{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.925,
  "eval_steps": 500,
  "global_step": 154,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 0.73828125,
      "learning_rate": 4.6875e-07,
      "loss": 1.7467,
      "step": 1
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.375,
      "learning_rate": 9.375e-07,
      "loss": 2.0932,
      "step": 2
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.92578125,
      "learning_rate": 1.40625e-06,
      "loss": 1.8037,
      "step": 3
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.61328125,
      "learning_rate": 1.875e-06,
      "loss": 1.8117,
      "step": 4
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.73046875,
      "learning_rate": 2.3437500000000002e-06,
      "loss": 1.9253,
      "step": 5
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.55078125,
      "learning_rate": 2.8125e-06,
      "loss": 1.8049,
      "step": 6
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9375,
      "learning_rate": 3.28125e-06,
      "loss": 1.9055,
      "step": 7
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.74609375,
      "learning_rate": 3.75e-06,
      "loss": 1.8487,
      "step": 8
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.71875,
      "learning_rate": 4.21875e-06,
      "loss": 1.6985,
      "step": 9
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.90625,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 1.8537,
      "step": 10
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.73046875,
      "learning_rate": 5.15625e-06,
      "loss": 1.952,
      "step": 11
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.474609375,
      "learning_rate": 5.625e-06,
      "loss": 2.1275,
      "step": 12
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.74609375,
      "learning_rate": 6.09375e-06,
      "loss": 1.7841,
      "step": 13
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.80859375,
      "learning_rate": 6.5625e-06,
      "loss": 1.782,
      "step": 14
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.703125,
      "learning_rate": 7.03125e-06,
      "loss": 1.7511,
      "step": 15
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.50390625,
      "learning_rate": 7.5e-06,
      "loss": 1.9407,
      "step": 16
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.48046875,
      "learning_rate": 7.96875e-06,
      "loss": 1.8245,
      "step": 17
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1875,
      "learning_rate": 8.4375e-06,
      "loss": 2.0168,
      "step": 18
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.93359375,
      "learning_rate": 8.90625e-06,
      "loss": 2.035,
      "step": 19
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.58203125,
      "learning_rate": 9.375000000000001e-06,
      "loss": 1.9485,
      "step": 20
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.466796875,
      "learning_rate": 9.84375e-06,
      "loss": 1.7688,
      "step": 21
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.52734375,
      "learning_rate": 1.03125e-05,
      "loss": 1.8852,
      "step": 22
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.484375,
      "learning_rate": 1.078125e-05,
      "loss": 1.8949,
      "step": 23
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3671875,
      "learning_rate": 1.125e-05,
      "loss": 1.7596,
      "step": 24
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6015625,
      "learning_rate": 1.171875e-05,
      "loss": 1.7431,
      "step": 25
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.63671875,
      "learning_rate": 1.21875e-05,
      "loss": 1.6513,
      "step": 26
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3671875,
      "learning_rate": 1.2656250000000001e-05,
      "loss": 1.8582,
      "step": 27
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6328125,
      "learning_rate": 1.3125e-05,
      "loss": 1.8404,
      "step": 28
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3671875,
      "learning_rate": 1.359375e-05,
      "loss": 1.9079,
      "step": 29
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.72265625,
      "learning_rate": 1.40625e-05,
      "loss": 1.7028,
      "step": 30
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.380859375,
      "learning_rate": 1.453125e-05,
      "loss": 1.8124,
      "step": 31
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.447265625,
      "learning_rate": 1.5e-05,
      "loss": 1.8654,
      "step": 32
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.384765625,
      "learning_rate": 1.4998192912177226e-05,
      "loss": 1.8245,
      "step": 33
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.56640625,
      "learning_rate": 1.4992772737231035e-05,
      "loss": 1.8048,
      "step": 34
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4140625,
      "learning_rate": 1.4983742740072142e-05,
      "loss": 1.751,
      "step": 35
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.404296875,
      "learning_rate": 1.4971108360033184e-05,
      "loss": 1.753,
      "step": 36
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33984375,
      "learning_rate": 1.495487720759226e-05,
      "loss": 1.6683,
      "step": 37
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.34375,
      "learning_rate": 1.4935059059788688e-05,
      "loss": 1.8784,
      "step": 38
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2294921875,
      "learning_rate": 1.4911665854333647e-05,
      "loss": 1.8926,
      "step": 39
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.353515625,
      "learning_rate": 1.4884711682419384e-05,
      "loss": 1.4877,
      "step": 40
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2373046875,
      "learning_rate": 1.485421278023117e-05,
      "loss": 1.7733,
      "step": 41
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.734375,
      "learning_rate": 1.4820187519167265e-05,
      "loss": 1.7075,
      "step": 42
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.546875,
      "learning_rate": 1.478265639477264e-05,
      "loss": 1.7193,
      "step": 43
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2294921875,
      "learning_rate": 1.4741642014393254e-05,
      "loss": 1.7998,
      "step": 44
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.259765625,
      "learning_rate": 1.4697169083558222e-05,
      "loss": 1.8215,
      "step": 45
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.330078125,
      "learning_rate": 1.4649264391098124e-05,
      "loss": 1.8873,
      "step": 46
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.26171875,
      "learning_rate": 1.4597956793008433e-05,
      "loss": 1.7815,
      "step": 47
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.255859375,
      "learning_rate": 1.454327719506772e-05,
      "loss": 1.8338,
      "step": 48
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.345703125,
      "learning_rate": 1.4485258534221185e-05,
      "loss": 1.8003,
      "step": 49
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.328125,
      "learning_rate": 1.4423935758740662e-05,
      "loss": 1.6483,
      "step": 50
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19140625,
      "learning_rate": 1.4359345807173091e-05,
      "loss": 1.8148,
      "step": 51
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.326171875,
      "learning_rate": 1.429152758609013e-05,
      "loss": 1.6243,
      "step": 52
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2431640625,
      "learning_rate": 1.4220521946652268e-05,
      "loss": 1.7367,
      "step": 53
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.388671875,
      "learning_rate": 1.4146371660001633e-05,
      "loss": 1.7469,
      "step": 54
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.32421875,
      "learning_rate": 1.4069121391498243e-05,
      "loss": 1.9459,
      "step": 55
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.251953125,
      "learning_rate": 1.3988817673815272e-05,
      "loss": 1.5953,
      "step": 56
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.3905508878909502e-05,
      "loss": 1.8231,
      "step": 57
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.42578125,
      "learning_rate": 1.381924518888387e-05,
      "loss": 1.7221,
      "step": 58
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.314453125,
      "learning_rate": 1.3730078565759639e-05,
      "loss": 1.7005,
      "step": 59
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.322265625,
      "learning_rate": 1.3638062720176423e-05,
      "loss": 1.757,
      "step": 60
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.337890625,
      "learning_rate": 1.354325307903891e-05,
      "loss": 1.6586,
      "step": 61
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.330078125,
      "learning_rate": 1.3445706752129755e-05,
      "loss": 1.5988,
      "step": 62
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.421875,
      "learning_rate": 1.3345482497708803e-05,
      "loss": 1.7875,
      "step": 63
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.208984375,
      "learning_rate": 1.3242640687119285e-05,
      "loss": 1.7303,
      "step": 64
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.279296875,
      "learning_rate": 1.3137243268422403e-05,
      "loss": 1.6414,
      "step": 65
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.265625,
      "learning_rate": 1.3029353729082113e-05,
      "loss": 1.6532,
      "step": 66
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.236328125,
      "learning_rate": 1.291903705772266e-05,
      "loss": 1.7693,
      "step": 67
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.478515625,
      "learning_rate": 1.2806359704981872e-05,
      "loss": 2.0534,
      "step": 68
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.15625,
      "learning_rate": 1.2691389543483761e-05,
      "loss": 1.6768,
      "step": 69
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2041015625,
      "learning_rate": 1.2574195826954602e-05,
      "loss": 1.6262,
      "step": 70
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2158203125,
      "learning_rate": 1.2454849148507073e-05,
      "loss": 1.6982,
      "step": 71
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.240234375,
      "learning_rate": 1.2333421398117615e-05,
      "loss": 1.7021,
      "step": 72
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1630859375,
      "learning_rate": 1.2209985719322584e-05,
      "loss": 1.6838,
      "step": 73
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.419921875,
      "learning_rate": 1.2084616465159332e-05,
      "loss": 1.727,
      "step": 74
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.3125,
      "learning_rate": 1.1957389153378707e-05,
      "loss": 1.6642,
      "step": 75
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3046875,
      "learning_rate": 1.182838042095599e-05,
      "loss": 1.6989,
      "step": 76
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.263671875,
      "learning_rate": 1.1697667977927642e-05,
      "loss": 1.6819,
      "step": 77
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.427734375,
      "learning_rate": 1.1565330560581693e-05,
      "loss": 1.7467,
      "step": 78
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.302734375,
      "learning_rate": 1.1431447884029941e-05,
      "loss": 1.6749,
      "step": 79
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.220703125,
      "learning_rate": 1.129610059419054e-05,
      "loss": 1.7539,
      "step": 80
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.322265625,
      "learning_rate": 1.115937021920993e-05,
      "loss": 1.6246,
      "step": 81
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1982421875,
      "learning_rate": 1.1021339120353321e-05,
      "loss": 1.7767,
      "step": 82
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.326171875,
      "learning_rate": 1.0882090442393351e-05,
      "loss": 1.7048,
      "step": 83
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.2109375,
      "learning_rate": 1.0741708063526774e-05,
      "loss": 1.653,
      "step": 84
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2080078125,
      "learning_rate": 1.0600276544849392e-05,
      "loss": 1.636,
      "step": 85
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1787109375,
      "learning_rate": 1.0457881079419584e-05,
      "loss": 1.6875,
      "step": 86
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.337890625,
      "learning_rate": 1.0314607440941219e-05,
      "loss": 1.4956,
      "step": 87
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.33984375,
      "learning_rate": 1.017054193209677e-05,
      "loss": 1.9225,
      "step": 88
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.234375,
      "learning_rate": 1.0025771332561807e-05,
      "loss": 1.5713,
      "step": 89
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.2578125,
      "learning_rate": 9.880382846732171e-06,
      "loss": 1.7293,
      "step": 90
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.29296875,
      "learning_rate": 9.734464051195298e-06,
      "loss": 1.7035,
      "step": 91
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.328125,
      "learning_rate": 9.588102841977366e-06,
      "loss": 1.6466,
      "step": 92
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1796875,
      "learning_rate": 9.441387381598006e-06,
      "loss": 1.8003,
      "step": 93
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.212890625,
      "learning_rate": 9.294406045964509e-06,
      "loss": 1.7824,
      "step": 94
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.150390625,
      "learning_rate": 9.147247371137474e-06,
      "loss": 1.7703,
      "step": 95
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1884765625,
      "learning_rate": 9.000000000000002e-06,
      "loss": 1.774,
      "step": 96
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.2412109375,
      "learning_rate": 8.852752628862527e-06,
      "loss": 1.5992,
      "step": 97
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.40234375,
      "learning_rate": 8.705593954035492e-06,
      "loss": 1.4456,
      "step": 98
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.201171875,
      "learning_rate": 8.558612618401998e-06,
      "loss": 1.6569,
      "step": 99
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.21484375,
      "learning_rate": 8.411897158022636e-06,
      "loss": 1.5053,
      "step": 100
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1826171875,
      "learning_rate": 8.265535948804704e-06,
      "loss": 1.6563,
      "step": 101
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.3125,
      "learning_rate": 8.119617153267831e-06,
      "loss": 1.7722,
      "step": 102
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.18359375,
      "learning_rate": 7.974228667438192e-06,
      "loss": 1.7319,
      "step": 103
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.3203125,
      "learning_rate": 7.829458067903232e-06,
      "loss": 1.7543,
      "step": 104
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.640625,
      "learning_rate": 7.685392559058783e-06,
      "loss": 1.7884,
      "step": 105
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.2138671875,
      "learning_rate": 7.5421189205804185e-06,
      "loss": 1.5919,
      "step": 106
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.2333984375,
      "learning_rate": 7.399723455150611e-06,
      "loss": 1.6299,
      "step": 107
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.1591796875,
      "learning_rate": 7.258291936473227e-06,
      "loss": 1.6468,
      "step": 108
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.2060546875,
      "learning_rate": 7.117909557606652e-06,
      "loss": 1.509,
      "step": 109
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.2236328125,
      "learning_rate": 6.978660879646681e-06,
      "loss": 1.7308,
      "step": 110
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.35546875,
      "learning_rate": 6.840629780790071e-06,
      "loss": 1.7327,
      "step": 111
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.2265625,
      "learning_rate": 6.7038994058094625e-06,
      "loss": 1.6166,
      "step": 112
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.283203125,
      "learning_rate": 6.568552115970062e-06,
      "loss": 1.8531,
      "step": 113
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.1923828125,
      "learning_rate": 6.434669439418309e-06,
      "loss": 1.612,
      "step": 114
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.37890625,
      "learning_rate": 6.3023320220723604e-06,
      "loss": 1.5659,
      "step": 115
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.2578125,
      "learning_rate": 6.171619579044014e-06,
      "loss": 1.741,
      "step": 116
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.3359375,
      "learning_rate": 6.042610846621297e-06,
      "loss": 1.6756,
      "step": 117
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.1748046875,
      "learning_rate": 5.915383534840671e-06,
      "loss": 1.6469,
      "step": 118
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.31640625,
      "learning_rate": 5.790014280677419e-06,
      "loss": 1.5172,
      "step": 119
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.2470703125,
      "learning_rate": 5.666578601882389e-06,
      "loss": 1.53,
      "step": 120
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.1796875,
      "learning_rate": 5.545150851492929e-06,
      "loss": 1.8531,
      "step": 121
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.4258041730454e-06,
      "loss": 1.6747,
      "step": 122
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.1328125,
      "learning_rate": 5.3086104565162404e-06,
      "loss": 1.7309,
      "step": 123
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.2265625,
      "learning_rate": 5.193640295018128e-06,
      "loss": 1.7294,
      "step": 124
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.24609375,
      "learning_rate": 5.080962942277341e-06,
      "loss": 1.4434,
      "step": 125
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.142578125,
      "learning_rate": 4.9706462709178905e-06,
      "loss": 1.6801,
      "step": 126
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.232421875,
      "learning_rate": 4.862756731577599e-06,
      "loss": 1.5832,
      "step": 127
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.259765625,
      "learning_rate": 4.757359312880715e-06,
      "loss": 1.8017,
      "step": 128
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.2294921875,
      "learning_rate": 4.6545175022912e-06,
      "loss": 1.7196,
      "step": 129
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.158203125,
      "learning_rate": 4.554293247870247e-06,
      "loss": 1.7319,
      "step": 130
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.2109375,
      "learning_rate": 4.4567469209610935e-06,
      "loss": 1.7938,
      "step": 131
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.44921875,
      "learning_rate": 4.3619372798235785e-06,
      "loss": 1.9002,
      "step": 132
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.25390625,
      "learning_rate": 4.269921434240363e-06,
      "loss": 1.6809,
      "step": 133
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.291015625,
      "learning_rate": 4.180754811116131e-06,
      "loss": 1.6639,
      "step": 134
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.31640625,
      "learning_rate": 4.0944911210904985e-06,
      "loss": 1.7448,
      "step": 135
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.011182326184728e-06,
      "loss": 1.6793,
      "step": 136
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.291015625,
      "learning_rate": 3.930878608501757e-06,
      "loss": 1.7793,
      "step": 137
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.2314453125,
      "learning_rate": 3.853628339998368e-06,
      "loss": 1.6064,
      "step": 138
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.2119140625,
      "learning_rate": 3.7794780533477323e-06,
      "loss": 1.6085,
      "step": 139
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.2578125,
      "learning_rate": 3.708472413909871e-06,
      "loss": 1.6289,
      "step": 140
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.453125,
      "learning_rate": 3.6406541928269087e-06,
      "loss": 1.6985,
      "step": 141
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.26171875,
      "learning_rate": 3.57606424125934e-06,
      "loss": 1.6574,
      "step": 142
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.5147414657788163e-06,
      "loss": 1.6721,
      "step": 143
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.357421875,
      "learning_rate": 3.4567228049322796e-06,
      "loss": 1.8479,
      "step": 144
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.4453125,
      "learning_rate": 3.4020432069915673e-06,
      "loss": 1.7129,
      "step": 145
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.169921875,
      "learning_rate": 3.3507356089018757e-06,
      "loss": 1.6321,
      "step": 146
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.23828125,
      "learning_rate": 3.30283091644178e-06,
      "loss": 1.6397,
      "step": 147
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.30859375,
      "learning_rate": 3.2583579856067474e-06,
      "loss": 1.6511,
      "step": 148
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.2255859375,
      "learning_rate": 3.217343605227361e-06,
      "loss": 1.7631,
      "step": 149
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.265625,
      "learning_rate": 3.1798124808327363e-06,
      "loss": 1.5645,
      "step": 150
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.322265625,
      "learning_rate": 3.1457872197688293e-06,
      "loss": 1.579,
      "step": 151
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.205078125,
      "learning_rate": 3.1152883175806177e-06,
      "loss": 1.6902,
      "step": 152
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2578125,
      "learning_rate": 3.0883341456663526e-06,
      "loss": 1.604,
      "step": 153
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.2578125,
      "learning_rate": 3.0649409402113143e-06,
      "loss": 1.551,
      "step": 154
    }
  ],
  "logging_steps": 1,
  "max_steps": 160,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 14,
  "total_flos": 5.355531142428623e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}