{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.11,
  "eval_steps": 500,
  "global_step": 11000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1e-05,
      "grad_norm": 1.631775788945859,
      "learning_rate": 3e-06,
      "loss": 10.8658,
      "step": 1
    },
    {
      "epoch": 2e-05,
      "grad_norm": 1.6173222705173065,
      "learning_rate": 6e-06,
      "loss": 10.8645,
      "step": 2
    },
    {
      "epoch": 3e-05,
      "grad_norm": 1.6387507243802044,
      "learning_rate": 9e-06,
      "loss": 10.8646,
      "step": 3
    },
    {
      "epoch": 4e-05,
      "grad_norm": 1.5975767011448438,
      "learning_rate": 1.2e-05,
      "loss": 10.8638,
      "step": 4
    },
    {
      "epoch": 5e-05,
      "grad_norm": 1.6454022013221787,
      "learning_rate": 1.5e-05,
      "loss": 10.8605,
      "step": 5
    },
    {
      "epoch": 6e-05,
      "grad_norm": 1.6407987097684302,
      "learning_rate": 1.8e-05,
      "loss": 10.8581,
      "step": 6
    },
    {
      "epoch": 7e-05,
      "grad_norm": 1.609586764602888,
      "learning_rate": 2.1000000000000002e-05,
      "loss": 10.8444,
      "step": 7
    },
    {
      "epoch": 8e-05,
      "grad_norm": 1.4683048428970586,
      "learning_rate": 2.4e-05,
      "loss": 10.8173,
      "step": 8
    },
    {
      "epoch": 9e-05,
      "grad_norm": 1.3933728304700357,
      "learning_rate": 2.7e-05,
      "loss": 10.8102,
      "step": 9
    },
    {
      "epoch": 0.0001,
      "grad_norm": 1.3326098319804733,
      "learning_rate": 3e-05,
      "loss": 10.7958,
      "step": 10
    },
    {
      "epoch": 0.00011,
      "grad_norm": 1.217237966643813,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 10.779,
      "step": 11
    },
    {
      "epoch": 0.00012,
      "grad_norm": 1.1764691634458027,
      "learning_rate": 3.6e-05,
      "loss": 10.7677,
      "step": 12
    },
    {
      "epoch": 0.00013,
      "grad_norm": 1.1304717895604097,
      "learning_rate": 3.9e-05,
      "loss": 10.7486,
      "step": 13
    },
    {
      "epoch": 0.00014,
      "grad_norm": 1.115888694899127,
      "learning_rate": 4.2000000000000004e-05,
      "loss": 10.7378,
      "step": 14
    },
    {
      "epoch": 0.00015,
      "grad_norm": 1.1065220670447153,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 10.7288,
      "step": 15
    },
    {
      "epoch": 0.00016,
      "grad_norm": 1.075226769828573,
      "learning_rate": 4.8e-05,
      "loss": 10.7115,
      "step": 16
    },
    {
      "epoch": 0.00017,
      "grad_norm": 1.0461893681391048,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 10.6956,
      "step": 17
    },
    {
      "epoch": 0.00018,
      "grad_norm": 1.0230892582928048,
      "learning_rate": 5.4e-05,
      "loss": 10.6795,
      "step": 18
    },
    {
      "epoch": 0.00019,
      "grad_norm": 0.9869910790408369,
      "learning_rate": 5.7e-05,
      "loss": 10.6637,
      "step": 19
    },
    {
      "epoch": 0.0002,
      "grad_norm": 0.9729421058544264,
      "learning_rate": 6e-05,
      "loss": 10.6515,
      "step": 20
    },
    {
      "epoch": 0.00021,
      "grad_norm": 0.943988636941647,
      "learning_rate": 6.3e-05,
      "loss": 10.6378,
      "step": 21
    },
    {
      "epoch": 0.00022,
      "grad_norm": 0.9278112602221015,
      "learning_rate": 6.599999999999999e-05,
      "loss": 10.6233,
      "step": 22
    },
    {
      "epoch": 0.00023,
      "grad_norm": 0.920877936489298,
      "learning_rate": 6.9e-05,
      "loss": 10.6091,
      "step": 23
    },
    {
      "epoch": 0.00024,
      "grad_norm": 0.9178015901962371,
      "learning_rate": 7.2e-05,
      "loss": 10.5981,
      "step": 24
    },
    {
      "epoch": 0.00025,
      "grad_norm": 0.9193853191136445,
      "learning_rate": 7.500000000000001e-05,
      "loss": 10.5835,
      "step": 25
    },
    {
      "epoch": 0.00026,
      "grad_norm": 0.9191743831022944,
      "learning_rate": 7.8e-05,
      "loss": 10.5705,
      "step": 26
    },
    {
      "epoch": 0.00027,
      "grad_norm": 0.9136913401152261,
      "learning_rate": 8.1e-05,
      "loss": 10.5585,
      "step": 27
    },
    {
      "epoch": 0.00028,
      "grad_norm": 0.9119853724531574,
      "learning_rate": 8.400000000000001e-05,
      "loss": 10.5456,
      "step": 28
    },
    {
      "epoch": 0.00029,
      "grad_norm": 0.9130829908837624,
      "learning_rate": 8.7e-05,
      "loss": 10.5312,
      "step": 29
    },
    {
      "epoch": 0.0003,
      "grad_norm": 0.9186726006674357,
      "learning_rate": 8.999999999999999e-05,
      "loss": 10.5159,
      "step": 30
    },
    {
      "epoch": 0.00031,
      "grad_norm": 0.9155120967133267,
      "learning_rate": 9.3e-05,
      "loss": 10.5012,
      "step": 31
    },
    {
      "epoch": 0.00032,
      "grad_norm": 0.9096055134642034,
      "learning_rate": 9.6e-05,
      "loss": 10.4871,
      "step": 32
    },
    {
      "epoch": 0.00033,
      "grad_norm": 0.91013596598753,
      "learning_rate": 9.900000000000001e-05,
      "loss": 10.4706,
      "step": 33
    },
    {
      "epoch": 0.00034,
      "grad_norm": 0.9103576711685224,
      "learning_rate": 0.00010200000000000001,
      "loss": 10.4543,
      "step": 34
    },
    {
      "epoch": 0.00035,
      "grad_norm": 0.9164612613814794,
      "learning_rate": 0.00010500000000000002,
      "loss": 10.4377,
      "step": 35
    },
    {
      "epoch": 0.00036,
      "grad_norm": 0.9129652960092816,
      "learning_rate": 0.000108,
      "loss": 10.4202,
      "step": 36
    },
    {
      "epoch": 0.00037,
      "grad_norm": 0.9029365420083331,
      "learning_rate": 0.000111,
      "loss": 10.4036,
      "step": 37
    },
    {
      "epoch": 0.00038,
      "grad_norm": 0.9075151789728013,
      "learning_rate": 0.000114,
      "loss": 10.3847,
      "step": 38
    },
    {
      "epoch": 0.00039,
      "grad_norm": 0.9102185085970206,
      "learning_rate": 0.000117,
      "loss": 10.3654,
      "step": 39
    },
    {
      "epoch": 0.0004,
      "grad_norm": 0.9144897951886523,
      "learning_rate": 0.00012,
      "loss": 10.3432,
      "step": 40
    },
    {
      "epoch": 0.00041,
      "grad_norm": 0.9044769084629607,
      "learning_rate": 0.000123,
      "loss": 10.3253,
      "step": 41
    },
    {
      "epoch": 0.00042,
      "grad_norm": 0.9101249031053068,
      "learning_rate": 0.000126,
      "loss": 10.3047,
      "step": 42
    },
    {
      "epoch": 0.00043,
      "grad_norm": 0.9147428510616606,
      "learning_rate": 0.000129,
      "loss": 10.282,
      "step": 43
    },
    {
      "epoch": 0.00044,
      "grad_norm": 0.9138907921510238,
      "learning_rate": 0.00013199999999999998,
      "loss": 10.2606,
      "step": 44
    },
    {
      "epoch": 0.00045,
      "grad_norm": 0.9165726299867081,
      "learning_rate": 0.000135,
      "loss": 10.2377,
      "step": 45
    },
    {
      "epoch": 0.00046,
      "grad_norm": 0.906013196308877,
      "learning_rate": 0.000138,
      "loss": 10.216,
      "step": 46
    },
    {
      "epoch": 0.00047,
      "grad_norm": 0.9116233570839986,
      "learning_rate": 0.000141,
      "loss": 10.1904,
      "step": 47
    },
    {
      "epoch": 0.00048,
      "grad_norm": 0.9086131145608887,
      "learning_rate": 0.000144,
      "loss": 10.1674,
      "step": 48
    },
    {
      "epoch": 0.00049,
      "grad_norm": 0.9094213201037699,
      "learning_rate": 0.000147,
      "loss": 10.1435,
      "step": 49
    },
    {
      "epoch": 0.0005,
      "grad_norm": 0.9128293203892458,
      "learning_rate": 0.00015000000000000001,
      "loss": 10.1173,
      "step": 50
    },
    {
      "epoch": 0.00051,
      "grad_norm": 0.9140261923858894,
      "learning_rate": 0.000153,
      "loss": 10.0916,
      "step": 51
    },
    {
      "epoch": 0.00052,
      "grad_norm": 0.9121253786029146,
      "learning_rate": 0.000156,
      "loss": 10.0657,
      "step": 52
    },
    {
      "epoch": 0.00053,
      "grad_norm": 0.9034950978450355,
      "learning_rate": 0.000159,
      "loss": 10.0413,
      "step": 53
    },
    {
      "epoch": 0.00054,
      "grad_norm": 0.9241206201798855,
      "learning_rate": 0.000162,
      "loss": 10.0092,
      "step": 54
    },
    {
      "epoch": 0.00055,
      "grad_norm": 0.9149154841379399,
      "learning_rate": 0.000165,
      "loss": 9.9852,
      "step": 55
    },
    {
      "epoch": 0.00056,
      "grad_norm": 0.9070429364957124,
      "learning_rate": 0.00016800000000000002,
      "loss": 9.9561,
      "step": 56
    },
    {
      "epoch": 0.00057,
      "grad_norm": 0.9122859536360214,
      "learning_rate": 0.000171,
      "loss": 9.9285,
      "step": 57
    },
    {
      "epoch": 0.00058,
      "grad_norm": 0.9101615769667563,
      "learning_rate": 0.000174,
      "loss": 9.9038,
      "step": 58
    },
    {
      "epoch": 0.00059,
      "grad_norm": 0.91462118557662,
      "learning_rate": 0.000177,
      "loss": 9.872,
      "step": 59
    },
    {
      "epoch": 0.0006,
      "grad_norm": 0.902238261354272,
      "learning_rate": 0.00017999999999999998,
      "loss": 9.846,
      "step": 60
    },
    {
      "epoch": 0.00061,
      "grad_norm": 0.9010263217287604,
      "learning_rate": 0.000183,
      "loss": 9.8177,
      "step": 61
    },
    {
      "epoch": 0.00062,
      "grad_norm": 0.9073977759483284,
      "learning_rate": 0.000186,
      "loss": 9.7857,
      "step": 62
    },
    {
      "epoch": 0.00063,
      "grad_norm": 0.9002035349554564,
      "learning_rate": 0.000189,
      "loss": 9.7585,
      "step": 63
    },
    {
      "epoch": 0.00064,
      "grad_norm": 0.8964472690991813,
      "learning_rate": 0.000192,
      "loss": 9.7283,
      "step": 64
    },
    {
      "epoch": 0.00065,
      "grad_norm": 0.9038015098943822,
      "learning_rate": 0.00019500000000000002,
      "loss": 9.696,
      "step": 65
    },
    {
      "epoch": 0.00066,
      "grad_norm": 0.8969444374465311,
      "learning_rate": 0.00019800000000000002,
      "loss": 9.6719,
      "step": 66
    },
    {
      "epoch": 0.00067,
      "grad_norm": 0.898249772566312,
      "learning_rate": 0.000201,
      "loss": 9.642,
      "step": 67
    },
    {
      "epoch": 0.00068,
      "grad_norm": 0.9048900549589218,
      "learning_rate": 0.00020400000000000003,
      "loss": 9.6091,
      "step": 68
    },
    {
      "epoch": 0.00069,
      "grad_norm": 0.8901553719298861,
      "learning_rate": 0.00020700000000000002,
      "loss": 9.5816,
      "step": 69
    },
    {
      "epoch": 0.0007,
      "grad_norm": 0.903386863411625,
      "learning_rate": 0.00021000000000000004,
      "loss": 9.5447,
      "step": 70
    },
    {
      "epoch": 0.00071,
      "grad_norm": 0.8933303377898588,
      "learning_rate": 0.00021299999999999997,
      "loss": 9.518,
      "step": 71
    },
    {
      "epoch": 0.00072,
      "grad_norm": 0.8969213907091691,
      "learning_rate": 0.000216,
      "loss": 9.4906,
      "step": 72
    },
    {
      "epoch": 0.00073,
      "grad_norm": 0.8960518018191052,
      "learning_rate": 0.00021899999999999998,
      "loss": 9.4566,
      "step": 73
    },
    {
      "epoch": 0.00074,
      "grad_norm": 0.8964935065687102,
      "learning_rate": 0.000222,
      "loss": 9.4297,
      "step": 74
    },
    {
      "epoch": 0.00075,
      "grad_norm": 0.8917043311520961,
      "learning_rate": 0.000225,
      "loss": 9.3947,
      "step": 75
    },
    {
      "epoch": 0.00076,
      "grad_norm": 0.8997723193205578,
      "learning_rate": 0.000228,
      "loss": 9.3676,
      "step": 76
    },
    {
      "epoch": 0.00077,
      "grad_norm": 0.8937482133829812,
      "learning_rate": 0.000231,
      "loss": 9.3393,
      "step": 77
    },
    {
      "epoch": 0.00078,
      "grad_norm": 0.886737139439046,
      "learning_rate": 0.000234,
      "loss": 9.3091,
      "step": 78
    },
    {
      "epoch": 0.00079,
      "grad_norm": 0.8895258541637481,
      "learning_rate": 0.00023700000000000001,
      "loss": 9.2733,
      "step": 79
    },
    {
      "epoch": 0.0008,
      "grad_norm": 0.8909958989870267,
      "learning_rate": 0.00024,
      "loss": 9.2384,
      "step": 80
    },
    {
      "epoch": 0.00081,
      "grad_norm": 0.8966003836256963,
      "learning_rate": 0.00024300000000000002,
      "loss": 9.2045,
      "step": 81
    },
    {
      "epoch": 0.00082,
      "grad_norm": 0.9055894629552318,
      "learning_rate": 0.000246,
      "loss": 9.1791,
      "step": 82
    },
    {
      "epoch": 0.00083,
      "grad_norm": 0.8961362604582432,
      "learning_rate": 0.00024900000000000004,
      "loss": 9.1431,
      "step": 83
    },
    {
      "epoch": 0.00084,
      "grad_norm": 0.8980147614185676,
      "learning_rate": 0.000252,
      "loss": 9.1116,
      "step": 84
    },
    {
      "epoch": 0.00085,
      "grad_norm": 0.8928314575907911,
      "learning_rate": 0.000255,
      "loss": 9.0884,
      "step": 85
    },
    {
      "epoch": 0.00086,
      "grad_norm": 0.8950046881177521,
      "learning_rate": 0.000258,
      "loss": 9.0518,
      "step": 86
    },
    {
      "epoch": 0.00087,
      "grad_norm": 0.8877891352831231,
      "learning_rate": 0.000261,
      "loss": 9.0297,
      "step": 87
    },
    {
      "epoch": 0.00088,
      "grad_norm": 0.8929677780146621,
      "learning_rate": 0.00026399999999999997,
      "loss": 8.992,
      "step": 88
    },
    {
      "epoch": 0.00089,
      "grad_norm": 0.8871031960638883,
      "learning_rate": 0.000267,
      "loss": 8.9663,
      "step": 89
    },
    {
      "epoch": 0.0009,
      "grad_norm": 0.8863614322091727,
      "learning_rate": 0.00027,
      "loss": 8.9387,
      "step": 90
    },
    {
      "epoch": 0.00091,
      "grad_norm": 0.8806256133072948,
      "learning_rate": 0.000273,
      "loss": 8.9083,
      "step": 91
    },
    {
      "epoch": 0.00092,
      "grad_norm": 0.8826316050497074,
      "learning_rate": 0.000276,
      "loss": 8.8757,
      "step": 92
    },
    {
      "epoch": 0.00093,
      "grad_norm": 0.8833603577684496,
      "learning_rate": 0.000279,
      "loss": 8.8461,
      "step": 93
    },
    {
      "epoch": 0.00094,
      "grad_norm": 0.8819538724809766,
      "learning_rate": 0.000282,
      "loss": 8.82,
      "step": 94
    },
    {
      "epoch": 0.00095,
      "grad_norm": 0.8776473753829749,
      "learning_rate": 0.000285,
      "loss": 8.7909,
      "step": 95
    },
    {
      "epoch": 0.00096,
      "grad_norm": 0.8854898687433331,
      "learning_rate": 0.000288,
      "loss": 8.7608,
      "step": 96
    },
    {
      "epoch": 0.00097,
      "grad_norm": 0.8763526561707659,
      "learning_rate": 0.000291,
      "loss": 8.7376,
      "step": 97
    },
    {
      "epoch": 0.00098,
      "grad_norm": 0.8773720509513535,
      "learning_rate": 0.000294,
      "loss": 8.7001,
      "step": 98
    },
    {
      "epoch": 0.00099,
      "grad_norm": 0.8782537783818637,
      "learning_rate": 0.000297,
      "loss": 8.6785,
      "step": 99
    },
    {
      "epoch": 0.001,
      "grad_norm": 0.876084031734807,
      "learning_rate": 0.00030000000000000003,
      "loss": 8.6509,
      "step": 100
    },
    {
      "epoch": 0.00101,
      "grad_norm": 0.8775511931302766,
      "learning_rate": 0.00030300000000000005,
      "loss": 8.6162,
      "step": 101
    },
    {
      "epoch": 0.00102,
      "grad_norm": 0.8666072056658197,
      "learning_rate": 0.000306,
      "loss": 8.5962,
      "step": 102
    },
    {
      "epoch": 0.00103,
      "grad_norm": 0.8733315070806934,
      "learning_rate": 0.000309,
      "loss": 8.5716,
      "step": 103
    },
    {
      "epoch": 0.00104,
      "grad_norm": 0.8664419648151436,
      "learning_rate": 0.000312,
      "loss": 8.5503,
      "step": 104
    },
    {
      "epoch": 0.00105,
      "grad_norm": 0.8699404252946213,
      "learning_rate": 0.000315,
      "loss": 8.5232,
      "step": 105
    },
    {
      "epoch": 0.00106,
      "grad_norm": 0.8630101619311507,
      "learning_rate": 0.000318,
      "loss": 8.4946,
      "step": 106
    },
    {
      "epoch": 0.00107,
      "grad_norm": 0.8533006978361278,
      "learning_rate": 0.000321,
      "loss": 8.4694,
      "step": 107
    },
    {
      "epoch": 0.00108,
      "grad_norm": 0.8571194919918376,
      "learning_rate": 0.000324,
      "loss": 8.4408,
      "step": 108
    },
    {
      "epoch": 0.00109,
      "grad_norm": 0.8496626885878062,
      "learning_rate": 0.000327,
      "loss": 8.4213,
      "step": 109
    },
    {
      "epoch": 0.0011,
      "grad_norm": 0.8617458268479945,
      "learning_rate": 0.00033,
      "loss": 8.3989,
      "step": 110
    },
    {
      "epoch": 0.00111,
      "grad_norm": 0.874730580725934,
      "learning_rate": 0.000333,
      "loss": 8.3705,
      "step": 111
    },
    {
      "epoch": 0.00112,
      "grad_norm": 0.9211403811176949,
      "learning_rate": 0.00033600000000000004,
      "loss": 8.351,
      "step": 112
    },
    {
      "epoch": 0.00113,
      "grad_norm": 0.9451163730301329,
      "learning_rate": 0.000339,
      "loss": 8.3126,
      "step": 113
    },
    {
      "epoch": 0.00114,
      "grad_norm": 0.8518853453666535,
      "learning_rate": 0.000342,
      "loss": 8.3025,
      "step": 114
    },
    {
      "epoch": 0.00115,
      "grad_norm": 0.8499246309464553,
      "learning_rate": 0.00034500000000000004,
      "loss": 8.2758,
      "step": 115
    },
    {
      "epoch": 0.00116,
      "grad_norm": 0.8769128820472754,
      "learning_rate": 0.000348,
      "loss": 8.2536,
      "step": 116
    },
    {
      "epoch": 0.00117,
      "grad_norm": 0.829578266212784,
      "learning_rate": 0.000351,
      "loss": 8.2211,
      "step": 117
    },
    {
      "epoch": 0.00118,
      "grad_norm": 0.8587574762862499,
      "learning_rate": 0.000354,
      "loss": 8.2068,
      "step": 118
    },
    {
      "epoch": 0.00119,
      "grad_norm": 0.8383808879241313,
      "learning_rate": 0.000357,
      "loss": 8.1942,
      "step": 119
    },
    {
      "epoch": 0.0012,
      "grad_norm": 0.8155263912622424,
      "learning_rate": 0.00035999999999999997,
      "loss": 8.1675,
      "step": 120
    },
    {
      "epoch": 0.00121,
      "grad_norm": 0.8344307084821188,
      "learning_rate": 0.000363,
      "loss": 8.1409,
      "step": 121
    },
    {
      "epoch": 0.00122,
      "grad_norm": 0.8097993043330719,
      "learning_rate": 0.000366,
      "loss": 8.1244,
      "step": 122
    },
    {
      "epoch": 0.00123,
      "grad_norm": 0.8029969793277704,
      "learning_rate": 0.000369,
      "loss": 8.102,
      "step": 123
    },
    {
      "epoch": 0.00124,
      "grad_norm": 0.7829455528112805,
      "learning_rate": 0.000372,
      "loss": 8.0811,
      "step": 124
    },
    {
      "epoch": 0.00125,
      "grad_norm": 0.8110394816024603,
      "learning_rate": 0.000375,
      "loss": 8.0581,
      "step": 125
    },
    {
      "epoch": 0.00126,
      "grad_norm": 0.8039928825408066,
      "learning_rate": 0.000378,
      "loss": 8.0463,
      "step": 126
    },
    {
      "epoch": 0.00127,
      "grad_norm": 0.8369020039958236,
      "learning_rate": 0.000381,
      "loss": 8.0243,
      "step": 127
    },
    {
      "epoch": 0.00128,
      "grad_norm": 0.9124681746054819,
      "learning_rate": 0.000384,
      "loss": 8.0062,
      "step": 128
    },
    {
      "epoch": 0.00129,
      "grad_norm": 0.9497790585671452,
      "learning_rate": 0.00038700000000000003,
      "loss": 7.997,
      "step": 129
    },
    {
      "epoch": 0.0013,
      "grad_norm": 0.895034767303024,
      "learning_rate": 0.00039000000000000005,
      "loss": 7.9709,
      "step": 130
    },
    {
      "epoch": 0.00131,
      "grad_norm": 0.7344919418773682,
      "learning_rate": 0.000393,
      "loss": 7.9348,
      "step": 131
    },
    {
      "epoch": 0.00132,
      "grad_norm": 0.8078203789615136,
      "learning_rate": 0.00039600000000000003,
      "loss": 7.9285,
      "step": 132
    },
    {
      "epoch": 0.00133,
      "grad_norm": 0.8012273324997825,
      "learning_rate": 0.00039900000000000005,
      "loss": 7.907,
      "step": 133
    },
    {
      "epoch": 0.00134,
      "grad_norm": 0.7227755159346104,
      "learning_rate": 0.000402,
      "loss": 7.8939,
      "step": 134
    },
    {
      "epoch": 0.00135,
      "grad_norm": 0.7235157354092677,
      "learning_rate": 0.00040500000000000003,
      "loss": 7.8661,
      "step": 135
    },
    {
      "epoch": 0.00136,
      "grad_norm": 0.7822895950244824,
      "learning_rate": 0.00040800000000000005,
      "loss": 7.852,
      "step": 136
    },
    {
      "epoch": 0.00137,
      "grad_norm": 0.7608062709985561,
      "learning_rate": 0.000411,
      "loss": 7.8315,
      "step": 137
    },
    {
      "epoch": 0.00138,
      "grad_norm": 0.7380360973204948,
      "learning_rate": 0.00041400000000000003,
      "loss": 7.8088,
      "step": 138
    },
    {
      "epoch": 0.00139,
      "grad_norm": 0.684851228306475,
      "learning_rate": 0.00041700000000000005,
      "loss": 7.7952,
      "step": 139
    },
    {
      "epoch": 0.0014,
      "grad_norm": 0.693462889960616,
      "learning_rate": 0.00042000000000000007,
      "loss": 7.7794,
      "step": 140
    },
    {
      "epoch": 0.00141,
      "grad_norm": 0.7360428489985282,
      "learning_rate": 0.000423,
      "loss": 7.7672,
      "step": 141
    },
    {
      "epoch": 0.00142,
      "grad_norm": 0.8088792714181905,
      "learning_rate": 0.00042599999999999995,
      "loss": 7.7572,
      "step": 142
    },
    {
      "epoch": 0.00143,
      "grad_norm": 0.8184868537412088,
      "learning_rate": 0.00042899999999999997,
      "loss": 7.7297,
      "step": 143
    },
    {
      "epoch": 0.00144,
      "grad_norm": 0.7328054873489724,
      "learning_rate": 0.000432,
      "loss": 7.7108,
      "step": 144
    },
    {
      "epoch": 0.00145,
      "grad_norm": 0.742383079085953,
      "learning_rate": 0.000435,
      "loss": 7.7067,
      "step": 145
    },
    {
      "epoch": 0.00146,
      "grad_norm": 0.8017743593965694,
      "learning_rate": 0.00043799999999999997,
      "loss": 7.6737,
      "step": 146
    },
    {
      "epoch": 0.00147,
      "grad_norm": 0.7202665590443219,
      "learning_rate": 0.000441,
      "loss": 7.6628,
      "step": 147
    },
    {
      "epoch": 0.00148,
      "grad_norm": 0.5746342281678257,
      "learning_rate": 0.000444,
      "loss": 7.6477,
      "step": 148
    },
    {
      "epoch": 0.00149,
      "grad_norm": 0.5929410187247641,
      "learning_rate": 0.00044699999999999997,
      "loss": 7.6424,
      "step": 149
    },
    {
      "epoch": 0.0015,
      "grad_norm": 0.7151318764270516,
      "learning_rate": 0.00045,
      "loss": 7.6133,
      "step": 150
    },
    {
      "epoch": 0.00151,
      "grad_norm": 0.6796514647913489,
      "learning_rate": 0.000453,
      "loss": 7.5947,
      "step": 151
    },
    {
      "epoch": 0.00152,
      "grad_norm": 0.543758313944587,
      "learning_rate": 0.000456,
      "loss": 7.5852,
      "step": 152
    },
    {
      "epoch": 0.00153,
      "grad_norm": 0.7299692440583161,
      "learning_rate": 0.000459,
      "loss": 7.5768,
      "step": 153
    },
    {
      "epoch": 0.00154,
      "grad_norm": 0.8631376359970574,
      "learning_rate": 0.000462,
      "loss": 7.5644,
      "step": 154
    },
    {
      "epoch": 0.00155,
      "grad_norm": 0.9628027146132815,
      "learning_rate": 0.000465,
      "loss": 7.546,
      "step": 155
    },
    {
      "epoch": 0.00156,
      "grad_norm": 1.6485250632015214,
      "learning_rate": 0.000468,
      "loss": 7.5336,
      "step": 156
    },
    {
      "epoch": 0.00157,
      "grad_norm": 0.9248857473926935,
      "learning_rate": 0.000471,
      "loss": 7.5216,
      "step": 157
    },
    {
      "epoch": 0.00158,
      "grad_norm": 0.7465186792591986,
      "learning_rate": 0.00047400000000000003,
      "loss": 7.4981,
      "step": 158
    },
    {
      "epoch": 0.00159,
      "grad_norm": 0.5902384572619932,
      "learning_rate": 0.000477,
      "loss": 7.4827,
      "step": 159
    },
    {
      "epoch": 0.0016,
      "grad_norm": 0.799266704152554,
      "learning_rate": 0.00048,
      "loss": 7.4675,
      "step": 160
    },
    {
      "epoch": 0.00161,
      "grad_norm": 0.7827143911710401,
      "learning_rate": 0.00048300000000000003,
      "loss": 7.4466,
      "step": 161
    },
    {
      "epoch": 0.00162,
      "grad_norm": 0.9218921339316959,
      "learning_rate": 0.00048600000000000005,
      "loss": 7.4513,
      "step": 162
    },
    {
      "epoch": 0.00163,
      "grad_norm": 0.7287660540574216,
      "learning_rate": 0.0004890000000000001,
      "loss": 7.4357,
      "step": 163
    },
    {
      "epoch": 0.00164,
      "grad_norm": 0.5250441215321361,
      "learning_rate": 0.000492,
      "loss": 7.4302,
      "step": 164
    },
    {
      "epoch": 0.00165,
      "grad_norm": 0.741808200483857,
      "learning_rate": 0.000495,
      "loss": 7.4077,
      "step": 165
    },
    {
      "epoch": 0.00166,
      "grad_norm": 0.6759515654254841,
      "learning_rate": 0.0004980000000000001,
      "loss": 7.3847,
      "step": 166
    },
    {
      "epoch": 0.00167,
      "grad_norm": 0.5877266102377413,
      "learning_rate": 0.000501,
      "loss": 7.373,
      "step": 167
    },
    {
      "epoch": 0.00168,
      "grad_norm": 0.571053515725499,
      "learning_rate": 0.000504,
      "loss": 7.3498,
      "step": 168
    },
    {
      "epoch": 0.00169,
      "grad_norm": 0.6044099115482432,
      "learning_rate": 0.0005070000000000001,
      "loss": 7.3473,
      "step": 169
    },
    {
      "epoch": 0.0017,
      "grad_norm": 0.46081682030640647,
      "learning_rate": 0.00051,
      "loss": 7.3262,
      "step": 170
    },
    {
      "epoch": 0.00171,
      "grad_norm": 0.6775750897944629,
      "learning_rate": 0.000513,
      "loss": 7.3116,
      "step": 171
    },
    {
      "epoch": 0.00172,
      "grad_norm": 0.57210862548929,
      "learning_rate": 0.000516,
      "loss": 7.3117,
      "step": 172
    },
    {
      "epoch": 0.00173,
      "grad_norm": 0.5300582190464731,
      "learning_rate": 0.0005189999999999999,
      "loss": 7.2934,
      "step": 173
    },
    {
      "epoch": 0.00174,
      "grad_norm": 0.7575900839335431,
      "learning_rate": 0.000522,
      "loss": 7.3114,
      "step": 174
    },
    {
      "epoch": 0.00175,
      "grad_norm": 0.7613961222663432,
      "learning_rate": 0.000525,
      "loss": 7.2747,
      "step": 175
    },
    {
      "epoch": 0.00176,
      "grad_norm": 0.6765759081090318,
      "learning_rate": 0.0005279999999999999,
      "loss": 7.2505,
      "step": 176
    },
    {
      "epoch": 0.00177,
      "grad_norm": 0.8933831251648804,
      "learning_rate": 0.000531,
      "loss": 7.2576,
      "step": 177
    },
    {
      "epoch": 0.00178,
      "grad_norm": 0.7076429855739662,
      "learning_rate": 0.000534,
      "loss": 7.2471,
      "step": 178
    },
    {
      "epoch": 0.00179,
      "grad_norm": 0.49163795680938555,
      "learning_rate": 0.000537,
      "loss": 7.2319,
      "step": 179
    },
    {
      "epoch": 0.0018,
      "grad_norm": 0.6284573091622804,
      "learning_rate": 0.00054,
      "loss": 7.2064,
      "step": 180
    },
    {
      "epoch": 0.00181,
      "grad_norm": 0.5715580305485367,
      "learning_rate": 0.000543,
      "loss": 7.1941,
      "step": 181
    },
    {
      "epoch": 0.00182,
      "grad_norm": 0.4282299340658738,
      "learning_rate": 0.000546,
      "loss": 7.1864,
      "step": 182
    },
    {
      "epoch": 0.00183,
      "grad_norm": 0.5948913741099119,
      "learning_rate": 0.000549,
      "loss": 7.1781,
      "step": 183
    },
    {
      "epoch": 0.00184,
      "grad_norm": 0.4755161712268706,
      "learning_rate": 0.000552,
      "loss": 7.1637,
      "step": 184
    },
    {
      "epoch": 0.00185,
      "grad_norm": 0.46412280386502286,
      "learning_rate": 0.000555,
      "loss": 7.14,
      "step": 185
    },
    {
      "epoch": 0.00186,
      "grad_norm": 0.526893652843914,
      "learning_rate": 0.000558,
      "loss": 7.1495,
      "step": 186
    },
    {
      "epoch": 0.00187,
      "grad_norm": 0.46987960377579885,
      "learning_rate": 0.000561,
      "loss": 7.1236,
      "step": 187
    },
    {
      "epoch": 0.00188,
      "grad_norm": 0.41993376006980737,
      "learning_rate": 0.000564,
      "loss": 7.1004,
      "step": 188
    },
    {
      "epoch": 0.00189,
      "grad_norm": 0.5433001024887105,
      "learning_rate": 0.000567,
      "loss": 7.0986,
      "step": 189
    },
    {
      "epoch": 0.0019,
      "grad_norm": 0.42284758794841465,
      "learning_rate": 0.00057,
      "loss": 7.0796,
      "step": 190
    },
    {
      "epoch": 0.00191,
      "grad_norm": 0.47158596225286625,
      "learning_rate": 0.000573,
      "loss": 7.0701,
      "step": 191
    },
    {
      "epoch": 0.00192,
      "grad_norm": 0.4590137866807191,
      "learning_rate": 0.000576,
      "loss": 7.0611,
      "step": 192
    },
    {
      "epoch": 0.00193,
      "grad_norm": 0.43849327433792495,
      "learning_rate": 0.000579,
      "loss": 7.0612,
      "step": 193
    },
    {
      "epoch": 0.00194,
      "grad_norm": 0.4326532382468588,
      "learning_rate": 0.000582,
      "loss": 7.0527,
      "step": 194
    },
    {
      "epoch": 0.00195,
      "grad_norm": 0.508356146068285,
      "learning_rate": 0.000585,
      "loss": 7.0348,
      "step": 195
    },
    {
      "epoch": 0.00196,
      "grad_norm": 0.534972126993015,
      "learning_rate": 0.000588,
      "loss": 7.0227,
      "step": 196
    },
    {
      "epoch": 0.00197,
      "grad_norm": 0.693386583628206,
      "learning_rate": 0.000591,
      "loss": 7.0198,
      "step": 197
    },
    {
      "epoch": 0.00198,
      "grad_norm": 1.4796610947894584,
      "learning_rate": 0.000594,
      "loss": 7.0311,
      "step": 198
    },
    {
      "epoch": 0.00199,
      "grad_norm": 0.8596271717690976,
      "learning_rate": 0.0005970000000000001,
      "loss": 7.0019,
      "step": 199
    },
    {
      "epoch": 0.002,
      "grad_norm": 0.5642900417004818,
      "learning_rate": 0.0006000000000000001,
      "loss": 6.991,
      "step": 200
    },
    {
      "epoch": 0.00201,
      "grad_norm": 0.8692961874823039,
      "learning_rate": 0.000603,
      "loss": 6.9753,
      "step": 201
    },
    {
      "epoch": 0.00202,
      "grad_norm": 0.6679575424248658,
      "learning_rate": 0.0006060000000000001,
      "loss": 6.9579,
      "step": 202
    },
    {
      "epoch": 0.00203,
      "grad_norm": 0.8457647455673973,
      "learning_rate": 0.0006090000000000001,
      "loss": 6.9683,
      "step": 203
    },
    {
      "epoch": 0.00204,
      "grad_norm": 0.9450482633193271,
      "learning_rate": 0.000612,
      "loss": 6.9556,
      "step": 204
    },
    {
      "epoch": 0.00205,
      "grad_norm": 1.3090292735766231,
      "learning_rate": 0.000615,
      "loss": 6.9523,
      "step": 205
    },
    {
      "epoch": 0.00206,
      "grad_norm": 0.6638095483572409,
      "learning_rate": 0.000618,
      "loss": 6.9303,
      "step": 206
    },
    {
      "epoch": 0.00207,
      "grad_norm": 0.4840297186697793,
      "learning_rate": 0.000621,
      "loss": 6.9174,
      "step": 207
    },
    {
      "epoch": 0.00208,
      "grad_norm": 0.7834386214905589,
      "learning_rate": 0.000624,
      "loss": 6.9192,
      "step": 208
    },
    {
      "epoch": 0.00209,
      "grad_norm": 0.69675327719343,
      "learning_rate": 0.000627,
      "loss": 6.9018,
      "step": 209
    },
    {
      "epoch": 0.0021,
      "grad_norm": 0.5517351762495105,
      "learning_rate": 0.00063,
      "loss": 6.8834,
      "step": 210
    },
    {
      "epoch": 0.00211,
      "grad_norm": 0.6866030941755482,
      "learning_rate": 0.000633,
      "loss": 6.8831,
      "step": 211
    },
    {
      "epoch": 0.00212,
      "grad_norm": 0.4815850149383259,
      "learning_rate": 0.000636,
      "loss": 6.8711,
      "step": 212
    },
    {
      "epoch": 0.00213,
      "grad_norm": 0.5445114180165818,
      "learning_rate": 0.000639,
      "loss": 6.8513,
      "step": 213
    },
    {
      "epoch": 0.00214,
      "grad_norm": 0.5893761744890885,
      "learning_rate": 0.000642,
      "loss": 6.8511,
      "step": 214
    },
    {
      "epoch": 0.00215,
      "grad_norm": 0.3989068076407255,
      "learning_rate": 0.000645,
      "loss": 6.8428,
      "step": 215
    },
    {
      "epoch": 0.00216,
      "grad_norm": 0.4674719209271809,
      "learning_rate": 0.000648,
      "loss": 6.8374,
      "step": 216
    },
    {
      "epoch": 0.00217,
      "grad_norm": 0.556887374004828,
      "learning_rate": 0.000651,
      "loss": 6.8123,
      "step": 217
    },
    {
      "epoch": 0.00218,
      "grad_norm": 0.5348764881519483,
      "learning_rate": 0.000654,
      "loss": 6.815,
      "step": 218
    },
    {
      "epoch": 0.00219,
      "grad_norm": 0.6261358280484484,
      "learning_rate": 0.000657,
      "loss": 6.7906,
      "step": 219
    },
    {
      "epoch": 0.0022,
      "grad_norm": 0.5490386646627615,
      "learning_rate": 0.00066,
      "loss": 6.8058,
      "step": 220
    },
    {
      "epoch": 0.00221,
      "grad_norm": 0.5297655781082383,
      "learning_rate": 0.0006630000000000001,
      "loss": 6.7838,
      "step": 221
    },
    {
      "epoch": 0.00222,
      "grad_norm": 0.53116953133404,
      "learning_rate": 0.000666,
      "loss": 6.7711,
      "step": 222
    },
    {
      "epoch": 0.00223,
      "grad_norm": 0.46985855580572156,
      "learning_rate": 0.000669,
      "loss": 6.7662,
      "step": 223
    },
    {
      "epoch": 0.00224,
      "grad_norm": 0.48892819667849163,
      "learning_rate": 0.0006720000000000001,
      "loss": 6.7615,
      "step": 224
    },
    {
      "epoch": 0.00225,
      "grad_norm": 0.5426443115029689,
      "learning_rate": 0.000675,
      "loss": 6.7505,
      "step": 225
    },
    {
      "epoch": 0.00226,
      "grad_norm": 0.47341745143430014,
      "learning_rate": 0.000678,
      "loss": 6.7457,
      "step": 226
    },
    {
      "epoch": 0.00227,
      "grad_norm": 0.47753897999990824,
      "learning_rate": 0.0006810000000000001,
      "loss": 6.7186,
      "step": 227
    },
    {
      "epoch": 0.00228,
      "grad_norm": 0.43835516232945165,
      "learning_rate": 0.000684,
      "loss": 6.721,
      "step": 228
    },
    {
      "epoch": 0.00229,
      "grad_norm": 0.3666587821660354,
      "learning_rate": 0.000687,
      "loss": 6.7162,
      "step": 229
    },
    {
      "epoch": 0.0023,
      "grad_norm": 0.5954344273025705,
      "learning_rate": 0.0006900000000000001,
      "loss": 6.6971,
      "step": 230
    },
    {
      "epoch": 0.00231,
      "grad_norm": 0.8324250780860898,
      "learning_rate": 0.000693,
      "loss": 6.6912,
      "step": 231
    },
    {
      "epoch": 0.00232,
      "grad_norm": 1.1082992895496584,
      "learning_rate": 0.000696,
      "loss": 6.7117,
      "step": 232
    },
    {
      "epoch": 0.00233,
      "grad_norm": 0.8989391942429391,
      "learning_rate": 0.0006990000000000001,
      "loss": 6.6931,
      "step": 233
    },
    {
      "epoch": 0.00234,
      "grad_norm": 0.8501087453831264,
      "learning_rate": 0.000702,
      "loss": 6.6816,
      "step": 234
    },
    {
      "epoch": 0.00235,
      "grad_norm": 0.9709457331919181,
      "learning_rate": 0.000705,
      "loss": 6.6715,
      "step": 235
    },
    {
      "epoch": 0.00236,
      "grad_norm": 0.8996312948341649,
      "learning_rate": 0.000708,
      "loss": 6.6542,
      "step": 236
    },
    {
      "epoch": 0.00237,
      "grad_norm": 0.7941572817187773,
      "learning_rate": 0.0007109999999999999,
      "loss": 6.6634,
      "step": 237
    },
    {
      "epoch": 0.00238,
      "grad_norm": 0.649710293154646,
      "learning_rate": 0.000714,
      "loss": 6.6483,
      "step": 238
    },
    {
      "epoch": 0.00239,
      "grad_norm": 0.7175873388046764,
      "learning_rate": 0.000717,
      "loss": 6.6317,
      "step": 239
    },
    {
      "epoch": 0.0024,
      "grad_norm": 1.0726439429102004,
      "learning_rate": 0.0007199999999999999,
      "loss": 6.6535,
      "step": 240
    },
    {
      "epoch": 0.00241,
      "grad_norm": 1.1551390926973517,
      "learning_rate": 0.000723,
      "loss": 6.6508,
      "step": 241
    },
    {
      "epoch": 0.00242,
      "grad_norm": 0.8245355038796127,
      "learning_rate": 0.000726,
      "loss": 6.615,
      "step": 242
    },
    {
      "epoch": 0.00243,
      "grad_norm": 0.7119399485811939,
      "learning_rate": 0.000729,
      "loss": 6.6026,
      "step": 243
    },
    {
      "epoch": 0.00244,
      "grad_norm": 0.6396700306701443,
      "learning_rate": 0.000732,
      "loss": 6.6042,
      "step": 244
    },
    {
      "epoch": 0.00245,
      "grad_norm": 0.668492143187707,
      "learning_rate": 0.000735,
      "loss": 6.5953,
      "step": 245
    },
    {
      "epoch": 0.00246,
      "grad_norm": 0.6209038847600604,
      "learning_rate": 0.000738,
      "loss": 6.5759,
      "step": 246
    },
    {
      "epoch": 0.00247,
      "grad_norm": 0.49470830317055475,
      "learning_rate": 0.000741,
      "loss": 6.5677,
      "step": 247
    },
    {
      "epoch": 0.00248,
      "grad_norm": 0.5745764116827149,
      "learning_rate": 0.000744,
      "loss": 6.5775,
      "step": 248
    },
    {
      "epoch": 0.00249,
      "grad_norm": 0.5319509172858093,
      "learning_rate": 0.000747,
      "loss": 6.5558,
      "step": 249
    },
    {
      "epoch": 0.0025,
      "grad_norm": 0.482084360804442,
      "learning_rate": 0.00075,
      "loss": 6.556,
      "step": 250
    },
    {
      "epoch": 0.00251,
      "grad_norm": 0.46516739275647623,
      "learning_rate": 0.000753,
      "loss": 6.521,
      "step": 251
    },
    {
      "epoch": 0.00252,
      "grad_norm": 0.4629119115625355,
      "learning_rate": 0.000756,
      "loss": 6.531,
      "step": 252
    },
    {
      "epoch": 0.00253,
      "grad_norm": 0.37719629506333596,
      "learning_rate": 0.000759,
      "loss": 6.519,
      "step": 253
    },
    {
      "epoch": 0.00254,
      "grad_norm": 0.44323602664762185,
      "learning_rate": 0.000762,
      "loss": 6.5149,
      "step": 254
    },
    {
      "epoch": 0.00255,
      "grad_norm": 0.38153047495099895,
      "learning_rate": 0.0007650000000000001,
      "loss": 6.5129,
      "step": 255
    },
    {
      "epoch": 0.00256,
      "grad_norm": 0.5270908471121704,
      "learning_rate": 0.000768,
      "loss": 6.4934,
      "step": 256
    },
    {
      "epoch": 0.00257,
      "grad_norm": 0.6201344591076082,
      "learning_rate": 0.000771,
      "loss": 6.4997,
      "step": 257
    },
    {
      "epoch": 0.00258,
      "grad_norm": 0.6391276132887356,
      "learning_rate": 0.0007740000000000001,
      "loss": 6.477,
      "step": 258
    },
    {
      "epoch": 0.00259,
      "grad_norm": 0.6374758421191778,
      "learning_rate": 0.000777,
      "loss": 6.4787,
      "step": 259
    },
    {
      "epoch": 0.0026,
      "grad_norm": 0.5597091224464362,
      "learning_rate": 0.0007800000000000001,
      "loss": 6.4607,
      "step": 260
    },
    {
      "epoch": 0.00261,
      "grad_norm": 0.587169694241395,
      "learning_rate": 0.0007830000000000001,
      "loss": 6.4722,
      "step": 261
    },
    {
      "epoch": 0.00262,
      "grad_norm": 0.6112267949829847,
      "learning_rate": 0.000786,
      "loss": 6.4511,
      "step": 262
    },
    {
      "epoch": 0.00263,
      "grad_norm": 0.5933922824160996,
      "learning_rate": 0.0007890000000000001,
      "loss": 6.4574,
      "step": 263
    },
    {
      "epoch": 0.00264,
      "grad_norm": 0.6560299493456899,
      "learning_rate": 0.0007920000000000001,
      "loss": 6.4408,
      "step": 264
    },
    {
      "epoch": 0.00265,
      "grad_norm": 0.9913628812090025,
      "learning_rate": 0.000795,
      "loss": 6.4422,
      "step": 265
    },
    {
      "epoch": 0.00266,
      "grad_norm": 1.340981155098937,
      "learning_rate": 0.0007980000000000001,
      "loss": 6.4533,
      "step": 266
    },
    {
      "epoch": 0.00267,
      "grad_norm": 0.8266116512325479,
      "learning_rate": 0.0008010000000000001,
      "loss": 6.4319,
      "step": 267
    },
    {
      "epoch": 0.00268,
      "grad_norm": 0.9896228951890642,
      "learning_rate": 0.000804,
      "loss": 6.4378,
      "step": 268
    },
    {
      "epoch": 0.00269,
      "grad_norm": 1.2352739008881923,
      "learning_rate": 0.0008070000000000001,
      "loss": 6.4279,
      "step": 269
    },
    {
      "epoch": 0.0027,
      "grad_norm": 1.1652427209458782,
      "learning_rate": 0.0008100000000000001,
      "loss": 6.4326,
      "step": 270
    },
    {
      "epoch": 0.00271,
      "grad_norm": 1.0407181933539849,
      "learning_rate": 0.000813,
      "loss": 6.4319,
      "step": 271
    },
    {
      "epoch": 0.00272,
      "grad_norm": 0.8880696455452757,
      "learning_rate": 0.0008160000000000001,
      "loss": 6.4138,
      "step": 272
    },
    {
      "epoch": 0.00273,
      "grad_norm": 0.8477724135782442,
      "learning_rate": 0.0008190000000000001,
      "loss": 6.404,
      "step": 273
    },
    {
      "epoch": 0.00274,
      "grad_norm": 0.7818547901656048,
      "learning_rate": 0.000822,
      "loss": 6.383,
      "step": 274
    },
    {
      "epoch": 0.00275,
      "grad_norm": 0.6915610404761925,
      "learning_rate": 0.0008250000000000001,
      "loss": 6.3888,
      "step": 275
    },
    {
      "epoch": 0.00276,
      "grad_norm": 0.6369714396426732,
      "learning_rate": 0.0008280000000000001,
      "loss": 6.3775,
      "step": 276
    },
    {
      "epoch": 0.00277,
      "grad_norm": 0.6792843462530734,
      "learning_rate": 0.0008310000000000001,
      "loss": 6.3726,
      "step": 277
    },
    {
      "epoch": 0.00278,
      "grad_norm": 0.6716653191335978,
      "learning_rate": 0.0008340000000000001,
      "loss": 6.3561,
      "step": 278
    },
    {
      "epoch": 0.00279,
      "grad_norm": 0.6104270551210891,
      "learning_rate": 0.0008370000000000001,
      "loss": 6.3562,
      "step": 279
    },
    {
      "epoch": 0.0028,
      "grad_norm": 0.5327216367370322,
      "learning_rate": 0.0008400000000000001,
      "loss": 6.3379,
      "step": 280
    },
    {
      "epoch": 0.00281,
      "grad_norm": 0.4495801132850456,
      "learning_rate": 0.0008430000000000001,
      "loss": 6.3253,
      "step": 281
    },
    {
      "epoch": 0.00282,
      "grad_norm": 0.4185635012011635,
      "learning_rate": 0.000846,
      "loss": 6.3254,
      "step": 282
    },
    {
      "epoch": 0.00283,
      "grad_norm": 0.41306707794715253,
      "learning_rate": 0.0008489999999999999,
      "loss": 6.3154,
      "step": 283
    },
    {
      "epoch": 0.00284,
      "grad_norm": 0.447351018324713,
      "learning_rate": 0.0008519999999999999,
      "loss": 6.3075,
      "step": 284
    },
    {
      "epoch": 0.00285,
      "grad_norm": 0.4656314656211844,
      "learning_rate": 0.000855,
      "loss": 6.3102,
      "step": 285
    },
    {
      "epoch": 0.00286,
      "grad_norm": 0.5287748664566101,
      "learning_rate": 0.0008579999999999999,
      "loss": 6.2881,
      "step": 286
    },
    {
      "epoch": 0.00287,
      "grad_norm": 0.60454227039484,
      "learning_rate": 0.000861,
      "loss": 6.2937,
      "step": 287
    },
    {
      "epoch": 0.00288,
      "grad_norm": 0.6409086244349441,
      "learning_rate": 0.000864,
      "loss": 6.2743,
      "step": 288
    },
    {
      "epoch": 0.00289,
      "grad_norm": 0.7540915605033448,
      "learning_rate": 0.0008669999999999999,
      "loss": 6.2909,
      "step": 289
    },
    {
      "epoch": 0.0029,
      "grad_norm": 0.9532532853232767,
      "learning_rate": 0.00087,
      "loss": 6.2777,
      "step": 290
    },
    {
      "epoch": 0.00291,
      "grad_norm": 0.9297633606905631,
      "learning_rate": 0.000873,
      "loss": 6.278,
      "step": 291
    },
    {
      "epoch": 0.00292,
      "grad_norm": 0.6544361038243887,
      "learning_rate": 0.0008759999999999999,
      "loss": 6.2635,
      "step": 292
    },
    {
      "epoch": 0.00293,
      "grad_norm": 0.9586546136156446,
      "learning_rate": 0.000879,
      "loss": 6.2582,
      "step": 293
    },
    {
      "epoch": 0.00294,
      "grad_norm": 0.8674924960686783,
      "learning_rate": 0.000882,
      "loss": 6.2684,
      "step": 294
    },
    {
      "epoch": 0.00295,
      "grad_norm": 0.8596325280201164,
      "learning_rate": 0.0008849999999999999,
      "loss": 6.2363,
      "step": 295
    },
    {
      "epoch": 0.00296,
      "grad_norm": 0.9927641151458286,
      "learning_rate": 0.000888,
      "loss": 6.2541,
      "step": 296
    },
    {
      "epoch": 0.00297,
      "grad_norm": 1.342485766358639,
      "learning_rate": 0.000891,
      "loss": 6.2408,
      "step": 297
    },
    {
      "epoch": 0.00298,
      "grad_norm": 1.1878316085061287,
      "learning_rate": 0.0008939999999999999,
      "loss": 6.2537,
      "step": 298
    },
    {
      "epoch": 0.00299,
      "grad_norm": 0.9496422749623654,
      "learning_rate": 0.000897,
      "loss": 6.2242,
      "step": 299
    },
    {
      "epoch": 0.003,
      "grad_norm": 1.122193003605518,
      "learning_rate": 0.0009,
      "loss": 6.2361,
      "step": 300
    },
    {
      "epoch": 0.00301,
      "grad_norm": 1.129970594986655,
      "learning_rate": 0.0009029999999999999,
      "loss": 6.2273,
      "step": 301
    },
    {
      "epoch": 0.00302,
      "grad_norm": 1.0740447263196922,
      "learning_rate": 0.000906,
      "loss": 6.2071,
      "step": 302
    },
    {
      "epoch": 0.00303,
      "grad_norm": 1.1900410452977912,
      "learning_rate": 0.000909,
      "loss": 6.2313,
      "step": 303
    },
    {
      "epoch": 0.00304,
      "grad_norm": 0.804691464481299,
      "learning_rate": 0.000912,
      "loss": 6.2111,
      "step": 304
    },
    {
      "epoch": 0.00305,
      "grad_norm": 0.7167209084416579,
      "learning_rate": 0.000915,
      "loss": 6.2106,
      "step": 305
    },
    {
      "epoch": 0.00306,
      "grad_norm": 0.5686498260282739,
      "learning_rate": 0.000918,
      "loss": 6.1897,
      "step": 306
    },
    {
      "epoch": 0.00307,
      "grad_norm": 0.5740516870647188,
      "learning_rate": 0.000921,
      "loss": 6.1847,
      "step": 307
    },
    {
      "epoch": 0.00308,
      "grad_norm": 0.5214022662741855,
      "learning_rate": 0.000924,
      "loss": 6.1668,
      "step": 308
    },
    {
      "epoch": 0.00309,
      "grad_norm": 0.489157506496739,
      "learning_rate": 0.000927,
      "loss": 6.1798,
      "step": 309
    },
    {
      "epoch": 0.0031,
      "grad_norm": 0.4872945232166538,
      "learning_rate": 0.00093,
      "loss": 6.1622,
      "step": 310
    },
    {
      "epoch": 0.00311,
      "grad_norm": 0.4909949625440354,
      "learning_rate": 0.000933,
      "loss": 6.1533,
      "step": 311
    },
    {
      "epoch": 0.00312,
      "grad_norm": 0.4186129744309998,
      "learning_rate": 0.000936,
      "loss": 6.1314,
      "step": 312
    },
    {
      "epoch": 0.00313,
      "grad_norm": 0.36050967020968366,
      "learning_rate": 0.0009390000000000001,
      "loss": 6.1442,
      "step": 313
    },
    {
      "epoch": 0.00314,
      "grad_norm": 0.3818285660239077,
      "learning_rate": 0.000942,
      "loss": 6.1495,
      "step": 314
    },
    {
      "epoch": 0.00315,
      "grad_norm": 0.42967169925093956,
      "learning_rate": 0.000945,
      "loss": 6.126,
      "step": 315
    },
    {
      "epoch": 0.00316,
      "grad_norm": 0.46511434454587514,
      "learning_rate": 0.0009480000000000001,
      "loss": 6.1004,
      "step": 316
    },
    {
      "epoch": 0.00317,
      "grad_norm": 0.5237888199450732,
      "learning_rate": 0.000951,
      "loss": 6.1201,
      "step": 317
    },
    {
      "epoch": 0.00318,
      "grad_norm": 0.6610672935792641,
      "learning_rate": 0.000954,
      "loss": 6.1161,
      "step": 318
    },
    {
      "epoch": 0.00319,
      "grad_norm": 0.9099491192879063,
      "learning_rate": 0.0009570000000000001,
      "loss": 6.1122,
      "step": 319
    },
    {
      "epoch": 0.0032,
      "grad_norm": 1.0329714723925014,
      "learning_rate": 0.00096,
      "loss": 6.1198,
      "step": 320
    },
    {
      "epoch": 0.00321,
      "grad_norm": 0.9944509511152075,
      "learning_rate": 0.000963,
      "loss": 6.1077,
      "step": 321
    },
    {
      "epoch": 0.00322,
      "grad_norm": 1.41191394849347,
      "learning_rate": 0.0009660000000000001,
      "loss": 6.1101,
      "step": 322
    },
    {
      "epoch": 0.00323,
      "grad_norm": 0.9172707652477707,
      "learning_rate": 0.000969,
      "loss": 6.097,
      "step": 323
    },
    {
      "epoch": 0.00324,
      "grad_norm": 1.1261423941310122,
      "learning_rate": 0.0009720000000000001,
      "loss": 6.1132,
      "step": 324
    },
    {
      "epoch": 0.00325,
      "grad_norm": 0.6623590740718236,
      "learning_rate": 0.0009750000000000001,
      "loss": 6.0626,
      "step": 325
    },
    {
      "epoch": 0.00326,
      "grad_norm": 0.7364273563271467,
      "learning_rate": 0.0009780000000000001,
      "loss": 6.0809,
      "step": 326
    },
    {
      "epoch": 0.00327,
      "grad_norm": 0.8106867198528734,
      "learning_rate": 0.000981,
      "loss": 6.0853,
      "step": 327
    },
    {
      "epoch": 0.00328,
      "grad_norm": 0.9008187294951384,
      "learning_rate": 0.000984,
      "loss": 6.0637,
      "step": 328
    },
    {
      "epoch": 0.00329,
      "grad_norm": 1.0311811602663732,
      "learning_rate": 0.000987,
      "loss": 6.0736,
      "step": 329
    },
    {
      "epoch": 0.0033,
      "grad_norm": 0.7413155368855245,
      "learning_rate": 0.00099,
      "loss": 6.0572,
      "step": 330
    },
    {
      "epoch": 0.00331,
      "grad_norm": 0.6745658849207387,
      "learning_rate": 0.0009930000000000002,
      "loss": 6.0599,
      "step": 331
    },
    {
      "epoch": 0.00332,
      "grad_norm": 0.5913240343902441,
      "learning_rate": 0.0009960000000000001,
      "loss": 6.041,
      "step": 332
    },
    {
      "epoch": 0.00333,
      "grad_norm": 0.5668749800176679,
      "learning_rate": 0.000999,
      "loss": 6.025,
      "step": 333
    },
    {
      "epoch": 0.00334,
      "grad_norm": 0.5007608052342689,
      "learning_rate": 0.001002,
      "loss": 6.0336,
      "step": 334
    },
    {
      "epoch": 0.00335,
      "grad_norm": 0.3983861566645405,
      "learning_rate": 0.001005,
      "loss": 6.0284,
      "step": 335
    },
    {
      "epoch": 0.00336,
      "grad_norm": 0.4274260388302738,
      "learning_rate": 0.001008,
      "loss": 6.0181,
      "step": 336
    },
    {
      "epoch": 0.00337,
      "grad_norm": 0.5335498119421307,
      "learning_rate": 0.0010110000000000002,
      "loss": 6.0004,
      "step": 337
    },
    {
      "epoch": 0.00338,
      "grad_norm": 0.4831054617031532,
      "learning_rate": 0.0010140000000000001,
      "loss": 6.0126,
      "step": 338
    },
    {
      "epoch": 0.00339,
      "grad_norm": 0.5437142240239534,
      "learning_rate": 0.0010170000000000001,
      "loss": 6.0197,
      "step": 339
    },
    {
      "epoch": 0.0034,
      "grad_norm": 0.5265472037464025,
      "learning_rate": 0.00102,
      "loss": 5.9884,
      "step": 340
    },
    {
      "epoch": 0.00341,
      "grad_norm": 0.4912689865797111,
      "learning_rate": 0.001023,
      "loss": 5.9692,
      "step": 341
    },
    {
      "epoch": 0.00342,
      "grad_norm": 0.42997171211054086,
      "learning_rate": 0.001026,
      "loss": 5.974,
      "step": 342
    },
    {
      "epoch": 0.00343,
      "grad_norm": 0.5197303002983154,
      "learning_rate": 0.0010290000000000002,
      "loss": 5.9763,
      "step": 343
    },
    {
      "epoch": 0.00344,
      "grad_norm": 0.8117900330313431,
      "learning_rate": 0.001032,
      "loss": 5.9747,
      "step": 344
    },
    {
      "epoch": 0.00345,
      "grad_norm": 1.1753813945983669,
      "learning_rate": 0.001035,
      "loss": 5.9788,
      "step": 345
    },
    {
      "epoch": 0.00346,
      "grad_norm": 0.8135676081857764,
      "learning_rate": 0.0010379999999999999,
      "loss": 5.9748,
      "step": 346
    },
    {
      "epoch": 0.00347,
      "grad_norm": 1.0182912247404574,
      "learning_rate": 0.001041,
      "loss": 5.9557,
      "step": 347
    },
    {
      "epoch": 0.00348,
      "grad_norm": 1.1407576555241683,
      "learning_rate": 0.001044,
      "loss": 5.978,
      "step": 348
    },
    {
      "epoch": 0.00349,
      "grad_norm": 0.7853068136249622,
      "learning_rate": 0.001047,
      "loss": 5.9412,
      "step": 349
    },
    {
      "epoch": 0.0035,
      "grad_norm": 1.0427704318540805,
      "learning_rate": 0.00105,
      "loss": 5.9779,
      "step": 350
    },
    {
      "epoch": 0.00351,
      "grad_norm": 0.8821399606009466,
      "learning_rate": 0.001053,
      "loss": 5.9701,
      "step": 351
    },
    {
      "epoch": 0.00352,
      "grad_norm": 0.9582157894617032,
      "learning_rate": 0.0010559999999999999,
      "loss": 5.955,
      "step": 352
    },
    {
      "epoch": 0.00353,
      "grad_norm": 1.0526665256553966,
      "learning_rate": 0.001059,
      "loss": 5.958,
      "step": 353
    },
    {
      "epoch": 0.00354,
      "grad_norm": 1.045275747166985,
      "learning_rate": 0.001062,
      "loss": 5.9353,
      "step": 354
    },
    {
      "epoch": 0.00355,
      "grad_norm": 1.1505195376317356,
      "learning_rate": 0.001065,
      "loss": 5.9542,
      "step": 355
    },
    {
      "epoch": 0.00356,
      "grad_norm": 1.0355197503433216,
      "learning_rate": 0.001068,
      "loss": 5.9425,
      "step": 356
    },
    {
      "epoch": 0.00357,
      "grad_norm": 1.127934795973434,
      "learning_rate": 0.001071,
      "loss": 5.9396,
      "step": 357
    },
    {
      "epoch": 0.00358,
      "grad_norm": 0.7430425972675007,
      "learning_rate": 0.001074,
      "loss": 5.9201,
      "step": 358
    },
    {
      "epoch": 0.00359,
      "grad_norm": 0.6597065121039739,
      "learning_rate": 0.001077,
      "loss": 5.9099,
      "step": 359
    },
    {
      "epoch": 0.0036,
      "grad_norm": 0.6034653307534226,
      "learning_rate": 0.00108,
      "loss": 5.9081,
      "step": 360
    },
    {
      "epoch": 0.00361,
      "grad_norm": 0.5960018023982208,
      "learning_rate": 0.001083,
      "loss": 5.9047,
      "step": 361
    },
    {
      "epoch": 0.00362,
      "grad_norm": 0.4563634882449727,
      "learning_rate": 0.001086,
      "loss": 5.884,
      "step": 362
    },
    {
      "epoch": 0.00363,
      "grad_norm": 0.49274399902142996,
      "learning_rate": 0.001089,
      "loss": 5.8759,
      "step": 363
    },
    {
      "epoch": 0.00364,
      "grad_norm": 0.4937234603270663,
      "learning_rate": 0.001092,
      "loss": 5.8901,
      "step": 364
    },
    {
      "epoch": 0.00365,
      "grad_norm": 0.5102012627619638,
      "learning_rate": 0.001095,
      "loss": 5.888,
      "step": 365
    },
    {
      "epoch": 0.00366,
      "grad_norm": 0.4676595798467989,
      "learning_rate": 0.001098,
      "loss": 5.862,
      "step": 366
    },
    {
      "epoch": 0.00367,
      "grad_norm": 0.49526135096535867,
      "learning_rate": 0.001101,
      "loss": 5.8667,
      "step": 367
    },
    {
      "epoch": 0.00368,
      "grad_norm": 0.47887378181150303,
      "learning_rate": 0.001104,
      "loss": 5.8643,
      "step": 368
    },
    {
      "epoch": 0.00369,
      "grad_norm": 0.48887117156741833,
      "learning_rate": 0.001107,
      "loss": 5.8686,
      "step": 369
    },
    {
      "epoch": 0.0037,
      "grad_norm": 0.4473709149836047,
      "learning_rate": 0.00111,
      "loss": 5.8472,
      "step": 370
    },
    {
      "epoch": 0.00371,
      "grad_norm": 0.38589559577094035,
      "learning_rate": 0.001113,
      "loss": 5.8158,
      "step": 371
    },
    {
      "epoch": 0.00372,
      "grad_norm": 0.3912315505838062,
      "learning_rate": 0.001116,
      "loss": 5.8379,
      "step": 372
    },
    {
      "epoch": 0.00373,
      "grad_norm": 0.38616823047071297,
      "learning_rate": 0.001119,
      "loss": 5.8267,
      "step": 373
    },
    {
      "epoch": 0.00374,
      "grad_norm": 0.45854090440574513,
      "learning_rate": 0.001122,
      "loss": 5.8316,
      "step": 374
    },
    {
      "epoch": 0.00375,
      "grad_norm": 0.5169440196993219,
      "learning_rate": 0.0011250000000000001,
      "loss": 5.8332,
      "step": 375
    },
    {
      "epoch": 0.00376,
      "grad_norm": 0.5067806568705457,
      "learning_rate": 0.001128,
      "loss": 5.8287,
      "step": 376
    },
    {
      "epoch": 0.00377,
      "grad_norm": 0.48558945502532774,
      "learning_rate": 0.001131,
      "loss": 5.8236,
      "step": 377
    },
    {
      "epoch": 0.00378,
      "grad_norm": 0.47384141098896654,
      "learning_rate": 0.001134,
      "loss": 5.8187,
      "step": 378
    },
    {
      "epoch": 0.00379,
      "grad_norm": 0.5705731390544022,
      "learning_rate": 0.001137,
      "loss": 5.8065,
      "step": 379
    },
    {
      "epoch": 0.0038,
      "grad_norm": 0.8415616570321116,
      "learning_rate": 0.00114,
      "loss": 5.8323,
      "step": 380
    },
    {
      "epoch": 0.00381,
      "grad_norm": 1.152388235651458,
      "learning_rate": 0.0011430000000000001,
      "loss": 5.8155,
      "step": 381
    },
    {
      "epoch": 0.00382,
      "grad_norm": 0.7784536663385624,
      "learning_rate": 0.001146,
      "loss": 5.7896,
      "step": 382
    },
    {
      "epoch": 0.00383,
      "grad_norm": 1.2096458575940454,
      "learning_rate": 0.001149,
      "loss": 5.8132,
      "step": 383
    },
    {
      "epoch": 0.00384,
      "grad_norm": 1.2032626959449177,
      "learning_rate": 0.001152,
      "loss": 5.8295,
      "step": 384
    },
    {
      "epoch": 0.00385,
      "grad_norm": 1.2258405640081835,
      "learning_rate": 0.001155,
      "loss": 5.8193,
      "step": 385
    },
    {
      "epoch": 0.00386,
      "grad_norm": 1.060557976067675,
      "learning_rate": 0.001158,
      "loss": 5.8182,
      "step": 386
    },
    {
      "epoch": 0.00387,
      "grad_norm": 1.6852101829047932,
      "learning_rate": 0.0011610000000000001,
      "loss": 5.8306,
      "step": 387
    },
    {
      "epoch": 0.00388,
      "grad_norm": 0.7125426173667109,
      "learning_rate": 0.001164,
      "loss": 5.7875,
      "step": 388
    },
    {
      "epoch": 0.00389,
      "grad_norm": 0.9333298966305301,
      "learning_rate": 0.001167,
      "loss": 5.8092,
      "step": 389
    },
    {
      "epoch": 0.0039,
      "grad_norm": 0.7871116791575423,
      "learning_rate": 0.00117,
      "loss": 5.7842,
      "step": 390
    },
    {
      "epoch": 0.00391,
      "grad_norm": 0.9033950769229127,
      "learning_rate": 0.001173,
      "loss": 5.7945,
      "step": 391
    },
    {
      "epoch": 0.00392,
      "grad_norm": 1.0985861295177402,
      "learning_rate": 0.001176,
      "loss": 5.8091,
      "step": 392
    },
    {
      "epoch": 0.00393,
      "grad_norm": 0.9893983760666882,
      "learning_rate": 0.0011790000000000001,
      "loss": 5.787,
      "step": 393
    },
    {
      "epoch": 0.00394,
      "grad_norm": 1.0087630537900902,
      "learning_rate": 0.001182,
      "loss": 5.7718,
      "step": 394
    },
    {
      "epoch": 0.00395,
      "grad_norm": 0.9357634093540522,
      "learning_rate": 0.001185,
      "loss": 5.7577,
      "step": 395
    },
    {
      "epoch": 0.00396,
      "grad_norm": 0.8613606742928634,
      "learning_rate": 0.001188,
      "loss": 5.7674,
      "step": 396
    },
    {
      "epoch": 0.00397,
      "grad_norm": 0.9393680367248612,
      "learning_rate": 0.001191,
      "loss": 5.7666,
      "step": 397
    },
    {
      "epoch": 0.00398,
      "grad_norm": 0.8380984764873387,
      "learning_rate": 0.0011940000000000002,
      "loss": 5.7669,
      "step": 398
    },
    {
      "epoch": 0.00399,
      "grad_norm": 0.7495495962771003,
      "learning_rate": 0.0011970000000000001,
      "loss": 5.7689,
      "step": 399
    },
    {
      "epoch": 0.004,
      "grad_norm": 0.6237821646680863,
      "learning_rate": 0.0012000000000000001,
      "loss": 5.751,
      "step": 400
    },
    {
      "epoch": 0.00401,
      "grad_norm": 0.6042562364668606,
      "learning_rate": 0.001203,
      "loss": 5.7286,
      "step": 401
    },
    {
      "epoch": 0.00402,
      "grad_norm": 0.6800421237430357,
      "learning_rate": 0.001206,
      "loss": 5.7387,
      "step": 402
    },
    {
      "epoch": 0.00403,
      "grad_norm": 0.5349967773183291,
      "learning_rate": 0.001209,
      "loss": 5.7296,
      "step": 403
    },
    {
      "epoch": 0.00404,
      "grad_norm": 0.4491885962138907,
      "learning_rate": 0.0012120000000000002,
      "loss": 5.7204,
      "step": 404
    },
    {
      "epoch": 0.00405,
      "grad_norm": 0.4231295613571548,
      "learning_rate": 0.0012150000000000002,
      "loss": 5.7133,
      "step": 405
    },
    {
      "epoch": 0.00406,
      "grad_norm": 0.4344671315280792,
      "learning_rate": 0.0012180000000000001,
      "loss": 5.7007,
      "step": 406
    },
    {
      "epoch": 0.00407,
      "grad_norm": 0.3917681066216858,
      "learning_rate": 0.0012209999999999999,
      "loss": 5.6947,
      "step": 407
    },
    {
      "epoch": 0.00408,
      "grad_norm": 0.3526446903520604,
      "learning_rate": 0.001224,
      "loss": 5.7113,
      "step": 408
    },
    {
      "epoch": 0.00409,
      "grad_norm": 0.3601774169994176,
      "learning_rate": 0.001227,
      "loss": 5.689,
      "step": 409
    },
    {
      "epoch": 0.0041,
      "grad_norm": 0.3999316895065895,
      "learning_rate": 0.00123,
      "loss": 5.6821,
      "step": 410
    },
    {
      "epoch": 0.00411,
      "grad_norm": 0.5215468120681382,
      "learning_rate": 0.001233,
      "loss": 5.6975,
      "step": 411
    },
    {
      "epoch": 0.00412,
      "grad_norm": 0.8731356141140694,
      "learning_rate": 0.001236,
      "loss": 5.697,
      "step": 412
    },
    {
      "epoch": 0.00413,
      "grad_norm": 1.2920248463477522,
      "learning_rate": 0.0012389999999999999,
      "loss": 5.7158,
      "step": 413
    },
    {
      "epoch": 0.00414,
      "grad_norm": 0.7474803494460109,
      "learning_rate": 0.001242,
      "loss": 5.6771,
      "step": 414
    },
    {
      "epoch": 0.00415,
      "grad_norm": 0.9736431117993121,
      "learning_rate": 0.001245,
      "loss": 5.6888,
      "step": 415
    },
    {
      "epoch": 0.00416,
      "grad_norm": 0.8653333577780613,
      "learning_rate": 0.001248,
      "loss": 5.6728,
      "step": 416
    },
    {
      "epoch": 0.00417,
      "grad_norm": 0.6891363999339204,
      "learning_rate": 0.001251,
      "loss": 5.695,
      "step": 417
    },
    {
      "epoch": 0.00418,
      "grad_norm": 0.7955125411502495,
      "learning_rate": 0.001254,
      "loss": 5.6734,
      "step": 418
    },
    {
      "epoch": 0.00419,
      "grad_norm": 0.8034523576562718,
      "learning_rate": 0.0012569999999999999,
      "loss": 5.6601,
      "step": 419
    },
    {
      "epoch": 0.0042,
      "grad_norm": 0.7731586474207807,
      "learning_rate": 0.00126,
      "loss": 5.6748,
      "step": 420
    },
    {
      "epoch": 0.00421,
      "grad_norm": 0.747486262420627,
      "learning_rate": 0.001263,
      "loss": 5.6666,
      "step": 421
    },
    {
      "epoch": 0.00422,
      "grad_norm": 0.7917981329409665,
      "learning_rate": 0.001266,
      "loss": 5.6544,
      "step": 422
    },
    {
      "epoch": 0.00423,
      "grad_norm": 1.0889555078416353,
      "learning_rate": 0.001269,
      "loss": 5.6655,
      "step": 423
    },
    {
      "epoch": 0.00424,
      "grad_norm": 0.9654337501414605,
      "learning_rate": 0.001272,
      "loss": 5.6614,
      "step": 424
    },
    {
      "epoch": 0.00425,
      "grad_norm": 0.9055610792467201,
      "learning_rate": 0.001275,
      "loss": 5.6624,
      "step": 425
    },
    {
      "epoch": 0.00426,
      "grad_norm": 0.8212981627676188,
      "learning_rate": 0.001278,
      "loss": 5.6637,
      "step": 426
    },
    {
      "epoch": 0.00427,
      "grad_norm": 0.7602647353125763,
      "learning_rate": 0.001281,
      "loss": 5.6467,
      "step": 427
    },
    {
      "epoch": 0.00428,
      "grad_norm": 0.677777266675102,
      "learning_rate": 0.001284,
      "loss": 5.6204,
      "step": 428
    },
    {
      "epoch": 0.00429,
      "grad_norm": 0.5947091658499406,
      "learning_rate": 0.001287,
      "loss": 5.6311,
      "step": 429
    },
    {
      "epoch": 0.0043,
      "grad_norm": 0.6377204770277832,
      "learning_rate": 0.00129,
      "loss": 5.6309,
      "step": 430
    },
    {
      "epoch": 0.00431,
      "grad_norm": 0.6897719280155576,
      "learning_rate": 0.001293,
      "loss": 5.6193,
      "step": 431
    },
    {
      "epoch": 0.00432,
      "grad_norm": 0.6884919593361081,
      "learning_rate": 0.001296,
      "loss": 5.6258,
      "step": 432
    },
    {
      "epoch": 0.00433,
      "grad_norm": 0.6913913571918432,
      "learning_rate": 0.001299,
      "loss": 5.6177,
      "step": 433
    },
    {
      "epoch": 0.00434,
      "grad_norm": 0.7261280979587743,
      "learning_rate": 0.001302,
      "loss": 5.6176,
      "step": 434
    },
    {
      "epoch": 0.00435,
      "grad_norm": 0.8547702731757605,
      "learning_rate": 0.001305,
      "loss": 5.6162,
      "step": 435
    },
    {
      "epoch": 0.00436,
      "grad_norm": 0.9457491419795808,
      "learning_rate": 0.001308,
      "loss": 5.5986,
      "step": 436
    },
    {
      "epoch": 0.00437,
      "grad_norm": 0.9092672289397813,
      "learning_rate": 0.001311,
      "loss": 5.6144,
      "step": 437
    },
    {
      "epoch": 0.00438,
      "grad_norm": 0.9049337850080227,
      "learning_rate": 0.001314,
      "loss": 5.6026,
      "step": 438
    },
    {
      "epoch": 0.00439,
      "grad_norm": 0.7237349559204094,
      "learning_rate": 0.001317,
      "loss": 5.622,
      "step": 439
    },
    {
      "epoch": 0.0044,
      "grad_norm": 0.8693791239531735,
      "learning_rate": 0.00132,
      "loss": 5.6008,
      "step": 440
    },
    {
      "epoch": 0.00441,
      "grad_norm": 0.8508838859779835,
      "learning_rate": 0.001323,
      "loss": 5.5985,
      "step": 441
    },
    {
      "epoch": 0.00442,
      "grad_norm": 0.6987140204651114,
      "learning_rate": 0.0013260000000000001,
      "loss": 5.587,
      "step": 442
    },
    {
      "epoch": 0.00443,
      "grad_norm": 0.5719525863328404,
      "learning_rate": 0.001329,
      "loss": 5.5843,
      "step": 443
    },
    {
      "epoch": 0.00444,
      "grad_norm": 0.5407139364493208,
      "learning_rate": 0.001332,
      "loss": 5.5841,
      "step": 444
    },
    {
      "epoch": 0.00445,
      "grad_norm": 0.5170533332401992,
      "learning_rate": 0.001335,
      "loss": 5.5667,
      "step": 445
    },
    {
      "epoch": 0.00446,
      "grad_norm": 0.43806698904849195,
      "learning_rate": 0.001338,
      "loss": 5.5666,
      "step": 446
    },
    {
      "epoch": 0.00447,
      "grad_norm": 0.49048920433285326,
      "learning_rate": 0.001341,
      "loss": 5.5671,
      "step": 447
    },
    {
      "epoch": 0.00448,
      "grad_norm": 0.46215050883864656,
      "learning_rate": 0.0013440000000000001,
      "loss": 5.5475,
      "step": 448
    },
    {
      "epoch": 0.00449,
      "grad_norm": 0.5259389714982564,
      "learning_rate": 0.001347,
      "loss": 5.5523,
      "step": 449
    },
    {
      "epoch": 0.0045,
      "grad_norm": 0.6261840891481112,
      "learning_rate": 0.00135,
      "loss": 5.5542,
      "step": 450
    },
    {
      "epoch": 0.00451,
      "grad_norm": 0.659112753346069,
      "learning_rate": 0.001353,
      "loss": 5.5431,
      "step": 451
    },
    {
      "epoch": 0.00452,
      "grad_norm": 0.6800042712218282,
      "learning_rate": 0.001356,
      "loss": 5.5311,
      "step": 452
    },
    {
      "epoch": 0.00453,
      "grad_norm": 0.5745322110996829,
      "learning_rate": 0.001359,
      "loss": 5.5188,
      "step": 453
    },
    {
      "epoch": 0.00454,
      "grad_norm": 0.5230006416449293,
      "learning_rate": 0.0013620000000000001,
      "loss": 5.5319,
      "step": 454
    },
    {
      "epoch": 0.00455,
      "grad_norm": 0.5858648520183006,
      "learning_rate": 0.0013650000000000001,
      "loss": 5.5314,
      "step": 455
    },
    {
      "epoch": 0.00456,
      "grad_norm": 0.5800568870191161,
      "learning_rate": 0.001368,
      "loss": 5.5269,
      "step": 456
    },
    {
      "epoch": 0.00457,
      "grad_norm": 0.5535894536098482,
      "learning_rate": 0.001371,
      "loss": 5.5162,
      "step": 457
    },
    {
      "epoch": 0.00458,
      "grad_norm": 0.56095957103827,
      "learning_rate": 0.001374,
      "loss": 5.5224,
      "step": 458
    },
    {
      "epoch": 0.00459,
      "grad_norm": 0.7492551931077938,
      "learning_rate": 0.0013770000000000002,
      "loss": 5.517,
      "step": 459
    },
    {
      "epoch": 0.0046,
      "grad_norm": 0.8251083608050601,
      "learning_rate": 0.0013800000000000002,
      "loss": 5.5084,
      "step": 460
    },
    {
      "epoch": 0.00461,
      "grad_norm": 0.7810512714683711,
      "learning_rate": 0.0013830000000000001,
      "loss": 5.5038,
      "step": 461
    },
    {
      "epoch": 0.00462,
      "grad_norm": 0.8065032793416945,
      "learning_rate": 0.001386,
      "loss": 5.5174,
      "step": 462
    },
    {
      "epoch": 0.00463,
      "grad_norm": 1.0894770209329594,
      "learning_rate": 0.001389,
      "loss": 5.5013,
      "step": 463
    },
    {
      "epoch": 0.00464,
      "grad_norm": 1.3225439160647088,
      "learning_rate": 0.001392,
      "loss": 5.5348,
      "step": 464
    },
    {
      "epoch": 0.00465,
      "grad_norm": 1.0604196603169807,
      "learning_rate": 0.0013950000000000002,
      "loss": 5.4973,
      "step": 465
    },
    {
      "epoch": 0.00466,
      "grad_norm": 1.279638390325454,
      "learning_rate": 0.0013980000000000002,
      "loss": 5.5365,
      "step": 466
    },
    {
      "epoch": 0.00467,
      "grad_norm": 0.8376473861337255,
      "learning_rate": 0.0014010000000000001,
      "loss": 5.5034,
      "step": 467
    },
    {
      "epoch": 0.00468,
      "grad_norm": 0.8787509291075707,
      "learning_rate": 0.001404,
      "loss": 5.4981,
      "step": 468
    },
    {
      "epoch": 0.00469,
      "grad_norm": 0.8315105482947757,
      "learning_rate": 0.001407,
      "loss": 5.4995,
      "step": 469
    },
    {
      "epoch": 0.0047,
      "grad_norm": 0.900233910053011,
      "learning_rate": 0.00141,
      "loss": 5.5098,
      "step": 470
    },
    {
      "epoch": 0.00471,
      "grad_norm": 1.1782268624389831,
      "learning_rate": 0.001413,
      "loss": 5.5031,
      "step": 471
    },
    {
      "epoch": 0.00472,
      "grad_norm": 0.8433457613569132,
      "learning_rate": 0.001416,
      "loss": 5.4989,
      "step": 472
    },
    {
      "epoch": 0.00473,
      "grad_norm": 0.8984284318795871,
      "learning_rate": 0.001419,
      "loss": 5.5107,
      "step": 473
    },
    {
      "epoch": 0.00474,
      "grad_norm": 0.8057807296189134,
      "learning_rate": 0.0014219999999999999,
      "loss": 5.4892,
      "step": 474
    },
    {
      "epoch": 0.00475,
      "grad_norm": 0.8485388443307728,
      "learning_rate": 0.001425,
      "loss": 5.4826,
      "step": 475
    },
    {
      "epoch": 0.00476,
      "grad_norm": 0.9809665505076786,
      "learning_rate": 0.001428,
      "loss": 5.5192,
      "step": 476
    },
    {
      "epoch": 0.00477,
      "grad_norm": 0.9686040040277449,
      "learning_rate": 0.001431,
      "loss": 5.4785,
      "step": 477
    },
    {
      "epoch": 0.00478,
      "grad_norm": 0.8580634771679295,
      "learning_rate": 0.001434,
      "loss": 5.4949,
      "step": 478
    },
    {
      "epoch": 0.00479,
      "grad_norm": 0.9699411511566143,
      "learning_rate": 0.001437,
      "loss": 5.4782,
      "step": 479
    },
    {
      "epoch": 0.0048,
      "grad_norm": 0.8190893004419723,
      "learning_rate": 0.0014399999999999999,
      "loss": 5.4711,
      "step": 480
    },
    {
      "epoch": 0.00481,
      "grad_norm": 0.7019568417012634,
      "learning_rate": 0.001443,
      "loss": 5.4711,
      "step": 481
    },
    {
      "epoch": 0.00482,
      "grad_norm": 0.6677085458766991,
      "learning_rate": 0.001446,
      "loss": 5.4413,
      "step": 482
    },
    {
      "epoch": 0.00483,
      "grad_norm": 0.6622223138809283,
      "learning_rate": 0.001449,
      "loss": 5.4499,
      "step": 483
    },
    {
      "epoch": 0.00484,
      "grad_norm": 0.6831178312322733,
      "learning_rate": 0.001452,
      "loss": 5.4529,
      "step": 484
    },
    {
      "epoch": 0.00485,
      "grad_norm": 0.6786720110326826,
      "learning_rate": 0.001455,
      "loss": 5.4548,
      "step": 485
    },
    {
      "epoch": 0.00486,
      "grad_norm": 0.639008514866701,
      "learning_rate": 0.001458,
      "loss": 5.4237,
      "step": 486
    },
    {
      "epoch": 0.00487,
      "grad_norm": 0.6663550617928226,
      "learning_rate": 0.001461,
      "loss": 5.4384,
      "step": 487
    },
    {
      "epoch": 0.00488,
      "grad_norm": 0.5492133897414355,
      "learning_rate": 0.001464,
      "loss": 5.4132,
      "step": 488
    },
    {
      "epoch": 0.00489,
      "grad_norm": 0.5801507624750007,
      "learning_rate": 0.001467,
      "loss": 5.4021,
      "step": 489
    },
    {
      "epoch": 0.0049,
      "grad_norm": 0.7369258331072537,
      "learning_rate": 0.00147,
      "loss": 5.4206,
      "step": 490
    },
    {
      "epoch": 0.00491,
      "grad_norm": 0.8149385883601376,
      "learning_rate": 0.001473,
      "loss": 5.4211,
      "step": 491
    },
    {
      "epoch": 0.00492,
      "grad_norm": 0.7605903135404127,
      "learning_rate": 0.001476,
      "loss": 5.4167,
      "step": 492
    },
    {
      "epoch": 0.00493,
      "grad_norm": 0.5930344173404182,
      "learning_rate": 0.001479,
      "loss": 5.4047,
      "step": 493
    },
    {
      "epoch": 0.00494,
      "grad_norm": 0.7636413305061132,
      "learning_rate": 0.001482,
      "loss": 5.4167,
      "step": 494
    },
    {
      "epoch": 0.00495,
      "grad_norm": 0.7369014833667976,
      "learning_rate": 0.001485,
      "loss": 5.4191,
      "step": 495
    },
    {
      "epoch": 0.00496,
      "grad_norm": 0.8504550792082386,
      "learning_rate": 0.001488,
      "loss": 5.4031,
      "step": 496
    },
    {
      "epoch": 0.00497,
      "grad_norm": 0.8843425860263048,
      "learning_rate": 0.001491,
      "loss": 5.3921,
      "step": 497
    },
    {
      "epoch": 0.00498,
      "grad_norm": 0.7438750793797253,
      "learning_rate": 0.001494,
      "loss": 5.4145,
      "step": 498
    },
    {
      "epoch": 0.00499,
      "grad_norm": 0.7036650868069556,
      "learning_rate": 0.001497,
      "loss": 5.3822,
      "step": 499
    },
    {
      "epoch": 0.005,
      "grad_norm": 0.7877256477029045,
      "learning_rate": 0.0015,
      "loss": 5.3896,
      "step": 500
    },
    {
      "epoch": 0.00501,
      "grad_norm": 0.7990985096145019,
      "learning_rate": 0.001503,
      "loss": 5.3912,
      "step": 501
    },
    {
      "epoch": 0.00502,
      "grad_norm": 0.5932348165440957,
      "learning_rate": 0.001506,
      "loss": 5.3876,
      "step": 502
    },
    {
      "epoch": 0.00503,
      "grad_norm": 0.6328380348360387,
      "learning_rate": 0.0015090000000000001,
      "loss": 5.391,
      "step": 503
    },
    {
      "epoch": 0.00504,
      "grad_norm": 0.5819727032922326,
      "learning_rate": 0.001512,
      "loss": 5.3693,
      "step": 504
    },
    {
      "epoch": 0.00505,
      "grad_norm": 0.5953710061568833,
      "learning_rate": 0.001515,
      "loss": 5.3594,
      "step": 505
    },
    {
      "epoch": 0.00506,
      "grad_norm": 0.5986845177656173,
      "learning_rate": 0.001518,
      "loss": 5.3624,
      "step": 506
    },
    {
      "epoch": 0.00507,
      "grad_norm": 0.623690249743195,
      "learning_rate": 0.001521,
      "loss": 5.3571,
      "step": 507
    },
    {
      "epoch": 0.00508,
      "grad_norm": 0.653996676321799,
      "learning_rate": 0.001524,
      "loss": 5.3588,
      "step": 508
    },
    {
      "epoch": 0.00509,
      "grad_norm": 0.7417086851753733,
      "learning_rate": 0.0015270000000000001,
      "loss": 5.3422,
      "step": 509
    },
    {
      "epoch": 0.0051,
      "grad_norm": 0.7033408638361137,
      "learning_rate": 0.0015300000000000001,
      "loss": 5.3598,
      "step": 510
    },
    {
      "epoch": 0.00511,
      "grad_norm": 0.7013752626190988,
      "learning_rate": 0.001533,
      "loss": 5.3361,
      "step": 511
    },
    {
      "epoch": 0.00512,
      "grad_norm": 0.7403626060663853,
      "learning_rate": 0.001536,
      "loss": 5.3344,
      "step": 512
    },
    {
      "epoch": 0.00513,
      "grad_norm": 0.7668257914395668,
      "learning_rate": 0.001539,
      "loss": 5.3488,
      "step": 513
    },
    {
      "epoch": 0.00514,
      "grad_norm": 0.8677889222141009,
      "learning_rate": 0.001542,
      "loss": 5.3327,
      "step": 514
    },
    {
      "epoch": 0.00515,
      "grad_norm": 0.896065553430359,
      "learning_rate": 0.0015450000000000001,
      "loss": 5.3422,
      "step": 515
    },
    {
      "epoch": 0.00516,
      "grad_norm": 1.0837566571017694,
      "learning_rate": 0.0015480000000000001,
      "loss": 5.3497,
      "step": 516
    },
    {
      "epoch": 0.00517,
      "grad_norm": 0.8071431981996826,
      "learning_rate": 0.001551,
      "loss": 5.3323,
      "step": 517
    },
    {
      "epoch": 0.00518,
      "grad_norm": 0.7918860105262308,
      "learning_rate": 0.001554,
      "loss": 5.3156,
      "step": 518
    },
    {
      "epoch": 0.00519,
      "grad_norm": 0.7777992304037674,
      "learning_rate": 0.001557,
      "loss": 5.3213,
      "step": 519
    },
    {
      "epoch": 0.0052,
      "grad_norm": 0.8275508154311308,
      "learning_rate": 0.0015600000000000002,
      "loss": 5.3297,
      "step": 520
    },
    {
      "epoch": 0.00521,
      "grad_norm": 1.081326682572488,
      "learning_rate": 0.0015630000000000002,
      "loss": 5.3161,
      "step": 521
    },
    {
      "epoch": 0.00522,
      "grad_norm": 1.0769033841173197,
      "learning_rate": 0.0015660000000000001,
      "loss": 5.3235,
      "step": 522
    },
    {
      "epoch": 0.00523,
      "grad_norm": 1.1142920349652348,
      "learning_rate": 0.001569,
      "loss": 5.3418,
      "step": 523
    },
    {
      "epoch": 0.00524,
      "grad_norm": 0.9680819380144772,
      "learning_rate": 0.001572,
      "loss": 5.3387,
      "step": 524
    },
    {
      "epoch": 0.00525,
      "grad_norm": 1.042843464512002,
      "learning_rate": 0.001575,
      "loss": 5.3364,
      "step": 525
    },
    {
      "epoch": 0.00526,
      "grad_norm": 0.8760110028730904,
      "learning_rate": 0.0015780000000000002,
      "loss": 5.3079,
      "step": 526
    },
    {
      "epoch": 0.00527,
      "grad_norm": 0.7131338611439731,
      "learning_rate": 0.0015810000000000002,
      "loss": 5.3127,
      "step": 527
    },
    {
      "epoch": 0.00528,
      "grad_norm": 0.6786352126644868,
      "learning_rate": 0.0015840000000000001,
      "loss": 5.2868,
      "step": 528
    },
    {
      "epoch": 0.00529,
      "grad_norm": 0.6952357562094686,
      "learning_rate": 0.001587,
      "loss": 5.2922,
      "step": 529
    },
    {
      "epoch": 0.0053,
      "grad_norm": 0.8086799159810172,
      "learning_rate": 0.00159,
      "loss": 5.3039,
      "step": 530
    },
    {
      "epoch": 0.00531,
      "grad_norm": 0.828973806141186,
      "learning_rate": 0.001593,
      "loss": 5.2873,
      "step": 531
    },
    {
      "epoch": 0.00532,
      "grad_norm": 0.7467931538676229,
      "learning_rate": 0.0015960000000000002,
      "loss": 5.2943,
      "step": 532
    },
    {
      "epoch": 0.00533,
      "grad_norm": 0.7141354989500697,
      "learning_rate": 0.0015990000000000002,
      "loss": 5.2786,
      "step": 533
    },
    {
      "epoch": 0.00534,
      "grad_norm": 0.9764493557723114,
      "learning_rate": 0.0016020000000000001,
      "loss": 5.2728,
      "step": 534
    },
    {
      "epoch": 0.00535,
      "grad_norm": 1.3880068471110967,
      "learning_rate": 0.001605,
      "loss": 5.2984,
      "step": 535
    },
    {
      "epoch": 0.00536,
      "grad_norm": 0.9041341459356813,
      "learning_rate": 0.001608,
      "loss": 5.2855,
      "step": 536
    },
    {
      "epoch": 0.00537,
      "grad_norm": 0.762740194970871,
      "learning_rate": 0.0016110000000000002,
      "loss": 5.2723,
      "step": 537
    },
    {
      "epoch": 0.00538,
      "grad_norm": 0.787312661332683,
      "learning_rate": 0.0016140000000000002,
      "loss": 5.2506,
      "step": 538
    },
    {
      "epoch": 0.00539,
      "grad_norm": 0.6102453826005042,
      "learning_rate": 0.0016170000000000002,
      "loss": 5.2365,
      "step": 539
    },
    {
      "epoch": 0.0054,
      "grad_norm": 0.6664103859218952,
      "learning_rate": 0.0016200000000000001,
      "loss": 5.2513,
      "step": 540
    },
    {
      "epoch": 0.00541,
      "grad_norm": 0.7228434818484509,
      "learning_rate": 0.001623,
      "loss": 5.2273,
      "step": 541
    },
    {
      "epoch": 0.00542,
      "grad_norm": 0.9646545444558308,
      "learning_rate": 0.001626,
      "loss": 5.2641,
      "step": 542
    },
    {
      "epoch": 0.00543,
      "grad_norm": 1.1121220265997553,
      "learning_rate": 0.0016290000000000002,
      "loss": 5.2329,
      "step": 543
    },
    {
      "epoch": 0.00544,
      "grad_norm": 0.7994777164441184,
      "learning_rate": 0.0016320000000000002,
      "loss": 5.2404,
      "step": 544
    },
    {
      "epoch": 0.00545,
      "grad_norm": 0.7226008260314222,
      "learning_rate": 0.0016350000000000002,
      "loss": 5.2461,
      "step": 545
    },
    {
      "epoch": 0.00546,
      "grad_norm": 0.7699535423166085,
      "learning_rate": 0.0016380000000000001,
      "loss": 5.2193,
      "step": 546
    },
    {
      "epoch": 0.00547,
      "grad_norm": 0.6548240326600666,
      "learning_rate": 0.001641,
      "loss": 5.2108,
      "step": 547
    },
    {
      "epoch": 0.00548,
      "grad_norm": 0.6332922946851393,
      "learning_rate": 0.001644,
      "loss": 5.2061,
      "step": 548
    },
    {
      "epoch": 0.00549,
      "grad_norm": 0.6231528959927674,
      "learning_rate": 0.0016470000000000002,
      "loss": 5.2,
      "step": 549
    },
    {
      "epoch": 0.0055,
      "grad_norm": 0.7419840881368932,
      "learning_rate": 0.0016500000000000002,
      "loss": 5.2179,
      "step": 550
    },
    {
      "epoch": 0.00551,
      "grad_norm": 0.7180205816820676,
      "learning_rate": 0.0016530000000000002,
      "loss": 5.2057,
      "step": 551
    },
    {
      "epoch": 0.00552,
      "grad_norm": 0.5920069574731561,
      "learning_rate": 0.0016560000000000001,
      "loss": 5.1823,
      "step": 552
    },
    {
      "epoch": 0.00553,
      "grad_norm": 0.7996998429214144,
      "learning_rate": 0.001659,
      "loss": 5.1806,
      "step": 553
    },
    {
      "epoch": 0.00554,
      "grad_norm": 1.0229110500291838,
      "learning_rate": 0.0016620000000000003,
      "loss": 5.1965,
      "step": 554
    },
    {
      "epoch": 0.00555,
      "grad_norm": 1.1118473608885646,
      "learning_rate": 0.0016650000000000002,
      "loss": 5.1994,
      "step": 555
    },
    {
      "epoch": 0.00556,
      "grad_norm": 0.9366759039894813,
      "learning_rate": 0.0016680000000000002,
      "loss": 5.1806,
      "step": 556
    },
    {
      "epoch": 0.00557,
      "grad_norm": 0.9046668934887724,
      "learning_rate": 0.0016710000000000002,
      "loss": 5.1671,
      "step": 557
    },
    {
      "epoch": 0.00558,
      "grad_norm": 1.142251826676036,
      "learning_rate": 0.0016740000000000001,
      "loss": 5.2009,
      "step": 558
    },
    {
      "epoch": 0.00559,
      "grad_norm": 1.0520781475504497,
      "learning_rate": 0.001677,
      "loss": 5.1865,
      "step": 559
    },
    {
      "epoch": 0.0056,
      "grad_norm": 1.0780070897638405,
      "learning_rate": 0.0016800000000000003,
      "loss": 5.1609,
      "step": 560
    },
    {
      "epoch": 0.00561,
      "grad_norm": 0.8904071170090557,
      "learning_rate": 0.0016830000000000003,
      "loss": 5.1755,
      "step": 561
    },
    {
      "epoch": 0.00562,
      "grad_norm": 0.8189640026396579,
      "learning_rate": 0.0016860000000000002,
      "loss": 5.168,
      "step": 562
    },
    {
      "epoch": 0.00563,
      "grad_norm": 0.746495696524217,
      "learning_rate": 0.001689,
      "loss": 5.1552,
      "step": 563
    },
    {
      "epoch": 0.00564,
      "grad_norm": 0.7249953066463264,
      "learning_rate": 0.001692,
      "loss": 5.1416,
      "step": 564
    },
    {
      "epoch": 0.00565,
      "grad_norm": 0.6193711615047397,
      "learning_rate": 0.001695,
      "loss": 5.1336,
      "step": 565
    },
    {
      "epoch": 0.00566,
      "grad_norm": 0.8661212922050541,
      "learning_rate": 0.0016979999999999999,
      "loss": 5.1381,
      "step": 566
    },
    {
      "epoch": 0.00567,
      "grad_norm": 0.9452019797636565,
      "learning_rate": 0.0017009999999999998,
      "loss": 5.1333,
      "step": 567
    },
    {
      "epoch": 0.00568,
      "grad_norm": 0.8863756714851743,
      "learning_rate": 0.0017039999999999998,
      "loss": 5.1455,
      "step": 568
    },
    {
      "epoch": 0.00569,
      "grad_norm": 0.8164512297006329,
      "learning_rate": 0.001707,
      "loss": 5.1087,
      "step": 569
    },
    {
      "epoch": 0.0057,
      "grad_norm": 0.8055756655780417,
      "learning_rate": 0.00171,
      "loss": 5.1416,
      "step": 570
    },
    {
      "epoch": 0.00571,
      "grad_norm": 0.9556127682537684,
      "learning_rate": 0.001713,
      "loss": 5.1421,
      "step": 571
    },
    {
      "epoch": 0.00572,
      "grad_norm": 1.1121438340859977,
      "learning_rate": 0.0017159999999999999,
      "loss": 5.1242,
      "step": 572
    },
    {
      "epoch": 0.00573,
      "grad_norm": 0.8538691427356556,
      "learning_rate": 0.0017189999999999998,
      "loss": 5.1261,
      "step": 573
    },
    {
      "epoch": 0.00574,
      "grad_norm": 0.754134808897758,
      "learning_rate": 0.001722,
      "loss": 5.1186,
      "step": 574
    },
    {
      "epoch": 0.00575,
      "grad_norm": 0.6045959777005846,
      "learning_rate": 0.001725,
      "loss": 5.0826,
      "step": 575
    },
    {
      "epoch": 0.00576,
      "grad_norm": 0.5849168439848929,
      "learning_rate": 0.001728,
      "loss": 5.0868,
      "step": 576
    },
    {
      "epoch": 0.00577,
      "grad_norm": 0.5881868482585118,
      "learning_rate": 0.001731,
      "loss": 5.0984,
      "step": 577
    },
    {
      "epoch": 0.00578,
      "grad_norm": 0.6496481817365951,
      "learning_rate": 0.0017339999999999999,
      "loss": 5.0795,
      "step": 578
    },
    {
      "epoch": 0.00579,
      "grad_norm": 0.7126042661301508,
      "learning_rate": 0.0017369999999999998,
      "loss": 5.0702,
      "step": 579
    },
    {
      "epoch": 0.0058,
      "grad_norm": 0.691634070596695,
      "learning_rate": 0.00174,
      "loss": 5.0826,
      "step": 580
    },
    {
      "epoch": 0.00581,
      "grad_norm": 0.6405819953602082,
      "learning_rate": 0.001743,
      "loss": 5.0809,
      "step": 581
    },
    {
      "epoch": 0.00582,
      "grad_norm": 0.6144348123489994,
      "learning_rate": 0.001746,
      "loss": 5.0509,
      "step": 582
    },
    {
      "epoch": 0.00583,
      "grad_norm": 0.5400038424579979,
      "learning_rate": 0.001749,
      "loss": 5.0752,
      "step": 583
    },
    {
      "epoch": 0.00584,
      "grad_norm": 0.4936797939946059,
      "learning_rate": 0.0017519999999999999,
      "loss": 5.0634,
      "step": 584
    },
    {
      "epoch": 0.00585,
      "grad_norm": 0.5420757595953297,
      "learning_rate": 0.0017549999999999998,
      "loss": 5.0509,
      "step": 585
    },
    {
      "epoch": 0.00586,
      "grad_norm": 0.6461298473240921,
      "learning_rate": 0.001758,
      "loss": 5.0463,
      "step": 586
    },
    {
      "epoch": 0.00587,
      "grad_norm": 1.0127747465457377,
      "learning_rate": 0.001761,
      "loss": 5.0477,
      "step": 587
    },
    {
      "epoch": 0.00588,
      "grad_norm": 1.312605638466154,
      "learning_rate": 0.001764,
      "loss": 5.0646,
      "step": 588
    },
    {
      "epoch": 0.00589,
      "grad_norm": 0.7336414507180539,
      "learning_rate": 0.001767,
      "loss": 5.0246,
      "step": 589
    },
    {
      "epoch": 0.0059,
      "grad_norm": 0.9403709566188089,
      "learning_rate": 0.0017699999999999999,
      "loss": 5.0344,
      "step": 590
    },
    {
      "epoch": 0.00591,
      "grad_norm": 1.1082528668309917,
      "learning_rate": 0.001773,
      "loss": 5.0455,
      "step": 591
    },
    {
      "epoch": 0.00592,
      "grad_norm": 1.2840924567627583,
      "learning_rate": 0.001776,
      "loss": 5.0904,
      "step": 592
    },
    {
      "epoch": 0.00593,
      "grad_norm": 0.7010705426983365,
      "learning_rate": 0.001779,
      "loss": 5.0507,
      "step": 593
    },
    {
      "epoch": 0.00594,
      "grad_norm": 0.7515579064184676,
      "learning_rate": 0.001782,
      "loss": 5.0452,
      "step": 594
    },
    {
      "epoch": 0.00595,
      "grad_norm": 0.8237589608472985,
      "learning_rate": 0.001785,
      "loss": 5.0574,
      "step": 595
    },
    {
      "epoch": 0.00596,
      "grad_norm": 0.7511193245039597,
      "learning_rate": 0.0017879999999999999,
      "loss": 5.0458,
      "step": 596
    },
    {
      "epoch": 0.00597,
      "grad_norm": 0.6951714106885373,
      "learning_rate": 0.001791,
      "loss": 5.0059,
      "step": 597
    },
    {
      "epoch": 0.00598,
      "grad_norm": 0.6637745885790589,
      "learning_rate": 0.001794,
      "loss": 5.0231,
      "step": 598
    },
    {
      "epoch": 0.00599,
      "grad_norm": 0.7127858481763457,
      "learning_rate": 0.001797,
      "loss": 5.0133,
      "step": 599
    },
    {
      "epoch": 0.006,
      "grad_norm": 0.6761974733345899,
      "learning_rate": 0.0018,
      "loss": 5.0135,
      "step": 600
    },
    {
      "epoch": 0.00601,
      "grad_norm": 0.6625605364634614,
      "learning_rate": 0.001803,
      "loss": 5.0015,
      "step": 601
    },
    {
      "epoch": 0.00602,
      "grad_norm": 0.6742478727145375,
      "learning_rate": 0.0018059999999999999,
      "loss": 4.9862,
      "step": 602
    },
    {
      "epoch": 0.00603,
      "grad_norm": 1.00015732542698,
      "learning_rate": 0.001809,
      "loss": 5.0193,
      "step": 603
    },
    {
      "epoch": 0.00604,
      "grad_norm": 1.387382192884798,
      "learning_rate": 0.001812,
      "loss": 5.0251,
      "step": 604
    },
    {
      "epoch": 0.00605,
      "grad_norm": 0.6727404947716551,
      "learning_rate": 0.001815,
      "loss": 5.0023,
      "step": 605
    },
    {
      "epoch": 0.00606,
      "grad_norm": 0.9044609854709968,
      "learning_rate": 0.001818,
      "loss": 5.0189,
      "step": 606
    },
    {
      "epoch": 0.00607,
      "grad_norm": 1.08596708759871,
      "learning_rate": 0.001821,
      "loss": 5.0221,
      "step": 607
    },
    {
      "epoch": 0.00608,
      "grad_norm": 1.0369936566425986,
      "learning_rate": 0.001824,
      "loss": 5.008,
      "step": 608
    },
    {
      "epoch": 0.00609,
      "grad_norm": 1.0935517991120203,
      "learning_rate": 0.001827,
      "loss": 5.0109,
      "step": 609
    },
    {
      "epoch": 0.0061,
      "grad_norm": 0.9727711844599547,
      "learning_rate": 0.00183,
      "loss": 4.9666,
      "step": 610
    },
    {
      "epoch": 0.00611,
      "grad_norm": 0.9492725313696737,
      "learning_rate": 0.001833,
      "loss": 4.9894,
      "step": 611
    },
    {
      "epoch": 0.00612,
      "grad_norm": 0.845936691035656,
      "learning_rate": 0.001836,
      "loss": 4.9768,
      "step": 612
    },
    {
      "epoch": 0.00613,
      "grad_norm": 0.917579763764549,
      "learning_rate": 0.001839,
      "loss": 4.9836,
      "step": 613
    },
    {
      "epoch": 0.00614,
      "grad_norm": 0.8975809320202123,
      "learning_rate": 0.001842,
      "loss": 5.0024,
      "step": 614
    },
    {
      "epoch": 0.00615,
      "grad_norm": 1.1935315831043936,
      "learning_rate": 0.001845,
      "loss": 5.0018,
      "step": 615
    },
    {
      "epoch": 0.00616,
      "grad_norm": 0.9948318214992812,
      "learning_rate": 0.001848,
      "loss": 4.9924,
      "step": 616
    },
    {
      "epoch": 0.00617,
      "grad_norm": 0.8063669388844663,
      "learning_rate": 0.001851,
      "loss": 4.9919,
      "step": 617
    },
    {
      "epoch": 0.00618,
      "grad_norm": 0.8184910219660716,
      "learning_rate": 0.001854,
      "loss": 4.9666,
      "step": 618
    },
    {
      "epoch": 0.00619,
      "grad_norm": 0.7780464806882716,
      "learning_rate": 0.001857,
      "loss": 4.9753,
      "step": 619
    },
    {
      "epoch": 0.0062,
      "grad_norm": 0.7430630101852395,
      "learning_rate": 0.00186,
      "loss": 4.9566,
      "step": 620
    },
    {
      "epoch": 0.00621,
      "grad_norm": 0.8040699289060931,
      "learning_rate": 0.001863,
      "loss": 4.9542,
      "step": 621
    },
    {
      "epoch": 0.00622,
      "grad_norm": 0.8423285566803137,
      "learning_rate": 0.001866,
      "loss": 4.9653,
      "step": 622
    },
    {
      "epoch": 0.00623,
      "grad_norm": 0.6802855865245334,
      "learning_rate": 0.001869,
      "loss": 4.9365,
      "step": 623
    },
    {
      "epoch": 0.00624,
      "grad_norm": 0.7045868643772514,
      "learning_rate": 0.001872,
      "loss": 4.9425,
      "step": 624
    },
    {
      "epoch": 0.00625,
      "grad_norm": 0.69605003901388,
      "learning_rate": 0.001875,
      "loss": 4.9397,
      "step": 625
    },
    {
      "epoch": 0.00626,
      "grad_norm": 0.8788947819856907,
      "learning_rate": 0.0018780000000000001,
      "loss": 4.9403,
      "step": 626
    },
    {
      "epoch": 0.00627,
      "grad_norm": 0.8580113274469313,
      "learning_rate": 0.001881,
      "loss": 4.9238,
      "step": 627
    },
    {
      "epoch": 0.00628,
      "grad_norm": 0.7437087045232712,
      "learning_rate": 0.001884,
      "loss": 4.9553,
      "step": 628
    },
    {
      "epoch": 0.00629,
      "grad_norm": 0.673794469112573,
      "learning_rate": 0.001887,
      "loss": 4.9059,
      "step": 629
    },
    {
      "epoch": 0.0063,
      "grad_norm": 0.7529443514224647,
      "learning_rate": 0.00189,
      "loss": 4.9225,
      "step": 630
    },
    {
      "epoch": 0.00631,
      "grad_norm": 0.7882316002133182,
      "learning_rate": 0.0018930000000000002,
      "loss": 4.9159,
      "step": 631
    },
    {
      "epoch": 0.00632,
      "grad_norm": 0.7345089369079263,
      "learning_rate": 0.0018960000000000001,
      "loss": 4.9318,
      "step": 632
    },
    {
      "epoch": 0.00633,
      "grad_norm": 0.807557335679046,
      "learning_rate": 0.001899,
      "loss": 4.9156,
      "step": 633
    },
    {
      "epoch": 0.00634,
      "grad_norm": 0.856273971211143,
      "learning_rate": 0.001902,
      "loss": 4.9086,
      "step": 634
    },
    {
      "epoch": 0.00635,
      "grad_norm": 0.8041095750030954,
      "learning_rate": 0.001905,
      "loss": 4.8993,
      "step": 635
    },
    {
      "epoch": 0.00636,
      "grad_norm": 0.8334087326563642,
      "learning_rate": 0.001908,
      "loss": 4.9117,
      "step": 636
    },
    {
      "epoch": 0.00637,
      "grad_norm": 0.8711627404236827,
      "learning_rate": 0.0019110000000000002,
      "loss": 4.9242,
      "step": 637
    },
    {
      "epoch": 0.00638,
      "grad_norm": 0.950273886749592,
      "learning_rate": 0.0019140000000000001,
      "loss": 4.918,
      "step": 638
    },
    {
      "epoch": 0.00639,
      "grad_norm": 0.9763758019156279,
      "learning_rate": 0.001917,
      "loss": 4.8946,
      "step": 639
    },
    {
      "epoch": 0.0064,
      "grad_norm": 0.9069546349974866,
      "learning_rate": 0.00192,
      "loss": 4.9049,
      "step": 640
    },
    {
      "epoch": 0.00641,
      "grad_norm": 0.7602914411110755,
      "learning_rate": 0.001923,
      "loss": 4.8967,
      "step": 641
    },
    {
      "epoch": 0.00642,
      "grad_norm": 0.6358369958975738,
      "learning_rate": 0.001926,
      "loss": 4.8823,
      "step": 642
    },
    {
      "epoch": 0.00643,
      "grad_norm": 0.6298192488222032,
      "learning_rate": 0.0019290000000000002,
      "loss": 4.8984,
      "step": 643
    },
    {
      "epoch": 0.00644,
      "grad_norm": 0.5835488483304159,
      "learning_rate": 0.0019320000000000001,
      "loss": 4.8719,
      "step": 644
    },
    {
      "epoch": 0.00645,
      "grad_norm": 0.509893487039198,
      "learning_rate": 0.001935,
      "loss": 4.8697,
      "step": 645
    },
    {
      "epoch": 0.00646,
      "grad_norm": 0.5016743009567477,
      "learning_rate": 0.001938,
      "loss": 4.8822,
      "step": 646
    },
    {
      "epoch": 0.00647,
      "grad_norm": 0.4834671196673339,
      "learning_rate": 0.001941,
      "loss": 4.8737,
      "step": 647
    },
    {
      "epoch": 0.00648,
      "grad_norm": 0.4900666309904975,
      "learning_rate": 0.0019440000000000002,
      "loss": 4.8665,
      "step": 648
    },
    {
      "epoch": 0.00649,
      "grad_norm": 0.6242032394190251,
      "learning_rate": 0.0019470000000000002,
      "loss": 4.8569,
      "step": 649
    },
    {
      "epoch": 0.0065,
      "grad_norm": 0.7946730011730083,
      "learning_rate": 0.0019500000000000001,
      "loss": 4.8944,
      "step": 650
    },
    {
      "epoch": 0.00651,
      "grad_norm": 0.7813096551019217,
      "learning_rate": 0.001953,
      "loss": 4.8797,
      "step": 651
    },
    {
      "epoch": 0.00652,
      "grad_norm": 0.5708054833927125,
      "learning_rate": 0.0019560000000000003,
      "loss": 4.8538,
      "step": 652
    },
    {
      "epoch": 0.00653,
      "grad_norm": 0.6416767177196502,
      "learning_rate": 0.0019590000000000002,
      "loss": 4.8612,
      "step": 653
    },
    {
      "epoch": 0.00654,
      "grad_norm": 0.6414247107018044,
      "learning_rate": 0.001962,
      "loss": 4.8324,
      "step": 654
    },
    {
      "epoch": 0.00655,
      "grad_norm": 0.5608777579230684,
      "learning_rate": 0.001965,
      "loss": 4.8581,
      "step": 655
    },
    {
      "epoch": 0.00656,
      "grad_norm": 0.4812696659686437,
      "learning_rate": 0.001968,
      "loss": 4.8497,
      "step": 656
    },
    {
      "epoch": 0.00657,
      "grad_norm": 0.5196607021803705,
      "learning_rate": 0.001971,
      "loss": 4.8212,
      "step": 657
    },
    {
      "epoch": 0.00658,
      "grad_norm": 0.5384007134004025,
      "learning_rate": 0.001974,
      "loss": 4.8422,
      "step": 658
    },
    {
      "epoch": 0.00659,
      "grad_norm": 0.6084877834513672,
      "learning_rate": 0.001977,
      "loss": 4.8215,
      "step": 659
    },
    {
      "epoch": 0.0066,
      "grad_norm": 0.7589081819730935,
      "learning_rate": 0.00198,
      "loss": 4.8434,
      "step": 660
    },
    {
      "epoch": 0.00661,
      "grad_norm": 0.7941713837035096,
      "learning_rate": 0.001983,
      "loss": 4.8155,
      "step": 661
    },
    {
      "epoch": 0.00662,
      "grad_norm": 0.7978547164974868,
      "learning_rate": 0.0019860000000000004,
      "loss": 4.8127,
      "step": 662
    },
    {
      "epoch": 0.00663,
      "grad_norm": 0.893378015840618,
      "learning_rate": 0.0019890000000000003,
      "loss": 4.8265,
      "step": 663
    },
    {
      "epoch": 0.00664,
      "grad_norm": 0.899949440268495,
      "learning_rate": 0.0019920000000000003,
      "loss": 4.8283,
      "step": 664
    },
    {
      "epoch": 0.00665,
      "grad_norm": 0.8127988991300923,
      "learning_rate": 0.0019950000000000002,
      "loss": 4.7986,
      "step": 665
    },
    {
      "epoch": 0.00666,
      "grad_norm": 0.9200900241429067,
      "learning_rate": 0.001998,
      "loss": 4.8226,
      "step": 666
    },
    {
      "epoch": 0.00667,
      "grad_norm": 1.037264674390151,
      "learning_rate": 0.002001,
      "loss": 4.8324,
      "step": 667
    },
    {
      "epoch": 0.00668,
      "grad_norm": 0.8082146942337904,
      "learning_rate": 0.002004,
      "loss": 4.8427,
      "step": 668
    },
    {
      "epoch": 0.00669,
      "grad_norm": 0.7033624756486074,
      "learning_rate": 0.002007,
      "loss": 4.8562,
      "step": 669
    },
    {
      "epoch": 0.0067,
      "grad_norm": 0.751969455636164,
      "learning_rate": 0.00201,
      "loss": 4.8525,
      "step": 670
    },
    {
      "epoch": 0.00671,
      "grad_norm": 0.736520529365372,
      "learning_rate": 0.002013,
      "loss": 4.8206,
      "step": 671
    },
    {
      "epoch": 0.00672,
      "grad_norm": 0.7466982774868701,
      "learning_rate": 0.002016,
      "loss": 4.8129,
      "step": 672
    },
    {
      "epoch": 0.00673,
      "grad_norm": 0.7025220166479262,
      "learning_rate": 0.002019,
      "loss": 4.8146,
      "step": 673
    },
    {
      "epoch": 0.00674,
      "grad_norm": 0.8461453039283889,
      "learning_rate": 0.0020220000000000004,
      "loss": 4.8144,
      "step": 674
    },
    {
      "epoch": 0.00675,
      "grad_norm": 0.9399881649158435,
      "learning_rate": 0.0020250000000000003,
      "loss": 4.8482,
      "step": 675
    },
    {
      "epoch": 0.00676,
      "grad_norm": 0.9357632097468723,
      "learning_rate": 0.0020280000000000003,
      "loss": 4.8268,
      "step": 676
    },
    {
      "epoch": 0.00677,
      "grad_norm": 0.7758960619033557,
      "learning_rate": 0.0020310000000000003,
      "loss": 4.8204,
      "step": 677
    },
    {
      "epoch": 0.00678,
      "grad_norm": 0.699292513140664,
      "learning_rate": 0.0020340000000000002,
      "loss": 4.8248,
      "step": 678
    },
    {
      "epoch": 0.00679,
      "grad_norm": 0.7370787957429817,
      "learning_rate": 0.002037,
      "loss": 4.816,
      "step": 679
    },
    {
      "epoch": 0.0068,
      "grad_norm": 0.8377547362902558,
      "learning_rate": 0.00204,
      "loss": 4.8174,
      "step": 680
    },
    {
      "epoch": 0.00681,
      "grad_norm": 0.8259782799224379,
      "learning_rate": 0.002043,
      "loss": 4.8155,
      "step": 681
    },
    {
      "epoch": 0.00682,
      "grad_norm": 0.7684261091318535,
      "learning_rate": 0.002046,
      "loss": 4.8082,
      "step": 682
    },
    {
      "epoch": 0.00683,
      "grad_norm": 0.8487367019402318,
      "learning_rate": 0.002049,
      "loss": 4.7989,
      "step": 683
    },
    {
      "epoch": 0.00684,
      "grad_norm": 0.8838018894616847,
      "learning_rate": 0.002052,
      "loss": 4.8194,
      "step": 684
    },
    {
      "epoch": 0.00685,
      "grad_norm": 0.8860697203395584,
      "learning_rate": 0.0020550000000000004,
      "loss": 4.8252,
      "step": 685
    },
    {
      "epoch": 0.00686,
      "grad_norm": 0.7336183086529302,
      "learning_rate": 0.0020580000000000004,
      "loss": 4.7837,
      "step": 686
    },
    {
      "epoch": 0.00687,
      "grad_norm": 0.8176630379413288,
      "learning_rate": 0.0020610000000000003,
      "loss": 4.7944,
      "step": 687
    },
    {
      "epoch": 0.00688,
      "grad_norm": 0.7703386551342313,
      "learning_rate": 0.002064,
      "loss": 4.7926,
      "step": 688
    },
    {
      "epoch": 0.00689,
      "grad_norm": 0.6919162061146223,
      "learning_rate": 0.002067,
      "loss": 4.7965,
      "step": 689
    },
    {
      "epoch": 0.0069,
      "grad_norm": 0.7424392154268248,
      "learning_rate": 0.00207,
      "loss": 4.7893,
      "step": 690
    },
    {
      "epoch": 0.00691,
      "grad_norm": 0.6515618524352145,
      "learning_rate": 0.0020729999999999998,
      "loss": 4.7559,
      "step": 691
    },
    {
      "epoch": 0.00692,
      "grad_norm": 0.6440846578393002,
      "learning_rate": 0.0020759999999999997,
      "loss": 4.776,
      "step": 692
    },
    {
      "epoch": 0.00693,
      "grad_norm": 0.6847536481828279,
      "learning_rate": 0.0020789999999999997,
      "loss": 4.7889,
      "step": 693
    },
    {
      "epoch": 0.00694,
      "grad_norm": 0.6321576161870056,
      "learning_rate": 0.002082,
      "loss": 4.7627,
      "step": 694
    },
    {
      "epoch": 0.00695,
      "grad_norm": 0.5791129920715202,
      "learning_rate": 0.002085,
      "loss": 4.7609,
      "step": 695
    },
    {
      "epoch": 0.00696,
      "grad_norm": 0.5895865438272808,
      "learning_rate": 0.002088,
      "loss": 4.7723,
      "step": 696
    },
    {
      "epoch": 0.00697,
      "grad_norm": 0.5008187604770619,
      "learning_rate": 0.002091,
      "loss": 4.7695,
      "step": 697
    },
    {
      "epoch": 0.00698,
      "grad_norm": 0.6970439697756265,
      "learning_rate": 0.002094,
      "loss": 4.7619,
      "step": 698
    },
    {
      "epoch": 0.00699,
      "grad_norm": 0.8941704543265332,
      "learning_rate": 0.002097,
      "loss": 4.7572,
      "step": 699
    },
    {
      "epoch": 0.007,
      "grad_norm": 0.9068627730041655,
      "learning_rate": 0.0021,
      "loss": 4.787,
      "step": 700
    },
    {
      "epoch": 0.00701,
      "grad_norm": 0.7146483381512303,
      "learning_rate": 0.002103,
      "loss": 4.7547,
      "step": 701
    },
    {
      "epoch": 0.00702,
      "grad_norm": 0.9172255209446268,
      "learning_rate": 0.002106,
      "loss": 4.77,
      "step": 702
    },
    {
      "epoch": 0.00703,
      "grad_norm": 0.9047172643914575,
      "learning_rate": 0.0021089999999999998,
      "loss": 4.7553,
      "step": 703
    },
    {
      "epoch": 0.00704,
      "grad_norm": 0.7853692419556185,
      "learning_rate": 0.0021119999999999997,
      "loss": 4.7583,
      "step": 704
    },
    {
      "epoch": 0.00705,
      "grad_norm": 0.7199878385614988,
      "learning_rate": 0.002115,
      "loss": 4.7725,
      "step": 705
    },
    {
      "epoch": 0.00706,
      "grad_norm": 0.7213393080579115,
      "learning_rate": 0.002118,
      "loss": 4.7581,
      "step": 706
    },
    {
      "epoch": 0.00707,
      "grad_norm": 0.7597119331851468,
      "learning_rate": 0.002121,
      "loss": 4.7413,
      "step": 707
    },
    {
      "epoch": 0.00708,
      "grad_norm": 0.6864102182118973,
      "learning_rate": 0.002124,
      "loss": 4.7187,
      "step": 708
    },
    {
      "epoch": 0.00709,
      "grad_norm": 0.7815902187763394,
      "learning_rate": 0.002127,
      "loss": 4.7572,
      "step": 709
    },
    {
      "epoch": 0.0071,
      "grad_norm": 0.8451784595752648,
      "learning_rate": 0.00213,
      "loss": 4.7552,
      "step": 710
    },
    {
      "epoch": 0.00711,
      "grad_norm": 1.0054655399528605,
      "learning_rate": 0.002133,
      "loss": 4.7414,
      "step": 711
    },
    {
      "epoch": 0.00712,
      "grad_norm": 0.9031323884556907,
      "learning_rate": 0.002136,
      "loss": 4.7728,
      "step": 712
    },
    {
      "epoch": 0.00713,
      "grad_norm": 0.961250906275713,
      "learning_rate": 0.002139,
      "loss": 4.7862,
      "step": 713
    },
    {
      "epoch": 0.00714,
      "grad_norm": 0.9556615314074448,
      "learning_rate": 0.002142,
      "loss": 4.7819,
      "step": 714
    },
    {
      "epoch": 0.00715,
      "grad_norm": 0.837203607680531,
      "learning_rate": 0.0021449999999999998,
      "loss": 4.7417,
      "step": 715
    },
    {
      "epoch": 0.00716,
      "grad_norm": 0.7607986282551458,
      "learning_rate": 0.002148,
      "loss": 4.7359,
      "step": 716
    },
    {
      "epoch": 0.00717,
      "grad_norm": 0.8703365352693242,
      "learning_rate": 0.002151,
      "loss": 4.7519,
      "step": 717
    },
    {
      "epoch": 0.00718,
      "grad_norm": 0.8830641357048177,
      "learning_rate": 0.002154,
      "loss": 4.7536,
      "step": 718
    },
    {
      "epoch": 0.00719,
      "grad_norm": 0.8090298717986324,
      "learning_rate": 0.002157,
      "loss": 4.7586,
      "step": 719
    },
    {
      "epoch": 0.0072,
      "grad_norm": 0.7002439324520396,
      "learning_rate": 0.00216,
      "loss": 4.7466,
      "step": 720
    },
    {
      "epoch": 0.00721,
      "grad_norm": 0.7540412799334538,
      "learning_rate": 0.002163,
      "loss": 4.7512,
      "step": 721
    },
    {
      "epoch": 0.00722,
      "grad_norm": 0.7234067697970273,
      "learning_rate": 0.002166,
      "loss": 4.7241,
      "step": 722
    },
    {
      "epoch": 0.00723,
      "grad_norm": 0.5796869415275953,
      "learning_rate": 0.002169,
      "loss": 4.73,
      "step": 723
    },
    {
      "epoch": 0.00724,
      "grad_norm": 0.6360613090935692,
      "learning_rate": 0.002172,
      "loss": 4.7294,
      "step": 724
    },
    {
      "epoch": 0.00725,
      "grad_norm": 0.6592111108932344,
      "learning_rate": 0.002175,
      "loss": 4.7232,
      "step": 725
    },
    {
      "epoch": 0.00726,
      "grad_norm": 0.7000176967246123,
      "learning_rate": 0.002178,
      "loss": 4.7406,
      "step": 726
    },
    {
      "epoch": 0.00727,
      "grad_norm": 0.6658154130327723,
      "learning_rate": 0.0021809999999999998,
      "loss": 4.7131,
      "step": 727
    },
    {
      "epoch": 0.00728,
      "grad_norm": 0.500886702178687,
      "learning_rate": 0.002184,
      "loss": 4.7222,
      "step": 728
    },
    {
      "epoch": 0.00729,
      "grad_norm": 0.553445989931654,
      "learning_rate": 0.002187,
      "loss": 4.7196,
      "step": 729
    },
    {
      "epoch": 0.0073,
      "grad_norm": 0.5928953773304845,
      "learning_rate": 0.00219,
      "loss": 4.7153,
      "step": 730
    },
    {
      "epoch": 0.00731,
      "grad_norm": 0.5280339609019513,
      "learning_rate": 0.002193,
      "loss": 4.7069,
      "step": 731
    },
    {
      "epoch": 0.00732,
      "grad_norm": 0.4601497488067425,
      "learning_rate": 0.002196,
      "loss": 4.7146,
      "step": 732
    },
    {
      "epoch": 0.00733,
      "grad_norm": 0.4831437067076967,
      "learning_rate": 0.002199,
      "loss": 4.6865,
      "step": 733
    },
    {
      "epoch": 0.00734,
      "grad_norm": 0.48957731222976764,
      "learning_rate": 0.002202,
      "loss": 4.7176,
      "step": 734
    },
    {
      "epoch": 0.00735,
      "grad_norm": 0.5029506248084066,
      "learning_rate": 0.002205,
      "loss": 4.7231,
      "step": 735
    },
    {
      "epoch": 0.00736,
      "grad_norm": 0.5300436466729722,
      "learning_rate": 0.002208,
      "loss": 4.7045,
      "step": 736
    },
    {
      "epoch": 0.00737,
      "grad_norm": 0.5354857814520255,
      "learning_rate": 0.002211,
      "loss": 4.6701,
      "step": 737
    },
    {
      "epoch": 0.00738,
      "grad_norm": 0.6855959285026678,
      "learning_rate": 0.002214,
      "loss": 4.6857,
      "step": 738
    },
    {
      "epoch": 0.00739,
      "grad_norm": 0.7193696222416395,
      "learning_rate": 0.0022170000000000002,
      "loss": 4.6773,
      "step": 739
    },
    {
      "epoch": 0.0074,
      "grad_norm": 0.7000843676029133,
      "learning_rate": 0.00222,
      "loss": 4.686,
      "step": 740
    },
    {
      "epoch": 0.00741,
      "grad_norm": 0.8262482718120322,
      "learning_rate": 0.002223,
      "loss": 4.6648,
      "step": 741
    },
    {
      "epoch": 0.00742,
      "grad_norm": 0.8068053565529363,
      "learning_rate": 0.002226,
      "loss": 4.71,
      "step": 742
    },
    {
      "epoch": 0.00743,
      "grad_norm": 0.7713935209386231,
      "learning_rate": 0.002229,
      "loss": 4.6667,
      "step": 743
    },
    {
      "epoch": 0.00744,
      "grad_norm": 0.6806090978340125,
      "learning_rate": 0.002232,
      "loss": 4.668,
      "step": 744
    },
    {
      "epoch": 0.00745,
      "grad_norm": 0.8150134425373752,
      "learning_rate": 0.002235,
      "loss": 4.6906,
      "step": 745
    },
    {
      "epoch": 0.00746,
      "grad_norm": 0.9083405480081935,
      "learning_rate": 0.002238,
      "loss": 4.6909,
      "step": 746
    },
    {
      "epoch": 0.00747,
      "grad_norm": 1.1312224699232258,
      "learning_rate": 0.002241,
      "loss": 4.6956,
      "step": 747
    },
    {
      "epoch": 0.00748,
      "grad_norm": 0.8174601291810354,
      "learning_rate": 0.002244,
      "loss": 4.6973,
      "step": 748
    },
    {
      "epoch": 0.00749,
      "grad_norm": 0.8423282795209855,
      "learning_rate": 0.002247,
      "loss": 4.6802,
      "step": 749
    },
    {
      "epoch": 0.0075,
      "grad_norm": 0.8679250685858194,
      "learning_rate": 0.0022500000000000003,
      "loss": 4.7268,
      "step": 750
    },
    {
      "epoch": 0.00751,
      "grad_norm": 1.0939871221004271,
      "learning_rate": 0.0022530000000000002,
      "loss": 4.7337,
      "step": 751
    },
    {
      "epoch": 0.00752,
      "grad_norm": 0.9886618564303525,
      "learning_rate": 0.002256,
      "loss": 4.6822,
      "step": 752
    },
    {
      "epoch": 0.00753,
      "grad_norm": 0.9258452428585717,
      "learning_rate": 0.002259,
      "loss": 4.7192,
      "step": 753
    },
    {
      "epoch": 0.00754,
      "grad_norm": 0.8790583060068752,
      "learning_rate": 0.002262,
      "loss": 4.7345,
      "step": 754
    },
    {
      "epoch": 0.00755,
      "grad_norm": 0.7727162685258321,
      "learning_rate": 0.002265,
      "loss": 4.6919,
      "step": 755
    },
    {
      "epoch": 0.00756,
      "grad_norm": 0.8048369196552551,
      "learning_rate": 0.002268,
      "loss": 4.6801,
      "step": 756
    },
    {
      "epoch": 0.00757,
      "grad_norm": 0.7307749692176225,
      "learning_rate": 0.002271,
      "loss": 4.6902,
      "step": 757
    },
    {
      "epoch": 0.00758,
      "grad_norm": 0.7628918458541498,
      "learning_rate": 0.002274,
      "loss": 4.6882,
      "step": 758
    },
    {
      "epoch": 0.00759,
      "grad_norm": 0.6811469027490352,
      "learning_rate": 0.002277,
      "loss": 4.693,
      "step": 759
    },
    {
      "epoch": 0.0076,
      "grad_norm": 0.5530412513377371,
      "learning_rate": 0.00228,
      "loss": 4.6735,
      "step": 760
    },
    {
      "epoch": 0.00761,
      "grad_norm": 0.5221449888101848,
      "learning_rate": 0.002283,
      "loss": 4.6405,
      "step": 761
    },
    {
      "epoch": 0.00762,
      "grad_norm": 0.5687089394846262,
      "learning_rate": 0.0022860000000000003,
      "loss": 4.6208,
      "step": 762
    },
    {
      "epoch": 0.00763,
      "grad_norm": 0.5814285508645848,
      "learning_rate": 0.0022890000000000002,
      "loss": 4.6718,
      "step": 763
    },
    {
      "epoch": 0.00764,
      "grad_norm": 0.6388540284979518,
      "learning_rate": 0.002292,
      "loss": 4.683,
      "step": 764
    },
    {
      "epoch": 0.00765,
      "grad_norm": 0.7278589788698283,
      "learning_rate": 0.002295,
      "loss": 4.6752,
      "step": 765
    },
    {
      "epoch": 0.00766,
      "grad_norm": 0.7050667087459527,
      "learning_rate": 0.002298,
      "loss": 4.6715,
      "step": 766
    },
    {
      "epoch": 0.00767,
      "grad_norm": 0.6023307971425665,
      "learning_rate": 0.002301,
      "loss": 4.6623,
      "step": 767
    },
    {
      "epoch": 0.00768,
      "grad_norm": 0.7162182495110988,
      "learning_rate": 0.002304,
      "loss": 4.6687,
      "step": 768
    },
    {
      "epoch": 0.00769,
      "grad_norm": 0.8754398583131885,
      "learning_rate": 0.002307,
      "loss": 4.6855,
      "step": 769
    },
    {
      "epoch": 0.0077,
      "grad_norm": 0.9282950433205286,
      "learning_rate": 0.00231,
      "loss": 4.656,
      "step": 770
    },
    {
      "epoch": 0.00771,
      "grad_norm": 0.8826983762517153,
      "learning_rate": 0.002313,
      "loss": 4.6927,
      "step": 771
    },
    {
      "epoch": 0.00772,
      "grad_norm": 0.7955428299875815,
      "learning_rate": 0.002316,
      "loss": 4.6752,
      "step": 772
    },
    {
      "epoch": 0.00773,
      "grad_norm": 0.6879017191297421,
      "learning_rate": 0.0023190000000000003,
      "loss": 4.6732,
      "step": 773
    },
    {
      "epoch": 0.00774,
      "grad_norm": 0.5805904836674535,
      "learning_rate": 0.0023220000000000003,
      "loss": 4.6842,
      "step": 774
    },
    {
      "epoch": 0.00775,
      "grad_norm": 0.5872309146101224,
      "learning_rate": 0.0023250000000000002,
      "loss": 4.6741,
      "step": 775
    },
    {
      "epoch": 0.00776,
      "grad_norm": 0.6663800275805344,
      "learning_rate": 0.002328,
      "loss": 4.6762,
      "step": 776
    },
    {
      "epoch": 0.00777,
      "grad_norm": 0.6475349218207965,
      "learning_rate": 0.002331,
      "loss": 4.6499,
      "step": 777
    },
    {
      "epoch": 0.00778,
      "grad_norm": 0.6498373909650491,
      "learning_rate": 0.002334,
      "loss": 4.6643,
      "step": 778
    },
    {
      "epoch": 0.00779,
      "grad_norm": 0.6357690952406082,
      "learning_rate": 0.002337,
      "loss": 4.6181,
      "step": 779
    },
    {
      "epoch": 0.0078,
      "grad_norm": 0.6241849680287349,
      "learning_rate": 0.00234,
      "loss": 4.6419,
      "step": 780
    },
    {
      "epoch": 0.00781,
      "grad_norm": 0.6808062655697796,
      "learning_rate": 0.002343,
      "loss": 4.6451,
      "step": 781
    },
    {
      "epoch": 0.00782,
      "grad_norm": 0.7065267585011001,
      "learning_rate": 0.002346,
      "loss": 4.6436,
      "step": 782
    },
    {
      "epoch": 0.00783,
      "grad_norm": 0.6381701638777938,
      "learning_rate": 0.002349,
      "loss": 4.6242,
      "step": 783
    },
    {
      "epoch": 0.00784,
      "grad_norm": 0.647841330234918,
      "learning_rate": 0.002352,
      "loss": 4.6355,
      "step": 784
    },
    {
      "epoch": 0.00785,
      "grad_norm": 0.60562299847623,
      "learning_rate": 0.0023550000000000003,
      "loss": 4.6345,
      "step": 785
    },
    {
      "epoch": 0.00786,
      "grad_norm": 0.5148036449557751,
      "learning_rate": 0.0023580000000000003,
      "loss": 4.628,
      "step": 786
    },
    {
      "epoch": 0.00787,
      "grad_norm": 0.7217903580979332,
      "learning_rate": 0.0023610000000000003,
      "loss": 4.641,
      "step": 787
    },
    {
      "epoch": 0.00788,
      "grad_norm": 0.8515178253715305,
      "learning_rate": 0.002364,
      "loss": 4.6466,
      "step": 788
    },
    {
      "epoch": 0.00789,
      "grad_norm": 0.8831947515557061,
      "learning_rate": 0.002367,
      "loss": 4.6753,
      "step": 789
    },
    {
      "epoch": 0.0079,
      "grad_norm": 0.9816312249435458,
      "learning_rate": 0.00237,
      "loss": 4.6574,
      "step": 790
    },
    {
      "epoch": 0.00791,
      "grad_norm": 0.8257428278422617,
      "learning_rate": 0.002373,
      "loss": 4.6081,
      "step": 791
    },
    {
      "epoch": 0.00792,
      "grad_norm": 0.6889392893349975,
      "learning_rate": 0.002376,
      "loss": 4.6372,
      "step": 792
    },
    {
      "epoch": 0.00793,
      "grad_norm": 0.5470340397913868,
      "learning_rate": 0.002379,
      "loss": 4.6272,
      "step": 793
    },
    {
      "epoch": 0.00794,
      "grad_norm": 0.5333909794818302,
      "learning_rate": 0.002382,
      "loss": 4.6359,
      "step": 794
    },
    {
      "epoch": 0.00795,
      "grad_norm": 0.47330116805918854,
      "learning_rate": 0.002385,
      "loss": 4.6105,
      "step": 795
    },
    {
      "epoch": 0.00796,
      "grad_norm": 0.46996625387544017,
      "learning_rate": 0.0023880000000000004,
      "loss": 4.6049,
      "step": 796
    },
    {
      "epoch": 0.00797,
      "grad_norm": 0.4648603328586337,
      "learning_rate": 0.0023910000000000003,
      "loss": 4.6461,
      "step": 797
    },
    {
      "epoch": 0.00798,
      "grad_norm": 0.48147045136320854,
      "learning_rate": 0.0023940000000000003,
      "loss": 4.6125,
      "step": 798
    },
    {
      "epoch": 0.00799,
      "grad_norm": 0.5520002928611956,
      "learning_rate": 0.0023970000000000003,
      "loss": 4.6461,
      "step": 799
    },
    {
      "epoch": 0.008,
      "grad_norm": 0.49520453644625784,
      "learning_rate": 0.0024000000000000002,
      "loss": 4.5958,
      "step": 800
    },
    {
      "epoch": 0.00801,
      "grad_norm": 0.4617883447215667,
      "learning_rate": 0.002403,
      "loss": 4.6244,
      "step": 801
    },
    {
      "epoch": 0.00802,
      "grad_norm": 0.6171622349156032,
      "learning_rate": 0.002406,
      "loss": 4.6206,
      "step": 802
    },
    {
      "epoch": 0.00803,
      "grad_norm": 0.7819651777797723,
      "learning_rate": 0.002409,
      "loss": 4.605,
      "step": 803
    },
    {
      "epoch": 0.00804,
      "grad_norm": 0.8072878053268496,
      "learning_rate": 0.002412,
      "loss": 4.6195,
      "step": 804
    },
    {
      "epoch": 0.00805,
      "grad_norm": 0.6878812715646375,
      "learning_rate": 0.002415,
      "loss": 4.6081,
      "step": 805
    },
    {
      "epoch": 0.00806,
      "grad_norm": 0.7628005366591507,
      "learning_rate": 0.002418,
      "loss": 4.6308,
      "step": 806
    },
    {
      "epoch": 0.00807,
      "grad_norm": 0.7958527408861041,
      "learning_rate": 0.0024210000000000004,
      "loss": 4.6228,
      "step": 807
    },
    {
      "epoch": 0.00808,
      "grad_norm": 0.7899389450719584,
      "learning_rate": 0.0024240000000000004,
      "loss": 4.6083,
      "step": 808
    },
    {
      "epoch": 0.00809,
      "grad_norm": 1.193539574978161,
      "learning_rate": 0.0024270000000000003,
      "loss": 4.6201,
      "step": 809
    },
    {
      "epoch": 0.0081,
      "grad_norm": 1.0067442084818319,
      "learning_rate": 0.0024300000000000003,
      "loss": 4.6554,
      "step": 810
    },
    {
      "epoch": 0.00811,
      "grad_norm": 1.0302946243785736,
      "learning_rate": 0.0024330000000000003,
      "loss": 4.6338,
      "step": 811
    },
    {
      "epoch": 0.00812,
      "grad_norm": 0.7966936317025509,
      "learning_rate": 0.0024360000000000002,
      "loss": 4.6044,
      "step": 812
    },
    {
      "epoch": 0.00813,
      "grad_norm": 0.7843512654561826,
      "learning_rate": 0.0024389999999999998,
      "loss": 4.6449,
      "step": 813
    },
    {
      "epoch": 0.00814,
      "grad_norm": 0.7981724618057067,
      "learning_rate": 0.0024419999999999997,
      "loss": 4.6482,
      "step": 814
    },
    {
      "epoch": 0.00815,
      "grad_norm": 0.8382113380987876,
      "learning_rate": 0.0024449999999999997,
      "loss": 4.6588,
      "step": 815
    },
    {
      "epoch": 0.00816,
      "grad_norm": 0.8204581966398267,
      "learning_rate": 0.002448,
      "loss": 4.6311,
      "step": 816
    },
    {
      "epoch": 0.00817,
      "grad_norm": 1.02336938073518,
      "learning_rate": 0.002451,
      "loss": 4.6583,
      "step": 817
    },
    {
      "epoch": 0.00818,
      "grad_norm": 0.815478540285485,
      "learning_rate": 0.002454,
      "loss": 4.6309,
      "step": 818
    },
    {
      "epoch": 0.00819,
      "grad_norm": 0.9108711148393207,
      "learning_rate": 0.002457,
      "loss": 4.6483,
      "step": 819
    },
    {
      "epoch": 0.0082,
      "grad_norm": 0.8773364113378127,
      "learning_rate": 0.00246,
      "loss": 4.6373,
      "step": 820
    },
    {
      "epoch": 0.00821,
      "grad_norm": 0.8093045902813614,
      "learning_rate": 0.002463,
      "loss": 4.6126,
      "step": 821
    },
    {
      "epoch": 0.00822,
      "grad_norm": 0.8775561270633004,
      "learning_rate": 0.002466,
      "loss": 4.6164,
      "step": 822
    },
    {
      "epoch": 0.00823,
      "grad_norm": 0.8916321669966187,
      "learning_rate": 0.002469,
      "loss": 4.6298,
      "step": 823
    },
    {
      "epoch": 0.00824,
      "grad_norm": 0.7939051533904264,
      "learning_rate": 0.002472,
      "loss": 4.6511,
      "step": 824
    },
    {
      "epoch": 0.00825,
      "grad_norm": 0.9509206049767348,
      "learning_rate": 0.0024749999999999998,
      "loss": 4.6625,
      "step": 825
    },
    {
      "epoch": 0.00826,
      "grad_norm": 1.0914612729506281,
      "learning_rate": 0.0024779999999999997,
      "loss": 4.6356,
      "step": 826
    },
    {
      "epoch": 0.00827,
      "grad_norm": 1.0440371247088225,
      "learning_rate": 0.002481,
      "loss": 4.6709,
      "step": 827
    },
    {
      "epoch": 0.00828,
      "grad_norm": 1.0292672719671891,
      "learning_rate": 0.002484,
      "loss": 4.6644,
      "step": 828
    },
    {
      "epoch": 0.00829,
      "grad_norm": 1.002549149572049,
      "learning_rate": 0.002487,
      "loss": 4.6608,
      "step": 829
    },
    {
      "epoch": 0.0083,
      "grad_norm": 0.8229445260626227,
      "learning_rate": 0.00249,
      "loss": 4.6642,
      "step": 830
    },
    {
      "epoch": 0.00831,
      "grad_norm": 0.7223475739297199,
      "learning_rate": 0.002493,
      "loss": 4.6379,
      "step": 831
    },
    {
      "epoch": 0.00832,
      "grad_norm": 0.6422365780456449,
      "learning_rate": 0.002496,
      "loss": 4.6349,
      "step": 832
    },
    {
      "epoch": 0.00833,
      "grad_norm": 0.6433276699815419,
      "learning_rate": 0.002499,
      "loss": 4.626,
      "step": 833
    },
    {
      "epoch": 0.00834,
      "grad_norm": 0.6136587911860008,
      "learning_rate": 0.002502,
      "loss": 4.6284,
      "step": 834
    },
    {
      "epoch": 0.00835,
      "grad_norm": 0.5892258230535582,
      "learning_rate": 0.002505,
      "loss": 4.617,
      "step": 835
    },
    {
      "epoch": 0.00836,
      "grad_norm": 0.5407454758774727,
      "learning_rate": 0.002508,
      "loss": 4.6416,
      "step": 836
    },
    {
      "epoch": 0.00837,
      "grad_norm": 0.5840603338652609,
      "learning_rate": 0.0025109999999999998,
      "loss": 4.582,
      "step": 837
    },
    {
      "epoch": 0.00838,
      "grad_norm": 0.5192725472759927,
      "learning_rate": 0.0025139999999999997,
      "loss": 4.6102,
      "step": 838
    },
    {
      "epoch": 0.00839,
      "grad_norm": 0.5064380785759203,
      "learning_rate": 0.002517,
      "loss": 4.6034,
      "step": 839
    },
    {
      "epoch": 0.0084,
      "grad_norm": 0.48476724687493267,
      "learning_rate": 0.00252,
      "loss": 4.5854,
      "step": 840
    },
    {
      "epoch": 0.00841,
      "grad_norm": 0.44496684540968734,
      "learning_rate": 0.002523,
      "loss": 4.5786,
      "step": 841
    },
    {
      "epoch": 0.00842,
      "grad_norm": 0.42609007840223895,
      "learning_rate": 0.002526,
      "loss": 4.5771,
      "step": 842
    },
    {
      "epoch": 0.00843,
      "grad_norm": 0.4703393687667864,
      "learning_rate": 0.002529,
      "loss": 4.5751,
      "step": 843
    },
    {
      "epoch": 0.00844,
      "grad_norm": 0.4976619641323943,
      "learning_rate": 0.002532,
      "loss": 4.5956,
      "step": 844
    },
    {
      "epoch": 0.00845,
      "grad_norm": 0.530881004052944,
      "learning_rate": 0.002535,
      "loss": 4.5625,
      "step": 845
    },
    {
      "epoch": 0.00846,
      "grad_norm": 0.5069253535552343,
      "learning_rate": 0.002538,
      "loss": 4.584,
      "step": 846
    },
    {
      "epoch": 0.00847,
      "grad_norm": 0.4900054637856495,
      "learning_rate": 0.002541,
      "loss": 4.5637,
      "step": 847
    },
    {
      "epoch": 0.00848,
      "grad_norm": 0.5361511355183629,
      "learning_rate": 0.002544,
      "loss": 4.5693,
      "step": 848
    },
    {
      "epoch": 0.00849,
      "grad_norm": 0.6067359238432654,
      "learning_rate": 0.002547,
      "loss": 4.5644,
      "step": 849
    },
    {
      "epoch": 0.0085,
      "grad_norm": 0.5519192846207763,
      "learning_rate": 0.00255,
      "loss": 4.573,
      "step": 850
    },
    {
      "epoch": 0.00851,
      "grad_norm": 0.46694877598438245,
      "learning_rate": 0.002553,
      "loss": 4.5875,
      "step": 851
    },
    {
      "epoch": 0.00852,
      "grad_norm": 0.477565098915178,
      "learning_rate": 0.002556,
      "loss": 4.5765,
      "step": 852
    },
    {
      "epoch": 0.00853,
      "grad_norm": 0.5020213435824815,
      "learning_rate": 0.002559,
      "loss": 4.5556,
      "step": 853
    },
    {
      "epoch": 0.00854,
      "grad_norm": 0.5171409161048013,
      "learning_rate": 0.002562,
      "loss": 4.5495,
      "step": 854
    },
    {
      "epoch": 0.00855,
      "grad_norm": 0.46627459343076927,
      "learning_rate": 0.002565,
      "loss": 4.5252,
      "step": 855
    },
    {
      "epoch": 0.00856,
      "grad_norm": 0.5139521756940325,
      "learning_rate": 0.002568,
      "loss": 4.5623,
      "step": 856
    },
    {
      "epoch": 0.00857,
      "grad_norm": 0.6011403998041547,
      "learning_rate": 0.002571,
      "loss": 4.5577,
      "step": 857
    },
    {
      "epoch": 0.00858,
      "grad_norm": 0.554768384377006,
      "learning_rate": 0.002574,
      "loss": 4.5487,
      "step": 858
    },
    {
      "epoch": 0.00859,
      "grad_norm": 0.539858542755145,
      "learning_rate": 0.002577,
      "loss": 4.5383,
      "step": 859
    },
    {
      "epoch": 0.0086,
      "grad_norm": 0.586599307397693,
      "learning_rate": 0.00258,
      "loss": 4.571,
      "step": 860
    },
    {
      "epoch": 0.00861,
      "grad_norm": 0.7154666400015554,
      "learning_rate": 0.0025830000000000002,
      "loss": 4.5928,
      "step": 861
    },
    {
      "epoch": 0.00862,
      "grad_norm": 0.7345971792604707,
      "learning_rate": 0.002586,
      "loss": 4.5402,
      "step": 862
    },
    {
      "epoch": 0.00863,
      "grad_norm": 0.7491760821516434,
      "learning_rate": 0.002589,
      "loss": 4.5765,
      "step": 863
    },
    {
      "epoch": 0.00864,
      "grad_norm": 0.9255705361922033,
      "learning_rate": 0.002592,
      "loss": 4.5322,
      "step": 864
    },
    {
      "epoch": 0.00865,
      "grad_norm": 0.9964986146275199,
      "learning_rate": 0.002595,
      "loss": 4.5664,
      "step": 865
    },
    {
      "epoch": 0.00866,
      "grad_norm": 0.7618488122087141,
      "learning_rate": 0.002598,
      "loss": 4.5836,
      "step": 866
    },
    {
      "epoch": 0.00867,
      "grad_norm": 0.8524780083566116,
      "learning_rate": 0.002601,
      "loss": 4.5742,
      "step": 867
    },
    {
      "epoch": 0.00868,
      "grad_norm": 0.9692981312410378,
      "learning_rate": 0.002604,
      "loss": 4.5808,
      "step": 868
    },
    {
      "epoch": 0.00869,
      "grad_norm": 1.1822252043975705,
      "learning_rate": 0.002607,
      "loss": 4.6073,
      "step": 869
    },
    {
      "epoch": 0.0087,
      "grad_norm": 0.9057663759386707,
      "learning_rate": 0.00261,
      "loss": 4.5844,
      "step": 870
    },
    {
      "epoch": 0.00871,
      "grad_norm": 0.9457338978675252,
      "learning_rate": 0.002613,
      "loss": 4.6115,
      "step": 871
    },
    {
      "epoch": 0.00872,
      "grad_norm": 0.9845348105394848,
      "learning_rate": 0.002616,
      "loss": 4.5975,
      "step": 872
    },
    {
      "epoch": 0.00873,
      "grad_norm": 0.8202179076205192,
      "learning_rate": 0.0026190000000000002,
      "loss": 4.5967,
      "step": 873
    },
    {
      "epoch": 0.00874,
      "grad_norm": 0.6587988147688274,
      "learning_rate": 0.002622,
      "loss": 4.6142,
      "step": 874
    },
    {
      "epoch": 0.00875,
      "grad_norm": 0.6312495265838277,
      "learning_rate": 0.002625,
      "loss": 4.5549,
      "step": 875
    },
    {
      "epoch": 0.00876,
      "grad_norm": 0.6646817876274769,
      "learning_rate": 0.002628,
      "loss": 4.5838,
      "step": 876
    },
    {
      "epoch": 0.00877,
      "grad_norm": 0.632354886157607,
      "learning_rate": 0.002631,
      "loss": 4.6011,
      "step": 877
    },
    {
      "epoch": 0.00878,
      "grad_norm": 0.5630676163174572,
      "learning_rate": 0.002634,
      "loss": 4.5288,
      "step": 878
    },
    {
      "epoch": 0.00879,
      "grad_norm": 0.4918531988998375,
      "learning_rate": 0.002637,
      "loss": 4.5559,
      "step": 879
    },
    {
      "epoch": 0.0088,
      "grad_norm": 0.42476181684324305,
      "learning_rate": 0.00264,
      "loss": 4.5634,
      "step": 880
    },
    {
      "epoch": 0.00881,
      "grad_norm": 0.4573573466722849,
      "learning_rate": 0.002643,
      "loss": 4.5644,
      "step": 881
    },
    {
      "epoch": 0.00882,
      "grad_norm": 0.5881448674370812,
      "learning_rate": 0.002646,
      "loss": 4.5659,
      "step": 882
    },
    {
      "epoch": 0.00883,
      "grad_norm": 0.7764456560266775,
      "learning_rate": 0.002649,
      "loss": 4.5645,
      "step": 883
    },
    {
      "epoch": 0.00884,
      "grad_norm": 0.9651176541039754,
      "learning_rate": 0.0026520000000000003,
      "loss": 4.5629,
      "step": 884
    },
    {
      "epoch": 0.00885,
      "grad_norm": 1.0453630458113787,
      "learning_rate": 0.0026550000000000002,
      "loss": 4.5675,
      "step": 885
    },
    {
      "epoch": 0.00886,
      "grad_norm": 0.8249295551343052,
      "learning_rate": 0.002658,
      "loss": 4.5811,
      "step": 886
    },
    {
      "epoch": 0.00887,
      "grad_norm": 0.6632170677276661,
      "learning_rate": 0.002661,
      "loss": 4.5643,
      "step": 887
    },
    {
      "epoch": 0.00888,
      "grad_norm": 0.7818922566742896,
      "learning_rate": 0.002664,
      "loss": 4.5709,
      "step": 888
    },
    {
      "epoch": 0.00889,
      "grad_norm": 0.8131057291041344,
      "learning_rate": 0.002667,
      "loss": 4.5276,
      "step": 889
    },
    {
      "epoch": 0.0089,
      "grad_norm": 0.7364786352062309,
      "learning_rate": 0.00267,
      "loss": 4.5735,
      "step": 890
    },
    {
      "epoch": 0.00891,
      "grad_norm": 0.6174568923006037,
      "learning_rate": 0.002673,
      "loss": 4.5538,
      "step": 891
    },
    {
      "epoch": 0.00892,
      "grad_norm": 0.6060396654742667,
      "learning_rate": 0.002676,
      "loss": 4.5678,
      "step": 892
    },
    {
      "epoch": 0.00893,
      "grad_norm": 0.6503337239639668,
      "learning_rate": 0.002679,
      "loss": 4.5453,
      "step": 893
    },
    {
      "epoch": 0.00894,
      "grad_norm": 0.6599395002026207,
      "learning_rate": 0.002682,
      "loss": 4.5291,
      "step": 894
    },
    {
      "epoch": 0.00895,
      "grad_norm": 0.5989877186645693,
      "learning_rate": 0.0026850000000000003,
      "loss": 4.5412,
      "step": 895
    },
    {
      "epoch": 0.00896,
      "grad_norm": 0.5286031214975206,
      "learning_rate": 0.0026880000000000003,
      "loss": 4.5273,
      "step": 896
    },
    {
      "epoch": 0.00897,
      "grad_norm": 0.6246596167729576,
      "learning_rate": 0.0026910000000000002,
      "loss": 4.5504,
      "step": 897
    },
    {
      "epoch": 0.00898,
      "grad_norm": 0.6886920087577523,
      "learning_rate": 0.002694,
      "loss": 4.5437,
      "step": 898
    },
    {
      "epoch": 0.00899,
      "grad_norm": 0.7603324493631337,
      "learning_rate": 0.002697,
      "loss": 4.5543,
      "step": 899
    },
    {
      "epoch": 0.009,
      "grad_norm": 0.7773743953648492,
      "learning_rate": 0.0027,
      "loss": 4.5794,
      "step": 900
    },
    {
      "epoch": 0.00901,
      "grad_norm": 0.683256197441996,
      "learning_rate": 0.002703,
      "loss": 4.5307,
      "step": 901
    },
    {
      "epoch": 0.00902,
      "grad_norm": 0.5681357763332335,
      "learning_rate": 0.002706,
      "loss": 4.5356,
      "step": 902
    },
    {
      "epoch": 0.00903,
      "grad_norm": 0.5420591540444755,
      "learning_rate": 0.002709,
      "loss": 4.5338,
      "step": 903
    },
    {
      "epoch": 0.00904,
      "grad_norm": 0.5224631659490503,
      "learning_rate": 0.002712,
      "loss": 4.5093,
      "step": 904
    },
    {
      "epoch": 0.00905,
      "grad_norm": 0.5026034590467293,
      "learning_rate": 0.002715,
      "loss": 4.5252,
      "step": 905
    },
    {
      "epoch": 0.00906,
      "grad_norm": 0.5177890071237494,
      "learning_rate": 0.002718,
      "loss": 4.5378,
      "step": 906
    },
    {
      "epoch": 0.00907,
      "grad_norm": 0.5764689015080159,
      "learning_rate": 0.0027210000000000003,
      "loss": 4.5536,
      "step": 907
    },
    {
      "epoch": 0.00908,
      "grad_norm": 0.6259624722487185,
      "learning_rate": 0.0027240000000000003,
      "loss": 4.5265,
      "step": 908
    },
    {
      "epoch": 0.00909,
      "grad_norm": 0.670091172363038,
      "learning_rate": 0.0027270000000000003,
      "loss": 4.5481,
      "step": 909
    },
    {
      "epoch": 0.0091,
      "grad_norm": 0.7211417475777565,
      "learning_rate": 0.0027300000000000002,
      "loss": 4.555,
      "step": 910
    },
    {
      "epoch": 0.00911,
      "grad_norm": 0.6734826041799787,
      "learning_rate": 0.002733,
      "loss": 4.5118,
      "step": 911
    },
    {
      "epoch": 0.00912,
      "grad_norm": 0.5721394003951694,
      "learning_rate": 0.002736,
      "loss": 4.5134,
      "step": 912
    },
    {
      "epoch": 0.00913,
      "grad_norm": 0.5576363751689392,
      "learning_rate": 0.002739,
      "loss": 4.5062,
      "step": 913
    },
    {
      "epoch": 0.00914,
      "grad_norm": 0.659556953854551,
      "learning_rate": 0.002742,
      "loss": 4.5426,
      "step": 914
    },
    {
      "epoch": 0.00915,
      "grad_norm": 0.7056386423863461,
      "learning_rate": 0.002745,
      "loss": 4.5041,
      "step": 915
    },
    {
      "epoch": 0.00916,
      "grad_norm": 0.6615396800381155,
      "learning_rate": 0.002748,
      "loss": 4.4829,
      "step": 916
    },
    {
      "epoch": 0.00917,
      "grad_norm": 0.6203895946828626,
      "learning_rate": 0.002751,
      "loss": 4.4952,
      "step": 917
    },
    {
      "epoch": 0.00918,
      "grad_norm": 0.6455523293432982,
      "learning_rate": 0.0027540000000000004,
      "loss": 4.5159,
      "step": 918
    },
    {
      "epoch": 0.00919,
      "grad_norm": 0.6153975854015812,
      "learning_rate": 0.0027570000000000003,
      "loss": 4.4956,
      "step": 919
    },
    {
      "epoch": 0.0092,
      "grad_norm": 0.6747074236078813,
      "learning_rate": 0.0027600000000000003,
      "loss": 4.516,
      "step": 920
    },
    {
      "epoch": 0.00921,
      "grad_norm": 0.7525938258933302,
      "learning_rate": 0.0027630000000000003,
      "loss": 4.5035,
      "step": 921
    },
    {
      "epoch": 0.00922,
      "grad_norm": 0.6613679668687953,
      "learning_rate": 0.0027660000000000002,
      "loss": 4.5161,
      "step": 922
    },
    {
      "epoch": 0.00923,
      "grad_norm": 0.657751627761755,
      "learning_rate": 0.002769,
      "loss": 4.4882,
      "step": 923
    },
    {
      "epoch": 0.00924,
      "grad_norm": 0.6747361910557791,
      "learning_rate": 0.002772,
      "loss": 4.515,
      "step": 924
    },
    {
      "epoch": 0.00925,
      "grad_norm": 0.6239646474738244,
      "learning_rate": 0.002775,
      "loss": 4.4929,
      "step": 925
    },
    {
      "epoch": 0.00926,
      "grad_norm": 0.599067117804374,
      "learning_rate": 0.002778,
      "loss": 4.5097,
      "step": 926
    },
    {
      "epoch": 0.00927,
      "grad_norm": 0.5594951339370652,
      "learning_rate": 0.002781,
      "loss": 4.4719,
      "step": 927
    },
    {
      "epoch": 0.00928,
      "grad_norm": 0.6063845847981192,
      "learning_rate": 0.002784,
      "loss": 4.518,
      "step": 928
    },
    {
      "epoch": 0.00929,
      "grad_norm": 0.6363503981385901,
      "learning_rate": 0.0027870000000000004,
      "loss": 4.5031,
      "step": 929
    },
    {
      "epoch": 0.0093,
      "grad_norm": 0.7396258221206569,
      "learning_rate": 0.0027900000000000004,
      "loss": 4.4944,
      "step": 930
    },
    {
      "epoch": 0.00931,
      "grad_norm": 0.8942550404249334,
      "learning_rate": 0.0027930000000000003,
      "loss": 4.517,
      "step": 931
    },
    {
      "epoch": 0.00932,
      "grad_norm": 1.0354660733966428,
      "learning_rate": 0.0027960000000000003,
      "loss": 4.5402,
      "step": 932
    },
    {
      "epoch": 0.00933,
      "grad_norm": 1.180367237398422,
      "learning_rate": 0.0027990000000000003,
      "loss": 4.525,
      "step": 933
    },
    {
      "epoch": 0.00934,
      "grad_norm": 0.9502642927196222,
      "learning_rate": 0.0028020000000000002,
      "loss": 4.5541,
      "step": 934
    },
    {
      "epoch": 0.00935,
      "grad_norm": 0.8858808588378486,
      "learning_rate": 0.002805,
      "loss": 4.5601,
      "step": 935
    },
    {
      "epoch": 0.00936,
      "grad_norm": 0.9244572615029755,
      "learning_rate": 0.002808,
      "loss": 4.569,
      "step": 936
    },
    {
      "epoch": 0.00937,
      "grad_norm": 0.9225067704838915,
      "learning_rate": 0.002811,
      "loss": 4.5806,
      "step": 937
    },
    {
      "epoch": 0.00938,
      "grad_norm": 0.8534895885659719,
      "learning_rate": 0.002814,
      "loss": 4.5604,
      "step": 938
    },
    {
      "epoch": 0.00939,
      "grad_norm": 0.9046137087836131,
      "learning_rate": 0.002817,
      "loss": 4.5554,
      "step": 939
    },
    {
      "epoch": 0.0094,
      "grad_norm": 0.7476401969651744,
      "learning_rate": 0.00282,
      "loss": 4.5482,
      "step": 940
    },
    {
      "epoch": 0.00941,
      "grad_norm": 0.7066455268972154,
      "learning_rate": 0.002823,
      "loss": 4.5225,
      "step": 941
    },
    {
      "epoch": 0.00942,
      "grad_norm": 0.6459119835251312,
      "learning_rate": 0.002826,
      "loss": 4.544,
      "step": 942
    },
    {
      "epoch": 0.00943,
      "grad_norm": 0.6055625869260791,
      "learning_rate": 0.002829,
      "loss": 4.5363,
      "step": 943
    },
    {
      "epoch": 0.00944,
      "grad_norm": 0.5293280202161804,
      "learning_rate": 0.002832,
      "loss": 4.5307,
      "step": 944
    },
    {
      "epoch": 0.00945,
      "grad_norm": 0.5438831110130244,
      "learning_rate": 0.002835,
      "loss": 4.5334,
      "step": 945
    },
    {
      "epoch": 0.00946,
      "grad_norm": 0.4766420743521973,
      "learning_rate": 0.002838,
      "loss": 4.5433,
      "step": 946
    },
    {
      "epoch": 0.00947,
      "grad_norm": 0.40195567884756706,
      "learning_rate": 0.0028409999999999998,
      "loss": 4.5081,
      "step": 947
    },
    {
      "epoch": 0.00948,
      "grad_norm": 0.3783844921089427,
      "learning_rate": 0.0028439999999999997,
      "loss": 4.4865,
      "step": 948
    },
    {
      "epoch": 0.00949,
      "grad_norm": 0.4197576569837563,
      "learning_rate": 0.002847,
      "loss": 4.5306,
      "step": 949
    },
    {
      "epoch": 0.0095,
      "grad_norm": 0.45947372263331304,
      "learning_rate": 0.00285,
      "loss": 4.5193,
      "step": 950
    },
    {
      "epoch": 0.00951,
      "grad_norm": 0.5187245758366383,
      "learning_rate": 0.002853,
      "loss": 4.4969,
      "step": 951
    },
    {
      "epoch": 0.00952,
      "grad_norm": 0.5515692080168162,
      "learning_rate": 0.002856,
      "loss": 4.5218,
      "step": 952
    },
    {
      "epoch": 0.00953,
      "grad_norm": 0.501582875002041,
      "learning_rate": 0.002859,
      "loss": 4.4606,
      "step": 953
    },
    {
      "epoch": 0.00954,
      "grad_norm": 0.5014106294436917,
      "learning_rate": 0.002862,
      "loss": 4.5197,
      "step": 954
    },
    {
      "epoch": 0.00955,
      "grad_norm": 0.6047606934565909,
      "learning_rate": 0.002865,
      "loss": 4.5086,
      "step": 955
    },
    {
      "epoch": 0.00956,
      "grad_norm": 0.6661868633369662,
      "learning_rate": 0.002868,
      "loss": 4.4921,
      "step": 956
    },
    {
      "epoch": 0.00957,
      "grad_norm": 0.6511713371124522,
      "learning_rate": 0.002871,
      "loss": 4.514,
      "step": 957
    },
    {
      "epoch": 0.00958,
      "grad_norm": 0.5733443203887492,
      "learning_rate": 0.002874,
      "loss": 4.4931,
      "step": 958
    },
    {
      "epoch": 0.00959,
      "grad_norm": 0.6024952806359369,
      "learning_rate": 0.002877,
      "loss": 4.4895,
      "step": 959
    },
    {
      "epoch": 0.0096,
      "grad_norm": 0.6029559818977924,
      "learning_rate": 0.0028799999999999997,
      "loss": 4.4868,
      "step": 960
    },
    {
      "epoch": 0.00961,
      "grad_norm": 0.5721073369283843,
      "learning_rate": 0.002883,
      "loss": 4.4604,
      "step": 961
    },
    {
      "epoch": 0.00962,
      "grad_norm": 0.5737900491823522,
      "learning_rate": 0.002886,
      "loss": 4.4898,
      "step": 962
    },
    {
      "epoch": 0.00963,
      "grad_norm": 0.5323481251626608,
      "learning_rate": 0.002889,
      "loss": 4.4867,
      "step": 963
    },
    {
      "epoch": 0.00964,
      "grad_norm": 0.5436801325002781,
      "learning_rate": 0.002892,
      "loss": 4.4807,
      "step": 964
    },
    {
      "epoch": 0.00965,
      "grad_norm": 0.60229729083351,
      "learning_rate": 0.002895,
      "loss": 4.4568,
      "step": 965
    },
    {
      "epoch": 0.00966,
      "grad_norm": 0.6629818387101306,
      "learning_rate": 0.002898,
      "loss": 4.4766,
      "step": 966
    },
    {
      "epoch": 0.00967,
      "grad_norm": 0.6748155978904155,
      "learning_rate": 0.002901,
      "loss": 4.5156,
      "step": 967
    },
    {
      "epoch": 0.00968,
      "grad_norm": 0.7427494012599226,
      "learning_rate": 0.002904,
      "loss": 4.4866,
      "step": 968
    },
    {
      "epoch": 0.00969,
      "grad_norm": 0.8794961931178971,
      "learning_rate": 0.002907,
      "loss": 4.5273,
      "step": 969
    },
    {
      "epoch": 0.0097,
      "grad_norm": 0.8586008854691127,
      "learning_rate": 0.00291,
      "loss": 4.4896,
      "step": 970
    },
    {
      "epoch": 0.00971,
      "grad_norm": 0.8273563438543869,
      "learning_rate": 0.002913,
      "loss": 4.4955,
      "step": 971
    },
    {
      "epoch": 0.00972,
      "grad_norm": 0.7536097688784559,
      "learning_rate": 0.002916,
      "loss": 4.5029,
      "step": 972
    },
    {
      "epoch": 0.00973,
      "grad_norm": 0.7541845251322323,
      "learning_rate": 0.002919,
      "loss": 4.4985,
      "step": 973
    },
    {
      "epoch": 0.00974,
      "grad_norm": 0.6473607436694337,
      "learning_rate": 0.002922,
      "loss": 4.4796,
      "step": 974
    },
    {
      "epoch": 0.00975,
      "grad_norm": 0.7361402706574074,
      "learning_rate": 0.002925,
      "loss": 4.5044,
      "step": 975
    },
    {
      "epoch": 0.00976,
      "grad_norm": 0.868936228763895,
      "learning_rate": 0.002928,
      "loss": 4.508,
      "step": 976
    },
    {
      "epoch": 0.00977,
      "grad_norm": 0.8813019375073942,
      "learning_rate": 0.002931,
      "loss": 4.5408,
      "step": 977
    },
    {
      "epoch": 0.00978,
      "grad_norm": 0.9426880168937273,
      "learning_rate": 0.002934,
      "loss": 4.5239,
      "step": 978
    },
    {
      "epoch": 0.00979,
      "grad_norm": 0.8776049562434768,
      "learning_rate": 0.002937,
      "loss": 4.5177,
      "step": 979
    },
    {
      "epoch": 0.0098,
      "grad_norm": 0.7621714209410982,
      "learning_rate": 0.00294,
      "loss": 4.5005,
      "step": 980
    },
    {
      "epoch": 0.00981,
      "grad_norm": 0.7607321859563556,
      "learning_rate": 0.002943,
      "loss": 4.5183,
      "step": 981
    },
    {
      "epoch": 0.00982,
      "grad_norm": 0.8148690145722087,
      "learning_rate": 0.002946,
      "loss": 4.5101,
      "step": 982
    },
    {
      "epoch": 0.00983,
      "grad_norm": 0.8602879224908239,
      "learning_rate": 0.0029490000000000002,
      "loss": 4.5186,
      "step": 983
    },
    {
      "epoch": 0.00984,
      "grad_norm": 0.9348586711938943,
      "learning_rate": 0.002952,
      "loss": 4.5105,
      "step": 984
    },
    {
      "epoch": 0.00985,
      "grad_norm": 0.9010607870226212,
      "learning_rate": 0.002955,
      "loss": 4.5341,
      "step": 985
    },
    {
      "epoch": 0.00986,
      "grad_norm": 0.8225410280635316,
      "learning_rate": 0.002958,
      "loss": 4.497,
      "step": 986
    },
    {
      "epoch": 0.00987,
      "grad_norm": 0.820560994458863,
      "learning_rate": 0.002961,
      "loss": 4.5111,
      "step": 987
    },
    {
      "epoch": 0.00988,
      "grad_norm": 0.7430257271274537,
      "learning_rate": 0.002964,
      "loss": 4.5437,
      "step": 988
    },
    {
      "epoch": 0.00989,
      "grad_norm": 0.7193873820034543,
      "learning_rate": 0.002967,
      "loss": 4.4838,
      "step": 989
    },
    {
      "epoch": 0.0099,
      "grad_norm": 0.7329220852773792,
      "learning_rate": 0.00297,
      "loss": 4.5219,
      "step": 990
    },
    {
      "epoch": 0.00991,
      "grad_norm": 0.7694030006138932,
      "learning_rate": 0.002973,
      "loss": 4.5213,
      "step": 991
    },
    {
      "epoch": 0.00992,
      "grad_norm": 0.7726534251991994,
      "learning_rate": 0.002976,
      "loss": 4.5153,
      "step": 992
    },
    {
      "epoch": 0.00993,
      "grad_norm": 0.6817474065224322,
      "learning_rate": 0.002979,
      "loss": 4.511,
      "step": 993
    },
    {
      "epoch": 0.00994,
      "grad_norm": 0.6628189744120299,
      "learning_rate": 0.002982,
      "loss": 4.5078,
      "step": 994
    },
    {
      "epoch": 0.00995,
      "grad_norm": 0.6249124245549155,
      "learning_rate": 0.0029850000000000002,
      "loss": 4.5069,
      "step": 995
    },
    {
      "epoch": 0.00996,
      "grad_norm": 0.54278083452404,
      "learning_rate": 0.002988,
      "loss": 4.503,
      "step": 996
    },
    {
      "epoch": 0.00997,
      "grad_norm": 0.5131542547273349,
      "learning_rate": 0.002991,
      "loss": 4.49,
      "step": 997
    },
    {
      "epoch": 0.00998,
      "grad_norm": 0.4760166868407609,
      "learning_rate": 0.002994,
      "loss": 4.4895,
      "step": 998
    },
    {
      "epoch": 0.00999,
      "grad_norm": 0.4855483581267517,
      "learning_rate": 0.002997,
      "loss": 4.4707,
      "step": 999
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.44929338024832627,
      "learning_rate": 0.003,
      "loss": 4.4897,
      "step": 1000
    },
    {
      "epoch": 0.01001,
      "grad_norm": 0.3731978010457433,
      "learning_rate": 0.003,
      "loss": 4.4851,
      "step": 1001
    },
    {
      "epoch": 0.01002,
      "grad_norm": 0.37113077487631946,
      "learning_rate": 0.003,
      "loss": 4.4516,
      "step": 1002
    },
    {
      "epoch": 0.01003,
      "grad_norm": 0.35590481029592896,
      "learning_rate": 0.003,
      "loss": 4.4696,
      "step": 1003
    },
    {
      "epoch": 0.01004,
      "grad_norm": 0.36275775448168,
      "learning_rate": 0.003,
      "loss": 4.4605,
      "step": 1004
    },
    {
      "epoch": 0.01005,
      "grad_norm": 0.43407423243746507,
      "learning_rate": 0.003,
      "loss": 4.4329,
      "step": 1005
    },
    {
      "epoch": 0.01006,
      "grad_norm": 0.5622897072513495,
      "learning_rate": 0.003,
      "loss": 4.4378,
      "step": 1006
    },
    {
      "epoch": 0.01007,
      "grad_norm": 0.742851457120247,
      "learning_rate": 0.003,
      "loss": 4.4915,
      "step": 1007
    },
    {
      "epoch": 0.01008,
      "grad_norm": 0.6898510838085624,
      "learning_rate": 0.003,
      "loss": 4.5119,
      "step": 1008
    },
    {
      "epoch": 0.01009,
      "grad_norm": 0.5850313111790646,
      "learning_rate": 0.003,
      "loss": 4.4686,
      "step": 1009
    },
    {
      "epoch": 0.0101,
      "grad_norm": 0.6689327239491644,
      "learning_rate": 0.003,
      "loss": 4.4833,
      "step": 1010
    },
    {
      "epoch": 0.01011,
      "grad_norm": 0.6177431531224691,
      "learning_rate": 0.003,
      "loss": 4.4907,
      "step": 1011
    },
    {
      "epoch": 0.01012,
      "grad_norm": 0.529876877427535,
      "learning_rate": 0.003,
      "loss": 4.4694,
      "step": 1012
    },
    {
      "epoch": 0.01013,
      "grad_norm": 0.5027890853405218,
      "learning_rate": 0.003,
      "loss": 4.4915,
      "step": 1013
    },
    {
      "epoch": 0.01014,
      "grad_norm": 0.4606528826158309,
      "learning_rate": 0.003,
      "loss": 4.4311,
      "step": 1014
    },
    {
      "epoch": 0.01015,
      "grad_norm": 0.43413097561878167,
      "learning_rate": 0.003,
      "loss": 4.4729,
      "step": 1015
    },
    {
      "epoch": 0.01016,
      "grad_norm": 0.45162191517129524,
      "learning_rate": 0.003,
      "loss": 4.4813,
      "step": 1016
    },
    {
      "epoch": 0.01017,
      "grad_norm": 0.4151722379204413,
      "learning_rate": 0.003,
      "loss": 4.4194,
      "step": 1017
    },
    {
      "epoch": 0.01018,
      "grad_norm": 0.42684710143489196,
      "learning_rate": 0.003,
      "loss": 4.4738,
      "step": 1018
    },
    {
      "epoch": 0.01019,
      "grad_norm": 0.49455469810835373,
      "learning_rate": 0.003,
      "loss": 4.4546,
      "step": 1019
    },
    {
      "epoch": 0.0102,
      "grad_norm": 0.44331421163833096,
      "learning_rate": 0.003,
      "loss": 4.4504,
      "step": 1020
    },
    {
      "epoch": 0.01021,
      "grad_norm": 0.44293820189086996,
      "learning_rate": 0.003,
      "loss": 4.4351,
      "step": 1021
    },
    {
      "epoch": 0.01022,
      "grad_norm": 0.471018484914238,
      "learning_rate": 0.003,
      "loss": 4.4454,
      "step": 1022
    },
    {
      "epoch": 0.01023,
      "grad_norm": 0.5245843463151362,
      "learning_rate": 0.003,
      "loss": 4.4439,
      "step": 1023
    },
    {
      "epoch": 0.01024,
      "grad_norm": 0.5471345601505091,
      "learning_rate": 0.003,
      "loss": 4.4059,
      "step": 1024
    },
    {
      "epoch": 0.01025,
      "grad_norm": 0.5561682271484449,
      "learning_rate": 0.003,
      "loss": 4.4357,
      "step": 1025
    },
    {
      "epoch": 0.01026,
      "grad_norm": 0.6750941633687737,
      "learning_rate": 0.003,
      "loss": 4.4343,
      "step": 1026
    },
    {
      "epoch": 0.01027,
      "grad_norm": 0.6614940222952833,
      "learning_rate": 0.003,
      "loss": 4.4593,
      "step": 1027
    },
    {
      "epoch": 0.01028,
      "grad_norm": 0.8232761879596748,
      "learning_rate": 0.003,
      "loss": 4.4444,
      "step": 1028
    },
    {
      "epoch": 0.01029,
      "grad_norm": 0.926962960927152,
      "learning_rate": 0.003,
      "loss": 4.4584,
      "step": 1029
    },
    {
      "epoch": 0.0103,
      "grad_norm": 0.7880165390972924,
      "learning_rate": 0.003,
      "loss": 4.4737,
      "step": 1030
    },
    {
      "epoch": 0.01031,
      "grad_norm": 0.8244417100638626,
      "learning_rate": 0.003,
      "loss": 4.493,
      "step": 1031
    },
    {
      "epoch": 0.01032,
      "grad_norm": 0.8457876785075499,
      "learning_rate": 0.003,
      "loss": 4.4747,
      "step": 1032
    },
    {
      "epoch": 0.01033,
      "grad_norm": 0.7537115798125514,
      "learning_rate": 0.003,
      "loss": 4.5075,
      "step": 1033
    },
    {
      "epoch": 0.01034,
      "grad_norm": 0.7955188357741166,
      "learning_rate": 0.003,
      "loss": 4.4759,
      "step": 1034
    },
    {
      "epoch": 0.01035,
      "grad_norm": 0.807776556733491,
      "learning_rate": 0.003,
      "loss": 4.4793,
      "step": 1035
    },
    {
      "epoch": 0.01036,
      "grad_norm": 0.8192211538243458,
      "learning_rate": 0.003,
      "loss": 4.4746,
      "step": 1036
    },
    {
      "epoch": 0.01037,
      "grad_norm": 0.8125711495155172,
      "learning_rate": 0.003,
      "loss": 4.5016,
      "step": 1037
    },
    {
      "epoch": 0.01038,
      "grad_norm": 0.8698834270185546,
      "learning_rate": 0.003,
      "loss": 4.5079,
      "step": 1038
    },
    {
      "epoch": 0.01039,
      "grad_norm": 0.8070907960047083,
      "learning_rate": 0.003,
      "loss": 4.4962,
      "step": 1039
    },
    {
      "epoch": 0.0104,
      "grad_norm": 0.8379908956267174,
      "learning_rate": 0.003,
      "loss": 4.5113,
      "step": 1040
    },
    {
      "epoch": 0.01041,
      "grad_norm": 0.9467065153215795,
      "learning_rate": 0.003,
      "loss": 4.4994,
      "step": 1041
    },
    {
      "epoch": 0.01042,
      "grad_norm": 1.0844879341171343,
      "learning_rate": 0.003,
      "loss": 4.5455,
      "step": 1042
    },
    {
      "epoch": 0.01043,
      "grad_norm": 0.8223563094033745,
      "learning_rate": 0.003,
      "loss": 4.4732,
      "step": 1043
    },
    {
      "epoch": 0.01044,
      "grad_norm": 0.7034538108566051,
      "learning_rate": 0.003,
      "loss": 4.5118,
      "step": 1044
    },
    {
      "epoch": 0.01045,
      "grad_norm": 0.7046831935605008,
      "learning_rate": 0.003,
      "loss": 4.5282,
      "step": 1045
    },
    {
      "epoch": 0.01046,
      "grad_norm": 0.5291717642722772,
      "learning_rate": 0.003,
      "loss": 4.4864,
      "step": 1046
    },
    {
      "epoch": 0.01047,
      "grad_norm": 0.517981866453337,
      "learning_rate": 0.003,
      "loss": 4.4886,
      "step": 1047
    },
    {
      "epoch": 0.01048,
      "grad_norm": 0.5100897334283181,
      "learning_rate": 0.003,
      "loss": 4.4775,
      "step": 1048
    },
    {
      "epoch": 0.01049,
      "grad_norm": 0.4646755867304285,
      "learning_rate": 0.003,
      "loss": 4.4516,
      "step": 1049
    },
    {
      "epoch": 0.0105,
      "grad_norm": 0.4688262914765259,
      "learning_rate": 0.003,
      "loss": 4.491,
      "step": 1050
    },
    {
      "epoch": 0.01051,
      "grad_norm": 0.40479473066275506,
      "learning_rate": 0.003,
      "loss": 4.4613,
      "step": 1051
    },
    {
      "epoch": 0.01052,
      "grad_norm": 0.37634413468362676,
      "learning_rate": 0.003,
      "loss": 4.4718,
      "step": 1052
    },
    {
      "epoch": 0.01053,
      "grad_norm": 0.35380747787174,
      "learning_rate": 0.003,
      "loss": 4.4768,
      "step": 1053
    },
    {
      "epoch": 0.01054,
      "grad_norm": 0.3233101912432746,
      "learning_rate": 0.003,
      "loss": 4.4657,
      "step": 1054
    },
    {
      "epoch": 0.01055,
      "grad_norm": 0.3051647539319004,
      "learning_rate": 0.003,
      "loss": 4.4412,
      "step": 1055
    },
    {
      "epoch": 0.01056,
      "grad_norm": 0.3183252525248436,
      "learning_rate": 0.003,
      "loss": 4.4395,
      "step": 1056
    },
    {
      "epoch": 0.01057,
      "grad_norm": 0.32244230779772487,
      "learning_rate": 0.003,
      "loss": 4.443,
      "step": 1057
    },
    {
      "epoch": 0.01058,
      "grad_norm": 0.37190643160876063,
      "learning_rate": 0.003,
      "loss": 4.4376,
      "step": 1058
    },
    {
      "epoch": 0.01059,
      "grad_norm": 0.4982331351188646,
      "learning_rate": 0.003,
      "loss": 4.4406,
      "step": 1059
    },
    {
      "epoch": 0.0106,
      "grad_norm": 0.630051966003807,
      "learning_rate": 0.003,
      "loss": 4.4377,
      "step": 1060
    },
    {
      "epoch": 0.01061,
      "grad_norm": 0.6685664249418685,
      "learning_rate": 0.003,
      "loss": 4.4357,
      "step": 1061
    },
    {
      "epoch": 0.01062,
      "grad_norm": 0.5702779525542006,
      "learning_rate": 0.003,
      "loss": 4.4281,
      "step": 1062
    },
    {
      "epoch": 0.01063,
      "grad_norm": 0.47516871800908467,
      "learning_rate": 0.003,
      "loss": 4.4193,
      "step": 1063
    },
    {
      "epoch": 0.01064,
      "grad_norm": 0.5362466950028499,
      "learning_rate": 0.003,
      "loss": 4.4381,
      "step": 1064
    },
    {
      "epoch": 0.01065,
      "grad_norm": 0.5373139287833821,
      "learning_rate": 0.003,
      "loss": 4.4375,
      "step": 1065
    },
    {
      "epoch": 0.01066,
      "grad_norm": 0.5486100821685685,
      "learning_rate": 0.003,
      "loss": 4.44,
      "step": 1066
    },
    {
      "epoch": 0.01067,
      "grad_norm": 0.4816436794229303,
      "learning_rate": 0.003,
      "loss": 4.4454,
      "step": 1067
    },
    {
      "epoch": 0.01068,
      "grad_norm": 0.429069056402875,
      "learning_rate": 0.003,
      "loss": 4.4286,
      "step": 1068
    },
    {
      "epoch": 0.01069,
      "grad_norm": 0.4352935464081401,
      "learning_rate": 0.003,
      "loss": 4.4347,
      "step": 1069
    },
    {
      "epoch": 0.0107,
      "grad_norm": 0.4882626803662278,
      "learning_rate": 0.003,
      "loss": 4.4874,
      "step": 1070
    },
    {
      "epoch": 0.01071,
      "grad_norm": 0.599919651347176,
      "learning_rate": 0.003,
      "loss": 4.4109,
      "step": 1071
    },
    {
      "epoch": 0.01072,
      "grad_norm": 0.6067483783044649,
      "learning_rate": 0.003,
      "loss": 4.4536,
      "step": 1072
    },
    {
      "epoch": 0.01073,
      "grad_norm": 0.49239058441195105,
      "learning_rate": 0.003,
      "loss": 4.4412,
      "step": 1073
    },
    {
      "epoch": 0.01074,
      "grad_norm": 0.49249754408010815,
      "learning_rate": 0.003,
      "loss": 4.4068,
      "step": 1074
    },
    {
      "epoch": 0.01075,
      "grad_norm": 0.4103440289244663,
      "learning_rate": 0.003,
      "loss": 4.4418,
      "step": 1075
    },
    {
      "epoch": 0.01076,
      "grad_norm": 0.41435198540181395,
      "learning_rate": 0.003,
      "loss": 4.4375,
      "step": 1076
    },
    {
      "epoch": 0.01077,
      "grad_norm": 0.42961621912720355,
      "learning_rate": 0.003,
      "loss": 4.4322,
      "step": 1077
    },
    {
      "epoch": 0.01078,
      "grad_norm": 0.47858570596245864,
      "learning_rate": 0.003,
      "loss": 4.3877,
      "step": 1078
    },
    {
      "epoch": 0.01079,
      "grad_norm": 0.5533382064380803,
      "learning_rate": 0.003,
      "loss": 4.4233,
      "step": 1079
    },
    {
      "epoch": 0.0108,
      "grad_norm": 0.6417480625144324,
      "learning_rate": 0.003,
      "loss": 4.4054,
      "step": 1080
    },
    {
      "epoch": 0.01081,
      "grad_norm": 0.7431146939801978,
      "learning_rate": 0.003,
      "loss": 4.4226,
      "step": 1081
    },
    {
      "epoch": 0.01082,
      "grad_norm": 0.9014359739424833,
      "learning_rate": 0.003,
      "loss": 4.4534,
      "step": 1082
    },
    {
      "epoch": 0.01083,
      "grad_norm": 0.8675952653065004,
      "learning_rate": 0.003,
      "loss": 4.4518,
      "step": 1083
    },
    {
      "epoch": 0.01084,
      "grad_norm": 0.7255782707521201,
      "learning_rate": 0.003,
      "loss": 4.448,
      "step": 1084
    },
    {
      "epoch": 0.01085,
      "grad_norm": 0.6450755897364046,
      "learning_rate": 0.003,
      "loss": 4.4349,
      "step": 1085
    },
    {
      "epoch": 0.01086,
      "grad_norm": 0.6814430736440221,
      "learning_rate": 0.003,
      "loss": 4.4125,
      "step": 1086
    },
    {
      "epoch": 0.01087,
      "grad_norm": 0.6399587640180319,
      "learning_rate": 0.003,
      "loss": 4.4286,
      "step": 1087
    },
    {
      "epoch": 0.01088,
      "grad_norm": 0.5625590086017324,
      "learning_rate": 0.003,
      "loss": 4.4612,
      "step": 1088
    },
    {
      "epoch": 0.01089,
      "grad_norm": 0.5476518731942144,
      "learning_rate": 0.003,
      "loss": 4.4282,
      "step": 1089
    },
    {
      "epoch": 0.0109,
      "grad_norm": 0.6962765199988842,
      "learning_rate": 0.003,
      "loss": 4.4603,
      "step": 1090
    },
    {
      "epoch": 0.01091,
      "grad_norm": 0.7992579884725848,
      "learning_rate": 0.003,
      "loss": 4.4391,
      "step": 1091
    },
    {
      "epoch": 0.01092,
      "grad_norm": 0.6922142003261247,
      "learning_rate": 0.003,
      "loss": 4.43,
      "step": 1092
    },
    {
      "epoch": 0.01093,
      "grad_norm": 0.7201776431280809,
      "learning_rate": 0.003,
      "loss": 4.4545,
      "step": 1093
    },
    {
      "epoch": 0.01094,
      "grad_norm": 0.7324411775727792,
      "learning_rate": 0.003,
      "loss": 4.4431,
      "step": 1094
    },
    {
      "epoch": 0.01095,
      "grad_norm": 0.7356072077097602,
      "learning_rate": 0.003,
      "loss": 4.4704,
      "step": 1095
    },
    {
      "epoch": 0.01096,
      "grad_norm": 0.7577305220019848,
      "learning_rate": 0.003,
      "loss": 4.445,
      "step": 1096
    },
    {
      "epoch": 0.01097,
      "grad_norm": 0.7290952400289119,
      "learning_rate": 0.003,
      "loss": 4.4352,
      "step": 1097
    },
    {
      "epoch": 0.01098,
      "grad_norm": 0.8085780813837044,
      "learning_rate": 0.003,
      "loss": 4.4555,
      "step": 1098
    },
    {
      "epoch": 0.01099,
      "grad_norm": 0.8944752292785944,
      "learning_rate": 0.003,
      "loss": 4.4361,
      "step": 1099
    },
    {
      "epoch": 0.011,
      "grad_norm": 0.7875559243106586,
      "learning_rate": 0.003,
      "loss": 4.4348,
      "step": 1100
    },
    {
      "epoch": 0.01101,
      "grad_norm": 0.8589536043002628,
      "learning_rate": 0.003,
      "loss": 4.4539,
      "step": 1101
    },
    {
      "epoch": 0.01102,
      "grad_norm": 0.8464001238353072,
      "learning_rate": 0.003,
      "loss": 4.4762,
      "step": 1102
    },
    {
      "epoch": 0.01103,
      "grad_norm": 0.8851736378981229,
      "learning_rate": 0.003,
      "loss": 4.4662,
      "step": 1103
    },
    {
      "epoch": 0.01104,
      "grad_norm": 0.7153611537759403,
      "learning_rate": 0.003,
      "loss": 4.4545,
      "step": 1104
    },
    {
      "epoch": 0.01105,
      "grad_norm": 0.5180783683134489,
      "learning_rate": 0.003,
      "loss": 4.4659,
      "step": 1105
    },
    {
      "epoch": 0.01106,
      "grad_norm": 0.5726422216533474,
      "learning_rate": 0.003,
      "loss": 4.4387,
      "step": 1106
    },
    {
      "epoch": 0.01107,
      "grad_norm": 0.5702919336798088,
      "learning_rate": 0.003,
      "loss": 4.4468,
      "step": 1107
    },
    {
      "epoch": 0.01108,
      "grad_norm": 0.48082946780834995,
      "learning_rate": 0.003,
      "loss": 4.4114,
      "step": 1108
    },
    {
      "epoch": 0.01109,
      "grad_norm": 0.48270985447367,
      "learning_rate": 0.003,
      "loss": 4.4251,
      "step": 1109
    },
    {
      "epoch": 0.0111,
      "grad_norm": 0.521002318978107,
      "learning_rate": 0.003,
      "loss": 4.4465,
      "step": 1110
    },
    {
      "epoch": 0.01111,
      "grad_norm": 0.5366151298263073,
      "learning_rate": 0.003,
      "loss": 4.4215,
      "step": 1111
    },
    {
      "epoch": 0.01112,
      "grad_norm": 0.4687552317220928,
      "learning_rate": 0.003,
      "loss": 4.4366,
      "step": 1112
    },
    {
      "epoch": 0.01113,
      "grad_norm": 0.41403168260550727,
      "learning_rate": 0.003,
      "loss": 4.403,
      "step": 1113
    },
    {
      "epoch": 0.01114,
      "grad_norm": 0.4499817067649614,
      "learning_rate": 0.003,
      "loss": 4.4155,
      "step": 1114
    },
    {
      "epoch": 0.01115,
      "grad_norm": 0.5079887140928414,
      "learning_rate": 0.003,
      "loss": 4.4407,
      "step": 1115
    },
    {
      "epoch": 0.01116,
      "grad_norm": 0.4835939068607683,
      "learning_rate": 0.003,
      "loss": 4.4267,
      "step": 1116
    },
    {
      "epoch": 0.01117,
      "grad_norm": 0.4596276551432057,
      "learning_rate": 0.003,
      "loss": 4.4124,
      "step": 1117
    },
    {
      "epoch": 0.01118,
      "grad_norm": 0.4299051580324855,
      "learning_rate": 0.003,
      "loss": 4.414,
      "step": 1118
    },
    {
      "epoch": 0.01119,
      "grad_norm": 0.43855200803943045,
      "learning_rate": 0.003,
      "loss": 4.4183,
      "step": 1119
    },
    {
      "epoch": 0.0112,
      "grad_norm": 0.47045620681482975,
      "learning_rate": 0.003,
      "loss": 4.4017,
      "step": 1120
    },
    {
      "epoch": 0.01121,
      "grad_norm": 0.47984890763909716,
      "learning_rate": 0.003,
      "loss": 4.4142,
      "step": 1121
    },
    {
      "epoch": 0.01122,
      "grad_norm": 0.4953392048385066,
      "learning_rate": 0.003,
      "loss": 4.3919,
      "step": 1122
    },
    {
      "epoch": 0.01123,
      "grad_norm": 0.5318471574599554,
      "learning_rate": 0.003,
      "loss": 4.3799,
      "step": 1123
    },
    {
      "epoch": 0.01124,
      "grad_norm": 0.5664861125554082,
      "learning_rate": 0.003,
      "loss": 4.4277,
      "step": 1124
    },
    {
      "epoch": 0.01125,
      "grad_norm": 0.5893995505624438,
      "learning_rate": 0.003,
      "loss": 4.3995,
      "step": 1125
    },
    {
      "epoch": 0.01126,
      "grad_norm": 0.5624864569583264,
      "learning_rate": 0.003,
      "loss": 4.4071,
      "step": 1126
    },
    {
      "epoch": 0.01127,
      "grad_norm": 0.6840590538183918,
      "learning_rate": 0.003,
      "loss": 4.442,
      "step": 1127
    },
    {
      "epoch": 0.01128,
      "grad_norm": 0.6290952877293927,
      "learning_rate": 0.003,
      "loss": 4.4221,
      "step": 1128
    },
    {
      "epoch": 0.01129,
      "grad_norm": 0.5439309858740465,
      "learning_rate": 0.003,
      "loss": 4.4089,
      "step": 1129
    },
    {
      "epoch": 0.0113,
      "grad_norm": 0.5640044781984855,
      "learning_rate": 0.003,
      "loss": 4.3915,
      "step": 1130
    },
    {
      "epoch": 0.01131,
      "grad_norm": 0.5291031908216413,
      "learning_rate": 0.003,
      "loss": 4.4024,
      "step": 1131
    },
    {
      "epoch": 0.01132,
      "grad_norm": 0.5330070088798269,
      "learning_rate": 0.003,
      "loss": 4.4286,
      "step": 1132
    },
    {
      "epoch": 0.01133,
      "grad_norm": 0.49343294212766914,
      "learning_rate": 0.003,
      "loss": 4.387,
      "step": 1133
    },
    {
      "epoch": 0.01134,
      "grad_norm": 0.5456485707169718,
      "learning_rate": 0.003,
      "loss": 4.4014,
      "step": 1134
    },
    {
      "epoch": 0.01135,
      "grad_norm": 0.5446260165807105,
      "learning_rate": 0.003,
      "loss": 4.4306,
      "step": 1135
    },
    {
      "epoch": 0.01136,
      "grad_norm": 0.4995675256097383,
      "learning_rate": 0.003,
      "loss": 4.4015,
      "step": 1136
    },
    {
      "epoch": 0.01137,
      "grad_norm": 0.5676213324650841,
      "learning_rate": 0.003,
      "loss": 4.4302,
      "step": 1137
    },
    {
      "epoch": 0.01138,
      "grad_norm": 0.5808574548770593,
      "learning_rate": 0.003,
      "loss": 4.3883,
      "step": 1138
    },
    {
      "epoch": 0.01139,
      "grad_norm": 0.5783839384535551,
      "learning_rate": 0.003,
      "loss": 4.3964,
      "step": 1139
    },
    {
      "epoch": 0.0114,
      "grad_norm": 0.5416297129288978,
      "learning_rate": 0.003,
      "loss": 4.3794,
      "step": 1140
    },
    {
      "epoch": 0.01141,
      "grad_norm": 0.5159183474876468,
      "learning_rate": 0.003,
      "loss": 4.4038,
      "step": 1141
    },
    {
      "epoch": 0.01142,
      "grad_norm": 0.5569718794922074,
      "learning_rate": 0.003,
      "loss": 4.3824,
      "step": 1142
    },
    {
      "epoch": 0.01143,
      "grad_norm": 0.5472188962598419,
      "learning_rate": 0.003,
      "loss": 4.3984,
      "step": 1143
    },
    {
      "epoch": 0.01144,
      "grad_norm": 0.5020345156310454,
      "learning_rate": 0.003,
      "loss": 4.3625,
      "step": 1144
    },
    {
      "epoch": 0.01145,
      "grad_norm": 0.5548649777515073,
      "learning_rate": 0.003,
      "loss": 4.4027,
      "step": 1145
    },
    {
      "epoch": 0.01146,
      "grad_norm": 0.5316552996009986,
      "learning_rate": 0.003,
      "loss": 4.3864,
      "step": 1146
    },
    {
      "epoch": 0.01147,
      "grad_norm": 0.580862846664092,
      "learning_rate": 0.003,
      "loss": 4.3809,
      "step": 1147
    },
    {
      "epoch": 0.01148,
      "grad_norm": 0.6205202997095316,
      "learning_rate": 0.003,
      "loss": 4.3827,
      "step": 1148
    },
    {
      "epoch": 0.01149,
      "grad_norm": 0.6731514287407836,
      "learning_rate": 0.003,
      "loss": 4.4075,
      "step": 1149
    },
    {
      "epoch": 0.0115,
      "grad_norm": 0.8717295567714907,
      "learning_rate": 0.003,
      "loss": 4.4098,
      "step": 1150
    },
    {
      "epoch": 0.01151,
      "grad_norm": 0.8032465834892395,
      "learning_rate": 0.003,
      "loss": 4.4148,
      "step": 1151
    },
    {
      "epoch": 0.01152,
      "grad_norm": 0.7381597286505803,
      "learning_rate": 0.003,
      "loss": 4.4217,
      "step": 1152
    },
    {
      "epoch": 0.01153,
      "grad_norm": 0.7429010521184413,
      "learning_rate": 0.003,
      "loss": 4.4395,
      "step": 1153
    },
    {
      "epoch": 0.01154,
      "grad_norm": 0.6900014831219232,
      "learning_rate": 0.003,
      "loss": 4.4154,
      "step": 1154
    },
    {
      "epoch": 0.01155,
      "grad_norm": 0.64430413026539,
      "learning_rate": 0.003,
      "loss": 4.4019,
      "step": 1155
    },
    {
      "epoch": 0.01156,
      "grad_norm": 0.5602174601436266,
      "learning_rate": 0.003,
      "loss": 4.3999,
      "step": 1156
    },
    {
      "epoch": 0.01157,
      "grad_norm": 0.5870036133781706,
      "learning_rate": 0.003,
      "loss": 4.4301,
      "step": 1157
    },
    {
      "epoch": 0.01158,
      "grad_norm": 0.49506132828209426,
      "learning_rate": 0.003,
      "loss": 4.3945,
      "step": 1158
    },
    {
      "epoch": 0.01159,
      "grad_norm": 0.4315860669439085,
      "learning_rate": 0.003,
      "loss": 4.3852,
      "step": 1159
    },
    {
      "epoch": 0.0116,
      "grad_norm": 0.48532942177035343,
      "learning_rate": 0.003,
      "loss": 4.4026,
      "step": 1160
    },
    {
      "epoch": 0.01161,
      "grad_norm": 0.441658698740049,
      "learning_rate": 0.003,
      "loss": 4.3739,
      "step": 1161
    },
    {
      "epoch": 0.01162,
      "grad_norm": 0.4784702532673203,
      "learning_rate": 0.003,
      "loss": 4.3867,
      "step": 1162
    },
    {
      "epoch": 0.01163,
      "grad_norm": 0.44588993451057907,
      "learning_rate": 0.003,
      "loss": 4.4046,
      "step": 1163
    },
    {
      "epoch": 0.01164,
      "grad_norm": 0.3982553498496815,
      "learning_rate": 0.003,
      "loss": 4.3917,
      "step": 1164
    },
    {
      "epoch": 0.01165,
      "grad_norm": 0.3715052589980553,
      "learning_rate": 0.003,
      "loss": 4.3843,
      "step": 1165
    },
    {
      "epoch": 0.01166,
      "grad_norm": 0.35254776690075756,
      "learning_rate": 0.003,
      "loss": 4.3817,
      "step": 1166
    },
    {
      "epoch": 0.01167,
      "grad_norm": 0.41049866169548815,
      "learning_rate": 0.003,
      "loss": 4.3499,
      "step": 1167
    },
    {
      "epoch": 0.01168,
      "grad_norm": 0.43814969571755685,
      "learning_rate": 0.003,
      "loss": 4.3577,
      "step": 1168
    },
    {
      "epoch": 0.01169,
      "grad_norm": 0.5199727636769494,
      "learning_rate": 0.003,
      "loss": 4.401,
      "step": 1169
    },
    {
      "epoch": 0.0117,
      "grad_norm": 0.7308392317868168,
      "learning_rate": 0.003,
      "loss": 4.3971,
      "step": 1170
    },
    {
      "epoch": 0.01171,
      "grad_norm": 0.8534230998766525,
      "learning_rate": 0.003,
      "loss": 4.3895,
      "step": 1171
    },
    {
      "epoch": 0.01172,
      "grad_norm": 0.7264855860795836,
      "learning_rate": 0.003,
      "loss": 4.3801,
      "step": 1172
    },
    {
      "epoch": 0.01173,
      "grad_norm": 0.7253809298036523,
      "learning_rate": 0.003,
      "loss": 4.4199,
      "step": 1173
    },
    {
      "epoch": 0.01174,
      "grad_norm": 0.7347248301973536,
      "learning_rate": 0.003,
      "loss": 4.3956,
      "step": 1174
    },
    {
      "epoch": 0.01175,
      "grad_norm": 0.837594156223567,
      "learning_rate": 0.003,
      "loss": 4.3885,
      "step": 1175
    },
    {
      "epoch": 0.01176,
      "grad_norm": 0.6959008759390608,
      "learning_rate": 0.003,
      "loss": 4.4162,
      "step": 1176
    },
    {
      "epoch": 0.01177,
      "grad_norm": 0.7136269362957118,
      "learning_rate": 0.003,
      "loss": 4.3914,
      "step": 1177
    },
    {
      "epoch": 0.01178,
      "grad_norm": 0.5923303437356635,
      "learning_rate": 0.003,
      "loss": 4.4002,
      "step": 1178
    },
    {
      "epoch": 0.01179,
      "grad_norm": 0.5305527199591268,
      "learning_rate": 0.003,
      "loss": 4.3783,
      "step": 1179
    },
    {
      "epoch": 0.0118,
      "grad_norm": 0.5526769016584179,
      "learning_rate": 0.003,
      "loss": 4.4415,
      "step": 1180
    },
    {
      "epoch": 0.01181,
      "grad_norm": 0.603168145699105,
      "learning_rate": 0.003,
      "loss": 4.4326,
      "step": 1181
    },
    {
      "epoch": 0.01182,
      "grad_norm": 0.5234478564139211,
      "learning_rate": 0.003,
      "loss": 4.3879,
      "step": 1182
    },
    {
      "epoch": 0.01183,
      "grad_norm": 0.571647059903747,
      "learning_rate": 0.003,
      "loss": 4.3807,
      "step": 1183
    },
    {
      "epoch": 0.01184,
      "grad_norm": 0.6778225586382575,
      "learning_rate": 0.003,
      "loss": 4.4261,
      "step": 1184
    },
    {
      "epoch": 0.01185,
      "grad_norm": 0.7498329664059358,
      "learning_rate": 0.003,
      "loss": 4.4017,
      "step": 1185
    },
    {
      "epoch": 0.01186,
      "grad_norm": 0.7043825932339357,
      "learning_rate": 0.003,
      "loss": 4.4247,
      "step": 1186
    },
    {
      "epoch": 0.01187,
      "grad_norm": 0.6471541142959871,
      "learning_rate": 0.003,
      "loss": 4.3867,
      "step": 1187
    },
    {
      "epoch": 0.01188,
      "grad_norm": 0.498120631856624,
      "learning_rate": 0.003,
      "loss": 4.3947,
      "step": 1188
    },
    {
      "epoch": 0.01189,
      "grad_norm": 0.4776874390876697,
      "learning_rate": 0.003,
      "loss": 4.3881,
      "step": 1189
    },
    {
      "epoch": 0.0119,
      "grad_norm": 0.5052480096056097,
      "learning_rate": 0.003,
      "loss": 4.3727,
      "step": 1190
    },
    {
      "epoch": 0.01191,
      "grad_norm": 0.554587298838607,
      "learning_rate": 0.003,
      "loss": 4.3534,
      "step": 1191
    },
    {
      "epoch": 0.01192,
      "grad_norm": 0.652885110576955,
      "learning_rate": 0.003,
      "loss": 4.3938,
      "step": 1192
    },
    {
      "epoch": 0.01193,
      "grad_norm": 0.814652753692073,
      "learning_rate": 0.003,
      "loss": 4.4033,
      "step": 1193
    },
    {
      "epoch": 0.01194,
      "grad_norm": 0.8149173654121779,
      "learning_rate": 0.003,
      "loss": 4.4185,
      "step": 1194
    },
    {
      "epoch": 0.01195,
      "grad_norm": 0.770137476772292,
      "learning_rate": 0.003,
      "loss": 4.3916,
      "step": 1195
    },
    {
      "epoch": 0.01196,
      "grad_norm": 0.7050379864259693,
      "learning_rate": 0.003,
      "loss": 4.3964,
      "step": 1196
    },
    {
      "epoch": 0.01197,
      "grad_norm": 0.5988825851814717,
      "learning_rate": 0.003,
      "loss": 4.3556,
      "step": 1197
    },
    {
      "epoch": 0.01198,
      "grad_norm": 0.5626266858189237,
      "learning_rate": 0.003,
      "loss": 4.4296,
      "step": 1198
    },
    {
      "epoch": 0.01199,
      "grad_norm": 0.5539141200560117,
      "learning_rate": 0.003,
      "loss": 4.4003,
      "step": 1199
    },
    {
      "epoch": 0.012,
      "grad_norm": 0.446871069304691,
      "learning_rate": 0.003,
      "loss": 4.3872,
      "step": 1200
    },
    {
      "epoch": 0.01201,
      "grad_norm": 0.41098308583368987,
      "learning_rate": 0.003,
      "loss": 4.4145,
      "step": 1201
    },
    {
      "epoch": 0.01202,
      "grad_norm": 0.40482114944892034,
      "learning_rate": 0.003,
      "loss": 4.3801,
      "step": 1202
    },
    {
      "epoch": 0.01203,
      "grad_norm": 0.3801203709283518,
      "learning_rate": 0.003,
      "loss": 4.3878,
      "step": 1203
    },
    {
      "epoch": 0.01204,
      "grad_norm": 0.4044230744585578,
      "learning_rate": 0.003,
      "loss": 4.3855,
      "step": 1204
    },
    {
      "epoch": 0.01205,
      "grad_norm": 0.4153028028477695,
      "learning_rate": 0.003,
      "loss": 4.3753,
      "step": 1205
    },
    {
      "epoch": 0.01206,
      "grad_norm": 0.4253601027948775,
      "learning_rate": 0.003,
      "loss": 4.4192,
      "step": 1206
    },
    {
      "epoch": 0.01207,
      "grad_norm": 0.42584001554399725,
      "learning_rate": 0.003,
      "loss": 4.3415,
      "step": 1207
    },
    {
      "epoch": 0.01208,
      "grad_norm": 0.38034721696084095,
      "learning_rate": 0.003,
      "loss": 4.3983,
      "step": 1208
    },
    {
      "epoch": 0.01209,
      "grad_norm": 0.38812168043845363,
      "learning_rate": 0.003,
      "loss": 4.3807,
      "step": 1209
    },
    {
      "epoch": 0.0121,
      "grad_norm": 0.48056015790597395,
      "learning_rate": 0.003,
      "loss": 4.3838,
      "step": 1210
    },
    {
      "epoch": 0.01211,
      "grad_norm": 0.5637402598516663,
      "learning_rate": 0.003,
      "loss": 4.3728,
      "step": 1211
    },
    {
      "epoch": 0.01212,
      "grad_norm": 0.678908053443676,
      "learning_rate": 0.003,
      "loss": 4.4009,
      "step": 1212
    },
    {
      "epoch": 0.01213,
      "grad_norm": 0.7872996276111477,
      "learning_rate": 0.003,
      "loss": 4.3904,
      "step": 1213
    },
    {
      "epoch": 0.01214,
      "grad_norm": 0.737245569468942,
      "learning_rate": 0.003,
      "loss": 4.3712,
      "step": 1214
    },
    {
      "epoch": 0.01215,
      "grad_norm": 0.7444446627290159,
      "learning_rate": 0.003,
      "loss": 4.3825,
      "step": 1215
    },
    {
      "epoch": 0.01216,
      "grad_norm": 0.7493613956712142,
      "learning_rate": 0.003,
      "loss": 4.3829,
      "step": 1216
    },
    {
      "epoch": 0.01217,
      "grad_norm": 0.7355810094680961,
      "learning_rate": 0.003,
      "loss": 4.3915,
      "step": 1217
    },
    {
      "epoch": 0.01218,
      "grad_norm": 0.8012910697526976,
      "learning_rate": 0.003,
      "loss": 4.3871,
      "step": 1218
    },
    {
      "epoch": 0.01219,
      "grad_norm": 0.8693012687676964,
      "learning_rate": 0.003,
      "loss": 4.3988,
      "step": 1219
    },
    {
      "epoch": 0.0122,
      "grad_norm": 0.8958839342651639,
      "learning_rate": 0.003,
      "loss": 4.3996,
      "step": 1220
    },
    {
      "epoch": 0.01221,
      "grad_norm": 1.0403569844700198,
      "learning_rate": 0.003,
      "loss": 4.4291,
      "step": 1221
    },
    {
      "epoch": 0.01222,
      "grad_norm": 1.1355134224594017,
      "learning_rate": 0.003,
      "loss": 4.4293,
      "step": 1222
    },
    {
      "epoch": 0.01223,
      "grad_norm": 0.9630650396281519,
      "learning_rate": 0.003,
      "loss": 4.4167,
      "step": 1223
    },
    {
      "epoch": 0.01224,
      "grad_norm": 0.8763852343022005,
      "learning_rate": 0.003,
      "loss": 4.4062,
      "step": 1224
    },
    {
      "epoch": 0.01225,
      "grad_norm": 0.8983930620848783,
      "learning_rate": 0.003,
      "loss": 4.4328,
      "step": 1225
    },
    {
      "epoch": 0.01226,
      "grad_norm": 0.7843004711214445,
      "learning_rate": 0.003,
      "loss": 4.4231,
      "step": 1226
    },
    {
      "epoch": 0.01227,
      "grad_norm": 0.6802825597762372,
      "learning_rate": 0.003,
      "loss": 4.4455,
      "step": 1227
    },
    {
      "epoch": 0.01228,
      "grad_norm": 0.7175774602889506,
      "learning_rate": 0.003,
      "loss": 4.433,
      "step": 1228
    },
    {
      "epoch": 0.01229,
      "grad_norm": 0.7124552785986754,
      "learning_rate": 0.003,
      "loss": 4.4177,
      "step": 1229
    },
    {
      "epoch": 0.0123,
      "grad_norm": 0.7173646190359848,
      "learning_rate": 0.003,
      "loss": 4.4462,
      "step": 1230
    },
    {
      "epoch": 0.01231,
      "grad_norm": 0.6710417652647681,
      "learning_rate": 0.003,
      "loss": 4.4192,
      "step": 1231
    },
    {
      "epoch": 0.01232,
      "grad_norm": 0.5319364475378244,
      "learning_rate": 0.003,
      "loss": 4.3837,
      "step": 1232
    },
    {
      "epoch": 0.01233,
      "grad_norm": 0.5107726510651173,
      "learning_rate": 0.003,
      "loss": 4.3892,
      "step": 1233
    },
    {
      "epoch": 0.01234,
      "grad_norm": 0.525523603676889,
      "learning_rate": 0.003,
      "loss": 4.4032,
      "step": 1234
    },
    {
      "epoch": 0.01235,
      "grad_norm": 0.4385933034868044,
      "learning_rate": 0.003,
      "loss": 4.3895,
      "step": 1235
    },
    {
      "epoch": 0.01236,
      "grad_norm": 0.4417827360571799,
      "learning_rate": 0.003,
      "loss": 4.3922,
      "step": 1236
    },
    {
      "epoch": 0.01237,
      "grad_norm": 0.3870170981569727,
      "learning_rate": 0.003,
      "loss": 4.3983,
      "step": 1237
    },
    {
      "epoch": 0.01238,
      "grad_norm": 0.3450028372069669,
      "learning_rate": 0.003,
      "loss": 4.3918,
      "step": 1238
    },
    {
      "epoch": 0.01239,
      "grad_norm": 0.32255856446948694,
      "learning_rate": 0.003,
      "loss": 4.3727,
      "step": 1239
    },
    {
      "epoch": 0.0124,
      "grad_norm": 0.3078475989244171,
      "learning_rate": 0.003,
      "loss": 4.3537,
      "step": 1240
    },
    {
      "epoch": 0.01241,
      "grad_norm": 0.34649646171341814,
      "learning_rate": 0.003,
      "loss": 4.3646,
      "step": 1241
    },
    {
      "epoch": 0.01242,
      "grad_norm": 0.3965097983444254,
      "learning_rate": 0.003,
      "loss": 4.3706,
      "step": 1242
    },
    {
      "epoch": 0.01243,
      "grad_norm": 0.4535687938279078,
      "learning_rate": 0.003,
      "loss": 4.3517,
      "step": 1243
    },
    {
      "epoch": 0.01244,
      "grad_norm": 0.5186172619894055,
      "learning_rate": 0.003,
      "loss": 4.3669,
      "step": 1244
    },
    {
      "epoch": 0.01245,
      "grad_norm": 0.6425018812487958,
      "learning_rate": 0.003,
      "loss": 4.363,
      "step": 1245
    },
    {
      "epoch": 0.01246,
      "grad_norm": 0.6245723386860975,
      "learning_rate": 0.003,
      "loss": 4.3446,
      "step": 1246
    },
    {
      "epoch": 0.01247,
      "grad_norm": 0.44377555171018157,
      "learning_rate": 0.003,
      "loss": 4.3915,
      "step": 1247
    },
    {
      "epoch": 0.01248,
      "grad_norm": 0.5278029739976049,
      "learning_rate": 0.003,
      "loss": 4.3761,
      "step": 1248
    },
    {
      "epoch": 0.01249,
      "grad_norm": 0.5102876241848929,
      "learning_rate": 0.003,
      "loss": 4.3677,
      "step": 1249
    },
    {
      "epoch": 0.0125,
      "grad_norm": 0.39559739569893493,
      "learning_rate": 0.003,
      "loss": 4.3348,
      "step": 1250
    },
    {
      "epoch": 0.01251,
      "grad_norm": 0.4774815455759189,
      "learning_rate": 0.003,
      "loss": 4.373,
      "step": 1251
    },
    {
      "epoch": 0.01252,
      "grad_norm": 0.40459109080676114,
      "learning_rate": 0.003,
      "loss": 4.3467,
      "step": 1252
    },
    {
      "epoch": 0.01253,
      "grad_norm": 0.3506155695588933,
      "learning_rate": 0.003,
      "loss": 4.3238,
      "step": 1253
    },
    {
      "epoch": 0.01254,
      "grad_norm": 0.36046868647854385,
      "learning_rate": 0.003,
      "loss": 4.3697,
      "step": 1254
    },
    {
      "epoch": 0.01255,
      "grad_norm": 0.400469314411457,
      "learning_rate": 0.003,
      "loss": 4.3202,
      "step": 1255
    },
    {
      "epoch": 0.01256,
      "grad_norm": 0.42579169102032727,
      "learning_rate": 0.003,
      "loss": 4.3478,
      "step": 1256
    },
    {
      "epoch": 0.01257,
      "grad_norm": 0.43666374954115567,
      "learning_rate": 0.003,
      "loss": 4.3342,
      "step": 1257
    },
    {
      "epoch": 0.01258,
      "grad_norm": 0.3885559357601015,
      "learning_rate": 0.003,
      "loss": 4.3411,
      "step": 1258
    },
    {
      "epoch": 0.01259,
      "grad_norm": 0.3450685274689168,
      "learning_rate": 0.003,
      "loss": 4.3417,
      "step": 1259
    },
    {
      "epoch": 0.0126,
      "grad_norm": 0.3296913229590147,
      "learning_rate": 0.003,
      "loss": 4.3689,
      "step": 1260
    },
    {
      "epoch": 0.01261,
      "grad_norm": 0.3469234561542926,
      "learning_rate": 0.003,
      "loss": 4.3628,
      "step": 1261
    },
    {
      "epoch": 0.01262,
      "grad_norm": 0.40339400226308325,
      "learning_rate": 0.003,
      "loss": 4.3558,
      "step": 1262
    },
    {
      "epoch": 0.01263,
      "grad_norm": 0.44597422220213867,
      "learning_rate": 0.003,
      "loss": 4.3512,
      "step": 1263
    },
    {
      "epoch": 0.01264,
      "grad_norm": 0.4427407408205576,
      "learning_rate": 0.003,
      "loss": 4.338,
      "step": 1264
    },
    {
      "epoch": 0.01265,
      "grad_norm": 0.41573445698488437,
      "learning_rate": 0.003,
      "loss": 4.3506,
      "step": 1265
    },
    {
      "epoch": 0.01266,
      "grad_norm": 0.45186560215837035,
      "learning_rate": 0.003,
      "loss": 4.349,
      "step": 1266
    },
    {
      "epoch": 0.01267,
      "grad_norm": 0.6290443655616889,
      "learning_rate": 0.003,
      "loss": 4.3767,
      "step": 1267
    },
    {
      "epoch": 0.01268,
      "grad_norm": 0.8841541603744382,
      "learning_rate": 0.003,
      "loss": 4.3763,
      "step": 1268
    },
    {
      "epoch": 0.01269,
      "grad_norm": 0.9324063273299563,
      "learning_rate": 0.003,
      "loss": 4.3633,
      "step": 1269
    },
    {
      "epoch": 0.0127,
      "grad_norm": 0.7531387064097018,
      "learning_rate": 0.003,
      "loss": 4.389,
      "step": 1270
    },
    {
      "epoch": 0.01271,
      "grad_norm": 0.7173511219844001,
      "learning_rate": 0.003,
      "loss": 4.3689,
      "step": 1271
    },
    {
      "epoch": 0.01272,
      "grad_norm": 0.6910307872241689,
      "learning_rate": 0.003,
      "loss": 4.3626,
      "step": 1272
    },
    {
      "epoch": 0.01273,
      "grad_norm": 0.6392091020786222,
      "learning_rate": 0.003,
      "loss": 4.3902,
      "step": 1273
    },
    {
      "epoch": 0.01274,
      "grad_norm": 0.5683923983758625,
      "learning_rate": 0.003,
      "loss": 4.3811,
      "step": 1274
    },
    {
      "epoch": 0.01275,
      "grad_norm": 0.48501080462586865,
      "learning_rate": 0.003,
      "loss": 4.3433,
      "step": 1275
    },
    {
      "epoch": 0.01276,
      "grad_norm": 0.4388665728327947,
      "learning_rate": 0.003,
      "loss": 4.3487,
      "step": 1276
    },
    {
      "epoch": 0.01277,
      "grad_norm": 0.5024100232553123,
      "learning_rate": 0.003,
      "loss": 4.358,
      "step": 1277
    },
    {
      "epoch": 0.01278,
      "grad_norm": 0.5072917507898231,
      "learning_rate": 0.003,
      "loss": 4.3646,
      "step": 1278
    },
    {
      "epoch": 0.01279,
      "grad_norm": 0.48942775338603506,
      "learning_rate": 0.003,
      "loss": 4.3461,
      "step": 1279
    },
    {
      "epoch": 0.0128,
      "grad_norm": 0.5180680805997396,
      "learning_rate": 0.003,
      "loss": 4.3453,
      "step": 1280
    },
    {
      "epoch": 0.01281,
      "grad_norm": 0.589882368752518,
      "learning_rate": 0.003,
      "loss": 4.3309,
      "step": 1281
    },
    {
      "epoch": 0.01282,
      "grad_norm": 0.7217688053955748,
      "learning_rate": 0.003,
      "loss": 4.3833,
      "step": 1282
    },
    {
      "epoch": 0.01283,
      "grad_norm": 0.849897015093446,
      "learning_rate": 0.003,
      "loss": 4.3595,
      "step": 1283
    },
    {
      "epoch": 0.01284,
      "grad_norm": 0.769319717001856,
      "learning_rate": 0.003,
      "loss": 4.3713,
      "step": 1284
    },
    {
      "epoch": 0.01285,
      "grad_norm": 0.7372234573055417,
      "learning_rate": 0.003,
      "loss": 4.3679,
      "step": 1285
    },
    {
      "epoch": 0.01286,
      "grad_norm": 0.6185984538132406,
      "learning_rate": 0.003,
      "loss": 4.3884,
      "step": 1286
    },
    {
      "epoch": 0.01287,
      "grad_norm": 0.566489365455937,
      "learning_rate": 0.003,
      "loss": 4.3487,
      "step": 1287
    },
    {
      "epoch": 0.01288,
      "grad_norm": 0.583268123970244,
      "learning_rate": 0.003,
      "loss": 4.3654,
      "step": 1288
    },
    {
      "epoch": 0.01289,
      "grad_norm": 0.5365604798494432,
      "learning_rate": 0.003,
      "loss": 4.3546,
      "step": 1289
    },
    {
      "epoch": 0.0129,
      "grad_norm": 0.5898461372899422,
      "learning_rate": 0.003,
      "loss": 4.3784,
      "step": 1290
    },
    {
      "epoch": 0.01291,
      "grad_norm": 0.7454645554405436,
      "learning_rate": 0.003,
      "loss": 4.3672,
      "step": 1291
    },
    {
      "epoch": 0.01292,
      "grad_norm": 0.8872276988300176,
      "learning_rate": 0.003,
      "loss": 4.4157,
      "step": 1292
    },
    {
      "epoch": 0.01293,
      "grad_norm": 0.8660558149247537,
      "learning_rate": 0.003,
      "loss": 4.4053,
      "step": 1293
    },
    {
      "epoch": 0.01294,
      "grad_norm": 0.7869826032444561,
      "learning_rate": 0.003,
      "loss": 4.3679,
      "step": 1294
    },
    {
      "epoch": 0.01295,
      "grad_norm": 0.7577761481466739,
      "learning_rate": 0.003,
      "loss": 4.3975,
      "step": 1295
    },
    {
      "epoch": 0.01296,
      "grad_norm": 0.6748799445050047,
      "learning_rate": 0.003,
      "loss": 4.3598,
      "step": 1296
    },
    {
      "epoch": 0.01297,
      "grad_norm": 0.670162676491956,
      "learning_rate": 0.003,
      "loss": 4.376,
      "step": 1297
    },
    {
      "epoch": 0.01298,
      "grad_norm": 0.5691583859101702,
      "learning_rate": 0.003,
      "loss": 4.3472,
      "step": 1298
    },
    {
      "epoch": 0.01299,
      "grad_norm": 0.5072926331461023,
      "learning_rate": 0.003,
      "loss": 4.3555,
      "step": 1299
    },
    {
      "epoch": 0.013,
      "grad_norm": 0.4828769685141814,
      "learning_rate": 0.003,
      "loss": 4.343,
      "step": 1300
    },
    {
      "epoch": 0.01301,
      "grad_norm": 0.4691822990462979,
      "learning_rate": 0.003,
      "loss": 4.3558,
      "step": 1301
    },
    {
      "epoch": 0.01302,
      "grad_norm": 0.4140113973052622,
      "learning_rate": 0.003,
      "loss": 4.4006,
      "step": 1302
    },
    {
      "epoch": 0.01303,
      "grad_norm": 0.3685109579455993,
      "learning_rate": 0.003,
      "loss": 4.3436,
      "step": 1303
    },
    {
      "epoch": 0.01304,
      "grad_norm": 0.3760405201721622,
      "learning_rate": 0.003,
      "loss": 4.3696,
      "step": 1304
    },
    {
      "epoch": 0.01305,
      "grad_norm": 0.3221665114274485,
      "learning_rate": 0.003,
      "loss": 4.3518,
      "step": 1305
    },
    {
      "epoch": 0.01306,
      "grad_norm": 0.33492469868124675,
      "learning_rate": 0.003,
      "loss": 4.3452,
      "step": 1306
    },
    {
      "epoch": 0.01307,
      "grad_norm": 0.33475611430641294,
      "learning_rate": 0.003,
      "loss": 4.3626,
      "step": 1307
    },
    {
      "epoch": 0.01308,
      "grad_norm": 0.4027154015406206,
      "learning_rate": 0.003,
      "loss": 4.3385,
      "step": 1308
    },
    {
      "epoch": 0.01309,
      "grad_norm": 0.5286332892700527,
      "learning_rate": 0.003,
      "loss": 4.3804,
      "step": 1309
    },
    {
      "epoch": 0.0131,
      "grad_norm": 0.7090250449949596,
      "learning_rate": 0.003,
      "loss": 4.3359,
      "step": 1310
    },
    {
      "epoch": 0.01311,
      "grad_norm": 0.9374592813554533,
      "learning_rate": 0.003,
      "loss": 4.3914,
      "step": 1311
    },
    {
      "epoch": 0.01312,
      "grad_norm": 0.8386895366440477,
      "learning_rate": 0.003,
      "loss": 4.3696,
      "step": 1312
    },
    {
      "epoch": 0.01313,
      "grad_norm": 0.7384206623774927,
      "learning_rate": 0.003,
      "loss": 4.3534,
      "step": 1313
    },
    {
      "epoch": 0.01314,
      "grad_norm": 0.7139194720668496,
      "learning_rate": 0.003,
      "loss": 4.3527,
      "step": 1314
    },
    {
      "epoch": 0.01315,
      "grad_norm": 0.7085227699461494,
      "learning_rate": 0.003,
      "loss": 4.3809,
      "step": 1315
    },
    {
      "epoch": 0.01316,
      "grad_norm": 0.6924444519815,
      "learning_rate": 0.003,
      "loss": 4.3653,
      "step": 1316
    },
    {
      "epoch": 0.01317,
      "grad_norm": 0.6387153931562799,
      "learning_rate": 0.003,
      "loss": 4.3902,
      "step": 1317
    },
    {
      "epoch": 0.01318,
      "grad_norm": 0.6344112280159295,
      "learning_rate": 0.003,
      "loss": 4.3652,
      "step": 1318
    },
    {
      "epoch": 0.01319,
      "grad_norm": 0.7270675767653099,
      "learning_rate": 0.003,
      "loss": 4.3841,
      "step": 1319
    },
    {
      "epoch": 0.0132,
      "grad_norm": 0.6321319111156113,
      "learning_rate": 0.003,
      "loss": 4.3725,
      "step": 1320
    },
    {
      "epoch": 0.01321,
      "grad_norm": 0.6141991678487796,
      "learning_rate": 0.003,
      "loss": 4.3508,
      "step": 1321
    },
    {
      "epoch": 0.01322,
      "grad_norm": 0.5694545226818948,
      "learning_rate": 0.003,
      "loss": 4.3756,
      "step": 1322
    },
    {
      "epoch": 0.01323,
      "grad_norm": 0.7445509945833578,
      "learning_rate": 0.003,
      "loss": 4.3766,
      "step": 1323
    },
    {
      "epoch": 0.01324,
      "grad_norm": 0.7345275618591681,
      "learning_rate": 0.003,
      "loss": 4.3604,
      "step": 1324
    },
    {
      "epoch": 0.01325,
      "grad_norm": 0.5731952380277242,
      "learning_rate": 0.003,
      "loss": 4.3595,
      "step": 1325
    },
    {
      "epoch": 0.01326,
      "grad_norm": 0.49399828185062894,
      "learning_rate": 0.003,
      "loss": 4.3432,
      "step": 1326
    },
    {
      "epoch": 0.01327,
      "grad_norm": 0.4625597253069682,
      "learning_rate": 0.003,
      "loss": 4.3586,
      "step": 1327
    },
    {
      "epoch": 0.01328,
      "grad_norm": 0.46176548306971715,
      "learning_rate": 0.003,
      "loss": 4.3477,
      "step": 1328
    },
    {
      "epoch": 0.01329,
      "grad_norm": 0.4120837821389968,
      "learning_rate": 0.003,
      "loss": 4.3099,
      "step": 1329
    },
    {
      "epoch": 0.0133,
      "grad_norm": 0.3625953571207396,
      "learning_rate": 0.003,
      "loss": 4.3494,
      "step": 1330
    },
    {
      "epoch": 0.01331,
      "grad_norm": 0.36184528613842726,
      "learning_rate": 0.003,
      "loss": 4.3351,
      "step": 1331
    },
    {
      "epoch": 0.01332,
      "grad_norm": 0.3568024153641722,
      "learning_rate": 0.003,
      "loss": 4.3532,
      "step": 1332
    },
    {
      "epoch": 0.01333,
      "grad_norm": 0.3204482353647099,
      "learning_rate": 0.003,
      "loss": 4.3252,
      "step": 1333
    },
    {
      "epoch": 0.01334,
      "grad_norm": 0.29957411498207237,
      "learning_rate": 0.003,
      "loss": 4.3114,
      "step": 1334
    },
    {
      "epoch": 0.01335,
      "grad_norm": 0.31171408266049466,
      "learning_rate": 0.003,
      "loss": 4.3334,
      "step": 1335
    },
    {
      "epoch": 0.01336,
      "grad_norm": 0.28708995347069843,
      "learning_rate": 0.003,
      "loss": 4.3294,
      "step": 1336
    },
    {
      "epoch": 0.01337,
      "grad_norm": 0.29278842241858566,
      "learning_rate": 0.003,
      "loss": 4.3434,
      "step": 1337
    },
    {
      "epoch": 0.01338,
      "grad_norm": 0.2729723674299326,
      "learning_rate": 0.003,
      "loss": 4.3088,
      "step": 1338
    },
    {
      "epoch": 0.01339,
      "grad_norm": 0.29431343213978217,
      "learning_rate": 0.003,
      "loss": 4.3327,
      "step": 1339
    },
    {
      "epoch": 0.0134,
      "grad_norm": 0.3294175545018703,
      "learning_rate": 0.003,
      "loss": 4.3159,
      "step": 1340
    },
    {
      "epoch": 0.01341,
      "grad_norm": 0.4346116539808444,
      "learning_rate": 0.003,
      "loss": 4.3329,
      "step": 1341
    },
    {
      "epoch": 0.01342,
      "grad_norm": 0.6006528789210607,
      "learning_rate": 0.003,
      "loss": 4.3107,
      "step": 1342
    },
    {
      "epoch": 0.01343,
      "grad_norm": 0.9468238356913852,
      "learning_rate": 0.003,
      "loss": 4.3323,
      "step": 1343
    },
    {
      "epoch": 0.01344,
      "grad_norm": 0.9600917389412905,
      "learning_rate": 0.003,
      "loss": 4.3573,
      "step": 1344
    },
    {
      "epoch": 0.01345,
      "grad_norm": 0.7454028532315747,
      "learning_rate": 0.003,
      "loss": 4.3824,
      "step": 1345
    },
    {
      "epoch": 0.01346,
      "grad_norm": 0.9412385374275949,
      "learning_rate": 0.003,
      "loss": 4.3817,
      "step": 1346
    },
    {
      "epoch": 0.01347,
      "grad_norm": 0.8019477836355356,
      "learning_rate": 0.003,
      "loss": 4.3621,
      "step": 1347
    },
    {
      "epoch": 0.01348,
      "grad_norm": 0.822147627695476,
      "learning_rate": 0.003,
      "loss": 4.3672,
      "step": 1348
    },
    {
      "epoch": 0.01349,
      "grad_norm": 0.7468015076519685,
      "learning_rate": 0.003,
      "loss": 4.3844,
      "step": 1349
    },
    {
      "epoch": 0.0135,
      "grad_norm": 0.7150496437877271,
      "learning_rate": 0.003,
      "loss": 4.3769,
      "step": 1350
    },
    {
      "epoch": 0.01351,
      "grad_norm": 0.6792275106065346,
      "learning_rate": 0.003,
      "loss": 4.4037,
      "step": 1351
    },
    {
      "epoch": 0.01352,
      "grad_norm": 0.6749873536633729,
      "learning_rate": 0.003,
      "loss": 4.3795,
      "step": 1352
    },
    {
      "epoch": 0.01353,
      "grad_norm": 0.5975751363464156,
      "learning_rate": 0.003,
      "loss": 4.3625,
      "step": 1353
    },
    {
      "epoch": 0.01354,
      "grad_norm": 0.6367703716959885,
      "learning_rate": 0.003,
      "loss": 4.3473,
      "step": 1354
    },
    {
      "epoch": 0.01355,
      "grad_norm": 0.6903253322513454,
      "learning_rate": 0.003,
      "loss": 4.3617,
      "step": 1355
    },
    {
      "epoch": 0.01356,
      "grad_norm": 0.7830002665034361,
      "learning_rate": 0.003,
      "loss": 4.3589,
      "step": 1356
    },
    {
      "epoch": 0.01357,
      "grad_norm": 0.7634637746675061,
      "learning_rate": 0.003,
      "loss": 4.351,
      "step": 1357
    },
    {
      "epoch": 0.01358,
      "grad_norm": 0.6912113286037168,
      "learning_rate": 0.003,
      "loss": 4.3426,
      "step": 1358
    },
    {
      "epoch": 0.01359,
      "grad_norm": 0.608100998568284,
      "learning_rate": 0.003,
      "loss": 4.3802,
      "step": 1359
    },
    {
      "epoch": 0.0136,
      "grad_norm": 0.4174874313338161,
      "learning_rate": 0.003,
      "loss": 4.3375,
      "step": 1360
    },
    {
      "epoch": 0.01361,
      "grad_norm": 0.4136844583981887,
      "learning_rate": 0.003,
      "loss": 4.3533,
      "step": 1361
    },
    {
      "epoch": 0.01362,
      "grad_norm": 0.34986686555598134,
      "learning_rate": 0.003,
      "loss": 4.3597,
      "step": 1362
    },
    {
      "epoch": 0.01363,
      "grad_norm": 0.39445259692219764,
      "learning_rate": 0.003,
      "loss": 4.3605,
      "step": 1363
    },
    {
      "epoch": 0.01364,
      "grad_norm": 0.3392587016329843,
      "learning_rate": 0.003,
      "loss": 4.3508,
      "step": 1364
    },
    {
      "epoch": 0.01365,
      "grad_norm": 0.3286929763598885,
      "learning_rate": 0.003,
      "loss": 4.3666,
      "step": 1365
    },
    {
      "epoch": 0.01366,
      "grad_norm": 0.30698836061044843,
      "learning_rate": 0.003,
      "loss": 4.3194,
      "step": 1366
    },
    {
      "epoch": 0.01367,
      "grad_norm": 0.3258356748399825,
      "learning_rate": 0.003,
      "loss": 4.3664,
      "step": 1367
    },
    {
      "epoch": 0.01368,
      "grad_norm": 0.28974975652956814,
      "learning_rate": 0.003,
      "loss": 4.3364,
      "step": 1368
    },
    {
      "epoch": 0.01369,
      "grad_norm": 0.3029711054161145,
      "learning_rate": 0.003,
      "loss": 4.3277,
      "step": 1369
    },
    {
      "epoch": 0.0137,
      "grad_norm": 0.2864574506759695,
      "learning_rate": 0.003,
      "loss": 4.3508,
      "step": 1370
    },
    {
      "epoch": 0.01371,
      "grad_norm": 0.2944593491964238,
      "learning_rate": 0.003,
      "loss": 4.3421,
      "step": 1371
    },
    {
      "epoch": 0.01372,
      "grad_norm": 0.29733135265028415,
      "learning_rate": 0.003,
      "loss": 4.29,
      "step": 1372
    },
    {
      "epoch": 0.01373,
      "grad_norm": 0.3030922885731935,
      "learning_rate": 0.003,
      "loss": 4.3287,
      "step": 1373
    },
    {
      "epoch": 0.01374,
      "grad_norm": 0.3135661353258684,
      "learning_rate": 0.003,
      "loss": 4.3353,
      "step": 1374
    },
    {
      "epoch": 0.01375,
      "grad_norm": 0.31552830475247895,
      "learning_rate": 0.003,
      "loss": 4.2954,
      "step": 1375
    },
    {
      "epoch": 0.01376,
      "grad_norm": 0.3753685901400331,
      "learning_rate": 0.003,
      "loss": 4.3218,
      "step": 1376
    },
    {
      "epoch": 0.01377,
      "grad_norm": 0.4931158688793232,
      "learning_rate": 0.003,
      "loss": 4.3247,
      "step": 1377
    },
    {
      "epoch": 0.01378,
      "grad_norm": 0.6194459766224392,
      "learning_rate": 0.003,
      "loss": 4.3586,
      "step": 1378
    },
    {
      "epoch": 0.01379,
      "grad_norm": 0.7157871954469385,
      "learning_rate": 0.003,
      "loss": 4.3344,
      "step": 1379
    },
    {
      "epoch": 0.0138,
      "grad_norm": 0.7778989267777976,
      "learning_rate": 0.003,
      "loss": 4.3379,
      "step": 1380
    },
    {
      "epoch": 0.01381,
      "grad_norm": 0.7540005493939272,
      "learning_rate": 0.003,
      "loss": 4.3478,
      "step": 1381
    },
    {
      "epoch": 0.01382,
      "grad_norm": 0.6552744600197745,
      "learning_rate": 0.003,
      "loss": 4.3434,
      "step": 1382
    },
    {
      "epoch": 0.01383,
      "grad_norm": 0.5244295500211882,
      "learning_rate": 0.003,
      "loss": 4.332,
      "step": 1383
    },
    {
      "epoch": 0.01384,
      "grad_norm": 0.5199048618667436,
      "learning_rate": 0.003,
      "loss": 4.3604,
      "step": 1384
    },
    {
      "epoch": 0.01385,
      "grad_norm": 0.5115879970370135,
      "learning_rate": 0.003,
      "loss": 4.3484,
      "step": 1385
    },
    {
      "epoch": 0.01386,
      "grad_norm": 0.5337761524536188,
      "learning_rate": 0.003,
      "loss": 4.3229,
      "step": 1386
    },
    {
      "epoch": 0.01387,
      "grad_norm": 0.49971027233062854,
      "learning_rate": 0.003,
      "loss": 4.3141,
      "step": 1387
    },
    {
      "epoch": 0.01388,
      "grad_norm": 0.45570239975200477,
      "learning_rate": 0.003,
      "loss": 4.3501,
      "step": 1388
    },
    {
      "epoch": 0.01389,
      "grad_norm": 0.4952533360995893,
      "learning_rate": 0.003,
      "loss": 4.329,
      "step": 1389
    },
    {
      "epoch": 0.0139,
      "grad_norm": 0.5836752219457788,
      "learning_rate": 0.003,
      "loss": 4.3354,
      "step": 1390
    },
    {
      "epoch": 0.01391,
      "grad_norm": 0.661197601877714,
      "learning_rate": 0.003,
      "loss": 4.3388,
      "step": 1391
    },
    {
      "epoch": 0.01392,
      "grad_norm": 0.7580066821140803,
      "learning_rate": 0.003,
      "loss": 4.3629,
      "step": 1392
    },
    {
      "epoch": 0.01393,
      "grad_norm": 0.864097109773997,
      "learning_rate": 0.003,
      "loss": 4.3351,
      "step": 1393
    },
    {
      "epoch": 0.01394,
      "grad_norm": 0.8137871207398635,
      "learning_rate": 0.003,
      "loss": 4.3329,
      "step": 1394
    },
    {
      "epoch": 0.01395,
      "grad_norm": 0.6452659589342782,
      "learning_rate": 0.003,
      "loss": 4.3528,
      "step": 1395
    },
    {
      "epoch": 0.01396,
      "grad_norm": 0.7169962932582273,
      "learning_rate": 0.003,
      "loss": 4.3526,
      "step": 1396
    },
    {
      "epoch": 0.01397,
      "grad_norm": 0.7722092445904787,
      "learning_rate": 0.003,
      "loss": 4.37,
      "step": 1397
    },
    {
      "epoch": 0.01398,
      "grad_norm": 0.9201404272023904,
      "learning_rate": 0.003,
      "loss": 4.3584,
      "step": 1398
    },
    {
      "epoch": 0.01399,
      "grad_norm": 0.8604376144856999,
      "learning_rate": 0.003,
      "loss": 4.3863,
      "step": 1399
    },
    {
      "epoch": 0.014,
      "grad_norm": 0.7356178947310503,
      "learning_rate": 0.003,
      "loss": 4.3371,
      "step": 1400
    },
    {
      "epoch": 0.01401,
      "grad_norm": 0.6698615745523159,
      "learning_rate": 0.003,
      "loss": 4.3548,
      "step": 1401
    },
    {
      "epoch": 0.01402,
      "grad_norm": 0.678651105292387,
      "learning_rate": 0.003,
      "loss": 4.367,
      "step": 1402
    },
    {
      "epoch": 0.01403,
      "grad_norm": 0.7164501675372886,
      "learning_rate": 0.003,
      "loss": 4.3428,
      "step": 1403
    },
    {
      "epoch": 0.01404,
      "grad_norm": 0.7780771367935839,
      "learning_rate": 0.003,
      "loss": 4.3706,
      "step": 1404
    },
    {
      "epoch": 0.01405,
      "grad_norm": 0.8097299555256038,
      "learning_rate": 0.003,
      "loss": 4.3921,
      "step": 1405
    },
    {
      "epoch": 0.01406,
      "grad_norm": 0.8397836379510005,
      "learning_rate": 0.003,
      "loss": 4.3748,
      "step": 1406
    },
    {
      "epoch": 0.01407,
      "grad_norm": 0.798434347116743,
      "learning_rate": 0.003,
      "loss": 4.3886,
      "step": 1407
    },
    {
      "epoch": 0.01408,
      "grad_norm": 0.690162960674035,
      "learning_rate": 0.003,
      "loss": 4.3438,
      "step": 1408
    },
    {
      "epoch": 0.01409,
      "grad_norm": 0.7158182494367376,
      "learning_rate": 0.003,
      "loss": 4.3755,
      "step": 1409
    },
    {
      "epoch": 0.0141,
      "grad_norm": 0.6626220406570043,
      "learning_rate": 0.003,
      "loss": 4.3521,
      "step": 1410
    },
    {
      "epoch": 0.01411,
      "grad_norm": 0.6080932877854873,
      "learning_rate": 0.003,
      "loss": 4.3227,
      "step": 1411
    },
    {
      "epoch": 0.01412,
      "grad_norm": 0.6342264759843587,
      "learning_rate": 0.003,
      "loss": 4.3642,
      "step": 1412
    },
    {
      "epoch": 0.01413,
      "grad_norm": 0.5857768731552475,
      "learning_rate": 0.003,
      "loss": 4.3559,
      "step": 1413
    },
    {
      "epoch": 0.01414,
      "grad_norm": 0.575683339939827,
      "learning_rate": 0.003,
      "loss": 4.336,
      "step": 1414
    },
    {
      "epoch": 0.01415,
      "grad_norm": 0.5478713782500327,
      "learning_rate": 0.003,
      "loss": 4.3284,
      "step": 1415
    },
    {
      "epoch": 0.01416,
      "grad_norm": 0.48495351731286235,
      "learning_rate": 0.003,
      "loss": 4.3497,
      "step": 1416
    },
    {
      "epoch": 0.01417,
      "grad_norm": 0.5582320312672604,
      "learning_rate": 0.003,
      "loss": 4.3419,
      "step": 1417
    },
    {
      "epoch": 0.01418,
      "grad_norm": 0.6202372623477388,
      "learning_rate": 0.003,
      "loss": 4.3622,
      "step": 1418
    },
    {
      "epoch": 0.01419,
      "grad_norm": 0.6521670079720687,
      "learning_rate": 0.003,
      "loss": 4.3534,
      "step": 1419
    },
    {
      "epoch": 0.0142,
      "grad_norm": 0.5549508977810897,
      "learning_rate": 0.003,
      "loss": 4.3578,
      "step": 1420
    },
    {
      "epoch": 0.01421,
      "grad_norm": 0.45613666396542774,
      "learning_rate": 0.003,
      "loss": 4.3197,
      "step": 1421
    },
    {
      "epoch": 0.01422,
      "grad_norm": 0.445372260007137,
      "learning_rate": 0.003,
      "loss": 4.322,
      "step": 1422
    },
    {
      "epoch": 0.01423,
      "grad_norm": 0.38402591989219925,
      "learning_rate": 0.003,
      "loss": 4.3246,
      "step": 1423
    },
    {
      "epoch": 0.01424,
      "grad_norm": 0.38806384974852803,
      "learning_rate": 0.003,
      "loss": 4.3401,
      "step": 1424
    },
    {
      "epoch": 0.01425,
      "grad_norm": 0.4132316355770368,
      "learning_rate": 0.003,
      "loss": 4.3436,
      "step": 1425
    },
    {
      "epoch": 0.01426,
      "grad_norm": 0.3529318435487905,
      "learning_rate": 0.003,
      "loss": 4.3165,
      "step": 1426
    },
    {
      "epoch": 0.01427,
      "grad_norm": 0.3824674454914564,
      "learning_rate": 0.003,
      "loss": 4.3137,
      "step": 1427
    },
    {
      "epoch": 0.01428,
      "grad_norm": 0.40903598720809137,
      "learning_rate": 0.003,
      "loss": 4.3085,
      "step": 1428
    },
    {
      "epoch": 0.01429,
      "grad_norm": 0.48492866215569996,
      "learning_rate": 0.003,
      "loss": 4.3211,
      "step": 1429
    },
    {
      "epoch": 0.0143,
      "grad_norm": 0.5657427651213296,
      "learning_rate": 0.003,
      "loss": 4.3342,
      "step": 1430
    },
    {
      "epoch": 0.01431,
      "grad_norm": 0.5675857894892379,
      "learning_rate": 0.003,
      "loss": 4.3246,
      "step": 1431
    },
    {
      "epoch": 0.01432,
      "grad_norm": 0.5138100030227308,
      "learning_rate": 0.003,
      "loss": 4.3123,
      "step": 1432
    },
    {
      "epoch": 0.01433,
      "grad_norm": 0.3834195074788409,
      "learning_rate": 0.003,
      "loss": 4.3046,
      "step": 1433
    },
    {
      "epoch": 0.01434,
      "grad_norm": 0.380142493390664,
      "learning_rate": 0.003,
      "loss": 4.2898,
      "step": 1434
    },
    {
      "epoch": 0.01435,
      "grad_norm": 0.40030955161206494,
      "learning_rate": 0.003,
      "loss": 4.2992,
      "step": 1435
    },
    {
      "epoch": 0.01436,
      "grad_norm": 0.3954780968032791,
      "learning_rate": 0.003,
      "loss": 4.3173,
      "step": 1436
    },
    {
      "epoch": 0.01437,
      "grad_norm": 0.3507385052853375,
      "learning_rate": 0.003,
      "loss": 4.3285,
      "step": 1437
    },
    {
      "epoch": 0.01438,
      "grad_norm": 0.37418173745501515,
      "learning_rate": 0.003,
      "loss": 4.3228,
      "step": 1438
    },
    {
      "epoch": 0.01439,
      "grad_norm": 0.3823240257736278,
      "learning_rate": 0.003,
      "loss": 4.2864,
      "step": 1439
    },
    {
      "epoch": 0.0144,
      "grad_norm": 0.4883909418304791,
      "learning_rate": 0.003,
      "loss": 4.3011,
      "step": 1440
    },
    {
      "epoch": 0.01441,
      "grad_norm": 0.5551175791026393,
      "learning_rate": 0.003,
      "loss": 4.3126,
      "step": 1441
    },
    {
      "epoch": 0.01442,
      "grad_norm": 0.5566257513382789,
      "learning_rate": 0.003,
      "loss": 4.2964,
      "step": 1442
    },
    {
      "epoch": 0.01443,
      "grad_norm": 0.5444967253409229,
      "learning_rate": 0.003,
      "loss": 4.3586,
      "step": 1443
    },
    {
      "epoch": 0.01444,
      "grad_norm": 0.5868401236318755,
      "learning_rate": 0.003,
      "loss": 4.3246,
      "step": 1444
    },
    {
      "epoch": 0.01445,
      "grad_norm": 0.49247571907636584,
      "learning_rate": 0.003,
      "loss": 4.282,
      "step": 1445
    },
    {
      "epoch": 0.01446,
      "grad_norm": 0.4611996904427917,
      "learning_rate": 0.003,
      "loss": 4.2968,
      "step": 1446
    },
    {
      "epoch": 0.01447,
      "grad_norm": 0.4704096683460519,
      "learning_rate": 0.003,
      "loss": 4.3067,
      "step": 1447
    },
    {
      "epoch": 0.01448,
      "grad_norm": 0.39306824337135715,
      "learning_rate": 0.003,
      "loss": 4.3134,
      "step": 1448
    },
    {
      "epoch": 0.01449,
      "grad_norm": 0.41700654092261286,
      "learning_rate": 0.003,
      "loss": 4.3094,
      "step": 1449
    },
    {
      "epoch": 0.0145,
      "grad_norm": 0.407843055117697,
      "learning_rate": 0.003,
      "loss": 4.3225,
      "step": 1450
    },
    {
      "epoch": 0.01451,
      "grad_norm": 0.46619408673841983,
      "learning_rate": 0.003,
      "loss": 4.304,
      "step": 1451
    },
    {
      "epoch": 0.01452,
      "grad_norm": 0.5845723252800918,
      "learning_rate": 0.003,
      "loss": 4.3078,
      "step": 1452
    },
    {
      "epoch": 0.01453,
      "grad_norm": 0.6930673151633832,
      "learning_rate": 0.003,
      "loss": 4.3299,
      "step": 1453
    },
    {
      "epoch": 0.01454,
      "grad_norm": 0.7096402728772245,
      "learning_rate": 0.003,
      "loss": 4.2995,
      "step": 1454
    },
    {
      "epoch": 0.01455,
      "grad_norm": 0.6610327702633573,
      "learning_rate": 0.003,
      "loss": 4.3327,
      "step": 1455
    },
    {
      "epoch": 0.01456,
      "grad_norm": 0.6903492208096148,
      "learning_rate": 0.003,
      "loss": 4.3118,
      "step": 1456
    },
    {
      "epoch": 0.01457,
      "grad_norm": 0.5695969137152378,
      "learning_rate": 0.003,
      "loss": 4.2978,
      "step": 1457
    },
    {
      "epoch": 0.01458,
      "grad_norm": 0.5248689907637836,
      "learning_rate": 0.003,
      "loss": 4.2909,
      "step": 1458
    },
    {
      "epoch": 0.01459,
      "grad_norm": 0.49793282629814667,
      "learning_rate": 0.003,
      "loss": 4.318,
      "step": 1459
    },
    {
      "epoch": 0.0146,
      "grad_norm": 0.5259290959891172,
      "learning_rate": 0.003,
      "loss": 4.3252,
      "step": 1460
    },
    {
      "epoch": 0.01461,
      "grad_norm": 0.5423165316350705,
      "learning_rate": 0.003,
      "loss": 4.3026,
      "step": 1461
    },
    {
      "epoch": 0.01462,
      "grad_norm": 0.5122102135827509,
      "learning_rate": 0.003,
      "loss": 4.3083,
      "step": 1462
    },
    {
      "epoch": 0.01463,
      "grad_norm": 0.48806792365665164,
      "learning_rate": 0.003,
      "loss": 4.304,
      "step": 1463
    },
    {
      "epoch": 0.01464,
      "grad_norm": 0.4826165334345253,
      "learning_rate": 0.003,
      "loss": 4.2803,
      "step": 1464
    },
    {
      "epoch": 0.01465,
      "grad_norm": 0.48806303528119876,
      "learning_rate": 0.003,
      "loss": 4.325,
      "step": 1465
    },
    {
      "epoch": 0.01466,
      "grad_norm": 0.5288880864983577,
      "learning_rate": 0.003,
      "loss": 4.2948,
      "step": 1466
    },
    {
      "epoch": 0.01467,
      "grad_norm": 0.6678963880365116,
      "learning_rate": 0.003,
      "loss": 4.3367,
      "step": 1467
    },
    {
      "epoch": 0.01468,
      "grad_norm": 0.8507141482558338,
      "learning_rate": 0.003,
      "loss": 4.3259,
      "step": 1468
    },
    {
      "epoch": 0.01469,
      "grad_norm": 0.9674104680499097,
      "learning_rate": 0.003,
      "loss": 4.3427,
      "step": 1469
    },
    {
      "epoch": 0.0147,
      "grad_norm": 0.9577096701541407,
      "learning_rate": 0.003,
      "loss": 4.3502,
      "step": 1470
    },
    {
      "epoch": 0.01471,
      "grad_norm": 0.9021599056855796,
      "learning_rate": 0.003,
      "loss": 4.3567,
      "step": 1471
    },
    {
      "epoch": 0.01472,
      "grad_norm": 0.8933064062959534,
      "learning_rate": 0.003,
      "loss": 4.3477,
      "step": 1472
    },
    {
      "epoch": 0.01473,
      "grad_norm": 0.8969620681941299,
      "learning_rate": 0.003,
      "loss": 4.3708,
      "step": 1473
    },
    {
      "epoch": 0.01474,
      "grad_norm": 0.6674847167774206,
      "learning_rate": 0.003,
      "loss": 4.3402,
      "step": 1474
    },
    {
      "epoch": 0.01475,
      "grad_norm": 0.7252535747045,
      "learning_rate": 0.003,
      "loss": 4.3675,
      "step": 1475
    },
    {
      "epoch": 0.01476,
      "grad_norm": 0.7222358160132806,
      "learning_rate": 0.003,
      "loss": 4.3503,
      "step": 1476
    },
    {
      "epoch": 0.01477,
      "grad_norm": 0.837925142735814,
      "learning_rate": 0.003,
      "loss": 4.3614,
      "step": 1477
    },
    {
      "epoch": 0.01478,
      "grad_norm": 0.8571430973709657,
      "learning_rate": 0.003,
      "loss": 4.3541,
      "step": 1478
    },
    {
      "epoch": 0.01479,
      "grad_norm": 0.7748951970932865,
      "learning_rate": 0.003,
      "loss": 4.3832,
      "step": 1479
    },
    {
      "epoch": 0.0148,
      "grad_norm": 0.5769653553075942,
      "learning_rate": 0.003,
      "loss": 4.3406,
      "step": 1480
    },
    {
      "epoch": 0.01481,
      "grad_norm": 0.5617397637544572,
      "learning_rate": 0.003,
      "loss": 4.343,
      "step": 1481
    },
    {
      "epoch": 0.01482,
      "grad_norm": 0.494589006608917,
      "learning_rate": 0.003,
      "loss": 4.3215,
      "step": 1482
    },
    {
      "epoch": 0.01483,
      "grad_norm": 0.5358352358514579,
      "learning_rate": 0.003,
      "loss": 4.3218,
      "step": 1483
    },
    {
      "epoch": 0.01484,
      "grad_norm": 0.4638148892097882,
      "learning_rate": 0.003,
      "loss": 4.3656,
      "step": 1484
    },
    {
      "epoch": 0.01485,
      "grad_norm": 0.5225092823087141,
      "learning_rate": 0.003,
      "loss": 4.3308,
      "step": 1485
    },
    {
      "epoch": 0.01486,
      "grad_norm": 0.5536760113665742,
      "learning_rate": 0.003,
      "loss": 4.3262,
      "step": 1486
    },
    {
      "epoch": 0.01487,
      "grad_norm": 0.5280609352965188,
      "learning_rate": 0.003,
      "loss": 4.3256,
      "step": 1487
    },
    {
      "epoch": 0.01488,
      "grad_norm": 0.6169285208705055,
      "learning_rate": 0.003,
      "loss": 4.3379,
      "step": 1488
    },
    {
      "epoch": 0.01489,
      "grad_norm": 0.7420878936102314,
      "learning_rate": 0.003,
      "loss": 4.3425,
      "step": 1489
    },
    {
      "epoch": 0.0149,
      "grad_norm": 0.7781493014062594,
      "learning_rate": 0.003,
      "loss": 4.3387,
      "step": 1490
    },
    {
      "epoch": 0.01491,
      "grad_norm": 0.6235654472208051,
      "learning_rate": 0.003,
      "loss": 4.3459,
      "step": 1491
    },
    {
      "epoch": 0.01492,
      "grad_norm": 0.5493424470537548,
      "learning_rate": 0.003,
      "loss": 4.3048,
      "step": 1492
    },
    {
      "epoch": 0.01493,
      "grad_norm": 0.5353895236188659,
      "learning_rate": 0.003,
      "loss": 4.3336,
      "step": 1493
    },
    {
      "epoch": 0.01494,
      "grad_norm": 0.4821593967850022,
      "learning_rate": 0.003,
      "loss": 4.3113,
      "step": 1494
    },
    {
      "epoch": 0.01495,
      "grad_norm": 0.4536978313935068,
      "learning_rate": 0.003,
      "loss": 4.3201,
      "step": 1495
    },
    {
      "epoch": 0.01496,
      "grad_norm": 0.4449517853772135,
      "learning_rate": 0.003,
      "loss": 4.294,
      "step": 1496
    },
    {
      "epoch": 0.01497,
      "grad_norm": 0.39867192565794446,
      "learning_rate": 0.003,
      "loss": 4.2656,
      "step": 1497
    },
    {
      "epoch": 0.01498,
      "grad_norm": 0.3910710132293917,
      "learning_rate": 0.003,
      "loss": 4.3202,
      "step": 1498
    },
    {
      "epoch": 0.01499,
      "grad_norm": 0.368047206254813,
      "learning_rate": 0.003,
      "loss": 4.339,
      "step": 1499
    },
    {
      "epoch": 0.015,
      "grad_norm": 0.34050808541322203,
      "learning_rate": 0.003,
      "loss": 4.2902,
      "step": 1500
    },
    {
      "epoch": 0.01501,
      "grad_norm": 0.36125568650846523,
      "learning_rate": 0.003,
      "loss": 4.2886,
      "step": 1501
    },
    {
      "epoch": 0.01502,
      "grad_norm": 0.3250619223936781,
      "learning_rate": 0.003,
      "loss": 4.3299,
      "step": 1502
    },
    {
      "epoch": 0.01503,
      "grad_norm": 0.362004630486072,
      "learning_rate": 0.003,
      "loss": 4.3091,
      "step": 1503
    },
    {
      "epoch": 0.01504,
      "grad_norm": 0.42640653384457183,
      "learning_rate": 0.003,
      "loss": 4.3092,
      "step": 1504
    },
    {
      "epoch": 0.01505,
      "grad_norm": 0.4655634074328374,
      "learning_rate": 0.003,
      "loss": 4.2935,
      "step": 1505
    },
    {
      "epoch": 0.01506,
      "grad_norm": 0.4830681135087923,
      "learning_rate": 0.003,
      "loss": 4.3131,
      "step": 1506
    },
    {
      "epoch": 0.01507,
      "grad_norm": 0.45072861631675815,
      "learning_rate": 0.003,
      "loss": 4.3252,
      "step": 1507
    },
    {
      "epoch": 0.01508,
      "grad_norm": 0.41884675318056874,
      "learning_rate": 0.003,
      "loss": 4.2996,
      "step": 1508
    },
    {
      "epoch": 0.01509,
      "grad_norm": 0.45582678472670524,
      "learning_rate": 0.003,
      "loss": 4.32,
      "step": 1509
    },
    {
      "epoch": 0.0151,
      "grad_norm": 0.4555994257034133,
      "learning_rate": 0.003,
      "loss": 4.2923,
      "step": 1510
    },
    {
      "epoch": 0.01511,
      "grad_norm": 0.47932862889061606,
      "learning_rate": 0.003,
      "loss": 4.2807,
      "step": 1511
    },
    {
      "epoch": 0.01512,
      "grad_norm": 0.4824629582897255,
      "learning_rate": 0.003,
      "loss": 4.2928,
      "step": 1512
    },
    {
      "epoch": 0.01513,
      "grad_norm": 0.4551520372558624,
      "learning_rate": 0.003,
      "loss": 4.2958,
      "step": 1513
    },
    {
      "epoch": 0.01514,
      "grad_norm": 0.40280658680118386,
      "learning_rate": 0.003,
      "loss": 4.3004,
      "step": 1514
    },
    {
      "epoch": 0.01515,
      "grad_norm": 0.4291367693804096,
      "learning_rate": 0.003,
      "loss": 4.2938,
      "step": 1515
    },
    {
      "epoch": 0.01516,
      "grad_norm": 0.5124558841404536,
      "learning_rate": 0.003,
      "loss": 4.3014,
      "step": 1516
    },
    {
      "epoch": 0.01517,
      "grad_norm": 0.662211893130953,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 1517
    },
    {
      "epoch": 0.01518,
      "grad_norm": 0.7783947565165081,
      "learning_rate": 0.003,
      "loss": 4.3029,
      "step": 1518
    },
    {
      "epoch": 0.01519,
      "grad_norm": 0.7115931895318321,
      "learning_rate": 0.003,
      "loss": 4.3425,
      "step": 1519
    },
    {
      "epoch": 0.0152,
      "grad_norm": 0.727509066973132,
      "learning_rate": 0.003,
      "loss": 4.2935,
      "step": 1520
    },
    {
      "epoch": 0.01521,
      "grad_norm": 0.8200160562394482,
      "learning_rate": 0.003,
      "loss": 4.3183,
      "step": 1521
    },
    {
      "epoch": 0.01522,
      "grad_norm": 0.785335657111215,
      "learning_rate": 0.003,
      "loss": 4.3154,
      "step": 1522
    },
    {
      "epoch": 0.01523,
      "grad_norm": 0.6269328513970822,
      "learning_rate": 0.003,
      "loss": 4.3029,
      "step": 1523
    },
    {
      "epoch": 0.01524,
      "grad_norm": 0.6280834911768081,
      "learning_rate": 0.003,
      "loss": 4.3392,
      "step": 1524
    },
    {
      "epoch": 0.01525,
      "grad_norm": 0.562699648791423,
      "learning_rate": 0.003,
      "loss": 4.3367,
      "step": 1525
    },
    {
      "epoch": 0.01526,
      "grad_norm": 0.6402790783462683,
      "learning_rate": 0.003,
      "loss": 4.3277,
      "step": 1526
    },
    {
      "epoch": 0.01527,
      "grad_norm": 0.6038705757661785,
      "learning_rate": 0.003,
      "loss": 4.2979,
      "step": 1527
    },
    {
      "epoch": 0.01528,
      "grad_norm": 0.636946924703658,
      "learning_rate": 0.003,
      "loss": 4.3174,
      "step": 1528
    },
    {
      "epoch": 0.01529,
      "grad_norm": 0.6521902920068134,
      "learning_rate": 0.003,
      "loss": 4.3136,
      "step": 1529
    },
    {
      "epoch": 0.0153,
      "grad_norm": 0.7380737016000076,
      "learning_rate": 0.003,
      "loss": 4.3247,
      "step": 1530
    },
    {
      "epoch": 0.01531,
      "grad_norm": 0.7548829272902333,
      "learning_rate": 0.003,
      "loss": 4.3243,
      "step": 1531
    },
    {
      "epoch": 0.01532,
      "grad_norm": 0.7137520123617993,
      "learning_rate": 0.003,
      "loss": 4.3014,
      "step": 1532
    },
    {
      "epoch": 0.01533,
      "grad_norm": 0.7523921063923493,
      "learning_rate": 0.003,
      "loss": 4.3327,
      "step": 1533
    },
    {
      "epoch": 0.01534,
      "grad_norm": 0.7506503205085225,
      "learning_rate": 0.003,
      "loss": 4.3336,
      "step": 1534
    },
    {
      "epoch": 0.01535,
      "grad_norm": 0.7303716243088675,
      "learning_rate": 0.003,
      "loss": 4.3418,
      "step": 1535
    },
    {
      "epoch": 0.01536,
      "grad_norm": 0.6839262408941584,
      "learning_rate": 0.003,
      "loss": 4.32,
      "step": 1536
    },
    {
      "epoch": 0.01537,
      "grad_norm": 0.7109733936082817,
      "learning_rate": 0.003,
      "loss": 4.3199,
      "step": 1537
    },
    {
      "epoch": 0.01538,
      "grad_norm": 0.7061208974263588,
      "learning_rate": 0.003,
      "loss": 4.296,
      "step": 1538
    },
    {
      "epoch": 0.01539,
      "grad_norm": 0.6986672189571356,
      "learning_rate": 0.003,
      "loss": 4.3349,
      "step": 1539
    },
    {
      "epoch": 0.0154,
      "grad_norm": 0.6534468982180698,
      "learning_rate": 0.003,
      "loss": 4.333,
      "step": 1540
    },
    {
      "epoch": 0.01541,
      "grad_norm": 0.49825734343698913,
      "learning_rate": 0.003,
      "loss": 4.3068,
      "step": 1541
    },
    {
      "epoch": 0.01542,
      "grad_norm": 0.44352017421476997,
      "learning_rate": 0.003,
      "loss": 4.3332,
      "step": 1542
    },
    {
      "epoch": 0.01543,
      "grad_norm": 0.38211515055441725,
      "learning_rate": 0.003,
      "loss": 4.3129,
      "step": 1543
    },
    {
      "epoch": 0.01544,
      "grad_norm": 0.38911843319302664,
      "learning_rate": 0.003,
      "loss": 4.2994,
      "step": 1544
    },
    {
      "epoch": 0.01545,
      "grad_norm": 0.3624309615071896,
      "learning_rate": 0.003,
      "loss": 4.3625,
      "step": 1545
    },
    {
      "epoch": 0.01546,
      "grad_norm": 0.4569500368392657,
      "learning_rate": 0.003,
      "loss": 4.3313,
      "step": 1546
    },
    {
      "epoch": 0.01547,
      "grad_norm": 0.5377623135091468,
      "learning_rate": 0.003,
      "loss": 4.3355,
      "step": 1547
    },
    {
      "epoch": 0.01548,
      "grad_norm": 0.6297787769238979,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1548
    },
    {
      "epoch": 0.01549,
      "grad_norm": 0.644439142847034,
      "learning_rate": 0.003,
      "loss": 4.2897,
      "step": 1549
    },
    {
      "epoch": 0.0155,
      "grad_norm": 0.618142851349942,
      "learning_rate": 0.003,
      "loss": 4.3109,
      "step": 1550
    },
    {
      "epoch": 0.01551,
      "grad_norm": 0.5448565014455535,
      "learning_rate": 0.003,
      "loss": 4.2869,
      "step": 1551
    },
    {
      "epoch": 0.01552,
      "grad_norm": 0.5723351701615428,
      "learning_rate": 0.003,
      "loss": 4.3071,
      "step": 1552
    },
    {
      "epoch": 0.01553,
      "grad_norm": 0.6022810466873098,
      "learning_rate": 0.003,
      "loss": 4.3328,
      "step": 1553
    },
    {
      "epoch": 0.01554,
      "grad_norm": 0.6989931225913981,
      "learning_rate": 0.003,
      "loss": 4.324,
      "step": 1554
    },
    {
      "epoch": 0.01555,
      "grad_norm": 0.6580206580270094,
      "learning_rate": 0.003,
      "loss": 4.2931,
      "step": 1555
    },
    {
      "epoch": 0.01556,
      "grad_norm": 0.6533983591917718,
      "learning_rate": 0.003,
      "loss": 4.3308,
      "step": 1556
    },
    {
      "epoch": 0.01557,
      "grad_norm": 0.6295945004011335,
      "learning_rate": 0.003,
      "loss": 4.3036,
      "step": 1557
    },
    {
      "epoch": 0.01558,
      "grad_norm": 0.5371854889087798,
      "learning_rate": 0.003,
      "loss": 4.2914,
      "step": 1558
    },
    {
      "epoch": 0.01559,
      "grad_norm": 0.4632942102113827,
      "learning_rate": 0.003,
      "loss": 4.326,
      "step": 1559
    },
    {
      "epoch": 0.0156,
      "grad_norm": 0.5051005361340883,
      "learning_rate": 0.003,
      "loss": 4.3169,
      "step": 1560
    },
    {
      "epoch": 0.01561,
      "grad_norm": 0.5283819882561609,
      "learning_rate": 0.003,
      "loss": 4.299,
      "step": 1561
    },
    {
      "epoch": 0.01562,
      "grad_norm": 0.591562612461458,
      "learning_rate": 0.003,
      "loss": 4.2958,
      "step": 1562
    },
    {
      "epoch": 0.01563,
      "grad_norm": 0.6357918082942466,
      "learning_rate": 0.003,
      "loss": 4.3041,
      "step": 1563
    },
    {
      "epoch": 0.01564,
      "grad_norm": 0.6669974739393074,
      "learning_rate": 0.003,
      "loss": 4.3336,
      "step": 1564
    },
    {
      "epoch": 0.01565,
      "grad_norm": 0.691137766433848,
      "learning_rate": 0.003,
      "loss": 4.3141,
      "step": 1565
    },
    {
      "epoch": 0.01566,
      "grad_norm": 0.5930174244490458,
      "learning_rate": 0.003,
      "loss": 4.3298,
      "step": 1566
    },
    {
      "epoch": 0.01567,
      "grad_norm": 0.551100674489701,
      "learning_rate": 0.003,
      "loss": 4.319,
      "step": 1567
    },
    {
      "epoch": 0.01568,
      "grad_norm": 0.5879757029751648,
      "learning_rate": 0.003,
      "loss": 4.3039,
      "step": 1568
    },
    {
      "epoch": 0.01569,
      "grad_norm": 0.5302512625540108,
      "learning_rate": 0.003,
      "loss": 4.3069,
      "step": 1569
    },
    {
      "epoch": 0.0157,
      "grad_norm": 0.49569383773361164,
      "learning_rate": 0.003,
      "loss": 4.3233,
      "step": 1570
    },
    {
      "epoch": 0.01571,
      "grad_norm": 0.4733165510784743,
      "learning_rate": 0.003,
      "loss": 4.3065,
      "step": 1571
    },
    {
      "epoch": 0.01572,
      "grad_norm": 0.4449250863339169,
      "learning_rate": 0.003,
      "loss": 4.3135,
      "step": 1572
    },
    {
      "epoch": 0.01573,
      "grad_norm": 0.3929931575857813,
      "learning_rate": 0.003,
      "loss": 4.2941,
      "step": 1573
    },
    {
      "epoch": 0.01574,
      "grad_norm": 0.4142053233129594,
      "learning_rate": 0.003,
      "loss": 4.3098,
      "step": 1574
    },
    {
      "epoch": 0.01575,
      "grad_norm": 0.3709436905014968,
      "learning_rate": 0.003,
      "loss": 4.2702,
      "step": 1575
    },
    {
      "epoch": 0.01576,
      "grad_norm": 0.36495283580117344,
      "learning_rate": 0.003,
      "loss": 4.2914,
      "step": 1576
    },
    {
      "epoch": 0.01577,
      "grad_norm": 0.3915297767332071,
      "learning_rate": 0.003,
      "loss": 4.2525,
      "step": 1577
    },
    {
      "epoch": 0.01578,
      "grad_norm": 0.5001638890424236,
      "learning_rate": 0.003,
      "loss": 4.284,
      "step": 1578
    },
    {
      "epoch": 0.01579,
      "grad_norm": 0.6179833768395876,
      "learning_rate": 0.003,
      "loss": 4.2911,
      "step": 1579
    },
    {
      "epoch": 0.0158,
      "grad_norm": 0.6342638472068223,
      "learning_rate": 0.003,
      "loss": 4.299,
      "step": 1580
    },
    {
      "epoch": 0.01581,
      "grad_norm": 0.5432322505669018,
      "learning_rate": 0.003,
      "loss": 4.3062,
      "step": 1581
    },
    {
      "epoch": 0.01582,
      "grad_norm": 0.4889222240722665,
      "learning_rate": 0.003,
      "loss": 4.2846,
      "step": 1582
    },
    {
      "epoch": 0.01583,
      "grad_norm": 0.5515618698965413,
      "learning_rate": 0.003,
      "loss": 4.274,
      "step": 1583
    },
    {
      "epoch": 0.01584,
      "grad_norm": 0.549844036531772,
      "learning_rate": 0.003,
      "loss": 4.3079,
      "step": 1584
    },
    {
      "epoch": 0.01585,
      "grad_norm": 0.48716680666102685,
      "learning_rate": 0.003,
      "loss": 4.2981,
      "step": 1585
    },
    {
      "epoch": 0.01586,
      "grad_norm": 0.623103221118658,
      "learning_rate": 0.003,
      "loss": 4.2774,
      "step": 1586
    },
    {
      "epoch": 0.01587,
      "grad_norm": 0.7984823726854448,
      "learning_rate": 0.003,
      "loss": 4.3077,
      "step": 1587
    },
    {
      "epoch": 0.01588,
      "grad_norm": 0.7627654453676783,
      "learning_rate": 0.003,
      "loss": 4.3371,
      "step": 1588
    },
    {
      "epoch": 0.01589,
      "grad_norm": 0.7599095947865857,
      "learning_rate": 0.003,
      "loss": 4.3312,
      "step": 1589
    },
    {
      "epoch": 0.0159,
      "grad_norm": 0.8159908089496294,
      "learning_rate": 0.003,
      "loss": 4.33,
      "step": 1590
    },
    {
      "epoch": 0.01591,
      "grad_norm": 0.8803872581529493,
      "learning_rate": 0.003,
      "loss": 4.3476,
      "step": 1591
    },
    {
      "epoch": 0.01592,
      "grad_norm": 0.9820230872746389,
      "learning_rate": 0.003,
      "loss": 4.349,
      "step": 1592
    },
    {
      "epoch": 0.01593,
      "grad_norm": 0.9673129749259365,
      "learning_rate": 0.003,
      "loss": 4.352,
      "step": 1593
    },
    {
      "epoch": 0.01594,
      "grad_norm": 0.9293561890551637,
      "learning_rate": 0.003,
      "loss": 4.3487,
      "step": 1594
    },
    {
      "epoch": 0.01595,
      "grad_norm": 0.9561454087951702,
      "learning_rate": 0.003,
      "loss": 4.3811,
      "step": 1595
    },
    {
      "epoch": 0.01596,
      "grad_norm": 1.0372746249351796,
      "learning_rate": 0.003,
      "loss": 4.3847,
      "step": 1596
    },
    {
      "epoch": 0.01597,
      "grad_norm": 0.9040299607313536,
      "learning_rate": 0.003,
      "loss": 4.3427,
      "step": 1597
    },
    {
      "epoch": 0.01598,
      "grad_norm": 1.1389328260746334,
      "learning_rate": 0.003,
      "loss": 4.3536,
      "step": 1598
    },
    {
      "epoch": 0.01599,
      "grad_norm": 1.097139764937509,
      "learning_rate": 0.003,
      "loss": 4.3856,
      "step": 1599
    },
    {
      "epoch": 0.016,
      "grad_norm": 0.9605869668937348,
      "learning_rate": 0.003,
      "loss": 4.3534,
      "step": 1600
    },
    {
      "epoch": 0.01601,
      "grad_norm": 0.7696982537009223,
      "learning_rate": 0.003,
      "loss": 4.3899,
      "step": 1601
    },
    {
      "epoch": 0.01602,
      "grad_norm": 0.7083103903580285,
      "learning_rate": 0.003,
      "loss": 4.3647,
      "step": 1602
    },
    {
      "epoch": 0.01603,
      "grad_norm": 0.5787396432841793,
      "learning_rate": 0.003,
      "loss": 4.3713,
      "step": 1603
    },
    {
      "epoch": 0.01604,
      "grad_norm": 0.5028516500547362,
      "learning_rate": 0.003,
      "loss": 4.3577,
      "step": 1604
    },
    {
      "epoch": 0.01605,
      "grad_norm": 0.4938813453051197,
      "learning_rate": 0.003,
      "loss": 4.3306,
      "step": 1605
    },
    {
      "epoch": 0.01606,
      "grad_norm": 0.5541639051688618,
      "learning_rate": 0.003,
      "loss": 4.3313,
      "step": 1606
    },
    {
      "epoch": 0.01607,
      "grad_norm": 0.5139661852261925,
      "learning_rate": 0.003,
      "loss": 4.3576,
      "step": 1607
    },
    {
      "epoch": 0.01608,
      "grad_norm": 0.48348767192426967,
      "learning_rate": 0.003,
      "loss": 4.3253,
      "step": 1608
    },
    {
      "epoch": 0.01609,
      "grad_norm": 0.40136757384139854,
      "learning_rate": 0.003,
      "loss": 4.3283,
      "step": 1609
    },
    {
      "epoch": 0.0161,
      "grad_norm": 0.4218183451690125,
      "learning_rate": 0.003,
      "loss": 4.2908,
      "step": 1610
    },
    {
      "epoch": 0.01611,
      "grad_norm": 0.460848545000983,
      "learning_rate": 0.003,
      "loss": 4.31,
      "step": 1611
    },
    {
      "epoch": 0.01612,
      "grad_norm": 0.4659444292137467,
      "learning_rate": 0.003,
      "loss": 4.3003,
      "step": 1612
    },
    {
      "epoch": 0.01613,
      "grad_norm": 0.4112363124807694,
      "learning_rate": 0.003,
      "loss": 4.3411,
      "step": 1613
    },
    {
      "epoch": 0.01614,
      "grad_norm": 0.3883756371913985,
      "learning_rate": 0.003,
      "loss": 4.3624,
      "step": 1614
    },
    {
      "epoch": 0.01615,
      "grad_norm": 0.3536906211225033,
      "learning_rate": 0.003,
      "loss": 4.3264,
      "step": 1615
    },
    {
      "epoch": 0.01616,
      "grad_norm": 0.31345780077250956,
      "learning_rate": 0.003,
      "loss": 4.3029,
      "step": 1616
    },
    {
      "epoch": 0.01617,
      "grad_norm": 0.30058841204286124,
      "learning_rate": 0.003,
      "loss": 4.2916,
      "step": 1617
    },
    {
      "epoch": 0.01618,
      "grad_norm": 0.27199761658041965,
      "learning_rate": 0.003,
      "loss": 4.2925,
      "step": 1618
    },
    {
      "epoch": 0.01619,
      "grad_norm": 0.24962840642226738,
      "learning_rate": 0.003,
      "loss": 4.2798,
      "step": 1619
    },
    {
      "epoch": 0.0162,
      "grad_norm": 0.2624146116891587,
      "learning_rate": 0.003,
      "loss": 4.2863,
      "step": 1620
    },
    {
      "epoch": 0.01621,
      "grad_norm": 0.2537767483196262,
      "learning_rate": 0.003,
      "loss": 4.2811,
      "step": 1621
    },
    {
      "epoch": 0.01622,
      "grad_norm": 0.3339010418458089,
      "learning_rate": 0.003,
      "loss": 4.3143,
      "step": 1622
    },
    {
      "epoch": 0.01623,
      "grad_norm": 0.4481356309307089,
      "learning_rate": 0.003,
      "loss": 4.2931,
      "step": 1623
    },
    {
      "epoch": 0.01624,
      "grad_norm": 0.6923232940858078,
      "learning_rate": 0.003,
      "loss": 4.3139,
      "step": 1624
    },
    {
      "epoch": 0.01625,
      "grad_norm": 0.796128597644668,
      "learning_rate": 0.003,
      "loss": 4.3039,
      "step": 1625
    },
    {
      "epoch": 0.01626,
      "grad_norm": 0.4972515352836982,
      "learning_rate": 0.003,
      "loss": 4.3127,
      "step": 1626
    },
    {
      "epoch": 0.01627,
      "grad_norm": 0.5373908693508559,
      "learning_rate": 0.003,
      "loss": 4.2845,
      "step": 1627
    },
    {
      "epoch": 0.01628,
      "grad_norm": 0.576906019908009,
      "learning_rate": 0.003,
      "loss": 4.2936,
      "step": 1628
    },
    {
      "epoch": 0.01629,
      "grad_norm": 0.4857532494288208,
      "learning_rate": 0.003,
      "loss": 4.2886,
      "step": 1629
    },
    {
      "epoch": 0.0163,
      "grad_norm": 0.6536351383709068,
      "learning_rate": 0.003,
      "loss": 4.284,
      "step": 1630
    },
    {
      "epoch": 0.01631,
      "grad_norm": 0.6136531914830949,
      "learning_rate": 0.003,
      "loss": 4.3036,
      "step": 1631
    },
    {
      "epoch": 0.01632,
      "grad_norm": 0.5441543421309533,
      "learning_rate": 0.003,
      "loss": 4.3133,
      "step": 1632
    },
    {
      "epoch": 0.01633,
      "grad_norm": 0.5546151460088474,
      "learning_rate": 0.003,
      "loss": 4.2852,
      "step": 1633
    },
    {
      "epoch": 0.01634,
      "grad_norm": 0.5283961307973114,
      "learning_rate": 0.003,
      "loss": 4.3028,
      "step": 1634
    },
    {
      "epoch": 0.01635,
      "grad_norm": 0.4867354243481517,
      "learning_rate": 0.003,
      "loss": 4.2796,
      "step": 1635
    },
    {
      "epoch": 0.01636,
      "grad_norm": 0.565294038636392,
      "learning_rate": 0.003,
      "loss": 4.2994,
      "step": 1636
    },
    {
      "epoch": 0.01637,
      "grad_norm": 0.571317155010679,
      "learning_rate": 0.003,
      "loss": 4.2844,
      "step": 1637
    },
    {
      "epoch": 0.01638,
      "grad_norm": 0.5573027415974858,
      "learning_rate": 0.003,
      "loss": 4.2921,
      "step": 1638
    },
    {
      "epoch": 0.01639,
      "grad_norm": 0.6223176506662222,
      "learning_rate": 0.003,
      "loss": 4.3165,
      "step": 1639
    },
    {
      "epoch": 0.0164,
      "grad_norm": 0.692259125142237,
      "learning_rate": 0.003,
      "loss": 4.3033,
      "step": 1640
    },
    {
      "epoch": 0.01641,
      "grad_norm": 0.5775190344518869,
      "learning_rate": 0.003,
      "loss": 4.2697,
      "step": 1641
    },
    {
      "epoch": 0.01642,
      "grad_norm": 0.5727565418624034,
      "learning_rate": 0.003,
      "loss": 4.316,
      "step": 1642
    },
    {
      "epoch": 0.01643,
      "grad_norm": 0.49414286165639626,
      "learning_rate": 0.003,
      "loss": 4.2959,
      "step": 1643
    },
    {
      "epoch": 0.01644,
      "grad_norm": 0.4716768356076658,
      "learning_rate": 0.003,
      "loss": 4.315,
      "step": 1644
    },
    {
      "epoch": 0.01645,
      "grad_norm": 0.4203296262888758,
      "learning_rate": 0.003,
      "loss": 4.2571,
      "step": 1645
    },
    {
      "epoch": 0.01646,
      "grad_norm": 0.42896262234184174,
      "learning_rate": 0.003,
      "loss": 4.2952,
      "step": 1646
    },
    {
      "epoch": 0.01647,
      "grad_norm": 0.4170543756256854,
      "learning_rate": 0.003,
      "loss": 4.2886,
      "step": 1647
    },
    {
      "epoch": 0.01648,
      "grad_norm": 0.42426821016871286,
      "learning_rate": 0.003,
      "loss": 4.2659,
      "step": 1648
    },
    {
      "epoch": 0.01649,
      "grad_norm": 0.4471925912241752,
      "learning_rate": 0.003,
      "loss": 4.3027,
      "step": 1649
    },
    {
      "epoch": 0.0165,
      "grad_norm": 0.512725534748134,
      "learning_rate": 0.003,
      "loss": 4.2806,
      "step": 1650
    },
    {
      "epoch": 0.01651,
      "grad_norm": 0.7005737333281123,
      "learning_rate": 0.003,
      "loss": 4.2999,
      "step": 1651
    },
    {
      "epoch": 0.01652,
      "grad_norm": 0.7594561646979391,
      "learning_rate": 0.003,
      "loss": 4.2643,
      "step": 1652
    },
    {
      "epoch": 0.01653,
      "grad_norm": 0.5870095922616082,
      "learning_rate": 0.003,
      "loss": 4.2994,
      "step": 1653
    },
    {
      "epoch": 0.01654,
      "grad_norm": 0.6476208006318457,
      "learning_rate": 0.003,
      "loss": 4.289,
      "step": 1654
    },
    {
      "epoch": 0.01655,
      "grad_norm": 0.6928825876927465,
      "learning_rate": 0.003,
      "loss": 4.3102,
      "step": 1655
    },
    {
      "epoch": 0.01656,
      "grad_norm": 0.6117569915496687,
      "learning_rate": 0.003,
      "loss": 4.2725,
      "step": 1656
    },
    {
      "epoch": 0.01657,
      "grad_norm": 0.614527863212692,
      "learning_rate": 0.003,
      "loss": 4.2915,
      "step": 1657
    },
    {
      "epoch": 0.01658,
      "grad_norm": 0.5818340523746026,
      "learning_rate": 0.003,
      "loss": 4.2809,
      "step": 1658
    },
    {
      "epoch": 0.01659,
      "grad_norm": 0.5122965623311254,
      "learning_rate": 0.003,
      "loss": 4.2829,
      "step": 1659
    },
    {
      "epoch": 0.0166,
      "grad_norm": 0.41721830635550766,
      "learning_rate": 0.003,
      "loss": 4.2624,
      "step": 1660
    },
    {
      "epoch": 0.01661,
      "grad_norm": 0.3974257488587888,
      "learning_rate": 0.003,
      "loss": 4.2659,
      "step": 1661
    },
    {
      "epoch": 0.01662,
      "grad_norm": 0.4134218787914048,
      "learning_rate": 0.003,
      "loss": 4.2699,
      "step": 1662
    },
    {
      "epoch": 0.01663,
      "grad_norm": 0.4716687664436207,
      "learning_rate": 0.003,
      "loss": 4.2937,
      "step": 1663
    },
    {
      "epoch": 0.01664,
      "grad_norm": 0.5068611880377488,
      "learning_rate": 0.003,
      "loss": 4.31,
      "step": 1664
    },
    {
      "epoch": 0.01665,
      "grad_norm": 0.49884944088963573,
      "learning_rate": 0.003,
      "loss": 4.2734,
      "step": 1665
    },
    {
      "epoch": 0.01666,
      "grad_norm": 0.5134497113162678,
      "learning_rate": 0.003,
      "loss": 4.277,
      "step": 1666
    },
    {
      "epoch": 0.01667,
      "grad_norm": 0.4951307012977702,
      "learning_rate": 0.003,
      "loss": 4.2916,
      "step": 1667
    },
    {
      "epoch": 0.01668,
      "grad_norm": 0.45857349182650015,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 1668
    },
    {
      "epoch": 0.01669,
      "grad_norm": 0.49049575707127974,
      "learning_rate": 0.003,
      "loss": 4.2801,
      "step": 1669
    },
    {
      "epoch": 0.0167,
      "grad_norm": 0.4594967732301719,
      "learning_rate": 0.003,
      "loss": 4.2692,
      "step": 1670
    },
    {
      "epoch": 0.01671,
      "grad_norm": 0.5516090968921173,
      "learning_rate": 0.003,
      "loss": 4.2597,
      "step": 1671
    },
    {
      "epoch": 0.01672,
      "grad_norm": 0.6238914771966855,
      "learning_rate": 0.003,
      "loss": 4.2771,
      "step": 1672
    },
    {
      "epoch": 0.01673,
      "grad_norm": 0.6444039300913702,
      "learning_rate": 0.003,
      "loss": 4.2759,
      "step": 1673
    },
    {
      "epoch": 0.01674,
      "grad_norm": 0.6140826000846973,
      "learning_rate": 0.003,
      "loss": 4.2936,
      "step": 1674
    },
    {
      "epoch": 0.01675,
      "grad_norm": 0.5952627086225425,
      "learning_rate": 0.003,
      "loss": 4.2797,
      "step": 1675
    },
    {
      "epoch": 0.01676,
      "grad_norm": 0.7397767991708332,
      "learning_rate": 0.003,
      "loss": 4.295,
      "step": 1676
    },
    {
      "epoch": 0.01677,
      "grad_norm": 0.6917176398613138,
      "learning_rate": 0.003,
      "loss": 4.287,
      "step": 1677
    },
    {
      "epoch": 0.01678,
      "grad_norm": 0.6139182789896808,
      "learning_rate": 0.003,
      "loss": 4.2961,
      "step": 1678
    },
    {
      "epoch": 0.01679,
      "grad_norm": 0.5929821413447228,
      "learning_rate": 0.003,
      "loss": 4.28,
      "step": 1679
    },
    {
      "epoch": 0.0168,
      "grad_norm": 0.5995626344940229,
      "learning_rate": 0.003,
      "loss": 4.2962,
      "step": 1680
    },
    {
      "epoch": 0.01681,
      "grad_norm": 0.5718502872460155,
      "learning_rate": 0.003,
      "loss": 4.2946,
      "step": 1681
    },
    {
      "epoch": 0.01682,
      "grad_norm": 0.5792048131668902,
      "learning_rate": 0.003,
      "loss": 4.2929,
      "step": 1682
    },
    {
      "epoch": 0.01683,
      "grad_norm": 0.5439228911454255,
      "learning_rate": 0.003,
      "loss": 4.3021,
      "step": 1683
    },
    {
      "epoch": 0.01684,
      "grad_norm": 0.5499788418796372,
      "learning_rate": 0.003,
      "loss": 4.2762,
      "step": 1684
    },
    {
      "epoch": 0.01685,
      "grad_norm": 0.5923021072734571,
      "learning_rate": 0.003,
      "loss": 4.2948,
      "step": 1685
    },
    {
      "epoch": 0.01686,
      "grad_norm": 0.5945832006013596,
      "learning_rate": 0.003,
      "loss": 4.2875,
      "step": 1686
    },
    {
      "epoch": 0.01687,
      "grad_norm": 0.5137142109366883,
      "learning_rate": 0.003,
      "loss": 4.2801,
      "step": 1687
    },
    {
      "epoch": 0.01688,
      "grad_norm": 0.5492618386501558,
      "learning_rate": 0.003,
      "loss": 4.2695,
      "step": 1688
    },
    {
      "epoch": 0.01689,
      "grad_norm": 0.5905725806017027,
      "learning_rate": 0.003,
      "loss": 4.2834,
      "step": 1689
    },
    {
      "epoch": 0.0169,
      "grad_norm": 0.6017735937539882,
      "learning_rate": 0.003,
      "loss": 4.2739,
      "step": 1690
    },
    {
      "epoch": 0.01691,
      "grad_norm": 0.5609944056824007,
      "learning_rate": 0.003,
      "loss": 4.2833,
      "step": 1691
    },
    {
      "epoch": 0.01692,
      "grad_norm": 0.640275407175671,
      "learning_rate": 0.003,
      "loss": 4.2898,
      "step": 1692
    },
    {
      "epoch": 0.01693,
      "grad_norm": 0.6644174640253917,
      "learning_rate": 0.003,
      "loss": 4.3205,
      "step": 1693
    },
    {
      "epoch": 0.01694,
      "grad_norm": 0.6415163103575349,
      "learning_rate": 0.003,
      "loss": 4.2685,
      "step": 1694
    },
    {
      "epoch": 0.01695,
      "grad_norm": 0.5843260495529953,
      "learning_rate": 0.003,
      "loss": 4.2981,
      "step": 1695
    },
    {
      "epoch": 0.01696,
      "grad_norm": 0.6242095105916499,
      "learning_rate": 0.003,
      "loss": 4.2919,
      "step": 1696
    },
    {
      "epoch": 0.01697,
      "grad_norm": 0.6266335557548692,
      "learning_rate": 0.003,
      "loss": 4.2683,
      "step": 1697
    },
    {
      "epoch": 0.01698,
      "grad_norm": 0.5465027100918087,
      "learning_rate": 0.003,
      "loss": 4.2606,
      "step": 1698
    },
    {
      "epoch": 0.01699,
      "grad_norm": 0.4757955199873524,
      "learning_rate": 0.003,
      "loss": 4.2671,
      "step": 1699
    },
    {
      "epoch": 0.017,
      "grad_norm": 0.5193155427226998,
      "learning_rate": 0.003,
      "loss": 4.2877,
      "step": 1700
    },
    {
      "epoch": 0.01701,
      "grad_norm": 0.5016711628234212,
      "learning_rate": 0.003,
      "loss": 4.2685,
      "step": 1701
    },
    {
      "epoch": 0.01702,
      "grad_norm": 0.5283542213739725,
      "learning_rate": 0.003,
      "loss": 4.2934,
      "step": 1702
    },
    {
      "epoch": 0.01703,
      "grad_norm": 0.5693915553392532,
      "learning_rate": 0.003,
      "loss": 4.2816,
      "step": 1703
    },
    {
      "epoch": 0.01704,
      "grad_norm": 0.5510842689706329,
      "learning_rate": 0.003,
      "loss": 4.2655,
      "step": 1704
    },
    {
      "epoch": 0.01705,
      "grad_norm": 0.5433284015620915,
      "learning_rate": 0.003,
      "loss": 4.2713,
      "step": 1705
    },
    {
      "epoch": 0.01706,
      "grad_norm": 0.5361651744968902,
      "learning_rate": 0.003,
      "loss": 4.2666,
      "step": 1706
    },
    {
      "epoch": 0.01707,
      "grad_norm": 0.49807673030468796,
      "learning_rate": 0.003,
      "loss": 4.2867,
      "step": 1707
    },
    {
      "epoch": 0.01708,
      "grad_norm": 0.47060326864649304,
      "learning_rate": 0.003,
      "loss": 4.2615,
      "step": 1708
    },
    {
      "epoch": 0.01709,
      "grad_norm": 0.5858233174308028,
      "learning_rate": 0.003,
      "loss": 4.2895,
      "step": 1709
    },
    {
      "epoch": 0.0171,
      "grad_norm": 0.6958457968419427,
      "learning_rate": 0.003,
      "loss": 4.2423,
      "step": 1710
    },
    {
      "epoch": 0.01711,
      "grad_norm": 0.6480228055861614,
      "learning_rate": 0.003,
      "loss": 4.2726,
      "step": 1711
    },
    {
      "epoch": 0.01712,
      "grad_norm": 0.5133833526312795,
      "learning_rate": 0.003,
      "loss": 4.2785,
      "step": 1712
    },
    {
      "epoch": 0.01713,
      "grad_norm": 0.6473419774088255,
      "learning_rate": 0.003,
      "loss": 4.2578,
      "step": 1713
    },
    {
      "epoch": 0.01714,
      "grad_norm": 0.6885593409140894,
      "learning_rate": 0.003,
      "loss": 4.3021,
      "step": 1714
    },
    {
      "epoch": 0.01715,
      "grad_norm": 0.7110176805262337,
      "learning_rate": 0.003,
      "loss": 4.3086,
      "step": 1715
    },
    {
      "epoch": 0.01716,
      "grad_norm": 0.6383663373581888,
      "learning_rate": 0.003,
      "loss": 4.284,
      "step": 1716
    },
    {
      "epoch": 0.01717,
      "grad_norm": 0.5609012358245258,
      "learning_rate": 0.003,
      "loss": 4.3151,
      "step": 1717
    },
    {
      "epoch": 0.01718,
      "grad_norm": 0.4893869305325425,
      "learning_rate": 0.003,
      "loss": 4.279,
      "step": 1718
    },
    {
      "epoch": 0.01719,
      "grad_norm": 0.4683591562635787,
      "learning_rate": 0.003,
      "loss": 4.3002,
      "step": 1719
    },
    {
      "epoch": 0.0172,
      "grad_norm": 0.43007495593928735,
      "learning_rate": 0.003,
      "loss": 4.2576,
      "step": 1720
    },
    {
      "epoch": 0.01721,
      "grad_norm": 0.4793855122225898,
      "learning_rate": 0.003,
      "loss": 4.2731,
      "step": 1721
    },
    {
      "epoch": 0.01722,
      "grad_norm": 0.5257529092437175,
      "learning_rate": 0.003,
      "loss": 4.2675,
      "step": 1722
    },
    {
      "epoch": 0.01723,
      "grad_norm": 0.6135435997236554,
      "learning_rate": 0.003,
      "loss": 4.2811,
      "step": 1723
    },
    {
      "epoch": 0.01724,
      "grad_norm": 0.6429236565141022,
      "learning_rate": 0.003,
      "loss": 4.2853,
      "step": 1724
    },
    {
      "epoch": 0.01725,
      "grad_norm": 0.6238007075890811,
      "learning_rate": 0.003,
      "loss": 4.27,
      "step": 1725
    },
    {
      "epoch": 0.01726,
      "grad_norm": 0.5743130238790891,
      "learning_rate": 0.003,
      "loss": 4.2614,
      "step": 1726
    },
    {
      "epoch": 0.01727,
      "grad_norm": 0.5512027746636241,
      "learning_rate": 0.003,
      "loss": 4.266,
      "step": 1727
    },
    {
      "epoch": 0.01728,
      "grad_norm": 0.5565544152920284,
      "learning_rate": 0.003,
      "loss": 4.2687,
      "step": 1728
    },
    {
      "epoch": 0.01729,
      "grad_norm": 0.6053220105417224,
      "learning_rate": 0.003,
      "loss": 4.2796,
      "step": 1729
    },
    {
      "epoch": 0.0173,
      "grad_norm": 0.7004864603830978,
      "learning_rate": 0.003,
      "loss": 4.256,
      "step": 1730
    },
    {
      "epoch": 0.01731,
      "grad_norm": 0.6677625209808086,
      "learning_rate": 0.003,
      "loss": 4.2797,
      "step": 1731
    },
    {
      "epoch": 0.01732,
      "grad_norm": 0.6458934000853458,
      "learning_rate": 0.003,
      "loss": 4.2877,
      "step": 1732
    },
    {
      "epoch": 0.01733,
      "grad_norm": 0.7455706085018527,
      "learning_rate": 0.003,
      "loss": 4.2732,
      "step": 1733
    },
    {
      "epoch": 0.01734,
      "grad_norm": 0.6265592537450486,
      "learning_rate": 0.003,
      "loss": 4.2823,
      "step": 1734
    },
    {
      "epoch": 0.01735,
      "grad_norm": 0.5392857728589878,
      "learning_rate": 0.003,
      "loss": 4.2489,
      "step": 1735
    },
    {
      "epoch": 0.01736,
      "grad_norm": 0.5707588496610313,
      "learning_rate": 0.003,
      "loss": 4.2503,
      "step": 1736
    },
    {
      "epoch": 0.01737,
      "grad_norm": 0.5560424700678082,
      "learning_rate": 0.003,
      "loss": 4.2812,
      "step": 1737
    },
    {
      "epoch": 0.01738,
      "grad_norm": 0.5730683002180033,
      "learning_rate": 0.003,
      "loss": 4.2686,
      "step": 1738
    },
    {
      "epoch": 0.01739,
      "grad_norm": 0.6224581195400443,
      "learning_rate": 0.003,
      "loss": 4.2874,
      "step": 1739
    },
    {
      "epoch": 0.0174,
      "grad_norm": 0.6572820925374313,
      "learning_rate": 0.003,
      "loss": 4.2884,
      "step": 1740
    },
    {
      "epoch": 0.01741,
      "grad_norm": 0.7158790913353127,
      "learning_rate": 0.003,
      "loss": 4.3015,
      "step": 1741
    },
    {
      "epoch": 0.01742,
      "grad_norm": 0.7666622653552904,
      "learning_rate": 0.003,
      "loss": 4.2911,
      "step": 1742
    },
    {
      "epoch": 0.01743,
      "grad_norm": 0.7426449847041513,
      "learning_rate": 0.003,
      "loss": 4.3102,
      "step": 1743
    },
    {
      "epoch": 0.01744,
      "grad_norm": 0.7275834660740842,
      "learning_rate": 0.003,
      "loss": 4.2916,
      "step": 1744
    },
    {
      "epoch": 0.01745,
      "grad_norm": 0.6002910482818848,
      "learning_rate": 0.003,
      "loss": 4.2539,
      "step": 1745
    },
    {
      "epoch": 0.01746,
      "grad_norm": 0.5594118226409661,
      "learning_rate": 0.003,
      "loss": 4.2741,
      "step": 1746
    },
    {
      "epoch": 0.01747,
      "grad_norm": 0.5290431452891758,
      "learning_rate": 0.003,
      "loss": 4.2708,
      "step": 1747
    },
    {
      "epoch": 0.01748,
      "grad_norm": 0.5055142153030036,
      "learning_rate": 0.003,
      "loss": 4.263,
      "step": 1748
    },
    {
      "epoch": 0.01749,
      "grad_norm": 0.5966872692997083,
      "learning_rate": 0.003,
      "loss": 4.2613,
      "step": 1749
    },
    {
      "epoch": 0.0175,
      "grad_norm": 0.5900848154260279,
      "learning_rate": 0.003,
      "loss": 4.2801,
      "step": 1750
    },
    {
      "epoch": 0.01751,
      "grad_norm": 0.6245345037465497,
      "learning_rate": 0.003,
      "loss": 4.2737,
      "step": 1751
    },
    {
      "epoch": 0.01752,
      "grad_norm": 0.5951652358284407,
      "learning_rate": 0.003,
      "loss": 4.3127,
      "step": 1752
    },
    {
      "epoch": 0.01753,
      "grad_norm": 0.5691433797032257,
      "learning_rate": 0.003,
      "loss": 4.2613,
      "step": 1753
    },
    {
      "epoch": 0.01754,
      "grad_norm": 0.5243977651759597,
      "learning_rate": 0.003,
      "loss": 4.2904,
      "step": 1754
    },
    {
      "epoch": 0.01755,
      "grad_norm": 0.5047009634705759,
      "learning_rate": 0.003,
      "loss": 4.23,
      "step": 1755
    },
    {
      "epoch": 0.01756,
      "grad_norm": 0.5542549196861234,
      "learning_rate": 0.003,
      "loss": 4.3093,
      "step": 1756
    },
    {
      "epoch": 0.01757,
      "grad_norm": 0.5493798919225148,
      "learning_rate": 0.003,
      "loss": 4.2887,
      "step": 1757
    },
    {
      "epoch": 0.01758,
      "grad_norm": 0.5731223181742938,
      "learning_rate": 0.003,
      "loss": 4.2949,
      "step": 1758
    },
    {
      "epoch": 0.01759,
      "grad_norm": 0.5640596577781694,
      "learning_rate": 0.003,
      "loss": 4.2619,
      "step": 1759
    },
    {
      "epoch": 0.0176,
      "grad_norm": 0.49712782984125864,
      "learning_rate": 0.003,
      "loss": 4.2816,
      "step": 1760
    },
    {
      "epoch": 0.01761,
      "grad_norm": 0.49486559076430775,
      "learning_rate": 0.003,
      "loss": 4.2648,
      "step": 1761
    },
    {
      "epoch": 0.01762,
      "grad_norm": 0.5110004857223106,
      "learning_rate": 0.003,
      "loss": 4.2854,
      "step": 1762
    },
    {
      "epoch": 0.01763,
      "grad_norm": 0.5633857856922079,
      "learning_rate": 0.003,
      "loss": 4.2792,
      "step": 1763
    },
    {
      "epoch": 0.01764,
      "grad_norm": 0.7662844615211925,
      "learning_rate": 0.003,
      "loss": 4.255,
      "step": 1764
    },
    {
      "epoch": 0.01765,
      "grad_norm": 0.8231706617421852,
      "learning_rate": 0.003,
      "loss": 4.2957,
      "step": 1765
    },
    {
      "epoch": 0.01766,
      "grad_norm": 0.6112884625612803,
      "learning_rate": 0.003,
      "loss": 4.2946,
      "step": 1766
    },
    {
      "epoch": 0.01767,
      "grad_norm": 0.5304115562781407,
      "learning_rate": 0.003,
      "loss": 4.2562,
      "step": 1767
    },
    {
      "epoch": 0.01768,
      "grad_norm": 0.5693392116057188,
      "learning_rate": 0.003,
      "loss": 4.2663,
      "step": 1768
    },
    {
      "epoch": 0.01769,
      "grad_norm": 0.5007419067309367,
      "learning_rate": 0.003,
      "loss": 4.2627,
      "step": 1769
    },
    {
      "epoch": 0.0177,
      "grad_norm": 0.5337006106644646,
      "learning_rate": 0.003,
      "loss": 4.2481,
      "step": 1770
    },
    {
      "epoch": 0.01771,
      "grad_norm": 0.4993647653872291,
      "learning_rate": 0.003,
      "loss": 4.278,
      "step": 1771
    },
    {
      "epoch": 0.01772,
      "grad_norm": 0.49378669154496685,
      "learning_rate": 0.003,
      "loss": 4.2492,
      "step": 1772
    },
    {
      "epoch": 0.01773,
      "grad_norm": 0.49668963256381116,
      "learning_rate": 0.003,
      "loss": 4.2435,
      "step": 1773
    },
    {
      "epoch": 0.01774,
      "grad_norm": 0.4890453055402567,
      "learning_rate": 0.003,
      "loss": 4.2647,
      "step": 1774
    },
    {
      "epoch": 0.01775,
      "grad_norm": 0.4981133550102386,
      "learning_rate": 0.003,
      "loss": 4.2509,
      "step": 1775
    },
    {
      "epoch": 0.01776,
      "grad_norm": 0.5331679722135182,
      "learning_rate": 0.003,
      "loss": 4.2445,
      "step": 1776
    },
    {
      "epoch": 0.01777,
      "grad_norm": 0.5614293963448983,
      "learning_rate": 0.003,
      "loss": 4.269,
      "step": 1777
    },
    {
      "epoch": 0.01778,
      "grad_norm": 0.5337530525849652,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 1778
    },
    {
      "epoch": 0.01779,
      "grad_norm": 0.5219382960786771,
      "learning_rate": 0.003,
      "loss": 4.277,
      "step": 1779
    },
    {
      "epoch": 0.0178,
      "grad_norm": 0.4801993403704907,
      "learning_rate": 0.003,
      "loss": 4.2462,
      "step": 1780
    },
    {
      "epoch": 0.01781,
      "grad_norm": 0.5353966732851616,
      "learning_rate": 0.003,
      "loss": 4.2633,
      "step": 1781
    },
    {
      "epoch": 0.01782,
      "grad_norm": 0.5707424190875742,
      "learning_rate": 0.003,
      "loss": 4.2884,
      "step": 1782
    },
    {
      "epoch": 0.01783,
      "grad_norm": 0.6083204008367944,
      "learning_rate": 0.003,
      "loss": 4.269,
      "step": 1783
    },
    {
      "epoch": 0.01784,
      "grad_norm": 0.5414123353142472,
      "learning_rate": 0.003,
      "loss": 4.2666,
      "step": 1784
    },
    {
      "epoch": 0.01785,
      "grad_norm": 0.6099552330414602,
      "learning_rate": 0.003,
      "loss": 4.2322,
      "step": 1785
    },
    {
      "epoch": 0.01786,
      "grad_norm": 0.6938449080529763,
      "learning_rate": 0.003,
      "loss": 4.2624,
      "step": 1786
    },
    {
      "epoch": 0.01787,
      "grad_norm": 0.7097606505143458,
      "learning_rate": 0.003,
      "loss": 4.2787,
      "step": 1787
    },
    {
      "epoch": 0.01788,
      "grad_norm": 0.7987231782657276,
      "learning_rate": 0.003,
      "loss": 4.29,
      "step": 1788
    },
    {
      "epoch": 0.01789,
      "grad_norm": 0.9170384713763422,
      "learning_rate": 0.003,
      "loss": 4.2752,
      "step": 1789
    },
    {
      "epoch": 0.0179,
      "grad_norm": 0.8912339815181158,
      "learning_rate": 0.003,
      "loss": 4.2973,
      "step": 1790
    },
    {
      "epoch": 0.01791,
      "grad_norm": 0.7681052193269913,
      "learning_rate": 0.003,
      "loss": 4.3056,
      "step": 1791
    },
    {
      "epoch": 0.01792,
      "grad_norm": 0.6429477629622415,
      "learning_rate": 0.003,
      "loss": 4.3028,
      "step": 1792
    },
    {
      "epoch": 0.01793,
      "grad_norm": 0.7186534811578004,
      "learning_rate": 0.003,
      "loss": 4.2729,
      "step": 1793
    },
    {
      "epoch": 0.01794,
      "grad_norm": 0.5765860920967305,
      "learning_rate": 0.003,
      "loss": 4.2799,
      "step": 1794
    },
    {
      "epoch": 0.01795,
      "grad_norm": 0.5000657697525336,
      "learning_rate": 0.003,
      "loss": 4.2762,
      "step": 1795
    },
    {
      "epoch": 0.01796,
      "grad_norm": 0.5320047446020336,
      "learning_rate": 0.003,
      "loss": 4.2904,
      "step": 1796
    },
    {
      "epoch": 0.01797,
      "grad_norm": 0.4504365845865774,
      "learning_rate": 0.003,
      "loss": 4.2509,
      "step": 1797
    },
    {
      "epoch": 0.01798,
      "grad_norm": 0.40462213723419516,
      "learning_rate": 0.003,
      "loss": 4.2636,
      "step": 1798
    },
    {
      "epoch": 0.01799,
      "grad_norm": 0.35282082316813834,
      "learning_rate": 0.003,
      "loss": 4.2763,
      "step": 1799
    },
    {
      "epoch": 0.018,
      "grad_norm": 0.3411687376041599,
      "learning_rate": 0.003,
      "loss": 4.2632,
      "step": 1800
    },
    {
      "epoch": 0.01801,
      "grad_norm": 0.3497200911452386,
      "learning_rate": 0.003,
      "loss": 4.2568,
      "step": 1801
    },
    {
      "epoch": 0.01802,
      "grad_norm": 0.35422809832207447,
      "learning_rate": 0.003,
      "loss": 4.2731,
      "step": 1802
    },
    {
      "epoch": 0.01803,
      "grad_norm": 0.3279859543333952,
      "learning_rate": 0.003,
      "loss": 4.277,
      "step": 1803
    },
    {
      "epoch": 0.01804,
      "grad_norm": 0.37278065596161997,
      "learning_rate": 0.003,
      "loss": 4.2479,
      "step": 1804
    },
    {
      "epoch": 0.01805,
      "grad_norm": 0.4167024120969314,
      "learning_rate": 0.003,
      "loss": 4.268,
      "step": 1805
    },
    {
      "epoch": 0.01806,
      "grad_norm": 0.45338026173808493,
      "learning_rate": 0.003,
      "loss": 4.228,
      "step": 1806
    },
    {
      "epoch": 0.01807,
      "grad_norm": 0.4492923467683076,
      "learning_rate": 0.003,
      "loss": 4.2649,
      "step": 1807
    },
    {
      "epoch": 0.01808,
      "grad_norm": 0.4519532920743694,
      "learning_rate": 0.003,
      "loss": 4.2552,
      "step": 1808
    },
    {
      "epoch": 0.01809,
      "grad_norm": 0.4794685836261005,
      "learning_rate": 0.003,
      "loss": 4.2751,
      "step": 1809
    },
    {
      "epoch": 0.0181,
      "grad_norm": 0.6301644557370442,
      "learning_rate": 0.003,
      "loss": 4.275,
      "step": 1810
    },
    {
      "epoch": 0.01811,
      "grad_norm": 0.8128009946323582,
      "learning_rate": 0.003,
      "loss": 4.2718,
      "step": 1811
    },
    {
      "epoch": 0.01812,
      "grad_norm": 0.8427115921852621,
      "learning_rate": 0.003,
      "loss": 4.2615,
      "step": 1812
    },
    {
      "epoch": 0.01813,
      "grad_norm": 0.7199083817416421,
      "learning_rate": 0.003,
      "loss": 4.3052,
      "step": 1813
    },
    {
      "epoch": 0.01814,
      "grad_norm": 0.7440856146842654,
      "learning_rate": 0.003,
      "loss": 4.2773,
      "step": 1814
    },
    {
      "epoch": 0.01815,
      "grad_norm": 0.6604668614903264,
      "learning_rate": 0.003,
      "loss": 4.2603,
      "step": 1815
    },
    {
      "epoch": 0.01816,
      "grad_norm": 0.7213183194060501,
      "learning_rate": 0.003,
      "loss": 4.289,
      "step": 1816
    },
    {
      "epoch": 0.01817,
      "grad_norm": 0.707160301614863,
      "learning_rate": 0.003,
      "loss": 4.2802,
      "step": 1817
    },
    {
      "epoch": 0.01818,
      "grad_norm": 0.636608110327377,
      "learning_rate": 0.003,
      "loss": 4.2785,
      "step": 1818
    },
    {
      "epoch": 0.01819,
      "grad_norm": 0.5804906977024753,
      "learning_rate": 0.003,
      "loss": 4.2581,
      "step": 1819
    },
    {
      "epoch": 0.0182,
      "grad_norm": 0.5372919614803818,
      "learning_rate": 0.003,
      "loss": 4.2768,
      "step": 1820
    },
    {
      "epoch": 0.01821,
      "grad_norm": 0.5873367244339422,
      "learning_rate": 0.003,
      "loss": 4.2652,
      "step": 1821
    },
    {
      "epoch": 0.01822,
      "grad_norm": 0.6367155622372229,
      "learning_rate": 0.003,
      "loss": 4.278,
      "step": 1822
    },
    {
      "epoch": 0.01823,
      "grad_norm": 0.7098801106756836,
      "learning_rate": 0.003,
      "loss": 4.2989,
      "step": 1823
    },
    {
      "epoch": 0.01824,
      "grad_norm": 0.6249291972074501,
      "learning_rate": 0.003,
      "loss": 4.2755,
      "step": 1824
    },
    {
      "epoch": 0.01825,
      "grad_norm": 0.5242106659607212,
      "learning_rate": 0.003,
      "loss": 4.2762,
      "step": 1825
    },
    {
      "epoch": 0.01826,
      "grad_norm": 0.644099441571583,
      "learning_rate": 0.003,
      "loss": 4.2915,
      "step": 1826
    },
    {
      "epoch": 0.01827,
      "grad_norm": 0.6224752218569206,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 1827
    },
    {
      "epoch": 0.01828,
      "grad_norm": 0.543186641981227,
      "learning_rate": 0.003,
      "loss": 4.2695,
      "step": 1828
    },
    {
      "epoch": 0.01829,
      "grad_norm": 0.6845506683310987,
      "learning_rate": 0.003,
      "loss": 4.2898,
      "step": 1829
    },
    {
      "epoch": 0.0183,
      "grad_norm": 0.879601600440537,
      "learning_rate": 0.003,
      "loss": 4.2972,
      "step": 1830
    },
    {
      "epoch": 0.01831,
      "grad_norm": 0.9844682114288815,
      "learning_rate": 0.003,
      "loss": 4.2992,
      "step": 1831
    },
    {
      "epoch": 0.01832,
      "grad_norm": 0.8821724273705098,
      "learning_rate": 0.003,
      "loss": 4.2978,
      "step": 1832
    },
    {
      "epoch": 0.01833,
      "grad_norm": 0.7330000924703792,
      "learning_rate": 0.003,
      "loss": 4.3267,
      "step": 1833
    },
    {
      "epoch": 0.01834,
      "grad_norm": 0.68997205114043,
      "learning_rate": 0.003,
      "loss": 4.2918,
      "step": 1834
    },
    {
      "epoch": 0.01835,
      "grad_norm": 0.7006645684897048,
      "learning_rate": 0.003,
      "loss": 4.2921,
      "step": 1835
    },
    {
      "epoch": 0.01836,
      "grad_norm": 0.7370828132611139,
      "learning_rate": 0.003,
      "loss": 4.2868,
      "step": 1836
    },
    {
      "epoch": 0.01837,
      "grad_norm": 0.8137814608022381,
      "learning_rate": 0.003,
      "loss": 4.2934,
      "step": 1837
    },
    {
      "epoch": 0.01838,
      "grad_norm": 0.6630708544495384,
      "learning_rate": 0.003,
      "loss": 4.2935,
      "step": 1838
    },
    {
      "epoch": 0.01839,
      "grad_norm": 0.6770506893324072,
      "learning_rate": 0.003,
      "loss": 4.3163,
      "step": 1839
    },
    {
      "epoch": 0.0184,
      "grad_norm": 0.772243829580562,
      "learning_rate": 0.003,
      "loss": 4.315,
      "step": 1840
    },
    {
      "epoch": 0.01841,
      "grad_norm": 0.7260652894274943,
      "learning_rate": 0.003,
      "loss": 4.2507,
      "step": 1841
    },
    {
      "epoch": 0.01842,
      "grad_norm": 0.7455618963662838,
      "learning_rate": 0.003,
      "loss": 4.2889,
      "step": 1842
    },
    {
      "epoch": 0.01843,
      "grad_norm": 0.5629263660723788,
      "learning_rate": 0.003,
      "loss": 4.2632,
      "step": 1843
    },
    {
      "epoch": 0.01844,
      "grad_norm": 0.44782374021160304,
      "learning_rate": 0.003,
      "loss": 4.2542,
      "step": 1844
    },
    {
      "epoch": 0.01845,
      "grad_norm": 0.44338335058672285,
      "learning_rate": 0.003,
      "loss": 4.2569,
      "step": 1845
    },
    {
      "epoch": 0.01846,
      "grad_norm": 0.3533576160559802,
      "learning_rate": 0.003,
      "loss": 4.2775,
      "step": 1846
    },
    {
      "epoch": 0.01847,
      "grad_norm": 0.36624369667968887,
      "learning_rate": 0.003,
      "loss": 4.2733,
      "step": 1847
    },
    {
      "epoch": 0.01848,
      "grad_norm": 0.3515311739049859,
      "learning_rate": 0.003,
      "loss": 4.2847,
      "step": 1848
    },
    {
      "epoch": 0.01849,
      "grad_norm": 0.31316963159896893,
      "learning_rate": 0.003,
      "loss": 4.2827,
      "step": 1849
    },
    {
      "epoch": 0.0185,
      "grad_norm": 0.3416820848274596,
      "learning_rate": 0.003,
      "loss": 4.2703,
      "step": 1850
    },
    {
      "epoch": 0.01851,
      "grad_norm": 0.3489689830102001,
      "learning_rate": 0.003,
      "loss": 4.2512,
      "step": 1851
    },
    {
      "epoch": 0.01852,
      "grad_norm": 0.3703418463232587,
      "learning_rate": 0.003,
      "loss": 4.2517,
      "step": 1852
    },
    {
      "epoch": 0.01853,
      "grad_norm": 0.41368285825954554,
      "learning_rate": 0.003,
      "loss": 4.2519,
      "step": 1853
    },
    {
      "epoch": 0.01854,
      "grad_norm": 0.43320899613747116,
      "learning_rate": 0.003,
      "loss": 4.2834,
      "step": 1854
    },
    {
      "epoch": 0.01855,
      "grad_norm": 0.5253237000651575,
      "learning_rate": 0.003,
      "loss": 4.2689,
      "step": 1855
    },
    {
      "epoch": 0.01856,
      "grad_norm": 0.5779002753152843,
      "learning_rate": 0.003,
      "loss": 4.2553,
      "step": 1856
    },
    {
      "epoch": 0.01857,
      "grad_norm": 0.593959858084494,
      "learning_rate": 0.003,
      "loss": 4.241,
      "step": 1857
    },
    {
      "epoch": 0.01858,
      "grad_norm": 0.5126648296135959,
      "learning_rate": 0.003,
      "loss": 4.2479,
      "step": 1858
    },
    {
      "epoch": 0.01859,
      "grad_norm": 0.45559130929525077,
      "learning_rate": 0.003,
      "loss": 4.2412,
      "step": 1859
    },
    {
      "epoch": 0.0186,
      "grad_norm": 0.4806055065781757,
      "learning_rate": 0.003,
      "loss": 4.2254,
      "step": 1860
    },
    {
      "epoch": 0.01861,
      "grad_norm": 0.48294706675854066,
      "learning_rate": 0.003,
      "loss": 4.2694,
      "step": 1861
    },
    {
      "epoch": 0.01862,
      "grad_norm": 0.4664947919884159,
      "learning_rate": 0.003,
      "loss": 4.2335,
      "step": 1862
    },
    {
      "epoch": 0.01863,
      "grad_norm": 0.4819198640587951,
      "learning_rate": 0.003,
      "loss": 4.2428,
      "step": 1863
    },
    {
      "epoch": 0.01864,
      "grad_norm": 0.5946115673610074,
      "learning_rate": 0.003,
      "loss": 4.2797,
      "step": 1864
    },
    {
      "epoch": 0.01865,
      "grad_norm": 0.7290398384671329,
      "learning_rate": 0.003,
      "loss": 4.2637,
      "step": 1865
    },
    {
      "epoch": 0.01866,
      "grad_norm": 0.8160446708452547,
      "learning_rate": 0.003,
      "loss": 4.2688,
      "step": 1866
    },
    {
      "epoch": 0.01867,
      "grad_norm": 0.7596737913536388,
      "learning_rate": 0.003,
      "loss": 4.2479,
      "step": 1867
    },
    {
      "epoch": 0.01868,
      "grad_norm": 0.7512897467429733,
      "learning_rate": 0.003,
      "loss": 4.265,
      "step": 1868
    },
    {
      "epoch": 0.01869,
      "grad_norm": 0.899865336798572,
      "learning_rate": 0.003,
      "loss": 4.2896,
      "step": 1869
    },
    {
      "epoch": 0.0187,
      "grad_norm": 0.8087963711485776,
      "learning_rate": 0.003,
      "loss": 4.2851,
      "step": 1870
    },
    {
      "epoch": 0.01871,
      "grad_norm": 0.6481673959520611,
      "learning_rate": 0.003,
      "loss": 4.2524,
      "step": 1871
    },
    {
      "epoch": 0.01872,
      "grad_norm": 0.5923778344000376,
      "learning_rate": 0.003,
      "loss": 4.2969,
      "step": 1872
    },
    {
      "epoch": 0.01873,
      "grad_norm": 0.5580453248479986,
      "learning_rate": 0.003,
      "loss": 4.2558,
      "step": 1873
    },
    {
      "epoch": 0.01874,
      "grad_norm": 0.6129883764486191,
      "learning_rate": 0.003,
      "loss": 4.2619,
      "step": 1874
    },
    {
      "epoch": 0.01875,
      "grad_norm": 0.5660814639175555,
      "learning_rate": 0.003,
      "loss": 4.2767,
      "step": 1875
    },
    {
      "epoch": 0.01876,
      "grad_norm": 0.5484550288366968,
      "learning_rate": 0.003,
      "loss": 4.2804,
      "step": 1876
    },
    {
      "epoch": 0.01877,
      "grad_norm": 0.5413620281670742,
      "learning_rate": 0.003,
      "loss": 4.2852,
      "step": 1877
    },
    {
      "epoch": 0.01878,
      "grad_norm": 0.5151059596185084,
      "learning_rate": 0.003,
      "loss": 4.2658,
      "step": 1878
    },
    {
      "epoch": 0.01879,
      "grad_norm": 0.5108820927917921,
      "learning_rate": 0.003,
      "loss": 4.2428,
      "step": 1879
    },
    {
      "epoch": 0.0188,
      "grad_norm": 0.480687658081291,
      "learning_rate": 0.003,
      "loss": 4.253,
      "step": 1880
    },
    {
      "epoch": 0.01881,
      "grad_norm": 0.4006710963490709,
      "learning_rate": 0.003,
      "loss": 4.2442,
      "step": 1881
    },
    {
      "epoch": 0.01882,
      "grad_norm": 0.4041555926045392,
      "learning_rate": 0.003,
      "loss": 4.2713,
      "step": 1882
    },
    {
      "epoch": 0.01883,
      "grad_norm": 0.35889370988759256,
      "learning_rate": 0.003,
      "loss": 4.2426,
      "step": 1883
    },
    {
      "epoch": 0.01884,
      "grad_norm": 0.3548857794793813,
      "learning_rate": 0.003,
      "loss": 4.2417,
      "step": 1884
    },
    {
      "epoch": 0.01885,
      "grad_norm": 0.35029061193490524,
      "learning_rate": 0.003,
      "loss": 4.2179,
      "step": 1885
    },
    {
      "epoch": 0.01886,
      "grad_norm": 0.44501447232013214,
      "learning_rate": 0.003,
      "loss": 4.2578,
      "step": 1886
    },
    {
      "epoch": 0.01887,
      "grad_norm": 0.5229549741250542,
      "learning_rate": 0.003,
      "loss": 4.2541,
      "step": 1887
    },
    {
      "epoch": 0.01888,
      "grad_norm": 0.6745619001945896,
      "learning_rate": 0.003,
      "loss": 4.2454,
      "step": 1888
    },
    {
      "epoch": 0.01889,
      "grad_norm": 0.6923119713487843,
      "learning_rate": 0.003,
      "loss": 4.265,
      "step": 1889
    },
    {
      "epoch": 0.0189,
      "grad_norm": 0.58679702829643,
      "learning_rate": 0.003,
      "loss": 4.2409,
      "step": 1890
    },
    {
      "epoch": 0.01891,
      "grad_norm": 0.646800864954939,
      "learning_rate": 0.003,
      "loss": 4.2919,
      "step": 1891
    },
    {
      "epoch": 0.01892,
      "grad_norm": 0.6502623001706804,
      "learning_rate": 0.003,
      "loss": 4.2707,
      "step": 1892
    },
    {
      "epoch": 0.01893,
      "grad_norm": 0.5615534003702374,
      "learning_rate": 0.003,
      "loss": 4.2469,
      "step": 1893
    },
    {
      "epoch": 0.01894,
      "grad_norm": 0.5991987877315249,
      "learning_rate": 0.003,
      "loss": 4.254,
      "step": 1894
    },
    {
      "epoch": 0.01895,
      "grad_norm": 0.5685602484818931,
      "learning_rate": 0.003,
      "loss": 4.2409,
      "step": 1895
    },
    {
      "epoch": 0.01896,
      "grad_norm": 0.5273740611308925,
      "learning_rate": 0.003,
      "loss": 4.2621,
      "step": 1896
    },
    {
      "epoch": 0.01897,
      "grad_norm": 0.4634723554903588,
      "learning_rate": 0.003,
      "loss": 4.2587,
      "step": 1897
    },
    {
      "epoch": 0.01898,
      "grad_norm": 0.49012578983884775,
      "learning_rate": 0.003,
      "loss": 4.2271,
      "step": 1898
    },
    {
      "epoch": 0.01899,
      "grad_norm": 0.5391967127125681,
      "learning_rate": 0.003,
      "loss": 4.2462,
      "step": 1899
    },
    {
      "epoch": 0.019,
      "grad_norm": 0.5314902668241436,
      "learning_rate": 0.003,
      "loss": 4.2384,
      "step": 1900
    },
    {
      "epoch": 0.01901,
      "grad_norm": 0.5694638994637579,
      "learning_rate": 0.003,
      "loss": 4.2871,
      "step": 1901
    },
    {
      "epoch": 0.01902,
      "grad_norm": 0.5590323874513298,
      "learning_rate": 0.003,
      "loss": 4.2641,
      "step": 1902
    },
    {
      "epoch": 0.01903,
      "grad_norm": 0.5475545528889447,
      "learning_rate": 0.003,
      "loss": 4.2196,
      "step": 1903
    },
    {
      "epoch": 0.01904,
      "grad_norm": 0.6218453695681301,
      "learning_rate": 0.003,
      "loss": 4.2791,
      "step": 1904
    },
    {
      "epoch": 0.01905,
      "grad_norm": 0.591162548550656,
      "learning_rate": 0.003,
      "loss": 4.2534,
      "step": 1905
    },
    {
      "epoch": 0.01906,
      "grad_norm": 0.5128889662513827,
      "learning_rate": 0.003,
      "loss": 4.2676,
      "step": 1906
    },
    {
      "epoch": 0.01907,
      "grad_norm": 0.5047288261347648,
      "learning_rate": 0.003,
      "loss": 4.2599,
      "step": 1907
    },
    {
      "epoch": 0.01908,
      "grad_norm": 0.4667283203300201,
      "learning_rate": 0.003,
      "loss": 4.2726,
      "step": 1908
    },
    {
      "epoch": 0.01909,
      "grad_norm": 0.5022149031562609,
      "learning_rate": 0.003,
      "loss": 4.2382,
      "step": 1909
    },
    {
      "epoch": 0.0191,
      "grad_norm": 0.5432392826551734,
      "learning_rate": 0.003,
      "loss": 4.2518,
      "step": 1910
    },
    {
      "epoch": 0.01911,
      "grad_norm": 0.49523295523680083,
      "learning_rate": 0.003,
      "loss": 4.2241,
      "step": 1911
    },
    {
      "epoch": 0.01912,
      "grad_norm": 0.5206775073204631,
      "learning_rate": 0.003,
      "loss": 4.2363,
      "step": 1912
    },
    {
      "epoch": 0.01913,
      "grad_norm": 0.523325426364381,
      "learning_rate": 0.003,
      "loss": 4.2544,
      "step": 1913
    },
    {
      "epoch": 0.01914,
      "grad_norm": 0.5843680788676122,
      "learning_rate": 0.003,
      "loss": 4.2743,
      "step": 1914
    },
    {
      "epoch": 0.01915,
      "grad_norm": 0.6905336390991205,
      "learning_rate": 0.003,
      "loss": 4.2632,
      "step": 1915
    },
    {
      "epoch": 0.01916,
      "grad_norm": 0.8314031188331161,
      "learning_rate": 0.003,
      "loss": 4.2512,
      "step": 1916
    },
    {
      "epoch": 0.01917,
      "grad_norm": 0.9750007322129252,
      "learning_rate": 0.003,
      "loss": 4.2904,
      "step": 1917
    },
    {
      "epoch": 0.01918,
      "grad_norm": 0.9797925708116749,
      "learning_rate": 0.003,
      "loss": 4.2842,
      "step": 1918
    },
    {
      "epoch": 0.01919,
      "grad_norm": 0.7747220152275403,
      "learning_rate": 0.003,
      "loss": 4.2834,
      "step": 1919
    },
    {
      "epoch": 0.0192,
      "grad_norm": 0.7189857079350012,
      "learning_rate": 0.003,
      "loss": 4.2819,
      "step": 1920
    },
    {
      "epoch": 0.01921,
      "grad_norm": 0.7219748871978388,
      "learning_rate": 0.003,
      "loss": 4.2705,
      "step": 1921
    },
    {
      "epoch": 0.01922,
      "grad_norm": 0.7146468956621638,
      "learning_rate": 0.003,
      "loss": 4.2862,
      "step": 1922
    },
    {
      "epoch": 0.01923,
      "grad_norm": 0.6697224209681402,
      "learning_rate": 0.003,
      "loss": 4.2923,
      "step": 1923
    },
    {
      "epoch": 0.01924,
      "grad_norm": 0.7063194554901026,
      "learning_rate": 0.003,
      "loss": 4.2924,
      "step": 1924
    },
    {
      "epoch": 0.01925,
      "grad_norm": 0.6608146496442958,
      "learning_rate": 0.003,
      "loss": 4.2782,
      "step": 1925
    },
    {
      "epoch": 0.01926,
      "grad_norm": 0.6058286924144092,
      "learning_rate": 0.003,
      "loss": 4.2871,
      "step": 1926
    },
    {
      "epoch": 0.01927,
      "grad_norm": 0.5582247211706641,
      "learning_rate": 0.003,
      "loss": 4.2731,
      "step": 1927
    },
    {
      "epoch": 0.01928,
      "grad_norm": 0.5497301592853885,
      "learning_rate": 0.003,
      "loss": 4.2438,
      "step": 1928
    },
    {
      "epoch": 0.01929,
      "grad_norm": 0.5697976403038864,
      "learning_rate": 0.003,
      "loss": 4.2661,
      "step": 1929
    },
    {
      "epoch": 0.0193,
      "grad_norm": 0.5669233105595727,
      "learning_rate": 0.003,
      "loss": 4.253,
      "step": 1930
    },
    {
      "epoch": 0.01931,
      "grad_norm": 0.4328335869920709,
      "learning_rate": 0.003,
      "loss": 4.2488,
      "step": 1931
    },
    {
      "epoch": 0.01932,
      "grad_norm": 0.37950505638182785,
      "learning_rate": 0.003,
      "loss": 4.2566,
      "step": 1932
    },
    {
      "epoch": 0.01933,
      "grad_norm": 0.3504964342922692,
      "learning_rate": 0.003,
      "loss": 4.2408,
      "step": 1933
    },
    {
      "epoch": 0.01934,
      "grad_norm": 0.3616114126450499,
      "learning_rate": 0.003,
      "loss": 4.2472,
      "step": 1934
    },
    {
      "epoch": 0.01935,
      "grad_norm": 0.3586196056492206,
      "learning_rate": 0.003,
      "loss": 4.259,
      "step": 1935
    },
    {
      "epoch": 0.01936,
      "grad_norm": 0.4036363099668179,
      "learning_rate": 0.003,
      "loss": 4.2404,
      "step": 1936
    },
    {
      "epoch": 0.01937,
      "grad_norm": 0.41909297435676146,
      "learning_rate": 0.003,
      "loss": 4.2527,
      "step": 1937
    },
    {
      "epoch": 0.01938,
      "grad_norm": 0.439340208584005,
      "learning_rate": 0.003,
      "loss": 4.2528,
      "step": 1938
    },
    {
      "epoch": 0.01939,
      "grad_norm": 0.4942036615177103,
      "learning_rate": 0.003,
      "loss": 4.2748,
      "step": 1939
    },
    {
      "epoch": 0.0194,
      "grad_norm": 0.5753579784669907,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 1940
    },
    {
      "epoch": 0.01941,
      "grad_norm": 0.7525206839134082,
      "learning_rate": 0.003,
      "loss": 4.2429,
      "step": 1941
    },
    {
      "epoch": 0.01942,
      "grad_norm": 0.8127296854745015,
      "learning_rate": 0.003,
      "loss": 4.2788,
      "step": 1942
    },
    {
      "epoch": 0.01943,
      "grad_norm": 0.7505954536075328,
      "learning_rate": 0.003,
      "loss": 4.2474,
      "step": 1943
    },
    {
      "epoch": 0.01944,
      "grad_norm": 0.6966879842895444,
      "learning_rate": 0.003,
      "loss": 4.292,
      "step": 1944
    },
    {
      "epoch": 0.01945,
      "grad_norm": 0.5317620511293283,
      "learning_rate": 0.003,
      "loss": 4.2513,
      "step": 1945
    },
    {
      "epoch": 0.01946,
      "grad_norm": 0.5396271588936495,
      "learning_rate": 0.003,
      "loss": 4.2483,
      "step": 1946
    },
    {
      "epoch": 0.01947,
      "grad_norm": 0.4953395916242608,
      "learning_rate": 0.003,
      "loss": 4.2539,
      "step": 1947
    },
    {
      "epoch": 0.01948,
      "grad_norm": 0.43095282528265144,
      "learning_rate": 0.003,
      "loss": 4.2724,
      "step": 1948
    },
    {
      "epoch": 0.01949,
      "grad_norm": 0.4581100416309599,
      "learning_rate": 0.003,
      "loss": 4.2647,
      "step": 1949
    },
    {
      "epoch": 0.0195,
      "grad_norm": 0.4138069104419874,
      "learning_rate": 0.003,
      "loss": 4.2509,
      "step": 1950
    },
    {
      "epoch": 0.01951,
      "grad_norm": 0.453383618119804,
      "learning_rate": 0.003,
      "loss": 4.2469,
      "step": 1951
    },
    {
      "epoch": 0.01952,
      "grad_norm": 0.5050716895331453,
      "learning_rate": 0.003,
      "loss": 4.2541,
      "step": 1952
    },
    {
      "epoch": 0.01953,
      "grad_norm": 0.5444826544170946,
      "learning_rate": 0.003,
      "loss": 4.2612,
      "step": 1953
    },
    {
      "epoch": 0.01954,
      "grad_norm": 0.5817368201949349,
      "learning_rate": 0.003,
      "loss": 4.2642,
      "step": 1954
    },
    {
      "epoch": 0.01955,
      "grad_norm": 0.5828427095685109,
      "learning_rate": 0.003,
      "loss": 4.257,
      "step": 1955
    },
    {
      "epoch": 0.01956,
      "grad_norm": 0.6033336279941961,
      "learning_rate": 0.003,
      "loss": 4.2559,
      "step": 1956
    },
    {
      "epoch": 0.01957,
      "grad_norm": 0.4961523479078479,
      "learning_rate": 0.003,
      "loss": 4.2179,
      "step": 1957
    },
    {
      "epoch": 0.01958,
      "grad_norm": 0.44186549075636594,
      "learning_rate": 0.003,
      "loss": 4.2115,
      "step": 1958
    },
    {
      "epoch": 0.01959,
      "grad_norm": 0.5087237373575997,
      "learning_rate": 0.003,
      "loss": 4.2394,
      "step": 1959
    },
    {
      "epoch": 0.0196,
      "grad_norm": 0.5262591797086301,
      "learning_rate": 0.003,
      "loss": 4.2322,
      "step": 1960
    },
    {
      "epoch": 0.01961,
      "grad_norm": 0.68395452085867,
      "learning_rate": 0.003,
      "loss": 4.2394,
      "step": 1961
    },
    {
      "epoch": 0.01962,
      "grad_norm": 0.8016832589931029,
      "learning_rate": 0.003,
      "loss": 4.2481,
      "step": 1962
    },
    {
      "epoch": 0.01963,
      "grad_norm": 0.75220900113814,
      "learning_rate": 0.003,
      "loss": 4.2388,
      "step": 1963
    },
    {
      "epoch": 0.01964,
      "grad_norm": 0.6357936481488424,
      "learning_rate": 0.003,
      "loss": 4.2611,
      "step": 1964
    },
    {
      "epoch": 0.01965,
      "grad_norm": 0.5451687729946599,
      "learning_rate": 0.003,
      "loss": 4.2493,
      "step": 1965
    },
    {
      "epoch": 0.01966,
      "grad_norm": 0.5793603889291398,
      "learning_rate": 0.003,
      "loss": 4.2538,
      "step": 1966
    },
    {
      "epoch": 0.01967,
      "grad_norm": 0.5157681292282557,
      "learning_rate": 0.003,
      "loss": 4.2253,
      "step": 1967
    },
    {
      "epoch": 0.01968,
      "grad_norm": 0.5440002063072249,
      "learning_rate": 0.003,
      "loss": 4.2502,
      "step": 1968
    },
    {
      "epoch": 0.01969,
      "grad_norm": 0.5300695855996388,
      "learning_rate": 0.003,
      "loss": 4.2446,
      "step": 1969
    },
    {
      "epoch": 0.0197,
      "grad_norm": 0.47595129333192576,
      "learning_rate": 0.003,
      "loss": 4.2426,
      "step": 1970
    },
    {
      "epoch": 0.01971,
      "grad_norm": 0.5285948280548484,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 1971
    },
    {
      "epoch": 0.01972,
      "grad_norm": 0.5935023814870325,
      "learning_rate": 0.003,
      "loss": 4.2369,
      "step": 1972
    },
    {
      "epoch": 0.01973,
      "grad_norm": 0.6689295429120606,
      "learning_rate": 0.003,
      "loss": 4.2228,
      "step": 1973
    },
    {
      "epoch": 0.01974,
      "grad_norm": 0.6805455931144389,
      "learning_rate": 0.003,
      "loss": 4.2627,
      "step": 1974
    },
    {
      "epoch": 0.01975,
      "grad_norm": 0.6851649356245609,
      "learning_rate": 0.003,
      "loss": 4.2574,
      "step": 1975
    },
    {
      "epoch": 0.01976,
      "grad_norm": 0.6359068486728622,
      "learning_rate": 0.003,
      "loss": 4.2445,
      "step": 1976
    },
    {
      "epoch": 0.01977,
      "grad_norm": 0.5752983961700786,
      "learning_rate": 0.003,
      "loss": 4.2775,
      "step": 1977
    },
    {
      "epoch": 0.01978,
      "grad_norm": 0.5885328424294697,
      "learning_rate": 0.003,
      "loss": 4.2773,
      "step": 1978
    },
    {
      "epoch": 0.01979,
      "grad_norm": 0.6114820249853535,
      "learning_rate": 0.003,
      "loss": 4.266,
      "step": 1979
    },
    {
      "epoch": 0.0198,
      "grad_norm": 0.6725244919020471,
      "learning_rate": 0.003,
      "loss": 4.2495,
      "step": 1980
    },
    {
      "epoch": 0.01981,
      "grad_norm": 0.6899410283984171,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 1981
    },
    {
      "epoch": 0.01982,
      "grad_norm": 0.7680678867968328,
      "learning_rate": 0.003,
      "loss": 4.2688,
      "step": 1982
    },
    {
      "epoch": 0.01983,
      "grad_norm": 0.8666192600294306,
      "learning_rate": 0.003,
      "loss": 4.2665,
      "step": 1983
    },
    {
      "epoch": 0.01984,
      "grad_norm": 0.7590492652886811,
      "learning_rate": 0.003,
      "loss": 4.2796,
      "step": 1984
    },
    {
      "epoch": 0.01985,
      "grad_norm": 0.6539060914690586,
      "learning_rate": 0.003,
      "loss": 4.2645,
      "step": 1985
    },
    {
      "epoch": 0.01986,
      "grad_norm": 0.715503037740265,
      "learning_rate": 0.003,
      "loss": 4.2958,
      "step": 1986
    },
    {
      "epoch": 0.01987,
      "grad_norm": 0.7397978022844682,
      "learning_rate": 0.003,
      "loss": 4.254,
      "step": 1987
    },
    {
      "epoch": 0.01988,
      "grad_norm": 0.7632034859099814,
      "learning_rate": 0.003,
      "loss": 4.2717,
      "step": 1988
    },
    {
      "epoch": 0.01989,
      "grad_norm": 0.7294621095424992,
      "learning_rate": 0.003,
      "loss": 4.2599,
      "step": 1989
    },
    {
      "epoch": 0.0199,
      "grad_norm": 0.7424352842692145,
      "learning_rate": 0.003,
      "loss": 4.274,
      "step": 1990
    },
    {
      "epoch": 0.01991,
      "grad_norm": 0.6805432395640066,
      "learning_rate": 0.003,
      "loss": 4.2481,
      "step": 1991
    },
    {
      "epoch": 0.01992,
      "grad_norm": 0.6518823852851582,
      "learning_rate": 0.003,
      "loss": 4.2906,
      "step": 1992
    },
    {
      "epoch": 0.01993,
      "grad_norm": 0.5385914746778909,
      "learning_rate": 0.003,
      "loss": 4.2663,
      "step": 1993
    },
    {
      "epoch": 0.01994,
      "grad_norm": 0.5600374553588182,
      "learning_rate": 0.003,
      "loss": 4.2843,
      "step": 1994
    },
    {
      "epoch": 0.01995,
      "grad_norm": 0.5777853734403589,
      "learning_rate": 0.003,
      "loss": 4.2772,
      "step": 1995
    },
    {
      "epoch": 0.01996,
      "grad_norm": 0.5901601073427315,
      "learning_rate": 0.003,
      "loss": 4.2466,
      "step": 1996
    },
    {
      "epoch": 0.01997,
      "grad_norm": 0.5583731211765088,
      "learning_rate": 0.003,
      "loss": 4.2745,
      "step": 1997
    },
    {
      "epoch": 0.01998,
      "grad_norm": 0.5499515734950207,
      "learning_rate": 0.003,
      "loss": 4.2592,
      "step": 1998
    },
    {
      "epoch": 0.01999,
      "grad_norm": 0.5390040915979398,
      "learning_rate": 0.003,
      "loss": 4.2488,
      "step": 1999
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6085613854495133,
      "learning_rate": 0.003,
      "loss": 4.2357,
      "step": 2000
    },
    {
      "epoch": 0.02001,
      "grad_norm": 0.6642300277845461,
      "learning_rate": 0.003,
      "loss": 4.2761,
      "step": 2001
    },
    {
      "epoch": 0.02002,
      "grad_norm": 0.6718563957297456,
      "learning_rate": 0.003,
      "loss": 4.2714,
      "step": 2002
    },
    {
      "epoch": 0.02003,
      "grad_norm": 0.5656785141879644,
      "learning_rate": 0.003,
      "loss": 4.2645,
      "step": 2003
    },
    {
      "epoch": 0.02004,
      "grad_norm": 0.5079141091367905,
      "learning_rate": 0.003,
      "loss": 4.2507,
      "step": 2004
    },
    {
      "epoch": 0.02005,
      "grad_norm": 0.5217306980533027,
      "learning_rate": 0.003,
      "loss": 4.254,
      "step": 2005
    },
    {
      "epoch": 0.02006,
      "grad_norm": 0.5701023982360506,
      "learning_rate": 0.003,
      "loss": 4.2527,
      "step": 2006
    },
    {
      "epoch": 0.02007,
      "grad_norm": 0.5456537194359483,
      "learning_rate": 0.003,
      "loss": 4.2671,
      "step": 2007
    },
    {
      "epoch": 0.02008,
      "grad_norm": 0.4424819860056267,
      "learning_rate": 0.003,
      "loss": 4.2538,
      "step": 2008
    },
    {
      "epoch": 0.02009,
      "grad_norm": 0.4167499120193345,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 2009
    },
    {
      "epoch": 0.0201,
      "grad_norm": 0.4192128686531188,
      "learning_rate": 0.003,
      "loss": 4.259,
      "step": 2010
    },
    {
      "epoch": 0.02011,
      "grad_norm": 0.3848165653499132,
      "learning_rate": 0.003,
      "loss": 4.2447,
      "step": 2011
    },
    {
      "epoch": 0.02012,
      "grad_norm": 0.433602941803945,
      "learning_rate": 0.003,
      "loss": 4.2428,
      "step": 2012
    },
    {
      "epoch": 0.02013,
      "grad_norm": 0.4888730590149885,
      "learning_rate": 0.003,
      "loss": 4.2535,
      "step": 2013
    },
    {
      "epoch": 0.02014,
      "grad_norm": 0.5673912976929151,
      "learning_rate": 0.003,
      "loss": 4.2387,
      "step": 2014
    },
    {
      "epoch": 0.02015,
      "grad_norm": 0.5541499948622309,
      "learning_rate": 0.003,
      "loss": 4.2361,
      "step": 2015
    },
    {
      "epoch": 0.02016,
      "grad_norm": 0.5126326745775238,
      "learning_rate": 0.003,
      "loss": 4.2417,
      "step": 2016
    },
    {
      "epoch": 0.02017,
      "grad_norm": 0.4179261623064346,
      "learning_rate": 0.003,
      "loss": 4.2316,
      "step": 2017
    },
    {
      "epoch": 0.02018,
      "grad_norm": 0.3898056312044461,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 2018
    },
    {
      "epoch": 0.02019,
      "grad_norm": 0.3954814595294976,
      "learning_rate": 0.003,
      "loss": 4.2373,
      "step": 2019
    },
    {
      "epoch": 0.0202,
      "grad_norm": 0.4213180029076697,
      "learning_rate": 0.003,
      "loss": 4.2455,
      "step": 2020
    },
    {
      "epoch": 0.02021,
      "grad_norm": 0.46415695965874665,
      "learning_rate": 0.003,
      "loss": 4.2601,
      "step": 2021
    },
    {
      "epoch": 0.02022,
      "grad_norm": 0.5255015046967251,
      "learning_rate": 0.003,
      "loss": 4.2204,
      "step": 2022
    },
    {
      "epoch": 0.02023,
      "grad_norm": 0.494786978730351,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2023
    },
    {
      "epoch": 0.02024,
      "grad_norm": 0.4706750110587683,
      "learning_rate": 0.003,
      "loss": 4.2183,
      "step": 2024
    },
    {
      "epoch": 0.02025,
      "grad_norm": 0.4845475406638408,
      "learning_rate": 0.003,
      "loss": 4.2698,
      "step": 2025
    },
    {
      "epoch": 0.02026,
      "grad_norm": 0.5870205306653891,
      "learning_rate": 0.003,
      "loss": 4.2539,
      "step": 2026
    },
    {
      "epoch": 0.02027,
      "grad_norm": 0.5994657871324464,
      "learning_rate": 0.003,
      "loss": 4.2502,
      "step": 2027
    },
    {
      "epoch": 0.02028,
      "grad_norm": 0.5969136086797949,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 2028
    },
    {
      "epoch": 0.02029,
      "grad_norm": 0.6010561087024289,
      "learning_rate": 0.003,
      "loss": 4.2419,
      "step": 2029
    },
    {
      "epoch": 0.0203,
      "grad_norm": 0.6795949256132597,
      "learning_rate": 0.003,
      "loss": 4.2385,
      "step": 2030
    },
    {
      "epoch": 0.02031,
      "grad_norm": 0.8760173401758673,
      "learning_rate": 0.003,
      "loss": 4.2538,
      "step": 2031
    },
    {
      "epoch": 0.02032,
      "grad_norm": 0.9553317966985612,
      "learning_rate": 0.003,
      "loss": 4.2369,
      "step": 2032
    },
    {
      "epoch": 0.02033,
      "grad_norm": 0.8062059561231115,
      "learning_rate": 0.003,
      "loss": 4.2804,
      "step": 2033
    },
    {
      "epoch": 0.02034,
      "grad_norm": 0.6544594771524366,
      "learning_rate": 0.003,
      "loss": 4.2639,
      "step": 2034
    },
    {
      "epoch": 0.02035,
      "grad_norm": 0.5631710363420367,
      "learning_rate": 0.003,
      "loss": 4.2451,
      "step": 2035
    },
    {
      "epoch": 0.02036,
      "grad_norm": 0.5975164123168384,
      "learning_rate": 0.003,
      "loss": 4.2483,
      "step": 2036
    },
    {
      "epoch": 0.02037,
      "grad_norm": 0.5354587611248767,
      "learning_rate": 0.003,
      "loss": 4.2582,
      "step": 2037
    },
    {
      "epoch": 0.02038,
      "grad_norm": 0.421237418399427,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 2038
    },
    {
      "epoch": 0.02039,
      "grad_norm": 0.4274320650935603,
      "learning_rate": 0.003,
      "loss": 4.2464,
      "step": 2039
    },
    {
      "epoch": 0.0204,
      "grad_norm": 0.4164176001256727,
      "learning_rate": 0.003,
      "loss": 4.2496,
      "step": 2040
    },
    {
      "epoch": 0.02041,
      "grad_norm": 0.4311810995214273,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2041
    },
    {
      "epoch": 0.02042,
      "grad_norm": 0.47918244910234625,
      "learning_rate": 0.003,
      "loss": 4.2125,
      "step": 2042
    },
    {
      "epoch": 0.02043,
      "grad_norm": 0.5311186564484092,
      "learning_rate": 0.003,
      "loss": 4.2587,
      "step": 2043
    },
    {
      "epoch": 0.02044,
      "grad_norm": 0.5775700762927194,
      "learning_rate": 0.003,
      "loss": 4.2349,
      "step": 2044
    },
    {
      "epoch": 0.02045,
      "grad_norm": 0.64581468780677,
      "learning_rate": 0.003,
      "loss": 4.2514,
      "step": 2045
    },
    {
      "epoch": 0.02046,
      "grad_norm": 0.6099280998923168,
      "learning_rate": 0.003,
      "loss": 4.2295,
      "step": 2046
    },
    {
      "epoch": 0.02047,
      "grad_norm": 0.49206984816447724,
      "learning_rate": 0.003,
      "loss": 4.2619,
      "step": 2047
    },
    {
      "epoch": 0.02048,
      "grad_norm": 0.6576416076375511,
      "learning_rate": 0.003,
      "loss": 4.2493,
      "step": 2048
    },
    {
      "epoch": 0.02049,
      "grad_norm": 0.720911622296069,
      "learning_rate": 0.003,
      "loss": 4.2776,
      "step": 2049
    },
    {
      "epoch": 0.0205,
      "grad_norm": 0.6221494156113543,
      "learning_rate": 0.003,
      "loss": 4.2218,
      "step": 2050
    },
    {
      "epoch": 0.02051,
      "grad_norm": 0.5686539010207549,
      "learning_rate": 0.003,
      "loss": 4.2223,
      "step": 2051
    },
    {
      "epoch": 0.02052,
      "grad_norm": 0.5121857318353746,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2052
    },
    {
      "epoch": 0.02053,
      "grad_norm": 0.4272252497200639,
      "learning_rate": 0.003,
      "loss": 4.2194,
      "step": 2053
    },
    {
      "epoch": 0.02054,
      "grad_norm": 0.4939477792323304,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 2054
    },
    {
      "epoch": 0.02055,
      "grad_norm": 0.5546078566627058,
      "learning_rate": 0.003,
      "loss": 4.2201,
      "step": 2055
    },
    {
      "epoch": 0.02056,
      "grad_norm": 0.6366201965593732,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2056
    },
    {
      "epoch": 0.02057,
      "grad_norm": 0.6110086842051798,
      "learning_rate": 0.003,
      "loss": 4.2608,
      "step": 2057
    },
    {
      "epoch": 0.02058,
      "grad_norm": 0.6217334659745377,
      "learning_rate": 0.003,
      "loss": 4.2367,
      "step": 2058
    },
    {
      "epoch": 0.02059,
      "grad_norm": 0.651819516754762,
      "learning_rate": 0.003,
      "loss": 4.222,
      "step": 2059
    },
    {
      "epoch": 0.0206,
      "grad_norm": 0.5754276638558378,
      "learning_rate": 0.003,
      "loss": 4.2638,
      "step": 2060
    },
    {
      "epoch": 0.02061,
      "grad_norm": 0.5646246506114238,
      "learning_rate": 0.003,
      "loss": 4.2372,
      "step": 2061
    },
    {
      "epoch": 0.02062,
      "grad_norm": 0.5560224904172448,
      "learning_rate": 0.003,
      "loss": 4.2373,
      "step": 2062
    },
    {
      "epoch": 0.02063,
      "grad_norm": 0.6303655478175813,
      "learning_rate": 0.003,
      "loss": 4.2352,
      "step": 2063
    },
    {
      "epoch": 0.02064,
      "grad_norm": 0.7300283957670998,
      "learning_rate": 0.003,
      "loss": 4.2505,
      "step": 2064
    },
    {
      "epoch": 0.02065,
      "grad_norm": 0.7574662177371257,
      "learning_rate": 0.003,
      "loss": 4.2359,
      "step": 2065
    },
    {
      "epoch": 0.02066,
      "grad_norm": 0.7411303035814747,
      "learning_rate": 0.003,
      "loss": 4.2531,
      "step": 2066
    },
    {
      "epoch": 0.02067,
      "grad_norm": 0.650869346843019,
      "learning_rate": 0.003,
      "loss": 4.2447,
      "step": 2067
    },
    {
      "epoch": 0.02068,
      "grad_norm": 0.6552127487874416,
      "learning_rate": 0.003,
      "loss": 4.2306,
      "step": 2068
    },
    {
      "epoch": 0.02069,
      "grad_norm": 0.592143853498531,
      "learning_rate": 0.003,
      "loss": 4.2516,
      "step": 2069
    },
    {
      "epoch": 0.0207,
      "grad_norm": 0.6207350896242446,
      "learning_rate": 0.003,
      "loss": 4.2263,
      "step": 2070
    },
    {
      "epoch": 0.02071,
      "grad_norm": 0.5661720697139818,
      "learning_rate": 0.003,
      "loss": 4.2543,
      "step": 2071
    },
    {
      "epoch": 0.02072,
      "grad_norm": 0.5613150829964545,
      "learning_rate": 0.003,
      "loss": 4.2589,
      "step": 2072
    },
    {
      "epoch": 0.02073,
      "grad_norm": 0.5913403223082936,
      "learning_rate": 0.003,
      "loss": 4.24,
      "step": 2073
    },
    {
      "epoch": 0.02074,
      "grad_norm": 0.5851338988165525,
      "learning_rate": 0.003,
      "loss": 4.2458,
      "step": 2074
    },
    {
      "epoch": 0.02075,
      "grad_norm": 0.6236167441400716,
      "learning_rate": 0.003,
      "loss": 4.2754,
      "step": 2075
    },
    {
      "epoch": 0.02076,
      "grad_norm": 0.6671767173763447,
      "learning_rate": 0.003,
      "loss": 4.2696,
      "step": 2076
    },
    {
      "epoch": 0.02077,
      "grad_norm": 0.7466794061160641,
      "learning_rate": 0.003,
      "loss": 4.2397,
      "step": 2077
    },
    {
      "epoch": 0.02078,
      "grad_norm": 0.7293207627425712,
      "learning_rate": 0.003,
      "loss": 4.2415,
      "step": 2078
    },
    {
      "epoch": 0.02079,
      "grad_norm": 0.7479234412446394,
      "learning_rate": 0.003,
      "loss": 4.2267,
      "step": 2079
    },
    {
      "epoch": 0.0208,
      "grad_norm": 0.8408782832743513,
      "learning_rate": 0.003,
      "loss": 4.2466,
      "step": 2080
    },
    {
      "epoch": 0.02081,
      "grad_norm": 0.6858851728314246,
      "learning_rate": 0.003,
      "loss": 4.2727,
      "step": 2081
    },
    {
      "epoch": 0.02082,
      "grad_norm": 0.5964291376338854,
      "learning_rate": 0.003,
      "loss": 4.2638,
      "step": 2082
    },
    {
      "epoch": 0.02083,
      "grad_norm": 0.6428002164481011,
      "learning_rate": 0.003,
      "loss": 4.2628,
      "step": 2083
    },
    {
      "epoch": 0.02084,
      "grad_norm": 0.5590580963015779,
      "learning_rate": 0.003,
      "loss": 4.2545,
      "step": 2084
    },
    {
      "epoch": 0.02085,
      "grad_norm": 0.5709511558813687,
      "learning_rate": 0.003,
      "loss": 4.2275,
      "step": 2085
    },
    {
      "epoch": 0.02086,
      "grad_norm": 0.49147509133747364,
      "learning_rate": 0.003,
      "loss": 4.2297,
      "step": 2086
    },
    {
      "epoch": 0.02087,
      "grad_norm": 0.4526072737335104,
      "learning_rate": 0.003,
      "loss": 4.2294,
      "step": 2087
    },
    {
      "epoch": 0.02088,
      "grad_norm": 0.4016974716447893,
      "learning_rate": 0.003,
      "loss": 4.2602,
      "step": 2088
    },
    {
      "epoch": 0.02089,
      "grad_norm": 0.3822543479885703,
      "learning_rate": 0.003,
      "loss": 4.2249,
      "step": 2089
    },
    {
      "epoch": 0.0209,
      "grad_norm": 0.36830581223819503,
      "learning_rate": 0.003,
      "loss": 4.2427,
      "step": 2090
    },
    {
      "epoch": 0.02091,
      "grad_norm": 0.35577804274859626,
      "learning_rate": 0.003,
      "loss": 4.2285,
      "step": 2091
    },
    {
      "epoch": 0.02092,
      "grad_norm": 0.4347176959080428,
      "learning_rate": 0.003,
      "loss": 4.2276,
      "step": 2092
    },
    {
      "epoch": 0.02093,
      "grad_norm": 0.5304539612640176,
      "learning_rate": 0.003,
      "loss": 4.1974,
      "step": 2093
    },
    {
      "epoch": 0.02094,
      "grad_norm": 0.7330731731573918,
      "learning_rate": 0.003,
      "loss": 4.2123,
      "step": 2094
    },
    {
      "epoch": 0.02095,
      "grad_norm": 0.7847225937930644,
      "learning_rate": 0.003,
      "loss": 4.2645,
      "step": 2095
    },
    {
      "epoch": 0.02096,
      "grad_norm": 0.6648124367791377,
      "learning_rate": 0.003,
      "loss": 4.2516,
      "step": 2096
    },
    {
      "epoch": 0.02097,
      "grad_norm": 0.5924239660138091,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2097
    },
    {
      "epoch": 0.02098,
      "grad_norm": 0.5602628841134751,
      "learning_rate": 0.003,
      "loss": 4.2465,
      "step": 2098
    },
    {
      "epoch": 0.02099,
      "grad_norm": 0.46052580380046326,
      "learning_rate": 0.003,
      "loss": 4.2512,
      "step": 2099
    },
    {
      "epoch": 0.021,
      "grad_norm": 0.4339841503760461,
      "learning_rate": 0.003,
      "loss": 4.263,
      "step": 2100
    },
    {
      "epoch": 0.02101,
      "grad_norm": 0.3839438876111581,
      "learning_rate": 0.003,
      "loss": 4.2511,
      "step": 2101
    },
    {
      "epoch": 0.02102,
      "grad_norm": 0.3754480167103612,
      "learning_rate": 0.003,
      "loss": 4.2531,
      "step": 2102
    },
    {
      "epoch": 0.02103,
      "grad_norm": 0.4162207299096809,
      "learning_rate": 0.003,
      "loss": 4.2151,
      "step": 2103
    },
    {
      "epoch": 0.02104,
      "grad_norm": 0.46199291536196674,
      "learning_rate": 0.003,
      "loss": 4.2197,
      "step": 2104
    },
    {
      "epoch": 0.02105,
      "grad_norm": 0.5379532841395008,
      "learning_rate": 0.003,
      "loss": 4.2264,
      "step": 2105
    },
    {
      "epoch": 0.02106,
      "grad_norm": 0.5254543741501657,
      "learning_rate": 0.003,
      "loss": 4.2187,
      "step": 2106
    },
    {
      "epoch": 0.02107,
      "grad_norm": 0.5366355457801288,
      "learning_rate": 0.003,
      "loss": 4.2328,
      "step": 2107
    },
    {
      "epoch": 0.02108,
      "grad_norm": 0.6703887406069353,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 2108
    },
    {
      "epoch": 0.02109,
      "grad_norm": 0.73797667148783,
      "learning_rate": 0.003,
      "loss": 4.2096,
      "step": 2109
    },
    {
      "epoch": 0.0211,
      "grad_norm": 0.7306602624016222,
      "learning_rate": 0.003,
      "loss": 4.2289,
      "step": 2110
    },
    {
      "epoch": 0.02111,
      "grad_norm": 0.7062809207206387,
      "learning_rate": 0.003,
      "loss": 4.2409,
      "step": 2111
    },
    {
      "epoch": 0.02112,
      "grad_norm": 0.6626082871730016,
      "learning_rate": 0.003,
      "loss": 4.2378,
      "step": 2112
    },
    {
      "epoch": 0.02113,
      "grad_norm": 0.7157807873847553,
      "learning_rate": 0.003,
      "loss": 4.238,
      "step": 2113
    },
    {
      "epoch": 0.02114,
      "grad_norm": 0.7509167267520314,
      "learning_rate": 0.003,
      "loss": 4.2327,
      "step": 2114
    },
    {
      "epoch": 0.02115,
      "grad_norm": 0.8631295846669226,
      "learning_rate": 0.003,
      "loss": 4.234,
      "step": 2115
    },
    {
      "epoch": 0.02116,
      "grad_norm": 0.8818862519491879,
      "learning_rate": 0.003,
      "loss": 4.2687,
      "step": 2116
    },
    {
      "epoch": 0.02117,
      "grad_norm": 0.9901874745713439,
      "learning_rate": 0.003,
      "loss": 4.2757,
      "step": 2117
    },
    {
      "epoch": 0.02118,
      "grad_norm": 1.0129584193183192,
      "learning_rate": 0.003,
      "loss": 4.28,
      "step": 2118
    },
    {
      "epoch": 0.02119,
      "grad_norm": 1.0428306836946426,
      "learning_rate": 0.003,
      "loss": 4.2999,
      "step": 2119
    },
    {
      "epoch": 0.0212,
      "grad_norm": 0.8765906436767104,
      "learning_rate": 0.003,
      "loss": 4.2719,
      "step": 2120
    },
    {
      "epoch": 0.02121,
      "grad_norm": 0.7948828861009111,
      "learning_rate": 0.003,
      "loss": 4.2627,
      "step": 2121
    },
    {
      "epoch": 0.02122,
      "grad_norm": 0.7472865629342507,
      "learning_rate": 0.003,
      "loss": 4.2726,
      "step": 2122
    },
    {
      "epoch": 0.02123,
      "grad_norm": 0.6385709948713884,
      "learning_rate": 0.003,
      "loss": 4.2438,
      "step": 2123
    },
    {
      "epoch": 0.02124,
      "grad_norm": 0.557924731350531,
      "learning_rate": 0.003,
      "loss": 4.2611,
      "step": 2124
    },
    {
      "epoch": 0.02125,
      "grad_norm": 0.5786579211625313,
      "learning_rate": 0.003,
      "loss": 4.2778,
      "step": 2125
    },
    {
      "epoch": 0.02126,
      "grad_norm": 0.6027087958052927,
      "learning_rate": 0.003,
      "loss": 4.2718,
      "step": 2126
    },
    {
      "epoch": 0.02127,
      "grad_norm": 0.5438113788648495,
      "learning_rate": 0.003,
      "loss": 4.2756,
      "step": 2127
    },
    {
      "epoch": 0.02128,
      "grad_norm": 0.4557902210786888,
      "learning_rate": 0.003,
      "loss": 4.261,
      "step": 2128
    },
    {
      "epoch": 0.02129,
      "grad_norm": 0.43397893502328583,
      "learning_rate": 0.003,
      "loss": 4.259,
      "step": 2129
    },
    {
      "epoch": 0.0213,
      "grad_norm": 0.4555779798497891,
      "learning_rate": 0.003,
      "loss": 4.233,
      "step": 2130
    },
    {
      "epoch": 0.02131,
      "grad_norm": 0.4779110278336917,
      "learning_rate": 0.003,
      "loss": 4.2569,
      "step": 2131
    },
    {
      "epoch": 0.02132,
      "grad_norm": 0.5370327391448694,
      "learning_rate": 0.003,
      "loss": 4.2673,
      "step": 2132
    },
    {
      "epoch": 0.02133,
      "grad_norm": 0.5952731573509253,
      "learning_rate": 0.003,
      "loss": 4.2603,
      "step": 2133
    },
    {
      "epoch": 0.02134,
      "grad_norm": 0.570127740822763,
      "learning_rate": 0.003,
      "loss": 4.2602,
      "step": 2134
    },
    {
      "epoch": 0.02135,
      "grad_norm": 0.4582370027457041,
      "learning_rate": 0.003,
      "loss": 4.2533,
      "step": 2135
    },
    {
      "epoch": 0.02136,
      "grad_norm": 0.36641117706775855,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 2136
    },
    {
      "epoch": 0.02137,
      "grad_norm": 0.34800353584932325,
      "learning_rate": 0.003,
      "loss": 4.2578,
      "step": 2137
    },
    {
      "epoch": 0.02138,
      "grad_norm": 0.30837616213325136,
      "learning_rate": 0.003,
      "loss": 4.2635,
      "step": 2138
    },
    {
      "epoch": 0.02139,
      "grad_norm": 0.3564245359330032,
      "learning_rate": 0.003,
      "loss": 4.2543,
      "step": 2139
    },
    {
      "epoch": 0.0214,
      "grad_norm": 0.35935426429256007,
      "learning_rate": 0.003,
      "loss": 4.2457,
      "step": 2140
    },
    {
      "epoch": 0.02141,
      "grad_norm": 0.44450585444098134,
      "learning_rate": 0.003,
      "loss": 4.2278,
      "step": 2141
    },
    {
      "epoch": 0.02142,
      "grad_norm": 0.5664462228803099,
      "learning_rate": 0.003,
      "loss": 4.2232,
      "step": 2142
    },
    {
      "epoch": 0.02143,
      "grad_norm": 0.6979889941355174,
      "learning_rate": 0.003,
      "loss": 4.2444,
      "step": 2143
    },
    {
      "epoch": 0.02144,
      "grad_norm": 0.6079500902139742,
      "learning_rate": 0.003,
      "loss": 4.2463,
      "step": 2144
    },
    {
      "epoch": 0.02145,
      "grad_norm": 0.4705352279704145,
      "learning_rate": 0.003,
      "loss": 4.2234,
      "step": 2145
    },
    {
      "epoch": 0.02146,
      "grad_norm": 0.6487033829979426,
      "learning_rate": 0.003,
      "loss": 4.242,
      "step": 2146
    },
    {
      "epoch": 0.02147,
      "grad_norm": 0.646803097358189,
      "learning_rate": 0.003,
      "loss": 4.2643,
      "step": 2147
    },
    {
      "epoch": 0.02148,
      "grad_norm": 0.5846288473404795,
      "learning_rate": 0.003,
      "loss": 4.2698,
      "step": 2148
    },
    {
      "epoch": 0.02149,
      "grad_norm": 0.6324004546408943,
      "learning_rate": 0.003,
      "loss": 4.2592,
      "step": 2149
    },
    {
      "epoch": 0.0215,
      "grad_norm": 0.6348968665767296,
      "learning_rate": 0.003,
      "loss": 4.2447,
      "step": 2150
    },
    {
      "epoch": 0.02151,
      "grad_norm": 0.6038850039685164,
      "learning_rate": 0.003,
      "loss": 4.2496,
      "step": 2151
    },
    {
      "epoch": 0.02152,
      "grad_norm": 0.6571231105206827,
      "learning_rate": 0.003,
      "loss": 4.229,
      "step": 2152
    },
    {
      "epoch": 0.02153,
      "grad_norm": 0.5810533963397566,
      "learning_rate": 0.003,
      "loss": 4.2363,
      "step": 2153
    },
    {
      "epoch": 0.02154,
      "grad_norm": 0.4956565749223054,
      "learning_rate": 0.003,
      "loss": 4.2226,
      "step": 2154
    },
    {
      "epoch": 0.02155,
      "grad_norm": 0.437847806481028,
      "learning_rate": 0.003,
      "loss": 4.252,
      "step": 2155
    },
    {
      "epoch": 0.02156,
      "grad_norm": 0.4180405852162412,
      "learning_rate": 0.003,
      "loss": 4.2371,
      "step": 2156
    },
    {
      "epoch": 0.02157,
      "grad_norm": 0.3760963967155856,
      "learning_rate": 0.003,
      "loss": 4.2006,
      "step": 2157
    },
    {
      "epoch": 0.02158,
      "grad_norm": 0.428171131535246,
      "learning_rate": 0.003,
      "loss": 4.2226,
      "step": 2158
    },
    {
      "epoch": 0.02159,
      "grad_norm": 0.4821995287398208,
      "learning_rate": 0.003,
      "loss": 4.2389,
      "step": 2159
    },
    {
      "epoch": 0.0216,
      "grad_norm": 0.6037348014921639,
      "learning_rate": 0.003,
      "loss": 4.2444,
      "step": 2160
    },
    {
      "epoch": 0.02161,
      "grad_norm": 0.6609447933471442,
      "learning_rate": 0.003,
      "loss": 4.2362,
      "step": 2161
    },
    {
      "epoch": 0.02162,
      "grad_norm": 0.6433252356071486,
      "learning_rate": 0.003,
      "loss": 4.2396,
      "step": 2162
    },
    {
      "epoch": 0.02163,
      "grad_norm": 0.5709983086388406,
      "learning_rate": 0.003,
      "loss": 4.2528,
      "step": 2163
    },
    {
      "epoch": 0.02164,
      "grad_norm": 0.6142803039373849,
      "learning_rate": 0.003,
      "loss": 4.2494,
      "step": 2164
    },
    {
      "epoch": 0.02165,
      "grad_norm": 0.5872422750407742,
      "learning_rate": 0.003,
      "loss": 4.2104,
      "step": 2165
    },
    {
      "epoch": 0.02166,
      "grad_norm": 0.5895046030386504,
      "learning_rate": 0.003,
      "loss": 4.2544,
      "step": 2166
    },
    {
      "epoch": 0.02167,
      "grad_norm": 0.5695409441331826,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 2167
    },
    {
      "epoch": 0.02168,
      "grad_norm": 0.5621779951550028,
      "learning_rate": 0.003,
      "loss": 4.2285,
      "step": 2168
    },
    {
      "epoch": 0.02169,
      "grad_norm": 0.4890626633165933,
      "learning_rate": 0.003,
      "loss": 4.2251,
      "step": 2169
    },
    {
      "epoch": 0.0217,
      "grad_norm": 0.5684142074498969,
      "learning_rate": 0.003,
      "loss": 4.2378,
      "step": 2170
    },
    {
      "epoch": 0.02171,
      "grad_norm": 0.6068809167714128,
      "learning_rate": 0.003,
      "loss": 4.2505,
      "step": 2171
    },
    {
      "epoch": 0.02172,
      "grad_norm": 0.5617111304615819,
      "learning_rate": 0.003,
      "loss": 4.2228,
      "step": 2172
    },
    {
      "epoch": 0.02173,
      "grad_norm": 0.5652162783140336,
      "learning_rate": 0.003,
      "loss": 4.2246,
      "step": 2173
    },
    {
      "epoch": 0.02174,
      "grad_norm": 0.6450142330375198,
      "learning_rate": 0.003,
      "loss": 4.2267,
      "step": 2174
    },
    {
      "epoch": 0.02175,
      "grad_norm": 0.6696169755702878,
      "learning_rate": 0.003,
      "loss": 4.2515,
      "step": 2175
    },
    {
      "epoch": 0.02176,
      "grad_norm": 0.6614934153481808,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 2176
    },
    {
      "epoch": 0.02177,
      "grad_norm": 0.7712758358266676,
      "learning_rate": 0.003,
      "loss": 4.2426,
      "step": 2177
    },
    {
      "epoch": 0.02178,
      "grad_norm": 0.8052241964952243,
      "learning_rate": 0.003,
      "loss": 4.234,
      "step": 2178
    },
    {
      "epoch": 0.02179,
      "grad_norm": 0.8190327773882939,
      "learning_rate": 0.003,
      "loss": 4.2474,
      "step": 2179
    },
    {
      "epoch": 0.0218,
      "grad_norm": 0.7633412782608947,
      "learning_rate": 0.003,
      "loss": 4.2656,
      "step": 2180
    },
    {
      "epoch": 0.02181,
      "grad_norm": 0.6487141963532598,
      "learning_rate": 0.003,
      "loss": 4.2624,
      "step": 2181
    },
    {
      "epoch": 0.02182,
      "grad_norm": 0.5938256063465354,
      "learning_rate": 0.003,
      "loss": 4.2443,
      "step": 2182
    },
    {
      "epoch": 0.02183,
      "grad_norm": 0.572941959470127,
      "learning_rate": 0.003,
      "loss": 4.2546,
      "step": 2183
    },
    {
      "epoch": 0.02184,
      "grad_norm": 0.5472840158022778,
      "learning_rate": 0.003,
      "loss": 4.2258,
      "step": 2184
    },
    {
      "epoch": 0.02185,
      "grad_norm": 0.48428509908133466,
      "learning_rate": 0.003,
      "loss": 4.2187,
      "step": 2185
    },
    {
      "epoch": 0.02186,
      "grad_norm": 0.41644305586561753,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 2186
    },
    {
      "epoch": 0.02187,
      "grad_norm": 0.4220079190945348,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2187
    },
    {
      "epoch": 0.02188,
      "grad_norm": 0.4040099253154041,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2188
    },
    {
      "epoch": 0.02189,
      "grad_norm": 0.384791417229758,
      "learning_rate": 0.003,
      "loss": 4.2412,
      "step": 2189
    },
    {
      "epoch": 0.0219,
      "grad_norm": 0.4200536352754627,
      "learning_rate": 0.003,
      "loss": 4.2271,
      "step": 2190
    },
    {
      "epoch": 0.02191,
      "grad_norm": 0.4219256289842046,
      "learning_rate": 0.003,
      "loss": 4.2197,
      "step": 2191
    },
    {
      "epoch": 0.02192,
      "grad_norm": 0.3796661113601938,
      "learning_rate": 0.003,
      "loss": 4.2468,
      "step": 2192
    },
    {
      "epoch": 0.02193,
      "grad_norm": 0.3951332402646691,
      "learning_rate": 0.003,
      "loss": 4.2124,
      "step": 2193
    },
    {
      "epoch": 0.02194,
      "grad_norm": 0.43785643693424503,
      "learning_rate": 0.003,
      "loss": 4.2632,
      "step": 2194
    },
    {
      "epoch": 0.02195,
      "grad_norm": 0.40101316917194174,
      "learning_rate": 0.003,
      "loss": 4.2286,
      "step": 2195
    },
    {
      "epoch": 0.02196,
      "grad_norm": 0.3895866783766733,
      "learning_rate": 0.003,
      "loss": 4.2337,
      "step": 2196
    },
    {
      "epoch": 0.02197,
      "grad_norm": 0.43978912346203514,
      "learning_rate": 0.003,
      "loss": 4.2283,
      "step": 2197
    },
    {
      "epoch": 0.02198,
      "grad_norm": 0.5128410451451384,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 2198
    },
    {
      "epoch": 0.02199,
      "grad_norm": 0.6337030003238012,
      "learning_rate": 0.003,
      "loss": 4.192,
      "step": 2199
    },
    {
      "epoch": 0.022,
      "grad_norm": 0.8998218740769401,
      "learning_rate": 0.003,
      "loss": 4.2327,
      "step": 2200
    },
    {
      "epoch": 0.02201,
      "grad_norm": 0.9431446506923681,
      "learning_rate": 0.003,
      "loss": 4.266,
      "step": 2201
    },
    {
      "epoch": 0.02202,
      "grad_norm": 0.7612219282493126,
      "learning_rate": 0.003,
      "loss": 4.2433,
      "step": 2202
    },
    {
      "epoch": 0.02203,
      "grad_norm": 0.6635684162527457,
      "learning_rate": 0.003,
      "loss": 4.2702,
      "step": 2203
    },
    {
      "epoch": 0.02204,
      "grad_norm": 0.7621929732197255,
      "learning_rate": 0.003,
      "loss": 4.2688,
      "step": 2204
    },
    {
      "epoch": 0.02205,
      "grad_norm": 0.8741388018917097,
      "learning_rate": 0.003,
      "loss": 4.2762,
      "step": 2205
    },
    {
      "epoch": 0.02206,
      "grad_norm": 0.9703192513292367,
      "learning_rate": 0.003,
      "loss": 4.26,
      "step": 2206
    },
    {
      "epoch": 0.02207,
      "grad_norm": 0.8841717513591386,
      "learning_rate": 0.003,
      "loss": 4.2677,
      "step": 2207
    },
    {
      "epoch": 0.02208,
      "grad_norm": 0.988099891532076,
      "learning_rate": 0.003,
      "loss": 4.2632,
      "step": 2208
    },
    {
      "epoch": 0.02209,
      "grad_norm": 1.1146750705911779,
      "learning_rate": 0.003,
      "loss": 4.3,
      "step": 2209
    },
    {
      "epoch": 0.0221,
      "grad_norm": 0.9650886464683209,
      "learning_rate": 0.003,
      "loss": 4.2786,
      "step": 2210
    },
    {
      "epoch": 0.02211,
      "grad_norm": 0.8938651751677317,
      "learning_rate": 0.003,
      "loss": 4.289,
      "step": 2211
    },
    {
      "epoch": 0.02212,
      "grad_norm": 0.8205609056746809,
      "learning_rate": 0.003,
      "loss": 4.2861,
      "step": 2212
    },
    {
      "epoch": 0.02213,
      "grad_norm": 0.8079312852613237,
      "learning_rate": 0.003,
      "loss": 4.2922,
      "step": 2213
    },
    {
      "epoch": 0.02214,
      "grad_norm": 0.6624076409511721,
      "learning_rate": 0.003,
      "loss": 4.2812,
      "step": 2214
    },
    {
      "epoch": 0.02215,
      "grad_norm": 0.6346284348339961,
      "learning_rate": 0.003,
      "loss": 4.2836,
      "step": 2215
    },
    {
      "epoch": 0.02216,
      "grad_norm": 0.6751649712191042,
      "learning_rate": 0.003,
      "loss": 4.262,
      "step": 2216
    },
    {
      "epoch": 0.02217,
      "grad_norm": 0.7610112582408145,
      "learning_rate": 0.003,
      "loss": 4.2609,
      "step": 2217
    },
    {
      "epoch": 0.02218,
      "grad_norm": 0.7718759523371377,
      "learning_rate": 0.003,
      "loss": 4.2931,
      "step": 2218
    },
    {
      "epoch": 0.02219,
      "grad_norm": 0.7213187726292779,
      "learning_rate": 0.003,
      "loss": 4.2813,
      "step": 2219
    },
    {
      "epoch": 0.0222,
      "grad_norm": 0.6931461020270434,
      "learning_rate": 0.003,
      "loss": 4.2766,
      "step": 2220
    },
    {
      "epoch": 0.02221,
      "grad_norm": 0.6335484790352024,
      "learning_rate": 0.003,
      "loss": 4.2612,
      "step": 2221
    },
    {
      "epoch": 0.02222,
      "grad_norm": 0.5200624241333043,
      "learning_rate": 0.003,
      "loss": 4.2557,
      "step": 2222
    },
    {
      "epoch": 0.02223,
      "grad_norm": 0.47510216931250304,
      "learning_rate": 0.003,
      "loss": 4.2633,
      "step": 2223
    },
    {
      "epoch": 0.02224,
      "grad_norm": 0.4708153372606544,
      "learning_rate": 0.003,
      "loss": 4.2436,
      "step": 2224
    },
    {
      "epoch": 0.02225,
      "grad_norm": 0.4713839636734649,
      "learning_rate": 0.003,
      "loss": 4.2455,
      "step": 2225
    },
    {
      "epoch": 0.02226,
      "grad_norm": 0.45242642141513834,
      "learning_rate": 0.003,
      "loss": 4.2572,
      "step": 2226
    },
    {
      "epoch": 0.02227,
      "grad_norm": 0.5403277615094669,
      "learning_rate": 0.003,
      "loss": 4.2506,
      "step": 2227
    },
    {
      "epoch": 0.02228,
      "grad_norm": 0.5419608649795276,
      "learning_rate": 0.003,
      "loss": 4.2378,
      "step": 2228
    },
    {
      "epoch": 0.02229,
      "grad_norm": 0.5074364347379132,
      "learning_rate": 0.003,
      "loss": 4.2163,
      "step": 2229
    },
    {
      "epoch": 0.0223,
      "grad_norm": 0.44694138444366344,
      "learning_rate": 0.003,
      "loss": 4.2496,
      "step": 2230
    },
    {
      "epoch": 0.02231,
      "grad_norm": 0.3954086546661809,
      "learning_rate": 0.003,
      "loss": 4.2523,
      "step": 2231
    },
    {
      "epoch": 0.02232,
      "grad_norm": 0.38774210113182334,
      "learning_rate": 0.003,
      "loss": 4.2551,
      "step": 2232
    },
    {
      "epoch": 0.02233,
      "grad_norm": 0.33836180367798796,
      "learning_rate": 0.003,
      "loss": 4.2207,
      "step": 2233
    },
    {
      "epoch": 0.02234,
      "grad_norm": 0.4026067706239707,
      "learning_rate": 0.003,
      "loss": 4.2423,
      "step": 2234
    },
    {
      "epoch": 0.02235,
      "grad_norm": 0.3714972499375364,
      "learning_rate": 0.003,
      "loss": 4.2633,
      "step": 2235
    },
    {
      "epoch": 0.02236,
      "grad_norm": 0.3656117183652505,
      "learning_rate": 0.003,
      "loss": 4.2142,
      "step": 2236
    },
    {
      "epoch": 0.02237,
      "grad_norm": 0.3860156575975231,
      "learning_rate": 0.003,
      "loss": 4.248,
      "step": 2237
    },
    {
      "epoch": 0.02238,
      "grad_norm": 0.40106624326868706,
      "learning_rate": 0.003,
      "loss": 4.227,
      "step": 2238
    },
    {
      "epoch": 0.02239,
      "grad_norm": 0.344415203397115,
      "learning_rate": 0.003,
      "loss": 4.2104,
      "step": 2239
    },
    {
      "epoch": 0.0224,
      "grad_norm": 0.30398409326848724,
      "learning_rate": 0.003,
      "loss": 4.1926,
      "step": 2240
    },
    {
      "epoch": 0.02241,
      "grad_norm": 0.29910610471963883,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2241
    },
    {
      "epoch": 0.02242,
      "grad_norm": 0.32765283542789003,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 2242
    },
    {
      "epoch": 0.02243,
      "grad_norm": 0.47068547305005354,
      "learning_rate": 0.003,
      "loss": 4.2397,
      "step": 2243
    },
    {
      "epoch": 0.02244,
      "grad_norm": 0.619857801317153,
      "learning_rate": 0.003,
      "loss": 4.2209,
      "step": 2244
    },
    {
      "epoch": 0.02245,
      "grad_norm": 0.7578137247550054,
      "learning_rate": 0.003,
      "loss": 4.214,
      "step": 2245
    },
    {
      "epoch": 0.02246,
      "grad_norm": 0.7174200654723875,
      "learning_rate": 0.003,
      "loss": 4.2474,
      "step": 2246
    },
    {
      "epoch": 0.02247,
      "grad_norm": 0.5627816069554801,
      "learning_rate": 0.003,
      "loss": 4.2147,
      "step": 2247
    },
    {
      "epoch": 0.02248,
      "grad_norm": 0.645246341768817,
      "learning_rate": 0.003,
      "loss": 4.2267,
      "step": 2248
    },
    {
      "epoch": 0.02249,
      "grad_norm": 0.5894073242429686,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2249
    },
    {
      "epoch": 0.0225,
      "grad_norm": 0.4708614120277191,
      "learning_rate": 0.003,
      "loss": 4.2264,
      "step": 2250
    },
    {
      "epoch": 0.02251,
      "grad_norm": 0.5732703006696422,
      "learning_rate": 0.003,
      "loss": 4.2516,
      "step": 2251
    },
    {
      "epoch": 0.02252,
      "grad_norm": 0.4876319361460763,
      "learning_rate": 0.003,
      "loss": 4.2397,
      "step": 2252
    },
    {
      "epoch": 0.02253,
      "grad_norm": 0.4518791398680973,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 2253
    },
    {
      "epoch": 0.02254,
      "grad_norm": 0.4686604716989295,
      "learning_rate": 0.003,
      "loss": 4.2242,
      "step": 2254
    },
    {
      "epoch": 0.02255,
      "grad_norm": 0.46315510081369105,
      "learning_rate": 0.003,
      "loss": 4.2076,
      "step": 2255
    },
    {
      "epoch": 0.02256,
      "grad_norm": 0.4115888436676222,
      "learning_rate": 0.003,
      "loss": 4.196,
      "step": 2256
    },
    {
      "epoch": 0.02257,
      "grad_norm": 0.4250881653478155,
      "learning_rate": 0.003,
      "loss": 4.213,
      "step": 2257
    },
    {
      "epoch": 0.02258,
      "grad_norm": 0.41687848922253873,
      "learning_rate": 0.003,
      "loss": 4.2225,
      "step": 2258
    },
    {
      "epoch": 0.02259,
      "grad_norm": 0.42540602594527765,
      "learning_rate": 0.003,
      "loss": 4.1835,
      "step": 2259
    },
    {
      "epoch": 0.0226,
      "grad_norm": 0.4465875569855491,
      "learning_rate": 0.003,
      "loss": 4.2393,
      "step": 2260
    },
    {
      "epoch": 0.02261,
      "grad_norm": 0.4515532217997496,
      "learning_rate": 0.003,
      "loss": 4.2163,
      "step": 2261
    },
    {
      "epoch": 0.02262,
      "grad_norm": 0.47647870630071837,
      "learning_rate": 0.003,
      "loss": 4.2167,
      "step": 2262
    },
    {
      "epoch": 0.02263,
      "grad_norm": 0.5731324666338504,
      "learning_rate": 0.003,
      "loss": 4.2339,
      "step": 2263
    },
    {
      "epoch": 0.02264,
      "grad_norm": 0.5964954733768894,
      "learning_rate": 0.003,
      "loss": 4.2281,
      "step": 2264
    },
    {
      "epoch": 0.02265,
      "grad_norm": 0.5712613063923222,
      "learning_rate": 0.003,
      "loss": 4.1855,
      "step": 2265
    },
    {
      "epoch": 0.02266,
      "grad_norm": 0.5312684023721527,
      "learning_rate": 0.003,
      "loss": 4.2007,
      "step": 2266
    },
    {
      "epoch": 0.02267,
      "grad_norm": 0.5398877397723375,
      "learning_rate": 0.003,
      "loss": 4.2294,
      "step": 2267
    },
    {
      "epoch": 0.02268,
      "grad_norm": 0.6097072952396848,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2268
    },
    {
      "epoch": 0.02269,
      "grad_norm": 0.667945707294233,
      "learning_rate": 0.003,
      "loss": 4.2339,
      "step": 2269
    },
    {
      "epoch": 0.0227,
      "grad_norm": 0.6780050185078733,
      "learning_rate": 0.003,
      "loss": 4.23,
      "step": 2270
    },
    {
      "epoch": 0.02271,
      "grad_norm": 0.6140341865228114,
      "learning_rate": 0.003,
      "loss": 4.2328,
      "step": 2271
    },
    {
      "epoch": 0.02272,
      "grad_norm": 0.6077740705952491,
      "learning_rate": 0.003,
      "loss": 4.2343,
      "step": 2272
    },
    {
      "epoch": 0.02273,
      "grad_norm": 0.6228614859955112,
      "learning_rate": 0.003,
      "loss": 4.237,
      "step": 2273
    },
    {
      "epoch": 0.02274,
      "grad_norm": 0.6497689324113504,
      "learning_rate": 0.003,
      "loss": 4.2221,
      "step": 2274
    },
    {
      "epoch": 0.02275,
      "grad_norm": 0.6593608950407358,
      "learning_rate": 0.003,
      "loss": 4.2201,
      "step": 2275
    },
    {
      "epoch": 0.02276,
      "grad_norm": 0.6819172040141601,
      "learning_rate": 0.003,
      "loss": 4.2138,
      "step": 2276
    },
    {
      "epoch": 0.02277,
      "grad_norm": 0.7620205189580369,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 2277
    },
    {
      "epoch": 0.02278,
      "grad_norm": 0.622995687715391,
      "learning_rate": 0.003,
      "loss": 4.2327,
      "step": 2278
    },
    {
      "epoch": 0.02279,
      "grad_norm": 0.5747298625450237,
      "learning_rate": 0.003,
      "loss": 4.2411,
      "step": 2279
    },
    {
      "epoch": 0.0228,
      "grad_norm": 0.5988670118192916,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 2280
    },
    {
      "epoch": 0.02281,
      "grad_norm": 0.611381401213052,
      "learning_rate": 0.003,
      "loss": 4.2272,
      "step": 2281
    },
    {
      "epoch": 0.02282,
      "grad_norm": 0.5846556638089334,
      "learning_rate": 0.003,
      "loss": 4.2409,
      "step": 2282
    },
    {
      "epoch": 0.02283,
      "grad_norm": 0.5475554139402415,
      "learning_rate": 0.003,
      "loss": 4.2536,
      "step": 2283
    },
    {
      "epoch": 0.02284,
      "grad_norm": 0.5238952449646747,
      "learning_rate": 0.003,
      "loss": 4.2482,
      "step": 2284
    },
    {
      "epoch": 0.02285,
      "grad_norm": 0.5449063604054842,
      "learning_rate": 0.003,
      "loss": 4.1851,
      "step": 2285
    },
    {
      "epoch": 0.02286,
      "grad_norm": 0.5626629932890803,
      "learning_rate": 0.003,
      "loss": 4.25,
      "step": 2286
    },
    {
      "epoch": 0.02287,
      "grad_norm": 0.5354326845671384,
      "learning_rate": 0.003,
      "loss": 4.2276,
      "step": 2287
    },
    {
      "epoch": 0.02288,
      "grad_norm": 0.5865237537888423,
      "learning_rate": 0.003,
      "loss": 4.2517,
      "step": 2288
    },
    {
      "epoch": 0.02289,
      "grad_norm": 0.6587817447859279,
      "learning_rate": 0.003,
      "loss": 4.2552,
      "step": 2289
    },
    {
      "epoch": 0.0229,
      "grad_norm": 0.763374859559503,
      "learning_rate": 0.003,
      "loss": 4.2369,
      "step": 2290
    },
    {
      "epoch": 0.02291,
      "grad_norm": 0.9238217255206519,
      "learning_rate": 0.003,
      "loss": 4.2327,
      "step": 2291
    },
    {
      "epoch": 0.02292,
      "grad_norm": 0.8375515673022389,
      "learning_rate": 0.003,
      "loss": 4.2276,
      "step": 2292
    },
    {
      "epoch": 0.02293,
      "grad_norm": 0.8229318263274273,
      "learning_rate": 0.003,
      "loss": 4.233,
      "step": 2293
    },
    {
      "epoch": 0.02294,
      "grad_norm": 0.767668173749955,
      "learning_rate": 0.003,
      "loss": 4.2363,
      "step": 2294
    },
    {
      "epoch": 0.02295,
      "grad_norm": 0.728858935661346,
      "learning_rate": 0.003,
      "loss": 4.2296,
      "step": 2295
    },
    {
      "epoch": 0.02296,
      "grad_norm": 0.74447993660391,
      "learning_rate": 0.003,
      "loss": 4.2662,
      "step": 2296
    },
    {
      "epoch": 0.02297,
      "grad_norm": 0.6968342842354955,
      "learning_rate": 0.003,
      "loss": 4.2327,
      "step": 2297
    },
    {
      "epoch": 0.02298,
      "grad_norm": 0.6411785233315757,
      "learning_rate": 0.003,
      "loss": 4.2441,
      "step": 2298
    },
    {
      "epoch": 0.02299,
      "grad_norm": 0.6765897982497444,
      "learning_rate": 0.003,
      "loss": 4.2472,
      "step": 2299
    },
    {
      "epoch": 0.023,
      "grad_norm": 0.7218316432570615,
      "learning_rate": 0.003,
      "loss": 4.2659,
      "step": 2300
    },
    {
      "epoch": 0.02301,
      "grad_norm": 0.6693972300748606,
      "learning_rate": 0.003,
      "loss": 4.2316,
      "step": 2301
    },
    {
      "epoch": 0.02302,
      "grad_norm": 0.6521855282359511,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 2302
    },
    {
      "epoch": 0.02303,
      "grad_norm": 0.5919587845688344,
      "learning_rate": 0.003,
      "loss": 4.224,
      "step": 2303
    },
    {
      "epoch": 0.02304,
      "grad_norm": 0.602764977489394,
      "learning_rate": 0.003,
      "loss": 4.2562,
      "step": 2304
    },
    {
      "epoch": 0.02305,
      "grad_norm": 0.5558668601822961,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 2305
    },
    {
      "epoch": 0.02306,
      "grad_norm": 0.5156139404377212,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2306
    },
    {
      "epoch": 0.02307,
      "grad_norm": 0.5662595240344765,
      "learning_rate": 0.003,
      "loss": 4.2364,
      "step": 2307
    },
    {
      "epoch": 0.02308,
      "grad_norm": 0.6173971988602593,
      "learning_rate": 0.003,
      "loss": 4.2276,
      "step": 2308
    },
    {
      "epoch": 0.02309,
      "grad_norm": 0.7369887546423364,
      "learning_rate": 0.003,
      "loss": 4.261,
      "step": 2309
    },
    {
      "epoch": 0.0231,
      "grad_norm": 0.7967301688365783,
      "learning_rate": 0.003,
      "loss": 4.2446,
      "step": 2310
    },
    {
      "epoch": 0.02311,
      "grad_norm": 0.6089871818858962,
      "learning_rate": 0.003,
      "loss": 4.245,
      "step": 2311
    },
    {
      "epoch": 0.02312,
      "grad_norm": 0.47586710546753136,
      "learning_rate": 0.003,
      "loss": 4.2111,
      "step": 2312
    },
    {
      "epoch": 0.02313,
      "grad_norm": 0.4974927545735252,
      "learning_rate": 0.003,
      "loss": 4.2079,
      "step": 2313
    },
    {
      "epoch": 0.02314,
      "grad_norm": 0.43769519573127175,
      "learning_rate": 0.003,
      "loss": 4.2204,
      "step": 2314
    },
    {
      "epoch": 0.02315,
      "grad_norm": 0.401786726516975,
      "learning_rate": 0.003,
      "loss": 4.222,
      "step": 2315
    },
    {
      "epoch": 0.02316,
      "grad_norm": 0.4259802780968188,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 2316
    },
    {
      "epoch": 0.02317,
      "grad_norm": 0.41182452876226056,
      "learning_rate": 0.003,
      "loss": 4.2152,
      "step": 2317
    },
    {
      "epoch": 0.02318,
      "grad_norm": 0.4136233759664175,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 2318
    },
    {
      "epoch": 0.02319,
      "grad_norm": 0.3949029424988165,
      "learning_rate": 0.003,
      "loss": 4.2247,
      "step": 2319
    },
    {
      "epoch": 0.0232,
      "grad_norm": 0.35504551442283205,
      "learning_rate": 0.003,
      "loss": 4.2312,
      "step": 2320
    },
    {
      "epoch": 0.02321,
      "grad_norm": 0.32986956749591967,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2321
    },
    {
      "epoch": 0.02322,
      "grad_norm": 0.2968681197139734,
      "learning_rate": 0.003,
      "loss": 4.2156,
      "step": 2322
    },
    {
      "epoch": 0.02323,
      "grad_norm": 0.3207801190374275,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 2323
    },
    {
      "epoch": 0.02324,
      "grad_norm": 0.33142643285723733,
      "learning_rate": 0.003,
      "loss": 4.1895,
      "step": 2324
    },
    {
      "epoch": 0.02325,
      "grad_norm": 0.3535725697371811,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 2325
    },
    {
      "epoch": 0.02326,
      "grad_norm": 0.3678851746936938,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2326
    },
    {
      "epoch": 0.02327,
      "grad_norm": 0.4101658150492574,
      "learning_rate": 0.003,
      "loss": 4.2161,
      "step": 2327
    },
    {
      "epoch": 0.02328,
      "grad_norm": 0.4519738717854165,
      "learning_rate": 0.003,
      "loss": 4.2155,
      "step": 2328
    },
    {
      "epoch": 0.02329,
      "grad_norm": 0.5647347753978986,
      "learning_rate": 0.003,
      "loss": 4.2089,
      "step": 2329
    },
    {
      "epoch": 0.0233,
      "grad_norm": 0.6839573886066393,
      "learning_rate": 0.003,
      "loss": 4.2105,
      "step": 2330
    },
    {
      "epoch": 0.02331,
      "grad_norm": 0.7998107409261247,
      "learning_rate": 0.003,
      "loss": 4.2543,
      "step": 2331
    },
    {
      "epoch": 0.02332,
      "grad_norm": 0.689815752941326,
      "learning_rate": 0.003,
      "loss": 4.2161,
      "step": 2332
    },
    {
      "epoch": 0.02333,
      "grad_norm": 0.6555251301313495,
      "learning_rate": 0.003,
      "loss": 4.222,
      "step": 2333
    },
    {
      "epoch": 0.02334,
      "grad_norm": 0.6263058920174612,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 2334
    },
    {
      "epoch": 0.02335,
      "grad_norm": 0.5873514688131092,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 2335
    },
    {
      "epoch": 0.02336,
      "grad_norm": 0.6178829826634938,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2336
    },
    {
      "epoch": 0.02337,
      "grad_norm": 0.6776239992532147,
      "learning_rate": 0.003,
      "loss": 4.2338,
      "step": 2337
    },
    {
      "epoch": 0.02338,
      "grad_norm": 0.6079153730498182,
      "learning_rate": 0.003,
      "loss": 4.2143,
      "step": 2338
    },
    {
      "epoch": 0.02339,
      "grad_norm": 0.5540805388644168,
      "learning_rate": 0.003,
      "loss": 4.2097,
      "step": 2339
    },
    {
      "epoch": 0.0234,
      "grad_norm": 0.4930522285796136,
      "learning_rate": 0.003,
      "loss": 4.2219,
      "step": 2340
    },
    {
      "epoch": 0.02341,
      "grad_norm": 0.5114159866478617,
      "learning_rate": 0.003,
      "loss": 4.198,
      "step": 2341
    },
    {
      "epoch": 0.02342,
      "grad_norm": 0.5088489673837915,
      "learning_rate": 0.003,
      "loss": 4.212,
      "step": 2342
    },
    {
      "epoch": 0.02343,
      "grad_norm": 0.5701232150213591,
      "learning_rate": 0.003,
      "loss": 4.2136,
      "step": 2343
    },
    {
      "epoch": 0.02344,
      "grad_norm": 0.5264496634765071,
      "learning_rate": 0.003,
      "loss": 4.2103,
      "step": 2344
    },
    {
      "epoch": 0.02345,
      "grad_norm": 0.463092136026227,
      "learning_rate": 0.003,
      "loss": 4.2037,
      "step": 2345
    },
    {
      "epoch": 0.02346,
      "grad_norm": 0.5119464591082599,
      "learning_rate": 0.003,
      "loss": 4.2041,
      "step": 2346
    },
    {
      "epoch": 0.02347,
      "grad_norm": 0.5680858879896561,
      "learning_rate": 0.003,
      "loss": 4.2412,
      "step": 2347
    },
    {
      "epoch": 0.02348,
      "grad_norm": 0.6575493478001748,
      "learning_rate": 0.003,
      "loss": 4.2098,
      "step": 2348
    },
    {
      "epoch": 0.02349,
      "grad_norm": 0.7446709988793854,
      "learning_rate": 0.003,
      "loss": 4.2329,
      "step": 2349
    },
    {
      "epoch": 0.0235,
      "grad_norm": 0.9616447862037343,
      "learning_rate": 0.003,
      "loss": 4.2417,
      "step": 2350
    },
    {
      "epoch": 0.02351,
      "grad_norm": 0.892365430816041,
      "learning_rate": 0.003,
      "loss": 4.2567,
      "step": 2351
    },
    {
      "epoch": 0.02352,
      "grad_norm": 0.7434115511547027,
      "learning_rate": 0.003,
      "loss": 4.2055,
      "step": 2352
    },
    {
      "epoch": 0.02353,
      "grad_norm": 0.7750549493995498,
      "learning_rate": 0.003,
      "loss": 4.242,
      "step": 2353
    },
    {
      "epoch": 0.02354,
      "grad_norm": 0.7464193045182832,
      "learning_rate": 0.003,
      "loss": 4.2509,
      "step": 2354
    },
    {
      "epoch": 0.02355,
      "grad_norm": 0.6497010807293944,
      "learning_rate": 0.003,
      "loss": 4.2393,
      "step": 2355
    },
    {
      "epoch": 0.02356,
      "grad_norm": 0.6914197602121488,
      "learning_rate": 0.003,
      "loss": 4.2239,
      "step": 2356
    },
    {
      "epoch": 0.02357,
      "grad_norm": 0.6846425351668688,
      "learning_rate": 0.003,
      "loss": 4.2295,
      "step": 2357
    },
    {
      "epoch": 0.02358,
      "grad_norm": 0.6290127860120783,
      "learning_rate": 0.003,
      "loss": 4.2459,
      "step": 2358
    },
    {
      "epoch": 0.02359,
      "grad_norm": 0.6298376689313879,
      "learning_rate": 0.003,
      "loss": 4.2228,
      "step": 2359
    },
    {
      "epoch": 0.0236,
      "grad_norm": 0.5166523858154795,
      "learning_rate": 0.003,
      "loss": 4.256,
      "step": 2360
    },
    {
      "epoch": 0.02361,
      "grad_norm": 0.5392240288462758,
      "learning_rate": 0.003,
      "loss": 4.2371,
      "step": 2361
    },
    {
      "epoch": 0.02362,
      "grad_norm": 0.46488034590567795,
      "learning_rate": 0.003,
      "loss": 4.2246,
      "step": 2362
    },
    {
      "epoch": 0.02363,
      "grad_norm": 0.45695605203462375,
      "learning_rate": 0.003,
      "loss": 4.1978,
      "step": 2363
    },
    {
      "epoch": 0.02364,
      "grad_norm": 0.450417146973962,
      "learning_rate": 0.003,
      "loss": 4.2171,
      "step": 2364
    },
    {
      "epoch": 0.02365,
      "grad_norm": 0.46003087118242425,
      "learning_rate": 0.003,
      "loss": 4.236,
      "step": 2365
    },
    {
      "epoch": 0.02366,
      "grad_norm": 0.48662211661939403,
      "learning_rate": 0.003,
      "loss": 4.2096,
      "step": 2366
    },
    {
      "epoch": 0.02367,
      "grad_norm": 0.5645419223124684,
      "learning_rate": 0.003,
      "loss": 4.2212,
      "step": 2367
    },
    {
      "epoch": 0.02368,
      "grad_norm": 0.5895187996336625,
      "learning_rate": 0.003,
      "loss": 4.1956,
      "step": 2368
    },
    {
      "epoch": 0.02369,
      "grad_norm": 0.6664022667321152,
      "learning_rate": 0.003,
      "loss": 4.211,
      "step": 2369
    },
    {
      "epoch": 0.0237,
      "grad_norm": 0.6992460452789251,
      "learning_rate": 0.003,
      "loss": 4.241,
      "step": 2370
    },
    {
      "epoch": 0.02371,
      "grad_norm": 0.6591985126074834,
      "learning_rate": 0.003,
      "loss": 4.223,
      "step": 2371
    },
    {
      "epoch": 0.02372,
      "grad_norm": 0.5265059072138487,
      "learning_rate": 0.003,
      "loss": 4.2082,
      "step": 2372
    },
    {
      "epoch": 0.02373,
      "grad_norm": 0.4951201634347875,
      "learning_rate": 0.003,
      "loss": 4.1933,
      "step": 2373
    },
    {
      "epoch": 0.02374,
      "grad_norm": 0.4876679210846313,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 2374
    },
    {
      "epoch": 0.02375,
      "grad_norm": 0.408164979920852,
      "learning_rate": 0.003,
      "loss": 4.2017,
      "step": 2375
    },
    {
      "epoch": 0.02376,
      "grad_norm": 0.3895969391943079,
      "learning_rate": 0.003,
      "loss": 4.2257,
      "step": 2376
    },
    {
      "epoch": 0.02377,
      "grad_norm": 0.4053428947086922,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 2377
    },
    {
      "epoch": 0.02378,
      "grad_norm": 0.43012507081905327,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 2378
    },
    {
      "epoch": 0.02379,
      "grad_norm": 0.4764061060377796,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 2379
    },
    {
      "epoch": 0.0238,
      "grad_norm": 0.4854688158872303,
      "learning_rate": 0.003,
      "loss": 4.1642,
      "step": 2380
    },
    {
      "epoch": 0.02381,
      "grad_norm": 0.4926226608113921,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 2381
    },
    {
      "epoch": 0.02382,
      "grad_norm": 0.5715519289577061,
      "learning_rate": 0.003,
      "loss": 4.2369,
      "step": 2382
    },
    {
      "epoch": 0.02383,
      "grad_norm": 0.5852310729435649,
      "learning_rate": 0.003,
      "loss": 4.2016,
      "step": 2383
    },
    {
      "epoch": 0.02384,
      "grad_norm": 0.5027933039192362,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 2384
    },
    {
      "epoch": 0.02385,
      "grad_norm": 0.5671992764095898,
      "learning_rate": 0.003,
      "loss": 4.2151,
      "step": 2385
    },
    {
      "epoch": 0.02386,
      "grad_norm": 0.5716874891458956,
      "learning_rate": 0.003,
      "loss": 4.1952,
      "step": 2386
    },
    {
      "epoch": 0.02387,
      "grad_norm": 0.5519898067526273,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 2387
    },
    {
      "epoch": 0.02388,
      "grad_norm": 0.5749264518166138,
      "learning_rate": 0.003,
      "loss": 4.2477,
      "step": 2388
    },
    {
      "epoch": 0.02389,
      "grad_norm": 0.6599062653773942,
      "learning_rate": 0.003,
      "loss": 4.2112,
      "step": 2389
    },
    {
      "epoch": 0.0239,
      "grad_norm": 0.8007810046111682,
      "learning_rate": 0.003,
      "loss": 4.2104,
      "step": 2390
    },
    {
      "epoch": 0.02391,
      "grad_norm": 0.7728699720546903,
      "learning_rate": 0.003,
      "loss": 4.2355,
      "step": 2391
    },
    {
      "epoch": 0.02392,
      "grad_norm": 0.637372032061665,
      "learning_rate": 0.003,
      "loss": 4.2375,
      "step": 2392
    },
    {
      "epoch": 0.02393,
      "grad_norm": 0.5856837925361781,
      "learning_rate": 0.003,
      "loss": 4.1967,
      "step": 2393
    },
    {
      "epoch": 0.02394,
      "grad_norm": 0.5796441189535547,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2394
    },
    {
      "epoch": 0.02395,
      "grad_norm": 0.49276837104854343,
      "learning_rate": 0.003,
      "loss": 4.2217,
      "step": 2395
    },
    {
      "epoch": 0.02396,
      "grad_norm": 0.4942734378430775,
      "learning_rate": 0.003,
      "loss": 4.2363,
      "step": 2396
    },
    {
      "epoch": 0.02397,
      "grad_norm": 0.4893677735684375,
      "learning_rate": 0.003,
      "loss": 4.2162,
      "step": 2397
    },
    {
      "epoch": 0.02398,
      "grad_norm": 0.5811783701919775,
      "learning_rate": 0.003,
      "loss": 4.2259,
      "step": 2398
    },
    {
      "epoch": 0.02399,
      "grad_norm": 0.7848440120318722,
      "learning_rate": 0.003,
      "loss": 4.2123,
      "step": 2399
    },
    {
      "epoch": 0.024,
      "grad_norm": 1.102610977529852,
      "learning_rate": 0.003,
      "loss": 4.2476,
      "step": 2400
    },
    {
      "epoch": 0.02401,
      "grad_norm": 0.8135265365179677,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 2401
    },
    {
      "epoch": 0.02402,
      "grad_norm": 0.583172308220775,
      "learning_rate": 0.003,
      "loss": 4.24,
      "step": 2402
    },
    {
      "epoch": 0.02403,
      "grad_norm": 0.5891421965222777,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 2403
    },
    {
      "epoch": 0.02404,
      "grad_norm": 0.5648336236466582,
      "learning_rate": 0.003,
      "loss": 4.2446,
      "step": 2404
    },
    {
      "epoch": 0.02405,
      "grad_norm": 0.5745395036103889,
      "learning_rate": 0.003,
      "loss": 4.2523,
      "step": 2405
    },
    {
      "epoch": 0.02406,
      "grad_norm": 0.5533135454038,
      "learning_rate": 0.003,
      "loss": 4.2331,
      "step": 2406
    },
    {
      "epoch": 0.02407,
      "grad_norm": 0.49929019237957983,
      "learning_rate": 0.003,
      "loss": 4.2073,
      "step": 2407
    },
    {
      "epoch": 0.02408,
      "grad_norm": 0.4943118983366084,
      "learning_rate": 0.003,
      "loss": 4.2144,
      "step": 2408
    },
    {
      "epoch": 0.02409,
      "grad_norm": 0.5449234611486594,
      "learning_rate": 0.003,
      "loss": 4.2212,
      "step": 2409
    },
    {
      "epoch": 0.0241,
      "grad_norm": 0.585465535756614,
      "learning_rate": 0.003,
      "loss": 4.2358,
      "step": 2410
    },
    {
      "epoch": 0.02411,
      "grad_norm": 0.6698135788102235,
      "learning_rate": 0.003,
      "loss": 4.2088,
      "step": 2411
    },
    {
      "epoch": 0.02412,
      "grad_norm": 0.7369339920597645,
      "learning_rate": 0.003,
      "loss": 4.217,
      "step": 2412
    },
    {
      "epoch": 0.02413,
      "grad_norm": 0.7305355993897886,
      "learning_rate": 0.003,
      "loss": 4.2166,
      "step": 2413
    },
    {
      "epoch": 0.02414,
      "grad_norm": 0.7556720083696307,
      "learning_rate": 0.003,
      "loss": 4.2325,
      "step": 2414
    },
    {
      "epoch": 0.02415,
      "grad_norm": 0.6579993879979832,
      "learning_rate": 0.003,
      "loss": 4.2326,
      "step": 2415
    },
    {
      "epoch": 0.02416,
      "grad_norm": 0.6108411981053206,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2416
    },
    {
      "epoch": 0.02417,
      "grad_norm": 0.5049292600507906,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 2417
    },
    {
      "epoch": 0.02418,
      "grad_norm": 0.4900011615352856,
      "learning_rate": 0.003,
      "loss": 4.2259,
      "step": 2418
    },
    {
      "epoch": 0.02419,
      "grad_norm": 0.4521560630393701,
      "learning_rate": 0.003,
      "loss": 4.2185,
      "step": 2419
    },
    {
      "epoch": 0.0242,
      "grad_norm": 0.4620681001428492,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 2420
    },
    {
      "epoch": 0.02421,
      "grad_norm": 0.4916389353511205,
      "learning_rate": 0.003,
      "loss": 4.1947,
      "step": 2421
    },
    {
      "epoch": 0.02422,
      "grad_norm": 0.5425187775211936,
      "learning_rate": 0.003,
      "loss": 4.2503,
      "step": 2422
    },
    {
      "epoch": 0.02423,
      "grad_norm": 0.6038360240873328,
      "learning_rate": 0.003,
      "loss": 4.2434,
      "step": 2423
    },
    {
      "epoch": 0.02424,
      "grad_norm": 0.6325505331996517,
      "learning_rate": 0.003,
      "loss": 4.2388,
      "step": 2424
    },
    {
      "epoch": 0.02425,
      "grad_norm": 0.5934465012445626,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2425
    },
    {
      "epoch": 0.02426,
      "grad_norm": 0.4852317425419543,
      "learning_rate": 0.003,
      "loss": 4.2259,
      "step": 2426
    },
    {
      "epoch": 0.02427,
      "grad_norm": 0.43424836675618983,
      "learning_rate": 0.003,
      "loss": 4.2507,
      "step": 2427
    },
    {
      "epoch": 0.02428,
      "grad_norm": 0.42558088050674436,
      "learning_rate": 0.003,
      "loss": 4.2206,
      "step": 2428
    },
    {
      "epoch": 0.02429,
      "grad_norm": 0.4007132170531262,
      "learning_rate": 0.003,
      "loss": 4.2052,
      "step": 2429
    },
    {
      "epoch": 0.0243,
      "grad_norm": 0.41461133903064823,
      "learning_rate": 0.003,
      "loss": 4.2277,
      "step": 2430
    },
    {
      "epoch": 0.02431,
      "grad_norm": 0.42272490135521923,
      "learning_rate": 0.003,
      "loss": 4.2037,
      "step": 2431
    },
    {
      "epoch": 0.02432,
      "grad_norm": 0.5028259204616563,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2432
    },
    {
      "epoch": 0.02433,
      "grad_norm": 0.6593220419533437,
      "learning_rate": 0.003,
      "loss": 4.2479,
      "step": 2433
    },
    {
      "epoch": 0.02434,
      "grad_norm": 0.8027427099690023,
      "learning_rate": 0.003,
      "loss": 4.2219,
      "step": 2434
    },
    {
      "epoch": 0.02435,
      "grad_norm": 0.8024207699638355,
      "learning_rate": 0.003,
      "loss": 4.2281,
      "step": 2435
    },
    {
      "epoch": 0.02436,
      "grad_norm": 0.7133615297883351,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2436
    },
    {
      "epoch": 0.02437,
      "grad_norm": 0.6439666201166401,
      "learning_rate": 0.003,
      "loss": 4.2142,
      "step": 2437
    },
    {
      "epoch": 0.02438,
      "grad_norm": 0.5858625518676239,
      "learning_rate": 0.003,
      "loss": 4.23,
      "step": 2438
    },
    {
      "epoch": 0.02439,
      "grad_norm": 0.5398100187318233,
      "learning_rate": 0.003,
      "loss": 4.1977,
      "step": 2439
    },
    {
      "epoch": 0.0244,
      "grad_norm": 0.5287852378026328,
      "learning_rate": 0.003,
      "loss": 4.2469,
      "step": 2440
    },
    {
      "epoch": 0.02441,
      "grad_norm": 0.5446152594519079,
      "learning_rate": 0.003,
      "loss": 4.2079,
      "step": 2441
    },
    {
      "epoch": 0.02442,
      "grad_norm": 0.5050019658896749,
      "learning_rate": 0.003,
      "loss": 4.2415,
      "step": 2442
    },
    {
      "epoch": 0.02443,
      "grad_norm": 0.5178076569823932,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 2443
    },
    {
      "epoch": 0.02444,
      "grad_norm": 0.48889067635139094,
      "learning_rate": 0.003,
      "loss": 4.2045,
      "step": 2444
    },
    {
      "epoch": 0.02445,
      "grad_norm": 0.5068159078989423,
      "learning_rate": 0.003,
      "loss": 4.2106,
      "step": 2445
    },
    {
      "epoch": 0.02446,
      "grad_norm": 0.5126341533127841,
      "learning_rate": 0.003,
      "loss": 4.2121,
      "step": 2446
    },
    {
      "epoch": 0.02447,
      "grad_norm": 0.4825218100648375,
      "learning_rate": 0.003,
      "loss": 4.2174,
      "step": 2447
    },
    {
      "epoch": 0.02448,
      "grad_norm": 0.46377191023349634,
      "learning_rate": 0.003,
      "loss": 4.2169,
      "step": 2448
    },
    {
      "epoch": 0.02449,
      "grad_norm": 0.5086550488721309,
      "learning_rate": 0.003,
      "loss": 4.2116,
      "step": 2449
    },
    {
      "epoch": 0.0245,
      "grad_norm": 0.5740571687783603,
      "learning_rate": 0.003,
      "loss": 4.2159,
      "step": 2450
    },
    {
      "epoch": 0.02451,
      "grad_norm": 0.5066757455693393,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 2451
    },
    {
      "epoch": 0.02452,
      "grad_norm": 0.40863119709622003,
      "learning_rate": 0.003,
      "loss": 4.1968,
      "step": 2452
    },
    {
      "epoch": 0.02453,
      "grad_norm": 0.44965288153972177,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 2453
    },
    {
      "epoch": 0.02454,
      "grad_norm": 0.5324108231499827,
      "learning_rate": 0.003,
      "loss": 4.1829,
      "step": 2454
    },
    {
      "epoch": 0.02455,
      "grad_norm": 0.6445398719560812,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2455
    },
    {
      "epoch": 0.02456,
      "grad_norm": 0.778972240900702,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 2456
    },
    {
      "epoch": 0.02457,
      "grad_norm": 0.6895291525630388,
      "learning_rate": 0.003,
      "loss": 4.2281,
      "step": 2457
    },
    {
      "epoch": 0.02458,
      "grad_norm": 0.5932547289808822,
      "learning_rate": 0.003,
      "loss": 4.1934,
      "step": 2458
    },
    {
      "epoch": 0.02459,
      "grad_norm": 0.5873468472438081,
      "learning_rate": 0.003,
      "loss": 4.2234,
      "step": 2459
    },
    {
      "epoch": 0.0246,
      "grad_norm": 0.6051334869486308,
      "learning_rate": 0.003,
      "loss": 4.2298,
      "step": 2460
    },
    {
      "epoch": 0.02461,
      "grad_norm": 0.598960720110885,
      "learning_rate": 0.003,
      "loss": 4.2047,
      "step": 2461
    },
    {
      "epoch": 0.02462,
      "grad_norm": 0.5336669037751343,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 2462
    },
    {
      "epoch": 0.02463,
      "grad_norm": 0.5634439620834016,
      "learning_rate": 0.003,
      "loss": 4.2301,
      "step": 2463
    },
    {
      "epoch": 0.02464,
      "grad_norm": 0.6026663982662849,
      "learning_rate": 0.003,
      "loss": 4.2123,
      "step": 2464
    },
    {
      "epoch": 0.02465,
      "grad_norm": 0.7167643023421324,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 2465
    },
    {
      "epoch": 0.02466,
      "grad_norm": 0.7713681863589867,
      "learning_rate": 0.003,
      "loss": 4.2172,
      "step": 2466
    },
    {
      "epoch": 0.02467,
      "grad_norm": 0.6973139861610792,
      "learning_rate": 0.003,
      "loss": 4.1958,
      "step": 2467
    },
    {
      "epoch": 0.02468,
      "grad_norm": 0.6102611547152099,
      "learning_rate": 0.003,
      "loss": 4.2374,
      "step": 2468
    },
    {
      "epoch": 0.02469,
      "grad_norm": 0.6541217572054333,
      "learning_rate": 0.003,
      "loss": 4.2207,
      "step": 2469
    },
    {
      "epoch": 0.0247,
      "grad_norm": 0.6679631943322341,
      "learning_rate": 0.003,
      "loss": 4.2165,
      "step": 2470
    },
    {
      "epoch": 0.02471,
      "grad_norm": 0.7159740472340401,
      "learning_rate": 0.003,
      "loss": 4.2074,
      "step": 2471
    },
    {
      "epoch": 0.02472,
      "grad_norm": 0.7905007318669144,
      "learning_rate": 0.003,
      "loss": 4.2152,
      "step": 2472
    },
    {
      "epoch": 0.02473,
      "grad_norm": 0.8760585933429962,
      "learning_rate": 0.003,
      "loss": 4.2103,
      "step": 2473
    },
    {
      "epoch": 0.02474,
      "grad_norm": 0.9178632648529088,
      "learning_rate": 0.003,
      "loss": 4.2472,
      "step": 2474
    },
    {
      "epoch": 0.02475,
      "grad_norm": 0.8011380404965301,
      "learning_rate": 0.003,
      "loss": 4.2214,
      "step": 2475
    },
    {
      "epoch": 0.02476,
      "grad_norm": 0.8709257402916254,
      "learning_rate": 0.003,
      "loss": 4.2334,
      "step": 2476
    },
    {
      "epoch": 0.02477,
      "grad_norm": 0.8785057383582693,
      "learning_rate": 0.003,
      "loss": 4.2701,
      "step": 2477
    },
    {
      "epoch": 0.02478,
      "grad_norm": 0.8265644129631496,
      "learning_rate": 0.003,
      "loss": 4.2383,
      "step": 2478
    },
    {
      "epoch": 0.02479,
      "grad_norm": 0.7447630055654997,
      "learning_rate": 0.003,
      "loss": 4.228,
      "step": 2479
    },
    {
      "epoch": 0.0248,
      "grad_norm": 0.7215953534640771,
      "learning_rate": 0.003,
      "loss": 4.2659,
      "step": 2480
    },
    {
      "epoch": 0.02481,
      "grad_norm": 0.7223104530962532,
      "learning_rate": 0.003,
      "loss": 4.2494,
      "step": 2481
    },
    {
      "epoch": 0.02482,
      "grad_norm": 0.7755415514978912,
      "learning_rate": 0.003,
      "loss": 4.2642,
      "step": 2482
    },
    {
      "epoch": 0.02483,
      "grad_norm": 0.8000293423835445,
      "learning_rate": 0.003,
      "loss": 4.2156,
      "step": 2483
    },
    {
      "epoch": 0.02484,
      "grad_norm": 0.6385982842425331,
      "learning_rate": 0.003,
      "loss": 4.2408,
      "step": 2484
    },
    {
      "epoch": 0.02485,
      "grad_norm": 0.7165633164736369,
      "learning_rate": 0.003,
      "loss": 4.2628,
      "step": 2485
    },
    {
      "epoch": 0.02486,
      "grad_norm": 0.666869104779761,
      "learning_rate": 0.003,
      "loss": 4.2845,
      "step": 2486
    },
    {
      "epoch": 0.02487,
      "grad_norm": 0.6526919813636585,
      "learning_rate": 0.003,
      "loss": 4.2376,
      "step": 2487
    },
    {
      "epoch": 0.02488,
      "grad_norm": 0.6689740712518896,
      "learning_rate": 0.003,
      "loss": 4.2153,
      "step": 2488
    },
    {
      "epoch": 0.02489,
      "grad_norm": 0.6762316340846642,
      "learning_rate": 0.003,
      "loss": 4.2054,
      "step": 2489
    },
    {
      "epoch": 0.0249,
      "grad_norm": 0.6381111552878137,
      "learning_rate": 0.003,
      "loss": 4.2081,
      "step": 2490
    },
    {
      "epoch": 0.02491,
      "grad_norm": 0.6575656650220585,
      "learning_rate": 0.003,
      "loss": 4.234,
      "step": 2491
    },
    {
      "epoch": 0.02492,
      "grad_norm": 0.6264756996157673,
      "learning_rate": 0.003,
      "loss": 4.243,
      "step": 2492
    },
    {
      "epoch": 0.02493,
      "grad_norm": 0.5475484048874111,
      "learning_rate": 0.003,
      "loss": 4.2085,
      "step": 2493
    },
    {
      "epoch": 0.02494,
      "grad_norm": 0.5832237103576846,
      "learning_rate": 0.003,
      "loss": 4.222,
      "step": 2494
    },
    {
      "epoch": 0.02495,
      "grad_norm": 0.49782209661562854,
      "learning_rate": 0.003,
      "loss": 4.2274,
      "step": 2495
    },
    {
      "epoch": 0.02496,
      "grad_norm": 0.403753171755924,
      "learning_rate": 0.003,
      "loss": 4.2141,
      "step": 2496
    },
    {
      "epoch": 0.02497,
      "grad_norm": 0.36378014443485196,
      "learning_rate": 0.003,
      "loss": 4.2165,
      "step": 2497
    },
    {
      "epoch": 0.02498,
      "grad_norm": 0.317230162546613,
      "learning_rate": 0.003,
      "loss": 4.2025,
      "step": 2498
    },
    {
      "epoch": 0.02499,
      "grad_norm": 0.3277491553637038,
      "learning_rate": 0.003,
      "loss": 4.2117,
      "step": 2499
    },
    {
      "epoch": 0.025,
      "grad_norm": 0.2882430663823555,
      "learning_rate": 0.003,
      "loss": 4.1956,
      "step": 2500
    },
    {
      "epoch": 0.02501,
      "grad_norm": 0.27906079169809134,
      "learning_rate": 0.003,
      "loss": 4.2251,
      "step": 2501
    },
    {
      "epoch": 0.02502,
      "grad_norm": 0.2871738503175601,
      "learning_rate": 0.003,
      "loss": 4.2056,
      "step": 2502
    },
    {
      "epoch": 0.02503,
      "grad_norm": 0.3203204666233,
      "learning_rate": 0.003,
      "loss": 4.2125,
      "step": 2503
    },
    {
      "epoch": 0.02504,
      "grad_norm": 0.38939921573272746,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 2504
    },
    {
      "epoch": 0.02505,
      "grad_norm": 0.5688786331558378,
      "learning_rate": 0.003,
      "loss": 4.1964,
      "step": 2505
    },
    {
      "epoch": 0.02506,
      "grad_norm": 0.7962360843283562,
      "learning_rate": 0.003,
      "loss": 4.2243,
      "step": 2506
    },
    {
      "epoch": 0.02507,
      "grad_norm": 0.8008164951690123,
      "learning_rate": 0.003,
      "loss": 4.2174,
      "step": 2507
    },
    {
      "epoch": 0.02508,
      "grad_norm": 0.5143170677845642,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 2508
    },
    {
      "epoch": 0.02509,
      "grad_norm": 0.6566394093706333,
      "learning_rate": 0.003,
      "loss": 4.2119,
      "step": 2509
    },
    {
      "epoch": 0.0251,
      "grad_norm": 0.7710410640765211,
      "learning_rate": 0.003,
      "loss": 4.2285,
      "step": 2510
    },
    {
      "epoch": 0.02511,
      "grad_norm": 0.6799202006125559,
      "learning_rate": 0.003,
      "loss": 4.2382,
      "step": 2511
    },
    {
      "epoch": 0.02512,
      "grad_norm": 0.6083471860257056,
      "learning_rate": 0.003,
      "loss": 4.2155,
      "step": 2512
    },
    {
      "epoch": 0.02513,
      "grad_norm": 0.6298329580111398,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2513
    },
    {
      "epoch": 0.02514,
      "grad_norm": 0.5557311367958748,
      "learning_rate": 0.003,
      "loss": 4.1968,
      "step": 2514
    },
    {
      "epoch": 0.02515,
      "grad_norm": 0.5172156322515098,
      "learning_rate": 0.003,
      "loss": 4.1861,
      "step": 2515
    },
    {
      "epoch": 0.02516,
      "grad_norm": 0.458365154058261,
      "learning_rate": 0.003,
      "loss": 4.2042,
      "step": 2516
    },
    {
      "epoch": 0.02517,
      "grad_norm": 0.47236102620897663,
      "learning_rate": 0.003,
      "loss": 4.2043,
      "step": 2517
    },
    {
      "epoch": 0.02518,
      "grad_norm": 0.3986825790599497,
      "learning_rate": 0.003,
      "loss": 4.2114,
      "step": 2518
    },
    {
      "epoch": 0.02519,
      "grad_norm": 0.402047193379362,
      "learning_rate": 0.003,
      "loss": 4.19,
      "step": 2519
    },
    {
      "epoch": 0.0252,
      "grad_norm": 0.3916406367114395,
      "learning_rate": 0.003,
      "loss": 4.1997,
      "step": 2520
    },
    {
      "epoch": 0.02521,
      "grad_norm": 0.41614255260417965,
      "learning_rate": 0.003,
      "loss": 4.196,
      "step": 2521
    },
    {
      "epoch": 0.02522,
      "grad_norm": 0.39122808601170617,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 2522
    },
    {
      "epoch": 0.02523,
      "grad_norm": 0.32265648253713153,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 2523
    },
    {
      "epoch": 0.02524,
      "grad_norm": 0.36648634510918254,
      "learning_rate": 0.003,
      "loss": 4.2042,
      "step": 2524
    },
    {
      "epoch": 0.02525,
      "grad_norm": 0.36515875028981976,
      "learning_rate": 0.003,
      "loss": 4.2171,
      "step": 2525
    },
    {
      "epoch": 0.02526,
      "grad_norm": 0.3879027742116084,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 2526
    },
    {
      "epoch": 0.02527,
      "grad_norm": 0.39717009730244557,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 2527
    },
    {
      "epoch": 0.02528,
      "grad_norm": 0.4631846026712896,
      "learning_rate": 0.003,
      "loss": 4.2219,
      "step": 2528
    },
    {
      "epoch": 0.02529,
      "grad_norm": 0.5301783594747518,
      "learning_rate": 0.003,
      "loss": 4.1932,
      "step": 2529
    },
    {
      "epoch": 0.0253,
      "grad_norm": 0.5779986723238283,
      "learning_rate": 0.003,
      "loss": 4.2154,
      "step": 2530
    },
    {
      "epoch": 0.02531,
      "grad_norm": 0.5503995974353456,
      "learning_rate": 0.003,
      "loss": 4.2161,
      "step": 2531
    },
    {
      "epoch": 0.02532,
      "grad_norm": 0.5718882840372597,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 2532
    },
    {
      "epoch": 0.02533,
      "grad_norm": 0.6687455514677941,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2533
    },
    {
      "epoch": 0.02534,
      "grad_norm": 0.6203399011184582,
      "learning_rate": 0.003,
      "loss": 4.2081,
      "step": 2534
    },
    {
      "epoch": 0.02535,
      "grad_norm": 0.7566323425219436,
      "learning_rate": 0.003,
      "loss": 4.205,
      "step": 2535
    },
    {
      "epoch": 0.02536,
      "grad_norm": 1.072654720911659,
      "learning_rate": 0.003,
      "loss": 4.2219,
      "step": 2536
    },
    {
      "epoch": 0.02537,
      "grad_norm": 1.0281329924391467,
      "learning_rate": 0.003,
      "loss": 4.2253,
      "step": 2537
    },
    {
      "epoch": 0.02538,
      "grad_norm": 0.9199583773276817,
      "learning_rate": 0.003,
      "loss": 4.2009,
      "step": 2538
    },
    {
      "epoch": 0.02539,
      "grad_norm": 0.8211807842712615,
      "learning_rate": 0.003,
      "loss": 4.2563,
      "step": 2539
    },
    {
      "epoch": 0.0254,
      "grad_norm": 0.8400545711442372,
      "learning_rate": 0.003,
      "loss": 4.2509,
      "step": 2540
    },
    {
      "epoch": 0.02541,
      "grad_norm": 0.820158631703877,
      "learning_rate": 0.003,
      "loss": 4.2399,
      "step": 2541
    },
    {
      "epoch": 0.02542,
      "grad_norm": 0.7974385353772132,
      "learning_rate": 0.003,
      "loss": 4.2171,
      "step": 2542
    },
    {
      "epoch": 0.02543,
      "grad_norm": 0.8594194387358546,
      "learning_rate": 0.003,
      "loss": 4.2551,
      "step": 2543
    },
    {
      "epoch": 0.02544,
      "grad_norm": 0.7910126968684995,
      "learning_rate": 0.003,
      "loss": 4.244,
      "step": 2544
    },
    {
      "epoch": 0.02545,
      "grad_norm": 0.6829853653468553,
      "learning_rate": 0.003,
      "loss": 4.216,
      "step": 2545
    },
    {
      "epoch": 0.02546,
      "grad_norm": 0.7120189223135196,
      "learning_rate": 0.003,
      "loss": 4.2304,
      "step": 2546
    },
    {
      "epoch": 0.02547,
      "grad_norm": 0.6042723068123944,
      "learning_rate": 0.003,
      "loss": 4.2606,
      "step": 2547
    },
    {
      "epoch": 0.02548,
      "grad_norm": 0.6298594843243801,
      "learning_rate": 0.003,
      "loss": 4.2602,
      "step": 2548
    },
    {
      "epoch": 0.02549,
      "grad_norm": 0.5999904649113489,
      "learning_rate": 0.003,
      "loss": 4.2469,
      "step": 2549
    },
    {
      "epoch": 0.0255,
      "grad_norm": 0.6271074405015903,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2550
    },
    {
      "epoch": 0.02551,
      "grad_norm": 0.6199575999348788,
      "learning_rate": 0.003,
      "loss": 4.2396,
      "step": 2551
    },
    {
      "epoch": 0.02552,
      "grad_norm": 0.6195762543397098,
      "learning_rate": 0.003,
      "loss": 4.2076,
      "step": 2552
    },
    {
      "epoch": 0.02553,
      "grad_norm": 0.6695668966322667,
      "learning_rate": 0.003,
      "loss": 4.2746,
      "step": 2553
    },
    {
      "epoch": 0.02554,
      "grad_norm": 0.7359151892369767,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2554
    },
    {
      "epoch": 0.02555,
      "grad_norm": 0.7444694590769985,
      "learning_rate": 0.003,
      "loss": 4.2187,
      "step": 2555
    },
    {
      "epoch": 0.02556,
      "grad_norm": 0.5563951149157678,
      "learning_rate": 0.003,
      "loss": 4.2212,
      "step": 2556
    },
    {
      "epoch": 0.02557,
      "grad_norm": 0.463326630335006,
      "learning_rate": 0.003,
      "loss": 4.2204,
      "step": 2557
    },
    {
      "epoch": 0.02558,
      "grad_norm": 0.4775061885903974,
      "learning_rate": 0.003,
      "loss": 4.2352,
      "step": 2558
    },
    {
      "epoch": 0.02559,
      "grad_norm": 0.45507178814749144,
      "learning_rate": 0.003,
      "loss": 4.229,
      "step": 2559
    },
    {
      "epoch": 0.0256,
      "grad_norm": 0.44496507541292724,
      "learning_rate": 0.003,
      "loss": 4.202,
      "step": 2560
    },
    {
      "epoch": 0.02561,
      "grad_norm": 0.3670426288233485,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 2561
    },
    {
      "epoch": 0.02562,
      "grad_norm": 0.38316899243773206,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 2562
    },
    {
      "epoch": 0.02563,
      "grad_norm": 0.3516799307048104,
      "learning_rate": 0.003,
      "loss": 4.1944,
      "step": 2563
    },
    {
      "epoch": 0.02564,
      "grad_norm": 0.3439157566474706,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2564
    },
    {
      "epoch": 0.02565,
      "grad_norm": 0.3530781028336598,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 2565
    },
    {
      "epoch": 0.02566,
      "grad_norm": 0.3535153102961858,
      "learning_rate": 0.003,
      "loss": 4.2125,
      "step": 2566
    },
    {
      "epoch": 0.02567,
      "grad_norm": 0.41816993390160107,
      "learning_rate": 0.003,
      "loss": 4.2115,
      "step": 2567
    },
    {
      "epoch": 0.02568,
      "grad_norm": 0.4566608866652098,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2568
    },
    {
      "epoch": 0.02569,
      "grad_norm": 0.5020275746642583,
      "learning_rate": 0.003,
      "loss": 4.1895,
      "step": 2569
    },
    {
      "epoch": 0.0257,
      "grad_norm": 0.45491911821241204,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 2570
    },
    {
      "epoch": 0.02571,
      "grad_norm": 0.4133535089644209,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2571
    },
    {
      "epoch": 0.02572,
      "grad_norm": 0.4042168971597752,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 2572
    },
    {
      "epoch": 0.02573,
      "grad_norm": 0.3823197589786481,
      "learning_rate": 0.003,
      "loss": 4.1933,
      "step": 2573
    },
    {
      "epoch": 0.02574,
      "grad_norm": 0.39218587765755697,
      "learning_rate": 0.003,
      "loss": 4.2238,
      "step": 2574
    },
    {
      "epoch": 0.02575,
      "grad_norm": 0.3438035712553242,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 2575
    },
    {
      "epoch": 0.02576,
      "grad_norm": 0.3842191419529875,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 2576
    },
    {
      "epoch": 0.02577,
      "grad_norm": 0.42583222296628487,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 2577
    },
    {
      "epoch": 0.02578,
      "grad_norm": 0.42273441184052274,
      "learning_rate": 0.003,
      "loss": 4.1902,
      "step": 2578
    },
    {
      "epoch": 0.02579,
      "grad_norm": 0.46892691434613215,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 2579
    },
    {
      "epoch": 0.0258,
      "grad_norm": 0.5709292370942015,
      "learning_rate": 0.003,
      "loss": 4.199,
      "step": 2580
    },
    {
      "epoch": 0.02581,
      "grad_norm": 0.7127132853972009,
      "learning_rate": 0.003,
      "loss": 4.2278,
      "step": 2581
    },
    {
      "epoch": 0.02582,
      "grad_norm": 0.8754198005529624,
      "learning_rate": 0.003,
      "loss": 4.2358,
      "step": 2582
    },
    {
      "epoch": 0.02583,
      "grad_norm": 0.8240175300819178,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2583
    },
    {
      "epoch": 0.02584,
      "grad_norm": 0.7842058343487789,
      "learning_rate": 0.003,
      "loss": 4.2267,
      "step": 2584
    },
    {
      "epoch": 0.02585,
      "grad_norm": 0.8326334504065149,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 2585
    },
    {
      "epoch": 0.02586,
      "grad_norm": 0.7754616837582553,
      "learning_rate": 0.003,
      "loss": 4.2322,
      "step": 2586
    },
    {
      "epoch": 0.02587,
      "grad_norm": 0.6511240448731711,
      "learning_rate": 0.003,
      "loss": 4.2206,
      "step": 2587
    },
    {
      "epoch": 0.02588,
      "grad_norm": 0.6418349994998516,
      "learning_rate": 0.003,
      "loss": 4.2155,
      "step": 2588
    },
    {
      "epoch": 0.02589,
      "grad_norm": 0.6950697428484538,
      "learning_rate": 0.003,
      "loss": 4.2254,
      "step": 2589
    },
    {
      "epoch": 0.0259,
      "grad_norm": 0.7453523656170742,
      "learning_rate": 0.003,
      "loss": 4.2359,
      "step": 2590
    },
    {
      "epoch": 0.02591,
      "grad_norm": 0.6556652428839654,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 2591
    },
    {
      "epoch": 0.02592,
      "grad_norm": 0.5716997523979154,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2592
    },
    {
      "epoch": 0.02593,
      "grad_norm": 0.5895740446059741,
      "learning_rate": 0.003,
      "loss": 4.2131,
      "step": 2593
    },
    {
      "epoch": 0.02594,
      "grad_norm": 0.6861270884763415,
      "learning_rate": 0.003,
      "loss": 4.2385,
      "step": 2594
    },
    {
      "epoch": 0.02595,
      "grad_norm": 0.6039685752573255,
      "learning_rate": 0.003,
      "loss": 4.1847,
      "step": 2595
    },
    {
      "epoch": 0.02596,
      "grad_norm": 0.5737417856631055,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2596
    },
    {
      "epoch": 0.02597,
      "grad_norm": 0.5807882188149026,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 2597
    },
    {
      "epoch": 0.02598,
      "grad_norm": 0.5478759076616263,
      "learning_rate": 0.003,
      "loss": 4.2296,
      "step": 2598
    },
    {
      "epoch": 0.02599,
      "grad_norm": 0.5274069809990996,
      "learning_rate": 0.003,
      "loss": 4.2218,
      "step": 2599
    },
    {
      "epoch": 0.026,
      "grad_norm": 0.5489291707939286,
      "learning_rate": 0.003,
      "loss": 4.2225,
      "step": 2600
    },
    {
      "epoch": 0.02601,
      "grad_norm": 0.5075235418141223,
      "learning_rate": 0.003,
      "loss": 4.2266,
      "step": 2601
    },
    {
      "epoch": 0.02602,
      "grad_norm": 0.5092543256680011,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2602
    },
    {
      "epoch": 0.02603,
      "grad_norm": 0.48420449669893695,
      "learning_rate": 0.003,
      "loss": 4.1912,
      "step": 2603
    },
    {
      "epoch": 0.02604,
      "grad_norm": 0.4845552481378335,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 2604
    },
    {
      "epoch": 0.02605,
      "grad_norm": 0.522772142438526,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2605
    },
    {
      "epoch": 0.02606,
      "grad_norm": 0.49986531502998316,
      "learning_rate": 0.003,
      "loss": 4.2158,
      "step": 2606
    },
    {
      "epoch": 0.02607,
      "grad_norm": 0.4992114205626333,
      "learning_rate": 0.003,
      "loss": 4.2071,
      "step": 2607
    },
    {
      "epoch": 0.02608,
      "grad_norm": 0.516906776877324,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 2608
    },
    {
      "epoch": 0.02609,
      "grad_norm": 0.5823377881437489,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 2609
    },
    {
      "epoch": 0.0261,
      "grad_norm": 0.6432677415233223,
      "learning_rate": 0.003,
      "loss": 4.1868,
      "step": 2610
    },
    {
      "epoch": 0.02611,
      "grad_norm": 0.6818797064218945,
      "learning_rate": 0.003,
      "loss": 4.1789,
      "step": 2611
    },
    {
      "epoch": 0.02612,
      "grad_norm": 0.6622646028172395,
      "learning_rate": 0.003,
      "loss": 4.2188,
      "step": 2612
    },
    {
      "epoch": 0.02613,
      "grad_norm": 0.6111717401030455,
      "learning_rate": 0.003,
      "loss": 4.2101,
      "step": 2613
    },
    {
      "epoch": 0.02614,
      "grad_norm": 0.5350928499489108,
      "learning_rate": 0.003,
      "loss": 4.1863,
      "step": 2614
    },
    {
      "epoch": 0.02615,
      "grad_norm": 0.4811589002895373,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 2615
    },
    {
      "epoch": 0.02616,
      "grad_norm": 0.4870958333875311,
      "learning_rate": 0.003,
      "loss": 4.1868,
      "step": 2616
    },
    {
      "epoch": 0.02617,
      "grad_norm": 0.48293626841319026,
      "learning_rate": 0.003,
      "loss": 4.1753,
      "step": 2617
    },
    {
      "epoch": 0.02618,
      "grad_norm": 0.5217244185547227,
      "learning_rate": 0.003,
      "loss": 4.2041,
      "step": 2618
    },
    {
      "epoch": 0.02619,
      "grad_norm": 0.5234547384625777,
      "learning_rate": 0.003,
      "loss": 4.2179,
      "step": 2619
    },
    {
      "epoch": 0.0262,
      "grad_norm": 0.4734488648814552,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 2620
    },
    {
      "epoch": 0.02621,
      "grad_norm": 0.4072710001568894,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 2621
    },
    {
      "epoch": 0.02622,
      "grad_norm": 0.3670914624450667,
      "learning_rate": 0.003,
      "loss": 4.1997,
      "step": 2622
    },
    {
      "epoch": 0.02623,
      "grad_norm": 0.3907209927814348,
      "learning_rate": 0.003,
      "loss": 4.2044,
      "step": 2623
    },
    {
      "epoch": 0.02624,
      "grad_norm": 0.4112397678843122,
      "learning_rate": 0.003,
      "loss": 4.2395,
      "step": 2624
    },
    {
      "epoch": 0.02625,
      "grad_norm": 0.4373318792363482,
      "learning_rate": 0.003,
      "loss": 4.2339,
      "step": 2625
    },
    {
      "epoch": 0.02626,
      "grad_norm": 0.43765342876243357,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 2626
    },
    {
      "epoch": 0.02627,
      "grad_norm": 0.44112134115596174,
      "learning_rate": 0.003,
      "loss": 4.1845,
      "step": 2627
    },
    {
      "epoch": 0.02628,
      "grad_norm": 0.5820511792087758,
      "learning_rate": 0.003,
      "loss": 4.2196,
      "step": 2628
    },
    {
      "epoch": 0.02629,
      "grad_norm": 0.6987525856490595,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 2629
    },
    {
      "epoch": 0.0263,
      "grad_norm": 0.8227125293487279,
      "learning_rate": 0.003,
      "loss": 4.2044,
      "step": 2630
    },
    {
      "epoch": 0.02631,
      "grad_norm": 0.8558111239802438,
      "learning_rate": 0.003,
      "loss": 4.2316,
      "step": 2631
    },
    {
      "epoch": 0.02632,
      "grad_norm": 0.7416342423752187,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 2632
    },
    {
      "epoch": 0.02633,
      "grad_norm": 0.7123350797241652,
      "learning_rate": 0.003,
      "loss": 4.2064,
      "step": 2633
    },
    {
      "epoch": 0.02634,
      "grad_norm": 0.7492282422336083,
      "learning_rate": 0.003,
      "loss": 4.2291,
      "step": 2634
    },
    {
      "epoch": 0.02635,
      "grad_norm": 0.9073437715289041,
      "learning_rate": 0.003,
      "loss": 4.2382,
      "step": 2635
    },
    {
      "epoch": 0.02636,
      "grad_norm": 0.8986238889024204,
      "learning_rate": 0.003,
      "loss": 4.2298,
      "step": 2636
    },
    {
      "epoch": 0.02637,
      "grad_norm": 0.803485832317335,
      "learning_rate": 0.003,
      "loss": 4.2558,
      "step": 2637
    },
    {
      "epoch": 0.02638,
      "grad_norm": 0.790197716438678,
      "learning_rate": 0.003,
      "loss": 4.2252,
      "step": 2638
    },
    {
      "epoch": 0.02639,
      "grad_norm": 0.6490246383904035,
      "learning_rate": 0.003,
      "loss": 4.2065,
      "step": 2639
    },
    {
      "epoch": 0.0264,
      "grad_norm": 0.6271932479026731,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 2640
    },
    {
      "epoch": 0.02641,
      "grad_norm": 0.5216470184091793,
      "learning_rate": 0.003,
      "loss": 4.212,
      "step": 2641
    },
    {
      "epoch": 0.02642,
      "grad_norm": 0.5340137157215237,
      "learning_rate": 0.003,
      "loss": 4.2208,
      "step": 2642
    },
    {
      "epoch": 0.02643,
      "grad_norm": 0.48600536528948673,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2643
    },
    {
      "epoch": 0.02644,
      "grad_norm": 0.4913843517241883,
      "learning_rate": 0.003,
      "loss": 4.2177,
      "step": 2644
    },
    {
      "epoch": 0.02645,
      "grad_norm": 0.5723820549842618,
      "learning_rate": 0.003,
      "loss": 4.2298,
      "step": 2645
    },
    {
      "epoch": 0.02646,
      "grad_norm": 0.6080093721456342,
      "learning_rate": 0.003,
      "loss": 4.1979,
      "step": 2646
    },
    {
      "epoch": 0.02647,
      "grad_norm": 0.6222106205330706,
      "learning_rate": 0.003,
      "loss": 4.2212,
      "step": 2647
    },
    {
      "epoch": 0.02648,
      "grad_norm": 0.5217769121416035,
      "learning_rate": 0.003,
      "loss": 4.2349,
      "step": 2648
    },
    {
      "epoch": 0.02649,
      "grad_norm": 0.4227997429869736,
      "learning_rate": 0.003,
      "loss": 4.2183,
      "step": 2649
    },
    {
      "epoch": 0.0265,
      "grad_norm": 0.47327414101860077,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 2650
    },
    {
      "epoch": 0.02651,
      "grad_norm": 0.5099068125596607,
      "learning_rate": 0.003,
      "loss": 4.1959,
      "step": 2651
    },
    {
      "epoch": 0.02652,
      "grad_norm": 0.60276863782341,
      "learning_rate": 0.003,
      "loss": 4.2278,
      "step": 2652
    },
    {
      "epoch": 0.02653,
      "grad_norm": 0.6740196425733591,
      "learning_rate": 0.003,
      "loss": 4.1869,
      "step": 2653
    },
    {
      "epoch": 0.02654,
      "grad_norm": 0.6718723086436392,
      "learning_rate": 0.003,
      "loss": 4.2008,
      "step": 2654
    },
    {
      "epoch": 0.02655,
      "grad_norm": 0.6985692134702927,
      "learning_rate": 0.003,
      "loss": 4.21,
      "step": 2655
    },
    {
      "epoch": 0.02656,
      "grad_norm": 0.7105425653160966,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 2656
    },
    {
      "epoch": 0.02657,
      "grad_norm": 0.6933600976197403,
      "learning_rate": 0.003,
      "loss": 4.2156,
      "step": 2657
    },
    {
      "epoch": 0.02658,
      "grad_norm": 0.6544303516003507,
      "learning_rate": 0.003,
      "loss": 4.2197,
      "step": 2658
    },
    {
      "epoch": 0.02659,
      "grad_norm": 0.5910001636604436,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2659
    },
    {
      "epoch": 0.0266,
      "grad_norm": 0.5163250162883284,
      "learning_rate": 0.003,
      "loss": 4.1915,
      "step": 2660
    },
    {
      "epoch": 0.02661,
      "grad_norm": 0.5302434720729938,
      "learning_rate": 0.003,
      "loss": 4.2129,
      "step": 2661
    },
    {
      "epoch": 0.02662,
      "grad_norm": 0.4750913186060552,
      "learning_rate": 0.003,
      "loss": 4.1991,
      "step": 2662
    },
    {
      "epoch": 0.02663,
      "grad_norm": 0.4491722077606405,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 2663
    },
    {
      "epoch": 0.02664,
      "grad_norm": 0.42258758049488826,
      "learning_rate": 0.003,
      "loss": 4.1803,
      "step": 2664
    },
    {
      "epoch": 0.02665,
      "grad_norm": 0.41774121759742056,
      "learning_rate": 0.003,
      "loss": 4.2287,
      "step": 2665
    },
    {
      "epoch": 0.02666,
      "grad_norm": 0.43325477990837064,
      "learning_rate": 0.003,
      "loss": 4.2184,
      "step": 2666
    },
    {
      "epoch": 0.02667,
      "grad_norm": 0.4508456123093455,
      "learning_rate": 0.003,
      "loss": 4.1974,
      "step": 2667
    },
    {
      "epoch": 0.02668,
      "grad_norm": 0.46262752082913233,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 2668
    },
    {
      "epoch": 0.02669,
      "grad_norm": 0.42577584585471717,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 2669
    },
    {
      "epoch": 0.0267,
      "grad_norm": 0.404720023333166,
      "learning_rate": 0.003,
      "loss": 4.1856,
      "step": 2670
    },
    {
      "epoch": 0.02671,
      "grad_norm": 0.3942334538580407,
      "learning_rate": 0.003,
      "loss": 4.2006,
      "step": 2671
    },
    {
      "epoch": 0.02672,
      "grad_norm": 0.40993974344347783,
      "learning_rate": 0.003,
      "loss": 4.1947,
      "step": 2672
    },
    {
      "epoch": 0.02673,
      "grad_norm": 0.4616934932283336,
      "learning_rate": 0.003,
      "loss": 4.1918,
      "step": 2673
    },
    {
      "epoch": 0.02674,
      "grad_norm": 0.5514329672225634,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 2674
    },
    {
      "epoch": 0.02675,
      "grad_norm": 0.5466818103576505,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 2675
    },
    {
      "epoch": 0.02676,
      "grad_norm": 0.6935187296359256,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 2676
    },
    {
      "epoch": 0.02677,
      "grad_norm": 0.8598433315495629,
      "learning_rate": 0.003,
      "loss": 4.231,
      "step": 2677
    },
    {
      "epoch": 0.02678,
      "grad_norm": 0.8883830728999605,
      "learning_rate": 0.003,
      "loss": 4.2344,
      "step": 2678
    },
    {
      "epoch": 0.02679,
      "grad_norm": 0.6405381161924845,
      "learning_rate": 0.003,
      "loss": 4.1941,
      "step": 2679
    },
    {
      "epoch": 0.0268,
      "grad_norm": 0.6915639860426047,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2680
    },
    {
      "epoch": 0.02681,
      "grad_norm": 0.5927165760939286,
      "learning_rate": 0.003,
      "loss": 4.2278,
      "step": 2681
    },
    {
      "epoch": 0.02682,
      "grad_norm": 0.5745893208237367,
      "learning_rate": 0.003,
      "loss": 4.2044,
      "step": 2682
    },
    {
      "epoch": 0.02683,
      "grad_norm": 0.5931786049326642,
      "learning_rate": 0.003,
      "loss": 4.2269,
      "step": 2683
    },
    {
      "epoch": 0.02684,
      "grad_norm": 0.5656789702546796,
      "learning_rate": 0.003,
      "loss": 4.1996,
      "step": 2684
    },
    {
      "epoch": 0.02685,
      "grad_norm": 0.6294556799351446,
      "learning_rate": 0.003,
      "loss": 4.211,
      "step": 2685
    },
    {
      "epoch": 0.02686,
      "grad_norm": 0.5832268033312072,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 2686
    },
    {
      "epoch": 0.02687,
      "grad_norm": 0.570290226028237,
      "learning_rate": 0.003,
      "loss": 4.1996,
      "step": 2687
    },
    {
      "epoch": 0.02688,
      "grad_norm": 0.5455818208368779,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 2688
    },
    {
      "epoch": 0.02689,
      "grad_norm": 0.5599812876010125,
      "learning_rate": 0.003,
      "loss": 4.2088,
      "step": 2689
    },
    {
      "epoch": 0.0269,
      "grad_norm": 0.5474516018001153,
      "learning_rate": 0.003,
      "loss": 4.2129,
      "step": 2690
    },
    {
      "epoch": 0.02691,
      "grad_norm": 0.5566067071138427,
      "learning_rate": 0.003,
      "loss": 4.2043,
      "step": 2691
    },
    {
      "epoch": 0.02692,
      "grad_norm": 0.526509280614571,
      "learning_rate": 0.003,
      "loss": 4.2054,
      "step": 2692
    },
    {
      "epoch": 0.02693,
      "grad_norm": 0.4827516603970495,
      "learning_rate": 0.003,
      "loss": 4.218,
      "step": 2693
    },
    {
      "epoch": 0.02694,
      "grad_norm": 0.46801137768639334,
      "learning_rate": 0.003,
      "loss": 4.1783,
      "step": 2694
    },
    {
      "epoch": 0.02695,
      "grad_norm": 0.4878539547841592,
      "learning_rate": 0.003,
      "loss": 4.1935,
      "step": 2695
    },
    {
      "epoch": 0.02696,
      "grad_norm": 0.47854515525549607,
      "learning_rate": 0.003,
      "loss": 4.1948,
      "step": 2696
    },
    {
      "epoch": 0.02697,
      "grad_norm": 0.5065481594378214,
      "learning_rate": 0.003,
      "loss": 4.2065,
      "step": 2697
    },
    {
      "epoch": 0.02698,
      "grad_norm": 0.5248263427239105,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 2698
    },
    {
      "epoch": 0.02699,
      "grad_norm": 0.5277451425908661,
      "learning_rate": 0.003,
      "loss": 4.2105,
      "step": 2699
    },
    {
      "epoch": 0.027,
      "grad_norm": 0.5538062675395776,
      "learning_rate": 0.003,
      "loss": 4.1806,
      "step": 2700
    },
    {
      "epoch": 0.02701,
      "grad_norm": 0.6185989339314542,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 2701
    },
    {
      "epoch": 0.02702,
      "grad_norm": 0.5874971422385314,
      "learning_rate": 0.003,
      "loss": 4.225,
      "step": 2702
    },
    {
      "epoch": 0.02703,
      "grad_norm": 0.5463427926178012,
      "learning_rate": 0.003,
      "loss": 4.2202,
      "step": 2703
    },
    {
      "epoch": 0.02704,
      "grad_norm": 0.5672786289651166,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 2704
    },
    {
      "epoch": 0.02705,
      "grad_norm": 0.6256411921927717,
      "learning_rate": 0.003,
      "loss": 4.2194,
      "step": 2705
    },
    {
      "epoch": 0.02706,
      "grad_norm": 0.652684226752671,
      "learning_rate": 0.003,
      "loss": 4.1841,
      "step": 2706
    },
    {
      "epoch": 0.02707,
      "grad_norm": 0.6029115328911401,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 2707
    },
    {
      "epoch": 0.02708,
      "grad_norm": 0.5287197350990817,
      "learning_rate": 0.003,
      "loss": 4.2087,
      "step": 2708
    },
    {
      "epoch": 0.02709,
      "grad_norm": 0.6492025140814462,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2709
    },
    {
      "epoch": 0.0271,
      "grad_norm": 0.6051042816144401,
      "learning_rate": 0.003,
      "loss": 4.1894,
      "step": 2710
    },
    {
      "epoch": 0.02711,
      "grad_norm": 0.5747478778857192,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 2711
    },
    {
      "epoch": 0.02712,
      "grad_norm": 0.5615755138550911,
      "learning_rate": 0.003,
      "loss": 4.1868,
      "step": 2712
    },
    {
      "epoch": 0.02713,
      "grad_norm": 0.5892906666649981,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 2713
    },
    {
      "epoch": 0.02714,
      "grad_norm": 0.6013481891384199,
      "learning_rate": 0.003,
      "loss": 4.1837,
      "step": 2714
    },
    {
      "epoch": 0.02715,
      "grad_norm": 0.6656155771795913,
      "learning_rate": 0.003,
      "loss": 4.1807,
      "step": 2715
    },
    {
      "epoch": 0.02716,
      "grad_norm": 0.8931240767996229,
      "learning_rate": 0.003,
      "loss": 4.2057,
      "step": 2716
    },
    {
      "epoch": 0.02717,
      "grad_norm": 0.9885029624018516,
      "learning_rate": 0.003,
      "loss": 4.2196,
      "step": 2717
    },
    {
      "epoch": 0.02718,
      "grad_norm": 0.8754295759055237,
      "learning_rate": 0.003,
      "loss": 4.2405,
      "step": 2718
    },
    {
      "epoch": 0.02719,
      "grad_norm": 0.7774225786110023,
      "learning_rate": 0.003,
      "loss": 4.2179,
      "step": 2719
    },
    {
      "epoch": 0.0272,
      "grad_norm": 0.6491659668608393,
      "learning_rate": 0.003,
      "loss": 4.2007,
      "step": 2720
    },
    {
      "epoch": 0.02721,
      "grad_norm": 0.6343965078219608,
      "learning_rate": 0.003,
      "loss": 4.2145,
      "step": 2721
    },
    {
      "epoch": 0.02722,
      "grad_norm": 0.6251153466031798,
      "learning_rate": 0.003,
      "loss": 4.2101,
      "step": 2722
    },
    {
      "epoch": 0.02723,
      "grad_norm": 0.6748255813851252,
      "learning_rate": 0.003,
      "loss": 4.2215,
      "step": 2723
    },
    {
      "epoch": 0.02724,
      "grad_norm": 0.6864377875600585,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2724
    },
    {
      "epoch": 0.02725,
      "grad_norm": 0.657540480175467,
      "learning_rate": 0.003,
      "loss": 4.2162,
      "step": 2725
    },
    {
      "epoch": 0.02726,
      "grad_norm": 0.67217283621503,
      "learning_rate": 0.003,
      "loss": 4.1914,
      "step": 2726
    },
    {
      "epoch": 0.02727,
      "grad_norm": 0.7144546093311819,
      "learning_rate": 0.003,
      "loss": 4.2253,
      "step": 2727
    },
    {
      "epoch": 0.02728,
      "grad_norm": 0.6685628291749064,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 2728
    },
    {
      "epoch": 0.02729,
      "grad_norm": 0.6958710627910398,
      "learning_rate": 0.003,
      "loss": 4.2431,
      "step": 2729
    },
    {
      "epoch": 0.0273,
      "grad_norm": 0.7029911290820989,
      "learning_rate": 0.003,
      "loss": 4.2118,
      "step": 2730
    },
    {
      "epoch": 0.02731,
      "grad_norm": 0.7388699066611659,
      "learning_rate": 0.003,
      "loss": 4.2131,
      "step": 2731
    },
    {
      "epoch": 0.02732,
      "grad_norm": 0.6457150813790012,
      "learning_rate": 0.003,
      "loss": 4.2055,
      "step": 2732
    },
    {
      "epoch": 0.02733,
      "grad_norm": 0.42245973261823044,
      "learning_rate": 0.003,
      "loss": 4.225,
      "step": 2733
    },
    {
      "epoch": 0.02734,
      "grad_norm": 0.4568223524843039,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 2734
    },
    {
      "epoch": 0.02735,
      "grad_norm": 0.46161311769792607,
      "learning_rate": 0.003,
      "loss": 4.2213,
      "step": 2735
    },
    {
      "epoch": 0.02736,
      "grad_norm": 0.43828909949858114,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2736
    },
    {
      "epoch": 0.02737,
      "grad_norm": 0.44381003049579976,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 2737
    },
    {
      "epoch": 0.02738,
      "grad_norm": 0.40770566309449235,
      "learning_rate": 0.003,
      "loss": 4.207,
      "step": 2738
    },
    {
      "epoch": 0.02739,
      "grad_norm": 0.3980139287869944,
      "learning_rate": 0.003,
      "loss": 4.1947,
      "step": 2739
    },
    {
      "epoch": 0.0274,
      "grad_norm": 0.4103938581926708,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2740
    },
    {
      "epoch": 0.02741,
      "grad_norm": 0.3977644284371821,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 2741
    },
    {
      "epoch": 0.02742,
      "grad_norm": 0.3910236660359437,
      "learning_rate": 0.003,
      "loss": 4.1829,
      "step": 2742
    },
    {
      "epoch": 0.02743,
      "grad_norm": 0.37057712717675134,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 2743
    },
    {
      "epoch": 0.02744,
      "grad_norm": 0.39171990335728823,
      "learning_rate": 0.003,
      "loss": 4.1862,
      "step": 2744
    },
    {
      "epoch": 0.02745,
      "grad_norm": 0.3931440185763024,
      "learning_rate": 0.003,
      "loss": 4.1978,
      "step": 2745
    },
    {
      "epoch": 0.02746,
      "grad_norm": 0.4493443882352147,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 2746
    },
    {
      "epoch": 0.02747,
      "grad_norm": 0.5239427386961047,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2747
    },
    {
      "epoch": 0.02748,
      "grad_norm": 0.48560097013750286,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 2748
    },
    {
      "epoch": 0.02749,
      "grad_norm": 0.4388250470896872,
      "learning_rate": 0.003,
      "loss": 4.1614,
      "step": 2749
    },
    {
      "epoch": 0.0275,
      "grad_norm": 0.425974848346712,
      "learning_rate": 0.003,
      "loss": 4.1976,
      "step": 2750
    },
    {
      "epoch": 0.02751,
      "grad_norm": 0.4487191491618812,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 2751
    },
    {
      "epoch": 0.02752,
      "grad_norm": 0.5262082176933003,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2752
    },
    {
      "epoch": 0.02753,
      "grad_norm": 0.5624164107816553,
      "learning_rate": 0.003,
      "loss": 4.1806,
      "step": 2753
    },
    {
      "epoch": 0.02754,
      "grad_norm": 0.6290188699871871,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 2754
    },
    {
      "epoch": 0.02755,
      "grad_norm": 0.6718718115523771,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 2755
    },
    {
      "epoch": 0.02756,
      "grad_norm": 0.689725298181599,
      "learning_rate": 0.003,
      "loss": 4.2042,
      "step": 2756
    },
    {
      "epoch": 0.02757,
      "grad_norm": 0.7456661709777471,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 2757
    },
    {
      "epoch": 0.02758,
      "grad_norm": 0.7434755613308037,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 2758
    },
    {
      "epoch": 0.02759,
      "grad_norm": 0.6285033589958848,
      "learning_rate": 0.003,
      "loss": 4.1982,
      "step": 2759
    },
    {
      "epoch": 0.0276,
      "grad_norm": 0.6775730198112929,
      "learning_rate": 0.003,
      "loss": 4.2094,
      "step": 2760
    },
    {
      "epoch": 0.02761,
      "grad_norm": 0.6910341516486844,
      "learning_rate": 0.003,
      "loss": 4.2268,
      "step": 2761
    },
    {
      "epoch": 0.02762,
      "grad_norm": 0.6650852460123947,
      "learning_rate": 0.003,
      "loss": 4.1938,
      "step": 2762
    },
    {
      "epoch": 0.02763,
      "grad_norm": 0.6011271334026552,
      "learning_rate": 0.003,
      "loss": 4.2029,
      "step": 2763
    },
    {
      "epoch": 0.02764,
      "grad_norm": 0.5886973411048619,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2764
    },
    {
      "epoch": 0.02765,
      "grad_norm": 0.6682064585302908,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2765
    },
    {
      "epoch": 0.02766,
      "grad_norm": 0.7004868212188018,
      "learning_rate": 0.003,
      "loss": 4.1783,
      "step": 2766
    },
    {
      "epoch": 0.02767,
      "grad_norm": 0.6391373162549133,
      "learning_rate": 0.003,
      "loss": 4.2317,
      "step": 2767
    },
    {
      "epoch": 0.02768,
      "grad_norm": 0.5711886608834656,
      "learning_rate": 0.003,
      "loss": 4.1613,
      "step": 2768
    },
    {
      "epoch": 0.02769,
      "grad_norm": 0.6028604983542873,
      "learning_rate": 0.003,
      "loss": 4.2212,
      "step": 2769
    },
    {
      "epoch": 0.0277,
      "grad_norm": 0.5836613470106289,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 2770
    },
    {
      "epoch": 0.02771,
      "grad_norm": 0.5619610082417591,
      "learning_rate": 0.003,
      "loss": 4.174,
      "step": 2771
    },
    {
      "epoch": 0.02772,
      "grad_norm": 0.55339038708748,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 2772
    },
    {
      "epoch": 0.02773,
      "grad_norm": 0.5034136342234373,
      "learning_rate": 0.003,
      "loss": 4.2003,
      "step": 2773
    },
    {
      "epoch": 0.02774,
      "grad_norm": 0.5193760267646912,
      "learning_rate": 0.003,
      "loss": 4.2225,
      "step": 2774
    },
    {
      "epoch": 0.02775,
      "grad_norm": 0.5211288768219952,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 2775
    },
    {
      "epoch": 0.02776,
      "grad_norm": 0.4784203907496932,
      "learning_rate": 0.003,
      "loss": 4.1844,
      "step": 2776
    },
    {
      "epoch": 0.02777,
      "grad_norm": 0.48277039376268643,
      "learning_rate": 0.003,
      "loss": 4.1625,
      "step": 2777
    },
    {
      "epoch": 0.02778,
      "grad_norm": 0.5520225977650862,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 2778
    },
    {
      "epoch": 0.02779,
      "grad_norm": 0.676790078006391,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 2779
    },
    {
      "epoch": 0.0278,
      "grad_norm": 0.7848726112221234,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 2780
    },
    {
      "epoch": 0.02781,
      "grad_norm": 0.7347482785122091,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 2781
    },
    {
      "epoch": 0.02782,
      "grad_norm": 0.6054417831566177,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 2782
    },
    {
      "epoch": 0.02783,
      "grad_norm": 0.5849343545068603,
      "learning_rate": 0.003,
      "loss": 4.1925,
      "step": 2783
    },
    {
      "epoch": 0.02784,
      "grad_norm": 0.6041156935596753,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 2784
    },
    {
      "epoch": 0.02785,
      "grad_norm": 0.595867835436581,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 2785
    },
    {
      "epoch": 0.02786,
      "grad_norm": 0.6928930555141227,
      "learning_rate": 0.003,
      "loss": 4.2052,
      "step": 2786
    },
    {
      "epoch": 0.02787,
      "grad_norm": 0.7406479460311315,
      "learning_rate": 0.003,
      "loss": 4.2115,
      "step": 2787
    },
    {
      "epoch": 0.02788,
      "grad_norm": 0.6940976950244777,
      "learning_rate": 0.003,
      "loss": 4.1982,
      "step": 2788
    },
    {
      "epoch": 0.02789,
      "grad_norm": 0.6273541949163579,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 2789
    },
    {
      "epoch": 0.0279,
      "grad_norm": 0.5979480136249415,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 2790
    },
    {
      "epoch": 0.02791,
      "grad_norm": 0.5594196614238619,
      "learning_rate": 0.003,
      "loss": 4.1864,
      "step": 2791
    },
    {
      "epoch": 0.02792,
      "grad_norm": 0.5206581117733323,
      "learning_rate": 0.003,
      "loss": 4.2009,
      "step": 2792
    },
    {
      "epoch": 0.02793,
      "grad_norm": 0.5275957045486279,
      "learning_rate": 0.003,
      "loss": 4.1994,
      "step": 2793
    },
    {
      "epoch": 0.02794,
      "grad_norm": 0.524155250681782,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 2794
    },
    {
      "epoch": 0.02795,
      "grad_norm": 0.5721629122704371,
      "learning_rate": 0.003,
      "loss": 4.2082,
      "step": 2795
    },
    {
      "epoch": 0.02796,
      "grad_norm": 0.6009605504824468,
      "learning_rate": 0.003,
      "loss": 4.2021,
      "step": 2796
    },
    {
      "epoch": 0.02797,
      "grad_norm": 0.5762272128929422,
      "learning_rate": 0.003,
      "loss": 4.2131,
      "step": 2797
    },
    {
      "epoch": 0.02798,
      "grad_norm": 0.5678013400030233,
      "learning_rate": 0.003,
      "loss": 4.2087,
      "step": 2798
    },
    {
      "epoch": 0.02799,
      "grad_norm": 0.5466966091962687,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2799
    },
    {
      "epoch": 0.028,
      "grad_norm": 0.5426896151230969,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2800
    },
    {
      "epoch": 0.02801,
      "grad_norm": 0.5309488273760117,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 2801
    },
    {
      "epoch": 0.02802,
      "grad_norm": 0.5742220089169363,
      "learning_rate": 0.003,
      "loss": 4.1824,
      "step": 2802
    },
    {
      "epoch": 0.02803,
      "grad_norm": 0.7893252753789815,
      "learning_rate": 0.003,
      "loss": 4.2101,
      "step": 2803
    },
    {
      "epoch": 0.02804,
      "grad_norm": 1.052487841789281,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 2804
    },
    {
      "epoch": 0.02805,
      "grad_norm": 0.8655535125529751,
      "learning_rate": 0.003,
      "loss": 4.2346,
      "step": 2805
    },
    {
      "epoch": 0.02806,
      "grad_norm": 0.6475434652146125,
      "learning_rate": 0.003,
      "loss": 4.202,
      "step": 2806
    },
    {
      "epoch": 0.02807,
      "grad_norm": 0.5991110542030375,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 2807
    },
    {
      "epoch": 0.02808,
      "grad_norm": 0.5007983686795169,
      "learning_rate": 0.003,
      "loss": 4.1818,
      "step": 2808
    },
    {
      "epoch": 0.02809,
      "grad_norm": 0.49095808154292164,
      "learning_rate": 0.003,
      "loss": 4.2217,
      "step": 2809
    },
    {
      "epoch": 0.0281,
      "grad_norm": 0.4819118161545627,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 2810
    },
    {
      "epoch": 0.02811,
      "grad_norm": 0.44009677007510295,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 2811
    },
    {
      "epoch": 0.02812,
      "grad_norm": 0.4080215487522985,
      "learning_rate": 0.003,
      "loss": 4.1953,
      "step": 2812
    },
    {
      "epoch": 0.02813,
      "grad_norm": 0.3937509668583135,
      "learning_rate": 0.003,
      "loss": 4.1839,
      "step": 2813
    },
    {
      "epoch": 0.02814,
      "grad_norm": 0.34050782014520087,
      "learning_rate": 0.003,
      "loss": 4.1814,
      "step": 2814
    },
    {
      "epoch": 0.02815,
      "grad_norm": 0.32268087418962427,
      "learning_rate": 0.003,
      "loss": 4.2181,
      "step": 2815
    },
    {
      "epoch": 0.02816,
      "grad_norm": 0.3255642062500674,
      "learning_rate": 0.003,
      "loss": 4.1944,
      "step": 2816
    },
    {
      "epoch": 0.02817,
      "grad_norm": 0.3381409922347926,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 2817
    },
    {
      "epoch": 0.02818,
      "grad_norm": 0.38168796766705765,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 2818
    },
    {
      "epoch": 0.02819,
      "grad_norm": 0.4325061454583649,
      "learning_rate": 0.003,
      "loss": 4.1913,
      "step": 2819
    },
    {
      "epoch": 0.0282,
      "grad_norm": 0.4532493695937539,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 2820
    },
    {
      "epoch": 0.02821,
      "grad_norm": 0.47151197099161557,
      "learning_rate": 0.003,
      "loss": 4.2029,
      "step": 2821
    },
    {
      "epoch": 0.02822,
      "grad_norm": 0.550967405378687,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 2822
    },
    {
      "epoch": 0.02823,
      "grad_norm": 0.5888233523434347,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2823
    },
    {
      "epoch": 0.02824,
      "grad_norm": 0.5616734570991271,
      "learning_rate": 0.003,
      "loss": 4.1907,
      "step": 2824
    },
    {
      "epoch": 0.02825,
      "grad_norm": 0.4685382550291958,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 2825
    },
    {
      "epoch": 0.02826,
      "grad_norm": 0.3984745923734355,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 2826
    },
    {
      "epoch": 0.02827,
      "grad_norm": 0.4466687377830192,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 2827
    },
    {
      "epoch": 0.02828,
      "grad_norm": 0.48090906385724086,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2828
    },
    {
      "epoch": 0.02829,
      "grad_norm": 0.4887994839877233,
      "learning_rate": 0.003,
      "loss": 4.2162,
      "step": 2829
    },
    {
      "epoch": 0.0283,
      "grad_norm": 0.5659608141025803,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 2830
    },
    {
      "epoch": 0.02831,
      "grad_norm": 0.6332182296297022,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 2831
    },
    {
      "epoch": 0.02832,
      "grad_norm": 0.7157897208149597,
      "learning_rate": 0.003,
      "loss": 4.2049,
      "step": 2832
    },
    {
      "epoch": 0.02833,
      "grad_norm": 0.7488210186844626,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 2833
    },
    {
      "epoch": 0.02834,
      "grad_norm": 0.721287579139196,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 2834
    },
    {
      "epoch": 0.02835,
      "grad_norm": 0.7107105075536025,
      "learning_rate": 0.003,
      "loss": 4.1873,
      "step": 2835
    },
    {
      "epoch": 0.02836,
      "grad_norm": 0.7041301874694449,
      "learning_rate": 0.003,
      "loss": 4.2059,
      "step": 2836
    },
    {
      "epoch": 0.02837,
      "grad_norm": 0.7504397142936561,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 2837
    },
    {
      "epoch": 0.02838,
      "grad_norm": 0.7162503617742563,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 2838
    },
    {
      "epoch": 0.02839,
      "grad_norm": 0.6629058755892886,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 2839
    },
    {
      "epoch": 0.0284,
      "grad_norm": 0.8045463625357235,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 2840
    },
    {
      "epoch": 0.02841,
      "grad_norm": 0.8128195799944198,
      "learning_rate": 0.003,
      "loss": 4.2167,
      "step": 2841
    },
    {
      "epoch": 0.02842,
      "grad_norm": 0.9021793870071328,
      "learning_rate": 0.003,
      "loss": 4.201,
      "step": 2842
    },
    {
      "epoch": 0.02843,
      "grad_norm": 0.9569629998728558,
      "learning_rate": 0.003,
      "loss": 4.2417,
      "step": 2843
    },
    {
      "epoch": 0.02844,
      "grad_norm": 0.9483564800535025,
      "learning_rate": 0.003,
      "loss": 4.2425,
      "step": 2844
    },
    {
      "epoch": 0.02845,
      "grad_norm": 0.9913358728662397,
      "learning_rate": 0.003,
      "loss": 4.2312,
      "step": 2845
    },
    {
      "epoch": 0.02846,
      "grad_norm": 1.063054241841927,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 2846
    },
    {
      "epoch": 0.02847,
      "grad_norm": 1.0645706257254368,
      "learning_rate": 0.003,
      "loss": 4.2982,
      "step": 2847
    },
    {
      "epoch": 0.02848,
      "grad_norm": 0.8386178601836739,
      "learning_rate": 0.003,
      "loss": 4.248,
      "step": 2848
    },
    {
      "epoch": 0.02849,
      "grad_norm": 0.764216257611563,
      "learning_rate": 0.003,
      "loss": 4.2302,
      "step": 2849
    },
    {
      "epoch": 0.0285,
      "grad_norm": 0.8296984070312703,
      "learning_rate": 0.003,
      "loss": 4.2406,
      "step": 2850
    },
    {
      "epoch": 0.02851,
      "grad_norm": 0.7660626969686088,
      "learning_rate": 0.003,
      "loss": 4.2727,
      "step": 2851
    },
    {
      "epoch": 0.02852,
      "grad_norm": 0.6699896951076655,
      "learning_rate": 0.003,
      "loss": 4.2432,
      "step": 2852
    },
    {
      "epoch": 0.02853,
      "grad_norm": 0.6448384261298653,
      "learning_rate": 0.003,
      "loss": 4.2061,
      "step": 2853
    },
    {
      "epoch": 0.02854,
      "grad_norm": 0.5738318430229328,
      "learning_rate": 0.003,
      "loss": 4.2452,
      "step": 2854
    },
    {
      "epoch": 0.02855,
      "grad_norm": 0.5220513862086724,
      "learning_rate": 0.003,
      "loss": 4.238,
      "step": 2855
    },
    {
      "epoch": 0.02856,
      "grad_norm": 0.487730279931273,
      "learning_rate": 0.003,
      "loss": 4.236,
      "step": 2856
    },
    {
      "epoch": 0.02857,
      "grad_norm": 0.5431060757677301,
      "learning_rate": 0.003,
      "loss": 4.2166,
      "step": 2857
    },
    {
      "epoch": 0.02858,
      "grad_norm": 0.5661595941123131,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 2858
    },
    {
      "epoch": 0.02859,
      "grad_norm": 0.4484066606331403,
      "learning_rate": 0.003,
      "loss": 4.2049,
      "step": 2859
    },
    {
      "epoch": 0.0286,
      "grad_norm": 0.4075396606562088,
      "learning_rate": 0.003,
      "loss": 4.2145,
      "step": 2860
    },
    {
      "epoch": 0.02861,
      "grad_norm": 0.38814897890981537,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 2861
    },
    {
      "epoch": 0.02862,
      "grad_norm": 0.35763606720502106,
      "learning_rate": 0.003,
      "loss": 4.1992,
      "step": 2862
    },
    {
      "epoch": 0.02863,
      "grad_norm": 0.33574165762339114,
      "learning_rate": 0.003,
      "loss": 4.2202,
      "step": 2863
    },
    {
      "epoch": 0.02864,
      "grad_norm": 0.2931133637710789,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2864
    },
    {
      "epoch": 0.02865,
      "grad_norm": 0.2747295563994342,
      "learning_rate": 0.003,
      "loss": 4.2198,
      "step": 2865
    },
    {
      "epoch": 0.02866,
      "grad_norm": 0.2878577462186846,
      "learning_rate": 0.003,
      "loss": 4.1997,
      "step": 2866
    },
    {
      "epoch": 0.02867,
      "grad_norm": 0.30967589778912374,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 2867
    },
    {
      "epoch": 0.02868,
      "grad_norm": 0.3317442746388715,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2868
    },
    {
      "epoch": 0.02869,
      "grad_norm": 0.41983761578290535,
      "learning_rate": 0.003,
      "loss": 4.1814,
      "step": 2869
    },
    {
      "epoch": 0.0287,
      "grad_norm": 0.5221052091755417,
      "learning_rate": 0.003,
      "loss": 4.2048,
      "step": 2870
    },
    {
      "epoch": 0.02871,
      "grad_norm": 0.6596505309045855,
      "learning_rate": 0.003,
      "loss": 4.1929,
      "step": 2871
    },
    {
      "epoch": 0.02872,
      "grad_norm": 0.6939156668954594,
      "learning_rate": 0.003,
      "loss": 4.2319,
      "step": 2872
    },
    {
      "epoch": 0.02873,
      "grad_norm": 0.5889408095632228,
      "learning_rate": 0.003,
      "loss": 4.1969,
      "step": 2873
    },
    {
      "epoch": 0.02874,
      "grad_norm": 0.5945812975211832,
      "learning_rate": 0.003,
      "loss": 4.2106,
      "step": 2874
    },
    {
      "epoch": 0.02875,
      "grad_norm": 0.5997681923067583,
      "learning_rate": 0.003,
      "loss": 4.2173,
      "step": 2875
    },
    {
      "epoch": 0.02876,
      "grad_norm": 0.5432684898721573,
      "learning_rate": 0.003,
      "loss": 4.1947,
      "step": 2876
    },
    {
      "epoch": 0.02877,
      "grad_norm": 0.5211950136895529,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 2877
    },
    {
      "epoch": 0.02878,
      "grad_norm": 0.4787079571485456,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 2878
    },
    {
      "epoch": 0.02879,
      "grad_norm": 0.45005415921350717,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 2879
    },
    {
      "epoch": 0.0288,
      "grad_norm": 0.43431149375473005,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 2880
    },
    {
      "epoch": 0.02881,
      "grad_norm": 0.4696763789342238,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 2881
    },
    {
      "epoch": 0.02882,
      "grad_norm": 0.5425923911480585,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 2882
    },
    {
      "epoch": 0.02883,
      "grad_norm": 0.4752679190892369,
      "learning_rate": 0.003,
      "loss": 4.1811,
      "step": 2883
    },
    {
      "epoch": 0.02884,
      "grad_norm": 0.4357972258734796,
      "learning_rate": 0.003,
      "loss": 4.2167,
      "step": 2884
    },
    {
      "epoch": 0.02885,
      "grad_norm": 0.42135010007795737,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 2885
    },
    {
      "epoch": 0.02886,
      "grad_norm": 0.4753116215584479,
      "learning_rate": 0.003,
      "loss": 4.1904,
      "step": 2886
    },
    {
      "epoch": 0.02887,
      "grad_norm": 0.4764421561290465,
      "learning_rate": 0.003,
      "loss": 4.1753,
      "step": 2887
    },
    {
      "epoch": 0.02888,
      "grad_norm": 0.4467467351995427,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 2888
    },
    {
      "epoch": 0.02889,
      "grad_norm": 0.4870006343939259,
      "learning_rate": 0.003,
      "loss": 4.1625,
      "step": 2889
    },
    {
      "epoch": 0.0289,
      "grad_norm": 0.4870936179251758,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2890
    },
    {
      "epoch": 0.02891,
      "grad_norm": 0.5804488605460525,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2891
    },
    {
      "epoch": 0.02892,
      "grad_norm": 0.6634466821898878,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 2892
    },
    {
      "epoch": 0.02893,
      "grad_norm": 0.5948043705095636,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 2893
    },
    {
      "epoch": 0.02894,
      "grad_norm": 0.6043920015751632,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 2894
    },
    {
      "epoch": 0.02895,
      "grad_norm": 0.6201059818349552,
      "learning_rate": 0.003,
      "loss": 4.2268,
      "step": 2895
    },
    {
      "epoch": 0.02896,
      "grad_norm": 0.6991042225102296,
      "learning_rate": 0.003,
      "loss": 4.1857,
      "step": 2896
    },
    {
      "epoch": 0.02897,
      "grad_norm": 0.7490225053396202,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 2897
    },
    {
      "epoch": 0.02898,
      "grad_norm": 0.7349838612699012,
      "learning_rate": 0.003,
      "loss": 4.2192,
      "step": 2898
    },
    {
      "epoch": 0.02899,
      "grad_norm": 0.7005801384832842,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 2899
    },
    {
      "epoch": 0.029,
      "grad_norm": 0.6982394096022596,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2900
    },
    {
      "epoch": 0.02901,
      "grad_norm": 0.7179332370923166,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 2901
    },
    {
      "epoch": 0.02902,
      "grad_norm": 0.6408206050082779,
      "learning_rate": 0.003,
      "loss": 4.2244,
      "step": 2902
    },
    {
      "epoch": 0.02903,
      "grad_norm": 0.5794248721426231,
      "learning_rate": 0.003,
      "loss": 4.1817,
      "step": 2903
    },
    {
      "epoch": 0.02904,
      "grad_norm": 0.6047919514195621,
      "learning_rate": 0.003,
      "loss": 4.1991,
      "step": 2904
    },
    {
      "epoch": 0.02905,
      "grad_norm": 0.6712321741283004,
      "learning_rate": 0.003,
      "loss": 4.2179,
      "step": 2905
    },
    {
      "epoch": 0.02906,
      "grad_norm": 0.7364218920572956,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2906
    },
    {
      "epoch": 0.02907,
      "grad_norm": 0.7800724882255463,
      "learning_rate": 0.003,
      "loss": 4.1811,
      "step": 2907
    },
    {
      "epoch": 0.02908,
      "grad_norm": 0.7350720938021648,
      "learning_rate": 0.003,
      "loss": 4.2049,
      "step": 2908
    },
    {
      "epoch": 0.02909,
      "grad_norm": 0.6300086850554201,
      "learning_rate": 0.003,
      "loss": 4.1942,
      "step": 2909
    },
    {
      "epoch": 0.0291,
      "grad_norm": 0.5419084839810532,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 2910
    },
    {
      "epoch": 0.02911,
      "grad_norm": 0.544140250655527,
      "learning_rate": 0.003,
      "loss": 4.1974,
      "step": 2911
    },
    {
      "epoch": 0.02912,
      "grad_norm": 0.5832074121091988,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 2912
    },
    {
      "epoch": 0.02913,
      "grad_norm": 0.46996544227828957,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 2913
    },
    {
      "epoch": 0.02914,
      "grad_norm": 0.5185065785536244,
      "learning_rate": 0.003,
      "loss": 4.1876,
      "step": 2914
    },
    {
      "epoch": 0.02915,
      "grad_norm": 0.5161041915363022,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 2915
    },
    {
      "epoch": 0.02916,
      "grad_norm": 0.5743632962294265,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 2916
    },
    {
      "epoch": 0.02917,
      "grad_norm": 0.5787314482733064,
      "learning_rate": 0.003,
      "loss": 4.2184,
      "step": 2917
    },
    {
      "epoch": 0.02918,
      "grad_norm": 0.6089620541454349,
      "learning_rate": 0.003,
      "loss": 4.1759,
      "step": 2918
    },
    {
      "epoch": 0.02919,
      "grad_norm": 0.5982379349158571,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 2919
    },
    {
      "epoch": 0.0292,
      "grad_norm": 0.5540097529270788,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 2920
    },
    {
      "epoch": 0.02921,
      "grad_norm": 0.5487736094530016,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 2921
    },
    {
      "epoch": 0.02922,
      "grad_norm": 0.46820953528164894,
      "learning_rate": 0.003,
      "loss": 4.1944,
      "step": 2922
    },
    {
      "epoch": 0.02923,
      "grad_norm": 0.3911917353150244,
      "learning_rate": 0.003,
      "loss": 4.1613,
      "step": 2923
    },
    {
      "epoch": 0.02924,
      "grad_norm": 0.3519948559374971,
      "learning_rate": 0.003,
      "loss": 4.1948,
      "step": 2924
    },
    {
      "epoch": 0.02925,
      "grad_norm": 0.37937109757353865,
      "learning_rate": 0.003,
      "loss": 4.1979,
      "step": 2925
    },
    {
      "epoch": 0.02926,
      "grad_norm": 0.4065842767583339,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 2926
    },
    {
      "epoch": 0.02927,
      "grad_norm": 0.475304983845517,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 2927
    },
    {
      "epoch": 0.02928,
      "grad_norm": 0.6734308844569649,
      "learning_rate": 0.003,
      "loss": 4.2095,
      "step": 2928
    },
    {
      "epoch": 0.02929,
      "grad_norm": 0.8546919556478062,
      "learning_rate": 0.003,
      "loss": 4.2048,
      "step": 2929
    },
    {
      "epoch": 0.0293,
      "grad_norm": 0.8348963936929714,
      "learning_rate": 0.003,
      "loss": 4.2035,
      "step": 2930
    },
    {
      "epoch": 0.02931,
      "grad_norm": 0.7454052973981365,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 2931
    },
    {
      "epoch": 0.02932,
      "grad_norm": 0.7467998755597431,
      "learning_rate": 0.003,
      "loss": 4.1986,
      "step": 2932
    },
    {
      "epoch": 0.02933,
      "grad_norm": 0.8069014916887333,
      "learning_rate": 0.003,
      "loss": 4.2246,
      "step": 2933
    },
    {
      "epoch": 0.02934,
      "grad_norm": 0.9491281331586824,
      "learning_rate": 0.003,
      "loss": 4.2274,
      "step": 2934
    },
    {
      "epoch": 0.02935,
      "grad_norm": 0.9112675124388164,
      "learning_rate": 0.003,
      "loss": 4.222,
      "step": 2935
    },
    {
      "epoch": 0.02936,
      "grad_norm": 0.8519647588347719,
      "learning_rate": 0.003,
      "loss": 4.2389,
      "step": 2936
    },
    {
      "epoch": 0.02937,
      "grad_norm": 0.9555587969650123,
      "learning_rate": 0.003,
      "loss": 4.2301,
      "step": 2937
    },
    {
      "epoch": 0.02938,
      "grad_norm": 0.9150311365393384,
      "learning_rate": 0.003,
      "loss": 4.2464,
      "step": 2938
    },
    {
      "epoch": 0.02939,
      "grad_norm": 1.0737440013454487,
      "learning_rate": 0.003,
      "loss": 4.2526,
      "step": 2939
    },
    {
      "epoch": 0.0294,
      "grad_norm": 1.0147261563008552,
      "learning_rate": 0.003,
      "loss": 4.2801,
      "step": 2940
    },
    {
      "epoch": 0.02941,
      "grad_norm": 1.022661673211658,
      "learning_rate": 0.003,
      "loss": 4.2879,
      "step": 2941
    },
    {
      "epoch": 0.02942,
      "grad_norm": 1.0444970801430784,
      "learning_rate": 0.003,
      "loss": 4.2748,
      "step": 2942
    },
    {
      "epoch": 0.02943,
      "grad_norm": 0.8455394657906586,
      "learning_rate": 0.003,
      "loss": 4.2533,
      "step": 2943
    },
    {
      "epoch": 0.02944,
      "grad_norm": 0.7628339588396233,
      "learning_rate": 0.003,
      "loss": 4.2739,
      "step": 2944
    },
    {
      "epoch": 0.02945,
      "grad_norm": 0.9388418061065686,
      "learning_rate": 0.003,
      "loss": 4.2636,
      "step": 2945
    },
    {
      "epoch": 0.02946,
      "grad_norm": 0.9857547654390597,
      "learning_rate": 0.003,
      "loss": 4.2709,
      "step": 2946
    },
    {
      "epoch": 0.02947,
      "grad_norm": 0.9078404718745537,
      "learning_rate": 0.003,
      "loss": 4.301,
      "step": 2947
    },
    {
      "epoch": 0.02948,
      "grad_norm": 0.9520758925981907,
      "learning_rate": 0.003,
      "loss": 4.2798,
      "step": 2948
    },
    {
      "epoch": 0.02949,
      "grad_norm": 1.0406544574219045,
      "learning_rate": 0.003,
      "loss": 4.2581,
      "step": 2949
    },
    {
      "epoch": 0.0295,
      "grad_norm": 0.9292248813352704,
      "learning_rate": 0.003,
      "loss": 4.2783,
      "step": 2950
    },
    {
      "epoch": 0.02951,
      "grad_norm": 0.8590093083304559,
      "learning_rate": 0.003,
      "loss": 4.2285,
      "step": 2951
    },
    {
      "epoch": 0.02952,
      "grad_norm": 0.6554290928742069,
      "learning_rate": 0.003,
      "loss": 4.2469,
      "step": 2952
    },
    {
      "epoch": 0.02953,
      "grad_norm": 0.8266033050494169,
      "learning_rate": 0.003,
      "loss": 4.2603,
      "step": 2953
    },
    {
      "epoch": 0.02954,
      "grad_norm": 0.9710367944205619,
      "learning_rate": 0.003,
      "loss": 4.3127,
      "step": 2954
    },
    {
      "epoch": 0.02955,
      "grad_norm": 0.9091757097495087,
      "learning_rate": 0.003,
      "loss": 4.2852,
      "step": 2955
    },
    {
      "epoch": 0.02956,
      "grad_norm": 0.7400776132762851,
      "learning_rate": 0.003,
      "loss": 4.2656,
      "step": 2956
    },
    {
      "epoch": 0.02957,
      "grad_norm": 0.5166853898536266,
      "learning_rate": 0.003,
      "loss": 4.2638,
      "step": 2957
    },
    {
      "epoch": 0.02958,
      "grad_norm": 0.49710477556743804,
      "learning_rate": 0.003,
      "loss": 4.276,
      "step": 2958
    },
    {
      "epoch": 0.02959,
      "grad_norm": 0.4499644234512779,
      "learning_rate": 0.003,
      "loss": 4.2565,
      "step": 2959
    },
    {
      "epoch": 0.0296,
      "grad_norm": 0.4150995513488321,
      "learning_rate": 0.003,
      "loss": 4.2574,
      "step": 2960
    },
    {
      "epoch": 0.02961,
      "grad_norm": 0.4170220343004301,
      "learning_rate": 0.003,
      "loss": 4.2257,
      "step": 2961
    },
    {
      "epoch": 0.02962,
      "grad_norm": 0.4582297929571511,
      "learning_rate": 0.003,
      "loss": 4.2309,
      "step": 2962
    },
    {
      "epoch": 0.02963,
      "grad_norm": 0.4661209765546486,
      "learning_rate": 0.003,
      "loss": 4.2018,
      "step": 2963
    },
    {
      "epoch": 0.02964,
      "grad_norm": 0.40893029477200127,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 2964
    },
    {
      "epoch": 0.02965,
      "grad_norm": 0.30952309242863746,
      "learning_rate": 0.003,
      "loss": 4.2259,
      "step": 2965
    },
    {
      "epoch": 0.02966,
      "grad_norm": 0.2925470804974261,
      "learning_rate": 0.003,
      "loss": 4.2331,
      "step": 2966
    },
    {
      "epoch": 0.02967,
      "grad_norm": 0.30877812791584447,
      "learning_rate": 0.003,
      "loss": 4.2188,
      "step": 2967
    },
    {
      "epoch": 0.02968,
      "grad_norm": 0.2982831000211414,
      "learning_rate": 0.003,
      "loss": 4.2128,
      "step": 2968
    },
    {
      "epoch": 0.02969,
      "grad_norm": 0.31941995962945985,
      "learning_rate": 0.003,
      "loss": 4.207,
      "step": 2969
    },
    {
      "epoch": 0.0297,
      "grad_norm": 0.4198384435174913,
      "learning_rate": 0.003,
      "loss": 4.1928,
      "step": 2970
    },
    {
      "epoch": 0.02971,
      "grad_norm": 0.4895528117346824,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2971
    },
    {
      "epoch": 0.02972,
      "grad_norm": 0.46403453609324874,
      "learning_rate": 0.003,
      "loss": 4.1791,
      "step": 2972
    },
    {
      "epoch": 0.02973,
      "grad_norm": 0.3996362985623369,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 2973
    },
    {
      "epoch": 0.02974,
      "grad_norm": 0.3669370591415718,
      "learning_rate": 0.003,
      "loss": 4.197,
      "step": 2974
    },
    {
      "epoch": 0.02975,
      "grad_norm": 0.3727347214399743,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 2975
    },
    {
      "epoch": 0.02976,
      "grad_norm": 0.40978542004123153,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2976
    },
    {
      "epoch": 0.02977,
      "grad_norm": 0.4390067885032759,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 2977
    },
    {
      "epoch": 0.02978,
      "grad_norm": 0.38270215912996663,
      "learning_rate": 0.003,
      "loss": 4.1974,
      "step": 2978
    },
    {
      "epoch": 0.02979,
      "grad_norm": 0.30944659309862094,
      "learning_rate": 0.003,
      "loss": 4.1765,
      "step": 2979
    },
    {
      "epoch": 0.0298,
      "grad_norm": 0.3395231398632022,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 2980
    },
    {
      "epoch": 0.02981,
      "grad_norm": 0.34592707909470244,
      "learning_rate": 0.003,
      "loss": 4.1907,
      "step": 2981
    },
    {
      "epoch": 0.02982,
      "grad_norm": 0.38035052902890665,
      "learning_rate": 0.003,
      "loss": 4.1864,
      "step": 2982
    },
    {
      "epoch": 0.02983,
      "grad_norm": 0.5707754810875516,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2983
    },
    {
      "epoch": 0.02984,
      "grad_norm": 0.8429776036729895,
      "learning_rate": 0.003,
      "loss": 4.1934,
      "step": 2984
    },
    {
      "epoch": 0.02985,
      "grad_norm": 0.9426647233611747,
      "learning_rate": 0.003,
      "loss": 4.2337,
      "step": 2985
    },
    {
      "epoch": 0.02986,
      "grad_norm": 0.5829174432301835,
      "learning_rate": 0.003,
      "loss": 4.1829,
      "step": 2986
    },
    {
      "epoch": 0.02987,
      "grad_norm": 0.5797297658595812,
      "learning_rate": 0.003,
      "loss": 4.1761,
      "step": 2987
    },
    {
      "epoch": 0.02988,
      "grad_norm": 0.5667463117618501,
      "learning_rate": 0.003,
      "loss": 4.1884,
      "step": 2988
    },
    {
      "epoch": 0.02989,
      "grad_norm": 0.46700979940535653,
      "learning_rate": 0.003,
      "loss": 4.196,
      "step": 2989
    },
    {
      "epoch": 0.0299,
      "grad_norm": 0.47425843632840237,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 2990
    },
    {
      "epoch": 0.02991,
      "grad_norm": 0.4585869479501072,
      "learning_rate": 0.003,
      "loss": 4.2064,
      "step": 2991
    },
    {
      "epoch": 0.02992,
      "grad_norm": 0.44091690071775774,
      "learning_rate": 0.003,
      "loss": 4.2006,
      "step": 2992
    },
    {
      "epoch": 0.02993,
      "grad_norm": 0.4427631334764928,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 2993
    },
    {
      "epoch": 0.02994,
      "grad_norm": 0.3680333066458911,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 2994
    },
    {
      "epoch": 0.02995,
      "grad_norm": 0.3313732790736943,
      "learning_rate": 0.003,
      "loss": 4.213,
      "step": 2995
    },
    {
      "epoch": 0.02996,
      "grad_norm": 0.33322944772224145,
      "learning_rate": 0.003,
      "loss": 4.1998,
      "step": 2996
    },
    {
      "epoch": 0.02997,
      "grad_norm": 0.35383706371010143,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 2997
    },
    {
      "epoch": 0.02998,
      "grad_norm": 0.3144018568097704,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 2998
    },
    {
      "epoch": 0.02999,
      "grad_norm": 0.2707448134611539,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 2999
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.31453822473769166,
      "learning_rate": 0.003,
      "loss": 4.1985,
      "step": 3000
    },
    {
      "epoch": 0.03001,
      "grad_norm": 0.3575380630906315,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 3001
    },
    {
      "epoch": 0.03002,
      "grad_norm": 0.37987123508738174,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 3002
    },
    {
      "epoch": 0.03003,
      "grad_norm": 0.43883399019288905,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 3003
    },
    {
      "epoch": 0.03004,
      "grad_norm": 0.4354037002724127,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 3004
    },
    {
      "epoch": 0.03005,
      "grad_norm": 0.44090216643634306,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 3005
    },
    {
      "epoch": 0.03006,
      "grad_norm": 0.4541062050925091,
      "learning_rate": 0.003,
      "loss": 4.1847,
      "step": 3006
    },
    {
      "epoch": 0.03007,
      "grad_norm": 0.5413252745038201,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 3007
    },
    {
      "epoch": 0.03008,
      "grad_norm": 0.5914463578547696,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 3008
    },
    {
      "epoch": 0.03009,
      "grad_norm": 0.6057328540964105,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 3009
    },
    {
      "epoch": 0.0301,
      "grad_norm": 0.6785032290357721,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 3010
    },
    {
      "epoch": 0.03011,
      "grad_norm": 0.7337865546330415,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3011
    },
    {
      "epoch": 0.03012,
      "grad_norm": 0.7469304524271401,
      "learning_rate": 0.003,
      "loss": 4.1847,
      "step": 3012
    },
    {
      "epoch": 0.03013,
      "grad_norm": 0.6647601453471846,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 3013
    },
    {
      "epoch": 0.03014,
      "grad_norm": 0.6507126425794799,
      "learning_rate": 0.003,
      "loss": 4.1729,
      "step": 3014
    },
    {
      "epoch": 0.03015,
      "grad_norm": 0.6380413738814859,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 3015
    },
    {
      "epoch": 0.03016,
      "grad_norm": 0.6182015477535148,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 3016
    },
    {
      "epoch": 0.03017,
      "grad_norm": 0.586737080148615,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 3017
    },
    {
      "epoch": 0.03018,
      "grad_norm": 0.5605400072157327,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 3018
    },
    {
      "epoch": 0.03019,
      "grad_norm": 0.5243331941511797,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 3019
    },
    {
      "epoch": 0.0302,
      "grad_norm": 0.49926755386002525,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 3020
    },
    {
      "epoch": 0.03021,
      "grad_norm": 0.5365217273161322,
      "learning_rate": 0.003,
      "loss": 4.2017,
      "step": 3021
    },
    {
      "epoch": 0.03022,
      "grad_norm": 0.49962846601806954,
      "learning_rate": 0.003,
      "loss": 4.1994,
      "step": 3022
    },
    {
      "epoch": 0.03023,
      "grad_norm": 0.44101225502614944,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 3023
    },
    {
      "epoch": 0.03024,
      "grad_norm": 0.4478587533564886,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 3024
    },
    {
      "epoch": 0.03025,
      "grad_norm": 0.36914153729322,
      "learning_rate": 0.003,
      "loss": 4.1809,
      "step": 3025
    },
    {
      "epoch": 0.03026,
      "grad_norm": 0.4189775215567249,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 3026
    },
    {
      "epoch": 0.03027,
      "grad_norm": 0.4812644816912879,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3027
    },
    {
      "epoch": 0.03028,
      "grad_norm": 0.5896483265711634,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 3028
    },
    {
      "epoch": 0.03029,
      "grad_norm": 0.7727132136494914,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 3029
    },
    {
      "epoch": 0.0303,
      "grad_norm": 0.8225314550231486,
      "learning_rate": 0.003,
      "loss": 4.184,
      "step": 3030
    },
    {
      "epoch": 0.03031,
      "grad_norm": 0.6946644336533113,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3031
    },
    {
      "epoch": 0.03032,
      "grad_norm": 0.7204147056292812,
      "learning_rate": 0.003,
      "loss": 4.192,
      "step": 3032
    },
    {
      "epoch": 0.03033,
      "grad_norm": 0.6929276241309884,
      "learning_rate": 0.003,
      "loss": 4.1876,
      "step": 3033
    },
    {
      "epoch": 0.03034,
      "grad_norm": 0.6811610794872539,
      "learning_rate": 0.003,
      "loss": 4.1933,
      "step": 3034
    },
    {
      "epoch": 0.03035,
      "grad_norm": 0.6796984084846434,
      "learning_rate": 0.003,
      "loss": 4.2232,
      "step": 3035
    },
    {
      "epoch": 0.03036,
      "grad_norm": 0.6389991482634575,
      "learning_rate": 0.003,
      "loss": 4.1949,
      "step": 3036
    },
    {
      "epoch": 0.03037,
      "grad_norm": 0.7107042944049744,
      "learning_rate": 0.003,
      "loss": 4.1819,
      "step": 3037
    },
    {
      "epoch": 0.03038,
      "grad_norm": 0.7261414033877567,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 3038
    },
    {
      "epoch": 0.03039,
      "grad_norm": 0.6927833025504285,
      "learning_rate": 0.003,
      "loss": 4.1894,
      "step": 3039
    },
    {
      "epoch": 0.0304,
      "grad_norm": 0.6355129937975011,
      "learning_rate": 0.003,
      "loss": 4.1938,
      "step": 3040
    },
    {
      "epoch": 0.03041,
      "grad_norm": 0.5305188998243404,
      "learning_rate": 0.003,
      "loss": 4.1953,
      "step": 3041
    },
    {
      "epoch": 0.03042,
      "grad_norm": 0.5277486902643708,
      "learning_rate": 0.003,
      "loss": 4.2187,
      "step": 3042
    },
    {
      "epoch": 0.03043,
      "grad_norm": 0.48072198596584337,
      "learning_rate": 0.003,
      "loss": 4.1691,
      "step": 3043
    },
    {
      "epoch": 0.03044,
      "grad_norm": 0.5305619681037871,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 3044
    },
    {
      "epoch": 0.03045,
      "grad_norm": 0.6019249816686142,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3045
    },
    {
      "epoch": 0.03046,
      "grad_norm": 0.6501665014846743,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3046
    },
    {
      "epoch": 0.03047,
      "grad_norm": 0.6512554720210705,
      "learning_rate": 0.003,
      "loss": 4.1981,
      "step": 3047
    },
    {
      "epoch": 0.03048,
      "grad_norm": 0.6687647551352641,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 3048
    },
    {
      "epoch": 0.03049,
      "grad_norm": 0.6495592187211907,
      "learning_rate": 0.003,
      "loss": 4.1692,
      "step": 3049
    },
    {
      "epoch": 0.0305,
      "grad_norm": 0.5742671707682592,
      "learning_rate": 0.003,
      "loss": 4.1811,
      "step": 3050
    },
    {
      "epoch": 0.03051,
      "grad_norm": 0.5413644439530771,
      "learning_rate": 0.003,
      "loss": 4.1904,
      "step": 3051
    },
    {
      "epoch": 0.03052,
      "grad_norm": 0.45841707001486987,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 3052
    },
    {
      "epoch": 0.03053,
      "grad_norm": 0.41417961125519964,
      "learning_rate": 0.003,
      "loss": 4.193,
      "step": 3053
    },
    {
      "epoch": 0.03054,
      "grad_norm": 0.43823010289384967,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3054
    },
    {
      "epoch": 0.03055,
      "grad_norm": 0.5090646995969303,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3055
    },
    {
      "epoch": 0.03056,
      "grad_norm": 0.5373409114504033,
      "learning_rate": 0.003,
      "loss": 4.1895,
      "step": 3056
    },
    {
      "epoch": 0.03057,
      "grad_norm": 0.6080154895830046,
      "learning_rate": 0.003,
      "loss": 4.1889,
      "step": 3057
    },
    {
      "epoch": 0.03058,
      "grad_norm": 0.6540368964023406,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 3058
    },
    {
      "epoch": 0.03059,
      "grad_norm": 0.5788424826468096,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 3059
    },
    {
      "epoch": 0.0306,
      "grad_norm": 0.4988555874907532,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 3060
    },
    {
      "epoch": 0.03061,
      "grad_norm": 0.4878251974397899,
      "learning_rate": 0.003,
      "loss": 4.1841,
      "step": 3061
    },
    {
      "epoch": 0.03062,
      "grad_norm": 0.5122227250314405,
      "learning_rate": 0.003,
      "loss": 4.2073,
      "step": 3062
    },
    {
      "epoch": 0.03063,
      "grad_norm": 0.48008913969979206,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 3063
    },
    {
      "epoch": 0.03064,
      "grad_norm": 0.46426896895718006,
      "learning_rate": 0.003,
      "loss": 4.1818,
      "step": 3064
    },
    {
      "epoch": 0.03065,
      "grad_norm": 0.4213766940672057,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 3065
    },
    {
      "epoch": 0.03066,
      "grad_norm": 0.3680402555018932,
      "learning_rate": 0.003,
      "loss": 4.1908,
      "step": 3066
    },
    {
      "epoch": 0.03067,
      "grad_norm": 0.3709473284252339,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 3067
    },
    {
      "epoch": 0.03068,
      "grad_norm": 0.3892404694228725,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 3068
    },
    {
      "epoch": 0.03069,
      "grad_norm": 0.421709346396654,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 3069
    },
    {
      "epoch": 0.0307,
      "grad_norm": 0.452536722715443,
      "learning_rate": 0.003,
      "loss": 4.1628,
      "step": 3070
    },
    {
      "epoch": 0.03071,
      "grad_norm": 0.43180254995245465,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 3071
    },
    {
      "epoch": 0.03072,
      "grad_norm": 0.4894410650771712,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3072
    },
    {
      "epoch": 0.03073,
      "grad_norm": 0.5403720255329184,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 3073
    },
    {
      "epoch": 0.03074,
      "grad_norm": 0.5846698892510205,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 3074
    },
    {
      "epoch": 0.03075,
      "grad_norm": 0.6574319904083324,
      "learning_rate": 0.003,
      "loss": 4.1977,
      "step": 3075
    },
    {
      "epoch": 0.03076,
      "grad_norm": 0.6619917640445645,
      "learning_rate": 0.003,
      "loss": 4.1936,
      "step": 3076
    },
    {
      "epoch": 0.03077,
      "grad_norm": 0.6043736206778582,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 3077
    },
    {
      "epoch": 0.03078,
      "grad_norm": 0.6204154957439805,
      "learning_rate": 0.003,
      "loss": 4.1729,
      "step": 3078
    },
    {
      "epoch": 0.03079,
      "grad_norm": 0.5593470356338103,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3079
    },
    {
      "epoch": 0.0308,
      "grad_norm": 0.5726380712603409,
      "learning_rate": 0.003,
      "loss": 4.1884,
      "step": 3080
    },
    {
      "epoch": 0.03081,
      "grad_norm": 0.6460072493113082,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 3081
    },
    {
      "epoch": 0.03082,
      "grad_norm": 0.6213962184440697,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 3082
    },
    {
      "epoch": 0.03083,
      "grad_norm": 0.7067435150796765,
      "learning_rate": 0.003,
      "loss": 4.1723,
      "step": 3083
    },
    {
      "epoch": 0.03084,
      "grad_norm": 0.7686911194186954,
      "learning_rate": 0.003,
      "loss": 4.1952,
      "step": 3084
    },
    {
      "epoch": 0.03085,
      "grad_norm": 0.7715187358287607,
      "learning_rate": 0.003,
      "loss": 4.193,
      "step": 3085
    },
    {
      "epoch": 0.03086,
      "grad_norm": 0.8253824375938966,
      "learning_rate": 0.003,
      "loss": 4.2186,
      "step": 3086
    },
    {
      "epoch": 0.03087,
      "grad_norm": 0.7572823291901524,
      "learning_rate": 0.003,
      "loss": 4.1941,
      "step": 3087
    },
    {
      "epoch": 0.03088,
      "grad_norm": 0.6756116771892228,
      "learning_rate": 0.003,
      "loss": 4.209,
      "step": 3088
    },
    {
      "epoch": 0.03089,
      "grad_norm": 0.6820329410441291,
      "learning_rate": 0.003,
      "loss": 4.1733,
      "step": 3089
    },
    {
      "epoch": 0.0309,
      "grad_norm": 0.7401927831575114,
      "learning_rate": 0.003,
      "loss": 4.1803,
      "step": 3090
    },
    {
      "epoch": 0.03091,
      "grad_norm": 0.6394165992316918,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 3091
    },
    {
      "epoch": 0.03092,
      "grad_norm": 0.6572468374220727,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 3092
    },
    {
      "epoch": 0.03093,
      "grad_norm": 0.6935160519017697,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 3093
    },
    {
      "epoch": 0.03094,
      "grad_norm": 0.6716436178909779,
      "learning_rate": 0.003,
      "loss": 4.188,
      "step": 3094
    },
    {
      "epoch": 0.03095,
      "grad_norm": 0.6591966230230418,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 3095
    },
    {
      "epoch": 0.03096,
      "grad_norm": 0.5959190368353036,
      "learning_rate": 0.003,
      "loss": 4.1776,
      "step": 3096
    },
    {
      "epoch": 0.03097,
      "grad_norm": 0.5866701878437197,
      "learning_rate": 0.003,
      "loss": 4.1994,
      "step": 3097
    },
    {
      "epoch": 0.03098,
      "grad_norm": 0.6431690827463121,
      "learning_rate": 0.003,
      "loss": 4.1981,
      "step": 3098
    },
    {
      "epoch": 0.03099,
      "grad_norm": 0.6249835235277311,
      "learning_rate": 0.003,
      "loss": 4.1902,
      "step": 3099
    },
    {
      "epoch": 0.031,
      "grad_norm": 0.6337483657376243,
      "learning_rate": 0.003,
      "loss": 4.1855,
      "step": 3100
    },
    {
      "epoch": 0.03101,
      "grad_norm": 0.6496496012737066,
      "learning_rate": 0.003,
      "loss": 4.1777,
      "step": 3101
    },
    {
      "epoch": 0.03102,
      "grad_norm": 0.6229765708037603,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 3102
    },
    {
      "epoch": 0.03103,
      "grad_norm": 0.6156795563561019,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 3103
    },
    {
      "epoch": 0.03104,
      "grad_norm": 0.5787370390357155,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 3104
    },
    {
      "epoch": 0.03105,
      "grad_norm": 0.46609179090900144,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 3105
    },
    {
      "epoch": 0.03106,
      "grad_norm": 0.39336258398622215,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 3106
    },
    {
      "epoch": 0.03107,
      "grad_norm": 0.4151072460907146,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 3107
    },
    {
      "epoch": 0.03108,
      "grad_norm": 0.3556637596181332,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 3108
    },
    {
      "epoch": 0.03109,
      "grad_norm": 0.3892938905312542,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3109
    },
    {
      "epoch": 0.0311,
      "grad_norm": 0.4475819635965094,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 3110
    },
    {
      "epoch": 0.03111,
      "grad_norm": 0.48864194527310445,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 3111
    },
    {
      "epoch": 0.03112,
      "grad_norm": 0.5382631741272804,
      "learning_rate": 0.003,
      "loss": 4.1803,
      "step": 3112
    },
    {
      "epoch": 0.03113,
      "grad_norm": 0.5134933039857796,
      "learning_rate": 0.003,
      "loss": 4.1686,
      "step": 3113
    },
    {
      "epoch": 0.03114,
      "grad_norm": 0.45531644644558,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 3114
    },
    {
      "epoch": 0.03115,
      "grad_norm": 0.409821580807442,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 3115
    },
    {
      "epoch": 0.03116,
      "grad_norm": 0.40691251051535354,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 3116
    },
    {
      "epoch": 0.03117,
      "grad_norm": 0.41357167905284664,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 3117
    },
    {
      "epoch": 0.03118,
      "grad_norm": 0.4338761124077823,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 3118
    },
    {
      "epoch": 0.03119,
      "grad_norm": 0.4549147699459294,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 3119
    },
    {
      "epoch": 0.0312,
      "grad_norm": 0.5032285577701964,
      "learning_rate": 0.003,
      "loss": 4.1839,
      "step": 3120
    },
    {
      "epoch": 0.03121,
      "grad_norm": 0.5982541074347542,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 3121
    },
    {
      "epoch": 0.03122,
      "grad_norm": 0.7093524896825744,
      "learning_rate": 0.003,
      "loss": 4.1884,
      "step": 3122
    },
    {
      "epoch": 0.03123,
      "grad_norm": 0.6935478647730015,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 3123
    },
    {
      "epoch": 0.03124,
      "grad_norm": 0.6414330115408258,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 3124
    },
    {
      "epoch": 0.03125,
      "grad_norm": 0.6053563143256356,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3125
    },
    {
      "epoch": 0.03126,
      "grad_norm": 0.5853887861604886,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3126
    },
    {
      "epoch": 0.03127,
      "grad_norm": 0.6449491393449573,
      "learning_rate": 0.003,
      "loss": 4.1802,
      "step": 3127
    },
    {
      "epoch": 0.03128,
      "grad_norm": 0.6502029339707546,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3128
    },
    {
      "epoch": 0.03129,
      "grad_norm": 0.7355040262252988,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 3129
    },
    {
      "epoch": 0.0313,
      "grad_norm": 0.7004678988862265,
      "learning_rate": 0.003,
      "loss": 4.1982,
      "step": 3130
    },
    {
      "epoch": 0.03131,
      "grad_norm": 0.545773725439425,
      "learning_rate": 0.003,
      "loss": 4.1723,
      "step": 3131
    },
    {
      "epoch": 0.03132,
      "grad_norm": 0.575890880552423,
      "learning_rate": 0.003,
      "loss": 4.1945,
      "step": 3132
    },
    {
      "epoch": 0.03133,
      "grad_norm": 0.6235572169219774,
      "learning_rate": 0.003,
      "loss": 4.1876,
      "step": 3133
    },
    {
      "epoch": 0.03134,
      "grad_norm": 0.6099749117272442,
      "learning_rate": 0.003,
      "loss": 4.1881,
      "step": 3134
    },
    {
      "epoch": 0.03135,
      "grad_norm": 0.6941243018172416,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 3135
    },
    {
      "epoch": 0.03136,
      "grad_norm": 0.70121888828651,
      "learning_rate": 0.003,
      "loss": 4.1772,
      "step": 3136
    },
    {
      "epoch": 0.03137,
      "grad_norm": 0.7085003488749844,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 3137
    },
    {
      "epoch": 0.03138,
      "grad_norm": 0.6535643891764703,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 3138
    },
    {
      "epoch": 0.03139,
      "grad_norm": 0.6308297801361256,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 3139
    },
    {
      "epoch": 0.0314,
      "grad_norm": 0.5309446473531373,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 3140
    },
    {
      "epoch": 0.03141,
      "grad_norm": 0.5693804160477011,
      "learning_rate": 0.003,
      "loss": 4.2097,
      "step": 3141
    },
    {
      "epoch": 0.03142,
      "grad_norm": 0.6211997698889902,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 3142
    },
    {
      "epoch": 0.03143,
      "grad_norm": 0.7073668422617321,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 3143
    },
    {
      "epoch": 0.03144,
      "grad_norm": 0.7851704146814469,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 3144
    },
    {
      "epoch": 0.03145,
      "grad_norm": 0.7954348048069579,
      "learning_rate": 0.003,
      "loss": 4.1863,
      "step": 3145
    },
    {
      "epoch": 0.03146,
      "grad_norm": 0.7034536386154087,
      "learning_rate": 0.003,
      "loss": 4.1929,
      "step": 3146
    },
    {
      "epoch": 0.03147,
      "grad_norm": 0.6509274147125733,
      "learning_rate": 0.003,
      "loss": 4.1902,
      "step": 3147
    },
    {
      "epoch": 0.03148,
      "grad_norm": 0.5546841496464855,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 3148
    },
    {
      "epoch": 0.03149,
      "grad_norm": 0.46984021974514056,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 3149
    },
    {
      "epoch": 0.0315,
      "grad_norm": 0.47872296043934637,
      "learning_rate": 0.003,
      "loss": 4.1773,
      "step": 3150
    },
    {
      "epoch": 0.03151,
      "grad_norm": 0.4383095460015223,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 3151
    },
    {
      "epoch": 0.03152,
      "grad_norm": 0.5147623540970385,
      "learning_rate": 0.003,
      "loss": 4.1863,
      "step": 3152
    },
    {
      "epoch": 0.03153,
      "grad_norm": 0.5162583950949946,
      "learning_rate": 0.003,
      "loss": 4.1959,
      "step": 3153
    },
    {
      "epoch": 0.03154,
      "grad_norm": 0.4461000862295419,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 3154
    },
    {
      "epoch": 0.03155,
      "grad_norm": 0.4554669455184566,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 3155
    },
    {
      "epoch": 0.03156,
      "grad_norm": 0.5334766080110984,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 3156
    },
    {
      "epoch": 0.03157,
      "grad_norm": 0.5732423454584378,
      "learning_rate": 0.003,
      "loss": 4.187,
      "step": 3157
    },
    {
      "epoch": 0.03158,
      "grad_norm": 0.4974411380590613,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 3158
    },
    {
      "epoch": 0.03159,
      "grad_norm": 0.44203642501633,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 3159
    },
    {
      "epoch": 0.0316,
      "grad_norm": 0.523140366094698,
      "learning_rate": 0.003,
      "loss": 4.1866,
      "step": 3160
    },
    {
      "epoch": 0.03161,
      "grad_norm": 0.5162200569611098,
      "learning_rate": 0.003,
      "loss": 4.1933,
      "step": 3161
    },
    {
      "epoch": 0.03162,
      "grad_norm": 0.5139469343607818,
      "learning_rate": 0.003,
      "loss": 4.197,
      "step": 3162
    },
    {
      "epoch": 0.03163,
      "grad_norm": 0.4948430436476383,
      "learning_rate": 0.003,
      "loss": 4.1778,
      "step": 3163
    },
    {
      "epoch": 0.03164,
      "grad_norm": 0.5669314442812998,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 3164
    },
    {
      "epoch": 0.03165,
      "grad_norm": 0.6170132015572393,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 3165
    },
    {
      "epoch": 0.03166,
      "grad_norm": 0.7007311191133752,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 3166
    },
    {
      "epoch": 0.03167,
      "grad_norm": 0.8595364667795417,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3167
    },
    {
      "epoch": 0.03168,
      "grad_norm": 0.7655669847205984,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 3168
    },
    {
      "epoch": 0.03169,
      "grad_norm": 0.6713498892330952,
      "learning_rate": 0.003,
      "loss": 4.2051,
      "step": 3169
    },
    {
      "epoch": 0.0317,
      "grad_norm": 0.677153969130682,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 3170
    },
    {
      "epoch": 0.03171,
      "grad_norm": 0.7390285282020239,
      "learning_rate": 0.003,
      "loss": 4.2052,
      "step": 3171
    },
    {
      "epoch": 0.03172,
      "grad_norm": 0.7569313720684826,
      "learning_rate": 0.003,
      "loss": 4.1948,
      "step": 3172
    },
    {
      "epoch": 0.03173,
      "grad_norm": 0.6724394145714953,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 3173
    },
    {
      "epoch": 0.03174,
      "grad_norm": 0.6566446253256072,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 3174
    },
    {
      "epoch": 0.03175,
      "grad_norm": 0.5308391109404651,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 3175
    },
    {
      "epoch": 0.03176,
      "grad_norm": 0.522270282662199,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 3176
    },
    {
      "epoch": 0.03177,
      "grad_norm": 0.5118875159250269,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 3177
    },
    {
      "epoch": 0.03178,
      "grad_norm": 0.5975750285946342,
      "learning_rate": 0.003,
      "loss": 4.2169,
      "step": 3178
    },
    {
      "epoch": 0.03179,
      "grad_norm": 0.6582127079564257,
      "learning_rate": 0.003,
      "loss": 4.1817,
      "step": 3179
    },
    {
      "epoch": 0.0318,
      "grad_norm": 0.6638267847544939,
      "learning_rate": 0.003,
      "loss": 4.1847,
      "step": 3180
    },
    {
      "epoch": 0.03181,
      "grad_norm": 0.633651026875217,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 3181
    },
    {
      "epoch": 0.03182,
      "grad_norm": 0.5922002752582246,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 3182
    },
    {
      "epoch": 0.03183,
      "grad_norm": 0.6660662970407278,
      "learning_rate": 0.003,
      "loss": 4.196,
      "step": 3183
    },
    {
      "epoch": 0.03184,
      "grad_norm": 0.6465621990428836,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 3184
    },
    {
      "epoch": 0.03185,
      "grad_norm": 0.6190012561941471,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 3185
    },
    {
      "epoch": 0.03186,
      "grad_norm": 0.5745629525057602,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3186
    },
    {
      "epoch": 0.03187,
      "grad_norm": 0.596914271669763,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 3187
    },
    {
      "epoch": 0.03188,
      "grad_norm": 0.5835579641541558,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 3188
    },
    {
      "epoch": 0.03189,
      "grad_norm": 0.576531148114492,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3189
    },
    {
      "epoch": 0.0319,
      "grad_norm": 0.46857212343029936,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 3190
    },
    {
      "epoch": 0.03191,
      "grad_norm": 0.44669804116834894,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 3191
    },
    {
      "epoch": 0.03192,
      "grad_norm": 0.5116025817651049,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3192
    },
    {
      "epoch": 0.03193,
      "grad_norm": 0.5017323740559085,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 3193
    },
    {
      "epoch": 0.03194,
      "grad_norm": 0.5506388551758917,
      "learning_rate": 0.003,
      "loss": 4.1961,
      "step": 3194
    },
    {
      "epoch": 0.03195,
      "grad_norm": 0.6905257500061265,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 3195
    },
    {
      "epoch": 0.03196,
      "grad_norm": 0.6814664188108994,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3196
    },
    {
      "epoch": 0.03197,
      "grad_norm": 0.6734616316609701,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 3197
    },
    {
      "epoch": 0.03198,
      "grad_norm": 0.5599435801670639,
      "learning_rate": 0.003,
      "loss": 4.1795,
      "step": 3198
    },
    {
      "epoch": 0.03199,
      "grad_norm": 0.4771733892435305,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 3199
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.4956367507190788,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3200
    },
    {
      "epoch": 0.03201,
      "grad_norm": 0.40451540956897725,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 3201
    },
    {
      "epoch": 0.03202,
      "grad_norm": 0.42753873923321506,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 3202
    },
    {
      "epoch": 0.03203,
      "grad_norm": 0.45457758457355213,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3203
    },
    {
      "epoch": 0.03204,
      "grad_norm": 0.49270456285505987,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 3204
    },
    {
      "epoch": 0.03205,
      "grad_norm": 0.551907056420435,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 3205
    },
    {
      "epoch": 0.03206,
      "grad_norm": 0.5848754633539133,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 3206
    },
    {
      "epoch": 0.03207,
      "grad_norm": 0.5646598219152835,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 3207
    },
    {
      "epoch": 0.03208,
      "grad_norm": 0.5057453481084084,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 3208
    },
    {
      "epoch": 0.03209,
      "grad_norm": 0.4868038489807391,
      "learning_rate": 0.003,
      "loss": 4.2026,
      "step": 3209
    },
    {
      "epoch": 0.0321,
      "grad_norm": 0.5084192511453576,
      "learning_rate": 0.003,
      "loss": 4.1839,
      "step": 3210
    },
    {
      "epoch": 0.03211,
      "grad_norm": 0.536979773481941,
      "learning_rate": 0.003,
      "loss": 4.1566,
      "step": 3211
    },
    {
      "epoch": 0.03212,
      "grad_norm": 0.6453864413806403,
      "learning_rate": 0.003,
      "loss": 4.171,
      "step": 3212
    },
    {
      "epoch": 0.03213,
      "grad_norm": 0.6979695060374274,
      "learning_rate": 0.003,
      "loss": 4.1767,
      "step": 3213
    },
    {
      "epoch": 0.03214,
      "grad_norm": 0.8362949946620015,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 3214
    },
    {
      "epoch": 0.03215,
      "grad_norm": 0.9000343317810777,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 3215
    },
    {
      "epoch": 0.03216,
      "grad_norm": 0.8252513255224121,
      "learning_rate": 0.003,
      "loss": 4.1822,
      "step": 3216
    },
    {
      "epoch": 0.03217,
      "grad_norm": 0.7553474851929083,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3217
    },
    {
      "epoch": 0.03218,
      "grad_norm": 0.7184034075019531,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 3218
    },
    {
      "epoch": 0.03219,
      "grad_norm": 0.7559505824598467,
      "learning_rate": 0.003,
      "loss": 4.1839,
      "step": 3219
    },
    {
      "epoch": 0.0322,
      "grad_norm": 0.7462679166089925,
      "learning_rate": 0.003,
      "loss": 4.1959,
      "step": 3220
    },
    {
      "epoch": 0.03221,
      "grad_norm": 0.8114417269245792,
      "learning_rate": 0.003,
      "loss": 4.2066,
      "step": 3221
    },
    {
      "epoch": 0.03222,
      "grad_norm": 0.7398684360618538,
      "learning_rate": 0.003,
      "loss": 4.1975,
      "step": 3222
    },
    {
      "epoch": 0.03223,
      "grad_norm": 0.6559376119246889,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 3223
    },
    {
      "epoch": 0.03224,
      "grad_norm": 0.6017029656710154,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3224
    },
    {
      "epoch": 0.03225,
      "grad_norm": 0.5653674381326333,
      "learning_rate": 0.003,
      "loss": 4.1866,
      "step": 3225
    },
    {
      "epoch": 0.03226,
      "grad_norm": 0.6502250085881689,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 3226
    },
    {
      "epoch": 0.03227,
      "grad_norm": 0.7143712957982395,
      "learning_rate": 0.003,
      "loss": 4.1994,
      "step": 3227
    },
    {
      "epoch": 0.03228,
      "grad_norm": 0.6315781743986074,
      "learning_rate": 0.003,
      "loss": 4.1692,
      "step": 3228
    },
    {
      "epoch": 0.03229,
      "grad_norm": 0.5567058811508958,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 3229
    },
    {
      "epoch": 0.0323,
      "grad_norm": 0.5544929832938678,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 3230
    },
    {
      "epoch": 0.03231,
      "grad_norm": 0.470332875801311,
      "learning_rate": 0.003,
      "loss": 4.1883,
      "step": 3231
    },
    {
      "epoch": 0.03232,
      "grad_norm": 0.4362924227260804,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 3232
    },
    {
      "epoch": 0.03233,
      "grad_norm": 0.4029479081021775,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 3233
    },
    {
      "epoch": 0.03234,
      "grad_norm": 0.48479141962478917,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 3234
    },
    {
      "epoch": 0.03235,
      "grad_norm": 0.5441326810507732,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 3235
    },
    {
      "epoch": 0.03236,
      "grad_norm": 0.5291856539541236,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 3236
    },
    {
      "epoch": 0.03237,
      "grad_norm": 0.4961166691014795,
      "learning_rate": 0.003,
      "loss": 4.1904,
      "step": 3237
    },
    {
      "epoch": 0.03238,
      "grad_norm": 0.4503193502587542,
      "learning_rate": 0.003,
      "loss": 4.1937,
      "step": 3238
    },
    {
      "epoch": 0.03239,
      "grad_norm": 0.49034979349671226,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 3239
    },
    {
      "epoch": 0.0324,
      "grad_norm": 0.5016168112544078,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 3240
    },
    {
      "epoch": 0.03241,
      "grad_norm": 0.576118532422288,
      "learning_rate": 0.003,
      "loss": 4.1878,
      "step": 3241
    },
    {
      "epoch": 0.03242,
      "grad_norm": 0.6819472138134928,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 3242
    },
    {
      "epoch": 0.03243,
      "grad_norm": 0.7729342734665511,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 3243
    },
    {
      "epoch": 0.03244,
      "grad_norm": 0.8596141569373975,
      "learning_rate": 0.003,
      "loss": 4.1873,
      "step": 3244
    },
    {
      "epoch": 0.03245,
      "grad_norm": 0.8099112239109375,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 3245
    },
    {
      "epoch": 0.03246,
      "grad_norm": 0.6033775070976766,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 3246
    },
    {
      "epoch": 0.03247,
      "grad_norm": 0.7122383502056996,
      "learning_rate": 0.003,
      "loss": 4.1951,
      "step": 3247
    },
    {
      "epoch": 0.03248,
      "grad_norm": 0.6134714767115921,
      "learning_rate": 0.003,
      "loss": 4.2117,
      "step": 3248
    },
    {
      "epoch": 0.03249,
      "grad_norm": 0.551686621394429,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3249
    },
    {
      "epoch": 0.0325,
      "grad_norm": 0.4692280069519903,
      "learning_rate": 0.003,
      "loss": 4.2016,
      "step": 3250
    },
    {
      "epoch": 0.03251,
      "grad_norm": 0.4762141737958913,
      "learning_rate": 0.003,
      "loss": 4.1968,
      "step": 3251
    },
    {
      "epoch": 0.03252,
      "grad_norm": 0.451291493723249,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 3252
    },
    {
      "epoch": 0.03253,
      "grad_norm": 0.5006035466184977,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 3253
    },
    {
      "epoch": 0.03254,
      "grad_norm": 0.49103882046130487,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 3254
    },
    {
      "epoch": 0.03255,
      "grad_norm": 0.528505966654426,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3255
    },
    {
      "epoch": 0.03256,
      "grad_norm": 0.5411019927206225,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3256
    },
    {
      "epoch": 0.03257,
      "grad_norm": 0.5426466711031169,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 3257
    },
    {
      "epoch": 0.03258,
      "grad_norm": 0.5458703600164347,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 3258
    },
    {
      "epoch": 0.03259,
      "grad_norm": 0.5203040924859128,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3259
    },
    {
      "epoch": 0.0326,
      "grad_norm": 0.45480155011690926,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 3260
    },
    {
      "epoch": 0.03261,
      "grad_norm": 0.4731935212642663,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 3261
    },
    {
      "epoch": 0.03262,
      "grad_norm": 0.47522534323877436,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 3262
    },
    {
      "epoch": 0.03263,
      "grad_norm": 0.4999694797882516,
      "learning_rate": 0.003,
      "loss": 4.1772,
      "step": 3263
    },
    {
      "epoch": 0.03264,
      "grad_norm": 0.5217567082705248,
      "learning_rate": 0.003,
      "loss": 4.189,
      "step": 3264
    },
    {
      "epoch": 0.03265,
      "grad_norm": 0.6014802197349467,
      "learning_rate": 0.003,
      "loss": 4.187,
      "step": 3265
    },
    {
      "epoch": 0.03266,
      "grad_norm": 0.6262382617606659,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 3266
    },
    {
      "epoch": 0.03267,
      "grad_norm": 0.6366837634406238,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 3267
    },
    {
      "epoch": 0.03268,
      "grad_norm": 0.724266364773543,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 3268
    },
    {
      "epoch": 0.03269,
      "grad_norm": 0.7761710629109877,
      "learning_rate": 0.003,
      "loss": 4.1789,
      "step": 3269
    },
    {
      "epoch": 0.0327,
      "grad_norm": 0.7383390055481723,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 3270
    },
    {
      "epoch": 0.03271,
      "grad_norm": 0.6168450072471018,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 3271
    },
    {
      "epoch": 0.03272,
      "grad_norm": 0.5388626714250451,
      "learning_rate": 0.003,
      "loss": 4.1712,
      "step": 3272
    },
    {
      "epoch": 0.03273,
      "grad_norm": 0.541266940860741,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 3273
    },
    {
      "epoch": 0.03274,
      "grad_norm": 0.5293709638699655,
      "learning_rate": 0.003,
      "loss": 4.1897,
      "step": 3274
    },
    {
      "epoch": 0.03275,
      "grad_norm": 0.45689882391191533,
      "learning_rate": 0.003,
      "loss": 4.1828,
      "step": 3275
    },
    {
      "epoch": 0.03276,
      "grad_norm": 0.341646145545343,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 3276
    },
    {
      "epoch": 0.03277,
      "grad_norm": 0.36838541029861377,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3277
    },
    {
      "epoch": 0.03278,
      "grad_norm": 0.3636023442028596,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 3278
    },
    {
      "epoch": 0.03279,
      "grad_norm": 0.3985620455372761,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3279
    },
    {
      "epoch": 0.0328,
      "grad_norm": 0.4543675760898354,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 3280
    },
    {
      "epoch": 0.03281,
      "grad_norm": 0.5083262289907728,
      "learning_rate": 0.003,
      "loss": 4.1897,
      "step": 3281
    },
    {
      "epoch": 0.03282,
      "grad_norm": 0.6315851302649976,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 3282
    },
    {
      "epoch": 0.03283,
      "grad_norm": 0.7322734724447769,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 3283
    },
    {
      "epoch": 0.03284,
      "grad_norm": 0.7647225152469821,
      "learning_rate": 0.003,
      "loss": 4.1977,
      "step": 3284
    },
    {
      "epoch": 0.03285,
      "grad_norm": 0.8483759241793429,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 3285
    },
    {
      "epoch": 0.03286,
      "grad_norm": 0.7999560566423449,
      "learning_rate": 0.003,
      "loss": 4.1904,
      "step": 3286
    },
    {
      "epoch": 0.03287,
      "grad_norm": 0.71589535526088,
      "learning_rate": 0.003,
      "loss": 4.196,
      "step": 3287
    },
    {
      "epoch": 0.03288,
      "grad_norm": 0.6336111011473581,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 3288
    },
    {
      "epoch": 0.03289,
      "grad_norm": 0.6641631109672658,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 3289
    },
    {
      "epoch": 0.0329,
      "grad_norm": 0.6647898720593706,
      "learning_rate": 0.003,
      "loss": 4.2094,
      "step": 3290
    },
    {
      "epoch": 0.03291,
      "grad_norm": 0.5807864922821838,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 3291
    },
    {
      "epoch": 0.03292,
      "grad_norm": 0.515754721038097,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3292
    },
    {
      "epoch": 0.03293,
      "grad_norm": 0.5498053265889223,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 3293
    },
    {
      "epoch": 0.03294,
      "grad_norm": 0.6841987985815203,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 3294
    },
    {
      "epoch": 0.03295,
      "grad_norm": 0.6093729695331228,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3295
    },
    {
      "epoch": 0.03296,
      "grad_norm": 0.6089661662715524,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 3296
    },
    {
      "epoch": 0.03297,
      "grad_norm": 0.5614137495318692,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3297
    },
    {
      "epoch": 0.03298,
      "grad_norm": 0.5134366817986185,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 3298
    },
    {
      "epoch": 0.03299,
      "grad_norm": 0.549963907837565,
      "learning_rate": 0.003,
      "loss": 4.1739,
      "step": 3299
    },
    {
      "epoch": 0.033,
      "grad_norm": 0.6519931831327299,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3300
    },
    {
      "epoch": 0.03301,
      "grad_norm": 0.7183972305593801,
      "learning_rate": 0.003,
      "loss": 4.2021,
      "step": 3301
    },
    {
      "epoch": 0.03302,
      "grad_norm": 0.7018171786354589,
      "learning_rate": 0.003,
      "loss": 4.1793,
      "step": 3302
    },
    {
      "epoch": 0.03303,
      "grad_norm": 0.5921722024443354,
      "learning_rate": 0.003,
      "loss": 4.1793,
      "step": 3303
    },
    {
      "epoch": 0.03304,
      "grad_norm": 0.5309946357381335,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3304
    },
    {
      "epoch": 0.03305,
      "grad_norm": 0.5654757181942384,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3305
    },
    {
      "epoch": 0.03306,
      "grad_norm": 0.5651614262192222,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 3306
    },
    {
      "epoch": 0.03307,
      "grad_norm": 0.6220742834040969,
      "learning_rate": 0.003,
      "loss": 4.1933,
      "step": 3307
    },
    {
      "epoch": 0.03308,
      "grad_norm": 0.6857411580561741,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3308
    },
    {
      "epoch": 0.03309,
      "grad_norm": 0.7647611362606778,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3309
    },
    {
      "epoch": 0.0331,
      "grad_norm": 0.9027911762824596,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 3310
    },
    {
      "epoch": 0.03311,
      "grad_norm": 0.9232745325192504,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 3311
    },
    {
      "epoch": 0.03312,
      "grad_norm": 1.0033276702218215,
      "learning_rate": 0.003,
      "loss": 4.1863,
      "step": 3312
    },
    {
      "epoch": 0.03313,
      "grad_norm": 0.8616181406278286,
      "learning_rate": 0.003,
      "loss": 4.217,
      "step": 3313
    },
    {
      "epoch": 0.03314,
      "grad_norm": 0.6538312191871826,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 3314
    },
    {
      "epoch": 0.03315,
      "grad_norm": 0.6513293592499171,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 3315
    },
    {
      "epoch": 0.03316,
      "grad_norm": 0.5753636988224785,
      "learning_rate": 0.003,
      "loss": 4.1986,
      "step": 3316
    },
    {
      "epoch": 0.03317,
      "grad_norm": 0.5660504612340471,
      "learning_rate": 0.003,
      "loss": 4.1897,
      "step": 3317
    },
    {
      "epoch": 0.03318,
      "grad_norm": 0.5321945122016645,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 3318
    },
    {
      "epoch": 0.03319,
      "grad_norm": 0.5341973320019645,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3319
    },
    {
      "epoch": 0.0332,
      "grad_norm": 0.47286473477321905,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3320
    },
    {
      "epoch": 0.03321,
      "grad_norm": 0.5050275992577558,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 3321
    },
    {
      "epoch": 0.03322,
      "grad_norm": 0.4844023535662593,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 3322
    },
    {
      "epoch": 0.03323,
      "grad_norm": 0.43309930004470687,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 3323
    },
    {
      "epoch": 0.03324,
      "grad_norm": 0.44890262885350884,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 3324
    },
    {
      "epoch": 0.03325,
      "grad_norm": 0.5095521797271855,
      "learning_rate": 0.003,
      "loss": 4.1917,
      "step": 3325
    },
    {
      "epoch": 0.03326,
      "grad_norm": 0.477471322433437,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 3326
    },
    {
      "epoch": 0.03327,
      "grad_norm": 0.4315337339422698,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3327
    },
    {
      "epoch": 0.03328,
      "grad_norm": 0.4171492646524173,
      "learning_rate": 0.003,
      "loss": 4.1912,
      "step": 3328
    },
    {
      "epoch": 0.03329,
      "grad_norm": 0.4506599360723401,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3329
    },
    {
      "epoch": 0.0333,
      "grad_norm": 0.43373534373844225,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 3330
    },
    {
      "epoch": 0.03331,
      "grad_norm": 0.48072192590135343,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 3331
    },
    {
      "epoch": 0.03332,
      "grad_norm": 0.5603763883718295,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 3332
    },
    {
      "epoch": 0.03333,
      "grad_norm": 0.5773402592299904,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 3333
    },
    {
      "epoch": 0.03334,
      "grad_norm": 0.5925239668699488,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 3334
    },
    {
      "epoch": 0.03335,
      "grad_norm": 0.5265113459144353,
      "learning_rate": 0.003,
      "loss": 4.1765,
      "step": 3335
    },
    {
      "epoch": 0.03336,
      "grad_norm": 0.4418521183409816,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 3336
    },
    {
      "epoch": 0.03337,
      "grad_norm": 0.42952166536927455,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 3337
    },
    {
      "epoch": 0.03338,
      "grad_norm": 0.43519066754716446,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 3338
    },
    {
      "epoch": 0.03339,
      "grad_norm": 0.4658156069803623,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3339
    },
    {
      "epoch": 0.0334,
      "grad_norm": 0.5171946739861909,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 3340
    },
    {
      "epoch": 0.03341,
      "grad_norm": 0.6356266827382321,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3341
    },
    {
      "epoch": 0.03342,
      "grad_norm": 0.6867092162549009,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 3342
    },
    {
      "epoch": 0.03343,
      "grad_norm": 0.704031363814926,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 3343
    },
    {
      "epoch": 0.03344,
      "grad_norm": 0.7513440673993107,
      "learning_rate": 0.003,
      "loss": 4.1949,
      "step": 3344
    },
    {
      "epoch": 0.03345,
      "grad_norm": 0.7519677795539157,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 3345
    },
    {
      "epoch": 0.03346,
      "grad_norm": 0.7713219880690217,
      "learning_rate": 0.003,
      "loss": 4.19,
      "step": 3346
    },
    {
      "epoch": 0.03347,
      "grad_norm": 0.8480035915924508,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 3347
    },
    {
      "epoch": 0.03348,
      "grad_norm": 0.8325415221214432,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 3348
    },
    {
      "epoch": 0.03349,
      "grad_norm": 0.7621143367252534,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 3349
    },
    {
      "epoch": 0.0335,
      "grad_norm": 0.7234514614978355,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3350
    },
    {
      "epoch": 0.03351,
      "grad_norm": 0.6715941237384234,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 3351
    },
    {
      "epoch": 0.03352,
      "grad_norm": 0.6771948560655807,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 3352
    },
    {
      "epoch": 0.03353,
      "grad_norm": 0.7374767561640508,
      "learning_rate": 0.003,
      "loss": 4.1997,
      "step": 3353
    },
    {
      "epoch": 0.03354,
      "grad_norm": 0.7924221564216216,
      "learning_rate": 0.003,
      "loss": 4.1685,
      "step": 3354
    },
    {
      "epoch": 0.03355,
      "grad_norm": 0.752343695322846,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 3355
    },
    {
      "epoch": 0.03356,
      "grad_norm": 0.6955515760269133,
      "learning_rate": 0.003,
      "loss": 4.2144,
      "step": 3356
    },
    {
      "epoch": 0.03357,
      "grad_norm": 0.6439226598770458,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3357
    },
    {
      "epoch": 0.03358,
      "grad_norm": 0.6072963552767238,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 3358
    },
    {
      "epoch": 0.03359,
      "grad_norm": 0.5544317009226006,
      "learning_rate": 0.003,
      "loss": 4.1868,
      "step": 3359
    },
    {
      "epoch": 0.0336,
      "grad_norm": 0.7584252977946636,
      "learning_rate": 0.003,
      "loss": 4.2055,
      "step": 3360
    },
    {
      "epoch": 0.03361,
      "grad_norm": 0.8675540717955526,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 3361
    },
    {
      "epoch": 0.03362,
      "grad_norm": 0.8109217912617241,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 3362
    },
    {
      "epoch": 0.03363,
      "grad_norm": 0.7666749182841046,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 3363
    },
    {
      "epoch": 0.03364,
      "grad_norm": 0.7823149796193858,
      "learning_rate": 0.003,
      "loss": 4.1828,
      "step": 3364
    },
    {
      "epoch": 0.03365,
      "grad_norm": 0.6387430725521948,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3365
    },
    {
      "epoch": 0.03366,
      "grad_norm": 0.5594833522786266,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3366
    },
    {
      "epoch": 0.03367,
      "grad_norm": 0.5181383097238287,
      "learning_rate": 0.003,
      "loss": 4.1575,
      "step": 3367
    },
    {
      "epoch": 0.03368,
      "grad_norm": 0.5760572623008274,
      "learning_rate": 0.003,
      "loss": 4.207,
      "step": 3368
    },
    {
      "epoch": 0.03369,
      "grad_norm": 0.550972005009095,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 3369
    },
    {
      "epoch": 0.0337,
      "grad_norm": 0.5261333264386098,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 3370
    },
    {
      "epoch": 0.03371,
      "grad_norm": 0.4162241207425965,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 3371
    },
    {
      "epoch": 0.03372,
      "grad_norm": 0.3964364873259744,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3372
    },
    {
      "epoch": 0.03373,
      "grad_norm": 0.39921807899478956,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 3373
    },
    {
      "epoch": 0.03374,
      "grad_norm": 0.40722470978586633,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 3374
    },
    {
      "epoch": 0.03375,
      "grad_norm": 0.4174362735293826,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 3375
    },
    {
      "epoch": 0.03376,
      "grad_norm": 0.4670589348527475,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 3376
    },
    {
      "epoch": 0.03377,
      "grad_norm": 0.5539342573100582,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 3377
    },
    {
      "epoch": 0.03378,
      "grad_norm": 0.6187862572290908,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 3378
    },
    {
      "epoch": 0.03379,
      "grad_norm": 0.6015622211172064,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 3379
    },
    {
      "epoch": 0.0338,
      "grad_norm": 0.4615401428395281,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 3380
    },
    {
      "epoch": 0.03381,
      "grad_norm": 0.32933941833228836,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 3381
    },
    {
      "epoch": 0.03382,
      "grad_norm": 0.3788346338425552,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 3382
    },
    {
      "epoch": 0.03383,
      "grad_norm": 0.376989210662061,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 3383
    },
    {
      "epoch": 0.03384,
      "grad_norm": 0.34387151748219785,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 3384
    },
    {
      "epoch": 0.03385,
      "grad_norm": 0.3352611282783692,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 3385
    },
    {
      "epoch": 0.03386,
      "grad_norm": 0.33892990368837295,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3386
    },
    {
      "epoch": 0.03387,
      "grad_norm": 0.37206580118259613,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 3387
    },
    {
      "epoch": 0.03388,
      "grad_norm": 0.41506855452840574,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 3388
    },
    {
      "epoch": 0.03389,
      "grad_norm": 0.42405220029053703,
      "learning_rate": 0.003,
      "loss": 4.1835,
      "step": 3389
    },
    {
      "epoch": 0.0339,
      "grad_norm": 0.4090594827115757,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 3390
    },
    {
      "epoch": 0.03391,
      "grad_norm": 0.4342402082004512,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 3391
    },
    {
      "epoch": 0.03392,
      "grad_norm": 0.49470807894733526,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 3392
    },
    {
      "epoch": 0.03393,
      "grad_norm": 0.7113520954651117,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3393
    },
    {
      "epoch": 0.03394,
      "grad_norm": 0.8695013133927905,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3394
    },
    {
      "epoch": 0.03395,
      "grad_norm": 0.937839460650371,
      "learning_rate": 0.003,
      "loss": 4.1994,
      "step": 3395
    },
    {
      "epoch": 0.03396,
      "grad_norm": 0.7904694460773314,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 3396
    },
    {
      "epoch": 0.03397,
      "grad_norm": 0.7235872457578897,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 3397
    },
    {
      "epoch": 0.03398,
      "grad_norm": 0.8062806717697898,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 3398
    },
    {
      "epoch": 0.03399,
      "grad_norm": 0.7158225027891656,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 3399
    },
    {
      "epoch": 0.034,
      "grad_norm": 0.6123275074220181,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 3400
    },
    {
      "epoch": 0.03401,
      "grad_norm": 0.6599152366758141,
      "learning_rate": 0.003,
      "loss": 4.1767,
      "step": 3401
    },
    {
      "epoch": 0.03402,
      "grad_norm": 0.6314070611427443,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 3402
    },
    {
      "epoch": 0.03403,
      "grad_norm": 0.6571294664516991,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 3403
    },
    {
      "epoch": 0.03404,
      "grad_norm": 0.6404238631170781,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 3404
    },
    {
      "epoch": 0.03405,
      "grad_norm": 0.6321854819314461,
      "learning_rate": 0.003,
      "loss": 4.1828,
      "step": 3405
    },
    {
      "epoch": 0.03406,
      "grad_norm": 0.6941079886283862,
      "learning_rate": 0.003,
      "loss": 4.1694,
      "step": 3406
    },
    {
      "epoch": 0.03407,
      "grad_norm": 0.8136592990040448,
      "learning_rate": 0.003,
      "loss": 4.1931,
      "step": 3407
    },
    {
      "epoch": 0.03408,
      "grad_norm": 0.8054752836427478,
      "learning_rate": 0.003,
      "loss": 4.1777,
      "step": 3408
    },
    {
      "epoch": 0.03409,
      "grad_norm": 0.7438300271168681,
      "learning_rate": 0.003,
      "loss": 4.1836,
      "step": 3409
    },
    {
      "epoch": 0.0341,
      "grad_norm": 0.7769660820773376,
      "learning_rate": 0.003,
      "loss": 4.1811,
      "step": 3410
    },
    {
      "epoch": 0.03411,
      "grad_norm": 0.7387445989360225,
      "learning_rate": 0.003,
      "loss": 4.1964,
      "step": 3411
    },
    {
      "epoch": 0.03412,
      "grad_norm": 0.7543102885862639,
      "learning_rate": 0.003,
      "loss": 4.2165,
      "step": 3412
    },
    {
      "epoch": 0.03413,
      "grad_norm": 0.7858533091946368,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 3413
    },
    {
      "epoch": 0.03414,
      "grad_norm": 0.8085223407654301,
      "learning_rate": 0.003,
      "loss": 4.2197,
      "step": 3414
    },
    {
      "epoch": 0.03415,
      "grad_norm": 0.7204513262789403,
      "learning_rate": 0.003,
      "loss": 4.1982,
      "step": 3415
    },
    {
      "epoch": 0.03416,
      "grad_norm": 0.669578263783182,
      "learning_rate": 0.003,
      "loss": 4.2009,
      "step": 3416
    },
    {
      "epoch": 0.03417,
      "grad_norm": 0.595729901049528,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 3417
    },
    {
      "epoch": 0.03418,
      "grad_norm": 0.6302485601669457,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 3418
    },
    {
      "epoch": 0.03419,
      "grad_norm": 0.48681001127779644,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3419
    },
    {
      "epoch": 0.0342,
      "grad_norm": 0.4381251825205999,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 3420
    },
    {
      "epoch": 0.03421,
      "grad_norm": 0.40842368527410894,
      "learning_rate": 0.003,
      "loss": 4.1733,
      "step": 3421
    },
    {
      "epoch": 0.03422,
      "grad_norm": 0.4158377016736793,
      "learning_rate": 0.003,
      "loss": 4.1888,
      "step": 3422
    },
    {
      "epoch": 0.03423,
      "grad_norm": 0.38251097705951487,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 3423
    },
    {
      "epoch": 0.03424,
      "grad_norm": 0.4199738680628049,
      "learning_rate": 0.003,
      "loss": 4.1889,
      "step": 3424
    },
    {
      "epoch": 0.03425,
      "grad_norm": 0.5152529015696985,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3425
    },
    {
      "epoch": 0.03426,
      "grad_norm": 0.5309877993913092,
      "learning_rate": 0.003,
      "loss": 4.1773,
      "step": 3426
    },
    {
      "epoch": 0.03427,
      "grad_norm": 0.4899224228978265,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 3427
    },
    {
      "epoch": 0.03428,
      "grad_norm": 0.5031074901419099,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 3428
    },
    {
      "epoch": 0.03429,
      "grad_norm": 0.6278449502427672,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3429
    },
    {
      "epoch": 0.0343,
      "grad_norm": 0.6739968965302988,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 3430
    },
    {
      "epoch": 0.03431,
      "grad_norm": 0.7060406786523291,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 3431
    },
    {
      "epoch": 0.03432,
      "grad_norm": 0.6678436460415758,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3432
    },
    {
      "epoch": 0.03433,
      "grad_norm": 0.47826597565957213,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 3433
    },
    {
      "epoch": 0.03434,
      "grad_norm": 0.4648969929741348,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3434
    },
    {
      "epoch": 0.03435,
      "grad_norm": 0.46695854146561,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 3435
    },
    {
      "epoch": 0.03436,
      "grad_norm": 0.5162770863422669,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 3436
    },
    {
      "epoch": 0.03437,
      "grad_norm": 0.48303509013107304,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 3437
    },
    {
      "epoch": 0.03438,
      "grad_norm": 0.5052348764127375,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 3438
    },
    {
      "epoch": 0.03439,
      "grad_norm": 0.4517303897020918,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 3439
    },
    {
      "epoch": 0.0344,
      "grad_norm": 0.44296664325342977,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 3440
    },
    {
      "epoch": 0.03441,
      "grad_norm": 0.4450842770576517,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3441
    },
    {
      "epoch": 0.03442,
      "grad_norm": 0.4941470142821772,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 3442
    },
    {
      "epoch": 0.03443,
      "grad_norm": 0.5954542724862985,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 3443
    },
    {
      "epoch": 0.03444,
      "grad_norm": 0.679140132757601,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 3444
    },
    {
      "epoch": 0.03445,
      "grad_norm": 0.6296700934818499,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 3445
    },
    {
      "epoch": 0.03446,
      "grad_norm": 0.5424802868176756,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 3446
    },
    {
      "epoch": 0.03447,
      "grad_norm": 0.4237052026562954,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3447
    },
    {
      "epoch": 0.03448,
      "grad_norm": 0.49663616285480033,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 3448
    },
    {
      "epoch": 0.03449,
      "grad_norm": 0.4965879325162881,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 3449
    },
    {
      "epoch": 0.0345,
      "grad_norm": 0.4699418192975183,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 3450
    },
    {
      "epoch": 0.03451,
      "grad_norm": 0.4342808651184125,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 3451
    },
    {
      "epoch": 0.03452,
      "grad_norm": 0.48981095252176143,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3452
    },
    {
      "epoch": 0.03453,
      "grad_norm": 0.5221855555116031,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 3453
    },
    {
      "epoch": 0.03454,
      "grad_norm": 0.5252422198675372,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 3454
    },
    {
      "epoch": 0.03455,
      "grad_norm": 0.5714349317836669,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 3455
    },
    {
      "epoch": 0.03456,
      "grad_norm": 0.5686518541913316,
      "learning_rate": 0.003,
      "loss": 4.154,
      "step": 3456
    },
    {
      "epoch": 0.03457,
      "grad_norm": 0.5777235599897509,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 3457
    },
    {
      "epoch": 0.03458,
      "grad_norm": 0.6023496924245035,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3458
    },
    {
      "epoch": 0.03459,
      "grad_norm": 0.6835034129659396,
      "learning_rate": 0.003,
      "loss": 4.1883,
      "step": 3459
    },
    {
      "epoch": 0.0346,
      "grad_norm": 0.7616829391144937,
      "learning_rate": 0.003,
      "loss": 4.195,
      "step": 3460
    },
    {
      "epoch": 0.03461,
      "grad_norm": 0.6951133371872186,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 3461
    },
    {
      "epoch": 0.03462,
      "grad_norm": 0.6533341057192256,
      "learning_rate": 0.003,
      "loss": 4.1653,
      "step": 3462
    },
    {
      "epoch": 0.03463,
      "grad_norm": 0.7614181817822457,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 3463
    },
    {
      "epoch": 0.03464,
      "grad_norm": 0.8154255816699014,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 3464
    },
    {
      "epoch": 0.03465,
      "grad_norm": 0.834367328138272,
      "learning_rate": 0.003,
      "loss": 4.1879,
      "step": 3465
    },
    {
      "epoch": 0.03466,
      "grad_norm": 0.7054161963486509,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 3466
    },
    {
      "epoch": 0.03467,
      "grad_norm": 0.6747042802423374,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 3467
    },
    {
      "epoch": 0.03468,
      "grad_norm": 0.6274506871293699,
      "learning_rate": 0.003,
      "loss": 4.1789,
      "step": 3468
    },
    {
      "epoch": 0.03469,
      "grad_norm": 0.574247040112387,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3469
    },
    {
      "epoch": 0.0347,
      "grad_norm": 0.5474693168535483,
      "learning_rate": 0.003,
      "loss": 4.1868,
      "step": 3470
    },
    {
      "epoch": 0.03471,
      "grad_norm": 0.5484056382186252,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3471
    },
    {
      "epoch": 0.03472,
      "grad_norm": 0.548625120805519,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 3472
    },
    {
      "epoch": 0.03473,
      "grad_norm": 0.637401868814706,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3473
    },
    {
      "epoch": 0.03474,
      "grad_norm": 0.7144883444351541,
      "learning_rate": 0.003,
      "loss": 4.2012,
      "step": 3474
    },
    {
      "epoch": 0.03475,
      "grad_norm": 0.7979670963387884,
      "learning_rate": 0.003,
      "loss": 4.1787,
      "step": 3475
    },
    {
      "epoch": 0.03476,
      "grad_norm": 0.7163973548344581,
      "learning_rate": 0.003,
      "loss": 4.1814,
      "step": 3476
    },
    {
      "epoch": 0.03477,
      "grad_norm": 0.5684898388546081,
      "learning_rate": 0.003,
      "loss": 4.1719,
      "step": 3477
    },
    {
      "epoch": 0.03478,
      "grad_norm": 0.5861592780845318,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 3478
    },
    {
      "epoch": 0.03479,
      "grad_norm": 0.5340943622573318,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 3479
    },
    {
      "epoch": 0.0348,
      "grad_norm": 0.4404622712681107,
      "learning_rate": 0.003,
      "loss": 4.1503,
      "step": 3480
    },
    {
      "epoch": 0.03481,
      "grad_norm": 0.4792980762796971,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3481
    },
    {
      "epoch": 0.03482,
      "grad_norm": 0.4579076307590615,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3482
    },
    {
      "epoch": 0.03483,
      "grad_norm": 0.42854750510723455,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 3483
    },
    {
      "epoch": 0.03484,
      "grad_norm": 0.4222950210388672,
      "learning_rate": 0.003,
      "loss": 4.1912,
      "step": 3484
    },
    {
      "epoch": 0.03485,
      "grad_norm": 0.47787917501254906,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 3485
    },
    {
      "epoch": 0.03486,
      "grad_norm": 0.6218177691290887,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 3486
    },
    {
      "epoch": 0.03487,
      "grad_norm": 0.7695122506406105,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 3487
    },
    {
      "epoch": 0.03488,
      "grad_norm": 0.811403718151084,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 3488
    },
    {
      "epoch": 0.03489,
      "grad_norm": 0.6478567488540471,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 3489
    },
    {
      "epoch": 0.0349,
      "grad_norm": 0.5349166330706258,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 3490
    },
    {
      "epoch": 0.03491,
      "grad_norm": 0.5539220367541061,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 3491
    },
    {
      "epoch": 0.03492,
      "grad_norm": 0.4809491803196432,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3492
    },
    {
      "epoch": 0.03493,
      "grad_norm": 0.4980699721500476,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 3493
    },
    {
      "epoch": 0.03494,
      "grad_norm": 0.6241704569078306,
      "learning_rate": 0.003,
      "loss": 4.161,
      "step": 3494
    },
    {
      "epoch": 0.03495,
      "grad_norm": 0.6045742174777644,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 3495
    },
    {
      "epoch": 0.03496,
      "grad_norm": 0.601044244149248,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 3496
    },
    {
      "epoch": 0.03497,
      "grad_norm": 0.5400771740929299,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 3497
    },
    {
      "epoch": 0.03498,
      "grad_norm": 0.713136808858786,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 3498
    },
    {
      "epoch": 0.03499,
      "grad_norm": 0.7906121138775885,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 3499
    },
    {
      "epoch": 0.035,
      "grad_norm": 0.7463261361411888,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 3500
    },
    {
      "epoch": 0.03501,
      "grad_norm": 0.6401328140167164,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 3501
    },
    {
      "epoch": 0.03502,
      "grad_norm": 0.5849954396740931,
      "learning_rate": 0.003,
      "loss": 4.1761,
      "step": 3502
    },
    {
      "epoch": 0.03503,
      "grad_norm": 0.5132547849361668,
      "learning_rate": 0.003,
      "loss": 4.1879,
      "step": 3503
    },
    {
      "epoch": 0.03504,
      "grad_norm": 0.5053019189532232,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3504
    },
    {
      "epoch": 0.03505,
      "grad_norm": 0.5412370735339649,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3505
    },
    {
      "epoch": 0.03506,
      "grad_norm": 0.5174741400250557,
      "learning_rate": 0.003,
      "loss": 4.1855,
      "step": 3506
    },
    {
      "epoch": 0.03507,
      "grad_norm": 0.5828929341462858,
      "learning_rate": 0.003,
      "loss": 4.1653,
      "step": 3507
    },
    {
      "epoch": 0.03508,
      "grad_norm": 0.6123621534119361,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 3508
    },
    {
      "epoch": 0.03509,
      "grad_norm": 0.529763454477017,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3509
    },
    {
      "epoch": 0.0351,
      "grad_norm": 0.4991440774069516,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 3510
    },
    {
      "epoch": 0.03511,
      "grad_norm": 0.5106927805679168,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 3511
    },
    {
      "epoch": 0.03512,
      "grad_norm": 0.5290709648712968,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 3512
    },
    {
      "epoch": 0.03513,
      "grad_norm": 0.5838748180600412,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 3513
    },
    {
      "epoch": 0.03514,
      "grad_norm": 0.8057246907688399,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3514
    },
    {
      "epoch": 0.03515,
      "grad_norm": 1.0135173467038623,
      "learning_rate": 0.003,
      "loss": 4.1945,
      "step": 3515
    },
    {
      "epoch": 0.03516,
      "grad_norm": 0.8727734267667479,
      "learning_rate": 0.003,
      "loss": 4.2288,
      "step": 3516
    },
    {
      "epoch": 0.03517,
      "grad_norm": 0.7059545473280222,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 3517
    },
    {
      "epoch": 0.03518,
      "grad_norm": 0.7000960321874965,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 3518
    },
    {
      "epoch": 0.03519,
      "grad_norm": 0.6627465613911154,
      "learning_rate": 0.003,
      "loss": 4.2036,
      "step": 3519
    },
    {
      "epoch": 0.0352,
      "grad_norm": 0.7079684591627484,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 3520
    },
    {
      "epoch": 0.03521,
      "grad_norm": 0.7564247406144935,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 3521
    },
    {
      "epoch": 0.03522,
      "grad_norm": 0.6970141789263714,
      "learning_rate": 0.003,
      "loss": 4.1807,
      "step": 3522
    },
    {
      "epoch": 0.03523,
      "grad_norm": 0.6458343838785704,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 3523
    },
    {
      "epoch": 0.03524,
      "grad_norm": 0.727411286446386,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 3524
    },
    {
      "epoch": 0.03525,
      "grad_norm": 0.6885063026654827,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 3525
    },
    {
      "epoch": 0.03526,
      "grad_norm": 0.503068225605494,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 3526
    },
    {
      "epoch": 0.03527,
      "grad_norm": 0.4984239052259041,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 3527
    },
    {
      "epoch": 0.03528,
      "grad_norm": 0.48402535036066135,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3528
    },
    {
      "epoch": 0.03529,
      "grad_norm": 0.4945614778246426,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 3529
    },
    {
      "epoch": 0.0353,
      "grad_norm": 0.488546537630184,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 3530
    },
    {
      "epoch": 0.03531,
      "grad_norm": 0.4615658807568464,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3531
    },
    {
      "epoch": 0.03532,
      "grad_norm": 0.5005896772829939,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3532
    },
    {
      "epoch": 0.03533,
      "grad_norm": 0.5146344370982837,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 3533
    },
    {
      "epoch": 0.03534,
      "grad_norm": 0.5278648337419918,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 3534
    },
    {
      "epoch": 0.03535,
      "grad_norm": 0.573214872592988,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 3535
    },
    {
      "epoch": 0.03536,
      "grad_norm": 0.6216412138977996,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 3536
    },
    {
      "epoch": 0.03537,
      "grad_norm": 0.641334014656523,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3537
    },
    {
      "epoch": 0.03538,
      "grad_norm": 0.7120173819031035,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 3538
    },
    {
      "epoch": 0.03539,
      "grad_norm": 0.7562087971985374,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 3539
    },
    {
      "epoch": 0.0354,
      "grad_norm": 0.745826443201787,
      "learning_rate": 0.003,
      "loss": 4.1925,
      "step": 3540
    },
    {
      "epoch": 0.03541,
      "grad_norm": 0.6880220783097426,
      "learning_rate": 0.003,
      "loss": 4.1777,
      "step": 3541
    },
    {
      "epoch": 0.03542,
      "grad_norm": 0.6660987603699731,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 3542
    },
    {
      "epoch": 0.03543,
      "grad_norm": 0.5966852103308192,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 3543
    },
    {
      "epoch": 0.03544,
      "grad_norm": 0.6147196669679205,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 3544
    },
    {
      "epoch": 0.03545,
      "grad_norm": 0.5876025358116799,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3545
    },
    {
      "epoch": 0.03546,
      "grad_norm": 0.5394946467408606,
      "learning_rate": 0.003,
      "loss": 4.161,
      "step": 3546
    },
    {
      "epoch": 0.03547,
      "grad_norm": 0.4760714738457914,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3547
    },
    {
      "epoch": 0.03548,
      "grad_norm": 0.511046276124325,
      "learning_rate": 0.003,
      "loss": 4.1712,
      "step": 3548
    },
    {
      "epoch": 0.03549,
      "grad_norm": 0.5497663152303965,
      "learning_rate": 0.003,
      "loss": 4.1676,
      "step": 3549
    },
    {
      "epoch": 0.0355,
      "grad_norm": 0.5734708015673942,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 3550
    },
    {
      "epoch": 0.03551,
      "grad_norm": 0.5704932035569535,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 3551
    },
    {
      "epoch": 0.03552,
      "grad_norm": 0.5413961705528875,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 3552
    },
    {
      "epoch": 0.03553,
      "grad_norm": 0.5629706262237084,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 3553
    },
    {
      "epoch": 0.03554,
      "grad_norm": 0.6847813269812701,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 3554
    },
    {
      "epoch": 0.03555,
      "grad_norm": 0.7442929788518159,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 3555
    },
    {
      "epoch": 0.03556,
      "grad_norm": 0.6995396794139108,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3556
    },
    {
      "epoch": 0.03557,
      "grad_norm": 0.6870977550247097,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 3557
    },
    {
      "epoch": 0.03558,
      "grad_norm": 0.4853466992283488,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 3558
    },
    {
      "epoch": 0.03559,
      "grad_norm": 0.4762880827808817,
      "learning_rate": 0.003,
      "loss": 4.1691,
      "step": 3559
    },
    {
      "epoch": 0.0356,
      "grad_norm": 0.44888469164051914,
      "learning_rate": 0.003,
      "loss": 4.1691,
      "step": 3560
    },
    {
      "epoch": 0.03561,
      "grad_norm": 0.4860772847324368,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 3561
    },
    {
      "epoch": 0.03562,
      "grad_norm": 0.609514740868879,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 3562
    },
    {
      "epoch": 0.03563,
      "grad_norm": 0.5645288584044179,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 3563
    },
    {
      "epoch": 0.03564,
      "grad_norm": 0.5713508531485367,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 3564
    },
    {
      "epoch": 0.03565,
      "grad_norm": 0.6809281314446707,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 3565
    },
    {
      "epoch": 0.03566,
      "grad_norm": 0.8446250147895774,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 3566
    },
    {
      "epoch": 0.03567,
      "grad_norm": 0.7816404227486906,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3567
    },
    {
      "epoch": 0.03568,
      "grad_norm": 0.565515569510055,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 3568
    },
    {
      "epoch": 0.03569,
      "grad_norm": 0.6392539102485247,
      "learning_rate": 0.003,
      "loss": 4.1604,
      "step": 3569
    },
    {
      "epoch": 0.0357,
      "grad_norm": 0.6094618901019504,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3570
    },
    {
      "epoch": 0.03571,
      "grad_norm": 0.516014245189601,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 3571
    },
    {
      "epoch": 0.03572,
      "grad_norm": 0.5649956960536694,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 3572
    },
    {
      "epoch": 0.03573,
      "grad_norm": 0.5285488438884934,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 3573
    },
    {
      "epoch": 0.03574,
      "grad_norm": 0.5059015354099536,
      "learning_rate": 0.003,
      "loss": 4.1778,
      "step": 3574
    },
    {
      "epoch": 0.03575,
      "grad_norm": 0.5626349168929159,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3575
    },
    {
      "epoch": 0.03576,
      "grad_norm": 0.5730355700030897,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 3576
    },
    {
      "epoch": 0.03577,
      "grad_norm": 0.5081514156321213,
      "learning_rate": 0.003,
      "loss": 4.1719,
      "step": 3577
    },
    {
      "epoch": 0.03578,
      "grad_norm": 0.4522800880743083,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 3578
    },
    {
      "epoch": 0.03579,
      "grad_norm": 0.4334604254761663,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3579
    },
    {
      "epoch": 0.0358,
      "grad_norm": 0.478675617542924,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 3580
    },
    {
      "epoch": 0.03581,
      "grad_norm": 0.5103431984153175,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 3581
    },
    {
      "epoch": 0.03582,
      "grad_norm": 0.5133704031348394,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 3582
    },
    {
      "epoch": 0.03583,
      "grad_norm": 0.5875445897748297,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 3583
    },
    {
      "epoch": 0.03584,
      "grad_norm": 0.5543663682322499,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 3584
    },
    {
      "epoch": 0.03585,
      "grad_norm": 0.49560990112191017,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 3585
    },
    {
      "epoch": 0.03586,
      "grad_norm": 0.5080536513623409,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3586
    },
    {
      "epoch": 0.03587,
      "grad_norm": 0.47897066887270817,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 3587
    },
    {
      "epoch": 0.03588,
      "grad_norm": 0.4859791474646478,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3588
    },
    {
      "epoch": 0.03589,
      "grad_norm": 0.5360881510224248,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 3589
    },
    {
      "epoch": 0.0359,
      "grad_norm": 0.6282942855775232,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 3590
    },
    {
      "epoch": 0.03591,
      "grad_norm": 0.7149898709877768,
      "learning_rate": 0.003,
      "loss": 4.1753,
      "step": 3591
    },
    {
      "epoch": 0.03592,
      "grad_norm": 0.6656767655390144,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 3592
    },
    {
      "epoch": 0.03593,
      "grad_norm": 0.5744352573697437,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 3593
    },
    {
      "epoch": 0.03594,
      "grad_norm": 0.5564608607734087,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 3594
    },
    {
      "epoch": 0.03595,
      "grad_norm": 0.6440817185713912,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3595
    },
    {
      "epoch": 0.03596,
      "grad_norm": 0.8208781786922031,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 3596
    },
    {
      "epoch": 0.03597,
      "grad_norm": 0.9222891823782106,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 3597
    },
    {
      "epoch": 0.03598,
      "grad_norm": 0.8427211641310154,
      "learning_rate": 0.003,
      "loss": 4.187,
      "step": 3598
    },
    {
      "epoch": 0.03599,
      "grad_norm": 0.852351507152875,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 3599
    },
    {
      "epoch": 0.036,
      "grad_norm": 0.8507559173234883,
      "learning_rate": 0.003,
      "loss": 4.2187,
      "step": 3600
    },
    {
      "epoch": 0.03601,
      "grad_norm": 0.9636100541843863,
      "learning_rate": 0.003,
      "loss": 4.1988,
      "step": 3601
    },
    {
      "epoch": 0.03602,
      "grad_norm": 1.043053481597312,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 3602
    },
    {
      "epoch": 0.03603,
      "grad_norm": 0.8586008867305468,
      "learning_rate": 0.003,
      "loss": 4.1772,
      "step": 3603
    },
    {
      "epoch": 0.03604,
      "grad_norm": 0.8633622676072363,
      "learning_rate": 0.003,
      "loss": 4.2215,
      "step": 3604
    },
    {
      "epoch": 0.03605,
      "grad_norm": 0.8221166093366499,
      "learning_rate": 0.003,
      "loss": 4.213,
      "step": 3605
    },
    {
      "epoch": 0.03606,
      "grad_norm": 0.7191787485946937,
      "learning_rate": 0.003,
      "loss": 4.1949,
      "step": 3606
    },
    {
      "epoch": 0.03607,
      "grad_norm": 0.6624835833221929,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 3607
    },
    {
      "epoch": 0.03608,
      "grad_norm": 0.6824569834328487,
      "learning_rate": 0.003,
      "loss": 4.1883,
      "step": 3608
    },
    {
      "epoch": 0.03609,
      "grad_norm": 0.6457675177801978,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 3609
    },
    {
      "epoch": 0.0361,
      "grad_norm": 0.6295884180353496,
      "learning_rate": 0.003,
      "loss": 4.2142,
      "step": 3610
    },
    {
      "epoch": 0.03611,
      "grad_norm": 0.590169158296821,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3611
    },
    {
      "epoch": 0.03612,
      "grad_norm": 0.5389921423268351,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 3612
    },
    {
      "epoch": 0.03613,
      "grad_norm": 0.5497537704200285,
      "learning_rate": 0.003,
      "loss": 4.1961,
      "step": 3613
    },
    {
      "epoch": 0.03614,
      "grad_norm": 0.4843702888139074,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 3614
    },
    {
      "epoch": 0.03615,
      "grad_norm": 0.48025803520879506,
      "learning_rate": 0.003,
      "loss": 4.1788,
      "step": 3615
    },
    {
      "epoch": 0.03616,
      "grad_norm": 0.46348349208320905,
      "learning_rate": 0.003,
      "loss": 4.1857,
      "step": 3616
    },
    {
      "epoch": 0.03617,
      "grad_norm": 0.39854001217238155,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 3617
    },
    {
      "epoch": 0.03618,
      "grad_norm": 0.4027281929968038,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3618
    },
    {
      "epoch": 0.03619,
      "grad_norm": 0.3985420323335125,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3619
    },
    {
      "epoch": 0.0362,
      "grad_norm": 0.3613468610024924,
      "learning_rate": 0.003,
      "loss": 4.1676,
      "step": 3620
    },
    {
      "epoch": 0.03621,
      "grad_norm": 0.33261587429198597,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 3621
    },
    {
      "epoch": 0.03622,
      "grad_norm": 0.3403864350330926,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 3622
    },
    {
      "epoch": 0.03623,
      "grad_norm": 0.3583311023359255,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 3623
    },
    {
      "epoch": 0.03624,
      "grad_norm": 0.32450649557353767,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 3624
    },
    {
      "epoch": 0.03625,
      "grad_norm": 0.3235232357698917,
      "learning_rate": 0.003,
      "loss": 4.1676,
      "step": 3625
    },
    {
      "epoch": 0.03626,
      "grad_norm": 0.3029552071156344,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 3626
    },
    {
      "epoch": 0.03627,
      "grad_norm": 0.3634806878840478,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 3627
    },
    {
      "epoch": 0.03628,
      "grad_norm": 0.5031788929713537,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 3628
    },
    {
      "epoch": 0.03629,
      "grad_norm": 0.7314661052924161,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 3629
    },
    {
      "epoch": 0.0363,
      "grad_norm": 0.9369830066901181,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 3630
    },
    {
      "epoch": 0.03631,
      "grad_norm": 0.7563300299386175,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3631
    },
    {
      "epoch": 0.03632,
      "grad_norm": 0.6272780473551262,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 3632
    },
    {
      "epoch": 0.03633,
      "grad_norm": 0.6625289405061192,
      "learning_rate": 0.003,
      "loss": 4.1907,
      "step": 3633
    },
    {
      "epoch": 0.03634,
      "grad_norm": 0.624342554157624,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 3634
    },
    {
      "epoch": 0.03635,
      "grad_norm": 0.583607515091326,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3635
    },
    {
      "epoch": 0.03636,
      "grad_norm": 0.5584832759878093,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3636
    },
    {
      "epoch": 0.03637,
      "grad_norm": 0.5837086725119016,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 3637
    },
    {
      "epoch": 0.03638,
      "grad_norm": 0.6095521315464627,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 3638
    },
    {
      "epoch": 0.03639,
      "grad_norm": 0.6657809000387206,
      "learning_rate": 0.003,
      "loss": 4.1922,
      "step": 3639
    },
    {
      "epoch": 0.0364,
      "grad_norm": 0.6289647060691178,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 3640
    },
    {
      "epoch": 0.03641,
      "grad_norm": 0.6059605963156967,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 3641
    },
    {
      "epoch": 0.03642,
      "grad_norm": 0.5705396060730448,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 3642
    },
    {
      "epoch": 0.03643,
      "grad_norm": 0.6035912533971294,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 3643
    },
    {
      "epoch": 0.03644,
      "grad_norm": 0.6309633478231041,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 3644
    },
    {
      "epoch": 0.03645,
      "grad_norm": 0.6355334191850317,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 3645
    },
    {
      "epoch": 0.03646,
      "grad_norm": 0.5786595124970733,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 3646
    },
    {
      "epoch": 0.03647,
      "grad_norm": 0.5949997765584221,
      "learning_rate": 0.003,
      "loss": 4.1642,
      "step": 3647
    },
    {
      "epoch": 0.03648,
      "grad_norm": 0.5464060834426805,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 3648
    },
    {
      "epoch": 0.03649,
      "grad_norm": 0.6161442180745826,
      "learning_rate": 0.003,
      "loss": 4.1777,
      "step": 3649
    },
    {
      "epoch": 0.0365,
      "grad_norm": 0.7276354515874378,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 3650
    },
    {
      "epoch": 0.03651,
      "grad_norm": 0.879246794128947,
      "learning_rate": 0.003,
      "loss": 4.1938,
      "step": 3651
    },
    {
      "epoch": 0.03652,
      "grad_norm": 0.8889526606107412,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 3652
    },
    {
      "epoch": 0.03653,
      "grad_norm": 0.7646404951633242,
      "learning_rate": 0.003,
      "loss": 4.162,
      "step": 3653
    },
    {
      "epoch": 0.03654,
      "grad_norm": 0.7290317207893501,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 3654
    },
    {
      "epoch": 0.03655,
      "grad_norm": 0.7063774172804314,
      "learning_rate": 0.003,
      "loss": 4.1865,
      "step": 3655
    },
    {
      "epoch": 0.03656,
      "grad_norm": 0.7160866471184082,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3656
    },
    {
      "epoch": 0.03657,
      "grad_norm": 0.5821614514235085,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 3657
    },
    {
      "epoch": 0.03658,
      "grad_norm": 0.5963209078366205,
      "learning_rate": 0.003,
      "loss": 4.2061,
      "step": 3658
    },
    {
      "epoch": 0.03659,
      "grad_norm": 0.5885163411512104,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 3659
    },
    {
      "epoch": 0.0366,
      "grad_norm": 0.6724966576985278,
      "learning_rate": 0.003,
      "loss": 4.1869,
      "step": 3660
    },
    {
      "epoch": 0.03661,
      "grad_norm": 0.70045611677563,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 3661
    },
    {
      "epoch": 0.03662,
      "grad_norm": 0.7038628116184068,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 3662
    },
    {
      "epoch": 0.03663,
      "grad_norm": 0.7338499926784156,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 3663
    },
    {
      "epoch": 0.03664,
      "grad_norm": 0.6003408827366159,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 3664
    },
    {
      "epoch": 0.03665,
      "grad_norm": 0.531193738984965,
      "learning_rate": 0.003,
      "loss": 4.1816,
      "step": 3665
    },
    {
      "epoch": 0.03666,
      "grad_norm": 0.502967345745313,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 3666
    },
    {
      "epoch": 0.03667,
      "grad_norm": 0.4403238489897598,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 3667
    },
    {
      "epoch": 0.03668,
      "grad_norm": 0.4214554821381715,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 3668
    },
    {
      "epoch": 0.03669,
      "grad_norm": 0.3786750496308303,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 3669
    },
    {
      "epoch": 0.0367,
      "grad_norm": 0.36562759077361934,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 3670
    },
    {
      "epoch": 0.03671,
      "grad_norm": 0.4132800551107581,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 3671
    },
    {
      "epoch": 0.03672,
      "grad_norm": 0.49256525507220217,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 3672
    },
    {
      "epoch": 0.03673,
      "grad_norm": 0.6349831894939928,
      "learning_rate": 0.003,
      "loss": 4.171,
      "step": 3673
    },
    {
      "epoch": 0.03674,
      "grad_norm": 0.6952242122376082,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 3674
    },
    {
      "epoch": 0.03675,
      "grad_norm": 0.5660824888307048,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 3675
    },
    {
      "epoch": 0.03676,
      "grad_norm": 0.4101727822433613,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 3676
    },
    {
      "epoch": 0.03677,
      "grad_norm": 0.4248844160991166,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 3677
    },
    {
      "epoch": 0.03678,
      "grad_norm": 0.40692106726610927,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 3678
    },
    {
      "epoch": 0.03679,
      "grad_norm": 0.43656051390996525,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3679
    },
    {
      "epoch": 0.0368,
      "grad_norm": 0.4651408055341787,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 3680
    },
    {
      "epoch": 0.03681,
      "grad_norm": 0.4431552362299314,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 3681
    },
    {
      "epoch": 0.03682,
      "grad_norm": 0.45670367087714675,
      "learning_rate": 0.003,
      "loss": 4.1622,
      "step": 3682
    },
    {
      "epoch": 0.03683,
      "grad_norm": 0.48732877269273184,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 3683
    },
    {
      "epoch": 0.03684,
      "grad_norm": 0.49569158763050314,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 3684
    },
    {
      "epoch": 0.03685,
      "grad_norm": 0.5827537649992007,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 3685
    },
    {
      "epoch": 0.03686,
      "grad_norm": 0.7130056023766307,
      "learning_rate": 0.003,
      "loss": 4.1622,
      "step": 3686
    },
    {
      "epoch": 0.03687,
      "grad_norm": 0.806080078993977,
      "learning_rate": 0.003,
      "loss": 4.1913,
      "step": 3687
    },
    {
      "epoch": 0.03688,
      "grad_norm": 0.8654301422344852,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 3688
    },
    {
      "epoch": 0.03689,
      "grad_norm": 0.824342247624797,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 3689
    },
    {
      "epoch": 0.0369,
      "grad_norm": 0.8399398898767244,
      "learning_rate": 0.003,
      "loss": 4.1897,
      "step": 3690
    },
    {
      "epoch": 0.03691,
      "grad_norm": 0.8376909355521104,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3691
    },
    {
      "epoch": 0.03692,
      "grad_norm": 0.7664237094520844,
      "learning_rate": 0.003,
      "loss": 4.2095,
      "step": 3692
    },
    {
      "epoch": 0.03693,
      "grad_norm": 0.6756940227740064,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 3693
    },
    {
      "epoch": 0.03694,
      "grad_norm": 0.665602669017162,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3694
    },
    {
      "epoch": 0.03695,
      "grad_norm": 0.6845010372537103,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 3695
    },
    {
      "epoch": 0.03696,
      "grad_norm": 0.8052078414912427,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3696
    },
    {
      "epoch": 0.03697,
      "grad_norm": 0.8787348855409773,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 3697
    },
    {
      "epoch": 0.03698,
      "grad_norm": 0.8247546282969409,
      "learning_rate": 0.003,
      "loss": 4.2121,
      "step": 3698
    },
    {
      "epoch": 0.03699,
      "grad_norm": 0.7409801258428377,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 3699
    },
    {
      "epoch": 0.037,
      "grad_norm": 0.6088451406774416,
      "learning_rate": 0.003,
      "loss": 4.1795,
      "step": 3700
    },
    {
      "epoch": 0.03701,
      "grad_norm": 0.6322208971568581,
      "learning_rate": 0.003,
      "loss": 4.1844,
      "step": 3701
    },
    {
      "epoch": 0.03702,
      "grad_norm": 0.5054476642910108,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3702
    },
    {
      "epoch": 0.03703,
      "grad_norm": 0.5443386034292382,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 3703
    },
    {
      "epoch": 0.03704,
      "grad_norm": 0.5944500200700061,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 3704
    },
    {
      "epoch": 0.03705,
      "grad_norm": 0.600099616356784,
      "learning_rate": 0.003,
      "loss": 4.1957,
      "step": 3705
    },
    {
      "epoch": 0.03706,
      "grad_norm": 0.6109788360926119,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3706
    },
    {
      "epoch": 0.03707,
      "grad_norm": 0.5293706856345912,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3707
    },
    {
      "epoch": 0.03708,
      "grad_norm": 0.47791373263744796,
      "learning_rate": 0.003,
      "loss": 4.1753,
      "step": 3708
    },
    {
      "epoch": 0.03709,
      "grad_norm": 0.49156286964072154,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 3709
    },
    {
      "epoch": 0.0371,
      "grad_norm": 0.5248101152432202,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 3710
    },
    {
      "epoch": 0.03711,
      "grad_norm": 0.6117031890743618,
      "learning_rate": 0.003,
      "loss": 4.1604,
      "step": 3711
    },
    {
      "epoch": 0.03712,
      "grad_norm": 0.6254066152554677,
      "learning_rate": 0.003,
      "loss": 4.1614,
      "step": 3712
    },
    {
      "epoch": 0.03713,
      "grad_norm": 0.6264703701067362,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 3713
    },
    {
      "epoch": 0.03714,
      "grad_norm": 0.620653200418187,
      "learning_rate": 0.003,
      "loss": 4.1614,
      "step": 3714
    },
    {
      "epoch": 0.03715,
      "grad_norm": 0.5248205552325349,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3715
    },
    {
      "epoch": 0.03716,
      "grad_norm": 0.5526084707582196,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 3716
    },
    {
      "epoch": 0.03717,
      "grad_norm": 0.49730839862475784,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3717
    },
    {
      "epoch": 0.03718,
      "grad_norm": 0.462962673452178,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 3718
    },
    {
      "epoch": 0.03719,
      "grad_norm": 0.3930444861352458,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 3719
    },
    {
      "epoch": 0.0372,
      "grad_norm": 0.40801634488970573,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 3720
    },
    {
      "epoch": 0.03721,
      "grad_norm": 0.44606614802333183,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3721
    },
    {
      "epoch": 0.03722,
      "grad_norm": 0.39111226545505906,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 3722
    },
    {
      "epoch": 0.03723,
      "grad_norm": 0.4408949322752162,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3723
    },
    {
      "epoch": 0.03724,
      "grad_norm": 0.5569410843865424,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 3724
    },
    {
      "epoch": 0.03725,
      "grad_norm": 0.6820788875058175,
      "learning_rate": 0.003,
      "loss": 4.1691,
      "step": 3725
    },
    {
      "epoch": 0.03726,
      "grad_norm": 0.7697397803396343,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 3726
    },
    {
      "epoch": 0.03727,
      "grad_norm": 0.7112467858706011,
      "learning_rate": 0.003,
      "loss": 4.1828,
      "step": 3727
    },
    {
      "epoch": 0.03728,
      "grad_norm": 0.6285060537833618,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 3728
    },
    {
      "epoch": 0.03729,
      "grad_norm": 0.5988137930316242,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3729
    },
    {
      "epoch": 0.0373,
      "grad_norm": 0.6000038779888225,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 3730
    },
    {
      "epoch": 0.03731,
      "grad_norm": 0.5727776342995737,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 3731
    },
    {
      "epoch": 0.03732,
      "grad_norm": 0.47697561747753886,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 3732
    },
    {
      "epoch": 0.03733,
      "grad_norm": 0.44272447557143685,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 3733
    },
    {
      "epoch": 0.03734,
      "grad_norm": 0.4247027107702777,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 3734
    },
    {
      "epoch": 0.03735,
      "grad_norm": 0.4479925161214012,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 3735
    },
    {
      "epoch": 0.03736,
      "grad_norm": 0.45973685869578795,
      "learning_rate": 0.003,
      "loss": 4.1686,
      "step": 3736
    },
    {
      "epoch": 0.03737,
      "grad_norm": 0.3920498911497691,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 3737
    },
    {
      "epoch": 0.03738,
      "grad_norm": 0.4063251593566824,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3738
    },
    {
      "epoch": 0.03739,
      "grad_norm": 0.46745772862008544,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 3739
    },
    {
      "epoch": 0.0374,
      "grad_norm": 0.46296090379338223,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 3740
    },
    {
      "epoch": 0.03741,
      "grad_norm": 0.512997084501291,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 3741
    },
    {
      "epoch": 0.03742,
      "grad_norm": 0.6523245986952474,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 3742
    },
    {
      "epoch": 0.03743,
      "grad_norm": 0.7442806374508512,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 3743
    },
    {
      "epoch": 0.03744,
      "grad_norm": 0.7320455988631113,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 3744
    },
    {
      "epoch": 0.03745,
      "grad_norm": 0.7624968676067105,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 3745
    },
    {
      "epoch": 0.03746,
      "grad_norm": 0.7031928628117834,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 3746
    },
    {
      "epoch": 0.03747,
      "grad_norm": 0.6687793682231787,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 3747
    },
    {
      "epoch": 0.03748,
      "grad_norm": 0.779694731392804,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 3748
    },
    {
      "epoch": 0.03749,
      "grad_norm": 0.7516521349267218,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3749
    },
    {
      "epoch": 0.0375,
      "grad_norm": 0.7321877471048407,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 3750
    },
    {
      "epoch": 0.03751,
      "grad_norm": 0.7130679243817225,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 3751
    },
    {
      "epoch": 0.03752,
      "grad_norm": 0.7452672741472625,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 3752
    },
    {
      "epoch": 0.03753,
      "grad_norm": 0.6724770863912444,
      "learning_rate": 0.003,
      "loss": 4.1729,
      "step": 3753
    },
    {
      "epoch": 0.03754,
      "grad_norm": 0.556817823672953,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 3754
    },
    {
      "epoch": 0.03755,
      "grad_norm": 0.5579634308288945,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 3755
    },
    {
      "epoch": 0.03756,
      "grad_norm": 0.6263580644537361,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 3756
    },
    {
      "epoch": 0.03757,
      "grad_norm": 0.6400148872312791,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 3757
    },
    {
      "epoch": 0.03758,
      "grad_norm": 0.6369106854442147,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 3758
    },
    {
      "epoch": 0.03759,
      "grad_norm": 0.5476477808333663,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 3759
    },
    {
      "epoch": 0.0376,
      "grad_norm": 0.6773838858228983,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 3760
    },
    {
      "epoch": 0.03761,
      "grad_norm": 0.7718542722389751,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 3761
    },
    {
      "epoch": 0.03762,
      "grad_norm": 0.6657955446674061,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 3762
    },
    {
      "epoch": 0.03763,
      "grad_norm": 0.6194620722466222,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3763
    },
    {
      "epoch": 0.03764,
      "grad_norm": 0.6595957894013659,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 3764
    },
    {
      "epoch": 0.03765,
      "grad_norm": 0.7419338588721475,
      "learning_rate": 0.003,
      "loss": 4.2111,
      "step": 3765
    },
    {
      "epoch": 0.03766,
      "grad_norm": 0.7575024078950962,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 3766
    },
    {
      "epoch": 0.03767,
      "grad_norm": 0.6434009404162437,
      "learning_rate": 0.003,
      "loss": 4.1896,
      "step": 3767
    },
    {
      "epoch": 0.03768,
      "grad_norm": 0.6051033538599371,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3768
    },
    {
      "epoch": 0.03769,
      "grad_norm": 0.581660518608867,
      "learning_rate": 0.003,
      "loss": 4.1566,
      "step": 3769
    },
    {
      "epoch": 0.0377,
      "grad_norm": 0.47329598157123676,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 3770
    },
    {
      "epoch": 0.03771,
      "grad_norm": 0.5353355178607329,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 3771
    },
    {
      "epoch": 0.03772,
      "grad_norm": 0.5081985686041285,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 3772
    },
    {
      "epoch": 0.03773,
      "grad_norm": 0.4994313050826756,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 3773
    },
    {
      "epoch": 0.03774,
      "grad_norm": 0.48919774898243473,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 3774
    },
    {
      "epoch": 0.03775,
      "grad_norm": 0.46915381954128726,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 3775
    },
    {
      "epoch": 0.03776,
      "grad_norm": 0.46730858925326924,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 3776
    },
    {
      "epoch": 0.03777,
      "grad_norm": 0.4868019030635906,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 3777
    },
    {
      "epoch": 0.03778,
      "grad_norm": 0.4826030300886077,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 3778
    },
    {
      "epoch": 0.03779,
      "grad_norm": 0.4347204008546579,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 3779
    },
    {
      "epoch": 0.0378,
      "grad_norm": 0.4013702620235084,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3780
    },
    {
      "epoch": 0.03781,
      "grad_norm": 0.4909799595844497,
      "learning_rate": 0.003,
      "loss": 4.1807,
      "step": 3781
    },
    {
      "epoch": 0.03782,
      "grad_norm": 0.5827570172997248,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 3782
    },
    {
      "epoch": 0.03783,
      "grad_norm": 0.7057652047796394,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 3783
    },
    {
      "epoch": 0.03784,
      "grad_norm": 0.7176846202765269,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 3784
    },
    {
      "epoch": 0.03785,
      "grad_norm": 0.6227082605371008,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 3785
    },
    {
      "epoch": 0.03786,
      "grad_norm": 0.5851856366466184,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 3786
    },
    {
      "epoch": 0.03787,
      "grad_norm": 0.5827445998978293,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 3787
    },
    {
      "epoch": 0.03788,
      "grad_norm": 0.5961875200035882,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 3788
    },
    {
      "epoch": 0.03789,
      "grad_norm": 0.5599813731691621,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 3789
    },
    {
      "epoch": 0.0379,
      "grad_norm": 0.5457520494795979,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 3790
    },
    {
      "epoch": 0.03791,
      "grad_norm": 0.5890367342982761,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 3791
    },
    {
      "epoch": 0.03792,
      "grad_norm": 0.6819220498588953,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 3792
    },
    {
      "epoch": 0.03793,
      "grad_norm": 0.7245933962883803,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 3793
    },
    {
      "epoch": 0.03794,
      "grad_norm": 0.6330889319321834,
      "learning_rate": 0.003,
      "loss": 4.1803,
      "step": 3794
    },
    {
      "epoch": 0.03795,
      "grad_norm": 0.6825420551312804,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3795
    },
    {
      "epoch": 0.03796,
      "grad_norm": 0.7213382120258517,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 3796
    },
    {
      "epoch": 0.03797,
      "grad_norm": 0.753053475755318,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 3797
    },
    {
      "epoch": 0.03798,
      "grad_norm": 0.8024496845778784,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 3798
    },
    {
      "epoch": 0.03799,
      "grad_norm": 0.6761853816469585,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 3799
    },
    {
      "epoch": 0.038,
      "grad_norm": 0.5760561821833747,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3800
    },
    {
      "epoch": 0.03801,
      "grad_norm": 0.5877988869253646,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 3801
    },
    {
      "epoch": 0.03802,
      "grad_norm": 0.6488463807188865,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 3802
    },
    {
      "epoch": 0.03803,
      "grad_norm": 0.5517421440712098,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3803
    },
    {
      "epoch": 0.03804,
      "grad_norm": 0.5096073674154153,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 3804
    },
    {
      "epoch": 0.03805,
      "grad_norm": 0.5032605295974328,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 3805
    },
    {
      "epoch": 0.03806,
      "grad_norm": 0.515161268146513,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 3806
    },
    {
      "epoch": 0.03807,
      "grad_norm": 0.5868281956315563,
      "learning_rate": 0.003,
      "loss": 4.154,
      "step": 3807
    },
    {
      "epoch": 0.03808,
      "grad_norm": 0.6126977458856417,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 3808
    },
    {
      "epoch": 0.03809,
      "grad_norm": 0.5958831138264383,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3809
    },
    {
      "epoch": 0.0381,
      "grad_norm": 0.5125381608413851,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 3810
    },
    {
      "epoch": 0.03811,
      "grad_norm": 0.46806634569742234,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3811
    },
    {
      "epoch": 0.03812,
      "grad_norm": 0.4708164911957734,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 3812
    },
    {
      "epoch": 0.03813,
      "grad_norm": 0.4451707961484543,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 3813
    },
    {
      "epoch": 0.03814,
      "grad_norm": 0.440545036907673,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 3814
    },
    {
      "epoch": 0.03815,
      "grad_norm": 0.4171348679565834,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 3815
    },
    {
      "epoch": 0.03816,
      "grad_norm": 0.553176653360458,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 3816
    },
    {
      "epoch": 0.03817,
      "grad_norm": 0.8249856806613923,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 3817
    },
    {
      "epoch": 0.03818,
      "grad_norm": 1.0849772720701052,
      "learning_rate": 0.003,
      "loss": 4.2073,
      "step": 3818
    },
    {
      "epoch": 0.03819,
      "grad_norm": 0.8469500316270079,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 3819
    },
    {
      "epoch": 0.0382,
      "grad_norm": 0.707642322044435,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 3820
    },
    {
      "epoch": 0.03821,
      "grad_norm": 0.747958147518101,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 3821
    },
    {
      "epoch": 0.03822,
      "grad_norm": 0.6648920321229015,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 3822
    },
    {
      "epoch": 0.03823,
      "grad_norm": 0.6528306662073633,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 3823
    },
    {
      "epoch": 0.03824,
      "grad_norm": 0.760735694975771,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 3824
    },
    {
      "epoch": 0.03825,
      "grad_norm": 0.8622307424344766,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 3825
    },
    {
      "epoch": 0.03826,
      "grad_norm": 0.8866334244520664,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 3826
    },
    {
      "epoch": 0.03827,
      "grad_norm": 0.8924274619717957,
      "learning_rate": 0.003,
      "loss": 4.206,
      "step": 3827
    },
    {
      "epoch": 0.03828,
      "grad_norm": 0.7065036502476953,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 3828
    },
    {
      "epoch": 0.03829,
      "grad_norm": 0.6969933866633966,
      "learning_rate": 0.003,
      "loss": 4.1837,
      "step": 3829
    },
    {
      "epoch": 0.0383,
      "grad_norm": 0.6465184385629209,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 3830
    },
    {
      "epoch": 0.03831,
      "grad_norm": 0.7285929638974894,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 3831
    },
    {
      "epoch": 0.03832,
      "grad_norm": 0.7578366971294306,
      "learning_rate": 0.003,
      "loss": 4.1714,
      "step": 3832
    },
    {
      "epoch": 0.03833,
      "grad_norm": 0.7591256877673351,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3833
    },
    {
      "epoch": 0.03834,
      "grad_norm": 0.6570705623790875,
      "learning_rate": 0.003,
      "loss": 4.2013,
      "step": 3834
    },
    {
      "epoch": 0.03835,
      "grad_norm": 0.6338048148231775,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 3835
    },
    {
      "epoch": 0.03836,
      "grad_norm": 0.579305584931518,
      "learning_rate": 0.003,
      "loss": 4.1872,
      "step": 3836
    },
    {
      "epoch": 0.03837,
      "grad_norm": 0.5000040915409231,
      "learning_rate": 0.003,
      "loss": 4.1685,
      "step": 3837
    },
    {
      "epoch": 0.03838,
      "grad_norm": 0.48980765754438066,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 3838
    },
    {
      "epoch": 0.03839,
      "grad_norm": 0.44391262328684045,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 3839
    },
    {
      "epoch": 0.0384,
      "grad_norm": 0.4581943030154349,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 3840
    },
    {
      "epoch": 0.03841,
      "grad_norm": 0.4824433708702651,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 3841
    },
    {
      "epoch": 0.03842,
      "grad_norm": 0.4830219498091642,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3842
    },
    {
      "epoch": 0.03843,
      "grad_norm": 0.45884626845612136,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 3843
    },
    {
      "epoch": 0.03844,
      "grad_norm": 0.3942504840023139,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 3844
    },
    {
      "epoch": 0.03845,
      "grad_norm": 0.35673387851966176,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 3845
    },
    {
      "epoch": 0.03846,
      "grad_norm": 0.3480023206810377,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3846
    },
    {
      "epoch": 0.03847,
      "grad_norm": 0.30772952512029117,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 3847
    },
    {
      "epoch": 0.03848,
      "grad_norm": 0.351053429769811,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 3848
    },
    {
      "epoch": 0.03849,
      "grad_norm": 0.3486839774780466,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3849
    },
    {
      "epoch": 0.0385,
      "grad_norm": 0.41203805300045565,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 3850
    },
    {
      "epoch": 0.03851,
      "grad_norm": 0.5052943186225257,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 3851
    },
    {
      "epoch": 0.03852,
      "grad_norm": 0.6267509920431387,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3852
    },
    {
      "epoch": 0.03853,
      "grad_norm": 0.7787688899298432,
      "learning_rate": 0.003,
      "loss": 4.1691,
      "step": 3853
    },
    {
      "epoch": 0.03854,
      "grad_norm": 0.7725109769746114,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 3854
    },
    {
      "epoch": 0.03855,
      "grad_norm": 0.6052355534312069,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3855
    },
    {
      "epoch": 0.03856,
      "grad_norm": 0.6299503142898346,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 3856
    },
    {
      "epoch": 0.03857,
      "grad_norm": 0.5706246944559047,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3857
    },
    {
      "epoch": 0.03858,
      "grad_norm": 0.5178970385116678,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 3858
    },
    {
      "epoch": 0.03859,
      "grad_norm": 0.6163137266901324,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 3859
    },
    {
      "epoch": 0.0386,
      "grad_norm": 0.5584392387461609,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 3860
    },
    {
      "epoch": 0.03861,
      "grad_norm": 0.5881922942664491,
      "learning_rate": 0.003,
      "loss": 4.174,
      "step": 3861
    },
    {
      "epoch": 0.03862,
      "grad_norm": 0.6070887580362981,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 3862
    },
    {
      "epoch": 0.03863,
      "grad_norm": 0.5412974380781377,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 3863
    },
    {
      "epoch": 0.03864,
      "grad_norm": 0.558239316328612,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 3864
    },
    {
      "epoch": 0.03865,
      "grad_norm": 0.5920559813538258,
      "learning_rate": 0.003,
      "loss": 4.1625,
      "step": 3865
    },
    {
      "epoch": 0.03866,
      "grad_norm": 0.65782610303018,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 3866
    },
    {
      "epoch": 0.03867,
      "grad_norm": 0.8001945832132011,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 3867
    },
    {
      "epoch": 0.03868,
      "grad_norm": 0.8093398470542099,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 3868
    },
    {
      "epoch": 0.03869,
      "grad_norm": 0.8272905784155873,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 3869
    },
    {
      "epoch": 0.0387,
      "grad_norm": 0.7805492019719857,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 3870
    },
    {
      "epoch": 0.03871,
      "grad_norm": 0.6463277445729251,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 3871
    },
    {
      "epoch": 0.03872,
      "grad_norm": 0.7130625269699555,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 3872
    },
    {
      "epoch": 0.03873,
      "grad_norm": 0.7604226657571129,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 3873
    },
    {
      "epoch": 0.03874,
      "grad_norm": 0.8264003715319567,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3874
    },
    {
      "epoch": 0.03875,
      "grad_norm": 0.8157133390523014,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3875
    },
    {
      "epoch": 0.03876,
      "grad_norm": 0.6633759602344576,
      "learning_rate": 0.003,
      "loss": 4.1981,
      "step": 3876
    },
    {
      "epoch": 0.03877,
      "grad_norm": 0.5945213411893223,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3877
    },
    {
      "epoch": 0.03878,
      "grad_norm": 0.6057373256262768,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 3878
    },
    {
      "epoch": 0.03879,
      "grad_norm": 0.632481493308064,
      "learning_rate": 0.003,
      "loss": 4.1887,
      "step": 3879
    },
    {
      "epoch": 0.0388,
      "grad_norm": 0.5936273104909171,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3880
    },
    {
      "epoch": 0.03881,
      "grad_norm": 0.5078096495379679,
      "learning_rate": 0.003,
      "loss": 4.1746,
      "step": 3881
    },
    {
      "epoch": 0.03882,
      "grad_norm": 0.47981556545878357,
      "learning_rate": 0.003,
      "loss": 4.1795,
      "step": 3882
    },
    {
      "epoch": 0.03883,
      "grad_norm": 0.5081271003225738,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 3883
    },
    {
      "epoch": 0.03884,
      "grad_norm": 0.5324630916346529,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 3884
    },
    {
      "epoch": 0.03885,
      "grad_norm": 0.5770047266124201,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 3885
    },
    {
      "epoch": 0.03886,
      "grad_norm": 0.5823626216641113,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 3886
    },
    {
      "epoch": 0.03887,
      "grad_norm": 0.5323044500683095,
      "learning_rate": 0.003,
      "loss": 4.1628,
      "step": 3887
    },
    {
      "epoch": 0.03888,
      "grad_norm": 0.6470253491308557,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 3888
    },
    {
      "epoch": 0.03889,
      "grad_norm": 0.7016901913145263,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 3889
    },
    {
      "epoch": 0.0389,
      "grad_norm": 0.6102524549297166,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 3890
    },
    {
      "epoch": 0.03891,
      "grad_norm": 0.5529440776190121,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 3891
    },
    {
      "epoch": 0.03892,
      "grad_norm": 0.5916779639985591,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3892
    },
    {
      "epoch": 0.03893,
      "grad_norm": 0.6828403164964358,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 3893
    },
    {
      "epoch": 0.03894,
      "grad_norm": 0.6193514242120052,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 3894
    },
    {
      "epoch": 0.03895,
      "grad_norm": 0.5599659326192736,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 3895
    },
    {
      "epoch": 0.03896,
      "grad_norm": 0.48762045043395263,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 3896
    },
    {
      "epoch": 0.03897,
      "grad_norm": 0.5042861977924956,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 3897
    },
    {
      "epoch": 0.03898,
      "grad_norm": 0.487281141563187,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 3898
    },
    {
      "epoch": 0.03899,
      "grad_norm": 0.5187455846326631,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 3899
    },
    {
      "epoch": 0.039,
      "grad_norm": 0.4540397798848912,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 3900
    },
    {
      "epoch": 0.03901,
      "grad_norm": 0.4735936186553674,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 3901
    },
    {
      "epoch": 0.03902,
      "grad_norm": 0.45921775589104796,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 3902
    },
    {
      "epoch": 0.03903,
      "grad_norm": 0.37493975839614513,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3903
    },
    {
      "epoch": 0.03904,
      "grad_norm": 0.3462652687764147,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 3904
    },
    {
      "epoch": 0.03905,
      "grad_norm": 0.35156300254567974,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 3905
    },
    {
      "epoch": 0.03906,
      "grad_norm": 0.3615192109464765,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 3906
    },
    {
      "epoch": 0.03907,
      "grad_norm": 0.4018823550496774,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 3907
    },
    {
      "epoch": 0.03908,
      "grad_norm": 0.5764826701055724,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 3908
    },
    {
      "epoch": 0.03909,
      "grad_norm": 0.7886220228031054,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 3909
    },
    {
      "epoch": 0.0391,
      "grad_norm": 0.8778590965436065,
      "learning_rate": 0.003,
      "loss": 4.1885,
      "step": 3910
    },
    {
      "epoch": 0.03911,
      "grad_norm": 0.7707501748681591,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 3911
    },
    {
      "epoch": 0.03912,
      "grad_norm": 0.7237965058473473,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3912
    },
    {
      "epoch": 0.03913,
      "grad_norm": 0.7684494783386372,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 3913
    },
    {
      "epoch": 0.03914,
      "grad_norm": 0.6559500648333249,
      "learning_rate": 0.003,
      "loss": 4.1776,
      "step": 3914
    },
    {
      "epoch": 0.03915,
      "grad_norm": 0.6438695734989948,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3915
    },
    {
      "epoch": 0.03916,
      "grad_norm": 0.710001018742664,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3916
    },
    {
      "epoch": 0.03917,
      "grad_norm": 0.7219014702450186,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 3917
    },
    {
      "epoch": 0.03918,
      "grad_norm": 0.6133683335088463,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 3918
    },
    {
      "epoch": 0.03919,
      "grad_norm": 0.4840575228811507,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 3919
    },
    {
      "epoch": 0.0392,
      "grad_norm": 0.4695659280359604,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 3920
    },
    {
      "epoch": 0.03921,
      "grad_norm": 0.4597028009194431,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 3921
    },
    {
      "epoch": 0.03922,
      "grad_norm": 0.4021059704956593,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 3922
    },
    {
      "epoch": 0.03923,
      "grad_norm": 0.45811748316459755,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 3923
    },
    {
      "epoch": 0.03924,
      "grad_norm": 0.4786080372228159,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 3924
    },
    {
      "epoch": 0.03925,
      "grad_norm": 0.5049165752143754,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 3925
    },
    {
      "epoch": 0.03926,
      "grad_norm": 0.5163104662428779,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 3926
    },
    {
      "epoch": 0.03927,
      "grad_norm": 0.5099171230784774,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 3927
    },
    {
      "epoch": 0.03928,
      "grad_norm": 0.572288419562524,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3928
    },
    {
      "epoch": 0.03929,
      "grad_norm": 0.6901560668624227,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 3929
    },
    {
      "epoch": 0.0393,
      "grad_norm": 0.6037962837223242,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 3930
    },
    {
      "epoch": 0.03931,
      "grad_norm": 0.5251957250436882,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3931
    },
    {
      "epoch": 0.03932,
      "grad_norm": 0.538831165664605,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3932
    },
    {
      "epoch": 0.03933,
      "grad_norm": 0.5209838890496463,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 3933
    },
    {
      "epoch": 0.03934,
      "grad_norm": 0.566519679434704,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 3934
    },
    {
      "epoch": 0.03935,
      "grad_norm": 0.646396237695394,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 3935
    },
    {
      "epoch": 0.03936,
      "grad_norm": 0.7750930535443175,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 3936
    },
    {
      "epoch": 0.03937,
      "grad_norm": 0.9869397924658896,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3937
    },
    {
      "epoch": 0.03938,
      "grad_norm": 1.0277905519905985,
      "learning_rate": 0.003,
      "loss": 4.1836,
      "step": 3938
    },
    {
      "epoch": 0.03939,
      "grad_norm": 0.7852266987902085,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 3939
    },
    {
      "epoch": 0.0394,
      "grad_norm": 0.8232677591358063,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 3940
    },
    {
      "epoch": 0.03941,
      "grad_norm": 0.7554935915430424,
      "learning_rate": 0.003,
      "loss": 4.1765,
      "step": 3941
    },
    {
      "epoch": 0.03942,
      "grad_norm": 0.5714665309206746,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 3942
    },
    {
      "epoch": 0.03943,
      "grad_norm": 0.5845658325896066,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 3943
    },
    {
      "epoch": 0.03944,
      "grad_norm": 0.6026107955836698,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 3944
    },
    {
      "epoch": 0.03945,
      "grad_norm": 0.5468006864617043,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 3945
    },
    {
      "epoch": 0.03946,
      "grad_norm": 0.5298095903489612,
      "learning_rate": 0.003,
      "loss": 4.1779,
      "step": 3946
    },
    {
      "epoch": 0.03947,
      "grad_norm": 0.5118480688773221,
      "learning_rate": 0.003,
      "loss": 4.1795,
      "step": 3947
    },
    {
      "epoch": 0.03948,
      "grad_norm": 0.5415983011837967,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 3948
    },
    {
      "epoch": 0.03949,
      "grad_norm": 0.6476511273829448,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 3949
    },
    {
      "epoch": 0.0395,
      "grad_norm": 0.5921282759053738,
      "learning_rate": 0.003,
      "loss": 4.1694,
      "step": 3950
    },
    {
      "epoch": 0.03951,
      "grad_norm": 0.4873072132844621,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 3951
    },
    {
      "epoch": 0.03952,
      "grad_norm": 0.47680587697404614,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 3952
    },
    {
      "epoch": 0.03953,
      "grad_norm": 0.4863348841149029,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 3953
    },
    {
      "epoch": 0.03954,
      "grad_norm": 0.4976353402048533,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3954
    },
    {
      "epoch": 0.03955,
      "grad_norm": 0.5951063481718434,
      "learning_rate": 0.003,
      "loss": 4.1575,
      "step": 3955
    },
    {
      "epoch": 0.03956,
      "grad_norm": 0.6359841084362824,
      "learning_rate": 0.003,
      "loss": 4.1616,
      "step": 3956
    },
    {
      "epoch": 0.03957,
      "grad_norm": 0.5941354341152176,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3957
    },
    {
      "epoch": 0.03958,
      "grad_norm": 0.6262454009808719,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 3958
    },
    {
      "epoch": 0.03959,
      "grad_norm": 0.7154633677223874,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 3959
    },
    {
      "epoch": 0.0396,
      "grad_norm": 0.7081920422281349,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 3960
    },
    {
      "epoch": 0.03961,
      "grad_norm": 0.7397573168118693,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 3961
    },
    {
      "epoch": 0.03962,
      "grad_norm": 0.711337260874705,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 3962
    },
    {
      "epoch": 0.03963,
      "grad_norm": 0.7123884385146033,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 3963
    },
    {
      "epoch": 0.03964,
      "grad_norm": 0.6152211153875224,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 3964
    },
    {
      "epoch": 0.03965,
      "grad_norm": 0.591222143630805,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 3965
    },
    {
      "epoch": 0.03966,
      "grad_norm": 0.5392802901033877,
      "learning_rate": 0.003,
      "loss": 4.1935,
      "step": 3966
    },
    {
      "epoch": 0.03967,
      "grad_norm": 0.524610568064759,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 3967
    },
    {
      "epoch": 0.03968,
      "grad_norm": 0.5121602181555871,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 3968
    },
    {
      "epoch": 0.03969,
      "grad_norm": 0.5008822737603466,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 3969
    },
    {
      "epoch": 0.0397,
      "grad_norm": 0.5150356493385037,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 3970
    },
    {
      "epoch": 0.03971,
      "grad_norm": 0.5181126377410241,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 3971
    },
    {
      "epoch": 0.03972,
      "grad_norm": 0.4952868090920291,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 3972
    },
    {
      "epoch": 0.03973,
      "grad_norm": 0.49146340749665485,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 3973
    },
    {
      "epoch": 0.03974,
      "grad_norm": 0.6037509933252784,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 3974
    },
    {
      "epoch": 0.03975,
      "grad_norm": 0.7194350601933044,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 3975
    },
    {
      "epoch": 0.03976,
      "grad_norm": 0.7807821793201645,
      "learning_rate": 0.003,
      "loss": 4.189,
      "step": 3976
    },
    {
      "epoch": 0.03977,
      "grad_norm": 0.7824814786642659,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3977
    },
    {
      "epoch": 0.03978,
      "grad_norm": 0.7382196107400837,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 3978
    },
    {
      "epoch": 0.03979,
      "grad_norm": 0.7187722022394311,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 3979
    },
    {
      "epoch": 0.0398,
      "grad_norm": 0.7054934205273465,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 3980
    },
    {
      "epoch": 0.03981,
      "grad_norm": 0.7032054871032929,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 3981
    },
    {
      "epoch": 0.03982,
      "grad_norm": 0.5823608521179618,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 3982
    },
    {
      "epoch": 0.03983,
      "grad_norm": 0.5786298940783224,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3983
    },
    {
      "epoch": 0.03984,
      "grad_norm": 0.6246079403898089,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 3984
    },
    {
      "epoch": 0.03985,
      "grad_norm": 0.6818628523024235,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 3985
    },
    {
      "epoch": 0.03986,
      "grad_norm": 0.6535731478840459,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 3986
    },
    {
      "epoch": 0.03987,
      "grad_norm": 0.6040151659435402,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 3987
    },
    {
      "epoch": 0.03988,
      "grad_norm": 0.6453741519043461,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 3988
    },
    {
      "epoch": 0.03989,
      "grad_norm": 0.764312719470282,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 3989
    },
    {
      "epoch": 0.0399,
      "grad_norm": 0.8495293415117621,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 3990
    },
    {
      "epoch": 0.03991,
      "grad_norm": 0.7402933337405657,
      "learning_rate": 0.003,
      "loss": 4.183,
      "step": 3991
    },
    {
      "epoch": 0.03992,
      "grad_norm": 0.5804226786236383,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 3992
    },
    {
      "epoch": 0.03993,
      "grad_norm": 0.586169093303552,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 3993
    },
    {
      "epoch": 0.03994,
      "grad_norm": 0.6041046229351503,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 3994
    },
    {
      "epoch": 0.03995,
      "grad_norm": 0.6875437069134418,
      "learning_rate": 0.003,
      "loss": 4.1441,
      "step": 3995
    },
    {
      "epoch": 0.03996,
      "grad_norm": 0.5847418717553203,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3996
    },
    {
      "epoch": 0.03997,
      "grad_norm": 0.5323079012210317,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3997
    },
    {
      "epoch": 0.03998,
      "grad_norm": 0.4356783084468044,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 3998
    },
    {
      "epoch": 0.03999,
      "grad_norm": 0.3763086485452268,
      "learning_rate": 0.003,
      "loss": 4.1692,
      "step": 3999
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.37795109500013657,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 4000
    },
    {
      "epoch": 0.04001,
      "grad_norm": 0.3830959820048188,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 4001
    },
    {
      "epoch": 0.04002,
      "grad_norm": 0.41146487178002505,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 4002
    },
    {
      "epoch": 0.04003,
      "grad_norm": 0.4313910001981467,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 4003
    },
    {
      "epoch": 0.04004,
      "grad_norm": 0.4459746247604169,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 4004
    },
    {
      "epoch": 0.04005,
      "grad_norm": 0.47718734644902294,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 4005
    },
    {
      "epoch": 0.04006,
      "grad_norm": 0.4814752963605302,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 4006
    },
    {
      "epoch": 0.04007,
      "grad_norm": 0.5099778221457298,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 4007
    },
    {
      "epoch": 0.04008,
      "grad_norm": 0.5043675363902501,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 4008
    },
    {
      "epoch": 0.04009,
      "grad_norm": 0.5439841205133068,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 4009
    },
    {
      "epoch": 0.0401,
      "grad_norm": 0.6497881885617821,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 4010
    },
    {
      "epoch": 0.04011,
      "grad_norm": 0.8245370470535895,
      "learning_rate": 0.003,
      "loss": 4.1694,
      "step": 4011
    },
    {
      "epoch": 0.04012,
      "grad_norm": 0.9674009929255976,
      "learning_rate": 0.003,
      "loss": 4.1789,
      "step": 4012
    },
    {
      "epoch": 0.04013,
      "grad_norm": 0.7353158349829942,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 4013
    },
    {
      "epoch": 0.04014,
      "grad_norm": 0.7081307887263109,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 4014
    },
    {
      "epoch": 0.04015,
      "grad_norm": 0.7515395751430206,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 4015
    },
    {
      "epoch": 0.04016,
      "grad_norm": 0.7357741460180915,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 4016
    },
    {
      "epoch": 0.04017,
      "grad_norm": 0.6881306027184227,
      "learning_rate": 0.003,
      "loss": 4.2004,
      "step": 4017
    },
    {
      "epoch": 0.04018,
      "grad_norm": 0.6328656616117202,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 4018
    },
    {
      "epoch": 0.04019,
      "grad_norm": 0.672111331766413,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4019
    },
    {
      "epoch": 0.0402,
      "grad_norm": 0.6935269300068305,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 4020
    },
    {
      "epoch": 0.04021,
      "grad_norm": 0.6368533158642944,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 4021
    },
    {
      "epoch": 0.04022,
      "grad_norm": 0.7051716536344202,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 4022
    },
    {
      "epoch": 0.04023,
      "grad_norm": 0.6469926590464656,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 4023
    },
    {
      "epoch": 0.04024,
      "grad_norm": 0.6694771111384129,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4024
    },
    {
      "epoch": 0.04025,
      "grad_norm": 0.6341741755537447,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 4025
    },
    {
      "epoch": 0.04026,
      "grad_norm": 0.6640885632809859,
      "learning_rate": 0.003,
      "loss": 4.1767,
      "step": 4026
    },
    {
      "epoch": 0.04027,
      "grad_norm": 0.6089194432075719,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 4027
    },
    {
      "epoch": 0.04028,
      "grad_norm": 0.6436111853565991,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 4028
    },
    {
      "epoch": 0.04029,
      "grad_norm": 0.601196906060423,
      "learning_rate": 0.003,
      "loss": 4.1851,
      "step": 4029
    },
    {
      "epoch": 0.0403,
      "grad_norm": 0.6008464331523574,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 4030
    },
    {
      "epoch": 0.04031,
      "grad_norm": 0.5735266100756939,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 4031
    },
    {
      "epoch": 0.04032,
      "grad_norm": 0.5248413725968446,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 4032
    },
    {
      "epoch": 0.04033,
      "grad_norm": 0.49046370370450504,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 4033
    },
    {
      "epoch": 0.04034,
      "grad_norm": 0.5090112067595912,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4034
    },
    {
      "epoch": 0.04035,
      "grad_norm": 0.4649874752654264,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 4035
    },
    {
      "epoch": 0.04036,
      "grad_norm": 0.4897958199323308,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 4036
    },
    {
      "epoch": 0.04037,
      "grad_norm": 0.6044670370792896,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 4037
    },
    {
      "epoch": 0.04038,
      "grad_norm": 0.8335470507296103,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4038
    },
    {
      "epoch": 0.04039,
      "grad_norm": 0.8645850468976461,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 4039
    },
    {
      "epoch": 0.0404,
      "grad_norm": 0.7530842619650773,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 4040
    },
    {
      "epoch": 0.04041,
      "grad_norm": 0.5842132443480766,
      "learning_rate": 0.003,
      "loss": 4.1457,
      "step": 4041
    },
    {
      "epoch": 0.04042,
      "grad_norm": 0.5354331761654865,
      "learning_rate": 0.003,
      "loss": 4.1835,
      "step": 4042
    },
    {
      "epoch": 0.04043,
      "grad_norm": 0.5185871828066747,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 4043
    },
    {
      "epoch": 0.04044,
      "grad_norm": 0.5097132412370223,
      "learning_rate": 0.003,
      "loss": 4.162,
      "step": 4044
    },
    {
      "epoch": 0.04045,
      "grad_norm": 0.553775594849916,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 4045
    },
    {
      "epoch": 0.04046,
      "grad_norm": 0.49141462695183785,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4046
    },
    {
      "epoch": 0.04047,
      "grad_norm": 0.41512219304607756,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 4047
    },
    {
      "epoch": 0.04048,
      "grad_norm": 0.4225143586443279,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 4048
    },
    {
      "epoch": 0.04049,
      "grad_norm": 0.3984988970584974,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 4049
    },
    {
      "epoch": 0.0405,
      "grad_norm": 0.3749419268913753,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 4050
    },
    {
      "epoch": 0.04051,
      "grad_norm": 0.39476946001794994,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4051
    },
    {
      "epoch": 0.04052,
      "grad_norm": 0.4554884596116926,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 4052
    },
    {
      "epoch": 0.04053,
      "grad_norm": 0.6242383494718413,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 4053
    },
    {
      "epoch": 0.04054,
      "grad_norm": 0.7860677992363685,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 4054
    },
    {
      "epoch": 0.04055,
      "grad_norm": 0.8126593846409768,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 4055
    },
    {
      "epoch": 0.04056,
      "grad_norm": 0.7391710444931487,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 4056
    },
    {
      "epoch": 0.04057,
      "grad_norm": 0.7425519640955306,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4057
    },
    {
      "epoch": 0.04058,
      "grad_norm": 0.7750101755361524,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 4058
    },
    {
      "epoch": 0.04059,
      "grad_norm": 0.6970458468533857,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 4059
    },
    {
      "epoch": 0.0406,
      "grad_norm": 0.6081117396919562,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 4060
    },
    {
      "epoch": 0.04061,
      "grad_norm": 0.539360737374925,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 4061
    },
    {
      "epoch": 0.04062,
      "grad_norm": 0.5166363975835321,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 4062
    },
    {
      "epoch": 0.04063,
      "grad_norm": 0.49611935522557493,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 4063
    },
    {
      "epoch": 0.04064,
      "grad_norm": 0.41285257049195684,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 4064
    },
    {
      "epoch": 0.04065,
      "grad_norm": 0.4763881231604338,
      "learning_rate": 0.003,
      "loss": 4.193,
      "step": 4065
    },
    {
      "epoch": 0.04066,
      "grad_norm": 0.4741067038687965,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4066
    },
    {
      "epoch": 0.04067,
      "grad_norm": 0.4426062959642298,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4067
    },
    {
      "epoch": 0.04068,
      "grad_norm": 0.42915467353070186,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 4068
    },
    {
      "epoch": 0.04069,
      "grad_norm": 0.45412191308027944,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4069
    },
    {
      "epoch": 0.0407,
      "grad_norm": 0.4671086228860357,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 4070
    },
    {
      "epoch": 0.04071,
      "grad_norm": 0.5692230139112596,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 4071
    },
    {
      "epoch": 0.04072,
      "grad_norm": 0.7764468905836547,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 4072
    },
    {
      "epoch": 0.04073,
      "grad_norm": 0.8448322221505851,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 4073
    },
    {
      "epoch": 0.04074,
      "grad_norm": 0.856020875384491,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 4074
    },
    {
      "epoch": 0.04075,
      "grad_norm": 0.9287167181671592,
      "learning_rate": 0.003,
      "loss": 4.1779,
      "step": 4075
    },
    {
      "epoch": 0.04076,
      "grad_norm": 0.873329112405605,
      "learning_rate": 0.003,
      "loss": 4.1789,
      "step": 4076
    },
    {
      "epoch": 0.04077,
      "grad_norm": 0.7601580397215363,
      "learning_rate": 0.003,
      "loss": 4.1881,
      "step": 4077
    },
    {
      "epoch": 0.04078,
      "grad_norm": 0.8042586607784782,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 4078
    },
    {
      "epoch": 0.04079,
      "grad_norm": 0.8176950474454038,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 4079
    },
    {
      "epoch": 0.0408,
      "grad_norm": 0.9030758743289882,
      "learning_rate": 0.003,
      "loss": 4.1811,
      "step": 4080
    },
    {
      "epoch": 0.04081,
      "grad_norm": 0.7986927315562776,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 4081
    },
    {
      "epoch": 0.04082,
      "grad_norm": 0.7456348575122663,
      "learning_rate": 0.003,
      "loss": 4.1933,
      "step": 4082
    },
    {
      "epoch": 0.04083,
      "grad_norm": 0.7464945818151806,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 4083
    },
    {
      "epoch": 0.04084,
      "grad_norm": 0.5910864081536713,
      "learning_rate": 0.003,
      "loss": 4.2007,
      "step": 4084
    },
    {
      "epoch": 0.04085,
      "grad_norm": 0.6374107110244114,
      "learning_rate": 0.003,
      "loss": 4.1813,
      "step": 4085
    },
    {
      "epoch": 0.04086,
      "grad_norm": 0.644544973240151,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 4086
    },
    {
      "epoch": 0.04087,
      "grad_norm": 0.6474810812977669,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 4087
    },
    {
      "epoch": 0.04088,
      "grad_norm": 0.48757799745230485,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 4088
    },
    {
      "epoch": 0.04089,
      "grad_norm": 0.5082604112884798,
      "learning_rate": 0.003,
      "loss": 4.1548,
      "step": 4089
    },
    {
      "epoch": 0.0409,
      "grad_norm": 0.4251634976124169,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 4090
    },
    {
      "epoch": 0.04091,
      "grad_norm": 0.40607507200162474,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 4091
    },
    {
      "epoch": 0.04092,
      "grad_norm": 0.4118371741461281,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 4092
    },
    {
      "epoch": 0.04093,
      "grad_norm": 0.379924610879027,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 4093
    },
    {
      "epoch": 0.04094,
      "grad_norm": 0.43768923530899545,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4094
    },
    {
      "epoch": 0.04095,
      "grad_norm": 0.411023776294053,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 4095
    },
    {
      "epoch": 0.04096,
      "grad_norm": 0.45768098717184347,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 4096
    },
    {
      "epoch": 0.04097,
      "grad_norm": 0.557314037287976,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4097
    },
    {
      "epoch": 0.04098,
      "grad_norm": 0.6699585119473419,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 4098
    },
    {
      "epoch": 0.04099,
      "grad_norm": 0.6624628981660053,
      "learning_rate": 0.003,
      "loss": 4.1765,
      "step": 4099
    },
    {
      "epoch": 0.041,
      "grad_norm": 0.4938757392791384,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 4100
    },
    {
      "epoch": 0.04101,
      "grad_norm": 0.41979137569186115,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 4101
    },
    {
      "epoch": 0.04102,
      "grad_norm": 0.49582634443681156,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 4102
    },
    {
      "epoch": 0.04103,
      "grad_norm": 0.46586436339945675,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 4103
    },
    {
      "epoch": 0.04104,
      "grad_norm": 0.45341411443238716,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 4104
    },
    {
      "epoch": 0.04105,
      "grad_norm": 0.39192920748405663,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 4105
    },
    {
      "epoch": 0.04106,
      "grad_norm": 0.40403661174930955,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 4106
    },
    {
      "epoch": 0.04107,
      "grad_norm": 0.46259866064643457,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4107
    },
    {
      "epoch": 0.04108,
      "grad_norm": 0.5224958035249453,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 4108
    },
    {
      "epoch": 0.04109,
      "grad_norm": 0.5925525541895633,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 4109
    },
    {
      "epoch": 0.0411,
      "grad_norm": 0.6870748923239856,
      "learning_rate": 0.003,
      "loss": 4.142,
      "step": 4110
    },
    {
      "epoch": 0.04111,
      "grad_norm": 0.6433679126097386,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 4111
    },
    {
      "epoch": 0.04112,
      "grad_norm": 0.5391529554622597,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 4112
    },
    {
      "epoch": 0.04113,
      "grad_norm": 0.5453895793045661,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 4113
    },
    {
      "epoch": 0.04114,
      "grad_norm": 0.6398919299706648,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 4114
    },
    {
      "epoch": 0.04115,
      "grad_norm": 0.5856750517364259,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 4115
    },
    {
      "epoch": 0.04116,
      "grad_norm": 0.5640052575174052,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 4116
    },
    {
      "epoch": 0.04117,
      "grad_norm": 0.5578995001301751,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 4117
    },
    {
      "epoch": 0.04118,
      "grad_norm": 0.6189606161266092,
      "learning_rate": 0.003,
      "loss": 4.1802,
      "step": 4118
    },
    {
      "epoch": 0.04119,
      "grad_norm": 0.69284741853494,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 4119
    },
    {
      "epoch": 0.0412,
      "grad_norm": 0.8122773604588729,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 4120
    },
    {
      "epoch": 0.04121,
      "grad_norm": 0.7788291402367661,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 4121
    },
    {
      "epoch": 0.04122,
      "grad_norm": 0.7205856618274178,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 4122
    },
    {
      "epoch": 0.04123,
      "grad_norm": 0.6723108027243482,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 4123
    },
    {
      "epoch": 0.04124,
      "grad_norm": 0.5742707663821022,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 4124
    },
    {
      "epoch": 0.04125,
      "grad_norm": 0.5921662326911094,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 4125
    },
    {
      "epoch": 0.04126,
      "grad_norm": 0.5065182010593955,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 4126
    },
    {
      "epoch": 0.04127,
      "grad_norm": 0.49270157976785267,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 4127
    },
    {
      "epoch": 0.04128,
      "grad_norm": 0.47833247711412197,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 4128
    },
    {
      "epoch": 0.04129,
      "grad_norm": 0.6245977843098431,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 4129
    },
    {
      "epoch": 0.0413,
      "grad_norm": 0.7299723124473214,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4130
    },
    {
      "epoch": 0.04131,
      "grad_norm": 0.7562574300910655,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 4131
    },
    {
      "epoch": 0.04132,
      "grad_norm": 0.7690429680218056,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 4132
    },
    {
      "epoch": 0.04133,
      "grad_norm": 0.7790262395321094,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 4133
    },
    {
      "epoch": 0.04134,
      "grad_norm": 0.7036209580453355,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 4134
    },
    {
      "epoch": 0.04135,
      "grad_norm": 0.7431461288088913,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 4135
    },
    {
      "epoch": 0.04136,
      "grad_norm": 0.7470037276613238,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4136
    },
    {
      "epoch": 0.04137,
      "grad_norm": 0.8449453146727118,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 4137
    },
    {
      "epoch": 0.04138,
      "grad_norm": 0.6568218214769733,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 4138
    },
    {
      "epoch": 0.04139,
      "grad_norm": 0.6371205783656457,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 4139
    },
    {
      "epoch": 0.0414,
      "grad_norm": 0.724881052006821,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 4140
    },
    {
      "epoch": 0.04141,
      "grad_norm": 0.7146032900181981,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 4141
    },
    {
      "epoch": 0.04142,
      "grad_norm": 0.8156917337253932,
      "learning_rate": 0.003,
      "loss": 4.1811,
      "step": 4142
    },
    {
      "epoch": 0.04143,
      "grad_norm": 0.8785836902260167,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 4143
    },
    {
      "epoch": 0.04144,
      "grad_norm": 0.7999443682311421,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 4144
    },
    {
      "epoch": 0.04145,
      "grad_norm": 0.82858324833918,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 4145
    },
    {
      "epoch": 0.04146,
      "grad_norm": 0.8235890608340272,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 4146
    },
    {
      "epoch": 0.04147,
      "grad_norm": 0.858747508934936,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 4147
    },
    {
      "epoch": 0.04148,
      "grad_norm": 0.7838271345732937,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 4148
    },
    {
      "epoch": 0.04149,
      "grad_norm": 0.7263149003922105,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 4149
    },
    {
      "epoch": 0.0415,
      "grad_norm": 0.6942802417486142,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 4150
    },
    {
      "epoch": 0.04151,
      "grad_norm": 0.6248553867172758,
      "learning_rate": 0.003,
      "loss": 4.1813,
      "step": 4151
    },
    {
      "epoch": 0.04152,
      "grad_norm": 0.6112599353340841,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 4152
    },
    {
      "epoch": 0.04153,
      "grad_norm": 0.5629861808054807,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 4153
    },
    {
      "epoch": 0.04154,
      "grad_norm": 0.5649468221105395,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 4154
    },
    {
      "epoch": 0.04155,
      "grad_norm": 0.5286657686017564,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4155
    },
    {
      "epoch": 0.04156,
      "grad_norm": 0.4616483876621516,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 4156
    },
    {
      "epoch": 0.04157,
      "grad_norm": 0.43208170797618906,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 4157
    },
    {
      "epoch": 0.04158,
      "grad_norm": 0.47690558549648143,
      "learning_rate": 0.003,
      "loss": 4.1765,
      "step": 4158
    },
    {
      "epoch": 0.04159,
      "grad_norm": 0.548409828074487,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4159
    },
    {
      "epoch": 0.0416,
      "grad_norm": 0.5628758636950728,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 4160
    },
    {
      "epoch": 0.04161,
      "grad_norm": 0.6122953755048903,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 4161
    },
    {
      "epoch": 0.04162,
      "grad_norm": 0.5313833560293301,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 4162
    },
    {
      "epoch": 0.04163,
      "grad_norm": 0.4545671521128938,
      "learning_rate": 0.003,
      "loss": 4.1485,
      "step": 4163
    },
    {
      "epoch": 0.04164,
      "grad_norm": 0.48581450867041054,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 4164
    },
    {
      "epoch": 0.04165,
      "grad_norm": 0.48634350202313903,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 4165
    },
    {
      "epoch": 0.04166,
      "grad_norm": 0.42940905174204824,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 4166
    },
    {
      "epoch": 0.04167,
      "grad_norm": 0.39230202494683986,
      "learning_rate": 0.003,
      "loss": 4.1526,
      "step": 4167
    },
    {
      "epoch": 0.04168,
      "grad_norm": 0.39538400082145897,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 4168
    },
    {
      "epoch": 0.04169,
      "grad_norm": 0.3447120216665393,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 4169
    },
    {
      "epoch": 0.0417,
      "grad_norm": 0.36603795136119954,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 4170
    },
    {
      "epoch": 0.04171,
      "grad_norm": 0.38374101547618106,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 4171
    },
    {
      "epoch": 0.04172,
      "grad_norm": 0.5162775730950696,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 4172
    },
    {
      "epoch": 0.04173,
      "grad_norm": 0.5760736714087062,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 4173
    },
    {
      "epoch": 0.04174,
      "grad_norm": 0.5576881465067848,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 4174
    },
    {
      "epoch": 0.04175,
      "grad_norm": 0.6077258689017496,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 4175
    },
    {
      "epoch": 0.04176,
      "grad_norm": 0.6416498201931042,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 4176
    },
    {
      "epoch": 0.04177,
      "grad_norm": 0.5865172170220208,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 4177
    },
    {
      "epoch": 0.04178,
      "grad_norm": 0.6121973272164818,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 4178
    },
    {
      "epoch": 0.04179,
      "grad_norm": 0.7466591370840786,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 4179
    },
    {
      "epoch": 0.0418,
      "grad_norm": 0.7407658974314641,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 4180
    },
    {
      "epoch": 0.04181,
      "grad_norm": 0.7876489798566603,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 4181
    },
    {
      "epoch": 0.04182,
      "grad_norm": 0.6878533187156349,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 4182
    },
    {
      "epoch": 0.04183,
      "grad_norm": 0.6899295854896734,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 4183
    },
    {
      "epoch": 0.04184,
      "grad_norm": 0.6688165143357564,
      "learning_rate": 0.003,
      "loss": 4.1979,
      "step": 4184
    },
    {
      "epoch": 0.04185,
      "grad_norm": 0.6585905689769214,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 4185
    },
    {
      "epoch": 0.04186,
      "grad_norm": 0.706982773019547,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 4186
    },
    {
      "epoch": 0.04187,
      "grad_norm": 0.7286914544536822,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 4187
    },
    {
      "epoch": 0.04188,
      "grad_norm": 0.7190589294145873,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 4188
    },
    {
      "epoch": 0.04189,
      "grad_norm": 0.7351396129481365,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 4189
    },
    {
      "epoch": 0.0419,
      "grad_norm": 0.7646912861138266,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 4190
    },
    {
      "epoch": 0.04191,
      "grad_norm": 0.8912342474994416,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 4191
    },
    {
      "epoch": 0.04192,
      "grad_norm": 0.8349167750236294,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 4192
    },
    {
      "epoch": 0.04193,
      "grad_norm": 0.8412143448929614,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 4193
    },
    {
      "epoch": 0.04194,
      "grad_norm": 0.7468627479626141,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 4194
    },
    {
      "epoch": 0.04195,
      "grad_norm": 0.675962819896126,
      "learning_rate": 0.003,
      "loss": 4.1829,
      "step": 4195
    },
    {
      "epoch": 0.04196,
      "grad_norm": 0.6834824762686134,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 4196
    },
    {
      "epoch": 0.04197,
      "grad_norm": 0.6470038410606324,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4197
    },
    {
      "epoch": 0.04198,
      "grad_norm": 0.6260474494865785,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 4198
    },
    {
      "epoch": 0.04199,
      "grad_norm": 0.6233390255040572,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 4199
    },
    {
      "epoch": 0.042,
      "grad_norm": 0.5419194177090347,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 4200
    },
    {
      "epoch": 0.04201,
      "grad_norm": 0.4956778815254003,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 4201
    },
    {
      "epoch": 0.04202,
      "grad_norm": 0.4424272529732883,
      "learning_rate": 0.003,
      "loss": 4.1485,
      "step": 4202
    },
    {
      "epoch": 0.04203,
      "grad_norm": 0.4059745715455596,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 4203
    },
    {
      "epoch": 0.04204,
      "grad_norm": 0.3551788544607894,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 4204
    },
    {
      "epoch": 0.04205,
      "grad_norm": 0.36710978367184766,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 4205
    },
    {
      "epoch": 0.04206,
      "grad_norm": 0.37067515609326535,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 4206
    },
    {
      "epoch": 0.04207,
      "grad_norm": 0.3909030628374286,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4207
    },
    {
      "epoch": 0.04208,
      "grad_norm": 0.3512353865469784,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 4208
    },
    {
      "epoch": 0.04209,
      "grad_norm": 0.3222350876286574,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 4209
    },
    {
      "epoch": 0.0421,
      "grad_norm": 0.33699670679266025,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 4210
    },
    {
      "epoch": 0.04211,
      "grad_norm": 0.397353308681094,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4211
    },
    {
      "epoch": 0.04212,
      "grad_norm": 0.4950752479622572,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 4212
    },
    {
      "epoch": 0.04213,
      "grad_norm": 0.6908228603245805,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 4213
    },
    {
      "epoch": 0.04214,
      "grad_norm": 0.8171734853482235,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 4214
    },
    {
      "epoch": 0.04215,
      "grad_norm": 0.8297471255167204,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4215
    },
    {
      "epoch": 0.04216,
      "grad_norm": 0.8171229032271536,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4216
    },
    {
      "epoch": 0.04217,
      "grad_norm": 0.7119480146884566,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 4217
    },
    {
      "epoch": 0.04218,
      "grad_norm": 0.7835951039751599,
      "learning_rate": 0.003,
      "loss": 4.1866,
      "step": 4218
    },
    {
      "epoch": 0.04219,
      "grad_norm": 0.7238721651116449,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 4219
    },
    {
      "epoch": 0.0422,
      "grad_norm": 0.6704096903245251,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 4220
    },
    {
      "epoch": 0.04221,
      "grad_norm": 0.6137772128626461,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4221
    },
    {
      "epoch": 0.04222,
      "grad_norm": 0.6312616992519636,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 4222
    },
    {
      "epoch": 0.04223,
      "grad_norm": 0.7309830489217892,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 4223
    },
    {
      "epoch": 0.04224,
      "grad_norm": 0.8353004708644942,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 4224
    },
    {
      "epoch": 0.04225,
      "grad_norm": 0.8671668423571915,
      "learning_rate": 0.003,
      "loss": 4.1793,
      "step": 4225
    },
    {
      "epoch": 0.04226,
      "grad_norm": 0.7341929503846356,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 4226
    },
    {
      "epoch": 0.04227,
      "grad_norm": 0.6287599565012932,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 4227
    },
    {
      "epoch": 0.04228,
      "grad_norm": 0.594514738745631,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 4228
    },
    {
      "epoch": 0.04229,
      "grad_norm": 0.6229152206319786,
      "learning_rate": 0.003,
      "loss": 4.1691,
      "step": 4229
    },
    {
      "epoch": 0.0423,
      "grad_norm": 0.6207671248596826,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 4230
    },
    {
      "epoch": 0.04231,
      "grad_norm": 0.622706916945179,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 4231
    },
    {
      "epoch": 0.04232,
      "grad_norm": 0.6141015860612352,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 4232
    },
    {
      "epoch": 0.04233,
      "grad_norm": 0.5575040241097531,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 4233
    },
    {
      "epoch": 0.04234,
      "grad_norm": 0.5779738605421891,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4234
    },
    {
      "epoch": 0.04235,
      "grad_norm": 0.6628268406822522,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 4235
    },
    {
      "epoch": 0.04236,
      "grad_norm": 0.5848188526492533,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 4236
    },
    {
      "epoch": 0.04237,
      "grad_norm": 0.4606869163911272,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 4237
    },
    {
      "epoch": 0.04238,
      "grad_norm": 0.44274972044788236,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 4238
    },
    {
      "epoch": 0.04239,
      "grad_norm": 0.4300796410055906,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4239
    },
    {
      "epoch": 0.0424,
      "grad_norm": 0.3931497226975424,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 4240
    },
    {
      "epoch": 0.04241,
      "grad_norm": 0.44468609283123023,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4241
    },
    {
      "epoch": 0.04242,
      "grad_norm": 0.42999491902207304,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 4242
    },
    {
      "epoch": 0.04243,
      "grad_norm": 0.45444243019546493,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 4243
    },
    {
      "epoch": 0.04244,
      "grad_norm": 0.4177997103887438,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 4244
    },
    {
      "epoch": 0.04245,
      "grad_norm": 0.4042766192791119,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 4245
    },
    {
      "epoch": 0.04246,
      "grad_norm": 0.49465297913164696,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 4246
    },
    {
      "epoch": 0.04247,
      "grad_norm": 0.7023548593149965,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 4247
    },
    {
      "epoch": 0.04248,
      "grad_norm": 0.944713514521764,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 4248
    },
    {
      "epoch": 0.04249,
      "grad_norm": 0.9299347494145838,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 4249
    },
    {
      "epoch": 0.0425,
      "grad_norm": 0.7051640509051822,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 4250
    },
    {
      "epoch": 0.04251,
      "grad_norm": 0.6812273114666765,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 4251
    },
    {
      "epoch": 0.04252,
      "grad_norm": 0.6230213953232125,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 4252
    },
    {
      "epoch": 0.04253,
      "grad_norm": 0.6107603990750144,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 4253
    },
    {
      "epoch": 0.04254,
      "grad_norm": 0.5714927077917459,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 4254
    },
    {
      "epoch": 0.04255,
      "grad_norm": 0.5988656811798401,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 4255
    },
    {
      "epoch": 0.04256,
      "grad_norm": 0.6394087469893323,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4256
    },
    {
      "epoch": 0.04257,
      "grad_norm": 0.5967251959756738,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 4257
    },
    {
      "epoch": 0.04258,
      "grad_norm": 0.5945289215936239,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 4258
    },
    {
      "epoch": 0.04259,
      "grad_norm": 0.5867660553093561,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4259
    },
    {
      "epoch": 0.0426,
      "grad_norm": 0.48714297403717527,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 4260
    },
    {
      "epoch": 0.04261,
      "grad_norm": 0.4560387546134075,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 4261
    },
    {
      "epoch": 0.04262,
      "grad_norm": 0.43694421386753557,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4262
    },
    {
      "epoch": 0.04263,
      "grad_norm": 0.474489343873346,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 4263
    },
    {
      "epoch": 0.04264,
      "grad_norm": 0.48969639817641303,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 4264
    },
    {
      "epoch": 0.04265,
      "grad_norm": 0.48935865768007264,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 4265
    },
    {
      "epoch": 0.04266,
      "grad_norm": 0.4739244721175001,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 4266
    },
    {
      "epoch": 0.04267,
      "grad_norm": 0.48494742028757803,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 4267
    },
    {
      "epoch": 0.04268,
      "grad_norm": 0.5422961243923582,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 4268
    },
    {
      "epoch": 0.04269,
      "grad_norm": 0.5460647987973027,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 4269
    },
    {
      "epoch": 0.0427,
      "grad_norm": 0.608961907674264,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4270
    },
    {
      "epoch": 0.04271,
      "grad_norm": 0.6558443912098731,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 4271
    },
    {
      "epoch": 0.04272,
      "grad_norm": 0.663079753729842,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 4272
    },
    {
      "epoch": 0.04273,
      "grad_norm": 0.7592770165276204,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 4273
    },
    {
      "epoch": 0.04274,
      "grad_norm": 0.6676368129361678,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 4274
    },
    {
      "epoch": 0.04275,
      "grad_norm": 0.6623028472426901,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 4275
    },
    {
      "epoch": 0.04276,
      "grad_norm": 0.6624560252528698,
      "learning_rate": 0.003,
      "loss": 4.1719,
      "step": 4276
    },
    {
      "epoch": 0.04277,
      "grad_norm": 0.6210345683298933,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 4277
    },
    {
      "epoch": 0.04278,
      "grad_norm": 0.5991391029139438,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 4278
    },
    {
      "epoch": 0.04279,
      "grad_norm": 0.6350138668928702,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 4279
    },
    {
      "epoch": 0.0428,
      "grad_norm": 0.6849523452457494,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 4280
    },
    {
      "epoch": 0.04281,
      "grad_norm": 0.6921302141996111,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 4281
    },
    {
      "epoch": 0.04282,
      "grad_norm": 0.756590885346882,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 4282
    },
    {
      "epoch": 0.04283,
      "grad_norm": 0.7245456429489706,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 4283
    },
    {
      "epoch": 0.04284,
      "grad_norm": 0.6672309834707763,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4284
    },
    {
      "epoch": 0.04285,
      "grad_norm": 0.6246574113348472,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 4285
    },
    {
      "epoch": 0.04286,
      "grad_norm": 0.4846532441701148,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 4286
    },
    {
      "epoch": 0.04287,
      "grad_norm": 0.49611202994371745,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 4287
    },
    {
      "epoch": 0.04288,
      "grad_norm": 0.49123248349204507,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 4288
    },
    {
      "epoch": 0.04289,
      "grad_norm": 0.4112901668688879,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 4289
    },
    {
      "epoch": 0.0429,
      "grad_norm": 0.38028576043319295,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 4290
    },
    {
      "epoch": 0.04291,
      "grad_norm": 0.3996251945749772,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 4291
    },
    {
      "epoch": 0.04292,
      "grad_norm": 0.43349568162385765,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 4292
    },
    {
      "epoch": 0.04293,
      "grad_norm": 0.5220793875805044,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4293
    },
    {
      "epoch": 0.04294,
      "grad_norm": 0.6050495154218054,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 4294
    },
    {
      "epoch": 0.04295,
      "grad_norm": 0.7860329987891421,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 4295
    },
    {
      "epoch": 0.04296,
      "grad_norm": 0.9004405617366277,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 4296
    },
    {
      "epoch": 0.04297,
      "grad_norm": 0.8479021177195584,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 4297
    },
    {
      "epoch": 0.04298,
      "grad_norm": 0.7336646368221134,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 4298
    },
    {
      "epoch": 0.04299,
      "grad_norm": 0.7078429400943396,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 4299
    },
    {
      "epoch": 0.043,
      "grad_norm": 0.7513235769819381,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 4300
    },
    {
      "epoch": 0.04301,
      "grad_norm": 0.8326441152753217,
      "learning_rate": 0.003,
      "loss": 4.1932,
      "step": 4301
    },
    {
      "epoch": 0.04302,
      "grad_norm": 0.6815941455473161,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 4302
    },
    {
      "epoch": 0.04303,
      "grad_norm": 0.6890232000405212,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 4303
    },
    {
      "epoch": 0.04304,
      "grad_norm": 0.6319808958937929,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4304
    },
    {
      "epoch": 0.04305,
      "grad_norm": 0.5899508711069417,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4305
    },
    {
      "epoch": 0.04306,
      "grad_norm": 0.5383422790486718,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 4306
    },
    {
      "epoch": 0.04307,
      "grad_norm": 0.48802486139669815,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 4307
    },
    {
      "epoch": 0.04308,
      "grad_norm": 0.5309500644286907,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4308
    },
    {
      "epoch": 0.04309,
      "grad_norm": 0.5454729884280429,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 4309
    },
    {
      "epoch": 0.0431,
      "grad_norm": 0.5943838417760108,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 4310
    },
    {
      "epoch": 0.04311,
      "grad_norm": 0.6278301410987507,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 4311
    },
    {
      "epoch": 0.04312,
      "grad_norm": 0.523985785894097,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4312
    },
    {
      "epoch": 0.04313,
      "grad_norm": 0.4476351084578838,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 4313
    },
    {
      "epoch": 0.04314,
      "grad_norm": 0.43202288706395875,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 4314
    },
    {
      "epoch": 0.04315,
      "grad_norm": 0.47058217727703,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 4315
    },
    {
      "epoch": 0.04316,
      "grad_norm": 0.5151993523236311,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 4316
    },
    {
      "epoch": 0.04317,
      "grad_norm": 0.6504034739996317,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 4317
    },
    {
      "epoch": 0.04318,
      "grad_norm": 0.722856030914651,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 4318
    },
    {
      "epoch": 0.04319,
      "grad_norm": 0.7408662854196157,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 4319
    },
    {
      "epoch": 0.0432,
      "grad_norm": 0.6108907111379617,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 4320
    },
    {
      "epoch": 0.04321,
      "grad_norm": 0.5580728193515315,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 4321
    },
    {
      "epoch": 0.04322,
      "grad_norm": 0.6293577640178539,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 4322
    },
    {
      "epoch": 0.04323,
      "grad_norm": 0.680052451325024,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 4323
    },
    {
      "epoch": 0.04324,
      "grad_norm": 0.6420666473310551,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4324
    },
    {
      "epoch": 0.04325,
      "grad_norm": 0.7103856885257132,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 4325
    },
    {
      "epoch": 0.04326,
      "grad_norm": 0.6921778251891293,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 4326
    },
    {
      "epoch": 0.04327,
      "grad_norm": 0.688475290776878,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 4327
    },
    {
      "epoch": 0.04328,
      "grad_norm": 0.6182648113705878,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 4328
    },
    {
      "epoch": 0.04329,
      "grad_norm": 0.6893506649601877,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 4329
    },
    {
      "epoch": 0.0433,
      "grad_norm": 0.7382201356692273,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4330
    },
    {
      "epoch": 0.04331,
      "grad_norm": 0.6908297901661757,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 4331
    },
    {
      "epoch": 0.04332,
      "grad_norm": 0.6538224132935573,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 4332
    },
    {
      "epoch": 0.04333,
      "grad_norm": 0.5839042658168669,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 4333
    },
    {
      "epoch": 0.04334,
      "grad_norm": 0.5113278780509889,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 4334
    },
    {
      "epoch": 0.04335,
      "grad_norm": 0.5145875368624392,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 4335
    },
    {
      "epoch": 0.04336,
      "grad_norm": 0.45945029823422817,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 4336
    },
    {
      "epoch": 0.04337,
      "grad_norm": 0.48662366939348867,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 4337
    },
    {
      "epoch": 0.04338,
      "grad_norm": 0.5497905856065477,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 4338
    },
    {
      "epoch": 0.04339,
      "grad_norm": 0.6427639317081701,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 4339
    },
    {
      "epoch": 0.0434,
      "grad_norm": 0.751540691985128,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 4340
    },
    {
      "epoch": 0.04341,
      "grad_norm": 0.8048212003756245,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 4341
    },
    {
      "epoch": 0.04342,
      "grad_norm": 0.7030263986542218,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 4342
    },
    {
      "epoch": 0.04343,
      "grad_norm": 0.6032636490778056,
      "learning_rate": 0.003,
      "loss": 4.1739,
      "step": 4343
    },
    {
      "epoch": 0.04344,
      "grad_norm": 0.6918101884406253,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 4344
    },
    {
      "epoch": 0.04345,
      "grad_norm": 0.822156034408674,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 4345
    },
    {
      "epoch": 0.04346,
      "grad_norm": 0.8112944749590358,
      "learning_rate": 0.003,
      "loss": 4.1692,
      "step": 4346
    },
    {
      "epoch": 0.04347,
      "grad_norm": 0.7773363458697162,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 4347
    },
    {
      "epoch": 0.04348,
      "grad_norm": 0.7166070078793744,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4348
    },
    {
      "epoch": 0.04349,
      "grad_norm": 0.7557376957323917,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 4349
    },
    {
      "epoch": 0.0435,
      "grad_norm": 0.7306221155287103,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 4350
    },
    {
      "epoch": 0.04351,
      "grad_norm": 0.7396583467273142,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 4351
    },
    {
      "epoch": 0.04352,
      "grad_norm": 0.7800507975012694,
      "learning_rate": 0.003,
      "loss": 4.1888,
      "step": 4352
    },
    {
      "epoch": 0.04353,
      "grad_norm": 1.021228379310444,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 4353
    },
    {
      "epoch": 0.04354,
      "grad_norm": 0.9464406672758259,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 4354
    },
    {
      "epoch": 0.04355,
      "grad_norm": 0.9680937005908226,
      "learning_rate": 0.003,
      "loss": 4.19,
      "step": 4355
    },
    {
      "epoch": 0.04356,
      "grad_norm": 0.8607469609935269,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 4356
    },
    {
      "epoch": 0.04357,
      "grad_norm": 0.823519837090474,
      "learning_rate": 0.003,
      "loss": 4.1828,
      "step": 4357
    },
    {
      "epoch": 0.04358,
      "grad_norm": 0.9595628631285592,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 4358
    },
    {
      "epoch": 0.04359,
      "grad_norm": 0.9518657368376723,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 4359
    },
    {
      "epoch": 0.0436,
      "grad_norm": 0.8367488853787286,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 4360
    },
    {
      "epoch": 0.04361,
      "grad_norm": 0.6767616158798689,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 4361
    },
    {
      "epoch": 0.04362,
      "grad_norm": 0.6754203563658906,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 4362
    },
    {
      "epoch": 0.04363,
      "grad_norm": 0.6302727609189303,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 4363
    },
    {
      "epoch": 0.04364,
      "grad_norm": 0.6320795297761237,
      "learning_rate": 0.003,
      "loss": 4.1873,
      "step": 4364
    },
    {
      "epoch": 0.04365,
      "grad_norm": 0.6155242365026946,
      "learning_rate": 0.003,
      "loss": 4.1835,
      "step": 4365
    },
    {
      "epoch": 0.04366,
      "grad_norm": 0.5750840343277154,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 4366
    },
    {
      "epoch": 0.04367,
      "grad_norm": 0.5395062951089282,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 4367
    },
    {
      "epoch": 0.04368,
      "grad_norm": 0.5070165557962105,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 4368
    },
    {
      "epoch": 0.04369,
      "grad_norm": 0.469120529046112,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 4369
    },
    {
      "epoch": 0.0437,
      "grad_norm": 0.42165323317321096,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 4370
    },
    {
      "epoch": 0.04371,
      "grad_norm": 0.4102155524937744,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 4371
    },
    {
      "epoch": 0.04372,
      "grad_norm": 0.45131221099147395,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 4372
    },
    {
      "epoch": 0.04373,
      "grad_norm": 0.5064889554273249,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 4373
    },
    {
      "epoch": 0.04374,
      "grad_norm": 0.6523859432482797,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 4374
    },
    {
      "epoch": 0.04375,
      "grad_norm": 0.7145171139398658,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 4375
    },
    {
      "epoch": 0.04376,
      "grad_norm": 0.6067647597260136,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 4376
    },
    {
      "epoch": 0.04377,
      "grad_norm": 0.4114371807638281,
      "learning_rate": 0.003,
      "loss": 4.1817,
      "step": 4377
    },
    {
      "epoch": 0.04378,
      "grad_norm": 0.4767337347167691,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 4378
    },
    {
      "epoch": 0.04379,
      "grad_norm": 0.5776389327364934,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 4379
    },
    {
      "epoch": 0.0438,
      "grad_norm": 0.5129468592408087,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 4380
    },
    {
      "epoch": 0.04381,
      "grad_norm": 0.4520287221841305,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 4381
    },
    {
      "epoch": 0.04382,
      "grad_norm": 0.5197515685911448,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4382
    },
    {
      "epoch": 0.04383,
      "grad_norm": 0.5177934608058117,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4383
    },
    {
      "epoch": 0.04384,
      "grad_norm": 0.5061115119488518,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 4384
    },
    {
      "epoch": 0.04385,
      "grad_norm": 0.4979837350700543,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 4385
    },
    {
      "epoch": 0.04386,
      "grad_norm": 0.5170426202552209,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 4386
    },
    {
      "epoch": 0.04387,
      "grad_norm": 0.497700525982779,
      "learning_rate": 0.003,
      "loss": 4.164,
      "step": 4387
    },
    {
      "epoch": 0.04388,
      "grad_norm": 0.5110241434141121,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 4388
    },
    {
      "epoch": 0.04389,
      "grad_norm": 0.47594850470553707,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 4389
    },
    {
      "epoch": 0.0439,
      "grad_norm": 0.5028500411846144,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4390
    },
    {
      "epoch": 0.04391,
      "grad_norm": 0.4958670611124616,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 4391
    },
    {
      "epoch": 0.04392,
      "grad_norm": 0.45433519309198606,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 4392
    },
    {
      "epoch": 0.04393,
      "grad_norm": 0.42745058981506323,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 4393
    },
    {
      "epoch": 0.04394,
      "grad_norm": 0.4270590633401501,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4394
    },
    {
      "epoch": 0.04395,
      "grad_norm": 0.48083329680294346,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 4395
    },
    {
      "epoch": 0.04396,
      "grad_norm": 0.4459452323390242,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 4396
    },
    {
      "epoch": 0.04397,
      "grad_norm": 0.4716168217418122,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 4397
    },
    {
      "epoch": 0.04398,
      "grad_norm": 0.5721086479882349,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 4398
    },
    {
      "epoch": 0.04399,
      "grad_norm": 0.629767772703653,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 4399
    },
    {
      "epoch": 0.044,
      "grad_norm": 0.6149549402399486,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 4400
    },
    {
      "epoch": 0.04401,
      "grad_norm": 0.573854062146023,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 4401
    },
    {
      "epoch": 0.04402,
      "grad_norm": 0.5937371470991296,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4402
    },
    {
      "epoch": 0.04403,
      "grad_norm": 0.7181194724479543,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 4403
    },
    {
      "epoch": 0.04404,
      "grad_norm": 0.7454554555523588,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 4404
    },
    {
      "epoch": 0.04405,
      "grad_norm": 0.7902792519841142,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 4405
    },
    {
      "epoch": 0.04406,
      "grad_norm": 0.7207216322892295,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 4406
    },
    {
      "epoch": 0.04407,
      "grad_norm": 0.6745132067112365,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 4407
    },
    {
      "epoch": 0.04408,
      "grad_norm": 0.6455801650691928,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 4408
    },
    {
      "epoch": 0.04409,
      "grad_norm": 0.6394593941217028,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 4409
    },
    {
      "epoch": 0.0441,
      "grad_norm": 0.49947842679511295,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4410
    },
    {
      "epoch": 0.04411,
      "grad_norm": 0.49572809609459384,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 4411
    },
    {
      "epoch": 0.04412,
      "grad_norm": 0.5979052104111249,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 4412
    },
    {
      "epoch": 0.04413,
      "grad_norm": 0.7297806765539856,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4413
    },
    {
      "epoch": 0.04414,
      "grad_norm": 0.8440545970233042,
      "learning_rate": 0.003,
      "loss": 4.1895,
      "step": 4414
    },
    {
      "epoch": 0.04415,
      "grad_norm": 0.8680059315838015,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 4415
    },
    {
      "epoch": 0.04416,
      "grad_norm": 0.6447487616588353,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 4416
    },
    {
      "epoch": 0.04417,
      "grad_norm": 0.5118950493406013,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 4417
    },
    {
      "epoch": 0.04418,
      "grad_norm": 0.524336962344827,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 4418
    },
    {
      "epoch": 0.04419,
      "grad_norm": 0.5573856100230878,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 4419
    },
    {
      "epoch": 0.0442,
      "grad_norm": 0.5231826126803166,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 4420
    },
    {
      "epoch": 0.04421,
      "grad_norm": 0.5267586062051023,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 4421
    },
    {
      "epoch": 0.04422,
      "grad_norm": 0.6148173783561381,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4422
    },
    {
      "epoch": 0.04423,
      "grad_norm": 0.639677614962493,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 4423
    },
    {
      "epoch": 0.04424,
      "grad_norm": 0.7269549825343812,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 4424
    },
    {
      "epoch": 0.04425,
      "grad_norm": 0.7800602356462868,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 4425
    },
    {
      "epoch": 0.04426,
      "grad_norm": 0.6552710671788101,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4426
    },
    {
      "epoch": 0.04427,
      "grad_norm": 0.6600251269485187,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4427
    },
    {
      "epoch": 0.04428,
      "grad_norm": 0.7074082478081082,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 4428
    },
    {
      "epoch": 0.04429,
      "grad_norm": 0.6687710065867836,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 4429
    },
    {
      "epoch": 0.0443,
      "grad_norm": 0.6237696770618166,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 4430
    },
    {
      "epoch": 0.04431,
      "grad_norm": 0.5184995550940078,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 4431
    },
    {
      "epoch": 0.04432,
      "grad_norm": 0.5237611236731565,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 4432
    },
    {
      "epoch": 0.04433,
      "grad_norm": 0.4836240843515611,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4433
    },
    {
      "epoch": 0.04434,
      "grad_norm": 0.4375263143804872,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 4434
    },
    {
      "epoch": 0.04435,
      "grad_norm": 0.4294571339856274,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 4435
    },
    {
      "epoch": 0.04436,
      "grad_norm": 0.4235691217798773,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 4436
    },
    {
      "epoch": 0.04437,
      "grad_norm": 0.38987028812840846,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 4437
    },
    {
      "epoch": 0.04438,
      "grad_norm": 0.366302357806493,
      "learning_rate": 0.003,
      "loss": 4.1767,
      "step": 4438
    },
    {
      "epoch": 0.04439,
      "grad_norm": 0.39300818537655363,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 4439
    },
    {
      "epoch": 0.0444,
      "grad_norm": 0.4338089650520033,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 4440
    },
    {
      "epoch": 0.04441,
      "grad_norm": 0.48573366379460514,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 4441
    },
    {
      "epoch": 0.04442,
      "grad_norm": 0.5334130138073742,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 4442
    },
    {
      "epoch": 0.04443,
      "grad_norm": 0.5670666571686344,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 4443
    },
    {
      "epoch": 0.04444,
      "grad_norm": 0.5320583177110734,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 4444
    },
    {
      "epoch": 0.04445,
      "grad_norm": 0.489885387553983,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4445
    },
    {
      "epoch": 0.04446,
      "grad_norm": 0.559048561777276,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 4446
    },
    {
      "epoch": 0.04447,
      "grad_norm": 0.6964141693448826,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 4447
    },
    {
      "epoch": 0.04448,
      "grad_norm": 0.7839348711029371,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 4448
    },
    {
      "epoch": 0.04449,
      "grad_norm": 0.8168410712463996,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 4449
    },
    {
      "epoch": 0.0445,
      "grad_norm": 0.7469909706513354,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 4450
    },
    {
      "epoch": 0.04451,
      "grad_norm": 0.7772290265480061,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 4451
    },
    {
      "epoch": 0.04452,
      "grad_norm": 0.750429713415397,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4452
    },
    {
      "epoch": 0.04453,
      "grad_norm": 0.8132686578476264,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 4453
    },
    {
      "epoch": 0.04454,
      "grad_norm": 0.7560392406490976,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 4454
    },
    {
      "epoch": 0.04455,
      "grad_norm": 0.6772193323346015,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 4455
    },
    {
      "epoch": 0.04456,
      "grad_norm": 0.6001846034669396,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 4456
    },
    {
      "epoch": 0.04457,
      "grad_norm": 0.555927693348537,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 4457
    },
    {
      "epoch": 0.04458,
      "grad_norm": 0.567937186369503,
      "learning_rate": 0.003,
      "loss": 4.1625,
      "step": 4458
    },
    {
      "epoch": 0.04459,
      "grad_norm": 0.6268058209460201,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 4459
    },
    {
      "epoch": 0.0446,
      "grad_norm": 0.6774954694650952,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 4460
    },
    {
      "epoch": 0.04461,
      "grad_norm": 0.7113381022778253,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 4461
    },
    {
      "epoch": 0.04462,
      "grad_norm": 0.7089477778475621,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 4462
    },
    {
      "epoch": 0.04463,
      "grad_norm": 0.7149044485381956,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 4463
    },
    {
      "epoch": 0.04464,
      "grad_norm": 0.7001719619214657,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 4464
    },
    {
      "epoch": 0.04465,
      "grad_norm": 0.641827326565005,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4465
    },
    {
      "epoch": 0.04466,
      "grad_norm": 0.5710205222869948,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 4466
    },
    {
      "epoch": 0.04467,
      "grad_norm": 0.5879622422772139,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4467
    },
    {
      "epoch": 0.04468,
      "grad_norm": 0.5833275454344445,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 4468
    },
    {
      "epoch": 0.04469,
      "grad_norm": 0.5035664280245044,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 4469
    },
    {
      "epoch": 0.0447,
      "grad_norm": 0.5330652391453462,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 4470
    },
    {
      "epoch": 0.04471,
      "grad_norm": 0.5415312535775053,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 4471
    },
    {
      "epoch": 0.04472,
      "grad_norm": 0.6201601279104281,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 4472
    },
    {
      "epoch": 0.04473,
      "grad_norm": 0.5823075422954469,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 4473
    },
    {
      "epoch": 0.04474,
      "grad_norm": 0.5211525998630241,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 4474
    },
    {
      "epoch": 0.04475,
      "grad_norm": 0.5787022872529871,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 4475
    },
    {
      "epoch": 0.04476,
      "grad_norm": 0.6268645285401386,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 4476
    },
    {
      "epoch": 0.04477,
      "grad_norm": 0.7238878952105227,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 4477
    },
    {
      "epoch": 0.04478,
      "grad_norm": 0.7625353869137804,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 4478
    },
    {
      "epoch": 0.04479,
      "grad_norm": 0.7572839617963258,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 4479
    },
    {
      "epoch": 0.0448,
      "grad_norm": 0.7894575112052167,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4480
    },
    {
      "epoch": 0.04481,
      "grad_norm": 0.74839779052581,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 4481
    },
    {
      "epoch": 0.04482,
      "grad_norm": 0.7334427098809556,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 4482
    },
    {
      "epoch": 0.04483,
      "grad_norm": 0.7262695320930564,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 4483
    },
    {
      "epoch": 0.04484,
      "grad_norm": 0.6647684381777006,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 4484
    },
    {
      "epoch": 0.04485,
      "grad_norm": 0.6381708748446918,
      "learning_rate": 0.003,
      "loss": 4.1719,
      "step": 4485
    },
    {
      "epoch": 0.04486,
      "grad_norm": 0.5675731699113569,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4486
    },
    {
      "epoch": 0.04487,
      "grad_norm": 0.6472424225913131,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 4487
    },
    {
      "epoch": 0.04488,
      "grad_norm": 0.694626079092874,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 4488
    },
    {
      "epoch": 0.04489,
      "grad_norm": 0.6658909000300576,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4489
    },
    {
      "epoch": 0.0449,
      "grad_norm": 0.5596108130436326,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 4490
    },
    {
      "epoch": 0.04491,
      "grad_norm": 0.4950212459279141,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4491
    },
    {
      "epoch": 0.04492,
      "grad_norm": 0.5347911305158913,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 4492
    },
    {
      "epoch": 0.04493,
      "grad_norm": 0.583677850500047,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 4493
    },
    {
      "epoch": 0.04494,
      "grad_norm": 0.5509029147435317,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 4494
    },
    {
      "epoch": 0.04495,
      "grad_norm": 0.5187917488469639,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 4495
    },
    {
      "epoch": 0.04496,
      "grad_norm": 0.530988798745176,
      "learning_rate": 0.003,
      "loss": 4.1746,
      "step": 4496
    },
    {
      "epoch": 0.04497,
      "grad_norm": 0.5020112887190497,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 4497
    },
    {
      "epoch": 0.04498,
      "grad_norm": 0.5882091094240027,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4498
    },
    {
      "epoch": 0.04499,
      "grad_norm": 0.6882774056835383,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 4499
    },
    {
      "epoch": 0.045,
      "grad_norm": 0.7302113825660921,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 4500
    },
    {
      "epoch": 0.04501,
      "grad_norm": 0.6431193209544874,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 4501
    },
    {
      "epoch": 0.04502,
      "grad_norm": 0.6128789446358794,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 4502
    },
    {
      "epoch": 0.04503,
      "grad_norm": 0.563632556792813,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 4503
    },
    {
      "epoch": 0.04504,
      "grad_norm": 0.49069503580557566,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 4504
    },
    {
      "epoch": 0.04505,
      "grad_norm": 0.5316897314034865,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 4505
    },
    {
      "epoch": 0.04506,
      "grad_norm": 0.485539206424938,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 4506
    },
    {
      "epoch": 0.04507,
      "grad_norm": 0.4651162125504691,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 4507
    },
    {
      "epoch": 0.04508,
      "grad_norm": 0.47136511588197955,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 4508
    },
    {
      "epoch": 0.04509,
      "grad_norm": 0.48078179210487143,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 4509
    },
    {
      "epoch": 0.0451,
      "grad_norm": 0.5697241941596817,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4510
    },
    {
      "epoch": 0.04511,
      "grad_norm": 0.6691571245585908,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 4511
    },
    {
      "epoch": 0.04512,
      "grad_norm": 0.7515863901593889,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 4512
    },
    {
      "epoch": 0.04513,
      "grad_norm": 0.7483616185624625,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 4513
    },
    {
      "epoch": 0.04514,
      "grad_norm": 0.703461108276439,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 4514
    },
    {
      "epoch": 0.04515,
      "grad_norm": 0.7246314655645058,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 4515
    },
    {
      "epoch": 0.04516,
      "grad_norm": 0.6368515120508723,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4516
    },
    {
      "epoch": 0.04517,
      "grad_norm": 0.5222304173800686,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 4517
    },
    {
      "epoch": 0.04518,
      "grad_norm": 0.48504759929098307,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 4518
    },
    {
      "epoch": 0.04519,
      "grad_norm": 0.43996749102619037,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 4519
    },
    {
      "epoch": 0.0452,
      "grad_norm": 0.4756958760047127,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 4520
    },
    {
      "epoch": 0.04521,
      "grad_norm": 0.49681842219333233,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 4521
    },
    {
      "epoch": 0.04522,
      "grad_norm": 0.5177557892182135,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 4522
    },
    {
      "epoch": 0.04523,
      "grad_norm": 0.4834475000365292,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 4523
    },
    {
      "epoch": 0.04524,
      "grad_norm": 0.4294821611706401,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 4524
    },
    {
      "epoch": 0.04525,
      "grad_norm": 0.4784372227541117,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 4525
    },
    {
      "epoch": 0.04526,
      "grad_norm": 0.5155297039314984,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 4526
    },
    {
      "epoch": 0.04527,
      "grad_norm": 0.6810222903041584,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 4527
    },
    {
      "epoch": 0.04528,
      "grad_norm": 0.8822750025167144,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 4528
    },
    {
      "epoch": 0.04529,
      "grad_norm": 0.8919184923159118,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4529
    },
    {
      "epoch": 0.0453,
      "grad_norm": 0.8374148469158906,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 4530
    },
    {
      "epoch": 0.04531,
      "grad_norm": 0.7926240119332635,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4531
    },
    {
      "epoch": 0.04532,
      "grad_norm": 0.8568328376834813,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 4532
    },
    {
      "epoch": 0.04533,
      "grad_norm": 1.0159060122619772,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4533
    },
    {
      "epoch": 0.04534,
      "grad_norm": 0.9536103887238125,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 4534
    },
    {
      "epoch": 0.04535,
      "grad_norm": 0.9201117164869018,
      "learning_rate": 0.003,
      "loss": 4.2131,
      "step": 4535
    },
    {
      "epoch": 0.04536,
      "grad_norm": 1.0425882626135783,
      "learning_rate": 0.003,
      "loss": 4.2164,
      "step": 4536
    },
    {
      "epoch": 0.04537,
      "grad_norm": 1.1171942484625808,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 4537
    },
    {
      "epoch": 0.04538,
      "grad_norm": 1.095605747573551,
      "learning_rate": 0.003,
      "loss": 4.2133,
      "step": 4538
    },
    {
      "epoch": 0.04539,
      "grad_norm": 0.9260638010898388,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 4539
    },
    {
      "epoch": 0.0454,
      "grad_norm": 0.7768806609844044,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 4540
    },
    {
      "epoch": 0.04541,
      "grad_norm": 0.819877783723442,
      "learning_rate": 0.003,
      "loss": 4.2224,
      "step": 4541
    },
    {
      "epoch": 0.04542,
      "grad_norm": 0.8805829563806863,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 4542
    },
    {
      "epoch": 0.04543,
      "grad_norm": 0.7923901917703942,
      "learning_rate": 0.003,
      "loss": 4.207,
      "step": 4543
    },
    {
      "epoch": 0.04544,
      "grad_norm": 0.5941044825763567,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 4544
    },
    {
      "epoch": 0.04545,
      "grad_norm": 0.6408411687252461,
      "learning_rate": 0.003,
      "loss": 4.1822,
      "step": 4545
    },
    {
      "epoch": 0.04546,
      "grad_norm": 0.8237308204937738,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 4546
    },
    {
      "epoch": 0.04547,
      "grad_norm": 0.9235544857511309,
      "learning_rate": 0.003,
      "loss": 4.1822,
      "step": 4547
    },
    {
      "epoch": 0.04548,
      "grad_norm": 0.7235487434091521,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 4548
    },
    {
      "epoch": 0.04549,
      "grad_norm": 0.6422371993419722,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 4549
    },
    {
      "epoch": 0.0455,
      "grad_norm": 0.5600832762072884,
      "learning_rate": 0.003,
      "loss": 4.1988,
      "step": 4550
    },
    {
      "epoch": 0.04551,
      "grad_norm": 0.5228348539847921,
      "learning_rate": 0.003,
      "loss": 4.206,
      "step": 4551
    },
    {
      "epoch": 0.04552,
      "grad_norm": 0.4730508563524038,
      "learning_rate": 0.003,
      "loss": 4.1829,
      "step": 4552
    },
    {
      "epoch": 0.04553,
      "grad_norm": 0.5620543138219484,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 4553
    },
    {
      "epoch": 0.04554,
      "grad_norm": 0.5765966592222326,
      "learning_rate": 0.003,
      "loss": 4.1553,
      "step": 4554
    },
    {
      "epoch": 0.04555,
      "grad_norm": 0.577781369783007,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 4555
    },
    {
      "epoch": 0.04556,
      "grad_norm": 0.5035922467137508,
      "learning_rate": 0.003,
      "loss": 4.1622,
      "step": 4556
    },
    {
      "epoch": 0.04557,
      "grad_norm": 0.5559102933120925,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 4557
    },
    {
      "epoch": 0.04558,
      "grad_norm": 0.7462315222900177,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 4558
    },
    {
      "epoch": 0.04559,
      "grad_norm": 0.7835969334333466,
      "learning_rate": 0.003,
      "loss": 4.1966,
      "step": 4559
    },
    {
      "epoch": 0.0456,
      "grad_norm": 0.6591837955511258,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4560
    },
    {
      "epoch": 0.04561,
      "grad_norm": 0.7187360930779703,
      "learning_rate": 0.003,
      "loss": 4.1604,
      "step": 4561
    },
    {
      "epoch": 0.04562,
      "grad_norm": 0.7184968771307666,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 4562
    },
    {
      "epoch": 0.04563,
      "grad_norm": 0.7171157380489706,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4563
    },
    {
      "epoch": 0.04564,
      "grad_norm": 0.6631256505733477,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 4564
    },
    {
      "epoch": 0.04565,
      "grad_norm": 0.5733198686410141,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4565
    },
    {
      "epoch": 0.04566,
      "grad_norm": 0.4813774167549479,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 4566
    },
    {
      "epoch": 0.04567,
      "grad_norm": 0.40853415838936963,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 4567
    },
    {
      "epoch": 0.04568,
      "grad_norm": 0.37364202822845655,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 4568
    },
    {
      "epoch": 0.04569,
      "grad_norm": 0.3618276444120114,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4569
    },
    {
      "epoch": 0.0457,
      "grad_norm": 0.3609968139572239,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4570
    },
    {
      "epoch": 0.04571,
      "grad_norm": 0.37747195577489445,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 4571
    },
    {
      "epoch": 0.04572,
      "grad_norm": 0.3874865960286465,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 4572
    },
    {
      "epoch": 0.04573,
      "grad_norm": 0.4109072807516651,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 4573
    },
    {
      "epoch": 0.04574,
      "grad_norm": 0.39179890606589207,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 4574
    },
    {
      "epoch": 0.04575,
      "grad_norm": 0.4386907207682461,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 4575
    },
    {
      "epoch": 0.04576,
      "grad_norm": 0.5065967670777235,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 4576
    },
    {
      "epoch": 0.04577,
      "grad_norm": 0.6143902378642073,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 4577
    },
    {
      "epoch": 0.04578,
      "grad_norm": 0.6827140793497046,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 4578
    },
    {
      "epoch": 0.04579,
      "grad_norm": 0.6929677517815375,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 4579
    },
    {
      "epoch": 0.0458,
      "grad_norm": 0.7109361429436778,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4580
    },
    {
      "epoch": 0.04581,
      "grad_norm": 0.5770733016902231,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 4581
    },
    {
      "epoch": 0.04582,
      "grad_norm": 0.4702915512162915,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 4582
    },
    {
      "epoch": 0.04583,
      "grad_norm": 0.45277658262029463,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 4583
    },
    {
      "epoch": 0.04584,
      "grad_norm": 0.4519208543100193,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 4584
    },
    {
      "epoch": 0.04585,
      "grad_norm": 0.4665143329624495,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4585
    },
    {
      "epoch": 0.04586,
      "grad_norm": 0.5475838831331661,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 4586
    },
    {
      "epoch": 0.04587,
      "grad_norm": 0.6898223965554029,
      "learning_rate": 0.003,
      "loss": 4.1575,
      "step": 4587
    },
    {
      "epoch": 0.04588,
      "grad_norm": 0.7350712744563861,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 4588
    },
    {
      "epoch": 0.04589,
      "grad_norm": 0.6924903850485442,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 4589
    },
    {
      "epoch": 0.0459,
      "grad_norm": 0.6678707302029508,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4590
    },
    {
      "epoch": 0.04591,
      "grad_norm": 0.6499473571050832,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 4591
    },
    {
      "epoch": 0.04592,
      "grad_norm": 0.6817526881561883,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4592
    },
    {
      "epoch": 0.04593,
      "grad_norm": 0.6518777120562186,
      "learning_rate": 0.003,
      "loss": 4.1644,
      "step": 4593
    },
    {
      "epoch": 0.04594,
      "grad_norm": 0.5747202505666615,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 4594
    },
    {
      "epoch": 0.04595,
      "grad_norm": 0.5397845618052467,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 4595
    },
    {
      "epoch": 0.04596,
      "grad_norm": 0.510854997128848,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4596
    },
    {
      "epoch": 0.04597,
      "grad_norm": 0.49627082518670596,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4597
    },
    {
      "epoch": 0.04598,
      "grad_norm": 0.530586170374032,
      "learning_rate": 0.003,
      "loss": 4.1793,
      "step": 4598
    },
    {
      "epoch": 0.04599,
      "grad_norm": 0.48072108059262636,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 4599
    },
    {
      "epoch": 0.046,
      "grad_norm": 0.4162968401651856,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4600
    },
    {
      "epoch": 0.04601,
      "grad_norm": 0.4060240320878838,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 4601
    },
    {
      "epoch": 0.04602,
      "grad_norm": 0.4911601212276765,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 4602
    },
    {
      "epoch": 0.04603,
      "grad_norm": 0.5112302821705641,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 4603
    },
    {
      "epoch": 0.04604,
      "grad_norm": 0.5642182423159343,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 4604
    },
    {
      "epoch": 0.04605,
      "grad_norm": 0.6193357979940537,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 4605
    },
    {
      "epoch": 0.04606,
      "grad_norm": 0.7047357474337432,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 4606
    },
    {
      "epoch": 0.04607,
      "grad_norm": 0.782876195240902,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4607
    },
    {
      "epoch": 0.04608,
      "grad_norm": 0.6985321303741415,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 4608
    },
    {
      "epoch": 0.04609,
      "grad_norm": 0.49949454925317704,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 4609
    },
    {
      "epoch": 0.0461,
      "grad_norm": 0.4635494280797984,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 4610
    },
    {
      "epoch": 0.04611,
      "grad_norm": 0.4523573678413289,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 4611
    },
    {
      "epoch": 0.04612,
      "grad_norm": 0.5042320328946598,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 4612
    },
    {
      "epoch": 0.04613,
      "grad_norm": 0.5210913810336651,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 4613
    },
    {
      "epoch": 0.04614,
      "grad_norm": 0.6245444112274343,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 4614
    },
    {
      "epoch": 0.04615,
      "grad_norm": 0.6535761170229347,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 4615
    },
    {
      "epoch": 0.04616,
      "grad_norm": 0.6420802226733096,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 4616
    },
    {
      "epoch": 0.04617,
      "grad_norm": 0.5687953058470812,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 4617
    },
    {
      "epoch": 0.04618,
      "grad_norm": 0.5083532285672069,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 4618
    },
    {
      "epoch": 0.04619,
      "grad_norm": 0.5265907169405639,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 4619
    },
    {
      "epoch": 0.0462,
      "grad_norm": 0.5890366123582629,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 4620
    },
    {
      "epoch": 0.04621,
      "grad_norm": 0.6504323601577833,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 4621
    },
    {
      "epoch": 0.04622,
      "grad_norm": 0.6707598130254062,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 4622
    },
    {
      "epoch": 0.04623,
      "grad_norm": 0.6302847440276245,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4623
    },
    {
      "epoch": 0.04624,
      "grad_norm": 0.6466984839849552,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 4624
    },
    {
      "epoch": 0.04625,
      "grad_norm": 0.6909336873679446,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4625
    },
    {
      "epoch": 0.04626,
      "grad_norm": 0.6100952148809866,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 4626
    },
    {
      "epoch": 0.04627,
      "grad_norm": 0.54274307554158,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 4627
    },
    {
      "epoch": 0.04628,
      "grad_norm": 0.5405464195031551,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 4628
    },
    {
      "epoch": 0.04629,
      "grad_norm": 0.5457077621265701,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 4629
    },
    {
      "epoch": 0.0463,
      "grad_norm": 0.5985608016616293,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 4630
    },
    {
      "epoch": 0.04631,
      "grad_norm": 0.6821487067722284,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4631
    },
    {
      "epoch": 0.04632,
      "grad_norm": 0.7372265744416058,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 4632
    },
    {
      "epoch": 0.04633,
      "grad_norm": 0.8415550301042956,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 4633
    },
    {
      "epoch": 0.04634,
      "grad_norm": 0.8373989629700342,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 4634
    },
    {
      "epoch": 0.04635,
      "grad_norm": 0.7581438335563392,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 4635
    },
    {
      "epoch": 0.04636,
      "grad_norm": 0.7125351605332876,
      "learning_rate": 0.003,
      "loss": 4.183,
      "step": 4636
    },
    {
      "epoch": 0.04637,
      "grad_norm": 0.7232757442184159,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4637
    },
    {
      "epoch": 0.04638,
      "grad_norm": 0.6757057573333891,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 4638
    },
    {
      "epoch": 0.04639,
      "grad_norm": 0.719912413703624,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 4639
    },
    {
      "epoch": 0.0464,
      "grad_norm": 0.7428965743218653,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4640
    },
    {
      "epoch": 0.04641,
      "grad_norm": 0.8598211728393256,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 4641
    },
    {
      "epoch": 0.04642,
      "grad_norm": 0.8907386217362441,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 4642
    },
    {
      "epoch": 0.04643,
      "grad_norm": 0.7795644135221422,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 4643
    },
    {
      "epoch": 0.04644,
      "grad_norm": 0.5620211951242756,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4644
    },
    {
      "epoch": 0.04645,
      "grad_norm": 0.5137991034790863,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 4645
    },
    {
      "epoch": 0.04646,
      "grad_norm": 0.5506672519917369,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 4646
    },
    {
      "epoch": 0.04647,
      "grad_norm": 0.545899380374656,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 4647
    },
    {
      "epoch": 0.04648,
      "grad_norm": 0.5926193733697508,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 4648
    },
    {
      "epoch": 0.04649,
      "grad_norm": 0.6418478720978149,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4649
    },
    {
      "epoch": 0.0465,
      "grad_norm": 0.6971029121589252,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 4650
    },
    {
      "epoch": 0.04651,
      "grad_norm": 0.7128225140702158,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4651
    },
    {
      "epoch": 0.04652,
      "grad_norm": 0.6817984613215538,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4652
    },
    {
      "epoch": 0.04653,
      "grad_norm": 0.6586749661533192,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 4653
    },
    {
      "epoch": 0.04654,
      "grad_norm": 0.5413982579807713,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 4654
    },
    {
      "epoch": 0.04655,
      "grad_norm": 0.4819381736457758,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4655
    },
    {
      "epoch": 0.04656,
      "grad_norm": 0.5090740776020435,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 4656
    },
    {
      "epoch": 0.04657,
      "grad_norm": 0.4529973940302403,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 4657
    },
    {
      "epoch": 0.04658,
      "grad_norm": 0.3863158282687969,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 4658
    },
    {
      "epoch": 0.04659,
      "grad_norm": 0.3846174175293322,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 4659
    },
    {
      "epoch": 0.0466,
      "grad_norm": 0.33629970185853586,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4660
    },
    {
      "epoch": 0.04661,
      "grad_norm": 0.3733824492556284,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 4661
    },
    {
      "epoch": 0.04662,
      "grad_norm": 0.40706146799826654,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 4662
    },
    {
      "epoch": 0.04663,
      "grad_norm": 0.4624049284667894,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 4663
    },
    {
      "epoch": 0.04664,
      "grad_norm": 0.6204385926386629,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 4664
    },
    {
      "epoch": 0.04665,
      "grad_norm": 0.8486297396587145,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 4665
    },
    {
      "epoch": 0.04666,
      "grad_norm": 0.8396332728949535,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4666
    },
    {
      "epoch": 0.04667,
      "grad_norm": 0.604029857762682,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 4667
    },
    {
      "epoch": 0.04668,
      "grad_norm": 0.6187443793003241,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 4668
    },
    {
      "epoch": 0.04669,
      "grad_norm": 0.671068656091709,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 4669
    },
    {
      "epoch": 0.0467,
      "grad_norm": 0.582982047763638,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 4670
    },
    {
      "epoch": 0.04671,
      "grad_norm": 0.5791042512401672,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4671
    },
    {
      "epoch": 0.04672,
      "grad_norm": 0.5822833904953063,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4672
    },
    {
      "epoch": 0.04673,
      "grad_norm": 0.6690185777600851,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 4673
    },
    {
      "epoch": 0.04674,
      "grad_norm": 0.6501772820492476,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 4674
    },
    {
      "epoch": 0.04675,
      "grad_norm": 0.5619140983596763,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 4675
    },
    {
      "epoch": 0.04676,
      "grad_norm": 0.4844274130103253,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 4676
    },
    {
      "epoch": 0.04677,
      "grad_norm": 0.46020296554025397,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 4677
    },
    {
      "epoch": 0.04678,
      "grad_norm": 0.4560378554438905,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 4678
    },
    {
      "epoch": 0.04679,
      "grad_norm": 0.48237840297163587,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 4679
    },
    {
      "epoch": 0.0468,
      "grad_norm": 0.545076677673247,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 4680
    },
    {
      "epoch": 0.04681,
      "grad_norm": 0.6224973549149077,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 4681
    },
    {
      "epoch": 0.04682,
      "grad_norm": 0.7719478560198243,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 4682
    },
    {
      "epoch": 0.04683,
      "grad_norm": 0.8064115060114956,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 4683
    },
    {
      "epoch": 0.04684,
      "grad_norm": 0.8855175249633898,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 4684
    },
    {
      "epoch": 0.04685,
      "grad_norm": 0.8534062578922067,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 4685
    },
    {
      "epoch": 0.04686,
      "grad_norm": 0.797573707115956,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 4686
    },
    {
      "epoch": 0.04687,
      "grad_norm": 0.713270606155539,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 4687
    },
    {
      "epoch": 0.04688,
      "grad_norm": 0.6241645872789137,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 4688
    },
    {
      "epoch": 0.04689,
      "grad_norm": 0.559937415479369,
      "learning_rate": 0.003,
      "loss": 4.1712,
      "step": 4689
    },
    {
      "epoch": 0.0469,
      "grad_norm": 0.5404635622147993,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 4690
    },
    {
      "epoch": 0.04691,
      "grad_norm": 0.486653564258482,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 4691
    },
    {
      "epoch": 0.04692,
      "grad_norm": 0.481468369732997,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 4692
    },
    {
      "epoch": 0.04693,
      "grad_norm": 0.5011365510399343,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 4693
    },
    {
      "epoch": 0.04694,
      "grad_norm": 0.48431784153510343,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 4694
    },
    {
      "epoch": 0.04695,
      "grad_norm": 0.5590735035376782,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 4695
    },
    {
      "epoch": 0.04696,
      "grad_norm": 0.5991808838927974,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 4696
    },
    {
      "epoch": 0.04697,
      "grad_norm": 0.6802689730153363,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 4697
    },
    {
      "epoch": 0.04698,
      "grad_norm": 0.6842952307392385,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 4698
    },
    {
      "epoch": 0.04699,
      "grad_norm": 0.6034505759619246,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 4699
    },
    {
      "epoch": 0.047,
      "grad_norm": 0.5907548018053599,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4700
    },
    {
      "epoch": 0.04701,
      "grad_norm": 0.5241329554797072,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 4701
    },
    {
      "epoch": 0.04702,
      "grad_norm": 0.5399206512572382,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 4702
    },
    {
      "epoch": 0.04703,
      "grad_norm": 0.5629949677153792,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 4703
    },
    {
      "epoch": 0.04704,
      "grad_norm": 0.4982163977250413,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 4704
    },
    {
      "epoch": 0.04705,
      "grad_norm": 0.5165399833831362,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 4705
    },
    {
      "epoch": 0.04706,
      "grad_norm": 0.5588859123289356,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 4706
    },
    {
      "epoch": 0.04707,
      "grad_norm": 0.6493712508280153,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4707
    },
    {
      "epoch": 0.04708,
      "grad_norm": 0.7592099124000022,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 4708
    },
    {
      "epoch": 0.04709,
      "grad_norm": 0.7643143840544172,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 4709
    },
    {
      "epoch": 0.0471,
      "grad_norm": 0.7117650865961848,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 4710
    },
    {
      "epoch": 0.04711,
      "grad_norm": 0.6222639694626878,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4711
    },
    {
      "epoch": 0.04712,
      "grad_norm": 0.5676066371017507,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 4712
    },
    {
      "epoch": 0.04713,
      "grad_norm": 0.5770424755113991,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 4713
    },
    {
      "epoch": 0.04714,
      "grad_norm": 0.5617675778124301,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4714
    },
    {
      "epoch": 0.04715,
      "grad_norm": 0.5635879231709099,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 4715
    },
    {
      "epoch": 0.04716,
      "grad_norm": 0.5756399495553071,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 4716
    },
    {
      "epoch": 0.04717,
      "grad_norm": 0.5516676091531315,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 4717
    },
    {
      "epoch": 0.04718,
      "grad_norm": 0.5287651595406879,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 4718
    },
    {
      "epoch": 0.04719,
      "grad_norm": 0.5253847377704106,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 4719
    },
    {
      "epoch": 0.0472,
      "grad_norm": 0.5350874385832446,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 4720
    },
    {
      "epoch": 0.04721,
      "grad_norm": 0.6004531613918229,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 4721
    },
    {
      "epoch": 0.04722,
      "grad_norm": 0.7164746962025985,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 4722
    },
    {
      "epoch": 0.04723,
      "grad_norm": 0.7305093269163504,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 4723
    },
    {
      "epoch": 0.04724,
      "grad_norm": 0.6386395939306334,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 4724
    },
    {
      "epoch": 0.04725,
      "grad_norm": 0.5477055183445729,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 4725
    },
    {
      "epoch": 0.04726,
      "grad_norm": 0.6698851087894558,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 4726
    },
    {
      "epoch": 0.04727,
      "grad_norm": 0.6952010909817607,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 4727
    },
    {
      "epoch": 0.04728,
      "grad_norm": 0.6237561354472855,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 4728
    },
    {
      "epoch": 0.04729,
      "grad_norm": 0.5824726583084653,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 4729
    },
    {
      "epoch": 0.0473,
      "grad_norm": 0.6234484457938285,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 4730
    },
    {
      "epoch": 0.04731,
      "grad_norm": 0.7040984462948884,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 4731
    },
    {
      "epoch": 0.04732,
      "grad_norm": 0.8060281611564052,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 4732
    },
    {
      "epoch": 0.04733,
      "grad_norm": 0.8908771747222946,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4733
    },
    {
      "epoch": 0.04734,
      "grad_norm": 0.8915755383205525,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4734
    },
    {
      "epoch": 0.04735,
      "grad_norm": 0.8597675990151213,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 4735
    },
    {
      "epoch": 0.04736,
      "grad_norm": 0.9421924904802637,
      "learning_rate": 0.003,
      "loss": 4.1714,
      "step": 4736
    },
    {
      "epoch": 0.04737,
      "grad_norm": 0.9723514275784639,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 4737
    },
    {
      "epoch": 0.04738,
      "grad_norm": 0.8765529889726066,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 4738
    },
    {
      "epoch": 0.04739,
      "grad_norm": 0.6891707628164019,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 4739
    },
    {
      "epoch": 0.0474,
      "grad_norm": 0.6320172894482899,
      "learning_rate": 0.003,
      "loss": 4.1783,
      "step": 4740
    },
    {
      "epoch": 0.04741,
      "grad_norm": 0.6185420901083515,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 4741
    },
    {
      "epoch": 0.04742,
      "grad_norm": 0.5743911871453854,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 4742
    },
    {
      "epoch": 0.04743,
      "grad_norm": 0.5220289188801445,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 4743
    },
    {
      "epoch": 0.04744,
      "grad_norm": 0.509105103411712,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 4744
    },
    {
      "epoch": 0.04745,
      "grad_norm": 0.47978510769114535,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 4745
    },
    {
      "epoch": 0.04746,
      "grad_norm": 0.5254663892408653,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 4746
    },
    {
      "epoch": 0.04747,
      "grad_norm": 0.5939317340288737,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 4747
    },
    {
      "epoch": 0.04748,
      "grad_norm": 0.5634019114968404,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 4748
    },
    {
      "epoch": 0.04749,
      "grad_norm": 0.5497840563348653,
      "learning_rate": 0.003,
      "loss": 4.1719,
      "step": 4749
    },
    {
      "epoch": 0.0475,
      "grad_norm": 0.5714725339017903,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 4750
    },
    {
      "epoch": 0.04751,
      "grad_norm": 0.6084741881095581,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 4751
    },
    {
      "epoch": 0.04752,
      "grad_norm": 0.6608722355986122,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 4752
    },
    {
      "epoch": 0.04753,
      "grad_norm": 0.7109251785941167,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4753
    },
    {
      "epoch": 0.04754,
      "grad_norm": 0.6628914079263116,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 4754
    },
    {
      "epoch": 0.04755,
      "grad_norm": 0.5644082779422211,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 4755
    },
    {
      "epoch": 0.04756,
      "grad_norm": 0.5287140638812969,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4756
    },
    {
      "epoch": 0.04757,
      "grad_norm": 0.49259530652550815,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 4757
    },
    {
      "epoch": 0.04758,
      "grad_norm": 0.5614284037719612,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 4758
    },
    {
      "epoch": 0.04759,
      "grad_norm": 0.5959601501901447,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 4759
    },
    {
      "epoch": 0.0476,
      "grad_norm": 0.6904403583130148,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 4760
    },
    {
      "epoch": 0.04761,
      "grad_norm": 0.7246423353646352,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 4761
    },
    {
      "epoch": 0.04762,
      "grad_norm": 0.6915319453958364,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 4762
    },
    {
      "epoch": 0.04763,
      "grad_norm": 0.6256527340482957,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4763
    },
    {
      "epoch": 0.04764,
      "grad_norm": 0.6439794931734228,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 4764
    },
    {
      "epoch": 0.04765,
      "grad_norm": 0.6077799243331904,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 4765
    },
    {
      "epoch": 0.04766,
      "grad_norm": 0.5987824824691682,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 4766
    },
    {
      "epoch": 0.04767,
      "grad_norm": 0.5323630132045762,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 4767
    },
    {
      "epoch": 0.04768,
      "grad_norm": 0.49097893919403107,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 4768
    },
    {
      "epoch": 0.04769,
      "grad_norm": 0.41538834375199724,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 4769
    },
    {
      "epoch": 0.0477,
      "grad_norm": 0.4062286279109976,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 4770
    },
    {
      "epoch": 0.04771,
      "grad_norm": 0.3969385937952626,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 4771
    },
    {
      "epoch": 0.04772,
      "grad_norm": 0.47881496679265567,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 4772
    },
    {
      "epoch": 0.04773,
      "grad_norm": 0.5857337921957999,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 4773
    },
    {
      "epoch": 0.04774,
      "grad_norm": 0.6545046565097566,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 4774
    },
    {
      "epoch": 0.04775,
      "grad_norm": 0.6909582502342425,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 4775
    },
    {
      "epoch": 0.04776,
      "grad_norm": 0.6830425397433746,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 4776
    },
    {
      "epoch": 0.04777,
      "grad_norm": 0.5407062720120711,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 4777
    },
    {
      "epoch": 0.04778,
      "grad_norm": 0.4928003344996406,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 4778
    },
    {
      "epoch": 0.04779,
      "grad_norm": 0.5540563562005282,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 4779
    },
    {
      "epoch": 0.0478,
      "grad_norm": 0.5919870365085376,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 4780
    },
    {
      "epoch": 0.04781,
      "grad_norm": 0.6502125533927232,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 4781
    },
    {
      "epoch": 0.04782,
      "grad_norm": 0.5583168452423385,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 4782
    },
    {
      "epoch": 0.04783,
      "grad_norm": 0.49694295687820944,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 4783
    },
    {
      "epoch": 0.04784,
      "grad_norm": 0.4979847074086822,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 4784
    },
    {
      "epoch": 0.04785,
      "grad_norm": 0.5392917734790091,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 4785
    },
    {
      "epoch": 0.04786,
      "grad_norm": 0.5426680689275899,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 4786
    },
    {
      "epoch": 0.04787,
      "grad_norm": 0.5518443356995324,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4787
    },
    {
      "epoch": 0.04788,
      "grad_norm": 0.6334713170503672,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 4788
    },
    {
      "epoch": 0.04789,
      "grad_norm": 0.6443157992564054,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 4789
    },
    {
      "epoch": 0.0479,
      "grad_norm": 0.6616151594742191,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 4790
    },
    {
      "epoch": 0.04791,
      "grad_norm": 0.7664115436248651,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 4791
    },
    {
      "epoch": 0.04792,
      "grad_norm": 0.8885911114335623,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 4792
    },
    {
      "epoch": 0.04793,
      "grad_norm": 1.0730053893316884,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 4793
    },
    {
      "epoch": 0.04794,
      "grad_norm": 0.7779447364165427,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 4794
    },
    {
      "epoch": 0.04795,
      "grad_norm": 0.588536399495119,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4795
    },
    {
      "epoch": 0.04796,
      "grad_norm": 0.68568663004306,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 4796
    },
    {
      "epoch": 0.04797,
      "grad_norm": 0.681720094933613,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 4797
    },
    {
      "epoch": 0.04798,
      "grad_norm": 0.7679505695308819,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 4798
    },
    {
      "epoch": 0.04799,
      "grad_norm": 0.7317665379567332,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 4799
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.6100825656074705,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 4800
    },
    {
      "epoch": 0.04801,
      "grad_norm": 0.5012414047123679,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 4801
    },
    {
      "epoch": 0.04802,
      "grad_norm": 0.4479983336578663,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 4802
    },
    {
      "epoch": 0.04803,
      "grad_norm": 0.4184616004474593,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 4803
    },
    {
      "epoch": 0.04804,
      "grad_norm": 0.41305580441153505,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4804
    },
    {
      "epoch": 0.04805,
      "grad_norm": 0.4398073604549804,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 4805
    },
    {
      "epoch": 0.04806,
      "grad_norm": 0.42743235225805976,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 4806
    },
    {
      "epoch": 0.04807,
      "grad_norm": 0.4727115258160579,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 4807
    },
    {
      "epoch": 0.04808,
      "grad_norm": 0.5919755301104812,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 4808
    },
    {
      "epoch": 0.04809,
      "grad_norm": 0.7289693894093642,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 4809
    },
    {
      "epoch": 0.0481,
      "grad_norm": 0.7839859175085296,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 4810
    },
    {
      "epoch": 0.04811,
      "grad_norm": 0.6884295236179085,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4811
    },
    {
      "epoch": 0.04812,
      "grad_norm": 0.7352119188470826,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 4812
    },
    {
      "epoch": 0.04813,
      "grad_norm": 0.695166473264558,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 4813
    },
    {
      "epoch": 0.04814,
      "grad_norm": 0.6553813909063907,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 4814
    },
    {
      "epoch": 0.04815,
      "grad_norm": 0.5872101085364063,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 4815
    },
    {
      "epoch": 0.04816,
      "grad_norm": 0.6233113378649032,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 4816
    },
    {
      "epoch": 0.04817,
      "grad_norm": 0.6150064661388728,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 4817
    },
    {
      "epoch": 0.04818,
      "grad_norm": 0.5836692065444768,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 4818
    },
    {
      "epoch": 0.04819,
      "grad_norm": 0.66525230376135,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 4819
    },
    {
      "epoch": 0.0482,
      "grad_norm": 0.734508856599843,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 4820
    },
    {
      "epoch": 0.04821,
      "grad_norm": 0.7602706927514201,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 4821
    },
    {
      "epoch": 0.04822,
      "grad_norm": 0.7876820121443141,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 4822
    },
    {
      "epoch": 0.04823,
      "grad_norm": 0.7863376722170511,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 4823
    },
    {
      "epoch": 0.04824,
      "grad_norm": 0.8763060513910201,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4824
    },
    {
      "epoch": 0.04825,
      "grad_norm": 0.8932088904498784,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 4825
    },
    {
      "epoch": 0.04826,
      "grad_norm": 0.7974932345630039,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 4826
    },
    {
      "epoch": 0.04827,
      "grad_norm": 0.787269582726668,
      "learning_rate": 0.003,
      "loss": 4.1569,
      "step": 4827
    },
    {
      "epoch": 0.04828,
      "grad_norm": 0.8334258319646332,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 4828
    },
    {
      "epoch": 0.04829,
      "grad_norm": 0.8348604948088247,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4829
    },
    {
      "epoch": 0.0483,
      "grad_norm": 0.8624019255671944,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 4830
    },
    {
      "epoch": 0.04831,
      "grad_norm": 0.9982066722626229,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 4831
    },
    {
      "epoch": 0.04832,
      "grad_norm": 0.7923904912144631,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 4832
    },
    {
      "epoch": 0.04833,
      "grad_norm": 0.7754468674602506,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 4833
    },
    {
      "epoch": 0.04834,
      "grad_norm": 0.6518452811577893,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4834
    },
    {
      "epoch": 0.04835,
      "grad_norm": 0.7302318791564857,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 4835
    },
    {
      "epoch": 0.04836,
      "grad_norm": 0.6773722956545634,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 4836
    },
    {
      "epoch": 0.04837,
      "grad_norm": 0.5824847865414027,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 4837
    },
    {
      "epoch": 0.04838,
      "grad_norm": 0.5995229116547709,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 4838
    },
    {
      "epoch": 0.04839,
      "grad_norm": 0.6433667604247222,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 4839
    },
    {
      "epoch": 0.0484,
      "grad_norm": 0.6015686841024611,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4840
    },
    {
      "epoch": 0.04841,
      "grad_norm": 0.513992549739537,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 4841
    },
    {
      "epoch": 0.04842,
      "grad_norm": 0.5245975343951695,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 4842
    },
    {
      "epoch": 0.04843,
      "grad_norm": 0.48743719353902437,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 4843
    },
    {
      "epoch": 0.04844,
      "grad_norm": 0.4307867246160189,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 4844
    },
    {
      "epoch": 0.04845,
      "grad_norm": 0.3751006510902314,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 4845
    },
    {
      "epoch": 0.04846,
      "grad_norm": 0.39953129595847076,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 4846
    },
    {
      "epoch": 0.04847,
      "grad_norm": 0.36070808802461135,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 4847
    },
    {
      "epoch": 0.04848,
      "grad_norm": 0.39594455872525836,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4848
    },
    {
      "epoch": 0.04849,
      "grad_norm": 0.38000994045173514,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4849
    },
    {
      "epoch": 0.0485,
      "grad_norm": 0.4205813881723009,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 4850
    },
    {
      "epoch": 0.04851,
      "grad_norm": 0.4045018073402591,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 4851
    },
    {
      "epoch": 0.04852,
      "grad_norm": 0.37568412955203784,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 4852
    },
    {
      "epoch": 0.04853,
      "grad_norm": 0.3455978343249278,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 4853
    },
    {
      "epoch": 0.04854,
      "grad_norm": 0.3952113128712618,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 4854
    },
    {
      "epoch": 0.04855,
      "grad_norm": 0.5622316469812096,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 4855
    },
    {
      "epoch": 0.04856,
      "grad_norm": 0.8797178488644849,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4856
    },
    {
      "epoch": 0.04857,
      "grad_norm": 1.1335123213922318,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 4857
    },
    {
      "epoch": 0.04858,
      "grad_norm": 0.6147110414360265,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 4858
    },
    {
      "epoch": 0.04859,
      "grad_norm": 0.6973892953945595,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 4859
    },
    {
      "epoch": 0.0486,
      "grad_norm": 0.7043154871004238,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 4860
    },
    {
      "epoch": 0.04861,
      "grad_norm": 0.7327036767884083,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 4861
    },
    {
      "epoch": 0.04862,
      "grad_norm": 0.712112403189955,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4862
    },
    {
      "epoch": 0.04863,
      "grad_norm": 0.7121663551182991,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 4863
    },
    {
      "epoch": 0.04864,
      "grad_norm": 0.6310157347751136,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 4864
    },
    {
      "epoch": 0.04865,
      "grad_norm": 0.6082126733812362,
      "learning_rate": 0.003,
      "loss": 4.1569,
      "step": 4865
    },
    {
      "epoch": 0.04866,
      "grad_norm": 0.569379004286016,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 4866
    },
    {
      "epoch": 0.04867,
      "grad_norm": 0.543390213497262,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 4867
    },
    {
      "epoch": 0.04868,
      "grad_norm": 0.5636989030319737,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4868
    },
    {
      "epoch": 0.04869,
      "grad_norm": 0.5250393396952697,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4869
    },
    {
      "epoch": 0.0487,
      "grad_norm": 0.4712102787061369,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 4870
    },
    {
      "epoch": 0.04871,
      "grad_norm": 0.4137409461601053,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 4871
    },
    {
      "epoch": 0.04872,
      "grad_norm": 0.34190339505881295,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 4872
    },
    {
      "epoch": 0.04873,
      "grad_norm": 0.3786358077114681,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 4873
    },
    {
      "epoch": 0.04874,
      "grad_norm": 0.3569458423955174,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 4874
    },
    {
      "epoch": 0.04875,
      "grad_norm": 0.34447042337429823,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 4875
    },
    {
      "epoch": 0.04876,
      "grad_norm": 0.40030916663729343,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 4876
    },
    {
      "epoch": 0.04877,
      "grad_norm": 0.4592360630373684,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 4877
    },
    {
      "epoch": 0.04878,
      "grad_norm": 0.47334555855094895,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 4878
    },
    {
      "epoch": 0.04879,
      "grad_norm": 0.4994212755863587,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 4879
    },
    {
      "epoch": 0.0488,
      "grad_norm": 0.5513466708644019,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 4880
    },
    {
      "epoch": 0.04881,
      "grad_norm": 0.6373315380482077,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 4881
    },
    {
      "epoch": 0.04882,
      "grad_norm": 0.6684216804530342,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 4882
    },
    {
      "epoch": 0.04883,
      "grad_norm": 0.5368064523422058,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 4883
    },
    {
      "epoch": 0.04884,
      "grad_norm": 0.5972330760546537,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 4884
    },
    {
      "epoch": 0.04885,
      "grad_norm": 0.8933715601802283,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 4885
    },
    {
      "epoch": 0.04886,
      "grad_norm": 0.898355015821335,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 4886
    },
    {
      "epoch": 0.04887,
      "grad_norm": 0.7173276903905438,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 4887
    },
    {
      "epoch": 0.04888,
      "grad_norm": 0.6726150274910551,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 4888
    },
    {
      "epoch": 0.04889,
      "grad_norm": 0.6876961916754174,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 4889
    },
    {
      "epoch": 0.0489,
      "grad_norm": 0.6584901279572161,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 4890
    },
    {
      "epoch": 0.04891,
      "grad_norm": 0.5884964182869652,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 4891
    },
    {
      "epoch": 0.04892,
      "grad_norm": 0.7007131660081303,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4892
    },
    {
      "epoch": 0.04893,
      "grad_norm": 0.856434726166467,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 4893
    },
    {
      "epoch": 0.04894,
      "grad_norm": 0.8623969209321906,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 4894
    },
    {
      "epoch": 0.04895,
      "grad_norm": 0.9386453675933166,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 4895
    },
    {
      "epoch": 0.04896,
      "grad_norm": 1.0225765758614995,
      "learning_rate": 0.003,
      "loss": 4.174,
      "step": 4896
    },
    {
      "epoch": 0.04897,
      "grad_norm": 1.0592162822847404,
      "learning_rate": 0.003,
      "loss": 4.1956,
      "step": 4897
    },
    {
      "epoch": 0.04898,
      "grad_norm": 0.9806770790608599,
      "learning_rate": 0.003,
      "loss": 4.1844,
      "step": 4898
    },
    {
      "epoch": 0.04899,
      "grad_norm": 1.0087204735328514,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 4899
    },
    {
      "epoch": 0.049,
      "grad_norm": 0.9807645706509096,
      "learning_rate": 0.003,
      "loss": 4.2071,
      "step": 4900
    },
    {
      "epoch": 0.04901,
      "grad_norm": 0.8655711245910307,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 4901
    },
    {
      "epoch": 0.04902,
      "grad_norm": 0.7265751482705578,
      "learning_rate": 0.003,
      "loss": 4.1956,
      "step": 4902
    },
    {
      "epoch": 0.04903,
      "grad_norm": 0.7751913689828299,
      "learning_rate": 0.003,
      "loss": 4.1967,
      "step": 4903
    },
    {
      "epoch": 0.04904,
      "grad_norm": 0.791355963573371,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 4904
    },
    {
      "epoch": 0.04905,
      "grad_norm": 0.7148922976648996,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 4905
    },
    {
      "epoch": 0.04906,
      "grad_norm": 0.6244001153297727,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 4906
    },
    {
      "epoch": 0.04907,
      "grad_norm": 0.766532703122461,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 4907
    },
    {
      "epoch": 0.04908,
      "grad_norm": 0.9239125390916663,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 4908
    },
    {
      "epoch": 0.04909,
      "grad_norm": 0.8062598058153201,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 4909
    },
    {
      "epoch": 0.0491,
      "grad_norm": 0.6645500078640559,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 4910
    },
    {
      "epoch": 0.04911,
      "grad_norm": 0.6285244715158727,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 4911
    },
    {
      "epoch": 0.04912,
      "grad_norm": 0.6953556617834721,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 4912
    },
    {
      "epoch": 0.04913,
      "grad_norm": 0.7284223880047193,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 4913
    },
    {
      "epoch": 0.04914,
      "grad_norm": 0.7207287527598001,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 4914
    },
    {
      "epoch": 0.04915,
      "grad_norm": 0.598886194111445,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 4915
    },
    {
      "epoch": 0.04916,
      "grad_norm": 0.575625916272145,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 4916
    },
    {
      "epoch": 0.04917,
      "grad_norm": 0.6136059699261334,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 4917
    },
    {
      "epoch": 0.04918,
      "grad_norm": 0.5509547477979617,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4918
    },
    {
      "epoch": 0.04919,
      "grad_norm": 0.4500315004245685,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4919
    },
    {
      "epoch": 0.0492,
      "grad_norm": 0.40527549449786815,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 4920
    },
    {
      "epoch": 0.04921,
      "grad_norm": 0.42108054279033674,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 4921
    },
    {
      "epoch": 0.04922,
      "grad_norm": 0.4404322304412163,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 4922
    },
    {
      "epoch": 0.04923,
      "grad_norm": 0.47502761296305623,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 4923
    },
    {
      "epoch": 0.04924,
      "grad_norm": 0.6308445729709988,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 4924
    },
    {
      "epoch": 0.04925,
      "grad_norm": 0.8223519029190577,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 4925
    },
    {
      "epoch": 0.04926,
      "grad_norm": 0.7978760513652026,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 4926
    },
    {
      "epoch": 0.04927,
      "grad_norm": 0.5978087975120926,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 4927
    },
    {
      "epoch": 0.04928,
      "grad_norm": 0.6116960538491134,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 4928
    },
    {
      "epoch": 0.04929,
      "grad_norm": 0.6956029366529681,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 4929
    },
    {
      "epoch": 0.0493,
      "grad_norm": 0.6193307396741099,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 4930
    },
    {
      "epoch": 0.04931,
      "grad_norm": 0.5548956756238017,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 4931
    },
    {
      "epoch": 0.04932,
      "grad_norm": 0.5257796103914479,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 4932
    },
    {
      "epoch": 0.04933,
      "grad_norm": 0.525537631722876,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 4933
    },
    {
      "epoch": 0.04934,
      "grad_norm": 0.5597601882963287,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 4934
    },
    {
      "epoch": 0.04935,
      "grad_norm": 0.5183399428365818,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 4935
    },
    {
      "epoch": 0.04936,
      "grad_norm": 0.45976573800777254,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 4936
    },
    {
      "epoch": 0.04937,
      "grad_norm": 0.4718016863136529,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 4937
    },
    {
      "epoch": 0.04938,
      "grad_norm": 0.5145237878577654,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 4938
    },
    {
      "epoch": 0.04939,
      "grad_norm": 0.49376291714918813,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 4939
    },
    {
      "epoch": 0.0494,
      "grad_norm": 0.40835265558787853,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4940
    },
    {
      "epoch": 0.04941,
      "grad_norm": 0.39634154664022514,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 4941
    },
    {
      "epoch": 0.04942,
      "grad_norm": 0.3697169007385673,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 4942
    },
    {
      "epoch": 0.04943,
      "grad_norm": 0.3331096726429694,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4943
    },
    {
      "epoch": 0.04944,
      "grad_norm": 0.35453569789946104,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 4944
    },
    {
      "epoch": 0.04945,
      "grad_norm": 0.3239762249913837,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4945
    },
    {
      "epoch": 0.04946,
      "grad_norm": 0.31174772171665677,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 4946
    },
    {
      "epoch": 0.04947,
      "grad_norm": 0.2863836162320785,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 4947
    },
    {
      "epoch": 0.04948,
      "grad_norm": 0.2929298130193931,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 4948
    },
    {
      "epoch": 0.04949,
      "grad_norm": 0.3283619905331677,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 4949
    },
    {
      "epoch": 0.0495,
      "grad_norm": 0.35583655037221473,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 4950
    },
    {
      "epoch": 0.04951,
      "grad_norm": 0.4306179937269439,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 4951
    },
    {
      "epoch": 0.04952,
      "grad_norm": 0.5484424313628553,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 4952
    },
    {
      "epoch": 0.04953,
      "grad_norm": 0.683881072985923,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 4953
    },
    {
      "epoch": 0.04954,
      "grad_norm": 0.8387871275000114,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 4954
    },
    {
      "epoch": 0.04955,
      "grad_norm": 1.0719745768190279,
      "learning_rate": 0.003,
      "loss": 4.142,
      "step": 4955
    },
    {
      "epoch": 0.04956,
      "grad_norm": 1.0142567681805608,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 4956
    },
    {
      "epoch": 0.04957,
      "grad_norm": 0.7860187886136664,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 4957
    },
    {
      "epoch": 0.04958,
      "grad_norm": 0.7666806226784963,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 4958
    },
    {
      "epoch": 0.04959,
      "grad_norm": 0.7024610691029676,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4959
    },
    {
      "epoch": 0.0496,
      "grad_norm": 0.634345285618044,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 4960
    },
    {
      "epoch": 0.04961,
      "grad_norm": 0.5794913569394096,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 4961
    },
    {
      "epoch": 0.04962,
      "grad_norm": 0.6259340689173926,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 4962
    },
    {
      "epoch": 0.04963,
      "grad_norm": 0.6609688779347547,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 4963
    },
    {
      "epoch": 0.04964,
      "grad_norm": 0.6719506081449138,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4964
    },
    {
      "epoch": 0.04965,
      "grad_norm": 0.6471802630352936,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 4965
    },
    {
      "epoch": 0.04966,
      "grad_norm": 0.7401257014237659,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 4966
    },
    {
      "epoch": 0.04967,
      "grad_norm": 0.7264688238953103,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 4967
    },
    {
      "epoch": 0.04968,
      "grad_norm": 0.6850630441933709,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 4968
    },
    {
      "epoch": 0.04969,
      "grad_norm": 0.7103603969004639,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 4969
    },
    {
      "epoch": 0.0497,
      "grad_norm": 0.7119697504065976,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 4970
    },
    {
      "epoch": 0.04971,
      "grad_norm": 0.6691773043773404,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4971
    },
    {
      "epoch": 0.04972,
      "grad_norm": 0.7584471656947329,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 4972
    },
    {
      "epoch": 0.04973,
      "grad_norm": 0.7074430853606097,
      "learning_rate": 0.003,
      "loss": 4.1896,
      "step": 4973
    },
    {
      "epoch": 0.04974,
      "grad_norm": 0.5713006659555607,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 4974
    },
    {
      "epoch": 0.04975,
      "grad_norm": 0.5130707295349296,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 4975
    },
    {
      "epoch": 0.04976,
      "grad_norm": 0.476744226744805,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4976
    },
    {
      "epoch": 0.04977,
      "grad_norm": 0.436532472120347,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4977
    },
    {
      "epoch": 0.04978,
      "grad_norm": 0.4481149086483841,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 4978
    },
    {
      "epoch": 0.04979,
      "grad_norm": 0.4658196526981218,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4979
    },
    {
      "epoch": 0.0498,
      "grad_norm": 0.48184153753418996,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 4980
    },
    {
      "epoch": 0.04981,
      "grad_norm": 0.5711200331385257,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 4981
    },
    {
      "epoch": 0.04982,
      "grad_norm": 0.7811982054881438,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 4982
    },
    {
      "epoch": 0.04983,
      "grad_norm": 0.9119195104356123,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 4983
    },
    {
      "epoch": 0.04984,
      "grad_norm": 0.809114284515937,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 4984
    },
    {
      "epoch": 0.04985,
      "grad_norm": 0.6537212838716787,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 4985
    },
    {
      "epoch": 0.04986,
      "grad_norm": 0.6869403529206495,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 4986
    },
    {
      "epoch": 0.04987,
      "grad_norm": 0.613841094604171,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 4987
    },
    {
      "epoch": 0.04988,
      "grad_norm": 0.6618484956538617,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4988
    },
    {
      "epoch": 0.04989,
      "grad_norm": 0.5936176795447469,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 4989
    },
    {
      "epoch": 0.0499,
      "grad_norm": 0.5392404418710565,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 4990
    },
    {
      "epoch": 0.04991,
      "grad_norm": 0.621246461554044,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 4991
    },
    {
      "epoch": 0.04992,
      "grad_norm": 0.670783622770409,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 4992
    },
    {
      "epoch": 0.04993,
      "grad_norm": 0.6045345128052574,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 4993
    },
    {
      "epoch": 0.04994,
      "grad_norm": 0.6101486114599416,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 4994
    },
    {
      "epoch": 0.04995,
      "grad_norm": 0.55705863270894,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 4995
    },
    {
      "epoch": 0.04996,
      "grad_norm": 0.44788775188911484,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 4996
    },
    {
      "epoch": 0.04997,
      "grad_norm": 0.4309530448591752,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 4997
    },
    {
      "epoch": 0.04998,
      "grad_norm": 0.428484725286708,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 4998
    },
    {
      "epoch": 0.04999,
      "grad_norm": 0.41683300725155314,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 4999
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3880481560630285,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 5000
    },
    {
      "epoch": 0.05001,
      "grad_norm": 0.44002892670528426,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 5001
    },
    {
      "epoch": 0.05002,
      "grad_norm": 0.5267413563943687,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 5002
    },
    {
      "epoch": 0.05003,
      "grad_norm": 0.6974673629709173,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 5003
    },
    {
      "epoch": 0.05004,
      "grad_norm": 0.858937261820554,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 5004
    },
    {
      "epoch": 0.05005,
      "grad_norm": 0.9141406787634108,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 5005
    },
    {
      "epoch": 0.05006,
      "grad_norm": 0.687864263963555,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 5006
    },
    {
      "epoch": 0.05007,
      "grad_norm": 0.5773822019308112,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 5007
    },
    {
      "epoch": 0.05008,
      "grad_norm": 0.5547957033613983,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 5008
    },
    {
      "epoch": 0.05009,
      "grad_norm": 0.5411686344353427,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 5009
    },
    {
      "epoch": 0.0501,
      "grad_norm": 0.489119317455081,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 5010
    },
    {
      "epoch": 0.05011,
      "grad_norm": 0.464838598175237,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 5011
    },
    {
      "epoch": 0.05012,
      "grad_norm": 0.4580943694467521,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 5012
    },
    {
      "epoch": 0.05013,
      "grad_norm": 0.4808521248590666,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 5013
    },
    {
      "epoch": 0.05014,
      "grad_norm": 0.5229887414294506,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 5014
    },
    {
      "epoch": 0.05015,
      "grad_norm": 0.5777151908604863,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 5015
    },
    {
      "epoch": 0.05016,
      "grad_norm": 0.5235213366994899,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 5016
    },
    {
      "epoch": 0.05017,
      "grad_norm": 0.445246255748685,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 5017
    },
    {
      "epoch": 0.05018,
      "grad_norm": 0.45246478410906943,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 5018
    },
    {
      "epoch": 0.05019,
      "grad_norm": 0.5231762403732283,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 5019
    },
    {
      "epoch": 0.0502,
      "grad_norm": 0.6594471849420648,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 5020
    },
    {
      "epoch": 0.05021,
      "grad_norm": 0.6904442298729323,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 5021
    },
    {
      "epoch": 0.05022,
      "grad_norm": 0.5909951697887482,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 5022
    },
    {
      "epoch": 0.05023,
      "grad_norm": 0.577685062094026,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 5023
    },
    {
      "epoch": 0.05024,
      "grad_norm": 0.5728157983655138,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 5024
    },
    {
      "epoch": 0.05025,
      "grad_norm": 0.5379702885678885,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 5025
    },
    {
      "epoch": 0.05026,
      "grad_norm": 0.6155371625202312,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 5026
    },
    {
      "epoch": 0.05027,
      "grad_norm": 0.7519388420060837,
      "learning_rate": 0.003,
      "loss": 4.1457,
      "step": 5027
    },
    {
      "epoch": 0.05028,
      "grad_norm": 0.9311839881105047,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 5028
    },
    {
      "epoch": 0.05029,
      "grad_norm": 0.9260888417245792,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 5029
    },
    {
      "epoch": 0.0503,
      "grad_norm": 0.7728492641666489,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 5030
    },
    {
      "epoch": 0.05031,
      "grad_norm": 0.777930269535098,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 5031
    },
    {
      "epoch": 0.05032,
      "grad_norm": 0.7350784470721563,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 5032
    },
    {
      "epoch": 0.05033,
      "grad_norm": 0.5291553968829333,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 5033
    },
    {
      "epoch": 0.05034,
      "grad_norm": 0.4919036368843819,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 5034
    },
    {
      "epoch": 0.05035,
      "grad_norm": 0.4975664864391519,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 5035
    },
    {
      "epoch": 0.05036,
      "grad_norm": 0.5163959831953332,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 5036
    },
    {
      "epoch": 0.05037,
      "grad_norm": 0.5569175575619792,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 5037
    },
    {
      "epoch": 0.05038,
      "grad_norm": 0.5227477080467972,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 5038
    },
    {
      "epoch": 0.05039,
      "grad_norm": 0.5597860102419677,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 5039
    },
    {
      "epoch": 0.0504,
      "grad_norm": 0.6074226279213635,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 5040
    },
    {
      "epoch": 0.05041,
      "grad_norm": 0.6238557653084295,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 5041
    },
    {
      "epoch": 0.05042,
      "grad_norm": 0.6153407136204087,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 5042
    },
    {
      "epoch": 0.05043,
      "grad_norm": 0.6514614471239298,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 5043
    },
    {
      "epoch": 0.05044,
      "grad_norm": 0.6930287002961413,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 5044
    },
    {
      "epoch": 0.05045,
      "grad_norm": 0.8304592547990185,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 5045
    },
    {
      "epoch": 0.05046,
      "grad_norm": 0.8026034540044529,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 5046
    },
    {
      "epoch": 0.05047,
      "grad_norm": 0.9245681532952602,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 5047
    },
    {
      "epoch": 0.05048,
      "grad_norm": 0.8960724676621618,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 5048
    },
    {
      "epoch": 0.05049,
      "grad_norm": 1.0151937627560343,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 5049
    },
    {
      "epoch": 0.0505,
      "grad_norm": 0.884162859927295,
      "learning_rate": 0.003,
      "loss": 4.2198,
      "step": 5050
    },
    {
      "epoch": 0.05051,
      "grad_norm": 0.8307433200396237,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 5051
    },
    {
      "epoch": 0.05052,
      "grad_norm": 0.7384842172109713,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 5052
    },
    {
      "epoch": 0.05053,
      "grad_norm": 0.7203012020157923,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 5053
    },
    {
      "epoch": 0.05054,
      "grad_norm": 0.7327188596143545,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 5054
    },
    {
      "epoch": 0.05055,
      "grad_norm": 0.7559923263710384,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 5055
    },
    {
      "epoch": 0.05056,
      "grad_norm": 0.6831499776250408,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 5056
    },
    {
      "epoch": 0.05057,
      "grad_norm": 0.6878997827607911,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 5057
    },
    {
      "epoch": 0.05058,
      "grad_norm": 0.6963511057251066,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 5058
    },
    {
      "epoch": 0.05059,
      "grad_norm": 0.7218382309881082,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 5059
    },
    {
      "epoch": 0.0506,
      "grad_norm": 0.7641404073482049,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 5060
    },
    {
      "epoch": 0.05061,
      "grad_norm": 0.7492054114953671,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 5061
    },
    {
      "epoch": 0.05062,
      "grad_norm": 0.580002983697803,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 5062
    },
    {
      "epoch": 0.05063,
      "grad_norm": 0.465857319923329,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 5063
    },
    {
      "epoch": 0.05064,
      "grad_norm": 0.4639475584360793,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 5064
    },
    {
      "epoch": 0.05065,
      "grad_norm": 0.4316878203187786,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 5065
    },
    {
      "epoch": 0.05066,
      "grad_norm": 0.4129716036113381,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 5066
    },
    {
      "epoch": 0.05067,
      "grad_norm": 0.3545337277972109,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 5067
    },
    {
      "epoch": 0.05068,
      "grad_norm": 0.35000428560320695,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 5068
    },
    {
      "epoch": 0.05069,
      "grad_norm": 0.28475691607233394,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 5069
    },
    {
      "epoch": 0.0507,
      "grad_norm": 0.3211180881977199,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 5070
    },
    {
      "epoch": 0.05071,
      "grad_norm": 0.31520195677598756,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 5071
    },
    {
      "epoch": 0.05072,
      "grad_norm": 0.33671460211119836,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 5072
    },
    {
      "epoch": 0.05073,
      "grad_norm": 0.33769028368475607,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 5073
    },
    {
      "epoch": 0.05074,
      "grad_norm": 0.3448787140101478,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 5074
    },
    {
      "epoch": 0.05075,
      "grad_norm": 0.3957609850918836,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 5075
    },
    {
      "epoch": 0.05076,
      "grad_norm": 0.48241503749103687,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 5076
    },
    {
      "epoch": 0.05077,
      "grad_norm": 0.6132233951932243,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 5077
    },
    {
      "epoch": 0.05078,
      "grad_norm": 0.7745174445055403,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 5078
    },
    {
      "epoch": 0.05079,
      "grad_norm": 0.8279381574948442,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 5079
    },
    {
      "epoch": 0.0508,
      "grad_norm": 0.8263557958238591,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 5080
    },
    {
      "epoch": 0.05081,
      "grad_norm": 0.8277412011902472,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 5081
    },
    {
      "epoch": 0.05082,
      "grad_norm": 0.7915182287182168,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 5082
    },
    {
      "epoch": 0.05083,
      "grad_norm": 0.7357570092313975,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 5083
    },
    {
      "epoch": 0.05084,
      "grad_norm": 0.9633741172415596,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 5084
    },
    {
      "epoch": 0.05085,
      "grad_norm": 1.0990777684274788,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 5085
    },
    {
      "epoch": 0.05086,
      "grad_norm": 0.7846549440538764,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5086
    },
    {
      "epoch": 0.05087,
      "grad_norm": 0.6846851816877524,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 5087
    },
    {
      "epoch": 0.05088,
      "grad_norm": 0.6318100802689848,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 5088
    },
    {
      "epoch": 0.05089,
      "grad_norm": 0.6520703017626633,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 5089
    },
    {
      "epoch": 0.0509,
      "grad_norm": 0.5565734582358871,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 5090
    },
    {
      "epoch": 0.05091,
      "grad_norm": 0.5349654586172464,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 5091
    },
    {
      "epoch": 0.05092,
      "grad_norm": 0.5297741402897067,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 5092
    },
    {
      "epoch": 0.05093,
      "grad_norm": 0.5575684104199916,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 5093
    },
    {
      "epoch": 0.05094,
      "grad_norm": 0.5605888909754495,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 5094
    },
    {
      "epoch": 0.05095,
      "grad_norm": 0.5343018695121468,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 5095
    },
    {
      "epoch": 0.05096,
      "grad_norm": 0.43949601788526343,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 5096
    },
    {
      "epoch": 0.05097,
      "grad_norm": 0.4810979086570852,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 5097
    },
    {
      "epoch": 0.05098,
      "grad_norm": 0.4654228949280776,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 5098
    },
    {
      "epoch": 0.05099,
      "grad_norm": 0.40501669029112214,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 5099
    },
    {
      "epoch": 0.051,
      "grad_norm": 0.4223377546263004,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 5100
    },
    {
      "epoch": 0.05101,
      "grad_norm": 0.44525788101780217,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 5101
    },
    {
      "epoch": 0.05102,
      "grad_norm": 0.4871872595505277,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 5102
    },
    {
      "epoch": 0.05103,
      "grad_norm": 0.5447347242656629,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 5103
    },
    {
      "epoch": 0.05104,
      "grad_norm": 0.5964115151305719,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 5104
    },
    {
      "epoch": 0.05105,
      "grad_norm": 0.5867716136824593,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 5105
    },
    {
      "epoch": 0.05106,
      "grad_norm": 0.5562979835891493,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 5106
    },
    {
      "epoch": 0.05107,
      "grad_norm": 0.5181396244592503,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 5107
    },
    {
      "epoch": 0.05108,
      "grad_norm": 0.5440116113449289,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 5108
    },
    {
      "epoch": 0.05109,
      "grad_norm": 0.5961075613550932,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 5109
    },
    {
      "epoch": 0.0511,
      "grad_norm": 0.533390436132316,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 5110
    },
    {
      "epoch": 0.05111,
      "grad_norm": 0.48024251864431305,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 5111
    },
    {
      "epoch": 0.05112,
      "grad_norm": 0.4598495045627523,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 5112
    },
    {
      "epoch": 0.05113,
      "grad_norm": 0.5112726037297928,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 5113
    },
    {
      "epoch": 0.05114,
      "grad_norm": 0.6162499535834111,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 5114
    },
    {
      "epoch": 0.05115,
      "grad_norm": 0.6330435450833568,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5115
    },
    {
      "epoch": 0.05116,
      "grad_norm": 0.7922157375423148,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 5116
    },
    {
      "epoch": 0.05117,
      "grad_norm": 0.8777485197768994,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5117
    },
    {
      "epoch": 0.05118,
      "grad_norm": 0.8487101677130807,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 5118
    },
    {
      "epoch": 0.05119,
      "grad_norm": 0.6672995585854586,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 5119
    },
    {
      "epoch": 0.0512,
      "grad_norm": 0.5347949292529923,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 5120
    },
    {
      "epoch": 0.05121,
      "grad_norm": 0.47651085513877506,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 5121
    },
    {
      "epoch": 0.05122,
      "grad_norm": 0.5586400651301906,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 5122
    },
    {
      "epoch": 0.05123,
      "grad_norm": 0.6300817408188087,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 5123
    },
    {
      "epoch": 0.05124,
      "grad_norm": 0.6701205648572075,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 5124
    },
    {
      "epoch": 0.05125,
      "grad_norm": 0.6468239470621561,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 5125
    },
    {
      "epoch": 0.05126,
      "grad_norm": 0.5445267421148194,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 5126
    },
    {
      "epoch": 0.05127,
      "grad_norm": 0.4799763707457368,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 5127
    },
    {
      "epoch": 0.05128,
      "grad_norm": 0.5803016836080052,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 5128
    },
    {
      "epoch": 0.05129,
      "grad_norm": 0.678952357079139,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 5129
    },
    {
      "epoch": 0.0513,
      "grad_norm": 0.8039176919182028,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 5130
    },
    {
      "epoch": 0.05131,
      "grad_norm": 0.8445292736195356,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 5131
    },
    {
      "epoch": 0.05132,
      "grad_norm": 0.8088539903803569,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 5132
    },
    {
      "epoch": 0.05133,
      "grad_norm": 0.8452742411739833,
      "learning_rate": 0.003,
      "loss": 4.1457,
      "step": 5133
    },
    {
      "epoch": 0.05134,
      "grad_norm": 0.7211547370019534,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 5134
    },
    {
      "epoch": 0.05135,
      "grad_norm": 0.7552143125752953,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 5135
    },
    {
      "epoch": 0.05136,
      "grad_norm": 0.8178560583882517,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 5136
    },
    {
      "epoch": 0.05137,
      "grad_norm": 0.7937255401431907,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 5137
    },
    {
      "epoch": 0.05138,
      "grad_norm": 0.7029385951002983,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 5138
    },
    {
      "epoch": 0.05139,
      "grad_norm": 0.755399264561879,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 5139
    },
    {
      "epoch": 0.0514,
      "grad_norm": 0.6372525763339177,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 5140
    },
    {
      "epoch": 0.05141,
      "grad_norm": 0.6168576930336179,
      "learning_rate": 0.003,
      "loss": 4.1854,
      "step": 5141
    },
    {
      "epoch": 0.05142,
      "grad_norm": 0.6036281974051024,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 5142
    },
    {
      "epoch": 0.05143,
      "grad_norm": 0.5036860750600824,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 5143
    },
    {
      "epoch": 0.05144,
      "grad_norm": 0.46857974439093014,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 5144
    },
    {
      "epoch": 0.05145,
      "grad_norm": 0.42494497123142294,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 5145
    },
    {
      "epoch": 0.05146,
      "grad_norm": 0.44380638929392596,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 5146
    },
    {
      "epoch": 0.05147,
      "grad_norm": 0.40327298093540914,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 5147
    },
    {
      "epoch": 0.05148,
      "grad_norm": 0.3953841068063316,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 5148
    },
    {
      "epoch": 0.05149,
      "grad_norm": 0.4012036518868883,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 5149
    },
    {
      "epoch": 0.0515,
      "grad_norm": 0.39949323869855685,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 5150
    },
    {
      "epoch": 0.05151,
      "grad_norm": 0.3876338989691726,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 5151
    },
    {
      "epoch": 0.05152,
      "grad_norm": 0.43621474925166587,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 5152
    },
    {
      "epoch": 0.05153,
      "grad_norm": 0.5301350008384232,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 5153
    },
    {
      "epoch": 0.05154,
      "grad_norm": 0.7035426773318153,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 5154
    },
    {
      "epoch": 0.05155,
      "grad_norm": 0.9094366977526885,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 5155
    },
    {
      "epoch": 0.05156,
      "grad_norm": 0.8124683719101861,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 5156
    },
    {
      "epoch": 0.05157,
      "grad_norm": 0.6963497652538516,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 5157
    },
    {
      "epoch": 0.05158,
      "grad_norm": 0.7426530943881816,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 5158
    },
    {
      "epoch": 0.05159,
      "grad_norm": 0.6705321816606814,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 5159
    },
    {
      "epoch": 0.0516,
      "grad_norm": 0.6026791060312885,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 5160
    },
    {
      "epoch": 0.05161,
      "grad_norm": 0.6460386669653281,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 5161
    },
    {
      "epoch": 0.05162,
      "grad_norm": 0.7274049490439595,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 5162
    },
    {
      "epoch": 0.05163,
      "grad_norm": 0.6827610438042071,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 5163
    },
    {
      "epoch": 0.05164,
      "grad_norm": 0.6633511077303998,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 5164
    },
    {
      "epoch": 0.05165,
      "grad_norm": 0.8069016890146392,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 5165
    },
    {
      "epoch": 0.05166,
      "grad_norm": 0.7564185954506445,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 5166
    },
    {
      "epoch": 0.05167,
      "grad_norm": 0.6729290579348203,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 5167
    },
    {
      "epoch": 0.05168,
      "grad_norm": 0.681973653519437,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 5168
    },
    {
      "epoch": 0.05169,
      "grad_norm": 0.6894995787523694,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 5169
    },
    {
      "epoch": 0.0517,
      "grad_norm": 0.6251406068036489,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 5170
    },
    {
      "epoch": 0.05171,
      "grad_norm": 0.6022206459158862,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 5171
    },
    {
      "epoch": 0.05172,
      "grad_norm": 0.5547021446284994,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 5172
    },
    {
      "epoch": 0.05173,
      "grad_norm": 0.5511637086768453,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 5173
    },
    {
      "epoch": 0.05174,
      "grad_norm": 0.549834258038828,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 5174
    },
    {
      "epoch": 0.05175,
      "grad_norm": 0.5143957250446763,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 5175
    },
    {
      "epoch": 0.05176,
      "grad_norm": 0.5016512485210078,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 5176
    },
    {
      "epoch": 0.05177,
      "grad_norm": 0.5010375401941776,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 5177
    },
    {
      "epoch": 0.05178,
      "grad_norm": 0.45886355153966824,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 5178
    },
    {
      "epoch": 0.05179,
      "grad_norm": 0.4774614456980692,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 5179
    },
    {
      "epoch": 0.0518,
      "grad_norm": 0.5343856749000508,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 5180
    },
    {
      "epoch": 0.05181,
      "grad_norm": 0.5847156904354145,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 5181
    },
    {
      "epoch": 0.05182,
      "grad_norm": 0.7281346718853275,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 5182
    },
    {
      "epoch": 0.05183,
      "grad_norm": 0.7959120436672125,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 5183
    },
    {
      "epoch": 0.05184,
      "grad_norm": 0.7708701874016135,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 5184
    },
    {
      "epoch": 0.05185,
      "grad_norm": 0.7830980202404018,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 5185
    },
    {
      "epoch": 0.05186,
      "grad_norm": 0.6793457202900599,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 5186
    },
    {
      "epoch": 0.05187,
      "grad_norm": 0.6855736266075582,
      "learning_rate": 0.003,
      "loss": 4.171,
      "step": 5187
    },
    {
      "epoch": 0.05188,
      "grad_norm": 0.760859985233772,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 5188
    },
    {
      "epoch": 0.05189,
      "grad_norm": 0.7264115896262829,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 5189
    },
    {
      "epoch": 0.0519,
      "grad_norm": 0.7175290962540748,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 5190
    },
    {
      "epoch": 0.05191,
      "grad_norm": 0.6496658085565145,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 5191
    },
    {
      "epoch": 0.05192,
      "grad_norm": 0.6267055244498101,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 5192
    },
    {
      "epoch": 0.05193,
      "grad_norm": 0.5727047155848178,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 5193
    },
    {
      "epoch": 0.05194,
      "grad_norm": 0.5491895800888712,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 5194
    },
    {
      "epoch": 0.05195,
      "grad_norm": 0.5312113948934261,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 5195
    },
    {
      "epoch": 0.05196,
      "grad_norm": 0.5838275239944962,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 5196
    },
    {
      "epoch": 0.05197,
      "grad_norm": 0.5319117287619517,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 5197
    },
    {
      "epoch": 0.05198,
      "grad_norm": 0.474231170755562,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 5198
    },
    {
      "epoch": 0.05199,
      "grad_norm": 0.5339097503934678,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 5199
    },
    {
      "epoch": 0.052,
      "grad_norm": 0.552513644020886,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 5200
    },
    {
      "epoch": 0.05201,
      "grad_norm": 0.5499912684018807,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 5201
    },
    {
      "epoch": 0.05202,
      "grad_norm": 0.5884166479851186,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 5202
    },
    {
      "epoch": 0.05203,
      "grad_norm": 0.6234832076161475,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 5203
    },
    {
      "epoch": 0.05204,
      "grad_norm": 0.6370950178244867,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 5204
    },
    {
      "epoch": 0.05205,
      "grad_norm": 0.7268995146225012,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 5205
    },
    {
      "epoch": 0.05206,
      "grad_norm": 0.8429209951426253,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 5206
    },
    {
      "epoch": 0.05207,
      "grad_norm": 0.891067417785423,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 5207
    },
    {
      "epoch": 0.05208,
      "grad_norm": 0.880801827049288,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 5208
    },
    {
      "epoch": 0.05209,
      "grad_norm": 0.743628891207245,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 5209
    },
    {
      "epoch": 0.0521,
      "grad_norm": 0.6585381210736347,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 5210
    },
    {
      "epoch": 0.05211,
      "grad_norm": 0.6590374105413944,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 5211
    },
    {
      "epoch": 0.05212,
      "grad_norm": 0.5838327606908597,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 5212
    },
    {
      "epoch": 0.05213,
      "grad_norm": 0.5577967457265012,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 5213
    },
    {
      "epoch": 0.05214,
      "grad_norm": 0.5831733721175694,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 5214
    },
    {
      "epoch": 0.05215,
      "grad_norm": 0.5947055436999698,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 5215
    },
    {
      "epoch": 0.05216,
      "grad_norm": 0.5162752873889312,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 5216
    },
    {
      "epoch": 0.05217,
      "grad_norm": 0.5215914347147623,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 5217
    },
    {
      "epoch": 0.05218,
      "grad_norm": 0.495811759691527,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 5218
    },
    {
      "epoch": 0.05219,
      "grad_norm": 0.5616102027479555,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 5219
    },
    {
      "epoch": 0.0522,
      "grad_norm": 0.596993154601823,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 5220
    },
    {
      "epoch": 0.05221,
      "grad_norm": 0.6135656226590566,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 5221
    },
    {
      "epoch": 0.05222,
      "grad_norm": 0.6651818813368179,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 5222
    },
    {
      "epoch": 0.05223,
      "grad_norm": 0.693814397802243,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 5223
    },
    {
      "epoch": 0.05224,
      "grad_norm": 0.6753245461392549,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 5224
    },
    {
      "epoch": 0.05225,
      "grad_norm": 0.5962298095800613,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 5225
    },
    {
      "epoch": 0.05226,
      "grad_norm": 0.5681174514487727,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 5226
    },
    {
      "epoch": 0.05227,
      "grad_norm": 0.6355387480928073,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 5227
    },
    {
      "epoch": 0.05228,
      "grad_norm": 0.63070568269053,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 5228
    },
    {
      "epoch": 0.05229,
      "grad_norm": 0.6541251553608277,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 5229
    },
    {
      "epoch": 0.0523,
      "grad_norm": 0.5976720272605445,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 5230
    },
    {
      "epoch": 0.05231,
      "grad_norm": 0.5666985347278075,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 5231
    },
    {
      "epoch": 0.05232,
      "grad_norm": 0.5817937128133905,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 5232
    },
    {
      "epoch": 0.05233,
      "grad_norm": 0.5938688559238984,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 5233
    },
    {
      "epoch": 0.05234,
      "grad_norm": 0.5954347096748437,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 5234
    },
    {
      "epoch": 0.05235,
      "grad_norm": 0.5732073934324781,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 5235
    },
    {
      "epoch": 0.05236,
      "grad_norm": 0.5696284586443667,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 5236
    },
    {
      "epoch": 0.05237,
      "grad_norm": 0.5556858968355276,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 5237
    },
    {
      "epoch": 0.05238,
      "grad_norm": 0.5512655986372486,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 5238
    },
    {
      "epoch": 0.05239,
      "grad_norm": 0.6544063071682276,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 5239
    },
    {
      "epoch": 0.0524,
      "grad_norm": 0.7356500161192003,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 5240
    },
    {
      "epoch": 0.05241,
      "grad_norm": 0.88373689710533,
      "learning_rate": 0.003,
      "loss": 4.1526,
      "step": 5241
    },
    {
      "epoch": 0.05242,
      "grad_norm": 0.811345798439545,
      "learning_rate": 0.003,
      "loss": 4.1503,
      "step": 5242
    },
    {
      "epoch": 0.05243,
      "grad_norm": 0.7905544586858798,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 5243
    },
    {
      "epoch": 0.05244,
      "grad_norm": 0.7298808104661723,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 5244
    },
    {
      "epoch": 0.05245,
      "grad_norm": 0.7094573896220808,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 5245
    },
    {
      "epoch": 0.05246,
      "grad_norm": 0.7963225003446655,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 5246
    },
    {
      "epoch": 0.05247,
      "grad_norm": 0.902516557295206,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 5247
    },
    {
      "epoch": 0.05248,
      "grad_norm": 0.7878245147008873,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 5248
    },
    {
      "epoch": 0.05249,
      "grad_norm": 0.6945422313261818,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 5249
    },
    {
      "epoch": 0.0525,
      "grad_norm": 0.622441402060283,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 5250
    },
    {
      "epoch": 0.05251,
      "grad_norm": 0.5779397363926364,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 5251
    },
    {
      "epoch": 0.05252,
      "grad_norm": 0.577099244293535,
      "learning_rate": 0.003,
      "loss": 4.1569,
      "step": 5252
    },
    {
      "epoch": 0.05253,
      "grad_norm": 0.5742421525793271,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 5253
    },
    {
      "epoch": 0.05254,
      "grad_norm": 0.5856615794186555,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 5254
    },
    {
      "epoch": 0.05255,
      "grad_norm": 0.5988982637518733,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 5255
    },
    {
      "epoch": 0.05256,
      "grad_norm": 0.6516682432466444,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 5256
    },
    {
      "epoch": 0.05257,
      "grad_norm": 0.7521657888490443,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 5257
    },
    {
      "epoch": 0.05258,
      "grad_norm": 0.7752727754327018,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 5258
    },
    {
      "epoch": 0.05259,
      "grad_norm": 0.7507817009513204,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 5259
    },
    {
      "epoch": 0.0526,
      "grad_norm": 0.8283788110748729,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 5260
    },
    {
      "epoch": 0.05261,
      "grad_norm": 0.9839152508479165,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 5261
    },
    {
      "epoch": 0.05262,
      "grad_norm": 0.8181134341889587,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 5262
    },
    {
      "epoch": 0.05263,
      "grad_norm": 0.8179493296972696,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 5263
    },
    {
      "epoch": 0.05264,
      "grad_norm": 0.8507075051576984,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 5264
    },
    {
      "epoch": 0.05265,
      "grad_norm": 0.7875040331282762,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 5265
    },
    {
      "epoch": 0.05266,
      "grad_norm": 0.7660878019896183,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 5266
    },
    {
      "epoch": 0.05267,
      "grad_norm": 0.7381530659889034,
      "learning_rate": 0.003,
      "loss": 4.1753,
      "step": 5267
    },
    {
      "epoch": 0.05268,
      "grad_norm": 0.7516718782373394,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 5268
    },
    {
      "epoch": 0.05269,
      "grad_norm": 0.7754342018650271,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 5269
    },
    {
      "epoch": 0.0527,
      "grad_norm": 0.7814902853941849,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 5270
    },
    {
      "epoch": 0.05271,
      "grad_norm": 0.7175519140149296,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 5271
    },
    {
      "epoch": 0.05272,
      "grad_norm": 0.6180907761221967,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 5272
    },
    {
      "epoch": 0.05273,
      "grad_norm": 0.5866907243765208,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 5273
    },
    {
      "epoch": 0.05274,
      "grad_norm": 0.7212098565817949,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 5274
    },
    {
      "epoch": 0.05275,
      "grad_norm": 0.745726351447074,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 5275
    },
    {
      "epoch": 0.05276,
      "grad_norm": 0.6574672838634651,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 5276
    },
    {
      "epoch": 0.05277,
      "grad_norm": 0.6169430231237321,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 5277
    },
    {
      "epoch": 0.05278,
      "grad_norm": 0.6357928590591313,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 5278
    },
    {
      "epoch": 0.05279,
      "grad_norm": 0.6803536948880607,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 5279
    },
    {
      "epoch": 0.0528,
      "grad_norm": 0.6367797725535698,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 5280
    },
    {
      "epoch": 0.05281,
      "grad_norm": 0.5426507335239904,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 5281
    },
    {
      "epoch": 0.05282,
      "grad_norm": 0.47424037246159206,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 5282
    },
    {
      "epoch": 0.05283,
      "grad_norm": 0.3930494949754586,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 5283
    },
    {
      "epoch": 0.05284,
      "grad_norm": 0.3934936059529288,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 5284
    },
    {
      "epoch": 0.05285,
      "grad_norm": 0.4015179406110254,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 5285
    },
    {
      "epoch": 0.05286,
      "grad_norm": 0.4480587464365292,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 5286
    },
    {
      "epoch": 0.05287,
      "grad_norm": 0.5070013234550016,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 5287
    },
    {
      "epoch": 0.05288,
      "grad_norm": 0.4813637426002363,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 5288
    },
    {
      "epoch": 0.05289,
      "grad_norm": 0.4573299309440364,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 5289
    },
    {
      "epoch": 0.0529,
      "grad_norm": 0.4657267342352843,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 5290
    },
    {
      "epoch": 0.05291,
      "grad_norm": 0.44753235119040924,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 5291
    },
    {
      "epoch": 0.05292,
      "grad_norm": 0.46480492942670476,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 5292
    },
    {
      "epoch": 0.05293,
      "grad_norm": 0.45730668556713944,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 5293
    },
    {
      "epoch": 0.05294,
      "grad_norm": 0.4237029704444932,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 5294
    },
    {
      "epoch": 0.05295,
      "grad_norm": 0.4482914371249565,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 5295
    },
    {
      "epoch": 0.05296,
      "grad_norm": 0.5493021494866288,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 5296
    },
    {
      "epoch": 0.05297,
      "grad_norm": 0.8031816093672703,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 5297
    },
    {
      "epoch": 0.05298,
      "grad_norm": 1.0196685307893005,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 5298
    },
    {
      "epoch": 0.05299,
      "grad_norm": 0.9550231337660582,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 5299
    },
    {
      "epoch": 0.053,
      "grad_norm": 0.8307857387808788,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 5300
    },
    {
      "epoch": 0.05301,
      "grad_norm": 0.7866950778572048,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 5301
    },
    {
      "epoch": 0.05302,
      "grad_norm": 0.6761619636450853,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 5302
    },
    {
      "epoch": 0.05303,
      "grad_norm": 0.665701232744064,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 5303
    },
    {
      "epoch": 0.05304,
      "grad_norm": 0.6461488983224262,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 5304
    },
    {
      "epoch": 0.05305,
      "grad_norm": 0.5665397238554841,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 5305
    },
    {
      "epoch": 0.05306,
      "grad_norm": 0.5309591176214379,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 5306
    },
    {
      "epoch": 0.05307,
      "grad_norm": 0.527472903799991,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 5307
    },
    {
      "epoch": 0.05308,
      "grad_norm": 0.44640273254720836,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 5308
    },
    {
      "epoch": 0.05309,
      "grad_norm": 0.4752479060075969,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 5309
    },
    {
      "epoch": 0.0531,
      "grad_norm": 0.4497310787484318,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 5310
    },
    {
      "epoch": 0.05311,
      "grad_norm": 0.47563156496846587,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 5311
    },
    {
      "epoch": 0.05312,
      "grad_norm": 0.508319969010842,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 5312
    },
    {
      "epoch": 0.05313,
      "grad_norm": 0.5258317632514994,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 5313
    },
    {
      "epoch": 0.05314,
      "grad_norm": 0.5419446505340356,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 5314
    },
    {
      "epoch": 0.05315,
      "grad_norm": 0.5617206811761004,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 5315
    },
    {
      "epoch": 0.05316,
      "grad_norm": 0.6150928093152392,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 5316
    },
    {
      "epoch": 0.05317,
      "grad_norm": 0.6470618227813498,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 5317
    },
    {
      "epoch": 0.05318,
      "grad_norm": 0.6497630105833105,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 5318
    },
    {
      "epoch": 0.05319,
      "grad_norm": 0.6589138797229622,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 5319
    },
    {
      "epoch": 0.0532,
      "grad_norm": 0.7898491657801758,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 5320
    },
    {
      "epoch": 0.05321,
      "grad_norm": 0.8933026805238407,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 5321
    },
    {
      "epoch": 0.05322,
      "grad_norm": 1.0398114885738736,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 5322
    },
    {
      "epoch": 0.05323,
      "grad_norm": 0.9043069047668215,
      "learning_rate": 0.003,
      "loss": 4.192,
      "step": 5323
    },
    {
      "epoch": 0.05324,
      "grad_norm": 0.7778016743549974,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 5324
    },
    {
      "epoch": 0.05325,
      "grad_norm": 0.6980624923718645,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5325
    },
    {
      "epoch": 0.05326,
      "grad_norm": 0.6326121552557883,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 5326
    },
    {
      "epoch": 0.05327,
      "grad_norm": 0.5930605167320887,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 5327
    },
    {
      "epoch": 0.05328,
      "grad_norm": 0.5642732817761132,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 5328
    },
    {
      "epoch": 0.05329,
      "grad_norm": 0.6305127051813995,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 5329
    },
    {
      "epoch": 0.0533,
      "grad_norm": 0.7052704472572536,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 5330
    },
    {
      "epoch": 0.05331,
      "grad_norm": 0.7552985640042235,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 5331
    },
    {
      "epoch": 0.05332,
      "grad_norm": 0.6810355156436282,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 5332
    },
    {
      "epoch": 0.05333,
      "grad_norm": 0.6053097534106456,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 5333
    },
    {
      "epoch": 0.05334,
      "grad_norm": 0.8017924702361167,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 5334
    },
    {
      "epoch": 0.05335,
      "grad_norm": 0.8242316752152679,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 5335
    },
    {
      "epoch": 0.05336,
      "grad_norm": 0.6320025894823955,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 5336
    },
    {
      "epoch": 0.05337,
      "grad_norm": 0.5224549291900125,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 5337
    },
    {
      "epoch": 0.05338,
      "grad_norm": 0.5296562283032598,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 5338
    },
    {
      "epoch": 0.05339,
      "grad_norm": 0.5274630915708725,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 5339
    },
    {
      "epoch": 0.0534,
      "grad_norm": 0.5537136834473111,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 5340
    },
    {
      "epoch": 0.05341,
      "grad_norm": 0.5459467303012407,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 5341
    },
    {
      "epoch": 0.05342,
      "grad_norm": 0.5994941640281616,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 5342
    },
    {
      "epoch": 0.05343,
      "grad_norm": 0.6071352996442535,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 5343
    },
    {
      "epoch": 0.05344,
      "grad_norm": 0.5345329157634405,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 5344
    },
    {
      "epoch": 0.05345,
      "grad_norm": 0.5116910157465104,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 5345
    },
    {
      "epoch": 0.05346,
      "grad_norm": 0.46468708763144895,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 5346
    },
    {
      "epoch": 0.05347,
      "grad_norm": 0.4409453577269599,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 5347
    },
    {
      "epoch": 0.05348,
      "grad_norm": 0.4414471535354094,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 5348
    },
    {
      "epoch": 0.05349,
      "grad_norm": 0.4595368190725746,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 5349
    },
    {
      "epoch": 0.0535,
      "grad_norm": 0.4867635564262319,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 5350
    },
    {
      "epoch": 0.05351,
      "grad_norm": 0.5386138190961993,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 5351
    },
    {
      "epoch": 0.05352,
      "grad_norm": 0.6349645099056564,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 5352
    },
    {
      "epoch": 0.05353,
      "grad_norm": 0.7163335707083324,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 5353
    },
    {
      "epoch": 0.05354,
      "grad_norm": 0.7284832017243276,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 5354
    },
    {
      "epoch": 0.05355,
      "grad_norm": 0.6051924935759244,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 5355
    },
    {
      "epoch": 0.05356,
      "grad_norm": 0.524816428418513,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 5356
    },
    {
      "epoch": 0.05357,
      "grad_norm": 0.5816829726245226,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 5357
    },
    {
      "epoch": 0.05358,
      "grad_norm": 0.5634033510447515,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 5358
    },
    {
      "epoch": 0.05359,
      "grad_norm": 0.6119701681573797,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 5359
    },
    {
      "epoch": 0.0536,
      "grad_norm": 0.7787011760550901,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 5360
    },
    {
      "epoch": 0.05361,
      "grad_norm": 0.8465182748674693,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 5361
    },
    {
      "epoch": 0.05362,
      "grad_norm": 0.8841512568835068,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 5362
    },
    {
      "epoch": 0.05363,
      "grad_norm": 0.7654054247828137,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 5363
    },
    {
      "epoch": 0.05364,
      "grad_norm": 0.6371159121774179,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 5364
    },
    {
      "epoch": 0.05365,
      "grad_norm": 0.6825354254162556,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 5365
    },
    {
      "epoch": 0.05366,
      "grad_norm": 0.8101026860948977,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 5366
    },
    {
      "epoch": 0.05367,
      "grad_norm": 0.8291532467226016,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 5367
    },
    {
      "epoch": 0.05368,
      "grad_norm": 0.8056447950741361,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 5368
    },
    {
      "epoch": 0.05369,
      "grad_norm": 0.8735378267002016,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 5369
    },
    {
      "epoch": 0.0537,
      "grad_norm": 0.9391702508573958,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 5370
    },
    {
      "epoch": 0.05371,
      "grad_norm": 0.8524226126886991,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 5371
    },
    {
      "epoch": 0.05372,
      "grad_norm": 0.803705266443803,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 5372
    },
    {
      "epoch": 0.05373,
      "grad_norm": 0.7560843151563093,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 5373
    },
    {
      "epoch": 0.05374,
      "grad_norm": 0.6930845720174758,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 5374
    },
    {
      "epoch": 0.05375,
      "grad_norm": 0.6546702545023289,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 5375
    },
    {
      "epoch": 0.05376,
      "grad_norm": 0.6000524634204305,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 5376
    },
    {
      "epoch": 0.05377,
      "grad_norm": 0.5834504831622397,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 5377
    },
    {
      "epoch": 0.05378,
      "grad_norm": 0.49712005322114705,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 5378
    },
    {
      "epoch": 0.05379,
      "grad_norm": 0.45036859975016763,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5379
    },
    {
      "epoch": 0.0538,
      "grad_norm": 0.43310401465589693,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 5380
    },
    {
      "epoch": 0.05381,
      "grad_norm": 0.43118280545438875,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 5381
    },
    {
      "epoch": 0.05382,
      "grad_norm": 0.4147511601446657,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 5382
    },
    {
      "epoch": 0.05383,
      "grad_norm": 0.41428955644326865,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 5383
    },
    {
      "epoch": 0.05384,
      "grad_norm": 0.46974708785278296,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 5384
    },
    {
      "epoch": 0.05385,
      "grad_norm": 0.4804439317498876,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 5385
    },
    {
      "epoch": 0.05386,
      "grad_norm": 0.4710597747403432,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 5386
    },
    {
      "epoch": 0.05387,
      "grad_norm": 0.5345471173023906,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 5387
    },
    {
      "epoch": 0.05388,
      "grad_norm": 0.527734554101038,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 5388
    },
    {
      "epoch": 0.05389,
      "grad_norm": 0.5157286475163666,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 5389
    },
    {
      "epoch": 0.0539,
      "grad_norm": 0.5289054154287074,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 5390
    },
    {
      "epoch": 0.05391,
      "grad_norm": 0.6349600394290038,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 5391
    },
    {
      "epoch": 0.05392,
      "grad_norm": 0.7287707209258841,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 5392
    },
    {
      "epoch": 0.05393,
      "grad_norm": 0.7525205292581727,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 5393
    },
    {
      "epoch": 0.05394,
      "grad_norm": 0.6536130075343365,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 5394
    },
    {
      "epoch": 0.05395,
      "grad_norm": 0.6426311365984038,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 5395
    },
    {
      "epoch": 0.05396,
      "grad_norm": 0.7277902979564174,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 5396
    },
    {
      "epoch": 0.05397,
      "grad_norm": 0.7417685288107195,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 5397
    },
    {
      "epoch": 0.05398,
      "grad_norm": 0.7086944915328566,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 5398
    },
    {
      "epoch": 0.05399,
      "grad_norm": 0.714762262058065,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 5399
    },
    {
      "epoch": 0.054,
      "grad_norm": 0.6803456287574434,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 5400
    },
    {
      "epoch": 0.05401,
      "grad_norm": 0.7115679618160242,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 5401
    },
    {
      "epoch": 0.05402,
      "grad_norm": 0.6178118046647656,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 5402
    },
    {
      "epoch": 0.05403,
      "grad_norm": 0.717834549171867,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 5403
    },
    {
      "epoch": 0.05404,
      "grad_norm": 0.7093324185375305,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 5404
    },
    {
      "epoch": 0.05405,
      "grad_norm": 0.6425829471366494,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 5405
    },
    {
      "epoch": 0.05406,
      "grad_norm": 0.6272170060637055,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 5406
    },
    {
      "epoch": 0.05407,
      "grad_norm": 0.599920594553199,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 5407
    },
    {
      "epoch": 0.05408,
      "grad_norm": 0.5820564980718425,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 5408
    },
    {
      "epoch": 0.05409,
      "grad_norm": 0.5111686298731917,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 5409
    },
    {
      "epoch": 0.0541,
      "grad_norm": 0.462193558686699,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 5410
    },
    {
      "epoch": 0.05411,
      "grad_norm": 0.4130402153846359,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 5411
    },
    {
      "epoch": 0.05412,
      "grad_norm": 0.4056983953065427,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 5412
    },
    {
      "epoch": 0.05413,
      "grad_norm": 0.37874815302403264,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 5413
    },
    {
      "epoch": 0.05414,
      "grad_norm": 0.44844074230106556,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 5414
    },
    {
      "epoch": 0.05415,
      "grad_norm": 0.5721190859806238,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 5415
    },
    {
      "epoch": 0.05416,
      "grad_norm": 0.6739294583356293,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 5416
    },
    {
      "epoch": 0.05417,
      "grad_norm": 0.6903098191439943,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 5417
    },
    {
      "epoch": 0.05418,
      "grad_norm": 0.63417118564695,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 5418
    },
    {
      "epoch": 0.05419,
      "grad_norm": 0.7191816724039328,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 5419
    },
    {
      "epoch": 0.0542,
      "grad_norm": 0.8004777313589135,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 5420
    },
    {
      "epoch": 0.05421,
      "grad_norm": 0.896985525441762,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 5421
    },
    {
      "epoch": 0.05422,
      "grad_norm": 0.7798643056239566,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 5422
    },
    {
      "epoch": 0.05423,
      "grad_norm": 0.7070552144440728,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 5423
    },
    {
      "epoch": 0.05424,
      "grad_norm": 0.5579018401845559,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 5424
    },
    {
      "epoch": 0.05425,
      "grad_norm": 0.5286925848297326,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 5425
    },
    {
      "epoch": 0.05426,
      "grad_norm": 0.5912337560467409,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 5426
    },
    {
      "epoch": 0.05427,
      "grad_norm": 0.5997777842806946,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 5427
    },
    {
      "epoch": 0.05428,
      "grad_norm": 0.6476094192756237,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 5428
    },
    {
      "epoch": 0.05429,
      "grad_norm": 0.6086875232467213,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 5429
    },
    {
      "epoch": 0.0543,
      "grad_norm": 0.681356494329053,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 5430
    },
    {
      "epoch": 0.05431,
      "grad_norm": 0.6615347259505286,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 5431
    },
    {
      "epoch": 0.05432,
      "grad_norm": 0.6347016018100827,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 5432
    },
    {
      "epoch": 0.05433,
      "grad_norm": 0.6345480892474629,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 5433
    },
    {
      "epoch": 0.05434,
      "grad_norm": 0.7111690564402451,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 5434
    },
    {
      "epoch": 0.05435,
      "grad_norm": 0.7921322712349017,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 5435
    },
    {
      "epoch": 0.05436,
      "grad_norm": 0.7564267955476854,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 5436
    },
    {
      "epoch": 0.05437,
      "grad_norm": 0.6706007148323503,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 5437
    },
    {
      "epoch": 0.05438,
      "grad_norm": 0.6450156634525958,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 5438
    },
    {
      "epoch": 0.05439,
      "grad_norm": 0.7387880792670661,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 5439
    },
    {
      "epoch": 0.0544,
      "grad_norm": 0.8351260726939442,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5440
    },
    {
      "epoch": 0.05441,
      "grad_norm": 0.8490525937722112,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 5441
    },
    {
      "epoch": 0.05442,
      "grad_norm": 0.674030590021176,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 5442
    },
    {
      "epoch": 0.05443,
      "grad_norm": 0.6505791851985235,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 5443
    },
    {
      "epoch": 0.05444,
      "grad_norm": 0.6106003708383616,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 5444
    },
    {
      "epoch": 0.05445,
      "grad_norm": 0.5632553032978198,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 5445
    },
    {
      "epoch": 0.05446,
      "grad_norm": 0.578137023972088,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 5446
    },
    {
      "epoch": 0.05447,
      "grad_norm": 0.6233499651264109,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 5447
    },
    {
      "epoch": 0.05448,
      "grad_norm": 0.7025863451489226,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 5448
    },
    {
      "epoch": 0.05449,
      "grad_norm": 0.6674289251362188,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 5449
    },
    {
      "epoch": 0.0545,
      "grad_norm": 0.68187200764662,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 5450
    },
    {
      "epoch": 0.05451,
      "grad_norm": 0.9009771762961888,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 5451
    },
    {
      "epoch": 0.05452,
      "grad_norm": 0.9180522733634838,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 5452
    },
    {
      "epoch": 0.05453,
      "grad_norm": 0.8476324692508597,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 5453
    },
    {
      "epoch": 0.05454,
      "grad_norm": 0.7252925157254209,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 5454
    },
    {
      "epoch": 0.05455,
      "grad_norm": 0.7663075676162417,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 5455
    },
    {
      "epoch": 0.05456,
      "grad_norm": 0.8409176235376956,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 5456
    },
    {
      "epoch": 0.05457,
      "grad_norm": 0.8283166405126581,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 5457
    },
    {
      "epoch": 0.05458,
      "grad_norm": 0.7230997777624161,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 5458
    },
    {
      "epoch": 0.05459,
      "grad_norm": 0.6649973615570582,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 5459
    },
    {
      "epoch": 0.0546,
      "grad_norm": 0.528365614611949,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 5460
    },
    {
      "epoch": 0.05461,
      "grad_norm": 0.5520584167184678,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 5461
    },
    {
      "epoch": 0.05462,
      "grad_norm": 0.5391041370594502,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 5462
    },
    {
      "epoch": 0.05463,
      "grad_norm": 0.5846088271091963,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 5463
    },
    {
      "epoch": 0.05464,
      "grad_norm": 0.7597079518484714,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 5464
    },
    {
      "epoch": 0.05465,
      "grad_norm": 0.8362666805772095,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 5465
    },
    {
      "epoch": 0.05466,
      "grad_norm": 0.8383727125474473,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 5466
    },
    {
      "epoch": 0.05467,
      "grad_norm": 0.8306791527668079,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 5467
    },
    {
      "epoch": 0.05468,
      "grad_norm": 0.6902496457265671,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 5468
    },
    {
      "epoch": 0.05469,
      "grad_norm": 0.6940477329883477,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 5469
    },
    {
      "epoch": 0.0547,
      "grad_norm": 0.6500396166893858,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 5470
    },
    {
      "epoch": 0.05471,
      "grad_norm": 0.5274542761046104,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 5471
    },
    {
      "epoch": 0.05472,
      "grad_norm": 0.5294637164309897,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 5472
    },
    {
      "epoch": 0.05473,
      "grad_norm": 0.5195706474802417,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 5473
    },
    {
      "epoch": 0.05474,
      "grad_norm": 0.5283189608781912,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 5474
    },
    {
      "epoch": 0.05475,
      "grad_norm": 0.5025729117153097,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 5475
    },
    {
      "epoch": 0.05476,
      "grad_norm": 0.4403020126149031,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 5476
    },
    {
      "epoch": 0.05477,
      "grad_norm": 0.4121449611240764,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 5477
    },
    {
      "epoch": 0.05478,
      "grad_norm": 0.39409300706271555,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 5478
    },
    {
      "epoch": 0.05479,
      "grad_norm": 0.4508711810401754,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 5479
    },
    {
      "epoch": 0.0548,
      "grad_norm": 0.5644355789014077,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 5480
    },
    {
      "epoch": 0.05481,
      "grad_norm": 0.6849613721636778,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 5481
    },
    {
      "epoch": 0.05482,
      "grad_norm": 0.7328991559368986,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 5482
    },
    {
      "epoch": 0.05483,
      "grad_norm": 0.6360361089895654,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 5483
    },
    {
      "epoch": 0.05484,
      "grad_norm": 0.5276598733333204,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 5484
    },
    {
      "epoch": 0.05485,
      "grad_norm": 0.43134255974381286,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 5485
    },
    {
      "epoch": 0.05486,
      "grad_norm": 0.49995198292073406,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 5486
    },
    {
      "epoch": 0.05487,
      "grad_norm": 0.5218900481549134,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 5487
    },
    {
      "epoch": 0.05488,
      "grad_norm": 0.6346419704768723,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 5488
    },
    {
      "epoch": 0.05489,
      "grad_norm": 0.7535739958723238,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 5489
    },
    {
      "epoch": 0.0549,
      "grad_norm": 0.7585637473534005,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 5490
    },
    {
      "epoch": 0.05491,
      "grad_norm": 0.6170687734506143,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 5491
    },
    {
      "epoch": 0.05492,
      "grad_norm": 0.5792989631670724,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 5492
    },
    {
      "epoch": 0.05493,
      "grad_norm": 0.5665640509165365,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 5493
    },
    {
      "epoch": 0.05494,
      "grad_norm": 0.6724865699209109,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 5494
    },
    {
      "epoch": 0.05495,
      "grad_norm": 0.6552527317872832,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 5495
    },
    {
      "epoch": 0.05496,
      "grad_norm": 0.5714103826811914,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 5496
    },
    {
      "epoch": 0.05497,
      "grad_norm": 0.5522465075739208,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 5497
    },
    {
      "epoch": 0.05498,
      "grad_norm": 0.5012005070964024,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 5498
    },
    {
      "epoch": 0.05499,
      "grad_norm": 0.4511204649500133,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 5499
    },
    {
      "epoch": 0.055,
      "grad_norm": 0.5056201476680349,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 5500
    },
    {
      "epoch": 0.05501,
      "grad_norm": 0.5484274718087955,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 5501
    },
    {
      "epoch": 0.05502,
      "grad_norm": 0.5221083178277119,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 5502
    },
    {
      "epoch": 0.05503,
      "grad_norm": 0.54151812556379,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 5503
    },
    {
      "epoch": 0.05504,
      "grad_norm": 0.5431761618729547,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 5504
    },
    {
      "epoch": 0.05505,
      "grad_norm": 0.677049395087336,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 5505
    },
    {
      "epoch": 0.05506,
      "grad_norm": 0.9158252562242566,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 5506
    },
    {
      "epoch": 0.05507,
      "grad_norm": 1.1004119550566662,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 5507
    },
    {
      "epoch": 0.05508,
      "grad_norm": 0.7690899803645341,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 5508
    },
    {
      "epoch": 0.05509,
      "grad_norm": 0.7409889310481907,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 5509
    },
    {
      "epoch": 0.0551,
      "grad_norm": 0.8413828044451898,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 5510
    },
    {
      "epoch": 0.05511,
      "grad_norm": 0.974031304462407,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 5511
    },
    {
      "epoch": 0.05512,
      "grad_norm": 0.9261519351358342,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 5512
    },
    {
      "epoch": 0.05513,
      "grad_norm": 0.7536753305180286,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 5513
    },
    {
      "epoch": 0.05514,
      "grad_norm": 0.682629393213208,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 5514
    },
    {
      "epoch": 0.05515,
      "grad_norm": 0.6438639256486987,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 5515
    },
    {
      "epoch": 0.05516,
      "grad_norm": 0.6370628235963453,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 5516
    },
    {
      "epoch": 0.05517,
      "grad_norm": 0.6380637575916892,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 5517
    },
    {
      "epoch": 0.05518,
      "grad_norm": 0.6722144275625301,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 5518
    },
    {
      "epoch": 0.05519,
      "grad_norm": 0.6613958013501796,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 5519
    },
    {
      "epoch": 0.0552,
      "grad_norm": 0.5988204576269331,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 5520
    },
    {
      "epoch": 0.05521,
      "grad_norm": 0.6572663799653967,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 5521
    },
    {
      "epoch": 0.05522,
      "grad_norm": 0.6243740084612656,
      "learning_rate": 0.003,
      "loss": 4.1628,
      "step": 5522
    },
    {
      "epoch": 0.05523,
      "grad_norm": 0.561465995263533,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 5523
    },
    {
      "epoch": 0.05524,
      "grad_norm": 0.5486658522043749,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 5524
    },
    {
      "epoch": 0.05525,
      "grad_norm": 0.5695350014121137,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 5525
    },
    {
      "epoch": 0.05526,
      "grad_norm": 0.5346598523218599,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 5526
    },
    {
      "epoch": 0.05527,
      "grad_norm": 0.4551658263055205,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 5527
    },
    {
      "epoch": 0.05528,
      "grad_norm": 0.4843464050850517,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 5528
    },
    {
      "epoch": 0.05529,
      "grad_norm": 0.585373677911223,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 5529
    },
    {
      "epoch": 0.0553,
      "grad_norm": 0.6825180764380449,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 5530
    },
    {
      "epoch": 0.05531,
      "grad_norm": 0.8045417801010512,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 5531
    },
    {
      "epoch": 0.05532,
      "grad_norm": 0.7195866881344775,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 5532
    },
    {
      "epoch": 0.05533,
      "grad_norm": 0.6162799867320136,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 5533
    },
    {
      "epoch": 0.05534,
      "grad_norm": 0.7904779000302894,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 5534
    },
    {
      "epoch": 0.05535,
      "grad_norm": 0.8644528986049678,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 5535
    },
    {
      "epoch": 0.05536,
      "grad_norm": 0.7766441044677856,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 5536
    },
    {
      "epoch": 0.05537,
      "grad_norm": 0.7239299173431871,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 5537
    },
    {
      "epoch": 0.05538,
      "grad_norm": 0.7001732822471084,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 5538
    },
    {
      "epoch": 0.05539,
      "grad_norm": 0.6623943623113974,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 5539
    },
    {
      "epoch": 0.0554,
      "grad_norm": 0.6471550442576808,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 5540
    },
    {
      "epoch": 0.05541,
      "grad_norm": 0.5868454097526585,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 5541
    },
    {
      "epoch": 0.05542,
      "grad_norm": 0.5047287366024358,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 5542
    },
    {
      "epoch": 0.05543,
      "grad_norm": 0.47745524075484136,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 5543
    },
    {
      "epoch": 0.05544,
      "grad_norm": 0.4985694332258811,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 5544
    },
    {
      "epoch": 0.05545,
      "grad_norm": 0.5073634801765811,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 5545
    },
    {
      "epoch": 0.05546,
      "grad_norm": 0.5237047792569028,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 5546
    },
    {
      "epoch": 0.05547,
      "grad_norm": 0.5914205238128549,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 5547
    },
    {
      "epoch": 0.05548,
      "grad_norm": 0.6165645927936478,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 5548
    },
    {
      "epoch": 0.05549,
      "grad_norm": 0.5898830700152652,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 5549
    },
    {
      "epoch": 0.0555,
      "grad_norm": 0.45665075102292346,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 5550
    },
    {
      "epoch": 0.05551,
      "grad_norm": 0.41856537923447845,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 5551
    },
    {
      "epoch": 0.05552,
      "grad_norm": 0.4227890713436711,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 5552
    },
    {
      "epoch": 0.05553,
      "grad_norm": 0.40800302608685,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 5553
    },
    {
      "epoch": 0.05554,
      "grad_norm": 0.4343242066304794,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 5554
    },
    {
      "epoch": 0.05555,
      "grad_norm": 0.49222919452357977,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 5555
    },
    {
      "epoch": 0.05556,
      "grad_norm": 0.5727061854119558,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 5556
    },
    {
      "epoch": 0.05557,
      "grad_norm": 0.6430410979228637,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 5557
    },
    {
      "epoch": 0.05558,
      "grad_norm": 0.7465449755259148,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 5558
    },
    {
      "epoch": 0.05559,
      "grad_norm": 0.8571034449603531,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 5559
    },
    {
      "epoch": 0.0556,
      "grad_norm": 0.9603542514944866,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 5560
    },
    {
      "epoch": 0.05561,
      "grad_norm": 1.1384407616016659,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 5561
    },
    {
      "epoch": 0.05562,
      "grad_norm": 0.8542948215220858,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 5562
    },
    {
      "epoch": 0.05563,
      "grad_norm": 0.6440539340616278,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 5563
    },
    {
      "epoch": 0.05564,
      "grad_norm": 0.5983189844345748,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 5564
    },
    {
      "epoch": 0.05565,
      "grad_norm": 0.6928412710848559,
      "learning_rate": 0.003,
      "loss": 4.1628,
      "step": 5565
    },
    {
      "epoch": 0.05566,
      "grad_norm": 0.7458788234271931,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 5566
    },
    {
      "epoch": 0.05567,
      "grad_norm": 0.6975211809831088,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 5567
    },
    {
      "epoch": 0.05568,
      "grad_norm": 0.7797367230472155,
      "learning_rate": 0.003,
      "loss": 4.1575,
      "step": 5568
    },
    {
      "epoch": 0.05569,
      "grad_norm": 0.8587086222029046,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 5569
    },
    {
      "epoch": 0.0557,
      "grad_norm": 0.842041289187777,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 5570
    },
    {
      "epoch": 0.05571,
      "grad_norm": 0.7876741380103197,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 5571
    },
    {
      "epoch": 0.05572,
      "grad_norm": 0.6444705359029274,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 5572
    },
    {
      "epoch": 0.05573,
      "grad_norm": 0.549050192063969,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 5573
    },
    {
      "epoch": 0.05574,
      "grad_norm": 0.5333806190349942,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 5574
    },
    {
      "epoch": 0.05575,
      "grad_norm": 0.5225404420588966,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 5575
    },
    {
      "epoch": 0.05576,
      "grad_norm": 0.5429894944172927,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 5576
    },
    {
      "epoch": 0.05577,
      "grad_norm": 0.5490427675347196,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 5577
    },
    {
      "epoch": 0.05578,
      "grad_norm": 0.6083016266092716,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 5578
    },
    {
      "epoch": 0.05579,
      "grad_norm": 0.6186932276076856,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 5579
    },
    {
      "epoch": 0.0558,
      "grad_norm": 0.5780512131591367,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 5580
    },
    {
      "epoch": 0.05581,
      "grad_norm": 0.5980239436138541,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 5581
    },
    {
      "epoch": 0.05582,
      "grad_norm": 0.5392719008352066,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 5582
    },
    {
      "epoch": 0.05583,
      "grad_norm": 0.5546273095860996,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 5583
    },
    {
      "epoch": 0.05584,
      "grad_norm": 0.6705838265732573,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 5584
    },
    {
      "epoch": 0.05585,
      "grad_norm": 0.8217396804436106,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 5585
    },
    {
      "epoch": 0.05586,
      "grad_norm": 0.9042169206281825,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 5586
    },
    {
      "epoch": 0.05587,
      "grad_norm": 0.9227619224934434,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 5587
    },
    {
      "epoch": 0.05588,
      "grad_norm": 0.7777466990352064,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 5588
    },
    {
      "epoch": 0.05589,
      "grad_norm": 0.6603780366654983,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 5589
    },
    {
      "epoch": 0.0559,
      "grad_norm": 0.6528519374382725,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 5590
    },
    {
      "epoch": 0.05591,
      "grad_norm": 0.6471996561956648,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 5591
    },
    {
      "epoch": 0.05592,
      "grad_norm": 0.6641363904608714,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 5592
    },
    {
      "epoch": 0.05593,
      "grad_norm": 0.6321306805343749,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 5593
    },
    {
      "epoch": 0.05594,
      "grad_norm": 0.5652755839521619,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 5594
    },
    {
      "epoch": 0.05595,
      "grad_norm": 0.5052820748237068,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 5595
    },
    {
      "epoch": 0.05596,
      "grad_norm": 0.47170632908810217,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 5596
    },
    {
      "epoch": 0.05597,
      "grad_norm": 0.4937678505983004,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 5597
    },
    {
      "epoch": 0.05598,
      "grad_norm": 0.5635649117022338,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 5598
    },
    {
      "epoch": 0.05599,
      "grad_norm": 0.6174104628038847,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 5599
    },
    {
      "epoch": 0.056,
      "grad_norm": 0.6539159328500083,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 5600
    },
    {
      "epoch": 0.05601,
      "grad_norm": 0.6679699951089212,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 5601
    },
    {
      "epoch": 0.05602,
      "grad_norm": 0.6546413158566504,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 5602
    },
    {
      "epoch": 0.05603,
      "grad_norm": 0.6946743527150302,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 5603
    },
    {
      "epoch": 0.05604,
      "grad_norm": 0.7552162101007626,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 5604
    },
    {
      "epoch": 0.05605,
      "grad_norm": 0.6696007064120673,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 5605
    },
    {
      "epoch": 0.05606,
      "grad_norm": 0.59562089846795,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 5606
    },
    {
      "epoch": 0.05607,
      "grad_norm": 0.6269393532345678,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 5607
    },
    {
      "epoch": 0.05608,
      "grad_norm": 0.6816772543279809,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 5608
    },
    {
      "epoch": 0.05609,
      "grad_norm": 0.6707354321971271,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 5609
    },
    {
      "epoch": 0.0561,
      "grad_norm": 0.6411808035499755,
      "learning_rate": 0.003,
      "loss": 4.1566,
      "step": 5610
    },
    {
      "epoch": 0.05611,
      "grad_norm": 0.6367039596675031,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5611
    },
    {
      "epoch": 0.05612,
      "grad_norm": 0.6159883542316671,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 5612
    },
    {
      "epoch": 0.05613,
      "grad_norm": 0.5688379331389054,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 5613
    },
    {
      "epoch": 0.05614,
      "grad_norm": 0.5623023558521582,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 5614
    },
    {
      "epoch": 0.05615,
      "grad_norm": 0.5064939834311122,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 5615
    },
    {
      "epoch": 0.05616,
      "grad_norm": 0.5555368115916227,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 5616
    },
    {
      "epoch": 0.05617,
      "grad_norm": 0.6951229882403026,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 5617
    },
    {
      "epoch": 0.05618,
      "grad_norm": 0.7979399037113596,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 5618
    },
    {
      "epoch": 0.05619,
      "grad_norm": 0.7776125407615552,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 5619
    },
    {
      "epoch": 0.0562,
      "grad_norm": 0.6817596297069429,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 5620
    },
    {
      "epoch": 0.05621,
      "grad_norm": 0.6549397635610529,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 5621
    },
    {
      "epoch": 0.05622,
      "grad_norm": 0.7329313978126628,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 5622
    },
    {
      "epoch": 0.05623,
      "grad_norm": 0.732086571615983,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 5623
    },
    {
      "epoch": 0.05624,
      "grad_norm": 0.6771095843567011,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 5624
    },
    {
      "epoch": 0.05625,
      "grad_norm": 0.6183411314364045,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 5625
    },
    {
      "epoch": 0.05626,
      "grad_norm": 0.6570365199694389,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 5626
    },
    {
      "epoch": 0.05627,
      "grad_norm": 0.6639084179326356,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 5627
    },
    {
      "epoch": 0.05628,
      "grad_norm": 0.6844026433487909,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 5628
    },
    {
      "epoch": 0.05629,
      "grad_norm": 0.82713889536583,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 5629
    },
    {
      "epoch": 0.0563,
      "grad_norm": 0.6942930662060884,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 5630
    },
    {
      "epoch": 0.05631,
      "grad_norm": 0.5294373134589537,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 5631
    },
    {
      "epoch": 0.05632,
      "grad_norm": 0.49155296329990794,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 5632
    },
    {
      "epoch": 0.05633,
      "grad_norm": 0.4835048319394899,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 5633
    },
    {
      "epoch": 0.05634,
      "grad_norm": 0.4557319202795549,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 5634
    },
    {
      "epoch": 0.05635,
      "grad_norm": 0.4698037555374509,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 5635
    },
    {
      "epoch": 0.05636,
      "grad_norm": 0.5988222861693672,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 5636
    },
    {
      "epoch": 0.05637,
      "grad_norm": 0.6571880685992473,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 5637
    },
    {
      "epoch": 0.05638,
      "grad_norm": 0.589395853147304,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 5638
    },
    {
      "epoch": 0.05639,
      "grad_norm": 0.5433844345616452,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 5639
    },
    {
      "epoch": 0.0564,
      "grad_norm": 0.5170019835368442,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 5640
    },
    {
      "epoch": 0.05641,
      "grad_norm": 0.4891316043455271,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 5641
    },
    {
      "epoch": 0.05642,
      "grad_norm": 0.4491321561168964,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 5642
    },
    {
      "epoch": 0.05643,
      "grad_norm": 0.4830092535839702,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 5643
    },
    {
      "epoch": 0.05644,
      "grad_norm": 0.5232904059467851,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 5644
    },
    {
      "epoch": 0.05645,
      "grad_norm": 0.567111969651976,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 5645
    },
    {
      "epoch": 0.05646,
      "grad_norm": 0.6631370135003563,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 5646
    },
    {
      "epoch": 0.05647,
      "grad_norm": 0.8154565355116417,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 5647
    },
    {
      "epoch": 0.05648,
      "grad_norm": 0.7782448973936725,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 5648
    },
    {
      "epoch": 0.05649,
      "grad_norm": 0.6294974249907404,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5649
    },
    {
      "epoch": 0.0565,
      "grad_norm": 0.658532379840979,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 5650
    },
    {
      "epoch": 0.05651,
      "grad_norm": 0.5881790419238977,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 5651
    },
    {
      "epoch": 0.05652,
      "grad_norm": 0.6487361989272737,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 5652
    },
    {
      "epoch": 0.05653,
      "grad_norm": 0.890297904598085,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 5653
    },
    {
      "epoch": 0.05654,
      "grad_norm": 1.122891362935015,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 5654
    },
    {
      "epoch": 0.05655,
      "grad_norm": 0.7784475530833843,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 5655
    },
    {
      "epoch": 0.05656,
      "grad_norm": 0.7039018333153184,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 5656
    },
    {
      "epoch": 0.05657,
      "grad_norm": 0.7611663979262661,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 5657
    },
    {
      "epoch": 0.05658,
      "grad_norm": 0.6301443417394809,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 5658
    },
    {
      "epoch": 0.05659,
      "grad_norm": 0.5799574664434455,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 5659
    },
    {
      "epoch": 0.0566,
      "grad_norm": 0.5398797700270416,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 5660
    },
    {
      "epoch": 0.05661,
      "grad_norm": 0.5607006474067805,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 5661
    },
    {
      "epoch": 0.05662,
      "grad_norm": 0.6312047749179707,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 5662
    },
    {
      "epoch": 0.05663,
      "grad_norm": 0.6691547203899587,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 5663
    },
    {
      "epoch": 0.05664,
      "grad_norm": 0.6839990390423485,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 5664
    },
    {
      "epoch": 0.05665,
      "grad_norm": 0.6483304454614884,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 5665
    },
    {
      "epoch": 0.05666,
      "grad_norm": 0.7336745114259869,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 5666
    },
    {
      "epoch": 0.05667,
      "grad_norm": 0.6601921754069777,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 5667
    },
    {
      "epoch": 0.05668,
      "grad_norm": 0.608008716479813,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 5668
    },
    {
      "epoch": 0.05669,
      "grad_norm": 0.6089029787701348,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 5669
    },
    {
      "epoch": 0.0567,
      "grad_norm": 0.5002563109306051,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 5670
    },
    {
      "epoch": 0.05671,
      "grad_norm": 0.5193596424869418,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 5671
    },
    {
      "epoch": 0.05672,
      "grad_norm": 0.6518666148309528,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 5672
    },
    {
      "epoch": 0.05673,
      "grad_norm": 0.7728626458240908,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 5673
    },
    {
      "epoch": 0.05674,
      "grad_norm": 0.8332874518712882,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 5674
    },
    {
      "epoch": 0.05675,
      "grad_norm": 0.8323964040847633,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 5675
    },
    {
      "epoch": 0.05676,
      "grad_norm": 0.7595324774497408,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 5676
    },
    {
      "epoch": 0.05677,
      "grad_norm": 0.6740603533153058,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 5677
    },
    {
      "epoch": 0.05678,
      "grad_norm": 0.5713494747040236,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 5678
    },
    {
      "epoch": 0.05679,
      "grad_norm": 0.6253195899590996,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 5679
    },
    {
      "epoch": 0.0568,
      "grad_norm": 0.6100710085093579,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 5680
    },
    {
      "epoch": 0.05681,
      "grad_norm": 0.6830054586678898,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 5681
    },
    {
      "epoch": 0.05682,
      "grad_norm": 0.7108282874400138,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 5682
    },
    {
      "epoch": 0.05683,
      "grad_norm": 0.5883840169923861,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 5683
    },
    {
      "epoch": 0.05684,
      "grad_norm": 0.6507038187659987,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 5684
    },
    {
      "epoch": 0.05685,
      "grad_norm": 0.6756833252452829,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 5685
    },
    {
      "epoch": 0.05686,
      "grad_norm": 0.583645658916334,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 5686
    },
    {
      "epoch": 0.05687,
      "grad_norm": 0.5953213292181491,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 5687
    },
    {
      "epoch": 0.05688,
      "grad_norm": 0.5475876848073131,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 5688
    },
    {
      "epoch": 0.05689,
      "grad_norm": 0.5025790725419514,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 5689
    },
    {
      "epoch": 0.0569,
      "grad_norm": 0.45433112969189393,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 5690
    },
    {
      "epoch": 0.05691,
      "grad_norm": 0.4140962723522923,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 5691
    },
    {
      "epoch": 0.05692,
      "grad_norm": 0.4662306842775959,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 5692
    },
    {
      "epoch": 0.05693,
      "grad_norm": 0.5891875363708755,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 5693
    },
    {
      "epoch": 0.05694,
      "grad_norm": 0.8176890746618,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 5694
    },
    {
      "epoch": 0.05695,
      "grad_norm": 1.0722082901443937,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 5695
    },
    {
      "epoch": 0.05696,
      "grad_norm": 0.9384436343645547,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 5696
    },
    {
      "epoch": 0.05697,
      "grad_norm": 0.7353450806650836,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 5697
    },
    {
      "epoch": 0.05698,
      "grad_norm": 0.6318836782831939,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 5698
    },
    {
      "epoch": 0.05699,
      "grad_norm": 0.5916316530186301,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 5699
    },
    {
      "epoch": 0.057,
      "grad_norm": 0.6535089788186025,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 5700
    },
    {
      "epoch": 0.05701,
      "grad_norm": 0.6546023694581979,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 5701
    },
    {
      "epoch": 0.05702,
      "grad_norm": 0.5084570882781324,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 5702
    },
    {
      "epoch": 0.05703,
      "grad_norm": 0.5133248015349259,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 5703
    },
    {
      "epoch": 0.05704,
      "grad_norm": 0.5358288934883745,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 5704
    },
    {
      "epoch": 0.05705,
      "grad_norm": 0.5900408511498093,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 5705
    },
    {
      "epoch": 0.05706,
      "grad_norm": 0.6627882566353759,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 5706
    },
    {
      "epoch": 0.05707,
      "grad_norm": 0.7192441694975358,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 5707
    },
    {
      "epoch": 0.05708,
      "grad_norm": 0.7445828354995511,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 5708
    },
    {
      "epoch": 0.05709,
      "grad_norm": 0.7483624354012154,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 5709
    },
    {
      "epoch": 0.0571,
      "grad_norm": 0.725767615807694,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 5710
    },
    {
      "epoch": 0.05711,
      "grad_norm": 0.7397045380207842,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 5711
    },
    {
      "epoch": 0.05712,
      "grad_norm": 0.7189079122457795,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 5712
    },
    {
      "epoch": 0.05713,
      "grad_norm": 0.7060174681207148,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 5713
    },
    {
      "epoch": 0.05714,
      "grad_norm": 0.6921393568332758,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 5714
    },
    {
      "epoch": 0.05715,
      "grad_norm": 0.6750383401887041,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 5715
    },
    {
      "epoch": 0.05716,
      "grad_norm": 0.7085886694681629,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5716
    },
    {
      "epoch": 0.05717,
      "grad_norm": 0.6663173016626173,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 5717
    },
    {
      "epoch": 0.05718,
      "grad_norm": 0.7941387320047305,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 5718
    },
    {
      "epoch": 0.05719,
      "grad_norm": 0.9177092410441985,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 5719
    },
    {
      "epoch": 0.0572,
      "grad_norm": 0.8971727958952642,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 5720
    },
    {
      "epoch": 0.05721,
      "grad_norm": 0.926040651918975,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 5721
    },
    {
      "epoch": 0.05722,
      "grad_norm": 0.899008075973923,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 5722
    },
    {
      "epoch": 0.05723,
      "grad_norm": 0.8257100704628738,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 5723
    },
    {
      "epoch": 0.05724,
      "grad_norm": 0.7778923288997804,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5724
    },
    {
      "epoch": 0.05725,
      "grad_norm": 0.6413770657071006,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 5725
    },
    {
      "epoch": 0.05726,
      "grad_norm": 0.6037306221880084,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 5726
    },
    {
      "epoch": 0.05727,
      "grad_norm": 0.5200633077018988,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 5727
    },
    {
      "epoch": 0.05728,
      "grad_norm": 0.5036743411797538,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 5728
    },
    {
      "epoch": 0.05729,
      "grad_norm": 0.4985411262542399,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 5729
    },
    {
      "epoch": 0.0573,
      "grad_norm": 0.4584917863448276,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 5730
    },
    {
      "epoch": 0.05731,
      "grad_norm": 0.5438247899724311,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 5731
    },
    {
      "epoch": 0.05732,
      "grad_norm": 0.5879044540226291,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 5732
    },
    {
      "epoch": 0.05733,
      "grad_norm": 0.5930153226482116,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 5733
    },
    {
      "epoch": 0.05734,
      "grad_norm": 0.551401918593653,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 5734
    },
    {
      "epoch": 0.05735,
      "grad_norm": 0.44849752852507807,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 5735
    },
    {
      "epoch": 0.05736,
      "grad_norm": 0.44905168397046813,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 5736
    },
    {
      "epoch": 0.05737,
      "grad_norm": 0.5268320731054086,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 5737
    },
    {
      "epoch": 0.05738,
      "grad_norm": 0.5127400458979028,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 5738
    },
    {
      "epoch": 0.05739,
      "grad_norm": 0.4584064056524342,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 5739
    },
    {
      "epoch": 0.0574,
      "grad_norm": 0.43315600343740956,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 5740
    },
    {
      "epoch": 0.05741,
      "grad_norm": 0.435296931858689,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 5741
    },
    {
      "epoch": 0.05742,
      "grad_norm": 0.4223546407108866,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 5742
    },
    {
      "epoch": 0.05743,
      "grad_norm": 0.3829337526851232,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 5743
    },
    {
      "epoch": 0.05744,
      "grad_norm": 0.42582439817855655,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 5744
    },
    {
      "epoch": 0.05745,
      "grad_norm": 0.5032675064997798,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 5745
    },
    {
      "epoch": 0.05746,
      "grad_norm": 0.5371812025082381,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 5746
    },
    {
      "epoch": 0.05747,
      "grad_norm": 0.5691353438016382,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 5747
    },
    {
      "epoch": 0.05748,
      "grad_norm": 0.5609944240328593,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 5748
    },
    {
      "epoch": 0.05749,
      "grad_norm": 0.5812063114829625,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 5749
    },
    {
      "epoch": 0.0575,
      "grad_norm": 0.59242400709736,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 5750
    },
    {
      "epoch": 0.05751,
      "grad_norm": 0.5892062367000988,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 5751
    },
    {
      "epoch": 0.05752,
      "grad_norm": 0.6411889904390121,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 5752
    },
    {
      "epoch": 0.05753,
      "grad_norm": 0.7623604527296114,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 5753
    },
    {
      "epoch": 0.05754,
      "grad_norm": 0.8190982577351572,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 5754
    },
    {
      "epoch": 0.05755,
      "grad_norm": 0.8129781285023648,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 5755
    },
    {
      "epoch": 0.05756,
      "grad_norm": 0.9645409102735437,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 5756
    },
    {
      "epoch": 0.05757,
      "grad_norm": 1.0613767929848879,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 5757
    },
    {
      "epoch": 0.05758,
      "grad_norm": 0.8340493918370546,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 5758
    },
    {
      "epoch": 0.05759,
      "grad_norm": 0.7501058117672132,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 5759
    },
    {
      "epoch": 0.0576,
      "grad_norm": 0.6678453188221255,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 5760
    },
    {
      "epoch": 0.05761,
      "grad_norm": 0.7222477111017501,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 5761
    },
    {
      "epoch": 0.05762,
      "grad_norm": 0.8358884638373766,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 5762
    },
    {
      "epoch": 0.05763,
      "grad_norm": 0.9017240989057173,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 5763
    },
    {
      "epoch": 0.05764,
      "grad_norm": 0.9525671514720745,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 5764
    },
    {
      "epoch": 0.05765,
      "grad_norm": 0.903840898655556,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 5765
    },
    {
      "epoch": 0.05766,
      "grad_norm": 0.841117793582558,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 5766
    },
    {
      "epoch": 0.05767,
      "grad_norm": 0.8184231215193009,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5767
    },
    {
      "epoch": 0.05768,
      "grad_norm": 0.7953651176604322,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 5768
    },
    {
      "epoch": 0.05769,
      "grad_norm": 0.8641222338040627,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 5769
    },
    {
      "epoch": 0.0577,
      "grad_norm": 0.8267946717518797,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 5770
    },
    {
      "epoch": 0.05771,
      "grad_norm": 0.7549731798996204,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 5771
    },
    {
      "epoch": 0.05772,
      "grad_norm": 0.8471888716859771,
      "learning_rate": 0.003,
      "loss": 4.174,
      "step": 5772
    },
    {
      "epoch": 0.05773,
      "grad_norm": 0.8004119994019282,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 5773
    },
    {
      "epoch": 0.05774,
      "grad_norm": 0.6795800808077224,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 5774
    },
    {
      "epoch": 0.05775,
      "grad_norm": 0.6426254130546161,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 5775
    },
    {
      "epoch": 0.05776,
      "grad_norm": 0.5457159419232533,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 5776
    },
    {
      "epoch": 0.05777,
      "grad_norm": 0.5193032203296611,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 5777
    },
    {
      "epoch": 0.05778,
      "grad_norm": 0.5430365890766845,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5778
    },
    {
      "epoch": 0.05779,
      "grad_norm": 0.5514106031103018,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 5779
    },
    {
      "epoch": 0.0578,
      "grad_norm": 0.5595076136345223,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 5780
    },
    {
      "epoch": 0.05781,
      "grad_norm": 0.6358454172899635,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 5781
    },
    {
      "epoch": 0.05782,
      "grad_norm": 0.7067244036587653,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 5782
    },
    {
      "epoch": 0.05783,
      "grad_norm": 0.6579510175517927,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 5783
    },
    {
      "epoch": 0.05784,
      "grad_norm": 0.4615349795672166,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 5784
    },
    {
      "epoch": 0.05785,
      "grad_norm": 0.40423330402508734,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 5785
    },
    {
      "epoch": 0.05786,
      "grad_norm": 0.4219551145978536,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 5786
    },
    {
      "epoch": 0.05787,
      "grad_norm": 0.47190796765263787,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 5787
    },
    {
      "epoch": 0.05788,
      "grad_norm": 0.48647627634592133,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 5788
    },
    {
      "epoch": 0.05789,
      "grad_norm": 0.45916564143367755,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 5789
    },
    {
      "epoch": 0.0579,
      "grad_norm": 0.38438157656491667,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 5790
    },
    {
      "epoch": 0.05791,
      "grad_norm": 0.380767444456616,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 5791
    },
    {
      "epoch": 0.05792,
      "grad_norm": 0.39494151370656566,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 5792
    },
    {
      "epoch": 0.05793,
      "grad_norm": 0.4521836450575782,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 5793
    },
    {
      "epoch": 0.05794,
      "grad_norm": 0.5585408775581093,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 5794
    },
    {
      "epoch": 0.05795,
      "grad_norm": 0.7037387927731176,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 5795
    },
    {
      "epoch": 0.05796,
      "grad_norm": 0.7512006570045856,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 5796
    },
    {
      "epoch": 0.05797,
      "grad_norm": 0.6139515253192398,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 5797
    },
    {
      "epoch": 0.05798,
      "grad_norm": 0.42741457828347207,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 5798
    },
    {
      "epoch": 0.05799,
      "grad_norm": 0.4697070878309648,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 5799
    },
    {
      "epoch": 0.058,
      "grad_norm": 0.5718953135900179,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 5800
    },
    {
      "epoch": 0.05801,
      "grad_norm": 0.6101112337124837,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 5801
    },
    {
      "epoch": 0.05802,
      "grad_norm": 0.518362856862478,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 5802
    },
    {
      "epoch": 0.05803,
      "grad_norm": 0.5319512529198976,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 5803
    },
    {
      "epoch": 0.05804,
      "grad_norm": 0.583888596545647,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 5804
    },
    {
      "epoch": 0.05805,
      "grad_norm": 0.6332538337272945,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 5805
    },
    {
      "epoch": 0.05806,
      "grad_norm": 0.6591146413634413,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 5806
    },
    {
      "epoch": 0.05807,
      "grad_norm": 0.5648403140301973,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 5807
    },
    {
      "epoch": 0.05808,
      "grad_norm": 0.5082305082390789,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 5808
    },
    {
      "epoch": 0.05809,
      "grad_norm": 0.45869354878914337,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 5809
    },
    {
      "epoch": 0.0581,
      "grad_norm": 0.5128621779633227,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 5810
    },
    {
      "epoch": 0.05811,
      "grad_norm": 0.4634974372718931,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 5811
    },
    {
      "epoch": 0.05812,
      "grad_norm": 0.5415369741433366,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 5812
    },
    {
      "epoch": 0.05813,
      "grad_norm": 0.6045041038302872,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 5813
    },
    {
      "epoch": 0.05814,
      "grad_norm": 0.781348878698672,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 5814
    },
    {
      "epoch": 0.05815,
      "grad_norm": 0.9049868579609852,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 5815
    },
    {
      "epoch": 0.05816,
      "grad_norm": 0.8496790014644907,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 5816
    },
    {
      "epoch": 0.05817,
      "grad_norm": 0.6848842285781443,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 5817
    },
    {
      "epoch": 0.05818,
      "grad_norm": 0.6972808127885899,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 5818
    },
    {
      "epoch": 0.05819,
      "grad_norm": 0.7928565409685948,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 5819
    },
    {
      "epoch": 0.0582,
      "grad_norm": 0.8531050086329389,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 5820
    },
    {
      "epoch": 0.05821,
      "grad_norm": 0.8181521585974338,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 5821
    },
    {
      "epoch": 0.05822,
      "grad_norm": 0.6685069393486631,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 5822
    },
    {
      "epoch": 0.05823,
      "grad_norm": 0.7178511889820098,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 5823
    },
    {
      "epoch": 0.05824,
      "grad_norm": 0.7316138274504331,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 5824
    },
    {
      "epoch": 0.05825,
      "grad_norm": 0.7577488477174734,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 5825
    },
    {
      "epoch": 0.05826,
      "grad_norm": 0.7352824443915027,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 5826
    },
    {
      "epoch": 0.05827,
      "grad_norm": 0.737466448111901,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 5827
    },
    {
      "epoch": 0.05828,
      "grad_norm": 0.5756572073483629,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 5828
    },
    {
      "epoch": 0.05829,
      "grad_norm": 0.5700830450607922,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 5829
    },
    {
      "epoch": 0.0583,
      "grad_norm": 0.6090643389167884,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 5830
    },
    {
      "epoch": 0.05831,
      "grad_norm": 0.6598337613053546,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 5831
    },
    {
      "epoch": 0.05832,
      "grad_norm": 0.7236800212652943,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 5832
    },
    {
      "epoch": 0.05833,
      "grad_norm": 0.8312662391240169,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 5833
    },
    {
      "epoch": 0.05834,
      "grad_norm": 0.8570982085297809,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 5834
    },
    {
      "epoch": 0.05835,
      "grad_norm": 0.7634242055631113,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 5835
    },
    {
      "epoch": 0.05836,
      "grad_norm": 0.8129339709531613,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 5836
    },
    {
      "epoch": 0.05837,
      "grad_norm": 0.7114562019507951,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 5837
    },
    {
      "epoch": 0.05838,
      "grad_norm": 0.7146446360282164,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 5838
    },
    {
      "epoch": 0.05839,
      "grad_norm": 0.6826153272700018,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 5839
    },
    {
      "epoch": 0.0584,
      "grad_norm": 0.6296145264005895,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 5840
    },
    {
      "epoch": 0.05841,
      "grad_norm": 0.6400001378596765,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 5841
    },
    {
      "epoch": 0.05842,
      "grad_norm": 0.6745147859755771,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 5842
    },
    {
      "epoch": 0.05843,
      "grad_norm": 0.7204687011170773,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 5843
    },
    {
      "epoch": 0.05844,
      "grad_norm": 0.6692907850570987,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 5844
    },
    {
      "epoch": 0.05845,
      "grad_norm": 0.6464847846170348,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 5845
    },
    {
      "epoch": 0.05846,
      "grad_norm": 0.5903179636360006,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 5846
    },
    {
      "epoch": 0.05847,
      "grad_norm": 0.5715707353163612,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 5847
    },
    {
      "epoch": 0.05848,
      "grad_norm": 0.6188778057459394,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 5848
    },
    {
      "epoch": 0.05849,
      "grad_norm": 0.679788170958821,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 5849
    },
    {
      "epoch": 0.0585,
      "grad_norm": 0.6866570136156571,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 5850
    },
    {
      "epoch": 0.05851,
      "grad_norm": 0.6035947467507179,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 5851
    },
    {
      "epoch": 0.05852,
      "grad_norm": 0.6646374634051921,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 5852
    },
    {
      "epoch": 0.05853,
      "grad_norm": 0.6845982684639399,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 5853
    },
    {
      "epoch": 0.05854,
      "grad_norm": 0.6479252813025695,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 5854
    },
    {
      "epoch": 0.05855,
      "grad_norm": 0.729313342535791,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 5855
    },
    {
      "epoch": 0.05856,
      "grad_norm": 0.8374311171433904,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 5856
    },
    {
      "epoch": 0.05857,
      "grad_norm": 0.8450077468218878,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 5857
    },
    {
      "epoch": 0.05858,
      "grad_norm": 0.7554133376301935,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 5858
    },
    {
      "epoch": 0.05859,
      "grad_norm": 0.6288985709272776,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 5859
    },
    {
      "epoch": 0.0586,
      "grad_norm": 0.6807746349622874,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 5860
    },
    {
      "epoch": 0.05861,
      "grad_norm": 0.6685473536685101,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 5861
    },
    {
      "epoch": 0.05862,
      "grad_norm": 0.5947123918251042,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 5862
    },
    {
      "epoch": 0.05863,
      "grad_norm": 0.6529519804727131,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 5863
    },
    {
      "epoch": 0.05864,
      "grad_norm": 0.7095951129146904,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 5864
    },
    {
      "epoch": 0.05865,
      "grad_norm": 0.69339783930652,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 5865
    },
    {
      "epoch": 0.05866,
      "grad_norm": 0.6690878299187139,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 5866
    },
    {
      "epoch": 0.05867,
      "grad_norm": 0.7086863427678343,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 5867
    },
    {
      "epoch": 0.05868,
      "grad_norm": 0.6786171909040684,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 5868
    },
    {
      "epoch": 0.05869,
      "grad_norm": 0.5232033904488403,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 5869
    },
    {
      "epoch": 0.0587,
      "grad_norm": 0.47022549323018026,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 5870
    },
    {
      "epoch": 0.05871,
      "grad_norm": 0.44108238286000917,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 5871
    },
    {
      "epoch": 0.05872,
      "grad_norm": 0.3913734511800551,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 5872
    },
    {
      "epoch": 0.05873,
      "grad_norm": 0.42138434666249186,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 5873
    },
    {
      "epoch": 0.05874,
      "grad_norm": 0.4288958519233743,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 5874
    },
    {
      "epoch": 0.05875,
      "grad_norm": 0.44577682626719106,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 5875
    },
    {
      "epoch": 0.05876,
      "grad_norm": 0.4510573231336446,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 5876
    },
    {
      "epoch": 0.05877,
      "grad_norm": 0.43564108247460703,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 5877
    },
    {
      "epoch": 0.05878,
      "grad_norm": 0.5419890900941077,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 5878
    },
    {
      "epoch": 0.05879,
      "grad_norm": 0.6244220397965549,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 5879
    },
    {
      "epoch": 0.0588,
      "grad_norm": 0.7402086193529186,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 5880
    },
    {
      "epoch": 0.05881,
      "grad_norm": 0.9386524101489326,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 5881
    },
    {
      "epoch": 0.05882,
      "grad_norm": 0.9522508870914118,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 5882
    },
    {
      "epoch": 0.05883,
      "grad_norm": 0.8075223632251547,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 5883
    },
    {
      "epoch": 0.05884,
      "grad_norm": 0.8799909048791514,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 5884
    },
    {
      "epoch": 0.05885,
      "grad_norm": 0.8530787571893335,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 5885
    },
    {
      "epoch": 0.05886,
      "grad_norm": 0.7329103122991582,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 5886
    },
    {
      "epoch": 0.05887,
      "grad_norm": 0.6748479851298841,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 5887
    },
    {
      "epoch": 0.05888,
      "grad_norm": 0.6343669394015016,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 5888
    },
    {
      "epoch": 0.05889,
      "grad_norm": 0.6277512686474859,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 5889
    },
    {
      "epoch": 0.0589,
      "grad_norm": 0.5871377194165454,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 5890
    },
    {
      "epoch": 0.05891,
      "grad_norm": 0.6464959924160834,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 5891
    },
    {
      "epoch": 0.05892,
      "grad_norm": 0.666522134105091,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 5892
    },
    {
      "epoch": 0.05893,
      "grad_norm": 0.6842219549286493,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 5893
    },
    {
      "epoch": 0.05894,
      "grad_norm": 0.6497740081031832,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 5894
    },
    {
      "epoch": 0.05895,
      "grad_norm": 0.6324488507002164,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 5895
    },
    {
      "epoch": 0.05896,
      "grad_norm": 0.6784791219839371,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 5896
    },
    {
      "epoch": 0.05897,
      "grad_norm": 0.633556875627449,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 5897
    },
    {
      "epoch": 0.05898,
      "grad_norm": 0.7611480962202484,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 5898
    },
    {
      "epoch": 0.05899,
      "grad_norm": 0.8755081899697468,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 5899
    },
    {
      "epoch": 0.059,
      "grad_norm": 0.7092368456575846,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 5900
    },
    {
      "epoch": 0.05901,
      "grad_norm": 0.7082980339584151,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 5901
    },
    {
      "epoch": 0.05902,
      "grad_norm": 0.5794941356444853,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 5902
    },
    {
      "epoch": 0.05903,
      "grad_norm": 0.6148369105501363,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 5903
    },
    {
      "epoch": 0.05904,
      "grad_norm": 0.49903767340028915,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 5904
    },
    {
      "epoch": 0.05905,
      "grad_norm": 0.5199067308874674,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 5905
    },
    {
      "epoch": 0.05906,
      "grad_norm": 0.555958949304611,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 5906
    },
    {
      "epoch": 0.05907,
      "grad_norm": 0.6558790760046532,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 5907
    },
    {
      "epoch": 0.05908,
      "grad_norm": 0.7027699977426964,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 5908
    },
    {
      "epoch": 0.05909,
      "grad_norm": 0.695163190283452,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 5909
    },
    {
      "epoch": 0.0591,
      "grad_norm": 0.6843743312475742,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 5910
    },
    {
      "epoch": 0.05911,
      "grad_norm": 0.8025018895941953,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 5911
    },
    {
      "epoch": 0.05912,
      "grad_norm": 0.7607180090206141,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 5912
    },
    {
      "epoch": 0.05913,
      "grad_norm": 0.6684077408320309,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 5913
    },
    {
      "epoch": 0.05914,
      "grad_norm": 0.5315953517109148,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 5914
    },
    {
      "epoch": 0.05915,
      "grad_norm": 0.5149999875273275,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 5915
    },
    {
      "epoch": 0.05916,
      "grad_norm": 0.5083952172925276,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 5916
    },
    {
      "epoch": 0.05917,
      "grad_norm": 0.6049621514969292,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 5917
    },
    {
      "epoch": 0.05918,
      "grad_norm": 0.6835808900733502,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 5918
    },
    {
      "epoch": 0.05919,
      "grad_norm": 0.6391964983836171,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 5919
    },
    {
      "epoch": 0.0592,
      "grad_norm": 0.6143559484021399,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 5920
    },
    {
      "epoch": 0.05921,
      "grad_norm": 0.5979521195261379,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 5921
    },
    {
      "epoch": 0.05922,
      "grad_norm": 0.5693281868266068,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 5922
    },
    {
      "epoch": 0.05923,
      "grad_norm": 0.4849615069695424,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 5923
    },
    {
      "epoch": 0.05924,
      "grad_norm": 0.46280219229994835,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 5924
    },
    {
      "epoch": 0.05925,
      "grad_norm": 0.48491396846071155,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 5925
    },
    {
      "epoch": 0.05926,
      "grad_norm": 0.4958640158739645,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 5926
    },
    {
      "epoch": 0.05927,
      "grad_norm": 0.4751378170054991,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 5927
    },
    {
      "epoch": 0.05928,
      "grad_norm": 0.4782244181945609,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 5928
    },
    {
      "epoch": 0.05929,
      "grad_norm": 0.5398915288946496,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 5929
    },
    {
      "epoch": 0.0593,
      "grad_norm": 0.5515514495551406,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 5930
    },
    {
      "epoch": 0.05931,
      "grad_norm": 0.5294701484364319,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 5931
    },
    {
      "epoch": 0.05932,
      "grad_norm": 0.636944020483588,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 5932
    },
    {
      "epoch": 0.05933,
      "grad_norm": 0.7543213645563572,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 5933
    },
    {
      "epoch": 0.05934,
      "grad_norm": 0.866274898856157,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 5934
    },
    {
      "epoch": 0.05935,
      "grad_norm": 0.8843523390695053,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 5935
    },
    {
      "epoch": 0.05936,
      "grad_norm": 0.8578932374136042,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 5936
    },
    {
      "epoch": 0.05937,
      "grad_norm": 0.9791083661921728,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 5937
    },
    {
      "epoch": 0.05938,
      "grad_norm": 1.091849551063209,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 5938
    },
    {
      "epoch": 0.05939,
      "grad_norm": 0.8366778414432169,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 5939
    },
    {
      "epoch": 0.0594,
      "grad_norm": 0.8548416780890227,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 5940
    },
    {
      "epoch": 0.05941,
      "grad_norm": 0.8185445127313269,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 5941
    },
    {
      "epoch": 0.05942,
      "grad_norm": 0.7989954573855615,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 5942
    },
    {
      "epoch": 0.05943,
      "grad_norm": 0.6699215014203319,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 5943
    },
    {
      "epoch": 0.05944,
      "grad_norm": 0.6793053946810763,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 5944
    },
    {
      "epoch": 0.05945,
      "grad_norm": 0.594609731292735,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5945
    },
    {
      "epoch": 0.05946,
      "grad_norm": 0.6220089630527496,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 5946
    },
    {
      "epoch": 0.05947,
      "grad_norm": 0.6014765437282273,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 5947
    },
    {
      "epoch": 0.05948,
      "grad_norm": 0.5605950560039095,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 5948
    },
    {
      "epoch": 0.05949,
      "grad_norm": 0.627701504083057,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 5949
    },
    {
      "epoch": 0.0595,
      "grad_norm": 0.825535661839418,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 5950
    },
    {
      "epoch": 0.05951,
      "grad_norm": 1.016734954201298,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 5951
    },
    {
      "epoch": 0.05952,
      "grad_norm": 0.9844508366684808,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 5952
    },
    {
      "epoch": 0.05953,
      "grad_norm": 0.7676164320052801,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 5953
    },
    {
      "epoch": 0.05954,
      "grad_norm": 0.6525936684271212,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 5954
    },
    {
      "epoch": 0.05955,
      "grad_norm": 0.6209927774101155,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 5955
    },
    {
      "epoch": 0.05956,
      "grad_norm": 0.5271299792696181,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 5956
    },
    {
      "epoch": 0.05957,
      "grad_norm": 0.4821600312349904,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 5957
    },
    {
      "epoch": 0.05958,
      "grad_norm": 0.4508961273097574,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 5958
    },
    {
      "epoch": 0.05959,
      "grad_norm": 0.39637824515281267,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 5959
    },
    {
      "epoch": 0.0596,
      "grad_norm": 0.38874299684571356,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 5960
    },
    {
      "epoch": 0.05961,
      "grad_norm": 0.3847016903534145,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 5961
    },
    {
      "epoch": 0.05962,
      "grad_norm": 0.5209058986494142,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 5962
    },
    {
      "epoch": 0.05963,
      "grad_norm": 0.6475184427595679,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 5963
    },
    {
      "epoch": 0.05964,
      "grad_norm": 0.8684223155180772,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 5964
    },
    {
      "epoch": 0.05965,
      "grad_norm": 0.9048617493837324,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 5965
    },
    {
      "epoch": 0.05966,
      "grad_norm": 0.6837588042230079,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 5966
    },
    {
      "epoch": 0.05967,
      "grad_norm": 0.7348291814277423,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 5967
    },
    {
      "epoch": 0.05968,
      "grad_norm": 0.8050791089461226,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 5968
    },
    {
      "epoch": 0.05969,
      "grad_norm": 0.6869212130630161,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 5969
    },
    {
      "epoch": 0.0597,
      "grad_norm": 0.6307870695677188,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 5970
    },
    {
      "epoch": 0.05971,
      "grad_norm": 0.5810377904286976,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 5971
    },
    {
      "epoch": 0.05972,
      "grad_norm": 0.6021740778839361,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 5972
    },
    {
      "epoch": 0.05973,
      "grad_norm": 0.6228448900216191,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 5973
    },
    {
      "epoch": 0.05974,
      "grad_norm": 0.6655393691311872,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 5974
    },
    {
      "epoch": 0.05975,
      "grad_norm": 0.7310132470202124,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 5975
    },
    {
      "epoch": 0.05976,
      "grad_norm": 0.644416431543167,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 5976
    },
    {
      "epoch": 0.05977,
      "grad_norm": 0.5496400386930255,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 5977
    },
    {
      "epoch": 0.05978,
      "grad_norm": 0.5085905535105046,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 5978
    },
    {
      "epoch": 0.05979,
      "grad_norm": 0.4862541012980576,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 5979
    },
    {
      "epoch": 0.0598,
      "grad_norm": 0.4826350144452677,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 5980
    },
    {
      "epoch": 0.05981,
      "grad_norm": 0.4741196340897285,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 5981
    },
    {
      "epoch": 0.05982,
      "grad_norm": 0.46060363145123095,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 5982
    },
    {
      "epoch": 0.05983,
      "grad_norm": 0.4820067683683615,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 5983
    },
    {
      "epoch": 0.05984,
      "grad_norm": 0.4691794530046038,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 5984
    },
    {
      "epoch": 0.05985,
      "grad_norm": 0.43934894841255706,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 5985
    },
    {
      "epoch": 0.05986,
      "grad_norm": 0.47930129072684313,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 5986
    },
    {
      "epoch": 0.05987,
      "grad_norm": 0.5100592025533284,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 5987
    },
    {
      "epoch": 0.05988,
      "grad_norm": 0.4911602197245711,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 5988
    },
    {
      "epoch": 0.05989,
      "grad_norm": 0.537756575806154,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 5989
    },
    {
      "epoch": 0.0599,
      "grad_norm": 0.6185246945024333,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 5990
    },
    {
      "epoch": 0.05991,
      "grad_norm": 0.6911938315101449,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 5991
    },
    {
      "epoch": 0.05992,
      "grad_norm": 0.7869203480784628,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 5992
    },
    {
      "epoch": 0.05993,
      "grad_norm": 0.8486504973239845,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 5993
    },
    {
      "epoch": 0.05994,
      "grad_norm": 0.7035449677364836,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 5994
    },
    {
      "epoch": 0.05995,
      "grad_norm": 0.6451536905550219,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 5995
    },
    {
      "epoch": 0.05996,
      "grad_norm": 0.5969532336692085,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 5996
    },
    {
      "epoch": 0.05997,
      "grad_norm": 0.5888895798721049,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 5997
    },
    {
      "epoch": 0.05998,
      "grad_norm": 0.5821959514538432,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 5998
    },
    {
      "epoch": 0.05999,
      "grad_norm": 0.6236789871480475,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 5999
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6956596076788343,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 6000
    },
    {
      "epoch": 0.06001,
      "grad_norm": 0.6802540034736559,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 6001
    },
    {
      "epoch": 0.06002,
      "grad_norm": 0.6479478065752834,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 6002
    },
    {
      "epoch": 0.06003,
      "grad_norm": 0.6764473408981168,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 6003
    },
    {
      "epoch": 0.06004,
      "grad_norm": 0.7085851864172142,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 6004
    },
    {
      "epoch": 0.06005,
      "grad_norm": 0.8647819122756384,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 6005
    },
    {
      "epoch": 0.06006,
      "grad_norm": 0.9638548780117764,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 6006
    },
    {
      "epoch": 0.06007,
      "grad_norm": 0.8220903209106395,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 6007
    },
    {
      "epoch": 0.06008,
      "grad_norm": 0.7201290584455559,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 6008
    },
    {
      "epoch": 0.06009,
      "grad_norm": 0.6481918637846485,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 6009
    },
    {
      "epoch": 0.0601,
      "grad_norm": 0.6606383295563236,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 6010
    },
    {
      "epoch": 0.06011,
      "grad_norm": 0.6045842842765333,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 6011
    },
    {
      "epoch": 0.06012,
      "grad_norm": 0.5311879872224512,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 6012
    },
    {
      "epoch": 0.06013,
      "grad_norm": 0.5661934584861681,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 6013
    },
    {
      "epoch": 0.06014,
      "grad_norm": 0.5103102427525318,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 6014
    },
    {
      "epoch": 0.06015,
      "grad_norm": 0.5332466894899932,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 6015
    },
    {
      "epoch": 0.06016,
      "grad_norm": 0.5814846632342617,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 6016
    },
    {
      "epoch": 0.06017,
      "grad_norm": 0.5584668704349169,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 6017
    },
    {
      "epoch": 0.06018,
      "grad_norm": 0.6487569147119433,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 6018
    },
    {
      "epoch": 0.06019,
      "grad_norm": 0.8160505827483635,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 6019
    },
    {
      "epoch": 0.0602,
      "grad_norm": 0.9505286189725469,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 6020
    },
    {
      "epoch": 0.06021,
      "grad_norm": 0.9008202254968174,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 6021
    },
    {
      "epoch": 0.06022,
      "grad_norm": 0.8318378400927622,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 6022
    },
    {
      "epoch": 0.06023,
      "grad_norm": 0.8535039615699999,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 6023
    },
    {
      "epoch": 0.06024,
      "grad_norm": 0.7266061447147755,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 6024
    },
    {
      "epoch": 0.06025,
      "grad_norm": 0.6534591173507039,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 6025
    },
    {
      "epoch": 0.06026,
      "grad_norm": 0.6882074560691404,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 6026
    },
    {
      "epoch": 0.06027,
      "grad_norm": 0.6152833498862915,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 6027
    },
    {
      "epoch": 0.06028,
      "grad_norm": 0.6917892126475053,
      "learning_rate": 0.003,
      "loss": 4.1526,
      "step": 6028
    },
    {
      "epoch": 0.06029,
      "grad_norm": 0.7744787029514602,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 6029
    },
    {
      "epoch": 0.0603,
      "grad_norm": 0.7298002661604193,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 6030
    },
    {
      "epoch": 0.06031,
      "grad_norm": 0.6529802283746298,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 6031
    },
    {
      "epoch": 0.06032,
      "grad_norm": 0.663361567248902,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 6032
    },
    {
      "epoch": 0.06033,
      "grad_norm": 0.6680068753946955,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 6033
    },
    {
      "epoch": 0.06034,
      "grad_norm": 0.5886111324153055,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 6034
    },
    {
      "epoch": 0.06035,
      "grad_norm": 0.5817861329776778,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 6035
    },
    {
      "epoch": 0.06036,
      "grad_norm": 0.5937836039304831,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 6036
    },
    {
      "epoch": 0.06037,
      "grad_norm": 0.5457734592349858,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 6037
    },
    {
      "epoch": 0.06038,
      "grad_norm": 0.5606694744948917,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 6038
    },
    {
      "epoch": 0.06039,
      "grad_norm": 0.5636222880096219,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 6039
    },
    {
      "epoch": 0.0604,
      "grad_norm": 0.499349110228407,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 6040
    },
    {
      "epoch": 0.06041,
      "grad_norm": 0.4402271663603966,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 6041
    },
    {
      "epoch": 0.06042,
      "grad_norm": 0.5428448624210946,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 6042
    },
    {
      "epoch": 0.06043,
      "grad_norm": 0.6553234947154927,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 6043
    },
    {
      "epoch": 0.06044,
      "grad_norm": 0.8600471908171244,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 6044
    },
    {
      "epoch": 0.06045,
      "grad_norm": 0.872706540571187,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 6045
    },
    {
      "epoch": 0.06046,
      "grad_norm": 0.821018845139322,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 6046
    },
    {
      "epoch": 0.06047,
      "grad_norm": 0.9233882633484479,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 6047
    },
    {
      "epoch": 0.06048,
      "grad_norm": 0.8442957052670749,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 6048
    },
    {
      "epoch": 0.06049,
      "grad_norm": 0.7302305536562084,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 6049
    },
    {
      "epoch": 0.0605,
      "grad_norm": 0.7221886690004308,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 6050
    },
    {
      "epoch": 0.06051,
      "grad_norm": 0.7360382510474752,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 6051
    },
    {
      "epoch": 0.06052,
      "grad_norm": 0.8032114038858249,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 6052
    },
    {
      "epoch": 0.06053,
      "grad_norm": 0.902493419850475,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 6053
    },
    {
      "epoch": 0.06054,
      "grad_norm": 0.806885544934829,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 6054
    },
    {
      "epoch": 0.06055,
      "grad_norm": 0.6869862769047108,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 6055
    },
    {
      "epoch": 0.06056,
      "grad_norm": 0.6404057857514671,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 6056
    },
    {
      "epoch": 0.06057,
      "grad_norm": 0.6290861083377466,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 6057
    },
    {
      "epoch": 0.06058,
      "grad_norm": 0.5444350086738762,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 6058
    },
    {
      "epoch": 0.06059,
      "grad_norm": 0.5143246073617321,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 6059
    },
    {
      "epoch": 0.0606,
      "grad_norm": 0.5226193521387048,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 6060
    },
    {
      "epoch": 0.06061,
      "grad_norm": 0.5393493288965573,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 6061
    },
    {
      "epoch": 0.06062,
      "grad_norm": 0.514502958864031,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 6062
    },
    {
      "epoch": 0.06063,
      "grad_norm": 0.5094815444270113,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 6063
    },
    {
      "epoch": 0.06064,
      "grad_norm": 0.565356316744902,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 6064
    },
    {
      "epoch": 0.06065,
      "grad_norm": 0.6136518813857853,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 6065
    },
    {
      "epoch": 0.06066,
      "grad_norm": 0.7283737698010649,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 6066
    },
    {
      "epoch": 0.06067,
      "grad_norm": 0.8404355722803185,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 6067
    },
    {
      "epoch": 0.06068,
      "grad_norm": 0.9467403298027881,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 6068
    },
    {
      "epoch": 0.06069,
      "grad_norm": 0.7197917961662677,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 6069
    },
    {
      "epoch": 0.0607,
      "grad_norm": 0.5774704901538709,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 6070
    },
    {
      "epoch": 0.06071,
      "grad_norm": 0.7311169673433349,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 6071
    },
    {
      "epoch": 0.06072,
      "grad_norm": 0.7458715679339405,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 6072
    },
    {
      "epoch": 0.06073,
      "grad_norm": 0.6582937125157123,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 6073
    },
    {
      "epoch": 0.06074,
      "grad_norm": 0.5828453881664645,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 6074
    },
    {
      "epoch": 0.06075,
      "grad_norm": 0.6029835041926357,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 6075
    },
    {
      "epoch": 0.06076,
      "grad_norm": 0.6620595916061112,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 6076
    },
    {
      "epoch": 0.06077,
      "grad_norm": 0.6564194015584814,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 6077
    },
    {
      "epoch": 0.06078,
      "grad_norm": 0.5186407499376411,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 6078
    },
    {
      "epoch": 0.06079,
      "grad_norm": 0.46056676475765035,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6079
    },
    {
      "epoch": 0.0608,
      "grad_norm": 0.47846889118582303,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 6080
    },
    {
      "epoch": 0.06081,
      "grad_norm": 0.5041919175905508,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 6081
    },
    {
      "epoch": 0.06082,
      "grad_norm": 0.46581276372898556,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 6082
    },
    {
      "epoch": 0.06083,
      "grad_norm": 0.44508131506221504,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 6083
    },
    {
      "epoch": 0.06084,
      "grad_norm": 0.4965899394145137,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 6084
    },
    {
      "epoch": 0.06085,
      "grad_norm": 0.54325827930552,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 6085
    },
    {
      "epoch": 0.06086,
      "grad_norm": 0.566296273115109,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 6086
    },
    {
      "epoch": 0.06087,
      "grad_norm": 0.6264732981988343,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 6087
    },
    {
      "epoch": 0.06088,
      "grad_norm": 0.7267394373677858,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 6088
    },
    {
      "epoch": 0.06089,
      "grad_norm": 0.6910375877495948,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 6089
    },
    {
      "epoch": 0.0609,
      "grad_norm": 0.6230953516324953,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 6090
    },
    {
      "epoch": 0.06091,
      "grad_norm": 0.6214986267493964,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 6091
    },
    {
      "epoch": 0.06092,
      "grad_norm": 0.5881606444052708,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 6092
    },
    {
      "epoch": 0.06093,
      "grad_norm": 0.6823668073031264,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 6093
    },
    {
      "epoch": 0.06094,
      "grad_norm": 0.8167174518561652,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 6094
    },
    {
      "epoch": 0.06095,
      "grad_norm": 0.7935802530659607,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 6095
    },
    {
      "epoch": 0.06096,
      "grad_norm": 0.8625378261886631,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 6096
    },
    {
      "epoch": 0.06097,
      "grad_norm": 0.8936840972481503,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 6097
    },
    {
      "epoch": 0.06098,
      "grad_norm": 0.9387438614632022,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 6098
    },
    {
      "epoch": 0.06099,
      "grad_norm": 0.929800973675657,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 6099
    },
    {
      "epoch": 0.061,
      "grad_norm": 0.7917197466465429,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 6100
    },
    {
      "epoch": 0.06101,
      "grad_norm": 0.696545220226268,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 6101
    },
    {
      "epoch": 0.06102,
      "grad_norm": 0.6949648238824377,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 6102
    },
    {
      "epoch": 0.06103,
      "grad_norm": 0.5850787846680839,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 6103
    },
    {
      "epoch": 0.06104,
      "grad_norm": 0.6577354660593795,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 6104
    },
    {
      "epoch": 0.06105,
      "grad_norm": 0.5544659377310093,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 6105
    },
    {
      "epoch": 0.06106,
      "grad_norm": 0.597694523433053,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 6106
    },
    {
      "epoch": 0.06107,
      "grad_norm": 0.6874240958175389,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 6107
    },
    {
      "epoch": 0.06108,
      "grad_norm": 0.7673053125459535,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 6108
    },
    {
      "epoch": 0.06109,
      "grad_norm": 0.8657897732774222,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 6109
    },
    {
      "epoch": 0.0611,
      "grad_norm": 0.8160350383434505,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 6110
    },
    {
      "epoch": 0.06111,
      "grad_norm": 0.7059937113818745,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 6111
    },
    {
      "epoch": 0.06112,
      "grad_norm": 0.6729024659212177,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 6112
    },
    {
      "epoch": 0.06113,
      "grad_norm": 0.6927676866964618,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 6113
    },
    {
      "epoch": 0.06114,
      "grad_norm": 0.6588192987110121,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 6114
    },
    {
      "epoch": 0.06115,
      "grad_norm": 0.6984192814248,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 6115
    },
    {
      "epoch": 0.06116,
      "grad_norm": 0.6898828248533009,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 6116
    },
    {
      "epoch": 0.06117,
      "grad_norm": 0.6133649995686336,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 6117
    },
    {
      "epoch": 0.06118,
      "grad_norm": 0.5717021418949526,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 6118
    },
    {
      "epoch": 0.06119,
      "grad_norm": 0.5743733423784956,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 6119
    },
    {
      "epoch": 0.0612,
      "grad_norm": 0.6414842059318022,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 6120
    },
    {
      "epoch": 0.06121,
      "grad_norm": 0.6294145948404021,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 6121
    },
    {
      "epoch": 0.06122,
      "grad_norm": 0.6183422089928163,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 6122
    },
    {
      "epoch": 0.06123,
      "grad_norm": 0.5371837518354552,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 6123
    },
    {
      "epoch": 0.06124,
      "grad_norm": 0.5226763311000785,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 6124
    },
    {
      "epoch": 0.06125,
      "grad_norm": 0.42639897195355503,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 6125
    },
    {
      "epoch": 0.06126,
      "grad_norm": 0.3629196367260766,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 6126
    },
    {
      "epoch": 0.06127,
      "grad_norm": 0.368265832283412,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 6127
    },
    {
      "epoch": 0.06128,
      "grad_norm": 0.38181801243432145,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 6128
    },
    {
      "epoch": 0.06129,
      "grad_norm": 0.40095129455194956,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 6129
    },
    {
      "epoch": 0.0613,
      "grad_norm": 0.4499458135634559,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 6130
    },
    {
      "epoch": 0.06131,
      "grad_norm": 0.5593938292242131,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 6131
    },
    {
      "epoch": 0.06132,
      "grad_norm": 0.68971097830543,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 6132
    },
    {
      "epoch": 0.06133,
      "grad_norm": 0.7970564003592729,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 6133
    },
    {
      "epoch": 0.06134,
      "grad_norm": 0.7946308123903568,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 6134
    },
    {
      "epoch": 0.06135,
      "grad_norm": 0.7243365229476787,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 6135
    },
    {
      "epoch": 0.06136,
      "grad_norm": 0.7248377814506896,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 6136
    },
    {
      "epoch": 0.06137,
      "grad_norm": 0.689387348680236,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 6137
    },
    {
      "epoch": 0.06138,
      "grad_norm": 0.6434616078916127,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 6138
    },
    {
      "epoch": 0.06139,
      "grad_norm": 0.6388715462619938,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 6139
    },
    {
      "epoch": 0.0614,
      "grad_norm": 0.5543186971635615,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 6140
    },
    {
      "epoch": 0.06141,
      "grad_norm": 0.5449818716272995,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 6141
    },
    {
      "epoch": 0.06142,
      "grad_norm": 0.5906892436672208,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 6142
    },
    {
      "epoch": 0.06143,
      "grad_norm": 0.6241199656978966,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 6143
    },
    {
      "epoch": 0.06144,
      "grad_norm": 0.691568154043788,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 6144
    },
    {
      "epoch": 0.06145,
      "grad_norm": 0.800383551611604,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 6145
    },
    {
      "epoch": 0.06146,
      "grad_norm": 0.8240824904436047,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 6146
    },
    {
      "epoch": 0.06147,
      "grad_norm": 0.8394359495307258,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 6147
    },
    {
      "epoch": 0.06148,
      "grad_norm": 0.8704187123192046,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 6148
    },
    {
      "epoch": 0.06149,
      "grad_norm": 0.8286112000168357,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 6149
    },
    {
      "epoch": 0.0615,
      "grad_norm": 0.7291708358830001,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 6150
    },
    {
      "epoch": 0.06151,
      "grad_norm": 0.6713497435304215,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 6151
    },
    {
      "epoch": 0.06152,
      "grad_norm": 0.6781308025527774,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6152
    },
    {
      "epoch": 0.06153,
      "grad_norm": 0.652938479870769,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 6153
    },
    {
      "epoch": 0.06154,
      "grad_norm": 0.6842620061629228,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 6154
    },
    {
      "epoch": 0.06155,
      "grad_norm": 0.6696388139082615,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 6155
    },
    {
      "epoch": 0.06156,
      "grad_norm": 0.661871708727772,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 6156
    },
    {
      "epoch": 0.06157,
      "grad_norm": 0.6706443949736604,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 6157
    },
    {
      "epoch": 0.06158,
      "grad_norm": 0.6730091441682898,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 6158
    },
    {
      "epoch": 0.06159,
      "grad_norm": 0.6516318216884822,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 6159
    },
    {
      "epoch": 0.0616,
      "grad_norm": 0.6776433090637509,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 6160
    },
    {
      "epoch": 0.06161,
      "grad_norm": 0.705245630542708,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 6161
    },
    {
      "epoch": 0.06162,
      "grad_norm": 0.667033646386603,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 6162
    },
    {
      "epoch": 0.06163,
      "grad_norm": 0.6115409987155483,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 6163
    },
    {
      "epoch": 0.06164,
      "grad_norm": 0.6022989592102578,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 6164
    },
    {
      "epoch": 0.06165,
      "grad_norm": 0.4995381931517796,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 6165
    },
    {
      "epoch": 0.06166,
      "grad_norm": 0.5481265388426909,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 6166
    },
    {
      "epoch": 0.06167,
      "grad_norm": 0.5609608519278377,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 6167
    },
    {
      "epoch": 0.06168,
      "grad_norm": 0.5825751863249862,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 6168
    },
    {
      "epoch": 0.06169,
      "grad_norm": 0.7711393078072618,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 6169
    },
    {
      "epoch": 0.0617,
      "grad_norm": 1.0686022680290213,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 6170
    },
    {
      "epoch": 0.06171,
      "grad_norm": 0.8988455270603705,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 6171
    },
    {
      "epoch": 0.06172,
      "grad_norm": 0.668550526380113,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 6172
    },
    {
      "epoch": 0.06173,
      "grad_norm": 0.6432161964405613,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 6173
    },
    {
      "epoch": 0.06174,
      "grad_norm": 0.6965904305758828,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 6174
    },
    {
      "epoch": 0.06175,
      "grad_norm": 0.6884542211695148,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 6175
    },
    {
      "epoch": 0.06176,
      "grad_norm": 0.722156592407858,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 6176
    },
    {
      "epoch": 0.06177,
      "grad_norm": 0.7406001109119403,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 6177
    },
    {
      "epoch": 0.06178,
      "grad_norm": 0.7298011712863863,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 6178
    },
    {
      "epoch": 0.06179,
      "grad_norm": 0.7122823712875358,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 6179
    },
    {
      "epoch": 0.0618,
      "grad_norm": 0.5964322698799788,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 6180
    },
    {
      "epoch": 0.06181,
      "grad_norm": 0.581223696651277,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 6181
    },
    {
      "epoch": 0.06182,
      "grad_norm": 0.6517698358193322,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 6182
    },
    {
      "epoch": 0.06183,
      "grad_norm": 0.82973155998321,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 6183
    },
    {
      "epoch": 0.06184,
      "grad_norm": 1.0291494249526294,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 6184
    },
    {
      "epoch": 0.06185,
      "grad_norm": 0.838098174913646,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 6185
    },
    {
      "epoch": 0.06186,
      "grad_norm": 0.7438011498693125,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 6186
    },
    {
      "epoch": 0.06187,
      "grad_norm": 0.6718856898807979,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 6187
    },
    {
      "epoch": 0.06188,
      "grad_norm": 0.5340636370837376,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 6188
    },
    {
      "epoch": 0.06189,
      "grad_norm": 0.5159651984761162,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 6189
    },
    {
      "epoch": 0.0619,
      "grad_norm": 0.4916863941661442,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 6190
    },
    {
      "epoch": 0.06191,
      "grad_norm": 0.4762063974718235,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 6191
    },
    {
      "epoch": 0.06192,
      "grad_norm": 0.46078596463013155,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 6192
    },
    {
      "epoch": 0.06193,
      "grad_norm": 0.45789803674311924,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 6193
    },
    {
      "epoch": 0.06194,
      "grad_norm": 0.3890805745658892,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 6194
    },
    {
      "epoch": 0.06195,
      "grad_norm": 0.39301414143303115,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 6195
    },
    {
      "epoch": 0.06196,
      "grad_norm": 0.4015529007182125,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 6196
    },
    {
      "epoch": 0.06197,
      "grad_norm": 0.4155210473439802,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 6197
    },
    {
      "epoch": 0.06198,
      "grad_norm": 0.4460266614970883,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 6198
    },
    {
      "epoch": 0.06199,
      "grad_norm": 0.4771646815553033,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 6199
    },
    {
      "epoch": 0.062,
      "grad_norm": 0.5196589692573884,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 6200
    },
    {
      "epoch": 0.06201,
      "grad_norm": 0.654622055901127,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 6201
    },
    {
      "epoch": 0.06202,
      "grad_norm": 0.7996732523950479,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 6202
    },
    {
      "epoch": 0.06203,
      "grad_norm": 0.9238177375399548,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 6203
    },
    {
      "epoch": 0.06204,
      "grad_norm": 0.8129041862310228,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 6204
    },
    {
      "epoch": 0.06205,
      "grad_norm": 0.6014391143548676,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 6205
    },
    {
      "epoch": 0.06206,
      "grad_norm": 0.6512606966939022,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 6206
    },
    {
      "epoch": 0.06207,
      "grad_norm": 0.6313513041064389,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 6207
    },
    {
      "epoch": 0.06208,
      "grad_norm": 0.5802624432054514,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 6208
    },
    {
      "epoch": 0.06209,
      "grad_norm": 0.5117728936425465,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 6209
    },
    {
      "epoch": 0.0621,
      "grad_norm": 0.5061612945373342,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 6210
    },
    {
      "epoch": 0.06211,
      "grad_norm": 0.5713900960332042,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 6211
    },
    {
      "epoch": 0.06212,
      "grad_norm": 0.6069757619115741,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 6212
    },
    {
      "epoch": 0.06213,
      "grad_norm": 0.7325742487754486,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 6213
    },
    {
      "epoch": 0.06214,
      "grad_norm": 0.9327829534552855,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 6214
    },
    {
      "epoch": 0.06215,
      "grad_norm": 1.035143571261494,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 6215
    },
    {
      "epoch": 0.06216,
      "grad_norm": 0.9038892809710738,
      "learning_rate": 0.003,
      "loss": 4.1712,
      "step": 6216
    },
    {
      "epoch": 0.06217,
      "grad_norm": 0.8175109198583256,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 6217
    },
    {
      "epoch": 0.06218,
      "grad_norm": 0.8799511997661614,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 6218
    },
    {
      "epoch": 0.06219,
      "grad_norm": 0.9186492040029288,
      "learning_rate": 0.003,
      "loss": 4.1569,
      "step": 6219
    },
    {
      "epoch": 0.0622,
      "grad_norm": 0.904845342853041,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 6220
    },
    {
      "epoch": 0.06221,
      "grad_norm": 0.8348080339462427,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 6221
    },
    {
      "epoch": 0.06222,
      "grad_norm": 0.7823721322877822,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 6222
    },
    {
      "epoch": 0.06223,
      "grad_norm": 0.8355859013287794,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 6223
    },
    {
      "epoch": 0.06224,
      "grad_norm": 0.9715063855754449,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 6224
    },
    {
      "epoch": 0.06225,
      "grad_norm": 0.7808199034581652,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 6225
    },
    {
      "epoch": 0.06226,
      "grad_norm": 0.5329204522038972,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 6226
    },
    {
      "epoch": 0.06227,
      "grad_norm": 0.5450265158322551,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 6227
    },
    {
      "epoch": 0.06228,
      "grad_norm": 0.5351380607743084,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 6228
    },
    {
      "epoch": 0.06229,
      "grad_norm": 0.4599058234618347,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 6229
    },
    {
      "epoch": 0.0623,
      "grad_norm": 0.4873390679451465,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 6230
    },
    {
      "epoch": 0.06231,
      "grad_norm": 0.5391956252106268,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 6231
    },
    {
      "epoch": 0.06232,
      "grad_norm": 0.5342591833135322,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 6232
    },
    {
      "epoch": 0.06233,
      "grad_norm": 0.5191689316659253,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 6233
    },
    {
      "epoch": 0.06234,
      "grad_norm": 0.5396807325147936,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 6234
    },
    {
      "epoch": 0.06235,
      "grad_norm": 0.5913027145203329,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 6235
    },
    {
      "epoch": 0.06236,
      "grad_norm": 0.5979504829726415,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 6236
    },
    {
      "epoch": 0.06237,
      "grad_norm": 0.5052384980096988,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 6237
    },
    {
      "epoch": 0.06238,
      "grad_norm": 0.6128825413246882,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 6238
    },
    {
      "epoch": 0.06239,
      "grad_norm": 0.7235535837216264,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 6239
    },
    {
      "epoch": 0.0624,
      "grad_norm": 0.7240803076746741,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 6240
    },
    {
      "epoch": 0.06241,
      "grad_norm": 0.6862973733458866,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 6241
    },
    {
      "epoch": 0.06242,
      "grad_norm": 0.7830627984578179,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 6242
    },
    {
      "epoch": 0.06243,
      "grad_norm": 0.7591018516451389,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 6243
    },
    {
      "epoch": 0.06244,
      "grad_norm": 0.7952190700820353,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 6244
    },
    {
      "epoch": 0.06245,
      "grad_norm": 0.6835256460654824,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 6245
    },
    {
      "epoch": 0.06246,
      "grad_norm": 0.6543359618448353,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 6246
    },
    {
      "epoch": 0.06247,
      "grad_norm": 0.5419452686920828,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 6247
    },
    {
      "epoch": 0.06248,
      "grad_norm": 0.4806235291826554,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 6248
    },
    {
      "epoch": 0.06249,
      "grad_norm": 0.453910776396139,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 6249
    },
    {
      "epoch": 0.0625,
      "grad_norm": 0.491783945365717,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 6250
    },
    {
      "epoch": 0.06251,
      "grad_norm": 0.5018941564042159,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 6251
    },
    {
      "epoch": 0.06252,
      "grad_norm": 0.4921472089644119,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 6252
    },
    {
      "epoch": 0.06253,
      "grad_norm": 0.5756543753609307,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 6253
    },
    {
      "epoch": 0.06254,
      "grad_norm": 0.754887950867572,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6254
    },
    {
      "epoch": 0.06255,
      "grad_norm": 0.8752296029209932,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 6255
    },
    {
      "epoch": 0.06256,
      "grad_norm": 0.9844123215090611,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 6256
    },
    {
      "epoch": 0.06257,
      "grad_norm": 0.8214120124061037,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 6257
    },
    {
      "epoch": 0.06258,
      "grad_norm": 0.7602595490679227,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 6258
    },
    {
      "epoch": 0.06259,
      "grad_norm": 0.9107431312102489,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 6259
    },
    {
      "epoch": 0.0626,
      "grad_norm": 0.8325776100808221,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 6260
    },
    {
      "epoch": 0.06261,
      "grad_norm": 0.7571831266663098,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 6261
    },
    {
      "epoch": 0.06262,
      "grad_norm": 0.8785267861570065,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 6262
    },
    {
      "epoch": 0.06263,
      "grad_norm": 0.8761411863659299,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 6263
    },
    {
      "epoch": 0.06264,
      "grad_norm": 0.7407405897110528,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 6264
    },
    {
      "epoch": 0.06265,
      "grad_norm": 0.5999131309688215,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 6265
    },
    {
      "epoch": 0.06266,
      "grad_norm": 0.5788028949497352,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 6266
    },
    {
      "epoch": 0.06267,
      "grad_norm": 0.5413831011099959,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 6267
    },
    {
      "epoch": 0.06268,
      "grad_norm": 0.5311326877936203,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 6268
    },
    {
      "epoch": 0.06269,
      "grad_norm": 0.46005639814155846,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 6269
    },
    {
      "epoch": 0.0627,
      "grad_norm": 0.4377136995334488,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 6270
    },
    {
      "epoch": 0.06271,
      "grad_norm": 0.45104310998898184,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 6271
    },
    {
      "epoch": 0.06272,
      "grad_norm": 0.5739468267833442,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 6272
    },
    {
      "epoch": 0.06273,
      "grad_norm": 0.6519100999767281,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 6273
    },
    {
      "epoch": 0.06274,
      "grad_norm": 0.7586632185001425,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 6274
    },
    {
      "epoch": 0.06275,
      "grad_norm": 0.8032215370564139,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 6275
    },
    {
      "epoch": 0.06276,
      "grad_norm": 0.808677786213581,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 6276
    },
    {
      "epoch": 0.06277,
      "grad_norm": 0.7705443100288017,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 6277
    },
    {
      "epoch": 0.06278,
      "grad_norm": 0.6723307868789453,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 6278
    },
    {
      "epoch": 0.06279,
      "grad_norm": 0.650139228463092,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 6279
    },
    {
      "epoch": 0.0628,
      "grad_norm": 0.6941961358979846,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 6280
    },
    {
      "epoch": 0.06281,
      "grad_norm": 0.6921836377836419,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 6281
    },
    {
      "epoch": 0.06282,
      "grad_norm": 0.7247534756382068,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 6282
    },
    {
      "epoch": 0.06283,
      "grad_norm": 0.698538662441215,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 6283
    },
    {
      "epoch": 0.06284,
      "grad_norm": 0.6478092027016852,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 6284
    },
    {
      "epoch": 0.06285,
      "grad_norm": 0.588832262527032,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 6285
    },
    {
      "epoch": 0.06286,
      "grad_norm": 0.471220817590307,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 6286
    },
    {
      "epoch": 0.06287,
      "grad_norm": 0.5253206409382251,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 6287
    },
    {
      "epoch": 0.06288,
      "grad_norm": 0.5700598976368867,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 6288
    },
    {
      "epoch": 0.06289,
      "grad_norm": 0.5480401509047442,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 6289
    },
    {
      "epoch": 0.0629,
      "grad_norm": 0.45176823851380066,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 6290
    },
    {
      "epoch": 0.06291,
      "grad_norm": 0.42244036366221777,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 6291
    },
    {
      "epoch": 0.06292,
      "grad_norm": 0.4757895925155192,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 6292
    },
    {
      "epoch": 0.06293,
      "grad_norm": 0.46687329270252953,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 6293
    },
    {
      "epoch": 0.06294,
      "grad_norm": 0.4692166334566838,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 6294
    },
    {
      "epoch": 0.06295,
      "grad_norm": 0.4964943486782987,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 6295
    },
    {
      "epoch": 0.06296,
      "grad_norm": 0.6023566234041191,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 6296
    },
    {
      "epoch": 0.06297,
      "grad_norm": 0.6562138349481754,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 6297
    },
    {
      "epoch": 0.06298,
      "grad_norm": 0.7252858308308101,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 6298
    },
    {
      "epoch": 0.06299,
      "grad_norm": 0.7441518369961619,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 6299
    },
    {
      "epoch": 0.063,
      "grad_norm": 0.643909205073487,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 6300
    },
    {
      "epoch": 0.06301,
      "grad_norm": 0.7076717275211952,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 6301
    },
    {
      "epoch": 0.06302,
      "grad_norm": 0.8610584882505145,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 6302
    },
    {
      "epoch": 0.06303,
      "grad_norm": 0.9134770715261514,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 6303
    },
    {
      "epoch": 0.06304,
      "grad_norm": 0.8308795580072063,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 6304
    },
    {
      "epoch": 0.06305,
      "grad_norm": 0.7703883250041841,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 6305
    },
    {
      "epoch": 0.06306,
      "grad_norm": 0.8074367440959146,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 6306
    },
    {
      "epoch": 0.06307,
      "grad_norm": 0.7821961103686115,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 6307
    },
    {
      "epoch": 0.06308,
      "grad_norm": 0.793746924122557,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 6308
    },
    {
      "epoch": 0.06309,
      "grad_norm": 0.8187320878830382,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 6309
    },
    {
      "epoch": 0.0631,
      "grad_norm": 0.8140608370958674,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 6310
    },
    {
      "epoch": 0.06311,
      "grad_norm": 0.7925215372022332,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 6311
    },
    {
      "epoch": 0.06312,
      "grad_norm": 0.6948175841091384,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 6312
    },
    {
      "epoch": 0.06313,
      "grad_norm": 0.6098450127577909,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 6313
    },
    {
      "epoch": 0.06314,
      "grad_norm": 0.601788808023616,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 6314
    },
    {
      "epoch": 0.06315,
      "grad_norm": 0.7002256479736269,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 6315
    },
    {
      "epoch": 0.06316,
      "grad_norm": 0.6702801162745691,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 6316
    },
    {
      "epoch": 0.06317,
      "grad_norm": 0.6767392466613616,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 6317
    },
    {
      "epoch": 0.06318,
      "grad_norm": 0.7396695999029177,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 6318
    },
    {
      "epoch": 0.06319,
      "grad_norm": 0.7873666098443218,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 6319
    },
    {
      "epoch": 0.0632,
      "grad_norm": 0.7968696734464397,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 6320
    },
    {
      "epoch": 0.06321,
      "grad_norm": 0.748485567878394,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 6321
    },
    {
      "epoch": 0.06322,
      "grad_norm": 0.700562547460408,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 6322
    },
    {
      "epoch": 0.06323,
      "grad_norm": 0.5674479127547497,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 6323
    },
    {
      "epoch": 0.06324,
      "grad_norm": 0.5979201027650989,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 6324
    },
    {
      "epoch": 0.06325,
      "grad_norm": 0.7374786634922282,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 6325
    },
    {
      "epoch": 0.06326,
      "grad_norm": 0.700878890093403,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 6326
    },
    {
      "epoch": 0.06327,
      "grad_norm": 0.6414815261782608,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 6327
    },
    {
      "epoch": 0.06328,
      "grad_norm": 0.5746442825430522,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 6328
    },
    {
      "epoch": 0.06329,
      "grad_norm": 0.4936256424209013,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 6329
    },
    {
      "epoch": 0.0633,
      "grad_norm": 0.5247583876058833,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 6330
    },
    {
      "epoch": 0.06331,
      "grad_norm": 0.5330239575249108,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 6331
    },
    {
      "epoch": 0.06332,
      "grad_norm": 0.5813651225344605,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 6332
    },
    {
      "epoch": 0.06333,
      "grad_norm": 0.597763757106397,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 6333
    },
    {
      "epoch": 0.06334,
      "grad_norm": 0.6306087415380968,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 6334
    },
    {
      "epoch": 0.06335,
      "grad_norm": 0.7275752434891477,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 6335
    },
    {
      "epoch": 0.06336,
      "grad_norm": 0.7536067920651794,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 6336
    },
    {
      "epoch": 0.06337,
      "grad_norm": 0.7434090606482405,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 6337
    },
    {
      "epoch": 0.06338,
      "grad_norm": 0.7917061963285527,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 6338
    },
    {
      "epoch": 0.06339,
      "grad_norm": 0.7963119547355659,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 6339
    },
    {
      "epoch": 0.0634,
      "grad_norm": 0.7009711549087663,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 6340
    },
    {
      "epoch": 0.06341,
      "grad_norm": 0.560288845949728,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 6341
    },
    {
      "epoch": 0.06342,
      "grad_norm": 0.6129201020379068,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 6342
    },
    {
      "epoch": 0.06343,
      "grad_norm": 0.6289890308043102,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 6343
    },
    {
      "epoch": 0.06344,
      "grad_norm": 0.5883345878311979,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 6344
    },
    {
      "epoch": 0.06345,
      "grad_norm": 0.6114359553520017,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 6345
    },
    {
      "epoch": 0.06346,
      "grad_norm": 0.6130452330266231,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 6346
    },
    {
      "epoch": 0.06347,
      "grad_norm": 0.5609868656725044,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 6347
    },
    {
      "epoch": 0.06348,
      "grad_norm": 0.5111538310088432,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 6348
    },
    {
      "epoch": 0.06349,
      "grad_norm": 0.47960134551277134,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 6349
    },
    {
      "epoch": 0.0635,
      "grad_norm": 0.47528221979889074,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 6350
    },
    {
      "epoch": 0.06351,
      "grad_norm": 0.44208488045258604,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 6351
    },
    {
      "epoch": 0.06352,
      "grad_norm": 0.5710585928560157,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 6352
    },
    {
      "epoch": 0.06353,
      "grad_norm": 0.8023230715536823,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 6353
    },
    {
      "epoch": 0.06354,
      "grad_norm": 1.0817933740124142,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 6354
    },
    {
      "epoch": 0.06355,
      "grad_norm": 0.839081793625545,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 6355
    },
    {
      "epoch": 0.06356,
      "grad_norm": 0.6498473408279323,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 6356
    },
    {
      "epoch": 0.06357,
      "grad_norm": 0.8657856383717981,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 6357
    },
    {
      "epoch": 0.06358,
      "grad_norm": 0.7367851380124152,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 6358
    },
    {
      "epoch": 0.06359,
      "grad_norm": 0.5580748213290369,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 6359
    },
    {
      "epoch": 0.0636,
      "grad_norm": 0.5715351960251037,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 6360
    },
    {
      "epoch": 0.06361,
      "grad_norm": 0.565402082000186,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 6361
    },
    {
      "epoch": 0.06362,
      "grad_norm": 0.6032607082774909,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 6362
    },
    {
      "epoch": 0.06363,
      "grad_norm": 0.5994033595373762,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 6363
    },
    {
      "epoch": 0.06364,
      "grad_norm": 0.5608720225119531,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 6364
    },
    {
      "epoch": 0.06365,
      "grad_norm": 0.5416673020756672,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 6365
    },
    {
      "epoch": 0.06366,
      "grad_norm": 0.5508838364908115,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 6366
    },
    {
      "epoch": 0.06367,
      "grad_norm": 0.6095798041890194,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 6367
    },
    {
      "epoch": 0.06368,
      "grad_norm": 0.582397991543009,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 6368
    },
    {
      "epoch": 0.06369,
      "grad_norm": 0.6111604358018735,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 6369
    },
    {
      "epoch": 0.0637,
      "grad_norm": 0.6887636290273739,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 6370
    },
    {
      "epoch": 0.06371,
      "grad_norm": 0.7312125926181495,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 6371
    },
    {
      "epoch": 0.06372,
      "grad_norm": 0.7404666495620588,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 6372
    },
    {
      "epoch": 0.06373,
      "grad_norm": 0.7812823153466856,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 6373
    },
    {
      "epoch": 0.06374,
      "grad_norm": 0.7678179937935664,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 6374
    },
    {
      "epoch": 0.06375,
      "grad_norm": 0.7856368742093315,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 6375
    },
    {
      "epoch": 0.06376,
      "grad_norm": 0.7933141620096883,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 6376
    },
    {
      "epoch": 0.06377,
      "grad_norm": 0.7911553024083666,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 6377
    },
    {
      "epoch": 0.06378,
      "grad_norm": 0.7097351170207112,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 6378
    },
    {
      "epoch": 0.06379,
      "grad_norm": 0.6022623562253553,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 6379
    },
    {
      "epoch": 0.0638,
      "grad_norm": 0.6161727993530954,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 6380
    },
    {
      "epoch": 0.06381,
      "grad_norm": 0.5896296208532763,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 6381
    },
    {
      "epoch": 0.06382,
      "grad_norm": 0.5666427713210422,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 6382
    },
    {
      "epoch": 0.06383,
      "grad_norm": 0.5280695299014706,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 6383
    },
    {
      "epoch": 0.06384,
      "grad_norm": 0.5315287442798651,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 6384
    },
    {
      "epoch": 0.06385,
      "grad_norm": 0.6180327194782872,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 6385
    },
    {
      "epoch": 0.06386,
      "grad_norm": 0.6877124042376932,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 6386
    },
    {
      "epoch": 0.06387,
      "grad_norm": 0.7049546795784516,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 6387
    },
    {
      "epoch": 0.06388,
      "grad_norm": 0.7560102533270469,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 6388
    },
    {
      "epoch": 0.06389,
      "grad_norm": 0.7727138622823404,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 6389
    },
    {
      "epoch": 0.0639,
      "grad_norm": 0.6854773910756959,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 6390
    },
    {
      "epoch": 0.06391,
      "grad_norm": 0.6116296991895176,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 6391
    },
    {
      "epoch": 0.06392,
      "grad_norm": 0.7278138368053769,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 6392
    },
    {
      "epoch": 0.06393,
      "grad_norm": 0.8102481077528341,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 6393
    },
    {
      "epoch": 0.06394,
      "grad_norm": 0.9128273035282637,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 6394
    },
    {
      "epoch": 0.06395,
      "grad_norm": 0.8577312836930895,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 6395
    },
    {
      "epoch": 0.06396,
      "grad_norm": 0.7745179973456292,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 6396
    },
    {
      "epoch": 0.06397,
      "grad_norm": 0.7881218719479312,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 6397
    },
    {
      "epoch": 0.06398,
      "grad_norm": 0.7024891362838419,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 6398
    },
    {
      "epoch": 0.06399,
      "grad_norm": 0.7110448250319877,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 6399
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.7111684482738276,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6400
    },
    {
      "epoch": 0.06401,
      "grad_norm": 0.693597181970454,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 6401
    },
    {
      "epoch": 0.06402,
      "grad_norm": 0.7922387757761263,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 6402
    },
    {
      "epoch": 0.06403,
      "grad_norm": 0.8260320747984292,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 6403
    },
    {
      "epoch": 0.06404,
      "grad_norm": 0.729663274761036,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 6404
    },
    {
      "epoch": 0.06405,
      "grad_norm": 0.7378732939090409,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 6405
    },
    {
      "epoch": 0.06406,
      "grad_norm": 0.6383629571059021,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 6406
    },
    {
      "epoch": 0.06407,
      "grad_norm": 0.6050162196732369,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6407
    },
    {
      "epoch": 0.06408,
      "grad_norm": 0.5983568485448366,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 6408
    },
    {
      "epoch": 0.06409,
      "grad_norm": 0.5642818531961227,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 6409
    },
    {
      "epoch": 0.0641,
      "grad_norm": 0.5874430801692443,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 6410
    },
    {
      "epoch": 0.06411,
      "grad_norm": 0.6330649262504477,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 6411
    },
    {
      "epoch": 0.06412,
      "grad_norm": 0.6618713014028005,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 6412
    },
    {
      "epoch": 0.06413,
      "grad_norm": 0.700072236260971,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 6413
    },
    {
      "epoch": 0.06414,
      "grad_norm": 0.6353084694205889,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 6414
    },
    {
      "epoch": 0.06415,
      "grad_norm": 0.5831173982008553,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 6415
    },
    {
      "epoch": 0.06416,
      "grad_norm": 0.5437779001226543,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 6416
    },
    {
      "epoch": 0.06417,
      "grad_norm": 0.6107145316832859,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 6417
    },
    {
      "epoch": 0.06418,
      "grad_norm": 0.6005928402730104,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 6418
    },
    {
      "epoch": 0.06419,
      "grad_norm": 0.583926884989267,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 6419
    },
    {
      "epoch": 0.0642,
      "grad_norm": 0.6742080868290792,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 6420
    },
    {
      "epoch": 0.06421,
      "grad_norm": 0.7855355256548284,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 6421
    },
    {
      "epoch": 0.06422,
      "grad_norm": 0.9319136699441484,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 6422
    },
    {
      "epoch": 0.06423,
      "grad_norm": 0.999583894911011,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 6423
    },
    {
      "epoch": 0.06424,
      "grad_norm": 0.9331342374240993,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 6424
    },
    {
      "epoch": 0.06425,
      "grad_norm": 0.7793159824304509,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 6425
    },
    {
      "epoch": 0.06426,
      "grad_norm": 0.6962948866885783,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 6426
    },
    {
      "epoch": 0.06427,
      "grad_norm": 0.683410446043668,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 6427
    },
    {
      "epoch": 0.06428,
      "grad_norm": 0.6512307290531658,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 6428
    },
    {
      "epoch": 0.06429,
      "grad_norm": 0.6377825140551957,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 6429
    },
    {
      "epoch": 0.0643,
      "grad_norm": 0.5759747531004582,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 6430
    },
    {
      "epoch": 0.06431,
      "grad_norm": 0.5676646229844707,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 6431
    },
    {
      "epoch": 0.06432,
      "grad_norm": 0.5461669723653775,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 6432
    },
    {
      "epoch": 0.06433,
      "grad_norm": 0.45797081849907423,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 6433
    },
    {
      "epoch": 0.06434,
      "grad_norm": 0.40792521706337564,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 6434
    },
    {
      "epoch": 0.06435,
      "grad_norm": 0.4730517544060023,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 6435
    },
    {
      "epoch": 0.06436,
      "grad_norm": 0.4766134688447076,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 6436
    },
    {
      "epoch": 0.06437,
      "grad_norm": 0.45378874087374643,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 6437
    },
    {
      "epoch": 0.06438,
      "grad_norm": 0.44043582431299527,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 6438
    },
    {
      "epoch": 0.06439,
      "grad_norm": 0.46296521989188577,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 6439
    },
    {
      "epoch": 0.0644,
      "grad_norm": 0.43339108687839656,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 6440
    },
    {
      "epoch": 0.06441,
      "grad_norm": 0.5476027827440176,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 6441
    },
    {
      "epoch": 0.06442,
      "grad_norm": 0.7595327046230422,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 6442
    },
    {
      "epoch": 0.06443,
      "grad_norm": 0.8816633245265507,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 6443
    },
    {
      "epoch": 0.06444,
      "grad_norm": 0.7973460889122836,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6444
    },
    {
      "epoch": 0.06445,
      "grad_norm": 0.6695114396139051,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 6445
    },
    {
      "epoch": 0.06446,
      "grad_norm": 0.7112577504861094,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 6446
    },
    {
      "epoch": 0.06447,
      "grad_norm": 0.6669997171392789,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 6447
    },
    {
      "epoch": 0.06448,
      "grad_norm": 0.7000529377796273,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 6448
    },
    {
      "epoch": 0.06449,
      "grad_norm": 0.7629296906277161,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 6449
    },
    {
      "epoch": 0.0645,
      "grad_norm": 0.8151936619117285,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 6450
    },
    {
      "epoch": 0.06451,
      "grad_norm": 0.7790440723976773,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 6451
    },
    {
      "epoch": 0.06452,
      "grad_norm": 0.7540024646037367,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 6452
    },
    {
      "epoch": 0.06453,
      "grad_norm": 0.7911961892915145,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 6453
    },
    {
      "epoch": 0.06454,
      "grad_norm": 0.8237841593306827,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 6454
    },
    {
      "epoch": 0.06455,
      "grad_norm": 0.8231550575684212,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 6455
    },
    {
      "epoch": 0.06456,
      "grad_norm": 0.7811541896303539,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 6456
    },
    {
      "epoch": 0.06457,
      "grad_norm": 0.8750658750124761,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 6457
    },
    {
      "epoch": 0.06458,
      "grad_norm": 0.9462957109484857,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 6458
    },
    {
      "epoch": 0.06459,
      "grad_norm": 0.951508448401663,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 6459
    },
    {
      "epoch": 0.0646,
      "grad_norm": 0.7887348617904948,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 6460
    },
    {
      "epoch": 0.06461,
      "grad_norm": 0.6561119853474549,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 6461
    },
    {
      "epoch": 0.06462,
      "grad_norm": 0.6564843031193351,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 6462
    },
    {
      "epoch": 0.06463,
      "grad_norm": 0.6866574328355095,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 6463
    },
    {
      "epoch": 0.06464,
      "grad_norm": 0.7412943846552807,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 6464
    },
    {
      "epoch": 0.06465,
      "grad_norm": 0.8607666448452919,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 6465
    },
    {
      "epoch": 0.06466,
      "grad_norm": 0.9805980509960028,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 6466
    },
    {
      "epoch": 0.06467,
      "grad_norm": 0.8240025115330769,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 6467
    },
    {
      "epoch": 0.06468,
      "grad_norm": 0.6915322018007451,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 6468
    },
    {
      "epoch": 0.06469,
      "grad_norm": 0.758205250001039,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 6469
    },
    {
      "epoch": 0.0647,
      "grad_norm": 0.6894610388291892,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 6470
    },
    {
      "epoch": 0.06471,
      "grad_norm": 0.5734326985005797,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 6471
    },
    {
      "epoch": 0.06472,
      "grad_norm": 0.6006963939375836,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 6472
    },
    {
      "epoch": 0.06473,
      "grad_norm": 0.5903102918749725,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 6473
    },
    {
      "epoch": 0.06474,
      "grad_norm": 0.6245812199248715,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 6474
    },
    {
      "epoch": 0.06475,
      "grad_norm": 0.6385947320236768,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 6475
    },
    {
      "epoch": 0.06476,
      "grad_norm": 0.5964051582023622,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 6476
    },
    {
      "epoch": 0.06477,
      "grad_norm": 0.4545337714425668,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 6477
    },
    {
      "epoch": 0.06478,
      "grad_norm": 0.4067813603673349,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 6478
    },
    {
      "epoch": 0.06479,
      "grad_norm": 0.35829021719441345,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 6479
    },
    {
      "epoch": 0.0648,
      "grad_norm": 0.36043755979794306,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 6480
    },
    {
      "epoch": 0.06481,
      "grad_norm": 0.379368972959528,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 6481
    },
    {
      "epoch": 0.06482,
      "grad_norm": 0.3991318139475991,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6482
    },
    {
      "epoch": 0.06483,
      "grad_norm": 0.5102756201687999,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 6483
    },
    {
      "epoch": 0.06484,
      "grad_norm": 0.6747208649376389,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 6484
    },
    {
      "epoch": 0.06485,
      "grad_norm": 0.818751842126756,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 6485
    },
    {
      "epoch": 0.06486,
      "grad_norm": 0.8421941035958429,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 6486
    },
    {
      "epoch": 0.06487,
      "grad_norm": 0.8130967059681043,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 6487
    },
    {
      "epoch": 0.06488,
      "grad_norm": 0.8190321300487033,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 6488
    },
    {
      "epoch": 0.06489,
      "grad_norm": 0.8618794228209637,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 6489
    },
    {
      "epoch": 0.0649,
      "grad_norm": 0.8167067584812114,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 6490
    },
    {
      "epoch": 0.06491,
      "grad_norm": 0.7120159141814162,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 6491
    },
    {
      "epoch": 0.06492,
      "grad_norm": 0.6966415188190852,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 6492
    },
    {
      "epoch": 0.06493,
      "grad_norm": 0.6279768739419431,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 6493
    },
    {
      "epoch": 0.06494,
      "grad_norm": 0.6729099375336287,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 6494
    },
    {
      "epoch": 0.06495,
      "grad_norm": 0.7138117731723219,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 6495
    },
    {
      "epoch": 0.06496,
      "grad_norm": 0.7481765813018704,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 6496
    },
    {
      "epoch": 0.06497,
      "grad_norm": 0.6676385968633429,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 6497
    },
    {
      "epoch": 0.06498,
      "grad_norm": 0.5497380735072145,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 6498
    },
    {
      "epoch": 0.06499,
      "grad_norm": 0.5123283023400911,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 6499
    },
    {
      "epoch": 0.065,
      "grad_norm": 0.4909211309507681,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 6500
    },
    {
      "epoch": 0.06501,
      "grad_norm": 0.47186804780732317,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6501
    },
    {
      "epoch": 0.06502,
      "grad_norm": 0.4586207578169618,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 6502
    },
    {
      "epoch": 0.06503,
      "grad_norm": 0.4734551401435721,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 6503
    },
    {
      "epoch": 0.06504,
      "grad_norm": 0.4518516829518515,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 6504
    },
    {
      "epoch": 0.06505,
      "grad_norm": 0.3926267599626706,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 6505
    },
    {
      "epoch": 0.06506,
      "grad_norm": 0.4116861037565272,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 6506
    },
    {
      "epoch": 0.06507,
      "grad_norm": 0.4869551083875778,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 6507
    },
    {
      "epoch": 0.06508,
      "grad_norm": 0.595601444957384,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 6508
    },
    {
      "epoch": 0.06509,
      "grad_norm": 0.8176861477648554,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 6509
    },
    {
      "epoch": 0.0651,
      "grad_norm": 0.9793486633870401,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 6510
    },
    {
      "epoch": 0.06511,
      "grad_norm": 1.101260179163563,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 6511
    },
    {
      "epoch": 0.06512,
      "grad_norm": 0.7048089149615372,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 6512
    },
    {
      "epoch": 0.06513,
      "grad_norm": 0.6606285760107887,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 6513
    },
    {
      "epoch": 0.06514,
      "grad_norm": 0.6936176900884794,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 6514
    },
    {
      "epoch": 0.06515,
      "grad_norm": 0.5802739054552889,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 6515
    },
    {
      "epoch": 0.06516,
      "grad_norm": 0.5957435080730257,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6516
    },
    {
      "epoch": 0.06517,
      "grad_norm": 0.6026894638436581,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 6517
    },
    {
      "epoch": 0.06518,
      "grad_norm": 0.6442432050579253,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 6518
    },
    {
      "epoch": 0.06519,
      "grad_norm": 0.6958415784014198,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 6519
    },
    {
      "epoch": 0.0652,
      "grad_norm": 0.687200073329854,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 6520
    },
    {
      "epoch": 0.06521,
      "grad_norm": 0.6695110160765739,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 6521
    },
    {
      "epoch": 0.06522,
      "grad_norm": 0.6730631239357413,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 6522
    },
    {
      "epoch": 0.06523,
      "grad_norm": 0.6680040550931838,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 6523
    },
    {
      "epoch": 0.06524,
      "grad_norm": 0.7000701390808372,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 6524
    },
    {
      "epoch": 0.06525,
      "grad_norm": 0.6591874414899291,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 6525
    },
    {
      "epoch": 0.06526,
      "grad_norm": 0.5974142684472702,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 6526
    },
    {
      "epoch": 0.06527,
      "grad_norm": 0.629624470137727,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 6527
    },
    {
      "epoch": 0.06528,
      "grad_norm": 0.7082086868906479,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 6528
    },
    {
      "epoch": 0.06529,
      "grad_norm": 0.7547993486095116,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 6529
    },
    {
      "epoch": 0.0653,
      "grad_norm": 0.7160694526225951,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 6530
    },
    {
      "epoch": 0.06531,
      "grad_norm": 0.7091031625495451,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 6531
    },
    {
      "epoch": 0.06532,
      "grad_norm": 0.5960779659612676,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 6532
    },
    {
      "epoch": 0.06533,
      "grad_norm": 0.569584737632218,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 6533
    },
    {
      "epoch": 0.06534,
      "grad_norm": 0.47053846099049507,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 6534
    },
    {
      "epoch": 0.06535,
      "grad_norm": 0.4876438724766613,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 6535
    },
    {
      "epoch": 0.06536,
      "grad_norm": 0.5421075581742479,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 6536
    },
    {
      "epoch": 0.06537,
      "grad_norm": 0.6138808767175822,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 6537
    },
    {
      "epoch": 0.06538,
      "grad_norm": 0.681850513199098,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 6538
    },
    {
      "epoch": 0.06539,
      "grad_norm": 0.7093467740008924,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 6539
    },
    {
      "epoch": 0.0654,
      "grad_norm": 0.6796947934262606,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 6540
    },
    {
      "epoch": 0.06541,
      "grad_norm": 0.5895536662879667,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 6541
    },
    {
      "epoch": 0.06542,
      "grad_norm": 0.5887252594422324,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 6542
    },
    {
      "epoch": 0.06543,
      "grad_norm": 0.6070666236587158,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 6543
    },
    {
      "epoch": 0.06544,
      "grad_norm": 0.6336458361165094,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 6544
    },
    {
      "epoch": 0.06545,
      "grad_norm": 0.5994933552644387,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 6545
    },
    {
      "epoch": 0.06546,
      "grad_norm": 0.5861867238542918,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 6546
    },
    {
      "epoch": 0.06547,
      "grad_norm": 0.7675266494691649,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 6547
    },
    {
      "epoch": 0.06548,
      "grad_norm": 0.7584810341137693,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 6548
    },
    {
      "epoch": 0.06549,
      "grad_norm": 0.7244422065216164,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 6549
    },
    {
      "epoch": 0.0655,
      "grad_norm": 0.7864205468003803,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 6550
    },
    {
      "epoch": 0.06551,
      "grad_norm": 0.7433546549614126,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 6551
    },
    {
      "epoch": 0.06552,
      "grad_norm": 0.8758299277222058,
      "learning_rate": 0.003,
      "loss": 4.142,
      "step": 6552
    },
    {
      "epoch": 0.06553,
      "grad_norm": 0.981060258172319,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 6553
    },
    {
      "epoch": 0.06554,
      "grad_norm": 1.0836670305325147,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 6554
    },
    {
      "epoch": 0.06555,
      "grad_norm": 0.893552519172031,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 6555
    },
    {
      "epoch": 0.06556,
      "grad_norm": 0.8133887797335478,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 6556
    },
    {
      "epoch": 0.06557,
      "grad_norm": 0.8038068218784429,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 6557
    },
    {
      "epoch": 0.06558,
      "grad_norm": 0.8563243567849327,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 6558
    },
    {
      "epoch": 0.06559,
      "grad_norm": 0.8357600944003225,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 6559
    },
    {
      "epoch": 0.0656,
      "grad_norm": 0.7471745091465914,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 6560
    },
    {
      "epoch": 0.06561,
      "grad_norm": 0.7197252510341869,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 6561
    },
    {
      "epoch": 0.06562,
      "grad_norm": 0.6961222717520075,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 6562
    },
    {
      "epoch": 0.06563,
      "grad_norm": 0.7969065869116986,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 6563
    },
    {
      "epoch": 0.06564,
      "grad_norm": 0.8609617020319454,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 6564
    },
    {
      "epoch": 0.06565,
      "grad_norm": 0.7706054033095626,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 6565
    },
    {
      "epoch": 0.06566,
      "grad_norm": 0.736160028183247,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 6566
    },
    {
      "epoch": 0.06567,
      "grad_norm": 0.8827507523317322,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 6567
    },
    {
      "epoch": 0.06568,
      "grad_norm": 0.9730014527594969,
      "learning_rate": 0.003,
      "loss": 4.1614,
      "step": 6568
    },
    {
      "epoch": 0.06569,
      "grad_norm": 0.8987588032009837,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 6569
    },
    {
      "epoch": 0.0657,
      "grad_norm": 0.7480401403974306,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 6570
    },
    {
      "epoch": 0.06571,
      "grad_norm": 0.5934268593799419,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 6571
    },
    {
      "epoch": 0.06572,
      "grad_norm": 0.6585081494040603,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 6572
    },
    {
      "epoch": 0.06573,
      "grad_norm": 0.7295365583733733,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 6573
    },
    {
      "epoch": 0.06574,
      "grad_norm": 0.756954183791856,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 6574
    },
    {
      "epoch": 0.06575,
      "grad_norm": 0.7099875197144789,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6575
    },
    {
      "epoch": 0.06576,
      "grad_norm": 0.6986413932325001,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 6576
    },
    {
      "epoch": 0.06577,
      "grad_norm": 0.6408480221530382,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 6577
    },
    {
      "epoch": 0.06578,
      "grad_norm": 0.5861819343962567,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 6578
    },
    {
      "epoch": 0.06579,
      "grad_norm": 0.5772678504283916,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 6579
    },
    {
      "epoch": 0.0658,
      "grad_norm": 0.4907489111870778,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 6580
    },
    {
      "epoch": 0.06581,
      "grad_norm": 0.4803454909907143,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 6581
    },
    {
      "epoch": 0.06582,
      "grad_norm": 0.5180246165607337,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 6582
    },
    {
      "epoch": 0.06583,
      "grad_norm": 0.63192504153945,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 6583
    },
    {
      "epoch": 0.06584,
      "grad_norm": 0.7212896215769731,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 6584
    },
    {
      "epoch": 0.06585,
      "grad_norm": 0.7725745929887623,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 6585
    },
    {
      "epoch": 0.06586,
      "grad_norm": 0.7150680524436269,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 6586
    },
    {
      "epoch": 0.06587,
      "grad_norm": 0.6454511286444826,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 6587
    },
    {
      "epoch": 0.06588,
      "grad_norm": 0.6209496060701972,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 6588
    },
    {
      "epoch": 0.06589,
      "grad_norm": 0.643641834535616,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 6589
    },
    {
      "epoch": 0.0659,
      "grad_norm": 0.6545001554853087,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 6590
    },
    {
      "epoch": 0.06591,
      "grad_norm": 0.6756223183950393,
      "learning_rate": 0.003,
      "loss": 4.162,
      "step": 6591
    },
    {
      "epoch": 0.06592,
      "grad_norm": 0.6681680196479507,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 6592
    },
    {
      "epoch": 0.06593,
      "grad_norm": 0.6370540979582453,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 6593
    },
    {
      "epoch": 0.06594,
      "grad_norm": 0.6592450103168598,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 6594
    },
    {
      "epoch": 0.06595,
      "grad_norm": 0.6488902280727541,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 6595
    },
    {
      "epoch": 0.06596,
      "grad_norm": 0.6078864404662637,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 6596
    },
    {
      "epoch": 0.06597,
      "grad_norm": 0.5826732446122348,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 6597
    },
    {
      "epoch": 0.06598,
      "grad_norm": 0.5786447854885438,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 6598
    },
    {
      "epoch": 0.06599,
      "grad_norm": 0.5203150176294418,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 6599
    },
    {
      "epoch": 0.066,
      "grad_norm": 0.5194458831005944,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 6600
    },
    {
      "epoch": 0.06601,
      "grad_norm": 0.5125280130771507,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 6601
    },
    {
      "epoch": 0.06602,
      "grad_norm": 0.5255374937732523,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 6602
    },
    {
      "epoch": 0.06603,
      "grad_norm": 0.5756998423978795,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6603
    },
    {
      "epoch": 0.06604,
      "grad_norm": 0.6322066603487145,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 6604
    },
    {
      "epoch": 0.06605,
      "grad_norm": 0.6466004843165204,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 6605
    },
    {
      "epoch": 0.06606,
      "grad_norm": 0.7167353080028168,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 6606
    },
    {
      "epoch": 0.06607,
      "grad_norm": 0.8469413810873132,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 6607
    },
    {
      "epoch": 0.06608,
      "grad_norm": 0.7995923933103859,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 6608
    },
    {
      "epoch": 0.06609,
      "grad_norm": 0.681750060363617,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 6609
    },
    {
      "epoch": 0.0661,
      "grad_norm": 0.6480858792190892,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 6610
    },
    {
      "epoch": 0.06611,
      "grad_norm": 0.658286589419674,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 6611
    },
    {
      "epoch": 0.06612,
      "grad_norm": 0.6793208367131326,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 6612
    },
    {
      "epoch": 0.06613,
      "grad_norm": 0.6454677330588496,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 6613
    },
    {
      "epoch": 0.06614,
      "grad_norm": 0.6649584285769843,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 6614
    },
    {
      "epoch": 0.06615,
      "grad_norm": 0.62805829678068,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 6615
    },
    {
      "epoch": 0.06616,
      "grad_norm": 0.5438812377408372,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 6616
    },
    {
      "epoch": 0.06617,
      "grad_norm": 0.5527092065688209,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 6617
    },
    {
      "epoch": 0.06618,
      "grad_norm": 0.6122283269442648,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 6618
    },
    {
      "epoch": 0.06619,
      "grad_norm": 0.742890868164231,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 6619
    },
    {
      "epoch": 0.0662,
      "grad_norm": 0.7283542861441105,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 6620
    },
    {
      "epoch": 0.06621,
      "grad_norm": 0.5718678667246628,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 6621
    },
    {
      "epoch": 0.06622,
      "grad_norm": 0.5903274463487566,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 6622
    },
    {
      "epoch": 0.06623,
      "grad_norm": 0.6512429289105603,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 6623
    },
    {
      "epoch": 0.06624,
      "grad_norm": 0.6598637861946588,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 6624
    },
    {
      "epoch": 0.06625,
      "grad_norm": 0.6856917785450871,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6625
    },
    {
      "epoch": 0.06626,
      "grad_norm": 0.6780772216393701,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 6626
    },
    {
      "epoch": 0.06627,
      "grad_norm": 0.6760700441536392,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 6627
    },
    {
      "epoch": 0.06628,
      "grad_norm": 0.6093236296424669,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 6628
    },
    {
      "epoch": 0.06629,
      "grad_norm": 0.54950173826257,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 6629
    },
    {
      "epoch": 0.0663,
      "grad_norm": 0.5194281298880598,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 6630
    },
    {
      "epoch": 0.06631,
      "grad_norm": 0.4311499243556024,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 6631
    },
    {
      "epoch": 0.06632,
      "grad_norm": 0.4475581432730069,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 6632
    },
    {
      "epoch": 0.06633,
      "grad_norm": 0.5189111671509972,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 6633
    },
    {
      "epoch": 0.06634,
      "grad_norm": 0.6506213384335201,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 6634
    },
    {
      "epoch": 0.06635,
      "grad_norm": 0.951821788755601,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 6635
    },
    {
      "epoch": 0.06636,
      "grad_norm": 1.1753961378034505,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 6636
    },
    {
      "epoch": 0.06637,
      "grad_norm": 0.8274131958669491,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 6637
    },
    {
      "epoch": 0.06638,
      "grad_norm": 0.9087736611088357,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 6638
    },
    {
      "epoch": 0.06639,
      "grad_norm": 0.8594515515830997,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 6639
    },
    {
      "epoch": 0.0664,
      "grad_norm": 0.7981389214332112,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 6640
    },
    {
      "epoch": 0.06641,
      "grad_norm": 0.7337294828850577,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 6641
    },
    {
      "epoch": 0.06642,
      "grad_norm": 0.6533577770906497,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 6642
    },
    {
      "epoch": 0.06643,
      "grad_norm": 0.7837625761612944,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 6643
    },
    {
      "epoch": 0.06644,
      "grad_norm": 0.7304936419918611,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 6644
    },
    {
      "epoch": 0.06645,
      "grad_norm": 0.7132250314025796,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 6645
    },
    {
      "epoch": 0.06646,
      "grad_norm": 0.7488926225230269,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 6646
    },
    {
      "epoch": 0.06647,
      "grad_norm": 0.793819493883866,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 6647
    },
    {
      "epoch": 0.06648,
      "grad_norm": 0.8562701883893814,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 6648
    },
    {
      "epoch": 0.06649,
      "grad_norm": 0.8550813838279994,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 6649
    },
    {
      "epoch": 0.0665,
      "grad_norm": 0.921797160027487,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 6650
    },
    {
      "epoch": 0.06651,
      "grad_norm": 0.8818469027010972,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 6651
    },
    {
      "epoch": 0.06652,
      "grad_norm": 0.8146093740190103,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 6652
    },
    {
      "epoch": 0.06653,
      "grad_norm": 0.7944236365667351,
      "learning_rate": 0.003,
      "loss": 4.1526,
      "step": 6653
    },
    {
      "epoch": 0.06654,
      "grad_norm": 0.8268460125317942,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 6654
    },
    {
      "epoch": 0.06655,
      "grad_norm": 0.8093935411338872,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 6655
    },
    {
      "epoch": 0.06656,
      "grad_norm": 0.7794404155644764,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 6656
    },
    {
      "epoch": 0.06657,
      "grad_norm": 0.867714800249613,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 6657
    },
    {
      "epoch": 0.06658,
      "grad_norm": 0.9142624375781937,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 6658
    },
    {
      "epoch": 0.06659,
      "grad_norm": 0.9521474355283531,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 6659
    },
    {
      "epoch": 0.0666,
      "grad_norm": 0.9916432590288042,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 6660
    },
    {
      "epoch": 0.06661,
      "grad_norm": 0.7955314202862873,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 6661
    },
    {
      "epoch": 0.06662,
      "grad_norm": 0.610721250942562,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 6662
    },
    {
      "epoch": 0.06663,
      "grad_norm": 0.6122676305957936,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 6663
    },
    {
      "epoch": 0.06664,
      "grad_norm": 0.5620102202705014,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 6664
    },
    {
      "epoch": 0.06665,
      "grad_norm": 0.5333334036557247,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 6665
    },
    {
      "epoch": 0.06666,
      "grad_norm": 0.4779105794437428,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 6666
    },
    {
      "epoch": 0.06667,
      "grad_norm": 0.4623720974757374,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 6667
    },
    {
      "epoch": 0.06668,
      "grad_norm": 0.43970305974094764,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 6668
    },
    {
      "epoch": 0.06669,
      "grad_norm": 0.4262049724870593,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6669
    },
    {
      "epoch": 0.0667,
      "grad_norm": 0.43473058147660226,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 6670
    },
    {
      "epoch": 0.06671,
      "grad_norm": 0.4811373905144156,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 6671
    },
    {
      "epoch": 0.06672,
      "grad_norm": 0.48294732369579685,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 6672
    },
    {
      "epoch": 0.06673,
      "grad_norm": 0.43487438428595204,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 6673
    },
    {
      "epoch": 0.06674,
      "grad_norm": 0.42810600796813436,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 6674
    },
    {
      "epoch": 0.06675,
      "grad_norm": 0.4187127469658615,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 6675
    },
    {
      "epoch": 0.06676,
      "grad_norm": 0.40648645104736636,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 6676
    },
    {
      "epoch": 0.06677,
      "grad_norm": 0.45239784588879955,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 6677
    },
    {
      "epoch": 0.06678,
      "grad_norm": 0.507095938002501,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 6678
    },
    {
      "epoch": 0.06679,
      "grad_norm": 0.5982462726379839,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 6679
    },
    {
      "epoch": 0.0668,
      "grad_norm": 0.7085661257950622,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 6680
    },
    {
      "epoch": 0.06681,
      "grad_norm": 0.7670644647417465,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 6681
    },
    {
      "epoch": 0.06682,
      "grad_norm": 0.666267524204535,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 6682
    },
    {
      "epoch": 0.06683,
      "grad_norm": 0.5601369401979518,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 6683
    },
    {
      "epoch": 0.06684,
      "grad_norm": 0.6101758101844771,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 6684
    },
    {
      "epoch": 0.06685,
      "grad_norm": 0.7821140255267783,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 6685
    },
    {
      "epoch": 0.06686,
      "grad_norm": 0.870444885006984,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 6686
    },
    {
      "epoch": 0.06687,
      "grad_norm": 0.885507452709397,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 6687
    },
    {
      "epoch": 0.06688,
      "grad_norm": 0.7868701754781342,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 6688
    },
    {
      "epoch": 0.06689,
      "grad_norm": 0.8462898511590071,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 6689
    },
    {
      "epoch": 0.0669,
      "grad_norm": 0.7964646938445537,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 6690
    },
    {
      "epoch": 0.06691,
      "grad_norm": 0.7632100671298492,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 6691
    },
    {
      "epoch": 0.06692,
      "grad_norm": 0.7875561157456419,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 6692
    },
    {
      "epoch": 0.06693,
      "grad_norm": 0.8456951569948422,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 6693
    },
    {
      "epoch": 0.06694,
      "grad_norm": 0.8930574483409586,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 6694
    },
    {
      "epoch": 0.06695,
      "grad_norm": 0.7514288949015006,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 6695
    },
    {
      "epoch": 0.06696,
      "grad_norm": 0.7702907982422175,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 6696
    },
    {
      "epoch": 0.06697,
      "grad_norm": 0.736140406934018,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 6697
    },
    {
      "epoch": 0.06698,
      "grad_norm": 0.7509814571017203,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 6698
    },
    {
      "epoch": 0.06699,
      "grad_norm": 0.6631263407471134,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 6699
    },
    {
      "epoch": 0.067,
      "grad_norm": 0.5871001542490745,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 6700
    },
    {
      "epoch": 0.06701,
      "grad_norm": 0.5435734122195582,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 6701
    },
    {
      "epoch": 0.06702,
      "grad_norm": 0.5252277352718284,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 6702
    },
    {
      "epoch": 0.06703,
      "grad_norm": 0.6019201583785915,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 6703
    },
    {
      "epoch": 0.06704,
      "grad_norm": 0.6629860196845838,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 6704
    },
    {
      "epoch": 0.06705,
      "grad_norm": 0.7076318403084291,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 6705
    },
    {
      "epoch": 0.06706,
      "grad_norm": 0.7718630424234669,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 6706
    },
    {
      "epoch": 0.06707,
      "grad_norm": 0.735500368867454,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 6707
    },
    {
      "epoch": 0.06708,
      "grad_norm": 0.6126034855042024,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 6708
    },
    {
      "epoch": 0.06709,
      "grad_norm": 0.5356173594265632,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6709
    },
    {
      "epoch": 0.0671,
      "grad_norm": 0.4923039482946903,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6710
    },
    {
      "epoch": 0.06711,
      "grad_norm": 0.496437675576904,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 6711
    },
    {
      "epoch": 0.06712,
      "grad_norm": 0.6267076354025424,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 6712
    },
    {
      "epoch": 0.06713,
      "grad_norm": 0.810419464520027,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 6713
    },
    {
      "epoch": 0.06714,
      "grad_norm": 1.0033614640972555,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 6714
    },
    {
      "epoch": 0.06715,
      "grad_norm": 1.034862211372409,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 6715
    },
    {
      "epoch": 0.06716,
      "grad_norm": 0.7891873870335114,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 6716
    },
    {
      "epoch": 0.06717,
      "grad_norm": 0.7583795710672536,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 6717
    },
    {
      "epoch": 0.06718,
      "grad_norm": 0.7115916620356093,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 6718
    },
    {
      "epoch": 0.06719,
      "grad_norm": 0.6809781091009515,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 6719
    },
    {
      "epoch": 0.0672,
      "grad_norm": 0.7352768877036369,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 6720
    },
    {
      "epoch": 0.06721,
      "grad_norm": 0.6315118398036524,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 6721
    },
    {
      "epoch": 0.06722,
      "grad_norm": 0.6185480236265406,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 6722
    },
    {
      "epoch": 0.06723,
      "grad_norm": 0.5845982684430056,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 6723
    },
    {
      "epoch": 0.06724,
      "grad_norm": 0.49159392452345974,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 6724
    },
    {
      "epoch": 0.06725,
      "grad_norm": 0.4513358433552748,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 6725
    },
    {
      "epoch": 0.06726,
      "grad_norm": 0.4366793957137453,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 6726
    },
    {
      "epoch": 0.06727,
      "grad_norm": 0.45039616788114,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 6727
    },
    {
      "epoch": 0.06728,
      "grad_norm": 0.42960667974723327,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 6728
    },
    {
      "epoch": 0.06729,
      "grad_norm": 0.3936495109654878,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 6729
    },
    {
      "epoch": 0.0673,
      "grad_norm": 0.4004584193437624,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 6730
    },
    {
      "epoch": 0.06731,
      "grad_norm": 0.43331873995398884,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 6731
    },
    {
      "epoch": 0.06732,
      "grad_norm": 0.5033918683059013,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 6732
    },
    {
      "epoch": 0.06733,
      "grad_norm": 0.5679439839709435,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 6733
    },
    {
      "epoch": 0.06734,
      "grad_norm": 0.6741835879416314,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 6734
    },
    {
      "epoch": 0.06735,
      "grad_norm": 0.7951013050676844,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 6735
    },
    {
      "epoch": 0.06736,
      "grad_norm": 0.8142625336131883,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 6736
    },
    {
      "epoch": 0.06737,
      "grad_norm": 0.6923418050164777,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 6737
    },
    {
      "epoch": 0.06738,
      "grad_norm": 0.6417437188150639,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 6738
    },
    {
      "epoch": 0.06739,
      "grad_norm": 0.6666846912447963,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 6739
    },
    {
      "epoch": 0.0674,
      "grad_norm": 0.7113245183023263,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 6740
    },
    {
      "epoch": 0.06741,
      "grad_norm": 0.696453096464101,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 6741
    },
    {
      "epoch": 0.06742,
      "grad_norm": 0.6475796676339262,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 6742
    },
    {
      "epoch": 0.06743,
      "grad_norm": 0.6335412186431436,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 6743
    },
    {
      "epoch": 0.06744,
      "grad_norm": 0.5563393921556343,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 6744
    },
    {
      "epoch": 0.06745,
      "grad_norm": 0.5419204393939647,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 6745
    },
    {
      "epoch": 0.06746,
      "grad_norm": 0.6275067206076368,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 6746
    },
    {
      "epoch": 0.06747,
      "grad_norm": 0.6946085858510286,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 6747
    },
    {
      "epoch": 0.06748,
      "grad_norm": 0.7589007060444024,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 6748
    },
    {
      "epoch": 0.06749,
      "grad_norm": 0.8157554526066001,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 6749
    },
    {
      "epoch": 0.0675,
      "grad_norm": 0.7942374978397424,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 6750
    },
    {
      "epoch": 0.06751,
      "grad_norm": 0.8980523962537803,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 6751
    },
    {
      "epoch": 0.06752,
      "grad_norm": 1.0269645356154302,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 6752
    },
    {
      "epoch": 0.06753,
      "grad_norm": 0.9257798372553527,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 6753
    },
    {
      "epoch": 0.06754,
      "grad_norm": 0.7843345022720416,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 6754
    },
    {
      "epoch": 0.06755,
      "grad_norm": 0.8626796100641331,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 6755
    },
    {
      "epoch": 0.06756,
      "grad_norm": 0.813278917534869,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 6756
    },
    {
      "epoch": 0.06757,
      "grad_norm": 0.7512724737472484,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 6757
    },
    {
      "epoch": 0.06758,
      "grad_norm": 0.6469087254992117,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 6758
    },
    {
      "epoch": 0.06759,
      "grad_norm": 0.6436740764232052,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 6759
    },
    {
      "epoch": 0.0676,
      "grad_norm": 0.6943244361400559,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 6760
    },
    {
      "epoch": 0.06761,
      "grad_norm": 0.6830986749187081,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 6761
    },
    {
      "epoch": 0.06762,
      "grad_norm": 0.6781853474637783,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 6762
    },
    {
      "epoch": 0.06763,
      "grad_norm": 0.6506611393135219,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 6763
    },
    {
      "epoch": 0.06764,
      "grad_norm": 0.7027611096936427,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 6764
    },
    {
      "epoch": 0.06765,
      "grad_norm": 0.6155749614693148,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 6765
    },
    {
      "epoch": 0.06766,
      "grad_norm": 0.6035208063328635,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 6766
    },
    {
      "epoch": 0.06767,
      "grad_norm": 0.5420944911927922,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 6767
    },
    {
      "epoch": 0.06768,
      "grad_norm": 0.6137703731544314,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 6768
    },
    {
      "epoch": 0.06769,
      "grad_norm": 0.7672588192354372,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 6769
    },
    {
      "epoch": 0.0677,
      "grad_norm": 0.8517644016624661,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 6770
    },
    {
      "epoch": 0.06771,
      "grad_norm": 0.8225325274897667,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 6771
    },
    {
      "epoch": 0.06772,
      "grad_norm": 0.8285914075732632,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 6772
    },
    {
      "epoch": 0.06773,
      "grad_norm": 0.7479732791491223,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 6773
    },
    {
      "epoch": 0.06774,
      "grad_norm": 0.7939299101707836,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 6774
    },
    {
      "epoch": 0.06775,
      "grad_norm": 0.8020629017874903,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 6775
    },
    {
      "epoch": 0.06776,
      "grad_norm": 0.7932490440340019,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 6776
    },
    {
      "epoch": 0.06777,
      "grad_norm": 0.794180803890291,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 6777
    },
    {
      "epoch": 0.06778,
      "grad_norm": 0.7991189828810096,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 6778
    },
    {
      "epoch": 0.06779,
      "grad_norm": 0.818786154364808,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 6779
    },
    {
      "epoch": 0.0678,
      "grad_norm": 0.712453918318275,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 6780
    },
    {
      "epoch": 0.06781,
      "grad_norm": 0.6542310526187651,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 6781
    },
    {
      "epoch": 0.06782,
      "grad_norm": 0.6334297545638243,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 6782
    },
    {
      "epoch": 0.06783,
      "grad_norm": 0.6222166600018698,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 6783
    },
    {
      "epoch": 0.06784,
      "grad_norm": 0.6173988646215278,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 6784
    },
    {
      "epoch": 0.06785,
      "grad_norm": 0.6089756191279135,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 6785
    },
    {
      "epoch": 0.06786,
      "grad_norm": 0.5533280430635569,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 6786
    },
    {
      "epoch": 0.06787,
      "grad_norm": 0.5466977622748588,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 6787
    },
    {
      "epoch": 0.06788,
      "grad_norm": 0.5709118469585935,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 6788
    },
    {
      "epoch": 0.06789,
      "grad_norm": 0.43615212913657647,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 6789
    },
    {
      "epoch": 0.0679,
      "grad_norm": 0.4445553735594529,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 6790
    },
    {
      "epoch": 0.06791,
      "grad_norm": 0.4404711279990138,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 6791
    },
    {
      "epoch": 0.06792,
      "grad_norm": 0.46477638175762537,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6792
    },
    {
      "epoch": 0.06793,
      "grad_norm": 0.48722297592983815,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 6793
    },
    {
      "epoch": 0.06794,
      "grad_norm": 0.4979919688575191,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 6794
    },
    {
      "epoch": 0.06795,
      "grad_norm": 0.49093393128752905,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 6795
    },
    {
      "epoch": 0.06796,
      "grad_norm": 0.5892149599358569,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 6796
    },
    {
      "epoch": 0.06797,
      "grad_norm": 0.7982288521309417,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 6797
    },
    {
      "epoch": 0.06798,
      "grad_norm": 0.9112336477160669,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 6798
    },
    {
      "epoch": 0.06799,
      "grad_norm": 0.8816688090009441,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 6799
    },
    {
      "epoch": 0.068,
      "grad_norm": 0.8279620352251622,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 6800
    },
    {
      "epoch": 0.06801,
      "grad_norm": 0.8613030262720641,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 6801
    },
    {
      "epoch": 0.06802,
      "grad_norm": 0.9654181970654568,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 6802
    },
    {
      "epoch": 0.06803,
      "grad_norm": 0.7831222168974614,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 6803
    },
    {
      "epoch": 0.06804,
      "grad_norm": 0.7859386559785015,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 6804
    },
    {
      "epoch": 0.06805,
      "grad_norm": 0.7680250323768286,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6805
    },
    {
      "epoch": 0.06806,
      "grad_norm": 0.7789471170116108,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 6806
    },
    {
      "epoch": 0.06807,
      "grad_norm": 0.6389939368418367,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 6807
    },
    {
      "epoch": 0.06808,
      "grad_norm": 0.708633773967604,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 6808
    },
    {
      "epoch": 0.06809,
      "grad_norm": 0.7180318843790542,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 6809
    },
    {
      "epoch": 0.0681,
      "grad_norm": 0.7301766381754865,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 6810
    },
    {
      "epoch": 0.06811,
      "grad_norm": 0.7471263469142446,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 6811
    },
    {
      "epoch": 0.06812,
      "grad_norm": 0.7142209108867097,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 6812
    },
    {
      "epoch": 0.06813,
      "grad_norm": 0.6372769592641551,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 6813
    },
    {
      "epoch": 0.06814,
      "grad_norm": 0.6379955586156076,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6814
    },
    {
      "epoch": 0.06815,
      "grad_norm": 0.6038210649447708,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 6815
    },
    {
      "epoch": 0.06816,
      "grad_norm": 0.5751849767984581,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 6816
    },
    {
      "epoch": 0.06817,
      "grad_norm": 0.5359776355709539,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 6817
    },
    {
      "epoch": 0.06818,
      "grad_norm": 0.5146004463623628,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 6818
    },
    {
      "epoch": 0.06819,
      "grad_norm": 0.5178577295105196,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 6819
    },
    {
      "epoch": 0.0682,
      "grad_norm": 0.5036407285943472,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 6820
    },
    {
      "epoch": 0.06821,
      "grad_norm": 0.5531066591245212,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 6821
    },
    {
      "epoch": 0.06822,
      "grad_norm": 0.6470177042281421,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 6822
    },
    {
      "epoch": 0.06823,
      "grad_norm": 0.6539637525535354,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 6823
    },
    {
      "epoch": 0.06824,
      "grad_norm": 0.7493448106054319,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 6824
    },
    {
      "epoch": 0.06825,
      "grad_norm": 0.7477095595533654,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 6825
    },
    {
      "epoch": 0.06826,
      "grad_norm": 0.616205641208411,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 6826
    },
    {
      "epoch": 0.06827,
      "grad_norm": 0.5343803407507347,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 6827
    },
    {
      "epoch": 0.06828,
      "grad_norm": 0.5516727566441831,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 6828
    },
    {
      "epoch": 0.06829,
      "grad_norm": 0.5678645003238811,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 6829
    },
    {
      "epoch": 0.0683,
      "grad_norm": 0.5736945535959624,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 6830
    },
    {
      "epoch": 0.06831,
      "grad_norm": 0.5766186753999193,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 6831
    },
    {
      "epoch": 0.06832,
      "grad_norm": 0.5412637917693169,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 6832
    },
    {
      "epoch": 0.06833,
      "grad_norm": 0.5354756199374949,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 6833
    },
    {
      "epoch": 0.06834,
      "grad_norm": 0.55758586869952,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 6834
    },
    {
      "epoch": 0.06835,
      "grad_norm": 0.5851752358292377,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 6835
    },
    {
      "epoch": 0.06836,
      "grad_norm": 0.5357594576872605,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 6836
    },
    {
      "epoch": 0.06837,
      "grad_norm": 0.5493992586473802,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 6837
    },
    {
      "epoch": 0.06838,
      "grad_norm": 0.6237990486728586,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 6838
    },
    {
      "epoch": 0.06839,
      "grad_norm": 0.9418401714608854,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 6839
    },
    {
      "epoch": 0.0684,
      "grad_norm": 1.3064642474920947,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 6840
    },
    {
      "epoch": 0.06841,
      "grad_norm": 0.6597660684920605,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 6841
    },
    {
      "epoch": 0.06842,
      "grad_norm": 0.6489661413957194,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 6842
    },
    {
      "epoch": 0.06843,
      "grad_norm": 0.6773408231334266,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 6843
    },
    {
      "epoch": 0.06844,
      "grad_norm": 0.6589160490897324,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 6844
    },
    {
      "epoch": 0.06845,
      "grad_norm": 0.5891507948339562,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 6845
    },
    {
      "epoch": 0.06846,
      "grad_norm": 0.5999658896363759,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 6846
    },
    {
      "epoch": 0.06847,
      "grad_norm": 0.5887881579441098,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 6847
    },
    {
      "epoch": 0.06848,
      "grad_norm": 0.5649925957595652,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 6848
    },
    {
      "epoch": 0.06849,
      "grad_norm": 0.5705197551609986,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 6849
    },
    {
      "epoch": 0.0685,
      "grad_norm": 0.5952753055541125,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 6850
    },
    {
      "epoch": 0.06851,
      "grad_norm": 0.6213756923220035,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 6851
    },
    {
      "epoch": 0.06852,
      "grad_norm": 0.6907364246047624,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 6852
    },
    {
      "epoch": 0.06853,
      "grad_norm": 0.6969902001664416,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 6853
    },
    {
      "epoch": 0.06854,
      "grad_norm": 0.662173631889165,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 6854
    },
    {
      "epoch": 0.06855,
      "grad_norm": 0.7232071834064421,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 6855
    },
    {
      "epoch": 0.06856,
      "grad_norm": 0.7279714519404217,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 6856
    },
    {
      "epoch": 0.06857,
      "grad_norm": 0.7721813680083663,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 6857
    },
    {
      "epoch": 0.06858,
      "grad_norm": 0.7425927331139939,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 6858
    },
    {
      "epoch": 0.06859,
      "grad_norm": 0.6721982817507133,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 6859
    },
    {
      "epoch": 0.0686,
      "grad_norm": 0.5774934823673541,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 6860
    },
    {
      "epoch": 0.06861,
      "grad_norm": 0.6138735149136986,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 6861
    },
    {
      "epoch": 0.06862,
      "grad_norm": 0.6955887372108659,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 6862
    },
    {
      "epoch": 0.06863,
      "grad_norm": 0.7889995228865956,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 6863
    },
    {
      "epoch": 0.06864,
      "grad_norm": 0.8319268473846568,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 6864
    },
    {
      "epoch": 0.06865,
      "grad_norm": 0.7332088452187622,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 6865
    },
    {
      "epoch": 0.06866,
      "grad_norm": 0.7569264658828823,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 6866
    },
    {
      "epoch": 0.06867,
      "grad_norm": 0.7482157424829708,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 6867
    },
    {
      "epoch": 0.06868,
      "grad_norm": 0.8067432575353655,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 6868
    },
    {
      "epoch": 0.06869,
      "grad_norm": 0.795724395615385,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 6869
    },
    {
      "epoch": 0.0687,
      "grad_norm": 0.7990167361942389,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 6870
    },
    {
      "epoch": 0.06871,
      "grad_norm": 0.751687574887691,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 6871
    },
    {
      "epoch": 0.06872,
      "grad_norm": 0.7941670622571882,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 6872
    },
    {
      "epoch": 0.06873,
      "grad_norm": 0.818073654767055,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 6873
    },
    {
      "epoch": 0.06874,
      "grad_norm": 0.8810084565313016,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 6874
    },
    {
      "epoch": 0.06875,
      "grad_norm": 0.9712435694596839,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 6875
    },
    {
      "epoch": 0.06876,
      "grad_norm": 1.1759386697440901,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 6876
    },
    {
      "epoch": 0.06877,
      "grad_norm": 0.8914085065829237,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 6877
    },
    {
      "epoch": 0.06878,
      "grad_norm": 0.6941362711827244,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 6878
    },
    {
      "epoch": 0.06879,
      "grad_norm": 0.6928015779389758,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 6879
    },
    {
      "epoch": 0.0688,
      "grad_norm": 0.738782255552787,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 6880
    },
    {
      "epoch": 0.06881,
      "grad_norm": 0.6910853720919746,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 6881
    },
    {
      "epoch": 0.06882,
      "grad_norm": 0.7254038162214872,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 6882
    },
    {
      "epoch": 0.06883,
      "grad_norm": 0.8207902946108598,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 6883
    },
    {
      "epoch": 0.06884,
      "grad_norm": 0.7716553726102112,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 6884
    },
    {
      "epoch": 0.06885,
      "grad_norm": 0.6958687421650827,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 6885
    },
    {
      "epoch": 0.06886,
      "grad_norm": 0.5821526100113974,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 6886
    },
    {
      "epoch": 0.06887,
      "grad_norm": 0.5981430384947571,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 6887
    },
    {
      "epoch": 0.06888,
      "grad_norm": 0.643772790056028,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 6888
    },
    {
      "epoch": 0.06889,
      "grad_norm": 0.6944406405800507,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 6889
    },
    {
      "epoch": 0.0689,
      "grad_norm": 0.6385881084418807,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 6890
    },
    {
      "epoch": 0.06891,
      "grad_norm": 0.6086144164229688,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 6891
    },
    {
      "epoch": 0.06892,
      "grad_norm": 0.6012480896266729,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 6892
    },
    {
      "epoch": 0.06893,
      "grad_norm": 0.561071573044732,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 6893
    },
    {
      "epoch": 0.06894,
      "grad_norm": 0.48860461588730103,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 6894
    },
    {
      "epoch": 0.06895,
      "grad_norm": 0.5246990023618513,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 6895
    },
    {
      "epoch": 0.06896,
      "grad_norm": 0.5887964150643885,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 6896
    },
    {
      "epoch": 0.06897,
      "grad_norm": 0.5241278786378765,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 6897
    },
    {
      "epoch": 0.06898,
      "grad_norm": 0.5243694688992354,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 6898
    },
    {
      "epoch": 0.06899,
      "grad_norm": 0.5729571128070601,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 6899
    },
    {
      "epoch": 0.069,
      "grad_norm": 0.5895390705006303,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 6900
    },
    {
      "epoch": 0.06901,
      "grad_norm": 0.6952354448422888,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 6901
    },
    {
      "epoch": 0.06902,
      "grad_norm": 0.6648257134889515,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 6902
    },
    {
      "epoch": 0.06903,
      "grad_norm": 0.6781063401244306,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 6903
    },
    {
      "epoch": 0.06904,
      "grad_norm": 0.6750907763884256,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 6904
    },
    {
      "epoch": 0.06905,
      "grad_norm": 0.701961664699957,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 6905
    },
    {
      "epoch": 0.06906,
      "grad_norm": 0.6469621356429142,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 6906
    },
    {
      "epoch": 0.06907,
      "grad_norm": 0.5765290038257871,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 6907
    },
    {
      "epoch": 0.06908,
      "grad_norm": 0.6197899283242572,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 6908
    },
    {
      "epoch": 0.06909,
      "grad_norm": 0.6933557799843862,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 6909
    },
    {
      "epoch": 0.0691,
      "grad_norm": 0.7562290219169062,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 6910
    },
    {
      "epoch": 0.06911,
      "grad_norm": 0.7734263332293689,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 6911
    },
    {
      "epoch": 0.06912,
      "grad_norm": 0.8992983547552829,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 6912
    },
    {
      "epoch": 0.06913,
      "grad_norm": 1.0208555338096066,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 6913
    },
    {
      "epoch": 0.06914,
      "grad_norm": 0.9933731989818703,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 6914
    },
    {
      "epoch": 0.06915,
      "grad_norm": 1.142744843357704,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 6915
    },
    {
      "epoch": 0.06916,
      "grad_norm": 0.671739611249756,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 6916
    },
    {
      "epoch": 0.06917,
      "grad_norm": 0.7866141531687029,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 6917
    },
    {
      "epoch": 0.06918,
      "grad_norm": 0.9611567239474412,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 6918
    },
    {
      "epoch": 0.06919,
      "grad_norm": 0.9336705733194112,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 6919
    },
    {
      "epoch": 0.0692,
      "grad_norm": 0.8550272945447894,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 6920
    },
    {
      "epoch": 0.06921,
      "grad_norm": 0.8255769935938679,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 6921
    },
    {
      "epoch": 0.06922,
      "grad_norm": 0.7887735703255148,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 6922
    },
    {
      "epoch": 0.06923,
      "grad_norm": 0.7586591684231498,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 6923
    },
    {
      "epoch": 0.06924,
      "grad_norm": 0.833512973215651,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 6924
    },
    {
      "epoch": 0.06925,
      "grad_norm": 0.8413623073932563,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 6925
    },
    {
      "epoch": 0.06926,
      "grad_norm": 0.7024551972837352,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 6926
    },
    {
      "epoch": 0.06927,
      "grad_norm": 0.5829899478703182,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 6927
    },
    {
      "epoch": 0.06928,
      "grad_norm": 0.5794264624289892,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 6928
    },
    {
      "epoch": 0.06929,
      "grad_norm": 0.5549181816114577,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6929
    },
    {
      "epoch": 0.0693,
      "grad_norm": 0.5430801494911696,
      "learning_rate": 0.003,
      "loss": 4.1566,
      "step": 6930
    },
    {
      "epoch": 0.06931,
      "grad_norm": 0.47027424052399663,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 6931
    },
    {
      "epoch": 0.06932,
      "grad_norm": 0.4173009048479236,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 6932
    },
    {
      "epoch": 0.06933,
      "grad_norm": 0.40150485804104163,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 6933
    },
    {
      "epoch": 0.06934,
      "grad_norm": 0.3862168071563836,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 6934
    },
    {
      "epoch": 0.06935,
      "grad_norm": 0.446771899060793,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 6935
    },
    {
      "epoch": 0.06936,
      "grad_norm": 0.5004501018374794,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 6936
    },
    {
      "epoch": 0.06937,
      "grad_norm": 0.5488151795943027,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 6937
    },
    {
      "epoch": 0.06938,
      "grad_norm": 0.5770212398497512,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 6938
    },
    {
      "epoch": 0.06939,
      "grad_norm": 0.558653999723117,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 6939
    },
    {
      "epoch": 0.0694,
      "grad_norm": 0.5466069373357318,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 6940
    },
    {
      "epoch": 0.06941,
      "grad_norm": 0.49853177456508263,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 6941
    },
    {
      "epoch": 0.06942,
      "grad_norm": 0.506244292367137,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 6942
    },
    {
      "epoch": 0.06943,
      "grad_norm": 0.49638439608831303,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 6943
    },
    {
      "epoch": 0.06944,
      "grad_norm": 0.5489964428142321,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 6944
    },
    {
      "epoch": 0.06945,
      "grad_norm": 0.7004086562916291,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 6945
    },
    {
      "epoch": 0.06946,
      "grad_norm": 0.8618988339111132,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 6946
    },
    {
      "epoch": 0.06947,
      "grad_norm": 0.8898968651463349,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 6947
    },
    {
      "epoch": 0.06948,
      "grad_norm": 0.8891675654883304,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6948
    },
    {
      "epoch": 0.06949,
      "grad_norm": 0.8415925076980938,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 6949
    },
    {
      "epoch": 0.0695,
      "grad_norm": 0.8358314389372723,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 6950
    },
    {
      "epoch": 0.06951,
      "grad_norm": 0.8991779236158414,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 6951
    },
    {
      "epoch": 0.06952,
      "grad_norm": 0.9143145994663924,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 6952
    },
    {
      "epoch": 0.06953,
      "grad_norm": 0.8582198172305813,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 6953
    },
    {
      "epoch": 0.06954,
      "grad_norm": 0.8532675006476708,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 6954
    },
    {
      "epoch": 0.06955,
      "grad_norm": 0.8361880964205014,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 6955
    },
    {
      "epoch": 0.06956,
      "grad_norm": 0.7160992886820927,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 6956
    },
    {
      "epoch": 0.06957,
      "grad_norm": 0.6572004040022867,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 6957
    },
    {
      "epoch": 0.06958,
      "grad_norm": 0.6687661358163699,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 6958
    },
    {
      "epoch": 0.06959,
      "grad_norm": 0.6817550869048531,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 6959
    },
    {
      "epoch": 0.0696,
      "grad_norm": 0.7535821582695273,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 6960
    },
    {
      "epoch": 0.06961,
      "grad_norm": 0.8825707617170178,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 6961
    },
    {
      "epoch": 0.06962,
      "grad_norm": 0.8710338520388163,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 6962
    },
    {
      "epoch": 0.06963,
      "grad_norm": 0.816650176387816,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 6963
    },
    {
      "epoch": 0.06964,
      "grad_norm": 0.7260316746191819,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 6964
    },
    {
      "epoch": 0.06965,
      "grad_norm": 0.6648041497655378,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 6965
    },
    {
      "epoch": 0.06966,
      "grad_norm": 0.6784510161564018,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 6966
    },
    {
      "epoch": 0.06967,
      "grad_norm": 0.6452607728019111,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 6967
    },
    {
      "epoch": 0.06968,
      "grad_norm": 0.6001265248194674,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 6968
    },
    {
      "epoch": 0.06969,
      "grad_norm": 0.5934642265206468,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 6969
    },
    {
      "epoch": 0.0697,
      "grad_norm": 0.5831037045271329,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 6970
    },
    {
      "epoch": 0.06971,
      "grad_norm": 0.6199075375261113,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 6971
    },
    {
      "epoch": 0.06972,
      "grad_norm": 0.6436556062187045,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 6972
    },
    {
      "epoch": 0.06973,
      "grad_norm": 0.5645264329417377,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 6973
    },
    {
      "epoch": 0.06974,
      "grad_norm": 0.6421089807938115,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 6974
    },
    {
      "epoch": 0.06975,
      "grad_norm": 0.7540210531000154,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 6975
    },
    {
      "epoch": 0.06976,
      "grad_norm": 0.7638332625980192,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 6976
    },
    {
      "epoch": 0.06977,
      "grad_norm": 0.8335885386349529,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 6977
    },
    {
      "epoch": 0.06978,
      "grad_norm": 0.8006226145225701,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 6978
    },
    {
      "epoch": 0.06979,
      "grad_norm": 0.7348797862910301,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 6979
    },
    {
      "epoch": 0.0698,
      "grad_norm": 0.8010250875166668,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 6980
    },
    {
      "epoch": 0.06981,
      "grad_norm": 0.7496972705067594,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 6981
    },
    {
      "epoch": 0.06982,
      "grad_norm": 0.7126647429947338,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 6982
    },
    {
      "epoch": 0.06983,
      "grad_norm": 0.676360984622482,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 6983
    },
    {
      "epoch": 0.06984,
      "grad_norm": 0.6167843107245039,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 6984
    },
    {
      "epoch": 0.06985,
      "grad_norm": 0.5226662640709087,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 6985
    },
    {
      "epoch": 0.06986,
      "grad_norm": 0.4870800595121638,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6986
    },
    {
      "epoch": 0.06987,
      "grad_norm": 0.4732107972316138,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 6987
    },
    {
      "epoch": 0.06988,
      "grad_norm": 0.5060749910774011,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 6988
    },
    {
      "epoch": 0.06989,
      "grad_norm": 0.6140987263407802,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 6989
    },
    {
      "epoch": 0.0699,
      "grad_norm": 0.6629504376964789,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 6990
    },
    {
      "epoch": 0.06991,
      "grad_norm": 0.7586325106957013,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 6991
    },
    {
      "epoch": 0.06992,
      "grad_norm": 0.6309648702561338,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 6992
    },
    {
      "epoch": 0.06993,
      "grad_norm": 0.5315699899335162,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 6993
    },
    {
      "epoch": 0.06994,
      "grad_norm": 0.48772801302005414,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 6994
    },
    {
      "epoch": 0.06995,
      "grad_norm": 0.5131657989899862,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 6995
    },
    {
      "epoch": 0.06996,
      "grad_norm": 0.5515157780044487,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 6996
    },
    {
      "epoch": 0.06997,
      "grad_norm": 0.5525209598406478,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 6997
    },
    {
      "epoch": 0.06998,
      "grad_norm": 0.479008437370693,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 6998
    },
    {
      "epoch": 0.06999,
      "grad_norm": 0.4505882666020481,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 6999
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.520401586803475,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 7000
    },
    {
      "epoch": 0.07001,
      "grad_norm": 0.5963833247005839,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 7001
    },
    {
      "epoch": 0.07002,
      "grad_norm": 0.7816965535130395,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 7002
    },
    {
      "epoch": 0.07003,
      "grad_norm": 0.8103989079782393,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 7003
    },
    {
      "epoch": 0.07004,
      "grad_norm": 0.7881315831755371,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 7004
    },
    {
      "epoch": 0.07005,
      "grad_norm": 0.8308231793794756,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 7005
    },
    {
      "epoch": 0.07006,
      "grad_norm": 0.8109007968000713,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 7006
    },
    {
      "epoch": 0.07007,
      "grad_norm": 0.6344467493229883,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 7007
    },
    {
      "epoch": 0.07008,
      "grad_norm": 0.6334215629519069,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 7008
    },
    {
      "epoch": 0.07009,
      "grad_norm": 0.6472001020552638,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 7009
    },
    {
      "epoch": 0.0701,
      "grad_norm": 0.6243314055174884,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 7010
    },
    {
      "epoch": 0.07011,
      "grad_norm": 0.6932842423649617,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 7011
    },
    {
      "epoch": 0.07012,
      "grad_norm": 0.8521696913542735,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 7012
    },
    {
      "epoch": 0.07013,
      "grad_norm": 1.0813751987049736,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 7013
    },
    {
      "epoch": 0.07014,
      "grad_norm": 0.8685320890891782,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 7014
    },
    {
      "epoch": 0.07015,
      "grad_norm": 0.714165389488949,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 7015
    },
    {
      "epoch": 0.07016,
      "grad_norm": 0.6240822477720798,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 7016
    },
    {
      "epoch": 0.07017,
      "grad_norm": 0.6300786532998714,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7017
    },
    {
      "epoch": 0.07018,
      "grad_norm": 0.6076108348038608,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7018
    },
    {
      "epoch": 0.07019,
      "grad_norm": 0.6908646612154652,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 7019
    },
    {
      "epoch": 0.0702,
      "grad_norm": 0.6690787244309745,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 7020
    },
    {
      "epoch": 0.07021,
      "grad_norm": 0.6214099299934482,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 7021
    },
    {
      "epoch": 0.07022,
      "grad_norm": 0.5406773132983848,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 7022
    },
    {
      "epoch": 0.07023,
      "grad_norm": 0.5564313613572602,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 7023
    },
    {
      "epoch": 0.07024,
      "grad_norm": 0.565839780513049,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 7024
    },
    {
      "epoch": 0.07025,
      "grad_norm": 0.6209168576950462,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 7025
    },
    {
      "epoch": 0.07026,
      "grad_norm": 0.6750589310489371,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 7026
    },
    {
      "epoch": 0.07027,
      "grad_norm": 0.683143198082187,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 7027
    },
    {
      "epoch": 0.07028,
      "grad_norm": 0.6337869888733688,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 7028
    },
    {
      "epoch": 0.07029,
      "grad_norm": 0.6399163304203913,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 7029
    },
    {
      "epoch": 0.0703,
      "grad_norm": 0.6748721551610196,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 7030
    },
    {
      "epoch": 0.07031,
      "grad_norm": 0.7321679893744895,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 7031
    },
    {
      "epoch": 0.07032,
      "grad_norm": 0.8299726001491252,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 7032
    },
    {
      "epoch": 0.07033,
      "grad_norm": 0.9060063098737187,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 7033
    },
    {
      "epoch": 0.07034,
      "grad_norm": 0.9522354378897846,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 7034
    },
    {
      "epoch": 0.07035,
      "grad_norm": 0.972269290200196,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 7035
    },
    {
      "epoch": 0.07036,
      "grad_norm": 0.924567971293714,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 7036
    },
    {
      "epoch": 0.07037,
      "grad_norm": 0.8651360365392121,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 7037
    },
    {
      "epoch": 0.07038,
      "grad_norm": 0.8060967376427206,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 7038
    },
    {
      "epoch": 0.07039,
      "grad_norm": 0.7677552044419784,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 7039
    },
    {
      "epoch": 0.0704,
      "grad_norm": 0.7934246766415343,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 7040
    },
    {
      "epoch": 0.07041,
      "grad_norm": 0.7702031563514626,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 7041
    },
    {
      "epoch": 0.07042,
      "grad_norm": 0.8294171357357263,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 7042
    },
    {
      "epoch": 0.07043,
      "grad_norm": 0.8699931370635664,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 7043
    },
    {
      "epoch": 0.07044,
      "grad_norm": 0.8662992135149028,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 7044
    },
    {
      "epoch": 0.07045,
      "grad_norm": 0.8556015774986996,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 7045
    },
    {
      "epoch": 0.07046,
      "grad_norm": 0.8195747843665967,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 7046
    },
    {
      "epoch": 0.07047,
      "grad_norm": 0.7213877816923308,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 7047
    },
    {
      "epoch": 0.07048,
      "grad_norm": 0.6149100905732247,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 7048
    },
    {
      "epoch": 0.07049,
      "grad_norm": 0.6732224266829641,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 7049
    },
    {
      "epoch": 0.0705,
      "grad_norm": 0.7014634218429036,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 7050
    },
    {
      "epoch": 0.07051,
      "grad_norm": 0.6522225830743739,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 7051
    },
    {
      "epoch": 0.07052,
      "grad_norm": 0.6500823083801979,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 7052
    },
    {
      "epoch": 0.07053,
      "grad_norm": 0.6368059007709821,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 7053
    },
    {
      "epoch": 0.07054,
      "grad_norm": 0.6937824763862059,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 7054
    },
    {
      "epoch": 0.07055,
      "grad_norm": 0.6678203074975709,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 7055
    },
    {
      "epoch": 0.07056,
      "grad_norm": 0.5999332348304974,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 7056
    },
    {
      "epoch": 0.07057,
      "grad_norm": 0.6207684817645435,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 7057
    },
    {
      "epoch": 0.07058,
      "grad_norm": 0.6334942997590186,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 7058
    },
    {
      "epoch": 0.07059,
      "grad_norm": 0.6248499550694564,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 7059
    },
    {
      "epoch": 0.0706,
      "grad_norm": 0.6690971751004321,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 7060
    },
    {
      "epoch": 0.07061,
      "grad_norm": 0.6681006456986236,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 7061
    },
    {
      "epoch": 0.07062,
      "grad_norm": 0.5873099188423571,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 7062
    },
    {
      "epoch": 0.07063,
      "grad_norm": 0.5660315602781396,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 7063
    },
    {
      "epoch": 0.07064,
      "grad_norm": 0.5354212695010325,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 7064
    },
    {
      "epoch": 0.07065,
      "grad_norm": 0.5541650096697861,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 7065
    },
    {
      "epoch": 0.07066,
      "grad_norm": 0.5737529936227163,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 7066
    },
    {
      "epoch": 0.07067,
      "grad_norm": 0.6443507476053769,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 7067
    },
    {
      "epoch": 0.07068,
      "grad_norm": 0.7079051560187819,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 7068
    },
    {
      "epoch": 0.07069,
      "grad_norm": 0.5681815389061051,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 7069
    },
    {
      "epoch": 0.0707,
      "grad_norm": 0.5054781241764719,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 7070
    },
    {
      "epoch": 0.07071,
      "grad_norm": 0.5363029539486839,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 7071
    },
    {
      "epoch": 0.07072,
      "grad_norm": 0.5662713105012626,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 7072
    },
    {
      "epoch": 0.07073,
      "grad_norm": 0.5886362212149604,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 7073
    },
    {
      "epoch": 0.07074,
      "grad_norm": 0.5383952438154733,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 7074
    },
    {
      "epoch": 0.07075,
      "grad_norm": 0.49108475293469883,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 7075
    },
    {
      "epoch": 0.07076,
      "grad_norm": 0.46684864123512826,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 7076
    },
    {
      "epoch": 0.07077,
      "grad_norm": 0.5465447445092506,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 7077
    },
    {
      "epoch": 0.07078,
      "grad_norm": 0.6015521724166992,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 7078
    },
    {
      "epoch": 0.07079,
      "grad_norm": 0.85890424119227,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 7079
    },
    {
      "epoch": 0.0708,
      "grad_norm": 1.282741314578649,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 7080
    },
    {
      "epoch": 0.07081,
      "grad_norm": 0.6908289270870185,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 7081
    },
    {
      "epoch": 0.07082,
      "grad_norm": 0.6179688734599202,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 7082
    },
    {
      "epoch": 0.07083,
      "grad_norm": 0.7995600194026101,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 7083
    },
    {
      "epoch": 0.07084,
      "grad_norm": 0.8857457001103319,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 7084
    },
    {
      "epoch": 0.07085,
      "grad_norm": 0.9143259065826967,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 7085
    },
    {
      "epoch": 0.07086,
      "grad_norm": 0.9810070731706124,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 7086
    },
    {
      "epoch": 0.07087,
      "grad_norm": 0.9123121727972462,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 7087
    },
    {
      "epoch": 0.07088,
      "grad_norm": 0.8624587235237147,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 7088
    },
    {
      "epoch": 0.07089,
      "grad_norm": 0.7897069970288599,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7089
    },
    {
      "epoch": 0.0709,
      "grad_norm": 0.6952949093057026,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7090
    },
    {
      "epoch": 0.07091,
      "grad_norm": 0.6864230472842601,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 7091
    },
    {
      "epoch": 0.07092,
      "grad_norm": 0.6649365850217969,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 7092
    },
    {
      "epoch": 0.07093,
      "grad_norm": 0.6841019927195824,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 7093
    },
    {
      "epoch": 0.07094,
      "grad_norm": 0.6127976340559059,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 7094
    },
    {
      "epoch": 0.07095,
      "grad_norm": 0.616886834283065,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 7095
    },
    {
      "epoch": 0.07096,
      "grad_norm": 0.6971966421797507,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 7096
    },
    {
      "epoch": 0.07097,
      "grad_norm": 0.6710301285847851,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 7097
    },
    {
      "epoch": 0.07098,
      "grad_norm": 0.61562136549296,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 7098
    },
    {
      "epoch": 0.07099,
      "grad_norm": 0.6419148521913456,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 7099
    },
    {
      "epoch": 0.071,
      "grad_norm": 0.6323659270253572,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 7100
    },
    {
      "epoch": 0.07101,
      "grad_norm": 0.6790975940780517,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 7101
    },
    {
      "epoch": 0.07102,
      "grad_norm": 0.7012073427223774,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 7102
    },
    {
      "epoch": 0.07103,
      "grad_norm": 0.8170376783966959,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 7103
    },
    {
      "epoch": 0.07104,
      "grad_norm": 0.9864196954449765,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 7104
    },
    {
      "epoch": 0.07105,
      "grad_norm": 0.9605522303404178,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 7105
    },
    {
      "epoch": 0.07106,
      "grad_norm": 0.8886566459127305,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 7106
    },
    {
      "epoch": 0.07107,
      "grad_norm": 0.8973256822058434,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 7107
    },
    {
      "epoch": 0.07108,
      "grad_norm": 0.8957355256957436,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 7108
    },
    {
      "epoch": 0.07109,
      "grad_norm": 0.9603444798838409,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 7109
    },
    {
      "epoch": 0.0711,
      "grad_norm": 1.071506569773605,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 7110
    },
    {
      "epoch": 0.07111,
      "grad_norm": 0.8800294655537059,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 7111
    },
    {
      "epoch": 0.07112,
      "grad_norm": 0.651480864700506,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 7112
    },
    {
      "epoch": 0.07113,
      "grad_norm": 0.5845950738429475,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 7113
    },
    {
      "epoch": 0.07114,
      "grad_norm": 0.5822849601468679,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 7114
    },
    {
      "epoch": 0.07115,
      "grad_norm": 0.5474586592685721,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 7115
    },
    {
      "epoch": 0.07116,
      "grad_norm": 0.5187520875086112,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 7116
    },
    {
      "epoch": 0.07117,
      "grad_norm": 0.4394900872650167,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 7117
    },
    {
      "epoch": 0.07118,
      "grad_norm": 0.3463005557935632,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 7118
    },
    {
      "epoch": 0.07119,
      "grad_norm": 0.3752207233130492,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 7119
    },
    {
      "epoch": 0.0712,
      "grad_norm": 0.4024093721623634,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 7120
    },
    {
      "epoch": 0.07121,
      "grad_norm": 0.4320668079849204,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 7121
    },
    {
      "epoch": 0.07122,
      "grad_norm": 0.43814596025832847,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 7122
    },
    {
      "epoch": 0.07123,
      "grad_norm": 0.4660611414965408,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 7123
    },
    {
      "epoch": 0.07124,
      "grad_norm": 0.4926215620935902,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 7124
    },
    {
      "epoch": 0.07125,
      "grad_norm": 0.4996717729986805,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 7125
    },
    {
      "epoch": 0.07126,
      "grad_norm": 0.5425463804440391,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 7126
    },
    {
      "epoch": 0.07127,
      "grad_norm": 0.6028489984656705,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 7127
    },
    {
      "epoch": 0.07128,
      "grad_norm": 0.6414630429602017,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 7128
    },
    {
      "epoch": 0.07129,
      "grad_norm": 0.6306384667736967,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 7129
    },
    {
      "epoch": 0.0713,
      "grad_norm": 0.6762556869839564,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 7130
    },
    {
      "epoch": 0.07131,
      "grad_norm": 0.782955807754172,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 7131
    },
    {
      "epoch": 0.07132,
      "grad_norm": 0.8150492320482005,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 7132
    },
    {
      "epoch": 0.07133,
      "grad_norm": 0.7371939952383894,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 7133
    },
    {
      "epoch": 0.07134,
      "grad_norm": 0.7676831192441333,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 7134
    },
    {
      "epoch": 0.07135,
      "grad_norm": 0.7690673723760045,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 7135
    },
    {
      "epoch": 0.07136,
      "grad_norm": 0.7766203756298646,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 7136
    },
    {
      "epoch": 0.07137,
      "grad_norm": 0.807723913043927,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 7137
    },
    {
      "epoch": 0.07138,
      "grad_norm": 0.7120609852111736,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 7138
    },
    {
      "epoch": 0.07139,
      "grad_norm": 0.6409244184377964,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 7139
    },
    {
      "epoch": 0.0714,
      "grad_norm": 0.6983765449622829,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 7140
    },
    {
      "epoch": 0.07141,
      "grad_norm": 0.6749975641693673,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 7141
    },
    {
      "epoch": 0.07142,
      "grad_norm": 0.7443875679801981,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 7142
    },
    {
      "epoch": 0.07143,
      "grad_norm": 0.8036162553734473,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 7143
    },
    {
      "epoch": 0.07144,
      "grad_norm": 0.8238780582102404,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 7144
    },
    {
      "epoch": 0.07145,
      "grad_norm": 0.7818657301386684,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 7145
    },
    {
      "epoch": 0.07146,
      "grad_norm": 0.589772103883361,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 7146
    },
    {
      "epoch": 0.07147,
      "grad_norm": 0.5061896394583955,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 7147
    },
    {
      "epoch": 0.07148,
      "grad_norm": 0.5563222484565196,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 7148
    },
    {
      "epoch": 0.07149,
      "grad_norm": 0.6306367666731154,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 7149
    },
    {
      "epoch": 0.0715,
      "grad_norm": 0.6045730081269641,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 7150
    },
    {
      "epoch": 0.07151,
      "grad_norm": 0.5783897582471499,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 7151
    },
    {
      "epoch": 0.07152,
      "grad_norm": 0.5932667743048016,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 7152
    },
    {
      "epoch": 0.07153,
      "grad_norm": 0.60473248923499,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 7153
    },
    {
      "epoch": 0.07154,
      "grad_norm": 0.6236732709390373,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 7154
    },
    {
      "epoch": 0.07155,
      "grad_norm": 0.7143317986692795,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 7155
    },
    {
      "epoch": 0.07156,
      "grad_norm": 0.806274009051711,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 7156
    },
    {
      "epoch": 0.07157,
      "grad_norm": 0.8013068941242575,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 7157
    },
    {
      "epoch": 0.07158,
      "grad_norm": 0.8062400294331641,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 7158
    },
    {
      "epoch": 0.07159,
      "grad_norm": 0.901901589441292,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 7159
    },
    {
      "epoch": 0.0716,
      "grad_norm": 0.9668482710989087,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 7160
    },
    {
      "epoch": 0.07161,
      "grad_norm": 0.905245384275865,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 7161
    },
    {
      "epoch": 0.07162,
      "grad_norm": 0.9280503905267976,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 7162
    },
    {
      "epoch": 0.07163,
      "grad_norm": 0.9016130134973356,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 7163
    },
    {
      "epoch": 0.07164,
      "grad_norm": 0.8238069946322247,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 7164
    },
    {
      "epoch": 0.07165,
      "grad_norm": 0.9229684257907796,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 7165
    },
    {
      "epoch": 0.07166,
      "grad_norm": 0.8602566555422861,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 7166
    },
    {
      "epoch": 0.07167,
      "grad_norm": 0.7912994410141391,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 7167
    },
    {
      "epoch": 0.07168,
      "grad_norm": 0.6645028530506353,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 7168
    },
    {
      "epoch": 0.07169,
      "grad_norm": 0.6619833308914435,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7169
    },
    {
      "epoch": 0.0717,
      "grad_norm": 0.6040314764385558,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 7170
    },
    {
      "epoch": 0.07171,
      "grad_norm": 0.6833108775226145,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 7171
    },
    {
      "epoch": 0.07172,
      "grad_norm": 0.7725987913634927,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 7172
    },
    {
      "epoch": 0.07173,
      "grad_norm": 0.7147302013430032,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 7173
    },
    {
      "epoch": 0.07174,
      "grad_norm": 0.7180021296498525,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 7174
    },
    {
      "epoch": 0.07175,
      "grad_norm": 0.7457543221263961,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 7175
    },
    {
      "epoch": 0.07176,
      "grad_norm": 0.6924904976395436,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 7176
    },
    {
      "epoch": 0.07177,
      "grad_norm": 0.5304650992546365,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 7177
    },
    {
      "epoch": 0.07178,
      "grad_norm": 0.5008427987279399,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 7178
    },
    {
      "epoch": 0.07179,
      "grad_norm": 0.4679265582031314,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 7179
    },
    {
      "epoch": 0.0718,
      "grad_norm": 0.43745452895304754,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 7180
    },
    {
      "epoch": 0.07181,
      "grad_norm": 0.45907357705620727,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 7181
    },
    {
      "epoch": 0.07182,
      "grad_norm": 0.49846152510871444,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 7182
    },
    {
      "epoch": 0.07183,
      "grad_norm": 0.5359985839562037,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 7183
    },
    {
      "epoch": 0.07184,
      "grad_norm": 0.6563790510844429,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 7184
    },
    {
      "epoch": 0.07185,
      "grad_norm": 0.6720180437894167,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 7185
    },
    {
      "epoch": 0.07186,
      "grad_norm": 0.5587772637850045,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 7186
    },
    {
      "epoch": 0.07187,
      "grad_norm": 0.5401622815209526,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 7187
    },
    {
      "epoch": 0.07188,
      "grad_norm": 0.5467984886698021,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 7188
    },
    {
      "epoch": 0.07189,
      "grad_norm": 0.5726158372896095,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 7189
    },
    {
      "epoch": 0.0719,
      "grad_norm": 0.7251779521154695,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 7190
    },
    {
      "epoch": 0.07191,
      "grad_norm": 0.8469754288010317,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 7191
    },
    {
      "epoch": 0.07192,
      "grad_norm": 0.9128848705622925,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 7192
    },
    {
      "epoch": 0.07193,
      "grad_norm": 0.8310324783740308,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 7193
    },
    {
      "epoch": 0.07194,
      "grad_norm": 0.6879939761909492,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 7194
    },
    {
      "epoch": 0.07195,
      "grad_norm": 0.5970160052475568,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 7195
    },
    {
      "epoch": 0.07196,
      "grad_norm": 0.5894178919524454,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 7196
    },
    {
      "epoch": 0.07197,
      "grad_norm": 0.5262073910189121,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 7197
    },
    {
      "epoch": 0.07198,
      "grad_norm": 0.5265600947376748,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 7198
    },
    {
      "epoch": 0.07199,
      "grad_norm": 0.541452819187127,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 7199
    },
    {
      "epoch": 0.072,
      "grad_norm": 0.6458006332096501,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 7200
    },
    {
      "epoch": 0.07201,
      "grad_norm": 0.8094071517689202,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 7201
    },
    {
      "epoch": 0.07202,
      "grad_norm": 0.7787699581874201,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 7202
    },
    {
      "epoch": 0.07203,
      "grad_norm": 0.668461238530481,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 7203
    },
    {
      "epoch": 0.07204,
      "grad_norm": 0.6519472171530598,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 7204
    },
    {
      "epoch": 0.07205,
      "grad_norm": 0.553477955345092,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 7205
    },
    {
      "epoch": 0.07206,
      "grad_norm": 0.576969241800161,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 7206
    },
    {
      "epoch": 0.07207,
      "grad_norm": 0.6130153388967977,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 7207
    },
    {
      "epoch": 0.07208,
      "grad_norm": 0.6120567702128528,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 7208
    },
    {
      "epoch": 0.07209,
      "grad_norm": 0.5925349870606897,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 7209
    },
    {
      "epoch": 0.0721,
      "grad_norm": 0.5460934958885093,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 7210
    },
    {
      "epoch": 0.07211,
      "grad_norm": 0.5617487245207603,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 7211
    },
    {
      "epoch": 0.07212,
      "grad_norm": 0.5415277872929839,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 7212
    },
    {
      "epoch": 0.07213,
      "grad_norm": 0.5502607732589133,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 7213
    },
    {
      "epoch": 0.07214,
      "grad_norm": 0.6409863289739841,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 7214
    },
    {
      "epoch": 0.07215,
      "grad_norm": 0.7570834086211581,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 7215
    },
    {
      "epoch": 0.07216,
      "grad_norm": 0.8132567876934826,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 7216
    },
    {
      "epoch": 0.07217,
      "grad_norm": 0.9345387667787567,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 7217
    },
    {
      "epoch": 0.07218,
      "grad_norm": 1.0833907300880954,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 7218
    },
    {
      "epoch": 0.07219,
      "grad_norm": 0.9260287293122886,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 7219
    },
    {
      "epoch": 0.0722,
      "grad_norm": 1.038503139941199,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 7220
    },
    {
      "epoch": 0.07221,
      "grad_norm": 0.9210746523660105,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 7221
    },
    {
      "epoch": 0.07222,
      "grad_norm": 0.9858808371441633,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 7222
    },
    {
      "epoch": 0.07223,
      "grad_norm": 0.9836284417759733,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 7223
    },
    {
      "epoch": 0.07224,
      "grad_norm": 0.985172160848877,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 7224
    },
    {
      "epoch": 0.07225,
      "grad_norm": 1.0092262989567715,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 7225
    },
    {
      "epoch": 0.07226,
      "grad_norm": 1.0380410348021303,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 7226
    },
    {
      "epoch": 0.07227,
      "grad_norm": 0.897496931177557,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 7227
    },
    {
      "epoch": 0.07228,
      "grad_norm": 0.8371397499199325,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 7228
    },
    {
      "epoch": 0.07229,
      "grad_norm": 0.68089602479955,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 7229
    },
    {
      "epoch": 0.0723,
      "grad_norm": 0.7440697649975428,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 7230
    },
    {
      "epoch": 0.07231,
      "grad_norm": 0.8546390577858282,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 7231
    },
    {
      "epoch": 0.07232,
      "grad_norm": 0.9911495928687625,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 7232
    },
    {
      "epoch": 0.07233,
      "grad_norm": 1.0402994652801865,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 7233
    },
    {
      "epoch": 0.07234,
      "grad_norm": 0.7834142777445765,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 7234
    },
    {
      "epoch": 0.07235,
      "grad_norm": 0.79697526236287,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 7235
    },
    {
      "epoch": 0.07236,
      "grad_norm": 0.8580397156652976,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 7236
    },
    {
      "epoch": 0.07237,
      "grad_norm": 0.8311156048197793,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 7237
    },
    {
      "epoch": 0.07238,
      "grad_norm": 0.7018648916636654,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 7238
    },
    {
      "epoch": 0.07239,
      "grad_norm": 0.6039008355156773,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 7239
    },
    {
      "epoch": 0.0724,
      "grad_norm": 0.5161419422417888,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 7240
    },
    {
      "epoch": 0.07241,
      "grad_norm": 0.4874109183137781,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 7241
    },
    {
      "epoch": 0.07242,
      "grad_norm": 0.6217288908705515,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 7242
    },
    {
      "epoch": 0.07243,
      "grad_norm": 0.6488298417249205,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 7243
    },
    {
      "epoch": 0.07244,
      "grad_norm": 0.6010925037781651,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 7244
    },
    {
      "epoch": 0.07245,
      "grad_norm": 0.5324768587577942,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 7245
    },
    {
      "epoch": 0.07246,
      "grad_norm": 0.49764740145745306,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 7246
    },
    {
      "epoch": 0.07247,
      "grad_norm": 0.4879501046529579,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 7247
    },
    {
      "epoch": 0.07248,
      "grad_norm": 0.44818383989320487,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 7248
    },
    {
      "epoch": 0.07249,
      "grad_norm": 0.4019767982051826,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 7249
    },
    {
      "epoch": 0.0725,
      "grad_norm": 0.37656345224727716,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 7250
    },
    {
      "epoch": 0.07251,
      "grad_norm": 0.39765646327210397,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 7251
    },
    {
      "epoch": 0.07252,
      "grad_norm": 0.4087153991886569,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 7252
    },
    {
      "epoch": 0.07253,
      "grad_norm": 0.4413249795378963,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 7253
    },
    {
      "epoch": 0.07254,
      "grad_norm": 0.49658754339700917,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 7254
    },
    {
      "epoch": 0.07255,
      "grad_norm": 0.6065287271274902,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7255
    },
    {
      "epoch": 0.07256,
      "grad_norm": 0.7783935834491852,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 7256
    },
    {
      "epoch": 0.07257,
      "grad_norm": 0.8736962562673857,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 7257
    },
    {
      "epoch": 0.07258,
      "grad_norm": 0.7615763185801613,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 7258
    },
    {
      "epoch": 0.07259,
      "grad_norm": 0.6225198130561953,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 7259
    },
    {
      "epoch": 0.0726,
      "grad_norm": 0.6660945346455424,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 7260
    },
    {
      "epoch": 0.07261,
      "grad_norm": 0.6994963663146176,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 7261
    },
    {
      "epoch": 0.07262,
      "grad_norm": 0.7558699256531053,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 7262
    },
    {
      "epoch": 0.07263,
      "grad_norm": 0.8389060291693147,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 7263
    },
    {
      "epoch": 0.07264,
      "grad_norm": 0.9199074585746712,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 7264
    },
    {
      "epoch": 0.07265,
      "grad_norm": 0.9046795786100689,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 7265
    },
    {
      "epoch": 0.07266,
      "grad_norm": 0.8984966553892594,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 7266
    },
    {
      "epoch": 0.07267,
      "grad_norm": 0.8779492964239948,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 7267
    },
    {
      "epoch": 0.07268,
      "grad_norm": 0.9458771160182521,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 7268
    },
    {
      "epoch": 0.07269,
      "grad_norm": 1.0021477216087307,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 7269
    },
    {
      "epoch": 0.0727,
      "grad_norm": 1.0517837764583857,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 7270
    },
    {
      "epoch": 0.07271,
      "grad_norm": 0.7757058905695857,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 7271
    },
    {
      "epoch": 0.07272,
      "grad_norm": 0.6955324357433823,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 7272
    },
    {
      "epoch": 0.07273,
      "grad_norm": 0.727966212131769,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 7273
    },
    {
      "epoch": 0.07274,
      "grad_norm": 0.6894516072772396,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 7274
    },
    {
      "epoch": 0.07275,
      "grad_norm": 0.6466287667559749,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 7275
    },
    {
      "epoch": 0.07276,
      "grad_norm": 0.5806437052644794,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 7276
    },
    {
      "epoch": 0.07277,
      "grad_norm": 0.5491496894626376,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 7277
    },
    {
      "epoch": 0.07278,
      "grad_norm": 0.5807535736273424,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 7278
    },
    {
      "epoch": 0.07279,
      "grad_norm": 0.5810337347879152,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 7279
    },
    {
      "epoch": 0.0728,
      "grad_norm": 0.593882083760101,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 7280
    },
    {
      "epoch": 0.07281,
      "grad_norm": 0.5145677719892531,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 7281
    },
    {
      "epoch": 0.07282,
      "grad_norm": 0.5310247033119996,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 7282
    },
    {
      "epoch": 0.07283,
      "grad_norm": 0.5752791694559306,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 7283
    },
    {
      "epoch": 0.07284,
      "grad_norm": 0.6120341790052407,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 7284
    },
    {
      "epoch": 0.07285,
      "grad_norm": 0.7608771862935181,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 7285
    },
    {
      "epoch": 0.07286,
      "grad_norm": 0.8881767066517879,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 7286
    },
    {
      "epoch": 0.07287,
      "grad_norm": 0.929008736891511,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 7287
    },
    {
      "epoch": 0.07288,
      "grad_norm": 0.7448189338903293,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 7288
    },
    {
      "epoch": 0.07289,
      "grad_norm": 0.6196893187777653,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 7289
    },
    {
      "epoch": 0.0729,
      "grad_norm": 0.6617627631781875,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 7290
    },
    {
      "epoch": 0.07291,
      "grad_norm": 0.5587104110160175,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 7291
    },
    {
      "epoch": 0.07292,
      "grad_norm": 0.5227848078058569,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 7292
    },
    {
      "epoch": 0.07293,
      "grad_norm": 0.4287989659731113,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 7293
    },
    {
      "epoch": 0.07294,
      "grad_norm": 0.39434672478777266,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 7294
    },
    {
      "epoch": 0.07295,
      "grad_norm": 0.35908636579555575,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 7295
    },
    {
      "epoch": 0.07296,
      "grad_norm": 0.3644756637139822,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 7296
    },
    {
      "epoch": 0.07297,
      "grad_norm": 0.3770629664236333,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 7297
    },
    {
      "epoch": 0.07298,
      "grad_norm": 0.48678207336741464,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 7298
    },
    {
      "epoch": 0.07299,
      "grad_norm": 0.5846170384424557,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 7299
    },
    {
      "epoch": 0.073,
      "grad_norm": 0.6595284998416144,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 7300
    },
    {
      "epoch": 0.07301,
      "grad_norm": 0.6211347463265787,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 7301
    },
    {
      "epoch": 0.07302,
      "grad_norm": 0.48150627502231,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 7302
    },
    {
      "epoch": 0.07303,
      "grad_norm": 0.5489707911727371,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 7303
    },
    {
      "epoch": 0.07304,
      "grad_norm": 0.5974812098818115,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 7304
    },
    {
      "epoch": 0.07305,
      "grad_norm": 0.7024670978584663,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 7305
    },
    {
      "epoch": 0.07306,
      "grad_norm": 0.7638676834438926,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 7306
    },
    {
      "epoch": 0.07307,
      "grad_norm": 0.8266201952032484,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 7307
    },
    {
      "epoch": 0.07308,
      "grad_norm": 0.8513186722362418,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 7308
    },
    {
      "epoch": 0.07309,
      "grad_norm": 1.0336678643068749,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 7309
    },
    {
      "epoch": 0.0731,
      "grad_norm": 1.1463223547284502,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 7310
    },
    {
      "epoch": 0.07311,
      "grad_norm": 0.7757914479144218,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 7311
    },
    {
      "epoch": 0.07312,
      "grad_norm": 0.7055116339180564,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 7312
    },
    {
      "epoch": 0.07313,
      "grad_norm": 0.6548456604315757,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 7313
    },
    {
      "epoch": 0.07314,
      "grad_norm": 0.7320846700471259,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 7314
    },
    {
      "epoch": 0.07315,
      "grad_norm": 0.8166597914868653,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 7315
    },
    {
      "epoch": 0.07316,
      "grad_norm": 0.7595697268852177,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 7316
    },
    {
      "epoch": 0.07317,
      "grad_norm": 0.6198775441688376,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 7317
    },
    {
      "epoch": 0.07318,
      "grad_norm": 0.5723028866588152,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 7318
    },
    {
      "epoch": 0.07319,
      "grad_norm": 0.6057245627345597,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 7319
    },
    {
      "epoch": 0.0732,
      "grad_norm": 0.7239900990155307,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 7320
    },
    {
      "epoch": 0.07321,
      "grad_norm": 0.6568845099790048,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 7321
    },
    {
      "epoch": 0.07322,
      "grad_norm": 0.5961986038574478,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 7322
    },
    {
      "epoch": 0.07323,
      "grad_norm": 0.6111250018556825,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 7323
    },
    {
      "epoch": 0.07324,
      "grad_norm": 0.6314202107843266,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 7324
    },
    {
      "epoch": 0.07325,
      "grad_norm": 0.5603304228911395,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 7325
    },
    {
      "epoch": 0.07326,
      "grad_norm": 0.6198242641884363,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 7326
    },
    {
      "epoch": 0.07327,
      "grad_norm": 0.6180461035522653,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 7327
    },
    {
      "epoch": 0.07328,
      "grad_norm": 0.5997168347855873,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 7328
    },
    {
      "epoch": 0.07329,
      "grad_norm": 0.7505933625758654,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 7329
    },
    {
      "epoch": 0.0733,
      "grad_norm": 0.95882955808798,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 7330
    },
    {
      "epoch": 0.07331,
      "grad_norm": 1.1625555883916676,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 7331
    },
    {
      "epoch": 0.07332,
      "grad_norm": 0.713532800458944,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 7332
    },
    {
      "epoch": 0.07333,
      "grad_norm": 0.6381195482238602,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 7333
    },
    {
      "epoch": 0.07334,
      "grad_norm": 0.5803769095988841,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 7334
    },
    {
      "epoch": 0.07335,
      "grad_norm": 0.5486263031540451,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 7335
    },
    {
      "epoch": 0.07336,
      "grad_norm": 0.5662029197744988,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 7336
    },
    {
      "epoch": 0.07337,
      "grad_norm": 0.4935245740209676,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 7337
    },
    {
      "epoch": 0.07338,
      "grad_norm": 0.5251534080906687,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 7338
    },
    {
      "epoch": 0.07339,
      "grad_norm": 0.5426405338587877,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 7339
    },
    {
      "epoch": 0.0734,
      "grad_norm": 0.572679387243017,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 7340
    },
    {
      "epoch": 0.07341,
      "grad_norm": 0.592683237309645,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 7341
    },
    {
      "epoch": 0.07342,
      "grad_norm": 0.6784287433146412,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 7342
    },
    {
      "epoch": 0.07343,
      "grad_norm": 0.8466966833790952,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 7343
    },
    {
      "epoch": 0.07344,
      "grad_norm": 0.9627040021124741,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 7344
    },
    {
      "epoch": 0.07345,
      "grad_norm": 0.8840562235128062,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 7345
    },
    {
      "epoch": 0.07346,
      "grad_norm": 0.7548058738161189,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 7346
    },
    {
      "epoch": 0.07347,
      "grad_norm": 0.6631446634048735,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 7347
    },
    {
      "epoch": 0.07348,
      "grad_norm": 0.7265237879004127,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 7348
    },
    {
      "epoch": 0.07349,
      "grad_norm": 0.7506619044246161,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 7349
    },
    {
      "epoch": 0.0735,
      "grad_norm": 0.8461103544372488,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 7350
    },
    {
      "epoch": 0.07351,
      "grad_norm": 0.990138732808367,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7351
    },
    {
      "epoch": 0.07352,
      "grad_norm": 1.033284964426441,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 7352
    },
    {
      "epoch": 0.07353,
      "grad_norm": 0.9900655416873149,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 7353
    },
    {
      "epoch": 0.07354,
      "grad_norm": 0.8494895382662214,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 7354
    },
    {
      "epoch": 0.07355,
      "grad_norm": 0.8514735721257103,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 7355
    },
    {
      "epoch": 0.07356,
      "grad_norm": 0.7642132298715535,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 7356
    },
    {
      "epoch": 0.07357,
      "grad_norm": 0.7217630714630613,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 7357
    },
    {
      "epoch": 0.07358,
      "grad_norm": 0.6372792070079403,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 7358
    },
    {
      "epoch": 0.07359,
      "grad_norm": 0.6640339672379827,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 7359
    },
    {
      "epoch": 0.0736,
      "grad_norm": 0.6688944780134485,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 7360
    },
    {
      "epoch": 0.07361,
      "grad_norm": 0.6165260490708869,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 7361
    },
    {
      "epoch": 0.07362,
      "grad_norm": 0.6046582282651884,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 7362
    },
    {
      "epoch": 0.07363,
      "grad_norm": 0.5897978442551399,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 7363
    },
    {
      "epoch": 0.07364,
      "grad_norm": 0.5332996805238572,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 7364
    },
    {
      "epoch": 0.07365,
      "grad_norm": 0.46010756257859925,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 7365
    },
    {
      "epoch": 0.07366,
      "grad_norm": 0.5374848389207422,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 7366
    },
    {
      "epoch": 0.07367,
      "grad_norm": 0.6597380812902667,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 7367
    },
    {
      "epoch": 0.07368,
      "grad_norm": 0.8518139437419292,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 7368
    },
    {
      "epoch": 0.07369,
      "grad_norm": 0.9618450628071217,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 7369
    },
    {
      "epoch": 0.0737,
      "grad_norm": 0.8620882223010047,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 7370
    },
    {
      "epoch": 0.07371,
      "grad_norm": 0.6611222884366442,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 7371
    },
    {
      "epoch": 0.07372,
      "grad_norm": 0.7543698836104294,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 7372
    },
    {
      "epoch": 0.07373,
      "grad_norm": 0.7294801364711708,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 7373
    },
    {
      "epoch": 0.07374,
      "grad_norm": 0.6703748242790802,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 7374
    },
    {
      "epoch": 0.07375,
      "grad_norm": 0.6644624708027584,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7375
    },
    {
      "epoch": 0.07376,
      "grad_norm": 0.7311681812607845,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 7376
    },
    {
      "epoch": 0.07377,
      "grad_norm": 0.7075965457515309,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 7377
    },
    {
      "epoch": 0.07378,
      "grad_norm": 0.6352699526430687,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 7378
    },
    {
      "epoch": 0.07379,
      "grad_norm": 0.6697856496315371,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 7379
    },
    {
      "epoch": 0.0738,
      "grad_norm": 0.6784741265654184,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 7380
    },
    {
      "epoch": 0.07381,
      "grad_norm": 0.5900914351363498,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 7381
    },
    {
      "epoch": 0.07382,
      "grad_norm": 0.608569663017179,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 7382
    },
    {
      "epoch": 0.07383,
      "grad_norm": 0.6350237567101997,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 7383
    },
    {
      "epoch": 0.07384,
      "grad_norm": 0.6424806458733962,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 7384
    },
    {
      "epoch": 0.07385,
      "grad_norm": 0.6720061518073006,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 7385
    },
    {
      "epoch": 0.07386,
      "grad_norm": 0.7795027909968454,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 7386
    },
    {
      "epoch": 0.07387,
      "grad_norm": 0.8143599209673019,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 7387
    },
    {
      "epoch": 0.07388,
      "grad_norm": 0.7807403224128945,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 7388
    },
    {
      "epoch": 0.07389,
      "grad_norm": 0.6943770705390825,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 7389
    },
    {
      "epoch": 0.0739,
      "grad_norm": 0.6749062613953275,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 7390
    },
    {
      "epoch": 0.07391,
      "grad_norm": 0.6980025211034937,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7391
    },
    {
      "epoch": 0.07392,
      "grad_norm": 0.8142287222162783,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 7392
    },
    {
      "epoch": 0.07393,
      "grad_norm": 0.848080762828504,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 7393
    },
    {
      "epoch": 0.07394,
      "grad_norm": 0.7052754295757714,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 7394
    },
    {
      "epoch": 0.07395,
      "grad_norm": 0.6918679030912211,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 7395
    },
    {
      "epoch": 0.07396,
      "grad_norm": 0.6396360507546505,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 7396
    },
    {
      "epoch": 0.07397,
      "grad_norm": 0.5575428300488946,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 7397
    },
    {
      "epoch": 0.07398,
      "grad_norm": 0.47819343358592886,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 7398
    },
    {
      "epoch": 0.07399,
      "grad_norm": 0.49123569869531275,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 7399
    },
    {
      "epoch": 0.074,
      "grad_norm": 0.4873818583702739,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 7400
    },
    {
      "epoch": 0.07401,
      "grad_norm": 0.5338185744914158,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 7401
    },
    {
      "epoch": 0.07402,
      "grad_norm": 0.6259981261871912,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 7402
    },
    {
      "epoch": 0.07403,
      "grad_norm": 0.7875202985260396,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 7403
    },
    {
      "epoch": 0.07404,
      "grad_norm": 0.9375888555588073,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 7404
    },
    {
      "epoch": 0.07405,
      "grad_norm": 0.916338808307593,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 7405
    },
    {
      "epoch": 0.07406,
      "grad_norm": 0.7809170835167528,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 7406
    },
    {
      "epoch": 0.07407,
      "grad_norm": 0.6654944055402259,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 7407
    },
    {
      "epoch": 0.07408,
      "grad_norm": 0.6915372132184653,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7408
    },
    {
      "epoch": 0.07409,
      "grad_norm": 0.7828828066535588,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 7409
    },
    {
      "epoch": 0.0741,
      "grad_norm": 0.8226790763634696,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 7410
    },
    {
      "epoch": 0.07411,
      "grad_norm": 0.8943168078020164,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 7411
    },
    {
      "epoch": 0.07412,
      "grad_norm": 0.8016790470869424,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 7412
    },
    {
      "epoch": 0.07413,
      "grad_norm": 0.7242923730785583,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 7413
    },
    {
      "epoch": 0.07414,
      "grad_norm": 0.7803236835873817,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 7414
    },
    {
      "epoch": 0.07415,
      "grad_norm": 0.809440770991952,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 7415
    },
    {
      "epoch": 0.07416,
      "grad_norm": 0.814374589546072,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7416
    },
    {
      "epoch": 0.07417,
      "grad_norm": 0.7472307812051249,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 7417
    },
    {
      "epoch": 0.07418,
      "grad_norm": 0.668144266675147,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 7418
    },
    {
      "epoch": 0.07419,
      "grad_norm": 0.7132902738686082,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 7419
    },
    {
      "epoch": 0.0742,
      "grad_norm": 0.6426183694835322,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 7420
    },
    {
      "epoch": 0.07421,
      "grad_norm": 0.6842452822488992,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 7421
    },
    {
      "epoch": 0.07422,
      "grad_norm": 0.6935109936445435,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7422
    },
    {
      "epoch": 0.07423,
      "grad_norm": 0.6977420494019977,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 7423
    },
    {
      "epoch": 0.07424,
      "grad_norm": 0.66836051737428,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 7424
    },
    {
      "epoch": 0.07425,
      "grad_norm": 0.6071918274550937,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 7425
    },
    {
      "epoch": 0.07426,
      "grad_norm": 0.5672292395284034,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 7426
    },
    {
      "epoch": 0.07427,
      "grad_norm": 0.6881340403118316,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 7427
    },
    {
      "epoch": 0.07428,
      "grad_norm": 0.8295995612479321,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 7428
    },
    {
      "epoch": 0.07429,
      "grad_norm": 1.0310591530396886,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 7429
    },
    {
      "epoch": 0.0743,
      "grad_norm": 1.1113844190744362,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 7430
    },
    {
      "epoch": 0.07431,
      "grad_norm": 0.8368950095459845,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 7431
    },
    {
      "epoch": 0.07432,
      "grad_norm": 0.7370171566843224,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 7432
    },
    {
      "epoch": 0.07433,
      "grad_norm": 0.7186305212736634,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 7433
    },
    {
      "epoch": 0.07434,
      "grad_norm": 0.6592511244859814,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 7434
    },
    {
      "epoch": 0.07435,
      "grad_norm": 0.6659908164403538,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 7435
    },
    {
      "epoch": 0.07436,
      "grad_norm": 0.7024784942727049,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 7436
    },
    {
      "epoch": 0.07437,
      "grad_norm": 0.750590282273592,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 7437
    },
    {
      "epoch": 0.07438,
      "grad_norm": 0.7342624547457341,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 7438
    },
    {
      "epoch": 0.07439,
      "grad_norm": 0.6552175598983562,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 7439
    },
    {
      "epoch": 0.0744,
      "grad_norm": 0.6670994817257487,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 7440
    },
    {
      "epoch": 0.07441,
      "grad_norm": 0.5868493465333204,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 7441
    },
    {
      "epoch": 0.07442,
      "grad_norm": 0.5749846079049408,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7442
    },
    {
      "epoch": 0.07443,
      "grad_norm": 0.616246467740031,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 7443
    },
    {
      "epoch": 0.07444,
      "grad_norm": 0.6314116396577024,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 7444
    },
    {
      "epoch": 0.07445,
      "grad_norm": 0.6891618905408962,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 7445
    },
    {
      "epoch": 0.07446,
      "grad_norm": 0.7665111909100017,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 7446
    },
    {
      "epoch": 0.07447,
      "grad_norm": 0.7960586954783623,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 7447
    },
    {
      "epoch": 0.07448,
      "grad_norm": 0.6932171237233758,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 7448
    },
    {
      "epoch": 0.07449,
      "grad_norm": 0.555157340249593,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 7449
    },
    {
      "epoch": 0.0745,
      "grad_norm": 0.5727726738711602,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 7450
    },
    {
      "epoch": 0.07451,
      "grad_norm": 0.6677189755758194,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 7451
    },
    {
      "epoch": 0.07452,
      "grad_norm": 0.7461108960326893,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 7452
    },
    {
      "epoch": 0.07453,
      "grad_norm": 0.7689716942440091,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 7453
    },
    {
      "epoch": 0.07454,
      "grad_norm": 0.6847633220363298,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 7454
    },
    {
      "epoch": 0.07455,
      "grad_norm": 0.6834076703366209,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 7455
    },
    {
      "epoch": 0.07456,
      "grad_norm": 0.5915374732547324,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 7456
    },
    {
      "epoch": 0.07457,
      "grad_norm": 0.5353509342865896,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 7457
    },
    {
      "epoch": 0.07458,
      "grad_norm": 0.6021411618752377,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 7458
    },
    {
      "epoch": 0.07459,
      "grad_norm": 0.6164057139569667,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7459
    },
    {
      "epoch": 0.0746,
      "grad_norm": 0.7212773204762993,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 7460
    },
    {
      "epoch": 0.07461,
      "grad_norm": 0.7976317064628464,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 7461
    },
    {
      "epoch": 0.07462,
      "grad_norm": 0.8064167886203844,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 7462
    },
    {
      "epoch": 0.07463,
      "grad_norm": 0.7632409751906114,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 7463
    },
    {
      "epoch": 0.07464,
      "grad_norm": 0.7512717490057481,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 7464
    },
    {
      "epoch": 0.07465,
      "grad_norm": 0.7943363392230436,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 7465
    },
    {
      "epoch": 0.07466,
      "grad_norm": 0.8637427260824427,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 7466
    },
    {
      "epoch": 0.07467,
      "grad_norm": 0.8498740214887592,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 7467
    },
    {
      "epoch": 0.07468,
      "grad_norm": 0.8357355406639847,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 7468
    },
    {
      "epoch": 0.07469,
      "grad_norm": 0.8142934568301395,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 7469
    },
    {
      "epoch": 0.0747,
      "grad_norm": 0.8503882150265013,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 7470
    },
    {
      "epoch": 0.07471,
      "grad_norm": 0.8925914077857183,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 7471
    },
    {
      "epoch": 0.07472,
      "grad_norm": 0.9101180407681418,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 7472
    },
    {
      "epoch": 0.07473,
      "grad_norm": 0.8580441555622639,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 7473
    },
    {
      "epoch": 0.07474,
      "grad_norm": 0.8896005983371758,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7474
    },
    {
      "epoch": 0.07475,
      "grad_norm": 0.7824169038965194,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 7475
    },
    {
      "epoch": 0.07476,
      "grad_norm": 0.7710890200443516,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 7476
    },
    {
      "epoch": 0.07477,
      "grad_norm": 0.7900469190579128,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 7477
    },
    {
      "epoch": 0.07478,
      "grad_norm": 0.6977697020555949,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 7478
    },
    {
      "epoch": 0.07479,
      "grad_norm": 0.6860967351037519,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 7479
    },
    {
      "epoch": 0.0748,
      "grad_norm": 0.6740365801183335,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 7480
    },
    {
      "epoch": 0.07481,
      "grad_norm": 0.7756716209864886,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 7481
    },
    {
      "epoch": 0.07482,
      "grad_norm": 0.7133092139283264,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 7482
    },
    {
      "epoch": 0.07483,
      "grad_norm": 0.6326111223603825,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 7483
    },
    {
      "epoch": 0.07484,
      "grad_norm": 0.5282476293532727,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 7484
    },
    {
      "epoch": 0.07485,
      "grad_norm": 0.5706931649668086,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 7485
    },
    {
      "epoch": 0.07486,
      "grad_norm": 0.606269213255391,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 7486
    },
    {
      "epoch": 0.07487,
      "grad_norm": 0.6105645025729305,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 7487
    },
    {
      "epoch": 0.07488,
      "grad_norm": 0.6237051677171727,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 7488
    },
    {
      "epoch": 0.07489,
      "grad_norm": 0.7811835813919411,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 7489
    },
    {
      "epoch": 0.0749,
      "grad_norm": 0.7579316329950212,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 7490
    },
    {
      "epoch": 0.07491,
      "grad_norm": 0.6336904857400839,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 7491
    },
    {
      "epoch": 0.07492,
      "grad_norm": 0.5762678308899742,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 7492
    },
    {
      "epoch": 0.07493,
      "grad_norm": 0.5896176068966966,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7493
    },
    {
      "epoch": 0.07494,
      "grad_norm": 0.5953859444943526,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 7494
    },
    {
      "epoch": 0.07495,
      "grad_norm": 0.6189501827559954,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 7495
    },
    {
      "epoch": 0.07496,
      "grad_norm": 0.6476212217965059,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 7496
    },
    {
      "epoch": 0.07497,
      "grad_norm": 0.590017860504856,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 7497
    },
    {
      "epoch": 0.07498,
      "grad_norm": 0.5711936608923054,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 7498
    },
    {
      "epoch": 0.07499,
      "grad_norm": 0.6570356392413452,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 7499
    },
    {
      "epoch": 0.075,
      "grad_norm": 0.6759155863873976,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 7500
    },
    {
      "epoch": 0.07501,
      "grad_norm": 0.8110049470072964,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 7501
    },
    {
      "epoch": 0.07502,
      "grad_norm": 0.9727707162810135,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 7502
    },
    {
      "epoch": 0.07503,
      "grad_norm": 1.1150877882227244,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 7503
    },
    {
      "epoch": 0.07504,
      "grad_norm": 0.7577757721039465,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 7504
    },
    {
      "epoch": 0.07505,
      "grad_norm": 0.6742452888810195,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 7505
    },
    {
      "epoch": 0.07506,
      "grad_norm": 0.6189390428317627,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 7506
    },
    {
      "epoch": 0.07507,
      "grad_norm": 0.6022087392016671,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 7507
    },
    {
      "epoch": 0.07508,
      "grad_norm": 0.5309580523879476,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 7508
    },
    {
      "epoch": 0.07509,
      "grad_norm": 0.5484502948250853,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 7509
    },
    {
      "epoch": 0.0751,
      "grad_norm": 0.6594506505512988,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 7510
    },
    {
      "epoch": 0.07511,
      "grad_norm": 0.6599050252229222,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 7511
    },
    {
      "epoch": 0.07512,
      "grad_norm": 0.5970442662703767,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 7512
    },
    {
      "epoch": 0.07513,
      "grad_norm": 0.5770492421013043,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 7513
    },
    {
      "epoch": 0.07514,
      "grad_norm": 0.5972414190275317,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 7514
    },
    {
      "epoch": 0.07515,
      "grad_norm": 0.5236448183039198,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 7515
    },
    {
      "epoch": 0.07516,
      "grad_norm": 0.491529472938158,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 7516
    },
    {
      "epoch": 0.07517,
      "grad_norm": 0.5269178228921855,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7517
    },
    {
      "epoch": 0.07518,
      "grad_norm": 0.49875426784289123,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 7518
    },
    {
      "epoch": 0.07519,
      "grad_norm": 0.4804224094730234,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 7519
    },
    {
      "epoch": 0.0752,
      "grad_norm": 0.4238021009088441,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 7520
    },
    {
      "epoch": 0.07521,
      "grad_norm": 0.4356047629744244,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 7521
    },
    {
      "epoch": 0.07522,
      "grad_norm": 0.4848019612577865,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 7522
    },
    {
      "epoch": 0.07523,
      "grad_norm": 0.5676769876951443,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 7523
    },
    {
      "epoch": 0.07524,
      "grad_norm": 0.7969402415607215,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 7524
    },
    {
      "epoch": 0.07525,
      "grad_norm": 1.158159401601464,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 7525
    },
    {
      "epoch": 0.07526,
      "grad_norm": 0.9956964848184818,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 7526
    },
    {
      "epoch": 0.07527,
      "grad_norm": 1.040836474640964,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 7527
    },
    {
      "epoch": 0.07528,
      "grad_norm": 0.9140741004270626,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 7528
    },
    {
      "epoch": 0.07529,
      "grad_norm": 1.0052666855509356,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 7529
    },
    {
      "epoch": 0.0753,
      "grad_norm": 0.9481670539724812,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 7530
    },
    {
      "epoch": 0.07531,
      "grad_norm": 0.9918379168573455,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 7531
    },
    {
      "epoch": 0.07532,
      "grad_norm": 1.0229542004769954,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 7532
    },
    {
      "epoch": 0.07533,
      "grad_norm": 1.0918782376363978,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 7533
    },
    {
      "epoch": 0.07534,
      "grad_norm": 0.9604541131474513,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 7534
    },
    {
      "epoch": 0.07535,
      "grad_norm": 1.0009685102812784,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 7535
    },
    {
      "epoch": 0.07536,
      "grad_norm": 1.0487571004898342,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 7536
    },
    {
      "epoch": 0.07537,
      "grad_norm": 0.9991469202887211,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 7537
    },
    {
      "epoch": 0.07538,
      "grad_norm": 0.8819852519452202,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 7538
    },
    {
      "epoch": 0.07539,
      "grad_norm": 0.8807197730769387,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 7539
    },
    {
      "epoch": 0.0754,
      "grad_norm": 0.877244025946411,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 7540
    },
    {
      "epoch": 0.07541,
      "grad_norm": 0.9740029088436687,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 7541
    },
    {
      "epoch": 0.07542,
      "grad_norm": 0.8600848931909216,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 7542
    },
    {
      "epoch": 0.07543,
      "grad_norm": 0.8280886875133882,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 7543
    },
    {
      "epoch": 0.07544,
      "grad_norm": 0.8071984724777687,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 7544
    },
    {
      "epoch": 0.07545,
      "grad_norm": 0.698095628354796,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 7545
    },
    {
      "epoch": 0.07546,
      "grad_norm": 0.6402526854429936,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 7546
    },
    {
      "epoch": 0.07547,
      "grad_norm": 0.6536241524830555,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 7547
    },
    {
      "epoch": 0.07548,
      "grad_norm": 0.5657943886646665,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 7548
    },
    {
      "epoch": 0.07549,
      "grad_norm": 0.6076370507493793,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 7549
    },
    {
      "epoch": 0.0755,
      "grad_norm": 0.5883507720311228,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 7550
    },
    {
      "epoch": 0.07551,
      "grad_norm": 0.619610373235812,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 7551
    },
    {
      "epoch": 0.07552,
      "grad_norm": 0.7123411482561006,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 7552
    },
    {
      "epoch": 0.07553,
      "grad_norm": 0.6469950236749169,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 7553
    },
    {
      "epoch": 0.07554,
      "grad_norm": 0.6316235881426905,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 7554
    },
    {
      "epoch": 0.07555,
      "grad_norm": 0.7256901233443528,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 7555
    },
    {
      "epoch": 0.07556,
      "grad_norm": 0.7618410256376742,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 7556
    },
    {
      "epoch": 0.07557,
      "grad_norm": 0.7239253341070478,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 7557
    },
    {
      "epoch": 0.07558,
      "grad_norm": 0.6931338512410227,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 7558
    },
    {
      "epoch": 0.07559,
      "grad_norm": 0.6724575608591329,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 7559
    },
    {
      "epoch": 0.0756,
      "grad_norm": 0.7061480953108663,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 7560
    },
    {
      "epoch": 0.07561,
      "grad_norm": 0.7787438183497947,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 7561
    },
    {
      "epoch": 0.07562,
      "grad_norm": 0.9044825545181286,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 7562
    },
    {
      "epoch": 0.07563,
      "grad_norm": 0.9934288625594392,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 7563
    },
    {
      "epoch": 0.07564,
      "grad_norm": 0.8635651395659641,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 7564
    },
    {
      "epoch": 0.07565,
      "grad_norm": 0.6895282281140435,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 7565
    },
    {
      "epoch": 0.07566,
      "grad_norm": 0.5363646891497745,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 7566
    },
    {
      "epoch": 0.07567,
      "grad_norm": 0.5387866394074117,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 7567
    },
    {
      "epoch": 0.07568,
      "grad_norm": 0.5090019275065167,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 7568
    },
    {
      "epoch": 0.07569,
      "grad_norm": 0.42405385696904213,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 7569
    },
    {
      "epoch": 0.0757,
      "grad_norm": 0.4110011928681986,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 7570
    },
    {
      "epoch": 0.07571,
      "grad_norm": 0.4413812175592004,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 7571
    },
    {
      "epoch": 0.07572,
      "grad_norm": 0.4976749366359514,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 7572
    },
    {
      "epoch": 0.07573,
      "grad_norm": 0.5537972928779329,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 7573
    },
    {
      "epoch": 0.07574,
      "grad_norm": 0.5308733184384575,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 7574
    },
    {
      "epoch": 0.07575,
      "grad_norm": 0.49041822167560783,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 7575
    },
    {
      "epoch": 0.07576,
      "grad_norm": 0.5081385300005833,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 7576
    },
    {
      "epoch": 0.07577,
      "grad_norm": 0.574336572955178,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 7577
    },
    {
      "epoch": 0.07578,
      "grad_norm": 0.5956477110919803,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 7578
    },
    {
      "epoch": 0.07579,
      "grad_norm": 0.6349724495359681,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 7579
    },
    {
      "epoch": 0.0758,
      "grad_norm": 0.5747733937772262,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 7580
    },
    {
      "epoch": 0.07581,
      "grad_norm": 0.5324184467786447,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 7581
    },
    {
      "epoch": 0.07582,
      "grad_norm": 0.5131944070203941,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 7582
    },
    {
      "epoch": 0.07583,
      "grad_norm": 0.5085404496587494,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 7583
    },
    {
      "epoch": 0.07584,
      "grad_norm": 0.541822111982503,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 7584
    },
    {
      "epoch": 0.07585,
      "grad_norm": 0.61290315130478,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 7585
    },
    {
      "epoch": 0.07586,
      "grad_norm": 0.7597528207897615,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 7586
    },
    {
      "epoch": 0.07587,
      "grad_norm": 1.0042272950879196,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 7587
    },
    {
      "epoch": 0.07588,
      "grad_norm": 1.1991037951904149,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 7588
    },
    {
      "epoch": 0.07589,
      "grad_norm": 0.7285232286835313,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 7589
    },
    {
      "epoch": 0.0759,
      "grad_norm": 0.7148833074382797,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 7590
    },
    {
      "epoch": 0.07591,
      "grad_norm": 0.7201822884141769,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 7591
    },
    {
      "epoch": 0.07592,
      "grad_norm": 0.6076702679439346,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 7592
    },
    {
      "epoch": 0.07593,
      "grad_norm": 0.5718081339296276,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 7593
    },
    {
      "epoch": 0.07594,
      "grad_norm": 0.6158587196440879,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 7594
    },
    {
      "epoch": 0.07595,
      "grad_norm": 0.6174048562726181,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 7595
    },
    {
      "epoch": 0.07596,
      "grad_norm": 0.6776269755619764,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 7596
    },
    {
      "epoch": 0.07597,
      "grad_norm": 0.7579861091139625,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 7597
    },
    {
      "epoch": 0.07598,
      "grad_norm": 0.821293646598518,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 7598
    },
    {
      "epoch": 0.07599,
      "grad_norm": 0.9172481452747592,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 7599
    },
    {
      "epoch": 0.076,
      "grad_norm": 0.9188185947900437,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 7600
    },
    {
      "epoch": 0.07601,
      "grad_norm": 0.8582002375162606,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 7601
    },
    {
      "epoch": 0.07602,
      "grad_norm": 0.7498431640329858,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 7602
    },
    {
      "epoch": 0.07603,
      "grad_norm": 0.727413692498518,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 7603
    },
    {
      "epoch": 0.07604,
      "grad_norm": 0.7128563866226183,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 7604
    },
    {
      "epoch": 0.07605,
      "grad_norm": 0.6925666833494799,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 7605
    },
    {
      "epoch": 0.07606,
      "grad_norm": 0.8051319465023743,
      "learning_rate": 0.003,
      "loss": 4.1653,
      "step": 7606
    },
    {
      "epoch": 0.07607,
      "grad_norm": 0.8182132405488027,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 7607
    },
    {
      "epoch": 0.07608,
      "grad_norm": 0.8255552624586233,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 7608
    },
    {
      "epoch": 0.07609,
      "grad_norm": 0.7970747267263344,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 7609
    },
    {
      "epoch": 0.0761,
      "grad_norm": 0.8035430769234121,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 7610
    },
    {
      "epoch": 0.07611,
      "grad_norm": 0.7414024082233376,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 7611
    },
    {
      "epoch": 0.07612,
      "grad_norm": 0.6738484686209927,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 7612
    },
    {
      "epoch": 0.07613,
      "grad_norm": 0.6313697370107805,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 7613
    },
    {
      "epoch": 0.07614,
      "grad_norm": 0.6779404296006304,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 7614
    },
    {
      "epoch": 0.07615,
      "grad_norm": 0.671043709478893,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 7615
    },
    {
      "epoch": 0.07616,
      "grad_norm": 0.648522037704203,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 7616
    },
    {
      "epoch": 0.07617,
      "grad_norm": 0.6992220289841787,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 7617
    },
    {
      "epoch": 0.07618,
      "grad_norm": 0.7961453102996263,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 7618
    },
    {
      "epoch": 0.07619,
      "grad_norm": 0.6675342147329035,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 7619
    },
    {
      "epoch": 0.0762,
      "grad_norm": 0.5717583537378645,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 7620
    },
    {
      "epoch": 0.07621,
      "grad_norm": 0.5677927273573751,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 7621
    },
    {
      "epoch": 0.07622,
      "grad_norm": 0.6433993271028621,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 7622
    },
    {
      "epoch": 0.07623,
      "grad_norm": 0.6940232002524971,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 7623
    },
    {
      "epoch": 0.07624,
      "grad_norm": 0.7880990684776099,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 7624
    },
    {
      "epoch": 0.07625,
      "grad_norm": 0.8639480293754638,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 7625
    },
    {
      "epoch": 0.07626,
      "grad_norm": 0.8696121558470246,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 7626
    },
    {
      "epoch": 0.07627,
      "grad_norm": 0.6599012014594735,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 7627
    },
    {
      "epoch": 0.07628,
      "grad_norm": 0.6055187589728076,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 7628
    },
    {
      "epoch": 0.07629,
      "grad_norm": 0.5602974302890468,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 7629
    },
    {
      "epoch": 0.0763,
      "grad_norm": 0.5483897493872879,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 7630
    },
    {
      "epoch": 0.07631,
      "grad_norm": 0.5313336121851904,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 7631
    },
    {
      "epoch": 0.07632,
      "grad_norm": 0.4946584392262865,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 7632
    },
    {
      "epoch": 0.07633,
      "grad_norm": 0.4679730906531611,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 7633
    },
    {
      "epoch": 0.07634,
      "grad_norm": 0.5163571664886781,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 7634
    },
    {
      "epoch": 0.07635,
      "grad_norm": 0.5766441343095514,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 7635
    },
    {
      "epoch": 0.07636,
      "grad_norm": 0.6356527311929989,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 7636
    },
    {
      "epoch": 0.07637,
      "grad_norm": 0.7120458591171308,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 7637
    },
    {
      "epoch": 0.07638,
      "grad_norm": 0.7256150029226924,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 7638
    },
    {
      "epoch": 0.07639,
      "grad_norm": 0.6685928629135915,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 7639
    },
    {
      "epoch": 0.0764,
      "grad_norm": 0.7914866586728011,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 7640
    },
    {
      "epoch": 0.07641,
      "grad_norm": 0.9183061054040322,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 7641
    },
    {
      "epoch": 0.07642,
      "grad_norm": 0.8815667821813041,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 7642
    },
    {
      "epoch": 0.07643,
      "grad_norm": 0.6789085139539063,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 7643
    },
    {
      "epoch": 0.07644,
      "grad_norm": 0.6670039351921899,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 7644
    },
    {
      "epoch": 0.07645,
      "grad_norm": 0.6793245893987427,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 7645
    },
    {
      "epoch": 0.07646,
      "grad_norm": 0.6629748478851417,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 7646
    },
    {
      "epoch": 0.07647,
      "grad_norm": 0.7288187662625558,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 7647
    },
    {
      "epoch": 0.07648,
      "grad_norm": 0.889407265953888,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 7648
    },
    {
      "epoch": 0.07649,
      "grad_norm": 0.9687671903338628,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7649
    },
    {
      "epoch": 0.0765,
      "grad_norm": 0.8049244708990047,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 7650
    },
    {
      "epoch": 0.07651,
      "grad_norm": 0.749604908748907,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 7651
    },
    {
      "epoch": 0.07652,
      "grad_norm": 0.7585566864542461,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 7652
    },
    {
      "epoch": 0.07653,
      "grad_norm": 0.9516693847937671,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 7653
    },
    {
      "epoch": 0.07654,
      "grad_norm": 1.0530156043085126,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 7654
    },
    {
      "epoch": 0.07655,
      "grad_norm": 0.8372129657311617,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 7655
    },
    {
      "epoch": 0.07656,
      "grad_norm": 0.7407223152315993,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 7656
    },
    {
      "epoch": 0.07657,
      "grad_norm": 0.7336956785627904,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 7657
    },
    {
      "epoch": 0.07658,
      "grad_norm": 0.672687938805935,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 7658
    },
    {
      "epoch": 0.07659,
      "grad_norm": 0.7385569673711774,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 7659
    },
    {
      "epoch": 0.0766,
      "grad_norm": 0.8172458837525138,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 7660
    },
    {
      "epoch": 0.07661,
      "grad_norm": 0.9105650740952083,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 7661
    },
    {
      "epoch": 0.07662,
      "grad_norm": 0.9795402229933363,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 7662
    },
    {
      "epoch": 0.07663,
      "grad_norm": 0.9659565857187145,
      "learning_rate": 0.003,
      "loss": 4.1485,
      "step": 7663
    },
    {
      "epoch": 0.07664,
      "grad_norm": 0.7426641164866011,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 7664
    },
    {
      "epoch": 0.07665,
      "grad_norm": 0.7654812929638697,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 7665
    },
    {
      "epoch": 0.07666,
      "grad_norm": 0.7454710802815059,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 7666
    },
    {
      "epoch": 0.07667,
      "grad_norm": 0.7064554873587587,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 7667
    },
    {
      "epoch": 0.07668,
      "grad_norm": 0.6257727862295976,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7668
    },
    {
      "epoch": 0.07669,
      "grad_norm": 0.5710711731797409,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 7669
    },
    {
      "epoch": 0.0767,
      "grad_norm": 0.528257520997173,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 7670
    },
    {
      "epoch": 0.07671,
      "grad_norm": 0.4959008285709095,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 7671
    },
    {
      "epoch": 0.07672,
      "grad_norm": 0.45441825603578956,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 7672
    },
    {
      "epoch": 0.07673,
      "grad_norm": 0.4429820345077854,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 7673
    },
    {
      "epoch": 0.07674,
      "grad_norm": 0.42319123978862466,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 7674
    },
    {
      "epoch": 0.07675,
      "grad_norm": 0.3845278102823501,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7675
    },
    {
      "epoch": 0.07676,
      "grad_norm": 0.41229574434392646,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 7676
    },
    {
      "epoch": 0.07677,
      "grad_norm": 0.43520964656733063,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 7677
    },
    {
      "epoch": 0.07678,
      "grad_norm": 0.44518003939309553,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 7678
    },
    {
      "epoch": 0.07679,
      "grad_norm": 0.5155228564191591,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 7679
    },
    {
      "epoch": 0.0768,
      "grad_norm": 0.5596883999233748,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 7680
    },
    {
      "epoch": 0.07681,
      "grad_norm": 0.5906511007621373,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 7681
    },
    {
      "epoch": 0.07682,
      "grad_norm": 0.6495617626958452,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 7682
    },
    {
      "epoch": 0.07683,
      "grad_norm": 0.813132646521943,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 7683
    },
    {
      "epoch": 0.07684,
      "grad_norm": 0.9701002240236403,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 7684
    },
    {
      "epoch": 0.07685,
      "grad_norm": 1.0224940781449272,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 7685
    },
    {
      "epoch": 0.07686,
      "grad_norm": 0.8278902572767264,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7686
    },
    {
      "epoch": 0.07687,
      "grad_norm": 0.8042154248471763,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 7687
    },
    {
      "epoch": 0.07688,
      "grad_norm": 0.8028648725581159,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 7688
    },
    {
      "epoch": 0.07689,
      "grad_norm": 0.7584652571831005,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 7689
    },
    {
      "epoch": 0.0769,
      "grad_norm": 0.668275319980837,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 7690
    },
    {
      "epoch": 0.07691,
      "grad_norm": 0.7293348633493727,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 7691
    },
    {
      "epoch": 0.07692,
      "grad_norm": 0.72632873093583,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 7692
    },
    {
      "epoch": 0.07693,
      "grad_norm": 0.7772222742789282,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 7693
    },
    {
      "epoch": 0.07694,
      "grad_norm": 0.844518033365835,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 7694
    },
    {
      "epoch": 0.07695,
      "grad_norm": 1.0562783279259773,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 7695
    },
    {
      "epoch": 0.07696,
      "grad_norm": 0.9678281861932773,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 7696
    },
    {
      "epoch": 0.07697,
      "grad_norm": 0.8583167779532994,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 7697
    },
    {
      "epoch": 0.07698,
      "grad_norm": 0.7254348536667942,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 7698
    },
    {
      "epoch": 0.07699,
      "grad_norm": 0.7025270777281947,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 7699
    },
    {
      "epoch": 0.077,
      "grad_norm": 0.6298853792744631,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7700
    },
    {
      "epoch": 0.07701,
      "grad_norm": 0.6529473251497263,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 7701
    },
    {
      "epoch": 0.07702,
      "grad_norm": 0.6618947976576033,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 7702
    },
    {
      "epoch": 0.07703,
      "grad_norm": 0.720492970627025,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 7703
    },
    {
      "epoch": 0.07704,
      "grad_norm": 0.7521627176606427,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 7704
    },
    {
      "epoch": 0.07705,
      "grad_norm": 0.6974134473890328,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 7705
    },
    {
      "epoch": 0.07706,
      "grad_norm": 0.6519559813236994,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 7706
    },
    {
      "epoch": 0.07707,
      "grad_norm": 0.7138065901393107,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 7707
    },
    {
      "epoch": 0.07708,
      "grad_norm": 0.7603897989016979,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 7708
    },
    {
      "epoch": 0.07709,
      "grad_norm": 0.7728171935175017,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 7709
    },
    {
      "epoch": 0.0771,
      "grad_norm": 0.6846056023357585,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 7710
    },
    {
      "epoch": 0.07711,
      "grad_norm": 0.6649417137508592,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 7711
    },
    {
      "epoch": 0.07712,
      "grad_norm": 0.7242895985336029,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 7712
    },
    {
      "epoch": 0.07713,
      "grad_norm": 0.7939906705321689,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 7713
    },
    {
      "epoch": 0.07714,
      "grad_norm": 0.8212321036889528,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 7714
    },
    {
      "epoch": 0.07715,
      "grad_norm": 0.8121576436599612,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 7715
    },
    {
      "epoch": 0.07716,
      "grad_norm": 0.7330469856352435,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 7716
    },
    {
      "epoch": 0.07717,
      "grad_norm": 0.6433922965206615,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 7717
    },
    {
      "epoch": 0.07718,
      "grad_norm": 0.5861977103583862,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 7718
    },
    {
      "epoch": 0.07719,
      "grad_norm": 0.5228180617616236,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 7719
    },
    {
      "epoch": 0.0772,
      "grad_norm": 0.4876233685514838,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 7720
    },
    {
      "epoch": 0.07721,
      "grad_norm": 0.5225866371833879,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 7721
    },
    {
      "epoch": 0.07722,
      "grad_norm": 0.49420226274527473,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 7722
    },
    {
      "epoch": 0.07723,
      "grad_norm": 0.5163143220392981,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 7723
    },
    {
      "epoch": 0.07724,
      "grad_norm": 0.5183943056733131,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 7724
    },
    {
      "epoch": 0.07725,
      "grad_norm": 0.544712728586757,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 7725
    },
    {
      "epoch": 0.07726,
      "grad_norm": 0.48659381970361326,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7726
    },
    {
      "epoch": 0.07727,
      "grad_norm": 0.48012804652416147,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 7727
    },
    {
      "epoch": 0.07728,
      "grad_norm": 0.611442740466274,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 7728
    },
    {
      "epoch": 0.07729,
      "grad_norm": 0.7927419006950845,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 7729
    },
    {
      "epoch": 0.0773,
      "grad_norm": 1.0647142259813651,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 7730
    },
    {
      "epoch": 0.07731,
      "grad_norm": 1.2555263685268876,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7731
    },
    {
      "epoch": 0.07732,
      "grad_norm": 0.7349062778385043,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 7732
    },
    {
      "epoch": 0.07733,
      "grad_norm": 0.662067738555623,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 7733
    },
    {
      "epoch": 0.07734,
      "grad_norm": 0.6622088276312383,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 7734
    },
    {
      "epoch": 0.07735,
      "grad_norm": 0.7190113835925798,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 7735
    },
    {
      "epoch": 0.07736,
      "grad_norm": 0.779921763945562,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 7736
    },
    {
      "epoch": 0.07737,
      "grad_norm": 0.7062610550634124,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 7737
    },
    {
      "epoch": 0.07738,
      "grad_norm": 0.781831296452957,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 7738
    },
    {
      "epoch": 0.07739,
      "grad_norm": 0.881490679155474,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 7739
    },
    {
      "epoch": 0.0774,
      "grad_norm": 0.9161291546076472,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 7740
    },
    {
      "epoch": 0.07741,
      "grad_norm": 0.9669089236249705,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 7741
    },
    {
      "epoch": 0.07742,
      "grad_norm": 0.8371183452364397,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 7742
    },
    {
      "epoch": 0.07743,
      "grad_norm": 0.6183081207697575,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 7743
    },
    {
      "epoch": 0.07744,
      "grad_norm": 0.6071723474314588,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 7744
    },
    {
      "epoch": 0.07745,
      "grad_norm": 0.6556013679404881,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 7745
    },
    {
      "epoch": 0.07746,
      "grad_norm": 0.6908482013298095,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 7746
    },
    {
      "epoch": 0.07747,
      "grad_norm": 0.7453245346858474,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 7747
    },
    {
      "epoch": 0.07748,
      "grad_norm": 0.779119331038319,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 7748
    },
    {
      "epoch": 0.07749,
      "grad_norm": 0.6901079756719614,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 7749
    },
    {
      "epoch": 0.0775,
      "grad_norm": 0.7554593110119233,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 7750
    },
    {
      "epoch": 0.07751,
      "grad_norm": 0.6857296299894233,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 7751
    },
    {
      "epoch": 0.07752,
      "grad_norm": 0.6922633638861819,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 7752
    },
    {
      "epoch": 0.07753,
      "grad_norm": 0.6594486091898183,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 7753
    },
    {
      "epoch": 0.07754,
      "grad_norm": 0.5934224681021092,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 7754
    },
    {
      "epoch": 0.07755,
      "grad_norm": 0.5188793093092156,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 7755
    },
    {
      "epoch": 0.07756,
      "grad_norm": 0.48414191971528364,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 7756
    },
    {
      "epoch": 0.07757,
      "grad_norm": 0.47273269138835095,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 7757
    },
    {
      "epoch": 0.07758,
      "grad_norm": 0.49740262524638973,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 7758
    },
    {
      "epoch": 0.07759,
      "grad_norm": 0.5065560507825277,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 7759
    },
    {
      "epoch": 0.0776,
      "grad_norm": 0.5375106355045344,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 7760
    },
    {
      "epoch": 0.07761,
      "grad_norm": 0.5513605901023222,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 7761
    },
    {
      "epoch": 0.07762,
      "grad_norm": 0.48015165347830113,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 7762
    },
    {
      "epoch": 0.07763,
      "grad_norm": 0.5167765335801594,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 7763
    },
    {
      "epoch": 0.07764,
      "grad_norm": 0.6089153068745908,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 7764
    },
    {
      "epoch": 0.07765,
      "grad_norm": 0.7391949344092991,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 7765
    },
    {
      "epoch": 0.07766,
      "grad_norm": 1.072503112575427,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 7766
    },
    {
      "epoch": 0.07767,
      "grad_norm": 1.2741364151057324,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 7767
    },
    {
      "epoch": 0.07768,
      "grad_norm": 0.8964766694765297,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 7768
    },
    {
      "epoch": 0.07769,
      "grad_norm": 1.049149406706748,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 7769
    },
    {
      "epoch": 0.0777,
      "grad_norm": 0.930557278971072,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 7770
    },
    {
      "epoch": 0.07771,
      "grad_norm": 0.8923728869689802,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 7771
    },
    {
      "epoch": 0.07772,
      "grad_norm": 0.7822123730589138,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 7772
    },
    {
      "epoch": 0.07773,
      "grad_norm": 0.7222823816657866,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 7773
    },
    {
      "epoch": 0.07774,
      "grad_norm": 0.7187204559865789,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 7774
    },
    {
      "epoch": 0.07775,
      "grad_norm": 0.7480635321015907,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 7775
    },
    {
      "epoch": 0.07776,
      "grad_norm": 0.760612608235149,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 7776
    },
    {
      "epoch": 0.07777,
      "grad_norm": 0.6882048442017037,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 7777
    },
    {
      "epoch": 0.07778,
      "grad_norm": 0.7510711884551454,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 7778
    },
    {
      "epoch": 0.07779,
      "grad_norm": 0.9393258215990358,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7779
    },
    {
      "epoch": 0.0778,
      "grad_norm": 0.9513634587238046,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 7780
    },
    {
      "epoch": 0.07781,
      "grad_norm": 0.9481811143998107,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 7781
    },
    {
      "epoch": 0.07782,
      "grad_norm": 0.9203734843556167,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 7782
    },
    {
      "epoch": 0.07783,
      "grad_norm": 0.7932108521278691,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 7783
    },
    {
      "epoch": 0.07784,
      "grad_norm": 0.7260500811767607,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 7784
    },
    {
      "epoch": 0.07785,
      "grad_norm": 0.7585529189874307,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 7785
    },
    {
      "epoch": 0.07786,
      "grad_norm": 0.6739329538427627,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 7786
    },
    {
      "epoch": 0.07787,
      "grad_norm": 0.5288700684147263,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 7787
    },
    {
      "epoch": 0.07788,
      "grad_norm": 0.5605208483007748,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 7788
    },
    {
      "epoch": 0.07789,
      "grad_norm": 0.5560131490937277,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 7789
    },
    {
      "epoch": 0.0779,
      "grad_norm": 0.5403302533220166,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 7790
    },
    {
      "epoch": 0.07791,
      "grad_norm": 0.5930794987356798,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 7791
    },
    {
      "epoch": 0.07792,
      "grad_norm": 0.6273285492908405,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 7792
    },
    {
      "epoch": 0.07793,
      "grad_norm": 0.6286748399924618,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 7793
    },
    {
      "epoch": 0.07794,
      "grad_norm": 0.6714313089602904,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 7794
    },
    {
      "epoch": 0.07795,
      "grad_norm": 0.7779410911099373,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 7795
    },
    {
      "epoch": 0.07796,
      "grad_norm": 0.9238625316835825,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 7796
    },
    {
      "epoch": 0.07797,
      "grad_norm": 0.9115617993434088,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 7797
    },
    {
      "epoch": 0.07798,
      "grad_norm": 0.8524686648853493,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 7798
    },
    {
      "epoch": 0.07799,
      "grad_norm": 0.6927867557827436,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 7799
    },
    {
      "epoch": 0.078,
      "grad_norm": 0.6008938961247131,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7800
    },
    {
      "epoch": 0.07801,
      "grad_norm": 0.6422683191074112,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 7801
    },
    {
      "epoch": 0.07802,
      "grad_norm": 0.6032600741877926,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 7802
    },
    {
      "epoch": 0.07803,
      "grad_norm": 0.5863000786481797,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 7803
    },
    {
      "epoch": 0.07804,
      "grad_norm": 0.5317230097198795,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7804
    },
    {
      "epoch": 0.07805,
      "grad_norm": 0.45647691543183516,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 7805
    },
    {
      "epoch": 0.07806,
      "grad_norm": 0.46699207392888487,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 7806
    },
    {
      "epoch": 0.07807,
      "grad_norm": 0.4393858321916459,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 7807
    },
    {
      "epoch": 0.07808,
      "grad_norm": 0.4646599647463825,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 7808
    },
    {
      "epoch": 0.07809,
      "grad_norm": 0.5658965842298878,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 7809
    },
    {
      "epoch": 0.0781,
      "grad_norm": 0.7181637740728304,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 7810
    },
    {
      "epoch": 0.07811,
      "grad_norm": 0.8109929763108267,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 7811
    },
    {
      "epoch": 0.07812,
      "grad_norm": 0.8368920397907048,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 7812
    },
    {
      "epoch": 0.07813,
      "grad_norm": 0.8903796018565326,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 7813
    },
    {
      "epoch": 0.07814,
      "grad_norm": 0.9722080091605632,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 7814
    },
    {
      "epoch": 0.07815,
      "grad_norm": 0.9736337448735193,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 7815
    },
    {
      "epoch": 0.07816,
      "grad_norm": 1.0961292869832548,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 7816
    },
    {
      "epoch": 0.07817,
      "grad_norm": 1.1565923993465608,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 7817
    },
    {
      "epoch": 0.07818,
      "grad_norm": 0.8585417973652822,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7818
    },
    {
      "epoch": 0.07819,
      "grad_norm": 0.7416217398177496,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 7819
    },
    {
      "epoch": 0.0782,
      "grad_norm": 0.6344240220641632,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 7820
    },
    {
      "epoch": 0.07821,
      "grad_norm": 0.6739464286664512,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 7821
    },
    {
      "epoch": 0.07822,
      "grad_norm": 0.5825705065071117,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 7822
    },
    {
      "epoch": 0.07823,
      "grad_norm": 0.5723600939543713,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 7823
    },
    {
      "epoch": 0.07824,
      "grad_norm": 0.5121389736529045,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7824
    },
    {
      "epoch": 0.07825,
      "grad_norm": 0.5355001404670079,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 7825
    },
    {
      "epoch": 0.07826,
      "grad_norm": 0.527193374309933,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 7826
    },
    {
      "epoch": 0.07827,
      "grad_norm": 0.46903753993293595,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 7827
    },
    {
      "epoch": 0.07828,
      "grad_norm": 0.5421978536054349,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 7828
    },
    {
      "epoch": 0.07829,
      "grad_norm": 0.5249081779647955,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 7829
    },
    {
      "epoch": 0.0783,
      "grad_norm": 0.5049597468526409,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 7830
    },
    {
      "epoch": 0.07831,
      "grad_norm": 0.5692015688709412,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 7831
    },
    {
      "epoch": 0.07832,
      "grad_norm": 0.6743332070780026,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 7832
    },
    {
      "epoch": 0.07833,
      "grad_norm": 0.7173712685283925,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7833
    },
    {
      "epoch": 0.07834,
      "grad_norm": 0.8370657093807944,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 7834
    },
    {
      "epoch": 0.07835,
      "grad_norm": 1.136204980987355,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 7835
    },
    {
      "epoch": 0.07836,
      "grad_norm": 0.8577818143218462,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 7836
    },
    {
      "epoch": 0.07837,
      "grad_norm": 0.7112654311205949,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 7837
    },
    {
      "epoch": 0.07838,
      "grad_norm": 0.6569683165529128,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 7838
    },
    {
      "epoch": 0.07839,
      "grad_norm": 0.7565840760711471,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 7839
    },
    {
      "epoch": 0.0784,
      "grad_norm": 0.7482225327831992,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7840
    },
    {
      "epoch": 0.07841,
      "grad_norm": 0.6711480993377672,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7841
    },
    {
      "epoch": 0.07842,
      "grad_norm": 0.6679564294971032,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 7842
    },
    {
      "epoch": 0.07843,
      "grad_norm": 0.59249526868013,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 7843
    },
    {
      "epoch": 0.07844,
      "grad_norm": 0.5591695681252433,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 7844
    },
    {
      "epoch": 0.07845,
      "grad_norm": 0.6153612137689568,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 7845
    },
    {
      "epoch": 0.07846,
      "grad_norm": 0.6618403605569363,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 7846
    },
    {
      "epoch": 0.07847,
      "grad_norm": 0.6781576972563144,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 7847
    },
    {
      "epoch": 0.07848,
      "grad_norm": 0.6768216849272073,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 7848
    },
    {
      "epoch": 0.07849,
      "grad_norm": 0.7621052101869075,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 7849
    },
    {
      "epoch": 0.0785,
      "grad_norm": 0.7067170147953223,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 7850
    },
    {
      "epoch": 0.07851,
      "grad_norm": 0.7326500117988808,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 7851
    },
    {
      "epoch": 0.07852,
      "grad_norm": 0.7009642585608814,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 7852
    },
    {
      "epoch": 0.07853,
      "grad_norm": 0.7619033300792631,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 7853
    },
    {
      "epoch": 0.07854,
      "grad_norm": 0.8435597472657554,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 7854
    },
    {
      "epoch": 0.07855,
      "grad_norm": 0.8978457164901018,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 7855
    },
    {
      "epoch": 0.07856,
      "grad_norm": 1.023677440950207,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 7856
    },
    {
      "epoch": 0.07857,
      "grad_norm": 0.9231166097330715,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 7857
    },
    {
      "epoch": 0.07858,
      "grad_norm": 0.8633661145024862,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 7858
    },
    {
      "epoch": 0.07859,
      "grad_norm": 0.7860454695774618,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 7859
    },
    {
      "epoch": 0.0786,
      "grad_norm": 0.7590473523974762,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 7860
    },
    {
      "epoch": 0.07861,
      "grad_norm": 0.7742422968395642,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 7861
    },
    {
      "epoch": 0.07862,
      "grad_norm": 0.6647634614520403,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 7862
    },
    {
      "epoch": 0.07863,
      "grad_norm": 0.6187463355818706,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 7863
    },
    {
      "epoch": 0.07864,
      "grad_norm": 0.5110625417422245,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 7864
    },
    {
      "epoch": 0.07865,
      "grad_norm": 0.5020168122945233,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 7865
    },
    {
      "epoch": 0.07866,
      "grad_norm": 0.5824358186306762,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 7866
    },
    {
      "epoch": 0.07867,
      "grad_norm": 0.6255923217504209,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 7867
    },
    {
      "epoch": 0.07868,
      "grad_norm": 0.7772954380441937,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 7868
    },
    {
      "epoch": 0.07869,
      "grad_norm": 0.9860133943188509,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 7869
    },
    {
      "epoch": 0.0787,
      "grad_norm": 1.0195783394153966,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7870
    },
    {
      "epoch": 0.07871,
      "grad_norm": 0.7559459514710412,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 7871
    },
    {
      "epoch": 0.07872,
      "grad_norm": 0.9076157878017646,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 7872
    },
    {
      "epoch": 0.07873,
      "grad_norm": 0.8226175326832436,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7873
    },
    {
      "epoch": 0.07874,
      "grad_norm": 0.7406414466218076,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 7874
    },
    {
      "epoch": 0.07875,
      "grad_norm": 0.7388682612769197,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 7875
    },
    {
      "epoch": 0.07876,
      "grad_norm": 0.904519114854065,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 7876
    },
    {
      "epoch": 0.07877,
      "grad_norm": 0.9368363860931168,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7877
    },
    {
      "epoch": 0.07878,
      "grad_norm": 0.8598751903328211,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 7878
    },
    {
      "epoch": 0.07879,
      "grad_norm": 0.819312740767117,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 7879
    },
    {
      "epoch": 0.0788,
      "grad_norm": 0.7760893771538491,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 7880
    },
    {
      "epoch": 0.07881,
      "grad_norm": 0.7187534747264536,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7881
    },
    {
      "epoch": 0.07882,
      "grad_norm": 0.6858619284030656,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 7882
    },
    {
      "epoch": 0.07883,
      "grad_norm": 0.6188431011405061,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 7883
    },
    {
      "epoch": 0.07884,
      "grad_norm": 0.49119386209036325,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 7884
    },
    {
      "epoch": 0.07885,
      "grad_norm": 0.4425644032822491,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 7885
    },
    {
      "epoch": 0.07886,
      "grad_norm": 0.45859576158776894,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 7886
    },
    {
      "epoch": 0.07887,
      "grad_norm": 0.4966339618463925,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 7887
    },
    {
      "epoch": 0.07888,
      "grad_norm": 0.5978659033312088,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 7888
    },
    {
      "epoch": 0.07889,
      "grad_norm": 0.7155472899570747,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 7889
    },
    {
      "epoch": 0.0789,
      "grad_norm": 0.7993574365094266,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 7890
    },
    {
      "epoch": 0.07891,
      "grad_norm": 0.7636085550101828,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 7891
    },
    {
      "epoch": 0.07892,
      "grad_norm": 0.6885425050961269,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 7892
    },
    {
      "epoch": 0.07893,
      "grad_norm": 0.6721526677142209,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 7893
    },
    {
      "epoch": 0.07894,
      "grad_norm": 0.6152441288738106,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 7894
    },
    {
      "epoch": 0.07895,
      "grad_norm": 0.561256579962152,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 7895
    },
    {
      "epoch": 0.07896,
      "grad_norm": 0.5735961480085187,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7896
    },
    {
      "epoch": 0.07897,
      "grad_norm": 0.676934691180459,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 7897
    },
    {
      "epoch": 0.07898,
      "grad_norm": 0.7401825684646072,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 7898
    },
    {
      "epoch": 0.07899,
      "grad_norm": 0.7621955625823885,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 7899
    },
    {
      "epoch": 0.079,
      "grad_norm": 0.7808977446083992,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 7900
    },
    {
      "epoch": 0.07901,
      "grad_norm": 0.8881664429218848,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 7901
    },
    {
      "epoch": 0.07902,
      "grad_norm": 0.8885276958518861,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 7902
    },
    {
      "epoch": 0.07903,
      "grad_norm": 0.8190648495783674,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 7903
    },
    {
      "epoch": 0.07904,
      "grad_norm": 0.7646472367126418,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 7904
    },
    {
      "epoch": 0.07905,
      "grad_norm": 0.768832626042469,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 7905
    },
    {
      "epoch": 0.07906,
      "grad_norm": 0.7344374945822847,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 7906
    },
    {
      "epoch": 0.07907,
      "grad_norm": 0.5806783775411604,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 7907
    },
    {
      "epoch": 0.07908,
      "grad_norm": 0.5703784952596479,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 7908
    },
    {
      "epoch": 0.07909,
      "grad_norm": 0.5524374992586529,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 7909
    },
    {
      "epoch": 0.0791,
      "grad_norm": 0.593207520021952,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 7910
    },
    {
      "epoch": 0.07911,
      "grad_norm": 0.6796537564703014,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 7911
    },
    {
      "epoch": 0.07912,
      "grad_norm": 0.811473537909756,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 7912
    },
    {
      "epoch": 0.07913,
      "grad_norm": 0.9594020178558872,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 7913
    },
    {
      "epoch": 0.07914,
      "grad_norm": 0.8781644497544638,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 7914
    },
    {
      "epoch": 0.07915,
      "grad_norm": 0.7673157421848797,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 7915
    },
    {
      "epoch": 0.07916,
      "grad_norm": 0.7231720388780313,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7916
    },
    {
      "epoch": 0.07917,
      "grad_norm": 0.747995150418676,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7917
    },
    {
      "epoch": 0.07918,
      "grad_norm": 0.7558357775396138,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 7918
    },
    {
      "epoch": 0.07919,
      "grad_norm": 0.7473819497926562,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 7919
    },
    {
      "epoch": 0.0792,
      "grad_norm": 0.7630023142885092,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 7920
    },
    {
      "epoch": 0.07921,
      "grad_norm": 0.661482132358115,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 7921
    },
    {
      "epoch": 0.07922,
      "grad_norm": 0.6198116791527155,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 7922
    },
    {
      "epoch": 0.07923,
      "grad_norm": 0.6551552600939399,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 7923
    },
    {
      "epoch": 0.07924,
      "grad_norm": 0.6696835280064822,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 7924
    },
    {
      "epoch": 0.07925,
      "grad_norm": 0.6925330545340292,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 7925
    },
    {
      "epoch": 0.07926,
      "grad_norm": 0.7284626249714012,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 7926
    },
    {
      "epoch": 0.07927,
      "grad_norm": 0.671892879773423,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 7927
    },
    {
      "epoch": 0.07928,
      "grad_norm": 0.6609056217998242,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 7928
    },
    {
      "epoch": 0.07929,
      "grad_norm": 0.6999670786024481,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 7929
    },
    {
      "epoch": 0.0793,
      "grad_norm": 0.6958735375819932,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 7930
    },
    {
      "epoch": 0.07931,
      "grad_norm": 0.7554404336036357,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 7931
    },
    {
      "epoch": 0.07932,
      "grad_norm": 0.8243101778229396,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 7932
    },
    {
      "epoch": 0.07933,
      "grad_norm": 0.9463359506341771,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 7933
    },
    {
      "epoch": 0.07934,
      "grad_norm": 1.013600457137422,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 7934
    },
    {
      "epoch": 0.07935,
      "grad_norm": 0.8419500712154685,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7935
    },
    {
      "epoch": 0.07936,
      "grad_norm": 0.6245467403233166,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 7936
    },
    {
      "epoch": 0.07937,
      "grad_norm": 0.7486505766341088,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 7937
    },
    {
      "epoch": 0.07938,
      "grad_norm": 0.8254464470107613,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 7938
    },
    {
      "epoch": 0.07939,
      "grad_norm": 0.8503311527457658,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 7939
    },
    {
      "epoch": 0.0794,
      "grad_norm": 0.7668471916028556,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 7940
    },
    {
      "epoch": 0.07941,
      "grad_norm": 0.6283204867133623,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 7941
    },
    {
      "epoch": 0.07942,
      "grad_norm": 0.5166138501693774,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 7942
    },
    {
      "epoch": 0.07943,
      "grad_norm": 0.5021949697098655,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 7943
    },
    {
      "epoch": 0.07944,
      "grad_norm": 0.5349404843772013,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 7944
    },
    {
      "epoch": 0.07945,
      "grad_norm": 0.5952661363698953,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 7945
    },
    {
      "epoch": 0.07946,
      "grad_norm": 0.5669655747114686,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 7946
    },
    {
      "epoch": 0.07947,
      "grad_norm": 0.5763265373404088,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 7947
    },
    {
      "epoch": 0.07948,
      "grad_norm": 0.5722110776640312,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 7948
    },
    {
      "epoch": 0.07949,
      "grad_norm": 0.6056595909010904,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 7949
    },
    {
      "epoch": 0.0795,
      "grad_norm": 0.6647078565086577,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 7950
    },
    {
      "epoch": 0.07951,
      "grad_norm": 0.6314215964588701,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 7951
    },
    {
      "epoch": 0.07952,
      "grad_norm": 0.5731615532451197,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 7952
    },
    {
      "epoch": 0.07953,
      "grad_norm": 0.5227418275474421,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 7953
    },
    {
      "epoch": 0.07954,
      "grad_norm": 0.5425211491008127,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 7954
    },
    {
      "epoch": 0.07955,
      "grad_norm": 0.5562460949888285,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 7955
    },
    {
      "epoch": 0.07956,
      "grad_norm": 0.5957051466560916,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 7956
    },
    {
      "epoch": 0.07957,
      "grad_norm": 0.6056607635018184,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 7957
    },
    {
      "epoch": 0.07958,
      "grad_norm": 0.7512975150926531,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 7958
    },
    {
      "epoch": 0.07959,
      "grad_norm": 1.0080192207485463,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 7959
    },
    {
      "epoch": 0.0796,
      "grad_norm": 1.2487463146426276,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7960
    },
    {
      "epoch": 0.07961,
      "grad_norm": 0.6826790561980052,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 7961
    },
    {
      "epoch": 0.07962,
      "grad_norm": 0.7451214520986289,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 7962
    },
    {
      "epoch": 0.07963,
      "grad_norm": 0.9631217756846927,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 7963
    },
    {
      "epoch": 0.07964,
      "grad_norm": 1.1941093560148814,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7964
    },
    {
      "epoch": 0.07965,
      "grad_norm": 0.7914311934718551,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 7965
    },
    {
      "epoch": 0.07966,
      "grad_norm": 0.6290523078446999,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 7966
    },
    {
      "epoch": 0.07967,
      "grad_norm": 0.6518156345361048,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 7967
    },
    {
      "epoch": 0.07968,
      "grad_norm": 0.6214813237738295,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 7968
    },
    {
      "epoch": 0.07969,
      "grad_norm": 0.5554977099941081,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 7969
    },
    {
      "epoch": 0.0797,
      "grad_norm": 0.5640600564781908,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 7970
    },
    {
      "epoch": 0.07971,
      "grad_norm": 0.6596519402139773,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 7971
    },
    {
      "epoch": 0.07972,
      "grad_norm": 0.8545861020949079,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 7972
    },
    {
      "epoch": 0.07973,
      "grad_norm": 0.865871265631871,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 7973
    },
    {
      "epoch": 0.07974,
      "grad_norm": 0.7460236260461683,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 7974
    },
    {
      "epoch": 0.07975,
      "grad_norm": 0.6850055866586392,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 7975
    },
    {
      "epoch": 0.07976,
      "grad_norm": 0.6735013134508674,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7976
    },
    {
      "epoch": 0.07977,
      "grad_norm": 0.7405183514912751,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 7977
    },
    {
      "epoch": 0.07978,
      "grad_norm": 0.8290886310120359,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 7978
    },
    {
      "epoch": 0.07979,
      "grad_norm": 0.8453917594571124,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 7979
    },
    {
      "epoch": 0.0798,
      "grad_norm": 0.8981121368882824,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 7980
    },
    {
      "epoch": 0.07981,
      "grad_norm": 0.8246540871692223,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 7981
    },
    {
      "epoch": 0.07982,
      "grad_norm": 0.8765802803687193,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 7982
    },
    {
      "epoch": 0.07983,
      "grad_norm": 0.9675677298543592,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 7983
    },
    {
      "epoch": 0.07984,
      "grad_norm": 1.0989989237488744,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 7984
    },
    {
      "epoch": 0.07985,
      "grad_norm": 0.865566009696521,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 7985
    },
    {
      "epoch": 0.07986,
      "grad_norm": 0.7217554053695046,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7986
    },
    {
      "epoch": 0.07987,
      "grad_norm": 0.7223309169228254,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 7987
    },
    {
      "epoch": 0.07988,
      "grad_norm": 0.7601248914831606,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 7988
    },
    {
      "epoch": 0.07989,
      "grad_norm": 0.754472367102963,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 7989
    },
    {
      "epoch": 0.0799,
      "grad_norm": 0.7962535081025417,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 7990
    },
    {
      "epoch": 0.07991,
      "grad_norm": 0.7457639189267391,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 7991
    },
    {
      "epoch": 0.07992,
      "grad_norm": 0.6851938375119424,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 7992
    },
    {
      "epoch": 0.07993,
      "grad_norm": 0.6269563237493135,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 7993
    },
    {
      "epoch": 0.07994,
      "grad_norm": 0.6070969865870026,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 7994
    },
    {
      "epoch": 0.07995,
      "grad_norm": 0.5627702965519503,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 7995
    },
    {
      "epoch": 0.07996,
      "grad_norm": 0.5112924542761613,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 7996
    },
    {
      "epoch": 0.07997,
      "grad_norm": 0.598007231362175,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 7997
    },
    {
      "epoch": 0.07998,
      "grad_norm": 0.6410257731351261,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 7998
    },
    {
      "epoch": 0.07999,
      "grad_norm": 0.7458379686223753,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 7999
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7034092807624223,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 8000
    },
    {
      "epoch": 0.08001,
      "grad_norm": 0.5871105361909563,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 8001
    },
    {
      "epoch": 0.08002,
      "grad_norm": 0.5995732080621035,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 8002
    },
    {
      "epoch": 0.08003,
      "grad_norm": 0.6613967789238772,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 8003
    },
    {
      "epoch": 0.08004,
      "grad_norm": 0.6523754319361283,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 8004
    },
    {
      "epoch": 0.08005,
      "grad_norm": 0.7036409698174545,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 8005
    },
    {
      "epoch": 0.08006,
      "grad_norm": 0.7212028299745142,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 8006
    },
    {
      "epoch": 0.08007,
      "grad_norm": 0.6666231743781851,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 8007
    },
    {
      "epoch": 0.08008,
      "grad_norm": 0.7849422476214978,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 8008
    },
    {
      "epoch": 0.08009,
      "grad_norm": 0.8161881054836299,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 8009
    },
    {
      "epoch": 0.0801,
      "grad_norm": 0.7682572142792697,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 8010
    },
    {
      "epoch": 0.08011,
      "grad_norm": 0.777422634390519,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 8011
    },
    {
      "epoch": 0.08012,
      "grad_norm": 0.8100758516563243,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 8012
    },
    {
      "epoch": 0.08013,
      "grad_norm": 0.7710777007779627,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 8013
    },
    {
      "epoch": 0.08014,
      "grad_norm": 0.7527995072769255,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 8014
    },
    {
      "epoch": 0.08015,
      "grad_norm": 0.7561651584547401,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 8015
    },
    {
      "epoch": 0.08016,
      "grad_norm": 0.7035897971730628,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 8016
    },
    {
      "epoch": 0.08017,
      "grad_norm": 0.7174150406944461,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 8017
    },
    {
      "epoch": 0.08018,
      "grad_norm": 0.7624585759410506,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 8018
    },
    {
      "epoch": 0.08019,
      "grad_norm": 0.7182168882426316,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 8019
    },
    {
      "epoch": 0.0802,
      "grad_norm": 0.7048358879364466,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 8020
    },
    {
      "epoch": 0.08021,
      "grad_norm": 0.7497958878822685,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 8021
    },
    {
      "epoch": 0.08022,
      "grad_norm": 0.821004036800071,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 8022
    },
    {
      "epoch": 0.08023,
      "grad_norm": 0.9809693423098361,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 8023
    },
    {
      "epoch": 0.08024,
      "grad_norm": 1.052787459999515,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 8024
    },
    {
      "epoch": 0.08025,
      "grad_norm": 0.8450811401393215,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 8025
    },
    {
      "epoch": 0.08026,
      "grad_norm": 0.807742847907378,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 8026
    },
    {
      "epoch": 0.08027,
      "grad_norm": 0.7663131570160219,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 8027
    },
    {
      "epoch": 0.08028,
      "grad_norm": 0.6840081645216585,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 8028
    },
    {
      "epoch": 0.08029,
      "grad_norm": 0.6282444638490934,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 8029
    },
    {
      "epoch": 0.0803,
      "grad_norm": 0.5930173859419846,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 8030
    },
    {
      "epoch": 0.08031,
      "grad_norm": 0.6043437323260166,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 8031
    },
    {
      "epoch": 0.08032,
      "grad_norm": 0.6211595627768416,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 8032
    },
    {
      "epoch": 0.08033,
      "grad_norm": 0.6326856189776532,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 8033
    },
    {
      "epoch": 0.08034,
      "grad_norm": 0.7414480977291886,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 8034
    },
    {
      "epoch": 0.08035,
      "grad_norm": 0.847555227207347,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 8035
    },
    {
      "epoch": 0.08036,
      "grad_norm": 0.8976992992805168,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 8036
    },
    {
      "epoch": 0.08037,
      "grad_norm": 0.7353446869433018,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8037
    },
    {
      "epoch": 0.08038,
      "grad_norm": 0.5737562105586347,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 8038
    },
    {
      "epoch": 0.08039,
      "grad_norm": 0.6223395377390066,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 8039
    },
    {
      "epoch": 0.0804,
      "grad_norm": 0.6460407278049516,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 8040
    },
    {
      "epoch": 0.08041,
      "grad_norm": 0.6792567658949662,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 8041
    },
    {
      "epoch": 0.08042,
      "grad_norm": 0.6631671232762439,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 8042
    },
    {
      "epoch": 0.08043,
      "grad_norm": 0.7543589264318384,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 8043
    },
    {
      "epoch": 0.08044,
      "grad_norm": 0.8201149824912215,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 8044
    },
    {
      "epoch": 0.08045,
      "grad_norm": 0.8899252198530396,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 8045
    },
    {
      "epoch": 0.08046,
      "grad_norm": 0.8349673707656354,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8046
    },
    {
      "epoch": 0.08047,
      "grad_norm": 0.7510934507359202,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 8047
    },
    {
      "epoch": 0.08048,
      "grad_norm": 0.7197955096059386,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 8048
    },
    {
      "epoch": 0.08049,
      "grad_norm": 0.6884550531131161,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 8049
    },
    {
      "epoch": 0.0805,
      "grad_norm": 0.6875283286820179,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 8050
    },
    {
      "epoch": 0.08051,
      "grad_norm": 0.6837157282878973,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 8051
    },
    {
      "epoch": 0.08052,
      "grad_norm": 0.6102356496634533,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 8052
    },
    {
      "epoch": 0.08053,
      "grad_norm": 0.5591025845354667,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 8053
    },
    {
      "epoch": 0.08054,
      "grad_norm": 0.6059431986010548,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 8054
    },
    {
      "epoch": 0.08055,
      "grad_norm": 0.5808217412748479,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 8055
    },
    {
      "epoch": 0.08056,
      "grad_norm": 0.6445970901420555,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8056
    },
    {
      "epoch": 0.08057,
      "grad_norm": 0.6726343017392604,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 8057
    },
    {
      "epoch": 0.08058,
      "grad_norm": 0.5999523335836248,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 8058
    },
    {
      "epoch": 0.08059,
      "grad_norm": 0.6543038111857801,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 8059
    },
    {
      "epoch": 0.0806,
      "grad_norm": 0.6982671331368794,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 8060
    },
    {
      "epoch": 0.08061,
      "grad_norm": 0.7521101541978558,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 8061
    },
    {
      "epoch": 0.08062,
      "grad_norm": 0.8023563510336984,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 8062
    },
    {
      "epoch": 0.08063,
      "grad_norm": 0.8276610456091178,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 8063
    },
    {
      "epoch": 0.08064,
      "grad_norm": 0.9822043974280728,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 8064
    },
    {
      "epoch": 0.08065,
      "grad_norm": 1.1637487862676754,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 8065
    },
    {
      "epoch": 0.08066,
      "grad_norm": 0.9728114020342268,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 8066
    },
    {
      "epoch": 0.08067,
      "grad_norm": 0.9136508947005387,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 8067
    },
    {
      "epoch": 0.08068,
      "grad_norm": 1.069372223697199,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 8068
    },
    {
      "epoch": 0.08069,
      "grad_norm": 1.045484111398971,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 8069
    },
    {
      "epoch": 0.0807,
      "grad_norm": 0.9148751637475432,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 8070
    },
    {
      "epoch": 0.08071,
      "grad_norm": 0.7828647369529286,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 8071
    },
    {
      "epoch": 0.08072,
      "grad_norm": 0.6962246302015118,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 8072
    },
    {
      "epoch": 0.08073,
      "grad_norm": 0.7338615119494287,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 8073
    },
    {
      "epoch": 0.08074,
      "grad_norm": 0.7372129834579164,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 8074
    },
    {
      "epoch": 0.08075,
      "grad_norm": 0.6637992164097991,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 8075
    },
    {
      "epoch": 0.08076,
      "grad_norm": 0.7364712485983946,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 8076
    },
    {
      "epoch": 0.08077,
      "grad_norm": 0.8373976570049751,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 8077
    },
    {
      "epoch": 0.08078,
      "grad_norm": 0.8081955608957938,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 8078
    },
    {
      "epoch": 0.08079,
      "grad_norm": 0.6241082470180894,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 8079
    },
    {
      "epoch": 0.0808,
      "grad_norm": 0.5925082920698637,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 8080
    },
    {
      "epoch": 0.08081,
      "grad_norm": 0.5771081693438284,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 8081
    },
    {
      "epoch": 0.08082,
      "grad_norm": 0.5644671641528122,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 8082
    },
    {
      "epoch": 0.08083,
      "grad_norm": 0.5574560207893343,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 8083
    },
    {
      "epoch": 0.08084,
      "grad_norm": 0.5516245286767989,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 8084
    },
    {
      "epoch": 0.08085,
      "grad_norm": 0.4994799258132661,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 8085
    },
    {
      "epoch": 0.08086,
      "grad_norm": 0.5387898877477255,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 8086
    },
    {
      "epoch": 0.08087,
      "grad_norm": 0.5279461986880152,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 8087
    },
    {
      "epoch": 0.08088,
      "grad_norm": 0.6653142602263723,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 8088
    },
    {
      "epoch": 0.08089,
      "grad_norm": 0.7838979326031138,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 8089
    },
    {
      "epoch": 0.0809,
      "grad_norm": 0.7172870093397796,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 8090
    },
    {
      "epoch": 0.08091,
      "grad_norm": 0.6970224864376466,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 8091
    },
    {
      "epoch": 0.08092,
      "grad_norm": 0.8698816808081968,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 8092
    },
    {
      "epoch": 0.08093,
      "grad_norm": 0.8903936647822183,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 8093
    },
    {
      "epoch": 0.08094,
      "grad_norm": 0.8170859896883738,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 8094
    },
    {
      "epoch": 0.08095,
      "grad_norm": 0.630198435688837,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 8095
    },
    {
      "epoch": 0.08096,
      "grad_norm": 0.561670164851885,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 8096
    },
    {
      "epoch": 0.08097,
      "grad_norm": 0.6003643063916327,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 8097
    },
    {
      "epoch": 0.08098,
      "grad_norm": 0.5928995388219754,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 8098
    },
    {
      "epoch": 0.08099,
      "grad_norm": 0.613396859162041,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 8099
    },
    {
      "epoch": 0.081,
      "grad_norm": 0.5937607661582928,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8100
    },
    {
      "epoch": 0.08101,
      "grad_norm": 0.5633649505477989,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 8101
    },
    {
      "epoch": 0.08102,
      "grad_norm": 0.517609472371696,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 8102
    },
    {
      "epoch": 0.08103,
      "grad_norm": 0.5099084616246116,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 8103
    },
    {
      "epoch": 0.08104,
      "grad_norm": 0.5423305382438013,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 8104
    },
    {
      "epoch": 0.08105,
      "grad_norm": 0.5677547112279588,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8105
    },
    {
      "epoch": 0.08106,
      "grad_norm": 0.6327286014925814,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8106
    },
    {
      "epoch": 0.08107,
      "grad_norm": 0.5935405615580722,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 8107
    },
    {
      "epoch": 0.08108,
      "grad_norm": 0.5906936959536371,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 8108
    },
    {
      "epoch": 0.08109,
      "grad_norm": 0.7713130902488691,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 8109
    },
    {
      "epoch": 0.0811,
      "grad_norm": 1.089819423816047,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 8110
    },
    {
      "epoch": 0.08111,
      "grad_norm": 0.9827588739551303,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 8111
    },
    {
      "epoch": 0.08112,
      "grad_norm": 1.0089600330173312,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 8112
    },
    {
      "epoch": 0.08113,
      "grad_norm": 0.9171567562711044,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 8113
    },
    {
      "epoch": 0.08114,
      "grad_norm": 0.8631573233521649,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 8114
    },
    {
      "epoch": 0.08115,
      "grad_norm": 0.8166838272437511,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 8115
    },
    {
      "epoch": 0.08116,
      "grad_norm": 0.788874725717097,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 8116
    },
    {
      "epoch": 0.08117,
      "grad_norm": 0.7767336993314875,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 8117
    },
    {
      "epoch": 0.08118,
      "grad_norm": 0.9294129096540645,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 8118
    },
    {
      "epoch": 0.08119,
      "grad_norm": 0.8162169856495585,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 8119
    },
    {
      "epoch": 0.0812,
      "grad_norm": 0.8009999394525374,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 8120
    },
    {
      "epoch": 0.08121,
      "grad_norm": 0.8865514748225773,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 8121
    },
    {
      "epoch": 0.08122,
      "grad_norm": 0.9884968101525333,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 8122
    },
    {
      "epoch": 0.08123,
      "grad_norm": 1.0369283181463518,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 8123
    },
    {
      "epoch": 0.08124,
      "grad_norm": 0.7962033997430238,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 8124
    },
    {
      "epoch": 0.08125,
      "grad_norm": 0.6823614488148949,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 8125
    },
    {
      "epoch": 0.08126,
      "grad_norm": 0.7327655636296029,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 8126
    },
    {
      "epoch": 0.08127,
      "grad_norm": 0.7378713020854091,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 8127
    },
    {
      "epoch": 0.08128,
      "grad_norm": 0.7764634911842385,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 8128
    },
    {
      "epoch": 0.08129,
      "grad_norm": 0.767500774593417,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 8129
    },
    {
      "epoch": 0.0813,
      "grad_norm": 0.7754642716859383,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8130
    },
    {
      "epoch": 0.08131,
      "grad_norm": 0.7938632639450202,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 8131
    },
    {
      "epoch": 0.08132,
      "grad_norm": 0.7890658387507653,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 8132
    },
    {
      "epoch": 0.08133,
      "grad_norm": 0.7935294120537186,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 8133
    },
    {
      "epoch": 0.08134,
      "grad_norm": 0.8391989214547845,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 8134
    },
    {
      "epoch": 0.08135,
      "grad_norm": 0.9554035900858945,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 8135
    },
    {
      "epoch": 0.08136,
      "grad_norm": 0.9663343129111622,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 8136
    },
    {
      "epoch": 0.08137,
      "grad_norm": 0.8734638206822923,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 8137
    },
    {
      "epoch": 0.08138,
      "grad_norm": 0.9659073383764333,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 8138
    },
    {
      "epoch": 0.08139,
      "grad_norm": 1.0174128187111928,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 8139
    },
    {
      "epoch": 0.0814,
      "grad_norm": 0.8902331333269284,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 8140
    },
    {
      "epoch": 0.08141,
      "grad_norm": 0.8349574760386372,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 8141
    },
    {
      "epoch": 0.08142,
      "grad_norm": 0.7575245098644054,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 8142
    },
    {
      "epoch": 0.08143,
      "grad_norm": 0.8507077034723728,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 8143
    },
    {
      "epoch": 0.08144,
      "grad_norm": 0.9743221963844606,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 8144
    },
    {
      "epoch": 0.08145,
      "grad_norm": 0.9476119009807066,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 8145
    },
    {
      "epoch": 0.08146,
      "grad_norm": 0.9002626878014724,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 8146
    },
    {
      "epoch": 0.08147,
      "grad_norm": 0.8588231090576398,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 8147
    },
    {
      "epoch": 0.08148,
      "grad_norm": 0.7514699828415461,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 8148
    },
    {
      "epoch": 0.08149,
      "grad_norm": 0.6673420124266218,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8149
    },
    {
      "epoch": 0.0815,
      "grad_norm": 0.7038005384460582,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 8150
    },
    {
      "epoch": 0.08151,
      "grad_norm": 0.7264508313079431,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 8151
    },
    {
      "epoch": 0.08152,
      "grad_norm": 0.8385234903173766,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 8152
    },
    {
      "epoch": 0.08153,
      "grad_norm": 0.8772929826726037,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8153
    },
    {
      "epoch": 0.08154,
      "grad_norm": 0.773933052101485,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 8154
    },
    {
      "epoch": 0.08155,
      "grad_norm": 0.5418340291312099,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 8155
    },
    {
      "epoch": 0.08156,
      "grad_norm": 0.612247602405888,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 8156
    },
    {
      "epoch": 0.08157,
      "grad_norm": 0.6995306526237685,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 8157
    },
    {
      "epoch": 0.08158,
      "grad_norm": 0.6914288085424243,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 8158
    },
    {
      "epoch": 0.08159,
      "grad_norm": 0.5639754454368973,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 8159
    },
    {
      "epoch": 0.0816,
      "grad_norm": 0.5308643800912878,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 8160
    },
    {
      "epoch": 0.08161,
      "grad_norm": 0.4794329031430582,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 8161
    },
    {
      "epoch": 0.08162,
      "grad_norm": 0.3941630661211312,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 8162
    },
    {
      "epoch": 0.08163,
      "grad_norm": 0.3500016460344925,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 8163
    },
    {
      "epoch": 0.08164,
      "grad_norm": 0.36562869444919177,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 8164
    },
    {
      "epoch": 0.08165,
      "grad_norm": 0.43629694271249264,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 8165
    },
    {
      "epoch": 0.08166,
      "grad_norm": 0.5708500396948257,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8166
    },
    {
      "epoch": 0.08167,
      "grad_norm": 0.7973454394619072,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 8167
    },
    {
      "epoch": 0.08168,
      "grad_norm": 1.0166589750002435,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 8168
    },
    {
      "epoch": 0.08169,
      "grad_norm": 1.0549264180537317,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 8169
    },
    {
      "epoch": 0.0817,
      "grad_norm": 0.71968050278163,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 8170
    },
    {
      "epoch": 0.08171,
      "grad_norm": 0.6543779608988335,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 8171
    },
    {
      "epoch": 0.08172,
      "grad_norm": 0.6293237774913742,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 8172
    },
    {
      "epoch": 0.08173,
      "grad_norm": 0.6399157682768067,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 8173
    },
    {
      "epoch": 0.08174,
      "grad_norm": 0.6938917181866926,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 8174
    },
    {
      "epoch": 0.08175,
      "grad_norm": 0.6568284294958178,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 8175
    },
    {
      "epoch": 0.08176,
      "grad_norm": 0.6514625676960912,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 8176
    },
    {
      "epoch": 0.08177,
      "grad_norm": 0.6436411792736433,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 8177
    },
    {
      "epoch": 0.08178,
      "grad_norm": 0.6566610790643876,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 8178
    },
    {
      "epoch": 0.08179,
      "grad_norm": 0.6775002449384784,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 8179
    },
    {
      "epoch": 0.0818,
      "grad_norm": 0.6047058885443021,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 8180
    },
    {
      "epoch": 0.08181,
      "grad_norm": 0.49255328216218053,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 8181
    },
    {
      "epoch": 0.08182,
      "grad_norm": 0.6513781988399094,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 8182
    },
    {
      "epoch": 0.08183,
      "grad_norm": 0.756600046387577,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 8183
    },
    {
      "epoch": 0.08184,
      "grad_norm": 0.8214276567397872,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 8184
    },
    {
      "epoch": 0.08185,
      "grad_norm": 0.8752383489148959,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 8185
    },
    {
      "epoch": 0.08186,
      "grad_norm": 0.8441519707828739,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8186
    },
    {
      "epoch": 0.08187,
      "grad_norm": 0.7672475601697205,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 8187
    },
    {
      "epoch": 0.08188,
      "grad_norm": 0.7282380600276215,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 8188
    },
    {
      "epoch": 0.08189,
      "grad_norm": 0.6846705238578715,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 8189
    },
    {
      "epoch": 0.0819,
      "grad_norm": 0.7207597619767919,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 8190
    },
    {
      "epoch": 0.08191,
      "grad_norm": 0.7101247154479835,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 8191
    },
    {
      "epoch": 0.08192,
      "grad_norm": 0.7686485468774331,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 8192
    },
    {
      "epoch": 0.08193,
      "grad_norm": 0.7889428905614293,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 8193
    },
    {
      "epoch": 0.08194,
      "grad_norm": 0.9153699308382264,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 8194
    },
    {
      "epoch": 0.08195,
      "grad_norm": 0.8233344904776073,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 8195
    },
    {
      "epoch": 0.08196,
      "grad_norm": 0.6057143745665352,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 8196
    },
    {
      "epoch": 0.08197,
      "grad_norm": 0.5644875985686773,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 8197
    },
    {
      "epoch": 0.08198,
      "grad_norm": 0.5894267537410771,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 8198
    },
    {
      "epoch": 0.08199,
      "grad_norm": 0.692674390552851,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 8199
    },
    {
      "epoch": 0.082,
      "grad_norm": 0.7242086618626092,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 8200
    },
    {
      "epoch": 0.08201,
      "grad_norm": 0.7495656742172861,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 8201
    },
    {
      "epoch": 0.08202,
      "grad_norm": 0.5902754450349533,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 8202
    },
    {
      "epoch": 0.08203,
      "grad_norm": 0.4939528228400696,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8203
    },
    {
      "epoch": 0.08204,
      "grad_norm": 0.5819717473152258,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 8204
    },
    {
      "epoch": 0.08205,
      "grad_norm": 0.6610578196934844,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 8205
    },
    {
      "epoch": 0.08206,
      "grad_norm": 0.6833022930778619,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 8206
    },
    {
      "epoch": 0.08207,
      "grad_norm": 0.7109226906588522,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 8207
    },
    {
      "epoch": 0.08208,
      "grad_norm": 0.6863440404860754,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 8208
    },
    {
      "epoch": 0.08209,
      "grad_norm": 0.7395650743809461,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 8209
    },
    {
      "epoch": 0.0821,
      "grad_norm": 0.9883215726717448,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 8210
    },
    {
      "epoch": 0.08211,
      "grad_norm": 1.2385395849984975,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 8211
    },
    {
      "epoch": 0.08212,
      "grad_norm": 0.8918876274434122,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 8212
    },
    {
      "epoch": 0.08213,
      "grad_norm": 0.8581489019919268,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 8213
    },
    {
      "epoch": 0.08214,
      "grad_norm": 0.8197346780442005,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 8214
    },
    {
      "epoch": 0.08215,
      "grad_norm": 0.762507621386095,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 8215
    },
    {
      "epoch": 0.08216,
      "grad_norm": 0.7489221002140064,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 8216
    },
    {
      "epoch": 0.08217,
      "grad_norm": 0.9152826477695661,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 8217
    },
    {
      "epoch": 0.08218,
      "grad_norm": 1.0382652565254082,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 8218
    },
    {
      "epoch": 0.08219,
      "grad_norm": 1.026804807644373,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 8219
    },
    {
      "epoch": 0.0822,
      "grad_norm": 1.0424560881490594,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 8220
    },
    {
      "epoch": 0.08221,
      "grad_norm": 0.8714512313495743,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 8221
    },
    {
      "epoch": 0.08222,
      "grad_norm": 0.7651631883270994,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 8222
    },
    {
      "epoch": 0.08223,
      "grad_norm": 0.729346414097472,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 8223
    },
    {
      "epoch": 0.08224,
      "grad_norm": 0.7979340889752132,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 8224
    },
    {
      "epoch": 0.08225,
      "grad_norm": 0.7645944759378901,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 8225
    },
    {
      "epoch": 0.08226,
      "grad_norm": 0.5448138887864806,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 8226
    },
    {
      "epoch": 0.08227,
      "grad_norm": 0.5856081470610704,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 8227
    },
    {
      "epoch": 0.08228,
      "grad_norm": 0.6128538428938678,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 8228
    },
    {
      "epoch": 0.08229,
      "grad_norm": 0.5828543069405947,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 8229
    },
    {
      "epoch": 0.0823,
      "grad_norm": 0.5252226962559838,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 8230
    },
    {
      "epoch": 0.08231,
      "grad_norm": 0.5557346682459137,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 8231
    },
    {
      "epoch": 0.08232,
      "grad_norm": 0.5650481326541311,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 8232
    },
    {
      "epoch": 0.08233,
      "grad_norm": 0.5634969006102941,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8233
    },
    {
      "epoch": 0.08234,
      "grad_norm": 0.5450247074135766,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 8234
    },
    {
      "epoch": 0.08235,
      "grad_norm": 0.499938817117624,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 8235
    },
    {
      "epoch": 0.08236,
      "grad_norm": 0.5296381965012722,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 8236
    },
    {
      "epoch": 0.08237,
      "grad_norm": 0.547808654408303,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 8237
    },
    {
      "epoch": 0.08238,
      "grad_norm": 0.5329237503159692,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 8238
    },
    {
      "epoch": 0.08239,
      "grad_norm": 0.6012253849536762,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 8239
    },
    {
      "epoch": 0.0824,
      "grad_norm": 0.7137927518105237,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 8240
    },
    {
      "epoch": 0.08241,
      "grad_norm": 0.9032854526366563,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8241
    },
    {
      "epoch": 0.08242,
      "grad_norm": 1.2043691943864299,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 8242
    },
    {
      "epoch": 0.08243,
      "grad_norm": 0.8643493496322304,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 8243
    },
    {
      "epoch": 0.08244,
      "grad_norm": 0.8144027974165489,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 8244
    },
    {
      "epoch": 0.08245,
      "grad_norm": 0.7635515232637834,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 8245
    },
    {
      "epoch": 0.08246,
      "grad_norm": 0.7163154200389772,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 8246
    },
    {
      "epoch": 0.08247,
      "grad_norm": 0.748090338264524,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 8247
    },
    {
      "epoch": 0.08248,
      "grad_norm": 0.5601792675667456,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 8248
    },
    {
      "epoch": 0.08249,
      "grad_norm": 0.5660826911158605,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 8249
    },
    {
      "epoch": 0.0825,
      "grad_norm": 0.49777269961971965,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 8250
    },
    {
      "epoch": 0.08251,
      "grad_norm": 0.4749763269578897,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 8251
    },
    {
      "epoch": 0.08252,
      "grad_norm": 0.5367890821268732,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 8252
    },
    {
      "epoch": 0.08253,
      "grad_norm": 0.6209172018514464,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8253
    },
    {
      "epoch": 0.08254,
      "grad_norm": 0.7460752011518125,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 8254
    },
    {
      "epoch": 0.08255,
      "grad_norm": 0.8859483471117526,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 8255
    },
    {
      "epoch": 0.08256,
      "grad_norm": 0.984289350625384,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 8256
    },
    {
      "epoch": 0.08257,
      "grad_norm": 1.0835433642153152,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8257
    },
    {
      "epoch": 0.08258,
      "grad_norm": 0.8632204878935943,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 8258
    },
    {
      "epoch": 0.08259,
      "grad_norm": 0.7275448679612547,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 8259
    },
    {
      "epoch": 0.0826,
      "grad_norm": 0.6773298349818122,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 8260
    },
    {
      "epoch": 0.08261,
      "grad_norm": 0.7281743646317274,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 8261
    },
    {
      "epoch": 0.08262,
      "grad_norm": 0.9373357855734575,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 8262
    },
    {
      "epoch": 0.08263,
      "grad_norm": 0.9851510358265243,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 8263
    },
    {
      "epoch": 0.08264,
      "grad_norm": 0.9213918489568741,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 8264
    },
    {
      "epoch": 0.08265,
      "grad_norm": 0.8995466212053259,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 8265
    },
    {
      "epoch": 0.08266,
      "grad_norm": 0.8095431576329132,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 8266
    },
    {
      "epoch": 0.08267,
      "grad_norm": 0.8603884649749075,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 8267
    },
    {
      "epoch": 0.08268,
      "grad_norm": 0.8697551596264255,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 8268
    },
    {
      "epoch": 0.08269,
      "grad_norm": 0.8766282707684185,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 8269
    },
    {
      "epoch": 0.0827,
      "grad_norm": 0.7471962671773691,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8270
    },
    {
      "epoch": 0.08271,
      "grad_norm": 0.6435084919846199,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 8271
    },
    {
      "epoch": 0.08272,
      "grad_norm": 0.5510784750295802,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 8272
    },
    {
      "epoch": 0.08273,
      "grad_norm": 0.5988908318224689,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 8273
    },
    {
      "epoch": 0.08274,
      "grad_norm": 0.6072551102960675,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 8274
    },
    {
      "epoch": 0.08275,
      "grad_norm": 0.6062453075526276,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8275
    },
    {
      "epoch": 0.08276,
      "grad_norm": 0.5278924020145801,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 8276
    },
    {
      "epoch": 0.08277,
      "grad_norm": 0.5216351289934781,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 8277
    },
    {
      "epoch": 0.08278,
      "grad_norm": 0.5235445369007313,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 8278
    },
    {
      "epoch": 0.08279,
      "grad_norm": 0.4302544923490237,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 8279
    },
    {
      "epoch": 0.0828,
      "grad_norm": 0.45314513906146026,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 8280
    },
    {
      "epoch": 0.08281,
      "grad_norm": 0.4947819664319072,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 8281
    },
    {
      "epoch": 0.08282,
      "grad_norm": 0.4700115344785109,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 8282
    },
    {
      "epoch": 0.08283,
      "grad_norm": 0.47458184523582214,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 8283
    },
    {
      "epoch": 0.08284,
      "grad_norm": 0.6217194123770353,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 8284
    },
    {
      "epoch": 0.08285,
      "grad_norm": 0.7679318114887002,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 8285
    },
    {
      "epoch": 0.08286,
      "grad_norm": 1.0091565333262968,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 8286
    },
    {
      "epoch": 0.08287,
      "grad_norm": 1.0308871533238988,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8287
    },
    {
      "epoch": 0.08288,
      "grad_norm": 0.7715553741553134,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 8288
    },
    {
      "epoch": 0.08289,
      "grad_norm": 0.7246211823017321,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 8289
    },
    {
      "epoch": 0.0829,
      "grad_norm": 0.6693388476539852,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 8290
    },
    {
      "epoch": 0.08291,
      "grad_norm": 0.6792730649441251,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 8291
    },
    {
      "epoch": 0.08292,
      "grad_norm": 0.715764254394465,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 8292
    },
    {
      "epoch": 0.08293,
      "grad_norm": 0.8330895077359063,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 8293
    },
    {
      "epoch": 0.08294,
      "grad_norm": 0.8999946195152977,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 8294
    },
    {
      "epoch": 0.08295,
      "grad_norm": 0.9722144185539187,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 8295
    },
    {
      "epoch": 0.08296,
      "grad_norm": 0.8945126241906354,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 8296
    },
    {
      "epoch": 0.08297,
      "grad_norm": 0.931976526589576,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 8297
    },
    {
      "epoch": 0.08298,
      "grad_norm": 1.0550600149143023,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 8298
    },
    {
      "epoch": 0.08299,
      "grad_norm": 1.0038590631217534,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 8299
    },
    {
      "epoch": 0.083,
      "grad_norm": 1.0483622801143657,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 8300
    },
    {
      "epoch": 0.08301,
      "grad_norm": 0.8663552032951759,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 8301
    },
    {
      "epoch": 0.08302,
      "grad_norm": 0.9615360827126928,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 8302
    },
    {
      "epoch": 0.08303,
      "grad_norm": 0.8375481767872405,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 8303
    },
    {
      "epoch": 0.08304,
      "grad_norm": 0.7453597337442939,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 8304
    },
    {
      "epoch": 0.08305,
      "grad_norm": 0.7131331051227506,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 8305
    },
    {
      "epoch": 0.08306,
      "grad_norm": 0.7124866649665565,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 8306
    },
    {
      "epoch": 0.08307,
      "grad_norm": 0.8250868166173586,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 8307
    },
    {
      "epoch": 0.08308,
      "grad_norm": 0.9061398647327097,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 8308
    },
    {
      "epoch": 0.08309,
      "grad_norm": 1.0307311740161893,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 8309
    },
    {
      "epoch": 0.0831,
      "grad_norm": 0.992744547897445,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 8310
    },
    {
      "epoch": 0.08311,
      "grad_norm": 0.7488997564809391,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 8311
    },
    {
      "epoch": 0.08312,
      "grad_norm": 0.7053261274926852,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 8312
    },
    {
      "epoch": 0.08313,
      "grad_norm": 0.6593948632063114,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8313
    },
    {
      "epoch": 0.08314,
      "grad_norm": 0.6995852846034631,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 8314
    },
    {
      "epoch": 0.08315,
      "grad_norm": 0.7002924867496525,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 8315
    },
    {
      "epoch": 0.08316,
      "grad_norm": 0.754368216614145,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 8316
    },
    {
      "epoch": 0.08317,
      "grad_norm": 0.7750729970073041,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8317
    },
    {
      "epoch": 0.08318,
      "grad_norm": 0.6364553809945898,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 8318
    },
    {
      "epoch": 0.08319,
      "grad_norm": 0.5773708268407741,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 8319
    },
    {
      "epoch": 0.0832,
      "grad_norm": 0.6001536418360034,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 8320
    },
    {
      "epoch": 0.08321,
      "grad_norm": 0.5089768075103857,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 8321
    },
    {
      "epoch": 0.08322,
      "grad_norm": 0.518428766042433,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 8322
    },
    {
      "epoch": 0.08323,
      "grad_norm": 0.5173998718635492,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 8323
    },
    {
      "epoch": 0.08324,
      "grad_norm": 0.4750014989168887,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 8324
    },
    {
      "epoch": 0.08325,
      "grad_norm": 0.5855181107353905,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 8325
    },
    {
      "epoch": 0.08326,
      "grad_norm": 0.7628006782684289,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 8326
    },
    {
      "epoch": 0.08327,
      "grad_norm": 1.0618146389847019,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 8327
    },
    {
      "epoch": 0.08328,
      "grad_norm": 0.9388504158158667,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 8328
    },
    {
      "epoch": 0.08329,
      "grad_norm": 0.6800766818090532,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 8329
    },
    {
      "epoch": 0.0833,
      "grad_norm": 0.5773008995872342,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 8330
    },
    {
      "epoch": 0.08331,
      "grad_norm": 0.5748792212675666,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 8331
    },
    {
      "epoch": 0.08332,
      "grad_norm": 0.5541285891852565,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8332
    },
    {
      "epoch": 0.08333,
      "grad_norm": 0.5416560426836903,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 8333
    },
    {
      "epoch": 0.08334,
      "grad_norm": 0.5432886614396639,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 8334
    },
    {
      "epoch": 0.08335,
      "grad_norm": 0.6393368606314609,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 8335
    },
    {
      "epoch": 0.08336,
      "grad_norm": 0.670088341618419,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 8336
    },
    {
      "epoch": 0.08337,
      "grad_norm": 0.759669728240697,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 8337
    },
    {
      "epoch": 0.08338,
      "grad_norm": 0.7670617917071586,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 8338
    },
    {
      "epoch": 0.08339,
      "grad_norm": 0.7599957118398699,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 8339
    },
    {
      "epoch": 0.0834,
      "grad_norm": 0.8032482037405414,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 8340
    },
    {
      "epoch": 0.08341,
      "grad_norm": 0.7742413499026626,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 8341
    },
    {
      "epoch": 0.08342,
      "grad_norm": 0.7647103140349684,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 8342
    },
    {
      "epoch": 0.08343,
      "grad_norm": 0.6857455040634841,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 8343
    },
    {
      "epoch": 0.08344,
      "grad_norm": 0.6285429482356256,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 8344
    },
    {
      "epoch": 0.08345,
      "grad_norm": 0.7221980546699931,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 8345
    },
    {
      "epoch": 0.08346,
      "grad_norm": 0.7143054001868141,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 8346
    },
    {
      "epoch": 0.08347,
      "grad_norm": 0.6385179914801554,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 8347
    },
    {
      "epoch": 0.08348,
      "grad_norm": 0.6990845577714458,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 8348
    },
    {
      "epoch": 0.08349,
      "grad_norm": 0.6913429696640464,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 8349
    },
    {
      "epoch": 0.0835,
      "grad_norm": 0.7192577771703562,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 8350
    },
    {
      "epoch": 0.08351,
      "grad_norm": 0.7760395598930231,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 8351
    },
    {
      "epoch": 0.08352,
      "grad_norm": 0.7993190167350477,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 8352
    },
    {
      "epoch": 0.08353,
      "grad_norm": 0.7398688625286849,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 8353
    },
    {
      "epoch": 0.08354,
      "grad_norm": 0.6834075220306864,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 8354
    },
    {
      "epoch": 0.08355,
      "grad_norm": 0.8175214131818854,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 8355
    },
    {
      "epoch": 0.08356,
      "grad_norm": 0.9414888867819124,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 8356
    },
    {
      "epoch": 0.08357,
      "grad_norm": 0.960843920438772,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 8357
    },
    {
      "epoch": 0.08358,
      "grad_norm": 0.9609403468914611,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 8358
    },
    {
      "epoch": 0.08359,
      "grad_norm": 0.9295346213960959,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 8359
    },
    {
      "epoch": 0.0836,
      "grad_norm": 0.9085274043267699,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 8360
    },
    {
      "epoch": 0.08361,
      "grad_norm": 0.8071363639869016,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8361
    },
    {
      "epoch": 0.08362,
      "grad_norm": 0.9382775907297787,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 8362
    },
    {
      "epoch": 0.08363,
      "grad_norm": 0.8709824514663582,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 8363
    },
    {
      "epoch": 0.08364,
      "grad_norm": 0.675740458123866,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 8364
    },
    {
      "epoch": 0.08365,
      "grad_norm": 0.6496018714951994,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 8365
    },
    {
      "epoch": 0.08366,
      "grad_norm": 0.6610790017908971,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 8366
    },
    {
      "epoch": 0.08367,
      "grad_norm": 0.6766179312441892,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 8367
    },
    {
      "epoch": 0.08368,
      "grad_norm": 0.6048963944008922,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 8368
    },
    {
      "epoch": 0.08369,
      "grad_norm": 0.5577980801642201,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 8369
    },
    {
      "epoch": 0.0837,
      "grad_norm": 0.5452065922923452,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 8370
    },
    {
      "epoch": 0.08371,
      "grad_norm": 0.5328227419403814,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 8371
    },
    {
      "epoch": 0.08372,
      "grad_norm": 0.5572436569854103,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 8372
    },
    {
      "epoch": 0.08373,
      "grad_norm": 0.595024680031319,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 8373
    },
    {
      "epoch": 0.08374,
      "grad_norm": 0.6171013061042033,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 8374
    },
    {
      "epoch": 0.08375,
      "grad_norm": 0.6632173727246423,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 8375
    },
    {
      "epoch": 0.08376,
      "grad_norm": 0.7586043969017534,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 8376
    },
    {
      "epoch": 0.08377,
      "grad_norm": 0.8201258736560947,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 8377
    },
    {
      "epoch": 0.08378,
      "grad_norm": 0.7660059240272924,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 8378
    },
    {
      "epoch": 0.08379,
      "grad_norm": 0.6820454577432933,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 8379
    },
    {
      "epoch": 0.0838,
      "grad_norm": 0.6391725474755902,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 8380
    },
    {
      "epoch": 0.08381,
      "grad_norm": 0.626073359728838,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 8381
    },
    {
      "epoch": 0.08382,
      "grad_norm": 0.6537353438521994,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 8382
    },
    {
      "epoch": 0.08383,
      "grad_norm": 0.5286621728800187,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 8383
    },
    {
      "epoch": 0.08384,
      "grad_norm": 0.49570196885919826,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 8384
    },
    {
      "epoch": 0.08385,
      "grad_norm": 0.4720817397929105,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 8385
    },
    {
      "epoch": 0.08386,
      "grad_norm": 0.4116758591839013,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 8386
    },
    {
      "epoch": 0.08387,
      "grad_norm": 0.45306876446030137,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 8387
    },
    {
      "epoch": 0.08388,
      "grad_norm": 0.6179847808239425,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 8388
    },
    {
      "epoch": 0.08389,
      "grad_norm": 0.9294947853793362,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 8389
    },
    {
      "epoch": 0.0839,
      "grad_norm": 1.2031306786524287,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 8390
    },
    {
      "epoch": 0.08391,
      "grad_norm": 0.8046262538632072,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 8391
    },
    {
      "epoch": 0.08392,
      "grad_norm": 0.882493419185855,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 8392
    },
    {
      "epoch": 0.08393,
      "grad_norm": 0.8613668633518297,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 8393
    },
    {
      "epoch": 0.08394,
      "grad_norm": 0.8844121298915572,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 8394
    },
    {
      "epoch": 0.08395,
      "grad_norm": 0.9156972707334563,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 8395
    },
    {
      "epoch": 0.08396,
      "grad_norm": 0.8431255266855052,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 8396
    },
    {
      "epoch": 0.08397,
      "grad_norm": 0.8830670687070312,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 8397
    },
    {
      "epoch": 0.08398,
      "grad_norm": 0.9314448769669744,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 8398
    },
    {
      "epoch": 0.08399,
      "grad_norm": 1.0449422905278423,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 8399
    },
    {
      "epoch": 0.084,
      "grad_norm": 1.0703912427847535,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 8400
    },
    {
      "epoch": 0.08401,
      "grad_norm": 0.99544544738125,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8401
    },
    {
      "epoch": 0.08402,
      "grad_norm": 0.9137811882497271,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 8402
    },
    {
      "epoch": 0.08403,
      "grad_norm": 0.8345353163377343,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 8403
    },
    {
      "epoch": 0.08404,
      "grad_norm": 0.7423619551465458,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 8404
    },
    {
      "epoch": 0.08405,
      "grad_norm": 0.6917338055216923,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 8405
    },
    {
      "epoch": 0.08406,
      "grad_norm": 0.6541321327571565,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 8406
    },
    {
      "epoch": 0.08407,
      "grad_norm": 0.6748689162799533,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 8407
    },
    {
      "epoch": 0.08408,
      "grad_norm": 0.6636547398677606,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 8408
    },
    {
      "epoch": 0.08409,
      "grad_norm": 0.6820702144071761,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 8409
    },
    {
      "epoch": 0.0841,
      "grad_norm": 0.6703929070086052,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 8410
    },
    {
      "epoch": 0.08411,
      "grad_norm": 0.6537163754004126,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 8411
    },
    {
      "epoch": 0.08412,
      "grad_norm": 0.6180851692722509,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 8412
    },
    {
      "epoch": 0.08413,
      "grad_norm": 0.5613447966463775,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 8413
    },
    {
      "epoch": 0.08414,
      "grad_norm": 0.5426935541427719,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 8414
    },
    {
      "epoch": 0.08415,
      "grad_norm": 0.5455362132623575,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 8415
    },
    {
      "epoch": 0.08416,
      "grad_norm": 0.5671062554035091,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 8416
    },
    {
      "epoch": 0.08417,
      "grad_norm": 0.6929581320753861,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 8417
    },
    {
      "epoch": 0.08418,
      "grad_norm": 0.7632658007580719,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 8418
    },
    {
      "epoch": 0.08419,
      "grad_norm": 0.8207298972699563,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 8419
    },
    {
      "epoch": 0.0842,
      "grad_norm": 0.8838166900993079,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 8420
    },
    {
      "epoch": 0.08421,
      "grad_norm": 0.8494239057110717,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 8421
    },
    {
      "epoch": 0.08422,
      "grad_norm": 0.7486051228897643,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 8422
    },
    {
      "epoch": 0.08423,
      "grad_norm": 0.6681036680494669,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 8423
    },
    {
      "epoch": 0.08424,
      "grad_norm": 0.6231615422695527,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 8424
    },
    {
      "epoch": 0.08425,
      "grad_norm": 0.5973371615272063,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 8425
    },
    {
      "epoch": 0.08426,
      "grad_norm": 0.590081914250373,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 8426
    },
    {
      "epoch": 0.08427,
      "grad_norm": 0.7124778189115221,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 8427
    },
    {
      "epoch": 0.08428,
      "grad_norm": 0.7887937264015177,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 8428
    },
    {
      "epoch": 0.08429,
      "grad_norm": 0.7678709133144459,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 8429
    },
    {
      "epoch": 0.0843,
      "grad_norm": 0.6332332135809398,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 8430
    },
    {
      "epoch": 0.08431,
      "grad_norm": 0.5107659296097526,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 8431
    },
    {
      "epoch": 0.08432,
      "grad_norm": 0.48858809488607446,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 8432
    },
    {
      "epoch": 0.08433,
      "grad_norm": 0.5110654376932778,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 8433
    },
    {
      "epoch": 0.08434,
      "grad_norm": 0.534360910387737,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 8434
    },
    {
      "epoch": 0.08435,
      "grad_norm": 0.5820631960297242,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 8435
    },
    {
      "epoch": 0.08436,
      "grad_norm": 0.661389118471186,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 8436
    },
    {
      "epoch": 0.08437,
      "grad_norm": 0.7111949328987919,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 8437
    },
    {
      "epoch": 0.08438,
      "grad_norm": 0.7522347549752256,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 8438
    },
    {
      "epoch": 0.08439,
      "grad_norm": 0.9493832565098858,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 8439
    },
    {
      "epoch": 0.0844,
      "grad_norm": 1.0648764993147652,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 8440
    },
    {
      "epoch": 0.08441,
      "grad_norm": 0.8414607606679769,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 8441
    },
    {
      "epoch": 0.08442,
      "grad_norm": 0.7257158697574967,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 8442
    },
    {
      "epoch": 0.08443,
      "grad_norm": 0.6399687963227431,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 8443
    },
    {
      "epoch": 0.08444,
      "grad_norm": 0.6122601189245481,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 8444
    },
    {
      "epoch": 0.08445,
      "grad_norm": 0.6421561307708404,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 8445
    },
    {
      "epoch": 0.08446,
      "grad_norm": 0.7720968877758178,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 8446
    },
    {
      "epoch": 0.08447,
      "grad_norm": 0.9969368300751122,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 8447
    },
    {
      "epoch": 0.08448,
      "grad_norm": 1.040230047556125,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 8448
    },
    {
      "epoch": 0.08449,
      "grad_norm": 0.9375823755529817,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 8449
    },
    {
      "epoch": 0.0845,
      "grad_norm": 0.8771502461318594,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 8450
    },
    {
      "epoch": 0.08451,
      "grad_norm": 0.7739620306437573,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 8451
    },
    {
      "epoch": 0.08452,
      "grad_norm": 0.673756075602562,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 8452
    },
    {
      "epoch": 0.08453,
      "grad_norm": 0.6543280916386777,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 8453
    },
    {
      "epoch": 0.08454,
      "grad_norm": 0.6889773663662369,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 8454
    },
    {
      "epoch": 0.08455,
      "grad_norm": 0.7838053719288101,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 8455
    },
    {
      "epoch": 0.08456,
      "grad_norm": 0.8119139733359273,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 8456
    },
    {
      "epoch": 0.08457,
      "grad_norm": 0.6837577523016988,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 8457
    },
    {
      "epoch": 0.08458,
      "grad_norm": 0.6923132881641682,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8458
    },
    {
      "epoch": 0.08459,
      "grad_norm": 0.7691218378497301,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 8459
    },
    {
      "epoch": 0.0846,
      "grad_norm": 0.7521614723982595,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 8460
    },
    {
      "epoch": 0.08461,
      "grad_norm": 0.7593018376927517,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 8461
    },
    {
      "epoch": 0.08462,
      "grad_norm": 0.8660621840990166,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 8462
    },
    {
      "epoch": 0.08463,
      "grad_norm": 0.832579407041644,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 8463
    },
    {
      "epoch": 0.08464,
      "grad_norm": 0.8496293669573718,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 8464
    },
    {
      "epoch": 0.08465,
      "grad_norm": 0.9447869224057244,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 8465
    },
    {
      "epoch": 0.08466,
      "grad_norm": 0.9693005746173255,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 8466
    },
    {
      "epoch": 0.08467,
      "grad_norm": 0.6866322634836225,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 8467
    },
    {
      "epoch": 0.08468,
      "grad_norm": 0.7903651480232442,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 8468
    },
    {
      "epoch": 0.08469,
      "grad_norm": 0.7645618738122506,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 8469
    },
    {
      "epoch": 0.0847,
      "grad_norm": 0.7413695835335838,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 8470
    },
    {
      "epoch": 0.08471,
      "grad_norm": 0.7231784570238742,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 8471
    },
    {
      "epoch": 0.08472,
      "grad_norm": 0.931451402440091,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 8472
    },
    {
      "epoch": 0.08473,
      "grad_norm": 1.1881300365948153,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 8473
    },
    {
      "epoch": 0.08474,
      "grad_norm": 0.7884945285700253,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 8474
    },
    {
      "epoch": 0.08475,
      "grad_norm": 0.6466505880641876,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 8475
    },
    {
      "epoch": 0.08476,
      "grad_norm": 0.6843402752034309,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 8476
    },
    {
      "epoch": 0.08477,
      "grad_norm": 0.6820809860569809,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 8477
    },
    {
      "epoch": 0.08478,
      "grad_norm": 0.6640365539259963,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 8478
    },
    {
      "epoch": 0.08479,
      "grad_norm": 0.6014643687487943,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 8479
    },
    {
      "epoch": 0.0848,
      "grad_norm": 0.6525358299661146,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 8480
    },
    {
      "epoch": 0.08481,
      "grad_norm": 0.706918999582302,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 8481
    },
    {
      "epoch": 0.08482,
      "grad_norm": 0.7468135222494826,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 8482
    },
    {
      "epoch": 0.08483,
      "grad_norm": 0.7732979758093033,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 8483
    },
    {
      "epoch": 0.08484,
      "grad_norm": 0.7492155564600703,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 8484
    },
    {
      "epoch": 0.08485,
      "grad_norm": 0.6786121897383414,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 8485
    },
    {
      "epoch": 0.08486,
      "grad_norm": 0.6273731355412779,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 8486
    },
    {
      "epoch": 0.08487,
      "grad_norm": 0.6720801463551171,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 8487
    },
    {
      "epoch": 0.08488,
      "grad_norm": 0.6335377972042376,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 8488
    },
    {
      "epoch": 0.08489,
      "grad_norm": 0.6719260396651473,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 8489
    },
    {
      "epoch": 0.0849,
      "grad_norm": 0.7312131012473647,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 8490
    },
    {
      "epoch": 0.08491,
      "grad_norm": 0.8249047120148688,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 8491
    },
    {
      "epoch": 0.08492,
      "grad_norm": 1.0407015940181197,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 8492
    },
    {
      "epoch": 0.08493,
      "grad_norm": 1.0867956366330738,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 8493
    },
    {
      "epoch": 0.08494,
      "grad_norm": 0.915620699057293,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 8494
    },
    {
      "epoch": 0.08495,
      "grad_norm": 0.89156893468348,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 8495
    },
    {
      "epoch": 0.08496,
      "grad_norm": 0.8517601087707647,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 8496
    },
    {
      "epoch": 0.08497,
      "grad_norm": 0.7993759869320393,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 8497
    },
    {
      "epoch": 0.08498,
      "grad_norm": 0.7017883148327033,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 8498
    },
    {
      "epoch": 0.08499,
      "grad_norm": 0.6236773075397719,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 8499
    },
    {
      "epoch": 0.085,
      "grad_norm": 0.6286035320618882,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 8500
    },
    {
      "epoch": 0.08501,
      "grad_norm": 0.5212820111963481,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 8501
    },
    {
      "epoch": 0.08502,
      "grad_norm": 0.5487558855281737,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 8502
    },
    {
      "epoch": 0.08503,
      "grad_norm": 0.5707569684852064,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 8503
    },
    {
      "epoch": 0.08504,
      "grad_norm": 0.6687161931414275,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 8504
    },
    {
      "epoch": 0.08505,
      "grad_norm": 0.6723026556005183,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 8505
    },
    {
      "epoch": 0.08506,
      "grad_norm": 0.6565086694246933,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 8506
    },
    {
      "epoch": 0.08507,
      "grad_norm": 0.6310941478662354,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 8507
    },
    {
      "epoch": 0.08508,
      "grad_norm": 0.5824407794995421,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 8508
    },
    {
      "epoch": 0.08509,
      "grad_norm": 0.570213160176025,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 8509
    },
    {
      "epoch": 0.0851,
      "grad_norm": 0.545653653993305,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 8510
    },
    {
      "epoch": 0.08511,
      "grad_norm": 0.6166091785522915,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 8511
    },
    {
      "epoch": 0.08512,
      "grad_norm": 0.7266811939494829,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 8512
    },
    {
      "epoch": 0.08513,
      "grad_norm": 1.0119925425784506,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 8513
    },
    {
      "epoch": 0.08514,
      "grad_norm": 1.0618994395885715,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 8514
    },
    {
      "epoch": 0.08515,
      "grad_norm": 0.8410180154663229,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 8515
    },
    {
      "epoch": 0.08516,
      "grad_norm": 0.7647905005211172,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 8516
    },
    {
      "epoch": 0.08517,
      "grad_norm": 0.7839452502836995,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8517
    },
    {
      "epoch": 0.08518,
      "grad_norm": 0.8837758490786244,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 8518
    },
    {
      "epoch": 0.08519,
      "grad_norm": 1.0026623503391587,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 8519
    },
    {
      "epoch": 0.0852,
      "grad_norm": 0.8804069304034173,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 8520
    },
    {
      "epoch": 0.08521,
      "grad_norm": 0.7391980251114973,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 8521
    },
    {
      "epoch": 0.08522,
      "grad_norm": 0.6582209926239634,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 8522
    },
    {
      "epoch": 0.08523,
      "grad_norm": 0.686774998714554,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 8523
    },
    {
      "epoch": 0.08524,
      "grad_norm": 0.6276671588615992,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 8524
    },
    {
      "epoch": 0.08525,
      "grad_norm": 0.6285944259527609,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 8525
    },
    {
      "epoch": 0.08526,
      "grad_norm": 0.6730183358064831,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 8526
    },
    {
      "epoch": 0.08527,
      "grad_norm": 0.7562963445445285,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 8527
    },
    {
      "epoch": 0.08528,
      "grad_norm": 0.7728429342633116,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 8528
    },
    {
      "epoch": 0.08529,
      "grad_norm": 0.765937905688893,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 8529
    },
    {
      "epoch": 0.0853,
      "grad_norm": 0.6845167962734888,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 8530
    },
    {
      "epoch": 0.08531,
      "grad_norm": 0.5726987582988281,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 8531
    },
    {
      "epoch": 0.08532,
      "grad_norm": 0.5597095167661914,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 8532
    },
    {
      "epoch": 0.08533,
      "grad_norm": 0.5041472063185792,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 8533
    },
    {
      "epoch": 0.08534,
      "grad_norm": 0.45851677887407977,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 8534
    },
    {
      "epoch": 0.08535,
      "grad_norm": 0.42358429319858426,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 8535
    },
    {
      "epoch": 0.08536,
      "grad_norm": 0.424584021481874,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 8536
    },
    {
      "epoch": 0.08537,
      "grad_norm": 0.5509844324435204,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 8537
    },
    {
      "epoch": 0.08538,
      "grad_norm": 0.7117815534754696,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 8538
    },
    {
      "epoch": 0.08539,
      "grad_norm": 0.8518554101874906,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 8539
    },
    {
      "epoch": 0.0854,
      "grad_norm": 0.9392658629366815,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 8540
    },
    {
      "epoch": 0.08541,
      "grad_norm": 0.9325601448881679,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 8541
    },
    {
      "epoch": 0.08542,
      "grad_norm": 0.8887155749787176,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 8542
    },
    {
      "epoch": 0.08543,
      "grad_norm": 0.9460176301826329,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 8543
    },
    {
      "epoch": 0.08544,
      "grad_norm": 0.8622072400696076,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 8544
    },
    {
      "epoch": 0.08545,
      "grad_norm": 0.8420027078116906,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8545
    },
    {
      "epoch": 0.08546,
      "grad_norm": 0.7341692131081741,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 8546
    },
    {
      "epoch": 0.08547,
      "grad_norm": 0.7543016815952612,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 8547
    },
    {
      "epoch": 0.08548,
      "grad_norm": 0.7607341436346121,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 8548
    },
    {
      "epoch": 0.08549,
      "grad_norm": 0.7660211842461696,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 8549
    },
    {
      "epoch": 0.0855,
      "grad_norm": 0.8328996624474356,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 8550
    },
    {
      "epoch": 0.08551,
      "grad_norm": 0.9887027893940781,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 8551
    },
    {
      "epoch": 0.08552,
      "grad_norm": 0.9138699679462117,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 8552
    },
    {
      "epoch": 0.08553,
      "grad_norm": 0.88796138098534,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 8553
    },
    {
      "epoch": 0.08554,
      "grad_norm": 0.7864887844624548,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 8554
    },
    {
      "epoch": 0.08555,
      "grad_norm": 0.8150705947614163,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 8555
    },
    {
      "epoch": 0.08556,
      "grad_norm": 0.8837241002629734,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 8556
    },
    {
      "epoch": 0.08557,
      "grad_norm": 0.9289452003464579,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 8557
    },
    {
      "epoch": 0.08558,
      "grad_norm": 0.9018548936220813,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 8558
    },
    {
      "epoch": 0.08559,
      "grad_norm": 0.8317138215654469,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 8559
    },
    {
      "epoch": 0.0856,
      "grad_norm": 0.7222621649910216,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 8560
    },
    {
      "epoch": 0.08561,
      "grad_norm": 0.6904128334298788,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 8561
    },
    {
      "epoch": 0.08562,
      "grad_norm": 0.6940485635066651,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 8562
    },
    {
      "epoch": 0.08563,
      "grad_norm": 0.7325452827552434,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 8563
    },
    {
      "epoch": 0.08564,
      "grad_norm": 0.8123368725671672,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 8564
    },
    {
      "epoch": 0.08565,
      "grad_norm": 0.9408503931561969,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 8565
    },
    {
      "epoch": 0.08566,
      "grad_norm": 1.0616374748054886,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 8566
    },
    {
      "epoch": 0.08567,
      "grad_norm": 1.047024424521824,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 8567
    },
    {
      "epoch": 0.08568,
      "grad_norm": 0.8468698345129514,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 8568
    },
    {
      "epoch": 0.08569,
      "grad_norm": 0.7944472675814765,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 8569
    },
    {
      "epoch": 0.0857,
      "grad_norm": 0.8045393385309881,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 8570
    },
    {
      "epoch": 0.08571,
      "grad_norm": 0.9098867331516741,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 8571
    },
    {
      "epoch": 0.08572,
      "grad_norm": 0.9195544563171938,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 8572
    },
    {
      "epoch": 0.08573,
      "grad_norm": 0.8423884949796944,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 8573
    },
    {
      "epoch": 0.08574,
      "grad_norm": 0.8214333018685611,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 8574
    },
    {
      "epoch": 0.08575,
      "grad_norm": 0.6350812617594055,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 8575
    },
    {
      "epoch": 0.08576,
      "grad_norm": 0.5568431689715675,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 8576
    },
    {
      "epoch": 0.08577,
      "grad_norm": 0.5632148561347211,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8577
    },
    {
      "epoch": 0.08578,
      "grad_norm": 0.5940159555341951,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 8578
    },
    {
      "epoch": 0.08579,
      "grad_norm": 0.5819842281049936,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 8579
    },
    {
      "epoch": 0.0858,
      "grad_norm": 0.6099409069272952,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 8580
    },
    {
      "epoch": 0.08581,
      "grad_norm": 0.6188694008443366,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 8581
    },
    {
      "epoch": 0.08582,
      "grad_norm": 0.59379978193371,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 8582
    },
    {
      "epoch": 0.08583,
      "grad_norm": 0.5744750135944172,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 8583
    },
    {
      "epoch": 0.08584,
      "grad_norm": 0.6722444371986878,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 8584
    },
    {
      "epoch": 0.08585,
      "grad_norm": 0.7411902899854854,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 8585
    },
    {
      "epoch": 0.08586,
      "grad_norm": 0.7792170934870236,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 8586
    },
    {
      "epoch": 0.08587,
      "grad_norm": 0.6853345642159595,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 8587
    },
    {
      "epoch": 0.08588,
      "grad_norm": 0.6170369949749352,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 8588
    },
    {
      "epoch": 0.08589,
      "grad_norm": 0.49708112672361066,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 8589
    },
    {
      "epoch": 0.0859,
      "grad_norm": 0.5152160677520728,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 8590
    },
    {
      "epoch": 0.08591,
      "grad_norm": 0.49741486517713757,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 8591
    },
    {
      "epoch": 0.08592,
      "grad_norm": 0.5910307086102645,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 8592
    },
    {
      "epoch": 0.08593,
      "grad_norm": 0.6649076276760316,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 8593
    },
    {
      "epoch": 0.08594,
      "grad_norm": 0.6902152865247433,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 8594
    },
    {
      "epoch": 0.08595,
      "grad_norm": 0.7200015025904549,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 8595
    },
    {
      "epoch": 0.08596,
      "grad_norm": 0.8369502085711913,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 8596
    },
    {
      "epoch": 0.08597,
      "grad_norm": 0.9081695796712485,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 8597
    },
    {
      "epoch": 0.08598,
      "grad_norm": 0.8574484765894195,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 8598
    },
    {
      "epoch": 0.08599,
      "grad_norm": 0.7141093112530459,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 8599
    },
    {
      "epoch": 0.086,
      "grad_norm": 0.7618404036518704,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 8600
    },
    {
      "epoch": 0.08601,
      "grad_norm": 0.7471246187217334,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 8601
    },
    {
      "epoch": 0.08602,
      "grad_norm": 0.7547952980026077,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 8602
    },
    {
      "epoch": 0.08603,
      "grad_norm": 0.7629507458741042,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 8603
    },
    {
      "epoch": 0.08604,
      "grad_norm": 0.8034444554972223,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 8604
    },
    {
      "epoch": 0.08605,
      "grad_norm": 0.9611109929996176,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 8605
    },
    {
      "epoch": 0.08606,
      "grad_norm": 1.1203243516217172,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 8606
    },
    {
      "epoch": 0.08607,
      "grad_norm": 0.7916021324508382,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 8607
    },
    {
      "epoch": 0.08608,
      "grad_norm": 0.5801180546262028,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 8608
    },
    {
      "epoch": 0.08609,
      "grad_norm": 0.5983815121428754,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 8609
    },
    {
      "epoch": 0.0861,
      "grad_norm": 0.6226924849818875,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 8610
    },
    {
      "epoch": 0.08611,
      "grad_norm": 0.5973475758608405,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 8611
    },
    {
      "epoch": 0.08612,
      "grad_norm": 0.6361887564027072,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 8612
    },
    {
      "epoch": 0.08613,
      "grad_norm": 0.6913314348371322,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 8613
    },
    {
      "epoch": 0.08614,
      "grad_norm": 0.6911565582790656,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 8614
    },
    {
      "epoch": 0.08615,
      "grad_norm": 0.6480201878105682,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 8615
    },
    {
      "epoch": 0.08616,
      "grad_norm": 0.621140757257243,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 8616
    },
    {
      "epoch": 0.08617,
      "grad_norm": 0.5814187395920917,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 8617
    },
    {
      "epoch": 0.08618,
      "grad_norm": 0.5949195662749583,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 8618
    },
    {
      "epoch": 0.08619,
      "grad_norm": 0.5474225046144992,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 8619
    },
    {
      "epoch": 0.0862,
      "grad_norm": 0.5884294701344107,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 8620
    },
    {
      "epoch": 0.08621,
      "grad_norm": 0.5414477831138819,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 8621
    },
    {
      "epoch": 0.08622,
      "grad_norm": 0.5853379182460213,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 8622
    },
    {
      "epoch": 0.08623,
      "grad_norm": 0.6748785821043584,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8623
    },
    {
      "epoch": 0.08624,
      "grad_norm": 0.8007913851325251,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 8624
    },
    {
      "epoch": 0.08625,
      "grad_norm": 0.9532221305960991,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 8625
    },
    {
      "epoch": 0.08626,
      "grad_norm": 0.9383143976468733,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 8626
    },
    {
      "epoch": 0.08627,
      "grad_norm": 0.8045743378009347,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8627
    },
    {
      "epoch": 0.08628,
      "grad_norm": 0.7402532485299439,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 8628
    },
    {
      "epoch": 0.08629,
      "grad_norm": 0.7187195965365745,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 8629
    },
    {
      "epoch": 0.0863,
      "grad_norm": 0.7749136270046196,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 8630
    },
    {
      "epoch": 0.08631,
      "grad_norm": 0.7803939926856307,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8631
    },
    {
      "epoch": 0.08632,
      "grad_norm": 0.812820737542032,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 8632
    },
    {
      "epoch": 0.08633,
      "grad_norm": 0.9418729107920155,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 8633
    },
    {
      "epoch": 0.08634,
      "grad_norm": 1.0595248522005765,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 8634
    },
    {
      "epoch": 0.08635,
      "grad_norm": 0.9627374328257062,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 8635
    },
    {
      "epoch": 0.08636,
      "grad_norm": 0.9731387011493425,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 8636
    },
    {
      "epoch": 0.08637,
      "grad_norm": 1.0615767557197648,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 8637
    },
    {
      "epoch": 0.08638,
      "grad_norm": 0.8969163900961645,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 8638
    },
    {
      "epoch": 0.08639,
      "grad_norm": 0.7836065427193415,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 8639
    },
    {
      "epoch": 0.0864,
      "grad_norm": 0.7397344262293751,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 8640
    },
    {
      "epoch": 0.08641,
      "grad_norm": 0.7324947749888412,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 8641
    },
    {
      "epoch": 0.08642,
      "grad_norm": 0.7288716146484082,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 8642
    },
    {
      "epoch": 0.08643,
      "grad_norm": 0.7339841907154421,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 8643
    },
    {
      "epoch": 0.08644,
      "grad_norm": 0.8107938826013737,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 8644
    },
    {
      "epoch": 0.08645,
      "grad_norm": 0.9556271679368206,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 8645
    },
    {
      "epoch": 0.08646,
      "grad_norm": 0.9674772661574546,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 8646
    },
    {
      "epoch": 0.08647,
      "grad_norm": 0.9453669207011212,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 8647
    },
    {
      "epoch": 0.08648,
      "grad_norm": 0.7876425030935472,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 8648
    },
    {
      "epoch": 0.08649,
      "grad_norm": 0.6724723608937945,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 8649
    },
    {
      "epoch": 0.0865,
      "grad_norm": 0.7454146637283006,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8650
    },
    {
      "epoch": 0.08651,
      "grad_norm": 0.8517115223988251,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 8651
    },
    {
      "epoch": 0.08652,
      "grad_norm": 0.8860899877841061,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 8652
    },
    {
      "epoch": 0.08653,
      "grad_norm": 0.6986085802414974,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 8653
    },
    {
      "epoch": 0.08654,
      "grad_norm": 0.6383445520115142,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 8654
    },
    {
      "epoch": 0.08655,
      "grad_norm": 0.6126837697223702,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 8655
    },
    {
      "epoch": 0.08656,
      "grad_norm": 0.6303132972548725,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 8656
    },
    {
      "epoch": 0.08657,
      "grad_norm": 0.7051801520736573,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 8657
    },
    {
      "epoch": 0.08658,
      "grad_norm": 0.7534311537900272,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 8658
    },
    {
      "epoch": 0.08659,
      "grad_norm": 0.6915534733946069,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 8659
    },
    {
      "epoch": 0.0866,
      "grad_norm": 0.6192069664136859,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 8660
    },
    {
      "epoch": 0.08661,
      "grad_norm": 0.5572612681652647,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 8661
    },
    {
      "epoch": 0.08662,
      "grad_norm": 0.518904918411408,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 8662
    },
    {
      "epoch": 0.08663,
      "grad_norm": 0.45643753659454445,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 8663
    },
    {
      "epoch": 0.08664,
      "grad_norm": 0.40977161141131885,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8664
    },
    {
      "epoch": 0.08665,
      "grad_norm": 0.3878668107895774,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 8665
    },
    {
      "epoch": 0.08666,
      "grad_norm": 0.413430960164737,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 8666
    },
    {
      "epoch": 0.08667,
      "grad_norm": 0.49272907946966926,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 8667
    },
    {
      "epoch": 0.08668,
      "grad_norm": 0.5766422713262668,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 8668
    },
    {
      "epoch": 0.08669,
      "grad_norm": 0.7090374371716733,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 8669
    },
    {
      "epoch": 0.0867,
      "grad_norm": 0.8903287925695574,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 8670
    },
    {
      "epoch": 0.08671,
      "grad_norm": 1.0465570588393474,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 8671
    },
    {
      "epoch": 0.08672,
      "grad_norm": 0.9634733776041171,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 8672
    },
    {
      "epoch": 0.08673,
      "grad_norm": 0.9660580246254519,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8673
    },
    {
      "epoch": 0.08674,
      "grad_norm": 0.904283629793719,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 8674
    },
    {
      "epoch": 0.08675,
      "grad_norm": 0.8438104142985071,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 8675
    },
    {
      "epoch": 0.08676,
      "grad_norm": 0.7857620375537205,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 8676
    },
    {
      "epoch": 0.08677,
      "grad_norm": 0.7969369610845204,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 8677
    },
    {
      "epoch": 0.08678,
      "grad_norm": 0.7396002144781133,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 8678
    },
    {
      "epoch": 0.08679,
      "grad_norm": 0.6673116262175337,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 8679
    },
    {
      "epoch": 0.0868,
      "grad_norm": 0.7555945744269758,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 8680
    },
    {
      "epoch": 0.08681,
      "grad_norm": 0.8839352666831195,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 8681
    },
    {
      "epoch": 0.08682,
      "grad_norm": 0.9051922142113155,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 8682
    },
    {
      "epoch": 0.08683,
      "grad_norm": 0.846027323488606,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 8683
    },
    {
      "epoch": 0.08684,
      "grad_norm": 0.7192665361560291,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 8684
    },
    {
      "epoch": 0.08685,
      "grad_norm": 0.643632971790141,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 8685
    },
    {
      "epoch": 0.08686,
      "grad_norm": 0.6466143794382575,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 8686
    },
    {
      "epoch": 0.08687,
      "grad_norm": 0.7039176614373789,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8687
    },
    {
      "epoch": 0.08688,
      "grad_norm": 0.7899249787925823,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 8688
    },
    {
      "epoch": 0.08689,
      "grad_norm": 0.8216415805263771,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 8689
    },
    {
      "epoch": 0.0869,
      "grad_norm": 0.7040479359385806,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 8690
    },
    {
      "epoch": 0.08691,
      "grad_norm": 0.6418926125651642,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 8691
    },
    {
      "epoch": 0.08692,
      "grad_norm": 0.7077297817088414,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 8692
    },
    {
      "epoch": 0.08693,
      "grad_norm": 0.7606697220731845,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 8693
    },
    {
      "epoch": 0.08694,
      "grad_norm": 0.7152450767122566,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 8694
    },
    {
      "epoch": 0.08695,
      "grad_norm": 0.6231380065971586,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 8695
    },
    {
      "epoch": 0.08696,
      "grad_norm": 0.5828288246085719,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 8696
    },
    {
      "epoch": 0.08697,
      "grad_norm": 0.5953032303327713,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 8697
    },
    {
      "epoch": 0.08698,
      "grad_norm": 0.6083787084008646,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 8698
    },
    {
      "epoch": 0.08699,
      "grad_norm": 0.7378994678558721,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 8699
    },
    {
      "epoch": 0.087,
      "grad_norm": 0.9161671171702003,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 8700
    },
    {
      "epoch": 0.08701,
      "grad_norm": 1.1377552843109822,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 8701
    },
    {
      "epoch": 0.08702,
      "grad_norm": 0.8079302042495463,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 8702
    },
    {
      "epoch": 0.08703,
      "grad_norm": 0.608371143181045,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 8703
    },
    {
      "epoch": 0.08704,
      "grad_norm": 0.6281858127290479,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 8704
    },
    {
      "epoch": 0.08705,
      "grad_norm": 0.6793561925183917,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 8705
    },
    {
      "epoch": 0.08706,
      "grad_norm": 0.7106800046852197,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 8706
    },
    {
      "epoch": 0.08707,
      "grad_norm": 0.7141650332904569,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 8707
    },
    {
      "epoch": 0.08708,
      "grad_norm": 0.8271404025231711,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 8708
    },
    {
      "epoch": 0.08709,
      "grad_norm": 1.0474926164726928,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 8709
    },
    {
      "epoch": 0.0871,
      "grad_norm": 0.9907087620885723,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 8710
    },
    {
      "epoch": 0.08711,
      "grad_norm": 1.0656047756713767,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 8711
    },
    {
      "epoch": 0.08712,
      "grad_norm": 0.881294661258653,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 8712
    },
    {
      "epoch": 0.08713,
      "grad_norm": 0.9263143666012618,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 8713
    },
    {
      "epoch": 0.08714,
      "grad_norm": 0.9652223357853811,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 8714
    },
    {
      "epoch": 0.08715,
      "grad_norm": 1.1556407233323303,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 8715
    },
    {
      "epoch": 0.08716,
      "grad_norm": 0.8600125406687056,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 8716
    },
    {
      "epoch": 0.08717,
      "grad_norm": 0.7952102732113174,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 8717
    },
    {
      "epoch": 0.08718,
      "grad_norm": 0.709911569034494,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 8718
    },
    {
      "epoch": 0.08719,
      "grad_norm": 0.6127044794162302,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 8719
    },
    {
      "epoch": 0.0872,
      "grad_norm": 0.5418597465972178,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 8720
    },
    {
      "epoch": 0.08721,
      "grad_norm": 0.545384052999374,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 8721
    },
    {
      "epoch": 0.08722,
      "grad_norm": 0.5854972859534596,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 8722
    },
    {
      "epoch": 0.08723,
      "grad_norm": 0.599752513559582,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 8723
    },
    {
      "epoch": 0.08724,
      "grad_norm": 0.5979573041484251,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 8724
    },
    {
      "epoch": 0.08725,
      "grad_norm": 0.6180900508607188,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 8725
    },
    {
      "epoch": 0.08726,
      "grad_norm": 0.6336982537847569,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 8726
    },
    {
      "epoch": 0.08727,
      "grad_norm": 0.7620135479461384,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 8727
    },
    {
      "epoch": 0.08728,
      "grad_norm": 0.8150934564907844,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 8728
    },
    {
      "epoch": 0.08729,
      "grad_norm": 0.6748250974417187,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 8729
    },
    {
      "epoch": 0.0873,
      "grad_norm": 0.5106549285627258,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 8730
    },
    {
      "epoch": 0.08731,
      "grad_norm": 0.616467259066863,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 8731
    },
    {
      "epoch": 0.08732,
      "grad_norm": 0.6409301951961746,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 8732
    },
    {
      "epoch": 0.08733,
      "grad_norm": 0.6531284409173848,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 8733
    },
    {
      "epoch": 0.08734,
      "grad_norm": 0.6909519673379502,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 8734
    },
    {
      "epoch": 0.08735,
      "grad_norm": 0.6949538564230802,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 8735
    },
    {
      "epoch": 0.08736,
      "grad_norm": 0.6353533341447787,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 8736
    },
    {
      "epoch": 0.08737,
      "grad_norm": 0.5766041576549661,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 8737
    },
    {
      "epoch": 0.08738,
      "grad_norm": 0.5636229267831031,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 8738
    },
    {
      "epoch": 0.08739,
      "grad_norm": 0.5714257910724869,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 8739
    },
    {
      "epoch": 0.0874,
      "grad_norm": 0.6664233704269024,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 8740
    },
    {
      "epoch": 0.08741,
      "grad_norm": 0.8499302285997681,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 8741
    },
    {
      "epoch": 0.08742,
      "grad_norm": 1.1223141440691569,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 8742
    },
    {
      "epoch": 0.08743,
      "grad_norm": 0.9050529775378228,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 8743
    },
    {
      "epoch": 0.08744,
      "grad_norm": 0.7951156037626225,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 8744
    },
    {
      "epoch": 0.08745,
      "grad_norm": 0.9924672009502984,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 8745
    },
    {
      "epoch": 0.08746,
      "grad_norm": 0.971675314061291,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 8746
    },
    {
      "epoch": 0.08747,
      "grad_norm": 0.9032139097191704,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 8747
    },
    {
      "epoch": 0.08748,
      "grad_norm": 0.798921343364292,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 8748
    },
    {
      "epoch": 0.08749,
      "grad_norm": 0.7958369681342767,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 8749
    },
    {
      "epoch": 0.0875,
      "grad_norm": 0.7880652528528818,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 8750
    },
    {
      "epoch": 0.08751,
      "grad_norm": 0.7117714704920254,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 8751
    },
    {
      "epoch": 0.08752,
      "grad_norm": 0.6564224414293092,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 8752
    },
    {
      "epoch": 0.08753,
      "grad_norm": 0.6251613210169059,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 8753
    },
    {
      "epoch": 0.08754,
      "grad_norm": 0.6545076658857255,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 8754
    },
    {
      "epoch": 0.08755,
      "grad_norm": 0.6299656295806964,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 8755
    },
    {
      "epoch": 0.08756,
      "grad_norm": 0.6937544705015782,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 8756
    },
    {
      "epoch": 0.08757,
      "grad_norm": 0.7613436036720337,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 8757
    },
    {
      "epoch": 0.08758,
      "grad_norm": 0.7935810078864296,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 8758
    },
    {
      "epoch": 0.08759,
      "grad_norm": 0.9218245052230097,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 8759
    },
    {
      "epoch": 0.0876,
      "grad_norm": 1.1133184341136968,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8760
    },
    {
      "epoch": 0.08761,
      "grad_norm": 0.8471729174022539,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 8761
    },
    {
      "epoch": 0.08762,
      "grad_norm": 0.6885885491623273,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 8762
    },
    {
      "epoch": 0.08763,
      "grad_norm": 0.6900728124677,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 8763
    },
    {
      "epoch": 0.08764,
      "grad_norm": 0.7448709364689655,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 8764
    },
    {
      "epoch": 0.08765,
      "grad_norm": 0.7003442871930288,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 8765
    },
    {
      "epoch": 0.08766,
      "grad_norm": 0.695189743891509,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 8766
    },
    {
      "epoch": 0.08767,
      "grad_norm": 0.6561447154517275,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 8767
    },
    {
      "epoch": 0.08768,
      "grad_norm": 0.6134378072581433,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 8768
    },
    {
      "epoch": 0.08769,
      "grad_norm": 0.6152250798453555,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 8769
    },
    {
      "epoch": 0.0877,
      "grad_norm": 0.6450512433536177,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 8770
    },
    {
      "epoch": 0.08771,
      "grad_norm": 0.7339302655862884,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 8771
    },
    {
      "epoch": 0.08772,
      "grad_norm": 0.8938660355068361,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 8772
    },
    {
      "epoch": 0.08773,
      "grad_norm": 0.918539366218562,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 8773
    },
    {
      "epoch": 0.08774,
      "grad_norm": 0.8161914119839463,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 8774
    },
    {
      "epoch": 0.08775,
      "grad_norm": 0.7485358173137795,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 8775
    },
    {
      "epoch": 0.08776,
      "grad_norm": 0.6501840423824952,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8776
    },
    {
      "epoch": 0.08777,
      "grad_norm": 0.6227297966691554,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8777
    },
    {
      "epoch": 0.08778,
      "grad_norm": 0.6147681929920195,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 8778
    },
    {
      "epoch": 0.08779,
      "grad_norm": 0.6711211259228609,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 8779
    },
    {
      "epoch": 0.0878,
      "grad_norm": 0.6873812876268875,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 8780
    },
    {
      "epoch": 0.08781,
      "grad_norm": 0.7894739230084857,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 8781
    },
    {
      "epoch": 0.08782,
      "grad_norm": 0.9319206036015761,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 8782
    },
    {
      "epoch": 0.08783,
      "grad_norm": 0.785166946779399,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 8783
    },
    {
      "epoch": 0.08784,
      "grad_norm": 0.6883174398483335,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 8784
    },
    {
      "epoch": 0.08785,
      "grad_norm": 0.7618236271622089,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 8785
    },
    {
      "epoch": 0.08786,
      "grad_norm": 0.7882058303319279,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 8786
    },
    {
      "epoch": 0.08787,
      "grad_norm": 0.8009435303009658,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 8787
    },
    {
      "epoch": 0.08788,
      "grad_norm": 0.8043157032401507,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 8788
    },
    {
      "epoch": 0.08789,
      "grad_norm": 0.8053019098728231,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 8789
    },
    {
      "epoch": 0.0879,
      "grad_norm": 0.7507680332954385,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 8790
    },
    {
      "epoch": 0.08791,
      "grad_norm": 0.742881908635139,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8791
    },
    {
      "epoch": 0.08792,
      "grad_norm": 0.6619963633805601,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 8792
    },
    {
      "epoch": 0.08793,
      "grad_norm": 0.6836794010275485,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 8793
    },
    {
      "epoch": 0.08794,
      "grad_norm": 0.8299110382089389,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 8794
    },
    {
      "epoch": 0.08795,
      "grad_norm": 1.0468573387811748,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8795
    },
    {
      "epoch": 0.08796,
      "grad_norm": 1.0109579043602372,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 8796
    },
    {
      "epoch": 0.08797,
      "grad_norm": 0.9354583933034025,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 8797
    },
    {
      "epoch": 0.08798,
      "grad_norm": 0.8801038141490901,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 8798
    },
    {
      "epoch": 0.08799,
      "grad_norm": 0.8373596391595058,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 8799
    },
    {
      "epoch": 0.088,
      "grad_norm": 0.8236058297216525,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 8800
    },
    {
      "epoch": 0.08801,
      "grad_norm": 0.9171639160570122,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 8801
    },
    {
      "epoch": 0.08802,
      "grad_norm": 0.950129352800799,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 8802
    },
    {
      "epoch": 0.08803,
      "grad_norm": 0.97417915794881,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 8803
    },
    {
      "epoch": 0.08804,
      "grad_norm": 0.9961244608205473,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 8804
    },
    {
      "epoch": 0.08805,
      "grad_norm": 0.8887568015028089,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 8805
    },
    {
      "epoch": 0.08806,
      "grad_norm": 0.8507811413672998,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 8806
    },
    {
      "epoch": 0.08807,
      "grad_norm": 0.7799651810021191,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 8807
    },
    {
      "epoch": 0.08808,
      "grad_norm": 0.6310859184139921,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 8808
    },
    {
      "epoch": 0.08809,
      "grad_norm": 0.5415343793054738,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 8809
    },
    {
      "epoch": 0.0881,
      "grad_norm": 0.5417480126046187,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 8810
    },
    {
      "epoch": 0.08811,
      "grad_norm": 0.5437548342900228,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 8811
    },
    {
      "epoch": 0.08812,
      "grad_norm": 0.5065008549086789,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 8812
    },
    {
      "epoch": 0.08813,
      "grad_norm": 0.44906733576252533,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 8813
    },
    {
      "epoch": 0.08814,
      "grad_norm": 0.4572901434044213,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 8814
    },
    {
      "epoch": 0.08815,
      "grad_norm": 0.39740682905597274,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 8815
    },
    {
      "epoch": 0.08816,
      "grad_norm": 0.41394278364223486,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 8816
    },
    {
      "epoch": 0.08817,
      "grad_norm": 0.40780800547558976,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 8817
    },
    {
      "epoch": 0.08818,
      "grad_norm": 0.45313876957887234,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 8818
    },
    {
      "epoch": 0.08819,
      "grad_norm": 0.47289697821709104,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 8819
    },
    {
      "epoch": 0.0882,
      "grad_norm": 0.5347028769974915,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 8820
    },
    {
      "epoch": 0.08821,
      "grad_norm": 0.6073667369132447,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 8821
    },
    {
      "epoch": 0.08822,
      "grad_norm": 0.631563807802477,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 8822
    },
    {
      "epoch": 0.08823,
      "grad_norm": 0.8219598063684808,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 8823
    },
    {
      "epoch": 0.08824,
      "grad_norm": 1.1113151394557685,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 8824
    },
    {
      "epoch": 0.08825,
      "grad_norm": 0.9855650333902289,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 8825
    },
    {
      "epoch": 0.08826,
      "grad_norm": 0.9778239420914661,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 8826
    },
    {
      "epoch": 0.08827,
      "grad_norm": 0.8179910297155395,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 8827
    },
    {
      "epoch": 0.08828,
      "grad_norm": 0.7966604826205193,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 8828
    },
    {
      "epoch": 0.08829,
      "grad_norm": 0.7977026742349369,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 8829
    },
    {
      "epoch": 0.0883,
      "grad_norm": 0.792129459764214,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 8830
    },
    {
      "epoch": 0.08831,
      "grad_norm": 0.9326888184375913,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 8831
    },
    {
      "epoch": 0.08832,
      "grad_norm": 0.866391958750175,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 8832
    },
    {
      "epoch": 0.08833,
      "grad_norm": 0.8717561470437578,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 8833
    },
    {
      "epoch": 0.08834,
      "grad_norm": 0.9086463815238154,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 8834
    },
    {
      "epoch": 0.08835,
      "grad_norm": 0.9982211117034512,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 8835
    },
    {
      "epoch": 0.08836,
      "grad_norm": 1.0116778830665036,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 8836
    },
    {
      "epoch": 0.08837,
      "grad_norm": 0.9047723332769526,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 8837
    },
    {
      "epoch": 0.08838,
      "grad_norm": 0.7218841347767363,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 8838
    },
    {
      "epoch": 0.08839,
      "grad_norm": 0.6219028863083907,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8839
    },
    {
      "epoch": 0.0884,
      "grad_norm": 0.6370526447568182,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 8840
    },
    {
      "epoch": 0.08841,
      "grad_norm": 0.6865207786249934,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 8841
    },
    {
      "epoch": 0.08842,
      "grad_norm": 0.8079884245973484,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 8842
    },
    {
      "epoch": 0.08843,
      "grad_norm": 0.7716447791785704,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 8843
    },
    {
      "epoch": 0.08844,
      "grad_norm": 0.6552595716080165,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 8844
    },
    {
      "epoch": 0.08845,
      "grad_norm": 0.6411659090784259,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 8845
    },
    {
      "epoch": 0.08846,
      "grad_norm": 0.6345516705136226,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 8846
    },
    {
      "epoch": 0.08847,
      "grad_norm": 0.6208423879645917,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 8847
    },
    {
      "epoch": 0.08848,
      "grad_norm": 0.6896343586488453,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 8848
    },
    {
      "epoch": 0.08849,
      "grad_norm": 0.6711212989454174,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 8849
    },
    {
      "epoch": 0.0885,
      "grad_norm": 0.787384336504839,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 8850
    },
    {
      "epoch": 0.08851,
      "grad_norm": 0.9296696469370899,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8851
    },
    {
      "epoch": 0.08852,
      "grad_norm": 0.9491351647008734,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 8852
    },
    {
      "epoch": 0.08853,
      "grad_norm": 0.8823989114621453,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 8853
    },
    {
      "epoch": 0.08854,
      "grad_norm": 0.7356047162096173,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 8854
    },
    {
      "epoch": 0.08855,
      "grad_norm": 0.6817590094654521,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 8855
    },
    {
      "epoch": 0.08856,
      "grad_norm": 0.6432697273240098,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 8856
    },
    {
      "epoch": 0.08857,
      "grad_norm": 0.5912854441908766,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 8857
    },
    {
      "epoch": 0.08858,
      "grad_norm": 0.6032547637065733,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 8858
    },
    {
      "epoch": 0.08859,
      "grad_norm": 0.5317724895750301,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 8859
    },
    {
      "epoch": 0.0886,
      "grad_norm": 0.490974624643615,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 8860
    },
    {
      "epoch": 0.08861,
      "grad_norm": 0.5134567395667045,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 8861
    },
    {
      "epoch": 0.08862,
      "grad_norm": 0.563666635376977,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 8862
    },
    {
      "epoch": 0.08863,
      "grad_norm": 0.7032312250587909,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 8863
    },
    {
      "epoch": 0.08864,
      "grad_norm": 0.8431923998213133,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 8864
    },
    {
      "epoch": 0.08865,
      "grad_norm": 0.9517906387901386,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 8865
    },
    {
      "epoch": 0.08866,
      "grad_norm": 0.894832461876128,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 8866
    },
    {
      "epoch": 0.08867,
      "grad_norm": 0.7886402628779215,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 8867
    },
    {
      "epoch": 0.08868,
      "grad_norm": 0.7427477147931492,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 8868
    },
    {
      "epoch": 0.08869,
      "grad_norm": 0.7801814362733093,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 8869
    },
    {
      "epoch": 0.0887,
      "grad_norm": 0.9378074963893244,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 8870
    },
    {
      "epoch": 0.08871,
      "grad_norm": 1.0785126885762188,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 8871
    },
    {
      "epoch": 0.08872,
      "grad_norm": 0.9686509189429044,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 8872
    },
    {
      "epoch": 0.08873,
      "grad_norm": 0.9359552632213535,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 8873
    },
    {
      "epoch": 0.08874,
      "grad_norm": 0.801952593241515,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 8874
    },
    {
      "epoch": 0.08875,
      "grad_norm": 0.7300737081360735,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 8875
    },
    {
      "epoch": 0.08876,
      "grad_norm": 0.6845784358472665,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 8876
    },
    {
      "epoch": 0.08877,
      "grad_norm": 0.5992699690439727,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 8877
    },
    {
      "epoch": 0.08878,
      "grad_norm": 0.6987685567227716,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 8878
    },
    {
      "epoch": 0.08879,
      "grad_norm": 0.755918105586785,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 8879
    },
    {
      "epoch": 0.0888,
      "grad_norm": 0.804896715950834,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 8880
    },
    {
      "epoch": 0.08881,
      "grad_norm": 0.8597307396006292,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 8881
    },
    {
      "epoch": 0.08882,
      "grad_norm": 0.890705623678484,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 8882
    },
    {
      "epoch": 0.08883,
      "grad_norm": 0.87665671220314,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 8883
    },
    {
      "epoch": 0.08884,
      "grad_norm": 0.9160504734206271,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 8884
    },
    {
      "epoch": 0.08885,
      "grad_norm": 0.9915450251876738,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 8885
    },
    {
      "epoch": 0.08886,
      "grad_norm": 0.9459675534923236,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 8886
    },
    {
      "epoch": 0.08887,
      "grad_norm": 0.8822638917509666,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 8887
    },
    {
      "epoch": 0.08888,
      "grad_norm": 0.9182248673162067,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 8888
    },
    {
      "epoch": 0.08889,
      "grad_norm": 0.8960795751997341,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 8889
    },
    {
      "epoch": 0.0889,
      "grad_norm": 0.7029348940189843,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 8890
    },
    {
      "epoch": 0.08891,
      "grad_norm": 0.5917270992086658,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 8891
    },
    {
      "epoch": 0.08892,
      "grad_norm": 0.5729686662161869,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 8892
    },
    {
      "epoch": 0.08893,
      "grad_norm": 0.4933602379272417,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 8893
    },
    {
      "epoch": 0.08894,
      "grad_norm": 0.4816956228568598,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 8894
    },
    {
      "epoch": 0.08895,
      "grad_norm": 0.5401089965190982,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 8895
    },
    {
      "epoch": 0.08896,
      "grad_norm": 0.6834674353769287,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 8896
    },
    {
      "epoch": 0.08897,
      "grad_norm": 0.9000006866364382,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 8897
    },
    {
      "epoch": 0.08898,
      "grad_norm": 1.0128344751677745,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 8898
    },
    {
      "epoch": 0.08899,
      "grad_norm": 0.8361225818664343,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8899
    },
    {
      "epoch": 0.089,
      "grad_norm": 0.7604422714843287,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 8900
    },
    {
      "epoch": 0.08901,
      "grad_norm": 0.7721795341276967,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 8901
    },
    {
      "epoch": 0.08902,
      "grad_norm": 0.900896957082942,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 8902
    },
    {
      "epoch": 0.08903,
      "grad_norm": 0.8920547801921579,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 8903
    },
    {
      "epoch": 0.08904,
      "grad_norm": 0.8420148885506412,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 8904
    },
    {
      "epoch": 0.08905,
      "grad_norm": 0.8570250538255739,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 8905
    },
    {
      "epoch": 0.08906,
      "grad_norm": 0.9205488642424341,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 8906
    },
    {
      "epoch": 0.08907,
      "grad_norm": 1.0405334047811572,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 8907
    },
    {
      "epoch": 0.08908,
      "grad_norm": 0.9468509956798719,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 8908
    },
    {
      "epoch": 0.08909,
      "grad_norm": 0.6926689852243239,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 8909
    },
    {
      "epoch": 0.0891,
      "grad_norm": 0.673009652964538,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 8910
    },
    {
      "epoch": 0.08911,
      "grad_norm": 0.6764278448657347,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 8911
    },
    {
      "epoch": 0.08912,
      "grad_norm": 0.713765564163065,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 8912
    },
    {
      "epoch": 0.08913,
      "grad_norm": 0.6427371366529472,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 8913
    },
    {
      "epoch": 0.08914,
      "grad_norm": 0.581053838640892,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8914
    },
    {
      "epoch": 0.08915,
      "grad_norm": 0.6154512813452391,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 8915
    },
    {
      "epoch": 0.08916,
      "grad_norm": 0.5811153917294403,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 8916
    },
    {
      "epoch": 0.08917,
      "grad_norm": 0.5826469378007894,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 8917
    },
    {
      "epoch": 0.08918,
      "grad_norm": 0.659868224790148,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 8918
    },
    {
      "epoch": 0.08919,
      "grad_norm": 0.7475809786764346,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 8919
    },
    {
      "epoch": 0.0892,
      "grad_norm": 0.8085506843447738,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 8920
    },
    {
      "epoch": 0.08921,
      "grad_norm": 0.8851417211555018,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 8921
    },
    {
      "epoch": 0.08922,
      "grad_norm": 0.8405802565396343,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 8922
    },
    {
      "epoch": 0.08923,
      "grad_norm": 0.7493224650962431,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 8923
    },
    {
      "epoch": 0.08924,
      "grad_norm": 0.6524585129634565,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 8924
    },
    {
      "epoch": 0.08925,
      "grad_norm": 0.6041119003990524,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 8925
    },
    {
      "epoch": 0.08926,
      "grad_norm": 0.6170325812343832,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 8926
    },
    {
      "epoch": 0.08927,
      "grad_norm": 0.5943880002209878,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 8927
    },
    {
      "epoch": 0.08928,
      "grad_norm": 0.6087426636265381,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 8928
    },
    {
      "epoch": 0.08929,
      "grad_norm": 0.6534682552946309,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 8929
    },
    {
      "epoch": 0.0893,
      "grad_norm": 0.7123983106901958,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 8930
    },
    {
      "epoch": 0.08931,
      "grad_norm": 0.6560939231760763,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 8931
    },
    {
      "epoch": 0.08932,
      "grad_norm": 0.5027522782217787,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 8932
    },
    {
      "epoch": 0.08933,
      "grad_norm": 0.4365213401530024,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 8933
    },
    {
      "epoch": 0.08934,
      "grad_norm": 0.5000766612819153,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 8934
    },
    {
      "epoch": 0.08935,
      "grad_norm": 0.5952483131640884,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 8935
    },
    {
      "epoch": 0.08936,
      "grad_norm": 0.6156724481814724,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 8936
    },
    {
      "epoch": 0.08937,
      "grad_norm": 0.5931999695378637,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 8937
    },
    {
      "epoch": 0.08938,
      "grad_norm": 0.6894161685221871,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 8938
    },
    {
      "epoch": 0.08939,
      "grad_norm": 0.7627353264346703,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 8939
    },
    {
      "epoch": 0.0894,
      "grad_norm": 0.7356138960546812,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 8940
    },
    {
      "epoch": 0.08941,
      "grad_norm": 0.6420508209011688,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 8941
    },
    {
      "epoch": 0.08942,
      "grad_norm": 0.6518833670204465,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 8942
    },
    {
      "epoch": 0.08943,
      "grad_norm": 0.6146219772409397,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 8943
    },
    {
      "epoch": 0.08944,
      "grad_norm": 0.6396512737329096,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 8944
    },
    {
      "epoch": 0.08945,
      "grad_norm": 0.6374698041220779,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 8945
    },
    {
      "epoch": 0.08946,
      "grad_norm": 0.7068156644089753,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 8946
    },
    {
      "epoch": 0.08947,
      "grad_norm": 0.845317356111474,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 8947
    },
    {
      "epoch": 0.08948,
      "grad_norm": 1.0368306661318563,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 8948
    },
    {
      "epoch": 0.08949,
      "grad_norm": 1.198555666195333,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 8949
    },
    {
      "epoch": 0.0895,
      "grad_norm": 0.5996385514208301,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 8950
    },
    {
      "epoch": 0.08951,
      "grad_norm": 0.7108449111037725,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 8951
    },
    {
      "epoch": 0.08952,
      "grad_norm": 0.855830753478774,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 8952
    },
    {
      "epoch": 0.08953,
      "grad_norm": 0.9206337595900704,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 8953
    },
    {
      "epoch": 0.08954,
      "grad_norm": 0.812196407142358,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 8954
    },
    {
      "epoch": 0.08955,
      "grad_norm": 0.7982919359283707,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 8955
    },
    {
      "epoch": 0.08956,
      "grad_norm": 0.8635629919504556,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 8956
    },
    {
      "epoch": 0.08957,
      "grad_norm": 0.9322894837024904,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 8957
    },
    {
      "epoch": 0.08958,
      "grad_norm": 0.9187549703391346,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 8958
    },
    {
      "epoch": 0.08959,
      "grad_norm": 0.9062695515862552,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8959
    },
    {
      "epoch": 0.0896,
      "grad_norm": 0.8138333486268774,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 8960
    },
    {
      "epoch": 0.08961,
      "grad_norm": 0.8118676867241735,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 8961
    },
    {
      "epoch": 0.08962,
      "grad_norm": 0.915869089395175,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 8962
    },
    {
      "epoch": 0.08963,
      "grad_norm": 1.105783806141911,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 8963
    },
    {
      "epoch": 0.08964,
      "grad_norm": 0.9871252968891374,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 8964
    },
    {
      "epoch": 0.08965,
      "grad_norm": 0.9621593755289617,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 8965
    },
    {
      "epoch": 0.08966,
      "grad_norm": 0.8845068329445493,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 8966
    },
    {
      "epoch": 0.08967,
      "grad_norm": 0.9262452245654436,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 8967
    },
    {
      "epoch": 0.08968,
      "grad_norm": 1.0690148845058771,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 8968
    },
    {
      "epoch": 0.08969,
      "grad_norm": 0.9874560303810536,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 8969
    },
    {
      "epoch": 0.0897,
      "grad_norm": 1.1933287056871495,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 8970
    },
    {
      "epoch": 0.08971,
      "grad_norm": 0.83496478128569,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 8971
    },
    {
      "epoch": 0.08972,
      "grad_norm": 0.8851270091462285,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 8972
    },
    {
      "epoch": 0.08973,
      "grad_norm": 0.7361881327430853,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 8973
    },
    {
      "epoch": 0.08974,
      "grad_norm": 0.6947872113441466,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 8974
    },
    {
      "epoch": 0.08975,
      "grad_norm": 0.6448548962054766,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 8975
    },
    {
      "epoch": 0.08976,
      "grad_norm": 0.6115477618993307,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 8976
    },
    {
      "epoch": 0.08977,
      "grad_norm": 0.6406580925867694,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 8977
    },
    {
      "epoch": 0.08978,
      "grad_norm": 0.5844796689209106,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 8978
    },
    {
      "epoch": 0.08979,
      "grad_norm": 0.4899670091468122,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 8979
    },
    {
      "epoch": 0.0898,
      "grad_norm": 0.45907274050387364,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 8980
    },
    {
      "epoch": 0.08981,
      "grad_norm": 0.4787661572615123,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 8981
    },
    {
      "epoch": 0.08982,
      "grad_norm": 0.5314004944865418,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 8982
    },
    {
      "epoch": 0.08983,
      "grad_norm": 0.5367270316206084,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 8983
    },
    {
      "epoch": 0.08984,
      "grad_norm": 0.6099160821198948,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 8984
    },
    {
      "epoch": 0.08985,
      "grad_norm": 0.6151453422803892,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 8985
    },
    {
      "epoch": 0.08986,
      "grad_norm": 0.6431362909917693,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 8986
    },
    {
      "epoch": 0.08987,
      "grad_norm": 0.6245835812002626,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 8987
    },
    {
      "epoch": 0.08988,
      "grad_norm": 0.65684747848799,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 8988
    },
    {
      "epoch": 0.08989,
      "grad_norm": 0.7624170463908312,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 8989
    },
    {
      "epoch": 0.0899,
      "grad_norm": 0.7803016187618669,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 8990
    },
    {
      "epoch": 0.08991,
      "grad_norm": 0.6199387610169692,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 8991
    },
    {
      "epoch": 0.08992,
      "grad_norm": 0.6328852031482564,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 8992
    },
    {
      "epoch": 0.08993,
      "grad_norm": 0.6444379781150428,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 8993
    },
    {
      "epoch": 0.08994,
      "grad_norm": 0.7004376193465449,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 8994
    },
    {
      "epoch": 0.08995,
      "grad_norm": 0.7021581496420412,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 8995
    },
    {
      "epoch": 0.08996,
      "grad_norm": 0.735896084022374,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 8996
    },
    {
      "epoch": 0.08997,
      "grad_norm": 0.7654208285833823,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 8997
    },
    {
      "epoch": 0.08998,
      "grad_norm": 0.8110039047646157,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 8998
    },
    {
      "epoch": 0.08999,
      "grad_norm": 0.9354160143582295,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 8999
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9976774961172428,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 9000
    },
    {
      "epoch": 0.09001,
      "grad_norm": 0.9736972254727577,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 9001
    },
    {
      "epoch": 0.09002,
      "grad_norm": 0.9273588401300575,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 9002
    },
    {
      "epoch": 0.09003,
      "grad_norm": 0.9369574420893962,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 9003
    },
    {
      "epoch": 0.09004,
      "grad_norm": 0.8974590725426277,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 9004
    },
    {
      "epoch": 0.09005,
      "grad_norm": 0.9593522302258589,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 9005
    },
    {
      "epoch": 0.09006,
      "grad_norm": 1.148323754626388,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 9006
    },
    {
      "epoch": 0.09007,
      "grad_norm": 0.7314690534751829,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 9007
    },
    {
      "epoch": 0.09008,
      "grad_norm": 0.7435332389856724,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 9008
    },
    {
      "epoch": 0.09009,
      "grad_norm": 0.7078932397563384,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 9009
    },
    {
      "epoch": 0.0901,
      "grad_norm": 0.7615661108195854,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 9010
    },
    {
      "epoch": 0.09011,
      "grad_norm": 0.6793695558724677,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 9011
    },
    {
      "epoch": 0.09012,
      "grad_norm": 0.5832604883899358,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 9012
    },
    {
      "epoch": 0.09013,
      "grad_norm": 0.6009269936700211,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 9013
    },
    {
      "epoch": 0.09014,
      "grad_norm": 0.5849630009375443,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 9014
    },
    {
      "epoch": 0.09015,
      "grad_norm": 0.6889294107758397,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 9015
    },
    {
      "epoch": 0.09016,
      "grad_norm": 0.7296383845886047,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 9016
    },
    {
      "epoch": 0.09017,
      "grad_norm": 0.6317455085918721,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 9017
    },
    {
      "epoch": 0.09018,
      "grad_norm": 0.6094236570193985,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 9018
    },
    {
      "epoch": 0.09019,
      "grad_norm": 0.6782840216217128,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 9019
    },
    {
      "epoch": 0.0902,
      "grad_norm": 0.7866077854724068,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 9020
    },
    {
      "epoch": 0.09021,
      "grad_norm": 0.9385063075453379,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 9021
    },
    {
      "epoch": 0.09022,
      "grad_norm": 1.0620409463514187,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 9022
    },
    {
      "epoch": 0.09023,
      "grad_norm": 0.7658741306051069,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 9023
    },
    {
      "epoch": 0.09024,
      "grad_norm": 0.6814552110746341,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 9024
    },
    {
      "epoch": 0.09025,
      "grad_norm": 0.7730019678384572,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 9025
    },
    {
      "epoch": 0.09026,
      "grad_norm": 0.7526237882680384,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 9026
    },
    {
      "epoch": 0.09027,
      "grad_norm": 0.6507122800031654,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 9027
    },
    {
      "epoch": 0.09028,
      "grad_norm": 0.6004305708920464,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 9028
    },
    {
      "epoch": 0.09029,
      "grad_norm": 0.524677033109727,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 9029
    },
    {
      "epoch": 0.0903,
      "grad_norm": 0.6033650045330714,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 9030
    },
    {
      "epoch": 0.09031,
      "grad_norm": 0.5871602565860976,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 9031
    },
    {
      "epoch": 0.09032,
      "grad_norm": 0.6156339772467885,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 9032
    },
    {
      "epoch": 0.09033,
      "grad_norm": 0.5735688885353999,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 9033
    },
    {
      "epoch": 0.09034,
      "grad_norm": 0.5650782637932048,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 9034
    },
    {
      "epoch": 0.09035,
      "grad_norm": 0.6052372993053132,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 9035
    },
    {
      "epoch": 0.09036,
      "grad_norm": 0.5476010632727801,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 9036
    },
    {
      "epoch": 0.09037,
      "grad_norm": 0.583873535910786,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 9037
    },
    {
      "epoch": 0.09038,
      "grad_norm": 0.7287923974736967,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 9038
    },
    {
      "epoch": 0.09039,
      "grad_norm": 0.9144241018735565,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 9039
    },
    {
      "epoch": 0.0904,
      "grad_norm": 1.0063720204208155,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 9040
    },
    {
      "epoch": 0.09041,
      "grad_norm": 0.9856961456597553,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 9041
    },
    {
      "epoch": 0.09042,
      "grad_norm": 0.854851014929688,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 9042
    },
    {
      "epoch": 0.09043,
      "grad_norm": 0.888431844356114,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 9043
    },
    {
      "epoch": 0.09044,
      "grad_norm": 1.017236877807041,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 9044
    },
    {
      "epoch": 0.09045,
      "grad_norm": 1.031807206439807,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 9045
    },
    {
      "epoch": 0.09046,
      "grad_norm": 0.85682286882739,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 9046
    },
    {
      "epoch": 0.09047,
      "grad_norm": 0.879826141603597,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 9047
    },
    {
      "epoch": 0.09048,
      "grad_norm": 0.8831965636813381,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 9048
    },
    {
      "epoch": 0.09049,
      "grad_norm": 0.9392095074031075,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 9049
    },
    {
      "epoch": 0.0905,
      "grad_norm": 0.7828170543785254,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 9050
    },
    {
      "epoch": 0.09051,
      "grad_norm": 0.8339468015119822,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 9051
    },
    {
      "epoch": 0.09052,
      "grad_norm": 1.0336225007587214,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 9052
    },
    {
      "epoch": 0.09053,
      "grad_norm": 1.1758495676250011,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 9053
    },
    {
      "epoch": 0.09054,
      "grad_norm": 0.6902237254747,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 9054
    },
    {
      "epoch": 0.09055,
      "grad_norm": 0.6385588548368929,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 9055
    },
    {
      "epoch": 0.09056,
      "grad_norm": 0.5682201761251565,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 9056
    },
    {
      "epoch": 0.09057,
      "grad_norm": 0.6058063245223638,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 9057
    },
    {
      "epoch": 0.09058,
      "grad_norm": 0.6296895298242686,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 9058
    },
    {
      "epoch": 0.09059,
      "grad_norm": 0.6548379189063352,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 9059
    },
    {
      "epoch": 0.0906,
      "grad_norm": 0.556996825598181,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 9060
    },
    {
      "epoch": 0.09061,
      "grad_norm": 0.521158274387155,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 9061
    },
    {
      "epoch": 0.09062,
      "grad_norm": 0.49652995554371476,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 9062
    },
    {
      "epoch": 0.09063,
      "grad_norm": 0.4672371999627178,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 9063
    },
    {
      "epoch": 0.09064,
      "grad_norm": 0.39411015877585076,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 9064
    },
    {
      "epoch": 0.09065,
      "grad_norm": 0.4138073282094934,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 9065
    },
    {
      "epoch": 0.09066,
      "grad_norm": 0.4761901801806996,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 9066
    },
    {
      "epoch": 0.09067,
      "grad_norm": 0.529331487923414,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 9067
    },
    {
      "epoch": 0.09068,
      "grad_norm": 0.5533215938891374,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 9068
    },
    {
      "epoch": 0.09069,
      "grad_norm": 0.6302304828764494,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 9069
    },
    {
      "epoch": 0.0907,
      "grad_norm": 0.7733370160113519,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 9070
    },
    {
      "epoch": 0.09071,
      "grad_norm": 0.9201950339828795,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 9071
    },
    {
      "epoch": 0.09072,
      "grad_norm": 1.0000279216059271,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 9072
    },
    {
      "epoch": 0.09073,
      "grad_norm": 0.9703960584155082,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 9073
    },
    {
      "epoch": 0.09074,
      "grad_norm": 0.7378599237261866,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 9074
    },
    {
      "epoch": 0.09075,
      "grad_norm": 0.6334621773210859,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 9075
    },
    {
      "epoch": 0.09076,
      "grad_norm": 0.7233595328453737,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 9076
    },
    {
      "epoch": 0.09077,
      "grad_norm": 0.6798870748945902,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 9077
    },
    {
      "epoch": 0.09078,
      "grad_norm": 0.6666031376042267,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 9078
    },
    {
      "epoch": 0.09079,
      "grad_norm": 0.8196363695329622,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 9079
    },
    {
      "epoch": 0.0908,
      "grad_norm": 0.8369563991262994,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 9080
    },
    {
      "epoch": 0.09081,
      "grad_norm": 0.7879366661568333,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 9081
    },
    {
      "epoch": 0.09082,
      "grad_norm": 0.9098587816411382,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 9082
    },
    {
      "epoch": 0.09083,
      "grad_norm": 0.9134982635005068,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 9083
    },
    {
      "epoch": 0.09084,
      "grad_norm": 0.9505568276434497,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 9084
    },
    {
      "epoch": 0.09085,
      "grad_norm": 0.8638717507088994,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 9085
    },
    {
      "epoch": 0.09086,
      "grad_norm": 0.8895115781730211,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 9086
    },
    {
      "epoch": 0.09087,
      "grad_norm": 1.060170360539398,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 9087
    },
    {
      "epoch": 0.09088,
      "grad_norm": 0.901927345203865,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 9088
    },
    {
      "epoch": 0.09089,
      "grad_norm": 0.785289100063639,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 9089
    },
    {
      "epoch": 0.0909,
      "grad_norm": 0.8620215806982862,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 9090
    },
    {
      "epoch": 0.09091,
      "grad_norm": 0.7979739519204804,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 9091
    },
    {
      "epoch": 0.09092,
      "grad_norm": 0.8103723620422554,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 9092
    },
    {
      "epoch": 0.09093,
      "grad_norm": 0.7761450446921329,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 9093
    },
    {
      "epoch": 0.09094,
      "grad_norm": 0.7568316928415706,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 9094
    },
    {
      "epoch": 0.09095,
      "grad_norm": 0.759347481387229,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 9095
    },
    {
      "epoch": 0.09096,
      "grad_norm": 0.766744250041468,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 9096
    },
    {
      "epoch": 0.09097,
      "grad_norm": 0.7055484847046865,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 9097
    },
    {
      "epoch": 0.09098,
      "grad_norm": 0.7249870657972629,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 9098
    },
    {
      "epoch": 0.09099,
      "grad_norm": 0.794584702637224,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 9099
    },
    {
      "epoch": 0.091,
      "grad_norm": 0.9292342497012281,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 9100
    },
    {
      "epoch": 0.09101,
      "grad_norm": 1.1151713569446484,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 9101
    },
    {
      "epoch": 0.09102,
      "grad_norm": 0.9856776803625353,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 9102
    },
    {
      "epoch": 0.09103,
      "grad_norm": 0.8883418202472645,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 9103
    },
    {
      "epoch": 0.09104,
      "grad_norm": 0.7209861355720855,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 9104
    },
    {
      "epoch": 0.09105,
      "grad_norm": 0.6974273614545563,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 9105
    },
    {
      "epoch": 0.09106,
      "grad_norm": 0.7573369118352933,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 9106
    },
    {
      "epoch": 0.09107,
      "grad_norm": 0.8573483561968565,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 9107
    },
    {
      "epoch": 0.09108,
      "grad_norm": 0.843416858845127,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9108
    },
    {
      "epoch": 0.09109,
      "grad_norm": 0.7331572974610379,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 9109
    },
    {
      "epoch": 0.0911,
      "grad_norm": 0.6866482073169092,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 9110
    },
    {
      "epoch": 0.09111,
      "grad_norm": 0.6146584810372685,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 9111
    },
    {
      "epoch": 0.09112,
      "grad_norm": 0.6233531858518306,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 9112
    },
    {
      "epoch": 0.09113,
      "grad_norm": 0.6630440885467346,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 9113
    },
    {
      "epoch": 0.09114,
      "grad_norm": 0.6675611226061425,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 9114
    },
    {
      "epoch": 0.09115,
      "grad_norm": 0.6851264099217901,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 9115
    },
    {
      "epoch": 0.09116,
      "grad_norm": 0.6743396804700748,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 9116
    },
    {
      "epoch": 0.09117,
      "grad_norm": 0.6349134476822885,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 9117
    },
    {
      "epoch": 0.09118,
      "grad_norm": 0.5734871976924563,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 9118
    },
    {
      "epoch": 0.09119,
      "grad_norm": 0.5722066825396821,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 9119
    },
    {
      "epoch": 0.0912,
      "grad_norm": 0.5747337761394117,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 9120
    },
    {
      "epoch": 0.09121,
      "grad_norm": 0.5703997923017384,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 9121
    },
    {
      "epoch": 0.09122,
      "grad_norm": 0.5937813885823265,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 9122
    },
    {
      "epoch": 0.09123,
      "grad_norm": 0.5793339846399808,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 9123
    },
    {
      "epoch": 0.09124,
      "grad_norm": 0.7079624241350131,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 9124
    },
    {
      "epoch": 0.09125,
      "grad_norm": 0.9363241418641047,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 9125
    },
    {
      "epoch": 0.09126,
      "grad_norm": 1.123535275176799,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 9126
    },
    {
      "epoch": 0.09127,
      "grad_norm": 0.9098450109303394,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 9127
    },
    {
      "epoch": 0.09128,
      "grad_norm": 0.7695802519035674,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 9128
    },
    {
      "epoch": 0.09129,
      "grad_norm": 0.6725773758279523,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 9129
    },
    {
      "epoch": 0.0913,
      "grad_norm": 0.7516703750695625,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 9130
    },
    {
      "epoch": 0.09131,
      "grad_norm": 0.7232356676372639,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 9131
    },
    {
      "epoch": 0.09132,
      "grad_norm": 0.6859783867977578,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 9132
    },
    {
      "epoch": 0.09133,
      "grad_norm": 0.6441924486606353,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 9133
    },
    {
      "epoch": 0.09134,
      "grad_norm": 0.655918695866971,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 9134
    },
    {
      "epoch": 0.09135,
      "grad_norm": 0.7860018696069528,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 9135
    },
    {
      "epoch": 0.09136,
      "grad_norm": 0.901439083024281,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 9136
    },
    {
      "epoch": 0.09137,
      "grad_norm": 1.0327495887186917,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 9137
    },
    {
      "epoch": 0.09138,
      "grad_norm": 0.9559641464369752,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 9138
    },
    {
      "epoch": 0.09139,
      "grad_norm": 0.8021385645026318,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 9139
    },
    {
      "epoch": 0.0914,
      "grad_norm": 0.6983879716817356,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 9140
    },
    {
      "epoch": 0.09141,
      "grad_norm": 0.6229466794575317,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 9141
    },
    {
      "epoch": 0.09142,
      "grad_norm": 0.5394263568511116,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 9142
    },
    {
      "epoch": 0.09143,
      "grad_norm": 0.5799460322868719,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 9143
    },
    {
      "epoch": 0.09144,
      "grad_norm": 0.7115794520642408,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 9144
    },
    {
      "epoch": 0.09145,
      "grad_norm": 0.9250322234700112,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 9145
    },
    {
      "epoch": 0.09146,
      "grad_norm": 1.1493765993374974,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 9146
    },
    {
      "epoch": 0.09147,
      "grad_norm": 0.6766953092233843,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 9147
    },
    {
      "epoch": 0.09148,
      "grad_norm": 0.615481970405836,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 9148
    },
    {
      "epoch": 0.09149,
      "grad_norm": 0.6468342965420243,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 9149
    },
    {
      "epoch": 0.0915,
      "grad_norm": 0.5695495986990226,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 9150
    },
    {
      "epoch": 0.09151,
      "grad_norm": 0.6322625762493587,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 9151
    },
    {
      "epoch": 0.09152,
      "grad_norm": 0.7095634250401223,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 9152
    },
    {
      "epoch": 0.09153,
      "grad_norm": 0.7562377869290767,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 9153
    },
    {
      "epoch": 0.09154,
      "grad_norm": 0.730100776007017,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 9154
    },
    {
      "epoch": 0.09155,
      "grad_norm": 0.6476046997499159,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 9155
    },
    {
      "epoch": 0.09156,
      "grad_norm": 0.5966611216224073,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 9156
    },
    {
      "epoch": 0.09157,
      "grad_norm": 0.729366774751956,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 9157
    },
    {
      "epoch": 0.09158,
      "grad_norm": 0.9225886030022502,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 9158
    },
    {
      "epoch": 0.09159,
      "grad_norm": 1.0003949696131078,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 9159
    },
    {
      "epoch": 0.0916,
      "grad_norm": 0.8838504831163226,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 9160
    },
    {
      "epoch": 0.09161,
      "grad_norm": 1.0435641927134112,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 9161
    },
    {
      "epoch": 0.09162,
      "grad_norm": 0.9849828677019289,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 9162
    },
    {
      "epoch": 0.09163,
      "grad_norm": 0.8423932570011732,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 9163
    },
    {
      "epoch": 0.09164,
      "grad_norm": 0.8118112808914887,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 9164
    },
    {
      "epoch": 0.09165,
      "grad_norm": 0.696647926319978,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 9165
    },
    {
      "epoch": 0.09166,
      "grad_norm": 0.7050748772822608,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 9166
    },
    {
      "epoch": 0.09167,
      "grad_norm": 0.6984810250947892,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 9167
    },
    {
      "epoch": 0.09168,
      "grad_norm": 0.7429002727610159,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 9168
    },
    {
      "epoch": 0.09169,
      "grad_norm": 0.8700828623089742,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 9169
    },
    {
      "epoch": 0.0917,
      "grad_norm": 0.8255131562046868,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 9170
    },
    {
      "epoch": 0.09171,
      "grad_norm": 0.8257708158169661,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 9171
    },
    {
      "epoch": 0.09172,
      "grad_norm": 0.9821171250112688,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 9172
    },
    {
      "epoch": 0.09173,
      "grad_norm": 1.1101867989693224,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 9173
    },
    {
      "epoch": 0.09174,
      "grad_norm": 0.8606212362625643,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 9174
    },
    {
      "epoch": 0.09175,
      "grad_norm": 0.7238484783389175,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 9175
    },
    {
      "epoch": 0.09176,
      "grad_norm": 0.7778471274442456,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 9176
    },
    {
      "epoch": 0.09177,
      "grad_norm": 0.7942453477102529,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 9177
    },
    {
      "epoch": 0.09178,
      "grad_norm": 0.6875439916755981,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 9178
    },
    {
      "epoch": 0.09179,
      "grad_norm": 0.7054405486618374,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9179
    },
    {
      "epoch": 0.0918,
      "grad_norm": 0.7085929286496793,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9180
    },
    {
      "epoch": 0.09181,
      "grad_norm": 0.6868386510039214,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 9181
    },
    {
      "epoch": 0.09182,
      "grad_norm": 0.693686693395854,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 9182
    },
    {
      "epoch": 0.09183,
      "grad_norm": 0.6969673134627956,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 9183
    },
    {
      "epoch": 0.09184,
      "grad_norm": 0.6607879678388875,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 9184
    },
    {
      "epoch": 0.09185,
      "grad_norm": 0.6097277133504103,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 9185
    },
    {
      "epoch": 0.09186,
      "grad_norm": 0.6707275075667879,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 9186
    },
    {
      "epoch": 0.09187,
      "grad_norm": 0.631051696187288,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 9187
    },
    {
      "epoch": 0.09188,
      "grad_norm": 0.5850735017399692,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 9188
    },
    {
      "epoch": 0.09189,
      "grad_norm": 0.5909062131280334,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 9189
    },
    {
      "epoch": 0.0919,
      "grad_norm": 0.5804461230556179,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 9190
    },
    {
      "epoch": 0.09191,
      "grad_norm": 0.74233665537322,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 9191
    },
    {
      "epoch": 0.09192,
      "grad_norm": 1.0070740958090563,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 9192
    },
    {
      "epoch": 0.09193,
      "grad_norm": 1.3162433610093374,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 9193
    },
    {
      "epoch": 0.09194,
      "grad_norm": 0.4948613105868042,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9194
    },
    {
      "epoch": 0.09195,
      "grad_norm": 0.683388327325886,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 9195
    },
    {
      "epoch": 0.09196,
      "grad_norm": 0.8065888189569945,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 9196
    },
    {
      "epoch": 0.09197,
      "grad_norm": 0.7584747106256147,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 9197
    },
    {
      "epoch": 0.09198,
      "grad_norm": 0.7753320297151186,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 9198
    },
    {
      "epoch": 0.09199,
      "grad_norm": 0.7824527448547508,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 9199
    },
    {
      "epoch": 0.092,
      "grad_norm": 0.7775469360725809,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 9200
    },
    {
      "epoch": 0.09201,
      "grad_norm": 0.7680517906058487,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 9201
    },
    {
      "epoch": 0.09202,
      "grad_norm": 0.8531435904229961,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 9202
    },
    {
      "epoch": 0.09203,
      "grad_norm": 0.8780606886961511,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 9203
    },
    {
      "epoch": 0.09204,
      "grad_norm": 0.7891522389637466,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 9204
    },
    {
      "epoch": 0.09205,
      "grad_norm": 0.7455931059386063,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 9205
    },
    {
      "epoch": 0.09206,
      "grad_norm": 0.7535666910826962,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 9206
    },
    {
      "epoch": 0.09207,
      "grad_norm": 0.6917320820801799,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 9207
    },
    {
      "epoch": 0.09208,
      "grad_norm": 0.704126699875504,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 9208
    },
    {
      "epoch": 0.09209,
      "grad_norm": 0.8523565990050219,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 9209
    },
    {
      "epoch": 0.0921,
      "grad_norm": 0.9685150489510521,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 9210
    },
    {
      "epoch": 0.09211,
      "grad_norm": 1.1138999636053162,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 9211
    },
    {
      "epoch": 0.09212,
      "grad_norm": 0.7870842232400849,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 9212
    },
    {
      "epoch": 0.09213,
      "grad_norm": 0.7697833258515777,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 9213
    },
    {
      "epoch": 0.09214,
      "grad_norm": 0.6846278043413059,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 9214
    },
    {
      "epoch": 0.09215,
      "grad_norm": 0.6743239347064265,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 9215
    },
    {
      "epoch": 0.09216,
      "grad_norm": 0.7010992201639914,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 9216
    },
    {
      "epoch": 0.09217,
      "grad_norm": 0.7476228942446806,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 9217
    },
    {
      "epoch": 0.09218,
      "grad_norm": 0.8348321650225655,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 9218
    },
    {
      "epoch": 0.09219,
      "grad_norm": 0.9157211023791693,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 9219
    },
    {
      "epoch": 0.0922,
      "grad_norm": 0.83049456838394,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 9220
    },
    {
      "epoch": 0.09221,
      "grad_norm": 0.6939065160104407,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 9221
    },
    {
      "epoch": 0.09222,
      "grad_norm": 0.692227128387877,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 9222
    },
    {
      "epoch": 0.09223,
      "grad_norm": 0.6206638261808557,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 9223
    },
    {
      "epoch": 0.09224,
      "grad_norm": 0.6155175720948994,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 9224
    },
    {
      "epoch": 0.09225,
      "grad_norm": 0.5763438884147025,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 9225
    },
    {
      "epoch": 0.09226,
      "grad_norm": 0.5792522714322638,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 9226
    },
    {
      "epoch": 0.09227,
      "grad_norm": 0.5698582837332627,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 9227
    },
    {
      "epoch": 0.09228,
      "grad_norm": 0.5554181806254156,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 9228
    },
    {
      "epoch": 0.09229,
      "grad_norm": 0.6496771344535306,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 9229
    },
    {
      "epoch": 0.0923,
      "grad_norm": 0.8079631630379661,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 9230
    },
    {
      "epoch": 0.09231,
      "grad_norm": 1.001327955726392,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 9231
    },
    {
      "epoch": 0.09232,
      "grad_norm": 1.0151152557949434,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 9232
    },
    {
      "epoch": 0.09233,
      "grad_norm": 0.7597476613283288,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 9233
    },
    {
      "epoch": 0.09234,
      "grad_norm": 0.7793623041860894,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 9234
    },
    {
      "epoch": 0.09235,
      "grad_norm": 0.8181572642649179,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 9235
    },
    {
      "epoch": 0.09236,
      "grad_norm": 0.7748209029826961,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 9236
    },
    {
      "epoch": 0.09237,
      "grad_norm": 0.7417587962678789,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 9237
    },
    {
      "epoch": 0.09238,
      "grad_norm": 0.7302829503235564,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 9238
    },
    {
      "epoch": 0.09239,
      "grad_norm": 0.8200299371821417,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 9239
    },
    {
      "epoch": 0.0924,
      "grad_norm": 0.7745701228788062,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 9240
    },
    {
      "epoch": 0.09241,
      "grad_norm": 0.8722251162417323,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 9241
    },
    {
      "epoch": 0.09242,
      "grad_norm": 0.890375969288489,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 9242
    },
    {
      "epoch": 0.09243,
      "grad_norm": 0.8425579306067554,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 9243
    },
    {
      "epoch": 0.09244,
      "grad_norm": 0.750374270081524,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 9244
    },
    {
      "epoch": 0.09245,
      "grad_norm": 0.7433474953602284,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 9245
    },
    {
      "epoch": 0.09246,
      "grad_norm": 0.8399736354443268,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 9246
    },
    {
      "epoch": 0.09247,
      "grad_norm": 0.8952276071336189,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 9247
    },
    {
      "epoch": 0.09248,
      "grad_norm": 0.9460058010349833,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 9248
    },
    {
      "epoch": 0.09249,
      "grad_norm": 1.028348281495935,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 9249
    },
    {
      "epoch": 0.0925,
      "grad_norm": 0.8994505559170306,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 9250
    },
    {
      "epoch": 0.09251,
      "grad_norm": 0.8661162024839089,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 9251
    },
    {
      "epoch": 0.09252,
      "grad_norm": 0.7576653757537349,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 9252
    },
    {
      "epoch": 0.09253,
      "grad_norm": 0.7300578749253667,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 9253
    },
    {
      "epoch": 0.09254,
      "grad_norm": 0.6655033765923661,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 9254
    },
    {
      "epoch": 0.09255,
      "grad_norm": 0.67210004385522,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 9255
    },
    {
      "epoch": 0.09256,
      "grad_norm": 0.6009374878800777,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 9256
    },
    {
      "epoch": 0.09257,
      "grad_norm": 0.5939860882674391,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 9257
    },
    {
      "epoch": 0.09258,
      "grad_norm": 0.6273131160536646,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 9258
    },
    {
      "epoch": 0.09259,
      "grad_norm": 0.6931906821491065,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 9259
    },
    {
      "epoch": 0.0926,
      "grad_norm": 0.6843836039174134,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 9260
    },
    {
      "epoch": 0.09261,
      "grad_norm": 0.664844852067105,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 9261
    },
    {
      "epoch": 0.09262,
      "grad_norm": 0.7149590284934766,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 9262
    },
    {
      "epoch": 0.09263,
      "grad_norm": 0.799686238491033,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 9263
    },
    {
      "epoch": 0.09264,
      "grad_norm": 0.9184087623202745,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 9264
    },
    {
      "epoch": 0.09265,
      "grad_norm": 0.8917999340633649,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 9265
    },
    {
      "epoch": 0.09266,
      "grad_norm": 1.0690273464224147,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 9266
    },
    {
      "epoch": 0.09267,
      "grad_norm": 0.9224213714197611,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 9267
    },
    {
      "epoch": 0.09268,
      "grad_norm": 0.7819872181594957,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 9268
    },
    {
      "epoch": 0.09269,
      "grad_norm": 0.8903337895871032,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 9269
    },
    {
      "epoch": 0.0927,
      "grad_norm": 0.8789410698039444,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 9270
    },
    {
      "epoch": 0.09271,
      "grad_norm": 0.8151226774452855,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 9271
    },
    {
      "epoch": 0.09272,
      "grad_norm": 0.6687056930008045,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 9272
    },
    {
      "epoch": 0.09273,
      "grad_norm": 0.6975950766902023,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 9273
    },
    {
      "epoch": 0.09274,
      "grad_norm": 0.704466186613381,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 9274
    },
    {
      "epoch": 0.09275,
      "grad_norm": 0.7505976895896498,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 9275
    },
    {
      "epoch": 0.09276,
      "grad_norm": 0.7764606026959862,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 9276
    },
    {
      "epoch": 0.09277,
      "grad_norm": 0.635369630015743,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9277
    },
    {
      "epoch": 0.09278,
      "grad_norm": 0.6241312910359976,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 9278
    },
    {
      "epoch": 0.09279,
      "grad_norm": 0.6872238724869792,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 9279
    },
    {
      "epoch": 0.0928,
      "grad_norm": 0.7388825536520087,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 9280
    },
    {
      "epoch": 0.09281,
      "grad_norm": 0.7389024774211659,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 9281
    },
    {
      "epoch": 0.09282,
      "grad_norm": 0.7721013112557448,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 9282
    },
    {
      "epoch": 0.09283,
      "grad_norm": 0.8336050701348083,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 9283
    },
    {
      "epoch": 0.09284,
      "grad_norm": 1.0158256883508041,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 9284
    },
    {
      "epoch": 0.09285,
      "grad_norm": 1.18058053789904,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 9285
    },
    {
      "epoch": 0.09286,
      "grad_norm": 0.8343774515252792,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 9286
    },
    {
      "epoch": 0.09287,
      "grad_norm": 0.7308200337417591,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 9287
    },
    {
      "epoch": 0.09288,
      "grad_norm": 0.6891633376361983,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 9288
    },
    {
      "epoch": 0.09289,
      "grad_norm": 0.6029232592146974,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 9289
    },
    {
      "epoch": 0.0929,
      "grad_norm": 0.5626416862126429,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9290
    },
    {
      "epoch": 0.09291,
      "grad_norm": 0.5779583984794753,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 9291
    },
    {
      "epoch": 0.09292,
      "grad_norm": 0.5276189332666765,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 9292
    },
    {
      "epoch": 0.09293,
      "grad_norm": 0.4601373246181229,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 9293
    },
    {
      "epoch": 0.09294,
      "grad_norm": 0.5858557148598581,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 9294
    },
    {
      "epoch": 0.09295,
      "grad_norm": 0.6570320858383965,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 9295
    },
    {
      "epoch": 0.09296,
      "grad_norm": 0.7058986690585867,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 9296
    },
    {
      "epoch": 0.09297,
      "grad_norm": 0.7237851928274343,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 9297
    },
    {
      "epoch": 0.09298,
      "grad_norm": 0.7244115678072608,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 9298
    },
    {
      "epoch": 0.09299,
      "grad_norm": 0.7023860071795822,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 9299
    },
    {
      "epoch": 0.093,
      "grad_norm": 0.6893885792882112,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 9300
    },
    {
      "epoch": 0.09301,
      "grad_norm": 0.7771637507723111,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 9301
    },
    {
      "epoch": 0.09302,
      "grad_norm": 0.8353111718797918,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 9302
    },
    {
      "epoch": 0.09303,
      "grad_norm": 0.9443868135718669,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 9303
    },
    {
      "epoch": 0.09304,
      "grad_norm": 0.9340284309555337,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 9304
    },
    {
      "epoch": 0.09305,
      "grad_norm": 0.821105987408757,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 9305
    },
    {
      "epoch": 0.09306,
      "grad_norm": 0.6798841287919474,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 9306
    },
    {
      "epoch": 0.09307,
      "grad_norm": 0.6308321339032982,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 9307
    },
    {
      "epoch": 0.09308,
      "grad_norm": 0.6601870406757323,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 9308
    },
    {
      "epoch": 0.09309,
      "grad_norm": 0.6442557609343686,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9309
    },
    {
      "epoch": 0.0931,
      "grad_norm": 0.6661782172293007,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 9310
    },
    {
      "epoch": 0.09311,
      "grad_norm": 0.694502121054022,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 9311
    },
    {
      "epoch": 0.09312,
      "grad_norm": 0.7404429863152544,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 9312
    },
    {
      "epoch": 0.09313,
      "grad_norm": 0.7933542321293575,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 9313
    },
    {
      "epoch": 0.09314,
      "grad_norm": 0.801634188665558,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 9314
    },
    {
      "epoch": 0.09315,
      "grad_norm": 0.8599424000518597,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 9315
    },
    {
      "epoch": 0.09316,
      "grad_norm": 0.8488962903419432,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 9316
    },
    {
      "epoch": 0.09317,
      "grad_norm": 0.8383529764637065,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 9317
    },
    {
      "epoch": 0.09318,
      "grad_norm": 0.8558462806350178,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 9318
    },
    {
      "epoch": 0.09319,
      "grad_norm": 0.9413535849970353,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 9319
    },
    {
      "epoch": 0.0932,
      "grad_norm": 1.2104907947773857,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 9320
    },
    {
      "epoch": 0.09321,
      "grad_norm": 0.9438406966956778,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 9321
    },
    {
      "epoch": 0.09322,
      "grad_norm": 0.8654878853235997,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 9322
    },
    {
      "epoch": 0.09323,
      "grad_norm": 0.7716832268633257,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 9323
    },
    {
      "epoch": 0.09324,
      "grad_norm": 0.9620871505328507,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 9324
    },
    {
      "epoch": 0.09325,
      "grad_norm": 1.033201062271869,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 9325
    },
    {
      "epoch": 0.09326,
      "grad_norm": 0.9180758393766172,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 9326
    },
    {
      "epoch": 0.09327,
      "grad_norm": 0.9604068125223971,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 9327
    },
    {
      "epoch": 0.09328,
      "grad_norm": 0.8487522338913149,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 9328
    },
    {
      "epoch": 0.09329,
      "grad_norm": 0.7140038231334875,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 9329
    },
    {
      "epoch": 0.0933,
      "grad_norm": 0.6871213262271845,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 9330
    },
    {
      "epoch": 0.09331,
      "grad_norm": 0.7546334768507528,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 9331
    },
    {
      "epoch": 0.09332,
      "grad_norm": 0.8370311645065874,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 9332
    },
    {
      "epoch": 0.09333,
      "grad_norm": 1.1267395055615177,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 9333
    },
    {
      "epoch": 0.09334,
      "grad_norm": 0.8851656694755474,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 9334
    },
    {
      "epoch": 0.09335,
      "grad_norm": 0.6895247005069166,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 9335
    },
    {
      "epoch": 0.09336,
      "grad_norm": 0.7529887012918819,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 9336
    },
    {
      "epoch": 0.09337,
      "grad_norm": 0.8968776641201162,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 9337
    },
    {
      "epoch": 0.09338,
      "grad_norm": 0.969190767069404,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 9338
    },
    {
      "epoch": 0.09339,
      "grad_norm": 0.9096354758398183,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 9339
    },
    {
      "epoch": 0.0934,
      "grad_norm": 1.0248240766167955,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 9340
    },
    {
      "epoch": 0.09341,
      "grad_norm": 0.8411462432609955,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 9341
    },
    {
      "epoch": 0.09342,
      "grad_norm": 0.7516218020852315,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 9342
    },
    {
      "epoch": 0.09343,
      "grad_norm": 0.8168975424156755,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 9343
    },
    {
      "epoch": 0.09344,
      "grad_norm": 0.8281071489261201,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 9344
    },
    {
      "epoch": 0.09345,
      "grad_norm": 0.7971176476434507,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 9345
    },
    {
      "epoch": 0.09346,
      "grad_norm": 0.7256703424187483,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 9346
    },
    {
      "epoch": 0.09347,
      "grad_norm": 0.625000462139122,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 9347
    },
    {
      "epoch": 0.09348,
      "grad_norm": 0.5650513642145435,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 9348
    },
    {
      "epoch": 0.09349,
      "grad_norm": 0.5637070822633384,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 9349
    },
    {
      "epoch": 0.0935,
      "grad_norm": 0.5477607998488464,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 9350
    },
    {
      "epoch": 0.09351,
      "grad_norm": 0.5967568854844185,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 9351
    },
    {
      "epoch": 0.09352,
      "grad_norm": 0.6324408848296686,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 9352
    },
    {
      "epoch": 0.09353,
      "grad_norm": 0.5988050971610268,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 9353
    },
    {
      "epoch": 0.09354,
      "grad_norm": 0.6587621555860957,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 9354
    },
    {
      "epoch": 0.09355,
      "grad_norm": 0.6913770140531089,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 9355
    },
    {
      "epoch": 0.09356,
      "grad_norm": 0.8268308580950648,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 9356
    },
    {
      "epoch": 0.09357,
      "grad_norm": 0.9832141390606829,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 9357
    },
    {
      "epoch": 0.09358,
      "grad_norm": 1.0264443343215965,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 9358
    },
    {
      "epoch": 0.09359,
      "grad_norm": 0.8022400111008331,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 9359
    },
    {
      "epoch": 0.0936,
      "grad_norm": 0.7232160641932127,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 9360
    },
    {
      "epoch": 0.09361,
      "grad_norm": 0.8345708225298007,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 9361
    },
    {
      "epoch": 0.09362,
      "grad_norm": 0.864583150263912,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 9362
    },
    {
      "epoch": 0.09363,
      "grad_norm": 0.8449596309061119,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 9363
    },
    {
      "epoch": 0.09364,
      "grad_norm": 0.721880849977252,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 9364
    },
    {
      "epoch": 0.09365,
      "grad_norm": 0.6721169386273598,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 9365
    },
    {
      "epoch": 0.09366,
      "grad_norm": 0.5176745603655055,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 9366
    },
    {
      "epoch": 0.09367,
      "grad_norm": 0.5663365559445198,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 9367
    },
    {
      "epoch": 0.09368,
      "grad_norm": 0.5588855629457294,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 9368
    },
    {
      "epoch": 0.09369,
      "grad_norm": 0.558318290254828,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 9369
    },
    {
      "epoch": 0.0937,
      "grad_norm": 0.5777087869339129,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 9370
    },
    {
      "epoch": 0.09371,
      "grad_norm": 0.6683550838053252,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 9371
    },
    {
      "epoch": 0.09372,
      "grad_norm": 0.8106915641791952,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 9372
    },
    {
      "epoch": 0.09373,
      "grad_norm": 1.0420652181706669,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 9373
    },
    {
      "epoch": 0.09374,
      "grad_norm": 0.9030870891200452,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 9374
    },
    {
      "epoch": 0.09375,
      "grad_norm": 0.6642846536487922,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 9375
    },
    {
      "epoch": 0.09376,
      "grad_norm": 0.5926013599625012,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 9376
    },
    {
      "epoch": 0.09377,
      "grad_norm": 0.7333731763197545,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 9377
    },
    {
      "epoch": 0.09378,
      "grad_norm": 0.7205911448453179,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 9378
    },
    {
      "epoch": 0.09379,
      "grad_norm": 0.6261002234170417,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 9379
    },
    {
      "epoch": 0.0938,
      "grad_norm": 0.5831220315530415,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 9380
    },
    {
      "epoch": 0.09381,
      "grad_norm": 0.6453696033833166,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 9381
    },
    {
      "epoch": 0.09382,
      "grad_norm": 0.7447834448687087,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 9382
    },
    {
      "epoch": 0.09383,
      "grad_norm": 0.8017088185500696,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 9383
    },
    {
      "epoch": 0.09384,
      "grad_norm": 0.749499774993177,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 9384
    },
    {
      "epoch": 0.09385,
      "grad_norm": 0.7694390492566549,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 9385
    },
    {
      "epoch": 0.09386,
      "grad_norm": 0.8555158072501666,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 9386
    },
    {
      "epoch": 0.09387,
      "grad_norm": 0.8776130218746526,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 9387
    },
    {
      "epoch": 0.09388,
      "grad_norm": 0.9970607046729825,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 9388
    },
    {
      "epoch": 0.09389,
      "grad_norm": 1.0116948164111348,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 9389
    },
    {
      "epoch": 0.0939,
      "grad_norm": 0.8607572473294256,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 9390
    },
    {
      "epoch": 0.09391,
      "grad_norm": 0.8141023279481875,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 9391
    },
    {
      "epoch": 0.09392,
      "grad_norm": 0.8221337642937177,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 9392
    },
    {
      "epoch": 0.09393,
      "grad_norm": 0.8631165883014514,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 9393
    },
    {
      "epoch": 0.09394,
      "grad_norm": 0.8542446307614803,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 9394
    },
    {
      "epoch": 0.09395,
      "grad_norm": 0.7984540124391786,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 9395
    },
    {
      "epoch": 0.09396,
      "grad_norm": 0.833132121750485,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 9396
    },
    {
      "epoch": 0.09397,
      "grad_norm": 0.9833442838193437,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 9397
    },
    {
      "epoch": 0.09398,
      "grad_norm": 1.003667576167635,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 9398
    },
    {
      "epoch": 0.09399,
      "grad_norm": 0.9559424149218053,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 9399
    },
    {
      "epoch": 0.094,
      "grad_norm": 0.9201692580630481,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 9400
    },
    {
      "epoch": 0.09401,
      "grad_norm": 0.8825427431804942,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 9401
    },
    {
      "epoch": 0.09402,
      "grad_norm": 0.7898577204069231,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 9402
    },
    {
      "epoch": 0.09403,
      "grad_norm": 0.7350505587815206,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 9403
    },
    {
      "epoch": 0.09404,
      "grad_norm": 0.7291684122942435,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 9404
    },
    {
      "epoch": 0.09405,
      "grad_norm": 0.8162712076769146,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 9405
    },
    {
      "epoch": 0.09406,
      "grad_norm": 0.9155173238132704,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 9406
    },
    {
      "epoch": 0.09407,
      "grad_norm": 0.8628871245046202,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 9407
    },
    {
      "epoch": 0.09408,
      "grad_norm": 0.7194018791800877,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 9408
    },
    {
      "epoch": 0.09409,
      "grad_norm": 0.7409924782459923,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 9409
    },
    {
      "epoch": 0.0941,
      "grad_norm": 0.7794190789815792,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 9410
    },
    {
      "epoch": 0.09411,
      "grad_norm": 0.8734670824027785,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 9411
    },
    {
      "epoch": 0.09412,
      "grad_norm": 0.8088836225010004,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 9412
    },
    {
      "epoch": 0.09413,
      "grad_norm": 0.7602993796261419,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9413
    },
    {
      "epoch": 0.09414,
      "grad_norm": 0.6810778606987302,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 9414
    },
    {
      "epoch": 0.09415,
      "grad_norm": 0.6032263201656475,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 9415
    },
    {
      "epoch": 0.09416,
      "grad_norm": 0.4846341007249403,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 9416
    },
    {
      "epoch": 0.09417,
      "grad_norm": 0.5073097020665416,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 9417
    },
    {
      "epoch": 0.09418,
      "grad_norm": 0.466812325936142,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 9418
    },
    {
      "epoch": 0.09419,
      "grad_norm": 0.4909393054849047,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 9419
    },
    {
      "epoch": 0.0942,
      "grad_norm": 0.4572606419258861,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 9420
    },
    {
      "epoch": 0.09421,
      "grad_norm": 0.5236689638625828,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 9421
    },
    {
      "epoch": 0.09422,
      "grad_norm": 0.663258154130139,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 9422
    },
    {
      "epoch": 0.09423,
      "grad_norm": 0.928899803447629,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 9423
    },
    {
      "epoch": 0.09424,
      "grad_norm": 1.1926656295302425,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 9424
    },
    {
      "epoch": 0.09425,
      "grad_norm": 0.6853441865571991,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 9425
    },
    {
      "epoch": 0.09426,
      "grad_norm": 0.5128317339031602,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 9426
    },
    {
      "epoch": 0.09427,
      "grad_norm": 0.7070587189992616,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 9427
    },
    {
      "epoch": 0.09428,
      "grad_norm": 0.773759955306798,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 9428
    },
    {
      "epoch": 0.09429,
      "grad_norm": 0.811718324620787,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 9429
    },
    {
      "epoch": 0.0943,
      "grad_norm": 0.7612959659265646,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 9430
    },
    {
      "epoch": 0.09431,
      "grad_norm": 0.705635244969301,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 9431
    },
    {
      "epoch": 0.09432,
      "grad_norm": 0.7185246612672866,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 9432
    },
    {
      "epoch": 0.09433,
      "grad_norm": 0.808636412731162,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9433
    },
    {
      "epoch": 0.09434,
      "grad_norm": 0.8588858555452056,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 9434
    },
    {
      "epoch": 0.09435,
      "grad_norm": 0.8829432922785366,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 9435
    },
    {
      "epoch": 0.09436,
      "grad_norm": 0.8229332972229458,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9436
    },
    {
      "epoch": 0.09437,
      "grad_norm": 0.7502237223787239,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 9437
    },
    {
      "epoch": 0.09438,
      "grad_norm": 0.6894340570648702,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 9438
    },
    {
      "epoch": 0.09439,
      "grad_norm": 0.7210313125158538,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 9439
    },
    {
      "epoch": 0.0944,
      "grad_norm": 0.7260397120709652,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 9440
    },
    {
      "epoch": 0.09441,
      "grad_norm": 0.7867602084595864,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 9441
    },
    {
      "epoch": 0.09442,
      "grad_norm": 0.8364820552569051,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 9442
    },
    {
      "epoch": 0.09443,
      "grad_norm": 0.9738365901239612,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 9443
    },
    {
      "epoch": 0.09444,
      "grad_norm": 1.162136240204578,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 9444
    },
    {
      "epoch": 0.09445,
      "grad_norm": 0.8068568465517928,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 9445
    },
    {
      "epoch": 0.09446,
      "grad_norm": 0.7524396274055184,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 9446
    },
    {
      "epoch": 0.09447,
      "grad_norm": 0.6473920221674216,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 9447
    },
    {
      "epoch": 0.09448,
      "grad_norm": 0.6745944542419227,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 9448
    },
    {
      "epoch": 0.09449,
      "grad_norm": 0.6655279612546542,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 9449
    },
    {
      "epoch": 0.0945,
      "grad_norm": 0.6127094330682528,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 9450
    },
    {
      "epoch": 0.09451,
      "grad_norm": 0.621361177849897,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9451
    },
    {
      "epoch": 0.09452,
      "grad_norm": 0.6362015223505509,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 9452
    },
    {
      "epoch": 0.09453,
      "grad_norm": 0.813045169964411,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 9453
    },
    {
      "epoch": 0.09454,
      "grad_norm": 1.1347224007728371,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 9454
    },
    {
      "epoch": 0.09455,
      "grad_norm": 0.8959042657473294,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9455
    },
    {
      "epoch": 0.09456,
      "grad_norm": 0.7685184606728515,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 9456
    },
    {
      "epoch": 0.09457,
      "grad_norm": 0.7342695223426569,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 9457
    },
    {
      "epoch": 0.09458,
      "grad_norm": 0.8272094526351071,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9458
    },
    {
      "epoch": 0.09459,
      "grad_norm": 0.8913799983064666,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 9459
    },
    {
      "epoch": 0.0946,
      "grad_norm": 0.7914931766921125,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 9460
    },
    {
      "epoch": 0.09461,
      "grad_norm": 0.7143667270257789,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 9461
    },
    {
      "epoch": 0.09462,
      "grad_norm": 0.7155076180457024,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 9462
    },
    {
      "epoch": 0.09463,
      "grad_norm": 0.689792898507005,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 9463
    },
    {
      "epoch": 0.09464,
      "grad_norm": 0.7023738600071371,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 9464
    },
    {
      "epoch": 0.09465,
      "grad_norm": 0.7187786079383829,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 9465
    },
    {
      "epoch": 0.09466,
      "grad_norm": 0.6302140939859505,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 9466
    },
    {
      "epoch": 0.09467,
      "grad_norm": 0.612265626337836,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 9467
    },
    {
      "epoch": 0.09468,
      "grad_norm": 0.6579286983017547,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 9468
    },
    {
      "epoch": 0.09469,
      "grad_norm": 0.5830435072856022,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 9469
    },
    {
      "epoch": 0.0947,
      "grad_norm": 0.5258507863729508,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 9470
    },
    {
      "epoch": 0.09471,
      "grad_norm": 0.46919009653098687,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 9471
    },
    {
      "epoch": 0.09472,
      "grad_norm": 0.5215402592850111,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 9472
    },
    {
      "epoch": 0.09473,
      "grad_norm": 0.6491591524568011,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 9473
    },
    {
      "epoch": 0.09474,
      "grad_norm": 0.7868164547787738,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 9474
    },
    {
      "epoch": 0.09475,
      "grad_norm": 0.9666742348116686,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 9475
    },
    {
      "epoch": 0.09476,
      "grad_norm": 1.0668974359449654,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 9476
    },
    {
      "epoch": 0.09477,
      "grad_norm": 0.8895495837200137,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 9477
    },
    {
      "epoch": 0.09478,
      "grad_norm": 0.7822408243265004,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 9478
    },
    {
      "epoch": 0.09479,
      "grad_norm": 0.7362350598274618,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 9479
    },
    {
      "epoch": 0.0948,
      "grad_norm": 0.7635551325432101,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 9480
    },
    {
      "epoch": 0.09481,
      "grad_norm": 0.8359593042143941,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 9481
    },
    {
      "epoch": 0.09482,
      "grad_norm": 0.8458302540004683,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 9482
    },
    {
      "epoch": 0.09483,
      "grad_norm": 0.8340671653924622,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 9483
    },
    {
      "epoch": 0.09484,
      "grad_norm": 0.7992408585225695,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 9484
    },
    {
      "epoch": 0.09485,
      "grad_norm": 0.7964148073286885,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 9485
    },
    {
      "epoch": 0.09486,
      "grad_norm": 0.8771624728266194,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 9486
    },
    {
      "epoch": 0.09487,
      "grad_norm": 0.9914369833754603,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 9487
    },
    {
      "epoch": 0.09488,
      "grad_norm": 0.92799842755361,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 9488
    },
    {
      "epoch": 0.09489,
      "grad_norm": 0.7239696176957987,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 9489
    },
    {
      "epoch": 0.0949,
      "grad_norm": 0.657621056297735,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 9490
    },
    {
      "epoch": 0.09491,
      "grad_norm": 0.5920365056277531,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 9491
    },
    {
      "epoch": 0.09492,
      "grad_norm": 0.6824599764976818,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 9492
    },
    {
      "epoch": 0.09493,
      "grad_norm": 0.8856580120578575,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 9493
    },
    {
      "epoch": 0.09494,
      "grad_norm": 0.9809438136763652,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 9494
    },
    {
      "epoch": 0.09495,
      "grad_norm": 1.0818725077217133,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 9495
    },
    {
      "epoch": 0.09496,
      "grad_norm": 0.8173342329103473,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 9496
    },
    {
      "epoch": 0.09497,
      "grad_norm": 0.6206180769695743,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 9497
    },
    {
      "epoch": 0.09498,
      "grad_norm": 0.5589853203174866,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 9498
    },
    {
      "epoch": 0.09499,
      "grad_norm": 0.5738351051872796,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 9499
    },
    {
      "epoch": 0.095,
      "grad_norm": 0.6306862899639631,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 9500
    },
    {
      "epoch": 0.09501,
      "grad_norm": 0.6733724428518848,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 9501
    },
    {
      "epoch": 0.09502,
      "grad_norm": 0.8461241073955719,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 9502
    },
    {
      "epoch": 0.09503,
      "grad_norm": 1.0158067722960233,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 9503
    },
    {
      "epoch": 0.09504,
      "grad_norm": 1.1190865130171455,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 9504
    },
    {
      "epoch": 0.09505,
      "grad_norm": 0.7932088206811884,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 9505
    },
    {
      "epoch": 0.09506,
      "grad_norm": 0.7129006027867597,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 9506
    },
    {
      "epoch": 0.09507,
      "grad_norm": 0.6024715566098497,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9507
    },
    {
      "epoch": 0.09508,
      "grad_norm": 0.6057657514430907,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 9508
    },
    {
      "epoch": 0.09509,
      "grad_norm": 0.6243112506876197,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 9509
    },
    {
      "epoch": 0.0951,
      "grad_norm": 0.6744699873109112,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 9510
    },
    {
      "epoch": 0.09511,
      "grad_norm": 0.5698773213648972,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 9511
    },
    {
      "epoch": 0.09512,
      "grad_norm": 0.5223468938440529,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9512
    },
    {
      "epoch": 0.09513,
      "grad_norm": 0.5295241801515409,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 9513
    },
    {
      "epoch": 0.09514,
      "grad_norm": 0.6195032274853618,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 9514
    },
    {
      "epoch": 0.09515,
      "grad_norm": 0.7923024129571904,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 9515
    },
    {
      "epoch": 0.09516,
      "grad_norm": 1.0748459000652908,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 9516
    },
    {
      "epoch": 0.09517,
      "grad_norm": 0.9652150547579573,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 9517
    },
    {
      "epoch": 0.09518,
      "grad_norm": 0.9317902587041311,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 9518
    },
    {
      "epoch": 0.09519,
      "grad_norm": 1.0011920568954655,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 9519
    },
    {
      "epoch": 0.0952,
      "grad_norm": 0.8162413257986282,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 9520
    },
    {
      "epoch": 0.09521,
      "grad_norm": 0.8511270427905199,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 9521
    },
    {
      "epoch": 0.09522,
      "grad_norm": 0.869150690242155,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 9522
    },
    {
      "epoch": 0.09523,
      "grad_norm": 0.7851073631966403,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 9523
    },
    {
      "epoch": 0.09524,
      "grad_norm": 0.7805600991791303,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 9524
    },
    {
      "epoch": 0.09525,
      "grad_norm": 0.6704373220450912,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 9525
    },
    {
      "epoch": 0.09526,
      "grad_norm": 0.6560833308141075,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 9526
    },
    {
      "epoch": 0.09527,
      "grad_norm": 0.6885732713810039,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 9527
    },
    {
      "epoch": 0.09528,
      "grad_norm": 0.9403396043081385,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 9528
    },
    {
      "epoch": 0.09529,
      "grad_norm": 1.00556887324088,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9529
    },
    {
      "epoch": 0.0953,
      "grad_norm": 0.888729666716305,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 9530
    },
    {
      "epoch": 0.09531,
      "grad_norm": 0.9418145584346838,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 9531
    },
    {
      "epoch": 0.09532,
      "grad_norm": 0.9872093380526542,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 9532
    },
    {
      "epoch": 0.09533,
      "grad_norm": 1.0096517466771673,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 9533
    },
    {
      "epoch": 0.09534,
      "grad_norm": 0.8896105143964483,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 9534
    },
    {
      "epoch": 0.09535,
      "grad_norm": 0.7675771159928804,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 9535
    },
    {
      "epoch": 0.09536,
      "grad_norm": 0.7682126422834153,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 9536
    },
    {
      "epoch": 0.09537,
      "grad_norm": 0.8056030277957967,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 9537
    },
    {
      "epoch": 0.09538,
      "grad_norm": 0.7162391293635292,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 9538
    },
    {
      "epoch": 0.09539,
      "grad_norm": 0.6683543592127842,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 9539
    },
    {
      "epoch": 0.0954,
      "grad_norm": 0.6001385230297466,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 9540
    },
    {
      "epoch": 0.09541,
      "grad_norm": 0.5520189491403364,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 9541
    },
    {
      "epoch": 0.09542,
      "grad_norm": 0.5771983745529633,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 9542
    },
    {
      "epoch": 0.09543,
      "grad_norm": 0.6070661140119326,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 9543
    },
    {
      "epoch": 0.09544,
      "grad_norm": 0.7384419185478289,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 9544
    },
    {
      "epoch": 0.09545,
      "grad_norm": 0.9657170883830423,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 9545
    },
    {
      "epoch": 0.09546,
      "grad_norm": 1.2681324157872516,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 9546
    },
    {
      "epoch": 0.09547,
      "grad_norm": 0.62049585194807,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 9547
    },
    {
      "epoch": 0.09548,
      "grad_norm": 0.795505396852106,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 9548
    },
    {
      "epoch": 0.09549,
      "grad_norm": 0.9041023642688241,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 9549
    },
    {
      "epoch": 0.0955,
      "grad_norm": 0.7920466015961556,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 9550
    },
    {
      "epoch": 0.09551,
      "grad_norm": 0.7664489290961656,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 9551
    },
    {
      "epoch": 0.09552,
      "grad_norm": 0.7297971620268919,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 9552
    },
    {
      "epoch": 0.09553,
      "grad_norm": 0.7231956245433028,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 9553
    },
    {
      "epoch": 0.09554,
      "grad_norm": 0.7096741017462413,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 9554
    },
    {
      "epoch": 0.09555,
      "grad_norm": 0.7188180696481433,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 9555
    },
    {
      "epoch": 0.09556,
      "grad_norm": 0.700959357888011,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 9556
    },
    {
      "epoch": 0.09557,
      "grad_norm": 0.6879426371558562,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 9557
    },
    {
      "epoch": 0.09558,
      "grad_norm": 0.6597031957541576,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 9558
    },
    {
      "epoch": 0.09559,
      "grad_norm": 0.6739137082114623,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 9559
    },
    {
      "epoch": 0.0956,
      "grad_norm": 0.6964119348231613,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 9560
    },
    {
      "epoch": 0.09561,
      "grad_norm": 0.7031263209660419,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 9561
    },
    {
      "epoch": 0.09562,
      "grad_norm": 0.9420485552641903,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 9562
    },
    {
      "epoch": 0.09563,
      "grad_norm": 6.130969416922354,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 9563
    },
    {
      "epoch": 0.09564,
      "grad_norm": 1.309865365129671,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 9564
    },
    {
      "epoch": 0.09565,
      "grad_norm": 1.0521711710301151,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 9565
    },
    {
      "epoch": 0.09566,
      "grad_norm": 0.9997826713731996,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 9566
    },
    {
      "epoch": 0.09567,
      "grad_norm": 1.3487618062811157,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 9567
    },
    {
      "epoch": 0.09568,
      "grad_norm": 0.9741842634354936,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 9568
    },
    {
      "epoch": 0.09569,
      "grad_norm": 1.0654371346008946,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 9569
    },
    {
      "epoch": 0.0957,
      "grad_norm": 0.8712382456617402,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 9570
    },
    {
      "epoch": 0.09571,
      "grad_norm": 0.8849374832653227,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 9571
    },
    {
      "epoch": 0.09572,
      "grad_norm": 1.0577800121636243,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 9572
    },
    {
      "epoch": 0.09573,
      "grad_norm": 1.206598624557028,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 9573
    },
    {
      "epoch": 0.09574,
      "grad_norm": 1.2512634504121267,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 9574
    },
    {
      "epoch": 0.09575,
      "grad_norm": 0.9501656930930779,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 9575
    },
    {
      "epoch": 0.09576,
      "grad_norm": 1.1728879533932677,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 9576
    },
    {
      "epoch": 0.09577,
      "grad_norm": 2.0768187878146875,
      "learning_rate": 0.003,
      "loss": 4.184,
      "step": 9577
    },
    {
      "epoch": 0.09578,
      "grad_norm": 1.105305608187662,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 9578
    },
    {
      "epoch": 0.09579,
      "grad_norm": 0.9689648072779388,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 9579
    },
    {
      "epoch": 0.0958,
      "grad_norm": 0.9360186586846221,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 9580
    },
    {
      "epoch": 0.09581,
      "grad_norm": 1.0800383998211391,
      "learning_rate": 0.003,
      "loss": 4.2082,
      "step": 9581
    },
    {
      "epoch": 0.09582,
      "grad_norm": 1.0704867786853698,
      "learning_rate": 0.003,
      "loss": 4.1485,
      "step": 9582
    },
    {
      "epoch": 0.09583,
      "grad_norm": 1.142877500779312,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 9583
    },
    {
      "epoch": 0.09584,
      "grad_norm": 1.3332645879072875,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 9584
    },
    {
      "epoch": 0.09585,
      "grad_norm": 1.5324381756718481,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 9585
    },
    {
      "epoch": 0.09586,
      "grad_norm": 0.8704649822421158,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 9586
    },
    {
      "epoch": 0.09587,
      "grad_norm": 0.9688340079251864,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 9587
    },
    {
      "epoch": 0.09588,
      "grad_norm": 0.8342771328110887,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 9588
    },
    {
      "epoch": 0.09589,
      "grad_norm": 0.7942887994349634,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 9589
    },
    {
      "epoch": 0.0959,
      "grad_norm": 0.8173041322090617,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 9590
    },
    {
      "epoch": 0.09591,
      "grad_norm": 0.7677670828705769,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 9591
    },
    {
      "epoch": 0.09592,
      "grad_norm": 0.7200915670319258,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 9592
    },
    {
      "epoch": 0.09593,
      "grad_norm": 0.7972359349347333,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 9593
    },
    {
      "epoch": 0.09594,
      "grad_norm": 0.8556890271092606,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 9594
    },
    {
      "epoch": 0.09595,
      "grad_norm": 0.8915728396380502,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 9595
    },
    {
      "epoch": 0.09596,
      "grad_norm": 0.9347507452393141,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 9596
    },
    {
      "epoch": 0.09597,
      "grad_norm": 1.0024508636239098,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 9597
    },
    {
      "epoch": 0.09598,
      "grad_norm": 1.2169763273515621,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 9598
    },
    {
      "epoch": 0.09599,
      "grad_norm": 0.7115963030817509,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 9599
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.6328045403895035,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 9600
    },
    {
      "epoch": 0.09601,
      "grad_norm": 0.5998687509571087,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 9601
    },
    {
      "epoch": 0.09602,
      "grad_norm": 0.5466887961482548,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 9602
    },
    {
      "epoch": 0.09603,
      "grad_norm": 0.7860448470509064,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 9603
    },
    {
      "epoch": 0.09604,
      "grad_norm": 0.8745624847605153,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 9604
    },
    {
      "epoch": 0.09605,
      "grad_norm": 0.8688163043711737,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 9605
    },
    {
      "epoch": 0.09606,
      "grad_norm": 0.7963417759657033,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 9606
    },
    {
      "epoch": 0.09607,
      "grad_norm": 0.7369028440038995,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 9607
    },
    {
      "epoch": 0.09608,
      "grad_norm": 0.8647736584876029,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 9608
    },
    {
      "epoch": 0.09609,
      "grad_norm": 1.0027577187055126,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 9609
    },
    {
      "epoch": 0.0961,
      "grad_norm": 0.8626178763464114,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 9610
    },
    {
      "epoch": 0.09611,
      "grad_norm": 0.5910337185914825,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 9611
    },
    {
      "epoch": 0.09612,
      "grad_norm": 0.49438202417856203,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 9612
    },
    {
      "epoch": 0.09613,
      "grad_norm": 0.484763634619797,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 9613
    },
    {
      "epoch": 0.09614,
      "grad_norm": 0.44675304497501767,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 9614
    },
    {
      "epoch": 0.09615,
      "grad_norm": 0.3830623042239131,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 9615
    },
    {
      "epoch": 0.09616,
      "grad_norm": 0.40447953537590603,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 9616
    },
    {
      "epoch": 0.09617,
      "grad_norm": 0.42749479774334626,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 9617
    },
    {
      "epoch": 0.09618,
      "grad_norm": 0.3807451005616551,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 9618
    },
    {
      "epoch": 0.09619,
      "grad_norm": 0.3715423418211784,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 9619
    },
    {
      "epoch": 0.0962,
      "grad_norm": 0.3425603111991135,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 9620
    },
    {
      "epoch": 0.09621,
      "grad_norm": 0.3507562163489838,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 9621
    },
    {
      "epoch": 0.09622,
      "grad_norm": 0.29748679852114185,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 9622
    },
    {
      "epoch": 0.09623,
      "grad_norm": 0.2839394173155373,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 9623
    },
    {
      "epoch": 0.09624,
      "grad_norm": 0.2794661039316224,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 9624
    },
    {
      "epoch": 0.09625,
      "grad_norm": 0.29629458143975185,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 9625
    },
    {
      "epoch": 0.09626,
      "grad_norm": 0.3292412126393909,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 9626
    },
    {
      "epoch": 0.09627,
      "grad_norm": 0.4497439282959039,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 9627
    },
    {
      "epoch": 0.09628,
      "grad_norm": 0.6526850258400877,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 9628
    },
    {
      "epoch": 0.09629,
      "grad_norm": 0.9634340689907828,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 9629
    },
    {
      "epoch": 0.0963,
      "grad_norm": 1.1516478385031867,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 9630
    },
    {
      "epoch": 0.09631,
      "grad_norm": 0.6401093793721576,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 9631
    },
    {
      "epoch": 0.09632,
      "grad_norm": 0.7636314345691994,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 9632
    },
    {
      "epoch": 0.09633,
      "grad_norm": 0.992571157461324,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 9633
    },
    {
      "epoch": 0.09634,
      "grad_norm": 0.7463855736399202,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 9634
    },
    {
      "epoch": 0.09635,
      "grad_norm": 0.6909299649125988,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 9635
    },
    {
      "epoch": 0.09636,
      "grad_norm": 0.7431003319653801,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 9636
    },
    {
      "epoch": 0.09637,
      "grad_norm": 0.6455512496885061,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 9637
    },
    {
      "epoch": 0.09638,
      "grad_norm": 0.5445783213282376,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 9638
    },
    {
      "epoch": 0.09639,
      "grad_norm": 0.5613727528028151,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9639
    },
    {
      "epoch": 0.0964,
      "grad_norm": 0.7244593336080651,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 9640
    },
    {
      "epoch": 0.09641,
      "grad_norm": 0.9030660564578993,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 9641
    },
    {
      "epoch": 0.09642,
      "grad_norm": 0.9887019736303208,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 9642
    },
    {
      "epoch": 0.09643,
      "grad_norm": 0.8838383160348036,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 9643
    },
    {
      "epoch": 0.09644,
      "grad_norm": 0.7515245578827098,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 9644
    },
    {
      "epoch": 0.09645,
      "grad_norm": 0.6649872915216315,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 9645
    },
    {
      "epoch": 0.09646,
      "grad_norm": 0.7116890218147718,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 9646
    },
    {
      "epoch": 0.09647,
      "grad_norm": 0.7019694940498846,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 9647
    },
    {
      "epoch": 0.09648,
      "grad_norm": 0.6596706388975605,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 9648
    },
    {
      "epoch": 0.09649,
      "grad_norm": 0.667100870547891,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 9649
    },
    {
      "epoch": 0.0965,
      "grad_norm": 0.6563616277238821,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 9650
    },
    {
      "epoch": 0.09651,
      "grad_norm": 0.6207032309420466,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 9651
    },
    {
      "epoch": 0.09652,
      "grad_norm": 0.6164765321685511,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 9652
    },
    {
      "epoch": 0.09653,
      "grad_norm": 0.6425858801124605,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 9653
    },
    {
      "epoch": 0.09654,
      "grad_norm": 0.6719434091175973,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 9654
    },
    {
      "epoch": 0.09655,
      "grad_norm": 0.6816595879015243,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 9655
    },
    {
      "epoch": 0.09656,
      "grad_norm": 0.75276222897346,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 9656
    },
    {
      "epoch": 0.09657,
      "grad_norm": 0.793810434998781,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 9657
    },
    {
      "epoch": 0.09658,
      "grad_norm": 0.8203985544035876,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 9658
    },
    {
      "epoch": 0.09659,
      "grad_norm": 0.7228109813703856,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 9659
    },
    {
      "epoch": 0.0966,
      "grad_norm": 0.6284036142391546,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 9660
    },
    {
      "epoch": 0.09661,
      "grad_norm": 0.5390823749375145,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 9661
    },
    {
      "epoch": 0.09662,
      "grad_norm": 0.4937468666738386,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 9662
    },
    {
      "epoch": 0.09663,
      "grad_norm": 0.5267363875644104,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 9663
    },
    {
      "epoch": 0.09664,
      "grad_norm": 0.5614869173293953,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 9664
    },
    {
      "epoch": 0.09665,
      "grad_norm": 0.6434988568327966,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 9665
    },
    {
      "epoch": 0.09666,
      "grad_norm": 0.6863099786692454,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 9666
    },
    {
      "epoch": 0.09667,
      "grad_norm": 0.6974192050812339,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 9667
    },
    {
      "epoch": 0.09668,
      "grad_norm": 0.7130895485506055,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 9668
    },
    {
      "epoch": 0.09669,
      "grad_norm": 0.7314830324385385,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 9669
    },
    {
      "epoch": 0.0967,
      "grad_norm": 0.7358599798360544,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 9670
    },
    {
      "epoch": 0.09671,
      "grad_norm": 0.6821836408473856,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 9671
    },
    {
      "epoch": 0.09672,
      "grad_norm": 0.6652927918131237,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 9672
    },
    {
      "epoch": 0.09673,
      "grad_norm": 0.7634068597168817,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 9673
    },
    {
      "epoch": 0.09674,
      "grad_norm": 0.7974126591485172,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 9674
    },
    {
      "epoch": 0.09675,
      "grad_norm": 0.8544438788166088,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 9675
    },
    {
      "epoch": 0.09676,
      "grad_norm": 0.7941264785151647,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9676
    },
    {
      "epoch": 0.09677,
      "grad_norm": 0.7203225810853267,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 9677
    },
    {
      "epoch": 0.09678,
      "grad_norm": 0.5740375147446136,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 9678
    },
    {
      "epoch": 0.09679,
      "grad_norm": 0.60439656922455,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 9679
    },
    {
      "epoch": 0.0968,
      "grad_norm": 0.7001578420854571,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 9680
    },
    {
      "epoch": 0.09681,
      "grad_norm": 0.8530019659681068,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 9681
    },
    {
      "epoch": 0.09682,
      "grad_norm": 1.0076727467524107,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 9682
    },
    {
      "epoch": 0.09683,
      "grad_norm": 1.1399405747494864,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 9683
    },
    {
      "epoch": 0.09684,
      "grad_norm": 0.6866617378578579,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 9684
    },
    {
      "epoch": 0.09685,
      "grad_norm": 0.6034395427034922,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 9685
    },
    {
      "epoch": 0.09686,
      "grad_norm": 0.6858126543232959,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 9686
    },
    {
      "epoch": 0.09687,
      "grad_norm": 0.7606273901161237,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 9687
    },
    {
      "epoch": 0.09688,
      "grad_norm": 0.7823206051561976,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 9688
    },
    {
      "epoch": 0.09689,
      "grad_norm": 0.7275273786792908,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 9689
    },
    {
      "epoch": 0.0969,
      "grad_norm": 0.7020173666361166,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 9690
    },
    {
      "epoch": 0.09691,
      "grad_norm": 0.668700670688743,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 9691
    },
    {
      "epoch": 0.09692,
      "grad_norm": 0.7183283334951155,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 9692
    },
    {
      "epoch": 0.09693,
      "grad_norm": 0.8228261565164465,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 9693
    },
    {
      "epoch": 0.09694,
      "grad_norm": 0.9398860669380985,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 9694
    },
    {
      "epoch": 0.09695,
      "grad_norm": 0.9679499359904971,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 9695
    },
    {
      "epoch": 0.09696,
      "grad_norm": 0.9831277733673565,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 9696
    },
    {
      "epoch": 0.09697,
      "grad_norm": 0.8922294143645066,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9697
    },
    {
      "epoch": 0.09698,
      "grad_norm": 0.8622941183892853,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 9698
    },
    {
      "epoch": 0.09699,
      "grad_norm": 0.849954807407499,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 9699
    },
    {
      "epoch": 0.097,
      "grad_norm": 0.8595604678823553,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 9700
    },
    {
      "epoch": 0.09701,
      "grad_norm": 0.7378223203944358,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 9701
    },
    {
      "epoch": 0.09702,
      "grad_norm": 0.6131607408869036,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 9702
    },
    {
      "epoch": 0.09703,
      "grad_norm": 0.5814032117502681,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 9703
    },
    {
      "epoch": 0.09704,
      "grad_norm": 0.6207822778251723,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 9704
    },
    {
      "epoch": 0.09705,
      "grad_norm": 0.5424790257052854,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 9705
    },
    {
      "epoch": 0.09706,
      "grad_norm": 0.5191042976935801,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 9706
    },
    {
      "epoch": 0.09707,
      "grad_norm": 0.4825162871461943,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 9707
    },
    {
      "epoch": 0.09708,
      "grad_norm": 0.5254613203667191,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 9708
    },
    {
      "epoch": 0.09709,
      "grad_norm": 0.5837214908960893,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 9709
    },
    {
      "epoch": 0.0971,
      "grad_norm": 0.5919660600745414,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 9710
    },
    {
      "epoch": 0.09711,
      "grad_norm": 0.5985906886004138,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 9711
    },
    {
      "epoch": 0.09712,
      "grad_norm": 0.6218596201649731,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 9712
    },
    {
      "epoch": 0.09713,
      "grad_norm": 0.6609701796902929,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 9713
    },
    {
      "epoch": 0.09714,
      "grad_norm": 0.5993265445219289,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 9714
    },
    {
      "epoch": 0.09715,
      "grad_norm": 0.5595560077596979,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 9715
    },
    {
      "epoch": 0.09716,
      "grad_norm": 0.5458771794387989,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 9716
    },
    {
      "epoch": 0.09717,
      "grad_norm": 0.5749644308670316,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 9717
    },
    {
      "epoch": 0.09718,
      "grad_norm": 0.7136367880167833,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 9718
    },
    {
      "epoch": 0.09719,
      "grad_norm": 0.8710577217664088,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 9719
    },
    {
      "epoch": 0.0972,
      "grad_norm": 1.1290953541399391,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 9720
    },
    {
      "epoch": 0.09721,
      "grad_norm": 0.9640405697464676,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 9721
    },
    {
      "epoch": 0.09722,
      "grad_norm": 0.947033121500074,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 9722
    },
    {
      "epoch": 0.09723,
      "grad_norm": 0.864058303867878,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 9723
    },
    {
      "epoch": 0.09724,
      "grad_norm": 0.7910749200085015,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 9724
    },
    {
      "epoch": 0.09725,
      "grad_norm": 0.6706273678022362,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 9725
    },
    {
      "epoch": 0.09726,
      "grad_norm": 0.6952803945114413,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 9726
    },
    {
      "epoch": 0.09727,
      "grad_norm": 0.7730372856532575,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 9727
    },
    {
      "epoch": 0.09728,
      "grad_norm": 0.6776321846761416,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 9728
    },
    {
      "epoch": 0.09729,
      "grad_norm": 0.656555599450724,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 9729
    },
    {
      "epoch": 0.0973,
      "grad_norm": 0.7397413425845409,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 9730
    },
    {
      "epoch": 0.09731,
      "grad_norm": 0.741668722027564,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 9731
    },
    {
      "epoch": 0.09732,
      "grad_norm": 0.9192765996179019,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 9732
    },
    {
      "epoch": 0.09733,
      "grad_norm": 0.9726160313328494,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 9733
    },
    {
      "epoch": 0.09734,
      "grad_norm": 0.8089628303200062,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 9734
    },
    {
      "epoch": 0.09735,
      "grad_norm": 0.9196854053701035,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 9735
    },
    {
      "epoch": 0.09736,
      "grad_norm": 1.0355781723270967,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 9736
    },
    {
      "epoch": 0.09737,
      "grad_norm": 1.390013473224088,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 9737
    },
    {
      "epoch": 0.09738,
      "grad_norm": 0.6677058435549194,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 9738
    },
    {
      "epoch": 0.09739,
      "grad_norm": 0.7149231696043877,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 9739
    },
    {
      "epoch": 0.0974,
      "grad_norm": 0.7596898377209321,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 9740
    },
    {
      "epoch": 0.09741,
      "grad_norm": 0.9445678693533357,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 9741
    },
    {
      "epoch": 0.09742,
      "grad_norm": 1.144761757205563,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 9742
    },
    {
      "epoch": 0.09743,
      "grad_norm": 0.7798904071068536,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 9743
    },
    {
      "epoch": 0.09744,
      "grad_norm": 0.7126950822384531,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 9744
    },
    {
      "epoch": 0.09745,
      "grad_norm": 0.804932675736049,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 9745
    },
    {
      "epoch": 0.09746,
      "grad_norm": 0.8275041917630538,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 9746
    },
    {
      "epoch": 0.09747,
      "grad_norm": 0.8654165216744595,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 9747
    },
    {
      "epoch": 0.09748,
      "grad_norm": 0.8483082244535185,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 9748
    },
    {
      "epoch": 0.09749,
      "grad_norm": 0.9655796883096281,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 9749
    },
    {
      "epoch": 0.0975,
      "grad_norm": 0.8375150794577596,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 9750
    },
    {
      "epoch": 0.09751,
      "grad_norm": 0.8357319865847818,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 9751
    },
    {
      "epoch": 0.09752,
      "grad_norm": 0.7805475045931051,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 9752
    },
    {
      "epoch": 0.09753,
      "grad_norm": 0.7014801049003869,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 9753
    },
    {
      "epoch": 0.09754,
      "grad_norm": 0.6856528434863608,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 9754
    },
    {
      "epoch": 0.09755,
      "grad_norm": 0.650848045634542,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 9755
    },
    {
      "epoch": 0.09756,
      "grad_norm": 0.5622036395323345,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 9756
    },
    {
      "epoch": 0.09757,
      "grad_norm": 0.5679917813675583,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 9757
    },
    {
      "epoch": 0.09758,
      "grad_norm": 0.5576435578489023,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 9758
    },
    {
      "epoch": 0.09759,
      "grad_norm": 0.6963321867164363,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 9759
    },
    {
      "epoch": 0.0976,
      "grad_norm": 0.7838079745448902,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 9760
    },
    {
      "epoch": 0.09761,
      "grad_norm": 0.9541884510786587,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 9761
    },
    {
      "epoch": 0.09762,
      "grad_norm": 1.121153313260097,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 9762
    },
    {
      "epoch": 0.09763,
      "grad_norm": 0.6664939878081045,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 9763
    },
    {
      "epoch": 0.09764,
      "grad_norm": 0.6067299530309102,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 9764
    },
    {
      "epoch": 0.09765,
      "grad_norm": 0.6916937760369664,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 9765
    },
    {
      "epoch": 0.09766,
      "grad_norm": 0.7187195565497518,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 9766
    },
    {
      "epoch": 0.09767,
      "grad_norm": 0.7297554741322773,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 9767
    },
    {
      "epoch": 0.09768,
      "grad_norm": 0.7197047154946628,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 9768
    },
    {
      "epoch": 0.09769,
      "grad_norm": 0.77558338050166,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 9769
    },
    {
      "epoch": 0.0977,
      "grad_norm": 0.776706503347036,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 9770
    },
    {
      "epoch": 0.09771,
      "grad_norm": 0.6629745622875617,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 9771
    },
    {
      "epoch": 0.09772,
      "grad_norm": 0.599070691762346,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 9772
    },
    {
      "epoch": 0.09773,
      "grad_norm": 0.6430974616537984,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 9773
    },
    {
      "epoch": 0.09774,
      "grad_norm": 0.65829695633147,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 9774
    },
    {
      "epoch": 0.09775,
      "grad_norm": 0.6436886424967752,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 9775
    },
    {
      "epoch": 0.09776,
      "grad_norm": 0.7272224412585245,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 9776
    },
    {
      "epoch": 0.09777,
      "grad_norm": 0.7810004465966248,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 9777
    },
    {
      "epoch": 0.09778,
      "grad_norm": 0.843156271360344,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 9778
    },
    {
      "epoch": 0.09779,
      "grad_norm": 0.8542309059649245,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 9779
    },
    {
      "epoch": 0.0978,
      "grad_norm": 0.8284073938545176,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 9780
    },
    {
      "epoch": 0.09781,
      "grad_norm": 0.7656977589639479,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 9781
    },
    {
      "epoch": 0.09782,
      "grad_norm": 0.642867591138632,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 9782
    },
    {
      "epoch": 0.09783,
      "grad_norm": 0.6315054401796648,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 9783
    },
    {
      "epoch": 0.09784,
      "grad_norm": 0.6506631017065446,
      "learning_rate": 0.003,
      "loss": 4.0418,
      "step": 9784
    },
    {
      "epoch": 0.09785,
      "grad_norm": 0.6870713649494674,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 9785
    },
    {
      "epoch": 0.09786,
      "grad_norm": 0.7447419206648687,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9786
    },
    {
      "epoch": 0.09787,
      "grad_norm": 0.8992521452563863,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 9787
    },
    {
      "epoch": 0.09788,
      "grad_norm": 0.8733619920631194,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 9788
    },
    {
      "epoch": 0.09789,
      "grad_norm": 0.829358760588995,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 9789
    },
    {
      "epoch": 0.0979,
      "grad_norm": 0.8166018992244876,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 9790
    },
    {
      "epoch": 0.09791,
      "grad_norm": 0.8700145284441482,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 9791
    },
    {
      "epoch": 0.09792,
      "grad_norm": 0.9672422465296193,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9792
    },
    {
      "epoch": 0.09793,
      "grad_norm": 0.8861730905617189,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 9793
    },
    {
      "epoch": 0.09794,
      "grad_norm": 0.8248882883309061,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 9794
    },
    {
      "epoch": 0.09795,
      "grad_norm": 0.7088842143033923,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 9795
    },
    {
      "epoch": 0.09796,
      "grad_norm": 0.7130954093466139,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 9796
    },
    {
      "epoch": 0.09797,
      "grad_norm": 0.7288682239466572,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 9797
    },
    {
      "epoch": 0.09798,
      "grad_norm": 0.7976440093808327,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 9798
    },
    {
      "epoch": 0.09799,
      "grad_norm": 0.9649496398862366,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 9799
    },
    {
      "epoch": 0.098,
      "grad_norm": 1.1331720133396581,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 9800
    },
    {
      "epoch": 0.09801,
      "grad_norm": 0.7034672135712403,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 9801
    },
    {
      "epoch": 0.09802,
      "grad_norm": 0.6773839521714846,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 9802
    },
    {
      "epoch": 0.09803,
      "grad_norm": 0.7355150714696657,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 9803
    },
    {
      "epoch": 0.09804,
      "grad_norm": 0.7174536168973166,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 9804
    },
    {
      "epoch": 0.09805,
      "grad_norm": 0.6507571892133279,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 9805
    },
    {
      "epoch": 0.09806,
      "grad_norm": 0.64602677559084,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 9806
    },
    {
      "epoch": 0.09807,
      "grad_norm": 0.7072086988189831,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 9807
    },
    {
      "epoch": 0.09808,
      "grad_norm": 0.6694715352849688,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 9808
    },
    {
      "epoch": 0.09809,
      "grad_norm": 0.641647359947315,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 9809
    },
    {
      "epoch": 0.0981,
      "grad_norm": 0.6231401766200729,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 9810
    },
    {
      "epoch": 0.09811,
      "grad_norm": 0.6349700167139162,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 9811
    },
    {
      "epoch": 0.09812,
      "grad_norm": 0.6948447910573521,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 9812
    },
    {
      "epoch": 0.09813,
      "grad_norm": 0.8393980996176053,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9813
    },
    {
      "epoch": 0.09814,
      "grad_norm": 0.9318454756189145,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 9814
    },
    {
      "epoch": 0.09815,
      "grad_norm": 0.9599372181289262,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 9815
    },
    {
      "epoch": 0.09816,
      "grad_norm": 0.8956701047579604,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 9816
    },
    {
      "epoch": 0.09817,
      "grad_norm": 0.7842843597137388,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9817
    },
    {
      "epoch": 0.09818,
      "grad_norm": 0.6893223770225535,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 9818
    },
    {
      "epoch": 0.09819,
      "grad_norm": 0.5764089558777323,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 9819
    },
    {
      "epoch": 0.0982,
      "grad_norm": 0.6831187750693034,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 9820
    },
    {
      "epoch": 0.09821,
      "grad_norm": 0.7420126000497227,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 9821
    },
    {
      "epoch": 0.09822,
      "grad_norm": 0.7918367231819561,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 9822
    },
    {
      "epoch": 0.09823,
      "grad_norm": 0.8307403908166644,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 9823
    },
    {
      "epoch": 0.09824,
      "grad_norm": 0.8384902251512314,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9824
    },
    {
      "epoch": 0.09825,
      "grad_norm": 0.9488542492159057,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9825
    },
    {
      "epoch": 0.09826,
      "grad_norm": 0.9814885356015456,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 9826
    },
    {
      "epoch": 0.09827,
      "grad_norm": 1.0404257496474245,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 9827
    },
    {
      "epoch": 0.09828,
      "grad_norm": 0.9374789182838859,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 9828
    },
    {
      "epoch": 0.09829,
      "grad_norm": 0.8352504231117999,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 9829
    },
    {
      "epoch": 0.0983,
      "grad_norm": 0.8712005573355397,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 9830
    },
    {
      "epoch": 0.09831,
      "grad_norm": 0.8221414372597472,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 9831
    },
    {
      "epoch": 0.09832,
      "grad_norm": 0.7097651197485733,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 9832
    },
    {
      "epoch": 0.09833,
      "grad_norm": 0.7171718585045,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 9833
    },
    {
      "epoch": 0.09834,
      "grad_norm": 0.7926851111043537,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 9834
    },
    {
      "epoch": 0.09835,
      "grad_norm": 0.8623151532060463,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9835
    },
    {
      "epoch": 0.09836,
      "grad_norm": 0.8199248380589881,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 9836
    },
    {
      "epoch": 0.09837,
      "grad_norm": 0.7755743905810927,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 9837
    },
    {
      "epoch": 0.09838,
      "grad_norm": 0.7705706750252852,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 9838
    },
    {
      "epoch": 0.09839,
      "grad_norm": 0.845394429517052,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 9839
    },
    {
      "epoch": 0.0984,
      "grad_norm": 0.9405467332911781,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 9840
    },
    {
      "epoch": 0.09841,
      "grad_norm": 0.8405964634828268,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 9841
    },
    {
      "epoch": 0.09842,
      "grad_norm": 0.7811416765302842,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 9842
    },
    {
      "epoch": 0.09843,
      "grad_norm": 0.8180023616159704,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 9843
    },
    {
      "epoch": 0.09844,
      "grad_norm": 0.7527374804721993,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 9844
    },
    {
      "epoch": 0.09845,
      "grad_norm": 0.7530390141763947,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 9845
    },
    {
      "epoch": 0.09846,
      "grad_norm": 0.8899981921507195,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 9846
    },
    {
      "epoch": 0.09847,
      "grad_norm": 1.1111905360253278,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 9847
    },
    {
      "epoch": 0.09848,
      "grad_norm": 0.8799351792098972,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 9848
    },
    {
      "epoch": 0.09849,
      "grad_norm": 0.6714966433483521,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 9849
    },
    {
      "epoch": 0.0985,
      "grad_norm": 0.660208855694013,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 9850
    },
    {
      "epoch": 0.09851,
      "grad_norm": 0.7037816807855953,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 9851
    },
    {
      "epoch": 0.09852,
      "grad_norm": 0.7854724491123188,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 9852
    },
    {
      "epoch": 0.09853,
      "grad_norm": 0.8934963259953501,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 9853
    },
    {
      "epoch": 0.09854,
      "grad_norm": 0.8084018134106394,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 9854
    },
    {
      "epoch": 0.09855,
      "grad_norm": 0.795443981883104,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 9855
    },
    {
      "epoch": 0.09856,
      "grad_norm": 0.7281963252673843,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 9856
    },
    {
      "epoch": 0.09857,
      "grad_norm": 0.7182745458303548,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 9857
    },
    {
      "epoch": 0.09858,
      "grad_norm": 0.7050569720124196,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 9858
    },
    {
      "epoch": 0.09859,
      "grad_norm": 0.6807473261285026,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 9859
    },
    {
      "epoch": 0.0986,
      "grad_norm": 0.6746756830722065,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 9860
    },
    {
      "epoch": 0.09861,
      "grad_norm": 0.7007907511364769,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 9861
    },
    {
      "epoch": 0.09862,
      "grad_norm": 0.7927126135369934,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 9862
    },
    {
      "epoch": 0.09863,
      "grad_norm": 0.8386321721308743,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 9863
    },
    {
      "epoch": 0.09864,
      "grad_norm": 0.7507936235349171,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 9864
    },
    {
      "epoch": 0.09865,
      "grad_norm": 0.6481226992818734,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 9865
    },
    {
      "epoch": 0.09866,
      "grad_norm": 0.5697115094410177,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 9866
    },
    {
      "epoch": 0.09867,
      "grad_norm": 0.5126449464627183,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 9867
    },
    {
      "epoch": 0.09868,
      "grad_norm": 0.5102549762756471,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 9868
    },
    {
      "epoch": 0.09869,
      "grad_norm": 0.5085779141402881,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 9869
    },
    {
      "epoch": 0.0987,
      "grad_norm": 0.5428508446899835,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 9870
    },
    {
      "epoch": 0.09871,
      "grad_norm": 0.6546855703044885,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 9871
    },
    {
      "epoch": 0.09872,
      "grad_norm": 0.6934739489178067,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 9872
    },
    {
      "epoch": 0.09873,
      "grad_norm": 0.6568023982508876,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 9873
    },
    {
      "epoch": 0.09874,
      "grad_norm": 0.7426087598644382,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 9874
    },
    {
      "epoch": 0.09875,
      "grad_norm": 0.9646739748869851,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 9875
    },
    {
      "epoch": 0.09876,
      "grad_norm": 1.1760726018799912,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9876
    },
    {
      "epoch": 0.09877,
      "grad_norm": 1.032505678325432,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9877
    },
    {
      "epoch": 0.09878,
      "grad_norm": 0.9520034852500903,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 9878
    },
    {
      "epoch": 0.09879,
      "grad_norm": 0.8304599751268602,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 9879
    },
    {
      "epoch": 0.0988,
      "grad_norm": 0.7973040789135167,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 9880
    },
    {
      "epoch": 0.09881,
      "grad_norm": 0.6749261652206336,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 9881
    },
    {
      "epoch": 0.09882,
      "grad_norm": 0.6720424310436138,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 9882
    },
    {
      "epoch": 0.09883,
      "grad_norm": 0.6552937594046303,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 9883
    },
    {
      "epoch": 0.09884,
      "grad_norm": 0.8385733301338101,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 9884
    },
    {
      "epoch": 0.09885,
      "grad_norm": 0.787180609022946,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 9885
    },
    {
      "epoch": 0.09886,
      "grad_norm": 0.7214374906120231,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 9886
    },
    {
      "epoch": 0.09887,
      "grad_norm": 0.813828996233088,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 9887
    },
    {
      "epoch": 0.09888,
      "grad_norm": 0.9395429521943715,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 9888
    },
    {
      "epoch": 0.09889,
      "grad_norm": 0.9594036018439058,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 9889
    },
    {
      "epoch": 0.0989,
      "grad_norm": 1.0604047721378298,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 9890
    },
    {
      "epoch": 0.09891,
      "grad_norm": 0.9760792138626977,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 9891
    },
    {
      "epoch": 0.09892,
      "grad_norm": 0.8055680873064225,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 9892
    },
    {
      "epoch": 0.09893,
      "grad_norm": 0.7150553503473565,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 9893
    },
    {
      "epoch": 0.09894,
      "grad_norm": 0.5919754883801267,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 9894
    },
    {
      "epoch": 0.09895,
      "grad_norm": 0.6042264333876616,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 9895
    },
    {
      "epoch": 0.09896,
      "grad_norm": 0.5430950181914646,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 9896
    },
    {
      "epoch": 0.09897,
      "grad_norm": 0.7064510242910831,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 9897
    },
    {
      "epoch": 0.09898,
      "grad_norm": 0.8003106649351045,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 9898
    },
    {
      "epoch": 0.09899,
      "grad_norm": 0.8270519641387873,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 9899
    },
    {
      "epoch": 0.099,
      "grad_norm": 0.7540500751340824,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 9900
    },
    {
      "epoch": 0.09901,
      "grad_norm": 0.6671609878427571,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 9901
    },
    {
      "epoch": 0.09902,
      "grad_norm": 0.7272528832102878,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 9902
    },
    {
      "epoch": 0.09903,
      "grad_norm": 0.749667895778945,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9903
    },
    {
      "epoch": 0.09904,
      "grad_norm": 0.6741213580842577,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 9904
    },
    {
      "epoch": 0.09905,
      "grad_norm": 0.6736961275412793,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 9905
    },
    {
      "epoch": 0.09906,
      "grad_norm": 0.6560958029133749,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 9906
    },
    {
      "epoch": 0.09907,
      "grad_norm": 0.631422253019129,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 9907
    },
    {
      "epoch": 0.09908,
      "grad_norm": 0.5392615341333712,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 9908
    },
    {
      "epoch": 0.09909,
      "grad_norm": 0.5811455319662365,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 9909
    },
    {
      "epoch": 0.0991,
      "grad_norm": 0.6707319041356397,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 9910
    },
    {
      "epoch": 0.09911,
      "grad_norm": 0.771243416598569,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 9911
    },
    {
      "epoch": 0.09912,
      "grad_norm": 0.8496996867713535,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 9912
    },
    {
      "epoch": 0.09913,
      "grad_norm": 0.8902788189271055,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 9913
    },
    {
      "epoch": 0.09914,
      "grad_norm": 0.9357066721293495,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 9914
    },
    {
      "epoch": 0.09915,
      "grad_norm": 0.9755645470634543,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 9915
    },
    {
      "epoch": 0.09916,
      "grad_norm": 1.0089094696696197,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 9916
    },
    {
      "epoch": 0.09917,
      "grad_norm": 0.8817642996354145,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 9917
    },
    {
      "epoch": 0.09918,
      "grad_norm": 0.7658568526663542,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 9918
    },
    {
      "epoch": 0.09919,
      "grad_norm": 0.8128521592081074,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 9919
    },
    {
      "epoch": 0.0992,
      "grad_norm": 0.8609166031572244,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 9920
    },
    {
      "epoch": 0.09921,
      "grad_norm": 1.0125750989658047,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 9921
    },
    {
      "epoch": 0.09922,
      "grad_norm": 1.0032822006219375,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 9922
    },
    {
      "epoch": 0.09923,
      "grad_norm": 1.2307924439733817,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 9923
    },
    {
      "epoch": 0.09924,
      "grad_norm": 1.0090705722158624,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 9924
    },
    {
      "epoch": 0.09925,
      "grad_norm": 1.0212794524503392,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 9925
    },
    {
      "epoch": 0.09926,
      "grad_norm": 0.9853180213898505,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 9926
    },
    {
      "epoch": 0.09927,
      "grad_norm": 0.758327811881651,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 9927
    },
    {
      "epoch": 0.09928,
      "grad_norm": 0.7659641009511424,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9928
    },
    {
      "epoch": 0.09929,
      "grad_norm": 0.7504067878828342,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 9929
    },
    {
      "epoch": 0.0993,
      "grad_norm": 0.6947364493882249,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 9930
    },
    {
      "epoch": 0.09931,
      "grad_norm": 0.6761909114866678,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 9931
    },
    {
      "epoch": 0.09932,
      "grad_norm": 0.6135941560379718,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9932
    },
    {
      "epoch": 0.09933,
      "grad_norm": 0.582085876114816,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 9933
    },
    {
      "epoch": 0.09934,
      "grad_norm": 0.6180077412932585,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 9934
    },
    {
      "epoch": 0.09935,
      "grad_norm": 0.6156612388696766,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 9935
    },
    {
      "epoch": 0.09936,
      "grad_norm": 0.6398409744821787,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 9936
    },
    {
      "epoch": 0.09937,
      "grad_norm": 0.6544534848994807,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 9937
    },
    {
      "epoch": 0.09938,
      "grad_norm": 0.8269077552442352,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 9938
    },
    {
      "epoch": 0.09939,
      "grad_norm": 0.9971115211192825,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 9939
    },
    {
      "epoch": 0.0994,
      "grad_norm": 0.9857415352546822,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 9940
    },
    {
      "epoch": 0.09941,
      "grad_norm": 0.8137608150128142,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 9941
    },
    {
      "epoch": 0.09942,
      "grad_norm": 0.593147335054599,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 9942
    },
    {
      "epoch": 0.09943,
      "grad_norm": 0.5875666526054226,
      "learning_rate": 0.003,
      "loss": 4.0301,
      "step": 9943
    },
    {
      "epoch": 0.09944,
      "grad_norm": 0.5879211483291883,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 9944
    },
    {
      "epoch": 0.09945,
      "grad_norm": 0.6905176331929537,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 9945
    },
    {
      "epoch": 0.09946,
      "grad_norm": 0.7198664855118773,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 9946
    },
    {
      "epoch": 0.09947,
      "grad_norm": 0.752440611386409,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9947
    },
    {
      "epoch": 0.09948,
      "grad_norm": 0.8192932165988773,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 9948
    },
    {
      "epoch": 0.09949,
      "grad_norm": 0.8209749794970975,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 9949
    },
    {
      "epoch": 0.0995,
      "grad_norm": 0.8358143036904804,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 9950
    },
    {
      "epoch": 0.09951,
      "grad_norm": 0.8867621716493573,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 9951
    },
    {
      "epoch": 0.09952,
      "grad_norm": 0.8466716064437273,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 9952
    },
    {
      "epoch": 0.09953,
      "grad_norm": 0.8254486041624173,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 9953
    },
    {
      "epoch": 0.09954,
      "grad_norm": 0.7712441746642007,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 9954
    },
    {
      "epoch": 0.09955,
      "grad_norm": 0.6903202853138538,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 9955
    },
    {
      "epoch": 0.09956,
      "grad_norm": 0.5906670263661717,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 9956
    },
    {
      "epoch": 0.09957,
      "grad_norm": 0.5743206687874738,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 9957
    },
    {
      "epoch": 0.09958,
      "grad_norm": 0.590105027476533,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 9958
    },
    {
      "epoch": 0.09959,
      "grad_norm": 0.6478146939289141,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 9959
    },
    {
      "epoch": 0.0996,
      "grad_norm": 0.6125903545494685,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 9960
    },
    {
      "epoch": 0.09961,
      "grad_norm": 0.6855179525277352,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 9961
    },
    {
      "epoch": 0.09962,
      "grad_norm": 0.8245686493079379,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 9962
    },
    {
      "epoch": 0.09963,
      "grad_norm": 1.1718745927375416,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 9963
    },
    {
      "epoch": 0.09964,
      "grad_norm": 0.7822113890013278,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 9964
    },
    {
      "epoch": 0.09965,
      "grad_norm": 0.6627168370652431,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 9965
    },
    {
      "epoch": 0.09966,
      "grad_norm": 0.6816975050009425,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 9966
    },
    {
      "epoch": 0.09967,
      "grad_norm": 0.6410163787617061,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9967
    },
    {
      "epoch": 0.09968,
      "grad_norm": 0.5513934862668493,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 9968
    },
    {
      "epoch": 0.09969,
      "grad_norm": 0.5717861304898563,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 9969
    },
    {
      "epoch": 0.0997,
      "grad_norm": 0.6378222450638118,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 9970
    },
    {
      "epoch": 0.09971,
      "grad_norm": 0.6879760044616384,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 9971
    },
    {
      "epoch": 0.09972,
      "grad_norm": 0.7703686261003476,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 9972
    },
    {
      "epoch": 0.09973,
      "grad_norm": 0.8455904106463509,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 9973
    },
    {
      "epoch": 0.09974,
      "grad_norm": 0.9031224948235831,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 9974
    },
    {
      "epoch": 0.09975,
      "grad_norm": 0.7834229298155747,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 9975
    },
    {
      "epoch": 0.09976,
      "grad_norm": 0.7374608282517201,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 9976
    },
    {
      "epoch": 0.09977,
      "grad_norm": 0.6475022518581292,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 9977
    },
    {
      "epoch": 0.09978,
      "grad_norm": 0.6655914091026776,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 9978
    },
    {
      "epoch": 0.09979,
      "grad_norm": 0.6092837390948476,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 9979
    },
    {
      "epoch": 0.0998,
      "grad_norm": 0.6222297729643492,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 9980
    },
    {
      "epoch": 0.09981,
      "grad_norm": 0.5644346622119025,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 9981
    },
    {
      "epoch": 0.09982,
      "grad_norm": 0.5746206355689046,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 9982
    },
    {
      "epoch": 0.09983,
      "grad_norm": 0.6579561700974546,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 9983
    },
    {
      "epoch": 0.09984,
      "grad_norm": 0.9489942242805092,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 9984
    },
    {
      "epoch": 0.09985,
      "grad_norm": 1.5512495843402707,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 9985
    },
    {
      "epoch": 0.09986,
      "grad_norm": 0.5642219821127367,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 9986
    },
    {
      "epoch": 0.09987,
      "grad_norm": 0.8878804991370798,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 9987
    },
    {
      "epoch": 0.09988,
      "grad_norm": 0.9616073363226949,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 9988
    },
    {
      "epoch": 0.09989,
      "grad_norm": 0.905824845539826,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9989
    },
    {
      "epoch": 0.0999,
      "grad_norm": 0.8647017499059604,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 9990
    },
    {
      "epoch": 0.09991,
      "grad_norm": 0.8032896275364889,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 9991
    },
    {
      "epoch": 0.09992,
      "grad_norm": 0.7279383274198165,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 9992
    },
    {
      "epoch": 0.09993,
      "grad_norm": 0.7799466970112703,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9993
    },
    {
      "epoch": 0.09994,
      "grad_norm": 0.9145234797734555,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 9994
    },
    {
      "epoch": 0.09995,
      "grad_norm": 0.8201103396095457,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 9995
    },
    {
      "epoch": 0.09996,
      "grad_norm": 0.9047552292555922,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 9996
    },
    {
      "epoch": 0.09997,
      "grad_norm": 0.8963430475859631,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 9997
    },
    {
      "epoch": 0.09998,
      "grad_norm": 0.813945518877336,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 9998
    },
    {
      "epoch": 0.09999,
      "grad_norm": 0.6549507314006496,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9999
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6519137253369942,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 10000
    },
    {
      "epoch": 0.10001,
      "grad_norm": 0.6310538887421997,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 10001
    },
    {
      "epoch": 0.10002,
      "grad_norm": 0.6391259755212524,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 10002
    },
    {
      "epoch": 0.10003,
      "grad_norm": 0.5978190100029864,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 10003
    },
    {
      "epoch": 0.10004,
      "grad_norm": 0.7353677870344891,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 10004
    },
    {
      "epoch": 0.10005,
      "grad_norm": 0.8600701195416798,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 10005
    },
    {
      "epoch": 0.10006,
      "grad_norm": 0.7651339491329487,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 10006
    },
    {
      "epoch": 0.10007,
      "grad_norm": 0.8281866504989451,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 10007
    },
    {
      "epoch": 0.10008,
      "grad_norm": 0.898976147912649,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 10008
    },
    {
      "epoch": 0.10009,
      "grad_norm": 0.9235345887520816,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 10009
    },
    {
      "epoch": 0.1001,
      "grad_norm": 1.1045884205459653,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 10010
    },
    {
      "epoch": 0.10011,
      "grad_norm": 0.8047977290401939,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 10011
    },
    {
      "epoch": 0.10012,
      "grad_norm": 0.8491840634842334,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 10012
    },
    {
      "epoch": 0.10013,
      "grad_norm": 1.0828080137474874,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 10013
    },
    {
      "epoch": 0.10014,
      "grad_norm": 1.1635619875301892,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 10014
    },
    {
      "epoch": 0.10015,
      "grad_norm": 0.897354796776556,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 10015
    },
    {
      "epoch": 0.10016,
      "grad_norm": 0.7780665867165246,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 10016
    },
    {
      "epoch": 0.10017,
      "grad_norm": 0.683817569961156,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 10017
    },
    {
      "epoch": 0.10018,
      "grad_norm": 0.7740532824418541,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 10018
    },
    {
      "epoch": 0.10019,
      "grad_norm": 0.8606450031261723,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 10019
    },
    {
      "epoch": 0.1002,
      "grad_norm": 0.808735430817491,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 10020
    },
    {
      "epoch": 0.10021,
      "grad_norm": 0.6620052775434282,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 10021
    },
    {
      "epoch": 0.10022,
      "grad_norm": 0.6154822144201489,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 10022
    },
    {
      "epoch": 0.10023,
      "grad_norm": 0.5187491125035918,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 10023
    },
    {
      "epoch": 0.10024,
      "grad_norm": 0.4942144582638463,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 10024
    },
    {
      "epoch": 0.10025,
      "grad_norm": 0.46654780612751734,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 10025
    },
    {
      "epoch": 0.10026,
      "grad_norm": 0.5004543480891657,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 10026
    },
    {
      "epoch": 0.10027,
      "grad_norm": 0.5513832529483311,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 10027
    },
    {
      "epoch": 0.10028,
      "grad_norm": 0.6352605183788917,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 10028
    },
    {
      "epoch": 0.10029,
      "grad_norm": 0.8367870123940754,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 10029
    },
    {
      "epoch": 0.1003,
      "grad_norm": 1.0581826188086585,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 10030
    },
    {
      "epoch": 0.10031,
      "grad_norm": 1.0220085066989417,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 10031
    },
    {
      "epoch": 0.10032,
      "grad_norm": 0.7783369834157056,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 10032
    },
    {
      "epoch": 0.10033,
      "grad_norm": 0.6709417988465439,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 10033
    },
    {
      "epoch": 0.10034,
      "grad_norm": 0.6669434095115762,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 10034
    },
    {
      "epoch": 0.10035,
      "grad_norm": 0.6147566105234249,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 10035
    },
    {
      "epoch": 0.10036,
      "grad_norm": 0.6529090399469691,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 10036
    },
    {
      "epoch": 0.10037,
      "grad_norm": 0.7107482522994412,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 10037
    },
    {
      "epoch": 0.10038,
      "grad_norm": 0.8197362150703974,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 10038
    },
    {
      "epoch": 0.10039,
      "grad_norm": 0.8295877869862686,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 10039
    },
    {
      "epoch": 0.1004,
      "grad_norm": 0.7374421686069658,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 10040
    },
    {
      "epoch": 0.10041,
      "grad_norm": 0.8214317115707387,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 10041
    },
    {
      "epoch": 0.10042,
      "grad_norm": 0.895440358479686,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 10042
    },
    {
      "epoch": 0.10043,
      "grad_norm": 0.9993692769809046,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 10043
    },
    {
      "epoch": 0.10044,
      "grad_norm": 1.072699158101983,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 10044
    },
    {
      "epoch": 0.10045,
      "grad_norm": 0.9388431886106355,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 10045
    },
    {
      "epoch": 0.10046,
      "grad_norm": 1.030884690149999,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 10046
    },
    {
      "epoch": 0.10047,
      "grad_norm": 1.0068468868079046,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 10047
    },
    {
      "epoch": 0.10048,
      "grad_norm": 0.7705600866643636,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 10048
    },
    {
      "epoch": 0.10049,
      "grad_norm": 0.8194123489368415,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 10049
    },
    {
      "epoch": 0.1005,
      "grad_norm": 0.869020771465238,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 10050
    },
    {
      "epoch": 0.10051,
      "grad_norm": 0.9307225832163902,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 10051
    },
    {
      "epoch": 0.10052,
      "grad_norm": 0.8621096104565541,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 10052
    },
    {
      "epoch": 0.10053,
      "grad_norm": 0.7248964302923027,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 10053
    },
    {
      "epoch": 0.10054,
      "grad_norm": 0.6696966942680252,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 10054
    },
    {
      "epoch": 0.10055,
      "grad_norm": 0.7706194800493945,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 10055
    },
    {
      "epoch": 0.10056,
      "grad_norm": 0.7476737464111006,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 10056
    },
    {
      "epoch": 0.10057,
      "grad_norm": 0.7734457893757593,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 10057
    },
    {
      "epoch": 0.10058,
      "grad_norm": 0.678252649945835,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 10058
    },
    {
      "epoch": 0.10059,
      "grad_norm": 0.6595095553472218,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 10059
    },
    {
      "epoch": 0.1006,
      "grad_norm": 0.7002337222649138,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 10060
    },
    {
      "epoch": 0.10061,
      "grad_norm": 0.7739912428589064,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 10061
    },
    {
      "epoch": 0.10062,
      "grad_norm": 0.8094580611217377,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 10062
    },
    {
      "epoch": 0.10063,
      "grad_norm": 0.7849235571724628,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 10063
    },
    {
      "epoch": 0.10064,
      "grad_norm": 0.71688513935201,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 10064
    },
    {
      "epoch": 0.10065,
      "grad_norm": 0.5387402250397292,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 10065
    },
    {
      "epoch": 0.10066,
      "grad_norm": 0.5738573805956311,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 10066
    },
    {
      "epoch": 0.10067,
      "grad_norm": 0.5128916053318231,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 10067
    },
    {
      "epoch": 0.10068,
      "grad_norm": 0.5303861773665701,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 10068
    },
    {
      "epoch": 0.10069,
      "grad_norm": 0.5363307400406016,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 10069
    },
    {
      "epoch": 0.1007,
      "grad_norm": 0.6677766567970822,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 10070
    },
    {
      "epoch": 0.10071,
      "grad_norm": 0.869800192886798,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 10071
    },
    {
      "epoch": 0.10072,
      "grad_norm": 1.241986467031261,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 10072
    },
    {
      "epoch": 0.10073,
      "grad_norm": 0.7044633440415046,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 10073
    },
    {
      "epoch": 0.10074,
      "grad_norm": 0.6016954396918823,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 10074
    },
    {
      "epoch": 0.10075,
      "grad_norm": 0.6506653532205113,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 10075
    },
    {
      "epoch": 0.10076,
      "grad_norm": 0.5913298031674521,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 10076
    },
    {
      "epoch": 0.10077,
      "grad_norm": 0.6612064518297085,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 10077
    },
    {
      "epoch": 0.10078,
      "grad_norm": 0.7381204773241105,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 10078
    },
    {
      "epoch": 0.10079,
      "grad_norm": 0.9136392204855105,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 10079
    },
    {
      "epoch": 0.1008,
      "grad_norm": 1.0551033651714792,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 10080
    },
    {
      "epoch": 0.10081,
      "grad_norm": 1.021241306196213,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 10081
    },
    {
      "epoch": 0.10082,
      "grad_norm": 0.9444310820937741,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 10082
    },
    {
      "epoch": 0.10083,
      "grad_norm": 0.8028394635763569,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 10083
    },
    {
      "epoch": 0.10084,
      "grad_norm": 0.6931291147221774,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 10084
    },
    {
      "epoch": 0.10085,
      "grad_norm": 0.6614930979877484,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 10085
    },
    {
      "epoch": 0.10086,
      "grad_norm": 0.6447028805806914,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 10086
    },
    {
      "epoch": 0.10087,
      "grad_norm": 0.7271300834569303,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 10087
    },
    {
      "epoch": 0.10088,
      "grad_norm": 0.8548799986148488,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 10088
    },
    {
      "epoch": 0.10089,
      "grad_norm": 0.8660557367235694,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 10089
    },
    {
      "epoch": 0.1009,
      "grad_norm": 0.8905488162446568,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 10090
    },
    {
      "epoch": 0.10091,
      "grad_norm": 0.9129923392768977,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 10091
    },
    {
      "epoch": 0.10092,
      "grad_norm": 0.9486091293972858,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 10092
    },
    {
      "epoch": 0.10093,
      "grad_norm": 0.9202478176786889,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 10093
    },
    {
      "epoch": 0.10094,
      "grad_norm": 0.9444438231215719,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 10094
    },
    {
      "epoch": 0.10095,
      "grad_norm": 1.0023114763189855,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 10095
    },
    {
      "epoch": 0.10096,
      "grad_norm": 0.9008830005116453,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 10096
    },
    {
      "epoch": 0.10097,
      "grad_norm": 0.7394970221020885,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 10097
    },
    {
      "epoch": 0.10098,
      "grad_norm": 0.7119640016754663,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 10098
    },
    {
      "epoch": 0.10099,
      "grad_norm": 0.6795956150432717,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 10099
    },
    {
      "epoch": 0.101,
      "grad_norm": 0.7985963421953124,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 10100
    },
    {
      "epoch": 0.10101,
      "grad_norm": 0.9676465329947534,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 10101
    },
    {
      "epoch": 0.10102,
      "grad_norm": 0.9402257720277828,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 10102
    },
    {
      "epoch": 0.10103,
      "grad_norm": 0.8874925029060675,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 10103
    },
    {
      "epoch": 0.10104,
      "grad_norm": 0.9117348708084444,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 10104
    },
    {
      "epoch": 0.10105,
      "grad_norm": 0.8479646908158324,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 10105
    },
    {
      "epoch": 0.10106,
      "grad_norm": 0.896593809928579,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 10106
    },
    {
      "epoch": 0.10107,
      "grad_norm": 0.8315621037343224,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 10107
    },
    {
      "epoch": 0.10108,
      "grad_norm": 0.7681718745034675,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 10108
    },
    {
      "epoch": 0.10109,
      "grad_norm": 0.8233892495101346,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 10109
    },
    {
      "epoch": 0.1011,
      "grad_norm": 0.9016640807163642,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 10110
    },
    {
      "epoch": 0.10111,
      "grad_norm": 0.9170672273091978,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 10111
    },
    {
      "epoch": 0.10112,
      "grad_norm": 0.8351702169958964,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 10112
    },
    {
      "epoch": 0.10113,
      "grad_norm": 0.8089461545174494,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 10113
    },
    {
      "epoch": 0.10114,
      "grad_norm": 0.9107383212481057,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 10114
    },
    {
      "epoch": 0.10115,
      "grad_norm": 0.8118768018834102,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 10115
    },
    {
      "epoch": 0.10116,
      "grad_norm": 0.7983571219601531,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 10116
    },
    {
      "epoch": 0.10117,
      "grad_norm": 0.7951177221933025,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 10117
    },
    {
      "epoch": 0.10118,
      "grad_norm": 0.8413030315926632,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 10118
    },
    {
      "epoch": 0.10119,
      "grad_norm": 0.716518761890454,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 10119
    },
    {
      "epoch": 0.1012,
      "grad_norm": 0.6486824283257824,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 10120
    },
    {
      "epoch": 0.10121,
      "grad_norm": 0.5880476149618893,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 10121
    },
    {
      "epoch": 0.10122,
      "grad_norm": 0.5866107750288472,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 10122
    },
    {
      "epoch": 0.10123,
      "grad_norm": 0.5811881253993841,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 10123
    },
    {
      "epoch": 0.10124,
      "grad_norm": 0.5569420366913549,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 10124
    },
    {
      "epoch": 0.10125,
      "grad_norm": 0.5942996797715101,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 10125
    },
    {
      "epoch": 0.10126,
      "grad_norm": 0.5670677013779793,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 10126
    },
    {
      "epoch": 0.10127,
      "grad_norm": 0.529089784800278,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 10127
    },
    {
      "epoch": 0.10128,
      "grad_norm": 0.4168566029164349,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 10128
    },
    {
      "epoch": 0.10129,
      "grad_norm": 0.4311731795710161,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 10129
    },
    {
      "epoch": 0.1013,
      "grad_norm": 0.4477041043125615,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 10130
    },
    {
      "epoch": 0.10131,
      "grad_norm": 0.5027821507640442,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 10131
    },
    {
      "epoch": 0.10132,
      "grad_norm": 0.5493559629395345,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 10132
    },
    {
      "epoch": 0.10133,
      "grad_norm": 0.6452629987298875,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 10133
    },
    {
      "epoch": 0.10134,
      "grad_norm": 0.8688443183887481,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 10134
    },
    {
      "epoch": 0.10135,
      "grad_norm": 1.1410657478991406,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 10135
    },
    {
      "epoch": 0.10136,
      "grad_norm": 0.923735834137093,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 10136
    },
    {
      "epoch": 0.10137,
      "grad_norm": 0.7985668392366627,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 10137
    },
    {
      "epoch": 0.10138,
      "grad_norm": 0.6893008020293587,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 10138
    },
    {
      "epoch": 0.10139,
      "grad_norm": 0.699027589097982,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 10139
    },
    {
      "epoch": 0.1014,
      "grad_norm": 0.7173481888193998,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 10140
    },
    {
      "epoch": 0.10141,
      "grad_norm": 0.6959173004694686,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 10141
    },
    {
      "epoch": 0.10142,
      "grad_norm": 0.813260923915381,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 10142
    },
    {
      "epoch": 0.10143,
      "grad_norm": 0.8889759915027781,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 10143
    },
    {
      "epoch": 0.10144,
      "grad_norm": 0.6930759800585614,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 10144
    },
    {
      "epoch": 0.10145,
      "grad_norm": 0.7031416657156874,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 10145
    },
    {
      "epoch": 0.10146,
      "grad_norm": 0.8186555112243362,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 10146
    },
    {
      "epoch": 0.10147,
      "grad_norm": 0.972515461562128,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 10147
    },
    {
      "epoch": 0.10148,
      "grad_norm": 1.2251324017059566,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 10148
    },
    {
      "epoch": 0.10149,
      "grad_norm": 0.7575767344901424,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 10149
    },
    {
      "epoch": 0.1015,
      "grad_norm": 0.7257739095105269,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 10150
    },
    {
      "epoch": 0.10151,
      "grad_norm": 0.6850837341823302,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 10151
    },
    {
      "epoch": 0.10152,
      "grad_norm": 0.6901837020889886,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 10152
    },
    {
      "epoch": 0.10153,
      "grad_norm": 0.7366995404122453,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 10153
    },
    {
      "epoch": 0.10154,
      "grad_norm": 0.7729200614601484,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 10154
    },
    {
      "epoch": 0.10155,
      "grad_norm": 0.9179351051298269,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 10155
    },
    {
      "epoch": 0.10156,
      "grad_norm": 1.079772577207657,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 10156
    },
    {
      "epoch": 0.10157,
      "grad_norm": 0.8478718907398735,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 10157
    },
    {
      "epoch": 0.10158,
      "grad_norm": 0.8076358725971214,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 10158
    },
    {
      "epoch": 0.10159,
      "grad_norm": 0.7182692438467507,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 10159
    },
    {
      "epoch": 0.1016,
      "grad_norm": 0.6354632732608569,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 10160
    },
    {
      "epoch": 0.10161,
      "grad_norm": 0.6096925827913215,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 10161
    },
    {
      "epoch": 0.10162,
      "grad_norm": 0.6412613510772903,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 10162
    },
    {
      "epoch": 0.10163,
      "grad_norm": 0.7228497893130814,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 10163
    },
    {
      "epoch": 0.10164,
      "grad_norm": 0.8189338574542239,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 10164
    },
    {
      "epoch": 0.10165,
      "grad_norm": 0.8802770642728063,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 10165
    },
    {
      "epoch": 0.10166,
      "grad_norm": 0.9340783981991572,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 10166
    },
    {
      "epoch": 0.10167,
      "grad_norm": 1.0522661488882454,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 10167
    },
    {
      "epoch": 0.10168,
      "grad_norm": 1.0883984803163758,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 10168
    },
    {
      "epoch": 0.10169,
      "grad_norm": 0.8718130916305842,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 10169
    },
    {
      "epoch": 0.1017,
      "grad_norm": 0.7267906765452536,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 10170
    },
    {
      "epoch": 0.10171,
      "grad_norm": 0.6741776316878704,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 10171
    },
    {
      "epoch": 0.10172,
      "grad_norm": 0.6366479102607301,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 10172
    },
    {
      "epoch": 0.10173,
      "grad_norm": 0.6255451070036055,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 10173
    },
    {
      "epoch": 0.10174,
      "grad_norm": 0.6833151370840097,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 10174
    },
    {
      "epoch": 0.10175,
      "grad_norm": 0.6980219428980601,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 10175
    },
    {
      "epoch": 0.10176,
      "grad_norm": 0.7375496501515941,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 10176
    },
    {
      "epoch": 0.10177,
      "grad_norm": 0.7898802215772395,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 10177
    },
    {
      "epoch": 0.10178,
      "grad_norm": 0.9839723040732539,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 10178
    },
    {
      "epoch": 0.10179,
      "grad_norm": 1.0558006196476055,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 10179
    },
    {
      "epoch": 0.1018,
      "grad_norm": 0.8839567388022763,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 10180
    },
    {
      "epoch": 0.10181,
      "grad_norm": 0.8323042106193205,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 10181
    },
    {
      "epoch": 0.10182,
      "grad_norm": 0.8791136712281088,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 10182
    },
    {
      "epoch": 0.10183,
      "grad_norm": 1.0917668790659707,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 10183
    },
    {
      "epoch": 0.10184,
      "grad_norm": 0.9025576624348437,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 10184
    },
    {
      "epoch": 0.10185,
      "grad_norm": 0.7798371617092108,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 10185
    },
    {
      "epoch": 0.10186,
      "grad_norm": 0.7713060566227249,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 10186
    },
    {
      "epoch": 0.10187,
      "grad_norm": 0.7563056494272968,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 10187
    },
    {
      "epoch": 0.10188,
      "grad_norm": 0.7772228424671542,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 10188
    },
    {
      "epoch": 0.10189,
      "grad_norm": 0.8260486212304557,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 10189
    },
    {
      "epoch": 0.1019,
      "grad_norm": 1.1171480943910872,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 10190
    },
    {
      "epoch": 0.10191,
      "grad_norm": 0.9843614175223876,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 10191
    },
    {
      "epoch": 0.10192,
      "grad_norm": 1.036663948755249,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 10192
    },
    {
      "epoch": 0.10193,
      "grad_norm": 0.8972801457141254,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 10193
    },
    {
      "epoch": 0.10194,
      "grad_norm": 0.8235311205411696,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 10194
    },
    {
      "epoch": 0.10195,
      "grad_norm": 0.709463331336577,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 10195
    },
    {
      "epoch": 0.10196,
      "grad_norm": 0.5235487875342617,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 10196
    },
    {
      "epoch": 0.10197,
      "grad_norm": 0.5246146886760584,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 10197
    },
    {
      "epoch": 0.10198,
      "grad_norm": 0.5530686360133628,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 10198
    },
    {
      "epoch": 0.10199,
      "grad_norm": 0.6533458323226683,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 10199
    },
    {
      "epoch": 0.102,
      "grad_norm": 0.8138884437121796,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 10200
    },
    {
      "epoch": 0.10201,
      "grad_norm": 0.8833147158811808,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 10201
    },
    {
      "epoch": 0.10202,
      "grad_norm": 0.8602495065910606,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 10202
    },
    {
      "epoch": 0.10203,
      "grad_norm": 0.8167686182900835,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 10203
    },
    {
      "epoch": 0.10204,
      "grad_norm": 0.69579025866685,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 10204
    },
    {
      "epoch": 0.10205,
      "grad_norm": 0.6952695693816464,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 10205
    },
    {
      "epoch": 0.10206,
      "grad_norm": 0.7178836748410159,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 10206
    },
    {
      "epoch": 0.10207,
      "grad_norm": 0.7051292005693832,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 10207
    },
    {
      "epoch": 0.10208,
      "grad_norm": 0.7123121092039982,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 10208
    },
    {
      "epoch": 0.10209,
      "grad_norm": 0.7357546073327017,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 10209
    },
    {
      "epoch": 0.1021,
      "grad_norm": 0.738960921368254,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 10210
    },
    {
      "epoch": 0.10211,
      "grad_norm": 0.9042302755628594,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 10211
    },
    {
      "epoch": 0.10212,
      "grad_norm": 0.9328835687364458,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 10212
    },
    {
      "epoch": 0.10213,
      "grad_norm": 1.064128555829592,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 10213
    },
    {
      "epoch": 0.10214,
      "grad_norm": 0.8423985032180068,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 10214
    },
    {
      "epoch": 0.10215,
      "grad_norm": 0.6894767482193619,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 10215
    },
    {
      "epoch": 0.10216,
      "grad_norm": 0.6407174824026426,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 10216
    },
    {
      "epoch": 0.10217,
      "grad_norm": 0.6930631872988513,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 10217
    },
    {
      "epoch": 0.10218,
      "grad_norm": 0.6736003865724515,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 10218
    },
    {
      "epoch": 0.10219,
      "grad_norm": 0.6391476717729888,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 10219
    },
    {
      "epoch": 0.1022,
      "grad_norm": 0.626075339436956,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 10220
    },
    {
      "epoch": 0.10221,
      "grad_norm": 0.6109786482845987,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 10221
    },
    {
      "epoch": 0.10222,
      "grad_norm": 0.6282338053842801,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 10222
    },
    {
      "epoch": 0.10223,
      "grad_norm": 0.6832246627842549,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 10223
    },
    {
      "epoch": 0.10224,
      "grad_norm": 0.7169113704064414,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 10224
    },
    {
      "epoch": 0.10225,
      "grad_norm": 0.7323081942187616,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 10225
    },
    {
      "epoch": 0.10226,
      "grad_norm": 0.7406008465439835,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 10226
    },
    {
      "epoch": 0.10227,
      "grad_norm": 0.785218333980764,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 10227
    },
    {
      "epoch": 0.10228,
      "grad_norm": 0.9898486890834313,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 10228
    },
    {
      "epoch": 0.10229,
      "grad_norm": 1.0742073532465095,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 10229
    },
    {
      "epoch": 0.1023,
      "grad_norm": 0.7265290601522981,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 10230
    },
    {
      "epoch": 0.10231,
      "grad_norm": 0.7367427007561755,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 10231
    },
    {
      "epoch": 0.10232,
      "grad_norm": 0.7481518210705496,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 10232
    },
    {
      "epoch": 0.10233,
      "grad_norm": 0.8005412014419764,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 10233
    },
    {
      "epoch": 0.10234,
      "grad_norm": 0.8303364545297103,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 10234
    },
    {
      "epoch": 0.10235,
      "grad_norm": 0.8310317993635913,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 10235
    },
    {
      "epoch": 0.10236,
      "grad_norm": 0.9054391176791641,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 10236
    },
    {
      "epoch": 0.10237,
      "grad_norm": 0.8435763363330704,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 10237
    },
    {
      "epoch": 0.10238,
      "grad_norm": 0.8115475750168918,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 10238
    },
    {
      "epoch": 0.10239,
      "grad_norm": 0.8628909482258778,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 10239
    },
    {
      "epoch": 0.1024,
      "grad_norm": 0.9364777145567061,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 10240
    },
    {
      "epoch": 0.10241,
      "grad_norm": 1.146465968559167,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 10241
    },
    {
      "epoch": 0.10242,
      "grad_norm": 0.854898755875994,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 10242
    },
    {
      "epoch": 0.10243,
      "grad_norm": 0.7511718291044952,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 10243
    },
    {
      "epoch": 0.10244,
      "grad_norm": 0.7603646485366432,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 10244
    },
    {
      "epoch": 0.10245,
      "grad_norm": 0.8105837774316208,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 10245
    },
    {
      "epoch": 0.10246,
      "grad_norm": 0.7977585495767636,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 10246
    },
    {
      "epoch": 0.10247,
      "grad_norm": 0.7506978997504432,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 10247
    },
    {
      "epoch": 0.10248,
      "grad_norm": 0.8759386294051426,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 10248
    },
    {
      "epoch": 0.10249,
      "grad_norm": 0.8431799228218545,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 10249
    },
    {
      "epoch": 0.1025,
      "grad_norm": 0.7864842623687947,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 10250
    },
    {
      "epoch": 0.10251,
      "grad_norm": 0.7740703845382346,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 10251
    },
    {
      "epoch": 0.10252,
      "grad_norm": 0.8321898093447786,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 10252
    },
    {
      "epoch": 0.10253,
      "grad_norm": 0.7894410369300876,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 10253
    },
    {
      "epoch": 0.10254,
      "grad_norm": 0.8133697650165276,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 10254
    },
    {
      "epoch": 0.10255,
      "grad_norm": 0.744829951039707,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 10255
    },
    {
      "epoch": 0.10256,
      "grad_norm": 0.7780534762171264,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 10256
    },
    {
      "epoch": 0.10257,
      "grad_norm": 0.8808046453133321,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 10257
    },
    {
      "epoch": 0.10258,
      "grad_norm": 0.9129780513694674,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 10258
    },
    {
      "epoch": 0.10259,
      "grad_norm": 0.8386788452118967,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 10259
    },
    {
      "epoch": 0.1026,
      "grad_norm": 0.7688367051425977,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 10260
    },
    {
      "epoch": 0.10261,
      "grad_norm": 0.6828172453485549,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 10261
    },
    {
      "epoch": 0.10262,
      "grad_norm": 0.6027470278842928,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 10262
    },
    {
      "epoch": 0.10263,
      "grad_norm": 0.6134625435010462,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 10263
    },
    {
      "epoch": 0.10264,
      "grad_norm": 0.6601613262512446,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 10264
    },
    {
      "epoch": 0.10265,
      "grad_norm": 0.6925742701606704,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 10265
    },
    {
      "epoch": 0.10266,
      "grad_norm": 0.7494871872755563,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 10266
    },
    {
      "epoch": 0.10267,
      "grad_norm": 0.7626957921447522,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 10267
    },
    {
      "epoch": 0.10268,
      "grad_norm": 0.8706405052637655,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 10268
    },
    {
      "epoch": 0.10269,
      "grad_norm": 0.9926896776195115,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 10269
    },
    {
      "epoch": 0.1027,
      "grad_norm": 0.9916626731747161,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 10270
    },
    {
      "epoch": 0.10271,
      "grad_norm": 0.96407920565692,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 10271
    },
    {
      "epoch": 0.10272,
      "grad_norm": 0.8960155500010829,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 10272
    },
    {
      "epoch": 0.10273,
      "grad_norm": 0.7192574545668858,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 10273
    },
    {
      "epoch": 0.10274,
      "grad_norm": 0.7829507156964814,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 10274
    },
    {
      "epoch": 0.10275,
      "grad_norm": 0.8067743579386221,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 10275
    },
    {
      "epoch": 0.10276,
      "grad_norm": 0.9012713038007449,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 10276
    },
    {
      "epoch": 0.10277,
      "grad_norm": 1.0017053952207466,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 10277
    },
    {
      "epoch": 0.10278,
      "grad_norm": 1.1196366403325528,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 10278
    },
    {
      "epoch": 0.10279,
      "grad_norm": 0.7036241002443819,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 10279
    },
    {
      "epoch": 0.1028,
      "grad_norm": 0.675873333438298,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 10280
    },
    {
      "epoch": 0.10281,
      "grad_norm": 0.7188694294424975,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 10281
    },
    {
      "epoch": 0.10282,
      "grad_norm": 0.6458206776287501,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 10282
    },
    {
      "epoch": 0.10283,
      "grad_norm": 0.6503790584246892,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 10283
    },
    {
      "epoch": 0.10284,
      "grad_norm": 0.7512150394872587,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 10284
    },
    {
      "epoch": 0.10285,
      "grad_norm": 0.8041418564413038,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 10285
    },
    {
      "epoch": 0.10286,
      "grad_norm": 0.7393185584957023,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 10286
    },
    {
      "epoch": 0.10287,
      "grad_norm": 0.6536497052383766,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 10287
    },
    {
      "epoch": 0.10288,
      "grad_norm": 0.6895525064317385,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 10288
    },
    {
      "epoch": 0.10289,
      "grad_norm": 0.71141474436633,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 10289
    },
    {
      "epoch": 0.1029,
      "grad_norm": 0.7863696654596326,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 10290
    },
    {
      "epoch": 0.10291,
      "grad_norm": 0.7433309300210272,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 10291
    },
    {
      "epoch": 0.10292,
      "grad_norm": 0.7228574492515069,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 10292
    },
    {
      "epoch": 0.10293,
      "grad_norm": 0.7328543775889537,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 10293
    },
    {
      "epoch": 0.10294,
      "grad_norm": 0.7119459365331811,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 10294
    },
    {
      "epoch": 0.10295,
      "grad_norm": 0.6902154836264375,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 10295
    },
    {
      "epoch": 0.10296,
      "grad_norm": 0.6673321189400658,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 10296
    },
    {
      "epoch": 0.10297,
      "grad_norm": 0.6206236886964629,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 10297
    },
    {
      "epoch": 0.10298,
      "grad_norm": 0.580549324398935,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 10298
    },
    {
      "epoch": 0.10299,
      "grad_norm": 0.5568108267250715,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 10299
    },
    {
      "epoch": 0.103,
      "grad_norm": 0.6459850251191824,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 10300
    },
    {
      "epoch": 0.10301,
      "grad_norm": 0.7870413569880143,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 10301
    },
    {
      "epoch": 0.10302,
      "grad_norm": 0.9424107245640844,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 10302
    },
    {
      "epoch": 0.10303,
      "grad_norm": 1.0280411938586382,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 10303
    },
    {
      "epoch": 0.10304,
      "grad_norm": 1.1122147913714624,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 10304
    },
    {
      "epoch": 0.10305,
      "grad_norm": 0.935437875996847,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 10305
    },
    {
      "epoch": 0.10306,
      "grad_norm": 0.9434216723466147,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 10306
    },
    {
      "epoch": 0.10307,
      "grad_norm": 1.077419206259069,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 10307
    },
    {
      "epoch": 0.10308,
      "grad_norm": 0.9260799956690462,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 10308
    },
    {
      "epoch": 0.10309,
      "grad_norm": 0.788475483287031,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 10309
    },
    {
      "epoch": 0.1031,
      "grad_norm": 0.7548519322564272,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 10310
    },
    {
      "epoch": 0.10311,
      "grad_norm": 0.8490099710145201,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 10311
    },
    {
      "epoch": 0.10312,
      "grad_norm": 0.9754889415393432,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 10312
    },
    {
      "epoch": 0.10313,
      "grad_norm": 0.9725068876707715,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 10313
    },
    {
      "epoch": 0.10314,
      "grad_norm": 1.008302977550403,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 10314
    },
    {
      "epoch": 0.10315,
      "grad_norm": 1.1068450571626192,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 10315
    },
    {
      "epoch": 0.10316,
      "grad_norm": 1.01689260859524,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 10316
    },
    {
      "epoch": 0.10317,
      "grad_norm": 1.0399611525497334,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 10317
    },
    {
      "epoch": 0.10318,
      "grad_norm": 1.0664621024838172,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 10318
    },
    {
      "epoch": 0.10319,
      "grad_norm": 0.8887351487113996,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 10319
    },
    {
      "epoch": 0.1032,
      "grad_norm": 0.9777260247175278,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 10320
    },
    {
      "epoch": 0.10321,
      "grad_norm": 1.0855049175533027,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 10321
    },
    {
      "epoch": 0.10322,
      "grad_norm": 0.963506360245415,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 10322
    },
    {
      "epoch": 0.10323,
      "grad_norm": 0.9881045744868919,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 10323
    },
    {
      "epoch": 0.10324,
      "grad_norm": 0.8742451143437673,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 10324
    },
    {
      "epoch": 0.10325,
      "grad_norm": 0.7592965201742712,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 10325
    },
    {
      "epoch": 0.10326,
      "grad_norm": 0.6924762638567085,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 10326
    },
    {
      "epoch": 0.10327,
      "grad_norm": 0.7996862828960193,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 10327
    },
    {
      "epoch": 0.10328,
      "grad_norm": 0.8472096443774338,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 10328
    },
    {
      "epoch": 0.10329,
      "grad_norm": 0.80863735984161,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 10329
    },
    {
      "epoch": 0.1033,
      "grad_norm": 0.7470728233214627,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 10330
    },
    {
      "epoch": 0.10331,
      "grad_norm": 0.6111385808771548,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 10331
    },
    {
      "epoch": 0.10332,
      "grad_norm": 0.5181245173636219,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 10332
    },
    {
      "epoch": 0.10333,
      "grad_norm": 0.5288066580362742,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 10333
    },
    {
      "epoch": 0.10334,
      "grad_norm": 0.5522664037256619,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 10334
    },
    {
      "epoch": 0.10335,
      "grad_norm": 0.6137655038412625,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 10335
    },
    {
      "epoch": 0.10336,
      "grad_norm": 0.7105459077980683,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 10336
    },
    {
      "epoch": 0.10337,
      "grad_norm": 0.8033414113689806,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 10337
    },
    {
      "epoch": 0.10338,
      "grad_norm": 0.76450882628852,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 10338
    },
    {
      "epoch": 0.10339,
      "grad_norm": 0.6766193484175521,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 10339
    },
    {
      "epoch": 0.1034,
      "grad_norm": 0.7046081840372489,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 10340
    },
    {
      "epoch": 0.10341,
      "grad_norm": 0.6881196394385097,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 10341
    },
    {
      "epoch": 0.10342,
      "grad_norm": 0.7019620860313454,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 10342
    },
    {
      "epoch": 0.10343,
      "grad_norm": 0.7669457179418919,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 10343
    },
    {
      "epoch": 0.10344,
      "grad_norm": 0.8562986464162007,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 10344
    },
    {
      "epoch": 0.10345,
      "grad_norm": 0.9678576326676694,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 10345
    },
    {
      "epoch": 0.10346,
      "grad_norm": 0.988840553225034,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 10346
    },
    {
      "epoch": 0.10347,
      "grad_norm": 0.8761665279793409,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 10347
    },
    {
      "epoch": 0.10348,
      "grad_norm": 0.8070465433399605,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 10348
    },
    {
      "epoch": 0.10349,
      "grad_norm": 0.746677695053773,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 10349
    },
    {
      "epoch": 0.1035,
      "grad_norm": 0.7251075414951131,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 10350
    },
    {
      "epoch": 0.10351,
      "grad_norm": 0.682523058274381,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 10351
    },
    {
      "epoch": 0.10352,
      "grad_norm": 0.5549826583410974,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 10352
    },
    {
      "epoch": 0.10353,
      "grad_norm": 0.6184831584530538,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 10353
    },
    {
      "epoch": 0.10354,
      "grad_norm": 0.6154810565062979,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 10354
    },
    {
      "epoch": 0.10355,
      "grad_norm": 0.6013274566749466,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 10355
    },
    {
      "epoch": 0.10356,
      "grad_norm": 0.5790494966726321,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 10356
    },
    {
      "epoch": 0.10357,
      "grad_norm": 0.5815749905417382,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 10357
    },
    {
      "epoch": 0.10358,
      "grad_norm": 0.5961417322223677,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 10358
    },
    {
      "epoch": 0.10359,
      "grad_norm": 0.612228084547915,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 10359
    },
    {
      "epoch": 0.1036,
      "grad_norm": 0.5632606844307874,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 10360
    },
    {
      "epoch": 0.10361,
      "grad_norm": 0.5944178213854017,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 10361
    },
    {
      "epoch": 0.10362,
      "grad_norm": 0.6050895286964695,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 10362
    },
    {
      "epoch": 0.10363,
      "grad_norm": 0.5893502071647355,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 10363
    },
    {
      "epoch": 0.10364,
      "grad_norm": 0.6480921596779424,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 10364
    },
    {
      "epoch": 0.10365,
      "grad_norm": 0.6655516728542986,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 10365
    },
    {
      "epoch": 0.10366,
      "grad_norm": 0.841939927056256,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 10366
    },
    {
      "epoch": 0.10367,
      "grad_norm": 1.2925940203305464,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 10367
    },
    {
      "epoch": 0.10368,
      "grad_norm": 1.0721423681105018,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 10368
    },
    {
      "epoch": 0.10369,
      "grad_norm": 0.8054815871522366,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 10369
    },
    {
      "epoch": 0.1037,
      "grad_norm": 0.6431254709101457,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 10370
    },
    {
      "epoch": 0.10371,
      "grad_norm": 0.6771185643178579,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 10371
    },
    {
      "epoch": 0.10372,
      "grad_norm": 0.6505722328550112,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 10372
    },
    {
      "epoch": 0.10373,
      "grad_norm": 0.649150897107484,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 10373
    },
    {
      "epoch": 0.10374,
      "grad_norm": 0.7655192327120619,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 10374
    },
    {
      "epoch": 0.10375,
      "grad_norm": 0.8276458451965146,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 10375
    },
    {
      "epoch": 0.10376,
      "grad_norm": 0.8401543111084332,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 10376
    },
    {
      "epoch": 0.10377,
      "grad_norm": 0.961981488825057,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 10377
    },
    {
      "epoch": 0.10378,
      "grad_norm": 1.0528802151842687,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 10378
    },
    {
      "epoch": 0.10379,
      "grad_norm": 0.9122713415226914,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 10379
    },
    {
      "epoch": 0.1038,
      "grad_norm": 0.8191396318548235,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 10380
    },
    {
      "epoch": 0.10381,
      "grad_norm": 0.7785335876041576,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 10381
    },
    {
      "epoch": 0.10382,
      "grad_norm": 0.8640344513391571,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 10382
    },
    {
      "epoch": 0.10383,
      "grad_norm": 1.0100767424979953,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 10383
    },
    {
      "epoch": 0.10384,
      "grad_norm": 1.0586243617599906,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 10384
    },
    {
      "epoch": 0.10385,
      "grad_norm": 0.8624828769831347,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 10385
    },
    {
      "epoch": 0.10386,
      "grad_norm": 0.8830596935983357,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 10386
    },
    {
      "epoch": 0.10387,
      "grad_norm": 0.8306326059978796,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 10387
    },
    {
      "epoch": 0.10388,
      "grad_norm": 0.8272101216281712,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 10388
    },
    {
      "epoch": 0.10389,
      "grad_norm": 0.9009021769238307,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 10389
    },
    {
      "epoch": 0.1039,
      "grad_norm": 1.0391034226876843,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 10390
    },
    {
      "epoch": 0.10391,
      "grad_norm": 0.9858486051647656,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 10391
    },
    {
      "epoch": 0.10392,
      "grad_norm": 0.817170907122026,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 10392
    },
    {
      "epoch": 0.10393,
      "grad_norm": 0.7813731942805362,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 10393
    },
    {
      "epoch": 0.10394,
      "grad_norm": 0.7491777949113574,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 10394
    },
    {
      "epoch": 0.10395,
      "grad_norm": 0.9168636740102267,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 10395
    },
    {
      "epoch": 0.10396,
      "grad_norm": 1.0881876341009609,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 10396
    },
    {
      "epoch": 0.10397,
      "grad_norm": 0.8483418535870134,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 10397
    },
    {
      "epoch": 0.10398,
      "grad_norm": 0.8004552390402355,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 10398
    },
    {
      "epoch": 0.10399,
      "grad_norm": 1.0550255797082628,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 10399
    },
    {
      "epoch": 0.104,
      "grad_norm": 1.1469901095195558,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 10400
    },
    {
      "epoch": 0.10401,
      "grad_norm": 0.776163770261994,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 10401
    },
    {
      "epoch": 0.10402,
      "grad_norm": 0.7191911818065254,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 10402
    },
    {
      "epoch": 0.10403,
      "grad_norm": 0.7757679538550868,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 10403
    },
    {
      "epoch": 0.10404,
      "grad_norm": 0.9017847369719004,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 10404
    },
    {
      "epoch": 0.10405,
      "grad_norm": 1.0205196229352549,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 10405
    },
    {
      "epoch": 0.10406,
      "grad_norm": 1.0561034007451116,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 10406
    },
    {
      "epoch": 0.10407,
      "grad_norm": 0.8382012374833445,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 10407
    },
    {
      "epoch": 0.10408,
      "grad_norm": 0.6440441204977746,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 10408
    },
    {
      "epoch": 0.10409,
      "grad_norm": 0.6231554547472592,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 10409
    },
    {
      "epoch": 0.1041,
      "grad_norm": 0.5895655256675674,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 10410
    },
    {
      "epoch": 0.10411,
      "grad_norm": 0.6078869262220562,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 10411
    },
    {
      "epoch": 0.10412,
      "grad_norm": 0.5714544500792731,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 10412
    },
    {
      "epoch": 0.10413,
      "grad_norm": 0.4888734408355746,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 10413
    },
    {
      "epoch": 0.10414,
      "grad_norm": 0.45177378648238387,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 10414
    },
    {
      "epoch": 0.10415,
      "grad_norm": 0.4464755195762157,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 10415
    },
    {
      "epoch": 0.10416,
      "grad_norm": 0.4119578434609014,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 10416
    },
    {
      "epoch": 0.10417,
      "grad_norm": 0.4530981054257146,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 10417
    },
    {
      "epoch": 0.10418,
      "grad_norm": 0.5625746495824158,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 10418
    },
    {
      "epoch": 0.10419,
      "grad_norm": 0.7189451960829651,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 10419
    },
    {
      "epoch": 0.1042,
      "grad_norm": 0.9831700376646376,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 10420
    },
    {
      "epoch": 0.10421,
      "grad_norm": 1.1445532836627261,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 10421
    },
    {
      "epoch": 0.10422,
      "grad_norm": 0.7082563767107952,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 10422
    },
    {
      "epoch": 0.10423,
      "grad_norm": 0.6874627848858307,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 10423
    },
    {
      "epoch": 0.10424,
      "grad_norm": 0.730815564511028,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 10424
    },
    {
      "epoch": 0.10425,
      "grad_norm": 0.6651685498466663,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 10425
    },
    {
      "epoch": 0.10426,
      "grad_norm": 0.629940706978873,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 10426
    },
    {
      "epoch": 0.10427,
      "grad_norm": 0.662314993288643,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 10427
    },
    {
      "epoch": 0.10428,
      "grad_norm": 0.7008715707921093,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 10428
    },
    {
      "epoch": 0.10429,
      "grad_norm": 0.774968411885718,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 10429
    },
    {
      "epoch": 0.1043,
      "grad_norm": 0.7183561816195599,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 10430
    },
    {
      "epoch": 0.10431,
      "grad_norm": 0.7603457171655063,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 10431
    },
    {
      "epoch": 0.10432,
      "grad_norm": 0.79045224783535,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 10432
    },
    {
      "epoch": 0.10433,
      "grad_norm": 0.9599162387244943,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 10433
    },
    {
      "epoch": 0.10434,
      "grad_norm": 0.8214963345150575,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 10434
    },
    {
      "epoch": 0.10435,
      "grad_norm": 0.7767373236471652,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 10435
    },
    {
      "epoch": 0.10436,
      "grad_norm": 0.7865430681912782,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 10436
    },
    {
      "epoch": 0.10437,
      "grad_norm": 0.7933364160947659,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 10437
    },
    {
      "epoch": 0.10438,
      "grad_norm": 0.7277652813568619,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 10438
    },
    {
      "epoch": 0.10439,
      "grad_norm": 0.6839207059096676,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 10439
    },
    {
      "epoch": 0.1044,
      "grad_norm": 0.6746530544011542,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 10440
    },
    {
      "epoch": 0.10441,
      "grad_norm": 0.7864041796225456,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 10441
    },
    {
      "epoch": 0.10442,
      "grad_norm": 0.8804608511981953,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 10442
    },
    {
      "epoch": 0.10443,
      "grad_norm": 1.0508300123164835,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 10443
    },
    {
      "epoch": 0.10444,
      "grad_norm": 1.051811976800046,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 10444
    },
    {
      "epoch": 0.10445,
      "grad_norm": 0.7888503995506437,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 10445
    },
    {
      "epoch": 0.10446,
      "grad_norm": 0.6603911887857837,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 10446
    },
    {
      "epoch": 0.10447,
      "grad_norm": 0.8408603404059548,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 10447
    },
    {
      "epoch": 0.10448,
      "grad_norm": 0.9695074227496694,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 10448
    },
    {
      "epoch": 0.10449,
      "grad_norm": 0.9822010322827212,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 10449
    },
    {
      "epoch": 0.1045,
      "grad_norm": 1.0012725608752013,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 10450
    },
    {
      "epoch": 0.10451,
      "grad_norm": 1.009392360148877,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 10451
    },
    {
      "epoch": 0.10452,
      "grad_norm": 1.0090387388437994,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 10452
    },
    {
      "epoch": 0.10453,
      "grad_norm": 0.8943599349231653,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 10453
    },
    {
      "epoch": 0.10454,
      "grad_norm": 0.8560707064477248,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 10454
    },
    {
      "epoch": 0.10455,
      "grad_norm": 0.8752369980580909,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 10455
    },
    {
      "epoch": 0.10456,
      "grad_norm": 0.9130498227436277,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 10456
    },
    {
      "epoch": 0.10457,
      "grad_norm": 0.8067284336559211,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 10457
    },
    {
      "epoch": 0.10458,
      "grad_norm": 0.7395291417558032,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 10458
    },
    {
      "epoch": 0.10459,
      "grad_norm": 0.7297382370955492,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 10459
    },
    {
      "epoch": 0.1046,
      "grad_norm": 0.8594227462287262,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 10460
    },
    {
      "epoch": 0.10461,
      "grad_norm": 0.8861318685991226,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 10461
    },
    {
      "epoch": 0.10462,
      "grad_norm": 0.8083739717834618,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 10462
    },
    {
      "epoch": 0.10463,
      "grad_norm": 0.855447088915927,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 10463
    },
    {
      "epoch": 0.10464,
      "grad_norm": 0.8309265331948995,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 10464
    },
    {
      "epoch": 0.10465,
      "grad_norm": 0.8950247767863003,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 10465
    },
    {
      "epoch": 0.10466,
      "grad_norm": 0.9702711674796212,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 10466
    },
    {
      "epoch": 0.10467,
      "grad_norm": 1.0808015279280165,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 10467
    },
    {
      "epoch": 0.10468,
      "grad_norm": 0.8796165704446247,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 10468
    },
    {
      "epoch": 0.10469,
      "grad_norm": 0.6864191337528118,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 10469
    },
    {
      "epoch": 0.1047,
      "grad_norm": 0.642911293497494,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 10470
    },
    {
      "epoch": 0.10471,
      "grad_norm": 0.6401286669093053,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 10471
    },
    {
      "epoch": 0.10472,
      "grad_norm": 0.7623940021861698,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 10472
    },
    {
      "epoch": 0.10473,
      "grad_norm": 0.7757677108816563,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 10473
    },
    {
      "epoch": 0.10474,
      "grad_norm": 0.8165254615464338,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 10474
    },
    {
      "epoch": 0.10475,
      "grad_norm": 0.6947649086407327,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 10475
    },
    {
      "epoch": 0.10476,
      "grad_norm": 0.609454645718345,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 10476
    },
    {
      "epoch": 0.10477,
      "grad_norm": 0.5156181671942854,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 10477
    },
    {
      "epoch": 0.10478,
      "grad_norm": 0.550404325359368,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 10478
    },
    {
      "epoch": 0.10479,
      "grad_norm": 0.5844190234215697,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 10479
    },
    {
      "epoch": 0.1048,
      "grad_norm": 0.6475683647990985,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 10480
    },
    {
      "epoch": 0.10481,
      "grad_norm": 0.765493534126949,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 10481
    },
    {
      "epoch": 0.10482,
      "grad_norm": 0.7460812863968737,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 10482
    },
    {
      "epoch": 0.10483,
      "grad_norm": 0.760367664758415,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 10483
    },
    {
      "epoch": 0.10484,
      "grad_norm": 0.7889324440591884,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 10484
    },
    {
      "epoch": 0.10485,
      "grad_norm": 0.7000125978773567,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 10485
    },
    {
      "epoch": 0.10486,
      "grad_norm": 0.697394489855323,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 10486
    },
    {
      "epoch": 0.10487,
      "grad_norm": 0.7734234905452742,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 10487
    },
    {
      "epoch": 0.10488,
      "grad_norm": 0.9578169878035425,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 10488
    },
    {
      "epoch": 0.10489,
      "grad_norm": 1.1043883747627603,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 10489
    },
    {
      "epoch": 0.1049,
      "grad_norm": 0.8806546383022705,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 10490
    },
    {
      "epoch": 0.10491,
      "grad_norm": 0.8581015926193526,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 10491
    },
    {
      "epoch": 0.10492,
      "grad_norm": 0.8562928611424699,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 10492
    },
    {
      "epoch": 0.10493,
      "grad_norm": 0.8058684115768378,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 10493
    },
    {
      "epoch": 0.10494,
      "grad_norm": 0.7770554096388587,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 10494
    },
    {
      "epoch": 0.10495,
      "grad_norm": 0.7334547382322241,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 10495
    },
    {
      "epoch": 0.10496,
      "grad_norm": 0.718416025393358,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 10496
    },
    {
      "epoch": 0.10497,
      "grad_norm": 0.7792435850503384,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 10497
    },
    {
      "epoch": 0.10498,
      "grad_norm": 0.8412922868706377,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 10498
    },
    {
      "epoch": 0.10499,
      "grad_norm": 0.9509337789006239,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 10499
    },
    {
      "epoch": 0.105,
      "grad_norm": 0.9912844508190685,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 10500
    },
    {
      "epoch": 0.10501,
      "grad_norm": 0.8580834861139071,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 10501
    },
    {
      "epoch": 0.10502,
      "grad_norm": 0.8167382602075896,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 10502
    },
    {
      "epoch": 0.10503,
      "grad_norm": 0.8743323585619591,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 10503
    },
    {
      "epoch": 0.10504,
      "grad_norm": 0.8327457707468423,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 10504
    },
    {
      "epoch": 0.10505,
      "grad_norm": 0.801391957128637,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 10505
    },
    {
      "epoch": 0.10506,
      "grad_norm": 0.6394241038195827,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 10506
    },
    {
      "epoch": 0.10507,
      "grad_norm": 0.6433059411947507,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 10507
    },
    {
      "epoch": 0.10508,
      "grad_norm": 0.6555621284685157,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 10508
    },
    {
      "epoch": 0.10509,
      "grad_norm": 0.6181176365447185,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 10509
    },
    {
      "epoch": 0.1051,
      "grad_norm": 0.65166605534626,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 10510
    },
    {
      "epoch": 0.10511,
      "grad_norm": 0.7356173809364867,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 10511
    },
    {
      "epoch": 0.10512,
      "grad_norm": 1.013079842957836,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 10512
    },
    {
      "epoch": 0.10513,
      "grad_norm": 1.1755143739073186,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 10513
    },
    {
      "epoch": 0.10514,
      "grad_norm": 0.8163557674329753,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 10514
    },
    {
      "epoch": 0.10515,
      "grad_norm": 0.7085350074934578,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 10515
    },
    {
      "epoch": 0.10516,
      "grad_norm": 0.6658791858624886,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 10516
    },
    {
      "epoch": 0.10517,
      "grad_norm": 0.6041515015315367,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 10517
    },
    {
      "epoch": 0.10518,
      "grad_norm": 0.7047204808794802,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 10518
    },
    {
      "epoch": 0.10519,
      "grad_norm": 0.7244412290591962,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 10519
    },
    {
      "epoch": 0.1052,
      "grad_norm": 0.7395861089702314,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 10520
    },
    {
      "epoch": 0.10521,
      "grad_norm": 0.7665016692200823,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 10521
    },
    {
      "epoch": 0.10522,
      "grad_norm": 0.7792629094064418,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 10522
    },
    {
      "epoch": 0.10523,
      "grad_norm": 0.7953210132825478,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 10523
    },
    {
      "epoch": 0.10524,
      "grad_norm": 0.7991555641259503,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 10524
    },
    {
      "epoch": 0.10525,
      "grad_norm": 0.8238402147042898,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 10525
    },
    {
      "epoch": 0.10526,
      "grad_norm": 0.7163491665622135,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 10526
    },
    {
      "epoch": 0.10527,
      "grad_norm": 0.6545420151814723,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 10527
    },
    {
      "epoch": 0.10528,
      "grad_norm": 0.7166141333279368,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 10528
    },
    {
      "epoch": 0.10529,
      "grad_norm": 0.9100729617540565,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 10529
    },
    {
      "epoch": 0.1053,
      "grad_norm": 1.0028018301025132,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 10530
    },
    {
      "epoch": 0.10531,
      "grad_norm": 1.0807675817900035,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 10531
    },
    {
      "epoch": 0.10532,
      "grad_norm": 0.8870026361329219,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 10532
    },
    {
      "epoch": 0.10533,
      "grad_norm": 0.893246559986549,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 10533
    },
    {
      "epoch": 0.10534,
      "grad_norm": 0.8877419641098879,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 10534
    },
    {
      "epoch": 0.10535,
      "grad_norm": 0.804120621184703,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 10535
    },
    {
      "epoch": 0.10536,
      "grad_norm": 0.731737971690129,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 10536
    },
    {
      "epoch": 0.10537,
      "grad_norm": 0.6685167604634952,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 10537
    },
    {
      "epoch": 0.10538,
      "grad_norm": 0.6041586651576828,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 10538
    },
    {
      "epoch": 0.10539,
      "grad_norm": 0.5722575391865711,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 10539
    },
    {
      "epoch": 0.1054,
      "grad_norm": 0.6001113282044517,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 10540
    },
    {
      "epoch": 0.10541,
      "grad_norm": 0.5762315876118429,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 10541
    },
    {
      "epoch": 0.10542,
      "grad_norm": 0.563179289202697,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 10542
    },
    {
      "epoch": 0.10543,
      "grad_norm": 0.5947299652169051,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 10543
    },
    {
      "epoch": 0.10544,
      "grad_norm": 0.6985463762361143,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 10544
    },
    {
      "epoch": 0.10545,
      "grad_norm": 0.8488458192713703,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 10545
    },
    {
      "epoch": 0.10546,
      "grad_norm": 1.134283035666868,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 10546
    },
    {
      "epoch": 0.10547,
      "grad_norm": 1.008621876461993,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 10547
    },
    {
      "epoch": 0.10548,
      "grad_norm": 0.8910932087633671,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 10548
    },
    {
      "epoch": 0.10549,
      "grad_norm": 0.809160397609863,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 10549
    },
    {
      "epoch": 0.1055,
      "grad_norm": 0.8218623812846083,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 10550
    },
    {
      "epoch": 0.10551,
      "grad_norm": 0.7992528000231008,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 10551
    },
    {
      "epoch": 0.10552,
      "grad_norm": 0.86719344382475,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 10552
    },
    {
      "epoch": 0.10553,
      "grad_norm": 0.908060487944853,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 10553
    },
    {
      "epoch": 0.10554,
      "grad_norm": 0.8652695049802602,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 10554
    },
    {
      "epoch": 0.10555,
      "grad_norm": 0.6860164989967895,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 10555
    },
    {
      "epoch": 0.10556,
      "grad_norm": 0.6378293392988055,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 10556
    },
    {
      "epoch": 0.10557,
      "grad_norm": 0.62800965514661,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 10557
    },
    {
      "epoch": 0.10558,
      "grad_norm": 0.6963803928423222,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 10558
    },
    {
      "epoch": 0.10559,
      "grad_norm": 0.6865270024171043,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 10559
    },
    {
      "epoch": 0.1056,
      "grad_norm": 0.737499509213761,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 10560
    },
    {
      "epoch": 0.10561,
      "grad_norm": 0.8984522058594936,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 10561
    },
    {
      "epoch": 0.10562,
      "grad_norm": 1.1339394563977538,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 10562
    },
    {
      "epoch": 0.10563,
      "grad_norm": 0.8198771181383394,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 10563
    },
    {
      "epoch": 0.10564,
      "grad_norm": 0.7658877274749886,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 10564
    },
    {
      "epoch": 0.10565,
      "grad_norm": 0.8482606637522958,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 10565
    },
    {
      "epoch": 0.10566,
      "grad_norm": 0.8347786010621645,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 10566
    },
    {
      "epoch": 0.10567,
      "grad_norm": 0.8843860041980399,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 10567
    },
    {
      "epoch": 0.10568,
      "grad_norm": 0.990139802584295,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 10568
    },
    {
      "epoch": 0.10569,
      "grad_norm": 0.9852662752867085,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 10569
    },
    {
      "epoch": 0.1057,
      "grad_norm": 0.9372987172274819,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 10570
    },
    {
      "epoch": 0.10571,
      "grad_norm": 0.988820397043408,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 10571
    },
    {
      "epoch": 0.10572,
      "grad_norm": 1.1678833347420248,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 10572
    },
    {
      "epoch": 0.10573,
      "grad_norm": 0.8517913634074189,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 10573
    },
    {
      "epoch": 0.10574,
      "grad_norm": 0.8659007094767834,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 10574
    },
    {
      "epoch": 0.10575,
      "grad_norm": 0.958183815886997,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 10575
    },
    {
      "epoch": 0.10576,
      "grad_norm": 0.9118557877430638,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 10576
    },
    {
      "epoch": 0.10577,
      "grad_norm": 0.9153226156451957,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 10577
    },
    {
      "epoch": 0.10578,
      "grad_norm": 0.9874735654072453,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 10578
    },
    {
      "epoch": 0.10579,
      "grad_norm": 1.0415079742593492,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 10579
    },
    {
      "epoch": 0.1058,
      "grad_norm": 0.9153784660111088,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 10580
    },
    {
      "epoch": 0.10581,
      "grad_norm": 0.9210262014080384,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 10581
    },
    {
      "epoch": 0.10582,
      "grad_norm": 0.9396705464198691,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 10582
    },
    {
      "epoch": 0.10583,
      "grad_norm": 0.9120879354056952,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 10583
    },
    {
      "epoch": 0.10584,
      "grad_norm": 0.8070610834160225,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 10584
    },
    {
      "epoch": 0.10585,
      "grad_norm": 0.8983503783260497,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 10585
    },
    {
      "epoch": 0.10586,
      "grad_norm": 0.942723580765965,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 10586
    },
    {
      "epoch": 0.10587,
      "grad_norm": 0.9654495812888432,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 10587
    },
    {
      "epoch": 0.10588,
      "grad_norm": 1.0137540717282463,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 10588
    },
    {
      "epoch": 0.10589,
      "grad_norm": 0.9211083933242845,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 10589
    },
    {
      "epoch": 0.1059,
      "grad_norm": 0.8886434692979224,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 10590
    },
    {
      "epoch": 0.10591,
      "grad_norm": 0.8414672791491735,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 10591
    },
    {
      "epoch": 0.10592,
      "grad_norm": 0.808437412569294,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 10592
    },
    {
      "epoch": 0.10593,
      "grad_norm": 0.6746674356577256,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 10593
    },
    {
      "epoch": 0.10594,
      "grad_norm": 0.6222040276891654,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 10594
    },
    {
      "epoch": 0.10595,
      "grad_norm": 0.6313983056054177,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 10595
    },
    {
      "epoch": 0.10596,
      "grad_norm": 0.6792929100891301,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 10596
    },
    {
      "epoch": 0.10597,
      "grad_norm": 0.6800719059268772,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 10597
    },
    {
      "epoch": 0.10598,
      "grad_norm": 0.770312479924072,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 10598
    },
    {
      "epoch": 0.10599,
      "grad_norm": 0.9764320147050246,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 10599
    },
    {
      "epoch": 0.106,
      "grad_norm": 0.9114216015686166,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 10600
    },
    {
      "epoch": 0.10601,
      "grad_norm": 0.6555189332915664,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 10601
    },
    {
      "epoch": 0.10602,
      "grad_norm": 0.5040307133652577,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 10602
    },
    {
      "epoch": 0.10603,
      "grad_norm": 0.5690484358230466,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 10603
    },
    {
      "epoch": 0.10604,
      "grad_norm": 0.5637735595856799,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 10604
    },
    {
      "epoch": 0.10605,
      "grad_norm": 0.48001727986025794,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 10605
    },
    {
      "epoch": 0.10606,
      "grad_norm": 0.4919296874299588,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 10606
    },
    {
      "epoch": 0.10607,
      "grad_norm": 0.4727741741475219,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 10607
    },
    {
      "epoch": 0.10608,
      "grad_norm": 0.5184442406668994,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 10608
    },
    {
      "epoch": 0.10609,
      "grad_norm": 0.5858630638238149,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 10609
    },
    {
      "epoch": 0.1061,
      "grad_norm": 0.6874353934764682,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 10610
    },
    {
      "epoch": 0.10611,
      "grad_norm": 0.8263741726587102,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 10611
    },
    {
      "epoch": 0.10612,
      "grad_norm": 0.8884315684622063,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 10612
    },
    {
      "epoch": 0.10613,
      "grad_norm": 0.8763453508937509,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 10613
    },
    {
      "epoch": 0.10614,
      "grad_norm": 0.8224298838859537,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 10614
    },
    {
      "epoch": 0.10615,
      "grad_norm": 0.7191046159980217,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 10615
    },
    {
      "epoch": 0.10616,
      "grad_norm": 0.7315824929912534,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 10616
    },
    {
      "epoch": 0.10617,
      "grad_norm": 0.916956377366291,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 10617
    },
    {
      "epoch": 0.10618,
      "grad_norm": 1.2629005890508869,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 10618
    },
    {
      "epoch": 0.10619,
      "grad_norm": 0.856578006061621,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 10619
    },
    {
      "epoch": 0.1062,
      "grad_norm": 0.7408131912141622,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 10620
    },
    {
      "epoch": 0.10621,
      "grad_norm": 0.6470727798627832,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 10621
    },
    {
      "epoch": 0.10622,
      "grad_norm": 0.5725004840385046,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 10622
    },
    {
      "epoch": 0.10623,
      "grad_norm": 0.553432799759805,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 10623
    },
    {
      "epoch": 0.10624,
      "grad_norm": 0.5403048416895829,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 10624
    },
    {
      "epoch": 0.10625,
      "grad_norm": 0.5104072292295226,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 10625
    },
    {
      "epoch": 0.10626,
      "grad_norm": 0.5823780352699905,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 10626
    },
    {
      "epoch": 0.10627,
      "grad_norm": 0.6962228994056245,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 10627
    },
    {
      "epoch": 0.10628,
      "grad_norm": 0.7935604816074494,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 10628
    },
    {
      "epoch": 0.10629,
      "grad_norm": 0.9695291887403658,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 10629
    },
    {
      "epoch": 0.1063,
      "grad_norm": 1.2894223340608182,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 10630
    },
    {
      "epoch": 0.10631,
      "grad_norm": 0.6621419698664075,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 10631
    },
    {
      "epoch": 0.10632,
      "grad_norm": 0.8074981429308187,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 10632
    },
    {
      "epoch": 0.10633,
      "grad_norm": 0.9026339646908454,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 10633
    },
    {
      "epoch": 0.10634,
      "grad_norm": 0.8417206941624089,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 10634
    },
    {
      "epoch": 0.10635,
      "grad_norm": 0.8020780791146946,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 10635
    },
    {
      "epoch": 0.10636,
      "grad_norm": 0.8075945654041848,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 10636
    },
    {
      "epoch": 0.10637,
      "grad_norm": 0.7764114257102779,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 10637
    },
    {
      "epoch": 0.10638,
      "grad_norm": 0.8095214640783682,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 10638
    },
    {
      "epoch": 0.10639,
      "grad_norm": 0.8784677433844998,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 10639
    },
    {
      "epoch": 0.1064,
      "grad_norm": 0.8355273252342663,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 10640
    },
    {
      "epoch": 0.10641,
      "grad_norm": 0.9458422852748879,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 10641
    },
    {
      "epoch": 0.10642,
      "grad_norm": 1.068264830551457,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 10642
    },
    {
      "epoch": 0.10643,
      "grad_norm": 1.0523298329122281,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 10643
    },
    {
      "epoch": 0.10644,
      "grad_norm": 0.7659982628957405,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 10644
    },
    {
      "epoch": 0.10645,
      "grad_norm": 0.652728924657321,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 10645
    },
    {
      "epoch": 0.10646,
      "grad_norm": 0.6456066494246264,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 10646
    },
    {
      "epoch": 0.10647,
      "grad_norm": 0.7464907408983558,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 10647
    },
    {
      "epoch": 0.10648,
      "grad_norm": 0.7405212525064351,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 10648
    },
    {
      "epoch": 0.10649,
      "grad_norm": 0.7478639631456179,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 10649
    },
    {
      "epoch": 0.1065,
      "grad_norm": 0.7550955604366715,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 10650
    },
    {
      "epoch": 0.10651,
      "grad_norm": 0.7110846298696153,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 10651
    },
    {
      "epoch": 0.10652,
      "grad_norm": 0.6402684922431795,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 10652
    },
    {
      "epoch": 0.10653,
      "grad_norm": 0.59791029327084,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 10653
    },
    {
      "epoch": 0.10654,
      "grad_norm": 0.6432662968161444,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 10654
    },
    {
      "epoch": 0.10655,
      "grad_norm": 0.6752567227234482,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 10655
    },
    {
      "epoch": 0.10656,
      "grad_norm": 0.6800244754630034,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 10656
    },
    {
      "epoch": 0.10657,
      "grad_norm": 0.5959513716804604,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 10657
    },
    {
      "epoch": 0.10658,
      "grad_norm": 0.6403859324045968,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 10658
    },
    {
      "epoch": 0.10659,
      "grad_norm": 0.8296924742157811,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 10659
    },
    {
      "epoch": 0.1066,
      "grad_norm": 1.0682115948141957,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 10660
    },
    {
      "epoch": 0.10661,
      "grad_norm": 1.1194993070021568,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 10661
    },
    {
      "epoch": 0.10662,
      "grad_norm": 0.9275062203048765,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 10662
    },
    {
      "epoch": 0.10663,
      "grad_norm": 0.8557195585149507,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 10663
    },
    {
      "epoch": 0.10664,
      "grad_norm": 0.7332891166349356,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 10664
    },
    {
      "epoch": 0.10665,
      "grad_norm": 0.7737620390856565,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 10665
    },
    {
      "epoch": 0.10666,
      "grad_norm": 0.8490683224003708,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 10666
    },
    {
      "epoch": 0.10667,
      "grad_norm": 0.8834112771739413,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 10667
    },
    {
      "epoch": 0.10668,
      "grad_norm": 0.9844633934429818,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 10668
    },
    {
      "epoch": 0.10669,
      "grad_norm": 0.932517734016524,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 10669
    },
    {
      "epoch": 0.1067,
      "grad_norm": 0.83578599313614,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 10670
    },
    {
      "epoch": 0.10671,
      "grad_norm": 0.9069085658024532,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 10671
    },
    {
      "epoch": 0.10672,
      "grad_norm": 0.9945862570463592,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 10672
    },
    {
      "epoch": 0.10673,
      "grad_norm": 1.09965107012104,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 10673
    },
    {
      "epoch": 0.10674,
      "grad_norm": 0.8268182899214279,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 10674
    },
    {
      "epoch": 0.10675,
      "grad_norm": 0.7034549781067129,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 10675
    },
    {
      "epoch": 0.10676,
      "grad_norm": 0.7890775923714155,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 10676
    },
    {
      "epoch": 0.10677,
      "grad_norm": 0.7900757896942041,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 10677
    },
    {
      "epoch": 0.10678,
      "grad_norm": 0.734976551432324,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 10678
    },
    {
      "epoch": 0.10679,
      "grad_norm": 0.7314744669978124,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 10679
    },
    {
      "epoch": 0.1068,
      "grad_norm": 0.6908817611613471,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 10680
    },
    {
      "epoch": 0.10681,
      "grad_norm": 0.6399548375986829,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 10681
    },
    {
      "epoch": 0.10682,
      "grad_norm": 0.7053930398751589,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 10682
    },
    {
      "epoch": 0.10683,
      "grad_norm": 0.8259488173340027,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 10683
    },
    {
      "epoch": 0.10684,
      "grad_norm": 0.8721003907230183,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 10684
    },
    {
      "epoch": 0.10685,
      "grad_norm": 0.8470217654691111,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 10685
    },
    {
      "epoch": 0.10686,
      "grad_norm": 0.90992443878563,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 10686
    },
    {
      "epoch": 0.10687,
      "grad_norm": 1.024100970580944,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 10687
    },
    {
      "epoch": 0.10688,
      "grad_norm": 1.0637806320585521,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 10688
    },
    {
      "epoch": 0.10689,
      "grad_norm": 1.1156946627292235,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 10689
    },
    {
      "epoch": 0.1069,
      "grad_norm": 0.8046943265183353,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 10690
    },
    {
      "epoch": 0.10691,
      "grad_norm": 0.7100711170524753,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 10691
    },
    {
      "epoch": 0.10692,
      "grad_norm": 0.6574317863899862,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 10692
    },
    {
      "epoch": 0.10693,
      "grad_norm": 0.6251555571305705,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 10693
    },
    {
      "epoch": 0.10694,
      "grad_norm": 0.679376390140605,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 10694
    },
    {
      "epoch": 0.10695,
      "grad_norm": 0.8280371560374931,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 10695
    },
    {
      "epoch": 0.10696,
      "grad_norm": 0.9932001330147407,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 10696
    },
    {
      "epoch": 0.10697,
      "grad_norm": 1.1062910064134184,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 10697
    },
    {
      "epoch": 0.10698,
      "grad_norm": 0.7722807520515808,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 10698
    },
    {
      "epoch": 0.10699,
      "grad_norm": 0.7081182958148078,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 10699
    },
    {
      "epoch": 0.107,
      "grad_norm": 0.7166968485485714,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 10700
    },
    {
      "epoch": 0.10701,
      "grad_norm": 0.7537508686418137,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 10701
    },
    {
      "epoch": 0.10702,
      "grad_norm": 0.8835708055800392,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 10702
    },
    {
      "epoch": 0.10703,
      "grad_norm": 0.7817691460438141,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 10703
    },
    {
      "epoch": 0.10704,
      "grad_norm": 0.645477228585237,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 10704
    },
    {
      "epoch": 0.10705,
      "grad_norm": 0.7710367164781207,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 10705
    },
    {
      "epoch": 0.10706,
      "grad_norm": 0.7901517958326595,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 10706
    },
    {
      "epoch": 0.10707,
      "grad_norm": 0.7891516356491817,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 10707
    },
    {
      "epoch": 0.10708,
      "grad_norm": 0.8615348206451179,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 10708
    },
    {
      "epoch": 0.10709,
      "grad_norm": 0.9888041530740351,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 10709
    },
    {
      "epoch": 0.1071,
      "grad_norm": 1.061660380792365,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 10710
    },
    {
      "epoch": 0.10711,
      "grad_norm": 0.9066601556149967,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 10711
    },
    {
      "epoch": 0.10712,
      "grad_norm": 0.8116455998652893,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 10712
    },
    {
      "epoch": 0.10713,
      "grad_norm": 0.8637113563116897,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 10713
    },
    {
      "epoch": 0.10714,
      "grad_norm": 0.7022220024601558,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 10714
    },
    {
      "epoch": 0.10715,
      "grad_norm": 0.6293443796442022,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 10715
    },
    {
      "epoch": 0.10716,
      "grad_norm": 0.6779094422266427,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 10716
    },
    {
      "epoch": 0.10717,
      "grad_norm": 0.7211066849511437,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 10717
    },
    {
      "epoch": 0.10718,
      "grad_norm": 0.6577616772264899,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 10718
    },
    {
      "epoch": 0.10719,
      "grad_norm": 0.6728898224028206,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 10719
    },
    {
      "epoch": 0.1072,
      "grad_norm": 0.6504562644268949,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 10720
    },
    {
      "epoch": 0.10721,
      "grad_norm": 0.5906580401216541,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 10721
    },
    {
      "epoch": 0.10722,
      "grad_norm": 0.6651104603087701,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 10722
    },
    {
      "epoch": 0.10723,
      "grad_norm": 0.8274844100078965,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 10723
    },
    {
      "epoch": 0.10724,
      "grad_norm": 1.107508585414254,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 10724
    },
    {
      "epoch": 0.10725,
      "grad_norm": 0.9770130165439336,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 10725
    },
    {
      "epoch": 0.10726,
      "grad_norm": 0.8609099367579911,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 10726
    },
    {
      "epoch": 0.10727,
      "grad_norm": 0.729890468071586,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 10727
    },
    {
      "epoch": 0.10728,
      "grad_norm": 0.6948485160734639,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 10728
    },
    {
      "epoch": 0.10729,
      "grad_norm": 0.6678861533303945,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 10729
    },
    {
      "epoch": 0.1073,
      "grad_norm": 0.7248961043248474,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 10730
    },
    {
      "epoch": 0.10731,
      "grad_norm": 0.7071667241504125,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 10731
    },
    {
      "epoch": 0.10732,
      "grad_norm": 0.7205983714963741,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 10732
    },
    {
      "epoch": 0.10733,
      "grad_norm": 0.688369875669321,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 10733
    },
    {
      "epoch": 0.10734,
      "grad_norm": 0.6741612928630552,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 10734
    },
    {
      "epoch": 0.10735,
      "grad_norm": 0.7199477700388638,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 10735
    },
    {
      "epoch": 0.10736,
      "grad_norm": 0.7876731105567968,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 10736
    },
    {
      "epoch": 0.10737,
      "grad_norm": 0.8324327281824561,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 10737
    },
    {
      "epoch": 0.10738,
      "grad_norm": 0.8709320081981367,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 10738
    },
    {
      "epoch": 0.10739,
      "grad_norm": 0.9562955220392207,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 10739
    },
    {
      "epoch": 0.1074,
      "grad_norm": 1.2304307954831826,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 10740
    },
    {
      "epoch": 0.10741,
      "grad_norm": 0.9300435353142966,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 10741
    },
    {
      "epoch": 0.10742,
      "grad_norm": 0.9335382984724416,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 10742
    },
    {
      "epoch": 0.10743,
      "grad_norm": 0.9877653745311188,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 10743
    },
    {
      "epoch": 0.10744,
      "grad_norm": 1.0405104571146349,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 10744
    },
    {
      "epoch": 0.10745,
      "grad_norm": 0.8419311900278269,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 10745
    },
    {
      "epoch": 0.10746,
      "grad_norm": 0.8478057480297286,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 10746
    },
    {
      "epoch": 0.10747,
      "grad_norm": 0.9138973791244602,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 10747
    },
    {
      "epoch": 0.10748,
      "grad_norm": 0.9349681802837756,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 10748
    },
    {
      "epoch": 0.10749,
      "grad_norm": 0.950015954392382,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 10749
    },
    {
      "epoch": 0.1075,
      "grad_norm": 0.8943074737592488,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 10750
    },
    {
      "epoch": 0.10751,
      "grad_norm": 0.8401604688025964,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 10751
    },
    {
      "epoch": 0.10752,
      "grad_norm": 0.8907490401952103,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 10752
    },
    {
      "epoch": 0.10753,
      "grad_norm": 0.8019963256041405,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 10753
    },
    {
      "epoch": 0.10754,
      "grad_norm": 0.7334278789249893,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 10754
    },
    {
      "epoch": 0.10755,
      "grad_norm": 0.7807073635179747,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 10755
    },
    {
      "epoch": 0.10756,
      "grad_norm": 0.9037897305755073,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 10756
    },
    {
      "epoch": 0.10757,
      "grad_norm": 1.0038789915507889,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 10757
    },
    {
      "epoch": 0.10758,
      "grad_norm": 0.8637668717135266,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 10758
    },
    {
      "epoch": 0.10759,
      "grad_norm": 0.7784709868978823,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 10759
    },
    {
      "epoch": 0.1076,
      "grad_norm": 0.7513385059899371,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 10760
    },
    {
      "epoch": 0.10761,
      "grad_norm": 0.658145040039802,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 10761
    },
    {
      "epoch": 0.10762,
      "grad_norm": 0.6319284675105509,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 10762
    },
    {
      "epoch": 0.10763,
      "grad_norm": 0.6484314688691583,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 10763
    },
    {
      "epoch": 0.10764,
      "grad_norm": 0.6447554772230162,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 10764
    },
    {
      "epoch": 0.10765,
      "grad_norm": 0.6947856450689127,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 10765
    },
    {
      "epoch": 0.10766,
      "grad_norm": 0.8271087875534372,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 10766
    },
    {
      "epoch": 0.10767,
      "grad_norm": 0.9248495761280467,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 10767
    },
    {
      "epoch": 0.10768,
      "grad_norm": 0.906365044623054,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 10768
    },
    {
      "epoch": 0.10769,
      "grad_norm": 0.9421556590622695,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 10769
    },
    {
      "epoch": 0.1077,
      "grad_norm": 0.9830274427766104,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 10770
    },
    {
      "epoch": 0.10771,
      "grad_norm": 0.7459901139183797,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 10771
    },
    {
      "epoch": 0.10772,
      "grad_norm": 0.8029037552458691,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 10772
    },
    {
      "epoch": 0.10773,
      "grad_norm": 0.8413257874108826,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 10773
    },
    {
      "epoch": 0.10774,
      "grad_norm": 0.8939967598650286,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 10774
    },
    {
      "epoch": 0.10775,
      "grad_norm": 0.9500930304152452,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 10775
    },
    {
      "epoch": 0.10776,
      "grad_norm": 0.9180361337301888,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 10776
    },
    {
      "epoch": 0.10777,
      "grad_norm": 0.8739106054853617,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 10777
    },
    {
      "epoch": 0.10778,
      "grad_norm": 0.7861751893427512,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 10778
    },
    {
      "epoch": 0.10779,
      "grad_norm": 0.6801288616685922,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 10779
    },
    {
      "epoch": 0.1078,
      "grad_norm": 0.6595454326458605,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 10780
    },
    {
      "epoch": 0.10781,
      "grad_norm": 0.677694591198545,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 10781
    },
    {
      "epoch": 0.10782,
      "grad_norm": 0.7322536223576103,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 10782
    },
    {
      "epoch": 0.10783,
      "grad_norm": 0.6949774146294931,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 10783
    },
    {
      "epoch": 0.10784,
      "grad_norm": 0.7112783123302691,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 10784
    },
    {
      "epoch": 0.10785,
      "grad_norm": 0.7624840584818994,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 10785
    },
    {
      "epoch": 0.10786,
      "grad_norm": 0.8307593955140588,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 10786
    },
    {
      "epoch": 0.10787,
      "grad_norm": 0.8234800415780226,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 10787
    },
    {
      "epoch": 0.10788,
      "grad_norm": 0.7063471007386206,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 10788
    },
    {
      "epoch": 0.10789,
      "grad_norm": 0.7003812666554212,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 10789
    },
    {
      "epoch": 0.1079,
      "grad_norm": 0.591255796389519,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 10790
    },
    {
      "epoch": 0.10791,
      "grad_norm": 0.6588407077575648,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 10791
    },
    {
      "epoch": 0.10792,
      "grad_norm": 0.7812948240216615,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 10792
    },
    {
      "epoch": 0.10793,
      "grad_norm": 0.875484648604842,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 10793
    },
    {
      "epoch": 0.10794,
      "grad_norm": 1.0819813451984712,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 10794
    },
    {
      "epoch": 0.10795,
      "grad_norm": 1.0106647386412784,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 10795
    },
    {
      "epoch": 0.10796,
      "grad_norm": 0.8727394703935979,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 10796
    },
    {
      "epoch": 0.10797,
      "grad_norm": 0.7493976900488687,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 10797
    },
    {
      "epoch": 0.10798,
      "grad_norm": 0.6778670897020321,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 10798
    },
    {
      "epoch": 0.10799,
      "grad_norm": 0.6886630260315241,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 10799
    },
    {
      "epoch": 0.108,
      "grad_norm": 0.6997924479763118,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 10800
    },
    {
      "epoch": 0.10801,
      "grad_norm": 0.6335312641192151,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 10801
    },
    {
      "epoch": 0.10802,
      "grad_norm": 0.5596975143808999,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 10802
    },
    {
      "epoch": 0.10803,
      "grad_norm": 0.5883654642439371,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 10803
    },
    {
      "epoch": 0.10804,
      "grad_norm": 0.6353797720617529,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 10804
    },
    {
      "epoch": 0.10805,
      "grad_norm": 0.766636901293996,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 10805
    },
    {
      "epoch": 0.10806,
      "grad_norm": 0.9243626908434707,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 10806
    },
    {
      "epoch": 0.10807,
      "grad_norm": 0.9974198487878154,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 10807
    },
    {
      "epoch": 0.10808,
      "grad_norm": 0.906105158565713,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 10808
    },
    {
      "epoch": 0.10809,
      "grad_norm": 0.9969514627729021,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 10809
    },
    {
      "epoch": 0.1081,
      "grad_norm": 0.834425050345777,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 10810
    },
    {
      "epoch": 0.10811,
      "grad_norm": 0.8783993703886297,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 10811
    },
    {
      "epoch": 0.10812,
      "grad_norm": 1.0486220426031836,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 10812
    },
    {
      "epoch": 0.10813,
      "grad_norm": 0.9635854562205466,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 10813
    },
    {
      "epoch": 0.10814,
      "grad_norm": 0.991913843257712,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 10814
    },
    {
      "epoch": 0.10815,
      "grad_norm": 1.03240484345877,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 10815
    },
    {
      "epoch": 0.10816,
      "grad_norm": 0.7999243463833863,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 10816
    },
    {
      "epoch": 0.10817,
      "grad_norm": 0.7146118966430277,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 10817
    },
    {
      "epoch": 0.10818,
      "grad_norm": 0.6760564583671121,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 10818
    },
    {
      "epoch": 0.10819,
      "grad_norm": 0.7437787761517194,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 10819
    },
    {
      "epoch": 0.1082,
      "grad_norm": 0.7667318805568926,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 10820
    },
    {
      "epoch": 0.10821,
      "grad_norm": 0.7578063636892496,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 10821
    },
    {
      "epoch": 0.10822,
      "grad_norm": 0.9722299248990864,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 10822
    },
    {
      "epoch": 0.10823,
      "grad_norm": 1.084422938439403,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 10823
    },
    {
      "epoch": 0.10824,
      "grad_norm": 0.7866493459633321,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 10824
    },
    {
      "epoch": 0.10825,
      "grad_norm": 0.6928064188763265,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 10825
    },
    {
      "epoch": 0.10826,
      "grad_norm": 0.7316035379733721,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 10826
    },
    {
      "epoch": 0.10827,
      "grad_norm": 0.7872692879605871,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 10827
    },
    {
      "epoch": 0.10828,
      "grad_norm": 0.6915375778555326,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 10828
    },
    {
      "epoch": 0.10829,
      "grad_norm": 0.748471989224979,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 10829
    },
    {
      "epoch": 0.1083,
      "grad_norm": 0.763563153780263,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 10830
    },
    {
      "epoch": 0.10831,
      "grad_norm": 0.7557051035736269,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 10831
    },
    {
      "epoch": 0.10832,
      "grad_norm": 0.8081559760247755,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 10832
    },
    {
      "epoch": 0.10833,
      "grad_norm": 0.7916475572460021,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 10833
    },
    {
      "epoch": 0.10834,
      "grad_norm": 0.7180126556558653,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 10834
    },
    {
      "epoch": 0.10835,
      "grad_norm": 0.6897522880798621,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 10835
    },
    {
      "epoch": 0.10836,
      "grad_norm": 0.720390447907444,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 10836
    },
    {
      "epoch": 0.10837,
      "grad_norm": 0.7729119904062143,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 10837
    },
    {
      "epoch": 0.10838,
      "grad_norm": 0.974189110491675,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 10838
    },
    {
      "epoch": 0.10839,
      "grad_norm": 1.1146789421849448,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 10839
    },
    {
      "epoch": 0.1084,
      "grad_norm": 0.7863144186085149,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 10840
    },
    {
      "epoch": 0.10841,
      "grad_norm": 0.6344950133109905,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 10841
    },
    {
      "epoch": 0.10842,
      "grad_norm": 0.6253130642537409,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 10842
    },
    {
      "epoch": 0.10843,
      "grad_norm": 0.642829186840075,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 10843
    },
    {
      "epoch": 0.10844,
      "grad_norm": 0.7172865995614057,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 10844
    },
    {
      "epoch": 0.10845,
      "grad_norm": 0.8387230458337437,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 10845
    },
    {
      "epoch": 0.10846,
      "grad_norm": 0.8272335404876863,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 10846
    },
    {
      "epoch": 0.10847,
      "grad_norm": 0.8339723408210796,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 10847
    },
    {
      "epoch": 0.10848,
      "grad_norm": 0.6960991280713948,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 10848
    },
    {
      "epoch": 0.10849,
      "grad_norm": 0.621407103825329,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 10849
    },
    {
      "epoch": 0.1085,
      "grad_norm": 0.6226763190146141,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 10850
    },
    {
      "epoch": 0.10851,
      "grad_norm": 0.7346253480020578,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 10851
    },
    {
      "epoch": 0.10852,
      "grad_norm": 0.8514943147557615,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 10852
    },
    {
      "epoch": 0.10853,
      "grad_norm": 0.9517437965287522,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 10853
    },
    {
      "epoch": 0.10854,
      "grad_norm": 0.9508741169410276,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 10854
    },
    {
      "epoch": 0.10855,
      "grad_norm": 0.7768027675885154,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 10855
    },
    {
      "epoch": 0.10856,
      "grad_norm": 0.7413209828275256,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 10856
    },
    {
      "epoch": 0.10857,
      "grad_norm": 0.7185402434742164,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 10857
    },
    {
      "epoch": 0.10858,
      "grad_norm": 0.6826430053195711,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 10858
    },
    {
      "epoch": 0.10859,
      "grad_norm": 0.761533780013182,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 10859
    },
    {
      "epoch": 0.1086,
      "grad_norm": 0.7642804260240755,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 10860
    },
    {
      "epoch": 0.10861,
      "grad_norm": 0.825548626289074,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 10861
    },
    {
      "epoch": 0.10862,
      "grad_norm": 0.9037697230091956,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 10862
    },
    {
      "epoch": 0.10863,
      "grad_norm": 0.8657754903618303,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 10863
    },
    {
      "epoch": 0.10864,
      "grad_norm": 0.8141322733115343,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 10864
    },
    {
      "epoch": 0.10865,
      "grad_norm": 0.7253469991497675,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 10865
    },
    {
      "epoch": 0.10866,
      "grad_norm": 0.7117167359362205,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 10866
    },
    {
      "epoch": 0.10867,
      "grad_norm": 0.7882818390575478,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 10867
    },
    {
      "epoch": 0.10868,
      "grad_norm": 0.9220190688619744,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 10868
    },
    {
      "epoch": 0.10869,
      "grad_norm": 1.2171062146891194,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 10869
    },
    {
      "epoch": 0.1087,
      "grad_norm": 0.9409103742766868,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 10870
    },
    {
      "epoch": 0.10871,
      "grad_norm": 0.7857575782459725,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 10871
    },
    {
      "epoch": 0.10872,
      "grad_norm": 0.7008071897173932,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 10872
    },
    {
      "epoch": 0.10873,
      "grad_norm": 0.6493494671071487,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 10873
    },
    {
      "epoch": 0.10874,
      "grad_norm": 0.7422045080755252,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 10874
    },
    {
      "epoch": 0.10875,
      "grad_norm": 0.7813850020221002,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 10875
    },
    {
      "epoch": 0.10876,
      "grad_norm": 0.7337134215205611,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 10876
    },
    {
      "epoch": 0.10877,
      "grad_norm": 0.5738251794128827,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 10877
    },
    {
      "epoch": 0.10878,
      "grad_norm": 0.5971880485148056,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 10878
    },
    {
      "epoch": 0.10879,
      "grad_norm": 0.6407269342613244,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 10879
    },
    {
      "epoch": 0.1088,
      "grad_norm": 0.7567936302504148,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 10880
    },
    {
      "epoch": 0.10881,
      "grad_norm": 0.8952438404265414,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 10881
    },
    {
      "epoch": 0.10882,
      "grad_norm": 1.016869262938487,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 10882
    },
    {
      "epoch": 0.10883,
      "grad_norm": 0.9089472739873565,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 10883
    },
    {
      "epoch": 0.10884,
      "grad_norm": 0.8192132716200892,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 10884
    },
    {
      "epoch": 0.10885,
      "grad_norm": 0.754863660292805,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 10885
    },
    {
      "epoch": 0.10886,
      "grad_norm": 0.7646838940972092,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 10886
    },
    {
      "epoch": 0.10887,
      "grad_norm": 0.8054792718032056,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 10887
    },
    {
      "epoch": 0.10888,
      "grad_norm": 0.807217157369814,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 10888
    },
    {
      "epoch": 0.10889,
      "grad_norm": 0.9596711545065643,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 10889
    },
    {
      "epoch": 0.1089,
      "grad_norm": 1.0281488211494914,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 10890
    },
    {
      "epoch": 0.10891,
      "grad_norm": 1.008425907314204,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 10891
    },
    {
      "epoch": 0.10892,
      "grad_norm": 0.9881878897319983,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 10892
    },
    {
      "epoch": 0.10893,
      "grad_norm": 0.9651702578236134,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 10893
    },
    {
      "epoch": 0.10894,
      "grad_norm": 0.91488242972445,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 10894
    },
    {
      "epoch": 0.10895,
      "grad_norm": 0.820190985605287,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 10895
    },
    {
      "epoch": 0.10896,
      "grad_norm": 0.8047362457919742,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 10896
    },
    {
      "epoch": 0.10897,
      "grad_norm": 0.8272579145202782,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 10897
    },
    {
      "epoch": 0.10898,
      "grad_norm": 0.8529516426412708,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 10898
    },
    {
      "epoch": 0.10899,
      "grad_norm": 0.792496666395517,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 10899
    },
    {
      "epoch": 0.109,
      "grad_norm": 0.938657528442336,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 10900
    },
    {
      "epoch": 0.10901,
      "grad_norm": 1.2363502284270567,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 10901
    },
    {
      "epoch": 0.10902,
      "grad_norm": 0.7903678185797423,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 10902
    },
    {
      "epoch": 0.10903,
      "grad_norm": 0.6789417264937385,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 10903
    },
    {
      "epoch": 0.10904,
      "grad_norm": 0.7466995321912779,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 10904
    },
    {
      "epoch": 0.10905,
      "grad_norm": 0.8446333893055047,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 10905
    },
    {
      "epoch": 0.10906,
      "grad_norm": 0.9434146945884782,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 10906
    },
    {
      "epoch": 0.10907,
      "grad_norm": 0.9448346223448993,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 10907
    },
    {
      "epoch": 0.10908,
      "grad_norm": 0.9492929430380965,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 10908
    },
    {
      "epoch": 0.10909,
      "grad_norm": 0.9638647144203306,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 10909
    },
    {
      "epoch": 0.1091,
      "grad_norm": 0.8114844498739628,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 10910
    },
    {
      "epoch": 0.10911,
      "grad_norm": 0.7861873196944332,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 10911
    },
    {
      "epoch": 0.10912,
      "grad_norm": 0.6992439758399313,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 10912
    },
    {
      "epoch": 0.10913,
      "grad_norm": 0.6815651781157617,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 10913
    },
    {
      "epoch": 0.10914,
      "grad_norm": 0.6536113835814653,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 10914
    },
    {
      "epoch": 0.10915,
      "grad_norm": 0.6938082032494179,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 10915
    },
    {
      "epoch": 0.10916,
      "grad_norm": 0.7202674599350765,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 10916
    },
    {
      "epoch": 0.10917,
      "grad_norm": 0.7491490948689121,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 10917
    },
    {
      "epoch": 0.10918,
      "grad_norm": 1.11036055240353,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 10918
    },
    {
      "epoch": 0.10919,
      "grad_norm": 1.1337221606450771,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 10919
    },
    {
      "epoch": 0.1092,
      "grad_norm": 0.8428547311068283,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 10920
    },
    {
      "epoch": 0.10921,
      "grad_norm": 0.6082082658833271,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 10921
    },
    {
      "epoch": 0.10922,
      "grad_norm": 0.6315066804935527,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 10922
    },
    {
      "epoch": 0.10923,
      "grad_norm": 0.6061406163799906,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 10923
    },
    {
      "epoch": 0.10924,
      "grad_norm": 0.6369014987608396,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 10924
    },
    {
      "epoch": 0.10925,
      "grad_norm": 0.6509900001681895,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 10925
    },
    {
      "epoch": 0.10926,
      "grad_norm": 0.5911101588927634,
      "learning_rate": 0.003,
      "loss": 4.037,
      "step": 10926
    },
    {
      "epoch": 0.10927,
      "grad_norm": 0.6090199769351918,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 10927
    },
    {
      "epoch": 0.10928,
      "grad_norm": 0.7583916555674786,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 10928
    },
    {
      "epoch": 0.10929,
      "grad_norm": 0.9306843772780642,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 10929
    },
    {
      "epoch": 0.1093,
      "grad_norm": 1.2362967491595924,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 10930
    },
    {
      "epoch": 0.10931,
      "grad_norm": 0.6502246716231995,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 10931
    },
    {
      "epoch": 0.10932,
      "grad_norm": 0.5470588873212986,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 10932
    },
    {
      "epoch": 0.10933,
      "grad_norm": 0.6038156560331971,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 10933
    },
    {
      "epoch": 0.10934,
      "grad_norm": 0.631710911079846,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 10934
    },
    {
      "epoch": 0.10935,
      "grad_norm": 0.6072788054734551,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 10935
    },
    {
      "epoch": 0.10936,
      "grad_norm": 0.5548572160813884,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 10936
    },
    {
      "epoch": 0.10937,
      "grad_norm": 0.5678083242728192,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 10937
    },
    {
      "epoch": 0.10938,
      "grad_norm": 0.5349364110178259,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 10938
    },
    {
      "epoch": 0.10939,
      "grad_norm": 0.5314674417062363,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 10939
    },
    {
      "epoch": 0.1094,
      "grad_norm": 0.5567007468861874,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 10940
    },
    {
      "epoch": 0.10941,
      "grad_norm": 0.6095491373372913,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 10941
    },
    {
      "epoch": 0.10942,
      "grad_norm": 0.7298043409485729,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 10942
    },
    {
      "epoch": 0.10943,
      "grad_norm": 0.786513510107108,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 10943
    },
    {
      "epoch": 0.10944,
      "grad_norm": 0.7774007804681506,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 10944
    },
    {
      "epoch": 0.10945,
      "grad_norm": 0.7495722469210125,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 10945
    },
    {
      "epoch": 0.10946,
      "grad_norm": 0.8064706771959848,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 10946
    },
    {
      "epoch": 0.10947,
      "grad_norm": 0.9393433262565961,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 10947
    },
    {
      "epoch": 0.10948,
      "grad_norm": 0.9387790046571931,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 10948
    },
    {
      "epoch": 0.10949,
      "grad_norm": 1.1410032379592778,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 10949
    },
    {
      "epoch": 0.1095,
      "grad_norm": 0.9129552961385187,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 10950
    },
    {
      "epoch": 0.10951,
      "grad_norm": 1.0435796410645963,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 10951
    },
    {
      "epoch": 0.10952,
      "grad_norm": 1.3358604841725472,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 10952
    },
    {
      "epoch": 0.10953,
      "grad_norm": 0.8310108373581879,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 10953
    },
    {
      "epoch": 0.10954,
      "grad_norm": 0.6723858624058059,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 10954
    },
    {
      "epoch": 0.10955,
      "grad_norm": 0.6611226964259507,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 10955
    },
    {
      "epoch": 0.10956,
      "grad_norm": 0.7011303501399112,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 10956
    },
    {
      "epoch": 0.10957,
      "grad_norm": 0.6836216086535878,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 10957
    },
    {
      "epoch": 0.10958,
      "grad_norm": 0.7592475674795014,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 10958
    },
    {
      "epoch": 0.10959,
      "grad_norm": 0.7522399573457461,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 10959
    },
    {
      "epoch": 0.1096,
      "grad_norm": 0.7583206213419711,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 10960
    },
    {
      "epoch": 0.10961,
      "grad_norm": 0.7948798372620559,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 10961
    },
    {
      "epoch": 0.10962,
      "grad_norm": 0.9520999820511873,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 10962
    },
    {
      "epoch": 0.10963,
      "grad_norm": 0.9934263182752433,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 10963
    },
    {
      "epoch": 0.10964,
      "grad_norm": 1.0404312099324224,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 10964
    },
    {
      "epoch": 0.10965,
      "grad_norm": 0.9314463760092233,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 10965
    },
    {
      "epoch": 0.10966,
      "grad_norm": 0.8350231443250099,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 10966
    },
    {
      "epoch": 0.10967,
      "grad_norm": 0.9320300124939774,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 10967
    },
    {
      "epoch": 0.10968,
      "grad_norm": 0.9215579233627603,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 10968
    },
    {
      "epoch": 0.10969,
      "grad_norm": 1.0604243559948778,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 10969
    },
    {
      "epoch": 0.1097,
      "grad_norm": 1.1138978634435108,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 10970
    },
    {
      "epoch": 0.10971,
      "grad_norm": 0.8951876392104671,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 10971
    },
    {
      "epoch": 0.10972,
      "grad_norm": 0.9676653328126739,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 10972
    },
    {
      "epoch": 0.10973,
      "grad_norm": 1.0352834124023993,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 10973
    },
    {
      "epoch": 0.10974,
      "grad_norm": 1.0675730058873871,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 10974
    },
    {
      "epoch": 0.10975,
      "grad_norm": 0.9740004559762198,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 10975
    },
    {
      "epoch": 0.10976,
      "grad_norm": 1.0447644987824045,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 10976
    },
    {
      "epoch": 0.10977,
      "grad_norm": 1.0200289761418742,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 10977
    },
    {
      "epoch": 0.10978,
      "grad_norm": 0.9365717545240859,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 10978
    },
    {
      "epoch": 0.10979,
      "grad_norm": 0.8362679752586221,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 10979
    },
    {
      "epoch": 0.1098,
      "grad_norm": 0.7172618437426157,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 10980
    },
    {
      "epoch": 0.10981,
      "grad_norm": 0.6864193346218213,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 10981
    },
    {
      "epoch": 0.10982,
      "grad_norm": 0.5895896832601729,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 10982
    },
    {
      "epoch": 0.10983,
      "grad_norm": 0.5787111289779593,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 10983
    },
    {
      "epoch": 0.10984,
      "grad_norm": 0.6057523165429262,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 10984
    },
    {
      "epoch": 0.10985,
      "grad_norm": 0.6514886745623131,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 10985
    },
    {
      "epoch": 0.10986,
      "grad_norm": 0.7435079048577373,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 10986
    },
    {
      "epoch": 0.10987,
      "grad_norm": 0.9200597665780822,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 10987
    },
    {
      "epoch": 0.10988,
      "grad_norm": 1.1265238934764898,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 10988
    },
    {
      "epoch": 0.10989,
      "grad_norm": 0.88644921715438,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 10989
    },
    {
      "epoch": 0.1099,
      "grad_norm": 0.8325133772180551,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 10990
    },
    {
      "epoch": 0.10991,
      "grad_norm": 0.8420468014984893,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 10991
    },
    {
      "epoch": 0.10992,
      "grad_norm": 0.8537316831757594,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 10992
    },
    {
      "epoch": 0.10993,
      "grad_norm": 0.7940387723272442,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 10993
    },
    {
      "epoch": 0.10994,
      "grad_norm": 0.6710457088949344,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 10994
    },
    {
      "epoch": 0.10995,
      "grad_norm": 0.6142598287538222,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 10995
    },
    {
      "epoch": 0.10996,
      "grad_norm": 0.6660630848750299,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 10996
    },
    {
      "epoch": 0.10997,
      "grad_norm": 0.7899825737081104,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 10997
    },
    {
      "epoch": 0.10998,
      "grad_norm": 0.8923751751449065,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 10998
    },
    {
      "epoch": 0.10999,
      "grad_norm": 0.9402386393787387,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 10999
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9050561963603275,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 11000
    }
  ],
  "logging_steps": 1,
  "max_steps": 100000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.36080071409664e+17,
  "train_batch_size": 1024,
  "trial_name": null,
  "trial_params": null
}