{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9987389659520807,
  "global_step": 297,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0.0,
      "loss": 8.8511,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0,
      "loss": 8.7808,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 6e-07,
      "loss": 8.804,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 6e-07,
      "loss": 8.7995,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 6e-07,
      "loss": 8.6549,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.2e-06,
      "loss": 8.7746,
      "step": 6
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8e-06,
      "loss": 8.7778,
      "step": 7
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.4e-06,
      "loss": 8.6186,
      "step": 8
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.9999999999999997e-06,
      "loss": 8.6414,
      "step": 9
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.6e-06,
      "loss": 8.2555,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.2e-06,
      "loss": 8.0864,
      "step": 11
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.8e-06,
      "loss": 7.9546,
      "step": 12
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.399999999999999e-06,
      "loss": 7.9313,
      "step": 13
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.399999999999999e-06,
      "loss": 7.8215,
      "step": 14
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.999999999999999e-06,
      "loss": 7.7881,
      "step": 15
    },
    {
      "epoch": 0.05,
      "learning_rate": 6.599999999999999e-06,
      "loss": 7.6596,
      "step": 16
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.2e-06,
      "loss": 7.6331,
      "step": 17
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.799999999999998e-06,
      "loss": 7.5963,
      "step": 18
    },
    {
      "epoch": 0.06,
      "learning_rate": 8.4e-06,
      "loss": 7.5002,
      "step": 19
    },
    {
      "epoch": 0.07,
      "learning_rate": 8.999999999999999e-06,
      "loss": 7.4665,
      "step": 20
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.6e-06,
      "loss": 7.4824,
      "step": 21
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.02e-05,
      "loss": 7.5147,
      "step": 22
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.0799999999999998e-05,
      "loss": 7.3182,
      "step": 23
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.14e-05,
      "loss": 7.3549,
      "step": 24
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.1999999999999999e-05,
      "loss": 7.2906,
      "step": 25
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.26e-05,
      "loss": 7.431,
      "step": 26
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.3199999999999997e-05,
      "loss": 7.3335,
      "step": 27
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.3799999999999998e-05,
      "loss": 7.285,
      "step": 28
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.44e-05,
      "loss": 7.2179,
      "step": 29
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 7.2501,
      "step": 30
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.5599999999999996e-05,
      "loss": 7.2811,
      "step": 31
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.6199999999999997e-05,
      "loss": 7.2661,
      "step": 32
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.68e-05,
      "loss": 7.2314,
      "step": 33
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.74e-05,
      "loss": 7.1629,
      "step": 34
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.7999999999999997e-05,
      "loss": 7.1379,
      "step": 35
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.8599999999999998e-05,
      "loss": 7.1309,
      "step": 36
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.92e-05,
      "loss": 7.2504,
      "step": 37
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.98e-05,
      "loss": 7.1629,
      "step": 38
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.04e-05,
      "loss": 7.2039,
      "step": 39
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.1e-05,
      "loss": 7.1528,
      "step": 40
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.1599999999999996e-05,
      "loss": 7.2184,
      "step": 41
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.2199999999999998e-05,
      "loss": 7.1632,
      "step": 42
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.28e-05,
      "loss": 7.0289,
      "step": 43
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.34e-05,
      "loss": 7.1943,
      "step": 44
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.3999999999999997e-05,
      "loss": 7.0794,
      "step": 45
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.4599999999999998e-05,
      "loss": 7.0557,
      "step": 46
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.52e-05,
      "loss": 7.0325,
      "step": 47
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.5799999999999997e-05,
      "loss": 6.916,
      "step": 48
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.6399999999999995e-05,
      "loss": 6.7969,
      "step": 49
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.6999999999999996e-05,
      "loss": 6.5192,
      "step": 50
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.7599999999999997e-05,
      "loss": 7.2997,
      "step": 51
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 7.2435,
      "step": 52
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.88e-05,
      "loss": 7.1027,
      "step": 53
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.94e-05,
      "loss": 7.0818,
      "step": 54
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 7.0494,
      "step": 55
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.06e-05,
      "loss": 7.0948,
      "step": 56
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.119999999999999e-05,
      "loss": 7.1433,
      "step": 57
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.1799999999999994e-05,
      "loss": 7.1552,
      "step": 58
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.2399999999999995e-05,
      "loss": 7.0391,
      "step": 59
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 7.0883,
      "step": 60
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.36e-05,
      "loss": 7.0281,
      "step": 61
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.42e-05,
      "loss": 6.9128,
      "step": 62
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.48e-05,
      "loss": 7.1093,
      "step": 63
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.539999999999999e-05,
      "loss": 7.045,
      "step": 64
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.5999999999999994e-05,
      "loss": 7.0655,
      "step": 65
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.6599999999999995e-05,
      "loss": 7.0467,
      "step": 66
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.7199999999999996e-05,
      "loss": 7.0838,
      "step": 67
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.78e-05,
      "loss": 7.0123,
      "step": 68
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.84e-05,
      "loss": 7.0332,
      "step": 69
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.9e-05,
      "loss": 7.0583,
      "step": 70
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.96e-05,
      "loss": 6.9911,
      "step": 71
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.02e-05,
      "loss": 7.0449,
      "step": 72
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.08e-05,
      "loss": 7.0333,
      "step": 73
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.14e-05,
      "loss": 7.0134,
      "step": 74
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.2e-05,
      "loss": 7.0769,
      "step": 75
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.259999999999999e-05,
      "loss": 7.0209,
      "step": 76
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.319999999999999e-05,
      "loss": 6.9658,
      "step": 77
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.3799999999999994e-05,
      "loss": 7.0328,
      "step": 78
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.4399999999999995e-05,
      "loss": 7.0265,
      "step": 79
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 7.0703,
      "step": 80
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.56e-05,
      "loss": 6.9814,
      "step": 81
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.62e-05,
      "loss": 7.0826,
      "step": 82
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.68e-05,
      "loss": 6.9739,
      "step": 83
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.7399999999999993e-05,
      "loss": 7.0487,
      "step": 84
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.7999999999999994e-05,
      "loss": 7.0852,
      "step": 85
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8599999999999995e-05,
      "loss": 6.975,
      "step": 86
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.9199999999999997e-05,
      "loss": 6.972,
      "step": 87
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.98e-05,
      "loss": 7.0417,
      "step": 88
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.04e-05,
      "loss": 6.952,
      "step": 89
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.1e-05,
      "loss": 7.0573,
      "step": 90
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.1599999999999994e-05,
      "loss": 6.9899,
      "step": 91
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.2199999999999995e-05,
      "loss": 7.1373,
      "step": 92
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.279999999999999e-05,
      "loss": 6.9841,
      "step": 93
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.339999999999999e-05,
      "loss": 6.9071,
      "step": 94
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.399999999999999e-05,
      "loss": 6.8056,
      "step": 95
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.459999999999999e-05,
      "loss": 6.9705,
      "step": 96
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.519999999999999e-05,
      "loss": 6.9171,
      "step": 97
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.5799999999999994e-05,
      "loss": 6.8158,
      "step": 98
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.6399999999999995e-05,
      "loss": 6.6855,
      "step": 99
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 6.4836,
      "step": 100
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.76e-05,
      "loss": 7.1406,
      "step": 101
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.82e-05,
      "loss": 7.1155,
      "step": 102
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.88e-05,
      "loss": 7.0086,
      "step": 103
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.94e-05,
      "loss": 7.0293,
      "step": 104
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 7.0157,
      "step": 105
    },
    {
      "epoch": 0.36,
      "learning_rate": 6.0599999999999996e-05,
      "loss": 7.0092,
      "step": 106
    },
    {
      "epoch": 0.36,
      "learning_rate": 6.12e-05,
      "loss": 7.0129,
      "step": 107
    },
    {
      "epoch": 0.36,
      "learning_rate": 6.18e-05,
      "loss": 7.0323,
      "step": 108
    },
    {
      "epoch": 0.37,
      "learning_rate": 6.239999999999999e-05,
      "loss": 7.0419,
      "step": 109
    },
    {
      "epoch": 0.37,
      "learning_rate": 6.299999999999999e-05,
      "loss": 7.0393,
      "step": 110
    },
    {
      "epoch": 0.37,
      "learning_rate": 6.359999999999999e-05,
      "loss": 7.018,
      "step": 111
    },
    {
      "epoch": 0.38,
      "learning_rate": 6.419999999999999e-05,
      "loss": 7.0016,
      "step": 112
    },
    {
      "epoch": 0.38,
      "learning_rate": 6.479999999999999e-05,
      "loss": 6.9545,
      "step": 113
    },
    {
      "epoch": 0.38,
      "learning_rate": 6.539999999999999e-05,
      "loss": 6.9617,
      "step": 114
    },
    {
      "epoch": 0.39,
      "learning_rate": 6.599999999999999e-05,
      "loss": 6.9617,
      "step": 115
    },
    {
      "epoch": 0.39,
      "learning_rate": 6.659999999999999e-05,
      "loss": 7.0031,
      "step": 116
    },
    {
      "epoch": 0.39,
      "learning_rate": 6.72e-05,
      "loss": 6.9382,
      "step": 117
    },
    {
      "epoch": 0.4,
      "learning_rate": 6.78e-05,
      "loss": 6.9027,
      "step": 118
    },
    {
      "epoch": 0.4,
      "learning_rate": 6.84e-05,
      "loss": 6.959,
      "step": 119
    },
    {
      "epoch": 0.4,
      "learning_rate": 6.9e-05,
      "loss": 7.0171,
      "step": 120
    },
    {
      "epoch": 0.41,
      "learning_rate": 6.96e-05,
      "loss": 7.0033,
      "step": 121
    },
    {
      "epoch": 0.41,
      "learning_rate": 7.02e-05,
      "loss": 7.0366,
      "step": 122
    },
    {
      "epoch": 0.41,
      "learning_rate": 7.079999999999999e-05,
      "loss": 6.955,
      "step": 123
    },
    {
      "epoch": 0.42,
      "learning_rate": 7.139999999999999e-05,
      "loss": 7.0449,
      "step": 124
    },
    {
      "epoch": 0.42,
      "learning_rate": 7.199999999999999e-05,
      "loss": 7.0208,
      "step": 125
    },
    {
      "epoch": 0.42,
      "learning_rate": 7.259999999999999e-05,
      "loss": 6.9296,
      "step": 126
    },
    {
      "epoch": 0.43,
      "learning_rate": 7.319999999999999e-05,
      "loss": 7.0125,
      "step": 127
    },
    {
      "epoch": 0.43,
      "learning_rate": 7.379999999999999e-05,
      "loss": 6.9242,
      "step": 128
    },
    {
      "epoch": 0.43,
      "learning_rate": 7.439999999999999e-05,
      "loss": 6.9626,
      "step": 129
    },
    {
      "epoch": 0.44,
      "learning_rate": 7.5e-05,
      "loss": 7.0513,
      "step": 130
    },
    {
      "epoch": 0.44,
      "learning_rate": 7.56e-05,
      "loss": 7.0241,
      "step": 131
    },
    {
      "epoch": 0.44,
      "learning_rate": 7.62e-05,
      "loss": 6.8046,
      "step": 132
    },
    {
      "epoch": 0.45,
      "learning_rate": 7.68e-05,
      "loss": 7.0524,
      "step": 133
    },
    {
      "epoch": 0.45,
      "learning_rate": 7.74e-05,
      "loss": 6.9242,
      "step": 134
    },
    {
      "epoch": 0.45,
      "learning_rate": 7.8e-05,
      "loss": 7.0188,
      "step": 135
    },
    {
      "epoch": 0.46,
      "learning_rate": 7.86e-05,
      "loss": 6.9299,
      "step": 136
    },
    {
      "epoch": 0.46,
      "learning_rate": 7.92e-05,
      "loss": 6.8704,
      "step": 137
    },
    {
      "epoch": 0.46,
      "learning_rate": 7.98e-05,
      "loss": 6.8663,
      "step": 138
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.04e-05,
      "loss": 6.9004,
      "step": 139
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.1e-05,
      "loss": 6.9129,
      "step": 140
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.16e-05,
      "loss": 7.0426,
      "step": 141
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.22e-05,
      "loss": 6.8777,
      "step": 142
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.28e-05,
      "loss": 6.9102,
      "step": 143
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.34e-05,
      "loss": 6.8713,
      "step": 144
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.4e-05,
      "loss": 6.7546,
      "step": 145
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.459999999999998e-05,
      "loss": 6.6504,
      "step": 146
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.519999999999998e-05,
      "loss": 6.7166,
      "step": 147
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.579999999999998e-05,
      "loss": 6.687,
      "step": 148
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.639999999999999e-05,
      "loss": 6.5554,
      "step": 149
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.699999999999999e-05,
      "loss": 6.396,
      "step": 150
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.759999999999999e-05,
      "loss": 7.3745,
      "step": 151
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.819999999999999e-05,
      "loss": 7.1846,
      "step": 152
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.879999999999999e-05,
      "loss": 7.0721,
      "step": 153
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.939999999999999e-05,
      "loss": 7.0759,
      "step": 154
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.999999999999999e-05,
      "loss": 7.0266,
      "step": 155
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.059999999999999e-05,
      "loss": 7.0283,
      "step": 156
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.12e-05,
      "loss": 6.914,
      "step": 157
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.18e-05,
      "loss": 6.9835,
      "step": 158
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.24e-05,
      "loss": 6.9416,
      "step": 159
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.3e-05,
      "loss": 6.9389,
      "step": 160
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.36e-05,
      "loss": 6.9157,
      "step": 161
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.419999999999999e-05,
      "loss": 6.9412,
      "step": 162
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.479999999999999e-05,
      "loss": 6.9444,
      "step": 163
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.539999999999999e-05,
      "loss": 6.9105,
      "step": 164
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.599999999999999e-05,
      "loss": 6.9351,
      "step": 165
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.659999999999999e-05,
      "loss": 6.8575,
      "step": 166
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.719999999999999e-05,
      "loss": 6.8637,
      "step": 167
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.779999999999999e-05,
      "loss": 6.928,
      "step": 168
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.839999999999999e-05,
      "loss": 6.9464,
      "step": 169
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.9e-05,
      "loss": 6.9578,
      "step": 170
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.96e-05,
      "loss": 6.8796,
      "step": 171
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0001002,
      "loss": 6.8743,
      "step": 172
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0001008,
      "loss": 6.8323,
      "step": 173
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001014,
      "loss": 6.76,
      "step": 174
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.000102,
      "loss": 7.0137,
      "step": 175
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001026,
      "loss": 6.8641,
      "step": 176
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00010319999999999999,
      "loss": 6.9469,
      "step": 177
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00010379999999999999,
      "loss": 6.887,
      "step": 178
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00010439999999999999,
      "loss": 6.9133,
      "step": 179
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00010499999999999999,
      "loss": 6.9685,
      "step": 180
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00010559999999999998,
      "loss": 6.8731,
      "step": 181
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00010619999999999998,
      "loss": 6.8619,
      "step": 182
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00010679999999999998,
      "loss": 6.8129,
      "step": 183
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00010739999999999998,
      "loss": 6.9164,
      "step": 184
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00010799999999999998,
      "loss": 6.8933,
      "step": 185
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00010859999999999998,
      "loss": 6.9132,
      "step": 186
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00010919999999999998,
      "loss": 6.9301,
      "step": 187
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00010979999999999999,
      "loss": 6.9065,
      "step": 188
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00011039999999999999,
      "loss": 6.9315,
      "step": 189
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00011099999999999999,
      "loss": 6.9694,
      "step": 190
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00011159999999999999,
      "loss": 6.8879,
      "step": 191
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00011219999999999999,
      "loss": 6.9392,
      "step": 192
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00011279999999999999,
      "loss": 6.8509,
      "step": 193
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00011339999999999999,
      "loss": 6.8079,
      "step": 194
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00011399999999999999,
      "loss": 6.7967,
      "step": 195
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001146,
      "loss": 6.716,
      "step": 196
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001152,
      "loss": 6.668,
      "step": 197
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0001158,
      "loss": 6.8254,
      "step": 198
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0001164,
      "loss": 6.546,
      "step": 199
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.000117,
      "loss": 6.3884,
      "step": 200
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0001176,
      "loss": 7.1095,
      "step": 201
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0001182,
      "loss": 7.0303,
      "step": 202
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0001188,
      "loss": 6.9859,
      "step": 203
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0001194,
      "loss": 6.8691,
      "step": 204
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00011999999999999999,
      "loss": 6.9824,
      "step": 205
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00012059999999999999,
      "loss": 6.8834,
      "step": 206
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00012119999999999999,
      "loss": 6.9037,
      "step": 207
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00012179999999999999,
      "loss": 7.0387,
      "step": 208
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0001224,
      "loss": 6.8654,
      "step": 209
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00012299999999999998,
      "loss": 6.9355,
      "step": 210
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0001236,
      "loss": 6.9106,
      "step": 211
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00012419999999999998,
      "loss": 6.8523,
      "step": 212
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00012479999999999997,
      "loss": 6.8772,
      "step": 213
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00012539999999999999,
      "loss": 6.8534,
      "step": 214
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00012599999999999997,
      "loss": 6.9606,
      "step": 215
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001266,
      "loss": 6.9924,
      "step": 216
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00012719999999999997,
      "loss": 6.8962,
      "step": 217
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001278,
      "loss": 6.8136,
      "step": 218
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00012839999999999998,
      "loss": 6.8424,
      "step": 219
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.000129,
      "loss": 6.8619,
      "step": 220
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00012959999999999998,
      "loss": 6.8796,
      "step": 221
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001302,
      "loss": 6.8676,
      "step": 222
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00013079999999999998,
      "loss": 6.8847,
      "step": 223
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001314,
      "loss": 6.8826,
      "step": 224
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00013199999999999998,
      "loss": 6.9246,
      "step": 225
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0001326,
      "loss": 6.9268,
      "step": 226
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00013319999999999999,
      "loss": 6.9359,
      "step": 227
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0001338,
      "loss": 6.8329,
      "step": 228
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0001344,
      "loss": 6.9592,
      "step": 229
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.000135,
      "loss": 6.8613,
      "step": 230
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0001356,
      "loss": 6.958,
      "step": 231
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0001362,
      "loss": 6.8579,
      "step": 232
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0001368,
      "loss": 6.9645,
      "step": 233
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0001374,
      "loss": 6.9207,
      "step": 234
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.000138,
      "loss": 7.005,
      "step": 235
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0001386,
      "loss": 6.8842,
      "step": 236
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0001392,
      "loss": 6.802,
      "step": 237
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00013979999999999998,
      "loss": 6.9791,
      "step": 238
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0001404,
      "loss": 6.8732,
      "step": 239
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00014099999999999998,
      "loss": 6.8599,
      "step": 240
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00014159999999999997,
      "loss": 6.8668,
      "step": 241
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001422,
      "loss": 6.9267,
      "step": 242
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00014279999999999997,
      "loss": 6.8407,
      "step": 243
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0001434,
      "loss": 6.7091,
      "step": 244
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00014399999999999998,
      "loss": 6.8146,
      "step": 245
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0001446,
      "loss": 6.8591,
      "step": 246
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00014519999999999998,
      "loss": 6.6833,
      "step": 247
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0001458,
      "loss": 6.5523,
      "step": 248
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00014639999999999998,
      "loss": 6.6264,
      "step": 249
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.000147,
      "loss": 6.3599,
      "step": 250
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00014759999999999998,
      "loss": 7.0614,
      "step": 251
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001482,
      "loss": 6.9579,
      "step": 252
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00014879999999999998,
      "loss": 7.0169,
      "step": 253
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001494,
      "loss": 6.949,
      "step": 254
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00015,
      "loss": 6.8869,
      "step": 255
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00015059999999999997,
      "loss": 6.878,
      "step": 256
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001512,
      "loss": 6.948,
      "step": 257
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00015179999999999998,
      "loss": 6.8967,
      "step": 258
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0001524,
      "loss": 6.8811,
      "step": 259
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00015299999999999998,
      "loss": 6.819,
      "step": 260
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001536,
      "loss": 6.8911,
      "step": 261
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00015419999999999998,
      "loss": 6.935,
      "step": 262
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001548,
      "loss": 6.7874,
      "step": 263
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00015539999999999998,
      "loss": 6.9012,
      "step": 264
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.000156,
      "loss": 6.838,
      "step": 265
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00015659999999999998,
      "loss": 6.9578,
      "step": 266
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0001572,
      "loss": 6.8522,
      "step": 267
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0001578,
      "loss": 6.8601,
      "step": 268
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0001584,
      "loss": 6.9054,
      "step": 269
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.000159,
      "loss": 6.876,
      "step": 270
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001596,
      "loss": 6.8331,
      "step": 271
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001602,
      "loss": 6.7783,
      "step": 272
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001608,
      "loss": 6.8498,
      "step": 273
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001614,
      "loss": 6.7725,
      "step": 274
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.000162,
      "loss": 6.932,
      "step": 275
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0001626,
      "loss": 6.839,
      "step": 276
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0001632,
      "loss": 6.862,
      "step": 277
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0001638,
      "loss": 6.8377,
      "step": 278
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0001644,
      "loss": 6.7948,
      "step": 279
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.000165,
      "loss": 6.7516,
      "step": 280
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0001656,
      "loss": 6.8518,
      "step": 281
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001662,
      "loss": 6.8144,
      "step": 282
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001668,
      "loss": 6.9245,
      "step": 283
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0001674,
      "loss": 6.7959,
      "step": 284
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.000168,
      "loss": 6.8214,
      "step": 285
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0001686,
      "loss": 6.9041,
      "step": 286
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00016919999999999997,
      "loss": 6.797,
      "step": 287
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00016979999999999998,
      "loss": 6.7532,
      "step": 288
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00017039999999999997,
      "loss": 6.7621,
      "step": 289
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00017099999999999998,
      "loss": 6.8274,
      "step": 290
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00017159999999999997,
      "loss": 6.7489,
      "step": 291
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00017219999999999998,
      "loss": 6.7898,
      "step": 292
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00017279999999999997,
      "loss": 6.7111,
      "step": 293
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00017339999999999996,
      "loss": 6.6793,
      "step": 294
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00017399999999999997,
      "loss": 6.5927,
      "step": 295
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00017459999999999996,
      "loss": 6.6744,
      "step": 296
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00017519999999999998,
      "loss": 6.3977,
      "step": 297
    },
    {
      "epoch": 1.0,
      "step": 297,
      "total_flos": 0.0,
      "train_loss": 7.029080853317723,
      "train_runtime": 4589.8897,
      "train_samples_per_second": 6.218,
      "train_steps_per_second": 0.065
    }
  ],
  "max_steps": 297,
  "num_train_epochs": 1,
  "total_flos": 0.0,
  "trial_name": null,
  "trial_params": null
}