "loss": 0.3424,
      "step": 39219
    },
    {
      "epoch": 2.196214581700078,
      "grad_norm": 1.2173939943313599,
      "learning_rate": 9.494657894736842e-05,
      "loss": 0.3585,
      "step": 39220
    },
    {
      "epoch": 2.1962705790122072,
      "grad_norm": 1.94076406955719,
      "learning_rate": 9.494631578947368e-05,
      "loss": 0.4395,
      "step": 39221
    },
    {
      "epoch": 2.1963265763243363,
      "grad_norm": 1.204633355140686,
      "learning_rate": 9.494605263157895e-05,
      "loss": 0.318,
      "step": 39222
    },
    {
      "epoch": 2.1963825736364653,
      "grad_norm": 1.2658069133758545,
      "learning_rate": 9.494578947368421e-05,
      "loss": 0.428,
      "step": 39223
    },
    {
      "epoch": 2.1964385709485943,
      "grad_norm": 0.978947103023529,
      "learning_rate": 9.494552631578947e-05,
      "loss": 0.3905,
      "step": 39224
    },
    {
      "epoch": 2.1964945682607233,
      "grad_norm": 1.4354674816131592,
      "learning_rate": 9.494526315789473e-05,
      "loss": 0.6673,
      "step": 39225
    },
    {
      "epoch": 2.1965505655728523,
      "grad_norm": 1.1528240442276,
      "learning_rate": 9.4945e-05,
      "loss": 0.4348,
      "step": 39226
    },
    {
      "epoch": 2.1966065628849814,
      "grad_norm": 1.365470290184021,
      "learning_rate": 9.494473684210527e-05,
      "loss": 0.4295,
      "step": 39227
    },
    {
      "epoch": 2.1966625601971104,
      "grad_norm": 1.1184728145599365,
      "learning_rate": 9.494447368421054e-05,
      "loss": 0.3364,
      "step": 39228
    },
    {
      "epoch": 2.1967185575092394,
      "grad_norm": 1.0508556365966797,
      "learning_rate": 9.494421052631579e-05,
      "loss": 0.3661,
      "step": 39229
    },
    {
      "epoch": 2.1967745548213684,
      "grad_norm": 1.1564157009124756,
      "learning_rate": 9.494394736842106e-05,
      "loss": 0.3553,
      "step": 39230
    },
    {
      "epoch": 2.1968305521334974,
      "grad_norm": 1.2126152515411377,
      "learning_rate": 9.494368421052632e-05,
      "loss": 0.3258,
      "step": 39231
    },
    {
      "epoch": 2.1968865494456264,
      "grad_norm": 1.1503323316574097,
      "learning_rate": 9.494342105263159e-05,
      "loss": 0.3937,
      "step": 39232
    },
    {
      "epoch": 2.1969425467577555,
      "grad_norm": 1.0257138013839722,
      "learning_rate": 9.494315789473684e-05,
      "loss": 0.5778,
      "step": 39233
    },
    {
      "epoch": 2.1969985440698845,
      "grad_norm": 1.0747004747390747,
      "learning_rate": 9.494289473684211e-05,
      "loss": 0.38,
      "step": 39234
    },
    {
      "epoch": 2.1970545413820135,
      "grad_norm": 1.3167614936828613,
      "learning_rate": 9.494263157894737e-05,
      "loss": 0.3817,
      "step": 39235
    },
    {
      "epoch": 2.1971105386941425,
      "grad_norm": 1.3896008729934692,
      "learning_rate": 9.494236842105264e-05,
      "loss": 0.3847,
      "step": 39236
    },
    {
      "epoch": 2.1971665360062715,
      "grad_norm": 1.2476284503936768,
      "learning_rate": 9.49421052631579e-05,
      "loss": 0.3669,
      "step": 39237
    },
    {
      "epoch": 2.1972225333184006,
      "grad_norm": 1.0637397766113281,
      "learning_rate": 9.494184210526316e-05,
      "loss": 0.3385,
      "step": 39238
    },
    {
      "epoch": 2.1972785306305296,
      "grad_norm": 1.0601918697357178,
      "learning_rate": 9.494157894736842e-05,
      "loss": 0.3928,
      "step": 39239
    },
    {
      "epoch": 2.1973345279426586,
      "grad_norm": 1.411753535270691,
      "learning_rate": 9.494131578947368e-05,
      "loss": 0.6168,
      "step": 39240
    },
    {
      "epoch": 2.1973905252547876,
      "grad_norm": 1.4000637531280518,
      "learning_rate": 9.494105263157896e-05,
      "loss": 0.4251,
      "step": 39241
    },
    {
      "epoch": 2.1974465225669166,
      "grad_norm": 1.333798885345459,
      "learning_rate": 9.494078947368422e-05,
      "loss": 0.4538,
      "step": 39242
    },
    {
      "epoch": 2.1975025198790457,
      "grad_norm": 1.1884304285049438,
      "learning_rate": 9.494052631578948e-05,
      "loss": 0.4104,
      "step": 39243
    },
    {
      "epoch": 2.1975585171911747,
      "grad_norm": 1.0855166912078857,
      "learning_rate": 9.494026315789474e-05,
      "loss": 0.3587,
      "step": 39244
    },
    {
      "epoch": 2.1976145145033037,
      "grad_norm": 1.1423125267028809,
      "learning_rate": 9.494000000000001e-05,
      "loss": 0.3943,
      "step": 39245
    },
    {
      "epoch": 2.1976705118154327,
      "grad_norm": 1.1682051420211792,
      "learning_rate": 9.493973684210527e-05,
      "loss": 0.4156,
      "step": 39246
    },
    {
      "epoch": 2.1977265091275617,
      "grad_norm": 1.141829252243042,
      "learning_rate": 9.493947368421053e-05,
      "loss": 0.3297,
      "step": 39247
    },
    {
      "epoch": 2.1977825064396908,
      "grad_norm": 1.6670130491256714,
      "learning_rate": 9.493921052631579e-05,
      "loss": 0.3846,
      "step": 39248
    },
    {
      "epoch": 2.1978385037518198,
      "grad_norm": 2.317133903503418,
      "learning_rate": 9.493894736842106e-05,
      "loss": 0.363,
      "step": 39249
    },
    {
      "epoch": 2.197894501063949,
      "grad_norm": Infinity,
      "learning_rate": 9.493894736842106e-05,
      "loss": 0.4071,
      "step": 39250
    },
    {
      "epoch": 2.197950498376078,
      "grad_norm": 1.3584693670272827,
      "learning_rate": 9.493868421052632e-05,
      "loss": 0.4368,
      "step": 39251
    },
    {
      "epoch": 2.198006495688207,
      "grad_norm": 1.3900644779205322,
      "learning_rate": 9.493842105263158e-05,
      "loss": 0.4187,
      "step": 39252
    },
    {
      "epoch": 2.198062493000336,
      "grad_norm": 1.392337441444397,
      "learning_rate": 9.493815789473684e-05,
      "loss": 0.3972,
      "step": 39253
    },
    {
      "epoch": 2.198118490312465,
      "grad_norm": 1.0016899108886719,
      "learning_rate": 9.493789473684211e-05,
      "loss": 0.3765,
      "step": 39254
    },
    {
      "epoch": 2.198174487624594,
      "grad_norm": 1.286842942237854,
      "learning_rate": 9.493763157894737e-05,
      "loss": 0.4873,
      "step": 39255
    },
    {
      "epoch": 2.198230484936723,
      "grad_norm": 1.3937572240829468,
      "learning_rate": 9.493736842105265e-05,
      "loss": 0.4057,
      "step": 39256
    },
    {
      "epoch": 2.198286482248852,
      "grad_norm": 1.6088981628417969,
      "learning_rate": 9.49371052631579e-05,
      "loss": 0.6128,
      "step": 39257
    },
    {
      "epoch": 2.198342479560981,
      "grad_norm": 1.049758791923523,
      "learning_rate": 9.493684210526315e-05,
      "loss": 0.4872,
      "step": 39258
    },
    {
      "epoch": 2.19839847687311,
      "grad_norm": 2.756300687789917,
      "learning_rate": 9.493657894736843e-05,
      "loss": 0.3316,
      "step": 39259
    },
    {
      "epoch": 2.198454474185239,
      "grad_norm": 0.9867526888847351,
      "learning_rate": 9.493631578947369e-05,
      "loss": 0.3947,
      "step": 39260
    },
    {
      "epoch": 2.198510471497368,
      "grad_norm": 1.358424186706543,
      "learning_rate": 9.493605263157896e-05,
      "loss": 0.3495,
      "step": 39261
    },
    {
      "epoch": 2.198566468809497,
      "grad_norm": 1.1801587343215942,
      "learning_rate": 9.49357894736842e-05,
      "loss": 0.3617,
      "step": 39262
    },
    {
      "epoch": 2.198622466121626,
      "grad_norm": 1.095976710319519,
      "learning_rate": 9.493552631578948e-05,
      "loss": 0.4444,
      "step": 39263
    },
    {
      "epoch": 2.198678463433755,
      "grad_norm": 1.1274425983428955,
      "learning_rate": 9.493526315789474e-05,
      "loss": 0.4946,
      "step": 39264
    },
    {
      "epoch": 2.198734460745884,
      "grad_norm": 1.2692400217056274,
      "learning_rate": 9.493500000000001e-05,
      "loss": 0.3954,
      "step": 39265
    },
    {
      "epoch": 2.198790458058013,
      "grad_norm": 1.3247352838516235,
      "learning_rate": 9.493473684210527e-05,
      "loss": 0.3894,
      "step": 39266
    },
    {
      "epoch": 2.198846455370142,
      "grad_norm": 1.5664595365524292,
      "learning_rate": 9.493447368421053e-05,
      "loss": 0.4534,
      "step": 39267
    },
    {
      "epoch": 2.198902452682271,
      "grad_norm": 1.2559038400650024,
      "learning_rate": 9.493421052631579e-05,
      "loss": 0.3604,
      "step": 39268
    },
    {
      "epoch": 2.1989584499944,
      "grad_norm": 1.5000864267349243,
      "learning_rate": 9.493394736842106e-05,
      "loss": 0.4337,
      "step": 39269
    },
    {
      "epoch": 2.199014447306529,
      "grad_norm": 1.262047529220581,
      "learning_rate": 9.493368421052632e-05,
      "loss": 0.3786,
      "step": 39270
    },
    {
      "epoch": 2.199070444618658,
      "grad_norm": 1.5206862688064575,
      "learning_rate": 9.493342105263158e-05,
      "loss": 0.4685,
      "step": 39271
    },
    {
      "epoch": 2.199126441930787,
      "grad_norm": 1.3647706508636475,
      "learning_rate": 9.493315789473684e-05,
      "loss": 0.4278,
      "step": 39272
    },
    {
      "epoch": 2.1991824392429162,
      "grad_norm": 1.1124950647354126,
      "learning_rate": 9.493289473684212e-05,
      "loss": 0.3485,
      "step": 39273
    },
    {
      "epoch": 2.1992384365550453,
      "grad_norm": 1.4288742542266846,
      "learning_rate": 9.493263157894738e-05,
      "loss": 0.4678,
      "step": 39274
    },
    {
      "epoch": 2.1992944338671743,
      "grad_norm": 1.0784953832626343,
      "learning_rate": 9.493236842105264e-05,
      "loss": 0.2871,
      "step": 39275
    },
    {
      "epoch": 2.1993504311793033,
      "grad_norm": 1.1378778219223022,
      "learning_rate": 9.49321052631579e-05,
      "loss": 0.4651,
      "step": 39276
    },
    {
      "epoch": 2.1994064284914323,
      "grad_norm": 1.417097806930542,
      "learning_rate": 9.493184210526316e-05,
      "loss": 0.4562,
      "step": 39277
    },
    {
      "epoch": 2.1994624258035613,
      "grad_norm": 1.0440655946731567,
      "learning_rate": 9.493157894736843e-05,
      "loss": 0.3712,
      "step": 39278
    },
    {
      "epoch": 2.1995184231156903,
      "grad_norm": 1.160477638244629,
      "learning_rate": 9.493131578947369e-05,
      "loss": 0.3591,
      "step": 39279
    },
    {
      "epoch": 2.1995744204278194,
      "grad_norm": 1.1765644550323486,
      "learning_rate": 9.493105263157895e-05,
      "loss": 0.4118,
      "step": 39280
    },
    {
      "epoch": 2.1996304177399484,
      "grad_norm": 1.2263824939727783,
      "learning_rate": 9.493078947368421e-05,
      "loss": 0.4364,
      "step": 39281
    },
    {
      "epoch": 2.1996864150520774,
      "grad_norm": 1.0986815690994263,
      "learning_rate": 9.493052631578948e-05,
      "loss": 0.4,
      "step": 39282
    },
    {
      "epoch": 2.1997424123642064,
      "grad_norm": 1.3523582220077515,
      "learning_rate": 9.493026315789474e-05,
      "loss": 0.4107,
      "step": 39283
    },
    {
      "epoch": 2.1997984096763354,
      "grad_norm": 1.6746745109558105,
      "learning_rate": 9.493000000000001e-05,
      "loss": 0.5522,
      "step": 39284
    },
    {
      "epoch": 2.1998544069884645,
      "grad_norm": 1.158724308013916,
      "learning_rate": 9.492973684210526e-05,
      "loss": 0.3857,
      "step": 39285
    },
    {
      "epoch": 2.1999104043005935,
      "grad_norm": 1.1654574871063232,
      "learning_rate": 9.492947368421053e-05,
      "loss": 0.46,
      "step": 39286
    },
    {
      "epoch": 2.1999664016127225,
      "grad_norm": 1.2558428049087524,
      "learning_rate": 9.49292105263158e-05,
      "loss": 0.4747,
      "step": 39287
    },
    {
      "epoch": 2.2000223989248515,
      "grad_norm": 1.0895527601242065,
      "learning_rate": 9.492894736842107e-05,
      "loss": 0.303,
      "step": 39288
    },
    {
      "epoch": 2.2000783962369805,
      "grad_norm": 1.1655685901641846,
      "learning_rate": 9.492868421052631e-05,
      "loss": 0.3701,
      "step": 39289
    },
    {
      "epoch": 2.2001343935491096,
      "grad_norm": 1.3279938697814941,
      "learning_rate": 9.492842105263159e-05,
      "loss": 0.3684,
      "step": 39290
    },
    {
      "epoch": 2.2001903908612386,
      "grad_norm": 1.2167469263076782,
      "learning_rate": 9.492815789473685e-05,
      "loss": 0.5036,
      "step": 39291
    },
    {
      "epoch": 2.2002463881733676,
      "grad_norm": 1.102217435836792,
      "learning_rate": 9.49278947368421e-05,
      "loss": 0.4345,
      "step": 39292
    },
    {
      "epoch": 2.2003023854854966,
      "grad_norm": 1.2142664194107056,
      "learning_rate": 9.492763157894738e-05,
      "loss": 0.3749,
      "step": 39293
    },
    {
      "epoch": 2.2003583827976256,
      "grad_norm": 1.1024116277694702,
      "learning_rate": 9.492736842105263e-05,
      "loss": 0.2944,
      "step": 39294
    },
    {
      "epoch": 2.2004143801097547,
      "grad_norm": 1.0679608583450317,
      "learning_rate": 9.49271052631579e-05,
      "loss": 0.309,
      "step": 39295
    },
    {
      "epoch": 2.2004703774218837,
      "grad_norm": 1.1432099342346191,
      "learning_rate": 9.492684210526316e-05,
      "loss": 0.47,
      "step": 39296
    },
    {
      "epoch": 2.2005263747340127,
      "grad_norm": 1.4059308767318726,
      "learning_rate": 9.492657894736843e-05,
      "loss": 0.4945,
      "step": 39297
    },
    {
      "epoch": 2.2005823720461417,
      "grad_norm": 1.4727104902267456,
      "learning_rate": 9.492631578947369e-05,
      "loss": 0.498,
      "step": 39298
    },
    {
      "epoch": 2.2006383693582707,
      "grad_norm": 1.1360934972763062,
      "learning_rate": 9.492605263157895e-05,
      "loss": 0.43,
      "step": 39299
    },
    {
      "epoch": 2.2006943666703997,
      "grad_norm": 1.139660120010376,
      "learning_rate": 9.492578947368421e-05,
      "loss": 0.4029,
      "step": 39300
    },
    {
      "epoch": 2.2007503639825288,
      "grad_norm": 1.252203106880188,
      "learning_rate": 9.492552631578948e-05,
      "loss": 0.3405,
      "step": 39301
    },
    {
      "epoch": 2.200806361294658,
      "grad_norm": 1.2618763446807861,
      "learning_rate": 9.492526315789474e-05,
      "loss": 0.4606,
      "step": 39302
    },
    {
      "epoch": 2.200862358606787,
      "grad_norm": 1.0292749404907227,
      "learning_rate": 9.4925e-05,
      "loss": 0.4432,
      "step": 39303
    },
    {
      "epoch": 2.200918355918916,
      "grad_norm": 1.155129313468933,
      "learning_rate": 9.492473684210526e-05,
      "loss": 0.3724,
      "step": 39304
    },
    {
      "epoch": 2.200974353231045,
      "grad_norm": 1.9602291584014893,
      "learning_rate": 9.492447368421054e-05,
      "loss": 0.6007,
      "step": 39305
    },
    {
      "epoch": 2.201030350543174,
      "grad_norm": 1.5157901048660278,
      "learning_rate": 9.49242105263158e-05,
      "loss": 0.5295,
      "step": 39306
    },
    {
      "epoch": 2.201086347855303,
      "grad_norm": 1.308485746383667,
      "learning_rate": 9.492394736842106e-05,
      "loss": 0.2816,
      "step": 39307
    },
    {
      "epoch": 2.201142345167432,
      "grad_norm": 1.3615050315856934,
      "learning_rate": 9.492368421052632e-05,
      "loss": 0.4409,
      "step": 39308
    },
    {
      "epoch": 2.201198342479561,
      "grad_norm": 1.2001307010650635,
      "learning_rate": 9.492342105263158e-05,
      "loss": 0.3931,
      "step": 39309
    },
    {
      "epoch": 2.20125433979169,
      "grad_norm": 1.1810969114303589,
      "learning_rate": 9.492315789473685e-05,
      "loss": 0.3792,
      "step": 39310
    },
    {
      "epoch": 2.201310337103819,
      "grad_norm": 1.0671933889389038,
      "learning_rate": 9.492289473684211e-05,
      "loss": 0.3758,
      "step": 39311
    },
    {
      "epoch": 2.201366334415948,
      "grad_norm": 1.090793490409851,
      "learning_rate": 9.492263157894737e-05,
      "loss": 0.3441,
      "step": 39312
    },
    {
      "epoch": 2.201422331728077,
      "grad_norm": 1.2792291641235352,
      "learning_rate": 9.492236842105263e-05,
      "loss": 0.3195,
      "step": 39313
    },
    {
      "epoch": 2.201478329040206,
      "grad_norm": 1.1196097135543823,
      "learning_rate": 9.49221052631579e-05,
      "loss": 0.3854,
      "step": 39314
    },
    {
      "epoch": 2.201534326352335,
      "grad_norm": 1.177736759185791,
      "learning_rate": 9.492184210526316e-05,
      "loss": 0.3877,
      "step": 39315
    },
    {
      "epoch": 2.201590323664464,
      "grad_norm": 1.6207741498947144,
      "learning_rate": 9.492157894736843e-05,
      "loss": 0.4691,
      "step": 39316
    },
    {
      "epoch": 2.201646320976593,
      "grad_norm": 1.186977744102478,
      "learning_rate": 9.492131578947368e-05,
      "loss": 0.3499,
      "step": 39317
    },
    {
      "epoch": 2.201702318288722,
      "grad_norm": 1.4378749132156372,
      "learning_rate": 9.492105263157895e-05,
      "loss": 0.4954,
      "step": 39318
    },
    {
      "epoch": 2.201758315600851,
      "grad_norm": 1.0880298614501953,
      "learning_rate": 9.492078947368421e-05,
      "loss": 0.3905,
      "step": 39319
    },
    {
      "epoch": 2.20181431291298,
      "grad_norm": 1.228933334350586,
      "learning_rate": 9.492052631578949e-05,
      "loss": 0.4962,
      "step": 39320
    },
    {
      "epoch": 2.201870310225109,
      "grad_norm": 1.3729842901229858,
      "learning_rate": 9.492026315789475e-05,
      "loss": 0.3732,
      "step": 39321
    },
    {
      "epoch": 2.201926307537238,
      "grad_norm": 1.1341028213500977,
      "learning_rate": 9.492e-05,
      "loss": 0.2977,
      "step": 39322
    },
    {
      "epoch": 2.201982304849367,
      "grad_norm": 1.1909477710723877,
      "learning_rate": 9.491973684210527e-05,
      "loss": 0.4542,
      "step": 39323
    },
    {
      "epoch": 2.202038302161496,
      "grad_norm": 1.1851598024368286,
      "learning_rate": 9.491947368421054e-05,
      "loss": 0.3342,
      "step": 39324
    },
    {
      "epoch": 2.2020942994736252,
      "grad_norm": 1.3212474584579468,
      "learning_rate": 9.49192105263158e-05,
      "loss": 0.4274,
      "step": 39325
    },
    {
      "epoch": 2.2021502967857542,
      "grad_norm": 1.1927757263183594,
      "learning_rate": 9.491894736842106e-05,
      "loss": 0.3609,
      "step": 39326
    },
    {
      "epoch": 2.2022062940978833,
      "grad_norm": 1.2361246347427368,
      "learning_rate": 9.491868421052632e-05,
      "loss": 0.3993,
      "step": 39327
    },
    {
      "epoch": 2.2022622914100123,
      "grad_norm": 1.221638798713684,
      "learning_rate": 9.491842105263158e-05,
      "loss": 0.2948,
      "step": 39328
    },
    {
      "epoch": 2.2023182887221413,
      "grad_norm": 2.991151809692383,
      "learning_rate": 9.491815789473685e-05,
      "loss": 0.3586,
      "step": 39329
    },
    {
      "epoch": 2.2023742860342703,
      "grad_norm": 1.0577765703201294,
      "learning_rate": 9.491789473684211e-05,
      "loss": 0.357,
      "step": 39330
    },
    {
      "epoch": 2.2024302833463993,
      "grad_norm": 1.4361063241958618,
      "learning_rate": 9.491763157894737e-05,
      "loss": 0.5449,
      "step": 39331
    },
    {
      "epoch": 2.2024862806585284,
      "grad_norm": 1.175093412399292,
      "learning_rate": 9.491736842105263e-05,
      "loss": 0.4037,
      "step": 39332
    },
    {
      "epoch": 2.2025422779706574,
      "grad_norm": 1.527953028678894,
      "learning_rate": 9.49171052631579e-05,
      "loss": 0.4733,
      "step": 39333
    },
    {
      "epoch": 2.2025982752827864,
      "grad_norm": 1.089242696762085,
      "learning_rate": 9.491684210526316e-05,
      "loss": 0.3371,
      "step": 39334
    },
    {
      "epoch": 2.2026542725949154,
      "grad_norm": 1.4622660875320435,
      "learning_rate": 9.491657894736842e-05,
      "loss": 0.5044,
      "step": 39335
    },
    {
      "epoch": 2.2027102699070444,
      "grad_norm": 1.4181054830551147,
      "learning_rate": 9.491631578947368e-05,
      "loss": 0.3694,
      "step": 39336
    },
    {
      "epoch": 2.2027662672191735,
      "grad_norm": 1.3265615701675415,
      "learning_rate": 9.491605263157896e-05,
      "loss": 0.4154,
      "step": 39337
    },
    {
      "epoch": 2.2028222645313025,
      "grad_norm": 1.2674179077148438,
      "learning_rate": 9.491578947368422e-05,
      "loss": 0.3843,
      "step": 39338
    },
    {
      "epoch": 2.2028782618434315,
      "grad_norm": 1.1680123805999756,
      "learning_rate": 9.491552631578949e-05,
      "loss": 0.4554,
      "step": 39339
    },
    {
      "epoch": 2.2029342591555605,
      "grad_norm": 1.3256821632385254,
      "learning_rate": 9.491526315789474e-05,
      "loss": 0.4979,
      "step": 39340
    },
    {
      "epoch": 2.2029902564676895,
      "grad_norm": 1.8294093608856201,
      "learning_rate": 9.491500000000001e-05,
      "loss": 0.4146,
      "step": 39341
    },
    {
      "epoch": 2.2030462537798186,
      "grad_norm": 1.1637344360351562,
      "learning_rate": 9.491473684210527e-05,
      "loss": 0.5407,
      "step": 39342
    },
    {
      "epoch": 2.2031022510919476,
      "grad_norm": 1.3018330335617065,
      "learning_rate": 9.491447368421053e-05,
      "loss": 0.3945,
      "step": 39343
    },
    {
      "epoch": 2.2031582484040766,
      "grad_norm": 1.7104288339614868,
      "learning_rate": 9.491421052631579e-05,
      "loss": 0.4961,
      "step": 39344
    },
    {
      "epoch": 2.2032142457162056,
      "grad_norm": 1.2187089920043945,
      "learning_rate": 9.491394736842105e-05,
      "loss": 0.391,
      "step": 39345
    },
    {
      "epoch": 2.2032702430283346,
      "grad_norm": 1.0079604387283325,
      "learning_rate": 9.491368421052632e-05,
      "loss": 0.2951,
      "step": 39346
    },
    {
      "epoch": 2.2033262403404636,
      "grad_norm": 0.9097388386726379,
      "learning_rate": 9.491342105263158e-05,
      "loss": 0.3989,
      "step": 39347
    },
    {
      "epoch": 2.2033822376525927,
      "grad_norm": 1.2129552364349365,
      "learning_rate": 9.491315789473685e-05,
      "loss": 0.4267,
      "step": 39348
    },
    {
      "epoch": 2.2034382349647217,
      "grad_norm": 1.247456669807434,
      "learning_rate": 9.49128947368421e-05,
      "loss": 0.3425,
      "step": 39349
    },
    {
      "epoch": 2.2034942322768507,
      "grad_norm": 1.3718011379241943,
      "learning_rate": 9.491263157894737e-05,
      "loss": 0.3578,
      "step": 39350
    },
    {
      "epoch": 2.2035502295889797,
      "grad_norm": 1.1155140399932861,
      "learning_rate": 9.491236842105263e-05,
      "loss": 0.346,
      "step": 39351
    },
    {
      "epoch": 2.2036062269011087,
      "grad_norm": 1.2574106454849243,
      "learning_rate": 9.49121052631579e-05,
      "loss": 0.4189,
      "step": 39352
    },
    {
      "epoch": 2.2036622242132378,
      "grad_norm": 1.05961012840271,
      "learning_rate": 9.491184210526317e-05,
      "loss": 0.3504,
      "step": 39353
    },
    {
      "epoch": 2.203718221525367,
      "grad_norm": 1.1112011671066284,
      "learning_rate": 9.491157894736843e-05,
      "loss": 0.3679,
      "step": 39354
    },
    {
      "epoch": 2.203774218837496,
      "grad_norm": 1.2634233236312866,
      "learning_rate": 9.491131578947369e-05,
      "loss": 0.5298,
      "step": 39355
    },
    {
      "epoch": 2.203830216149625,
      "grad_norm": 1.2223896980285645,
      "learning_rate": 9.491105263157896e-05,
      "loss": 0.493,
      "step": 39356
    },
    {
      "epoch": 2.203886213461754,
      "grad_norm": 1.292396068572998,
      "learning_rate": 9.491078947368422e-05,
      "loss": 0.4107,
      "step": 39357
    },
    {
      "epoch": 2.203942210773883,
      "grad_norm": 1.210006833076477,
      "learning_rate": 9.491052631578948e-05,
      "loss": 0.3662,
      "step": 39358
    },
    {
      "epoch": 2.203998208086012,
      "grad_norm": 1.2033412456512451,
      "learning_rate": 9.491026315789474e-05,
      "loss": 0.3618,
      "step": 39359
    },
    {
      "epoch": 2.204054205398141,
      "grad_norm": 1.2836880683898926,
      "learning_rate": 9.491000000000001e-05,
      "loss": 0.2981,
      "step": 39360
    },
    {
      "epoch": 2.20411020271027,
      "grad_norm": 0.9685962796211243,
      "learning_rate": 9.490973684210527e-05,
      "loss": 0.2838,
      "step": 39361
    },
    {
      "epoch": 2.204166200022399,
      "grad_norm": 1.1783638000488281,
      "learning_rate": 9.490947368421053e-05,
      "loss": 0.4193,
      "step": 39362
    },
    {
      "epoch": 2.204222197334528,
      "grad_norm": 1.2935866117477417,
      "learning_rate": 9.490921052631579e-05,
      "loss": 0.3465,
      "step": 39363
    },
    {
      "epoch": 2.204278194646657,
      "grad_norm": 1.152005910873413,
      "learning_rate": 9.490894736842105e-05,
      "loss": 0.403,
      "step": 39364
    },
    {
      "epoch": 2.204334191958786,
      "grad_norm": 1.0610591173171997,
      "learning_rate": 9.490868421052632e-05,
      "loss": 0.3927,
      "step": 39365
    },
    {
      "epoch": 2.204390189270915,
      "grad_norm": 1.2709193229675293,
      "learning_rate": 9.490842105263158e-05,
      "loss": 0.3874,
      "step": 39366
    },
    {
      "epoch": 2.204446186583044,
      "grad_norm": 1.2130619287490845,
      "learning_rate": 9.490815789473684e-05,
      "loss": 0.482,
      "step": 39367
    },
    {
      "epoch": 2.204502183895173,
      "grad_norm": 1.4525854587554932,
      "learning_rate": 9.49078947368421e-05,
      "loss": 0.4421,
      "step": 39368
    },
    {
      "epoch": 2.204558181207302,
      "grad_norm": 1.3734872341156006,
      "learning_rate": 9.490763157894738e-05,
      "loss": 0.4058,
      "step": 39369
    },
    {
      "epoch": 2.204614178519431,
      "grad_norm": 1.0020607709884644,
      "learning_rate": 9.490736842105264e-05,
      "loss": 0.3527,
      "step": 39370
    },
    {
      "epoch": 2.20467017583156,
      "grad_norm": 1.1850841045379639,
      "learning_rate": 9.490710526315791e-05,
      "loss": 0.4669,
      "step": 39371
    },
    {
      "epoch": 2.204726173143689,
      "grad_norm": 1.3022788763046265,
      "learning_rate": 9.490684210526316e-05,
      "loss": 0.4011,
      "step": 39372
    },
    {
      "epoch": 2.204782170455818,
      "grad_norm": 1.42355215549469,
      "learning_rate": 9.490657894736843e-05,
      "loss": 0.5983,
      "step": 39373
    },
    {
      "epoch": 2.204838167767947,
      "grad_norm": 1.6269807815551758,
      "learning_rate": 9.490631578947369e-05,
      "loss": 0.4742,
      "step": 39374
    },
    {
      "epoch": 2.204894165080076,
      "grad_norm": 1.2100167274475098,
      "learning_rate": 9.490605263157896e-05,
      "loss": 0.3567,
      "step": 39375
    },
    {
      "epoch": 2.204950162392205,
      "grad_norm": 1.0779049396514893,
      "learning_rate": 9.490578947368422e-05,
      "loss": 0.3919,
      "step": 39376
    },
    {
      "epoch": 2.2050061597043342,
      "grad_norm": 1.4696465730667114,
      "learning_rate": 9.490552631578948e-05,
      "loss": 0.4987,
      "step": 39377
    },
    {
      "epoch": 2.2050621570164632,
      "grad_norm": 1.1518709659576416,
      "learning_rate": 9.490526315789474e-05,
      "loss": 0.4005,
      "step": 39378
    },
    {
      "epoch": 2.2051181543285923,
      "grad_norm": 1.4353548288345337,
      "learning_rate": 9.4905e-05,
      "loss": 0.4269,
      "step": 39379
    },
    {
      "epoch": 2.2051741516407213,
      "grad_norm": 1.6050125360488892,
      "learning_rate": 9.490473684210527e-05,
      "loss": 0.3598,
      "step": 39380
    },
    {
      "epoch": 2.2052301489528503,
      "grad_norm": 1.313718557357788,
      "learning_rate": 9.490447368421052e-05,
      "loss": 0.4787,
      "step": 39381
    },
    {
      "epoch": 2.2052861462649793,
      "grad_norm": 1.4907491207122803,
      "learning_rate": 9.490421052631579e-05,
      "loss": 0.4583,
      "step": 39382
    },
    {
      "epoch": 2.2053421435771083,
      "grad_norm": 1.0511032342910767,
      "learning_rate": 9.490394736842105e-05,
      "loss": 0.428,
      "step": 39383
    },
    {
      "epoch": 2.2053981408892374,
      "grad_norm": 1.29319429397583,
      "learning_rate": 9.490368421052633e-05,
      "loss": 0.4183,
      "step": 39384
    },
    {
      "epoch": 2.2054541382013664,
      "grad_norm": 1.3380197286605835,
      "learning_rate": 9.490342105263159e-05,
      "loss": 0.4372,
      "step": 39385
    },
    {
      "epoch": 2.2055101355134954,
      "grad_norm": 1.205701231956482,
      "learning_rate": 9.490315789473685e-05,
      "loss": 0.6668,
      "step": 39386
    },
    {
      "epoch": 2.2055661328256244,
      "grad_norm": 1.1658575534820557,
      "learning_rate": 9.49028947368421e-05,
      "loss": 0.3206,
      "step": 39387
    },
    {
      "epoch": 2.2056221301377534,
      "grad_norm": 1.552689552307129,
      "learning_rate": 9.490263157894738e-05,
      "loss": 0.5339,
      "step": 39388
    },
    {
      "epoch": 2.2056781274498825,
      "grad_norm": 1.019373893737793,
      "learning_rate": 9.490236842105264e-05,
      "loss": 0.2995,
      "step": 39389
    },
    {
      "epoch": 2.2057341247620115,
      "grad_norm": 1.1684759855270386,
      "learning_rate": 9.49021052631579e-05,
      "loss": 0.409,
      "step": 39390
    },
    {
      "epoch": 2.2057901220741405,
      "grad_norm": 1.038171410560608,
      "learning_rate": 9.490184210526316e-05,
      "loss": 0.3349,
      "step": 39391
    },
    {
      "epoch": 2.2058461193862695,
      "grad_norm": 1.7563350200653076,
      "learning_rate": 9.490157894736843e-05,
      "loss": 0.5792,
      "step": 39392
    },
    {
      "epoch": 2.2059021166983985,
      "grad_norm": 1.1089731454849243,
      "learning_rate": 9.490131578947369e-05,
      "loss": 0.3798,
      "step": 39393
    },
    {
      "epoch": 2.2059581140105275,
      "grad_norm": 1.2701400518417358,
      "learning_rate": 9.490105263157896e-05,
      "loss": 0.4079,
      "step": 39394
    },
    {
      "epoch": 2.2060141113226566,
      "grad_norm": 1.291379451751709,
      "learning_rate": 9.490078947368421e-05,
      "loss": 0.4018,
      "step": 39395
    },
    {
      "epoch": 2.2060701086347856,
      "grad_norm": 1.1315112113952637,
      "learning_rate": 9.490052631578947e-05,
      "loss": 0.3452,
      "step": 39396
    },
    {
      "epoch": 2.2061261059469146,
      "grad_norm": 1.3090662956237793,
      "learning_rate": 9.490026315789474e-05,
      "loss": 0.3291,
      "step": 39397
    },
    {
      "epoch": 2.2061821032590436,
      "grad_norm": 1.2487105131149292,
      "learning_rate": 9.49e-05,
      "loss": 0.3743,
      "step": 39398
    },
    {
      "epoch": 2.2062381005711726,
      "grad_norm": 1.1568964719772339,
      "learning_rate": 9.489973684210526e-05,
      "loss": 0.3617,
      "step": 39399
    },
    {
      "epoch": 2.2062940978833017,
      "grad_norm": 1.7552618980407715,
      "learning_rate": 9.489947368421052e-05,
      "loss": 0.5455,
      "step": 39400
    },
    {
      "epoch": 2.2063500951954307,
      "grad_norm": 0.9996274709701538,
      "learning_rate": 9.48992105263158e-05,
      "loss": 0.3291,
      "step": 39401
    },
    {
      "epoch": 2.2064060925075597,
      "grad_norm": 1.0297887325286865,
      "learning_rate": 9.489894736842106e-05,
      "loss": 0.4223,
      "step": 39402
    },
    {
      "epoch": 2.2064620898196887,
      "grad_norm": 0.9772664308547974,
      "learning_rate": 9.489868421052633e-05,
      "loss": 0.2732,
      "step": 39403
    },
    {
      "epoch": 2.2065180871318177,
      "grad_norm": 1.4791831970214844,
      "learning_rate": 9.489842105263157e-05,
      "loss": 0.3542,
      "step": 39404
    },
    {
      "epoch": 2.2065740844439468,
      "grad_norm": 1.216740369796753,
      "learning_rate": 9.489815789473685e-05,
      "loss": 0.4073,
      "step": 39405
    },
    {
      "epoch": 2.206630081756076,
      "grad_norm": 1.59219491481781,
      "learning_rate": 9.489789473684211e-05,
      "loss": 0.4708,
      "step": 39406
    },
    {
      "epoch": 2.206686079068205,
      "grad_norm": 1.1547629833221436,
      "learning_rate": 9.489763157894738e-05,
      "loss": 0.4151,
      "step": 39407
    },
    {
      "epoch": 2.206742076380334,
      "grad_norm": 1.1807096004486084,
      "learning_rate": 9.489736842105264e-05,
      "loss": 0.4442,
      "step": 39408
    },
    {
      "epoch": 2.206798073692463,
      "grad_norm": 1.4545261859893799,
      "learning_rate": 9.48971052631579e-05,
      "loss": 0.3936,
      "step": 39409
    },
    {
      "epoch": 2.206854071004592,
      "grad_norm": 1.263182520866394,
      "learning_rate": 9.489684210526316e-05,
      "loss": 0.3769,
      "step": 39410
    },
    {
      "epoch": 2.206910068316721,
      "grad_norm": 1.2876176834106445,
      "learning_rate": 9.489657894736843e-05,
      "loss": 0.4633,
      "step": 39411
    },
    {
      "epoch": 2.20696606562885,
      "grad_norm": 1.2935539484024048,
      "learning_rate": 9.489631578947369e-05,
      "loss": 0.5254,
      "step": 39412
    },
    {
      "epoch": 2.207022062940979,
      "grad_norm": 1.2221156358718872,
      "learning_rate": 9.489605263157895e-05,
      "loss": 0.4282,
      "step": 39413
    },
    {
      "epoch": 2.207078060253108,
      "grad_norm": 1.3922674655914307,
      "learning_rate": 9.489578947368421e-05,
      "loss": 0.3174,
      "step": 39414
    },
    {
      "epoch": 2.207134057565237,
      "grad_norm": 1.2614103555679321,
      "learning_rate": 9.489552631578947e-05,
      "loss": 0.4641,
      "step": 39415
    },
    {
      "epoch": 2.207190054877366,
      "grad_norm": 1.3582285642623901,
      "learning_rate": 9.489526315789475e-05,
      "loss": 0.4834,
      "step": 39416
    },
    {
      "epoch": 2.207246052189495,
      "grad_norm": 1.2694936990737915,
      "learning_rate": 9.4895e-05,
      "loss": 0.3946,
      "step": 39417
    },
    {
      "epoch": 2.207302049501624,
      "grad_norm": 1.1047664880752563,
      "learning_rate": 9.489473684210527e-05,
      "loss": 0.3991,
      "step": 39418
    },
    {
      "epoch": 2.207358046813753,
      "grad_norm": 1.309397578239441,
      "learning_rate": 9.489447368421052e-05,
      "loss": 0.4047,
      "step": 39419
    },
    {
      "epoch": 2.207414044125882,
      "grad_norm": 0.965138852596283,
      "learning_rate": 9.48942105263158e-05,
      "loss": 0.2978,
      "step": 39420
    },
    {
      "epoch": 2.207470041438011,
      "grad_norm": 1.0652178525924683,
      "learning_rate": 9.489394736842106e-05,
      "loss": 0.4868,
      "step": 39421
    },
    {
      "epoch": 2.20752603875014,
      "grad_norm": 1.2504161596298218,
      "learning_rate": 9.489368421052632e-05,
      "loss": 0.5275,
      "step": 39422
    },
    {
      "epoch": 2.207582036062269,
      "grad_norm": 1.2136930227279663,
      "learning_rate": 9.489342105263158e-05,
      "loss": 0.3844,
      "step": 39423
    },
    {
      "epoch": 2.207638033374398,
      "grad_norm": 1.0742990970611572,
      "learning_rate": 9.489315789473685e-05,
      "loss": 0.3827,
      "step": 39424
    },
    {
      "epoch": 2.207694030686527,
      "grad_norm": 1.3198479413986206,
      "learning_rate": 9.489289473684211e-05,
      "loss": 0.5481,
      "step": 39425
    },
    {
      "epoch": 2.207750027998656,
      "grad_norm": 1.1638554334640503,
      "learning_rate": 9.489263157894738e-05,
      "loss": 0.35,
      "step": 39426
    },
    {
      "epoch": 2.207806025310785,
      "grad_norm": 1.0478335618972778,
      "learning_rate": 9.489236842105263e-05,
      "loss": 0.3603,
      "step": 39427
    },
    {
      "epoch": 2.207862022622914,
      "grad_norm": 1.3544954061508179,
      "learning_rate": 9.48921052631579e-05,
      "loss": 0.444,
      "step": 39428
    },
    {
      "epoch": 2.207918019935043,
      "grad_norm": 1.2961604595184326,
      "learning_rate": 9.489184210526316e-05,
      "loss": 0.459,
      "step": 39429
    },
    {
      "epoch": 2.2079740172471722,
      "grad_norm": 1.2026969194412231,
      "learning_rate": 9.489157894736842e-05,
      "loss": 0.3337,
      "step": 39430
    },
    {
      "epoch": 2.2080300145593013,
      "grad_norm": 1.0703009366989136,
      "learning_rate": 9.48913157894737e-05,
      "loss": 0.3582,
      "step": 39431
    },
    {
      "epoch": 2.2080860118714303,
      "grad_norm": 1.4483842849731445,
      "learning_rate": 9.489105263157894e-05,
      "loss": 0.5261,
      "step": 39432
    },
    {
      "epoch": 2.2081420091835593,
      "grad_norm": 1.122277021408081,
      "learning_rate": 9.489078947368422e-05,
      "loss": 0.3199,
      "step": 39433
    },
    {
      "epoch": 2.2081980064956883,
      "grad_norm": 1.0623100996017456,
      "learning_rate": 9.489052631578947e-05,
      "loss": 0.3993,
      "step": 39434
    },
    {
      "epoch": 2.2082540038078173,
      "grad_norm": 1.1094510555267334,
      "learning_rate": 9.489026315789475e-05,
      "loss": 0.3147,
      "step": 39435
    },
    {
      "epoch": 2.2083100011199464,
      "grad_norm": 1.2570363283157349,
      "learning_rate": 9.489e-05,
      "loss": 0.4363,
      "step": 39436
    },
    {
      "epoch": 2.2083659984320754,
      "grad_norm": 1.1157420873641968,
      "learning_rate": 9.488973684210527e-05,
      "loss": 0.4852,
      "step": 39437
    },
    {
      "epoch": 2.2084219957442044,
      "grad_norm": 1.2953319549560547,
      "learning_rate": 9.488947368421053e-05,
      "loss": 0.5739,
      "step": 39438
    },
    {
      "epoch": 2.2084779930563334,
      "grad_norm": 1.335352897644043,
      "learning_rate": 9.48892105263158e-05,
      "loss": 0.3707,
      "step": 39439
    },
    {
      "epoch": 2.2085339903684624,
      "grad_norm": 1.5566648244857788,
      "learning_rate": 9.488894736842106e-05,
      "loss": 0.3212,
      "step": 39440
    },
    {
      "epoch": 2.2085899876805914,
      "grad_norm": 1.1545898914337158,
      "learning_rate": 9.488868421052632e-05,
      "loss": 0.4212,
      "step": 39441
    },
    {
      "epoch": 2.2086459849927205,
      "grad_norm": 1.1226252317428589,
      "learning_rate": 9.488842105263158e-05,
      "loss": 0.3335,
      "step": 39442
    },
    {
      "epoch": 2.2087019823048495,
      "grad_norm": 1.2611626386642456,
      "learning_rate": 9.488815789473685e-05,
      "loss": 0.3461,
      "step": 39443
    },
    {
      "epoch": 2.2087579796169785,
      "grad_norm": 1.2921968698501587,
      "learning_rate": 9.488789473684211e-05,
      "loss": 0.482,
      "step": 39444
    },
    {
      "epoch": 2.2088139769291075,
      "grad_norm": 1.3033865690231323,
      "learning_rate": 9.488763157894737e-05,
      "loss": 0.4343,
      "step": 39445
    },
    {
      "epoch": 2.2088699742412365,
      "grad_norm": 1.3479501008987427,
      "learning_rate": 9.488736842105263e-05,
      "loss": 0.3744,
      "step": 39446
    },
    {
      "epoch": 2.2089259715533656,
      "grad_norm": 1.0953289270401,
      "learning_rate": 9.488710526315789e-05,
      "loss": 0.3697,
      "step": 39447
    },
    {
      "epoch": 2.2089819688654946,
      "grad_norm": 1.4060044288635254,
      "learning_rate": 9.488684210526317e-05,
      "loss": 0.3674,
      "step": 39448
    },
    {
      "epoch": 2.2090379661776236,
      "grad_norm": 1.0578078031539917,
      "learning_rate": 9.488657894736843e-05,
      "loss": 0.3018,
      "step": 39449
    },
    {
      "epoch": 2.2090939634897526,
      "grad_norm": 1.0024700164794922,
      "learning_rate": 9.488631578947368e-05,
      "loss": 0.39,
      "step": 39450
    },
    {
      "epoch": 2.2091499608018816,
      "grad_norm": 1.2217293977737427,
      "learning_rate": 9.488605263157894e-05,
      "loss": 0.4692,
      "step": 39451
    },
    {
      "epoch": 2.2092059581140107,
      "grad_norm": 2.155561685562134,
      "learning_rate": 9.488578947368422e-05,
      "loss": 0.4475,
      "step": 39452
    },
    {
      "epoch": 2.2092619554261397,
      "grad_norm": 1.280143141746521,
      "learning_rate": 9.488552631578948e-05,
      "loss": 0.3999,
      "step": 39453
    },
    {
      "epoch": 2.2093179527382687,
      "grad_norm": 1.1410243511199951,
      "learning_rate": 9.488526315789474e-05,
      "loss": 0.4895,
      "step": 39454
    },
    {
      "epoch": 2.2093739500503977,
      "grad_norm": 1.0631921291351318,
      "learning_rate": 9.4885e-05,
      "loss": 0.3316,
      "step": 39455
    },
    {
      "epoch": 2.2094299473625267,
      "grad_norm": 1.2528951168060303,
      "learning_rate": 9.488473684210527e-05,
      "loss": 0.3505,
      "step": 39456
    },
    {
      "epoch": 2.2094859446746558,
      "grad_norm": 1.5276639461517334,
      "learning_rate": 9.488447368421053e-05,
      "loss": 0.4463,
      "step": 39457
    },
    {
      "epoch": 2.2095419419867848,
      "grad_norm": 1.1898967027664185,
      "learning_rate": 9.48842105263158e-05,
      "loss": 0.397,
      "step": 39458
    },
    {
      "epoch": 2.209597939298914,
      "grad_norm": 1.1977025270462036,
      "learning_rate": 9.488394736842105e-05,
      "loss": 0.3885,
      "step": 39459
    },
    {
      "epoch": 2.209653936611043,
      "grad_norm": 1.207180380821228,
      "learning_rate": 9.488368421052632e-05,
      "loss": 0.3495,
      "step": 39460
    },
    {
      "epoch": 2.209709933923172,
      "grad_norm": 2.1690587997436523,
      "learning_rate": 9.488342105263158e-05,
      "loss": 0.4943,
      "step": 39461
    },
    {
      "epoch": 2.209765931235301,
      "grad_norm": 1.2840505838394165,
      "learning_rate": 9.488315789473686e-05,
      "loss": 0.3654,
      "step": 39462
    },
    {
      "epoch": 2.20982192854743,
      "grad_norm": 1.235318899154663,
      "learning_rate": 9.488289473684212e-05,
      "loss": 0.4723,
      "step": 39463
    },
    {
      "epoch": 2.209877925859559,
      "grad_norm": 0.914313793182373,
      "learning_rate": 9.488263157894736e-05,
      "loss": 0.3448,
      "step": 39464
    },
    {
      "epoch": 2.209933923171688,
      "grad_norm": 1.4670906066894531,
      "learning_rate": 9.488236842105263e-05,
      "loss": 0.4202,
      "step": 39465
    },
    {
      "epoch": 2.209989920483817,
      "grad_norm": 1.1392887830734253,
      "learning_rate": 9.48821052631579e-05,
      "loss": 0.3634,
      "step": 39466
    },
    {
      "epoch": 2.210045917795946,
      "grad_norm": 1.3882942199707031,
      "learning_rate": 9.488184210526317e-05,
      "loss": 0.4386,
      "step": 39467
    },
    {
      "epoch": 2.210101915108075,
      "grad_norm": 1.3332775831222534,
      "learning_rate": 9.488157894736843e-05,
      "loss": 0.402,
      "step": 39468
    },
    {
      "epoch": 2.210157912420204,
      "grad_norm": 1.3088740110397339,
      "learning_rate": 9.488131578947369e-05,
      "loss": 0.4818,
      "step": 39469
    },
    {
      "epoch": 2.210213909732333,
      "grad_norm": 1.0668730735778809,
      "learning_rate": 9.488105263157895e-05,
      "loss": 0.3802,
      "step": 39470
    },
    {
      "epoch": 2.210269907044462,
      "grad_norm": 1.3302572965621948,
      "learning_rate": 9.488078947368422e-05,
      "loss": 0.3559,
      "step": 39471
    },
    {
      "epoch": 2.210325904356591,
      "grad_norm": 1.135267734527588,
      "learning_rate": 9.488052631578948e-05,
      "loss": 0.3098,
      "step": 39472
    },
    {
      "epoch": 2.21038190166872,
      "grad_norm": 1.3553977012634277,
      "learning_rate": 9.488026315789474e-05,
      "loss": 0.3405,
      "step": 39473
    },
    {
      "epoch": 2.210437898980849,
      "grad_norm": 1.0660375356674194,
      "learning_rate": 9.488e-05,
      "loss": 0.3221,
      "step": 39474
    },
    {
      "epoch": 2.210493896292978,
      "grad_norm": 1.2229137420654297,
      "learning_rate": 9.487973684210527e-05,
      "loss": 0.3875,
      "step": 39475
    },
    {
      "epoch": 2.210549893605107,
      "grad_norm": 1.1356780529022217,
      "learning_rate": 9.487947368421053e-05,
      "loss": 0.4331,
      "step": 39476
    },
    {
      "epoch": 2.210605890917236,
      "grad_norm": 1.1612269878387451,
      "learning_rate": 9.487921052631579e-05,
      "loss": 0.3993,
      "step": 39477
    },
    {
      "epoch": 2.210661888229365,
      "grad_norm": 1.132050633430481,
      "learning_rate": 9.487894736842105e-05,
      "loss": 0.3471,
      "step": 39478
    },
    {
      "epoch": 2.210717885541494,
      "grad_norm": 1.2439926862716675,
      "learning_rate": 9.487868421052633e-05,
      "loss": 0.5064,
      "step": 39479
    },
    {
      "epoch": 2.210773882853623,
      "grad_norm": 1.0388453006744385,
      "learning_rate": 9.487842105263159e-05,
      "loss": 0.3963,
      "step": 39480
    },
    {
      "epoch": 2.210829880165752,
      "grad_norm": 1.0763019323349,
      "learning_rate": 9.487815789473686e-05,
      "loss": 0.3492,
      "step": 39481
    },
    {
      "epoch": 2.2108858774778812,
      "grad_norm": 1.2014552354812622,
      "learning_rate": 9.48778947368421e-05,
      "loss": 0.3836,
      "step": 39482
    },
    {
      "epoch": 2.2109418747900103,
      "grad_norm": 1.0644665956497192,
      "learning_rate": 9.487763157894736e-05,
      "loss": 0.3975,
      "step": 39483
    },
    {
      "epoch": 2.2109978721021393,
      "grad_norm": 1.3811395168304443,
      "learning_rate": 9.487736842105264e-05,
      "loss": 0.3515,
      "step": 39484
    },
    {
      "epoch": 2.2110538694142683,
      "grad_norm": 1.309718370437622,
      "learning_rate": 9.48771052631579e-05,
      "loss": 0.3616,
      "step": 39485
    },
    {
      "epoch": 2.2111098667263973,
      "grad_norm": 1.0296283960342407,
      "learning_rate": 9.487684210526317e-05,
      "loss": 0.3913,
      "step": 39486
    },
    {
      "epoch": 2.2111658640385263,
      "grad_norm": 1.2111494541168213,
      "learning_rate": 9.487657894736842e-05,
      "loss": 0.374,
      "step": 39487
    },
    {
      "epoch": 2.2112218613506553,
      "grad_norm": 1.1305137872695923,
      "learning_rate": 9.487631578947369e-05,
      "loss": 0.4628,
      "step": 39488
    },
    {
      "epoch": 2.2112778586627844,
      "grad_norm": 1.375943899154663,
      "learning_rate": 9.487605263157895e-05,
      "loss": 0.4683,
      "step": 39489
    },
    {
      "epoch": 2.2113338559749134,
      "grad_norm": 1.403503656387329,
      "learning_rate": 9.487578947368422e-05,
      "loss": 0.3749,
      "step": 39490
    },
    {
      "epoch": 2.2113898532870424,
      "grad_norm": 1.1193042993545532,
      "learning_rate": 9.487552631578947e-05,
      "loss": 0.3039,
      "step": 39491
    },
    {
      "epoch": 2.2114458505991714,
      "grad_norm": 1.232711672782898,
      "learning_rate": 9.487526315789474e-05,
      "loss": 0.4333,
      "step": 39492
    },
    {
      "epoch": 2.2115018479113004,
      "grad_norm": 1.1490355730056763,
      "learning_rate": 9.4875e-05,
      "loss": 0.4038,
      "step": 39493
    },
    {
      "epoch": 2.2115578452234295,
      "grad_norm": 1.578195571899414,
      "learning_rate": 9.487473684210528e-05,
      "loss": 0.4094,
      "step": 39494
    },
    {
      "epoch": 2.2116138425355585,
      "grad_norm": 1.1531920433044434,
      "learning_rate": 9.487447368421054e-05,
      "loss": 0.386,
      "step": 39495
    },
    {
      "epoch": 2.2116698398476875,
      "grad_norm": 1.2632719278335571,
      "learning_rate": 9.48742105263158e-05,
      "loss": 0.4551,
      "step": 39496
    },
    {
      "epoch": 2.2117258371598165,
      "grad_norm": 0.9915551543235779,
      "learning_rate": 9.487394736842105e-05,
      "loss": 0.335,
      "step": 39497
    },
    {
      "epoch": 2.2117818344719455,
      "grad_norm": 1.3023353815078735,
      "learning_rate": 9.487368421052633e-05,
      "loss": 0.4608,
      "step": 39498
    },
    {
      "epoch": 2.2118378317840746,
      "grad_norm": 1.3370416164398193,
      "learning_rate": 9.487342105263159e-05,
      "loss": 0.3906,
      "step": 39499
    },
    {
      "epoch": 2.2118938290962036,
      "grad_norm": 1.300503134727478,
      "learning_rate": 9.487315789473685e-05,
      "loss": 0.3701,
      "step": 39500
    },
    {
      "epoch": 2.2119498264083326,
      "grad_norm": 1.0996440649032593,
      "learning_rate": 9.487289473684211e-05,
      "loss": 0.4575,
      "step": 39501
    },
    {
      "epoch": 2.2120058237204616,
      "grad_norm": 1.1221290826797485,
      "learning_rate": 9.487263157894737e-05,
      "loss": 0.3514,
      "step": 39502
    },
    {
      "epoch": 2.2120618210325906,
      "grad_norm": 1.1745353937149048,
      "learning_rate": 9.487236842105264e-05,
      "loss": 0.4257,
      "step": 39503
    },
    {
      "epoch": 2.2121178183447197,
      "grad_norm": 1.1432949304580688,
      "learning_rate": 9.48721052631579e-05,
      "loss": 0.3615,
      "step": 39504
    },
    {
      "epoch": 2.2121738156568487,
      "grad_norm": 1.1346967220306396,
      "learning_rate": 9.487184210526316e-05,
      "loss": 0.4955,
      "step": 39505
    },
    {
      "epoch": 2.2122298129689777,
      "grad_norm": 1.1889140605926514,
      "learning_rate": 9.487157894736842e-05,
      "loss": 0.4462,
      "step": 39506
    },
    {
      "epoch": 2.2122858102811067,
      "grad_norm": 1.1017791032791138,
      "learning_rate": 9.487131578947369e-05,
      "loss": 0.413,
      "step": 39507
    },
    {
      "epoch": 2.2123418075932357,
      "grad_norm": 1.10360586643219,
      "learning_rate": 9.487105263157895e-05,
      "loss": 0.2981,
      "step": 39508
    },
    {
      "epoch": 2.2123978049053648,
      "grad_norm": 0.9922963380813599,
      "learning_rate": 9.487078947368421e-05,
      "loss": 0.3166,
      "step": 39509
    },
    {
      "epoch": 2.2124538022174933,
      "grad_norm": 1.1659842729568481,
      "learning_rate": 9.487052631578947e-05,
      "loss": 0.4249,
      "step": 39510
    },
    {
      "epoch": 2.212509799529623,
      "grad_norm": 1.0374927520751953,
      "learning_rate": 9.487026315789475e-05,
      "loss": 0.3453,
      "step": 39511
    },
    {
      "epoch": 2.2125657968417514,
      "grad_norm": 1.374549388885498,
      "learning_rate": 9.487e-05,
      "loss": 0.3526,
      "step": 39512
    },
    {
      "epoch": 2.212621794153881,
      "grad_norm": 1.4134262800216675,
      "learning_rate": 9.486973684210528e-05,
      "loss": 0.4607,
      "step": 39513
    },
    {
      "epoch": 2.2126777914660094,
      "grad_norm": 0.9947406053543091,
      "learning_rate": 9.486947368421052e-05,
      "loss": 0.4176,
      "step": 39514
    },
    {
      "epoch": 2.212733788778139,
      "grad_norm": 1.235861897468567,
      "learning_rate": 9.48692105263158e-05,
      "loss": 0.3164,
      "step": 39515
    },
    {
      "epoch": 2.2127897860902674,
      "grad_norm": 1.3603397607803345,
      "learning_rate": 9.486894736842106e-05,
      "loss": 0.2996,
      "step": 39516
    },
    {
      "epoch": 2.212845783402397,
      "grad_norm": 1.2442145347595215,
      "learning_rate": 9.486868421052632e-05,
      "loss": 0.4031,
      "step": 39517
    },
    {
      "epoch": 2.2129017807145255,
      "grad_norm": 1.1878242492675781,
      "learning_rate": 9.486842105263159e-05,
      "loss": 0.3654,
      "step": 39518
    },
    {
      "epoch": 2.212957778026655,
      "grad_norm": 1.181290626525879,
      "learning_rate": 9.486815789473684e-05,
      "loss": 0.4246,
      "step": 39519
    },
    {
      "epoch": 2.2130137753387835,
      "grad_norm": 1.4528449773788452,
      "learning_rate": 9.486789473684211e-05,
      "loss": 0.4401,
      "step": 39520
    },
    {
      "epoch": 2.213069772650913,
      "grad_norm": 1.2494711875915527,
      "learning_rate": 9.486763157894737e-05,
      "loss": 0.4755,
      "step": 39521
    },
    {
      "epoch": 2.2131257699630416,
      "grad_norm": 1.2395983934402466,
      "learning_rate": 9.486736842105264e-05,
      "loss": 0.5522,
      "step": 39522
    },
    {
      "epoch": 2.213181767275171,
      "grad_norm": 1.205452561378479,
      "learning_rate": 9.48671052631579e-05,
      "loss": 0.3622,
      "step": 39523
    },
    {
      "epoch": 2.2132377645872996,
      "grad_norm": 1.2171417474746704,
      "learning_rate": 9.486684210526316e-05,
      "loss": 0.3705,
      "step": 39524
    },
    {
      "epoch": 2.213293761899429,
      "grad_norm": 1.2575563192367554,
      "learning_rate": 9.486657894736842e-05,
      "loss": 0.413,
      "step": 39525
    },
    {
      "epoch": 2.2133497592115576,
      "grad_norm": 1.1577190160751343,
      "learning_rate": 9.48663157894737e-05,
      "loss": 0.36,
      "step": 39526
    },
    {
      "epoch": 2.213405756523687,
      "grad_norm": 1.4515182971954346,
      "learning_rate": 9.486605263157895e-05,
      "loss": 0.5074,
      "step": 39527
    },
    {
      "epoch": 2.2134617538358157,
      "grad_norm": 1.1130716800689697,
      "learning_rate": 9.486578947368421e-05,
      "loss": 0.3852,
      "step": 39528
    },
    {
      "epoch": 2.213517751147945,
      "grad_norm": 1.2099227905273438,
      "learning_rate": 9.486552631578947e-05,
      "loss": 0.3614,
      "step": 39529
    },
    {
      "epoch": 2.2135737484600737,
      "grad_norm": 1.2388745546340942,
      "learning_rate": 9.486526315789475e-05,
      "loss": 0.3864,
      "step": 39530
    },
    {
      "epoch": 2.2136297457722027,
      "grad_norm": 1.3295307159423828,
      "learning_rate": 9.486500000000001e-05,
      "loss": 0.3432,
      "step": 39531
    },
    {
      "epoch": 2.2136857430843317,
      "grad_norm": 1.3419257402420044,
      "learning_rate": 9.486473684210527e-05,
      "loss": 0.3558,
      "step": 39532
    },
    {
      "epoch": 2.2137417403964608,
      "grad_norm": 1.0854109525680542,
      "learning_rate": 9.486447368421053e-05,
      "loss": 0.3938,
      "step": 39533
    },
    {
      "epoch": 2.21379773770859,
      "grad_norm": 1.218134880065918,
      "learning_rate": 9.486421052631579e-05,
      "loss": 0.4503,
      "step": 39534
    },
    {
      "epoch": 2.213853735020719,
      "grad_norm": 1.2031400203704834,
      "learning_rate": 9.486394736842106e-05,
      "loss": 0.3679,
      "step": 39535
    },
    {
      "epoch": 2.213909732332848,
      "grad_norm": 1.1427584886550903,
      "learning_rate": 9.486368421052632e-05,
      "loss": 0.3866,
      "step": 39536
    },
    {
      "epoch": 2.213965729644977,
      "grad_norm": 1.0621387958526611,
      "learning_rate": 9.486342105263158e-05,
      "loss": 0.3412,
      "step": 39537
    },
    {
      "epoch": 2.214021726957106,
      "grad_norm": 1.129280686378479,
      "learning_rate": 9.486315789473684e-05,
      "loss": 0.3086,
      "step": 39538
    },
    {
      "epoch": 2.214077724269235,
      "grad_norm": 1.1368530988693237,
      "learning_rate": 9.486289473684211e-05,
      "loss": 0.3482,
      "step": 39539
    },
    {
      "epoch": 2.214133721581364,
      "grad_norm": 1.1958513259887695,
      "learning_rate": 9.486263157894737e-05,
      "loss": 0.3904,
      "step": 39540
    },
    {
      "epoch": 2.214189718893493,
      "grad_norm": 1.5152448415756226,
      "learning_rate": 9.486236842105265e-05,
      "loss": 0.4525,
      "step": 39541
    },
    {
      "epoch": 2.214245716205622,
      "grad_norm": 1.0344181060791016,
      "learning_rate": 9.486210526315789e-05,
      "loss": 0.3189,
      "step": 39542
    },
    {
      "epoch": 2.214301713517751,
      "grad_norm": 1.2310240268707275,
      "learning_rate": 9.486184210526316e-05,
      "loss": 0.4751,
      "step": 39543
    },
    {
      "epoch": 2.21435771082988,
      "grad_norm": 1.3463168144226074,
      "learning_rate": 9.486157894736842e-05,
      "loss": 0.3635,
      "step": 39544
    },
    {
      "epoch": 2.214413708142009,
      "grad_norm": 1.1757280826568604,
      "learning_rate": 9.48613157894737e-05,
      "loss": 0.4357,
      "step": 39545
    },
    {
      "epoch": 2.214469705454138,
      "grad_norm": 1.423470377922058,
      "learning_rate": 9.486105263157894e-05,
      "loss": 0.4359,
      "step": 39546
    },
    {
      "epoch": 2.214525702766267,
      "grad_norm": 1.0382089614868164,
      "learning_rate": 9.486078947368422e-05,
      "loss": 0.4562,
      "step": 39547
    },
    {
      "epoch": 2.214581700078396,
      "grad_norm": 1.0313842296600342,
      "learning_rate": 9.486052631578948e-05,
      "loss": 0.3448,
      "step": 39548
    },
    {
      "epoch": 2.214637697390525,
      "grad_norm": 1.2458701133728027,
      "learning_rate": 9.486026315789475e-05,
      "loss": 0.3934,
      "step": 39549
    },
    {
      "epoch": 2.214693694702654,
      "grad_norm": 1.136637568473816,
      "learning_rate": 9.486000000000001e-05,
      "loss": 0.3829,
      "step": 39550
    },
    {
      "epoch": 2.214749692014783,
      "grad_norm": 1.133142113685608,
      "learning_rate": 9.485973684210526e-05,
      "loss": 0.4146,
      "step": 39551
    },
    {
      "epoch": 2.214805689326912,
      "grad_norm": 1.0305284261703491,
      "learning_rate": 9.485947368421053e-05,
      "loss": 0.3061,
      "step": 39552
    },
    {
      "epoch": 2.214861686639041,
      "grad_norm": 1.3573909997940063,
      "learning_rate": 9.485921052631579e-05,
      "loss": 0.551,
      "step": 39553
    },
    {
      "epoch": 2.21491768395117,
      "grad_norm": 2.0785410404205322,
      "learning_rate": 9.485894736842106e-05,
      "loss": 0.4419,
      "step": 39554
    },
    {
      "epoch": 2.214973681263299,
      "grad_norm": 1.4044862985610962,
      "learning_rate": 9.485868421052632e-05,
      "loss": 0.48,
      "step": 39555
    },
    {
      "epoch": 2.215029678575428,
      "grad_norm": 1.051600456237793,
      "learning_rate": 9.485842105263158e-05,
      "loss": 0.3426,
      "step": 39556
    },
    {
      "epoch": 2.2150856758875572,
      "grad_norm": 1.3114019632339478,
      "learning_rate": 9.485815789473684e-05,
      "loss": 0.4629,
      "step": 39557
    },
    {
      "epoch": 2.2151416731996862,
      "grad_norm": 1.1474425792694092,
      "learning_rate": 9.485789473684211e-05,
      "loss": 0.3169,
      "step": 39558
    },
    {
      "epoch": 2.2151976705118153,
      "grad_norm": 1.1465036869049072,
      "learning_rate": 9.485763157894737e-05,
      "loss": 0.4715,
      "step": 39559
    },
    {
      "epoch": 2.2152536678239443,
      "grad_norm": 1.058963656425476,
      "learning_rate": 9.485736842105263e-05,
      "loss": 0.3828,
      "step": 39560
    },
    {
      "epoch": 2.2153096651360733,
      "grad_norm": 1.0208237171173096,
      "learning_rate": 9.48571052631579e-05,
      "loss": 0.3754,
      "step": 39561
    },
    {
      "epoch": 2.2153656624482023,
      "grad_norm": 1.2445060014724731,
      "learning_rate": 9.485684210526317e-05,
      "loss": 0.4185,
      "step": 39562
    },
    {
      "epoch": 2.2154216597603313,
      "grad_norm": 1.2862858772277832,
      "learning_rate": 9.485657894736843e-05,
      "loss": 0.4077,
      "step": 39563
    },
    {
      "epoch": 2.2154776570724604,
      "grad_norm": 1.5391099452972412,
      "learning_rate": 9.485631578947369e-05,
      "loss": 0.4952,
      "step": 39564
    },
    {
      "epoch": 2.2155336543845894,
      "grad_norm": 1.4818860292434692,
      "learning_rate": 9.485605263157895e-05,
      "loss": 0.4438,
      "step": 39565
    },
    {
      "epoch": 2.2155896516967184,
      "grad_norm": 0.9798838496208191,
      "learning_rate": 9.485578947368422e-05,
      "loss": 0.3535,
      "step": 39566
    },
    {
      "epoch": 2.2156456490088474,
      "grad_norm": 1.1640608310699463,
      "learning_rate": 9.485552631578948e-05,
      "loss": 0.3377,
      "step": 39567
    },
    {
      "epoch": 2.2157016463209764,
      "grad_norm": 0.9555137157440186,
      "learning_rate": 9.485526315789474e-05,
      "loss": 0.3356,
      "step": 39568
    },
    {
      "epoch": 2.2157576436331055,
      "grad_norm": 1.3477306365966797,
      "learning_rate": 9.4855e-05,
      "loss": 0.4165,
      "step": 39569
    },
    {
      "epoch": 2.2158136409452345,
      "grad_norm": 1.3791073560714722,
      "learning_rate": 9.485473684210526e-05,
      "loss": 0.3652,
      "step": 39570
    },
    {
      "epoch": 2.2158696382573635,
      "grad_norm": 1.3857711553573608,
      "learning_rate": 9.485447368421053e-05,
      "loss": 0.5188,
      "step": 39571
    },
    {
      "epoch": 2.2159256355694925,
      "grad_norm": 1.678026556968689,
      "learning_rate": 9.485421052631579e-05,
      "loss": 0.4708,
      "step": 39572
    },
    {
      "epoch": 2.2159816328816215,
      "grad_norm": 1.1528973579406738,
      "learning_rate": 9.485394736842107e-05,
      "loss": 0.5218,
      "step": 39573
    },
    {
      "epoch": 2.2160376301937506,
      "grad_norm": 1.249748706817627,
      "learning_rate": 9.485368421052631e-05,
      "loss": 0.37,
      "step": 39574
    },
    {
      "epoch": 2.2160936275058796,
      "grad_norm": 1.2241184711456299,
      "learning_rate": 9.485342105263158e-05,
      "loss": 0.47,
      "step": 39575
    },
    {
      "epoch": 2.2161496248180086,
      "grad_norm": 1.190361738204956,
      "learning_rate": 9.485315789473684e-05,
      "loss": 0.5866,
      "step": 39576
    },
    {
      "epoch": 2.2162056221301376,
      "grad_norm": 1.234858751296997,
      "learning_rate": 9.485289473684212e-05,
      "loss": 0.5482,
      "step": 39577
    },
    {
      "epoch": 2.2162616194422666,
      "grad_norm": 1.2922357320785522,
      "learning_rate": 9.485263157894738e-05,
      "loss": 0.355,
      "step": 39578
    },
    {
      "epoch": 2.2163176167543956,
      "grad_norm": 1.18617844581604,
      "learning_rate": 9.485236842105264e-05,
      "loss": 0.4831,
      "step": 39579
    },
    {
      "epoch": 2.2163736140665247,
      "grad_norm": 1.3128308057785034,
      "learning_rate": 9.48521052631579e-05,
      "loss": 0.4147,
      "step": 39580
    },
    {
      "epoch": 2.2164296113786537,
      "grad_norm": 1.2681304216384888,
      "learning_rate": 9.485184210526317e-05,
      "loss": 0.3738,
      "step": 39581
    },
    {
      "epoch": 2.2164856086907827,
      "grad_norm": 1.4341285228729248,
      "learning_rate": 9.485157894736843e-05,
      "loss": 0.5228,
      "step": 39582
    },
    {
      "epoch": 2.2165416060029117,
      "grad_norm": 1.6567810773849487,
      "learning_rate": 9.485131578947369e-05,
      "loss": 0.4377,
      "step": 39583
    },
    {
      "epoch": 2.2165976033150407,
      "grad_norm": 1.1970773935317993,
      "learning_rate": 9.485105263157895e-05,
      "loss": 0.3151,
      "step": 39584
    },
    {
      "epoch": 2.2166536006271698,
      "grad_norm": 1.059559941291809,
      "learning_rate": 9.485078947368421e-05,
      "loss": 0.4379,
      "step": 39585
    },
    {
      "epoch": 2.216709597939299,
      "grad_norm": 1.123970866203308,
      "learning_rate": 9.485052631578948e-05,
      "loss": 0.3411,
      "step": 39586
    },
    {
      "epoch": 2.216765595251428,
      "grad_norm": 1.0425021648406982,
      "learning_rate": 9.485026315789474e-05,
      "loss": 0.353,
      "step": 39587
    },
    {
      "epoch": 2.216821592563557,
      "grad_norm": 1.0684791803359985,
      "learning_rate": 9.485e-05,
      "loss": 0.5801,
      "step": 39588
    },
    {
      "epoch": 2.216877589875686,
      "grad_norm": 1.1544586420059204,
      "learning_rate": 9.484973684210526e-05,
      "loss": 0.343,
      "step": 39589
    },
    {
      "epoch": 2.216933587187815,
      "grad_norm": 1.0893439054489136,
      "learning_rate": 9.484947368421053e-05,
      "loss": 0.3112,
      "step": 39590
    },
    {
      "epoch": 2.216989584499944,
      "grad_norm": 1.119444727897644,
      "learning_rate": 9.48492105263158e-05,
      "loss": 0.5263,
      "step": 39591
    },
    {
      "epoch": 2.217045581812073,
      "grad_norm": 1.2116923332214355,
      "learning_rate": 9.484894736842105e-05,
      "loss": 0.3842,
      "step": 39592
    },
    {
      "epoch": 2.217101579124202,
      "grad_norm": 1.315686821937561,
      "learning_rate": 9.484868421052631e-05,
      "loss": 0.4364,
      "step": 39593
    },
    {
      "epoch": 2.217157576436331,
      "grad_norm": 1.096604824066162,
      "learning_rate": 9.484842105263159e-05,
      "loss": 0.3816,
      "step": 39594
    },
    {
      "epoch": 2.21721357374846,
      "grad_norm": 1.1477854251861572,
      "learning_rate": 9.484815789473685e-05,
      "loss": 0.4196,
      "step": 39595
    },
    {
      "epoch": 2.217269571060589,
      "grad_norm": 1.1825330257415771,
      "learning_rate": 9.484789473684212e-05,
      "loss": 0.3366,
      "step": 39596
    },
    {
      "epoch": 2.217325568372718,
      "grad_norm": 1.2780394554138184,
      "learning_rate": 9.484763157894737e-05,
      "loss": 0.376,
      "step": 39597
    },
    {
      "epoch": 2.217381565684847,
      "grad_norm": 1.1259257793426514,
      "learning_rate": 9.484736842105264e-05,
      "loss": 0.3643,
      "step": 39598
    },
    {
      "epoch": 2.217437562996976,
      "grad_norm": 1.4169061183929443,
      "learning_rate": 9.48471052631579e-05,
      "loss": 0.3612,
      "step": 39599
    },
    {
      "epoch": 2.217493560309105,
      "grad_norm": 1.1657676696777344,
      "learning_rate": 9.484684210526317e-05,
      "loss": 0.3814,
      "step": 39600
    },
    {
      "epoch": 2.217549557621234,
      "grad_norm": 1.1967030763626099,
      "learning_rate": 9.484657894736842e-05,
      "loss": 0.3648,
      "step": 39601
    },
    {
      "epoch": 2.217605554933363,
      "grad_norm": 1.3778002262115479,
      "learning_rate": 9.484631578947369e-05,
      "loss": 0.4008,
      "step": 39602
    },
    {
      "epoch": 2.217661552245492,
      "grad_norm": 1.1689342260360718,
      "learning_rate": 9.484605263157895e-05,
      "loss": 0.3479,
      "step": 39603
    },
    {
      "epoch": 2.217717549557621,
      "grad_norm": 1.2687549591064453,
      "learning_rate": 9.484578947368421e-05,
      "loss": 0.3335,
      "step": 39604
    },
    {
      "epoch": 2.21777354686975,
      "grad_norm": 1.307100534439087,
      "learning_rate": 9.484552631578948e-05,
      "loss": 0.4053,
      "step": 39605
    },
    {
      "epoch": 2.217829544181879,
      "grad_norm": 1.1839298009872437,
      "learning_rate": 9.484526315789473e-05,
      "loss": 0.3905,
      "step": 39606
    },
    {
      "epoch": 2.217885541494008,
      "grad_norm": 1.2022268772125244,
      "learning_rate": 9.4845e-05,
      "loss": 0.3875,
      "step": 39607
    },
    {
      "epoch": 2.217941538806137,
      "grad_norm": 1.246770977973938,
      "learning_rate": 9.484473684210526e-05,
      "loss": 0.5234,
      "step": 39608
    },
    {
      "epoch": 2.217997536118266,
      "grad_norm": 1.1393014192581177,
      "learning_rate": 9.484447368421054e-05,
      "loss": 0.3716,
      "step": 39609
    },
    {
      "epoch": 2.2180535334303952,
      "grad_norm": 1.2752878665924072,
      "learning_rate": 9.48442105263158e-05,
      "loss": 0.3931,
      "step": 39610
    },
    {
      "epoch": 2.2181095307425243,
      "grad_norm": 1.000220537185669,
      "learning_rate": 9.484394736842106e-05,
      "loss": 0.3441,
      "step": 39611
    },
    {
      "epoch": 2.2181655280546533,
      "grad_norm": 1.2714444398880005,
      "learning_rate": 9.484368421052632e-05,
      "loss": 0.3639,
      "step": 39612
    },
    {
      "epoch": 2.2182215253667823,
      "grad_norm": 1.600055456161499,
      "learning_rate": 9.484342105263159e-05,
      "loss": 0.4784,
      "step": 39613
    },
    {
      "epoch": 2.2182775226789113,
      "grad_norm": 1.0380969047546387,
      "learning_rate": 9.484315789473685e-05,
      "loss": 0.3703,
      "step": 39614
    },
    {
      "epoch": 2.2183335199910403,
      "grad_norm": 1.3356578350067139,
      "learning_rate": 9.484289473684211e-05,
      "loss": 0.4546,
      "step": 39615
    },
    {
      "epoch": 2.2183895173031694,
      "grad_norm": 1.2550917863845825,
      "learning_rate": 9.484263157894737e-05,
      "loss": 0.4955,
      "step": 39616
    },
    {
      "epoch": 2.2184455146152984,
      "grad_norm": 1.0455214977264404,
      "learning_rate": 9.484236842105264e-05,
      "loss": 0.3334,
      "step": 39617
    },
    {
      "epoch": 2.2185015119274274,
      "grad_norm": 1.1249582767486572,
      "learning_rate": 9.48421052631579e-05,
      "loss": 0.3429,
      "step": 39618
    },
    {
      "epoch": 2.2185575092395564,
      "grad_norm": 1.170705795288086,
      "learning_rate": 9.484184210526316e-05,
      "loss": 0.4165,
      "step": 39619
    },
    {
      "epoch": 2.2186135065516854,
      "grad_norm": 0.9637539386749268,
      "learning_rate": 9.484157894736842e-05,
      "loss": 0.3251,
      "step": 39620
    },
    {
      "epoch": 2.2186695038638145,
      "grad_norm": 1.1349906921386719,
      "learning_rate": 9.484131578947368e-05,
      "loss": 0.3594,
      "step": 39621
    },
    {
      "epoch": 2.2187255011759435,
      "grad_norm": 1.3575472831726074,
      "learning_rate": 9.484105263157895e-05,
      "loss": 0.5506,
      "step": 39622
    },
    {
      "epoch": 2.2187814984880725,
      "grad_norm": 1.261404037475586,
      "learning_rate": 9.484078947368421e-05,
      "loss": 0.4343,
      "step": 39623
    },
    {
      "epoch": 2.2188374958002015,
      "grad_norm": 1.1019033193588257,
      "learning_rate": 9.484052631578947e-05,
      "loss": 0.3609,
      "step": 39624
    },
    {
      "epoch": 2.2188934931123305,
      "grad_norm": 1.121375560760498,
      "learning_rate": 9.484026315789473e-05,
      "loss": 0.4969,
      "step": 39625
    },
    {
      "epoch": 2.2189494904244595,
      "grad_norm": 1.2595089673995972,
      "learning_rate": 9.484e-05,
      "loss": 0.4708,
      "step": 39626
    },
    {
      "epoch": 2.2190054877365886,
      "grad_norm": 1.1131263971328735,
      "learning_rate": 9.483973684210527e-05,
      "loss": 0.4694,
      "step": 39627
    },
    {
      "epoch": 2.2190614850487176,
      "grad_norm": 1.2495172023773193,
      "learning_rate": 9.483947368421054e-05,
      "loss": 0.3802,
      "step": 39628
    },
    {
      "epoch": 2.2191174823608466,
      "grad_norm": 1.1700639724731445,
      "learning_rate": 9.483921052631579e-05,
      "loss": 0.3091,
      "step": 39629
    },
    {
      "epoch": 2.2191734796729756,
      "grad_norm": 1.186990737915039,
      "learning_rate": 9.483894736842106e-05,
      "loss": 0.3972,
      "step": 39630
    },
    {
      "epoch": 2.2192294769851046,
      "grad_norm": 1.2280267477035522,
      "learning_rate": 9.483868421052632e-05,
      "loss": 0.2716,
      "step": 39631
    },
    {
      "epoch": 2.2192854742972337,
      "grad_norm": 1.2524919509887695,
      "learning_rate": 9.483842105263159e-05,
      "loss": 0.4729,
      "step": 39632
    },
    {
      "epoch": 2.2193414716093627,
      "grad_norm": 1.2634778022766113,
      "learning_rate": 9.483815789473685e-05,
      "loss": 0.4062,
      "step": 39633
    },
    {
      "epoch": 2.2193974689214917,
      "grad_norm": 1.1456775665283203,
      "learning_rate": 9.483789473684211e-05,
      "loss": 0.3915,
      "step": 39634
    },
    {
      "epoch": 2.2194534662336207,
      "grad_norm": 1.087252140045166,
      "learning_rate": 9.483763157894737e-05,
      "loss": 0.416,
      "step": 39635
    },
    {
      "epoch": 2.2195094635457497,
      "grad_norm": 2.1726810932159424,
      "learning_rate": 9.483736842105264e-05,
      "loss": 0.5216,
      "step": 39636
    },
    {
      "epoch": 2.2195654608578788,
      "grad_norm": 1.1706876754760742,
      "learning_rate": 9.48371052631579e-05,
      "loss": 0.4654,
      "step": 39637
    },
    {
      "epoch": 2.2196214581700078,
      "grad_norm": 1.371441125869751,
      "learning_rate": 9.483684210526315e-05,
      "loss": 0.3873,
      "step": 39638
    },
    {
      "epoch": 2.219677455482137,
      "grad_norm": 1.204808235168457,
      "learning_rate": 9.483657894736842e-05,
      "loss": 0.404,
      "step": 39639
    },
    {
      "epoch": 2.219733452794266,
      "grad_norm": 1.261924147605896,
      "learning_rate": 9.483631578947368e-05,
      "loss": 0.4299,
      "step": 39640
    },
    {
      "epoch": 2.219789450106395,
      "grad_norm": 1.1421394348144531,
      "learning_rate": 9.483605263157896e-05,
      "loss": 0.3555,
      "step": 39641
    },
    {
      "epoch": 2.219845447418524,
      "grad_norm": 1.214897871017456,
      "learning_rate": 9.483578947368422e-05,
      "loss": 0.6148,
      "step": 39642
    },
    {
      "epoch": 2.219901444730653,
      "grad_norm": 1.0573283433914185,
      "learning_rate": 9.483552631578948e-05,
      "loss": 0.4294,
      "step": 39643
    },
    {
      "epoch": 2.219957442042782,
      "grad_norm": 10.791913986206055,
      "learning_rate": 9.483526315789474e-05,
      "loss": 0.4417,
      "step": 39644
    },
    {
      "epoch": 2.220013439354911,
      "grad_norm": 1.390397548675537,
      "learning_rate": 9.483500000000001e-05,
      "loss": 0.529,
      "step": 39645
    },
    {
      "epoch": 2.22006943666704,
      "grad_norm": 1.1331439018249512,
      "learning_rate": 9.483473684210527e-05,
      "loss": 0.4047,
      "step": 39646
    },
    {
      "epoch": 2.220125433979169,
      "grad_norm": 1.2922428846359253,
      "learning_rate": 9.483447368421053e-05,
      "loss": 0.418,
      "step": 39647
    },
    {
      "epoch": 2.220181431291298,
      "grad_norm": 1.1824982166290283,
      "learning_rate": 9.483421052631579e-05,
      "loss": 0.3895,
      "step": 39648
    },
    {
      "epoch": 2.220237428603427,
      "grad_norm": 1.2164486646652222,
      "learning_rate": 9.483394736842106e-05,
      "loss": 0.4417,
      "step": 39649
    },
    {
      "epoch": 2.220293425915556,
      "grad_norm": 1.3991725444793701,
      "learning_rate": 9.483368421052632e-05,
      "loss": 0.4443,
      "step": 39650
    },
    {
      "epoch": 2.220349423227685,
      "grad_norm": 1.0671916007995605,
      "learning_rate": 9.48334210526316e-05,
      "loss": 0.4276,
      "step": 39651
    },
    {
      "epoch": 2.220405420539814,
      "grad_norm": 1.4109306335449219,
      "learning_rate": 9.483315789473684e-05,
      "loss": 0.409,
      "step": 39652
    },
    {
      "epoch": 2.220461417851943,
      "grad_norm": 1.2509561777114868,
      "learning_rate": 9.483289473684211e-05,
      "loss": 0.4706,
      "step": 39653
    },
    {
      "epoch": 2.220517415164072,
      "grad_norm": 1.0618574619293213,
      "learning_rate": 9.483263157894737e-05,
      "loss": 0.4609,
      "step": 39654
    },
    {
      "epoch": 2.220573412476201,
      "grad_norm": 1.336771845817566,
      "learning_rate": 9.483236842105263e-05,
      "loss": 0.5094,
      "step": 39655
    },
    {
      "epoch": 2.22062940978833,
      "grad_norm": 1.4276797771453857,
      "learning_rate": 9.48321052631579e-05,
      "loss": 0.3624,
      "step": 39656
    },
    {
      "epoch": 2.220685407100459,
      "grad_norm": 1.7243316173553467,
      "learning_rate": 9.483184210526315e-05,
      "loss": 0.4573,
      "step": 39657
    },
    {
      "epoch": 2.220741404412588,
      "grad_norm": 1.1294041872024536,
      "learning_rate": 9.483157894736843e-05,
      "loss": 0.4825,
      "step": 39658
    },
    {
      "epoch": 2.220797401724717,
      "grad_norm": 0.9865116477012634,
      "learning_rate": 9.483131578947369e-05,
      "loss": 0.3246,
      "step": 39659
    },
    {
      "epoch": 2.220853399036846,
      "grad_norm": 1.2761858701705933,
      "learning_rate": 9.483105263157896e-05,
      "loss": 0.3819,
      "step": 39660
    },
    {
      "epoch": 2.220909396348975,
      "grad_norm": 1.1085267066955566,
      "learning_rate": 9.48307894736842e-05,
      "loss": 0.3212,
      "step": 39661
    },
    {
      "epoch": 2.2209653936611042,
      "grad_norm": 1.7972567081451416,
      "learning_rate": 9.483052631578948e-05,
      "loss": 0.6801,
      "step": 39662
    },
    {
      "epoch": 2.2210213909732333,
      "grad_norm": 1.2082934379577637,
      "learning_rate": 9.483026315789474e-05,
      "loss": 0.4397,
      "step": 39663
    },
    {
      "epoch": 2.2210773882853623,
      "grad_norm": 1.0723037719726562,
      "learning_rate": 9.483000000000001e-05,
      "loss": 0.3937,
      "step": 39664
    },
    {
      "epoch": 2.2211333855974913,
      "grad_norm": 1.058102011680603,
      "learning_rate": 9.482973684210527e-05,
      "loss": 0.3651,
      "step": 39665
    },
    {
      "epoch": 2.2211893829096203,
      "grad_norm": 2.7301034927368164,
      "learning_rate": 9.482947368421053e-05,
      "loss": 0.5489,
      "step": 39666
    },
    {
      "epoch": 2.2212453802217493,
      "grad_norm": 1.2046235799789429,
      "learning_rate": 9.482921052631579e-05,
      "loss": 0.4115,
      "step": 39667
    },
    {
      "epoch": 2.2213013775338784,
      "grad_norm": 1.2329860925674438,
      "learning_rate": 9.482894736842106e-05,
      "loss": 0.3013,
      "step": 39668
    },
    {
      "epoch": 2.2213573748460074,
      "grad_norm": 1.0649844408035278,
      "learning_rate": 9.482868421052632e-05,
      "loss": 0.3596,
      "step": 39669
    },
    {
      "epoch": 2.2214133721581364,
      "grad_norm": 1.4467285871505737,
      "learning_rate": 9.482842105263158e-05,
      "loss": 0.3552,
      "step": 39670
    },
    {
      "epoch": 2.2214693694702654,
      "grad_norm": 1.0206724405288696,
      "learning_rate": 9.482815789473684e-05,
      "loss": 0.3166,
      "step": 39671
    },
    {
      "epoch": 2.2215253667823944,
      "grad_norm": 1.1924445629119873,
      "learning_rate": 9.48278947368421e-05,
      "loss": 0.4182,
      "step": 39672
    },
    {
      "epoch": 2.2215813640945234,
      "grad_norm": 1.177398443222046,
      "learning_rate": 9.482763157894738e-05,
      "loss": 0.4786,
      "step": 39673
    },
    {
      "epoch": 2.2216373614066525,
      "grad_norm": 1.2489194869995117,
      "learning_rate": 9.482736842105264e-05,
      "loss": 0.4167,
      "step": 39674
    },
    {
      "epoch": 2.2216933587187815,
      "grad_norm": 1.434828519821167,
      "learning_rate": 9.48271052631579e-05,
      "loss": 0.4523,
      "step": 39675
    },
    {
      "epoch": 2.2217493560309105,
      "grad_norm": 1.1750794649124146,
      "learning_rate": 9.482684210526316e-05,
      "loss": 0.487,
      "step": 39676
    },
    {
      "epoch": 2.2218053533430395,
      "grad_norm": 1.2070955038070679,
      "learning_rate": 9.482657894736843e-05,
      "loss": 0.3655,
      "step": 39677
    },
    {
      "epoch": 2.2218613506551685,
      "grad_norm": 1.1159030199050903,
      "learning_rate": 9.482631578947369e-05,
      "loss": 0.3658,
      "step": 39678
    },
    {
      "epoch": 2.2219173479672976,
      "grad_norm": 1.3631054162979126,
      "learning_rate": 9.482605263157895e-05,
      "loss": 0.4561,
      "step": 39679
    },
    {
      "epoch": 2.2219733452794266,
      "grad_norm": 1.663633942604065,
      "learning_rate": 9.482578947368421e-05,
      "loss": 0.4248,
      "step": 39680
    },
    {
      "epoch": 2.2220293425915556,
      "grad_norm": 1.2630255222320557,
      "learning_rate": 9.482552631578948e-05,
      "loss": 0.3751,
      "step": 39681
    },
    {
      "epoch": 2.2220853399036846,
      "grad_norm": 1.3706482648849487,
      "learning_rate": 9.482526315789474e-05,
      "loss": 0.3752,
      "step": 39682
    },
    {
      "epoch": 2.2221413372158136,
      "grad_norm": 1.5206102132797241,
      "learning_rate": 9.482500000000001e-05,
      "loss": 0.4022,
      "step": 39683
    },
    {
      "epoch": 2.2221973345279427,
      "grad_norm": 1.2914458513259888,
      "learning_rate": 9.482473684210526e-05,
      "loss": 0.4704,
      "step": 39684
    },
    {
      "epoch": 2.2222533318400717,
      "grad_norm": 0.9695793390274048,
      "learning_rate": 9.482447368421053e-05,
      "loss": 0.3478,
      "step": 39685
    },
    {
      "epoch": 2.2223093291522007,
      "grad_norm": 1.1576822996139526,
      "learning_rate": 9.48242105263158e-05,
      "loss": 0.384,
      "step": 39686
    },
    {
      "epoch": 2.2223653264643297,
      "grad_norm": 1.2858967781066895,
      "learning_rate": 9.482394736842107e-05,
      "loss": 0.5366,
      "step": 39687
    },
    {
      "epoch": 2.2224213237764587,
      "grad_norm": 1.0221651792526245,
      "learning_rate": 9.482368421052633e-05,
      "loss": 0.376,
      "step": 39688
    },
    {
      "epoch": 2.2224773210885878,
      "grad_norm": 1.1719318628311157,
      "learning_rate": 9.482342105263157e-05,
      "loss": 0.3026,
      "step": 39689
    },
    {
      "epoch": 2.2225333184007168,
      "grad_norm": 1.4125910997390747,
      "learning_rate": 9.482315789473685e-05,
      "loss": 0.338,
      "step": 39690
    },
    {
      "epoch": 2.222589315712846,
      "grad_norm": 1.1082124710083008,
      "learning_rate": 9.48228947368421e-05,
      "loss": 0.3626,
      "step": 39691
    },
    {
      "epoch": 2.222645313024975,
      "grad_norm": 1.2429324388504028,
      "learning_rate": 9.482263157894738e-05,
      "loss": 0.5297,
      "step": 39692
    },
    {
      "epoch": 2.222701310337104,
      "grad_norm": 1.2607945203781128,
      "learning_rate": 9.482236842105263e-05,
      "loss": 0.4846,
      "step": 39693
    },
    {
      "epoch": 2.222757307649233,
      "grad_norm": 1.5364521741867065,
      "learning_rate": 9.48221052631579e-05,
      "loss": 0.4344,
      "step": 39694
    },
    {
      "epoch": 2.222813304961362,
      "grad_norm": 1.3926931619644165,
      "learning_rate": 9.482184210526316e-05,
      "loss": 0.3599,
      "step": 39695
    },
    {
      "epoch": 2.222869302273491,
      "grad_norm": 1.4049726724624634,
      "learning_rate": 9.482157894736843e-05,
      "loss": 0.3878,
      "step": 39696
    },
    {
      "epoch": 2.22292529958562,
      "grad_norm": 1.5774027109146118,
      "learning_rate": 9.482131578947369e-05,
      "loss": 0.5274,
      "step": 39697
    },
    {
      "epoch": 2.222981296897749,
      "grad_norm": 1.7156578302383423,
      "learning_rate": 9.482105263157895e-05,
      "loss": 0.4011,
      "step": 39698
    },
    {
      "epoch": 2.223037294209878,
      "grad_norm": 1.2779486179351807,
      "learning_rate": 9.482078947368421e-05,
      "loss": 0.4061,
      "step": 39699
    },
    {
      "epoch": 2.223093291522007,
      "grad_norm": 1.2724679708480835,
      "learning_rate": 9.482052631578948e-05,
      "loss": 0.3558,
      "step": 39700
    },
    {
      "epoch": 2.223149288834136,
      "grad_norm": 1.2752137184143066,
      "learning_rate": 9.482026315789474e-05,
      "loss": 0.4258,
      "step": 39701
    },
    {
      "epoch": 2.223205286146265,
      "grad_norm": 1.357906699180603,
      "learning_rate": 9.482e-05,
      "loss": 0.3723,
      "step": 39702
    },
    {
      "epoch": 2.223261283458394,
      "grad_norm": 1.6095585823059082,
      "learning_rate": 9.481973684210526e-05,
      "loss": 0.4974,
      "step": 39703
    },
    {
      "epoch": 2.223317280770523,
      "grad_norm": 1.19790518283844,
      "learning_rate": 9.481947368421054e-05,
      "loss": 0.4012,
      "step": 39704
    },
    {
      "epoch": 2.223373278082652,
      "grad_norm": 1.2199758291244507,
      "learning_rate": 9.48192105263158e-05,
      "loss": 0.5005,
      "step": 39705
    },
    {
      "epoch": 2.223429275394781,
      "grad_norm": 1.2912514209747314,
      "learning_rate": 9.481894736842106e-05,
      "loss": 0.3701,
      "step": 39706
    },
    {
      "epoch": 2.22348527270691,
      "grad_norm": 1.1865085363388062,
      "learning_rate": 9.481868421052632e-05,
      "loss": 0.4453,
      "step": 39707
    },
    {
      "epoch": 2.223541270019039,
      "grad_norm": 2.6361682415008545,
      "learning_rate": 9.481842105263158e-05,
      "loss": 0.3508,
      "step": 39708
    },
    {
      "epoch": 2.223597267331168,
      "grad_norm": 1.3985697031021118,
      "learning_rate": 9.481815789473685e-05,
      "loss": 0.4197,
      "step": 39709
    },
    {
      "epoch": 2.223653264643297,
      "grad_norm": 1.170430302619934,
      "learning_rate": 9.481789473684211e-05,
      "loss": 0.4483,
      "step": 39710
    },
    {
      "epoch": 2.223709261955426,
      "grad_norm": 1.0090118646621704,
      "learning_rate": 9.481763157894737e-05,
      "loss": 0.3575,
      "step": 39711
    },
    {
      "epoch": 2.223765259267555,
      "grad_norm": 1.0705267190933228,
      "learning_rate": 9.481736842105263e-05,
      "loss": 0.348,
      "step": 39712
    },
    {
      "epoch": 2.223821256579684,
      "grad_norm": 1.1835777759552002,
      "learning_rate": 9.48171052631579e-05,
      "loss": 0.4068,
      "step": 39713
    },
    {
      "epoch": 2.2238772538918132,
      "grad_norm": 1.257071852684021,
      "learning_rate": 9.481684210526316e-05,
      "loss": 0.4582,
      "step": 39714
    },
    {
      "epoch": 2.2239332512039423,
      "grad_norm": 2.141129732131958,
      "learning_rate": 9.481657894736843e-05,
      "loss": 0.4844,
      "step": 39715
    },
    {
      "epoch": 2.2239892485160713,
      "grad_norm": 1.0961389541625977,
      "learning_rate": 9.481631578947368e-05,
      "loss": 0.4245,
      "step": 39716
    },
    {
      "epoch": 2.2240452458282003,
      "grad_norm": 1.2191020250320435,
      "learning_rate": 9.481605263157895e-05,
      "loss": 0.4326,
      "step": 39717
    },
    {
      "epoch": 2.2241012431403293,
      "grad_norm": 1.1012532711029053,
      "learning_rate": 9.481578947368421e-05,
      "loss": 0.3926,
      "step": 39718
    },
    {
      "epoch": 2.2241572404524583,
      "grad_norm": 1.0334488153457642,
      "learning_rate": 9.481552631578949e-05,
      "loss": 0.4717,
      "step": 39719
    },
    {
      "epoch": 2.2242132377645873,
      "grad_norm": 1.1681933403015137,
      "learning_rate": 9.481526315789475e-05,
      "loss": 0.3487,
      "step": 39720
    },
    {
      "epoch": 2.2242692350767164,
      "grad_norm": 1.2991100549697876,
      "learning_rate": 9.4815e-05,
      "loss": 0.4101,
      "step": 39721
    },
    {
      "epoch": 2.2243252323888454,
      "grad_norm": 1.236640214920044,
      "learning_rate": 9.481473684210527e-05,
      "loss": 0.5678,
      "step": 39722
    },
    {
      "epoch": 2.2243812297009744,
      "grad_norm": 1.0643407106399536,
      "learning_rate": 9.481447368421054e-05,
      "loss": 0.3461,
      "step": 39723
    },
    {
      "epoch": 2.2244372270131034,
      "grad_norm": 1.4532872438430786,
      "learning_rate": 9.48142105263158e-05,
      "loss": 0.3079,
      "step": 39724
    },
    {
      "epoch": 2.2244932243252324,
      "grad_norm": 1.170744776725769,
      "learning_rate": 9.481394736842106e-05,
      "loss": 0.4741,
      "step": 39725
    },
    {
      "epoch": 2.2245492216373615,
      "grad_norm": 1.157811164855957,
      "learning_rate": 9.481368421052632e-05,
      "loss": 0.3857,
      "step": 39726
    },
    {
      "epoch": 2.2246052189494905,
      "grad_norm": 1.125324010848999,
      "learning_rate": 9.481342105263158e-05,
      "loss": 0.4201,
      "step": 39727
    },
    {
      "epoch": 2.2246612162616195,
      "grad_norm": 1.4654470682144165,
      "learning_rate": 9.481315789473685e-05,
      "loss": 0.3172,
      "step": 39728
    },
    {
      "epoch": 2.2247172135737485,
      "grad_norm": 1.0262280702590942,
      "learning_rate": 9.481289473684211e-05,
      "loss": 0.4554,
      "step": 39729
    },
    {
      "epoch": 2.2247732108858775,
      "grad_norm": 1.249243974685669,
      "learning_rate": 9.481263157894737e-05,
      "loss": 0.3718,
      "step": 39730
    },
    {
      "epoch": 2.2248292081980066,
      "grad_norm": 1.0469605922698975,
      "learning_rate": 9.481236842105263e-05,
      "loss": 0.3856,
      "step": 39731
    },
    {
      "epoch": 2.2248852055101356,
      "grad_norm": 1.4074395895004272,
      "learning_rate": 9.48121052631579e-05,
      "loss": 0.4509,
      "step": 39732
    },
    {
      "epoch": 2.2249412028222646,
      "grad_norm": 1.3245704174041748,
      "learning_rate": 9.481184210526316e-05,
      "loss": 0.478,
      "step": 39733
    },
    {
      "epoch": 2.2249972001343936,
      "grad_norm": 1.2466191053390503,
      "learning_rate": 9.481157894736842e-05,
      "loss": 0.3827,
      "step": 39734
    },
    {
      "epoch": 2.2250531974465226,
      "grad_norm": 1.398132085800171,
      "learning_rate": 9.481131578947368e-05,
      "loss": 0.4098,
      "step": 39735
    },
    {
      "epoch": 2.2251091947586517,
      "grad_norm": 1.1671637296676636,
      "learning_rate": 9.481105263157896e-05,
      "loss": 0.3573,
      "step": 39736
    },
    {
      "epoch": 2.2251651920707807,
      "grad_norm": 1.032700777053833,
      "learning_rate": 9.481078947368422e-05,
      "loss": 0.4409,
      "step": 39737
    },
    {
      "epoch": 2.2252211893829097,
      "grad_norm": 1.1913005113601685,
      "learning_rate": 9.481052631578949e-05,
      "loss": 0.4002,
      "step": 39738
    },
    {
      "epoch": 2.2252771866950387,
      "grad_norm": 2.002268075942993,
      "learning_rate": 9.481026315789474e-05,
      "loss": 0.4355,
      "step": 39739
    },
    {
      "epoch": 2.2253331840071677,
      "grad_norm": 1.1960407495498657,
      "learning_rate": 9.481000000000001e-05,
      "loss": 0.4236,
      "step": 39740
    },
    {
      "epoch": 2.2253891813192967,
      "grad_norm": 1.2726662158966064,
      "learning_rate": 9.480973684210527e-05,
      "loss": 0.3861,
      "step": 39741
    },
    {
      "epoch": 2.2254451786314258,
      "grad_norm": 1.1637592315673828,
      "learning_rate": 9.480947368421053e-05,
      "loss": 0.4101,
      "step": 39742
    },
    {
      "epoch": 2.225501175943555,
      "grad_norm": 0.940432608127594,
      "learning_rate": 9.48092105263158e-05,
      "loss": 0.3208,
      "step": 39743
    },
    {
      "epoch": 2.225557173255684,
      "grad_norm": 1.6558113098144531,
      "learning_rate": 9.480894736842105e-05,
      "loss": 0.3767,
      "step": 39744
    },
    {
      "epoch": 2.225613170567813,
      "grad_norm": 1.0391144752502441,
      "learning_rate": 9.480868421052632e-05,
      "loss": 0.4304,
      "step": 39745
    },
    {
      "epoch": 2.225669167879942,
      "grad_norm": 1.2625582218170166,
      "learning_rate": 9.480842105263158e-05,
      "loss": 0.5164,
      "step": 39746
    },
    {
      "epoch": 2.225725165192071,
      "grad_norm": 1.1927005052566528,
      "learning_rate": 9.480815789473685e-05,
      "loss": 0.4818,
      "step": 39747
    },
    {
      "epoch": 2.2257811625042,
      "grad_norm": 1.2076629400253296,
      "learning_rate": 9.48078947368421e-05,
      "loss": 0.3365,
      "step": 39748
    },
    {
      "epoch": 2.225837159816329,
      "grad_norm": 1.2277437448501587,
      "learning_rate": 9.480763157894737e-05,
      "loss": 0.3782,
      "step": 39749
    },
    {
      "epoch": 2.225893157128458,
      "grad_norm": 1.1514946222305298,
      "learning_rate": 9.480736842105263e-05,
      "loss": 0.4731,
      "step": 39750
    },
    {
      "epoch": 2.225949154440587,
      "grad_norm": 1.470760703086853,
      "learning_rate": 9.48071052631579e-05,
      "loss": 0.4129,
      "step": 39751
    },
    {
      "epoch": 2.226005151752716,
      "grad_norm": 0.8810442686080933,
      "learning_rate": 9.480684210526317e-05,
      "loss": 0.2546,
      "step": 39752
    },
    {
      "epoch": 2.226061149064845,
      "grad_norm": 1.1701087951660156,
      "learning_rate": 9.480657894736843e-05,
      "loss": 0.3746,
      "step": 39753
    },
    {
      "epoch": 2.226117146376974,
      "grad_norm": 1.3707770109176636,
      "learning_rate": 9.480631578947369e-05,
      "loss": 0.362,
      "step": 39754
    },
    {
      "epoch": 2.226173143689103,
      "grad_norm": 1.1347661018371582,
      "learning_rate": 9.480605263157896e-05,
      "loss": 0.3165,
      "step": 39755
    },
    {
      "epoch": 2.226229141001232,
      "grad_norm": 1.1795978546142578,
      "learning_rate": 9.480578947368422e-05,
      "loss": 0.4668,
      "step": 39756
    },
    {
      "epoch": 2.226285138313361,
      "grad_norm": 1.299878478050232,
      "learning_rate": 9.480552631578948e-05,
      "loss": 0.4173,
      "step": 39757
    },
    {
      "epoch": 2.22634113562549,
      "grad_norm": 0.974886417388916,
      "learning_rate": 9.480526315789474e-05,
      "loss": 0.318,
      "step": 39758
    },
    {
      "epoch": 2.226397132937619,
      "grad_norm": 1.0609098672866821,
      "learning_rate": 9.4805e-05,
      "loss": 0.4361,
      "step": 39759
    },
    {
      "epoch": 2.226453130249748,
      "grad_norm": 1.8939884901046753,
      "learning_rate": 9.480473684210527e-05,
      "loss": 0.3306,
      "step": 39760
    },
    {
      "epoch": 2.226509127561877,
      "grad_norm": 1.3681687116622925,
      "learning_rate": 9.480447368421053e-05,
      "loss": 0.426,
      "step": 39761
    },
    {
      "epoch": 2.226565124874006,
      "grad_norm": 1.3573007583618164,
      "learning_rate": 9.480421052631579e-05,
      "loss": 0.5548,
      "step": 39762
    },
    {
      "epoch": 2.226621122186135,
      "grad_norm": 1.0843397378921509,
      "learning_rate": 9.480394736842105e-05,
      "loss": 0.3726,
      "step": 39763
    },
    {
      "epoch": 2.226677119498264,
      "grad_norm": 1.2007956504821777,
      "learning_rate": 9.480368421052632e-05,
      "loss": 0.4688,
      "step": 39764
    },
    {
      "epoch": 2.226733116810393,
      "grad_norm": 1.2222439050674438,
      "learning_rate": 9.480342105263158e-05,
      "loss": 0.3997,
      "step": 39765
    },
    {
      "epoch": 2.2267891141225222,
      "grad_norm": 1.286411166191101,
      "learning_rate": 9.480315789473684e-05,
      "loss": 0.6286,
      "step": 39766
    },
    {
      "epoch": 2.2268451114346512,
      "grad_norm": 1.085105538368225,
      "learning_rate": 9.48028947368421e-05,
      "loss": 0.4206,
      "step": 39767
    },
    {
      "epoch": 2.2269011087467803,
      "grad_norm": 1.2870198488235474,
      "learning_rate": 9.480263157894738e-05,
      "loss": 0.384,
      "step": 39768
    },
    {
      "epoch": 2.2269571060589093,
      "grad_norm": 1.1900490522384644,
      "learning_rate": 9.480236842105264e-05,
      "loss": 0.3915,
      "step": 39769
    },
    {
      "epoch": 2.2270131033710383,
      "grad_norm": 1.1406409740447998,
      "learning_rate": 9.480210526315791e-05,
      "loss": 0.4217,
      "step": 39770
    },
    {
      "epoch": 2.2270691006831673,
      "grad_norm": 1.1910468339920044,
      "learning_rate": 9.480184210526316e-05,
      "loss": 0.3716,
      "step": 39771
    },
    {
      "epoch": 2.2271250979952963,
      "grad_norm": 1.4051614999771118,
      "learning_rate": 9.480157894736843e-05,
      "loss": 0.433,
      "step": 39772
    },
    {
      "epoch": 2.2271810953074254,
      "grad_norm": 1.2196649312973022,
      "learning_rate": 9.480131578947369e-05,
      "loss": 0.3273,
      "step": 39773
    },
    {
      "epoch": 2.2272370926195544,
      "grad_norm": 1.033158302307129,
      "learning_rate": 9.480105263157896e-05,
      "loss": 0.3351,
      "step": 39774
    },
    {
      "epoch": 2.2272930899316834,
      "grad_norm": 1.2032259702682495,
      "learning_rate": 9.480078947368422e-05,
      "loss": 0.3464,
      "step": 39775
    },
    {
      "epoch": 2.2273490872438124,
      "grad_norm": 1.281961441040039,
      "learning_rate": 9.480052631578947e-05,
      "loss": 0.4609,
      "step": 39776
    },
    {
      "epoch": 2.2274050845559414,
      "grad_norm": 4.507180213928223,
      "learning_rate": 9.480026315789474e-05,
      "loss": 0.3777,
      "step": 39777
    },
    {
      "epoch": 2.2274610818680705,
      "grad_norm": 1.2982254028320312,
      "learning_rate": 9.48e-05,
      "loss": 0.4758,
      "step": 39778
    },
    {
      "epoch": 2.2275170791801995,
      "grad_norm": 1.4358621835708618,
      "learning_rate": 9.479973684210527e-05,
      "loss": 0.406,
      "step": 39779
    },
    {
      "epoch": 2.2275730764923285,
      "grad_norm": 1.2582206726074219,
      "learning_rate": 9.479947368421053e-05,
      "loss": 0.4762,
      "step": 39780
    },
    {
      "epoch": 2.2276290738044575,
      "grad_norm": 1.2534246444702148,
      "learning_rate": 9.479921052631579e-05,
      "loss": 0.4386,
      "step": 39781
    },
    {
      "epoch": 2.2276850711165865,
      "grad_norm": 2.333282232284546,
      "learning_rate": 9.479894736842105e-05,
      "loss": 0.3885,
      "step": 39782
    },
    {
      "epoch": 2.2277410684287156,
      "grad_norm": 1.3362810611724854,
      "learning_rate": 9.479868421052633e-05,
      "loss": 0.464,
      "step": 39783
    },
    {
      "epoch": 2.2277970657408446,
      "grad_norm": 1.3119702339172363,
      "learning_rate": 9.479842105263159e-05,
      "loss": 0.6372,
      "step": 39784
    },
    {
      "epoch": 2.2278530630529736,
      "grad_norm": 1.0718625783920288,
      "learning_rate": 9.479815789473685e-05,
      "loss": 0.3374,
      "step": 39785
    },
    {
      "epoch": 2.2279090603651026,
      "grad_norm": 1.34573495388031,
      "learning_rate": 9.47978947368421e-05,
      "loss": 0.3578,
      "step": 39786
    },
    {
      "epoch": 2.2279650576772316,
      "grad_norm": 1.118297815322876,
      "learning_rate": 9.479763157894738e-05,
      "loss": 0.291,
      "step": 39787
    },
    {
      "epoch": 2.2280210549893606,
      "grad_norm": 1.1476413011550903,
      "learning_rate": 9.479736842105264e-05,
      "loss": 0.3062,
      "step": 39788
    },
    {
      "epoch": 2.2280770523014897,
      "grad_norm": 1.116910696029663,
      "learning_rate": 9.47971052631579e-05,
      "loss": 0.4255,
      "step": 39789
    },
    {
      "epoch": 2.2281330496136187,
      "grad_norm": 1.252384066581726,
      "learning_rate": 9.479684210526316e-05,
      "loss": 0.4015,
      "step": 39790
    },
    {
      "epoch": 2.2281890469257477,
      "grad_norm": 1.2172892093658447,
      "learning_rate": 9.479657894736843e-05,
      "loss": 0.4348,
      "step": 39791
    },
    {
      "epoch": 2.2282450442378767,
      "grad_norm": 1.406664490699768,
      "learning_rate": 9.479631578947369e-05,
      "loss": 0.4189,
      "step": 39792
    },
    {
      "epoch": 2.2283010415500057,
      "grad_norm": 1.3978335857391357,
      "learning_rate": 9.479605263157895e-05,
      "loss": 0.3727,
      "step": 39793
    },
    {
      "epoch": 2.2283570388621348,
      "grad_norm": 1.137801170349121,
      "learning_rate": 9.479578947368421e-05,
      "loss": 0.3744,
      "step": 39794
    },
    {
      "epoch": 2.228413036174264,
      "grad_norm": 1.4093049764633179,
      "learning_rate": 9.479552631578947e-05,
      "loss": 0.6216,
      "step": 39795
    },
    {
      "epoch": 2.228469033486393,
      "grad_norm": 1.2021464109420776,
      "learning_rate": 9.479526315789474e-05,
      "loss": 0.3637,
      "step": 39796
    },
    {
      "epoch": 2.228525030798522,
      "grad_norm": 1.1721134185791016,
      "learning_rate": 9.4795e-05,
      "loss": 0.3977,
      "step": 39797
    },
    {
      "epoch": 2.228581028110651,
      "grad_norm": 0.9753643274307251,
      "learning_rate": 9.479473684210528e-05,
      "loss": 0.2699,
      "step": 39798
    },
    {
      "epoch": 2.22863702542278,
      "grad_norm": 1.1609952449798584,
      "learning_rate": 9.479447368421052e-05,
      "loss": 0.4145,
      "step": 39799
    },
    {
      "epoch": 2.228693022734909,
      "grad_norm": 1.2368007898330688,
      "learning_rate": 9.47942105263158e-05,
      "loss": 0.5363,
      "step": 39800
    },
    {
      "epoch": 2.228749020047038,
      "grad_norm": 0.9663000106811523,
      "learning_rate": 9.479394736842106e-05,
      "loss": 0.249,
      "step": 39801
    },
    {
      "epoch": 2.228805017359167,
      "grad_norm": 1.2016947269439697,
      "learning_rate": 9.479368421052633e-05,
      "loss": 0.3795,
      "step": 39802
    },
    {
      "epoch": 2.228861014671296,
      "grad_norm": 1.0489708185195923,
      "learning_rate": 9.479342105263157e-05,
      "loss": 0.3504,
      "step": 39803
    },
    {
      "epoch": 2.228917011983425,
      "grad_norm": 0.964414656162262,
      "learning_rate": 9.479315789473685e-05,
      "loss": 0.3381,
      "step": 39804
    },
    {
      "epoch": 2.228973009295554,
      "grad_norm": 1.1644946336746216,
      "learning_rate": 9.479289473684211e-05,
      "loss": 0.3507,
      "step": 39805
    },
    {
      "epoch": 2.229029006607683,
      "grad_norm": 1.5401169061660767,
      "learning_rate": 9.479263157894738e-05,
      "loss": 0.4503,
      "step": 39806
    },
    {
      "epoch": 2.229085003919812,
      "grad_norm": 1.1361335515975952,
      "learning_rate": 9.479236842105264e-05,
      "loss": 0.4594,
      "step": 39807
    },
    {
      "epoch": 2.229141001231941,
      "grad_norm": 1.2504023313522339,
      "learning_rate": 9.47921052631579e-05,
      "loss": 0.3669,
      "step": 39808
    },
    {
      "epoch": 2.22919699854407,
      "grad_norm": 1.2397444248199463,
      "learning_rate": 9.479184210526316e-05,
      "loss": 0.2995,
      "step": 39809
    },
    {
      "epoch": 2.229252995856199,
      "grad_norm": 1.14100980758667,
      "learning_rate": 9.479157894736842e-05,
      "loss": 0.3375,
      "step": 39810
    },
    {
      "epoch": 2.229308993168328,
      "grad_norm": 1.2952340841293335,
      "learning_rate": 9.47913157894737e-05,
      "loss": 0.4742,
      "step": 39811
    },
    {
      "epoch": 2.229364990480457,
      "grad_norm": 1.1374386548995972,
      "learning_rate": 9.479105263157895e-05,
      "loss": 0.385,
      "step": 39812
    },
    {
      "epoch": 2.229420987792586,
      "grad_norm": 1.1735543012619019,
      "learning_rate": 9.479078947368421e-05,
      "loss": 0.4437,
      "step": 39813
    },
    {
      "epoch": 2.229476985104715,
      "grad_norm": 1.3908323049545288,
      "learning_rate": 9.479052631578947e-05,
      "loss": 0.4092,
      "step": 39814
    },
    {
      "epoch": 2.229532982416844,
      "grad_norm": 1.7981433868408203,
      "learning_rate": 9.479026315789475e-05,
      "loss": 0.3561,
      "step": 39815
    },
    {
      "epoch": 2.229588979728973,
      "grad_norm": 1.3346965312957764,
      "learning_rate": 9.479e-05,
      "loss": 0.4796,
      "step": 39816
    },
    {
      "epoch": 2.229644977041102,
      "grad_norm": 1.8139790296554565,
      "learning_rate": 9.478973684210527e-05,
      "loss": 0.3339,
      "step": 39817
    },
    {
      "epoch": 2.2297009743532312,
      "grad_norm": 1.2758008241653442,
      "learning_rate": 9.478947368421053e-05,
      "loss": 0.4748,
      "step": 39818
    },
    {
      "epoch": 2.2297569716653602,
      "grad_norm": 1.1799649000167847,
      "learning_rate": 9.47892105263158e-05,
      "loss": 0.3774,
      "step": 39819
    },
    {
      "epoch": 2.2298129689774893,
      "grad_norm": 1.022952675819397,
      "learning_rate": 9.478894736842106e-05,
      "loss": 0.3983,
      "step": 39820
    },
    {
      "epoch": 2.2298689662896183,
      "grad_norm": 1.3135348558425903,
      "learning_rate": 9.478868421052632e-05,
      "loss": 0.3605,
      "step": 39821
    },
    {
      "epoch": 2.2299249636017473,
      "grad_norm": 1.201823115348816,
      "learning_rate": 9.478842105263158e-05,
      "loss": 0.3748,
      "step": 39822
    },
    {
      "epoch": 2.2299809609138763,
      "grad_norm": 1.1420440673828125,
      "learning_rate": 9.478815789473685e-05,
      "loss": 0.4037,
      "step": 39823
    },
    {
      "epoch": 2.2300369582260053,
      "grad_norm": 1.2136605978012085,
      "learning_rate": 9.478789473684211e-05,
      "loss": 0.3921,
      "step": 39824
    },
    {
      "epoch": 2.2300929555381344,
      "grad_norm": 1.9451848268508911,
      "learning_rate": 9.478763157894738e-05,
      "loss": 0.5273,
      "step": 39825
    },
    {
      "epoch": 2.2301489528502634,
      "grad_norm": 1.184277892112732,
      "learning_rate": 9.478736842105263e-05,
      "loss": 0.31,
      "step": 39826
    },
    {
      "epoch": 2.2302049501623924,
      "grad_norm": 1.0472304821014404,
      "learning_rate": 9.478710526315789e-05,
      "loss": 0.4147,
      "step": 39827
    },
    {
      "epoch": 2.2302609474745214,
      "grad_norm": 1.1181080341339111,
      "learning_rate": 9.478684210526316e-05,
      "loss": 0.398,
      "step": 39828
    },
    {
      "epoch": 2.2303169447866504,
      "grad_norm": 1.3620513677597046,
      "learning_rate": 9.478657894736842e-05,
      "loss": 0.4324,
      "step": 39829
    },
    {
      "epoch": 2.2303729420987795,
      "grad_norm": 1.1015474796295166,
      "learning_rate": 9.47863157894737e-05,
      "loss": 0.4513,
      "step": 39830
    },
    {
      "epoch": 2.2304289394109085,
      "grad_norm": 1.3489550352096558,
      "learning_rate": 9.478605263157894e-05,
      "loss": 0.3804,
      "step": 39831
    },
    {
      "epoch": 2.2304849367230375,
      "grad_norm": 1.2574396133422852,
      "learning_rate": 9.478578947368422e-05,
      "loss": 0.3538,
      "step": 39832
    },
    {
      "epoch": 2.2305409340351665,
      "grad_norm": 1.5626978874206543,
      "learning_rate": 9.478552631578948e-05,
      "loss": 0.3698,
      "step": 39833
    },
    {
      "epoch": 2.2305969313472955,
      "grad_norm": 1.0399703979492188,
      "learning_rate": 9.478526315789475e-05,
      "loss": 0.4223,
      "step": 39834
    },
    {
      "epoch": 2.2306529286594245,
      "grad_norm": 1.1375476121902466,
      "learning_rate": 9.478500000000001e-05,
      "loss": 0.5115,
      "step": 39835
    },
    {
      "epoch": 2.2307089259715536,
      "grad_norm": 1.6256465911865234,
      "learning_rate": 9.478473684210527e-05,
      "loss": 0.4987,
      "step": 39836
    },
    {
      "epoch": 2.2307649232836826,
      "grad_norm": 1.2412623167037964,
      "learning_rate": 9.478447368421053e-05,
      "loss": 0.3802,
      "step": 39837
    },
    {
      "epoch": 2.2308209205958116,
      "grad_norm": 1.166092038154602,
      "learning_rate": 9.47842105263158e-05,
      "loss": 0.4695,
      "step": 39838
    },
    {
      "epoch": 2.2308769179079406,
      "grad_norm": 1.1605539321899414,
      "learning_rate": 9.478394736842106e-05,
      "loss": 0.4198,
      "step": 39839
    },
    {
      "epoch": 2.2309329152200696,
      "grad_norm": 1.2228286266326904,
      "learning_rate": 9.478368421052632e-05,
      "loss": 0.4252,
      "step": 39840
    },
    {
      "epoch": 2.230988912532198,
      "grad_norm": 0.9917532801628113,
      "learning_rate": 9.478342105263158e-05,
      "loss": 0.3579,
      "step": 39841
    },
    {
      "epoch": 2.2310449098443277,
      "grad_norm": 1.3003849983215332,
      "learning_rate": 9.478315789473685e-05,
      "loss": 0.3199,
      "step": 39842
    },
    {
      "epoch": 2.2311009071564563,
      "grad_norm": 1.5952361822128296,
      "learning_rate": 9.478289473684211e-05,
      "loss": 0.3744,
      "step": 39843
    },
    {
      "epoch": 2.2311569044685857,
      "grad_norm": 1.2072569131851196,
      "learning_rate": 9.478263157894737e-05,
      "loss": 0.5561,
      "step": 39844
    },
    {
      "epoch": 2.2312129017807143,
      "grad_norm": 1.322372555732727,
      "learning_rate": 9.478236842105263e-05,
      "loss": 0.4333,
      "step": 39845
    },
    {
      "epoch": 2.2312688990928438,
      "grad_norm": 1.1622461080551147,
      "learning_rate": 9.478210526315789e-05,
      "loss": 0.4416,
      "step": 39846
    },
    {
      "epoch": 2.2313248964049723,
      "grad_norm": 20.153928756713867,
      "learning_rate": 9.478184210526317e-05,
      "loss": 0.3336,
      "step": 39847
    },
    {
      "epoch": 2.231380893717102,
      "grad_norm": 1.6961157321929932,
      "learning_rate": 9.478157894736843e-05,
      "loss": 0.4616,
      "step": 39848
    },
    {
      "epoch": 2.2314368910292304,
      "grad_norm": 1.068688988685608,
      "learning_rate": 9.478131578947368e-05,
      "loss": 0.4314,
      "step": 39849
    },
    {
      "epoch": 2.23149288834136,
      "grad_norm": 1.40049147605896,
      "learning_rate": 9.478105263157894e-05,
      "loss": 0.3476,
      "step": 39850
    },
    {
      "epoch": 2.2315488856534884,
      "grad_norm": 1.318068504333496,
      "learning_rate": 9.478078947368422e-05,
      "loss": 0.381,
      "step": 39851
    },
    {
      "epoch": 2.231604882965618,
      "grad_norm": 1.2210620641708374,
      "learning_rate": 9.478052631578948e-05,
      "loss": 0.3842,
      "step": 39852
    },
    {
      "epoch": 2.2316608802777465,
      "grad_norm": 1.0520250797271729,
      "learning_rate": 9.478026315789474e-05,
      "loss": 0.3699,
      "step": 39853
    },
    {
      "epoch": 2.231716877589876,
      "grad_norm": 1.6190332174301147,
      "learning_rate": 9.478e-05,
      "loss": 0.4258,
      "step": 39854
    },
    {
      "epoch": 2.2317728749020045,
      "grad_norm": 1.3141025304794312,
      "learning_rate": 9.477973684210527e-05,
      "loss": 0.3708,
      "step": 39855
    },
    {
      "epoch": 2.231828872214134,
      "grad_norm": 1.2058528661727905,
      "learning_rate": 9.477947368421053e-05,
      "loss": 0.3888,
      "step": 39856
    },
    {
      "epoch": 2.2318848695262625,
      "grad_norm": 1.110175371170044,
      "learning_rate": 9.47792105263158e-05,
      "loss": 0.3429,
      "step": 39857
    },
    {
      "epoch": 2.231940866838392,
      "grad_norm": 1.2571953535079956,
      "learning_rate": 9.477894736842105e-05,
      "loss": 0.2947,
      "step": 39858
    },
    {
      "epoch": 2.2319968641505206,
      "grad_norm": 1.224549651145935,
      "learning_rate": 9.477868421052632e-05,
      "loss": 0.348,
      "step": 39859
    },
    {
      "epoch": 2.23205286146265,
      "grad_norm": 1.143818736076355,
      "learning_rate": 9.477842105263158e-05,
      "loss": 0.4194,
      "step": 39860
    },
    {
      "epoch": 2.2321088587747786,
      "grad_norm": 1.4943702220916748,
      "learning_rate": 9.477815789473686e-05,
      "loss": 0.4468,
      "step": 39861
    },
    {
      "epoch": 2.2321648560869076,
      "grad_norm": 1.168671727180481,
      "learning_rate": 9.477789473684212e-05,
      "loss": 0.4272,
      "step": 39862
    },
    {
      "epoch": 2.2322208533990366,
      "grad_norm": 1.1281013488769531,
      "learning_rate": 9.477763157894736e-05,
      "loss": 0.4344,
      "step": 39863
    },
    {
      "epoch": 2.2322768507111657,
      "grad_norm": 1.409969687461853,
      "learning_rate": 9.477736842105264e-05,
      "loss": 0.4032,
      "step": 39864
    },
    {
      "epoch": 2.2323328480232947,
      "grad_norm": 1.2034969329833984,
      "learning_rate": 9.47771052631579e-05,
      "loss": 0.4199,
      "step": 39865
    },
    {
      "epoch": 2.2323888453354237,
      "grad_norm": 1.4730056524276733,
      "learning_rate": 9.477684210526317e-05,
      "loss": 0.5229,
      "step": 39866
    },
    {
      "epoch": 2.2324448426475527,
      "grad_norm": 1.0847047567367554,
      "learning_rate": 9.477657894736843e-05,
      "loss": 0.4607,
      "step": 39867
    },
    {
      "epoch": 2.2325008399596817,
      "grad_norm": 1.590936303138733,
      "learning_rate": 9.477631578947369e-05,
      "loss": 0.4402,
      "step": 39868
    },
    {
      "epoch": 2.2325568372718108,
      "grad_norm": 1.1979938745498657,
      "learning_rate": 9.477605263157895e-05,
      "loss": 0.4177,
      "step": 39869
    },
    {
      "epoch": 2.2326128345839398,
      "grad_norm": 1.2597712278366089,
      "learning_rate": 9.477578947368422e-05,
      "loss": 0.5459,
      "step": 39870
    },
    {
      "epoch": 2.232668831896069,
      "grad_norm": 1.1838299036026,
      "learning_rate": 9.477552631578948e-05,
      "loss": 0.3987,
      "step": 39871
    },
    {
      "epoch": 2.232724829208198,
      "grad_norm": 1.2514230012893677,
      "learning_rate": 9.477526315789474e-05,
      "loss": 0.3958,
      "step": 39872
    },
    {
      "epoch": 2.232780826520327,
      "grad_norm": 1.5621144771575928,
      "learning_rate": 9.4775e-05,
      "loss": 0.3974,
      "step": 39873
    },
    {
      "epoch": 2.232836823832456,
      "grad_norm": 1.7287871837615967,
      "learning_rate": 9.477473684210527e-05,
      "loss": 0.5622,
      "step": 39874
    },
    {
      "epoch": 2.232892821144585,
      "grad_norm": 1.0921645164489746,
      "learning_rate": 9.477447368421053e-05,
      "loss": 0.337,
      "step": 39875
    },
    {
      "epoch": 2.232948818456714,
      "grad_norm": 1.237606406211853,
      "learning_rate": 9.477421052631579e-05,
      "loss": 0.4316,
      "step": 39876
    },
    {
      "epoch": 2.233004815768843,
      "grad_norm": 1.2418818473815918,
      "learning_rate": 9.477394736842105e-05,
      "loss": 0.4315,
      "step": 39877
    },
    {
      "epoch": 2.233060813080972,
      "grad_norm": 1.1087077856063843,
      "learning_rate": 9.477368421052633e-05,
      "loss": 0.3977,
      "step": 39878
    },
    {
      "epoch": 2.233116810393101,
      "grad_norm": 118.37350463867188,
      "learning_rate": 9.477342105263159e-05,
      "loss": 0.5059,
      "step": 39879
    },
    {
      "epoch": 2.23317280770523,
      "grad_norm": 1.2595902681350708,
      "learning_rate": 9.477315789473684e-05,
      "loss": 0.5157,
      "step": 39880
    },
    {
      "epoch": 2.233228805017359,
      "grad_norm": 1.1393039226531982,
      "learning_rate": 9.47728947368421e-05,
      "loss": 0.5149,
      "step": 39881
    },
    {
      "epoch": 2.233284802329488,
      "grad_norm": 1.2310459613800049,
      "learning_rate": 9.477263157894736e-05,
      "loss": 0.3994,
      "step": 39882
    },
    {
      "epoch": 2.233340799641617,
      "grad_norm": 1.1022762060165405,
      "learning_rate": 9.477236842105264e-05,
      "loss": 0.348,
      "step": 39883
    },
    {
      "epoch": 2.233396796953746,
      "grad_norm": 1.0758891105651855,
      "learning_rate": 9.47721052631579e-05,
      "loss": 0.4134,
      "step": 39884
    },
    {
      "epoch": 2.233452794265875,
      "grad_norm": 1.349063515663147,
      "learning_rate": 9.477184210526317e-05,
      "loss": 0.6077,
      "step": 39885
    },
    {
      "epoch": 2.233508791578004,
      "grad_norm": 1.756942868232727,
      "learning_rate": 9.477157894736842e-05,
      "loss": 0.5718,
      "step": 39886
    },
    {
      "epoch": 2.233564788890133,
      "grad_norm": 1.3244483470916748,
      "learning_rate": 9.477131578947369e-05,
      "loss": 0.3469,
      "step": 39887
    },
    {
      "epoch": 2.233620786202262,
      "grad_norm": 1.1796387434005737,
      "learning_rate": 9.477105263157895e-05,
      "loss": 0.457,
      "step": 39888
    },
    {
      "epoch": 2.233676783514391,
      "grad_norm": 1.3100197315216064,
      "learning_rate": 9.477078947368422e-05,
      "loss": 0.4546,
      "step": 39889
    },
    {
      "epoch": 2.23373278082652,
      "grad_norm": 1.0772711038589478,
      "learning_rate": 9.477052631578948e-05,
      "loss": 0.312,
      "step": 39890
    },
    {
      "epoch": 2.233788778138649,
      "grad_norm": 1.1480755805969238,
      "learning_rate": 9.477026315789474e-05,
      "loss": 0.3906,
      "step": 39891
    },
    {
      "epoch": 2.233844775450778,
      "grad_norm": 1.4124020338058472,
      "learning_rate": 9.477e-05,
      "loss": 0.3352,
      "step": 39892
    },
    {
      "epoch": 2.233900772762907,
      "grad_norm": 1.0873157978057861,
      "learning_rate": 9.476973684210528e-05,
      "loss": 0.4241,
      "step": 39893
    },
    {
      "epoch": 2.2339567700750362,
      "grad_norm": 1.2109360694885254,
      "learning_rate": 9.476947368421054e-05,
      "loss": 0.3599,
      "step": 39894
    },
    {
      "epoch": 2.2340127673871653,
      "grad_norm": 1.368592619895935,
      "learning_rate": 9.47692105263158e-05,
      "loss": 0.4567,
      "step": 39895
    },
    {
      "epoch": 2.2340687646992943,
      "grad_norm": 1.1569875478744507,
      "learning_rate": 9.476894736842105e-05,
      "loss": 0.3024,
      "step": 39896
    },
    {
      "epoch": 2.2341247620114233,
      "grad_norm": 1.054959774017334,
      "learning_rate": 9.476868421052631e-05,
      "loss": 0.417,
      "step": 39897
    },
    {
      "epoch": 2.2341807593235523,
      "grad_norm": 1.1425422430038452,
      "learning_rate": 9.476842105263159e-05,
      "loss": 0.3577,
      "step": 39898
    },
    {
      "epoch": 2.2342367566356813,
      "grad_norm": 1.1021727323532104,
      "learning_rate": 9.476815789473685e-05,
      "loss": 0.4438,
      "step": 39899
    },
    {
      "epoch": 2.2342927539478104,
      "grad_norm": 1.489829182624817,
      "learning_rate": 9.476789473684211e-05,
      "loss": 0.4864,
      "step": 39900
    },
    {
      "epoch": 2.2343487512599394,
      "grad_norm": 1.2514277696609497,
      "learning_rate": 9.476763157894737e-05,
      "loss": 0.3668,
      "step": 39901
    },
    {
      "epoch": 2.2344047485720684,
      "grad_norm": 1.1762288808822632,
      "learning_rate": 9.476736842105264e-05,
      "loss": 0.47,
      "step": 39902
    },
    {
      "epoch": 2.2344607458841974,
      "grad_norm": 1.1313693523406982,
      "learning_rate": 9.47671052631579e-05,
      "loss": 0.3328,
      "step": 39903
    },
    {
      "epoch": 2.2345167431963264,
      "grad_norm": 1.2157329320907593,
      "learning_rate": 9.476684210526316e-05,
      "loss": 0.5045,
      "step": 39904
    },
    {
      "epoch": 2.2345727405084554,
      "grad_norm": 1.1704727411270142,
      "learning_rate": 9.476657894736842e-05,
      "loss": 0.3563,
      "step": 39905
    },
    {
      "epoch": 2.2346287378205845,
      "grad_norm": 1.3883517980575562,
      "learning_rate": 9.476631578947369e-05,
      "loss": 0.3895,
      "step": 39906
    },
    {
      "epoch": 2.2346847351327135,
      "grad_norm": 1.3662335872650146,
      "learning_rate": 9.476605263157895e-05,
      "loss": 0.5314,
      "step": 39907
    },
    {
      "epoch": 2.2347407324448425,
      "grad_norm": 1.518214464187622,
      "learning_rate": 9.476578947368421e-05,
      "loss": 0.4034,
      "step": 39908
    },
    {
      "epoch": 2.2347967297569715,
      "grad_norm": 1.1782102584838867,
      "learning_rate": 9.476552631578947e-05,
      "loss": 0.3626,
      "step": 39909
    },
    {
      "epoch": 2.2348527270691005,
      "grad_norm": 1.2837440967559814,
      "learning_rate": 9.476526315789475e-05,
      "loss": 0.4021,
      "step": 39910
    },
    {
      "epoch": 2.2349087243812296,
      "grad_norm": 1.225203037261963,
      "learning_rate": 9.4765e-05,
      "loss": 0.4332,
      "step": 39911
    },
    {
      "epoch": 2.2349647216933586,
      "grad_norm": 0.9944750070571899,
      "learning_rate": 9.476473684210528e-05,
      "loss": 0.3155,
      "step": 39912
    },
    {
      "epoch": 2.2350207190054876,
      "grad_norm": 1.1196249723434448,
      "learning_rate": 9.476447368421052e-05,
      "loss": 0.4272,
      "step": 39913
    },
    {
      "epoch": 2.2350767163176166,
      "grad_norm": 1.8444879055023193,
      "learning_rate": 9.476421052631578e-05,
      "loss": 0.6322,
      "step": 39914
    },
    {
      "epoch": 2.2351327136297456,
      "grad_norm": 1.1092880964279175,
      "learning_rate": 9.476394736842106e-05,
      "loss": 0.3613,
      "step": 39915
    },
    {
      "epoch": 2.2351887109418747,
      "grad_norm": 1.4292500019073486,
      "learning_rate": 9.476368421052632e-05,
      "loss": 0.4643,
      "step": 39916
    },
    {
      "epoch": 2.2352447082540037,
      "grad_norm": 1.431752324104309,
      "learning_rate": 9.476342105263159e-05,
      "loss": 0.4556,
      "step": 39917
    },
    {
      "epoch": 2.2353007055661327,
      "grad_norm": 1.2649872303009033,
      "learning_rate": 9.476315789473684e-05,
      "loss": 0.4774,
      "step": 39918
    },
    {
      "epoch": 2.2353567028782617,
      "grad_norm": 1.07740318775177,
      "learning_rate": 9.476289473684211e-05,
      "loss": 0.4053,
      "step": 39919
    },
    {
      "epoch": 2.2354127001903907,
      "grad_norm": 1.1882028579711914,
      "learning_rate": 9.476263157894737e-05,
      "loss": 0.3765,
      "step": 39920
    },
    {
      "epoch": 2.2354686975025198,
      "grad_norm": 1.1409432888031006,
      "learning_rate": 9.476236842105264e-05,
      "loss": 0.5393,
      "step": 39921
    },
    {
      "epoch": 2.2355246948146488,
      "grad_norm": 1.3845298290252686,
      "learning_rate": 9.47621052631579e-05,
      "loss": 0.3877,
      "step": 39922
    },
    {
      "epoch": 2.235580692126778,
      "grad_norm": 1.367463231086731,
      "learning_rate": 9.476184210526316e-05,
      "loss": 0.331,
      "step": 39923
    },
    {
      "epoch": 2.235636689438907,
      "grad_norm": 1.0793558359146118,
      "learning_rate": 9.476157894736842e-05,
      "loss": 0.3899,
      "step": 39924
    },
    {
      "epoch": 2.235692686751036,
      "grad_norm": 1.2754768133163452,
      "learning_rate": 9.47613157894737e-05,
      "loss": 0.3708,
      "step": 39925
    },
    {
      "epoch": 2.235748684063165,
      "grad_norm": 1.3791537284851074,
      "learning_rate": 9.476105263157896e-05,
      "loss": 0.6432,
      "step": 39926
    },
    {
      "epoch": 2.235804681375294,
      "grad_norm": 0.948826014995575,
      "learning_rate": 9.476078947368421e-05,
      "loss": 0.3876,
      "step": 39927
    },
    {
      "epoch": 2.235860678687423,
      "grad_norm": 1.3538728952407837,
      "learning_rate": 9.476052631578947e-05,
      "loss": 0.3793,
      "step": 39928
    },
    {
      "epoch": 2.235916675999552,
      "grad_norm": 1.072240948677063,
      "learning_rate": 9.476026315789475e-05,
      "loss": 0.413,
      "step": 39929
    },
    {
      "epoch": 2.235972673311681,
      "grad_norm": 1.527756929397583,
      "learning_rate": 9.476000000000001e-05,
      "loss": 0.3499,
      "step": 39930
    },
    {
      "epoch": 2.23602867062381,
      "grad_norm": 1.3422818183898926,
      "learning_rate": 9.475973684210527e-05,
      "loss": 0.346,
      "step": 39931
    },
    {
      "epoch": 2.236084667935939,
      "grad_norm": 1.402057409286499,
      "learning_rate": 9.475947368421053e-05,
      "loss": 0.4758,
      "step": 39932
    },
    {
      "epoch": 2.236140665248068,
      "grad_norm": 1.0964053869247437,
      "learning_rate": 9.475921052631579e-05,
      "loss": 0.3429,
      "step": 39933
    },
    {
      "epoch": 2.236196662560197,
      "grad_norm": 1.1753878593444824,
      "learning_rate": 9.475894736842106e-05,
      "loss": 0.4079,
      "step": 39934
    },
    {
      "epoch": 2.236252659872326,
      "grad_norm": 1.103694200515747,
      "learning_rate": 9.475868421052632e-05,
      "loss": 0.3765,
      "step": 39935
    },
    {
      "epoch": 2.236308657184455,
      "grad_norm": 1.4737013578414917,
      "learning_rate": 9.475842105263158e-05,
      "loss": 0.4378,
      "step": 39936
    },
    {
      "epoch": 2.236364654496584,
      "grad_norm": 1.2695138454437256,
      "learning_rate": 9.475815789473684e-05,
      "loss": 0.3272,
      "step": 39937
    },
    {
      "epoch": 2.236420651808713,
      "grad_norm": 1.065146803855896,
      "learning_rate": 9.475789473684211e-05,
      "loss": 0.3768,
      "step": 39938
    },
    {
      "epoch": 2.236476649120842,
      "grad_norm": 1.373531699180603,
      "learning_rate": 9.475763157894737e-05,
      "loss": 0.5587,
      "step": 39939
    },
    {
      "epoch": 2.236532646432971,
      "grad_norm": 1.2843073606491089,
      "learning_rate": 9.475736842105265e-05,
      "loss": 0.5004,
      "step": 39940
    },
    {
      "epoch": 2.2365886437451,
      "grad_norm": 1.2023910284042358,
      "learning_rate": 9.475710526315789e-05,
      "loss": 0.2787,
      "step": 39941
    },
    {
      "epoch": 2.236644641057229,
      "grad_norm": 1.529686689376831,
      "learning_rate": 9.475684210526316e-05,
      "loss": 0.5828,
      "step": 39942
    },
    {
      "epoch": 2.236700638369358,
      "grad_norm": 1.269740104675293,
      "learning_rate": 9.475657894736842e-05,
      "loss": 0.4057,
      "step": 39943
    },
    {
      "epoch": 2.236756635681487,
      "grad_norm": 1.2110581398010254,
      "learning_rate": 9.47563157894737e-05,
      "loss": 0.3941,
      "step": 39944
    },
    {
      "epoch": 2.236812632993616,
      "grad_norm": 1.1980048418045044,
      "learning_rate": 9.475605263157896e-05,
      "loss": 0.433,
      "step": 39945
    },
    {
      "epoch": 2.2368686303057452,
      "grad_norm": 1.1548186540603638,
      "learning_rate": 9.475578947368422e-05,
      "loss": 0.3879,
      "step": 39946
    },
    {
      "epoch": 2.2369246276178743,
      "grad_norm": 1.1750766038894653,
      "learning_rate": 9.475552631578948e-05,
      "loss": 0.4583,
      "step": 39947
    },
    {
      "epoch": 2.2369806249300033,
      "grad_norm": 1.3085960149765015,
      "learning_rate": 9.475526315789474e-05,
      "loss": 0.4658,
      "step": 39948
    },
    {
      "epoch": 2.2370366222421323,
      "grad_norm": Infinity,
      "learning_rate": 9.475526315789474e-05,
      "loss": 0.4228,
      "step": 39949
    },
    {
      "epoch": 2.2370926195542613,
      "grad_norm": 1.4823559522628784,
      "learning_rate": 9.475500000000001e-05,
      "loss": 0.2996,
      "step": 39950
    },
    {
      "epoch": 2.2371486168663903,
      "grad_norm": 1.205166220664978,
      "learning_rate": 9.475473684210526e-05,
      "loss": 0.3887,
      "step": 39951
    },
    {
      "epoch": 2.2372046141785193,
      "grad_norm": 1.1702510118484497,
      "learning_rate": 9.475447368421053e-05,
      "loss": 0.351,
      "step": 39952
    },
    {
      "epoch": 2.2372606114906484,
      "grad_norm": 1.0614465475082397,
      "learning_rate": 9.475421052631579e-05,
      "loss": 0.393,
      "step": 39953
    },
    {
      "epoch": 2.2373166088027774,
      "grad_norm": 1.2083666324615479,
      "learning_rate": 9.475394736842106e-05,
      "loss": 0.5644,
      "step": 39954
    },
    {
      "epoch": 2.2373726061149064,
      "grad_norm": 1.180579662322998,
      "learning_rate": 9.475368421052632e-05,
      "loss": 0.3806,
      "step": 39955
    },
    {
      "epoch": 2.2374286034270354,
      "grad_norm": 1.1403300762176514,
      "learning_rate": 9.475342105263158e-05,
      "loss": 0.4456,
      "step": 39956
    },
    {
      "epoch": 2.2374846007391644,
      "grad_norm": 1.1023598909378052,
      "learning_rate": 9.475315789473684e-05,
      "loss": 0.4709,
      "step": 39957
    },
    {
      "epoch": 2.2375405980512935,
      "grad_norm": 1.2078845500946045,
      "learning_rate": 9.475289473684212e-05,
      "loss": 0.3839,
      "step": 39958
    },
    {
      "epoch": 2.2375965953634225,
      "grad_norm": 1.3416873216629028,
      "learning_rate": 9.475263157894737e-05,
      "loss": 0.3914,
      "step": 39959
    },
    {
      "epoch": 2.2376525926755515,
      "grad_norm": 1.1283265352249146,
      "learning_rate": 9.475236842105263e-05,
      "loss": 0.5546,
      "step": 39960
    },
    {
      "epoch": 2.2377085899876805,
      "grad_norm": 1.2295030355453491,
      "learning_rate": 9.47521052631579e-05,
      "loss": 0.516,
      "step": 39961
    },
    {
      "epoch": 2.2377645872998095,
      "grad_norm": 1.1981478929519653,
      "learning_rate": 9.475184210526317e-05,
      "loss": 0.4325,
      "step": 39962
    },
    {
      "epoch": 2.2378205846119386,
      "grad_norm": 0.8648694753646851,
      "learning_rate": 9.475157894736843e-05,
      "loss": 0.2625,
      "step": 39963
    },
    {
      "epoch": 2.2378765819240676,
      "grad_norm": 1.5383596420288086,
      "learning_rate": 9.475131578947369e-05,
      "loss": 0.3375,
      "step": 39964
    },
    {
      "epoch": 2.2379325792361966,
      "grad_norm": 1.240010380744934,
      "learning_rate": 9.475105263157895e-05,
      "loss": 0.5168,
      "step": 39965
    },
    {
      "epoch": 2.2379885765483256,
      "grad_norm": 1.1895915269851685,
      "learning_rate": 9.475078947368422e-05,
      "loss": 0.4048,
      "step": 39966
    },
    {
      "epoch": 2.2380445738604546,
      "grad_norm": 1.3029874563217163,
      "learning_rate": 9.475052631578948e-05,
      "loss": 0.4319,
      "step": 39967
    },
    {
      "epoch": 2.2381005711725837,
      "grad_norm": 1.290453314781189,
      "learning_rate": 9.475026315789474e-05,
      "loss": 0.3719,
      "step": 39968
    },
    {
      "epoch": 2.2381565684847127,
      "grad_norm": 1.1601901054382324,
      "learning_rate": 9.475e-05,
      "loss": 0.2659,
      "step": 39969
    },
    {
      "epoch": 2.2382125657968417,
      "grad_norm": 1.4407151937484741,
      "learning_rate": 9.474973684210526e-05,
      "loss": 0.5851,
      "step": 39970
    },
    {
      "epoch": 2.2382685631089707,
      "grad_norm": 1.149189829826355,
      "learning_rate": 9.474947368421053e-05,
      "loss": 0.5252,
      "step": 39971
    },
    {
      "epoch": 2.2383245604210997,
      "grad_norm": 1.588834524154663,
      "learning_rate": 9.474921052631579e-05,
      "loss": 0.3904,
      "step": 39972
    },
    {
      "epoch": 2.2383805577332287,
      "grad_norm": 1.182956576347351,
      "learning_rate": 9.474894736842107e-05,
      "loss": 0.3434,
      "step": 39973
    },
    {
      "epoch": 2.2384365550453578,
      "grad_norm": 1.3318060636520386,
      "learning_rate": 9.474868421052631e-05,
      "loss": 0.6055,
      "step": 39974
    },
    {
      "epoch": 2.238492552357487,
      "grad_norm": 1.1993985176086426,
      "learning_rate": 9.474842105263158e-05,
      "loss": 0.4704,
      "step": 39975
    },
    {
      "epoch": 2.238548549669616,
      "grad_norm": 1.2488253116607666,
      "learning_rate": 9.474815789473684e-05,
      "loss": 0.408,
      "step": 39976
    },
    {
      "epoch": 2.238604546981745,
      "grad_norm": 1.151971697807312,
      "learning_rate": 9.474789473684212e-05,
      "loss": 0.3721,
      "step": 39977
    },
    {
      "epoch": 2.238660544293874,
      "grad_norm": 1.1755130290985107,
      "learning_rate": 9.474763157894738e-05,
      "loss": 0.4729,
      "step": 39978
    },
    {
      "epoch": 2.238716541606003,
      "grad_norm": 1.3706068992614746,
      "learning_rate": 9.474736842105264e-05,
      "loss": 0.3167,
      "step": 39979
    },
    {
      "epoch": 2.238772538918132,
      "grad_norm": 1.2356736660003662,
      "learning_rate": 9.47471052631579e-05,
      "loss": 0.3977,
      "step": 39980
    },
    {
      "epoch": 2.238828536230261,
      "grad_norm": 1.4225574731826782,
      "learning_rate": 9.474684210526317e-05,
      "loss": 0.3663,
      "step": 39981
    },
    {
      "epoch": 2.23888453354239,
      "grad_norm": 1.1724600791931152,
      "learning_rate": 9.474657894736843e-05,
      "loss": 0.4472,
      "step": 39982
    },
    {
      "epoch": 2.238940530854519,
      "grad_norm": 1.2882767915725708,
      "learning_rate": 9.474631578947369e-05,
      "loss": 0.4334,
      "step": 39983
    },
    {
      "epoch": 2.238996528166648,
      "grad_norm": 2.4172704219818115,
      "learning_rate": 9.474605263157895e-05,
      "loss": 0.3863,
      "step": 39984
    },
    {
      "epoch": 2.239052525478777,
      "grad_norm": 1.551350474357605,
      "learning_rate": 9.474578947368421e-05,
      "loss": 0.4781,
      "step": 39985
    },
    {
      "epoch": 2.239108522790906,
      "grad_norm": 1.3170713186264038,
      "learning_rate": 9.474552631578948e-05,
      "loss": 0.4631,
      "step": 39986
    },
    {
      "epoch": 2.239164520103035,
      "grad_norm": 1.2197309732437134,
      "learning_rate": 9.474526315789474e-05,
      "loss": 0.4882,
      "step": 39987
    },
    {
      "epoch": 2.239220517415164,
      "grad_norm": 1.1837671995162964,
      "learning_rate": 9.4745e-05,
      "loss": 0.3603,
      "step": 39988
    },
    {
      "epoch": 2.239276514727293,
      "grad_norm": 1.223314881324768,
      "learning_rate": 9.474473684210526e-05,
      "loss": 0.3685,
      "step": 39989
    },
    {
      "epoch": 2.239332512039422,
      "grad_norm": 2.763145685195923,
      "learning_rate": 9.474447368421053e-05,
      "loss": 0.4145,
      "step": 39990
    },
    {
      "epoch": 2.239388509351551,
      "grad_norm": 1.1128594875335693,
      "learning_rate": 9.47442105263158e-05,
      "loss": 0.4004,
      "step": 39991
    },
    {
      "epoch": 2.23944450666368,
      "grad_norm": 1.231052279472351,
      "learning_rate": 9.474394736842105e-05,
      "loss": 0.474,
      "step": 39992
    },
    {
      "epoch": 2.239500503975809,
      "grad_norm": 1.0675373077392578,
      "learning_rate": 9.474368421052631e-05,
      "loss": 0.4333,
      "step": 39993
    },
    {
      "epoch": 2.239556501287938,
      "grad_norm": 1.4970506429672241,
      "learning_rate": 9.474342105263159e-05,
      "loss": 0.5533,
      "step": 39994
    },
    {
      "epoch": 2.239612498600067,
      "grad_norm": 1.4095773696899414,
      "learning_rate": 9.474315789473685e-05,
      "loss": 0.3557,
      "step": 39995
    },
    {
      "epoch": 2.239668495912196,
      "grad_norm": 1.0520011186599731,
      "learning_rate": 9.474289473684212e-05,
      "loss": 0.2956,
      "step": 39996
    },
    {
      "epoch": 2.239724493224325,
      "grad_norm": 1.2317395210266113,
      "learning_rate": 9.474263157894737e-05,
      "loss": 0.3938,
      "step": 39997
    },
    {
      "epoch": 2.2397804905364542,
      "grad_norm": 2.2235002517700195,
      "learning_rate": 9.474236842105264e-05,
      "loss": 0.4064,
      "step": 39998
    },
    {
      "epoch": 2.2398364878485832,
      "grad_norm": 1.1575438976287842,
      "learning_rate": 9.47421052631579e-05,
      "loss": 0.4056,
      "step": 39999
    },
    {
      "epoch": 2.2398924851607123,
      "grad_norm": 1.2392680644989014,
      "learning_rate": 9.474184210526317e-05,
      "loss": 0.4986,
      "step": 40000
    },
    {
      "epoch": 2.2399484824728413,
      "grad_norm": 0.9851133823394775,
      "learning_rate": 9.474157894736843e-05,
      "loss": 0.3659,
      "step": 40001
    },
    {
      "epoch": 2.2400044797849703,
      "grad_norm": 1.1650434732437134,
      "learning_rate": 9.474131578947368e-05,
      "loss": 0.3928,
      "step": 40002
    },
    {
      "epoch": 2.2400604770970993,
      "grad_norm": 1.099534511566162,
      "learning_rate": 9.474105263157895e-05,
      "loss": 0.3755,
      "step": 40003
    },
    {
      "epoch": 2.2401164744092283,
      "grad_norm": 1.2335864305496216,
      "learning_rate": 9.474078947368421e-05,
      "loss": 0.5132,
      "step": 40004
    },
    {
      "epoch": 2.2401724717213574,
      "grad_norm": 1.2972909212112427,
      "learning_rate": 9.474052631578948e-05,
      "loss": 0.3832,
      "step": 40005
    },
    {
      "epoch": 2.2402284690334864,
      "grad_norm": 1.3466465473175049,
      "learning_rate": 9.474026315789473e-05,
      "loss": 0.3817,
      "step": 40006
    },
    {
      "epoch": 2.2402844663456154,
      "grad_norm": 1.375908613204956,
      "learning_rate": 9.474e-05,
      "loss": 0.3668,
      "step": 40007
    },
    {
      "epoch": 2.2403404636577444,
      "grad_norm": 1.1419041156768799,
      "learning_rate": 9.473973684210526e-05,
      "loss": 0.3017,
      "step": 40008
    },
    {
      "epoch": 2.2403964609698734,
      "grad_norm": 1.1640428304672241,
      "learning_rate": 9.473947368421054e-05,
      "loss": 0.3712,
      "step": 40009
    },
    {
      "epoch": 2.2404524582820025,
      "grad_norm": 1.5585756301879883,
      "learning_rate": 9.47392105263158e-05,
      "loss": 0.473,
      "step": 40010
    },
    {
      "epoch": 2.2405084555941315,
      "grad_norm": 1.0717350244522095,
      "learning_rate": 9.473894736842106e-05,
      "loss": 0.366,
      "step": 40011
    },
    {
      "epoch": 2.2405644529062605,
      "grad_norm": 1.302870512008667,
      "learning_rate": 9.473868421052632e-05,
      "loss": 0.4169,
      "step": 40012
    },
    {
      "epoch": 2.2406204502183895,
      "grad_norm": 1.2220220565795898,
      "learning_rate": 9.473842105263159e-05,
      "loss": 0.3958,
      "step": 40013
    },
    {
      "epoch": 2.2406764475305185,
      "grad_norm": 1.4140089750289917,
      "learning_rate": 9.473815789473685e-05,
      "loss": 0.4479,
      "step": 40014
    },
    {
      "epoch": 2.2407324448426476,
      "grad_norm": 1.7141547203063965,
      "learning_rate": 9.473789473684211e-05,
      "loss": 0.5872,
      "step": 40015
    },
    {
      "epoch": 2.2407884421547766,
      "grad_norm": 1.467725157737732,
      "learning_rate": 9.473763157894737e-05,
      "loss": 0.3752,
      "step": 40016
    },
    {
      "epoch": 2.2408444394669056,
      "grad_norm": 1.1830037832260132,
      "learning_rate": 9.473736842105264e-05,
      "loss": 0.447,
      "step": 40017
    },
    {
      "epoch": 2.2409004367790346,
      "grad_norm": 1.0790154933929443,
      "learning_rate": 9.47371052631579e-05,
      "loss": 0.4001,
      "step": 40018
    },
    {
      "epoch": 2.2409564340911636,
      "grad_norm": 1.1032518148422241,
      "learning_rate": 9.473684210526316e-05,
      "loss": 0.4083,
      "step": 40019
    },
    {
      "epoch": 2.2410124314032926,
      "grad_norm": 1.2645224332809448,
      "learning_rate": 9.473657894736842e-05,
      "loss": 0.4652,
      "step": 40020
    },
    {
      "epoch": 2.2410684287154217,
      "grad_norm": 1.3790812492370605,
      "learning_rate": 9.473631578947368e-05,
      "loss": 0.3428,
      "step": 40021
    },
    {
      "epoch": 2.2411244260275507,
      "grad_norm": 1.375135898590088,
      "learning_rate": 9.473605263157895e-05,
      "loss": 0.3707,
      "step": 40022
    },
    {
      "epoch": 2.2411804233396797,
      "grad_norm": 1.0942384004592896,
      "learning_rate": 9.473578947368421e-05,
      "loss": 0.3139,
      "step": 40023
    },
    {
      "epoch": 2.2412364206518087,
      "grad_norm": 1.4653470516204834,
      "learning_rate": 9.473552631578947e-05,
      "loss": 0.4462,
      "step": 40024
    },
    {
      "epoch": 2.2412924179639377,
      "grad_norm": 1.3072495460510254,
      "learning_rate": 9.473526315789473e-05,
      "loss": 0.4028,
      "step": 40025
    },
    {
      "epoch": 2.2413484152760668,
      "grad_norm": 1.0115150213241577,
      "learning_rate": 9.473500000000001e-05,
      "loss": 0.2975,
      "step": 40026
    },
    {
      "epoch": 2.241404412588196,
      "grad_norm": 1.0322843790054321,
      "learning_rate": 9.473473684210527e-05,
      "loss": 0.2533,
      "step": 40027
    },
    {
      "epoch": 2.241460409900325,
      "grad_norm": 1.1061087846755981,
      "learning_rate": 9.473447368421054e-05,
      "loss": 0.3353,
      "step": 40028
    },
    {
      "epoch": 2.241516407212454,
      "grad_norm": 1.1929560899734497,
      "learning_rate": 9.473421052631579e-05,
      "loss": 0.2924,
      "step": 40029
    },
    {
      "epoch": 2.241572404524583,
      "grad_norm": 1.4066029787063599,
      "learning_rate": 9.473394736842106e-05,
      "loss": 0.401,
      "step": 40030
    },
    {
      "epoch": 2.241628401836712,
      "grad_norm": 1.2358721494674683,
      "learning_rate": 9.473368421052632e-05,
      "loss": 0.4111,
      "step": 40031
    },
    {
      "epoch": 2.241684399148841,
      "grad_norm": 1.1237452030181885,
      "learning_rate": 9.473342105263159e-05,
      "loss": 0.4697,
      "step": 40032
    },
    {
      "epoch": 2.24174039646097,
      "grad_norm": 1.0696967840194702,
      "learning_rate": 9.473315789473685e-05,
      "loss": 0.3536,
      "step": 40033
    },
    {
      "epoch": 2.241796393773099,
      "grad_norm": 1.179937481880188,
      "learning_rate": 9.473289473684211e-05,
      "loss": 0.3543,
      "step": 40034
    },
    {
      "epoch": 2.241852391085228,
      "grad_norm": 1.2309807538986206,
      "learning_rate": 9.473263157894737e-05,
      "loss": 0.3823,
      "step": 40035
    },
    {
      "epoch": 2.241908388397357,
      "grad_norm": 2.0140416622161865,
      "learning_rate": 9.473236842105263e-05,
      "loss": 0.5147,
      "step": 40036
    },
    {
      "epoch": 2.241964385709486,
      "grad_norm": 1.1582443714141846,
      "learning_rate": 9.47321052631579e-05,
      "loss": 0.3891,
      "step": 40037
    },
    {
      "epoch": 2.242020383021615,
      "grad_norm": 1.3058408498764038,
      "learning_rate": 9.473184210526316e-05,
      "loss": 0.4588,
      "step": 40038
    },
    {
      "epoch": 2.242076380333744,
      "grad_norm": 1.4526749849319458,
      "learning_rate": 9.473157894736842e-05,
      "loss": 0.5216,
      "step": 40039
    },
    {
      "epoch": 2.242132377645873,
      "grad_norm": 1.1712088584899902,
      "learning_rate": 9.473131578947368e-05,
      "loss": 0.3896,
      "step": 40040
    },
    {
      "epoch": 2.242188374958002,
      "grad_norm": 1.3635497093200684,
      "learning_rate": 9.473105263157896e-05,
      "loss": 0.3978,
      "step": 40041
    },
    {
      "epoch": 2.242244372270131,
      "grad_norm": 1.1317417621612549,
      "learning_rate": 9.473078947368422e-05,
      "loss": 0.3616,
      "step": 40042
    },
    {
      "epoch": 2.24230036958226,
      "grad_norm": 1.096992015838623,
      "learning_rate": 9.473052631578948e-05,
      "loss": 0.412,
      "step": 40043
    },
    {
      "epoch": 2.242356366894389,
      "grad_norm": 1.5048686265945435,
      "learning_rate": 9.473026315789474e-05,
      "loss": 0.3999,
      "step": 40044
    },
    {
      "epoch": 2.242412364206518,
      "grad_norm": 1.1228364706039429,
      "learning_rate": 9.473000000000001e-05,
      "loss": 0.3503,
      "step": 40045
    },
    {
      "epoch": 2.242468361518647,
      "grad_norm": 1.1628942489624023,
      "learning_rate": 9.472973684210527e-05,
      "loss": 0.403,
      "step": 40046
    },
    {
      "epoch": 2.242524358830776,
      "grad_norm": 1.1514170169830322,
      "learning_rate": 9.472947368421053e-05,
      "loss": 0.3823,
      "step": 40047
    },
    {
      "epoch": 2.242580356142905,
      "grad_norm": 1.3031609058380127,
      "learning_rate": 9.472921052631579e-05,
      "loss": 0.4264,
      "step": 40048
    },
    {
      "epoch": 2.242636353455034,
      "grad_norm": 1.132583498954773,
      "learning_rate": 9.472894736842106e-05,
      "loss": 0.4152,
      "step": 40049
    },
    {
      "epoch": 2.2426923507671632,
      "grad_norm": 1.7693240642547607,
      "learning_rate": 9.472868421052632e-05,
      "loss": 0.387,
      "step": 40050
    },
    {
      "epoch": 2.2427483480792922,
      "grad_norm": 7.094480991363525,
      "learning_rate": 9.47284210526316e-05,
      "loss": 0.4944,
      "step": 40051
    },
    {
      "epoch": 2.2428043453914213,
      "grad_norm": 1.3027894496917725,
      "learning_rate": 9.472815789473684e-05,
      "loss": 0.387,
      "step": 40052
    },
    {
      "epoch": 2.2428603427035503,
      "grad_norm": 1.1837313175201416,
      "learning_rate": 9.47278947368421e-05,
      "loss": 0.3046,
      "step": 40053
    },
    {
      "epoch": 2.2429163400156793,
      "grad_norm": 1.136681318283081,
      "learning_rate": 9.472763157894737e-05,
      "loss": 0.378,
      "step": 40054
    },
    {
      "epoch": 2.2429723373278083,
      "grad_norm": 1.2850608825683594,
      "learning_rate": 9.472736842105263e-05,
      "loss": 0.3835,
      "step": 40055
    },
    {
      "epoch": 2.2430283346399373,
      "grad_norm": 1.178913950920105,
      "learning_rate": 9.47271052631579e-05,
      "loss": 0.4344,
      "step": 40056
    },
    {
      "epoch": 2.2430843319520664,
      "grad_norm": 1.373615026473999,
      "learning_rate": 9.472684210526315e-05,
      "loss": 0.4389,
      "step": 40057
    },
    {
      "epoch": 2.2431403292641954,
      "grad_norm": 1.1545320749282837,
      "learning_rate": 9.472657894736843e-05,
      "loss": 0.3852,
      "step": 40058
    },
    {
      "epoch": 2.2431963265763244,
      "grad_norm": 1.3827202320098877,
      "learning_rate": 9.472631578947369e-05,
      "loss": 0.3811,
      "step": 40059
    },
    {
      "epoch": 2.2432523238884534,
      "grad_norm": 1.240789532661438,
      "learning_rate": 9.472605263157896e-05,
      "loss": 0.3543,
      "step": 40060
    },
    {
      "epoch": 2.2433083212005824,
      "grad_norm": 1.1640691757202148,
      "learning_rate": 9.47257894736842e-05,
      "loss": 0.4291,
      "step": 40061
    },
    {
      "epoch": 2.2433643185127115,
      "grad_norm": 1.270259141921997,
      "learning_rate": 9.472552631578948e-05,
      "loss": 0.3339,
      "step": 40062
    },
    {
      "epoch": 2.2434203158248405,
      "grad_norm": 1.1392079591751099,
      "learning_rate": 9.472526315789474e-05,
      "loss": 0.3278,
      "step": 40063
    },
    {
      "epoch": 2.2434763131369695,
      "grad_norm": 1.4896596670150757,
      "learning_rate": 9.472500000000001e-05,
      "loss": 0.4646,
      "step": 40064
    },
    {
      "epoch": 2.2435323104490985,
      "grad_norm": 1.2585833072662354,
      "learning_rate": 9.472473684210527e-05,
      "loss": 0.4276,
      "step": 40065
    },
    {
      "epoch": 2.2435883077612275,
      "grad_norm": 1.1571033000946045,
      "learning_rate": 9.472447368421053e-05,
      "loss": 0.4073,
      "step": 40066
    },
    {
      "epoch": 2.2436443050733565,
      "grad_norm": 1.4096288681030273,
      "learning_rate": 9.472421052631579e-05,
      "loss": 0.4808,
      "step": 40067
    },
    {
      "epoch": 2.2437003023854856,
      "grad_norm": 1.2043886184692383,
      "learning_rate": 9.472394736842106e-05,
      "loss": 0.3484,
      "step": 40068
    },
    {
      "epoch": 2.2437562996976146,
      "grad_norm": 1.0547136068344116,
      "learning_rate": 9.472368421052632e-05,
      "loss": 0.2978,
      "step": 40069
    },
    {
      "epoch": 2.2438122970097436,
      "grad_norm": 1.116759181022644,
      "learning_rate": 9.472342105263158e-05,
      "loss": 0.3285,
      "step": 40070
    },
    {
      "epoch": 2.2438682943218726,
      "grad_norm": 1.1092449426651,
      "learning_rate": 9.472315789473684e-05,
      "loss": 0.3875,
      "step": 40071
    },
    {
      "epoch": 2.2439242916340016,
      "grad_norm": 1.1607588529586792,
      "learning_rate": 9.47228947368421e-05,
      "loss": 0.4198,
      "step": 40072
    },
    {
      "epoch": 2.2439802889461307,
      "grad_norm": 1.1589411497116089,
      "learning_rate": 9.472263157894738e-05,
      "loss": 0.382,
      "step": 40073
    },
    {
      "epoch": 2.2440362862582597,
      "grad_norm": 1.0209678411483765,
      "learning_rate": 9.472236842105264e-05,
      "loss": 0.446,
      "step": 40074
    },
    {
      "epoch": 2.2440922835703887,
      "grad_norm": 1.3072453737258911,
      "learning_rate": 9.47221052631579e-05,
      "loss": 0.4451,
      "step": 40075
    },
    {
      "epoch": 2.2441482808825177,
      "grad_norm": 0.9194907546043396,
      "learning_rate": 9.472184210526316e-05,
      "loss": 0.3515,
      "step": 40076
    },
    {
      "epoch": 2.2442042781946467,
      "grad_norm": 1.2015539407730103,
      "learning_rate": 9.472157894736843e-05,
      "loss": 0.5375,
      "step": 40077
    },
    {
      "epoch": 2.2442602755067758,
      "grad_norm": 1.1443856954574585,
      "learning_rate": 9.472131578947369e-05,
      "loss": 0.3863,
      "step": 40078
    },
    {
      "epoch": 2.244316272818905,
      "grad_norm": 1.0491079092025757,
      "learning_rate": 9.472105263157895e-05,
      "loss": 0.336,
      "step": 40079
    },
    {
      "epoch": 2.244372270131034,
      "grad_norm": 1.2625688314437866,
      "learning_rate": 9.472078947368421e-05,
      "loss": 0.4045,
      "step": 40080
    },
    {
      "epoch": 2.244428267443163,
      "grad_norm": 1.3699678182601929,
      "learning_rate": 9.472052631578948e-05,
      "loss": 0.451,
      "step": 40081
    },
    {
      "epoch": 2.244484264755292,
      "grad_norm": 1.2473140954971313,
      "learning_rate": 9.472026315789474e-05,
      "loss": 0.306,
      "step": 40082
    },
    {
      "epoch": 2.244540262067421,
      "grad_norm": 1.1685445308685303,
      "learning_rate": 9.472000000000001e-05,
      "loss": 0.3856,
      "step": 40083
    },
    {
      "epoch": 2.24459625937955,
      "grad_norm": 0.9222745895385742,
      "learning_rate": 9.471973684210526e-05,
      "loss": 0.3257,
      "step": 40084
    },
    {
      "epoch": 2.244652256691679,
      "grad_norm": 1.3740324974060059,
      "learning_rate": 9.471947368421053e-05,
      "loss": 0.4357,
      "step": 40085
    },
    {
      "epoch": 2.244708254003808,
      "grad_norm": 1.3621305227279663,
      "learning_rate": 9.47192105263158e-05,
      "loss": 0.4403,
      "step": 40086
    },
    {
      "epoch": 2.244764251315937,
      "grad_norm": 1.295389175415039,
      "learning_rate": 9.471894736842107e-05,
      "loss": 0.3244,
      "step": 40087
    },
    {
      "epoch": 2.244820248628066,
      "grad_norm": 1.4892868995666504,
      "learning_rate": 9.471868421052633e-05,
      "loss": 0.3576,
      "step": 40088
    },
    {
      "epoch": 2.244876245940195,
      "grad_norm": 1.1565486192703247,
      "learning_rate": 9.471842105263157e-05,
      "loss": 0.3935,
      "step": 40089
    },
    {
      "epoch": 2.244932243252324,
      "grad_norm": 1.5260311365127563,
      "learning_rate": 9.471815789473685e-05,
      "loss": 0.4058,
      "step": 40090
    },
    {
      "epoch": 2.244988240564453,
      "grad_norm": 1.0575448274612427,
      "learning_rate": 9.47178947368421e-05,
      "loss": 0.3802,
      "step": 40091
    },
    {
      "epoch": 2.245044237876582,
      "grad_norm": 1.2589242458343506,
      "learning_rate": 9.471763157894738e-05,
      "loss": 0.3704,
      "step": 40092
    },
    {
      "epoch": 2.245100235188711,
      "grad_norm": 1.0562865734100342,
      "learning_rate": 9.471736842105264e-05,
      "loss": 0.336,
      "step": 40093
    },
    {
      "epoch": 2.24515623250084,
      "grad_norm": 1.2664527893066406,
      "learning_rate": 9.47171052631579e-05,
      "loss": 0.5396,
      "step": 40094
    },
    {
      "epoch": 2.245212229812969,
      "grad_norm": 1.2666444778442383,
      "learning_rate": 9.471684210526316e-05,
      "loss": 0.4644,
      "step": 40095
    },
    {
      "epoch": 2.245268227125098,
      "grad_norm": 1.139235496520996,
      "learning_rate": 9.471657894736843e-05,
      "loss": 0.3456,
      "step": 40096
    },
    {
      "epoch": 2.245324224437227,
      "grad_norm": 1.319272756576538,
      "learning_rate": 9.471631578947369e-05,
      "loss": 0.5157,
      "step": 40097
    },
    {
      "epoch": 2.245380221749356,
      "grad_norm": 1.3000242710113525,
      "learning_rate": 9.471605263157895e-05,
      "loss": 0.4559,
      "step": 40098
    },
    {
      "epoch": 2.245436219061485,
      "grad_norm": 1.203012228012085,
      "learning_rate": 9.471578947368421e-05,
      "loss": 0.3529,
      "step": 40099
    },
    {
      "epoch": 2.245492216373614,
      "grad_norm": 1.2658575773239136,
      "learning_rate": 9.471552631578948e-05,
      "loss": 0.4043,
      "step": 40100
    },
    {
      "epoch": 2.245548213685743,
      "grad_norm": 1.316209077835083,
      "learning_rate": 9.471526315789474e-05,
      "loss": 0.4743,
      "step": 40101
    },
    {
      "epoch": 2.245604210997872,
      "grad_norm": 0.9757815599441528,
      "learning_rate": 9.4715e-05,
      "loss": 0.3781,
      "step": 40102
    },
    {
      "epoch": 2.2456602083100012,
      "grad_norm": 1.2216956615447998,
      "learning_rate": 9.471473684210526e-05,
      "loss": 0.4249,
      "step": 40103
    },
    {
      "epoch": 2.2457162056221303,
      "grad_norm": 1.0190409421920776,
      "learning_rate": 9.471447368421054e-05,
      "loss": 0.3108,
      "step": 40104
    },
    {
      "epoch": 2.2457722029342593,
      "grad_norm": 1.2755038738250732,
      "learning_rate": 9.47142105263158e-05,
      "loss": 0.4462,
      "step": 40105
    },
    {
      "epoch": 2.2458282002463883,
      "grad_norm": 1.097569465637207,
      "learning_rate": 9.471394736842106e-05,
      "loss": 0.3708,
      "step": 40106
    },
    {
      "epoch": 2.2458841975585173,
      "grad_norm": 1.544690489768982,
      "learning_rate": 9.471368421052632e-05,
      "loss": 0.7099,
      "step": 40107
    },
    {
      "epoch": 2.2459401948706463,
      "grad_norm": 1.197120189666748,
      "learning_rate": 9.471342105263158e-05,
      "loss": 0.4852,
      "step": 40108
    },
    {
      "epoch": 2.2459961921827754,
      "grad_norm": 1.1146851778030396,
      "learning_rate": 9.471315789473685e-05,
      "loss": 0.3941,
      "step": 40109
    },
    {
      "epoch": 2.2460521894949044,
      "grad_norm": 1.3033556938171387,
      "learning_rate": 9.471289473684211e-05,
      "loss": 0.4565,
      "step": 40110
    },
    {
      "epoch": 2.2461081868070334,
      "grad_norm": 1.4440224170684814,
      "learning_rate": 9.471263157894737e-05,
      "loss": 0.4238,
      "step": 40111
    },
    {
      "epoch": 2.2461641841191624,
      "grad_norm": 1.2164455652236938,
      "learning_rate": 9.471236842105263e-05,
      "loss": 0.5021,
      "step": 40112
    },
    {
      "epoch": 2.2462201814312914,
      "grad_norm": 1.3465259075164795,
      "learning_rate": 9.47121052631579e-05,
      "loss": 0.3254,
      "step": 40113
    },
    {
      "epoch": 2.2462761787434204,
      "grad_norm": 1.0940874814987183,
      "learning_rate": 9.471184210526316e-05,
      "loss": 0.3525,
      "step": 40114
    },
    {
      "epoch": 2.2463321760555495,
      "grad_norm": 1.3678523302078247,
      "learning_rate": 9.471157894736843e-05,
      "loss": 0.3569,
      "step": 40115
    },
    {
      "epoch": 2.2463881733676785,
      "grad_norm": 1.0850518941879272,
      "learning_rate": 9.471131578947368e-05,
      "loss": 0.4202,
      "step": 40116
    },
    {
      "epoch": 2.2464441706798075,
      "grad_norm": 1.2148479223251343,
      "learning_rate": 9.471105263157895e-05,
      "loss": 0.2912,
      "step": 40117
    },
    {
      "epoch": 2.2465001679919365,
      "grad_norm": 1.4207943677902222,
      "learning_rate": 9.471078947368421e-05,
      "loss": 0.5088,
      "step": 40118
    },
    {
      "epoch": 2.2465561653040655,
      "grad_norm": 1.3358690738677979,
      "learning_rate": 9.471052631578949e-05,
      "loss": 0.3653,
      "step": 40119
    },
    {
      "epoch": 2.2466121626161946,
      "grad_norm": 1.0093083381652832,
      "learning_rate": 9.471026315789475e-05,
      "loss": 0.3814,
      "step": 40120
    },
    {
      "epoch": 2.2466681599283236,
      "grad_norm": 2.148082733154297,
      "learning_rate": 9.471e-05,
      "loss": 0.5028,
      "step": 40121
    },
    {
      "epoch": 2.2467241572404526,
      "grad_norm": 1.4454413652420044,
      "learning_rate": 9.470973684210527e-05,
      "loss": 0.376,
      "step": 40122
    },
    {
      "epoch": 2.2467801545525816,
      "grad_norm": 1.3570654392242432,
      "learning_rate": 9.470947368421053e-05,
      "loss": 0.3734,
      "step": 40123
    },
    {
      "epoch": 2.2468361518647106,
      "grad_norm": 1.3323249816894531,
      "learning_rate": 9.47092105263158e-05,
      "loss": 0.4194,
      "step": 40124
    },
    {
      "epoch": 2.2468921491768397,
      "grad_norm": 1.6291946172714233,
      "learning_rate": 9.470894736842106e-05,
      "loss": 0.468,
      "step": 40125
    },
    {
      "epoch": 2.2469481464889687,
      "grad_norm": 1.167348861694336,
      "learning_rate": 9.470868421052632e-05,
      "loss": 0.3023,
      "step": 40126
    },
    {
      "epoch": 2.2470041438010977,
      "grad_norm": 1.0483983755111694,
      "learning_rate": 9.470842105263158e-05,
      "loss": 0.329,
      "step": 40127
    },
    {
      "epoch": 2.2470601411132267,
      "grad_norm": 1.2004703283309937,
      "learning_rate": 9.470815789473685e-05,
      "loss": 0.4033,
      "step": 40128
    },
    {
      "epoch": 2.2471161384253557,
      "grad_norm": 1.2403535842895508,
      "learning_rate": 9.470789473684211e-05,
      "loss": 0.3976,
      "step": 40129
    },
    {
      "epoch": 2.2471721357374848,
      "grad_norm": 1.2348365783691406,
      "learning_rate": 9.470763157894737e-05,
      "loss": 0.3982,
      "step": 40130
    },
    {
      "epoch": 2.2472281330496138,
      "grad_norm": 1.0892412662506104,
      "learning_rate": 9.470736842105263e-05,
      "loss": 0.3602,
      "step": 40131
    },
    {
      "epoch": 2.247284130361743,
      "grad_norm": 1.4445762634277344,
      "learning_rate": 9.47071052631579e-05,
      "loss": 0.3874,
      "step": 40132
    },
    {
      "epoch": 2.247340127673872,
      "grad_norm": 1.0186032056808472,
      "learning_rate": 9.470684210526316e-05,
      "loss": 0.358,
      "step": 40133
    },
    {
      "epoch": 2.247396124986001,
      "grad_norm": 0.9626522660255432,
      "learning_rate": 9.470657894736842e-05,
      "loss": 0.2884,
      "step": 40134
    },
    {
      "epoch": 2.24745212229813,
      "grad_norm": 1.0658365488052368,
      "learning_rate": 9.470631578947368e-05,
      "loss": 0.4058,
      "step": 40135
    },
    {
      "epoch": 2.247508119610259,
      "grad_norm": 1.147383689880371,
      "learning_rate": 9.470605263157896e-05,
      "loss": 0.5104,
      "step": 40136
    },
    {
      "epoch": 2.247564116922388,
      "grad_norm": 1.7567428350448608,
      "learning_rate": 9.470578947368422e-05,
      "loss": 0.4483,
      "step": 40137
    },
    {
      "epoch": 2.247620114234517,
      "grad_norm": 1.270355224609375,
      "learning_rate": 9.470552631578949e-05,
      "loss": 0.3691,
      "step": 40138
    },
    {
      "epoch": 2.247676111546646,
      "grad_norm": 1.3380851745605469,
      "learning_rate": 9.470526315789474e-05,
      "loss": 0.3938,
      "step": 40139
    },
    {
      "epoch": 2.247732108858775,
      "grad_norm": 1.293134093284607,
      "learning_rate": 9.4705e-05,
      "loss": 0.4256,
      "step": 40140
    },
    {
      "epoch": 2.247788106170904,
      "grad_norm": 1.433640718460083,
      "learning_rate": 9.470473684210527e-05,
      "loss": 0.5922,
      "step": 40141
    },
    {
      "epoch": 2.247844103483033,
      "grad_norm": 1.2602661848068237,
      "learning_rate": 9.470447368421053e-05,
      "loss": 0.4579,
      "step": 40142
    },
    {
      "epoch": 2.247900100795162,
      "grad_norm": 1.115237832069397,
      "learning_rate": 9.47042105263158e-05,
      "loss": 0.3688,
      "step": 40143
    },
    {
      "epoch": 2.247956098107291,
      "grad_norm": 1.0808993577957153,
      "learning_rate": 9.470394736842105e-05,
      "loss": 0.5651,
      "step": 40144
    },
    {
      "epoch": 2.24801209541942,
      "grad_norm": 1.1818220615386963,
      "learning_rate": 9.470368421052632e-05,
      "loss": 0.3476,
      "step": 40145
    },
    {
      "epoch": 2.248068092731549,
      "grad_norm": 1.3889023065567017,
      "learning_rate": 9.470342105263158e-05,
      "loss": 0.3479,
      "step": 40146
    },
    {
      "epoch": 2.248124090043678,
      "grad_norm": 1.410182237625122,
      "learning_rate": 9.470315789473685e-05,
      "loss": 0.5013,
      "step": 40147
    },
    {
      "epoch": 2.248180087355807,
      "grad_norm": 1.351880669593811,
      "learning_rate": 9.470289473684211e-05,
      "loss": 0.5259,
      "step": 40148
    },
    {
      "epoch": 2.248236084667936,
      "grad_norm": 1.4977388381958008,
      "learning_rate": 9.470263157894737e-05,
      "loss": 0.485,
      "step": 40149
    },
    {
      "epoch": 2.248292081980065,
      "grad_norm": 1.0728261470794678,
      "learning_rate": 9.470236842105263e-05,
      "loss": 0.3275,
      "step": 40150
    },
    {
      "epoch": 2.248348079292194,
      "grad_norm": 1.4816807508468628,
      "learning_rate": 9.47021052631579e-05,
      "loss": 0.4071,
      "step": 40151
    },
    {
      "epoch": 2.248404076604323,
      "grad_norm": 1.1847679615020752,
      "learning_rate": 9.470184210526317e-05,
      "loss": 0.5818,
      "step": 40152
    },
    {
      "epoch": 2.248460073916452,
      "grad_norm": 1.2820178270339966,
      "learning_rate": 9.470157894736843e-05,
      "loss": 0.4254,
      "step": 40153
    },
    {
      "epoch": 2.248516071228581,
      "grad_norm": 1.3149336576461792,
      "learning_rate": 9.470131578947369e-05,
      "loss": 0.505,
      "step": 40154
    },
    {
      "epoch": 2.2485720685407102,
      "grad_norm": 1.2017433643341064,
      "learning_rate": 9.470105263157896e-05,
      "loss": 0.2538,
      "step": 40155
    },
    {
      "epoch": 2.2486280658528393,
      "grad_norm": 1.2277536392211914,
      "learning_rate": 9.470078947368422e-05,
      "loss": 0.4682,
      "step": 40156
    },
    {
      "epoch": 2.2486840631649683,
      "grad_norm": 1.5223044157028198,
      "learning_rate": 9.470052631578948e-05,
      "loss": 0.4615,
      "step": 40157
    },
    {
      "epoch": 2.2487400604770973,
      "grad_norm": 1.10268235206604,
      "learning_rate": 9.470026315789474e-05,
      "loss": 0.4101,
      "step": 40158
    },
    {
      "epoch": 2.2487960577892263,
      "grad_norm": 1.0358903408050537,
      "learning_rate": 9.47e-05,
      "loss": 0.3913,
      "step": 40159
    },
    {
      "epoch": 2.2488520551013553,
      "grad_norm": 1.1859701871871948,
      "learning_rate": 9.469973684210527e-05,
      "loss": 0.4007,
      "step": 40160
    },
    {
      "epoch": 2.2489080524134843,
      "grad_norm": 1.2227067947387695,
      "learning_rate": 9.469947368421053e-05,
      "loss": 0.4533,
      "step": 40161
    },
    {
      "epoch": 2.2489640497256134,
      "grad_norm": 1.1936038732528687,
      "learning_rate": 9.469921052631579e-05,
      "loss": 0.4481,
      "step": 40162
    },
    {
      "epoch": 2.2490200470377424,
      "grad_norm": 1.214200496673584,
      "learning_rate": 9.469894736842105e-05,
      "loss": 0.3699,
      "step": 40163
    },
    {
      "epoch": 2.2490760443498714,
      "grad_norm": 1.3284720182418823,
      "learning_rate": 9.469868421052632e-05,
      "loss": 0.3362,
      "step": 40164
    },
    {
      "epoch": 2.2491320416620004,
      "grad_norm": 1.2022281885147095,
      "learning_rate": 9.469842105263158e-05,
      "loss": 0.4207,
      "step": 40165
    },
    {
      "epoch": 2.2491880389741294,
      "grad_norm": 1.1746233701705933,
      "learning_rate": 9.469815789473684e-05,
      "loss": 0.4223,
      "step": 40166
    },
    {
      "epoch": 2.2492440362862585,
      "grad_norm": 1.3932158946990967,
      "learning_rate": 9.46978947368421e-05,
      "loss": 0.3824,
      "step": 40167
    },
    {
      "epoch": 2.2493000335983875,
      "grad_norm": 1.443080186843872,
      "learning_rate": 9.469763157894738e-05,
      "loss": 0.3932,
      "step": 40168
    },
    {
      "epoch": 2.2493560309105165,
      "grad_norm": 1.0938193798065186,
      "learning_rate": 9.469736842105264e-05,
      "loss": 0.3954,
      "step": 40169
    },
    {
      "epoch": 2.2494120282226455,
      "grad_norm": 1.5580129623413086,
      "learning_rate": 9.469710526315791e-05,
      "loss": 0.5555,
      "step": 40170
    },
    {
      "epoch": 2.2494680255347745,
      "grad_norm": 1.2123533487319946,
      "learning_rate": 9.469684210526316e-05,
      "loss": 0.4642,
      "step": 40171
    },
    {
      "epoch": 2.249524022846903,
      "grad_norm": 1.0812416076660156,
      "learning_rate": 9.469657894736843e-05,
      "loss": 0.3178,
      "step": 40172
    },
    {
      "epoch": 2.2495800201590326,
      "grad_norm": 1.3480777740478516,
      "learning_rate": 9.469631578947369e-05,
      "loss": 0.4252,
      "step": 40173
    },
    {
      "epoch": 2.249636017471161,
      "grad_norm": 1.2388789653778076,
      "learning_rate": 9.469605263157895e-05,
      "loss": 0.3595,
      "step": 40174
    },
    {
      "epoch": 2.2496920147832906,
      "grad_norm": 2.250688076019287,
      "learning_rate": 9.469578947368422e-05,
      "loss": 0.3539,
      "step": 40175
    },
    {
      "epoch": 2.249748012095419,
      "grad_norm": 1.377152919769287,
      "learning_rate": 9.469552631578947e-05,
      "loss": 0.3855,
      "step": 40176
    },
    {
      "epoch": 2.2498040094075487,
      "grad_norm": 1.3823412656784058,
      "learning_rate": 9.469526315789474e-05,
      "loss": 0.4807,
      "step": 40177
    },
    {
      "epoch": 2.2498600067196772,
      "grad_norm": 1.090590238571167,
      "learning_rate": 9.4695e-05,
      "loss": 0.3301,
      "step": 40178
    },
    {
      "epoch": 2.2499160040318067,
      "grad_norm": 1.5525349378585815,
      "learning_rate": 9.469473684210527e-05,
      "loss": 0.6467,
      "step": 40179
    },
    {
      "epoch": 2.2499720013439353,
      "grad_norm": 1.1669665575027466,
      "learning_rate": 9.469447368421053e-05,
      "loss": 0.3872,
      "step": 40180
    },
    {
      "epoch": 2.2500279986560647,
      "grad_norm": 1.456902027130127,
      "learning_rate": 9.46942105263158e-05,
      "loss": 0.3851,
      "step": 40181
    },
    {
      "epoch": 2.2500839959681933,
      "grad_norm": 1.1647282838821411,
      "learning_rate": 9.469394736842105e-05,
      "loss": 0.4387,
      "step": 40182
    },
    {
      "epoch": 2.2501399932803228,
      "grad_norm": 1.1139322519302368,
      "learning_rate": 9.469368421052633e-05,
      "loss": 0.3383,
      "step": 40183
    },
    {
      "epoch": 2.2501959905924513,
      "grad_norm": 1.3475444316864014,
      "learning_rate": 9.469342105263159e-05,
      "loss": 0.3843,
      "step": 40184
    },
    {
      "epoch": 2.250251987904581,
      "grad_norm": 1.1799030303955078,
      "learning_rate": 9.469315789473685e-05,
      "loss": 0.429,
      "step": 40185
    },
    {
      "epoch": 2.2503079852167094,
      "grad_norm": 1.2678601741790771,
      "learning_rate": 9.46928947368421e-05,
      "loss": 0.4107,
      "step": 40186
    },
    {
      "epoch": 2.250363982528839,
      "grad_norm": 1.0544410943984985,
      "learning_rate": 9.469263157894738e-05,
      "loss": 0.3521,
      "step": 40187
    },
    {
      "epoch": 2.2504199798409674,
      "grad_norm": 1.2537574768066406,
      "learning_rate": 9.469236842105264e-05,
      "loss": 0.3666,
      "step": 40188
    },
    {
      "epoch": 2.250475977153097,
      "grad_norm": 1.1284570693969727,
      "learning_rate": 9.46921052631579e-05,
      "loss": 0.3437,
      "step": 40189
    },
    {
      "epoch": 2.2505319744652255,
      "grad_norm": 2.3447911739349365,
      "learning_rate": 9.469184210526316e-05,
      "loss": 0.4281,
      "step": 40190
    },
    {
      "epoch": 2.250587971777355,
      "grad_norm": 1.790476679801941,
      "learning_rate": 9.469157894736842e-05,
      "loss": 0.5354,
      "step": 40191
    },
    {
      "epoch": 2.2506439690894835,
      "grad_norm": 1.093313455581665,
      "learning_rate": 9.469131578947369e-05,
      "loss": 0.4061,
      "step": 40192
    },
    {
      "epoch": 2.250699966401613,
      "grad_norm": 1.2129058837890625,
      "learning_rate": 9.469105263157895e-05,
      "loss": 0.3507,
      "step": 40193
    },
    {
      "epoch": 2.2507559637137415,
      "grad_norm": 3.002373456954956,
      "learning_rate": 9.469078947368421e-05,
      "loss": 0.4381,
      "step": 40194
    },
    {
      "epoch": 2.250811961025871,
      "grad_norm": 1.1771005392074585,
      "learning_rate": 9.469052631578947e-05,
      "loss": 0.3605,
      "step": 40195
    },
    {
      "epoch": 2.2508679583379996,
      "grad_norm": 1.1493512392044067,
      "learning_rate": 9.469026315789474e-05,
      "loss": 0.4266,
      "step": 40196
    },
    {
      "epoch": 2.250923955650129,
      "grad_norm": 1.136483907699585,
      "learning_rate": 9.469e-05,
      "loss": 0.4038,
      "step": 40197
    },
    {
      "epoch": 2.2509799529622576,
      "grad_norm": 1.11309814453125,
      "learning_rate": 9.468973684210528e-05,
      "loss": 0.359,
      "step": 40198
    },
    {
      "epoch": 2.251035950274387,
      "grad_norm": 1.3423651456832886,
      "learning_rate": 9.468947368421052e-05,
      "loss": 0.5096,
      "step": 40199
    },
    {
      "epoch": 2.2510919475865157,
      "grad_norm": 1.347076177597046,
      "learning_rate": 9.46892105263158e-05,
      "loss": 0.6856,
      "step": 40200
    },
    {
      "epoch": 2.251147944898645,
      "grad_norm": 1.193207859992981,
      "learning_rate": 9.468894736842106e-05,
      "loss": 0.4806,
      "step": 40201
    },
    {
      "epoch": 2.2512039422107737,
      "grad_norm": 1.0027910470962524,
      "learning_rate": 9.468868421052633e-05,
      "loss": 0.3848,
      "step": 40202
    },
    {
      "epoch": 2.2512599395229027,
      "grad_norm": 1.05929434299469,
      "learning_rate": 9.468842105263158e-05,
      "loss": 0.3007,
      "step": 40203
    },
    {
      "epoch": 2.2513159368350317,
      "grad_norm": 1.8180017471313477,
      "learning_rate": 9.468815789473685e-05,
      "loss": 0.4107,
      "step": 40204
    },
    {
      "epoch": 2.2513719341471607,
      "grad_norm": 1.191176176071167,
      "learning_rate": 9.468789473684211e-05,
      "loss": 0.3798,
      "step": 40205
    },
    {
      "epoch": 2.2514279314592898,
      "grad_norm": 1.7255076169967651,
      "learning_rate": 9.468763157894738e-05,
      "loss": 0.3711,
      "step": 40206
    },
    {
      "epoch": 2.251483928771419,
      "grad_norm": 1.1258769035339355,
      "learning_rate": 9.468736842105264e-05,
      "loss": 0.4427,
      "step": 40207
    },
    {
      "epoch": 2.251539926083548,
      "grad_norm": 1.1882961988449097,
      "learning_rate": 9.46871052631579e-05,
      "loss": 0.4297,
      "step": 40208
    },
    {
      "epoch": 2.251595923395677,
      "grad_norm": 1.151963710784912,
      "learning_rate": 9.468684210526316e-05,
      "loss": 0.4622,
      "step": 40209
    },
    {
      "epoch": 2.251651920707806,
      "grad_norm": 1.1382732391357422,
      "learning_rate": 9.468657894736842e-05,
      "loss": 0.425,
      "step": 40210
    },
    {
      "epoch": 2.251707918019935,
      "grad_norm": 1.2088879346847534,
      "learning_rate": 9.46863157894737e-05,
      "loss": 0.4749,
      "step": 40211
    },
    {
      "epoch": 2.251763915332064,
      "grad_norm": 1.764775276184082,
      "learning_rate": 9.468605263157895e-05,
      "loss": 0.6663,
      "step": 40212
    },
    {
      "epoch": 2.251819912644193,
      "grad_norm": 1.2178705930709839,
      "learning_rate": 9.468578947368421e-05,
      "loss": 0.3944,
      "step": 40213
    },
    {
      "epoch": 2.251875909956322,
      "grad_norm": 1.1462750434875488,
      "learning_rate": 9.468552631578947e-05,
      "loss": 0.3637,
      "step": 40214
    },
    {
      "epoch": 2.251931907268451,
      "grad_norm": 1.1129882335662842,
      "learning_rate": 9.468526315789475e-05,
      "loss": 0.3065,
      "step": 40215
    },
    {
      "epoch": 2.25198790458058,
      "grad_norm": 1.1299878358840942,
      "learning_rate": 9.4685e-05,
      "loss": 0.3973,
      "step": 40216
    },
    {
      "epoch": 2.252043901892709,
      "grad_norm": 1.2472347021102905,
      "learning_rate": 9.468473684210527e-05,
      "loss": 0.4698,
      "step": 40217
    },
    {
      "epoch": 2.252099899204838,
      "grad_norm": 1.0947084426879883,
      "learning_rate": 9.468447368421053e-05,
      "loss": 0.3481,
      "step": 40218
    },
    {
      "epoch": 2.252155896516967,
      "grad_norm": 1.1644936800003052,
      "learning_rate": 9.46842105263158e-05,
      "loss": 0.4058,
      "step": 40219
    },
    {
      "epoch": 2.252211893829096,
      "grad_norm": 1.2407244443893433,
      "learning_rate": 9.468394736842106e-05,
      "loss": 0.2998,
      "step": 40220
    },
    {
      "epoch": 2.252267891141225,
      "grad_norm": 1.096522569656372,
      "learning_rate": 9.468368421052632e-05,
      "loss": 0.3788,
      "step": 40221
    },
    {
      "epoch": 2.252323888453354,
      "grad_norm": 1.0371482372283936,
      "learning_rate": 9.468342105263158e-05,
      "loss": 0.392,
      "step": 40222
    },
    {
      "epoch": 2.252379885765483,
      "grad_norm": 1.0562124252319336,
      "learning_rate": 9.468315789473685e-05,
      "loss": 0.3896,
      "step": 40223
    },
    {
      "epoch": 2.252435883077612,
      "grad_norm": 1.2212586402893066,
      "learning_rate": 9.468289473684211e-05,
      "loss": 0.4116,
      "step": 40224
    },
    {
      "epoch": 2.252491880389741,
      "grad_norm": 1.3088864088058472,
      "learning_rate": 9.468263157894738e-05,
      "loss": 0.3429,
      "step": 40225
    },
    {
      "epoch": 2.25254787770187,
      "grad_norm": 1.155081033706665,
      "learning_rate": 9.468236842105263e-05,
      "loss": 0.423,
      "step": 40226
    },
    {
      "epoch": 2.252603875013999,
      "grad_norm": 1.1861810684204102,
      "learning_rate": 9.468210526315789e-05,
      "loss": 0.3501,
      "step": 40227
    },
    {
      "epoch": 2.252659872326128,
      "grad_norm": 1.2616063356399536,
      "learning_rate": 9.468184210526316e-05,
      "loss": 0.397,
      "step": 40228
    },
    {
      "epoch": 2.252715869638257,
      "grad_norm": 1.175142526626587,
      "learning_rate": 9.468157894736842e-05,
      "loss": 0.2777,
      "step": 40229
    },
    {
      "epoch": 2.2527718669503862,
      "grad_norm": 1.2170854806900024,
      "learning_rate": 9.46813157894737e-05,
      "loss": 0.3856,
      "step": 40230
    },
    {
      "epoch": 2.2528278642625152,
      "grad_norm": 1.0473250150680542,
      "learning_rate": 9.468105263157894e-05,
      "loss": 0.3207,
      "step": 40231
    },
    {
      "epoch": 2.2528838615746443,
      "grad_norm": 1.2947707176208496,
      "learning_rate": 9.468078947368422e-05,
      "loss": 0.3708,
      "step": 40232
    },
    {
      "epoch": 2.2529398588867733,
      "grad_norm": 1.213927149772644,
      "learning_rate": 9.468052631578948e-05,
      "loss": 0.3919,
      "step": 40233
    },
    {
      "epoch": 2.2529958561989023,
      "grad_norm": 1.257867693901062,
      "learning_rate": 9.468026315789475e-05,
      "loss": 0.4879,
      "step": 40234
    },
    {
      "epoch": 2.2530518535110313,
      "grad_norm": 1.1220237016677856,
      "learning_rate": 9.468000000000001e-05,
      "loss": 0.3626,
      "step": 40235
    },
    {
      "epoch": 2.2531078508231603,
      "grad_norm": 1.1531041860580444,
      "learning_rate": 9.467973684210527e-05,
      "loss": 0.3942,
      "step": 40236
    },
    {
      "epoch": 2.2531638481352894,
      "grad_norm": 1.3512048721313477,
      "learning_rate": 9.467947368421053e-05,
      "loss": 0.461,
      "step": 40237
    },
    {
      "epoch": 2.2532198454474184,
      "grad_norm": 1.1911240816116333,
      "learning_rate": 9.46792105263158e-05,
      "loss": 0.4034,
      "step": 40238
    },
    {
      "epoch": 2.2532758427595474,
      "grad_norm": 1.36972975730896,
      "learning_rate": 9.467894736842106e-05,
      "loss": 0.4378,
      "step": 40239
    },
    {
      "epoch": 2.2533318400716764,
      "grad_norm": 1.178636908531189,
      "learning_rate": 9.467868421052632e-05,
      "loss": 0.3935,
      "step": 40240
    },
    {
      "epoch": 2.2533878373838054,
      "grad_norm": 1.2232637405395508,
      "learning_rate": 9.467842105263158e-05,
      "loss": 0.5019,
      "step": 40241
    },
    {
      "epoch": 2.2534438346959345,
      "grad_norm": 1.074726939201355,
      "learning_rate": 9.467815789473685e-05,
      "loss": 0.3484,
      "step": 40242
    },
    {
      "epoch": 2.2534998320080635,
      "grad_norm": 1.0410679578781128,
      "learning_rate": 9.467789473684211e-05,
      "loss": 0.3195,
      "step": 40243
    },
    {
      "epoch": 2.2535558293201925,
      "grad_norm": 1.1891978979110718,
      "learning_rate": 9.467763157894737e-05,
      "loss": 0.4273,
      "step": 40244
    },
    {
      "epoch": 2.2536118266323215,
      "grad_norm": 1.1611050367355347,
      "learning_rate": 9.467736842105263e-05,
      "loss": 0.3517,
      "step": 40245
    },
    {
      "epoch": 2.2536678239444505,
      "grad_norm": 1.3147691488265991,
      "learning_rate": 9.467710526315789e-05,
      "loss": 0.4235,
      "step": 40246
    },
    {
      "epoch": 2.2537238212565796,
      "grad_norm": 1.5211437940597534,
      "learning_rate": 9.467684210526317e-05,
      "loss": 0.3539,
      "step": 40247
    },
    {
      "epoch": 2.2537798185687086,
      "grad_norm": 1.1932721138000488,
      "learning_rate": 9.467657894736843e-05,
      "loss": 0.4261,
      "step": 40248
    },
    {
      "epoch": 2.2538358158808376,
      "grad_norm": 1.2296850681304932,
      "learning_rate": 9.467631578947369e-05,
      "loss": 0.4086,
      "step": 40249
    },
    {
      "epoch": 2.2538918131929666,
      "grad_norm": 1.157375693321228,
      "learning_rate": 9.467605263157894e-05,
      "loss": 0.3681,
      "step": 40250
    },
    {
      "epoch": 2.2539478105050956,
      "grad_norm": 1.4234784841537476,
      "learning_rate": 9.467578947368422e-05,
      "loss": 0.4119,
      "step": 40251
    },
    {
      "epoch": 2.2540038078172246,
      "grad_norm": 1.0864088535308838,
      "learning_rate": 9.467552631578948e-05,
      "loss": 0.3212,
      "step": 40252
    },
    {
      "epoch": 2.2540598051293537,
      "grad_norm": 1.1244655847549438,
      "learning_rate": 9.467526315789475e-05,
      "loss": 0.3245,
      "step": 40253
    },
    {
      "epoch": 2.2541158024414827,
      "grad_norm": 1.0265889167785645,
      "learning_rate": 9.4675e-05,
      "loss": 0.3804,
      "step": 40254
    },
    {
      "epoch": 2.2541717997536117,
      "grad_norm": 1.337349534034729,
      "learning_rate": 9.467473684210527e-05,
      "loss": 0.4125,
      "step": 40255
    },
    {
      "epoch": 2.2542277970657407,
      "grad_norm": 1.4713741540908813,
      "learning_rate": 9.467447368421053e-05,
      "loss": 0.4126,
      "step": 40256
    },
    {
      "epoch": 2.2542837943778697,
      "grad_norm": 1.6377911567687988,
      "learning_rate": 9.46742105263158e-05,
      "loss": 0.5054,
      "step": 40257
    },
    {
      "epoch": 2.2543397916899988,
      "grad_norm": 1.0416444540023804,
      "learning_rate": 9.467394736842105e-05,
      "loss": 0.3019,
      "step": 40258
    },
    {
      "epoch": 2.254395789002128,
      "grad_norm": 1.5094548463821411,
      "learning_rate": 9.467368421052632e-05,
      "loss": 0.5026,
      "step": 40259
    },
    {
      "epoch": 2.254451786314257,
      "grad_norm": 1.0562242269515991,
      "learning_rate": 9.467342105263158e-05,
      "loss": 0.3072,
      "step": 40260
    },
    {
      "epoch": 2.254507783626386,
      "grad_norm": 1.186156988143921,
      "learning_rate": 9.467315789473684e-05,
      "loss": 0.3499,
      "step": 40261
    },
    {
      "epoch": 2.254563780938515,
      "grad_norm": 1.5425621271133423,
      "learning_rate": 9.467289473684212e-05,
      "loss": 0.3366,
      "step": 40262
    },
    {
      "epoch": 2.254619778250644,
      "grad_norm": 1.4950631856918335,
      "learning_rate": 9.467263157894736e-05,
      "loss": 0.4938,
      "step": 40263
    },
    {
      "epoch": 2.254675775562773,
      "grad_norm": 1.323747158050537,
      "learning_rate": 9.467236842105264e-05,
      "loss": 0.3857,
      "step": 40264
    },
    {
      "epoch": 2.254731772874902,
      "grad_norm": 1.1153520345687866,
      "learning_rate": 9.46721052631579e-05,
      "loss": 0.3747,
      "step": 40265
    },
    {
      "epoch": 2.254787770187031,
      "grad_norm": 1.0667577981948853,
      "learning_rate": 9.467184210526317e-05,
      "loss": 0.3097,
      "step": 40266
    },
    {
      "epoch": 2.25484376749916,
      "grad_norm": 1.3149584531784058,
      "learning_rate": 9.467157894736843e-05,
      "loss": 0.4503,
      "step": 40267
    },
    {
      "epoch": 2.254899764811289,
      "grad_norm": 1.2549396753311157,
      "learning_rate": 9.467131578947369e-05,
      "loss": 0.369,
      "step": 40268
    },
    {
      "epoch": 2.254955762123418,
      "grad_norm": 1.2764784097671509,
      "learning_rate": 9.467105263157895e-05,
      "loss": 0.4124,
      "step": 40269
    },
    {
      "epoch": 2.255011759435547,
      "grad_norm": 1.1376515626907349,
      "learning_rate": 9.467078947368422e-05,
      "loss": 0.4005,
      "step": 40270
    },
    {
      "epoch": 2.255067756747676,
      "grad_norm": 1.490296483039856,
      "learning_rate": 9.467052631578948e-05,
      "loss": 0.3302,
      "step": 40271
    },
    {
      "epoch": 2.255123754059805,
      "grad_norm": 2.2487077713012695,
      "learning_rate": 9.467026315789474e-05,
      "loss": 0.3226,
      "step": 40272
    },
    {
      "epoch": 2.255179751371934,
      "grad_norm": 1.6115819215774536,
      "learning_rate": 9.467e-05,
      "loss": 0.4234,
      "step": 40273
    },
    {
      "epoch": 2.255235748684063,
      "grad_norm": 8.818623542785645,
      "learning_rate": 9.466973684210527e-05,
      "loss": 0.4371,
      "step": 40274
    },
    {
      "epoch": 2.255291745996192,
      "grad_norm": 1.279303789138794,
      "learning_rate": 9.466947368421053e-05,
      "loss": 0.4366,
      "step": 40275
    },
    {
      "epoch": 2.255347743308321,
      "grad_norm": 1.1217670440673828,
      "learning_rate": 9.466921052631579e-05,
      "loss": 0.3366,
      "step": 40276
    },
    {
      "epoch": 2.25540374062045,
      "grad_norm": 1.648451805114746,
      "learning_rate": 9.466894736842105e-05,
      "loss": 0.4528,
      "step": 40277
    },
    {
      "epoch": 2.255459737932579,
      "grad_norm": 1.1057618856430054,
      "learning_rate": 9.466868421052631e-05,
      "loss": 0.3194,
      "step": 40278
    },
    {
      "epoch": 2.255515735244708,
      "grad_norm": 1.3770878314971924,
      "learning_rate": 9.466842105263159e-05,
      "loss": 0.4615,
      "step": 40279
    },
    {
      "epoch": 2.255571732556837,
      "grad_norm": 1.1005779504776,
      "learning_rate": 9.466815789473685e-05,
      "loss": 0.364,
      "step": 40280
    },
    {
      "epoch": 2.255627729868966,
      "grad_norm": 1.367957353591919,
      "learning_rate": 9.46678947368421e-05,
      "loss": 0.4468,
      "step": 40281
    },
    {
      "epoch": 2.255683727181095,
      "grad_norm": 1.3481450080871582,
      "learning_rate": 9.466763157894736e-05,
      "loss": 0.461,
      "step": 40282
    },
    {
      "epoch": 2.2557397244932242,
      "grad_norm": 1.098899245262146,
      "learning_rate": 9.466736842105264e-05,
      "loss": 0.4193,
      "step": 40283
    },
    {
      "epoch": 2.2557957218053533,
      "grad_norm": 1.4182229042053223,
      "learning_rate": 9.46671052631579e-05,
      "loss": 0.4216,
      "step": 40284
    },
    {
      "epoch": 2.2558517191174823,
      "grad_norm": 3.334995746612549,
      "learning_rate": 9.466684210526317e-05,
      "loss": 0.5931,
      "step": 40285
    },
    {
      "epoch": 2.2559077164296113,
      "grad_norm": 1.3568634986877441,
      "learning_rate": 9.466657894736842e-05,
      "loss": 0.442,
      "step": 40286
    },
    {
      "epoch": 2.2559637137417403,
      "grad_norm": 0.9288855791091919,
      "learning_rate": 9.466631578947369e-05,
      "loss": 0.3407,
      "step": 40287
    },
    {
      "epoch": 2.2560197110538693,
      "grad_norm": 1.2287007570266724,
      "learning_rate": 9.466605263157895e-05,
      "loss": 0.3445,
      "step": 40288
    },
    {
      "epoch": 2.2560757083659984,
      "grad_norm": 0.9728140830993652,
      "learning_rate": 9.466578947368422e-05,
      "loss": 0.3269,
      "step": 40289
    },
    {
      "epoch": 2.2561317056781274,
      "grad_norm": 0.9737794995307922,
      "learning_rate": 9.466552631578948e-05,
      "loss": 0.3516,
      "step": 40290
    },
    {
      "epoch": 2.2561877029902564,
      "grad_norm": 1.1371206045150757,
      "learning_rate": 9.466526315789474e-05,
      "loss": 0.4607,
      "step": 40291
    },
    {
      "epoch": 2.2562437003023854,
      "grad_norm": 1.1774191856384277,
      "learning_rate": 9.4665e-05,
      "loss": 0.4708,
      "step": 40292
    },
    {
      "epoch": 2.2562996976145144,
      "grad_norm": 1.1353883743286133,
      "learning_rate": 9.466473684210528e-05,
      "loss": 0.539,
      "step": 40293
    },
    {
      "epoch": 2.2563556949266435,
      "grad_norm": 1.2183647155761719,
      "learning_rate": 9.466447368421054e-05,
      "loss": 0.2985,
      "step": 40294
    },
    {
      "epoch": 2.2564116922387725,
      "grad_norm": 2.177011728286743,
      "learning_rate": 9.46642105263158e-05,
      "loss": 0.4216,
      "step": 40295
    },
    {
      "epoch": 2.2564676895509015,
      "grad_norm": 1.1379708051681519,
      "learning_rate": 9.466394736842105e-05,
      "loss": 0.4859,
      "step": 40296
    },
    {
      "epoch": 2.2565236868630305,
      "grad_norm": 1.1925945281982422,
      "learning_rate": 9.466368421052631e-05,
      "loss": 0.4342,
      "step": 40297
    },
    {
      "epoch": 2.2565796841751595,
      "grad_norm": 1.1074904203414917,
      "learning_rate": 9.466342105263159e-05,
      "loss": 0.469,
      "step": 40298
    },
    {
      "epoch": 2.2566356814872885,
      "grad_norm": 1.2941091060638428,
      "learning_rate": 9.466315789473685e-05,
      "loss": 0.412,
      "step": 40299
    },
    {
      "epoch": 2.2566916787994176,
      "grad_norm": 1.2547640800476074,
      "learning_rate": 9.466289473684211e-05,
      "loss": 0.3441,
      "step": 40300
    },
    {
      "epoch": 2.2567476761115466,
      "grad_norm": 1.072322964668274,
      "learning_rate": 9.466263157894737e-05,
      "loss": 0.4617,
      "step": 40301
    },
    {
      "epoch": 2.2568036734236756,
      "grad_norm": 1.2649054527282715,
      "learning_rate": 9.466236842105264e-05,
      "loss": 0.4337,
      "step": 40302
    },
    {
      "epoch": 2.2568596707358046,
      "grad_norm": 1.1271538734436035,
      "learning_rate": 9.46621052631579e-05,
      "loss": 0.4349,
      "step": 40303
    },
    {
      "epoch": 2.2569156680479336,
      "grad_norm": 1.3438583612442017,
      "learning_rate": 9.466184210526316e-05,
      "loss": 0.4786,
      "step": 40304
    },
    {
      "epoch": 2.2569716653600627,
      "grad_norm": 0.9407152533531189,
      "learning_rate": 9.466157894736842e-05,
      "loss": 0.2909,
      "step": 40305
    },
    {
      "epoch": 2.2570276626721917,
      "grad_norm": 1.1129287481307983,
      "learning_rate": 9.466131578947369e-05,
      "loss": 0.3138,
      "step": 40306
    },
    {
      "epoch": 2.2570836599843207,
      "grad_norm": 1.1620537042617798,
      "learning_rate": 9.466105263157895e-05,
      "loss": 0.3559,
      "step": 40307
    },
    {
      "epoch": 2.2571396572964497,
      "grad_norm": 1.3283755779266357,
      "learning_rate": 9.466078947368423e-05,
      "loss": 0.4498,
      "step": 40308
    },
    {
      "epoch": 2.2571956546085787,
      "grad_norm": 1.0311189889907837,
      "learning_rate": 9.466052631578947e-05,
      "loss": 0.3601,
      "step": 40309
    },
    {
      "epoch": 2.2572516519207078,
      "grad_norm": 1.1707583665847778,
      "learning_rate": 9.466026315789475e-05,
      "loss": 0.426,
      "step": 40310
    },
    {
      "epoch": 2.2573076492328368,
      "grad_norm": 1.4164403676986694,
      "learning_rate": 9.466e-05,
      "loss": 0.4428,
      "step": 40311
    },
    {
      "epoch": 2.257363646544966,
      "grad_norm": 1.3374484777450562,
      "learning_rate": 9.465973684210526e-05,
      "loss": 0.5566,
      "step": 40312
    },
    {
      "epoch": 2.257419643857095,
      "grad_norm": 1.245092511177063,
      "learning_rate": 9.465947368421052e-05,
      "loss": 0.4054,
      "step": 40313
    },
    {
      "epoch": 2.257475641169224,
      "grad_norm": 1.0508105754852295,
      "learning_rate": 9.465921052631578e-05,
      "loss": 0.2421,
      "step": 40314
    },
    {
      "epoch": 2.257531638481353,
      "grad_norm": 1.4905184507369995,
      "learning_rate": 9.465894736842106e-05,
      "loss": 0.4139,
      "step": 40315
    },
    {
      "epoch": 2.257587635793482,
      "grad_norm": 1.2063349485397339,
      "learning_rate": 9.465868421052632e-05,
      "loss": 0.3856,
      "step": 40316
    },
    {
      "epoch": 2.257643633105611,
      "grad_norm": 1.3707929849624634,
      "learning_rate": 9.465842105263159e-05,
      "loss": 0.5647,
      "step": 40317
    },
    {
      "epoch": 2.25769963041774,
      "grad_norm": 1.251119613647461,
      "learning_rate": 9.465815789473684e-05,
      "loss": 0.5111,
      "step": 40318
    },
    {
      "epoch": 2.257755627729869,
      "grad_norm": 1.135854959487915,
      "learning_rate": 9.465789473684211e-05,
      "loss": 0.3353,
      "step": 40319
    },
    {
      "epoch": 2.257811625041998,
      "grad_norm": 1.2801518440246582,
      "learning_rate": 9.465763157894737e-05,
      "loss": 0.4317,
      "step": 40320
    },
    {
      "epoch": 2.257867622354127,
      "grad_norm": 1.2633455991744995,
      "learning_rate": 9.465736842105264e-05,
      "loss": 0.4702,
      "step": 40321
    },
    {
      "epoch": 2.257923619666256,
      "grad_norm": 1.1757678985595703,
      "learning_rate": 9.46571052631579e-05,
      "loss": 0.3274,
      "step": 40322
    },
    {
      "epoch": 2.257979616978385,
      "grad_norm": 1.1845859289169312,
      "learning_rate": 9.465684210526316e-05,
      "loss": 0.3826,
      "step": 40323
    },
    {
      "epoch": 2.258035614290514,
      "grad_norm": 1.0530648231506348,
      "learning_rate": 9.465657894736842e-05,
      "loss": 0.3302,
      "step": 40324
    },
    {
      "epoch": 2.258091611602643,
      "grad_norm": 1.1436012983322144,
      "learning_rate": 9.46563157894737e-05,
      "loss": 0.4458,
      "step": 40325
    },
    {
      "epoch": 2.258147608914772,
      "grad_norm": 1.3375295400619507,
      "learning_rate": 9.465605263157896e-05,
      "loss": 0.4548,
      "step": 40326
    },
    {
      "epoch": 2.258203606226901,
      "grad_norm": 1.704437494277954,
      "learning_rate": 9.465578947368421e-05,
      "loss": 0.4973,
      "step": 40327
    },
    {
      "epoch": 2.25825960353903,
      "grad_norm": 1.1635740995407104,
      "learning_rate": 9.465552631578947e-05,
      "loss": 0.3054,
      "step": 40328
    },
    {
      "epoch": 2.258315600851159,
      "grad_norm": 1.2312188148498535,
      "learning_rate": 9.465526315789475e-05,
      "loss": 0.3454,
      "step": 40329
    },
    {
      "epoch": 2.258371598163288,
      "grad_norm": 1.2998744249343872,
      "learning_rate": 9.465500000000001e-05,
      "loss": 0.4072,
      "step": 40330
    },
    {
      "epoch": 2.258427595475417,
      "grad_norm": 1.1218258142471313,
      "learning_rate": 9.465473684210527e-05,
      "loss": 0.4692,
      "step": 40331
    },
    {
      "epoch": 2.258483592787546,
      "grad_norm": 1.1797640323638916,
      "learning_rate": 9.465447368421053e-05,
      "loss": 0.3704,
      "step": 40332
    },
    {
      "epoch": 2.258539590099675,
      "grad_norm": 1.1451855897903442,
      "learning_rate": 9.465421052631579e-05,
      "loss": 0.3644,
      "step": 40333
    },
    {
      "epoch": 2.258595587411804,
      "grad_norm": 1.065679907798767,
      "learning_rate": 9.465394736842106e-05,
      "loss": 0.4223,
      "step": 40334
    },
    {
      "epoch": 2.2586515847239332,
      "grad_norm": 1.1771676540374756,
      "learning_rate": 9.465368421052632e-05,
      "loss": 0.3959,
      "step": 40335
    },
    {
      "epoch": 2.2587075820360623,
      "grad_norm": 1.444936990737915,
      "learning_rate": 9.465342105263158e-05,
      "loss": 0.6433,
      "step": 40336
    },
    {
      "epoch": 2.2587635793481913,
      "grad_norm": 1.2181119918823242,
      "learning_rate": 9.465315789473684e-05,
      "loss": 0.3704,
      "step": 40337
    },
    {
      "epoch": 2.2588195766603203,
      "grad_norm": 1.3059271574020386,
      "learning_rate": 9.465289473684211e-05,
      "loss": 0.3059,
      "step": 40338
    },
    {
      "epoch": 2.2588755739724493,
      "grad_norm": 1.0524498224258423,
      "learning_rate": 9.465263157894737e-05,
      "loss": 0.3513,
      "step": 40339
    },
    {
      "epoch": 2.2589315712845783,
      "grad_norm": 1.3416804075241089,
      "learning_rate": 9.465236842105265e-05,
      "loss": 0.3838,
      "step": 40340
    },
    {
      "epoch": 2.2589875685967074,
      "grad_norm": 1.1031877994537354,
      "learning_rate": 9.465210526315789e-05,
      "loss": 0.4401,
      "step": 40341
    },
    {
      "epoch": 2.2590435659088364,
      "grad_norm": 1.066445231437683,
      "learning_rate": 9.465184210526317e-05,
      "loss": 0.3803,
      "step": 40342
    },
    {
      "epoch": 2.2590995632209654,
      "grad_norm": 1.0381954908370972,
      "learning_rate": 9.465157894736842e-05,
      "loss": 0.3713,
      "step": 40343
    },
    {
      "epoch": 2.2591555605330944,
      "grad_norm": 1.3082499504089355,
      "learning_rate": 9.46513157894737e-05,
      "loss": 0.5798,
      "step": 40344
    },
    {
      "epoch": 2.2592115578452234,
      "grad_norm": 1.2993909120559692,
      "learning_rate": 9.465105263157896e-05,
      "loss": 0.4107,
      "step": 40345
    },
    {
      "epoch": 2.2592675551573524,
      "grad_norm": 1.0972895622253418,
      "learning_rate": 9.465078947368422e-05,
      "loss": 0.4262,
      "step": 40346
    },
    {
      "epoch": 2.2593235524694815,
      "grad_norm": 1.192597508430481,
      "learning_rate": 9.465052631578948e-05,
      "loss": 0.3273,
      "step": 40347
    },
    {
      "epoch": 2.2593795497816105,
      "grad_norm": 1.2875967025756836,
      "learning_rate": 9.465026315789474e-05,
      "loss": 0.3857,
      "step": 40348
    },
    {
      "epoch": 2.2594355470937395,
      "grad_norm": 1.0806066989898682,
      "learning_rate": 9.465000000000001e-05,
      "loss": 0.3981,
      "step": 40349
    },
    {
      "epoch": 2.2594915444058685,
      "grad_norm": 1.3211942911148071,
      "learning_rate": 9.464973684210527e-05,
      "loss": 0.5012,
      "step": 40350
    },
    {
      "epoch": 2.2595475417179975,
      "grad_norm": 1.0952271223068237,
      "learning_rate": 9.464947368421053e-05,
      "loss": 0.3828,
      "step": 40351
    },
    {
      "epoch": 2.2596035390301266,
      "grad_norm": 1.0847833156585693,
      "learning_rate": 9.464921052631579e-05,
      "loss": 0.448,
      "step": 40352
    },
    {
      "epoch": 2.2596595363422556,
      "grad_norm": 1.0694583654403687,
      "learning_rate": 9.464894736842106e-05,
      "loss": 0.3886,
      "step": 40353
    },
    {
      "epoch": 2.2597155336543846,
      "grad_norm": 0.9983018040657043,
      "learning_rate": 9.464868421052632e-05,
      "loss": 0.3799,
      "step": 40354
    },
    {
      "epoch": 2.2597715309665136,
      "grad_norm": 1.216844916343689,
      "learning_rate": 9.464842105263158e-05,
      "loss": 0.4187,
      "step": 40355
    },
    {
      "epoch": 2.2598275282786426,
      "grad_norm": 1.175490140914917,
      "learning_rate": 9.464815789473684e-05,
      "loss": 0.3263,
      "step": 40356
    },
    {
      "epoch": 2.2598835255907717,
      "grad_norm": 1.2974120378494263,
      "learning_rate": 9.464789473684212e-05,
      "loss": 0.3872,
      "step": 40357
    },
    {
      "epoch": 2.2599395229029007,
      "grad_norm": 1.2057908773422241,
      "learning_rate": 9.464763157894737e-05,
      "loss": 0.4072,
      "step": 40358
    },
    {
      "epoch": 2.2599955202150297,
      "grad_norm": 1.2761505842208862,
      "learning_rate": 9.464736842105263e-05,
      "loss": 0.382,
      "step": 40359
    },
    {
      "epoch": 2.2600515175271587,
      "grad_norm": 1.1110893487930298,
      "learning_rate": 9.46471052631579e-05,
      "loss": 0.3967,
      "step": 40360
    },
    {
      "epoch": 2.2601075148392877,
      "grad_norm": 1.2542656660079956,
      "learning_rate": 9.464684210526317e-05,
      "loss": 0.3321,
      "step": 40361
    },
    {
      "epoch": 2.2601635121514168,
      "grad_norm": 1.3461135625839233,
      "learning_rate": 9.464657894736843e-05,
      "loss": 0.506,
      "step": 40362
    },
    {
      "epoch": 2.2602195094635458,
      "grad_norm": 3.6581709384918213,
      "learning_rate": 9.46463157894737e-05,
      "loss": 0.4516,
      "step": 40363
    },
    {
      "epoch": 2.260275506775675,
      "grad_norm": 1.1049401760101318,
      "learning_rate": 9.464605263157895e-05,
      "loss": 0.2598,
      "step": 40364
    },
    {
      "epoch": 2.260331504087804,
      "grad_norm": 1.2461767196655273,
      "learning_rate": 9.46457894736842e-05,
      "loss": 0.33,
      "step": 40365
    },
    {
      "epoch": 2.260387501399933,
      "grad_norm": 1.1858155727386475,
      "learning_rate": 9.464552631578948e-05,
      "loss": 0.3229,
      "step": 40366
    },
    {
      "epoch": 2.260443498712062,
      "grad_norm": 1.3696266412734985,
      "learning_rate": 9.464526315789474e-05,
      "loss": 0.435,
      "step": 40367
    },
    {
      "epoch": 2.260499496024191,
      "grad_norm": 1.6595209836959839,
      "learning_rate": 9.4645e-05,
      "loss": 0.4714,
      "step": 40368
    },
    {
      "epoch": 2.26055549333632,
      "grad_norm": 1.3161598443984985,
      "learning_rate": 9.464473684210526e-05,
      "loss": 0.3896,
      "step": 40369
    },
    {
      "epoch": 2.260611490648449,
      "grad_norm": 0.9250859022140503,
      "learning_rate": 9.464447368421053e-05,
      "loss": 0.2896,
      "step": 40370
    },
    {
      "epoch": 2.260667487960578,
      "grad_norm": 1.1213774681091309,
      "learning_rate": 9.464421052631579e-05,
      "loss": 0.443,
      "step": 40371
    },
    {
      "epoch": 2.260723485272707,
      "grad_norm": 1.0855780839920044,
      "learning_rate": 9.464394736842107e-05,
      "loss": 0.3396,
      "step": 40372
    },
    {
      "epoch": 2.260779482584836,
      "grad_norm": 1.2259312868118286,
      "learning_rate": 9.464368421052631e-05,
      "loss": 0.4426,
      "step": 40373
    },
    {
      "epoch": 2.260835479896965,
      "grad_norm": 1.1146873235702515,
      "learning_rate": 9.464342105263158e-05,
      "loss": 0.3586,
      "step": 40374
    },
    {
      "epoch": 2.260891477209094,
      "grad_norm": 1.0434610843658447,
      "learning_rate": 9.464315789473684e-05,
      "loss": 0.3922,
      "step": 40375
    },
    {
      "epoch": 2.260947474521223,
      "grad_norm": 1.2282809019088745,
      "learning_rate": 9.464289473684212e-05,
      "loss": 0.3515,
      "step": 40376
    },
    {
      "epoch": 2.261003471833352,
      "grad_norm": 1.2720481157302856,
      "learning_rate": 9.464263157894738e-05,
      "loss": 0.4477,
      "step": 40377
    },
    {
      "epoch": 2.261059469145481,
      "grad_norm": 1.222194790840149,
      "learning_rate": 9.464236842105264e-05,
      "loss": 0.3408,
      "step": 40378
    },
    {
      "epoch": 2.26111546645761,
      "grad_norm": 1.0590826272964478,
      "learning_rate": 9.46421052631579e-05,
      "loss": 0.3593,
      "step": 40379
    },
    {
      "epoch": 2.261171463769739,
      "grad_norm": 1.039536476135254,
      "learning_rate": 9.464184210526317e-05,
      "loss": 0.3553,
      "step": 40380
    },
    {
      "epoch": 2.261227461081868,
      "grad_norm": 1.315877079963684,
      "learning_rate": 9.464157894736843e-05,
      "loss": 0.416,
      "step": 40381
    },
    {
      "epoch": 2.261283458393997,
      "grad_norm": 1.3325190544128418,
      "learning_rate": 9.464131578947369e-05,
      "loss": 0.4003,
      "step": 40382
    },
    {
      "epoch": 2.261339455706126,
      "grad_norm": 1.7807756662368774,
      "learning_rate": 9.464105263157895e-05,
      "loss": 0.5673,
      "step": 40383
    },
    {
      "epoch": 2.261395453018255,
      "grad_norm": 1.0976921319961548,
      "learning_rate": 9.464078947368421e-05,
      "loss": 0.4525,
      "step": 40384
    },
    {
      "epoch": 2.261451450330384,
      "grad_norm": 1.4316296577453613,
      "learning_rate": 9.464052631578948e-05,
      "loss": 0.4938,
      "step": 40385
    },
    {
      "epoch": 2.261507447642513,
      "grad_norm": 1.2819199562072754,
      "learning_rate": 9.464026315789474e-05,
      "loss": 0.4582,
      "step": 40386
    },
    {
      "epoch": 2.2615634449546422,
      "grad_norm": 1.1583997011184692,
      "learning_rate": 9.464e-05,
      "loss": 0.4459,
      "step": 40387
    },
    {
      "epoch": 2.2616194422667713,
      "grad_norm": 1.0067702531814575,
      "learning_rate": 9.463973684210526e-05,
      "loss": 0.3078,
      "step": 40388
    },
    {
      "epoch": 2.2616754395789003,
      "grad_norm": 1.244781494140625,
      "learning_rate": 9.463947368421053e-05,
      "loss": 0.3075,
      "step": 40389
    },
    {
      "epoch": 2.2617314368910293,
      "grad_norm": 1.3263671398162842,
      "learning_rate": 9.46392105263158e-05,
      "loss": 0.4379,
      "step": 40390
    },
    {
      "epoch": 2.2617874342031583,
      "grad_norm": 1.2320064306259155,
      "learning_rate": 9.463894736842105e-05,
      "loss": 0.3751,
      "step": 40391
    },
    {
      "epoch": 2.2618434315152873,
      "grad_norm": 1.0125446319580078,
      "learning_rate": 9.463868421052631e-05,
      "loss": 0.3668,
      "step": 40392
    },
    {
      "epoch": 2.2618994288274163,
      "grad_norm": 0.9972684383392334,
      "learning_rate": 9.463842105263159e-05,
      "loss": 0.3232,
      "step": 40393
    },
    {
      "epoch": 2.2619554261395454,
      "grad_norm": 1.8667103052139282,
      "learning_rate": 9.463815789473685e-05,
      "loss": 0.4368,
      "step": 40394
    },
    {
      "epoch": 2.2620114234516744,
      "grad_norm": 1.273384928703308,
      "learning_rate": 9.463789473684212e-05,
      "loss": 0.4138,
      "step": 40395
    },
    {
      "epoch": 2.2620674207638034,
      "grad_norm": 1.73623526096344,
      "learning_rate": 9.463763157894737e-05,
      "loss": 0.4158,
      "step": 40396
    },
    {
      "epoch": 2.2621234180759324,
      "grad_norm": 1.2916810512542725,
      "learning_rate": 9.463736842105264e-05,
      "loss": 0.3893,
      "step": 40397
    },
    {
      "epoch": 2.2621794153880614,
      "grad_norm": 1.079209566116333,
      "learning_rate": 9.46371052631579e-05,
      "loss": 0.4067,
      "step": 40398
    },
    {
      "epoch": 2.2622354127001905,
      "grad_norm": 1.1295397281646729,
      "learning_rate": 9.463684210526316e-05,
      "loss": 0.453,
      "step": 40399
    },
    {
      "epoch": 2.2622914100123195,
      "grad_norm": 1.0388671159744263,
      "learning_rate": 9.463657894736843e-05,
      "loss": 0.4025,
      "step": 40400
    },
    {
      "epoch": 2.2623474073244485,
      "grad_norm": 1.066413164138794,
      "learning_rate": 9.463631578947368e-05,
      "loss": 0.3365,
      "step": 40401
    },
    {
      "epoch": 2.2624034046365775,
      "grad_norm": 1.284334421157837,
      "learning_rate": 9.463605263157895e-05,
      "loss": 0.3746,
      "step": 40402
    },
    {
      "epoch": 2.2624594019487065,
      "grad_norm": 1.3797918558120728,
      "learning_rate": 9.463578947368421e-05,
      "loss": 0.4389,
      "step": 40403
    },
    {
      "epoch": 2.2625153992608356,
      "grad_norm": 1.4297658205032349,
      "learning_rate": 9.463552631578949e-05,
      "loss": 0.4545,
      "step": 40404
    },
    {
      "epoch": 2.2625713965729646,
      "grad_norm": 1.23444664478302,
      "learning_rate": 9.463526315789473e-05,
      "loss": 0.4252,
      "step": 40405
    },
    {
      "epoch": 2.2626273938850936,
      "grad_norm": 3.6973371505737305,
      "learning_rate": 9.4635e-05,
      "loss": 0.44,
      "step": 40406
    },
    {
      "epoch": 2.2626833911972226,
      "grad_norm": 1.0619100332260132,
      "learning_rate": 9.463473684210526e-05,
      "loss": 0.3907,
      "step": 40407
    },
    {
      "epoch": 2.2627393885093516,
      "grad_norm": 1.301879644393921,
      "learning_rate": 9.463447368421054e-05,
      "loss": 0.4198,
      "step": 40408
    },
    {
      "epoch": 2.2627953858214807,
      "grad_norm": 1.2094963788986206,
      "learning_rate": 9.46342105263158e-05,
      "loss": 0.486,
      "step": 40409
    },
    {
      "epoch": 2.2628513831336097,
      "grad_norm": 1.3144015073776245,
      "learning_rate": 9.463394736842106e-05,
      "loss": 0.4568,
      "step": 40410
    },
    {
      "epoch": 2.2629073804457387,
      "grad_norm": 1.2742695808410645,
      "learning_rate": 9.463368421052632e-05,
      "loss": 0.4925,
      "step": 40411
    },
    {
      "epoch": 2.2629633777578677,
      "grad_norm": 1.131796956062317,
      "learning_rate": 9.463342105263159e-05,
      "loss": 0.539,
      "step": 40412
    },
    {
      "epoch": 2.2630193750699967,
      "grad_norm": 2.9456851482391357,
      "learning_rate": 9.463315789473685e-05,
      "loss": 0.3464,
      "step": 40413
    },
    {
      "epoch": 2.2630753723821257,
      "grad_norm": 1.27320396900177,
      "learning_rate": 9.463289473684211e-05,
      "loss": 0.5058,
      "step": 40414
    },
    {
      "epoch": 2.2631313696942548,
      "grad_norm": 1.1681808233261108,
      "learning_rate": 9.463263157894737e-05,
      "loss": 0.3627,
      "step": 40415
    },
    {
      "epoch": 2.263187367006384,
      "grad_norm": 1.185450553894043,
      "learning_rate": 9.463236842105263e-05,
      "loss": 0.4167,
      "step": 40416
    },
    {
      "epoch": 2.263243364318513,
      "grad_norm": 1.0709630250930786,
      "learning_rate": 9.46321052631579e-05,
      "loss": 0.4486,
      "step": 40417
    },
    {
      "epoch": 2.263299361630642,
      "grad_norm": 1.3015860319137573,
      "learning_rate": 9.463184210526316e-05,
      "loss": 0.4456,
      "step": 40418
    },
    {
      "epoch": 2.263355358942771,
      "grad_norm": 1.0957012176513672,
      "learning_rate": 9.463157894736842e-05,
      "loss": 0.3832,
      "step": 40419
    },
    {
      "epoch": 2.2634113562549,
      "grad_norm": 1.168670654296875,
      "learning_rate": 9.463131578947368e-05,
      "loss": 0.4307,
      "step": 40420
    },
    {
      "epoch": 2.263467353567029,
      "grad_norm": 1.1268547773361206,
      "learning_rate": 9.463105263157895e-05,
      "loss": 0.4312,
      "step": 40421
    },
    {
      "epoch": 2.263523350879158,
      "grad_norm": 1.2061339616775513,
      "learning_rate": 9.463078947368421e-05,
      "loss": 0.3836,
      "step": 40422
    },
    {
      "epoch": 2.263579348191287,
      "grad_norm": 1.1966536045074463,
      "learning_rate": 9.463052631578947e-05,
      "loss": 0.3374,
      "step": 40423
    },
    {
      "epoch": 2.263635345503416,
      "grad_norm": 1.177609920501709,
      "learning_rate": 9.463026315789473e-05,
      "loss": 0.4126,
      "step": 40424
    },
    {
      "epoch": 2.263691342815545,
      "grad_norm": 1.6852208375930786,
      "learning_rate": 9.463000000000001e-05,
      "loss": 0.387,
      "step": 40425
    },
    {
      "epoch": 2.263747340127674,
      "grad_norm": 1.2249035835266113,
      "learning_rate": 9.462973684210527e-05,
      "loss": 0.444,
      "step": 40426
    },
    {
      "epoch": 2.263803337439803,
      "grad_norm": 8.048696517944336,
      "learning_rate": 9.462947368421054e-05,
      "loss": 0.336,
      "step": 40427
    },
    {
      "epoch": 2.263859334751932,
      "grad_norm": 1.4287192821502686,
      "learning_rate": 9.462921052631579e-05,
      "loss": 0.4392,
      "step": 40428
    },
    {
      "epoch": 2.263915332064061,
      "grad_norm": 1.277269721031189,
      "learning_rate": 9.462894736842106e-05,
      "loss": 0.4229,
      "step": 40429
    },
    {
      "epoch": 2.26397132937619,
      "grad_norm": 1.132293462753296,
      "learning_rate": 9.462868421052632e-05,
      "loss": 0.4926,
      "step": 40430
    },
    {
      "epoch": 2.264027326688319,
      "grad_norm": 1.2950248718261719,
      "learning_rate": 9.462842105263159e-05,
      "loss": 0.3702,
      "step": 40431
    },
    {
      "epoch": 2.264083324000448,
      "grad_norm": 1.1158510446548462,
      "learning_rate": 9.462815789473685e-05,
      "loss": 0.4292,
      "step": 40432
    },
    {
      "epoch": 2.264139321312577,
      "grad_norm": 2.381382465362549,
      "learning_rate": 9.46278947368421e-05,
      "loss": 0.4134,
      "step": 40433
    },
    {
      "epoch": 2.264195318624706,
      "grad_norm": 1.0139827728271484,
      "learning_rate": 9.462763157894737e-05,
      "loss": 0.388,
      "step": 40434
    },
    {
      "epoch": 2.264251315936835,
      "grad_norm": 1.3971186876296997,
      "learning_rate": 9.462736842105263e-05,
      "loss": 0.3968,
      "step": 40435
    },
    {
      "epoch": 2.264307313248964,
      "grad_norm": 1.136900782585144,
      "learning_rate": 9.46271052631579e-05,
      "loss": 0.4134,
      "step": 40436
    },
    {
      "epoch": 2.264363310561093,
      "grad_norm": 1.0802419185638428,
      "learning_rate": 9.462684210526316e-05,
      "loss": 0.3914,
      "step": 40437
    },
    {
      "epoch": 2.264419307873222,
      "grad_norm": 1.1812375783920288,
      "learning_rate": 9.462657894736842e-05,
      "loss": 0.3496,
      "step": 40438
    },
    {
      "epoch": 2.2644753051853512,
      "grad_norm": 1.2728197574615479,
      "learning_rate": 9.462631578947368e-05,
      "loss": 0.3968,
      "step": 40439
    },
    {
      "epoch": 2.2645313024974802,
      "grad_norm": 1.2735815048217773,
      "learning_rate": 9.462605263157896e-05,
      "loss": 0.4051,
      "step": 40440
    },
    {
      "epoch": 2.2645872998096093,
      "grad_norm": 1.0759103298187256,
      "learning_rate": 9.462578947368422e-05,
      "loss": 0.3644,
      "step": 40441
    },
    {
      "epoch": 2.2646432971217383,
      "grad_norm": 1.2403099536895752,
      "learning_rate": 9.462552631578948e-05,
      "loss": 0.351,
      "step": 40442
    },
    {
      "epoch": 2.2646992944338673,
      "grad_norm": 1.233041763305664,
      "learning_rate": 9.462526315789474e-05,
      "loss": 0.402,
      "step": 40443
    },
    {
      "epoch": 2.2647552917459963,
      "grad_norm": 1.0860545635223389,
      "learning_rate": 9.462500000000001e-05,
      "loss": 0.4781,
      "step": 40444
    },
    {
      "epoch": 2.2648112890581253,
      "grad_norm": 1.46695077419281,
      "learning_rate": 9.462473684210527e-05,
      "loss": 0.4261,
      "step": 40445
    },
    {
      "epoch": 2.2648672863702544,
      "grad_norm": 1.1049014329910278,
      "learning_rate": 9.462447368421053e-05,
      "loss": 0.3832,
      "step": 40446
    },
    {
      "epoch": 2.2649232836823834,
      "grad_norm": 1.1958290338516235,
      "learning_rate": 9.462421052631579e-05,
      "loss": 0.3442,
      "step": 40447
    },
    {
      "epoch": 2.2649792809945124,
      "grad_norm": 1.3269994258880615,
      "learning_rate": 9.462394736842106e-05,
      "loss": 0.4438,
      "step": 40448
    },
    {
      "epoch": 2.2650352783066414,
      "grad_norm": 1.1083450317382812,
      "learning_rate": 9.462368421052632e-05,
      "loss": 0.3506,
      "step": 40449
    },
    {
      "epoch": 2.2650912756187704,
      "grad_norm": 1.1331969499588013,
      "learning_rate": 9.46234210526316e-05,
      "loss": 0.3819,
      "step": 40450
    },
    {
      "epoch": 2.2651472729308995,
      "grad_norm": 1.1612178087234497,
      "learning_rate": 9.462315789473684e-05,
      "loss": 0.4164,
      "step": 40451
    },
    {
      "epoch": 2.2652032702430285,
      "grad_norm": 1.1315405368804932,
      "learning_rate": 9.46228947368421e-05,
      "loss": 0.4671,
      "step": 40452
    },
    {
      "epoch": 2.2652592675551575,
      "grad_norm": 1.1833078861236572,
      "learning_rate": 9.462263157894737e-05,
      "loss": 0.3384,
      "step": 40453
    },
    {
      "epoch": 2.2653152648672865,
      "grad_norm": 1.2797057628631592,
      "learning_rate": 9.462236842105263e-05,
      "loss": 0.4459,
      "step": 40454
    },
    {
      "epoch": 2.2653712621794155,
      "grad_norm": 1.142492651939392,
      "learning_rate": 9.462210526315791e-05,
      "loss": 0.3413,
      "step": 40455
    },
    {
      "epoch": 2.2654272594915446,
      "grad_norm": 1.2397264242172241,
      "learning_rate": 9.462184210526315e-05,
      "loss": 0.4774,
      "step": 40456
    },
    {
      "epoch": 2.2654832568036736,
      "grad_norm": 1.2352741956710815,
      "learning_rate": 9.462157894736843e-05,
      "loss": 0.3767,
      "step": 40457
    },
    {
      "epoch": 2.2655392541158026,
      "grad_norm": 1.2771943807601929,
      "learning_rate": 9.462131578947369e-05,
      "loss": 0.445,
      "step": 40458
    },
    {
      "epoch": 2.2655952514279316,
      "grad_norm": 0.9388750195503235,
      "learning_rate": 9.462105263157896e-05,
      "loss": 0.2893,
      "step": 40459
    },
    {
      "epoch": 2.2656512487400606,
      "grad_norm": 1.263070821762085,
      "learning_rate": 9.46207894736842e-05,
      "loss": 0.3609,
      "step": 40460
    },
    {
      "epoch": 2.2657072460521896,
      "grad_norm": 1.284348487854004,
      "learning_rate": 9.462052631578948e-05,
      "loss": 0.3801,
      "step": 40461
    },
    {
      "epoch": 2.2657632433643187,
      "grad_norm": 1.3505194187164307,
      "learning_rate": 9.462026315789474e-05,
      "loss": 0.3853,
      "step": 40462
    },
    {
      "epoch": 2.2658192406764477,
      "grad_norm": 1.1799821853637695,
      "learning_rate": 9.462000000000001e-05,
      "loss": 0.4255,
      "step": 40463
    },
    {
      "epoch": 2.2658752379885767,
      "grad_norm": 1.0638905763626099,
      "learning_rate": 9.461973684210527e-05,
      "loss": 0.4123,
      "step": 40464
    },
    {
      "epoch": 2.2659312353007057,
      "grad_norm": 1.027500867843628,
      "learning_rate": 9.461947368421053e-05,
      "loss": 0.3315,
      "step": 40465
    },
    {
      "epoch": 2.2659872326128347,
      "grad_norm": 1.6638792753219604,
      "learning_rate": 9.461921052631579e-05,
      "loss": 0.418,
      "step": 40466
    },
    {
      "epoch": 2.2660432299249638,
      "grad_norm": 2.853843927383423,
      "learning_rate": 9.461894736842106e-05,
      "loss": 0.4194,
      "step": 40467
    },
    {
      "epoch": 2.266099227237093,
      "grad_norm": 1.109104871749878,
      "learning_rate": 9.461868421052632e-05,
      "loss": 0.3686,
      "step": 40468
    },
    {
      "epoch": 2.266155224549222,
      "grad_norm": 1.1648691892623901,
      "learning_rate": 9.461842105263158e-05,
      "loss": 0.3273,
      "step": 40469
    },
    {
      "epoch": 2.266211221861351,
      "grad_norm": 1.2241833209991455,
      "learning_rate": 9.461815789473684e-05,
      "loss": 0.4193,
      "step": 40470
    },
    {
      "epoch": 2.26626721917348,
      "grad_norm": 1.2876853942871094,
      "learning_rate": 9.46178947368421e-05,
      "loss": 0.382,
      "step": 40471
    },
    {
      "epoch": 2.266323216485609,
      "grad_norm": 1.217707633972168,
      "learning_rate": 9.461763157894738e-05,
      "loss": 0.3919,
      "step": 40472
    },
    {
      "epoch": 2.266379213797738,
      "grad_norm": 1.123657464981079,
      "learning_rate": 9.461736842105264e-05,
      "loss": 0.4051,
      "step": 40473
    },
    {
      "epoch": 2.266435211109867,
      "grad_norm": 1.0729305744171143,
      "learning_rate": 9.46171052631579e-05,
      "loss": 0.4951,
      "step": 40474
    },
    {
      "epoch": 2.266491208421996,
      "grad_norm": 1.163252592086792,
      "learning_rate": 9.461684210526316e-05,
      "loss": 0.3706,
      "step": 40475
    },
    {
      "epoch": 2.266547205734125,
      "grad_norm": 1.1236474514007568,
      "learning_rate": 9.461657894736843e-05,
      "loss": 0.3724,
      "step": 40476
    },
    {
      "epoch": 2.266603203046254,
      "grad_norm": 1.0196162462234497,
      "learning_rate": 9.461631578947369e-05,
      "loss": 0.3453,
      "step": 40477
    },
    {
      "epoch": 2.266659200358383,
      "grad_norm": 1.349471092224121,
      "learning_rate": 9.461605263157895e-05,
      "loss": 0.4287,
      "step": 40478
    },
    {
      "epoch": 2.266715197670512,
      "grad_norm": 1.0240190029144287,
      "learning_rate": 9.461578947368421e-05,
      "loss": 0.3366,
      "step": 40479
    },
    {
      "epoch": 2.266771194982641,
      "grad_norm": 1.178196907043457,
      "learning_rate": 9.461552631578948e-05,
      "loss": 0.3333,
      "step": 40480
    },
    {
      "epoch": 2.26682719229477,
      "grad_norm": 1.16752028465271,
      "learning_rate": 9.461526315789474e-05,
      "loss": 0.4048,
      "step": 40481
    },
    {
      "epoch": 2.266883189606899,
      "grad_norm": 1.1811375617980957,
      "learning_rate": 9.461500000000001e-05,
      "loss": 0.4129,
      "step": 40482
    },
    {
      "epoch": 2.266939186919028,
      "grad_norm": 1.3402822017669678,
      "learning_rate": 9.461473684210526e-05,
      "loss": 0.5252,
      "step": 40483
    },
    {
      "epoch": 2.266995184231157,
      "grad_norm": 1.2851439714431763,
      "learning_rate": 9.461447368421053e-05,
      "loss": 0.4547,
      "step": 40484
    },
    {
      "epoch": 2.267051181543286,
      "grad_norm": 1.2237930297851562,
      "learning_rate": 9.46142105263158e-05,
      "loss": 0.3724,
      "step": 40485
    },
    {
      "epoch": 2.267107178855415,
      "grad_norm": 1.1725502014160156,
      "learning_rate": 9.461394736842105e-05,
      "loss": 0.3866,
      "step": 40486
    },
    {
      "epoch": 2.267163176167544,
      "grad_norm": 1.3695721626281738,
      "learning_rate": 9.461368421052633e-05,
      "loss": 0.5176,
      "step": 40487
    },
    {
      "epoch": 2.267219173479673,
      "grad_norm": 1.1344494819641113,
      "learning_rate": 9.461342105263157e-05,
      "loss": 0.3788,
      "step": 40488
    },
    {
      "epoch": 2.267275170791802,
      "grad_norm": 1.2121444940567017,
      "learning_rate": 9.461315789473685e-05,
      "loss": 0.5308,
      "step": 40489
    },
    {
      "epoch": 2.267331168103931,
      "grad_norm": 0.9950360059738159,
      "learning_rate": 9.46128947368421e-05,
      "loss": 0.3252,
      "step": 40490
    },
    {
      "epoch": 2.26738716541606,
      "grad_norm": 1.164266586303711,
      "learning_rate": 9.461263157894738e-05,
      "loss": 0.3423,
      "step": 40491
    },
    {
      "epoch": 2.2674431627281892,
      "grad_norm": 0.986965000629425,
      "learning_rate": 9.461236842105264e-05,
      "loss": 0.368,
      "step": 40492
    },
    {
      "epoch": 2.267499160040318,
      "grad_norm": 1.242859959602356,
      "learning_rate": 9.46121052631579e-05,
      "loss": 0.3715,
      "step": 40493
    },
    {
      "epoch": 2.2675551573524473,
      "grad_norm": 1.0006341934204102,
      "learning_rate": 9.461184210526316e-05,
      "loss": 0.4417,
      "step": 40494
    },
    {
      "epoch": 2.267611154664576,
      "grad_norm": 1.089477777481079,
      "learning_rate": 9.461157894736843e-05,
      "loss": 0.429,
      "step": 40495
    },
    {
      "epoch": 2.2676671519767053,
      "grad_norm": 1.4148645401000977,
      "learning_rate": 9.461131578947369e-05,
      "loss": 0.4258,
      "step": 40496
    },
    {
      "epoch": 2.267723149288834,
      "grad_norm": 1.2728643417358398,
      "learning_rate": 9.461105263157895e-05,
      "loss": 0.3921,
      "step": 40497
    },
    {
      "epoch": 2.2677791466009634,
      "grad_norm": 1.278693675994873,
      "learning_rate": 9.461078947368421e-05,
      "loss": 0.4192,
      "step": 40498
    },
    {
      "epoch": 2.267835143913092,
      "grad_norm": 1.3750426769256592,
      "learning_rate": 9.461052631578948e-05,
      "loss": 0.5112,
      "step": 40499
    },
    {
      "epoch": 2.2678911412252214,
      "grad_norm": 1.2014737129211426,
      "learning_rate": 9.461026315789474e-05,
      "loss": 0.3825,
      "step": 40500
    },
    {
      "epoch": 2.26794713853735,
      "grad_norm": 1.263054609298706,
      "learning_rate": 9.461e-05,
      "loss": 0.4629,
      "step": 40501
    },
    {
      "epoch": 2.2680031358494794,
      "grad_norm": 1.3502540588378906,
      "learning_rate": 9.460973684210526e-05,
      "loss": 0.3915,
      "step": 40502
    },
    {
      "epoch": 2.268059133161608,
      "grad_norm": 1.1903547048568726,
      "learning_rate": 9.460947368421052e-05,
      "loss": 0.383,
      "step": 40503
    },
    {
      "epoch": 2.2681151304737375,
      "grad_norm": 1.1257144212722778,
      "learning_rate": 9.46092105263158e-05,
      "loss": 0.3255,
      "step": 40504
    },
    {
      "epoch": 2.268171127785866,
      "grad_norm": 1.2431561946868896,
      "learning_rate": 9.460894736842106e-05,
      "loss": 0.3416,
      "step": 40505
    },
    {
      "epoch": 2.2682271250979955,
      "grad_norm": 1.1336997747421265,
      "learning_rate": 9.460868421052632e-05,
      "loss": 0.4381,
      "step": 40506
    },
    {
      "epoch": 2.268283122410124,
      "grad_norm": 1.1583088636398315,
      "learning_rate": 9.460842105263158e-05,
      "loss": 0.3545,
      "step": 40507
    },
    {
      "epoch": 2.2683391197222535,
      "grad_norm": 1.2995786666870117,
      "learning_rate": 9.460815789473685e-05,
      "loss": 0.3637,
      "step": 40508
    },
    {
      "epoch": 2.268395117034382,
      "grad_norm": 1.337058424949646,
      "learning_rate": 9.460789473684211e-05,
      "loss": 0.376,
      "step": 40509
    },
    {
      "epoch": 2.2684511143465116,
      "grad_norm": 1.2869439125061035,
      "learning_rate": 9.460763157894738e-05,
      "loss": 0.3914,
      "step": 40510
    },
    {
      "epoch": 2.26850711165864,
      "grad_norm": 1.2116807699203491,
      "learning_rate": 9.460736842105263e-05,
      "loss": 0.4164,
      "step": 40511
    },
    {
      "epoch": 2.2685631089707696,
      "grad_norm": 1.1212992668151855,
      "learning_rate": 9.46071052631579e-05,
      "loss": 0.3749,
      "step": 40512
    },
    {
      "epoch": 2.268619106282898,
      "grad_norm": 1.5295838117599487,
      "learning_rate": 9.460684210526316e-05,
      "loss": 0.3856,
      "step": 40513
    },
    {
      "epoch": 2.2686751035950277,
      "grad_norm": 1.5487151145935059,
      "learning_rate": 9.460657894736843e-05,
      "loss": 0.3304,
      "step": 40514
    },
    {
      "epoch": 2.2687311009071562,
      "grad_norm": 1.5145213603973389,
      "learning_rate": 9.460631578947368e-05,
      "loss": 0.5556,
      "step": 40515
    },
    {
      "epoch": 2.2687870982192857,
      "grad_norm": 1.3443989753723145,
      "learning_rate": 9.460605263157895e-05,
      "loss": 0.5611,
      "step": 40516
    },
    {
      "epoch": 2.2688430955314143,
      "grad_norm": 1.3281117677688599,
      "learning_rate": 9.460578947368421e-05,
      "loss": 0.4301,
      "step": 40517
    },
    {
      "epoch": 2.2688990928435437,
      "grad_norm": 1.243749737739563,
      "learning_rate": 9.460552631578949e-05,
      "loss": 0.4215,
      "step": 40518
    },
    {
      "epoch": 2.2689550901556723,
      "grad_norm": 1.2872587442398071,
      "learning_rate": 9.460526315789475e-05,
      "loss": 0.4699,
      "step": 40519
    },
    {
      "epoch": 2.269011087467802,
      "grad_norm": 1.2896764278411865,
      "learning_rate": 9.460499999999999e-05,
      "loss": 0.4627,
      "step": 40520
    },
    {
      "epoch": 2.2690670847799304,
      "grad_norm": 1.0949015617370605,
      "learning_rate": 9.460473684210527e-05,
      "loss": 0.3285,
      "step": 40521
    },
    {
      "epoch": 2.26912308209206,
      "grad_norm": 1.007161021232605,
      "learning_rate": 9.460447368421053e-05,
      "loss": 0.3132,
      "step": 40522
    },
    {
      "epoch": 2.2691790794041884,
      "grad_norm": 1.0987603664398193,
      "learning_rate": 9.46042105263158e-05,
      "loss": 0.5,
      "step": 40523
    },
    {
      "epoch": 2.269235076716318,
      "grad_norm": 1.708152174949646,
      "learning_rate": 9.460394736842106e-05,
      "loss": 0.3719,
      "step": 40524
    },
    {
      "epoch": 2.2692910740284464,
      "grad_norm": 1.13357675075531,
      "learning_rate": 9.460368421052632e-05,
      "loss": 0.3864,
      "step": 40525
    },
    {
      "epoch": 2.269347071340576,
      "grad_norm": 1.1700465679168701,
      "learning_rate": 9.460342105263158e-05,
      "loss": 0.2584,
      "step": 40526
    },
    {
      "epoch": 2.2694030686527045,
      "grad_norm": 1.2960734367370605,
      "learning_rate": 9.460315789473685e-05,
      "loss": 0.4427,
      "step": 40527
    },
    {
      "epoch": 2.269459065964834,
      "grad_norm": 1.1186174154281616,
      "learning_rate": 9.460289473684211e-05,
      "loss": 0.448,
      "step": 40528
    },
    {
      "epoch": 2.2695150632769625,
      "grad_norm": 1.0102180242538452,
      "learning_rate": 9.460263157894737e-05,
      "loss": 0.3312,
      "step": 40529
    },
    {
      "epoch": 2.269571060589092,
      "grad_norm": 1.3226977586746216,
      "learning_rate": 9.460236842105263e-05,
      "loss": 0.3526,
      "step": 40530
    },
    {
      "epoch": 2.2696270579012205,
      "grad_norm": 1.0980631113052368,
      "learning_rate": 9.46021052631579e-05,
      "loss": 0.3637,
      "step": 40531
    },
    {
      "epoch": 2.2696830552133496,
      "grad_norm": 1.2014403343200684,
      "learning_rate": 9.460184210526316e-05,
      "loss": 0.368,
      "step": 40532
    },
    {
      "epoch": 2.2697390525254786,
      "grad_norm": 1.1677125692367554,
      "learning_rate": 9.460157894736842e-05,
      "loss": 0.3874,
      "step": 40533
    },
    {
      "epoch": 2.2697950498376076,
      "grad_norm": 1.1828606128692627,
      "learning_rate": 9.460131578947368e-05,
      "loss": 0.4269,
      "step": 40534
    },
    {
      "epoch": 2.2698510471497366,
      "grad_norm": 1.1326069831848145,
      "learning_rate": 9.460105263157896e-05,
      "loss": 0.2152,
      "step": 40535
    },
    {
      "epoch": 2.2699070444618656,
      "grad_norm": 1.3309108018875122,
      "learning_rate": 9.460078947368422e-05,
      "loss": 0.4462,
      "step": 40536
    },
    {
      "epoch": 2.2699630417739947,
      "grad_norm": 1.1551995277404785,
      "learning_rate": 9.460052631578948e-05,
      "loss": 0.3438,
      "step": 40537
    },
    {
      "epoch": 2.2700190390861237,
      "grad_norm": 1.0498015880584717,
      "learning_rate": 9.460026315789474e-05,
      "loss": 0.3138,
      "step": 40538
    },
    {
      "epoch": 2.2700750363982527,
      "grad_norm": 1.1026535034179688,
      "learning_rate": 9.46e-05,
      "loss": 0.4305,
      "step": 40539
    },
    {
      "epoch": 2.2701310337103817,
      "grad_norm": 1.0027085542678833,
      "learning_rate": 9.459973684210527e-05,
      "loss": 0.3159,
      "step": 40540
    },
    {
      "epoch": 2.2701870310225107,
      "grad_norm": 1.3759174346923828,
      "learning_rate": 9.459947368421053e-05,
      "loss": 0.3919,
      "step": 40541
    },
    {
      "epoch": 2.2702430283346398,
      "grad_norm": 1.138127326965332,
      "learning_rate": 9.45992105263158e-05,
      "loss": 0.4041,
      "step": 40542
    },
    {
      "epoch": 2.2702990256467688,
      "grad_norm": 1.0994898080825806,
      "learning_rate": 9.459894736842105e-05,
      "loss": 0.3442,
      "step": 40543
    },
    {
      "epoch": 2.270355022958898,
      "grad_norm": 1.5775060653686523,
      "learning_rate": 9.459868421052632e-05,
      "loss": 0.4103,
      "step": 40544
    },
    {
      "epoch": 2.270411020271027,
      "grad_norm": 1.1225507259368896,
      "learning_rate": 9.459842105263158e-05,
      "loss": 0.3927,
      "step": 40545
    },
    {
      "epoch": 2.270467017583156,
      "grad_norm": 1.0932636260986328,
      "learning_rate": 9.459815789473685e-05,
      "loss": 0.3331,
      "step": 40546
    },
    {
      "epoch": 2.270523014895285,
      "grad_norm": 1.2866544723510742,
      "learning_rate": 9.459789473684211e-05,
      "loss": 0.4559,
      "step": 40547
    },
    {
      "epoch": 2.270579012207414,
      "grad_norm": 1.8256518840789795,
      "learning_rate": 9.459763157894737e-05,
      "loss": 0.3918,
      "step": 40548
    },
    {
      "epoch": 2.270635009519543,
      "grad_norm": 1.1217502355575562,
      "learning_rate": 9.459736842105263e-05,
      "loss": 0.3729,
      "step": 40549
    },
    {
      "epoch": 2.270691006831672,
      "grad_norm": 1.2543939352035522,
      "learning_rate": 9.45971052631579e-05,
      "loss": 0.453,
      "step": 40550
    },
    {
      "epoch": 2.270747004143801,
      "grad_norm": 1.3210906982421875,
      "learning_rate": 9.459684210526317e-05,
      "loss": 0.544,
      "step": 40551
    },
    {
      "epoch": 2.27080300145593,
      "grad_norm": 1.0610665082931519,
      "learning_rate": 9.459657894736843e-05,
      "loss": 0.3944,
      "step": 40552
    },
    {
      "epoch": 2.270858998768059,
      "grad_norm": 1.028406023979187,
      "learning_rate": 9.459631578947369e-05,
      "loss": 0.3958,
      "step": 40553
    },
    {
      "epoch": 2.270914996080188,
      "grad_norm": 1.5856077671051025,
      "learning_rate": 9.459605263157895e-05,
      "loss": 0.4057,
      "step": 40554
    },
    {
      "epoch": 2.270970993392317,
      "grad_norm": 1.3397217988967896,
      "learning_rate": 9.459578947368422e-05,
      "loss": 0.5113,
      "step": 40555
    },
    {
      "epoch": 2.271026990704446,
      "grad_norm": 1.700391411781311,
      "learning_rate": 9.459552631578948e-05,
      "loss": 0.4107,
      "step": 40556
    },
    {
      "epoch": 2.271082988016575,
      "grad_norm": 1.009935975074768,
      "learning_rate": 9.459526315789474e-05,
      "loss": 0.3437,
      "step": 40557
    },
    {
      "epoch": 2.271138985328704,
      "grad_norm": 1.384375810623169,
      "learning_rate": 9.4595e-05,
      "loss": 0.4215,
      "step": 40558
    },
    {
      "epoch": 2.271194982640833,
      "grad_norm": 1.545481562614441,
      "learning_rate": 9.459473684210527e-05,
      "loss": 0.4502,
      "step": 40559
    },
    {
      "epoch": 2.271250979952962,
      "grad_norm": 1.2553125619888306,
      "learning_rate": 9.459447368421053e-05,
      "loss": 0.5871,
      "step": 40560
    },
    {
      "epoch": 2.271306977265091,
      "grad_norm": 1.3203444480895996,
      "learning_rate": 9.459421052631579e-05,
      "loss": 0.635,
      "step": 40561
    },
    {
      "epoch": 2.27136297457722,
      "grad_norm": 1.2080284357070923,
      "learning_rate": 9.459394736842105e-05,
      "loss": 0.3949,
      "step": 40562
    },
    {
      "epoch": 2.271418971889349,
      "grad_norm": 1.0241115093231201,
      "learning_rate": 9.459368421052632e-05,
      "loss": 0.4426,
      "step": 40563
    },
    {
      "epoch": 2.271474969201478,
      "grad_norm": 1.3481581211090088,
      "learning_rate": 9.459342105263158e-05,
      "loss": 0.396,
      "step": 40564
    },
    {
      "epoch": 2.271530966513607,
      "grad_norm": 1.2400517463684082,
      "learning_rate": 9.459315789473686e-05,
      "loss": 0.4219,
      "step": 40565
    },
    {
      "epoch": 2.271586963825736,
      "grad_norm": 1.135642409324646,
      "learning_rate": 9.45928947368421e-05,
      "loss": 0.2816,
      "step": 40566
    },
    {
      "epoch": 2.2716429611378652,
      "grad_norm": 1.3617733716964722,
      "learning_rate": 9.459263157894738e-05,
      "loss": 0.4161,
      "step": 40567
    },
    {
      "epoch": 2.2716989584499943,
      "grad_norm": 1.1379430294036865,
      "learning_rate": 9.459236842105264e-05,
      "loss": 0.4298,
      "step": 40568
    },
    {
      "epoch": 2.2717549557621233,
      "grad_norm": 1.3969991207122803,
      "learning_rate": 9.459210526315791e-05,
      "loss": 0.4807,
      "step": 40569
    },
    {
      "epoch": 2.2718109530742523,
      "grad_norm": 1.384647011756897,
      "learning_rate": 9.459184210526316e-05,
      "loss": 0.387,
      "step": 40570
    },
    {
      "epoch": 2.2718669503863813,
      "grad_norm": 1.4508918523788452,
      "learning_rate": 9.459157894736843e-05,
      "loss": 0.5153,
      "step": 40571
    },
    {
      "epoch": 2.2719229476985103,
      "grad_norm": 1.0793359279632568,
      "learning_rate": 9.459131578947369e-05,
      "loss": 0.3294,
      "step": 40572
    },
    {
      "epoch": 2.2719789450106394,
      "grad_norm": 1.2309826612472534,
      "learning_rate": 9.459105263157895e-05,
      "loss": 0.5021,
      "step": 40573
    },
    {
      "epoch": 2.2720349423227684,
      "grad_norm": 1.0962035655975342,
      "learning_rate": 9.459078947368422e-05,
      "loss": 0.4171,
      "step": 40574
    },
    {
      "epoch": 2.2720909396348974,
      "grad_norm": 1.2114489078521729,
      "learning_rate": 9.459052631578947e-05,
      "loss": 0.3636,
      "step": 40575
    },
    {
      "epoch": 2.2721469369470264,
      "grad_norm": 1.2173047065734863,
      "learning_rate": 9.459026315789474e-05,
      "loss": 0.3545,
      "step": 40576
    },
    {
      "epoch": 2.2722029342591554,
      "grad_norm": 1.0975050926208496,
      "learning_rate": 9.459e-05,
      "loss": 0.5374,
      "step": 40577
    },
    {
      "epoch": 2.2722589315712844,
      "grad_norm": 1.1415431499481201,
      "learning_rate": 9.458973684210527e-05,
      "loss": 0.3797,
      "step": 40578
    },
    {
      "epoch": 2.2723149288834135,
      "grad_norm": 1.1166616678237915,
      "learning_rate": 9.458947368421053e-05,
      "loss": 0.4344,
      "step": 40579
    },
    {
      "epoch": 2.2723709261955425,
      "grad_norm": 1.2645405530929565,
      "learning_rate": 9.45892105263158e-05,
      "loss": 0.4048,
      "step": 40580
    },
    {
      "epoch": 2.2724269235076715,
      "grad_norm": 1.1214083433151245,
      "learning_rate": 9.458894736842105e-05,
      "loss": 0.4528,
      "step": 40581
    },
    {
      "epoch": 2.2724829208198005,
      "grad_norm": 1.2971934080123901,
      "learning_rate": 9.458868421052633e-05,
      "loss": 0.5001,
      "step": 40582
    },
    {
      "epoch": 2.2725389181319295,
      "grad_norm": 1.2930711507797241,
      "learning_rate": 9.458842105263159e-05,
      "loss": 0.4363,
      "step": 40583
    },
    {
      "epoch": 2.2725949154440586,
      "grad_norm": 1.1945903301239014,
      "learning_rate": 9.458815789473685e-05,
      "loss": 0.4326,
      "step": 40584
    },
    {
      "epoch": 2.2726509127561876,
      "grad_norm": 1.2399060726165771,
      "learning_rate": 9.45878947368421e-05,
      "loss": 0.5048,
      "step": 40585
    },
    {
      "epoch": 2.2727069100683166,
      "grad_norm": 1.334643840789795,
      "learning_rate": 9.458763157894738e-05,
      "loss": 0.4086,
      "step": 40586
    },
    {
      "epoch": 2.2727629073804456,
      "grad_norm": 1.189266324043274,
      "learning_rate": 9.458736842105264e-05,
      "loss": 0.3829,
      "step": 40587
    },
    {
      "epoch": 2.2728189046925746,
      "grad_norm": 1.0801719427108765,
      "learning_rate": 9.45871052631579e-05,
      "loss": 0.4455,
      "step": 40588
    },
    {
      "epoch": 2.2728749020047037,
      "grad_norm": 1.4720373153686523,
      "learning_rate": 9.458684210526316e-05,
      "loss": 0.5109,
      "step": 40589
    },
    {
      "epoch": 2.2729308993168327,
      "grad_norm": 1.4487801790237427,
      "learning_rate": 9.458657894736842e-05,
      "loss": 0.3974,
      "step": 40590
    },
    {
      "epoch": 2.2729868966289617,
      "grad_norm": 1.0977919101715088,
      "learning_rate": 9.458631578947369e-05,
      "loss": 0.3584,
      "step": 40591
    },
    {
      "epoch": 2.2730428939410907,
      "grad_norm": 1.474965214729309,
      "learning_rate": 9.458605263157895e-05,
      "loss": 0.3883,
      "step": 40592
    },
    {
      "epoch": 2.2730988912532197,
      "grad_norm": 1.1310701370239258,
      "learning_rate": 9.458578947368421e-05,
      "loss": 0.3491,
      "step": 40593
    },
    {
      "epoch": 2.2731548885653488,
      "grad_norm": 1.2242786884307861,
      "learning_rate": 9.458552631578947e-05,
      "loss": 0.3747,
      "step": 40594
    },
    {
      "epoch": 2.2732108858774778,
      "grad_norm": 2.974498748779297,
      "learning_rate": 9.458526315789474e-05,
      "loss": 0.3417,
      "step": 40595
    },
    {
      "epoch": 2.273266883189607,
      "grad_norm": 1.2573018074035645,
      "learning_rate": 9.4585e-05,
      "loss": 0.3961,
      "step": 40596
    },
    {
      "epoch": 2.273322880501736,
      "grad_norm": 1.2831525802612305,
      "learning_rate": 9.458473684210528e-05,
      "loss": 0.4377,
      "step": 40597
    },
    {
      "epoch": 2.273378877813865,
      "grad_norm": 1.4229695796966553,
      "learning_rate": 9.458447368421052e-05,
      "loss": 0.4332,
      "step": 40598
    },
    {
      "epoch": 2.273434875125994,
      "grad_norm": 1.1408116817474365,
      "learning_rate": 9.45842105263158e-05,
      "loss": 0.3922,
      "step": 40599
    },
    {
      "epoch": 2.273490872438123,
      "grad_norm": 1.0934343338012695,
      "learning_rate": 9.458394736842106e-05,
      "loss": 0.3424,
      "step": 40600
    },
    {
      "epoch": 2.273546869750252,
      "grad_norm": 1.2211631536483765,
      "learning_rate": 9.458368421052633e-05,
      "loss": 0.4017,
      "step": 40601
    },
    {
      "epoch": 2.273602867062381,
      "grad_norm": 1.3245149850845337,
      "learning_rate": 9.458342105263159e-05,
      "loss": 0.4802,
      "step": 40602
    },
    {
      "epoch": 2.27365886437451,
      "grad_norm": 1.1212351322174072,
      "learning_rate": 9.458315789473685e-05,
      "loss": 0.5004,
      "step": 40603
    },
    {
      "epoch": 2.273714861686639,
      "grad_norm": 1.0691895484924316,
      "learning_rate": 9.458289473684211e-05,
      "loss": 0.4036,
      "step": 40604
    },
    {
      "epoch": 2.273770858998768,
      "grad_norm": 1.0293790102005005,
      "learning_rate": 9.458263157894738e-05,
      "loss": 0.2551,
      "step": 40605
    },
    {
      "epoch": 2.273826856310897,
      "grad_norm": 1.2396889925003052,
      "learning_rate": 9.458236842105264e-05,
      "loss": 0.4525,
      "step": 40606
    },
    {
      "epoch": 2.273882853623026,
      "grad_norm": 1.2029087543487549,
      "learning_rate": 9.458210526315789e-05,
      "loss": 0.3525,
      "step": 40607
    },
    {
      "epoch": 2.273938850935155,
      "grad_norm": 1.2619563341140747,
      "learning_rate": 9.458184210526316e-05,
      "loss": 0.4911,
      "step": 40608
    },
    {
      "epoch": 2.273994848247284,
      "grad_norm": 1.2013729810714722,
      "learning_rate": 9.458157894736842e-05,
      "loss": 0.5835,
      "step": 40609
    },
    {
      "epoch": 2.274050845559413,
      "grad_norm": 1.0333744287490845,
      "learning_rate": 9.45813157894737e-05,
      "loss": 0.4243,
      "step": 40610
    },
    {
      "epoch": 2.274106842871542,
      "grad_norm": 1.1525123119354248,
      "learning_rate": 9.458105263157895e-05,
      "loss": 0.5303,
      "step": 40611
    },
    {
      "epoch": 2.274162840183671,
      "grad_norm": 1.3624944686889648,
      "learning_rate": 9.458078947368421e-05,
      "loss": 0.4247,
      "step": 40612
    },
    {
      "epoch": 2.2742188374958,
      "grad_norm": 1.298219084739685,
      "learning_rate": 9.458052631578947e-05,
      "loss": 0.4992,
      "step": 40613
    },
    {
      "epoch": 2.274274834807929,
      "grad_norm": 1.0188695192337036,
      "learning_rate": 9.458026315789475e-05,
      "loss": 0.29,
      "step": 40614
    },
    {
      "epoch": 2.274330832120058,
      "grad_norm": 1.324236512184143,
      "learning_rate": 9.458e-05,
      "loss": 0.3992,
      "step": 40615
    },
    {
      "epoch": 2.274386829432187,
      "grad_norm": 1.5384267568588257,
      "learning_rate": 9.457973684210527e-05,
      "loss": 0.4577,
      "step": 40616
    },
    {
      "epoch": 2.274442826744316,
      "grad_norm": 0.99504154920578,
      "learning_rate": 9.457947368421053e-05,
      "loss": 0.2792,
      "step": 40617
    },
    {
      "epoch": 2.274498824056445,
      "grad_norm": 1.2706249952316284,
      "learning_rate": 9.45792105263158e-05,
      "loss": 0.3899,
      "step": 40618
    },
    {
      "epoch": 2.2745548213685742,
      "grad_norm": 1.286297082901001,
      "learning_rate": 9.457894736842106e-05,
      "loss": 0.385,
      "step": 40619
    },
    {
      "epoch": 2.2746108186807033,
      "grad_norm": 1.2250655889511108,
      "learning_rate": 9.457868421052633e-05,
      "loss": 0.361,
      "step": 40620
    },
    {
      "epoch": 2.2746668159928323,
      "grad_norm": 1.2352221012115479,
      "learning_rate": 9.457842105263158e-05,
      "loss": 0.3888,
      "step": 40621
    },
    {
      "epoch": 2.2747228133049613,
      "grad_norm": 2.1670773029327393,
      "learning_rate": 9.457815789473685e-05,
      "loss": 0.4272,
      "step": 40622
    },
    {
      "epoch": 2.2747788106170903,
      "grad_norm": 1.369737148284912,
      "learning_rate": 9.457789473684211e-05,
      "loss": 0.3345,
      "step": 40623
    },
    {
      "epoch": 2.2748348079292193,
      "grad_norm": 1.2595831155776978,
      "learning_rate": 9.457763157894737e-05,
      "loss": 0.587,
      "step": 40624
    },
    {
      "epoch": 2.2748908052413483,
      "grad_norm": 1.288218379020691,
      "learning_rate": 9.457736842105263e-05,
      "loss": 0.4151,
      "step": 40625
    },
    {
      "epoch": 2.2749468025534774,
      "grad_norm": 1.1666189432144165,
      "learning_rate": 9.457710526315789e-05,
      "loss": 0.3919,
      "step": 40626
    },
    {
      "epoch": 2.2750027998656064,
      "grad_norm": 1.8396273851394653,
      "learning_rate": 9.457684210526316e-05,
      "loss": 0.3797,
      "step": 40627
    },
    {
      "epoch": 2.2750587971777354,
      "grad_norm": 1.4156901836395264,
      "learning_rate": 9.457657894736842e-05,
      "loss": 0.3822,
      "step": 40628
    },
    {
      "epoch": 2.2751147944898644,
      "grad_norm": 1.175866723060608,
      "learning_rate": 9.45763157894737e-05,
      "loss": 0.4732,
      "step": 40629
    },
    {
      "epoch": 2.2751707918019934,
      "grad_norm": 1.1439058780670166,
      "learning_rate": 9.457605263157894e-05,
      "loss": 0.2878,
      "step": 40630
    },
    {
      "epoch": 2.2752267891141225,
      "grad_norm": 1.288842797279358,
      "learning_rate": 9.457578947368422e-05,
      "loss": 0.3905,
      "step": 40631
    },
    {
      "epoch": 2.2752827864262515,
      "grad_norm": 1.33756685256958,
      "learning_rate": 9.457552631578948e-05,
      "loss": 0.3562,
      "step": 40632
    },
    {
      "epoch": 2.2753387837383805,
      "grad_norm": 1.22990083694458,
      "learning_rate": 9.457526315789475e-05,
      "loss": 0.2758,
      "step": 40633
    },
    {
      "epoch": 2.2753947810505095,
      "grad_norm": 1.1799225807189941,
      "learning_rate": 9.457500000000001e-05,
      "loss": 0.5032,
      "step": 40634
    },
    {
      "epoch": 2.2754507783626385,
      "grad_norm": 1.7661281824111938,
      "learning_rate": 9.457473684210527e-05,
      "loss": 0.5858,
      "step": 40635
    },
    {
      "epoch": 2.2755067756747676,
      "grad_norm": 1.152991771697998,
      "learning_rate": 9.457447368421053e-05,
      "loss": 0.4368,
      "step": 40636
    },
    {
      "epoch": 2.2755627729868966,
      "grad_norm": 1.489823818206787,
      "learning_rate": 9.45742105263158e-05,
      "loss": 0.559,
      "step": 40637
    },
    {
      "epoch": 2.2756187702990256,
      "grad_norm": 1.209084391593933,
      "learning_rate": 9.457394736842106e-05,
      "loss": 0.4946,
      "step": 40638
    },
    {
      "epoch": 2.2756747676111546,
      "grad_norm": 1.1643140316009521,
      "learning_rate": 9.457368421052632e-05,
      "loss": 0.3482,
      "step": 40639
    },
    {
      "epoch": 2.2757307649232836,
      "grad_norm": 1.270938515663147,
      "learning_rate": 9.457342105263158e-05,
      "loss": 0.3369,
      "step": 40640
    },
    {
      "epoch": 2.2757867622354127,
      "grad_norm": 1.0088104009628296,
      "learning_rate": 9.457315789473684e-05,
      "loss": 0.2605,
      "step": 40641
    },
    {
      "epoch": 2.2758427595475417,
      "grad_norm": 1.1692498922348022,
      "learning_rate": 9.457289473684211e-05,
      "loss": 0.478,
      "step": 40642
    },
    {
      "epoch": 2.2758987568596707,
      "grad_norm": 1.4544577598571777,
      "learning_rate": 9.457263157894737e-05,
      "loss": 0.3604,
      "step": 40643
    },
    {
      "epoch": 2.2759547541717997,
      "grad_norm": 1.437093734741211,
      "learning_rate": 9.457236842105263e-05,
      "loss": 0.4022,
      "step": 40644
    },
    {
      "epoch": 2.2760107514839287,
      "grad_norm": 1.5041439533233643,
      "learning_rate": 9.457210526315789e-05,
      "loss": 0.4109,
      "step": 40645
    },
    {
      "epoch": 2.2760667487960577,
      "grad_norm": 1.5819566249847412,
      "learning_rate": 9.457184210526317e-05,
      "loss": 0.4624,
      "step": 40646
    },
    {
      "epoch": 2.2761227461081868,
      "grad_norm": 1.1363409757614136,
      "learning_rate": 9.457157894736843e-05,
      "loss": 0.4082,
      "step": 40647
    },
    {
      "epoch": 2.276178743420316,
      "grad_norm": 1.1435924768447876,
      "learning_rate": 9.457131578947369e-05,
      "loss": 0.4487,
      "step": 40648
    },
    {
      "epoch": 2.276234740732445,
      "grad_norm": 1.104856014251709,
      "learning_rate": 9.457105263157895e-05,
      "loss": 0.3443,
      "step": 40649
    },
    {
      "epoch": 2.276290738044574,
      "grad_norm": 1.4004374742507935,
      "learning_rate": 9.457078947368422e-05,
      "loss": 0.5505,
      "step": 40650
    },
    {
      "epoch": 2.276346735356703,
      "grad_norm": 1.9768024682998657,
      "learning_rate": 9.457052631578948e-05,
      "loss": 0.2911,
      "step": 40651
    },
    {
      "epoch": 2.276402732668832,
      "grad_norm": 1.1699367761611938,
      "learning_rate": 9.457026315789475e-05,
      "loss": 0.4692,
      "step": 40652
    },
    {
      "epoch": 2.276458729980961,
      "grad_norm": 1.182948350906372,
      "learning_rate": 9.457e-05,
      "loss": 0.4232,
      "step": 40653
    },
    {
      "epoch": 2.27651472729309,
      "grad_norm": 1.4045296907424927,
      "learning_rate": 9.456973684210527e-05,
      "loss": 0.3406,
      "step": 40654
    },
    {
      "epoch": 2.276570724605219,
      "grad_norm": 3.0734546184539795,
      "learning_rate": 9.456947368421053e-05,
      "loss": 0.5694,
      "step": 40655
    },
    {
      "epoch": 2.276626721917348,
      "grad_norm": 1.249147653579712,
      "learning_rate": 9.45692105263158e-05,
      "loss": 0.442,
      "step": 40656
    },
    {
      "epoch": 2.276682719229477,
      "grad_norm": 1.1361415386199951,
      "learning_rate": 9.456894736842106e-05,
      "loss": 0.4814,
      "step": 40657
    },
    {
      "epoch": 2.276738716541606,
      "grad_norm": 1.0875663757324219,
      "learning_rate": 9.456868421052631e-05,
      "loss": 0.3289,
      "step": 40658
    },
    {
      "epoch": 2.276794713853735,
      "grad_norm": 1.055518388748169,
      "learning_rate": 9.456842105263158e-05,
      "loss": 0.414,
      "step": 40659
    },
    {
      "epoch": 2.276850711165864,
      "grad_norm": 1.5574101209640503,
      "learning_rate": 9.456815789473684e-05,
      "loss": 0.4048,
      "step": 40660
    },
    {
      "epoch": 2.276906708477993,
      "grad_norm": 1.1019195318222046,
      "learning_rate": 9.456789473684212e-05,
      "loss": 0.3016,
      "step": 40661
    },
    {
      "epoch": 2.276962705790122,
      "grad_norm": 1.1658498048782349,
      "learning_rate": 9.456763157894736e-05,
      "loss": 0.4819,
      "step": 40662
    },
    {
      "epoch": 2.277018703102251,
      "grad_norm": 1.2599719762802124,
      "learning_rate": 9.456736842105264e-05,
      "loss": 0.378,
      "step": 40663
    },
    {
      "epoch": 2.27707470041438,
      "grad_norm": 1.1545414924621582,
      "learning_rate": 9.45671052631579e-05,
      "loss": 0.3704,
      "step": 40664
    },
    {
      "epoch": 2.277130697726509,
      "grad_norm": 1.086358904838562,
      "learning_rate": 9.456684210526317e-05,
      "loss": 0.3505,
      "step": 40665
    },
    {
      "epoch": 2.277186695038638,
      "grad_norm": 1.1106089353561401,
      "learning_rate": 9.456657894736843e-05,
      "loss": 0.3949,
      "step": 40666
    },
    {
      "epoch": 2.277242692350767,
      "grad_norm": 1.1631847620010376,
      "learning_rate": 9.456631578947369e-05,
      "loss": 0.3261,
      "step": 40667
    },
    {
      "epoch": 2.277298689662896,
      "grad_norm": 1.504260540008545,
      "learning_rate": 9.456605263157895e-05,
      "loss": 0.5226,
      "step": 40668
    },
    {
      "epoch": 2.277354686975025,
      "grad_norm": 1.1442697048187256,
      "learning_rate": 9.456578947368422e-05,
      "loss": 0.3669,
      "step": 40669
    },
    {
      "epoch": 2.277410684287154,
      "grad_norm": 1.2757880687713623,
      "learning_rate": 9.456552631578948e-05,
      "loss": 0.4514,
      "step": 40670
    },
    {
      "epoch": 2.2774666815992832,
      "grad_norm": 1.204907774925232,
      "learning_rate": 9.456526315789474e-05,
      "loss": 0.4126,
      "step": 40671
    },
    {
      "epoch": 2.2775226789114122,
      "grad_norm": 1.2308409214019775,
      "learning_rate": 9.4565e-05,
      "loss": 0.5865,
      "step": 40672
    },
    {
      "epoch": 2.2775786762235413,
      "grad_norm": 1.2333112955093384,
      "learning_rate": 9.456473684210527e-05,
      "loss": 0.371,
      "step": 40673
    },
    {
      "epoch": 2.2776346735356703,
      "grad_norm": 1.0535119771957397,
      "learning_rate": 9.456447368421053e-05,
      "loss": 0.3727,
      "step": 40674
    },
    {
      "epoch": 2.2776906708477993,
      "grad_norm": 1.2088721990585327,
      "learning_rate": 9.456421052631579e-05,
      "loss": 0.3457,
      "step": 40675
    },
    {
      "epoch": 2.2777466681599283,
      "grad_norm": 2.702401876449585,
      "learning_rate": 9.456394736842105e-05,
      "loss": 0.4453,
      "step": 40676
    },
    {
      "epoch": 2.2778026654720573,
      "grad_norm": 1.2911460399627686,
      "learning_rate": 9.456368421052631e-05,
      "loss": 0.5286,
      "step": 40677
    },
    {
      "epoch": 2.2778586627841864,
      "grad_norm": 1.0863419771194458,
      "learning_rate": 9.456342105263159e-05,
      "loss": 0.4829,
      "step": 40678
    },
    {
      "epoch": 2.2779146600963154,
      "grad_norm": 1.0396877527236938,
      "learning_rate": 9.456315789473685e-05,
      "loss": 0.3943,
      "step": 40679
    },
    {
      "epoch": 2.2779706574084444,
      "grad_norm": 1.3092750310897827,
      "learning_rate": 9.45628947368421e-05,
      "loss": 0.4931,
      "step": 40680
    },
    {
      "epoch": 2.2780266547205734,
      "grad_norm": 2.3488402366638184,
      "learning_rate": 9.456263157894736e-05,
      "loss": 0.3927,
      "step": 40681
    },
    {
      "epoch": 2.2780826520327024,
      "grad_norm": 1.1009044647216797,
      "learning_rate": 9.456236842105264e-05,
      "loss": 0.311,
      "step": 40682
    },
    {
      "epoch": 2.2781386493448315,
      "grad_norm": 1.2242518663406372,
      "learning_rate": 9.45621052631579e-05,
      "loss": 0.5009,
      "step": 40683
    },
    {
      "epoch": 2.2781946466569605,
      "grad_norm": 1.48647940158844,
      "learning_rate": 9.456184210526317e-05,
      "loss": 0.4011,
      "step": 40684
    },
    {
      "epoch": 2.2782506439690895,
      "grad_norm": 1.2681246995925903,
      "learning_rate": 9.456157894736842e-05,
      "loss": 0.4263,
      "step": 40685
    },
    {
      "epoch": 2.2783066412812185,
      "grad_norm": 1.281705379486084,
      "learning_rate": 9.456131578947369e-05,
      "loss": 0.334,
      "step": 40686
    },
    {
      "epoch": 2.2783626385933475,
      "grad_norm": 1.189846396446228,
      "learning_rate": 9.456105263157895e-05,
      "loss": 0.3196,
      "step": 40687
    },
    {
      "epoch": 2.2784186359054766,
      "grad_norm": 1.3390530347824097,
      "learning_rate": 9.456078947368422e-05,
      "loss": 0.3496,
      "step": 40688
    },
    {
      "epoch": 2.2784746332176056,
      "grad_norm": 0.9478362798690796,
      "learning_rate": 9.456052631578948e-05,
      "loss": 0.3264,
      "step": 40689
    },
    {
      "epoch": 2.2785306305297346,
      "grad_norm": 1.1544431447982788,
      "learning_rate": 9.456026315789474e-05,
      "loss": 0.4445,
      "step": 40690
    },
    {
      "epoch": 2.2785866278418636,
      "grad_norm": 1.1023313999176025,
      "learning_rate": 9.456e-05,
      "loss": 0.2768,
      "step": 40691
    },
    {
      "epoch": 2.2786426251539926,
      "grad_norm": 1.0048623085021973,
      "learning_rate": 9.455973684210528e-05,
      "loss": 0.3045,
      "step": 40692
    },
    {
      "epoch": 2.2786986224661216,
      "grad_norm": 1.1986424922943115,
      "learning_rate": 9.455947368421054e-05,
      "loss": 0.3915,
      "step": 40693
    },
    {
      "epoch": 2.2787546197782507,
      "grad_norm": 1.5429383516311646,
      "learning_rate": 9.45592105263158e-05,
      "loss": 0.3852,
      "step": 40694
    },
    {
      "epoch": 2.2788106170903797,
      "grad_norm": 1.5468636751174927,
      "learning_rate": 9.455894736842106e-05,
      "loss": 0.5161,
      "step": 40695
    },
    {
      "epoch": 2.2788666144025087,
      "grad_norm": 1.1783115863800049,
      "learning_rate": 9.455868421052631e-05,
      "loss": 0.3406,
      "step": 40696
    },
    {
      "epoch": 2.2789226117146377,
      "grad_norm": 1.3493643999099731,
      "learning_rate": 9.455842105263159e-05,
      "loss": 0.3999,
      "step": 40697
    },
    {
      "epoch": 2.2789786090267667,
      "grad_norm": 1.3171998262405396,
      "learning_rate": 9.455815789473685e-05,
      "loss": 0.3947,
      "step": 40698
    },
    {
      "epoch": 2.2790346063388958,
      "grad_norm": 1.8221908807754517,
      "learning_rate": 9.455789473684211e-05,
      "loss": 0.4198,
      "step": 40699
    },
    {
      "epoch": 2.279090603651025,
      "grad_norm": 1.0413686037063599,
      "learning_rate": 9.455763157894737e-05,
      "loss": 0.3787,
      "step": 40700
    },
    {
      "epoch": 2.279146600963154,
      "grad_norm": 1.8679466247558594,
      "learning_rate": 9.455736842105264e-05,
      "loss": 0.3968,
      "step": 40701
    },
    {
      "epoch": 2.279202598275283,
      "grad_norm": 1.2845100164413452,
      "learning_rate": 9.45571052631579e-05,
      "loss": 0.3852,
      "step": 40702
    },
    {
      "epoch": 2.279258595587412,
      "grad_norm": 1.473031759262085,
      "learning_rate": 9.455684210526316e-05,
      "loss": 0.3869,
      "step": 40703
    },
    {
      "epoch": 2.279314592899541,
      "grad_norm": 1.3209993839263916,
      "learning_rate": 9.455657894736842e-05,
      "loss": 0.5044,
      "step": 40704
    },
    {
      "epoch": 2.27937059021167,
      "grad_norm": 1.190691351890564,
      "learning_rate": 9.455631578947369e-05,
      "loss": 0.3455,
      "step": 40705
    },
    {
      "epoch": 2.279426587523799,
      "grad_norm": 1.1404445171356201,
      "learning_rate": 9.455605263157895e-05,
      "loss": 0.3455,
      "step": 40706
    },
    {
      "epoch": 2.279482584835928,
      "grad_norm": 1.1950658559799194,
      "learning_rate": 9.455578947368423e-05,
      "loss": 0.4867,
      "step": 40707
    },
    {
      "epoch": 2.279538582148057,
      "grad_norm": 1.141538381576538,
      "learning_rate": 9.455552631578947e-05,
      "loss": 0.4024,
      "step": 40708
    },
    {
      "epoch": 2.279594579460186,
      "grad_norm": 1.139651894569397,
      "learning_rate": 9.455526315789475e-05,
      "loss": 0.3851,
      "step": 40709
    },
    {
      "epoch": 2.279650576772315,
      "grad_norm": 2.500432252883911,
      "learning_rate": 9.4555e-05,
      "loss": 0.4448,
      "step": 40710
    },
    {
      "epoch": 2.279706574084444,
      "grad_norm": 1.292118787765503,
      "learning_rate": 9.455473684210526e-05,
      "loss": 0.3999,
      "step": 40711
    },
    {
      "epoch": 2.279762571396573,
      "grad_norm": 1.3038464784622192,
      "learning_rate": 9.455447368421054e-05,
      "loss": 0.4593,
      "step": 40712
    },
    {
      "epoch": 2.279818568708702,
      "grad_norm": 1.586011290550232,
      "learning_rate": 9.455421052631578e-05,
      "loss": 0.6508,
      "step": 40713
    },
    {
      "epoch": 2.279874566020831,
      "grad_norm": 1.3113987445831299,
      "learning_rate": 9.455394736842106e-05,
      "loss": 0.4406,
      "step": 40714
    },
    {
      "epoch": 2.27993056333296,
      "grad_norm": 1.0662182569503784,
      "learning_rate": 9.455368421052632e-05,
      "loss": 0.2824,
      "step": 40715
    },
    {
      "epoch": 2.279986560645089,
      "grad_norm": 1.3647671937942505,
      "learning_rate": 9.455342105263159e-05,
      "loss": 0.4789,
      "step": 40716
    },
    {
      "epoch": 2.280042557957218,
      "grad_norm": 1.1081174612045288,
      "learning_rate": 9.455315789473684e-05,
      "loss": 0.3911,
      "step": 40717
    },
    {
      "epoch": 2.280098555269347,
      "grad_norm": 1.1749792098999023,
      "learning_rate": 9.455289473684211e-05,
      "loss": 0.4746,
      "step": 40718
    },
    {
      "epoch": 2.280154552581476,
      "grad_norm": 1.1469017267227173,
      "learning_rate": 9.455263157894737e-05,
      "loss": 0.4871,
      "step": 40719
    },
    {
      "epoch": 2.280210549893605,
      "grad_norm": 1.0866130590438843,
      "learning_rate": 9.455236842105264e-05,
      "loss": 0.3739,
      "step": 40720
    },
    {
      "epoch": 2.280266547205734,
      "grad_norm": 1.3231158256530762,
      "learning_rate": 9.45521052631579e-05,
      "loss": 0.452,
      "step": 40721
    },
    {
      "epoch": 2.280322544517863,
      "grad_norm": 1.2544708251953125,
      "learning_rate": 9.455184210526316e-05,
      "loss": 0.4238,
      "step": 40722
    },
    {
      "epoch": 2.2803785418299922,
      "grad_norm": 1.5187524557113647,
      "learning_rate": 9.455157894736842e-05,
      "loss": 0.3369,
      "step": 40723
    },
    {
      "epoch": 2.2804345391421212,
      "grad_norm": 1.0395478010177612,
      "learning_rate": 9.45513157894737e-05,
      "loss": 0.3294,
      "step": 40724
    },
    {
      "epoch": 2.2804905364542503,
      "grad_norm": 1.2137303352355957,
      "learning_rate": 9.455105263157896e-05,
      "loss": 0.4857,
      "step": 40725
    },
    {
      "epoch": 2.2805465337663793,
      "grad_norm": 1.1185930967330933,
      "learning_rate": 9.455078947368422e-05,
      "loss": 0.4514,
      "step": 40726
    },
    {
      "epoch": 2.2806025310785083,
      "grad_norm": 1.0367804765701294,
      "learning_rate": 9.455052631578947e-05,
      "loss": 0.4904,
      "step": 40727
    },
    {
      "epoch": 2.2806585283906373,
      "grad_norm": 1.1859992742538452,
      "learning_rate": 9.455026315789473e-05,
      "loss": 0.4533,
      "step": 40728
    },
    {
      "epoch": 2.2807145257027663,
      "grad_norm": 1.1731269359588623,
      "learning_rate": 9.455000000000001e-05,
      "loss": 0.3564,
      "step": 40729
    },
    {
      "epoch": 2.2807705230148954,
      "grad_norm": 1.441381573677063,
      "learning_rate": 9.454973684210527e-05,
      "loss": 0.6787,
      "step": 40730
    },
    {
      "epoch": 2.2808265203270244,
      "grad_norm": 1.266228437423706,
      "learning_rate": 9.454947368421053e-05,
      "loss": 0.4046,
      "step": 40731
    },
    {
      "epoch": 2.2808825176391534,
      "grad_norm": 1.1656194925308228,
      "learning_rate": 9.454921052631579e-05,
      "loss": 0.4161,
      "step": 40732
    },
    {
      "epoch": 2.2809385149512824,
      "grad_norm": 0.9893287420272827,
      "learning_rate": 9.454894736842106e-05,
      "loss": 0.3206,
      "step": 40733
    },
    {
      "epoch": 2.2809945122634114,
      "grad_norm": 1.1631611585617065,
      "learning_rate": 9.454868421052632e-05,
      "loss": 0.4332,
      "step": 40734
    },
    {
      "epoch": 2.2810505095755405,
      "grad_norm": 1.1853488683700562,
      "learning_rate": 9.454842105263158e-05,
      "loss": 0.4037,
      "step": 40735
    },
    {
      "epoch": 2.2811065068876695,
      "grad_norm": 1.2882390022277832,
      "learning_rate": 9.454815789473684e-05,
      "loss": 0.452,
      "step": 40736
    },
    {
      "epoch": 2.2811625041997985,
      "grad_norm": 1.1939971446990967,
      "learning_rate": 9.454789473684211e-05,
      "loss": 0.3749,
      "step": 40737
    },
    {
      "epoch": 2.2812185015119275,
      "grad_norm": 1.4257616996765137,
      "learning_rate": 9.454763157894737e-05,
      "loss": 0.7062,
      "step": 40738
    },
    {
      "epoch": 2.2812744988240565,
      "grad_norm": 1.230320930480957,
      "learning_rate": 9.454736842105265e-05,
      "loss": 0.491,
      "step": 40739
    },
    {
      "epoch": 2.2813304961361855,
      "grad_norm": 1.7072973251342773,
      "learning_rate": 9.454710526315789e-05,
      "loss": 0.6571,
      "step": 40740
    },
    {
      "epoch": 2.2813864934483146,
      "grad_norm": 1.2500269412994385,
      "learning_rate": 9.454684210526317e-05,
      "loss": 0.409,
      "step": 40741
    },
    {
      "epoch": 2.2814424907604436,
      "grad_norm": 1.2978328466415405,
      "learning_rate": 9.454657894736842e-05,
      "loss": 0.4459,
      "step": 40742
    },
    {
      "epoch": 2.2814984880725726,
      "grad_norm": 1.1234685182571411,
      "learning_rate": 9.45463157894737e-05,
      "loss": 0.3741,
      "step": 40743
    },
    {
      "epoch": 2.2815544853847016,
      "grad_norm": 1.0350223779678345,
      "learning_rate": 9.454605263157896e-05,
      "loss": 0.3605,
      "step": 40744
    },
    {
      "epoch": 2.2816104826968306,
      "grad_norm": 0.9962635040283203,
      "learning_rate": 9.45457894736842e-05,
      "loss": 0.3641,
      "step": 40745
    },
    {
      "epoch": 2.2816664800089597,
      "grad_norm": 1.1440106630325317,
      "learning_rate": 9.454552631578948e-05,
      "loss": 0.4514,
      "step": 40746
    },
    {
      "epoch": 2.2817224773210887,
      "grad_norm": 1.0001720190048218,
      "learning_rate": 9.454526315789474e-05,
      "loss": 0.3663,
      "step": 40747
    },
    {
      "epoch": 2.2817784746332177,
      "grad_norm": 1.0957376956939697,
      "learning_rate": 9.454500000000001e-05,
      "loss": 0.3766,
      "step": 40748
    },
    {
      "epoch": 2.2818344719453467,
      "grad_norm": 1.0857912302017212,
      "learning_rate": 9.454473684210527e-05,
      "loss": 0.3578,
      "step": 40749
    },
    {
      "epoch": 2.2818904692574757,
      "grad_norm": 1.0717769861221313,
      "learning_rate": 9.454447368421053e-05,
      "loss": 0.3777,
      "step": 40750
    },
    {
      "epoch": 2.2819464665696048,
      "grad_norm": 1.1021376848220825,
      "learning_rate": 9.454421052631579e-05,
      "loss": 0.3118,
      "step": 40751
    },
    {
      "epoch": 2.282002463881734,
      "grad_norm": 1.1121015548706055,
      "learning_rate": 9.454394736842106e-05,
      "loss": 0.4028,
      "step": 40752
    },
    {
      "epoch": 2.282058461193863,
      "grad_norm": 1.2617932558059692,
      "learning_rate": 9.454368421052632e-05,
      "loss": 0.3605,
      "step": 40753
    },
    {
      "epoch": 2.282114458505992,
      "grad_norm": 1.328307032585144,
      "learning_rate": 9.454342105263158e-05,
      "loss": 0.4156,
      "step": 40754
    },
    {
      "epoch": 2.282170455818121,
      "grad_norm": 1.6096069812774658,
      "learning_rate": 9.454315789473684e-05,
      "loss": 0.5602,
      "step": 40755
    },
    {
      "epoch": 2.28222645313025,
      "grad_norm": 1.155880093574524,
      "learning_rate": 9.454289473684212e-05,
      "loss": 0.3975,
      "step": 40756
    },
    {
      "epoch": 2.282282450442379,
      "grad_norm": 1.1766146421432495,
      "learning_rate": 9.454263157894738e-05,
      "loss": 0.3232,
      "step": 40757
    },
    {
      "epoch": 2.282338447754508,
      "grad_norm": 1.078818440437317,
      "learning_rate": 9.454236842105263e-05,
      "loss": 0.3293,
      "step": 40758
    },
    {
      "epoch": 2.282394445066637,
      "grad_norm": 1.1642459630966187,
      "learning_rate": 9.45421052631579e-05,
      "loss": 0.5092,
      "step": 40759
    },
    {
      "epoch": 2.282450442378766,
      "grad_norm": 1.0969308614730835,
      "learning_rate": 9.454184210526317e-05,
      "loss": 0.4029,
      "step": 40760
    },
    {
      "epoch": 2.282506439690895,
      "grad_norm": 1.1417568922042847,
      "learning_rate": 9.454157894736843e-05,
      "loss": 0.4593,
      "step": 40761
    },
    {
      "epoch": 2.282562437003024,
      "grad_norm": 1.3384737968444824,
      "learning_rate": 9.454131578947369e-05,
      "loss": 0.3998,
      "step": 40762
    },
    {
      "epoch": 2.282618434315153,
      "grad_norm": 1.0474281311035156,
      "learning_rate": 9.454105263157895e-05,
      "loss": 0.3799,
      "step": 40763
    },
    {
      "epoch": 2.282674431627282,
      "grad_norm": 1.0527393817901611,
      "learning_rate": 9.45407894736842e-05,
      "loss": 0.3328,
      "step": 40764
    },
    {
      "epoch": 2.282730428939411,
      "grad_norm": 1.5475727319717407,
      "learning_rate": 9.454052631578948e-05,
      "loss": 0.5563,
      "step": 40765
    },
    {
      "epoch": 2.28278642625154,
      "grad_norm": 1.1792128086090088,
      "learning_rate": 9.454026315789474e-05,
      "loss": 0.3018,
      "step": 40766
    },
    {
      "epoch": 2.282842423563669,
      "grad_norm": 1.3407564163208008,
      "learning_rate": 9.454000000000001e-05,
      "loss": 0.5293,
      "step": 40767
    },
    {
      "epoch": 2.282898420875798,
      "grad_norm": 1.006574273109436,
      "learning_rate": 9.453973684210526e-05,
      "loss": 0.3691,
      "step": 40768
    },
    {
      "epoch": 2.282954418187927,
      "grad_norm": 1.1501902341842651,
      "learning_rate": 9.453947368421053e-05,
      "loss": 0.4222,
      "step": 40769
    },
    {
      "epoch": 2.283010415500056,
      "grad_norm": 1.2683444023132324,
      "learning_rate": 9.453921052631579e-05,
      "loss": 0.4697,
      "step": 40770
    },
    {
      "epoch": 2.283066412812185,
      "grad_norm": 1.2929222583770752,
      "learning_rate": 9.453894736842107e-05,
      "loss": 0.5122,
      "step": 40771
    },
    {
      "epoch": 2.283122410124314,
      "grad_norm": 1.339768886566162,
      "learning_rate": 9.453868421052631e-05,
      "loss": 0.4908,
      "step": 40772
    },
    {
      "epoch": 2.283178407436443,
      "grad_norm": 1.1752787828445435,
      "learning_rate": 9.453842105263158e-05,
      "loss": 0.336,
      "step": 40773
    },
    {
      "epoch": 2.283234404748572,
      "grad_norm": 1.1361424922943115,
      "learning_rate": 9.453815789473684e-05,
      "loss": 0.3768,
      "step": 40774
    },
    {
      "epoch": 2.283290402060701,
      "grad_norm": 1.2001701593399048,
      "learning_rate": 9.453789473684212e-05,
      "loss": 0.4433,
      "step": 40775
    },
    {
      "epoch": 2.2833463993728302,
      "grad_norm": 1.187086582183838,
      "learning_rate": 9.453763157894738e-05,
      "loss": 0.3738,
      "step": 40776
    },
    {
      "epoch": 2.2834023966849593,
      "grad_norm": 1.1378377676010132,
      "learning_rate": 9.453736842105264e-05,
      "loss": 0.3925,
      "step": 40777
    },
    {
      "epoch": 2.2834583939970883,
      "grad_norm": 1.069530963897705,
      "learning_rate": 9.45371052631579e-05,
      "loss": 0.3852,
      "step": 40778
    },
    {
      "epoch": 2.2835143913092173,
      "grad_norm": 1.092367172241211,
      "learning_rate": 9.453684210526316e-05,
      "loss": 0.3679,
      "step": 40779
    },
    {
      "epoch": 2.2835703886213463,
      "grad_norm": 1.2742834091186523,
      "learning_rate": 9.453657894736843e-05,
      "loss": 0.475,
      "step": 40780
    },
    {
      "epoch": 2.2836263859334753,
      "grad_norm": 1.0827237367630005,
      "learning_rate": 9.453631578947369e-05,
      "loss": 0.4066,
      "step": 40781
    },
    {
      "epoch": 2.2836823832456044,
      "grad_norm": 1.3026787042617798,
      "learning_rate": 9.453605263157895e-05,
      "loss": 0.5782,
      "step": 40782
    },
    {
      "epoch": 2.2837383805577334,
      "grad_norm": 1.1385407447814941,
      "learning_rate": 9.453578947368421e-05,
      "loss": 0.3384,
      "step": 40783
    },
    {
      "epoch": 2.2837943778698624,
      "grad_norm": 1.182676911354065,
      "learning_rate": 9.453552631578948e-05,
      "loss": 0.4198,
      "step": 40784
    },
    {
      "epoch": 2.2838503751819914,
      "grad_norm": 1.4524996280670166,
      "learning_rate": 9.453526315789474e-05,
      "loss": 0.3952,
      "step": 40785
    },
    {
      "epoch": 2.2839063724941204,
      "grad_norm": 1.1759510040283203,
      "learning_rate": 9.4535e-05,
      "loss": 0.3721,
      "step": 40786
    },
    {
      "epoch": 2.2839623698062494,
      "grad_norm": 1.1353518962860107,
      "learning_rate": 9.453473684210526e-05,
      "loss": 0.4739,
      "step": 40787
    },
    {
      "epoch": 2.2840183671183785,
      "grad_norm": 1.1772749423980713,
      "learning_rate": 9.453447368421054e-05,
      "loss": 0.3984,
      "step": 40788
    },
    {
      "epoch": 2.2840743644305075,
      "grad_norm": 1.1025346517562866,
      "learning_rate": 9.45342105263158e-05,
      "loss": 0.365,
      "step": 40789
    },
    {
      "epoch": 2.2841303617426365,
      "grad_norm": 1.0741193294525146,
      "learning_rate": 9.453394736842105e-05,
      "loss": 0.4014,
      "step": 40790
    },
    {
      "epoch": 2.2841863590547655,
      "grad_norm": 1.2959717512130737,
      "learning_rate": 9.453368421052631e-05,
      "loss": 0.5235,
      "step": 40791
    },
    {
      "epoch": 2.2842423563668945,
      "grad_norm": 1.1121904850006104,
      "learning_rate": 9.453342105263159e-05,
      "loss": 0.3379,
      "step": 40792
    },
    {
      "epoch": 2.2842983536790236,
      "grad_norm": 1.1972206830978394,
      "learning_rate": 9.453315789473685e-05,
      "loss": 0.4776,
      "step": 40793
    },
    {
      "epoch": 2.2843543509911526,
      "grad_norm": 1.1832555532455444,
      "learning_rate": 9.453289473684212e-05,
      "loss": 0.4313,
      "step": 40794
    },
    {
      "epoch": 2.2844103483032816,
      "grad_norm": 1.0814447402954102,
      "learning_rate": 9.453263157894737e-05,
      "loss": 0.3868,
      "step": 40795
    },
    {
      "epoch": 2.2844663456154106,
      "grad_norm": 1.6931625604629517,
      "learning_rate": 9.453236842105263e-05,
      "loss": 0.3943,
      "step": 40796
    },
    {
      "epoch": 2.2845223429275396,
      "grad_norm": 1.4816912412643433,
      "learning_rate": 9.45321052631579e-05,
      "loss": 0.428,
      "step": 40797
    },
    {
      "epoch": 2.2845783402396687,
      "grad_norm": 1.384658694267273,
      "learning_rate": 9.453184210526316e-05,
      "loss": 0.545,
      "step": 40798
    },
    {
      "epoch": 2.2846343375517977,
      "grad_norm": 1.3358010053634644,
      "learning_rate": 9.453157894736843e-05,
      "loss": 0.3701,
      "step": 40799
    },
    {
      "epoch": 2.2846903348639267,
      "grad_norm": 1.3422331809997559,
      "learning_rate": 9.453131578947368e-05,
      "loss": 0.3349,
      "step": 40800
    },
    {
      "epoch": 2.2847463321760557,
      "grad_norm": 1.3526791334152222,
      "learning_rate": 9.453105263157895e-05,
      "loss": 0.3471,
      "step": 40801
    },
    {
      "epoch": 2.2848023294881847,
      "grad_norm": 1.410909652709961,
      "learning_rate": 9.453078947368421e-05,
      "loss": 0.4392,
      "step": 40802
    },
    {
      "epoch": 2.2848583268003138,
      "grad_norm": 1.245595932006836,
      "learning_rate": 9.453052631578949e-05,
      "loss": 0.5317,
      "step": 40803
    },
    {
      "epoch": 2.2849143241124428,
      "grad_norm": 1.268906831741333,
      "learning_rate": 9.453026315789474e-05,
      "loss": 0.3763,
      "step": 40804
    },
    {
      "epoch": 2.284970321424572,
      "grad_norm": 1.3148695230484009,
      "learning_rate": 9.453e-05,
      "loss": 0.3801,
      "step": 40805
    },
    {
      "epoch": 2.285026318736701,
      "grad_norm": 1.3297442197799683,
      "learning_rate": 9.452973684210526e-05,
      "loss": 0.5929,
      "step": 40806
    },
    {
      "epoch": 2.28508231604883,
      "grad_norm": 1.0440672636032104,
      "learning_rate": 9.452947368421054e-05,
      "loss": 0.4379,
      "step": 40807
    },
    {
      "epoch": 2.285138313360959,
      "grad_norm": 1.1177469491958618,
      "learning_rate": 9.45292105263158e-05,
      "loss": 0.4573,
      "step": 40808
    },
    {
      "epoch": 2.285194310673088,
      "grad_norm": 1.1262434720993042,
      "learning_rate": 9.452894736842106e-05,
      "loss": 0.3983,
      "step": 40809
    },
    {
      "epoch": 2.285250307985217,
      "grad_norm": 0.913974404335022,
      "learning_rate": 9.452868421052632e-05,
      "loss": 0.3562,
      "step": 40810
    },
    {
      "epoch": 2.285306305297346,
      "grad_norm": 1.361995816230774,
      "learning_rate": 9.452842105263159e-05,
      "loss": 0.4027,
      "step": 40811
    },
    {
      "epoch": 2.285362302609475,
      "grad_norm": 1.4298921823501587,
      "learning_rate": 9.452815789473685e-05,
      "loss": 0.4842,
      "step": 40812
    },
    {
      "epoch": 2.285418299921604,
      "grad_norm": 1.1082196235656738,
      "learning_rate": 9.452789473684211e-05,
      "loss": 0.4625,
      "step": 40813
    },
    {
      "epoch": 2.285474297233733,
      "grad_norm": 1.6947755813598633,
      "learning_rate": 9.452763157894737e-05,
      "loss": 0.4076,
      "step": 40814
    },
    {
      "epoch": 2.285530294545862,
      "grad_norm": 2.012397289276123,
      "learning_rate": 9.452736842105263e-05,
      "loss": 0.5032,
      "step": 40815
    },
    {
      "epoch": 2.285586291857991,
      "grad_norm": 1.3171069622039795,
      "learning_rate": 9.45271052631579e-05,
      "loss": 0.5199,
      "step": 40816
    },
    {
      "epoch": 2.28564228917012,
      "grad_norm": 1.0655114650726318,
      "learning_rate": 9.452684210526316e-05,
      "loss": 0.358,
      "step": 40817
    },
    {
      "epoch": 2.285698286482249,
      "grad_norm": 1.25496506690979,
      "learning_rate": 9.452657894736842e-05,
      "loss": 0.4259,
      "step": 40818
    },
    {
      "epoch": 2.285754283794378,
      "grad_norm": 1.226019024848938,
      "learning_rate": 9.452631578947368e-05,
      "loss": 0.4369,
      "step": 40819
    },
    {
      "epoch": 2.285810281106507,
      "grad_norm": 1.29500412940979,
      "learning_rate": 9.452605263157895e-05,
      "loss": 0.3374,
      "step": 40820
    },
    {
      "epoch": 2.285866278418636,
      "grad_norm": 1.312785029411316,
      "learning_rate": 9.452578947368421e-05,
      "loss": 0.5795,
      "step": 40821
    },
    {
      "epoch": 2.2859222757307647,
      "grad_norm": 1.2990490198135376,
      "learning_rate": 9.452552631578949e-05,
      "loss": 0.5413,
      "step": 40822
    },
    {
      "epoch": 2.285978273042894,
      "grad_norm": 1.215506672859192,
      "learning_rate": 9.452526315789473e-05,
      "loss": 0.4853,
      "step": 40823
    },
    {
      "epoch": 2.2860342703550227,
      "grad_norm": 1.4231382608413696,
      "learning_rate": 9.452500000000001e-05,
      "loss": 0.4386,
      "step": 40824
    },
    {
      "epoch": 2.286090267667152,
      "grad_norm": 1.7093122005462646,
      "learning_rate": 9.452473684210527e-05,
      "loss": 0.4625,
      "step": 40825
    },
    {
      "epoch": 2.2861462649792808,
      "grad_norm": 1.0981152057647705,
      "learning_rate": 9.452447368421054e-05,
      "loss": 0.4368,
      "step": 40826
    },
    {
      "epoch": 2.28620226229141,
      "grad_norm": 1.358931064605713,
      "learning_rate": 9.452421052631579e-05,
      "loss": 0.351,
      "step": 40827
    },
    {
      "epoch": 2.286258259603539,
      "grad_norm": 1.16800856590271,
      "learning_rate": 9.452394736842106e-05,
      "loss": 0.4189,
      "step": 40828
    },
    {
      "epoch": 2.2863142569156683,
      "grad_norm": 1.1592634916305542,
      "learning_rate": 9.452368421052632e-05,
      "loss": 0.5028,
      "step": 40829
    },
    {
      "epoch": 2.286370254227797,
      "grad_norm": 1.0210527181625366,
      "learning_rate": 9.452342105263159e-05,
      "loss": 0.3342,
      "step": 40830
    },
    {
      "epoch": 2.2864262515399263,
      "grad_norm": 1.3599320650100708,
      "learning_rate": 9.452315789473685e-05,
      "loss": 0.435,
      "step": 40831
    },
    {
      "epoch": 2.286482248852055,
      "grad_norm": 1.6425976753234863,
      "learning_rate": 9.45228947368421e-05,
      "loss": 0.6303,
      "step": 40832
    },
    {
      "epoch": 2.2865382461641843,
      "grad_norm": 1.7112220525741577,
      "learning_rate": 9.452263157894737e-05,
      "loss": 0.534,
      "step": 40833
    },
    {
      "epoch": 2.286594243476313,
      "grad_norm": 1.0830254554748535,
      "learning_rate": 9.452236842105263e-05,
      "loss": 0.3632,
      "step": 40834
    },
    {
      "epoch": 2.2866502407884424,
      "grad_norm": 1.2991468906402588,
      "learning_rate": 9.45221052631579e-05,
      "loss": 0.4075,
      "step": 40835
    },
    {
      "epoch": 2.286706238100571,
      "grad_norm": 1.2036579847335815,
      "learning_rate": 9.452184210526316e-05,
      "loss": 0.4785,
      "step": 40836
    },
    {
      "epoch": 2.2867622354127004,
      "grad_norm": 1.0331439971923828,
      "learning_rate": 9.452157894736842e-05,
      "loss": 0.42,
      "step": 40837
    },
    {
      "epoch": 2.286818232724829,
      "grad_norm": 1.3257039785385132,
      "learning_rate": 9.452131578947368e-05,
      "loss": 0.5174,
      "step": 40838
    },
    {
      "epoch": 2.2868742300369584,
      "grad_norm": 1.1616357564926147,
      "learning_rate": 9.452105263157896e-05,
      "loss": 0.4571,
      "step": 40839
    },
    {
      "epoch": 2.286930227349087,
      "grad_norm": 1.7315592765808105,
      "learning_rate": 9.452078947368422e-05,
      "loss": 0.3595,
      "step": 40840
    },
    {
      "epoch": 2.2869862246612165,
      "grad_norm": 1.304601788520813,
      "learning_rate": 9.452052631578948e-05,
      "loss": 0.3848,
      "step": 40841
    },
    {
      "epoch": 2.287042221973345,
      "grad_norm": 1.7512010335922241,
      "learning_rate": 9.452026315789474e-05,
      "loss": 0.3855,
      "step": 40842
    },
    {
      "epoch": 2.2870982192854745,
      "grad_norm": 1.8160886764526367,
      "learning_rate": 9.452000000000001e-05,
      "loss": 0.3164,
      "step": 40843
    },
    {
      "epoch": 2.287154216597603,
      "grad_norm": 1.1015946865081787,
      "learning_rate": 9.451973684210527e-05,
      "loss": 0.413,
      "step": 40844
    },
    {
      "epoch": 2.2872102139097326,
      "grad_norm": 1.353179931640625,
      "learning_rate": 9.451947368421053e-05,
      "loss": 0.4529,
      "step": 40845
    },
    {
      "epoch": 2.287266211221861,
      "grad_norm": 1.2537550926208496,
      "learning_rate": 9.451921052631579e-05,
      "loss": 0.4895,
      "step": 40846
    },
    {
      "epoch": 2.2873222085339906,
      "grad_norm": 1.1315968036651611,
      "learning_rate": 9.451894736842106e-05,
      "loss": 0.4106,
      "step": 40847
    },
    {
      "epoch": 2.287378205846119,
      "grad_norm": 1.1098504066467285,
      "learning_rate": 9.451868421052632e-05,
      "loss": 0.4442,
      "step": 40848
    },
    {
      "epoch": 2.2874342031582486,
      "grad_norm": 1.4122551679611206,
      "learning_rate": 9.451842105263158e-05,
      "loss": 0.628,
      "step": 40849
    },
    {
      "epoch": 2.287490200470377,
      "grad_norm": 1.446104645729065,
      "learning_rate": 9.451815789473684e-05,
      "loss": 0.472,
      "step": 40850
    },
    {
      "epoch": 2.2875461977825067,
      "grad_norm": 1.0111216306686401,
      "learning_rate": 9.45178947368421e-05,
      "loss": 0.3676,
      "step": 40851
    },
    {
      "epoch": 2.2876021950946352,
      "grad_norm": 3.5224239826202393,
      "learning_rate": 9.451763157894737e-05,
      "loss": 0.3782,
      "step": 40852
    },
    {
      "epoch": 2.2876581924067647,
      "grad_norm": 1.0094807147979736,
      "learning_rate": 9.451736842105263e-05,
      "loss": 0.3587,
      "step": 40853
    },
    {
      "epoch": 2.2877141897188933,
      "grad_norm": 1.1357190608978271,
      "learning_rate": 9.451710526315791e-05,
      "loss": 0.4085,
      "step": 40854
    },
    {
      "epoch": 2.2877701870310228,
      "grad_norm": 1.5335227251052856,
      "learning_rate": 9.451684210526315e-05,
      "loss": 0.5235,
      "step": 40855
    },
    {
      "epoch": 2.2878261843431513,
      "grad_norm": 1.2523075342178345,
      "learning_rate": 9.451657894736843e-05,
      "loss": 0.444,
      "step": 40856
    },
    {
      "epoch": 2.287882181655281,
      "grad_norm": 1.159908652305603,
      "learning_rate": 9.451631578947369e-05,
      "loss": 0.4259,
      "step": 40857
    },
    {
      "epoch": 2.2879381789674094,
      "grad_norm": 1.0728968381881714,
      "learning_rate": 9.451605263157896e-05,
      "loss": 0.3896,
      "step": 40858
    },
    {
      "epoch": 2.287994176279539,
      "grad_norm": 1.1651214361190796,
      "learning_rate": 9.451578947368422e-05,
      "loss": 0.3782,
      "step": 40859
    },
    {
      "epoch": 2.2880501735916674,
      "grad_norm": 1.0190309286117554,
      "learning_rate": 9.451552631578948e-05,
      "loss": 0.3151,
      "step": 40860
    },
    {
      "epoch": 2.288106170903797,
      "grad_norm": 1.1230497360229492,
      "learning_rate": 9.451526315789474e-05,
      "loss": 0.413,
      "step": 40861
    },
    {
      "epoch": 2.2881621682159254,
      "grad_norm": 1.2884025573730469,
      "learning_rate": 9.451500000000001e-05,
      "loss": 0.4299,
      "step": 40862
    },
    {
      "epoch": 2.2882181655280545,
      "grad_norm": 1.2730110883712769,
      "learning_rate": 9.451473684210527e-05,
      "loss": 0.3841,
      "step": 40863
    },
    {
      "epoch": 2.2882741628401835,
      "grad_norm": 1.1795185804367065,
      "learning_rate": 9.451447368421053e-05,
      "loss": 0.4143,
      "step": 40864
    },
    {
      "epoch": 2.2883301601523125,
      "grad_norm": 1.2776638269424438,
      "learning_rate": 9.451421052631579e-05,
      "loss": 0.3611,
      "step": 40865
    },
    {
      "epoch": 2.2883861574644415,
      "grad_norm": 1.0309574604034424,
      "learning_rate": 9.451394736842105e-05,
      "loss": 0.3548,
      "step": 40866
    },
    {
      "epoch": 2.2884421547765705,
      "grad_norm": 1.2780596017837524,
      "learning_rate": 9.451368421052632e-05,
      "loss": 0.4396,
      "step": 40867
    },
    {
      "epoch": 2.2884981520886996,
      "grad_norm": 1.0982019901275635,
      "learning_rate": 9.451342105263158e-05,
      "loss": 0.4266,
      "step": 40868
    },
    {
      "epoch": 2.2885541494008286,
      "grad_norm": 1.0824683904647827,
      "learning_rate": 9.451315789473684e-05,
      "loss": 0.3387,
      "step": 40869
    },
    {
      "epoch": 2.2886101467129576,
      "grad_norm": 1.2712594270706177,
      "learning_rate": 9.45128947368421e-05,
      "loss": 0.362,
      "step": 40870
    },
    {
      "epoch": 2.2886661440250866,
      "grad_norm": 1.3995704650878906,
      "learning_rate": 9.451263157894738e-05,
      "loss": 0.4478,
      "step": 40871
    },
    {
      "epoch": 2.2887221413372156,
      "grad_norm": 1.255401849746704,
      "learning_rate": 9.451236842105264e-05,
      "loss": 0.491,
      "step": 40872
    },
    {
      "epoch": 2.2887781386493447,
      "grad_norm": 1.3592734336853027,
      "learning_rate": 9.45121052631579e-05,
      "loss": 0.4026,
      "step": 40873
    },
    {
      "epoch": 2.2888341359614737,
      "grad_norm": 1.1194995641708374,
      "learning_rate": 9.451184210526316e-05,
      "loss": 0.4288,
      "step": 40874
    },
    {
      "epoch": 2.2888901332736027,
      "grad_norm": 1.3842111825942993,
      "learning_rate": 9.451157894736843e-05,
      "loss": 0.4076,
      "step": 40875
    },
    {
      "epoch": 2.2889461305857317,
      "grad_norm": 1.0915096998214722,
      "learning_rate": 9.451131578947369e-05,
      "loss": 0.4239,
      "step": 40876
    },
    {
      "epoch": 2.2890021278978607,
      "grad_norm": 1.2498295307159424,
      "learning_rate": 9.451105263157895e-05,
      "loss": 0.6339,
      "step": 40877
    },
    {
      "epoch": 2.2890581252099897,
      "grad_norm": 1.2229845523834229,
      "learning_rate": 9.451078947368421e-05,
      "loss": 0.4647,
      "step": 40878
    },
    {
      "epoch": 2.2891141225221188,
      "grad_norm": 1.2996368408203125,
      "learning_rate": 9.451052631578948e-05,
      "loss": 0.3963,
      "step": 40879
    },
    {
      "epoch": 2.289170119834248,
      "grad_norm": 1.2182092666625977,
      "learning_rate": 9.451026315789474e-05,
      "loss": 0.3182,
      "step": 40880
    },
    {
      "epoch": 2.289226117146377,
      "grad_norm": 1.193029522895813,
      "learning_rate": 9.451000000000002e-05,
      "loss": 0.3691,
      "step": 40881
    },
    {
      "epoch": 2.289282114458506,
      "grad_norm": 1.087503433227539,
      "learning_rate": 9.450973684210526e-05,
      "loss": 0.3632,
      "step": 40882
    },
    {
      "epoch": 2.289338111770635,
      "grad_norm": 1.507468819618225,
      "learning_rate": 9.450947368421052e-05,
      "loss": 0.4712,
      "step": 40883
    },
    {
      "epoch": 2.289394109082764,
      "grad_norm": 1.2011854648590088,
      "learning_rate": 9.45092105263158e-05,
      "loss": 0.4968,
      "step": 40884
    },
    {
      "epoch": 2.289450106394893,
      "grad_norm": 1.2221673727035522,
      "learning_rate": 9.450894736842105e-05,
      "loss": 0.4666,
      "step": 40885
    },
    {
      "epoch": 2.289506103707022,
      "grad_norm": 1.7016428709030151,
      "learning_rate": 9.450868421052633e-05,
      "loss": 0.4986,
      "step": 40886
    },
    {
      "epoch": 2.289562101019151,
      "grad_norm": 1.347862958908081,
      "learning_rate": 9.450842105263157e-05,
      "loss": 0.4446,
      "step": 40887
    },
    {
      "epoch": 2.28961809833128,
      "grad_norm": 1.1245880126953125,
      "learning_rate": 9.450815789473685e-05,
      "loss": 0.3261,
      "step": 40888
    },
    {
      "epoch": 2.289674095643409,
      "grad_norm": 1.355403184890747,
      "learning_rate": 9.45078947368421e-05,
      "loss": 0.5013,
      "step": 40889
    },
    {
      "epoch": 2.289730092955538,
      "grad_norm": 1.2152338027954102,
      "learning_rate": 9.450763157894738e-05,
      "loss": 0.4256,
      "step": 40890
    },
    {
      "epoch": 2.289786090267667,
      "grad_norm": 1.1112534999847412,
      "learning_rate": 9.450736842105264e-05,
      "loss": 0.3098,
      "step": 40891
    },
    {
      "epoch": 2.289842087579796,
      "grad_norm": 1.1031806468963623,
      "learning_rate": 9.45071052631579e-05,
      "loss": 0.4814,
      "step": 40892
    },
    {
      "epoch": 2.289898084891925,
      "grad_norm": 1.1614916324615479,
      "learning_rate": 9.450684210526316e-05,
      "loss": 0.3306,
      "step": 40893
    },
    {
      "epoch": 2.289954082204054,
      "grad_norm": 1.2880679368972778,
      "learning_rate": 9.450657894736843e-05,
      "loss": 0.4621,
      "step": 40894
    },
    {
      "epoch": 2.290010079516183,
      "grad_norm": 1.1163889169692993,
      "learning_rate": 9.450631578947369e-05,
      "loss": 0.2877,
      "step": 40895
    },
    {
      "epoch": 2.290066076828312,
      "grad_norm": 2.1183884143829346,
      "learning_rate": 9.450605263157895e-05,
      "loss": 0.3459,
      "step": 40896
    },
    {
      "epoch": 2.290122074140441,
      "grad_norm": 1.1605061292648315,
      "learning_rate": 9.450578947368421e-05,
      "loss": 0.5265,
      "step": 40897
    },
    {
      "epoch": 2.29017807145257,
      "grad_norm": 1.2426642179489136,
      "learning_rate": 9.450552631578948e-05,
      "loss": 0.4247,
      "step": 40898
    },
    {
      "epoch": 2.290234068764699,
      "grad_norm": 1.0339198112487793,
      "learning_rate": 9.450526315789474e-05,
      "loss": 0.3855,
      "step": 40899
    },
    {
      "epoch": 2.290290066076828,
      "grad_norm": 1.3488032817840576,
      "learning_rate": 9.4505e-05,
      "loss": 0.3739,
      "step": 40900
    },
    {
      "epoch": 2.290346063388957,
      "grad_norm": 1.2192391157150269,
      "learning_rate": 9.450473684210526e-05,
      "loss": 0.3077,
      "step": 40901
    },
    {
      "epoch": 2.290402060701086,
      "grad_norm": 1.304444670677185,
      "learning_rate": 9.450447368421052e-05,
      "loss": 0.632,
      "step": 40902
    },
    {
      "epoch": 2.2904580580132152,
      "grad_norm": 1.239427089691162,
      "learning_rate": 9.45042105263158e-05,
      "loss": 0.4729,
      "step": 40903
    },
    {
      "epoch": 2.2905140553253442,
      "grad_norm": 1.1766020059585571,
      "learning_rate": 9.450394736842106e-05,
      "loss": 0.3786,
      "step": 40904
    },
    {
      "epoch": 2.2905700526374733,
      "grad_norm": 1.263381004333496,
      "learning_rate": 9.450368421052632e-05,
      "loss": 0.4117,
      "step": 40905
    },
    {
      "epoch": 2.2906260499496023,
      "grad_norm": 1.1620476245880127,
      "learning_rate": 9.450342105263158e-05,
      "loss": 0.3772,
      "step": 40906
    },
    {
      "epoch": 2.2906820472617313,
      "grad_norm": 1.5683238506317139,
      "learning_rate": 9.450315789473685e-05,
      "loss": 0.449,
      "step": 40907
    },
    {
      "epoch": 2.2907380445738603,
      "grad_norm": 1.2682077884674072,
      "learning_rate": 9.450289473684211e-05,
      "loss": 0.3034,
      "step": 40908
    },
    {
      "epoch": 2.2907940418859893,
      "grad_norm": 1.0831515789031982,
      "learning_rate": 9.450263157894738e-05,
      "loss": 0.3772,
      "step": 40909
    },
    {
      "epoch": 2.2908500391981184,
      "grad_norm": 1.1901028156280518,
      "learning_rate": 9.450236842105263e-05,
      "loss": 0.467,
      "step": 40910
    },
    {
      "epoch": 2.2909060365102474,
      "grad_norm": 1.1299101114273071,
      "learning_rate": 9.45021052631579e-05,
      "loss": 0.3466,
      "step": 40911
    },
    {
      "epoch": 2.2909620338223764,
      "grad_norm": 1.2004923820495605,
      "learning_rate": 9.450184210526316e-05,
      "loss": 0.4476,
      "step": 40912
    },
    {
      "epoch": 2.2910180311345054,
      "grad_norm": 1.2486859560012817,
      "learning_rate": 9.450157894736843e-05,
      "loss": 0.4379,
      "step": 40913
    },
    {
      "epoch": 2.2910740284466344,
      "grad_norm": 1.0966185331344604,
      "learning_rate": 9.45013157894737e-05,
      "loss": 0.3944,
      "step": 40914
    },
    {
      "epoch": 2.2911300257587635,
      "grad_norm": 1.1059114933013916,
      "learning_rate": 9.450105263157895e-05,
      "loss": 0.2733,
      "step": 40915
    },
    {
      "epoch": 2.2911860230708925,
      "grad_norm": 1.1275370121002197,
      "learning_rate": 9.450078947368421e-05,
      "loss": 0.3725,
      "step": 40916
    },
    {
      "epoch": 2.2912420203830215,
      "grad_norm": 1.7922306060791016,
      "learning_rate": 9.450052631578947e-05,
      "loss": 0.3991,
      "step": 40917
    },
    {
      "epoch": 2.2912980176951505,
      "grad_norm": 1.0369951725006104,
      "learning_rate": 9.450026315789475e-05,
      "loss": 0.3563,
      "step": 40918
    },
    {
      "epoch": 2.2913540150072795,
      "grad_norm": 1.0638270378112793,
      "learning_rate": 9.449999999999999e-05,
      "loss": 0.3632,
      "step": 40919
    },
    {
      "epoch": 2.2914100123194086,
      "grad_norm": 1.3651202917099,
      "learning_rate": 9.449973684210527e-05,
      "loss": 0.4033,
      "step": 40920
    },
    {
      "epoch": 2.2914660096315376,
      "grad_norm": 1.327662467956543,
      "learning_rate": 9.449947368421053e-05,
      "loss": 0.4075,
      "step": 40921
    },
    {
      "epoch": 2.2915220069436666,
      "grad_norm": 1.2228999137878418,
      "learning_rate": 9.44992105263158e-05,
      "loss": 0.4145,
      "step": 40922
    },
    {
      "epoch": 2.2915780042557956,
      "grad_norm": 1.3536463975906372,
      "learning_rate": 9.449894736842106e-05,
      "loss": 0.3919,
      "step": 40923
    },
    {
      "epoch": 2.2916340015679246,
      "grad_norm": 1.0907925367355347,
      "learning_rate": 9.449868421052632e-05,
      "loss": 0.3847,
      "step": 40924
    },
    {
      "epoch": 2.2916899988800536,
      "grad_norm": 1.1760512590408325,
      "learning_rate": 9.449842105263158e-05,
      "loss": 0.425,
      "step": 40925
    },
    {
      "epoch": 2.2917459961921827,
      "grad_norm": 1.0818884372711182,
      "learning_rate": 9.449815789473685e-05,
      "loss": 0.369,
      "step": 40926
    },
    {
      "epoch": 2.2918019935043117,
      "grad_norm": 1.1519336700439453,
      "learning_rate": 9.449789473684211e-05,
      "loss": 0.3967,
      "step": 40927
    },
    {
      "epoch": 2.2918579908164407,
      "grad_norm": 1.2027822732925415,
      "learning_rate": 9.449763157894737e-05,
      "loss": 0.407,
      "step": 40928
    },
    {
      "epoch": 2.2919139881285697,
      "grad_norm": 1.3721596002578735,
      "learning_rate": 9.449736842105263e-05,
      "loss": 0.3189,
      "step": 40929
    },
    {
      "epoch": 2.2919699854406987,
      "grad_norm": 1.0069239139556885,
      "learning_rate": 9.44971052631579e-05,
      "loss": 0.448,
      "step": 40930
    },
    {
      "epoch": 2.2920259827528278,
      "grad_norm": 1.4787769317626953,
      "learning_rate": 9.449684210526316e-05,
      "loss": 0.4343,
      "step": 40931
    },
    {
      "epoch": 2.292081980064957,
      "grad_norm": 1.2266465425491333,
      "learning_rate": 9.449657894736842e-05,
      "loss": 0.3329,
      "step": 40932
    },
    {
      "epoch": 2.292137977377086,
      "grad_norm": 1.3142961263656616,
      "learning_rate": 9.449631578947368e-05,
      "loss": 0.4719,
      "step": 40933
    },
    {
      "epoch": 2.292193974689215,
      "grad_norm": 1.2013123035430908,
      "learning_rate": 9.449605263157896e-05,
      "loss": 0.3198,
      "step": 40934
    },
    {
      "epoch": 2.292249972001344,
      "grad_norm": 1.2666776180267334,
      "learning_rate": 9.449578947368422e-05,
      "loss": 0.5273,
      "step": 40935
    },
    {
      "epoch": 2.292305969313473,
      "grad_norm": 1.22313392162323,
      "learning_rate": 9.449552631578948e-05,
      "loss": 0.3404,
      "step": 40936
    },
    {
      "epoch": 2.292361966625602,
      "grad_norm": 1.190106749534607,
      "learning_rate": 9.449526315789474e-05,
      "loss": 0.3736,
      "step": 40937
    },
    {
      "epoch": 2.292417963937731,
      "grad_norm": 1.1512961387634277,
      "learning_rate": 9.4495e-05,
      "loss": 0.4433,
      "step": 40938
    },
    {
      "epoch": 2.29247396124986,
      "grad_norm": 1.0428340435028076,
      "learning_rate": 9.449473684210527e-05,
      "loss": 0.3289,
      "step": 40939
    },
    {
      "epoch": 2.292529958561989,
      "grad_norm": 1.0771962404251099,
      "learning_rate": 9.449447368421053e-05,
      "loss": 0.3793,
      "step": 40940
    },
    {
      "epoch": 2.292585955874118,
      "grad_norm": 1.0007634162902832,
      "learning_rate": 9.44942105263158e-05,
      "loss": 0.3915,
      "step": 40941
    },
    {
      "epoch": 2.292641953186247,
      "grad_norm": 1.365447759628296,
      "learning_rate": 9.449394736842105e-05,
      "loss": 0.3978,
      "step": 40942
    },
    {
      "epoch": 2.292697950498376,
      "grad_norm": 1.527063250541687,
      "learning_rate": 9.449368421052632e-05,
      "loss": 0.4195,
      "step": 40943
    },
    {
      "epoch": 2.292753947810505,
      "grad_norm": 1.4016728401184082,
      "learning_rate": 9.449342105263158e-05,
      "loss": 0.4597,
      "step": 40944
    },
    {
      "epoch": 2.292809945122634,
      "grad_norm": 1.1718723773956299,
      "learning_rate": 9.449315789473685e-05,
      "loss": 0.4199,
      "step": 40945
    },
    {
      "epoch": 2.292865942434763,
      "grad_norm": 1.3043171167373657,
      "learning_rate": 9.449289473684211e-05,
      "loss": 0.3205,
      "step": 40946
    },
    {
      "epoch": 2.292921939746892,
      "grad_norm": 1.2200672626495361,
      "learning_rate": 9.449263157894737e-05,
      "loss": 0.3675,
      "step": 40947
    },
    {
      "epoch": 2.292977937059021,
      "grad_norm": 1.3172346353530884,
      "learning_rate": 9.449236842105263e-05,
      "loss": 0.3676,
      "step": 40948
    },
    {
      "epoch": 2.29303393437115,
      "grad_norm": 1.0622167587280273,
      "learning_rate": 9.449210526315791e-05,
      "loss": 0.3939,
      "step": 40949
    },
    {
      "epoch": 2.293089931683279,
      "grad_norm": 1.3013421297073364,
      "learning_rate": 9.449184210526317e-05,
      "loss": 0.2635,
      "step": 40950
    },
    {
      "epoch": 2.293145928995408,
      "grad_norm": 1.1558432579040527,
      "learning_rate": 9.449157894736843e-05,
      "loss": 0.4306,
      "step": 40951
    },
    {
      "epoch": 2.293201926307537,
      "grad_norm": 1.2057414054870605,
      "learning_rate": 9.449131578947369e-05,
      "loss": 0.3858,
      "step": 40952
    },
    {
      "epoch": 2.293257923619666,
      "grad_norm": 1.0726661682128906,
      "learning_rate": 9.449105263157895e-05,
      "loss": 0.3789,
      "step": 40953
    },
    {
      "epoch": 2.293313920931795,
      "grad_norm": 1.287476658821106,
      "learning_rate": 9.449078947368422e-05,
      "loss": 0.4155,
      "step": 40954
    },
    {
      "epoch": 2.293369918243924,
      "grad_norm": 1.2825465202331543,
      "learning_rate": 9.449052631578948e-05,
      "loss": 0.4285,
      "step": 40955
    },
    {
      "epoch": 2.2934259155560532,
      "grad_norm": 1.1929147243499756,
      "learning_rate": 9.449026315789474e-05,
      "loss": 0.5072,
      "step": 40956
    },
    {
      "epoch": 2.2934819128681823,
      "grad_norm": 1.1540424823760986,
      "learning_rate": 9.449e-05,
      "loss": 0.3393,
      "step": 40957
    },
    {
      "epoch": 2.2935379101803113,
      "grad_norm": 1.2304447889328003,
      "learning_rate": 9.448973684210527e-05,
      "loss": 0.355,
      "step": 40958
    },
    {
      "epoch": 2.2935939074924403,
      "grad_norm": 1.1898878812789917,
      "learning_rate": 9.448947368421053e-05,
      "loss": 0.3964,
      "step": 40959
    },
    {
      "epoch": 2.2936499048045693,
      "grad_norm": 1.2480679750442505,
      "learning_rate": 9.448921052631579e-05,
      "loss": 0.3654,
      "step": 40960
    },
    {
      "epoch": 2.2937059021166983,
      "grad_norm": 1.1460193395614624,
      "learning_rate": 9.448894736842105e-05,
      "loss": 0.4169,
      "step": 40961
    },
    {
      "epoch": 2.2937618994288274,
      "grad_norm": 1.3420658111572266,
      "learning_rate": 9.448868421052632e-05,
      "loss": 0.4806,
      "step": 40962
    },
    {
      "epoch": 2.2938178967409564,
      "grad_norm": 1.2498208284378052,
      "learning_rate": 9.448842105263158e-05,
      "loss": 0.3124,
      "step": 40963
    },
    {
      "epoch": 2.2938738940530854,
      "grad_norm": 1.3126887083053589,
      "learning_rate": 9.448815789473686e-05,
      "loss": 0.3718,
      "step": 40964
    },
    {
      "epoch": 2.2939298913652144,
      "grad_norm": 0.9935676455497742,
      "learning_rate": 9.44878947368421e-05,
      "loss": 0.3567,
      "step": 40965
    },
    {
      "epoch": 2.2939858886773434,
      "grad_norm": 1.150750756263733,
      "learning_rate": 9.448763157894738e-05,
      "loss": 0.5106,
      "step": 40966
    },
    {
      "epoch": 2.2940418859894725,
      "grad_norm": 1.1576104164123535,
      "learning_rate": 9.448736842105264e-05,
      "loss": 0.4623,
      "step": 40967
    },
    {
      "epoch": 2.2940978833016015,
      "grad_norm": 1.4180045127868652,
      "learning_rate": 9.448710526315791e-05,
      "loss": 0.4302,
      "step": 40968
    },
    {
      "epoch": 2.2941538806137305,
      "grad_norm": 1.1114437580108643,
      "learning_rate": 9.448684210526317e-05,
      "loss": 0.4735,
      "step": 40969
    },
    {
      "epoch": 2.2942098779258595,
      "grad_norm": 1.166338324546814,
      "learning_rate": 9.448657894736842e-05,
      "loss": 0.3872,
      "step": 40970
    },
    {
      "epoch": 2.2942658752379885,
      "grad_norm": 1.473102331161499,
      "learning_rate": 9.448631578947369e-05,
      "loss": 0.4464,
      "step": 40971
    },
    {
      "epoch": 2.2943218725501175,
      "grad_norm": 1.3719292879104614,
      "learning_rate": 9.448605263157895e-05,
      "loss": 0.3943,
      "step": 40972
    },
    {
      "epoch": 2.2943778698622466,
      "grad_norm": 1.0014541149139404,
      "learning_rate": 9.448578947368422e-05,
      "loss": 0.344,
      "step": 40973
    },
    {
      "epoch": 2.2944338671743756,
      "grad_norm": 1.3095749616622925,
      "learning_rate": 9.448552631578947e-05,
      "loss": 0.4144,
      "step": 40974
    },
    {
      "epoch": 2.2944898644865046,
      "grad_norm": 1.5319305658340454,
      "learning_rate": 9.448526315789474e-05,
      "loss": 0.4001,
      "step": 40975
    },
    {
      "epoch": 2.2945458617986336,
      "grad_norm": 1.2399495840072632,
      "learning_rate": 9.4485e-05,
      "loss": 0.38,
      "step": 40976
    },
    {
      "epoch": 2.2946018591107626,
      "grad_norm": 0.9646350741386414,
      "learning_rate": 9.448473684210527e-05,
      "loss": 0.3228,
      "step": 40977
    },
    {
      "epoch": 2.2946578564228917,
      "grad_norm": 1.1930832862854004,
      "learning_rate": 9.448447368421053e-05,
      "loss": 0.5378,
      "step": 40978
    },
    {
      "epoch": 2.2947138537350207,
      "grad_norm": 3.9676475524902344,
      "learning_rate": 9.44842105263158e-05,
      "loss": 0.4193,
      "step": 40979
    },
    {
      "epoch": 2.2947698510471497,
      "grad_norm": 1.2792505025863647,
      "learning_rate": 9.448394736842105e-05,
      "loss": 0.4197,
      "step": 40980
    },
    {
      "epoch": 2.2948258483592787,
      "grad_norm": 1.1944348812103271,
      "learning_rate": 9.448368421052633e-05,
      "loss": 0.5138,
      "step": 40981
    },
    {
      "epoch": 2.2948818456714077,
      "grad_norm": 1.1937428712844849,
      "learning_rate": 9.448342105263159e-05,
      "loss": 0.4034,
      "step": 40982
    },
    {
      "epoch": 2.2949378429835368,
      "grad_norm": 1.27170729637146,
      "learning_rate": 9.448315789473685e-05,
      "loss": 0.437,
      "step": 40983
    },
    {
      "epoch": 2.2949938402956658,
      "grad_norm": 2.1695430278778076,
      "learning_rate": 9.44828947368421e-05,
      "loss": 0.4922,
      "step": 40984
    },
    {
      "epoch": 2.295049837607795,
      "grad_norm": 1.2298381328582764,
      "learning_rate": 9.448263157894738e-05,
      "loss": 0.4957,
      "step": 40985
    },
    {
      "epoch": 2.295105834919924,
      "grad_norm": 1.2365107536315918,
      "learning_rate": 9.448236842105264e-05,
      "loss": 0.4557,
      "step": 40986
    },
    {
      "epoch": 2.295161832232053,
      "grad_norm": 1.1257683038711548,
      "learning_rate": 9.44821052631579e-05,
      "loss": 0.4618,
      "step": 40987
    },
    {
      "epoch": 2.295217829544182,
      "grad_norm": 1.2703577280044556,
      "learning_rate": 9.448184210526316e-05,
      "loss": 0.368,
      "step": 40988
    },
    {
      "epoch": 2.295273826856311,
      "grad_norm": 1.1775504350662231,
      "learning_rate": 9.448157894736842e-05,
      "loss": 0.4538,
      "step": 40989
    },
    {
      "epoch": 2.29532982416844,
      "grad_norm": 1.211552381515503,
      "learning_rate": 9.448131578947369e-05,
      "loss": 0.442,
      "step": 40990
    },
    {
      "epoch": 2.295385821480569,
      "grad_norm": 0.9949976205825806,
      "learning_rate": 9.448105263157895e-05,
      "loss": 0.2922,
      "step": 40991
    },
    {
      "epoch": 2.295441818792698,
      "grad_norm": 1.3057242631912231,
      "learning_rate": 9.448078947368421e-05,
      "loss": 0.4271,
      "step": 40992
    },
    {
      "epoch": 2.295497816104827,
      "grad_norm": 1.3530786037445068,
      "learning_rate": 9.448052631578947e-05,
      "loss": 0.3554,
      "step": 40993
    },
    {
      "epoch": 2.295553813416956,
      "grad_norm": 1.249964714050293,
      "learning_rate": 9.448026315789474e-05,
      "loss": 0.4115,
      "step": 40994
    },
    {
      "epoch": 2.295609810729085,
      "grad_norm": 1.3308093547821045,
      "learning_rate": 9.448e-05,
      "loss": 0.3706,
      "step": 40995
    },
    {
      "epoch": 2.295665808041214,
      "grad_norm": 1.0598113536834717,
      "learning_rate": 9.447973684210528e-05,
      "loss": 0.3008,
      "step": 40996
    },
    {
      "epoch": 2.295721805353343,
      "grad_norm": 1.2341864109039307,
      "learning_rate": 9.447947368421052e-05,
      "loss": 0.4196,
      "step": 40997
    },
    {
      "epoch": 2.295777802665472,
      "grad_norm": 1.2854304313659668,
      "learning_rate": 9.44792105263158e-05,
      "loss": 0.4727,
      "step": 40998
    },
    {
      "epoch": 2.295833799977601,
      "grad_norm": 1.2498843669891357,
      "learning_rate": 9.447894736842106e-05,
      "loss": 0.4855,
      "step": 40999
    },
    {
      "epoch": 2.29588979728973,
      "grad_norm": 0.9384483098983765,
      "learning_rate": 9.447868421052633e-05,
      "loss": 0.3439,
      "step": 41000
    },
    {
      "epoch": 2.295945794601859,
      "grad_norm": 1.5126700401306152,
      "learning_rate": 9.447842105263159e-05,
      "loss": 0.4676,
      "step": 41001
    },
    {
      "epoch": 2.296001791913988,
      "grad_norm": 1.1356618404388428,
      "learning_rate": 9.447815789473685e-05,
      "loss": 0.3712,
      "step": 41002
    },
    {
      "epoch": 2.296057789226117,
      "grad_norm": 1.1557092666625977,
      "learning_rate": 9.447789473684211e-05,
      "loss": 0.3532,
      "step": 41003
    },
    {
      "epoch": 2.296113786538246,
      "grad_norm": 1.2167025804519653,
      "learning_rate": 9.447763157894737e-05,
      "loss": 0.3081,
      "step": 41004
    },
    {
      "epoch": 2.296169783850375,
      "grad_norm": 1.0456291437149048,
      "learning_rate": 9.447736842105264e-05,
      "loss": 0.3286,
      "step": 41005
    },
    {
      "epoch": 2.296225781162504,
      "grad_norm": 1.3038848638534546,
      "learning_rate": 9.44771052631579e-05,
      "loss": 0.3303,
      "step": 41006
    },
    {
      "epoch": 2.296281778474633,
      "grad_norm": 1.148787260055542,
      "learning_rate": 9.447684210526316e-05,
      "loss": 0.3593,
      "step": 41007
    },
    {
      "epoch": 2.2963377757867622,
      "grad_norm": 1.431427001953125,
      "learning_rate": 9.447657894736842e-05,
      "loss": 0.3813,
      "step": 41008
    },
    {
      "epoch": 2.2963937730988913,
      "grad_norm": 1.318831443786621,
      "learning_rate": 9.44763157894737e-05,
      "loss": 0.6031,
      "step": 41009
    },
    {
      "epoch": 2.2964497704110203,
      "grad_norm": 1.2120219469070435,
      "learning_rate": 9.447605263157895e-05,
      "loss": 0.3939,
      "step": 41010
    },
    {
      "epoch": 2.2965057677231493,
      "grad_norm": 1.4941589832305908,
      "learning_rate": 9.447578947368421e-05,
      "loss": 0.3676,
      "step": 41011
    },
    {
      "epoch": 2.2965617650352783,
      "grad_norm": 1.1603403091430664,
      "learning_rate": 9.447552631578947e-05,
      "loss": 0.4713,
      "step": 41012
    },
    {
      "epoch": 2.2966177623474073,
      "grad_norm": 1.2495685815811157,
      "learning_rate": 9.447526315789475e-05,
      "loss": 0.3814,
      "step": 41013
    },
    {
      "epoch": 2.2966737596595364,
      "grad_norm": 1.3770884275436401,
      "learning_rate": 9.4475e-05,
      "loss": 0.4228,
      "step": 41014
    },
    {
      "epoch": 2.2967297569716654,
      "grad_norm": 1.1375032663345337,
      "learning_rate": 9.447473684210527e-05,
      "loss": 0.4001,
      "step": 41015
    },
    {
      "epoch": 2.2967857542837944,
      "grad_norm": 1.1974637508392334,
      "learning_rate": 9.447447368421053e-05,
      "loss": 0.4937,
      "step": 41016
    },
    {
      "epoch": 2.2968417515959234,
      "grad_norm": 1.4289354085922241,
      "learning_rate": 9.44742105263158e-05,
      "loss": 0.3866,
      "step": 41017
    },
    {
      "epoch": 2.2968977489080524,
      "grad_norm": 1.3628568649291992,
      "learning_rate": 9.447394736842106e-05,
      "loss": 0.4045,
      "step": 41018
    },
    {
      "epoch": 2.2969537462201814,
      "grad_norm": 1.4367363452911377,
      "learning_rate": 9.447368421052633e-05,
      "loss": 0.6259,
      "step": 41019
    },
    {
      "epoch": 2.2970097435323105,
      "grad_norm": 1.2419058084487915,
      "learning_rate": 9.447342105263158e-05,
      "loss": 0.4216,
      "step": 41020
    },
    {
      "epoch": 2.2970657408444395,
      "grad_norm": 1.2421067953109741,
      "learning_rate": 9.447315789473684e-05,
      "loss": 0.4085,
      "step": 41021
    },
    {
      "epoch": 2.2971217381565685,
      "grad_norm": 1.1407455205917358,
      "learning_rate": 9.447289473684211e-05,
      "loss": 0.4231,
      "step": 41022
    },
    {
      "epoch": 2.2971777354686975,
      "grad_norm": 1.1749318838119507,
      "learning_rate": 9.447263157894737e-05,
      "loss": 0.3661,
      "step": 41023
    },
    {
      "epoch": 2.2972337327808265,
      "grad_norm": 1.1493127346038818,
      "learning_rate": 9.447236842105264e-05,
      "loss": 0.3681,
      "step": 41024
    },
    {
      "epoch": 2.2972897300929556,
      "grad_norm": 1.308562994003296,
      "learning_rate": 9.447210526315789e-05,
      "loss": 0.4233,
      "step": 41025
    },
    {
      "epoch": 2.2973457274050846,
      "grad_norm": 1.1703099012374878,
      "learning_rate": 9.447184210526316e-05,
      "loss": 0.4349,
      "step": 41026
    },
    {
      "epoch": 2.2974017247172136,
      "grad_norm": 1.1787515878677368,
      "learning_rate": 9.447157894736842e-05,
      "loss": 0.3317,
      "step": 41027
    },
    {
      "epoch": 2.2974577220293426,
      "grad_norm": 1.4087570905685425,
      "learning_rate": 9.44713157894737e-05,
      "loss": 0.456,
      "step": 41028
    },
    {
      "epoch": 2.2975137193414716,
      "grad_norm": 1.1689895391464233,
      "learning_rate": 9.447105263157894e-05,
      "loss": 0.3891,
      "step": 41029
    },
    {
      "epoch": 2.2975697166536007,
      "grad_norm": 1.2607016563415527,
      "learning_rate": 9.447078947368422e-05,
      "loss": 0.4144,
      "step": 41030
    },
    {
      "epoch": 2.2976257139657297,
      "grad_norm": 1.0939470529556274,
      "learning_rate": 9.447052631578948e-05,
      "loss": 0.4209,
      "step": 41031
    },
    {
      "epoch": 2.2976817112778587,
      "grad_norm": 1.0828138589859009,
      "learning_rate": 9.447026315789475e-05,
      "loss": 0.41,
      "step": 41032
    },
    {
      "epoch": 2.2977377085899877,
      "grad_norm": 1.824317216873169,
      "learning_rate": 9.447000000000001e-05,
      "loss": 0.4542,
      "step": 41033
    },
    {
      "epoch": 2.2977937059021167,
      "grad_norm": 1.4709928035736084,
      "learning_rate": 9.446973684210527e-05,
      "loss": 0.4738,
      "step": 41034
    },
    {
      "epoch": 2.2978497032142458,
      "grad_norm": 1.1033259630203247,
      "learning_rate": 9.446947368421053e-05,
      "loss": 0.4526,
      "step": 41035
    },
    {
      "epoch": 2.2979057005263748,
      "grad_norm": 1.1319345235824585,
      "learning_rate": 9.44692105263158e-05,
      "loss": 0.3799,
      "step": 41036
    },
    {
      "epoch": 2.297961697838504,
      "grad_norm": 1.0476441383361816,
      "learning_rate": 9.446894736842106e-05,
      "loss": 0.3174,
      "step": 41037
    },
    {
      "epoch": 2.298017695150633,
      "grad_norm": 1.1341053247451782,
      "learning_rate": 9.446868421052632e-05,
      "loss": 0.4399,
      "step": 41038
    },
    {
      "epoch": 2.298073692462762,
      "grad_norm": 1.166084885597229,
      "learning_rate": 9.446842105263158e-05,
      "loss": 0.3245,
      "step": 41039
    },
    {
      "epoch": 2.298129689774891,
      "grad_norm": 1.4878582954406738,
      "learning_rate": 9.446815789473684e-05,
      "loss": 0.5112,
      "step": 41040
    },
    {
      "epoch": 2.29818568708702,
      "grad_norm": 1.1163021326065063,
      "learning_rate": 9.446789473684211e-05,
      "loss": 0.3523,
      "step": 41041
    },
    {
      "epoch": 2.298241684399149,
      "grad_norm": 1.2826194763183594,
      "learning_rate": 9.446763157894737e-05,
      "loss": 0.4249,
      "step": 41042
    },
    {
      "epoch": 2.298297681711278,
      "grad_norm": 0.9728094339370728,
      "learning_rate": 9.446736842105263e-05,
      "loss": 0.307,
      "step": 41043
    },
    {
      "epoch": 2.298353679023407,
      "grad_norm": 1.2379939556121826,
      "learning_rate": 9.446710526315789e-05,
      "loss": 0.3998,
      "step": 41044
    },
    {
      "epoch": 2.298409676335536,
      "grad_norm": 1.08194100856781,
      "learning_rate": 9.446684210526317e-05,
      "loss": 0.3588,
      "step": 41045
    },
    {
      "epoch": 2.298465673647665,
      "grad_norm": 1.02166748046875,
      "learning_rate": 9.446657894736843e-05,
      "loss": 0.294,
      "step": 41046
    },
    {
      "epoch": 2.298521670959794,
      "grad_norm": 1.039847731590271,
      "learning_rate": 9.446631578947369e-05,
      "loss": 0.3131,
      "step": 41047
    },
    {
      "epoch": 2.298577668271923,
      "grad_norm": 1.292528748512268,
      "learning_rate": 9.446605263157895e-05,
      "loss": 0.4246,
      "step": 41048
    },
    {
      "epoch": 2.298633665584052,
      "grad_norm": 2.0334975719451904,
      "learning_rate": 9.446578947368422e-05,
      "loss": 0.6701,
      "step": 41049
    },
    {
      "epoch": 2.298689662896181,
      "grad_norm": 1.2254564762115479,
      "learning_rate": 9.446552631578948e-05,
      "loss": 0.3212,
      "step": 41050
    },
    {
      "epoch": 2.29874566020831,
      "grad_norm": 1.154109239578247,
      "learning_rate": 9.446526315789475e-05,
      "loss": 0.2934,
      "step": 41051
    },
    {
      "epoch": 2.298801657520439,
      "grad_norm": 1.0780616998672485,
      "learning_rate": 9.4465e-05,
      "loss": 0.4078,
      "step": 41052
    },
    {
      "epoch": 2.298857654832568,
      "grad_norm": 1.3531138896942139,
      "learning_rate": 9.446473684210527e-05,
      "loss": 0.429,
      "step": 41053
    },
    {
      "epoch": 2.298913652144697,
      "grad_norm": 1.2903035879135132,
      "learning_rate": 9.446447368421053e-05,
      "loss": 0.4173,
      "step": 41054
    },
    {
      "epoch": 2.298969649456826,
      "grad_norm": 1.4337186813354492,
      "learning_rate": 9.44642105263158e-05,
      "loss": 0.4609,
      "step": 41055
    },
    {
      "epoch": 2.299025646768955,
      "grad_norm": 1.255073070526123,
      "learning_rate": 9.446394736842106e-05,
      "loss": 0.4492,
      "step": 41056
    },
    {
      "epoch": 2.299081644081084,
      "grad_norm": 1.4382634162902832,
      "learning_rate": 9.446368421052631e-05,
      "loss": 0.421,
      "step": 41057
    },
    {
      "epoch": 2.299137641393213,
      "grad_norm": 1.0598331689834595,
      "learning_rate": 9.446342105263158e-05,
      "loss": 0.4544,
      "step": 41058
    },
    {
      "epoch": 2.299193638705342,
      "grad_norm": 1.15713632106781,
      "learning_rate": 9.446315789473684e-05,
      "loss": 0.3135,
      "step": 41059
    },
    {
      "epoch": 2.2992496360174712,
      "grad_norm": 1.1746466159820557,
      "learning_rate": 9.446289473684212e-05,
      "loss": 0.4169,
      "step": 41060
    },
    {
      "epoch": 2.2993056333296003,
      "grad_norm": 1.4122806787490845,
      "learning_rate": 9.446263157894738e-05,
      "loss": 0.3712,
      "step": 41061
    },
    {
      "epoch": 2.2993616306417293,
      "grad_norm": 1.0100699663162231,
      "learning_rate": 9.446236842105264e-05,
      "loss": 0.4306,
      "step": 41062
    },
    {
      "epoch": 2.2994176279538583,
      "grad_norm": 1.4227796792984009,
      "learning_rate": 9.44621052631579e-05,
      "loss": 0.4145,
      "step": 41063
    },
    {
      "epoch": 2.2994736252659873,
      "grad_norm": 1.1188349723815918,
      "learning_rate": 9.446184210526317e-05,
      "loss": 0.5237,
      "step": 41064
    },
    {
      "epoch": 2.2995296225781163,
      "grad_norm": 1.7654399871826172,
      "learning_rate": 9.446157894736843e-05,
      "loss": 0.4551,
      "step": 41065
    },
    {
      "epoch": 2.2995856198902453,
      "grad_norm": 1.2685762643814087,
      "learning_rate": 9.446131578947369e-05,
      "loss": 0.6306,
      "step": 41066
    },
    {
      "epoch": 2.2996416172023744,
      "grad_norm": 1.1727772951126099,
      "learning_rate": 9.446105263157895e-05,
      "loss": 0.3253,
      "step": 41067
    },
    {
      "epoch": 2.2996976145145034,
      "grad_norm": 1.4301499128341675,
      "learning_rate": 9.446078947368422e-05,
      "loss": 0.4329,
      "step": 41068
    },
    {
      "epoch": 2.2997536118266324,
      "grad_norm": 1.3428634405136108,
      "learning_rate": 9.446052631578948e-05,
      "loss": 0.4392,
      "step": 41069
    },
    {
      "epoch": 2.2998096091387614,
      "grad_norm": 1.1013458967208862,
      "learning_rate": 9.446026315789474e-05,
      "loss": 0.3832,
      "step": 41070
    },
    {
      "epoch": 2.2998656064508904,
      "grad_norm": 1.1627110242843628,
      "learning_rate": 9.446e-05,
      "loss": 0.3285,
      "step": 41071
    },
    {
      "epoch": 2.2999216037630195,
      "grad_norm": 1.4366910457611084,
      "learning_rate": 9.445973684210527e-05,
      "loss": 0.4185,
      "step": 41072
    },
    {
      "epoch": 2.2999776010751485,
      "grad_norm": 1.4668974876403809,
      "learning_rate": 9.445947368421053e-05,
      "loss": 0.4888,
      "step": 41073
    },
    {
      "epoch": 2.3000335983872775,
      "grad_norm": 1.4171216487884521,
      "learning_rate": 9.445921052631579e-05,
      "loss": 0.6389,
      "step": 41074
    },
    {
      "epoch": 2.3000895956994065,
      "grad_norm": 1.1692252159118652,
      "learning_rate": 9.445894736842105e-05,
      "loss": 0.4283,
      "step": 41075
    },
    {
      "epoch": 2.3001455930115355,
      "grad_norm": 1.0869266986846924,
      "learning_rate": 9.445868421052631e-05,
      "loss": 0.3723,
      "step": 41076
    },
    {
      "epoch": 2.3002015903236646,
      "grad_norm": 1.2284107208251953,
      "learning_rate": 9.445842105263159e-05,
      "loss": 0.4374,
      "step": 41077
    },
    {
      "epoch": 2.3002575876357936,
      "grad_norm": 1.244626760482788,
      "learning_rate": 9.445815789473685e-05,
      "loss": 0.2871,
      "step": 41078
    },
    {
      "epoch": 2.3003135849479226,
      "grad_norm": 1.3934588432312012,
      "learning_rate": 9.44578947368421e-05,
      "loss": 0.4364,
      "step": 41079
    },
    {
      "epoch": 2.3003695822600516,
      "grad_norm": 1.210655689239502,
      "learning_rate": 9.445763157894736e-05,
      "loss": 0.4166,
      "step": 41080
    },
    {
      "epoch": 2.3004255795721806,
      "grad_norm": 1.0206670761108398,
      "learning_rate": 9.445736842105264e-05,
      "loss": 0.2758,
      "step": 41081
    },
    {
      "epoch": 2.3004815768843097,
      "grad_norm": 1.387534260749817,
      "learning_rate": 9.44571052631579e-05,
      "loss": 0.3962,
      "step": 41082
    },
    {
      "epoch": 2.3005375741964387,
      "grad_norm": 1.2012280225753784,
      "learning_rate": 9.445684210526317e-05,
      "loss": 0.4016,
      "step": 41083
    },
    {
      "epoch": 2.3005935715085677,
      "grad_norm": 1.517020344734192,
      "learning_rate": 9.445657894736842e-05,
      "loss": 0.4156,
      "step": 41084
    },
    {
      "epoch": 2.3006495688206967,
      "grad_norm": 1.1856268644332886,
      "learning_rate": 9.445631578947369e-05,
      "loss": 0.3997,
      "step": 41085
    },
    {
      "epoch": 2.3007055661328257,
      "grad_norm": 1.2983989715576172,
      "learning_rate": 9.445605263157895e-05,
      "loss": 0.4981,
      "step": 41086
    },
    {
      "epoch": 2.3007615634449547,
      "grad_norm": 1.1210360527038574,
      "learning_rate": 9.445578947368422e-05,
      "loss": 0.3545,
      "step": 41087
    },
    {
      "epoch": 2.3008175607570838,
      "grad_norm": 1.2245783805847168,
      "learning_rate": 9.445552631578948e-05,
      "loss": 0.4266,
      "step": 41088
    },
    {
      "epoch": 2.300873558069213,
      "grad_norm": 1.170212745666504,
      "learning_rate": 9.445526315789474e-05,
      "loss": 0.4732,
      "step": 41089
    },
    {
      "epoch": 2.300929555381342,
      "grad_norm": 1.2895238399505615,
      "learning_rate": 9.4455e-05,
      "loss": 0.4146,
      "step": 41090
    },
    {
      "epoch": 2.300985552693471,
      "grad_norm": 1.3115994930267334,
      "learning_rate": 9.445473684210526e-05,
      "loss": 0.3908,
      "step": 41091
    },
    {
      "epoch": 2.3010415500056,
      "grad_norm": 1.357094645500183,
      "learning_rate": 9.445447368421054e-05,
      "loss": 0.4171,
      "step": 41092
    },
    {
      "epoch": 2.301097547317729,
      "grad_norm": 1.2858004570007324,
      "learning_rate": 9.44542105263158e-05,
      "loss": 0.4663,
      "step": 41093
    },
    {
      "epoch": 2.301153544629858,
      "grad_norm": 1.167243242263794,
      "learning_rate": 9.445394736842106e-05,
      "loss": 0.478,
      "step": 41094
    },
    {
      "epoch": 2.301209541941987,
      "grad_norm": 1.3510496616363525,
      "learning_rate": 9.445368421052632e-05,
      "loss": 0.334,
      "step": 41095
    },
    {
      "epoch": 2.301265539254116,
      "grad_norm": 1.2992651462554932,
      "learning_rate": 9.445342105263159e-05,
      "loss": 0.4249,
      "step": 41096
    },
    {
      "epoch": 2.301321536566245,
      "grad_norm": 1.0918200016021729,
      "learning_rate": 9.445315789473685e-05,
      "loss": 0.3791,
      "step": 41097
    },
    {
      "epoch": 2.301377533878374,
      "grad_norm": 1.1572182178497314,
      "learning_rate": 9.445289473684211e-05,
      "loss": 0.4501,
      "step": 41098
    },
    {
      "epoch": 2.301433531190503,
      "grad_norm": 1.1787786483764648,
      "learning_rate": 9.445263157894737e-05,
      "loss": 0.5839,
      "step": 41099
    },
    {
      "epoch": 2.301489528502632,
      "grad_norm": 1.1431785821914673,
      "learning_rate": 9.445236842105264e-05,
      "loss": 0.4091,
      "step": 41100
    },
    {
      "epoch": 2.301545525814761,
      "grad_norm": 0.9629998803138733,
      "learning_rate": 9.44521052631579e-05,
      "loss": 0.261,
      "step": 41101
    },
    {
      "epoch": 2.30160152312689,
      "grad_norm": 1.0944408178329468,
      "learning_rate": 9.445184210526316e-05,
      "loss": 0.4321,
      "step": 41102
    },
    {
      "epoch": 2.301657520439019,
      "grad_norm": 1.130164623260498,
      "learning_rate": 9.445157894736842e-05,
      "loss": 0.4404,
      "step": 41103
    },
    {
      "epoch": 2.301713517751148,
      "grad_norm": 1.2889387607574463,
      "learning_rate": 9.445131578947369e-05,
      "loss": 0.4271,
      "step": 41104
    },
    {
      "epoch": 2.301769515063277,
      "grad_norm": 1.2469089031219482,
      "learning_rate": 9.445105263157895e-05,
      "loss": 0.4932,
      "step": 41105
    },
    {
      "epoch": 2.301825512375406,
      "grad_norm": 1.561651349067688,
      "learning_rate": 9.445078947368423e-05,
      "loss": 0.4857,
      "step": 41106
    },
    {
      "epoch": 2.301881509687535,
      "grad_norm": 1.207578182220459,
      "learning_rate": 9.445052631578947e-05,
      "loss": 0.3862,
      "step": 41107
    },
    {
      "epoch": 2.301937506999664,
      "grad_norm": 1.0754132270812988,
      "learning_rate": 9.445026315789473e-05,
      "loss": 0.4123,
      "step": 41108
    },
    {
      "epoch": 2.301993504311793,
      "grad_norm": 1.0970557928085327,
      "learning_rate": 9.445e-05,
      "loss": 0.3689,
      "step": 41109
    },
    {
      "epoch": 2.302049501623922,
      "grad_norm": 1.1462322473526,
      "learning_rate": 9.444973684210527e-05,
      "loss": 0.389,
      "step": 41110
    },
    {
      "epoch": 2.302105498936051,
      "grad_norm": 1.1647051572799683,
      "learning_rate": 9.444947368421054e-05,
      "loss": 0.4134,
      "step": 41111
    },
    {
      "epoch": 2.3021614962481802,
      "grad_norm": 1.1929714679718018,
      "learning_rate": 9.444921052631578e-05,
      "loss": 0.3728,
      "step": 41112
    },
    {
      "epoch": 2.3022174935603092,
      "grad_norm": 1.1695880889892578,
      "learning_rate": 9.444894736842106e-05,
      "loss": 0.3382,
      "step": 41113
    },
    {
      "epoch": 2.3022734908724383,
      "grad_norm": 1.2483124732971191,
      "learning_rate": 9.444868421052632e-05,
      "loss": 0.3711,
      "step": 41114
    },
    {
      "epoch": 2.3023294881845673,
      "grad_norm": 1.0774250030517578,
      "learning_rate": 9.444842105263159e-05,
      "loss": 0.3927,
      "step": 41115
    },
    {
      "epoch": 2.3023854854966963,
      "grad_norm": 1.1876918077468872,
      "learning_rate": 9.444815789473685e-05,
      "loss": 0.4074,
      "step": 41116
    },
    {
      "epoch": 2.3024414828088253,
      "grad_norm": 1.1963183879852295,
      "learning_rate": 9.444789473684211e-05,
      "loss": 0.2942,
      "step": 41117
    },
    {
      "epoch": 2.3024974801209543,
      "grad_norm": 1.2483224868774414,
      "learning_rate": 9.444763157894737e-05,
      "loss": 0.3444,
      "step": 41118
    },
    {
      "epoch": 2.3025534774330834,
      "grad_norm": 1.2213937044143677,
      "learning_rate": 9.444736842105264e-05,
      "loss": 0.5581,
      "step": 41119
    },
    {
      "epoch": 2.3026094747452124,
      "grad_norm": 1.1406149864196777,
      "learning_rate": 9.44471052631579e-05,
      "loss": 0.4119,
      "step": 41120
    },
    {
      "epoch": 2.3026654720573414,
      "grad_norm": 1.4735124111175537,
      "learning_rate": 9.444684210526316e-05,
      "loss": 0.4359,
      "step": 41121
    },
    {
      "epoch": 2.3027214693694704,
      "grad_norm": 1.2479605674743652,
      "learning_rate": 9.444657894736842e-05,
      "loss": 0.4266,
      "step": 41122
    },
    {
      "epoch": 2.3027774666815994,
      "grad_norm": 1.0820972919464111,
      "learning_rate": 9.44463157894737e-05,
      "loss": 0.3723,
      "step": 41123
    },
    {
      "epoch": 2.3028334639937285,
      "grad_norm": 1.0464650392532349,
      "learning_rate": 9.444605263157896e-05,
      "loss": 0.5619,
      "step": 41124
    },
    {
      "epoch": 2.3028894613058575,
      "grad_norm": 3.076843023300171,
      "learning_rate": 9.444578947368422e-05,
      "loss": 0.5786,
      "step": 41125
    },
    {
      "epoch": 2.3029454586179865,
      "grad_norm": 1.3389883041381836,
      "learning_rate": 9.444552631578947e-05,
      "loss": 0.4176,
      "step": 41126
    },
    {
      "epoch": 2.3030014559301155,
      "grad_norm": 1.0942277908325195,
      "learning_rate": 9.444526315789473e-05,
      "loss": 0.4362,
      "step": 41127
    },
    {
      "epoch": 2.3030574532422445,
      "grad_norm": 1.4207924604415894,
      "learning_rate": 9.444500000000001e-05,
      "loss": 0.5054,
      "step": 41128
    },
    {
      "epoch": 2.3031134505543736,
      "grad_norm": 1.9212104082107544,
      "learning_rate": 9.444473684210527e-05,
      "loss": 0.4975,
      "step": 41129
    },
    {
      "epoch": 2.3031694478665026,
      "grad_norm": 1.1983555555343628,
      "learning_rate": 9.444447368421053e-05,
      "loss": 0.3388,
      "step": 41130
    },
    {
      "epoch": 2.3032254451786316,
      "grad_norm": 1.1969877481460571,
      "learning_rate": 9.444421052631579e-05,
      "loss": 0.4088,
      "step": 41131
    },
    {
      "epoch": 2.3032814424907606,
      "grad_norm": 1.1954209804534912,
      "learning_rate": 9.444394736842106e-05,
      "loss": 0.5604,
      "step": 41132
    },
    {
      "epoch": 2.3033374398028896,
      "grad_norm": 1.2000771760940552,
      "learning_rate": 9.444368421052632e-05,
      "loss": 0.4022,
      "step": 41133
    },
    {
      "epoch": 2.3033934371150186,
      "grad_norm": 1.2491918802261353,
      "learning_rate": 9.444342105263158e-05,
      "loss": 0.3094,
      "step": 41134
    },
    {
      "epoch": 2.3034494344271477,
      "grad_norm": 1.0597612857818604,
      "learning_rate": 9.444315789473684e-05,
      "loss": 0.3546,
      "step": 41135
    },
    {
      "epoch": 2.3035054317392767,
      "grad_norm": 1.6120266914367676,
      "learning_rate": 9.444289473684211e-05,
      "loss": 0.5958,
      "step": 41136
    },
    {
      "epoch": 2.3035614290514057,
      "grad_norm": 1.248927116394043,
      "learning_rate": 9.444263157894737e-05,
      "loss": 0.3942,
      "step": 41137
    },
    {
      "epoch": 2.3036174263635347,
      "grad_norm": 1.0413066148757935,
      "learning_rate": 9.444236842105265e-05,
      "loss": 0.3408,
      "step": 41138
    },
    {
      "epoch": 2.3036734236756637,
      "grad_norm": 1.2440885305404663,
      "learning_rate": 9.444210526315789e-05,
      "loss": 0.4456,
      "step": 41139
    },
    {
      "epoch": 2.3037294209877928,
      "grad_norm": 1.1932263374328613,
      "learning_rate": 9.444184210526317e-05,
      "loss": 0.3805,
      "step": 41140
    },
    {
      "epoch": 2.303785418299922,
      "grad_norm": 1.8755558729171753,
      "learning_rate": 9.444157894736843e-05,
      "loss": 0.657,
      "step": 41141
    },
    {
      "epoch": 2.303841415612051,
      "grad_norm": 1.25771963596344,
      "learning_rate": 9.444131578947368e-05,
      "loss": 0.3562,
      "step": 41142
    },
    {
      "epoch": 2.30389741292418,
      "grad_norm": 1.5962650775909424,
      "learning_rate": 9.444105263157896e-05,
      "loss": 0.4554,
      "step": 41143
    },
    {
      "epoch": 2.303953410236309,
      "grad_norm": 1.016640067100525,
      "learning_rate": 9.44407894736842e-05,
      "loss": 0.2997,
      "step": 41144
    },
    {
      "epoch": 2.304009407548438,
      "grad_norm": 1.0921821594238281,
      "learning_rate": 9.444052631578948e-05,
      "loss": 0.3552,
      "step": 41145
    },
    {
      "epoch": 2.304065404860567,
      "grad_norm": 1.1198456287384033,
      "learning_rate": 9.444026315789474e-05,
      "loss": 0.4067,
      "step": 41146
    },
    {
      "epoch": 2.304121402172696,
      "grad_norm": 1.135983943939209,
      "learning_rate": 9.444000000000001e-05,
      "loss": 0.3517,
      "step": 41147
    },
    {
      "epoch": 2.304177399484825,
      "grad_norm": 1.1520676612854004,
      "learning_rate": 9.443973684210527e-05,
      "loss": 0.3625,
      "step": 41148
    },
    {
      "epoch": 2.304233396796954,
      "grad_norm": 1.096591591835022,
      "learning_rate": 9.443947368421053e-05,
      "loss": 0.3884,
      "step": 41149
    },
    {
      "epoch": 2.304289394109083,
      "grad_norm": 1.1004701852798462,
      "learning_rate": 9.443921052631579e-05,
      "loss": 0.587,
      "step": 41150
    },
    {
      "epoch": 2.304345391421212,
      "grad_norm": 1.0026360750198364,
      "learning_rate": 9.443894736842106e-05,
      "loss": 0.2592,
      "step": 41151
    },
    {
      "epoch": 2.304401388733341,
      "grad_norm": 1.2916796207427979,
      "learning_rate": 9.443868421052632e-05,
      "loss": 0.3808,
      "step": 41152
    },
    {
      "epoch": 2.3044573860454696,
      "grad_norm": 1.241463541984558,
      "learning_rate": 9.443842105263158e-05,
      "loss": 0.4439,
      "step": 41153
    },
    {
      "epoch": 2.304513383357599,
      "grad_norm": 1.0523325204849243,
      "learning_rate": 9.443815789473684e-05,
      "loss": 0.2913,
      "step": 41154
    },
    {
      "epoch": 2.3045693806697276,
      "grad_norm": 1.4757956266403198,
      "learning_rate": 9.443789473684212e-05,
      "loss": 0.602,
      "step": 41155
    },
    {
      "epoch": 2.304625377981857,
      "grad_norm": 1.5350556373596191,
      "learning_rate": 9.443763157894738e-05,
      "loss": 0.5242,
      "step": 41156
    },
    {
      "epoch": 2.3046813752939856,
      "grad_norm": 1.3382859230041504,
      "learning_rate": 9.443736842105263e-05,
      "loss": 0.3939,
      "step": 41157
    },
    {
      "epoch": 2.304737372606115,
      "grad_norm": 1.2251603603363037,
      "learning_rate": 9.44371052631579e-05,
      "loss": 0.3432,
      "step": 41158
    },
    {
      "epoch": 2.3047933699182437,
      "grad_norm": 1.0954320430755615,
      "learning_rate": 9.443684210526315e-05,
      "loss": 0.393,
      "step": 41159
    },
    {
      "epoch": 2.304849367230373,
      "grad_norm": 1.1894170045852661,
      "learning_rate": 9.443657894736843e-05,
      "loss": 0.4276,
      "step": 41160
    },
    {
      "epoch": 2.3049053645425017,
      "grad_norm": 1.2310396432876587,
      "learning_rate": 9.443631578947369e-05,
      "loss": 0.3287,
      "step": 41161
    },
    {
      "epoch": 2.304961361854631,
      "grad_norm": 1.019013524055481,
      "learning_rate": 9.443605263157895e-05,
      "loss": 0.3896,
      "step": 41162
    },
    {
      "epoch": 2.3050173591667598,
      "grad_norm": 1.1283564567565918,
      "learning_rate": 9.443578947368421e-05,
      "loss": 0.3744,
      "step": 41163
    },
    {
      "epoch": 2.3050733564788892,
      "grad_norm": 1.1618505716323853,
      "learning_rate": 9.443552631578948e-05,
      "loss": 0.4102,
      "step": 41164
    },
    {
      "epoch": 2.305129353791018,
      "grad_norm": 1.096095323562622,
      "learning_rate": 9.443526315789474e-05,
      "loss": 0.4126,
      "step": 41165
    },
    {
      "epoch": 2.3051853511031473,
      "grad_norm": 1.2699341773986816,
      "learning_rate": 9.443500000000001e-05,
      "loss": 0.4288,
      "step": 41166
    },
    {
      "epoch": 2.305241348415276,
      "grad_norm": 1.2683535814285278,
      "learning_rate": 9.443473684210526e-05,
      "loss": 0.4654,
      "step": 41167
    },
    {
      "epoch": 2.3052973457274053,
      "grad_norm": 1.0817099809646606,
      "learning_rate": 9.443447368421053e-05,
      "loss": 0.3669,
      "step": 41168
    },
    {
      "epoch": 2.305353343039534,
      "grad_norm": 1.0629780292510986,
      "learning_rate": 9.443421052631579e-05,
      "loss": 0.3186,
      "step": 41169
    },
    {
      "epoch": 2.3054093403516633,
      "grad_norm": 1.2280093431472778,
      "learning_rate": 9.443394736842107e-05,
      "loss": 0.3481,
      "step": 41170
    },
    {
      "epoch": 2.305465337663792,
      "grad_norm": 1.1792889833450317,
      "learning_rate": 9.443368421052633e-05,
      "loss": 0.3871,
      "step": 41171
    },
    {
      "epoch": 2.3055213349759214,
      "grad_norm": 1.1607532501220703,
      "learning_rate": 9.443342105263159e-05,
      "loss": 0.3518,
      "step": 41172
    },
    {
      "epoch": 2.30557733228805,
      "grad_norm": 1.382346510887146,
      "learning_rate": 9.443315789473684e-05,
      "loss": 0.3422,
      "step": 41173
    },
    {
      "epoch": 2.3056333296001794,
      "grad_norm": 2.358943462371826,
      "learning_rate": 9.443289473684212e-05,
      "loss": 0.4702,
      "step": 41174
    },
    {
      "epoch": 2.305689326912308,
      "grad_norm": 1.1040019989013672,
      "learning_rate": 9.443263157894738e-05,
      "loss": 0.396,
      "step": 41175
    },
    {
      "epoch": 2.3057453242244375,
      "grad_norm": 1.385833740234375,
      "learning_rate": 9.443236842105264e-05,
      "loss": 0.4665,
      "step": 41176
    },
    {
      "epoch": 2.305801321536566,
      "grad_norm": 1.2313884496688843,
      "learning_rate": 9.44321052631579e-05,
      "loss": 0.4431,
      "step": 41177
    },
    {
      "epoch": 2.3058573188486955,
      "grad_norm": 1.314171552658081,
      "learning_rate": 9.443184210526316e-05,
      "loss": 0.5751,
      "step": 41178
    },
    {
      "epoch": 2.305913316160824,
      "grad_norm": 1.1755602359771729,
      "learning_rate": 9.443157894736843e-05,
      "loss": 0.3631,
      "step": 41179
    },
    {
      "epoch": 2.3059693134729535,
      "grad_norm": 1.1333853006362915,
      "learning_rate": 9.443131578947369e-05,
      "loss": 0.457,
      "step": 41180
    },
    {
      "epoch": 2.306025310785082,
      "grad_norm": 1.4316025972366333,
      "learning_rate": 9.443105263157895e-05,
      "loss": 0.4215,
      "step": 41181
    },
    {
      "epoch": 2.3060813080972116,
      "grad_norm": 1.150217890739441,
      "learning_rate": 9.443078947368421e-05,
      "loss": 0.33,
      "step": 41182
    },
    {
      "epoch": 2.30613730540934,
      "grad_norm": 1.2990572452545166,
      "learning_rate": 9.443052631578948e-05,
      "loss": 0.4061,
      "step": 41183
    },
    {
      "epoch": 2.3061933027214696,
      "grad_norm": 1.056262731552124,
      "learning_rate": 9.443026315789474e-05,
      "loss": 0.4119,
      "step": 41184
    },
    {
      "epoch": 2.306249300033598,
      "grad_norm": 1.326519250869751,
      "learning_rate": 9.443e-05,
      "loss": 0.4717,
      "step": 41185
    },
    {
      "epoch": 2.3063052973457276,
      "grad_norm": 1.003314733505249,
      "learning_rate": 9.442973684210526e-05,
      "loss": 0.3695,
      "step": 41186
    },
    {
      "epoch": 2.306361294657856,
      "grad_norm": 1.3201234340667725,
      "learning_rate": 9.442947368421054e-05,
      "loss": 0.6548,
      "step": 41187
    },
    {
      "epoch": 2.3064172919699857,
      "grad_norm": 1.1729272603988647,
      "learning_rate": 9.44292105263158e-05,
      "loss": 0.3128,
      "step": 41188
    },
    {
      "epoch": 2.3064732892821143,
      "grad_norm": 1.1012417078018188,
      "learning_rate": 9.442894736842105e-05,
      "loss": 0.2822,
      "step": 41189
    },
    {
      "epoch": 2.3065292865942437,
      "grad_norm": 1.3226896524429321,
      "learning_rate": 9.442868421052631e-05,
      "loss": 0.521,
      "step": 41190
    },
    {
      "epoch": 2.3065852839063723,
      "grad_norm": 1.0767464637756348,
      "learning_rate": 9.442842105263159e-05,
      "loss": 0.3531,
      "step": 41191
    },
    {
      "epoch": 2.3066412812185018,
      "grad_norm": 1.212410807609558,
      "learning_rate": 9.442815789473685e-05,
      "loss": 0.3683,
      "step": 41192
    },
    {
      "epoch": 2.3066972785306303,
      "grad_norm": 1.1131529808044434,
      "learning_rate": 9.442789473684212e-05,
      "loss": 0.2997,
      "step": 41193
    },
    {
      "epoch": 2.3067532758427594,
      "grad_norm": 1.0302926301956177,
      "learning_rate": 9.442763157894737e-05,
      "loss": 0.3515,
      "step": 41194
    },
    {
      "epoch": 2.3068092731548884,
      "grad_norm": 1.3277411460876465,
      "learning_rate": 9.442736842105263e-05,
      "loss": 0.3476,
      "step": 41195
    },
    {
      "epoch": 2.3068652704670174,
      "grad_norm": 1.1428031921386719,
      "learning_rate": 9.44271052631579e-05,
      "loss": 0.3007,
      "step": 41196
    },
    {
      "epoch": 2.3069212677791464,
      "grad_norm": 1.12355637550354,
      "learning_rate": 9.442684210526316e-05,
      "loss": 0.476,
      "step": 41197
    },
    {
      "epoch": 2.3069772650912754,
      "grad_norm": 1.4822880029678345,
      "learning_rate": 9.442657894736843e-05,
      "loss": 0.497,
      "step": 41198
    },
    {
      "epoch": 2.3070332624034044,
      "grad_norm": 1.2447497844696045,
      "learning_rate": 9.442631578947368e-05,
      "loss": 0.4572,
      "step": 41199
    },
    {
      "epoch": 2.3070892597155335,
      "grad_norm": 3.252469778060913,
      "learning_rate": 9.442605263157895e-05,
      "loss": 0.4996,
      "step": 41200
    },
    {
      "epoch": 2.3071452570276625,
      "grad_norm": 1.3244056701660156,
      "learning_rate": 9.442578947368421e-05,
      "loss": 0.4706,
      "step": 41201
    },
    {
      "epoch": 2.3072012543397915,
      "grad_norm": 5.475678443908691,
      "learning_rate": 9.442552631578949e-05,
      "loss": 0.3791,
      "step": 41202
    },
    {
      "epoch": 2.3072572516519205,
      "grad_norm": 1.1146981716156006,
      "learning_rate": 9.442526315789475e-05,
      "loss": 0.6018,
      "step": 41203
    },
    {
      "epoch": 2.3073132489640495,
      "grad_norm": 1.054796576499939,
      "learning_rate": 9.4425e-05,
      "loss": 0.3865,
      "step": 41204
    },
    {
      "epoch": 2.3073692462761786,
      "grad_norm": 1.4048603773117065,
      "learning_rate": 9.442473684210526e-05,
      "loss": 0.4875,
      "step": 41205
    },
    {
      "epoch": 2.3074252435883076,
      "grad_norm": 1.4637945890426636,
      "learning_rate": 9.442447368421054e-05,
      "loss": 0.4676,
      "step": 41206
    },
    {
      "epoch": 2.3074812409004366,
      "grad_norm": 1.1658724546432495,
      "learning_rate": 9.44242105263158e-05,
      "loss": 0.4202,
      "step": 41207
    },
    {
      "epoch": 2.3075372382125656,
      "grad_norm": 1.299358606338501,
      "learning_rate": 9.442394736842106e-05,
      "loss": 0.6069,
      "step": 41208
    },
    {
      "epoch": 2.3075932355246946,
      "grad_norm": 1.7072272300720215,
      "learning_rate": 9.442368421052632e-05,
      "loss": 0.634,
      "step": 41209
    },
    {
      "epoch": 2.3076492328368237,
      "grad_norm": 1.3296971321105957,
      "learning_rate": 9.442342105263159e-05,
      "loss": 0.5488,
      "step": 41210
    },
    {
      "epoch": 2.3077052301489527,
      "grad_norm": 1.1389737129211426,
      "learning_rate": 9.442315789473685e-05,
      "loss": 0.3998,
      "step": 41211
    },
    {
      "epoch": 2.3077612274610817,
      "grad_norm": 1.3433154821395874,
      "learning_rate": 9.442289473684211e-05,
      "loss": 0.4376,
      "step": 41212
    },
    {
      "epoch": 2.3078172247732107,
      "grad_norm": 1.2962489128112793,
      "learning_rate": 9.442263157894737e-05,
      "loss": 0.3559,
      "step": 41213
    },
    {
      "epoch": 2.3078732220853397,
      "grad_norm": 1.194562315940857,
      "learning_rate": 9.442236842105263e-05,
      "loss": 0.3917,
      "step": 41214
    },
    {
      "epoch": 2.3079292193974688,
      "grad_norm": 1.3812541961669922,
      "learning_rate": 9.44221052631579e-05,
      "loss": 0.3673,
      "step": 41215
    },
    {
      "epoch": 2.3079852167095978,
      "grad_norm": 1.4755765199661255,
      "learning_rate": 9.442184210526316e-05,
      "loss": 0.5047,
      "step": 41216
    },
    {
      "epoch": 2.308041214021727,
      "grad_norm": 1.1226434707641602,
      "learning_rate": 9.442157894736842e-05,
      "loss": 0.3681,
      "step": 41217
    },
    {
      "epoch": 2.308097211333856,
      "grad_norm": 1.3759899139404297,
      "learning_rate": 9.442131578947368e-05,
      "loss": 0.4198,
      "step": 41218
    },
    {
      "epoch": 2.308153208645985,
      "grad_norm": 1.372986912727356,
      "learning_rate": 9.442105263157895e-05,
      "loss": 0.4164,
      "step": 41219
    },
    {
      "epoch": 2.308209205958114,
      "grad_norm": 1.2326743602752686,
      "learning_rate": 9.442078947368421e-05,
      "loss": 0.3656,
      "step": 41220
    },
    {
      "epoch": 2.308265203270243,
      "grad_norm": 1.4366059303283691,
      "learning_rate": 9.442052631578949e-05,
      "loss": 0.5534,
      "step": 41221
    },
    {
      "epoch": 2.308321200582372,
      "grad_norm": 1.2622298002243042,
      "learning_rate": 9.442026315789473e-05,
      "loss": 0.3881,
      "step": 41222
    },
    {
      "epoch": 2.308377197894501,
      "grad_norm": 1.1661293506622314,
      "learning_rate": 9.442000000000001e-05,
      "loss": 0.3973,
      "step": 41223
    },
    {
      "epoch": 2.30843319520663,
      "grad_norm": 1.1904220581054688,
      "learning_rate": 9.441973684210527e-05,
      "loss": 0.4137,
      "step": 41224
    },
    {
      "epoch": 2.308489192518759,
      "grad_norm": 1.3862000703811646,
      "learning_rate": 9.441947368421054e-05,
      "loss": 0.3895,
      "step": 41225
    },
    {
      "epoch": 2.308545189830888,
      "grad_norm": 1.1853761672973633,
      "learning_rate": 9.441921052631579e-05,
      "loss": 0.4411,
      "step": 41226
    },
    {
      "epoch": 2.308601187143017,
      "grad_norm": 1.4763569831848145,
      "learning_rate": 9.441894736842106e-05,
      "loss": 0.5152,
      "step": 41227
    },
    {
      "epoch": 2.308657184455146,
      "grad_norm": 1.10613214969635,
      "learning_rate": 9.441868421052632e-05,
      "loss": 0.423,
      "step": 41228
    },
    {
      "epoch": 2.308713181767275,
      "grad_norm": 1.2948658466339111,
      "learning_rate": 9.441842105263158e-05,
      "loss": 0.367,
      "step": 41229
    },
    {
      "epoch": 2.308769179079404,
      "grad_norm": 1.1493228673934937,
      "learning_rate": 9.441815789473685e-05,
      "loss": 0.3077,
      "step": 41230
    },
    {
      "epoch": 2.308825176391533,
      "grad_norm": 1.3492008447647095,
      "learning_rate": 9.44178947368421e-05,
      "loss": 0.4266,
      "step": 41231
    },
    {
      "epoch": 2.308881173703662,
      "grad_norm": 1.0226320028305054,
      "learning_rate": 9.441763157894737e-05,
      "loss": 0.3619,
      "step": 41232
    },
    {
      "epoch": 2.308937171015791,
      "grad_norm": 1.1736783981323242,
      "learning_rate": 9.441736842105263e-05,
      "loss": 0.4611,
      "step": 41233
    },
    {
      "epoch": 2.30899316832792,
      "grad_norm": 1.4075218439102173,
      "learning_rate": 9.44171052631579e-05,
      "loss": 0.601,
      "step": 41234
    },
    {
      "epoch": 2.309049165640049,
      "grad_norm": 1.0453821420669556,
      "learning_rate": 9.441684210526316e-05,
      "loss": 0.2935,
      "step": 41235
    },
    {
      "epoch": 2.309105162952178,
      "grad_norm": 1.3486816883087158,
      "learning_rate": 9.441657894736842e-05,
      "loss": 0.5075,
      "step": 41236
    },
    {
      "epoch": 2.309161160264307,
      "grad_norm": 1.1429842710494995,
      "learning_rate": 9.441631578947368e-05,
      "loss": 0.4051,
      "step": 41237
    },
    {
      "epoch": 2.309217157576436,
      "grad_norm": 1.2654139995574951,
      "learning_rate": 9.441605263157896e-05,
      "loss": 0.436,
      "step": 41238
    },
    {
      "epoch": 2.309273154888565,
      "grad_norm": 1.1187604665756226,
      "learning_rate": 9.441578947368422e-05,
      "loss": 0.4698,
      "step": 41239
    },
    {
      "epoch": 2.3093291522006942,
      "grad_norm": 1.3010759353637695,
      "learning_rate": 9.441552631578948e-05,
      "loss": 0.4163,
      "step": 41240
    },
    {
      "epoch": 2.3093851495128233,
      "grad_norm": 1.2733290195465088,
      "learning_rate": 9.441526315789474e-05,
      "loss": 0.3791,
      "step": 41241
    },
    {
      "epoch": 2.3094411468249523,
      "grad_norm": 1.2594940662384033,
      "learning_rate": 9.441500000000001e-05,
      "loss": 0.4574,
      "step": 41242
    },
    {
      "epoch": 2.3094971441370813,
      "grad_norm": 1.2382731437683105,
      "learning_rate": 9.441473684210527e-05,
      "loss": 0.4378,
      "step": 41243
    },
    {
      "epoch": 2.3095531414492103,
      "grad_norm": 1.1887248754501343,
      "learning_rate": 9.441447368421053e-05,
      "loss": 0.3804,
      "step": 41244
    },
    {
      "epoch": 2.3096091387613393,
      "grad_norm": 1.1392658948898315,
      "learning_rate": 9.441421052631579e-05,
      "loss": 0.3676,
      "step": 41245
    },
    {
      "epoch": 2.3096651360734684,
      "grad_norm": 1.0222123861312866,
      "learning_rate": 9.441394736842105e-05,
      "loss": 0.4313,
      "step": 41246
    },
    {
      "epoch": 2.3097211333855974,
      "grad_norm": 1.0896024703979492,
      "learning_rate": 9.441368421052632e-05,
      "loss": 0.3613,
      "step": 41247
    },
    {
      "epoch": 2.3097771306977264,
      "grad_norm": 1.2499682903289795,
      "learning_rate": 9.441342105263158e-05,
      "loss": 0.3702,
      "step": 41248
    },
    {
      "epoch": 2.3098331280098554,
      "grad_norm": 1.5982550382614136,
      "learning_rate": 9.441315789473684e-05,
      "loss": 0.3645,
      "step": 41249
    },
    {
      "epoch": 2.3098891253219844,
      "grad_norm": 1.1547956466674805,
      "learning_rate": 9.44128947368421e-05,
      "loss": 0.4226,
      "step": 41250
    },
    {
      "epoch": 2.3099451226341134,
      "grad_norm": 1.1691392660140991,
      "learning_rate": 9.441263157894737e-05,
      "loss": 0.3692,
      "step": 41251
    },
    {
      "epoch": 2.3100011199462425,
      "grad_norm": 1.3276208639144897,
      "learning_rate": 9.441236842105263e-05,
      "loss": 0.427,
      "step": 41252
    },
    {
      "epoch": 2.3100571172583715,
      "grad_norm": 1.1660151481628418,
      "learning_rate": 9.441210526315791e-05,
      "loss": 0.4639,
      "step": 41253
    },
    {
      "epoch": 2.3101131145705005,
      "grad_norm": 1.5728514194488525,
      "learning_rate": 9.441184210526315e-05,
      "loss": 0.4579,
      "step": 41254
    },
    {
      "epoch": 2.3101691118826295,
      "grad_norm": 1.5247868299484253,
      "learning_rate": 9.441157894736843e-05,
      "loss": 0.5588,
      "step": 41255
    },
    {
      "epoch": 2.3102251091947585,
      "grad_norm": 0.9623777270317078,
      "learning_rate": 9.441131578947369e-05,
      "loss": 0.2966,
      "step": 41256
    },
    {
      "epoch": 2.3102811065068876,
      "grad_norm": 15.850016593933105,
      "learning_rate": 9.441105263157896e-05,
      "loss": 0.398,
      "step": 41257
    },
    {
      "epoch": 2.3103371038190166,
      "grad_norm": 1.3107852935791016,
      "learning_rate": 9.441078947368422e-05,
      "loss": 0.5054,
      "step": 41258
    },
    {
      "epoch": 2.3103931011311456,
      "grad_norm": 1.3028587102890015,
      "learning_rate": 9.441052631578948e-05,
      "loss": 0.4868,
      "step": 41259
    },
    {
      "epoch": 2.3104490984432746,
      "grad_norm": 1.3456355333328247,
      "learning_rate": 9.441026315789474e-05,
      "loss": 0.5253,
      "step": 41260
    },
    {
      "epoch": 2.3105050957554036,
      "grad_norm": 1.1589744091033936,
      "learning_rate": 9.441000000000001e-05,
      "loss": 0.3649,
      "step": 41261
    },
    {
      "epoch": 2.3105610930675327,
      "grad_norm": 1.3102086782455444,
      "learning_rate": 9.440973684210527e-05,
      "loss": 0.4942,
      "step": 41262
    },
    {
      "epoch": 2.3106170903796617,
      "grad_norm": 0.9877236485481262,
      "learning_rate": 9.440947368421053e-05,
      "loss": 0.2984,
      "step": 41263
    },
    {
      "epoch": 2.3106730876917907,
      "grad_norm": 1.7010084390640259,
      "learning_rate": 9.440921052631579e-05,
      "loss": 0.5229,
      "step": 41264
    },
    {
      "epoch": 2.3107290850039197,
      "grad_norm": 1.4839822053909302,
      "learning_rate": 9.440894736842105e-05,
      "loss": 0.3877,
      "step": 41265
    },
    {
      "epoch": 2.3107850823160487,
      "grad_norm": 1.3590247631072998,
      "learning_rate": 9.440868421052632e-05,
      "loss": 0.4533,
      "step": 41266
    },
    {
      "epoch": 2.3108410796281778,
      "grad_norm": 1.2635881900787354,
      "learning_rate": 9.440842105263158e-05,
      "loss": 0.4093,
      "step": 41267
    },
    {
      "epoch": 2.3108970769403068,
      "grad_norm": 1.3417646884918213,
      "learning_rate": 9.440815789473684e-05,
      "loss": 0.3533,
      "step": 41268
    },
    {
      "epoch": 2.310953074252436,
      "grad_norm": 1.9569975137710571,
      "learning_rate": 9.44078947368421e-05,
      "loss": 0.4248,
      "step": 41269
    },
    {
      "epoch": 2.311009071564565,
      "grad_norm": 1.1856062412261963,
      "learning_rate": 9.440763157894738e-05,
      "loss": 0.3642,
      "step": 41270
    },
    {
      "epoch": 2.311065068876694,
      "grad_norm": 1.1207315921783447,
      "learning_rate": 9.440736842105264e-05,
      "loss": 0.3766,
      "step": 41271
    },
    {
      "epoch": 2.311121066188823,
      "grad_norm": 1.1116873025894165,
      "learning_rate": 9.44071052631579e-05,
      "loss": 0.3486,
      "step": 41272
    },
    {
      "epoch": 2.311177063500952,
      "grad_norm": 1.0138052701950073,
      "learning_rate": 9.440684210526316e-05,
      "loss": 0.3452,
      "step": 41273
    },
    {
      "epoch": 2.311233060813081,
      "grad_norm": 3.0268807411193848,
      "learning_rate": 9.440657894736843e-05,
      "loss": 0.3628,
      "step": 41274
    },
    {
      "epoch": 2.31128905812521,
      "grad_norm": 1.2343382835388184,
      "learning_rate": 9.440631578947369e-05,
      "loss": 0.5134,
      "step": 41275
    },
    {
      "epoch": 2.311345055437339,
      "grad_norm": 1.1045588254928589,
      "learning_rate": 9.440605263157896e-05,
      "loss": 0.4906,
      "step": 41276
    },
    {
      "epoch": 2.311401052749468,
      "grad_norm": 1.7074835300445557,
      "learning_rate": 9.440578947368421e-05,
      "loss": 0.5393,
      "step": 41277
    },
    {
      "epoch": 2.311457050061597,
      "grad_norm": 1.5158981084823608,
      "learning_rate": 9.440552631578948e-05,
      "loss": 0.3874,
      "step": 41278
    },
    {
      "epoch": 2.311513047373726,
      "grad_norm": 1.138958215713501,
      "learning_rate": 9.440526315789474e-05,
      "loss": 0.3779,
      "step": 41279
    },
    {
      "epoch": 2.311569044685855,
      "grad_norm": 1.3646339178085327,
      "learning_rate": 9.4405e-05,
      "loss": 0.4031,
      "step": 41280
    },
    {
      "epoch": 2.311625041997984,
      "grad_norm": 1.1444863080978394,
      "learning_rate": 9.440473684210526e-05,
      "loss": 0.3771,
      "step": 41281
    },
    {
      "epoch": 2.311681039310113,
      "grad_norm": 1.1277296543121338,
      "learning_rate": 9.440447368421052e-05,
      "loss": 0.4182,
      "step": 41282
    },
    {
      "epoch": 2.311737036622242,
      "grad_norm": 1.7238092422485352,
      "learning_rate": 9.44042105263158e-05,
      "loss": 0.451,
      "step": 41283
    },
    {
      "epoch": 2.311793033934371,
      "grad_norm": 1.0843058824539185,
      "learning_rate": 9.440394736842105e-05,
      "loss": 0.4729,
      "step": 41284
    },
    {
      "epoch": 2.3118490312465,
      "grad_norm": 1.1807172298431396,
      "learning_rate": 9.440368421052633e-05,
      "loss": 0.5245,
      "step": 41285
    },
    {
      "epoch": 2.311905028558629,
      "grad_norm": 1.1698341369628906,
      "learning_rate": 9.440342105263157e-05,
      "loss": 0.4166,
      "step": 41286
    },
    {
      "epoch": 2.311961025870758,
      "grad_norm": 1.026491403579712,
      "learning_rate": 9.440315789473685e-05,
      "loss": 0.3069,
      "step": 41287
    },
    {
      "epoch": 2.312017023182887,
      "grad_norm": 1.4497567415237427,
      "learning_rate": 9.44028947368421e-05,
      "loss": 0.3443,
      "step": 41288
    },
    {
      "epoch": 2.312073020495016,
      "grad_norm": 1.2242244482040405,
      "learning_rate": 9.440263157894738e-05,
      "loss": 0.3821,
      "step": 41289
    },
    {
      "epoch": 2.312129017807145,
      "grad_norm": 1.2092628479003906,
      "learning_rate": 9.440236842105264e-05,
      "loss": 0.4558,
      "step": 41290
    },
    {
      "epoch": 2.312185015119274,
      "grad_norm": 1.1276986598968506,
      "learning_rate": 9.44021052631579e-05,
      "loss": 0.3331,
      "step": 41291
    },
    {
      "epoch": 2.3122410124314032,
      "grad_norm": 1.174345850944519,
      "learning_rate": 9.440184210526316e-05,
      "loss": 0.3326,
      "step": 41292
    },
    {
      "epoch": 2.3122970097435323,
      "grad_norm": 1.0791987180709839,
      "learning_rate": 9.440157894736843e-05,
      "loss": 0.3382,
      "step": 41293
    },
    {
      "epoch": 2.3123530070556613,
      "grad_norm": 1.0942697525024414,
      "learning_rate": 9.440131578947369e-05,
      "loss": 0.3754,
      "step": 41294
    },
    {
      "epoch": 2.3124090043677903,
      "grad_norm": 1.3148592710494995,
      "learning_rate": 9.440105263157895e-05,
      "loss": 0.5432,
      "step": 41295
    },
    {
      "epoch": 2.3124650016799193,
      "grad_norm": 1.2570302486419678,
      "learning_rate": 9.440078947368421e-05,
      "loss": 0.4145,
      "step": 41296
    },
    {
      "epoch": 2.3125209989920483,
      "grad_norm": 1.1952694654464722,
      "learning_rate": 9.440052631578948e-05,
      "loss": 0.2893,
      "step": 41297
    },
    {
      "epoch": 2.3125769963041773,
      "grad_norm": 1.060887098312378,
      "learning_rate": 9.440026315789474e-05,
      "loss": 0.3543,
      "step": 41298
    },
    {
      "epoch": 2.3126329936163064,
      "grad_norm": 1.3482955694198608,
      "learning_rate": 9.44e-05,
      "loss": 0.3677,
      "step": 41299
    },
    {
      "epoch": 2.3126889909284354,
      "grad_norm": 1.1082282066345215,
      "learning_rate": 9.439973684210526e-05,
      "loss": 0.3493,
      "step": 41300
    },
    {
      "epoch": 2.3127449882405644,
      "grad_norm": 0.9989748597145081,
      "learning_rate": 9.439947368421052e-05,
      "loss": 0.3754,
      "step": 41301
    },
    {
      "epoch": 2.3128009855526934,
      "grad_norm": 1.4040677547454834,
      "learning_rate": 9.43992105263158e-05,
      "loss": 0.5438,
      "step": 41302
    },
    {
      "epoch": 2.3128569828648224,
      "grad_norm": 1.5152626037597656,
      "learning_rate": 9.439894736842106e-05,
      "loss": 0.5051,
      "step": 41303
    },
    {
      "epoch": 2.3129129801769515,
      "grad_norm": 1.204582691192627,
      "learning_rate": 9.439868421052632e-05,
      "loss": 0.365,
      "step": 41304
    },
    {
      "epoch": 2.3129689774890805,
      "grad_norm": 1.354763150215149,
      "learning_rate": 9.439842105263158e-05,
      "loss": 0.3932,
      "step": 41305
    },
    {
      "epoch": 2.3130249748012095,
      "grad_norm": 1.3283936977386475,
      "learning_rate": 9.439815789473685e-05,
      "loss": 0.3773,
      "step": 41306
    },
    {
      "epoch": 2.3130809721133385,
      "grad_norm": 1.0832765102386475,
      "learning_rate": 9.439789473684211e-05,
      "loss": 0.3317,
      "step": 41307
    },
    {
      "epoch": 2.3131369694254675,
      "grad_norm": 1.1603450775146484,
      "learning_rate": 9.439763157894738e-05,
      "loss": 0.3382,
      "step": 41308
    },
    {
      "epoch": 2.3131929667375966,
      "grad_norm": 1.197878360748291,
      "learning_rate": 9.439736842105263e-05,
      "loss": 0.3889,
      "step": 41309
    },
    {
      "epoch": 2.3132489640497256,
      "grad_norm": 1.4172712564468384,
      "learning_rate": 9.43971052631579e-05,
      "loss": 0.3217,
      "step": 41310
    },
    {
      "epoch": 2.3133049613618546,
      "grad_norm": 1.192231297492981,
      "learning_rate": 9.439684210526316e-05,
      "loss": 0.5098,
      "step": 41311
    },
    {
      "epoch": 2.3133609586739836,
      "grad_norm": 1.0548523664474487,
      "learning_rate": 9.439657894736843e-05,
      "loss": 0.3614,
      "step": 41312
    },
    {
      "epoch": 2.3134169559861126,
      "grad_norm": 1.1855655908584595,
      "learning_rate": 9.43963157894737e-05,
      "loss": 0.3765,
      "step": 41313
    },
    {
      "epoch": 2.3134729532982417,
      "grad_norm": 1.0742753744125366,
      "learning_rate": 9.439605263157895e-05,
      "loss": 0.4044,
      "step": 41314
    },
    {
      "epoch": 2.3135289506103707,
      "grad_norm": 1.7576572895050049,
      "learning_rate": 9.439578947368421e-05,
      "loss": 0.6979,
      "step": 41315
    },
    {
      "epoch": 2.3135849479224997,
      "grad_norm": 1.232686161994934,
      "learning_rate": 9.439552631578947e-05,
      "loss": 0.3823,
      "step": 41316
    },
    {
      "epoch": 2.3136409452346287,
      "grad_norm": 1.371396541595459,
      "learning_rate": 9.439526315789475e-05,
      "loss": 0.4225,
      "step": 41317
    },
    {
      "epoch": 2.3136969425467577,
      "grad_norm": 1.556419849395752,
      "learning_rate": 9.439500000000001e-05,
      "loss": 0.562,
      "step": 41318
    },
    {
      "epoch": 2.3137529398588867,
      "grad_norm": 3.6189262866973877,
      "learning_rate": 9.439473684210527e-05,
      "loss": 0.415,
      "step": 41319
    },
    {
      "epoch": 2.3138089371710158,
      "grad_norm": 1.14079749584198,
      "learning_rate": 9.439447368421053e-05,
      "loss": 0.3497,
      "step": 41320
    },
    {
      "epoch": 2.313864934483145,
      "grad_norm": 0.9735620617866516,
      "learning_rate": 9.43942105263158e-05,
      "loss": 0.3399,
      "step": 41321
    },
    {
      "epoch": 2.313920931795274,
      "grad_norm": 1.159853458404541,
      "learning_rate": 9.439394736842106e-05,
      "loss": 0.3901,
      "step": 41322
    },
    {
      "epoch": 2.313976929107403,
      "grad_norm": 1.1504665613174438,
      "learning_rate": 9.439368421052632e-05,
      "loss": 0.372,
      "step": 41323
    },
    {
      "epoch": 2.314032926419532,
      "grad_norm": 1.0760533809661865,
      "learning_rate": 9.439342105263158e-05,
      "loss": 0.3904,
      "step": 41324
    },
    {
      "epoch": 2.314088923731661,
      "grad_norm": 1.1629009246826172,
      "learning_rate": 9.439315789473685e-05,
      "loss": 0.5327,
      "step": 41325
    },
    {
      "epoch": 2.31414492104379,
      "grad_norm": 1.285649061203003,
      "learning_rate": 9.439289473684211e-05,
      "loss": 0.477,
      "step": 41326
    },
    {
      "epoch": 2.314200918355919,
      "grad_norm": 1.1474159955978394,
      "learning_rate": 9.439263157894737e-05,
      "loss": 0.3918,
      "step": 41327
    },
    {
      "epoch": 2.314256915668048,
      "grad_norm": 1.158265233039856,
      "learning_rate": 9.439236842105263e-05,
      "loss": 0.4011,
      "step": 41328
    },
    {
      "epoch": 2.314312912980177,
      "grad_norm": 1.3388100862503052,
      "learning_rate": 9.43921052631579e-05,
      "loss": 0.4331,
      "step": 41329
    },
    {
      "epoch": 2.314368910292306,
      "grad_norm": 1.1474487781524658,
      "learning_rate": 9.439184210526316e-05,
      "loss": 0.3586,
      "step": 41330
    },
    {
      "epoch": 2.314424907604435,
      "grad_norm": 1.3449229001998901,
      "learning_rate": 9.439157894736844e-05,
      "loss": 0.4732,
      "step": 41331
    },
    {
      "epoch": 2.314480904916564,
      "grad_norm": 0.9961864352226257,
      "learning_rate": 9.439131578947368e-05,
      "loss": 0.4983,
      "step": 41332
    },
    {
      "epoch": 2.314536902228693,
      "grad_norm": 1.0431329011917114,
      "learning_rate": 9.439105263157894e-05,
      "loss": 0.3561,
      "step": 41333
    },
    {
      "epoch": 2.314592899540822,
      "grad_norm": 1.059946060180664,
      "learning_rate": 9.439078947368422e-05,
      "loss": 0.3425,
      "step": 41334
    },
    {
      "epoch": 2.314648896852951,
      "grad_norm": 1.4289476871490479,
      "learning_rate": 9.439052631578948e-05,
      "loss": 0.4408,
      "step": 41335
    },
    {
      "epoch": 2.31470489416508,
      "grad_norm": 1.2132060527801514,
      "learning_rate": 9.439026315789474e-05,
      "loss": 0.4205,
      "step": 41336
    },
    {
      "epoch": 2.314760891477209,
      "grad_norm": 0.9894360899925232,
      "learning_rate": 9.439e-05,
      "loss": 0.3407,
      "step": 41337
    },
    {
      "epoch": 2.314816888789338,
      "grad_norm": 1.1171969175338745,
      "learning_rate": 9.438973684210527e-05,
      "loss": 0.4879,
      "step": 41338
    },
    {
      "epoch": 2.314872886101467,
      "grad_norm": 1.1477209329605103,
      "learning_rate": 9.438947368421053e-05,
      "loss": 0.4461,
      "step": 41339
    },
    {
      "epoch": 2.314928883413596,
      "grad_norm": 1.2204196453094482,
      "learning_rate": 9.43892105263158e-05,
      "loss": 0.5072,
      "step": 41340
    },
    {
      "epoch": 2.314984880725725,
      "grad_norm": 1.3188509941101074,
      "learning_rate": 9.438894736842105e-05,
      "loss": 0.6795,
      "step": 41341
    },
    {
      "epoch": 2.315040878037854,
      "grad_norm": 1.0996758937835693,
      "learning_rate": 9.438868421052632e-05,
      "loss": 0.3322,
      "step": 41342
    },
    {
      "epoch": 2.315096875349983,
      "grad_norm": 1.3580321073532104,
      "learning_rate": 9.438842105263158e-05,
      "loss": 0.4266,
      "step": 41343
    },
    {
      "epoch": 2.3151528726621122,
      "grad_norm": 1.1975034475326538,
      "learning_rate": 9.438815789473685e-05,
      "loss": 0.4652,
      "step": 41344
    },
    {
      "epoch": 2.3152088699742412,
      "grad_norm": 1.2128863334655762,
      "learning_rate": 9.438789473684211e-05,
      "loss": 0.4818,
      "step": 41345
    },
    {
      "epoch": 2.3152648672863703,
      "grad_norm": 1.1107937097549438,
      "learning_rate": 9.438763157894737e-05,
      "loss": 0.4031,
      "step": 41346
    },
    {
      "epoch": 2.3153208645984993,
      "grad_norm": 1.0903804302215576,
      "learning_rate": 9.438736842105263e-05,
      "loss": 0.3444,
      "step": 41347
    },
    {
      "epoch": 2.3153768619106283,
      "grad_norm": 1.4978266954421997,
      "learning_rate": 9.438710526315791e-05,
      "loss": 0.3729,
      "step": 41348
    },
    {
      "epoch": 2.3154328592227573,
      "grad_norm": 1.3032159805297852,
      "learning_rate": 9.438684210526317e-05,
      "loss": 0.5783,
      "step": 41349
    },
    {
      "epoch": 2.3154888565348863,
      "grad_norm": 1.316075086593628,
      "learning_rate": 9.438657894736843e-05,
      "loss": 0.3711,
      "step": 41350
    },
    {
      "epoch": 2.3155448538470154,
      "grad_norm": 1.1292375326156616,
      "learning_rate": 9.438631578947369e-05,
      "loss": 0.3986,
      "step": 41351
    },
    {
      "epoch": 2.3156008511591444,
      "grad_norm": 1.2872331142425537,
      "learning_rate": 9.438605263157895e-05,
      "loss": 0.3025,
      "step": 41352
    },
    {
      "epoch": 2.3156568484712734,
      "grad_norm": 1.060409665107727,
      "learning_rate": 9.438578947368422e-05,
      "loss": 0.3404,
      "step": 41353
    },
    {
      "epoch": 2.3157128457834024,
      "grad_norm": 1.3974651098251343,
      "learning_rate": 9.438552631578948e-05,
      "loss": 0.3759,
      "step": 41354
    },
    {
      "epoch": 2.3157688430955314,
      "grad_norm": 1.255693793296814,
      "learning_rate": 9.438526315789474e-05,
      "loss": 0.4282,
      "step": 41355
    },
    {
      "epoch": 2.3158248404076605,
      "grad_norm": 1.0562257766723633,
      "learning_rate": 9.4385e-05,
      "loss": 0.3347,
      "step": 41356
    },
    {
      "epoch": 2.3158808377197895,
      "grad_norm": 1.2200738191604614,
      "learning_rate": 9.438473684210527e-05,
      "loss": 0.3947,
      "step": 41357
    },
    {
      "epoch": 2.3159368350319185,
      "grad_norm": 1.255780816078186,
      "learning_rate": 9.438447368421053e-05,
      "loss": 0.4157,
      "step": 41358
    },
    {
      "epoch": 2.3159928323440475,
      "grad_norm": 1.2522015571594238,
      "learning_rate": 9.438421052631579e-05,
      "loss": 0.3716,
      "step": 41359
    },
    {
      "epoch": 2.3160488296561765,
      "grad_norm": 1.108750820159912,
      "learning_rate": 9.438394736842105e-05,
      "loss": 0.4209,
      "step": 41360
    },
    {
      "epoch": 2.3161048269683056,
      "grad_norm": 1.1700783967971802,
      "learning_rate": 9.438368421052632e-05,
      "loss": 0.3629,
      "step": 41361
    },
    {
      "epoch": 2.3161608242804346,
      "grad_norm": 1.0911403894424438,
      "learning_rate": 9.438342105263158e-05,
      "loss": 0.3663,
      "step": 41362
    },
    {
      "epoch": 2.3162168215925636,
      "grad_norm": 1.2353960275650024,
      "learning_rate": 9.438315789473686e-05,
      "loss": 0.3926,
      "step": 41363
    },
    {
      "epoch": 2.3162728189046926,
      "grad_norm": 1.389467477798462,
      "learning_rate": 9.43828947368421e-05,
      "loss": 0.449,
      "step": 41364
    },
    {
      "epoch": 2.3163288162168216,
      "grad_norm": 1.1134833097457886,
      "learning_rate": 9.438263157894738e-05,
      "loss": 0.3548,
      "step": 41365
    },
    {
      "epoch": 2.3163848135289506,
      "grad_norm": 13.240059852600098,
      "learning_rate": 9.438236842105264e-05,
      "loss": 0.4089,
      "step": 41366
    },
    {
      "epoch": 2.3164408108410797,
      "grad_norm": 1.4368284940719604,
      "learning_rate": 9.43821052631579e-05,
      "loss": 0.4111,
      "step": 41367
    },
    {
      "epoch": 2.3164968081532087,
      "grad_norm": 1.2925031185150146,
      "learning_rate": 9.438184210526317e-05,
      "loss": 0.4966,
      "step": 41368
    },
    {
      "epoch": 2.3165528054653377,
      "grad_norm": 1.3029847145080566,
      "learning_rate": 9.438157894736842e-05,
      "loss": 0.3637,
      "step": 41369
    },
    {
      "epoch": 2.3166088027774667,
      "grad_norm": 1.3450571298599243,
      "learning_rate": 9.438131578947369e-05,
      "loss": 0.5743,
      "step": 41370
    },
    {
      "epoch": 2.3166648000895957,
      "grad_norm": 1.221514105796814,
      "learning_rate": 9.438105263157895e-05,
      "loss": 0.4085,
      "step": 41371
    },
    {
      "epoch": 2.3167207974017248,
      "grad_norm": 1.167541742324829,
      "learning_rate": 9.438078947368422e-05,
      "loss": 0.3486,
      "step": 41372
    },
    {
      "epoch": 2.316776794713854,
      "grad_norm": 1.3627599477767944,
      "learning_rate": 9.438052631578948e-05,
      "loss": 0.4794,
      "step": 41373
    },
    {
      "epoch": 2.316832792025983,
      "grad_norm": 1.270850658416748,
      "learning_rate": 9.438026315789474e-05,
      "loss": 0.3536,
      "step": 41374
    },
    {
      "epoch": 2.316888789338112,
      "grad_norm": 1.3900926113128662,
      "learning_rate": 9.438e-05,
      "loss": 0.3641,
      "step": 41375
    },
    {
      "epoch": 2.316944786650241,
      "grad_norm": 1.3106361627578735,
      "learning_rate": 9.437973684210527e-05,
      "loss": 0.4269,
      "step": 41376
    },
    {
      "epoch": 2.31700078396237,
      "grad_norm": 1.3721694946289062,
      "learning_rate": 9.437947368421053e-05,
      "loss": 0.3696,
      "step": 41377
    },
    {
      "epoch": 2.317056781274499,
      "grad_norm": 1.198875904083252,
      "learning_rate": 9.43792105263158e-05,
      "loss": 0.4302,
      "step": 41378
    },
    {
      "epoch": 2.317112778586628,
      "grad_norm": 1.159014344215393,
      "learning_rate": 9.437894736842105e-05,
      "loss": 0.358,
      "step": 41379
    },
    {
      "epoch": 2.317168775898757,
      "grad_norm": 1.1493645906448364,
      "learning_rate": 9.437868421052633e-05,
      "loss": 0.3787,
      "step": 41380
    },
    {
      "epoch": 2.317224773210886,
      "grad_norm": 1.2999175786972046,
      "learning_rate": 9.437842105263159e-05,
      "loss": 0.4319,
      "step": 41381
    },
    {
      "epoch": 2.317280770523015,
      "grad_norm": 1.0108214616775513,
      "learning_rate": 9.437815789473685e-05,
      "loss": 0.3313,
      "step": 41382
    },
    {
      "epoch": 2.317336767835144,
      "grad_norm": 0.9442539811134338,
      "learning_rate": 9.43778947368421e-05,
      "loss": 0.3817,
      "step": 41383
    },
    {
      "epoch": 2.317392765147273,
      "grad_norm": 1.1305065155029297,
      "learning_rate": 9.437763157894737e-05,
      "loss": 0.3805,
      "step": 41384
    },
    {
      "epoch": 2.317448762459402,
      "grad_norm": 1.7780417203903198,
      "learning_rate": 9.437736842105264e-05,
      "loss": 0.499,
      "step": 41385
    },
    {
      "epoch": 2.317504759771531,
      "grad_norm": 1.2618051767349243,
      "learning_rate": 9.43771052631579e-05,
      "loss": 0.3869,
      "step": 41386
    },
    {
      "epoch": 2.31756075708366,
      "grad_norm": 1.7549107074737549,
      "learning_rate": 9.437684210526316e-05,
      "loss": 0.2933,
      "step": 41387
    },
    {
      "epoch": 2.317616754395789,
      "grad_norm": 1.3269871473312378,
      "learning_rate": 9.437657894736842e-05,
      "loss": 0.4403,
      "step": 41388
    },
    {
      "epoch": 2.317672751707918,
      "grad_norm": 1.320339560508728,
      "learning_rate": 9.437631578947369e-05,
      "loss": 0.5584,
      "step": 41389
    },
    {
      "epoch": 2.317728749020047,
      "grad_norm": 15.979132652282715,
      "learning_rate": 9.437605263157895e-05,
      "loss": 0.3821,
      "step": 41390
    },
    {
      "epoch": 2.317784746332176,
      "grad_norm": 1.1556161642074585,
      "learning_rate": 9.437578947368421e-05,
      "loss": 0.3353,
      "step": 41391
    },
    {
      "epoch": 2.317840743644305,
      "grad_norm": 0.9923631548881531,
      "learning_rate": 9.437552631578947e-05,
      "loss": 0.3342,
      "step": 41392
    },
    {
      "epoch": 2.317896740956434,
      "grad_norm": 1.2832163572311401,
      "learning_rate": 9.437526315789474e-05,
      "loss": 0.4033,
      "step": 41393
    },
    {
      "epoch": 2.317952738268563,
      "grad_norm": 1.159406304359436,
      "learning_rate": 9.4375e-05,
      "loss": 0.3435,
      "step": 41394
    },
    {
      "epoch": 2.318008735580692,
      "grad_norm": 1.163432240486145,
      "learning_rate": 9.437473684210528e-05,
      "loss": 0.2772,
      "step": 41395
    },
    {
      "epoch": 2.3180647328928212,
      "grad_norm": 1.2027949094772339,
      "learning_rate": 9.437447368421052e-05,
      "loss": 0.3307,
      "step": 41396
    },
    {
      "epoch": 2.3181207302049502,
      "grad_norm": 1.23633873462677,
      "learning_rate": 9.43742105263158e-05,
      "loss": 0.3789,
      "step": 41397
    },
    {
      "epoch": 2.3181767275170793,
      "grad_norm": 1.8226385116577148,
      "learning_rate": 9.437394736842106e-05,
      "loss": 0.3312,
      "step": 41398
    },
    {
      "epoch": 2.3182327248292083,
      "grad_norm": 1.2587995529174805,
      "learning_rate": 9.437368421052633e-05,
      "loss": 0.4833,
      "step": 41399
    },
    {
      "epoch": 2.3182887221413373,
      "grad_norm": 1.324337363243103,
      "learning_rate": 9.437342105263159e-05,
      "loss": 0.4096,
      "step": 41400
    },
    {
      "epoch": 2.3183447194534663,
      "grad_norm": 1.2407947778701782,
      "learning_rate": 9.437315789473684e-05,
      "loss": 0.4042,
      "step": 41401
    },
    {
      "epoch": 2.3184007167655953,
      "grad_norm": 2.1488277912139893,
      "learning_rate": 9.437289473684211e-05,
      "loss": 0.5565,
      "step": 41402
    },
    {
      "epoch": 2.3184567140777244,
      "grad_norm": 1.1434844732284546,
      "learning_rate": 9.437263157894737e-05,
      "loss": 0.3578,
      "step": 41403
    },
    {
      "epoch": 2.3185127113898534,
      "grad_norm": 1.7001357078552246,
      "learning_rate": 9.437236842105264e-05,
      "loss": 0.5056,
      "step": 41404
    },
    {
      "epoch": 2.3185687087019824,
      "grad_norm": 1.2785147428512573,
      "learning_rate": 9.43721052631579e-05,
      "loss": 0.4721,
      "step": 41405
    },
    {
      "epoch": 2.3186247060141114,
      "grad_norm": 1.1670337915420532,
      "learning_rate": 9.437184210526316e-05,
      "loss": 0.4096,
      "step": 41406
    },
    {
      "epoch": 2.3186807033262404,
      "grad_norm": 1.314004898071289,
      "learning_rate": 9.437157894736842e-05,
      "loss": 0.3949,
      "step": 41407
    },
    {
      "epoch": 2.3187367006383695,
      "grad_norm": 1.181923508644104,
      "learning_rate": 9.43713157894737e-05,
      "loss": 0.3221,
      "step": 41408
    },
    {
      "epoch": 2.3187926979504985,
      "grad_norm": 1.2725424766540527,
      "learning_rate": 9.437105263157895e-05,
      "loss": 0.4359,
      "step": 41409
    },
    {
      "epoch": 2.3188486952626275,
      "grad_norm": 1.3356045484542847,
      "learning_rate": 9.437078947368421e-05,
      "loss": 0.4832,
      "step": 41410
    },
    {
      "epoch": 2.3189046925747565,
      "grad_norm": 1.111527681350708,
      "learning_rate": 9.437052631578947e-05,
      "loss": 0.3469,
      "step": 41411
    },
    {
      "epoch": 2.3189606898868855,
      "grad_norm": 1.1512436866760254,
      "learning_rate": 9.437026315789475e-05,
      "loss": 0.4355,
      "step": 41412
    },
    {
      "epoch": 2.3190166871990145,
      "grad_norm": 1.1099945306777954,
      "learning_rate": 9.437e-05,
      "loss": 0.3806,
      "step": 41413
    },
    {
      "epoch": 2.3190726845111436,
      "grad_norm": 1.3678765296936035,
      "learning_rate": 9.436973684210527e-05,
      "loss": 0.4302,
      "step": 41414
    },
    {
      "epoch": 2.3191286818232726,
      "grad_norm": 1.2139514684677124,
      "learning_rate": 9.436947368421053e-05,
      "loss": 0.379,
      "step": 41415
    },
    {
      "epoch": 2.3191846791354016,
      "grad_norm": 5.6088690757751465,
      "learning_rate": 9.43692105263158e-05,
      "loss": 0.6149,
      "step": 41416
    },
    {
      "epoch": 2.3192406764475306,
      "grad_norm": 1.440214991569519,
      "learning_rate": 9.436894736842106e-05,
      "loss": 0.4052,
      "step": 41417
    },
    {
      "epoch": 2.3192966737596596,
      "grad_norm": 1.5163050889968872,
      "learning_rate": 9.436868421052633e-05,
      "loss": 0.407,
      "step": 41418
    },
    {
      "epoch": 2.3193526710717887,
      "grad_norm": 6.795169830322266,
      "learning_rate": 9.436842105263158e-05,
      "loss": 0.3733,
      "step": 41419
    },
    {
      "epoch": 2.3194086683839177,
      "grad_norm": 1.2232766151428223,
      "learning_rate": 9.436815789473684e-05,
      "loss": 0.4208,
      "step": 41420
    },
    {
      "epoch": 2.3194646656960467,
      "grad_norm": 1.642359972000122,
      "learning_rate": 9.436789473684211e-05,
      "loss": 0.4789,
      "step": 41421
    },
    {
      "epoch": 2.3195206630081757,
      "grad_norm": 1.1009441614151,
      "learning_rate": 9.436763157894737e-05,
      "loss": 0.4159,
      "step": 41422
    },
    {
      "epoch": 2.3195766603203047,
      "grad_norm": 1.159757375717163,
      "learning_rate": 9.436736842105264e-05,
      "loss": 0.376,
      "step": 41423
    },
    {
      "epoch": 2.3196326576324338,
      "grad_norm": 1.082768201828003,
      "learning_rate": 9.436710526315789e-05,
      "loss": 0.3205,
      "step": 41424
    },
    {
      "epoch": 2.3196886549445628,
      "grad_norm": 1.617936134338379,
      "learning_rate": 9.436684210526316e-05,
      "loss": 0.4031,
      "step": 41425
    },
    {
      "epoch": 2.319744652256692,
      "grad_norm": 1.8554003238677979,
      "learning_rate": 9.436657894736842e-05,
      "loss": 0.472,
      "step": 41426
    },
    {
      "epoch": 2.319800649568821,
      "grad_norm": 1.246032476425171,
      "learning_rate": 9.43663157894737e-05,
      "loss": 0.3915,
      "step": 41427
    },
    {
      "epoch": 2.31985664688095,
      "grad_norm": 1.2154436111450195,
      "learning_rate": 9.436605263157894e-05,
      "loss": 0.3353,
      "step": 41428
    },
    {
      "epoch": 2.319912644193079,
      "grad_norm": 1.1615893840789795,
      "learning_rate": 9.436578947368422e-05,
      "loss": 0.3052,
      "step": 41429
    },
    {
      "epoch": 2.319968641505208,
      "grad_norm": 1.9925775527954102,
      "learning_rate": 9.436552631578948e-05,
      "loss": 0.2431,
      "step": 41430
    },
    {
      "epoch": 2.320024638817337,
      "grad_norm": 1.0813637971878052,
      "learning_rate": 9.436526315789475e-05,
      "loss": 0.3694,
      "step": 41431
    },
    {
      "epoch": 2.320080636129466,
      "grad_norm": 1.2801499366760254,
      "learning_rate": 9.436500000000001e-05,
      "loss": 0.3849,
      "step": 41432
    },
    {
      "epoch": 2.320136633441595,
      "grad_norm": 1.333127737045288,
      "learning_rate": 9.436473684210527e-05,
      "loss": 0.3736,
      "step": 41433
    },
    {
      "epoch": 2.320192630753724,
      "grad_norm": 1.1614184379577637,
      "learning_rate": 9.436447368421053e-05,
      "loss": 0.4216,
      "step": 41434
    },
    {
      "epoch": 2.320248628065853,
      "grad_norm": 2.762162208557129,
      "learning_rate": 9.43642105263158e-05,
      "loss": 0.4516,
      "step": 41435
    },
    {
      "epoch": 2.320304625377982,
      "grad_norm": 1.6449096202850342,
      "learning_rate": 9.436394736842106e-05,
      "loss": 0.4867,
      "step": 41436
    },
    {
      "epoch": 2.320360622690111,
      "grad_norm": 1.2274833917617798,
      "learning_rate": 9.436368421052632e-05,
      "loss": 0.3648,
      "step": 41437
    },
    {
      "epoch": 2.32041662000224,
      "grad_norm": 1.7491999864578247,
      "learning_rate": 9.436342105263158e-05,
      "loss": 0.5316,
      "step": 41438
    },
    {
      "epoch": 2.320472617314369,
      "grad_norm": 1.2867528200149536,
      "learning_rate": 9.436315789473684e-05,
      "loss": 0.4215,
      "step": 41439
    },
    {
      "epoch": 2.320528614626498,
      "grad_norm": 1.2793112993240356,
      "learning_rate": 9.436289473684211e-05,
      "loss": 0.3446,
      "step": 41440
    },
    {
      "epoch": 2.320584611938627,
      "grad_norm": 2.504293203353882,
      "learning_rate": 9.436263157894737e-05,
      "loss": 0.4524,
      "step": 41441
    },
    {
      "epoch": 2.320640609250756,
      "grad_norm": 1.3625446557998657,
      "learning_rate": 9.436236842105263e-05,
      "loss": 0.4193,
      "step": 41442
    },
    {
      "epoch": 2.320696606562885,
      "grad_norm": 1.12100088596344,
      "learning_rate": 9.436210526315789e-05,
      "loss": 0.3409,
      "step": 41443
    },
    {
      "epoch": 2.320752603875014,
      "grad_norm": 1.2281224727630615,
      "learning_rate": 9.436184210526317e-05,
      "loss": 0.3814,
      "step": 41444
    },
    {
      "epoch": 2.320808601187143,
      "grad_norm": 1.0855356454849243,
      "learning_rate": 9.436157894736843e-05,
      "loss": 0.4202,
      "step": 41445
    },
    {
      "epoch": 2.320864598499272,
      "grad_norm": 1.2163196802139282,
      "learning_rate": 9.436131578947369e-05,
      "loss": 0.3915,
      "step": 41446
    },
    {
      "epoch": 2.320920595811401,
      "grad_norm": 0.9154990911483765,
      "learning_rate": 9.436105263157895e-05,
      "loss": 0.3104,
      "step": 41447
    },
    {
      "epoch": 2.32097659312353,
      "grad_norm": 1.186795711517334,
      "learning_rate": 9.436078947368422e-05,
      "loss": 0.3597,
      "step": 41448
    },
    {
      "epoch": 2.3210325904356592,
      "grad_norm": 1.3870964050292969,
      "learning_rate": 9.436052631578948e-05,
      "loss": 0.4255,
      "step": 41449
    },
    {
      "epoch": 2.3210885877477883,
      "grad_norm": 1.103212833404541,
      "learning_rate": 9.436026315789475e-05,
      "loss": 0.3273,
      "step": 41450
    },
    {
      "epoch": 2.3211445850599173,
      "grad_norm": 2.326517105102539,
      "learning_rate": 9.436e-05,
      "loss": 0.6455,
      "step": 41451
    },
    {
      "epoch": 2.3212005823720463,
      "grad_norm": 1.1867144107818604,
      "learning_rate": 9.435973684210527e-05,
      "loss": 0.3826,
      "step": 41452
    },
    {
      "epoch": 2.3212565796841753,
      "grad_norm": 1.2935075759887695,
      "learning_rate": 9.435947368421053e-05,
      "loss": 0.4337,
      "step": 41453
    },
    {
      "epoch": 2.3213125769963043,
      "grad_norm": 1.2231755256652832,
      "learning_rate": 9.435921052631579e-05,
      "loss": 0.413,
      "step": 41454
    },
    {
      "epoch": 2.3213685743084334,
      "grad_norm": 1.3426060676574707,
      "learning_rate": 9.435894736842106e-05,
      "loss": 0.6001,
      "step": 41455
    },
    {
      "epoch": 2.3214245716205624,
      "grad_norm": 1.193281650543213,
      "learning_rate": 9.435868421052631e-05,
      "loss": 0.4322,
      "step": 41456
    },
    {
      "epoch": 2.3214805689326914,
      "grad_norm": 1.1700212955474854,
      "learning_rate": 9.435842105263158e-05,
      "loss": 0.3625,
      "step": 41457
    },
    {
      "epoch": 2.3215365662448204,
      "grad_norm": 1.2244272232055664,
      "learning_rate": 9.435815789473684e-05,
      "loss": 0.395,
      "step": 41458
    },
    {
      "epoch": 2.3215925635569494,
      "grad_norm": 1.6634242534637451,
      "learning_rate": 9.435789473684212e-05,
      "loss": 0.4192,
      "step": 41459
    },
    {
      "epoch": 2.3216485608690784,
      "grad_norm": 1.1649669408798218,
      "learning_rate": 9.435763157894738e-05,
      "loss": 0.3864,
      "step": 41460
    },
    {
      "epoch": 2.3217045581812075,
      "grad_norm": 1.0379775762557983,
      "learning_rate": 9.435736842105264e-05,
      "loss": 0.3756,
      "step": 41461
    },
    {
      "epoch": 2.3217605554933365,
      "grad_norm": 1.249643087387085,
      "learning_rate": 9.43571052631579e-05,
      "loss": 0.3745,
      "step": 41462
    },
    {
      "epoch": 2.3218165528054655,
      "grad_norm": 1.0316903591156006,
      "learning_rate": 9.435684210526317e-05,
      "loss": 0.4327,
      "step": 41463
    },
    {
      "epoch": 2.3218725501175945,
      "grad_norm": 1.4474607706069946,
      "learning_rate": 9.435657894736843e-05,
      "loss": 0.4343,
      "step": 41464
    },
    {
      "epoch": 2.3219285474297235,
      "grad_norm": 1.108749508857727,
      "learning_rate": 9.435631578947369e-05,
      "loss": 0.4515,
      "step": 41465
    },
    {
      "epoch": 2.3219845447418526,
      "grad_norm": 0.9205725789070129,
      "learning_rate": 9.435605263157895e-05,
      "loss": 0.3079,
      "step": 41466
    },
    {
      "epoch": 2.3220405420539816,
      "grad_norm": 1.2063063383102417,
      "learning_rate": 9.435578947368422e-05,
      "loss": 0.4234,
      "step": 41467
    },
    {
      "epoch": 2.3220965393661106,
      "grad_norm": 1.2108074426651,
      "learning_rate": 9.435552631578948e-05,
      "loss": 0.3142,
      "step": 41468
    },
    {
      "epoch": 2.3221525366782396,
      "grad_norm": 1.282341718673706,
      "learning_rate": 9.435526315789474e-05,
      "loss": 0.4147,
      "step": 41469
    },
    {
      "epoch": 2.3222085339903686,
      "grad_norm": 1.2531129121780396,
      "learning_rate": 9.4355e-05,
      "loss": 0.3466,
      "step": 41470
    },
    {
      "epoch": 2.3222645313024977,
      "grad_norm": 1.1117655038833618,
      "learning_rate": 9.435473684210526e-05,
      "loss": 0.3235,
      "step": 41471
    },
    {
      "epoch": 2.3223205286146267,
      "grad_norm": 1.381260871887207,
      "learning_rate": 9.435447368421053e-05,
      "loss": 0.3854,
      "step": 41472
    },
    {
      "epoch": 2.3223765259267557,
      "grad_norm": 1.1924549341201782,
      "learning_rate": 9.435421052631579e-05,
      "loss": 0.3963,
      "step": 41473
    },
    {
      "epoch": 2.3224325232388847,
      "grad_norm": 1.2785983085632324,
      "learning_rate": 9.435394736842105e-05,
      "loss": 0.3829,
      "step": 41474
    },
    {
      "epoch": 2.3224885205510137,
      "grad_norm": 1.138791799545288,
      "learning_rate": 9.435368421052631e-05,
      "loss": 0.3686,
      "step": 41475
    },
    {
      "epoch": 2.3225445178631428,
      "grad_norm": 1.3743484020233154,
      "learning_rate": 9.435342105263159e-05,
      "loss": 0.4304,
      "step": 41476
    },
    {
      "epoch": 2.3226005151752718,
      "grad_norm": 1.4038283824920654,
      "learning_rate": 9.435315789473685e-05,
      "loss": 0.4705,
      "step": 41477
    },
    {
      "epoch": 2.322656512487401,
      "grad_norm": 1.0943456888198853,
      "learning_rate": 9.435289473684212e-05,
      "loss": 0.3759,
      "step": 41478
    },
    {
      "epoch": 2.32271250979953,
      "grad_norm": 0.9064270853996277,
      "learning_rate": 9.435263157894737e-05,
      "loss": 0.446,
      "step": 41479
    },
    {
      "epoch": 2.322768507111659,
      "grad_norm": 1.477450966835022,
      "learning_rate": 9.435236842105264e-05,
      "loss": 0.4595,
      "step": 41480
    },
    {
      "epoch": 2.322824504423788,
      "grad_norm": 1.3120715618133545,
      "learning_rate": 9.43521052631579e-05,
      "loss": 0.3375,
      "step": 41481
    },
    {
      "epoch": 2.322880501735917,
      "grad_norm": 1.0751261711120605,
      "learning_rate": 9.435184210526317e-05,
      "loss": 0.4121,
      "step": 41482
    },
    {
      "epoch": 2.322936499048046,
      "grad_norm": 1.099856972694397,
      "learning_rate": 9.435157894736842e-05,
      "loss": 0.3891,
      "step": 41483
    },
    {
      "epoch": 2.3229924963601745,
      "grad_norm": 1.1082332134246826,
      "learning_rate": 9.435131578947369e-05,
      "loss": 0.4049,
      "step": 41484
    },
    {
      "epoch": 2.323048493672304,
      "grad_norm": 1.136395812034607,
      "learning_rate": 9.435105263157895e-05,
      "loss": 0.3487,
      "step": 41485
    },
    {
      "epoch": 2.3231044909844325,
      "grad_norm": 1.3666706085205078,
      "learning_rate": 9.435078947368422e-05,
      "loss": 0.4336,
      "step": 41486
    },
    {
      "epoch": 2.323160488296562,
      "grad_norm": 1.1374821662902832,
      "learning_rate": 9.435052631578948e-05,
      "loss": 0.355,
      "step": 41487
    },
    {
      "epoch": 2.3232164856086905,
      "grad_norm": 1.497208833694458,
      "learning_rate": 9.435026315789473e-05,
      "loss": 0.5575,
      "step": 41488
    },
    {
      "epoch": 2.32327248292082,
      "grad_norm": 1.1022244691848755,
      "learning_rate": 9.435e-05,
      "loss": 0.4633,
      "step": 41489
    },
    {
      "epoch": 2.3233284802329486,
      "grad_norm": 1.2131279706954956,
      "learning_rate": 9.434973684210526e-05,
      "loss": 0.3468,
      "step": 41490
    },
    {
      "epoch": 2.323384477545078,
      "grad_norm": 1.1669669151306152,
      "learning_rate": 9.434947368421054e-05,
      "loss": 0.4671,
      "step": 41491
    },
    {
      "epoch": 2.3234404748572066,
      "grad_norm": 5.152338027954102,
      "learning_rate": 9.43492105263158e-05,
      "loss": 0.3361,
      "step": 41492
    },
    {
      "epoch": 2.323496472169336,
      "grad_norm": 1.3370871543884277,
      "learning_rate": 9.434894736842106e-05,
      "loss": 0.4213,
      "step": 41493
    },
    {
      "epoch": 2.3235524694814647,
      "grad_norm": 1.6586989164352417,
      "learning_rate": 9.434868421052632e-05,
      "loss": 0.6026,
      "step": 41494
    },
    {
      "epoch": 2.323608466793594,
      "grad_norm": 1.40994393825531,
      "learning_rate": 9.434842105263159e-05,
      "loss": 0.5499,
      "step": 41495
    },
    {
      "epoch": 2.3236644641057227,
      "grad_norm": 1.0669114589691162,
      "learning_rate": 9.434815789473685e-05,
      "loss": 0.4114,
      "step": 41496
    },
    {
      "epoch": 2.323720461417852,
      "grad_norm": 1.7415821552276611,
      "learning_rate": 9.434789473684211e-05,
      "loss": 0.4394,
      "step": 41497
    },
    {
      "epoch": 2.3237764587299807,
      "grad_norm": 1.217822551727295,
      "learning_rate": 9.434763157894737e-05,
      "loss": 0.4051,
      "step": 41498
    },
    {
      "epoch": 2.32383245604211,
      "grad_norm": 1.5085703134536743,
      "learning_rate": 9.434736842105264e-05,
      "loss": 0.5867,
      "step": 41499
    },
    {
      "epoch": 2.3238884533542388,
      "grad_norm": 1.2492339611053467,
      "learning_rate": 9.43471052631579e-05,
      "loss": 0.4121,
      "step": 41500
    },
    {
      "epoch": 2.3239444506663682,
      "grad_norm": 1.3328871726989746,
      "learning_rate": 9.434684210526316e-05,
      "loss": 0.5951,
      "step": 41501
    },
    {
      "epoch": 2.324000447978497,
      "grad_norm": 1.1705149412155151,
      "learning_rate": 9.434657894736842e-05,
      "loss": 0.3892,
      "step": 41502
    },
    {
      "epoch": 2.3240564452906263,
      "grad_norm": 1.110447883605957,
      "learning_rate": 9.43463157894737e-05,
      "loss": 0.3708,
      "step": 41503
    },
    {
      "epoch": 2.324112442602755,
      "grad_norm": 1.2638144493103027,
      "learning_rate": 9.434605263157895e-05,
      "loss": 0.3803,
      "step": 41504
    },
    {
      "epoch": 2.3241684399148843,
      "grad_norm": 1.018725872039795,
      "learning_rate": 9.434578947368421e-05,
      "loss": 0.2935,
      "step": 41505
    },
    {
      "epoch": 2.324224437227013,
      "grad_norm": 1.7340035438537598,
      "learning_rate": 9.434552631578947e-05,
      "loss": 0.435,
      "step": 41506
    },
    {
      "epoch": 2.3242804345391423,
      "grad_norm": 1.3757978677749634,
      "learning_rate": 9.434526315789473e-05,
      "loss": 0.3157,
      "step": 41507
    },
    {
      "epoch": 2.324336431851271,
      "grad_norm": 1.095306634902954,
      "learning_rate": 9.4345e-05,
      "loss": 0.4941,
      "step": 41508
    },
    {
      "epoch": 2.3243924291634004,
      "grad_norm": 1.3000998497009277,
      "learning_rate": 9.434473684210527e-05,
      "loss": 0.3635,
      "step": 41509
    },
    {
      "epoch": 2.324448426475529,
      "grad_norm": 1.2867426872253418,
      "learning_rate": 9.434447368421054e-05,
      "loss": 0.337,
      "step": 41510
    },
    {
      "epoch": 2.3245044237876584,
      "grad_norm": 1.1369102001190186,
      "learning_rate": 9.434421052631578e-05,
      "loss": 0.4177,
      "step": 41511
    },
    {
      "epoch": 2.324560421099787,
      "grad_norm": 1.1665189266204834,
      "learning_rate": 9.434394736842106e-05,
      "loss": 0.4357,
      "step": 41512
    },
    {
      "epoch": 2.3246164184119165,
      "grad_norm": 1.141129970550537,
      "learning_rate": 9.434368421052632e-05,
      "loss": 0.3353,
      "step": 41513
    },
    {
      "epoch": 2.324672415724045,
      "grad_norm": 1.2560420036315918,
      "learning_rate": 9.434342105263159e-05,
      "loss": 0.4167,
      "step": 41514
    },
    {
      "epoch": 2.3247284130361745,
      "grad_norm": 1.153064250946045,
      "learning_rate": 9.434315789473685e-05,
      "loss": 0.326,
      "step": 41515
    },
    {
      "epoch": 2.324784410348303,
      "grad_norm": 1.0928597450256348,
      "learning_rate": 9.434289473684211e-05,
      "loss": 0.3567,
      "step": 41516
    },
    {
      "epoch": 2.3248404076604325,
      "grad_norm": 1.3983395099639893,
      "learning_rate": 9.434263157894737e-05,
      "loss": 0.3523,
      "step": 41517
    },
    {
      "epoch": 2.324896404972561,
      "grad_norm": 1.4594054222106934,
      "learning_rate": 9.434236842105264e-05,
      "loss": 0.4962,
      "step": 41518
    },
    {
      "epoch": 2.3249524022846906,
      "grad_norm": 1.3389967679977417,
      "learning_rate": 9.43421052631579e-05,
      "loss": 0.3855,
      "step": 41519
    },
    {
      "epoch": 2.325008399596819,
      "grad_norm": 1.1690731048583984,
      "learning_rate": 9.434184210526316e-05,
      "loss": 0.4012,
      "step": 41520
    },
    {
      "epoch": 2.3250643969089486,
      "grad_norm": 1.3864506483078003,
      "learning_rate": 9.434157894736842e-05,
      "loss": 0.5261,
      "step": 41521
    },
    {
      "epoch": 2.325120394221077,
      "grad_norm": 1.56587553024292,
      "learning_rate": 9.434131578947368e-05,
      "loss": 0.4393,
      "step": 41522
    },
    {
      "epoch": 2.3251763915332067,
      "grad_norm": 1.2505574226379395,
      "learning_rate": 9.434105263157896e-05,
      "loss": 0.4088,
      "step": 41523
    },
    {
      "epoch": 2.3252323888453352,
      "grad_norm": 1.073076605796814,
      "learning_rate": 9.434078947368422e-05,
      "loss": 0.38,
      "step": 41524
    },
    {
      "epoch": 2.3252883861574642,
      "grad_norm": 0.9981218576431274,
      "learning_rate": 9.434052631578948e-05,
      "loss": 0.3153,
      "step": 41525
    },
    {
      "epoch": 2.3253443834695933,
      "grad_norm": 1.4683637619018555,
      "learning_rate": 9.434026315789473e-05,
      "loss": 0.3522,
      "step": 41526
    },
    {
      "epoch": 2.3254003807817223,
      "grad_norm": 1.2398388385772705,
      "learning_rate": 9.434000000000001e-05,
      "loss": 0.4713,
      "step": 41527
    },
    {
      "epoch": 2.3254563780938513,
      "grad_norm": 1.4454247951507568,
      "learning_rate": 9.433973684210527e-05,
      "loss": 0.4499,
      "step": 41528
    },
    {
      "epoch": 2.3255123754059803,
      "grad_norm": 1.2900307178497314,
      "learning_rate": 9.433947368421053e-05,
      "loss": 0.3776,
      "step": 41529
    },
    {
      "epoch": 2.3255683727181093,
      "grad_norm": 1.1652144193649292,
      "learning_rate": 9.433921052631579e-05,
      "loss": 0.3361,
      "step": 41530
    },
    {
      "epoch": 2.3256243700302384,
      "grad_norm": 1.1878324747085571,
      "learning_rate": 9.433894736842106e-05,
      "loss": 0.4921,
      "step": 41531
    },
    {
      "epoch": 2.3256803673423674,
      "grad_norm": 1.1960644721984863,
      "learning_rate": 9.433868421052632e-05,
      "loss": 0.3676,
      "step": 41532
    },
    {
      "epoch": 2.3257363646544964,
      "grad_norm": 1.0765178203582764,
      "learning_rate": 9.43384210526316e-05,
      "loss": 0.3853,
      "step": 41533
    },
    {
      "epoch": 2.3257923619666254,
      "grad_norm": 1.0752185583114624,
      "learning_rate": 9.433815789473684e-05,
      "loss": 0.4312,
      "step": 41534
    },
    {
      "epoch": 2.3258483592787544,
      "grad_norm": 1.1072521209716797,
      "learning_rate": 9.433789473684211e-05,
      "loss": 0.3946,
      "step": 41535
    },
    {
      "epoch": 2.3259043565908835,
      "grad_norm": 1.0504088401794434,
      "learning_rate": 9.433763157894737e-05,
      "loss": 0.3921,
      "step": 41536
    },
    {
      "epoch": 2.3259603539030125,
      "grad_norm": 1.1277610063552856,
      "learning_rate": 9.433736842105265e-05,
      "loss": 0.3754,
      "step": 41537
    },
    {
      "epoch": 2.3260163512151415,
      "grad_norm": 1.1618610620498657,
      "learning_rate": 9.433710526315789e-05,
      "loss": 0.3718,
      "step": 41538
    },
    {
      "epoch": 2.3260723485272705,
      "grad_norm": 1.2163972854614258,
      "learning_rate": 9.433684210526317e-05,
      "loss": 0.4246,
      "step": 41539
    },
    {
      "epoch": 2.3261283458393995,
      "grad_norm": 1.088391900062561,
      "learning_rate": 9.433657894736843e-05,
      "loss": 0.3969,
      "step": 41540
    },
    {
      "epoch": 2.3261843431515286,
      "grad_norm": 0.9387152194976807,
      "learning_rate": 9.433631578947369e-05,
      "loss": 0.3819,
      "step": 41541
    },
    {
      "epoch": 2.3262403404636576,
      "grad_norm": 0.9201839566230774,
      "learning_rate": 9.433605263157896e-05,
      "loss": 0.2971,
      "step": 41542
    },
    {
      "epoch": 2.3262963377757866,
      "grad_norm": 1.156425952911377,
      "learning_rate": 9.43357894736842e-05,
      "loss": 0.4129,
      "step": 41543
    },
    {
      "epoch": 2.3263523350879156,
      "grad_norm": 1.094070553779602,
      "learning_rate": 9.433552631578948e-05,
      "loss": 0.3659,
      "step": 41544
    },
    {
      "epoch": 2.3264083324000446,
      "grad_norm": 1.2249170541763306,
      "learning_rate": 9.433526315789474e-05,
      "loss": 0.4709,
      "step": 41545
    },
    {
      "epoch": 2.3264643297121737,
      "grad_norm": 0.9159320592880249,
      "learning_rate": 9.433500000000001e-05,
      "loss": 0.2835,
      "step": 41546
    },
    {
      "epoch": 2.3265203270243027,
      "grad_norm": 1.1559115648269653,
      "learning_rate": 9.433473684210527e-05,
      "loss": 0.3011,
      "step": 41547
    },
    {
      "epoch": 2.3265763243364317,
      "grad_norm": 1.3106906414031982,
      "learning_rate": 9.433447368421053e-05,
      "loss": 0.3989,
      "step": 41548
    },
    {
      "epoch": 2.3266323216485607,
      "grad_norm": 1.0658910274505615,
      "learning_rate": 9.433421052631579e-05,
      "loss": 0.3484,
      "step": 41549
    },
    {
      "epoch": 2.3266883189606897,
      "grad_norm": 1.2402797937393188,
      "learning_rate": 9.433394736842106e-05,
      "loss": 0.3697,
      "step": 41550
    },
    {
      "epoch": 2.3267443162728187,
      "grad_norm": 1.0892387628555298,
      "learning_rate": 9.433368421052632e-05,
      "loss": 0.3676,
      "step": 41551
    },
    {
      "epoch": 2.3268003135849478,
      "grad_norm": 1.3662179708480835,
      "learning_rate": 9.433342105263158e-05,
      "loss": 0.3938,
      "step": 41552
    },
    {
      "epoch": 2.326856310897077,
      "grad_norm": 3.620800256729126,
      "learning_rate": 9.433315789473684e-05,
      "loss": 0.2778,
      "step": 41553
    },
    {
      "epoch": 2.326912308209206,
      "grad_norm": 1.1394436359405518,
      "learning_rate": 9.433289473684212e-05,
      "loss": 0.3365,
      "step": 41554
    },
    {
      "epoch": 2.326968305521335,
      "grad_norm": 1.2460987567901611,
      "learning_rate": 9.433263157894738e-05,
      "loss": 0.3224,
      "step": 41555
    },
    {
      "epoch": 2.327024302833464,
      "grad_norm": 1.1246585845947266,
      "learning_rate": 9.433236842105264e-05,
      "loss": 0.5338,
      "step": 41556
    },
    {
      "epoch": 2.327080300145593,
      "grad_norm": 1.3045283555984497,
      "learning_rate": 9.43321052631579e-05,
      "loss": 0.4315,
      "step": 41557
    },
    {
      "epoch": 2.327136297457722,
      "grad_norm": 1.094956874847412,
      "learning_rate": 9.433184210526315e-05,
      "loss": 0.4847,
      "step": 41558
    },
    {
      "epoch": 2.327192294769851,
      "grad_norm": 1.4461864233016968,
      "learning_rate": 9.433157894736843e-05,
      "loss": 0.4119,
      "step": 41559
    },
    {
      "epoch": 2.32724829208198,
      "grad_norm": 1.1447547674179077,
      "learning_rate": 9.433131578947369e-05,
      "loss": 0.3574,
      "step": 41560
    },
    {
      "epoch": 2.327304289394109,
      "grad_norm": 1.2475696802139282,
      "learning_rate": 9.433105263157895e-05,
      "loss": 0.4133,
      "step": 41561
    },
    {
      "epoch": 2.327360286706238,
      "grad_norm": 1.1497441530227661,
      "learning_rate": 9.433078947368421e-05,
      "loss": 0.433,
      "step": 41562
    },
    {
      "epoch": 2.327416284018367,
      "grad_norm": 1.386181354522705,
      "learning_rate": 9.433052631578948e-05,
      "loss": 0.3228,
      "step": 41563
    },
    {
      "epoch": 2.327472281330496,
      "grad_norm": 1.1684449911117554,
      "learning_rate": 9.433026315789474e-05,
      "loss": 0.3917,
      "step": 41564
    },
    {
      "epoch": 2.327528278642625,
      "grad_norm": 1.3723171949386597,
      "learning_rate": 9.433000000000001e-05,
      "loss": 0.4866,
      "step": 41565
    },
    {
      "epoch": 2.327584275954754,
      "grad_norm": 0.9769529700279236,
      "learning_rate": 9.432973684210526e-05,
      "loss": 0.3378,
      "step": 41566
    },
    {
      "epoch": 2.327640273266883,
      "grad_norm": 1.1759470701217651,
      "learning_rate": 9.432947368421053e-05,
      "loss": 0.3337,
      "step": 41567
    },
    {
      "epoch": 2.327696270579012,
      "grad_norm": 1.3004335165023804,
      "learning_rate": 9.432921052631579e-05,
      "loss": 0.4525,
      "step": 41568
    },
    {
      "epoch": 2.327752267891141,
      "grad_norm": 1.381335735321045,
      "learning_rate": 9.432894736842107e-05,
      "loss": 0.4564,
      "step": 41569
    },
    {
      "epoch": 2.32780826520327,
      "grad_norm": 1.2472684383392334,
      "learning_rate": 9.432868421052633e-05,
      "loss": 0.3269,
      "step": 41570
    },
    {
      "epoch": 2.327864262515399,
      "grad_norm": 1.7341516017913818,
      "learning_rate": 9.432842105263159e-05,
      "loss": 0.6316,
      "step": 41571
    },
    {
      "epoch": 2.327920259827528,
      "grad_norm": 0.9813318252563477,
      "learning_rate": 9.432815789473684e-05,
      "loss": 0.2936,
      "step": 41572
    },
    {
      "epoch": 2.327976257139657,
      "grad_norm": 1.0791974067687988,
      "learning_rate": 9.432789473684212e-05,
      "loss": 0.359,
      "step": 41573
    },
    {
      "epoch": 2.328032254451786,
      "grad_norm": 1.2258621454238892,
      "learning_rate": 9.432763157894738e-05,
      "loss": 0.4433,
      "step": 41574
    },
    {
      "epoch": 2.328088251763915,
      "grad_norm": 1.5598926544189453,
      "learning_rate": 9.432736842105262e-05,
      "loss": 0.5025,
      "step": 41575
    },
    {
      "epoch": 2.3281442490760442,
      "grad_norm": 1.1922926902770996,
      "learning_rate": 9.43271052631579e-05,
      "loss": 0.4191,
      "step": 41576
    },
    {
      "epoch": 2.3282002463881732,
      "grad_norm": 1.0238518714904785,
      "learning_rate": 9.432684210526316e-05,
      "loss": 0.3593,
      "step": 41577
    },
    {
      "epoch": 2.3282562437003023,
      "grad_norm": 1.4590539932250977,
      "learning_rate": 9.432657894736843e-05,
      "loss": 0.3404,
      "step": 41578
    },
    {
      "epoch": 2.3283122410124313,
      "grad_norm": 1.292555332183838,
      "learning_rate": 9.432631578947369e-05,
      "loss": 0.3438,
      "step": 41579
    },
    {
      "epoch": 2.3283682383245603,
      "grad_norm": 1.245064616203308,
      "learning_rate": 9.432605263157895e-05,
      "loss": 0.3831,
      "step": 41580
    },
    {
      "epoch": 2.3284242356366893,
      "grad_norm": 1.2050551176071167,
      "learning_rate": 9.432578947368421e-05,
      "loss": 0.4587,
      "step": 41581
    },
    {
      "epoch": 2.3284802329488183,
      "grad_norm": 1.0992491245269775,
      "learning_rate": 9.432552631578948e-05,
      "loss": 0.3585,
      "step": 41582
    },
    {
      "epoch": 2.3285362302609474,
      "grad_norm": 1.3423775434494019,
      "learning_rate": 9.432526315789474e-05,
      "loss": 0.5745,
      "step": 41583
    },
    {
      "epoch": 2.3285922275730764,
      "grad_norm": 1.1654114723205566,
      "learning_rate": 9.4325e-05,
      "loss": 0.3859,
      "step": 41584
    },
    {
      "epoch": 2.3286482248852054,
      "grad_norm": 1.1433289051055908,
      "learning_rate": 9.432473684210526e-05,
      "loss": 0.3157,
      "step": 41585
    },
    {
      "epoch": 2.3287042221973344,
      "grad_norm": 1.4299864768981934,
      "learning_rate": 9.432447368421054e-05,
      "loss": 0.4653,
      "step": 41586
    },
    {
      "epoch": 2.3287602195094634,
      "grad_norm": 3.0687637329101562,
      "learning_rate": 9.43242105263158e-05,
      "loss": 0.3564,
      "step": 41587
    },
    {
      "epoch": 2.3288162168215925,
      "grad_norm": 1.122779130935669,
      "learning_rate": 9.432394736842107e-05,
      "loss": 0.409,
      "step": 41588
    },
    {
      "epoch": 2.3288722141337215,
      "grad_norm": 2.0268421173095703,
      "learning_rate": 9.432368421052631e-05,
      "loss": 0.4409,
      "step": 41589
    },
    {
      "epoch": 2.3289282114458505,
      "grad_norm": 0.9438332915306091,
      "learning_rate": 9.432342105263159e-05,
      "loss": 0.2831,
      "step": 41590
    },
    {
      "epoch": 2.3289842087579795,
      "grad_norm": 1.3705345392227173,
      "learning_rate": 9.432315789473685e-05,
      "loss": 0.4815,
      "step": 41591
    },
    {
      "epoch": 2.3290402060701085,
      "grad_norm": 1.087262511253357,
      "learning_rate": 9.432289473684211e-05,
      "loss": 0.3322,
      "step": 41592
    },
    {
      "epoch": 2.3290962033822376,
      "grad_norm": 1.2847710847854614,
      "learning_rate": 9.432263157894737e-05,
      "loss": 0.3894,
      "step": 41593
    },
    {
      "epoch": 2.3291522006943666,
      "grad_norm": 1.3494991064071655,
      "learning_rate": 9.432236842105263e-05,
      "loss": 0.3591,
      "step": 41594
    },
    {
      "epoch": 2.3292081980064956,
      "grad_norm": 1.268125295639038,
      "learning_rate": 9.43221052631579e-05,
      "loss": 0.359,
      "step": 41595
    },
    {
      "epoch": 2.3292641953186246,
      "grad_norm": 1.2688404321670532,
      "learning_rate": 9.432184210526316e-05,
      "loss": 0.4193,
      "step": 41596
    },
    {
      "epoch": 2.3293201926307536,
      "grad_norm": 1.0668623447418213,
      "learning_rate": 9.432157894736843e-05,
      "loss": 0.3658,
      "step": 41597
    },
    {
      "epoch": 2.3293761899428826,
      "grad_norm": 1.1841307878494263,
      "learning_rate": 9.432131578947368e-05,
      "loss": 0.3154,
      "step": 41598
    },
    {
      "epoch": 2.3294321872550117,
      "grad_norm": 1.1780558824539185,
      "learning_rate": 9.432105263157895e-05,
      "loss": 0.4356,
      "step": 41599
    },
    {
      "epoch": 2.3294881845671407,
      "grad_norm": 2.430851936340332,
      "learning_rate": 9.432078947368421e-05,
      "loss": 0.5697,
      "step": 41600
    },
    {
      "epoch": 2.3295441818792697,
      "grad_norm": 1.1251972913742065,
      "learning_rate": 9.432052631578949e-05,
      "loss": 0.3537,
      "step": 41601
    },
    {
      "epoch": 2.3296001791913987,
      "grad_norm": 1.195039987564087,
      "learning_rate": 9.432026315789475e-05,
      "loss": 0.3496,
      "step": 41602
    },
    {
      "epoch": 2.3296561765035277,
      "grad_norm": 0.9922090172767639,
      "learning_rate": 9.432e-05,
      "loss": 0.3585,
      "step": 41603
    },
    {
      "epoch": 2.3297121738156568,
      "grad_norm": 1.1157501935958862,
      "learning_rate": 9.431973684210526e-05,
      "loss": 0.4565,
      "step": 41604
    },
    {
      "epoch": 2.329768171127786,
      "grad_norm": 1.3803671598434448,
      "learning_rate": 9.431947368421054e-05,
      "loss": 0.4088,
      "step": 41605
    },
    {
      "epoch": 2.329824168439915,
      "grad_norm": 1.2145215272903442,
      "learning_rate": 9.43192105263158e-05,
      "loss": 0.4503,
      "step": 41606
    },
    {
      "epoch": 2.329880165752044,
      "grad_norm": 1.2467933893203735,
      "learning_rate": 9.431894736842106e-05,
      "loss": 0.4943,
      "step": 41607
    },
    {
      "epoch": 2.329936163064173,
      "grad_norm": 1.352020502090454,
      "learning_rate": 9.431868421052632e-05,
      "loss": 0.4142,
      "step": 41608
    },
    {
      "epoch": 2.329992160376302,
      "grad_norm": 1.3673547506332397,
      "learning_rate": 9.431842105263158e-05,
      "loss": 0.5611,
      "step": 41609
    },
    {
      "epoch": 2.330048157688431,
      "grad_norm": 1.2323366403579712,
      "learning_rate": 9.431815789473685e-05,
      "loss": 0.698,
      "step": 41610
    },
    {
      "epoch": 2.33010415500056,
      "grad_norm": 1.1359715461730957,
      "learning_rate": 9.431789473684211e-05,
      "loss": 0.3438,
      "step": 41611
    },
    {
      "epoch": 2.330160152312689,
      "grad_norm": 1.3788206577301025,
      "learning_rate": 9.431763157894737e-05,
      "loss": 0.4516,
      "step": 41612
    },
    {
      "epoch": 2.330216149624818,
      "grad_norm": 1.0749812126159668,
      "learning_rate": 9.431736842105263e-05,
      "loss": 0.3207,
      "step": 41613
    },
    {
      "epoch": 2.330272146936947,
      "grad_norm": 1.2539621591567993,
      "learning_rate": 9.43171052631579e-05,
      "loss": 0.3878,
      "step": 41614
    },
    {
      "epoch": 2.330328144249076,
      "grad_norm": 1.1313960552215576,
      "learning_rate": 9.431684210526316e-05,
      "loss": 0.3392,
      "step": 41615
    },
    {
      "epoch": 2.330384141561205,
      "grad_norm": 1.1177096366882324,
      "learning_rate": 9.431657894736842e-05,
      "loss": 0.3661,
      "step": 41616
    },
    {
      "epoch": 2.330440138873334,
      "grad_norm": 1.251929521560669,
      "learning_rate": 9.431631578947368e-05,
      "loss": 0.3735,
      "step": 41617
    },
    {
      "epoch": 2.330496136185463,
      "grad_norm": 1.1779056787490845,
      "learning_rate": 9.431605263157896e-05,
      "loss": 0.4895,
      "step": 41618
    },
    {
      "epoch": 2.330552133497592,
      "grad_norm": 1.118046760559082,
      "learning_rate": 9.431578947368421e-05,
      "loss": 0.437,
      "step": 41619
    },
    {
      "epoch": 2.330608130809721,
      "grad_norm": 1.448420524597168,
      "learning_rate": 9.431552631578949e-05,
      "loss": 0.5376,
      "step": 41620
    },
    {
      "epoch": 2.33066412812185,
      "grad_norm": 1.3400373458862305,
      "learning_rate": 9.431526315789473e-05,
      "loss": 0.3729,
      "step": 41621
    },
    {
      "epoch": 2.330720125433979,
      "grad_norm": 1.1248011589050293,
      "learning_rate": 9.431500000000001e-05,
      "loss": 0.3674,
      "step": 41622
    },
    {
      "epoch": 2.330776122746108,
      "grad_norm": 1.1546565294265747,
      "learning_rate": 9.431473684210527e-05,
      "loss": 0.4526,
      "step": 41623
    },
    {
      "epoch": 2.330832120058237,
      "grad_norm": 1.0657941102981567,
      "learning_rate": 9.431447368421054e-05,
      "loss": 0.3928,
      "step": 41624
    },
    {
      "epoch": 2.330888117370366,
      "grad_norm": 1.1658271551132202,
      "learning_rate": 9.43142105263158e-05,
      "loss": 0.4677,
      "step": 41625
    },
    {
      "epoch": 2.330944114682495,
      "grad_norm": 1.356210708618164,
      "learning_rate": 9.431394736842105e-05,
      "loss": 0.4965,
      "step": 41626
    },
    {
      "epoch": 2.331000111994624,
      "grad_norm": 1.2530122995376587,
      "learning_rate": 9.431368421052632e-05,
      "loss": 0.4313,
      "step": 41627
    },
    {
      "epoch": 2.331056109306753,
      "grad_norm": 0.990557849407196,
      "learning_rate": 9.431342105263158e-05,
      "loss": 0.2755,
      "step": 41628
    },
    {
      "epoch": 2.3311121066188822,
      "grad_norm": 1.1237517595291138,
      "learning_rate": 9.431315789473685e-05,
      "loss": 0.3367,
      "step": 41629
    },
    {
      "epoch": 2.3311681039310113,
      "grad_norm": 1.4609808921813965,
      "learning_rate": 9.43128947368421e-05,
      "loss": 0.4101,
      "step": 41630
    },
    {
      "epoch": 2.3312241012431403,
      "grad_norm": 1.1659746170043945,
      "learning_rate": 9.431263157894737e-05,
      "loss": 0.4217,
      "step": 41631
    },
    {
      "epoch": 2.3312800985552693,
      "grad_norm": 0.9995316863059998,
      "learning_rate": 9.431236842105263e-05,
      "loss": 0.3789,
      "step": 41632
    },
    {
      "epoch": 2.3313360958673983,
      "grad_norm": 1.2200870513916016,
      "learning_rate": 9.43121052631579e-05,
      "loss": 0.3935,
      "step": 41633
    },
    {
      "epoch": 2.3313920931795273,
      "grad_norm": 1.1122318506240845,
      "learning_rate": 9.431184210526316e-05,
      "loss": 0.4057,
      "step": 41634
    },
    {
      "epoch": 2.3314480904916564,
      "grad_norm": 1.36678147315979,
      "learning_rate": 9.431157894736842e-05,
      "loss": 0.3027,
      "step": 41635
    },
    {
      "epoch": 2.3315040878037854,
      "grad_norm": 1.1468138694763184,
      "learning_rate": 9.431131578947368e-05,
      "loss": 0.3935,
      "step": 41636
    },
    {
      "epoch": 2.3315600851159144,
      "grad_norm": 1.0361790657043457,
      "learning_rate": 9.431105263157896e-05,
      "loss": 0.318,
      "step": 41637
    },
    {
      "epoch": 2.3316160824280434,
      "grad_norm": 1.0989990234375,
      "learning_rate": 9.431078947368422e-05,
      "loss": 0.3625,
      "step": 41638
    },
    {
      "epoch": 2.3316720797401724,
      "grad_norm": 1.1724388599395752,
      "learning_rate": 9.431052631578948e-05,
      "loss": 0.3974,
      "step": 41639
    },
    {
      "epoch": 2.3317280770523015,
      "grad_norm": 1.3609899282455444,
      "learning_rate": 9.431026315789474e-05,
      "loss": 0.4279,
      "step": 41640
    },
    {
      "epoch": 2.3317840743644305,
      "grad_norm": 1.1316336393356323,
      "learning_rate": 9.431000000000001e-05,
      "loss": 0.3428,
      "step": 41641
    },
    {
      "epoch": 2.3318400716765595,
      "grad_norm": 1.0280351638793945,
      "learning_rate": 9.430973684210527e-05,
      "loss": 0.3498,
      "step": 41642
    },
    {
      "epoch": 2.3318960689886885,
      "grad_norm": 1.697363257408142,
      "learning_rate": 9.430947368421053e-05,
      "loss": 0.3986,
      "step": 41643
    },
    {
      "epoch": 2.3319520663008175,
      "grad_norm": 1.0848455429077148,
      "learning_rate": 9.430921052631579e-05,
      "loss": 0.4206,
      "step": 41644
    },
    {
      "epoch": 2.3320080636129465,
      "grad_norm": 1.131593942642212,
      "learning_rate": 9.430894736842105e-05,
      "loss": 0.3853,
      "step": 41645
    },
    {
      "epoch": 2.3320640609250756,
      "grad_norm": 1.2424079179763794,
      "learning_rate": 9.430868421052632e-05,
      "loss": 0.5361,
      "step": 41646
    },
    {
      "epoch": 2.3321200582372046,
      "grad_norm": 1.182169795036316,
      "learning_rate": 9.430842105263158e-05,
      "loss": 0.5731,
      "step": 41647
    },
    {
      "epoch": 2.3321760555493336,
      "grad_norm": 1.1431505680084229,
      "learning_rate": 9.430815789473684e-05,
      "loss": 0.3826,
      "step": 41648
    },
    {
      "epoch": 2.3322320528614626,
      "grad_norm": 1.4568727016448975,
      "learning_rate": 9.43078947368421e-05,
      "loss": 0.4087,
      "step": 41649
    },
    {
      "epoch": 2.3322880501735916,
      "grad_norm": 1.3378961086273193,
      "learning_rate": 9.430763157894737e-05,
      "loss": 0.6345,
      "step": 41650
    },
    {
      "epoch": 2.3323440474857207,
      "grad_norm": 1.6190663576126099,
      "learning_rate": 9.430736842105263e-05,
      "loss": 0.4164,
      "step": 41651
    },
    {
      "epoch": 2.3324000447978497,
      "grad_norm": 1.3478869199752808,
      "learning_rate": 9.430710526315791e-05,
      "loss": 0.4849,
      "step": 41652
    },
    {
      "epoch": 2.3324560421099787,
      "grad_norm": 1.2431246042251587,
      "learning_rate": 9.430684210526315e-05,
      "loss": 0.4154,
      "step": 41653
    },
    {
      "epoch": 2.3325120394221077,
      "grad_norm": 1.2233268022537231,
      "learning_rate": 9.430657894736843e-05,
      "loss": 0.3562,
      "step": 41654
    },
    {
      "epoch": 2.3325680367342367,
      "grad_norm": 0.9987894296646118,
      "learning_rate": 9.430631578947369e-05,
      "loss": 0.338,
      "step": 41655
    },
    {
      "epoch": 2.3326240340463658,
      "grad_norm": 1.0406979322433472,
      "learning_rate": 9.430605263157896e-05,
      "loss": 0.36,
      "step": 41656
    },
    {
      "epoch": 2.3326800313584948,
      "grad_norm": 1.4256278276443481,
      "learning_rate": 9.430578947368422e-05,
      "loss": 0.4429,
      "step": 41657
    },
    {
      "epoch": 2.332736028670624,
      "grad_norm": 1.1722667217254639,
      "learning_rate": 9.430552631578948e-05,
      "loss": 0.3298,
      "step": 41658
    },
    {
      "epoch": 2.332792025982753,
      "grad_norm": 1.0652486085891724,
      "learning_rate": 9.430526315789474e-05,
      "loss": 0.3419,
      "step": 41659
    },
    {
      "epoch": 2.332848023294882,
      "grad_norm": 1.029771089553833,
      "learning_rate": 9.430500000000001e-05,
      "loss": 0.2534,
      "step": 41660
    },
    {
      "epoch": 2.332904020607011,
      "grad_norm": 1.1265232563018799,
      "learning_rate": 9.430473684210527e-05,
      "loss": 0.3277,
      "step": 41661
    },
    {
      "epoch": 2.33296001791914,
      "grad_norm": 1.3172229528427124,
      "learning_rate": 9.430447368421053e-05,
      "loss": 0.3412,
      "step": 41662
    },
    {
      "epoch": 2.333016015231269,
      "grad_norm": 1.2827367782592773,
      "learning_rate": 9.430421052631579e-05,
      "loss": 0.3825,
      "step": 41663
    },
    {
      "epoch": 2.333072012543398,
      "grad_norm": 1.4889280796051025,
      "learning_rate": 9.430394736842105e-05,
      "loss": 0.5843,
      "step": 41664
    },
    {
      "epoch": 2.333128009855527,
      "grad_norm": 1.1209124326705933,
      "learning_rate": 9.430368421052632e-05,
      "loss": 0.383,
      "step": 41665
    },
    {
      "epoch": 2.333184007167656,
      "grad_norm": 1.3554283380508423,
      "learning_rate": 9.430342105263158e-05,
      "loss": 0.4178,
      "step": 41666
    },
    {
      "epoch": 2.333240004479785,
      "grad_norm": 1.1092591285705566,
      "learning_rate": 9.430315789473684e-05,
      "loss": 0.3603,
      "step": 41667
    },
    {
      "epoch": 2.333296001791914,
      "grad_norm": 1.1824424266815186,
      "learning_rate": 9.43028947368421e-05,
      "loss": 0.4009,
      "step": 41668
    },
    {
      "epoch": 2.333351999104043,
      "grad_norm": 1.3623435497283936,
      "learning_rate": 9.430263157894738e-05,
      "loss": 0.3503,
      "step": 41669
    },
    {
      "epoch": 2.333407996416172,
      "grad_norm": 1.3782472610473633,
      "learning_rate": 9.430236842105264e-05,
      "loss": 0.4621,
      "step": 41670
    },
    {
      "epoch": 2.333463993728301,
      "grad_norm": 1.1193665266036987,
      "learning_rate": 9.43021052631579e-05,
      "loss": 0.4595,
      "step": 41671
    },
    {
      "epoch": 2.33351999104043,
      "grad_norm": 1.1852681636810303,
      "learning_rate": 9.430184210526316e-05,
      "loss": 0.3594,
      "step": 41672
    },
    {
      "epoch": 2.333575988352559,
      "grad_norm": 1.169629454612732,
      "learning_rate": 9.430157894736843e-05,
      "loss": 0.3724,
      "step": 41673
    },
    {
      "epoch": 2.333631985664688,
      "grad_norm": 0.9822991490364075,
      "learning_rate": 9.430131578947369e-05,
      "loss": 0.4527,
      "step": 41674
    },
    {
      "epoch": 2.333687982976817,
      "grad_norm": 1.3232277631759644,
      "learning_rate": 9.430105263157896e-05,
      "loss": 0.4737,
      "step": 41675
    },
    {
      "epoch": 2.333743980288946,
      "grad_norm": 1.0209532976150513,
      "learning_rate": 9.430078947368421e-05,
      "loss": 0.3734,
      "step": 41676
    },
    {
      "epoch": 2.333799977601075,
      "grad_norm": 1.3224495649337769,
      "learning_rate": 9.430052631578948e-05,
      "loss": 0.5143,
      "step": 41677
    },
    {
      "epoch": 2.333855974913204,
      "grad_norm": 1.3637382984161377,
      "learning_rate": 9.430026315789474e-05,
      "loss": 0.5713,
      "step": 41678
    },
    {
      "epoch": 2.333911972225333,
      "grad_norm": 1.3716785907745361,
      "learning_rate": 9.43e-05,
      "loss": 0.4359,
      "step": 41679
    },
    {
      "epoch": 2.333967969537462,
      "grad_norm": 1.008894920349121,
      "learning_rate": 9.429973684210528e-05,
      "loss": 0.3966,
      "step": 41680
    },
    {
      "epoch": 2.3340239668495912,
      "grad_norm": 1.3727296590805054,
      "learning_rate": 9.429947368421052e-05,
      "loss": 0.498,
      "step": 41681
    },
    {
      "epoch": 2.3340799641617203,
      "grad_norm": 1.355534315109253,
      "learning_rate": 9.42992105263158e-05,
      "loss": 0.3918,
      "step": 41682
    },
    {
      "epoch": 2.3341359614738493,
      "grad_norm": 1.2827754020690918,
      "learning_rate": 9.429894736842105e-05,
      "loss": 0.3398,
      "step": 41683
    },
    {
      "epoch": 2.3341919587859783,
      "grad_norm": 1.1970857381820679,
      "learning_rate": 9.429868421052633e-05,
      "loss": 0.4814,
      "step": 41684
    },
    {
      "epoch": 2.3342479560981073,
      "grad_norm": 1.4424879550933838,
      "learning_rate": 9.429842105263157e-05,
      "loss": 0.3725,
      "step": 41685
    },
    {
      "epoch": 2.3343039534102363,
      "grad_norm": 1.116942048072815,
      "learning_rate": 9.429815789473685e-05,
      "loss": 0.4007,
      "step": 41686
    },
    {
      "epoch": 2.3343599507223654,
      "grad_norm": 1.3346049785614014,
      "learning_rate": 9.42978947368421e-05,
      "loss": 0.6027,
      "step": 41687
    },
    {
      "epoch": 2.3344159480344944,
      "grad_norm": 0.9552223086357117,
      "learning_rate": 9.429763157894738e-05,
      "loss": 0.3323,
      "step": 41688
    },
    {
      "epoch": 2.3344719453466234,
      "grad_norm": 1.2311464548110962,
      "learning_rate": 9.429736842105264e-05,
      "loss": 0.3994,
      "step": 41689
    },
    {
      "epoch": 2.3345279426587524,
      "grad_norm": 1.2025054693222046,
      "learning_rate": 9.42971052631579e-05,
      "loss": 0.469,
      "step": 41690
    },
    {
      "epoch": 2.3345839399708814,
      "grad_norm": 1.4589117765426636,
      "learning_rate": 9.429684210526316e-05,
      "loss": 0.4501,
      "step": 41691
    },
    {
      "epoch": 2.3346399372830104,
      "grad_norm": 1.066486120223999,
      "learning_rate": 9.429657894736843e-05,
      "loss": 0.299,
      "step": 41692
    },
    {
      "epoch": 2.3346959345951395,
      "grad_norm": 1.087945580482483,
      "learning_rate": 9.429631578947369e-05,
      "loss": 0.4147,
      "step": 41693
    },
    {
      "epoch": 2.3347519319072685,
      "grad_norm": 1.1757198572158813,
      "learning_rate": 9.429605263157895e-05,
      "loss": 0.3714,
      "step": 41694
    },
    {
      "epoch": 2.3348079292193975,
      "grad_norm": 1.5330173969268799,
      "learning_rate": 9.429578947368421e-05,
      "loss": 0.5794,
      "step": 41695
    },
    {
      "epoch": 2.3348639265315265,
      "grad_norm": 0.9312339425086975,
      "learning_rate": 9.429552631578947e-05,
      "loss": 0.3141,
      "step": 41696
    },
    {
      "epoch": 2.3349199238436555,
      "grad_norm": 1.1804759502410889,
      "learning_rate": 9.429526315789474e-05,
      "loss": 0.408,
      "step": 41697
    },
    {
      "epoch": 2.3349759211557846,
      "grad_norm": 1.4757715463638306,
      "learning_rate": 9.4295e-05,
      "loss": 0.4644,
      "step": 41698
    },
    {
      "epoch": 2.3350319184679136,
      "grad_norm": 1.1422994136810303,
      "learning_rate": 9.429473684210526e-05,
      "loss": 0.4288,
      "step": 41699
    },
    {
      "epoch": 2.3350879157800426,
      "grad_norm": 1.0504841804504395,
      "learning_rate": 9.429447368421052e-05,
      "loss": 0.3136,
      "step": 41700
    },
    {
      "epoch": 2.3351439130921716,
      "grad_norm": 1.280503749847412,
      "learning_rate": 9.42942105263158e-05,
      "loss": 0.3732,
      "step": 41701
    },
    {
      "epoch": 2.3351999104043006,
      "grad_norm": 1.050684928894043,
      "learning_rate": 9.429394736842106e-05,
      "loss": 0.433,
      "step": 41702
    },
    {
      "epoch": 2.3352559077164297,
      "grad_norm": 1.3107224702835083,
      "learning_rate": 9.429368421052632e-05,
      "loss": 0.4073,
      "step": 41703
    },
    {
      "epoch": 2.3353119050285587,
      "grad_norm": 1.3444058895111084,
      "learning_rate": 9.429342105263158e-05,
      "loss": 0.4727,
      "step": 41704
    },
    {
      "epoch": 2.3353679023406877,
      "grad_norm": 1.0133336782455444,
      "learning_rate": 9.429315789473685e-05,
      "loss": 0.3862,
      "step": 41705
    },
    {
      "epoch": 2.3354238996528167,
      "grad_norm": 1.2451754808425903,
      "learning_rate": 9.429289473684211e-05,
      "loss": 0.5589,
      "step": 41706
    },
    {
      "epoch": 2.3354798969649457,
      "grad_norm": 1.3266135454177856,
      "learning_rate": 9.429263157894738e-05,
      "loss": 0.4292,
      "step": 41707
    },
    {
      "epoch": 2.3355358942770748,
      "grad_norm": 1.4028217792510986,
      "learning_rate": 9.429236842105263e-05,
      "loss": 0.3688,
      "step": 41708
    },
    {
      "epoch": 2.3355918915892038,
      "grad_norm": 1.3867204189300537,
      "learning_rate": 9.42921052631579e-05,
      "loss": 0.3798,
      "step": 41709
    },
    {
      "epoch": 2.335647888901333,
      "grad_norm": 0.9747878313064575,
      "learning_rate": 9.429184210526316e-05,
      "loss": 0.3073,
      "step": 41710
    },
    {
      "epoch": 2.335703886213462,
      "grad_norm": 8.884810447692871,
      "learning_rate": 9.429157894736844e-05,
      "loss": 0.3939,
      "step": 41711
    },
    {
      "epoch": 2.335759883525591,
      "grad_norm": 1.3453997373580933,
      "learning_rate": 9.42913157894737e-05,
      "loss": 0.479,
      "step": 41712
    },
    {
      "epoch": 2.33581588083772,
      "grad_norm": 1.3915377855300903,
      "learning_rate": 9.429105263157894e-05,
      "loss": 0.4703,
      "step": 41713
    },
    {
      "epoch": 2.335871878149849,
      "grad_norm": 1.4205951690673828,
      "learning_rate": 9.429078947368421e-05,
      "loss": 0.416,
      "step": 41714
    },
    {
      "epoch": 2.335927875461978,
      "grad_norm": 1.204131007194519,
      "learning_rate": 9.429052631578947e-05,
      "loss": 0.4131,
      "step": 41715
    },
    {
      "epoch": 2.335983872774107,
      "grad_norm": 1.440616250038147,
      "learning_rate": 9.429026315789475e-05,
      "loss": 0.3045,
      "step": 41716
    },
    {
      "epoch": 2.336039870086236,
      "grad_norm": 1.2486579418182373,
      "learning_rate": 9.429000000000001e-05,
      "loss": 0.3971,
      "step": 41717
    },
    {
      "epoch": 2.336095867398365,
      "grad_norm": 1.1286476850509644,
      "learning_rate": 9.428973684210527e-05,
      "loss": 0.4534,
      "step": 41718
    },
    {
      "epoch": 2.336151864710494,
      "grad_norm": 1.3379883766174316,
      "learning_rate": 9.428947368421053e-05,
      "loss": 0.3618,
      "step": 41719
    },
    {
      "epoch": 2.336207862022623,
      "grad_norm": 1.2174293994903564,
      "learning_rate": 9.42892105263158e-05,
      "loss": 0.3415,
      "step": 41720
    },
    {
      "epoch": 2.336263859334752,
      "grad_norm": 1.2182146310806274,
      "learning_rate": 9.428894736842106e-05,
      "loss": 0.426,
      "step": 41721
    },
    {
      "epoch": 2.336319856646881,
      "grad_norm": 1.4156049489974976,
      "learning_rate": 9.428868421052632e-05,
      "loss": 0.416,
      "step": 41722
    },
    {
      "epoch": 2.33637585395901,
      "grad_norm": 1.4497874975204468,
      "learning_rate": 9.428842105263158e-05,
      "loss": 0.5433,
      "step": 41723
    },
    {
      "epoch": 2.336431851271139,
      "grad_norm": 1.6712723970413208,
      "learning_rate": 9.428815789473685e-05,
      "loss": 0.3794,
      "step": 41724
    },
    {
      "epoch": 2.336487848583268,
      "grad_norm": 1.330540657043457,
      "learning_rate": 9.428789473684211e-05,
      "loss": 0.353,
      "step": 41725
    },
    {
      "epoch": 2.336543845895397,
      "grad_norm": 1.1170587539672852,
      "learning_rate": 9.428763157894737e-05,
      "loss": 0.3398,
      "step": 41726
    },
    {
      "epoch": 2.336599843207526,
      "grad_norm": 1.41217839717865,
      "learning_rate": 9.428736842105263e-05,
      "loss": 0.4369,
      "step": 41727
    },
    {
      "epoch": 2.336655840519655,
      "grad_norm": 9.436219215393066,
      "learning_rate": 9.42871052631579e-05,
      "loss": 0.4083,
      "step": 41728
    },
    {
      "epoch": 2.336711837831784,
      "grad_norm": 1.4224368333816528,
      "learning_rate": 9.428684210526316e-05,
      "loss": 0.5299,
      "step": 41729
    },
    {
      "epoch": 2.336767835143913,
      "grad_norm": 1.157005786895752,
      "learning_rate": 9.428657894736842e-05,
      "loss": 0.3881,
      "step": 41730
    },
    {
      "epoch": 2.336823832456042,
      "grad_norm": 1.6119577884674072,
      "learning_rate": 9.428631578947368e-05,
      "loss": 0.4736,
      "step": 41731
    },
    {
      "epoch": 2.336879829768171,
      "grad_norm": 1.1501855850219727,
      "learning_rate": 9.428605263157894e-05,
      "loss": 0.3721,
      "step": 41732
    },
    {
      "epoch": 2.3369358270803002,
      "grad_norm": 1.53270423412323,
      "learning_rate": 9.428578947368422e-05,
      "loss": 0.4571,
      "step": 41733
    },
    {
      "epoch": 2.3369918243924293,
      "grad_norm": 1.7130926847457886,
      "learning_rate": 9.428552631578948e-05,
      "loss": 0.2826,
      "step": 41734
    },
    {
      "epoch": 2.3370478217045583,
      "grad_norm": 1.2406684160232544,
      "learning_rate": 9.428526315789475e-05,
      "loss": 0.5091,
      "step": 41735
    },
    {
      "epoch": 2.3371038190166873,
      "grad_norm": 1.0346183776855469,
      "learning_rate": 9.4285e-05,
      "loss": 0.2716,
      "step": 41736
    },
    {
      "epoch": 2.3371598163288163,
      "grad_norm": 1.2530306577682495,
      "learning_rate": 9.428473684210527e-05,
      "loss": 0.3613,
      "step": 41737
    },
    {
      "epoch": 2.3372158136409453,
      "grad_norm": 1.282768964767456,
      "learning_rate": 9.428447368421053e-05,
      "loss": 0.417,
      "step": 41738
    },
    {
      "epoch": 2.3372718109530743,
      "grad_norm": 1.467400074005127,
      "learning_rate": 9.42842105263158e-05,
      "loss": 0.4691,
      "step": 41739
    },
    {
      "epoch": 2.3373278082652034,
      "grad_norm": 1.1563502550125122,
      "learning_rate": 9.428394736842105e-05,
      "loss": 0.3668,
      "step": 41740
    },
    {
      "epoch": 2.3373838055773324,
      "grad_norm": 1.3010636568069458,
      "learning_rate": 9.428368421052632e-05,
      "loss": 0.4101,
      "step": 41741
    },
    {
      "epoch": 2.3374398028894614,
      "grad_norm": 1.5787529945373535,
      "learning_rate": 9.428342105263158e-05,
      "loss": 0.4249,
      "step": 41742
    },
    {
      "epoch": 2.3374958002015904,
      "grad_norm": 1.2126593589782715,
      "learning_rate": 9.428315789473685e-05,
      "loss": 0.3829,
      "step": 41743
    },
    {
      "epoch": 2.3375517975137194,
      "grad_norm": 1.2742772102355957,
      "learning_rate": 9.428289473684211e-05,
      "loss": 0.3764,
      "step": 41744
    },
    {
      "epoch": 2.3376077948258485,
      "grad_norm": 1.1307629346847534,
      "learning_rate": 9.428263157894737e-05,
      "loss": 0.3727,
      "step": 41745
    },
    {
      "epoch": 2.3376637921379775,
      "grad_norm": 1.0303871631622314,
      "learning_rate": 9.428236842105263e-05,
      "loss": 0.275,
      "step": 41746
    },
    {
      "epoch": 2.3377197894501065,
      "grad_norm": 1.8317618370056152,
      "learning_rate": 9.42821052631579e-05,
      "loss": 0.465,
      "step": 41747
    },
    {
      "epoch": 2.3377757867622355,
      "grad_norm": 1.093740701675415,
      "learning_rate": 9.428184210526317e-05,
      "loss": 0.3903,
      "step": 41748
    },
    {
      "epoch": 2.3378317840743645,
      "grad_norm": 1.3149365186691284,
      "learning_rate": 9.428157894736843e-05,
      "loss": 0.3396,
      "step": 41749
    },
    {
      "epoch": 2.3378877813864936,
      "grad_norm": 1.2754952907562256,
      "learning_rate": 9.428131578947369e-05,
      "loss": 0.3512,
      "step": 41750
    },
    {
      "epoch": 2.3379437786986226,
      "grad_norm": 1.2109102010726929,
      "learning_rate": 9.428105263157895e-05,
      "loss": 0.3263,
      "step": 41751
    },
    {
      "epoch": 2.3379997760107516,
      "grad_norm": 1.2914378643035889,
      "learning_rate": 9.428078947368422e-05,
      "loss": 0.503,
      "step": 41752
    },
    {
      "epoch": 2.3380557733228806,
      "grad_norm": 1.2562527656555176,
      "learning_rate": 9.428052631578948e-05,
      "loss": 0.4047,
      "step": 41753
    },
    {
      "epoch": 2.3381117706350096,
      "grad_norm": 1.27664315700531,
      "learning_rate": 9.428026315789474e-05,
      "loss": 0.5224,
      "step": 41754
    },
    {
      "epoch": 2.3381677679471387,
      "grad_norm": 1.3295894861221313,
      "learning_rate": 9.428e-05,
      "loss": 0.3922,
      "step": 41755
    },
    {
      "epoch": 2.3382237652592677,
      "grad_norm": 1.0919817686080933,
      "learning_rate": 9.427973684210527e-05,
      "loss": 0.3577,
      "step": 41756
    },
    {
      "epoch": 2.3382797625713967,
      "grad_norm": 1.1978970766067505,
      "learning_rate": 9.427947368421053e-05,
      "loss": 0.4259,
      "step": 41757
    },
    {
      "epoch": 2.3383357598835257,
      "grad_norm": 1.1360406875610352,
      "learning_rate": 9.427921052631579e-05,
      "loss": 0.3695,
      "step": 41758
    },
    {
      "epoch": 2.3383917571956547,
      "grad_norm": 1.6929552555084229,
      "learning_rate": 9.427894736842105e-05,
      "loss": 0.3602,
      "step": 41759
    },
    {
      "epoch": 2.3384477545077837,
      "grad_norm": 1.345205545425415,
      "learning_rate": 9.427868421052632e-05,
      "loss": 0.361,
      "step": 41760
    },
    {
      "epoch": 2.3385037518199128,
      "grad_norm": 1.705392837524414,
      "learning_rate": 9.427842105263158e-05,
      "loss": 0.5616,
      "step": 41761
    },
    {
      "epoch": 2.338559749132042,
      "grad_norm": 1.2249919176101685,
      "learning_rate": 9.427815789473686e-05,
      "loss": 0.3584,
      "step": 41762
    },
    {
      "epoch": 2.338615746444171,
      "grad_norm": 1.1815836429595947,
      "learning_rate": 9.42778947368421e-05,
      "loss": 0.337,
      "step": 41763
    },
    {
      "epoch": 2.3386717437563,
      "grad_norm": 1.1074751615524292,
      "learning_rate": 9.427763157894736e-05,
      "loss": 0.4065,
      "step": 41764
    },
    {
      "epoch": 2.338727741068429,
      "grad_norm": 1.1780734062194824,
      "learning_rate": 9.427736842105264e-05,
      "loss": 0.3084,
      "step": 41765
    },
    {
      "epoch": 2.338783738380558,
      "grad_norm": 1.3323748111724854,
      "learning_rate": 9.42771052631579e-05,
      "loss": 0.4491,
      "step": 41766
    },
    {
      "epoch": 2.338839735692687,
      "grad_norm": 1.1028201580047607,
      "learning_rate": 9.427684210526317e-05,
      "loss": 0.3188,
      "step": 41767
    },
    {
      "epoch": 2.338895733004816,
      "grad_norm": 1.160706877708435,
      "learning_rate": 9.427657894736842e-05,
      "loss": 0.3737,
      "step": 41768
    },
    {
      "epoch": 2.338951730316945,
      "grad_norm": 1.1706403493881226,
      "learning_rate": 9.427631578947369e-05,
      "loss": 0.5354,
      "step": 41769
    },
    {
      "epoch": 2.339007727629074,
      "grad_norm": 1.269150733947754,
      "learning_rate": 9.427605263157895e-05,
      "loss": 0.3022,
      "step": 41770
    },
    {
      "epoch": 2.339063724941203,
      "grad_norm": 1.3719485998153687,
      "learning_rate": 9.427578947368422e-05,
      "loss": 0.4369,
      "step": 41771
    },
    {
      "epoch": 2.339119722253332,
      "grad_norm": 1.270066738128662,
      "learning_rate": 9.427552631578948e-05,
      "loss": 0.3907,
      "step": 41772
    },
    {
      "epoch": 2.339175719565461,
      "grad_norm": 1.4558935165405273,
      "learning_rate": 9.427526315789474e-05,
      "loss": 0.5376,
      "step": 41773
    },
    {
      "epoch": 2.33923171687759,
      "grad_norm": 1.1652445793151855,
      "learning_rate": 9.4275e-05,
      "loss": 0.4387,
      "step": 41774
    },
    {
      "epoch": 2.339287714189719,
      "grad_norm": 1.228729248046875,
      "learning_rate": 9.427473684210527e-05,
      "loss": 0.4562,
      "step": 41775
    },
    {
      "epoch": 2.339343711501848,
      "grad_norm": 1.0535378456115723,
      "learning_rate": 9.427447368421053e-05,
      "loss": 0.5205,
      "step": 41776
    },
    {
      "epoch": 2.339399708813977,
      "grad_norm": 1.1124930381774902,
      "learning_rate": 9.42742105263158e-05,
      "loss": 0.3259,
      "step": 41777
    },
    {
      "epoch": 2.339455706126106,
      "grad_norm": 1.1334924697875977,
      "learning_rate": 9.427394736842105e-05,
      "loss": 0.3837,
      "step": 41778
    },
    {
      "epoch": 2.339511703438235,
      "grad_norm": 1.1416373252868652,
      "learning_rate": 9.427368421052633e-05,
      "loss": 0.3076,
      "step": 41779
    },
    {
      "epoch": 2.339567700750364,
      "grad_norm": 1.187648892402649,
      "learning_rate": 9.427342105263159e-05,
      "loss": 0.4436,
      "step": 41780
    },
    {
      "epoch": 2.339623698062493,
      "grad_norm": 0.938332736492157,
      "learning_rate": 9.427315789473685e-05,
      "loss": 0.2651,
      "step": 41781
    },
    {
      "epoch": 2.339679695374622,
      "grad_norm": 1.1859545707702637,
      "learning_rate": 9.42728947368421e-05,
      "loss": 0.4166,
      "step": 41782
    },
    {
      "epoch": 2.339735692686751,
      "grad_norm": 1.0829342603683472,
      "learning_rate": 9.427263157894737e-05,
      "loss": 0.3966,
      "step": 41783
    },
    {
      "epoch": 2.33979168999888,
      "grad_norm": 1.1341698169708252,
      "learning_rate": 9.427236842105264e-05,
      "loss": 0.3065,
      "step": 41784
    },
    {
      "epoch": 2.3398476873110092,
      "grad_norm": 1.224109411239624,
      "learning_rate": 9.42721052631579e-05,
      "loss": 0.4558,
      "step": 41785
    },
    {
      "epoch": 2.3399036846231382,
      "grad_norm": 1.0407896041870117,
      "learning_rate": 9.427184210526316e-05,
      "loss": 0.4261,
      "step": 41786
    },
    {
      "epoch": 2.3399596819352673,
      "grad_norm": 1.2565711736679077,
      "learning_rate": 9.427157894736842e-05,
      "loss": 0.4373,
      "step": 41787
    },
    {
      "epoch": 2.3400156792473963,
      "grad_norm": 1.186922311782837,
      "learning_rate": 9.427131578947369e-05,
      "loss": 0.3888,
      "step": 41788
    },
    {
      "epoch": 2.3400716765595253,
      "grad_norm": 1.1499872207641602,
      "learning_rate": 9.427105263157895e-05,
      "loss": 0.3495,
      "step": 41789
    },
    {
      "epoch": 2.3401276738716543,
      "grad_norm": 1.2514384984970093,
      "learning_rate": 9.427078947368422e-05,
      "loss": 0.4684,
      "step": 41790
    },
    {
      "epoch": 2.3401836711837833,
      "grad_norm": 1.0329967737197876,
      "learning_rate": 9.427052631578947e-05,
      "loss": 0.3001,
      "step": 41791
    },
    {
      "epoch": 2.3402396684959124,
      "grad_norm": 1.080004334449768,
      "learning_rate": 9.427026315789474e-05,
      "loss": 0.297,
      "step": 41792
    },
    {
      "epoch": 2.3402956658080414,
      "grad_norm": 1.3767447471618652,
      "learning_rate": 9.427e-05,
      "loss": 0.5396,
      "step": 41793
    },
    {
      "epoch": 2.3403516631201704,
      "grad_norm": 1.1343660354614258,
      "learning_rate": 9.426973684210528e-05,
      "loss": 0.3601,
      "step": 41794
    },
    {
      "epoch": 2.3404076604322994,
      "grad_norm": 1.0282236337661743,
      "learning_rate": 9.426947368421052e-05,
      "loss": 0.2492,
      "step": 41795
    },
    {
      "epoch": 2.3404636577444284,
      "grad_norm": 1.2647206783294678,
      "learning_rate": 9.42692105263158e-05,
      "loss": 0.3798,
      "step": 41796
    },
    {
      "epoch": 2.3405196550565575,
      "grad_norm": 1.4174648523330688,
      "learning_rate": 9.426894736842106e-05,
      "loss": 0.3754,
      "step": 41797
    },
    {
      "epoch": 2.3405756523686865,
      "grad_norm": 1.20793879032135,
      "learning_rate": 9.426868421052633e-05,
      "loss": 0.3814,
      "step": 41798
    },
    {
      "epoch": 2.3406316496808155,
      "grad_norm": 1.113721489906311,
      "learning_rate": 9.426842105263159e-05,
      "loss": 0.3805,
      "step": 41799
    },
    {
      "epoch": 2.3406876469929445,
      "grad_norm": 1.3276747465133667,
      "learning_rate": 9.426815789473684e-05,
      "loss": 0.4756,
      "step": 41800
    },
    {
      "epoch": 2.3407436443050735,
      "grad_norm": 1.2933099269866943,
      "learning_rate": 9.426789473684211e-05,
      "loss": 0.6164,
      "step": 41801
    },
    {
      "epoch": 2.3407996416172026,
      "grad_norm": 1.3743950128555298,
      "learning_rate": 9.426763157894737e-05,
      "loss": 0.3269,
      "step": 41802
    },
    {
      "epoch": 2.3408556389293316,
      "grad_norm": 1.3877071142196655,
      "learning_rate": 9.426736842105264e-05,
      "loss": 0.4525,
      "step": 41803
    },
    {
      "epoch": 2.3409116362414606,
      "grad_norm": 0.9714048504829407,
      "learning_rate": 9.42671052631579e-05,
      "loss": 0.312,
      "step": 41804
    },
    {
      "epoch": 2.3409676335535896,
      "grad_norm": 1.0771499872207642,
      "learning_rate": 9.426684210526316e-05,
      "loss": 0.3177,
      "step": 41805
    },
    {
      "epoch": 2.3410236308657186,
      "grad_norm": 1.1779916286468506,
      "learning_rate": 9.426657894736842e-05,
      "loss": 0.3441,
      "step": 41806
    },
    {
      "epoch": 2.3410796281778476,
      "grad_norm": 1.23214852809906,
      "learning_rate": 9.42663157894737e-05,
      "loss": 0.6082,
      "step": 41807
    },
    {
      "epoch": 2.3411356254899767,
      "grad_norm": 1.1305887699127197,
      "learning_rate": 9.426605263157895e-05,
      "loss": 0.3718,
      "step": 41808
    },
    {
      "epoch": 2.3411916228021057,
      "grad_norm": 0.9967213869094849,
      "learning_rate": 9.426578947368421e-05,
      "loss": 0.3536,
      "step": 41809
    },
    {
      "epoch": 2.3412476201142347,
      "grad_norm": 1.8401435613632202,
      "learning_rate": 9.426552631578947e-05,
      "loss": 0.5358,
      "step": 41810
    },
    {
      "epoch": 2.3413036174263637,
      "grad_norm": 1.336448311805725,
      "learning_rate": 9.426526315789475e-05,
      "loss": 0.3524,
      "step": 41811
    },
    {
      "epoch": 2.3413596147384927,
      "grad_norm": 1.1528698205947876,
      "learning_rate": 9.4265e-05,
      "loss": 0.3462,
      "step": 41812
    },
    {
      "epoch": 2.3414156120506218,
      "grad_norm": 1.3161871433258057,
      "learning_rate": 9.426473684210527e-05,
      "loss": 0.4334,
      "step": 41813
    },
    {
      "epoch": 2.341471609362751,
      "grad_norm": 1.3091539144515991,
      "learning_rate": 9.426447368421053e-05,
      "loss": 0.5229,
      "step": 41814
    },
    {
      "epoch": 2.3415276066748794,
      "grad_norm": 1.1306521892547607,
      "learning_rate": 9.42642105263158e-05,
      "loss": 0.4743,
      "step": 41815
    },
    {
      "epoch": 2.341583603987009,
      "grad_norm": 1.1712980270385742,
      "learning_rate": 9.426394736842106e-05,
      "loss": 0.2625,
      "step": 41816
    },
    {
      "epoch": 2.3416396012991374,
      "grad_norm": 1.3377574682235718,
      "learning_rate": 9.426368421052632e-05,
      "loss": 0.4132,
      "step": 41817
    },
    {
      "epoch": 2.341695598611267,
      "grad_norm": 1.3885612487792969,
      "learning_rate": 9.426342105263158e-05,
      "loss": 0.5286,
      "step": 41818
    },
    {
      "epoch": 2.3417515959233954,
      "grad_norm": 1.1134899854660034,
      "learning_rate": 9.426315789473684e-05,
      "loss": 0.3455,
      "step": 41819
    },
    {
      "epoch": 2.341807593235525,
      "grad_norm": 1.1279454231262207,
      "learning_rate": 9.426289473684211e-05,
      "loss": 0.391,
      "step": 41820
    },
    {
      "epoch": 2.3418635905476535,
      "grad_norm": 1.1154600381851196,
      "learning_rate": 9.426263157894737e-05,
      "loss": 0.396,
      "step": 41821
    },
    {
      "epoch": 2.341919587859783,
      "grad_norm": 1.1974982023239136,
      "learning_rate": 9.426236842105264e-05,
      "loss": 0.4784,
      "step": 41822
    },
    {
      "epoch": 2.3419755851719115,
      "grad_norm": 1.4913843870162964,
      "learning_rate": 9.426210526315789e-05,
      "loss": 0.4432,
      "step": 41823
    },
    {
      "epoch": 2.342031582484041,
      "grad_norm": 1.224095344543457,
      "learning_rate": 9.426184210526316e-05,
      "loss": 0.4479,
      "step": 41824
    },
    {
      "epoch": 2.3420875797961695,
      "grad_norm": 1.6423996686935425,
      "learning_rate": 9.426157894736842e-05,
      "loss": 0.4092,
      "step": 41825
    },
    {
      "epoch": 2.342143577108299,
      "grad_norm": 1.4122017621994019,
      "learning_rate": 9.42613157894737e-05,
      "loss": 0.4754,
      "step": 41826
    },
    {
      "epoch": 2.3421995744204276,
      "grad_norm": 1.071643590927124,
      "learning_rate": 9.426105263157896e-05,
      "loss": 0.3118,
      "step": 41827
    },
    {
      "epoch": 2.342255571732557,
      "grad_norm": 1.2291921377182007,
      "learning_rate": 9.426078947368422e-05,
      "loss": 0.485,
      "step": 41828
    },
    {
      "epoch": 2.3423115690446856,
      "grad_norm": 1.5928658246994019,
      "learning_rate": 9.426052631578948e-05,
      "loss": 0.4086,
      "step": 41829
    },
    {
      "epoch": 2.342367566356815,
      "grad_norm": 1.258038878440857,
      "learning_rate": 9.426026315789475e-05,
      "loss": 0.3436,
      "step": 41830
    },
    {
      "epoch": 2.3424235636689437,
      "grad_norm": 1.3400644063949585,
      "learning_rate": 9.426000000000001e-05,
      "loss": 0.6097,
      "step": 41831
    },
    {
      "epoch": 2.342479560981073,
      "grad_norm": 1.2180894613265991,
      "learning_rate": 9.425973684210527e-05,
      "loss": 0.3986,
      "step": 41832
    },
    {
      "epoch": 2.3425355582932017,
      "grad_norm": 1.7089297771453857,
      "learning_rate": 9.425947368421053e-05,
      "loss": 0.4027,
      "step": 41833
    },
    {
      "epoch": 2.342591555605331,
      "grad_norm": 1.1832506656646729,
      "learning_rate": 9.425921052631579e-05,
      "loss": 0.3882,
      "step": 41834
    },
    {
      "epoch": 2.3426475529174597,
      "grad_norm": 1.1357338428497314,
      "learning_rate": 9.425894736842106e-05,
      "loss": 0.34,
      "step": 41835
    },
    {
      "epoch": 2.342703550229589,
      "grad_norm": 1.2460598945617676,
      "learning_rate": 9.425868421052632e-05,
      "loss": 0.4762,
      "step": 41836
    },
    {
      "epoch": 2.342759547541718,
      "grad_norm": 1.6425243616104126,
      "learning_rate": 9.425842105263158e-05,
      "loss": 0.3345,
      "step": 41837
    },
    {
      "epoch": 2.3428155448538472,
      "grad_norm": 1.2084277868270874,
      "learning_rate": 9.425815789473684e-05,
      "loss": 0.3793,
      "step": 41838
    },
    {
      "epoch": 2.342871542165976,
      "grad_norm": 1.2540837526321411,
      "learning_rate": 9.425789473684211e-05,
      "loss": 0.3491,
      "step": 41839
    },
    {
      "epoch": 2.3429275394781053,
      "grad_norm": 1.1292132139205933,
      "learning_rate": 9.425763157894737e-05,
      "loss": 0.3796,
      "step": 41840
    },
    {
      "epoch": 2.342983536790234,
      "grad_norm": 3.4217171669006348,
      "learning_rate": 9.425736842105263e-05,
      "loss": 0.3949,
      "step": 41841
    },
    {
      "epoch": 2.3430395341023633,
      "grad_norm": 1.3985133171081543,
      "learning_rate": 9.425710526315789e-05,
      "loss": 0.4345,
      "step": 41842
    },
    {
      "epoch": 2.343095531414492,
      "grad_norm": 1.1672624349594116,
      "learning_rate": 9.425684210526317e-05,
      "loss": 0.4105,
      "step": 41843
    },
    {
      "epoch": 2.3431515287266214,
      "grad_norm": 1.179754376411438,
      "learning_rate": 9.425657894736843e-05,
      "loss": 0.3785,
      "step": 41844
    },
    {
      "epoch": 2.34320752603875,
      "grad_norm": 1.3450326919555664,
      "learning_rate": 9.42563157894737e-05,
      "loss": 0.4209,
      "step": 41845
    },
    {
      "epoch": 2.3432635233508794,
      "grad_norm": 1.2324138879776,
      "learning_rate": 9.425605263157895e-05,
      "loss": 0.4119,
      "step": 41846
    },
    {
      "epoch": 2.343319520663008,
      "grad_norm": 0.9802883863449097,
      "learning_rate": 9.425578947368422e-05,
      "loss": 0.295,
      "step": 41847
    },
    {
      "epoch": 2.3433755179751374,
      "grad_norm": 1.1618484258651733,
      "learning_rate": 9.425552631578948e-05,
      "loss": 0.5352,
      "step": 41848
    },
    {
      "epoch": 2.343431515287266,
      "grad_norm": 1.0275614261627197,
      "learning_rate": 9.425526315789475e-05,
      "loss": 0.3691,
      "step": 41849
    },
    {
      "epoch": 2.3434875125993955,
      "grad_norm": 1.4349318742752075,
      "learning_rate": 9.4255e-05,
      "loss": 0.3887,
      "step": 41850
    },
    {
      "epoch": 2.343543509911524,
      "grad_norm": 1.4337527751922607,
      "learning_rate": 9.425473684210526e-05,
      "loss": 0.4679,
      "step": 41851
    },
    {
      "epoch": 2.3435995072236535,
      "grad_norm": 1.142625331878662,
      "learning_rate": 9.425447368421053e-05,
      "loss": 0.4438,
      "step": 41852
    },
    {
      "epoch": 2.343655504535782,
      "grad_norm": 1.2787381410598755,
      "learning_rate": 9.425421052631579e-05,
      "loss": 0.3788,
      "step": 41853
    },
    {
      "epoch": 2.3437115018479115,
      "grad_norm": 1.2361544370651245,
      "learning_rate": 9.425394736842106e-05,
      "loss": 0.4464,
      "step": 41854
    },
    {
      "epoch": 2.34376749916004,
      "grad_norm": 1.086199164390564,
      "learning_rate": 9.425368421052631e-05,
      "loss": 0.2947,
      "step": 41855
    },
    {
      "epoch": 2.343823496472169,
      "grad_norm": 1.2735209465026855,
      "learning_rate": 9.425342105263158e-05,
      "loss": 0.496,
      "step": 41856
    },
    {
      "epoch": 2.343879493784298,
      "grad_norm": 1.3250583410263062,
      "learning_rate": 9.425315789473684e-05,
      "loss": 0.3788,
      "step": 41857
    },
    {
      "epoch": 2.343935491096427,
      "grad_norm": 1.5316301584243774,
      "learning_rate": 9.425289473684212e-05,
      "loss": 0.6071,
      "step": 41858
    },
    {
      "epoch": 2.343991488408556,
      "grad_norm": 1.1828645467758179,
      "learning_rate": 9.425263157894738e-05,
      "loss": 0.4172,
      "step": 41859
    },
    {
      "epoch": 2.344047485720685,
      "grad_norm": 1.4202466011047363,
      "learning_rate": 9.425236842105264e-05,
      "loss": 0.4974,
      "step": 41860
    },
    {
      "epoch": 2.3441034830328142,
      "grad_norm": 1.1320104598999023,
      "learning_rate": 9.42521052631579e-05,
      "loss": 0.3741,
      "step": 41861
    },
    {
      "epoch": 2.3441594803449433,
      "grad_norm": 1.3960785865783691,
      "learning_rate": 9.425184210526317e-05,
      "loss": 0.447,
      "step": 41862
    },
    {
      "epoch": 2.3442154776570723,
      "grad_norm": 1.4883716106414795,
      "learning_rate": 9.425157894736843e-05,
      "loss": 0.4523,
      "step": 41863
    },
    {
      "epoch": 2.3442714749692013,
      "grad_norm": 1.328608751296997,
      "learning_rate": 9.425131578947369e-05,
      "loss": 0.3863,
      "step": 41864
    },
    {
      "epoch": 2.3443274722813303,
      "grad_norm": 1.705578088760376,
      "learning_rate": 9.425105263157895e-05,
      "loss": 0.4822,
      "step": 41865
    },
    {
      "epoch": 2.3443834695934593,
      "grad_norm": 1.3447991609573364,
      "learning_rate": 9.425078947368422e-05,
      "loss": 0.3543,
      "step": 41866
    },
    {
      "epoch": 2.3444394669055884,
      "grad_norm": 1.1524291038513184,
      "learning_rate": 9.425052631578948e-05,
      "loss": 0.4113,
      "step": 41867
    },
    {
      "epoch": 2.3444954642177174,
      "grad_norm": 1.7251302003860474,
      "learning_rate": 9.425026315789474e-05,
      "loss": 0.5049,
      "step": 41868
    },
    {
      "epoch": 2.3445514615298464,
      "grad_norm": 1.0794435739517212,
      "learning_rate": 9.425e-05,
      "loss": 0.421,
      "step": 41869
    },
    {
      "epoch": 2.3446074588419754,
      "grad_norm": 1.2601521015167236,
      "learning_rate": 9.424973684210526e-05,
      "loss": 0.4124,
      "step": 41870
    },
    {
      "epoch": 2.3446634561541044,
      "grad_norm": 1.3364636898040771,
      "learning_rate": 9.424947368421053e-05,
      "loss": 0.4945,
      "step": 41871
    },
    {
      "epoch": 2.3447194534662334,
      "grad_norm": 1.2101716995239258,
      "learning_rate": 9.42492105263158e-05,
      "loss": 0.4145,
      "step": 41872
    },
    {
      "epoch": 2.3447754507783625,
      "grad_norm": 1.1150449514389038,
      "learning_rate": 9.424894736842105e-05,
      "loss": 0.3268,
      "step": 41873
    },
    {
      "epoch": 2.3448314480904915,
      "grad_norm": 1.2457066774368286,
      "learning_rate": 9.424868421052631e-05,
      "loss": 0.471,
      "step": 41874
    },
    {
      "epoch": 2.3448874454026205,
      "grad_norm": 1.1699256896972656,
      "learning_rate": 9.424842105263159e-05,
      "loss": 0.4224,
      "step": 41875
    },
    {
      "epoch": 2.3449434427147495,
      "grad_norm": 1.1539231538772583,
      "learning_rate": 9.424815789473685e-05,
      "loss": 0.391,
      "step": 41876
    },
    {
      "epoch": 2.3449994400268785,
      "grad_norm": 1.3850818872451782,
      "learning_rate": 9.424789473684212e-05,
      "loss": 0.5928,
      "step": 41877
    },
    {
      "epoch": 2.3450554373390076,
      "grad_norm": 1.1350572109222412,
      "learning_rate": 9.424763157894737e-05,
      "loss": 0.3388,
      "step": 41878
    },
    {
      "epoch": 2.3451114346511366,
      "grad_norm": 1.3524158000946045,
      "learning_rate": 9.424736842105264e-05,
      "loss": 0.5431,
      "step": 41879
    },
    {
      "epoch": 2.3451674319632656,
      "grad_norm": 1.4524420499801636,
      "learning_rate": 9.42471052631579e-05,
      "loss": 0.4179,
      "step": 41880
    },
    {
      "epoch": 2.3452234292753946,
      "grad_norm": 1.5528044700622559,
      "learning_rate": 9.424684210526317e-05,
      "loss": 0.3886,
      "step": 41881
    },
    {
      "epoch": 2.3452794265875236,
      "grad_norm": 1.3053419589996338,
      "learning_rate": 9.424657894736843e-05,
      "loss": 0.4776,
      "step": 41882
    },
    {
      "epoch": 2.3453354238996527,
      "grad_norm": 1.2407245635986328,
      "learning_rate": 9.424631578947369e-05,
      "loss": 0.5198,
      "step": 41883
    },
    {
      "epoch": 2.3453914212117817,
      "grad_norm": 1.1918450593948364,
      "learning_rate": 9.424605263157895e-05,
      "loss": 0.409,
      "step": 41884
    },
    {
      "epoch": 2.3454474185239107,
      "grad_norm": 1.6173343658447266,
      "learning_rate": 9.424578947368421e-05,
      "loss": 0.4784,
      "step": 41885
    },
    {
      "epoch": 2.3455034158360397,
      "grad_norm": 1.0414845943450928,
      "learning_rate": 9.424552631578948e-05,
      "loss": 0.3483,
      "step": 41886
    },
    {
      "epoch": 2.3455594131481687,
      "grad_norm": 1.2562373876571655,
      "learning_rate": 9.424526315789473e-05,
      "loss": 0.4984,
      "step": 41887
    },
    {
      "epoch": 2.3456154104602978,
      "grad_norm": 1.0403567552566528,
      "learning_rate": 9.4245e-05,
      "loss": 0.4442,
      "step": 41888
    },
    {
      "epoch": 2.3456714077724268,
      "grad_norm": 1.015623688697815,
      "learning_rate": 9.424473684210526e-05,
      "loss": 0.3414,
      "step": 41889
    },
    {
      "epoch": 2.345727405084556,
      "grad_norm": 1.4809482097625732,
      "learning_rate": 9.424447368421054e-05,
      "loss": 0.4381,
      "step": 41890
    },
    {
      "epoch": 2.345783402396685,
      "grad_norm": 1.1635187864303589,
      "learning_rate": 9.42442105263158e-05,
      "loss": 0.4136,
      "step": 41891
    },
    {
      "epoch": 2.345839399708814,
      "grad_norm": 1.2481491565704346,
      "learning_rate": 9.424394736842106e-05,
      "loss": 0.3558,
      "step": 41892
    },
    {
      "epoch": 2.345895397020943,
      "grad_norm": 1.7884985208511353,
      "learning_rate": 9.424368421052632e-05,
      "loss": 0.7679,
      "step": 41893
    },
    {
      "epoch": 2.345951394333072,
      "grad_norm": 1.0589234828948975,
      "learning_rate": 9.424342105263159e-05,
      "loss": 0.3458,
      "step": 41894
    },
    {
      "epoch": 2.346007391645201,
      "grad_norm": 1.3798965215682983,
      "learning_rate": 9.424315789473685e-05,
      "loss": 0.5949,
      "step": 41895
    },
    {
      "epoch": 2.34606338895733,
      "grad_norm": 1.2889463901519775,
      "learning_rate": 9.424289473684211e-05,
      "loss": 0.3607,
      "step": 41896
    },
    {
      "epoch": 2.346119386269459,
      "grad_norm": 1.4177976846694946,
      "learning_rate": 9.424263157894737e-05,
      "loss": 0.739,
      "step": 41897
    },
    {
      "epoch": 2.346175383581588,
      "grad_norm": 1.213110327720642,
      "learning_rate": 9.424236842105264e-05,
      "loss": 0.5117,
      "step": 41898
    },
    {
      "epoch": 2.346231380893717,
      "grad_norm": 1.0544935464859009,
      "learning_rate": 9.42421052631579e-05,
      "loss": 0.3846,
      "step": 41899
    },
    {
      "epoch": 2.346287378205846,
      "grad_norm": 1.1342898607254028,
      "learning_rate": 9.424184210526316e-05,
      "loss": 0.3124,
      "step": 41900
    },
    {
      "epoch": 2.346343375517975,
      "grad_norm": 1.0577930212020874,
      "learning_rate": 9.424157894736842e-05,
      "loss": 0.3406,
      "step": 41901
    },
    {
      "epoch": 2.346399372830104,
      "grad_norm": 1.6600995063781738,
      "learning_rate": 9.42413157894737e-05,
      "loss": 0.3875,
      "step": 41902
    },
    {
      "epoch": 2.346455370142233,
      "grad_norm": 1.125165343284607,
      "learning_rate": 9.424105263157895e-05,
      "loss": 0.3791,
      "step": 41903
    },
    {
      "epoch": 2.346511367454362,
      "grad_norm": 1.086573600769043,
      "learning_rate": 9.424078947368421e-05,
      "loss": 0.331,
      "step": 41904
    },
    {
      "epoch": 2.346567364766491,
      "grad_norm": 1.5938458442687988,
      "learning_rate": 9.424052631578947e-05,
      "loss": 0.4114,
      "step": 41905
    },
    {
      "epoch": 2.34662336207862,
      "grad_norm": 1.1174705028533936,
      "learning_rate": 9.424026315789473e-05,
      "loss": 0.3278,
      "step": 41906
    },
    {
      "epoch": 2.346679359390749,
      "grad_norm": 0.9441462755203247,
      "learning_rate": 9.424e-05,
      "loss": 0.3228,
      "step": 41907
    },
    {
      "epoch": 2.346735356702878,
      "grad_norm": 1.315874457359314,
      "learning_rate": 9.423973684210527e-05,
      "loss": 0.4691,
      "step": 41908
    },
    {
      "epoch": 2.346791354015007,
      "grad_norm": 1.1920456886291504,
      "learning_rate": 9.423947368421054e-05,
      "loss": 0.4221,
      "step": 41909
    },
    {
      "epoch": 2.346847351327136,
      "grad_norm": 1.08306884765625,
      "learning_rate": 9.423921052631578e-05,
      "loss": 0.4482,
      "step": 41910
    },
    {
      "epoch": 2.346903348639265,
      "grad_norm": 1.1478420495986938,
      "learning_rate": 9.423894736842106e-05,
      "loss": 0.4323,
      "step": 41911
    },
    {
      "epoch": 2.346959345951394,
      "grad_norm": 1.2103220224380493,
      "learning_rate": 9.423868421052632e-05,
      "loss": 0.5561,
      "step": 41912
    },
    {
      "epoch": 2.3470153432635232,
      "grad_norm": 1.527949333190918,
      "learning_rate": 9.423842105263159e-05,
      "loss": 0.4219,
      "step": 41913
    },
    {
      "epoch": 2.3470713405756523,
      "grad_norm": 1.2173051834106445,
      "learning_rate": 9.423815789473685e-05,
      "loss": 0.4368,
      "step": 41914
    },
    {
      "epoch": 2.3471273378877813,
      "grad_norm": 3.7206315994262695,
      "learning_rate": 9.423789473684211e-05,
      "loss": 0.3615,
      "step": 41915
    },
    {
      "epoch": 2.3471833351999103,
      "grad_norm": 1.016814112663269,
      "learning_rate": 9.423763157894737e-05,
      "loss": 0.3292,
      "step": 41916
    },
    {
      "epoch": 2.3472393325120393,
      "grad_norm": 1.2219231128692627,
      "learning_rate": 9.423736842105264e-05,
      "loss": 0.5443,
      "step": 41917
    },
    {
      "epoch": 2.3472953298241683,
      "grad_norm": 1.270013689994812,
      "learning_rate": 9.42371052631579e-05,
      "loss": 0.4485,
      "step": 41918
    },
    {
      "epoch": 2.3473513271362973,
      "grad_norm": 1.19881272315979,
      "learning_rate": 9.423684210526316e-05,
      "loss": 0.3319,
      "step": 41919
    },
    {
      "epoch": 2.3474073244484264,
      "grad_norm": 1.1045656204223633,
      "learning_rate": 9.423657894736842e-05,
      "loss": 0.3807,
      "step": 41920
    },
    {
      "epoch": 2.3474633217605554,
      "grad_norm": 0.9773579835891724,
      "learning_rate": 9.423631578947368e-05,
      "loss": 0.3254,
      "step": 41921
    },
    {
      "epoch": 2.3475193190726844,
      "grad_norm": 1.30803382396698,
      "learning_rate": 9.423605263157896e-05,
      "loss": 0.3976,
      "step": 41922
    },
    {
      "epoch": 2.3475753163848134,
      "grad_norm": 1.4182490110397339,
      "learning_rate": 9.423578947368422e-05,
      "loss": 0.4168,
      "step": 41923
    },
    {
      "epoch": 2.3476313136969424,
      "grad_norm": 1.389750361442566,
      "learning_rate": 9.423552631578948e-05,
      "loss": 0.4313,
      "step": 41924
    },
    {
      "epoch": 2.3476873110090715,
      "grad_norm": 1.2765437364578247,
      "learning_rate": 9.423526315789474e-05,
      "loss": 0.6314,
      "step": 41925
    },
    {
      "epoch": 2.3477433083212005,
      "grad_norm": 1.0952752828598022,
      "learning_rate": 9.423500000000001e-05,
      "loss": 0.3257,
      "step": 41926
    },
    {
      "epoch": 2.3477993056333295,
      "grad_norm": 1.1169794797897339,
      "learning_rate": 9.423473684210527e-05,
      "loss": 0.425,
      "step": 41927
    },
    {
      "epoch": 2.3478553029454585,
      "grad_norm": 1.1962910890579224,
      "learning_rate": 9.423447368421053e-05,
      "loss": 0.3431,
      "step": 41928
    },
    {
      "epoch": 2.3479113002575875,
      "grad_norm": 1.6997184753417969,
      "learning_rate": 9.423421052631579e-05,
      "loss": 0.3816,
      "step": 41929
    },
    {
      "epoch": 2.3479672975697166,
      "grad_norm": 1.2387455701828003,
      "learning_rate": 9.423394736842106e-05,
      "loss": 0.3539,
      "step": 41930
    },
    {
      "epoch": 2.3480232948818456,
      "grad_norm": 1.4194833040237427,
      "learning_rate": 9.423368421052632e-05,
      "loss": 0.5217,
      "step": 41931
    },
    {
      "epoch": 2.3480792921939746,
      "grad_norm": 1.1208289861679077,
      "learning_rate": 9.42334210526316e-05,
      "loss": 0.4095,
      "step": 41932
    },
    {
      "epoch": 2.3481352895061036,
      "grad_norm": 2.0579662322998047,
      "learning_rate": 9.423315789473684e-05,
      "loss": 0.3798,
      "step": 41933
    },
    {
      "epoch": 2.3481912868182326,
      "grad_norm": 1.0255206823349,
      "learning_rate": 9.423289473684211e-05,
      "loss": 0.4104,
      "step": 41934
    },
    {
      "epoch": 2.3482472841303617,
      "grad_norm": 1.035516619682312,
      "learning_rate": 9.423263157894737e-05,
      "loss": 0.4242,
      "step": 41935
    },
    {
      "epoch": 2.3483032814424907,
      "grad_norm": 1.3136500120162964,
      "learning_rate": 9.423236842105265e-05,
      "loss": 0.3947,
      "step": 41936
    },
    {
      "epoch": 2.3483592787546197,
      "grad_norm": 1.2971669435501099,
      "learning_rate": 9.42321052631579e-05,
      "loss": 0.4815,
      "step": 41937
    },
    {
      "epoch": 2.3484152760667487,
      "grad_norm": 1.1738402843475342,
      "learning_rate": 9.423184210526315e-05,
      "loss": 0.2907,
      "step": 41938
    },
    {
      "epoch": 2.3484712733788777,
      "grad_norm": 1.0700563192367554,
      "learning_rate": 9.423157894736843e-05,
      "loss": 0.3582,
      "step": 41939
    },
    {
      "epoch": 2.3485272706910068,
      "grad_norm": 1.2877978086471558,
      "learning_rate": 9.423131578947369e-05,
      "loss": 0.3896,
      "step": 41940
    },
    {
      "epoch": 2.3485832680031358,
      "grad_norm": 1.0687603950500488,
      "learning_rate": 9.423105263157896e-05,
      "loss": 0.4127,
      "step": 41941
    },
    {
      "epoch": 2.348639265315265,
      "grad_norm": 1.3275868892669678,
      "learning_rate": 9.42307894736842e-05,
      "loss": 0.3145,
      "step": 41942
    },
    {
      "epoch": 2.348695262627394,
      "grad_norm": 1.0240685939788818,
      "learning_rate": 9.423052631578948e-05,
      "loss": 0.3168,
      "step": 41943
    },
    {
      "epoch": 2.348751259939523,
      "grad_norm": 1.4507956504821777,
      "learning_rate": 9.423026315789474e-05,
      "loss": 0.3863,
      "step": 41944
    },
    {
      "epoch": 2.348807257251652,
      "grad_norm": 1.2680301666259766,
      "learning_rate": 9.423000000000001e-05,
      "loss": 0.4111,
      "step": 41945
    },
    {
      "epoch": 2.348863254563781,
      "grad_norm": 1.1888868808746338,
      "learning_rate": 9.422973684210527e-05,
      "loss": 0.2686,
      "step": 41946
    },
    {
      "epoch": 2.34891925187591,
      "grad_norm": 1.0212517976760864,
      "learning_rate": 9.422947368421053e-05,
      "loss": 0.3038,
      "step": 41947
    },
    {
      "epoch": 2.348975249188039,
      "grad_norm": 1.2542644739151,
      "learning_rate": 9.422921052631579e-05,
      "loss": 0.3456,
      "step": 41948
    },
    {
      "epoch": 2.349031246500168,
      "grad_norm": 1.1059317588806152,
      "learning_rate": 9.422894736842106e-05,
      "loss": 0.3615,
      "step": 41949
    },
    {
      "epoch": 2.349087243812297,
      "grad_norm": 1.2373396158218384,
      "learning_rate": 9.422868421052632e-05,
      "loss": 0.5558,
      "step": 41950
    },
    {
      "epoch": 2.349143241124426,
      "grad_norm": 1.3789201974868774,
      "learning_rate": 9.422842105263158e-05,
      "loss": 0.4783,
      "step": 41951
    },
    {
      "epoch": 2.349199238436555,
      "grad_norm": 1.4175273180007935,
      "learning_rate": 9.422815789473684e-05,
      "loss": 0.3201,
      "step": 41952
    },
    {
      "epoch": 2.349255235748684,
      "grad_norm": 1.0703898668289185,
      "learning_rate": 9.422789473684212e-05,
      "loss": 0.3748,
      "step": 41953
    },
    {
      "epoch": 2.349311233060813,
      "grad_norm": 1.4079670906066895,
      "learning_rate": 9.422763157894738e-05,
      "loss": 0.4475,
      "step": 41954
    },
    {
      "epoch": 2.349367230372942,
      "grad_norm": 1.2110671997070312,
      "learning_rate": 9.422736842105264e-05,
      "loss": 0.3905,
      "step": 41955
    },
    {
      "epoch": 2.349423227685071,
      "grad_norm": 1.2391728162765503,
      "learning_rate": 9.42271052631579e-05,
      "loss": 0.3827,
      "step": 41956
    },
    {
      "epoch": 2.3494792249972,
      "grad_norm": 1.0653246641159058,
      "learning_rate": 9.422684210526315e-05,
      "loss": 0.3572,
      "step": 41957
    },
    {
      "epoch": 2.349535222309329,
      "grad_norm": 1.1151396036148071,
      "learning_rate": 9.422657894736843e-05,
      "loss": 0.4081,
      "step": 41958
    },
    {
      "epoch": 2.349591219621458,
      "grad_norm": 1.2151936292648315,
      "learning_rate": 9.422631578947369e-05,
      "loss": 0.3841,
      "step": 41959
    },
    {
      "epoch": 2.349647216933587,
      "grad_norm": 1.547565221786499,
      "learning_rate": 9.422605263157895e-05,
      "loss": 0.4742,
      "step": 41960
    },
    {
      "epoch": 2.349703214245716,
      "grad_norm": 1.3111594915390015,
      "learning_rate": 9.422578947368421e-05,
      "loss": 0.3881,
      "step": 41961
    },
    {
      "epoch": 2.349759211557845,
      "grad_norm": 1.1176769733428955,
      "learning_rate": 9.422552631578948e-05,
      "loss": 0.3866,
      "step": 41962
    },
    {
      "epoch": 2.349815208869974,
      "grad_norm": 2.228069305419922,
      "learning_rate": 9.422526315789474e-05,
      "loss": 0.4115,
      "step": 41963
    },
    {
      "epoch": 2.349871206182103,
      "grad_norm": 1.9174860715866089,
      "learning_rate": 9.422500000000001e-05,
      "loss": 0.3668,
      "step": 41964
    },
    {
      "epoch": 2.3499272034942322,
      "grad_norm": 1.223373532295227,
      "learning_rate": 9.422473684210526e-05,
      "loss": 0.4554,
      "step": 41965
    },
    {
      "epoch": 2.3499832008063612,
      "grad_norm": 1.7548977136611938,
      "learning_rate": 9.422447368421053e-05,
      "loss": 0.4082,
      "step": 41966
    },
    {
      "epoch": 2.3500391981184903,
      "grad_norm": 10.118800163269043,
      "learning_rate": 9.422421052631579e-05,
      "loss": 0.3732,
      "step": 41967
    },
    {
      "epoch": 2.3500951954306193,
      "grad_norm": 1.1329727172851562,
      "learning_rate": 9.422394736842107e-05,
      "loss": 0.3455,
      "step": 41968
    },
    {
      "epoch": 2.3501511927427483,
      "grad_norm": 1.1047589778900146,
      "learning_rate": 9.422368421052633e-05,
      "loss": 0.3214,
      "step": 41969
    },
    {
      "epoch": 2.3502071900548773,
      "grad_norm": 1.398796796798706,
      "learning_rate": 9.422342105263159e-05,
      "loss": 0.468,
      "step": 41970
    },
    {
      "epoch": 2.3502631873670063,
      "grad_norm": 1.2286341190338135,
      "learning_rate": 9.422315789473685e-05,
      "loss": 0.5296,
      "step": 41971
    },
    {
      "epoch": 2.3503191846791354,
      "grad_norm": 1.1639487743377686,
      "learning_rate": 9.42228947368421e-05,
      "loss": 0.4148,
      "step": 41972
    },
    {
      "epoch": 2.3503751819912644,
      "grad_norm": 1.1125766038894653,
      "learning_rate": 9.422263157894738e-05,
      "loss": 0.384,
      "step": 41973
    },
    {
      "epoch": 2.3504311793033934,
      "grad_norm": 1.1336102485656738,
      "learning_rate": 9.422236842105264e-05,
      "loss": 0.4169,
      "step": 41974
    },
    {
      "epoch": 2.3504871766155224,
      "grad_norm": 1.2123184204101562,
      "learning_rate": 9.42221052631579e-05,
      "loss": 0.4211,
      "step": 41975
    },
    {
      "epoch": 2.3505431739276514,
      "grad_norm": 1.0760911703109741,
      "learning_rate": 9.422184210526316e-05,
      "loss": 0.3141,
      "step": 41976
    },
    {
      "epoch": 2.3505991712397805,
      "grad_norm": 1.1022170782089233,
      "learning_rate": 9.422157894736843e-05,
      "loss": 0.3608,
      "step": 41977
    },
    {
      "epoch": 2.3506551685519095,
      "grad_norm": 1.0313996076583862,
      "learning_rate": 9.422131578947369e-05,
      "loss": 0.3936,
      "step": 41978
    },
    {
      "epoch": 2.3507111658640385,
      "grad_norm": 1.2951610088348389,
      "learning_rate": 9.422105263157895e-05,
      "loss": 0.3751,
      "step": 41979
    },
    {
      "epoch": 2.3507671631761675,
      "grad_norm": 1.344642162322998,
      "learning_rate": 9.422078947368421e-05,
      "loss": 0.5303,
      "step": 41980
    },
    {
      "epoch": 2.3508231604882965,
      "grad_norm": 1.1803078651428223,
      "learning_rate": 9.422052631578948e-05,
      "loss": 0.354,
      "step": 41981
    },
    {
      "epoch": 2.3508791578004256,
      "grad_norm": 1.8724883794784546,
      "learning_rate": 9.422026315789474e-05,
      "loss": 0.4203,
      "step": 41982
    },
    {
      "epoch": 2.3509351551125546,
      "grad_norm": 1.3569070100784302,
      "learning_rate": 9.422e-05,
      "loss": 0.3127,
      "step": 41983
    },
    {
      "epoch": 2.3509911524246836,
      "grad_norm": 1.0621393918991089,
      "learning_rate": 9.421973684210526e-05,
      "loss": 0.3003,
      "step": 41984
    },
    {
      "epoch": 2.3510471497368126,
      "grad_norm": 1.196302890777588,
      "learning_rate": 9.421947368421054e-05,
      "loss": 0.4334,
      "step": 41985
    },
    {
      "epoch": 2.3511031470489416,
      "grad_norm": 1.472033143043518,
      "learning_rate": 9.42192105263158e-05,
      "loss": 0.4106,
      "step": 41986
    },
    {
      "epoch": 2.3511591443610707,
      "grad_norm": 1.4869413375854492,
      "learning_rate": 9.421894736842107e-05,
      "loss": 0.4798,
      "step": 41987
    },
    {
      "epoch": 2.3512151416731997,
      "grad_norm": 1.1627663373947144,
      "learning_rate": 9.421868421052631e-05,
      "loss": 0.3811,
      "step": 41988
    },
    {
      "epoch": 2.3512711389853287,
      "grad_norm": 1.048711895942688,
      "learning_rate": 9.421842105263157e-05,
      "loss": 0.4891,
      "step": 41989
    },
    {
      "epoch": 2.3513271362974577,
      "grad_norm": 1.013312578201294,
      "learning_rate": 9.421815789473685e-05,
      "loss": 0.3595,
      "step": 41990
    },
    {
      "epoch": 2.3513831336095867,
      "grad_norm": 1.2516334056854248,
      "learning_rate": 9.421789473684211e-05,
      "loss": 0.4513,
      "step": 41991
    },
    {
      "epoch": 2.3514391309217157,
      "grad_norm": 1.0890311002731323,
      "learning_rate": 9.421763157894738e-05,
      "loss": 0.3958,
      "step": 41992
    },
    {
      "epoch": 2.3514951282338448,
      "grad_norm": 1.163549542427063,
      "learning_rate": 9.421736842105263e-05,
      "loss": 0.4203,
      "step": 41993
    },
    {
      "epoch": 2.351551125545974,
      "grad_norm": 1.35733962059021,
      "learning_rate": 9.42171052631579e-05,
      "loss": 0.4986,
      "step": 41994
    },
    {
      "epoch": 2.351607122858103,
      "grad_norm": 1.1108150482177734,
      "learning_rate": 9.421684210526316e-05,
      "loss": 0.3442,
      "step": 41995
    },
    {
      "epoch": 2.351663120170232,
      "grad_norm": 0.950831949710846,
      "learning_rate": 9.421657894736843e-05,
      "loss": 0.3652,
      "step": 41996
    },
    {
      "epoch": 2.351719117482361,
      "grad_norm": 1.182410478591919,
      "learning_rate": 9.421631578947368e-05,
      "loss": 0.423,
      "step": 41997
    },
    {
      "epoch": 2.35177511479449,
      "grad_norm": 1.0937422513961792,
      "learning_rate": 9.421605263157895e-05,
      "loss": 0.3506,
      "step": 41998
    },
    {
      "epoch": 2.351831112106619,
      "grad_norm": 1.310194969177246,
      "learning_rate": 9.421578947368421e-05,
      "loss": 0.41,
      "step": 41999
    },
    {
      "epoch": 2.351887109418748,
      "grad_norm": 1.4692976474761963,
      "learning_rate": 9.421552631578949e-05,
      "loss": 0.4413,
      "step": 42000
    },
    {
      "epoch": 2.351943106730877,
      "grad_norm": 1.204043984413147,
      "learning_rate": 9.421526315789475e-05,
      "loss": 0.4192,
      "step": 42001
    },
    {
      "epoch": 2.351999104043006,
      "grad_norm": 1.1480443477630615,
      "learning_rate": 9.4215e-05,
      "loss": 0.3172,
      "step": 42002
    },
    {
      "epoch": 2.352055101355135,
      "grad_norm": 1.1406035423278809,
      "learning_rate": 9.421473684210526e-05,
      "loss": 0.3259,
      "step": 42003
    },
    {
      "epoch": 2.352111098667264,
      "grad_norm": 1.1293482780456543,
      "learning_rate": 9.421447368421054e-05,
      "loss": 0.3216,
      "step": 42004
    },
    {
      "epoch": 2.352167095979393,
      "grad_norm": 1.1756552457809448,
      "learning_rate": 9.42142105263158e-05,
      "loss": 0.3967,
      "step": 42005
    },
    {
      "epoch": 2.352223093291522,
      "grad_norm": 1.1172279119491577,
      "learning_rate": 9.421394736842106e-05,
      "loss": 0.3094,
      "step": 42006
    },
    {
      "epoch": 2.352279090603651,
      "grad_norm": 1.8310580253601074,
      "learning_rate": 9.421368421052632e-05,
      "loss": 0.4468,
      "step": 42007
    },
    {
      "epoch": 2.35233508791578,
      "grad_norm": 1.2626774311065674,
      "learning_rate": 9.421342105263158e-05,
      "loss": 0.3546,
      "step": 42008
    },
    {
      "epoch": 2.352391085227909,
      "grad_norm": 1.250585913658142,
      "learning_rate": 9.421315789473685e-05,
      "loss": 0.4734,
      "step": 42009
    },
    {
      "epoch": 2.352447082540038,
      "grad_norm": 1.2268357276916504,
      "learning_rate": 9.421289473684211e-05,
      "loss": 0.4409,
      "step": 42010
    },
    {
      "epoch": 2.352503079852167,
      "grad_norm": 1.2325795888900757,
      "learning_rate": 9.421263157894737e-05,
      "loss": 0.3902,
      "step": 42011
    },
    {
      "epoch": 2.352559077164296,
      "grad_norm": 1.3361058235168457,
      "learning_rate": 9.421236842105263e-05,
      "loss": 0.3958,
      "step": 42012
    },
    {
      "epoch": 2.352615074476425,
      "grad_norm": 1.0456767082214355,
      "learning_rate": 9.42121052631579e-05,
      "loss": 0.4249,
      "step": 42013
    },
    {
      "epoch": 2.352671071788554,
      "grad_norm": 1.154228925704956,
      "learning_rate": 9.421184210526316e-05,
      "loss": 0.4413,
      "step": 42014
    },
    {
      "epoch": 2.352727069100683,
      "grad_norm": 1.2956551313400269,
      "learning_rate": 9.421157894736842e-05,
      "loss": 0.442,
      "step": 42015
    },
    {
      "epoch": 2.352783066412812,
      "grad_norm": 1.205148458480835,
      "learning_rate": 9.421131578947368e-05,
      "loss": 0.4808,
      "step": 42016
    },
    {
      "epoch": 2.3528390637249412,
      "grad_norm": 1.5038695335388184,
      "learning_rate": 9.421105263157896e-05,
      "loss": 0.4367,
      "step": 42017
    },
    {
      "epoch": 2.3528950610370702,
      "grad_norm": 1.2735575437545776,
      "learning_rate": 9.421078947368421e-05,
      "loss": 0.4128,
      "step": 42018
    },
    {
      "epoch": 2.3529510583491993,
      "grad_norm": 1.2765793800354004,
      "learning_rate": 9.421052631578949e-05,
      "loss": 0.439,
      "step": 42019
    },
    {
      "epoch": 2.3530070556613283,
      "grad_norm": 1.1689181327819824,
      "learning_rate": 9.421026315789473e-05,
      "loss": 0.4519,
      "step": 42020
    },
    {
      "epoch": 2.3530630529734573,
      "grad_norm": 1.1140270233154297,
      "learning_rate": 9.421000000000001e-05,
      "loss": 0.3811,
      "step": 42021
    },
    {
      "epoch": 2.3531190502855863,
      "grad_norm": 1.6903316974639893,
      "learning_rate": 9.420973684210527e-05,
      "loss": 0.6076,
      "step": 42022
    },
    {
      "epoch": 2.3531750475977153,
      "grad_norm": 1.437776803970337,
      "learning_rate": 9.420947368421054e-05,
      "loss": 0.6009,
      "step": 42023
    },
    {
      "epoch": 2.3532310449098444,
      "grad_norm": 1.2875986099243164,
      "learning_rate": 9.42092105263158e-05,
      "loss": 0.5224,
      "step": 42024
    },
    {
      "epoch": 2.3532870422219734,
      "grad_norm": 1.072899580001831,
      "learning_rate": 9.420894736842105e-05,
      "loss": 0.3386,
      "step": 42025
    },
    {
      "epoch": 2.3533430395341024,
      "grad_norm": 1.104203701019287,
      "learning_rate": 9.420868421052632e-05,
      "loss": 0.3261,
      "step": 42026
    },
    {
      "epoch": 2.3533990368462314,
      "grad_norm": 1.3839703798294067,
      "learning_rate": 9.420842105263158e-05,
      "loss": 0.3843,
      "step": 42027
    },
    {
      "epoch": 2.3534550341583604,
      "grad_norm": 1.1921048164367676,
      "learning_rate": 9.420815789473685e-05,
      "loss": 0.5402,
      "step": 42028
    },
    {
      "epoch": 2.3535110314704895,
      "grad_norm": 1.1254487037658691,
      "learning_rate": 9.420789473684211e-05,
      "loss": 0.4031,
      "step": 42029
    },
    {
      "epoch": 2.3535670287826185,
      "grad_norm": 1.0913447141647339,
      "learning_rate": 9.420763157894737e-05,
      "loss": 0.4249,
      "step": 42030
    },
    {
      "epoch": 2.3536230260947475,
      "grad_norm": 1.4986735582351685,
      "learning_rate": 9.420736842105263e-05,
      "loss": 0.4284,
      "step": 42031
    },
    {
      "epoch": 2.3536790234068765,
      "grad_norm": 1.283302664756775,
      "learning_rate": 9.42071052631579e-05,
      "loss": 0.3525,
      "step": 42032
    },
    {
      "epoch": 2.3537350207190055,
      "grad_norm": 1.1544184684753418,
      "learning_rate": 9.420684210526317e-05,
      "loss": 0.4365,
      "step": 42033
    },
    {
      "epoch": 2.3537910180311346,
      "grad_norm": 1.0214793682098389,
      "learning_rate": 9.420657894736842e-05,
      "loss": 0.2939,
      "step": 42034
    },
    {
      "epoch": 2.3538470153432636,
      "grad_norm": 1.1046987771987915,
      "learning_rate": 9.420631578947368e-05,
      "loss": 0.3814,
      "step": 42035
    },
    {
      "epoch": 2.3539030126553926,
      "grad_norm": 1.1168564558029175,
      "learning_rate": 9.420605263157896e-05,
      "loss": 0.28,
      "step": 42036
    },
    {
      "epoch": 2.3539590099675216,
      "grad_norm": 1.2675065994262695,
      "learning_rate": 9.420578947368422e-05,
      "loss": 0.4099,
      "step": 42037
    },
    {
      "epoch": 2.3540150072796506,
      "grad_norm": 1.2064424753189087,
      "learning_rate": 9.420552631578948e-05,
      "loss": 0.4263,
      "step": 42038
    },
    {
      "epoch": 2.3540710045917796,
      "grad_norm": 1.1632829904556274,
      "learning_rate": 9.420526315789474e-05,
      "loss": 0.4582,
      "step": 42039
    },
    {
      "epoch": 2.3541270019039087,
      "grad_norm": 1.1134392023086548,
      "learning_rate": 9.420500000000001e-05,
      "loss": 0.3127,
      "step": 42040
    },
    {
      "epoch": 2.3541829992160377,
      "grad_norm": 1.4776973724365234,
      "learning_rate": 9.420473684210527e-05,
      "loss": 0.4227,
      "step": 42041
    },
    {
      "epoch": 2.3542389965281667,
      "grad_norm": 1.2541221380233765,
      "learning_rate": 9.420447368421053e-05,
      "loss": 0.453,
      "step": 42042
    },
    {
      "epoch": 2.3542949938402957,
      "grad_norm": 1.1257274150848389,
      "learning_rate": 9.420421052631579e-05,
      "loss": 0.3714,
      "step": 42043
    },
    {
      "epoch": 2.3543509911524247,
      "grad_norm": 1.248753547668457,
      "learning_rate": 9.420394736842105e-05,
      "loss": 0.4053,
      "step": 42044
    },
    {
      "epoch": 2.3544069884645538,
      "grad_norm": 1.2743364572525024,
      "learning_rate": 9.420368421052632e-05,
      "loss": 0.4346,
      "step": 42045
    },
    {
      "epoch": 2.354462985776683,
      "grad_norm": 1.0889259576797485,
      "learning_rate": 9.420342105263158e-05,
      "loss": 0.3278,
      "step": 42046
    },
    {
      "epoch": 2.354518983088812,
      "grad_norm": 1.3129757642745972,
      "learning_rate": 9.420315789473684e-05,
      "loss": 0.3579,
      "step": 42047
    },
    {
      "epoch": 2.354574980400941,
      "grad_norm": 1.2878180742263794,
      "learning_rate": 9.42028947368421e-05,
      "loss": 0.4799,
      "step": 42048
    },
    {
      "epoch": 2.35463097771307,
      "grad_norm": 1.365297555923462,
      "learning_rate": 9.420263157894737e-05,
      "loss": 0.3907,
      "step": 42049
    },
    {
      "epoch": 2.354686975025199,
      "grad_norm": 1.35836923122406,
      "learning_rate": 9.420236842105263e-05,
      "loss": 0.3393,
      "step": 42050
    },
    {
      "epoch": 2.354742972337328,
      "grad_norm": 1.517835259437561,
      "learning_rate": 9.420210526315791e-05,
      "loss": 0.4224,
      "step": 42051
    },
    {
      "epoch": 2.354798969649457,
      "grad_norm": 1.301274061203003,
      "learning_rate": 9.420184210526315e-05,
      "loss": 0.3745,
      "step": 42052
    },
    {
      "epoch": 2.354854966961586,
      "grad_norm": 1.34084951877594,
      "learning_rate": 9.420157894736843e-05,
      "loss": 0.5806,
      "step": 42053
    },
    {
      "epoch": 2.354910964273715,
      "grad_norm": 1.339669942855835,
      "learning_rate": 9.420131578947369e-05,
      "loss": 0.5756,
      "step": 42054
    },
    {
      "epoch": 2.354966961585844,
      "grad_norm": 1.5642307996749878,
      "learning_rate": 9.420105263157896e-05,
      "loss": 0.42,
      "step": 42055
    },
    {
      "epoch": 2.355022958897973,
      "grad_norm": 3.180543899536133,
      "learning_rate": 9.420078947368422e-05,
      "loss": 0.3992,
      "step": 42056
    },
    {
      "epoch": 2.355078956210102,
      "grad_norm": 1.3625800609588623,
      "learning_rate": 9.420052631578948e-05,
      "loss": 0.3916,
      "step": 42057
    },
    {
      "epoch": 2.355134953522231,
      "grad_norm": 2.2966105937957764,
      "learning_rate": 9.420026315789474e-05,
      "loss": 0.4625,
      "step": 42058
    },
    {
      "epoch": 2.35519095083436,
      "grad_norm": 1.1040472984313965,
      "learning_rate": 9.42e-05,
      "loss": 0.4012,
      "step": 42059
    },
    {
      "epoch": 2.355246948146489,
      "grad_norm": 1.2432657480239868,
      "learning_rate": 9.419973684210527e-05,
      "loss": 0.4039,
      "step": 42060
    },
    {
      "epoch": 2.355302945458618,
      "grad_norm": 1.4047868251800537,
      "learning_rate": 9.419947368421053e-05,
      "loss": 0.5086,
      "step": 42061
    },
    {
      "epoch": 2.355358942770747,
      "grad_norm": 1.0367718935012817,
      "learning_rate": 9.419921052631579e-05,
      "loss": 0.3648,
      "step": 42062
    },
    {
      "epoch": 2.355414940082876,
      "grad_norm": 1.008502721786499,
      "learning_rate": 9.419894736842105e-05,
      "loss": 0.3302,
      "step": 42063
    },
    {
      "epoch": 2.355470937395005,
      "grad_norm": 1.1072465181350708,
      "learning_rate": 9.419868421052633e-05,
      "loss": 0.3738,
      "step": 42064
    },
    {
      "epoch": 2.355526934707134,
      "grad_norm": 1.0874302387237549,
      "learning_rate": 9.419842105263158e-05,
      "loss": 0.4086,
      "step": 42065
    },
    {
      "epoch": 2.355582932019263,
      "grad_norm": 1.5220630168914795,
      "learning_rate": 9.419815789473684e-05,
      "loss": 0.6325,
      "step": 42066
    },
    {
      "epoch": 2.355638929331392,
      "grad_norm": 1.1535512208938599,
      "learning_rate": 9.41978947368421e-05,
      "loss": 0.32,
      "step": 42067
    },
    {
      "epoch": 2.355694926643521,
      "grad_norm": 1.1065874099731445,
      "learning_rate": 9.419763157894738e-05,
      "loss": 0.4099,
      "step": 42068
    },
    {
      "epoch": 2.3557509239556502,
      "grad_norm": 1.4733980894088745,
      "learning_rate": 9.419736842105264e-05,
      "loss": 0.471,
      "step": 42069
    },
    {
      "epoch": 2.3558069212677792,
      "grad_norm": 1.2396273612976074,
      "learning_rate": 9.41971052631579e-05,
      "loss": 0.4677,
      "step": 42070
    },
    {
      "epoch": 2.3558629185799083,
      "grad_norm": 1.0827510356903076,
      "learning_rate": 9.419684210526316e-05,
      "loss": 0.3787,
      "step": 42071
    },
    {
      "epoch": 2.3559189158920373,
      "grad_norm": 1.5058119297027588,
      "learning_rate": 9.419657894736843e-05,
      "loss": 0.4911,
      "step": 42072
    },
    {
      "epoch": 2.3559749132041663,
      "grad_norm": 1.165886640548706,
      "learning_rate": 9.419631578947369e-05,
      "loss": 0.3993,
      "step": 42073
    },
    {
      "epoch": 2.3560309105162953,
      "grad_norm": 1.0766451358795166,
      "learning_rate": 9.419605263157896e-05,
      "loss": 0.4162,
      "step": 42074
    },
    {
      "epoch": 2.3560869078284243,
      "grad_norm": 0.9640364050865173,
      "learning_rate": 9.419578947368421e-05,
      "loss": 0.2919,
      "step": 42075
    },
    {
      "epoch": 2.3561429051405534,
      "grad_norm": 1.1855136156082153,
      "learning_rate": 9.419552631578947e-05,
      "loss": 0.4452,
      "step": 42076
    },
    {
      "epoch": 2.3561989024526824,
      "grad_norm": 1.188442587852478,
      "learning_rate": 9.419526315789474e-05,
      "loss": 0.3969,
      "step": 42077
    },
    {
      "epoch": 2.3562548997648114,
      "grad_norm": 1.849393606185913,
      "learning_rate": 9.4195e-05,
      "loss": 0.5059,
      "step": 42078
    },
    {
      "epoch": 2.3563108970769404,
      "grad_norm": 1.0141866207122803,
      "learning_rate": 9.419473684210528e-05,
      "loss": 0.3143,
      "step": 42079
    },
    {
      "epoch": 2.3563668943890694,
      "grad_norm": 1.200398325920105,
      "learning_rate": 9.419447368421052e-05,
      "loss": 0.4128,
      "step": 42080
    },
    {
      "epoch": 2.3564228917011985,
      "grad_norm": 1.2005687952041626,
      "learning_rate": 9.41942105263158e-05,
      "loss": 0.3887,
      "step": 42081
    },
    {
      "epoch": 2.3564788890133275,
      "grad_norm": 1.1121059656143188,
      "learning_rate": 9.419394736842105e-05,
      "loss": 0.361,
      "step": 42082
    },
    {
      "epoch": 2.3565348863254565,
      "grad_norm": 1.4347631931304932,
      "learning_rate": 9.419368421052633e-05,
      "loss": 0.5222,
      "step": 42083
    },
    {
      "epoch": 2.3565908836375855,
      "grad_norm": 1.2278953790664673,
      "learning_rate": 9.419342105263159e-05,
      "loss": 0.3644,
      "step": 42084
    },
    {
      "epoch": 2.3566468809497145,
      "grad_norm": 2.2126948833465576,
      "learning_rate": 9.419315789473685e-05,
      "loss": 0.5783,
      "step": 42085
    },
    {
      "epoch": 2.3567028782618435,
      "grad_norm": 1.3801368474960327,
      "learning_rate": 9.419289473684211e-05,
      "loss": 0.5011,
      "step": 42086
    },
    {
      "epoch": 2.3567588755739726,
      "grad_norm": 1.2987030744552612,
      "learning_rate": 9.419263157894738e-05,
      "loss": 0.5352,
      "step": 42087
    },
    {
      "epoch": 2.3568148728861016,
      "grad_norm": 1.2596839666366577,
      "learning_rate": 9.419236842105264e-05,
      "loss": 0.477,
      "step": 42088
    },
    {
      "epoch": 2.3568708701982306,
      "grad_norm": 1.0862349271774292,
      "learning_rate": 9.41921052631579e-05,
      "loss": 0.2845,
      "step": 42089
    },
    {
      "epoch": 2.3569268675103596,
      "grad_norm": 1.1161876916885376,
      "learning_rate": 9.419184210526316e-05,
      "loss": 0.388,
      "step": 42090
    },
    {
      "epoch": 2.3569828648224886,
      "grad_norm": 1.650629997253418,
      "learning_rate": 9.419157894736843e-05,
      "loss": 0.3566,
      "step": 42091
    },
    {
      "epoch": 2.3570388621346177,
      "grad_norm": 1.850733995437622,
      "learning_rate": 9.419131578947369e-05,
      "loss": 0.4441,
      "step": 42092
    },
    {
      "epoch": 2.3570948594467467,
      "grad_norm": 2.5519347190856934,
      "learning_rate": 9.419105263157895e-05,
      "loss": 0.3978,
      "step": 42093
    },
    {
      "epoch": 2.3571508567588757,
      "grad_norm": 1.5656487941741943,
      "learning_rate": 9.419078947368421e-05,
      "loss": 0.7014,
      "step": 42094
    },
    {
      "epoch": 2.3572068540710047,
      "grad_norm": 1.715987205505371,
      "learning_rate": 9.419052631578947e-05,
      "loss": 0.3879,
      "step": 42095
    },
    {
      "epoch": 2.3572628513831337,
      "grad_norm": 1.540969729423523,
      "learning_rate": 9.419026315789474e-05,
      "loss": 0.4568,
      "step": 42096
    },
    {
      "epoch": 2.3573188486952628,
      "grad_norm": 1.50597083568573,
      "learning_rate": 9.419e-05,
      "loss": 0.3852,
      "step": 42097
    },
    {
      "epoch": 2.3573748460073918,
      "grad_norm": 1.5645270347595215,
      "learning_rate": 9.418973684210526e-05,
      "loss": 0.4804,
      "step": 42098
    },
    {
      "epoch": 2.357430843319521,
      "grad_norm": 1.2777581214904785,
      "learning_rate": 9.418947368421052e-05,
      "loss": 0.3347,
      "step": 42099
    },
    {
      "epoch": 2.35748684063165,
      "grad_norm": 1.2117984294891357,
      "learning_rate": 9.41892105263158e-05,
      "loss": 0.4275,
      "step": 42100
    },
    {
      "epoch": 2.357542837943779,
      "grad_norm": 1.1498421430587769,
      "learning_rate": 9.418894736842106e-05,
      "loss": 0.3176,
      "step": 42101
    },
    {
      "epoch": 2.357598835255908,
      "grad_norm": 1.6608003377914429,
      "learning_rate": 9.418868421052632e-05,
      "loss": 0.4514,
      "step": 42102
    },
    {
      "epoch": 2.357654832568037,
      "grad_norm": 1.249788522720337,
      "learning_rate": 9.418842105263158e-05,
      "loss": 0.4094,
      "step": 42103
    },
    {
      "epoch": 2.357710829880166,
      "grad_norm": 1.2765742540359497,
      "learning_rate": 9.418815789473685e-05,
      "loss": 0.3854,
      "step": 42104
    },
    {
      "epoch": 2.357766827192295,
      "grad_norm": 1.3121942281723022,
      "learning_rate": 9.418789473684211e-05,
      "loss": 0.4046,
      "step": 42105
    },
    {
      "epoch": 2.357822824504424,
      "grad_norm": 1.1774343252182007,
      "learning_rate": 9.418763157894738e-05,
      "loss": 0.4232,
      "step": 42106
    },
    {
      "epoch": 2.357878821816553,
      "grad_norm": 1.0485992431640625,
      "learning_rate": 9.418736842105263e-05,
      "loss": 0.3953,
      "step": 42107
    },
    {
      "epoch": 2.357934819128682,
      "grad_norm": 1.2635159492492676,
      "learning_rate": 9.41871052631579e-05,
      "loss": 0.4725,
      "step": 42108
    },
    {
      "epoch": 2.357990816440811,
      "grad_norm": 1.3943434953689575,
      "learning_rate": 9.418684210526316e-05,
      "loss": 0.4141,
      "step": 42109
    },
    {
      "epoch": 2.35804681375294,
      "grad_norm": 1.1756008863449097,
      "learning_rate": 9.418657894736842e-05,
      "loss": 0.42,
      "step": 42110
    },
    {
      "epoch": 2.358102811065069,
      "grad_norm": 0.9625444412231445,
      "learning_rate": 9.41863157894737e-05,
      "loss": 0.278,
      "step": 42111
    },
    {
      "epoch": 2.358158808377198,
      "grad_norm": 1.0112862586975098,
      "learning_rate": 9.418605263157894e-05,
      "loss": 0.3779,
      "step": 42112
    },
    {
      "epoch": 2.358214805689327,
      "grad_norm": 1.143189549446106,
      "learning_rate": 9.418578947368421e-05,
      "loss": 0.3644,
      "step": 42113
    },
    {
      "epoch": 2.358270803001456,
      "grad_norm": 1.2035561800003052,
      "learning_rate": 9.418552631578947e-05,
      "loss": 0.3764,
      "step": 42114
    },
    {
      "epoch": 2.358326800313585,
      "grad_norm": 61.36873245239258,
      "learning_rate": 9.418526315789475e-05,
      "loss": 0.4916,
      "step": 42115
    },
    {
      "epoch": 2.358382797625714,
      "grad_norm": 1.449543833732605,
      "learning_rate": 9.418500000000001e-05,
      "loss": 0.4049,
      "step": 42116
    },
    {
      "epoch": 2.358438794937843,
      "grad_norm": 1.035894513130188,
      "learning_rate": 9.418473684210527e-05,
      "loss": 0.3785,
      "step": 42117
    },
    {
      "epoch": 2.358494792249972,
      "grad_norm": 1.1107745170593262,
      "learning_rate": 9.418447368421053e-05,
      "loss": 0.3628,
      "step": 42118
    },
    {
      "epoch": 2.358550789562101,
      "grad_norm": 3.718616247177124,
      "learning_rate": 9.41842105263158e-05,
      "loss": 0.3685,
      "step": 42119
    },
    {
      "epoch": 2.35860678687423,
      "grad_norm": 1.372156023979187,
      "learning_rate": 9.418394736842106e-05,
      "loss": 0.2991,
      "step": 42120
    },
    {
      "epoch": 2.358662784186359,
      "grad_norm": 1.1128473281860352,
      "learning_rate": 9.418368421052632e-05,
      "loss": 0.3907,
      "step": 42121
    },
    {
      "epoch": 2.3587187814984882,
      "grad_norm": 1.4412513971328735,
      "learning_rate": 9.418342105263158e-05,
      "loss": 0.5189,
      "step": 42122
    },
    {
      "epoch": 2.3587747788106173,
      "grad_norm": 1.358426570892334,
      "learning_rate": 9.418315789473685e-05,
      "loss": 0.3482,
      "step": 42123
    },
    {
      "epoch": 2.3588307761227463,
      "grad_norm": 1.280336856842041,
      "learning_rate": 9.418289473684211e-05,
      "loss": 0.5239,
      "step": 42124
    },
    {
      "epoch": 2.3588867734348753,
      "grad_norm": 1.1203536987304688,
      "learning_rate": 9.418263157894737e-05,
      "loss": 0.3963,
      "step": 42125
    },
    {
      "epoch": 2.3589427707470043,
      "grad_norm": 1.1906802654266357,
      "learning_rate": 9.418236842105263e-05,
      "loss": 0.4178,
      "step": 42126
    },
    {
      "epoch": 2.3589987680591333,
      "grad_norm": 1.1729192733764648,
      "learning_rate": 9.418210526315789e-05,
      "loss": 0.4484,
      "step": 42127
    },
    {
      "epoch": 2.3590547653712624,
      "grad_norm": 1.3363909721374512,
      "learning_rate": 9.418184210526316e-05,
      "loss": 0.4531,
      "step": 42128
    },
    {
      "epoch": 2.3591107626833914,
      "grad_norm": 0.9300652146339417,
      "learning_rate": 9.418157894736842e-05,
      "loss": 0.3279,
      "step": 42129
    },
    {
      "epoch": 2.3591667599955204,
      "grad_norm": 1.2821087837219238,
      "learning_rate": 9.418131578947368e-05,
      "loss": 0.4305,
      "step": 42130
    },
    {
      "epoch": 2.3592227573076494,
      "grad_norm": 1.1366273164749146,
      "learning_rate": 9.418105263157894e-05,
      "loss": 0.4396,
      "step": 42131
    },
    {
      "epoch": 2.3592787546197784,
      "grad_norm": 1.0845974683761597,
      "learning_rate": 9.418078947368422e-05,
      "loss": 0.3359,
      "step": 42132
    },
    {
      "epoch": 2.3593347519319074,
      "grad_norm": 1.292511224746704,
      "learning_rate": 9.418052631578948e-05,
      "loss": 0.4877,
      "step": 42133
    },
    {
      "epoch": 2.3593907492440365,
      "grad_norm": 1.1286345720291138,
      "learning_rate": 9.418026315789475e-05,
      "loss": 0.3599,
      "step": 42134
    },
    {
      "epoch": 2.3594467465561655,
      "grad_norm": 1.0115923881530762,
      "learning_rate": 9.418e-05,
      "loss": 0.38,
      "step": 42135
    },
    {
      "epoch": 2.3595027438682945,
      "grad_norm": 1.2855521440505981,
      "learning_rate": 9.417973684210527e-05,
      "loss": 0.3605,
      "step": 42136
    },
    {
      "epoch": 2.3595587411804235,
      "grad_norm": 1.0307698249816895,
      "learning_rate": 9.417947368421053e-05,
      "loss": 0.3303,
      "step": 42137
    },
    {
      "epoch": 2.3596147384925525,
      "grad_norm": 1.32102370262146,
      "learning_rate": 9.41792105263158e-05,
      "loss": 0.4074,
      "step": 42138
    },
    {
      "epoch": 2.3596707358046816,
      "grad_norm": 1.0507911443710327,
      "learning_rate": 9.417894736842106e-05,
      "loss": 0.3955,
      "step": 42139
    },
    {
      "epoch": 2.3597267331168106,
      "grad_norm": 1.2132923603057861,
      "learning_rate": 9.417868421052632e-05,
      "loss": 0.3623,
      "step": 42140
    },
    {
      "epoch": 2.3597827304289396,
      "grad_norm": 1.2888809442520142,
      "learning_rate": 9.417842105263158e-05,
      "loss": 0.3651,
      "step": 42141
    },
    {
      "epoch": 2.3598387277410686,
      "grad_norm": 1.3733798265457153,
      "learning_rate": 9.417815789473685e-05,
      "loss": 0.5268,
      "step": 42142
    },
    {
      "epoch": 2.3598947250531976,
      "grad_norm": 1.2760508060455322,
      "learning_rate": 9.417789473684211e-05,
      "loss": 0.4916,
      "step": 42143
    },
    {
      "epoch": 2.3599507223653267,
      "grad_norm": 1.8532829284667969,
      "learning_rate": 9.417763157894737e-05,
      "loss": 0.5818,
      "step": 42144
    },
    {
      "epoch": 2.3600067196774557,
      "grad_norm": 1.2099051475524902,
      "learning_rate": 9.417736842105263e-05,
      "loss": 0.4142,
      "step": 42145
    },
    {
      "epoch": 2.3600627169895843,
      "grad_norm": 0.9829007387161255,
      "learning_rate": 9.41771052631579e-05,
      "loss": 0.3263,
      "step": 42146
    },
    {
      "epoch": 2.3601187143017137,
      "grad_norm": 1.2856959104537964,
      "learning_rate": 9.417684210526317e-05,
      "loss": 0.4431,
      "step": 42147
    },
    {
      "epoch": 2.3601747116138423,
      "grad_norm": 1.0858981609344482,
      "learning_rate": 9.417657894736843e-05,
      "loss": 0.4336,
      "step": 42148
    },
    {
      "epoch": 2.3602307089259718,
      "grad_norm": 1.0360996723175049,
      "learning_rate": 9.417631578947369e-05,
      "loss": 0.3676,
      "step": 42149
    },
    {
      "epoch": 2.3602867062381003,
      "grad_norm": 1.082420825958252,
      "learning_rate": 9.417605263157895e-05,
      "loss": 0.3381,
      "step": 42150
    },
    {
      "epoch": 2.36034270355023,
      "grad_norm": 1.4514939785003662,
      "learning_rate": 9.417578947368422e-05,
      "loss": 0.4968,
      "step": 42151
    },
    {
      "epoch": 2.3603987008623584,
      "grad_norm": 3.6139161586761475,
      "learning_rate": 9.417552631578948e-05,
      "loss": 0.4063,
      "step": 42152
    },
    {
      "epoch": 2.360454698174488,
      "grad_norm": 1.0487850904464722,
      "learning_rate": 9.417526315789474e-05,
      "loss": 0.441,
      "step": 42153
    },
    {
      "epoch": 2.3605106954866164,
      "grad_norm": 1.2696326971054077,
      "learning_rate": 9.4175e-05,
      "loss": 0.3713,
      "step": 42154
    },
    {
      "epoch": 2.360566692798746,
      "grad_norm": 1.152755618095398,
      "learning_rate": 9.417473684210527e-05,
      "loss": 0.3529,
      "step": 42155
    },
    {
      "epoch": 2.3606226901108744,
      "grad_norm": 1.2065237760543823,
      "learning_rate": 9.417447368421053e-05,
      "loss": 0.4083,
      "step": 42156
    },
    {
      "epoch": 2.360678687423004,
      "grad_norm": 1.2593610286712646,
      "learning_rate": 9.417421052631579e-05,
      "loss": 0.4469,
      "step": 42157
    },
    {
      "epoch": 2.3607346847351325,
      "grad_norm": 1.5991078615188599,
      "learning_rate": 9.417394736842105e-05,
      "loss": 0.7718,
      "step": 42158
    },
    {
      "epoch": 2.360790682047262,
      "grad_norm": 1.1831023693084717,
      "learning_rate": 9.417368421052632e-05,
      "loss": 0.4062,
      "step": 42159
    },
    {
      "epoch": 2.3608466793593905,
      "grad_norm": 1.384891390800476,
      "learning_rate": 9.417342105263158e-05,
      "loss": 0.4493,
      "step": 42160
    },
    {
      "epoch": 2.36090267667152,
      "grad_norm": 1.5018233060836792,
      "learning_rate": 9.417315789473686e-05,
      "loss": 0.4251,
      "step": 42161
    },
    {
      "epoch": 2.3609586739836486,
      "grad_norm": 1.1330299377441406,
      "learning_rate": 9.41728947368421e-05,
      "loss": 0.3526,
      "step": 42162
    },
    {
      "epoch": 2.361014671295778,
      "grad_norm": 1.162330150604248,
      "learning_rate": 9.417263157894736e-05,
      "loss": 0.3637,
      "step": 42163
    },
    {
      "epoch": 2.3610706686079066,
      "grad_norm": 1.1777554750442505,
      "learning_rate": 9.417236842105264e-05,
      "loss": 0.5096,
      "step": 42164
    },
    {
      "epoch": 2.361126665920036,
      "grad_norm": 1.178244948387146,
      "learning_rate": 9.41721052631579e-05,
      "loss": 0.4498,
      "step": 42165
    },
    {
      "epoch": 2.3611826632321646,
      "grad_norm": 1.3178316354751587,
      "learning_rate": 9.417184210526317e-05,
      "loss": 0.3688,
      "step": 42166
    },
    {
      "epoch": 2.361238660544294,
      "grad_norm": 1.0979162454605103,
      "learning_rate": 9.417157894736842e-05,
      "loss": 0.3657,
      "step": 42167
    },
    {
      "epoch": 2.3612946578564227,
      "grad_norm": 1.0116496086120605,
      "learning_rate": 9.417131578947369e-05,
      "loss": 0.3186,
      "step": 42168
    },
    {
      "epoch": 2.361350655168552,
      "grad_norm": 1.1470129489898682,
      "learning_rate": 9.417105263157895e-05,
      "loss": 0.3409,
      "step": 42169
    },
    {
      "epoch": 2.3614066524806807,
      "grad_norm": 1.3460580110549927,
      "learning_rate": 9.417078947368422e-05,
      "loss": 0.4685,
      "step": 42170
    },
    {
      "epoch": 2.36146264979281,
      "grad_norm": 1.2768175601959229,
      "learning_rate": 9.417052631578948e-05,
      "loss": 0.4529,
      "step": 42171
    },
    {
      "epoch": 2.3615186471049388,
      "grad_norm": 1.029980182647705,
      "learning_rate": 9.417026315789474e-05,
      "loss": 0.3452,
      "step": 42172
    },
    {
      "epoch": 2.361574644417068,
      "grad_norm": 1.3561220169067383,
      "learning_rate": 9.417e-05,
      "loss": 0.4805,
      "step": 42173
    },
    {
      "epoch": 2.361630641729197,
      "grad_norm": 1.126116156578064,
      "learning_rate": 9.416973684210527e-05,
      "loss": 0.2943,
      "step": 42174
    },
    {
      "epoch": 2.3616866390413263,
      "grad_norm": 1.1219772100448608,
      "learning_rate": 9.416947368421053e-05,
      "loss": 0.376,
      "step": 42175
    },
    {
      "epoch": 2.361742636353455,
      "grad_norm": 1.1999244689941406,
      "learning_rate": 9.41692105263158e-05,
      "loss": 0.3814,
      "step": 42176
    },
    {
      "epoch": 2.3617986336655843,
      "grad_norm": 1.2493246793746948,
      "learning_rate": 9.416894736842105e-05,
      "loss": 0.3363,
      "step": 42177
    },
    {
      "epoch": 2.361854630977713,
      "grad_norm": 1.4358148574829102,
      "learning_rate": 9.416868421052633e-05,
      "loss": 0.3995,
      "step": 42178
    },
    {
      "epoch": 2.3619106282898423,
      "grad_norm": 1.453559398651123,
      "learning_rate": 9.416842105263159e-05,
      "loss": 0.5668,
      "step": 42179
    },
    {
      "epoch": 2.361966625601971,
      "grad_norm": 1.3580994606018066,
      "learning_rate": 9.416815789473685e-05,
      "loss": 0.5791,
      "step": 42180
    },
    {
      "epoch": 2.3620226229141004,
      "grad_norm": 1.1576915979385376,
      "learning_rate": 9.41678947368421e-05,
      "loss": 0.3847,
      "step": 42181
    },
    {
      "epoch": 2.362078620226229,
      "grad_norm": 1.1490558385849,
      "learning_rate": 9.416763157894737e-05,
      "loss": 0.6186,
      "step": 42182
    },
    {
      "epoch": 2.3621346175383584,
      "grad_norm": 1.0884122848510742,
      "learning_rate": 9.416736842105264e-05,
      "loss": 0.2773,
      "step": 42183
    },
    {
      "epoch": 2.362190614850487,
      "grad_norm": 1.2221380472183228,
      "learning_rate": 9.41671052631579e-05,
      "loss": 0.3557,
      "step": 42184
    },
    {
      "epoch": 2.3622466121626164,
      "grad_norm": 1.364810824394226,
      "learning_rate": 9.416684210526316e-05,
      "loss": 0.4715,
      "step": 42185
    },
    {
      "epoch": 2.362302609474745,
      "grad_norm": 1.260488510131836,
      "learning_rate": 9.416657894736842e-05,
      "loss": 0.594,
      "step": 42186
    },
    {
      "epoch": 2.362358606786874,
      "grad_norm": 1.143228530883789,
      "learning_rate": 9.416631578947369e-05,
      "loss": 0.5061,
      "step": 42187
    },
    {
      "epoch": 2.362414604099003,
      "grad_norm": 1.1019113063812256,
      "learning_rate": 9.416605263157895e-05,
      "loss": 0.3515,
      "step": 42188
    },
    {
      "epoch": 2.362470601411132,
      "grad_norm": 1.1494735479354858,
      "learning_rate": 9.416578947368422e-05,
      "loss": 0.3531,
      "step": 42189
    },
    {
      "epoch": 2.362526598723261,
      "grad_norm": 1.0774478912353516,
      "learning_rate": 9.416552631578947e-05,
      "loss": 0.3085,
      "step": 42190
    },
    {
      "epoch": 2.36258259603539,
      "grad_norm": 1.184244155883789,
      "learning_rate": 9.416526315789474e-05,
      "loss": 0.4209,
      "step": 42191
    },
    {
      "epoch": 2.362638593347519,
      "grad_norm": 1.2900294065475464,
      "learning_rate": 9.4165e-05,
      "loss": 0.5229,
      "step": 42192
    },
    {
      "epoch": 2.362694590659648,
      "grad_norm": 0.9700883030891418,
      "learning_rate": 9.416473684210528e-05,
      "loss": 0.3744,
      "step": 42193
    },
    {
      "epoch": 2.362750587971777,
      "grad_norm": 1.0989584922790527,
      "learning_rate": 9.416447368421054e-05,
      "loss": 0.433,
      "step": 42194
    },
    {
      "epoch": 2.362806585283906,
      "grad_norm": 1.1285914182662964,
      "learning_rate": 9.41642105263158e-05,
      "loss": 0.3869,
      "step": 42195
    },
    {
      "epoch": 2.362862582596035,
      "grad_norm": 1.1750162839889526,
      "learning_rate": 9.416394736842106e-05,
      "loss": 0.3492,
      "step": 42196
    },
    {
      "epoch": 2.3629185799081642,
      "grad_norm": 1.3149181604385376,
      "learning_rate": 9.416368421052632e-05,
      "loss": 0.4276,
      "step": 42197
    },
    {
      "epoch": 2.3629745772202932,
      "grad_norm": 1.1668169498443604,
      "learning_rate": 9.416342105263159e-05,
      "loss": 0.3422,
      "step": 42198
    },
    {
      "epoch": 2.3630305745324223,
      "grad_norm": 1.0885361433029175,
      "learning_rate": 9.416315789473684e-05,
      "loss": 0.3907,
      "step": 42199
    },
    {
      "epoch": 2.3630865718445513,
      "grad_norm": 0.9846569895744324,
      "learning_rate": 9.416289473684211e-05,
      "loss": 0.3883,
      "step": 42200
    },
    {
      "epoch": 2.3631425691566803,
      "grad_norm": 1.1640223264694214,
      "learning_rate": 9.416263157894737e-05,
      "loss": 0.4835,
      "step": 42201
    },
    {
      "epoch": 2.3631985664688093,
      "grad_norm": 1.1281291246414185,
      "learning_rate": 9.416236842105264e-05,
      "loss": 0.4183,
      "step": 42202
    },
    {
      "epoch": 2.3632545637809383,
      "grad_norm": 1.3642750978469849,
      "learning_rate": 9.41621052631579e-05,
      "loss": 0.4044,
      "step": 42203
    },
    {
      "epoch": 2.3633105610930674,
      "grad_norm": 1.5359916687011719,
      "learning_rate": 9.416184210526316e-05,
      "loss": 0.421,
      "step": 42204
    },
    {
      "epoch": 2.3633665584051964,
      "grad_norm": 1.3497382402420044,
      "learning_rate": 9.416157894736842e-05,
      "loss": 0.4016,
      "step": 42205
    },
    {
      "epoch": 2.3634225557173254,
      "grad_norm": 1.1716583967208862,
      "learning_rate": 9.41613157894737e-05,
      "loss": 0.339,
      "step": 42206
    },
    {
      "epoch": 2.3634785530294544,
      "grad_norm": 1.4831912517547607,
      "learning_rate": 9.416105263157895e-05,
      "loss": 0.3675,
      "step": 42207
    },
    {
      "epoch": 2.3635345503415834,
      "grad_norm": 1.106115460395813,
      "learning_rate": 9.416078947368421e-05,
      "loss": 0.3889,
      "step": 42208
    },
    {
      "epoch": 2.3635905476537125,
      "grad_norm": 1.136399745941162,
      "learning_rate": 9.416052631578947e-05,
      "loss": 0.5022,
      "step": 42209
    },
    {
      "epoch": 2.3636465449658415,
      "grad_norm": 1.2835071086883545,
      "learning_rate": 9.416026315789475e-05,
      "loss": 0.4914,
      "step": 42210
    },
    {
      "epoch": 2.3637025422779705,
      "grad_norm": 1.346935510635376,
      "learning_rate": 9.416e-05,
      "loss": 0.4141,
      "step": 42211
    },
    {
      "epoch": 2.3637585395900995,
      "grad_norm": 1.3580408096313477,
      "learning_rate": 9.415973684210527e-05,
      "loss": 0.3836,
      "step": 42212
    },
    {
      "epoch": 2.3638145369022285,
      "grad_norm": 1.3355134725570679,
      "learning_rate": 9.415947368421053e-05,
      "loss": 0.3028,
      "step": 42213
    },
    {
      "epoch": 2.3638705342143576,
      "grad_norm": 1.0368287563323975,
      "learning_rate": 9.415921052631579e-05,
      "loss": 0.3456,
      "step": 42214
    },
    {
      "epoch": 2.3639265315264866,
      "grad_norm": 1.2260124683380127,
      "learning_rate": 9.415894736842106e-05,
      "loss": 0.3136,
      "step": 42215
    },
    {
      "epoch": 2.3639825288386156,
      "grad_norm": 1.2879105806350708,
      "learning_rate": 9.415868421052632e-05,
      "loss": 0.3998,
      "step": 42216
    },
    {
      "epoch": 2.3640385261507446,
      "grad_norm": 1.2197303771972656,
      "learning_rate": 9.415842105263158e-05,
      "loss": 0.5165,
      "step": 42217
    },
    {
      "epoch": 2.3640945234628736,
      "grad_norm": 1.231621503829956,
      "learning_rate": 9.415815789473684e-05,
      "loss": 0.4964,
      "step": 42218
    },
    {
      "epoch": 2.3641505207750027,
      "grad_norm": 1.1322216987609863,
      "learning_rate": 9.415789473684211e-05,
      "loss": 0.3759,
      "step": 42219
    },
    {
      "epoch": 2.3642065180871317,
      "grad_norm": 1.1338282823562622,
      "learning_rate": 9.415763157894737e-05,
      "loss": 0.3694,
      "step": 42220
    },
    {
      "epoch": 2.3642625153992607,
      "grad_norm": 1.3787081241607666,
      "learning_rate": 9.415736842105264e-05,
      "loss": 0.4801,
      "step": 42221
    },
    {
      "epoch": 2.3643185127113897,
      "grad_norm": 1.359574794769287,
      "learning_rate": 9.415710526315789e-05,
      "loss": 0.4804,
      "step": 42222
    },
    {
      "epoch": 2.3643745100235187,
      "grad_norm": 1.0256779193878174,
      "learning_rate": 9.415684210526316e-05,
      "loss": 0.318,
      "step": 42223
    },
    {
      "epoch": 2.3644305073356477,
      "grad_norm": 1.412192702293396,
      "learning_rate": 9.415657894736842e-05,
      "loss": 0.3921,
      "step": 42224
    },
    {
      "epoch": 2.3644865046477768,
      "grad_norm": 1.0518834590911865,
      "learning_rate": 9.41563157894737e-05,
      "loss": 0.5457,
      "step": 42225
    },
    {
      "epoch": 2.364542501959906,
      "grad_norm": 1.1536314487457275,
      "learning_rate": 9.415605263157896e-05,
      "loss": 0.363,
      "step": 42226
    },
    {
      "epoch": 2.364598499272035,
      "grad_norm": 2.0204622745513916,
      "learning_rate": 9.415578947368422e-05,
      "loss": 0.5185,
      "step": 42227
    },
    {
      "epoch": 2.364654496584164,
      "grad_norm": 1.1709697246551514,
      "learning_rate": 9.415552631578948e-05,
      "loss": 0.392,
      "step": 42228
    },
    {
      "epoch": 2.364710493896293,
      "grad_norm": 1.1572654247283936,
      "learning_rate": 9.415526315789475e-05,
      "loss": 0.4597,
      "step": 42229
    },
    {
      "epoch": 2.364766491208422,
      "grad_norm": 1.1046416759490967,
      "learning_rate": 9.415500000000001e-05,
      "loss": 0.3784,
      "step": 42230
    },
    {
      "epoch": 2.364822488520551,
      "grad_norm": 1.4480878114700317,
      "learning_rate": 9.415473684210527e-05,
      "loss": 0.399,
      "step": 42231
    },
    {
      "epoch": 2.36487848583268,
      "grad_norm": 1.3057136535644531,
      "learning_rate": 9.415447368421053e-05,
      "loss": 0.5041,
      "step": 42232
    },
    {
      "epoch": 2.364934483144809,
      "grad_norm": 1.4253910779953003,
      "learning_rate": 9.415421052631579e-05,
      "loss": 0.5961,
      "step": 42233
    },
    {
      "epoch": 2.364990480456938,
      "grad_norm": 1.3882946968078613,
      "learning_rate": 9.415394736842106e-05,
      "loss": 0.5922,
      "step": 42234
    },
    {
      "epoch": 2.365046477769067,
      "grad_norm": 1.0331233739852905,
      "learning_rate": 9.415368421052632e-05,
      "loss": 0.3107,
      "step": 42235
    },
    {
      "epoch": 2.365102475081196,
      "grad_norm": 1.6579644680023193,
      "learning_rate": 9.415342105263158e-05,
      "loss": 0.4178,
      "step": 42236
    },
    {
      "epoch": 2.365158472393325,
      "grad_norm": 1.7633724212646484,
      "learning_rate": 9.415315789473684e-05,
      "loss": 0.4294,
      "step": 42237
    },
    {
      "epoch": 2.365214469705454,
      "grad_norm": 0.9841364622116089,
      "learning_rate": 9.415289473684211e-05,
      "loss": 0.3487,
      "step": 42238
    },
    {
      "epoch": 2.365270467017583,
      "grad_norm": 1.0259346961975098,
      "learning_rate": 9.415263157894737e-05,
      "loss": 0.332,
      "step": 42239
    },
    {
      "epoch": 2.365326464329712,
      "grad_norm": 1.0068045854568481,
      "learning_rate": 9.415236842105263e-05,
      "loss": 0.3138,
      "step": 42240
    },
    {
      "epoch": 2.365382461641841,
      "grad_norm": 1.7210932970046997,
      "learning_rate": 9.415210526315789e-05,
      "loss": 0.3268,
      "step": 42241
    },
    {
      "epoch": 2.36543845895397,
      "grad_norm": 1.2664185762405396,
      "learning_rate": 9.415184210526317e-05,
      "loss": 0.3914,
      "step": 42242
    },
    {
      "epoch": 2.365494456266099,
      "grad_norm": 1.252374529838562,
      "learning_rate": 9.415157894736843e-05,
      "loss": 0.4191,
      "step": 42243
    },
    {
      "epoch": 2.365550453578228,
      "grad_norm": 1.5683587789535522,
      "learning_rate": 9.41513157894737e-05,
      "loss": 0.4306,
      "step": 42244
    },
    {
      "epoch": 2.365606450890357,
      "grad_norm": 1.137418270111084,
      "learning_rate": 9.415105263157895e-05,
      "loss": 0.4835,
      "step": 42245
    },
    {
      "epoch": 2.365662448202486,
      "grad_norm": 1.2689480781555176,
      "learning_rate": 9.415078947368422e-05,
      "loss": 0.4379,
      "step": 42246
    },
    {
      "epoch": 2.365718445514615,
      "grad_norm": 1.120169758796692,
      "learning_rate": 9.415052631578948e-05,
      "loss": 0.344,
      "step": 42247
    },
    {
      "epoch": 2.365774442826744,
      "grad_norm": 1.237959861755371,
      "learning_rate": 9.415026315789474e-05,
      "loss": 0.3245,
      "step": 42248
    },
    {
      "epoch": 2.3658304401388732,
      "grad_norm": 1.3555324077606201,
      "learning_rate": 9.415e-05,
      "loss": 0.3726,
      "step": 42249
    },
    {
      "epoch": 2.3658864374510022,
      "grad_norm": 1.4046891927719116,
      "learning_rate": 9.414973684210526e-05,
      "loss": 0.5583,
      "step": 42250
    },
    {
      "epoch": 2.3659424347631313,
      "grad_norm": 1.1015177965164185,
      "learning_rate": 9.414947368421053e-05,
      "loss": 0.373,
      "step": 42251
    },
    {
      "epoch": 2.3659984320752603,
      "grad_norm": 1.1048450469970703,
      "learning_rate": 9.414921052631579e-05,
      "loss": 0.3414,
      "step": 42252
    },
    {
      "epoch": 2.3660544293873893,
      "grad_norm": 1.351989507675171,
      "learning_rate": 9.414894736842106e-05,
      "loss": 0.4545,
      "step": 42253
    },
    {
      "epoch": 2.3661104266995183,
      "grad_norm": 1.4576549530029297,
      "learning_rate": 9.414868421052631e-05,
      "loss": 0.6975,
      "step": 42254
    },
    {
      "epoch": 2.3661664240116473,
      "grad_norm": 1.3782958984375,
      "learning_rate": 9.414842105263158e-05,
      "loss": 0.4966,
      "step": 42255
    },
    {
      "epoch": 2.3662224213237764,
      "grad_norm": 1.2241895198822021,
      "learning_rate": 9.414815789473684e-05,
      "loss": 0.5219,
      "step": 42256
    },
    {
      "epoch": 2.3662784186359054,
      "grad_norm": 1.0890096426010132,
      "learning_rate": 9.414789473684212e-05,
      "loss": 0.3904,
      "step": 42257
    },
    {
      "epoch": 2.3663344159480344,
      "grad_norm": 1.2898129224777222,
      "learning_rate": 9.414763157894738e-05,
      "loss": 0.4234,
      "step": 42258
    },
    {
      "epoch": 2.3663904132601634,
      "grad_norm": 1.3337053060531616,
      "learning_rate": 9.414736842105264e-05,
      "loss": 0.5256,
      "step": 42259
    },
    {
      "epoch": 2.3664464105722924,
      "grad_norm": 1.0534831285476685,
      "learning_rate": 9.41471052631579e-05,
      "loss": 0.39,
      "step": 42260
    },
    {
      "epoch": 2.3665024078844215,
      "grad_norm": 1.1585514545440674,
      "learning_rate": 9.414684210526317e-05,
      "loss": 0.51,
      "step": 42261
    },
    {
      "epoch": 2.3665584051965505,
      "grad_norm": 1.321932077407837,
      "learning_rate": 9.414657894736843e-05,
      "loss": 0.5089,
      "step": 42262
    },
    {
      "epoch": 2.3666144025086795,
      "grad_norm": 1.4394193887710571,
      "learning_rate": 9.414631578947369e-05,
      "loss": 0.4224,
      "step": 42263
    },
    {
      "epoch": 2.3666703998208085,
      "grad_norm": 1.437076210975647,
      "learning_rate": 9.414605263157895e-05,
      "loss": 0.7174,
      "step": 42264
    },
    {
      "epoch": 2.3667263971329375,
      "grad_norm": 1.1430895328521729,
      "learning_rate": 9.414578947368422e-05,
      "loss": 0.4302,
      "step": 42265
    },
    {
      "epoch": 2.3667823944450666,
      "grad_norm": 1.5252368450164795,
      "learning_rate": 9.414552631578948e-05,
      "loss": 0.3615,
      "step": 42266
    },
    {
      "epoch": 2.3668383917571956,
      "grad_norm": 1.2454910278320312,
      "learning_rate": 9.414526315789474e-05,
      "loss": 0.3977,
      "step": 42267
    },
    {
      "epoch": 2.3668943890693246,
      "grad_norm": 1.3482586145401,
      "learning_rate": 9.4145e-05,
      "loss": 0.4887,
      "step": 42268
    },
    {
      "epoch": 2.3669503863814536,
      "grad_norm": 1.1646373271942139,
      "learning_rate": 9.414473684210526e-05,
      "loss": 0.3855,
      "step": 42269
    },
    {
      "epoch": 2.3670063836935826,
      "grad_norm": 1.4197959899902344,
      "learning_rate": 9.414447368421053e-05,
      "loss": 0.4413,
      "step": 42270
    },
    {
      "epoch": 2.3670623810057116,
      "grad_norm": 1.0230342149734497,
      "learning_rate": 9.41442105263158e-05,
      "loss": 0.365,
      "step": 42271
    },
    {
      "epoch": 2.3671183783178407,
      "grad_norm": 1.1333682537078857,
      "learning_rate": 9.414394736842105e-05,
      "loss": 0.385,
      "step": 42272
    },
    {
      "epoch": 2.3671743756299697,
      "grad_norm": 1.3012174367904663,
      "learning_rate": 9.414368421052631e-05,
      "loss": 0.4023,
      "step": 42273
    },
    {
      "epoch": 2.3672303729420987,
      "grad_norm": 1.4104245901107788,
      "learning_rate": 9.414342105263159e-05,
      "loss": 0.398,
      "step": 42274
    },
    {
      "epoch": 2.3672863702542277,
      "grad_norm": 1.1881835460662842,
      "learning_rate": 9.414315789473685e-05,
      "loss": 0.4203,
      "step": 42275
    },
    {
      "epoch": 2.3673423675663567,
      "grad_norm": 1.6257612705230713,
      "learning_rate": 9.414289473684212e-05,
      "loss": 0.5006,
      "step": 42276
    },
    {
      "epoch": 2.3673983648784858,
      "grad_norm": 1.0663199424743652,
      "learning_rate": 9.414263157894737e-05,
      "loss": 0.3501,
      "step": 42277
    },
    {
      "epoch": 2.367454362190615,
      "grad_norm": 1.2903213500976562,
      "learning_rate": 9.414236842105264e-05,
      "loss": 0.4218,
      "step": 42278
    },
    {
      "epoch": 2.367510359502744,
      "grad_norm": 1.1001280546188354,
      "learning_rate": 9.41421052631579e-05,
      "loss": 0.3916,
      "step": 42279
    },
    {
      "epoch": 2.367566356814873,
      "grad_norm": 1.3239824771881104,
      "learning_rate": 9.414184210526317e-05,
      "loss": 0.3983,
      "step": 42280
    },
    {
      "epoch": 2.367622354127002,
      "grad_norm": 1.1877532005310059,
      "learning_rate": 9.414157894736843e-05,
      "loss": 0.4268,
      "step": 42281
    },
    {
      "epoch": 2.367678351439131,
      "grad_norm": 1.2388200759887695,
      "learning_rate": 9.414131578947369e-05,
      "loss": 0.3759,
      "step": 42282
    },
    {
      "epoch": 2.36773434875126,
      "grad_norm": 1.0127414464950562,
      "learning_rate": 9.414105263157895e-05,
      "loss": 0.4168,
      "step": 42283
    },
    {
      "epoch": 2.367790346063389,
      "grad_norm": 1.015407681465149,
      "learning_rate": 9.414078947368421e-05,
      "loss": 0.3612,
      "step": 42284
    },
    {
      "epoch": 2.367846343375518,
      "grad_norm": 1.247032642364502,
      "learning_rate": 9.414052631578948e-05,
      "loss": 0.4044,
      "step": 42285
    },
    {
      "epoch": 2.367902340687647,
      "grad_norm": 1.3325552940368652,
      "learning_rate": 9.414026315789474e-05,
      "loss": 0.3992,
      "step": 42286
    },
    {
      "epoch": 2.367958337999776,
      "grad_norm": 1.2160544395446777,
      "learning_rate": 9.414e-05,
      "loss": 0.4365,
      "step": 42287
    },
    {
      "epoch": 2.368014335311905,
      "grad_norm": 1.3640742301940918,
      "learning_rate": 9.413973684210526e-05,
      "loss": 0.5011,
      "step": 42288
    },
    {
      "epoch": 2.368070332624034,
      "grad_norm": 1.4170483350753784,
      "learning_rate": 9.413947368421054e-05,
      "loss": 0.2931,
      "step": 42289
    },
    {
      "epoch": 2.368126329936163,
      "grad_norm": 1.3146421909332275,
      "learning_rate": 9.41392105263158e-05,
      "loss": 0.4019,
      "step": 42290
    },
    {
      "epoch": 2.368182327248292,
      "grad_norm": 1.4473234415054321,
      "learning_rate": 9.413894736842106e-05,
      "loss": 0.4005,
      "step": 42291
    },
    {
      "epoch": 2.368238324560421,
      "grad_norm": 1.0588699579238892,
      "learning_rate": 9.413868421052632e-05,
      "loss": 0.3635,
      "step": 42292
    },
    {
      "epoch": 2.36829432187255,
      "grad_norm": 1.165635585784912,
      "learning_rate": 9.413842105263159e-05,
      "loss": 0.3652,
      "step": 42293
    },
    {
      "epoch": 2.368350319184679,
      "grad_norm": 1.156206727027893,
      "learning_rate": 9.413815789473685e-05,
      "loss": 0.3808,
      "step": 42294
    },
    {
      "epoch": 2.368406316496808,
      "grad_norm": 1.1734799146652222,
      "learning_rate": 9.413789473684211e-05,
      "loss": 0.4195,
      "step": 42295
    },
    {
      "epoch": 2.368462313808937,
      "grad_norm": 1.3753827810287476,
      "learning_rate": 9.413763157894737e-05,
      "loss": 0.2852,
      "step": 42296
    },
    {
      "epoch": 2.368518311121066,
      "grad_norm": 1.1885277032852173,
      "learning_rate": 9.413736842105264e-05,
      "loss": 0.4409,
      "step": 42297
    },
    {
      "epoch": 2.368574308433195,
      "grad_norm": 1.0572854280471802,
      "learning_rate": 9.41371052631579e-05,
      "loss": 0.3468,
      "step": 42298
    },
    {
      "epoch": 2.368630305745324,
      "grad_norm": 1.1223461627960205,
      "learning_rate": 9.413684210526317e-05,
      "loss": 0.3509,
      "step": 42299
    },
    {
      "epoch": 2.368686303057453,
      "grad_norm": 1.2320494651794434,
      "learning_rate": 9.413657894736842e-05,
      "loss": 0.3432,
      "step": 42300
    },
    {
      "epoch": 2.368742300369582,
      "grad_norm": 1.1564735174179077,
      "learning_rate": 9.413631578947368e-05,
      "loss": 0.3376,
      "step": 42301
    },
    {
      "epoch": 2.3687982976817112,
      "grad_norm": 1.2554796934127808,
      "learning_rate": 9.413605263157895e-05,
      "loss": 0.4869,
      "step": 42302
    },
    {
      "epoch": 2.3688542949938403,
      "grad_norm": 1.1750661134719849,
      "learning_rate": 9.413578947368421e-05,
      "loss": 0.5216,
      "step": 42303
    },
    {
      "epoch": 2.3689102923059693,
      "grad_norm": 1.294796109199524,
      "learning_rate": 9.413552631578947e-05,
      "loss": 0.4229,
      "step": 42304
    },
    {
      "epoch": 2.3689662896180983,
      "grad_norm": 1.187984824180603,
      "learning_rate": 9.413526315789473e-05,
      "loss": 0.3875,
      "step": 42305
    },
    {
      "epoch": 2.3690222869302273,
      "grad_norm": 1.3266453742980957,
      "learning_rate": 9.4135e-05,
      "loss": 0.3631,
      "step": 42306
    },
    {
      "epoch": 2.3690782842423563,
      "grad_norm": 1.2450937032699585,
      "learning_rate": 9.413473684210527e-05,
      "loss": 0.3293,
      "step": 42307
    },
    {
      "epoch": 2.3691342815544854,
      "grad_norm": 1.4017856121063232,
      "learning_rate": 9.413447368421054e-05,
      "loss": 0.4375,
      "step": 42308
    },
    {
      "epoch": 2.3691902788666144,
      "grad_norm": 1.0897020101547241,
      "learning_rate": 9.413421052631579e-05,
      "loss": 0.3458,
      "step": 42309
    },
    {
      "epoch": 2.3692462761787434,
      "grad_norm": 1.0226852893829346,
      "learning_rate": 9.413394736842106e-05,
      "loss": 0.3255,
      "step": 42310
    },
    {
      "epoch": 2.3693022734908724,
      "grad_norm": 1.2814875841140747,
      "learning_rate": 9.413368421052632e-05,
      "loss": 0.5398,
      "step": 42311
    },
    {
      "epoch": 2.3693582708030014,
      "grad_norm": 1.5108939409255981,
      "learning_rate": 9.413342105263159e-05,
      "loss": 0.4587,
      "step": 42312
    },
    {
      "epoch": 2.3694142681151305,
      "grad_norm": 1.310226321220398,
      "learning_rate": 9.413315789473685e-05,
      "loss": 0.4365,
      "step": 42313
    },
    {
      "epoch": 2.3694702654272595,
      "grad_norm": 1.241117238998413,
      "learning_rate": 9.413289473684211e-05,
      "loss": 0.3804,
      "step": 42314
    },
    {
      "epoch": 2.3695262627393885,
      "grad_norm": 1.2082135677337646,
      "learning_rate": 9.413263157894737e-05,
      "loss": 0.3618,
      "step": 42315
    },
    {
      "epoch": 2.3695822600515175,
      "grad_norm": 0.9718478322029114,
      "learning_rate": 9.413236842105264e-05,
      "loss": 0.3222,
      "step": 42316
    },
    {
      "epoch": 2.3696382573636465,
      "grad_norm": 1.446513056755066,
      "learning_rate": 9.41321052631579e-05,
      "loss": 0.6055,
      "step": 42317
    },
    {
      "epoch": 2.3696942546757755,
      "grad_norm": 1.2139391899108887,
      "learning_rate": 9.413184210526316e-05,
      "loss": 0.4942,
      "step": 42318
    },
    {
      "epoch": 2.3697502519879046,
      "grad_norm": 2.036512613296509,
      "learning_rate": 9.413157894736842e-05,
      "loss": 0.5179,
      "step": 42319
    },
    {
      "epoch": 2.3698062493000336,
      "grad_norm": 1.020703911781311,
      "learning_rate": 9.413131578947368e-05,
      "loss": 0.4201,
      "step": 42320
    },
    {
      "epoch": 2.3698622466121626,
      "grad_norm": 1.8046185970306396,
      "learning_rate": 9.413105263157896e-05,
      "loss": 0.382,
      "step": 42321
    },
    {
      "epoch": 2.3699182439242916,
      "grad_norm": 1.198351263999939,
      "learning_rate": 9.413078947368422e-05,
      "loss": 0.398,
      "step": 42322
    },
    {
      "epoch": 2.3699742412364206,
      "grad_norm": 1.2732659578323364,
      "learning_rate": 9.413052631578948e-05,
      "loss": 0.471,
      "step": 42323
    },
    {
      "epoch": 2.3700302385485497,
      "grad_norm": 1.4621856212615967,
      "learning_rate": 9.413026315789474e-05,
      "loss": 0.4931,
      "step": 42324
    },
    {
      "epoch": 2.3700862358606787,
      "grad_norm": 1.0360468626022339,
      "learning_rate": 9.413000000000001e-05,
      "loss": 0.4257,
      "step": 42325
    },
    {
      "epoch": 2.3701422331728077,
      "grad_norm": 1.0498589277267456,
      "learning_rate": 9.412973684210527e-05,
      "loss": 0.3706,
      "step": 42326
    },
    {
      "epoch": 2.3701982304849367,
      "grad_norm": 1.2263721227645874,
      "learning_rate": 9.412947368421053e-05,
      "loss": 0.3882,
      "step": 42327
    },
    {
      "epoch": 2.3702542277970657,
      "grad_norm": 1.3455969095230103,
      "learning_rate": 9.412921052631579e-05,
      "loss": 0.4953,
      "step": 42328
    },
    {
      "epoch": 2.3703102251091948,
      "grad_norm": 1.1746537685394287,
      "learning_rate": 9.412894736842106e-05,
      "loss": 0.4029,
      "step": 42329
    },
    {
      "epoch": 2.3703662224213238,
      "grad_norm": 1.2489250898361206,
      "learning_rate": 9.412868421052632e-05,
      "loss": 0.429,
      "step": 42330
    },
    {
      "epoch": 2.370422219733453,
      "grad_norm": 1.1374889612197876,
      "learning_rate": 9.41284210526316e-05,
      "loss": 0.3791,
      "step": 42331
    },
    {
      "epoch": 2.370478217045582,
      "grad_norm": 1.2026957273483276,
      "learning_rate": 9.412815789473684e-05,
      "loss": 0.3491,
      "step": 42332
    },
    {
      "epoch": 2.370534214357711,
      "grad_norm": 1.35395085811615,
      "learning_rate": 9.412789473684211e-05,
      "loss": 0.5071,
      "step": 42333
    },
    {
      "epoch": 2.37059021166984,
      "grad_norm": 1.2361044883728027,
      "learning_rate": 9.412763157894737e-05,
      "loss": 0.5443,
      "step": 42334
    },
    {
      "epoch": 2.370646208981969,
      "grad_norm": 1.1043180227279663,
      "learning_rate": 9.412736842105263e-05,
      "loss": 0.3432,
      "step": 42335
    },
    {
      "epoch": 2.370702206294098,
      "grad_norm": 1.1496894359588623,
      "learning_rate": 9.41271052631579e-05,
      "loss": 0.4516,
      "step": 42336
    },
    {
      "epoch": 2.370758203606227,
      "grad_norm": 1.928362250328064,
      "learning_rate": 9.412684210526315e-05,
      "loss": 0.4151,
      "step": 42337
    },
    {
      "epoch": 2.370814200918356,
      "grad_norm": 1.3049509525299072,
      "learning_rate": 9.412657894736843e-05,
      "loss": 0.5352,
      "step": 42338
    },
    {
      "epoch": 2.370870198230485,
      "grad_norm": 1.2352666854858398,
      "learning_rate": 9.412631578947369e-05,
      "loss": 0.3984,
      "step": 42339
    },
    {
      "epoch": 2.370926195542614,
      "grad_norm": 1.3515156507492065,
      "learning_rate": 9.412605263157896e-05,
      "loss": 0.4786,
      "step": 42340
    },
    {
      "epoch": 2.370982192854743,
      "grad_norm": 1.0450198650360107,
      "learning_rate": 9.412578947368422e-05,
      "loss": 0.4316,
      "step": 42341
    },
    {
      "epoch": 2.371038190166872,
      "grad_norm": 1.3953994512557983,
      "learning_rate": 9.412552631578948e-05,
      "loss": 0.3837,
      "step": 42342
    },
    {
      "epoch": 2.371094187479001,
      "grad_norm": 1.0789560079574585,
      "learning_rate": 9.412526315789474e-05,
      "loss": 0.3729,
      "step": 42343
    },
    {
      "epoch": 2.37115018479113,
      "grad_norm": 1.348918080329895,
      "learning_rate": 9.412500000000001e-05,
      "loss": 0.3958,
      "step": 42344
    },
    {
      "epoch": 2.371206182103259,
      "grad_norm": 1.452298879623413,
      "learning_rate": 9.412473684210527e-05,
      "loss": 0.5401,
      "step": 42345
    },
    {
      "epoch": 2.371262179415388,
      "grad_norm": 1.172171950340271,
      "learning_rate": 9.412447368421053e-05,
      "loss": 0.3964,
      "step": 42346
    },
    {
      "epoch": 2.371318176727517,
      "grad_norm": 1.212975263595581,
      "learning_rate": 9.412421052631579e-05,
      "loss": 0.4021,
      "step": 42347
    },
    {
      "epoch": 2.371374174039646,
      "grad_norm": 1.376387119293213,
      "learning_rate": 9.412394736842106e-05,
      "loss": 0.5202,
      "step": 42348
    },
    {
      "epoch": 2.371430171351775,
      "grad_norm": 1.3071963787078857,
      "learning_rate": 9.412368421052632e-05,
      "loss": 0.3982,
      "step": 42349
    },
    {
      "epoch": 2.371486168663904,
      "grad_norm": 1.24335777759552,
      "learning_rate": 9.412342105263158e-05,
      "loss": 0.4069,
      "step": 42350
    },
    {
      "epoch": 2.371542165976033,
      "grad_norm": 1.262821912765503,
      "learning_rate": 9.412315789473684e-05,
      "loss": 0.5295,
      "step": 42351
    },
    {
      "epoch": 2.371598163288162,
      "grad_norm": 1.224733829498291,
      "learning_rate": 9.41228947368421e-05,
      "loss": 0.3901,
      "step": 42352
    },
    {
      "epoch": 2.371654160600291,
      "grad_norm": 1.1541788578033447,
      "learning_rate": 9.412263157894738e-05,
      "loss": 0.4325,
      "step": 42353
    },
    {
      "epoch": 2.3717101579124202,
      "grad_norm": 1.331683874130249,
      "learning_rate": 9.412236842105264e-05,
      "loss": 0.5881,
      "step": 42354
    },
    {
      "epoch": 2.3717661552245493,
      "grad_norm": 1.3279318809509277,
      "learning_rate": 9.41221052631579e-05,
      "loss": 0.4407,
      "step": 42355
    },
    {
      "epoch": 2.3718221525366783,
      "grad_norm": 1.3541243076324463,
      "learning_rate": 9.412184210526315e-05,
      "loss": 0.511,
      "step": 42356
    },
    {
      "epoch": 2.3718781498488073,
      "grad_norm": 1.2629435062408447,
      "learning_rate": 9.412157894736843e-05,
      "loss": 0.4459,
      "step": 42357
    },
    {
      "epoch": 2.3719341471609363,
      "grad_norm": 1.2103248834609985,
      "learning_rate": 9.412131578947369e-05,
      "loss": 0.4809,
      "step": 42358
    },
    {
      "epoch": 2.3719901444730653,
      "grad_norm": 1.0771657228469849,
      "learning_rate": 9.412105263157895e-05,
      "loss": 0.2962,
      "step": 42359
    },
    {
      "epoch": 2.3720461417851944,
      "grad_norm": 1.3111639022827148,
      "learning_rate": 9.412078947368421e-05,
      "loss": 0.396,
      "step": 42360
    },
    {
      "epoch": 2.3721021390973234,
      "grad_norm": 1.2420310974121094,
      "learning_rate": 9.412052631578948e-05,
      "loss": 0.2855,
      "step": 42361
    },
    {
      "epoch": 2.3721581364094524,
      "grad_norm": 1.4291789531707764,
      "learning_rate": 9.412026315789474e-05,
      "loss": 0.5448,
      "step": 42362
    },
    {
      "epoch": 2.3722141337215814,
      "grad_norm": 1.33905827999115,
      "learning_rate": 9.412000000000001e-05,
      "loss": 0.358,
      "step": 42363
    },
    {
      "epoch": 2.3722701310337104,
      "grad_norm": 1.0521758794784546,
      "learning_rate": 9.411973684210526e-05,
      "loss": 0.3166,
      "step": 42364
    },
    {
      "epoch": 2.3723261283458394,
      "grad_norm": 1.2340853214263916,
      "learning_rate": 9.411947368421053e-05,
      "loss": 0.5367,
      "step": 42365
    },
    {
      "epoch": 2.3723821256579685,
      "grad_norm": 1.116244912147522,
      "learning_rate": 9.411921052631579e-05,
      "loss": 0.2931,
      "step": 42366
    },
    {
      "epoch": 2.3724381229700975,
      "grad_norm": 1.0315062999725342,
      "learning_rate": 9.411894736842107e-05,
      "loss": 0.3077,
      "step": 42367
    },
    {
      "epoch": 2.3724941202822265,
      "grad_norm": 1.2503843307495117,
      "learning_rate": 9.411868421052633e-05,
      "loss": 0.4035,
      "step": 42368
    },
    {
      "epoch": 2.3725501175943555,
      "grad_norm": 1.4868385791778564,
      "learning_rate": 9.411842105263157e-05,
      "loss": 0.5422,
      "step": 42369
    },
    {
      "epoch": 2.3726061149064845,
      "grad_norm": 1.2327184677124023,
      "learning_rate": 9.411815789473685e-05,
      "loss": 0.4454,
      "step": 42370
    },
    {
      "epoch": 2.3726621122186136,
      "grad_norm": 1.0259407758712769,
      "learning_rate": 9.41178947368421e-05,
      "loss": 0.3556,
      "step": 42371
    },
    {
      "epoch": 2.3727181095307426,
      "grad_norm": 1.0937976837158203,
      "learning_rate": 9.411763157894738e-05,
      "loss": 0.3771,
      "step": 42372
    },
    {
      "epoch": 2.3727741068428716,
      "grad_norm": 1.10593843460083,
      "learning_rate": 9.411736842105264e-05,
      "loss": 0.3912,
      "step": 42373
    },
    {
      "epoch": 2.3728301041550006,
      "grad_norm": 1.1883816719055176,
      "learning_rate": 9.41171052631579e-05,
      "loss": 0.3656,
      "step": 42374
    },
    {
      "epoch": 2.3728861014671296,
      "grad_norm": 1.1932767629623413,
      "learning_rate": 9.411684210526316e-05,
      "loss": 0.3913,
      "step": 42375
    },
    {
      "epoch": 2.3729420987792587,
      "grad_norm": 1.00204336643219,
      "learning_rate": 9.411657894736843e-05,
      "loss": 0.3639,
      "step": 42376
    },
    {
      "epoch": 2.3729980960913877,
      "grad_norm": 1.1682087182998657,
      "learning_rate": 9.411631578947369e-05,
      "loss": 0.3734,
      "step": 42377
    },
    {
      "epoch": 2.3730540934035167,
      "grad_norm": 1.6911283731460571,
      "learning_rate": 9.411605263157895e-05,
      "loss": 0.4758,
      "step": 42378
    },
    {
      "epoch": 2.3731100907156457,
      "grad_norm": 1.0432487726211548,
      "learning_rate": 9.411578947368421e-05,
      "loss": 0.3029,
      "step": 42379
    },
    {
      "epoch": 2.3731660880277747,
      "grad_norm": 1.0524612665176392,
      "learning_rate": 9.411552631578948e-05,
      "loss": 0.3323,
      "step": 42380
    },
    {
      "epoch": 2.3732220853399038,
      "grad_norm": 0.9982818961143494,
      "learning_rate": 9.411526315789474e-05,
      "loss": 0.4652,
      "step": 42381
    },
    {
      "epoch": 2.3732780826520328,
      "grad_norm": 1.2826111316680908,
      "learning_rate": 9.4115e-05,
      "loss": 0.4497,
      "step": 42382
    },
    {
      "epoch": 2.373334079964162,
      "grad_norm": 1.3327100276947021,
      "learning_rate": 9.411473684210526e-05,
      "loss": 0.4338,
      "step": 42383
    },
    {
      "epoch": 2.373390077276291,
      "grad_norm": 1.1968905925750732,
      "learning_rate": 9.411447368421054e-05,
      "loss": 0.4065,
      "step": 42384
    },
    {
      "epoch": 2.37344607458842,
      "grad_norm": 1.0256004333496094,
      "learning_rate": 9.41142105263158e-05,
      "loss": 0.4193,
      "step": 42385
    },
    {
      "epoch": 2.373502071900549,
      "grad_norm": 1.3103188276290894,
      "learning_rate": 9.411394736842107e-05,
      "loss": 0.4892,
      "step": 42386
    },
    {
      "epoch": 2.373558069212678,
      "grad_norm": 1.2221312522888184,
      "learning_rate": 9.411368421052631e-05,
      "loss": 0.4555,
      "step": 42387
    },
    {
      "epoch": 2.373614066524807,
      "grad_norm": 1.1479402780532837,
      "learning_rate": 9.411342105263157e-05,
      "loss": 0.2965,
      "step": 42388
    },
    {
      "epoch": 2.373670063836936,
      "grad_norm": 1.3358964920043945,
      "learning_rate": 9.411315789473685e-05,
      "loss": 0.4131,
      "step": 42389
    },
    {
      "epoch": 2.373726061149065,
      "grad_norm": 1.0753743648529053,
      "learning_rate": 9.411289473684211e-05,
      "loss": 0.3822,
      "step": 42390
    },
    {
      "epoch": 2.373782058461194,
      "grad_norm": 1.2815124988555908,
      "learning_rate": 9.411263157894738e-05,
      "loss": 0.3596,
      "step": 42391
    },
    {
      "epoch": 2.373838055773323,
      "grad_norm": 1.650792121887207,
      "learning_rate": 9.411236842105263e-05,
      "loss": 0.3503,
      "step": 42392
    },
    {
      "epoch": 2.373894053085452,
      "grad_norm": 1.120585322380066,
      "learning_rate": 9.41121052631579e-05,
      "loss": 0.323,
      "step": 42393
    },
    {
      "epoch": 2.373950050397581,
      "grad_norm": 1.1304606199264526,
      "learning_rate": 9.411184210526316e-05,
      "loss": 0.4634,
      "step": 42394
    },
    {
      "epoch": 2.37400604770971,
      "grad_norm": 1.1531093120574951,
      "learning_rate": 9.411157894736843e-05,
      "loss": 0.47,
      "step": 42395
    },
    {
      "epoch": 2.374062045021839,
      "grad_norm": 1.1587066650390625,
      "learning_rate": 9.411131578947369e-05,
      "loss": 0.4124,
      "step": 42396
    },
    {
      "epoch": 2.374118042333968,
      "grad_norm": 1.1382609605789185,
      "learning_rate": 9.411105263157895e-05,
      "loss": 0.3733,
      "step": 42397
    },
    {
      "epoch": 2.374174039646097,
      "grad_norm": 1.3380318880081177,
      "learning_rate": 9.411078947368421e-05,
      "loss": 0.449,
      "step": 42398
    },
    {
      "epoch": 2.374230036958226,
      "grad_norm": 1.1523327827453613,
      "learning_rate": 9.411052631578949e-05,
      "loss": 0.5129,
      "step": 42399
    },
    {
      "epoch": 2.374286034270355,
      "grad_norm": 1.366464614868164,
      "learning_rate": 9.411026315789475e-05,
      "loss": 0.347,
      "step": 42400
    },
    {
      "epoch": 2.374342031582484,
      "grad_norm": 1.1939518451690674,
      "learning_rate": 9.411e-05,
      "loss": 0.3971,
      "step": 42401
    },
    {
      "epoch": 2.374398028894613,
      "grad_norm": 1.1485081911087036,
      "learning_rate": 9.410973684210527e-05,
      "loss": 0.4003,
      "step": 42402
    },
    {
      "epoch": 2.374454026206742,
      "grad_norm": 1.0558888912200928,
      "learning_rate": 9.410947368421054e-05,
      "loss": 0.2693,
      "step": 42403
    },
    {
      "epoch": 2.374510023518871,
      "grad_norm": 1.3160074949264526,
      "learning_rate": 9.41092105263158e-05,
      "loss": 0.373,
      "step": 42404
    },
    {
      "epoch": 2.374566020831,
      "grad_norm": 1.238418459892273,
      "learning_rate": 9.410894736842106e-05,
      "loss": 0.3897,
      "step": 42405
    },
    {
      "epoch": 2.3746220181431292,
      "grad_norm": 1.578658103942871,
      "learning_rate": 9.410868421052632e-05,
      "loss": 0.3431,
      "step": 42406
    },
    {
      "epoch": 2.3746780154552583,
      "grad_norm": 1.2434436082839966,
      "learning_rate": 9.410842105263158e-05,
      "loss": 0.5052,
      "step": 42407
    },
    {
      "epoch": 2.3747340127673873,
      "grad_norm": 1.179065465927124,
      "learning_rate": 9.410815789473685e-05,
      "loss": 0.3723,
      "step": 42408
    },
    {
      "epoch": 2.3747900100795163,
      "grad_norm": 1.3952171802520752,
      "learning_rate": 9.410789473684211e-05,
      "loss": 0.4517,
      "step": 42409
    },
    {
      "epoch": 2.3748460073916453,
      "grad_norm": 1.4435498714447021,
      "learning_rate": 9.410763157894737e-05,
      "loss": 0.3141,
      "step": 42410
    },
    {
      "epoch": 2.3749020047037743,
      "grad_norm": 1.4181057214736938,
      "learning_rate": 9.410736842105263e-05,
      "loss": 0.5501,
      "step": 42411
    },
    {
      "epoch": 2.3749580020159033,
      "grad_norm": 1.1330512762069702,
      "learning_rate": 9.41071052631579e-05,
      "loss": 0.398,
      "step": 42412
    },
    {
      "epoch": 2.3750139993280324,
      "grad_norm": 1.0308467149734497,
      "learning_rate": 9.410684210526316e-05,
      "loss": 0.338,
      "step": 42413
    },
    {
      "epoch": 2.3750699966401614,
      "grad_norm": 1.4484411478042603,
      "learning_rate": 9.410657894736842e-05,
      "loss": 0.5694,
      "step": 42414
    },
    {
      "epoch": 2.3751259939522904,
      "grad_norm": 1.166049838066101,
      "learning_rate": 9.410631578947368e-05,
      "loss": 0.429,
      "step": 42415
    },
    {
      "epoch": 2.3751819912644194,
      "grad_norm": 1.5028506517410278,
      "learning_rate": 9.410605263157896e-05,
      "loss": 0.5987,
      "step": 42416
    },
    {
      "epoch": 2.3752379885765484,
      "grad_norm": 1.2491739988327026,
      "learning_rate": 9.410578947368422e-05,
      "loss": 0.4342,
      "step": 42417
    },
    {
      "epoch": 2.3752939858886775,
      "grad_norm": 1.2192823886871338,
      "learning_rate": 9.410552631578949e-05,
      "loss": 0.3977,
      "step": 42418
    },
    {
      "epoch": 2.3753499832008065,
      "grad_norm": 1.2274869680404663,
      "learning_rate": 9.410526315789473e-05,
      "loss": 0.3943,
      "step": 42419
    },
    {
      "epoch": 2.3754059805129355,
      "grad_norm": 1.4463802576065063,
      "learning_rate": 9.410500000000001e-05,
      "loss": 0.3577,
      "step": 42420
    },
    {
      "epoch": 2.3754619778250645,
      "grad_norm": 0.9956678748130798,
      "learning_rate": 9.410473684210527e-05,
      "loss": 0.2971,
      "step": 42421
    },
    {
      "epoch": 2.3755179751371935,
      "grad_norm": 2.575307846069336,
      "learning_rate": 9.410447368421053e-05,
      "loss": 0.3574,
      "step": 42422
    },
    {
      "epoch": 2.3755739724493226,
      "grad_norm": 1.0277279615402222,
      "learning_rate": 9.41042105263158e-05,
      "loss": 0.3396,
      "step": 42423
    },
    {
      "epoch": 2.3756299697614516,
      "grad_norm": 1.3075816631317139,
      "learning_rate": 9.410394736842105e-05,
      "loss": 0.4444,
      "step": 42424
    },
    {
      "epoch": 2.3756859670735806,
      "grad_norm": 1.1444107294082642,
      "learning_rate": 9.410368421052632e-05,
      "loss": 0.3799,
      "step": 42425
    },
    {
      "epoch": 2.3757419643857096,
      "grad_norm": 1.200604796409607,
      "learning_rate": 9.410342105263158e-05,
      "loss": 0.3877,
      "step": 42426
    },
    {
      "epoch": 2.3757979616978386,
      "grad_norm": 1.075493335723877,
      "learning_rate": 9.410315789473685e-05,
      "loss": 0.3021,
      "step": 42427
    },
    {
      "epoch": 2.3758539590099677,
      "grad_norm": 1.1089084148406982,
      "learning_rate": 9.410289473684211e-05,
      "loss": 0.2997,
      "step": 42428
    },
    {
      "epoch": 2.3759099563220967,
      "grad_norm": 1.238358974456787,
      "learning_rate": 9.410263157894737e-05,
      "loss": 0.4752,
      "step": 42429
    },
    {
      "epoch": 2.3759659536342257,
      "grad_norm": 1.1755434274673462,
      "learning_rate": 9.410236842105263e-05,
      "loss": 0.4905,
      "step": 42430
    },
    {
      "epoch": 2.3760219509463547,
      "grad_norm": 2.001124382019043,
      "learning_rate": 9.41021052631579e-05,
      "loss": 0.5663,
      "step": 42431
    },
    {
      "epoch": 2.3760779482584837,
      "grad_norm": 1.191519856452942,
      "learning_rate": 9.410184210526317e-05,
      "loss": 0.4035,
      "step": 42432
    },
    {
      "epoch": 2.3761339455706127,
      "grad_norm": 1.160946249961853,
      "learning_rate": 9.410157894736842e-05,
      "loss": 0.3357,
      "step": 42433
    },
    {
      "epoch": 2.3761899428827418,
      "grad_norm": 1.127658724784851,
      "learning_rate": 9.410131578947368e-05,
      "loss": 0.3297,
      "step": 42434
    },
    {
      "epoch": 2.376245940194871,
      "grad_norm": 1.3732601404190063,
      "learning_rate": 9.410105263157896e-05,
      "loss": 0.5153,
      "step": 42435
    },
    {
      "epoch": 2.376301937507,
      "grad_norm": 1.3933708667755127,
      "learning_rate": 9.410078947368422e-05,
      "loss": 0.3478,
      "step": 42436
    },
    {
      "epoch": 2.376357934819129,
      "grad_norm": 1.1052151918411255,
      "learning_rate": 9.410052631578948e-05,
      "loss": 0.5795,
      "step": 42437
    },
    {
      "epoch": 2.376413932131258,
      "grad_norm": 1.2381658554077148,
      "learning_rate": 9.410026315789474e-05,
      "loss": 0.3963,
      "step": 42438
    },
    {
      "epoch": 2.376469929443387,
      "grad_norm": 1.3573307991027832,
      "learning_rate": 9.41e-05,
      "loss": 0.4522,
      "step": 42439
    },
    {
      "epoch": 2.376525926755516,
      "grad_norm": 1.3043878078460693,
      "learning_rate": 9.409973684210527e-05,
      "loss": 0.3714,
      "step": 42440
    },
    {
      "epoch": 2.376581924067645,
      "grad_norm": 1.2031733989715576,
      "learning_rate": 9.409947368421053e-05,
      "loss": 0.4421,
      "step": 42441
    },
    {
      "epoch": 2.376637921379774,
      "grad_norm": 1.2736499309539795,
      "learning_rate": 9.409921052631579e-05,
      "loss": 0.51,
      "step": 42442
    },
    {
      "epoch": 2.376693918691903,
      "grad_norm": 0.9469358325004578,
      "learning_rate": 9.409894736842105e-05,
      "loss": 0.259,
      "step": 42443
    },
    {
      "epoch": 2.376749916004032,
      "grad_norm": 1.1957001686096191,
      "learning_rate": 9.409868421052632e-05,
      "loss": 0.4244,
      "step": 42444
    },
    {
      "epoch": 2.376805913316161,
      "grad_norm": 1.5072258710861206,
      "learning_rate": 9.409842105263158e-05,
      "loss": 0.4276,
      "step": 42445
    },
    {
      "epoch": 2.37686191062829,
      "grad_norm": 1.1035648584365845,
      "learning_rate": 9.409815789473686e-05,
      "loss": 0.3905,
      "step": 42446
    },
    {
      "epoch": 2.376917907940419,
      "grad_norm": 1.0916134119033813,
      "learning_rate": 9.40978947368421e-05,
      "loss": 0.3051,
      "step": 42447
    },
    {
      "epoch": 2.376973905252548,
      "grad_norm": 1.3853813409805298,
      "learning_rate": 9.409763157894738e-05,
      "loss": 0.3954,
      "step": 42448
    },
    {
      "epoch": 2.377029902564677,
      "grad_norm": 1.0745974779129028,
      "learning_rate": 9.409736842105263e-05,
      "loss": 0.3295,
      "step": 42449
    },
    {
      "epoch": 2.377085899876806,
      "grad_norm": 1.5301626920700073,
      "learning_rate": 9.409710526315791e-05,
      "loss": 0.5941,
      "step": 42450
    },
    {
      "epoch": 2.377141897188935,
      "grad_norm": 1.2304660081863403,
      "learning_rate": 9.409684210526315e-05,
      "loss": 0.4205,
      "step": 42451
    },
    {
      "epoch": 2.377197894501064,
      "grad_norm": 1.218997597694397,
      "learning_rate": 9.409657894736843e-05,
      "loss": 0.341,
      "step": 42452
    },
    {
      "epoch": 2.377253891813193,
      "grad_norm": 3.525034189224243,
      "learning_rate": 9.409631578947369e-05,
      "loss": 0.4506,
      "step": 42453
    },
    {
      "epoch": 2.377309889125322,
      "grad_norm": 1.222209095954895,
      "learning_rate": 9.409605263157896e-05,
      "loss": 0.3776,
      "step": 42454
    },
    {
      "epoch": 2.377365886437451,
      "grad_norm": 1.5149116516113281,
      "learning_rate": 9.409578947368422e-05,
      "loss": 0.5059,
      "step": 42455
    },
    {
      "epoch": 2.37742188374958,
      "grad_norm": 1.2607446908950806,
      "learning_rate": 9.409552631578947e-05,
      "loss": 0.4005,
      "step": 42456
    },
    {
      "epoch": 2.377477881061709,
      "grad_norm": 1.5401476621627808,
      "learning_rate": 9.409526315789474e-05,
      "loss": 0.5519,
      "step": 42457
    },
    {
      "epoch": 2.3775338783738382,
      "grad_norm": 1.2779308557510376,
      "learning_rate": 9.4095e-05,
      "loss": 0.4528,
      "step": 42458
    },
    {
      "epoch": 2.3775898756859672,
      "grad_norm": 1.2017914056777954,
      "learning_rate": 9.409473684210527e-05,
      "loss": 0.3509,
      "step": 42459
    },
    {
      "epoch": 2.3776458729980963,
      "grad_norm": 1.1976125240325928,
      "learning_rate": 9.409447368421053e-05,
      "loss": 0.4927,
      "step": 42460
    },
    {
      "epoch": 2.3777018703102253,
      "grad_norm": 1.3936169147491455,
      "learning_rate": 9.409421052631579e-05,
      "loss": 0.4801,
      "step": 42461
    },
    {
      "epoch": 2.3777578676223543,
      "grad_norm": 1.4349616765975952,
      "learning_rate": 9.409394736842105e-05,
      "loss": 0.4154,
      "step": 42462
    },
    {
      "epoch": 2.3778138649344833,
      "grad_norm": 1.1959031820297241,
      "learning_rate": 9.409368421052633e-05,
      "loss": 0.4593,
      "step": 42463
    },
    {
      "epoch": 2.3778698622466123,
      "grad_norm": 1.3643070459365845,
      "learning_rate": 9.409342105263158e-05,
      "loss": 0.3677,
      "step": 42464
    },
    {
      "epoch": 2.3779258595587414,
      "grad_norm": 1.3468074798583984,
      "learning_rate": 9.409315789473684e-05,
      "loss": 0.3999,
      "step": 42465
    },
    {
      "epoch": 2.3779818568708704,
      "grad_norm": 1.089212417602539,
      "learning_rate": 9.40928947368421e-05,
      "loss": 0.366,
      "step": 42466
    },
    {
      "epoch": 2.3780378541829994,
      "grad_norm": 1.1597673892974854,
      "learning_rate": 9.409263157894738e-05,
      "loss": 0.3565,
      "step": 42467
    },
    {
      "epoch": 2.3780938514951284,
      "grad_norm": 1.2557344436645508,
      "learning_rate": 9.409236842105264e-05,
      "loss": 0.3627,
      "step": 42468
    },
    {
      "epoch": 2.3781498488072574,
      "grad_norm": 1.2739707231521606,
      "learning_rate": 9.40921052631579e-05,
      "loss": 0.4697,
      "step": 42469
    },
    {
      "epoch": 2.3782058461193865,
      "grad_norm": 1.112221360206604,
      "learning_rate": 9.409184210526316e-05,
      "loss": 0.3037,
      "step": 42470
    },
    {
      "epoch": 2.3782618434315155,
      "grad_norm": 1.2911196947097778,
      "learning_rate": 9.409157894736843e-05,
      "loss": 0.429,
      "step": 42471
    },
    {
      "epoch": 2.3783178407436445,
      "grad_norm": 1.1795971393585205,
      "learning_rate": 9.409131578947369e-05,
      "loss": 0.3821,
      "step": 42472
    },
    {
      "epoch": 2.3783738380557735,
      "grad_norm": 1.0450501441955566,
      "learning_rate": 9.409105263157895e-05,
      "loss": 0.383,
      "step": 42473
    },
    {
      "epoch": 2.3784298353679025,
      "grad_norm": 1.216058373451233,
      "learning_rate": 9.409078947368421e-05,
      "loss": 0.4603,
      "step": 42474
    },
    {
      "epoch": 2.378485832680031,
      "grad_norm": 1.1050595045089722,
      "learning_rate": 9.409052631578947e-05,
      "loss": 0.3484,
      "step": 42475
    },
    {
      "epoch": 2.3785418299921606,
      "grad_norm": 1.096500039100647,
      "learning_rate": 9.409026315789474e-05,
      "loss": 0.3869,
      "step": 42476
    },
    {
      "epoch": 2.378597827304289,
      "grad_norm": 1.244277000427246,
      "learning_rate": 9.409e-05,
      "loss": 0.4237,
      "step": 42477
    },
    {
      "epoch": 2.3786538246164186,
      "grad_norm": 1.6338012218475342,
      "learning_rate": 9.408973684210528e-05,
      "loss": 0.4734,
      "step": 42478
    },
    {
      "epoch": 2.378709821928547,
      "grad_norm": 1.1416494846343994,
      "learning_rate": 9.408947368421052e-05,
      "loss": 0.4312,
      "step": 42479
    },
    {
      "epoch": 2.3787658192406766,
      "grad_norm": 1.9732004404067993,
      "learning_rate": 9.40892105263158e-05,
      "loss": 0.3966,
      "step": 42480
    },
    {
      "epoch": 2.3788218165528052,
      "grad_norm": 1.327549695968628,
      "learning_rate": 9.408894736842105e-05,
      "loss": 0.3474,
      "step": 42481
    },
    {
      "epoch": 2.3788778138649347,
      "grad_norm": 1.1186048984527588,
      "learning_rate": 9.408868421052633e-05,
      "loss": 0.3206,
      "step": 42482
    },
    {
      "epoch": 2.3789338111770633,
      "grad_norm": 1.1910033226013184,
      "learning_rate": 9.408842105263159e-05,
      "loss": 0.4344,
      "step": 42483
    },
    {
      "epoch": 2.3789898084891927,
      "grad_norm": 1.308523416519165,
      "learning_rate": 9.408815789473685e-05,
      "loss": 0.514,
      "step": 42484
    },
    {
      "epoch": 2.3790458058013213,
      "grad_norm": 1.563263177871704,
      "learning_rate": 9.408789473684211e-05,
      "loss": 0.5107,
      "step": 42485
    },
    {
      "epoch": 2.3791018031134508,
      "grad_norm": 1.1659225225448608,
      "learning_rate": 9.408763157894738e-05,
      "loss": 0.3893,
      "step": 42486
    },
    {
      "epoch": 2.3791578004255793,
      "grad_norm": 1.2152005434036255,
      "learning_rate": 9.408736842105264e-05,
      "loss": 0.4992,
      "step": 42487
    },
    {
      "epoch": 2.379213797737709,
      "grad_norm": 1.2288497686386108,
      "learning_rate": 9.40871052631579e-05,
      "loss": 0.5169,
      "step": 42488
    },
    {
      "epoch": 2.3792697950498374,
      "grad_norm": 1.1491831541061401,
      "learning_rate": 9.408684210526316e-05,
      "loss": 0.3503,
      "step": 42489
    },
    {
      "epoch": 2.379325792361967,
      "grad_norm": 2.056318521499634,
      "learning_rate": 9.408657894736842e-05,
      "loss": 0.4091,
      "step": 42490
    },
    {
      "epoch": 2.3793817896740954,
      "grad_norm": 1.4687947034835815,
      "learning_rate": 9.408631578947369e-05,
      "loss": 0.5416,
      "step": 42491
    },
    {
      "epoch": 2.379437786986225,
      "grad_norm": 1.312280535697937,
      "learning_rate": 9.408605263157895e-05,
      "loss": 0.4215,
      "step": 42492
    },
    {
      "epoch": 2.3794937842983535,
      "grad_norm": 1.140360713005066,
      "learning_rate": 9.408578947368421e-05,
      "loss": 0.3485,
      "step": 42493
    },
    {
      "epoch": 2.379549781610483,
      "grad_norm": 1.2055162191390991,
      "learning_rate": 9.408552631578947e-05,
      "loss": 0.3483,
      "step": 42494
    },
    {
      "epoch": 2.3796057789226115,
      "grad_norm": 1.2068771123886108,
      "learning_rate": 9.408526315789474e-05,
      "loss": 0.3799,
      "step": 42495
    },
    {
      "epoch": 2.379661776234741,
      "grad_norm": 1.384659767150879,
      "learning_rate": 9.4085e-05,
      "loss": 0.5501,
      "step": 42496
    },
    {
      "epoch": 2.3797177735468695,
      "grad_norm": 1.3155443668365479,
      "learning_rate": 9.408473684210526e-05,
      "loss": 0.3832,
      "step": 42497
    },
    {
      "epoch": 2.379773770858999,
      "grad_norm": 1.3432371616363525,
      "learning_rate": 9.408447368421052e-05,
      "loss": 0.3812,
      "step": 42498
    },
    {
      "epoch": 2.3798297681711276,
      "grad_norm": 0.9414938688278198,
      "learning_rate": 9.40842105263158e-05,
      "loss": 0.3228,
      "step": 42499
    },
    {
      "epoch": 2.379885765483257,
      "grad_norm": 1.392841100692749,
      "learning_rate": 9.408394736842106e-05,
      "loss": 0.4655,
      "step": 42500
    },
    {
      "epoch": 2.3799417627953856,
      "grad_norm": 1.1543760299682617,
      "learning_rate": 9.408368421052633e-05,
      "loss": 0.3398,
      "step": 42501
    },
    {
      "epoch": 2.379997760107515,
      "grad_norm": 1.2505989074707031,
      "learning_rate": 9.408342105263158e-05,
      "loss": 0.3968,
      "step": 42502
    },
    {
      "epoch": 2.3800537574196436,
      "grad_norm": 1.1237807273864746,
      "learning_rate": 9.408315789473685e-05,
      "loss": 0.3717,
      "step": 42503
    },
    {
      "epoch": 2.380109754731773,
      "grad_norm": 1.085248589515686,
      "learning_rate": 9.408289473684211e-05,
      "loss": 0.4845,
      "step": 42504
    },
    {
      "epoch": 2.3801657520439017,
      "grad_norm": 1.2724963426589966,
      "learning_rate": 9.408263157894738e-05,
      "loss": 0.4373,
      "step": 42505
    },
    {
      "epoch": 2.380221749356031,
      "grad_norm": 1.1509380340576172,
      "learning_rate": 9.408236842105263e-05,
      "loss": 0.463,
      "step": 42506
    },
    {
      "epoch": 2.3802777466681597,
      "grad_norm": 1.275742530822754,
      "learning_rate": 9.40821052631579e-05,
      "loss": 0.5805,
      "step": 42507
    },
    {
      "epoch": 2.380333743980289,
      "grad_norm": 1.1412025690078735,
      "learning_rate": 9.408184210526316e-05,
      "loss": 0.3379,
      "step": 42508
    },
    {
      "epoch": 2.3803897412924178,
      "grad_norm": 1.1979143619537354,
      "learning_rate": 9.408157894736842e-05,
      "loss": 0.3587,
      "step": 42509
    },
    {
      "epoch": 2.3804457386045472,
      "grad_norm": 1.125076413154602,
      "learning_rate": 9.40813157894737e-05,
      "loss": 0.4127,
      "step": 42510
    },
    {
      "epoch": 2.380501735916676,
      "grad_norm": 1.0476491451263428,
      "learning_rate": 9.408105263157894e-05,
      "loss": 0.3689,
      "step": 42511
    },
    {
      "epoch": 2.3805577332288053,
      "grad_norm": 1.202276349067688,
      "learning_rate": 9.408078947368421e-05,
      "loss": 0.3619,
      "step": 42512
    },
    {
      "epoch": 2.380613730540934,
      "grad_norm": 1.092563509941101,
      "learning_rate": 9.408052631578947e-05,
      "loss": 0.5527,
      "step": 42513
    },
    {
      "epoch": 2.3806697278530633,
      "grad_norm": 1.2603657245635986,
      "learning_rate": 9.408026315789475e-05,
      "loss": 0.5164,
      "step": 42514
    },
    {
      "epoch": 2.380725725165192,
      "grad_norm": 1.2806247472763062,
      "learning_rate": 9.408000000000001e-05,
      "loss": 0.5413,
      "step": 42515
    },
    {
      "epoch": 2.3807817224773213,
      "grad_norm": 1.360559344291687,
      "learning_rate": 9.407973684210527e-05,
      "loss": 0.5411,
      "step": 42516
    },
    {
      "epoch": 2.38083771978945,
      "grad_norm": 1.0450483560562134,
      "learning_rate": 9.407947368421053e-05,
      "loss": 0.4087,
      "step": 42517
    },
    {
      "epoch": 2.380893717101579,
      "grad_norm": 1.2660740613937378,
      "learning_rate": 9.40792105263158e-05,
      "loss": 0.5154,
      "step": 42518
    },
    {
      "epoch": 2.380949714413708,
      "grad_norm": 1.2018334865570068,
      "learning_rate": 9.407894736842106e-05,
      "loss": 0.3713,
      "step": 42519
    },
    {
      "epoch": 2.381005711725837,
      "grad_norm": 1.1157623529434204,
      "learning_rate": 9.407868421052632e-05,
      "loss": 0.3616,
      "step": 42520
    },
    {
      "epoch": 2.381061709037966,
      "grad_norm": 1.9918997287750244,
      "learning_rate": 9.407842105263158e-05,
      "loss": 0.3589,
      "step": 42521
    },
    {
      "epoch": 2.381117706350095,
      "grad_norm": 1.1965969800949097,
      "learning_rate": 9.407815789473685e-05,
      "loss": 0.4041,
      "step": 42522
    },
    {
      "epoch": 2.381173703662224,
      "grad_norm": 1.4048922061920166,
      "learning_rate": 9.407789473684211e-05,
      "loss": 0.3891,
      "step": 42523
    },
    {
      "epoch": 2.381229700974353,
      "grad_norm": 1.3953217267990112,
      "learning_rate": 9.407763157894737e-05,
      "loss": 0.3885,
      "step": 42524
    },
    {
      "epoch": 2.381285698286482,
      "grad_norm": 1.2107235193252563,
      "learning_rate": 9.407736842105263e-05,
      "loss": 0.3978,
      "step": 42525
    },
    {
      "epoch": 2.381341695598611,
      "grad_norm": 1.2201377153396606,
      "learning_rate": 9.407710526315789e-05,
      "loss": 0.5128,
      "step": 42526
    },
    {
      "epoch": 2.38139769291074,
      "grad_norm": 1.235162615776062,
      "learning_rate": 9.407684210526316e-05,
      "loss": 0.4138,
      "step": 42527
    },
    {
      "epoch": 2.381453690222869,
      "grad_norm": 1.0170131921768188,
      "learning_rate": 9.407657894736842e-05,
      "loss": 0.4483,
      "step": 42528
    },
    {
      "epoch": 2.381509687534998,
      "grad_norm": 1.326956868171692,
      "learning_rate": 9.407631578947368e-05,
      "loss": 0.4156,
      "step": 42529
    },
    {
      "epoch": 2.381565684847127,
      "grad_norm": 1.121862530708313,
      "learning_rate": 9.407605263157894e-05,
      "loss": 0.4415,
      "step": 42530
    },
    {
      "epoch": 2.381621682159256,
      "grad_norm": 1.4975086450576782,
      "learning_rate": 9.407578947368422e-05,
      "loss": 0.3848,
      "step": 42531
    },
    {
      "epoch": 2.381677679471385,
      "grad_norm": 1.1271705627441406,
      "learning_rate": 9.407552631578948e-05,
      "loss": 0.5332,
      "step": 42532
    },
    {
      "epoch": 2.381733676783514,
      "grad_norm": 1.3706332445144653,
      "learning_rate": 9.407526315789475e-05,
      "loss": 0.3456,
      "step": 42533
    },
    {
      "epoch": 2.3817896740956432,
      "grad_norm": 1.7513446807861328,
      "learning_rate": 9.4075e-05,
      "loss": 0.4118,
      "step": 42534
    },
    {
      "epoch": 2.3818456714077723,
      "grad_norm": 1.1036343574523926,
      "learning_rate": 9.407473684210527e-05,
      "loss": 0.4785,
      "step": 42535
    },
    {
      "epoch": 2.3819016687199013,
      "grad_norm": 1.2073208093643188,
      "learning_rate": 9.407447368421053e-05,
      "loss": 0.4064,
      "step": 42536
    },
    {
      "epoch": 2.3819576660320303,
      "grad_norm": 1.003194808959961,
      "learning_rate": 9.40742105263158e-05,
      "loss": 0.3475,
      "step": 42537
    },
    {
      "epoch": 2.3820136633441593,
      "grad_norm": 1.0659860372543335,
      "learning_rate": 9.407394736842106e-05,
      "loss": 0.3201,
      "step": 42538
    },
    {
      "epoch": 2.3820696606562883,
      "grad_norm": 1.2434395551681519,
      "learning_rate": 9.407368421052632e-05,
      "loss": 0.3965,
      "step": 42539
    },
    {
      "epoch": 2.3821256579684174,
      "grad_norm": 2.6625494956970215,
      "learning_rate": 9.407342105263158e-05,
      "loss": 0.4299,
      "step": 42540
    },
    {
      "epoch": 2.3821816552805464,
      "grad_norm": 1.6783849000930786,
      "learning_rate": 9.407315789473686e-05,
      "loss": 0.3971,
      "step": 42541
    },
    {
      "epoch": 2.3822376525926754,
      "grad_norm": 1.285970687866211,
      "learning_rate": 9.407289473684211e-05,
      "loss": 0.4381,
      "step": 42542
    },
    {
      "epoch": 2.3822936499048044,
      "grad_norm": 1.2205157279968262,
      "learning_rate": 9.407263157894737e-05,
      "loss": 0.4058,
      "step": 42543
    },
    {
      "epoch": 2.3823496472169334,
      "grad_norm": 1.2811921834945679,
      "learning_rate": 9.407236842105263e-05,
      "loss": 0.393,
      "step": 42544
    },
    {
      "epoch": 2.3824056445290624,
      "grad_norm": 1.2970865964889526,
      "learning_rate": 9.40721052631579e-05,
      "loss": 0.6065,
      "step": 42545
    },
    {
      "epoch": 2.3824616418411915,
      "grad_norm": 1.2408443689346313,
      "learning_rate": 9.407184210526317e-05,
      "loss": 0.3807,
      "step": 42546
    },
    {
      "epoch": 2.3825176391533205,
      "grad_norm": 1.3409698009490967,
      "learning_rate": 9.407157894736843e-05,
      "loss": 0.4483,
      "step": 42547
    },
    {
      "epoch": 2.3825736364654495,
      "grad_norm": 1.4250755310058594,
      "learning_rate": 9.407131578947369e-05,
      "loss": 0.5103,
      "step": 42548
    },
    {
      "epoch": 2.3826296337775785,
      "grad_norm": 1.2405428886413574,
      "learning_rate": 9.407105263157895e-05,
      "loss": 0.3803,
      "step": 42549
    },
    {
      "epoch": 2.3826856310897075,
      "grad_norm": 1.17203688621521,
      "learning_rate": 9.407078947368422e-05,
      "loss": 0.393,
      "step": 42550
    },
    {
      "epoch": 2.3827416284018366,
      "grad_norm": 1.2503646612167358,
      "learning_rate": 9.407052631578948e-05,
      "loss": 0.4466,
      "step": 42551
    },
    {
      "epoch": 2.3827976257139656,
      "grad_norm": 3.6451518535614014,
      "learning_rate": 9.407026315789474e-05,
      "loss": 0.3539,
      "step": 42552
    },
    {
      "epoch": 2.3828536230260946,
      "grad_norm": 1.1758249998092651,
      "learning_rate": 9.407e-05,
      "loss": 0.384,
      "step": 42553
    },
    {
      "epoch": 2.3829096203382236,
      "grad_norm": 1.2939223051071167,
      "learning_rate": 9.406973684210527e-05,
      "loss": 0.4589,
      "step": 42554
    },
    {
      "epoch": 2.3829656176503526,
      "grad_norm": 1.0699198246002197,
      "learning_rate": 9.406947368421053e-05,
      "loss": 0.3628,
      "step": 42555
    },
    {
      "epoch": 2.3830216149624817,
      "grad_norm": 1.463165283203125,
      "learning_rate": 9.40692105263158e-05,
      "loss": 0.4143,
      "step": 42556
    },
    {
      "epoch": 2.3830776122746107,
      "grad_norm": 1.2300323247909546,
      "learning_rate": 9.406894736842105e-05,
      "loss": 0.3432,
      "step": 42557
    },
    {
      "epoch": 2.3831336095867397,
      "grad_norm": 1.3511840105056763,
      "learning_rate": 9.406868421052632e-05,
      "loss": 0.3803,
      "step": 42558
    },
    {
      "epoch": 2.3831896068988687,
      "grad_norm": 1.0859596729278564,
      "learning_rate": 9.406842105263158e-05,
      "loss": 0.3732,
      "step": 42559
    },
    {
      "epoch": 2.3832456042109977,
      "grad_norm": 1.1830259561538696,
      "learning_rate": 9.406815789473684e-05,
      "loss": 0.4137,
      "step": 42560
    },
    {
      "epoch": 2.3833016015231268,
      "grad_norm": 1.1586086750030518,
      "learning_rate": 9.40678947368421e-05,
      "loss": 0.3749,
      "step": 42561
    },
    {
      "epoch": 2.3833575988352558,
      "grad_norm": 1.0648373365402222,
      "learning_rate": 9.406763157894736e-05,
      "loss": 0.4581,
      "step": 42562
    },
    {
      "epoch": 2.383413596147385,
      "grad_norm": 1.062685489654541,
      "learning_rate": 9.406736842105264e-05,
      "loss": 0.3683,
      "step": 42563
    },
    {
      "epoch": 2.383469593459514,
      "grad_norm": 1.1497598886489868,
      "learning_rate": 9.40671052631579e-05,
      "loss": 0.3664,
      "step": 42564
    },
    {
      "epoch": 2.383525590771643,
      "grad_norm": 1.1968621015548706,
      "learning_rate": 9.406684210526317e-05,
      "loss": 0.3181,
      "step": 42565
    },
    {
      "epoch": 2.383581588083772,
      "grad_norm": 1.191324234008789,
      "learning_rate": 9.406657894736842e-05,
      "loss": 0.4757,
      "step": 42566
    },
    {
      "epoch": 2.383637585395901,
      "grad_norm": 1.2028124332427979,
      "learning_rate": 9.406631578947369e-05,
      "loss": 0.3237,
      "step": 42567
    },
    {
      "epoch": 2.38369358270803,
      "grad_norm": 1.3055126667022705,
      "learning_rate": 9.406605263157895e-05,
      "loss": 0.4884,
      "step": 42568
    },
    {
      "epoch": 2.383749580020159,
      "grad_norm": 1.5982452630996704,
      "learning_rate": 9.406578947368422e-05,
      "loss": 0.489,
      "step": 42569
    },
    {
      "epoch": 2.383805577332288,
      "grad_norm": 0.9566874504089355,
      "learning_rate": 9.406552631578948e-05,
      "loss": 0.33,
      "step": 42570
    },
    {
      "epoch": 2.383861574644417,
      "grad_norm": 1.3176177740097046,
      "learning_rate": 9.406526315789474e-05,
      "loss": 0.3599,
      "step": 42571
    },
    {
      "epoch": 2.383917571956546,
      "grad_norm": 0.9685880541801453,
      "learning_rate": 9.4065e-05,
      "loss": 0.3114,
      "step": 42572
    },
    {
      "epoch": 2.383973569268675,
      "grad_norm": 1.288273572921753,
      "learning_rate": 9.406473684210527e-05,
      "loss": 0.4814,
      "step": 42573
    },
    {
      "epoch": 2.384029566580804,
      "grad_norm": 1.2276841402053833,
      "learning_rate": 9.406447368421053e-05,
      "loss": 0.4001,
      "step": 42574
    },
    {
      "epoch": 2.384085563892933,
      "grad_norm": 1.0506620407104492,
      "learning_rate": 9.40642105263158e-05,
      "loss": 0.3761,
      "step": 42575
    },
    {
      "epoch": 2.384141561205062,
      "grad_norm": 1.33298921585083,
      "learning_rate": 9.406394736842105e-05,
      "loss": 0.4378,
      "step": 42576
    },
    {
      "epoch": 2.384197558517191,
      "grad_norm": 1.2699986696243286,
      "learning_rate": 9.406368421052631e-05,
      "loss": 0.3371,
      "step": 42577
    },
    {
      "epoch": 2.38425355582932,
      "grad_norm": 1.2369009256362915,
      "learning_rate": 9.406342105263159e-05,
      "loss": 0.3554,
      "step": 42578
    },
    {
      "epoch": 2.384309553141449,
      "grad_norm": 1.1824365854263306,
      "learning_rate": 9.406315789473685e-05,
      "loss": 0.4967,
      "step": 42579
    },
    {
      "epoch": 2.384365550453578,
      "grad_norm": 1.4971169233322144,
      "learning_rate": 9.40628947368421e-05,
      "loss": 0.3747,
      "step": 42580
    },
    {
      "epoch": 2.384421547765707,
      "grad_norm": 1.3762418031692505,
      "learning_rate": 9.406263157894737e-05,
      "loss": 0.4584,
      "step": 42581
    },
    {
      "epoch": 2.384477545077836,
      "grad_norm": 1.2758166790008545,
      "learning_rate": 9.406236842105264e-05,
      "loss": 0.3064,
      "step": 42582
    },
    {
      "epoch": 2.384533542389965,
      "grad_norm": 1.3122650384902954,
      "learning_rate": 9.40621052631579e-05,
      "loss": 0.397,
      "step": 42583
    },
    {
      "epoch": 2.384589539702094,
      "grad_norm": 1.0488005876541138,
      "learning_rate": 9.406184210526316e-05,
      "loss": 0.3866,
      "step": 42584
    },
    {
      "epoch": 2.384645537014223,
      "grad_norm": 1.020864486694336,
      "learning_rate": 9.406157894736842e-05,
      "loss": 0.2957,
      "step": 42585
    },
    {
      "epoch": 2.3847015343263522,
      "grad_norm": 1.4510560035705566,
      "learning_rate": 9.406131578947369e-05,
      "loss": 0.4678,
      "step": 42586
    },
    {
      "epoch": 2.3847575316384813,
      "grad_norm": 1.2683963775634766,
      "learning_rate": 9.406105263157895e-05,
      "loss": 0.452,
      "step": 42587
    },
    {
      "epoch": 2.3848135289506103,
      "grad_norm": 1.1072547435760498,
      "learning_rate": 9.406078947368422e-05,
      "loss": 0.3309,
      "step": 42588
    },
    {
      "epoch": 2.3848695262627393,
      "grad_norm": 1.121264934539795,
      "learning_rate": 9.406052631578947e-05,
      "loss": 0.5253,
      "step": 42589
    },
    {
      "epoch": 2.3849255235748683,
      "grad_norm": 1.1909416913986206,
      "learning_rate": 9.406026315789474e-05,
      "loss": 0.4868,
      "step": 42590
    },
    {
      "epoch": 2.3849815208869973,
      "grad_norm": 1.237779974937439,
      "learning_rate": 9.406e-05,
      "loss": 0.3394,
      "step": 42591
    },
    {
      "epoch": 2.3850375181991263,
      "grad_norm": 1.1321485042572021,
      "learning_rate": 9.405973684210528e-05,
      "loss": 0.4951,
      "step": 42592
    },
    {
      "epoch": 2.3850935155112554,
      "grad_norm": 1.1843172311782837,
      "learning_rate": 9.405947368421054e-05,
      "loss": 0.3951,
      "step": 42593
    },
    {
      "epoch": 2.3851495128233844,
      "grad_norm": 1.0003044605255127,
      "learning_rate": 9.405921052631578e-05,
      "loss": 0.3542,
      "step": 42594
    },
    {
      "epoch": 2.3852055101355134,
      "grad_norm": 15.324576377868652,
      "learning_rate": 9.405894736842106e-05,
      "loss": 0.27,
      "step": 42595
    },
    {
      "epoch": 2.3852615074476424,
      "grad_norm": 1.124688982963562,
      "learning_rate": 9.405868421052632e-05,
      "loss": 0.3357,
      "step": 42596
    },
    {
      "epoch": 2.3853175047597714,
      "grad_norm": 0.9334547519683838,
      "learning_rate": 9.405842105263159e-05,
      "loss": 0.2975,
      "step": 42597
    },
    {
      "epoch": 2.3853735020719005,
      "grad_norm": 1.2480884790420532,
      "learning_rate": 9.405815789473684e-05,
      "loss": 0.4539,
      "step": 42598
    },
    {
      "epoch": 2.3854294993840295,
      "grad_norm": 5.411787986755371,
      "learning_rate": 9.405789473684211e-05,
      "loss": 0.5638,
      "step": 42599
    },
    {
      "epoch": 2.3854854966961585,
      "grad_norm": 1.331764578819275,
      "learning_rate": 9.405763157894737e-05,
      "loss": 0.4644,
      "step": 42600
    },
    {
      "epoch": 2.3855414940082875,
      "grad_norm": 1.3296133279800415,
      "learning_rate": 9.405736842105264e-05,
      "loss": 0.6881,
      "step": 42601
    },
    {
      "epoch": 2.3855974913204165,
      "grad_norm": 1.2573027610778809,
      "learning_rate": 9.40571052631579e-05,
      "loss": 0.6897,
      "step": 42602
    },
    {
      "epoch": 2.3856534886325456,
      "grad_norm": 1.117430329322815,
      "learning_rate": 9.405684210526316e-05,
      "loss": 0.3679,
      "step": 42603
    },
    {
      "epoch": 2.3857094859446746,
      "grad_norm": 1.3276596069335938,
      "learning_rate": 9.405657894736842e-05,
      "loss": 0.5267,
      "step": 42604
    },
    {
      "epoch": 2.3857654832568036,
      "grad_norm": 1.1627373695373535,
      "learning_rate": 9.40563157894737e-05,
      "loss": 0.3476,
      "step": 42605
    },
    {
      "epoch": 2.3858214805689326,
      "grad_norm": 1.2422236204147339,
      "learning_rate": 9.405605263157895e-05,
      "loss": 0.4312,
      "step": 42606
    },
    {
      "epoch": 2.3858774778810616,
      "grad_norm": 1.2037944793701172,
      "learning_rate": 9.405578947368421e-05,
      "loss": 0.4309,
      "step": 42607
    },
    {
      "epoch": 2.3859334751931907,
      "grad_norm": 1.30731999874115,
      "learning_rate": 9.405552631578947e-05,
      "loss": 0.5137,
      "step": 42608
    },
    {
      "epoch": 2.3859894725053197,
      "grad_norm": 1.1168564558029175,
      "learning_rate": 9.405526315789475e-05,
      "loss": 0.4747,
      "step": 42609
    },
    {
      "epoch": 2.3860454698174487,
      "grad_norm": 1.1194229125976562,
      "learning_rate": 9.4055e-05,
      "loss": 0.3362,
      "step": 42610
    },
    {
      "epoch": 2.3861014671295777,
      "grad_norm": 1.008150339126587,
      "learning_rate": 9.405473684210527e-05,
      "loss": 0.5189,
      "step": 42611
    },
    {
      "epoch": 2.3861574644417067,
      "grad_norm": 1.055415153503418,
      "learning_rate": 9.405447368421053e-05,
      "loss": 0.3264,
      "step": 42612
    },
    {
      "epoch": 2.3862134617538358,
      "grad_norm": 1.0336294174194336,
      "learning_rate": 9.405421052631579e-05,
      "loss": 0.4551,
      "step": 42613
    },
    {
      "epoch": 2.3862694590659648,
      "grad_norm": 1.0394479036331177,
      "learning_rate": 9.405394736842106e-05,
      "loss": 0.3806,
      "step": 42614
    },
    {
      "epoch": 2.386325456378094,
      "grad_norm": 1.166113018989563,
      "learning_rate": 9.405368421052632e-05,
      "loss": 0.4929,
      "step": 42615
    },
    {
      "epoch": 2.386381453690223,
      "grad_norm": 4.127678394317627,
      "learning_rate": 9.405342105263158e-05,
      "loss": 0.4001,
      "step": 42616
    },
    {
      "epoch": 2.386437451002352,
      "grad_norm": 1.1513208150863647,
      "learning_rate": 9.405315789473684e-05,
      "loss": 0.3832,
      "step": 42617
    },
    {
      "epoch": 2.386493448314481,
      "grad_norm": 1.2024736404418945,
      "learning_rate": 9.405289473684211e-05,
      "loss": 0.3868,
      "step": 42618
    },
    {
      "epoch": 2.38654944562661,
      "grad_norm": 1.1587923765182495,
      "learning_rate": 9.405263157894737e-05,
      "loss": 0.4126,
      "step": 42619
    },
    {
      "epoch": 2.386605442938739,
      "grad_norm": 1.6466926336288452,
      "learning_rate": 9.405236842105264e-05,
      "loss": 0.5869,
      "step": 42620
    },
    {
      "epoch": 2.386661440250868,
      "grad_norm": 1.379857063293457,
      "learning_rate": 9.405210526315789e-05,
      "loss": 0.3721,
      "step": 42621
    },
    {
      "epoch": 2.386717437562997,
      "grad_norm": 1.1543580293655396,
      "learning_rate": 9.405184210526316e-05,
      "loss": 0.3289,
      "step": 42622
    },
    {
      "epoch": 2.386773434875126,
      "grad_norm": 0.9915464520454407,
      "learning_rate": 9.405157894736842e-05,
      "loss": 0.3488,
      "step": 42623
    },
    {
      "epoch": 2.386829432187255,
      "grad_norm": 1.0548279285430908,
      "learning_rate": 9.40513157894737e-05,
      "loss": 0.3626,
      "step": 42624
    },
    {
      "epoch": 2.386885429499384,
      "grad_norm": 1.111660122871399,
      "learning_rate": 9.405105263157896e-05,
      "loss": 0.3327,
      "step": 42625
    },
    {
      "epoch": 2.386941426811513,
      "grad_norm": 1.109354853630066,
      "learning_rate": 9.405078947368422e-05,
      "loss": 0.3759,
      "step": 42626
    },
    {
      "epoch": 2.386997424123642,
      "grad_norm": 1.059809923171997,
      "learning_rate": 9.405052631578948e-05,
      "loss": 0.3487,
      "step": 42627
    },
    {
      "epoch": 2.387053421435771,
      "grad_norm": 1.2506803274154663,
      "learning_rate": 9.405026315789475e-05,
      "loss": 0.3445,
      "step": 42628
    },
    {
      "epoch": 2.3871094187479,
      "grad_norm": 0.9957774877548218,
      "learning_rate": 9.405000000000001e-05,
      "loss": 0.3543,
      "step": 42629
    },
    {
      "epoch": 2.387165416060029,
      "grad_norm": 1.4011187553405762,
      "learning_rate": 9.404973684210527e-05,
      "loss": 0.3819,
      "step": 42630
    },
    {
      "epoch": 2.387221413372158,
      "grad_norm": 1.187614917755127,
      "learning_rate": 9.404947368421053e-05,
      "loss": 0.3956,
      "step": 42631
    },
    {
      "epoch": 2.387277410684287,
      "grad_norm": 1.0884028673171997,
      "learning_rate": 9.404921052631579e-05,
      "loss": 0.4755,
      "step": 42632
    },
    {
      "epoch": 2.387333407996416,
      "grad_norm": 1.2023686170578003,
      "learning_rate": 9.404894736842106e-05,
      "loss": 0.4647,
      "step": 42633
    },
    {
      "epoch": 2.387389405308545,
      "grad_norm": 1.1974496841430664,
      "learning_rate": 9.404868421052632e-05,
      "loss": 0.3666,
      "step": 42634
    },
    {
      "epoch": 2.387445402620674,
      "grad_norm": 1.119666337966919,
      "learning_rate": 9.404842105263158e-05,
      "loss": 0.357,
      "step": 42635
    },
    {
      "epoch": 2.387501399932803,
      "grad_norm": 1.0321756601333618,
      "learning_rate": 9.404815789473684e-05,
      "loss": 0.3378,
      "step": 42636
    },
    {
      "epoch": 2.387557397244932,
      "grad_norm": 1.3267462253570557,
      "learning_rate": 9.404789473684211e-05,
      "loss": 0.3869,
      "step": 42637
    },
    {
      "epoch": 2.3876133945570612,
      "grad_norm": 1.1031581163406372,
      "learning_rate": 9.404763157894737e-05,
      "loss": 0.4331,
      "step": 42638
    },
    {
      "epoch": 2.3876693918691902,
      "grad_norm": 1.022200345993042,
      "learning_rate": 9.404736842105263e-05,
      "loss": 0.3178,
      "step": 42639
    },
    {
      "epoch": 2.3877253891813193,
      "grad_norm": 1.3006731271743774,
      "learning_rate": 9.40471052631579e-05,
      "loss": 0.3602,
      "step": 42640
    },
    {
      "epoch": 2.3877813864934483,
      "grad_norm": 1.0300768613815308,
      "learning_rate": 9.404684210526317e-05,
      "loss": 0.3414,
      "step": 42641
    },
    {
      "epoch": 2.3878373838055773,
      "grad_norm": 1.1395182609558105,
      "learning_rate": 9.404657894736843e-05,
      "loss": 0.3352,
      "step": 42642
    },
    {
      "epoch": 2.3878933811177063,
      "grad_norm": 1.2421512603759766,
      "learning_rate": 9.40463157894737e-05,
      "loss": 0.3343,
      "step": 42643
    },
    {
      "epoch": 2.3879493784298353,
      "grad_norm": 1.195260763168335,
      "learning_rate": 9.404605263157895e-05,
      "loss": 0.3609,
      "step": 42644
    },
    {
      "epoch": 2.3880053757419644,
      "grad_norm": 1.1531985998153687,
      "learning_rate": 9.404578947368422e-05,
      "loss": 0.3348,
      "step": 42645
    },
    {
      "epoch": 2.3880613730540934,
      "grad_norm": 1.0802340507507324,
      "learning_rate": 9.404552631578948e-05,
      "loss": 0.3581,
      "step": 42646
    },
    {
      "epoch": 2.3881173703662224,
      "grad_norm": 1.3586745262145996,
      "learning_rate": 9.404526315789474e-05,
      "loss": 0.3773,
      "step": 42647
    },
    {
      "epoch": 2.3881733676783514,
      "grad_norm": 1.2025184631347656,
      "learning_rate": 9.404500000000001e-05,
      "loss": 0.4027,
      "step": 42648
    },
    {
      "epoch": 2.3882293649904804,
      "grad_norm": 1.1350380182266235,
      "learning_rate": 9.404473684210526e-05,
      "loss": 0.3496,
      "step": 42649
    },
    {
      "epoch": 2.3882853623026095,
      "grad_norm": 1.095055341720581,
      "learning_rate": 9.404447368421053e-05,
      "loss": 0.3094,
      "step": 42650
    },
    {
      "epoch": 2.3883413596147385,
      "grad_norm": 1.1523957252502441,
      "learning_rate": 9.404421052631579e-05,
      "loss": 0.3901,
      "step": 42651
    },
    {
      "epoch": 2.3883973569268675,
      "grad_norm": 1.1032826900482178,
      "learning_rate": 9.404394736842106e-05,
      "loss": 0.3929,
      "step": 42652
    },
    {
      "epoch": 2.3884533542389965,
      "grad_norm": 1.4672939777374268,
      "learning_rate": 9.404368421052631e-05,
      "loss": 0.4425,
      "step": 42653
    },
    {
      "epoch": 2.3885093515511255,
      "grad_norm": 1.3077774047851562,
      "learning_rate": 9.404342105263158e-05,
      "loss": 0.3643,
      "step": 42654
    },
    {
      "epoch": 2.3885653488632546,
      "grad_norm": 1.5778776407241821,
      "learning_rate": 9.404315789473684e-05,
      "loss": 0.4809,
      "step": 42655
    },
    {
      "epoch": 2.3886213461753836,
      "grad_norm": 1.1037013530731201,
      "learning_rate": 9.404289473684212e-05,
      "loss": 0.4742,
      "step": 42656
    },
    {
      "epoch": 2.3886773434875126,
      "grad_norm": 1.0530924797058105,
      "learning_rate": 9.404263157894738e-05,
      "loss": 0.3816,
      "step": 42657
    },
    {
      "epoch": 2.3887333407996416,
      "grad_norm": 1.268921136856079,
      "learning_rate": 9.404236842105264e-05,
      "loss": 0.5544,
      "step": 42658
    },
    {
      "epoch": 2.3887893381117706,
      "grad_norm": 1.18560791015625,
      "learning_rate": 9.40421052631579e-05,
      "loss": 0.4251,
      "step": 42659
    },
    {
      "epoch": 2.3888453354238997,
      "grad_norm": 1.2642385959625244,
      "learning_rate": 9.404184210526317e-05,
      "loss": 0.4166,
      "step": 42660
    },
    {
      "epoch": 2.3889013327360287,
      "grad_norm": 1.18563973903656,
      "learning_rate": 9.404157894736843e-05,
      "loss": 0.5054,
      "step": 42661
    },
    {
      "epoch": 2.3889573300481577,
      "grad_norm": 1.1250606775283813,
      "learning_rate": 9.404131578947369e-05,
      "loss": 0.3635,
      "step": 42662
    },
    {
      "epoch": 2.3890133273602867,
      "grad_norm": 1.5753247737884521,
      "learning_rate": 9.404105263157895e-05,
      "loss": 0.3921,
      "step": 42663
    },
    {
      "epoch": 2.3890693246724157,
      "grad_norm": 1.2122334241867065,
      "learning_rate": 9.404078947368421e-05,
      "loss": 0.4131,
      "step": 42664
    },
    {
      "epoch": 2.3891253219845447,
      "grad_norm": 1.4192795753479004,
      "learning_rate": 9.404052631578948e-05,
      "loss": 0.495,
      "step": 42665
    },
    {
      "epoch": 2.3891813192966738,
      "grad_norm": 1.0998648405075073,
      "learning_rate": 9.404026315789474e-05,
      "loss": 0.2548,
      "step": 42666
    },
    {
      "epoch": 2.389237316608803,
      "grad_norm": 1.1161388158798218,
      "learning_rate": 9.404e-05,
      "loss": 0.4076,
      "step": 42667
    },
    {
      "epoch": 2.389293313920932,
      "grad_norm": 1.1583874225616455,
      "learning_rate": 9.403973684210526e-05,
      "loss": 0.3951,
      "step": 42668
    },
    {
      "epoch": 2.389349311233061,
      "grad_norm": 1.1109627485275269,
      "learning_rate": 9.403947368421053e-05,
      "loss": 0.4145,
      "step": 42669
    },
    {
      "epoch": 2.38940530854519,
      "grad_norm": 1.3675705194473267,
      "learning_rate": 9.40392105263158e-05,
      "loss": 0.5858,
      "step": 42670
    },
    {
      "epoch": 2.389461305857319,
      "grad_norm": 1.4983429908752441,
      "learning_rate": 9.403894736842105e-05,
      "loss": 0.4275,
      "step": 42671
    },
    {
      "epoch": 2.389517303169448,
      "grad_norm": 1.1537508964538574,
      "learning_rate": 9.403868421052631e-05,
      "loss": 0.6054,
      "step": 42672
    },
    {
      "epoch": 2.389573300481577,
      "grad_norm": 1.521843671798706,
      "learning_rate": 9.403842105263159e-05,
      "loss": 0.4285,
      "step": 42673
    },
    {
      "epoch": 2.389629297793706,
      "grad_norm": 1.1374387741088867,
      "learning_rate": 9.403815789473685e-05,
      "loss": 0.4711,
      "step": 42674
    },
    {
      "epoch": 2.389685295105835,
      "grad_norm": 1.174505591392517,
      "learning_rate": 9.403789473684212e-05,
      "loss": 0.3622,
      "step": 42675
    },
    {
      "epoch": 2.389741292417964,
      "grad_norm": 1.2992359399795532,
      "learning_rate": 9.403763157894737e-05,
      "loss": 0.4013,
      "step": 42676
    },
    {
      "epoch": 2.389797289730093,
      "grad_norm": 1.2300549745559692,
      "learning_rate": 9.403736842105264e-05,
      "loss": 0.4557,
      "step": 42677
    },
    {
      "epoch": 2.389853287042222,
      "grad_norm": 1.2163190841674805,
      "learning_rate": 9.40371052631579e-05,
      "loss": 0.3282,
      "step": 42678
    },
    {
      "epoch": 2.389909284354351,
      "grad_norm": 1.3333773612976074,
      "learning_rate": 9.403684210526317e-05,
      "loss": 0.5762,
      "step": 42679
    },
    {
      "epoch": 2.38996528166648,
      "grad_norm": 1.0658111572265625,
      "learning_rate": 9.403657894736843e-05,
      "loss": 0.397,
      "step": 42680
    },
    {
      "epoch": 2.390021278978609,
      "grad_norm": 1.051979899406433,
      "learning_rate": 9.403631578947368e-05,
      "loss": 0.4385,
      "step": 42681
    },
    {
      "epoch": 2.390077276290738,
      "grad_norm": 1.2111361026763916,
      "learning_rate": 9.403605263157895e-05,
      "loss": 0.3194,
      "step": 42682
    },
    {
      "epoch": 2.390133273602867,
      "grad_norm": 1.2240217924118042,
      "learning_rate": 9.403578947368421e-05,
      "loss": 0.3632,
      "step": 42683
    },
    {
      "epoch": 2.390189270914996,
      "grad_norm": 1.3208426237106323,
      "learning_rate": 9.403552631578948e-05,
      "loss": 0.3418,
      "step": 42684
    },
    {
      "epoch": 2.390245268227125,
      "grad_norm": 1.2668181657791138,
      "learning_rate": 9.403526315789474e-05,
      "loss": 0.5016,
      "step": 42685
    },
    {
      "epoch": 2.390301265539254,
      "grad_norm": 1.1530085802078247,
      "learning_rate": 9.4035e-05,
      "loss": 0.388,
      "step": 42686
    },
    {
      "epoch": 2.390357262851383,
      "grad_norm": 1.2501213550567627,
      "learning_rate": 9.403473684210526e-05,
      "loss": 0.5076,
      "step": 42687
    },
    {
      "epoch": 2.390413260163512,
      "grad_norm": 1.1210938692092896,
      "learning_rate": 9.403447368421054e-05,
      "loss": 0.3942,
      "step": 42688
    },
    {
      "epoch": 2.390469257475641,
      "grad_norm": 1.1361290216445923,
      "learning_rate": 9.40342105263158e-05,
      "loss": 0.3223,
      "step": 42689
    },
    {
      "epoch": 2.3905252547877702,
      "grad_norm": 1.1473275423049927,
      "learning_rate": 9.403394736842106e-05,
      "loss": 0.4319,
      "step": 42690
    },
    {
      "epoch": 2.3905812520998992,
      "grad_norm": 1.6550055742263794,
      "learning_rate": 9.403368421052632e-05,
      "loss": 0.3256,
      "step": 42691
    },
    {
      "epoch": 2.3906372494120283,
      "grad_norm": 1.1310539245605469,
      "learning_rate": 9.403342105263159e-05,
      "loss": 0.292,
      "step": 42692
    },
    {
      "epoch": 2.3906932467241573,
      "grad_norm": 1.851605772972107,
      "learning_rate": 9.403315789473685e-05,
      "loss": 0.3793,
      "step": 42693
    },
    {
      "epoch": 2.3907492440362863,
      "grad_norm": 1.2337329387664795,
      "learning_rate": 9.403289473684211e-05,
      "loss": 0.4356,
      "step": 42694
    },
    {
      "epoch": 2.3908052413484153,
      "grad_norm": 1.118699073791504,
      "learning_rate": 9.403263157894737e-05,
      "loss": 0.3398,
      "step": 42695
    },
    {
      "epoch": 2.3908612386605443,
      "grad_norm": 0.984032392501831,
      "learning_rate": 9.403236842105264e-05,
      "loss": 0.3362,
      "step": 42696
    },
    {
      "epoch": 2.3909172359726734,
      "grad_norm": 1.1673319339752197,
      "learning_rate": 9.40321052631579e-05,
      "loss": 0.3786,
      "step": 42697
    },
    {
      "epoch": 2.3909732332848024,
      "grad_norm": 1.3165054321289062,
      "learning_rate": 9.403184210526316e-05,
      "loss": 0.492,
      "step": 42698
    },
    {
      "epoch": 2.3910292305969314,
      "grad_norm": 1.2805838584899902,
      "learning_rate": 9.403157894736842e-05,
      "loss": 0.5496,
      "step": 42699
    },
    {
      "epoch": 2.3910852279090604,
      "grad_norm": 1.298423409461975,
      "learning_rate": 9.403131578947368e-05,
      "loss": 0.3035,
      "step": 42700
    },
    {
      "epoch": 2.3911412252211894,
      "grad_norm": 1.1271740198135376,
      "learning_rate": 9.403105263157895e-05,
      "loss": 0.3902,
      "step": 42701
    },
    {
      "epoch": 2.3911972225333185,
      "grad_norm": 1.182063341140747,
      "learning_rate": 9.403078947368421e-05,
      "loss": 0.4121,
      "step": 42702
    },
    {
      "epoch": 2.3912532198454475,
      "grad_norm": 1.1168649196624756,
      "learning_rate": 9.403052631578949e-05,
      "loss": 0.3589,
      "step": 42703
    },
    {
      "epoch": 2.3913092171575765,
      "grad_norm": 1.0168119668960571,
      "learning_rate": 9.403026315789473e-05,
      "loss": 0.3186,
      "step": 42704
    },
    {
      "epoch": 2.3913652144697055,
      "grad_norm": 1.2867718935012817,
      "learning_rate": 9.403e-05,
      "loss": 0.494,
      "step": 42705
    },
    {
      "epoch": 2.3914212117818345,
      "grad_norm": 1.3506160974502563,
      "learning_rate": 9.402973684210527e-05,
      "loss": 0.3393,
      "step": 42706
    },
    {
      "epoch": 2.3914772090939636,
      "grad_norm": 1.448885202407837,
      "learning_rate": 9.402947368421054e-05,
      "loss": 0.3793,
      "step": 42707
    },
    {
      "epoch": 2.3915332064060926,
      "grad_norm": 1.1891429424285889,
      "learning_rate": 9.402921052631579e-05,
      "loss": 0.3184,
      "step": 42708
    },
    {
      "epoch": 2.3915892037182216,
      "grad_norm": 1.5013341903686523,
      "learning_rate": 9.402894736842106e-05,
      "loss": 0.5169,
      "step": 42709
    },
    {
      "epoch": 2.3916452010303506,
      "grad_norm": 1.4108986854553223,
      "learning_rate": 9.402868421052632e-05,
      "loss": 0.4365,
      "step": 42710
    },
    {
      "epoch": 2.3917011983424796,
      "grad_norm": 1.2258822917938232,
      "learning_rate": 9.402842105263159e-05,
      "loss": 0.4077,
      "step": 42711
    },
    {
      "epoch": 2.3917571956546086,
      "grad_norm": 1.0955442190170288,
      "learning_rate": 9.402815789473685e-05,
      "loss": 0.4137,
      "step": 42712
    },
    {
      "epoch": 2.3918131929667377,
      "grad_norm": 0.9462907314300537,
      "learning_rate": 9.402789473684211e-05,
      "loss": 0.2962,
      "step": 42713
    },
    {
      "epoch": 2.3918691902788667,
      "grad_norm": 1.1818344593048096,
      "learning_rate": 9.402763157894737e-05,
      "loss": 0.4124,
      "step": 42714
    },
    {
      "epoch": 2.3919251875909957,
      "grad_norm": 1.0877208709716797,
      "learning_rate": 9.402736842105263e-05,
      "loss": 0.4221,
      "step": 42715
    },
    {
      "epoch": 2.3919811849031247,
      "grad_norm": 1.0634098052978516,
      "learning_rate": 9.40271052631579e-05,
      "loss": 0.3711,
      "step": 42716
    },
    {
      "epoch": 2.3920371822152537,
      "grad_norm": 1.2608414888381958,
      "learning_rate": 9.402684210526316e-05,
      "loss": 0.4463,
      "step": 42717
    },
    {
      "epoch": 2.3920931795273828,
      "grad_norm": 1.2866356372833252,
      "learning_rate": 9.402657894736842e-05,
      "loss": 0.3731,
      "step": 42718
    },
    {
      "epoch": 2.392149176839512,
      "grad_norm": 1.1856920719146729,
      "learning_rate": 9.402631578947368e-05,
      "loss": 0.322,
      "step": 42719
    },
    {
      "epoch": 2.392205174151641,
      "grad_norm": 1.1983140707015991,
      "learning_rate": 9.402605263157896e-05,
      "loss": 0.3605,
      "step": 42720
    },
    {
      "epoch": 2.39226117146377,
      "grad_norm": 1.392439603805542,
      "learning_rate": 9.402578947368422e-05,
      "loss": 0.5023,
      "step": 42721
    },
    {
      "epoch": 2.392317168775899,
      "grad_norm": 1.2660714387893677,
      "learning_rate": 9.402552631578948e-05,
      "loss": 0.403,
      "step": 42722
    },
    {
      "epoch": 2.392373166088028,
      "grad_norm": 1.411842703819275,
      "learning_rate": 9.402526315789474e-05,
      "loss": 0.4247,
      "step": 42723
    },
    {
      "epoch": 2.392429163400157,
      "grad_norm": 1.3333075046539307,
      "learning_rate": 9.402500000000001e-05,
      "loss": 0.4288,
      "step": 42724
    },
    {
      "epoch": 2.392485160712286,
      "grad_norm": 1.2108970880508423,
      "learning_rate": 9.402473684210527e-05,
      "loss": 0.4055,
      "step": 42725
    },
    {
      "epoch": 2.392541158024415,
      "grad_norm": 1.088212013244629,
      "learning_rate": 9.402447368421053e-05,
      "loss": 0.4001,
      "step": 42726
    },
    {
      "epoch": 2.392597155336544,
      "grad_norm": 1.3110018968582153,
      "learning_rate": 9.402421052631579e-05,
      "loss": 0.377,
      "step": 42727
    },
    {
      "epoch": 2.392653152648673,
      "grad_norm": 1.108091950416565,
      "learning_rate": 9.402394736842106e-05,
      "loss": 0.4061,
      "step": 42728
    },
    {
      "epoch": 2.392709149960802,
      "grad_norm": 1.158329725265503,
      "learning_rate": 9.402368421052632e-05,
      "loss": 0.4217,
      "step": 42729
    },
    {
      "epoch": 2.392765147272931,
      "grad_norm": 1.5274122953414917,
      "learning_rate": 9.40234210526316e-05,
      "loss": 0.4145,
      "step": 42730
    },
    {
      "epoch": 2.39282114458506,
      "grad_norm": 1.365785837173462,
      "learning_rate": 9.402315789473684e-05,
      "loss": 0.4783,
      "step": 42731
    },
    {
      "epoch": 2.392877141897189,
      "grad_norm": 1.085085153579712,
      "learning_rate": 9.40228947368421e-05,
      "loss": 0.3383,
      "step": 42732
    },
    {
      "epoch": 2.392933139209318,
      "grad_norm": 1.1834867000579834,
      "learning_rate": 9.402263157894737e-05,
      "loss": 0.4196,
      "step": 42733
    },
    {
      "epoch": 2.392989136521447,
      "grad_norm": 1.3948897123336792,
      "learning_rate": 9.402236842105263e-05,
      "loss": 0.4615,
      "step": 42734
    },
    {
      "epoch": 2.393045133833576,
      "grad_norm": 1.0753192901611328,
      "learning_rate": 9.40221052631579e-05,
      "loss": 0.3919,
      "step": 42735
    },
    {
      "epoch": 2.393101131145705,
      "grad_norm": 1.6030347347259521,
      "learning_rate": 9.402184210526315e-05,
      "loss": 0.3469,
      "step": 42736
    },
    {
      "epoch": 2.393157128457834,
      "grad_norm": 1.2505477666854858,
      "learning_rate": 9.402157894736843e-05,
      "loss": 0.3151,
      "step": 42737
    },
    {
      "epoch": 2.393213125769963,
      "grad_norm": 1.2580486536026,
      "learning_rate": 9.402131578947369e-05,
      "loss": 0.2761,
      "step": 42738
    },
    {
      "epoch": 2.393269123082092,
      "grad_norm": 1.143646240234375,
      "learning_rate": 9.402105263157896e-05,
      "loss": 0.3778,
      "step": 42739
    },
    {
      "epoch": 2.393325120394221,
      "grad_norm": 1.2043582201004028,
      "learning_rate": 9.402078947368422e-05,
      "loss": 0.4391,
      "step": 42740
    },
    {
      "epoch": 2.39338111770635,
      "grad_norm": 0.9684794545173645,
      "learning_rate": 9.402052631578948e-05,
      "loss": 0.3339,
      "step": 42741
    },
    {
      "epoch": 2.3934371150184792,
      "grad_norm": 1.1173056364059448,
      "learning_rate": 9.402026315789474e-05,
      "loss": 0.2831,
      "step": 42742
    },
    {
      "epoch": 2.3934931123306082,
      "grad_norm": 0.9866859912872314,
      "learning_rate": 9.402000000000001e-05,
      "loss": 0.286,
      "step": 42743
    },
    {
      "epoch": 2.3935491096427373,
      "grad_norm": 3.10390305519104,
      "learning_rate": 9.401973684210527e-05,
      "loss": 0.5762,
      "step": 42744
    },
    {
      "epoch": 2.3936051069548663,
      "grad_norm": 1.401552438735962,
      "learning_rate": 9.401947368421053e-05,
      "loss": 0.3413,
      "step": 42745
    },
    {
      "epoch": 2.3936611042669953,
      "grad_norm": 1.2560654878616333,
      "learning_rate": 9.401921052631579e-05,
      "loss": 0.3923,
      "step": 42746
    },
    {
      "epoch": 2.3937171015791243,
      "grad_norm": 1.161210298538208,
      "learning_rate": 9.401894736842106e-05,
      "loss": 0.3351,
      "step": 42747
    },
    {
      "epoch": 2.3937730988912533,
      "grad_norm": 1.2119914293289185,
      "learning_rate": 9.401868421052632e-05,
      "loss": 0.4222,
      "step": 42748
    },
    {
      "epoch": 2.3938290962033824,
      "grad_norm": 1.0822558403015137,
      "learning_rate": 9.401842105263158e-05,
      "loss": 0.4639,
      "step": 42749
    },
    {
      "epoch": 2.3938850935155114,
      "grad_norm": 1.3759263753890991,
      "learning_rate": 9.401815789473684e-05,
      "loss": 0.4945,
      "step": 42750
    },
    {
      "epoch": 2.3939410908276404,
      "grad_norm": 1.3513514995574951,
      "learning_rate": 9.40178947368421e-05,
      "loss": 0.3742,
      "step": 42751
    },
    {
      "epoch": 2.3939970881397694,
      "grad_norm": 1.5572748184204102,
      "learning_rate": 9.401763157894738e-05,
      "loss": 0.3887,
      "step": 42752
    },
    {
      "epoch": 2.3940530854518984,
      "grad_norm": 1.1274296045303345,
      "learning_rate": 9.401736842105264e-05,
      "loss": 0.4432,
      "step": 42753
    },
    {
      "epoch": 2.3941090827640275,
      "grad_norm": 1.2218854427337646,
      "learning_rate": 9.40171052631579e-05,
      "loss": 0.4089,
      "step": 42754
    },
    {
      "epoch": 2.3941650800761565,
      "grad_norm": 1.073961615562439,
      "learning_rate": 9.401684210526316e-05,
      "loss": 0.5157,
      "step": 42755
    },
    {
      "epoch": 2.3942210773882855,
      "grad_norm": 1.107279896736145,
      "learning_rate": 9.401657894736843e-05,
      "loss": 0.495,
      "step": 42756
    },
    {
      "epoch": 2.3942770747004145,
      "grad_norm": 1.038818597793579,
      "learning_rate": 9.401631578947369e-05,
      "loss": 0.395,
      "step": 42757
    },
    {
      "epoch": 2.3943330720125435,
      "grad_norm": 1.2185393571853638,
      "learning_rate": 9.401605263157896e-05,
      "loss": 0.3753,
      "step": 42758
    },
    {
      "epoch": 2.3943890693246725,
      "grad_norm": 1.1542401313781738,
      "learning_rate": 9.401578947368421e-05,
      "loss": 0.4189,
      "step": 42759
    },
    {
      "epoch": 2.3944450666368016,
      "grad_norm": 1.1466140747070312,
      "learning_rate": 9.401552631578948e-05,
      "loss": 0.3728,
      "step": 42760
    },
    {
      "epoch": 2.3945010639489306,
      "grad_norm": 1.2688628435134888,
      "learning_rate": 9.401526315789474e-05,
      "loss": 0.3835,
      "step": 42761
    },
    {
      "epoch": 2.3945570612610596,
      "grad_norm": 1.1185503005981445,
      "learning_rate": 9.401500000000001e-05,
      "loss": 0.3116,
      "step": 42762
    },
    {
      "epoch": 2.3946130585731886,
      "grad_norm": 1.1198681592941284,
      "learning_rate": 9.401473684210526e-05,
      "loss": 0.3635,
      "step": 42763
    },
    {
      "epoch": 2.3946690558853176,
      "grad_norm": 1.3092260360717773,
      "learning_rate": 9.401447368421053e-05,
      "loss": 0.567,
      "step": 42764
    },
    {
      "epoch": 2.3947250531974467,
      "grad_norm": 1.058984637260437,
      "learning_rate": 9.401421052631579e-05,
      "loss": 0.342,
      "step": 42765
    },
    {
      "epoch": 2.3947810505095757,
      "grad_norm": 1.2620432376861572,
      "learning_rate": 9.401394736842107e-05,
      "loss": 0.3833,
      "step": 42766
    },
    {
      "epoch": 2.3948370478217047,
      "grad_norm": 1.0415794849395752,
      "learning_rate": 9.401368421052633e-05,
      "loss": 0.3985,
      "step": 42767
    },
    {
      "epoch": 2.3948930451338337,
      "grad_norm": 1.145587682723999,
      "learning_rate": 9.401342105263157e-05,
      "loss": 0.5011,
      "step": 42768
    },
    {
      "epoch": 2.3949490424459627,
      "grad_norm": 1.3376713991165161,
      "learning_rate": 9.401315789473685e-05,
      "loss": 0.3777,
      "step": 42769
    },
    {
      "epoch": 2.3950050397580918,
      "grad_norm": 1.276750922203064,
      "learning_rate": 9.40128947368421e-05,
      "loss": 0.3519,
      "step": 42770
    },
    {
      "epoch": 2.3950610370702208,
      "grad_norm": 1.0923223495483398,
      "learning_rate": 9.401263157894738e-05,
      "loss": 0.3671,
      "step": 42771
    },
    {
      "epoch": 2.39511703438235,
      "grad_norm": 1.375555157661438,
      "learning_rate": 9.401236842105264e-05,
      "loss": 0.4479,
      "step": 42772
    },
    {
      "epoch": 2.395173031694479,
      "grad_norm": 1.1508938074111938,
      "learning_rate": 9.40121052631579e-05,
      "loss": 0.375,
      "step": 42773
    },
    {
      "epoch": 2.395229029006608,
      "grad_norm": 1.0027307271957397,
      "learning_rate": 9.401184210526316e-05,
      "loss": 0.3353,
      "step": 42774
    },
    {
      "epoch": 2.395285026318737,
      "grad_norm": 1.1454983949661255,
      "learning_rate": 9.401157894736843e-05,
      "loss": 0.3756,
      "step": 42775
    },
    {
      "epoch": 2.395341023630866,
      "grad_norm": 1.2472492456436157,
      "learning_rate": 9.401131578947369e-05,
      "loss": 0.4752,
      "step": 42776
    },
    {
      "epoch": 2.395397020942995,
      "grad_norm": 1.1967554092407227,
      "learning_rate": 9.401105263157895e-05,
      "loss": 0.4373,
      "step": 42777
    },
    {
      "epoch": 2.395453018255124,
      "grad_norm": 1.2511521577835083,
      "learning_rate": 9.401078947368421e-05,
      "loss": 0.3579,
      "step": 42778
    },
    {
      "epoch": 2.395509015567253,
      "grad_norm": 1.672969102859497,
      "learning_rate": 9.401052631578948e-05,
      "loss": 0.441,
      "step": 42779
    },
    {
      "epoch": 2.395565012879382,
      "grad_norm": 1.1752283573150635,
      "learning_rate": 9.401026315789474e-05,
      "loss": 0.3223,
      "step": 42780
    },
    {
      "epoch": 2.395621010191511,
      "grad_norm": 1.1414518356323242,
      "learning_rate": 9.401e-05,
      "loss": 0.4768,
      "step": 42781
    },
    {
      "epoch": 2.39567700750364,
      "grad_norm": 1.2581430673599243,
      "learning_rate": 9.400973684210526e-05,
      "loss": 0.3748,
      "step": 42782
    },
    {
      "epoch": 2.395733004815769,
      "grad_norm": 1.402573585510254,
      "learning_rate": 9.400947368421054e-05,
      "loss": 0.5812,
      "step": 42783
    },
    {
      "epoch": 2.395789002127898,
      "grad_norm": 1.3187615871429443,
      "learning_rate": 9.40092105263158e-05,
      "loss": 0.4218,
      "step": 42784
    },
    {
      "epoch": 2.395844999440027,
      "grad_norm": 1.286502480506897,
      "learning_rate": 9.400894736842106e-05,
      "loss": 0.3305,
      "step": 42785
    },
    {
      "epoch": 2.395900996752156,
      "grad_norm": 1.0662767887115479,
      "learning_rate": 9.400868421052632e-05,
      "loss": 0.3683,
      "step": 42786
    },
    {
      "epoch": 2.395956994064285,
      "grad_norm": 1.2989014387130737,
      "learning_rate": 9.400842105263157e-05,
      "loss": 0.4997,
      "step": 42787
    },
    {
      "epoch": 2.396012991376414,
      "grad_norm": 1.1781584024429321,
      "learning_rate": 9.400815789473685e-05,
      "loss": 0.4547,
      "step": 42788
    },
    {
      "epoch": 2.396068988688543,
      "grad_norm": 1.5581963062286377,
      "learning_rate": 9.400789473684211e-05,
      "loss": 0.4403,
      "step": 42789
    },
    {
      "epoch": 2.396124986000672,
      "grad_norm": 1.0447933673858643,
      "learning_rate": 9.400763157894738e-05,
      "loss": 0.3238,
      "step": 42790
    },
    {
      "epoch": 2.396180983312801,
      "grad_norm": 1.1410270929336548,
      "learning_rate": 9.400736842105263e-05,
      "loss": 0.2957,
      "step": 42791
    },
    {
      "epoch": 2.39623698062493,
      "grad_norm": 1.290082573890686,
      "learning_rate": 9.40071052631579e-05,
      "loss": 0.4667,
      "step": 42792
    },
    {
      "epoch": 2.396292977937059,
      "grad_norm": 1.031409740447998,
      "learning_rate": 9.400684210526316e-05,
      "loss": 0.408,
      "step": 42793
    },
    {
      "epoch": 2.396348975249188,
      "grad_norm": 1.5443731546401978,
      "learning_rate": 9.400657894736843e-05,
      "loss": 0.761,
      "step": 42794
    },
    {
      "epoch": 2.3964049725613172,
      "grad_norm": 1.1245025396347046,
      "learning_rate": 9.40063157894737e-05,
      "loss": 0.2859,
      "step": 42795
    },
    {
      "epoch": 2.3964609698734463,
      "grad_norm": 1.1500089168548584,
      "learning_rate": 9.400605263157895e-05,
      "loss": 0.422,
      "step": 42796
    },
    {
      "epoch": 2.3965169671855753,
      "grad_norm": 1.508986234664917,
      "learning_rate": 9.400578947368421e-05,
      "loss": 0.589,
      "step": 42797
    },
    {
      "epoch": 2.3965729644977043,
      "grad_norm": 1.029876708984375,
      "learning_rate": 9.400552631578949e-05,
      "loss": 0.3333,
      "step": 42798
    },
    {
      "epoch": 2.3966289618098333,
      "grad_norm": 1.2869879007339478,
      "learning_rate": 9.400526315789475e-05,
      "loss": 0.5261,
      "step": 42799
    },
    {
      "epoch": 2.3966849591219623,
      "grad_norm": 1.2624270915985107,
      "learning_rate": 9.4005e-05,
      "loss": 0.3782,
      "step": 42800
    },
    {
      "epoch": 2.3967409564340914,
      "grad_norm": 1.1485683917999268,
      "learning_rate": 9.400473684210527e-05,
      "loss": 0.4573,
      "step": 42801
    },
    {
      "epoch": 2.3967969537462204,
      "grad_norm": 1.140895962715149,
      "learning_rate": 9.400447368421052e-05,
      "loss": 0.4054,
      "step": 42802
    },
    {
      "epoch": 2.3968529510583494,
      "grad_norm": 1.046038031578064,
      "learning_rate": 9.40042105263158e-05,
      "loss": 0.3919,
      "step": 42803
    },
    {
      "epoch": 2.3969089483704784,
      "grad_norm": 1.0892093181610107,
      "learning_rate": 9.400394736842106e-05,
      "loss": 0.3999,
      "step": 42804
    },
    {
      "epoch": 2.3969649456826074,
      "grad_norm": 0.9485438466072083,
      "learning_rate": 9.400368421052632e-05,
      "loss": 0.3217,
      "step": 42805
    },
    {
      "epoch": 2.397020942994736,
      "grad_norm": 1.319278597831726,
      "learning_rate": 9.400342105263158e-05,
      "loss": 0.4454,
      "step": 42806
    },
    {
      "epoch": 2.3970769403068655,
      "grad_norm": 1.1633769273757935,
      "learning_rate": 9.400315789473685e-05,
      "loss": 0.3626,
      "step": 42807
    },
    {
      "epoch": 2.397132937618994,
      "grad_norm": 1.0471646785736084,
      "learning_rate": 9.400289473684211e-05,
      "loss": 0.4241,
      "step": 42808
    },
    {
      "epoch": 2.3971889349311235,
      "grad_norm": 1.117857813835144,
      "learning_rate": 9.400263157894737e-05,
      "loss": 0.4194,
      "step": 42809
    },
    {
      "epoch": 2.397244932243252,
      "grad_norm": 1.3718310594558716,
      "learning_rate": 9.400236842105263e-05,
      "loss": 0.4809,
      "step": 42810
    },
    {
      "epoch": 2.3973009295553815,
      "grad_norm": 1.4049574136734009,
      "learning_rate": 9.40021052631579e-05,
      "loss": 0.4892,
      "step": 42811
    },
    {
      "epoch": 2.39735692686751,
      "grad_norm": 1.1366477012634277,
      "learning_rate": 9.400184210526316e-05,
      "loss": 0.3659,
      "step": 42812
    },
    {
      "epoch": 2.3974129241796396,
      "grad_norm": 1.2459602355957031,
      "learning_rate": 9.400157894736844e-05,
      "loss": 0.3963,
      "step": 42813
    },
    {
      "epoch": 2.397468921491768,
      "grad_norm": 1.1670079231262207,
      "learning_rate": 9.400131578947368e-05,
      "loss": 0.3154,
      "step": 42814
    },
    {
      "epoch": 2.3975249188038976,
      "grad_norm": 1.280793309211731,
      "learning_rate": 9.400105263157896e-05,
      "loss": 0.3072,
      "step": 42815
    },
    {
      "epoch": 2.397580916116026,
      "grad_norm": 1.2907675504684448,
      "learning_rate": 9.400078947368422e-05,
      "loss": 0.4184,
      "step": 42816
    },
    {
      "epoch": 2.3976369134281557,
      "grad_norm": 1.3640002012252808,
      "learning_rate": 9.400052631578949e-05,
      "loss": 0.4823,
      "step": 42817
    },
    {
      "epoch": 2.3976929107402842,
      "grad_norm": 1.5399712324142456,
      "learning_rate": 9.400026315789473e-05,
      "loss": 0.5543,
      "step": 42818
    },
    {
      "epoch": 2.3977489080524137,
      "grad_norm": 1.2145538330078125,
      "learning_rate": 9.4e-05,
      "loss": 0.3665,
      "step": 42819
    },
    {
      "epoch": 2.3978049053645423,
      "grad_norm": 1.2149910926818848,
      "learning_rate": 9.399973684210527e-05,
      "loss": 0.3596,
      "step": 42820
    },
    {
      "epoch": 2.3978609026766717,
      "grad_norm": 1.4620509147644043,
      "learning_rate": 9.399947368421053e-05,
      "loss": 0.4531,
      "step": 42821
    },
    {
      "epoch": 2.3979168999888003,
      "grad_norm": 1.2913742065429688,
      "learning_rate": 9.39992105263158e-05,
      "loss": 0.3643,
      "step": 42822
    },
    {
      "epoch": 2.3979728973009298,
      "grad_norm": 1.3904821872711182,
      "learning_rate": 9.399894736842105e-05,
      "loss": 0.3905,
      "step": 42823
    },
    {
      "epoch": 2.3980288946130583,
      "grad_norm": 1.1223782300949097,
      "learning_rate": 9.399868421052632e-05,
      "loss": 0.3046,
      "step": 42824
    },
    {
      "epoch": 2.398084891925188,
      "grad_norm": 1.0365148782730103,
      "learning_rate": 9.399842105263158e-05,
      "loss": 0.2993,
      "step": 42825
    },
    {
      "epoch": 2.3981408892373164,
      "grad_norm": 1.1204408407211304,
      "learning_rate": 9.399815789473685e-05,
      "loss": 0.3714,
      "step": 42826
    },
    {
      "epoch": 2.398196886549446,
      "grad_norm": 1.2624417543411255,
      "learning_rate": 9.399789473684211e-05,
      "loss": 0.3913,
      "step": 42827
    },
    {
      "epoch": 2.3982528838615744,
      "grad_norm": 1.0731436014175415,
      "learning_rate": 9.399763157894737e-05,
      "loss": 0.4125,
      "step": 42828
    },
    {
      "epoch": 2.398308881173704,
      "grad_norm": 1.4700095653533936,
      "learning_rate": 9.399736842105263e-05,
      "loss": 0.4481,
      "step": 42829
    },
    {
      "epoch": 2.3983648784858325,
      "grad_norm": 1.7787823677062988,
      "learning_rate": 9.39971052631579e-05,
      "loss": 0.321,
      "step": 42830
    },
    {
      "epoch": 2.398420875797962,
      "grad_norm": 1.1218106746673584,
      "learning_rate": 9.399684210526317e-05,
      "loss": 0.3995,
      "step": 42831
    },
    {
      "epoch": 2.3984768731100905,
      "grad_norm": 1.0557903051376343,
      "learning_rate": 9.399657894736843e-05,
      "loss": 0.3747,
      "step": 42832
    },
    {
      "epoch": 2.39853287042222,
      "grad_norm": 1.2945040464401245,
      "learning_rate": 9.399631578947368e-05,
      "loss": 0.547,
      "step": 42833
    },
    {
      "epoch": 2.3985888677343485,
      "grad_norm": 1.332270622253418,
      "learning_rate": 9.399605263157896e-05,
      "loss": 0.5035,
      "step": 42834
    },
    {
      "epoch": 2.398644865046478,
      "grad_norm": 0.9269304871559143,
      "learning_rate": 9.399578947368422e-05,
      "loss": 0.3199,
      "step": 42835
    },
    {
      "epoch": 2.3987008623586066,
      "grad_norm": 1.1885533332824707,
      "learning_rate": 9.399552631578948e-05,
      "loss": 0.4381,
      "step": 42836
    },
    {
      "epoch": 2.398756859670736,
      "grad_norm": 1.3204103708267212,
      "learning_rate": 9.399526315789474e-05,
      "loss": 0.4148,
      "step": 42837
    },
    {
      "epoch": 2.3988128569828646,
      "grad_norm": 1.2235286235809326,
      "learning_rate": 9.3995e-05,
      "loss": 0.4217,
      "step": 42838
    },
    {
      "epoch": 2.398868854294994,
      "grad_norm": 1.1143721342086792,
      "learning_rate": 9.399473684210527e-05,
      "loss": 0.3371,
      "step": 42839
    },
    {
      "epoch": 2.3989248516071227,
      "grad_norm": 1.0690478086471558,
      "learning_rate": 9.399447368421053e-05,
      "loss": 0.3062,
      "step": 42840
    },
    {
      "epoch": 2.398980848919252,
      "grad_norm": 1.2560781240463257,
      "learning_rate": 9.399421052631579e-05,
      "loss": 0.3706,
      "step": 42841
    },
    {
      "epoch": 2.3990368462313807,
      "grad_norm": 1.0828628540039062,
      "learning_rate": 9.399394736842105e-05,
      "loss": 0.4368,
      "step": 42842
    },
    {
      "epoch": 2.39909284354351,
      "grad_norm": 1.2249202728271484,
      "learning_rate": 9.399368421052632e-05,
      "loss": 0.4066,
      "step": 42843
    },
    {
      "epoch": 2.3991488408556387,
      "grad_norm": 1.4011949300765991,
      "learning_rate": 9.399342105263158e-05,
      "loss": 0.4649,
      "step": 42844
    },
    {
      "epoch": 2.399204838167768,
      "grad_norm": 1.3579423427581787,
      "learning_rate": 9.399315789473686e-05,
      "loss": 0.4712,
      "step": 42845
    },
    {
      "epoch": 2.3992608354798968,
      "grad_norm": 1.1339137554168701,
      "learning_rate": 9.39928947368421e-05,
      "loss": 0.4603,
      "step": 42846
    },
    {
      "epoch": 2.399316832792026,
      "grad_norm": 1.0509252548217773,
      "learning_rate": 9.399263157894738e-05,
      "loss": 0.4475,
      "step": 42847
    },
    {
      "epoch": 2.399372830104155,
      "grad_norm": 1.2603141069412231,
      "learning_rate": 9.399236842105263e-05,
      "loss": 0.4419,
      "step": 42848
    },
    {
      "epoch": 2.399428827416284,
      "grad_norm": 0.9640393257141113,
      "learning_rate": 9.399210526315791e-05,
      "loss": 0.426,
      "step": 42849
    },
    {
      "epoch": 2.399484824728413,
      "grad_norm": 1.1849970817565918,
      "learning_rate": 9.399184210526317e-05,
      "loss": 0.3901,
      "step": 42850
    },
    {
      "epoch": 2.399540822040542,
      "grad_norm": 1.0341917276382446,
      "learning_rate": 9.399157894736843e-05,
      "loss": 0.3527,
      "step": 42851
    },
    {
      "epoch": 2.399596819352671,
      "grad_norm": 1.13348388671875,
      "learning_rate": 9.399131578947369e-05,
      "loss": 0.3958,
      "step": 42852
    },
    {
      "epoch": 2.3996528166648,
      "grad_norm": 1.149422526359558,
      "learning_rate": 9.399105263157895e-05,
      "loss": 0.308,
      "step": 42853
    },
    {
      "epoch": 2.399708813976929,
      "grad_norm": 1.173488736152649,
      "learning_rate": 9.399078947368422e-05,
      "loss": 0.3812,
      "step": 42854
    },
    {
      "epoch": 2.399764811289058,
      "grad_norm": 1.33085298538208,
      "learning_rate": 9.399052631578947e-05,
      "loss": 0.4558,
      "step": 42855
    },
    {
      "epoch": 2.399820808601187,
      "grad_norm": 1.2319053411483765,
      "learning_rate": 9.399026315789474e-05,
      "loss": 0.4156,
      "step": 42856
    },
    {
      "epoch": 2.399876805913316,
      "grad_norm": 1.6285816431045532,
      "learning_rate": 9.399e-05,
      "loss": 0.4663,
      "step": 42857
    },
    {
      "epoch": 2.399932803225445,
      "grad_norm": 1.435790777206421,
      "learning_rate": 9.398973684210527e-05,
      "loss": 0.4234,
      "step": 42858
    },
    {
      "epoch": 2.399988800537574,
      "grad_norm": 1.1825608015060425,
      "learning_rate": 9.398947368421053e-05,
      "loss": 0.3957,
      "step": 42859
    },
    {
      "epoch": 2.400044797849703,
      "grad_norm": 1.055742859840393,
      "learning_rate": 9.398921052631579e-05,
      "loss": 0.3827,
      "step": 42860
    },
    {
      "epoch": 2.400100795161832,
      "grad_norm": 1.2493926286697388,
      "learning_rate": 9.398894736842105e-05,
      "loss": 0.5136,
      "step": 42861
    },
    {
      "epoch": 2.400156792473961,
      "grad_norm": 0.9950408935546875,
      "learning_rate": 9.398868421052633e-05,
      "loss": 0.3472,
      "step": 42862
    },
    {
      "epoch": 2.40021278978609,
      "grad_norm": 1.24483323097229,
      "learning_rate": 9.398842105263159e-05,
      "loss": 0.4914,
      "step": 42863
    },
    {
      "epoch": 2.400268787098219,
      "grad_norm": 1.0699388980865479,
      "learning_rate": 9.398815789473684e-05,
      "loss": 0.4175,
      "step": 42864
    },
    {
      "epoch": 2.400324784410348,
      "grad_norm": 1.0724236965179443,
      "learning_rate": 9.39878947368421e-05,
      "loss": 0.4878,
      "step": 42865
    },
    {
      "epoch": 2.400380781722477,
      "grad_norm": 1.6201810836791992,
      "learning_rate": 9.398763157894738e-05,
      "loss": 0.2849,
      "step": 42866
    },
    {
      "epoch": 2.400436779034606,
      "grad_norm": 0.9813246726989746,
      "learning_rate": 9.398736842105264e-05,
      "loss": 0.3453,
      "step": 42867
    },
    {
      "epoch": 2.400492776346735,
      "grad_norm": 1.2604049444198608,
      "learning_rate": 9.398710526315791e-05,
      "loss": 0.4922,
      "step": 42868
    },
    {
      "epoch": 2.400548773658864,
      "grad_norm": 1.163133144378662,
      "learning_rate": 9.398684210526316e-05,
      "loss": 0.4883,
      "step": 42869
    },
    {
      "epoch": 2.4006047709709932,
      "grad_norm": 1.082651972770691,
      "learning_rate": 9.398657894736843e-05,
      "loss": 0.3854,
      "step": 42870
    },
    {
      "epoch": 2.4006607682831222,
      "grad_norm": 1.1151775121688843,
      "learning_rate": 9.398631578947369e-05,
      "loss": 0.3727,
      "step": 42871
    },
    {
      "epoch": 2.4007167655952513,
      "grad_norm": 1.364314079284668,
      "learning_rate": 9.398605263157895e-05,
      "loss": 0.3411,
      "step": 42872
    },
    {
      "epoch": 2.4007727629073803,
      "grad_norm": 1.1152544021606445,
      "learning_rate": 9.398578947368421e-05,
      "loss": 0.4127,
      "step": 42873
    },
    {
      "epoch": 2.4008287602195093,
      "grad_norm": 1.131212830543518,
      "learning_rate": 9.398552631578947e-05,
      "loss": 0.33,
      "step": 42874
    },
    {
      "epoch": 2.4008847575316383,
      "grad_norm": 1.6210938692092896,
      "learning_rate": 9.398526315789474e-05,
      "loss": 0.4159,
      "step": 42875
    },
    {
      "epoch": 2.4009407548437673,
      "grad_norm": 1.2321794033050537,
      "learning_rate": 9.3985e-05,
      "loss": 0.4258,
      "step": 42876
    },
    {
      "epoch": 2.4009967521558964,
      "grad_norm": 1.2900965213775635,
      "learning_rate": 9.398473684210528e-05,
      "loss": 0.4819,
      "step": 42877
    },
    {
      "epoch": 2.4010527494680254,
      "grad_norm": 1.0988104343414307,
      "learning_rate": 9.398447368421052e-05,
      "loss": 0.5025,
      "step": 42878
    },
    {
      "epoch": 2.4011087467801544,
      "grad_norm": 1.0248181819915771,
      "learning_rate": 9.39842105263158e-05,
      "loss": 0.3141,
      "step": 42879
    },
    {
      "epoch": 2.4011647440922834,
      "grad_norm": 1.2771834135055542,
      "learning_rate": 9.398394736842105e-05,
      "loss": 0.4338,
      "step": 42880
    },
    {
      "epoch": 2.4012207414044124,
      "grad_norm": 1.1962817907333374,
      "learning_rate": 9.398368421052633e-05,
      "loss": 0.3978,
      "step": 42881
    },
    {
      "epoch": 2.4012767387165415,
      "grad_norm": 1.3730812072753906,
      "learning_rate": 9.398342105263159e-05,
      "loss": 0.3833,
      "step": 42882
    },
    {
      "epoch": 2.4013327360286705,
      "grad_norm": 1.0591527223587036,
      "learning_rate": 9.398315789473685e-05,
      "loss": 0.3921,
      "step": 42883
    },
    {
      "epoch": 2.4013887333407995,
      "grad_norm": 1.1018654108047485,
      "learning_rate": 9.398289473684211e-05,
      "loss": 0.3963,
      "step": 42884
    },
    {
      "epoch": 2.4014447306529285,
      "grad_norm": 1.2255334854125977,
      "learning_rate": 9.398263157894738e-05,
      "loss": 0.4485,
      "step": 42885
    },
    {
      "epoch": 2.4015007279650575,
      "grad_norm": 1.2397135496139526,
      "learning_rate": 9.398236842105264e-05,
      "loss": 0.4718,
      "step": 42886
    },
    {
      "epoch": 2.4015567252771866,
      "grad_norm": 1.2437766790390015,
      "learning_rate": 9.39821052631579e-05,
      "loss": 0.4592,
      "step": 42887
    },
    {
      "epoch": 2.4016127225893156,
      "grad_norm": 1.1579355001449585,
      "learning_rate": 9.398184210526316e-05,
      "loss": 0.3638,
      "step": 42888
    },
    {
      "epoch": 2.4016687199014446,
      "grad_norm": 1.391806960105896,
      "learning_rate": 9.398157894736842e-05,
      "loss": 0.3318,
      "step": 42889
    },
    {
      "epoch": 2.4017247172135736,
      "grad_norm": 1.3346620798110962,
      "learning_rate": 9.398131578947369e-05,
      "loss": 0.3775,
      "step": 42890
    },
    {
      "epoch": 2.4017807145257026,
      "grad_norm": 1.1103512048721313,
      "learning_rate": 9.398105263157895e-05,
      "loss": 0.4334,
      "step": 42891
    },
    {
      "epoch": 2.4018367118378317,
      "grad_norm": 1.2742112874984741,
      "learning_rate": 9.398078947368421e-05,
      "loss": 0.4332,
      "step": 42892
    },
    {
      "epoch": 2.4018927091499607,
      "grad_norm": 1.362546443939209,
      "learning_rate": 9.398052631578947e-05,
      "loss": 0.4832,
      "step": 42893
    },
    {
      "epoch": 2.4019487064620897,
      "grad_norm": 1.0513864755630493,
      "learning_rate": 9.398026315789475e-05,
      "loss": 0.3855,
      "step": 42894
    },
    {
      "epoch": 2.4020047037742187,
      "grad_norm": 1.298464059829712,
      "learning_rate": 9.398e-05,
      "loss": 0.43,
      "step": 42895
    },
    {
      "epoch": 2.4020607010863477,
      "grad_norm": 1.3281333446502686,
      "learning_rate": 9.397973684210526e-05,
      "loss": 0.4325,
      "step": 42896
    },
    {
      "epoch": 2.4021166983984767,
      "grad_norm": 1.0945121049880981,
      "learning_rate": 9.397947368421052e-05,
      "loss": 0.3481,
      "step": 42897
    },
    {
      "epoch": 2.4021726957106058,
      "grad_norm": 1.1759122610092163,
      "learning_rate": 9.39792105263158e-05,
      "loss": 0.402,
      "step": 42898
    },
    {
      "epoch": 2.402228693022735,
      "grad_norm": 1.086368203163147,
      "learning_rate": 9.397894736842106e-05,
      "loss": 0.3861,
      "step": 42899
    },
    {
      "epoch": 2.402284690334864,
      "grad_norm": 1.3255102634429932,
      "learning_rate": 9.397868421052633e-05,
      "loss": 0.386,
      "step": 42900
    },
    {
      "epoch": 2.402340687646993,
      "grad_norm": 1.1373614072799683,
      "learning_rate": 9.397842105263158e-05,
      "loss": 0.3563,
      "step": 42901
    },
    {
      "epoch": 2.402396684959122,
      "grad_norm": 1.3802698850631714,
      "learning_rate": 9.397815789473685e-05,
      "loss": 0.3763,
      "step": 42902
    },
    {
      "epoch": 2.402452682271251,
      "grad_norm": 1.1138663291931152,
      "learning_rate": 9.397789473684211e-05,
      "loss": 0.3746,
      "step": 42903
    },
    {
      "epoch": 2.40250867958338,
      "grad_norm": 1.2412904500961304,
      "learning_rate": 9.397763157894738e-05,
      "loss": 0.414,
      "step": 42904
    },
    {
      "epoch": 2.402564676895509,
      "grad_norm": 1.4699715375900269,
      "learning_rate": 9.397736842105264e-05,
      "loss": 0.3551,
      "step": 42905
    },
    {
      "epoch": 2.402620674207638,
      "grad_norm": 1.1739765405654907,
      "learning_rate": 9.397710526315789e-05,
      "loss": 0.5009,
      "step": 42906
    },
    {
      "epoch": 2.402676671519767,
      "grad_norm": 1.2483924627304077,
      "learning_rate": 9.397684210526316e-05,
      "loss": 0.4668,
      "step": 42907
    },
    {
      "epoch": 2.402732668831896,
      "grad_norm": 1.153741478919983,
      "learning_rate": 9.397657894736842e-05,
      "loss": 0.5129,
      "step": 42908
    },
    {
      "epoch": 2.402788666144025,
      "grad_norm": 1.2788336277008057,
      "learning_rate": 9.39763157894737e-05,
      "loss": 0.3661,
      "step": 42909
    },
    {
      "epoch": 2.402844663456154,
      "grad_norm": 1.0794240236282349,
      "learning_rate": 9.397605263157894e-05,
      "loss": 0.3934,
      "step": 42910
    },
    {
      "epoch": 2.402900660768283,
      "grad_norm": 1.0847769975662231,
      "learning_rate": 9.397578947368421e-05,
      "loss": 0.4521,
      "step": 42911
    },
    {
      "epoch": 2.402956658080412,
      "grad_norm": 1.2194803953170776,
      "learning_rate": 9.397552631578947e-05,
      "loss": 0.4577,
      "step": 42912
    },
    {
      "epoch": 2.403012655392541,
      "grad_norm": 1.2085070610046387,
      "learning_rate": 9.397526315789475e-05,
      "loss": 0.3837,
      "step": 42913
    },
    {
      "epoch": 2.40306865270467,
      "grad_norm": 1.1162357330322266,
      "learning_rate": 9.397500000000001e-05,
      "loss": 0.3928,
      "step": 42914
    },
    {
      "epoch": 2.403124650016799,
      "grad_norm": 1.1461759805679321,
      "learning_rate": 9.397473684210527e-05,
      "loss": 0.3563,
      "step": 42915
    },
    {
      "epoch": 2.403180647328928,
      "grad_norm": 1.1112070083618164,
      "learning_rate": 9.397447368421053e-05,
      "loss": 0.4599,
      "step": 42916
    },
    {
      "epoch": 2.403236644641057,
      "grad_norm": 1.3734626770019531,
      "learning_rate": 9.39742105263158e-05,
      "loss": 0.5404,
      "step": 42917
    },
    {
      "epoch": 2.403292641953186,
      "grad_norm": 1.5203425884246826,
      "learning_rate": 9.397394736842106e-05,
      "loss": 0.406,
      "step": 42918
    },
    {
      "epoch": 2.403348639265315,
      "grad_norm": 1.3640928268432617,
      "learning_rate": 9.397368421052632e-05,
      "loss": 0.6159,
      "step": 42919
    },
    {
      "epoch": 2.403404636577444,
      "grad_norm": 1.1265184879302979,
      "learning_rate": 9.397342105263158e-05,
      "loss": 0.415,
      "step": 42920
    },
    {
      "epoch": 2.403460633889573,
      "grad_norm": 1.3865538835525513,
      "learning_rate": 9.397315789473685e-05,
      "loss": 0.4299,
      "step": 42921
    },
    {
      "epoch": 2.4035166312017022,
      "grad_norm": 1.5282878875732422,
      "learning_rate": 9.397289473684211e-05,
      "loss": 0.3583,
      "step": 42922
    },
    {
      "epoch": 2.4035726285138312,
      "grad_norm": 1.1397448778152466,
      "learning_rate": 9.397263157894737e-05,
      "loss": 0.4033,
      "step": 42923
    },
    {
      "epoch": 2.4036286258259603,
      "grad_norm": 1.0645920038223267,
      "learning_rate": 9.397236842105263e-05,
      "loss": 0.3553,
      "step": 42924
    },
    {
      "epoch": 2.4036846231380893,
      "grad_norm": 1.0781527757644653,
      "learning_rate": 9.397210526315789e-05,
      "loss": 0.3036,
      "step": 42925
    },
    {
      "epoch": 2.4037406204502183,
      "grad_norm": 1.1391663551330566,
      "learning_rate": 9.397184210526316e-05,
      "loss": 0.3779,
      "step": 42926
    },
    {
      "epoch": 2.4037966177623473,
      "grad_norm": 1.1351583003997803,
      "learning_rate": 9.397157894736842e-05,
      "loss": 0.4974,
      "step": 42927
    },
    {
      "epoch": 2.4038526150744763,
      "grad_norm": 1.0221221446990967,
      "learning_rate": 9.397131578947368e-05,
      "loss": 0.3608,
      "step": 42928
    },
    {
      "epoch": 2.4039086123866054,
      "grad_norm": 1.3087012767791748,
      "learning_rate": 9.397105263157894e-05,
      "loss": 0.4143,
      "step": 42929
    },
    {
      "epoch": 2.4039646096987344,
      "grad_norm": 1.138741374015808,
      "learning_rate": 9.397078947368422e-05,
      "loss": 0.4244,
      "step": 42930
    },
    {
      "epoch": 2.4040206070108634,
      "grad_norm": 1.221858024597168,
      "learning_rate": 9.397052631578948e-05,
      "loss": 0.4279,
      "step": 42931
    },
    {
      "epoch": 2.4040766043229924,
      "grad_norm": 1.3469194173812866,
      "learning_rate": 9.397026315789475e-05,
      "loss": 0.4146,
      "step": 42932
    },
    {
      "epoch": 2.4041326016351214,
      "grad_norm": 1.137859582901001,
      "learning_rate": 9.397e-05,
      "loss": 0.4299,
      "step": 42933
    },
    {
      "epoch": 2.4041885989472505,
      "grad_norm": 1.1727951765060425,
      "learning_rate": 9.396973684210527e-05,
      "loss": 0.3589,
      "step": 42934
    },
    {
      "epoch": 2.4042445962593795,
      "grad_norm": 1.0487192869186401,
      "learning_rate": 9.396947368421053e-05,
      "loss": 0.4531,
      "step": 42935
    },
    {
      "epoch": 2.4043005935715085,
      "grad_norm": 1.4081909656524658,
      "learning_rate": 9.39692105263158e-05,
      "loss": 0.4299,
      "step": 42936
    },
    {
      "epoch": 2.4043565908836375,
      "grad_norm": 1.3632649183273315,
      "learning_rate": 9.396894736842106e-05,
      "loss": 0.3999,
      "step": 42937
    },
    {
      "epoch": 2.4044125881957665,
      "grad_norm": 1.4034035205841064,
      "learning_rate": 9.396868421052632e-05,
      "loss": 0.4279,
      "step": 42938
    },
    {
      "epoch": 2.4044685855078956,
      "grad_norm": 1.3694361448287964,
      "learning_rate": 9.396842105263158e-05,
      "loss": 0.4226,
      "step": 42939
    },
    {
      "epoch": 2.4045245828200246,
      "grad_norm": 1.2090771198272705,
      "learning_rate": 9.396815789473684e-05,
      "loss": 0.4353,
      "step": 42940
    },
    {
      "epoch": 2.4045805801321536,
      "grad_norm": 1.2322231531143188,
      "learning_rate": 9.396789473684211e-05,
      "loss": 0.3873,
      "step": 42941
    },
    {
      "epoch": 2.4046365774442826,
      "grad_norm": 1.3113725185394287,
      "learning_rate": 9.396763157894737e-05,
      "loss": 0.4536,
      "step": 42942
    },
    {
      "epoch": 2.4046925747564116,
      "grad_norm": 1.192201852798462,
      "learning_rate": 9.396736842105263e-05,
      "loss": 0.384,
      "step": 42943
    },
    {
      "epoch": 2.4047485720685406,
      "grad_norm": 1.15830397605896,
      "learning_rate": 9.39671052631579e-05,
      "loss": 0.3974,
      "step": 42944
    },
    {
      "epoch": 2.4048045693806697,
      "grad_norm": 1.2198481559753418,
      "learning_rate": 9.396684210526317e-05,
      "loss": 0.3559,
      "step": 42945
    },
    {
      "epoch": 2.4048605666927987,
      "grad_norm": 1.2232214212417603,
      "learning_rate": 9.396657894736843e-05,
      "loss": 0.4375,
      "step": 42946
    },
    {
      "epoch": 2.4049165640049277,
      "grad_norm": 1.1937845945358276,
      "learning_rate": 9.396631578947369e-05,
      "loss": 0.532,
      "step": 42947
    },
    {
      "epoch": 2.4049725613170567,
      "grad_norm": 1.3005129098892212,
      "learning_rate": 9.396605263157895e-05,
      "loss": 0.3676,
      "step": 42948
    },
    {
      "epoch": 2.4050285586291857,
      "grad_norm": 1.0055527687072754,
      "learning_rate": 9.396578947368422e-05,
      "loss": 0.3824,
      "step": 42949
    },
    {
      "epoch": 2.4050845559413148,
      "grad_norm": 1.0742515325546265,
      "learning_rate": 9.396552631578948e-05,
      "loss": 0.402,
      "step": 42950
    },
    {
      "epoch": 2.405140553253444,
      "grad_norm": 1.2483686208724976,
      "learning_rate": 9.396526315789474e-05,
      "loss": 0.3754,
      "step": 42951
    },
    {
      "epoch": 2.405196550565573,
      "grad_norm": 1.295030951499939,
      "learning_rate": 9.3965e-05,
      "loss": 0.3978,
      "step": 42952
    },
    {
      "epoch": 2.405252547877702,
      "grad_norm": 1.2300734519958496,
      "learning_rate": 9.396473684210527e-05,
      "loss": 0.3723,
      "step": 42953
    },
    {
      "epoch": 2.405308545189831,
      "grad_norm": 1.435346245765686,
      "learning_rate": 9.396447368421053e-05,
      "loss": 0.5706,
      "step": 42954
    },
    {
      "epoch": 2.40536454250196,
      "grad_norm": 1.1682757139205933,
      "learning_rate": 9.39642105263158e-05,
      "loss": 0.3188,
      "step": 42955
    },
    {
      "epoch": 2.405420539814089,
      "grad_norm": 1.174091100692749,
      "learning_rate": 9.396394736842105e-05,
      "loss": 0.4594,
      "step": 42956
    },
    {
      "epoch": 2.405476537126218,
      "grad_norm": 1.4960664510726929,
      "learning_rate": 9.396368421052631e-05,
      "loss": 0.4271,
      "step": 42957
    },
    {
      "epoch": 2.405532534438347,
      "grad_norm": 1.2805312871932983,
      "learning_rate": 9.396342105263158e-05,
      "loss": 0.4597,
      "step": 42958
    },
    {
      "epoch": 2.405588531750476,
      "grad_norm": 1.1715357303619385,
      "learning_rate": 9.396315789473684e-05,
      "loss": 0.4632,
      "step": 42959
    },
    {
      "epoch": 2.405644529062605,
      "grad_norm": 1.0852584838867188,
      "learning_rate": 9.396289473684212e-05,
      "loss": 0.4034,
      "step": 42960
    },
    {
      "epoch": 2.405700526374734,
      "grad_norm": 1.5558000802993774,
      "learning_rate": 9.396263157894736e-05,
      "loss": 0.4439,
      "step": 42961
    },
    {
      "epoch": 2.405756523686863,
      "grad_norm": 1.12416672706604,
      "learning_rate": 9.396236842105264e-05,
      "loss": 0.413,
      "step": 42962
    },
    {
      "epoch": 2.405812520998992,
      "grad_norm": 1.1439340114593506,
      "learning_rate": 9.39621052631579e-05,
      "loss": 0.3941,
      "step": 42963
    },
    {
      "epoch": 2.405868518311121,
      "grad_norm": 1.1268935203552246,
      "learning_rate": 9.396184210526317e-05,
      "loss": 0.3592,
      "step": 42964
    },
    {
      "epoch": 2.40592451562325,
      "grad_norm": 1.2831907272338867,
      "learning_rate": 9.396157894736842e-05,
      "loss": 0.4107,
      "step": 42965
    },
    {
      "epoch": 2.405980512935379,
      "grad_norm": 1.372144341468811,
      "learning_rate": 9.396131578947369e-05,
      "loss": 0.5267,
      "step": 42966
    },
    {
      "epoch": 2.406036510247508,
      "grad_norm": 1.155949354171753,
      "learning_rate": 9.396105263157895e-05,
      "loss": 0.4104,
      "step": 42967
    },
    {
      "epoch": 2.406092507559637,
      "grad_norm": 1.1827507019042969,
      "learning_rate": 9.396078947368422e-05,
      "loss": 0.3735,
      "step": 42968
    },
    {
      "epoch": 2.406148504871766,
      "grad_norm": 1.2791862487792969,
      "learning_rate": 9.396052631578948e-05,
      "loss": 0.4888,
      "step": 42969
    },
    {
      "epoch": 2.406204502183895,
      "grad_norm": 1.4813575744628906,
      "learning_rate": 9.396026315789474e-05,
      "loss": 0.4679,
      "step": 42970
    },
    {
      "epoch": 2.406260499496024,
      "grad_norm": 1.0876219272613525,
      "learning_rate": 9.396e-05,
      "loss": 0.3157,
      "step": 42971
    },
    {
      "epoch": 2.406316496808153,
      "grad_norm": 1.0424891710281372,
      "learning_rate": 9.395973684210527e-05,
      "loss": 0.3645,
      "step": 42972
    },
    {
      "epoch": 2.406372494120282,
      "grad_norm": 1.0234819650650024,
      "learning_rate": 9.395947368421053e-05,
      "loss": 0.322,
      "step": 42973
    },
    {
      "epoch": 2.406428491432411,
      "grad_norm": 1.3803379535675049,
      "learning_rate": 9.39592105263158e-05,
      "loss": 0.4928,
      "step": 42974
    },
    {
      "epoch": 2.4064844887445402,
      "grad_norm": 1.032228708267212,
      "learning_rate": 9.395894736842105e-05,
      "loss": 0.3183,
      "step": 42975
    },
    {
      "epoch": 2.4065404860566693,
      "grad_norm": 1.038776159286499,
      "learning_rate": 9.395868421052631e-05,
      "loss": 0.3472,
      "step": 42976
    },
    {
      "epoch": 2.4065964833687983,
      "grad_norm": 1.3813427686691284,
      "learning_rate": 9.395842105263159e-05,
      "loss": 0.3408,
      "step": 42977
    },
    {
      "epoch": 2.4066524806809273,
      "grad_norm": 1.138115406036377,
      "learning_rate": 9.395815789473685e-05,
      "loss": 0.3637,
      "step": 42978
    },
    {
      "epoch": 2.4067084779930563,
      "grad_norm": 0.9865298271179199,
      "learning_rate": 9.39578947368421e-05,
      "loss": 0.3841,
      "step": 42979
    },
    {
      "epoch": 2.4067644753051853,
      "grad_norm": 1.16624915599823,
      "learning_rate": 9.395763157894737e-05,
      "loss": 0.2743,
      "step": 42980
    },
    {
      "epoch": 2.4068204726173144,
      "grad_norm": 1.3586690425872803,
      "learning_rate": 9.395736842105264e-05,
      "loss": 0.4359,
      "step": 42981
    },
    {
      "epoch": 2.4068764699294434,
      "grad_norm": 1.3061840534210205,
      "learning_rate": 9.39571052631579e-05,
      "loss": 0.3645,
      "step": 42982
    },
    {
      "epoch": 2.4069324672415724,
      "grad_norm": 1.1973698139190674,
      "learning_rate": 9.395684210526316e-05,
      "loss": 0.4012,
      "step": 42983
    },
    {
      "epoch": 2.4069884645537014,
      "grad_norm": 1.3382766246795654,
      "learning_rate": 9.395657894736842e-05,
      "loss": 0.3996,
      "step": 42984
    },
    {
      "epoch": 2.4070444618658304,
      "grad_norm": 1.1486364603042603,
      "learning_rate": 9.395631578947369e-05,
      "loss": 0.3254,
      "step": 42985
    },
    {
      "epoch": 2.4071004591779595,
      "grad_norm": 1.4335806369781494,
      "learning_rate": 9.395605263157895e-05,
      "loss": 0.4245,
      "step": 42986
    },
    {
      "epoch": 2.4071564564900885,
      "grad_norm": 1.1917120218276978,
      "learning_rate": 9.395578947368423e-05,
      "loss": 0.5286,
      "step": 42987
    },
    {
      "epoch": 2.4072124538022175,
      "grad_norm": 1.2304513454437256,
      "learning_rate": 9.395552631578947e-05,
      "loss": 0.4125,
      "step": 42988
    },
    {
      "epoch": 2.4072684511143465,
      "grad_norm": 1.0390325784683228,
      "learning_rate": 9.395526315789474e-05,
      "loss": 0.3033,
      "step": 42989
    },
    {
      "epoch": 2.4073244484264755,
      "grad_norm": 1.2280079126358032,
      "learning_rate": 9.3955e-05,
      "loss": 0.3685,
      "step": 42990
    },
    {
      "epoch": 2.4073804457386045,
      "grad_norm": 1.1059726476669312,
      "learning_rate": 9.395473684210528e-05,
      "loss": 0.2905,
      "step": 42991
    },
    {
      "epoch": 2.4074364430507336,
      "grad_norm": 1.502891182899475,
      "learning_rate": 9.395447368421054e-05,
      "loss": 0.4896,
      "step": 42992
    },
    {
      "epoch": 2.4074924403628626,
      "grad_norm": 1.170148491859436,
      "learning_rate": 9.395421052631578e-05,
      "loss": 0.3585,
      "step": 42993
    },
    {
      "epoch": 2.4075484376749916,
      "grad_norm": 1.3524712324142456,
      "learning_rate": 9.395394736842106e-05,
      "loss": 0.4593,
      "step": 42994
    },
    {
      "epoch": 2.4076044349871206,
      "grad_norm": 1.226854681968689,
      "learning_rate": 9.395368421052632e-05,
      "loss": 0.3867,
      "step": 42995
    },
    {
      "epoch": 2.4076604322992496,
      "grad_norm": 1.2388516664505005,
      "learning_rate": 9.395342105263159e-05,
      "loss": 0.4901,
      "step": 42996
    },
    {
      "epoch": 2.4077164296113787,
      "grad_norm": 1.0370386838912964,
      "learning_rate": 9.395315789473685e-05,
      "loss": 0.364,
      "step": 42997
    },
    {
      "epoch": 2.4077724269235077,
      "grad_norm": 1.2158148288726807,
      "learning_rate": 9.395289473684211e-05,
      "loss": 0.3682,
      "step": 42998
    },
    {
      "epoch": 2.4078284242356367,
      "grad_norm": 1.2620946168899536,
      "learning_rate": 9.395263157894737e-05,
      "loss": 0.4716,
      "step": 42999
    },
    {
      "epoch": 2.4078844215477657,
      "grad_norm": 1.3492554426193237,
      "learning_rate": 9.395236842105264e-05,
      "loss": 0.3541,
      "step": 43000
    },
    {
      "epoch": 2.4079404188598947,
      "grad_norm": 1.1032800674438477,
      "learning_rate": 9.39521052631579e-05,
      "loss": 0.526,
      "step": 43001
    },
    {
      "epoch": 2.4079964161720238,
      "grad_norm": 1.1166880130767822,
      "learning_rate": 9.395184210526316e-05,
      "loss": 0.4579,
      "step": 43002
    },
    {
      "epoch": 2.4080524134841528,
      "grad_norm": 1.3685531616210938,
      "learning_rate": 9.395157894736842e-05,
      "loss": 0.4222,
      "step": 43003
    },
    {
      "epoch": 2.408108410796282,
      "grad_norm": 1.185243010520935,
      "learning_rate": 9.39513157894737e-05,
      "loss": 0.3348,
      "step": 43004
    },
    {
      "epoch": 2.408164408108411,
      "grad_norm": 1.3580867052078247,
      "learning_rate": 9.395105263157895e-05,
      "loss": 0.4418,
      "step": 43005
    },
    {
      "epoch": 2.40822040542054,
      "grad_norm": 1.2270675897598267,
      "learning_rate": 9.395078947368421e-05,
      "loss": 0.3925,
      "step": 43006
    },
    {
      "epoch": 2.408276402732669,
      "grad_norm": 1.0914605855941772,
      "learning_rate": 9.395052631578947e-05,
      "loss": 0.3445,
      "step": 43007
    },
    {
      "epoch": 2.408332400044798,
      "grad_norm": 1.897592544555664,
      "learning_rate": 9.395026315789475e-05,
      "loss": 0.4612,
      "step": 43008
    },
    {
      "epoch": 2.408388397356927,
      "grad_norm": 1.2961806058883667,
      "learning_rate": 9.395000000000001e-05,
      "loss": 0.4088,
      "step": 43009
    },
    {
      "epoch": 2.408444394669056,
      "grad_norm": 1.277518391609192,
      "learning_rate": 9.394973684210527e-05,
      "loss": 0.3839,
      "step": 43010
    },
    {
      "epoch": 2.408500391981185,
      "grad_norm": 1.1530386209487915,
      "learning_rate": 9.394947368421053e-05,
      "loss": 0.3807,
      "step": 43011
    },
    {
      "epoch": 2.408556389293314,
      "grad_norm": 1.3407140970230103,
      "learning_rate": 9.394921052631579e-05,
      "loss": 0.3558,
      "step": 43012
    },
    {
      "epoch": 2.408612386605443,
      "grad_norm": 1.0811842679977417,
      "learning_rate": 9.394894736842106e-05,
      "loss": 0.3878,
      "step": 43013
    },
    {
      "epoch": 2.408668383917572,
      "grad_norm": 1.2993879318237305,
      "learning_rate": 9.394868421052632e-05,
      "loss": 0.5521,
      "step": 43014
    },
    {
      "epoch": 2.408724381229701,
      "grad_norm": 1.1716740131378174,
      "learning_rate": 9.394842105263159e-05,
      "loss": 0.4078,
      "step": 43015
    },
    {
      "epoch": 2.40878037854183,
      "grad_norm": 1.3194535970687866,
      "learning_rate": 9.394815789473684e-05,
      "loss": 0.6376,
      "step": 43016
    },
    {
      "epoch": 2.408836375853959,
      "grad_norm": 1.1337637901306152,
      "learning_rate": 9.394789473684211e-05,
      "loss": 0.3578,
      "step": 43017
    },
    {
      "epoch": 2.408892373166088,
      "grad_norm": 1.2077672481536865,
      "learning_rate": 9.394763157894737e-05,
      "loss": 0.3894,
      "step": 43018
    },
    {
      "epoch": 2.408948370478217,
      "grad_norm": 1.4268771409988403,
      "learning_rate": 9.394736842105264e-05,
      "loss": 0.4158,
      "step": 43019
    },
    {
      "epoch": 2.409004367790346,
      "grad_norm": 1.1748284101486206,
      "learning_rate": 9.394710526315789e-05,
      "loss": 0.3332,
      "step": 43020
    },
    {
      "epoch": 2.409060365102475,
      "grad_norm": 1.2950421571731567,
      "learning_rate": 9.394684210526316e-05,
      "loss": 0.4042,
      "step": 43021
    },
    {
      "epoch": 2.409116362414604,
      "grad_norm": 1.2509827613830566,
      "learning_rate": 9.394657894736842e-05,
      "loss": 0.4134,
      "step": 43022
    },
    {
      "epoch": 2.409172359726733,
      "grad_norm": 1.3471120595932007,
      "learning_rate": 9.39463157894737e-05,
      "loss": 0.5651,
      "step": 43023
    },
    {
      "epoch": 2.409228357038862,
      "grad_norm": 1.3577544689178467,
      "learning_rate": 9.394605263157896e-05,
      "loss": 0.498,
      "step": 43024
    },
    {
      "epoch": 2.409284354350991,
      "grad_norm": 1.008226990699768,
      "learning_rate": 9.394578947368422e-05,
      "loss": 0.2879,
      "step": 43025
    },
    {
      "epoch": 2.40934035166312,
      "grad_norm": 1.1439552307128906,
      "learning_rate": 9.394552631578948e-05,
      "loss": 0.4208,
      "step": 43026
    },
    {
      "epoch": 2.4093963489752492,
      "grad_norm": 1.055285930633545,
      "learning_rate": 9.394526315789474e-05,
      "loss": 0.3195,
      "step": 43027
    },
    {
      "epoch": 2.4094523462873783,
      "grad_norm": 0.9813190698623657,
      "learning_rate": 9.394500000000001e-05,
      "loss": 0.2721,
      "step": 43028
    },
    {
      "epoch": 2.4095083435995073,
      "grad_norm": 1.2558999061584473,
      "learning_rate": 9.394473684210527e-05,
      "loss": 0.4237,
      "step": 43029
    },
    {
      "epoch": 2.4095643409116363,
      "grad_norm": 1.0759652853012085,
      "learning_rate": 9.394447368421053e-05,
      "loss": 0.2916,
      "step": 43030
    },
    {
      "epoch": 2.4096203382237653,
      "grad_norm": 0.8966822624206543,
      "learning_rate": 9.394421052631579e-05,
      "loss": 0.3355,
      "step": 43031
    },
    {
      "epoch": 2.4096763355358943,
      "grad_norm": 1.3577170372009277,
      "learning_rate": 9.394394736842106e-05,
      "loss": 0.3826,
      "step": 43032
    },
    {
      "epoch": 2.4097323328480234,
      "grad_norm": 1.3411133289337158,
      "learning_rate": 9.394368421052632e-05,
      "loss": 0.3963,
      "step": 43033
    },
    {
      "epoch": 2.4097883301601524,
      "grad_norm": 1.043069839477539,
      "learning_rate": 9.394342105263158e-05,
      "loss": 0.3662,
      "step": 43034
    },
    {
      "epoch": 2.4098443274722814,
      "grad_norm": 1.3937034606933594,
      "learning_rate": 9.394315789473684e-05,
      "loss": 0.5147,
      "step": 43035
    },
    {
      "epoch": 2.4099003247844104,
      "grad_norm": 1.065588355064392,
      "learning_rate": 9.394289473684211e-05,
      "loss": 0.3697,
      "step": 43036
    },
    {
      "epoch": 2.4099563220965394,
      "grad_norm": 1.199974775314331,
      "learning_rate": 9.394263157894737e-05,
      "loss": 0.3463,
      "step": 43037
    },
    {
      "epoch": 2.4100123194086684,
      "grad_norm": 1.6323134899139404,
      "learning_rate": 9.394236842105263e-05,
      "loss": 0.3932,
      "step": 43038
    },
    {
      "epoch": 2.4100683167207975,
      "grad_norm": 1.3040753602981567,
      "learning_rate": 9.39421052631579e-05,
      "loss": 0.3669,
      "step": 43039
    },
    {
      "epoch": 2.4101243140329265,
      "grad_norm": 1.1835899353027344,
      "learning_rate": 9.394184210526317e-05,
      "loss": 0.5383,
      "step": 43040
    },
    {
      "epoch": 2.4101803113450555,
      "grad_norm": 1.078391194343567,
      "learning_rate": 9.394157894736843e-05,
      "loss": 0.4446,
      "step": 43041
    },
    {
      "epoch": 2.4102363086571845,
      "grad_norm": 1.1840434074401855,
      "learning_rate": 9.39413157894737e-05,
      "loss": 0.3895,
      "step": 43042
    },
    {
      "epoch": 2.4102923059693135,
      "grad_norm": 1.3172699213027954,
      "learning_rate": 9.394105263157895e-05,
      "loss": 0.4833,
      "step": 43043
    },
    {
      "epoch": 2.4103483032814426,
      "grad_norm": 1.2804224491119385,
      "learning_rate": 9.39407894736842e-05,
      "loss": 0.3817,
      "step": 43044
    },
    {
      "epoch": 2.4104043005935716,
      "grad_norm": 1.3820586204528809,
      "learning_rate": 9.394052631578948e-05,
      "loss": 0.3568,
      "step": 43045
    },
    {
      "epoch": 2.4104602979057006,
      "grad_norm": 1.1864701509475708,
      "learning_rate": 9.394026315789474e-05,
      "loss": 0.5526,
      "step": 43046
    },
    {
      "epoch": 2.4105162952178296,
      "grad_norm": 1.1147346496582031,
      "learning_rate": 9.394000000000001e-05,
      "loss": 0.4278,
      "step": 43047
    },
    {
      "epoch": 2.4105722925299586,
      "grad_norm": 1.197258710861206,
      "learning_rate": 9.393973684210526e-05,
      "loss": 0.4131,
      "step": 43048
    },
    {
      "epoch": 2.4106282898420877,
      "grad_norm": 1.2473978996276855,
      "learning_rate": 9.393947368421053e-05,
      "loss": 0.3866,
      "step": 43049
    },
    {
      "epoch": 2.4106842871542167,
      "grad_norm": 1.3464152812957764,
      "learning_rate": 9.393921052631579e-05,
      "loss": 0.5236,
      "step": 43050
    },
    {
      "epoch": 2.4107402844663457,
      "grad_norm": 1.2231892347335815,
      "learning_rate": 9.393894736842106e-05,
      "loss": 0.4314,
      "step": 43051
    },
    {
      "epoch": 2.4107962817784747,
      "grad_norm": 1.1388556957244873,
      "learning_rate": 9.393868421052632e-05,
      "loss": 0.3246,
      "step": 43052
    },
    {
      "epoch": 2.4108522790906037,
      "grad_norm": 1.3741780519485474,
      "learning_rate": 9.393842105263158e-05,
      "loss": 0.4131,
      "step": 43053
    },
    {
      "epoch": 2.4109082764027328,
      "grad_norm": 1.1298327445983887,
      "learning_rate": 9.393815789473684e-05,
      "loss": 0.3995,
      "step": 43054
    },
    {
      "epoch": 2.4109642737148618,
      "grad_norm": 1.1314697265625,
      "learning_rate": 9.393789473684212e-05,
      "loss": 0.3856,
      "step": 43055
    },
    {
      "epoch": 2.411020271026991,
      "grad_norm": 1.2970670461654663,
      "learning_rate": 9.393763157894738e-05,
      "loss": 0.4078,
      "step": 43056
    },
    {
      "epoch": 2.41107626833912,
      "grad_norm": 1.1381648778915405,
      "learning_rate": 9.393736842105264e-05,
      "loss": 0.497,
      "step": 43057
    },
    {
      "epoch": 2.411132265651249,
      "grad_norm": 1.1786407232284546,
      "learning_rate": 9.39371052631579e-05,
      "loss": 0.3873,
      "step": 43058
    },
    {
      "epoch": 2.411188262963378,
      "grad_norm": 1.4625476598739624,
      "learning_rate": 9.393684210526317e-05,
      "loss": 0.4412,
      "step": 43059
    },
    {
      "epoch": 2.411244260275507,
      "grad_norm": 1.138588786125183,
      "learning_rate": 9.393657894736843e-05,
      "loss": 0.3834,
      "step": 43060
    },
    {
      "epoch": 2.411300257587636,
      "grad_norm": 1.2320101261138916,
      "learning_rate": 9.393631578947369e-05,
      "loss": 0.482,
      "step": 43061
    },
    {
      "epoch": 2.411356254899765,
      "grad_norm": 1.3663620948791504,
      "learning_rate": 9.393605263157895e-05,
      "loss": 0.4771,
      "step": 43062
    },
    {
      "epoch": 2.411412252211894,
      "grad_norm": 1.1438909769058228,
      "learning_rate": 9.393578947368421e-05,
      "loss": 0.3449,
      "step": 43063
    },
    {
      "epoch": 2.411468249524023,
      "grad_norm": 1.2409121990203857,
      "learning_rate": 9.393552631578948e-05,
      "loss": 0.3754,
      "step": 43064
    },
    {
      "epoch": 2.411524246836152,
      "grad_norm": 1.1740427017211914,
      "learning_rate": 9.393526315789474e-05,
      "loss": 0.4087,
      "step": 43065
    },
    {
      "epoch": 2.411580244148281,
      "grad_norm": 1.0233221054077148,
      "learning_rate": 9.3935e-05,
      "loss": 0.3297,
      "step": 43066
    },
    {
      "epoch": 2.41163624146041,
      "grad_norm": 1.2422679662704468,
      "learning_rate": 9.393473684210526e-05,
      "loss": 0.39,
      "step": 43067
    },
    {
      "epoch": 2.411692238772539,
      "grad_norm": 1.1759917736053467,
      "learning_rate": 9.393447368421053e-05,
      "loss": 0.3853,
      "step": 43068
    },
    {
      "epoch": 2.411748236084668,
      "grad_norm": 1.04758620262146,
      "learning_rate": 9.39342105263158e-05,
      "loss": 0.3977,
      "step": 43069
    },
    {
      "epoch": 2.411804233396797,
      "grad_norm": 1.1623505353927612,
      "learning_rate": 9.393394736842105e-05,
      "loss": 0.3438,
      "step": 43070
    },
    {
      "epoch": 2.411860230708926,
      "grad_norm": 1.240695834159851,
      "learning_rate": 9.393368421052631e-05,
      "loss": 0.4,
      "step": 43071
    },
    {
      "epoch": 2.411916228021055,
      "grad_norm": 1.4051753282546997,
      "learning_rate": 9.393342105263159e-05,
      "loss": 0.3885,
      "step": 43072
    },
    {
      "epoch": 2.411972225333184,
      "grad_norm": 1.0265486240386963,
      "learning_rate": 9.393315789473685e-05,
      "loss": 0.3625,
      "step": 43073
    },
    {
      "epoch": 2.412028222645313,
      "grad_norm": 1.183159351348877,
      "learning_rate": 9.393289473684212e-05,
      "loss": 0.3719,
      "step": 43074
    },
    {
      "epoch": 2.412084219957442,
      "grad_norm": 1.3969826698303223,
      "learning_rate": 9.393263157894737e-05,
      "loss": 0.4985,
      "step": 43075
    },
    {
      "epoch": 2.412140217269571,
      "grad_norm": 1.1585941314697266,
      "learning_rate": 9.393236842105264e-05,
      "loss": 0.3914,
      "step": 43076
    },
    {
      "epoch": 2.4121962145817,
      "grad_norm": 1.1349459886550903,
      "learning_rate": 9.39321052631579e-05,
      "loss": 0.3289,
      "step": 43077
    },
    {
      "epoch": 2.412252211893829,
      "grad_norm": 1.0545337200164795,
      "learning_rate": 9.393184210526316e-05,
      "loss": 0.4249,
      "step": 43078
    },
    {
      "epoch": 2.4123082092059582,
      "grad_norm": 1.295855164527893,
      "learning_rate": 9.393157894736843e-05,
      "loss": 0.4178,
      "step": 43079
    },
    {
      "epoch": 2.4123642065180873,
      "grad_norm": 1.2078099250793457,
      "learning_rate": 9.393131578947368e-05,
      "loss": 0.3209,
      "step": 43080
    },
    {
      "epoch": 2.4124202038302163,
      "grad_norm": 1.2538615465164185,
      "learning_rate": 9.393105263157895e-05,
      "loss": 0.5204,
      "step": 43081
    },
    {
      "epoch": 2.4124762011423453,
      "grad_norm": 1.229345440864563,
      "learning_rate": 9.393078947368421e-05,
      "loss": 0.3593,
      "step": 43082
    },
    {
      "epoch": 2.4125321984544743,
      "grad_norm": 1.5337655544281006,
      "learning_rate": 9.393052631578948e-05,
      "loss": 0.5777,
      "step": 43083
    },
    {
      "epoch": 2.4125881957666033,
      "grad_norm": 1.2354308366775513,
      "learning_rate": 9.393026315789474e-05,
      "loss": 0.5183,
      "step": 43084
    },
    {
      "epoch": 2.4126441930787323,
      "grad_norm": 1.061542272567749,
      "learning_rate": 9.393e-05,
      "loss": 0.3403,
      "step": 43085
    },
    {
      "epoch": 2.4127001903908614,
      "grad_norm": 1.2376623153686523,
      "learning_rate": 9.392973684210526e-05,
      "loss": 0.4698,
      "step": 43086
    },
    {
      "epoch": 2.4127561877029904,
      "grad_norm": 1.1181284189224243,
      "learning_rate": 9.392947368421054e-05,
      "loss": 0.3205,
      "step": 43087
    },
    {
      "epoch": 2.4128121850151194,
      "grad_norm": 1.375240683555603,
      "learning_rate": 9.39292105263158e-05,
      "loss": 0.3852,
      "step": 43088
    },
    {
      "epoch": 2.4128681823272484,
      "grad_norm": 1.3245576620101929,
      "learning_rate": 9.392894736842106e-05,
      "loss": 0.4236,
      "step": 43089
    },
    {
      "epoch": 2.4129241796393774,
      "grad_norm": 1.2057394981384277,
      "learning_rate": 9.392868421052632e-05,
      "loss": 0.4361,
      "step": 43090
    },
    {
      "epoch": 2.4129801769515065,
      "grad_norm": 0.9364051818847656,
      "learning_rate": 9.392842105263159e-05,
      "loss": 0.3343,
      "step": 43091
    },
    {
      "epoch": 2.4130361742636355,
      "grad_norm": 1.3495407104492188,
      "learning_rate": 9.392815789473685e-05,
      "loss": 0.4471,
      "step": 43092
    },
    {
      "epoch": 2.4130921715757645,
      "grad_norm": 1.1332517862319946,
      "learning_rate": 9.392789473684211e-05,
      "loss": 0.4485,
      "step": 43093
    },
    {
      "epoch": 2.4131481688878935,
      "grad_norm": 1.1557482481002808,
      "learning_rate": 9.392763157894737e-05,
      "loss": 0.3675,
      "step": 43094
    },
    {
      "epoch": 2.4132041662000225,
      "grad_norm": 1.103502631187439,
      "learning_rate": 9.392736842105263e-05,
      "loss": 0.5136,
      "step": 43095
    },
    {
      "epoch": 2.4132601635121516,
      "grad_norm": 1.2479134798049927,
      "learning_rate": 9.39271052631579e-05,
      "loss": 0.4389,
      "step": 43096
    },
    {
      "epoch": 2.4133161608242806,
      "grad_norm": 1.2877532243728638,
      "learning_rate": 9.392684210526316e-05,
      "loss": 0.4599,
      "step": 43097
    },
    {
      "epoch": 2.4133721581364096,
      "grad_norm": 1.163008689880371,
      "learning_rate": 9.392657894736842e-05,
      "loss": 0.3782,
      "step": 43098
    },
    {
      "epoch": 2.4134281554485386,
      "grad_norm": 1.0230954885482788,
      "learning_rate": 9.392631578947368e-05,
      "loss": 0.2977,
      "step": 43099
    },
    {
      "epoch": 2.4134841527606676,
      "grad_norm": 1.3225160837173462,
      "learning_rate": 9.392605263157895e-05,
      "loss": 0.3995,
      "step": 43100
    },
    {
      "epoch": 2.4135401500727967,
      "grad_norm": 1.1815807819366455,
      "learning_rate": 9.392578947368421e-05,
      "loss": 0.462,
      "step": 43101
    },
    {
      "epoch": 2.4135961473849257,
      "grad_norm": 1.3101767301559448,
      "learning_rate": 9.392552631578949e-05,
      "loss": 0.522,
      "step": 43102
    },
    {
      "epoch": 2.4136521446970547,
      "grad_norm": 1.246085286140442,
      "learning_rate": 9.392526315789473e-05,
      "loss": 0.5063,
      "step": 43103
    },
    {
      "epoch": 2.4137081420091837,
      "grad_norm": 1.0921101570129395,
      "learning_rate": 9.3925e-05,
      "loss": 0.3595,
      "step": 43104
    },
    {
      "epoch": 2.4137641393213127,
      "grad_norm": 1.2914799451828003,
      "learning_rate": 9.392473684210527e-05,
      "loss": 0.3731,
      "step": 43105
    },
    {
      "epoch": 2.4138201366334417,
      "grad_norm": 1.195311427116394,
      "learning_rate": 9.392447368421054e-05,
      "loss": 0.4376,
      "step": 43106
    },
    {
      "epoch": 2.4138761339455708,
      "grad_norm": 1.1262593269348145,
      "learning_rate": 9.39242105263158e-05,
      "loss": 0.4426,
      "step": 43107
    },
    {
      "epoch": 2.4139321312577,
      "grad_norm": 1.2530790567398071,
      "learning_rate": 9.392394736842106e-05,
      "loss": 0.374,
      "step": 43108
    },
    {
      "epoch": 2.413988128569829,
      "grad_norm": 1.1239497661590576,
      "learning_rate": 9.392368421052632e-05,
      "loss": 0.333,
      "step": 43109
    },
    {
      "epoch": 2.414044125881958,
      "grad_norm": 1.190722107887268,
      "learning_rate": 9.392342105263159e-05,
      "loss": 0.4351,
      "step": 43110
    },
    {
      "epoch": 2.414100123194087,
      "grad_norm": 1.214995265007019,
      "learning_rate": 9.392315789473685e-05,
      "loss": 0.431,
      "step": 43111
    },
    {
      "epoch": 2.414156120506216,
      "grad_norm": 1.322631597518921,
      "learning_rate": 9.392289473684211e-05,
      "loss": 0.4152,
      "step": 43112
    },
    {
      "epoch": 2.414212117818345,
      "grad_norm": 1.034173607826233,
      "learning_rate": 9.392263157894737e-05,
      "loss": 0.4139,
      "step": 43113
    },
    {
      "epoch": 2.414268115130474,
      "grad_norm": 1.2792304754257202,
      "learning_rate": 9.392236842105263e-05,
      "loss": 0.4924,
      "step": 43114
    },
    {
      "epoch": 2.414324112442603,
      "grad_norm": 0.9484776258468628,
      "learning_rate": 9.39221052631579e-05,
      "loss": 0.3114,
      "step": 43115
    },
    {
      "epoch": 2.414380109754732,
      "grad_norm": 1.2107387781143188,
      "learning_rate": 9.392184210526316e-05,
      "loss": 0.3528,
      "step": 43116
    },
    {
      "epoch": 2.414436107066861,
      "grad_norm": 1.1360641717910767,
      "learning_rate": 9.392157894736842e-05,
      "loss": 0.4449,
      "step": 43117
    },
    {
      "epoch": 2.41449210437899,
      "grad_norm": 1.410962700843811,
      "learning_rate": 9.392131578947368e-05,
      "loss": 0.3334,
      "step": 43118
    },
    {
      "epoch": 2.414548101691119,
      "grad_norm": 1.2803066968917847,
      "learning_rate": 9.392105263157896e-05,
      "loss": 0.4316,
      "step": 43119
    },
    {
      "epoch": 2.414604099003248,
      "grad_norm": 1.0435315370559692,
      "learning_rate": 9.392078947368422e-05,
      "loss": 0.4116,
      "step": 43120
    },
    {
      "epoch": 2.414660096315377,
      "grad_norm": 1.2325584888458252,
      "learning_rate": 9.392052631578948e-05,
      "loss": 0.6071,
      "step": 43121
    },
    {
      "epoch": 2.414716093627506,
      "grad_norm": 1.3006629943847656,
      "learning_rate": 9.392026315789474e-05,
      "loss": 0.4262,
      "step": 43122
    },
    {
      "epoch": 2.414772090939635,
      "grad_norm": 1.2508822679519653,
      "learning_rate": 9.392000000000001e-05,
      "loss": 0.5309,
      "step": 43123
    },
    {
      "epoch": 2.414828088251764,
      "grad_norm": 0.9068685173988342,
      "learning_rate": 9.391973684210527e-05,
      "loss": 0.3575,
      "step": 43124
    },
    {
      "epoch": 2.414884085563893,
      "grad_norm": 1.3196194171905518,
      "learning_rate": 9.391947368421053e-05,
      "loss": 0.5246,
      "step": 43125
    },
    {
      "epoch": 2.414940082876022,
      "grad_norm": 1.0987013578414917,
      "learning_rate": 9.391921052631579e-05,
      "loss": 0.3174,
      "step": 43126
    },
    {
      "epoch": 2.414996080188151,
      "grad_norm": 1.372594952583313,
      "learning_rate": 9.391894736842106e-05,
      "loss": 0.3966,
      "step": 43127
    },
    {
      "epoch": 2.41505207750028,
      "grad_norm": 1.1103476285934448,
      "learning_rate": 9.391868421052632e-05,
      "loss": 0.427,
      "step": 43128
    },
    {
      "epoch": 2.415108074812409,
      "grad_norm": 0.9660261273384094,
      "learning_rate": 9.39184210526316e-05,
      "loss": 0.3716,
      "step": 43129
    },
    {
      "epoch": 2.415164072124538,
      "grad_norm": 5.481461524963379,
      "learning_rate": 9.391815789473684e-05,
      "loss": 0.4418,
      "step": 43130
    },
    {
      "epoch": 2.4152200694366672,
      "grad_norm": 1.4513428211212158,
      "learning_rate": 9.39178947368421e-05,
      "loss": 0.4423,
      "step": 43131
    },
    {
      "epoch": 2.4152760667487962,
      "grad_norm": 1.5506590604782104,
      "learning_rate": 9.391763157894737e-05,
      "loss": 0.4848,
      "step": 43132
    },
    {
      "epoch": 2.4153320640609253,
      "grad_norm": 1.2460384368896484,
      "learning_rate": 9.391736842105263e-05,
      "loss": 0.5426,
      "step": 43133
    },
    {
      "epoch": 2.4153880613730543,
      "grad_norm": 1.2605429887771606,
      "learning_rate": 9.39171052631579e-05,
      "loss": 0.4097,
      "step": 43134
    },
    {
      "epoch": 2.4154440586851833,
      "grad_norm": 1.403633952140808,
      "learning_rate": 9.391684210526315e-05,
      "loss": 0.4767,
      "step": 43135
    },
    {
      "epoch": 2.4155000559973123,
      "grad_norm": 1.1948398351669312,
      "learning_rate": 9.391657894736843e-05,
      "loss": 0.43,
      "step": 43136
    },
    {
      "epoch": 2.415556053309441,
      "grad_norm": 1.2926198244094849,
      "learning_rate": 9.391631578947369e-05,
      "loss": 0.4129,
      "step": 43137
    },
    {
      "epoch": 2.4156120506215704,
      "grad_norm": 1.553526520729065,
      "learning_rate": 9.391605263157896e-05,
      "loss": 0.4335,
      "step": 43138
    },
    {
      "epoch": 2.415668047933699,
      "grad_norm": 1.153843879699707,
      "learning_rate": 9.391578947368422e-05,
      "loss": 0.3703,
      "step": 43139
    },
    {
      "epoch": 2.4157240452458284,
      "grad_norm": 1.134920358657837,
      "learning_rate": 9.391552631578948e-05,
      "loss": 0.4737,
      "step": 43140
    },
    {
      "epoch": 2.415780042557957,
      "grad_norm": 1.0548110008239746,
      "learning_rate": 9.391526315789474e-05,
      "loss": 0.3695,
      "step": 43141
    },
    {
      "epoch": 2.4158360398700864,
      "grad_norm": 1.035057783126831,
      "learning_rate": 9.391500000000001e-05,
      "loss": 0.3744,
      "step": 43142
    },
    {
      "epoch": 2.415892037182215,
      "grad_norm": 1.5538004636764526,
      "learning_rate": 9.391473684210527e-05,
      "loss": 0.4031,
      "step": 43143
    },
    {
      "epoch": 2.4159480344943445,
      "grad_norm": 1.3576792478561401,
      "learning_rate": 9.391447368421053e-05,
      "loss": 0.4627,
      "step": 43144
    },
    {
      "epoch": 2.416004031806473,
      "grad_norm": 1.1172876358032227,
      "learning_rate": 9.391421052631579e-05,
      "loss": 0.3467,
      "step": 43145
    },
    {
      "epoch": 2.4160600291186025,
      "grad_norm": 1.0625871419906616,
      "learning_rate": 9.391394736842106e-05,
      "loss": 0.3746,
      "step": 43146
    },
    {
      "epoch": 2.416116026430731,
      "grad_norm": 1.0962773561477661,
      "learning_rate": 9.391368421052632e-05,
      "loss": 0.4264,
      "step": 43147
    },
    {
      "epoch": 2.4161720237428606,
      "grad_norm": 1.1038094758987427,
      "learning_rate": 9.391342105263158e-05,
      "loss": 0.427,
      "step": 43148
    },
    {
      "epoch": 2.416228021054989,
      "grad_norm": 1.1239385604858398,
      "learning_rate": 9.391315789473684e-05,
      "loss": 0.4856,
      "step": 43149
    },
    {
      "epoch": 2.4162840183671186,
      "grad_norm": 1.0793992280960083,
      "learning_rate": 9.39128947368421e-05,
      "loss": 0.3883,
      "step": 43150
    },
    {
      "epoch": 2.416340015679247,
      "grad_norm": 1.1953121423721313,
      "learning_rate": 9.391263157894738e-05,
      "loss": 0.3327,
      "step": 43151
    },
    {
      "epoch": 2.4163960129913766,
      "grad_norm": 1.2115827798843384,
      "learning_rate": 9.391236842105264e-05,
      "loss": 0.3148,
      "step": 43152
    },
    {
      "epoch": 2.416452010303505,
      "grad_norm": 1.3389991521835327,
      "learning_rate": 9.39121052631579e-05,
      "loss": 0.4906,
      "step": 43153
    },
    {
      "epoch": 2.4165080076156347,
      "grad_norm": 1.2485898733139038,
      "learning_rate": 9.391184210526316e-05,
      "loss": 0.4216,
      "step": 43154
    },
    {
      "epoch": 2.4165640049277632,
      "grad_norm": 1.3658658266067505,
      "learning_rate": 9.391157894736843e-05,
      "loss": 0.4177,
      "step": 43155
    },
    {
      "epoch": 2.4166200022398927,
      "grad_norm": 1.165670394897461,
      "learning_rate": 9.391131578947369e-05,
      "loss": 0.4061,
      "step": 43156
    },
    {
      "epoch": 2.4166759995520213,
      "grad_norm": 1.6250052452087402,
      "learning_rate": 9.391105263157896e-05,
      "loss": 0.5847,
      "step": 43157
    },
    {
      "epoch": 2.4167319968641507,
      "grad_norm": 60.535221099853516,
      "learning_rate": 9.391078947368421e-05,
      "loss": 0.3729,
      "step": 43158
    },
    {
      "epoch": 2.4167879941762793,
      "grad_norm": 1.3058147430419922,
      "learning_rate": 9.391052631578948e-05,
      "loss": 0.3938,
      "step": 43159
    },
    {
      "epoch": 2.416843991488409,
      "grad_norm": 0.9565534591674805,
      "learning_rate": 9.391026315789474e-05,
      "loss": 0.3505,
      "step": 43160
    },
    {
      "epoch": 2.4168999888005374,
      "grad_norm": 1.2298884391784668,
      "learning_rate": 9.391000000000001e-05,
      "loss": 0.4616,
      "step": 43161
    },
    {
      "epoch": 2.416955986112667,
      "grad_norm": 0.9534597992897034,
      "learning_rate": 9.390973684210527e-05,
      "loss": 0.3291,
      "step": 43162
    },
    {
      "epoch": 2.4170119834247954,
      "grad_norm": 1.2636332511901855,
      "learning_rate": 9.390947368421053e-05,
      "loss": 0.4012,
      "step": 43163
    },
    {
      "epoch": 2.417067980736925,
      "grad_norm": 1.173264503479004,
      "learning_rate": 9.390921052631579e-05,
      "loss": 0.4676,
      "step": 43164
    },
    {
      "epoch": 2.4171239780490534,
      "grad_norm": 1.1306235790252686,
      "learning_rate": 9.390894736842105e-05,
      "loss": 0.439,
      "step": 43165
    },
    {
      "epoch": 2.417179975361183,
      "grad_norm": 1.6614991426467896,
      "learning_rate": 9.390868421052633e-05,
      "loss": 0.3796,
      "step": 43166
    },
    {
      "epoch": 2.4172359726733115,
      "grad_norm": 1.211554765701294,
      "learning_rate": 9.390842105263157e-05,
      "loss": 0.4156,
      "step": 43167
    },
    {
      "epoch": 2.417291969985441,
      "grad_norm": 1.3854492902755737,
      "learning_rate": 9.390815789473685e-05,
      "loss": 0.4769,
      "step": 43168
    },
    {
      "epoch": 2.4173479672975695,
      "grad_norm": 1.379655361175537,
      "learning_rate": 9.39078947368421e-05,
      "loss": 0.4333,
      "step": 43169
    },
    {
      "epoch": 2.417403964609699,
      "grad_norm": 0.9838935136795044,
      "learning_rate": 9.390763157894738e-05,
      "loss": 0.3197,
      "step": 43170
    },
    {
      "epoch": 2.4174599619218275,
      "grad_norm": 1.2255523204803467,
      "learning_rate": 9.390736842105264e-05,
      "loss": 0.4104,
      "step": 43171
    },
    {
      "epoch": 2.417515959233957,
      "grad_norm": 1.3336806297302246,
      "learning_rate": 9.39071052631579e-05,
      "loss": 0.4586,
      "step": 43172
    },
    {
      "epoch": 2.4175719565460856,
      "grad_norm": 1.1134190559387207,
      "learning_rate": 9.390684210526316e-05,
      "loss": 0.4706,
      "step": 43173
    },
    {
      "epoch": 2.417627953858215,
      "grad_norm": 1.2328906059265137,
      "learning_rate": 9.390657894736843e-05,
      "loss": 0.47,
      "step": 43174
    },
    {
      "epoch": 2.4176839511703436,
      "grad_norm": 1.3155572414398193,
      "learning_rate": 9.390631578947369e-05,
      "loss": 0.4967,
      "step": 43175
    },
    {
      "epoch": 2.417739948482473,
      "grad_norm": 1.430331826210022,
      "learning_rate": 9.390605263157895e-05,
      "loss": 0.4034,
      "step": 43176
    },
    {
      "epoch": 2.4177959457946017,
      "grad_norm": 1.0663114786148071,
      "learning_rate": 9.390578947368421e-05,
      "loss": 0.3639,
      "step": 43177
    },
    {
      "epoch": 2.4178519431067307,
      "grad_norm": 1.2175368070602417,
      "learning_rate": 9.390552631578948e-05,
      "loss": 0.3805,
      "step": 43178
    },
    {
      "epoch": 2.4179079404188597,
      "grad_norm": 1.1363192796707153,
      "learning_rate": 9.390526315789474e-05,
      "loss": 0.4373,
      "step": 43179
    },
    {
      "epoch": 2.4179639377309887,
      "grad_norm": 1.1539466381072998,
      "learning_rate": 9.3905e-05,
      "loss": 0.425,
      "step": 43180
    },
    {
      "epoch": 2.4180199350431177,
      "grad_norm": 1.173219919204712,
      "learning_rate": 9.390473684210526e-05,
      "loss": 0.4532,
      "step": 43181
    },
    {
      "epoch": 2.4180759323552468,
      "grad_norm": 1.0323699712753296,
      "learning_rate": 9.390447368421052e-05,
      "loss": 0.3398,
      "step": 43182
    },
    {
      "epoch": 2.418131929667376,
      "grad_norm": 1.0396913290023804,
      "learning_rate": 9.39042105263158e-05,
      "loss": 0.257,
      "step": 43183
    },
    {
      "epoch": 2.418187926979505,
      "grad_norm": 1.4664250612258911,
      "learning_rate": 9.390394736842106e-05,
      "loss": 0.645,
      "step": 43184
    },
    {
      "epoch": 2.418243924291634,
      "grad_norm": 1.12051260471344,
      "learning_rate": 9.390368421052632e-05,
      "loss": 0.3298,
      "step": 43185
    },
    {
      "epoch": 2.418299921603763,
      "grad_norm": 1.1920262575149536,
      "learning_rate": 9.390342105263157e-05,
      "loss": 0.4536,
      "step": 43186
    },
    {
      "epoch": 2.418355918915892,
      "grad_norm": 1.313310146331787,
      "learning_rate": 9.390315789473685e-05,
      "loss": 0.3502,
      "step": 43187
    },
    {
      "epoch": 2.418411916228021,
      "grad_norm": 1.0927398204803467,
      "learning_rate": 9.390289473684211e-05,
      "loss": 0.4967,
      "step": 43188
    },
    {
      "epoch": 2.41846791354015,
      "grad_norm": 1.2346994876861572,
      "learning_rate": 9.390263157894738e-05,
      "loss": 0.4715,
      "step": 43189
    },
    {
      "epoch": 2.418523910852279,
      "grad_norm": 1.2547309398651123,
      "learning_rate": 9.390236842105263e-05,
      "loss": 0.3551,
      "step": 43190
    },
    {
      "epoch": 2.418579908164408,
      "grad_norm": 1.3142169713974,
      "learning_rate": 9.39021052631579e-05,
      "loss": 0.3755,
      "step": 43191
    },
    {
      "epoch": 2.418635905476537,
      "grad_norm": 1.2423584461212158,
      "learning_rate": 9.390184210526316e-05,
      "loss": 0.3715,
      "step": 43192
    },
    {
      "epoch": 2.418691902788666,
      "grad_norm": 1.1156691312789917,
      "learning_rate": 9.390157894736843e-05,
      "loss": 0.4065,
      "step": 43193
    },
    {
      "epoch": 2.418747900100795,
      "grad_norm": 1.2996442317962646,
      "learning_rate": 9.39013157894737e-05,
      "loss": 0.3434,
      "step": 43194
    },
    {
      "epoch": 2.418803897412924,
      "grad_norm": 1.118552565574646,
      "learning_rate": 9.390105263157895e-05,
      "loss": 0.3606,
      "step": 43195
    },
    {
      "epoch": 2.418859894725053,
      "grad_norm": 1.155348539352417,
      "learning_rate": 9.390078947368421e-05,
      "loss": 0.3741,
      "step": 43196
    },
    {
      "epoch": 2.418915892037182,
      "grad_norm": 1.3397969007492065,
      "learning_rate": 9.390052631578949e-05,
      "loss": 0.4408,
      "step": 43197
    },
    {
      "epoch": 2.418971889349311,
      "grad_norm": 1.0667774677276611,
      "learning_rate": 9.390026315789475e-05,
      "loss": 0.3677,
      "step": 43198
    },
    {
      "epoch": 2.41902788666144,
      "grad_norm": 1.1114362478256226,
      "learning_rate": 9.39e-05,
      "loss": 0.3472,
      "step": 43199
    },
    {
      "epoch": 2.419083883973569,
      "grad_norm": 1.0219911336898804,
      "learning_rate": 9.389973684210527e-05,
      "loss": 0.3541,
      "step": 43200
    },
    {
      "epoch": 2.419139881285698,
      "grad_norm": 1.1505446434020996,
      "learning_rate": 9.389947368421053e-05,
      "loss": 0.4004,
      "step": 43201
    },
    {
      "epoch": 2.419195878597827,
      "grad_norm": 1.1320921182632446,
      "learning_rate": 9.38992105263158e-05,
      "loss": 0.3508,
      "step": 43202
    },
    {
      "epoch": 2.419251875909956,
      "grad_norm": 1.3824846744537354,
      "learning_rate": 9.389894736842106e-05,
      "loss": 0.42,
      "step": 43203
    },
    {
      "epoch": 2.419307873222085,
      "grad_norm": 1.0686715841293335,
      "learning_rate": 9.389868421052632e-05,
      "loss": 0.3867,
      "step": 43204
    },
    {
      "epoch": 2.419363870534214,
      "grad_norm": 1.3476555347442627,
      "learning_rate": 9.389842105263158e-05,
      "loss": 0.4533,
      "step": 43205
    },
    {
      "epoch": 2.419419867846343,
      "grad_norm": 1.0201270580291748,
      "learning_rate": 9.389815789473685e-05,
      "loss": 0.3794,
      "step": 43206
    },
    {
      "epoch": 2.4194758651584722,
      "grad_norm": 1.5151606798171997,
      "learning_rate": 9.389789473684211e-05,
      "loss": 0.5095,
      "step": 43207
    },
    {
      "epoch": 2.4195318624706013,
      "grad_norm": 1.0412055253982544,
      "learning_rate": 9.389763157894737e-05,
      "loss": 0.4149,
      "step": 43208
    },
    {
      "epoch": 2.4195878597827303,
      "grad_norm": 1.1569360494613647,
      "learning_rate": 9.389736842105263e-05,
      "loss": 0.4052,
      "step": 43209
    },
    {
      "epoch": 2.4196438570948593,
      "grad_norm": 1.0770893096923828,
      "learning_rate": 9.38971052631579e-05,
      "loss": 0.4049,
      "step": 43210
    },
    {
      "epoch": 2.4196998544069883,
      "grad_norm": 1.2463077306747437,
      "learning_rate": 9.389684210526316e-05,
      "loss": 0.3686,
      "step": 43211
    },
    {
      "epoch": 2.4197558517191173,
      "grad_norm": 1.2184151411056519,
      "learning_rate": 9.389657894736844e-05,
      "loss": 0.3882,
      "step": 43212
    },
    {
      "epoch": 2.4198118490312464,
      "grad_norm": 1.1225277185440063,
      "learning_rate": 9.389631578947368e-05,
      "loss": 0.3753,
      "step": 43213
    },
    {
      "epoch": 2.4198678463433754,
      "grad_norm": 1.248574137687683,
      "learning_rate": 9.389605263157896e-05,
      "loss": 0.3381,
      "step": 43214
    },
    {
      "epoch": 2.4199238436555044,
      "grad_norm": 1.0137771368026733,
      "learning_rate": 9.389578947368422e-05,
      "loss": 0.3379,
      "step": 43215
    },
    {
      "epoch": 2.4199798409676334,
      "grad_norm": 1.0141942501068115,
      "learning_rate": 9.389552631578948e-05,
      "loss": 0.3385,
      "step": 43216
    },
    {
      "epoch": 2.4200358382797624,
      "grad_norm": 1.3201202154159546,
      "learning_rate": 9.389526315789475e-05,
      "loss": 0.4165,
      "step": 43217
    },
    {
      "epoch": 2.4200918355918914,
      "grad_norm": 1.6294525861740112,
      "learning_rate": 9.3895e-05,
      "loss": 0.5434,
      "step": 43218
    },
    {
      "epoch": 2.4201478329040205,
      "grad_norm": 1.3794221878051758,
      "learning_rate": 9.389473684210527e-05,
      "loss": 0.6643,
      "step": 43219
    },
    {
      "epoch": 2.4202038302161495,
      "grad_norm": 1.267309308052063,
      "learning_rate": 9.389447368421053e-05,
      "loss": 0.3686,
      "step": 43220
    },
    {
      "epoch": 2.4202598275282785,
      "grad_norm": 1.4485201835632324,
      "learning_rate": 9.38942105263158e-05,
      "loss": 0.423,
      "step": 43221
    },
    {
      "epoch": 2.4203158248404075,
      "grad_norm": 1.0893577337265015,
      "learning_rate": 9.389394736842105e-05,
      "loss": 0.2788,
      "step": 43222
    },
    {
      "epoch": 2.4203718221525365,
      "grad_norm": 1.3437695503234863,
      "learning_rate": 9.389368421052632e-05,
      "loss": 0.3517,
      "step": 43223
    },
    {
      "epoch": 2.4204278194646656,
      "grad_norm": 1.0807229280471802,
      "learning_rate": 9.389342105263158e-05,
      "loss": 0.3304,
      "step": 43224
    },
    {
      "epoch": 2.4204838167767946,
      "grad_norm": 1.154120683670044,
      "learning_rate": 9.389315789473685e-05,
      "loss": 0.3893,
      "step": 43225
    },
    {
      "epoch": 2.4205398140889236,
      "grad_norm": 1.243463397026062,
      "learning_rate": 9.389289473684211e-05,
      "loss": 0.4865,
      "step": 43226
    },
    {
      "epoch": 2.4205958114010526,
      "grad_norm": 1.0998270511627197,
      "learning_rate": 9.389263157894737e-05,
      "loss": 0.27,
      "step": 43227
    },
    {
      "epoch": 2.4206518087131816,
      "grad_norm": 1.1756845712661743,
      "learning_rate": 9.389236842105263e-05,
      "loss": 0.4328,
      "step": 43228
    },
    {
      "epoch": 2.4207078060253107,
      "grad_norm": 1.040419578552246,
      "learning_rate": 9.38921052631579e-05,
      "loss": 0.4173,
      "step": 43229
    },
    {
      "epoch": 2.4207638033374397,
      "grad_norm": 1.5336278676986694,
      "learning_rate": 9.389184210526317e-05,
      "loss": 0.7211,
      "step": 43230
    },
    {
      "epoch": 2.4208198006495687,
      "grad_norm": 1.1336097717285156,
      "learning_rate": 9.389157894736843e-05,
      "loss": 0.3522,
      "step": 43231
    },
    {
      "epoch": 2.4208757979616977,
      "grad_norm": 1.3105348348617554,
      "learning_rate": 9.389131578947369e-05,
      "loss": 0.4495,
      "step": 43232
    },
    {
      "epoch": 2.4209317952738267,
      "grad_norm": 1.0838714838027954,
      "learning_rate": 9.389105263157896e-05,
      "loss": 0.4578,
      "step": 43233
    },
    {
      "epoch": 2.4209877925859558,
      "grad_norm": 1.4369572401046753,
      "learning_rate": 9.389078947368422e-05,
      "loss": 0.4893,
      "step": 43234
    },
    {
      "epoch": 2.4210437898980848,
      "grad_norm": 1.0797373056411743,
      "learning_rate": 9.389052631578948e-05,
      "loss": 0.3482,
      "step": 43235
    },
    {
      "epoch": 2.421099787210214,
      "grad_norm": 0.9420746564865112,
      "learning_rate": 9.389026315789474e-05,
      "loss": 0.2995,
      "step": 43236
    },
    {
      "epoch": 2.421155784522343,
      "grad_norm": 1.3257580995559692,
      "learning_rate": 9.389e-05,
      "loss": 0.4331,
      "step": 43237
    },
    {
      "epoch": 2.421211781834472,
      "grad_norm": 1.193414568901062,
      "learning_rate": 9.388973684210527e-05,
      "loss": 0.4675,
      "step": 43238
    },
    {
      "epoch": 2.421267779146601,
      "grad_norm": 1.1686961650848389,
      "learning_rate": 9.388947368421053e-05,
      "loss": 0.3843,
      "step": 43239
    },
    {
      "epoch": 2.42132377645873,
      "grad_norm": 1.0356837511062622,
      "learning_rate": 9.388921052631579e-05,
      "loss": 0.4006,
      "step": 43240
    },
    {
      "epoch": 2.421379773770859,
      "grad_norm": 1.1721603870391846,
      "learning_rate": 9.388894736842105e-05,
      "loss": 0.4251,
      "step": 43241
    },
    {
      "epoch": 2.421435771082988,
      "grad_norm": 1.0868178606033325,
      "learning_rate": 9.388868421052632e-05,
      "loss": 0.3316,
      "step": 43242
    },
    {
      "epoch": 2.421491768395117,
      "grad_norm": 1.1554374694824219,
      "learning_rate": 9.388842105263158e-05,
      "loss": 0.4383,
      "step": 43243
    },
    {
      "epoch": 2.421547765707246,
      "grad_norm": 1.1034214496612549,
      "learning_rate": 9.388815789473686e-05,
      "loss": 0.4773,
      "step": 43244
    },
    {
      "epoch": 2.421603763019375,
      "grad_norm": 1.2092292308807373,
      "learning_rate": 9.38878947368421e-05,
      "loss": 0.4074,
      "step": 43245
    },
    {
      "epoch": 2.421659760331504,
      "grad_norm": 1.39845609664917,
      "learning_rate": 9.388763157894738e-05,
      "loss": 0.6073,
      "step": 43246
    },
    {
      "epoch": 2.421715757643633,
      "grad_norm": 1.2786108255386353,
      "learning_rate": 9.388736842105264e-05,
      "loss": 0.5254,
      "step": 43247
    },
    {
      "epoch": 2.421771754955762,
      "grad_norm": 1.2143055200576782,
      "learning_rate": 9.388710526315791e-05,
      "loss": 0.4087,
      "step": 43248
    },
    {
      "epoch": 2.421827752267891,
      "grad_norm": 1.3010919094085693,
      "learning_rate": 9.388684210526317e-05,
      "loss": 0.4788,
      "step": 43249
    },
    {
      "epoch": 2.42188374958002,
      "grad_norm": 1.2912427186965942,
      "learning_rate": 9.388657894736843e-05,
      "loss": 0.3812,
      "step": 43250
    },
    {
      "epoch": 2.421939746892149,
      "grad_norm": 1.0862281322479248,
      "learning_rate": 9.388631578947369e-05,
      "loss": 0.3291,
      "step": 43251
    },
    {
      "epoch": 2.421995744204278,
      "grad_norm": 1.0091866254806519,
      "learning_rate": 9.388605263157895e-05,
      "loss": 0.358,
      "step": 43252
    },
    {
      "epoch": 2.422051741516407,
      "grad_norm": 1.1845595836639404,
      "learning_rate": 9.388578947368422e-05,
      "loss": 0.2845,
      "step": 43253
    },
    {
      "epoch": 2.422107738828536,
      "grad_norm": 1.8282955884933472,
      "learning_rate": 9.388552631578948e-05,
      "loss": 0.5202,
      "step": 43254
    },
    {
      "epoch": 2.422163736140665,
      "grad_norm": 1.4279855489730835,
      "learning_rate": 9.388526315789474e-05,
      "loss": 0.3703,
      "step": 43255
    },
    {
      "epoch": 2.422219733452794,
      "grad_norm": 1.1744455099105835,
      "learning_rate": 9.3885e-05,
      "loss": 0.3577,
      "step": 43256
    },
    {
      "epoch": 2.422275730764923,
      "grad_norm": 1.2386982440948486,
      "learning_rate": 9.388473684210527e-05,
      "loss": 0.6016,
      "step": 43257
    },
    {
      "epoch": 2.422331728077052,
      "grad_norm": 1.3044960498809814,
      "learning_rate": 9.388447368421053e-05,
      "loss": 0.4669,
      "step": 43258
    },
    {
      "epoch": 2.4223877253891812,
      "grad_norm": 1.5267186164855957,
      "learning_rate": 9.388421052631579e-05,
      "loss": 0.4866,
      "step": 43259
    },
    {
      "epoch": 2.4224437227013103,
      "grad_norm": 1.190309762954712,
      "learning_rate": 9.388394736842105e-05,
      "loss": 0.3326,
      "step": 43260
    },
    {
      "epoch": 2.4224997200134393,
      "grad_norm": 1.3032572269439697,
      "learning_rate": 9.388368421052633e-05,
      "loss": 0.3926,
      "step": 43261
    },
    {
      "epoch": 2.4225557173255683,
      "grad_norm": 1.1624886989593506,
      "learning_rate": 9.388342105263159e-05,
      "loss": 0.3717,
      "step": 43262
    },
    {
      "epoch": 2.4226117146376973,
      "grad_norm": 1.152766466140747,
      "learning_rate": 9.388315789473685e-05,
      "loss": 0.4647,
      "step": 43263
    },
    {
      "epoch": 2.4226677119498263,
      "grad_norm": 1.572954773902893,
      "learning_rate": 9.38828947368421e-05,
      "loss": 0.3511,
      "step": 43264
    },
    {
      "epoch": 2.4227237092619553,
      "grad_norm": 1.1790647506713867,
      "learning_rate": 9.388263157894738e-05,
      "loss": 0.397,
      "step": 43265
    },
    {
      "epoch": 2.4227797065740844,
      "grad_norm": 1.1409581899642944,
      "learning_rate": 9.388236842105264e-05,
      "loss": 0.4568,
      "step": 43266
    },
    {
      "epoch": 2.4228357038862134,
      "grad_norm": 1.3559352159500122,
      "learning_rate": 9.388210526315791e-05,
      "loss": 0.4505,
      "step": 43267
    },
    {
      "epoch": 2.4228917011983424,
      "grad_norm": 1.1805355548858643,
      "learning_rate": 9.388184210526316e-05,
      "loss": 0.3235,
      "step": 43268
    },
    {
      "epoch": 2.4229476985104714,
      "grad_norm": 1.1075761318206787,
      "learning_rate": 9.388157894736842e-05,
      "loss": 0.3424,
      "step": 43269
    },
    {
      "epoch": 2.4230036958226004,
      "grad_norm": 1.2438340187072754,
      "learning_rate": 9.388131578947369e-05,
      "loss": 0.5587,
      "step": 43270
    },
    {
      "epoch": 2.4230596931347295,
      "grad_norm": 1.3456114530563354,
      "learning_rate": 9.388105263157895e-05,
      "loss": 0.4115,
      "step": 43271
    },
    {
      "epoch": 2.4231156904468585,
      "grad_norm": 1.4254484176635742,
      "learning_rate": 9.388078947368421e-05,
      "loss": 0.4292,
      "step": 43272
    },
    {
      "epoch": 2.4231716877589875,
      "grad_norm": 1.6074918508529663,
      "learning_rate": 9.388052631578947e-05,
      "loss": 0.4058,
      "step": 43273
    },
    {
      "epoch": 2.4232276850711165,
      "grad_norm": 1.2041456699371338,
      "learning_rate": 9.388026315789474e-05,
      "loss": 0.4227,
      "step": 43274
    },
    {
      "epoch": 2.4232836823832455,
      "grad_norm": 1.1134364604949951,
      "learning_rate": 9.388e-05,
      "loss": 0.3612,
      "step": 43275
    },
    {
      "epoch": 2.4233396796953746,
      "grad_norm": 1.0012108087539673,
      "learning_rate": 9.387973684210528e-05,
      "loss": 0.3828,
      "step": 43276
    },
    {
      "epoch": 2.4233956770075036,
      "grad_norm": 1.6649832725524902,
      "learning_rate": 9.387947368421052e-05,
      "loss": 0.3591,
      "step": 43277
    },
    {
      "epoch": 2.4234516743196326,
      "grad_norm": 1.1950238943099976,
      "learning_rate": 9.38792105263158e-05,
      "loss": 0.4079,
      "step": 43278
    },
    {
      "epoch": 2.4235076716317616,
      "grad_norm": 1.4727941751480103,
      "learning_rate": 9.387894736842105e-05,
      "loss": 0.4825,
      "step": 43279
    },
    {
      "epoch": 2.4235636689438906,
      "grad_norm": 1.3424957990646362,
      "learning_rate": 9.387868421052633e-05,
      "loss": 0.4561,
      "step": 43280
    },
    {
      "epoch": 2.4236196662560197,
      "grad_norm": 1.1803845167160034,
      "learning_rate": 9.387842105263159e-05,
      "loss": 0.3833,
      "step": 43281
    },
    {
      "epoch": 2.4236756635681487,
      "grad_norm": 1.0076804161071777,
      "learning_rate": 9.387815789473685e-05,
      "loss": 0.2925,
      "step": 43282
    },
    {
      "epoch": 2.4237316608802777,
      "grad_norm": 1.3036189079284668,
      "learning_rate": 9.387789473684211e-05,
      "loss": 0.412,
      "step": 43283
    },
    {
      "epoch": 2.4237876581924067,
      "grad_norm": 1.1030062437057495,
      "learning_rate": 9.387763157894738e-05,
      "loss": 0.3355,
      "step": 43284
    },
    {
      "epoch": 2.4238436555045357,
      "grad_norm": 1.480510950088501,
      "learning_rate": 9.387736842105264e-05,
      "loss": 0.3778,
      "step": 43285
    },
    {
      "epoch": 2.4238996528166648,
      "grad_norm": 1.049233078956604,
      "learning_rate": 9.38771052631579e-05,
      "loss": 0.3181,
      "step": 43286
    },
    {
      "epoch": 2.4239556501287938,
      "grad_norm": 1.2352315187454224,
      "learning_rate": 9.387684210526316e-05,
      "loss": 0.397,
      "step": 43287
    },
    {
      "epoch": 2.424011647440923,
      "grad_norm": 1.1681936979293823,
      "learning_rate": 9.387657894736842e-05,
      "loss": 0.4868,
      "step": 43288
    },
    {
      "epoch": 2.424067644753052,
      "grad_norm": 1.1199984550476074,
      "learning_rate": 9.387631578947369e-05,
      "loss": 0.5695,
      "step": 43289
    },
    {
      "epoch": 2.424123642065181,
      "grad_norm": 1.2349029779434204,
      "learning_rate": 9.387605263157895e-05,
      "loss": 0.3664,
      "step": 43290
    },
    {
      "epoch": 2.42417963937731,
      "grad_norm": 1.2325079441070557,
      "learning_rate": 9.387578947368421e-05,
      "loss": 0.3464,
      "step": 43291
    },
    {
      "epoch": 2.424235636689439,
      "grad_norm": 1.2472261190414429,
      "learning_rate": 9.387552631578947e-05,
      "loss": 0.4071,
      "step": 43292
    },
    {
      "epoch": 2.424291634001568,
      "grad_norm": 1.2833224534988403,
      "learning_rate": 9.387526315789475e-05,
      "loss": 0.5953,
      "step": 43293
    },
    {
      "epoch": 2.424347631313697,
      "grad_norm": 1.0062750577926636,
      "learning_rate": 9.3875e-05,
      "loss": 0.3378,
      "step": 43294
    },
    {
      "epoch": 2.424403628625826,
      "grad_norm": 1.3772644996643066,
      "learning_rate": 9.387473684210526e-05,
      "loss": 0.6255,
      "step": 43295
    },
    {
      "epoch": 2.424459625937955,
      "grad_norm": 1.2342292070388794,
      "learning_rate": 9.387447368421052e-05,
      "loss": 0.364,
      "step": 43296
    },
    {
      "epoch": 2.424515623250084,
      "grad_norm": 1.0805096626281738,
      "learning_rate": 9.38742105263158e-05,
      "loss": 0.3374,
      "step": 43297
    },
    {
      "epoch": 2.424571620562213,
      "grad_norm": 1.118391752243042,
      "learning_rate": 9.387394736842106e-05,
      "loss": 0.3436,
      "step": 43298
    },
    {
      "epoch": 2.424627617874342,
      "grad_norm": 1.0967470407485962,
      "learning_rate": 9.387368421052633e-05,
      "loss": 0.474,
      "step": 43299
    },
    {
      "epoch": 2.424683615186471,
      "grad_norm": 1.2771642208099365,
      "learning_rate": 9.387342105263158e-05,
      "loss": 0.4183,
      "step": 43300
    },
    {
      "epoch": 2.4247396124986,
      "grad_norm": 1.3030600547790527,
      "learning_rate": 9.387315789473685e-05,
      "loss": 0.3994,
      "step": 43301
    },
    {
      "epoch": 2.424795609810729,
      "grad_norm": 1.1603407859802246,
      "learning_rate": 9.387289473684211e-05,
      "loss": 0.3234,
      "step": 43302
    },
    {
      "epoch": 2.424851607122858,
      "grad_norm": 1.1143550872802734,
      "learning_rate": 9.387263157894737e-05,
      "loss": 0.3263,
      "step": 43303
    },
    {
      "epoch": 2.424907604434987,
      "grad_norm": 1.081955909729004,
      "learning_rate": 9.387236842105264e-05,
      "loss": 0.4103,
      "step": 43304
    },
    {
      "epoch": 2.424963601747116,
      "grad_norm": 1.1078665256500244,
      "learning_rate": 9.387210526315789e-05,
      "loss": 0.4375,
      "step": 43305
    },
    {
      "epoch": 2.425019599059245,
      "grad_norm": 1.1512471437454224,
      "learning_rate": 9.387184210526316e-05,
      "loss": 0.3524,
      "step": 43306
    },
    {
      "epoch": 2.425075596371374,
      "grad_norm": 1.2474839687347412,
      "learning_rate": 9.387157894736842e-05,
      "loss": 0.4101,
      "step": 43307
    },
    {
      "epoch": 2.425131593683503,
      "grad_norm": 1.4544689655303955,
      "learning_rate": 9.38713157894737e-05,
      "loss": 0.417,
      "step": 43308
    },
    {
      "epoch": 2.425187590995632,
      "grad_norm": 1.1838221549987793,
      "learning_rate": 9.387105263157896e-05,
      "loss": 0.3753,
      "step": 43309
    },
    {
      "epoch": 2.425243588307761,
      "grad_norm": 1.2453458309173584,
      "learning_rate": 9.387078947368421e-05,
      "loss": 0.356,
      "step": 43310
    },
    {
      "epoch": 2.4252995856198902,
      "grad_norm": 2.3535826206207275,
      "learning_rate": 9.387052631578947e-05,
      "loss": 0.3519,
      "step": 43311
    },
    {
      "epoch": 2.4253555829320192,
      "grad_norm": 1.2159059047698975,
      "learning_rate": 9.387026315789475e-05,
      "loss": 0.4917,
      "step": 43312
    },
    {
      "epoch": 2.4254115802441483,
      "grad_norm": 1.44333016872406,
      "learning_rate": 9.387000000000001e-05,
      "loss": 0.6589,
      "step": 43313
    },
    {
      "epoch": 2.4254675775562773,
      "grad_norm": 1.2149183750152588,
      "learning_rate": 9.386973684210527e-05,
      "loss": 0.3693,
      "step": 43314
    },
    {
      "epoch": 2.4255235748684063,
      "grad_norm": 1.102017879486084,
      "learning_rate": 9.386947368421053e-05,
      "loss": 0.4015,
      "step": 43315
    },
    {
      "epoch": 2.4255795721805353,
      "grad_norm": 1.4537397623062134,
      "learning_rate": 9.38692105263158e-05,
      "loss": 0.4418,
      "step": 43316
    },
    {
      "epoch": 2.4256355694926643,
      "grad_norm": 1.1533023118972778,
      "learning_rate": 9.386894736842106e-05,
      "loss": 0.3911,
      "step": 43317
    },
    {
      "epoch": 2.4256915668047934,
      "grad_norm": 1.1877247095108032,
      "learning_rate": 9.386868421052632e-05,
      "loss": 0.3713,
      "step": 43318
    },
    {
      "epoch": 2.4257475641169224,
      "grad_norm": 1.0765410661697388,
      "learning_rate": 9.386842105263158e-05,
      "loss": 0.4446,
      "step": 43319
    },
    {
      "epoch": 2.4258035614290514,
      "grad_norm": 1.4328455924987793,
      "learning_rate": 9.386815789473684e-05,
      "loss": 0.4334,
      "step": 43320
    },
    {
      "epoch": 2.4258595587411804,
      "grad_norm": 1.2991150617599487,
      "learning_rate": 9.386789473684211e-05,
      "loss": 0.4606,
      "step": 43321
    },
    {
      "epoch": 2.4259155560533094,
      "grad_norm": 1.202183485031128,
      "learning_rate": 9.386763157894737e-05,
      "loss": 0.3483,
      "step": 43322
    },
    {
      "epoch": 2.4259715533654385,
      "grad_norm": 1.1124930381774902,
      "learning_rate": 9.386736842105263e-05,
      "loss": 0.3801,
      "step": 43323
    },
    {
      "epoch": 2.4260275506775675,
      "grad_norm": 1.2466683387756348,
      "learning_rate": 9.386710526315789e-05,
      "loss": 0.3554,
      "step": 43324
    },
    {
      "epoch": 2.4260835479896965,
      "grad_norm": 1.3240309953689575,
      "learning_rate": 9.386684210526316e-05,
      "loss": 0.4564,
      "step": 43325
    },
    {
      "epoch": 2.4261395453018255,
      "grad_norm": 1.1996407508850098,
      "learning_rate": 9.386657894736842e-05,
      "loss": 0.3253,
      "step": 43326
    },
    {
      "epoch": 2.4261955426139545,
      "grad_norm": 1.1670150756835938,
      "learning_rate": 9.386631578947368e-05,
      "loss": 0.5599,
      "step": 43327
    },
    {
      "epoch": 2.4262515399260836,
      "grad_norm": 0.950490415096283,
      "learning_rate": 9.386605263157894e-05,
      "loss": 0.3096,
      "step": 43328
    },
    {
      "epoch": 2.4263075372382126,
      "grad_norm": 1.210343599319458,
      "learning_rate": 9.386578947368422e-05,
      "loss": 0.4202,
      "step": 43329
    },
    {
      "epoch": 2.4263635345503416,
      "grad_norm": 1.4921112060546875,
      "learning_rate": 9.386552631578948e-05,
      "loss": 0.3783,
      "step": 43330
    },
    {
      "epoch": 2.4264195318624706,
      "grad_norm": 1.0691190958023071,
      "learning_rate": 9.386526315789475e-05,
      "loss": 0.4163,
      "step": 43331
    },
    {
      "epoch": 2.4264755291745996,
      "grad_norm": 1.0907038450241089,
      "learning_rate": 9.3865e-05,
      "loss": 0.3056,
      "step": 43332
    },
    {
      "epoch": 2.4265315264867287,
      "grad_norm": 1.3283377885818481,
      "learning_rate": 9.386473684210527e-05,
      "loss": 0.4833,
      "step": 43333
    },
    {
      "epoch": 2.4265875237988577,
      "grad_norm": 1.1758837699890137,
      "learning_rate": 9.386447368421053e-05,
      "loss": 0.4387,
      "step": 43334
    },
    {
      "epoch": 2.4266435211109867,
      "grad_norm": 1.3361684083938599,
      "learning_rate": 9.38642105263158e-05,
      "loss": 0.403,
      "step": 43335
    },
    {
      "epoch": 2.4266995184231157,
      "grad_norm": 1.2100074291229248,
      "learning_rate": 9.386394736842106e-05,
      "loss": 0.4033,
      "step": 43336
    },
    {
      "epoch": 2.4267555157352447,
      "grad_norm": 1.314420461654663,
      "learning_rate": 9.386368421052631e-05,
      "loss": 0.4736,
      "step": 43337
    },
    {
      "epoch": 2.4268115130473737,
      "grad_norm": 1.1464412212371826,
      "learning_rate": 9.386342105263158e-05,
      "loss": 0.4867,
      "step": 43338
    },
    {
      "epoch": 2.4268675103595028,
      "grad_norm": 1.3981119394302368,
      "learning_rate": 9.386315789473684e-05,
      "loss": 0.522,
      "step": 43339
    },
    {
      "epoch": 2.426923507671632,
      "grad_norm": 1.0674493312835693,
      "learning_rate": 9.386289473684212e-05,
      "loss": 0.4018,
      "step": 43340
    },
    {
      "epoch": 2.426979504983761,
      "grad_norm": 1.4946500062942505,
      "learning_rate": 9.386263157894737e-05,
      "loss": 0.4493,
      "step": 43341
    },
    {
      "epoch": 2.42703550229589,
      "grad_norm": 1.2862634658813477,
      "learning_rate": 9.386236842105263e-05,
      "loss": 0.4656,
      "step": 43342
    },
    {
      "epoch": 2.427091499608019,
      "grad_norm": 1.1617116928100586,
      "learning_rate": 9.38621052631579e-05,
      "loss": 0.4247,
      "step": 43343
    },
    {
      "epoch": 2.427147496920148,
      "grad_norm": 1.376528024673462,
      "learning_rate": 9.386184210526317e-05,
      "loss": 0.3746,
      "step": 43344
    },
    {
      "epoch": 2.427203494232277,
      "grad_norm": 1.2471561431884766,
      "learning_rate": 9.386157894736843e-05,
      "loss": 0.3677,
      "step": 43345
    },
    {
      "epoch": 2.427259491544406,
      "grad_norm": 1.153925895690918,
      "learning_rate": 9.386131578947369e-05,
      "loss": 0.4013,
      "step": 43346
    },
    {
      "epoch": 2.427315488856535,
      "grad_norm": 1.207470178604126,
      "learning_rate": 9.386105263157895e-05,
      "loss": 0.5585,
      "step": 43347
    },
    {
      "epoch": 2.427371486168664,
      "grad_norm": 1.0314432382583618,
      "learning_rate": 9.386078947368422e-05,
      "loss": 0.3651,
      "step": 43348
    },
    {
      "epoch": 2.427427483480793,
      "grad_norm": 1.4034732580184937,
      "learning_rate": 9.386052631578948e-05,
      "loss": 0.367,
      "step": 43349
    },
    {
      "epoch": 2.427483480792922,
      "grad_norm": 1.182443380355835,
      "learning_rate": 9.386026315789474e-05,
      "loss": 0.4221,
      "step": 43350
    },
    {
      "epoch": 2.427539478105051,
      "grad_norm": 1.2632993459701538,
      "learning_rate": 9.386e-05,
      "loss": 0.4657,
      "step": 43351
    },
    {
      "epoch": 2.42759547541718,
      "grad_norm": 1.04806649684906,
      "learning_rate": 9.385973684210527e-05,
      "loss": 0.4245,
      "step": 43352
    },
    {
      "epoch": 2.427651472729309,
      "grad_norm": 1.074133276939392,
      "learning_rate": 9.385947368421053e-05,
      "loss": 0.3508,
      "step": 43353
    },
    {
      "epoch": 2.427707470041438,
      "grad_norm": 1.3391444683074951,
      "learning_rate": 9.38592105263158e-05,
      "loss": 0.3284,
      "step": 43354
    },
    {
      "epoch": 2.427763467353567,
      "grad_norm": 1.1975843906402588,
      "learning_rate": 9.385894736842105e-05,
      "loss": 0.4113,
      "step": 43355
    },
    {
      "epoch": 2.427819464665696,
      "grad_norm": 1.0653481483459473,
      "learning_rate": 9.385868421052631e-05,
      "loss": 0.3907,
      "step": 43356
    },
    {
      "epoch": 2.427875461977825,
      "grad_norm": 1.3801605701446533,
      "learning_rate": 9.385842105263158e-05,
      "loss": 0.4362,
      "step": 43357
    },
    {
      "epoch": 2.427931459289954,
      "grad_norm": 0.9844357967376709,
      "learning_rate": 9.385815789473684e-05,
      "loss": 0.3961,
      "step": 43358
    },
    {
      "epoch": 2.427987456602083,
      "grad_norm": 1.298036813735962,
      "learning_rate": 9.385789473684212e-05,
      "loss": 0.4182,
      "step": 43359
    },
    {
      "epoch": 2.428043453914212,
      "grad_norm": 1.1005222797393799,
      "learning_rate": 9.385763157894736e-05,
      "loss": 0.4708,
      "step": 43360
    },
    {
      "epoch": 2.428099451226341,
      "grad_norm": 1.1731996536254883,
      "learning_rate": 9.385736842105264e-05,
      "loss": 0.4222,
      "step": 43361
    },
    {
      "epoch": 2.42815544853847,
      "grad_norm": 1.137898564338684,
      "learning_rate": 9.38571052631579e-05,
      "loss": 0.3345,
      "step": 43362
    },
    {
      "epoch": 2.4282114458505992,
      "grad_norm": 1.2865368127822876,
      "learning_rate": 9.385684210526317e-05,
      "loss": 0.41,
      "step": 43363
    },
    {
      "epoch": 2.4282674431627282,
      "grad_norm": 1.557415246963501,
      "learning_rate": 9.385657894736843e-05,
      "loss": 0.4678,
      "step": 43364
    },
    {
      "epoch": 2.4283234404748573,
      "grad_norm": 1.3160344362258911,
      "learning_rate": 9.385631578947369e-05,
      "loss": 0.4289,
      "step": 43365
    },
    {
      "epoch": 2.4283794377869863,
      "grad_norm": 1.038507342338562,
      "learning_rate": 9.385605263157895e-05,
      "loss": 0.4918,
      "step": 43366
    },
    {
      "epoch": 2.4284354350991153,
      "grad_norm": 1.4419865608215332,
      "learning_rate": 9.385578947368422e-05,
      "loss": 0.4504,
      "step": 43367
    },
    {
      "epoch": 2.4284914324112443,
      "grad_norm": 1.0739492177963257,
      "learning_rate": 9.385552631578948e-05,
      "loss": 0.331,
      "step": 43368
    },
    {
      "epoch": 2.4285474297233733,
      "grad_norm": 1.1418654918670654,
      "learning_rate": 9.385526315789474e-05,
      "loss": 0.417,
      "step": 43369
    },
    {
      "epoch": 2.4286034270355024,
      "grad_norm": 1.1604127883911133,
      "learning_rate": 9.3855e-05,
      "loss": 0.3685,
      "step": 43370
    },
    {
      "epoch": 2.4286594243476314,
      "grad_norm": 3.397444009780884,
      "learning_rate": 9.385473684210528e-05,
      "loss": 0.4888,
      "step": 43371
    },
    {
      "epoch": 2.4287154216597604,
      "grad_norm": 1.2603759765625,
      "learning_rate": 9.385447368421053e-05,
      "loss": 0.5675,
      "step": 43372
    },
    {
      "epoch": 2.4287714189718894,
      "grad_norm": 1.1021838188171387,
      "learning_rate": 9.38542105263158e-05,
      "loss": 0.4028,
      "step": 43373
    },
    {
      "epoch": 2.4288274162840184,
      "grad_norm": 1.0308390855789185,
      "learning_rate": 9.385394736842105e-05,
      "loss": 0.3346,
      "step": 43374
    },
    {
      "epoch": 2.4288834135961475,
      "grad_norm": 1.1870626211166382,
      "learning_rate": 9.385368421052631e-05,
      "loss": 0.5407,
      "step": 43375
    },
    {
      "epoch": 2.4289394109082765,
      "grad_norm": 1.2564014196395874,
      "learning_rate": 9.385342105263159e-05,
      "loss": 0.4037,
      "step": 43376
    },
    {
      "epoch": 2.4289954082204055,
      "grad_norm": 1.1590462923049927,
      "learning_rate": 9.385315789473685e-05,
      "loss": 0.375,
      "step": 43377
    },
    {
      "epoch": 2.4290514055325345,
      "grad_norm": 1.1806632280349731,
      "learning_rate": 9.38528947368421e-05,
      "loss": 0.5893,
      "step": 43378
    },
    {
      "epoch": 2.4291074028446635,
      "grad_norm": 1.3714137077331543,
      "learning_rate": 9.385263157894737e-05,
      "loss": 0.4624,
      "step": 43379
    },
    {
      "epoch": 2.4291634001567926,
      "grad_norm": 1.3838032484054565,
      "learning_rate": 9.385236842105264e-05,
      "loss": 0.313,
      "step": 43380
    },
    {
      "epoch": 2.4292193974689216,
      "grad_norm": 1.2200109958648682,
      "learning_rate": 9.38521052631579e-05,
      "loss": 0.4538,
      "step": 43381
    },
    {
      "epoch": 2.4292753947810506,
      "grad_norm": 1.0471696853637695,
      "learning_rate": 9.385184210526316e-05,
      "loss": 0.3655,
      "step": 43382
    },
    {
      "epoch": 2.4293313920931796,
      "grad_norm": 1.198943018913269,
      "learning_rate": 9.385157894736842e-05,
      "loss": 0.4556,
      "step": 43383
    },
    {
      "epoch": 2.4293873894053086,
      "grad_norm": 1.4081692695617676,
      "learning_rate": 9.385131578947369e-05,
      "loss": 0.4586,
      "step": 43384
    },
    {
      "epoch": 2.4294433867174376,
      "grad_norm": 1.3346951007843018,
      "learning_rate": 9.385105263157895e-05,
      "loss": 0.3244,
      "step": 43385
    },
    {
      "epoch": 2.4294993840295667,
      "grad_norm": 1.3078590631484985,
      "learning_rate": 9.385078947368423e-05,
      "loss": 0.3655,
      "step": 43386
    },
    {
      "epoch": 2.4295553813416957,
      "grad_norm": 1.1638308763504028,
      "learning_rate": 9.385052631578947e-05,
      "loss": 0.4038,
      "step": 43387
    },
    {
      "epoch": 2.4296113786538247,
      "grad_norm": 1.1087274551391602,
      "learning_rate": 9.385026315789474e-05,
      "loss": 0.3192,
      "step": 43388
    },
    {
      "epoch": 2.4296673759659537,
      "grad_norm": 1.1162607669830322,
      "learning_rate": 9.385e-05,
      "loss": 0.43,
      "step": 43389
    },
    {
      "epoch": 2.4297233732780827,
      "grad_norm": 1.0834661722183228,
      "learning_rate": 9.384973684210526e-05,
      "loss": 0.3336,
      "step": 43390
    },
    {
      "epoch": 2.4297793705902118,
      "grad_norm": 1.3958379030227661,
      "learning_rate": 9.384947368421054e-05,
      "loss": 0.4832,
      "step": 43391
    },
    {
      "epoch": 2.429835367902341,
      "grad_norm": 1.2369455099105835,
      "learning_rate": 9.384921052631578e-05,
      "loss": 0.5863,
      "step": 43392
    },
    {
      "epoch": 2.42989136521447,
      "grad_norm": 1.0227656364440918,
      "learning_rate": 9.384894736842106e-05,
      "loss": 0.3514,
      "step": 43393
    },
    {
      "epoch": 2.429947362526599,
      "grad_norm": 1.047366976737976,
      "learning_rate": 9.384868421052632e-05,
      "loss": 0.3925,
      "step": 43394
    },
    {
      "epoch": 2.430003359838728,
      "grad_norm": 1.0129384994506836,
      "learning_rate": 9.384842105263159e-05,
      "loss": 0.4072,
      "step": 43395
    },
    {
      "epoch": 2.430059357150857,
      "grad_norm": 0.9804482460021973,
      "learning_rate": 9.384815789473685e-05,
      "loss": 0.3473,
      "step": 43396
    },
    {
      "epoch": 2.430115354462986,
      "grad_norm": 1.4185374975204468,
      "learning_rate": 9.384789473684211e-05,
      "loss": 0.3489,
      "step": 43397
    },
    {
      "epoch": 2.430171351775115,
      "grad_norm": 1.0564411878585815,
      "learning_rate": 9.384763157894737e-05,
      "loss": 0.3962,
      "step": 43398
    },
    {
      "epoch": 2.430227349087244,
      "grad_norm": 1.1915143728256226,
      "learning_rate": 9.384736842105264e-05,
      "loss": 0.3832,
      "step": 43399
    },
    {
      "epoch": 2.430283346399373,
      "grad_norm": 1.1948678493499756,
      "learning_rate": 9.38471052631579e-05,
      "loss": 0.4527,
      "step": 43400
    },
    {
      "epoch": 2.430339343711502,
      "grad_norm": 1.2793842554092407,
      "learning_rate": 9.384684210526316e-05,
      "loss": 0.5447,
      "step": 43401
    },
    {
      "epoch": 2.430395341023631,
      "grad_norm": 1.0344980955123901,
      "learning_rate": 9.384657894736842e-05,
      "loss": 0.4154,
      "step": 43402
    },
    {
      "epoch": 2.43045133833576,
      "grad_norm": 1.1797325611114502,
      "learning_rate": 9.38463157894737e-05,
      "loss": 0.3862,
      "step": 43403
    },
    {
      "epoch": 2.430507335647889,
      "grad_norm": 1.1524313688278198,
      "learning_rate": 9.384605263157895e-05,
      "loss": 0.3115,
      "step": 43404
    },
    {
      "epoch": 2.430563332960018,
      "grad_norm": 1.114667534828186,
      "learning_rate": 9.384578947368421e-05,
      "loss": 0.4526,
      "step": 43405
    },
    {
      "epoch": 2.430619330272147,
      "grad_norm": 1.274276852607727,
      "learning_rate": 9.384552631578947e-05,
      "loss": 0.3929,
      "step": 43406
    },
    {
      "epoch": 2.430675327584276,
      "grad_norm": 1.360005259513855,
      "learning_rate": 9.384526315789473e-05,
      "loss": 0.4642,
      "step": 43407
    },
    {
      "epoch": 2.430731324896405,
      "grad_norm": 7.713287830352783,
      "learning_rate": 9.384500000000001e-05,
      "loss": 0.398,
      "step": 43408
    },
    {
      "epoch": 2.430787322208534,
      "grad_norm": 1.1749053001403809,
      "learning_rate": 9.384473684210527e-05,
      "loss": 0.453,
      "step": 43409
    },
    {
      "epoch": 2.430843319520663,
      "grad_norm": 1.1834076642990112,
      "learning_rate": 9.384447368421053e-05,
      "loss": 0.4047,
      "step": 43410
    },
    {
      "epoch": 2.430899316832792,
      "grad_norm": 1.1666511297225952,
      "learning_rate": 9.384421052631579e-05,
      "loss": 0.3991,
      "step": 43411
    },
    {
      "epoch": 2.430955314144921,
      "grad_norm": 1.813346266746521,
      "learning_rate": 9.384394736842106e-05,
      "loss": 0.4145,
      "step": 43412
    },
    {
      "epoch": 2.43101131145705,
      "grad_norm": 0.9740927815437317,
      "learning_rate": 9.384368421052632e-05,
      "loss": 0.5302,
      "step": 43413
    },
    {
      "epoch": 2.431067308769179,
      "grad_norm": 1.1508328914642334,
      "learning_rate": 9.384342105263159e-05,
      "loss": 0.4138,
      "step": 43414
    },
    {
      "epoch": 2.431123306081308,
      "grad_norm": 1.4510514736175537,
      "learning_rate": 9.384315789473684e-05,
      "loss": 0.3605,
      "step": 43415
    },
    {
      "epoch": 2.4311793033934372,
      "grad_norm": 1.2821334600448608,
      "learning_rate": 9.384289473684211e-05,
      "loss": 0.4614,
      "step": 43416
    },
    {
      "epoch": 2.4312353007055663,
      "grad_norm": 1.3336087465286255,
      "learning_rate": 9.384263157894737e-05,
      "loss": 0.4015,
      "step": 43417
    },
    {
      "epoch": 2.4312912980176953,
      "grad_norm": 1.3297868967056274,
      "learning_rate": 9.384236842105264e-05,
      "loss": 0.3959,
      "step": 43418
    },
    {
      "epoch": 2.4313472953298243,
      "grad_norm": 1.1191812753677368,
      "learning_rate": 9.384210526315789e-05,
      "loss": 0.3282,
      "step": 43419
    },
    {
      "epoch": 2.4314032926419533,
      "grad_norm": 1.187023401260376,
      "learning_rate": 9.384184210526316e-05,
      "loss": 0.4345,
      "step": 43420
    },
    {
      "epoch": 2.4314592899540823,
      "grad_norm": 1.2757049798965454,
      "learning_rate": 9.384157894736842e-05,
      "loss": 0.3805,
      "step": 43421
    },
    {
      "epoch": 2.4315152872662114,
      "grad_norm": 1.1709263324737549,
      "learning_rate": 9.38413157894737e-05,
      "loss": 0.3898,
      "step": 43422
    },
    {
      "epoch": 2.4315712845783404,
      "grad_norm": 1.2875988483428955,
      "learning_rate": 9.384105263157896e-05,
      "loss": 0.4805,
      "step": 43423
    },
    {
      "epoch": 2.4316272818904694,
      "grad_norm": 1.3478721380233765,
      "learning_rate": 9.38407894736842e-05,
      "loss": 0.4032,
      "step": 43424
    },
    {
      "epoch": 2.4316832792025984,
      "grad_norm": 1.7640914916992188,
      "learning_rate": 9.384052631578948e-05,
      "loss": 0.4563,
      "step": 43425
    },
    {
      "epoch": 2.4317392765147274,
      "grad_norm": 1.093003749847412,
      "learning_rate": 9.384026315789474e-05,
      "loss": 0.3327,
      "step": 43426
    },
    {
      "epoch": 2.4317952738268565,
      "grad_norm": 1.1120656728744507,
      "learning_rate": 9.384000000000001e-05,
      "loss": 0.4486,
      "step": 43427
    },
    {
      "epoch": 2.4318512711389855,
      "grad_norm": 1.2630828619003296,
      "learning_rate": 9.383973684210527e-05,
      "loss": 0.4555,
      "step": 43428
    },
    {
      "epoch": 2.4319072684511145,
      "grad_norm": 0.9723885655403137,
      "learning_rate": 9.383947368421053e-05,
      "loss": 0.3497,
      "step": 43429
    },
    {
      "epoch": 2.4319632657632435,
      "grad_norm": 1.4140599966049194,
      "learning_rate": 9.383921052631579e-05,
      "loss": 0.491,
      "step": 43430
    },
    {
      "epoch": 2.4320192630753725,
      "grad_norm": 1.132109522819519,
      "learning_rate": 9.383894736842106e-05,
      "loss": 0.4306,
      "step": 43431
    },
    {
      "epoch": 2.4320752603875015,
      "grad_norm": 1.426221251487732,
      "learning_rate": 9.383868421052632e-05,
      "loss": 0.586,
      "step": 43432
    },
    {
      "epoch": 2.4321312576996306,
      "grad_norm": 1.649672031402588,
      "learning_rate": 9.383842105263158e-05,
      "loss": 0.4764,
      "step": 43433
    },
    {
      "epoch": 2.4321872550117596,
      "grad_norm": 1.1716945171356201,
      "learning_rate": 9.383815789473684e-05,
      "loss": 0.3141,
      "step": 43434
    },
    {
      "epoch": 2.4322432523238886,
      "grad_norm": 1.1834070682525635,
      "learning_rate": 9.383789473684211e-05,
      "loss": 0.4313,
      "step": 43435
    },
    {
      "epoch": 2.4322992496360176,
      "grad_norm": 1.052927017211914,
      "learning_rate": 9.383763157894737e-05,
      "loss": 0.3753,
      "step": 43436
    },
    {
      "epoch": 2.4323552469481466,
      "grad_norm": 1.1493428945541382,
      "learning_rate": 9.383736842105263e-05,
      "loss": 0.3408,
      "step": 43437
    },
    {
      "epoch": 2.4324112442602757,
      "grad_norm": 1.233973503112793,
      "learning_rate": 9.38371052631579e-05,
      "loss": 0.2602,
      "step": 43438
    },
    {
      "epoch": 2.4324672415724047,
      "grad_norm": 1.2100218534469604,
      "learning_rate": 9.383684210526317e-05,
      "loss": 0.4832,
      "step": 43439
    },
    {
      "epoch": 2.4325232388845337,
      "grad_norm": 1.162550687789917,
      "learning_rate": 9.383657894736843e-05,
      "loss": 0.3711,
      "step": 43440
    },
    {
      "epoch": 2.4325792361966627,
      "grad_norm": 1.1472750902175903,
      "learning_rate": 9.383631578947369e-05,
      "loss": 0.3832,
      "step": 43441
    },
    {
      "epoch": 2.4326352335087917,
      "grad_norm": 1.3634388446807861,
      "learning_rate": 9.383605263157895e-05,
      "loss": 0.4464,
      "step": 43442
    },
    {
      "epoch": 2.4326912308209208,
      "grad_norm": 1.2247101068496704,
      "learning_rate": 9.38357894736842e-05,
      "loss": 0.3486,
      "step": 43443
    },
    {
      "epoch": 2.4327472281330498,
      "grad_norm": 1.378816843032837,
      "learning_rate": 9.383552631578948e-05,
      "loss": 0.4521,
      "step": 43444
    },
    {
      "epoch": 2.432803225445179,
      "grad_norm": 1.5216033458709717,
      "learning_rate": 9.383526315789474e-05,
      "loss": 0.4524,
      "step": 43445
    },
    {
      "epoch": 2.432859222757308,
      "grad_norm": 1.3239271640777588,
      "learning_rate": 9.383500000000001e-05,
      "loss": 0.413,
      "step": 43446
    },
    {
      "epoch": 2.432915220069437,
      "grad_norm": 1.3717900514602661,
      "learning_rate": 9.383473684210526e-05,
      "loss": 0.5352,
      "step": 43447
    },
    {
      "epoch": 2.432971217381566,
      "grad_norm": 1.1783660650253296,
      "learning_rate": 9.383447368421053e-05,
      "loss": 0.3006,
      "step": 43448
    },
    {
      "epoch": 2.433027214693695,
      "grad_norm": 1.282853364944458,
      "learning_rate": 9.383421052631579e-05,
      "loss": 0.3663,
      "step": 43449
    },
    {
      "epoch": 2.433083212005824,
      "grad_norm": 1.1573971509933472,
      "learning_rate": 9.383394736842106e-05,
      "loss": 0.4207,
      "step": 43450
    },
    {
      "epoch": 2.433139209317953,
      "grad_norm": 1.1459968090057373,
      "learning_rate": 9.383368421052632e-05,
      "loss": 0.3567,
      "step": 43451
    },
    {
      "epoch": 2.433195206630082,
      "grad_norm": 1.0860177278518677,
      "learning_rate": 9.383342105263158e-05,
      "loss": 0.391,
      "step": 43452
    },
    {
      "epoch": 2.433251203942211,
      "grad_norm": 1.3015120029449463,
      "learning_rate": 9.383315789473684e-05,
      "loss": 0.3851,
      "step": 43453
    },
    {
      "epoch": 2.43330720125434,
      "grad_norm": 1.2159581184387207,
      "learning_rate": 9.383289473684212e-05,
      "loss": 0.3658,
      "step": 43454
    },
    {
      "epoch": 2.433363198566469,
      "grad_norm": 1.1927125453948975,
      "learning_rate": 9.383263157894738e-05,
      "loss": 0.409,
      "step": 43455
    },
    {
      "epoch": 2.433419195878598,
      "grad_norm": 1.0856728553771973,
      "learning_rate": 9.383236842105264e-05,
      "loss": 0.4171,
      "step": 43456
    },
    {
      "epoch": 2.433475193190727,
      "grad_norm": 1.1201846599578857,
      "learning_rate": 9.38321052631579e-05,
      "loss": 0.4162,
      "step": 43457
    },
    {
      "epoch": 2.433531190502856,
      "grad_norm": 0.9723695516586304,
      "learning_rate": 9.383184210526316e-05,
      "loss": 0.3514,
      "step": 43458
    },
    {
      "epoch": 2.433587187814985,
      "grad_norm": 1.7527860403060913,
      "learning_rate": 9.383157894736843e-05,
      "loss": 0.5715,
      "step": 43459
    },
    {
      "epoch": 2.433643185127114,
      "grad_norm": 1.070355772972107,
      "learning_rate": 9.383131578947369e-05,
      "loss": 0.4033,
      "step": 43460
    },
    {
      "epoch": 2.433699182439243,
      "grad_norm": 1.1626781225204468,
      "learning_rate": 9.383105263157895e-05,
      "loss": 0.3528,
      "step": 43461
    },
    {
      "epoch": 2.433755179751372,
      "grad_norm": 1.237432599067688,
      "learning_rate": 9.383078947368421e-05,
      "loss": 0.3989,
      "step": 43462
    },
    {
      "epoch": 2.433811177063501,
      "grad_norm": 1.386903166770935,
      "learning_rate": 9.383052631578948e-05,
      "loss": 0.4431,
      "step": 43463
    },
    {
      "epoch": 2.43386717437563,
      "grad_norm": 1.3324124813079834,
      "learning_rate": 9.383026315789474e-05,
      "loss": 0.5316,
      "step": 43464
    },
    {
      "epoch": 2.433923171687759,
      "grad_norm": 1.1503700017929077,
      "learning_rate": 9.383e-05,
      "loss": 0.3643,
      "step": 43465
    },
    {
      "epoch": 2.433979168999888,
      "grad_norm": 1.1987820863723755,
      "learning_rate": 9.382973684210526e-05,
      "loss": 0.3456,
      "step": 43466
    },
    {
      "epoch": 2.434035166312017,
      "grad_norm": 1.2245421409606934,
      "learning_rate": 9.382947368421053e-05,
      "loss": 0.3478,
      "step": 43467
    },
    {
      "epoch": 2.434091163624146,
      "grad_norm": 1.0515055656433105,
      "learning_rate": 9.38292105263158e-05,
      "loss": 0.3727,
      "step": 43468
    },
    {
      "epoch": 2.4341471609362753,
      "grad_norm": 1.222813367843628,
      "learning_rate": 9.382894736842107e-05,
      "loss": 0.4322,
      "step": 43469
    },
    {
      "epoch": 2.434203158248404,
      "grad_norm": 0.9088592529296875,
      "learning_rate": 9.382868421052631e-05,
      "loss": 0.3074,
      "step": 43470
    },
    {
      "epoch": 2.4342591555605333,
      "grad_norm": 1.2925727367401123,
      "learning_rate": 9.382842105263159e-05,
      "loss": 0.4125,
      "step": 43471
    },
    {
      "epoch": 2.434315152872662,
      "grad_norm": 1.3260629177093506,
      "learning_rate": 9.382815789473685e-05,
      "loss": 0.4035,
      "step": 43472
    },
    {
      "epoch": 2.4343711501847913,
      "grad_norm": 1.110970377922058,
      "learning_rate": 9.382789473684212e-05,
      "loss": 0.3778,
      "step": 43473
    },
    {
      "epoch": 2.43442714749692,
      "grad_norm": 0.9924740791320801,
      "learning_rate": 9.382763157894737e-05,
      "loss": 0.3712,
      "step": 43474
    },
    {
      "epoch": 2.4344831448090494,
      "grad_norm": 1.052088975906372,
      "learning_rate": 9.382736842105264e-05,
      "loss": 0.3325,
      "step": 43475
    },
    {
      "epoch": 2.434539142121178,
      "grad_norm": 1.1123687028884888,
      "learning_rate": 9.38271052631579e-05,
      "loss": 0.3348,
      "step": 43476
    },
    {
      "epoch": 2.4345951394333074,
      "grad_norm": 1.0884625911712646,
      "learning_rate": 9.382684210526316e-05,
      "loss": 0.3563,
      "step": 43477
    },
    {
      "epoch": 2.434651136745436,
      "grad_norm": 1.2509570121765137,
      "learning_rate": 9.382657894736843e-05,
      "loss": 0.4616,
      "step": 43478
    },
    {
      "epoch": 2.4347071340575654,
      "grad_norm": 1.0460319519042969,
      "learning_rate": 9.382631578947368e-05,
      "loss": 0.2775,
      "step": 43479
    },
    {
      "epoch": 2.434763131369694,
      "grad_norm": 1.2151391506195068,
      "learning_rate": 9.382605263157895e-05,
      "loss": 0.4061,
      "step": 43480
    },
    {
      "epoch": 2.4348191286818235,
      "grad_norm": 0.98110032081604,
      "learning_rate": 9.382578947368421e-05,
      "loss": 0.3884,
      "step": 43481
    },
    {
      "epoch": 2.434875125993952,
      "grad_norm": 1.818224549293518,
      "learning_rate": 9.382552631578948e-05,
      "loss": 0.5784,
      "step": 43482
    },
    {
      "epoch": 2.4349311233060815,
      "grad_norm": 1.1985265016555786,
      "learning_rate": 9.382526315789474e-05,
      "loss": 0.4301,
      "step": 43483
    },
    {
      "epoch": 2.43498712061821,
      "grad_norm": 1.1882362365722656,
      "learning_rate": 9.3825e-05,
      "loss": 0.4501,
      "step": 43484
    },
    {
      "epoch": 2.4350431179303396,
      "grad_norm": 1.0997867584228516,
      "learning_rate": 9.382473684210526e-05,
      "loss": 0.3908,
      "step": 43485
    },
    {
      "epoch": 2.435099115242468,
      "grad_norm": 1.283348560333252,
      "learning_rate": 9.382447368421054e-05,
      "loss": 0.4327,
      "step": 43486
    },
    {
      "epoch": 2.4351551125545976,
      "grad_norm": 1.1015461683273315,
      "learning_rate": 9.38242105263158e-05,
      "loss": 0.3612,
      "step": 43487
    },
    {
      "epoch": 2.435211109866726,
      "grad_norm": 1.0880892276763916,
      "learning_rate": 9.382394736842106e-05,
      "loss": 0.4804,
      "step": 43488
    },
    {
      "epoch": 2.4352671071788556,
      "grad_norm": 1.246701955795288,
      "learning_rate": 9.382368421052632e-05,
      "loss": 0.4529,
      "step": 43489
    },
    {
      "epoch": 2.435323104490984,
      "grad_norm": 1.1859369277954102,
      "learning_rate": 9.382342105263159e-05,
      "loss": 0.3615,
      "step": 43490
    },
    {
      "epoch": 2.4353791018031137,
      "grad_norm": 1.1748050451278687,
      "learning_rate": 9.382315789473685e-05,
      "loss": 0.3863,
      "step": 43491
    },
    {
      "epoch": 2.4354350991152423,
      "grad_norm": 1.1295069456100464,
      "learning_rate": 9.382289473684211e-05,
      "loss": 0.3886,
      "step": 43492
    },
    {
      "epoch": 2.4354910964273717,
      "grad_norm": 1.160613775253296,
      "learning_rate": 9.382263157894737e-05,
      "loss": 0.416,
      "step": 43493
    },
    {
      "epoch": 2.4355470937395003,
      "grad_norm": 1.717684030532837,
      "learning_rate": 9.382236842105263e-05,
      "loss": 0.4655,
      "step": 43494
    },
    {
      "epoch": 2.4356030910516298,
      "grad_norm": 1.2494808435440063,
      "learning_rate": 9.38221052631579e-05,
      "loss": 0.5969,
      "step": 43495
    },
    {
      "epoch": 2.4356590883637583,
      "grad_norm": 1.0145930051803589,
      "learning_rate": 9.382184210526316e-05,
      "loss": 0.3805,
      "step": 43496
    },
    {
      "epoch": 2.435715085675888,
      "grad_norm": 1.2939825057983398,
      "learning_rate": 9.382157894736842e-05,
      "loss": 0.4869,
      "step": 43497
    },
    {
      "epoch": 2.4357710829880164,
      "grad_norm": 1.1067980527877808,
      "learning_rate": 9.382131578947368e-05,
      "loss": 0.5308,
      "step": 43498
    },
    {
      "epoch": 2.435827080300146,
      "grad_norm": 1.2373117208480835,
      "learning_rate": 9.382105263157895e-05,
      "loss": 0.3543,
      "step": 43499
    },
    {
      "epoch": 2.4358830776122744,
      "grad_norm": 1.1672812700271606,
      "learning_rate": 9.382078947368421e-05,
      "loss": 0.3616,
      "step": 43500
    },
    {
      "epoch": 2.435939074924404,
      "grad_norm": 1.0591599941253662,
      "learning_rate": 9.382052631578949e-05,
      "loss": 0.3633,
      "step": 43501
    },
    {
      "epoch": 2.4359950722365324,
      "grad_norm": 1.1283249855041504,
      "learning_rate": 9.382026315789473e-05,
      "loss": 0.4081,
      "step": 43502
    },
    {
      "epoch": 2.436051069548662,
      "grad_norm": 1.4259402751922607,
      "learning_rate": 9.382e-05,
      "loss": 0.3711,
      "step": 43503
    },
    {
      "epoch": 2.4361070668607905,
      "grad_norm": 1.2054909467697144,
      "learning_rate": 9.381973684210527e-05,
      "loss": 0.564,
      "step": 43504
    },
    {
      "epoch": 2.43616306417292,
      "grad_norm": 1.467523455619812,
      "learning_rate": 9.381947368421054e-05,
      "loss": 0.4992,
      "step": 43505
    },
    {
      "epoch": 2.4362190614850485,
      "grad_norm": 1.2247976064682007,
      "learning_rate": 9.38192105263158e-05,
      "loss": 0.3736,
      "step": 43506
    },
    {
      "epoch": 2.436275058797178,
      "grad_norm": 1.647125244140625,
      "learning_rate": 9.381894736842106e-05,
      "loss": 0.3764,
      "step": 43507
    },
    {
      "epoch": 2.4363310561093066,
      "grad_norm": 1.184555172920227,
      "learning_rate": 9.381868421052632e-05,
      "loss": 0.3848,
      "step": 43508
    },
    {
      "epoch": 2.4363870534214356,
      "grad_norm": 1.1530085802078247,
      "learning_rate": 9.381842105263159e-05,
      "loss": 0.5664,
      "step": 43509
    },
    {
      "epoch": 2.4364430507335646,
      "grad_norm": 1.1183990240097046,
      "learning_rate": 9.381815789473685e-05,
      "loss": 0.3885,
      "step": 43510
    },
    {
      "epoch": 2.4364990480456936,
      "grad_norm": 1.1145797967910767,
      "learning_rate": 9.381789473684211e-05,
      "loss": 0.335,
      "step": 43511
    },
    {
      "epoch": 2.4365550453578226,
      "grad_norm": 1.2208172082901,
      "learning_rate": 9.381763157894737e-05,
      "loss": 0.3936,
      "step": 43512
    },
    {
      "epoch": 2.4366110426699517,
      "grad_norm": 1.1112356185913086,
      "learning_rate": 9.381736842105263e-05,
      "loss": 0.3554,
      "step": 43513
    },
    {
      "epoch": 2.4366670399820807,
      "grad_norm": 1.2250603437423706,
      "learning_rate": 9.38171052631579e-05,
      "loss": 0.5187,
      "step": 43514
    },
    {
      "epoch": 2.4367230372942097,
      "grad_norm": 1.2865691184997559,
      "learning_rate": 9.381684210526316e-05,
      "loss": 0.4522,
      "step": 43515
    },
    {
      "epoch": 2.4367790346063387,
      "grad_norm": 1.0813944339752197,
      "learning_rate": 9.381657894736842e-05,
      "loss": 0.3706,
      "step": 43516
    },
    {
      "epoch": 2.4368350319184677,
      "grad_norm": 1.374730110168457,
      "learning_rate": 9.381631578947368e-05,
      "loss": 0.5834,
      "step": 43517
    },
    {
      "epoch": 2.4368910292305968,
      "grad_norm": 1.0086525678634644,
      "learning_rate": 9.381605263157896e-05,
      "loss": 0.3376,
      "step": 43518
    },
    {
      "epoch": 2.4369470265427258,
      "grad_norm": 1.1249088048934937,
      "learning_rate": 9.381578947368422e-05,
      "loss": 0.3933,
      "step": 43519
    },
    {
      "epoch": 2.437003023854855,
      "grad_norm": 1.0572181940078735,
      "learning_rate": 9.381552631578948e-05,
      "loss": 0.308,
      "step": 43520
    },
    {
      "epoch": 2.437059021166984,
      "grad_norm": 1.2151226997375488,
      "learning_rate": 9.381526315789474e-05,
      "loss": 0.4717,
      "step": 43521
    },
    {
      "epoch": 2.437115018479113,
      "grad_norm": 1.1393799781799316,
      "learning_rate": 9.381500000000001e-05,
      "loss": 0.3594,
      "step": 43522
    },
    {
      "epoch": 2.437171015791242,
      "grad_norm": 1.05832040309906,
      "learning_rate": 9.381473684210527e-05,
      "loss": 0.3329,
      "step": 43523
    },
    {
      "epoch": 2.437227013103371,
      "grad_norm": 1.2633845806121826,
      "learning_rate": 9.381447368421054e-05,
      "loss": 0.4283,
      "step": 43524
    },
    {
      "epoch": 2.4372830104155,
      "grad_norm": 1.177384376525879,
      "learning_rate": 9.381421052631579e-05,
      "loss": 0.3734,
      "step": 43525
    },
    {
      "epoch": 2.437339007727629,
      "grad_norm": 1.1382733583450317,
      "learning_rate": 9.381394736842106e-05,
      "loss": 0.4136,
      "step": 43526
    },
    {
      "epoch": 2.437395005039758,
      "grad_norm": 1.2176085710525513,
      "learning_rate": 9.381368421052632e-05,
      "loss": 0.4514,
      "step": 43527
    },
    {
      "epoch": 2.437451002351887,
      "grad_norm": 1.3244165182113647,
      "learning_rate": 9.381342105263158e-05,
      "loss": 0.4183,
      "step": 43528
    },
    {
      "epoch": 2.437506999664016,
      "grad_norm": 1.1814717054367065,
      "learning_rate": 9.381315789473684e-05,
      "loss": 0.5872,
      "step": 43529
    },
    {
      "epoch": 2.437562996976145,
      "grad_norm": 1.2787936925888062,
      "learning_rate": 9.38128947368421e-05,
      "loss": 0.441,
      "step": 43530
    },
    {
      "epoch": 2.437618994288274,
      "grad_norm": 1.197939157485962,
      "learning_rate": 9.381263157894737e-05,
      "loss": 0.5021,
      "step": 43531
    },
    {
      "epoch": 2.437674991600403,
      "grad_norm": 1.0469249486923218,
      "learning_rate": 9.381236842105263e-05,
      "loss": 0.422,
      "step": 43532
    },
    {
      "epoch": 2.437730988912532,
      "grad_norm": 1.389747142791748,
      "learning_rate": 9.38121052631579e-05,
      "loss": 0.3812,
      "step": 43533
    },
    {
      "epoch": 2.437786986224661,
      "grad_norm": 1.1925057172775269,
      "learning_rate": 9.381184210526315e-05,
      "loss": 0.4221,
      "step": 43534
    },
    {
      "epoch": 2.43784298353679,
      "grad_norm": 0.9684349298477173,
      "learning_rate": 9.381157894736843e-05,
      "loss": 0.4149,
      "step": 43535
    },
    {
      "epoch": 2.437898980848919,
      "grad_norm": 1.0804325342178345,
      "learning_rate": 9.381131578947369e-05,
      "loss": 0.3414,
      "step": 43536
    },
    {
      "epoch": 2.437954978161048,
      "grad_norm": 1.3908649682998657,
      "learning_rate": 9.381105263157896e-05,
      "loss": 0.4029,
      "step": 43537
    },
    {
      "epoch": 2.438010975473177,
      "grad_norm": 1.1876524686813354,
      "learning_rate": 9.381078947368422e-05,
      "loss": 0.4228,
      "step": 43538
    },
    {
      "epoch": 2.438066972785306,
      "grad_norm": 1.1097257137298584,
      "learning_rate": 9.381052631578948e-05,
      "loss": 0.3393,
      "step": 43539
    },
    {
      "epoch": 2.438122970097435,
      "grad_norm": 1.1329421997070312,
      "learning_rate": 9.381026315789474e-05,
      "loss": 0.4043,
      "step": 43540
    },
    {
      "epoch": 2.438178967409564,
      "grad_norm": 1.111767053604126,
      "learning_rate": 9.381000000000001e-05,
      "loss": 0.4088,
      "step": 43541
    },
    {
      "epoch": 2.438234964721693,
      "grad_norm": 1.3643239736557007,
      "learning_rate": 9.380973684210527e-05,
      "loss": 0.4138,
      "step": 43542
    },
    {
      "epoch": 2.4382909620338222,
      "grad_norm": 0.9669643640518188,
      "learning_rate": 9.380947368421053e-05,
      "loss": 0.3597,
      "step": 43543
    },
    {
      "epoch": 2.4383469593459512,
      "grad_norm": 1.2528917789459229,
      "learning_rate": 9.380921052631579e-05,
      "loss": 0.3543,
      "step": 43544
    },
    {
      "epoch": 2.4384029566580803,
      "grad_norm": 1.1798632144927979,
      "learning_rate": 9.380894736842105e-05,
      "loss": 0.4696,
      "step": 43545
    },
    {
      "epoch": 2.4384589539702093,
      "grad_norm": 1.4390419721603394,
      "learning_rate": 9.380868421052632e-05,
      "loss": 0.6272,
      "step": 43546
    },
    {
      "epoch": 2.4385149512823383,
      "grad_norm": 1.1236462593078613,
      "learning_rate": 9.380842105263158e-05,
      "loss": 0.3224,
      "step": 43547
    },
    {
      "epoch": 2.4385709485944673,
      "grad_norm": 1.0145539045333862,
      "learning_rate": 9.380815789473684e-05,
      "loss": 0.2538,
      "step": 43548
    },
    {
      "epoch": 2.4386269459065963,
      "grad_norm": 1.1043665409088135,
      "learning_rate": 9.38078947368421e-05,
      "loss": 0.3901,
      "step": 43549
    },
    {
      "epoch": 2.4386829432187254,
      "grad_norm": 1.0802520513534546,
      "learning_rate": 9.380763157894738e-05,
      "loss": 0.2895,
      "step": 43550
    },
    {
      "epoch": 2.4387389405308544,
      "grad_norm": 1.249882459640503,
      "learning_rate": 9.380736842105264e-05,
      "loss": 0.4197,
      "step": 43551
    },
    {
      "epoch": 2.4387949378429834,
      "grad_norm": 0.9152452945709229,
      "learning_rate": 9.38071052631579e-05,
      "loss": 0.344,
      "step": 43552
    },
    {
      "epoch": 2.4388509351551124,
      "grad_norm": 1.2954599857330322,
      "learning_rate": 9.380684210526316e-05,
      "loss": 0.4432,
      "step": 43553
    },
    {
      "epoch": 2.4389069324672414,
      "grad_norm": 1.2975479364395142,
      "learning_rate": 9.380657894736843e-05,
      "loss": 0.388,
      "step": 43554
    },
    {
      "epoch": 2.4389629297793705,
      "grad_norm": 1.1526026725769043,
      "learning_rate": 9.380631578947369e-05,
      "loss": 0.2732,
      "step": 43555
    },
    {
      "epoch": 2.4390189270914995,
      "grad_norm": 1.2810592651367188,
      "learning_rate": 9.380605263157896e-05,
      "loss": 0.4176,
      "step": 43556
    },
    {
      "epoch": 2.4390749244036285,
      "grad_norm": 1.2511413097381592,
      "learning_rate": 9.380578947368421e-05,
      "loss": 0.5216,
      "step": 43557
    },
    {
      "epoch": 2.4391309217157575,
      "grad_norm": 1.2433569431304932,
      "learning_rate": 9.380552631578948e-05,
      "loss": 0.452,
      "step": 43558
    },
    {
      "epoch": 2.4391869190278865,
      "grad_norm": 1.1853517293930054,
      "learning_rate": 9.380526315789474e-05,
      "loss": 0.3632,
      "step": 43559
    },
    {
      "epoch": 2.4392429163400156,
      "grad_norm": 1.0360019207000732,
      "learning_rate": 9.380500000000001e-05,
      "loss": 0.3465,
      "step": 43560
    },
    {
      "epoch": 2.4392989136521446,
      "grad_norm": 0.971160352230072,
      "learning_rate": 9.380473684210527e-05,
      "loss": 0.274,
      "step": 43561
    },
    {
      "epoch": 2.4393549109642736,
      "grad_norm": 1.2742706537246704,
      "learning_rate": 9.380447368421052e-05,
      "loss": 0.4756,
      "step": 43562
    },
    {
      "epoch": 2.4394109082764026,
      "grad_norm": 1.0161525011062622,
      "learning_rate": 9.38042105263158e-05,
      "loss": 0.3122,
      "step": 43563
    },
    {
      "epoch": 2.4394669055885316,
      "grad_norm": 1.3556708097457886,
      "learning_rate": 9.380394736842105e-05,
      "loss": 0.4056,
      "step": 43564
    },
    {
      "epoch": 2.4395229029006607,
      "grad_norm": 1.2303104400634766,
      "learning_rate": 9.380368421052633e-05,
      "loss": 0.3931,
      "step": 43565
    },
    {
      "epoch": 2.4395789002127897,
      "grad_norm": 1.114974856376648,
      "learning_rate": 9.380342105263159e-05,
      "loss": 0.3985,
      "step": 43566
    },
    {
      "epoch": 2.4396348975249187,
      "grad_norm": 1.4099518060684204,
      "learning_rate": 9.380315789473685e-05,
      "loss": 0.4567,
      "step": 43567
    },
    {
      "epoch": 2.4396908948370477,
      "grad_norm": 1.172695279121399,
      "learning_rate": 9.38028947368421e-05,
      "loss": 0.4288,
      "step": 43568
    },
    {
      "epoch": 2.4397468921491767,
      "grad_norm": 1.2217788696289062,
      "learning_rate": 9.380263157894738e-05,
      "loss": 0.4117,
      "step": 43569
    },
    {
      "epoch": 2.4398028894613057,
      "grad_norm": 1.101059079170227,
      "learning_rate": 9.380236842105264e-05,
      "loss": 0.3787,
      "step": 43570
    },
    {
      "epoch": 2.4398588867734348,
      "grad_norm": 1.3017903566360474,
      "learning_rate": 9.38021052631579e-05,
      "loss": 0.3785,
      "step": 43571
    },
    {
      "epoch": 2.439914884085564,
      "grad_norm": 1.130303978919983,
      "learning_rate": 9.380184210526316e-05,
      "loss": 0.4085,
      "step": 43572
    },
    {
      "epoch": 2.439970881397693,
      "grad_norm": 1.1646997928619385,
      "learning_rate": 9.380157894736843e-05,
      "loss": 0.3446,
      "step": 43573
    },
    {
      "epoch": 2.440026878709822,
      "grad_norm": 1.3437261581420898,
      "learning_rate": 9.380131578947369e-05,
      "loss": 0.4407,
      "step": 43574
    },
    {
      "epoch": 2.440082876021951,
      "grad_norm": 1.3229933977127075,
      "learning_rate": 9.380105263157895e-05,
      "loss": 0.3774,
      "step": 43575
    },
    {
      "epoch": 2.44013887333408,
      "grad_norm": 1.1961010694503784,
      "learning_rate": 9.380078947368421e-05,
      "loss": 0.4461,
      "step": 43576
    },
    {
      "epoch": 2.440194870646209,
      "grad_norm": 1.038870930671692,
      "learning_rate": 9.380052631578948e-05,
      "loss": 0.4113,
      "step": 43577
    },
    {
      "epoch": 2.440250867958338,
      "grad_norm": 1.1936339139938354,
      "learning_rate": 9.380026315789474e-05,
      "loss": 0.3882,
      "step": 43578
    },
    {
      "epoch": 2.440306865270467,
      "grad_norm": 1.1477329730987549,
      "learning_rate": 9.38e-05,
      "loss": 0.3641,
      "step": 43579
    },
    {
      "epoch": 2.440362862582596,
      "grad_norm": 1.2114328145980835,
      "learning_rate": 9.379973684210526e-05,
      "loss": 0.4105,
      "step": 43580
    },
    {
      "epoch": 2.440418859894725,
      "grad_norm": 1.136233925819397,
      "learning_rate": 9.379947368421052e-05,
      "loss": 0.34,
      "step": 43581
    },
    {
      "epoch": 2.440474857206854,
      "grad_norm": 1.2263675928115845,
      "learning_rate": 9.37992105263158e-05,
      "loss": 0.3782,
      "step": 43582
    },
    {
      "epoch": 2.440530854518983,
      "grad_norm": 1.0855965614318848,
      "learning_rate": 9.379894736842106e-05,
      "loss": 0.3841,
      "step": 43583
    },
    {
      "epoch": 2.440586851831112,
      "grad_norm": 0.9905179142951965,
      "learning_rate": 9.379868421052632e-05,
      "loss": 0.3653,
      "step": 43584
    },
    {
      "epoch": 2.440642849143241,
      "grad_norm": 1.2677885293960571,
      "learning_rate": 9.379842105263158e-05,
      "loss": 0.4063,
      "step": 43585
    },
    {
      "epoch": 2.44069884645537,
      "grad_norm": 1.1108193397521973,
      "learning_rate": 9.379815789473685e-05,
      "loss": 0.3224,
      "step": 43586
    },
    {
      "epoch": 2.440754843767499,
      "grad_norm": 1.31383216381073,
      "learning_rate": 9.379789473684211e-05,
      "loss": 0.4106,
      "step": 43587
    },
    {
      "epoch": 2.440810841079628,
      "grad_norm": 1.2637027502059937,
      "learning_rate": 9.379763157894738e-05,
      "loss": 0.4832,
      "step": 43588
    },
    {
      "epoch": 2.440866838391757,
      "grad_norm": 1.2205462455749512,
      "learning_rate": 9.379736842105263e-05,
      "loss": 0.4223,
      "step": 43589
    },
    {
      "epoch": 2.440922835703886,
      "grad_norm": 1.1361775398254395,
      "learning_rate": 9.37971052631579e-05,
      "loss": 0.3303,
      "step": 43590
    },
    {
      "epoch": 2.440978833016015,
      "grad_norm": 0.9849117994308472,
      "learning_rate": 9.379684210526316e-05,
      "loss": 0.2706,
      "step": 43591
    },
    {
      "epoch": 2.441034830328144,
      "grad_norm": 1.071748971939087,
      "learning_rate": 9.379657894736843e-05,
      "loss": 0.6465,
      "step": 43592
    },
    {
      "epoch": 2.441090827640273,
      "grad_norm": 1.0876084566116333,
      "learning_rate": 9.37963157894737e-05,
      "loss": 0.3637,
      "step": 43593
    },
    {
      "epoch": 2.441146824952402,
      "grad_norm": 1.1235402822494507,
      "learning_rate": 9.379605263157895e-05,
      "loss": 0.3101,
      "step": 43594
    },
    {
      "epoch": 2.4412028222645312,
      "grad_norm": 1.5383445024490356,
      "learning_rate": 9.379578947368421e-05,
      "loss": 0.3591,
      "step": 43595
    },
    {
      "epoch": 2.4412588195766602,
      "grad_norm": 1.0634770393371582,
      "learning_rate": 9.379552631578949e-05,
      "loss": 0.5025,
      "step": 43596
    },
    {
      "epoch": 2.4413148168887893,
      "grad_norm": 0.9967201352119446,
      "learning_rate": 9.379526315789475e-05,
      "loss": 0.3714,
      "step": 43597
    },
    {
      "epoch": 2.4413708142009183,
      "grad_norm": 1.3237299919128418,
      "learning_rate": 9.3795e-05,
      "loss": 0.541,
      "step": 43598
    },
    {
      "epoch": 2.4414268115130473,
      "grad_norm": 1.0238609313964844,
      "learning_rate": 9.379473684210527e-05,
      "loss": 0.3675,
      "step": 43599
    },
    {
      "epoch": 2.4414828088251763,
      "grad_norm": 1.0351039171218872,
      "learning_rate": 9.379447368421053e-05,
      "loss": 0.3415,
      "step": 43600
    },
    {
      "epoch": 2.4415388061373053,
      "grad_norm": 1.2138872146606445,
      "learning_rate": 9.37942105263158e-05,
      "loss": 0.4554,
      "step": 43601
    },
    {
      "epoch": 2.4415948034494344,
      "grad_norm": 1.0285530090332031,
      "learning_rate": 9.379394736842106e-05,
      "loss": 0.3276,
      "step": 43602
    },
    {
      "epoch": 2.4416508007615634,
      "grad_norm": 1.209574580192566,
      "learning_rate": 9.379368421052632e-05,
      "loss": 0.3885,
      "step": 43603
    },
    {
      "epoch": 2.4417067980736924,
      "grad_norm": 1.61426842212677,
      "learning_rate": 9.379342105263158e-05,
      "loss": 0.4751,
      "step": 43604
    },
    {
      "epoch": 2.4417627953858214,
      "grad_norm": 1.2000374794006348,
      "learning_rate": 9.379315789473685e-05,
      "loss": 0.3743,
      "step": 43605
    },
    {
      "epoch": 2.4418187926979504,
      "grad_norm": 2.280125141143799,
      "learning_rate": 9.379289473684211e-05,
      "loss": 0.4071,
      "step": 43606
    },
    {
      "epoch": 2.4418747900100795,
      "grad_norm": 1.0987000465393066,
      "learning_rate": 9.379263157894737e-05,
      "loss": 0.4392,
      "step": 43607
    },
    {
      "epoch": 2.4419307873222085,
      "grad_norm": 1.9124033451080322,
      "learning_rate": 9.379236842105263e-05,
      "loss": 0.4704,
      "step": 43608
    },
    {
      "epoch": 2.4419867846343375,
      "grad_norm": 1.328802466392517,
      "learning_rate": 9.37921052631579e-05,
      "loss": 0.4052,
      "step": 43609
    },
    {
      "epoch": 2.4420427819464665,
      "grad_norm": 1.102630376815796,
      "learning_rate": 9.379184210526316e-05,
      "loss": 0.3419,
      "step": 43610
    },
    {
      "epoch": 2.4420987792585955,
      "grad_norm": 1.2645013332366943,
      "learning_rate": 9.379157894736844e-05,
      "loss": 0.372,
      "step": 43611
    },
    {
      "epoch": 2.4421547765707246,
      "grad_norm": 1.1576825380325317,
      "learning_rate": 9.379131578947368e-05,
      "loss": 0.5108,
      "step": 43612
    },
    {
      "epoch": 2.4422107738828536,
      "grad_norm": 21.2763671875,
      "learning_rate": 9.379105263157896e-05,
      "loss": 0.3379,
      "step": 43613
    },
    {
      "epoch": 2.4422667711949826,
      "grad_norm": 1.2121535539627075,
      "learning_rate": 9.379078947368422e-05,
      "loss": 0.4324,
      "step": 43614
    },
    {
      "epoch": 2.4423227685071116,
      "grad_norm": 1.1602247953414917,
      "learning_rate": 9.379052631578948e-05,
      "loss": 0.3681,
      "step": 43615
    },
    {
      "epoch": 2.4423787658192406,
      "grad_norm": 1.2490981817245483,
      "learning_rate": 9.379026315789475e-05,
      "loss": 0.4606,
      "step": 43616
    },
    {
      "epoch": 2.4424347631313696,
      "grad_norm": 1.1423240900039673,
      "learning_rate": 9.379e-05,
      "loss": 0.3683,
      "step": 43617
    },
    {
      "epoch": 2.4424907604434987,
      "grad_norm": 1.0834897756576538,
      "learning_rate": 9.378973684210527e-05,
      "loss": 0.4426,
      "step": 43618
    },
    {
      "epoch": 2.4425467577556277,
      "grad_norm": 1.1412252187728882,
      "learning_rate": 9.378947368421053e-05,
      "loss": 0.5135,
      "step": 43619
    },
    {
      "epoch": 2.4426027550677567,
      "grad_norm": 1.2741209268569946,
      "learning_rate": 9.37892105263158e-05,
      "loss": 0.3978,
      "step": 43620
    },
    {
      "epoch": 2.4426587523798857,
      "grad_norm": 1.284626841545105,
      "learning_rate": 9.378894736842105e-05,
      "loss": 0.3712,
      "step": 43621
    },
    {
      "epoch": 2.4427147496920147,
      "grad_norm": 1.2163032293319702,
      "learning_rate": 9.378868421052632e-05,
      "loss": 0.3384,
      "step": 43622
    },
    {
      "epoch": 2.4427707470041438,
      "grad_norm": 1.0085527896881104,
      "learning_rate": 9.378842105263158e-05,
      "loss": 0.2756,
      "step": 43623
    },
    {
      "epoch": 2.442826744316273,
      "grad_norm": 1.2884711027145386,
      "learning_rate": 9.378815789473685e-05,
      "loss": 0.4086,
      "step": 43624
    },
    {
      "epoch": 2.442882741628402,
      "grad_norm": 1.1734811067581177,
      "learning_rate": 9.378789473684211e-05,
      "loss": 0.3701,
      "step": 43625
    },
    {
      "epoch": 2.442938738940531,
      "grad_norm": 1.1677095890045166,
      "learning_rate": 9.378763157894737e-05,
      "loss": 0.5154,
      "step": 43626
    },
    {
      "epoch": 2.44299473625266,
      "grad_norm": 1.0862445831298828,
      "learning_rate": 9.378736842105263e-05,
      "loss": 0.4135,
      "step": 43627
    },
    {
      "epoch": 2.443050733564789,
      "grad_norm": 1.1216758489608765,
      "learning_rate": 9.37871052631579e-05,
      "loss": 0.4393,
      "step": 43628
    },
    {
      "epoch": 2.443106730876918,
      "grad_norm": 0.9754014015197754,
      "learning_rate": 9.378684210526317e-05,
      "loss": 0.4312,
      "step": 43629
    },
    {
      "epoch": 2.443162728189047,
      "grad_norm": 1.6072847843170166,
      "learning_rate": 9.378657894736843e-05,
      "loss": 0.3918,
      "step": 43630
    },
    {
      "epoch": 2.443218725501176,
      "grad_norm": 1.1401312351226807,
      "learning_rate": 9.378631578947369e-05,
      "loss": 0.3287,
      "step": 43631
    },
    {
      "epoch": 2.443274722813305,
      "grad_norm": 1.3051751852035522,
      "learning_rate": 9.378605263157894e-05,
      "loss": 0.5795,
      "step": 43632
    },
    {
      "epoch": 2.443330720125434,
      "grad_norm": 1.1379828453063965,
      "learning_rate": 9.378578947368422e-05,
      "loss": 0.4222,
      "step": 43633
    },
    {
      "epoch": 2.443386717437563,
      "grad_norm": 1.2749556303024292,
      "learning_rate": 9.378552631578948e-05,
      "loss": 0.4336,
      "step": 43634
    },
    {
      "epoch": 2.443442714749692,
      "grad_norm": 1.154718279838562,
      "learning_rate": 9.378526315789474e-05,
      "loss": 0.3251,
      "step": 43635
    },
    {
      "epoch": 2.443498712061821,
      "grad_norm": 1.1960395574569702,
      "learning_rate": 9.3785e-05,
      "loss": 0.6221,
      "step": 43636
    },
    {
      "epoch": 2.44355470937395,
      "grad_norm": 1.3199608325958252,
      "learning_rate": 9.378473684210527e-05,
      "loss": 0.4653,
      "step": 43637
    },
    {
      "epoch": 2.443610706686079,
      "grad_norm": 1.226493000984192,
      "learning_rate": 9.378447368421053e-05,
      "loss": 0.4635,
      "step": 43638
    },
    {
      "epoch": 2.443666703998208,
      "grad_norm": 1.4030107259750366,
      "learning_rate": 9.378421052631579e-05,
      "loss": 0.4308,
      "step": 43639
    },
    {
      "epoch": 2.443722701310337,
      "grad_norm": 1.2558891773223877,
      "learning_rate": 9.378394736842105e-05,
      "loss": 0.4089,
      "step": 43640
    },
    {
      "epoch": 2.443778698622466,
      "grad_norm": 1.0963397026062012,
      "learning_rate": 9.378368421052632e-05,
      "loss": 0.4336,
      "step": 43641
    },
    {
      "epoch": 2.443834695934595,
      "grad_norm": 1.1146467924118042,
      "learning_rate": 9.378342105263158e-05,
      "loss": 0.3727,
      "step": 43642
    },
    {
      "epoch": 2.443890693246724,
      "grad_norm": 1.049160361289978,
      "learning_rate": 9.378315789473686e-05,
      "loss": 0.3913,
      "step": 43643
    },
    {
      "epoch": 2.443946690558853,
      "grad_norm": 1.0531044006347656,
      "learning_rate": 9.37828947368421e-05,
      "loss": 0.3898,
      "step": 43644
    },
    {
      "epoch": 2.444002687870982,
      "grad_norm": 1.0650125741958618,
      "learning_rate": 9.378263157894738e-05,
      "loss": 0.4223,
      "step": 43645
    },
    {
      "epoch": 2.444058685183111,
      "grad_norm": 1.0833420753479004,
      "learning_rate": 9.378236842105264e-05,
      "loss": 0.3063,
      "step": 43646
    },
    {
      "epoch": 2.44411468249524,
      "grad_norm": 1.0019596815109253,
      "learning_rate": 9.378210526315791e-05,
      "loss": 0.3058,
      "step": 43647
    },
    {
      "epoch": 2.4441706798073692,
      "grad_norm": 1.0865423679351807,
      "learning_rate": 9.378184210526317e-05,
      "loss": 0.3333,
      "step": 43648
    },
    {
      "epoch": 2.4442266771194983,
      "grad_norm": 1.5698976516723633,
      "learning_rate": 9.378157894736841e-05,
      "loss": 0.6244,
      "step": 43649
    },
    {
      "epoch": 2.4442826744316273,
      "grad_norm": 1.148758053779602,
      "learning_rate": 9.378131578947369e-05,
      "loss": 0.4434,
      "step": 43650
    },
    {
      "epoch": 2.4443386717437563,
      "grad_norm": 1.1846868991851807,
      "learning_rate": 9.378105263157895e-05,
      "loss": 0.3566,
      "step": 43651
    },
    {
      "epoch": 2.4443946690558853,
      "grad_norm": 1.0910898447036743,
      "learning_rate": 9.378078947368422e-05,
      "loss": 0.3292,
      "step": 43652
    },
    {
      "epoch": 2.4444506663680143,
      "grad_norm": 1.2427153587341309,
      "learning_rate": 9.378052631578948e-05,
      "loss": 0.3736,
      "step": 43653
    },
    {
      "epoch": 2.4445066636801434,
      "grad_norm": 1.362568974494934,
      "learning_rate": 9.378026315789474e-05,
      "loss": 0.3326,
      "step": 43654
    },
    {
      "epoch": 2.4445626609922724,
      "grad_norm": 1.3020566701889038,
      "learning_rate": 9.378e-05,
      "loss": 0.5816,
      "step": 43655
    },
    {
      "epoch": 2.4446186583044014,
      "grad_norm": 1.0465363264083862,
      "learning_rate": 9.377973684210527e-05,
      "loss": 0.36,
      "step": 43656
    },
    {
      "epoch": 2.4446746556165304,
      "grad_norm": 1.1095346212387085,
      "learning_rate": 9.377947368421053e-05,
      "loss": 0.3467,
      "step": 43657
    },
    {
      "epoch": 2.4447306529286594,
      "grad_norm": 1.3338831663131714,
      "learning_rate": 9.377921052631579e-05,
      "loss": 0.5412,
      "step": 43658
    },
    {
      "epoch": 2.4447866502407885,
      "grad_norm": 1.0864962339401245,
      "learning_rate": 9.377894736842105e-05,
      "loss": 0.3924,
      "step": 43659
    },
    {
      "epoch": 2.4448426475529175,
      "grad_norm": 1.048994541168213,
      "learning_rate": 9.377868421052633e-05,
      "loss": 0.392,
      "step": 43660
    },
    {
      "epoch": 2.4448986448650465,
      "grad_norm": 1.2112869024276733,
      "learning_rate": 9.377842105263159e-05,
      "loss": 0.4144,
      "step": 43661
    },
    {
      "epoch": 2.4449546421771755,
      "grad_norm": 0.9742178320884705,
      "learning_rate": 9.377815789473685e-05,
      "loss": 0.3506,
      "step": 43662
    },
    {
      "epoch": 2.4450106394893045,
      "grad_norm": 1.2229830026626587,
      "learning_rate": 9.37778947368421e-05,
      "loss": 0.3812,
      "step": 43663
    },
    {
      "epoch": 2.4450666368014335,
      "grad_norm": 1.1158320903778076,
      "learning_rate": 9.377763157894738e-05,
      "loss": 0.3382,
      "step": 43664
    },
    {
      "epoch": 2.4451226341135626,
      "grad_norm": 1.1286373138427734,
      "learning_rate": 9.377736842105264e-05,
      "loss": 0.341,
      "step": 43665
    },
    {
      "epoch": 2.4451786314256916,
      "grad_norm": 1.093476414680481,
      "learning_rate": 9.37771052631579e-05,
      "loss": 0.2808,
      "step": 43666
    },
    {
      "epoch": 2.4452346287378206,
      "grad_norm": 1.164022445678711,
      "learning_rate": 9.377684210526316e-05,
      "loss": 0.4055,
      "step": 43667
    },
    {
      "epoch": 2.4452906260499496,
      "grad_norm": 1.1908897161483765,
      "learning_rate": 9.377657894736842e-05,
      "loss": 0.4668,
      "step": 43668
    },
    {
      "epoch": 2.4453466233620786,
      "grad_norm": 1.337965965270996,
      "learning_rate": 9.377631578947369e-05,
      "loss": 0.4346,
      "step": 43669
    },
    {
      "epoch": 2.4454026206742077,
      "grad_norm": 1.3720922470092773,
      "learning_rate": 9.377605263157895e-05,
      "loss": 0.4393,
      "step": 43670
    },
    {
      "epoch": 2.4454586179863367,
      "grad_norm": 1.5693382024765015,
      "learning_rate": 9.377578947368422e-05,
      "loss": 0.3939,
      "step": 43671
    },
    {
      "epoch": 2.4455146152984657,
      "grad_norm": 1.0617972612380981,
      "learning_rate": 9.377552631578947e-05,
      "loss": 0.3239,
      "step": 43672
    },
    {
      "epoch": 2.4455706126105947,
      "grad_norm": 1.1115037202835083,
      "learning_rate": 9.377526315789474e-05,
      "loss": 0.3882,
      "step": 43673
    },
    {
      "epoch": 2.4456266099227237,
      "grad_norm": 0.9725748300552368,
      "learning_rate": 9.3775e-05,
      "loss": 0.3312,
      "step": 43674
    },
    {
      "epoch": 2.4456826072348528,
      "grad_norm": 1.3639880418777466,
      "learning_rate": 9.377473684210528e-05,
      "loss": 0.5066,
      "step": 43675
    },
    {
      "epoch": 2.4457386045469818,
      "grad_norm": 1.265601396560669,
      "learning_rate": 9.377447368421052e-05,
      "loss": 0.4099,
      "step": 43676
    },
    {
      "epoch": 2.445794601859111,
      "grad_norm": 1.0266481637954712,
      "learning_rate": 9.37742105263158e-05,
      "loss": 0.4265,
      "step": 43677
    },
    {
      "epoch": 2.44585059917124,
      "grad_norm": 1.1295057535171509,
      "learning_rate": 9.377394736842106e-05,
      "loss": 0.4222,
      "step": 43678
    },
    {
      "epoch": 2.445906596483369,
      "grad_norm": 0.9731562733650208,
      "learning_rate": 9.377368421052633e-05,
      "loss": 0.2532,
      "step": 43679
    },
    {
      "epoch": 2.445962593795498,
      "grad_norm": 1.1277998685836792,
      "learning_rate": 9.377342105263159e-05,
      "loss": 0.4147,
      "step": 43680
    },
    {
      "epoch": 2.446018591107627,
      "grad_norm": 1.2166575193405151,
      "learning_rate": 9.377315789473685e-05,
      "loss": 0.4434,
      "step": 43681
    },
    {
      "epoch": 2.446074588419756,
      "grad_norm": 1.089971899986267,
      "learning_rate": 9.377289473684211e-05,
      "loss": 0.3621,
      "step": 43682
    },
    {
      "epoch": 2.446130585731885,
      "grad_norm": 1.1873785257339478,
      "learning_rate": 9.377263157894737e-05,
      "loss": 0.335,
      "step": 43683
    },
    {
      "epoch": 2.446186583044014,
      "grad_norm": 1.5514419078826904,
      "learning_rate": 9.377236842105264e-05,
      "loss": 0.5034,
      "step": 43684
    },
    {
      "epoch": 2.446242580356143,
      "grad_norm": 1.0871385335922241,
      "learning_rate": 9.37721052631579e-05,
      "loss": 0.3635,
      "step": 43685
    },
    {
      "epoch": 2.446298577668272,
      "grad_norm": 3.2459747791290283,
      "learning_rate": 9.377184210526316e-05,
      "loss": 0.4355,
      "step": 43686
    },
    {
      "epoch": 2.446354574980401,
      "grad_norm": 1.1414971351623535,
      "learning_rate": 9.377157894736842e-05,
      "loss": 0.3802,
      "step": 43687
    },
    {
      "epoch": 2.44641057229253,
      "grad_norm": 1.1510952711105347,
      "learning_rate": 9.377131578947369e-05,
      "loss": 0.3774,
      "step": 43688
    },
    {
      "epoch": 2.446466569604659,
      "grad_norm": 1.1318522691726685,
      "learning_rate": 9.377105263157895e-05,
      "loss": 0.2816,
      "step": 43689
    },
    {
      "epoch": 2.446522566916788,
      "grad_norm": 1.124312162399292,
      "learning_rate": 9.377078947368421e-05,
      "loss": 0.3201,
      "step": 43690
    },
    {
      "epoch": 2.446578564228917,
      "grad_norm": 2.025885581970215,
      "learning_rate": 9.377052631578947e-05,
      "loss": 0.561,
      "step": 43691
    },
    {
      "epoch": 2.446634561541046,
      "grad_norm": 1.2253518104553223,
      "learning_rate": 9.377026315789475e-05,
      "loss": 0.3734,
      "step": 43692
    },
    {
      "epoch": 2.446690558853175,
      "grad_norm": 1.26633882522583,
      "learning_rate": 9.377e-05,
      "loss": 0.4172,
      "step": 43693
    },
    {
      "epoch": 2.446746556165304,
      "grad_norm": 1.433893084526062,
      "learning_rate": 9.376973684210526e-05,
      "loss": 0.4917,
      "step": 43694
    },
    {
      "epoch": 2.446802553477433,
      "grad_norm": 1.0203039646148682,
      "learning_rate": 9.376947368421052e-05,
      "loss": 0.3831,
      "step": 43695
    },
    {
      "epoch": 2.446858550789562,
      "grad_norm": 0.9234296679496765,
      "learning_rate": 9.37692105263158e-05,
      "loss": 0.3233,
      "step": 43696
    },
    {
      "epoch": 2.446914548101691,
      "grad_norm": 1.1691572666168213,
      "learning_rate": 9.376894736842106e-05,
      "loss": 0.5234,
      "step": 43697
    },
    {
      "epoch": 2.44697054541382,
      "grad_norm": 1.1446528434753418,
      "learning_rate": 9.376868421052633e-05,
      "loss": 0.3824,
      "step": 43698
    },
    {
      "epoch": 2.447026542725949,
      "grad_norm": 1.535370945930481,
      "learning_rate": 9.376842105263158e-05,
      "loss": 0.4807,
      "step": 43699
    },
    {
      "epoch": 2.4470825400380782,
      "grad_norm": 1.0322017669677734,
      "learning_rate": 9.376815789473684e-05,
      "loss": 0.3254,
      "step": 43700
    },
    {
      "epoch": 2.4471385373502073,
      "grad_norm": 1.879762887954712,
      "learning_rate": 9.376789473684211e-05,
      "loss": 0.384,
      "step": 43701
    },
    {
      "epoch": 2.4471945346623363,
      "grad_norm": 1.1668339967727661,
      "learning_rate": 9.376763157894737e-05,
      "loss": 0.3251,
      "step": 43702
    },
    {
      "epoch": 2.4472505319744653,
      "grad_norm": 1.5099295377731323,
      "learning_rate": 9.376736842105264e-05,
      "loss": 0.4476,
      "step": 43703
    },
    {
      "epoch": 2.4473065292865943,
      "grad_norm": 44.37250900268555,
      "learning_rate": 9.376710526315789e-05,
      "loss": 0.3932,
      "step": 43704
    },
    {
      "epoch": 2.4473625265987233,
      "grad_norm": 1.140372633934021,
      "learning_rate": 9.376684210526316e-05,
      "loss": 0.368,
      "step": 43705
    },
    {
      "epoch": 2.4474185239108524,
      "grad_norm": 1.0529611110687256,
      "learning_rate": 9.376657894736842e-05,
      "loss": 0.3392,
      "step": 43706
    },
    {
      "epoch": 2.4474745212229814,
      "grad_norm": 1.2084894180297852,
      "learning_rate": 9.37663157894737e-05,
      "loss": 0.35,
      "step": 43707
    },
    {
      "epoch": 2.4475305185351104,
      "grad_norm": 1.6079727411270142,
      "learning_rate": 9.376605263157896e-05,
      "loss": 0.5559,
      "step": 43708
    },
    {
      "epoch": 2.4475865158472394,
      "grad_norm": 1.3256789445877075,
      "learning_rate": 9.376578947368422e-05,
      "loss": 0.3361,
      "step": 43709
    },
    {
      "epoch": 2.4476425131593684,
      "grad_norm": 1.705016016960144,
      "learning_rate": 9.376552631578947e-05,
      "loss": 0.4373,
      "step": 43710
    },
    {
      "epoch": 2.4476985104714974,
      "grad_norm": 1.2192429304122925,
      "learning_rate": 9.376526315789475e-05,
      "loss": 0.5857,
      "step": 43711
    },
    {
      "epoch": 2.4477545077836265,
      "grad_norm": 1.0645999908447266,
      "learning_rate": 9.376500000000001e-05,
      "loss": 0.3985,
      "step": 43712
    },
    {
      "epoch": 2.4478105050957555,
      "grad_norm": 1.1529145240783691,
      "learning_rate": 9.376473684210527e-05,
      "loss": 0.3655,
      "step": 43713
    },
    {
      "epoch": 2.4478665024078845,
      "grad_norm": 1.1662991046905518,
      "learning_rate": 9.376447368421053e-05,
      "loss": 0.3829,
      "step": 43714
    },
    {
      "epoch": 2.4479224997200135,
      "grad_norm": 0.9849196672439575,
      "learning_rate": 9.37642105263158e-05,
      "loss": 0.2998,
      "step": 43715
    },
    {
      "epoch": 2.4479784970321425,
      "grad_norm": 1.6906744241714478,
      "learning_rate": 9.376394736842106e-05,
      "loss": 0.4748,
      "step": 43716
    },
    {
      "epoch": 2.4480344943442716,
      "grad_norm": 1.300154209136963,
      "learning_rate": 9.376368421052632e-05,
      "loss": 0.4192,
      "step": 43717
    },
    {
      "epoch": 2.4480904916564006,
      "grad_norm": 1.0675753355026245,
      "learning_rate": 9.376342105263158e-05,
      "loss": 0.3314,
      "step": 43718
    },
    {
      "epoch": 2.4481464889685296,
      "grad_norm": 1.8462144136428833,
      "learning_rate": 9.376315789473684e-05,
      "loss": 0.4346,
      "step": 43719
    },
    {
      "epoch": 2.4482024862806586,
      "grad_norm": 1.3585484027862549,
      "learning_rate": 9.376289473684211e-05,
      "loss": 0.4246,
      "step": 43720
    },
    {
      "epoch": 2.4482584835927876,
      "grad_norm": 1.0886788368225098,
      "learning_rate": 9.376263157894737e-05,
      "loss": 0.3985,
      "step": 43721
    },
    {
      "epoch": 2.4483144809049167,
      "grad_norm": 1.0004119873046875,
      "learning_rate": 9.376236842105263e-05,
      "loss": 0.2763,
      "step": 43722
    },
    {
      "epoch": 2.4483704782170457,
      "grad_norm": 1.22940993309021,
      "learning_rate": 9.376210526315789e-05,
      "loss": 0.4011,
      "step": 43723
    },
    {
      "epoch": 2.4484264755291747,
      "grad_norm": 1.137768030166626,
      "learning_rate": 9.376184210526317e-05,
      "loss": 0.3554,
      "step": 43724
    },
    {
      "epoch": 2.4484824728413037,
      "grad_norm": 1.2076971530914307,
      "learning_rate": 9.376157894736842e-05,
      "loss": 0.4111,
      "step": 43725
    },
    {
      "epoch": 2.4485384701534327,
      "grad_norm": 1.3236175775527954,
      "learning_rate": 9.37613157894737e-05,
      "loss": 0.4733,
      "step": 43726
    },
    {
      "epoch": 2.4485944674655618,
      "grad_norm": 1.5193358659744263,
      "learning_rate": 9.376105263157894e-05,
      "loss": 0.4315,
      "step": 43727
    },
    {
      "epoch": 2.4486504647776908,
      "grad_norm": 1.0656602382659912,
      "learning_rate": 9.376078947368422e-05,
      "loss": 0.3344,
      "step": 43728
    },
    {
      "epoch": 2.44870646208982,
      "grad_norm": 1.3186131715774536,
      "learning_rate": 9.376052631578948e-05,
      "loss": 0.4576,
      "step": 43729
    },
    {
      "epoch": 2.448762459401949,
      "grad_norm": 1.5465681552886963,
      "learning_rate": 9.376026315789475e-05,
      "loss": 0.4415,
      "step": 43730
    },
    {
      "epoch": 2.448818456714078,
      "grad_norm": 1.1777374744415283,
      "learning_rate": 9.376e-05,
      "loss": 0.3712,
      "step": 43731
    },
    {
      "epoch": 2.448874454026207,
      "grad_norm": 1.0133075714111328,
      "learning_rate": 9.375973684210527e-05,
      "loss": 0.3226,
      "step": 43732
    },
    {
      "epoch": 2.448930451338336,
      "grad_norm": 1.2057561874389648,
      "learning_rate": 9.375947368421053e-05,
      "loss": 0.335,
      "step": 43733
    },
    {
      "epoch": 2.448986448650465,
      "grad_norm": 1.2212294340133667,
      "learning_rate": 9.37592105263158e-05,
      "loss": 0.3468,
      "step": 43734
    },
    {
      "epoch": 2.449042445962594,
      "grad_norm": 1.236613392829895,
      "learning_rate": 9.375894736842106e-05,
      "loss": 0.5029,
      "step": 43735
    },
    {
      "epoch": 2.449098443274723,
      "grad_norm": 1.046769142150879,
      "learning_rate": 9.375868421052631e-05,
      "loss": 0.4269,
      "step": 43736
    },
    {
      "epoch": 2.449154440586852,
      "grad_norm": 1.3350902795791626,
      "learning_rate": 9.375842105263158e-05,
      "loss": 0.5204,
      "step": 43737
    },
    {
      "epoch": 2.449210437898981,
      "grad_norm": 1.2115131616592407,
      "learning_rate": 9.375815789473684e-05,
      "loss": 0.456,
      "step": 43738
    },
    {
      "epoch": 2.44926643521111,
      "grad_norm": 1.0237935781478882,
      "learning_rate": 9.375789473684212e-05,
      "loss": 0.3462,
      "step": 43739
    },
    {
      "epoch": 2.449322432523239,
      "grad_norm": 1.6484836339950562,
      "learning_rate": 9.375763157894737e-05,
      "loss": 0.4813,
      "step": 43740
    },
    {
      "epoch": 2.449378429835368,
      "grad_norm": 2.320385694503784,
      "learning_rate": 9.375736842105263e-05,
      "loss": 0.4933,
      "step": 43741
    },
    {
      "epoch": 2.449434427147497,
      "grad_norm": 1.1416598558425903,
      "learning_rate": 9.37571052631579e-05,
      "loss": 0.3579,
      "step": 43742
    },
    {
      "epoch": 2.449490424459626,
      "grad_norm": 1.3637193441390991,
      "learning_rate": 9.375684210526317e-05,
      "loss": 0.4303,
      "step": 43743
    },
    {
      "epoch": 2.449546421771755,
      "grad_norm": 1.2515645027160645,
      "learning_rate": 9.375657894736843e-05,
      "loss": 0.3739,
      "step": 43744
    },
    {
      "epoch": 2.449602419083884,
      "grad_norm": 1.1284507513046265,
      "learning_rate": 9.375631578947369e-05,
      "loss": 0.3438,
      "step": 43745
    },
    {
      "epoch": 2.449658416396013,
      "grad_norm": 1.0731176137924194,
      "learning_rate": 9.375605263157895e-05,
      "loss": 0.3492,
      "step": 43746
    },
    {
      "epoch": 2.449714413708142,
      "grad_norm": 1.2167754173278809,
      "learning_rate": 9.375578947368422e-05,
      "loss": 0.5333,
      "step": 43747
    },
    {
      "epoch": 2.449770411020271,
      "grad_norm": 1.2244106531143188,
      "learning_rate": 9.375552631578948e-05,
      "loss": 0.3766,
      "step": 43748
    },
    {
      "epoch": 2.4498264083324,
      "grad_norm": 1.012938380241394,
      "learning_rate": 9.375526315789474e-05,
      "loss": 0.381,
      "step": 43749
    },
    {
      "epoch": 2.449882405644529,
      "grad_norm": 1.1563268899917603,
      "learning_rate": 9.3755e-05,
      "loss": 0.3678,
      "step": 43750
    },
    {
      "epoch": 2.449938402956658,
      "grad_norm": 1.0605182647705078,
      "learning_rate": 9.375473684210527e-05,
      "loss": 0.3274,
      "step": 43751
    },
    {
      "epoch": 2.4499944002687872,
      "grad_norm": 2.353602886199951,
      "learning_rate": 9.375447368421053e-05,
      "loss": 0.3984,
      "step": 43752
    },
    {
      "epoch": 2.4500503975809163,
      "grad_norm": 1.3002004623413086,
      "learning_rate": 9.375421052631579e-05,
      "loss": 0.573,
      "step": 43753
    },
    {
      "epoch": 2.4501063948930453,
      "grad_norm": 1.1317799091339111,
      "learning_rate": 9.375394736842105e-05,
      "loss": 0.3771,
      "step": 43754
    },
    {
      "epoch": 2.4501623922051743,
      "grad_norm": 1.151565670967102,
      "learning_rate": 9.375368421052631e-05,
      "loss": 0.3281,
      "step": 43755
    },
    {
      "epoch": 2.4502183895173033,
      "grad_norm": 1.2632449865341187,
      "learning_rate": 9.375342105263158e-05,
      "loss": 0.3834,
      "step": 43756
    },
    {
      "epoch": 2.4502743868294323,
      "grad_norm": 1.2566313743591309,
      "learning_rate": 9.375315789473684e-05,
      "loss": 0.4457,
      "step": 43757
    },
    {
      "epoch": 2.4503303841415613,
      "grad_norm": 1.334071397781372,
      "learning_rate": 9.375289473684212e-05,
      "loss": 0.4616,
      "step": 43758
    },
    {
      "epoch": 2.4503863814536904,
      "grad_norm": 1.0823010206222534,
      "learning_rate": 9.375263157894736e-05,
      "loss": 0.3203,
      "step": 43759
    },
    {
      "epoch": 2.4504423787658194,
      "grad_norm": 1.4388163089752197,
      "learning_rate": 9.375236842105264e-05,
      "loss": 0.4122,
      "step": 43760
    },
    {
      "epoch": 2.4504983760779484,
      "grad_norm": 1.2031493186950684,
      "learning_rate": 9.37521052631579e-05,
      "loss": 0.3863,
      "step": 43761
    },
    {
      "epoch": 2.4505543733900774,
      "grad_norm": 1.0397971868515015,
      "learning_rate": 9.375184210526317e-05,
      "loss": 0.3096,
      "step": 43762
    },
    {
      "epoch": 2.4506103707022064,
      "grad_norm": 1.2586244344711304,
      "learning_rate": 9.375157894736843e-05,
      "loss": 0.4188,
      "step": 43763
    },
    {
      "epoch": 2.4506663680143355,
      "grad_norm": 9.512496948242188,
      "learning_rate": 9.375131578947369e-05,
      "loss": 0.354,
      "step": 43764
    },
    {
      "epoch": 2.4507223653264645,
      "grad_norm": 1.300660490989685,
      "learning_rate": 9.375105263157895e-05,
      "loss": 0.353,
      "step": 43765
    },
    {
      "epoch": 2.4507783626385935,
      "grad_norm": 1.4918419122695923,
      "learning_rate": 9.375078947368422e-05,
      "loss": 0.495,
      "step": 43766
    },
    {
      "epoch": 2.4508343599507225,
      "grad_norm": 1.2638053894042969,
      "learning_rate": 9.375052631578948e-05,
      "loss": 0.3409,
      "step": 43767
    },
    {
      "epoch": 2.4508903572628515,
      "grad_norm": 1.2711232900619507,
      "learning_rate": 9.375026315789474e-05,
      "loss": 0.4545,
      "step": 43768
    },
    {
      "epoch": 2.4509463545749806,
      "grad_norm": 1.112298846244812,
      "learning_rate": 9.375e-05,
      "loss": 0.38,
      "step": 43769
    },
    {
      "epoch": 2.4510023518871096,
      "grad_norm": 1.0446643829345703,
      "learning_rate": 9.374973684210526e-05,
      "loss": 0.3301,
      "step": 43770
    },
    {
      "epoch": 2.4510583491992386,
      "grad_norm": 1.1426150798797607,
      "learning_rate": 9.374947368421053e-05,
      "loss": 0.3669,
      "step": 43771
    },
    {
      "epoch": 2.4511143465113676,
      "grad_norm": 1.2353901863098145,
      "learning_rate": 9.37492105263158e-05,
      "loss": 0.3856,
      "step": 43772
    },
    {
      "epoch": 2.4511703438234966,
      "grad_norm": 1.0847564935684204,
      "learning_rate": 9.374894736842105e-05,
      "loss": 0.4021,
      "step": 43773
    },
    {
      "epoch": 2.4512263411356257,
      "grad_norm": 1.307093620300293,
      "learning_rate": 9.374868421052631e-05,
      "loss": 0.5238,
      "step": 43774
    },
    {
      "epoch": 2.4512823384477547,
      "grad_norm": 1.017579436302185,
      "learning_rate": 9.374842105263159e-05,
      "loss": 0.3229,
      "step": 43775
    },
    {
      "epoch": 2.4513383357598837,
      "grad_norm": 1.171047329902649,
      "learning_rate": 9.374815789473685e-05,
      "loss": 0.4943,
      "step": 43776
    },
    {
      "epoch": 2.4513943330720127,
      "grad_norm": 1.1344552040100098,
      "learning_rate": 9.374789473684211e-05,
      "loss": 0.4714,
      "step": 43777
    },
    {
      "epoch": 2.4514503303841417,
      "grad_norm": 1.21371328830719,
      "learning_rate": 9.374763157894737e-05,
      "loss": 0.3875,
      "step": 43778
    },
    {
      "epoch": 2.4515063276962707,
      "grad_norm": 1.0660808086395264,
      "learning_rate": 9.374736842105264e-05,
      "loss": 0.4513,
      "step": 43779
    },
    {
      "epoch": 2.4515623250083998,
      "grad_norm": 1.0837348699569702,
      "learning_rate": 9.37471052631579e-05,
      "loss": 0.3179,
      "step": 43780
    },
    {
      "epoch": 2.451618322320529,
      "grad_norm": 1.1555426120758057,
      "learning_rate": 9.374684210526317e-05,
      "loss": 0.393,
      "step": 43781
    },
    {
      "epoch": 2.451674319632658,
      "grad_norm": 1.4543371200561523,
      "learning_rate": 9.374657894736842e-05,
      "loss": 0.4406,
      "step": 43782
    },
    {
      "epoch": 2.451730316944787,
      "grad_norm": 1.4915732145309448,
      "learning_rate": 9.374631578947369e-05,
      "loss": 0.477,
      "step": 43783
    },
    {
      "epoch": 2.451786314256916,
      "grad_norm": 0.9504106044769287,
      "learning_rate": 9.374605263157895e-05,
      "loss": 0.3675,
      "step": 43784
    },
    {
      "epoch": 2.451842311569045,
      "grad_norm": 1.3691825866699219,
      "learning_rate": 9.374578947368423e-05,
      "loss": 0.3154,
      "step": 43785
    },
    {
      "epoch": 2.451898308881174,
      "grad_norm": 1.1487942934036255,
      "learning_rate": 9.374552631578947e-05,
      "loss": 0.352,
      "step": 43786
    },
    {
      "epoch": 2.451954306193303,
      "grad_norm": 1.5841172933578491,
      "learning_rate": 9.374526315789473e-05,
      "loss": 0.46,
      "step": 43787
    },
    {
      "epoch": 2.452010303505432,
      "grad_norm": 1.1862643957138062,
      "learning_rate": 9.3745e-05,
      "loss": 0.3524,
      "step": 43788
    },
    {
      "epoch": 2.452066300817561,
      "grad_norm": 1.168288230895996,
      "learning_rate": 9.374473684210526e-05,
      "loss": 0.2511,
      "step": 43789
    },
    {
      "epoch": 2.45212229812969,
      "grad_norm": 1.2948198318481445,
      "learning_rate": 9.374447368421054e-05,
      "loss": 0.344,
      "step": 43790
    },
    {
      "epoch": 2.452178295441819,
      "grad_norm": 0.9435927271842957,
      "learning_rate": 9.374421052631578e-05,
      "loss": 0.3356,
      "step": 43791
    },
    {
      "epoch": 2.452234292753948,
      "grad_norm": 1.1332719326019287,
      "learning_rate": 9.374394736842106e-05,
      "loss": 0.3971,
      "step": 43792
    },
    {
      "epoch": 2.452290290066077,
      "grad_norm": 1.0861185789108276,
      "learning_rate": 9.374368421052632e-05,
      "loss": 0.4332,
      "step": 43793
    },
    {
      "epoch": 2.452346287378206,
      "grad_norm": 1.4115861654281616,
      "learning_rate": 9.374342105263159e-05,
      "loss": 0.5223,
      "step": 43794
    },
    {
      "epoch": 2.452402284690335,
      "grad_norm": 1.119736909866333,
      "learning_rate": 9.374315789473685e-05,
      "loss": 0.3643,
      "step": 43795
    },
    {
      "epoch": 2.452458282002464,
      "grad_norm": 1.207983136177063,
      "learning_rate": 9.374289473684211e-05,
      "loss": 0.3963,
      "step": 43796
    },
    {
      "epoch": 2.452514279314593,
      "grad_norm": 1.0166454315185547,
      "learning_rate": 9.374263157894737e-05,
      "loss": 0.4215,
      "step": 43797
    },
    {
      "epoch": 2.452570276626722,
      "grad_norm": 1.1283265352249146,
      "learning_rate": 9.374236842105264e-05,
      "loss": 0.3324,
      "step": 43798
    },
    {
      "epoch": 2.4526262739388507,
      "grad_norm": 1.3217812776565552,
      "learning_rate": 9.37421052631579e-05,
      "loss": 0.3715,
      "step": 43799
    },
    {
      "epoch": 2.45268227125098,
      "grad_norm": 1.312993049621582,
      "learning_rate": 9.374184210526316e-05,
      "loss": 0.3835,
      "step": 43800
    },
    {
      "epoch": 2.4527382685631087,
      "grad_norm": 1.273610234260559,
      "learning_rate": 9.374157894736842e-05,
      "loss": 0.5424,
      "step": 43801
    },
    {
      "epoch": 2.452794265875238,
      "grad_norm": 1.1342265605926514,
      "learning_rate": 9.37413157894737e-05,
      "loss": 0.457,
      "step": 43802
    },
    {
      "epoch": 2.4528502631873668,
      "grad_norm": 1.2340644598007202,
      "learning_rate": 9.374105263157895e-05,
      "loss": 0.4694,
      "step": 43803
    },
    {
      "epoch": 2.4529062604994962,
      "grad_norm": 1.126397728919983,
      "learning_rate": 9.374078947368421e-05,
      "loss": 0.288,
      "step": 43804
    },
    {
      "epoch": 2.452962257811625,
      "grad_norm": 1.1056026220321655,
      "learning_rate": 9.374052631578947e-05,
      "loss": 0.4957,
      "step": 43805
    },
    {
      "epoch": 2.4530182551237543,
      "grad_norm": 1.2103379964828491,
      "learning_rate": 9.374026315789473e-05,
      "loss": 0.4769,
      "step": 43806
    },
    {
      "epoch": 2.453074252435883,
      "grad_norm": 1.147001028060913,
      "learning_rate": 9.374000000000001e-05,
      "loss": 0.3714,
      "step": 43807
    },
    {
      "epoch": 2.4531302497480123,
      "grad_norm": 1.0231484174728394,
      "learning_rate": 9.373973684210527e-05,
      "loss": 0.3541,
      "step": 43808
    },
    {
      "epoch": 2.453186247060141,
      "grad_norm": 1.1612920761108398,
      "learning_rate": 9.373947368421053e-05,
      "loss": 0.4203,
      "step": 43809
    },
    {
      "epoch": 2.4532422443722703,
      "grad_norm": 1.039345622062683,
      "learning_rate": 9.373921052631579e-05,
      "loss": 0.3911,
      "step": 43810
    },
    {
      "epoch": 2.453298241684399,
      "grad_norm": 1.053196907043457,
      "learning_rate": 9.373894736842106e-05,
      "loss": 0.4328,
      "step": 43811
    },
    {
      "epoch": 2.4533542389965284,
      "grad_norm": 1.1567507982254028,
      "learning_rate": 9.373868421052632e-05,
      "loss": 0.3564,
      "step": 43812
    },
    {
      "epoch": 2.453410236308657,
      "grad_norm": 1.066456913948059,
      "learning_rate": 9.373842105263159e-05,
      "loss": 0.3967,
      "step": 43813
    },
    {
      "epoch": 2.4534662336207864,
      "grad_norm": 1.111541986465454,
      "learning_rate": 9.373815789473684e-05,
      "loss": 0.4746,
      "step": 43814
    },
    {
      "epoch": 2.453522230932915,
      "grad_norm": 1.2412431240081787,
      "learning_rate": 9.373789473684211e-05,
      "loss": 0.4545,
      "step": 43815
    },
    {
      "epoch": 2.4535782282450445,
      "grad_norm": 1.2947889566421509,
      "learning_rate": 9.373763157894737e-05,
      "loss": 0.4872,
      "step": 43816
    },
    {
      "epoch": 2.453634225557173,
      "grad_norm": 1.2588623762130737,
      "learning_rate": 9.373736842105265e-05,
      "loss": 0.6162,
      "step": 43817
    },
    {
      "epoch": 2.4536902228693025,
      "grad_norm": 1.038870096206665,
      "learning_rate": 9.37371052631579e-05,
      "loss": 0.2991,
      "step": 43818
    },
    {
      "epoch": 2.453746220181431,
      "grad_norm": 1.0160081386566162,
      "learning_rate": 9.373684210526316e-05,
      "loss": 0.37,
      "step": 43819
    },
    {
      "epoch": 2.4538022174935605,
      "grad_norm": 1.3640297651290894,
      "learning_rate": 9.373657894736842e-05,
      "loss": 0.4391,
      "step": 43820
    },
    {
      "epoch": 2.453858214805689,
      "grad_norm": 1.1888067722320557,
      "learning_rate": 9.373631578947368e-05,
      "loss": 0.3819,
      "step": 43821
    },
    {
      "epoch": 2.4539142121178186,
      "grad_norm": 1.3223631381988525,
      "learning_rate": 9.373605263157896e-05,
      "loss": 0.4584,
      "step": 43822
    },
    {
      "epoch": 2.453970209429947,
      "grad_norm": 1.1852532625198364,
      "learning_rate": 9.37357894736842e-05,
      "loss": 0.3437,
      "step": 43823
    },
    {
      "epoch": 2.4540262067420766,
      "grad_norm": 1.1907708644866943,
      "learning_rate": 9.373552631578948e-05,
      "loss": 0.3982,
      "step": 43824
    },
    {
      "epoch": 2.454082204054205,
      "grad_norm": 2.5197865962982178,
      "learning_rate": 9.373526315789474e-05,
      "loss": 0.5213,
      "step": 43825
    },
    {
      "epoch": 2.4541382013663346,
      "grad_norm": 1.2427020072937012,
      "learning_rate": 9.373500000000001e-05,
      "loss": 0.3831,
      "step": 43826
    },
    {
      "epoch": 2.4541941986784632,
      "grad_norm": 1.1632657051086426,
      "learning_rate": 9.373473684210527e-05,
      "loss": 0.3876,
      "step": 43827
    },
    {
      "epoch": 2.4542501959905927,
      "grad_norm": 1.2760387659072876,
      "learning_rate": 9.373447368421053e-05,
      "loss": 0.5675,
      "step": 43828
    },
    {
      "epoch": 2.4543061933027213,
      "grad_norm": 0.9990751147270203,
      "learning_rate": 9.373421052631579e-05,
      "loss": 0.3503,
      "step": 43829
    },
    {
      "epoch": 2.4543621906148507,
      "grad_norm": 1.0173438787460327,
      "learning_rate": 9.373394736842106e-05,
      "loss": 0.3304,
      "step": 43830
    },
    {
      "epoch": 2.4544181879269793,
      "grad_norm": 1.1189181804656982,
      "learning_rate": 9.373368421052632e-05,
      "loss": 0.3402,
      "step": 43831
    },
    {
      "epoch": 2.4544741852391088,
      "grad_norm": 1.1344925165176392,
      "learning_rate": 9.373342105263158e-05,
      "loss": 0.3977,
      "step": 43832
    },
    {
      "epoch": 2.4545301825512373,
      "grad_norm": 1.3588571548461914,
      "learning_rate": 9.373315789473684e-05,
      "loss": 0.3711,
      "step": 43833
    },
    {
      "epoch": 2.454586179863367,
      "grad_norm": 1.3361027240753174,
      "learning_rate": 9.373289473684211e-05,
      "loss": 0.4532,
      "step": 43834
    },
    {
      "epoch": 2.4546421771754954,
      "grad_norm": 1.056199073791504,
      "learning_rate": 9.373263157894737e-05,
      "loss": 0.4686,
      "step": 43835
    },
    {
      "epoch": 2.454698174487625,
      "grad_norm": 1.0404422283172607,
      "learning_rate": 9.373236842105265e-05,
      "loss": 0.3905,
      "step": 43836
    },
    {
      "epoch": 2.4547541717997534,
      "grad_norm": 1.199611783027649,
      "learning_rate": 9.37321052631579e-05,
      "loss": 0.4326,
      "step": 43837
    },
    {
      "epoch": 2.454810169111883,
      "grad_norm": 1.0852619409561157,
      "learning_rate": 9.373184210526317e-05,
      "loss": 0.5105,
      "step": 43838
    },
    {
      "epoch": 2.4548661664240115,
      "grad_norm": 1.2011563777923584,
      "learning_rate": 9.373157894736843e-05,
      "loss": 0.3648,
      "step": 43839
    },
    {
      "epoch": 2.4549221637361405,
      "grad_norm": 1.1973915100097656,
      "learning_rate": 9.373131578947369e-05,
      "loss": 0.5492,
      "step": 43840
    },
    {
      "epoch": 2.4549781610482695,
      "grad_norm": 1.3122315406799316,
      "learning_rate": 9.373105263157895e-05,
      "loss": 0.3891,
      "step": 43841
    },
    {
      "epoch": 2.4550341583603985,
      "grad_norm": 1.1656357049942017,
      "learning_rate": 9.37307894736842e-05,
      "loss": 0.3554,
      "step": 43842
    },
    {
      "epoch": 2.4550901556725275,
      "grad_norm": 1.22404146194458,
      "learning_rate": 9.373052631578948e-05,
      "loss": 0.405,
      "step": 43843
    },
    {
      "epoch": 2.4551461529846565,
      "grad_norm": 1.2471683025360107,
      "learning_rate": 9.373026315789474e-05,
      "loss": 0.3592,
      "step": 43844
    },
    {
      "epoch": 2.4552021502967856,
      "grad_norm": 1.3199968338012695,
      "learning_rate": 9.373000000000001e-05,
      "loss": 0.4145,
      "step": 43845
    },
    {
      "epoch": 2.4552581476089146,
      "grad_norm": 1.490583062171936,
      "learning_rate": 9.372973684210526e-05,
      "loss": 0.4096,
      "step": 43846
    },
    {
      "epoch": 2.4553141449210436,
      "grad_norm": 1.3668901920318604,
      "learning_rate": 9.372947368421053e-05,
      "loss": 0.418,
      "step": 43847
    },
    {
      "epoch": 2.4553701422331726,
      "grad_norm": 1.2287410497665405,
      "learning_rate": 9.372921052631579e-05,
      "loss": 0.3582,
      "step": 43848
    },
    {
      "epoch": 2.4554261395453016,
      "grad_norm": 1.4730790853500366,
      "learning_rate": 9.372894736842106e-05,
      "loss": 0.5673,
      "step": 43849
    },
    {
      "epoch": 2.4554821368574307,
      "grad_norm": 1.4054229259490967,
      "learning_rate": 9.372868421052632e-05,
      "loss": 0.3533,
      "step": 43850
    },
    {
      "epoch": 2.4555381341695597,
      "grad_norm": 1.4618122577667236,
      "learning_rate": 9.372842105263158e-05,
      "loss": 0.4733,
      "step": 43851
    },
    {
      "epoch": 2.4555941314816887,
      "grad_norm": 1.2669718265533447,
      "learning_rate": 9.372815789473684e-05,
      "loss": 0.2999,
      "step": 43852
    },
    {
      "epoch": 2.4556501287938177,
      "grad_norm": 1.144573450088501,
      "learning_rate": 9.372789473684212e-05,
      "loss": 0.4246,
      "step": 43853
    },
    {
      "epoch": 2.4557061261059467,
      "grad_norm": 1.2047260999679565,
      "learning_rate": 9.372763157894738e-05,
      "loss": 0.4535,
      "step": 43854
    },
    {
      "epoch": 2.4557621234180758,
      "grad_norm": 1.1501902341842651,
      "learning_rate": 9.372736842105264e-05,
      "loss": 0.4079,
      "step": 43855
    },
    {
      "epoch": 2.455818120730205,
      "grad_norm": 0.9730649590492249,
      "learning_rate": 9.37271052631579e-05,
      "loss": 0.3059,
      "step": 43856
    },
    {
      "epoch": 2.455874118042334,
      "grad_norm": 1.3352222442626953,
      "learning_rate": 9.372684210526316e-05,
      "loss": 0.3857,
      "step": 43857
    },
    {
      "epoch": 2.455930115354463,
      "grad_norm": 1.0371432304382324,
      "learning_rate": 9.372657894736843e-05,
      "loss": 0.3403,
      "step": 43858
    },
    {
      "epoch": 2.455986112666592,
      "grad_norm": 1.271681547164917,
      "learning_rate": 9.372631578947369e-05,
      "loss": 0.4078,
      "step": 43859
    },
    {
      "epoch": 2.456042109978721,
      "grad_norm": 0.9867258667945862,
      "learning_rate": 9.372605263157895e-05,
      "loss": 0.2883,
      "step": 43860
    },
    {
      "epoch": 2.45609810729085,
      "grad_norm": 1.5032049417495728,
      "learning_rate": 9.372578947368421e-05,
      "loss": 0.5264,
      "step": 43861
    },
    {
      "epoch": 2.456154104602979,
      "grad_norm": 1.1476589441299438,
      "learning_rate": 9.372552631578948e-05,
      "loss": 0.5132,
      "step": 43862
    },
    {
      "epoch": 2.456210101915108,
      "grad_norm": 1.3256691694259644,
      "learning_rate": 9.372526315789474e-05,
      "loss": 0.5342,
      "step": 43863
    },
    {
      "epoch": 2.456266099227237,
      "grad_norm": 1.1948165893554688,
      "learning_rate": 9.3725e-05,
      "loss": 0.5206,
      "step": 43864
    },
    {
      "epoch": 2.456322096539366,
      "grad_norm": 0.9295013546943665,
      "learning_rate": 9.372473684210526e-05,
      "loss": 0.4475,
      "step": 43865
    },
    {
      "epoch": 2.456378093851495,
      "grad_norm": 1.3460439443588257,
      "learning_rate": 9.372447368421053e-05,
      "loss": 0.4908,
      "step": 43866
    },
    {
      "epoch": 2.456434091163624,
      "grad_norm": 1.1859809160232544,
      "learning_rate": 9.37242105263158e-05,
      "loss": 0.3331,
      "step": 43867
    },
    {
      "epoch": 2.456490088475753,
      "grad_norm": 1.2419772148132324,
      "learning_rate": 9.372394736842107e-05,
      "loss": 0.3966,
      "step": 43868
    },
    {
      "epoch": 2.456546085787882,
      "grad_norm": 1.3940320014953613,
      "learning_rate": 9.372368421052631e-05,
      "loss": 0.4059,
      "step": 43869
    },
    {
      "epoch": 2.456602083100011,
      "grad_norm": 1.1169556379318237,
      "learning_rate": 9.372342105263159e-05,
      "loss": 0.4131,
      "step": 43870
    },
    {
      "epoch": 2.45665808041214,
      "grad_norm": 1.2106316089630127,
      "learning_rate": 9.372315789473685e-05,
      "loss": 0.4508,
      "step": 43871
    },
    {
      "epoch": 2.456714077724269,
      "grad_norm": 1.2936195135116577,
      "learning_rate": 9.372289473684212e-05,
      "loss": 0.4039,
      "step": 43872
    },
    {
      "epoch": 2.456770075036398,
      "grad_norm": 1.208203911781311,
      "learning_rate": 9.372263157894738e-05,
      "loss": 0.4323,
      "step": 43873
    },
    {
      "epoch": 2.456826072348527,
      "grad_norm": 1.451072096824646,
      "learning_rate": 9.372236842105263e-05,
      "loss": 0.4534,
      "step": 43874
    },
    {
      "epoch": 2.456882069660656,
      "grad_norm": 1.1147087812423706,
      "learning_rate": 9.37221052631579e-05,
      "loss": 0.4012,
      "step": 43875
    },
    {
      "epoch": 2.456938066972785,
      "grad_norm": 1.38563072681427,
      "learning_rate": 9.372184210526316e-05,
      "loss": 0.4124,
      "step": 43876
    },
    {
      "epoch": 2.456994064284914,
      "grad_norm": 1.0656094551086426,
      "learning_rate": 9.372157894736843e-05,
      "loss": 0.3636,
      "step": 43877
    },
    {
      "epoch": 2.457050061597043,
      "grad_norm": 1.2050896883010864,
      "learning_rate": 9.372131578947368e-05,
      "loss": 0.3482,
      "step": 43878
    },
    {
      "epoch": 2.457106058909172,
      "grad_norm": 1.5593667030334473,
      "learning_rate": 9.372105263157895e-05,
      "loss": 0.5082,
      "step": 43879
    },
    {
      "epoch": 2.4571620562213012,
      "grad_norm": 1.3880250453948975,
      "learning_rate": 9.372078947368421e-05,
      "loss": 0.4241,
      "step": 43880
    },
    {
      "epoch": 2.4572180535334303,
      "grad_norm": 1.1753108501434326,
      "learning_rate": 9.372052631578948e-05,
      "loss": 0.5359,
      "step": 43881
    },
    {
      "epoch": 2.4572740508455593,
      "grad_norm": 1.5554665327072144,
      "learning_rate": 9.372026315789474e-05,
      "loss": 0.4002,
      "step": 43882
    },
    {
      "epoch": 2.4573300481576883,
      "grad_norm": 1.3856898546218872,
      "learning_rate": 9.372e-05,
      "loss": 0.4492,
      "step": 43883
    },
    {
      "epoch": 2.4573860454698173,
      "grad_norm": 1.2469426393508911,
      "learning_rate": 9.371973684210526e-05,
      "loss": 0.4838,
      "step": 43884
    },
    {
      "epoch": 2.4574420427819463,
      "grad_norm": 1.2526464462280273,
      "learning_rate": 9.371947368421054e-05,
      "loss": 0.4892,
      "step": 43885
    },
    {
      "epoch": 2.4574980400940754,
      "grad_norm": 1.2267673015594482,
      "learning_rate": 9.37192105263158e-05,
      "loss": 0.3422,
      "step": 43886
    },
    {
      "epoch": 2.4575540374062044,
      "grad_norm": 1.4503768682479858,
      "learning_rate": 9.371894736842106e-05,
      "loss": 0.5103,
      "step": 43887
    },
    {
      "epoch": 2.4576100347183334,
      "grad_norm": 3.7832815647125244,
      "learning_rate": 9.371868421052632e-05,
      "loss": 0.397,
      "step": 43888
    },
    {
      "epoch": 2.4576660320304624,
      "grad_norm": 0.9889214038848877,
      "learning_rate": 9.371842105263159e-05,
      "loss": 0.2748,
      "step": 43889
    },
    {
      "epoch": 2.4577220293425914,
      "grad_norm": 1.1794129610061646,
      "learning_rate": 9.371815789473685e-05,
      "loss": 0.3459,
      "step": 43890
    },
    {
      "epoch": 2.4577780266547204,
      "grad_norm": 1.0466742515563965,
      "learning_rate": 9.371789473684211e-05,
      "loss": 0.3532,
      "step": 43891
    },
    {
      "epoch": 2.4578340239668495,
      "grad_norm": 1.0683820247650146,
      "learning_rate": 9.371763157894737e-05,
      "loss": 0.2881,
      "step": 43892
    },
    {
      "epoch": 2.4578900212789785,
      "grad_norm": 1.264441967010498,
      "learning_rate": 9.371736842105263e-05,
      "loss": 0.4009,
      "step": 43893
    },
    {
      "epoch": 2.4579460185911075,
      "grad_norm": 1.0755852460861206,
      "learning_rate": 9.37171052631579e-05,
      "loss": 0.2855,
      "step": 43894
    },
    {
      "epoch": 2.4580020159032365,
      "grad_norm": 1.2200274467468262,
      "learning_rate": 9.371684210526316e-05,
      "loss": 0.3453,
      "step": 43895
    },
    {
      "epoch": 2.4580580132153655,
      "grad_norm": 1.0976800918579102,
      "learning_rate": 9.371657894736842e-05,
      "loss": 0.4506,
      "step": 43896
    },
    {
      "epoch": 2.4581140105274946,
      "grad_norm": 1.2620670795440674,
      "learning_rate": 9.371631578947368e-05,
      "loss": 0.4072,
      "step": 43897
    },
    {
      "epoch": 2.4581700078396236,
      "grad_norm": 1.0903087854385376,
      "learning_rate": 9.371605263157895e-05,
      "loss": 0.5302,
      "step": 43898
    },
    {
      "epoch": 2.4582260051517526,
      "grad_norm": 1.069486379623413,
      "learning_rate": 9.371578947368421e-05,
      "loss": 0.4267,
      "step": 43899
    },
    {
      "epoch": 2.4582820024638816,
      "grad_norm": 1.0436335802078247,
      "learning_rate": 9.371552631578949e-05,
      "loss": 0.3439,
      "step": 43900
    },
    {
      "epoch": 2.4583379997760106,
      "grad_norm": 1.3194355964660645,
      "learning_rate": 9.371526315789473e-05,
      "loss": 0.428,
      "step": 43901
    },
    {
      "epoch": 2.4583939970881397,
      "grad_norm": 1.2429838180541992,
      "learning_rate": 9.3715e-05,
      "loss": 0.3492,
      "step": 43902
    },
    {
      "epoch": 2.4584499944002687,
      "grad_norm": 1.0790207386016846,
      "learning_rate": 9.371473684210527e-05,
      "loss": 0.3143,
      "step": 43903
    },
    {
      "epoch": 2.4585059917123977,
      "grad_norm": 1.4567341804504395,
      "learning_rate": 9.371447368421054e-05,
      "loss": 0.5508,
      "step": 43904
    },
    {
      "epoch": 2.4585619890245267,
      "grad_norm": 1.1903563737869263,
      "learning_rate": 9.37142105263158e-05,
      "loss": 0.3575,
      "step": 43905
    },
    {
      "epoch": 2.4586179863366557,
      "grad_norm": 1.1110758781433105,
      "learning_rate": 9.371394736842106e-05,
      "loss": 0.4926,
      "step": 43906
    },
    {
      "epoch": 2.4586739836487848,
      "grad_norm": 1.6739675998687744,
      "learning_rate": 9.371368421052632e-05,
      "loss": 0.4816,
      "step": 43907
    },
    {
      "epoch": 2.4587299809609138,
      "grad_norm": 2.0655746459960938,
      "learning_rate": 9.371342105263158e-05,
      "loss": 0.664,
      "step": 43908
    },
    {
      "epoch": 2.458785978273043,
      "grad_norm": 1.246039867401123,
      "learning_rate": 9.371315789473685e-05,
      "loss": 0.4272,
      "step": 43909
    },
    {
      "epoch": 2.458841975585172,
      "grad_norm": 1.3805861473083496,
      "learning_rate": 9.371289473684211e-05,
      "loss": 0.4314,
      "step": 43910
    },
    {
      "epoch": 2.458897972897301,
      "grad_norm": 1.2322567701339722,
      "learning_rate": 9.371263157894737e-05,
      "loss": 0.3976,
      "step": 43911
    },
    {
      "epoch": 2.45895397020943,
      "grad_norm": 1.2185637950897217,
      "learning_rate": 9.371236842105263e-05,
      "loss": 0.5363,
      "step": 43912
    },
    {
      "epoch": 2.459009967521559,
      "grad_norm": 1.1679600477218628,
      "learning_rate": 9.37121052631579e-05,
      "loss": 0.3417,
      "step": 43913
    },
    {
      "epoch": 2.459065964833688,
      "grad_norm": 1.1826345920562744,
      "learning_rate": 9.371184210526316e-05,
      "loss": 0.3998,
      "step": 43914
    },
    {
      "epoch": 2.459121962145817,
      "grad_norm": 1.0912599563598633,
      "learning_rate": 9.371157894736842e-05,
      "loss": 0.3837,
      "step": 43915
    },
    {
      "epoch": 2.459177959457946,
      "grad_norm": 1.1540943384170532,
      "learning_rate": 9.371131578947368e-05,
      "loss": 0.3983,
      "step": 43916
    },
    {
      "epoch": 2.459233956770075,
      "grad_norm": 1.4016510248184204,
      "learning_rate": 9.371105263157896e-05,
      "loss": 0.4861,
      "step": 43917
    },
    {
      "epoch": 2.459289954082204,
      "grad_norm": 1.6697447299957275,
      "learning_rate": 9.371078947368422e-05,
      "loss": 0.4754,
      "step": 43918
    },
    {
      "epoch": 2.459345951394333,
      "grad_norm": 1.7862989902496338,
      "learning_rate": 9.371052631578948e-05,
      "loss": 0.3556,
      "step": 43919
    },
    {
      "epoch": 2.459401948706462,
      "grad_norm": 1.163803219795227,
      "learning_rate": 9.371026315789474e-05,
      "loss": 0.4877,
      "step": 43920
    },
    {
      "epoch": 2.459457946018591,
      "grad_norm": 1.2600648403167725,
      "learning_rate": 9.371000000000001e-05,
      "loss": 0.4212,
      "step": 43921
    },
    {
      "epoch": 2.45951394333072,
      "grad_norm": 1.0461797714233398,
      "learning_rate": 9.370973684210527e-05,
      "loss": 0.3764,
      "step": 43922
    },
    {
      "epoch": 2.459569940642849,
      "grad_norm": 1.1030359268188477,
      "learning_rate": 9.370947368421054e-05,
      "loss": 0.3828,
      "step": 43923
    },
    {
      "epoch": 2.459625937954978,
      "grad_norm": 1.1080652475357056,
      "learning_rate": 9.370921052631579e-05,
      "loss": 0.2918,
      "step": 43924
    },
    {
      "epoch": 2.459681935267107,
      "grad_norm": 1.3141148090362549,
      "learning_rate": 9.370894736842105e-05,
      "loss": 0.4478,
      "step": 43925
    },
    {
      "epoch": 2.459737932579236,
      "grad_norm": 1.2177132368087769,
      "learning_rate": 9.370868421052632e-05,
      "loss": 0.3594,
      "step": 43926
    },
    {
      "epoch": 2.459793929891365,
      "grad_norm": 1.1319793462753296,
      "learning_rate": 9.370842105263158e-05,
      "loss": 0.327,
      "step": 43927
    },
    {
      "epoch": 2.459849927203494,
      "grad_norm": 1.0238929986953735,
      "learning_rate": 9.370815789473685e-05,
      "loss": 0.4054,
      "step": 43928
    },
    {
      "epoch": 2.459905924515623,
      "grad_norm": 1.152561068534851,
      "learning_rate": 9.37078947368421e-05,
      "loss": 0.475,
      "step": 43929
    },
    {
      "epoch": 2.459961921827752,
      "grad_norm": 0.9773671627044678,
      "learning_rate": 9.370763157894737e-05,
      "loss": 0.3104,
      "step": 43930
    },
    {
      "epoch": 2.460017919139881,
      "grad_norm": 1.2545851469039917,
      "learning_rate": 9.370736842105263e-05,
      "loss": 0.3792,
      "step": 43931
    },
    {
      "epoch": 2.4600739164520102,
      "grad_norm": 1.4108887910842896,
      "learning_rate": 9.37071052631579e-05,
      "loss": 0.3936,
      "step": 43932
    },
    {
      "epoch": 2.4601299137641393,
      "grad_norm": 1.1425073146820068,
      "learning_rate": 9.370684210526315e-05,
      "loss": 0.3553,
      "step": 43933
    },
    {
      "epoch": 2.4601859110762683,
      "grad_norm": 1.1321817636489868,
      "learning_rate": 9.370657894736843e-05,
      "loss": 0.3609,
      "step": 43934
    },
    {
      "epoch": 2.4602419083883973,
      "grad_norm": 1.2184488773345947,
      "learning_rate": 9.370631578947369e-05,
      "loss": 0.4172,
      "step": 43935
    },
    {
      "epoch": 2.4602979057005263,
      "grad_norm": 1.021667242050171,
      "learning_rate": 9.370605263157896e-05,
      "loss": 0.3046,
      "step": 43936
    },
    {
      "epoch": 2.4603539030126553,
      "grad_norm": 1.2242690324783325,
      "learning_rate": 9.370578947368422e-05,
      "loss": 0.467,
      "step": 43937
    },
    {
      "epoch": 2.4604099003247843,
      "grad_norm": 1.1548478603363037,
      "learning_rate": 9.370552631578948e-05,
      "loss": 0.4465,
      "step": 43938
    },
    {
      "epoch": 2.4604658976369134,
      "grad_norm": 1.1805572509765625,
      "learning_rate": 9.370526315789474e-05,
      "loss": 0.5234,
      "step": 43939
    },
    {
      "epoch": 2.4605218949490424,
      "grad_norm": 1.3229966163635254,
      "learning_rate": 9.370500000000001e-05,
      "loss": 0.4428,
      "step": 43940
    },
    {
      "epoch": 2.4605778922611714,
      "grad_norm": 1.2532237768173218,
      "learning_rate": 9.370473684210527e-05,
      "loss": 0.4855,
      "step": 43941
    },
    {
      "epoch": 2.4606338895733004,
      "grad_norm": 1.0509499311447144,
      "learning_rate": 9.370447368421053e-05,
      "loss": 0.3579,
      "step": 43942
    },
    {
      "epoch": 2.4606898868854294,
      "grad_norm": 0.9565574526786804,
      "learning_rate": 9.370421052631579e-05,
      "loss": 0.3573,
      "step": 43943
    },
    {
      "epoch": 2.4607458841975585,
      "grad_norm": 1.2125788927078247,
      "learning_rate": 9.370394736842105e-05,
      "loss": 0.3405,
      "step": 43944
    },
    {
      "epoch": 2.4608018815096875,
      "grad_norm": 1.1953964233398438,
      "learning_rate": 9.370368421052632e-05,
      "loss": 0.5305,
      "step": 43945
    },
    {
      "epoch": 2.4608578788218165,
      "grad_norm": 1.26026451587677,
      "learning_rate": 9.370342105263158e-05,
      "loss": 0.3852,
      "step": 43946
    },
    {
      "epoch": 2.4609138761339455,
      "grad_norm": 1.179924726486206,
      "learning_rate": 9.370315789473684e-05,
      "loss": 0.3899,
      "step": 43947
    },
    {
      "epoch": 2.4609698734460745,
      "grad_norm": 1.1868436336517334,
      "learning_rate": 9.37028947368421e-05,
      "loss": 0.3986,
      "step": 43948
    },
    {
      "epoch": 2.4610258707582036,
      "grad_norm": 1.2180452346801758,
      "learning_rate": 9.370263157894738e-05,
      "loss": 0.3813,
      "step": 43949
    },
    {
      "epoch": 2.4610818680703326,
      "grad_norm": 1.1777856349945068,
      "learning_rate": 9.370236842105264e-05,
      "loss": 0.4419,
      "step": 43950
    },
    {
      "epoch": 2.4611378653824616,
      "grad_norm": 1.1629559993743896,
      "learning_rate": 9.37021052631579e-05,
      "loss": 0.4021,
      "step": 43951
    },
    {
      "epoch": 2.4611938626945906,
      "grad_norm": 2.236706256866455,
      "learning_rate": 9.370184210526316e-05,
      "loss": 0.4782,
      "step": 43952
    },
    {
      "epoch": 2.4612498600067196,
      "grad_norm": 1.864449143409729,
      "learning_rate": 9.370157894736843e-05,
      "loss": 0.3784,
      "step": 43953
    },
    {
      "epoch": 2.4613058573188487,
      "grad_norm": 1.3919776678085327,
      "learning_rate": 9.370131578947369e-05,
      "loss": 0.4957,
      "step": 43954
    },
    {
      "epoch": 2.4613618546309777,
      "grad_norm": 1.4784114360809326,
      "learning_rate": 9.370105263157896e-05,
      "loss": 0.3823,
      "step": 43955
    },
    {
      "epoch": 2.4614178519431067,
      "grad_norm": 1.1372689008712769,
      "learning_rate": 9.370078947368421e-05,
      "loss": 0.4858,
      "step": 43956
    },
    {
      "epoch": 2.4614738492552357,
      "grad_norm": 1.2237818241119385,
      "learning_rate": 9.370052631578948e-05,
      "loss": 0.4301,
      "step": 43957
    },
    {
      "epoch": 2.4615298465673647,
      "grad_norm": 1.2508577108383179,
      "learning_rate": 9.370026315789474e-05,
      "loss": 0.3645,
      "step": 43958
    },
    {
      "epoch": 2.4615858438794938,
      "grad_norm": 1.1041606664657593,
      "learning_rate": 9.370000000000001e-05,
      "loss": 0.3234,
      "step": 43959
    },
    {
      "epoch": 2.4616418411916228,
      "grad_norm": 1.509909987449646,
      "learning_rate": 9.369973684210527e-05,
      "loss": 0.4552,
      "step": 43960
    },
    {
      "epoch": 2.461697838503752,
      "grad_norm": 1.1359837055206299,
      "learning_rate": 9.369947368421052e-05,
      "loss": 0.4749,
      "step": 43961
    },
    {
      "epoch": 2.461753835815881,
      "grad_norm": 1.4343856573104858,
      "learning_rate": 9.36992105263158e-05,
      "loss": 0.2567,
      "step": 43962
    },
    {
      "epoch": 2.46180983312801,
      "grad_norm": 1.3366953134536743,
      "learning_rate": 9.369894736842105e-05,
      "loss": 0.3146,
      "step": 43963
    },
    {
      "epoch": 2.461865830440139,
      "grad_norm": 1.5013960599899292,
      "learning_rate": 9.369868421052633e-05,
      "loss": 0.427,
      "step": 43964
    },
    {
      "epoch": 2.461921827752268,
      "grad_norm": 1.2389426231384277,
      "learning_rate": 9.369842105263159e-05,
      "loss": 0.3601,
      "step": 43965
    },
    {
      "epoch": 2.461977825064397,
      "grad_norm": 1.0977588891983032,
      "learning_rate": 9.369815789473685e-05,
      "loss": 0.4056,
      "step": 43966
    },
    {
      "epoch": 2.462033822376526,
      "grad_norm": 1.286423683166504,
      "learning_rate": 9.36978947368421e-05,
      "loss": 0.512,
      "step": 43967
    },
    {
      "epoch": 2.462089819688655,
      "grad_norm": 1.1975669860839844,
      "learning_rate": 9.369763157894738e-05,
      "loss": 0.3562,
      "step": 43968
    },
    {
      "epoch": 2.462145817000784,
      "grad_norm": 1.1896835565567017,
      "learning_rate": 9.369736842105264e-05,
      "loss": 0.362,
      "step": 43969
    },
    {
      "epoch": 2.462201814312913,
      "grad_norm": 1.2846862077713013,
      "learning_rate": 9.36971052631579e-05,
      "loss": 0.4659,
      "step": 43970
    },
    {
      "epoch": 2.462257811625042,
      "grad_norm": 1.1826756000518799,
      "learning_rate": 9.369684210526316e-05,
      "loss": 0.3571,
      "step": 43971
    },
    {
      "epoch": 2.462313808937171,
      "grad_norm": 1.100092887878418,
      "learning_rate": 9.369657894736843e-05,
      "loss": 0.327,
      "step": 43972
    },
    {
      "epoch": 2.4623698062493,
      "grad_norm": 1.0298722982406616,
      "learning_rate": 9.369631578947369e-05,
      "loss": 0.461,
      "step": 43973
    },
    {
      "epoch": 2.462425803561429,
      "grad_norm": 1.1048294305801392,
      "learning_rate": 9.369605263157895e-05,
      "loss": 0.3424,
      "step": 43974
    },
    {
      "epoch": 2.462481800873558,
      "grad_norm": 0.9605061411857605,
      "learning_rate": 9.369578947368421e-05,
      "loss": 0.3005,
      "step": 43975
    },
    {
      "epoch": 2.462537798185687,
      "grad_norm": 1.3487586975097656,
      "learning_rate": 9.369552631578948e-05,
      "loss": 0.4386,
      "step": 43976
    },
    {
      "epoch": 2.462593795497816,
      "grad_norm": 1.4309014081954956,
      "learning_rate": 9.369526315789474e-05,
      "loss": 0.4834,
      "step": 43977
    },
    {
      "epoch": 2.462649792809945,
      "grad_norm": 1.056044101715088,
      "learning_rate": 9.3695e-05,
      "loss": 0.391,
      "step": 43978
    },
    {
      "epoch": 2.462705790122074,
      "grad_norm": 1.4049144983291626,
      "learning_rate": 9.369473684210526e-05,
      "loss": 0.3939,
      "step": 43979
    },
    {
      "epoch": 2.462761787434203,
      "grad_norm": 1.1205466985702515,
      "learning_rate": 9.369447368421052e-05,
      "loss": 0.3747,
      "step": 43980
    },
    {
      "epoch": 2.462817784746332,
      "grad_norm": 1.1049559116363525,
      "learning_rate": 9.36942105263158e-05,
      "loss": 0.4836,
      "step": 43981
    },
    {
      "epoch": 2.462873782058461,
      "grad_norm": 1.0575199127197266,
      "learning_rate": 9.369394736842106e-05,
      "loss": 0.3208,
      "step": 43982
    },
    {
      "epoch": 2.46292977937059,
      "grad_norm": 1.0496854782104492,
      "learning_rate": 9.369368421052633e-05,
      "loss": 0.3444,
      "step": 43983
    },
    {
      "epoch": 2.4629857766827192,
      "grad_norm": 1.1299480199813843,
      "learning_rate": 9.369342105263158e-05,
      "loss": 0.3904,
      "step": 43984
    },
    {
      "epoch": 2.4630417739948482,
      "grad_norm": 1.074558138847351,
      "learning_rate": 9.369315789473685e-05,
      "loss": 0.3424,
      "step": 43985
    },
    {
      "epoch": 2.4630977713069773,
      "grad_norm": 1.2925041913986206,
      "learning_rate": 9.369289473684211e-05,
      "loss": 0.551,
      "step": 43986
    },
    {
      "epoch": 2.4631537686191063,
      "grad_norm": 1.4883556365966797,
      "learning_rate": 9.369263157894738e-05,
      "loss": 0.4566,
      "step": 43987
    },
    {
      "epoch": 2.4632097659312353,
      "grad_norm": 1.2466665506362915,
      "learning_rate": 9.369236842105263e-05,
      "loss": 0.4134,
      "step": 43988
    },
    {
      "epoch": 2.4632657632433643,
      "grad_norm": 1.0996952056884766,
      "learning_rate": 9.36921052631579e-05,
      "loss": 0.3135,
      "step": 43989
    },
    {
      "epoch": 2.4633217605554933,
      "grad_norm": 1.114341139793396,
      "learning_rate": 9.369184210526316e-05,
      "loss": 0.3493,
      "step": 43990
    },
    {
      "epoch": 2.4633777578676224,
      "grad_norm": 1.1812667846679688,
      "learning_rate": 9.369157894736843e-05,
      "loss": 0.446,
      "step": 43991
    },
    {
      "epoch": 2.4634337551797514,
      "grad_norm": 1.137360692024231,
      "learning_rate": 9.36913157894737e-05,
      "loss": 0.4037,
      "step": 43992
    },
    {
      "epoch": 2.4634897524918804,
      "grad_norm": 1.022881269454956,
      "learning_rate": 9.369105263157895e-05,
      "loss": 0.3637,
      "step": 43993
    },
    {
      "epoch": 2.4635457498040094,
      "grad_norm": 1.2217137813568115,
      "learning_rate": 9.369078947368421e-05,
      "loss": 0.4461,
      "step": 43994
    },
    {
      "epoch": 2.4636017471161384,
      "grad_norm": 0.9796823263168335,
      "learning_rate": 9.369052631578947e-05,
      "loss": 0.3404,
      "step": 43995
    },
    {
      "epoch": 2.4636577444282675,
      "grad_norm": 1.036768913269043,
      "learning_rate": 9.369026315789475e-05,
      "loss": 0.297,
      "step": 43996
    },
    {
      "epoch": 2.4637137417403965,
      "grad_norm": 1.3559069633483887,
      "learning_rate": 9.369e-05,
      "loss": 0.3296,
      "step": 43997
    },
    {
      "epoch": 2.4637697390525255,
      "grad_norm": 1.221373200416565,
      "learning_rate": 9.368973684210527e-05,
      "loss": 0.4202,
      "step": 43998
    },
    {
      "epoch": 2.4638257363646545,
      "grad_norm": 1.3434100151062012,
      "learning_rate": 9.368947368421053e-05,
      "loss": 0.4098,
      "step": 43999
    },
    {
      "epoch": 2.4638817336767835,
      "grad_norm": 1.3566709756851196,
      "learning_rate": 9.36892105263158e-05,
      "loss": 0.4464,
      "step": 44000
    },
    {
      "epoch": 2.4639377309889126,
      "grad_norm": 0.9783819317817688,
      "learning_rate": 9.368894736842106e-05,
      "loss": 0.3247,
      "step": 44001
    },
    {
      "epoch": 2.4639937283010416,
      "grad_norm": 1.1412975788116455,
      "learning_rate": 9.368868421052632e-05,
      "loss": 0.342,
      "step": 44002
    },
    {
      "epoch": 2.4640497256131706,
      "grad_norm": 1.256348729133606,
      "learning_rate": 9.368842105263158e-05,
      "loss": 0.4867,
      "step": 44003
    },
    {
      "epoch": 2.4641057229252996,
      "grad_norm": 1.1195480823516846,
      "learning_rate": 9.368815789473685e-05,
      "loss": 0.3567,
      "step": 44004
    },
    {
      "epoch": 2.4641617202374286,
      "grad_norm": 1.0207374095916748,
      "learning_rate": 9.368789473684211e-05,
      "loss": 0.387,
      "step": 44005
    },
    {
      "epoch": 2.4642177175495577,
      "grad_norm": 1.0712746381759644,
      "learning_rate": 9.368763157894737e-05,
      "loss": 0.3746,
      "step": 44006
    },
    {
      "epoch": 2.4642737148616867,
      "grad_norm": 1.0711150169372559,
      "learning_rate": 9.368736842105263e-05,
      "loss": 0.3524,
      "step": 44007
    },
    {
      "epoch": 2.4643297121738157,
      "grad_norm": 1.1193681955337524,
      "learning_rate": 9.36871052631579e-05,
      "loss": 0.3801,
      "step": 44008
    },
    {
      "epoch": 2.4643857094859447,
      "grad_norm": 2.050351858139038,
      "learning_rate": 9.368684210526316e-05,
      "loss": 0.4216,
      "step": 44009
    },
    {
      "epoch": 2.4644417067980737,
      "grad_norm": 1.061379075050354,
      "learning_rate": 9.368657894736844e-05,
      "loss": 0.3953,
      "step": 44010
    },
    {
      "epoch": 2.4644977041102027,
      "grad_norm": 1.0632109642028809,
      "learning_rate": 9.368631578947368e-05,
      "loss": 0.2962,
      "step": 44011
    },
    {
      "epoch": 2.4645537014223318,
      "grad_norm": 1.1900075674057007,
      "learning_rate": 9.368605263157894e-05,
      "loss": 0.3209,
      "step": 44012
    },
    {
      "epoch": 2.464609698734461,
      "grad_norm": 1.0738804340362549,
      "learning_rate": 9.368578947368422e-05,
      "loss": 0.3785,
      "step": 44013
    },
    {
      "epoch": 2.46466569604659,
      "grad_norm": 1.4291799068450928,
      "learning_rate": 9.368552631578948e-05,
      "loss": 0.39,
      "step": 44014
    },
    {
      "epoch": 2.464721693358719,
      "grad_norm": 1.1958266496658325,
      "learning_rate": 9.368526315789475e-05,
      "loss": 0.4006,
      "step": 44015
    },
    {
      "epoch": 2.464777690670848,
      "grad_norm": 1.3450801372528076,
      "learning_rate": 9.3685e-05,
      "loss": 0.5331,
      "step": 44016
    },
    {
      "epoch": 2.464833687982977,
      "grad_norm": 1.0440376996994019,
      "learning_rate": 9.368473684210527e-05,
      "loss": 0.2854,
      "step": 44017
    },
    {
      "epoch": 2.464889685295106,
      "grad_norm": 1.272728681564331,
      "learning_rate": 9.368447368421053e-05,
      "loss": 0.5144,
      "step": 44018
    },
    {
      "epoch": 2.464945682607235,
      "grad_norm": 1.1235953569412231,
      "learning_rate": 9.36842105263158e-05,
      "loss": 0.2873,
      "step": 44019
    },
    {
      "epoch": 2.465001679919364,
      "grad_norm": 1.303578495979309,
      "learning_rate": 9.368394736842106e-05,
      "loss": 0.4773,
      "step": 44020
    },
    {
      "epoch": 2.465057677231493,
      "grad_norm": 1.3914977312088013,
      "learning_rate": 9.368368421052632e-05,
      "loss": 0.2957,
      "step": 44021
    },
    {
      "epoch": 2.465113674543622,
      "grad_norm": 1.169516682624817,
      "learning_rate": 9.368342105263158e-05,
      "loss": 0.3974,
      "step": 44022
    },
    {
      "epoch": 2.465169671855751,
      "grad_norm": 1.167578101158142,
      "learning_rate": 9.368315789473685e-05,
      "loss": 0.3527,
      "step": 44023
    },
    {
      "epoch": 2.46522566916788,
      "grad_norm": 1.1339764595031738,
      "learning_rate": 9.368289473684211e-05,
      "loss": 0.4494,
      "step": 44024
    },
    {
      "epoch": 2.465281666480009,
      "grad_norm": 1.078296422958374,
      "learning_rate": 9.368263157894737e-05,
      "loss": 0.3479,
      "step": 44025
    },
    {
      "epoch": 2.465337663792138,
      "grad_norm": 1.1302907466888428,
      "learning_rate": 9.368236842105263e-05,
      "loss": 0.3762,
      "step": 44026
    },
    {
      "epoch": 2.465393661104267,
      "grad_norm": 1.1961250305175781,
      "learning_rate": 9.36821052631579e-05,
      "loss": 0.334,
      "step": 44027
    },
    {
      "epoch": 2.465449658416396,
      "grad_norm": 1.1655254364013672,
      "learning_rate": 9.368184210526317e-05,
      "loss": 0.3819,
      "step": 44028
    },
    {
      "epoch": 2.465505655728525,
      "grad_norm": 1.359838843345642,
      "learning_rate": 9.368157894736843e-05,
      "loss": 0.3357,
      "step": 44029
    },
    {
      "epoch": 2.465561653040654,
      "grad_norm": 0.9917736053466797,
      "learning_rate": 9.368131578947369e-05,
      "loss": 0.3988,
      "step": 44030
    },
    {
      "epoch": 2.465617650352783,
      "grad_norm": 1.0773760080337524,
      "learning_rate": 9.368105263157895e-05,
      "loss": 0.3112,
      "step": 44031
    },
    {
      "epoch": 2.465673647664912,
      "grad_norm": 1.0271515846252441,
      "learning_rate": 9.368078947368422e-05,
      "loss": 0.3213,
      "step": 44032
    },
    {
      "epoch": 2.465729644977041,
      "grad_norm": 1.1921385526657104,
      "learning_rate": 9.368052631578948e-05,
      "loss": 0.2775,
      "step": 44033
    },
    {
      "epoch": 2.46578564228917,
      "grad_norm": 1.25236177444458,
      "learning_rate": 9.368026315789474e-05,
      "loss": 0.3778,
      "step": 44034
    },
    {
      "epoch": 2.465841639601299,
      "grad_norm": 1.499922275543213,
      "learning_rate": 9.368e-05,
      "loss": 0.6028,
      "step": 44035
    },
    {
      "epoch": 2.4658976369134282,
      "grad_norm": 1.3520774841308594,
      "learning_rate": 9.367973684210527e-05,
      "loss": 0.5467,
      "step": 44036
    },
    {
      "epoch": 2.4659536342255572,
      "grad_norm": 1.1964569091796875,
      "learning_rate": 9.367947368421053e-05,
      "loss": 0.5296,
      "step": 44037
    },
    {
      "epoch": 2.4660096315376863,
      "grad_norm": 1.1248208284378052,
      "learning_rate": 9.36792105263158e-05,
      "loss": 0.4742,
      "step": 44038
    },
    {
      "epoch": 2.4660656288498153,
      "grad_norm": 1.0359379053115845,
      "learning_rate": 9.367894736842105e-05,
      "loss": 0.3229,
      "step": 44039
    },
    {
      "epoch": 2.4661216261619443,
      "grad_norm": 1.373462200164795,
      "learning_rate": 9.367868421052632e-05,
      "loss": 0.4026,
      "step": 44040
    },
    {
      "epoch": 2.4661776234740733,
      "grad_norm": 1.202039361000061,
      "learning_rate": 9.367842105263158e-05,
      "loss": 0.4507,
      "step": 44041
    },
    {
      "epoch": 2.4662336207862023,
      "grad_norm": 1.3657056093215942,
      "learning_rate": 9.367815789473686e-05,
      "loss": 0.4638,
      "step": 44042
    },
    {
      "epoch": 2.4662896180983314,
      "grad_norm": 1.024149775505066,
      "learning_rate": 9.36778947368421e-05,
      "loss": 0.3326,
      "step": 44043
    },
    {
      "epoch": 2.4663456154104604,
      "grad_norm": 1.1358633041381836,
      "learning_rate": 9.367763157894738e-05,
      "loss": 0.4382,
      "step": 44044
    },
    {
      "epoch": 2.4664016127225894,
      "grad_norm": 1.0708404779434204,
      "learning_rate": 9.367736842105264e-05,
      "loss": 0.318,
      "step": 44045
    },
    {
      "epoch": 2.4664576100347184,
      "grad_norm": 1.3292200565338135,
      "learning_rate": 9.36771052631579e-05,
      "loss": 0.4684,
      "step": 44046
    },
    {
      "epoch": 2.4665136073468474,
      "grad_norm": 1.2605080604553223,
      "learning_rate": 9.367684210526317e-05,
      "loss": 0.4051,
      "step": 44047
    },
    {
      "epoch": 2.4665696046589765,
      "grad_norm": 1.0108362436294556,
      "learning_rate": 9.367657894736841e-05,
      "loss": 0.3772,
      "step": 44048
    },
    {
      "epoch": 2.4666256019711055,
      "grad_norm": 1.531085729598999,
      "learning_rate": 9.367631578947369e-05,
      "loss": 0.4447,
      "step": 44049
    },
    {
      "epoch": 2.4666815992832345,
      "grad_norm": 1.2041574716567993,
      "learning_rate": 9.367605263157895e-05,
      "loss": 0.4847,
      "step": 44050
    },
    {
      "epoch": 2.4667375965953635,
      "grad_norm": 1.3494386672973633,
      "learning_rate": 9.367578947368422e-05,
      "loss": 0.4032,
      "step": 44051
    },
    {
      "epoch": 2.4667935939074925,
      "grad_norm": 1.2513763904571533,
      "learning_rate": 9.367552631578948e-05,
      "loss": 0.5364,
      "step": 44052
    },
    {
      "epoch": 2.4668495912196216,
      "grad_norm": 1.1867985725402832,
      "learning_rate": 9.367526315789474e-05,
      "loss": 0.4274,
      "step": 44053
    },
    {
      "epoch": 2.4669055885317506,
      "grad_norm": 1.1427605152130127,
      "learning_rate": 9.3675e-05,
      "loss": 0.3654,
      "step": 44054
    },
    {
      "epoch": 2.4669615858438796,
      "grad_norm": 1.0306271314620972,
      "learning_rate": 9.367473684210527e-05,
      "loss": 0.3622,
      "step": 44055
    },
    {
      "epoch": 2.4670175831560086,
      "grad_norm": 1.0763897895812988,
      "learning_rate": 9.367447368421053e-05,
      "loss": 0.4505,
      "step": 44056
    },
    {
      "epoch": 2.4670735804681376,
      "grad_norm": 1.433589220046997,
      "learning_rate": 9.367421052631579e-05,
      "loss": 0.5643,
      "step": 44057
    },
    {
      "epoch": 2.4671295777802666,
      "grad_norm": 0.9540082812309265,
      "learning_rate": 9.367394736842105e-05,
      "loss": 0.311,
      "step": 44058
    },
    {
      "epoch": 2.4671855750923957,
      "grad_norm": 1.257386565208435,
      "learning_rate": 9.367368421052633e-05,
      "loss": 0.4465,
      "step": 44059
    },
    {
      "epoch": 2.4672415724045247,
      "grad_norm": 0.9749913811683655,
      "learning_rate": 9.367342105263159e-05,
      "loss": 0.396,
      "step": 44060
    },
    {
      "epoch": 2.4672975697166537,
      "grad_norm": 1.0280036926269531,
      "learning_rate": 9.367315789473685e-05,
      "loss": 0.3229,
      "step": 44061
    },
    {
      "epoch": 2.4673535670287827,
      "grad_norm": 1.1539928913116455,
      "learning_rate": 9.36728947368421e-05,
      "loss": 0.314,
      "step": 44062
    },
    {
      "epoch": 2.4674095643409117,
      "grad_norm": 0.958448052406311,
      "learning_rate": 9.367263157894736e-05,
      "loss": 0.3348,
      "step": 44063
    },
    {
      "epoch": 2.4674655616530408,
      "grad_norm": 1.329084038734436,
      "learning_rate": 9.367236842105264e-05,
      "loss": 0.4809,
      "step": 44064
    },
    {
      "epoch": 2.46752155896517,
      "grad_norm": 1.1161242723464966,
      "learning_rate": 9.36721052631579e-05,
      "loss": 0.4009,
      "step": 44065
    },
    {
      "epoch": 2.467577556277299,
      "grad_norm": 1.1887012720108032,
      "learning_rate": 9.367184210526316e-05,
      "loss": 0.4353,
      "step": 44066
    },
    {
      "epoch": 2.467633553589428,
      "grad_norm": 1.0540107488632202,
      "learning_rate": 9.367157894736842e-05,
      "loss": 0.3163,
      "step": 44067
    },
    {
      "epoch": 2.467689550901557,
      "grad_norm": 1.3905054330825806,
      "learning_rate": 9.367131578947369e-05,
      "loss": 0.5134,
      "step": 44068
    },
    {
      "epoch": 2.467745548213686,
      "grad_norm": 1.1003040075302124,
      "learning_rate": 9.367105263157895e-05,
      "loss": 0.3423,
      "step": 44069
    },
    {
      "epoch": 2.467801545525815,
      "grad_norm": 1.286909580230713,
      "learning_rate": 9.367078947368422e-05,
      "loss": 0.5266,
      "step": 44070
    },
    {
      "epoch": 2.467857542837944,
      "grad_norm": 1.0045418739318848,
      "learning_rate": 9.367052631578947e-05,
      "loss": 0.317,
      "step": 44071
    },
    {
      "epoch": 2.467913540150073,
      "grad_norm": 1.1579242944717407,
      "learning_rate": 9.367026315789474e-05,
      "loss": 0.4013,
      "step": 44072
    },
    {
      "epoch": 2.467969537462202,
      "grad_norm": 1.1702934503555298,
      "learning_rate": 9.367e-05,
      "loss": 0.3754,
      "step": 44073
    },
    {
      "epoch": 2.468025534774331,
      "grad_norm": 1.4780206680297852,
      "learning_rate": 9.366973684210528e-05,
      "loss": 0.4585,
      "step": 44074
    },
    {
      "epoch": 2.46808153208646,
      "grad_norm": 1.3680706024169922,
      "learning_rate": 9.366947368421054e-05,
      "loss": 0.4865,
      "step": 44075
    },
    {
      "epoch": 2.468137529398589,
      "grad_norm": 1.005650520324707,
      "learning_rate": 9.36692105263158e-05,
      "loss": 0.3618,
      "step": 44076
    },
    {
      "epoch": 2.468193526710718,
      "grad_norm": 1.2440506219863892,
      "learning_rate": 9.366894736842106e-05,
      "loss": 0.4509,
      "step": 44077
    },
    {
      "epoch": 2.468249524022847,
      "grad_norm": 3.961688995361328,
      "learning_rate": 9.366868421052633e-05,
      "loss": 0.4074,
      "step": 44078
    },
    {
      "epoch": 2.468305521334976,
      "grad_norm": 1.5190232992172241,
      "learning_rate": 9.366842105263159e-05,
      "loss": 0.4102,
      "step": 44079
    },
    {
      "epoch": 2.468361518647105,
      "grad_norm": 1.1501225233078003,
      "learning_rate": 9.366815789473685e-05,
      "loss": 0.3552,
      "step": 44080
    },
    {
      "epoch": 2.468417515959234,
      "grad_norm": 1.5113422870635986,
      "learning_rate": 9.366789473684211e-05,
      "loss": 0.5726,
      "step": 44081
    },
    {
      "epoch": 2.468473513271363,
      "grad_norm": 1.124323844909668,
      "learning_rate": 9.366763157894737e-05,
      "loss": 0.4517,
      "step": 44082
    },
    {
      "epoch": 2.468529510583492,
      "grad_norm": 1.2362663745880127,
      "learning_rate": 9.366736842105264e-05,
      "loss": 0.3874,
      "step": 44083
    },
    {
      "epoch": 2.468585507895621,
      "grad_norm": 1.3162689208984375,
      "learning_rate": 9.36671052631579e-05,
      "loss": 0.3758,
      "step": 44084
    },
    {
      "epoch": 2.46864150520775,
      "grad_norm": 1.3905032873153687,
      "learning_rate": 9.366684210526316e-05,
      "loss": 0.5052,
      "step": 44085
    },
    {
      "epoch": 2.468697502519879,
      "grad_norm": 1.4345086812973022,
      "learning_rate": 9.366657894736842e-05,
      "loss": 0.4463,
      "step": 44086
    },
    {
      "epoch": 2.468753499832008,
      "grad_norm": 1.2156461477279663,
      "learning_rate": 9.366631578947369e-05,
      "loss": 0.4629,
      "step": 44087
    },
    {
      "epoch": 2.468809497144137,
      "grad_norm": 1.0959806442260742,
      "learning_rate": 9.366605263157895e-05,
      "loss": 0.3785,
      "step": 44088
    },
    {
      "epoch": 2.4688654944562662,
      "grad_norm": 1.2783684730529785,
      "learning_rate": 9.366578947368421e-05,
      "loss": 0.4519,
      "step": 44089
    },
    {
      "epoch": 2.4689214917683953,
      "grad_norm": 1.615684986114502,
      "learning_rate": 9.366552631578947e-05,
      "loss": 0.3396,
      "step": 44090
    },
    {
      "epoch": 2.4689774890805243,
      "grad_norm": 1.2498834133148193,
      "learning_rate": 9.366526315789475e-05,
      "loss": 0.3973,
      "step": 44091
    },
    {
      "epoch": 2.4690334863926533,
      "grad_norm": 1.0827817916870117,
      "learning_rate": 9.3665e-05,
      "loss": 0.4252,
      "step": 44092
    },
    {
      "epoch": 2.4690894837047823,
      "grad_norm": 1.1794648170471191,
      "learning_rate": 9.366473684210527e-05,
      "loss": 0.3929,
      "step": 44093
    },
    {
      "epoch": 2.4691454810169113,
      "grad_norm": 1.0532640218734741,
      "learning_rate": 9.366447368421052e-05,
      "loss": 0.4309,
      "step": 44094
    },
    {
      "epoch": 2.4692014783290404,
      "grad_norm": 1.1812111139297485,
      "learning_rate": 9.36642105263158e-05,
      "loss": 0.399,
      "step": 44095
    },
    {
      "epoch": 2.4692574756411694,
      "grad_norm": 1.3980095386505127,
      "learning_rate": 9.366394736842106e-05,
      "loss": 0.4568,
      "step": 44096
    },
    {
      "epoch": 2.4693134729532984,
      "grad_norm": 0.947557270526886,
      "learning_rate": 9.366368421052633e-05,
      "loss": 0.3584,
      "step": 44097
    },
    {
      "epoch": 2.4693694702654274,
      "grad_norm": 1.2157189846038818,
      "learning_rate": 9.366342105263158e-05,
      "loss": 0.3158,
      "step": 44098
    },
    {
      "epoch": 2.4694254675775564,
      "grad_norm": 1.3564356565475464,
      "learning_rate": 9.366315789473684e-05,
      "loss": 0.4215,
      "step": 44099
    },
    {
      "epoch": 2.4694814648896855,
      "grad_norm": 1.1626747846603394,
      "learning_rate": 9.366289473684211e-05,
      "loss": 0.3075,
      "step": 44100
    },
    {
      "epoch": 2.4695374622018145,
      "grad_norm": 1.2534950971603394,
      "learning_rate": 9.366263157894737e-05,
      "loss": 0.3094,
      "step": 44101
    },
    {
      "epoch": 2.4695934595139435,
      "grad_norm": 1.0521891117095947,
      "learning_rate": 9.366236842105264e-05,
      "loss": 0.4207,
      "step": 44102
    },
    {
      "epoch": 2.4696494568260725,
      "grad_norm": 1.131402611732483,
      "learning_rate": 9.366210526315789e-05,
      "loss": 0.4155,
      "step": 44103
    },
    {
      "epoch": 2.4697054541382015,
      "grad_norm": 1.890564203262329,
      "learning_rate": 9.366184210526316e-05,
      "loss": 0.4706,
      "step": 44104
    },
    {
      "epoch": 2.4697614514503305,
      "grad_norm": 1.1770954132080078,
      "learning_rate": 9.366157894736842e-05,
      "loss": 0.4354,
      "step": 44105
    },
    {
      "epoch": 2.4698174487624596,
      "grad_norm": 1.0822374820709229,
      "learning_rate": 9.36613157894737e-05,
      "loss": 0.3286,
      "step": 44106
    },
    {
      "epoch": 2.4698734460745886,
      "grad_norm": 1.3864142894744873,
      "learning_rate": 9.366105263157896e-05,
      "loss": 0.3798,
      "step": 44107
    },
    {
      "epoch": 2.4699294433867176,
      "grad_norm": 1.3354579210281372,
      "learning_rate": 9.366078947368422e-05,
      "loss": 0.5007,
      "step": 44108
    },
    {
      "epoch": 2.4699854406988466,
      "grad_norm": 1.2717374563217163,
      "learning_rate": 9.366052631578947e-05,
      "loss": 0.4068,
      "step": 44109
    },
    {
      "epoch": 2.4700414380109756,
      "grad_norm": 0.9810419678688049,
      "learning_rate": 9.366026315789475e-05,
      "loss": 0.4736,
      "step": 44110
    },
    {
      "epoch": 2.4700974353231047,
      "grad_norm": 1.5323796272277832,
      "learning_rate": 9.366000000000001e-05,
      "loss": 0.4021,
      "step": 44111
    },
    {
      "epoch": 2.4701534326352337,
      "grad_norm": 1.0392820835113525,
      "learning_rate": 9.365973684210527e-05,
      "loss": 0.4277,
      "step": 44112
    },
    {
      "epoch": 2.4702094299473627,
      "grad_norm": 1.3933950662612915,
      "learning_rate": 9.365947368421053e-05,
      "loss": 0.3735,
      "step": 44113
    },
    {
      "epoch": 2.4702654272594917,
      "grad_norm": 1.2772082090377808,
      "learning_rate": 9.36592105263158e-05,
      "loss": 0.4558,
      "step": 44114
    },
    {
      "epoch": 2.4703214245716207,
      "grad_norm": 1.1120884418487549,
      "learning_rate": 9.365894736842106e-05,
      "loss": 0.3334,
      "step": 44115
    },
    {
      "epoch": 2.4703774218837498,
      "grad_norm": 1.3755912780761719,
      "learning_rate": 9.365868421052632e-05,
      "loss": 0.4307,
      "step": 44116
    },
    {
      "epoch": 2.4704334191958788,
      "grad_norm": 0.9504393935203552,
      "learning_rate": 9.365842105263158e-05,
      "loss": 0.3934,
      "step": 44117
    },
    {
      "epoch": 2.470489416508008,
      "grad_norm": 1.0620770454406738,
      "learning_rate": 9.365815789473684e-05,
      "loss": 0.3357,
      "step": 44118
    },
    {
      "epoch": 2.470545413820137,
      "grad_norm": 1.0509591102600098,
      "learning_rate": 9.365789473684211e-05,
      "loss": 0.3773,
      "step": 44119
    },
    {
      "epoch": 2.470601411132266,
      "grad_norm": 1.3002393245697021,
      "learning_rate": 9.365763157894737e-05,
      "loss": 0.5345,
      "step": 44120
    },
    {
      "epoch": 2.470657408444395,
      "grad_norm": 1.1038538217544556,
      "learning_rate": 9.365736842105263e-05,
      "loss": 0.3716,
      "step": 44121
    },
    {
      "epoch": 2.470713405756524,
      "grad_norm": 1.2519797086715698,
      "learning_rate": 9.365710526315789e-05,
      "loss": 0.4457,
      "step": 44122
    },
    {
      "epoch": 2.470769403068653,
      "grad_norm": 1.2866604328155518,
      "learning_rate": 9.365684210526317e-05,
      "loss": 0.3703,
      "step": 44123
    },
    {
      "epoch": 2.470825400380782,
      "grad_norm": 1.1615679264068604,
      "learning_rate": 9.365657894736843e-05,
      "loss": 0.3675,
      "step": 44124
    },
    {
      "epoch": 2.470881397692911,
      "grad_norm": 1.277629017829895,
      "learning_rate": 9.36563157894737e-05,
      "loss": 0.3952,
      "step": 44125
    },
    {
      "epoch": 2.47093739500504,
      "grad_norm": 0.9527827501296997,
      "learning_rate": 9.365605263157894e-05,
      "loss": 0.3516,
      "step": 44126
    },
    {
      "epoch": 2.470993392317169,
      "grad_norm": 1.086543321609497,
      "learning_rate": 9.365578947368422e-05,
      "loss": 0.4292,
      "step": 44127
    },
    {
      "epoch": 2.471049389629298,
      "grad_norm": 1.107340931892395,
      "learning_rate": 9.365552631578948e-05,
      "loss": 0.3998,
      "step": 44128
    },
    {
      "epoch": 2.471105386941427,
      "grad_norm": 1.5575259923934937,
      "learning_rate": 9.365526315789475e-05,
      "loss": 0.4101,
      "step": 44129
    },
    {
      "epoch": 2.4711613842535556,
      "grad_norm": 1.187996745109558,
      "learning_rate": 9.365500000000001e-05,
      "loss": 0.3771,
      "step": 44130
    },
    {
      "epoch": 2.471217381565685,
      "grad_norm": 1.0467430353164673,
      "learning_rate": 9.365473684210527e-05,
      "loss": 0.2957,
      "step": 44131
    },
    {
      "epoch": 2.4712733788778136,
      "grad_norm": 1.0744919776916504,
      "learning_rate": 9.365447368421053e-05,
      "loss": 0.3675,
      "step": 44132
    },
    {
      "epoch": 2.471329376189943,
      "grad_norm": 1.1523994207382202,
      "learning_rate": 9.365421052631579e-05,
      "loss": 0.4402,
      "step": 44133
    },
    {
      "epoch": 2.4713853735020717,
      "grad_norm": 1.3272831439971924,
      "learning_rate": 9.365394736842106e-05,
      "loss": 0.3924,
      "step": 44134
    },
    {
      "epoch": 2.471441370814201,
      "grad_norm": 1.1133606433868408,
      "learning_rate": 9.365368421052631e-05,
      "loss": 0.3858,
      "step": 44135
    },
    {
      "epoch": 2.4714973681263297,
      "grad_norm": 1.1648343801498413,
      "learning_rate": 9.365342105263158e-05,
      "loss": 0.5602,
      "step": 44136
    },
    {
      "epoch": 2.471553365438459,
      "grad_norm": 1.038198709487915,
      "learning_rate": 9.365315789473684e-05,
      "loss": 0.2836,
      "step": 44137
    },
    {
      "epoch": 2.4716093627505877,
      "grad_norm": 1.1249364614486694,
      "learning_rate": 9.365289473684212e-05,
      "loss": 0.3407,
      "step": 44138
    },
    {
      "epoch": 2.471665360062717,
      "grad_norm": 1.039136290550232,
      "learning_rate": 9.365263157894738e-05,
      "loss": 0.3837,
      "step": 44139
    },
    {
      "epoch": 2.4717213573748458,
      "grad_norm": 1.1571974754333496,
      "learning_rate": 9.365236842105263e-05,
      "loss": 0.3395,
      "step": 44140
    },
    {
      "epoch": 2.4717773546869752,
      "grad_norm": 1.267731785774231,
      "learning_rate": 9.36521052631579e-05,
      "loss": 0.4256,
      "step": 44141
    },
    {
      "epoch": 2.471833351999104,
      "grad_norm": 1.6461541652679443,
      "learning_rate": 9.365184210526317e-05,
      "loss": 0.4764,
      "step": 44142
    },
    {
      "epoch": 2.4718893493112333,
      "grad_norm": 1.1013351678848267,
      "learning_rate": 9.365157894736843e-05,
      "loss": 0.3444,
      "step": 44143
    },
    {
      "epoch": 2.471945346623362,
      "grad_norm": 1.2162506580352783,
      "learning_rate": 9.365131578947369e-05,
      "loss": 0.3778,
      "step": 44144
    },
    {
      "epoch": 2.4720013439354913,
      "grad_norm": 1.0435081720352173,
      "learning_rate": 9.365105263157895e-05,
      "loss": 0.5001,
      "step": 44145
    },
    {
      "epoch": 2.47205734124762,
      "grad_norm": 1.201692819595337,
      "learning_rate": 9.365078947368422e-05,
      "loss": 0.3647,
      "step": 44146
    },
    {
      "epoch": 2.4721133385597494,
      "grad_norm": 1.0861254930496216,
      "learning_rate": 9.365052631578948e-05,
      "loss": 0.3028,
      "step": 44147
    },
    {
      "epoch": 2.472169335871878,
      "grad_norm": 1.2334767580032349,
      "learning_rate": 9.365026315789474e-05,
      "loss": 0.4915,
      "step": 44148
    },
    {
      "epoch": 2.4722253331840074,
      "grad_norm": 1.0025677680969238,
      "learning_rate": 9.365e-05,
      "loss": 0.3859,
      "step": 44149
    },
    {
      "epoch": 2.472281330496136,
      "grad_norm": 1.5126653909683228,
      "learning_rate": 9.364973684210526e-05,
      "loss": 0.4494,
      "step": 44150
    },
    {
      "epoch": 2.4723373278082654,
      "grad_norm": 1.2949076890945435,
      "learning_rate": 9.364947368421053e-05,
      "loss": 0.3585,
      "step": 44151
    },
    {
      "epoch": 2.472393325120394,
      "grad_norm": 1.5175106525421143,
      "learning_rate": 9.364921052631579e-05,
      "loss": 0.5013,
      "step": 44152
    },
    {
      "epoch": 2.4724493224325235,
      "grad_norm": 1.2476967573165894,
      "learning_rate": 9.364894736842105e-05,
      "loss": 0.2908,
      "step": 44153
    },
    {
      "epoch": 2.472505319744652,
      "grad_norm": 1.2612708806991577,
      "learning_rate": 9.364868421052631e-05,
      "loss": 0.5451,
      "step": 44154
    },
    {
      "epoch": 2.4725613170567815,
      "grad_norm": 1.6813753843307495,
      "learning_rate": 9.364842105263158e-05,
      "loss": 0.3972,
      "step": 44155
    },
    {
      "epoch": 2.47261731436891,
      "grad_norm": 1.061497449874878,
      "learning_rate": 9.364815789473684e-05,
      "loss": 0.4859,
      "step": 44156
    },
    {
      "epoch": 2.4726733116810395,
      "grad_norm": 1.2612429857254028,
      "learning_rate": 9.364789473684212e-05,
      "loss": 0.3339,
      "step": 44157
    },
    {
      "epoch": 2.472729308993168,
      "grad_norm": 1.1108336448669434,
      "learning_rate": 9.364763157894736e-05,
      "loss": 0.3639,
      "step": 44158
    },
    {
      "epoch": 2.4727853063052976,
      "grad_norm": 1.0925554037094116,
      "learning_rate": 9.364736842105264e-05,
      "loss": 0.3867,
      "step": 44159
    },
    {
      "epoch": 2.472841303617426,
      "grad_norm": 1.3325393199920654,
      "learning_rate": 9.36471052631579e-05,
      "loss": 0.3766,
      "step": 44160
    },
    {
      "epoch": 2.4728973009295556,
      "grad_norm": 1.1267825365066528,
      "learning_rate": 9.364684210526317e-05,
      "loss": 0.3133,
      "step": 44161
    },
    {
      "epoch": 2.472953298241684,
      "grad_norm": 1.1658260822296143,
      "learning_rate": 9.364657894736843e-05,
      "loss": 0.4255,
      "step": 44162
    },
    {
      "epoch": 2.4730092955538137,
      "grad_norm": 1.2096806764602661,
      "learning_rate": 9.364631578947369e-05,
      "loss": 0.3835,
      "step": 44163
    },
    {
      "epoch": 2.4730652928659422,
      "grad_norm": 0.9385897517204285,
      "learning_rate": 9.364605263157895e-05,
      "loss": 0.4209,
      "step": 44164
    },
    {
      "epoch": 2.4731212901780717,
      "grad_norm": 0.9884217381477356,
      "learning_rate": 9.364578947368422e-05,
      "loss": 0.3795,
      "step": 44165
    },
    {
      "epoch": 2.4731772874902003,
      "grad_norm": 1.325321078300476,
      "learning_rate": 9.364552631578948e-05,
      "loss": 0.5624,
      "step": 44166
    },
    {
      "epoch": 2.4732332848023297,
      "grad_norm": 1.0410025119781494,
      "learning_rate": 9.364526315789474e-05,
      "loss": 0.367,
      "step": 44167
    },
    {
      "epoch": 2.4732892821144583,
      "grad_norm": 1.3467849493026733,
      "learning_rate": 9.3645e-05,
      "loss": 0.3696,
      "step": 44168
    },
    {
      "epoch": 2.4733452794265878,
      "grad_norm": 1.2354958057403564,
      "learning_rate": 9.364473684210526e-05,
      "loss": 0.4545,
      "step": 44169
    },
    {
      "epoch": 2.4734012767387163,
      "grad_norm": 1.1850742101669312,
      "learning_rate": 9.364447368421054e-05,
      "loss": 0.4972,
      "step": 44170
    },
    {
      "epoch": 2.4734572740508454,
      "grad_norm": 1.09269380569458,
      "learning_rate": 9.36442105263158e-05,
      "loss": 0.306,
      "step": 44171
    },
    {
      "epoch": 2.4735132713629744,
      "grad_norm": 1.181783676147461,
      "learning_rate": 9.364394736842105e-05,
      "loss": 0.3975,
      "step": 44172
    },
    {
      "epoch": 2.4735692686751034,
      "grad_norm": 1.132027506828308,
      "learning_rate": 9.364368421052631e-05,
      "loss": 0.4291,
      "step": 44173
    },
    {
      "epoch": 2.4736252659872324,
      "grad_norm": 1.5134623050689697,
      "learning_rate": 9.364342105263159e-05,
      "loss": 0.4832,
      "step": 44174
    },
    {
      "epoch": 2.4736812632993614,
      "grad_norm": 1.2445316314697266,
      "learning_rate": 9.364315789473685e-05,
      "loss": 0.3755,
      "step": 44175
    },
    {
      "epoch": 2.4737372606114905,
      "grad_norm": 1.2487881183624268,
      "learning_rate": 9.364289473684211e-05,
      "loss": 0.3557,
      "step": 44176
    },
    {
      "epoch": 2.4737932579236195,
      "grad_norm": 1.0553139448165894,
      "learning_rate": 9.364263157894737e-05,
      "loss": 0.4807,
      "step": 44177
    },
    {
      "epoch": 2.4738492552357485,
      "grad_norm": 1.2687548398971558,
      "learning_rate": 9.364236842105264e-05,
      "loss": 0.3559,
      "step": 44178
    },
    {
      "epoch": 2.4739052525478775,
      "grad_norm": 1.063984990119934,
      "learning_rate": 9.36421052631579e-05,
      "loss": 0.4786,
      "step": 44179
    },
    {
      "epoch": 2.4739612498600065,
      "grad_norm": 1.2033805847167969,
      "learning_rate": 9.364184210526317e-05,
      "loss": 0.3296,
      "step": 44180
    },
    {
      "epoch": 2.4740172471721356,
      "grad_norm": 0.958592414855957,
      "learning_rate": 9.364157894736842e-05,
      "loss": 0.3607,
      "step": 44181
    },
    {
      "epoch": 2.4740732444842646,
      "grad_norm": 1.4361246824264526,
      "learning_rate": 9.364131578947369e-05,
      "loss": 0.3993,
      "step": 44182
    },
    {
      "epoch": 2.4741292417963936,
      "grad_norm": 1.2696774005889893,
      "learning_rate": 9.364105263157895e-05,
      "loss": 0.4031,
      "step": 44183
    },
    {
      "epoch": 2.4741852391085226,
      "grad_norm": 1.1495460271835327,
      "learning_rate": 9.364078947368421e-05,
      "loss": 0.3673,
      "step": 44184
    },
    {
      "epoch": 2.4742412364206516,
      "grad_norm": 1.0882701873779297,
      "learning_rate": 9.364052631578949e-05,
      "loss": 0.2845,
      "step": 44185
    },
    {
      "epoch": 2.4742972337327807,
      "grad_norm": 1.030644178390503,
      "learning_rate": 9.364026315789473e-05,
      "loss": 0.3428,
      "step": 44186
    },
    {
      "epoch": 2.4743532310449097,
      "grad_norm": 1.1731072664260864,
      "learning_rate": 9.364e-05,
      "loss": 0.4346,
      "step": 44187
    },
    {
      "epoch": 2.4744092283570387,
      "grad_norm": 1.0636916160583496,
      "learning_rate": 9.363973684210526e-05,
      "loss": 0.4252,
      "step": 44188
    },
    {
      "epoch": 2.4744652256691677,
      "grad_norm": 1.3042874336242676,
      "learning_rate": 9.363947368421054e-05,
      "loss": 0.4403,
      "step": 44189
    },
    {
      "epoch": 2.4745212229812967,
      "grad_norm": 1.261055827140808,
      "learning_rate": 9.363921052631578e-05,
      "loss": 0.3018,
      "step": 44190
    },
    {
      "epoch": 2.4745772202934257,
      "grad_norm": 1.2624691724777222,
      "learning_rate": 9.363894736842106e-05,
      "loss": 0.3558,
      "step": 44191
    },
    {
      "epoch": 2.4746332176055548,
      "grad_norm": 1.0622167587280273,
      "learning_rate": 9.363868421052632e-05,
      "loss": 0.3872,
      "step": 44192
    },
    {
      "epoch": 2.474689214917684,
      "grad_norm": 1.302704095840454,
      "learning_rate": 9.363842105263159e-05,
      "loss": 0.34,
      "step": 44193
    },
    {
      "epoch": 2.474745212229813,
      "grad_norm": 1.1235733032226562,
      "learning_rate": 9.363815789473685e-05,
      "loss": 0.4324,
      "step": 44194
    },
    {
      "epoch": 2.474801209541942,
      "grad_norm": 1.168811559677124,
      "learning_rate": 9.363789473684211e-05,
      "loss": 0.381,
      "step": 44195
    },
    {
      "epoch": 2.474857206854071,
      "grad_norm": 1.1778302192687988,
      "learning_rate": 9.363763157894737e-05,
      "loss": 0.4337,
      "step": 44196
    },
    {
      "epoch": 2.4749132041662,
      "grad_norm": 1.262765884399414,
      "learning_rate": 9.363736842105264e-05,
      "loss": 0.4214,
      "step": 44197
    },
    {
      "epoch": 2.474969201478329,
      "grad_norm": 1.1716381311416626,
      "learning_rate": 9.36371052631579e-05,
      "loss": 0.357,
      "step": 44198
    },
    {
      "epoch": 2.475025198790458,
      "grad_norm": 1.2705764770507812,
      "learning_rate": 9.363684210526316e-05,
      "loss": 0.3563,
      "step": 44199
    },
    {
      "epoch": 2.475081196102587,
      "grad_norm": 1.1583493947982788,
      "learning_rate": 9.363657894736842e-05,
      "loss": 0.4596,
      "step": 44200
    },
    {
      "epoch": 2.475137193414716,
      "grad_norm": 1.0806114673614502,
      "learning_rate": 9.36363157894737e-05,
      "loss": 0.373,
      "step": 44201
    },
    {
      "epoch": 2.475193190726845,
      "grad_norm": 1.2327076196670532,
      "learning_rate": 9.363605263157895e-05,
      "loss": 0.3775,
      "step": 44202
    },
    {
      "epoch": 2.475249188038974,
      "grad_norm": 1.133095145225525,
      "learning_rate": 9.363578947368421e-05,
      "loss": 0.3428,
      "step": 44203
    },
    {
      "epoch": 2.475305185351103,
      "grad_norm": 1.2337861061096191,
      "learning_rate": 9.363552631578947e-05,
      "loss": 0.4825,
      "step": 44204
    },
    {
      "epoch": 2.475361182663232,
      "grad_norm": 1.3687037229537964,
      "learning_rate": 9.363526315789473e-05,
      "loss": 0.4871,
      "step": 44205
    },
    {
      "epoch": 2.475417179975361,
      "grad_norm": 1.0706144571304321,
      "learning_rate": 9.363500000000001e-05,
      "loss": 0.3895,
      "step": 44206
    },
    {
      "epoch": 2.47547317728749,
      "grad_norm": 1.185217261314392,
      "learning_rate": 9.363473684210527e-05,
      "loss": 0.3375,
      "step": 44207
    },
    {
      "epoch": 2.475529174599619,
      "grad_norm": 1.0831892490386963,
      "learning_rate": 9.363447368421053e-05,
      "loss": 0.4064,
      "step": 44208
    },
    {
      "epoch": 2.475585171911748,
      "grad_norm": 1.218835473060608,
      "learning_rate": 9.363421052631579e-05,
      "loss": 0.3733,
      "step": 44209
    },
    {
      "epoch": 2.475641169223877,
      "grad_norm": 1.0665627717971802,
      "learning_rate": 9.363394736842106e-05,
      "loss": 0.4019,
      "step": 44210
    },
    {
      "epoch": 2.475697166536006,
      "grad_norm": 1.2511937618255615,
      "learning_rate": 9.363368421052632e-05,
      "loss": 0.5153,
      "step": 44211
    },
    {
      "epoch": 2.475753163848135,
      "grad_norm": 1.4110581874847412,
      "learning_rate": 9.363342105263159e-05,
      "loss": 0.3775,
      "step": 44212
    },
    {
      "epoch": 2.475809161160264,
      "grad_norm": 1.0463111400604248,
      "learning_rate": 9.363315789473684e-05,
      "loss": 0.3598,
      "step": 44213
    },
    {
      "epoch": 2.475865158472393,
      "grad_norm": 1.0675930976867676,
      "learning_rate": 9.363289473684211e-05,
      "loss": 0.4516,
      "step": 44214
    },
    {
      "epoch": 2.475921155784522,
      "grad_norm": 1.0578423738479614,
      "learning_rate": 9.363263157894737e-05,
      "loss": 0.3859,
      "step": 44215
    },
    {
      "epoch": 2.4759771530966512,
      "grad_norm": 1.3587926626205444,
      "learning_rate": 9.363236842105265e-05,
      "loss": 0.43,
      "step": 44216
    },
    {
      "epoch": 2.4760331504087802,
      "grad_norm": 1.0513360500335693,
      "learning_rate": 9.36321052631579e-05,
      "loss": 0.286,
      "step": 44217
    },
    {
      "epoch": 2.4760891477209093,
      "grad_norm": 1.070373296737671,
      "learning_rate": 9.363184210526316e-05,
      "loss": 0.3816,
      "step": 44218
    },
    {
      "epoch": 2.4761451450330383,
      "grad_norm": 2.3888747692108154,
      "learning_rate": 9.363157894736842e-05,
      "loss": 0.4975,
      "step": 44219
    },
    {
      "epoch": 2.4762011423451673,
      "grad_norm": 1.155505895614624,
      "learning_rate": 9.363131578947368e-05,
      "loss": 0.3417,
      "step": 44220
    },
    {
      "epoch": 2.4762571396572963,
      "grad_norm": 1.20297110080719,
      "learning_rate": 9.363105263157896e-05,
      "loss": 0.4159,
      "step": 44221
    },
    {
      "epoch": 2.4763131369694253,
      "grad_norm": 1.0364887714385986,
      "learning_rate": 9.363078947368422e-05,
      "loss": 0.4141,
      "step": 44222
    },
    {
      "epoch": 2.4763691342815544,
      "grad_norm": 1.1430013179779053,
      "learning_rate": 9.363052631578948e-05,
      "loss": 0.3749,
      "step": 44223
    },
    {
      "epoch": 2.4764251315936834,
      "grad_norm": 1.2423121929168701,
      "learning_rate": 9.363026315789474e-05,
      "loss": 0.387,
      "step": 44224
    },
    {
      "epoch": 2.4764811289058124,
      "grad_norm": 1.3047313690185547,
      "learning_rate": 9.363000000000001e-05,
      "loss": 0.5771,
      "step": 44225
    },
    {
      "epoch": 2.4765371262179414,
      "grad_norm": 1.0119483470916748,
      "learning_rate": 9.362973684210527e-05,
      "loss": 0.3837,
      "step": 44226
    },
    {
      "epoch": 2.4765931235300704,
      "grad_norm": 1.500756859779358,
      "learning_rate": 9.362947368421053e-05,
      "loss": 0.4075,
      "step": 44227
    },
    {
      "epoch": 2.4766491208421995,
      "grad_norm": 1.291379690170288,
      "learning_rate": 9.362921052631579e-05,
      "loss": 0.3999,
      "step": 44228
    },
    {
      "epoch": 2.4767051181543285,
      "grad_norm": 2.538275718688965,
      "learning_rate": 9.362894736842106e-05,
      "loss": 0.5308,
      "step": 44229
    },
    {
      "epoch": 2.4767611154664575,
      "grad_norm": 1.202265739440918,
      "learning_rate": 9.362868421052632e-05,
      "loss": 0.5129,
      "step": 44230
    },
    {
      "epoch": 2.4768171127785865,
      "grad_norm": 1.124449610710144,
      "learning_rate": 9.362842105263158e-05,
      "loss": 0.3865,
      "step": 44231
    },
    {
      "epoch": 2.4768731100907155,
      "grad_norm": 6.918768405914307,
      "learning_rate": 9.362815789473684e-05,
      "loss": 0.3774,
      "step": 44232
    },
    {
      "epoch": 2.4769291074028446,
      "grad_norm": 0.9559996724128723,
      "learning_rate": 9.362789473684211e-05,
      "loss": 0.2914,
      "step": 44233
    },
    {
      "epoch": 2.4769851047149736,
      "grad_norm": 1.4014638662338257,
      "learning_rate": 9.362763157894737e-05,
      "loss": 0.3846,
      "step": 44234
    },
    {
      "epoch": 2.4770411020271026,
      "grad_norm": 1.219598650932312,
      "learning_rate": 9.362736842105265e-05,
      "loss": 0.5325,
      "step": 44235
    },
    {
      "epoch": 2.4770970993392316,
      "grad_norm": 1.401444673538208,
      "learning_rate": 9.36271052631579e-05,
      "loss": 0.3956,
      "step": 44236
    },
    {
      "epoch": 2.4771530966513606,
      "grad_norm": 1.232041835784912,
      "learning_rate": 9.362684210526315e-05,
      "loss": 0.4369,
      "step": 44237
    },
    {
      "epoch": 2.4772090939634897,
      "grad_norm": 1.0963729619979858,
      "learning_rate": 9.362657894736843e-05,
      "loss": 0.4629,
      "step": 44238
    },
    {
      "epoch": 2.4772650912756187,
      "grad_norm": 1.120522141456604,
      "learning_rate": 9.362631578947369e-05,
      "loss": 0.3754,
      "step": 44239
    },
    {
      "epoch": 2.4773210885877477,
      "grad_norm": 1.3960210084915161,
      "learning_rate": 9.362605263157896e-05,
      "loss": 0.5175,
      "step": 44240
    },
    {
      "epoch": 2.4773770858998767,
      "grad_norm": 1.3212816715240479,
      "learning_rate": 9.36257894736842e-05,
      "loss": 0.3626,
      "step": 44241
    },
    {
      "epoch": 2.4774330832120057,
      "grad_norm": 1.3703598976135254,
      "learning_rate": 9.362552631578948e-05,
      "loss": 0.349,
      "step": 44242
    },
    {
      "epoch": 2.4774890805241347,
      "grad_norm": 1.1253827810287476,
      "learning_rate": 9.362526315789474e-05,
      "loss": 0.2938,
      "step": 44243
    },
    {
      "epoch": 2.4775450778362638,
      "grad_norm": 1.2143992185592651,
      "learning_rate": 9.362500000000001e-05,
      "loss": 0.4504,
      "step": 44244
    },
    {
      "epoch": 2.477601075148393,
      "grad_norm": 1.4462101459503174,
      "learning_rate": 9.362473684210526e-05,
      "loss": 0.4675,
      "step": 44245
    },
    {
      "epoch": 2.477657072460522,
      "grad_norm": 1.0151829719543457,
      "learning_rate": 9.362447368421053e-05,
      "loss": 0.2997,
      "step": 44246
    },
    {
      "epoch": 2.477713069772651,
      "grad_norm": 1.3268916606903076,
      "learning_rate": 9.362421052631579e-05,
      "loss": 0.5082,
      "step": 44247
    },
    {
      "epoch": 2.47776906708478,
      "grad_norm": 1.262639045715332,
      "learning_rate": 9.362394736842106e-05,
      "loss": 0.4398,
      "step": 44248
    },
    {
      "epoch": 2.477825064396909,
      "grad_norm": 1.350880742073059,
      "learning_rate": 9.362368421052632e-05,
      "loss": 0.4232,
      "step": 44249
    },
    {
      "epoch": 2.477881061709038,
      "grad_norm": 1.3605902194976807,
      "learning_rate": 9.362342105263158e-05,
      "loss": 0.5403,
      "step": 44250
    },
    {
      "epoch": 2.477937059021167,
      "grad_norm": 1.1158623695373535,
      "learning_rate": 9.362315789473684e-05,
      "loss": 0.4137,
      "step": 44251
    },
    {
      "epoch": 2.477993056333296,
      "grad_norm": 1.388620376586914,
      "learning_rate": 9.362289473684212e-05,
      "loss": 0.4514,
      "step": 44252
    },
    {
      "epoch": 2.478049053645425,
      "grad_norm": 1.1969324350357056,
      "learning_rate": 9.362263157894738e-05,
      "loss": 0.4217,
      "step": 44253
    },
    {
      "epoch": 2.478105050957554,
      "grad_norm": 1.4513752460479736,
      "learning_rate": 9.362236842105264e-05,
      "loss": 0.486,
      "step": 44254
    },
    {
      "epoch": 2.478161048269683,
      "grad_norm": 1.3357194662094116,
      "learning_rate": 9.36221052631579e-05,
      "loss": 0.5557,
      "step": 44255
    },
    {
      "epoch": 2.478217045581812,
      "grad_norm": 1.2365554571151733,
      "learning_rate": 9.362184210526316e-05,
      "loss": 0.3582,
      "step": 44256
    },
    {
      "epoch": 2.478273042893941,
      "grad_norm": 1.4263801574707031,
      "learning_rate": 9.362157894736843e-05,
      "loss": 0.4174,
      "step": 44257
    },
    {
      "epoch": 2.47832904020607,
      "grad_norm": 1.1818263530731201,
      "learning_rate": 9.362131578947369e-05,
      "loss": 0.4068,
      "step": 44258
    },
    {
      "epoch": 2.478385037518199,
      "grad_norm": 1.1868363618850708,
      "learning_rate": 9.362105263157895e-05,
      "loss": 0.4086,
      "step": 44259
    },
    {
      "epoch": 2.478441034830328,
      "grad_norm": 1.3890413045883179,
      "learning_rate": 9.362078947368421e-05,
      "loss": 0.3857,
      "step": 44260
    },
    {
      "epoch": 2.478497032142457,
      "grad_norm": 1.1044574975967407,
      "learning_rate": 9.362052631578948e-05,
      "loss": 0.3654,
      "step": 44261
    },
    {
      "epoch": 2.478553029454586,
      "grad_norm": 1.3509091138839722,
      "learning_rate": 9.362026315789474e-05,
      "loss": 0.394,
      "step": 44262
    },
    {
      "epoch": 2.478609026766715,
      "grad_norm": 1.177695631980896,
      "learning_rate": 9.362e-05,
      "loss": 0.3574,
      "step": 44263
    },
    {
      "epoch": 2.478665024078844,
      "grad_norm": 1.0882335901260376,
      "learning_rate": 9.361973684210526e-05,
      "loss": 0.4333,
      "step": 44264
    },
    {
      "epoch": 2.478721021390973,
      "grad_norm": 1.234005093574524,
      "learning_rate": 9.361947368421053e-05,
      "loss": 0.4111,
      "step": 44265
    },
    {
      "epoch": 2.478777018703102,
      "grad_norm": 1.3114949464797974,
      "learning_rate": 9.36192105263158e-05,
      "loss": 0.546,
      "step": 44266
    },
    {
      "epoch": 2.478833016015231,
      "grad_norm": 1.1203795671463013,
      "learning_rate": 9.361894736842107e-05,
      "loss": 0.4014,
      "step": 44267
    },
    {
      "epoch": 2.4788890133273602,
      "grad_norm": 1.273461103439331,
      "learning_rate": 9.361868421052631e-05,
      "loss": 0.3747,
      "step": 44268
    },
    {
      "epoch": 2.4789450106394892,
      "grad_norm": 1.461968183517456,
      "learning_rate": 9.361842105263159e-05,
      "loss": 0.4591,
      "step": 44269
    },
    {
      "epoch": 2.4790010079516183,
      "grad_norm": 1.5604476928710938,
      "learning_rate": 9.361815789473685e-05,
      "loss": 0.5595,
      "step": 44270
    },
    {
      "epoch": 2.4790570052637473,
      "grad_norm": 1.204005479812622,
      "learning_rate": 9.36178947368421e-05,
      "loss": 0.6587,
      "step": 44271
    },
    {
      "epoch": 2.4791130025758763,
      "grad_norm": 1.3592901229858398,
      "learning_rate": 9.361763157894738e-05,
      "loss": 0.4579,
      "step": 44272
    },
    {
      "epoch": 2.4791689998880053,
      "grad_norm": 0.9730402827262878,
      "learning_rate": 9.361736842105263e-05,
      "loss": 0.44,
      "step": 44273
    },
    {
      "epoch": 2.4792249972001343,
      "grad_norm": 1.2401785850524902,
      "learning_rate": 9.36171052631579e-05,
      "loss": 0.4138,
      "step": 44274
    },
    {
      "epoch": 2.4792809945122634,
      "grad_norm": 1.1003730297088623,
      "learning_rate": 9.361684210526316e-05,
      "loss": 0.4546,
      "step": 44275
    },
    {
      "epoch": 2.4793369918243924,
      "grad_norm": 1.1668202877044678,
      "learning_rate": 9.361657894736843e-05,
      "loss": 0.3441,
      "step": 44276
    },
    {
      "epoch": 2.4793929891365214,
      "grad_norm": 1.307818055152893,
      "learning_rate": 9.361631578947369e-05,
      "loss": 0.3807,
      "step": 44277
    },
    {
      "epoch": 2.4794489864486504,
      "grad_norm": 1.1633470058441162,
      "learning_rate": 9.361605263157895e-05,
      "loss": 0.3264,
      "step": 44278
    },
    {
      "epoch": 2.4795049837607794,
      "grad_norm": 1.1844751834869385,
      "learning_rate": 9.361578947368421e-05,
      "loss": 0.399,
      "step": 44279
    },
    {
      "epoch": 2.4795609810729085,
      "grad_norm": 1.1914105415344238,
      "learning_rate": 9.361552631578948e-05,
      "loss": 0.4115,
      "step": 44280
    },
    {
      "epoch": 2.4796169783850375,
      "grad_norm": 1.1621674299240112,
      "learning_rate": 9.361526315789474e-05,
      "loss": 0.4789,
      "step": 44281
    },
    {
      "epoch": 2.4796729756971665,
      "grad_norm": 1.2318912744522095,
      "learning_rate": 9.3615e-05,
      "loss": 0.3727,
      "step": 44282
    },
    {
      "epoch": 2.4797289730092955,
      "grad_norm": 1.42116117477417,
      "learning_rate": 9.361473684210526e-05,
      "loss": 0.382,
      "step": 44283
    },
    {
      "epoch": 2.4797849703214245,
      "grad_norm": 1.2754536867141724,
      "learning_rate": 9.361447368421054e-05,
      "loss": 0.3931,
      "step": 44284
    },
    {
      "epoch": 2.4798409676335536,
      "grad_norm": 1.2895548343658447,
      "learning_rate": 9.36142105263158e-05,
      "loss": 0.4431,
      "step": 44285
    },
    {
      "epoch": 2.4798969649456826,
      "grad_norm": 1.221282958984375,
      "learning_rate": 9.361394736842106e-05,
      "loss": 0.5005,
      "step": 44286
    },
    {
      "epoch": 2.4799529622578116,
      "grad_norm": 1.1840317249298096,
      "learning_rate": 9.361368421052632e-05,
      "loss": 0.3752,
      "step": 44287
    },
    {
      "epoch": 2.4800089595699406,
      "grad_norm": 1.2022030353546143,
      "learning_rate": 9.361342105263158e-05,
      "loss": 0.3922,
      "step": 44288
    },
    {
      "epoch": 2.4800649568820696,
      "grad_norm": 1.2402735948562622,
      "learning_rate": 9.361315789473685e-05,
      "loss": 0.4245,
      "step": 44289
    },
    {
      "epoch": 2.4801209541941986,
      "grad_norm": 1.180904746055603,
      "learning_rate": 9.361289473684211e-05,
      "loss": 0.3804,
      "step": 44290
    },
    {
      "epoch": 2.4801769515063277,
      "grad_norm": 1.1709727048873901,
      "learning_rate": 9.361263157894737e-05,
      "loss": 0.3607,
      "step": 44291
    },
    {
      "epoch": 2.4802329488184567,
      "grad_norm": 1.1259634494781494,
      "learning_rate": 9.361236842105263e-05,
      "loss": 0.3738,
      "step": 44292
    },
    {
      "epoch": 2.4802889461305857,
      "grad_norm": 1.236454725265503,
      "learning_rate": 9.36121052631579e-05,
      "loss": 0.4646,
      "step": 44293
    },
    {
      "epoch": 2.4803449434427147,
      "grad_norm": 1.212053656578064,
      "learning_rate": 9.361184210526316e-05,
      "loss": 0.4888,
      "step": 44294
    },
    {
      "epoch": 2.4804009407548437,
      "grad_norm": 1.0567636489868164,
      "learning_rate": 9.361157894736842e-05,
      "loss": 0.4504,
      "step": 44295
    },
    {
      "epoch": 2.4804569380669728,
      "grad_norm": 1.4017983675003052,
      "learning_rate": 9.361131578947368e-05,
      "loss": 0.3912,
      "step": 44296
    },
    {
      "epoch": 2.480512935379102,
      "grad_norm": 1.2582666873931885,
      "learning_rate": 9.361105263157895e-05,
      "loss": 0.4469,
      "step": 44297
    },
    {
      "epoch": 2.480568932691231,
      "grad_norm": 1.0491713285446167,
      "learning_rate": 9.361078947368421e-05,
      "loss": 0.3459,
      "step": 44298
    },
    {
      "epoch": 2.48062493000336,
      "grad_norm": 1.0057142972946167,
      "learning_rate": 9.361052631578949e-05,
      "loss": 0.3196,
      "step": 44299
    },
    {
      "epoch": 2.480680927315489,
      "grad_norm": 1.0935282707214355,
      "learning_rate": 9.361026315789473e-05,
      "loss": 0.333,
      "step": 44300
    },
    {
      "epoch": 2.480736924627618,
      "grad_norm": 1.250519871711731,
      "learning_rate": 9.361e-05,
      "loss": 0.3471,
      "step": 44301
    },
    {
      "epoch": 2.480792921939747,
      "grad_norm": 1.3426438570022583,
      "learning_rate": 9.360973684210527e-05,
      "loss": 0.448,
      "step": 44302
    },
    {
      "epoch": 2.480848919251876,
      "grad_norm": 1.2694522142410278,
      "learning_rate": 9.360947368421054e-05,
      "loss": 0.4841,
      "step": 44303
    },
    {
      "epoch": 2.480904916564005,
      "grad_norm": 1.098730444908142,
      "learning_rate": 9.36092105263158e-05,
      "loss": 0.5077,
      "step": 44304
    },
    {
      "epoch": 2.480960913876134,
      "grad_norm": 1.1663686037063599,
      "learning_rate": 9.360894736842105e-05,
      "loss": 0.4174,
      "step": 44305
    },
    {
      "epoch": 2.481016911188263,
      "grad_norm": 1.2516649961471558,
      "learning_rate": 9.360868421052632e-05,
      "loss": 0.3566,
      "step": 44306
    },
    {
      "epoch": 2.481072908500392,
      "grad_norm": 1.1248210668563843,
      "learning_rate": 9.360842105263158e-05,
      "loss": 0.3192,
      "step": 44307
    },
    {
      "epoch": 2.481128905812521,
      "grad_norm": 2.4669246673583984,
      "learning_rate": 9.360815789473685e-05,
      "loss": 0.3732,
      "step": 44308
    },
    {
      "epoch": 2.48118490312465,
      "grad_norm": 1.19463050365448,
      "learning_rate": 9.360789473684211e-05,
      "loss": 0.3554,
      "step": 44309
    },
    {
      "epoch": 2.481240900436779,
      "grad_norm": 1.0442301034927368,
      "learning_rate": 9.360763157894737e-05,
      "loss": 0.3112,
      "step": 44310
    },
    {
      "epoch": 2.481296897748908,
      "grad_norm": 3.028778553009033,
      "learning_rate": 9.360736842105263e-05,
      "loss": 0.4936,
      "step": 44311
    },
    {
      "epoch": 2.481352895061037,
      "grad_norm": 1.1400574445724487,
      "learning_rate": 9.36071052631579e-05,
      "loss": 0.488,
      "step": 44312
    },
    {
      "epoch": 2.481408892373166,
      "grad_norm": 1.0928895473480225,
      "learning_rate": 9.360684210526316e-05,
      "loss": 0.3176,
      "step": 44313
    },
    {
      "epoch": 2.481464889685295,
      "grad_norm": 1.1874134540557861,
      "learning_rate": 9.360657894736842e-05,
      "loss": 0.3583,
      "step": 44314
    },
    {
      "epoch": 2.481520886997424,
      "grad_norm": 1.341428518295288,
      "learning_rate": 9.360631578947368e-05,
      "loss": 0.577,
      "step": 44315
    },
    {
      "epoch": 2.481576884309553,
      "grad_norm": 1.1175403594970703,
      "learning_rate": 9.360605263157896e-05,
      "loss": 0.3954,
      "step": 44316
    },
    {
      "epoch": 2.481632881621682,
      "grad_norm": 1.0590373277664185,
      "learning_rate": 9.360578947368422e-05,
      "loss": 0.3935,
      "step": 44317
    },
    {
      "epoch": 2.481688878933811,
      "grad_norm": 1.0659778118133545,
      "learning_rate": 9.360552631578948e-05,
      "loss": 0.3345,
      "step": 44318
    },
    {
      "epoch": 2.48174487624594,
      "grad_norm": 0.9667007327079773,
      "learning_rate": 9.360526315789474e-05,
      "loss": 0.375,
      "step": 44319
    },
    {
      "epoch": 2.481800873558069,
      "grad_norm": 1.5114572048187256,
      "learning_rate": 9.360500000000001e-05,
      "loss": 0.6055,
      "step": 44320
    },
    {
      "epoch": 2.4818568708701982,
      "grad_norm": 1.123246669769287,
      "learning_rate": 9.360473684210527e-05,
      "loss": 0.3809,
      "step": 44321
    },
    {
      "epoch": 2.4819128681823273,
      "grad_norm": 1.472427248954773,
      "learning_rate": 9.360447368421054e-05,
      "loss": 0.4051,
      "step": 44322
    },
    {
      "epoch": 2.4819688654944563,
      "grad_norm": 1.1469322443008423,
      "learning_rate": 9.360421052631579e-05,
      "loss": 0.3426,
      "step": 44323
    },
    {
      "epoch": 2.4820248628065853,
      "grad_norm": 1.1927857398986816,
      "learning_rate": 9.360394736842105e-05,
      "loss": 0.5035,
      "step": 44324
    },
    {
      "epoch": 2.4820808601187143,
      "grad_norm": 1.2909719944000244,
      "learning_rate": 9.360368421052632e-05,
      "loss": 0.4026,
      "step": 44325
    },
    {
      "epoch": 2.4821368574308433,
      "grad_norm": 1.3235340118408203,
      "learning_rate": 9.360342105263158e-05,
      "loss": 0.5212,
      "step": 44326
    },
    {
      "epoch": 2.4821928547429724,
      "grad_norm": 1.2182832956314087,
      "learning_rate": 9.360315789473685e-05,
      "loss": 0.5021,
      "step": 44327
    },
    {
      "epoch": 2.4822488520551014,
      "grad_norm": 1.1385419368743896,
      "learning_rate": 9.36028947368421e-05,
      "loss": 0.3244,
      "step": 44328
    },
    {
      "epoch": 2.4823048493672304,
      "grad_norm": 1.1761189699172974,
      "learning_rate": 9.360263157894737e-05,
      "loss": 0.4057,
      "step": 44329
    },
    {
      "epoch": 2.4823608466793594,
      "grad_norm": 1.1188102960586548,
      "learning_rate": 9.360236842105263e-05,
      "loss": 0.3156,
      "step": 44330
    },
    {
      "epoch": 2.4824168439914884,
      "grad_norm": 1.6709442138671875,
      "learning_rate": 9.360210526315791e-05,
      "loss": 0.3776,
      "step": 44331
    },
    {
      "epoch": 2.4824728413036175,
      "grad_norm": 1.4009681940078735,
      "learning_rate": 9.360184210526317e-05,
      "loss": 0.4567,
      "step": 44332
    },
    {
      "epoch": 2.4825288386157465,
      "grad_norm": 1.3581562042236328,
      "learning_rate": 9.360157894736843e-05,
      "loss": 0.3986,
      "step": 44333
    },
    {
      "epoch": 2.4825848359278755,
      "grad_norm": 1.220747947692871,
      "learning_rate": 9.360131578947369e-05,
      "loss": 0.4543,
      "step": 44334
    },
    {
      "epoch": 2.4826408332400045,
      "grad_norm": 1.236267328262329,
      "learning_rate": 9.360105263157896e-05,
      "loss": 0.4253,
      "step": 44335
    },
    {
      "epoch": 2.4826968305521335,
      "grad_norm": 1.0665589570999146,
      "learning_rate": 9.360078947368422e-05,
      "loss": 0.3402,
      "step": 44336
    },
    {
      "epoch": 2.4827528278642625,
      "grad_norm": 1.1624139547348022,
      "learning_rate": 9.360052631578948e-05,
      "loss": 0.3634,
      "step": 44337
    },
    {
      "epoch": 2.4828088251763916,
      "grad_norm": 1.1849192380905151,
      "learning_rate": 9.360026315789474e-05,
      "loss": 0.3726,
      "step": 44338
    },
    {
      "epoch": 2.4828648224885206,
      "grad_norm": 1.114926815032959,
      "learning_rate": 9.360000000000001e-05,
      "loss": 0.2879,
      "step": 44339
    },
    {
      "epoch": 2.4829208198006496,
      "grad_norm": 1.1868789196014404,
      "learning_rate": 9.359973684210527e-05,
      "loss": 0.4337,
      "step": 44340
    },
    {
      "epoch": 2.4829768171127786,
      "grad_norm": 2.4211010932922363,
      "learning_rate": 9.359947368421053e-05,
      "loss": 0.4841,
      "step": 44341
    },
    {
      "epoch": 2.4830328144249076,
      "grad_norm": 1.0909616947174072,
      "learning_rate": 9.359921052631579e-05,
      "loss": 0.3437,
      "step": 44342
    },
    {
      "epoch": 2.4830888117370367,
      "grad_norm": 1.056530475616455,
      "learning_rate": 9.359894736842105e-05,
      "loss": 0.3347,
      "step": 44343
    },
    {
      "epoch": 2.4831448090491657,
      "grad_norm": 1.4640955924987793,
      "learning_rate": 9.359868421052632e-05,
      "loss": 0.4662,
      "step": 44344
    },
    {
      "epoch": 2.4832008063612947,
      "grad_norm": 1.1382948160171509,
      "learning_rate": 9.359842105263158e-05,
      "loss": 0.4165,
      "step": 44345
    },
    {
      "epoch": 2.4832568036734237,
      "grad_norm": 2.338960647583008,
      "learning_rate": 9.359815789473684e-05,
      "loss": 0.3593,
      "step": 44346
    },
    {
      "epoch": 2.4833128009855527,
      "grad_norm": 1.2627880573272705,
      "learning_rate": 9.35978947368421e-05,
      "loss": 0.4196,
      "step": 44347
    },
    {
      "epoch": 2.4833687982976818,
      "grad_norm": 1.4457350969314575,
      "learning_rate": 9.359763157894738e-05,
      "loss": 0.4471,
      "step": 44348
    },
    {
      "epoch": 2.4834247956098108,
      "grad_norm": 1.1647441387176514,
      "learning_rate": 9.359736842105264e-05,
      "loss": 0.3835,
      "step": 44349
    },
    {
      "epoch": 2.48348079292194,
      "grad_norm": 1.1431159973144531,
      "learning_rate": 9.35971052631579e-05,
      "loss": 0.3496,
      "step": 44350
    },
    {
      "epoch": 2.483536790234069,
      "grad_norm": 1.9090896844863892,
      "learning_rate": 9.359684210526316e-05,
      "loss": 0.3803,
      "step": 44351
    },
    {
      "epoch": 2.483592787546198,
      "grad_norm": 1.4625990390777588,
      "learning_rate": 9.359657894736843e-05,
      "loss": 0.3953,
      "step": 44352
    },
    {
      "epoch": 2.483648784858327,
      "grad_norm": 1.1794495582580566,
      "learning_rate": 9.359631578947369e-05,
      "loss": 0.4426,
      "step": 44353
    },
    {
      "epoch": 2.483704782170456,
      "grad_norm": 1.1374478340148926,
      "learning_rate": 9.359605263157896e-05,
      "loss": 0.3319,
      "step": 44354
    },
    {
      "epoch": 2.483760779482585,
      "grad_norm": 1.2206623554229736,
      "learning_rate": 9.359578947368421e-05,
      "loss": 0.309,
      "step": 44355
    },
    {
      "epoch": 2.483816776794714,
      "grad_norm": 1.0533288717269897,
      "learning_rate": 9.359552631578948e-05,
      "loss": 0.3853,
      "step": 44356
    },
    {
      "epoch": 2.483872774106843,
      "grad_norm": 1.0481966733932495,
      "learning_rate": 9.359526315789474e-05,
      "loss": 0.3109,
      "step": 44357
    },
    {
      "epoch": 2.483928771418972,
      "grad_norm": 1.1098259687423706,
      "learning_rate": 9.3595e-05,
      "loss": 0.3584,
      "step": 44358
    },
    {
      "epoch": 2.483984768731101,
      "grad_norm": 1.2437944412231445,
      "learning_rate": 9.359473684210527e-05,
      "loss": 0.4526,
      "step": 44359
    },
    {
      "epoch": 2.48404076604323,
      "grad_norm": 1.125532627105713,
      "learning_rate": 9.359447368421052e-05,
      "loss": 0.3306,
      "step": 44360
    },
    {
      "epoch": 2.484096763355359,
      "grad_norm": 1.1381503343582153,
      "learning_rate": 9.35942105263158e-05,
      "loss": 0.342,
      "step": 44361
    },
    {
      "epoch": 2.484152760667488,
      "grad_norm": 1.5417871475219727,
      "learning_rate": 9.359394736842105e-05,
      "loss": 0.3742,
      "step": 44362
    },
    {
      "epoch": 2.484208757979617,
      "grad_norm": 1.2284289598464966,
      "learning_rate": 9.359368421052633e-05,
      "loss": 0.4808,
      "step": 44363
    },
    {
      "epoch": 2.484264755291746,
      "grad_norm": 1.0769321918487549,
      "learning_rate": 9.359342105263159e-05,
      "loss": 0.405,
      "step": 44364
    },
    {
      "epoch": 2.484320752603875,
      "grad_norm": 1.147834300994873,
      "learning_rate": 9.359315789473685e-05,
      "loss": 0.3571,
      "step": 44365
    },
    {
      "epoch": 2.484376749916004,
      "grad_norm": 1.1369510889053345,
      "learning_rate": 9.35928947368421e-05,
      "loss": 0.5292,
      "step": 44366
    },
    {
      "epoch": 2.484432747228133,
      "grad_norm": 1.2867659330368042,
      "learning_rate": 9.359263157894738e-05,
      "loss": 0.4032,
      "step": 44367
    },
    {
      "epoch": 2.484488744540262,
      "grad_norm": 1.1284868717193604,
      "learning_rate": 9.359236842105264e-05,
      "loss": 0.3291,
      "step": 44368
    },
    {
      "epoch": 2.484544741852391,
      "grad_norm": 1.228381633758545,
      "learning_rate": 9.35921052631579e-05,
      "loss": 0.4156,
      "step": 44369
    },
    {
      "epoch": 2.48460073916452,
      "grad_norm": 1.2863619327545166,
      "learning_rate": 9.359184210526316e-05,
      "loss": 0.4195,
      "step": 44370
    },
    {
      "epoch": 2.484656736476649,
      "grad_norm": 1.2525149583816528,
      "learning_rate": 9.359157894736843e-05,
      "loss": 0.5231,
      "step": 44371
    },
    {
      "epoch": 2.484712733788778,
      "grad_norm": 1.1156443357467651,
      "learning_rate": 9.359131578947369e-05,
      "loss": 0.4538,
      "step": 44372
    },
    {
      "epoch": 2.4847687311009072,
      "grad_norm": 1.1146354675292969,
      "learning_rate": 9.359105263157895e-05,
      "loss": 0.3729,
      "step": 44373
    },
    {
      "epoch": 2.4848247284130363,
      "grad_norm": 1.3225160837173462,
      "learning_rate": 9.359078947368421e-05,
      "loss": 0.4583,
      "step": 44374
    },
    {
      "epoch": 2.4848807257251653,
      "grad_norm": 1.6498847007751465,
      "learning_rate": 9.359052631578947e-05,
      "loss": 0.4204,
      "step": 44375
    },
    {
      "epoch": 2.4849367230372943,
      "grad_norm": 1.392510175704956,
      "learning_rate": 9.359026315789474e-05,
      "loss": 0.355,
      "step": 44376
    },
    {
      "epoch": 2.4849927203494233,
      "grad_norm": 44.35410690307617,
      "learning_rate": 9.359e-05,
      "loss": 0.4076,
      "step": 44377
    },
    {
      "epoch": 2.4850487176615523,
      "grad_norm": 1.1210507154464722,
      "learning_rate": 9.358973684210526e-05,
      "loss": 0.3591,
      "step": 44378
    },
    {
      "epoch": 2.4851047149736814,
      "grad_norm": 1.152748942375183,
      "learning_rate": 9.358947368421052e-05,
      "loss": 0.4241,
      "step": 44379
    },
    {
      "epoch": 2.4851607122858104,
      "grad_norm": 1.199645757675171,
      "learning_rate": 9.35892105263158e-05,
      "loss": 0.5081,
      "step": 44380
    },
    {
      "epoch": 2.4852167095979394,
      "grad_norm": 1.1402279138565063,
      "learning_rate": 9.358894736842106e-05,
      "loss": 0.4445,
      "step": 44381
    },
    {
      "epoch": 2.4852727069100684,
      "grad_norm": 1.1898512840270996,
      "learning_rate": 9.358868421052633e-05,
      "loss": 0.4172,
      "step": 44382
    },
    {
      "epoch": 2.4853287042221974,
      "grad_norm": 1.2633826732635498,
      "learning_rate": 9.358842105263158e-05,
      "loss": 0.3905,
      "step": 44383
    },
    {
      "epoch": 2.4853847015343264,
      "grad_norm": 1.1433947086334229,
      "learning_rate": 9.358815789473685e-05,
      "loss": 0.4654,
      "step": 44384
    },
    {
      "epoch": 2.4854406988464555,
      "grad_norm": 1.388482689857483,
      "learning_rate": 9.358789473684211e-05,
      "loss": 0.473,
      "step": 44385
    },
    {
      "epoch": 2.4854966961585845,
      "grad_norm": 1.2269740104675293,
      "learning_rate": 9.358763157894738e-05,
      "loss": 0.3988,
      "step": 44386
    },
    {
      "epoch": 2.4855526934707135,
      "grad_norm": 1.3058655261993408,
      "learning_rate": 9.358736842105264e-05,
      "loss": 0.4723,
      "step": 44387
    },
    {
      "epoch": 2.4856086907828425,
      "grad_norm": 1.2200459241867065,
      "learning_rate": 9.35871052631579e-05,
      "loss": 0.3827,
      "step": 44388
    },
    {
      "epoch": 2.4856646880949715,
      "grad_norm": 1.1663531064987183,
      "learning_rate": 9.358684210526316e-05,
      "loss": 0.4147,
      "step": 44389
    },
    {
      "epoch": 2.4857206854071006,
      "grad_norm": 1.1792305707931519,
      "learning_rate": 9.358657894736843e-05,
      "loss": 0.3393,
      "step": 44390
    },
    {
      "epoch": 2.4857766827192296,
      "grad_norm": 1.4000658988952637,
      "learning_rate": 9.35863157894737e-05,
      "loss": 0.4596,
      "step": 44391
    },
    {
      "epoch": 2.4858326800313586,
      "grad_norm": 1.0567255020141602,
      "learning_rate": 9.358605263157894e-05,
      "loss": 0.3018,
      "step": 44392
    },
    {
      "epoch": 2.4858886773434876,
      "grad_norm": 1.1478970050811768,
      "learning_rate": 9.358578947368421e-05,
      "loss": 0.5257,
      "step": 44393
    },
    {
      "epoch": 2.4859446746556166,
      "grad_norm": 1.1812740564346313,
      "learning_rate": 9.358552631578947e-05,
      "loss": 0.4569,
      "step": 44394
    },
    {
      "epoch": 2.4860006719677457,
      "grad_norm": 1.1951205730438232,
      "learning_rate": 9.358526315789475e-05,
      "loss": 0.4418,
      "step": 44395
    },
    {
      "epoch": 2.4860566692798747,
      "grad_norm": 1.0341664552688599,
      "learning_rate": 9.3585e-05,
      "loss": 0.4201,
      "step": 44396
    },
    {
      "epoch": 2.4861126665920037,
      "grad_norm": 1.0356801748275757,
      "learning_rate": 9.358473684210527e-05,
      "loss": 0.429,
      "step": 44397
    },
    {
      "epoch": 2.4861686639041327,
      "grad_norm": 1.1959925889968872,
      "learning_rate": 9.358447368421053e-05,
      "loss": 0.3728,
      "step": 44398
    },
    {
      "epoch": 2.4862246612162617,
      "grad_norm": 1.2789123058319092,
      "learning_rate": 9.35842105263158e-05,
      "loss": 0.3894,
      "step": 44399
    },
    {
      "epoch": 2.4862806585283908,
      "grad_norm": 1.1927591562271118,
      "learning_rate": 9.358394736842106e-05,
      "loss": 0.3223,
      "step": 44400
    },
    {
      "epoch": 2.4863366558405198,
      "grad_norm": 1.1825623512268066,
      "learning_rate": 9.358368421052632e-05,
      "loss": 0.4201,
      "step": 44401
    },
    {
      "epoch": 2.486392653152649,
      "grad_norm": 1.1208559274673462,
      "learning_rate": 9.358342105263158e-05,
      "loss": 0.4757,
      "step": 44402
    },
    {
      "epoch": 2.486448650464778,
      "grad_norm": 1.0040638446807861,
      "learning_rate": 9.358315789473685e-05,
      "loss": 0.4577,
      "step": 44403
    },
    {
      "epoch": 2.486504647776907,
      "grad_norm": 1.1077594757080078,
      "learning_rate": 9.358289473684211e-05,
      "loss": 0.4893,
      "step": 44404
    },
    {
      "epoch": 2.486560645089036,
      "grad_norm": 1.1230523586273193,
      "learning_rate": 9.358263157894737e-05,
      "loss": 0.3368,
      "step": 44405
    },
    {
      "epoch": 2.486616642401165,
      "grad_norm": 1.061773419380188,
      "learning_rate": 9.358236842105263e-05,
      "loss": 0.4284,
      "step": 44406
    },
    {
      "epoch": 2.486672639713294,
      "grad_norm": 1.078866720199585,
      "learning_rate": 9.35821052631579e-05,
      "loss": 0.3328,
      "step": 44407
    },
    {
      "epoch": 2.486728637025423,
      "grad_norm": 1.2729166746139526,
      "learning_rate": 9.358184210526316e-05,
      "loss": 0.4318,
      "step": 44408
    },
    {
      "epoch": 2.486784634337552,
      "grad_norm": 1.183813452720642,
      "learning_rate": 9.358157894736842e-05,
      "loss": 0.4281,
      "step": 44409
    },
    {
      "epoch": 2.486840631649681,
      "grad_norm": 1.2210932970046997,
      "learning_rate": 9.358131578947368e-05,
      "loss": 0.4753,
      "step": 44410
    },
    {
      "epoch": 2.48689662896181,
      "grad_norm": 1.0906797647476196,
      "learning_rate": 9.358105263157894e-05,
      "loss": 0.3994,
      "step": 44411
    },
    {
      "epoch": 2.486952626273939,
      "grad_norm": 1.1253408193588257,
      "learning_rate": 9.358078947368422e-05,
      "loss": 0.2988,
      "step": 44412
    },
    {
      "epoch": 2.487008623586068,
      "grad_norm": 1.3081265687942505,
      "learning_rate": 9.358052631578948e-05,
      "loss": 0.3908,
      "step": 44413
    },
    {
      "epoch": 2.487064620898197,
      "grad_norm": 1.0889675617218018,
      "learning_rate": 9.358026315789475e-05,
      "loss": 0.413,
      "step": 44414
    },
    {
      "epoch": 2.487120618210326,
      "grad_norm": 1.3809175491333008,
      "learning_rate": 9.358e-05,
      "loss": 0.45,
      "step": 44415
    },
    {
      "epoch": 2.487176615522455,
      "grad_norm": 1.1167020797729492,
      "learning_rate": 9.357973684210527e-05,
      "loss": 0.3413,
      "step": 44416
    },
    {
      "epoch": 2.487232612834584,
      "grad_norm": 1.3193563222885132,
      "learning_rate": 9.357947368421053e-05,
      "loss": 0.4888,
      "step": 44417
    },
    {
      "epoch": 2.487288610146713,
      "grad_norm": 1.1543149948120117,
      "learning_rate": 9.35792105263158e-05,
      "loss": 0.3694,
      "step": 44418
    },
    {
      "epoch": 2.487344607458842,
      "grad_norm": 1.077239990234375,
      "learning_rate": 9.357894736842106e-05,
      "loss": 0.3586,
      "step": 44419
    },
    {
      "epoch": 2.487400604770971,
      "grad_norm": 1.0811024904251099,
      "learning_rate": 9.357868421052632e-05,
      "loss": 0.3176,
      "step": 44420
    },
    {
      "epoch": 2.4874566020831,
      "grad_norm": 1.4588935375213623,
      "learning_rate": 9.357842105263158e-05,
      "loss": 0.4131,
      "step": 44421
    },
    {
      "epoch": 2.487512599395229,
      "grad_norm": 1.4578564167022705,
      "learning_rate": 9.357815789473685e-05,
      "loss": 0.4867,
      "step": 44422
    },
    {
      "epoch": 2.487568596707358,
      "grad_norm": 1.0522583723068237,
      "learning_rate": 9.357789473684211e-05,
      "loss": 0.3762,
      "step": 44423
    },
    {
      "epoch": 2.487624594019487,
      "grad_norm": 1.0071390867233276,
      "learning_rate": 9.357763157894737e-05,
      "loss": 0.2678,
      "step": 44424
    },
    {
      "epoch": 2.4876805913316162,
      "grad_norm": 1.1334043741226196,
      "learning_rate": 9.357736842105263e-05,
      "loss": 0.3766,
      "step": 44425
    },
    {
      "epoch": 2.4877365886437453,
      "grad_norm": 1.4347904920578003,
      "learning_rate": 9.357710526315789e-05,
      "loss": 0.4893,
      "step": 44426
    },
    {
      "epoch": 2.4877925859558743,
      "grad_norm": 1.2108877897262573,
      "learning_rate": 9.357684210526317e-05,
      "loss": 0.4045,
      "step": 44427
    },
    {
      "epoch": 2.4878485832680033,
      "grad_norm": 1.4925814867019653,
      "learning_rate": 9.357657894736843e-05,
      "loss": 0.3663,
      "step": 44428
    },
    {
      "epoch": 2.4879045805801323,
      "grad_norm": 1.2763339281082153,
      "learning_rate": 9.357631578947369e-05,
      "loss": 0.4363,
      "step": 44429
    },
    {
      "epoch": 2.4879605778922613,
      "grad_norm": 1.0145639181137085,
      "learning_rate": 9.357605263157895e-05,
      "loss": 0.4099,
      "step": 44430
    },
    {
      "epoch": 2.4880165752043903,
      "grad_norm": 1.1439982652664185,
      "learning_rate": 9.357578947368422e-05,
      "loss": 0.445,
      "step": 44431
    },
    {
      "epoch": 2.4880725725165194,
      "grad_norm": 1.1238441467285156,
      "learning_rate": 9.357552631578948e-05,
      "loss": 0.4545,
      "step": 44432
    },
    {
      "epoch": 2.4881285698286484,
      "grad_norm": 1.0151135921478271,
      "learning_rate": 9.357526315789474e-05,
      "loss": 0.4035,
      "step": 44433
    },
    {
      "epoch": 2.4881845671407774,
      "grad_norm": 1.2869915962219238,
      "learning_rate": 9.3575e-05,
      "loss": 0.6252,
      "step": 44434
    },
    {
      "epoch": 2.4882405644529064,
      "grad_norm": 1.1926716566085815,
      "learning_rate": 9.357473684210527e-05,
      "loss": 0.4034,
      "step": 44435
    },
    {
      "epoch": 2.4882965617650354,
      "grad_norm": 1.498978853225708,
      "learning_rate": 9.357447368421053e-05,
      "loss": 0.462,
      "step": 44436
    },
    {
      "epoch": 2.4883525590771645,
      "grad_norm": 1.2526978254318237,
      "learning_rate": 9.35742105263158e-05,
      "loss": 0.4297,
      "step": 44437
    },
    {
      "epoch": 2.4884085563892935,
      "grad_norm": 1.225989580154419,
      "learning_rate": 9.357394736842105e-05,
      "loss": 0.4025,
      "step": 44438
    },
    {
      "epoch": 2.4884645537014225,
      "grad_norm": 1.126609206199646,
      "learning_rate": 9.357368421052632e-05,
      "loss": 0.4147,
      "step": 44439
    },
    {
      "epoch": 2.4885205510135515,
      "grad_norm": 1.2837029695510864,
      "learning_rate": 9.357342105263158e-05,
      "loss": 0.4329,
      "step": 44440
    },
    {
      "epoch": 2.4885765483256805,
      "grad_norm": 1.1106743812561035,
      "learning_rate": 9.357315789473686e-05,
      "loss": 0.4723,
      "step": 44441
    },
    {
      "epoch": 2.4886325456378096,
      "grad_norm": 1.2484773397445679,
      "learning_rate": 9.35728947368421e-05,
      "loss": 0.386,
      "step": 44442
    },
    {
      "epoch": 2.4886885429499386,
      "grad_norm": 1.164198398590088,
      "learning_rate": 9.357263157894738e-05,
      "loss": 0.4549,
      "step": 44443
    },
    {
      "epoch": 2.4887445402620676,
      "grad_norm": 1.0683376789093018,
      "learning_rate": 9.357236842105264e-05,
      "loss": 0.4195,
      "step": 44444
    },
    {
      "epoch": 2.4888005375741966,
      "grad_norm": 1.1411572694778442,
      "learning_rate": 9.35721052631579e-05,
      "loss": 0.3504,
      "step": 44445
    },
    {
      "epoch": 2.4888565348863256,
      "grad_norm": 1.2010785341262817,
      "learning_rate": 9.357184210526317e-05,
      "loss": 0.444,
      "step": 44446
    },
    {
      "epoch": 2.4889125321984547,
      "grad_norm": 1.1552280187606812,
      "learning_rate": 9.357157894736841e-05,
      "loss": 0.4045,
      "step": 44447
    },
    {
      "epoch": 2.4889685295105837,
      "grad_norm": 1.1287347078323364,
      "learning_rate": 9.357131578947369e-05,
      "loss": 0.4258,
      "step": 44448
    },
    {
      "epoch": 2.4890245268227127,
      "grad_norm": 1.2057421207427979,
      "learning_rate": 9.357105263157895e-05,
      "loss": 0.3446,
      "step": 44449
    },
    {
      "epoch": 2.4890805241348417,
      "grad_norm": 1.0982046127319336,
      "learning_rate": 9.357078947368422e-05,
      "loss": 0.2673,
      "step": 44450
    },
    {
      "epoch": 2.4891365214469707,
      "grad_norm": 1.7271478176116943,
      "learning_rate": 9.357052631578948e-05,
      "loss": 0.4236,
      "step": 44451
    },
    {
      "epoch": 2.4891925187590997,
      "grad_norm": 1.3551247119903564,
      "learning_rate": 9.357026315789474e-05,
      "loss": 0.4806,
      "step": 44452
    },
    {
      "epoch": 2.4892485160712288,
      "grad_norm": 1.125567078590393,
      "learning_rate": 9.357e-05,
      "loss": 0.357,
      "step": 44453
    },
    {
      "epoch": 2.489304513383358,
      "grad_norm": 1.27196204662323,
      "learning_rate": 9.356973684210527e-05,
      "loss": 0.4471,
      "step": 44454
    },
    {
      "epoch": 2.489360510695487,
      "grad_norm": 1.3074246644973755,
      "learning_rate": 9.356947368421053e-05,
      "loss": 0.3845,
      "step": 44455
    },
    {
      "epoch": 2.489416508007616,
      "grad_norm": 1.0610613822937012,
      "learning_rate": 9.356921052631579e-05,
      "loss": 0.322,
      "step": 44456
    },
    {
      "epoch": 2.489472505319745,
      "grad_norm": 1.4347412586212158,
      "learning_rate": 9.356894736842105e-05,
      "loss": 0.4696,
      "step": 44457
    },
    {
      "epoch": 2.489528502631874,
      "grad_norm": 1.164192795753479,
      "learning_rate": 9.356868421052633e-05,
      "loss": 0.5126,
      "step": 44458
    },
    {
      "epoch": 2.4895844999440024,
      "grad_norm": 1.386047601699829,
      "learning_rate": 9.356842105263159e-05,
      "loss": 0.4049,
      "step": 44459
    },
    {
      "epoch": 2.489640497256132,
      "grad_norm": 1.360996127128601,
      "learning_rate": 9.356815789473685e-05,
      "loss": 0.3766,
      "step": 44460
    },
    {
      "epoch": 2.4896964945682605,
      "grad_norm": 0.997880220413208,
      "learning_rate": 9.35678947368421e-05,
      "loss": 0.4326,
      "step": 44461
    },
    {
      "epoch": 2.48975249188039,
      "grad_norm": 1.3211970329284668,
      "learning_rate": 9.356763157894736e-05,
      "loss": 0.4903,
      "step": 44462
    },
    {
      "epoch": 2.4898084891925185,
      "grad_norm": 1.4319353103637695,
      "learning_rate": 9.356736842105264e-05,
      "loss": 0.4173,
      "step": 44463
    },
    {
      "epoch": 2.489864486504648,
      "grad_norm": 0.9493651390075684,
      "learning_rate": 9.35671052631579e-05,
      "loss": 0.3284,
      "step": 44464
    },
    {
      "epoch": 2.4899204838167766,
      "grad_norm": 1.0870492458343506,
      "learning_rate": 9.356684210526316e-05,
      "loss": 0.4141,
      "step": 44465
    },
    {
      "epoch": 2.489976481128906,
      "grad_norm": 1.049599051475525,
      "learning_rate": 9.356657894736842e-05,
      "loss": 0.3829,
      "step": 44466
    },
    {
      "epoch": 2.4900324784410346,
      "grad_norm": 1.1602377891540527,
      "learning_rate": 9.356631578947369e-05,
      "loss": 0.3894,
      "step": 44467
    },
    {
      "epoch": 2.490088475753164,
      "grad_norm": 1.1313836574554443,
      "learning_rate": 9.356605263157895e-05,
      "loss": 0.3796,
      "step": 44468
    },
    {
      "epoch": 2.4901444730652926,
      "grad_norm": 1.1797488927841187,
      "learning_rate": 9.356578947368422e-05,
      "loss": 0.33,
      "step": 44469
    },
    {
      "epoch": 2.490200470377422,
      "grad_norm": 1.3110485076904297,
      "learning_rate": 9.356552631578947e-05,
      "loss": 0.3458,
      "step": 44470
    },
    {
      "epoch": 2.4902564676895507,
      "grad_norm": 1.1925069093704224,
      "learning_rate": 9.356526315789474e-05,
      "loss": 0.4582,
      "step": 44471
    },
    {
      "epoch": 2.49031246500168,
      "grad_norm": 1.1130527257919312,
      "learning_rate": 9.3565e-05,
      "loss": 0.4093,
      "step": 44472
    },
    {
      "epoch": 2.4903684623138087,
      "grad_norm": 1.208340048789978,
      "learning_rate": 9.356473684210528e-05,
      "loss": 0.3557,
      "step": 44473
    },
    {
      "epoch": 2.490424459625938,
      "grad_norm": 1.1466593742370605,
      "learning_rate": 9.356447368421054e-05,
      "loss": 0.3907,
      "step": 44474
    },
    {
      "epoch": 2.4904804569380667,
      "grad_norm": 1.4521417617797852,
      "learning_rate": 9.35642105263158e-05,
      "loss": 0.4391,
      "step": 44475
    },
    {
      "epoch": 2.490536454250196,
      "grad_norm": 1.3337092399597168,
      "learning_rate": 9.356394736842106e-05,
      "loss": 0.4372,
      "step": 44476
    },
    {
      "epoch": 2.490592451562325,
      "grad_norm": 1.2849644422531128,
      "learning_rate": 9.356368421052633e-05,
      "loss": 0.3156,
      "step": 44477
    },
    {
      "epoch": 2.4906484488744542,
      "grad_norm": 1.3593333959579468,
      "learning_rate": 9.356342105263159e-05,
      "loss": 0.4368,
      "step": 44478
    },
    {
      "epoch": 2.490704446186583,
      "grad_norm": 1.2579338550567627,
      "learning_rate": 9.356315789473685e-05,
      "loss": 0.3806,
      "step": 44479
    },
    {
      "epoch": 2.4907604434987123,
      "grad_norm": 1.108458161354065,
      "learning_rate": 9.356289473684211e-05,
      "loss": 0.4546,
      "step": 44480
    },
    {
      "epoch": 2.490816440810841,
      "grad_norm": 1.0899685621261597,
      "learning_rate": 9.356263157894737e-05,
      "loss": 0.3982,
      "step": 44481
    },
    {
      "epoch": 2.4908724381229703,
      "grad_norm": 1.3837579488754272,
      "learning_rate": 9.356236842105264e-05,
      "loss": 0.4146,
      "step": 44482
    },
    {
      "epoch": 2.490928435435099,
      "grad_norm": 1.0728747844696045,
      "learning_rate": 9.35621052631579e-05,
      "loss": 0.3834,
      "step": 44483
    },
    {
      "epoch": 2.4909844327472284,
      "grad_norm": 1.3314669132232666,
      "learning_rate": 9.356184210526316e-05,
      "loss": 0.4853,
      "step": 44484
    },
    {
      "epoch": 2.491040430059357,
      "grad_norm": 1.450663447380066,
      "learning_rate": 9.356157894736842e-05,
      "loss": 0.3598,
      "step": 44485
    },
    {
      "epoch": 2.4910964273714864,
      "grad_norm": 1.1993194818496704,
      "learning_rate": 9.35613157894737e-05,
      "loss": 0.4042,
      "step": 44486
    },
    {
      "epoch": 2.491152424683615,
      "grad_norm": 1.5242265462875366,
      "learning_rate": 9.356105263157895e-05,
      "loss": 0.5459,
      "step": 44487
    },
    {
      "epoch": 2.4912084219957444,
      "grad_norm": 1.3309662342071533,
      "learning_rate": 9.356078947368421e-05,
      "loss": 0.565,
      "step": 44488
    },
    {
      "epoch": 2.491264419307873,
      "grad_norm": 1.1130262613296509,
      "learning_rate": 9.356052631578947e-05,
      "loss": 0.3753,
      "step": 44489
    },
    {
      "epoch": 2.4913204166200025,
      "grad_norm": 1.0250605344772339,
      "learning_rate": 9.356026315789475e-05,
      "loss": 0.3533,
      "step": 44490
    },
    {
      "epoch": 2.491376413932131,
      "grad_norm": 1.19575035572052,
      "learning_rate": 9.356e-05,
      "loss": 0.302,
      "step": 44491
    },
    {
      "epoch": 2.4914324112442605,
      "grad_norm": 1.419180989265442,
      "learning_rate": 9.355973684210528e-05,
      "loss": 0.4853,
      "step": 44492
    },
    {
      "epoch": 2.491488408556389,
      "grad_norm": 1.2002583742141724,
      "learning_rate": 9.355947368421052e-05,
      "loss": 0.3595,
      "step": 44493
    },
    {
      "epoch": 2.4915444058685186,
      "grad_norm": 1.1833614110946655,
      "learning_rate": 9.35592105263158e-05,
      "loss": 0.3289,
      "step": 44494
    },
    {
      "epoch": 2.491600403180647,
      "grad_norm": 1.2692651748657227,
      "learning_rate": 9.355894736842106e-05,
      "loss": 0.4062,
      "step": 44495
    },
    {
      "epoch": 2.4916564004927766,
      "grad_norm": 0.9845141172409058,
      "learning_rate": 9.355868421052632e-05,
      "loss": 0.3014,
      "step": 44496
    },
    {
      "epoch": 2.491712397804905,
      "grad_norm": 7.51622200012207,
      "learning_rate": 9.355842105263158e-05,
      "loss": 0.4298,
      "step": 44497
    },
    {
      "epoch": 2.4917683951170346,
      "grad_norm": 1.1542017459869385,
      "learning_rate": 9.355815789473684e-05,
      "loss": 0.3768,
      "step": 44498
    },
    {
      "epoch": 2.491824392429163,
      "grad_norm": 1.2763515710830688,
      "learning_rate": 9.355789473684211e-05,
      "loss": 0.3632,
      "step": 44499
    },
    {
      "epoch": 2.4918803897412927,
      "grad_norm": 1.0262624025344849,
      "learning_rate": 9.355763157894737e-05,
      "loss": 0.3591,
      "step": 44500
    },
    {
      "epoch": 2.4919363870534212,
      "grad_norm": 1.1824690103530884,
      "learning_rate": 9.355736842105264e-05,
      "loss": 0.3901,
      "step": 44501
    },
    {
      "epoch": 2.4919923843655503,
      "grad_norm": 1.105539321899414,
      "learning_rate": 9.355710526315789e-05,
      "loss": 0.355,
      "step": 44502
    },
    {
      "epoch": 2.4920483816776793,
      "grad_norm": 1.0151619911193848,
      "learning_rate": 9.355684210526316e-05,
      "loss": 0.358,
      "step": 44503
    },
    {
      "epoch": 2.4921043789898083,
      "grad_norm": 1.5544389486312866,
      "learning_rate": 9.355657894736842e-05,
      "loss": 0.4465,
      "step": 44504
    },
    {
      "epoch": 2.4921603763019373,
      "grad_norm": 1.2862645387649536,
      "learning_rate": 9.35563157894737e-05,
      "loss": 0.399,
      "step": 44505
    },
    {
      "epoch": 2.4922163736140663,
      "grad_norm": 1.3210413455963135,
      "learning_rate": 9.355605263157896e-05,
      "loss": 0.6167,
      "step": 44506
    },
    {
      "epoch": 2.4922723709261954,
      "grad_norm": 1.354833960533142,
      "learning_rate": 9.355578947368422e-05,
      "loss": 0.5499,
      "step": 44507
    },
    {
      "epoch": 2.4923283682383244,
      "grad_norm": 1.7716625928878784,
      "learning_rate": 9.355552631578948e-05,
      "loss": 0.3527,
      "step": 44508
    },
    {
      "epoch": 2.4923843655504534,
      "grad_norm": 1.17331063747406,
      "learning_rate": 9.355526315789475e-05,
      "loss": 0.5329,
      "step": 44509
    },
    {
      "epoch": 2.4924403628625824,
      "grad_norm": 1.0916378498077393,
      "learning_rate": 9.355500000000001e-05,
      "loss": 0.3668,
      "step": 44510
    },
    {
      "epoch": 2.4924963601747114,
      "grad_norm": 1.2058136463165283,
      "learning_rate": 9.355473684210527e-05,
      "loss": 0.3862,
      "step": 44511
    },
    {
      "epoch": 2.4925523574868405,
      "grad_norm": 0.986126720905304,
      "learning_rate": 9.355447368421053e-05,
      "loss": 0.4441,
      "step": 44512
    },
    {
      "epoch": 2.4926083547989695,
      "grad_norm": 1.314240574836731,
      "learning_rate": 9.355421052631579e-05,
      "loss": 0.5099,
      "step": 44513
    },
    {
      "epoch": 2.4926643521110985,
      "grad_norm": 1.1080907583236694,
      "learning_rate": 9.355394736842106e-05,
      "loss": 0.4354,
      "step": 44514
    },
    {
      "epoch": 2.4927203494232275,
      "grad_norm": 1.100200891494751,
      "learning_rate": 9.355368421052632e-05,
      "loss": 0.3223,
      "step": 44515
    },
    {
      "epoch": 2.4927763467353565,
      "grad_norm": 1.2496306896209717,
      "learning_rate": 9.355342105263158e-05,
      "loss": 0.3977,
      "step": 44516
    },
    {
      "epoch": 2.4928323440474855,
      "grad_norm": 1.0086182355880737,
      "learning_rate": 9.355315789473684e-05,
      "loss": 0.3195,
      "step": 44517
    },
    {
      "epoch": 2.4928883413596146,
      "grad_norm": 1.2935634851455688,
      "learning_rate": 9.355289473684211e-05,
      "loss": 0.2798,
      "step": 44518
    },
    {
      "epoch": 2.4929443386717436,
      "grad_norm": 1.1985141038894653,
      "learning_rate": 9.355263157894737e-05,
      "loss": 0.4648,
      "step": 44519
    },
    {
      "epoch": 2.4930003359838726,
      "grad_norm": 1.1259708404541016,
      "learning_rate": 9.355236842105263e-05,
      "loss": 0.4877,
      "step": 44520
    },
    {
      "epoch": 2.4930563332960016,
      "grad_norm": 1.1808613538742065,
      "learning_rate": 9.355210526315789e-05,
      "loss": 0.4061,
      "step": 44521
    },
    {
      "epoch": 2.4931123306081306,
      "grad_norm": 1.2323607206344604,
      "learning_rate": 9.355184210526317e-05,
      "loss": 0.4406,
      "step": 44522
    },
    {
      "epoch": 2.4931683279202597,
      "grad_norm": 7.535837173461914,
      "learning_rate": 9.355157894736843e-05,
      "loss": 0.5043,
      "step": 44523
    },
    {
      "epoch": 2.4932243252323887,
      "grad_norm": 1.151171088218689,
      "learning_rate": 9.35513157894737e-05,
      "loss": 0.3319,
      "step": 44524
    },
    {
      "epoch": 2.4932803225445177,
      "grad_norm": 1.2301807403564453,
      "learning_rate": 9.355105263157894e-05,
      "loss": 0.505,
      "step": 44525
    },
    {
      "epoch": 2.4933363198566467,
      "grad_norm": 1.2326233386993408,
      "learning_rate": 9.355078947368422e-05,
      "loss": 0.3669,
      "step": 44526
    },
    {
      "epoch": 2.4933923171687757,
      "grad_norm": 1.208547830581665,
      "learning_rate": 9.355052631578948e-05,
      "loss": 0.3658,
      "step": 44527
    },
    {
      "epoch": 2.4934483144809048,
      "grad_norm": 1.1443120241165161,
      "learning_rate": 9.355026315789475e-05,
      "loss": 0.6528,
      "step": 44528
    },
    {
      "epoch": 2.493504311793034,
      "grad_norm": 1.2341182231903076,
      "learning_rate": 9.355000000000001e-05,
      "loss": 0.3968,
      "step": 44529
    },
    {
      "epoch": 2.493560309105163,
      "grad_norm": 1.1549293994903564,
      "learning_rate": 9.354973684210526e-05,
      "loss": 0.3434,
      "step": 44530
    },
    {
      "epoch": 2.493616306417292,
      "grad_norm": 1.357999563217163,
      "learning_rate": 9.354947368421053e-05,
      "loss": 0.4809,
      "step": 44531
    },
    {
      "epoch": 2.493672303729421,
      "grad_norm": 1.1309576034545898,
      "learning_rate": 9.354921052631579e-05,
      "loss": 0.3897,
      "step": 44532
    },
    {
      "epoch": 2.49372830104155,
      "grad_norm": 1.0351967811584473,
      "learning_rate": 9.354894736842106e-05,
      "loss": 0.4121,
      "step": 44533
    },
    {
      "epoch": 2.493784298353679,
      "grad_norm": 1.3602430820465088,
      "learning_rate": 9.354868421052632e-05,
      "loss": 0.5525,
      "step": 44534
    },
    {
      "epoch": 2.493840295665808,
      "grad_norm": 1.6055549383163452,
      "learning_rate": 9.354842105263158e-05,
      "loss": 0.4374,
      "step": 44535
    },
    {
      "epoch": 2.493896292977937,
      "grad_norm": 1.559765100479126,
      "learning_rate": 9.354815789473684e-05,
      "loss": 0.3399,
      "step": 44536
    },
    {
      "epoch": 2.493952290290066,
      "grad_norm": 1.170188069343567,
      "learning_rate": 9.354789473684212e-05,
      "loss": 0.3473,
      "step": 44537
    },
    {
      "epoch": 2.494008287602195,
      "grad_norm": 1.37400484085083,
      "learning_rate": 9.354763157894738e-05,
      "loss": 0.5714,
      "step": 44538
    },
    {
      "epoch": 2.494064284914324,
      "grad_norm": 1.215146541595459,
      "learning_rate": 9.354736842105264e-05,
      "loss": 0.3237,
      "step": 44539
    },
    {
      "epoch": 2.494120282226453,
      "grad_norm": 1.2419079542160034,
      "learning_rate": 9.35471052631579e-05,
      "loss": 0.5179,
      "step": 44540
    },
    {
      "epoch": 2.494176279538582,
      "grad_norm": 1.1104885339736938,
      "learning_rate": 9.354684210526317e-05,
      "loss": 0.3554,
      "step": 44541
    },
    {
      "epoch": 2.494232276850711,
      "grad_norm": 1.4985460042953491,
      "learning_rate": 9.354657894736843e-05,
      "loss": 0.512,
      "step": 44542
    },
    {
      "epoch": 2.49428827416284,
      "grad_norm": 1.1210670471191406,
      "learning_rate": 9.354631578947369e-05,
      "loss": 0.3436,
      "step": 44543
    },
    {
      "epoch": 2.494344271474969,
      "grad_norm": 1.16505765914917,
      "learning_rate": 9.354605263157895e-05,
      "loss": 0.3715,
      "step": 44544
    },
    {
      "epoch": 2.494400268787098,
      "grad_norm": 1.1544876098632812,
      "learning_rate": 9.354578947368422e-05,
      "loss": 0.331,
      "step": 44545
    },
    {
      "epoch": 2.494456266099227,
      "grad_norm": 1.709804892539978,
      "learning_rate": 9.354552631578948e-05,
      "loss": 0.544,
      "step": 44546
    },
    {
      "epoch": 2.494512263411356,
      "grad_norm": 0.9656328558921814,
      "learning_rate": 9.354526315789474e-05,
      "loss": 0.3138,
      "step": 44547
    },
    {
      "epoch": 2.494568260723485,
      "grad_norm": 1.2797094583511353,
      "learning_rate": 9.3545e-05,
      "loss": 0.4767,
      "step": 44548
    },
    {
      "epoch": 2.494624258035614,
      "grad_norm": 1.151874303817749,
      "learning_rate": 9.354473684210526e-05,
      "loss": 0.3544,
      "step": 44549
    },
    {
      "epoch": 2.494680255347743,
      "grad_norm": 1.486685037612915,
      "learning_rate": 9.354447368421053e-05,
      "loss": 0.3793,
      "step": 44550
    },
    {
      "epoch": 2.494736252659872,
      "grad_norm": 1.0361285209655762,
      "learning_rate": 9.354421052631579e-05,
      "loss": 0.4764,
      "step": 44551
    },
    {
      "epoch": 2.494792249972001,
      "grad_norm": 1.218870997428894,
      "learning_rate": 9.354394736842105e-05,
      "loss": 0.4632,
      "step": 44552
    },
    {
      "epoch": 2.4948482472841302,
      "grad_norm": 1.3622161149978638,
      "learning_rate": 9.354368421052631e-05,
      "loss": 0.3988,
      "step": 44553
    },
    {
      "epoch": 2.4949042445962593,
      "grad_norm": 1.3340386152267456,
      "learning_rate": 9.354342105263159e-05,
      "loss": 0.4068,
      "step": 44554
    },
    {
      "epoch": 2.4949602419083883,
      "grad_norm": 1.1575779914855957,
      "learning_rate": 9.354315789473684e-05,
      "loss": 0.398,
      "step": 44555
    },
    {
      "epoch": 2.4950162392205173,
      "grad_norm": 1.2215622663497925,
      "learning_rate": 9.354289473684212e-05,
      "loss": 0.4176,
      "step": 44556
    },
    {
      "epoch": 2.4950722365326463,
      "grad_norm": 1.2361446619033813,
      "learning_rate": 9.354263157894736e-05,
      "loss": 0.4508,
      "step": 44557
    },
    {
      "epoch": 2.4951282338447753,
      "grad_norm": 1.2982866764068604,
      "learning_rate": 9.354236842105264e-05,
      "loss": 0.411,
      "step": 44558
    },
    {
      "epoch": 2.4951842311569044,
      "grad_norm": 1.1623055934906006,
      "learning_rate": 9.35421052631579e-05,
      "loss": 0.3802,
      "step": 44559
    },
    {
      "epoch": 2.4952402284690334,
      "grad_norm": 1.106780767440796,
      "learning_rate": 9.354184210526317e-05,
      "loss": 0.301,
      "step": 44560
    },
    {
      "epoch": 2.4952962257811624,
      "grad_norm": 1.0043548345565796,
      "learning_rate": 9.354157894736843e-05,
      "loss": 0.3523,
      "step": 44561
    },
    {
      "epoch": 2.4953522230932914,
      "grad_norm": 1.3794631958007812,
      "learning_rate": 9.354131578947369e-05,
      "loss": 0.3433,
      "step": 44562
    },
    {
      "epoch": 2.4954082204054204,
      "grad_norm": 1.2128015756607056,
      "learning_rate": 9.354105263157895e-05,
      "loss": 0.477,
      "step": 44563
    },
    {
      "epoch": 2.4954642177175494,
      "grad_norm": 1.1903910636901855,
      "learning_rate": 9.354078947368422e-05,
      "loss": 0.3965,
      "step": 44564
    },
    {
      "epoch": 2.4955202150296785,
      "grad_norm": 1.2743923664093018,
      "learning_rate": 9.354052631578948e-05,
      "loss": 0.5635,
      "step": 44565
    },
    {
      "epoch": 2.4955762123418075,
      "grad_norm": 1.1189897060394287,
      "learning_rate": 9.354026315789474e-05,
      "loss": 0.3201,
      "step": 44566
    },
    {
      "epoch": 2.4956322096539365,
      "grad_norm": 1.3290870189666748,
      "learning_rate": 9.354e-05,
      "loss": 0.4668,
      "step": 44567
    },
    {
      "epoch": 2.4956882069660655,
      "grad_norm": 0.9568538069725037,
      "learning_rate": 9.353973684210526e-05,
      "loss": 0.3379,
      "step": 44568
    },
    {
      "epoch": 2.4957442042781945,
      "grad_norm": 1.2523893117904663,
      "learning_rate": 9.353947368421054e-05,
      "loss": 0.3831,
      "step": 44569
    },
    {
      "epoch": 2.4958002015903236,
      "grad_norm": 1.2217408418655396,
      "learning_rate": 9.35392105263158e-05,
      "loss": 0.4256,
      "step": 44570
    },
    {
      "epoch": 2.4958561989024526,
      "grad_norm": 1.054660439491272,
      "learning_rate": 9.353894736842105e-05,
      "loss": 0.4205,
      "step": 44571
    },
    {
      "epoch": 2.4959121962145816,
      "grad_norm": 1.337009072303772,
      "learning_rate": 9.353868421052631e-05,
      "loss": 0.5731,
      "step": 44572
    },
    {
      "epoch": 2.4959681935267106,
      "grad_norm": 1.045000433921814,
      "learning_rate": 9.353842105263159e-05,
      "loss": 0.3476,
      "step": 44573
    },
    {
      "epoch": 2.4960241908388396,
      "grad_norm": 1.155574917793274,
      "learning_rate": 9.353815789473685e-05,
      "loss": 0.3721,
      "step": 44574
    },
    {
      "epoch": 2.4960801881509687,
      "grad_norm": 1.5082155466079712,
      "learning_rate": 9.353789473684211e-05,
      "loss": 0.3882,
      "step": 44575
    },
    {
      "epoch": 2.4961361854630977,
      "grad_norm": 1.3360319137573242,
      "learning_rate": 9.353763157894737e-05,
      "loss": 0.4484,
      "step": 44576
    },
    {
      "epoch": 2.4961921827752267,
      "grad_norm": 1.323848009109497,
      "learning_rate": 9.353736842105264e-05,
      "loss": 0.4051,
      "step": 44577
    },
    {
      "epoch": 2.4962481800873557,
      "grad_norm": 9.819396018981934,
      "learning_rate": 9.35371052631579e-05,
      "loss": 0.4802,
      "step": 44578
    },
    {
      "epoch": 2.4963041773994847,
      "grad_norm": 1.1856976747512817,
      "learning_rate": 9.353684210526317e-05,
      "loss": 0.4157,
      "step": 44579
    },
    {
      "epoch": 2.4963601747116138,
      "grad_norm": 1.045804500579834,
      "learning_rate": 9.353657894736842e-05,
      "loss": 0.3884,
      "step": 44580
    },
    {
      "epoch": 2.4964161720237428,
      "grad_norm": 1.268300175666809,
      "learning_rate": 9.353631578947369e-05,
      "loss": 0.503,
      "step": 44581
    },
    {
      "epoch": 2.496472169335872,
      "grad_norm": 1.4763926267623901,
      "learning_rate": 9.353605263157895e-05,
      "loss": 0.5203,
      "step": 44582
    },
    {
      "epoch": 2.496528166648001,
      "grad_norm": 1.544094204902649,
      "learning_rate": 9.353578947368421e-05,
      "loss": 0.3919,
      "step": 44583
    },
    {
      "epoch": 2.49658416396013,
      "grad_norm": 1.068217158317566,
      "learning_rate": 9.353552631578949e-05,
      "loss": 0.3457,
      "step": 44584
    },
    {
      "epoch": 2.496640161272259,
      "grad_norm": 1.3168879747390747,
      "learning_rate": 9.353526315789473e-05,
      "loss": 0.3624,
      "step": 44585
    },
    {
      "epoch": 2.496696158584388,
      "grad_norm": 1.2534552812576294,
      "learning_rate": 9.3535e-05,
      "loss": 0.3504,
      "step": 44586
    },
    {
      "epoch": 2.496752155896517,
      "grad_norm": 1.0770039558410645,
      "learning_rate": 9.353473684210526e-05,
      "loss": 0.4354,
      "step": 44587
    },
    {
      "epoch": 2.496808153208646,
      "grad_norm": 1.4050458669662476,
      "learning_rate": 9.353447368421054e-05,
      "loss": 0.4579,
      "step": 44588
    },
    {
      "epoch": 2.496864150520775,
      "grad_norm": 1.3040722608566284,
      "learning_rate": 9.35342105263158e-05,
      "loss": 0.4504,
      "step": 44589
    },
    {
      "epoch": 2.496920147832904,
      "grad_norm": 1.0375791788101196,
      "learning_rate": 9.353394736842106e-05,
      "loss": 0.4131,
      "step": 44590
    },
    {
      "epoch": 2.496976145145033,
      "grad_norm": 1.1411267518997192,
      "learning_rate": 9.353368421052632e-05,
      "loss": 0.38,
      "step": 44591
    },
    {
      "epoch": 2.497032142457162,
      "grad_norm": 1.0989536046981812,
      "learning_rate": 9.353342105263159e-05,
      "loss": 0.4544,
      "step": 44592
    },
    {
      "epoch": 2.497088139769291,
      "grad_norm": 1.0053738355636597,
      "learning_rate": 9.353315789473685e-05,
      "loss": 0.367,
      "step": 44593
    },
    {
      "epoch": 2.49714413708142,
      "grad_norm": 1.3539074659347534,
      "learning_rate": 9.353289473684211e-05,
      "loss": 0.4807,
      "step": 44594
    },
    {
      "epoch": 2.497200134393549,
      "grad_norm": 1.2240935564041138,
      "learning_rate": 9.353263157894737e-05,
      "loss": 0.5392,
      "step": 44595
    },
    {
      "epoch": 2.497256131705678,
      "grad_norm": 1.0673106908798218,
      "learning_rate": 9.353236842105264e-05,
      "loss": 0.3482,
      "step": 44596
    },
    {
      "epoch": 2.497312129017807,
      "grad_norm": 1.2816349267959595,
      "learning_rate": 9.35321052631579e-05,
      "loss": 0.6039,
      "step": 44597
    },
    {
      "epoch": 2.497368126329936,
      "grad_norm": 1.082727074623108,
      "learning_rate": 9.353184210526316e-05,
      "loss": 0.4048,
      "step": 44598
    },
    {
      "epoch": 2.497424123642065,
      "grad_norm": 1.451478362083435,
      "learning_rate": 9.353157894736842e-05,
      "loss": 0.3925,
      "step": 44599
    },
    {
      "epoch": 2.497480120954194,
      "grad_norm": 1.020350694656372,
      "learning_rate": 9.353131578947368e-05,
      "loss": 0.325,
      "step": 44600
    },
    {
      "epoch": 2.497536118266323,
      "grad_norm": 1.119593858718872,
      "learning_rate": 9.353105263157895e-05,
      "loss": 0.3799,
      "step": 44601
    },
    {
      "epoch": 2.497592115578452,
      "grad_norm": 1.174675464630127,
      "learning_rate": 9.353078947368421e-05,
      "loss": 0.3132,
      "step": 44602
    },
    {
      "epoch": 2.497648112890581,
      "grad_norm": 1.2800489664077759,
      "learning_rate": 9.353052631578947e-05,
      "loss": 0.4283,
      "step": 44603
    },
    {
      "epoch": 2.49770411020271,
      "grad_norm": 1.4173389673233032,
      "learning_rate": 9.353026315789473e-05,
      "loss": 0.4295,
      "step": 44604
    },
    {
      "epoch": 2.4977601075148392,
      "grad_norm": 1.1316266059875488,
      "learning_rate": 9.353000000000001e-05,
      "loss": 0.4051,
      "step": 44605
    },
    {
      "epoch": 2.4978161048269683,
      "grad_norm": 1.1593562364578247,
      "learning_rate": 9.352973684210527e-05,
      "loss": 0.3948,
      "step": 44606
    },
    {
      "epoch": 2.4978721021390973,
      "grad_norm": 1.2337074279785156,
      "learning_rate": 9.352947368421053e-05,
      "loss": 0.4017,
      "step": 44607
    },
    {
      "epoch": 2.4979280994512263,
      "grad_norm": 0.9460651874542236,
      "learning_rate": 9.352921052631579e-05,
      "loss": 0.4544,
      "step": 44608
    },
    {
      "epoch": 2.4979840967633553,
      "grad_norm": 1.1809152364730835,
      "learning_rate": 9.352894736842106e-05,
      "loss": 0.3425,
      "step": 44609
    },
    {
      "epoch": 2.4980400940754843,
      "grad_norm": 1.2045084238052368,
      "learning_rate": 9.352868421052632e-05,
      "loss": 0.3817,
      "step": 44610
    },
    {
      "epoch": 2.4980960913876133,
      "grad_norm": 1.1365395784378052,
      "learning_rate": 9.352842105263159e-05,
      "loss": 0.3534,
      "step": 44611
    },
    {
      "epoch": 2.4981520886997424,
      "grad_norm": 1.476328730583191,
      "learning_rate": 9.352815789473684e-05,
      "loss": 0.4412,
      "step": 44612
    },
    {
      "epoch": 2.4982080860118714,
      "grad_norm": 1.2751727104187012,
      "learning_rate": 9.352789473684211e-05,
      "loss": 0.439,
      "step": 44613
    },
    {
      "epoch": 2.4982640833240004,
      "grad_norm": 1.4420068264007568,
      "learning_rate": 9.352763157894737e-05,
      "loss": 0.6271,
      "step": 44614
    },
    {
      "epoch": 2.4983200806361294,
      "grad_norm": 1.2181980609893799,
      "learning_rate": 9.352736842105265e-05,
      "loss": 0.3874,
      "step": 44615
    },
    {
      "epoch": 2.4983760779482584,
      "grad_norm": 1.0598855018615723,
      "learning_rate": 9.35271052631579e-05,
      "loss": 0.3678,
      "step": 44616
    },
    {
      "epoch": 2.4984320752603875,
      "grad_norm": 1.1201733350753784,
      "learning_rate": 9.352684210526315e-05,
      "loss": 0.4082,
      "step": 44617
    },
    {
      "epoch": 2.4984880725725165,
      "grad_norm": 1.218275547027588,
      "learning_rate": 9.352657894736842e-05,
      "loss": 0.3683,
      "step": 44618
    },
    {
      "epoch": 2.4985440698846455,
      "grad_norm": 1.186571717262268,
      "learning_rate": 9.352631578947368e-05,
      "loss": 0.4404,
      "step": 44619
    },
    {
      "epoch": 2.4986000671967745,
      "grad_norm": 1.2111152410507202,
      "learning_rate": 9.352605263157896e-05,
      "loss": 0.3227,
      "step": 44620
    },
    {
      "epoch": 2.4986560645089035,
      "grad_norm": 1.0418857336044312,
      "learning_rate": 9.352578947368422e-05,
      "loss": 0.3649,
      "step": 44621
    },
    {
      "epoch": 2.4987120618210326,
      "grad_norm": 1.175389051437378,
      "learning_rate": 9.352552631578948e-05,
      "loss": 0.381,
      "step": 44622
    },
    {
      "epoch": 2.4987680591331616,
      "grad_norm": 1.2156866788864136,
      "learning_rate": 9.352526315789474e-05,
      "loss": 0.3367,
      "step": 44623
    },
    {
      "epoch": 2.4988240564452906,
      "grad_norm": 1.110910415649414,
      "learning_rate": 9.352500000000001e-05,
      "loss": 0.427,
      "step": 44624
    },
    {
      "epoch": 2.4988800537574196,
      "grad_norm": 1.2534229755401611,
      "learning_rate": 9.352473684210527e-05,
      "loss": 0.3823,
      "step": 44625
    },
    {
      "epoch": 2.4989360510695486,
      "grad_norm": 1.2188228368759155,
      "learning_rate": 9.352447368421053e-05,
      "loss": 0.4874,
      "step": 44626
    },
    {
      "epoch": 2.4989920483816777,
      "grad_norm": 1.1332389116287231,
      "learning_rate": 9.352421052631579e-05,
      "loss": 0.38,
      "step": 44627
    },
    {
      "epoch": 2.4990480456938067,
      "grad_norm": 1.098968505859375,
      "learning_rate": 9.352394736842106e-05,
      "loss": 0.5178,
      "step": 44628
    },
    {
      "epoch": 2.4991040430059357,
      "grad_norm": 1.2652230262756348,
      "learning_rate": 9.352368421052632e-05,
      "loss": 0.4355,
      "step": 44629
    },
    {
      "epoch": 2.4991600403180647,
      "grad_norm": 1.156794786453247,
      "learning_rate": 9.352342105263158e-05,
      "loss": 0.5021,
      "step": 44630
    },
    {
      "epoch": 2.4992160376301937,
      "grad_norm": 1.033754825592041,
      "learning_rate": 9.352315789473684e-05,
      "loss": 0.3085,
      "step": 44631
    },
    {
      "epoch": 2.4992720349423228,
      "grad_norm": 1.0400224924087524,
      "learning_rate": 9.352289473684211e-05,
      "loss": 0.313,
      "step": 44632
    },
    {
      "epoch": 2.4993280322544518,
      "grad_norm": 1.517809510231018,
      "learning_rate": 9.352263157894737e-05,
      "loss": 0.3092,
      "step": 44633
    },
    {
      "epoch": 2.499384029566581,
      "grad_norm": 1.084779977798462,
      "learning_rate": 9.352236842105263e-05,
      "loss": 0.3796,
      "step": 44634
    },
    {
      "epoch": 2.49944002687871,
      "grad_norm": 0.949918270111084,
      "learning_rate": 9.35221052631579e-05,
      "loss": 0.3283,
      "step": 44635
    },
    {
      "epoch": 2.499496024190839,
      "grad_norm": 1.3239248991012573,
      "learning_rate": 9.352184210526315e-05,
      "loss": 0.494,
      "step": 44636
    },
    {
      "epoch": 2.499552021502968,
      "grad_norm": 19.85563850402832,
      "learning_rate": 9.352157894736843e-05,
      "loss": 0.4355,
      "step": 44637
    },
    {
      "epoch": 2.499608018815097,
      "grad_norm": 1.2303223609924316,
      "learning_rate": 9.352131578947369e-05,
      "loss": 0.615,
      "step": 44638
    },
    {
      "epoch": 2.499664016127226,
      "grad_norm": 1.1118488311767578,
      "learning_rate": 9.352105263157896e-05,
      "loss": 0.4323,
      "step": 44639
    },
    {
      "epoch": 2.499720013439355,
      "grad_norm": 1.187358021736145,
      "learning_rate": 9.35207894736842e-05,
      "loss": 0.3848,
      "step": 44640
    },
    {
      "epoch": 2.499776010751484,
      "grad_norm": 1.2090603113174438,
      "learning_rate": 9.352052631578948e-05,
      "loss": 0.3726,
      "step": 44641
    },
    {
      "epoch": 2.499832008063613,
      "grad_norm": 1.228467583656311,
      "learning_rate": 9.352026315789474e-05,
      "loss": 0.3517,
      "step": 44642
    },
    {
      "epoch": 2.499888005375742,
      "grad_norm": 1.5968700647354126,
      "learning_rate": 9.352000000000001e-05,
      "loss": 0.379,
      "step": 44643
    },
    {
      "epoch": 2.499944002687871,
      "grad_norm": 1.2593424320220947,
      "learning_rate": 9.351973684210526e-05,
      "loss": 0.4138,
      "step": 44644
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3016573190689087,
      "learning_rate": 9.351947368421053e-05,
      "loss": 0.3671,
      "step": 44645
    },
    {
      "epoch": 2.500055997312129,
      "grad_norm": 1.1840263605117798,
      "learning_rate": 9.351921052631579e-05,
      "loss": 0.4773,
      "step": 44646
    },
    {
      "epoch": 2.500111994624258,
      "grad_norm": 1.2678768634796143,
      "learning_rate": 9.351894736842107e-05,
      "loss": 0.3908,
      "step": 44647
    },
    {
      "epoch": 2.500167991936387,
      "grad_norm": 0.993491530418396,
      "learning_rate": 9.351868421052632e-05,
      "loss": 0.3165,
      "step": 44648
    },
    {
      "epoch": 2.500223989248516,
      "grad_norm": 2.2869937419891357,
      "learning_rate": 9.351842105263158e-05,
      "loss": 0.4759,
      "step": 44649
    },
    {
      "epoch": 2.500279986560645,
      "grad_norm": 0.9921106100082397,
      "learning_rate": 9.351815789473684e-05,
      "loss": 0.377,
      "step": 44650
    },
    {
      "epoch": 2.500335983872774,
      "grad_norm": 1.2296686172485352,
      "learning_rate": 9.35178947368421e-05,
      "loss": 0.4371,
      "step": 44651
    },
    {
      "epoch": 2.500391981184903,
      "grad_norm": 1.1348974704742432,
      "learning_rate": 9.351763157894738e-05,
      "loss": 0.3865,
      "step": 44652
    },
    {
      "epoch": 2.500447978497032,
      "grad_norm": 1.2140257358551025,
      "learning_rate": 9.351736842105264e-05,
      "loss": 0.4715,
      "step": 44653
    },
    {
      "epoch": 2.500503975809161,
      "grad_norm": 1.4428361654281616,
      "learning_rate": 9.35171052631579e-05,
      "loss": 0.4581,
      "step": 44654
    },
    {
      "epoch": 2.50055997312129,
      "grad_norm": 1.1700103282928467,
      "learning_rate": 9.351684210526316e-05,
      "loss": 0.3407,
      "step": 44655
    },
    {
      "epoch": 2.500615970433419,
      "grad_norm": 1.299353003501892,
      "learning_rate": 9.351657894736843e-05,
      "loss": 0.4395,
      "step": 44656
    },
    {
      "epoch": 2.5006719677455482,
      "grad_norm": 1.2602280378341675,
      "learning_rate": 9.351631578947369e-05,
      "loss": 0.4957,
      "step": 44657
    },
    {
      "epoch": 2.5007279650576772,
      "grad_norm": 1.3617912530899048,
      "learning_rate": 9.351605263157895e-05,
      "loss": 0.4178,
      "step": 44658
    },
    {
      "epoch": 2.5007839623698063,
      "grad_norm": 1.1706573963165283,
      "learning_rate": 9.351578947368421e-05,
      "loss": 0.316,
      "step": 44659
    },
    {
      "epoch": 2.5008399596819353,
      "grad_norm": 1.237616777420044,
      "learning_rate": 9.351552631578948e-05,
      "loss": 0.3374,
      "step": 44660
    },
    {
      "epoch": 2.5008959569940643,
      "grad_norm": 1.6490988731384277,
      "learning_rate": 9.351526315789474e-05,
      "loss": 0.4088,
      "step": 44661
    },
    {
      "epoch": 2.5009519543061933,
      "grad_norm": 1.238945484161377,
      "learning_rate": 9.3515e-05,
      "loss": 0.6542,
      "step": 44662
    },
    {
      "epoch": 2.5010079516183223,
      "grad_norm": 1.5903511047363281,
      "learning_rate": 9.351473684210526e-05,
      "loss": 0.4516,
      "step": 44663
    },
    {
      "epoch": 2.5010639489304514,
      "grad_norm": 1.311757206916809,
      "learning_rate": 9.351447368421053e-05,
      "loss": 0.4234,
      "step": 44664
    },
    {
      "epoch": 2.5011199462425804,
      "grad_norm": 1.1380364894866943,
      "learning_rate": 9.35142105263158e-05,
      "loss": 0.3385,
      "step": 44665
    },
    {
      "epoch": 2.5011759435547094,
      "grad_norm": 1.1873936653137207,
      "learning_rate": 9.351394736842107e-05,
      "loss": 0.4796,
      "step": 44666
    },
    {
      "epoch": 2.5012319408668384,
      "grad_norm": 1.6725373268127441,
      "learning_rate": 9.351368421052631e-05,
      "loss": 0.4425,
      "step": 44667
    },
    {
      "epoch": 2.5012879381789674,
      "grad_norm": 1.152907371520996,
      "learning_rate": 9.351342105263157e-05,
      "loss": 0.3858,
      "step": 44668
    },
    {
      "epoch": 2.5013439354910965,
      "grad_norm": 1.0271967649459839,
      "learning_rate": 9.351315789473685e-05,
      "loss": 0.441,
      "step": 44669
    },
    {
      "epoch": 2.5013999328032255,
      "grad_norm": 1.0780562162399292,
      "learning_rate": 9.35128947368421e-05,
      "loss": 0.3655,
      "step": 44670
    },
    {
      "epoch": 2.5014559301153545,
      "grad_norm": 0.9843218326568604,
      "learning_rate": 9.351263157894738e-05,
      "loss": 0.2872,
      "step": 44671
    },
    {
      "epoch": 2.5015119274274835,
      "grad_norm": 1.1521226167678833,
      "learning_rate": 9.351236842105263e-05,
      "loss": 0.3923,
      "step": 44672
    },
    {
      "epoch": 2.5015679247396125,
      "grad_norm": 1.3495076894760132,
      "learning_rate": 9.35121052631579e-05,
      "loss": 0.4406,
      "step": 44673
    },
    {
      "epoch": 2.5016239220517416,
      "grad_norm": 1.1498416662216187,
      "learning_rate": 9.351184210526316e-05,
      "loss": 0.3942,
      "step": 44674
    },
    {
      "epoch": 2.5016799193638706,
      "grad_norm": 1.2754565477371216,
      "learning_rate": 9.351157894736843e-05,
      "loss": 0.2719,
      "step": 44675
    },
    {
      "epoch": 2.5017359166759996,
      "grad_norm": 1.2649343013763428,
      "learning_rate": 9.351131578947369e-05,
      "loss": 0.476,
      "step": 44676
    },
    {
      "epoch": 2.5017919139881286,
      "grad_norm": 0.9628691673278809,
      "learning_rate": 9.351105263157895e-05,
      "loss": 0.3764,
      "step": 44677
    },
    {
      "epoch": 2.5018479113002576,
      "grad_norm": 1.0442384481430054,
      "learning_rate": 9.351078947368421e-05,
      "loss": 0.3362,
      "step": 44678
    },
    {
      "epoch": 2.5019039086123867,
      "grad_norm": 1.1974157094955444,
      "learning_rate": 9.351052631578948e-05,
      "loss": 0.4438,
      "step": 44679
    },
    {
      "epoch": 2.5019599059245157,
      "grad_norm": 1.1462780237197876,
      "learning_rate": 9.351026315789474e-05,
      "loss": 0.5253,
      "step": 44680
    },
    {
      "epoch": 2.5020159032366447,
      "grad_norm": 1.182629108428955,
      "learning_rate": 9.351e-05,
      "loss": 0.3761,
      "step": 44681
    },
    {
      "epoch": 2.5020719005487737,
      "grad_norm": 1.0513659715652466,
      "learning_rate": 9.350973684210526e-05,
      "loss": 0.3578,
      "step": 44682
    },
    {
      "epoch": 2.5021278978609027,
      "grad_norm": 1.19379723072052,
      "learning_rate": 9.350947368421054e-05,
      "loss": 0.4054,
      "step": 44683
    },
    {
      "epoch": 2.5021838951730317,
      "grad_norm": 1.318214774131775,
      "learning_rate": 9.35092105263158e-05,
      "loss": 0.4571,
      "step": 44684
    },
    {
      "epoch": 2.5022398924851608,
      "grad_norm": 1.1367806196212769,
      "learning_rate": 9.350894736842106e-05,
      "loss": 0.4094,
      "step": 44685
    },
    {
      "epoch": 2.50229588979729,
      "grad_norm": 1.2601944208145142,
      "learning_rate": 9.350868421052632e-05,
      "loss": 0.4854,
      "step": 44686
    },
    {
      "epoch": 2.502351887109419,
      "grad_norm": 1.2000831365585327,
      "learning_rate": 9.350842105263158e-05,
      "loss": 0.389,
      "step": 44687
    },
    {
      "epoch": 2.502407884421548,
      "grad_norm": 1.3486360311508179,
      "learning_rate": 9.350815789473685e-05,
      "loss": 0.3434,
      "step": 44688
    },
    {
      "epoch": 2.502463881733677,
      "grad_norm": 1.184953212738037,
      "learning_rate": 9.350789473684211e-05,
      "loss": 0.348,
      "step": 44689
    },
    {
      "epoch": 2.502519879045806,
      "grad_norm": 1.2773815393447876,
      "learning_rate": 9.350763157894737e-05,
      "loss": 0.5166,
      "step": 44690
    },
    {
      "epoch": 2.502575876357935,
      "grad_norm": 1.2060847282409668,
      "learning_rate": 9.350736842105263e-05,
      "loss": 0.4637,
      "step": 44691
    },
    {
      "epoch": 2.502631873670064,
      "grad_norm": 1.0748361349105835,
      "learning_rate": 9.35071052631579e-05,
      "loss": 0.3526,
      "step": 44692
    },
    {
      "epoch": 2.502687870982193,
      "grad_norm": 1.1506118774414062,
      "learning_rate": 9.350684210526316e-05,
      "loss": 0.3724,
      "step": 44693
    },
    {
      "epoch": 2.502743868294322,
      "grad_norm": 1.3560246229171753,
      "learning_rate": 9.350657894736843e-05,
      "loss": 0.4857,
      "step": 44694
    },
    {
      "epoch": 2.502799865606451,
      "grad_norm": 1.2747424840927124,
      "learning_rate": 9.350631578947368e-05,
      "loss": 0.3745,
      "step": 44695
    },
    {
      "epoch": 2.50285586291858,
      "grad_norm": 1.3685498237609863,
      "learning_rate": 9.350605263157895e-05,
      "loss": 0.4849,
      "step": 44696
    },
    {
      "epoch": 2.502911860230709,
      "grad_norm": 1.009788155555725,
      "learning_rate": 9.350578947368421e-05,
      "loss": 0.3425,
      "step": 44697
    },
    {
      "epoch": 2.502967857542838,
      "grad_norm": 1.2830582857131958,
      "learning_rate": 9.350552631578949e-05,
      "loss": 0.4866,
      "step": 44698
    },
    {
      "epoch": 2.503023854854967,
      "grad_norm": 1.2266055345535278,
      "learning_rate": 9.350526315789473e-05,
      "loss": 0.342,
      "step": 44699
    },
    {
      "epoch": 2.503079852167096,
      "grad_norm": 1.1327704191207886,
      "learning_rate": 9.350500000000001e-05,
      "loss": 0.3717,
      "step": 44700
    },
    {
      "epoch": 2.503135849479225,
      "grad_norm": 1.1795837879180908,
      "learning_rate": 9.350473684210527e-05,
      "loss": 0.4293,
      "step": 44701
    },
    {
      "epoch": 2.503191846791354,
      "grad_norm": 1.298890233039856,
      "learning_rate": 9.350447368421054e-05,
      "loss": 0.4682,
      "step": 44702
    },
    {
      "epoch": 2.503247844103483,
      "grad_norm": 1.1279315948486328,
      "learning_rate": 9.35042105263158e-05,
      "loss": 0.3738,
      "step": 44703
    },
    {
      "epoch": 2.503303841415612,
      "grad_norm": 1.2725718021392822,
      "learning_rate": 9.350394736842105e-05,
      "loss": 0.3567,
      "step": 44704
    },
    {
      "epoch": 2.503359838727741,
      "grad_norm": 1.3284419775009155,
      "learning_rate": 9.350368421052632e-05,
      "loss": 0.3593,
      "step": 44705
    },
    {
      "epoch": 2.50341583603987,
      "grad_norm": 1.8396999835968018,
      "learning_rate": 9.350342105263158e-05,
      "loss": 0.5817,
      "step": 44706
    },
    {
      "epoch": 2.503471833351999,
      "grad_norm": 1.316655158996582,
      "learning_rate": 9.350315789473685e-05,
      "loss": 0.4341,
      "step": 44707
    },
    {
      "epoch": 2.503527830664128,
      "grad_norm": 1.2661175727844238,
      "learning_rate": 9.350289473684211e-05,
      "loss": 0.3674,
      "step": 44708
    },
    {
      "epoch": 2.5035838279762572,
      "grad_norm": 1.917382001876831,
      "learning_rate": 9.350263157894737e-05,
      "loss": 0.3928,
      "step": 44709
    },
    {
      "epoch": 2.5036398252883862,
      "grad_norm": 1.1785186529159546,
      "learning_rate": 9.350236842105263e-05,
      "loss": 0.4208,
      "step": 44710
    },
    {
      "epoch": 2.5036958226005153,
      "grad_norm": 1.1549183130264282,
      "learning_rate": 9.35021052631579e-05,
      "loss": 0.4218,
      "step": 44711
    },
    {
      "epoch": 2.5037518199126443,
      "grad_norm": 1.2087838649749756,
      "learning_rate": 9.350184210526316e-05,
      "loss": 0.368,
      "step": 44712
    },
    {
      "epoch": 2.5038078172247733,
      "grad_norm": 1.0444931983947754,
      "learning_rate": 9.350157894736842e-05,
      "loss": 0.3661,
      "step": 44713
    },
    {
      "epoch": 2.5038638145369023,
      "grad_norm": 1.3926641941070557,
      "learning_rate": 9.350131578947368e-05,
      "loss": 0.3647,
      "step": 44714
    },
    {
      "epoch": 2.5039198118490313,
      "grad_norm": 1.0948355197906494,
      "learning_rate": 9.350105263157896e-05,
      "loss": 0.2584,
      "step": 44715
    },
    {
      "epoch": 2.5039758091611604,
      "grad_norm": 1.3165448904037476,
      "learning_rate": 9.350078947368422e-05,
      "loss": 0.5379,
      "step": 44716
    },
    {
      "epoch": 2.5040318064732894,
      "grad_norm": 1.1503769159317017,
      "learning_rate": 9.350052631578948e-05,
      "loss": 0.3543,
      "step": 44717
    },
    {
      "epoch": 2.5040878037854184,
      "grad_norm": 1.5424970388412476,
      "learning_rate": 9.350026315789474e-05,
      "loss": 0.4101,
      "step": 44718
    },
    {
      "epoch": 2.5041438010975474,
      "grad_norm": 1.2243847846984863,
      "learning_rate": 9.350000000000001e-05,
      "loss": 0.3538,
      "step": 44719
    },
    {
      "epoch": 2.5041997984096764,
      "grad_norm": 1.2283695936203003,
      "learning_rate": 9.349973684210527e-05,
      "loss": 0.3983,
      "step": 44720
    },
    {
      "epoch": 2.5042557957218055,
      "grad_norm": 1.2716865539550781,
      "learning_rate": 9.349947368421053e-05,
      "loss": 0.5418,
      "step": 44721
    },
    {
      "epoch": 2.5043117930339345,
      "grad_norm": 1.2942513227462769,
      "learning_rate": 9.349921052631579e-05,
      "loss": 0.3613,
      "step": 44722
    },
    {
      "epoch": 2.5043677903460635,
      "grad_norm": 1.0981354713439941,
      "learning_rate": 9.349894736842105e-05,
      "loss": 0.3238,
      "step": 44723
    },
    {
      "epoch": 2.5044237876581925,
      "grad_norm": 1.3400903940200806,
      "learning_rate": 9.349868421052632e-05,
      "loss": 0.3353,
      "step": 44724
    },
    {
      "epoch": 2.5044797849703215,
      "grad_norm": 1.2113410234451294,
      "learning_rate": 9.349842105263158e-05,
      "loss": 0.3542,
      "step": 44725
    },
    {
      "epoch": 2.5045357822824506,
      "grad_norm": 1.2446471452713013,
      "learning_rate": 9.349815789473685e-05,
      "loss": 0.3027,
      "step": 44726
    },
    {
      "epoch": 2.5045917795945796,
      "grad_norm": 1.138442039489746,
      "learning_rate": 9.34978947368421e-05,
      "loss": 0.3691,
      "step": 44727
    },
    {
      "epoch": 2.5046477769067086,
      "grad_norm": 1.3231689929962158,
      "learning_rate": 9.349763157894737e-05,
      "loss": 0.3512,
      "step": 44728
    },
    {
      "epoch": 2.5047037742188376,
      "grad_norm": 1.287824273109436,
      "learning_rate": 9.349736842105263e-05,
      "loss": 0.381,
      "step": 44729
    },
    {
      "epoch": 2.5047597715309666,
      "grad_norm": 1.1315996646881104,
      "learning_rate": 9.349710526315791e-05,
      "loss": 0.3756,
      "step": 44730
    },
    {
      "epoch": 2.5048157688430956,
      "grad_norm": 1.2799986600875854,
      "learning_rate": 9.349684210526317e-05,
      "loss": 0.3874,
      "step": 44731
    },
    {
      "epoch": 2.5048717661552247,
      "grad_norm": 1.2432392835617065,
      "learning_rate": 9.349657894736843e-05,
      "loss": 0.2736,
      "step": 44732
    },
    {
      "epoch": 2.5049277634673537,
      "grad_norm": 1.2550952434539795,
      "learning_rate": 9.349631578947369e-05,
      "loss": 0.4735,
      "step": 44733
    },
    {
      "epoch": 2.5049837607794827,
      "grad_norm": 1.1441066265106201,
      "learning_rate": 9.349605263157896e-05,
      "loss": 0.372,
      "step": 44734
    },
    {
      "epoch": 2.5050397580916117,
      "grad_norm": 1.1870216131210327,
      "learning_rate": 9.349578947368422e-05,
      "loss": 0.386,
      "step": 44735
    },
    {
      "epoch": 2.5050957554037407,
      "grad_norm": 1.2146191596984863,
      "learning_rate": 9.349552631578948e-05,
      "loss": 0.3392,
      "step": 44736
    },
    {
      "epoch": 2.5051517527158698,
      "grad_norm": 1.1965885162353516,
      "learning_rate": 9.349526315789474e-05,
      "loss": 0.4497,
      "step": 44737
    },
    {
      "epoch": 2.505207750027999,
      "grad_norm": 1.1423410177230835,
      "learning_rate": 9.3495e-05,
      "loss": 0.3101,
      "step": 44738
    },
    {
      "epoch": 2.505263747340128,
      "grad_norm": 1.023838996887207,
      "learning_rate": 9.349473684210527e-05,
      "loss": 0.2957,
      "step": 44739
    },
    {
      "epoch": 2.505319744652257,
      "grad_norm": 1.3482131958007812,
      "learning_rate": 9.349447368421053e-05,
      "loss": 0.5415,
      "step": 44740
    },
    {
      "epoch": 2.505375741964386,
      "grad_norm": 1.3883073329925537,
      "learning_rate": 9.349421052631579e-05,
      "loss": 0.5359,
      "step": 44741
    },
    {
      "epoch": 2.505431739276515,
      "grad_norm": 1.2522249221801758,
      "learning_rate": 9.349394736842105e-05,
      "loss": 0.3965,
      "step": 44742
    },
    {
      "epoch": 2.505487736588644,
      "grad_norm": 1.0210261344909668,
      "learning_rate": 9.349368421052632e-05,
      "loss": 0.2961,
      "step": 44743
    },
    {
      "epoch": 2.505543733900773,
      "grad_norm": 1.5919770002365112,
      "learning_rate": 9.349342105263158e-05,
      "loss": 0.4089,
      "step": 44744
    },
    {
      "epoch": 2.505599731212902,
      "grad_norm": 1.1701679229736328,
      "learning_rate": 9.349315789473684e-05,
      "loss": 0.3503,
      "step": 44745
    },
    {
      "epoch": 2.505655728525031,
      "grad_norm": 1.1423298120498657,
      "learning_rate": 9.34928947368421e-05,
      "loss": 0.4315,
      "step": 44746
    },
    {
      "epoch": 2.50571172583716,
      "grad_norm": 1.3764071464538574,
      "learning_rate": 9.349263157894738e-05,
      "loss": 0.3889,
      "step": 44747
    },
    {
      "epoch": 2.505767723149289,
      "grad_norm": 1.380503535270691,
      "learning_rate": 9.349236842105264e-05,
      "loss": 0.4675,
      "step": 44748
    },
    {
      "epoch": 2.505823720461418,
      "grad_norm": 1.128568410873413,
      "learning_rate": 9.349210526315791e-05,
      "loss": 0.4233,
      "step": 44749
    },
    {
      "epoch": 2.505879717773547,
      "grad_norm": 1.4740972518920898,
      "learning_rate": 9.349184210526316e-05,
      "loss": 0.4184,
      "step": 44750
    },
    {
      "epoch": 2.505935715085676,
      "grad_norm": 1.5354292392730713,
      "learning_rate": 9.349157894736843e-05,
      "loss": 0.6251,
      "step": 44751
    },
    {
      "epoch": 2.505991712397805,
      "grad_norm": 1.0475372076034546,
      "learning_rate": 9.349131578947369e-05,
      "loss": 0.3222,
      "step": 44752
    },
    {
      "epoch": 2.506047709709934,
      "grad_norm": 1.1628245115280151,
      "learning_rate": 9.349105263157896e-05,
      "loss": 0.4823,
      "step": 44753
    },
    {
      "epoch": 2.506103707022063,
      "grad_norm": 2.2967488765716553,
      "learning_rate": 9.349078947368421e-05,
      "loss": 0.3765,
      "step": 44754
    },
    {
      "epoch": 2.506159704334192,
      "grad_norm": 1.2709400653839111,
      "learning_rate": 9.349052631578947e-05,
      "loss": 0.4373,
      "step": 44755
    },
    {
      "epoch": 2.506215701646321,
      "grad_norm": 0.9283302426338196,
      "learning_rate": 9.349026315789474e-05,
      "loss": 0.3354,
      "step": 44756
    },
    {
      "epoch": 2.50627169895845,
      "grad_norm": 1.3800677061080933,
      "learning_rate": 9.349e-05,
      "loss": 0.5264,
      "step": 44757
    },
    {
      "epoch": 2.506327696270579,
      "grad_norm": 1.177882194519043,
      "learning_rate": 9.348973684210527e-05,
      "loss": 0.4655,
      "step": 44758
    },
    {
      "epoch": 2.506383693582708,
      "grad_norm": 1.2900073528289795,
      "learning_rate": 9.348947368421052e-05,
      "loss": 0.3857,
      "step": 44759
    },
    {
      "epoch": 2.506439690894837,
      "grad_norm": 1.0525318384170532,
      "learning_rate": 9.34892105263158e-05,
      "loss": 0.3985,
      "step": 44760
    },
    {
      "epoch": 2.506495688206966,
      "grad_norm": 1.3303323984146118,
      "learning_rate": 9.348894736842105e-05,
      "loss": 0.3265,
      "step": 44761
    },
    {
      "epoch": 2.5065516855190952,
      "grad_norm": 1.3088613748550415,
      "learning_rate": 9.348868421052633e-05,
      "loss": 0.3647,
      "step": 44762
    },
    {
      "epoch": 2.5066076828312243,
      "grad_norm": 1.0863131284713745,
      "learning_rate": 9.348842105263159e-05,
      "loss": 0.4025,
      "step": 44763
    },
    {
      "epoch": 2.5066636801433533,
      "grad_norm": 1.4102165699005127,
      "learning_rate": 9.348815789473685e-05,
      "loss": 0.3959,
      "step": 44764
    },
    {
      "epoch": 2.5067196774554823,
      "grad_norm": 1.0793423652648926,
      "learning_rate": 9.34878947368421e-05,
      "loss": 0.5526,
      "step": 44765
    },
    {
      "epoch": 2.5067756747676113,
      "grad_norm": 1.2461094856262207,
      "learning_rate": 9.348763157894738e-05,
      "loss": 0.3942,
      "step": 44766
    },
    {
      "epoch": 2.5068316720797403,
      "grad_norm": 2.523122549057007,
      "learning_rate": 9.348736842105264e-05,
      "loss": 0.3373,
      "step": 44767
    },
    {
      "epoch": 2.5068876693918694,
      "grad_norm": 1.1954014301300049,
      "learning_rate": 9.34871052631579e-05,
      "loss": 0.3472,
      "step": 44768
    },
    {
      "epoch": 2.5069436667039984,
      "grad_norm": 1.023280382156372,
      "learning_rate": 9.348684210526316e-05,
      "loss": 0.3267,
      "step": 44769
    },
    {
      "epoch": 2.506999664016127,
      "grad_norm": 1.2781025171279907,
      "learning_rate": 9.348657894736843e-05,
      "loss": 0.3043,
      "step": 44770
    },
    {
      "epoch": 2.5070556613282564,
      "grad_norm": 1.1376464366912842,
      "learning_rate": 9.348631578947369e-05,
      "loss": 0.3542,
      "step": 44771
    },
    {
      "epoch": 2.507111658640385,
      "grad_norm": 1.0966333150863647,
      "learning_rate": 9.348605263157895e-05,
      "loss": 0.3635,
      "step": 44772
    },
    {
      "epoch": 2.5071676559525145,
      "grad_norm": 1.1147621870040894,
      "learning_rate": 9.348578947368421e-05,
      "loss": 0.4341,
      "step": 44773
    },
    {
      "epoch": 2.507223653264643,
      "grad_norm": 0.9573778510093689,
      "learning_rate": 9.348552631578947e-05,
      "loss": 0.2992,
      "step": 44774
    },
    {
      "epoch": 2.5072796505767725,
      "grad_norm": 1.322076678276062,
      "learning_rate": 9.348526315789474e-05,
      "loss": 0.4546,
      "step": 44775
    },
    {
      "epoch": 2.507335647888901,
      "grad_norm": 1.108537197113037,
      "learning_rate": 9.3485e-05,
      "loss": 0.3718,
      "step": 44776
    },
    {
      "epoch": 2.5073916452010305,
      "grad_norm": 1.3583650588989258,
      "learning_rate": 9.348473684210526e-05,
      "loss": 0.5452,
      "step": 44777
    },
    {
      "epoch": 2.507447642513159,
      "grad_norm": 1.4060158729553223,
      "learning_rate": 9.348447368421052e-05,
      "loss": 0.3256,
      "step": 44778
    },
    {
      "epoch": 2.5075036398252886,
      "grad_norm": 0.9820559024810791,
      "learning_rate": 9.34842105263158e-05,
      "loss": 0.3332,
      "step": 44779
    },
    {
      "epoch": 2.507559637137417,
      "grad_norm": 1.1356019973754883,
      "learning_rate": 9.348394736842106e-05,
      "loss": 0.4063,
      "step": 44780
    },
    {
      "epoch": 2.5076156344495466,
      "grad_norm": 0.9587997794151306,
      "learning_rate": 9.348368421052633e-05,
      "loss": 0.3454,
      "step": 44781
    },
    {
      "epoch": 2.507671631761675,
      "grad_norm": 1.2400751113891602,
      "learning_rate": 9.348342105263158e-05,
      "loss": 0.429,
      "step": 44782
    },
    {
      "epoch": 2.5077276290738046,
      "grad_norm": 1.120600938796997,
      "learning_rate": 9.348315789473685e-05,
      "loss": 0.5239,
      "step": 44783
    },
    {
      "epoch": 2.507783626385933,
      "grad_norm": 1.1395894289016724,
      "learning_rate": 9.348289473684211e-05,
      "loss": 0.4214,
      "step": 44784
    },
    {
      "epoch": 2.5078396236980627,
      "grad_norm": 1.3470524549484253,
      "learning_rate": 9.348263157894738e-05,
      "loss": 0.3757,
      "step": 44785
    },
    {
      "epoch": 2.5078956210101913,
      "grad_norm": 1.1209582090377808,
      "learning_rate": 9.348236842105264e-05,
      "loss": 0.3477,
      "step": 44786
    },
    {
      "epoch": 2.5079516183223207,
      "grad_norm": 1.6194506883621216,
      "learning_rate": 9.34821052631579e-05,
      "loss": 0.3662,
      "step": 44787
    },
    {
      "epoch": 2.5080076156344493,
      "grad_norm": 1.4752724170684814,
      "learning_rate": 9.348184210526316e-05,
      "loss": 0.4196,
      "step": 44788
    },
    {
      "epoch": 2.5080636129465788,
      "grad_norm": 1.1000336408615112,
      "learning_rate": 9.348157894736842e-05,
      "loss": 0.3127,
      "step": 44789
    },
    {
      "epoch": 2.5081196102587073,
      "grad_norm": 1.4938992261886597,
      "learning_rate": 9.34813157894737e-05,
      "loss": 0.4267,
      "step": 44790
    },
    {
      "epoch": 2.508175607570837,
      "grad_norm": 1.1985728740692139,
      "learning_rate": 9.348105263157895e-05,
      "loss": 0.3426,
      "step": 44791
    },
    {
      "epoch": 2.5082316048829654,
      "grad_norm": 1.3682639598846436,
      "learning_rate": 9.348078947368421e-05,
      "loss": 0.4814,
      "step": 44792
    },
    {
      "epoch": 2.508287602195095,
      "grad_norm": 1.1228935718536377,
      "learning_rate": 9.348052631578947e-05,
      "loss": 0.3589,
      "step": 44793
    },
    {
      "epoch": 2.5083435995072234,
      "grad_norm": 1.0209373235702515,
      "learning_rate": 9.348026315789475e-05,
      "loss": 0.4057,
      "step": 44794
    },
    {
      "epoch": 2.508399596819353,
      "grad_norm": 1.5994699001312256,
      "learning_rate": 9.348e-05,
      "loss": 0.565,
      "step": 44795
    },
    {
      "epoch": 2.5084555941314814,
      "grad_norm": 1.1077467203140259,
      "learning_rate": 9.347973684210527e-05,
      "loss": 0.4683,
      "step": 44796
    },
    {
      "epoch": 2.508511591443611,
      "grad_norm": 1.3415753841400146,
      "learning_rate": 9.347947368421053e-05,
      "loss": 0.4564,
      "step": 44797
    },
    {
      "epoch": 2.5085675887557395,
      "grad_norm": 1.2240289449691772,
      "learning_rate": 9.34792105263158e-05,
      "loss": 0.4228,
      "step": 44798
    },
    {
      "epoch": 2.508623586067869,
      "grad_norm": 1.2324711084365845,
      "learning_rate": 9.347894736842106e-05,
      "loss": 0.4194,
      "step": 44799
    },
    {
      "epoch": 2.5086795833799975,
      "grad_norm": 1.3035613298416138,
      "learning_rate": 9.347868421052632e-05,
      "loss": 0.4778,
      "step": 44800
    },
    {
      "epoch": 2.508735580692127,
      "grad_norm": 1.1946133375167847,
      "learning_rate": 9.347842105263158e-05,
      "loss": 0.5276,
      "step": 44801
    },
    {
      "epoch": 2.5087915780042556,
      "grad_norm": 1.1371045112609863,
      "learning_rate": 9.347815789473685e-05,
      "loss": 0.3981,
      "step": 44802
    },
    {
      "epoch": 2.508847575316385,
      "grad_norm": 1.5162835121154785,
      "learning_rate": 9.347789473684211e-05,
      "loss": 0.4246,
      "step": 44803
    },
    {
      "epoch": 2.5089035726285136,
      "grad_norm": 1.1307345628738403,
      "learning_rate": 9.347763157894738e-05,
      "loss": 0.3362,
      "step": 44804
    },
    {
      "epoch": 2.508959569940643,
      "grad_norm": 1.1523785591125488,
      "learning_rate": 9.347736842105263e-05,
      "loss": 0.3283,
      "step": 44805
    },
    {
      "epoch": 2.5090155672527716,
      "grad_norm": 0.9287214875221252,
      "learning_rate": 9.34771052631579e-05,
      "loss": 0.3207,
      "step": 44806
    },
    {
      "epoch": 2.509071564564901,
      "grad_norm": 0.9575591683387756,
      "learning_rate": 9.347684210526316e-05,
      "loss": 0.2962,
      "step": 44807
    },
    {
      "epoch": 2.5091275618770297,
      "grad_norm": 1.4418352842330933,
      "learning_rate": 9.347657894736842e-05,
      "loss": 0.4925,
      "step": 44808
    },
    {
      "epoch": 2.509183559189159,
      "grad_norm": 1.2317215204238892,
      "learning_rate": 9.347631578947368e-05,
      "loss": 0.345,
      "step": 44809
    },
    {
      "epoch": 2.5092395565012877,
      "grad_norm": 1.242807388305664,
      "learning_rate": 9.347605263157894e-05,
      "loss": 0.3959,
      "step": 44810
    },
    {
      "epoch": 2.509295553813417,
      "grad_norm": 1.2883909940719604,
      "learning_rate": 9.347578947368422e-05,
      "loss": 0.4667,
      "step": 44811
    },
    {
      "epoch": 2.5093515511255458,
      "grad_norm": 1.6157764196395874,
      "learning_rate": 9.347552631578948e-05,
      "loss": 0.6144,
      "step": 44812
    },
    {
      "epoch": 2.509407548437675,
      "grad_norm": 1.1221263408660889,
      "learning_rate": 9.347526315789475e-05,
      "loss": 0.3846,
      "step": 44813
    },
    {
      "epoch": 2.509463545749804,
      "grad_norm": 1.9636478424072266,
      "learning_rate": 9.3475e-05,
      "loss": 0.4537,
      "step": 44814
    },
    {
      "epoch": 2.5095195430619333,
      "grad_norm": 1.1219887733459473,
      "learning_rate": 9.347473684210527e-05,
      "loss": 0.4722,
      "step": 44815
    },
    {
      "epoch": 2.509575540374062,
      "grad_norm": 1.2659757137298584,
      "learning_rate": 9.347447368421053e-05,
      "loss": 0.4594,
      "step": 44816
    },
    {
      "epoch": 2.5096315376861913,
      "grad_norm": 1.1388801336288452,
      "learning_rate": 9.34742105263158e-05,
      "loss": 0.3963,
      "step": 44817
    },
    {
      "epoch": 2.50968753499832,
      "grad_norm": 1.4837017059326172,
      "learning_rate": 9.347394736842106e-05,
      "loss": 0.4176,
      "step": 44818
    },
    {
      "epoch": 2.5097435323104493,
      "grad_norm": 1.1181418895721436,
      "learning_rate": 9.347368421052632e-05,
      "loss": 0.3433,
      "step": 44819
    },
    {
      "epoch": 2.509799529622578,
      "grad_norm": 1.1035248041152954,
      "learning_rate": 9.347342105263158e-05,
      "loss": 0.3439,
      "step": 44820
    },
    {
      "epoch": 2.5098555269347074,
      "grad_norm": 1.2026946544647217,
      "learning_rate": 9.347315789473685e-05,
      "loss": 0.398,
      "step": 44821
    },
    {
      "epoch": 2.509911524246836,
      "grad_norm": 1.1365171670913696,
      "learning_rate": 9.347289473684211e-05,
      "loss": 0.3844,
      "step": 44822
    },
    {
      "epoch": 2.5099675215589654,
      "grad_norm": 1.1773253679275513,
      "learning_rate": 9.347263157894737e-05,
      "loss": 0.3002,
      "step": 44823
    },
    {
      "epoch": 2.510023518871094,
      "grad_norm": 1.6119105815887451,
      "learning_rate": 9.347236842105263e-05,
      "loss": 0.383,
      "step": 44824
    },
    {
      "epoch": 2.5100795161832234,
      "grad_norm": 1.4468108415603638,
      "learning_rate": 9.347210526315789e-05,
      "loss": 0.4419,
      "step": 44825
    },
    {
      "epoch": 2.510135513495352,
      "grad_norm": 1.131170630455017,
      "learning_rate": 9.347184210526317e-05,
      "loss": 0.4063,
      "step": 44826
    },
    {
      "epoch": 2.5101915108074815,
      "grad_norm": 0.9969378113746643,
      "learning_rate": 9.347157894736843e-05,
      "loss": 0.3209,
      "step": 44827
    },
    {
      "epoch": 2.51024750811961,
      "grad_norm": 1.2613555192947388,
      "learning_rate": 9.347131578947369e-05,
      "loss": 0.5424,
      "step": 44828
    },
    {
      "epoch": 2.5103035054317395,
      "grad_norm": 1.34848952293396,
      "learning_rate": 9.347105263157895e-05,
      "loss": 0.5583,
      "step": 44829
    },
    {
      "epoch": 2.510359502743868,
      "grad_norm": 1.1752198934555054,
      "learning_rate": 9.347078947368422e-05,
      "loss": 0.4375,
      "step": 44830
    },
    {
      "epoch": 2.5104155000559976,
      "grad_norm": 1.29131281375885,
      "learning_rate": 9.347052631578948e-05,
      "loss": 0.4499,
      "step": 44831
    },
    {
      "epoch": 2.510471497368126,
      "grad_norm": 1.0799561738967896,
      "learning_rate": 9.347026315789474e-05,
      "loss": 0.4558,
      "step": 44832
    },
    {
      "epoch": 2.5105274946802556,
      "grad_norm": 1.5851231813430786,
      "learning_rate": 9.347e-05,
      "loss": 0.3901,
      "step": 44833
    },
    {
      "epoch": 2.510583491992384,
      "grad_norm": 1.1060611009597778,
      "learning_rate": 9.346973684210527e-05,
      "loss": 0.3985,
      "step": 44834
    },
    {
      "epoch": 2.5106394893045136,
      "grad_norm": 1.2519795894622803,
      "learning_rate": 9.346947368421053e-05,
      "loss": 0.3983,
      "step": 44835
    },
    {
      "epoch": 2.510695486616642,
      "grad_norm": 1.2525273561477661,
      "learning_rate": 9.34692105263158e-05,
      "loss": 0.4165,
      "step": 44836
    },
    {
      "epoch": 2.5107514839287717,
      "grad_norm": 0.97767174243927,
      "learning_rate": 9.346894736842105e-05,
      "loss": 0.2777,
      "step": 44837
    },
    {
      "epoch": 2.5108074812409003,
      "grad_norm": 1.2371069192886353,
      "learning_rate": 9.346868421052632e-05,
      "loss": 0.4148,
      "step": 44838
    },
    {
      "epoch": 2.5108634785530297,
      "grad_norm": 1.32941472530365,
      "learning_rate": 9.346842105263158e-05,
      "loss": 0.615,
      "step": 44839
    },
    {
      "epoch": 2.5109194758651583,
      "grad_norm": 0.9924670457839966,
      "learning_rate": 9.346815789473686e-05,
      "loss": 0.3319,
      "step": 44840
    },
    {
      "epoch": 2.5109754731772878,
      "grad_norm": 1.2255840301513672,
      "learning_rate": 9.346789473684212e-05,
      "loss": 0.3632,
      "step": 44841
    },
    {
      "epoch": 2.5110314704894163,
      "grad_norm": 1.4669734239578247,
      "learning_rate": 9.346763157894736e-05,
      "loss": 0.5695,
      "step": 44842
    },
    {
      "epoch": 2.511087467801546,
      "grad_norm": 1.2088528871536255,
      "learning_rate": 9.346736842105264e-05,
      "loss": 0.3351,
      "step": 44843
    },
    {
      "epoch": 2.5111434651136744,
      "grad_norm": 1.3070580959320068,
      "learning_rate": 9.34671052631579e-05,
      "loss": 0.5017,
      "step": 44844
    },
    {
      "epoch": 2.511199462425804,
      "grad_norm": 1.03866708278656,
      "learning_rate": 9.346684210526317e-05,
      "loss": 0.3976,
      "step": 44845
    },
    {
      "epoch": 2.5112554597379324,
      "grad_norm": 1.590061902999878,
      "learning_rate": 9.346657894736841e-05,
      "loss": 0.28,
      "step": 44846
    },
    {
      "epoch": 2.511311457050062,
      "grad_norm": 1.2179813385009766,
      "learning_rate": 9.346631578947369e-05,
      "loss": 0.5433,
      "step": 44847
    },
    {
      "epoch": 2.5113674543621904,
      "grad_norm": 2.9194672107696533,
      "learning_rate": 9.346605263157895e-05,
      "loss": 0.4616,
      "step": 44848
    },
    {
      "epoch": 2.51142345167432,
      "grad_norm": 1.089234709739685,
      "learning_rate": 9.346578947368422e-05,
      "loss": 0.3061,
      "step": 44849
    },
    {
      "epoch": 2.5114794489864485,
      "grad_norm": 1.0947996377944946,
      "learning_rate": 9.346552631578948e-05,
      "loss": 0.3551,
      "step": 44850
    },
    {
      "epoch": 2.511535446298578,
      "grad_norm": 1.2092132568359375,
      "learning_rate": 9.346526315789474e-05,
      "loss": 0.3332,
      "step": 44851
    },
    {
      "epoch": 2.5115914436107065,
      "grad_norm": 1.2373971939086914,
      "learning_rate": 9.3465e-05,
      "loss": 0.3388,
      "step": 44852
    },
    {
      "epoch": 2.5116474409228355,
      "grad_norm": 8.125524520874023,
      "learning_rate": 9.346473684210527e-05,
      "loss": 0.538,
      "step": 44853
    },
    {
      "epoch": 2.5117034382349646,
      "grad_norm": 1.4691716432571411,
      "learning_rate": 9.346447368421053e-05,
      "loss": 0.4294,
      "step": 44854
    },
    {
      "epoch": 2.5117594355470936,
      "grad_norm": 1.2349040508270264,
      "learning_rate": 9.346421052631579e-05,
      "loss": 0.5085,
      "step": 44855
    },
    {
      "epoch": 2.5118154328592226,
      "grad_norm": 1.2856624126434326,
      "learning_rate": 9.346394736842105e-05,
      "loss": 0.3606,
      "step": 44856
    },
    {
      "epoch": 2.5118714301713516,
      "grad_norm": 1.2632639408111572,
      "learning_rate": 9.346368421052633e-05,
      "loss": 0.4337,
      "step": 44857
    },
    {
      "epoch": 2.5119274274834806,
      "grad_norm": 1.580068588256836,
      "learning_rate": 9.346342105263159e-05,
      "loss": 0.4892,
      "step": 44858
    },
    {
      "epoch": 2.5119834247956097,
      "grad_norm": 1.4663323163986206,
      "learning_rate": 9.346315789473685e-05,
      "loss": 0.4151,
      "step": 44859
    },
    {
      "epoch": 2.5120394221077387,
      "grad_norm": 1.3076223134994507,
      "learning_rate": 9.34628947368421e-05,
      "loss": 0.4489,
      "step": 44860
    },
    {
      "epoch": 2.5120954194198677,
      "grad_norm": 1.6174418926239014,
      "learning_rate": 9.346263157894737e-05,
      "loss": 0.4643,
      "step": 44861
    },
    {
      "epoch": 2.5121514167319967,
      "grad_norm": 1.0677871704101562,
      "learning_rate": 9.346236842105264e-05,
      "loss": 0.3172,
      "step": 44862
    },
    {
      "epoch": 2.5122074140441257,
      "grad_norm": 1.259846568107605,
      "learning_rate": 9.34621052631579e-05,
      "loss": 0.3394,
      "step": 44863
    },
    {
      "epoch": 2.5122634113562547,
      "grad_norm": 1.1495903730392456,
      "learning_rate": 9.346184210526316e-05,
      "loss": 0.3107,
      "step": 44864
    },
    {
      "epoch": 2.5123194086683838,
      "grad_norm": 1.021652102470398,
      "learning_rate": 9.346157894736842e-05,
      "loss": 0.3296,
      "step": 44865
    },
    {
      "epoch": 2.512375405980513,
      "grad_norm": 1.2996822595596313,
      "learning_rate": 9.346131578947369e-05,
      "loss": 0.4432,
      "step": 44866
    },
    {
      "epoch": 2.512431403292642,
      "grad_norm": 1.3068476915359497,
      "learning_rate": 9.346105263157895e-05,
      "loss": 0.4548,
      "step": 44867
    },
    {
      "epoch": 2.512487400604771,
      "grad_norm": 1.1853389739990234,
      "learning_rate": 9.346078947368422e-05,
      "loss": 0.4578,
      "step": 44868
    },
    {
      "epoch": 2.5125433979169,
      "grad_norm": 1.3115187883377075,
      "learning_rate": 9.346052631578947e-05,
      "loss": 0.4174,
      "step": 44869
    },
    {
      "epoch": 2.512599395229029,
      "grad_norm": 1.3815999031066895,
      "learning_rate": 9.346026315789474e-05,
      "loss": 0.3732,
      "step": 44870
    },
    {
      "epoch": 2.512655392541158,
      "grad_norm": 1.068616271018982,
      "learning_rate": 9.346e-05,
      "loss": 0.346,
      "step": 44871
    },
    {
      "epoch": 2.512711389853287,
      "grad_norm": 1.1265902519226074,
      "learning_rate": 9.345973684210528e-05,
      "loss": 0.3245,
      "step": 44872
    },
    {
      "epoch": 2.512767387165416,
      "grad_norm": 1.0503796339035034,
      "learning_rate": 9.345947368421054e-05,
      "loss": 0.3052,
      "step": 44873
    },
    {
      "epoch": 2.512823384477545,
      "grad_norm": 1.1018083095550537,
      "learning_rate": 9.34592105263158e-05,
      "loss": 0.4371,
      "step": 44874
    },
    {
      "epoch": 2.512879381789674,
      "grad_norm": 1.4471663236618042,
      "learning_rate": 9.345894736842106e-05,
      "loss": 0.4307,
      "step": 44875
    },
    {
      "epoch": 2.512935379101803,
      "grad_norm": 1.0819631814956665,
      "learning_rate": 9.345868421052632e-05,
      "loss": 0.4652,
      "step": 44876
    },
    {
      "epoch": 2.512991376413932,
      "grad_norm": 1.0855740308761597,
      "learning_rate": 9.345842105263159e-05,
      "loss": 0.3975,
      "step": 44877
    },
    {
      "epoch": 2.513047373726061,
      "grad_norm": 1.1854274272918701,
      "learning_rate": 9.345815789473685e-05,
      "loss": 0.3471,
      "step": 44878
    },
    {
      "epoch": 2.51310337103819,
      "grad_norm": 1.0236424207687378,
      "learning_rate": 9.345789473684211e-05,
      "loss": 0.3814,
      "step": 44879
    },
    {
      "epoch": 2.513159368350319,
      "grad_norm": 1.3927698135375977,
      "learning_rate": 9.345763157894737e-05,
      "loss": 0.4134,
      "step": 44880
    },
    {
      "epoch": 2.513215365662448,
      "grad_norm": 8.184261322021484,
      "learning_rate": 9.345736842105264e-05,
      "loss": 0.4522,
      "step": 44881
    },
    {
      "epoch": 2.513271362974577,
      "grad_norm": 2.255575180053711,
      "learning_rate": 9.34571052631579e-05,
      "loss": 0.5953,
      "step": 44882
    },
    {
      "epoch": 2.513327360286706,
      "grad_norm": 1.0129317045211792,
      "learning_rate": 9.345684210526316e-05,
      "loss": 0.3147,
      "step": 44883
    },
    {
      "epoch": 2.513383357598835,
      "grad_norm": 1.1146271228790283,
      "learning_rate": 9.345657894736842e-05,
      "loss": 0.5345,
      "step": 44884
    },
    {
      "epoch": 2.513439354910964,
      "grad_norm": 1.20524263381958,
      "learning_rate": 9.34563157894737e-05,
      "loss": 0.4027,
      "step": 44885
    },
    {
      "epoch": 2.513495352223093,
      "grad_norm": 1.0501710176467896,
      "learning_rate": 9.345605263157895e-05,
      "loss": 0.3371,
      "step": 44886
    },
    {
      "epoch": 2.513551349535222,
      "grad_norm": 1.3221904039382935,
      "learning_rate": 9.345578947368421e-05,
      "loss": 0.4243,
      "step": 44887
    },
    {
      "epoch": 2.513607346847351,
      "grad_norm": 1.3419839143753052,
      "learning_rate": 9.345552631578947e-05,
      "loss": 0.3276,
      "step": 44888
    },
    {
      "epoch": 2.5136633441594802,
      "grad_norm": 0.968989908695221,
      "learning_rate": 9.345526315789475e-05,
      "loss": 0.3418,
      "step": 44889
    },
    {
      "epoch": 2.5137193414716092,
      "grad_norm": 1.2165015935897827,
      "learning_rate": 9.3455e-05,
      "loss": 0.4474,
      "step": 44890
    },
    {
      "epoch": 2.5137753387837383,
      "grad_norm": 1.1653826236724854,
      "learning_rate": 9.345473684210528e-05,
      "loss": 0.4719,
      "step": 44891
    },
    {
      "epoch": 2.5138313360958673,
      "grad_norm": 1.0929332971572876,
      "learning_rate": 9.345447368421053e-05,
      "loss": 0.4604,
      "step": 44892
    },
    {
      "epoch": 2.5138873334079963,
      "grad_norm": 1.0278501510620117,
      "learning_rate": 9.345421052631578e-05,
      "loss": 0.4443,
      "step": 44893
    },
    {
      "epoch": 2.5139433307201253,
      "grad_norm": 1.2402970790863037,
      "learning_rate": 9.345394736842106e-05,
      "loss": 0.4572,
      "step": 44894
    },
    {
      "epoch": 2.5139993280322543,
      "grad_norm": 1.3332836627960205,
      "learning_rate": 9.345368421052632e-05,
      "loss": 0.3434,
      "step": 44895
    },
    {
      "epoch": 2.5140553253443834,
      "grad_norm": 1.216914176940918,
      "learning_rate": 9.345342105263159e-05,
      "loss": 0.3555,
      "step": 44896
    },
    {
      "epoch": 2.5141113226565124,
      "grad_norm": 1.308079481124878,
      "learning_rate": 9.345315789473684e-05,
      "loss": 0.4229,
      "step": 44897
    },
    {
      "epoch": 2.5141673199686414,
      "grad_norm": 1.1437238454818726,
      "learning_rate": 9.345289473684211e-05,
      "loss": 0.4043,
      "step": 44898
    },
    {
      "epoch": 2.5142233172807704,
      "grad_norm": 1.3466852903366089,
      "learning_rate": 9.345263157894737e-05,
      "loss": 0.4086,
      "step": 44899
    },
    {
      "epoch": 2.5142793145928994,
      "grad_norm": 1.2115637063980103,
      "learning_rate": 9.345236842105264e-05,
      "loss": 0.567,
      "step": 44900
    },
    {
      "epoch": 2.5143353119050285,
      "grad_norm": 1.2190706729888916,
      "learning_rate": 9.345210526315789e-05,
      "loss": 0.6623,
      "step": 44901
    },
    {
      "epoch": 2.5143913092171575,
      "grad_norm": 1.0553839206695557,
      "learning_rate": 9.345184210526316e-05,
      "loss": 0.4239,
      "step": 44902
    },
    {
      "epoch": 2.5144473065292865,
      "grad_norm": 1.414591908454895,
      "learning_rate": 9.345157894736842e-05,
      "loss": 0.3563,
      "step": 44903
    },
    {
      "epoch": 2.5145033038414155,
      "grad_norm": 1.1586596965789795,
      "learning_rate": 9.34513157894737e-05,
      "loss": 0.4577,
      "step": 44904
    },
    {
      "epoch": 2.5145593011535445,
      "grad_norm": 1.2129076719284058,
      "learning_rate": 9.345105263157896e-05,
      "loss": 0.6084,
      "step": 44905
    },
    {
      "epoch": 2.5146152984656736,
      "grad_norm": 1.448001503944397,
      "learning_rate": 9.345078947368422e-05,
      "loss": 0.4939,
      "step": 44906
    },
    {
      "epoch": 2.5146712957778026,
      "grad_norm": 1.0205128192901611,
      "learning_rate": 9.345052631578948e-05,
      "loss": 0.3671,
      "step": 44907
    },
    {
      "epoch": 2.5147272930899316,
      "grad_norm": 1.4580501317977905,
      "learning_rate": 9.345026315789475e-05,
      "loss": 0.4371,
      "step": 44908
    },
    {
      "epoch": 2.5147832904020606,
      "grad_norm": 0.9909213185310364,
      "learning_rate": 9.345000000000001e-05,
      "loss": 0.3443,
      "step": 44909
    },
    {
      "epoch": 2.5148392877141896,
      "grad_norm": 1.4379419088363647,
      "learning_rate": 9.344973684210527e-05,
      "loss": 0.3435,
      "step": 44910
    },
    {
      "epoch": 2.5148952850263186,
      "grad_norm": 1.00141179561615,
      "learning_rate": 9.344947368421053e-05,
      "loss": 0.3569,
      "step": 44911
    },
    {
      "epoch": 2.5149512823384477,
      "grad_norm": 1.019567608833313,
      "learning_rate": 9.344921052631579e-05,
      "loss": 0.3208,
      "step": 44912
    },
    {
      "epoch": 2.5150072796505767,
      "grad_norm": 1.207808494567871,
      "learning_rate": 9.344894736842106e-05,
      "loss": 0.4349,
      "step": 44913
    },
    {
      "epoch": 2.5150632769627057,
      "grad_norm": 1.4679515361785889,
      "learning_rate": 9.344868421052632e-05,
      "loss": 0.3923,
      "step": 44914
    },
    {
      "epoch": 2.5151192742748347,
      "grad_norm": 1.0824780464172363,
      "learning_rate": 9.344842105263158e-05,
      "loss": 0.2899,
      "step": 44915
    },
    {
      "epoch": 2.5151752715869637,
      "grad_norm": 1.0728927850723267,
      "learning_rate": 9.344815789473684e-05,
      "loss": 0.3198,
      "step": 44916
    },
    {
      "epoch": 2.5152312688990928,
      "grad_norm": 1.2754501104354858,
      "learning_rate": 9.344789473684211e-05,
      "loss": 0.4084,
      "step": 44917
    },
    {
      "epoch": 2.515287266211222,
      "grad_norm": 1.4581232070922852,
      "learning_rate": 9.344763157894737e-05,
      "loss": 0.487,
      "step": 44918
    },
    {
      "epoch": 2.515343263523351,
      "grad_norm": 0.9967120289802551,
      "learning_rate": 9.344736842105263e-05,
      "loss": 0.4334,
      "step": 44919
    },
    {
      "epoch": 2.51539926083548,
      "grad_norm": 1.0986638069152832,
      "learning_rate": 9.344710526315789e-05,
      "loss": 0.439,
      "step": 44920
    },
    {
      "epoch": 2.515455258147609,
      "grad_norm": 1.3770689964294434,
      "learning_rate": 9.344684210526317e-05,
      "loss": 0.4033,
      "step": 44921
    },
    {
      "epoch": 2.515511255459738,
      "grad_norm": 0.9774295091629028,
      "learning_rate": 9.344657894736843e-05,
      "loss": 0.3056,
      "step": 44922
    },
    {
      "epoch": 2.515567252771867,
      "grad_norm": 1.3876203298568726,
      "learning_rate": 9.34463157894737e-05,
      "loss": 0.5283,
      "step": 44923
    },
    {
      "epoch": 2.515623250083996,
      "grad_norm": 1.1342518329620361,
      "learning_rate": 9.344605263157894e-05,
      "loss": 0.4154,
      "step": 44924
    },
    {
      "epoch": 2.515679247396125,
      "grad_norm": 3.111009120941162,
      "learning_rate": 9.344578947368422e-05,
      "loss": 0.3657,
      "step": 44925
    },
    {
      "epoch": 2.515735244708254,
      "grad_norm": 1.144946575164795,
      "learning_rate": 9.344552631578948e-05,
      "loss": 0.3353,
      "step": 44926
    },
    {
      "epoch": 2.515791242020383,
      "grad_norm": 0.9770836234092712,
      "learning_rate": 9.344526315789475e-05,
      "loss": 0.323,
      "step": 44927
    },
    {
      "epoch": 2.515847239332512,
      "grad_norm": 1.0280795097351074,
      "learning_rate": 9.344500000000001e-05,
      "loss": 0.334,
      "step": 44928
    },
    {
      "epoch": 2.515903236644641,
      "grad_norm": 1.0440059900283813,
      "learning_rate": 9.344473684210526e-05,
      "loss": 0.3485,
      "step": 44929
    },
    {
      "epoch": 2.51595923395677,
      "grad_norm": 1.4271368980407715,
      "learning_rate": 9.344447368421053e-05,
      "loss": 0.376,
      "step": 44930
    },
    {
      "epoch": 2.516015231268899,
      "grad_norm": 1.1543761491775513,
      "learning_rate": 9.344421052631579e-05,
      "loss": 0.3957,
      "step": 44931
    },
    {
      "epoch": 2.516071228581028,
      "grad_norm": 1.255721092224121,
      "learning_rate": 9.344394736842106e-05,
      "loss": 0.5025,
      "step": 44932
    },
    {
      "epoch": 2.516127225893157,
      "grad_norm": 1.1022852659225464,
      "learning_rate": 9.344368421052632e-05,
      "loss": 0.3901,
      "step": 44933
    },
    {
      "epoch": 2.516183223205286,
      "grad_norm": 1.430487036705017,
      "learning_rate": 9.344342105263158e-05,
      "loss": 0.4536,
      "step": 44934
    },
    {
      "epoch": 2.516239220517415,
      "grad_norm": 1.1101151704788208,
      "learning_rate": 9.344315789473684e-05,
      "loss": 0.3305,
      "step": 44935
    },
    {
      "epoch": 2.516295217829544,
      "grad_norm": 1.1055845022201538,
      "learning_rate": 9.344289473684212e-05,
      "loss": 0.3563,
      "step": 44936
    },
    {
      "epoch": 2.516351215141673,
      "grad_norm": 1.3700309991836548,
      "learning_rate": 9.344263157894738e-05,
      "loss": 0.388,
      "step": 44937
    },
    {
      "epoch": 2.516407212453802,
      "grad_norm": 1.1935478448867798,
      "learning_rate": 9.344236842105264e-05,
      "loss": 0.5133,
      "step": 44938
    },
    {
      "epoch": 2.516463209765931,
      "grad_norm": 1.3323646783828735,
      "learning_rate": 9.34421052631579e-05,
      "loss": 0.4344,
      "step": 44939
    },
    {
      "epoch": 2.51651920707806,
      "grad_norm": 1.147402048110962,
      "learning_rate": 9.344184210526317e-05,
      "loss": 0.3988,
      "step": 44940
    },
    {
      "epoch": 2.5165752043901892,
      "grad_norm": 1.0571316480636597,
      "learning_rate": 9.344157894736843e-05,
      "loss": 0.2771,
      "step": 44941
    },
    {
      "epoch": 2.5166312017023182,
      "grad_norm": 1.1549065113067627,
      "learning_rate": 9.344131578947369e-05,
      "loss": 0.4805,
      "step": 44942
    },
    {
      "epoch": 2.5166871990144473,
      "grad_norm": 1.6336346864700317,
      "learning_rate": 9.344105263157895e-05,
      "loss": 0.4339,
      "step": 44943
    },
    {
      "epoch": 2.5167431963265763,
      "grad_norm": 1.232505440711975,
      "learning_rate": 9.344078947368422e-05,
      "loss": 0.4188,
      "step": 44944
    },
    {
      "epoch": 2.5167991936387053,
      "grad_norm": 1.0953078269958496,
      "learning_rate": 9.344052631578948e-05,
      "loss": 0.4922,
      "step": 44945
    },
    {
      "epoch": 2.5168551909508343,
      "grad_norm": 1.0453640222549438,
      "learning_rate": 9.344026315789474e-05,
      "loss": 0.3212,
      "step": 44946
    },
    {
      "epoch": 2.5169111882629633,
      "grad_norm": 1.1152838468551636,
      "learning_rate": 9.344e-05,
      "loss": 0.4817,
      "step": 44947
    },
    {
      "epoch": 2.5169671855750924,
      "grad_norm": 0.9481313824653625,
      "learning_rate": 9.343973684210526e-05,
      "loss": 0.3571,
      "step": 44948
    },
    {
      "epoch": 2.5170231828872214,
      "grad_norm": 1.3109378814697266,
      "learning_rate": 9.343947368421053e-05,
      "loss": 0.4554,
      "step": 44949
    },
    {
      "epoch": 2.5170791801993504,
      "grad_norm": 1.0466681718826294,
      "learning_rate": 9.343921052631579e-05,
      "loss": 0.3763,
      "step": 44950
    },
    {
      "epoch": 2.5171351775114794,
      "grad_norm": 1.1284700632095337,
      "learning_rate": 9.343894736842107e-05,
      "loss": 0.351,
      "step": 44951
    },
    {
      "epoch": 2.5171911748236084,
      "grad_norm": 1.2319542169570923,
      "learning_rate": 9.343868421052631e-05,
      "loss": 0.4552,
      "step": 44952
    },
    {
      "epoch": 2.5172471721357375,
      "grad_norm": 1.0170358419418335,
      "learning_rate": 9.343842105263159e-05,
      "loss": 0.3283,
      "step": 44953
    },
    {
      "epoch": 2.5173031694478665,
      "grad_norm": 1.0471830368041992,
      "learning_rate": 9.343815789473685e-05,
      "loss": 0.3781,
      "step": 44954
    },
    {
      "epoch": 2.5173591667599955,
      "grad_norm": 1.6155041456222534,
      "learning_rate": 9.343789473684212e-05,
      "loss": 0.4377,
      "step": 44955
    },
    {
      "epoch": 2.5174151640721245,
      "grad_norm": 1.072693109512329,
      "learning_rate": 9.343763157894736e-05,
      "loss": 0.4797,
      "step": 44956
    },
    {
      "epoch": 2.5174711613842535,
      "grad_norm": 1.1484203338623047,
      "learning_rate": 9.343736842105264e-05,
      "loss": 0.3885,
      "step": 44957
    },
    {
      "epoch": 2.5175271586963825,
      "grad_norm": 1.503772258758545,
      "learning_rate": 9.34371052631579e-05,
      "loss": 0.4369,
      "step": 44958
    },
    {
      "epoch": 2.5175831560085116,
      "grad_norm": 1.1823770999908447,
      "learning_rate": 9.343684210526317e-05,
      "loss": 0.3695,
      "step": 44959
    },
    {
      "epoch": 2.5176391533206406,
      "grad_norm": 1.205325961112976,
      "learning_rate": 9.343657894736843e-05,
      "loss": 0.3792,
      "step": 44960
    },
    {
      "epoch": 2.5176951506327696,
      "grad_norm": 1.4646451473236084,
      "learning_rate": 9.343631578947369e-05,
      "loss": 0.491,
      "step": 44961
    },
    {
      "epoch": 2.5177511479448986,
      "grad_norm": 1.1776790618896484,
      "learning_rate": 9.343605263157895e-05,
      "loss": 0.3946,
      "step": 44962
    },
    {
      "epoch": 2.5178071452570276,
      "grad_norm": 0.9953280687332153,
      "learning_rate": 9.343578947368421e-05,
      "loss": 0.4312,
      "step": 44963
    },
    {
      "epoch": 2.5178631425691567,
      "grad_norm": 1.0994231700897217,
      "learning_rate": 9.343552631578948e-05,
      "loss": 0.4054,
      "step": 44964
    },
    {
      "epoch": 2.5179191398812857,
      "grad_norm": 0.9563860893249512,
      "learning_rate": 9.343526315789474e-05,
      "loss": 0.2767,
      "step": 44965
    },
    {
      "epoch": 2.5179751371934147,
      "grad_norm": 1.2504534721374512,
      "learning_rate": 9.3435e-05,
      "loss": 0.3696,
      "step": 44966
    },
    {
      "epoch": 2.5180311345055437,
      "grad_norm": 0.9428294897079468,
      "learning_rate": 9.343473684210526e-05,
      "loss": 0.3675,
      "step": 44967
    },
    {
      "epoch": 2.5180871318176727,
      "grad_norm": 1.3348417282104492,
      "learning_rate": 9.343447368421054e-05,
      "loss": 0.4198,
      "step": 44968
    },
    {
      "epoch": 2.5181431291298018,
      "grad_norm": 1.2012553215026855,
      "learning_rate": 9.34342105263158e-05,
      "loss": 0.4658,
      "step": 44969
    },
    {
      "epoch": 2.518199126441931,
      "grad_norm": 1.1559463739395142,
      "learning_rate": 9.343394736842105e-05,
      "loss": 0.3906,
      "step": 44970
    },
    {
      "epoch": 2.51825512375406,
      "grad_norm": 1.1877522468566895,
      "learning_rate": 9.343368421052631e-05,
      "loss": 0.4526,
      "step": 44971
    },
    {
      "epoch": 2.518311121066189,
      "grad_norm": 1.5210577249526978,
      "learning_rate": 9.343342105263159e-05,
      "loss": 0.3942,
      "step": 44972
    },
    {
      "epoch": 2.518367118378318,
      "grad_norm": 1.4321980476379395,
      "learning_rate": 9.343315789473685e-05,
      "loss": 0.4002,
      "step": 44973
    },
    {
      "epoch": 2.518423115690447,
      "grad_norm": 1.023194670677185,
      "learning_rate": 9.343289473684211e-05,
      "loss": 0.3982,
      "step": 44974
    },
    {
      "epoch": 2.518479113002576,
      "grad_norm": 1.208418607711792,
      "learning_rate": 9.343263157894737e-05,
      "loss": 0.3328,
      "step": 44975
    },
    {
      "epoch": 2.518535110314705,
      "grad_norm": 1.31124746799469,
      "learning_rate": 9.343236842105264e-05,
      "loss": 0.3582,
      "step": 44976
    },
    {
      "epoch": 2.518591107626834,
      "grad_norm": 1.5660958290100098,
      "learning_rate": 9.34321052631579e-05,
      "loss": 0.3897,
      "step": 44977
    },
    {
      "epoch": 2.518647104938963,
      "grad_norm": 1.4612643718719482,
      "learning_rate": 9.343184210526317e-05,
      "loss": 0.4789,
      "step": 44978
    },
    {
      "epoch": 2.518703102251092,
      "grad_norm": 1.2671443223953247,
      "learning_rate": 9.343157894736842e-05,
      "loss": 0.3856,
      "step": 44979
    },
    {
      "epoch": 2.518759099563221,
      "grad_norm": 1.078467607498169,
      "learning_rate": 9.343131578947368e-05,
      "loss": 0.2855,
      "step": 44980
    },
    {
      "epoch": 2.51881509687535,
      "grad_norm": 1.160021424293518,
      "learning_rate": 9.343105263157895e-05,
      "loss": 0.4486,
      "step": 44981
    },
    {
      "epoch": 2.518871094187479,
      "grad_norm": 1.1333954334259033,
      "learning_rate": 9.343078947368421e-05,
      "loss": 0.4205,
      "step": 44982
    },
    {
      "epoch": 2.518927091499608,
      "grad_norm": 1.1214267015457153,
      "learning_rate": 9.343052631578949e-05,
      "loss": 0.3243,
      "step": 44983
    },
    {
      "epoch": 2.518983088811737,
      "grad_norm": 1.0001105070114136,
      "learning_rate": 9.343026315789473e-05,
      "loss": 0.3383,
      "step": 44984
    },
    {
      "epoch": 2.519039086123866,
      "grad_norm": 1.1627229452133179,
      "learning_rate": 9.343e-05,
      "loss": 0.4093,
      "step": 44985
    },
    {
      "epoch": 2.519095083435995,
      "grad_norm": 1.2763195037841797,
      "learning_rate": 9.342973684210526e-05,
      "loss": 0.525,
      "step": 44986
    },
    {
      "epoch": 2.519151080748124,
      "grad_norm": 1.234586477279663,
      "learning_rate": 9.342947368421054e-05,
      "loss": 0.4677,
      "step": 44987
    },
    {
      "epoch": 2.519207078060253,
      "grad_norm": 1.3622779846191406,
      "learning_rate": 9.34292105263158e-05,
      "loss": 0.5741,
      "step": 44988
    },
    {
      "epoch": 2.519263075372382,
      "grad_norm": 1.3100652694702148,
      "learning_rate": 9.342894736842106e-05,
      "loss": 0.6675,
      "step": 44989
    },
    {
      "epoch": 2.519319072684511,
      "grad_norm": 1.146007776260376,
      "learning_rate": 9.342868421052632e-05,
      "loss": 0.3324,
      "step": 44990
    },
    {
      "epoch": 2.51937506999664,
      "grad_norm": 1.2564313411712646,
      "learning_rate": 9.342842105263159e-05,
      "loss": 0.4336,
      "step": 44991
    },
    {
      "epoch": 2.519431067308769,
      "grad_norm": 1.1478283405303955,
      "learning_rate": 9.342815789473685e-05,
      "loss": 0.3981,
      "step": 44992
    },
    {
      "epoch": 2.519487064620898,
      "grad_norm": 1.2597934007644653,
      "learning_rate": 9.342789473684211e-05,
      "loss": 0.4504,
      "step": 44993
    },
    {
      "epoch": 2.5195430619330272,
      "grad_norm": 1.3104852437973022,
      "learning_rate": 9.342763157894737e-05,
      "loss": 0.4432,
      "step": 44994
    },
    {
      "epoch": 2.5195990592451563,
      "grad_norm": 1.2330958843231201,
      "learning_rate": 9.342736842105264e-05,
      "loss": 0.3808,
      "step": 44995
    },
    {
      "epoch": 2.5196550565572853,
      "grad_norm": 1.0490154027938843,
      "learning_rate": 9.34271052631579e-05,
      "loss": 0.3291,
      "step": 44996
    },
    {
      "epoch": 2.5197110538694143,
      "grad_norm": 1.1007615327835083,
      "learning_rate": 9.342684210526316e-05,
      "loss": 0.4098,
      "step": 44997
    },
    {
      "epoch": 2.5197670511815433,
      "grad_norm": 1.170037865638733,
      "learning_rate": 9.342657894736842e-05,
      "loss": 0.3506,
      "step": 44998
    },
    {
      "epoch": 2.5198230484936723,
      "grad_norm": 2.0011634826660156,
      "learning_rate": 9.342631578947368e-05,
      "loss": 0.4867,
      "step": 44999
    },
    {
      "epoch": 2.5198790458058014,
      "grad_norm": 1.2240971326828003,
      "learning_rate": 9.342605263157896e-05,
      "loss": 0.4368,
      "step": 45000
    },
    {
      "epoch": 2.5199350431179304,
      "grad_norm": 1.1403335332870483,
      "learning_rate": 9.342578947368421e-05,
      "loss": 0.4468,
      "step": 45001
    },
    {
      "epoch": 2.5199910404300594,
      "grad_norm": 1.1571733951568604,
      "learning_rate": 9.342552631578947e-05,
      "loss": 0.3807,
      "step": 45002
    },
    {
      "epoch": 2.5200470377421884,
      "grad_norm": 0.9375881552696228,
      "learning_rate": 9.342526315789473e-05,
      "loss": 0.3517,
      "step": 45003
    },
    {
      "epoch": 2.5201030350543174,
      "grad_norm": 1.1599394083023071,
      "learning_rate": 9.342500000000001e-05,
      "loss": 0.436,
      "step": 45004
    },
    {
      "epoch": 2.5201590323664464,
      "grad_norm": 6.312375545501709,
      "learning_rate": 9.342473684210527e-05,
      "loss": 0.3994,
      "step": 45005
    },
    {
      "epoch": 2.5202150296785755,
      "grad_norm": 1.3332339525222778,
      "learning_rate": 9.342447368421054e-05,
      "loss": 0.5155,
      "step": 45006
    },
    {
      "epoch": 2.5202710269907045,
      "grad_norm": 1.2522649765014648,
      "learning_rate": 9.342421052631579e-05,
      "loss": 0.5232,
      "step": 45007
    },
    {
      "epoch": 2.5203270243028335,
      "grad_norm": 1.0837116241455078,
      "learning_rate": 9.342394736842106e-05,
      "loss": 0.3897,
      "step": 45008
    },
    {
      "epoch": 2.5203830216149625,
      "grad_norm": 1.2225641012191772,
      "learning_rate": 9.342368421052632e-05,
      "loss": 0.4053,
      "step": 45009
    },
    {
      "epoch": 2.5204390189270915,
      "grad_norm": 1.3245056867599487,
      "learning_rate": 9.342342105263159e-05,
      "loss": 0.3482,
      "step": 45010
    },
    {
      "epoch": 2.5204950162392206,
      "grad_norm": 1.0236687660217285,
      "learning_rate": 9.342315789473684e-05,
      "loss": 0.4086,
      "step": 45011
    },
    {
      "epoch": 2.5205510135513496,
      "grad_norm": 1.4843497276306152,
      "learning_rate": 9.342289473684211e-05,
      "loss": 0.5144,
      "step": 45012
    },
    {
      "epoch": 2.5206070108634786,
      "grad_norm": 1.1718429327011108,
      "learning_rate": 9.342263157894737e-05,
      "loss": 0.3906,
      "step": 45013
    },
    {
      "epoch": 2.5206630081756076,
      "grad_norm": 1.0507522821426392,
      "learning_rate": 9.342236842105263e-05,
      "loss": 0.3336,
      "step": 45014
    },
    {
      "epoch": 2.5207190054877366,
      "grad_norm": 1.1859264373779297,
      "learning_rate": 9.34221052631579e-05,
      "loss": 0.3965,
      "step": 45015
    },
    {
      "epoch": 2.5207750027998657,
      "grad_norm": 0.9355223178863525,
      "learning_rate": 9.342184210526315e-05,
      "loss": 0.3265,
      "step": 45016
    },
    {
      "epoch": 2.5208310001119947,
      "grad_norm": 1.4237927198410034,
      "learning_rate": 9.342157894736842e-05,
      "loss": 0.3464,
      "step": 45017
    },
    {
      "epoch": 2.5208869974241237,
      "grad_norm": 0.9275553226470947,
      "learning_rate": 9.342131578947368e-05,
      "loss": 0.3218,
      "step": 45018
    },
    {
      "epoch": 2.5209429947362527,
      "grad_norm": 1.118915319442749,
      "learning_rate": 9.342105263157896e-05,
      "loss": 0.3699,
      "step": 45019
    },
    {
      "epoch": 2.5209989920483817,
      "grad_norm": 1.1228718757629395,
      "learning_rate": 9.342078947368422e-05,
      "loss": 0.3493,
      "step": 45020
    },
    {
      "epoch": 2.5210549893605108,
      "grad_norm": 1.3951319456100464,
      "learning_rate": 9.342052631578948e-05,
      "loss": 0.3988,
      "step": 45021
    },
    {
      "epoch": 2.5211109866726398,
      "grad_norm": 1.1483542919158936,
      "learning_rate": 9.342026315789474e-05,
      "loss": 0.4051,
      "step": 45022
    },
    {
      "epoch": 2.521166983984769,
      "grad_norm": 1.1666837930679321,
      "learning_rate": 9.342000000000001e-05,
      "loss": 0.4742,
      "step": 45023
    },
    {
      "epoch": 2.521222981296898,
      "grad_norm": 1.1819350719451904,
      "learning_rate": 9.341973684210527e-05,
      "loss": 0.2938,
      "step": 45024
    },
    {
      "epoch": 2.521278978609027,
      "grad_norm": 1.0625483989715576,
      "learning_rate": 9.341947368421053e-05,
      "loss": 0.4375,
      "step": 45025
    },
    {
      "epoch": 2.521334975921156,
      "grad_norm": 1.3652863502502441,
      "learning_rate": 9.341921052631579e-05,
      "loss": 0.4873,
      "step": 45026
    },
    {
      "epoch": 2.521390973233285,
      "grad_norm": 0.968160092830658,
      "learning_rate": 9.341894736842106e-05,
      "loss": 0.327,
      "step": 45027
    },
    {
      "epoch": 2.521446970545414,
      "grad_norm": 1.3170801401138306,
      "learning_rate": 9.341868421052632e-05,
      "loss": 0.4211,
      "step": 45028
    },
    {
      "epoch": 2.521502967857543,
      "grad_norm": 0.946506917476654,
      "learning_rate": 9.341842105263158e-05,
      "loss": 0.4307,
      "step": 45029
    },
    {
      "epoch": 2.521558965169672,
      "grad_norm": 1.1729804277420044,
      "learning_rate": 9.341815789473684e-05,
      "loss": 0.4079,
      "step": 45030
    },
    {
      "epoch": 2.521614962481801,
      "grad_norm": 1.6278482675552368,
      "learning_rate": 9.34178947368421e-05,
      "loss": 0.493,
      "step": 45031
    },
    {
      "epoch": 2.52167095979393,
      "grad_norm": 1.1343801021575928,
      "learning_rate": 9.341763157894737e-05,
      "loss": 0.3559,
      "step": 45032
    },
    {
      "epoch": 2.521726957106059,
      "grad_norm": 1.1425701379776,
      "learning_rate": 9.341736842105263e-05,
      "loss": 0.34,
      "step": 45033
    },
    {
      "epoch": 2.521782954418188,
      "grad_norm": 1.3116127252578735,
      "learning_rate": 9.34171052631579e-05,
      "loss": 0.3707,
      "step": 45034
    },
    {
      "epoch": 2.521838951730317,
      "grad_norm": 1.2800461053848267,
      "learning_rate": 9.341684210526315e-05,
      "loss": 0.3339,
      "step": 45035
    },
    {
      "epoch": 2.521894949042446,
      "grad_norm": 1.0552483797073364,
      "learning_rate": 9.341657894736843e-05,
      "loss": 0.4507,
      "step": 45036
    },
    {
      "epoch": 2.521950946354575,
      "grad_norm": 1.2434345483779907,
      "learning_rate": 9.341631578947369e-05,
      "loss": 0.4956,
      "step": 45037
    },
    {
      "epoch": 2.522006943666704,
      "grad_norm": 1.1293774843215942,
      "learning_rate": 9.341605263157896e-05,
      "loss": 0.3899,
      "step": 45038
    },
    {
      "epoch": 2.522062940978833,
      "grad_norm": 1.1378203630447388,
      "learning_rate": 9.34157894736842e-05,
      "loss": 0.4032,
      "step": 45039
    },
    {
      "epoch": 2.522118938290962,
      "grad_norm": 1.2980105876922607,
      "learning_rate": 9.341552631578948e-05,
      "loss": 0.3017,
      "step": 45040
    },
    {
      "epoch": 2.522174935603091,
      "grad_norm": 1.2770673036575317,
      "learning_rate": 9.341526315789474e-05,
      "loss": 0.4895,
      "step": 45041
    },
    {
      "epoch": 2.52223093291522,
      "grad_norm": 1.2009093761444092,
      "learning_rate": 9.341500000000001e-05,
      "loss": 0.3455,
      "step": 45042
    },
    {
      "epoch": 2.522286930227349,
      "grad_norm": 0.9977799654006958,
      "learning_rate": 9.341473684210527e-05,
      "loss": 0.3911,
      "step": 45043
    },
    {
      "epoch": 2.522342927539478,
      "grad_norm": 1.513598918914795,
      "learning_rate": 9.341447368421053e-05,
      "loss": 0.4044,
      "step": 45044
    },
    {
      "epoch": 2.522398924851607,
      "grad_norm": 0.9576770067214966,
      "learning_rate": 9.341421052631579e-05,
      "loss": 0.358,
      "step": 45045
    },
    {
      "epoch": 2.5224549221637362,
      "grad_norm": 1.1021870374679565,
      "learning_rate": 9.341394736842107e-05,
      "loss": 0.3989,
      "step": 45046
    },
    {
      "epoch": 2.5225109194758653,
      "grad_norm": 1.2287181615829468,
      "learning_rate": 9.341368421052632e-05,
      "loss": 0.3621,
      "step": 45047
    },
    {
      "epoch": 2.5225669167879943,
      "grad_norm": 1.1992675065994263,
      "learning_rate": 9.341342105263158e-05,
      "loss": 0.3832,
      "step": 45048
    },
    {
      "epoch": 2.5226229141001233,
      "grad_norm": 1.4756604433059692,
      "learning_rate": 9.341315789473684e-05,
      "loss": 0.3589,
      "step": 45049
    },
    {
      "epoch": 2.5226789114122523,
      "grad_norm": 1.2160685062408447,
      "learning_rate": 9.34128947368421e-05,
      "loss": 0.4059,
      "step": 45050
    },
    {
      "epoch": 2.5227349087243813,
      "grad_norm": 1.4132702350616455,
      "learning_rate": 9.341263157894738e-05,
      "loss": 0.6246,
      "step": 45051
    },
    {
      "epoch": 2.5227909060365103,
      "grad_norm": 1.3800989389419556,
      "learning_rate": 9.341236842105264e-05,
      "loss": 0.6802,
      "step": 45052
    },
    {
      "epoch": 2.5228469033486394,
      "grad_norm": 1.4345715045928955,
      "learning_rate": 9.34121052631579e-05,
      "loss": 0.4364,
      "step": 45053
    },
    {
      "epoch": 2.5229029006607684,
      "grad_norm": 0.9079887270927429,
      "learning_rate": 9.341184210526316e-05,
      "loss": 0.3043,
      "step": 45054
    },
    {
      "epoch": 2.5229588979728974,
      "grad_norm": 1.3440006971359253,
      "learning_rate": 9.341157894736843e-05,
      "loss": 0.5654,
      "step": 45055
    },
    {
      "epoch": 2.5230148952850264,
      "grad_norm": 1.4064229726791382,
      "learning_rate": 9.341131578947369e-05,
      "loss": 0.5295,
      "step": 45056
    },
    {
      "epoch": 2.5230708925971554,
      "grad_norm": 1.1477257013320923,
      "learning_rate": 9.341105263157895e-05,
      "loss": 0.4029,
      "step": 45057
    },
    {
      "epoch": 2.5231268899092845,
      "grad_norm": 1.649611234664917,
      "learning_rate": 9.341078947368421e-05,
      "loss": 0.3661,
      "step": 45058
    },
    {
      "epoch": 2.5231828872214135,
      "grad_norm": 1.430660367012024,
      "learning_rate": 9.341052631578948e-05,
      "loss": 0.53,
      "step": 45059
    },
    {
      "epoch": 2.5232388845335425,
      "grad_norm": 0.9649778604507446,
      "learning_rate": 9.341026315789474e-05,
      "loss": 0.3671,
      "step": 45060
    },
    {
      "epoch": 2.5232948818456715,
      "grad_norm": 1.4579644203186035,
      "learning_rate": 9.341000000000002e-05,
      "loss": 0.4924,
      "step": 45061
    },
    {
      "epoch": 2.5233508791578005,
      "grad_norm": 1.1260203123092651,
      "learning_rate": 9.340973684210526e-05,
      "loss": 0.3724,
      "step": 45062
    },
    {
      "epoch": 2.5234068764699296,
      "grad_norm": 1.1087557077407837,
      "learning_rate": 9.340947368421053e-05,
      "loss": 0.3606,
      "step": 45063
    },
    {
      "epoch": 2.5234628737820586,
      "grad_norm": 1.1141988039016724,
      "learning_rate": 9.34092105263158e-05,
      "loss": 0.4391,
      "step": 45064
    },
    {
      "epoch": 2.5235188710941876,
      "grad_norm": 1.3191722631454468,
      "learning_rate": 9.340894736842107e-05,
      "loss": 0.4731,
      "step": 45065
    },
    {
      "epoch": 2.5235748684063166,
      "grad_norm": 1.41512131690979,
      "learning_rate": 9.340868421052631e-05,
      "loss": 0.3902,
      "step": 45066
    },
    {
      "epoch": 2.5236308657184456,
      "grad_norm": 1.687557339668274,
      "learning_rate": 9.340842105263157e-05,
      "loss": 0.3659,
      "step": 45067
    },
    {
      "epoch": 2.5236868630305747,
      "grad_norm": 1.1788462400436401,
      "learning_rate": 9.340815789473685e-05,
      "loss": 0.3276,
      "step": 45068
    },
    {
      "epoch": 2.5237428603427037,
      "grad_norm": 1.1703671216964722,
      "learning_rate": 9.34078947368421e-05,
      "loss": 0.3723,
      "step": 45069
    },
    {
      "epoch": 2.5237988576548327,
      "grad_norm": 1.241195559501648,
      "learning_rate": 9.340763157894738e-05,
      "loss": 0.422,
      "step": 45070
    },
    {
      "epoch": 2.5238548549669617,
      "grad_norm": 0.9738800525665283,
      "learning_rate": 9.340736842105263e-05,
      "loss": 0.3217,
      "step": 45071
    },
    {
      "epoch": 2.5239108522790907,
      "grad_norm": 1.2739169597625732,
      "learning_rate": 9.34071052631579e-05,
      "loss": 0.4885,
      "step": 45072
    },
    {
      "epoch": 2.5239668495912198,
      "grad_norm": 0.9963579177856445,
      "learning_rate": 9.340684210526316e-05,
      "loss": 0.3149,
      "step": 45073
    },
    {
      "epoch": 2.5240228469033488,
      "grad_norm": 0.9164132475852966,
      "learning_rate": 9.340657894736843e-05,
      "loss": 0.3975,
      "step": 45074
    },
    {
      "epoch": 2.524078844215478,
      "grad_norm": 1.0604761838912964,
      "learning_rate": 9.340631578947369e-05,
      "loss": 0.424,
      "step": 45075
    },
    {
      "epoch": 2.524134841527607,
      "grad_norm": 1.1388243436813354,
      "learning_rate": 9.340605263157895e-05,
      "loss": 0.4245,
      "step": 45076
    },
    {
      "epoch": 2.524190838839736,
      "grad_norm": 1.5361894369125366,
      "learning_rate": 9.340578947368421e-05,
      "loss": 0.5117,
      "step": 45077
    },
    {
      "epoch": 2.524246836151865,
      "grad_norm": 1.1749378442764282,
      "learning_rate": 9.340552631578948e-05,
      "loss": 0.3992,
      "step": 45078
    },
    {
      "epoch": 2.524302833463994,
      "grad_norm": 1.165282964706421,
      "learning_rate": 9.340526315789474e-05,
      "loss": 0.482,
      "step": 45079
    },
    {
      "epoch": 2.524358830776123,
      "grad_norm": 2.387977123260498,
      "learning_rate": 9.3405e-05,
      "loss": 0.3764,
      "step": 45080
    },
    {
      "epoch": 2.524414828088252,
      "grad_norm": 1.0958046913146973,
      "learning_rate": 9.340473684210526e-05,
      "loss": 0.3359,
      "step": 45081
    },
    {
      "epoch": 2.524470825400381,
      "grad_norm": 1.565321683883667,
      "learning_rate": 9.340447368421054e-05,
      "loss": 0.2915,
      "step": 45082
    },
    {
      "epoch": 2.52452682271251,
      "grad_norm": 1.3294605016708374,
      "learning_rate": 9.34042105263158e-05,
      "loss": 0.4456,
      "step": 45083
    },
    {
      "epoch": 2.524582820024639,
      "grad_norm": 1.3427128791809082,
      "learning_rate": 9.340394736842106e-05,
      "loss": 0.3722,
      "step": 45084
    },
    {
      "epoch": 2.524638817336768,
      "grad_norm": 1.0191469192504883,
      "learning_rate": 9.340368421052632e-05,
      "loss": 0.3192,
      "step": 45085
    },
    {
      "epoch": 2.524694814648897,
      "grad_norm": 1.278325080871582,
      "learning_rate": 9.340342105263158e-05,
      "loss": 0.3817,
      "step": 45086
    },
    {
      "epoch": 2.524750811961026,
      "grad_norm": 1.2492661476135254,
      "learning_rate": 9.340315789473685e-05,
      "loss": 0.3923,
      "step": 45087
    },
    {
      "epoch": 2.524806809273155,
      "grad_norm": 1.042992353439331,
      "learning_rate": 9.340289473684211e-05,
      "loss": 0.3939,
      "step": 45088
    },
    {
      "epoch": 2.524862806585284,
      "grad_norm": 1.5448247194290161,
      "learning_rate": 9.340263157894737e-05,
      "loss": 0.3493,
      "step": 45089
    },
    {
      "epoch": 2.524918803897413,
      "grad_norm": 1.5317858457565308,
      "learning_rate": 9.340236842105263e-05,
      "loss": 0.4325,
      "step": 45090
    },
    {
      "epoch": 2.524974801209542,
      "grad_norm": 1.2117873430252075,
      "learning_rate": 9.34021052631579e-05,
      "loss": 0.4355,
      "step": 45091
    },
    {
      "epoch": 2.525030798521671,
      "grad_norm": 1.4078155755996704,
      "learning_rate": 9.340184210526316e-05,
      "loss": 0.3512,
      "step": 45092
    },
    {
      "epoch": 2.5250867958338,
      "grad_norm": 1.2072077989578247,
      "learning_rate": 9.340157894736844e-05,
      "loss": 0.5043,
      "step": 45093
    },
    {
      "epoch": 2.525142793145929,
      "grad_norm": 1.0573042631149292,
      "learning_rate": 9.340131578947368e-05,
      "loss": 0.3367,
      "step": 45094
    },
    {
      "epoch": 2.525198790458058,
      "grad_norm": 0.9739222526550293,
      "learning_rate": 9.340105263157895e-05,
      "loss": 0.3845,
      "step": 45095
    },
    {
      "epoch": 2.525254787770187,
      "grad_norm": 1.2553620338439941,
      "learning_rate": 9.340078947368421e-05,
      "loss": 0.4464,
      "step": 45096
    },
    {
      "epoch": 2.525310785082316,
      "grad_norm": 1.2545933723449707,
      "learning_rate": 9.340052631578949e-05,
      "loss": 0.4025,
      "step": 45097
    },
    {
      "epoch": 2.5253667823944452,
      "grad_norm": 1.13052499294281,
      "learning_rate": 9.340026315789475e-05,
      "loss": 0.453,
      "step": 45098
    },
    {
      "epoch": 2.5254227797065743,
      "grad_norm": 1.2982596158981323,
      "learning_rate": 9.340000000000001e-05,
      "loss": 0.5166,
      "step": 45099
    },
    {
      "epoch": 2.5254787770187033,
      "grad_norm": 1.1480048894882202,
      "learning_rate": 9.339973684210527e-05,
      "loss": 0.3518,
      "step": 45100
    },
    {
      "epoch": 2.525534774330832,
      "grad_norm": 1.0768511295318604,
      "learning_rate": 9.339947368421053e-05,
      "loss": 0.3888,
      "step": 45101
    },
    {
      "epoch": 2.5255907716429613,
      "grad_norm": 1.5143009424209595,
      "learning_rate": 9.33992105263158e-05,
      "loss": 0.4153,
      "step": 45102
    },
    {
      "epoch": 2.52564676895509,
      "grad_norm": 0.9787368774414062,
      "learning_rate": 9.339894736842105e-05,
      "loss": 0.3268,
      "step": 45103
    },
    {
      "epoch": 2.5257027662672193,
      "grad_norm": 1.5540978908538818,
      "learning_rate": 9.339868421052632e-05,
      "loss": 0.4002,
      "step": 45104
    },
    {
      "epoch": 2.525758763579348,
      "grad_norm": 1.0439560413360596,
      "learning_rate": 9.339842105263158e-05,
      "loss": 0.408,
      "step": 45105
    },
    {
      "epoch": 2.5258147608914774,
      "grad_norm": 1.1765691041946411,
      "learning_rate": 9.339815789473685e-05,
      "loss": 0.436,
      "step": 45106
    },
    {
      "epoch": 2.525870758203606,
      "grad_norm": 1.1970053911209106,
      "learning_rate": 9.339789473684211e-05,
      "loss": 0.3781,
      "step": 45107
    },
    {
      "epoch": 2.5259267555157354,
      "grad_norm": 1.1379094123840332,
      "learning_rate": 9.339763157894737e-05,
      "loss": 0.4421,
      "step": 45108
    },
    {
      "epoch": 2.525982752827864,
      "grad_norm": 1.3179931640625,
      "learning_rate": 9.339736842105263e-05,
      "loss": 0.3257,
      "step": 45109
    },
    {
      "epoch": 2.5260387501399935,
      "grad_norm": 1.0589743852615356,
      "learning_rate": 9.33971052631579e-05,
      "loss": 0.2979,
      "step": 45110
    },
    {
      "epoch": 2.526094747452122,
      "grad_norm": 1.413193941116333,
      "learning_rate": 9.339684210526316e-05,
      "loss": 0.4256,
      "step": 45111
    },
    {
      "epoch": 2.5261507447642515,
      "grad_norm": 1.419501543045044,
      "learning_rate": 9.339657894736842e-05,
      "loss": 0.4531,
      "step": 45112
    },
    {
      "epoch": 2.52620674207638,
      "grad_norm": 1.2457424402236938,
      "learning_rate": 9.339631578947368e-05,
      "loss": 0.3737,
      "step": 45113
    },
    {
      "epoch": 2.5262627393885095,
      "grad_norm": 1.2079306840896606,
      "learning_rate": 9.339605263157896e-05,
      "loss": 0.3401,
      "step": 45114
    },
    {
      "epoch": 2.526318736700638,
      "grad_norm": 1.0016108751296997,
      "learning_rate": 9.339578947368422e-05,
      "loss": 0.3256,
      "step": 45115
    },
    {
      "epoch": 2.5263747340127676,
      "grad_norm": 1.2053142786026,
      "learning_rate": 9.339552631578948e-05,
      "loss": 0.4198,
      "step": 45116
    },
    {
      "epoch": 2.526430731324896,
      "grad_norm": 1.1074261665344238,
      "learning_rate": 9.339526315789474e-05,
      "loss": 0.4297,
      "step": 45117
    },
    {
      "epoch": 2.5264867286370256,
      "grad_norm": 1.0737534761428833,
      "learning_rate": 9.3395e-05,
      "loss": 0.396,
      "step": 45118
    },
    {
      "epoch": 2.526542725949154,
      "grad_norm": 1.097792148590088,
      "learning_rate": 9.339473684210527e-05,
      "loss": 0.4653,
      "step": 45119
    },
    {
      "epoch": 2.5265987232612837,
      "grad_norm": 1.238970398902893,
      "learning_rate": 9.339447368421053e-05,
      "loss": 0.4188,
      "step": 45120
    },
    {
      "epoch": 2.5266547205734122,
      "grad_norm": 1.3464114665985107,
      "learning_rate": 9.339421052631579e-05,
      "loss": 0.5059,
      "step": 45121
    },
    {
      "epoch": 2.5267107178855417,
      "grad_norm": 1.4598034620285034,
      "learning_rate": 9.339394736842105e-05,
      "loss": 0.3773,
      "step": 45122
    },
    {
      "epoch": 2.5267667151976703,
      "grad_norm": 1.0902695655822754,
      "learning_rate": 9.339368421052632e-05,
      "loss": 0.5127,
      "step": 45123
    },
    {
      "epoch": 2.5268227125097997,
      "grad_norm": 1.230257511138916,
      "learning_rate": 9.339342105263158e-05,
      "loss": 0.455,
      "step": 45124
    },
    {
      "epoch": 2.5268787098219283,
      "grad_norm": 1.2285648584365845,
      "learning_rate": 9.339315789473685e-05,
      "loss": 0.3085,
      "step": 45125
    },
    {
      "epoch": 2.5269347071340578,
      "grad_norm": 1.129361629486084,
      "learning_rate": 9.33928947368421e-05,
      "loss": 0.4479,
      "step": 45126
    },
    {
      "epoch": 2.5269907044461863,
      "grad_norm": 1.2754857540130615,
      "learning_rate": 9.339263157894737e-05,
      "loss": 0.3292,
      "step": 45127
    },
    {
      "epoch": 2.527046701758316,
      "grad_norm": 1.2990093231201172,
      "learning_rate": 9.339236842105263e-05,
      "loss": 0.3503,
      "step": 45128
    },
    {
      "epoch": 2.5271026990704444,
      "grad_norm": 1.4103270769119263,
      "learning_rate": 9.339210526315791e-05,
      "loss": 0.4367,
      "step": 45129
    },
    {
      "epoch": 2.527158696382574,
      "grad_norm": 1.131492018699646,
      "learning_rate": 9.339184210526317e-05,
      "loss": 0.3644,
      "step": 45130
    },
    {
      "epoch": 2.5272146936947024,
      "grad_norm": 1.294791579246521,
      "learning_rate": 9.339157894736843e-05,
      "loss": 0.4252,
      "step": 45131
    },
    {
      "epoch": 2.527270691006832,
      "grad_norm": 1.4098106622695923,
      "learning_rate": 9.339131578947369e-05,
      "loss": 0.3889,
      "step": 45132
    },
    {
      "epoch": 2.5273266883189605,
      "grad_norm": 1.0930731296539307,
      "learning_rate": 9.339105263157896e-05,
      "loss": 0.4311,
      "step": 45133
    },
    {
      "epoch": 2.52738268563109,
      "grad_norm": 1.3909887075424194,
      "learning_rate": 9.339078947368422e-05,
      "loss": 0.515,
      "step": 45134
    },
    {
      "epoch": 2.5274386829432185,
      "grad_norm": 1.3683050870895386,
      "learning_rate": 9.339052631578948e-05,
      "loss": 0.6142,
      "step": 45135
    },
    {
      "epoch": 2.527494680255348,
      "grad_norm": 1.382804274559021,
      "learning_rate": 9.339026315789474e-05,
      "loss": 0.461,
      "step": 45136
    },
    {
      "epoch": 2.5275506775674765,
      "grad_norm": 1.2536604404449463,
      "learning_rate": 9.339e-05,
      "loss": 0.3702,
      "step": 45137
    },
    {
      "epoch": 2.527606674879606,
      "grad_norm": 1.1602362394332886,
      "learning_rate": 9.338973684210527e-05,
      "loss": 0.3101,
      "step": 45138
    },
    {
      "epoch": 2.5276626721917346,
      "grad_norm": 0.9738381505012512,
      "learning_rate": 9.338947368421053e-05,
      "loss": 0.2954,
      "step": 45139
    },
    {
      "epoch": 2.527718669503864,
      "grad_norm": 1.1538786888122559,
      "learning_rate": 9.338921052631579e-05,
      "loss": 0.4015,
      "step": 45140
    },
    {
      "epoch": 2.5277746668159926,
      "grad_norm": 1.1494847536087036,
      "learning_rate": 9.338894736842105e-05,
      "loss": 0.3916,
      "step": 45141
    },
    {
      "epoch": 2.527830664128122,
      "grad_norm": 1.0634634494781494,
      "learning_rate": 9.338868421052632e-05,
      "loss": 0.3858,
      "step": 45142
    },
    {
      "epoch": 2.5278866614402506,
      "grad_norm": 1.1732077598571777,
      "learning_rate": 9.338842105263158e-05,
      "loss": 0.4547,
      "step": 45143
    },
    {
      "epoch": 2.52794265875238,
      "grad_norm": 1.3576562404632568,
      "learning_rate": 9.338815789473684e-05,
      "loss": 0.4196,
      "step": 45144
    },
    {
      "epoch": 2.5279986560645087,
      "grad_norm": 1.6606502532958984,
      "learning_rate": 9.33878947368421e-05,
      "loss": 0.4791,
      "step": 45145
    },
    {
      "epoch": 2.528054653376638,
      "grad_norm": 1.2869861125946045,
      "learning_rate": 9.338763157894738e-05,
      "loss": 0.3888,
      "step": 45146
    },
    {
      "epoch": 2.5281106506887667,
      "grad_norm": 0.9239200353622437,
      "learning_rate": 9.338736842105264e-05,
      "loss": 0.3941,
      "step": 45147
    },
    {
      "epoch": 2.528166648000896,
      "grad_norm": 1.1052995920181274,
      "learning_rate": 9.338710526315791e-05,
      "loss": 0.3587,
      "step": 45148
    },
    {
      "epoch": 2.5282226453130248,
      "grad_norm": 1.1069886684417725,
      "learning_rate": 9.338684210526316e-05,
      "loss": 0.3702,
      "step": 45149
    },
    {
      "epoch": 2.5282786426251542,
      "grad_norm": 4.872007369995117,
      "learning_rate": 9.338657894736843e-05,
      "loss": 0.4661,
      "step": 45150
    },
    {
      "epoch": 2.528334639937283,
      "grad_norm": 1.1149044036865234,
      "learning_rate": 9.338631578947369e-05,
      "loss": 0.3993,
      "step": 45151
    },
    {
      "epoch": 2.5283906372494123,
      "grad_norm": 1.0310707092285156,
      "learning_rate": 9.338605263157895e-05,
      "loss": 0.3898,
      "step": 45152
    },
    {
      "epoch": 2.528446634561541,
      "grad_norm": 1.0805010795593262,
      "learning_rate": 9.338578947368422e-05,
      "loss": 0.4205,
      "step": 45153
    },
    {
      "epoch": 2.5285026318736703,
      "grad_norm": 1.1837612390518188,
      "learning_rate": 9.338552631578947e-05,
      "loss": 0.4698,
      "step": 45154
    },
    {
      "epoch": 2.528558629185799,
      "grad_norm": 1.3161191940307617,
      "learning_rate": 9.338526315789474e-05,
      "loss": 0.4502,
      "step": 45155
    },
    {
      "epoch": 2.5286146264979283,
      "grad_norm": 1.1102139949798584,
      "learning_rate": 9.3385e-05,
      "loss": 0.359,
      "step": 45156
    },
    {
      "epoch": 2.528670623810057,
      "grad_norm": 1.18354070186615,
      "learning_rate": 9.338473684210527e-05,
      "loss": 0.3612,
      "step": 45157
    },
    {
      "epoch": 2.5287266211221864,
      "grad_norm": 1.3239307403564453,
      "learning_rate": 9.338447368421052e-05,
      "loss": 0.474,
      "step": 45158
    },
    {
      "epoch": 2.528782618434315,
      "grad_norm": 1.0875134468078613,
      "learning_rate": 9.33842105263158e-05,
      "loss": 0.3601,
      "step": 45159
    },
    {
      "epoch": 2.5288386157464444,
      "grad_norm": 1.3045045137405396,
      "learning_rate": 9.338394736842105e-05,
      "loss": 0.4849,
      "step": 45160
    },
    {
      "epoch": 2.528894613058573,
      "grad_norm": 1.6577404737472534,
      "learning_rate": 9.338368421052633e-05,
      "loss": 0.6871,
      "step": 45161
    },
    {
      "epoch": 2.5289506103707025,
      "grad_norm": 1.1601085662841797,
      "learning_rate": 9.338342105263159e-05,
      "loss": 0.3436,
      "step": 45162
    },
    {
      "epoch": 2.529006607682831,
      "grad_norm": 1.2022532224655151,
      "learning_rate": 9.338315789473685e-05,
      "loss": 0.4046,
      "step": 45163
    },
    {
      "epoch": 2.5290626049949605,
      "grad_norm": 1.071622371673584,
      "learning_rate": 9.33828947368421e-05,
      "loss": 0.3933,
      "step": 45164
    },
    {
      "epoch": 2.529118602307089,
      "grad_norm": 1.0860626697540283,
      "learning_rate": 9.338263157894738e-05,
      "loss": 0.3671,
      "step": 45165
    },
    {
      "epoch": 2.5291745996192185,
      "grad_norm": 1.3062572479248047,
      "learning_rate": 9.338236842105264e-05,
      "loss": 0.4975,
      "step": 45166
    },
    {
      "epoch": 2.529230596931347,
      "grad_norm": 1.188965082168579,
      "learning_rate": 9.33821052631579e-05,
      "loss": 0.3607,
      "step": 45167
    },
    {
      "epoch": 2.5292865942434766,
      "grad_norm": 1.1517301797866821,
      "learning_rate": 9.338184210526316e-05,
      "loss": 0.3599,
      "step": 45168
    },
    {
      "epoch": 2.529342591555605,
      "grad_norm": 1.232240080833435,
      "learning_rate": 9.338157894736843e-05,
      "loss": 0.4651,
      "step": 45169
    },
    {
      "epoch": 2.5293985888677346,
      "grad_norm": 1.176153540611267,
      "learning_rate": 9.338131578947369e-05,
      "loss": 0.3881,
      "step": 45170
    },
    {
      "epoch": 2.529454586179863,
      "grad_norm": 1.0519949197769165,
      "learning_rate": 9.338105263157895e-05,
      "loss": 0.3461,
      "step": 45171
    },
    {
      "epoch": 2.5295105834919926,
      "grad_norm": 1.444745421409607,
      "learning_rate": 9.338078947368421e-05,
      "loss": 0.5805,
      "step": 45172
    },
    {
      "epoch": 2.5295665808041212,
      "grad_norm": 1.1891999244689941,
      "learning_rate": 9.338052631578947e-05,
      "loss": 0.4302,
      "step": 45173
    },
    {
      "epoch": 2.5296225781162507,
      "grad_norm": 1.153363823890686,
      "learning_rate": 9.338026315789474e-05,
      "loss": 0.3532,
      "step": 45174
    },
    {
      "epoch": 2.5296785754283793,
      "grad_norm": 1.2139067649841309,
      "learning_rate": 9.338e-05,
      "loss": 0.4526,
      "step": 45175
    },
    {
      "epoch": 2.5297345727405087,
      "grad_norm": 1.1376354694366455,
      "learning_rate": 9.337973684210526e-05,
      "loss": 0.4158,
      "step": 45176
    },
    {
      "epoch": 2.5297905700526373,
      "grad_norm": 1.1992076635360718,
      "learning_rate": 9.337947368421052e-05,
      "loss": 0.5269,
      "step": 45177
    },
    {
      "epoch": 2.5298465673647668,
      "grad_norm": 1.1391613483428955,
      "learning_rate": 9.33792105263158e-05,
      "loss": 0.4084,
      "step": 45178
    },
    {
      "epoch": 2.5299025646768953,
      "grad_norm": 1.122583270072937,
      "learning_rate": 9.337894736842106e-05,
      "loss": 0.3477,
      "step": 45179
    },
    {
      "epoch": 2.529958561989025,
      "grad_norm": 1.106078863143921,
      "learning_rate": 9.337868421052633e-05,
      "loss": 0.3117,
      "step": 45180
    },
    {
      "epoch": 2.5300145593011534,
      "grad_norm": 1.2008103132247925,
      "learning_rate": 9.337842105263158e-05,
      "loss": 0.3407,
      "step": 45181
    },
    {
      "epoch": 2.530070556613283,
      "grad_norm": 1.2463667392730713,
      "learning_rate": 9.337815789473685e-05,
      "loss": 0.36,
      "step": 45182
    },
    {
      "epoch": 2.5301265539254114,
      "grad_norm": 1.1172571182250977,
      "learning_rate": 9.337789473684211e-05,
      "loss": 0.3473,
      "step": 45183
    },
    {
      "epoch": 2.5301825512375404,
      "grad_norm": 1.3725645542144775,
      "learning_rate": 9.337763157894738e-05,
      "loss": 0.4122,
      "step": 45184
    },
    {
      "epoch": 2.5302385485496695,
      "grad_norm": 0.9529877305030823,
      "learning_rate": 9.337736842105264e-05,
      "loss": 0.4425,
      "step": 45185
    },
    {
      "epoch": 2.5302945458617985,
      "grad_norm": 1.0737224817276,
      "learning_rate": 9.33771052631579e-05,
      "loss": 0.416,
      "step": 45186
    },
    {
      "epoch": 2.5303505431739275,
      "grad_norm": 1.338065266609192,
      "learning_rate": 9.337684210526316e-05,
      "loss": 0.3699,
      "step": 45187
    },
    {
      "epoch": 2.5304065404860565,
      "grad_norm": 1.1875625848770142,
      "learning_rate": 9.337657894736842e-05,
      "loss": 0.316,
      "step": 45188
    },
    {
      "epoch": 2.5304625377981855,
      "grad_norm": 1.0232272148132324,
      "learning_rate": 9.33763157894737e-05,
      "loss": 0.4504,
      "step": 45189
    },
    {
      "epoch": 2.5305185351103145,
      "grad_norm": 1.1358455419540405,
      "learning_rate": 9.337605263157895e-05,
      "loss": 0.3543,
      "step": 45190
    },
    {
      "epoch": 2.5305745324224436,
      "grad_norm": 1.235042929649353,
      "learning_rate": 9.337578947368421e-05,
      "loss": 0.3676,
      "step": 45191
    },
    {
      "epoch": 2.5306305297345726,
      "grad_norm": 1.7193940877914429,
      "learning_rate": 9.337552631578947e-05,
      "loss": 0.5011,
      "step": 45192
    },
    {
      "epoch": 2.5306865270467016,
      "grad_norm": 1.0871737003326416,
      "learning_rate": 9.337526315789475e-05,
      "loss": 0.3449,
      "step": 45193
    },
    {
      "epoch": 2.5307425243588306,
      "grad_norm": 1.1899616718292236,
      "learning_rate": 9.3375e-05,
      "loss": 0.4164,
      "step": 45194
    },
    {
      "epoch": 2.5307985216709596,
      "grad_norm": 1.2257051467895508,
      "learning_rate": 9.337473684210527e-05,
      "loss": 0.3094,
      "step": 45195
    },
    {
      "epoch": 2.5308545189830887,
      "grad_norm": 1.1478500366210938,
      "learning_rate": 9.337447368421053e-05,
      "loss": 0.3197,
      "step": 45196
    },
    {
      "epoch": 2.5309105162952177,
      "grad_norm": 1.124822974205017,
      "learning_rate": 9.33742105263158e-05,
      "loss": 0.4217,
      "step": 45197
    },
    {
      "epoch": 2.5309665136073467,
      "grad_norm": 1.260844111442566,
      "learning_rate": 9.337394736842106e-05,
      "loss": 0.3711,
      "step": 45198
    },
    {
      "epoch": 2.5310225109194757,
      "grad_norm": 1.2106292247772217,
      "learning_rate": 9.337368421052632e-05,
      "loss": 0.4464,
      "step": 45199
    },
    {
      "epoch": 2.5310785082316047,
      "grad_norm": 1.400875449180603,
      "learning_rate": 9.337342105263158e-05,
      "loss": 0.28,
      "step": 45200
    },
    {
      "epoch": 2.5311345055437338,
      "grad_norm": 1.1131926774978638,
      "learning_rate": 9.337315789473685e-05,
      "loss": 0.651,
      "step": 45201
    },
    {
      "epoch": 2.531190502855863,
      "grad_norm": 1.451791524887085,
      "learning_rate": 9.337289473684211e-05,
      "loss": 0.297,
      "step": 45202
    },
    {
      "epoch": 2.531246500167992,
      "grad_norm": 1.2021558284759521,
      "learning_rate": 9.337263157894738e-05,
      "loss": 0.4299,
      "step": 45203
    },
    {
      "epoch": 2.531302497480121,
      "grad_norm": 1.460461139678955,
      "learning_rate": 9.337236842105263e-05,
      "loss": 0.3899,
      "step": 45204
    },
    {
      "epoch": 2.53135849479225,
      "grad_norm": 1.3322944641113281,
      "learning_rate": 9.337210526315789e-05,
      "loss": 0.3108,
      "step": 45205
    },
    {
      "epoch": 2.531414492104379,
      "grad_norm": 1.264677882194519,
      "learning_rate": 9.337184210526316e-05,
      "loss": 0.354,
      "step": 45206
    },
    {
      "epoch": 2.531470489416508,
      "grad_norm": 1.2151342630386353,
      "learning_rate": 9.337157894736842e-05,
      "loss": 0.49,
      "step": 45207
    },
    {
      "epoch": 2.531526486728637,
      "grad_norm": 1.2918343544006348,
      "learning_rate": 9.33713157894737e-05,
      "loss": 0.4085,
      "step": 45208
    },
    {
      "epoch": 2.531582484040766,
      "grad_norm": 1.3364993333816528,
      "learning_rate": 9.337105263157894e-05,
      "loss": 0.3112,
      "step": 45209
    },
    {
      "epoch": 2.531638481352895,
      "grad_norm": 1.2031617164611816,
      "learning_rate": 9.337078947368422e-05,
      "loss": 0.4125,
      "step": 45210
    },
    {
      "epoch": 2.531694478665024,
      "grad_norm": 1.063269853591919,
      "learning_rate": 9.337052631578948e-05,
      "loss": 0.3567,
      "step": 45211
    },
    {
      "epoch": 2.531750475977153,
      "grad_norm": 1.306514024734497,
      "learning_rate": 9.337026315789475e-05,
      "loss": 0.4826,
      "step": 45212
    },
    {
      "epoch": 2.531806473289282,
      "grad_norm": 1.0742863416671753,
      "learning_rate": 9.337e-05,
      "loss": 0.2993,
      "step": 45213
    },
    {
      "epoch": 2.531862470601411,
      "grad_norm": 1.084478497505188,
      "learning_rate": 9.336973684210527e-05,
      "loss": 0.4825,
      "step": 45214
    },
    {
      "epoch": 2.53191846791354,
      "grad_norm": 1.1511906385421753,
      "learning_rate": 9.336947368421053e-05,
      "loss": 0.4134,
      "step": 45215
    },
    {
      "epoch": 2.531974465225669,
      "grad_norm": 1.4297105073928833,
      "learning_rate": 9.33692105263158e-05,
      "loss": 0.3668,
      "step": 45216
    },
    {
      "epoch": 2.532030462537798,
      "grad_norm": 1.131028175354004,
      "learning_rate": 9.336894736842106e-05,
      "loss": 0.341,
      "step": 45217
    },
    {
      "epoch": 2.532086459849927,
      "grad_norm": 0.9926726818084717,
      "learning_rate": 9.336868421052632e-05,
      "loss": 0.3477,
      "step": 45218
    },
    {
      "epoch": 2.532142457162056,
      "grad_norm": 1.5053209066390991,
      "learning_rate": 9.336842105263158e-05,
      "loss": 0.3904,
      "step": 45219
    },
    {
      "epoch": 2.532198454474185,
      "grad_norm": 1.4318617582321167,
      "learning_rate": 9.336815789473685e-05,
      "loss": 0.5467,
      "step": 45220
    },
    {
      "epoch": 2.532254451786314,
      "grad_norm": 1.2721027135849,
      "learning_rate": 9.336789473684211e-05,
      "loss": 0.3927,
      "step": 45221
    },
    {
      "epoch": 2.532310449098443,
      "grad_norm": 1.144276738166809,
      "learning_rate": 9.336763157894737e-05,
      "loss": 0.4427,
      "step": 45222
    },
    {
      "epoch": 2.532366446410572,
      "grad_norm": 0.998488187789917,
      "learning_rate": 9.336736842105263e-05,
      "loss": 0.4444,
      "step": 45223
    },
    {
      "epoch": 2.532422443722701,
      "grad_norm": 1.9380106925964355,
      "learning_rate": 9.336710526315789e-05,
      "loss": 0.3967,
      "step": 45224
    },
    {
      "epoch": 2.53247844103483,
      "grad_norm": 1.3981908559799194,
      "learning_rate": 9.336684210526317e-05,
      "loss": 0.34,
      "step": 45225
    },
    {
      "epoch": 2.5325344383469592,
      "grad_norm": 1.2559336423873901,
      "learning_rate": 9.336657894736843e-05,
      "loss": 0.3915,
      "step": 45226
    },
    {
      "epoch": 2.5325904356590883,
      "grad_norm": 1.497085452079773,
      "learning_rate": 9.336631578947369e-05,
      "loss": 0.4074,
      "step": 45227
    },
    {
      "epoch": 2.5326464329712173,
      "grad_norm": 1.298133373260498,
      "learning_rate": 9.336605263157895e-05,
      "loss": 0.4729,
      "step": 45228
    },
    {
      "epoch": 2.5327024302833463,
      "grad_norm": 1.0105904340744019,
      "learning_rate": 9.336578947368422e-05,
      "loss": 0.353,
      "step": 45229
    },
    {
      "epoch": 2.5327584275954753,
      "grad_norm": 1.1941754817962646,
      "learning_rate": 9.336552631578948e-05,
      "loss": 0.3754,
      "step": 45230
    },
    {
      "epoch": 2.5328144249076043,
      "grad_norm": 1.2362611293792725,
      "learning_rate": 9.336526315789474e-05,
      "loss": 0.5003,
      "step": 45231
    },
    {
      "epoch": 2.5328704222197334,
      "grad_norm": 1.0890926122665405,
      "learning_rate": 9.3365e-05,
      "loss": 0.4168,
      "step": 45232
    },
    {
      "epoch": 2.5329264195318624,
      "grad_norm": 1.344289779663086,
      "learning_rate": 9.336473684210527e-05,
      "loss": 0.4641,
      "step": 45233
    },
    {
      "epoch": 2.5329824168439914,
      "grad_norm": 1.1892579793930054,
      "learning_rate": 9.336447368421053e-05,
      "loss": 0.3389,
      "step": 45234
    },
    {
      "epoch": 2.5330384141561204,
      "grad_norm": 1.1842209100723267,
      "learning_rate": 9.33642105263158e-05,
      "loss": 0.3071,
      "step": 45235
    },
    {
      "epoch": 2.5330944114682494,
      "grad_norm": 1.0685248374938965,
      "learning_rate": 9.336394736842105e-05,
      "loss": 0.3465,
      "step": 45236
    },
    {
      "epoch": 2.5331504087803784,
      "grad_norm": 1.8892714977264404,
      "learning_rate": 9.336368421052632e-05,
      "loss": 0.6618,
      "step": 45237
    },
    {
      "epoch": 2.5332064060925075,
      "grad_norm": 0.9832775592803955,
      "learning_rate": 9.336342105263158e-05,
      "loss": 0.4002,
      "step": 45238
    },
    {
      "epoch": 2.5332624034046365,
      "grad_norm": 1.1432660818099976,
      "learning_rate": 9.336315789473684e-05,
      "loss": 0.3566,
      "step": 45239
    },
    {
      "epoch": 2.5333184007167655,
      "grad_norm": 1.0462316274642944,
      "learning_rate": 9.336289473684212e-05,
      "loss": 0.334,
      "step": 45240
    },
    {
      "epoch": 2.5333743980288945,
      "grad_norm": 1.3578897714614868,
      "learning_rate": 9.336263157894736e-05,
      "loss": 0.5442,
      "step": 45241
    },
    {
      "epoch": 2.5334303953410235,
      "grad_norm": 1.9784818887710571,
      "learning_rate": 9.336236842105264e-05,
      "loss": 0.6246,
      "step": 45242
    },
    {
      "epoch": 2.5334863926531526,
      "grad_norm": 1.3815408945083618,
      "learning_rate": 9.33621052631579e-05,
      "loss": 0.49,
      "step": 45243
    },
    {
      "epoch": 2.5335423899652816,
      "grad_norm": 1.2610872983932495,
      "learning_rate": 9.336184210526317e-05,
      "loss": 0.3834,
      "step": 45244
    },
    {
      "epoch": 2.5335983872774106,
      "grad_norm": 1.2550121545791626,
      "learning_rate": 9.336157894736843e-05,
      "loss": 0.4933,
      "step": 45245
    },
    {
      "epoch": 2.5336543845895396,
      "grad_norm": 1.279241681098938,
      "learning_rate": 9.336131578947369e-05,
      "loss": 0.2893,
      "step": 45246
    },
    {
      "epoch": 2.5337103819016686,
      "grad_norm": 1.0732061862945557,
      "learning_rate": 9.336105263157895e-05,
      "loss": 0.3774,
      "step": 45247
    },
    {
      "epoch": 2.5337663792137977,
      "grad_norm": 1.111973524093628,
      "learning_rate": 9.336078947368422e-05,
      "loss": 0.3549,
      "step": 45248
    },
    {
      "epoch": 2.5338223765259267,
      "grad_norm": 1.1784390211105347,
      "learning_rate": 9.336052631578948e-05,
      "loss": 0.4648,
      "step": 45249
    },
    {
      "epoch": 2.5338783738380557,
      "grad_norm": 1.2090518474578857,
      "learning_rate": 9.336026315789474e-05,
      "loss": 0.4341,
      "step": 45250
    },
    {
      "epoch": 2.5339343711501847,
      "grad_norm": 2.4343459606170654,
      "learning_rate": 9.336e-05,
      "loss": 0.4336,
      "step": 45251
    },
    {
      "epoch": 2.5339903684623137,
      "grad_norm": 1.2436964511871338,
      "learning_rate": 9.335973684210527e-05,
      "loss": 0.4915,
      "step": 45252
    },
    {
      "epoch": 2.5340463657744428,
      "grad_norm": 0.9614510536193848,
      "learning_rate": 9.335947368421053e-05,
      "loss": 0.2904,
      "step": 45253
    },
    {
      "epoch": 2.5341023630865718,
      "grad_norm": 1.4795252084732056,
      "learning_rate": 9.33592105263158e-05,
      "loss": 0.5812,
      "step": 45254
    },
    {
      "epoch": 2.534158360398701,
      "grad_norm": 1.1069657802581787,
      "learning_rate": 9.335894736842105e-05,
      "loss": 0.3125,
      "step": 45255
    },
    {
      "epoch": 2.53421435771083,
      "grad_norm": 1.0722161531448364,
      "learning_rate": 9.335868421052631e-05,
      "loss": 0.3727,
      "step": 45256
    },
    {
      "epoch": 2.534270355022959,
      "grad_norm": 1.054937720298767,
      "learning_rate": 9.335842105263159e-05,
      "loss": 0.2395,
      "step": 45257
    },
    {
      "epoch": 2.534326352335088,
      "grad_norm": 1.1573792695999146,
      "learning_rate": 9.335815789473685e-05,
      "loss": 0.4847,
      "step": 45258
    },
    {
      "epoch": 2.534382349647217,
      "grad_norm": 1.2153877019882202,
      "learning_rate": 9.33578947368421e-05,
      "loss": 0.3613,
      "step": 45259
    },
    {
      "epoch": 2.534438346959346,
      "grad_norm": 1.262445330619812,
      "learning_rate": 9.335763157894737e-05,
      "loss": 0.364,
      "step": 45260
    },
    {
      "epoch": 2.534494344271475,
      "grad_norm": 1.3474647998809814,
      "learning_rate": 9.335736842105264e-05,
      "loss": 0.4819,
      "step": 45261
    },
    {
      "epoch": 2.534550341583604,
      "grad_norm": 1.1217551231384277,
      "learning_rate": 9.33571052631579e-05,
      "loss": 0.4093,
      "step": 45262
    },
    {
      "epoch": 2.534606338895733,
      "grad_norm": 1.5300803184509277,
      "learning_rate": 9.335684210526316e-05,
      "loss": 0.5184,
      "step": 45263
    },
    {
      "epoch": 2.534662336207862,
      "grad_norm": 1.216748595237732,
      "learning_rate": 9.335657894736842e-05,
      "loss": 0.3831,
      "step": 45264
    },
    {
      "epoch": 2.534718333519991,
      "grad_norm": 1.0298793315887451,
      "learning_rate": 9.335631578947369e-05,
      "loss": 0.354,
      "step": 45265
    },
    {
      "epoch": 2.53477433083212,
      "grad_norm": 1.0117430686950684,
      "learning_rate": 9.335605263157895e-05,
      "loss": 0.424,
      "step": 45266
    },
    {
      "epoch": 2.534830328144249,
      "grad_norm": 1.319228172302246,
      "learning_rate": 9.335578947368422e-05,
      "loss": 0.4444,
      "step": 45267
    },
    {
      "epoch": 2.534886325456378,
      "grad_norm": 1.2935110330581665,
      "learning_rate": 9.335552631578947e-05,
      "loss": 0.6328,
      "step": 45268
    },
    {
      "epoch": 2.534942322768507,
      "grad_norm": 1.1519806385040283,
      "learning_rate": 9.335526315789474e-05,
      "loss": 0.3794,
      "step": 45269
    },
    {
      "epoch": 2.534998320080636,
      "grad_norm": 1.0636451244354248,
      "learning_rate": 9.3355e-05,
      "loss": 0.3567,
      "step": 45270
    },
    {
      "epoch": 2.535054317392765,
      "grad_norm": 1.390102505683899,
      "learning_rate": 9.335473684210528e-05,
      "loss": 0.4264,
      "step": 45271
    },
    {
      "epoch": 2.535110314704894,
      "grad_norm": 1.598382830619812,
      "learning_rate": 9.335447368421054e-05,
      "loss": 0.4391,
      "step": 45272
    },
    {
      "epoch": 2.535166312017023,
      "grad_norm": 3.8355326652526855,
      "learning_rate": 9.335421052631578e-05,
      "loss": 0.6235,
      "step": 45273
    },
    {
      "epoch": 2.535222309329152,
      "grad_norm": 1.801727294921875,
      "learning_rate": 9.335394736842106e-05,
      "loss": 0.6115,
      "step": 45274
    },
    {
      "epoch": 2.535278306641281,
      "grad_norm": 1.1578021049499512,
      "learning_rate": 9.335368421052632e-05,
      "loss": 0.358,
      "step": 45275
    },
    {
      "epoch": 2.53533430395341,
      "grad_norm": 1.3351460695266724,
      "learning_rate": 9.335342105263159e-05,
      "loss": 0.4415,
      "step": 45276
    },
    {
      "epoch": 2.535390301265539,
      "grad_norm": 1.5685043334960938,
      "learning_rate": 9.335315789473685e-05,
      "loss": 0.3998,
      "step": 45277
    },
    {
      "epoch": 2.5354462985776682,
      "grad_norm": 1.455068588256836,
      "learning_rate": 9.335289473684211e-05,
      "loss": 0.4018,
      "step": 45278
    },
    {
      "epoch": 2.5355022958897973,
      "grad_norm": 1.0965807437896729,
      "learning_rate": 9.335263157894737e-05,
      "loss": 0.3251,
      "step": 45279
    },
    {
      "epoch": 2.5355582932019263,
      "grad_norm": 1.163636565208435,
      "learning_rate": 9.335236842105264e-05,
      "loss": 0.4291,
      "step": 45280
    },
    {
      "epoch": 2.5356142905140553,
      "grad_norm": 1.3461495637893677,
      "learning_rate": 9.33521052631579e-05,
      "loss": 0.5616,
      "step": 45281
    },
    {
      "epoch": 2.5356702878261843,
      "grad_norm": 1.5996973514556885,
      "learning_rate": 9.335184210526316e-05,
      "loss": 0.6035,
      "step": 45282
    },
    {
      "epoch": 2.5357262851383133,
      "grad_norm": 1.376313328742981,
      "learning_rate": 9.335157894736842e-05,
      "loss": 0.4519,
      "step": 45283
    },
    {
      "epoch": 2.5357822824504423,
      "grad_norm": 1.0798572301864624,
      "learning_rate": 9.33513157894737e-05,
      "loss": 0.3787,
      "step": 45284
    },
    {
      "epoch": 2.5358382797625714,
      "grad_norm": 1.2517855167388916,
      "learning_rate": 9.335105263157895e-05,
      "loss": 0.3608,
      "step": 45285
    },
    {
      "epoch": 2.5358942770747004,
      "grad_norm": 1.2725660800933838,
      "learning_rate": 9.335078947368421e-05,
      "loss": 0.3656,
      "step": 45286
    },
    {
      "epoch": 2.5359502743868294,
      "grad_norm": 1.1492847204208374,
      "learning_rate": 9.335052631578947e-05,
      "loss": 0.412,
      "step": 45287
    },
    {
      "epoch": 2.5360062716989584,
      "grad_norm": 1.1615641117095947,
      "learning_rate": 9.335026315789475e-05,
      "loss": 0.4044,
      "step": 45288
    },
    {
      "epoch": 2.5360622690110874,
      "grad_norm": 1.2334094047546387,
      "learning_rate": 9.335e-05,
      "loss": 0.3515,
      "step": 45289
    },
    {
      "epoch": 2.5361182663232165,
      "grad_norm": 1.7674583196640015,
      "learning_rate": 9.334973684210528e-05,
      "loss": 0.4485,
      "step": 45290
    },
    {
      "epoch": 2.5361742636353455,
      "grad_norm": 1.076705813407898,
      "learning_rate": 9.334947368421053e-05,
      "loss": 0.3818,
      "step": 45291
    },
    {
      "epoch": 2.5362302609474745,
      "grad_norm": 1.0605013370513916,
      "learning_rate": 9.334921052631578e-05,
      "loss": 0.5019,
      "step": 45292
    },
    {
      "epoch": 2.5362862582596035,
      "grad_norm": 1.206882357597351,
      "learning_rate": 9.334894736842106e-05,
      "loss": 0.357,
      "step": 45293
    },
    {
      "epoch": 2.5363422555717325,
      "grad_norm": 1.3598541021347046,
      "learning_rate": 9.334868421052632e-05,
      "loss": 0.4731,
      "step": 45294
    },
    {
      "epoch": 2.5363982528838616,
      "grad_norm": 1.396519422531128,
      "learning_rate": 9.334842105263159e-05,
      "loss": 0.549,
      "step": 45295
    },
    {
      "epoch": 2.5364542501959906,
      "grad_norm": 0.9747830033302307,
      "learning_rate": 9.334815789473684e-05,
      "loss": 0.3111,
      "step": 45296
    },
    {
      "epoch": 2.5365102475081196,
      "grad_norm": 1.3529704809188843,
      "learning_rate": 9.334789473684211e-05,
      "loss": 0.4898,
      "step": 45297
    },
    {
      "epoch": 2.5365662448202486,
      "grad_norm": 1.397033452987671,
      "learning_rate": 9.334763157894737e-05,
      "loss": 0.4368,
      "step": 45298
    },
    {
      "epoch": 2.5366222421323776,
      "grad_norm": 1.7389013767242432,
      "learning_rate": 9.334736842105264e-05,
      "loss": 0.5111,
      "step": 45299
    },
    {
      "epoch": 2.5366782394445067,
      "grad_norm": 1.1006819009780884,
      "learning_rate": 9.33471052631579e-05,
      "loss": 0.432,
      "step": 45300
    },
    {
      "epoch": 2.5367342367566357,
      "grad_norm": 0.9172592759132385,
      "learning_rate": 9.334684210526316e-05,
      "loss": 0.2798,
      "step": 45301
    },
    {
      "epoch": 2.5367902340687647,
      "grad_norm": 1.7873425483703613,
      "learning_rate": 9.334657894736842e-05,
      "loss": 0.506,
      "step": 45302
    },
    {
      "epoch": 2.5368462313808937,
      "grad_norm": 1.2786563634872437,
      "learning_rate": 9.33463157894737e-05,
      "loss": 0.4152,
      "step": 45303
    },
    {
      "epoch": 2.5369022286930227,
      "grad_norm": 1.6191579103469849,
      "learning_rate": 9.334605263157896e-05,
      "loss": 0.3895,
      "step": 45304
    },
    {
      "epoch": 2.5369582260051518,
      "grad_norm": 1.182167887687683,
      "learning_rate": 9.334578947368422e-05,
      "loss": 0.3104,
      "step": 45305
    },
    {
      "epoch": 2.5370142233172808,
      "grad_norm": 1.30289888381958,
      "learning_rate": 9.334552631578948e-05,
      "loss": 0.4209,
      "step": 45306
    },
    {
      "epoch": 2.53707022062941,
      "grad_norm": 1.35088312625885,
      "learning_rate": 9.334526315789475e-05,
      "loss": 0.4159,
      "step": 45307
    },
    {
      "epoch": 2.537126217941539,
      "grad_norm": 1.144099473953247,
      "learning_rate": 9.334500000000001e-05,
      "loss": 0.336,
      "step": 45308
    },
    {
      "epoch": 2.537182215253668,
      "grad_norm": 1.0986301898956299,
      "learning_rate": 9.334473684210527e-05,
      "loss": 0.436,
      "step": 45309
    },
    {
      "epoch": 2.537238212565797,
      "grad_norm": 1.4699419736862183,
      "learning_rate": 9.334447368421053e-05,
      "loss": 0.4374,
      "step": 45310
    },
    {
      "epoch": 2.537294209877926,
      "grad_norm": 1.1298091411590576,
      "learning_rate": 9.334421052631579e-05,
      "loss": 0.341,
      "step": 45311
    },
    {
      "epoch": 2.537350207190055,
      "grad_norm": 1.1286169290542603,
      "learning_rate": 9.334394736842106e-05,
      "loss": 0.396,
      "step": 45312
    },
    {
      "epoch": 2.537406204502184,
      "grad_norm": 1.0614938735961914,
      "learning_rate": 9.334368421052632e-05,
      "loss": 0.3653,
      "step": 45313
    },
    {
      "epoch": 2.537462201814313,
      "grad_norm": 1.449526309967041,
      "learning_rate": 9.334342105263158e-05,
      "loss": 0.3266,
      "step": 45314
    },
    {
      "epoch": 2.537518199126442,
      "grad_norm": 1.019828200340271,
      "learning_rate": 9.334315789473684e-05,
      "loss": 0.4084,
      "step": 45315
    },
    {
      "epoch": 2.537574196438571,
      "grad_norm": 1.0963102579116821,
      "learning_rate": 9.334289473684211e-05,
      "loss": 0.294,
      "step": 45316
    },
    {
      "epoch": 2.5376301937507,
      "grad_norm": 1.0415846109390259,
      "learning_rate": 9.334263157894737e-05,
      "loss": 0.4018,
      "step": 45317
    },
    {
      "epoch": 2.537686191062829,
      "grad_norm": 1.3552651405334473,
      "learning_rate": 9.334236842105263e-05,
      "loss": 0.3391,
      "step": 45318
    },
    {
      "epoch": 2.537742188374958,
      "grad_norm": 1.0553205013275146,
      "learning_rate": 9.334210526315789e-05,
      "loss": 0.4012,
      "step": 45319
    },
    {
      "epoch": 2.537798185687087,
      "grad_norm": 1.4028840065002441,
      "learning_rate": 9.334184210526317e-05,
      "loss": 0.5818,
      "step": 45320
    },
    {
      "epoch": 2.537854182999216,
      "grad_norm": 1.035807490348816,
      "learning_rate": 9.334157894736843e-05,
      "loss": 0.3006,
      "step": 45321
    },
    {
      "epoch": 2.537910180311345,
      "grad_norm": 0.9817575216293335,
      "learning_rate": 9.33413157894737e-05,
      "loss": 0.3093,
      "step": 45322
    },
    {
      "epoch": 2.537966177623474,
      "grad_norm": 1.3165676593780518,
      "learning_rate": 9.334105263157894e-05,
      "loss": 0.509,
      "step": 45323
    },
    {
      "epoch": 2.538022174935603,
      "grad_norm": 1.2350512742996216,
      "learning_rate": 9.334078947368422e-05,
      "loss": 0.4871,
      "step": 45324
    },
    {
      "epoch": 2.538078172247732,
      "grad_norm": 1.2546879053115845,
      "learning_rate": 9.334052631578948e-05,
      "loss": 0.3917,
      "step": 45325
    },
    {
      "epoch": 2.538134169559861,
      "grad_norm": 1.2261762619018555,
      "learning_rate": 9.334026315789474e-05,
      "loss": 0.3834,
      "step": 45326
    },
    {
      "epoch": 2.53819016687199,
      "grad_norm": 1.1783288717269897,
      "learning_rate": 9.334000000000001e-05,
      "loss": 0.3996,
      "step": 45327
    },
    {
      "epoch": 2.538246164184119,
      "grad_norm": 1.246935486793518,
      "learning_rate": 9.333973684210526e-05,
      "loss": 0.3269,
      "step": 45328
    },
    {
      "epoch": 2.538302161496248,
      "grad_norm": 1.1102646589279175,
      "learning_rate": 9.333947368421053e-05,
      "loss": 0.3952,
      "step": 45329
    },
    {
      "epoch": 2.5383581588083772,
      "grad_norm": 1.8977389335632324,
      "learning_rate": 9.333921052631579e-05,
      "loss": 0.3676,
      "step": 45330
    },
    {
      "epoch": 2.5384141561205062,
      "grad_norm": 1.0553873777389526,
      "learning_rate": 9.333894736842106e-05,
      "loss": 0.3199,
      "step": 45331
    },
    {
      "epoch": 2.5384701534326353,
      "grad_norm": 1.1308423280715942,
      "learning_rate": 9.333868421052632e-05,
      "loss": 0.4141,
      "step": 45332
    },
    {
      "epoch": 2.5385261507447643,
      "grad_norm": 1.1441901922225952,
      "learning_rate": 9.333842105263158e-05,
      "loss": 0.3888,
      "step": 45333
    },
    {
      "epoch": 2.5385821480568933,
      "grad_norm": 1.1931356191635132,
      "learning_rate": 9.333815789473684e-05,
      "loss": 0.3902,
      "step": 45334
    },
    {
      "epoch": 2.5386381453690223,
      "grad_norm": 1.1282448768615723,
      "learning_rate": 9.333789473684212e-05,
      "loss": 0.3455,
      "step": 45335
    },
    {
      "epoch": 2.5386941426811513,
      "grad_norm": 1.3573325872421265,
      "learning_rate": 9.333763157894738e-05,
      "loss": 0.3766,
      "step": 45336
    },
    {
      "epoch": 2.5387501399932804,
      "grad_norm": 1.047177791595459,
      "learning_rate": 9.333736842105264e-05,
      "loss": 0.4228,
      "step": 45337
    },
    {
      "epoch": 2.5388061373054094,
      "grad_norm": 1.1990817785263062,
      "learning_rate": 9.33371052631579e-05,
      "loss": 0.408,
      "step": 45338
    },
    {
      "epoch": 2.5388621346175384,
      "grad_norm": 1.1718990802764893,
      "learning_rate": 9.333684210526317e-05,
      "loss": 0.4653,
      "step": 45339
    },
    {
      "epoch": 2.5389181319296674,
      "grad_norm": 1.2509409189224243,
      "learning_rate": 9.333657894736843e-05,
      "loss": 0.4103,
      "step": 45340
    },
    {
      "epoch": 2.5389741292417964,
      "grad_norm": 1.145836353302002,
      "learning_rate": 9.333631578947369e-05,
      "loss": 0.3526,
      "step": 45341
    },
    {
      "epoch": 2.5390301265539255,
      "grad_norm": 1.0795141458511353,
      "learning_rate": 9.333605263157895e-05,
      "loss": 0.2988,
      "step": 45342
    },
    {
      "epoch": 2.5390861238660545,
      "grad_norm": 1.1540536880493164,
      "learning_rate": 9.333578947368421e-05,
      "loss": 0.3905,
      "step": 45343
    },
    {
      "epoch": 2.5391421211781835,
      "grad_norm": 1.049790620803833,
      "learning_rate": 9.333552631578948e-05,
      "loss": 0.4103,
      "step": 45344
    },
    {
      "epoch": 2.5391981184903125,
      "grad_norm": 1.0093870162963867,
      "learning_rate": 9.333526315789474e-05,
      "loss": 0.3091,
      "step": 45345
    },
    {
      "epoch": 2.5392541158024415,
      "grad_norm": 1.8718820810317993,
      "learning_rate": 9.3335e-05,
      "loss": 0.4476,
      "step": 45346
    },
    {
      "epoch": 2.5393101131145706,
      "grad_norm": 0.969196081161499,
      "learning_rate": 9.333473684210526e-05,
      "loss": 0.3074,
      "step": 45347
    },
    {
      "epoch": 2.5393661104266996,
      "grad_norm": 1.3815734386444092,
      "learning_rate": 9.333447368421053e-05,
      "loss": 0.3527,
      "step": 45348
    },
    {
      "epoch": 2.5394221077388286,
      "grad_norm": 1.0754064321517944,
      "learning_rate": 9.333421052631579e-05,
      "loss": 0.3193,
      "step": 45349
    },
    {
      "epoch": 2.5394781050509576,
      "grad_norm": 1.1894993782043457,
      "learning_rate": 9.333394736842107e-05,
      "loss": 0.4357,
      "step": 45350
    },
    {
      "epoch": 2.5395341023630866,
      "grad_norm": 1.2097920179367065,
      "learning_rate": 9.333368421052631e-05,
      "loss": 0.4072,
      "step": 45351
    },
    {
      "epoch": 2.5395900996752157,
      "grad_norm": 1.1850552558898926,
      "learning_rate": 9.333342105263159e-05,
      "loss": 0.4633,
      "step": 45352
    },
    {
      "epoch": 2.5396460969873447,
      "grad_norm": 1.4764612913131714,
      "learning_rate": 9.333315789473685e-05,
      "loss": 0.4032,
      "step": 45353
    },
    {
      "epoch": 2.5397020942994737,
      "grad_norm": 1.0935046672821045,
      "learning_rate": 9.333289473684212e-05,
      "loss": 0.3736,
      "step": 45354
    },
    {
      "epoch": 2.5397580916116027,
      "grad_norm": 1.6461457014083862,
      "learning_rate": 9.333263157894738e-05,
      "loss": 0.3943,
      "step": 45355
    },
    {
      "epoch": 2.5398140889237317,
      "grad_norm": 1.2088853120803833,
      "learning_rate": 9.333236842105264e-05,
      "loss": 0.3462,
      "step": 45356
    },
    {
      "epoch": 2.5398700862358607,
      "grad_norm": 1.3382636308670044,
      "learning_rate": 9.33321052631579e-05,
      "loss": 0.4208,
      "step": 45357
    },
    {
      "epoch": 2.5399260835479898,
      "grad_norm": 1.1970593929290771,
      "learning_rate": 9.333184210526317e-05,
      "loss": 0.3519,
      "step": 45358
    },
    {
      "epoch": 2.539982080860119,
      "grad_norm": 1.3002225160598755,
      "learning_rate": 9.333157894736843e-05,
      "loss": 0.4825,
      "step": 45359
    },
    {
      "epoch": 2.540038078172248,
      "grad_norm": 1.05095374584198,
      "learning_rate": 9.333131578947368e-05,
      "loss": 0.4767,
      "step": 45360
    },
    {
      "epoch": 2.540094075484377,
      "grad_norm": 1.097488284111023,
      "learning_rate": 9.333105263157895e-05,
      "loss": 0.4171,
      "step": 45361
    },
    {
      "epoch": 2.540150072796506,
      "grad_norm": 1.2171536684036255,
      "learning_rate": 9.333078947368421e-05,
      "loss": 0.4819,
      "step": 45362
    },
    {
      "epoch": 2.540206070108635,
      "grad_norm": 1.0202728509902954,
      "learning_rate": 9.333052631578948e-05,
      "loss": 0.2963,
      "step": 45363
    },
    {
      "epoch": 2.540262067420764,
      "grad_norm": 1.006606101989746,
      "learning_rate": 9.333026315789474e-05,
      "loss": 0.4239,
      "step": 45364
    },
    {
      "epoch": 2.540318064732893,
      "grad_norm": 1.2740468978881836,
      "learning_rate": 9.333e-05,
      "loss": 0.4306,
      "step": 45365
    },
    {
      "epoch": 2.540374062045022,
      "grad_norm": 0.925267219543457,
      "learning_rate": 9.332973684210526e-05,
      "loss": 0.2875,
      "step": 45366
    },
    {
      "epoch": 2.540430059357151,
      "grad_norm": 1.3783094882965088,
      "learning_rate": 9.332947368421054e-05,
      "loss": 0.3498,
      "step": 45367
    },
    {
      "epoch": 2.54048605666928,
      "grad_norm": 1.3869962692260742,
      "learning_rate": 9.33292105263158e-05,
      "loss": 0.3389,
      "step": 45368
    },
    {
      "epoch": 2.540542053981409,
      "grad_norm": 1.250387191772461,
      "learning_rate": 9.332894736842106e-05,
      "loss": 0.4545,
      "step": 45369
    },
    {
      "epoch": 2.540598051293538,
      "grad_norm": 1.0628342628479004,
      "learning_rate": 9.332868421052631e-05,
      "loss": 0.3775,
      "step": 45370
    },
    {
      "epoch": 2.540654048605667,
      "grad_norm": 1.237439751625061,
      "learning_rate": 9.332842105263159e-05,
      "loss": 0.3827,
      "step": 45371
    },
    {
      "epoch": 2.540710045917796,
      "grad_norm": 1.0984333753585815,
      "learning_rate": 9.332815789473685e-05,
      "loss": 0.4277,
      "step": 45372
    },
    {
      "epoch": 2.540766043229925,
      "grad_norm": 1.289589524269104,
      "learning_rate": 9.332789473684211e-05,
      "loss": 0.4805,
      "step": 45373
    },
    {
      "epoch": 2.540822040542054,
      "grad_norm": 1.5857789516448975,
      "learning_rate": 9.332763157894737e-05,
      "loss": 0.5066,
      "step": 45374
    },
    {
      "epoch": 2.540878037854183,
      "grad_norm": 1.4003747701644897,
      "learning_rate": 9.332736842105264e-05,
      "loss": 0.4243,
      "step": 45375
    },
    {
      "epoch": 2.540934035166312,
      "grad_norm": 1.0537551641464233,
      "learning_rate": 9.33271052631579e-05,
      "loss": 0.3499,
      "step": 45376
    },
    {
      "epoch": 2.540990032478441,
      "grad_norm": 1.1186370849609375,
      "learning_rate": 9.332684210526316e-05,
      "loss": 0.3504,
      "step": 45377
    },
    {
      "epoch": 2.54104602979057,
      "grad_norm": 1.3935378789901733,
      "learning_rate": 9.332657894736842e-05,
      "loss": 0.5814,
      "step": 45378
    },
    {
      "epoch": 2.541102027102699,
      "grad_norm": 1.0203542709350586,
      "learning_rate": 9.332631578947368e-05,
      "loss": 0.4926,
      "step": 45379
    },
    {
      "epoch": 2.541158024414828,
      "grad_norm": 1.0868418216705322,
      "learning_rate": 9.332605263157895e-05,
      "loss": 0.2937,
      "step": 45380
    },
    {
      "epoch": 2.541214021726957,
      "grad_norm": 1.2829493284225464,
      "learning_rate": 9.332578947368421e-05,
      "loss": 0.5876,
      "step": 45381
    },
    {
      "epoch": 2.5412700190390862,
      "grad_norm": 1.0392296314239502,
      "learning_rate": 9.332552631578949e-05,
      "loss": 0.3307,
      "step": 45382
    },
    {
      "epoch": 2.5413260163512152,
      "grad_norm": 1.1340253353118896,
      "learning_rate": 9.332526315789473e-05,
      "loss": 0.4648,
      "step": 45383
    },
    {
      "epoch": 2.5413820136633443,
      "grad_norm": 1.3888187408447266,
      "learning_rate": 9.3325e-05,
      "loss": 0.3755,
      "step": 45384
    },
    {
      "epoch": 2.5414380109754733,
      "grad_norm": 1.1885852813720703,
      "learning_rate": 9.332473684210526e-05,
      "loss": 0.5954,
      "step": 45385
    },
    {
      "epoch": 2.5414940082876023,
      "grad_norm": 0.9921424984931946,
      "learning_rate": 9.332447368421054e-05,
      "loss": 0.4178,
      "step": 45386
    },
    {
      "epoch": 2.5415500055997313,
      "grad_norm": 0.9037467241287231,
      "learning_rate": 9.33242105263158e-05,
      "loss": 0.3812,
      "step": 45387
    },
    {
      "epoch": 2.5416060029118603,
      "grad_norm": 1.3360651731491089,
      "learning_rate": 9.332394736842106e-05,
      "loss": 0.3397,
      "step": 45388
    },
    {
      "epoch": 2.5416620002239894,
      "grad_norm": 1.2790435552597046,
      "learning_rate": 9.332368421052632e-05,
      "loss": 0.3841,
      "step": 45389
    },
    {
      "epoch": 2.5417179975361184,
      "grad_norm": 1.0188674926757812,
      "learning_rate": 9.332342105263159e-05,
      "loss": 0.3353,
      "step": 45390
    },
    {
      "epoch": 2.5417739948482474,
      "grad_norm": 1.1912813186645508,
      "learning_rate": 9.332315789473685e-05,
      "loss": 0.3891,
      "step": 45391
    },
    {
      "epoch": 2.5418299921603764,
      "grad_norm": 1.2114315032958984,
      "learning_rate": 9.332289473684211e-05,
      "loss": 0.3174,
      "step": 45392
    },
    {
      "epoch": 2.5418859894725054,
      "grad_norm": 1.1330804824829102,
      "learning_rate": 9.332263157894737e-05,
      "loss": 0.4282,
      "step": 45393
    },
    {
      "epoch": 2.5419419867846345,
      "grad_norm": 1.0338659286499023,
      "learning_rate": 9.332236842105263e-05,
      "loss": 0.4529,
      "step": 45394
    },
    {
      "epoch": 2.5419979840967635,
      "grad_norm": 1.4313313961029053,
      "learning_rate": 9.33221052631579e-05,
      "loss": 0.3806,
      "step": 45395
    },
    {
      "epoch": 2.5420539814088925,
      "grad_norm": 1.3672512769699097,
      "learning_rate": 9.332184210526316e-05,
      "loss": 0.3381,
      "step": 45396
    },
    {
      "epoch": 2.5421099787210215,
      "grad_norm": 1.4325401782989502,
      "learning_rate": 9.332157894736842e-05,
      "loss": 0.5508,
      "step": 45397
    },
    {
      "epoch": 2.5421659760331505,
      "grad_norm": 1.0843567848205566,
      "learning_rate": 9.332131578947368e-05,
      "loss": 0.3739,
      "step": 45398
    },
    {
      "epoch": 2.5422219733452796,
      "grad_norm": 1.3448868989944458,
      "learning_rate": 9.332105263157896e-05,
      "loss": 0.4028,
      "step": 45399
    },
    {
      "epoch": 2.5422779706574086,
      "grad_norm": 0.9155479073524475,
      "learning_rate": 9.332078947368422e-05,
      "loss": 0.313,
      "step": 45400
    },
    {
      "epoch": 2.5423339679695376,
      "grad_norm": 1.1040098667144775,
      "learning_rate": 9.332052631578947e-05,
      "loss": 0.6229,
      "step": 45401
    },
    {
      "epoch": 2.5423899652816666,
      "grad_norm": 1.0508986711502075,
      "learning_rate": 9.332026315789473e-05,
      "loss": 0.3536,
      "step": 45402
    },
    {
      "epoch": 2.5424459625937956,
      "grad_norm": 1.090962529182434,
      "learning_rate": 9.332000000000001e-05,
      "loss": 0.3271,
      "step": 45403
    },
    {
      "epoch": 2.5425019599059246,
      "grad_norm": 1.3420130014419556,
      "learning_rate": 9.331973684210527e-05,
      "loss": 0.3727,
      "step": 45404
    },
    {
      "epoch": 2.5425579572180537,
      "grad_norm": 1.089132308959961,
      "learning_rate": 9.331947368421054e-05,
      "loss": 0.346,
      "step": 45405
    },
    {
      "epoch": 2.5426139545301827,
      "grad_norm": 1.1936652660369873,
      "learning_rate": 9.331921052631579e-05,
      "loss": 0.3804,
      "step": 45406
    },
    {
      "epoch": 2.5426699518423117,
      "grad_norm": 1.0668883323669434,
      "learning_rate": 9.331894736842106e-05,
      "loss": 0.3809,
      "step": 45407
    },
    {
      "epoch": 2.5427259491544407,
      "grad_norm": 1.3738937377929688,
      "learning_rate": 9.331868421052632e-05,
      "loss": 0.4322,
      "step": 45408
    },
    {
      "epoch": 2.5427819464665697,
      "grad_norm": 1.1574946641921997,
      "learning_rate": 9.33184210526316e-05,
      "loss": 0.3818,
      "step": 45409
    },
    {
      "epoch": 2.5428379437786988,
      "grad_norm": 1.4381228685379028,
      "learning_rate": 9.331815789473685e-05,
      "loss": 0.4719,
      "step": 45410
    },
    {
      "epoch": 2.542893941090828,
      "grad_norm": 1.4009007215499878,
      "learning_rate": 9.331789473684211e-05,
      "loss": 0.4715,
      "step": 45411
    },
    {
      "epoch": 2.542949938402957,
      "grad_norm": 1.1210962533950806,
      "learning_rate": 9.331763157894737e-05,
      "loss": 0.309,
      "step": 45412
    },
    {
      "epoch": 2.543005935715086,
      "grad_norm": 1.0451971292495728,
      "learning_rate": 9.331736842105263e-05,
      "loss": 0.3305,
      "step": 45413
    },
    {
      "epoch": 2.543061933027215,
      "grad_norm": 1.4638595581054688,
      "learning_rate": 9.33171052631579e-05,
      "loss": 0.4155,
      "step": 45414
    },
    {
      "epoch": 2.543117930339344,
      "grad_norm": 1.1173802614212036,
      "learning_rate": 9.331684210526315e-05,
      "loss": 0.5106,
      "step": 45415
    },
    {
      "epoch": 2.543173927651473,
      "grad_norm": 1.135046124458313,
      "learning_rate": 9.331657894736842e-05,
      "loss": 0.3677,
      "step": 45416
    },
    {
      "epoch": 2.543229924963602,
      "grad_norm": 1.26047945022583,
      "learning_rate": 9.331631578947368e-05,
      "loss": 0.3525,
      "step": 45417
    },
    {
      "epoch": 2.543285922275731,
      "grad_norm": 1.547135591506958,
      "learning_rate": 9.331605263157896e-05,
      "loss": 0.5662,
      "step": 45418
    },
    {
      "epoch": 2.54334191958786,
      "grad_norm": 1.1728248596191406,
      "learning_rate": 9.331578947368422e-05,
      "loss": 0.3539,
      "step": 45419
    },
    {
      "epoch": 2.543397916899989,
      "grad_norm": 1.2535096406936646,
      "learning_rate": 9.331552631578948e-05,
      "loss": 0.4595,
      "step": 45420
    },
    {
      "epoch": 2.543453914212118,
      "grad_norm": 1.23871648311615,
      "learning_rate": 9.331526315789474e-05,
      "loss": 0.3997,
      "step": 45421
    },
    {
      "epoch": 2.543509911524247,
      "grad_norm": 1.249550700187683,
      "learning_rate": 9.331500000000001e-05,
      "loss": 0.3602,
      "step": 45422
    },
    {
      "epoch": 2.543565908836376,
      "grad_norm": 1.2181986570358276,
      "learning_rate": 9.331473684210527e-05,
      "loss": 0.3148,
      "step": 45423
    },
    {
      "epoch": 2.543621906148505,
      "grad_norm": 1.0738900899887085,
      "learning_rate": 9.331447368421053e-05,
      "loss": 0.3604,
      "step": 45424
    },
    {
      "epoch": 2.543677903460634,
      "grad_norm": 1.245262861251831,
      "learning_rate": 9.331421052631579e-05,
      "loss": 0.4391,
      "step": 45425
    },
    {
      "epoch": 2.543733900772763,
      "grad_norm": 1.3473206758499146,
      "learning_rate": 9.331394736842106e-05,
      "loss": 0.5701,
      "step": 45426
    },
    {
      "epoch": 2.543789898084892,
      "grad_norm": 1.2382371425628662,
      "learning_rate": 9.331368421052632e-05,
      "loss": 0.3456,
      "step": 45427
    },
    {
      "epoch": 2.543845895397021,
      "grad_norm": 1.2661610841751099,
      "learning_rate": 9.331342105263158e-05,
      "loss": 0.4297,
      "step": 45428
    },
    {
      "epoch": 2.54390189270915,
      "grad_norm": 1.0677636861801147,
      "learning_rate": 9.331315789473684e-05,
      "loss": 0.3873,
      "step": 45429
    },
    {
      "epoch": 2.543957890021279,
      "grad_norm": 1.1617891788482666,
      "learning_rate": 9.33128947368421e-05,
      "loss": 0.3592,
      "step": 45430
    },
    {
      "epoch": 2.544013887333408,
      "grad_norm": 1.2983496189117432,
      "learning_rate": 9.331263157894738e-05,
      "loss": 0.4615,
      "step": 45431
    },
    {
      "epoch": 2.5440698846455367,
      "grad_norm": 1.3749406337738037,
      "learning_rate": 9.331236842105263e-05,
      "loss": 0.5818,
      "step": 45432
    },
    {
      "epoch": 2.544125881957666,
      "grad_norm": 1.2134195566177368,
      "learning_rate": 9.33121052631579e-05,
      "loss": 0.4888,
      "step": 45433
    },
    {
      "epoch": 2.5441818792697948,
      "grad_norm": 0.9438391923904419,
      "learning_rate": 9.331184210526315e-05,
      "loss": 0.3715,
      "step": 45434
    },
    {
      "epoch": 2.5442378765819242,
      "grad_norm": 1.4753665924072266,
      "learning_rate": 9.331157894736843e-05,
      "loss": 0.4589,
      "step": 45435
    },
    {
      "epoch": 2.544293873894053,
      "grad_norm": 1.2656809091567993,
      "learning_rate": 9.331131578947369e-05,
      "loss": 0.4044,
      "step": 45436
    },
    {
      "epoch": 2.5443498712061823,
      "grad_norm": 1.1619596481323242,
      "learning_rate": 9.331105263157896e-05,
      "loss": 0.3703,
      "step": 45437
    },
    {
      "epoch": 2.544405868518311,
      "grad_norm": 1.2375251054763794,
      "learning_rate": 9.33107894736842e-05,
      "loss": 0.4455,
      "step": 45438
    },
    {
      "epoch": 2.5444618658304403,
      "grad_norm": 1.629641056060791,
      "learning_rate": 9.331052631578948e-05,
      "loss": 0.548,
      "step": 45439
    },
    {
      "epoch": 2.544517863142569,
      "grad_norm": 1.3591606616973877,
      "learning_rate": 9.331026315789474e-05,
      "loss": 0.4843,
      "step": 45440
    },
    {
      "epoch": 2.5445738604546984,
      "grad_norm": 1.1018253564834595,
      "learning_rate": 9.331000000000001e-05,
      "loss": 0.41,
      "step": 45441
    },
    {
      "epoch": 2.544629857766827,
      "grad_norm": 1.372522234916687,
      "learning_rate": 9.330973684210527e-05,
      "loss": 0.5011,
      "step": 45442
    },
    {
      "epoch": 2.5446858550789564,
      "grad_norm": 2.0608558654785156,
      "learning_rate": 9.330947368421053e-05,
      "loss": 0.4013,
      "step": 45443
    },
    {
      "epoch": 2.544741852391085,
      "grad_norm": 1.164616584777832,
      "learning_rate": 9.330921052631579e-05,
      "loss": 0.5258,
      "step": 45444
    },
    {
      "epoch": 2.5447978497032144,
      "grad_norm": 1.1393369436264038,
      "learning_rate": 9.330894736842107e-05,
      "loss": 0.3221,
      "step": 45445
    },
    {
      "epoch": 2.544853847015343,
      "grad_norm": 1.0013407468795776,
      "learning_rate": 9.330868421052633e-05,
      "loss": 0.4308,
      "step": 45446
    },
    {
      "epoch": 2.5449098443274725,
      "grad_norm": 1.4113047122955322,
      "learning_rate": 9.330842105263158e-05,
      "loss": 0.5197,
      "step": 45447
    },
    {
      "epoch": 2.544965841639601,
      "grad_norm": 1.1284620761871338,
      "learning_rate": 9.330815789473684e-05,
      "loss": 0.4599,
      "step": 45448
    },
    {
      "epoch": 2.5450218389517305,
      "grad_norm": 1.1629842519760132,
      "learning_rate": 9.33078947368421e-05,
      "loss": 0.3811,
      "step": 45449
    },
    {
      "epoch": 2.545077836263859,
      "grad_norm": 0.9989349246025085,
      "learning_rate": 9.330763157894738e-05,
      "loss": 0.2139,
      "step": 45450
    },
    {
      "epoch": 2.5451338335759885,
      "grad_norm": 1.1623722314834595,
      "learning_rate": 9.330736842105264e-05,
      "loss": 0.4573,
      "step": 45451
    },
    {
      "epoch": 2.545189830888117,
      "grad_norm": 1.3423378467559814,
      "learning_rate": 9.33071052631579e-05,
      "loss": 0.39,
      "step": 45452
    },
    {
      "epoch": 2.5452458282002466,
      "grad_norm": 1.486692190170288,
      "learning_rate": 9.330684210526316e-05,
      "loss": 0.6149,
      "step": 45453
    },
    {
      "epoch": 2.545301825512375,
      "grad_norm": 1.2401117086410522,
      "learning_rate": 9.330657894736843e-05,
      "loss": 0.4917,
      "step": 45454
    },
    {
      "epoch": 2.5453578228245046,
      "grad_norm": 1.0880720615386963,
      "learning_rate": 9.330631578947369e-05,
      "loss": 0.4575,
      "step": 45455
    },
    {
      "epoch": 2.545413820136633,
      "grad_norm": 1.3601131439208984,
      "learning_rate": 9.330605263157895e-05,
      "loss": 0.5534,
      "step": 45456
    },
    {
      "epoch": 2.5454698174487627,
      "grad_norm": 1.218003749847412,
      "learning_rate": 9.330578947368421e-05,
      "loss": 0.372,
      "step": 45457
    },
    {
      "epoch": 2.5455258147608912,
      "grad_norm": 1.161191701889038,
      "learning_rate": 9.330552631578948e-05,
      "loss": 0.352,
      "step": 45458
    },
    {
      "epoch": 2.5455818120730207,
      "grad_norm": 1.1991173028945923,
      "learning_rate": 9.330526315789474e-05,
      "loss": 0.4329,
      "step": 45459
    },
    {
      "epoch": 2.5456378093851493,
      "grad_norm": 1.2213611602783203,
      "learning_rate": 9.330500000000002e-05,
      "loss": 0.3869,
      "step": 45460
    },
    {
      "epoch": 2.5456938066972787,
      "grad_norm": 1.490718126296997,
      "learning_rate": 9.330473684210526e-05,
      "loss": 0.4245,
      "step": 45461
    },
    {
      "epoch": 2.5457498040094073,
      "grad_norm": 1.0505620241165161,
      "learning_rate": 9.330447368421053e-05,
      "loss": 0.3715,
      "step": 45462
    },
    {
      "epoch": 2.5458058013215368,
      "grad_norm": 1.304104208946228,
      "learning_rate": 9.33042105263158e-05,
      "loss": 0.4309,
      "step": 45463
    },
    {
      "epoch": 2.5458617986336654,
      "grad_norm": 1.1874130964279175,
      "learning_rate": 9.330394736842105e-05,
      "loss": 0.6305,
      "step": 45464
    },
    {
      "epoch": 2.545917795945795,
      "grad_norm": 1.0002580881118774,
      "learning_rate": 9.330368421052631e-05,
      "loss": 0.3999,
      "step": 45465
    },
    {
      "epoch": 2.5459737932579234,
      "grad_norm": 1.0980788469314575,
      "learning_rate": 9.330342105263157e-05,
      "loss": 0.3493,
      "step": 45466
    },
    {
      "epoch": 2.546029790570053,
      "grad_norm": 1.0435987710952759,
      "learning_rate": 9.330315789473685e-05,
      "loss": 0.3904,
      "step": 45467
    },
    {
      "epoch": 2.5460857878821814,
      "grad_norm": 1.568314552307129,
      "learning_rate": 9.330289473684211e-05,
      "loss": 0.383,
      "step": 45468
    },
    {
      "epoch": 2.546141785194311,
      "grad_norm": 1.3833121061325073,
      "learning_rate": 9.330263157894738e-05,
      "loss": 0.4097,
      "step": 45469
    },
    {
      "epoch": 2.5461977825064395,
      "grad_norm": 1.0114573240280151,
      "learning_rate": 9.330236842105263e-05,
      "loss": 0.3791,
      "step": 45470
    },
    {
      "epoch": 2.546253779818569,
      "grad_norm": 0.9822787046432495,
      "learning_rate": 9.33021052631579e-05,
      "loss": 0.3211,
      "step": 45471
    },
    {
      "epoch": 2.5463097771306975,
      "grad_norm": 1.2400672435760498,
      "learning_rate": 9.330184210526316e-05,
      "loss": 0.5172,
      "step": 45472
    },
    {
      "epoch": 2.546365774442827,
      "grad_norm": 1.0592838525772095,
      "learning_rate": 9.330157894736843e-05,
      "loss": 0.3034,
      "step": 45473
    },
    {
      "epoch": 2.5464217717549555,
      "grad_norm": 1.0443229675292969,
      "learning_rate": 9.330131578947369e-05,
      "loss": 0.4256,
      "step": 45474
    },
    {
      "epoch": 2.546477769067085,
      "grad_norm": 1.2022860050201416,
      "learning_rate": 9.330105263157895e-05,
      "loss": 0.3589,
      "step": 45475
    },
    {
      "epoch": 2.5465337663792136,
      "grad_norm": 1.1963139772415161,
      "learning_rate": 9.330078947368421e-05,
      "loss": 0.4322,
      "step": 45476
    },
    {
      "epoch": 2.546589763691343,
      "grad_norm": 1.0549054145812988,
      "learning_rate": 9.330052631578949e-05,
      "loss": 0.3288,
      "step": 45477
    },
    {
      "epoch": 2.5466457610034716,
      "grad_norm": 1.3584388494491577,
      "learning_rate": 9.330026315789474e-05,
      "loss": 0.4263,
      "step": 45478
    },
    {
      "epoch": 2.546701758315601,
      "grad_norm": 1.3955621719360352,
      "learning_rate": 9.33e-05,
      "loss": 0.4291,
      "step": 45479
    },
    {
      "epoch": 2.5467577556277297,
      "grad_norm": 1.454166293144226,
      "learning_rate": 9.329973684210526e-05,
      "loss": 0.5663,
      "step": 45480
    },
    {
      "epoch": 2.546813752939859,
      "grad_norm": 1.1491954326629639,
      "learning_rate": 9.329947368421052e-05,
      "loss": 0.398,
      "step": 45481
    },
    {
      "epoch": 2.5468697502519877,
      "grad_norm": 1.2691184282302856,
      "learning_rate": 9.32992105263158e-05,
      "loss": 0.3803,
      "step": 45482
    },
    {
      "epoch": 2.546925747564117,
      "grad_norm": 1.129529595375061,
      "learning_rate": 9.329894736842106e-05,
      "loss": 0.3835,
      "step": 45483
    },
    {
      "epoch": 2.5469817448762457,
      "grad_norm": 1.1957905292510986,
      "learning_rate": 9.329868421052632e-05,
      "loss": 0.4623,
      "step": 45484
    },
    {
      "epoch": 2.547037742188375,
      "grad_norm": 1.1748223304748535,
      "learning_rate": 9.329842105263158e-05,
      "loss": 0.4202,
      "step": 45485
    },
    {
      "epoch": 2.5470937395005038,
      "grad_norm": 1.213054895401001,
      "learning_rate": 9.329815789473685e-05,
      "loss": 0.4349,
      "step": 45486
    },
    {
      "epoch": 2.5471497368126332,
      "grad_norm": 1.1977232694625854,
      "learning_rate": 9.329789473684211e-05,
      "loss": 0.316,
      "step": 45487
    },
    {
      "epoch": 2.547205734124762,
      "grad_norm": 1.4981600046157837,
      "learning_rate": 9.329763157894737e-05,
      "loss": 0.3926,
      "step": 45488
    },
    {
      "epoch": 2.5472617314368913,
      "grad_norm": 1.3002710342407227,
      "learning_rate": 9.329736842105263e-05,
      "loss": 0.4907,
      "step": 45489
    },
    {
      "epoch": 2.54731772874902,
      "grad_norm": 1.3732635974884033,
      "learning_rate": 9.32971052631579e-05,
      "loss": 0.5696,
      "step": 45490
    },
    {
      "epoch": 2.5473737260611493,
      "grad_norm": 1.5697863101959229,
      "learning_rate": 9.329684210526316e-05,
      "loss": 0.362,
      "step": 45491
    },
    {
      "epoch": 2.547429723373278,
      "grad_norm": 1.3127086162567139,
      "learning_rate": 9.329657894736844e-05,
      "loss": 0.3902,
      "step": 45492
    },
    {
      "epoch": 2.5474857206854074,
      "grad_norm": 1.1133718490600586,
      "learning_rate": 9.329631578947368e-05,
      "loss": 0.4,
      "step": 45493
    },
    {
      "epoch": 2.547541717997536,
      "grad_norm": 1.1892738342285156,
      "learning_rate": 9.329605263157895e-05,
      "loss": 0.3888,
      "step": 45494
    },
    {
      "epoch": 2.5475977153096654,
      "grad_norm": 1.3525183200836182,
      "learning_rate": 9.329578947368421e-05,
      "loss": 0.4122,
      "step": 45495
    },
    {
      "epoch": 2.547653712621794,
      "grad_norm": 1.0153789520263672,
      "learning_rate": 9.329552631578949e-05,
      "loss": 0.4387,
      "step": 45496
    },
    {
      "epoch": 2.5477097099339234,
      "grad_norm": 1.2033063173294067,
      "learning_rate": 9.329526315789475e-05,
      "loss": 0.456,
      "step": 45497
    },
    {
      "epoch": 2.547765707246052,
      "grad_norm": 1.0247381925582886,
      "learning_rate": 9.3295e-05,
      "loss": 0.4086,
      "step": 45498
    },
    {
      "epoch": 2.5478217045581815,
      "grad_norm": 1.1156585216522217,
      "learning_rate": 9.329473684210527e-05,
      "loss": 0.4221,
      "step": 45499
    },
    {
      "epoch": 2.54787770187031,
      "grad_norm": 1.5074691772460938,
      "learning_rate": 9.329447368421053e-05,
      "loss": 0.3805,
      "step": 45500
    },
    {
      "epoch": 2.5479336991824395,
      "grad_norm": 1.043045997619629,
      "learning_rate": 9.32942105263158e-05,
      "loss": 0.3723,
      "step": 45501
    },
    {
      "epoch": 2.547989696494568,
      "grad_norm": 1.202847957611084,
      "learning_rate": 9.329394736842106e-05,
      "loss": 0.3136,
      "step": 45502
    },
    {
      "epoch": 2.5480456938066975,
      "grad_norm": 1.126823902130127,
      "learning_rate": 9.329368421052632e-05,
      "loss": 0.2795,
      "step": 45503
    },
    {
      "epoch": 2.548101691118826,
      "grad_norm": 1.3956373929977417,
      "learning_rate": 9.329342105263158e-05,
      "loss": 0.3798,
      "step": 45504
    },
    {
      "epoch": 2.5481576884309556,
      "grad_norm": 1.2856290340423584,
      "learning_rate": 9.329315789473685e-05,
      "loss": 0.4415,
      "step": 45505
    },
    {
      "epoch": 2.548213685743084,
      "grad_norm": 1.1042677164077759,
      "learning_rate": 9.329289473684211e-05,
      "loss": 0.3779,
      "step": 45506
    },
    {
      "epoch": 2.5482696830552136,
      "grad_norm": 1.2296535968780518,
      "learning_rate": 9.329263157894737e-05,
      "loss": 0.4062,
      "step": 45507
    },
    {
      "epoch": 2.548325680367342,
      "grad_norm": 1.1364452838897705,
      "learning_rate": 9.329236842105263e-05,
      "loss": 0.4145,
      "step": 45508
    },
    {
      "epoch": 2.5483816776794717,
      "grad_norm": 1.105017066001892,
      "learning_rate": 9.32921052631579e-05,
      "loss": 0.374,
      "step": 45509
    },
    {
      "epoch": 2.5484376749916002,
      "grad_norm": 1.396712064743042,
      "learning_rate": 9.329184210526316e-05,
      "loss": 0.4227,
      "step": 45510
    },
    {
      "epoch": 2.5484936723037297,
      "grad_norm": 1.1925255060195923,
      "learning_rate": 9.329157894736842e-05,
      "loss": 0.4084,
      "step": 45511
    },
    {
      "epoch": 2.5485496696158583,
      "grad_norm": 1.0134319067001343,
      "learning_rate": 9.329131578947368e-05,
      "loss": 0.2773,
      "step": 45512
    },
    {
      "epoch": 2.5486056669279877,
      "grad_norm": 1.4131910800933838,
      "learning_rate": 9.329105263157896e-05,
      "loss": 0.4967,
      "step": 45513
    },
    {
      "epoch": 2.5486616642401163,
      "grad_norm": 1.197297215461731,
      "learning_rate": 9.329078947368422e-05,
      "loss": 0.4896,
      "step": 45514
    },
    {
      "epoch": 2.5487176615522453,
      "grad_norm": 1.1274511814117432,
      "learning_rate": 9.329052631578948e-05,
      "loss": 0.451,
      "step": 45515
    },
    {
      "epoch": 2.5487736588643743,
      "grad_norm": 1.1043752431869507,
      "learning_rate": 9.329026315789474e-05,
      "loss": 0.3595,
      "step": 45516
    },
    {
      "epoch": 2.5488296561765034,
      "grad_norm": 1.386168122291565,
      "learning_rate": 9.329e-05,
      "loss": 0.4609,
      "step": 45517
    },
    {
      "epoch": 2.5488856534886324,
      "grad_norm": 1.2402522563934326,
      "learning_rate": 9.328973684210527e-05,
      "loss": 0.3086,
      "step": 45518
    },
    {
      "epoch": 2.5489416508007614,
      "grad_norm": 1.0370854139328003,
      "learning_rate": 9.328947368421053e-05,
      "loss": 0.3099,
      "step": 45519
    },
    {
      "epoch": 2.5489976481128904,
      "grad_norm": 3.557767152786255,
      "learning_rate": 9.328921052631579e-05,
      "loss": 0.4006,
      "step": 45520
    },
    {
      "epoch": 2.5490536454250194,
      "grad_norm": 1.1421376466751099,
      "learning_rate": 9.328894736842105e-05,
      "loss": 0.3455,
      "step": 45521
    },
    {
      "epoch": 2.5491096427371485,
      "grad_norm": 1.2116714715957642,
      "learning_rate": 9.328868421052632e-05,
      "loss": 0.4625,
      "step": 45522
    },
    {
      "epoch": 2.5491656400492775,
      "grad_norm": 1.2807080745697021,
      "learning_rate": 9.328842105263158e-05,
      "loss": 0.3261,
      "step": 45523
    },
    {
      "epoch": 2.5492216373614065,
      "grad_norm": 1.1547012329101562,
      "learning_rate": 9.328815789473685e-05,
      "loss": 0.4012,
      "step": 45524
    },
    {
      "epoch": 2.5492776346735355,
      "grad_norm": 1.322356104850769,
      "learning_rate": 9.32878947368421e-05,
      "loss": 0.4321,
      "step": 45525
    },
    {
      "epoch": 2.5493336319856645,
      "grad_norm": 1.0448459386825562,
      "learning_rate": 9.328763157894737e-05,
      "loss": 0.3675,
      "step": 45526
    },
    {
      "epoch": 2.5493896292977936,
      "grad_norm": 1.3963772058486938,
      "learning_rate": 9.328736842105263e-05,
      "loss": 0.6169,
      "step": 45527
    },
    {
      "epoch": 2.5494456266099226,
      "grad_norm": 1.1270365715026855,
      "learning_rate": 9.328710526315791e-05,
      "loss": 0.3338,
      "step": 45528
    },
    {
      "epoch": 2.5495016239220516,
      "grad_norm": 1.1276499032974243,
      "learning_rate": 9.328684210526317e-05,
      "loss": 0.3833,
      "step": 45529
    },
    {
      "epoch": 2.5495576212341806,
      "grad_norm": 1.1946803331375122,
      "learning_rate": 9.328657894736843e-05,
      "loss": 0.4669,
      "step": 45530
    },
    {
      "epoch": 2.5496136185463096,
      "grad_norm": 1.0077214241027832,
      "learning_rate": 9.328631578947369e-05,
      "loss": 0.3313,
      "step": 45531
    },
    {
      "epoch": 2.5496696158584387,
      "grad_norm": 1.118607521057129,
      "learning_rate": 9.328605263157896e-05,
      "loss": 0.4058,
      "step": 45532
    },
    {
      "epoch": 2.5497256131705677,
      "grad_norm": 1.1313105821609497,
      "learning_rate": 9.328578947368422e-05,
      "loss": 0.3715,
      "step": 45533
    },
    {
      "epoch": 2.5497816104826967,
      "grad_norm": 1.1950281858444214,
      "learning_rate": 9.328552631578948e-05,
      "loss": 0.3671,
      "step": 45534
    },
    {
      "epoch": 2.5498376077948257,
      "grad_norm": 0.9588315486907959,
      "learning_rate": 9.328526315789474e-05,
      "loss": 0.28,
      "step": 45535
    },
    {
      "epoch": 2.5498936051069547,
      "grad_norm": 1.04447340965271,
      "learning_rate": 9.3285e-05,
      "loss": 0.3435,
      "step": 45536
    },
    {
      "epoch": 2.5499496024190837,
      "grad_norm": 1.1422456502914429,
      "learning_rate": 9.328473684210527e-05,
      "loss": 0.3618,
      "step": 45537
    },
    {
      "epoch": 2.5500055997312128,
      "grad_norm": 1.268076777458191,
      "learning_rate": 9.328447368421053e-05,
      "loss": 0.4105,
      "step": 45538
    },
    {
      "epoch": 2.550061597043342,
      "grad_norm": 1.2519358396530151,
      "learning_rate": 9.328421052631579e-05,
      "loss": 0.382,
      "step": 45539
    },
    {
      "epoch": 2.550117594355471,
      "grad_norm": 1.3188177347183228,
      "learning_rate": 9.328394736842105e-05,
      "loss": 0.2919,
      "step": 45540
    },
    {
      "epoch": 2.5501735916676,
      "grad_norm": 1.4831511974334717,
      "learning_rate": 9.328368421052632e-05,
      "loss": 0.4791,
      "step": 45541
    },
    {
      "epoch": 2.550229588979729,
      "grad_norm": 1.0797375440597534,
      "learning_rate": 9.328342105263158e-05,
      "loss": 0.4061,
      "step": 45542
    },
    {
      "epoch": 2.550285586291858,
      "grad_norm": 1.144763469696045,
      "learning_rate": 9.328315789473684e-05,
      "loss": 0.424,
      "step": 45543
    },
    {
      "epoch": 2.550341583603987,
      "grad_norm": 1.2958213090896606,
      "learning_rate": 9.32828947368421e-05,
      "loss": 0.3746,
      "step": 45544
    },
    {
      "epoch": 2.550397580916116,
      "grad_norm": 1.3364577293395996,
      "learning_rate": 9.328263157894738e-05,
      "loss": 0.4446,
      "step": 45545
    },
    {
      "epoch": 2.550453578228245,
      "grad_norm": 0.9293643832206726,
      "learning_rate": 9.328236842105264e-05,
      "loss": 0.4455,
      "step": 45546
    },
    {
      "epoch": 2.550509575540374,
      "grad_norm": 1.1867369413375854,
      "learning_rate": 9.328210526315791e-05,
      "loss": 0.5685,
      "step": 45547
    },
    {
      "epoch": 2.550565572852503,
      "grad_norm": 1.7388741970062256,
      "learning_rate": 9.328184210526316e-05,
      "loss": 0.4748,
      "step": 45548
    },
    {
      "epoch": 2.550621570164632,
      "grad_norm": 1.1298727989196777,
      "learning_rate": 9.328157894736843e-05,
      "loss": 0.369,
      "step": 45549
    },
    {
      "epoch": 2.550677567476761,
      "grad_norm": 1.265620231628418,
      "learning_rate": 9.328131578947369e-05,
      "loss": 0.3635,
      "step": 45550
    },
    {
      "epoch": 2.55073356478889,
      "grad_norm": 1.1824848651885986,
      "learning_rate": 9.328105263157895e-05,
      "loss": 0.5387,
      "step": 45551
    },
    {
      "epoch": 2.550789562101019,
      "grad_norm": 1.5040555000305176,
      "learning_rate": 9.328078947368422e-05,
      "loss": 0.591,
      "step": 45552
    },
    {
      "epoch": 2.550845559413148,
      "grad_norm": 0.9612807035446167,
      "learning_rate": 9.328052631578947e-05,
      "loss": 0.3174,
      "step": 45553
    },
    {
      "epoch": 2.550901556725277,
      "grad_norm": 1.1754109859466553,
      "learning_rate": 9.328026315789474e-05,
      "loss": 0.4166,
      "step": 45554
    },
    {
      "epoch": 2.550957554037406,
      "grad_norm": 1.1892164945602417,
      "learning_rate": 9.328e-05,
      "loss": 0.5019,
      "step": 45555
    },
    {
      "epoch": 2.551013551349535,
      "grad_norm": 0.9745883345603943,
      "learning_rate": 9.327973684210527e-05,
      "loss": 0.4013,
      "step": 45556
    },
    {
      "epoch": 2.551069548661664,
      "grad_norm": 0.9784499406814575,
      "learning_rate": 9.327947368421053e-05,
      "loss": 0.3361,
      "step": 45557
    },
    {
      "epoch": 2.551125545973793,
      "grad_norm": 0.9853188991546631,
      "learning_rate": 9.32792105263158e-05,
      "loss": 0.333,
      "step": 45558
    },
    {
      "epoch": 2.551181543285922,
      "grad_norm": 1.1957056522369385,
      "learning_rate": 9.327894736842105e-05,
      "loss": 0.4235,
      "step": 45559
    },
    {
      "epoch": 2.551237540598051,
      "grad_norm": 1.2127811908721924,
      "learning_rate": 9.327868421052633e-05,
      "loss": 0.3969,
      "step": 45560
    },
    {
      "epoch": 2.55129353791018,
      "grad_norm": 1.0176527500152588,
      "learning_rate": 9.327842105263159e-05,
      "loss": 0.3465,
      "step": 45561
    },
    {
      "epoch": 2.5513495352223092,
      "grad_norm": 1.3686647415161133,
      "learning_rate": 9.327815789473685e-05,
      "loss": 0.4607,
      "step": 45562
    },
    {
      "epoch": 2.5514055325344382,
      "grad_norm": 1.4335973262786865,
      "learning_rate": 9.32778947368421e-05,
      "loss": 0.3478,
      "step": 45563
    },
    {
      "epoch": 2.5514615298465673,
      "grad_norm": 1.0871343612670898,
      "learning_rate": 9.327763157894738e-05,
      "loss": 0.3685,
      "step": 45564
    },
    {
      "epoch": 2.5515175271586963,
      "grad_norm": 1.2457151412963867,
      "learning_rate": 9.327736842105264e-05,
      "loss": 0.4463,
      "step": 45565
    },
    {
      "epoch": 2.5515735244708253,
      "grad_norm": 1.2166354656219482,
      "learning_rate": 9.32771052631579e-05,
      "loss": 0.5207,
      "step": 45566
    },
    {
      "epoch": 2.5516295217829543,
      "grad_norm": 1.0356742143630981,
      "learning_rate": 9.327684210526316e-05,
      "loss": 0.3431,
      "step": 45567
    },
    {
      "epoch": 2.5516855190950833,
      "grad_norm": 1.560911774635315,
      "learning_rate": 9.327657894736842e-05,
      "loss": 0.529,
      "step": 45568
    },
    {
      "epoch": 2.5517415164072124,
      "grad_norm": 1.2377440929412842,
      "learning_rate": 9.327631578947369e-05,
      "loss": 0.4086,
      "step": 45569
    },
    {
      "epoch": 2.5517975137193414,
      "grad_norm": 0.9890557527542114,
      "learning_rate": 9.327605263157895e-05,
      "loss": 0.3385,
      "step": 45570
    },
    {
      "epoch": 2.5518535110314704,
      "grad_norm": 1.3722487688064575,
      "learning_rate": 9.327578947368421e-05,
      "loss": 0.4394,
      "step": 45571
    },
    {
      "epoch": 2.5519095083435994,
      "grad_norm": 1.1656464338302612,
      "learning_rate": 9.327552631578947e-05,
      "loss": 0.4531,
      "step": 45572
    },
    {
      "epoch": 2.5519655056557284,
      "grad_norm": 1.2346620559692383,
      "learning_rate": 9.327526315789474e-05,
      "loss": 0.436,
      "step": 45573
    },
    {
      "epoch": 2.5520215029678575,
      "grad_norm": 1.304569125175476,
      "learning_rate": 9.3275e-05,
      "loss": 0.3939,
      "step": 45574
    },
    {
      "epoch": 2.5520775002799865,
      "grad_norm": 1.3827192783355713,
      "learning_rate": 9.327473684210526e-05,
      "loss": 0.393,
      "step": 45575
    },
    {
      "epoch": 2.5521334975921155,
      "grad_norm": 1.0160528421401978,
      "learning_rate": 9.327447368421052e-05,
      "loss": 0.3062,
      "step": 45576
    },
    {
      "epoch": 2.5521894949042445,
      "grad_norm": 1.3612617254257202,
      "learning_rate": 9.32742105263158e-05,
      "loss": 0.4366,
      "step": 45577
    },
    {
      "epoch": 2.5522454922163735,
      "grad_norm": 1.078266978263855,
      "learning_rate": 9.327394736842106e-05,
      "loss": 0.4817,
      "step": 45578
    },
    {
      "epoch": 2.5523014895285026,
      "grad_norm": 1.2146620750427246,
      "learning_rate": 9.327368421052633e-05,
      "loss": 0.3633,
      "step": 45579
    },
    {
      "epoch": 2.5523574868406316,
      "grad_norm": 1.4247937202453613,
      "learning_rate": 9.327342105263158e-05,
      "loss": 0.5016,
      "step": 45580
    },
    {
      "epoch": 2.5524134841527606,
      "grad_norm": 1.3972868919372559,
      "learning_rate": 9.327315789473685e-05,
      "loss": 0.4021,
      "step": 45581
    },
    {
      "epoch": 2.5524694814648896,
      "grad_norm": 1.277218222618103,
      "learning_rate": 9.327289473684211e-05,
      "loss": 0.4911,
      "step": 45582
    },
    {
      "epoch": 2.5525254787770186,
      "grad_norm": 1.1201694011688232,
      "learning_rate": 9.327263157894738e-05,
      "loss": 0.4254,
      "step": 45583
    },
    {
      "epoch": 2.5525814760891476,
      "grad_norm": 1.1593981981277466,
      "learning_rate": 9.327236842105264e-05,
      "loss": 0.4175,
      "step": 45584
    },
    {
      "epoch": 2.5526374734012767,
      "grad_norm": 1.2248313426971436,
      "learning_rate": 9.327210526315789e-05,
      "loss": 0.4873,
      "step": 45585
    },
    {
      "epoch": 2.5526934707134057,
      "grad_norm": 0.9427081942558289,
      "learning_rate": 9.327184210526316e-05,
      "loss": 0.2833,
      "step": 45586
    },
    {
      "epoch": 2.5527494680255347,
      "grad_norm": 1.0200705528259277,
      "learning_rate": 9.327157894736842e-05,
      "loss": 0.3746,
      "step": 45587
    },
    {
      "epoch": 2.5528054653376637,
      "grad_norm": 1.0756677389144897,
      "learning_rate": 9.32713157894737e-05,
      "loss": 0.3722,
      "step": 45588
    },
    {
      "epoch": 2.5528614626497927,
      "grad_norm": 1.0629841089248657,
      "learning_rate": 9.327105263157895e-05,
      "loss": 0.4169,
      "step": 45589
    },
    {
      "epoch": 2.5529174599619218,
      "grad_norm": 1.1962857246398926,
      "learning_rate": 9.327078947368421e-05,
      "loss": 0.4261,
      "step": 45590
    },
    {
      "epoch": 2.552973457274051,
      "grad_norm": 1.4027111530303955,
      "learning_rate": 9.327052631578947e-05,
      "loss": 0.5373,
      "step": 45591
    },
    {
      "epoch": 2.55302945458618,
      "grad_norm": 1.379607915878296,
      "learning_rate": 9.327026315789475e-05,
      "loss": 0.4226,
      "step": 45592
    },
    {
      "epoch": 2.553085451898309,
      "grad_norm": 1.4633805751800537,
      "learning_rate": 9.327e-05,
      "loss": 0.4066,
      "step": 45593
    },
    {
      "epoch": 2.553141449210438,
      "grad_norm": 1.0058832168579102,
      "learning_rate": 9.326973684210527e-05,
      "loss": 0.3139,
      "step": 45594
    },
    {
      "epoch": 2.553197446522567,
      "grad_norm": 1.4055863618850708,
      "learning_rate": 9.326947368421053e-05,
      "loss": 0.4012,
      "step": 45595
    },
    {
      "epoch": 2.553253443834696,
      "grad_norm": 0.9868448376655579,
      "learning_rate": 9.32692105263158e-05,
      "loss": 0.3782,
      "step": 45596
    },
    {
      "epoch": 2.553309441146825,
      "grad_norm": 1.2650110721588135,
      "learning_rate": 9.326894736842106e-05,
      "loss": 0.5607,
      "step": 45597
    },
    {
      "epoch": 2.553365438458954,
      "grad_norm": 1.39922034740448,
      "learning_rate": 9.326868421052632e-05,
      "loss": 0.437,
      "step": 45598
    },
    {
      "epoch": 2.553421435771083,
      "grad_norm": 1.281574010848999,
      "learning_rate": 9.326842105263158e-05,
      "loss": 0.348,
      "step": 45599
    },
    {
      "epoch": 2.553477433083212,
      "grad_norm": 1.137755274772644,
      "learning_rate": 9.326815789473685e-05,
      "loss": 0.461,
      "step": 45600
    },
    {
      "epoch": 2.553533430395341,
      "grad_norm": 1.351910948753357,
      "learning_rate": 9.326789473684211e-05,
      "loss": 0.4383,
      "step": 45601
    },
    {
      "epoch": 2.55358942770747,
      "grad_norm": 1.5636895895004272,
      "learning_rate": 9.326763157894737e-05,
      "loss": 0.4454,
      "step": 45602
    },
    {
      "epoch": 2.553645425019599,
      "grad_norm": 1.0131797790527344,
      "learning_rate": 9.326736842105263e-05,
      "loss": 0.3249,
      "step": 45603
    },
    {
      "epoch": 2.553701422331728,
      "grad_norm": 1.34159255027771,
      "learning_rate": 9.326710526315789e-05,
      "loss": 0.5007,
      "step": 45604
    },
    {
      "epoch": 2.553757419643857,
      "grad_norm": 1.2082847356796265,
      "learning_rate": 9.326684210526316e-05,
      "loss": 0.4757,
      "step": 45605
    },
    {
      "epoch": 2.553813416955986,
      "grad_norm": 1.1693072319030762,
      "learning_rate": 9.326657894736842e-05,
      "loss": 0.312,
      "step": 45606
    },
    {
      "epoch": 2.553869414268115,
      "grad_norm": 1.2204186916351318,
      "learning_rate": 9.32663157894737e-05,
      "loss": 0.3884,
      "step": 45607
    },
    {
      "epoch": 2.553925411580244,
      "grad_norm": 2.8457987308502197,
      "learning_rate": 9.326605263157894e-05,
      "loss": 0.3027,
      "step": 45608
    },
    {
      "epoch": 2.553981408892373,
      "grad_norm": 1.3017693758010864,
      "learning_rate": 9.326578947368422e-05,
      "loss": 0.3121,
      "step": 45609
    },
    {
      "epoch": 2.554037406204502,
      "grad_norm": 1.1814875602722168,
      "learning_rate": 9.326552631578948e-05,
      "loss": 0.392,
      "step": 45610
    },
    {
      "epoch": 2.554093403516631,
      "grad_norm": 0.9689959287643433,
      "learning_rate": 9.326526315789475e-05,
      "loss": 0.3155,
      "step": 45611
    },
    {
      "epoch": 2.55414940082876,
      "grad_norm": 1.1015560626983643,
      "learning_rate": 9.326500000000001e-05,
      "loss": 0.5243,
      "step": 45612
    },
    {
      "epoch": 2.554205398140889,
      "grad_norm": 1.238231897354126,
      "learning_rate": 9.326473684210527e-05,
      "loss": 0.4462,
      "step": 45613
    },
    {
      "epoch": 2.5542613954530182,
      "grad_norm": 1.0320967435836792,
      "learning_rate": 9.326447368421053e-05,
      "loss": 0.4327,
      "step": 45614
    },
    {
      "epoch": 2.5543173927651472,
      "grad_norm": 1.3268393278121948,
      "learning_rate": 9.32642105263158e-05,
      "loss": 0.3836,
      "step": 45615
    },
    {
      "epoch": 2.5543733900772763,
      "grad_norm": 1.2125952243804932,
      "learning_rate": 9.326394736842106e-05,
      "loss": 0.3865,
      "step": 45616
    },
    {
      "epoch": 2.5544293873894053,
      "grad_norm": 1.5733534097671509,
      "learning_rate": 9.326368421052632e-05,
      "loss": 0.6021,
      "step": 45617
    },
    {
      "epoch": 2.5544853847015343,
      "grad_norm": 1.2419317960739136,
      "learning_rate": 9.326342105263158e-05,
      "loss": 0.4697,
      "step": 45618
    },
    {
      "epoch": 2.5545413820136633,
      "grad_norm": 1.016759991645813,
      "learning_rate": 9.326315789473684e-05,
      "loss": 0.2857,
      "step": 45619
    },
    {
      "epoch": 2.5545973793257923,
      "grad_norm": 1.1140037775039673,
      "learning_rate": 9.326289473684211e-05,
      "loss": 0.3096,
      "step": 45620
    },
    {
      "epoch": 2.5546533766379214,
      "grad_norm": 1.2843455076217651,
      "learning_rate": 9.326263157894737e-05,
      "loss": 0.3755,
      "step": 45621
    },
    {
      "epoch": 2.5547093739500504,
      "grad_norm": 1.187127947807312,
      "learning_rate": 9.326236842105263e-05,
      "loss": 0.4097,
      "step": 45622
    },
    {
      "epoch": 2.5547653712621794,
      "grad_norm": 1.1217719316482544,
      "learning_rate": 9.32621052631579e-05,
      "loss": 0.39,
      "step": 45623
    },
    {
      "epoch": 2.5548213685743084,
      "grad_norm": 1.2366663217544556,
      "learning_rate": 9.326184210526317e-05,
      "loss": 0.4215,
      "step": 45624
    },
    {
      "epoch": 2.5548773658864374,
      "grad_norm": 1.2192249298095703,
      "learning_rate": 9.326157894736843e-05,
      "loss": 0.4296,
      "step": 45625
    },
    {
      "epoch": 2.5549333631985665,
      "grad_norm": 1.794351577758789,
      "learning_rate": 9.326131578947369e-05,
      "loss": 0.6779,
      "step": 45626
    },
    {
      "epoch": 2.5549893605106955,
      "grad_norm": 1.2011408805847168,
      "learning_rate": 9.326105263157895e-05,
      "loss": 0.3865,
      "step": 45627
    },
    {
      "epoch": 2.5550453578228245,
      "grad_norm": 1.2304136753082275,
      "learning_rate": 9.326078947368422e-05,
      "loss": 0.492,
      "step": 45628
    },
    {
      "epoch": 2.5551013551349535,
      "grad_norm": 1.0056356191635132,
      "learning_rate": 9.326052631578948e-05,
      "loss": 0.3073,
      "step": 45629
    },
    {
      "epoch": 2.5551573524470825,
      "grad_norm": 1.4691132307052612,
      "learning_rate": 9.326026315789474e-05,
      "loss": 0.4013,
      "step": 45630
    },
    {
      "epoch": 2.5552133497592115,
      "grad_norm": 1.1730116605758667,
      "learning_rate": 9.326e-05,
      "loss": 0.589,
      "step": 45631
    },
    {
      "epoch": 2.5552693470713406,
      "grad_norm": 1.1387131214141846,
      "learning_rate": 9.325973684210527e-05,
      "loss": 0.3943,
      "step": 45632
    },
    {
      "epoch": 2.5553253443834696,
      "grad_norm": 1.13905668258667,
      "learning_rate": 9.325947368421053e-05,
      "loss": 0.346,
      "step": 45633
    },
    {
      "epoch": 2.5553813416955986,
      "grad_norm": 1.4302362203598022,
      "learning_rate": 9.32592105263158e-05,
      "loss": 0.4856,
      "step": 45634
    },
    {
      "epoch": 2.5554373390077276,
      "grad_norm": 1.1664444208145142,
      "learning_rate": 9.325894736842105e-05,
      "loss": 0.2877,
      "step": 45635
    },
    {
      "epoch": 2.5554933363198566,
      "grad_norm": 1.1313159465789795,
      "learning_rate": 9.325868421052631e-05,
      "loss": 0.3581,
      "step": 45636
    },
    {
      "epoch": 2.5555493336319857,
      "grad_norm": 1.4714298248291016,
      "learning_rate": 9.325842105263158e-05,
      "loss": 0.4252,
      "step": 45637
    },
    {
      "epoch": 2.5556053309441147,
      "grad_norm": 1.4931342601776123,
      "learning_rate": 9.325815789473684e-05,
      "loss": 0.4858,
      "step": 45638
    },
    {
      "epoch": 2.5556613282562437,
      "grad_norm": 1.1312881708145142,
      "learning_rate": 9.325789473684212e-05,
      "loss": 0.3891,
      "step": 45639
    },
    {
      "epoch": 2.5557173255683727,
      "grad_norm": 1.3210203647613525,
      "learning_rate": 9.325763157894736e-05,
      "loss": 0.4444,
      "step": 45640
    },
    {
      "epoch": 2.5557733228805017,
      "grad_norm": 1.2131695747375488,
      "learning_rate": 9.325736842105264e-05,
      "loss": 0.3467,
      "step": 45641
    },
    {
      "epoch": 2.5558293201926308,
      "grad_norm": 1.1313854455947876,
      "learning_rate": 9.32571052631579e-05,
      "loss": 0.458,
      "step": 45642
    },
    {
      "epoch": 2.55588531750476,
      "grad_norm": 1.3144233226776123,
      "learning_rate": 9.325684210526317e-05,
      "loss": 0.3914,
      "step": 45643
    },
    {
      "epoch": 2.555941314816889,
      "grad_norm": 1.18085515499115,
      "learning_rate": 9.325657894736843e-05,
      "loss": 0.3232,
      "step": 45644
    },
    {
      "epoch": 2.555997312129018,
      "grad_norm": 1.4017281532287598,
      "learning_rate": 9.325631578947369e-05,
      "loss": 0.5163,
      "step": 45645
    },
    {
      "epoch": 2.556053309441147,
      "grad_norm": 1.2561012506484985,
      "learning_rate": 9.325605263157895e-05,
      "loss": 0.4668,
      "step": 45646
    },
    {
      "epoch": 2.556109306753276,
      "grad_norm": 1.210850715637207,
      "learning_rate": 9.325578947368422e-05,
      "loss": 0.3906,
      "step": 45647
    },
    {
      "epoch": 2.556165304065405,
      "grad_norm": 1.1935440301895142,
      "learning_rate": 9.325552631578948e-05,
      "loss": 0.3574,
      "step": 45648
    },
    {
      "epoch": 2.556221301377534,
      "grad_norm": 1.3698830604553223,
      "learning_rate": 9.325526315789474e-05,
      "loss": 0.3904,
      "step": 45649
    },
    {
      "epoch": 2.556277298689663,
      "grad_norm": 1.1001482009887695,
      "learning_rate": 9.3255e-05,
      "loss": 0.3224,
      "step": 45650
    },
    {
      "epoch": 2.556333296001792,
      "grad_norm": 1.2250148057937622,
      "learning_rate": 9.325473684210527e-05,
      "loss": 0.3642,
      "step": 45651
    },
    {
      "epoch": 2.556389293313921,
      "grad_norm": 1.1564087867736816,
      "learning_rate": 9.325447368421053e-05,
      "loss": 0.5367,
      "step": 45652
    },
    {
      "epoch": 2.55644529062605,
      "grad_norm": 1.075554370880127,
      "learning_rate": 9.32542105263158e-05,
      "loss": 0.3751,
      "step": 45653
    },
    {
      "epoch": 2.556501287938179,
      "grad_norm": 1.1332156658172607,
      "learning_rate": 9.325394736842105e-05,
      "loss": 0.5071,
      "step": 45654
    },
    {
      "epoch": 2.556557285250308,
      "grad_norm": 1.383529543876648,
      "learning_rate": 9.325368421052631e-05,
      "loss": 0.5627,
      "step": 45655
    },
    {
      "epoch": 2.556613282562437,
      "grad_norm": 1.28536057472229,
      "learning_rate": 9.325342105263159e-05,
      "loss": 0.2972,
      "step": 45656
    },
    {
      "epoch": 2.556669279874566,
      "grad_norm": 1.1395752429962158,
      "learning_rate": 9.325315789473685e-05,
      "loss": 0.4714,
      "step": 45657
    },
    {
      "epoch": 2.556725277186695,
      "grad_norm": 1.3974889516830444,
      "learning_rate": 9.32528947368421e-05,
      "loss": 0.3769,
      "step": 45658
    },
    {
      "epoch": 2.556781274498824,
      "grad_norm": 0.9640372395515442,
      "learning_rate": 9.325263157894737e-05,
      "loss": 0.3609,
      "step": 45659
    },
    {
      "epoch": 2.556837271810953,
      "grad_norm": 1.2757364511489868,
      "learning_rate": 9.325236842105264e-05,
      "loss": 0.4384,
      "step": 45660
    },
    {
      "epoch": 2.556893269123082,
      "grad_norm": 1.1255967617034912,
      "learning_rate": 9.32521052631579e-05,
      "loss": 0.3759,
      "step": 45661
    },
    {
      "epoch": 2.556949266435211,
      "grad_norm": 1.0359216928482056,
      "learning_rate": 9.325184210526317e-05,
      "loss": 0.3848,
      "step": 45662
    },
    {
      "epoch": 2.55700526374734,
      "grad_norm": 1.2480062246322632,
      "learning_rate": 9.325157894736842e-05,
      "loss": 0.3645,
      "step": 45663
    },
    {
      "epoch": 2.557061261059469,
      "grad_norm": 1.1469781398773193,
      "learning_rate": 9.325131578947369e-05,
      "loss": 0.4041,
      "step": 45664
    },
    {
      "epoch": 2.557117258371598,
      "grad_norm": 1.367587685585022,
      "learning_rate": 9.325105263157895e-05,
      "loss": 0.3825,
      "step": 45665
    },
    {
      "epoch": 2.557173255683727,
      "grad_norm": 1.728960394859314,
      "learning_rate": 9.325078947368422e-05,
      "loss": 0.4866,
      "step": 45666
    },
    {
      "epoch": 2.5572292529958562,
      "grad_norm": 1.1842454671859741,
      "learning_rate": 9.325052631578947e-05,
      "loss": 0.3705,
      "step": 45667
    },
    {
      "epoch": 2.5572852503079853,
      "grad_norm": 1.195404052734375,
      "learning_rate": 9.325026315789474e-05,
      "loss": 0.3402,
      "step": 45668
    },
    {
      "epoch": 2.5573412476201143,
      "grad_norm": 1.595546841621399,
      "learning_rate": 9.325e-05,
      "loss": 0.3296,
      "step": 45669
    },
    {
      "epoch": 2.5573972449322433,
      "grad_norm": 1.2033277750015259,
      "learning_rate": 9.324973684210528e-05,
      "loss": 0.384,
      "step": 45670
    },
    {
      "epoch": 2.5574532422443723,
      "grad_norm": 3.279566526412964,
      "learning_rate": 9.324947368421054e-05,
      "loss": 0.3414,
      "step": 45671
    },
    {
      "epoch": 2.5575092395565013,
      "grad_norm": 1.2063119411468506,
      "learning_rate": 9.324921052631578e-05,
      "loss": 0.3675,
      "step": 45672
    },
    {
      "epoch": 2.5575652368686304,
      "grad_norm": 1.162808895111084,
      "learning_rate": 9.324894736842106e-05,
      "loss": 0.4085,
      "step": 45673
    },
    {
      "epoch": 2.5576212341807594,
      "grad_norm": 1.3233859539031982,
      "learning_rate": 9.324868421052632e-05,
      "loss": 0.3672,
      "step": 45674
    },
    {
      "epoch": 2.5576772314928884,
      "grad_norm": 1.0006930828094482,
      "learning_rate": 9.324842105263159e-05,
      "loss": 0.3662,
      "step": 45675
    },
    {
      "epoch": 2.5577332288050174,
      "grad_norm": 1.1017684936523438,
      "learning_rate": 9.324815789473685e-05,
      "loss": 0.3555,
      "step": 45676
    },
    {
      "epoch": 2.5577892261171464,
      "grad_norm": 1.2516436576843262,
      "learning_rate": 9.324789473684211e-05,
      "loss": 0.5026,
      "step": 45677
    },
    {
      "epoch": 2.5578452234292754,
      "grad_norm": 1.251975655555725,
      "learning_rate": 9.324763157894737e-05,
      "loss": 0.4134,
      "step": 45678
    },
    {
      "epoch": 2.5579012207414045,
      "grad_norm": 0.9582602977752686,
      "learning_rate": 9.324736842105264e-05,
      "loss": 0.3548,
      "step": 45679
    },
    {
      "epoch": 2.5579572180535335,
      "grad_norm": 1.4859552383422852,
      "learning_rate": 9.32471052631579e-05,
      "loss": 0.5554,
      "step": 45680
    },
    {
      "epoch": 2.5580132153656625,
      "grad_norm": 1.015962839126587,
      "learning_rate": 9.324684210526316e-05,
      "loss": 0.4235,
      "step": 45681
    },
    {
      "epoch": 2.5580692126777915,
      "grad_norm": 1.1603929996490479,
      "learning_rate": 9.324657894736842e-05,
      "loss": 0.3881,
      "step": 45682
    },
    {
      "epoch": 2.5581252099899205,
      "grad_norm": 1.2697588205337524,
      "learning_rate": 9.32463157894737e-05,
      "loss": 0.5235,
      "step": 45683
    },
    {
      "epoch": 2.5581812073020496,
      "grad_norm": 1.1522265672683716,
      "learning_rate": 9.324605263157895e-05,
      "loss": 0.3555,
      "step": 45684
    },
    {
      "epoch": 2.5582372046141786,
      "grad_norm": 1.1480109691619873,
      "learning_rate": 9.324578947368421e-05,
      "loss": 0.335,
      "step": 45685
    },
    {
      "epoch": 2.5582932019263076,
      "grad_norm": 1.244274616241455,
      "learning_rate": 9.324552631578947e-05,
      "loss": 0.3773,
      "step": 45686
    },
    {
      "epoch": 2.5583491992384366,
      "grad_norm": 1.1693480014801025,
      "learning_rate": 9.324526315789475e-05,
      "loss": 0.6179,
      "step": 45687
    },
    {
      "epoch": 2.5584051965505656,
      "grad_norm": 1.3929036855697632,
      "learning_rate": 9.3245e-05,
      "loss": 0.3648,
      "step": 45688
    },
    {
      "epoch": 2.5584611938626947,
      "grad_norm": 1.0592700242996216,
      "learning_rate": 9.324473684210527e-05,
      "loss": 0.4285,
      "step": 45689
    },
    {
      "epoch": 2.5585171911748237,
      "grad_norm": 1.013587474822998,
      "learning_rate": 9.324447368421053e-05,
      "loss": 0.4833,
      "step": 45690
    },
    {
      "epoch": 2.5585731884869527,
      "grad_norm": 1.4980398416519165,
      "learning_rate": 9.324421052631579e-05,
      "loss": 0.3609,
      "step": 45691
    },
    {
      "epoch": 2.5586291857990817,
      "grad_norm": 1.169128656387329,
      "learning_rate": 9.324394736842106e-05,
      "loss": 0.3945,
      "step": 45692
    },
    {
      "epoch": 2.5586851831112107,
      "grad_norm": 1.200084924697876,
      "learning_rate": 9.324368421052632e-05,
      "loss": 0.4536,
      "step": 45693
    },
    {
      "epoch": 2.5587411804233398,
      "grad_norm": 1.5824177265167236,
      "learning_rate": 9.324342105263159e-05,
      "loss": 0.4062,
      "step": 45694
    },
    {
      "epoch": 2.5587971777354688,
      "grad_norm": 0.9825482964515686,
      "learning_rate": 9.324315789473684e-05,
      "loss": 0.2983,
      "step": 45695
    },
    {
      "epoch": 2.558853175047598,
      "grad_norm": 1.2140988111495972,
      "learning_rate": 9.324289473684211e-05,
      "loss": 0.3609,
      "step": 45696
    },
    {
      "epoch": 2.558909172359727,
      "grad_norm": 0.875757098197937,
      "learning_rate": 9.324263157894737e-05,
      "loss": 0.268,
      "step": 45697
    },
    {
      "epoch": 2.558965169671856,
      "grad_norm": 1.0031931400299072,
      "learning_rate": 9.324236842105264e-05,
      "loss": 0.338,
      "step": 45698
    },
    {
      "epoch": 2.559021166983985,
      "grad_norm": 1.2444010972976685,
      "learning_rate": 9.32421052631579e-05,
      "loss": 0.4075,
      "step": 45699
    },
    {
      "epoch": 2.559077164296114,
      "grad_norm": 1.4407545328140259,
      "learning_rate": 9.324184210526316e-05,
      "loss": 0.3802,
      "step": 45700
    },
    {
      "epoch": 2.559133161608243,
      "grad_norm": 0.972081184387207,
      "learning_rate": 9.324157894736842e-05,
      "loss": 0.2922,
      "step": 45701
    },
    {
      "epoch": 2.559189158920372,
      "grad_norm": 1.0640215873718262,
      "learning_rate": 9.32413157894737e-05,
      "loss": 0.3579,
      "step": 45702
    },
    {
      "epoch": 2.559245156232501,
      "grad_norm": 1.2619779109954834,
      "learning_rate": 9.324105263157896e-05,
      "loss": 0.4187,
      "step": 45703
    },
    {
      "epoch": 2.55930115354463,
      "grad_norm": 1.8751230239868164,
      "learning_rate": 9.324078947368422e-05,
      "loss": 0.4124,
      "step": 45704
    },
    {
      "epoch": 2.559357150856759,
      "grad_norm": 1.206122636795044,
      "learning_rate": 9.324052631578948e-05,
      "loss": 0.4839,
      "step": 45705
    },
    {
      "epoch": 2.559413148168888,
      "grad_norm": 1.3412203788757324,
      "learning_rate": 9.324026315789474e-05,
      "loss": 0.5816,
      "step": 45706
    },
    {
      "epoch": 2.559469145481017,
      "grad_norm": 1.2568408250808716,
      "learning_rate": 9.324000000000001e-05,
      "loss": 0.506,
      "step": 45707
    },
    {
      "epoch": 2.559525142793146,
      "grad_norm": 1.2885011434555054,
      "learning_rate": 9.323973684210527e-05,
      "loss": 0.4702,
      "step": 45708
    },
    {
      "epoch": 2.559581140105275,
      "grad_norm": 1.052280306816101,
      "learning_rate": 9.323947368421053e-05,
      "loss": 0.4138,
      "step": 45709
    },
    {
      "epoch": 2.559637137417404,
      "grad_norm": 1.1961079835891724,
      "learning_rate": 9.323921052631579e-05,
      "loss": 0.4005,
      "step": 45710
    },
    {
      "epoch": 2.559693134729533,
      "grad_norm": 1.0916801691055298,
      "learning_rate": 9.323894736842106e-05,
      "loss": 0.3366,
      "step": 45711
    },
    {
      "epoch": 2.559749132041662,
      "grad_norm": 1.1278434991836548,
      "learning_rate": 9.323868421052632e-05,
      "loss": 0.4107,
      "step": 45712
    },
    {
      "epoch": 2.559805129353791,
      "grad_norm": 2.5425703525543213,
      "learning_rate": 9.323842105263158e-05,
      "loss": 0.3212,
      "step": 45713
    },
    {
      "epoch": 2.55986112666592,
      "grad_norm": 1.120042085647583,
      "learning_rate": 9.323815789473684e-05,
      "loss": 0.425,
      "step": 45714
    },
    {
      "epoch": 2.559917123978049,
      "grad_norm": 1.0947351455688477,
      "learning_rate": 9.323789473684211e-05,
      "loss": 0.3601,
      "step": 45715
    },
    {
      "epoch": 2.559973121290178,
      "grad_norm": 1.2361760139465332,
      "learning_rate": 9.323763157894737e-05,
      "loss": 0.4042,
      "step": 45716
    },
    {
      "epoch": 2.560029118602307,
      "grad_norm": 1.1049026250839233,
      "learning_rate": 9.323736842105265e-05,
      "loss": 0.4318,
      "step": 45717
    },
    {
      "epoch": 2.560085115914436,
      "grad_norm": 1.195594310760498,
      "learning_rate": 9.323710526315789e-05,
      "loss": 0.3597,
      "step": 45718
    },
    {
      "epoch": 2.5601411132265652,
      "grad_norm": 1.2355653047561646,
      "learning_rate": 9.323684210526317e-05,
      "loss": 0.3344,
      "step": 45719
    },
    {
      "epoch": 2.5601971105386943,
      "grad_norm": 1.0339802503585815,
      "learning_rate": 9.323657894736843e-05,
      "loss": 0.3592,
      "step": 45720
    },
    {
      "epoch": 2.5602531078508233,
      "grad_norm": 1.7107774019241333,
      "learning_rate": 9.32363157894737e-05,
      "loss": 0.4588,
      "step": 45721
    },
    {
      "epoch": 2.5603091051629523,
      "grad_norm": 1.2407349348068237,
      "learning_rate": 9.323605263157895e-05,
      "loss": 0.3533,
      "step": 45722
    },
    {
      "epoch": 2.5603651024750813,
      "grad_norm": 1.1276711225509644,
      "learning_rate": 9.32357894736842e-05,
      "loss": 0.3698,
      "step": 45723
    },
    {
      "epoch": 2.5604210997872103,
      "grad_norm": 1.4450361728668213,
      "learning_rate": 9.323552631578948e-05,
      "loss": 0.4524,
      "step": 45724
    },
    {
      "epoch": 2.5604770970993393,
      "grad_norm": 1.1825019121170044,
      "learning_rate": 9.323526315789474e-05,
      "loss": 0.5107,
      "step": 45725
    },
    {
      "epoch": 2.5605330944114684,
      "grad_norm": 1.134027123451233,
      "learning_rate": 9.323500000000001e-05,
      "loss": 0.417,
      "step": 45726
    },
    {
      "epoch": 2.5605890917235974,
      "grad_norm": 1.032018780708313,
      "learning_rate": 9.323473684210526e-05,
      "loss": 0.3408,
      "step": 45727
    },
    {
      "epoch": 2.5606450890357264,
      "grad_norm": 1.6317919492721558,
      "learning_rate": 9.323447368421053e-05,
      "loss": 0.5422,
      "step": 45728
    },
    {
      "epoch": 2.5607010863478554,
      "grad_norm": 1.0493611097335815,
      "learning_rate": 9.323421052631579e-05,
      "loss": 0.3266,
      "step": 45729
    },
    {
      "epoch": 2.5607570836599844,
      "grad_norm": 1.3089109659194946,
      "learning_rate": 9.323394736842106e-05,
      "loss": 0.4636,
      "step": 45730
    },
    {
      "epoch": 2.5608130809721135,
      "grad_norm": 1.0117363929748535,
      "learning_rate": 9.323368421052632e-05,
      "loss": 0.3684,
      "step": 45731
    },
    {
      "epoch": 2.5608690782842425,
      "grad_norm": 1.2079764604568481,
      "learning_rate": 9.323342105263158e-05,
      "loss": 0.2532,
      "step": 45732
    },
    {
      "epoch": 2.5609250755963715,
      "grad_norm": 1.0275832414627075,
      "learning_rate": 9.323315789473684e-05,
      "loss": 0.343,
      "step": 45733
    },
    {
      "epoch": 2.5609810729085005,
      "grad_norm": 1.2589458227157593,
      "learning_rate": 9.323289473684212e-05,
      "loss": 0.3578,
      "step": 45734
    },
    {
      "epoch": 2.5610370702206295,
      "grad_norm": 1.2149940729141235,
      "learning_rate": 9.323263157894738e-05,
      "loss": 0.3968,
      "step": 45735
    },
    {
      "epoch": 2.5610930675327586,
      "grad_norm": 1.1274718046188354,
      "learning_rate": 9.323236842105264e-05,
      "loss": 0.3794,
      "step": 45736
    },
    {
      "epoch": 2.5611490648448876,
      "grad_norm": 1.2266300916671753,
      "learning_rate": 9.32321052631579e-05,
      "loss": 0.4552,
      "step": 45737
    },
    {
      "epoch": 2.5612050621570166,
      "grad_norm": 0.9906600713729858,
      "learning_rate": 9.323184210526317e-05,
      "loss": 0.3175,
      "step": 45738
    },
    {
      "epoch": 2.5612610594691456,
      "grad_norm": 1.0905526876449585,
      "learning_rate": 9.323157894736843e-05,
      "loss": 0.4285,
      "step": 45739
    },
    {
      "epoch": 2.5613170567812746,
      "grad_norm": 0.9535433650016785,
      "learning_rate": 9.323131578947369e-05,
      "loss": 0.3932,
      "step": 45740
    },
    {
      "epoch": 2.5613730540934037,
      "grad_norm": 1.311492919921875,
      "learning_rate": 9.323105263157895e-05,
      "loss": 0.4295,
      "step": 45741
    },
    {
      "epoch": 2.5614290514055327,
      "grad_norm": 1.2669093608856201,
      "learning_rate": 9.323078947368421e-05,
      "loss": 0.4549,
      "step": 45742
    },
    {
      "epoch": 2.5614850487176617,
      "grad_norm": 1.3176826238632202,
      "learning_rate": 9.323052631578948e-05,
      "loss": 0.528,
      "step": 45743
    },
    {
      "epoch": 2.5615410460297907,
      "grad_norm": 1.1536734104156494,
      "learning_rate": 9.323026315789474e-05,
      "loss": 0.4261,
      "step": 45744
    },
    {
      "epoch": 2.5615970433419197,
      "grad_norm": 1.254226803779602,
      "learning_rate": 9.323e-05,
      "loss": 0.317,
      "step": 45745
    },
    {
      "epoch": 2.5616530406540488,
      "grad_norm": 1.5932658910751343,
      "learning_rate": 9.322973684210526e-05,
      "loss": 0.5025,
      "step": 45746
    },
    {
      "epoch": 2.5617090379661778,
      "grad_norm": 1.348116159439087,
      "learning_rate": 9.322947368421053e-05,
      "loss": 0.4817,
      "step": 45747
    },
    {
      "epoch": 2.561765035278307,
      "grad_norm": 1.0877364873886108,
      "learning_rate": 9.322921052631579e-05,
      "loss": 0.3022,
      "step": 45748
    },
    {
      "epoch": 2.561821032590436,
      "grad_norm": 1.4115625619888306,
      "learning_rate": 9.322894736842107e-05,
      "loss": 0.4522,
      "step": 45749
    },
    {
      "epoch": 2.561877029902565,
      "grad_norm": 1.363369345664978,
      "learning_rate": 9.322868421052631e-05,
      "loss": 0.4922,
      "step": 45750
    },
    {
      "epoch": 2.561933027214694,
      "grad_norm": 1.1387078762054443,
      "learning_rate": 9.322842105263159e-05,
      "loss": 0.3572,
      "step": 45751
    },
    {
      "epoch": 2.561989024526823,
      "grad_norm": 1.1272447109222412,
      "learning_rate": 9.322815789473685e-05,
      "loss": 0.3686,
      "step": 45752
    },
    {
      "epoch": 2.562045021838952,
      "grad_norm": 1.0637720823287964,
      "learning_rate": 9.322789473684212e-05,
      "loss": 0.4098,
      "step": 45753
    },
    {
      "epoch": 2.562101019151081,
      "grad_norm": 1.205729603767395,
      "learning_rate": 9.322763157894738e-05,
      "loss": 0.4362,
      "step": 45754
    },
    {
      "epoch": 2.56215701646321,
      "grad_norm": 1.383683681488037,
      "learning_rate": 9.322736842105264e-05,
      "loss": 0.3607,
      "step": 45755
    },
    {
      "epoch": 2.562213013775339,
      "grad_norm": 0.9735696315765381,
      "learning_rate": 9.32271052631579e-05,
      "loss": 0.2919,
      "step": 45756
    },
    {
      "epoch": 2.562269011087468,
      "grad_norm": 1.0302846431732178,
      "learning_rate": 9.322684210526316e-05,
      "loss": 0.3606,
      "step": 45757
    },
    {
      "epoch": 2.562325008399597,
      "grad_norm": 1.5440552234649658,
      "learning_rate": 9.322657894736843e-05,
      "loss": 0.5488,
      "step": 45758
    },
    {
      "epoch": 2.562381005711726,
      "grad_norm": 1.5616835355758667,
      "learning_rate": 9.322631578947369e-05,
      "loss": 0.449,
      "step": 45759
    },
    {
      "epoch": 2.562437003023855,
      "grad_norm": 1.2558298110961914,
      "learning_rate": 9.322605263157895e-05,
      "loss": 0.4412,
      "step": 45760
    },
    {
      "epoch": 2.562493000335984,
      "grad_norm": 1.0505138635635376,
      "learning_rate": 9.322578947368421e-05,
      "loss": 0.3367,
      "step": 45761
    },
    {
      "epoch": 2.562548997648113,
      "grad_norm": 1.183586597442627,
      "learning_rate": 9.322552631578948e-05,
      "loss": 0.3887,
      "step": 45762
    },
    {
      "epoch": 2.5626049949602416,
      "grad_norm": 1.3686237335205078,
      "learning_rate": 9.322526315789474e-05,
      "loss": 0.3549,
      "step": 45763
    },
    {
      "epoch": 2.562660992272371,
      "grad_norm": 1.3428999185562134,
      "learning_rate": 9.3225e-05,
      "loss": 0.4532,
      "step": 45764
    },
    {
      "epoch": 2.5627169895844997,
      "grad_norm": 1.0158320665359497,
      "learning_rate": 9.322473684210526e-05,
      "loss": 0.3652,
      "step": 45765
    },
    {
      "epoch": 2.562772986896629,
      "grad_norm": 1.281711459159851,
      "learning_rate": 9.322447368421054e-05,
      "loss": 0.3821,
      "step": 45766
    },
    {
      "epoch": 2.5628289842087577,
      "grad_norm": 1.1996203660964966,
      "learning_rate": 9.32242105263158e-05,
      "loss": 0.4204,
      "step": 45767
    },
    {
      "epoch": 2.562884981520887,
      "grad_norm": 1.2329365015029907,
      "learning_rate": 9.322394736842106e-05,
      "loss": 0.5543,
      "step": 45768
    },
    {
      "epoch": 2.5629409788330157,
      "grad_norm": 1.2657426595687866,
      "learning_rate": 9.322368421052631e-05,
      "loss": 0.4744,
      "step": 45769
    },
    {
      "epoch": 2.562996976145145,
      "grad_norm": 0.9845755696296692,
      "learning_rate": 9.322342105263159e-05,
      "loss": 0.3641,
      "step": 45770
    },
    {
      "epoch": 2.563052973457274,
      "grad_norm": 1.2035232782363892,
      "learning_rate": 9.322315789473685e-05,
      "loss": 0.4131,
      "step": 45771
    },
    {
      "epoch": 2.5631089707694032,
      "grad_norm": 1.303586483001709,
      "learning_rate": 9.322289473684212e-05,
      "loss": 0.3969,
      "step": 45772
    },
    {
      "epoch": 2.563164968081532,
      "grad_norm": 1.195119023323059,
      "learning_rate": 9.322263157894737e-05,
      "loss": 0.459,
      "step": 45773
    },
    {
      "epoch": 2.5632209653936613,
      "grad_norm": 1.2336194515228271,
      "learning_rate": 9.322236842105264e-05,
      "loss": 0.4525,
      "step": 45774
    },
    {
      "epoch": 2.56327696270579,
      "grad_norm": 1.088029384613037,
      "learning_rate": 9.32221052631579e-05,
      "loss": 0.3257,
      "step": 45775
    },
    {
      "epoch": 2.5633329600179193,
      "grad_norm": 1.2099227905273438,
      "learning_rate": 9.322184210526316e-05,
      "loss": 0.377,
      "step": 45776
    },
    {
      "epoch": 2.563388957330048,
      "grad_norm": 1.1198279857635498,
      "learning_rate": 9.322157894736842e-05,
      "loss": 0.3777,
      "step": 45777
    },
    {
      "epoch": 2.5634449546421774,
      "grad_norm": 1.1287174224853516,
      "learning_rate": 9.322131578947368e-05,
      "loss": 0.4526,
      "step": 45778
    },
    {
      "epoch": 2.563500951954306,
      "grad_norm": 1.0839780569076538,
      "learning_rate": 9.322105263157895e-05,
      "loss": 0.4133,
      "step": 45779
    },
    {
      "epoch": 2.5635569492664354,
      "grad_norm": 1.1413198709487915,
      "learning_rate": 9.322078947368421e-05,
      "loss": 0.3812,
      "step": 45780
    },
    {
      "epoch": 2.563612946578564,
      "grad_norm": 1.3592926263809204,
      "learning_rate": 9.322052631578949e-05,
      "loss": 0.4621,
      "step": 45781
    },
    {
      "epoch": 2.5636689438906934,
      "grad_norm": 1.391784429550171,
      "learning_rate": 9.322026315789473e-05,
      "loss": 0.503,
      "step": 45782
    },
    {
      "epoch": 2.563724941202822,
      "grad_norm": 2.1052563190460205,
      "learning_rate": 9.322e-05,
      "loss": 0.4295,
      "step": 45783
    },
    {
      "epoch": 2.5637809385149515,
      "grad_norm": 1.1694837808609009,
      "learning_rate": 9.321973684210527e-05,
      "loss": 0.4858,
      "step": 45784
    },
    {
      "epoch": 2.56383693582708,
      "grad_norm": 1.1886311769485474,
      "learning_rate": 9.321947368421054e-05,
      "loss": 0.3549,
      "step": 45785
    },
    {
      "epoch": 2.5638929331392095,
      "grad_norm": 1.321065902709961,
      "learning_rate": 9.32192105263158e-05,
      "loss": 0.422,
      "step": 45786
    },
    {
      "epoch": 2.563948930451338,
      "grad_norm": 15.169246673583984,
      "learning_rate": 9.321894736842106e-05,
      "loss": 0.3438,
      "step": 45787
    },
    {
      "epoch": 2.5640049277634676,
      "grad_norm": 1.134333610534668,
      "learning_rate": 9.321868421052632e-05,
      "loss": 0.4274,
      "step": 45788
    },
    {
      "epoch": 2.564060925075596,
      "grad_norm": 1.0968347787857056,
      "learning_rate": 9.321842105263159e-05,
      "loss": 0.3614,
      "step": 45789
    },
    {
      "epoch": 2.5641169223877256,
      "grad_norm": 1.3577218055725098,
      "learning_rate": 9.321815789473685e-05,
      "loss": 0.2972,
      "step": 45790
    },
    {
      "epoch": 2.564172919699854,
      "grad_norm": 1.853683352470398,
      "learning_rate": 9.321789473684211e-05,
      "loss": 0.4007,
      "step": 45791
    },
    {
      "epoch": 2.5642289170119836,
      "grad_norm": 0.9404269456863403,
      "learning_rate": 9.321763157894737e-05,
      "loss": 0.3209,
      "step": 45792
    },
    {
      "epoch": 2.564284914324112,
      "grad_norm": 1.1967439651489258,
      "learning_rate": 9.321736842105263e-05,
      "loss": 0.5329,
      "step": 45793
    },
    {
      "epoch": 2.5643409116362417,
      "grad_norm": 1.1776896715164185,
      "learning_rate": 9.32171052631579e-05,
      "loss": 0.3775,
      "step": 45794
    },
    {
      "epoch": 2.5643969089483702,
      "grad_norm": 1.7608062028884888,
      "learning_rate": 9.321684210526316e-05,
      "loss": 0.369,
      "step": 45795
    },
    {
      "epoch": 2.5644529062604997,
      "grad_norm": 1.0942223072052002,
      "learning_rate": 9.321657894736842e-05,
      "loss": 0.3538,
      "step": 45796
    },
    {
      "epoch": 2.5645089035726283,
      "grad_norm": 1.1120693683624268,
      "learning_rate": 9.321631578947368e-05,
      "loss": 0.4077,
      "step": 45797
    },
    {
      "epoch": 2.5645649008847577,
      "grad_norm": 1.82485032081604,
      "learning_rate": 9.321605263157896e-05,
      "loss": 0.4356,
      "step": 45798
    },
    {
      "epoch": 2.5646208981968863,
      "grad_norm": 1.3394153118133545,
      "learning_rate": 9.321578947368422e-05,
      "loss": 0.4355,
      "step": 45799
    },
    {
      "epoch": 2.564676895509016,
      "grad_norm": 1.3118377923965454,
      "learning_rate": 9.321552631578947e-05,
      "loss": 0.5102,
      "step": 45800
    },
    {
      "epoch": 2.5647328928211444,
      "grad_norm": 1.3070013523101807,
      "learning_rate": 9.321526315789473e-05,
      "loss": 0.3757,
      "step": 45801
    },
    {
      "epoch": 2.564788890133274,
      "grad_norm": 1.694876790046692,
      "learning_rate": 9.321500000000001e-05,
      "loss": 0.4263,
      "step": 45802
    },
    {
      "epoch": 2.5648448874454024,
      "grad_norm": 1.2137254476547241,
      "learning_rate": 9.321473684210527e-05,
      "loss": 0.5445,
      "step": 45803
    },
    {
      "epoch": 2.564900884757532,
      "grad_norm": 1.4015522003173828,
      "learning_rate": 9.321447368421054e-05,
      "loss": 0.4854,
      "step": 45804
    },
    {
      "epoch": 2.5649568820696604,
      "grad_norm": 1.118233561515808,
      "learning_rate": 9.321421052631579e-05,
      "loss": 0.3749,
      "step": 45805
    },
    {
      "epoch": 2.56501287938179,
      "grad_norm": 1.0573805570602417,
      "learning_rate": 9.321394736842106e-05,
      "loss": 0.3215,
      "step": 45806
    },
    {
      "epoch": 2.5650688766939185,
      "grad_norm": 1.412376046180725,
      "learning_rate": 9.321368421052632e-05,
      "loss": 0.3604,
      "step": 45807
    },
    {
      "epoch": 2.565124874006048,
      "grad_norm": 1.00432550907135,
      "learning_rate": 9.32134210526316e-05,
      "loss": 0.3239,
      "step": 45808
    },
    {
      "epoch": 2.5651808713181765,
      "grad_norm": 1.1237467527389526,
      "learning_rate": 9.321315789473685e-05,
      "loss": 0.3716,
      "step": 45809
    },
    {
      "epoch": 2.565236868630306,
      "grad_norm": 1.2478457689285278,
      "learning_rate": 9.32128947368421e-05,
      "loss": 0.4387,
      "step": 45810
    },
    {
      "epoch": 2.5652928659424346,
      "grad_norm": 1.244915246963501,
      "learning_rate": 9.321263157894737e-05,
      "loss": 0.356,
      "step": 45811
    },
    {
      "epoch": 2.565348863254564,
      "grad_norm": 1.1220879554748535,
      "learning_rate": 9.321236842105263e-05,
      "loss": 0.413,
      "step": 45812
    },
    {
      "epoch": 2.5654048605666926,
      "grad_norm": 1.0878968238830566,
      "learning_rate": 9.32121052631579e-05,
      "loss": 0.4896,
      "step": 45813
    },
    {
      "epoch": 2.565460857878822,
      "grad_norm": 1.2028474807739258,
      "learning_rate": 9.321184210526315e-05,
      "loss": 0.378,
      "step": 45814
    },
    {
      "epoch": 2.5655168551909506,
      "grad_norm": 1.1307753324508667,
      "learning_rate": 9.321157894736843e-05,
      "loss": 0.3998,
      "step": 45815
    },
    {
      "epoch": 2.56557285250308,
      "grad_norm": 1.0618451833724976,
      "learning_rate": 9.321131578947368e-05,
      "loss": 0.3663,
      "step": 45816
    },
    {
      "epoch": 2.5656288498152087,
      "grad_norm": 1.0767853260040283,
      "learning_rate": 9.321105263157896e-05,
      "loss": 0.3767,
      "step": 45817
    },
    {
      "epoch": 2.565684847127338,
      "grad_norm": 1.1243631839752197,
      "learning_rate": 9.321078947368422e-05,
      "loss": 0.3373,
      "step": 45818
    },
    {
      "epoch": 2.5657408444394667,
      "grad_norm": 1.1264264583587646,
      "learning_rate": 9.321052631578948e-05,
      "loss": 0.3969,
      "step": 45819
    },
    {
      "epoch": 2.565796841751596,
      "grad_norm": 1.7147566080093384,
      "learning_rate": 9.321026315789474e-05,
      "loss": 0.4581,
      "step": 45820
    },
    {
      "epoch": 2.5658528390637247,
      "grad_norm": 1.2335530519485474,
      "learning_rate": 9.321000000000001e-05,
      "loss": 0.359,
      "step": 45821
    },
    {
      "epoch": 2.565908836375854,
      "grad_norm": 1.323560118675232,
      "learning_rate": 9.320973684210527e-05,
      "loss": 0.3521,
      "step": 45822
    },
    {
      "epoch": 2.565964833687983,
      "grad_norm": 1.0332026481628418,
      "learning_rate": 9.320947368421053e-05,
      "loss": 0.3351,
      "step": 45823
    },
    {
      "epoch": 2.5660208310001122,
      "grad_norm": 1.0913318395614624,
      "learning_rate": 9.320921052631579e-05,
      "loss": 0.3768,
      "step": 45824
    },
    {
      "epoch": 2.566076828312241,
      "grad_norm": 1.2505122423171997,
      "learning_rate": 9.320894736842106e-05,
      "loss": 0.4706,
      "step": 45825
    },
    {
      "epoch": 2.5661328256243703,
      "grad_norm": 1.257021427154541,
      "learning_rate": 9.320868421052632e-05,
      "loss": 0.4098,
      "step": 45826
    },
    {
      "epoch": 2.566188822936499,
      "grad_norm": 1.1770654916763306,
      "learning_rate": 9.320842105263158e-05,
      "loss": 0.3293,
      "step": 45827
    },
    {
      "epoch": 2.5662448202486283,
      "grad_norm": 1.200415849685669,
      "learning_rate": 9.320815789473684e-05,
      "loss": 0.4818,
      "step": 45828
    },
    {
      "epoch": 2.566300817560757,
      "grad_norm": 1.106445074081421,
      "learning_rate": 9.32078947368421e-05,
      "loss": 0.3376,
      "step": 45829
    },
    {
      "epoch": 2.5663568148728864,
      "grad_norm": 1.2312828302383423,
      "learning_rate": 9.320763157894738e-05,
      "loss": 0.3006,
      "step": 45830
    },
    {
      "epoch": 2.566412812185015,
      "grad_norm": 0.9670117497444153,
      "learning_rate": 9.320736842105263e-05,
      "loss": 0.3307,
      "step": 45831
    },
    {
      "epoch": 2.5664688094971444,
      "grad_norm": 1.3457633256912231,
      "learning_rate": 9.32071052631579e-05,
      "loss": 0.3607,
      "step": 45832
    },
    {
      "epoch": 2.566524806809273,
      "grad_norm": 1.153135895729065,
      "learning_rate": 9.320684210526315e-05,
      "loss": 0.4922,
      "step": 45833
    },
    {
      "epoch": 2.5665808041214024,
      "grad_norm": 1.2689006328582764,
      "learning_rate": 9.320657894736843e-05,
      "loss": 0.3541,
      "step": 45834
    },
    {
      "epoch": 2.566636801433531,
      "grad_norm": 1.425445795059204,
      "learning_rate": 9.320631578947369e-05,
      "loss": 0.3613,
      "step": 45835
    },
    {
      "epoch": 2.5666927987456605,
      "grad_norm": 1.100764274597168,
      "learning_rate": 9.320605263157896e-05,
      "loss": 0.4537,
      "step": 45836
    },
    {
      "epoch": 2.566748796057789,
      "grad_norm": 1.0209444761276245,
      "learning_rate": 9.320578947368421e-05,
      "loss": 0.3195,
      "step": 45837
    },
    {
      "epoch": 2.5668047933699185,
      "grad_norm": 1.1242780685424805,
      "learning_rate": 9.320552631578948e-05,
      "loss": 0.3592,
      "step": 45838
    },
    {
      "epoch": 2.566860790682047,
      "grad_norm": 1.2996630668640137,
      "learning_rate": 9.320526315789474e-05,
      "loss": 0.5063,
      "step": 45839
    },
    {
      "epoch": 2.5669167879941766,
      "grad_norm": 1.2525056600570679,
      "learning_rate": 9.320500000000001e-05,
      "loss": 0.4167,
      "step": 45840
    },
    {
      "epoch": 2.566972785306305,
      "grad_norm": 1.1569304466247559,
      "learning_rate": 9.320473684210527e-05,
      "loss": 0.3451,
      "step": 45841
    },
    {
      "epoch": 2.5670287826184346,
      "grad_norm": 1.0828945636749268,
      "learning_rate": 9.320447368421053e-05,
      "loss": 0.4607,
      "step": 45842
    },
    {
      "epoch": 2.567084779930563,
      "grad_norm": 1.0482077598571777,
      "learning_rate": 9.320421052631579e-05,
      "loss": 0.3929,
      "step": 45843
    },
    {
      "epoch": 2.567140777242692,
      "grad_norm": 1.3216757774353027,
      "learning_rate": 9.320394736842105e-05,
      "loss": 0.3432,
      "step": 45844
    },
    {
      "epoch": 2.567196774554821,
      "grad_norm": 1.2227122783660889,
      "learning_rate": 9.320368421052633e-05,
      "loss": 0.4154,
      "step": 45845
    },
    {
      "epoch": 2.5672527718669502,
      "grad_norm": 1.1086390018463135,
      "learning_rate": 9.320342105263159e-05,
      "loss": 0.4039,
      "step": 45846
    },
    {
      "epoch": 2.5673087691790792,
      "grad_norm": 1.3790932893753052,
      "learning_rate": 9.320315789473684e-05,
      "loss": 0.3644,
      "step": 45847
    },
    {
      "epoch": 2.5673647664912083,
      "grad_norm": 1.1006743907928467,
      "learning_rate": 9.32028947368421e-05,
      "loss": 0.3973,
      "step": 45848
    },
    {
      "epoch": 2.5674207638033373,
      "grad_norm": 1.2658334970474243,
      "learning_rate": 9.320263157894738e-05,
      "loss": 0.4366,
      "step": 45849
    },
    {
      "epoch": 2.5674767611154663,
      "grad_norm": 1.237228274345398,
      "learning_rate": 9.320236842105264e-05,
      "loss": 0.4115,
      "step": 45850
    },
    {
      "epoch": 2.5675327584275953,
      "grad_norm": 0.971743106842041,
      "learning_rate": 9.32021052631579e-05,
      "loss": 0.5167,
      "step": 45851
    },
    {
      "epoch": 2.5675887557397243,
      "grad_norm": 0.9546078443527222,
      "learning_rate": 9.320184210526316e-05,
      "loss": 0.3766,
      "step": 45852
    },
    {
      "epoch": 2.5676447530518534,
      "grad_norm": 1.215753197669983,
      "learning_rate": 9.320157894736843e-05,
      "loss": 0.3179,
      "step": 45853
    },
    {
      "epoch": 2.5677007503639824,
      "grad_norm": 1.1478941440582275,
      "learning_rate": 9.320131578947369e-05,
      "loss": 0.3614,
      "step": 45854
    },
    {
      "epoch": 2.5677567476761114,
      "grad_norm": 1.1370004415512085,
      "learning_rate": 9.320105263157895e-05,
      "loss": 0.3273,
      "step": 45855
    },
    {
      "epoch": 2.5678127449882404,
      "grad_norm": 1.2187919616699219,
      "learning_rate": 9.320078947368421e-05,
      "loss": 0.4245,
      "step": 45856
    },
    {
      "epoch": 2.5678687423003694,
      "grad_norm": 0.967246413230896,
      "learning_rate": 9.320052631578948e-05,
      "loss": 0.3181,
      "step": 45857
    },
    {
      "epoch": 2.5679247396124985,
      "grad_norm": 1.1272865533828735,
      "learning_rate": 9.320026315789474e-05,
      "loss": 0.3703,
      "step": 45858
    },
    {
      "epoch": 2.5679807369246275,
      "grad_norm": 1.245540738105774,
      "learning_rate": 9.320000000000002e-05,
      "loss": 0.3682,
      "step": 45859
    },
    {
      "epoch": 2.5680367342367565,
      "grad_norm": 1.0926406383514404,
      "learning_rate": 9.319973684210526e-05,
      "loss": 0.3463,
      "step": 45860
    },
    {
      "epoch": 2.5680927315488855,
      "grad_norm": 1.1934795379638672,
      "learning_rate": 9.319947368421052e-05,
      "loss": 0.4253,
      "step": 45861
    },
    {
      "epoch": 2.5681487288610145,
      "grad_norm": 0.8828509449958801,
      "learning_rate": 9.31992105263158e-05,
      "loss": 0.3313,
      "step": 45862
    },
    {
      "epoch": 2.5682047261731435,
      "grad_norm": 1.3817951679229736,
      "learning_rate": 9.319894736842105e-05,
      "loss": 0.3679,
      "step": 45863
    },
    {
      "epoch": 2.5682607234852726,
      "grad_norm": 1.1344029903411865,
      "learning_rate": 9.319868421052633e-05,
      "loss": 0.3592,
      "step": 45864
    },
    {
      "epoch": 2.5683167207974016,
      "grad_norm": 1.2899951934814453,
      "learning_rate": 9.319842105263157e-05,
      "loss": 0.4868,
      "step": 45865
    },
    {
      "epoch": 2.5683727181095306,
      "grad_norm": 1.247365117073059,
      "learning_rate": 9.319815789473685e-05,
      "loss": 0.3539,
      "step": 45866
    },
    {
      "epoch": 2.5684287154216596,
      "grad_norm": 1.1153408288955688,
      "learning_rate": 9.319789473684211e-05,
      "loss": 0.4292,
      "step": 45867
    },
    {
      "epoch": 2.5684847127337886,
      "grad_norm": 1.203562617301941,
      "learning_rate": 9.319763157894738e-05,
      "loss": 0.3415,
      "step": 45868
    },
    {
      "epoch": 2.5685407100459177,
      "grad_norm": 1.2654542922973633,
      "learning_rate": 9.319736842105263e-05,
      "loss": 0.5575,
      "step": 45869
    },
    {
      "epoch": 2.5685967073580467,
      "grad_norm": 1.1862399578094482,
      "learning_rate": 9.31971052631579e-05,
      "loss": 0.4209,
      "step": 45870
    },
    {
      "epoch": 2.5686527046701757,
      "grad_norm": 1.116708755493164,
      "learning_rate": 9.319684210526316e-05,
      "loss": 0.4045,
      "step": 45871
    },
    {
      "epoch": 2.5687087019823047,
      "grad_norm": 1.1129900217056274,
      "learning_rate": 9.319657894736843e-05,
      "loss": 0.4008,
      "step": 45872
    },
    {
      "epoch": 2.5687646992944337,
      "grad_norm": 1.2307490110397339,
      "learning_rate": 9.319631578947369e-05,
      "loss": 0.4568,
      "step": 45873
    },
    {
      "epoch": 2.5688206966065628,
      "grad_norm": 1.087794542312622,
      "learning_rate": 9.319605263157895e-05,
      "loss": 0.3188,
      "step": 45874
    },
    {
      "epoch": 2.568876693918692,
      "grad_norm": 1.0588098764419556,
      "learning_rate": 9.319578947368421e-05,
      "loss": 0.3026,
      "step": 45875
    },
    {
      "epoch": 2.568932691230821,
      "grad_norm": 1.0869641304016113,
      "learning_rate": 9.319552631578949e-05,
      "loss": 0.4921,
      "step": 45876
    },
    {
      "epoch": 2.56898868854295,
      "grad_norm": 1.0289782285690308,
      "learning_rate": 9.319526315789474e-05,
      "loss": 0.5033,
      "step": 45877
    },
    {
      "epoch": 2.569044685855079,
      "grad_norm": 1.1767892837524414,
      "learning_rate": 9.3195e-05,
      "loss": 0.3714,
      "step": 45878
    },
    {
      "epoch": 2.569100683167208,
      "grad_norm": 1.3125910758972168,
      "learning_rate": 9.319473684210526e-05,
      "loss": 0.5378,
      "step": 45879
    },
    {
      "epoch": 2.569156680479337,
      "grad_norm": 0.9992678165435791,
      "learning_rate": 9.319447368421052e-05,
      "loss": 0.3305,
      "step": 45880
    },
    {
      "epoch": 2.569212677791466,
      "grad_norm": 0.9862795472145081,
      "learning_rate": 9.31942105263158e-05,
      "loss": 0.4197,
      "step": 45881
    },
    {
      "epoch": 2.569268675103595,
      "grad_norm": 1.3451968431472778,
      "learning_rate": 9.319394736842106e-05,
      "loss": 0.3356,
      "step": 45882
    },
    {
      "epoch": 2.569324672415724,
      "grad_norm": 1.1789580583572388,
      "learning_rate": 9.319368421052632e-05,
      "loss": 0.5989,
      "step": 45883
    },
    {
      "epoch": 2.569380669727853,
      "grad_norm": 1.1294902563095093,
      "learning_rate": 9.319342105263158e-05,
      "loss": 0.3753,
      "step": 45884
    },
    {
      "epoch": 2.569436667039982,
      "grad_norm": 1.2693742513656616,
      "learning_rate": 9.319315789473685e-05,
      "loss": 0.4586,
      "step": 45885
    },
    {
      "epoch": 2.569492664352111,
      "grad_norm": 0.9450874924659729,
      "learning_rate": 9.319289473684211e-05,
      "loss": 0.313,
      "step": 45886
    },
    {
      "epoch": 2.56954866166424,
      "grad_norm": 1.0429608821868896,
      "learning_rate": 9.319263157894737e-05,
      "loss": 0.3548,
      "step": 45887
    },
    {
      "epoch": 2.569604658976369,
      "grad_norm": 1.0232843160629272,
      "learning_rate": 9.319236842105263e-05,
      "loss": 0.2763,
      "step": 45888
    },
    {
      "epoch": 2.569660656288498,
      "grad_norm": 1.0586025714874268,
      "learning_rate": 9.31921052631579e-05,
      "loss": 0.3469,
      "step": 45889
    },
    {
      "epoch": 2.569716653600627,
      "grad_norm": 1.1040581464767456,
      "learning_rate": 9.319184210526316e-05,
      "loss": 0.4477,
      "step": 45890
    },
    {
      "epoch": 2.569772650912756,
      "grad_norm": 1.170853853225708,
      "learning_rate": 9.319157894736844e-05,
      "loss": 0.3692,
      "step": 45891
    },
    {
      "epoch": 2.569828648224885,
      "grad_norm": 1.1636574268341064,
      "learning_rate": 9.319131578947368e-05,
      "loss": 0.4486,
      "step": 45892
    },
    {
      "epoch": 2.569884645537014,
      "grad_norm": 1.180637001991272,
      "learning_rate": 9.319105263157895e-05,
      "loss": 0.4515,
      "step": 45893
    },
    {
      "epoch": 2.569940642849143,
      "grad_norm": 1.3041411638259888,
      "learning_rate": 9.319078947368421e-05,
      "loss": 0.3805,
      "step": 45894
    },
    {
      "epoch": 2.569996640161272,
      "grad_norm": 1.2125961780548096,
      "learning_rate": 9.319052631578949e-05,
      "loss": 0.5076,
      "step": 45895
    },
    {
      "epoch": 2.570052637473401,
      "grad_norm": 1.3724616765975952,
      "learning_rate": 9.319026315789475e-05,
      "loss": 0.4547,
      "step": 45896
    },
    {
      "epoch": 2.57010863478553,
      "grad_norm": 1.2791975736618042,
      "learning_rate": 9.319e-05,
      "loss": 0.3991,
      "step": 45897
    },
    {
      "epoch": 2.570164632097659,
      "grad_norm": 1.160436749458313,
      "learning_rate": 9.318973684210527e-05,
      "loss": 0.376,
      "step": 45898
    },
    {
      "epoch": 2.5702206294097882,
      "grad_norm": 1.408887505531311,
      "learning_rate": 9.318947368421053e-05,
      "loss": 0.3605,
      "step": 45899
    },
    {
      "epoch": 2.5702766267219173,
      "grad_norm": 1.084800362586975,
      "learning_rate": 9.31892105263158e-05,
      "loss": 0.3598,
      "step": 45900
    },
    {
      "epoch": 2.5703326240340463,
      "grad_norm": 1.1411126852035522,
      "learning_rate": 9.318894736842106e-05,
      "loss": 0.4537,
      "step": 45901
    },
    {
      "epoch": 2.5703886213461753,
      "grad_norm": 1.1892144680023193,
      "learning_rate": 9.318868421052632e-05,
      "loss": 0.4685,
      "step": 45902
    },
    {
      "epoch": 2.5704446186583043,
      "grad_norm": 1.0740100145339966,
      "learning_rate": 9.318842105263158e-05,
      "loss": 0.3666,
      "step": 45903
    },
    {
      "epoch": 2.5705006159704333,
      "grad_norm": 0.806837797164917,
      "learning_rate": 9.318815789473685e-05,
      "loss": 0.2794,
      "step": 45904
    },
    {
      "epoch": 2.5705566132825624,
      "grad_norm": 1.1996101140975952,
      "learning_rate": 9.318789473684211e-05,
      "loss": 0.3135,
      "step": 45905
    },
    {
      "epoch": 2.5706126105946914,
      "grad_norm": 1.085951566696167,
      "learning_rate": 9.318763157894737e-05,
      "loss": 0.3917,
      "step": 45906
    },
    {
      "epoch": 2.5706686079068204,
      "grad_norm": 1.1220403909683228,
      "learning_rate": 9.318736842105263e-05,
      "loss": 0.4032,
      "step": 45907
    },
    {
      "epoch": 2.5707246052189494,
      "grad_norm": 1.2868386507034302,
      "learning_rate": 9.31871052631579e-05,
      "loss": 0.3779,
      "step": 45908
    },
    {
      "epoch": 2.5707806025310784,
      "grad_norm": 0.8794397711753845,
      "learning_rate": 9.318684210526316e-05,
      "loss": 0.2525,
      "step": 45909
    },
    {
      "epoch": 2.5708365998432074,
      "grad_norm": 1.1922544240951538,
      "learning_rate": 9.318657894736842e-05,
      "loss": 0.4778,
      "step": 45910
    },
    {
      "epoch": 2.5708925971553365,
      "grad_norm": 2.926259756088257,
      "learning_rate": 9.318631578947368e-05,
      "loss": 0.5772,
      "step": 45911
    },
    {
      "epoch": 2.5709485944674655,
      "grad_norm": 1.1517329216003418,
      "learning_rate": 9.318605263157896e-05,
      "loss": 0.4383,
      "step": 45912
    },
    {
      "epoch": 2.5710045917795945,
      "grad_norm": 1.367051124572754,
      "learning_rate": 9.318578947368422e-05,
      "loss": 0.4252,
      "step": 45913
    },
    {
      "epoch": 2.5710605890917235,
      "grad_norm": 1.1536588668823242,
      "learning_rate": 9.318552631578948e-05,
      "loss": 0.398,
      "step": 45914
    },
    {
      "epoch": 2.5711165864038525,
      "grad_norm": 1.1822625398635864,
      "learning_rate": 9.318526315789474e-05,
      "loss": 0.471,
      "step": 45915
    },
    {
      "epoch": 2.5711725837159816,
      "grad_norm": 1.2724826335906982,
      "learning_rate": 9.3185e-05,
      "loss": 0.3487,
      "step": 45916
    },
    {
      "epoch": 2.5712285810281106,
      "grad_norm": 1.4357938766479492,
      "learning_rate": 9.318473684210527e-05,
      "loss": 0.48,
      "step": 45917
    },
    {
      "epoch": 2.5712845783402396,
      "grad_norm": 1.1453899145126343,
      "learning_rate": 9.318447368421053e-05,
      "loss": 0.3791,
      "step": 45918
    },
    {
      "epoch": 2.5713405756523686,
      "grad_norm": 1.3468393087387085,
      "learning_rate": 9.31842105263158e-05,
      "loss": 0.325,
      "step": 45919
    },
    {
      "epoch": 2.5713965729644976,
      "grad_norm": 1.0968509912490845,
      "learning_rate": 9.318394736842105e-05,
      "loss": 0.3483,
      "step": 45920
    },
    {
      "epoch": 2.5714525702766267,
      "grad_norm": 1.1521761417388916,
      "learning_rate": 9.318368421052632e-05,
      "loss": 0.4365,
      "step": 45921
    },
    {
      "epoch": 2.5715085675887557,
      "grad_norm": 1.0848356485366821,
      "learning_rate": 9.318342105263158e-05,
      "loss": 0.3655,
      "step": 45922
    },
    {
      "epoch": 2.5715645649008847,
      "grad_norm": 1.0041016340255737,
      "learning_rate": 9.318315789473686e-05,
      "loss": 0.3306,
      "step": 45923
    },
    {
      "epoch": 2.5716205622130137,
      "grad_norm": 1.3201125860214233,
      "learning_rate": 9.31828947368421e-05,
      "loss": 0.4463,
      "step": 45924
    },
    {
      "epoch": 2.5716765595251427,
      "grad_norm": 1.4077279567718506,
      "learning_rate": 9.318263157894737e-05,
      "loss": 0.472,
      "step": 45925
    },
    {
      "epoch": 2.5717325568372718,
      "grad_norm": 1.3431774377822876,
      "learning_rate": 9.318236842105263e-05,
      "loss": 0.3716,
      "step": 45926
    },
    {
      "epoch": 2.5717885541494008,
      "grad_norm": 1.0158594846725464,
      "learning_rate": 9.318210526315791e-05,
      "loss": 0.3127,
      "step": 45927
    },
    {
      "epoch": 2.57184455146153,
      "grad_norm": 1.1353245973587036,
      "learning_rate": 9.318184210526317e-05,
      "loss": 0.4089,
      "step": 45928
    },
    {
      "epoch": 2.571900548773659,
      "grad_norm": 1.1977567672729492,
      "learning_rate": 9.318157894736843e-05,
      "loss": 0.4002,
      "step": 45929
    },
    {
      "epoch": 2.571956546085788,
      "grad_norm": 1.6051511764526367,
      "learning_rate": 9.318131578947369e-05,
      "loss": 0.4603,
      "step": 45930
    },
    {
      "epoch": 2.572012543397917,
      "grad_norm": 1.0861529111862183,
      "learning_rate": 9.318105263157895e-05,
      "loss": 0.266,
      "step": 45931
    },
    {
      "epoch": 2.572068540710046,
      "grad_norm": 1.0562223196029663,
      "learning_rate": 9.318078947368422e-05,
      "loss": 0.359,
      "step": 45932
    },
    {
      "epoch": 2.572124538022175,
      "grad_norm": 0.9891675114631653,
      "learning_rate": 9.318052631578948e-05,
      "loss": 0.3565,
      "step": 45933
    },
    {
      "epoch": 2.572180535334304,
      "grad_norm": 1.4393901824951172,
      "learning_rate": 9.318026315789474e-05,
      "loss": 0.4011,
      "step": 45934
    },
    {
      "epoch": 2.572236532646433,
      "grad_norm": 1.237891435623169,
      "learning_rate": 9.318e-05,
      "loss": 0.4604,
      "step": 45935
    },
    {
      "epoch": 2.572292529958562,
      "grad_norm": 1.0912176370620728,
      "learning_rate": 9.317973684210527e-05,
      "loss": 0.3481,
      "step": 45936
    },
    {
      "epoch": 2.572348527270691,
      "grad_norm": 1.2254101037979126,
      "learning_rate": 9.317947368421053e-05,
      "loss": 0.2802,
      "step": 45937
    },
    {
      "epoch": 2.57240452458282,
      "grad_norm": 1.397782325744629,
      "learning_rate": 9.317921052631579e-05,
      "loss": 0.4698,
      "step": 45938
    },
    {
      "epoch": 2.572460521894949,
      "grad_norm": 1.6142774820327759,
      "learning_rate": 9.317894736842105e-05,
      "loss": 0.4267,
      "step": 45939
    },
    {
      "epoch": 2.572516519207078,
      "grad_norm": 1.0425869226455688,
      "learning_rate": 9.317868421052632e-05,
      "loss": 0.3316,
      "step": 45940
    },
    {
      "epoch": 2.572572516519207,
      "grad_norm": 1.012616753578186,
      "learning_rate": 9.317842105263158e-05,
      "loss": 0.508,
      "step": 45941
    },
    {
      "epoch": 2.572628513831336,
      "grad_norm": 1.1643704175949097,
      "learning_rate": 9.317815789473684e-05,
      "loss": 0.3385,
      "step": 45942
    },
    {
      "epoch": 2.572684511143465,
      "grad_norm": 1.028082251548767,
      "learning_rate": 9.31778947368421e-05,
      "loss": 0.343,
      "step": 45943
    },
    {
      "epoch": 2.572740508455594,
      "grad_norm": 1.3522393703460693,
      "learning_rate": 9.317763157894738e-05,
      "loss": 0.4234,
      "step": 45944
    },
    {
      "epoch": 2.572796505767723,
      "grad_norm": 1.1051863431930542,
      "learning_rate": 9.317736842105264e-05,
      "loss": 0.4046,
      "step": 45945
    },
    {
      "epoch": 2.572852503079852,
      "grad_norm": 1.5609067678451538,
      "learning_rate": 9.317710526315791e-05,
      "loss": 0.4757,
      "step": 45946
    },
    {
      "epoch": 2.572908500391981,
      "grad_norm": 1.1490943431854248,
      "learning_rate": 9.317684210526316e-05,
      "loss": 0.5026,
      "step": 45947
    },
    {
      "epoch": 2.57296449770411,
      "grad_norm": 1.4425662755966187,
      "learning_rate": 9.317657894736842e-05,
      "loss": 0.5196,
      "step": 45948
    },
    {
      "epoch": 2.573020495016239,
      "grad_norm": 1.1500816345214844,
      "learning_rate": 9.317631578947369e-05,
      "loss": 0.3398,
      "step": 45949
    },
    {
      "epoch": 2.573076492328368,
      "grad_norm": 0.9664220809936523,
      "learning_rate": 9.317605263157895e-05,
      "loss": 0.356,
      "step": 45950
    },
    {
      "epoch": 2.5731324896404972,
      "grad_norm": 1.1630465984344482,
      "learning_rate": 9.317578947368422e-05,
      "loss": 0.3445,
      "step": 45951
    },
    {
      "epoch": 2.5731884869526263,
      "grad_norm": 1.26423180103302,
      "learning_rate": 9.317552631578947e-05,
      "loss": 0.4185,
      "step": 45952
    },
    {
      "epoch": 2.5732444842647553,
      "grad_norm": 1.2639659643173218,
      "learning_rate": 9.317526315789474e-05,
      "loss": 0.3768,
      "step": 45953
    },
    {
      "epoch": 2.5733004815768843,
      "grad_norm": 1.079248309135437,
      "learning_rate": 9.3175e-05,
      "loss": 0.3106,
      "step": 45954
    },
    {
      "epoch": 2.5733564788890133,
      "grad_norm": 1.1732382774353027,
      "learning_rate": 9.317473684210527e-05,
      "loss": 0.4007,
      "step": 45955
    },
    {
      "epoch": 2.5734124762011423,
      "grad_norm": 1.1750617027282715,
      "learning_rate": 9.317447368421053e-05,
      "loss": 0.3312,
      "step": 45956
    },
    {
      "epoch": 2.5734684735132713,
      "grad_norm": 0.8681473731994629,
      "learning_rate": 9.31742105263158e-05,
      "loss": 0.2851,
      "step": 45957
    },
    {
      "epoch": 2.5735244708254004,
      "grad_norm": 1.1687676906585693,
      "learning_rate": 9.317394736842105e-05,
      "loss": 0.3381,
      "step": 45958
    },
    {
      "epoch": 2.5735804681375294,
      "grad_norm": 0.9861500263214111,
      "learning_rate": 9.317368421052633e-05,
      "loss": 0.3698,
      "step": 45959
    },
    {
      "epoch": 2.5736364654496584,
      "grad_norm": 1.2362630367279053,
      "learning_rate": 9.317342105263159e-05,
      "loss": 0.4299,
      "step": 45960
    },
    {
      "epoch": 2.5736924627617874,
      "grad_norm": 1.0838663578033447,
      "learning_rate": 9.317315789473685e-05,
      "loss": 0.4099,
      "step": 45961
    },
    {
      "epoch": 2.5737484600739164,
      "grad_norm": 1.8621318340301514,
      "learning_rate": 9.31728947368421e-05,
      "loss": 0.3261,
      "step": 45962
    },
    {
      "epoch": 2.5738044573860455,
      "grad_norm": 1.1353189945220947,
      "learning_rate": 9.317263157894738e-05,
      "loss": 0.3837,
      "step": 45963
    },
    {
      "epoch": 2.5738604546981745,
      "grad_norm": 0.9756107926368713,
      "learning_rate": 9.317236842105264e-05,
      "loss": 0.4376,
      "step": 45964
    },
    {
      "epoch": 2.5739164520103035,
      "grad_norm": 1.469584345817566,
      "learning_rate": 9.31721052631579e-05,
      "loss": 0.4687,
      "step": 45965
    },
    {
      "epoch": 2.5739724493224325,
      "grad_norm": 1.4445240497589111,
      "learning_rate": 9.317184210526316e-05,
      "loss": 0.6256,
      "step": 45966
    },
    {
      "epoch": 2.5740284466345615,
      "grad_norm": 1.159963607788086,
      "learning_rate": 9.317157894736842e-05,
      "loss": 0.3382,
      "step": 45967
    },
    {
      "epoch": 2.5740844439466906,
      "grad_norm": 1.350013256072998,
      "learning_rate": 9.317131578947369e-05,
      "loss": 0.4519,
      "step": 45968
    },
    {
      "epoch": 2.5741404412588196,
      "grad_norm": 1.1050174236297607,
      "learning_rate": 9.317105263157895e-05,
      "loss": 0.3941,
      "step": 45969
    },
    {
      "epoch": 2.5741964385709486,
      "grad_norm": 1.1208648681640625,
      "learning_rate": 9.317078947368421e-05,
      "loss": 0.4117,
      "step": 45970
    },
    {
      "epoch": 2.5742524358830776,
      "grad_norm": 1.2012273073196411,
      "learning_rate": 9.317052631578947e-05,
      "loss": 0.3426,
      "step": 45971
    },
    {
      "epoch": 2.5743084331952066,
      "grad_norm": 1.1187759637832642,
      "learning_rate": 9.317026315789474e-05,
      "loss": 0.377,
      "step": 45972
    },
    {
      "epoch": 2.5743644305073357,
      "grad_norm": 1.0412623882293701,
      "learning_rate": 9.317e-05,
      "loss": 0.396,
      "step": 45973
    },
    {
      "epoch": 2.5744204278194647,
      "grad_norm": 1.1677817106246948,
      "learning_rate": 9.316973684210528e-05,
      "loss": 0.3081,
      "step": 45974
    },
    {
      "epoch": 2.5744764251315937,
      "grad_norm": 1.2372729778289795,
      "learning_rate": 9.316947368421052e-05,
      "loss": 0.5085,
      "step": 45975
    },
    {
      "epoch": 2.5745324224437227,
      "grad_norm": 1.1914616823196411,
      "learning_rate": 9.31692105263158e-05,
      "loss": 0.4636,
      "step": 45976
    },
    {
      "epoch": 2.5745884197558517,
      "grad_norm": 1.2914350032806396,
      "learning_rate": 9.316894736842106e-05,
      "loss": 0.4887,
      "step": 45977
    },
    {
      "epoch": 2.5746444170679808,
      "grad_norm": 1.4277184009552002,
      "learning_rate": 9.316868421052633e-05,
      "loss": 0.6,
      "step": 45978
    },
    {
      "epoch": 2.5747004143801098,
      "grad_norm": 1.4837602376937866,
      "learning_rate": 9.316842105263158e-05,
      "loss": 0.4269,
      "step": 45979
    },
    {
      "epoch": 2.574756411692239,
      "grad_norm": 1.2476377487182617,
      "learning_rate": 9.316815789473685e-05,
      "loss": 0.4336,
      "step": 45980
    },
    {
      "epoch": 2.574812409004368,
      "grad_norm": 1.2525088787078857,
      "learning_rate": 9.316789473684211e-05,
      "loss": 0.325,
      "step": 45981
    },
    {
      "epoch": 2.574868406316497,
      "grad_norm": 1.3420413732528687,
      "learning_rate": 9.316763157894737e-05,
      "loss": 0.3898,
      "step": 45982
    },
    {
      "epoch": 2.574924403628626,
      "grad_norm": 1.2477458715438843,
      "learning_rate": 9.316736842105264e-05,
      "loss": 0.3686,
      "step": 45983
    },
    {
      "epoch": 2.574980400940755,
      "grad_norm": 1.1343430280685425,
      "learning_rate": 9.316710526315789e-05,
      "loss": 0.3928,
      "step": 45984
    },
    {
      "epoch": 2.575036398252884,
      "grad_norm": 1.3381307125091553,
      "learning_rate": 9.316684210526316e-05,
      "loss": 0.3457,
      "step": 45985
    },
    {
      "epoch": 2.575092395565013,
      "grad_norm": 1.1925050020217896,
      "learning_rate": 9.316657894736842e-05,
      "loss": 0.4944,
      "step": 45986
    },
    {
      "epoch": 2.575148392877142,
      "grad_norm": 2.661663055419922,
      "learning_rate": 9.31663157894737e-05,
      "loss": 0.3457,
      "step": 45987
    },
    {
      "epoch": 2.575204390189271,
      "grad_norm": 1.2135288715362549,
      "learning_rate": 9.316605263157895e-05,
      "loss": 0.3453,
      "step": 45988
    },
    {
      "epoch": 2.5752603875014,
      "grad_norm": 1.183806300163269,
      "learning_rate": 9.316578947368421e-05,
      "loss": 0.4,
      "step": 45989
    },
    {
      "epoch": 2.575316384813529,
      "grad_norm": 1.4379099607467651,
      "learning_rate": 9.316552631578947e-05,
      "loss": 0.5303,
      "step": 45990
    },
    {
      "epoch": 2.575372382125658,
      "grad_norm": 1.1737977266311646,
      "learning_rate": 9.316526315789475e-05,
      "loss": 0.3619,
      "step": 45991
    },
    {
      "epoch": 2.575428379437787,
      "grad_norm": 1.280013918876648,
      "learning_rate": 9.3165e-05,
      "loss": 0.4399,
      "step": 45992
    },
    {
      "epoch": 2.575484376749916,
      "grad_norm": 1.1782958507537842,
      "learning_rate": 9.316473684210527e-05,
      "loss": 0.3295,
      "step": 45993
    },
    {
      "epoch": 2.575540374062045,
      "grad_norm": 1.1879894733428955,
      "learning_rate": 9.316447368421053e-05,
      "loss": 0.4469,
      "step": 45994
    },
    {
      "epoch": 2.575596371374174,
      "grad_norm": 1.0662089586257935,
      "learning_rate": 9.31642105263158e-05,
      "loss": 0.3716,
      "step": 45995
    },
    {
      "epoch": 2.575652368686303,
      "grad_norm": 1.3476603031158447,
      "learning_rate": 9.316394736842106e-05,
      "loss": 0.36,
      "step": 45996
    },
    {
      "epoch": 2.575708365998432,
      "grad_norm": 1.289571762084961,
      "learning_rate": 9.316368421052632e-05,
      "loss": 0.3726,
      "step": 45997
    },
    {
      "epoch": 2.575764363310561,
      "grad_norm": 1.0155049562454224,
      "learning_rate": 9.316342105263158e-05,
      "loss": 0.3355,
      "step": 45998
    },
    {
      "epoch": 2.57582036062269,
      "grad_norm": 0.9930593967437744,
      "learning_rate": 9.316315789473684e-05,
      "loss": 0.3609,
      "step": 45999
    },
    {
      "epoch": 2.575876357934819,
      "grad_norm": 1.5688420534133911,
      "learning_rate": 9.316289473684211e-05,
      "loss": 0.3788,
      "step": 46000
    },
    {
      "epoch": 2.575932355246948,
      "grad_norm": 1.0519797801971436,
      "learning_rate": 9.316263157894737e-05,
      "loss": 0.3114,
      "step": 46001
    },
    {
      "epoch": 2.575988352559077,
      "grad_norm": 1.3388243913650513,
      "learning_rate": 9.316236842105263e-05,
      "loss": 0.3639,
      "step": 46002
    },
    {
      "epoch": 2.5760443498712062,
      "grad_norm": 1.2444498538970947,
      "learning_rate": 9.316210526315789e-05,
      "loss": 0.5302,
      "step": 46003
    },
    {
      "epoch": 2.5761003471833352,
      "grad_norm": 0.947036862373352,
      "learning_rate": 9.316184210526316e-05,
      "loss": 0.3336,
      "step": 46004
    },
    {
      "epoch": 2.5761563444954643,
      "grad_norm": 1.1413090229034424,
      "learning_rate": 9.316157894736842e-05,
      "loss": 0.3762,
      "step": 46005
    },
    {
      "epoch": 2.5762123418075933,
      "grad_norm": 1.0925520658493042,
      "learning_rate": 9.31613157894737e-05,
      "loss": 0.4306,
      "step": 46006
    },
    {
      "epoch": 2.5762683391197223,
      "grad_norm": 1.22615385055542,
      "learning_rate": 9.316105263157894e-05,
      "loss": 0.4755,
      "step": 46007
    },
    {
      "epoch": 2.5763243364318513,
      "grad_norm": 1.077869176864624,
      "learning_rate": 9.316078947368422e-05,
      "loss": 0.3231,
      "step": 46008
    },
    {
      "epoch": 2.5763803337439803,
      "grad_norm": 1.1102615594863892,
      "learning_rate": 9.316052631578948e-05,
      "loss": 0.3705,
      "step": 46009
    },
    {
      "epoch": 2.5764363310561094,
      "grad_norm": 1.5360053777694702,
      "learning_rate": 9.316026315789475e-05,
      "loss": 0.4663,
      "step": 46010
    },
    {
      "epoch": 2.5764923283682384,
      "grad_norm": 1.1477519273757935,
      "learning_rate": 9.316000000000001e-05,
      "loss": 0.3577,
      "step": 46011
    },
    {
      "epoch": 2.5765483256803674,
      "grad_norm": 1.1828967332839966,
      "learning_rate": 9.315973684210527e-05,
      "loss": 0.3282,
      "step": 46012
    },
    {
      "epoch": 2.5766043229924964,
      "grad_norm": 1.071303129196167,
      "learning_rate": 9.315947368421053e-05,
      "loss": 0.3449,
      "step": 46013
    },
    {
      "epoch": 2.5766603203046254,
      "grad_norm": 1.121272325515747,
      "learning_rate": 9.31592105263158e-05,
      "loss": 0.4479,
      "step": 46014
    },
    {
      "epoch": 2.5767163176167545,
      "grad_norm": 1.4459490776062012,
      "learning_rate": 9.315894736842106e-05,
      "loss": 0.445,
      "step": 46015
    },
    {
      "epoch": 2.5767723149288835,
      "grad_norm": 1.2250949144363403,
      "learning_rate": 9.315868421052632e-05,
      "loss": 0.5084,
      "step": 46016
    },
    {
      "epoch": 2.5768283122410125,
      "grad_norm": 0.9811687469482422,
      "learning_rate": 9.315842105263158e-05,
      "loss": 0.4029,
      "step": 46017
    },
    {
      "epoch": 2.5768843095531415,
      "grad_norm": 1.3154581785202026,
      "learning_rate": 9.315815789473684e-05,
      "loss": 0.3411,
      "step": 46018
    },
    {
      "epoch": 2.5769403068652705,
      "grad_norm": 1.0479449033737183,
      "learning_rate": 9.315789473684211e-05,
      "loss": 0.3482,
      "step": 46019
    },
    {
      "epoch": 2.5769963041773996,
      "grad_norm": 1.1492435932159424,
      "learning_rate": 9.315763157894737e-05,
      "loss": 0.3607,
      "step": 46020
    },
    {
      "epoch": 2.5770523014895286,
      "grad_norm": 1.0657223463058472,
      "learning_rate": 9.315736842105263e-05,
      "loss": 0.3465,
      "step": 46021
    },
    {
      "epoch": 2.5771082988016576,
      "grad_norm": 1.4544436931610107,
      "learning_rate": 9.31571052631579e-05,
      "loss": 0.6357,
      "step": 46022
    },
    {
      "epoch": 2.5771642961137866,
      "grad_norm": 1.054081916809082,
      "learning_rate": 9.315684210526317e-05,
      "loss": 0.3657,
      "step": 46023
    },
    {
      "epoch": 2.5772202934259156,
      "grad_norm": 1.0402076244354248,
      "learning_rate": 9.315657894736843e-05,
      "loss": 0.3238,
      "step": 46024
    },
    {
      "epoch": 2.5772762907380447,
      "grad_norm": 1.1456807851791382,
      "learning_rate": 9.315631578947369e-05,
      "loss": 0.4309,
      "step": 46025
    },
    {
      "epoch": 2.5773322880501737,
      "grad_norm": 1.136291742324829,
      "learning_rate": 9.315605263157895e-05,
      "loss": 0.2956,
      "step": 46026
    },
    {
      "epoch": 2.5773882853623027,
      "grad_norm": 1.1384942531585693,
      "learning_rate": 9.315578947368422e-05,
      "loss": 0.3905,
      "step": 46027
    },
    {
      "epoch": 2.5774442826744317,
      "grad_norm": 1.0591740608215332,
      "learning_rate": 9.315552631578948e-05,
      "loss": 0.4322,
      "step": 46028
    },
    {
      "epoch": 2.5775002799865607,
      "grad_norm": 1.1473219394683838,
      "learning_rate": 9.315526315789475e-05,
      "loss": 0.3987,
      "step": 46029
    },
    {
      "epoch": 2.5775562772986897,
      "grad_norm": 1.072749376296997,
      "learning_rate": 9.3155e-05,
      "loss": 0.3893,
      "step": 46030
    },
    {
      "epoch": 2.5776122746108188,
      "grad_norm": 1.4761978387832642,
      "learning_rate": 9.315473684210527e-05,
      "loss": 0.5563,
      "step": 46031
    },
    {
      "epoch": 2.577668271922948,
      "grad_norm": 1.0590451955795288,
      "learning_rate": 9.315447368421053e-05,
      "loss": 0.3499,
      "step": 46032
    },
    {
      "epoch": 2.577724269235077,
      "grad_norm": 0.8593736886978149,
      "learning_rate": 9.31542105263158e-05,
      "loss": 0.2561,
      "step": 46033
    },
    {
      "epoch": 2.577780266547206,
      "grad_norm": 1.2652361392974854,
      "learning_rate": 9.315394736842105e-05,
      "loss": 0.5819,
      "step": 46034
    },
    {
      "epoch": 2.577836263859335,
      "grad_norm": 1.181604266166687,
      "learning_rate": 9.315368421052631e-05,
      "loss": 0.4735,
      "step": 46035
    },
    {
      "epoch": 2.577892261171464,
      "grad_norm": 1.280901551246643,
      "learning_rate": 9.315342105263158e-05,
      "loss": 0.5325,
      "step": 46036
    },
    {
      "epoch": 2.577948258483593,
      "grad_norm": 1.1786962747573853,
      "learning_rate": 9.315315789473684e-05,
      "loss": 0.4014,
      "step": 46037
    },
    {
      "epoch": 2.578004255795722,
      "grad_norm": 1.397752046585083,
      "learning_rate": 9.315289473684212e-05,
      "loss": 0.3402,
      "step": 46038
    },
    {
      "epoch": 2.578060253107851,
      "grad_norm": 1.1130037307739258,
      "learning_rate": 9.315263157894736e-05,
      "loss": 0.4848,
      "step": 46039
    },
    {
      "epoch": 2.57811625041998,
      "grad_norm": 1.360951542854309,
      "learning_rate": 9.315236842105264e-05,
      "loss": 0.449,
      "step": 46040
    },
    {
      "epoch": 2.578172247732109,
      "grad_norm": 1.142729640007019,
      "learning_rate": 9.31521052631579e-05,
      "loss": 0.3708,
      "step": 46041
    },
    {
      "epoch": 2.578228245044238,
      "grad_norm": 0.9992817044258118,
      "learning_rate": 9.315184210526317e-05,
      "loss": 0.3213,
      "step": 46042
    },
    {
      "epoch": 2.578284242356367,
      "grad_norm": 1.3108927011489868,
      "learning_rate": 9.315157894736843e-05,
      "loss": 0.4167,
      "step": 46043
    },
    {
      "epoch": 2.578340239668496,
      "grad_norm": 1.5837466716766357,
      "learning_rate": 9.315131578947369e-05,
      "loss": 0.5196,
      "step": 46044
    },
    {
      "epoch": 2.578396236980625,
      "grad_norm": 0.927727222442627,
      "learning_rate": 9.315105263157895e-05,
      "loss": 0.3432,
      "step": 46045
    },
    {
      "epoch": 2.578452234292754,
      "grad_norm": 1.1483770608901978,
      "learning_rate": 9.315078947368422e-05,
      "loss": 0.4136,
      "step": 46046
    },
    {
      "epoch": 2.578508231604883,
      "grad_norm": 1.0409961938858032,
      "learning_rate": 9.315052631578948e-05,
      "loss": 0.4282,
      "step": 46047
    },
    {
      "epoch": 2.578564228917012,
      "grad_norm": 0.9694766402244568,
      "learning_rate": 9.315026315789474e-05,
      "loss": 0.4193,
      "step": 46048
    },
    {
      "epoch": 2.578620226229141,
      "grad_norm": 1.0620146989822388,
      "learning_rate": 9.315e-05,
      "loss": 0.3902,
      "step": 46049
    },
    {
      "epoch": 2.57867622354127,
      "grad_norm": 1.1515355110168457,
      "learning_rate": 9.314973684210527e-05,
      "loss": 0.376,
      "step": 46050
    },
    {
      "epoch": 2.578732220853399,
      "grad_norm": 1.056361436843872,
      "learning_rate": 9.314947368421053e-05,
      "loss": 0.3444,
      "step": 46051
    },
    {
      "epoch": 2.578788218165528,
      "grad_norm": 1.074299931526184,
      "learning_rate": 9.31492105263158e-05,
      "loss": 0.4389,
      "step": 46052
    },
    {
      "epoch": 2.578844215477657,
      "grad_norm": 1.1565958261489868,
      "learning_rate": 9.314894736842105e-05,
      "loss": 0.4372,
      "step": 46053
    },
    {
      "epoch": 2.578900212789786,
      "grad_norm": 1.2462416887283325,
      "learning_rate": 9.314868421052631e-05,
      "loss": 0.3845,
      "step": 46054
    },
    {
      "epoch": 2.5789562101019152,
      "grad_norm": 1.2445703744888306,
      "learning_rate": 9.314842105263159e-05,
      "loss": 0.3762,
      "step": 46055
    },
    {
      "epoch": 2.5790122074140442,
      "grad_norm": 1.1718096733093262,
      "learning_rate": 9.314815789473685e-05,
      "loss": 0.3526,
      "step": 46056
    },
    {
      "epoch": 2.5790682047261733,
      "grad_norm": 2.2784805297851562,
      "learning_rate": 9.31478947368421e-05,
      "loss": 0.6718,
      "step": 46057
    },
    {
      "epoch": 2.5791242020383023,
      "grad_norm": 1.2471497058868408,
      "learning_rate": 9.314763157894737e-05,
      "loss": 0.3402,
      "step": 46058
    },
    {
      "epoch": 2.5791801993504313,
      "grad_norm": 1.159052848815918,
      "learning_rate": 9.314736842105264e-05,
      "loss": 0.3907,
      "step": 46059
    },
    {
      "epoch": 2.5792361966625603,
      "grad_norm": 1.467471718788147,
      "learning_rate": 9.31471052631579e-05,
      "loss": 0.4823,
      "step": 46060
    },
    {
      "epoch": 2.5792921939746893,
      "grad_norm": 1.1169527769088745,
      "learning_rate": 9.314684210526317e-05,
      "loss": 0.3697,
      "step": 46061
    },
    {
      "epoch": 2.5793481912868184,
      "grad_norm": 1.257900357246399,
      "learning_rate": 9.314657894736842e-05,
      "loss": 0.4065,
      "step": 46062
    },
    {
      "epoch": 2.5794041885989474,
      "grad_norm": 1.2235114574432373,
      "learning_rate": 9.314631578947369e-05,
      "loss": 0.3677,
      "step": 46063
    },
    {
      "epoch": 2.5794601859110764,
      "grad_norm": 1.4424211978912354,
      "learning_rate": 9.314605263157895e-05,
      "loss": 0.4321,
      "step": 46064
    },
    {
      "epoch": 2.5795161832232054,
      "grad_norm": 1.1728360652923584,
      "learning_rate": 9.314578947368422e-05,
      "loss": 0.5346,
      "step": 46065
    },
    {
      "epoch": 2.5795721805353344,
      "grad_norm": 1.0435181856155396,
      "learning_rate": 9.314552631578948e-05,
      "loss": 0.3737,
      "step": 46066
    },
    {
      "epoch": 2.5796281778474635,
      "grad_norm": 1.06194269657135,
      "learning_rate": 9.314526315789474e-05,
      "loss": 0.3952,
      "step": 46067
    },
    {
      "epoch": 2.5796841751595925,
      "grad_norm": 1.3419098854064941,
      "learning_rate": 9.3145e-05,
      "loss": 0.3814,
      "step": 46068
    },
    {
      "epoch": 2.5797401724717215,
      "grad_norm": 1.0676994323730469,
      "learning_rate": 9.314473684210526e-05,
      "loss": 0.4272,
      "step": 46069
    },
    {
      "epoch": 2.5797961697838505,
      "grad_norm": 1.0949664115905762,
      "learning_rate": 9.314447368421054e-05,
      "loss": 0.3657,
      "step": 46070
    },
    {
      "epoch": 2.5798521670959795,
      "grad_norm": 1.0061603784561157,
      "learning_rate": 9.314421052631578e-05,
      "loss": 0.2596,
      "step": 46071
    },
    {
      "epoch": 2.5799081644081086,
      "grad_norm": 1.1637463569641113,
      "learning_rate": 9.314394736842106e-05,
      "loss": 0.4634,
      "step": 46072
    },
    {
      "epoch": 2.5799641617202376,
      "grad_norm": 0.9348063468933105,
      "learning_rate": 9.314368421052632e-05,
      "loss": 0.3751,
      "step": 46073
    },
    {
      "epoch": 2.5800201590323666,
      "grad_norm": 1.1514455080032349,
      "learning_rate": 9.314342105263159e-05,
      "loss": 0.3176,
      "step": 46074
    },
    {
      "epoch": 2.5800761563444956,
      "grad_norm": 1.101204752922058,
      "learning_rate": 9.314315789473685e-05,
      "loss": 0.3418,
      "step": 46075
    },
    {
      "epoch": 2.5801321536566246,
      "grad_norm": 0.9068188667297363,
      "learning_rate": 9.314289473684211e-05,
      "loss": 0.2782,
      "step": 46076
    },
    {
      "epoch": 2.5801881509687536,
      "grad_norm": 1.2096669673919678,
      "learning_rate": 9.314263157894737e-05,
      "loss": 0.4085,
      "step": 46077
    },
    {
      "epoch": 2.5802441482808827,
      "grad_norm": 2.3137753009796143,
      "learning_rate": 9.314236842105264e-05,
      "loss": 0.3432,
      "step": 46078
    },
    {
      "epoch": 2.5803001455930117,
      "grad_norm": 1.6223537921905518,
      "learning_rate": 9.31421052631579e-05,
      "loss": 0.427,
      "step": 46079
    },
    {
      "epoch": 2.5803561429051407,
      "grad_norm": 1.9210237264633179,
      "learning_rate": 9.314184210526316e-05,
      "loss": 0.5861,
      "step": 46080
    },
    {
      "epoch": 2.5804121402172697,
      "grad_norm": 1.4032903909683228,
      "learning_rate": 9.314157894736842e-05,
      "loss": 0.3324,
      "step": 46081
    },
    {
      "epoch": 2.5804681375293987,
      "grad_norm": 1.6977425813674927,
      "learning_rate": 9.31413157894737e-05,
      "loss": 0.5505,
      "step": 46082
    },
    {
      "epoch": 2.5805241348415278,
      "grad_norm": 1.3876904249191284,
      "learning_rate": 9.314105263157895e-05,
      "loss": 0.3835,
      "step": 46083
    },
    {
      "epoch": 2.580580132153657,
      "grad_norm": 1.3035751581192017,
      "learning_rate": 9.314078947368423e-05,
      "loss": 0.4425,
      "step": 46084
    },
    {
      "epoch": 2.580636129465786,
      "grad_norm": 3.136500597000122,
      "learning_rate": 9.314052631578947e-05,
      "loss": 0.345,
      "step": 46085
    },
    {
      "epoch": 2.580692126777915,
      "grad_norm": 1.2417433261871338,
      "learning_rate": 9.314026315789473e-05,
      "loss": 0.4438,
      "step": 46086
    },
    {
      "epoch": 2.580748124090044,
      "grad_norm": 1.2220112085342407,
      "learning_rate": 9.314e-05,
      "loss": 0.3782,
      "step": 46087
    },
    {
      "epoch": 2.580804121402173,
      "grad_norm": 1.2444084882736206,
      "learning_rate": 9.313973684210527e-05,
      "loss": 0.4214,
      "step": 46088
    },
    {
      "epoch": 2.580860118714302,
      "grad_norm": 1.8742244243621826,
      "learning_rate": 9.313947368421053e-05,
      "loss": 0.528,
      "step": 46089
    },
    {
      "epoch": 2.580916116026431,
      "grad_norm": 1.5566651821136475,
      "learning_rate": 9.313921052631579e-05,
      "loss": 0.5614,
      "step": 46090
    },
    {
      "epoch": 2.58097211333856,
      "grad_norm": 1.311989188194275,
      "learning_rate": 9.313894736842106e-05,
      "loss": 0.5929,
      "step": 46091
    },
    {
      "epoch": 2.5810281106506885,
      "grad_norm": 1.1537903547286987,
      "learning_rate": 9.313868421052632e-05,
      "loss": 0.3887,
      "step": 46092
    },
    {
      "epoch": 2.581084107962818,
      "grad_norm": 1.5106817483901978,
      "learning_rate": 9.313842105263159e-05,
      "loss": 0.4065,
      "step": 46093
    },
    {
      "epoch": 2.5811401052749465,
      "grad_norm": 1.0126575231552124,
      "learning_rate": 9.313815789473684e-05,
      "loss": 0.3849,
      "step": 46094
    },
    {
      "epoch": 2.581196102587076,
      "grad_norm": 1.2632516622543335,
      "learning_rate": 9.313789473684211e-05,
      "loss": 0.4506,
      "step": 46095
    },
    {
      "epoch": 2.5812520998992046,
      "grad_norm": 1.5963244438171387,
      "learning_rate": 9.313763157894737e-05,
      "loss": 0.406,
      "step": 46096
    },
    {
      "epoch": 2.581308097211334,
      "grad_norm": 1.444861888885498,
      "learning_rate": 9.313736842105264e-05,
      "loss": 0.6319,
      "step": 46097
    },
    {
      "epoch": 2.5813640945234626,
      "grad_norm": 1.2981599569320679,
      "learning_rate": 9.31371052631579e-05,
      "loss": 0.4995,
      "step": 46098
    },
    {
      "epoch": 2.581420091835592,
      "grad_norm": 0.8989390134811401,
      "learning_rate": 9.313684210526316e-05,
      "loss": 0.3089,
      "step": 46099
    },
    {
      "epoch": 2.5814760891477206,
      "grad_norm": 1.0667085647583008,
      "learning_rate": 9.313657894736842e-05,
      "loss": 0.3027,
      "step": 46100
    },
    {
      "epoch": 2.58153208645985,
      "grad_norm": 1.2525547742843628,
      "learning_rate": 9.31363157894737e-05,
      "loss": 0.3871,
      "step": 46101
    },
    {
      "epoch": 2.5815880837719787,
      "grad_norm": 1.4255220890045166,
      "learning_rate": 9.313605263157896e-05,
      "loss": 0.4652,
      "step": 46102
    },
    {
      "epoch": 2.581644081084108,
      "grad_norm": 1.0932177305221558,
      "learning_rate": 9.313578947368422e-05,
      "loss": 0.3531,
      "step": 46103
    },
    {
      "epoch": 2.5817000783962367,
      "grad_norm": 1.2329206466674805,
      "learning_rate": 9.313552631578948e-05,
      "loss": 0.4215,
      "step": 46104
    },
    {
      "epoch": 2.581756075708366,
      "grad_norm": 1.2955242395401,
      "learning_rate": 9.313526315789474e-05,
      "loss": 0.3689,
      "step": 46105
    },
    {
      "epoch": 2.5818120730204948,
      "grad_norm": 1.2032455205917358,
      "learning_rate": 9.313500000000001e-05,
      "loss": 0.4864,
      "step": 46106
    },
    {
      "epoch": 2.581868070332624,
      "grad_norm": 1.3263942003250122,
      "learning_rate": 9.313473684210527e-05,
      "loss": 0.3865,
      "step": 46107
    },
    {
      "epoch": 2.581924067644753,
      "grad_norm": 1.122671365737915,
      "learning_rate": 9.313447368421053e-05,
      "loss": 0.248,
      "step": 46108
    },
    {
      "epoch": 2.5819800649568823,
      "grad_norm": 1.2542387247085571,
      "learning_rate": 9.313421052631579e-05,
      "loss": 0.5328,
      "step": 46109
    },
    {
      "epoch": 2.582036062269011,
      "grad_norm": 1.110645055770874,
      "learning_rate": 9.313394736842106e-05,
      "loss": 0.3752,
      "step": 46110
    },
    {
      "epoch": 2.5820920595811403,
      "grad_norm": 1.1280505657196045,
      "learning_rate": 9.313368421052632e-05,
      "loss": 0.4218,
      "step": 46111
    },
    {
      "epoch": 2.582148056893269,
      "grad_norm": 1.1789436340332031,
      "learning_rate": 9.313342105263158e-05,
      "loss": 0.4076,
      "step": 46112
    },
    {
      "epoch": 2.5822040542053983,
      "grad_norm": 1.2088063955307007,
      "learning_rate": 9.313315789473684e-05,
      "loss": 0.4068,
      "step": 46113
    },
    {
      "epoch": 2.582260051517527,
      "grad_norm": 1.3133759498596191,
      "learning_rate": 9.313289473684211e-05,
      "loss": 0.363,
      "step": 46114
    },
    {
      "epoch": 2.5823160488296564,
      "grad_norm": 1.0569279193878174,
      "learning_rate": 9.313263157894737e-05,
      "loss": 0.3917,
      "step": 46115
    },
    {
      "epoch": 2.582372046141785,
      "grad_norm": 1.1340553760528564,
      "learning_rate": 9.313236842105265e-05,
      "loss": 0.4839,
      "step": 46116
    },
    {
      "epoch": 2.5824280434539144,
      "grad_norm": 1.097480058670044,
      "learning_rate": 9.313210526315789e-05,
      "loss": 0.4591,
      "step": 46117
    },
    {
      "epoch": 2.582484040766043,
      "grad_norm": 1.458385944366455,
      "learning_rate": 9.313184210526317e-05,
      "loss": 0.5556,
      "step": 46118
    },
    {
      "epoch": 2.5825400380781725,
      "grad_norm": 1.0876173973083496,
      "learning_rate": 9.313157894736843e-05,
      "loss": 0.3459,
      "step": 46119
    },
    {
      "epoch": 2.582596035390301,
      "grad_norm": 1.0457082986831665,
      "learning_rate": 9.313131578947369e-05,
      "loss": 0.3347,
      "step": 46120
    },
    {
      "epoch": 2.5826520327024305,
      "grad_norm": 1.3583276271820068,
      "learning_rate": 9.313105263157896e-05,
      "loss": 0.4201,
      "step": 46121
    },
    {
      "epoch": 2.582708030014559,
      "grad_norm": 1.4155386686325073,
      "learning_rate": 9.31307894736842e-05,
      "loss": 0.4608,
      "step": 46122
    },
    {
      "epoch": 2.5827640273266885,
      "grad_norm": 0.9848421216011047,
      "learning_rate": 9.313052631578948e-05,
      "loss": 0.3612,
      "step": 46123
    },
    {
      "epoch": 2.582820024638817,
      "grad_norm": 1.5432130098342896,
      "learning_rate": 9.313026315789474e-05,
      "loss": 0.5421,
      "step": 46124
    },
    {
      "epoch": 2.5828760219509466,
      "grad_norm": 1.2114622592926025,
      "learning_rate": 9.313000000000001e-05,
      "loss": 0.5826,
      "step": 46125
    },
    {
      "epoch": 2.582932019263075,
      "grad_norm": 1.0866502523422241,
      "learning_rate": 9.312973684210526e-05,
      "loss": 0.3279,
      "step": 46126
    },
    {
      "epoch": 2.5829880165752046,
      "grad_norm": 1.450933575630188,
      "learning_rate": 9.312947368421053e-05,
      "loss": 0.4545,
      "step": 46127
    },
    {
      "epoch": 2.583044013887333,
      "grad_norm": 1.2252963781356812,
      "learning_rate": 9.312921052631579e-05,
      "loss": 0.3875,
      "step": 46128
    },
    {
      "epoch": 2.5831000111994626,
      "grad_norm": 1.4160548448562622,
      "learning_rate": 9.312894736842106e-05,
      "loss": 0.4755,
      "step": 46129
    },
    {
      "epoch": 2.583156008511591,
      "grad_norm": 1.111814022064209,
      "learning_rate": 9.312868421052632e-05,
      "loss": 0.4226,
      "step": 46130
    },
    {
      "epoch": 2.5832120058237207,
      "grad_norm": 1.1783521175384521,
      "learning_rate": 9.312842105263158e-05,
      "loss": 0.4091,
      "step": 46131
    },
    {
      "epoch": 2.5832680031358493,
      "grad_norm": 1.3160145282745361,
      "learning_rate": 9.312815789473684e-05,
      "loss": 0.3478,
      "step": 46132
    },
    {
      "epoch": 2.5833240004479787,
      "grad_norm": 1.1390950679779053,
      "learning_rate": 9.312789473684212e-05,
      "loss": 0.3533,
      "step": 46133
    },
    {
      "epoch": 2.5833799977601073,
      "grad_norm": 1.1132807731628418,
      "learning_rate": 9.312763157894738e-05,
      "loss": 0.408,
      "step": 46134
    },
    {
      "epoch": 2.5834359950722368,
      "grad_norm": 1.3719650506973267,
      "learning_rate": 9.312736842105264e-05,
      "loss": 0.392,
      "step": 46135
    },
    {
      "epoch": 2.5834919923843653,
      "grad_norm": 1.7874808311462402,
      "learning_rate": 9.31271052631579e-05,
      "loss": 0.393,
      "step": 46136
    },
    {
      "epoch": 2.583547989696495,
      "grad_norm": 0.940536618232727,
      "learning_rate": 9.312684210526317e-05,
      "loss": 0.3134,
      "step": 46137
    },
    {
      "epoch": 2.5836039870086234,
      "grad_norm": 1.33277428150177,
      "learning_rate": 9.312657894736843e-05,
      "loss": 0.4385,
      "step": 46138
    },
    {
      "epoch": 2.583659984320753,
      "grad_norm": 1.152375340461731,
      "learning_rate": 9.312631578947369e-05,
      "loss": 0.4978,
      "step": 46139
    },
    {
      "epoch": 2.5837159816328814,
      "grad_norm": 1.1312285661697388,
      "learning_rate": 9.312605263157895e-05,
      "loss": 0.446,
      "step": 46140
    },
    {
      "epoch": 2.583771978945011,
      "grad_norm": 1.226535439491272,
      "learning_rate": 9.312578947368421e-05,
      "loss": 0.4768,
      "step": 46141
    },
    {
      "epoch": 2.5838279762571394,
      "grad_norm": 1.7996746301651,
      "learning_rate": 9.312552631578948e-05,
      "loss": 0.4884,
      "step": 46142
    },
    {
      "epoch": 2.583883973569269,
      "grad_norm": 1.5521812438964844,
      "learning_rate": 9.312526315789474e-05,
      "loss": 0.6233,
      "step": 46143
    },
    {
      "epoch": 2.5839399708813975,
      "grad_norm": 1.3173505067825317,
      "learning_rate": 9.3125e-05,
      "loss": 0.2906,
      "step": 46144
    },
    {
      "epoch": 2.583995968193527,
      "grad_norm": 1.1461753845214844,
      "learning_rate": 9.312473684210526e-05,
      "loss": 0.4117,
      "step": 46145
    },
    {
      "epoch": 2.5840519655056555,
      "grad_norm": 1.248136281967163,
      "learning_rate": 9.312447368421053e-05,
      "loss": 0.4202,
      "step": 46146
    },
    {
      "epoch": 2.584107962817785,
      "grad_norm": 1.1915944814682007,
      "learning_rate": 9.312421052631579e-05,
      "loss": 0.3287,
      "step": 46147
    },
    {
      "epoch": 2.5841639601299136,
      "grad_norm": 1.2087528705596924,
      "learning_rate": 9.312394736842107e-05,
      "loss": 0.4107,
      "step": 46148
    },
    {
      "epoch": 2.584219957442043,
      "grad_norm": 1.3676587343215942,
      "learning_rate": 9.312368421052631e-05,
      "loss": 0.3929,
      "step": 46149
    },
    {
      "epoch": 2.5842759547541716,
      "grad_norm": 1.058803915977478,
      "learning_rate": 9.312342105263159e-05,
      "loss": 0.3817,
      "step": 46150
    },
    {
      "epoch": 2.584331952066301,
      "grad_norm": 1.0382416248321533,
      "learning_rate": 9.312315789473685e-05,
      "loss": 0.3762,
      "step": 46151
    },
    {
      "epoch": 2.5843879493784296,
      "grad_norm": 1.0956649780273438,
      "learning_rate": 9.312289473684212e-05,
      "loss": 0.3801,
      "step": 46152
    },
    {
      "epoch": 2.584443946690559,
      "grad_norm": 1.1819037199020386,
      "learning_rate": 9.312263157894738e-05,
      "loss": 0.349,
      "step": 46153
    },
    {
      "epoch": 2.5844999440026877,
      "grad_norm": 1.3597872257232666,
      "learning_rate": 9.312236842105264e-05,
      "loss": 0.3677,
      "step": 46154
    },
    {
      "epoch": 2.584555941314817,
      "grad_norm": 1.3350191116333008,
      "learning_rate": 9.31221052631579e-05,
      "loss": 0.3295,
      "step": 46155
    },
    {
      "epoch": 2.5846119386269457,
      "grad_norm": 1.059969425201416,
      "learning_rate": 9.312184210526316e-05,
      "loss": 0.361,
      "step": 46156
    },
    {
      "epoch": 2.584667935939075,
      "grad_norm": 0.9629393815994263,
      "learning_rate": 9.312157894736843e-05,
      "loss": 0.3669,
      "step": 46157
    },
    {
      "epoch": 2.5847239332512038,
      "grad_norm": 1.1319193840026855,
      "learning_rate": 9.312131578947369e-05,
      "loss": 0.3581,
      "step": 46158
    },
    {
      "epoch": 2.584779930563333,
      "grad_norm": 1.2861007452011108,
      "learning_rate": 9.312105263157895e-05,
      "loss": 0.5574,
      "step": 46159
    },
    {
      "epoch": 2.584835927875462,
      "grad_norm": 1.279954195022583,
      "learning_rate": 9.312078947368421e-05,
      "loss": 0.4105,
      "step": 46160
    },
    {
      "epoch": 2.5848919251875913,
      "grad_norm": 1.2599095106124878,
      "learning_rate": 9.312052631578948e-05,
      "loss": 0.5204,
      "step": 46161
    },
    {
      "epoch": 2.58494792249972,
      "grad_norm": 1.1875824928283691,
      "learning_rate": 9.312026315789474e-05,
      "loss": 0.3558,
      "step": 46162
    },
    {
      "epoch": 2.5850039198118493,
      "grad_norm": 1.1571764945983887,
      "learning_rate": 9.312e-05,
      "loss": 0.3278,
      "step": 46163
    },
    {
      "epoch": 2.585059917123978,
      "grad_norm": 1.2332196235656738,
      "learning_rate": 9.311973684210526e-05,
      "loss": 0.4497,
      "step": 46164
    },
    {
      "epoch": 2.5851159144361073,
      "grad_norm": 1.0362366437911987,
      "learning_rate": 9.311947368421054e-05,
      "loss": 0.3395,
      "step": 46165
    },
    {
      "epoch": 2.585171911748236,
      "grad_norm": 1.149551272392273,
      "learning_rate": 9.31192105263158e-05,
      "loss": 0.3847,
      "step": 46166
    },
    {
      "epoch": 2.5852279090603654,
      "grad_norm": 1.1010514497756958,
      "learning_rate": 9.311894736842106e-05,
      "loss": 0.3965,
      "step": 46167
    },
    {
      "epoch": 2.585283906372494,
      "grad_norm": 1.1411404609680176,
      "learning_rate": 9.311868421052632e-05,
      "loss": 0.4159,
      "step": 46168
    },
    {
      "epoch": 2.5853399036846234,
      "grad_norm": 1.258015513420105,
      "learning_rate": 9.311842105263159e-05,
      "loss": 0.4268,
      "step": 46169
    },
    {
      "epoch": 2.585395900996752,
      "grad_norm": 1.1511561870574951,
      "learning_rate": 9.311815789473685e-05,
      "loss": 0.4248,
      "step": 46170
    },
    {
      "epoch": 2.5854518983088814,
      "grad_norm": 1.1136687994003296,
      "learning_rate": 9.311789473684212e-05,
      "loss": 0.3705,
      "step": 46171
    },
    {
      "epoch": 2.58550789562101,
      "grad_norm": 1.070802927017212,
      "learning_rate": 9.311763157894737e-05,
      "loss": 0.3682,
      "step": 46172
    },
    {
      "epoch": 2.5855638929331395,
      "grad_norm": 1.627197265625,
      "learning_rate": 9.311736842105263e-05,
      "loss": 0.5591,
      "step": 46173
    },
    {
      "epoch": 2.585619890245268,
      "grad_norm": 1.195107102394104,
      "learning_rate": 9.31171052631579e-05,
      "loss": 0.437,
      "step": 46174
    },
    {
      "epoch": 2.585675887557397,
      "grad_norm": 1.1197826862335205,
      "learning_rate": 9.311684210526316e-05,
      "loss": 0.377,
      "step": 46175
    },
    {
      "epoch": 2.585731884869526,
      "grad_norm": 1.4083714485168457,
      "learning_rate": 9.311657894736843e-05,
      "loss": 0.3296,
      "step": 46176
    },
    {
      "epoch": 2.585787882181655,
      "grad_norm": 1.4512770175933838,
      "learning_rate": 9.311631578947368e-05,
      "loss": 0.3438,
      "step": 46177
    },
    {
      "epoch": 2.585843879493784,
      "grad_norm": 1.2137832641601562,
      "learning_rate": 9.311605263157895e-05,
      "loss": 0.3822,
      "step": 46178
    },
    {
      "epoch": 2.585899876805913,
      "grad_norm": 0.9990248084068298,
      "learning_rate": 9.311578947368421e-05,
      "loss": 0.3782,
      "step": 46179
    },
    {
      "epoch": 2.585955874118042,
      "grad_norm": 1.4267797470092773,
      "learning_rate": 9.311552631578949e-05,
      "loss": 0.4335,
      "step": 46180
    },
    {
      "epoch": 2.586011871430171,
      "grad_norm": 1.2732597589492798,
      "learning_rate": 9.311526315789473e-05,
      "loss": 0.5702,
      "step": 46181
    },
    {
      "epoch": 2.5860678687423,
      "grad_norm": 1.0875052213668823,
      "learning_rate": 9.3115e-05,
      "loss": 0.4149,
      "step": 46182
    },
    {
      "epoch": 2.5861238660544292,
      "grad_norm": 1.1585497856140137,
      "learning_rate": 9.311473684210527e-05,
      "loss": 0.4281,
      "step": 46183
    },
    {
      "epoch": 2.5861798633665583,
      "grad_norm": 1.0399203300476074,
      "learning_rate": 9.311447368421054e-05,
      "loss": 0.432,
      "step": 46184
    },
    {
      "epoch": 2.5862358606786873,
      "grad_norm": 1.38935387134552,
      "learning_rate": 9.31142105263158e-05,
      "loss": 0.4513,
      "step": 46185
    },
    {
      "epoch": 2.5862918579908163,
      "grad_norm": 1.4753776788711548,
      "learning_rate": 9.311394736842106e-05,
      "loss": 0.4661,
      "step": 46186
    },
    {
      "epoch": 2.5863478553029453,
      "grad_norm": 1.1953052282333374,
      "learning_rate": 9.311368421052632e-05,
      "loss": 0.4839,
      "step": 46187
    },
    {
      "epoch": 2.5864038526150743,
      "grad_norm": 1.0560799837112427,
      "learning_rate": 9.311342105263159e-05,
      "loss": 0.4439,
      "step": 46188
    },
    {
      "epoch": 2.5864598499272033,
      "grad_norm": 1.1305561065673828,
      "learning_rate": 9.311315789473685e-05,
      "loss": 0.403,
      "step": 46189
    },
    {
      "epoch": 2.5865158472393324,
      "grad_norm": 1.1195857524871826,
      "learning_rate": 9.311289473684211e-05,
      "loss": 0.3444,
      "step": 46190
    },
    {
      "epoch": 2.5865718445514614,
      "grad_norm": 1.049867033958435,
      "learning_rate": 9.311263157894737e-05,
      "loss": 0.3421,
      "step": 46191
    },
    {
      "epoch": 2.5866278418635904,
      "grad_norm": 1.3990519046783447,
      "learning_rate": 9.311236842105263e-05,
      "loss": 0.4355,
      "step": 46192
    },
    {
      "epoch": 2.5866838391757194,
      "grad_norm": 1.2467957735061646,
      "learning_rate": 9.31121052631579e-05,
      "loss": 0.3774,
      "step": 46193
    },
    {
      "epoch": 2.5867398364878484,
      "grad_norm": 1.0563324689865112,
      "learning_rate": 9.311184210526316e-05,
      "loss": 0.3194,
      "step": 46194
    },
    {
      "epoch": 2.5867958337999775,
      "grad_norm": 1.231909990310669,
      "learning_rate": 9.311157894736842e-05,
      "loss": 0.402,
      "step": 46195
    },
    {
      "epoch": 2.5868518311121065,
      "grad_norm": 1.0885509252548218,
      "learning_rate": 9.311131578947368e-05,
      "loss": 0.4035,
      "step": 46196
    },
    {
      "epoch": 2.5869078284242355,
      "grad_norm": 1.448405146598816,
      "learning_rate": 9.311105263157896e-05,
      "loss": 0.4949,
      "step": 46197
    },
    {
      "epoch": 2.5869638257363645,
      "grad_norm": 1.551013708114624,
      "learning_rate": 9.311078947368422e-05,
      "loss": 0.4266,
      "step": 46198
    },
    {
      "epoch": 2.5870198230484935,
      "grad_norm": 1.246457576751709,
      "learning_rate": 9.311052631578948e-05,
      "loss": 0.4558,
      "step": 46199
    },
    {
      "epoch": 2.5870758203606226,
      "grad_norm": 1.4119164943695068,
      "learning_rate": 9.311026315789473e-05,
      "loss": 0.3869,
      "step": 46200
    },
    {
      "epoch": 2.5871318176727516,
      "grad_norm": 1.4596666097640991,
      "learning_rate": 9.311000000000001e-05,
      "loss": 0.4522,
      "step": 46201
    },
    {
      "epoch": 2.5871878149848806,
      "grad_norm": 1.3969022035598755,
      "learning_rate": 9.310973684210527e-05,
      "loss": 0.5272,
      "step": 46202
    },
    {
      "epoch": 2.5872438122970096,
      "grad_norm": 1.2330166101455688,
      "learning_rate": 9.310947368421054e-05,
      "loss": 0.4256,
      "step": 46203
    },
    {
      "epoch": 2.5872998096091386,
      "grad_norm": 1.050488829612732,
      "learning_rate": 9.310921052631579e-05,
      "loss": 0.4009,
      "step": 46204
    },
    {
      "epoch": 2.5873558069212677,
      "grad_norm": 1.5017423629760742,
      "learning_rate": 9.310894736842106e-05,
      "loss": 0.4551,
      "step": 46205
    },
    {
      "epoch": 2.5874118042333967,
      "grad_norm": 1.1235839128494263,
      "learning_rate": 9.310868421052632e-05,
      "loss": 0.3989,
      "step": 46206
    },
    {
      "epoch": 2.5874678015455257,
      "grad_norm": 1.1039791107177734,
      "learning_rate": 9.310842105263158e-05,
      "loss": 0.3786,
      "step": 46207
    },
    {
      "epoch": 2.5875237988576547,
      "grad_norm": 1.1936520338058472,
      "learning_rate": 9.310815789473685e-05,
      "loss": 0.3841,
      "step": 46208
    },
    {
      "epoch": 2.5875797961697837,
      "grad_norm": 1.161730170249939,
      "learning_rate": 9.31078947368421e-05,
      "loss": 0.3911,
      "step": 46209
    },
    {
      "epoch": 2.5876357934819127,
      "grad_norm": 0.9859711527824402,
      "learning_rate": 9.310763157894737e-05,
      "loss": 0.3708,
      "step": 46210
    },
    {
      "epoch": 2.5876917907940418,
      "grad_norm": 1.176864743232727,
      "learning_rate": 9.310736842105263e-05,
      "loss": 0.4341,
      "step": 46211
    },
    {
      "epoch": 2.587747788106171,
      "grad_norm": 1.0070438385009766,
      "learning_rate": 9.31071052631579e-05,
      "loss": 0.318,
      "step": 46212
    },
    {
      "epoch": 2.5878037854183,
      "grad_norm": 1.3358042240142822,
      "learning_rate": 9.310684210526317e-05,
      "loss": 0.4245,
      "step": 46213
    },
    {
      "epoch": 2.587859782730429,
      "grad_norm": 1.1505718231201172,
      "learning_rate": 9.310657894736843e-05,
      "loss": 0.4625,
      "step": 46214
    },
    {
      "epoch": 2.587915780042558,
      "grad_norm": 1.3368526697158813,
      "learning_rate": 9.310631578947368e-05,
      "loss": 0.5161,
      "step": 46215
    },
    {
      "epoch": 2.587971777354687,
      "grad_norm": 1.2531532049179077,
      "learning_rate": 9.310605263157896e-05,
      "loss": 0.3739,
      "step": 46216
    },
    {
      "epoch": 2.588027774666816,
      "grad_norm": 1.1150315999984741,
      "learning_rate": 9.310578947368422e-05,
      "loss": 0.4028,
      "step": 46217
    },
    {
      "epoch": 2.588083771978945,
      "grad_norm": 1.269049882888794,
      "learning_rate": 9.310552631578948e-05,
      "loss": 0.3915,
      "step": 46218
    },
    {
      "epoch": 2.588139769291074,
      "grad_norm": 1.1772435903549194,
      "learning_rate": 9.310526315789474e-05,
      "loss": 0.3252,
      "step": 46219
    },
    {
      "epoch": 2.588195766603203,
      "grad_norm": 1.1974480152130127,
      "learning_rate": 9.310500000000001e-05,
      "loss": 0.4289,
      "step": 46220
    },
    {
      "epoch": 2.588251763915332,
      "grad_norm": 1.2302343845367432,
      "learning_rate": 9.310473684210527e-05,
      "loss": 0.379,
      "step": 46221
    },
    {
      "epoch": 2.588307761227461,
      "grad_norm": 1.2236132621765137,
      "learning_rate": 9.310447368421053e-05,
      "loss": 0.4447,
      "step": 46222
    },
    {
      "epoch": 2.58836375853959,
      "grad_norm": 1.1185723543167114,
      "learning_rate": 9.310421052631579e-05,
      "loss": 0.3863,
      "step": 46223
    },
    {
      "epoch": 2.588419755851719,
      "grad_norm": 1.0115447044372559,
      "learning_rate": 9.310394736842105e-05,
      "loss": 0.331,
      "step": 46224
    },
    {
      "epoch": 2.588475753163848,
      "grad_norm": 1.1912463903427124,
      "learning_rate": 9.310368421052632e-05,
      "loss": 0.3945,
      "step": 46225
    },
    {
      "epoch": 2.588531750475977,
      "grad_norm": 1.0816136598587036,
      "learning_rate": 9.310342105263158e-05,
      "loss": 0.3278,
      "step": 46226
    },
    {
      "epoch": 2.588587747788106,
      "grad_norm": 1.2076407670974731,
      "learning_rate": 9.310315789473684e-05,
      "loss": 0.403,
      "step": 46227
    },
    {
      "epoch": 2.588643745100235,
      "grad_norm": 1.2909005880355835,
      "learning_rate": 9.31028947368421e-05,
      "loss": 0.3652,
      "step": 46228
    },
    {
      "epoch": 2.588699742412364,
      "grad_norm": 1.0006271600723267,
      "learning_rate": 9.310263157894738e-05,
      "loss": 0.3912,
      "step": 46229
    },
    {
      "epoch": 2.588755739724493,
      "grad_norm": 1.245719313621521,
      "learning_rate": 9.310236842105264e-05,
      "loss": 0.373,
      "step": 46230
    },
    {
      "epoch": 2.588811737036622,
      "grad_norm": 1.5838624238967896,
      "learning_rate": 9.310210526315791e-05,
      "loss": 0.4055,
      "step": 46231
    },
    {
      "epoch": 2.588867734348751,
      "grad_norm": 1.228395938873291,
      "learning_rate": 9.310184210526315e-05,
      "loss": 0.5576,
      "step": 46232
    },
    {
      "epoch": 2.58892373166088,
      "grad_norm": 1.313234567642212,
      "learning_rate": 9.310157894736843e-05,
      "loss": 0.4007,
      "step": 46233
    },
    {
      "epoch": 2.588979728973009,
      "grad_norm": 1.3885167837142944,
      "learning_rate": 9.310131578947369e-05,
      "loss": 0.5922,
      "step": 46234
    },
    {
      "epoch": 2.5890357262851382,
      "grad_norm": 1.1877297163009644,
      "learning_rate": 9.310105263157896e-05,
      "loss": 0.4966,
      "step": 46235
    },
    {
      "epoch": 2.5890917235972672,
      "grad_norm": 1.1633297204971313,
      "learning_rate": 9.310078947368421e-05,
      "loss": 0.3257,
      "step": 46236
    },
    {
      "epoch": 2.5891477209093963,
      "grad_norm": 1.0928453207015991,
      "learning_rate": 9.310052631578948e-05,
      "loss": 0.2998,
      "step": 46237
    },
    {
      "epoch": 2.5892037182215253,
      "grad_norm": 1.0469619035720825,
      "learning_rate": 9.310026315789474e-05,
      "loss": 0.3869,
      "step": 46238
    },
    {
      "epoch": 2.5892597155336543,
      "grad_norm": 1.0504008531570435,
      "learning_rate": 9.310000000000001e-05,
      "loss": 0.3194,
      "step": 46239
    },
    {
      "epoch": 2.5893157128457833,
      "grad_norm": 1.1703003644943237,
      "learning_rate": 9.309973684210527e-05,
      "loss": 0.3038,
      "step": 46240
    },
    {
      "epoch": 2.5893717101579123,
      "grad_norm": 1.3680009841918945,
      "learning_rate": 9.309947368421052e-05,
      "loss": 0.4307,
      "step": 46241
    },
    {
      "epoch": 2.5894277074700414,
      "grad_norm": 1.7781453132629395,
      "learning_rate": 9.309921052631579e-05,
      "loss": 0.3945,
      "step": 46242
    },
    {
      "epoch": 2.5894837047821704,
      "grad_norm": 1.120333194732666,
      "learning_rate": 9.309894736842105e-05,
      "loss": 0.3481,
      "step": 46243
    },
    {
      "epoch": 2.5895397020942994,
      "grad_norm": 1.2128700017929077,
      "learning_rate": 9.309868421052633e-05,
      "loss": 0.3608,
      "step": 46244
    },
    {
      "epoch": 2.5895956994064284,
      "grad_norm": 1.1390490531921387,
      "learning_rate": 9.309842105263159e-05,
      "loss": 0.497,
      "step": 46245
    },
    {
      "epoch": 2.5896516967185574,
      "grad_norm": 1.183670997619629,
      "learning_rate": 9.309815789473684e-05,
      "loss": 0.411,
      "step": 46246
    },
    {
      "epoch": 2.5897076940306865,
      "grad_norm": 1.2987509965896606,
      "learning_rate": 9.30978947368421e-05,
      "loss": 0.3711,
      "step": 46247
    },
    {
      "epoch": 2.5897636913428155,
      "grad_norm": 1.0237271785736084,
      "learning_rate": 9.309763157894738e-05,
      "loss": 0.3306,
      "step": 46248
    },
    {
      "epoch": 2.5898196886549445,
      "grad_norm": 1.253071904182434,
      "learning_rate": 9.309736842105264e-05,
      "loss": 0.4109,
      "step": 46249
    },
    {
      "epoch": 2.5898756859670735,
      "grad_norm": 1.2068551778793335,
      "learning_rate": 9.30971052631579e-05,
      "loss": 0.3682,
      "step": 46250
    },
    {
      "epoch": 2.5899316832792025,
      "grad_norm": 0.9989849925041199,
      "learning_rate": 9.309684210526316e-05,
      "loss": 0.3073,
      "step": 46251
    },
    {
      "epoch": 2.5899876805913316,
      "grad_norm": 1.171868085861206,
      "learning_rate": 9.309657894736843e-05,
      "loss": 0.3735,
      "step": 46252
    },
    {
      "epoch": 2.5900436779034606,
      "grad_norm": 1.1079397201538086,
      "learning_rate": 9.309631578947369e-05,
      "loss": 0.4156,
      "step": 46253
    },
    {
      "epoch": 2.5900996752155896,
      "grad_norm": 1.6420271396636963,
      "learning_rate": 9.309605263157895e-05,
      "loss": 0.4815,
      "step": 46254
    },
    {
      "epoch": 2.5901556725277186,
      "grad_norm": 1.2622206211090088,
      "learning_rate": 9.309578947368421e-05,
      "loss": 0.5426,
      "step": 46255
    },
    {
      "epoch": 2.5902116698398476,
      "grad_norm": 1.3164116144180298,
      "learning_rate": 9.309552631578948e-05,
      "loss": 0.3236,
      "step": 46256
    },
    {
      "epoch": 2.5902676671519766,
      "grad_norm": 1.7366224527359009,
      "learning_rate": 9.309526315789474e-05,
      "loss": 0.4633,
      "step": 46257
    },
    {
      "epoch": 2.5903236644641057,
      "grad_norm": 0.9842000603675842,
      "learning_rate": 9.309500000000002e-05,
      "loss": 0.2767,
      "step": 46258
    },
    {
      "epoch": 2.5903796617762347,
      "grad_norm": 1.1537973880767822,
      "learning_rate": 9.309473684210526e-05,
      "loss": 0.3579,
      "step": 46259
    },
    {
      "epoch": 2.5904356590883637,
      "grad_norm": 1.2546542882919312,
      "learning_rate": 9.309447368421052e-05,
      "loss": 0.4915,
      "step": 46260
    },
    {
      "epoch": 2.5904916564004927,
      "grad_norm": 1.2191046476364136,
      "learning_rate": 9.30942105263158e-05,
      "loss": 0.4695,
      "step": 46261
    },
    {
      "epoch": 2.5905476537126217,
      "grad_norm": 1.3442881107330322,
      "learning_rate": 9.309394736842105e-05,
      "loss": 0.4343,
      "step": 46262
    },
    {
      "epoch": 2.5906036510247508,
      "grad_norm": 1.3232883214950562,
      "learning_rate": 9.309368421052633e-05,
      "loss": 0.4336,
      "step": 46263
    },
    {
      "epoch": 2.59065964833688,
      "grad_norm": 0.9567542672157288,
      "learning_rate": 9.309342105263157e-05,
      "loss": 0.3133,
      "step": 46264
    },
    {
      "epoch": 2.590715645649009,
      "grad_norm": 1.2464120388031006,
      "learning_rate": 9.309315789473685e-05,
      "loss": 0.4245,
      "step": 46265
    },
    {
      "epoch": 2.590771642961138,
      "grad_norm": 1.1727780103683472,
      "learning_rate": 9.309289473684211e-05,
      "loss": 0.442,
      "step": 46266
    },
    {
      "epoch": 2.590827640273267,
      "grad_norm": 1.4074660539627075,
      "learning_rate": 9.309263157894738e-05,
      "loss": 0.4253,
      "step": 46267
    },
    {
      "epoch": 2.590883637585396,
      "grad_norm": 1.0359115600585938,
      "learning_rate": 9.309236842105264e-05,
      "loss": 0.2886,
      "step": 46268
    },
    {
      "epoch": 2.590939634897525,
      "grad_norm": 1.2708663940429688,
      "learning_rate": 9.30921052631579e-05,
      "loss": 0.4186,
      "step": 46269
    },
    {
      "epoch": 2.590995632209654,
      "grad_norm": 1.2987852096557617,
      "learning_rate": 9.309184210526316e-05,
      "loss": 0.5822,
      "step": 46270
    },
    {
      "epoch": 2.591051629521783,
      "grad_norm": 1.1190284490585327,
      "learning_rate": 9.309157894736843e-05,
      "loss": 0.6175,
      "step": 46271
    },
    {
      "epoch": 2.591107626833912,
      "grad_norm": 1.1067454814910889,
      "learning_rate": 9.309131578947369e-05,
      "loss": 0.3865,
      "step": 46272
    },
    {
      "epoch": 2.591163624146041,
      "grad_norm": 1.2553926706314087,
      "learning_rate": 9.309105263157895e-05,
      "loss": 0.4105,
      "step": 46273
    },
    {
      "epoch": 2.59121962145817,
      "grad_norm": 1.1765193939208984,
      "learning_rate": 9.309078947368421e-05,
      "loss": 0.4001,
      "step": 46274
    },
    {
      "epoch": 2.591275618770299,
      "grad_norm": 1.1803947687149048,
      "learning_rate": 9.309052631578949e-05,
      "loss": 0.3658,
      "step": 46275
    },
    {
      "epoch": 2.591331616082428,
      "grad_norm": 1.1997201442718506,
      "learning_rate": 9.309026315789475e-05,
      "loss": 0.5434,
      "step": 46276
    },
    {
      "epoch": 2.591387613394557,
      "grad_norm": 1.0450609922409058,
      "learning_rate": 9.309e-05,
      "loss": 0.3593,
      "step": 46277
    },
    {
      "epoch": 2.591443610706686,
      "grad_norm": 1.0744925737380981,
      "learning_rate": 9.308973684210526e-05,
      "loss": 0.4222,
      "step": 46278
    },
    {
      "epoch": 2.591499608018815,
      "grad_norm": 0.9694228172302246,
      "learning_rate": 9.308947368421052e-05,
      "loss": 0.339,
      "step": 46279
    },
    {
      "epoch": 2.591555605330944,
      "grad_norm": 0.9629693627357483,
      "learning_rate": 9.30892105263158e-05,
      "loss": 0.3697,
      "step": 46280
    },
    {
      "epoch": 2.591611602643073,
      "grad_norm": 1.0900285243988037,
      "learning_rate": 9.308894736842106e-05,
      "loss": 0.4263,
      "step": 46281
    },
    {
      "epoch": 2.591667599955202,
      "grad_norm": 1.033509612083435,
      "learning_rate": 9.308868421052632e-05,
      "loss": 0.2765,
      "step": 46282
    },
    {
      "epoch": 2.591723597267331,
      "grad_norm": 1.081179141998291,
      "learning_rate": 9.308842105263158e-05,
      "loss": 0.2802,
      "step": 46283
    },
    {
      "epoch": 2.59177959457946,
      "grad_norm": 1.3849984407424927,
      "learning_rate": 9.308815789473685e-05,
      "loss": 0.4324,
      "step": 46284
    },
    {
      "epoch": 2.591835591891589,
      "grad_norm": 1.608033299446106,
      "learning_rate": 9.308789473684211e-05,
      "loss": 0.335,
      "step": 46285
    },
    {
      "epoch": 2.591891589203718,
      "grad_norm": 2.0281853675842285,
      "learning_rate": 9.308763157894737e-05,
      "loss": 0.3771,
      "step": 46286
    },
    {
      "epoch": 2.5919475865158472,
      "grad_norm": 1.3969900608062744,
      "learning_rate": 9.308736842105263e-05,
      "loss": 0.4417,
      "step": 46287
    },
    {
      "epoch": 2.5920035838279762,
      "grad_norm": 1.2181472778320312,
      "learning_rate": 9.30871052631579e-05,
      "loss": 0.4052,
      "step": 46288
    },
    {
      "epoch": 2.5920595811401053,
      "grad_norm": 0.9348074197769165,
      "learning_rate": 9.308684210526316e-05,
      "loss": 0.2563,
      "step": 46289
    },
    {
      "epoch": 2.5921155784522343,
      "grad_norm": 1.2464137077331543,
      "learning_rate": 9.308657894736844e-05,
      "loss": 0.545,
      "step": 46290
    },
    {
      "epoch": 2.5921715757643633,
      "grad_norm": 0.9846921563148499,
      "learning_rate": 9.308631578947368e-05,
      "loss": 0.3108,
      "step": 46291
    },
    {
      "epoch": 2.5922275730764923,
      "grad_norm": 1.2812780141830444,
      "learning_rate": 9.308605263157896e-05,
      "loss": 0.4695,
      "step": 46292
    },
    {
      "epoch": 2.5922835703886213,
      "grad_norm": 1.041433572769165,
      "learning_rate": 9.308578947368421e-05,
      "loss": 0.2697,
      "step": 46293
    },
    {
      "epoch": 2.5923395677007504,
      "grad_norm": 1.2597488164901733,
      "learning_rate": 9.308552631578947e-05,
      "loss": 0.4214,
      "step": 46294
    },
    {
      "epoch": 2.5923955650128794,
      "grad_norm": 1.0989595651626587,
      "learning_rate": 9.308526315789475e-05,
      "loss": 0.5186,
      "step": 46295
    },
    {
      "epoch": 2.5924515623250084,
      "grad_norm": 1.103491187095642,
      "learning_rate": 9.3085e-05,
      "loss": 0.3612,
      "step": 46296
    },
    {
      "epoch": 2.5925075596371374,
      "grad_norm": 1.0412403345108032,
      "learning_rate": 9.308473684210527e-05,
      "loss": 0.3184,
      "step": 46297
    },
    {
      "epoch": 2.5925635569492664,
      "grad_norm": 1.0847171545028687,
      "learning_rate": 9.308447368421053e-05,
      "loss": 0.3547,
      "step": 46298
    },
    {
      "epoch": 2.5926195542613955,
      "grad_norm": 1.170149803161621,
      "learning_rate": 9.30842105263158e-05,
      "loss": 0.4565,
      "step": 46299
    },
    {
      "epoch": 2.5926755515735245,
      "grad_norm": 1.2003085613250732,
      "learning_rate": 9.308394736842106e-05,
      "loss": 0.5156,
      "step": 46300
    },
    {
      "epoch": 2.5927315488856535,
      "grad_norm": 1.15707528591156,
      "learning_rate": 9.308368421052632e-05,
      "loss": 0.4768,
      "step": 46301
    },
    {
      "epoch": 2.5927875461977825,
      "grad_norm": 1.2563356161117554,
      "learning_rate": 9.308342105263158e-05,
      "loss": 0.4962,
      "step": 46302
    },
    {
      "epoch": 2.5928435435099115,
      "grad_norm": 1.0945630073547363,
      "learning_rate": 9.308315789473685e-05,
      "loss": 0.3812,
      "step": 46303
    },
    {
      "epoch": 2.5928995408220405,
      "grad_norm": 0.9414119720458984,
      "learning_rate": 9.308289473684211e-05,
      "loss": 0.2923,
      "step": 46304
    },
    {
      "epoch": 2.5929555381341696,
      "grad_norm": 1.2058321237564087,
      "learning_rate": 9.308263157894737e-05,
      "loss": 0.463,
      "step": 46305
    },
    {
      "epoch": 2.5930115354462986,
      "grad_norm": 1.1503084897994995,
      "learning_rate": 9.308236842105263e-05,
      "loss": 0.3851,
      "step": 46306
    },
    {
      "epoch": 2.5930675327584276,
      "grad_norm": 1.205010175704956,
      "learning_rate": 9.30821052631579e-05,
      "loss": 0.4971,
      "step": 46307
    },
    {
      "epoch": 2.5931235300705566,
      "grad_norm": 1.0483094453811646,
      "learning_rate": 9.308184210526316e-05,
      "loss": 0.3803,
      "step": 46308
    },
    {
      "epoch": 2.5931795273826856,
      "grad_norm": 1.314060926437378,
      "learning_rate": 9.308157894736842e-05,
      "loss": 0.4673,
      "step": 46309
    },
    {
      "epoch": 2.5932355246948147,
      "grad_norm": 1.3096795082092285,
      "learning_rate": 9.308131578947368e-05,
      "loss": 0.3963,
      "step": 46310
    },
    {
      "epoch": 2.5932915220069437,
      "grad_norm": 1.3893641233444214,
      "learning_rate": 9.308105263157894e-05,
      "loss": 0.4069,
      "step": 46311
    },
    {
      "epoch": 2.5933475193190727,
      "grad_norm": 1.2928149700164795,
      "learning_rate": 9.308078947368422e-05,
      "loss": 0.4374,
      "step": 46312
    },
    {
      "epoch": 2.5934035166312017,
      "grad_norm": 1.154284954071045,
      "learning_rate": 9.308052631578948e-05,
      "loss": 0.4114,
      "step": 46313
    },
    {
      "epoch": 2.5934595139433307,
      "grad_norm": 1.2459124326705933,
      "learning_rate": 9.308026315789474e-05,
      "loss": 0.3966,
      "step": 46314
    },
    {
      "epoch": 2.5935155112554598,
      "grad_norm": 1.0978296995162964,
      "learning_rate": 9.308e-05,
      "loss": 0.4238,
      "step": 46315
    },
    {
      "epoch": 2.593571508567589,
      "grad_norm": 1.4941835403442383,
      "learning_rate": 9.307973684210527e-05,
      "loss": 0.3848,
      "step": 46316
    },
    {
      "epoch": 2.593627505879718,
      "grad_norm": 1.096929669380188,
      "learning_rate": 9.307947368421053e-05,
      "loss": 0.3566,
      "step": 46317
    },
    {
      "epoch": 2.593683503191847,
      "grad_norm": 1.2642258405685425,
      "learning_rate": 9.30792105263158e-05,
      "loss": 0.5874,
      "step": 46318
    },
    {
      "epoch": 2.593739500503976,
      "grad_norm": 1.2297660112380981,
      "learning_rate": 9.307894736842105e-05,
      "loss": 0.4101,
      "step": 46319
    },
    {
      "epoch": 2.593795497816105,
      "grad_norm": 1.0441217422485352,
      "learning_rate": 9.307868421052632e-05,
      "loss": 0.3274,
      "step": 46320
    },
    {
      "epoch": 2.593851495128234,
      "grad_norm": 1.0500421524047852,
      "learning_rate": 9.307842105263158e-05,
      "loss": 0.4838,
      "step": 46321
    },
    {
      "epoch": 2.593907492440363,
      "grad_norm": 1.1367409229278564,
      "learning_rate": 9.307815789473686e-05,
      "loss": 0.3975,
      "step": 46322
    },
    {
      "epoch": 2.593963489752492,
      "grad_norm": 1.3118699789047241,
      "learning_rate": 9.307789473684211e-05,
      "loss": 0.3542,
      "step": 46323
    },
    {
      "epoch": 2.594019487064621,
      "grad_norm": 1.1071178913116455,
      "learning_rate": 9.307763157894737e-05,
      "loss": 0.3284,
      "step": 46324
    },
    {
      "epoch": 2.59407548437675,
      "grad_norm": 1.0018930435180664,
      "learning_rate": 9.307736842105263e-05,
      "loss": 0.2769,
      "step": 46325
    },
    {
      "epoch": 2.594131481688879,
      "grad_norm": 1.2303153276443481,
      "learning_rate": 9.307710526315791e-05,
      "loss": 0.329,
      "step": 46326
    },
    {
      "epoch": 2.594187479001008,
      "grad_norm": 1.1183476448059082,
      "learning_rate": 9.307684210526317e-05,
      "loss": 0.3309,
      "step": 46327
    },
    {
      "epoch": 2.594243476313137,
      "grad_norm": 1.2909375429153442,
      "learning_rate": 9.307657894736841e-05,
      "loss": 0.3618,
      "step": 46328
    },
    {
      "epoch": 2.594299473625266,
      "grad_norm": 1.3322277069091797,
      "learning_rate": 9.307631578947369e-05,
      "loss": 0.4223,
      "step": 46329
    },
    {
      "epoch": 2.594355470937395,
      "grad_norm": 1.0959221124649048,
      "learning_rate": 9.307605263157895e-05,
      "loss": 0.4415,
      "step": 46330
    },
    {
      "epoch": 2.594411468249524,
      "grad_norm": 1.1785094738006592,
      "learning_rate": 9.307578947368422e-05,
      "loss": 0.4333,
      "step": 46331
    },
    {
      "epoch": 2.594467465561653,
      "grad_norm": 1.228363275527954,
      "learning_rate": 9.307552631578948e-05,
      "loss": 0.4117,
      "step": 46332
    },
    {
      "epoch": 2.594523462873782,
      "grad_norm": 1.4216464757919312,
      "learning_rate": 9.307526315789474e-05,
      "loss": 0.3942,
      "step": 46333
    },
    {
      "epoch": 2.594579460185911,
      "grad_norm": 1.234973669052124,
      "learning_rate": 9.3075e-05,
      "loss": 0.3639,
      "step": 46334
    },
    {
      "epoch": 2.59463545749804,
      "grad_norm": 1.4586713314056396,
      "learning_rate": 9.307473684210527e-05,
      "loss": 0.3931,
      "step": 46335
    },
    {
      "epoch": 2.594691454810169,
      "grad_norm": 1.3273762464523315,
      "learning_rate": 9.307447368421053e-05,
      "loss": 0.411,
      "step": 46336
    },
    {
      "epoch": 2.594747452122298,
      "grad_norm": 1.116397500038147,
      "learning_rate": 9.307421052631579e-05,
      "loss": 0.4085,
      "step": 46337
    },
    {
      "epoch": 2.594803449434427,
      "grad_norm": 1.081879734992981,
      "learning_rate": 9.307394736842105e-05,
      "loss": 0.3788,
      "step": 46338
    },
    {
      "epoch": 2.594859446746556,
      "grad_norm": 1.1855541467666626,
      "learning_rate": 9.307368421052632e-05,
      "loss": 0.3918,
      "step": 46339
    },
    {
      "epoch": 2.5949154440586852,
      "grad_norm": 1.2762482166290283,
      "learning_rate": 9.307342105263158e-05,
      "loss": 0.4214,
      "step": 46340
    },
    {
      "epoch": 2.5949714413708143,
      "grad_norm": 1.2305614948272705,
      "learning_rate": 9.307315789473684e-05,
      "loss": 0.3271,
      "step": 46341
    },
    {
      "epoch": 2.5950274386829433,
      "grad_norm": 1.743653655052185,
      "learning_rate": 9.30728947368421e-05,
      "loss": 0.7206,
      "step": 46342
    },
    {
      "epoch": 2.5950834359950723,
      "grad_norm": 1.3808716535568237,
      "learning_rate": 9.307263157894738e-05,
      "loss": 0.5562,
      "step": 46343
    },
    {
      "epoch": 2.5951394333072013,
      "grad_norm": 1.1509453058242798,
      "learning_rate": 9.307236842105264e-05,
      "loss": 0.3664,
      "step": 46344
    },
    {
      "epoch": 2.5951954306193303,
      "grad_norm": 1.0026576519012451,
      "learning_rate": 9.30721052631579e-05,
      "loss": 0.3386,
      "step": 46345
    },
    {
      "epoch": 2.5952514279314594,
      "grad_norm": 1.435781717300415,
      "learning_rate": 9.307184210526316e-05,
      "loss": 0.5191,
      "step": 46346
    },
    {
      "epoch": 2.5953074252435884,
      "grad_norm": 1.3051142692565918,
      "learning_rate": 9.307157894736842e-05,
      "loss": 0.6892,
      "step": 46347
    },
    {
      "epoch": 2.5953634225557174,
      "grad_norm": 1.2356524467468262,
      "learning_rate": 9.307131578947369e-05,
      "loss": 0.3988,
      "step": 46348
    },
    {
      "epoch": 2.5954194198678464,
      "grad_norm": 1.0640677213668823,
      "learning_rate": 9.307105263157895e-05,
      "loss": 0.4973,
      "step": 46349
    },
    {
      "epoch": 2.5954754171799754,
      "grad_norm": 1.3630106449127197,
      "learning_rate": 9.307078947368422e-05,
      "loss": 0.5907,
      "step": 46350
    },
    {
      "epoch": 2.5955314144921044,
      "grad_norm": 1.2732597589492798,
      "learning_rate": 9.307052631578947e-05,
      "loss": 0.4914,
      "step": 46351
    },
    {
      "epoch": 2.5955874118042335,
      "grad_norm": 1.0415436029434204,
      "learning_rate": 9.307026315789474e-05,
      "loss": 0.4629,
      "step": 46352
    },
    {
      "epoch": 2.5956434091163625,
      "grad_norm": 1.2142295837402344,
      "learning_rate": 9.307e-05,
      "loss": 0.4085,
      "step": 46353
    },
    {
      "epoch": 2.5956994064284915,
      "grad_norm": 1.2304552793502808,
      "learning_rate": 9.306973684210527e-05,
      "loss": 0.3546,
      "step": 46354
    },
    {
      "epoch": 2.5957554037406205,
      "grad_norm": 1.071213722229004,
      "learning_rate": 9.306947368421053e-05,
      "loss": 0.3885,
      "step": 46355
    },
    {
      "epoch": 2.5958114010527495,
      "grad_norm": 1.065073013305664,
      "learning_rate": 9.30692105263158e-05,
      "loss": 0.3458,
      "step": 46356
    },
    {
      "epoch": 2.5958673983648786,
      "grad_norm": 5.032269477844238,
      "learning_rate": 9.306894736842105e-05,
      "loss": 0.3571,
      "step": 46357
    },
    {
      "epoch": 2.5959233956770076,
      "grad_norm": 1.0765151977539062,
      "learning_rate": 9.306868421052633e-05,
      "loss": 0.4723,
      "step": 46358
    },
    {
      "epoch": 2.5959793929891366,
      "grad_norm": 1.0870074033737183,
      "learning_rate": 9.306842105263159e-05,
      "loss": 0.504,
      "step": 46359
    },
    {
      "epoch": 2.5960353903012656,
      "grad_norm": 1.2023396492004395,
      "learning_rate": 9.306815789473685e-05,
      "loss": 0.3668,
      "step": 46360
    },
    {
      "epoch": 2.5960913876133946,
      "grad_norm": 1.0600939989089966,
      "learning_rate": 9.30678947368421e-05,
      "loss": 0.4239,
      "step": 46361
    },
    {
      "epoch": 2.5961473849255237,
      "grad_norm": 3.0582919120788574,
      "learning_rate": 9.306763157894737e-05,
      "loss": 0.3892,
      "step": 46362
    },
    {
      "epoch": 2.5962033822376527,
      "grad_norm": 1.0472710132598877,
      "learning_rate": 9.306736842105264e-05,
      "loss": 0.4776,
      "step": 46363
    },
    {
      "epoch": 2.5962593795497817,
      "grad_norm": 1.1434522867202759,
      "learning_rate": 9.30671052631579e-05,
      "loss": 0.5073,
      "step": 46364
    },
    {
      "epoch": 2.5963153768619107,
      "grad_norm": 1.149087905883789,
      "learning_rate": 9.306684210526316e-05,
      "loss": 0.5352,
      "step": 46365
    },
    {
      "epoch": 2.5963713741740397,
      "grad_norm": 1.2497172355651855,
      "learning_rate": 9.306657894736842e-05,
      "loss": 0.429,
      "step": 46366
    },
    {
      "epoch": 2.5964273714861688,
      "grad_norm": 1.3501276969909668,
      "learning_rate": 9.306631578947369e-05,
      "loss": 0.3541,
      "step": 46367
    },
    {
      "epoch": 2.5964833687982978,
      "grad_norm": 1.020113229751587,
      "learning_rate": 9.306605263157895e-05,
      "loss": 0.3417,
      "step": 46368
    },
    {
      "epoch": 2.596539366110427,
      "grad_norm": 1.4978958368301392,
      "learning_rate": 9.306578947368421e-05,
      "loss": 0.5658,
      "step": 46369
    },
    {
      "epoch": 2.596595363422556,
      "grad_norm": 1.1411269903182983,
      "learning_rate": 9.306552631578947e-05,
      "loss": 0.4128,
      "step": 46370
    },
    {
      "epoch": 2.596651360734685,
      "grad_norm": 1.3018317222595215,
      "learning_rate": 9.306526315789474e-05,
      "loss": 0.4062,
      "step": 46371
    },
    {
      "epoch": 2.596707358046814,
      "grad_norm": 1.0208609104156494,
      "learning_rate": 9.3065e-05,
      "loss": 0.3452,
      "step": 46372
    },
    {
      "epoch": 2.596763355358943,
      "grad_norm": 1.126666784286499,
      "learning_rate": 9.306473684210528e-05,
      "loss": 0.3561,
      "step": 46373
    },
    {
      "epoch": 2.596819352671072,
      "grad_norm": 1.3486653566360474,
      "learning_rate": 9.306447368421052e-05,
      "loss": 0.4562,
      "step": 46374
    },
    {
      "epoch": 2.596875349983201,
      "grad_norm": 1.3658181428909302,
      "learning_rate": 9.30642105263158e-05,
      "loss": 0.3916,
      "step": 46375
    },
    {
      "epoch": 2.59693134729533,
      "grad_norm": 1.173493504524231,
      "learning_rate": 9.306394736842106e-05,
      "loss": 0.363,
      "step": 46376
    },
    {
      "epoch": 2.596987344607459,
      "grad_norm": 1.268128752708435,
      "learning_rate": 9.306368421052633e-05,
      "loss": 0.5235,
      "step": 46377
    },
    {
      "epoch": 2.597043341919588,
      "grad_norm": 1.022345781326294,
      "learning_rate": 9.306342105263159e-05,
      "loss": 0.3314,
      "step": 46378
    },
    {
      "epoch": 2.597099339231717,
      "grad_norm": 1.3354579210281372,
      "learning_rate": 9.306315789473685e-05,
      "loss": 0.5914,
      "step": 46379
    },
    {
      "epoch": 2.597155336543846,
      "grad_norm": 1.1834393739700317,
      "learning_rate": 9.306289473684211e-05,
      "loss": 0.3164,
      "step": 46380
    },
    {
      "epoch": 2.597211333855975,
      "grad_norm": 1.0589256286621094,
      "learning_rate": 9.306263157894737e-05,
      "loss": 0.3791,
      "step": 46381
    },
    {
      "epoch": 2.597267331168104,
      "grad_norm": 1.0989223718643188,
      "learning_rate": 9.306236842105264e-05,
      "loss": 0.406,
      "step": 46382
    },
    {
      "epoch": 2.597323328480233,
      "grad_norm": 1.840376615524292,
      "learning_rate": 9.306210526315789e-05,
      "loss": 0.4698,
      "step": 46383
    },
    {
      "epoch": 2.597379325792362,
      "grad_norm": 1.2043107748031616,
      "learning_rate": 9.306184210526316e-05,
      "loss": 0.3954,
      "step": 46384
    },
    {
      "epoch": 2.597435323104491,
      "grad_norm": 1.103837013244629,
      "learning_rate": 9.306157894736842e-05,
      "loss": 0.3648,
      "step": 46385
    },
    {
      "epoch": 2.59749132041662,
      "grad_norm": 1.0096098184585571,
      "learning_rate": 9.30613157894737e-05,
      "loss": 0.346,
      "step": 46386
    },
    {
      "epoch": 2.597547317728749,
      "grad_norm": 1.1829688549041748,
      "learning_rate": 9.306105263157895e-05,
      "loss": 0.3753,
      "step": 46387
    },
    {
      "epoch": 2.597603315040878,
      "grad_norm": 1.4263122081756592,
      "learning_rate": 9.306078947368421e-05,
      "loss": 0.3576,
      "step": 46388
    },
    {
      "epoch": 2.597659312353007,
      "grad_norm": 1.3190265893936157,
      "learning_rate": 9.306052631578947e-05,
      "loss": 0.3338,
      "step": 46389
    },
    {
      "epoch": 2.597715309665136,
      "grad_norm": 0.9917435050010681,
      "learning_rate": 9.306026315789475e-05,
      "loss": 0.3164,
      "step": 46390
    },
    {
      "epoch": 2.597771306977265,
      "grad_norm": 1.0683341026306152,
      "learning_rate": 9.306000000000001e-05,
      "loss": 0.444,
      "step": 46391
    },
    {
      "epoch": 2.5978273042893942,
      "grad_norm": 1.2895150184631348,
      "learning_rate": 9.305973684210527e-05,
      "loss": 0.3916,
      "step": 46392
    },
    {
      "epoch": 2.5978833016015233,
      "grad_norm": 0.968153178691864,
      "learning_rate": 9.305947368421053e-05,
      "loss": 0.3465,
      "step": 46393
    },
    {
      "epoch": 2.5979392989136523,
      "grad_norm": 1.0791574716567993,
      "learning_rate": 9.30592105263158e-05,
      "loss": 0.3661,
      "step": 46394
    },
    {
      "epoch": 2.5979952962257813,
      "grad_norm": 1.0966107845306396,
      "learning_rate": 9.305894736842106e-05,
      "loss": 0.3442,
      "step": 46395
    },
    {
      "epoch": 2.5980512935379103,
      "grad_norm": 1.0598368644714355,
      "learning_rate": 9.305868421052632e-05,
      "loss": 0.2701,
      "step": 46396
    },
    {
      "epoch": 2.5981072908500393,
      "grad_norm": 1.027726173400879,
      "learning_rate": 9.305842105263158e-05,
      "loss": 0.2789,
      "step": 46397
    },
    {
      "epoch": 2.5981632881621683,
      "grad_norm": 1.2512030601501465,
      "learning_rate": 9.305815789473684e-05,
      "loss": 0.4415,
      "step": 46398
    },
    {
      "epoch": 2.5982192854742974,
      "grad_norm": 1.259636640548706,
      "learning_rate": 9.305789473684211e-05,
      "loss": 0.4764,
      "step": 46399
    },
    {
      "epoch": 2.5982752827864264,
      "grad_norm": 1.3739681243896484,
      "learning_rate": 9.305763157894737e-05,
      "loss": 0.4599,
      "step": 46400
    },
    {
      "epoch": 2.5983312800985554,
      "grad_norm": 1.242702603340149,
      "learning_rate": 9.305736842105263e-05,
      "loss": 0.4708,
      "step": 46401
    },
    {
      "epoch": 2.5983872774106844,
      "grad_norm": 1.1186217069625854,
      "learning_rate": 9.305710526315789e-05,
      "loss": 0.4253,
      "step": 46402
    },
    {
      "epoch": 2.5984432747228134,
      "grad_norm": 1.3703936338424683,
      "learning_rate": 9.305684210526316e-05,
      "loss": 0.4427,
      "step": 46403
    },
    {
      "epoch": 2.5984992720349425,
      "grad_norm": 1.2901157140731812,
      "learning_rate": 9.305657894736842e-05,
      "loss": 0.4259,
      "step": 46404
    },
    {
      "epoch": 2.5985552693470715,
      "grad_norm": 1.2709455490112305,
      "learning_rate": 9.30563157894737e-05,
      "loss": 0.4528,
      "step": 46405
    },
    {
      "epoch": 2.5986112666592005,
      "grad_norm": 1.4229942560195923,
      "learning_rate": 9.305605263157894e-05,
      "loss": 0.5051,
      "step": 46406
    },
    {
      "epoch": 2.5986672639713295,
      "grad_norm": 0.9639158248901367,
      "learning_rate": 9.305578947368422e-05,
      "loss": 0.3377,
      "step": 46407
    },
    {
      "epoch": 2.5987232612834585,
      "grad_norm": 1.2738722562789917,
      "learning_rate": 9.305552631578948e-05,
      "loss": 0.6525,
      "step": 46408
    },
    {
      "epoch": 2.5987792585955876,
      "grad_norm": 1.1387035846710205,
      "learning_rate": 9.305526315789475e-05,
      "loss": 0.4028,
      "step": 46409
    },
    {
      "epoch": 2.5988352559077166,
      "grad_norm": 1.1342895030975342,
      "learning_rate": 9.305500000000001e-05,
      "loss": 0.3728,
      "step": 46410
    },
    {
      "epoch": 2.5988912532198456,
      "grad_norm": 1.2227965593338013,
      "learning_rate": 9.305473684210527e-05,
      "loss": 0.4896,
      "step": 46411
    },
    {
      "epoch": 2.5989472505319746,
      "grad_norm": 1.304675817489624,
      "learning_rate": 9.305447368421053e-05,
      "loss": 0.4238,
      "step": 46412
    },
    {
      "epoch": 2.5990032478441036,
      "grad_norm": 1.3673406839370728,
      "learning_rate": 9.30542105263158e-05,
      "loss": 0.4155,
      "step": 46413
    },
    {
      "epoch": 2.5990592451562327,
      "grad_norm": 1.2016870975494385,
      "learning_rate": 9.305394736842106e-05,
      "loss": 0.4107,
      "step": 46414
    },
    {
      "epoch": 2.5991152424683617,
      "grad_norm": 1.2177479267120361,
      "learning_rate": 9.305368421052632e-05,
      "loss": 0.3277,
      "step": 46415
    },
    {
      "epoch": 2.5991712397804907,
      "grad_norm": 1.3576984405517578,
      "learning_rate": 9.305342105263158e-05,
      "loss": 0.4359,
      "step": 46416
    },
    {
      "epoch": 2.5992272370926197,
      "grad_norm": 1.2905004024505615,
      "learning_rate": 9.305315789473684e-05,
      "loss": 0.335,
      "step": 46417
    },
    {
      "epoch": 2.5992832344047487,
      "grad_norm": 1.2379571199417114,
      "learning_rate": 9.305289473684211e-05,
      "loss": 0.34,
      "step": 46418
    },
    {
      "epoch": 2.5993392317168778,
      "grad_norm": 1.1430139541625977,
      "learning_rate": 9.305263157894737e-05,
      "loss": 0.3974,
      "step": 46419
    },
    {
      "epoch": 2.5993952290290068,
      "grad_norm": 1.2210012674331665,
      "learning_rate": 9.305236842105263e-05,
      "loss": 0.4135,
      "step": 46420
    },
    {
      "epoch": 2.599451226341136,
      "grad_norm": 1.2502466440200806,
      "learning_rate": 9.30521052631579e-05,
      "loss": 0.3892,
      "step": 46421
    },
    {
      "epoch": 2.599507223653265,
      "grad_norm": 1.5734598636627197,
      "learning_rate": 9.305184210526317e-05,
      "loss": 0.4753,
      "step": 46422
    },
    {
      "epoch": 2.5995632209653934,
      "grad_norm": 1.3160457611083984,
      "learning_rate": 9.305157894736843e-05,
      "loss": 0.4401,
      "step": 46423
    },
    {
      "epoch": 2.599619218277523,
      "grad_norm": 1.126574158668518,
      "learning_rate": 9.305131578947369e-05,
      "loss": 0.4322,
      "step": 46424
    },
    {
      "epoch": 2.5996752155896514,
      "grad_norm": 0.979223906993866,
      "learning_rate": 9.305105263157895e-05,
      "loss": 0.3052,
      "step": 46425
    },
    {
      "epoch": 2.599731212901781,
      "grad_norm": 1.0468416213989258,
      "learning_rate": 9.305078947368422e-05,
      "loss": 0.2731,
      "step": 46426
    },
    {
      "epoch": 2.5997872102139095,
      "grad_norm": 1.0872056484222412,
      "learning_rate": 9.305052631578948e-05,
      "loss": 0.4303,
      "step": 46427
    },
    {
      "epoch": 2.599843207526039,
      "grad_norm": 1.2613859176635742,
      "learning_rate": 9.305026315789475e-05,
      "loss": 0.391,
      "step": 46428
    },
    {
      "epoch": 2.5998992048381675,
      "grad_norm": 1.1846903562545776,
      "learning_rate": 9.305e-05,
      "loss": 0.3151,
      "step": 46429
    },
    {
      "epoch": 2.599955202150297,
      "grad_norm": 1.0874963998794556,
      "learning_rate": 9.304973684210527e-05,
      "loss": 0.3574,
      "step": 46430
    },
    {
      "epoch": 2.6000111994624255,
      "grad_norm": 1.1916323900222778,
      "learning_rate": 9.304947368421053e-05,
      "loss": 0.3601,
      "step": 46431
    },
    {
      "epoch": 2.600067196774555,
      "grad_norm": 1.1626559495925903,
      "learning_rate": 9.304921052631579e-05,
      "loss": 0.3617,
      "step": 46432
    },
    {
      "epoch": 2.6001231940866836,
      "grad_norm": 1.147560477256775,
      "learning_rate": 9.304894736842106e-05,
      "loss": 0.4632,
      "step": 46433
    },
    {
      "epoch": 2.600179191398813,
      "grad_norm": 1.1132581233978271,
      "learning_rate": 9.304868421052631e-05,
      "loss": 0.663,
      "step": 46434
    },
    {
      "epoch": 2.6002351887109416,
      "grad_norm": 1.254677414894104,
      "learning_rate": 9.304842105263158e-05,
      "loss": 0.4212,
      "step": 46435
    },
    {
      "epoch": 2.600291186023071,
      "grad_norm": 1.2145379781723022,
      "learning_rate": 9.304815789473684e-05,
      "loss": 0.368,
      "step": 46436
    },
    {
      "epoch": 2.6003471833351997,
      "grad_norm": 1.1810592412948608,
      "learning_rate": 9.304789473684212e-05,
      "loss": 0.3461,
      "step": 46437
    },
    {
      "epoch": 2.600403180647329,
      "grad_norm": 1.1920262575149536,
      "learning_rate": 9.304763157894736e-05,
      "loss": 0.3849,
      "step": 46438
    },
    {
      "epoch": 2.6004591779594577,
      "grad_norm": 1.1273354291915894,
      "learning_rate": 9.304736842105264e-05,
      "loss": 0.3623,
      "step": 46439
    },
    {
      "epoch": 2.600515175271587,
      "grad_norm": 1.3927035331726074,
      "learning_rate": 9.30471052631579e-05,
      "loss": 0.3894,
      "step": 46440
    },
    {
      "epoch": 2.6005711725837157,
      "grad_norm": 1.291559100151062,
      "learning_rate": 9.304684210526317e-05,
      "loss": 0.374,
      "step": 46441
    },
    {
      "epoch": 2.600627169895845,
      "grad_norm": 1.0641018152236938,
      "learning_rate": 9.304657894736843e-05,
      "loss": 0.3254,
      "step": 46442
    },
    {
      "epoch": 2.6006831672079738,
      "grad_norm": 1.4104937314987183,
      "learning_rate": 9.304631578947369e-05,
      "loss": 0.4582,
      "step": 46443
    },
    {
      "epoch": 2.6007391645201032,
      "grad_norm": 1.227135181427002,
      "learning_rate": 9.304605263157895e-05,
      "loss": 0.5633,
      "step": 46444
    },
    {
      "epoch": 2.600795161832232,
      "grad_norm": 1.0533151626586914,
      "learning_rate": 9.304578947368422e-05,
      "loss": 0.3805,
      "step": 46445
    },
    {
      "epoch": 2.6008511591443613,
      "grad_norm": 1.1004160642623901,
      "learning_rate": 9.304552631578948e-05,
      "loss": 0.4832,
      "step": 46446
    },
    {
      "epoch": 2.60090715645649,
      "grad_norm": 1.0386230945587158,
      "learning_rate": 9.304526315789474e-05,
      "loss": 0.3892,
      "step": 46447
    },
    {
      "epoch": 2.6009631537686193,
      "grad_norm": 1.1947985887527466,
      "learning_rate": 9.3045e-05,
      "loss": 0.2936,
      "step": 46448
    },
    {
      "epoch": 2.601019151080748,
      "grad_norm": 1.3050929307937622,
      "learning_rate": 9.304473684210526e-05,
      "loss": 0.3935,
      "step": 46449
    },
    {
      "epoch": 2.6010751483928773,
      "grad_norm": 1.1932874917984009,
      "learning_rate": 9.304447368421053e-05,
      "loss": 0.4036,
      "step": 46450
    },
    {
      "epoch": 2.601131145705006,
      "grad_norm": 0.9661457538604736,
      "learning_rate": 9.30442105263158e-05,
      "loss": 0.3628,
      "step": 46451
    },
    {
      "epoch": 2.6011871430171354,
      "grad_norm": 1.4228384494781494,
      "learning_rate": 9.304394736842105e-05,
      "loss": 0.7101,
      "step": 46452
    },
    {
      "epoch": 2.601243140329264,
      "grad_norm": 1.2521458864212036,
      "learning_rate": 9.304368421052631e-05,
      "loss": 0.4048,
      "step": 46453
    },
    {
      "epoch": 2.6012991376413934,
      "grad_norm": 1.1667956113815308,
      "learning_rate": 9.304342105263159e-05,
      "loss": 0.4032,
      "step": 46454
    },
    {
      "epoch": 2.601355134953522,
      "grad_norm": 1.0820139646530151,
      "learning_rate": 9.304315789473685e-05,
      "loss": 0.3675,
      "step": 46455
    },
    {
      "epoch": 2.6014111322656515,
      "grad_norm": 1.1574163436889648,
      "learning_rate": 9.30428947368421e-05,
      "loss": 0.4527,
      "step": 46456
    },
    {
      "epoch": 2.60146712957778,
      "grad_norm": 1.113939642906189,
      "learning_rate": 9.304263157894737e-05,
      "loss": 0.414,
      "step": 46457
    },
    {
      "epoch": 2.6015231268899095,
      "grad_norm": 1.214937448501587,
      "learning_rate": 9.304236842105264e-05,
      "loss": 0.3992,
      "step": 46458
    },
    {
      "epoch": 2.601579124202038,
      "grad_norm": 1.1163384914398193,
      "learning_rate": 9.30421052631579e-05,
      "loss": 0.4927,
      "step": 46459
    },
    {
      "epoch": 2.6016351215141675,
      "grad_norm": 1.4073883295059204,
      "learning_rate": 9.304184210526317e-05,
      "loss": 0.6369,
      "step": 46460
    },
    {
      "epoch": 2.601691118826296,
      "grad_norm": 1.188210129737854,
      "learning_rate": 9.304157894736842e-05,
      "loss": 0.3661,
      "step": 46461
    },
    {
      "epoch": 2.6017471161384256,
      "grad_norm": 1.5695980787277222,
      "learning_rate": 9.304131578947369e-05,
      "loss": 0.3368,
      "step": 46462
    },
    {
      "epoch": 2.601803113450554,
      "grad_norm": 1.061631202697754,
      "learning_rate": 9.304105263157895e-05,
      "loss": 0.2741,
      "step": 46463
    },
    {
      "epoch": 2.6018591107626836,
      "grad_norm": 1.2739530801773071,
      "learning_rate": 9.304078947368422e-05,
      "loss": 0.5952,
      "step": 46464
    },
    {
      "epoch": 2.601915108074812,
      "grad_norm": 1.2000675201416016,
      "learning_rate": 9.304052631578948e-05,
      "loss": 0.3457,
      "step": 46465
    },
    {
      "epoch": 2.6019711053869417,
      "grad_norm": 0.9580949544906616,
      "learning_rate": 9.304026315789473e-05,
      "loss": 0.4259,
      "step": 46466
    },
    {
      "epoch": 2.6020271026990702,
      "grad_norm": 1.2437034845352173,
      "learning_rate": 9.304e-05,
      "loss": 0.4251,
      "step": 46467
    },
    {
      "epoch": 2.6020831000111997,
      "grad_norm": 1.0650449991226196,
      "learning_rate": 9.303973684210526e-05,
      "loss": 0.4338,
      "step": 46468
    },
    {
      "epoch": 2.6021390973233283,
      "grad_norm": 1.244703769683838,
      "learning_rate": 9.303947368421054e-05,
      "loss": 0.3511,
      "step": 46469
    },
    {
      "epoch": 2.6021950946354577,
      "grad_norm": 1.3416006565093994,
      "learning_rate": 9.30392105263158e-05,
      "loss": 0.4138,
      "step": 46470
    },
    {
      "epoch": 2.6022510919475863,
      "grad_norm": 1.353318691253662,
      "learning_rate": 9.303894736842106e-05,
      "loss": 0.352,
      "step": 46471
    },
    {
      "epoch": 2.6023070892597158,
      "grad_norm": 1.4492400884628296,
      "learning_rate": 9.303868421052632e-05,
      "loss": 0.4157,
      "step": 46472
    },
    {
      "epoch": 2.6023630865718443,
      "grad_norm": 1.1647073030471802,
      "learning_rate": 9.303842105263159e-05,
      "loss": 0.3766,
      "step": 46473
    },
    {
      "epoch": 2.602419083883974,
      "grad_norm": 1.231884479522705,
      "learning_rate": 9.303815789473685e-05,
      "loss": 0.3742,
      "step": 46474
    },
    {
      "epoch": 2.6024750811961024,
      "grad_norm": 1.00637948513031,
      "learning_rate": 9.303789473684211e-05,
      "loss": 0.3254,
      "step": 46475
    },
    {
      "epoch": 2.602531078508232,
      "grad_norm": 1.198676347732544,
      "learning_rate": 9.303763157894737e-05,
      "loss": 0.4207,
      "step": 46476
    },
    {
      "epoch": 2.6025870758203604,
      "grad_norm": 1.2479243278503418,
      "learning_rate": 9.303736842105264e-05,
      "loss": 0.4484,
      "step": 46477
    },
    {
      "epoch": 2.60264307313249,
      "grad_norm": 1.097208023071289,
      "learning_rate": 9.30371052631579e-05,
      "loss": 0.4814,
      "step": 46478
    },
    {
      "epoch": 2.6026990704446185,
      "grad_norm": 1.277184009552002,
      "learning_rate": 9.303684210526316e-05,
      "loss": 0.3587,
      "step": 46479
    },
    {
      "epoch": 2.602755067756748,
      "grad_norm": 1.0342180728912354,
      "learning_rate": 9.303657894736842e-05,
      "loss": 0.3902,
      "step": 46480
    },
    {
      "epoch": 2.6028110650688765,
      "grad_norm": 1.42534339427948,
      "learning_rate": 9.30363157894737e-05,
      "loss": 0.4532,
      "step": 46481
    },
    {
      "epoch": 2.602867062381006,
      "grad_norm": 1.2270914316177368,
      "learning_rate": 9.303605263157895e-05,
      "loss": 0.4947,
      "step": 46482
    },
    {
      "epoch": 2.6029230596931345,
      "grad_norm": 1.6300156116485596,
      "learning_rate": 9.303578947368421e-05,
      "loss": 0.377,
      "step": 46483
    },
    {
      "epoch": 2.602979057005264,
      "grad_norm": 0.9486250877380371,
      "learning_rate": 9.303552631578947e-05,
      "loss": 0.2987,
      "step": 46484
    },
    {
      "epoch": 2.6030350543173926,
      "grad_norm": 1.0906494855880737,
      "learning_rate": 9.303526315789473e-05,
      "loss": 0.3443,
      "step": 46485
    },
    {
      "epoch": 2.603091051629522,
      "grad_norm": 1.5641968250274658,
      "learning_rate": 9.3035e-05,
      "loss": 0.5295,
      "step": 46486
    },
    {
      "epoch": 2.6031470489416506,
      "grad_norm": 1.4293843507766724,
      "learning_rate": 9.303473684210527e-05,
      "loss": 0.3326,
      "step": 46487
    },
    {
      "epoch": 2.60320304625378,
      "grad_norm": 1.3640592098236084,
      "learning_rate": 9.303447368421053e-05,
      "loss": 0.4591,
      "step": 46488
    },
    {
      "epoch": 2.6032590435659086,
      "grad_norm": 1.1550309658050537,
      "learning_rate": 9.303421052631579e-05,
      "loss": 0.3343,
      "step": 46489
    },
    {
      "epoch": 2.603315040878038,
      "grad_norm": 1.0223761796951294,
      "learning_rate": 9.303394736842106e-05,
      "loss": 0.2891,
      "step": 46490
    },
    {
      "epoch": 2.6033710381901667,
      "grad_norm": 1.1564793586730957,
      "learning_rate": 9.303368421052632e-05,
      "loss": 0.3646,
      "step": 46491
    },
    {
      "epoch": 2.603427035502296,
      "grad_norm": 4.680053234100342,
      "learning_rate": 9.303342105263159e-05,
      "loss": 0.4865,
      "step": 46492
    },
    {
      "epoch": 2.6034830328144247,
      "grad_norm": 1.1159172058105469,
      "learning_rate": 9.303315789473684e-05,
      "loss": 0.2847,
      "step": 46493
    },
    {
      "epoch": 2.603539030126554,
      "grad_norm": 1.0027815103530884,
      "learning_rate": 9.303289473684211e-05,
      "loss": 0.4123,
      "step": 46494
    },
    {
      "epoch": 2.6035950274386828,
      "grad_norm": 1.0499452352523804,
      "learning_rate": 9.303263157894737e-05,
      "loss": 0.2889,
      "step": 46495
    },
    {
      "epoch": 2.6036510247508122,
      "grad_norm": 1.4142584800720215,
      "learning_rate": 9.303236842105264e-05,
      "loss": 0.5174,
      "step": 46496
    },
    {
      "epoch": 2.603707022062941,
      "grad_norm": 0.989594578742981,
      "learning_rate": 9.30321052631579e-05,
      "loss": 0.3169,
      "step": 46497
    },
    {
      "epoch": 2.6037630193750703,
      "grad_norm": 1.1057268381118774,
      "learning_rate": 9.303184210526316e-05,
      "loss": 0.2985,
      "step": 46498
    },
    {
      "epoch": 2.603819016687199,
      "grad_norm": 1.039628028869629,
      "learning_rate": 9.303157894736842e-05,
      "loss": 0.4595,
      "step": 46499
    },
    {
      "epoch": 2.6038750139993283,
      "grad_norm": 1.202059030532837,
      "learning_rate": 9.30313157894737e-05,
      "loss": 0.422,
      "step": 46500
    },
    {
      "epoch": 2.603931011311457,
      "grad_norm": 1.0177092552185059,
      "learning_rate": 9.303105263157896e-05,
      "loss": 0.3472,
      "step": 46501
    },
    {
      "epoch": 2.6039870086235863,
      "grad_norm": 1.3162777423858643,
      "learning_rate": 9.303078947368422e-05,
      "loss": 0.3866,
      "step": 46502
    },
    {
      "epoch": 2.604043005935715,
      "grad_norm": 1.2071207761764526,
      "learning_rate": 9.303052631578948e-05,
      "loss": 0.4198,
      "step": 46503
    },
    {
      "epoch": 2.6040990032478444,
      "grad_norm": 1.9643195867538452,
      "learning_rate": 9.303026315789474e-05,
      "loss": 0.5239,
      "step": 46504
    },
    {
      "epoch": 2.604155000559973,
      "grad_norm": 1.4066940546035767,
      "learning_rate": 9.303000000000001e-05,
      "loss": 0.412,
      "step": 46505
    },
    {
      "epoch": 2.604210997872102,
      "grad_norm": 1.056431770324707,
      "learning_rate": 9.302973684210527e-05,
      "loss": 0.2899,
      "step": 46506
    },
    {
      "epoch": 2.604266995184231,
      "grad_norm": 0.9736373424530029,
      "learning_rate": 9.302947368421053e-05,
      "loss": 0.3523,
      "step": 46507
    },
    {
      "epoch": 2.60432299249636,
      "grad_norm": 1.0358986854553223,
      "learning_rate": 9.302921052631579e-05,
      "loss": 0.302,
      "step": 46508
    },
    {
      "epoch": 2.604378989808489,
      "grad_norm": 1.272763967514038,
      "learning_rate": 9.302894736842106e-05,
      "loss": 0.4631,
      "step": 46509
    },
    {
      "epoch": 2.604434987120618,
      "grad_norm": 1.051343560218811,
      "learning_rate": 9.302868421052632e-05,
      "loss": 0.3608,
      "step": 46510
    },
    {
      "epoch": 2.604490984432747,
      "grad_norm": 1.0620665550231934,
      "learning_rate": 9.302842105263158e-05,
      "loss": 0.3109,
      "step": 46511
    },
    {
      "epoch": 2.604546981744876,
      "grad_norm": 1.5440928936004639,
      "learning_rate": 9.302815789473684e-05,
      "loss": 0.4604,
      "step": 46512
    },
    {
      "epoch": 2.604602979057005,
      "grad_norm": 1.2319453954696655,
      "learning_rate": 9.302789473684211e-05,
      "loss": 0.3864,
      "step": 46513
    },
    {
      "epoch": 2.604658976369134,
      "grad_norm": 1.358185887336731,
      "learning_rate": 9.302763157894737e-05,
      "loss": 0.3582,
      "step": 46514
    },
    {
      "epoch": 2.604714973681263,
      "grad_norm": 1.1510616540908813,
      "learning_rate": 9.302736842105265e-05,
      "loss": 0.3906,
      "step": 46515
    },
    {
      "epoch": 2.604770970993392,
      "grad_norm": 1.4592418670654297,
      "learning_rate": 9.302710526315789e-05,
      "loss": 0.3838,
      "step": 46516
    },
    {
      "epoch": 2.604826968305521,
      "grad_norm": 1.2418780326843262,
      "learning_rate": 9.302684210526317e-05,
      "loss": 0.4018,
      "step": 46517
    },
    {
      "epoch": 2.60488296561765,
      "grad_norm": 1.6895040273666382,
      "learning_rate": 9.302657894736843e-05,
      "loss": 0.3782,
      "step": 46518
    },
    {
      "epoch": 2.6049389629297792,
      "grad_norm": 1.0250135660171509,
      "learning_rate": 9.302631578947369e-05,
      "loss": 0.3604,
      "step": 46519
    },
    {
      "epoch": 2.6049949602419082,
      "grad_norm": 1.4300905466079712,
      "learning_rate": 9.302605263157896e-05,
      "loss": 0.3191,
      "step": 46520
    },
    {
      "epoch": 2.6050509575540373,
      "grad_norm": 1.3257346153259277,
      "learning_rate": 9.30257894736842e-05,
      "loss": 0.5257,
      "step": 46521
    },
    {
      "epoch": 2.6051069548661663,
      "grad_norm": 1.2085860967636108,
      "learning_rate": 9.302552631578948e-05,
      "loss": 0.4289,
      "step": 46522
    },
    {
      "epoch": 2.6051629521782953,
      "grad_norm": 1.2155348062515259,
      "learning_rate": 9.302526315789474e-05,
      "loss": 0.4195,
      "step": 46523
    },
    {
      "epoch": 2.6052189494904243,
      "grad_norm": 1.0436477661132812,
      "learning_rate": 9.302500000000001e-05,
      "loss": 0.3814,
      "step": 46524
    },
    {
      "epoch": 2.6052749468025533,
      "grad_norm": 1.1576848030090332,
      "learning_rate": 9.302473684210527e-05,
      "loss": 0.3968,
      "step": 46525
    },
    {
      "epoch": 2.6053309441146824,
      "grad_norm": 6.950581073760986,
      "learning_rate": 9.302447368421053e-05,
      "loss": 0.5232,
      "step": 46526
    },
    {
      "epoch": 2.6053869414268114,
      "grad_norm": 1.756662368774414,
      "learning_rate": 9.302421052631579e-05,
      "loss": 0.4306,
      "step": 46527
    },
    {
      "epoch": 2.6054429387389404,
      "grad_norm": 1.0839502811431885,
      "learning_rate": 9.302394736842106e-05,
      "loss": 0.3517,
      "step": 46528
    },
    {
      "epoch": 2.6054989360510694,
      "grad_norm": 1.1785544157028198,
      "learning_rate": 9.302368421052632e-05,
      "loss": 0.3655,
      "step": 46529
    },
    {
      "epoch": 2.6055549333631984,
      "grad_norm": 1.1357667446136475,
      "learning_rate": 9.302342105263158e-05,
      "loss": 0.3231,
      "step": 46530
    },
    {
      "epoch": 2.6056109306753275,
      "grad_norm": 1.2374035120010376,
      "learning_rate": 9.302315789473684e-05,
      "loss": 0.3943,
      "step": 46531
    },
    {
      "epoch": 2.6056669279874565,
      "grad_norm": 1.360485553741455,
      "learning_rate": 9.302289473684212e-05,
      "loss": 0.4535,
      "step": 46532
    },
    {
      "epoch": 2.6057229252995855,
      "grad_norm": 1.2514512538909912,
      "learning_rate": 9.302263157894738e-05,
      "loss": 0.4974,
      "step": 46533
    },
    {
      "epoch": 2.6057789226117145,
      "grad_norm": 1.1692370176315308,
      "learning_rate": 9.302236842105264e-05,
      "loss": 0.3499,
      "step": 46534
    },
    {
      "epoch": 2.6058349199238435,
      "grad_norm": 0.9827505946159363,
      "learning_rate": 9.30221052631579e-05,
      "loss": 0.3726,
      "step": 46535
    },
    {
      "epoch": 2.6058909172359725,
      "grad_norm": 1.0020781755447388,
      "learning_rate": 9.302184210526316e-05,
      "loss": 0.2868,
      "step": 46536
    },
    {
      "epoch": 2.6059469145481016,
      "grad_norm": 1.3046605587005615,
      "learning_rate": 9.302157894736843e-05,
      "loss": 0.3484,
      "step": 46537
    },
    {
      "epoch": 2.6060029118602306,
      "grad_norm": 1.155605435371399,
      "learning_rate": 9.302131578947369e-05,
      "loss": 0.4737,
      "step": 46538
    },
    {
      "epoch": 2.6060589091723596,
      "grad_norm": 1.1728681325912476,
      "learning_rate": 9.302105263157895e-05,
      "loss": 0.4589,
      "step": 46539
    },
    {
      "epoch": 2.6061149064844886,
      "grad_norm": 1.2775704860687256,
      "learning_rate": 9.302078947368421e-05,
      "loss": 0.4299,
      "step": 46540
    },
    {
      "epoch": 2.6061709037966176,
      "grad_norm": 1.2015646696090698,
      "learning_rate": 9.302052631578948e-05,
      "loss": 0.4057,
      "step": 46541
    },
    {
      "epoch": 2.6062269011087467,
      "grad_norm": 1.1919351816177368,
      "learning_rate": 9.302026315789474e-05,
      "loss": 0.5465,
      "step": 46542
    },
    {
      "epoch": 2.6062828984208757,
      "grad_norm": 1.0784245729446411,
      "learning_rate": 9.302e-05,
      "loss": 0.4397,
      "step": 46543
    },
    {
      "epoch": 2.6063388957330047,
      "grad_norm": 1.0491892099380493,
      "learning_rate": 9.301973684210526e-05,
      "loss": 0.4944,
      "step": 46544
    },
    {
      "epoch": 2.6063948930451337,
      "grad_norm": 1.2368894815444946,
      "learning_rate": 9.301947368421053e-05,
      "loss": 0.5697,
      "step": 46545
    },
    {
      "epoch": 2.6064508903572627,
      "grad_norm": 0.9837244153022766,
      "learning_rate": 9.301921052631579e-05,
      "loss": 0.4152,
      "step": 46546
    },
    {
      "epoch": 2.6065068876693918,
      "grad_norm": 1.315573811531067,
      "learning_rate": 9.301894736842107e-05,
      "loss": 0.3764,
      "step": 46547
    },
    {
      "epoch": 2.6065628849815208,
      "grad_norm": 1.3167028427124023,
      "learning_rate": 9.301868421052631e-05,
      "loss": 0.4615,
      "step": 46548
    },
    {
      "epoch": 2.60661888229365,
      "grad_norm": 1.3782520294189453,
      "learning_rate": 9.301842105263159e-05,
      "loss": 0.3486,
      "step": 46549
    },
    {
      "epoch": 2.606674879605779,
      "grad_norm": 1.0889307260513306,
      "learning_rate": 9.301815789473685e-05,
      "loss": 0.3447,
      "step": 46550
    },
    {
      "epoch": 2.606730876917908,
      "grad_norm": 1.2331639528274536,
      "learning_rate": 9.301789473684212e-05,
      "loss": 0.3692,
      "step": 46551
    },
    {
      "epoch": 2.606786874230037,
      "grad_norm": 1.1452991962432861,
      "learning_rate": 9.301763157894738e-05,
      "loss": 0.4368,
      "step": 46552
    },
    {
      "epoch": 2.606842871542166,
      "grad_norm": 1.2455683946609497,
      "learning_rate": 9.301736842105262e-05,
      "loss": 0.4391,
      "step": 46553
    },
    {
      "epoch": 2.606898868854295,
      "grad_norm": 1.530213475227356,
      "learning_rate": 9.30171052631579e-05,
      "loss": 0.4022,
      "step": 46554
    },
    {
      "epoch": 2.606954866166424,
      "grad_norm": 1.0618746280670166,
      "learning_rate": 9.301684210526316e-05,
      "loss": 0.44,
      "step": 46555
    },
    {
      "epoch": 2.607010863478553,
      "grad_norm": 1.0888922214508057,
      "learning_rate": 9.301657894736843e-05,
      "loss": 0.4074,
      "step": 46556
    },
    {
      "epoch": 2.607066860790682,
      "grad_norm": 1.3950855731964111,
      "learning_rate": 9.301631578947369e-05,
      "loss": 0.3681,
      "step": 46557
    },
    {
      "epoch": 2.607122858102811,
      "grad_norm": 1.1784887313842773,
      "learning_rate": 9.301605263157895e-05,
      "loss": 0.3617,
      "step": 46558
    },
    {
      "epoch": 2.60717885541494,
      "grad_norm": 1.236409306526184,
      "learning_rate": 9.301578947368421e-05,
      "loss": 0.4029,
      "step": 46559
    },
    {
      "epoch": 2.607234852727069,
      "grad_norm": 1.3023930788040161,
      "learning_rate": 9.301552631578948e-05,
      "loss": 0.477,
      "step": 46560
    },
    {
      "epoch": 2.607290850039198,
      "grad_norm": 1.027319073677063,
      "learning_rate": 9.301526315789474e-05,
      "loss": 0.3757,
      "step": 46561
    },
    {
      "epoch": 2.607346847351327,
      "grad_norm": 1.0279603004455566,
      "learning_rate": 9.3015e-05,
      "loss": 0.3608,
      "step": 46562
    },
    {
      "epoch": 2.607402844663456,
      "grad_norm": 1.2154029607772827,
      "learning_rate": 9.301473684210526e-05,
      "loss": 0.5579,
      "step": 46563
    },
    {
      "epoch": 2.607458841975585,
      "grad_norm": 2.1505751609802246,
      "learning_rate": 9.301447368421054e-05,
      "loss": 0.3735,
      "step": 46564
    },
    {
      "epoch": 2.607514839287714,
      "grad_norm": 1.3641637563705444,
      "learning_rate": 9.30142105263158e-05,
      "loss": 0.5112,
      "step": 46565
    },
    {
      "epoch": 2.607570836599843,
      "grad_norm": 1.1326463222503662,
      "learning_rate": 9.301394736842106e-05,
      "loss": 0.2707,
      "step": 46566
    },
    {
      "epoch": 2.607626833911972,
      "grad_norm": 1.152876615524292,
      "learning_rate": 9.301368421052632e-05,
      "loss": 0.3703,
      "step": 46567
    },
    {
      "epoch": 2.607682831224101,
      "grad_norm": 1.386849045753479,
      "learning_rate": 9.301342105263159e-05,
      "loss": 0.3763,
      "step": 46568
    },
    {
      "epoch": 2.60773882853623,
      "grad_norm": 1.6253869533538818,
      "learning_rate": 9.301315789473685e-05,
      "loss": 0.4848,
      "step": 46569
    },
    {
      "epoch": 2.607794825848359,
      "grad_norm": 1.4386910200119019,
      "learning_rate": 9.301289473684211e-05,
      "loss": 0.3897,
      "step": 46570
    },
    {
      "epoch": 2.607850823160488,
      "grad_norm": 1.4965099096298218,
      "learning_rate": 9.301263157894737e-05,
      "loss": 0.4671,
      "step": 46571
    },
    {
      "epoch": 2.6079068204726172,
      "grad_norm": 1.1429165601730347,
      "learning_rate": 9.301236842105263e-05,
      "loss": 0.5394,
      "step": 46572
    },
    {
      "epoch": 2.6079628177847463,
      "grad_norm": 1.2738810777664185,
      "learning_rate": 9.30121052631579e-05,
      "loss": 0.4492,
      "step": 46573
    },
    {
      "epoch": 2.6080188150968753,
      "grad_norm": 1.5004088878631592,
      "learning_rate": 9.301184210526316e-05,
      "loss": 0.4748,
      "step": 46574
    },
    {
      "epoch": 2.6080748124090043,
      "grad_norm": 3.550349235534668,
      "learning_rate": 9.301157894736843e-05,
      "loss": 0.3674,
      "step": 46575
    },
    {
      "epoch": 2.6081308097211333,
      "grad_norm": 1.0376020669937134,
      "learning_rate": 9.301131578947368e-05,
      "loss": 0.3438,
      "step": 46576
    },
    {
      "epoch": 2.6081868070332623,
      "grad_norm": 1.4004347324371338,
      "learning_rate": 9.301105263157895e-05,
      "loss": 0.4473,
      "step": 46577
    },
    {
      "epoch": 2.6082428043453914,
      "grad_norm": 1.0414180755615234,
      "learning_rate": 9.301078947368421e-05,
      "loss": 0.3527,
      "step": 46578
    },
    {
      "epoch": 2.6082988016575204,
      "grad_norm": 1.1772812604904175,
      "learning_rate": 9.301052631578949e-05,
      "loss": 0.282,
      "step": 46579
    },
    {
      "epoch": 2.6083547989696494,
      "grad_norm": 1.2752467393875122,
      "learning_rate": 9.301026315789475e-05,
      "loss": 0.476,
      "step": 46580
    },
    {
      "epoch": 2.6084107962817784,
      "grad_norm": 1.4263345003128052,
      "learning_rate": 9.301e-05,
      "loss": 0.5376,
      "step": 46581
    },
    {
      "epoch": 2.6084667935939074,
      "grad_norm": 1.284131407737732,
      "learning_rate": 9.300973684210527e-05,
      "loss": 0.3322,
      "step": 46582
    },
    {
      "epoch": 2.6085227909060364,
      "grad_norm": 1.3838568925857544,
      "learning_rate": 9.300947368421054e-05,
      "loss": 0.4009,
      "step": 46583
    },
    {
      "epoch": 2.6085787882181655,
      "grad_norm": 1.0283442735671997,
      "learning_rate": 9.30092105263158e-05,
      "loss": 0.3557,
      "step": 46584
    },
    {
      "epoch": 2.6086347855302945,
      "grad_norm": 1.3276817798614502,
      "learning_rate": 9.300894736842106e-05,
      "loss": 0.4271,
      "step": 46585
    },
    {
      "epoch": 2.6086907828424235,
      "grad_norm": 1.7550311088562012,
      "learning_rate": 9.300868421052632e-05,
      "loss": 0.3818,
      "step": 46586
    },
    {
      "epoch": 2.6087467801545525,
      "grad_norm": 1.219070553779602,
      "learning_rate": 9.300842105263158e-05,
      "loss": 0.5494,
      "step": 46587
    },
    {
      "epoch": 2.6088027774666815,
      "grad_norm": 1.201698660850525,
      "learning_rate": 9.300815789473685e-05,
      "loss": 0.4026,
      "step": 46588
    },
    {
      "epoch": 2.6088587747788106,
      "grad_norm": 1.2091761827468872,
      "learning_rate": 9.300789473684211e-05,
      "loss": 0.4615,
      "step": 46589
    },
    {
      "epoch": 2.6089147720909396,
      "grad_norm": 1.363063097000122,
      "learning_rate": 9.300763157894737e-05,
      "loss": 0.4248,
      "step": 46590
    },
    {
      "epoch": 2.6089707694030686,
      "grad_norm": 1.4333062171936035,
      "learning_rate": 9.300736842105263e-05,
      "loss": 0.5165,
      "step": 46591
    },
    {
      "epoch": 2.6090267667151976,
      "grad_norm": 1.2628583908081055,
      "learning_rate": 9.30071052631579e-05,
      "loss": 0.3636,
      "step": 46592
    },
    {
      "epoch": 2.6090827640273266,
      "grad_norm": 1.2157214879989624,
      "learning_rate": 9.300684210526316e-05,
      "loss": 0.4027,
      "step": 46593
    },
    {
      "epoch": 2.6091387613394557,
      "grad_norm": 1.1543291807174683,
      "learning_rate": 9.300657894736842e-05,
      "loss": 0.4073,
      "step": 46594
    },
    {
      "epoch": 2.6091947586515847,
      "grad_norm": 1.238649606704712,
      "learning_rate": 9.300631578947368e-05,
      "loss": 0.3677,
      "step": 46595
    },
    {
      "epoch": 2.6092507559637137,
      "grad_norm": 1.2070821523666382,
      "learning_rate": 9.300605263157896e-05,
      "loss": 0.3631,
      "step": 46596
    },
    {
      "epoch": 2.6093067532758427,
      "grad_norm": 1.1323351860046387,
      "learning_rate": 9.300578947368422e-05,
      "loss": 0.3464,
      "step": 46597
    },
    {
      "epoch": 2.6093627505879717,
      "grad_norm": 1.1080574989318848,
      "learning_rate": 9.300552631578948e-05,
      "loss": 0.5159,
      "step": 46598
    },
    {
      "epoch": 2.6094187479001008,
      "grad_norm": 1.1234997510910034,
      "learning_rate": 9.300526315789473e-05,
      "loss": 0.3592,
      "step": 46599
    },
    {
      "epoch": 2.6094747452122298,
      "grad_norm": 1.3733960390090942,
      "learning_rate": 9.300500000000001e-05,
      "loss": 0.3532,
      "step": 46600
    },
    {
      "epoch": 2.609530742524359,
      "grad_norm": 1.1492043733596802,
      "learning_rate": 9.300473684210527e-05,
      "loss": 0.4223,
      "step": 46601
    },
    {
      "epoch": 2.609586739836488,
      "grad_norm": 1.1460093259811401,
      "learning_rate": 9.300447368421054e-05,
      "loss": 0.4638,
      "step": 46602
    },
    {
      "epoch": 2.609642737148617,
      "grad_norm": 1.3171253204345703,
      "learning_rate": 9.300421052631579e-05,
      "loss": 0.3983,
      "step": 46603
    },
    {
      "epoch": 2.609698734460746,
      "grad_norm": 1.28238844871521,
      "learning_rate": 9.300394736842105e-05,
      "loss": 0.3827,
      "step": 46604
    },
    {
      "epoch": 2.609754731772875,
      "grad_norm": 1.1583778858184814,
      "learning_rate": 9.300368421052632e-05,
      "loss": 0.4178,
      "step": 46605
    },
    {
      "epoch": 2.609810729085004,
      "grad_norm": 1.1171513795852661,
      "learning_rate": 9.300342105263158e-05,
      "loss": 0.4351,
      "step": 46606
    },
    {
      "epoch": 2.609866726397133,
      "grad_norm": 1.218218445777893,
      "learning_rate": 9.300315789473685e-05,
      "loss": 0.4024,
      "step": 46607
    },
    {
      "epoch": 2.609922723709262,
      "grad_norm": 1.0216279029846191,
      "learning_rate": 9.30028947368421e-05,
      "loss": 0.4262,
      "step": 46608
    },
    {
      "epoch": 2.609978721021391,
      "grad_norm": 2.543121814727783,
      "learning_rate": 9.300263157894737e-05,
      "loss": 0.429,
      "step": 46609
    },
    {
      "epoch": 2.61003471833352,
      "grad_norm": 1.2610794305801392,
      "learning_rate": 9.300236842105263e-05,
      "loss": 0.374,
      "step": 46610
    },
    {
      "epoch": 2.610090715645649,
      "grad_norm": 1.0212806463241577,
      "learning_rate": 9.30021052631579e-05,
      "loss": 0.317,
      "step": 46611
    },
    {
      "epoch": 2.610146712957778,
      "grad_norm": 1.213247299194336,
      "learning_rate": 9.300184210526317e-05,
      "loss": 0.4415,
      "step": 46612
    },
    {
      "epoch": 2.610202710269907,
      "grad_norm": 1.2385320663452148,
      "learning_rate": 9.300157894736843e-05,
      "loss": 0.3539,
      "step": 46613
    },
    {
      "epoch": 2.610258707582036,
      "grad_norm": 1.3352187871932983,
      "learning_rate": 9.300131578947369e-05,
      "loss": 0.3995,
      "step": 46614
    },
    {
      "epoch": 2.610314704894165,
      "grad_norm": 3.493345022201538,
      "learning_rate": 9.300105263157896e-05,
      "loss": 0.3789,
      "step": 46615
    },
    {
      "epoch": 2.610370702206294,
      "grad_norm": 1.8478645086288452,
      "learning_rate": 9.300078947368422e-05,
      "loss": 0.4978,
      "step": 46616
    },
    {
      "epoch": 2.610426699518423,
      "grad_norm": 1.2096174955368042,
      "learning_rate": 9.300052631578948e-05,
      "loss": 0.286,
      "step": 46617
    },
    {
      "epoch": 2.610482696830552,
      "grad_norm": 1.294213056564331,
      "learning_rate": 9.300026315789474e-05,
      "loss": 0.3751,
      "step": 46618
    },
    {
      "epoch": 2.610538694142681,
      "grad_norm": 1.1654341220855713,
      "learning_rate": 9.300000000000001e-05,
      "loss": 0.4262,
      "step": 46619
    },
    {
      "epoch": 2.61059469145481,
      "grad_norm": 1.1719697713851929,
      "learning_rate": 9.299973684210527e-05,
      "loss": 0.3955,
      "step": 46620
    },
    {
      "epoch": 2.610650688766939,
      "grad_norm": 1.2508690357208252,
      "learning_rate": 9.299947368421053e-05,
      "loss": 0.4435,
      "step": 46621
    },
    {
      "epoch": 2.610706686079068,
      "grad_norm": 1.1091339588165283,
      "learning_rate": 9.299921052631579e-05,
      "loss": 0.4597,
      "step": 46622
    },
    {
      "epoch": 2.610762683391197,
      "grad_norm": 1.0975714921951294,
      "learning_rate": 9.299894736842105e-05,
      "loss": 0.3784,
      "step": 46623
    },
    {
      "epoch": 2.6108186807033262,
      "grad_norm": 0.9454960823059082,
      "learning_rate": 9.299868421052632e-05,
      "loss": 0.2611,
      "step": 46624
    },
    {
      "epoch": 2.6108746780154553,
      "grad_norm": 1.1642026901245117,
      "learning_rate": 9.299842105263158e-05,
      "loss": 0.4509,
      "step": 46625
    },
    {
      "epoch": 2.6109306753275843,
      "grad_norm": 1.5556083917617798,
      "learning_rate": 9.299815789473684e-05,
      "loss": 0.6253,
      "step": 46626
    },
    {
      "epoch": 2.6109866726397133,
      "grad_norm": 1.224578619003296,
      "learning_rate": 9.29978947368421e-05,
      "loss": 0.3381,
      "step": 46627
    },
    {
      "epoch": 2.6110426699518423,
      "grad_norm": 3.255194902420044,
      "learning_rate": 9.299763157894738e-05,
      "loss": 0.5234,
      "step": 46628
    },
    {
      "epoch": 2.6110986672639713,
      "grad_norm": 1.1428229808807373,
      "learning_rate": 9.299736842105264e-05,
      "loss": 0.3073,
      "step": 46629
    },
    {
      "epoch": 2.6111546645761003,
      "grad_norm": 1.4078631401062012,
      "learning_rate": 9.299710526315791e-05,
      "loss": 0.2542,
      "step": 46630
    },
    {
      "epoch": 2.6112106618882294,
      "grad_norm": 1.068902611732483,
      "learning_rate": 9.299684210526315e-05,
      "loss": 0.396,
      "step": 46631
    },
    {
      "epoch": 2.6112666592003584,
      "grad_norm": 1.0733650922775269,
      "learning_rate": 9.299657894736843e-05,
      "loss": 0.3553,
      "step": 46632
    },
    {
      "epoch": 2.6113226565124874,
      "grad_norm": 1.222968578338623,
      "learning_rate": 9.299631578947369e-05,
      "loss": 0.334,
      "step": 46633
    },
    {
      "epoch": 2.6113786538246164,
      "grad_norm": 1.3344064950942993,
      "learning_rate": 9.299605263157896e-05,
      "loss": 0.4478,
      "step": 46634
    },
    {
      "epoch": 2.6114346511367454,
      "grad_norm": 1.1899425983428955,
      "learning_rate": 9.299578947368422e-05,
      "loss": 0.3186,
      "step": 46635
    },
    {
      "epoch": 2.6114906484488745,
      "grad_norm": 1.0654300451278687,
      "learning_rate": 9.299552631578948e-05,
      "loss": 0.3854,
      "step": 46636
    },
    {
      "epoch": 2.6115466457610035,
      "grad_norm": 1.1135302782058716,
      "learning_rate": 9.299526315789474e-05,
      "loss": 0.5509,
      "step": 46637
    },
    {
      "epoch": 2.6116026430731325,
      "grad_norm": 1.1014275550842285,
      "learning_rate": 9.299500000000001e-05,
      "loss": 0.4745,
      "step": 46638
    },
    {
      "epoch": 2.6116586403852615,
      "grad_norm": 1.0376029014587402,
      "learning_rate": 9.299473684210527e-05,
      "loss": 0.3105,
      "step": 46639
    },
    {
      "epoch": 2.6117146376973905,
      "grad_norm": 1.4303823709487915,
      "learning_rate": 9.299447368421052e-05,
      "loss": 0.4444,
      "step": 46640
    },
    {
      "epoch": 2.6117706350095196,
      "grad_norm": 1.406695008277893,
      "learning_rate": 9.299421052631579e-05,
      "loss": 0.3542,
      "step": 46641
    },
    {
      "epoch": 2.6118266323216486,
      "grad_norm": 1.0023406744003296,
      "learning_rate": 9.299394736842105e-05,
      "loss": 0.3404,
      "step": 46642
    },
    {
      "epoch": 2.6118826296337776,
      "grad_norm": 1.2660740613937378,
      "learning_rate": 9.299368421052633e-05,
      "loss": 0.5272,
      "step": 46643
    },
    {
      "epoch": 2.6119386269459066,
      "grad_norm": 1.1643739938735962,
      "learning_rate": 9.299342105263159e-05,
      "loss": 0.4506,
      "step": 46644
    },
    {
      "epoch": 2.6119946242580356,
      "grad_norm": 1.2932565212249756,
      "learning_rate": 9.299315789473685e-05,
      "loss": 0.3818,
      "step": 46645
    },
    {
      "epoch": 2.6120506215701647,
      "grad_norm": 1.4373708963394165,
      "learning_rate": 9.29928947368421e-05,
      "loss": 0.5381,
      "step": 46646
    },
    {
      "epoch": 2.6121066188822937,
      "grad_norm": 1.4364473819732666,
      "learning_rate": 9.299263157894738e-05,
      "loss": 0.5348,
      "step": 46647
    },
    {
      "epoch": 2.6121626161944227,
      "grad_norm": 1.1308902502059937,
      "learning_rate": 9.299236842105264e-05,
      "loss": 0.5657,
      "step": 46648
    },
    {
      "epoch": 2.6122186135065517,
      "grad_norm": 1.3225475549697876,
      "learning_rate": 9.29921052631579e-05,
      "loss": 0.465,
      "step": 46649
    },
    {
      "epoch": 2.6122746108186807,
      "grad_norm": 1.1010724306106567,
      "learning_rate": 9.299184210526316e-05,
      "loss": 0.3833,
      "step": 46650
    },
    {
      "epoch": 2.6123306081308098,
      "grad_norm": 1.1573151350021362,
      "learning_rate": 9.299157894736843e-05,
      "loss": 0.3548,
      "step": 46651
    },
    {
      "epoch": 2.6123866054429388,
      "grad_norm": 1.2825130224227905,
      "learning_rate": 9.299131578947369e-05,
      "loss": 0.4014,
      "step": 46652
    },
    {
      "epoch": 2.612442602755068,
      "grad_norm": 1.2304491996765137,
      "learning_rate": 9.299105263157895e-05,
      "loss": 0.475,
      "step": 46653
    },
    {
      "epoch": 2.612498600067197,
      "grad_norm": 0.9783585667610168,
      "learning_rate": 9.299078947368421e-05,
      "loss": 0.3348,
      "step": 46654
    },
    {
      "epoch": 2.612554597379326,
      "grad_norm": 1.2408232688903809,
      "learning_rate": 9.299052631578948e-05,
      "loss": 0.4249,
      "step": 46655
    },
    {
      "epoch": 2.612610594691455,
      "grad_norm": 1.1084270477294922,
      "learning_rate": 9.299026315789474e-05,
      "loss": 0.408,
      "step": 46656
    },
    {
      "epoch": 2.612666592003584,
      "grad_norm": 1.1768100261688232,
      "learning_rate": 9.299e-05,
      "loss": 0.3857,
      "step": 46657
    },
    {
      "epoch": 2.612722589315713,
      "grad_norm": 1.1828205585479736,
      "learning_rate": 9.298973684210526e-05,
      "loss": 0.3799,
      "step": 46658
    },
    {
      "epoch": 2.612778586627842,
      "grad_norm": 1.1990939378738403,
      "learning_rate": 9.298947368421052e-05,
      "loss": 0.4627,
      "step": 46659
    },
    {
      "epoch": 2.612834583939971,
      "grad_norm": 1.1285840272903442,
      "learning_rate": 9.29892105263158e-05,
      "loss": 0.3348,
      "step": 46660
    },
    {
      "epoch": 2.6128905812521,
      "grad_norm": 1.2289525270462036,
      "learning_rate": 9.298894736842105e-05,
      "loss": 0.363,
      "step": 46661
    },
    {
      "epoch": 2.612946578564229,
      "grad_norm": 1.4180353879928589,
      "learning_rate": 9.298868421052633e-05,
      "loss": 0.3898,
      "step": 46662
    },
    {
      "epoch": 2.613002575876358,
      "grad_norm": 1.1110385656356812,
      "learning_rate": 9.298842105263157e-05,
      "loss": 0.3456,
      "step": 46663
    },
    {
      "epoch": 2.613058573188487,
      "grad_norm": 1.4971394538879395,
      "learning_rate": 9.298815789473685e-05,
      "loss": 0.4093,
      "step": 46664
    },
    {
      "epoch": 2.613114570500616,
      "grad_norm": 1.0828325748443604,
      "learning_rate": 9.298789473684211e-05,
      "loss": 0.443,
      "step": 46665
    },
    {
      "epoch": 2.613170567812745,
      "grad_norm": 1.202742338180542,
      "learning_rate": 9.298763157894738e-05,
      "loss": 0.4079,
      "step": 46666
    },
    {
      "epoch": 2.613226565124874,
      "grad_norm": 1.0951130390167236,
      "learning_rate": 9.298736842105264e-05,
      "loss": 0.4235,
      "step": 46667
    },
    {
      "epoch": 2.613282562437003,
      "grad_norm": 1.319145917892456,
      "learning_rate": 9.29871052631579e-05,
      "loss": 0.4171,
      "step": 46668
    },
    {
      "epoch": 2.613338559749132,
      "grad_norm": 1.2181124687194824,
      "learning_rate": 9.298684210526316e-05,
      "loss": 0.4251,
      "step": 46669
    },
    {
      "epoch": 2.613394557061261,
      "grad_norm": 1.2492749691009521,
      "learning_rate": 9.298657894736843e-05,
      "loss": 0.4055,
      "step": 46670
    },
    {
      "epoch": 2.61345055437339,
      "grad_norm": 1.1912407875061035,
      "learning_rate": 9.298631578947369e-05,
      "loss": 0.441,
      "step": 46671
    },
    {
      "epoch": 2.613506551685519,
      "grad_norm": 1.2228444814682007,
      "learning_rate": 9.298605263157895e-05,
      "loss": 0.4807,
      "step": 46672
    },
    {
      "epoch": 2.613562548997648,
      "grad_norm": 0.9490548372268677,
      "learning_rate": 9.298578947368421e-05,
      "loss": 0.2669,
      "step": 46673
    },
    {
      "epoch": 2.613618546309777,
      "grad_norm": 1.1885175704956055,
      "learning_rate": 9.298552631578947e-05,
      "loss": 0.4388,
      "step": 46674
    },
    {
      "epoch": 2.613674543621906,
      "grad_norm": 1.39473557472229,
      "learning_rate": 9.298526315789475e-05,
      "loss": 0.432,
      "step": 46675
    },
    {
      "epoch": 2.6137305409340352,
      "grad_norm": 1.1211905479431152,
      "learning_rate": 9.2985e-05,
      "loss": 0.5124,
      "step": 46676
    },
    {
      "epoch": 2.6137865382461642,
      "grad_norm": 1.2203218936920166,
      "learning_rate": 9.298473684210526e-05,
      "loss": 0.4477,
      "step": 46677
    },
    {
      "epoch": 2.6138425355582933,
      "grad_norm": 1.369370460510254,
      "learning_rate": 9.298447368421052e-05,
      "loss": 0.3663,
      "step": 46678
    },
    {
      "epoch": 2.6138985328704223,
      "grad_norm": 1.027858853340149,
      "learning_rate": 9.29842105263158e-05,
      "loss": 0.4191,
      "step": 46679
    },
    {
      "epoch": 2.6139545301825513,
      "grad_norm": 1.3899892568588257,
      "learning_rate": 9.298394736842106e-05,
      "loss": 0.4292,
      "step": 46680
    },
    {
      "epoch": 2.6140105274946803,
      "grad_norm": 1.0923852920532227,
      "learning_rate": 9.298368421052632e-05,
      "loss": 0.3487,
      "step": 46681
    },
    {
      "epoch": 2.6140665248068093,
      "grad_norm": 1.1200538873672485,
      "learning_rate": 9.298342105263158e-05,
      "loss": 0.3073,
      "step": 46682
    },
    {
      "epoch": 2.6141225221189384,
      "grad_norm": 1.6595849990844727,
      "learning_rate": 9.298315789473685e-05,
      "loss": 0.5936,
      "step": 46683
    },
    {
      "epoch": 2.6141785194310674,
      "grad_norm": 1.1285338401794434,
      "learning_rate": 9.298289473684211e-05,
      "loss": 0.3796,
      "step": 46684
    },
    {
      "epoch": 2.6142345167431964,
      "grad_norm": 1.2834473848342896,
      "learning_rate": 9.298263157894738e-05,
      "loss": 0.3864,
      "step": 46685
    },
    {
      "epoch": 2.6142905140553254,
      "grad_norm": 1.5202257633209229,
      "learning_rate": 9.298236842105263e-05,
      "loss": 0.5937,
      "step": 46686
    },
    {
      "epoch": 2.6143465113674544,
      "grad_norm": 1.0306605100631714,
      "learning_rate": 9.29821052631579e-05,
      "loss": 0.3639,
      "step": 46687
    },
    {
      "epoch": 2.6144025086795835,
      "grad_norm": 1.1619771718978882,
      "learning_rate": 9.298184210526316e-05,
      "loss": 0.3971,
      "step": 46688
    },
    {
      "epoch": 2.6144585059917125,
      "grad_norm": 1.212755560874939,
      "learning_rate": 9.298157894736844e-05,
      "loss": 0.4113,
      "step": 46689
    },
    {
      "epoch": 2.6145145033038415,
      "grad_norm": 1.5457602739334106,
      "learning_rate": 9.298131578947368e-05,
      "loss": 0.395,
      "step": 46690
    },
    {
      "epoch": 2.6145705006159705,
      "grad_norm": 0.9748319387435913,
      "learning_rate": 9.298105263157894e-05,
      "loss": 0.371,
      "step": 46691
    },
    {
      "epoch": 2.6146264979280995,
      "grad_norm": 1.1145292520523071,
      "learning_rate": 9.298078947368421e-05,
      "loss": 0.4219,
      "step": 46692
    },
    {
      "epoch": 2.6146824952402286,
      "grad_norm": 1.1746282577514648,
      "learning_rate": 9.298052631578947e-05,
      "loss": 0.3978,
      "step": 46693
    },
    {
      "epoch": 2.6147384925523576,
      "grad_norm": 1.0057501792907715,
      "learning_rate": 9.298026315789475e-05,
      "loss": 0.3702,
      "step": 46694
    },
    {
      "epoch": 2.6147944898644866,
      "grad_norm": 1.2233495712280273,
      "learning_rate": 9.298e-05,
      "loss": 0.4914,
      "step": 46695
    },
    {
      "epoch": 2.6148504871766156,
      "grad_norm": 2.321775197982788,
      "learning_rate": 9.297973684210527e-05,
      "loss": 0.521,
      "step": 46696
    },
    {
      "epoch": 2.6149064844887446,
      "grad_norm": 1.1047756671905518,
      "learning_rate": 9.297947368421053e-05,
      "loss": 0.2826,
      "step": 46697
    },
    {
      "epoch": 2.6149624818008737,
      "grad_norm": 1.0921597480773926,
      "learning_rate": 9.29792105263158e-05,
      "loss": 0.4113,
      "step": 46698
    },
    {
      "epoch": 2.6150184791130027,
      "grad_norm": 1.0332170724868774,
      "learning_rate": 9.297894736842106e-05,
      "loss": 0.3345,
      "step": 46699
    },
    {
      "epoch": 2.6150744764251317,
      "grad_norm": 0.9269431233406067,
      "learning_rate": 9.297868421052632e-05,
      "loss": 0.2807,
      "step": 46700
    },
    {
      "epoch": 2.6151304737372607,
      "grad_norm": 1.647600531578064,
      "learning_rate": 9.297842105263158e-05,
      "loss": 0.3338,
      "step": 46701
    },
    {
      "epoch": 2.6151864710493897,
      "grad_norm": 1.2438921928405762,
      "learning_rate": 9.297815789473685e-05,
      "loss": 0.3225,
      "step": 46702
    },
    {
      "epoch": 2.6152424683615187,
      "grad_norm": 0.9726871252059937,
      "learning_rate": 9.297789473684211e-05,
      "loss": 0.3034,
      "step": 46703
    },
    {
      "epoch": 2.6152984656736478,
      "grad_norm": 1.1678202152252197,
      "learning_rate": 9.297763157894737e-05,
      "loss": 0.3353,
      "step": 46704
    },
    {
      "epoch": 2.615354462985777,
      "grad_norm": 1.0944322347640991,
      "learning_rate": 9.297736842105263e-05,
      "loss": 0.3435,
      "step": 46705
    },
    {
      "epoch": 2.615410460297906,
      "grad_norm": 1.160312533378601,
      "learning_rate": 9.29771052631579e-05,
      "loss": 0.4908,
      "step": 46706
    },
    {
      "epoch": 2.615466457610035,
      "grad_norm": 1.211646318435669,
      "learning_rate": 9.297684210526317e-05,
      "loss": 0.4731,
      "step": 46707
    },
    {
      "epoch": 2.615522454922164,
      "grad_norm": 1.1735705137252808,
      "learning_rate": 9.297657894736842e-05,
      "loss": 0.3869,
      "step": 46708
    },
    {
      "epoch": 2.615578452234293,
      "grad_norm": 1.3701469898223877,
      "learning_rate": 9.297631578947368e-05,
      "loss": 0.3438,
      "step": 46709
    },
    {
      "epoch": 2.615634449546422,
      "grad_norm": 2.0871517658233643,
      "learning_rate": 9.297605263157894e-05,
      "loss": 0.3194,
      "step": 46710
    },
    {
      "epoch": 2.615690446858551,
      "grad_norm": 1.24550199508667,
      "learning_rate": 9.297578947368422e-05,
      "loss": 0.3119,
      "step": 46711
    },
    {
      "epoch": 2.61574644417068,
      "grad_norm": 1.0324658155441284,
      "learning_rate": 9.297552631578948e-05,
      "loss": 0.3611,
      "step": 46712
    },
    {
      "epoch": 2.615802441482809,
      "grad_norm": 1.21067476272583,
      "learning_rate": 9.297526315789474e-05,
      "loss": 0.3275,
      "step": 46713
    },
    {
      "epoch": 2.615858438794938,
      "grad_norm": 1.3327478170394897,
      "learning_rate": 9.2975e-05,
      "loss": 0.3432,
      "step": 46714
    },
    {
      "epoch": 2.615914436107067,
      "grad_norm": 1.0227373838424683,
      "learning_rate": 9.297473684210527e-05,
      "loss": 0.3751,
      "step": 46715
    },
    {
      "epoch": 2.615970433419196,
      "grad_norm": 1.2362167835235596,
      "learning_rate": 9.297447368421053e-05,
      "loss": 0.4402,
      "step": 46716
    },
    {
      "epoch": 2.616026430731325,
      "grad_norm": 1.2102203369140625,
      "learning_rate": 9.29742105263158e-05,
      "loss": 0.4519,
      "step": 46717
    },
    {
      "epoch": 2.616082428043454,
      "grad_norm": 1.2363210916519165,
      "learning_rate": 9.297394736842105e-05,
      "loss": 0.446,
      "step": 46718
    },
    {
      "epoch": 2.616138425355583,
      "grad_norm": 1.0664886236190796,
      "learning_rate": 9.297368421052632e-05,
      "loss": 0.4039,
      "step": 46719
    },
    {
      "epoch": 2.616194422667712,
      "grad_norm": 1.5525882244110107,
      "learning_rate": 9.297342105263158e-05,
      "loss": 0.395,
      "step": 46720
    },
    {
      "epoch": 2.616250419979841,
      "grad_norm": 1.2278224229812622,
      "learning_rate": 9.297315789473686e-05,
      "loss": 0.4857,
      "step": 46721
    },
    {
      "epoch": 2.61630641729197,
      "grad_norm": 1.246382236480713,
      "learning_rate": 9.297289473684212e-05,
      "loss": 0.4867,
      "step": 46722
    },
    {
      "epoch": 2.616362414604099,
      "grad_norm": 1.1361393928527832,
      "learning_rate": 9.297263157894737e-05,
      "loss": 0.5091,
      "step": 46723
    },
    {
      "epoch": 2.616418411916228,
      "grad_norm": 1.275478720664978,
      "learning_rate": 9.297236842105263e-05,
      "loss": 0.542,
      "step": 46724
    },
    {
      "epoch": 2.616474409228357,
      "grad_norm": 1.2570104598999023,
      "learning_rate": 9.29721052631579e-05,
      "loss": 0.4444,
      "step": 46725
    },
    {
      "epoch": 2.616530406540486,
      "grad_norm": 1.2875245809555054,
      "learning_rate": 9.297184210526317e-05,
      "loss": 0.5084,
      "step": 46726
    },
    {
      "epoch": 2.616586403852615,
      "grad_norm": 1.2645480632781982,
      "learning_rate": 9.297157894736843e-05,
      "loss": 0.4481,
      "step": 46727
    },
    {
      "epoch": 2.6166424011647442,
      "grad_norm": 1.5479607582092285,
      "learning_rate": 9.297131578947369e-05,
      "loss": 0.5425,
      "step": 46728
    },
    {
      "epoch": 2.6166983984768732,
      "grad_norm": 1.2086495161056519,
      "learning_rate": 9.297105263157895e-05,
      "loss": 0.4281,
      "step": 46729
    },
    {
      "epoch": 2.6167543957890023,
      "grad_norm": 1.4190393686294556,
      "learning_rate": 9.297078947368422e-05,
      "loss": 0.6183,
      "step": 46730
    },
    {
      "epoch": 2.6168103931011313,
      "grad_norm": 1.3730733394622803,
      "learning_rate": 9.297052631578948e-05,
      "loss": 0.4142,
      "step": 46731
    },
    {
      "epoch": 2.6168663904132603,
      "grad_norm": 1.212880253791809,
      "learning_rate": 9.297026315789474e-05,
      "loss": 0.365,
      "step": 46732
    },
    {
      "epoch": 2.6169223877253893,
      "grad_norm": 1.1551216840744019,
      "learning_rate": 9.297e-05,
      "loss": 0.4941,
      "step": 46733
    },
    {
      "epoch": 2.6169783850375183,
      "grad_norm": 1.172528624534607,
      "learning_rate": 9.296973684210527e-05,
      "loss": 0.4252,
      "step": 46734
    },
    {
      "epoch": 2.6170343823496474,
      "grad_norm": 1.14329195022583,
      "learning_rate": 9.296947368421053e-05,
      "loss": 0.5225,
      "step": 46735
    },
    {
      "epoch": 2.6170903796617764,
      "grad_norm": 1.9598976373672485,
      "learning_rate": 9.296921052631579e-05,
      "loss": 0.4284,
      "step": 46736
    },
    {
      "epoch": 2.6171463769739054,
      "grad_norm": 0.9865350127220154,
      "learning_rate": 9.296894736842105e-05,
      "loss": 0.3303,
      "step": 46737
    },
    {
      "epoch": 2.6172023742860344,
      "grad_norm": 1.1210765838623047,
      "learning_rate": 9.296868421052632e-05,
      "loss": 0.3316,
      "step": 46738
    },
    {
      "epoch": 2.6172583715981634,
      "grad_norm": 1.0898633003234863,
      "learning_rate": 9.296842105263158e-05,
      "loss": 0.4742,
      "step": 46739
    },
    {
      "epoch": 2.6173143689102925,
      "grad_norm": 1.0996369123458862,
      "learning_rate": 9.296815789473686e-05,
      "loss": 0.2971,
      "step": 46740
    },
    {
      "epoch": 2.6173703662224215,
      "grad_norm": 1.3314951658248901,
      "learning_rate": 9.29678947368421e-05,
      "loss": 0.4739,
      "step": 46741
    },
    {
      "epoch": 2.6174263635345505,
      "grad_norm": 1.0952012538909912,
      "learning_rate": 9.296763157894738e-05,
      "loss": 0.3667,
      "step": 46742
    },
    {
      "epoch": 2.6174823608466795,
      "grad_norm": 1.3140828609466553,
      "learning_rate": 9.296736842105264e-05,
      "loss": 0.3883,
      "step": 46743
    },
    {
      "epoch": 2.6175383581588085,
      "grad_norm": 1.1674870252609253,
      "learning_rate": 9.29671052631579e-05,
      "loss": 0.3587,
      "step": 46744
    },
    {
      "epoch": 2.6175943554709376,
      "grad_norm": 1.0534330606460571,
      "learning_rate": 9.296684210526316e-05,
      "loss": 0.3889,
      "step": 46745
    },
    {
      "epoch": 2.6176503527830666,
      "grad_norm": 1.3798774480819702,
      "learning_rate": 9.296657894736842e-05,
      "loss": 0.518,
      "step": 46746
    },
    {
      "epoch": 2.6177063500951956,
      "grad_norm": 1.4897041320800781,
      "learning_rate": 9.296631578947369e-05,
      "loss": 0.3668,
      "step": 46747
    },
    {
      "epoch": 2.6177623474073246,
      "grad_norm": 1.3185303211212158,
      "learning_rate": 9.296605263157895e-05,
      "loss": 0.2848,
      "step": 46748
    },
    {
      "epoch": 2.6178183447194536,
      "grad_norm": 0.9920732378959656,
      "learning_rate": 9.296578947368422e-05,
      "loss": 0.3637,
      "step": 46749
    },
    {
      "epoch": 2.6178743420315826,
      "grad_norm": 1.2335031032562256,
      "learning_rate": 9.296552631578947e-05,
      "loss": 0.5493,
      "step": 46750
    },
    {
      "epoch": 2.6179303393437117,
      "grad_norm": 1.6651972532272339,
      "learning_rate": 9.296526315789474e-05,
      "loss": 0.4698,
      "step": 46751
    },
    {
      "epoch": 2.6179863366558407,
      "grad_norm": 1.4533491134643555,
      "learning_rate": 9.2965e-05,
      "loss": 0.4126,
      "step": 46752
    },
    {
      "epoch": 2.6180423339679697,
      "grad_norm": 1.130131483078003,
      "learning_rate": 9.296473684210528e-05,
      "loss": 0.3973,
      "step": 46753
    },
    {
      "epoch": 2.6180983312800983,
      "grad_norm": 1.3289220333099365,
      "learning_rate": 9.296447368421053e-05,
      "loss": 0.478,
      "step": 46754
    },
    {
      "epoch": 2.6181543285922277,
      "grad_norm": 1.2845627069473267,
      "learning_rate": 9.29642105263158e-05,
      "loss": 0.4362,
      "step": 46755
    },
    {
      "epoch": 2.6182103259043563,
      "grad_norm": 1.0131199359893799,
      "learning_rate": 9.296394736842105e-05,
      "loss": 0.414,
      "step": 46756
    },
    {
      "epoch": 2.618266323216486,
      "grad_norm": 1.1362448930740356,
      "learning_rate": 9.296368421052633e-05,
      "loss": 0.3894,
      "step": 46757
    },
    {
      "epoch": 2.6183223205286144,
      "grad_norm": 1.0010775327682495,
      "learning_rate": 9.296342105263159e-05,
      "loss": 0.3246,
      "step": 46758
    },
    {
      "epoch": 2.618378317840744,
      "grad_norm": 1.145743727684021,
      "learning_rate": 9.296315789473685e-05,
      "loss": 0.3619,
      "step": 46759
    },
    {
      "epoch": 2.6184343151528724,
      "grad_norm": 1.2314954996109009,
      "learning_rate": 9.29628947368421e-05,
      "loss": 0.4092,
      "step": 46760
    },
    {
      "epoch": 2.618490312465002,
      "grad_norm": 1.763024926185608,
      "learning_rate": 9.296263157894737e-05,
      "loss": 0.4066,
      "step": 46761
    },
    {
      "epoch": 2.6185463097771304,
      "grad_norm": 1.14560866355896,
      "learning_rate": 9.296236842105264e-05,
      "loss": 0.3184,
      "step": 46762
    },
    {
      "epoch": 2.61860230708926,
      "grad_norm": 1.3619943857192993,
      "learning_rate": 9.29621052631579e-05,
      "loss": 0.4676,
      "step": 46763
    },
    {
      "epoch": 2.6186583044013885,
      "grad_norm": 1.2551202774047852,
      "learning_rate": 9.296184210526316e-05,
      "loss": 0.4017,
      "step": 46764
    },
    {
      "epoch": 2.618714301713518,
      "grad_norm": 1.1253986358642578,
      "learning_rate": 9.296157894736842e-05,
      "loss": 0.3926,
      "step": 46765
    },
    {
      "epoch": 2.6187702990256465,
      "grad_norm": 1.3530446290969849,
      "learning_rate": 9.296131578947369e-05,
      "loss": 0.5482,
      "step": 46766
    },
    {
      "epoch": 2.618826296337776,
      "grad_norm": 1.4351688623428345,
      "learning_rate": 9.296105263157895e-05,
      "loss": 0.3681,
      "step": 46767
    },
    {
      "epoch": 2.6188822936499045,
      "grad_norm": 1.9886863231658936,
      "learning_rate": 9.296078947368421e-05,
      "loss": 0.4713,
      "step": 46768
    },
    {
      "epoch": 2.618938290962034,
      "grad_norm": 1.2003042697906494,
      "learning_rate": 9.296052631578947e-05,
      "loss": 0.4211,
      "step": 46769
    },
    {
      "epoch": 2.6189942882741626,
      "grad_norm": 1.2783092260360718,
      "learning_rate": 9.296026315789474e-05,
      "loss": 0.5489,
      "step": 46770
    },
    {
      "epoch": 2.619050285586292,
      "grad_norm": 1.1137007474899292,
      "learning_rate": 9.296e-05,
      "loss": 0.2743,
      "step": 46771
    },
    {
      "epoch": 2.6191062828984206,
      "grad_norm": 0.9794380068778992,
      "learning_rate": 9.295973684210528e-05,
      "loss": 0.3035,
      "step": 46772
    },
    {
      "epoch": 2.61916228021055,
      "grad_norm": 1.138496994972229,
      "learning_rate": 9.295947368421052e-05,
      "loss": 0.4509,
      "step": 46773
    },
    {
      "epoch": 2.6192182775226787,
      "grad_norm": 1.253280520439148,
      "learning_rate": 9.29592105263158e-05,
      "loss": 0.3246,
      "step": 46774
    },
    {
      "epoch": 2.619274274834808,
      "grad_norm": 1.2355107069015503,
      "learning_rate": 9.295894736842106e-05,
      "loss": 0.442,
      "step": 46775
    },
    {
      "epoch": 2.6193302721469367,
      "grad_norm": 1.708880066871643,
      "learning_rate": 9.295868421052633e-05,
      "loss": 0.4466,
      "step": 46776
    },
    {
      "epoch": 2.619386269459066,
      "grad_norm": 1.1865224838256836,
      "learning_rate": 9.295842105263159e-05,
      "loss": 0.3459,
      "step": 46777
    },
    {
      "epoch": 2.6194422667711947,
      "grad_norm": 1.3466836214065552,
      "learning_rate": 9.295815789473684e-05,
      "loss": 0.4544,
      "step": 46778
    },
    {
      "epoch": 2.619498264083324,
      "grad_norm": 1.2605745792388916,
      "learning_rate": 9.295789473684211e-05,
      "loss": 0.6802,
      "step": 46779
    },
    {
      "epoch": 2.6195542613954528,
      "grad_norm": 1.003353238105774,
      "learning_rate": 9.295763157894737e-05,
      "loss": 0.3653,
      "step": 46780
    },
    {
      "epoch": 2.6196102587075822,
      "grad_norm": 1.4013078212738037,
      "learning_rate": 9.295736842105264e-05,
      "loss": 0.3567,
      "step": 46781
    },
    {
      "epoch": 2.619666256019711,
      "grad_norm": 1.143011450767517,
      "learning_rate": 9.29571052631579e-05,
      "loss": 0.4626,
      "step": 46782
    },
    {
      "epoch": 2.6197222533318403,
      "grad_norm": 1.1356501579284668,
      "learning_rate": 9.295684210526316e-05,
      "loss": 0.3323,
      "step": 46783
    },
    {
      "epoch": 2.619778250643969,
      "grad_norm": 1.2056152820587158,
      "learning_rate": 9.295657894736842e-05,
      "loss": 0.52,
      "step": 46784
    },
    {
      "epoch": 2.6198342479560983,
      "grad_norm": 1.6221848726272583,
      "learning_rate": 9.29563157894737e-05,
      "loss": 0.4295,
      "step": 46785
    },
    {
      "epoch": 2.619890245268227,
      "grad_norm": 1.4846909046173096,
      "learning_rate": 9.295605263157895e-05,
      "loss": 0.6415,
      "step": 46786
    },
    {
      "epoch": 2.6199462425803564,
      "grad_norm": 1.3210195302963257,
      "learning_rate": 9.295578947368421e-05,
      "loss": 0.561,
      "step": 46787
    },
    {
      "epoch": 2.620002239892485,
      "grad_norm": 1.148607850074768,
      "learning_rate": 9.295552631578947e-05,
      "loss": 0.5336,
      "step": 46788
    },
    {
      "epoch": 2.6200582372046144,
      "grad_norm": 1.1712279319763184,
      "learning_rate": 9.295526315789475e-05,
      "loss": 0.5014,
      "step": 46789
    },
    {
      "epoch": 2.620114234516743,
      "grad_norm": 1.163711667060852,
      "learning_rate": 9.295500000000001e-05,
      "loss": 0.2981,
      "step": 46790
    },
    {
      "epoch": 2.6201702318288724,
      "grad_norm": 1.3419890403747559,
      "learning_rate": 9.295473684210527e-05,
      "loss": 0.4093,
      "step": 46791
    },
    {
      "epoch": 2.620226229141001,
      "grad_norm": 1.1486859321594238,
      "learning_rate": 9.295447368421053e-05,
      "loss": 0.3793,
      "step": 46792
    },
    {
      "epoch": 2.6202822264531305,
      "grad_norm": 1.024530291557312,
      "learning_rate": 9.29542105263158e-05,
      "loss": 0.4485,
      "step": 46793
    },
    {
      "epoch": 2.620338223765259,
      "grad_norm": 1.150249719619751,
      "learning_rate": 9.295394736842106e-05,
      "loss": 0.3444,
      "step": 46794
    },
    {
      "epoch": 2.6203942210773885,
      "grad_norm": 1.1997238397598267,
      "learning_rate": 9.295368421052632e-05,
      "loss": 0.4375,
      "step": 46795
    },
    {
      "epoch": 2.620450218389517,
      "grad_norm": 1.2190262079238892,
      "learning_rate": 9.295342105263158e-05,
      "loss": 0.3787,
      "step": 46796
    },
    {
      "epoch": 2.6205062157016465,
      "grad_norm": 1.053983211517334,
      "learning_rate": 9.295315789473684e-05,
      "loss": 0.3154,
      "step": 46797
    },
    {
      "epoch": 2.620562213013775,
      "grad_norm": 1.0090112686157227,
      "learning_rate": 9.295289473684211e-05,
      "loss": 0.3573,
      "step": 46798
    },
    {
      "epoch": 2.6206182103259046,
      "grad_norm": 1.162980556488037,
      "learning_rate": 9.295263157894737e-05,
      "loss": 0.3862,
      "step": 46799
    },
    {
      "epoch": 2.620674207638033,
      "grad_norm": 1.1985747814178467,
      "learning_rate": 9.295236842105263e-05,
      "loss": 0.4161,
      "step": 46800
    },
    {
      "epoch": 2.6207302049501626,
      "grad_norm": 1.1420365571975708,
      "learning_rate": 9.295210526315789e-05,
      "loss": 0.4306,
      "step": 46801
    },
    {
      "epoch": 2.620786202262291,
      "grad_norm": 0.9453822374343872,
      "learning_rate": 9.295184210526316e-05,
      "loss": 0.3628,
      "step": 46802
    },
    {
      "epoch": 2.6208421995744207,
      "grad_norm": 1.2848963737487793,
      "learning_rate": 9.295157894736842e-05,
      "loss": 0.3084,
      "step": 46803
    },
    {
      "epoch": 2.6208981968865492,
      "grad_norm": 1.2864497900009155,
      "learning_rate": 9.29513157894737e-05,
      "loss": 0.38,
      "step": 46804
    },
    {
      "epoch": 2.6209541941986787,
      "grad_norm": 1.0188215970993042,
      "learning_rate": 9.295105263157894e-05,
      "loss": 0.358,
      "step": 46805
    },
    {
      "epoch": 2.6210101915108073,
      "grad_norm": 1.063781976699829,
      "learning_rate": 9.295078947368422e-05,
      "loss": 0.3996,
      "step": 46806
    },
    {
      "epoch": 2.6210661888229367,
      "grad_norm": 1.3126211166381836,
      "learning_rate": 9.295052631578948e-05,
      "loss": 0.4277,
      "step": 46807
    },
    {
      "epoch": 2.6211221861350653,
      "grad_norm": 1.275341510772705,
      "learning_rate": 9.295026315789475e-05,
      "loss": 0.3897,
      "step": 46808
    },
    {
      "epoch": 2.6211781834471948,
      "grad_norm": 1.46393620967865,
      "learning_rate": 9.295000000000001e-05,
      "loss": 0.3813,
      "step": 46809
    },
    {
      "epoch": 2.6212341807593234,
      "grad_norm": 1.3119186162948608,
      "learning_rate": 9.294973684210527e-05,
      "loss": 0.4948,
      "step": 46810
    },
    {
      "epoch": 2.621290178071453,
      "grad_norm": 1.063950777053833,
      "learning_rate": 9.294947368421053e-05,
      "loss": 0.4053,
      "step": 46811
    },
    {
      "epoch": 2.6213461753835814,
      "grad_norm": 1.1416627168655396,
      "learning_rate": 9.294921052631579e-05,
      "loss": 0.4408,
      "step": 46812
    },
    {
      "epoch": 2.621402172695711,
      "grad_norm": 1.0680553913116455,
      "learning_rate": 9.294894736842106e-05,
      "loss": 0.4368,
      "step": 46813
    },
    {
      "epoch": 2.6214581700078394,
      "grad_norm": 1.3229248523712158,
      "learning_rate": 9.294868421052632e-05,
      "loss": 0.3357,
      "step": 46814
    },
    {
      "epoch": 2.621514167319969,
      "grad_norm": 1.49494206905365,
      "learning_rate": 9.294842105263158e-05,
      "loss": 0.3979,
      "step": 46815
    },
    {
      "epoch": 2.6215701646320975,
      "grad_norm": 1.183313012123108,
      "learning_rate": 9.294815789473684e-05,
      "loss": 0.3195,
      "step": 46816
    },
    {
      "epoch": 2.621626161944227,
      "grad_norm": 1.0830750465393066,
      "learning_rate": 9.294789473684211e-05,
      "loss": 0.4149,
      "step": 46817
    },
    {
      "epoch": 2.6216821592563555,
      "grad_norm": 1.0247219800949097,
      "learning_rate": 9.294763157894737e-05,
      "loss": 0.2976,
      "step": 46818
    },
    {
      "epoch": 2.621738156568485,
      "grad_norm": 1.2096703052520752,
      "learning_rate": 9.294736842105263e-05,
      "loss": 0.3536,
      "step": 46819
    },
    {
      "epoch": 2.6217941538806135,
      "grad_norm": 1.2946943044662476,
      "learning_rate": 9.29471052631579e-05,
      "loss": 0.5974,
      "step": 46820
    },
    {
      "epoch": 2.621850151192743,
      "grad_norm": 1.3831393718719482,
      "learning_rate": 9.294684210526317e-05,
      "loss": 0.4063,
      "step": 46821
    },
    {
      "epoch": 2.6219061485048716,
      "grad_norm": 1.3210722208023071,
      "learning_rate": 9.294657894736843e-05,
      "loss": 0.4165,
      "step": 46822
    },
    {
      "epoch": 2.621962145817001,
      "grad_norm": 1.2941522598266602,
      "learning_rate": 9.294631578947369e-05,
      "loss": 0.4364,
      "step": 46823
    },
    {
      "epoch": 2.6220181431291296,
      "grad_norm": 1.3078821897506714,
      "learning_rate": 9.294605263157895e-05,
      "loss": 0.4047,
      "step": 46824
    },
    {
      "epoch": 2.622074140441259,
      "grad_norm": 1.2820583581924438,
      "learning_rate": 9.294578947368422e-05,
      "loss": 0.4297,
      "step": 46825
    },
    {
      "epoch": 2.6221301377533877,
      "grad_norm": 1.212156891822815,
      "learning_rate": 9.294552631578948e-05,
      "loss": 0.4591,
      "step": 46826
    },
    {
      "epoch": 2.622186135065517,
      "grad_norm": 1.0013808012008667,
      "learning_rate": 9.294526315789475e-05,
      "loss": 0.3915,
      "step": 46827
    },
    {
      "epoch": 2.6222421323776457,
      "grad_norm": 1.1047852039337158,
      "learning_rate": 9.2945e-05,
      "loss": 0.3849,
      "step": 46828
    },
    {
      "epoch": 2.622298129689775,
      "grad_norm": 1.251317024230957,
      "learning_rate": 9.294473684210526e-05,
      "loss": 0.4171,
      "step": 46829
    },
    {
      "epoch": 2.6223541270019037,
      "grad_norm": 1.0286364555358887,
      "learning_rate": 9.294447368421053e-05,
      "loss": 0.3906,
      "step": 46830
    },
    {
      "epoch": 2.622410124314033,
      "grad_norm": 1.1840567588806152,
      "learning_rate": 9.294421052631579e-05,
      "loss": 0.4472,
      "step": 46831
    },
    {
      "epoch": 2.6224661216261618,
      "grad_norm": 1.273462176322937,
      "learning_rate": 9.294394736842106e-05,
      "loss": 0.6608,
      "step": 46832
    },
    {
      "epoch": 2.6225221189382912,
      "grad_norm": 1.1367168426513672,
      "learning_rate": 9.294368421052631e-05,
      "loss": 0.3169,
      "step": 46833
    },
    {
      "epoch": 2.62257811625042,
      "grad_norm": 1.3610601425170898,
      "learning_rate": 9.294342105263158e-05,
      "loss": 0.4391,
      "step": 46834
    },
    {
      "epoch": 2.6226341135625493,
      "grad_norm": 1.3763405084609985,
      "learning_rate": 9.294315789473684e-05,
      "loss": 0.4942,
      "step": 46835
    },
    {
      "epoch": 2.622690110874678,
      "grad_norm": 1.277869701385498,
      "learning_rate": 9.294289473684212e-05,
      "loss": 0.4401,
      "step": 46836
    },
    {
      "epoch": 2.622746108186807,
      "grad_norm": 1.535495400428772,
      "learning_rate": 9.294263157894736e-05,
      "loss": 0.3942,
      "step": 46837
    },
    {
      "epoch": 2.622802105498936,
      "grad_norm": 1.0069700479507446,
      "learning_rate": 9.294236842105264e-05,
      "loss": 0.4211,
      "step": 46838
    },
    {
      "epoch": 2.622858102811065,
      "grad_norm": 1.3738114833831787,
      "learning_rate": 9.29421052631579e-05,
      "loss": 0.444,
      "step": 46839
    },
    {
      "epoch": 2.622914100123194,
      "grad_norm": 1.6657222509384155,
      "learning_rate": 9.294184210526317e-05,
      "loss": 0.5791,
      "step": 46840
    },
    {
      "epoch": 2.622970097435323,
      "grad_norm": 1.159819483757019,
      "learning_rate": 9.294157894736843e-05,
      "loss": 0.3734,
      "step": 46841
    },
    {
      "epoch": 2.623026094747452,
      "grad_norm": 1.6650866270065308,
      "learning_rate": 9.294131578947369e-05,
      "loss": 0.3606,
      "step": 46842
    },
    {
      "epoch": 2.623082092059581,
      "grad_norm": 1.0191733837127686,
      "learning_rate": 9.294105263157895e-05,
      "loss": 0.3686,
      "step": 46843
    },
    {
      "epoch": 2.62313808937171,
      "grad_norm": 1.2151050567626953,
      "learning_rate": 9.294078947368422e-05,
      "loss": 0.421,
      "step": 46844
    },
    {
      "epoch": 2.623194086683839,
      "grad_norm": 1.4377672672271729,
      "learning_rate": 9.294052631578948e-05,
      "loss": 0.3522,
      "step": 46845
    },
    {
      "epoch": 2.623250083995968,
      "grad_norm": 1.1337298154830933,
      "learning_rate": 9.294026315789474e-05,
      "loss": 0.4244,
      "step": 46846
    },
    {
      "epoch": 2.623306081308097,
      "grad_norm": 0.9996371865272522,
      "learning_rate": 9.294e-05,
      "loss": 0.3824,
      "step": 46847
    },
    {
      "epoch": 2.623362078620226,
      "grad_norm": 1.2200084924697876,
      "learning_rate": 9.293973684210526e-05,
      "loss": 0.4108,
      "step": 46848
    },
    {
      "epoch": 2.623418075932355,
      "grad_norm": 1.142985463142395,
      "learning_rate": 9.293947368421053e-05,
      "loss": 0.4335,
      "step": 46849
    },
    {
      "epoch": 2.623474073244484,
      "grad_norm": 1.1654499769210815,
      "learning_rate": 9.29392105263158e-05,
      "loss": 0.4249,
      "step": 46850
    },
    {
      "epoch": 2.623530070556613,
      "grad_norm": 0.9991153478622437,
      "learning_rate": 9.293894736842105e-05,
      "loss": 0.2508,
      "step": 46851
    },
    {
      "epoch": 2.623586067868742,
      "grad_norm": 1.164332628250122,
      "learning_rate": 9.293868421052631e-05,
      "loss": 0.3906,
      "step": 46852
    },
    {
      "epoch": 2.623642065180871,
      "grad_norm": 1.0930718183517456,
      "learning_rate": 9.293842105263159e-05,
      "loss": 0.4893,
      "step": 46853
    },
    {
      "epoch": 2.623698062493,
      "grad_norm": 1.1107914447784424,
      "learning_rate": 9.293815789473685e-05,
      "loss": 0.2895,
      "step": 46854
    },
    {
      "epoch": 2.623754059805129,
      "grad_norm": 1.153120517730713,
      "learning_rate": 9.29378947368421e-05,
      "loss": 0.469,
      "step": 46855
    },
    {
      "epoch": 2.6238100571172582,
      "grad_norm": 1.1557214260101318,
      "learning_rate": 9.293763157894737e-05,
      "loss": 0.3707,
      "step": 46856
    },
    {
      "epoch": 2.6238660544293873,
      "grad_norm": 1.144906759262085,
      "learning_rate": 9.293736842105264e-05,
      "loss": 0.3731,
      "step": 46857
    },
    {
      "epoch": 2.6239220517415163,
      "grad_norm": 1.194730520248413,
      "learning_rate": 9.29371052631579e-05,
      "loss": 0.4565,
      "step": 46858
    },
    {
      "epoch": 2.6239780490536453,
      "grad_norm": 1.2038601636886597,
      "learning_rate": 9.293684210526317e-05,
      "loss": 0.4596,
      "step": 46859
    },
    {
      "epoch": 2.6240340463657743,
      "grad_norm": 0.9323737025260925,
      "learning_rate": 9.293657894736842e-05,
      "loss": 0.3498,
      "step": 46860
    },
    {
      "epoch": 2.6240900436779033,
      "grad_norm": 1.0257657766342163,
      "learning_rate": 9.293631578947369e-05,
      "loss": 0.3493,
      "step": 46861
    },
    {
      "epoch": 2.6241460409900323,
      "grad_norm": 1.3857697248458862,
      "learning_rate": 9.293605263157895e-05,
      "loss": 0.3482,
      "step": 46862
    },
    {
      "epoch": 2.6242020383021614,
      "grad_norm": 1.1450566053390503,
      "learning_rate": 9.293578947368422e-05,
      "loss": 0.3862,
      "step": 46863
    },
    {
      "epoch": 2.6242580356142904,
      "grad_norm": 0.9813950061798096,
      "learning_rate": 9.293552631578948e-05,
      "loss": 0.3308,
      "step": 46864
    },
    {
      "epoch": 2.6243140329264194,
      "grad_norm": 1.206701397895813,
      "learning_rate": 9.293526315789473e-05,
      "loss": 0.384,
      "step": 46865
    },
    {
      "epoch": 2.6243700302385484,
      "grad_norm": 1.4730498790740967,
      "learning_rate": 9.2935e-05,
      "loss": 0.5002,
      "step": 46866
    },
    {
      "epoch": 2.6244260275506774,
      "grad_norm": 25.71799087524414,
      "learning_rate": 9.293473684210526e-05,
      "loss": 0.3861,
      "step": 46867
    },
    {
      "epoch": 2.6244820248628065,
      "grad_norm": 1.2271591424942017,
      "learning_rate": 9.293447368421054e-05,
      "loss": 0.3942,
      "step": 46868
    },
    {
      "epoch": 2.6245380221749355,
      "grad_norm": 1.3096116781234741,
      "learning_rate": 9.29342105263158e-05,
      "loss": 0.4031,
      "step": 46869
    },
    {
      "epoch": 2.6245940194870645,
      "grad_norm": 1.063765048980713,
      "learning_rate": 9.293394736842106e-05,
      "loss": 0.3613,
      "step": 46870
    },
    {
      "epoch": 2.6246500167991935,
      "grad_norm": 1.2308342456817627,
      "learning_rate": 9.293368421052632e-05,
      "loss": 0.4228,
      "step": 46871
    },
    {
      "epoch": 2.6247060141113225,
      "grad_norm": 1.0634870529174805,
      "learning_rate": 9.293342105263159e-05,
      "loss": 0.388,
      "step": 46872
    },
    {
      "epoch": 2.6247620114234516,
      "grad_norm": 1.1167595386505127,
      "learning_rate": 9.293315789473685e-05,
      "loss": 0.4569,
      "step": 46873
    },
    {
      "epoch": 2.6248180087355806,
      "grad_norm": 1.3785927295684814,
      "learning_rate": 9.293289473684211e-05,
      "loss": 0.4373,
      "step": 46874
    },
    {
      "epoch": 2.6248740060477096,
      "grad_norm": 1.1960278749465942,
      "learning_rate": 9.293263157894737e-05,
      "loss": 0.4188,
      "step": 46875
    },
    {
      "epoch": 2.6249300033598386,
      "grad_norm": 1.3218187093734741,
      "learning_rate": 9.293236842105264e-05,
      "loss": 0.3839,
      "step": 46876
    },
    {
      "epoch": 2.6249860006719676,
      "grad_norm": 0.9911853671073914,
      "learning_rate": 9.29321052631579e-05,
      "loss": 0.3853,
      "step": 46877
    },
    {
      "epoch": 2.6250419979840967,
      "grad_norm": 1.4362967014312744,
      "learning_rate": 9.293184210526316e-05,
      "loss": 0.3796,
      "step": 46878
    },
    {
      "epoch": 2.6250979952962257,
      "grad_norm": 1.0741127729415894,
      "learning_rate": 9.293157894736842e-05,
      "loss": 0.411,
      "step": 46879
    },
    {
      "epoch": 2.6251539926083547,
      "grad_norm": 1.2929226160049438,
      "learning_rate": 9.29313157894737e-05,
      "loss": 0.3197,
      "step": 46880
    },
    {
      "epoch": 2.6252099899204837,
      "grad_norm": 1.2102035284042358,
      "learning_rate": 9.293105263157895e-05,
      "loss": 0.2452,
      "step": 46881
    },
    {
      "epoch": 2.6252659872326127,
      "grad_norm": 1.6133804321289062,
      "learning_rate": 9.293078947368421e-05,
      "loss": 0.376,
      "step": 46882
    },
    {
      "epoch": 2.6253219845447417,
      "grad_norm": 1.11672842502594,
      "learning_rate": 9.293052631578947e-05,
      "loss": 0.4684,
      "step": 46883
    },
    {
      "epoch": 2.6253779818568708,
      "grad_norm": 1.1760737895965576,
      "learning_rate": 9.293026315789473e-05,
      "loss": 0.4643,
      "step": 46884
    },
    {
      "epoch": 2.625433979169,
      "grad_norm": 1.3039157390594482,
      "learning_rate": 9.293e-05,
      "loss": 0.375,
      "step": 46885
    },
    {
      "epoch": 2.625489976481129,
      "grad_norm": 1.256610631942749,
      "learning_rate": 9.292973684210527e-05,
      "loss": 0.3342,
      "step": 46886
    },
    {
      "epoch": 2.625545973793258,
      "grad_norm": 1.287553071975708,
      "learning_rate": 9.292947368421054e-05,
      "loss": 0.3702,
      "step": 46887
    },
    {
      "epoch": 2.625601971105387,
      "grad_norm": 1.1641291379928589,
      "learning_rate": 9.292921052631579e-05,
      "loss": 0.3848,
      "step": 46888
    },
    {
      "epoch": 2.625657968417516,
      "grad_norm": 1.2423171997070312,
      "learning_rate": 9.292894736842106e-05,
      "loss": 0.4469,
      "step": 46889
    },
    {
      "epoch": 2.625713965729645,
      "grad_norm": 1.170389175415039,
      "learning_rate": 9.292868421052632e-05,
      "loss": 0.3731,
      "step": 46890
    },
    {
      "epoch": 2.625769963041774,
      "grad_norm": 1.1667860746383667,
      "learning_rate": 9.292842105263159e-05,
      "loss": 0.4254,
      "step": 46891
    },
    {
      "epoch": 2.625825960353903,
      "grad_norm": 1.0317459106445312,
      "learning_rate": 9.292815789473684e-05,
      "loss": 0.3392,
      "step": 46892
    },
    {
      "epoch": 2.625881957666032,
      "grad_norm": 0.99285489320755,
      "learning_rate": 9.292789473684211e-05,
      "loss": 0.3249,
      "step": 46893
    },
    {
      "epoch": 2.625937954978161,
      "grad_norm": 1.5093568563461304,
      "learning_rate": 9.292763157894737e-05,
      "loss": 0.3336,
      "step": 46894
    },
    {
      "epoch": 2.62599395229029,
      "grad_norm": 0.966926097869873,
      "learning_rate": 9.292736842105264e-05,
      "loss": 0.3285,
      "step": 46895
    },
    {
      "epoch": 2.626049949602419,
      "grad_norm": 1.4612423181533813,
      "learning_rate": 9.29271052631579e-05,
      "loss": 0.4652,
      "step": 46896
    },
    {
      "epoch": 2.626105946914548,
      "grad_norm": 1.4244133234024048,
      "learning_rate": 9.292684210526316e-05,
      "loss": 0.4029,
      "step": 46897
    },
    {
      "epoch": 2.626161944226677,
      "grad_norm": 1.1992512941360474,
      "learning_rate": 9.292657894736842e-05,
      "loss": 0.3859,
      "step": 46898
    },
    {
      "epoch": 2.626217941538806,
      "grad_norm": 1.3435248136520386,
      "learning_rate": 9.292631578947368e-05,
      "loss": 0.4226,
      "step": 46899
    },
    {
      "epoch": 2.626273938850935,
      "grad_norm": 1.3188532590866089,
      "learning_rate": 9.292605263157896e-05,
      "loss": 0.3834,
      "step": 46900
    },
    {
      "epoch": 2.626329936163064,
      "grad_norm": 1.232077956199646,
      "learning_rate": 9.292578947368422e-05,
      "loss": 0.4165,
      "step": 46901
    },
    {
      "epoch": 2.626385933475193,
      "grad_norm": 1.9849222898483276,
      "learning_rate": 9.292552631578948e-05,
      "loss": 0.4392,
      "step": 46902
    },
    {
      "epoch": 2.626441930787322,
      "grad_norm": 1.1978445053100586,
      "learning_rate": 9.292526315789474e-05,
      "loss": 0.4123,
      "step": 46903
    },
    {
      "epoch": 2.626497928099451,
      "grad_norm": 1.1746307611465454,
      "learning_rate": 9.292500000000001e-05,
      "loss": 0.42,
      "step": 46904
    },
    {
      "epoch": 2.62655392541158,
      "grad_norm": 0.9686213731765747,
      "learning_rate": 9.292473684210527e-05,
      "loss": 0.2971,
      "step": 46905
    },
    {
      "epoch": 2.626609922723709,
      "grad_norm": 1.1451046466827393,
      "learning_rate": 9.292447368421053e-05,
      "loss": 0.4534,
      "step": 46906
    },
    {
      "epoch": 2.626665920035838,
      "grad_norm": 1.5036507844924927,
      "learning_rate": 9.292421052631579e-05,
      "loss": 0.5551,
      "step": 46907
    },
    {
      "epoch": 2.6267219173479672,
      "grad_norm": 0.944596529006958,
      "learning_rate": 9.292394736842106e-05,
      "loss": 0.258,
      "step": 46908
    },
    {
      "epoch": 2.6267779146600962,
      "grad_norm": 1.2620248794555664,
      "learning_rate": 9.292368421052632e-05,
      "loss": 0.3711,
      "step": 46909
    },
    {
      "epoch": 2.6268339119722253,
      "grad_norm": 1.1045030355453491,
      "learning_rate": 9.292342105263158e-05,
      "loss": 0.395,
      "step": 46910
    },
    {
      "epoch": 2.6268899092843543,
      "grad_norm": 1.62193763256073,
      "learning_rate": 9.292315789473684e-05,
      "loss": 0.394,
      "step": 46911
    },
    {
      "epoch": 2.6269459065964833,
      "grad_norm": 1.3000679016113281,
      "learning_rate": 9.292289473684211e-05,
      "loss": 0.4478,
      "step": 46912
    },
    {
      "epoch": 2.6270019039086123,
      "grad_norm": 0.984430730342865,
      "learning_rate": 9.292263157894737e-05,
      "loss": 0.2648,
      "step": 46913
    },
    {
      "epoch": 2.6270579012207413,
      "grad_norm": 1.4571577310562134,
      "learning_rate": 9.292236842105265e-05,
      "loss": 0.5023,
      "step": 46914
    },
    {
      "epoch": 2.6271138985328704,
      "grad_norm": 1.1449741125106812,
      "learning_rate": 9.292210526315789e-05,
      "loss": 0.3326,
      "step": 46915
    },
    {
      "epoch": 2.6271698958449994,
      "grad_norm": 1.0675387382507324,
      "learning_rate": 9.292184210526315e-05,
      "loss": 0.4177,
      "step": 46916
    },
    {
      "epoch": 2.6272258931571284,
      "grad_norm": 1.390217661857605,
      "learning_rate": 9.292157894736843e-05,
      "loss": 0.377,
      "step": 46917
    },
    {
      "epoch": 2.6272818904692574,
      "grad_norm": 1.4298555850982666,
      "learning_rate": 9.292131578947369e-05,
      "loss": 0.3869,
      "step": 46918
    },
    {
      "epoch": 2.6273378877813864,
      "grad_norm": 1.584778904914856,
      "learning_rate": 9.292105263157896e-05,
      "loss": 0.4475,
      "step": 46919
    },
    {
      "epoch": 2.6273938850935155,
      "grad_norm": 1.3577972650527954,
      "learning_rate": 9.29207894736842e-05,
      "loss": 0.5336,
      "step": 46920
    },
    {
      "epoch": 2.6274498824056445,
      "grad_norm": 1.4191336631774902,
      "learning_rate": 9.292052631578948e-05,
      "loss": 0.3963,
      "step": 46921
    },
    {
      "epoch": 2.6275058797177735,
      "grad_norm": 1.4518589973449707,
      "learning_rate": 9.292026315789474e-05,
      "loss": 0.3855,
      "step": 46922
    },
    {
      "epoch": 2.6275618770299025,
      "grad_norm": 9.198808670043945,
      "learning_rate": 9.292000000000001e-05,
      "loss": 0.4031,
      "step": 46923
    },
    {
      "epoch": 2.6276178743420315,
      "grad_norm": 1.1096291542053223,
      "learning_rate": 9.291973684210527e-05,
      "loss": 0.4039,
      "step": 46924
    },
    {
      "epoch": 2.6276738716541606,
      "grad_norm": 1.6860275268554688,
      "learning_rate": 9.291947368421053e-05,
      "loss": 0.4522,
      "step": 46925
    },
    {
      "epoch": 2.6277298689662896,
      "grad_norm": 1.2366893291473389,
      "learning_rate": 9.291921052631579e-05,
      "loss": 0.5135,
      "step": 46926
    },
    {
      "epoch": 2.6277858662784186,
      "grad_norm": 1.3470619916915894,
      "learning_rate": 9.291894736842106e-05,
      "loss": 0.4922,
      "step": 46927
    },
    {
      "epoch": 2.6278418635905476,
      "grad_norm": 1.3473389148712158,
      "learning_rate": 9.291868421052632e-05,
      "loss": 0.3981,
      "step": 46928
    },
    {
      "epoch": 2.6278978609026766,
      "grad_norm": 1.1883580684661865,
      "learning_rate": 9.291842105263158e-05,
      "loss": 0.3534,
      "step": 46929
    },
    {
      "epoch": 2.6279538582148056,
      "grad_norm": 1.0331083536148071,
      "learning_rate": 9.291815789473684e-05,
      "loss": 0.3202,
      "step": 46930
    },
    {
      "epoch": 2.6280098555269347,
      "grad_norm": 0.9996214509010315,
      "learning_rate": 9.291789473684212e-05,
      "loss": 0.3337,
      "step": 46931
    },
    {
      "epoch": 2.6280658528390637,
      "grad_norm": 1.5348716974258423,
      "learning_rate": 9.291763157894738e-05,
      "loss": 0.3376,
      "step": 46932
    },
    {
      "epoch": 2.6281218501511927,
      "grad_norm": 1.3197225332260132,
      "learning_rate": 9.291736842105264e-05,
      "loss": 0.4879,
      "step": 46933
    },
    {
      "epoch": 2.6281778474633217,
      "grad_norm": 1.050466775894165,
      "learning_rate": 9.29171052631579e-05,
      "loss": 0.3923,
      "step": 46934
    },
    {
      "epoch": 2.6282338447754507,
      "grad_norm": 1.2297800779342651,
      "learning_rate": 9.291684210526316e-05,
      "loss": 0.3992,
      "step": 46935
    },
    {
      "epoch": 2.6282898420875798,
      "grad_norm": 1.0194079875946045,
      "learning_rate": 9.291657894736843e-05,
      "loss": 0.3898,
      "step": 46936
    },
    {
      "epoch": 2.628345839399709,
      "grad_norm": 0.9917104244232178,
      "learning_rate": 9.291631578947369e-05,
      "loss": 0.3579,
      "step": 46937
    },
    {
      "epoch": 2.628401836711838,
      "grad_norm": 0.9514667987823486,
      "learning_rate": 9.291605263157895e-05,
      "loss": 0.3832,
      "step": 46938
    },
    {
      "epoch": 2.628457834023967,
      "grad_norm": 1.1570740938186646,
      "learning_rate": 9.291578947368421e-05,
      "loss": 0.46,
      "step": 46939
    },
    {
      "epoch": 2.628513831336096,
      "grad_norm": 1.1597615480422974,
      "learning_rate": 9.291552631578948e-05,
      "loss": 0.4247,
      "step": 46940
    },
    {
      "epoch": 2.628569828648225,
      "grad_norm": 1.1936694383621216,
      "learning_rate": 9.291526315789474e-05,
      "loss": 0.4533,
      "step": 46941
    },
    {
      "epoch": 2.628625825960354,
      "grad_norm": 1.2868821620941162,
      "learning_rate": 9.291500000000001e-05,
      "loss": 0.359,
      "step": 46942
    },
    {
      "epoch": 2.628681823272483,
      "grad_norm": 1.1153076887130737,
      "learning_rate": 9.291473684210526e-05,
      "loss": 0.3036,
      "step": 46943
    },
    {
      "epoch": 2.628737820584612,
      "grad_norm": 1.0310460329055786,
      "learning_rate": 9.291447368421053e-05,
      "loss": 0.4035,
      "step": 46944
    },
    {
      "epoch": 2.628793817896741,
      "grad_norm": 1.1082276105880737,
      "learning_rate": 9.29142105263158e-05,
      "loss": 0.5377,
      "step": 46945
    },
    {
      "epoch": 2.62884981520887,
      "grad_norm": 1.0089575052261353,
      "learning_rate": 9.291394736842107e-05,
      "loss": 0.306,
      "step": 46946
    },
    {
      "epoch": 2.628905812520999,
      "grad_norm": 1.1472532749176025,
      "learning_rate": 9.291368421052631e-05,
      "loss": 0.3529,
      "step": 46947
    },
    {
      "epoch": 2.628961809833128,
      "grad_norm": 1.1784005165100098,
      "learning_rate": 9.291342105263159e-05,
      "loss": 0.4072,
      "step": 46948
    },
    {
      "epoch": 2.629017807145257,
      "grad_norm": 1.0315799713134766,
      "learning_rate": 9.291315789473685e-05,
      "loss": 0.4009,
      "step": 46949
    },
    {
      "epoch": 2.629073804457386,
      "grad_norm": 1.257932186126709,
      "learning_rate": 9.29128947368421e-05,
      "loss": 0.4242,
      "step": 46950
    },
    {
      "epoch": 2.629129801769515,
      "grad_norm": 1.4812002182006836,
      "learning_rate": 9.291263157894738e-05,
      "loss": 0.4213,
      "step": 46951
    },
    {
      "epoch": 2.629185799081644,
      "grad_norm": 1.1574113368988037,
      "learning_rate": 9.291236842105263e-05,
      "loss": 0.4464,
      "step": 46952
    },
    {
      "epoch": 2.629241796393773,
      "grad_norm": 1.3412389755249023,
      "learning_rate": 9.29121052631579e-05,
      "loss": 0.4203,
      "step": 46953
    },
    {
      "epoch": 2.629297793705902,
      "grad_norm": 1.7697086334228516,
      "learning_rate": 9.291184210526316e-05,
      "loss": 0.3715,
      "step": 46954
    },
    {
      "epoch": 2.629353791018031,
      "grad_norm": 1.2484581470489502,
      "learning_rate": 9.291157894736843e-05,
      "loss": 0.4023,
      "step": 46955
    },
    {
      "epoch": 2.62940978833016,
      "grad_norm": 1.1666266918182373,
      "learning_rate": 9.291131578947369e-05,
      "loss": 0.408,
      "step": 46956
    },
    {
      "epoch": 2.629465785642289,
      "grad_norm": 1.3479361534118652,
      "learning_rate": 9.291105263157895e-05,
      "loss": 0.4975,
      "step": 46957
    },
    {
      "epoch": 2.629521782954418,
      "grad_norm": 1.1643496751785278,
      "learning_rate": 9.291078947368421e-05,
      "loss": 0.4184,
      "step": 46958
    },
    {
      "epoch": 2.629577780266547,
      "grad_norm": 1.085070252418518,
      "learning_rate": 9.291052631578948e-05,
      "loss": 0.3462,
      "step": 46959
    },
    {
      "epoch": 2.6296337775786762,
      "grad_norm": 1.0632199048995972,
      "learning_rate": 9.291026315789474e-05,
      "loss": 0.4678,
      "step": 46960
    },
    {
      "epoch": 2.6296897748908052,
      "grad_norm": 1.7379429340362549,
      "learning_rate": 9.291e-05,
      "loss": 0.5269,
      "step": 46961
    },
    {
      "epoch": 2.6297457722029343,
      "grad_norm": 1.1484042406082153,
      "learning_rate": 9.290973684210526e-05,
      "loss": 0.3737,
      "step": 46962
    },
    {
      "epoch": 2.6298017695150633,
      "grad_norm": 1.0907198190689087,
      "learning_rate": 9.290947368421054e-05,
      "loss": 0.3051,
      "step": 46963
    },
    {
      "epoch": 2.6298577668271923,
      "grad_norm": 1.4166656732559204,
      "learning_rate": 9.29092105263158e-05,
      "loss": 0.4188,
      "step": 46964
    },
    {
      "epoch": 2.6299137641393213,
      "grad_norm": 1.0852526426315308,
      "learning_rate": 9.290894736842106e-05,
      "loss": 0.4816,
      "step": 46965
    },
    {
      "epoch": 2.6299697614514503,
      "grad_norm": 1.251293420791626,
      "learning_rate": 9.290868421052632e-05,
      "loss": 0.5106,
      "step": 46966
    },
    {
      "epoch": 2.6300257587635794,
      "grad_norm": 1.3656463623046875,
      "learning_rate": 9.290842105263158e-05,
      "loss": 0.5036,
      "step": 46967
    },
    {
      "epoch": 2.6300817560757084,
      "grad_norm": 1.1081115007400513,
      "learning_rate": 9.290815789473685e-05,
      "loss": 0.433,
      "step": 46968
    },
    {
      "epoch": 2.6301377533878374,
      "grad_norm": 1.1234749555587769,
      "learning_rate": 9.290789473684211e-05,
      "loss": 0.3871,
      "step": 46969
    },
    {
      "epoch": 2.6301937506999664,
      "grad_norm": 1.28562331199646,
      "learning_rate": 9.290763157894737e-05,
      "loss": 0.5969,
      "step": 46970
    },
    {
      "epoch": 2.6302497480120954,
      "grad_norm": 1.0516748428344727,
      "learning_rate": 9.290736842105263e-05,
      "loss": 0.3274,
      "step": 46971
    },
    {
      "epoch": 2.6303057453242245,
      "grad_norm": 2.0585405826568604,
      "learning_rate": 9.29071052631579e-05,
      "loss": 0.4967,
      "step": 46972
    },
    {
      "epoch": 2.6303617426363535,
      "grad_norm": 1.1156389713287354,
      "learning_rate": 9.290684210526316e-05,
      "loss": 0.3613,
      "step": 46973
    },
    {
      "epoch": 2.6304177399484825,
      "grad_norm": 1.154118299484253,
      "learning_rate": 9.290657894736843e-05,
      "loss": 0.3427,
      "step": 46974
    },
    {
      "epoch": 2.6304737372606115,
      "grad_norm": 1.2301157712936401,
      "learning_rate": 9.290631578947368e-05,
      "loss": 0.4436,
      "step": 46975
    },
    {
      "epoch": 2.6305297345727405,
      "grad_norm": 5.588713645935059,
      "learning_rate": 9.290605263157895e-05,
      "loss": 0.3457,
      "step": 46976
    },
    {
      "epoch": 2.6305857318848695,
      "grad_norm": 1.3277336359024048,
      "learning_rate": 9.290578947368421e-05,
      "loss": 0.4595,
      "step": 46977
    },
    {
      "epoch": 2.6306417291969986,
      "grad_norm": 1.2962141036987305,
      "learning_rate": 9.290552631578949e-05,
      "loss": 0.4179,
      "step": 46978
    },
    {
      "epoch": 2.6306977265091276,
      "grad_norm": 1.1249175071716309,
      "learning_rate": 9.290526315789475e-05,
      "loss": 0.5147,
      "step": 46979
    },
    {
      "epoch": 2.6307537238212566,
      "grad_norm": 1.299116611480713,
      "learning_rate": 9.2905e-05,
      "loss": 0.4371,
      "step": 46980
    },
    {
      "epoch": 2.6308097211333856,
      "grad_norm": 1.0215312242507935,
      "learning_rate": 9.290473684210527e-05,
      "loss": 0.3774,
      "step": 46981
    },
    {
      "epoch": 2.6308657184455146,
      "grad_norm": 1.327256679534912,
      "learning_rate": 9.290447368421054e-05,
      "loss": 0.3434,
      "step": 46982
    },
    {
      "epoch": 2.6309217157576437,
      "grad_norm": 1.212263584136963,
      "learning_rate": 9.29042105263158e-05,
      "loss": 0.4863,
      "step": 46983
    },
    {
      "epoch": 2.6309777130697727,
      "grad_norm": 1.2237985134124756,
      "learning_rate": 9.290394736842106e-05,
      "loss": 0.3657,
      "step": 46984
    },
    {
      "epoch": 2.6310337103819017,
      "grad_norm": 1.1181721687316895,
      "learning_rate": 9.290368421052632e-05,
      "loss": 0.4082,
      "step": 46985
    },
    {
      "epoch": 2.6310897076940307,
      "grad_norm": 1.048292636871338,
      "learning_rate": 9.290342105263158e-05,
      "loss": 0.3604,
      "step": 46986
    },
    {
      "epoch": 2.6311457050061597,
      "grad_norm": 1.0829049348831177,
      "learning_rate": 9.290315789473685e-05,
      "loss": 0.4133,
      "step": 46987
    },
    {
      "epoch": 2.6312017023182888,
      "grad_norm": 1.189784049987793,
      "learning_rate": 9.290289473684211e-05,
      "loss": 0.3572,
      "step": 46988
    },
    {
      "epoch": 2.631257699630418,
      "grad_norm": 1.4107248783111572,
      "learning_rate": 9.290263157894737e-05,
      "loss": 0.4099,
      "step": 46989
    },
    {
      "epoch": 2.631313696942547,
      "grad_norm": 1.120826244354248,
      "learning_rate": 9.290236842105263e-05,
      "loss": 0.4595,
      "step": 46990
    },
    {
      "epoch": 2.631369694254676,
      "grad_norm": 1.0824512243270874,
      "learning_rate": 9.29021052631579e-05,
      "loss": 0.3438,
      "step": 46991
    },
    {
      "epoch": 2.631425691566805,
      "grad_norm": 1.2773828506469727,
      "learning_rate": 9.290184210526316e-05,
      "loss": 0.4036,
      "step": 46992
    },
    {
      "epoch": 2.631481688878934,
      "grad_norm": 1.039052128791809,
      "learning_rate": 9.290157894736842e-05,
      "loss": 0.4689,
      "step": 46993
    },
    {
      "epoch": 2.631537686191063,
      "grad_norm": 1.19552743434906,
      "learning_rate": 9.290131578947368e-05,
      "loss": 0.3705,
      "step": 46994
    },
    {
      "epoch": 2.631593683503192,
      "grad_norm": 1.3612709045410156,
      "learning_rate": 9.290105263157896e-05,
      "loss": 0.4747,
      "step": 46995
    },
    {
      "epoch": 2.631649680815321,
      "grad_norm": 1.4303760528564453,
      "learning_rate": 9.290078947368422e-05,
      "loss": 0.4465,
      "step": 46996
    },
    {
      "epoch": 2.63170567812745,
      "grad_norm": 1.2032630443572998,
      "learning_rate": 9.290052631578949e-05,
      "loss": 0.4565,
      "step": 46997
    },
    {
      "epoch": 2.631761675439579,
      "grad_norm": 1.4642654657363892,
      "learning_rate": 9.290026315789474e-05,
      "loss": 0.4018,
      "step": 46998
    },
    {
      "epoch": 2.631817672751708,
      "grad_norm": 1.0528929233551025,
      "learning_rate": 9.290000000000001e-05,
      "loss": 0.3331,
      "step": 46999
    },
    {
      "epoch": 2.631873670063837,
      "grad_norm": 1.539127230644226,
      "learning_rate": 9.289973684210527e-05,
      "loss": 0.4537,
      "step": 47000
    },
    {
      "epoch": 2.631929667375966,
      "grad_norm": 1.1344317197799683,
      "learning_rate": 9.289947368421054e-05,
      "loss": 0.3488,
      "step": 47001
    },
    {
      "epoch": 2.631985664688095,
      "grad_norm": 1.1277116537094116,
      "learning_rate": 9.289921052631579e-05,
      "loss": 0.3822,
      "step": 47002
    },
    {
      "epoch": 2.632041662000224,
      "grad_norm": 0.9893087148666382,
      "learning_rate": 9.289894736842105e-05,
      "loss": 0.3022,
      "step": 47003
    },
    {
      "epoch": 2.632097659312353,
      "grad_norm": 1.4053288698196411,
      "learning_rate": 9.289868421052632e-05,
      "loss": 0.4629,
      "step": 47004
    },
    {
      "epoch": 2.632153656624482,
      "grad_norm": 1.2331140041351318,
      "learning_rate": 9.289842105263158e-05,
      "loss": 0.4773,
      "step": 47005
    },
    {
      "epoch": 2.632209653936611,
      "grad_norm": 1.2761956453323364,
      "learning_rate": 9.289815789473685e-05,
      "loss": 0.4183,
      "step": 47006
    },
    {
      "epoch": 2.63226565124874,
      "grad_norm": 1.1187763214111328,
      "learning_rate": 9.28978947368421e-05,
      "loss": 0.4508,
      "step": 47007
    },
    {
      "epoch": 2.632321648560869,
      "grad_norm": 0.9472700953483582,
      "learning_rate": 9.289763157894737e-05,
      "loss": 0.3241,
      "step": 47008
    },
    {
      "epoch": 2.632377645872998,
      "grad_norm": 1.1563911437988281,
      "learning_rate": 9.289736842105263e-05,
      "loss": 0.4682,
      "step": 47009
    },
    {
      "epoch": 2.632433643185127,
      "grad_norm": 1.0195175409317017,
      "learning_rate": 9.28971052631579e-05,
      "loss": 0.3863,
      "step": 47010
    },
    {
      "epoch": 2.632489640497256,
      "grad_norm": 1.56032395362854,
      "learning_rate": 9.289684210526317e-05,
      "loss": 0.426,
      "step": 47011
    },
    {
      "epoch": 2.632545637809385,
      "grad_norm": 1.059169054031372,
      "learning_rate": 9.289657894736843e-05,
      "loss": 0.3269,
      "step": 47012
    },
    {
      "epoch": 2.6326016351215142,
      "grad_norm": 1.08601713180542,
      "learning_rate": 9.289631578947369e-05,
      "loss": 0.4132,
      "step": 47013
    },
    {
      "epoch": 2.6326576324336433,
      "grad_norm": 1.2722054719924927,
      "learning_rate": 9.289605263157896e-05,
      "loss": 0.337,
      "step": 47014
    },
    {
      "epoch": 2.6327136297457723,
      "grad_norm": 0.9643957018852234,
      "learning_rate": 9.289578947368422e-05,
      "loss": 0.3015,
      "step": 47015
    },
    {
      "epoch": 2.6327696270579013,
      "grad_norm": 1.1986796855926514,
      "learning_rate": 9.289552631578948e-05,
      "loss": 0.4105,
      "step": 47016
    },
    {
      "epoch": 2.6328256243700303,
      "grad_norm": 1.449905276298523,
      "learning_rate": 9.289526315789474e-05,
      "loss": 0.4174,
      "step": 47017
    },
    {
      "epoch": 2.6328816216821593,
      "grad_norm": 1.181523323059082,
      "learning_rate": 9.289500000000001e-05,
      "loss": 0.326,
      "step": 47018
    },
    {
      "epoch": 2.6329376189942884,
      "grad_norm": 1.2053459882736206,
      "learning_rate": 9.289473684210527e-05,
      "loss": 0.4061,
      "step": 47019
    },
    {
      "epoch": 2.6329936163064174,
      "grad_norm": 1.5102391242980957,
      "learning_rate": 9.289447368421053e-05,
      "loss": 0.6364,
      "step": 47020
    },
    {
      "epoch": 2.6330496136185464,
      "grad_norm": 1.1346380710601807,
      "learning_rate": 9.289421052631579e-05,
      "loss": 0.3587,
      "step": 47021
    },
    {
      "epoch": 2.6331056109306754,
      "grad_norm": 1.4418588876724243,
      "learning_rate": 9.289394736842105e-05,
      "loss": 0.3857,
      "step": 47022
    },
    {
      "epoch": 2.6331616082428044,
      "grad_norm": 0.9776993989944458,
      "learning_rate": 9.289368421052632e-05,
      "loss": 0.3339,
      "step": 47023
    },
    {
      "epoch": 2.6332176055549334,
      "grad_norm": 1.1112648248672485,
      "learning_rate": 9.289342105263158e-05,
      "loss": 0.3305,
      "step": 47024
    },
    {
      "epoch": 2.6332736028670625,
      "grad_norm": 1.172960877418518,
      "learning_rate": 9.289315789473684e-05,
      "loss": 0.4012,
      "step": 47025
    },
    {
      "epoch": 2.6333296001791915,
      "grad_norm": 1.1801588535308838,
      "learning_rate": 9.28928947368421e-05,
      "loss": 0.3398,
      "step": 47026
    },
    {
      "epoch": 2.6333855974913205,
      "grad_norm": 1.0496631860733032,
      "learning_rate": 9.289263157894738e-05,
      "loss": 0.3505,
      "step": 47027
    },
    {
      "epoch": 2.6334415948034495,
      "grad_norm": 1.0360312461853027,
      "learning_rate": 9.289236842105264e-05,
      "loss": 0.382,
      "step": 47028
    },
    {
      "epoch": 2.6334975921155785,
      "grad_norm": 1.087490200996399,
      "learning_rate": 9.289210526315791e-05,
      "loss": 0.4375,
      "step": 47029
    },
    {
      "epoch": 2.6335535894277076,
      "grad_norm": 1.371126413345337,
      "learning_rate": 9.289184210526315e-05,
      "loss": 0.4592,
      "step": 47030
    },
    {
      "epoch": 2.6336095867398366,
      "grad_norm": 1.1268630027770996,
      "learning_rate": 9.289157894736843e-05,
      "loss": 0.4559,
      "step": 47031
    },
    {
      "epoch": 2.6336655840519656,
      "grad_norm": 1.2383745908737183,
      "learning_rate": 9.289131578947369e-05,
      "loss": 0.4004,
      "step": 47032
    },
    {
      "epoch": 2.6337215813640946,
      "grad_norm": 1.1104247570037842,
      "learning_rate": 9.289105263157896e-05,
      "loss": 0.4466,
      "step": 47033
    },
    {
      "epoch": 2.6337775786762236,
      "grad_norm": 1.1377804279327393,
      "learning_rate": 9.289078947368422e-05,
      "loss": 0.4069,
      "step": 47034
    },
    {
      "epoch": 2.6338335759883527,
      "grad_norm": 1.3107728958129883,
      "learning_rate": 9.289052631578948e-05,
      "loss": 0.6453,
      "step": 47035
    },
    {
      "epoch": 2.6338895733004817,
      "grad_norm": 1.147043228149414,
      "learning_rate": 9.289026315789474e-05,
      "loss": 0.4879,
      "step": 47036
    },
    {
      "epoch": 2.6339455706126107,
      "grad_norm": 1.217816948890686,
      "learning_rate": 9.289e-05,
      "loss": 0.3928,
      "step": 47037
    },
    {
      "epoch": 2.6340015679247397,
      "grad_norm": 1.022105097770691,
      "learning_rate": 9.288973684210527e-05,
      "loss": 0.4093,
      "step": 47038
    },
    {
      "epoch": 2.6340575652368687,
      "grad_norm": 1.2954506874084473,
      "learning_rate": 9.288947368421052e-05,
      "loss": 0.4198,
      "step": 47039
    },
    {
      "epoch": 2.6341135625489978,
      "grad_norm": 1.022426724433899,
      "learning_rate": 9.288921052631579e-05,
      "loss": 0.3374,
      "step": 47040
    },
    {
      "epoch": 2.6341695598611268,
      "grad_norm": 1.114538311958313,
      "learning_rate": 9.288894736842105e-05,
      "loss": 0.3343,
      "step": 47041
    },
    {
      "epoch": 2.634225557173256,
      "grad_norm": 1.1801730394363403,
      "learning_rate": 9.288868421052633e-05,
      "loss": 0.3972,
      "step": 47042
    },
    {
      "epoch": 2.634281554485385,
      "grad_norm": 1.5636847019195557,
      "learning_rate": 9.288842105263159e-05,
      "loss": 0.4531,
      "step": 47043
    },
    {
      "epoch": 2.634337551797514,
      "grad_norm": 1.074082851409912,
      "learning_rate": 9.288815789473685e-05,
      "loss": 0.5799,
      "step": 47044
    },
    {
      "epoch": 2.634393549109643,
      "grad_norm": 1.2119306325912476,
      "learning_rate": 9.28878947368421e-05,
      "loss": 0.3589,
      "step": 47045
    },
    {
      "epoch": 2.634449546421772,
      "grad_norm": 1.1573002338409424,
      "learning_rate": 9.288763157894738e-05,
      "loss": 0.3342,
      "step": 47046
    },
    {
      "epoch": 2.634505543733901,
      "grad_norm": 1.228115439414978,
      "learning_rate": 9.288736842105264e-05,
      "loss": 0.3552,
      "step": 47047
    },
    {
      "epoch": 2.63456154104603,
      "grad_norm": 1.1376709938049316,
      "learning_rate": 9.28871052631579e-05,
      "loss": 0.459,
      "step": 47048
    },
    {
      "epoch": 2.634617538358159,
      "grad_norm": 1.2990806102752686,
      "learning_rate": 9.288684210526316e-05,
      "loss": 0.3231,
      "step": 47049
    },
    {
      "epoch": 2.634673535670288,
      "grad_norm": 1.468453049659729,
      "learning_rate": 9.288657894736843e-05,
      "loss": 0.272,
      "step": 47050
    },
    {
      "epoch": 2.634729532982417,
      "grad_norm": 1.0406954288482666,
      "learning_rate": 9.288631578947369e-05,
      "loss": 0.3387,
      "step": 47051
    },
    {
      "epoch": 2.634785530294546,
      "grad_norm": 1.073775053024292,
      "learning_rate": 9.288605263157896e-05,
      "loss": 0.293,
      "step": 47052
    },
    {
      "epoch": 2.634841527606675,
      "grad_norm": 1.3651704788208008,
      "learning_rate": 9.288578947368421e-05,
      "loss": 0.4932,
      "step": 47053
    },
    {
      "epoch": 2.634897524918804,
      "grad_norm": 1.257690668106079,
      "learning_rate": 9.288552631578947e-05,
      "loss": 0.5209,
      "step": 47054
    },
    {
      "epoch": 2.634953522230933,
      "grad_norm": 1.138351321220398,
      "learning_rate": 9.288526315789474e-05,
      "loss": 0.4218,
      "step": 47055
    },
    {
      "epoch": 2.635009519543062,
      "grad_norm": 1.2450518608093262,
      "learning_rate": 9.2885e-05,
      "loss": 0.5495,
      "step": 47056
    },
    {
      "epoch": 2.635065516855191,
      "grad_norm": Infinity,
      "learning_rate": 9.2885e-05,
      "loss": 0.3253,
      "step": 47057
    },
    {
      "epoch": 2.63512151416732,
      "grad_norm": 1.440476655960083,
      "learning_rate": 9.288473684210526e-05,
      "loss": 0.4147,
      "step": 47058
    },
    {
      "epoch": 2.635177511479449,
      "grad_norm": 1.3521078824996948,
      "learning_rate": 9.288447368421052e-05,
      "loss": 0.5,
      "step": 47059
    },
    {
      "epoch": 2.635233508791578,
      "grad_norm": 1.2027658224105835,
      "learning_rate": 9.28842105263158e-05,
      "loss": 0.4045,
      "step": 47060
    },
    {
      "epoch": 2.635289506103707,
      "grad_norm": 1.1683554649353027,
      "learning_rate": 9.288394736842106e-05,
      "loss": 0.422,
      "step": 47061
    },
    {
      "epoch": 2.635345503415836,
      "grad_norm": 1.333648920059204,
      "learning_rate": 9.288368421052633e-05,
      "loss": 0.5873,
      "step": 47062
    },
    {
      "epoch": 2.635401500727965,
      "grad_norm": 1.1779141426086426,
      "learning_rate": 9.288342105263157e-05,
      "loss": 0.3586,
      "step": 47063
    },
    {
      "epoch": 2.635457498040094,
      "grad_norm": 1.0689085721969604,
      "learning_rate": 9.288315789473685e-05,
      "loss": 0.3111,
      "step": 47064
    },
    {
      "epoch": 2.6355134953522232,
      "grad_norm": 1.0622735023498535,
      "learning_rate": 9.288289473684211e-05,
      "loss": 0.3866,
      "step": 47065
    },
    {
      "epoch": 2.6355694926643523,
      "grad_norm": 1.306572437286377,
      "learning_rate": 9.288263157894738e-05,
      "loss": 0.4758,
      "step": 47066
    },
    {
      "epoch": 2.6356254899764813,
      "grad_norm": 1.3974851369857788,
      "learning_rate": 9.288236842105264e-05,
      "loss": 0.4602,
      "step": 47067
    },
    {
      "epoch": 2.6356814872886103,
      "grad_norm": 1.0201671123504639,
      "learning_rate": 9.28821052631579e-05,
      "loss": 0.3692,
      "step": 47068
    },
    {
      "epoch": 2.6357374846007393,
      "grad_norm": 1.1542476415634155,
      "learning_rate": 9.288184210526316e-05,
      "loss": 0.3613,
      "step": 47069
    },
    {
      "epoch": 2.6357934819128683,
      "grad_norm": 1.2347222566604614,
      "learning_rate": 9.288157894736843e-05,
      "loss": 0.5051,
      "step": 47070
    },
    {
      "epoch": 2.6358494792249973,
      "grad_norm": 1.1261557340621948,
      "learning_rate": 9.288131578947369e-05,
      "loss": 0.3834,
      "step": 47071
    },
    {
      "epoch": 2.6359054765371264,
      "grad_norm": 1.3250292539596558,
      "learning_rate": 9.288105263157895e-05,
      "loss": 0.4385,
      "step": 47072
    },
    {
      "epoch": 2.6359614738492554,
      "grad_norm": 1.7122801542282104,
      "learning_rate": 9.288078947368421e-05,
      "loss": 0.4768,
      "step": 47073
    },
    {
      "epoch": 2.6360174711613844,
      "grad_norm": 1.144468069076538,
      "learning_rate": 9.288052631578947e-05,
      "loss": 0.3499,
      "step": 47074
    },
    {
      "epoch": 2.6360734684735134,
      "grad_norm": 1.2208809852600098,
      "learning_rate": 9.288026315789475e-05,
      "loss": 0.3005,
      "step": 47075
    },
    {
      "epoch": 2.6361294657856424,
      "grad_norm": 1.7636222839355469,
      "learning_rate": 9.288e-05,
      "loss": 0.435,
      "step": 47076
    },
    {
      "epoch": 2.6361854630977715,
      "grad_norm": 1.6449847221374512,
      "learning_rate": 9.287973684210526e-05,
      "loss": 0.2522,
      "step": 47077
    },
    {
      "epoch": 2.6362414604099005,
      "grad_norm": 1.0474190711975098,
      "learning_rate": 9.287947368421052e-05,
      "loss": 0.3655,
      "step": 47078
    },
    {
      "epoch": 2.6362974577220295,
      "grad_norm": 1.1536836624145508,
      "learning_rate": 9.28792105263158e-05,
      "loss": 0.4376,
      "step": 47079
    },
    {
      "epoch": 2.6363534550341585,
      "grad_norm": 1.1514642238616943,
      "learning_rate": 9.287894736842106e-05,
      "loss": 0.3202,
      "step": 47080
    },
    {
      "epoch": 2.6364094523462875,
      "grad_norm": 1.098402500152588,
      "learning_rate": 9.287868421052632e-05,
      "loss": 0.3741,
      "step": 47081
    },
    {
      "epoch": 2.6364654496584166,
      "grad_norm": 1.1826136112213135,
      "learning_rate": 9.287842105263158e-05,
      "loss": 0.4155,
      "step": 47082
    },
    {
      "epoch": 2.6365214469705456,
      "grad_norm": 1.149810552597046,
      "learning_rate": 9.287815789473685e-05,
      "loss": 0.4187,
      "step": 47083
    },
    {
      "epoch": 2.6365774442826746,
      "grad_norm": 1.220630168914795,
      "learning_rate": 9.287789473684211e-05,
      "loss": 0.3424,
      "step": 47084
    },
    {
      "epoch": 2.636633441594803,
      "grad_norm": 1.2126140594482422,
      "learning_rate": 9.287763157894738e-05,
      "loss": 0.4373,
      "step": 47085
    },
    {
      "epoch": 2.6366894389069326,
      "grad_norm": 1.2777427434921265,
      "learning_rate": 9.287736842105263e-05,
      "loss": 0.598,
      "step": 47086
    },
    {
      "epoch": 2.636745436219061,
      "grad_norm": 1.172796607017517,
      "learning_rate": 9.28771052631579e-05,
      "loss": 0.4064,
      "step": 47087
    },
    {
      "epoch": 2.6368014335311907,
      "grad_norm": 1.424055576324463,
      "learning_rate": 9.287684210526316e-05,
      "loss": 0.4675,
      "step": 47088
    },
    {
      "epoch": 2.6368574308433192,
      "grad_norm": 1.1924517154693604,
      "learning_rate": 9.287657894736842e-05,
      "loss": 0.3833,
      "step": 47089
    },
    {
      "epoch": 2.6369134281554487,
      "grad_norm": 1.1879324913024902,
      "learning_rate": 9.28763157894737e-05,
      "loss": 0.34,
      "step": 47090
    },
    {
      "epoch": 2.6369694254675773,
      "grad_norm": 1.2103761434555054,
      "learning_rate": 9.287605263157894e-05,
      "loss": 0.4749,
      "step": 47091
    },
    {
      "epoch": 2.6370254227797068,
      "grad_norm": 0.9686588048934937,
      "learning_rate": 9.287578947368422e-05,
      "loss": 0.3235,
      "step": 47092
    },
    {
      "epoch": 2.6370814200918353,
      "grad_norm": 1.100850224494934,
      "learning_rate": 9.287552631578947e-05,
      "loss": 0.3143,
      "step": 47093
    },
    {
      "epoch": 2.637137417403965,
      "grad_norm": 1.1561366319656372,
      "learning_rate": 9.287526315789475e-05,
      "loss": 0.3784,
      "step": 47094
    },
    {
      "epoch": 2.6371934147160934,
      "grad_norm": 1.2292882204055786,
      "learning_rate": 9.2875e-05,
      "loss": 0.3818,
      "step": 47095
    },
    {
      "epoch": 2.637249412028223,
      "grad_norm": 1.344846248626709,
      "learning_rate": 9.287473684210527e-05,
      "loss": 0.3279,
      "step": 47096
    },
    {
      "epoch": 2.6373054093403514,
      "grad_norm": 1.0718538761138916,
      "learning_rate": 9.287447368421053e-05,
      "loss": 0.3904,
      "step": 47097
    },
    {
      "epoch": 2.637361406652481,
      "grad_norm": 1.2847731113433838,
      "learning_rate": 9.28742105263158e-05,
      "loss": 0.4544,
      "step": 47098
    },
    {
      "epoch": 2.6374174039646094,
      "grad_norm": 1.1354014873504639,
      "learning_rate": 9.287394736842106e-05,
      "loss": 0.4271,
      "step": 47099
    },
    {
      "epoch": 2.637473401276739,
      "grad_norm": 1.6589752435684204,
      "learning_rate": 9.287368421052632e-05,
      "loss": 0.4755,
      "step": 47100
    },
    {
      "epoch": 2.6375293985888675,
      "grad_norm": 1.148756742477417,
      "learning_rate": 9.287342105263158e-05,
      "loss": 0.4227,
      "step": 47101
    },
    {
      "epoch": 2.637585395900997,
      "grad_norm": 1.616754412651062,
      "learning_rate": 9.287315789473685e-05,
      "loss": 0.3574,
      "step": 47102
    },
    {
      "epoch": 2.6376413932131255,
      "grad_norm": 1.1399976015090942,
      "learning_rate": 9.287289473684211e-05,
      "loss": 0.4349,
      "step": 47103
    },
    {
      "epoch": 2.637697390525255,
      "grad_norm": 1.0837591886520386,
      "learning_rate": 9.287263157894737e-05,
      "loss": 0.4025,
      "step": 47104
    },
    {
      "epoch": 2.6377533878373836,
      "grad_norm": 1.160154104232788,
      "learning_rate": 9.287236842105263e-05,
      "loss": 0.3851,
      "step": 47105
    },
    {
      "epoch": 2.637809385149513,
      "grad_norm": 1.4484425783157349,
      "learning_rate": 9.28721052631579e-05,
      "loss": 0.5181,
      "step": 47106
    },
    {
      "epoch": 2.6378653824616416,
      "grad_norm": 1.1300444602966309,
      "learning_rate": 9.287184210526317e-05,
      "loss": 0.3668,
      "step": 47107
    },
    {
      "epoch": 2.637921379773771,
      "grad_norm": 1.387725591659546,
      "learning_rate": 9.287157894736842e-05,
      "loss": 0.4759,
      "step": 47108
    },
    {
      "epoch": 2.6379773770858996,
      "grad_norm": 1.018106460571289,
      "learning_rate": 9.287131578947368e-05,
      "loss": 0.3314,
      "step": 47109
    },
    {
      "epoch": 2.638033374398029,
      "grad_norm": 1.1239522695541382,
      "learning_rate": 9.287105263157894e-05,
      "loss": 0.4261,
      "step": 47110
    },
    {
      "epoch": 2.6380893717101577,
      "grad_norm": 1.288457989692688,
      "learning_rate": 9.287078947368422e-05,
      "loss": 0.3915,
      "step": 47111
    },
    {
      "epoch": 2.638145369022287,
      "grad_norm": 1.1860618591308594,
      "learning_rate": 9.287052631578948e-05,
      "loss": 0.3751,
      "step": 47112
    },
    {
      "epoch": 2.6382013663344157,
      "grad_norm": 1.0909477472305298,
      "learning_rate": 9.287026315789474e-05,
      "loss": 0.395,
      "step": 47113
    },
    {
      "epoch": 2.638257363646545,
      "grad_norm": 1.2550222873687744,
      "learning_rate": 9.287e-05,
      "loss": 0.3579,
      "step": 47114
    },
    {
      "epoch": 2.6383133609586737,
      "grad_norm": 1.2246642112731934,
      "learning_rate": 9.286973684210527e-05,
      "loss": 0.4079,
      "step": 47115
    },
    {
      "epoch": 2.638369358270803,
      "grad_norm": 1.1546550989151,
      "learning_rate": 9.286947368421053e-05,
      "loss": 0.3901,
      "step": 47116
    },
    {
      "epoch": 2.638425355582932,
      "grad_norm": 1.117661476135254,
      "learning_rate": 9.28692105263158e-05,
      "loss": 0.3684,
      "step": 47117
    },
    {
      "epoch": 2.6384813528950612,
      "grad_norm": 1.0887564420700073,
      "learning_rate": 9.286894736842105e-05,
      "loss": 0.3612,
      "step": 47118
    },
    {
      "epoch": 2.63853735020719,
      "grad_norm": 1.2764697074890137,
      "learning_rate": 9.286868421052632e-05,
      "loss": 0.5214,
      "step": 47119
    },
    {
      "epoch": 2.6385933475193193,
      "grad_norm": 1.1876147985458374,
      "learning_rate": 9.286842105263158e-05,
      "loss": 0.3824,
      "step": 47120
    },
    {
      "epoch": 2.638649344831448,
      "grad_norm": 1.1166566610336304,
      "learning_rate": 9.286815789473686e-05,
      "loss": 0.3498,
      "step": 47121
    },
    {
      "epoch": 2.6387053421435773,
      "grad_norm": 1.0656471252441406,
      "learning_rate": 9.286789473684212e-05,
      "loss": 0.3418,
      "step": 47122
    },
    {
      "epoch": 2.638761339455706,
      "grad_norm": 1.378637671470642,
      "learning_rate": 9.286763157894738e-05,
      "loss": 0.3847,
      "step": 47123
    },
    {
      "epoch": 2.6388173367678354,
      "grad_norm": 1.19815194606781,
      "learning_rate": 9.286736842105263e-05,
      "loss": 0.3878,
      "step": 47124
    },
    {
      "epoch": 2.638873334079964,
      "grad_norm": 1.042303442955017,
      "learning_rate": 9.28671052631579e-05,
      "loss": 0.4004,
      "step": 47125
    },
    {
      "epoch": 2.6389293313920934,
      "grad_norm": 1.1364494562149048,
      "learning_rate": 9.286684210526317e-05,
      "loss": 0.4205,
      "step": 47126
    },
    {
      "epoch": 2.638985328704222,
      "grad_norm": 1.3343371152877808,
      "learning_rate": 9.286657894736843e-05,
      "loss": 0.3113,
      "step": 47127
    },
    {
      "epoch": 2.6390413260163514,
      "grad_norm": 1.4137060642242432,
      "learning_rate": 9.286631578947369e-05,
      "loss": 0.378,
      "step": 47128
    },
    {
      "epoch": 2.63909732332848,
      "grad_norm": 1.2229394912719727,
      "learning_rate": 9.286605263157895e-05,
      "loss": 0.4349,
      "step": 47129
    },
    {
      "epoch": 2.6391533206406095,
      "grad_norm": 1.1136841773986816,
      "learning_rate": 9.286578947368422e-05,
      "loss": 0.2989,
      "step": 47130
    },
    {
      "epoch": 2.639209317952738,
      "grad_norm": 1.4584095478057861,
      "learning_rate": 9.286552631578948e-05,
      "loss": 0.5938,
      "step": 47131
    },
    {
      "epoch": 2.6392653152648675,
      "grad_norm": 1.0897856950759888,
      "learning_rate": 9.286526315789474e-05,
      "loss": 0.4156,
      "step": 47132
    },
    {
      "epoch": 2.639321312576996,
      "grad_norm": 1.0507068634033203,
      "learning_rate": 9.2865e-05,
      "loss": 0.3409,
      "step": 47133
    },
    {
      "epoch": 2.6393773098891256,
      "grad_norm": 1.0902528762817383,
      "learning_rate": 9.286473684210527e-05,
      "loss": 0.3685,
      "step": 47134
    },
    {
      "epoch": 2.639433307201254,
      "grad_norm": 1.1191939115524292,
      "learning_rate": 9.286447368421053e-05,
      "loss": 0.3469,
      "step": 47135
    },
    {
      "epoch": 2.6394893045133836,
      "grad_norm": 1.2916542291641235,
      "learning_rate": 9.286421052631579e-05,
      "loss": 0.3909,
      "step": 47136
    },
    {
      "epoch": 2.639545301825512,
      "grad_norm": 1.1472058296203613,
      "learning_rate": 9.286394736842105e-05,
      "loss": 0.3312,
      "step": 47137
    },
    {
      "epoch": 2.6396012991376416,
      "grad_norm": 1.1692471504211426,
      "learning_rate": 9.286368421052633e-05,
      "loss": 0.4288,
      "step": 47138
    },
    {
      "epoch": 2.63965729644977,
      "grad_norm": 1.3263696432113647,
      "learning_rate": 9.286342105263158e-05,
      "loss": 0.3914,
      "step": 47139
    },
    {
      "epoch": 2.6397132937618997,
      "grad_norm": 1.0097308158874512,
      "learning_rate": 9.286315789473686e-05,
      "loss": 0.4063,
      "step": 47140
    },
    {
      "epoch": 2.6397692910740282,
      "grad_norm": 1.333343505859375,
      "learning_rate": 9.28628947368421e-05,
      "loss": 0.4624,
      "step": 47141
    },
    {
      "epoch": 2.6398252883861577,
      "grad_norm": 0.9409430027008057,
      "learning_rate": 9.286263157894736e-05,
      "loss": 0.399,
      "step": 47142
    },
    {
      "epoch": 2.6398812856982863,
      "grad_norm": 1.144340991973877,
      "learning_rate": 9.286236842105264e-05,
      "loss": 0.371,
      "step": 47143
    },
    {
      "epoch": 2.6399372830104157,
      "grad_norm": 1.2190780639648438,
      "learning_rate": 9.28621052631579e-05,
      "loss": 0.3923,
      "step": 47144
    },
    {
      "epoch": 2.6399932803225443,
      "grad_norm": 1.3309259414672852,
      "learning_rate": 9.286184210526317e-05,
      "loss": 0.4158,
      "step": 47145
    },
    {
      "epoch": 2.640049277634674,
      "grad_norm": 1.1447874307632446,
      "learning_rate": 9.286157894736842e-05,
      "loss": 0.4537,
      "step": 47146
    },
    {
      "epoch": 2.6401052749468024,
      "grad_norm": 1.3446931838989258,
      "learning_rate": 9.286131578947369e-05,
      "loss": 0.4532,
      "step": 47147
    },
    {
      "epoch": 2.640161272258932,
      "grad_norm": 1.4104149341583252,
      "learning_rate": 9.286105263157895e-05,
      "loss": 0.4165,
      "step": 47148
    },
    {
      "epoch": 2.6402172695710604,
      "grad_norm": 1.412948727607727,
      "learning_rate": 9.286078947368422e-05,
      "loss": 0.5424,
      "step": 47149
    },
    {
      "epoch": 2.64027326688319,
      "grad_norm": 1.2615450620651245,
      "learning_rate": 9.286052631578947e-05,
      "loss": 0.4352,
      "step": 47150
    },
    {
      "epoch": 2.6403292641953184,
      "grad_norm": 1.1721049547195435,
      "learning_rate": 9.286026315789474e-05,
      "loss": 0.3884,
      "step": 47151
    },
    {
      "epoch": 2.640385261507448,
      "grad_norm": 1.5796034336090088,
      "learning_rate": 9.286e-05,
      "loss": 0.5045,
      "step": 47152
    },
    {
      "epoch": 2.6404412588195765,
      "grad_norm": 1.7254384756088257,
      "learning_rate": 9.285973684210528e-05,
      "loss": 0.4435,
      "step": 47153
    },
    {
      "epoch": 2.640497256131706,
      "grad_norm": 1.2196675539016724,
      "learning_rate": 9.285947368421054e-05,
      "loss": 0.3874,
      "step": 47154
    },
    {
      "epoch": 2.6405532534438345,
      "grad_norm": 1.4416377544403076,
      "learning_rate": 9.28592105263158e-05,
      "loss": 0.4471,
      "step": 47155
    },
    {
      "epoch": 2.640609250755964,
      "grad_norm": 1.1695175170898438,
      "learning_rate": 9.285894736842105e-05,
      "loss": 0.3903,
      "step": 47156
    },
    {
      "epoch": 2.6406652480680926,
      "grad_norm": 1.232383131980896,
      "learning_rate": 9.285868421052633e-05,
      "loss": 0.4447,
      "step": 47157
    },
    {
      "epoch": 2.640721245380222,
      "grad_norm": 1.3042259216308594,
      "learning_rate": 9.285842105263159e-05,
      "loss": 0.4469,
      "step": 47158
    },
    {
      "epoch": 2.6407772426923506,
      "grad_norm": 1.0352526903152466,
      "learning_rate": 9.285815789473685e-05,
      "loss": 0.4154,
      "step": 47159
    },
    {
      "epoch": 2.64083324000448,
      "grad_norm": 1.2063462734222412,
      "learning_rate": 9.285789473684211e-05,
      "loss": 0.3043,
      "step": 47160
    },
    {
      "epoch": 2.6408892373166086,
      "grad_norm": 2.4879558086395264,
      "learning_rate": 9.285763157894737e-05,
      "loss": 0.5878,
      "step": 47161
    },
    {
      "epoch": 2.640945234628738,
      "grad_norm": 1.1312165260314941,
      "learning_rate": 9.285736842105264e-05,
      "loss": 0.3508,
      "step": 47162
    },
    {
      "epoch": 2.6410012319408667,
      "grad_norm": 1.5276029109954834,
      "learning_rate": 9.28571052631579e-05,
      "loss": 0.406,
      "step": 47163
    },
    {
      "epoch": 2.641057229252996,
      "grad_norm": 1.3302150964736938,
      "learning_rate": 9.285684210526316e-05,
      "loss": 0.5139,
      "step": 47164
    },
    {
      "epoch": 2.6411132265651247,
      "grad_norm": 1.1590670347213745,
      "learning_rate": 9.285657894736842e-05,
      "loss": 0.3705,
      "step": 47165
    },
    {
      "epoch": 2.641169223877254,
      "grad_norm": 1.179772973060608,
      "learning_rate": 9.285631578947369e-05,
      "loss": 0.313,
      "step": 47166
    },
    {
      "epoch": 2.6412252211893827,
      "grad_norm": 1.2565361261367798,
      "learning_rate": 9.285605263157895e-05,
      "loss": 0.481,
      "step": 47167
    },
    {
      "epoch": 2.6412812185015118,
      "grad_norm": 1.1663827896118164,
      "learning_rate": 9.285578947368421e-05,
      "loss": 0.3625,
      "step": 47168
    },
    {
      "epoch": 2.641337215813641,
      "grad_norm": 1.2567182779312134,
      "learning_rate": 9.285552631578947e-05,
      "loss": 0.3578,
      "step": 47169
    },
    {
      "epoch": 2.64139321312577,
      "grad_norm": 1.1425122022628784,
      "learning_rate": 9.285526315789474e-05,
      "loss": 0.3073,
      "step": 47170
    },
    {
      "epoch": 2.641449210437899,
      "grad_norm": 1.2015430927276611,
      "learning_rate": 9.2855e-05,
      "loss": 0.3737,
      "step": 47171
    },
    {
      "epoch": 2.641505207750028,
      "grad_norm": 1.0744110345840454,
      "learning_rate": 9.285473684210528e-05,
      "loss": 0.3106,
      "step": 47172
    },
    {
      "epoch": 2.641561205062157,
      "grad_norm": 1.582350254058838,
      "learning_rate": 9.285447368421052e-05,
      "loss": 0.4133,
      "step": 47173
    },
    {
      "epoch": 2.641617202374286,
      "grad_norm": 1.3005602359771729,
      "learning_rate": 9.28542105263158e-05,
      "loss": 0.327,
      "step": 47174
    },
    {
      "epoch": 2.641673199686415,
      "grad_norm": 1.4784561395645142,
      "learning_rate": 9.285394736842106e-05,
      "loss": 0.4519,
      "step": 47175
    },
    {
      "epoch": 2.641729196998544,
      "grad_norm": 0.9423432350158691,
      "learning_rate": 9.285368421052632e-05,
      "loss": 0.3176,
      "step": 47176
    },
    {
      "epoch": 2.641785194310673,
      "grad_norm": 1.397945761680603,
      "learning_rate": 9.285342105263159e-05,
      "loss": 0.41,
      "step": 47177
    },
    {
      "epoch": 2.641841191622802,
      "grad_norm": 1.6130889654159546,
      "learning_rate": 9.285315789473684e-05,
      "loss": 0.4107,
      "step": 47178
    },
    {
      "epoch": 2.641897188934931,
      "grad_norm": 1.1258587837219238,
      "learning_rate": 9.285289473684211e-05,
      "loss": 0.3231,
      "step": 47179
    },
    {
      "epoch": 2.64195318624706,
      "grad_norm": 1.605398416519165,
      "learning_rate": 9.285263157894737e-05,
      "loss": 0.3592,
      "step": 47180
    },
    {
      "epoch": 2.642009183559189,
      "grad_norm": 1.09932541847229,
      "learning_rate": 9.285236842105264e-05,
      "loss": 0.3887,
      "step": 47181
    },
    {
      "epoch": 2.642065180871318,
      "grad_norm": 1.2176731824874878,
      "learning_rate": 9.28521052631579e-05,
      "loss": 0.3711,
      "step": 47182
    },
    {
      "epoch": 2.642121178183447,
      "grad_norm": 1.0824843645095825,
      "learning_rate": 9.285184210526316e-05,
      "loss": 0.4474,
      "step": 47183
    },
    {
      "epoch": 2.642177175495576,
      "grad_norm": 0.9846916794776917,
      "learning_rate": 9.285157894736842e-05,
      "loss": 0.327,
      "step": 47184
    },
    {
      "epoch": 2.642233172807705,
      "grad_norm": 1.218879222869873,
      "learning_rate": 9.28513157894737e-05,
      "loss": 0.423,
      "step": 47185
    },
    {
      "epoch": 2.642289170119834,
      "grad_norm": 1.1151536703109741,
      "learning_rate": 9.285105263157895e-05,
      "loss": 0.4367,
      "step": 47186
    },
    {
      "epoch": 2.642345167431963,
      "grad_norm": 1.048687219619751,
      "learning_rate": 9.285078947368421e-05,
      "loss": 0.3578,
      "step": 47187
    },
    {
      "epoch": 2.642401164744092,
      "grad_norm": 1.1753637790679932,
      "learning_rate": 9.285052631578947e-05,
      "loss": 0.3725,
      "step": 47188
    },
    {
      "epoch": 2.642457162056221,
      "grad_norm": 1.3802372217178345,
      "learning_rate": 9.285026315789475e-05,
      "loss": 0.4557,
      "step": 47189
    },
    {
      "epoch": 2.64251315936835,
      "grad_norm": 1.2999836206436157,
      "learning_rate": 9.285000000000001e-05,
      "loss": 0.2983,
      "step": 47190
    },
    {
      "epoch": 2.642569156680479,
      "grad_norm": 0.8757368922233582,
      "learning_rate": 9.284973684210527e-05,
      "loss": 0.38,
      "step": 47191
    },
    {
      "epoch": 2.6426251539926082,
      "grad_norm": 0.9733116626739502,
      "learning_rate": 9.284947368421053e-05,
      "loss": 0.3875,
      "step": 47192
    },
    {
      "epoch": 2.6426811513047372,
      "grad_norm": 1.4432021379470825,
      "learning_rate": 9.284921052631579e-05,
      "loss": 0.38,
      "step": 47193
    },
    {
      "epoch": 2.6427371486168663,
      "grad_norm": 1.0729929208755493,
      "learning_rate": 9.284894736842106e-05,
      "loss": 0.3593,
      "step": 47194
    },
    {
      "epoch": 2.6427931459289953,
      "grad_norm": 1.0462443828582764,
      "learning_rate": 9.284868421052632e-05,
      "loss": 0.3983,
      "step": 47195
    },
    {
      "epoch": 2.6428491432411243,
      "grad_norm": 1.0544977188110352,
      "learning_rate": 9.284842105263158e-05,
      "loss": 0.4023,
      "step": 47196
    },
    {
      "epoch": 2.6429051405532533,
      "grad_norm": 1.2211987972259521,
      "learning_rate": 9.284815789473684e-05,
      "loss": 0.329,
      "step": 47197
    },
    {
      "epoch": 2.6429611378653823,
      "grad_norm": 1.1181702613830566,
      "learning_rate": 9.284789473684211e-05,
      "loss": 0.4007,
      "step": 47198
    },
    {
      "epoch": 2.6430171351775114,
      "grad_norm": 1.0558886528015137,
      "learning_rate": 9.284763157894737e-05,
      "loss": 0.374,
      "step": 47199
    },
    {
      "epoch": 2.6430731324896404,
      "grad_norm": 1.1354585886001587,
      "learning_rate": 9.284736842105265e-05,
      "loss": 0.385,
      "step": 47200
    },
    {
      "epoch": 2.6431291298017694,
      "grad_norm": 1.1070374250411987,
      "learning_rate": 9.284710526315789e-05,
      "loss": 0.3705,
      "step": 47201
    },
    {
      "epoch": 2.6431851271138984,
      "grad_norm": 1.0600956678390503,
      "learning_rate": 9.284684210526316e-05,
      "loss": 0.3822,
      "step": 47202
    },
    {
      "epoch": 2.6432411244260274,
      "grad_norm": 1.2256706953048706,
      "learning_rate": 9.284657894736842e-05,
      "loss": 0.5034,
      "step": 47203
    },
    {
      "epoch": 2.6432971217381565,
      "grad_norm": 1.125280499458313,
      "learning_rate": 9.28463157894737e-05,
      "loss": 0.3256,
      "step": 47204
    },
    {
      "epoch": 2.6433531190502855,
      "grad_norm": 1.1259568929672241,
      "learning_rate": 9.284605263157894e-05,
      "loss": 0.4374,
      "step": 47205
    },
    {
      "epoch": 2.6434091163624145,
      "grad_norm": 1.5906413793563843,
      "learning_rate": 9.284578947368422e-05,
      "loss": 0.3999,
      "step": 47206
    },
    {
      "epoch": 2.6434651136745435,
      "grad_norm": 1.1292152404785156,
      "learning_rate": 9.284552631578948e-05,
      "loss": 0.3563,
      "step": 47207
    },
    {
      "epoch": 2.6435211109866725,
      "grad_norm": 1.183517575263977,
      "learning_rate": 9.284526315789475e-05,
      "loss": 0.3691,
      "step": 47208
    },
    {
      "epoch": 2.6435771082988015,
      "grad_norm": 1.117695689201355,
      "learning_rate": 9.284500000000001e-05,
      "loss": 0.3178,
      "step": 47209
    },
    {
      "epoch": 2.6436331056109306,
      "grad_norm": 1.180788516998291,
      "learning_rate": 9.284473684210526e-05,
      "loss": 0.3938,
      "step": 47210
    },
    {
      "epoch": 2.6436891029230596,
      "grad_norm": 0.9863051772117615,
      "learning_rate": 9.284447368421053e-05,
      "loss": 0.3903,
      "step": 47211
    },
    {
      "epoch": 2.6437451002351886,
      "grad_norm": 1.2895848751068115,
      "learning_rate": 9.284421052631579e-05,
      "loss": 0.3234,
      "step": 47212
    },
    {
      "epoch": 2.6438010975473176,
      "grad_norm": 1.3952655792236328,
      "learning_rate": 9.284394736842106e-05,
      "loss": 0.4951,
      "step": 47213
    },
    {
      "epoch": 2.6438570948594466,
      "grad_norm": 1.3119497299194336,
      "learning_rate": 9.284368421052632e-05,
      "loss": 0.3507,
      "step": 47214
    },
    {
      "epoch": 2.6439130921715757,
      "grad_norm": 5.615248203277588,
      "learning_rate": 9.284342105263158e-05,
      "loss": 0.4002,
      "step": 47215
    },
    {
      "epoch": 2.6439690894837047,
      "grad_norm": 1.4216052293777466,
      "learning_rate": 9.284315789473684e-05,
      "loss": 0.4313,
      "step": 47216
    },
    {
      "epoch": 2.6440250867958337,
      "grad_norm": 1.1733524799346924,
      "learning_rate": 9.284289473684211e-05,
      "loss": 0.3226,
      "step": 47217
    },
    {
      "epoch": 2.6440810841079627,
      "grad_norm": 1.182119607925415,
      "learning_rate": 9.284263157894737e-05,
      "loss": 0.4195,
      "step": 47218
    },
    {
      "epoch": 2.6441370814200917,
      "grad_norm": 1.0446248054504395,
      "learning_rate": 9.284236842105263e-05,
      "loss": 0.3429,
      "step": 47219
    },
    {
      "epoch": 2.6441930787322208,
      "grad_norm": 1.0906153917312622,
      "learning_rate": 9.28421052631579e-05,
      "loss": 0.3823,
      "step": 47220
    },
    {
      "epoch": 2.6442490760443498,
      "grad_norm": 1.1202232837677002,
      "learning_rate": 9.284184210526317e-05,
      "loss": 0.3345,
      "step": 47221
    },
    {
      "epoch": 2.644305073356479,
      "grad_norm": 1.2506822347640991,
      "learning_rate": 9.284157894736843e-05,
      "loss": 0.4919,
      "step": 47222
    },
    {
      "epoch": 2.644361070668608,
      "grad_norm": 1.3186287879943848,
      "learning_rate": 9.284131578947369e-05,
      "loss": 0.3928,
      "step": 47223
    },
    {
      "epoch": 2.644417067980737,
      "grad_norm": 1.183869481086731,
      "learning_rate": 9.284105263157895e-05,
      "loss": 0.2906,
      "step": 47224
    },
    {
      "epoch": 2.644473065292866,
      "grad_norm": 1.106387734413147,
      "learning_rate": 9.284078947368422e-05,
      "loss": 0.3984,
      "step": 47225
    },
    {
      "epoch": 2.644529062604995,
      "grad_norm": 1.3914591073989868,
      "learning_rate": 9.284052631578948e-05,
      "loss": 0.5154,
      "step": 47226
    },
    {
      "epoch": 2.644585059917124,
      "grad_norm": 0.9328710436820984,
      "learning_rate": 9.284026315789475e-05,
      "loss": 0.4217,
      "step": 47227
    },
    {
      "epoch": 2.644641057229253,
      "grad_norm": 1.1350332498550415,
      "learning_rate": 9.284e-05,
      "loss": 0.4972,
      "step": 47228
    },
    {
      "epoch": 2.644697054541382,
      "grad_norm": 1.0180718898773193,
      "learning_rate": 9.283973684210526e-05,
      "loss": 0.4106,
      "step": 47229
    },
    {
      "epoch": 2.644753051853511,
      "grad_norm": 2.093587875366211,
      "learning_rate": 9.283947368421053e-05,
      "loss": 0.5163,
      "step": 47230
    },
    {
      "epoch": 2.64480904916564,
      "grad_norm": 1.3041139841079712,
      "learning_rate": 9.283921052631579e-05,
      "loss": 0.4097,
      "step": 47231
    },
    {
      "epoch": 2.644865046477769,
      "grad_norm": 1.0746735334396362,
      "learning_rate": 9.283894736842106e-05,
      "loss": 0.329,
      "step": 47232
    },
    {
      "epoch": 2.644921043789898,
      "grad_norm": 1.0773961544036865,
      "learning_rate": 9.283868421052631e-05,
      "loss": 0.3002,
      "step": 47233
    },
    {
      "epoch": 2.644977041102027,
      "grad_norm": 1.2955607175827026,
      "learning_rate": 9.283842105263158e-05,
      "loss": 0.413,
      "step": 47234
    },
    {
      "epoch": 2.645033038414156,
      "grad_norm": 1.051275372505188,
      "learning_rate": 9.283815789473684e-05,
      "loss": 0.3119,
      "step": 47235
    },
    {
      "epoch": 2.645089035726285,
      "grad_norm": 1.5349481105804443,
      "learning_rate": 9.283789473684212e-05,
      "loss": 0.5248,
      "step": 47236
    },
    {
      "epoch": 2.645145033038414,
      "grad_norm": 1.2725114822387695,
      "learning_rate": 9.283763157894738e-05,
      "loss": 0.3738,
      "step": 47237
    },
    {
      "epoch": 2.645201030350543,
      "grad_norm": 1.241368293762207,
      "learning_rate": 9.283736842105264e-05,
      "loss": 0.4822,
      "step": 47238
    },
    {
      "epoch": 2.645257027662672,
      "grad_norm": 1.1658673286437988,
      "learning_rate": 9.28371052631579e-05,
      "loss": 0.416,
      "step": 47239
    },
    {
      "epoch": 2.645313024974801,
      "grad_norm": 1.1237760782241821,
      "learning_rate": 9.283684210526317e-05,
      "loss": 0.411,
      "step": 47240
    },
    {
      "epoch": 2.64536902228693,
      "grad_norm": 1.3881566524505615,
      "learning_rate": 9.283657894736843e-05,
      "loss": 0.5757,
      "step": 47241
    },
    {
      "epoch": 2.645425019599059,
      "grad_norm": 1.1334192752838135,
      "learning_rate": 9.283631578947369e-05,
      "loss": 0.301,
      "step": 47242
    },
    {
      "epoch": 2.645481016911188,
      "grad_norm": 1.0927727222442627,
      "learning_rate": 9.283605263157895e-05,
      "loss": 0.3437,
      "step": 47243
    },
    {
      "epoch": 2.645537014223317,
      "grad_norm": 1.1004325151443481,
      "learning_rate": 9.283578947368422e-05,
      "loss": 0.3649,
      "step": 47244
    },
    {
      "epoch": 2.6455930115354462,
      "grad_norm": 1.276138424873352,
      "learning_rate": 9.283552631578948e-05,
      "loss": 0.4153,
      "step": 47245
    },
    {
      "epoch": 2.6456490088475753,
      "grad_norm": 0.998623251914978,
      "learning_rate": 9.283526315789474e-05,
      "loss": 0.3307,
      "step": 47246
    },
    {
      "epoch": 2.6457050061597043,
      "grad_norm": 1.175546407699585,
      "learning_rate": 9.2835e-05,
      "loss": 0.5867,
      "step": 47247
    },
    {
      "epoch": 2.6457610034718333,
      "grad_norm": 1.2784581184387207,
      "learning_rate": 9.283473684210526e-05,
      "loss": 0.3392,
      "step": 47248
    },
    {
      "epoch": 2.6458170007839623,
      "grad_norm": 1.1743813753128052,
      "learning_rate": 9.283447368421053e-05,
      "loss": 0.3941,
      "step": 47249
    },
    {
      "epoch": 2.6458729980960913,
      "grad_norm": 1.2884882688522339,
      "learning_rate": 9.28342105263158e-05,
      "loss": 0.3728,
      "step": 47250
    },
    {
      "epoch": 2.6459289954082204,
      "grad_norm": 1.058132290840149,
      "learning_rate": 9.283394736842105e-05,
      "loss": 0.4094,
      "step": 47251
    },
    {
      "epoch": 2.6459849927203494,
      "grad_norm": 1.3293882608413696,
      "learning_rate": 9.283368421052631e-05,
      "loss": 0.381,
      "step": 47252
    },
    {
      "epoch": 2.6460409900324784,
      "grad_norm": 1.1831815242767334,
      "learning_rate": 9.283342105263159e-05,
      "loss": 0.3488,
      "step": 47253
    },
    {
      "epoch": 2.6460969873446074,
      "grad_norm": 6.530035972595215,
      "learning_rate": 9.283315789473685e-05,
      "loss": 0.344,
      "step": 47254
    },
    {
      "epoch": 2.6461529846567364,
      "grad_norm": 1.131874918937683,
      "learning_rate": 9.283289473684212e-05,
      "loss": 0.374,
      "step": 47255
    },
    {
      "epoch": 2.6462089819688654,
      "grad_norm": 1.2201600074768066,
      "learning_rate": 9.283263157894737e-05,
      "loss": 0.3764,
      "step": 47256
    },
    {
      "epoch": 2.6462649792809945,
      "grad_norm": 1.1607797145843506,
      "learning_rate": 9.283236842105264e-05,
      "loss": 0.3651,
      "step": 47257
    },
    {
      "epoch": 2.6463209765931235,
      "grad_norm": 1.3369439840316772,
      "learning_rate": 9.28321052631579e-05,
      "loss": 0.4542,
      "step": 47258
    },
    {
      "epoch": 2.6463769739052525,
      "grad_norm": 1.062454104423523,
      "learning_rate": 9.283184210526317e-05,
      "loss": 0.3453,
      "step": 47259
    },
    {
      "epoch": 2.6464329712173815,
      "grad_norm": 1.051140546798706,
      "learning_rate": 9.283157894736842e-05,
      "loss": 0.3722,
      "step": 47260
    },
    {
      "epoch": 2.6464889685295105,
      "grad_norm": 1.0293678045272827,
      "learning_rate": 9.283131578947369e-05,
      "loss": 0.3157,
      "step": 47261
    },
    {
      "epoch": 2.6465449658416396,
      "grad_norm": 1.340517520904541,
      "learning_rate": 9.283105263157895e-05,
      "loss": 0.4257,
      "step": 47262
    },
    {
      "epoch": 2.6466009631537686,
      "grad_norm": 1.098403811454773,
      "learning_rate": 9.283078947368421e-05,
      "loss": 0.3506,
      "step": 47263
    },
    {
      "epoch": 2.6466569604658976,
      "grad_norm": 1.0847928524017334,
      "learning_rate": 9.283052631578948e-05,
      "loss": 0.4435,
      "step": 47264
    },
    {
      "epoch": 2.6467129577780266,
      "grad_norm": 1.1143977642059326,
      "learning_rate": 9.283026315789473e-05,
      "loss": 0.3855,
      "step": 47265
    },
    {
      "epoch": 2.6467689550901556,
      "grad_norm": 1.0994781255722046,
      "learning_rate": 9.283e-05,
      "loss": 0.4448,
      "step": 47266
    },
    {
      "epoch": 2.6468249524022847,
      "grad_norm": 1.099743366241455,
      "learning_rate": 9.282973684210526e-05,
      "loss": 0.3969,
      "step": 47267
    },
    {
      "epoch": 2.6468809497144137,
      "grad_norm": 0.950365424156189,
      "learning_rate": 9.282947368421054e-05,
      "loss": 0.3622,
      "step": 47268
    },
    {
      "epoch": 2.6469369470265427,
      "grad_norm": 1.0168462991714478,
      "learning_rate": 9.28292105263158e-05,
      "loss": 0.3827,
      "step": 47269
    },
    {
      "epoch": 2.6469929443386717,
      "grad_norm": 1.253541111946106,
      "learning_rate": 9.282894736842106e-05,
      "loss": 0.4529,
      "step": 47270
    },
    {
      "epoch": 2.6470489416508007,
      "grad_norm": 1.2013014554977417,
      "learning_rate": 9.282868421052632e-05,
      "loss": 0.4195,
      "step": 47271
    },
    {
      "epoch": 2.6471049389629298,
      "grad_norm": 1.0248223543167114,
      "learning_rate": 9.282842105263159e-05,
      "loss": 0.3701,
      "step": 47272
    },
    {
      "epoch": 2.6471609362750588,
      "grad_norm": 1.272473692893982,
      "learning_rate": 9.282815789473685e-05,
      "loss": 0.485,
      "step": 47273
    },
    {
      "epoch": 2.647216933587188,
      "grad_norm": 1.1103805303573608,
      "learning_rate": 9.282789473684211e-05,
      "loss": 0.3753,
      "step": 47274
    },
    {
      "epoch": 2.647272930899317,
      "grad_norm": 1.204677939414978,
      "learning_rate": 9.282763157894737e-05,
      "loss": 0.4391,
      "step": 47275
    },
    {
      "epoch": 2.647328928211446,
      "grad_norm": 1.1374101638793945,
      "learning_rate": 9.282736842105264e-05,
      "loss": 0.3061,
      "step": 47276
    },
    {
      "epoch": 2.647384925523575,
      "grad_norm": 1.308250904083252,
      "learning_rate": 9.28271052631579e-05,
      "loss": 0.3883,
      "step": 47277
    },
    {
      "epoch": 2.647440922835704,
      "grad_norm": 1.36335289478302,
      "learning_rate": 9.282684210526316e-05,
      "loss": 0.4856,
      "step": 47278
    },
    {
      "epoch": 2.647496920147833,
      "grad_norm": 1.4866420030593872,
      "learning_rate": 9.282657894736842e-05,
      "loss": 0.5682,
      "step": 47279
    },
    {
      "epoch": 2.647552917459962,
      "grad_norm": 1.0454217195510864,
      "learning_rate": 9.282631578947368e-05,
      "loss": 0.3042,
      "step": 47280
    },
    {
      "epoch": 2.647608914772091,
      "grad_norm": 1.0047128200531006,
      "learning_rate": 9.282605263157895e-05,
      "loss": 0.4155,
      "step": 47281
    },
    {
      "epoch": 2.64766491208422,
      "grad_norm": 1.2558155059814453,
      "learning_rate": 9.282578947368421e-05,
      "loss": 0.368,
      "step": 47282
    },
    {
      "epoch": 2.647720909396349,
      "grad_norm": 1.1668663024902344,
      "learning_rate": 9.282552631578947e-05,
      "loss": 0.4067,
      "step": 47283
    },
    {
      "epoch": 2.647776906708478,
      "grad_norm": 1.1425472497940063,
      "learning_rate": 9.282526315789473e-05,
      "loss": 0.3402,
      "step": 47284
    },
    {
      "epoch": 2.647832904020607,
      "grad_norm": 0.9857205152511597,
      "learning_rate": 9.2825e-05,
      "loss": 0.2803,
      "step": 47285
    },
    {
      "epoch": 2.647888901332736,
      "grad_norm": 1.0854580402374268,
      "learning_rate": 9.282473684210527e-05,
      "loss": 0.3682,
      "step": 47286
    },
    {
      "epoch": 2.647944898644865,
      "grad_norm": 1.3884446620941162,
      "learning_rate": 9.282447368421054e-05,
      "loss": 0.3748,
      "step": 47287
    },
    {
      "epoch": 2.648000895956994,
      "grad_norm": 1.1964740753173828,
      "learning_rate": 9.282421052631579e-05,
      "loss": 0.3485,
      "step": 47288
    },
    {
      "epoch": 2.648056893269123,
      "grad_norm": 1.0712134838104248,
      "learning_rate": 9.282394736842106e-05,
      "loss": 0.3677,
      "step": 47289
    },
    {
      "epoch": 2.648112890581252,
      "grad_norm": 1.2058523893356323,
      "learning_rate": 9.282368421052632e-05,
      "loss": 0.4677,
      "step": 47290
    },
    {
      "epoch": 2.648168887893381,
      "grad_norm": 1.2572628259658813,
      "learning_rate": 9.282342105263159e-05,
      "loss": 0.6057,
      "step": 47291
    },
    {
      "epoch": 2.64822488520551,
      "grad_norm": 1.2058522701263428,
      "learning_rate": 9.282315789473685e-05,
      "loss": 0.3912,
      "step": 47292
    },
    {
      "epoch": 2.648280882517639,
      "grad_norm": 1.2105376720428467,
      "learning_rate": 9.282289473684211e-05,
      "loss": 0.4281,
      "step": 47293
    },
    {
      "epoch": 2.648336879829768,
      "grad_norm": 1.2736655473709106,
      "learning_rate": 9.282263157894737e-05,
      "loss": 0.33,
      "step": 47294
    },
    {
      "epoch": 2.648392877141897,
      "grad_norm": 1.4279433488845825,
      "learning_rate": 9.282236842105264e-05,
      "loss": 0.4527,
      "step": 47295
    },
    {
      "epoch": 2.648448874454026,
      "grad_norm": 1.18088960647583,
      "learning_rate": 9.28221052631579e-05,
      "loss": 0.3482,
      "step": 47296
    },
    {
      "epoch": 2.6485048717661552,
      "grad_norm": 1.1353497505187988,
      "learning_rate": 9.282184210526315e-05,
      "loss": 0.4168,
      "step": 47297
    },
    {
      "epoch": 2.6485608690782843,
      "grad_norm": 0.9586924910545349,
      "learning_rate": 9.282157894736842e-05,
      "loss": 0.3707,
      "step": 47298
    },
    {
      "epoch": 2.6486168663904133,
      "grad_norm": 1.0255540609359741,
      "learning_rate": 9.282131578947368e-05,
      "loss": 0.3196,
      "step": 47299
    },
    {
      "epoch": 2.6486728637025423,
      "grad_norm": 1.0755798816680908,
      "learning_rate": 9.282105263157896e-05,
      "loss": 0.3349,
      "step": 47300
    },
    {
      "epoch": 2.6487288610146713,
      "grad_norm": 0.945899248123169,
      "learning_rate": 9.282078947368422e-05,
      "loss": 0.3058,
      "step": 47301
    },
    {
      "epoch": 2.6487848583268003,
      "grad_norm": 1.1393210887908936,
      "learning_rate": 9.282052631578948e-05,
      "loss": 0.3303,
      "step": 47302
    },
    {
      "epoch": 2.6488408556389293,
      "grad_norm": 1.3186050653457642,
      "learning_rate": 9.282026315789474e-05,
      "loss": 0.3951,
      "step": 47303
    },
    {
      "epoch": 2.6488968529510584,
      "grad_norm": 1.1286953687667847,
      "learning_rate": 9.282000000000001e-05,
      "loss": 0.5325,
      "step": 47304
    },
    {
      "epoch": 2.6489528502631874,
      "grad_norm": 1.2224489450454712,
      "learning_rate": 9.281973684210527e-05,
      "loss": 0.4545,
      "step": 47305
    },
    {
      "epoch": 2.6490088475753164,
      "grad_norm": 1.1597079038619995,
      "learning_rate": 9.281947368421053e-05,
      "loss": 0.3806,
      "step": 47306
    },
    {
      "epoch": 2.6490648448874454,
      "grad_norm": 1.3673837184906006,
      "learning_rate": 9.281921052631579e-05,
      "loss": 0.4738,
      "step": 47307
    },
    {
      "epoch": 2.6491208421995744,
      "grad_norm": 1.3606059551239014,
      "learning_rate": 9.281894736842106e-05,
      "loss": 0.5845,
      "step": 47308
    },
    {
      "epoch": 2.6491768395117035,
      "grad_norm": 1.2028276920318604,
      "learning_rate": 9.281868421052632e-05,
      "loss": 0.5625,
      "step": 47309
    },
    {
      "epoch": 2.6492328368238325,
      "grad_norm": 1.1546616554260254,
      "learning_rate": 9.281842105263158e-05,
      "loss": 0.394,
      "step": 47310
    },
    {
      "epoch": 2.6492888341359615,
      "grad_norm": 1.2212735414505005,
      "learning_rate": 9.281815789473684e-05,
      "loss": 0.5187,
      "step": 47311
    },
    {
      "epoch": 2.6493448314480905,
      "grad_norm": 1.46455979347229,
      "learning_rate": 9.281789473684211e-05,
      "loss": 0.4435,
      "step": 47312
    },
    {
      "epoch": 2.6494008287602195,
      "grad_norm": 1.1484616994857788,
      "learning_rate": 9.281763157894737e-05,
      "loss": 0.4464,
      "step": 47313
    },
    {
      "epoch": 2.6494568260723486,
      "grad_norm": 1.334641695022583,
      "learning_rate": 9.281736842105263e-05,
      "loss": 0.5189,
      "step": 47314
    },
    {
      "epoch": 2.6495128233844776,
      "grad_norm": 1.1570926904678345,
      "learning_rate": 9.28171052631579e-05,
      "loss": 0.3425,
      "step": 47315
    },
    {
      "epoch": 2.6495688206966066,
      "grad_norm": 1.4560905694961548,
      "learning_rate": 9.281684210526315e-05,
      "loss": 0.3559,
      "step": 47316
    },
    {
      "epoch": 2.6496248180087356,
      "grad_norm": 0.9309477210044861,
      "learning_rate": 9.281657894736843e-05,
      "loss": 0.369,
      "step": 47317
    },
    {
      "epoch": 2.6496808153208646,
      "grad_norm": 1.1213810443878174,
      "learning_rate": 9.281631578947369e-05,
      "loss": 0.3781,
      "step": 47318
    },
    {
      "epoch": 2.6497368126329937,
      "grad_norm": 1.2483534812927246,
      "learning_rate": 9.281605263157896e-05,
      "loss": 0.5511,
      "step": 47319
    },
    {
      "epoch": 2.6497928099451227,
      "grad_norm": 1.2032537460327148,
      "learning_rate": 9.28157894736842e-05,
      "loss": 0.5573,
      "step": 47320
    },
    {
      "epoch": 2.6498488072572517,
      "grad_norm": 1.1514328718185425,
      "learning_rate": 9.281552631578948e-05,
      "loss": 0.4671,
      "step": 47321
    },
    {
      "epoch": 2.6499048045693807,
      "grad_norm": 1.3692694902420044,
      "learning_rate": 9.281526315789474e-05,
      "loss": 0.4237,
      "step": 47322
    },
    {
      "epoch": 2.6499608018815097,
      "grad_norm": 1.2762064933776855,
      "learning_rate": 9.281500000000001e-05,
      "loss": 0.4395,
      "step": 47323
    },
    {
      "epoch": 2.6500167991936388,
      "grad_norm": 1.1670019626617432,
      "learning_rate": 9.281473684210527e-05,
      "loss": 0.4248,
      "step": 47324
    },
    {
      "epoch": 2.6500727965057678,
      "grad_norm": 1.1209040880203247,
      "learning_rate": 9.281447368421053e-05,
      "loss": 0.3066,
      "step": 47325
    },
    {
      "epoch": 2.650128793817897,
      "grad_norm": 1.0088346004486084,
      "learning_rate": 9.281421052631579e-05,
      "loss": 0.3912,
      "step": 47326
    },
    {
      "epoch": 2.650184791130026,
      "grad_norm": 1.4989216327667236,
      "learning_rate": 9.281394736842106e-05,
      "loss": 0.6362,
      "step": 47327
    },
    {
      "epoch": 2.650240788442155,
      "grad_norm": 1.0243817567825317,
      "learning_rate": 9.281368421052632e-05,
      "loss": 0.4693,
      "step": 47328
    },
    {
      "epoch": 2.650296785754284,
      "grad_norm": 1.275313138961792,
      "learning_rate": 9.281342105263158e-05,
      "loss": 0.3666,
      "step": 47329
    },
    {
      "epoch": 2.650352783066413,
      "grad_norm": 1.2796465158462524,
      "learning_rate": 9.281315789473684e-05,
      "loss": 0.4374,
      "step": 47330
    },
    {
      "epoch": 2.650408780378542,
      "grad_norm": 1.2002606391906738,
      "learning_rate": 9.28128947368421e-05,
      "loss": 0.4139,
      "step": 47331
    },
    {
      "epoch": 2.650464777690671,
      "grad_norm": 1.2478233575820923,
      "learning_rate": 9.281263157894738e-05,
      "loss": 0.3436,
      "step": 47332
    },
    {
      "epoch": 2.6505207750028,
      "grad_norm": 1.0271979570388794,
      "learning_rate": 9.281236842105264e-05,
      "loss": 0.3169,
      "step": 47333
    },
    {
      "epoch": 2.650576772314929,
      "grad_norm": 1.110460877418518,
      "learning_rate": 9.28121052631579e-05,
      "loss": 0.5511,
      "step": 47334
    },
    {
      "epoch": 2.650632769627058,
      "grad_norm": 1.2558114528656006,
      "learning_rate": 9.281184210526316e-05,
      "loss": 0.384,
      "step": 47335
    },
    {
      "epoch": 2.650688766939187,
      "grad_norm": 1.1179598569869995,
      "learning_rate": 9.281157894736843e-05,
      "loss": 0.4063,
      "step": 47336
    },
    {
      "epoch": 2.650744764251316,
      "grad_norm": 1.2945671081542969,
      "learning_rate": 9.281131578947369e-05,
      "loss": 0.5105,
      "step": 47337
    },
    {
      "epoch": 2.650800761563445,
      "grad_norm": 1.469787836074829,
      "learning_rate": 9.281105263157895e-05,
      "loss": 0.4152,
      "step": 47338
    },
    {
      "epoch": 2.650856758875574,
      "grad_norm": 1.1698426008224487,
      "learning_rate": 9.281078947368421e-05,
      "loss": 0.4583,
      "step": 47339
    },
    {
      "epoch": 2.650912756187703,
      "grad_norm": 1.0619993209838867,
      "learning_rate": 9.281052631578948e-05,
      "loss": 0.477,
      "step": 47340
    },
    {
      "epoch": 2.650968753499832,
      "grad_norm": 1.285013198852539,
      "learning_rate": 9.281026315789474e-05,
      "loss": 0.4247,
      "step": 47341
    },
    {
      "epoch": 2.651024750811961,
      "grad_norm": 1.3380341529846191,
      "learning_rate": 9.281000000000001e-05,
      "loss": 0.4429,
      "step": 47342
    },
    {
      "epoch": 2.65108074812409,
      "grad_norm": 1.215882420539856,
      "learning_rate": 9.280973684210526e-05,
      "loss": 0.3162,
      "step": 47343
    },
    {
      "epoch": 2.651136745436219,
      "grad_norm": 1.1475400924682617,
      "learning_rate": 9.280947368421053e-05,
      "loss": 0.3682,
      "step": 47344
    },
    {
      "epoch": 2.651192742748348,
      "grad_norm": 1.3506118059158325,
      "learning_rate": 9.28092105263158e-05,
      "loss": 0.4964,
      "step": 47345
    },
    {
      "epoch": 2.651248740060477,
      "grad_norm": 1.0748189687728882,
      "learning_rate": 9.280894736842107e-05,
      "loss": 0.3636,
      "step": 47346
    },
    {
      "epoch": 2.651304737372606,
      "grad_norm": 1.2732067108154297,
      "learning_rate": 9.280868421052633e-05,
      "loss": 0.5344,
      "step": 47347
    },
    {
      "epoch": 2.651360734684735,
      "grad_norm": 1.168340802192688,
      "learning_rate": 9.280842105263159e-05,
      "loss": 0.3395,
      "step": 47348
    },
    {
      "epoch": 2.6514167319968642,
      "grad_norm": 0.9707564115524292,
      "learning_rate": 9.280815789473685e-05,
      "loss": 0.2856,
      "step": 47349
    },
    {
      "epoch": 2.6514727293089932,
      "grad_norm": 0.9960217475891113,
      "learning_rate": 9.28078947368421e-05,
      "loss": 0.4184,
      "step": 47350
    },
    {
      "epoch": 2.6515287266211223,
      "grad_norm": 1.2831878662109375,
      "learning_rate": 9.280763157894738e-05,
      "loss": 0.4136,
      "step": 47351
    },
    {
      "epoch": 2.6515847239332513,
      "grad_norm": 1.343133568763733,
      "learning_rate": 9.280736842105263e-05,
      "loss": 0.4377,
      "step": 47352
    },
    {
      "epoch": 2.6516407212453803,
      "grad_norm": 1.416107177734375,
      "learning_rate": 9.28071052631579e-05,
      "loss": 0.3826,
      "step": 47353
    },
    {
      "epoch": 2.6516967185575093,
      "grad_norm": 1.010925054550171,
      "learning_rate": 9.280684210526316e-05,
      "loss": 0.3297,
      "step": 47354
    },
    {
      "epoch": 2.6517527158696383,
      "grad_norm": 1.0699340105056763,
      "learning_rate": 9.280657894736843e-05,
      "loss": 0.3486,
      "step": 47355
    },
    {
      "epoch": 2.6518087131817674,
      "grad_norm": 1.038496971130371,
      "learning_rate": 9.280631578947369e-05,
      "loss": 0.3042,
      "step": 47356
    },
    {
      "epoch": 2.6518647104938964,
      "grad_norm": 1.157675862312317,
      "learning_rate": 9.280605263157895e-05,
      "loss": 0.4852,
      "step": 47357
    },
    {
      "epoch": 2.6519207078060254,
      "grad_norm": 1.3081722259521484,
      "learning_rate": 9.280578947368421e-05,
      "loss": 0.3504,
      "step": 47358
    },
    {
      "epoch": 2.6519767051181544,
      "grad_norm": 1.2764074802398682,
      "learning_rate": 9.280552631578948e-05,
      "loss": 0.3753,
      "step": 47359
    },
    {
      "epoch": 2.6520327024302834,
      "grad_norm": 1.314063310623169,
      "learning_rate": 9.280526315789474e-05,
      "loss": 0.4127,
      "step": 47360
    },
    {
      "epoch": 2.6520886997424125,
      "grad_norm": 1.0930436849594116,
      "learning_rate": 9.2805e-05,
      "loss": 0.3436,
      "step": 47361
    },
    {
      "epoch": 2.6521446970545415,
      "grad_norm": 1.1373069286346436,
      "learning_rate": 9.280473684210526e-05,
      "loss": 0.3545,
      "step": 47362
    },
    {
      "epoch": 2.6522006943666705,
      "grad_norm": 1.5980607271194458,
      "learning_rate": 9.280447368421054e-05,
      "loss": 0.3955,
      "step": 47363
    },
    {
      "epoch": 2.6522566916787995,
      "grad_norm": 1.351307988166809,
      "learning_rate": 9.28042105263158e-05,
      "loss": 0.5676,
      "step": 47364
    },
    {
      "epoch": 2.6523126889909285,
      "grad_norm": 1.281726360321045,
      "learning_rate": 9.280394736842106e-05,
      "loss": 0.4463,
      "step": 47365
    },
    {
      "epoch": 2.6523686863030576,
      "grad_norm": 1.12949538230896,
      "learning_rate": 9.280368421052632e-05,
      "loss": 0.3685,
      "step": 47366
    },
    {
      "epoch": 2.6524246836151866,
      "grad_norm": 1.0199346542358398,
      "learning_rate": 9.280342105263158e-05,
      "loss": 0.2803,
      "step": 47367
    },
    {
      "epoch": 2.6524806809273156,
      "grad_norm": 1.6842765808105469,
      "learning_rate": 9.280315789473685e-05,
      "loss": 0.4908,
      "step": 47368
    },
    {
      "epoch": 2.6525366782394446,
      "grad_norm": 0.9752827286720276,
      "learning_rate": 9.280289473684211e-05,
      "loss": 0.3323,
      "step": 47369
    },
    {
      "epoch": 2.6525926755515736,
      "grad_norm": 1.1426132917404175,
      "learning_rate": 9.280263157894737e-05,
      "loss": 0.4139,
      "step": 47370
    },
    {
      "epoch": 2.6526486728637027,
      "grad_norm": 6.249517917633057,
      "learning_rate": 9.280236842105263e-05,
      "loss": 0.4077,
      "step": 47371
    },
    {
      "epoch": 2.6527046701758317,
      "grad_norm": 1.3404966592788696,
      "learning_rate": 9.28021052631579e-05,
      "loss": 0.3328,
      "step": 47372
    },
    {
      "epoch": 2.6527606674879607,
      "grad_norm": 1.247960090637207,
      "learning_rate": 9.280184210526316e-05,
      "loss": 0.4228,
      "step": 47373
    },
    {
      "epoch": 2.6528166648000897,
      "grad_norm": 0.9732924699783325,
      "learning_rate": 9.280157894736843e-05,
      "loss": 0.2848,
      "step": 47374
    },
    {
      "epoch": 2.6528726621122187,
      "grad_norm": 1.1455532312393188,
      "learning_rate": 9.280131578947368e-05,
      "loss": 0.5429,
      "step": 47375
    },
    {
      "epoch": 2.6529286594243477,
      "grad_norm": 1.296518087387085,
      "learning_rate": 9.280105263157895e-05,
      "loss": 0.3566,
      "step": 47376
    },
    {
      "epoch": 2.6529846567364768,
      "grad_norm": 1.4415994882583618,
      "learning_rate": 9.280078947368421e-05,
      "loss": 0.2905,
      "step": 47377
    },
    {
      "epoch": 2.653040654048606,
      "grad_norm": 1.1553828716278076,
      "learning_rate": 9.280052631578949e-05,
      "loss": 0.4326,
      "step": 47378
    },
    {
      "epoch": 2.653096651360735,
      "grad_norm": 1.3821310997009277,
      "learning_rate": 9.280026315789475e-05,
      "loss": 0.5058,
      "step": 47379
    },
    {
      "epoch": 2.653152648672864,
      "grad_norm": 1.188806176185608,
      "learning_rate": 9.28e-05,
      "loss": 0.5167,
      "step": 47380
    },
    {
      "epoch": 2.653208645984993,
      "grad_norm": 1.2384036779403687,
      "learning_rate": 9.279973684210527e-05,
      "loss": 0.4195,
      "step": 47381
    },
    {
      "epoch": 2.653264643297122,
      "grad_norm": 1.2527974843978882,
      "learning_rate": 9.279947368421054e-05,
      "loss": 0.4913,
      "step": 47382
    },
    {
      "epoch": 2.653320640609251,
      "grad_norm": 1.0805578231811523,
      "learning_rate": 9.27992105263158e-05,
      "loss": 0.315,
      "step": 47383
    },
    {
      "epoch": 2.65337663792138,
      "grad_norm": 1.1112421751022339,
      "learning_rate": 9.279894736842106e-05,
      "loss": 0.4358,
      "step": 47384
    },
    {
      "epoch": 2.653432635233509,
      "grad_norm": 1.6203702688217163,
      "learning_rate": 9.279868421052632e-05,
      "loss": 0.4184,
      "step": 47385
    },
    {
      "epoch": 2.653488632545638,
      "grad_norm": 1.3104215860366821,
      "learning_rate": 9.279842105263158e-05,
      "loss": 0.4823,
      "step": 47386
    },
    {
      "epoch": 2.653544629857767,
      "grad_norm": 1.146422266960144,
      "learning_rate": 9.279815789473685e-05,
      "loss": 0.4654,
      "step": 47387
    },
    {
      "epoch": 2.653600627169896,
      "grad_norm": 1.1565003395080566,
      "learning_rate": 9.279789473684211e-05,
      "loss": 0.3961,
      "step": 47388
    },
    {
      "epoch": 2.653656624482025,
      "grad_norm": 1.336767315864563,
      "learning_rate": 9.279763157894737e-05,
      "loss": 0.4712,
      "step": 47389
    },
    {
      "epoch": 2.653712621794154,
      "grad_norm": 1.2935867309570312,
      "learning_rate": 9.279736842105263e-05,
      "loss": 0.4177,
      "step": 47390
    },
    {
      "epoch": 2.653768619106283,
      "grad_norm": 1.2471948862075806,
      "learning_rate": 9.27971052631579e-05,
      "loss": 0.4183,
      "step": 47391
    },
    {
      "epoch": 2.653824616418412,
      "grad_norm": 1.2142927646636963,
      "learning_rate": 9.279684210526316e-05,
      "loss": 0.3329,
      "step": 47392
    },
    {
      "epoch": 2.653880613730541,
      "grad_norm": 1.2951856851577759,
      "learning_rate": 9.279657894736842e-05,
      "loss": 0.5949,
      "step": 47393
    },
    {
      "epoch": 2.65393661104267,
      "grad_norm": 1.2066380977630615,
      "learning_rate": 9.279631578947368e-05,
      "loss": 0.3499,
      "step": 47394
    },
    {
      "epoch": 2.653992608354799,
      "grad_norm": 1.0970025062561035,
      "learning_rate": 9.279605263157896e-05,
      "loss": 0.3028,
      "step": 47395
    },
    {
      "epoch": 2.654048605666928,
      "grad_norm": 1.1269798278808594,
      "learning_rate": 9.279578947368422e-05,
      "loss": 0.3506,
      "step": 47396
    },
    {
      "epoch": 2.654104602979057,
      "grad_norm": 1.2115707397460938,
      "learning_rate": 9.279552631578949e-05,
      "loss": 0.3385,
      "step": 47397
    },
    {
      "epoch": 2.654160600291186,
      "grad_norm": 1.1550346612930298,
      "learning_rate": 9.279526315789474e-05,
      "loss": 0.2769,
      "step": 47398
    },
    {
      "epoch": 2.654216597603315,
      "grad_norm": 1.2555100917816162,
      "learning_rate": 9.279500000000001e-05,
      "loss": 0.5388,
      "step": 47399
    },
    {
      "epoch": 2.654272594915444,
      "grad_norm": 1.220895528793335,
      "learning_rate": 9.279473684210527e-05,
      "loss": 0.462,
      "step": 47400
    },
    {
      "epoch": 2.6543285922275732,
      "grad_norm": 1.2442971467971802,
      "learning_rate": 9.279447368421053e-05,
      "loss": 0.3198,
      "step": 47401
    },
    {
      "epoch": 2.6543845895397022,
      "grad_norm": 1.3912742137908936,
      "learning_rate": 9.27942105263158e-05,
      "loss": 0.5326,
      "step": 47402
    },
    {
      "epoch": 2.6544405868518313,
      "grad_norm": 1.3022576570510864,
      "learning_rate": 9.279394736842105e-05,
      "loss": 0.3787,
      "step": 47403
    },
    {
      "epoch": 2.6544965841639603,
      "grad_norm": 2.773176670074463,
      "learning_rate": 9.279368421052632e-05,
      "loss": 0.5761,
      "step": 47404
    },
    {
      "epoch": 2.6545525814760893,
      "grad_norm": 1.1172339916229248,
      "learning_rate": 9.279342105263158e-05,
      "loss": 0.4468,
      "step": 47405
    },
    {
      "epoch": 2.6546085787882183,
      "grad_norm": 1.4781614542007446,
      "learning_rate": 9.279315789473685e-05,
      "loss": 0.3895,
      "step": 47406
    },
    {
      "epoch": 2.6546645761003473,
      "grad_norm": 1.0744173526763916,
      "learning_rate": 9.27928947368421e-05,
      "loss": 0.4567,
      "step": 47407
    },
    {
      "epoch": 2.6547205734124764,
      "grad_norm": 1.2259507179260254,
      "learning_rate": 9.279263157894737e-05,
      "loss": 0.3969,
      "step": 47408
    },
    {
      "epoch": 2.6547765707246054,
      "grad_norm": 1.0328724384307861,
      "learning_rate": 9.279236842105263e-05,
      "loss": 0.3437,
      "step": 47409
    },
    {
      "epoch": 2.6548325680367344,
      "grad_norm": 1.122828722000122,
      "learning_rate": 9.27921052631579e-05,
      "loss": 0.4198,
      "step": 47410
    },
    {
      "epoch": 2.6548885653488634,
      "grad_norm": 1.177795171737671,
      "learning_rate": 9.279184210526317e-05,
      "loss": 0.428,
      "step": 47411
    },
    {
      "epoch": 2.6549445626609924,
      "grad_norm": 0.9679808616638184,
      "learning_rate": 9.279157894736843e-05,
      "loss": 0.3814,
      "step": 47412
    },
    {
      "epoch": 2.6550005599731215,
      "grad_norm": 1.1406450271606445,
      "learning_rate": 9.279131578947369e-05,
      "loss": 0.4707,
      "step": 47413
    },
    {
      "epoch": 2.6550565572852505,
      "grad_norm": 1.7606542110443115,
      "learning_rate": 9.279105263157896e-05,
      "loss": 0.3474,
      "step": 47414
    },
    {
      "epoch": 2.6551125545973795,
      "grad_norm": 1.4737811088562012,
      "learning_rate": 9.279078947368422e-05,
      "loss": 0.3827,
      "step": 47415
    },
    {
      "epoch": 2.655168551909508,
      "grad_norm": 1.1982736587524414,
      "learning_rate": 9.279052631578948e-05,
      "loss": 0.3369,
      "step": 47416
    },
    {
      "epoch": 2.6552245492216375,
      "grad_norm": 1.1504876613616943,
      "learning_rate": 9.279026315789474e-05,
      "loss": 0.4056,
      "step": 47417
    },
    {
      "epoch": 2.655280546533766,
      "grad_norm": 1.2437138557434082,
      "learning_rate": 9.279e-05,
      "loss": 0.5052,
      "step": 47418
    },
    {
      "epoch": 2.6553365438458956,
      "grad_norm": 1.0265378952026367,
      "learning_rate": 9.278973684210527e-05,
      "loss": 0.3347,
      "step": 47419
    },
    {
      "epoch": 2.655392541158024,
      "grad_norm": 4.583118438720703,
      "learning_rate": 9.278947368421053e-05,
      "loss": 0.4028,
      "step": 47420
    },
    {
      "epoch": 2.6554485384701536,
      "grad_norm": 1.1025794744491577,
      "learning_rate": 9.278921052631579e-05,
      "loss": 0.3467,
      "step": 47421
    },
    {
      "epoch": 2.655504535782282,
      "grad_norm": 1.2353644371032715,
      "learning_rate": 9.278894736842105e-05,
      "loss": 0.3243,
      "step": 47422
    },
    {
      "epoch": 2.6555605330944116,
      "grad_norm": 1.1974589824676514,
      "learning_rate": 9.278868421052632e-05,
      "loss": 0.3577,
      "step": 47423
    },
    {
      "epoch": 2.65561653040654,
      "grad_norm": 1.1033002138137817,
      "learning_rate": 9.278842105263158e-05,
      "loss": 0.3417,
      "step": 47424
    },
    {
      "epoch": 2.6556725277186697,
      "grad_norm": 1.198957085609436,
      "learning_rate": 9.278815789473684e-05,
      "loss": 0.3737,
      "step": 47425
    },
    {
      "epoch": 2.6557285250307983,
      "grad_norm": 1.387298345565796,
      "learning_rate": 9.27878947368421e-05,
      "loss": 0.5178,
      "step": 47426
    },
    {
      "epoch": 2.6557845223429277,
      "grad_norm": 0.951326310634613,
      "learning_rate": 9.278763157894738e-05,
      "loss": 0.3306,
      "step": 47427
    },
    {
      "epoch": 2.6558405196550563,
      "grad_norm": 1.0141007900238037,
      "learning_rate": 9.278736842105264e-05,
      "loss": 0.3155,
      "step": 47428
    },
    {
      "epoch": 2.6558965169671858,
      "grad_norm": 1.4019529819488525,
      "learning_rate": 9.278710526315791e-05,
      "loss": 0.4005,
      "step": 47429
    },
    {
      "epoch": 2.6559525142793143,
      "grad_norm": 1.391941785812378,
      "learning_rate": 9.278684210526315e-05,
      "loss": 0.4454,
      "step": 47430
    },
    {
      "epoch": 2.656008511591444,
      "grad_norm": 1.25059175491333,
      "learning_rate": 9.278657894736843e-05,
      "loss": 0.4378,
      "step": 47431
    },
    {
      "epoch": 2.6560645089035724,
      "grad_norm": 1.3228929042816162,
      "learning_rate": 9.278631578947369e-05,
      "loss": 0.4045,
      "step": 47432
    },
    {
      "epoch": 2.656120506215702,
      "grad_norm": 1.3107473850250244,
      "learning_rate": 9.278605263157896e-05,
      "loss": 0.4424,
      "step": 47433
    },
    {
      "epoch": 2.6561765035278304,
      "grad_norm": 1.0154026746749878,
      "learning_rate": 9.278578947368422e-05,
      "loss": 0.4263,
      "step": 47434
    },
    {
      "epoch": 2.65623250083996,
      "grad_norm": 1.5786575078964233,
      "learning_rate": 9.278552631578947e-05,
      "loss": 0.5746,
      "step": 47435
    },
    {
      "epoch": 2.6562884981520885,
      "grad_norm": 1.0666754245758057,
      "learning_rate": 9.278526315789474e-05,
      "loss": 0.3852,
      "step": 47436
    },
    {
      "epoch": 2.656344495464218,
      "grad_norm": 1.1458840370178223,
      "learning_rate": 9.2785e-05,
      "loss": 0.3973,
      "step": 47437
    },
    {
      "epoch": 2.6564004927763465,
      "grad_norm": 1.4595284461975098,
      "learning_rate": 9.278473684210527e-05,
      "loss": 0.4544,
      "step": 47438
    },
    {
      "epoch": 2.656456490088476,
      "grad_norm": 1.2383803129196167,
      "learning_rate": 9.278447368421053e-05,
      "loss": 0.4303,
      "step": 47439
    },
    {
      "epoch": 2.6565124874006045,
      "grad_norm": 1.1134872436523438,
      "learning_rate": 9.278421052631579e-05,
      "loss": 0.3184,
      "step": 47440
    },
    {
      "epoch": 2.656568484712734,
      "grad_norm": 1.103294849395752,
      "learning_rate": 9.278394736842105e-05,
      "loss": 0.4878,
      "step": 47441
    },
    {
      "epoch": 2.6566244820248626,
      "grad_norm": 1.2575212717056274,
      "learning_rate": 9.278368421052633e-05,
      "loss": 0.3501,
      "step": 47442
    },
    {
      "epoch": 2.656680479336992,
      "grad_norm": 1.1125930547714233,
      "learning_rate": 9.278342105263159e-05,
      "loss": 0.361,
      "step": 47443
    },
    {
      "epoch": 2.6567364766491206,
      "grad_norm": 1.5357236862182617,
      "learning_rate": 9.278315789473685e-05,
      "loss": 0.53,
      "step": 47444
    },
    {
      "epoch": 2.65679247396125,
      "grad_norm": 1.138389229774475,
      "learning_rate": 9.27828947368421e-05,
      "loss": 0.3939,
      "step": 47445
    },
    {
      "epoch": 2.6568484712733786,
      "grad_norm": 1.6490459442138672,
      "learning_rate": 9.278263157894738e-05,
      "loss": 0.4478,
      "step": 47446
    },
    {
      "epoch": 2.656904468585508,
      "grad_norm": 1.3572652339935303,
      "learning_rate": 9.278236842105264e-05,
      "loss": 0.5394,
      "step": 47447
    },
    {
      "epoch": 2.6569604658976367,
      "grad_norm": 0.926978349685669,
      "learning_rate": 9.27821052631579e-05,
      "loss": 0.2687,
      "step": 47448
    },
    {
      "epoch": 2.657016463209766,
      "grad_norm": 1.1402649879455566,
      "learning_rate": 9.278184210526316e-05,
      "loss": 0.394,
      "step": 47449
    },
    {
      "epoch": 2.6570724605218947,
      "grad_norm": 1.4772206544876099,
      "learning_rate": 9.278157894736843e-05,
      "loss": 0.5178,
      "step": 47450
    },
    {
      "epoch": 2.657128457834024,
      "grad_norm": 1.31520676612854,
      "learning_rate": 9.278131578947369e-05,
      "loss": 0.5268,
      "step": 47451
    },
    {
      "epoch": 2.6571844551461528,
      "grad_norm": 1.0168590545654297,
      "learning_rate": 9.278105263157895e-05,
      "loss": 0.3349,
      "step": 47452
    },
    {
      "epoch": 2.657240452458282,
      "grad_norm": 1.129348874092102,
      "learning_rate": 9.278078947368421e-05,
      "loss": 0.3473,
      "step": 47453
    },
    {
      "epoch": 2.657296449770411,
      "grad_norm": 1.2584019899368286,
      "learning_rate": 9.278052631578947e-05,
      "loss": 0.3661,
      "step": 47454
    },
    {
      "epoch": 2.6573524470825403,
      "grad_norm": 1.8298060894012451,
      "learning_rate": 9.278026315789474e-05,
      "loss": 0.6074,
      "step": 47455
    },
    {
      "epoch": 2.657408444394669,
      "grad_norm": 1.0128612518310547,
      "learning_rate": 9.278e-05,
      "loss": 0.3593,
      "step": 47456
    },
    {
      "epoch": 2.6574644417067983,
      "grad_norm": 1.1211118698120117,
      "learning_rate": 9.277973684210528e-05,
      "loss": 0.4933,
      "step": 47457
    },
    {
      "epoch": 2.657520439018927,
      "grad_norm": 1.2811659574508667,
      "learning_rate": 9.277947368421052e-05,
      "loss": 0.375,
      "step": 47458
    },
    {
      "epoch": 2.6575764363310563,
      "grad_norm": 1.6519309282302856,
      "learning_rate": 9.27792105263158e-05,
      "loss": 0.4228,
      "step": 47459
    },
    {
      "epoch": 2.657632433643185,
      "grad_norm": 1.163385272026062,
      "learning_rate": 9.277894736842106e-05,
      "loss": 0.3622,
      "step": 47460
    },
    {
      "epoch": 2.6576884309553144,
      "grad_norm": 1.2200430631637573,
      "learning_rate": 9.277868421052633e-05,
      "loss": 0.3449,
      "step": 47461
    },
    {
      "epoch": 2.657744428267443,
      "grad_norm": 1.2044298648834229,
      "learning_rate": 9.277842105263157e-05,
      "loss": 0.4778,
      "step": 47462
    },
    {
      "epoch": 2.6578004255795724,
      "grad_norm": 1.1120169162750244,
      "learning_rate": 9.277815789473685e-05,
      "loss": 0.4296,
      "step": 47463
    },
    {
      "epoch": 2.657856422891701,
      "grad_norm": 1.1753687858581543,
      "learning_rate": 9.277789473684211e-05,
      "loss": 0.434,
      "step": 47464
    },
    {
      "epoch": 2.6579124202038305,
      "grad_norm": 1.3142558336257935,
      "learning_rate": 9.277763157894738e-05,
      "loss": 0.3449,
      "step": 47465
    },
    {
      "epoch": 2.657968417515959,
      "grad_norm": 1.0876353979110718,
      "learning_rate": 9.277736842105264e-05,
      "loss": 0.3579,
      "step": 47466
    },
    {
      "epoch": 2.6580244148280885,
      "grad_norm": 1.6637771129608154,
      "learning_rate": 9.27771052631579e-05,
      "loss": 0.5025,
      "step": 47467
    },
    {
      "epoch": 2.658080412140217,
      "grad_norm": 1.139370322227478,
      "learning_rate": 9.277684210526316e-05,
      "loss": 0.3388,
      "step": 47468
    },
    {
      "epoch": 2.6581364094523465,
      "grad_norm": 4.984150409698486,
      "learning_rate": 9.277657894736843e-05,
      "loss": 0.4036,
      "step": 47469
    },
    {
      "epoch": 2.658192406764475,
      "grad_norm": 1.0501899719238281,
      "learning_rate": 9.277631578947369e-05,
      "loss": 0.3863,
      "step": 47470
    },
    {
      "epoch": 2.6582484040766046,
      "grad_norm": 1.1435717344284058,
      "learning_rate": 9.277605263157895e-05,
      "loss": 0.564,
      "step": 47471
    },
    {
      "epoch": 2.658304401388733,
      "grad_norm": 1.5279697179794312,
      "learning_rate": 9.277578947368421e-05,
      "loss": 0.416,
      "step": 47472
    },
    {
      "epoch": 2.6583603987008626,
      "grad_norm": 1.0355006456375122,
      "learning_rate": 9.277552631578947e-05,
      "loss": 0.3201,
      "step": 47473
    },
    {
      "epoch": 2.658416396012991,
      "grad_norm": 1.1237143278121948,
      "learning_rate": 9.277526315789475e-05,
      "loss": 0.4628,
      "step": 47474
    },
    {
      "epoch": 2.6584723933251206,
      "grad_norm": 1.1637531518936157,
      "learning_rate": 9.2775e-05,
      "loss": 0.4,
      "step": 47475
    },
    {
      "epoch": 2.658528390637249,
      "grad_norm": 1.2168872356414795,
      "learning_rate": 9.277473684210527e-05,
      "loss": 0.3764,
      "step": 47476
    },
    {
      "epoch": 2.6585843879493787,
      "grad_norm": 1.3048585653305054,
      "learning_rate": 9.277447368421052e-05,
      "loss": 0.4718,
      "step": 47477
    },
    {
      "epoch": 2.6586403852615073,
      "grad_norm": 1.0684360265731812,
      "learning_rate": 9.27742105263158e-05,
      "loss": 0.5087,
      "step": 47478
    },
    {
      "epoch": 2.6586963825736367,
      "grad_norm": 1.2147250175476074,
      "learning_rate": 9.277394736842106e-05,
      "loss": 0.4338,
      "step": 47479
    },
    {
      "epoch": 2.6587523798857653,
      "grad_norm": 1.0636107921600342,
      "learning_rate": 9.277368421052632e-05,
      "loss": 0.2823,
      "step": 47480
    },
    {
      "epoch": 2.6588083771978948,
      "grad_norm": 1.0709655284881592,
      "learning_rate": 9.277342105263158e-05,
      "loss": 0.3091,
      "step": 47481
    },
    {
      "epoch": 2.6588643745100233,
      "grad_norm": 1.2647533416748047,
      "learning_rate": 9.277315789473685e-05,
      "loss": 0.4359,
      "step": 47482
    },
    {
      "epoch": 2.658920371822153,
      "grad_norm": 1.1466622352600098,
      "learning_rate": 9.277289473684211e-05,
      "loss": 0.4158,
      "step": 47483
    },
    {
      "epoch": 2.6589763691342814,
      "grad_norm": 1.3347880840301514,
      "learning_rate": 9.277263157894738e-05,
      "loss": 0.413,
      "step": 47484
    },
    {
      "epoch": 2.659032366446411,
      "grad_norm": 2.219022035598755,
      "learning_rate": 9.277236842105263e-05,
      "loss": 0.4263,
      "step": 47485
    },
    {
      "epoch": 2.6590883637585394,
      "grad_norm": 1.1262099742889404,
      "learning_rate": 9.27721052631579e-05,
      "loss": 0.3577,
      "step": 47486
    },
    {
      "epoch": 2.659144361070669,
      "grad_norm": 1.1133038997650146,
      "learning_rate": 9.277184210526316e-05,
      "loss": 0.353,
      "step": 47487
    },
    {
      "epoch": 2.6592003583827974,
      "grad_norm": 1.2435643672943115,
      "learning_rate": 9.277157894736842e-05,
      "loss": 0.4069,
      "step": 47488
    },
    {
      "epoch": 2.659256355694927,
      "grad_norm": 1.1728836297988892,
      "learning_rate": 9.27713157894737e-05,
      "loss": 0.3877,
      "step": 47489
    },
    {
      "epoch": 2.6593123530070555,
      "grad_norm": 1.1074012517929077,
      "learning_rate": 9.277105263157894e-05,
      "loss": 0.3398,
      "step": 47490
    },
    {
      "epoch": 2.659368350319185,
      "grad_norm": 1.1784555912017822,
      "learning_rate": 9.277078947368422e-05,
      "loss": 0.3061,
      "step": 47491
    },
    {
      "epoch": 2.6594243476313135,
      "grad_norm": 1.2618969678878784,
      "learning_rate": 9.277052631578947e-05,
      "loss": 0.5589,
      "step": 47492
    },
    {
      "epoch": 2.659480344943443,
      "grad_norm": 2.4528260231018066,
      "learning_rate": 9.277026315789475e-05,
      "loss": 0.4601,
      "step": 47493
    },
    {
      "epoch": 2.6595363422555716,
      "grad_norm": 1.0737504959106445,
      "learning_rate": 9.277000000000001e-05,
      "loss": 0.4018,
      "step": 47494
    },
    {
      "epoch": 2.659592339567701,
      "grad_norm": 1.3025546073913574,
      "learning_rate": 9.276973684210527e-05,
      "loss": 0.4166,
      "step": 47495
    },
    {
      "epoch": 2.6596483368798296,
      "grad_norm": 1.1772878170013428,
      "learning_rate": 9.276947368421053e-05,
      "loss": 0.421,
      "step": 47496
    },
    {
      "epoch": 2.659704334191959,
      "grad_norm": 1.1376831531524658,
      "learning_rate": 9.27692105263158e-05,
      "loss": 0.4561,
      "step": 47497
    },
    {
      "epoch": 2.6597603315040876,
      "grad_norm": 0.9929221272468567,
      "learning_rate": 9.276894736842106e-05,
      "loss": 0.3392,
      "step": 47498
    },
    {
      "epoch": 2.6598163288162167,
      "grad_norm": 1.189436674118042,
      "learning_rate": 9.276868421052632e-05,
      "loss": 0.3948,
      "step": 47499
    },
    {
      "epoch": 2.6598723261283457,
      "grad_norm": 1.2430880069732666,
      "learning_rate": 9.276842105263158e-05,
      "loss": 0.4323,
      "step": 47500
    },
    {
      "epoch": 2.6599283234404747,
      "grad_norm": 1.106285810470581,
      "learning_rate": 9.276815789473685e-05,
      "loss": 0.3353,
      "step": 47501
    },
    {
      "epoch": 2.6599843207526037,
      "grad_norm": 1.01632821559906,
      "learning_rate": 9.276789473684211e-05,
      "loss": 0.2796,
      "step": 47502
    },
    {
      "epoch": 2.6600403180647327,
      "grad_norm": 0.9593456983566284,
      "learning_rate": 9.276763157894737e-05,
      "loss": 0.3113,
      "step": 47503
    },
    {
      "epoch": 2.6600963153768618,
      "grad_norm": 1.1246867179870605,
      "learning_rate": 9.276736842105263e-05,
      "loss": 0.3786,
      "step": 47504
    },
    {
      "epoch": 2.6601523126889908,
      "grad_norm": 1.0139086246490479,
      "learning_rate": 9.276710526315789e-05,
      "loss": 0.3511,
      "step": 47505
    },
    {
      "epoch": 2.66020831000112,
      "grad_norm": 1.1895309686660767,
      "learning_rate": 9.276684210526317e-05,
      "loss": 0.3669,
      "step": 47506
    },
    {
      "epoch": 2.660264307313249,
      "grad_norm": 1.460293173789978,
      "learning_rate": 9.276657894736843e-05,
      "loss": 0.4111,
      "step": 47507
    },
    {
      "epoch": 2.660320304625378,
      "grad_norm": 1.1458388566970825,
      "learning_rate": 9.276631578947368e-05,
      "loss": 0.3038,
      "step": 47508
    },
    {
      "epoch": 2.660376301937507,
      "grad_norm": 1.026963472366333,
      "learning_rate": 9.276605263157894e-05,
      "loss": 0.3554,
      "step": 47509
    },
    {
      "epoch": 2.660432299249636,
      "grad_norm": 1.1522372961044312,
      "learning_rate": 9.276578947368422e-05,
      "loss": 0.3597,
      "step": 47510
    },
    {
      "epoch": 2.660488296561765,
      "grad_norm": 1.4190419912338257,
      "learning_rate": 9.276552631578948e-05,
      "loss": 0.5335,
      "step": 47511
    },
    {
      "epoch": 2.660544293873894,
      "grad_norm": 1.1921638250350952,
      "learning_rate": 9.276526315789474e-05,
      "loss": 0.2936,
      "step": 47512
    },
    {
      "epoch": 2.660600291186023,
      "grad_norm": 1.2409456968307495,
      "learning_rate": 9.2765e-05,
      "loss": 0.3491,
      "step": 47513
    },
    {
      "epoch": 2.660656288498152,
      "grad_norm": 1.6299793720245361,
      "learning_rate": 9.276473684210527e-05,
      "loss": 0.4457,
      "step": 47514
    },
    {
      "epoch": 2.660712285810281,
      "grad_norm": 1.0544073581695557,
      "learning_rate": 9.276447368421053e-05,
      "loss": 0.3843,
      "step": 47515
    },
    {
      "epoch": 2.66076828312241,
      "grad_norm": 1.4418237209320068,
      "learning_rate": 9.27642105263158e-05,
      "loss": 0.4643,
      "step": 47516
    },
    {
      "epoch": 2.660824280434539,
      "grad_norm": 1.3180662393569946,
      "learning_rate": 9.276394736842105e-05,
      "loss": 0.4648,
      "step": 47517
    },
    {
      "epoch": 2.660880277746668,
      "grad_norm": 1.3610684871673584,
      "learning_rate": 9.276368421052632e-05,
      "loss": 0.409,
      "step": 47518
    },
    {
      "epoch": 2.660936275058797,
      "grad_norm": 1.2219200134277344,
      "learning_rate": 9.276342105263158e-05,
      "loss": 0.3288,
      "step": 47519
    },
    {
      "epoch": 2.660992272370926,
      "grad_norm": 1.127295970916748,
      "learning_rate": 9.276315789473686e-05,
      "loss": 0.3613,
      "step": 47520
    },
    {
      "epoch": 2.661048269683055,
      "grad_norm": 0.9292310476303101,
      "learning_rate": 9.276289473684212e-05,
      "loss": 0.3013,
      "step": 47521
    },
    {
      "epoch": 2.661104266995184,
      "grad_norm": 1.5516446828842163,
      "learning_rate": 9.276263157894736e-05,
      "loss": 0.4484,
      "step": 47522
    },
    {
      "epoch": 2.661160264307313,
      "grad_norm": 1.1948965787887573,
      "learning_rate": 9.276236842105263e-05,
      "loss": 0.3744,
      "step": 47523
    },
    {
      "epoch": 2.661216261619442,
      "grad_norm": 1.0692087411880493,
      "learning_rate": 9.27621052631579e-05,
      "loss": 0.4457,
      "step": 47524
    },
    {
      "epoch": 2.661272258931571,
      "grad_norm": 1.2923316955566406,
      "learning_rate": 9.276184210526317e-05,
      "loss": 0.4898,
      "step": 47525
    },
    {
      "epoch": 2.6613282562437,
      "grad_norm": 1.3638238906860352,
      "learning_rate": 9.276157894736843e-05,
      "loss": 0.4599,
      "step": 47526
    },
    {
      "epoch": 2.661384253555829,
      "grad_norm": 1.0682158470153809,
      "learning_rate": 9.276131578947369e-05,
      "loss": 0.3927,
      "step": 47527
    },
    {
      "epoch": 2.661440250867958,
      "grad_norm": 1.2292553186416626,
      "learning_rate": 9.276105263157895e-05,
      "loss": 0.4903,
      "step": 47528
    },
    {
      "epoch": 2.6614962481800872,
      "grad_norm": 1.2993570566177368,
      "learning_rate": 9.276078947368422e-05,
      "loss": 0.4181,
      "step": 47529
    },
    {
      "epoch": 2.6615522454922163,
      "grad_norm": 1.3693057298660278,
      "learning_rate": 9.276052631578948e-05,
      "loss": 0.4961,
      "step": 47530
    },
    {
      "epoch": 2.6616082428043453,
      "grad_norm": 1.1759835481643677,
      "learning_rate": 9.276026315789474e-05,
      "loss": 0.3923,
      "step": 47531
    },
    {
      "epoch": 2.6616642401164743,
      "grad_norm": 1.3165513277053833,
      "learning_rate": 9.276e-05,
      "loss": 0.5496,
      "step": 47532
    },
    {
      "epoch": 2.6617202374286033,
      "grad_norm": 1.412943720817566,
      "learning_rate": 9.275973684210527e-05,
      "loss": 0.3417,
      "step": 47533
    },
    {
      "epoch": 2.6617762347407323,
      "grad_norm": 1.6130561828613281,
      "learning_rate": 9.275947368421053e-05,
      "loss": 0.4891,
      "step": 47534
    },
    {
      "epoch": 2.6618322320528613,
      "grad_norm": 1.1690733432769775,
      "learning_rate": 9.275921052631579e-05,
      "loss": 0.5798,
      "step": 47535
    },
    {
      "epoch": 2.6618882293649904,
      "grad_norm": 1.4799929857254028,
      "learning_rate": 9.275894736842105e-05,
      "loss": 0.5649,
      "step": 47536
    },
    {
      "epoch": 2.6619442266771194,
      "grad_norm": 2.2744240760803223,
      "learning_rate": 9.275868421052633e-05,
      "loss": 0.4301,
      "step": 47537
    },
    {
      "epoch": 2.6620002239892484,
      "grad_norm": 1.0771297216415405,
      "learning_rate": 9.275842105263159e-05,
      "loss": 0.4837,
      "step": 47538
    },
    {
      "epoch": 2.6620562213013774,
      "grad_norm": 1.273617148399353,
      "learning_rate": 9.275815789473684e-05,
      "loss": 0.4142,
      "step": 47539
    },
    {
      "epoch": 2.6621122186135064,
      "grad_norm": 1.191590666770935,
      "learning_rate": 9.27578947368421e-05,
      "loss": 0.4429,
      "step": 47540
    },
    {
      "epoch": 2.6621682159256355,
      "grad_norm": 1.0914942026138306,
      "learning_rate": 9.275763157894736e-05,
      "loss": 0.329,
      "step": 47541
    },
    {
      "epoch": 2.6622242132377645,
      "grad_norm": 1.109630823135376,
      "learning_rate": 9.275736842105264e-05,
      "loss": 0.3849,
      "step": 47542
    },
    {
      "epoch": 2.6622802105498935,
      "grad_norm": 1.02004873752594,
      "learning_rate": 9.27571052631579e-05,
      "loss": 0.3487,
      "step": 47543
    },
    {
      "epoch": 2.6623362078620225,
      "grad_norm": 1.0554287433624268,
      "learning_rate": 9.275684210526317e-05,
      "loss": 0.5054,
      "step": 47544
    },
    {
      "epoch": 2.6623922051741515,
      "grad_norm": 1.2499020099639893,
      "learning_rate": 9.275657894736842e-05,
      "loss": 0.4873,
      "step": 47545
    },
    {
      "epoch": 2.6624482024862806,
      "grad_norm": 1.192051649093628,
      "learning_rate": 9.275631578947369e-05,
      "loss": 0.5205,
      "step": 47546
    },
    {
      "epoch": 2.6625041997984096,
      "grad_norm": 1.1780436038970947,
      "learning_rate": 9.275605263157895e-05,
      "loss": 0.4158,
      "step": 47547
    },
    {
      "epoch": 2.6625601971105386,
      "grad_norm": 1.2634131908416748,
      "learning_rate": 9.275578947368422e-05,
      "loss": 0.4462,
      "step": 47548
    },
    {
      "epoch": 2.6626161944226676,
      "grad_norm": 1.068765640258789,
      "learning_rate": 9.275552631578948e-05,
      "loss": 0.4764,
      "step": 47549
    },
    {
      "epoch": 2.6626721917347966,
      "grad_norm": 1.3456987142562866,
      "learning_rate": 9.275526315789474e-05,
      "loss": 0.3999,
      "step": 47550
    },
    {
      "epoch": 2.6627281890469257,
      "grad_norm": 1.2645747661590576,
      "learning_rate": 9.2755e-05,
      "loss": 0.5458,
      "step": 47551
    },
    {
      "epoch": 2.6627841863590547,
      "grad_norm": 1.092883586883545,
      "learning_rate": 9.275473684210528e-05,
      "loss": 0.3779,
      "step": 47552
    },
    {
      "epoch": 2.6628401836711837,
      "grad_norm": 1.0276927947998047,
      "learning_rate": 9.275447368421054e-05,
      "loss": 0.3754,
      "step": 47553
    },
    {
      "epoch": 2.6628961809833127,
      "grad_norm": 0.9178904294967651,
      "learning_rate": 9.27542105263158e-05,
      "loss": 0.2767,
      "step": 47554
    },
    {
      "epoch": 2.6629521782954417,
      "grad_norm": 1.1357015371322632,
      "learning_rate": 9.275394736842105e-05,
      "loss": 0.5254,
      "step": 47555
    },
    {
      "epoch": 2.6630081756075707,
      "grad_norm": 0.9717265963554382,
      "learning_rate": 9.275368421052631e-05,
      "loss": 0.2569,
      "step": 47556
    },
    {
      "epoch": 2.6630641729196998,
      "grad_norm": 1.4066895246505737,
      "learning_rate": 9.275342105263159e-05,
      "loss": 0.4267,
      "step": 47557
    },
    {
      "epoch": 2.663120170231829,
      "grad_norm": 1.2285960912704468,
      "learning_rate": 9.275315789473685e-05,
      "loss": 0.4188,
      "step": 47558
    },
    {
      "epoch": 2.663176167543958,
      "grad_norm": 1.316941738128662,
      "learning_rate": 9.275289473684211e-05,
      "loss": 0.4087,
      "step": 47559
    },
    {
      "epoch": 2.663232164856087,
      "grad_norm": 1.264624834060669,
      "learning_rate": 9.275263157894737e-05,
      "loss": 0.4329,
      "step": 47560
    },
    {
      "epoch": 2.663288162168216,
      "grad_norm": 1.1164597272872925,
      "learning_rate": 9.275236842105264e-05,
      "loss": 0.2901,
      "step": 47561
    },
    {
      "epoch": 2.663344159480345,
      "grad_norm": 1.1543498039245605,
      "learning_rate": 9.27521052631579e-05,
      "loss": 0.363,
      "step": 47562
    },
    {
      "epoch": 2.663400156792474,
      "grad_norm": 1.302572250366211,
      "learning_rate": 9.275184210526316e-05,
      "loss": 0.4103,
      "step": 47563
    },
    {
      "epoch": 2.663456154104603,
      "grad_norm": 1.5761260986328125,
      "learning_rate": 9.275157894736842e-05,
      "loss": 0.5826,
      "step": 47564
    },
    {
      "epoch": 2.663512151416732,
      "grad_norm": 1.9424961805343628,
      "learning_rate": 9.275131578947369e-05,
      "loss": 0.4995,
      "step": 47565
    },
    {
      "epoch": 2.663568148728861,
      "grad_norm": 1.548863410949707,
      "learning_rate": 9.275105263157895e-05,
      "loss": 0.4664,
      "step": 47566
    },
    {
      "epoch": 2.66362414604099,
      "grad_norm": 1.0366572141647339,
      "learning_rate": 9.275078947368421e-05,
      "loss": 0.3418,
      "step": 47567
    },
    {
      "epoch": 2.663680143353119,
      "grad_norm": 1.2684104442596436,
      "learning_rate": 9.275052631578947e-05,
      "loss": 0.6641,
      "step": 47568
    },
    {
      "epoch": 2.663736140665248,
      "grad_norm": 1.0629364252090454,
      "learning_rate": 9.275026315789475e-05,
      "loss": 0.4015,
      "step": 47569
    },
    {
      "epoch": 2.663792137977377,
      "grad_norm": 1.089124083518982,
      "learning_rate": 9.275e-05,
      "loss": 0.484,
      "step": 47570
    },
    {
      "epoch": 2.663848135289506,
      "grad_norm": 1.1646091938018799,
      "learning_rate": 9.274973684210528e-05,
      "loss": 0.3841,
      "step": 47571
    },
    {
      "epoch": 2.663904132601635,
      "grad_norm": 1.1608027219772339,
      "learning_rate": 9.274947368421052e-05,
      "loss": 0.445,
      "step": 47572
    },
    {
      "epoch": 2.663960129913764,
      "grad_norm": 1.2543586492538452,
      "learning_rate": 9.274921052631578e-05,
      "loss": 0.4621,
      "step": 47573
    },
    {
      "epoch": 2.664016127225893,
      "grad_norm": 1.136059284210205,
      "learning_rate": 9.274894736842106e-05,
      "loss": 0.4294,
      "step": 47574
    },
    {
      "epoch": 2.664072124538022,
      "grad_norm": 1.2578800916671753,
      "learning_rate": 9.274868421052632e-05,
      "loss": 0.3795,
      "step": 47575
    },
    {
      "epoch": 2.664128121850151,
      "grad_norm": 0.8623762726783752,
      "learning_rate": 9.274842105263159e-05,
      "loss": 0.2558,
      "step": 47576
    },
    {
      "epoch": 2.66418411916228,
      "grad_norm": 1.2714905738830566,
      "learning_rate": 9.274815789473684e-05,
      "loss": 0.4006,
      "step": 47577
    },
    {
      "epoch": 2.664240116474409,
      "grad_norm": 1.0374336242675781,
      "learning_rate": 9.274789473684211e-05,
      "loss": 0.3227,
      "step": 47578
    },
    {
      "epoch": 2.664296113786538,
      "grad_norm": 1.0971360206604004,
      "learning_rate": 9.274763157894737e-05,
      "loss": 0.4454,
      "step": 47579
    },
    {
      "epoch": 2.664352111098667,
      "grad_norm": 1.3018372058868408,
      "learning_rate": 9.274736842105264e-05,
      "loss": 0.4517,
      "step": 47580
    },
    {
      "epoch": 2.6644081084107962,
      "grad_norm": 1.4711925983428955,
      "learning_rate": 9.27471052631579e-05,
      "loss": 0.4173,
      "step": 47581
    },
    {
      "epoch": 2.6644641057229252,
      "grad_norm": 1.0012773275375366,
      "learning_rate": 9.274684210526316e-05,
      "loss": 0.3926,
      "step": 47582
    },
    {
      "epoch": 2.6645201030350543,
      "grad_norm": 1.1273266077041626,
      "learning_rate": 9.274657894736842e-05,
      "loss": 0.4498,
      "step": 47583
    },
    {
      "epoch": 2.6645761003471833,
      "grad_norm": 1.4104565382003784,
      "learning_rate": 9.27463157894737e-05,
      "loss": 0.3628,
      "step": 47584
    },
    {
      "epoch": 2.6646320976593123,
      "grad_norm": 1.2100521326065063,
      "learning_rate": 9.274605263157895e-05,
      "loss": 0.3483,
      "step": 47585
    },
    {
      "epoch": 2.6646880949714413,
      "grad_norm": 1.0484782457351685,
      "learning_rate": 9.274578947368421e-05,
      "loss": 0.479,
      "step": 47586
    },
    {
      "epoch": 2.6647440922835703,
      "grad_norm": 1.1991688013076782,
      "learning_rate": 9.274552631578947e-05,
      "loss": 0.3608,
      "step": 47587
    },
    {
      "epoch": 2.6648000895956994,
      "grad_norm": 1.3280956745147705,
      "learning_rate": 9.274526315789475e-05,
      "loss": 0.3743,
      "step": 47588
    },
    {
      "epoch": 2.6648560869078284,
      "grad_norm": 1.2627779245376587,
      "learning_rate": 9.274500000000001e-05,
      "loss": 0.4256,
      "step": 47589
    },
    {
      "epoch": 2.6649120842199574,
      "grad_norm": 1.196451187133789,
      "learning_rate": 9.274473684210527e-05,
      "loss": 0.3947,
      "step": 47590
    },
    {
      "epoch": 2.6649680815320864,
      "grad_norm": 1.1108018159866333,
      "learning_rate": 9.274447368421053e-05,
      "loss": 0.3601,
      "step": 47591
    },
    {
      "epoch": 2.6650240788442154,
      "grad_norm": 1.2697044610977173,
      "learning_rate": 9.274421052631579e-05,
      "loss": 0.5211,
      "step": 47592
    },
    {
      "epoch": 2.6650800761563445,
      "grad_norm": 1.1213499307632446,
      "learning_rate": 9.274394736842106e-05,
      "loss": 0.2913,
      "step": 47593
    },
    {
      "epoch": 2.6651360734684735,
      "grad_norm": 1.0762097835540771,
      "learning_rate": 9.274368421052632e-05,
      "loss": 0.4065,
      "step": 47594
    },
    {
      "epoch": 2.6651920707806025,
      "grad_norm": 1.1248291730880737,
      "learning_rate": 9.274342105263158e-05,
      "loss": 0.3523,
      "step": 47595
    },
    {
      "epoch": 2.6652480680927315,
      "grad_norm": 2.2400875091552734,
      "learning_rate": 9.274315789473684e-05,
      "loss": 0.5946,
      "step": 47596
    },
    {
      "epoch": 2.6653040654048605,
      "grad_norm": 0.987905740737915,
      "learning_rate": 9.274289473684211e-05,
      "loss": 0.3015,
      "step": 47597
    },
    {
      "epoch": 2.6653600627169896,
      "grad_norm": 1.1405946016311646,
      "learning_rate": 9.274263157894737e-05,
      "loss": 0.3567,
      "step": 47598
    },
    {
      "epoch": 2.6654160600291186,
      "grad_norm": 1.245200276374817,
      "learning_rate": 9.274236842105265e-05,
      "loss": 0.3658,
      "step": 47599
    },
    {
      "epoch": 2.6654720573412476,
      "grad_norm": 1.2087043523788452,
      "learning_rate": 9.274210526315789e-05,
      "loss": 0.3692,
      "step": 47600
    },
    {
      "epoch": 2.6655280546533766,
      "grad_norm": 1.0968115329742432,
      "learning_rate": 9.274184210526316e-05,
      "loss": 0.3851,
      "step": 47601
    },
    {
      "epoch": 2.6655840519655056,
      "grad_norm": 1.2881309986114502,
      "learning_rate": 9.274157894736842e-05,
      "loss": 0.3776,
      "step": 47602
    },
    {
      "epoch": 2.6656400492776346,
      "grad_norm": 1.0842046737670898,
      "learning_rate": 9.27413157894737e-05,
      "loss": 0.3982,
      "step": 47603
    },
    {
      "epoch": 2.6656960465897637,
      "grad_norm": 1.14646315574646,
      "learning_rate": 9.274105263157896e-05,
      "loss": 0.3668,
      "step": 47604
    },
    {
      "epoch": 2.6657520439018927,
      "grad_norm": 1.6181786060333252,
      "learning_rate": 9.274078947368422e-05,
      "loss": 0.3516,
      "step": 47605
    },
    {
      "epoch": 2.6658080412140217,
      "grad_norm": 1.264983057975769,
      "learning_rate": 9.274052631578948e-05,
      "loss": 0.3816,
      "step": 47606
    },
    {
      "epoch": 2.6658640385261507,
      "grad_norm": 1.2994030714035034,
      "learning_rate": 9.274026315789475e-05,
      "loss": 0.3726,
      "step": 47607
    },
    {
      "epoch": 2.6659200358382797,
      "grad_norm": 1.1678959131240845,
      "learning_rate": 9.274000000000001e-05,
      "loss": 0.414,
      "step": 47608
    },
    {
      "epoch": 2.6659760331504088,
      "grad_norm": 1.1982603073120117,
      "learning_rate": 9.273973684210526e-05,
      "loss": 0.4165,
      "step": 47609
    },
    {
      "epoch": 2.666032030462538,
      "grad_norm": 1.1993447542190552,
      "learning_rate": 9.273947368421053e-05,
      "loss": 0.4076,
      "step": 47610
    },
    {
      "epoch": 2.666088027774667,
      "grad_norm": 1.4822980165481567,
      "learning_rate": 9.273921052631579e-05,
      "loss": 0.5479,
      "step": 47611
    },
    {
      "epoch": 2.666144025086796,
      "grad_norm": 1.056164026260376,
      "learning_rate": 9.273894736842106e-05,
      "loss": 0.3972,
      "step": 47612
    },
    {
      "epoch": 2.666200022398925,
      "grad_norm": 1.2561849355697632,
      "learning_rate": 9.273868421052632e-05,
      "loss": 0.3351,
      "step": 47613
    },
    {
      "epoch": 2.666256019711054,
      "grad_norm": 1.2079919576644897,
      "learning_rate": 9.273842105263158e-05,
      "loss": 0.4385,
      "step": 47614
    },
    {
      "epoch": 2.666312017023183,
      "grad_norm": 0.975650429725647,
      "learning_rate": 9.273815789473684e-05,
      "loss": 0.3503,
      "step": 47615
    },
    {
      "epoch": 2.666368014335312,
      "grad_norm": 1.1701608896255493,
      "learning_rate": 9.273789473684211e-05,
      "loss": 0.3421,
      "step": 47616
    },
    {
      "epoch": 2.666424011647441,
      "grad_norm": 9.973010063171387,
      "learning_rate": 9.273763157894737e-05,
      "loss": 0.5261,
      "step": 47617
    },
    {
      "epoch": 2.66648000895957,
      "grad_norm": 1.176804542541504,
      "learning_rate": 9.273736842105263e-05,
      "loss": 0.3659,
      "step": 47618
    },
    {
      "epoch": 2.666536006271699,
      "grad_norm": 1.129718542098999,
      "learning_rate": 9.27371052631579e-05,
      "loss": 0.3772,
      "step": 47619
    },
    {
      "epoch": 2.666592003583828,
      "grad_norm": 1.212834358215332,
      "learning_rate": 9.273684210526317e-05,
      "loss": 0.5393,
      "step": 47620
    },
    {
      "epoch": 2.666648000895957,
      "grad_norm": 1.045158863067627,
      "learning_rate": 9.273657894736843e-05,
      "loss": 0.382,
      "step": 47621
    },
    {
      "epoch": 2.666703998208086,
      "grad_norm": 1.3724884986877441,
      "learning_rate": 9.273631578947369e-05,
      "loss": 0.3346,
      "step": 47622
    },
    {
      "epoch": 2.666759995520215,
      "grad_norm": 1.428062915802002,
      "learning_rate": 9.273605263157895e-05,
      "loss": 0.2812,
      "step": 47623
    },
    {
      "epoch": 2.666815992832344,
      "grad_norm": 1.3064239025115967,
      "learning_rate": 9.273578947368422e-05,
      "loss": 0.3652,
      "step": 47624
    },
    {
      "epoch": 2.666871990144473,
      "grad_norm": 1.2269973754882812,
      "learning_rate": 9.273552631578948e-05,
      "loss": 0.4256,
      "step": 47625
    },
    {
      "epoch": 2.666927987456602,
      "grad_norm": 1.1800638437271118,
      "learning_rate": 9.273526315789474e-05,
      "loss": 0.3497,
      "step": 47626
    },
    {
      "epoch": 2.666983984768731,
      "grad_norm": 1.008231520652771,
      "learning_rate": 9.2735e-05,
      "loss": 0.3556,
      "step": 47627
    },
    {
      "epoch": 2.66703998208086,
      "grad_norm": 1.3462977409362793,
      "learning_rate": 9.273473684210526e-05,
      "loss": 0.3348,
      "step": 47628
    },
    {
      "epoch": 2.667095979392989,
      "grad_norm": 1.1017541885375977,
      "learning_rate": 9.273447368421053e-05,
      "loss": 0.3892,
      "step": 47629
    },
    {
      "epoch": 2.667151976705118,
      "grad_norm": 1.2252403497695923,
      "learning_rate": 9.273421052631579e-05,
      "loss": 0.3412,
      "step": 47630
    },
    {
      "epoch": 2.667207974017247,
      "grad_norm": 1.3084675073623657,
      "learning_rate": 9.273394736842106e-05,
      "loss": 0.5525,
      "step": 47631
    },
    {
      "epoch": 2.667263971329376,
      "grad_norm": 1.0266379117965698,
      "learning_rate": 9.273368421052631e-05,
      "loss": 0.4568,
      "step": 47632
    },
    {
      "epoch": 2.6673199686415052,
      "grad_norm": 1.2163810729980469,
      "learning_rate": 9.273342105263158e-05,
      "loss": 0.5049,
      "step": 47633
    },
    {
      "epoch": 2.6673759659536342,
      "grad_norm": 1.2116118669509888,
      "learning_rate": 9.273315789473684e-05,
      "loss": 0.3168,
      "step": 47634
    },
    {
      "epoch": 2.6674319632657633,
      "grad_norm": 1.1918892860412598,
      "learning_rate": 9.273289473684212e-05,
      "loss": 0.3678,
      "step": 47635
    },
    {
      "epoch": 2.6674879605778923,
      "grad_norm": 1.1233115196228027,
      "learning_rate": 9.273263157894738e-05,
      "loss": 0.3353,
      "step": 47636
    },
    {
      "epoch": 2.6675439578900213,
      "grad_norm": 1.2912232875823975,
      "learning_rate": 9.273236842105264e-05,
      "loss": 0.3578,
      "step": 47637
    },
    {
      "epoch": 2.6675999552021503,
      "grad_norm": 1.139631748199463,
      "learning_rate": 9.27321052631579e-05,
      "loss": 0.4722,
      "step": 47638
    },
    {
      "epoch": 2.6676559525142793,
      "grad_norm": 1.1909219026565552,
      "learning_rate": 9.273184210526317e-05,
      "loss": 0.368,
      "step": 47639
    },
    {
      "epoch": 2.6677119498264084,
      "grad_norm": 1.0619488954544067,
      "learning_rate": 9.273157894736843e-05,
      "loss": 0.3735,
      "step": 47640
    },
    {
      "epoch": 2.6677679471385374,
      "grad_norm": 1.5012789964675903,
      "learning_rate": 9.273131578947369e-05,
      "loss": 0.4321,
      "step": 47641
    },
    {
      "epoch": 2.6678239444506664,
      "grad_norm": 0.9738088250160217,
      "learning_rate": 9.273105263157895e-05,
      "loss": 0.4287,
      "step": 47642
    },
    {
      "epoch": 2.6678799417627954,
      "grad_norm": 0.9064452648162842,
      "learning_rate": 9.273078947368421e-05,
      "loss": 0.2431,
      "step": 47643
    },
    {
      "epoch": 2.6679359390749244,
      "grad_norm": 1.1912959814071655,
      "learning_rate": 9.273052631578948e-05,
      "loss": 0.4859,
      "step": 47644
    },
    {
      "epoch": 2.6679919363870535,
      "grad_norm": 1.25887930393219,
      "learning_rate": 9.273026315789474e-05,
      "loss": 0.4192,
      "step": 47645
    },
    {
      "epoch": 2.6680479336991825,
      "grad_norm": 1.5504930019378662,
      "learning_rate": 9.273e-05,
      "loss": 0.4707,
      "step": 47646
    },
    {
      "epoch": 2.6681039310113115,
      "grad_norm": 1.2293529510498047,
      "learning_rate": 9.272973684210526e-05,
      "loss": 0.4578,
      "step": 47647
    },
    {
      "epoch": 2.6681599283234405,
      "grad_norm": 1.8161752223968506,
      "learning_rate": 9.272947368421053e-05,
      "loss": 0.4477,
      "step": 47648
    },
    {
      "epoch": 2.6682159256355695,
      "grad_norm": 1.1746166944503784,
      "learning_rate": 9.27292105263158e-05,
      "loss": 0.3473,
      "step": 47649
    },
    {
      "epoch": 2.6682719229476985,
      "grad_norm": 1.2853899002075195,
      "learning_rate": 9.272894736842105e-05,
      "loss": 0.5221,
      "step": 47650
    },
    {
      "epoch": 2.6683279202598276,
      "grad_norm": 0.9979504346847534,
      "learning_rate": 9.272868421052631e-05,
      "loss": 0.4125,
      "step": 47651
    },
    {
      "epoch": 2.6683839175719566,
      "grad_norm": 1.3434330224990845,
      "learning_rate": 9.272842105263159e-05,
      "loss": 0.4953,
      "step": 47652
    },
    {
      "epoch": 2.6684399148840856,
      "grad_norm": 1.3204206228256226,
      "learning_rate": 9.272815789473685e-05,
      "loss": 0.4644,
      "step": 47653
    },
    {
      "epoch": 2.6684959121962146,
      "grad_norm": 1.3389912843704224,
      "learning_rate": 9.272789473684212e-05,
      "loss": 0.5843,
      "step": 47654
    },
    {
      "epoch": 2.6685519095083436,
      "grad_norm": 1.1455268859863281,
      "learning_rate": 9.272763157894737e-05,
      "loss": 0.4536,
      "step": 47655
    },
    {
      "epoch": 2.6686079068204727,
      "grad_norm": 1.0680243968963623,
      "learning_rate": 9.272736842105264e-05,
      "loss": 0.3066,
      "step": 47656
    },
    {
      "epoch": 2.6686639041326017,
      "grad_norm": 1.2559292316436768,
      "learning_rate": 9.27271052631579e-05,
      "loss": 0.3111,
      "step": 47657
    },
    {
      "epoch": 2.6687199014447307,
      "grad_norm": 1.0821160078048706,
      "learning_rate": 9.272684210526317e-05,
      "loss": 0.387,
      "step": 47658
    },
    {
      "epoch": 2.6687758987568597,
      "grad_norm": 1.4170722961425781,
      "learning_rate": 9.272657894736842e-05,
      "loss": 0.4487,
      "step": 47659
    },
    {
      "epoch": 2.6688318960689887,
      "grad_norm": 1.3915019035339355,
      "learning_rate": 9.272631578947368e-05,
      "loss": 0.5105,
      "step": 47660
    },
    {
      "epoch": 2.6688878933811178,
      "grad_norm": 1.410212755203247,
      "learning_rate": 9.272605263157895e-05,
      "loss": 0.4084,
      "step": 47661
    },
    {
      "epoch": 2.668943890693247,
      "grad_norm": 1.3044352531433105,
      "learning_rate": 9.272578947368421e-05,
      "loss": 0.4028,
      "step": 47662
    },
    {
      "epoch": 2.668999888005376,
      "grad_norm": 1.2218925952911377,
      "learning_rate": 9.272552631578948e-05,
      "loss": 0.3886,
      "step": 47663
    },
    {
      "epoch": 2.669055885317505,
      "grad_norm": 1.5004249811172485,
      "learning_rate": 9.272526315789473e-05,
      "loss": 0.3935,
      "step": 47664
    },
    {
      "epoch": 2.669111882629634,
      "grad_norm": 1.314431071281433,
      "learning_rate": 9.2725e-05,
      "loss": 0.3841,
      "step": 47665
    },
    {
      "epoch": 2.669167879941763,
      "grad_norm": 0.9777299761772156,
      "learning_rate": 9.272473684210526e-05,
      "loss": 0.3477,
      "step": 47666
    },
    {
      "epoch": 2.669223877253892,
      "grad_norm": 1.036908745765686,
      "learning_rate": 9.272447368421054e-05,
      "loss": 0.396,
      "step": 47667
    },
    {
      "epoch": 2.669279874566021,
      "grad_norm": 1.1879122257232666,
      "learning_rate": 9.27242105263158e-05,
      "loss": 0.2804,
      "step": 47668
    },
    {
      "epoch": 2.66933587187815,
      "grad_norm": 1.2770706415176392,
      "learning_rate": 9.272394736842106e-05,
      "loss": 0.5406,
      "step": 47669
    },
    {
      "epoch": 2.669391869190279,
      "grad_norm": 1.174457311630249,
      "learning_rate": 9.272368421052632e-05,
      "loss": 0.4791,
      "step": 47670
    },
    {
      "epoch": 2.669447866502408,
      "grad_norm": 1.2456554174423218,
      "learning_rate": 9.272342105263159e-05,
      "loss": 0.3471,
      "step": 47671
    },
    {
      "epoch": 2.669503863814537,
      "grad_norm": 1.3959336280822754,
      "learning_rate": 9.272315789473685e-05,
      "loss": 0.6184,
      "step": 47672
    },
    {
      "epoch": 2.669559861126666,
      "grad_norm": 1.3390270471572876,
      "learning_rate": 9.272289473684211e-05,
      "loss": 0.3971,
      "step": 47673
    },
    {
      "epoch": 2.669615858438795,
      "grad_norm": 1.325117588043213,
      "learning_rate": 9.272263157894737e-05,
      "loss": 0.39,
      "step": 47674
    },
    {
      "epoch": 2.669671855750924,
      "grad_norm": 1.154538631439209,
      "learning_rate": 9.272236842105264e-05,
      "loss": 0.3239,
      "step": 47675
    },
    {
      "epoch": 2.669727853063053,
      "grad_norm": 1.6201868057250977,
      "learning_rate": 9.27221052631579e-05,
      "loss": 0.6737,
      "step": 47676
    },
    {
      "epoch": 2.669783850375182,
      "grad_norm": 1.2993992567062378,
      "learning_rate": 9.272184210526316e-05,
      "loss": 0.3818,
      "step": 47677
    },
    {
      "epoch": 2.669839847687311,
      "grad_norm": 1.2092458009719849,
      "learning_rate": 9.272157894736842e-05,
      "loss": 0.4103,
      "step": 47678
    },
    {
      "epoch": 2.66989584499944,
      "grad_norm": 1.2663904428482056,
      "learning_rate": 9.272131578947368e-05,
      "loss": 0.4258,
      "step": 47679
    },
    {
      "epoch": 2.669951842311569,
      "grad_norm": 1.0981320142745972,
      "learning_rate": 9.272105263157895e-05,
      "loss": 0.4174,
      "step": 47680
    },
    {
      "epoch": 2.670007839623698,
      "grad_norm": 1.3481465578079224,
      "learning_rate": 9.272078947368421e-05,
      "loss": 0.4323,
      "step": 47681
    },
    {
      "epoch": 2.670063836935827,
      "grad_norm": 1.2047775983810425,
      "learning_rate": 9.272052631578947e-05,
      "loss": 0.4473,
      "step": 47682
    },
    {
      "epoch": 2.670119834247956,
      "grad_norm": 1.2452057600021362,
      "learning_rate": 9.272026315789473e-05,
      "loss": 0.3302,
      "step": 47683
    },
    {
      "epoch": 2.670175831560085,
      "grad_norm": 1.124908447265625,
      "learning_rate": 9.272e-05,
      "loss": 0.347,
      "step": 47684
    },
    {
      "epoch": 2.670231828872214,
      "grad_norm": 1.1847048997879028,
      "learning_rate": 9.271973684210527e-05,
      "loss": 0.451,
      "step": 47685
    },
    {
      "epoch": 2.6702878261843432,
      "grad_norm": 1.3739248514175415,
      "learning_rate": 9.271947368421054e-05,
      "loss": 0.4343,
      "step": 47686
    },
    {
      "epoch": 2.6703438234964723,
      "grad_norm": 1.4640392065048218,
      "learning_rate": 9.271921052631579e-05,
      "loss": 0.3897,
      "step": 47687
    },
    {
      "epoch": 2.6703998208086013,
      "grad_norm": 2.0532164573669434,
      "learning_rate": 9.271894736842106e-05,
      "loss": 0.4406,
      "step": 47688
    },
    {
      "epoch": 2.6704558181207303,
      "grad_norm": 1.177236795425415,
      "learning_rate": 9.271868421052632e-05,
      "loss": 0.4548,
      "step": 47689
    },
    {
      "epoch": 2.6705118154328593,
      "grad_norm": 1.1050159931182861,
      "learning_rate": 9.271842105263159e-05,
      "loss": 0.3929,
      "step": 47690
    },
    {
      "epoch": 2.6705678127449883,
      "grad_norm": 1.2016139030456543,
      "learning_rate": 9.271815789473685e-05,
      "loss": 0.3876,
      "step": 47691
    },
    {
      "epoch": 2.6706238100571174,
      "grad_norm": 1.090924859046936,
      "learning_rate": 9.271789473684211e-05,
      "loss": 0.3816,
      "step": 47692
    },
    {
      "epoch": 2.6706798073692464,
      "grad_norm": 1.4255425930023193,
      "learning_rate": 9.271763157894737e-05,
      "loss": 0.5017,
      "step": 47693
    },
    {
      "epoch": 2.6707358046813754,
      "grad_norm": 1.1507585048675537,
      "learning_rate": 9.271736842105263e-05,
      "loss": 0.3509,
      "step": 47694
    },
    {
      "epoch": 2.6707918019935044,
      "grad_norm": 0.9805853962898254,
      "learning_rate": 9.27171052631579e-05,
      "loss": 0.2738,
      "step": 47695
    },
    {
      "epoch": 2.6708477993056334,
      "grad_norm": 1.2460591793060303,
      "learning_rate": 9.271684210526316e-05,
      "loss": 0.505,
      "step": 47696
    },
    {
      "epoch": 2.6709037966177624,
      "grad_norm": 1.3697493076324463,
      "learning_rate": 9.271657894736842e-05,
      "loss": 0.5751,
      "step": 47697
    },
    {
      "epoch": 2.6709597939298915,
      "grad_norm": 1.0968867540359497,
      "learning_rate": 9.271631578947368e-05,
      "loss": 0.4676,
      "step": 47698
    },
    {
      "epoch": 2.6710157912420205,
      "grad_norm": 1.3214744329452515,
      "learning_rate": 9.271605263157896e-05,
      "loss": 0.5485,
      "step": 47699
    },
    {
      "epoch": 2.6710717885541495,
      "grad_norm": 1.2098684310913086,
      "learning_rate": 9.271578947368422e-05,
      "loss": 0.5283,
      "step": 47700
    },
    {
      "epoch": 2.6711277858662785,
      "grad_norm": 1.2724970579147339,
      "learning_rate": 9.271552631578948e-05,
      "loss": 0.5644,
      "step": 47701
    },
    {
      "epoch": 2.6711837831784075,
      "grad_norm": 1.4909911155700684,
      "learning_rate": 9.271526315789474e-05,
      "loss": 0.381,
      "step": 47702
    },
    {
      "epoch": 2.6712397804905366,
      "grad_norm": 0.9529215097427368,
      "learning_rate": 9.271500000000001e-05,
      "loss": 0.3875,
      "step": 47703
    },
    {
      "epoch": 2.6712957778026656,
      "grad_norm": 1.017738699913025,
      "learning_rate": 9.271473684210527e-05,
      "loss": 0.3505,
      "step": 47704
    },
    {
      "epoch": 2.6713517751147946,
      "grad_norm": 1.2561829090118408,
      "learning_rate": 9.271447368421053e-05,
      "loss": 0.5288,
      "step": 47705
    },
    {
      "epoch": 2.6714077724269236,
      "grad_norm": 0.9466624855995178,
      "learning_rate": 9.271421052631579e-05,
      "loss": 0.3294,
      "step": 47706
    },
    {
      "epoch": 2.6714637697390526,
      "grad_norm": 1.1562076807022095,
      "learning_rate": 9.271394736842106e-05,
      "loss": 0.3638,
      "step": 47707
    },
    {
      "epoch": 2.6715197670511817,
      "grad_norm": 1.2548937797546387,
      "learning_rate": 9.271368421052632e-05,
      "loss": 0.4375,
      "step": 47708
    },
    {
      "epoch": 2.6715757643633107,
      "grad_norm": 1.2531824111938477,
      "learning_rate": 9.27134210526316e-05,
      "loss": 0.3676,
      "step": 47709
    },
    {
      "epoch": 2.6716317616754397,
      "grad_norm": 1.2538422346115112,
      "learning_rate": 9.271315789473684e-05,
      "loss": 0.4406,
      "step": 47710
    },
    {
      "epoch": 2.6716877589875687,
      "grad_norm": 2.7898976802825928,
      "learning_rate": 9.271289473684211e-05,
      "loss": 0.3839,
      "step": 47711
    },
    {
      "epoch": 2.6717437562996977,
      "grad_norm": 1.247151255607605,
      "learning_rate": 9.271263157894737e-05,
      "loss": 0.3611,
      "step": 47712
    },
    {
      "epoch": 2.6717997536118268,
      "grad_norm": 1.3634530305862427,
      "learning_rate": 9.271236842105263e-05,
      "loss": 0.6133,
      "step": 47713
    },
    {
      "epoch": 2.6718557509239558,
      "grad_norm": 1.2633709907531738,
      "learning_rate": 9.27121052631579e-05,
      "loss": 0.3984,
      "step": 47714
    },
    {
      "epoch": 2.671911748236085,
      "grad_norm": 1.0147438049316406,
      "learning_rate": 9.271184210526315e-05,
      "loss": 0.3739,
      "step": 47715
    },
    {
      "epoch": 2.671967745548214,
      "grad_norm": 1.1683586835861206,
      "learning_rate": 9.271157894736843e-05,
      "loss": 0.3168,
      "step": 47716
    },
    {
      "epoch": 2.672023742860343,
      "grad_norm": 1.554422378540039,
      "learning_rate": 9.271131578947369e-05,
      "loss": 0.4177,
      "step": 47717
    },
    {
      "epoch": 2.672079740172472,
      "grad_norm": 1.0523382425308228,
      "learning_rate": 9.271105263157896e-05,
      "loss": 0.3649,
      "step": 47718
    },
    {
      "epoch": 2.672135737484601,
      "grad_norm": 1.0413740873336792,
      "learning_rate": 9.27107894736842e-05,
      "loss": 0.3302,
      "step": 47719
    },
    {
      "epoch": 2.67219173479673,
      "grad_norm": 1.0472010374069214,
      "learning_rate": 9.271052631578948e-05,
      "loss": 0.4117,
      "step": 47720
    },
    {
      "epoch": 2.672247732108859,
      "grad_norm": 1.0614008903503418,
      "learning_rate": 9.271026315789474e-05,
      "loss": 0.3228,
      "step": 47721
    },
    {
      "epoch": 2.672303729420988,
      "grad_norm": 1.2798857688903809,
      "learning_rate": 9.271000000000001e-05,
      "loss": 0.47,
      "step": 47722
    },
    {
      "epoch": 2.672359726733117,
      "grad_norm": 1.3511453866958618,
      "learning_rate": 9.270973684210527e-05,
      "loss": 0.509,
      "step": 47723
    },
    {
      "epoch": 2.672415724045246,
      "grad_norm": 1.0842405557632446,
      "learning_rate": 9.270947368421053e-05,
      "loss": 0.3972,
      "step": 47724
    },
    {
      "epoch": 2.672471721357375,
      "grad_norm": 1.1405714750289917,
      "learning_rate": 9.270921052631579e-05,
      "loss": 0.4452,
      "step": 47725
    },
    {
      "epoch": 2.672527718669504,
      "grad_norm": 1.2329267263412476,
      "learning_rate": 9.270894736842106e-05,
      "loss": 0.3298,
      "step": 47726
    },
    {
      "epoch": 2.672583715981633,
      "grad_norm": 1.4733500480651855,
      "learning_rate": 9.270868421052632e-05,
      "loss": 0.4643,
      "step": 47727
    },
    {
      "epoch": 2.672639713293762,
      "grad_norm": 1.2314311265945435,
      "learning_rate": 9.270842105263158e-05,
      "loss": 0.4534,
      "step": 47728
    },
    {
      "epoch": 2.672695710605891,
      "grad_norm": 1.321105718612671,
      "learning_rate": 9.270815789473684e-05,
      "loss": 0.3554,
      "step": 47729
    },
    {
      "epoch": 2.67275170791802,
      "grad_norm": 1.2114466428756714,
      "learning_rate": 9.27078947368421e-05,
      "loss": 0.2903,
      "step": 47730
    },
    {
      "epoch": 2.672807705230149,
      "grad_norm": 1.116355538368225,
      "learning_rate": 9.270763157894738e-05,
      "loss": 0.3899,
      "step": 47731
    },
    {
      "epoch": 2.672863702542278,
      "grad_norm": 0.9207002520561218,
      "learning_rate": 9.270736842105264e-05,
      "loss": 0.2722,
      "step": 47732
    },
    {
      "epoch": 2.672919699854407,
      "grad_norm": 2.3024168014526367,
      "learning_rate": 9.27071052631579e-05,
      "loss": 0.3817,
      "step": 47733
    },
    {
      "epoch": 2.672975697166536,
      "grad_norm": 1.370310664176941,
      "learning_rate": 9.270684210526316e-05,
      "loss": 0.3584,
      "step": 47734
    },
    {
      "epoch": 2.673031694478665,
      "grad_norm": 1.244425654411316,
      "learning_rate": 9.270657894736843e-05,
      "loss": 0.427,
      "step": 47735
    },
    {
      "epoch": 2.673087691790794,
      "grad_norm": 1.5430843830108643,
      "learning_rate": 9.270631578947369e-05,
      "loss": 0.3776,
      "step": 47736
    },
    {
      "epoch": 2.673143689102923,
      "grad_norm": 1.0521957874298096,
      "learning_rate": 9.270605263157895e-05,
      "loss": 0.3554,
      "step": 47737
    },
    {
      "epoch": 2.6731996864150522,
      "grad_norm": 1.0691338777542114,
      "learning_rate": 9.270578947368421e-05,
      "loss": 0.2817,
      "step": 47738
    },
    {
      "epoch": 2.6732556837271813,
      "grad_norm": 1.0008784532546997,
      "learning_rate": 9.270552631578948e-05,
      "loss": 0.2828,
      "step": 47739
    },
    {
      "epoch": 2.6733116810393103,
      "grad_norm": 1.1579856872558594,
      "learning_rate": 9.270526315789474e-05,
      "loss": 0.4353,
      "step": 47740
    },
    {
      "epoch": 2.6733676783514393,
      "grad_norm": 0.9530447125434875,
      "learning_rate": 9.270500000000001e-05,
      "loss": 0.3462,
      "step": 47741
    },
    {
      "epoch": 2.6734236756635683,
      "grad_norm": 1.923862338066101,
      "learning_rate": 9.270473684210526e-05,
      "loss": 0.4766,
      "step": 47742
    },
    {
      "epoch": 2.6734796729756973,
      "grad_norm": 1.365543246269226,
      "learning_rate": 9.270447368421053e-05,
      "loss": 0.4353,
      "step": 47743
    },
    {
      "epoch": 2.6735356702878263,
      "grad_norm": 1.2928060293197632,
      "learning_rate": 9.27042105263158e-05,
      "loss": 0.5077,
      "step": 47744
    },
    {
      "epoch": 2.6735916675999554,
      "grad_norm": 1.5102511644363403,
      "learning_rate": 9.270394736842107e-05,
      "loss": 0.4113,
      "step": 47745
    },
    {
      "epoch": 2.6736476649120844,
      "grad_norm": 28.981225967407227,
      "learning_rate": 9.270368421052633e-05,
      "loss": 0.2987,
      "step": 47746
    },
    {
      "epoch": 2.673703662224213,
      "grad_norm": 0.9775345325469971,
      "learning_rate": 9.270342105263157e-05,
      "loss": 0.3122,
      "step": 47747
    },
    {
      "epoch": 2.6737596595363424,
      "grad_norm": 1.050163984298706,
      "learning_rate": 9.270315789473685e-05,
      "loss": 0.3286,
      "step": 47748
    },
    {
      "epoch": 2.673815656848471,
      "grad_norm": 1.158350944519043,
      "learning_rate": 9.27028947368421e-05,
      "loss": 0.4127,
      "step": 47749
    },
    {
      "epoch": 2.6738716541606005,
      "grad_norm": 1.257048487663269,
      "learning_rate": 9.270263157894738e-05,
      "loss": 0.3862,
      "step": 47750
    },
    {
      "epoch": 2.673927651472729,
      "grad_norm": 1.4238168001174927,
      "learning_rate": 9.270236842105264e-05,
      "loss": 0.4913,
      "step": 47751
    },
    {
      "epoch": 2.6739836487848585,
      "grad_norm": 1.4065065383911133,
      "learning_rate": 9.27021052631579e-05,
      "loss": 0.4144,
      "step": 47752
    },
    {
      "epoch": 2.674039646096987,
      "grad_norm": 1.1196796894073486,
      "learning_rate": 9.270184210526316e-05,
      "loss": 0.4319,
      "step": 47753
    },
    {
      "epoch": 2.6740956434091165,
      "grad_norm": 1.1342880725860596,
      "learning_rate": 9.270157894736843e-05,
      "loss": 0.363,
      "step": 47754
    },
    {
      "epoch": 2.674151640721245,
      "grad_norm": 1.2477728128433228,
      "learning_rate": 9.270131578947369e-05,
      "loss": 0.4862,
      "step": 47755
    },
    {
      "epoch": 2.6742076380333746,
      "grad_norm": 1.174963116645813,
      "learning_rate": 9.270105263157895e-05,
      "loss": 0.524,
      "step": 47756
    },
    {
      "epoch": 2.674263635345503,
      "grad_norm": 1.1364448070526123,
      "learning_rate": 9.270078947368421e-05,
      "loss": 0.3597,
      "step": 47757
    },
    {
      "epoch": 2.6743196326576326,
      "grad_norm": 1.0814162492752075,
      "learning_rate": 9.270052631578948e-05,
      "loss": 0.3927,
      "step": 47758
    },
    {
      "epoch": 2.674375629969761,
      "grad_norm": 1.194346308708191,
      "learning_rate": 9.270026315789474e-05,
      "loss": 0.409,
      "step": 47759
    },
    {
      "epoch": 2.6744316272818907,
      "grad_norm": 1.2346943616867065,
      "learning_rate": 9.27e-05,
      "loss": 0.3805,
      "step": 47760
    },
    {
      "epoch": 2.6744876245940192,
      "grad_norm": 1.2514721155166626,
      "learning_rate": 9.269973684210526e-05,
      "loss": 0.581,
      "step": 47761
    },
    {
      "epoch": 2.6745436219061487,
      "grad_norm": 1.1615569591522217,
      "learning_rate": 9.269947368421054e-05,
      "loss": 0.3865,
      "step": 47762
    },
    {
      "epoch": 2.6745996192182773,
      "grad_norm": 1.43671715259552,
      "learning_rate": 9.26992105263158e-05,
      "loss": 0.3051,
      "step": 47763
    },
    {
      "epoch": 2.6746556165304067,
      "grad_norm": 1.1561893224716187,
      "learning_rate": 9.269894736842106e-05,
      "loss": 0.433,
      "step": 47764
    },
    {
      "epoch": 2.6747116138425353,
      "grad_norm": 1.2000234127044678,
      "learning_rate": 9.269868421052632e-05,
      "loss": 0.3716,
      "step": 47765
    },
    {
      "epoch": 2.6747676111546648,
      "grad_norm": 1.0337213277816772,
      "learning_rate": 9.269842105263158e-05,
      "loss": 0.3815,
      "step": 47766
    },
    {
      "epoch": 2.6748236084667933,
      "grad_norm": 1.2272530794143677,
      "learning_rate": 9.269815789473685e-05,
      "loss": 0.5235,
      "step": 47767
    },
    {
      "epoch": 2.674879605778923,
      "grad_norm": 1.3122800588607788,
      "learning_rate": 9.269789473684211e-05,
      "loss": 0.3988,
      "step": 47768
    },
    {
      "epoch": 2.6749356030910514,
      "grad_norm": 1.0575898885726929,
      "learning_rate": 9.269763157894737e-05,
      "loss": 0.3546,
      "step": 47769
    },
    {
      "epoch": 2.674991600403181,
      "grad_norm": 1.240477442741394,
      "learning_rate": 9.269736842105263e-05,
      "loss": 0.3454,
      "step": 47770
    },
    {
      "epoch": 2.6750475977153094,
      "grad_norm": 1.0041407346725464,
      "learning_rate": 9.26971052631579e-05,
      "loss": 0.2848,
      "step": 47771
    },
    {
      "epoch": 2.675103595027439,
      "grad_norm": 1.426337718963623,
      "learning_rate": 9.269684210526316e-05,
      "loss": 0.4202,
      "step": 47772
    },
    {
      "epoch": 2.6751595923395675,
      "grad_norm": 1.2755252122879028,
      "learning_rate": 9.269657894736843e-05,
      "loss": 0.3307,
      "step": 47773
    },
    {
      "epoch": 2.675215589651697,
      "grad_norm": 1.0364798307418823,
      "learning_rate": 9.269631578947368e-05,
      "loss": 0.5004,
      "step": 47774
    },
    {
      "epoch": 2.6752715869638255,
      "grad_norm": 1.1622111797332764,
      "learning_rate": 9.269605263157895e-05,
      "loss": 0.3771,
      "step": 47775
    },
    {
      "epoch": 2.675327584275955,
      "grad_norm": 1.380347728729248,
      "learning_rate": 9.269578947368421e-05,
      "loss": 0.4674,
      "step": 47776
    },
    {
      "epoch": 2.6753835815880835,
      "grad_norm": 1.1976853609085083,
      "learning_rate": 9.269552631578949e-05,
      "loss": 0.3967,
      "step": 47777
    },
    {
      "epoch": 2.675439578900213,
      "grad_norm": 1.1495815515518188,
      "learning_rate": 9.269526315789475e-05,
      "loss": 0.429,
      "step": 47778
    },
    {
      "epoch": 2.6754955762123416,
      "grad_norm": 1.116594672203064,
      "learning_rate": 9.2695e-05,
      "loss": 0.385,
      "step": 47779
    },
    {
      "epoch": 2.675551573524471,
      "grad_norm": 0.999987781047821,
      "learning_rate": 9.269473684210527e-05,
      "loss": 0.3361,
      "step": 47780
    },
    {
      "epoch": 2.6756075708365996,
      "grad_norm": 1.139424443244934,
      "learning_rate": 9.269447368421053e-05,
      "loss": 0.4051,
      "step": 47781
    },
    {
      "epoch": 2.675663568148729,
      "grad_norm": 1.2612146139144897,
      "learning_rate": 9.26942105263158e-05,
      "loss": 0.4113,
      "step": 47782
    },
    {
      "epoch": 2.6757195654608577,
      "grad_norm": 1.6711788177490234,
      "learning_rate": 9.269394736842106e-05,
      "loss": 0.4349,
      "step": 47783
    },
    {
      "epoch": 2.675775562772987,
      "grad_norm": 1.3835583925247192,
      "learning_rate": 9.269368421052632e-05,
      "loss": 0.4294,
      "step": 47784
    },
    {
      "epoch": 2.6758315600851157,
      "grad_norm": 1.3838880062103271,
      "learning_rate": 9.269342105263158e-05,
      "loss": 0.3625,
      "step": 47785
    },
    {
      "epoch": 2.675887557397245,
      "grad_norm": 1.1828957796096802,
      "learning_rate": 9.269315789473685e-05,
      "loss": 0.4543,
      "step": 47786
    },
    {
      "epoch": 2.6759435547093737,
      "grad_norm": 1.4447481632232666,
      "learning_rate": 9.269289473684211e-05,
      "loss": 0.3617,
      "step": 47787
    },
    {
      "epoch": 2.675999552021503,
      "grad_norm": 1.4834697246551514,
      "learning_rate": 9.269263157894737e-05,
      "loss": 0.457,
      "step": 47788
    },
    {
      "epoch": 2.6760555493336318,
      "grad_norm": 1.3348218202590942,
      "learning_rate": 9.269236842105263e-05,
      "loss": 0.363,
      "step": 47789
    },
    {
      "epoch": 2.6761115466457612,
      "grad_norm": 1.2873613834381104,
      "learning_rate": 9.26921052631579e-05,
      "loss": 0.4467,
      "step": 47790
    },
    {
      "epoch": 2.67616754395789,
      "grad_norm": 1.587181568145752,
      "learning_rate": 9.269184210526316e-05,
      "loss": 0.4814,
      "step": 47791
    },
    {
      "epoch": 2.6762235412700193,
      "grad_norm": 0.9964888095855713,
      "learning_rate": 9.269157894736842e-05,
      "loss": 0.4351,
      "step": 47792
    },
    {
      "epoch": 2.676279538582148,
      "grad_norm": 1.1419146060943604,
      "learning_rate": 9.269131578947368e-05,
      "loss": 0.3941,
      "step": 47793
    },
    {
      "epoch": 2.6763355358942773,
      "grad_norm": 2.001640796661377,
      "learning_rate": 9.269105263157896e-05,
      "loss": 0.3904,
      "step": 47794
    },
    {
      "epoch": 2.676391533206406,
      "grad_norm": 1.2208929061889648,
      "learning_rate": 9.269078947368422e-05,
      "loss": 0.3498,
      "step": 47795
    },
    {
      "epoch": 2.6764475305185353,
      "grad_norm": 1.1475390195846558,
      "learning_rate": 9.269052631578949e-05,
      "loss": 0.3543,
      "step": 47796
    },
    {
      "epoch": 2.676503527830664,
      "grad_norm": 1.1290451288223267,
      "learning_rate": 9.269026315789474e-05,
      "loss": 0.3907,
      "step": 47797
    },
    {
      "epoch": 2.6765595251427934,
      "grad_norm": 2.4756264686584473,
      "learning_rate": 9.269e-05,
      "loss": 0.4052,
      "step": 47798
    },
    {
      "epoch": 2.676615522454922,
      "grad_norm": 1.2377738952636719,
      "learning_rate": 9.268973684210527e-05,
      "loss": 0.3952,
      "step": 47799
    },
    {
      "epoch": 2.6766715197670514,
      "grad_norm": 1.3301308155059814,
      "learning_rate": 9.268947368421053e-05,
      "loss": 0.4608,
      "step": 47800
    },
    {
      "epoch": 2.67672751707918,
      "grad_norm": 1.5669701099395752,
      "learning_rate": 9.26892105263158e-05,
      "loss": 0.383,
      "step": 47801
    },
    {
      "epoch": 2.6767835143913095,
      "grad_norm": 1.0153595209121704,
      "learning_rate": 9.268894736842105e-05,
      "loss": 0.3226,
      "step": 47802
    },
    {
      "epoch": 2.676839511703438,
      "grad_norm": 1.3522647619247437,
      "learning_rate": 9.268868421052632e-05,
      "loss": 0.5446,
      "step": 47803
    },
    {
      "epoch": 2.6768955090155675,
      "grad_norm": 1.0125373601913452,
      "learning_rate": 9.268842105263158e-05,
      "loss": 0.2999,
      "step": 47804
    },
    {
      "epoch": 2.676951506327696,
      "grad_norm": 1.5722360610961914,
      "learning_rate": 9.268815789473685e-05,
      "loss": 0.5023,
      "step": 47805
    },
    {
      "epoch": 2.6770075036398255,
      "grad_norm": 1.3264153003692627,
      "learning_rate": 9.268789473684211e-05,
      "loss": 0.4459,
      "step": 47806
    },
    {
      "epoch": 2.677063500951954,
      "grad_norm": 1.2622379064559937,
      "learning_rate": 9.268763157894737e-05,
      "loss": 0.4154,
      "step": 47807
    },
    {
      "epoch": 2.6771194982640836,
      "grad_norm": 1.1937501430511475,
      "learning_rate": 9.268736842105263e-05,
      "loss": 0.3245,
      "step": 47808
    },
    {
      "epoch": 2.677175495576212,
      "grad_norm": 1.1173481941223145,
      "learning_rate": 9.26871052631579e-05,
      "loss": 0.5287,
      "step": 47809
    },
    {
      "epoch": 2.6772314928883416,
      "grad_norm": 1.136512041091919,
      "learning_rate": 9.268684210526317e-05,
      "loss": 0.3518,
      "step": 47810
    },
    {
      "epoch": 2.67728749020047,
      "grad_norm": 1.8841850757598877,
      "learning_rate": 9.268657894736843e-05,
      "loss": 0.4851,
      "step": 47811
    },
    {
      "epoch": 2.6773434875125997,
      "grad_norm": 1.0801796913146973,
      "learning_rate": 9.268631578947369e-05,
      "loss": 0.343,
      "step": 47812
    },
    {
      "epoch": 2.6773994848247282,
      "grad_norm": 1.3997802734375,
      "learning_rate": 9.268605263157896e-05,
      "loss": 0.3583,
      "step": 47813
    },
    {
      "epoch": 2.6774554821368577,
      "grad_norm": 1.2265769243240356,
      "learning_rate": 9.268578947368422e-05,
      "loss": 0.3904,
      "step": 47814
    },
    {
      "epoch": 2.6775114794489863,
      "grad_norm": 1.3216334581375122,
      "learning_rate": 9.268552631578948e-05,
      "loss": 0.4882,
      "step": 47815
    },
    {
      "epoch": 2.6775674767611157,
      "grad_norm": 1.2003467082977295,
      "learning_rate": 9.268526315789474e-05,
      "loss": 0.3937,
      "step": 47816
    },
    {
      "epoch": 2.6776234740732443,
      "grad_norm": 1.140363097190857,
      "learning_rate": 9.2685e-05,
      "loss": 0.3342,
      "step": 47817
    },
    {
      "epoch": 2.6776794713853738,
      "grad_norm": 1.2253869771957397,
      "learning_rate": 9.268473684210527e-05,
      "loss": 0.4294,
      "step": 47818
    },
    {
      "epoch": 2.6777354686975023,
      "grad_norm": 1.280599594116211,
      "learning_rate": 9.268447368421053e-05,
      "loss": 0.4078,
      "step": 47819
    },
    {
      "epoch": 2.677791466009632,
      "grad_norm": 1.1165099143981934,
      "learning_rate": 9.268421052631579e-05,
      "loss": 0.3207,
      "step": 47820
    },
    {
      "epoch": 2.6778474633217604,
      "grad_norm": 1.1123110055923462,
      "learning_rate": 9.268394736842105e-05,
      "loss": 0.4881,
      "step": 47821
    },
    {
      "epoch": 2.67790346063389,
      "grad_norm": 1.076734185218811,
      "learning_rate": 9.268368421052632e-05,
      "loss": 0.4382,
      "step": 47822
    },
    {
      "epoch": 2.6779594579460184,
      "grad_norm": 1.2299330234527588,
      "learning_rate": 9.268342105263158e-05,
      "loss": 0.3107,
      "step": 47823
    },
    {
      "epoch": 2.678015455258148,
      "grad_norm": 1.223860263824463,
      "learning_rate": 9.268315789473684e-05,
      "loss": 0.4295,
      "step": 47824
    },
    {
      "epoch": 2.6780714525702765,
      "grad_norm": 1.0006104707717896,
      "learning_rate": 9.26828947368421e-05,
      "loss": 0.3384,
      "step": 47825
    },
    {
      "epoch": 2.678127449882406,
      "grad_norm": 1.1087127923965454,
      "learning_rate": 9.268263157894738e-05,
      "loss": 0.3521,
      "step": 47826
    },
    {
      "epoch": 2.6781834471945345,
      "grad_norm": 1.0691715478897095,
      "learning_rate": 9.268236842105264e-05,
      "loss": 0.3602,
      "step": 47827
    },
    {
      "epoch": 2.678239444506664,
      "grad_norm": 0.965553879737854,
      "learning_rate": 9.268210526315791e-05,
      "loss": 0.3315,
      "step": 47828
    },
    {
      "epoch": 2.6782954418187925,
      "grad_norm": 1.1629024744033813,
      "learning_rate": 9.268184210526316e-05,
      "loss": 0.3639,
      "step": 47829
    },
    {
      "epoch": 2.6783514391309216,
      "grad_norm": 1.5215692520141602,
      "learning_rate": 9.268157894736843e-05,
      "loss": 0.4933,
      "step": 47830
    },
    {
      "epoch": 2.6784074364430506,
      "grad_norm": 1.2850102186203003,
      "learning_rate": 9.268131578947369e-05,
      "loss": 0.3993,
      "step": 47831
    },
    {
      "epoch": 2.6784634337551796,
      "grad_norm": 1.3956345319747925,
      "learning_rate": 9.268105263157896e-05,
      "loss": 0.4413,
      "step": 47832
    },
    {
      "epoch": 2.6785194310673086,
      "grad_norm": 1.1221729516983032,
      "learning_rate": 9.268078947368422e-05,
      "loss": 0.334,
      "step": 47833
    },
    {
      "epoch": 2.6785754283794376,
      "grad_norm": 1.2323111295700073,
      "learning_rate": 9.268052631578947e-05,
      "loss": 0.4356,
      "step": 47834
    },
    {
      "epoch": 2.6786314256915666,
      "grad_norm": 1.4334977865219116,
      "learning_rate": 9.268026315789474e-05,
      "loss": 0.456,
      "step": 47835
    },
    {
      "epoch": 2.6786874230036957,
      "grad_norm": 1.2918404340744019,
      "learning_rate": 9.268e-05,
      "loss": 0.4275,
      "step": 47836
    },
    {
      "epoch": 2.6787434203158247,
      "grad_norm": 1.4713538885116577,
      "learning_rate": 9.267973684210527e-05,
      "loss": 0.5327,
      "step": 47837
    },
    {
      "epoch": 2.6787994176279537,
      "grad_norm": 1.343296766281128,
      "learning_rate": 9.267947368421053e-05,
      "loss": 0.5149,
      "step": 47838
    },
    {
      "epoch": 2.6788554149400827,
      "grad_norm": 1.1042585372924805,
      "learning_rate": 9.267921052631579e-05,
      "loss": 0.4674,
      "step": 47839
    },
    {
      "epoch": 2.6789114122522117,
      "grad_norm": 1.247188687324524,
      "learning_rate": 9.267894736842105e-05,
      "loss": 0.349,
      "step": 47840
    },
    {
      "epoch": 2.6789674095643408,
      "grad_norm": 1.980407953262329,
      "learning_rate": 9.267868421052633e-05,
      "loss": 0.3829,
      "step": 47841
    },
    {
      "epoch": 2.67902340687647,
      "grad_norm": 1.2237497568130493,
      "learning_rate": 9.267842105263159e-05,
      "loss": 0.5391,
      "step": 47842
    },
    {
      "epoch": 2.679079404188599,
      "grad_norm": 1.0744506120681763,
      "learning_rate": 9.267815789473685e-05,
      "loss": 0.458,
      "step": 47843
    },
    {
      "epoch": 2.679135401500728,
      "grad_norm": 1.7879174947738647,
      "learning_rate": 9.26778947368421e-05,
      "loss": 0.4406,
      "step": 47844
    },
    {
      "epoch": 2.679191398812857,
      "grad_norm": 1.2716103792190552,
      "learning_rate": 9.267763157894738e-05,
      "loss": 0.3587,
      "step": 47845
    },
    {
      "epoch": 2.679247396124986,
      "grad_norm": 1.2254962921142578,
      "learning_rate": 9.267736842105264e-05,
      "loss": 0.4714,
      "step": 47846
    },
    {
      "epoch": 2.679303393437115,
      "grad_norm": 1.1475086212158203,
      "learning_rate": 9.26771052631579e-05,
      "loss": 0.498,
      "step": 47847
    },
    {
      "epoch": 2.679359390749244,
      "grad_norm": 1.4059903621673584,
      "learning_rate": 9.267684210526316e-05,
      "loss": 0.3839,
      "step": 47848
    },
    {
      "epoch": 2.679415388061373,
      "grad_norm": 1.29823637008667,
      "learning_rate": 9.267657894736843e-05,
      "loss": 0.3926,
      "step": 47849
    },
    {
      "epoch": 2.679471385373502,
      "grad_norm": 1.236661434173584,
      "learning_rate": 9.267631578947369e-05,
      "loss": 0.3623,
      "step": 47850
    },
    {
      "epoch": 2.679527382685631,
      "grad_norm": 1.1398160457611084,
      "learning_rate": 9.267605263157895e-05,
      "loss": 0.2873,
      "step": 47851
    },
    {
      "epoch": 2.67958337999776,
      "grad_norm": 1.5209718942642212,
      "learning_rate": 9.267578947368421e-05,
      "loss": 0.4466,
      "step": 47852
    },
    {
      "epoch": 2.679639377309889,
      "grad_norm": 1.498822569847107,
      "learning_rate": 9.267552631578947e-05,
      "loss": 0.433,
      "step": 47853
    },
    {
      "epoch": 2.679695374622018,
      "grad_norm": 1.8030474185943604,
      "learning_rate": 9.267526315789474e-05,
      "loss": 0.4509,
      "step": 47854
    },
    {
      "epoch": 2.679751371934147,
      "grad_norm": 1.1488314867019653,
      "learning_rate": 9.2675e-05,
      "loss": 0.3625,
      "step": 47855
    },
    {
      "epoch": 2.679807369246276,
      "grad_norm": 1.2726997137069702,
      "learning_rate": 9.267473684210528e-05,
      "loss": 0.4461,
      "step": 47856
    },
    {
      "epoch": 2.679863366558405,
      "grad_norm": 1.2456368207931519,
      "learning_rate": 9.267447368421052e-05,
      "loss": 0.5879,
      "step": 47857
    },
    {
      "epoch": 2.679919363870534,
      "grad_norm": 1.3350324630737305,
      "learning_rate": 9.26742105263158e-05,
      "loss": 0.4616,
      "step": 47858
    },
    {
      "epoch": 2.679975361182663,
      "grad_norm": 1.104844093322754,
      "learning_rate": 9.267394736842106e-05,
      "loss": 0.3642,
      "step": 47859
    },
    {
      "epoch": 2.680031358494792,
      "grad_norm": 1.3494511842727661,
      "learning_rate": 9.267368421052633e-05,
      "loss": 0.4125,
      "step": 47860
    },
    {
      "epoch": 2.680087355806921,
      "grad_norm": 1.1848931312561035,
      "learning_rate": 9.267342105263157e-05,
      "loss": 0.4639,
      "step": 47861
    },
    {
      "epoch": 2.68014335311905,
      "grad_norm": 1.1642677783966064,
      "learning_rate": 9.267315789473685e-05,
      "loss": 0.3487,
      "step": 47862
    },
    {
      "epoch": 2.680199350431179,
      "grad_norm": 1.3135396242141724,
      "learning_rate": 9.267289473684211e-05,
      "loss": 0.4558,
      "step": 47863
    },
    {
      "epoch": 2.680255347743308,
      "grad_norm": 1.04069185256958,
      "learning_rate": 9.267263157894738e-05,
      "loss": 0.3909,
      "step": 47864
    },
    {
      "epoch": 2.6803113450554372,
      "grad_norm": 1.0357638597488403,
      "learning_rate": 9.267236842105264e-05,
      "loss": 0.3585,
      "step": 47865
    },
    {
      "epoch": 2.6803673423675662,
      "grad_norm": 1.0681413412094116,
      "learning_rate": 9.26721052631579e-05,
      "loss": 0.3252,
      "step": 47866
    },
    {
      "epoch": 2.6804233396796953,
      "grad_norm": 1.5507864952087402,
      "learning_rate": 9.267184210526316e-05,
      "loss": 0.4486,
      "step": 47867
    },
    {
      "epoch": 2.6804793369918243,
      "grad_norm": 1.301295280456543,
      "learning_rate": 9.267157894736842e-05,
      "loss": 0.3784,
      "step": 47868
    },
    {
      "epoch": 2.6805353343039533,
      "grad_norm": 1.8679041862487793,
      "learning_rate": 9.26713157894737e-05,
      "loss": 0.5409,
      "step": 47869
    },
    {
      "epoch": 2.6805913316160823,
      "grad_norm": 1.0675432682037354,
      "learning_rate": 9.267105263157895e-05,
      "loss": 0.3748,
      "step": 47870
    },
    {
      "epoch": 2.6806473289282113,
      "grad_norm": 1.2178475856781006,
      "learning_rate": 9.267078947368421e-05,
      "loss": 0.4539,
      "step": 47871
    },
    {
      "epoch": 2.6807033262403404,
      "grad_norm": 1.1309034824371338,
      "learning_rate": 9.267052631578947e-05,
      "loss": 0.4314,
      "step": 47872
    },
    {
      "epoch": 2.6807593235524694,
      "grad_norm": 1.1379342079162598,
      "learning_rate": 9.267026315789475e-05,
      "loss": 0.3011,
      "step": 47873
    },
    {
      "epoch": 2.6808153208645984,
      "grad_norm": 1.3592448234558105,
      "learning_rate": 9.267e-05,
      "loss": 0.4765,
      "step": 47874
    },
    {
      "epoch": 2.6808713181767274,
      "grad_norm": 1.4272902011871338,
      "learning_rate": 9.266973684210527e-05,
      "loss": 0.4197,
      "step": 47875
    },
    {
      "epoch": 2.6809273154888564,
      "grad_norm": 1.0565543174743652,
      "learning_rate": 9.266947368421052e-05,
      "loss": 0.4647,
      "step": 47876
    },
    {
      "epoch": 2.6809833128009855,
      "grad_norm": 1.3128750324249268,
      "learning_rate": 9.26692105263158e-05,
      "loss": 0.4921,
      "step": 47877
    },
    {
      "epoch": 2.6810393101131145,
      "grad_norm": 1.1479016542434692,
      "learning_rate": 9.266894736842106e-05,
      "loss": 0.341,
      "step": 47878
    },
    {
      "epoch": 2.6810953074252435,
      "grad_norm": 1.0815821886062622,
      "learning_rate": 9.266868421052632e-05,
      "loss": 0.4417,
      "step": 47879
    },
    {
      "epoch": 2.6811513047373725,
      "grad_norm": 1.3256213665008545,
      "learning_rate": 9.266842105263158e-05,
      "loss": 0.3821,
      "step": 47880
    },
    {
      "epoch": 2.6812073020495015,
      "grad_norm": 1.0697332620620728,
      "learning_rate": 9.266815789473685e-05,
      "loss": 0.3436,
      "step": 47881
    },
    {
      "epoch": 2.6812632993616305,
      "grad_norm": 1.280110239982605,
      "learning_rate": 9.266789473684211e-05,
      "loss": 0.4328,
      "step": 47882
    },
    {
      "epoch": 2.6813192966737596,
      "grad_norm": 1.5841667652130127,
      "learning_rate": 9.266763157894738e-05,
      "loss": 0.3989,
      "step": 47883
    },
    {
      "epoch": 2.6813752939858886,
      "grad_norm": 1.0869314670562744,
      "learning_rate": 9.266736842105263e-05,
      "loss": 0.3225,
      "step": 47884
    },
    {
      "epoch": 2.6814312912980176,
      "grad_norm": 1.1324384212493896,
      "learning_rate": 9.266710526315789e-05,
      "loss": 0.3942,
      "step": 47885
    },
    {
      "epoch": 2.6814872886101466,
      "grad_norm": 1.2003108263015747,
      "learning_rate": 9.266684210526316e-05,
      "loss": 0.3851,
      "step": 47886
    },
    {
      "epoch": 2.6815432859222756,
      "grad_norm": 1.059016466140747,
      "learning_rate": 9.266657894736842e-05,
      "loss": 0.4677,
      "step": 47887
    },
    {
      "epoch": 2.6815992832344047,
      "grad_norm": 1.198194146156311,
      "learning_rate": 9.26663157894737e-05,
      "loss": 0.4202,
      "step": 47888
    },
    {
      "epoch": 2.6816552805465337,
      "grad_norm": 1.0208699703216553,
      "learning_rate": 9.266605263157894e-05,
      "loss": 0.3789,
      "step": 47889
    },
    {
      "epoch": 2.6817112778586627,
      "grad_norm": 1.0194646120071411,
      "learning_rate": 9.266578947368422e-05,
      "loss": 0.3313,
      "step": 47890
    },
    {
      "epoch": 2.6817672751707917,
      "grad_norm": 1.2013812065124512,
      "learning_rate": 9.266552631578948e-05,
      "loss": 0.3816,
      "step": 47891
    },
    {
      "epoch": 2.6818232724829207,
      "grad_norm": 1.0947545766830444,
      "learning_rate": 9.266526315789475e-05,
      "loss": 0.3801,
      "step": 47892
    },
    {
      "epoch": 2.6818792697950498,
      "grad_norm": 1.2629832029342651,
      "learning_rate": 9.266500000000001e-05,
      "loss": 0.3579,
      "step": 47893
    },
    {
      "epoch": 2.6819352671071788,
      "grad_norm": 0.9853864908218384,
      "learning_rate": 9.266473684210527e-05,
      "loss": 0.2782,
      "step": 47894
    },
    {
      "epoch": 2.681991264419308,
      "grad_norm": 1.2657511234283447,
      "learning_rate": 9.266447368421053e-05,
      "loss": 0.5076,
      "step": 47895
    },
    {
      "epoch": 2.682047261731437,
      "grad_norm": 1.0073182582855225,
      "learning_rate": 9.26642105263158e-05,
      "loss": 0.3402,
      "step": 47896
    },
    {
      "epoch": 2.682103259043566,
      "grad_norm": 2.022655963897705,
      "learning_rate": 9.266394736842106e-05,
      "loss": 0.5163,
      "step": 47897
    },
    {
      "epoch": 2.682159256355695,
      "grad_norm": 1.2612284421920776,
      "learning_rate": 9.266368421052632e-05,
      "loss": 0.6015,
      "step": 47898
    },
    {
      "epoch": 2.682215253667824,
      "grad_norm": 1.1210434436798096,
      "learning_rate": 9.266342105263158e-05,
      "loss": 0.3964,
      "step": 47899
    },
    {
      "epoch": 2.682271250979953,
      "grad_norm": 1.2173187732696533,
      "learning_rate": 9.266315789473685e-05,
      "loss": 0.44,
      "step": 47900
    },
    {
      "epoch": 2.682327248292082,
      "grad_norm": 1.2398945093154907,
      "learning_rate": 9.266289473684211e-05,
      "loss": 0.3909,
      "step": 47901
    },
    {
      "epoch": 2.682383245604211,
      "grad_norm": 1.4878675937652588,
      "learning_rate": 9.266263157894737e-05,
      "loss": 0.4104,
      "step": 47902
    },
    {
      "epoch": 2.68243924291634,
      "grad_norm": 1.236393690109253,
      "learning_rate": 9.266236842105263e-05,
      "loss": 0.5328,
      "step": 47903
    },
    {
      "epoch": 2.682495240228469,
      "grad_norm": 1.2498973608016968,
      "learning_rate": 9.266210526315789e-05,
      "loss": 0.3093,
      "step": 47904
    },
    {
      "epoch": 2.682551237540598,
      "grad_norm": 1.0993890762329102,
      "learning_rate": 9.266184210526317e-05,
      "loss": 0.3808,
      "step": 47905
    },
    {
      "epoch": 2.682607234852727,
      "grad_norm": 1.2274836301803589,
      "learning_rate": 9.266157894736843e-05,
      "loss": 0.4328,
      "step": 47906
    },
    {
      "epoch": 2.682663232164856,
      "grad_norm": 1.1571154594421387,
      "learning_rate": 9.266131578947368e-05,
      "loss": 0.4065,
      "step": 47907
    },
    {
      "epoch": 2.682719229476985,
      "grad_norm": 1.1565924882888794,
      "learning_rate": 9.266105263157894e-05,
      "loss": 0.5049,
      "step": 47908
    },
    {
      "epoch": 2.682775226789114,
      "grad_norm": 1.1699274778366089,
      "learning_rate": 9.266078947368422e-05,
      "loss": 0.3033,
      "step": 47909
    },
    {
      "epoch": 2.682831224101243,
      "grad_norm": 1.2623804807662964,
      "learning_rate": 9.266052631578948e-05,
      "loss": 0.3946,
      "step": 47910
    },
    {
      "epoch": 2.682887221413372,
      "grad_norm": 1.051317811012268,
      "learning_rate": 9.266026315789475e-05,
      "loss": 0.3333,
      "step": 47911
    },
    {
      "epoch": 2.682943218725501,
      "grad_norm": 1.1511017084121704,
      "learning_rate": 9.266e-05,
      "loss": 0.3221,
      "step": 47912
    },
    {
      "epoch": 2.68299921603763,
      "grad_norm": 1.2766140699386597,
      "learning_rate": 9.265973684210527e-05,
      "loss": 0.4356,
      "step": 47913
    },
    {
      "epoch": 2.683055213349759,
      "grad_norm": 1.5672428607940674,
      "learning_rate": 9.265947368421053e-05,
      "loss": 0.4113,
      "step": 47914
    },
    {
      "epoch": 2.683111210661888,
      "grad_norm": 1.0446810722351074,
      "learning_rate": 9.26592105263158e-05,
      "loss": 0.3586,
      "step": 47915
    },
    {
      "epoch": 2.683167207974017,
      "grad_norm": 1.1152465343475342,
      "learning_rate": 9.265894736842105e-05,
      "loss": 0.386,
      "step": 47916
    },
    {
      "epoch": 2.683223205286146,
      "grad_norm": 0.9407862424850464,
      "learning_rate": 9.265868421052632e-05,
      "loss": 0.3601,
      "step": 47917
    },
    {
      "epoch": 2.6832792025982752,
      "grad_norm": 1.156243085861206,
      "learning_rate": 9.265842105263158e-05,
      "loss": 0.3488,
      "step": 47918
    },
    {
      "epoch": 2.6833351999104043,
      "grad_norm": 2.416597604751587,
      "learning_rate": 9.265815789473684e-05,
      "loss": 0.3958,
      "step": 47919
    },
    {
      "epoch": 2.6833911972225333,
      "grad_norm": 1.2758125066757202,
      "learning_rate": 9.265789473684212e-05,
      "loss": 0.3963,
      "step": 47920
    },
    {
      "epoch": 2.6834471945346623,
      "grad_norm": 1.1852244138717651,
      "learning_rate": 9.265763157894736e-05,
      "loss": 0.4184,
      "step": 47921
    },
    {
      "epoch": 2.6835031918467913,
      "grad_norm": 1.0322381258010864,
      "learning_rate": 9.265736842105264e-05,
      "loss": 0.2977,
      "step": 47922
    },
    {
      "epoch": 2.6835591891589203,
      "grad_norm": 1.0341237783432007,
      "learning_rate": 9.26571052631579e-05,
      "loss": 0.3016,
      "step": 47923
    },
    {
      "epoch": 2.6836151864710494,
      "grad_norm": 0.9271747469902039,
      "learning_rate": 9.265684210526317e-05,
      "loss": 0.3435,
      "step": 47924
    },
    {
      "epoch": 2.6836711837831784,
      "grad_norm": 1.0085557699203491,
      "learning_rate": 9.265657894736843e-05,
      "loss": 0.3356,
      "step": 47925
    },
    {
      "epoch": 2.6837271810953074,
      "grad_norm": 1.129323124885559,
      "learning_rate": 9.265631578947369e-05,
      "loss": 0.3355,
      "step": 47926
    },
    {
      "epoch": 2.6837831784074364,
      "grad_norm": 1.3734979629516602,
      "learning_rate": 9.265605263157895e-05,
      "loss": 0.4125,
      "step": 47927
    },
    {
      "epoch": 2.6838391757195654,
      "grad_norm": 1.2266839742660522,
      "learning_rate": 9.265578947368422e-05,
      "loss": 0.41,
      "step": 47928
    },
    {
      "epoch": 2.6838951730316944,
      "grad_norm": 1.1913975477218628,
      "learning_rate": 9.265552631578948e-05,
      "loss": 0.3633,
      "step": 47929
    },
    {
      "epoch": 2.6839511703438235,
      "grad_norm": 1.0053199529647827,
      "learning_rate": 9.265526315789474e-05,
      "loss": 0.3481,
      "step": 47930
    },
    {
      "epoch": 2.6840071676559525,
      "grad_norm": 1.1489588022232056,
      "learning_rate": 9.2655e-05,
      "loss": 0.4388,
      "step": 47931
    },
    {
      "epoch": 2.6840631649680815,
      "grad_norm": 1.1263055801391602,
      "learning_rate": 9.265473684210527e-05,
      "loss": 0.3868,
      "step": 47932
    },
    {
      "epoch": 2.6841191622802105,
      "grad_norm": 1.1484137773513794,
      "learning_rate": 9.265447368421053e-05,
      "loss": 0.4087,
      "step": 47933
    },
    {
      "epoch": 2.6841751595923395,
      "grad_norm": 1.2428553104400635,
      "learning_rate": 9.265421052631579e-05,
      "loss": 0.4271,
      "step": 47934
    },
    {
      "epoch": 2.6842311569044686,
      "grad_norm": 1.0594239234924316,
      "learning_rate": 9.265394736842105e-05,
      "loss": 0.4011,
      "step": 47935
    },
    {
      "epoch": 2.6842871542165976,
      "grad_norm": 1.0110453367233276,
      "learning_rate": 9.265368421052631e-05,
      "loss": 0.3102,
      "step": 47936
    },
    {
      "epoch": 2.6843431515287266,
      "grad_norm": 1.1090874671936035,
      "learning_rate": 9.265342105263159e-05,
      "loss": 0.4184,
      "step": 47937
    },
    {
      "epoch": 2.6843991488408556,
      "grad_norm": 4.073392868041992,
      "learning_rate": 9.265315789473684e-05,
      "loss": 0.4728,
      "step": 47938
    },
    {
      "epoch": 2.6844551461529846,
      "grad_norm": 1.1305532455444336,
      "learning_rate": 9.26528947368421e-05,
      "loss": 0.4704,
      "step": 47939
    },
    {
      "epoch": 2.6845111434651137,
      "grad_norm": 1.3449867963790894,
      "learning_rate": 9.265263157894736e-05,
      "loss": 0.3639,
      "step": 47940
    },
    {
      "epoch": 2.6845671407772427,
      "grad_norm": 1.0622413158416748,
      "learning_rate": 9.265236842105264e-05,
      "loss": 0.3348,
      "step": 47941
    },
    {
      "epoch": 2.6846231380893717,
      "grad_norm": 1.2072728872299194,
      "learning_rate": 9.26521052631579e-05,
      "loss": 0.3167,
      "step": 47942
    },
    {
      "epoch": 2.6846791354015007,
      "grad_norm": 1.2234331369400024,
      "learning_rate": 9.265184210526317e-05,
      "loss": 0.397,
      "step": 47943
    },
    {
      "epoch": 2.6847351327136297,
      "grad_norm": 1.2786176204681396,
      "learning_rate": 9.265157894736842e-05,
      "loss": 0.4768,
      "step": 47944
    },
    {
      "epoch": 2.6847911300257588,
      "grad_norm": 1.0533453226089478,
      "learning_rate": 9.265131578947369e-05,
      "loss": 0.4114,
      "step": 47945
    },
    {
      "epoch": 2.6848471273378878,
      "grad_norm": 1.072516679763794,
      "learning_rate": 9.265105263157895e-05,
      "loss": 0.3881,
      "step": 47946
    },
    {
      "epoch": 2.684903124650017,
      "grad_norm": 1.0557605028152466,
      "learning_rate": 9.265078947368422e-05,
      "loss": 0.4408,
      "step": 47947
    },
    {
      "epoch": 2.684959121962146,
      "grad_norm": 1.082983374595642,
      "learning_rate": 9.265052631578948e-05,
      "loss": 0.2921,
      "step": 47948
    },
    {
      "epoch": 2.685015119274275,
      "grad_norm": 0.924119770526886,
      "learning_rate": 9.265026315789474e-05,
      "loss": 0.3537,
      "step": 47949
    },
    {
      "epoch": 2.685071116586404,
      "grad_norm": 1.0329347848892212,
      "learning_rate": 9.265e-05,
      "loss": 0.3673,
      "step": 47950
    },
    {
      "epoch": 2.685127113898533,
      "grad_norm": 1.097707748413086,
      "learning_rate": 9.264973684210528e-05,
      "loss": 0.497,
      "step": 47951
    },
    {
      "epoch": 2.685183111210662,
      "grad_norm": 1.0307432413101196,
      "learning_rate": 9.264947368421054e-05,
      "loss": 0.4287,
      "step": 47952
    },
    {
      "epoch": 2.685239108522791,
      "grad_norm": 1.6949459314346313,
      "learning_rate": 9.26492105263158e-05,
      "loss": 0.412,
      "step": 47953
    },
    {
      "epoch": 2.68529510583492,
      "grad_norm": 1.2851179838180542,
      "learning_rate": 9.264894736842105e-05,
      "loss": 0.3064,
      "step": 47954
    },
    {
      "epoch": 2.685351103147049,
      "grad_norm": 1.037749171257019,
      "learning_rate": 9.264868421052631e-05,
      "loss": 0.3579,
      "step": 47955
    },
    {
      "epoch": 2.685407100459178,
      "grad_norm": 1.116367220878601,
      "learning_rate": 9.264842105263159e-05,
      "loss": 0.3778,
      "step": 47956
    },
    {
      "epoch": 2.685463097771307,
      "grad_norm": 1.2896662950515747,
      "learning_rate": 9.264815789473685e-05,
      "loss": 0.3838,
      "step": 47957
    },
    {
      "epoch": 2.685519095083436,
      "grad_norm": 0.9503831267356873,
      "learning_rate": 9.264789473684211e-05,
      "loss": 0.2932,
      "step": 47958
    },
    {
      "epoch": 2.685575092395565,
      "grad_norm": 1.1508657932281494,
      "learning_rate": 9.264763157894737e-05,
      "loss": 0.425,
      "step": 47959
    },
    {
      "epoch": 2.685631089707694,
      "grad_norm": 1.316573143005371,
      "learning_rate": 9.264736842105264e-05,
      "loss": 0.4829,
      "step": 47960
    },
    {
      "epoch": 2.685687087019823,
      "grad_norm": 0.9538385272026062,
      "learning_rate": 9.26471052631579e-05,
      "loss": 0.4012,
      "step": 47961
    },
    {
      "epoch": 2.685743084331952,
      "grad_norm": 1.1991348266601562,
      "learning_rate": 9.264684210526316e-05,
      "loss": 0.4134,
      "step": 47962
    },
    {
      "epoch": 2.685799081644081,
      "grad_norm": 1.221527099609375,
      "learning_rate": 9.264657894736842e-05,
      "loss": 0.4077,
      "step": 47963
    },
    {
      "epoch": 2.68585507895621,
      "grad_norm": 1.2710380554199219,
      "learning_rate": 9.264631578947369e-05,
      "loss": 0.3191,
      "step": 47964
    },
    {
      "epoch": 2.685911076268339,
      "grad_norm": 1.318284273147583,
      "learning_rate": 9.264605263157895e-05,
      "loss": 0.4607,
      "step": 47965
    },
    {
      "epoch": 2.685967073580468,
      "grad_norm": 1.359891414642334,
      "learning_rate": 9.264578947368423e-05,
      "loss": 0.5648,
      "step": 47966
    },
    {
      "epoch": 2.686023070892597,
      "grad_norm": 1.0615447759628296,
      "learning_rate": 9.264552631578947e-05,
      "loss": 0.3833,
      "step": 47967
    },
    {
      "epoch": 2.686079068204726,
      "grad_norm": 1.019260048866272,
      "learning_rate": 9.264526315789475e-05,
      "loss": 0.3797,
      "step": 47968
    },
    {
      "epoch": 2.686135065516855,
      "grad_norm": 1.2075841426849365,
      "learning_rate": 9.2645e-05,
      "loss": 0.4394,
      "step": 47969
    },
    {
      "epoch": 2.6861910628289842,
      "grad_norm": 1.1413755416870117,
      "learning_rate": 9.264473684210528e-05,
      "loss": 0.4114,
      "step": 47970
    },
    {
      "epoch": 2.6862470601411133,
      "grad_norm": 1.404134750366211,
      "learning_rate": 9.264447368421052e-05,
      "loss": 0.3621,
      "step": 47971
    },
    {
      "epoch": 2.6863030574532423,
      "grad_norm": 1.2854489088058472,
      "learning_rate": 9.264421052631578e-05,
      "loss": 0.3837,
      "step": 47972
    },
    {
      "epoch": 2.6863590547653713,
      "grad_norm": 1.9556503295898438,
      "learning_rate": 9.264394736842106e-05,
      "loss": 0.4353,
      "step": 47973
    },
    {
      "epoch": 2.6864150520775003,
      "grad_norm": 1.364250659942627,
      "learning_rate": 9.264368421052632e-05,
      "loss": 0.5261,
      "step": 47974
    },
    {
      "epoch": 2.6864710493896293,
      "grad_norm": 1.2445117235183716,
      "learning_rate": 9.264342105263159e-05,
      "loss": 0.2905,
      "step": 47975
    },
    {
      "epoch": 2.6865270467017583,
      "grad_norm": 1.1323368549346924,
      "learning_rate": 9.264315789473684e-05,
      "loss": 0.424,
      "step": 47976
    },
    {
      "epoch": 2.6865830440138874,
      "grad_norm": 1.2794936895370483,
      "learning_rate": 9.264289473684211e-05,
      "loss": 0.5011,
      "step": 47977
    },
    {
      "epoch": 2.6866390413260164,
      "grad_norm": 1.2332236766815186,
      "learning_rate": 9.264263157894737e-05,
      "loss": 0.4695,
      "step": 47978
    },
    {
      "epoch": 2.6866950386381454,
      "grad_norm": 1.0451505184173584,
      "learning_rate": 9.264236842105264e-05,
      "loss": 0.4082,
      "step": 47979
    },
    {
      "epoch": 2.6867510359502744,
      "grad_norm": 1.5541996955871582,
      "learning_rate": 9.26421052631579e-05,
      "loss": 0.6527,
      "step": 47980
    },
    {
      "epoch": 2.6868070332624034,
      "grad_norm": 1.4219582080841064,
      "learning_rate": 9.264184210526316e-05,
      "loss": 0.5364,
      "step": 47981
    },
    {
      "epoch": 2.6868630305745325,
      "grad_norm": 1.116012692451477,
      "learning_rate": 9.264157894736842e-05,
      "loss": 0.353,
      "step": 47982
    },
    {
      "epoch": 2.6869190278866615,
      "grad_norm": 1.1991639137268066,
      "learning_rate": 9.26413157894737e-05,
      "loss": 0.3888,
      "step": 47983
    },
    {
      "epoch": 2.6869750251987905,
      "grad_norm": 1.2383548021316528,
      "learning_rate": 9.264105263157896e-05,
      "loss": 0.3745,
      "step": 47984
    },
    {
      "epoch": 2.6870310225109195,
      "grad_norm": 1.2030359506607056,
      "learning_rate": 9.264078947368421e-05,
      "loss": 0.3804,
      "step": 47985
    },
    {
      "epoch": 2.6870870198230485,
      "grad_norm": 1.30626380443573,
      "learning_rate": 9.264052631578947e-05,
      "loss": 0.4919,
      "step": 47986
    },
    {
      "epoch": 2.6871430171351776,
      "grad_norm": 1.3223153352737427,
      "learning_rate": 9.264026315789475e-05,
      "loss": 0.4822,
      "step": 47987
    },
    {
      "epoch": 2.6871990144473066,
      "grad_norm": 1.3755075931549072,
      "learning_rate": 9.264000000000001e-05,
      "loss": 0.4073,
      "step": 47988
    },
    {
      "epoch": 2.6872550117594356,
      "grad_norm": 1.0453438758850098,
      "learning_rate": 9.263973684210527e-05,
      "loss": 0.4705,
      "step": 47989
    },
    {
      "epoch": 2.6873110090715646,
      "grad_norm": 1.2186061143875122,
      "learning_rate": 9.263947368421053e-05,
      "loss": 0.6236,
      "step": 47990
    },
    {
      "epoch": 2.6873670063836936,
      "grad_norm": 1.1164683103561401,
      "learning_rate": 9.263921052631579e-05,
      "loss": 0.3955,
      "step": 47991
    },
    {
      "epoch": 2.6874230036958227,
      "grad_norm": 1.1177412271499634,
      "learning_rate": 9.263894736842106e-05,
      "loss": 0.378,
      "step": 47992
    },
    {
      "epoch": 2.6874790010079517,
      "grad_norm": 1.1956828832626343,
      "learning_rate": 9.263868421052632e-05,
      "loss": 0.367,
      "step": 47993
    },
    {
      "epoch": 2.6875349983200807,
      "grad_norm": 1.2705851793289185,
      "learning_rate": 9.263842105263158e-05,
      "loss": 0.3338,
      "step": 47994
    },
    {
      "epoch": 2.6875909956322097,
      "grad_norm": 1.599159598350525,
      "learning_rate": 9.263815789473684e-05,
      "loss": 0.4086,
      "step": 47995
    },
    {
      "epoch": 2.6876469929443387,
      "grad_norm": 1.3865232467651367,
      "learning_rate": 9.263789473684211e-05,
      "loss": 0.4835,
      "step": 47996
    },
    {
      "epoch": 2.6877029902564677,
      "grad_norm": 1.4632568359375,
      "learning_rate": 9.263763157894737e-05,
      "loss": 0.4259,
      "step": 47997
    },
    {
      "epoch": 2.6877589875685968,
      "grad_norm": 1.3144479990005493,
      "learning_rate": 9.263736842105265e-05,
      "loss": 0.5303,
      "step": 47998
    },
    {
      "epoch": 2.687814984880726,
      "grad_norm": 1.648030161857605,
      "learning_rate": 9.263710526315789e-05,
      "loss": 0.3871,
      "step": 47999
    },
    {
      "epoch": 2.687870982192855,
      "grad_norm": 1.5107238292694092,
      "learning_rate": 9.263684210526316e-05,
      "loss": 0.5785,
      "step": 48000
    }
  ],
  "logging_steps": 1,
  "max_steps": 400000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 23,
  "save_steps": 4000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 8.470653447395082e+17,
  "train_batch_size": 28,
  "trial_name": null,
  "trial_params": null
}