{
  "best_metric": 0.15221890807151794,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.14711290915777858,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000735564545788893,
      "grad_norm": 1.3955460786819458,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.8061,
      "step": 1
    },
    {
      "epoch": 0.000735564545788893,
      "eval_loss": 1.0986895561218262,
      "eval_runtime": 191.9666,
      "eval_samples_per_second": 11.929,
      "eval_steps_per_second": 5.965,
      "step": 1
    },
    {
      "epoch": 0.001471129091577786,
      "grad_norm": 1.407787799835205,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.8431,
      "step": 2
    },
    {
      "epoch": 0.002206693637366679,
      "grad_norm": 1.375799536705017,
      "learning_rate": 1e-05,
      "loss": 0.8597,
      "step": 3
    },
    {
      "epoch": 0.002942258183155572,
      "grad_norm": 1.491884469985962,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.8988,
      "step": 4
    },
    {
      "epoch": 0.003677822728944465,
      "grad_norm": 1.3589227199554443,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.853,
      "step": 5
    },
    {
      "epoch": 0.004413387274733358,
      "grad_norm": 1.2300374507904053,
      "learning_rate": 2e-05,
      "loss": 0.8203,
      "step": 6
    },
    {
      "epoch": 0.0051489518205222505,
      "grad_norm": 1.02793550491333,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 0.7321,
      "step": 7
    },
    {
      "epoch": 0.005884516366311144,
      "grad_norm": 0.952317476272583,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.7058,
      "step": 8
    },
    {
      "epoch": 0.0066200809121000365,
      "grad_norm": 0.8238752484321594,
      "learning_rate": 3e-05,
      "loss": 0.6184,
      "step": 9
    },
    {
      "epoch": 0.00735564545788893,
      "grad_norm": 0.9686686992645264,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.5712,
      "step": 10
    },
    {
      "epoch": 0.008091210003677823,
      "grad_norm": 0.9226304888725281,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 0.4854,
      "step": 11
    },
    {
      "epoch": 0.008826774549466716,
      "grad_norm": 0.8993305563926697,
      "learning_rate": 4e-05,
      "loss": 0.4479,
      "step": 12
    },
    {
      "epoch": 0.009562339095255609,
      "grad_norm": 0.8207470178604126,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 0.3879,
      "step": 13
    },
    {
      "epoch": 0.010297903641044501,
      "grad_norm": 0.7917346954345703,
      "learning_rate": 4.666666666666667e-05,
      "loss": 0.3442,
      "step": 14
    },
    {
      "epoch": 0.011033468186833395,
      "grad_norm": 0.6463001370429993,
      "learning_rate": 5e-05,
      "loss": 0.2924,
      "step": 15
    },
    {
      "epoch": 0.011769032732622288,
      "grad_norm": 0.524644672870636,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.2662,
      "step": 16
    },
    {
      "epoch": 0.01250459727841118,
      "grad_norm": 0.5327229499816895,
      "learning_rate": 5.666666666666667e-05,
      "loss": 0.2513,
      "step": 17
    },
    {
      "epoch": 0.013240161824200073,
      "grad_norm": 0.6008023619651794,
      "learning_rate": 6e-05,
      "loss": 0.2499,
      "step": 18
    },
    {
      "epoch": 0.013975726369988967,
      "grad_norm": 0.4902108311653137,
      "learning_rate": 6.333333333333333e-05,
      "loss": 0.2138,
      "step": 19
    },
    {
      "epoch": 0.01471129091577786,
      "grad_norm": 0.5153264999389648,
      "learning_rate": 6.666666666666667e-05,
      "loss": 0.2068,
      "step": 20
    },
    {
      "epoch": 0.015446855461566752,
      "grad_norm": 0.5366236567497253,
      "learning_rate": 7e-05,
      "loss": 0.197,
      "step": 21
    },
    {
      "epoch": 0.016182420007355647,
      "grad_norm": 0.407263845205307,
      "learning_rate": 7.333333333333333e-05,
      "loss": 0.1966,
      "step": 22
    },
    {
      "epoch": 0.016917984553144538,
      "grad_norm": 0.36304229497909546,
      "learning_rate": 7.666666666666667e-05,
      "loss": 0.2016,
      "step": 23
    },
    {
      "epoch": 0.017653549098933432,
      "grad_norm": 0.3669580817222595,
      "learning_rate": 8e-05,
      "loss": 0.2059,
      "step": 24
    },
    {
      "epoch": 0.018389113644722323,
      "grad_norm": 0.22362849116325378,
      "learning_rate": 8.333333333333334e-05,
      "loss": 0.1527,
      "step": 25
    },
    {
      "epoch": 0.019124678190511217,
      "grad_norm": 0.21722978353500366,
      "learning_rate": 8.666666666666667e-05,
      "loss": 0.1503,
      "step": 26
    },
    {
      "epoch": 0.01986024273630011,
      "grad_norm": 0.2847565710544586,
      "learning_rate": 9e-05,
      "loss": 0.1685,
      "step": 27
    },
    {
      "epoch": 0.020595807282089002,
      "grad_norm": 0.2842387557029724,
      "learning_rate": 9.333333333333334e-05,
      "loss": 0.1553,
      "step": 28
    },
    {
      "epoch": 0.021331371827877896,
      "grad_norm": 0.25112128257751465,
      "learning_rate": 9.666666666666667e-05,
      "loss": 0.151,
      "step": 29
    },
    {
      "epoch": 0.02206693637366679,
      "grad_norm": 0.2505984306335449,
      "learning_rate": 0.0001,
      "loss": 0.1433,
      "step": 30
    },
    {
      "epoch": 0.02280250091945568,
      "grad_norm": 0.245281383395195,
      "learning_rate": 9.999146252290264e-05,
      "loss": 0.1618,
      "step": 31
    },
    {
      "epoch": 0.023538065465244576,
      "grad_norm": 0.31131860613822937,
      "learning_rate": 9.996585300715116e-05,
      "loss": 0.1587,
      "step": 32
    },
    {
      "epoch": 0.024273630011033467,
      "grad_norm": 0.2021295130252838,
      "learning_rate": 9.99231801983717e-05,
      "loss": 0.1654,
      "step": 33
    },
    {
      "epoch": 0.02500919455682236,
      "grad_norm": 0.23759086430072784,
      "learning_rate": 9.986345866928941e-05,
      "loss": 0.1569,
      "step": 34
    },
    {
      "epoch": 0.025744759102611255,
      "grad_norm": 0.23068475723266602,
      "learning_rate": 9.978670881475172e-05,
      "loss": 0.1664,
      "step": 35
    },
    {
      "epoch": 0.026480323648400146,
      "grad_norm": 0.23946265876293182,
      "learning_rate": 9.96929568447637e-05,
      "loss": 0.1418,
      "step": 36
    },
    {
      "epoch": 0.02721588819418904,
      "grad_norm": 0.16092844307422638,
      "learning_rate": 9.958223477553714e-05,
      "loss": 0.1519,
      "step": 37
    },
    {
      "epoch": 0.027951452739977935,
      "grad_norm": 0.20561547577381134,
      "learning_rate": 9.94545804185573e-05,
      "loss": 0.1334,
      "step": 38
    },
    {
      "epoch": 0.028687017285766826,
      "grad_norm": 0.24364995956420898,
      "learning_rate": 9.931003736767013e-05,
      "loss": 0.158,
      "step": 39
    },
    {
      "epoch": 0.02942258183155572,
      "grad_norm": 0.2280334234237671,
      "learning_rate": 9.91486549841951e-05,
      "loss": 0.1429,
      "step": 40
    },
    {
      "epoch": 0.03015814637734461,
      "grad_norm": 0.25129419565200806,
      "learning_rate": 9.89704883800683e-05,
      "loss": 0.1689,
      "step": 41
    },
    {
      "epoch": 0.030893710923133505,
      "grad_norm": 0.15832379460334778,
      "learning_rate": 9.877559839902184e-05,
      "loss": 0.1559,
      "step": 42
    },
    {
      "epoch": 0.0316292754689224,
      "grad_norm": 0.26977041363716125,
      "learning_rate": 9.85640515958057e-05,
      "loss": 0.1665,
      "step": 43
    },
    {
      "epoch": 0.032364840014711294,
      "grad_norm": 0.18031959235668182,
      "learning_rate": 9.833592021345937e-05,
      "loss": 0.1779,
      "step": 44
    },
    {
      "epoch": 0.03310040456050018,
      "grad_norm": 0.22443000972270966,
      "learning_rate": 9.809128215864097e-05,
      "loss": 0.1847,
      "step": 45
    },
    {
      "epoch": 0.033835969106289075,
      "grad_norm": 0.2750892639160156,
      "learning_rate": 9.783022097502204e-05,
      "loss": 0.1664,
      "step": 46
    },
    {
      "epoch": 0.03457153365207797,
      "grad_norm": 0.2007928043603897,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.1082,
      "step": 47
    },
    {
      "epoch": 0.035307098197866864,
      "grad_norm": 0.20176196098327637,
      "learning_rate": 9.725919140804099e-05,
      "loss": 0.126,
      "step": 48
    },
    {
      "epoch": 0.03604266274365576,
      "grad_norm": 0.19546356797218323,
      "learning_rate": 9.694941803075283e-05,
      "loss": 0.1979,
      "step": 49
    },
    {
      "epoch": 0.036778227289444645,
      "grad_norm": 0.4081459641456604,
      "learning_rate": 9.662361147021779e-05,
      "loss": 0.1876,
      "step": 50
    },
    {
      "epoch": 0.036778227289444645,
      "eval_loss": 0.19655592739582062,
      "eval_runtime": 192.9534,
      "eval_samples_per_second": 11.868,
      "eval_steps_per_second": 5.934,
      "step": 50
    },
    {
      "epoch": 0.03751379183523354,
      "grad_norm": 0.9154587388038635,
      "learning_rate": 9.628188298907782e-05,
      "loss": 0.1907,
      "step": 51
    },
    {
      "epoch": 0.038249356381022434,
      "grad_norm": 0.3832627832889557,
      "learning_rate": 9.592434928729616e-05,
      "loss": 0.1698,
      "step": 52
    },
    {
      "epoch": 0.03898492092681133,
      "grad_norm": 0.344266414642334,
      "learning_rate": 9.555113246230442e-05,
      "loss": 0.1614,
      "step": 53
    },
    {
      "epoch": 0.03972048547260022,
      "grad_norm": 0.3299565017223358,
      "learning_rate": 9.516235996730645e-05,
      "loss": 0.1651,
      "step": 54
    },
    {
      "epoch": 0.04045605001838912,
      "grad_norm": 0.2757066488265991,
      "learning_rate": 9.475816456775313e-05,
      "loss": 0.1715,
      "step": 55
    },
    {
      "epoch": 0.041191614564178004,
      "grad_norm": 0.26629284024238586,
      "learning_rate": 9.43386842960031e-05,
      "loss": 0.1632,
      "step": 56
    },
    {
      "epoch": 0.0419271791099669,
      "grad_norm": 0.2985883057117462,
      "learning_rate": 9.39040624041849e-05,
      "loss": 0.177,
      "step": 57
    },
    {
      "epoch": 0.04266274365575579,
      "grad_norm": 0.1407419890165329,
      "learning_rate": 9.345444731527642e-05,
      "loss": 0.1631,
      "step": 58
    },
    {
      "epoch": 0.04339830820154469,
      "grad_norm": 0.16616809368133545,
      "learning_rate": 9.298999257241863e-05,
      "loss": 0.1715,
      "step": 59
    },
    {
      "epoch": 0.04413387274733358,
      "grad_norm": 0.18962794542312622,
      "learning_rate": 9.251085678648072e-05,
      "loss": 0.181,
      "step": 60
    },
    {
      "epoch": 0.04486943729312247,
      "grad_norm": 0.17522728443145752,
      "learning_rate": 9.201720358189464e-05,
      "loss": 0.1759,
      "step": 61
    },
    {
      "epoch": 0.04560500183891136,
      "grad_norm": 0.14696826040744781,
      "learning_rate": 9.150920154077754e-05,
      "loss": 0.1894,
      "step": 62
    },
    {
      "epoch": 0.04634056638470026,
      "grad_norm": 0.13528281450271606,
      "learning_rate": 9.098702414536107e-05,
      "loss": 0.1743,
      "step": 63
    },
    {
      "epoch": 0.04707613093048915,
      "grad_norm": 0.13108627498149872,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.1666,
      "step": 64
    },
    {
      "epoch": 0.047811695476278046,
      "grad_norm": 0.15503785014152527,
      "learning_rate": 8.9900861364012e-05,
      "loss": 0.1721,
      "step": 65
    },
    {
      "epoch": 0.04854726002206693,
      "grad_norm": 0.13304218649864197,
      "learning_rate": 8.933724690167417e-05,
      "loss": 0.158,
      "step": 66
    },
    {
      "epoch": 0.04928282456785583,
      "grad_norm": 0.1355734020471573,
      "learning_rate": 8.876019880555649e-05,
      "loss": 0.1752,
      "step": 67
    },
    {
      "epoch": 0.05001838911364472,
      "grad_norm": 0.1821872442960739,
      "learning_rate": 8.816991413705516e-05,
      "loss": 0.1667,
      "step": 68
    },
    {
      "epoch": 0.050753953659433616,
      "grad_norm": 0.12791278958320618,
      "learning_rate": 8.756659447784368e-05,
      "loss": 0.1646,
      "step": 69
    },
    {
      "epoch": 0.05148951820522251,
      "grad_norm": 0.12836161255836487,
      "learning_rate": 8.695044586103296e-05,
      "loss": 0.165,
      "step": 70
    },
    {
      "epoch": 0.0522250827510114,
      "grad_norm": 0.11118108779191971,
      "learning_rate": 8.632167870081121e-05,
      "loss": 0.1767,
      "step": 71
    },
    {
      "epoch": 0.05296064729680029,
      "grad_norm": 0.18004432320594788,
      "learning_rate": 8.568050772058762e-05,
      "loss": 0.1536,
      "step": 72
    },
    {
      "epoch": 0.05369621184258919,
      "grad_norm": 0.16035598516464233,
      "learning_rate": 8.502715187966455e-05,
      "loss": 0.1501,
      "step": 73
    },
    {
      "epoch": 0.05443177638837808,
      "grad_norm": 0.11390311270952225,
      "learning_rate": 8.436183429846313e-05,
      "loss": 0.1473,
      "step": 74
    },
    {
      "epoch": 0.055167340934166975,
      "grad_norm": 0.11487963795661926,
      "learning_rate": 8.368478218232787e-05,
      "loss": 0.1371,
      "step": 75
    },
    {
      "epoch": 0.05590290547995587,
      "grad_norm": 0.1358136385679245,
      "learning_rate": 8.299622674393614e-05,
      "loss": 0.1354,
      "step": 76
    },
    {
      "epoch": 0.05663847002574476,
      "grad_norm": 0.11609425395727158,
      "learning_rate": 8.229640312433937e-05,
      "loss": 0.1339,
      "step": 77
    },
    {
      "epoch": 0.05737403457153365,
      "grad_norm": 0.10956840217113495,
      "learning_rate": 8.158555031266254e-05,
      "loss": 0.1329,
      "step": 78
    },
    {
      "epoch": 0.058109599117322545,
      "grad_norm": 0.15083730220794678,
      "learning_rate": 8.086391106448965e-05,
      "loss": 0.1221,
      "step": 79
    },
    {
      "epoch": 0.05884516366311144,
      "grad_norm": 0.13061225414276123,
      "learning_rate": 8.013173181896283e-05,
      "loss": 0.1356,
      "step": 80
    },
    {
      "epoch": 0.059580728208900334,
      "grad_norm": 0.09824299067258835,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.1425,
      "step": 81
    },
    {
      "epoch": 0.06031629275468922,
      "grad_norm": 0.23265309631824493,
      "learning_rate": 7.863675700402526e-05,
      "loss": 0.1478,
      "step": 82
    },
    {
      "epoch": 0.061051857300478116,
      "grad_norm": 0.10225420445203781,
      "learning_rate": 7.787447196714427e-05,
      "loss": 0.1348,
      "step": 83
    },
    {
      "epoch": 0.06178742184626701,
      "grad_norm": 0.14389212429523468,
      "learning_rate": 7.710266782362247e-05,
      "loss": 0.1449,
      "step": 84
    },
    {
      "epoch": 0.0625229863920559,
      "grad_norm": 0.22097352147102356,
      "learning_rate": 7.63216081438678e-05,
      "loss": 0.1407,
      "step": 85
    },
    {
      "epoch": 0.0632585509378448,
      "grad_norm": 0.13521602749824524,
      "learning_rate": 7.553155965904535e-05,
      "loss": 0.1591,
      "step": 86
    },
    {
      "epoch": 0.06399411548363369,
      "grad_norm": 0.09170956909656525,
      "learning_rate": 7.473279216998895e-05,
      "loss": 0.1366,
      "step": 87
    },
    {
      "epoch": 0.06472968002942259,
      "grad_norm": 0.09572591632604599,
      "learning_rate": 7.392557845506432e-05,
      "loss": 0.1557,
      "step": 88
    },
    {
      "epoch": 0.06546524457521148,
      "grad_norm": 0.12399427592754364,
      "learning_rate": 7.311019417701566e-05,
      "loss": 0.1534,
      "step": 89
    },
    {
      "epoch": 0.06620080912100036,
      "grad_norm": 0.09619427472352982,
      "learning_rate": 7.228691778882693e-05,
      "loss": 0.1561,
      "step": 90
    },
    {
      "epoch": 0.06693637366678926,
      "grad_norm": 0.09777440130710602,
      "learning_rate": 7.145603043863045e-05,
      "loss": 0.1528,
      "step": 91
    },
    {
      "epoch": 0.06767193821257815,
      "grad_norm": 0.14338025450706482,
      "learning_rate": 7.061781587369519e-05,
      "loss": 0.1577,
      "step": 92
    },
    {
      "epoch": 0.06840750275836704,
      "grad_norm": 0.08410169184207916,
      "learning_rate": 6.977256034352712e-05,
      "loss": 0.1427,
      "step": 93
    },
    {
      "epoch": 0.06914306730415594,
      "grad_norm": 0.09892772138118744,
      "learning_rate": 6.892055250211552e-05,
      "loss": 0.1435,
      "step": 94
    },
    {
      "epoch": 0.06987863184994483,
      "grad_norm": 0.2183755338191986,
      "learning_rate": 6.806208330935766e-05,
      "loss": 0.1738,
      "step": 95
    },
    {
      "epoch": 0.07061419639573373,
      "grad_norm": 0.14419087767601013,
      "learning_rate": 6.719744593169641e-05,
      "loss": 0.1714,
      "step": 96
    },
    {
      "epoch": 0.07134976094152262,
      "grad_norm": 0.0980491042137146,
      "learning_rate": 6.632693564200416e-05,
      "loss": 0.0685,
      "step": 97
    },
    {
      "epoch": 0.07208532548731152,
      "grad_norm": 0.08917222917079926,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.1289,
      "step": 98
    },
    {
      "epoch": 0.07282089003310041,
      "grad_norm": 0.1148543581366539,
      "learning_rate": 6.456948734446624e-05,
      "loss": 0.1919,
      "step": 99
    },
    {
      "epoch": 0.07355645457888929,
      "grad_norm": 0.1381547749042511,
      "learning_rate": 6.368314950360415e-05,
      "loss": 0.1883,
      "step": 100
    },
    {
      "epoch": 0.07355645457888929,
      "eval_loss": 0.1586042046546936,
      "eval_runtime": 193.0598,
      "eval_samples_per_second": 11.862,
      "eval_steps_per_second": 5.931,
      "step": 100
    },
    {
      "epoch": 0.07429201912467819,
      "grad_norm": 0.48300158977508545,
      "learning_rate": 6.279213887972179e-05,
      "loss": 0.1455,
      "step": 101
    },
    {
      "epoch": 0.07502758367046708,
      "grad_norm": 0.13084329664707184,
      "learning_rate": 6.189675975213094e-05,
      "loss": 0.1361,
      "step": 102
    },
    {
      "epoch": 0.07576314821625597,
      "grad_norm": 0.11135324835777283,
      "learning_rate": 6.099731789198344e-05,
      "loss": 0.1476,
      "step": 103
    },
    {
      "epoch": 0.07649871276204487,
      "grad_norm": 0.17674945294857025,
      "learning_rate": 6.009412045785051e-05,
      "loss": 0.1504,
      "step": 104
    },
    {
      "epoch": 0.07723427730783376,
      "grad_norm": 0.31758978962898254,
      "learning_rate": 5.918747589082853e-05,
      "loss": 0.1508,
      "step": 105
    },
    {
      "epoch": 0.07796984185362266,
      "grad_norm": 0.14566396176815033,
      "learning_rate": 5.82776938092065e-05,
      "loss": 0.1425,
      "step": 106
    },
    {
      "epoch": 0.07870540639941155,
      "grad_norm": 0.18799343705177307,
      "learning_rate": 5.736508490273188e-05,
      "loss": 0.1478,
      "step": 107
    },
    {
      "epoch": 0.07944097094520045,
      "grad_norm": 0.14357222616672516,
      "learning_rate": 5.644996082651017e-05,
      "loss": 0.1617,
      "step": 108
    },
    {
      "epoch": 0.08017653549098934,
      "grad_norm": 0.1759078949689865,
      "learning_rate": 5.553263409457504e-05,
      "loss": 0.1652,
      "step": 109
    },
    {
      "epoch": 0.08091210003677823,
      "grad_norm": 0.13642680644989014,
      "learning_rate": 5.4613417973165106e-05,
      "loss": 0.158,
      "step": 110
    },
    {
      "epoch": 0.08164766458256711,
      "grad_norm": 0.0958728939294815,
      "learning_rate": 5.3692626373743706e-05,
      "loss": 0.1736,
      "step": 111
    },
    {
      "epoch": 0.08238322912835601,
      "grad_norm": 0.10906694084405899,
      "learning_rate": 5.27705737457985e-05,
      "loss": 0.1725,
      "step": 112
    },
    {
      "epoch": 0.0831187936741449,
      "grad_norm": 0.09963790327310562,
      "learning_rate": 5.184757496945726e-05,
      "loss": 0.1853,
      "step": 113
    },
    {
      "epoch": 0.0838543582199338,
      "grad_norm": 0.10227838158607483,
      "learning_rate": 5.092394524795649e-05,
      "loss": 0.1506,
      "step": 114
    },
    {
      "epoch": 0.08458992276572269,
      "grad_norm": 0.11423751711845398,
      "learning_rate": 5e-05,
      "loss": 0.1647,
      "step": 115
    },
    {
      "epoch": 0.08532548731151159,
      "grad_norm": 0.1167915016412735,
      "learning_rate": 4.907605475204352e-05,
      "loss": 0.1701,
      "step": 116
    },
    {
      "epoch": 0.08606105185730048,
      "grad_norm": 0.10179633647203445,
      "learning_rate": 4.8152425030542766e-05,
      "loss": 0.1784,
      "step": 117
    },
    {
      "epoch": 0.08679661640308937,
      "grad_norm": 0.07918064296245575,
      "learning_rate": 4.72294262542015e-05,
      "loss": 0.1655,
      "step": 118
    },
    {
      "epoch": 0.08753218094887827,
      "grad_norm": 0.09195617586374283,
      "learning_rate": 4.6307373626256306e-05,
      "loss": 0.1673,
      "step": 119
    },
    {
      "epoch": 0.08826774549466716,
      "grad_norm": 0.08825042843818665,
      "learning_rate": 4.5386582026834906e-05,
      "loss": 0.1792,
      "step": 120
    },
    {
      "epoch": 0.08900331004045604,
      "grad_norm": 0.10082146525382996,
      "learning_rate": 4.446736590542497e-05,
      "loss": 0.1701,
      "step": 121
    },
    {
      "epoch": 0.08973887458624494,
      "grad_norm": 0.09268957376480103,
      "learning_rate": 4.3550039173489845e-05,
      "loss": 0.1681,
      "step": 122
    },
    {
      "epoch": 0.09047443913203383,
      "grad_norm": 0.18448396027088165,
      "learning_rate": 4.2634915097268115e-05,
      "loss": 0.159,
      "step": 123
    },
    {
      "epoch": 0.09121000367782273,
      "grad_norm": 0.08122457563877106,
      "learning_rate": 4.1722306190793495e-05,
      "loss": 0.152,
      "step": 124
    },
    {
      "epoch": 0.09194556822361162,
      "grad_norm": 0.11715678125619888,
      "learning_rate": 4.0812524109171476e-05,
      "loss": 0.1563,
      "step": 125
    },
    {
      "epoch": 0.09268113276940051,
      "grad_norm": 0.06476876884698868,
      "learning_rate": 3.99058795421495e-05,
      "loss": 0.1444,
      "step": 126
    },
    {
      "epoch": 0.09341669731518941,
      "grad_norm": 0.08869259804487228,
      "learning_rate": 3.9002682108016585e-05,
      "loss": 0.141,
      "step": 127
    },
    {
      "epoch": 0.0941522618609783,
      "grad_norm": 0.09998400509357452,
      "learning_rate": 3.8103240247869075e-05,
      "loss": 0.1466,
      "step": 128
    },
    {
      "epoch": 0.0948878264067672,
      "grad_norm": 0.07661914825439453,
      "learning_rate": 3.720786112027822e-05,
      "loss": 0.1294,
      "step": 129
    },
    {
      "epoch": 0.09562339095255609,
      "grad_norm": 0.07216975837945938,
      "learning_rate": 3.631685049639586e-05,
      "loss": 0.1293,
      "step": 130
    },
    {
      "epoch": 0.09635895549834499,
      "grad_norm": 0.09876257926225662,
      "learning_rate": 3.543051265553377e-05,
      "loss": 0.153,
      "step": 131
    },
    {
      "epoch": 0.09709452004413387,
      "grad_norm": 0.08432336151599884,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.1561,
      "step": 132
    },
    {
      "epoch": 0.09783008458992276,
      "grad_norm": 0.15363752841949463,
      "learning_rate": 3.367306435799584e-05,
      "loss": 0.1451,
      "step": 133
    },
    {
      "epoch": 0.09856564913571166,
      "grad_norm": 0.0780152752995491,
      "learning_rate": 3.2802554068303596e-05,
      "loss": 0.1562,
      "step": 134
    },
    {
      "epoch": 0.09930121368150055,
      "grad_norm": 0.07892629504203796,
      "learning_rate": 3.1937916690642356e-05,
      "loss": 0.1575,
      "step": 135
    },
    {
      "epoch": 0.10003677822728944,
      "grad_norm": 0.06880857050418854,
      "learning_rate": 3.107944749788449e-05,
      "loss": 0.1461,
      "step": 136
    },
    {
      "epoch": 0.10077234277307834,
      "grad_norm": 0.11363841593265533,
      "learning_rate": 3.0227439656472877e-05,
      "loss": 0.163,
      "step": 137
    },
    {
      "epoch": 0.10150790731886723,
      "grad_norm": 0.05595669150352478,
      "learning_rate": 2.9382184126304834e-05,
      "loss": 0.1535,
      "step": 138
    },
    {
      "epoch": 0.10224347186465613,
      "grad_norm": 0.054943084716796875,
      "learning_rate": 2.8543969561369556e-05,
      "loss": 0.1232,
      "step": 139
    },
    {
      "epoch": 0.10297903641044502,
      "grad_norm": 0.07336664199829102,
      "learning_rate": 2.771308221117309e-05,
      "loss": 0.1348,
      "step": 140
    },
    {
      "epoch": 0.10371460095623392,
      "grad_norm": 0.044373560696840286,
      "learning_rate": 2.688980582298435e-05,
      "loss": 0.1183,
      "step": 141
    },
    {
      "epoch": 0.1044501655020228,
      "grad_norm": 0.0573909729719162,
      "learning_rate": 2.607442154493568e-05,
      "loss": 0.1536,
      "step": 142
    },
    {
      "epoch": 0.10518573004781169,
      "grad_norm": 0.06830648332834244,
      "learning_rate": 2.5267207830011068e-05,
      "loss": 0.1489,
      "step": 143
    },
    {
      "epoch": 0.10592129459360058,
      "grad_norm": 0.06195629760622978,
      "learning_rate": 2.446844034095466e-05,
      "loss": 0.1512,
      "step": 144
    },
    {
      "epoch": 0.10665685913938948,
      "grad_norm": 0.06268982589244843,
      "learning_rate": 2.3678391856132204e-05,
      "loss": 0.1738,
      "step": 145
    },
    {
      "epoch": 0.10739242368517837,
      "grad_norm": 0.08042903244495392,
      "learning_rate": 2.2897332176377528e-05,
      "loss": 0.1464,
      "step": 146
    },
    {
      "epoch": 0.10812798823096727,
      "grad_norm": 0.07065901905298233,
      "learning_rate": 2.2125528032855724e-05,
      "loss": 0.1503,
      "step": 147
    },
    {
      "epoch": 0.10886355277675616,
      "grad_norm": 0.04282858222723007,
      "learning_rate": 2.136324299597474e-05,
      "loss": 0.113,
      "step": 148
    },
    {
      "epoch": 0.10959911732254506,
      "grad_norm": 0.07070647180080414,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.1903,
      "step": 149
    },
    {
      "epoch": 0.11033468186833395,
      "grad_norm": 0.2172415852546692,
      "learning_rate": 1.9868268181037185e-05,
      "loss": 0.1915,
      "step": 150
    },
    {
      "epoch": 0.11033468186833395,
      "eval_loss": 0.1532428115606308,
      "eval_runtime": 192.9593,
      "eval_samples_per_second": 11.868,
      "eval_steps_per_second": 5.934,
      "step": 150
    },
    {
      "epoch": 0.11107024641412284,
      "grad_norm": 0.12797152996063232,
      "learning_rate": 1.9136088935510362e-05,
      "loss": 0.1411,
      "step": 151
    },
    {
      "epoch": 0.11180581095991174,
      "grad_norm": 0.1103443130850792,
      "learning_rate": 1.8414449687337464e-05,
      "loss": 0.1444,
      "step": 152
    },
    {
      "epoch": 0.11254137550570062,
      "grad_norm": 0.08198443800210953,
      "learning_rate": 1.7703596875660645e-05,
      "loss": 0.1422,
      "step": 153
    },
    {
      "epoch": 0.11327694005148951,
      "grad_norm": 0.10905614495277405,
      "learning_rate": 1.700377325606388e-05,
      "loss": 0.1533,
      "step": 154
    },
    {
      "epoch": 0.11401250459727841,
      "grad_norm": 0.13996587693691254,
      "learning_rate": 1.631521781767214e-05,
      "loss": 0.1418,
      "step": 155
    },
    {
      "epoch": 0.1147480691430673,
      "grad_norm": 0.07795163244009018,
      "learning_rate": 1.5638165701536868e-05,
      "loss": 0.1514,
      "step": 156
    },
    {
      "epoch": 0.1154836336888562,
      "grad_norm": 0.07931917160749435,
      "learning_rate": 1.4972848120335453e-05,
      "loss": 0.1464,
      "step": 157
    },
    {
      "epoch": 0.11621919823464509,
      "grad_norm": 0.0686430111527443,
      "learning_rate": 1.4319492279412388e-05,
      "loss": 0.1472,
      "step": 158
    },
    {
      "epoch": 0.11695476278043399,
      "grad_norm": 0.083328977227211,
      "learning_rate": 1.3678321299188801e-05,
      "loss": 0.1532,
      "step": 159
    },
    {
      "epoch": 0.11769032732622288,
      "grad_norm": 0.09695445746183395,
      "learning_rate": 1.3049554138967051e-05,
      "loss": 0.1645,
      "step": 160
    },
    {
      "epoch": 0.11842589187201177,
      "grad_norm": 0.1066075786948204,
      "learning_rate": 1.2433405522156332e-05,
      "loss": 0.1596,
      "step": 161
    },
    {
      "epoch": 0.11916145641780067,
      "grad_norm": 0.08189854025840759,
      "learning_rate": 1.183008586294485e-05,
      "loss": 0.1734,
      "step": 162
    },
    {
      "epoch": 0.11989702096358955,
      "grad_norm": 0.06908301264047623,
      "learning_rate": 1.1239801194443506e-05,
      "loss": 0.1726,
      "step": 163
    },
    {
      "epoch": 0.12063258550937844,
      "grad_norm": 0.08635183423757553,
      "learning_rate": 1.066275309832584e-05,
      "loss": 0.1892,
      "step": 164
    },
    {
      "epoch": 0.12136815005516734,
      "grad_norm": 0.1024232879281044,
      "learning_rate": 1.0099138635988026e-05,
      "loss": 0.174,
      "step": 165
    },
    {
      "epoch": 0.12210371460095623,
      "grad_norm": 0.09103770554065704,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.1606,
      "step": 166
    },
    {
      "epoch": 0.12283927914674513,
      "grad_norm": 0.07270883768796921,
      "learning_rate": 9.012975854638949e-06,
      "loss": 0.1572,
      "step": 167
    },
    {
      "epoch": 0.12357484369253402,
      "grad_norm": 0.0831652358174324,
      "learning_rate": 8.490798459222476e-06,
      "loss": 0.172,
      "step": 168
    },
    {
      "epoch": 0.12431040823832291,
      "grad_norm": 0.11702089756727219,
      "learning_rate": 7.982796418105371e-06,
      "loss": 0.1667,
      "step": 169
    },
    {
      "epoch": 0.1250459727841118,
      "grad_norm": 0.08802536129951477,
      "learning_rate": 7.489143213519301e-06,
      "loss": 0.1639,
      "step": 170
    },
    {
      "epoch": 0.1257815373299007,
      "grad_norm": 0.08707013726234436,
      "learning_rate": 7.010007427581378e-06,
      "loss": 0.1817,
      "step": 171
    },
    {
      "epoch": 0.1265171018756896,
      "grad_norm": 0.0758250504732132,
      "learning_rate": 6.5455526847235825e-06,
      "loss": 0.1584,
      "step": 172
    },
    {
      "epoch": 0.12725266642147848,
      "grad_norm": 0.08350931107997894,
      "learning_rate": 6.0959375958151045e-06,
      "loss": 0.1696,
      "step": 173
    },
    {
      "epoch": 0.12798823096726739,
      "grad_norm": 0.07942811399698257,
      "learning_rate": 5.6613157039969055e-06,
      "loss": 0.1441,
      "step": 174
    },
    {
      "epoch": 0.12872379551305627,
      "grad_norm": 0.062362104654312134,
      "learning_rate": 5.241835432246889e-06,
      "loss": 0.1323,
      "step": 175
    },
    {
      "epoch": 0.12945936005884517,
      "grad_norm": 0.07514508813619614,
      "learning_rate": 4.837640032693558e-06,
      "loss": 0.1618,
      "step": 176
    },
    {
      "epoch": 0.13019492460463405,
      "grad_norm": 0.07202325016260147,
      "learning_rate": 4.448867537695578e-06,
      "loss": 0.1222,
      "step": 177
    },
    {
      "epoch": 0.13093048915042296,
      "grad_norm": 0.0819399282336235,
      "learning_rate": 4.075650712703849e-06,
      "loss": 0.1395,
      "step": 178
    },
    {
      "epoch": 0.13166605369621184,
      "grad_norm": 0.092143714427948,
      "learning_rate": 3.71811701092219e-06,
      "loss": 0.1463,
      "step": 179
    },
    {
      "epoch": 0.13240161824200072,
      "grad_norm": 0.0711476281285286,
      "learning_rate": 3.376388529782215e-06,
      "loss": 0.1356,
      "step": 180
    },
    {
      "epoch": 0.13313718278778963,
      "grad_norm": 0.06504477560520172,
      "learning_rate": 3.0505819692471792e-06,
      "loss": 0.1335,
      "step": 181
    },
    {
      "epoch": 0.1338727473335785,
      "grad_norm": 0.07897264510393143,
      "learning_rate": 2.7408085919590264e-06,
      "loss": 0.1623,
      "step": 182
    },
    {
      "epoch": 0.13460831187936742,
      "grad_norm": 0.08557404577732086,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.1515,
      "step": 183
    },
    {
      "epoch": 0.1353438764251563,
      "grad_norm": 0.06471476703882217,
      "learning_rate": 2.1697790249779636e-06,
      "loss": 0.134,
      "step": 184
    },
    {
      "epoch": 0.1360794409709452,
      "grad_norm": 0.07636171579360962,
      "learning_rate": 1.908717841359048e-06,
      "loss": 0.1438,
      "step": 185
    },
    {
      "epoch": 0.1368150055167341,
      "grad_norm": 0.05042177066206932,
      "learning_rate": 1.6640797865406288e-06,
      "loss": 0.1344,
      "step": 186
    },
    {
      "epoch": 0.137550570062523,
      "grad_norm": 0.05630411207675934,
      "learning_rate": 1.4359484041943038e-06,
      "loss": 0.1498,
      "step": 187
    },
    {
      "epoch": 0.13828613460831188,
      "grad_norm": 0.04595191031694412,
      "learning_rate": 1.2244016009781701e-06,
      "loss": 0.1455,
      "step": 188
    },
    {
      "epoch": 0.13902169915410076,
      "grad_norm": 0.07169954478740692,
      "learning_rate": 1.0295116199317057e-06,
      "loss": 0.1542,
      "step": 189
    },
    {
      "epoch": 0.13975726369988967,
      "grad_norm": 0.05570828542113304,
      "learning_rate": 8.513450158049108e-07,
      "loss": 0.1326,
      "step": 190
    },
    {
      "epoch": 0.14049282824567855,
      "grad_norm": 0.09222900122404099,
      "learning_rate": 6.899626323298713e-07,
      "loss": 0.1656,
      "step": 191
    },
    {
      "epoch": 0.14122839279146746,
      "grad_norm": 0.09059873968362808,
      "learning_rate": 5.454195814427021e-07,
      "loss": 0.1732,
      "step": 192
    },
    {
      "epoch": 0.14196395733725634,
      "grad_norm": 0.11539854854345322,
      "learning_rate": 4.177652244628627e-07,
      "loss": 0.157,
      "step": 193
    },
    {
      "epoch": 0.14269952188304524,
      "grad_norm": 0.08133802562952042,
      "learning_rate": 3.0704315523631953e-07,
      "loss": 0.1733,
      "step": 194
    },
    {
      "epoch": 0.14343508642883412,
      "grad_norm": 0.0870978981256485,
      "learning_rate": 2.1329118524827662e-07,
      "loss": 0.1564,
      "step": 195
    },
    {
      "epoch": 0.14417065097462303,
      "grad_norm": 0.07082478702068329,
      "learning_rate": 1.3654133071059893e-07,
      "loss": 0.1379,
      "step": 196
    },
    {
      "epoch": 0.1449062155204119,
      "grad_norm": 0.11655426025390625,
      "learning_rate": 7.681980162830282e-08,
      "loss": 0.1102,
      "step": 197
    },
    {
      "epoch": 0.14564178006620082,
      "grad_norm": 0.07025150209665298,
      "learning_rate": 3.4146992848854695e-08,
      "loss": 0.1203,
      "step": 198
    },
    {
      "epoch": 0.1463773446119897,
      "grad_norm": 0.08077063411474228,
      "learning_rate": 8.537477097364522e-09,
      "loss": 0.1903,
      "step": 199
    },
    {
      "epoch": 0.14711290915777858,
      "grad_norm": 0.10279268771409988,
      "learning_rate": 0.0,
      "loss": 0.194,
      "step": 200
    },
    {
      "epoch": 0.14711290915777858,
      "eval_loss": 0.15221890807151794,
      "eval_runtime": 193.1181,
      "eval_samples_per_second": 11.858,
      "eval_steps_per_second": 5.929,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.0321122932791706e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}